From 849369d6c66d3054688672f97d31fceb8e8230fb Mon Sep 17 00:00:00 2001 From: root Date: Fri, 25 Dec 2015 04:40:36 +0000 Subject: initial_commit --- drivers/gpu/Makefile | 1 + drivers/gpu/drm/Kconfig | 43 + drivers/gpu/drm/Makefile | 60 + drivers/gpu/drm/README.drm | 43 + drivers/gpu/drm/ati_pcigart.c | 201 + drivers/gpu/drm/drm_agpsupport.c | 469 ++ drivers/gpu/drm/drm_auth.c | 194 + drivers/gpu/drm/drm_buffer.c | 184 + drivers/gpu/drm/drm_bufs.c | 1629 +++++ drivers/gpu/drm/drm_cache.c | 99 + drivers/gpu/drm/drm_context.c | 461 ++ drivers/gpu/drm/drm_crtc.c | 2781 ++++++++ drivers/gpu/drm/drm_crtc_helper.c | 964 +++ drivers/gpu/drm/drm_debugfs.c | 236 + drivers/gpu/drm/drm_dma.c | 161 + drivers/gpu/drm/drm_dp_i2c_helper.c | 208 + drivers/gpu/drm/drm_drv.c | 490 ++ drivers/gpu/drm/drm_edid.c | 1600 +++++ drivers/gpu/drm/drm_edid_modes.h | 380 ++ drivers/gpu/drm/drm_encoder_slave.c | 123 + drivers/gpu/drm/drm_fb_helper.c | 1461 ++++ drivers/gpu/drm/drm_fops.c | 658 ++ drivers/gpu/drm/drm_gem.c | 584 ++ drivers/gpu/drm/drm_global.c | 112 + drivers/gpu/drm/drm_hashtab.c | 196 + drivers/gpu/drm/drm_info.c | 315 + drivers/gpu/drm/drm_ioc32.c | 1083 +++ drivers/gpu/drm/drm_ioctl.c | 355 + drivers/gpu/drm/drm_irq.c | 1338 ++++ drivers/gpu/drm/drm_lock.c | 374 + drivers/gpu/drm/drm_memory.c | 162 + drivers/gpu/drm/drm_mm.c | 716 ++ drivers/gpu/drm/drm_modes.c | 1132 ++++ drivers/gpu/drm/drm_pci.c | 468 ++ drivers/gpu/drm/drm_platform.c | 196 + drivers/gpu/drm/drm_proc.c | 221 + drivers/gpu/drm/drm_scatter.c | 217 + drivers/gpu/drm/drm_sman.c | 350 + drivers/gpu/drm/drm_stub.c | 494 ++ drivers/gpu/drm/drm_sysfs.c | 561 ++ drivers/gpu/drm/drm_trace.h | 66 + drivers/gpu/drm/drm_trace_points.c | 4 + drivers/gpu/drm/drm_usb.c | 117 + drivers/gpu/drm/drm_vm.c | 690 ++ drivers/gpu/drm/i2c/Makefile | 7 + drivers/gpu/drm/i2c/ch7006_drv.c | 555 ++ drivers/gpu/drm/i2c/ch7006_mode.c | 471 ++ drivers/gpu/drm/i2c/ch7006_priv.h | 345 + drivers/gpu/drm/i2c/sil164_drv.c | 462 ++ drivers/gpu/drm/i810/Makefile | 8 + drivers/gpu/drm/i810/i810_dma.c | 1276 ++++ drivers/gpu/drm/i810/i810_drv.c | 100 + drivers/gpu/drm/i810/i810_drv.h | 245 + drivers/gpu/drm/i915/Makefile | 44 + drivers/gpu/drm/i915/dvo.h | 144 + drivers/gpu/drm/i915/dvo_ch7017.c | 401 ++ drivers/gpu/drm/i915/dvo_ch7xxx.c | 331 + drivers/gpu/drm/i915/dvo_ivch.c | 421 ++ drivers/gpu/drm/i915/dvo_sil164.c | 263 + drivers/gpu/drm/i915/dvo_tfp410.c | 304 + drivers/gpu/drm/i915/i915_debugfs.c | 1470 ++++ drivers/gpu/drm/i915/i915_dma.c | 2321 +++++++ drivers/gpu/drm/i915/i915_drv.c | 869 +++ drivers/gpu/drm/i915/i915_drv.h | 1403 ++++ drivers/gpu/drm/i915/i915_gem.c | 4149 ++++++++++++ drivers/gpu/drm/i915/i915_gem_debug.c | 203 + drivers/gpu/drm/i915/i915_gem_evict.c | 224 + drivers/gpu/drm/i915/i915_gem_execbuffer.c | 1348 ++++ drivers/gpu/drm/i915/i915_gem_gtt.c | 122 + drivers/gpu/drm/i915/i915_gem_tiling.c | 491 ++ drivers/gpu/drm/i915/i915_ioc32.c | 219 + drivers/gpu/drm/i915/i915_irq.c | 2088 ++++++ drivers/gpu/drm/i915/i915_mem.c | 387 ++ drivers/gpu/drm/i915/i915_reg.h | 3484 ++++++++++ drivers/gpu/drm/i915/i915_suspend.c | 912 +++ drivers/gpu/drm/i915/i915_trace.h | 418 ++ drivers/gpu/drm/i915/i915_trace_points.c | 11 + drivers/gpu/drm/i915/intel_acpi.c | 252 + drivers/gpu/drm/i915/intel_bios.c | 703 ++ drivers/gpu/drm/i915/intel_bios.h | 611 ++ drivers/gpu/drm/i915/intel_crt.c | 609 ++ drivers/gpu/drm/i915/intel_display.c | 8240 +++++++++++++++++++++++ drivers/gpu/drm/i915/intel_dp.c | 2083 ++++++ drivers/gpu/drm/i915/intel_drv.h | 350 + drivers/gpu/drm/i915/intel_dvo.c | 448 ++ drivers/gpu/drm/i915/intel_fb.c | 286 + drivers/gpu/drm/i915/intel_hdmi.c | 449 ++ drivers/gpu/drm/i915/intel_i2c.c | 472 ++ drivers/gpu/drm/i915/intel_lvds.c | 1069 +++ drivers/gpu/drm/i915/intel_modes.c | 142 + drivers/gpu/drm/i915/intel_opregion.c | 516 ++ drivers/gpu/drm/i915/intel_overlay.c | 1617 +++++ drivers/gpu/drm/i915/intel_panel.c | 316 + drivers/gpu/drm/i915/intel_ringbuffer.c | 1373 ++++ drivers/gpu/drm/i915/intel_ringbuffer.h | 203 + drivers/gpu/drm/i915/intel_sdvo.c | 2618 +++++++ drivers/gpu/drm/i915/intel_sdvo_regs.h | 728 ++ drivers/gpu/drm/i915/intel_tv.c | 1730 +++++ drivers/gpu/drm/mga/Makefile | 11 + drivers/gpu/drm/mga/mga_dma.c | 1154 ++++ drivers/gpu/drm/mga/mga_drv.c | 142 + drivers/gpu/drm/mga/mga_drv.h | 666 ++ drivers/gpu/drm/mga/mga_ioc32.c | 226 + drivers/gpu/drm/mga/mga_irq.c | 174 + drivers/gpu/drm/mga/mga_state.c | 1103 +++ drivers/gpu/drm/mga/mga_warp.c | 169 + drivers/gpu/drm/nouveau/Kconfig | 56 + drivers/gpu/drm/nouveau/Makefile | 42 + drivers/gpu/drm/nouveau/nouveau_acpi.c | 387 ++ drivers/gpu/drm/nouveau/nouveau_backlight.c | 174 + drivers/gpu/drm/nouveau/nouveau_bios.c | 6930 +++++++++++++++++++ drivers/gpu/drm/nouveau/nouveau_bios.h | 336 + drivers/gpu/drm/nouveau/nouveau_bo.c | 1115 +++ drivers/gpu/drm/nouveau/nouveau_calc.c | 478 ++ drivers/gpu/drm/nouveau/nouveau_channel.c | 491 ++ drivers/gpu/drm/nouveau/nouveau_connector.c | 942 +++ drivers/gpu/drm/nouveau/nouveau_connector.h | 58 + drivers/gpu/drm/nouveau/nouveau_crtc.h | 97 + drivers/gpu/drm/nouveau/nouveau_debugfs.c | 197 + drivers/gpu/drm/nouveau/nouveau_display.c | 368 + drivers/gpu/drm/nouveau/nouveau_dma.c | 338 + drivers/gpu/drm/nouveau/nouveau_dma.h | 175 + drivers/gpu/drm/nouveau/nouveau_dp.c | 637 ++ drivers/gpu/drm/nouveau/nouveau_drv.c | 479 ++ drivers/gpu/drm/nouveau/nouveau_drv.h | 1583 +++++ drivers/gpu/drm/nouveau/nouveau_encoder.h | 103 + drivers/gpu/drm/nouveau/nouveau_fb.h | 48 + drivers/gpu/drm/nouveau/nouveau_fbcon.c | 537 ++ drivers/gpu/drm/nouveau/nouveau_fbcon.h | 69 + drivers/gpu/drm/nouveau/nouveau_fence.c | 617 ++ drivers/gpu/drm/nouveau/nouveau_gem.c | 808 +++ drivers/gpu/drm/nouveau/nouveau_grctx.h | 133 + drivers/gpu/drm/nouveau/nouveau_hw.c | 1087 +++ drivers/gpu/drm/nouveau/nouveau_hw.h | 474 ++ drivers/gpu/drm/nouveau/nouveau_i2c.c | 323 + drivers/gpu/drm/nouveau/nouveau_i2c.h | 53 + drivers/gpu/drm/nouveau/nouveau_ioc32.c | 70 + drivers/gpu/drm/nouveau/nouveau_irq.c | 151 + drivers/gpu/drm/nouveau/nouveau_mem.c | 930 +++ drivers/gpu/drm/nouveau/nouveau_mm.c | 171 + drivers/gpu/drm/nouveau/nouveau_mm.h | 65 + drivers/gpu/drm/nouveau/nouveau_notifier.c | 194 + drivers/gpu/drm/nouveau/nouveau_object.c | 998 +++ drivers/gpu/drm/nouveau/nouveau_perf.c | 296 + drivers/gpu/drm/nouveau/nouveau_pm.c | 557 ++ drivers/gpu/drm/nouveau/nouveau_pm.h | 74 + drivers/gpu/drm/nouveau/nouveau_ramht.c | 309 + drivers/gpu/drm/nouveau/nouveau_ramht.h | 55 + drivers/gpu/drm/nouveau/nouveau_reg.h | 855 +++ drivers/gpu/drm/nouveau/nouveau_sgdma.c | 538 ++ drivers/gpu/drm/nouveau/nouveau_state.c | 1140 ++++ drivers/gpu/drm/nouveau/nouveau_temp.c | 307 + drivers/gpu/drm/nouveau/nouveau_ttm.c | 103 + drivers/gpu/drm/nouveau/nouveau_util.c | 78 + drivers/gpu/drm/nouveau/nouveau_util.h | 49 + drivers/gpu/drm/nouveau/nouveau_vm.c | 432 ++ drivers/gpu/drm/nouveau/nouveau_vm.h | 113 + drivers/gpu/drm/nouveau/nouveau_volt.c | 220 + drivers/gpu/drm/nouveau/nv04_crtc.c | 1052 +++ drivers/gpu/drm/nouveau/nv04_cursor.c | 71 + drivers/gpu/drm/nouveau/nv04_dac.c | 544 ++ drivers/gpu/drm/nouveau/nv04_dfp.c | 712 ++ drivers/gpu/drm/nouveau/nv04_display.c | 281 + drivers/gpu/drm/nouveau/nv04_fb.c | 21 + drivers/gpu/drm/nouveau/nv04_fbcon.c | 275 + drivers/gpu/drm/nouveau/nv04_fifo.c | 543 ++ drivers/gpu/drm/nouveau/nv04_graph.c | 1359 ++++ drivers/gpu/drm/nouveau/nv04_instmem.c | 166 + drivers/gpu/drm/nouveau/nv04_mc.c | 24 + drivers/gpu/drm/nouveau/nv04_pm.c | 90 + drivers/gpu/drm/nouveau/nv04_timer.c | 51 + drivers/gpu/drm/nouveau/nv04_tv.c | 254 + drivers/gpu/drm/nouveau/nv10_fb.c | 144 + drivers/gpu/drm/nouveau/nv10_fifo.c | 246 + drivers/gpu/drm/nouveau/nv10_gpio.c | 92 + drivers/gpu/drm/nouveau/nv10_graph.c | 1189 ++++ drivers/gpu/drm/nouveau/nv17_tv.c | 827 +++ drivers/gpu/drm/nouveau/nv17_tv.h | 161 + drivers/gpu/drm/nouveau/nv17_tv_modes.c | 593 ++ drivers/gpu/drm/nouveau/nv20_graph.c | 835 +++ drivers/gpu/drm/nouveau/nv30_fb.c | 116 + drivers/gpu/drm/nouveau/nv40_fb.c | 118 + drivers/gpu/drm/nouveau/nv40_fifo.c | 307 + drivers/gpu/drm/nouveau/nv40_graph.c | 562 ++ drivers/gpu/drm/nouveau/nv40_grctx.c | 662 ++ drivers/gpu/drm/nouveau/nv40_mc.c | 28 + drivers/gpu/drm/nouveau/nv40_mpeg.c | 311 + drivers/gpu/drm/nouveau/nv50_calc.c | 97 + drivers/gpu/drm/nouveau/nv50_crtc.c | 788 +++ drivers/gpu/drm/nouveau/nv50_cursor.c | 157 + drivers/gpu/drm/nouveau/nv50_dac.c | 319 + drivers/gpu/drm/nouveau/nv50_display.c | 1011 +++ drivers/gpu/drm/nouveau/nv50_display.h | 87 + drivers/gpu/drm/nouveau/nv50_evo.c | 425 ++ drivers/gpu/drm/nouveau/nv50_evo.h | 119 + drivers/gpu/drm/nouveau/nv50_fb.c | 294 + drivers/gpu/drm/nouveau/nv50_fbcon.c | 264 + drivers/gpu/drm/nouveau/nv50_fifo.c | 504 ++ drivers/gpu/drm/nouveau/nv50_gpio.c | 302 + drivers/gpu/drm/nouveau/nv50_graph.c | 1121 +++ drivers/gpu/drm/nouveau/nv50_grctx.c | 3322 +++++++++ drivers/gpu/drm/nouveau/nv50_instmem.c | 428 ++ drivers/gpu/drm/nouveau/nv50_mc.c | 40 + drivers/gpu/drm/nouveau/nv50_mpeg.c | 256 + drivers/gpu/drm/nouveau/nv50_pm.c | 146 + drivers/gpu/drm/nouveau/nv50_sor.c | 341 + drivers/gpu/drm/nouveau/nv50_vm.c | 182 + drivers/gpu/drm/nouveau/nv50_vram.c | 211 + drivers/gpu/drm/nouveau/nv84_crypt.c | 193 + drivers/gpu/drm/nouveau/nva3_copy.c | 226 + drivers/gpu/drm/nouveau/nva3_copy.fuc | 870 +++ drivers/gpu/drm/nouveau/nva3_copy.fuc.h | 534 ++ drivers/gpu/drm/nouveau/nva3_pm.c | 204 + drivers/gpu/drm/nouveau/nvc0_copy.c | 243 + drivers/gpu/drm/nouveau/nvc0_copy.fuc.h | 527 ++ drivers/gpu/drm/nouveau/nvc0_fb.c | 38 + drivers/gpu/drm/nouveau/nvc0_fbcon.c | 269 + drivers/gpu/drm/nouveau/nvc0_fifo.c | 508 ++ drivers/gpu/drm/nouveau/nvc0_graph.c | 796 +++ drivers/gpu/drm/nouveau/nvc0_graph.h | 75 + drivers/gpu/drm/nouveau/nvc0_grctx.c | 2874 ++++++++ drivers/gpu/drm/nouveau/nvc0_instmem.c | 231 + drivers/gpu/drm/nouveau/nvc0_vm.c | 130 + drivers/gpu/drm/nouveau/nvc0_vram.c | 113 + drivers/gpu/drm/nouveau/nvreg.h | 517 ++ drivers/gpu/drm/r128/Makefile | 10 + drivers/gpu/drm/r128/r128_cce.c | 937 +++ drivers/gpu/drm/r128/r128_drv.c | 111 + drivers/gpu/drm/r128/r128_drv.h | 530 ++ drivers/gpu/drm/r128/r128_ioc32.c | 215 + drivers/gpu/drm/r128/r128_irq.c | 116 + drivers/gpu/drm/r128/r128_state.c | 1667 +++++ drivers/gpu/drm/radeon/.gitignore | 3 + drivers/gpu/drm/radeon/Kconfig | 31 + drivers/gpu/drm/radeon/Makefile | 80 + drivers/gpu/drm/radeon/ObjectID.h | 691 ++ drivers/gpu/drm/radeon/atom-bits.h | 48 + drivers/gpu/drm/radeon/atom-names.h | 100 + drivers/gpu/drm/radeon/atom-types.h | 42 + drivers/gpu/drm/radeon/atom.c | 1401 ++++ drivers/gpu/drm/radeon/atom.h | 159 + drivers/gpu/drm/radeon/atombios.h | 7023 +++++++++++++++++++ drivers/gpu/drm/radeon/atombios_crtc.c | 1636 +++++ drivers/gpu/drm/radeon/atombios_dp.c | 968 +++ drivers/gpu/drm/radeon/avivod.h | 62 + drivers/gpu/drm/radeon/cayman_blit_shaders.c | 373 + drivers/gpu/drm/radeon/cayman_blit_shaders.h | 35 + drivers/gpu/drm/radeon/evergreen.c | 3348 +++++++++ drivers/gpu/drm/radeon/evergreen_blit_kms.c | 988 +++ drivers/gpu/drm/radeon/evergreen_blit_shaders.c | 356 + drivers/gpu/drm/radeon/evergreen_blit_shaders.h | 35 + drivers/gpu/drm/radeon/evergreen_cs.c | 1442 ++++ drivers/gpu/drm/radeon/evergreen_reg.h | 194 + drivers/gpu/drm/radeon/evergreend.h | 1129 ++++ drivers/gpu/drm/radeon/mkregtable.c | 725 ++ drivers/gpu/drm/radeon/ni.c | 1566 +++++ drivers/gpu/drm/radeon/ni_reg.h | 86 + drivers/gpu/drm/radeon/nid.h | 538 ++ drivers/gpu/drm/radeon/r100.c | 3972 +++++++++++ drivers/gpu/drm/radeon/r100_track.h | 197 + drivers/gpu/drm/radeon/r100d.h | 880 +++ drivers/gpu/drm/radeon/r200.c | 535 ++ drivers/gpu/drm/radeon/r300.c | 1558 +++++ drivers/gpu/drm/radeon/r300_cmdbuf.c | 1185 ++++ drivers/gpu/drm/radeon/r300_reg.h | 1789 +++++ drivers/gpu/drm/radeon/r300d.h | 354 + drivers/gpu/drm/radeon/r420.c | 468 ++ drivers/gpu/drm/radeon/r420d.h | 249 + drivers/gpu/drm/radeon/r500_reg.h | 793 +++ drivers/gpu/drm/radeon/r520.c | 309 + drivers/gpu/drm/radeon/r520d.h | 187 + drivers/gpu/drm/radeon/r600.c | 3776 +++++++++++ drivers/gpu/drm/radeon/r600_audio.c | 282 + drivers/gpu/drm/radeon/r600_blit.c | 868 +++ drivers/gpu/drm/radeon/r600_blit_kms.c | 840 +++ drivers/gpu/drm/radeon/r600_blit_shaders.c | 718 ++ drivers/gpu/drm/radeon/r600_blit_shaders.h | 38 + drivers/gpu/drm/radeon/r600_cp.c | 2667 ++++++++ drivers/gpu/drm/radeon/r600_cs.c | 1917 ++++++ drivers/gpu/drm/radeon/r600_hdmi.c | 593 ++ drivers/gpu/drm/radeon/r600_reg.h | 200 + drivers/gpu/drm/radeon/r600d.h | 1520 +++++ drivers/gpu/drm/radeon/radeon.h | 1523 +++++ drivers/gpu/drm/radeon/radeon_acpi.c | 67 + drivers/gpu/drm/radeon/radeon_agp.c | 283 + drivers/gpu/drm/radeon/radeon_asic.c | 1066 +++ drivers/gpu/drm/radeon/radeon_asic.h | 443 ++ drivers/gpu/drm/radeon/radeon_atombios.c | 3100 +++++++++ drivers/gpu/drm/radeon/radeon_atpx_handler.c | 283 + drivers/gpu/drm/radeon/radeon_benchmark.c | 160 + drivers/gpu/drm/radeon/radeon_bios.c | 527 ++ drivers/gpu/drm/radeon/radeon_clocks.c | 912 +++ drivers/gpu/drm/radeon/radeon_combios.c | 3508 ++++++++++ drivers/gpu/drm/radeon/radeon_connectors.c | 1959 ++++++ drivers/gpu/drm/radeon/radeon_cp.c | 2257 +++++++ drivers/gpu/drm/radeon/radeon_cs.c | 341 + drivers/gpu/drm/radeon/radeon_cursor.c | 299 + drivers/gpu/drm/radeon/radeon_device.c | 1034 +++ drivers/gpu/drm/radeon/radeon_display.c | 1651 +++++ drivers/gpu/drm/radeon/radeon_drv.c | 426 ++ drivers/gpu/drm/radeon/radeon_drv.h | 2171 ++++++ drivers/gpu/drm/radeon/radeon_encoders.c | 2421 +++++++ drivers/gpu/drm/radeon/radeon_family.h | 110 + drivers/gpu/drm/radeon/radeon_fb.c | 419 ++ drivers/gpu/drm/radeon/radeon_fence.c | 437 ++ drivers/gpu/drm/radeon/radeon_gart.c | 290 + drivers/gpu/drm/radeon/radeon_gem.c | 390 ++ drivers/gpu/drm/radeon/radeon_i2c.c | 1168 ++++ drivers/gpu/drm/radeon/radeon_ioc32.c | 425 ++ drivers/gpu/drm/radeon/radeon_irq.c | 401 ++ drivers/gpu/drm/radeon/radeon_irq_kms.c | 264 + drivers/gpu/drm/radeon/radeon_kms.c | 442 ++ drivers/gpu/drm/radeon/radeon_legacy_crtc.c | 1075 +++ drivers/gpu/drm/radeon/radeon_legacy_encoders.c | 1666 +++++ drivers/gpu/drm/radeon/radeon_legacy_tv.c | 923 +++ drivers/gpu/drm/radeon/radeon_mem.c | 302 + drivers/gpu/drm/radeon/radeon_mode.h | 700 ++ drivers/gpu/drm/radeon/radeon_object.c | 533 ++ drivers/gpu/drm/radeon/radeon_object.h | 169 + drivers/gpu/drm/radeon/radeon_pm.c | 868 +++ drivers/gpu/drm/radeon/radeon_reg.h | 3706 ++++++++++ drivers/gpu/drm/radeon/radeon_ring.c | 465 ++ drivers/gpu/drm/radeon/radeon_state.c | 3261 +++++++++ drivers/gpu/drm/radeon/radeon_test.c | 218 + drivers/gpu/drm/radeon/radeon_trace.h | 82 + drivers/gpu/drm/radeon/radeon_trace_points.c | 9 + drivers/gpu/drm/radeon/radeon_ttm.c | 828 +++ drivers/gpu/drm/radeon/reg_srcs/cayman | 620 ++ drivers/gpu/drm/radeon/reg_srcs/evergreen | 622 ++ drivers/gpu/drm/radeon/reg_srcs/r100 | 105 + drivers/gpu/drm/radeon/reg_srcs/r200 | 186 + drivers/gpu/drm/radeon/reg_srcs/r300 | 714 ++ drivers/gpu/drm/radeon/reg_srcs/r420 | 780 +++ drivers/gpu/drm/radeon/reg_srcs/r600 | 762 +++ drivers/gpu/drm/radeon/reg_srcs/rn50 | 30 + drivers/gpu/drm/radeon/reg_srcs/rs600 | 780 +++ drivers/gpu/drm/radeon/reg_srcs/rv515 | 494 ++ drivers/gpu/drm/radeon/rs100d.h | 40 + drivers/gpu/drm/radeon/rs400.c | 542 ++ drivers/gpu/drm/radeon/rs400d.h | 160 + drivers/gpu/drm/radeon/rs600.c | 995 +++ drivers/gpu/drm/radeon/rs600d.h | 671 ++ drivers/gpu/drm/radeon/rs690.c | 765 +++ drivers/gpu/drm/radeon/rs690d.h | 310 + drivers/gpu/drm/radeon/rv200d.h | 36 + drivers/gpu/drm/radeon/rv250d.h | 123 + drivers/gpu/drm/radeon/rv350d.h | 52 + drivers/gpu/drm/radeon/rv515.c | 1104 +++ drivers/gpu/drm/radeon/rv515d.h | 649 ++ drivers/gpu/drm/radeon/rv770.c | 1413 ++++ drivers/gpu/drm/radeon/rv770d.h | 402 ++ drivers/gpu/drm/savage/Makefile | 9 + drivers/gpu/drm/savage/savage_bci.c | 1088 +++ drivers/gpu/drm/savage/savage_drv.c | 87 + drivers/gpu/drm/savage/savage_drv.h | 575 ++ drivers/gpu/drm/savage/savage_state.c | 1162 ++++ drivers/gpu/drm/sis/Makefile | 10 + drivers/gpu/drm/sis/sis_drv.c | 115 + drivers/gpu/drm/sis/sis_drv.h | 73 + drivers/gpu/drm/sis/sis_mm.c | 331 + drivers/gpu/drm/tdfx/Makefile | 8 + drivers/gpu/drm/tdfx/tdfx_drv.c | 84 + drivers/gpu/drm/tdfx/tdfx_drv.h | 47 + drivers/gpu/drm/ttm/Makefile | 10 + drivers/gpu/drm/ttm/ttm_agp_backend.c | 152 + drivers/gpu/drm/ttm/ttm_bo.c | 1892 ++++++ drivers/gpu/drm/ttm/ttm_bo_manager.c | 157 + drivers/gpu/drm/ttm/ttm_bo_util.c | 683 ++ drivers/gpu/drm/ttm/ttm_bo_vm.c | 459 ++ drivers/gpu/drm/ttm/ttm_execbuf_util.c | 238 + drivers/gpu/drm/ttm/ttm_lock.c | 310 + drivers/gpu/drm/ttm/ttm_memory.c | 600 ++ drivers/gpu/drm/ttm/ttm_module.c | 102 + drivers/gpu/drm/ttm/ttm_object.c | 452 ++ drivers/gpu/drm/ttm/ttm_page_alloc.c | 859 +++ drivers/gpu/drm/ttm/ttm_tt.c | 590 ++ drivers/gpu/drm/via/Makefile | 8 + drivers/gpu/drm/via/via_3d_reg.h | 1650 +++++ drivers/gpu/drm/via/via_dma.c | 741 ++ drivers/gpu/drm/via/via_dmablit.c | 808 +++ drivers/gpu/drm/via/via_dmablit.h | 140 + drivers/gpu/drm/via/via_drv.c | 96 + drivers/gpu/drm/via/via_drv.h | 157 + drivers/gpu/drm/via/via_irq.c | 392 ++ drivers/gpu/drm/via/via_map.c | 132 + drivers/gpu/drm/via/via_mm.c | 192 + drivers/gpu/drm/via/via_verifier.c | 1111 +++ drivers/gpu/drm/via/via_verifier.h | 62 + drivers/gpu/drm/via/via_video.c | 93 + drivers/gpu/drm/vivante/Makefile | 29 + drivers/gpu/drm/vivante/vivante_drv.c | 111 + drivers/gpu/drm/vivante/vivante_drv.h | 66 + drivers/gpu/drm/vmwgfx/Kconfig | 13 + drivers/gpu/drm/vmwgfx/Makefile | 9 + drivers/gpu/drm/vmwgfx/svga3d_reg.h | 1793 +++++ drivers/gpu/drm/vmwgfx/svga_escape.h | 89 + drivers/gpu/drm/vmwgfx/svga_overlay.h | 201 + drivers/gpu/drm/vmwgfx/svga_reg.h | 1346 ++++ drivers/gpu/drm/vmwgfx/svga_types.h | 45 + drivers/gpu/drm/vmwgfx/vmwgfx_buffer.c | 329 + drivers/gpu/drm/vmwgfx/vmwgfx_drv.c | 958 +++ drivers/gpu/drm/vmwgfx/vmwgfx_drv.h | 579 ++ drivers/gpu/drm/vmwgfx/vmwgfx_execbuf.c | 719 ++ drivers/gpu/drm/vmwgfx/vmwgfx_fb.c | 717 ++ drivers/gpu/drm/vmwgfx/vmwgfx_fence.c | 173 + drivers/gpu/drm/vmwgfx/vmwgfx_fifo.c | 562 ++ drivers/gpu/drm/vmwgfx/vmwgfx_gmr.c | 201 + drivers/gpu/drm/vmwgfx/vmwgfx_gmrid_manager.c | 137 + drivers/gpu/drm/vmwgfx/vmwgfx_ioctl.c | 90 + drivers/gpu/drm/vmwgfx/vmwgfx_irq.c | 297 + drivers/gpu/drm/vmwgfx/vmwgfx_kms.c | 1154 ++++ drivers/gpu/drm/vmwgfx/vmwgfx_kms.h | 104 + drivers/gpu/drm/vmwgfx/vmwgfx_ldu.c | 638 ++ drivers/gpu/drm/vmwgfx/vmwgfx_overlay.c | 627 ++ drivers/gpu/drm/vmwgfx/vmwgfx_reg.h | 57 + drivers/gpu/drm/vmwgfx/vmwgfx_resource.c | 1122 +++ drivers/gpu/drm/vmwgfx/vmwgfx_ttm_glue.c | 99 + drivers/gpu/ion/Kconfig | 18 + drivers/gpu/ion/Makefile | 3 + drivers/gpu/ion/ion.c | 1205 ++++ drivers/gpu/ion/ion_carveout_heap.c | 162 + drivers/gpu/ion/ion_heap.c | 72 + drivers/gpu/ion/ion_priv.h | 184 + drivers/gpu/ion/ion_system_heap.c | 198 + drivers/gpu/ion/ion_system_mapper.c | 114 + drivers/gpu/ion/mxc/Makefile | 1 + drivers/gpu/ion/mxc/mxc_ion.c | 101 + drivers/gpu/ion/tegra/Makefile | 1 + drivers/gpu/ion/tegra/tegra_ion.c | 96 + drivers/gpu/stub/Kconfig | 18 + drivers/gpu/stub/Makefile | 1 + drivers/gpu/stub/poulsbo.c | 64 + drivers/gpu/vga/Kconfig | 29 + drivers/gpu/vga/Makefile | 2 + drivers/gpu/vga/vga_switcheroo.c | 492 ++ drivers/gpu/vga/vgaarb.c | 1328 ++++ 436 files changed, 273895 insertions(+) create mode 100644 drivers/gpu/Makefile create mode 100644 drivers/gpu/drm/Kconfig create mode 100644 drivers/gpu/drm/Makefile create mode 100644 drivers/gpu/drm/README.drm create mode 100644 drivers/gpu/drm/ati_pcigart.c create mode 100644 drivers/gpu/drm/drm_agpsupport.c create mode 100644 drivers/gpu/drm/drm_auth.c create mode 100644 drivers/gpu/drm/drm_buffer.c create mode 100644 drivers/gpu/drm/drm_bufs.c create mode 100644 drivers/gpu/drm/drm_cache.c create mode 100644 drivers/gpu/drm/drm_context.c create mode 100644 drivers/gpu/drm/drm_crtc.c create mode 100644 drivers/gpu/drm/drm_crtc_helper.c create mode 100644 drivers/gpu/drm/drm_debugfs.c create mode 100644 drivers/gpu/drm/drm_dma.c create mode 100644 drivers/gpu/drm/drm_dp_i2c_helper.c create mode 100644 drivers/gpu/drm/drm_drv.c create mode 100644 drivers/gpu/drm/drm_edid.c create mode 100644 drivers/gpu/drm/drm_edid_modes.h create mode 100644 drivers/gpu/drm/drm_encoder_slave.c create mode 100644 drivers/gpu/drm/drm_fb_helper.c create mode 100644 drivers/gpu/drm/drm_fops.c create mode 100644 drivers/gpu/drm/drm_gem.c create mode 100644 drivers/gpu/drm/drm_global.c create mode 100644 drivers/gpu/drm/drm_hashtab.c create mode 100644 drivers/gpu/drm/drm_info.c create mode 100644 drivers/gpu/drm/drm_ioc32.c create mode 100644 drivers/gpu/drm/drm_ioctl.c create mode 100644 drivers/gpu/drm/drm_irq.c create mode 100644 drivers/gpu/drm/drm_lock.c create mode 100644 drivers/gpu/drm/drm_memory.c create mode 100644 drivers/gpu/drm/drm_mm.c create mode 100644 drivers/gpu/drm/drm_modes.c create mode 100644 drivers/gpu/drm/drm_pci.c create mode 100644 drivers/gpu/drm/drm_platform.c create mode 100644 drivers/gpu/drm/drm_proc.c create mode 100644 drivers/gpu/drm/drm_scatter.c create mode 100644 drivers/gpu/drm/drm_sman.c create mode 100644 drivers/gpu/drm/drm_stub.c create mode 100644 drivers/gpu/drm/drm_sysfs.c create mode 100644 drivers/gpu/drm/drm_trace.h create mode 100644 drivers/gpu/drm/drm_trace_points.c create mode 100644 drivers/gpu/drm/drm_usb.c create mode 100644 drivers/gpu/drm/drm_vm.c create mode 100644 drivers/gpu/drm/i2c/Makefile create mode 100644 drivers/gpu/drm/i2c/ch7006_drv.c create mode 100644 drivers/gpu/drm/i2c/ch7006_mode.c create mode 100644 drivers/gpu/drm/i2c/ch7006_priv.h create mode 100644 drivers/gpu/drm/i2c/sil164_drv.c create mode 100644 drivers/gpu/drm/i810/Makefile create mode 100644 drivers/gpu/drm/i810/i810_dma.c create mode 100644 drivers/gpu/drm/i810/i810_drv.c create mode 100644 drivers/gpu/drm/i810/i810_drv.h create mode 100644 drivers/gpu/drm/i915/Makefile create mode 100644 drivers/gpu/drm/i915/dvo.h create mode 100644 drivers/gpu/drm/i915/dvo_ch7017.c create mode 100644 drivers/gpu/drm/i915/dvo_ch7xxx.c create mode 100644 drivers/gpu/drm/i915/dvo_ivch.c create mode 100644 drivers/gpu/drm/i915/dvo_sil164.c create mode 100644 drivers/gpu/drm/i915/dvo_tfp410.c create mode 100644 drivers/gpu/drm/i915/i915_debugfs.c create mode 100644 drivers/gpu/drm/i915/i915_dma.c create mode 100644 drivers/gpu/drm/i915/i915_drv.c create mode 100644 drivers/gpu/drm/i915/i915_drv.h create mode 100644 drivers/gpu/drm/i915/i915_gem.c create mode 100644 drivers/gpu/drm/i915/i915_gem_debug.c create mode 100644 drivers/gpu/drm/i915/i915_gem_evict.c create mode 100644 drivers/gpu/drm/i915/i915_gem_execbuffer.c create mode 100644 drivers/gpu/drm/i915/i915_gem_gtt.c create mode 100644 drivers/gpu/drm/i915/i915_gem_tiling.c create mode 100644 drivers/gpu/drm/i915/i915_ioc32.c create mode 100644 drivers/gpu/drm/i915/i915_irq.c create mode 100644 drivers/gpu/drm/i915/i915_mem.c create mode 100644 drivers/gpu/drm/i915/i915_reg.h create mode 100644 drivers/gpu/drm/i915/i915_suspend.c create mode 100644 drivers/gpu/drm/i915/i915_trace.h create mode 100644 drivers/gpu/drm/i915/i915_trace_points.c create mode 100644 drivers/gpu/drm/i915/intel_acpi.c create mode 100644 drivers/gpu/drm/i915/intel_bios.c create mode 100644 drivers/gpu/drm/i915/intel_bios.h create mode 100644 drivers/gpu/drm/i915/intel_crt.c create mode 100644 drivers/gpu/drm/i915/intel_display.c create mode 100644 drivers/gpu/drm/i915/intel_dp.c create mode 100644 drivers/gpu/drm/i915/intel_drv.h create mode 100644 drivers/gpu/drm/i915/intel_dvo.c create mode 100644 drivers/gpu/drm/i915/intel_fb.c create mode 100644 drivers/gpu/drm/i915/intel_hdmi.c create mode 100644 drivers/gpu/drm/i915/intel_i2c.c create mode 100644 drivers/gpu/drm/i915/intel_lvds.c create mode 100644 drivers/gpu/drm/i915/intel_modes.c create mode 100644 drivers/gpu/drm/i915/intel_opregion.c create mode 100644 drivers/gpu/drm/i915/intel_overlay.c create mode 100644 drivers/gpu/drm/i915/intel_panel.c create mode 100644 drivers/gpu/drm/i915/intel_ringbuffer.c create mode 100644 drivers/gpu/drm/i915/intel_ringbuffer.h create mode 100644 drivers/gpu/drm/i915/intel_sdvo.c create mode 100644 drivers/gpu/drm/i915/intel_sdvo_regs.h create mode 100644 drivers/gpu/drm/i915/intel_tv.c create mode 100644 drivers/gpu/drm/mga/Makefile create mode 100644 drivers/gpu/drm/mga/mga_dma.c create mode 100644 drivers/gpu/drm/mga/mga_drv.c create mode 100644 drivers/gpu/drm/mga/mga_drv.h create mode 100644 drivers/gpu/drm/mga/mga_ioc32.c create mode 100644 drivers/gpu/drm/mga/mga_irq.c create mode 100644 drivers/gpu/drm/mga/mga_state.c create mode 100644 drivers/gpu/drm/mga/mga_warp.c create mode 100644 drivers/gpu/drm/nouveau/Kconfig create mode 100644 drivers/gpu/drm/nouveau/Makefile create mode 100644 drivers/gpu/drm/nouveau/nouveau_acpi.c create mode 100644 drivers/gpu/drm/nouveau/nouveau_backlight.c create mode 100644 drivers/gpu/drm/nouveau/nouveau_bios.c create mode 100644 drivers/gpu/drm/nouveau/nouveau_bios.h create mode 100644 drivers/gpu/drm/nouveau/nouveau_bo.c create mode 100644 drivers/gpu/drm/nouveau/nouveau_calc.c create mode 100644 drivers/gpu/drm/nouveau/nouveau_channel.c create mode 100644 drivers/gpu/drm/nouveau/nouveau_connector.c create mode 100644 drivers/gpu/drm/nouveau/nouveau_connector.h create mode 100644 drivers/gpu/drm/nouveau/nouveau_crtc.h create mode 100644 drivers/gpu/drm/nouveau/nouveau_debugfs.c create mode 100644 drivers/gpu/drm/nouveau/nouveau_display.c create mode 100644 drivers/gpu/drm/nouveau/nouveau_dma.c create mode 100644 drivers/gpu/drm/nouveau/nouveau_dma.h create mode 100644 drivers/gpu/drm/nouveau/nouveau_dp.c create mode 100644 drivers/gpu/drm/nouveau/nouveau_drv.c create mode 100644 drivers/gpu/drm/nouveau/nouveau_drv.h create mode 100644 drivers/gpu/drm/nouveau/nouveau_encoder.h create mode 100644 drivers/gpu/drm/nouveau/nouveau_fb.h create mode 100644 drivers/gpu/drm/nouveau/nouveau_fbcon.c create mode 100644 drivers/gpu/drm/nouveau/nouveau_fbcon.h create mode 100644 drivers/gpu/drm/nouveau/nouveau_fence.c create mode 100644 drivers/gpu/drm/nouveau/nouveau_gem.c create mode 100644 drivers/gpu/drm/nouveau/nouveau_grctx.h create mode 100644 drivers/gpu/drm/nouveau/nouveau_hw.c create mode 100644 drivers/gpu/drm/nouveau/nouveau_hw.h create mode 100644 drivers/gpu/drm/nouveau/nouveau_i2c.c create mode 100644 drivers/gpu/drm/nouveau/nouveau_i2c.h create mode 100644 drivers/gpu/drm/nouveau/nouveau_ioc32.c create mode 100644 drivers/gpu/drm/nouveau/nouveau_irq.c create mode 100644 drivers/gpu/drm/nouveau/nouveau_mem.c create mode 100644 drivers/gpu/drm/nouveau/nouveau_mm.c create mode 100644 drivers/gpu/drm/nouveau/nouveau_mm.h create mode 100644 drivers/gpu/drm/nouveau/nouveau_notifier.c create mode 100644 drivers/gpu/drm/nouveau/nouveau_object.c create mode 100644 drivers/gpu/drm/nouveau/nouveau_perf.c create mode 100644 drivers/gpu/drm/nouveau/nouveau_pm.c create mode 100644 drivers/gpu/drm/nouveau/nouveau_pm.h create mode 100644 drivers/gpu/drm/nouveau/nouveau_ramht.c create mode 100644 drivers/gpu/drm/nouveau/nouveau_ramht.h create mode 100644 drivers/gpu/drm/nouveau/nouveau_reg.h create mode 100644 drivers/gpu/drm/nouveau/nouveau_sgdma.c create mode 100644 drivers/gpu/drm/nouveau/nouveau_state.c create mode 100644 drivers/gpu/drm/nouveau/nouveau_temp.c create mode 100644 drivers/gpu/drm/nouveau/nouveau_ttm.c create mode 100644 drivers/gpu/drm/nouveau/nouveau_util.c create mode 100644 drivers/gpu/drm/nouveau/nouveau_util.h create mode 100644 drivers/gpu/drm/nouveau/nouveau_vm.c create mode 100644 drivers/gpu/drm/nouveau/nouveau_vm.h create mode 100644 drivers/gpu/drm/nouveau/nouveau_volt.c create mode 100644 drivers/gpu/drm/nouveau/nv04_crtc.c create mode 100644 drivers/gpu/drm/nouveau/nv04_cursor.c create mode 100644 drivers/gpu/drm/nouveau/nv04_dac.c create mode 100644 drivers/gpu/drm/nouveau/nv04_dfp.c create mode 100644 drivers/gpu/drm/nouveau/nv04_display.c create mode 100644 drivers/gpu/drm/nouveau/nv04_fb.c create mode 100644 drivers/gpu/drm/nouveau/nv04_fbcon.c create mode 100644 drivers/gpu/drm/nouveau/nv04_fifo.c create mode 100644 drivers/gpu/drm/nouveau/nv04_graph.c create mode 100644 drivers/gpu/drm/nouveau/nv04_instmem.c create mode 100644 drivers/gpu/drm/nouveau/nv04_mc.c create mode 100644 drivers/gpu/drm/nouveau/nv04_pm.c create mode 100644 drivers/gpu/drm/nouveau/nv04_timer.c create mode 100644 drivers/gpu/drm/nouveau/nv04_tv.c create mode 100644 drivers/gpu/drm/nouveau/nv10_fb.c create mode 100644 drivers/gpu/drm/nouveau/nv10_fifo.c create mode 100644 drivers/gpu/drm/nouveau/nv10_gpio.c create mode 100644 drivers/gpu/drm/nouveau/nv10_graph.c create mode 100644 drivers/gpu/drm/nouveau/nv17_tv.c create mode 100644 drivers/gpu/drm/nouveau/nv17_tv.h create mode 100644 drivers/gpu/drm/nouveau/nv17_tv_modes.c create mode 100644 drivers/gpu/drm/nouveau/nv20_graph.c create mode 100644 drivers/gpu/drm/nouveau/nv30_fb.c create mode 100644 drivers/gpu/drm/nouveau/nv40_fb.c create mode 100644 drivers/gpu/drm/nouveau/nv40_fifo.c create mode 100644 drivers/gpu/drm/nouveau/nv40_graph.c create mode 100644 drivers/gpu/drm/nouveau/nv40_grctx.c create mode 100644 drivers/gpu/drm/nouveau/nv40_mc.c create mode 100644 drivers/gpu/drm/nouveau/nv40_mpeg.c create mode 100644 drivers/gpu/drm/nouveau/nv50_calc.c create mode 100644 drivers/gpu/drm/nouveau/nv50_crtc.c create mode 100644 drivers/gpu/drm/nouveau/nv50_cursor.c create mode 100644 drivers/gpu/drm/nouveau/nv50_dac.c create mode 100644 drivers/gpu/drm/nouveau/nv50_display.c create mode 100644 drivers/gpu/drm/nouveau/nv50_display.h create mode 100644 drivers/gpu/drm/nouveau/nv50_evo.c create mode 100644 drivers/gpu/drm/nouveau/nv50_evo.h create mode 100644 drivers/gpu/drm/nouveau/nv50_fb.c create mode 100644 drivers/gpu/drm/nouveau/nv50_fbcon.c create mode 100644 drivers/gpu/drm/nouveau/nv50_fifo.c create mode 100644 drivers/gpu/drm/nouveau/nv50_gpio.c create mode 100644 drivers/gpu/drm/nouveau/nv50_graph.c create mode 100644 drivers/gpu/drm/nouveau/nv50_grctx.c create mode 100644 drivers/gpu/drm/nouveau/nv50_instmem.c create mode 100644 drivers/gpu/drm/nouveau/nv50_mc.c create mode 100644 drivers/gpu/drm/nouveau/nv50_mpeg.c create mode 100644 drivers/gpu/drm/nouveau/nv50_pm.c create mode 100644 drivers/gpu/drm/nouveau/nv50_sor.c create mode 100644 drivers/gpu/drm/nouveau/nv50_vm.c create mode 100644 drivers/gpu/drm/nouveau/nv50_vram.c create mode 100644 drivers/gpu/drm/nouveau/nv84_crypt.c create mode 100644 drivers/gpu/drm/nouveau/nva3_copy.c create mode 100644 drivers/gpu/drm/nouveau/nva3_copy.fuc create mode 100644 drivers/gpu/drm/nouveau/nva3_copy.fuc.h create mode 100644 drivers/gpu/drm/nouveau/nva3_pm.c create mode 100644 drivers/gpu/drm/nouveau/nvc0_copy.c create mode 100644 drivers/gpu/drm/nouveau/nvc0_copy.fuc.h create mode 100644 drivers/gpu/drm/nouveau/nvc0_fb.c create mode 100644 drivers/gpu/drm/nouveau/nvc0_fbcon.c create mode 100644 drivers/gpu/drm/nouveau/nvc0_fifo.c create mode 100644 drivers/gpu/drm/nouveau/nvc0_graph.c create mode 100644 drivers/gpu/drm/nouveau/nvc0_graph.h create mode 100644 drivers/gpu/drm/nouveau/nvc0_grctx.c create mode 100644 drivers/gpu/drm/nouveau/nvc0_instmem.c create mode 100644 drivers/gpu/drm/nouveau/nvc0_vm.c create mode 100644 drivers/gpu/drm/nouveau/nvc0_vram.c create mode 100644 drivers/gpu/drm/nouveau/nvreg.h create mode 100644 drivers/gpu/drm/r128/Makefile create mode 100644 drivers/gpu/drm/r128/r128_cce.c create mode 100644 drivers/gpu/drm/r128/r128_drv.c create mode 100644 drivers/gpu/drm/r128/r128_drv.h create mode 100644 drivers/gpu/drm/r128/r128_ioc32.c create mode 100644 drivers/gpu/drm/r128/r128_irq.c create mode 100644 drivers/gpu/drm/r128/r128_state.c create mode 100644 drivers/gpu/drm/radeon/.gitignore create mode 100644 drivers/gpu/drm/radeon/Kconfig create mode 100644 drivers/gpu/drm/radeon/Makefile create mode 100644 drivers/gpu/drm/radeon/ObjectID.h create mode 100644 drivers/gpu/drm/radeon/atom-bits.h create mode 100644 drivers/gpu/drm/radeon/atom-names.h create mode 100644 drivers/gpu/drm/radeon/atom-types.h create mode 100644 drivers/gpu/drm/radeon/atom.c create mode 100644 drivers/gpu/drm/radeon/atom.h create mode 100644 drivers/gpu/drm/radeon/atombios.h create mode 100644 drivers/gpu/drm/radeon/atombios_crtc.c create mode 100644 drivers/gpu/drm/radeon/atombios_dp.c create mode 100644 drivers/gpu/drm/radeon/avivod.h create mode 100644 drivers/gpu/drm/radeon/cayman_blit_shaders.c create mode 100644 drivers/gpu/drm/radeon/cayman_blit_shaders.h create mode 100644 drivers/gpu/drm/radeon/evergreen.c create mode 100644 drivers/gpu/drm/radeon/evergreen_blit_kms.c create mode 100644 drivers/gpu/drm/radeon/evergreen_blit_shaders.c create mode 100644 drivers/gpu/drm/radeon/evergreen_blit_shaders.h create mode 100644 drivers/gpu/drm/radeon/evergreen_cs.c create mode 100644 drivers/gpu/drm/radeon/evergreen_reg.h create mode 100644 drivers/gpu/drm/radeon/evergreend.h create mode 100644 drivers/gpu/drm/radeon/mkregtable.c create mode 100644 drivers/gpu/drm/radeon/ni.c create mode 100644 drivers/gpu/drm/radeon/ni_reg.h create mode 100644 drivers/gpu/drm/radeon/nid.h create mode 100644 drivers/gpu/drm/radeon/r100.c create mode 100644 drivers/gpu/drm/radeon/r100_track.h create mode 100644 drivers/gpu/drm/radeon/r100d.h create mode 100644 drivers/gpu/drm/radeon/r200.c create mode 100644 drivers/gpu/drm/radeon/r300.c create mode 100644 drivers/gpu/drm/radeon/r300_cmdbuf.c create mode 100644 drivers/gpu/drm/radeon/r300_reg.h create mode 100644 drivers/gpu/drm/radeon/r300d.h create mode 100644 drivers/gpu/drm/radeon/r420.c create mode 100644 drivers/gpu/drm/radeon/r420d.h create mode 100644 drivers/gpu/drm/radeon/r500_reg.h create mode 100644 drivers/gpu/drm/radeon/r520.c create mode 100644 drivers/gpu/drm/radeon/r520d.h create mode 100644 drivers/gpu/drm/radeon/r600.c create mode 100644 drivers/gpu/drm/radeon/r600_audio.c create mode 100644 drivers/gpu/drm/radeon/r600_blit.c create mode 100644 drivers/gpu/drm/radeon/r600_blit_kms.c create mode 100644 drivers/gpu/drm/radeon/r600_blit_shaders.c create mode 100644 drivers/gpu/drm/radeon/r600_blit_shaders.h create mode 100644 drivers/gpu/drm/radeon/r600_cp.c create mode 100644 drivers/gpu/drm/radeon/r600_cs.c create mode 100644 drivers/gpu/drm/radeon/r600_hdmi.c create mode 100644 drivers/gpu/drm/radeon/r600_reg.h create mode 100644 drivers/gpu/drm/radeon/r600d.h create mode 100644 drivers/gpu/drm/radeon/radeon.h create mode 100644 drivers/gpu/drm/radeon/radeon_acpi.c create mode 100644 drivers/gpu/drm/radeon/radeon_agp.c create mode 100644 drivers/gpu/drm/radeon/radeon_asic.c create mode 100644 drivers/gpu/drm/radeon/radeon_asic.h create mode 100644 drivers/gpu/drm/radeon/radeon_atombios.c create mode 100644 drivers/gpu/drm/radeon/radeon_atpx_handler.c create mode 100644 drivers/gpu/drm/radeon/radeon_benchmark.c create mode 100644 drivers/gpu/drm/radeon/radeon_bios.c create mode 100644 drivers/gpu/drm/radeon/radeon_clocks.c create mode 100644 drivers/gpu/drm/radeon/radeon_combios.c create mode 100644 drivers/gpu/drm/radeon/radeon_connectors.c create mode 100644 drivers/gpu/drm/radeon/radeon_cp.c create mode 100644 drivers/gpu/drm/radeon/radeon_cs.c create mode 100644 drivers/gpu/drm/radeon/radeon_cursor.c create mode 100644 drivers/gpu/drm/radeon/radeon_device.c create mode 100644 drivers/gpu/drm/radeon/radeon_display.c create mode 100644 drivers/gpu/drm/radeon/radeon_drv.c create mode 100644 drivers/gpu/drm/radeon/radeon_drv.h create mode 100644 drivers/gpu/drm/radeon/radeon_encoders.c create mode 100644 drivers/gpu/drm/radeon/radeon_family.h create mode 100644 drivers/gpu/drm/radeon/radeon_fb.c create mode 100644 drivers/gpu/drm/radeon/radeon_fence.c create mode 100644 drivers/gpu/drm/radeon/radeon_gart.c create mode 100644 drivers/gpu/drm/radeon/radeon_gem.c create mode 100644 drivers/gpu/drm/radeon/radeon_i2c.c create mode 100644 drivers/gpu/drm/radeon/radeon_ioc32.c create mode 100644 drivers/gpu/drm/radeon/radeon_irq.c create mode 100644 drivers/gpu/drm/radeon/radeon_irq_kms.c create mode 100644 drivers/gpu/drm/radeon/radeon_kms.c create mode 100644 drivers/gpu/drm/radeon/radeon_legacy_crtc.c create mode 100644 drivers/gpu/drm/radeon/radeon_legacy_encoders.c create mode 100644 drivers/gpu/drm/radeon/radeon_legacy_tv.c create mode 100644 drivers/gpu/drm/radeon/radeon_mem.c create mode 100644 drivers/gpu/drm/radeon/radeon_mode.h create mode 100644 drivers/gpu/drm/radeon/radeon_object.c create mode 100644 drivers/gpu/drm/radeon/radeon_object.h create mode 100644 drivers/gpu/drm/radeon/radeon_pm.c create mode 100644 drivers/gpu/drm/radeon/radeon_reg.h create mode 100644 drivers/gpu/drm/radeon/radeon_ring.c create mode 100644 drivers/gpu/drm/radeon/radeon_state.c create mode 100644 drivers/gpu/drm/radeon/radeon_test.c create mode 100644 drivers/gpu/drm/radeon/radeon_trace.h create mode 100644 drivers/gpu/drm/radeon/radeon_trace_points.c create mode 100644 drivers/gpu/drm/radeon/radeon_ttm.c create mode 100644 drivers/gpu/drm/radeon/reg_srcs/cayman create mode 100644 drivers/gpu/drm/radeon/reg_srcs/evergreen create mode 100644 drivers/gpu/drm/radeon/reg_srcs/r100 create mode 100644 drivers/gpu/drm/radeon/reg_srcs/r200 create mode 100644 drivers/gpu/drm/radeon/reg_srcs/r300 create mode 100644 drivers/gpu/drm/radeon/reg_srcs/r420 create mode 100644 drivers/gpu/drm/radeon/reg_srcs/r600 create mode 100644 drivers/gpu/drm/radeon/reg_srcs/rn50 create mode 100644 drivers/gpu/drm/radeon/reg_srcs/rs600 create mode 100644 drivers/gpu/drm/radeon/reg_srcs/rv515 create mode 100644 drivers/gpu/drm/radeon/rs100d.h create mode 100644 drivers/gpu/drm/radeon/rs400.c create mode 100644 drivers/gpu/drm/radeon/rs400d.h create mode 100644 drivers/gpu/drm/radeon/rs600.c create mode 100644 drivers/gpu/drm/radeon/rs600d.h create mode 100644 drivers/gpu/drm/radeon/rs690.c create mode 100644 drivers/gpu/drm/radeon/rs690d.h create mode 100644 drivers/gpu/drm/radeon/rv200d.h create mode 100644 drivers/gpu/drm/radeon/rv250d.h create mode 100644 drivers/gpu/drm/radeon/rv350d.h create mode 100644 drivers/gpu/drm/radeon/rv515.c create mode 100644 drivers/gpu/drm/radeon/rv515d.h create mode 100644 drivers/gpu/drm/radeon/rv770.c create mode 100644 drivers/gpu/drm/radeon/rv770d.h create mode 100644 drivers/gpu/drm/savage/Makefile create mode 100644 drivers/gpu/drm/savage/savage_bci.c create mode 100644 drivers/gpu/drm/savage/savage_drv.c create mode 100644 drivers/gpu/drm/savage/savage_drv.h create mode 100644 drivers/gpu/drm/savage/savage_state.c create mode 100644 drivers/gpu/drm/sis/Makefile create mode 100644 drivers/gpu/drm/sis/sis_drv.c create mode 100644 drivers/gpu/drm/sis/sis_drv.h create mode 100644 drivers/gpu/drm/sis/sis_mm.c create mode 100644 drivers/gpu/drm/tdfx/Makefile create mode 100644 drivers/gpu/drm/tdfx/tdfx_drv.c create mode 100644 drivers/gpu/drm/tdfx/tdfx_drv.h create mode 100644 drivers/gpu/drm/ttm/Makefile create mode 100644 drivers/gpu/drm/ttm/ttm_agp_backend.c create mode 100644 drivers/gpu/drm/ttm/ttm_bo.c create mode 100644 drivers/gpu/drm/ttm/ttm_bo_manager.c create mode 100644 drivers/gpu/drm/ttm/ttm_bo_util.c create mode 100644 drivers/gpu/drm/ttm/ttm_bo_vm.c create mode 100644 drivers/gpu/drm/ttm/ttm_execbuf_util.c create mode 100644 drivers/gpu/drm/ttm/ttm_lock.c create mode 100644 drivers/gpu/drm/ttm/ttm_memory.c create mode 100644 drivers/gpu/drm/ttm/ttm_module.c create mode 100644 drivers/gpu/drm/ttm/ttm_object.c create mode 100644 drivers/gpu/drm/ttm/ttm_page_alloc.c create mode 100644 drivers/gpu/drm/ttm/ttm_tt.c create mode 100644 drivers/gpu/drm/via/Makefile create mode 100644 drivers/gpu/drm/via/via_3d_reg.h create mode 100644 drivers/gpu/drm/via/via_dma.c create mode 100644 drivers/gpu/drm/via/via_dmablit.c create mode 100644 drivers/gpu/drm/via/via_dmablit.h create mode 100644 drivers/gpu/drm/via/via_drv.c create mode 100644 drivers/gpu/drm/via/via_drv.h create mode 100644 drivers/gpu/drm/via/via_irq.c create mode 100644 drivers/gpu/drm/via/via_map.c create mode 100644 drivers/gpu/drm/via/via_mm.c create mode 100644 drivers/gpu/drm/via/via_verifier.c create mode 100644 drivers/gpu/drm/via/via_verifier.h create mode 100644 drivers/gpu/drm/via/via_video.c create mode 100644 drivers/gpu/drm/vivante/Makefile create mode 100644 drivers/gpu/drm/vivante/vivante_drv.c create mode 100644 drivers/gpu/drm/vivante/vivante_drv.h create mode 100644 drivers/gpu/drm/vmwgfx/Kconfig create mode 100644 drivers/gpu/drm/vmwgfx/Makefile create mode 100644 drivers/gpu/drm/vmwgfx/svga3d_reg.h create mode 100644 drivers/gpu/drm/vmwgfx/svga_escape.h create mode 100644 drivers/gpu/drm/vmwgfx/svga_overlay.h create mode 100644 drivers/gpu/drm/vmwgfx/svga_reg.h create mode 100644 drivers/gpu/drm/vmwgfx/svga_types.h create mode 100644 drivers/gpu/drm/vmwgfx/vmwgfx_buffer.c create mode 100644 drivers/gpu/drm/vmwgfx/vmwgfx_drv.c create mode 100644 drivers/gpu/drm/vmwgfx/vmwgfx_drv.h create mode 100644 drivers/gpu/drm/vmwgfx/vmwgfx_execbuf.c create mode 100644 drivers/gpu/drm/vmwgfx/vmwgfx_fb.c create mode 100644 drivers/gpu/drm/vmwgfx/vmwgfx_fence.c create mode 100644 drivers/gpu/drm/vmwgfx/vmwgfx_fifo.c create mode 100644 drivers/gpu/drm/vmwgfx/vmwgfx_gmr.c create mode 100644 drivers/gpu/drm/vmwgfx/vmwgfx_gmrid_manager.c create mode 100644 drivers/gpu/drm/vmwgfx/vmwgfx_ioctl.c create mode 100644 drivers/gpu/drm/vmwgfx/vmwgfx_irq.c create mode 100644 drivers/gpu/drm/vmwgfx/vmwgfx_kms.c create mode 100644 drivers/gpu/drm/vmwgfx/vmwgfx_kms.h create mode 100644 drivers/gpu/drm/vmwgfx/vmwgfx_ldu.c create mode 100644 drivers/gpu/drm/vmwgfx/vmwgfx_overlay.c create mode 100644 drivers/gpu/drm/vmwgfx/vmwgfx_reg.h create mode 100644 drivers/gpu/drm/vmwgfx/vmwgfx_resource.c create mode 100644 drivers/gpu/drm/vmwgfx/vmwgfx_ttm_glue.c create mode 100644 drivers/gpu/ion/Kconfig create mode 100644 drivers/gpu/ion/Makefile create mode 100644 drivers/gpu/ion/ion.c create mode 100644 drivers/gpu/ion/ion_carveout_heap.c create mode 100644 drivers/gpu/ion/ion_heap.c create mode 100644 drivers/gpu/ion/ion_priv.h create mode 100644 drivers/gpu/ion/ion_system_heap.c create mode 100644 drivers/gpu/ion/ion_system_mapper.c create mode 100644 drivers/gpu/ion/mxc/Makefile create mode 100644 drivers/gpu/ion/mxc/mxc_ion.c create mode 100644 drivers/gpu/ion/tegra/Makefile create mode 100644 drivers/gpu/ion/tegra/tegra_ion.c create mode 100644 drivers/gpu/stub/Kconfig create mode 100644 drivers/gpu/stub/Makefile create mode 100644 drivers/gpu/stub/poulsbo.c create mode 100644 drivers/gpu/vga/Kconfig create mode 100644 drivers/gpu/vga/Makefile create mode 100644 drivers/gpu/vga/vga_switcheroo.c create mode 100644 drivers/gpu/vga/vgaarb.c (limited to 'drivers/gpu') diff --git a/drivers/gpu/Makefile b/drivers/gpu/Makefile new file mode 100644 index 00000000..ca2d3b34 --- /dev/null +++ b/drivers/gpu/Makefile @@ -0,0 +1 @@ +obj-y += drm/ vga/ stub/ ion/ diff --git a/drivers/gpu/drm/Kconfig b/drivers/gpu/drm/Kconfig new file mode 100644 index 00000000..e93b8e0c --- /dev/null +++ b/drivers/gpu/drm/Kconfig @@ -0,0 +1,43 @@ +# +# Drm device configuration +# +# This driver provides support for the +# Direct Rendering Infrastructure (DRI) in XFree86 4.1.0 and higher. +# +menuconfig DRM + tristate "Direct Rendering Manager (XFree86 4.1.0 and higher DRI support)" + select I2C + select I2C_ALGOBIT + select SLOW_WORK + help + Kernel-level support for the Direct Rendering Infrastructure (DRI) + introduced in XFree86 4.0. If you say Y here, you need to select + the module that's right for your graphics card from the list below. + These modules provide support for synchronization, security, and + DMA transfers. Please see for more + details. You should also select and configure AGP + (/dev/agpgart) support if it is available for your platform. + +config DRM_KMS_HELPER + tristate + depends on DRM + select FB + select FRAMEBUFFER_CONSOLE if !EXPERT + help + FB and CRTC helpers for KMS drivers. + +config DRM_TTM + tristate + depends on DRM + help + GPU memory management subsystem for devices with multiple + GPU memory types. Will be enabled automatically if a device driver + uses it. + +config DRM_VIVANTE + tristate "Vivante GCCore" + depends on DRM + help + Choose this option if you have a Vivante graphics card. + If M is selected, the module will be called vivante. + diff --git a/drivers/gpu/drm/Makefile b/drivers/gpu/drm/Makefile new file mode 100644 index 00000000..d2f5809f --- /dev/null +++ b/drivers/gpu/drm/Makefile @@ -0,0 +1,60 @@ +############################################################################## +# +# Copyright (C) 2005 - 2012 by Vivante Corp. +# +# This program is free software; you can redistribute it and/or modify +# it under the terms of the GNU General Public License as published by +# the Free Software Foundation; either version 2 of the license, or +# (at your option) any later version. +# +# This program is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU General Public License for more details. +# +# You should have received a copy of the GNU General Public License +# along with this program; if not write to the Free Software +# Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. +# +############################################################################## + + +# +# Makefile for the drm device driver. This driver provides support for the +# Direct Rendering Infrastructure (DRI) in XFree86 4.1.0 and higher. + +ccflags-y := -Iinclude/drm + +drm-y := drm_auth.o drm_buffer.o drm_bufs.o drm_cache.o \ + drm_context.o drm_dma.o \ + drm_drv.o drm_fops.o drm_gem.o drm_ioctl.o drm_irq.o \ + drm_lock.o drm_memory.o drm_proc.o drm_stub.o drm_vm.o \ + drm_agpsupport.o drm_scatter.o ati_pcigart.o drm_pci.o \ + drm_platform.o drm_sysfs.o drm_hashtab.o drm_sman.o drm_mm.o \ + drm_crtc.o drm_modes.o drm_edid.o \ + drm_info.o drm_debugfs.o drm_encoder_slave.o \ + drm_trace_points.o drm_global.o + +drm-$(CONFIG_COMPAT) += drm_ioc32.o + +drm_kms_helper-y := drm_fb_helper.o drm_crtc_helper.o drm_dp_i2c_helper.o + +obj-$(CONFIG_DRM_KMS_HELPER) += drm_kms_helper.o + +CFLAGS_drm_trace_points.o := -I$(src) + +obj-$(CONFIG_DRM) += drm.o +obj-$(CONFIG_DRM_TTM) += ttm/ +obj-$(CONFIG_DRM_VIVANTE) += vivante/ +obj-$(CONFIG_DRM_TDFX) += tdfx/ +obj-$(CONFIG_DRM_R128) += r128/ +obj-$(CONFIG_DRM_RADEON)+= radeon/ +obj-$(CONFIG_DRM_MGA) += mga/ +obj-$(CONFIG_DRM_I810) += i810/ +obj-$(CONFIG_DRM_I915) += i915/ +obj-$(CONFIG_DRM_SIS) += sis/ +obj-$(CONFIG_DRM_SAVAGE)+= savage/ +obj-$(CONFIG_DRM_VMWGFX)+= vmwgfx/ +obj-$(CONFIG_DRM_VIA) +=via/ +obj-$(CONFIG_DRM_NOUVEAU) +=nouveau/ +obj-y += i2c/ diff --git a/drivers/gpu/drm/README.drm b/drivers/gpu/drm/README.drm new file mode 100644 index 00000000..b5b33272 --- /dev/null +++ b/drivers/gpu/drm/README.drm @@ -0,0 +1,43 @@ +************************************************************ +* For the very latest on DRI development, please see: * +* http://dri.freedesktop.org/ * +************************************************************ + +The Direct Rendering Manager (drm) is a device-independent kernel-level +device driver that provides support for the XFree86 Direct Rendering +Infrastructure (DRI). + +The DRM supports the Direct Rendering Infrastructure (DRI) in four major +ways: + + 1. The DRM provides synchronized access to the graphics hardware via + the use of an optimized two-tiered lock. + + 2. The DRM enforces the DRI security policy for access to the graphics + hardware by only allowing authenticated X11 clients access to + restricted regions of memory. + + 3. The DRM provides a generic DMA engine, complete with multiple + queues and the ability to detect the need for an OpenGL context + switch. + + 4. The DRM is extensible via the use of small device-specific modules + that rely extensively on the API exported by the DRM module. + + +Documentation on the DRI is available from: + http://dri.freedesktop.org/wiki/Documentation + http://sourceforge.net/project/showfiles.php?group_id=387 + http://dri.sourceforge.net/doc/ + +For specific information about kernel-level support, see: + + The Direct Rendering Manager, Kernel Support for the Direct Rendering + Infrastructure + http://dri.sourceforge.net/doc/drm_low_level.html + + Hardware Locking for the Direct Rendering Infrastructure + http://dri.sourceforge.net/doc/hardware_locking_low_level.html + + A Security Analysis of the Direct Rendering Infrastructure + http://dri.sourceforge.net/doc/security_low_level.html diff --git a/drivers/gpu/drm/ati_pcigart.c b/drivers/gpu/drm/ati_pcigart.c new file mode 100644 index 00000000..1c364924 --- /dev/null +++ b/drivers/gpu/drm/ati_pcigart.c @@ -0,0 +1,201 @@ +/** + * \file ati_pcigart.c + * ATI PCI GART support + * + * \author Gareth Hughes + */ + +/* + * Created: Wed Dec 13 21:52:19 2000 by gareth@valinux.com + * + * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + */ + +#include "drmP.h" + +# define ATI_PCIGART_PAGE_SIZE 4096 /**< PCI GART page size */ + +static int drm_ati_alloc_pcigart_table(struct drm_device *dev, + struct drm_ati_pcigart_info *gart_info) +{ + gart_info->table_handle = drm_pci_alloc(dev, gart_info->table_size, + PAGE_SIZE); + if (gart_info->table_handle == NULL) + return -ENOMEM; + + return 0; +} + +static void drm_ati_free_pcigart_table(struct drm_device *dev, + struct drm_ati_pcigart_info *gart_info) +{ + drm_pci_free(dev, gart_info->table_handle); + gart_info->table_handle = NULL; +} + +int drm_ati_pcigart_cleanup(struct drm_device *dev, struct drm_ati_pcigart_info *gart_info) +{ + struct drm_sg_mem *entry = dev->sg; + unsigned long pages; + int i; + int max_pages; + + /* we need to support large memory configurations */ + if (!entry) { + DRM_ERROR("no scatter/gather memory!\n"); + return 0; + } + + if (gart_info->bus_addr) { + + max_pages = (gart_info->table_size / sizeof(u32)); + pages = (entry->pages <= max_pages) + ? entry->pages : max_pages; + + for (i = 0; i < pages; i++) { + if (!entry->busaddr[i]) + break; + pci_unmap_page(dev->pdev, entry->busaddr[i], + PAGE_SIZE, PCI_DMA_BIDIRECTIONAL); + } + + if (gart_info->gart_table_location == DRM_ATI_GART_MAIN) + gart_info->bus_addr = 0; + } + + if (gart_info->gart_table_location == DRM_ATI_GART_MAIN && + gart_info->table_handle) { + drm_ati_free_pcigart_table(dev, gart_info); + } + + return 1; +} +EXPORT_SYMBOL(drm_ati_pcigart_cleanup); + +int drm_ati_pcigart_init(struct drm_device *dev, struct drm_ati_pcigart_info *gart_info) +{ + struct drm_local_map *map = &gart_info->mapping; + struct drm_sg_mem *entry = dev->sg; + void *address = NULL; + unsigned long pages; + u32 *pci_gart = NULL, page_base, gart_idx; + dma_addr_t bus_address = 0; + int i, j, ret = 0; + int max_ati_pages, max_real_pages; + + if (!entry) { + DRM_ERROR("no scatter/gather memory!\n"); + goto done; + } + + if (gart_info->gart_table_location == DRM_ATI_GART_MAIN) { + DRM_DEBUG("PCI: no table in VRAM: using normal RAM\n"); + + if (pci_set_dma_mask(dev->pdev, gart_info->table_mask)) { + DRM_ERROR("fail to set dma mask to 0x%Lx\n", + (unsigned long long)gart_info->table_mask); + ret = 1; + goto done; + } + + ret = drm_ati_alloc_pcigart_table(dev, gart_info); + if (ret) { + DRM_ERROR("cannot allocate PCI GART page!\n"); + goto done; + } + + pci_gart = gart_info->table_handle->vaddr; + address = gart_info->table_handle->vaddr; + bus_address = gart_info->table_handle->busaddr; + } else { + address = gart_info->addr; + bus_address = gart_info->bus_addr; + DRM_DEBUG("PCI: Gart Table: VRAM %08LX mapped at %08lX\n", + (unsigned long long)bus_address, + (unsigned long)address); + } + + + max_ati_pages = (gart_info->table_size / sizeof(u32)); + max_real_pages = max_ati_pages / (PAGE_SIZE / ATI_PCIGART_PAGE_SIZE); + pages = (entry->pages <= max_real_pages) + ? entry->pages : max_real_pages; + + if (gart_info->gart_table_location == DRM_ATI_GART_MAIN) { + memset(pci_gart, 0, max_ati_pages * sizeof(u32)); + } else { + memset_io((void __iomem *)map->handle, 0, max_ati_pages * sizeof(u32)); + } + + gart_idx = 0; + for (i = 0; i < pages; i++) { + /* we need to support large memory configurations */ + entry->busaddr[i] = pci_map_page(dev->pdev, entry->pagelist[i], + 0, PAGE_SIZE, PCI_DMA_BIDIRECTIONAL); + if (pci_dma_mapping_error(dev->pdev, entry->busaddr[i])) { + DRM_ERROR("unable to map PCIGART pages!\n"); + drm_ati_pcigart_cleanup(dev, gart_info); + address = NULL; + bus_address = 0; + goto done; + } + page_base = (u32) entry->busaddr[i]; + + for (j = 0; j < (PAGE_SIZE / ATI_PCIGART_PAGE_SIZE); j++) { + u32 val; + + switch(gart_info->gart_reg_if) { + case DRM_ATI_GART_IGP: + val = page_base | 0xc; + break; + case DRM_ATI_GART_PCIE: + val = (page_base >> 8) | 0xc; + break; + default: + case DRM_ATI_GART_PCI: + val = page_base; + break; + } + if (gart_info->gart_table_location == + DRM_ATI_GART_MAIN) + pci_gart[gart_idx] = cpu_to_le32(val); + else + DRM_WRITE32(map, gart_idx * sizeof(u32), val); + gart_idx++; + page_base += ATI_PCIGART_PAGE_SIZE; + } + } + ret = 1; + +#if defined(__i386__) || defined(__x86_64__) + wbinvd(); +#else + mb(); +#endif + + done: + gart_info->addr = address; + gart_info->bus_addr = bus_address; + return ret; +} +EXPORT_SYMBOL(drm_ati_pcigart_init); diff --git a/drivers/gpu/drm/drm_agpsupport.c b/drivers/gpu/drm/drm_agpsupport.c new file mode 100644 index 00000000..0cb2ba50 --- /dev/null +++ b/drivers/gpu/drm/drm_agpsupport.c @@ -0,0 +1,469 @@ +/** + * \file drm_agpsupport.c + * DRM support for AGP/GART backend + * + * \author Rickard E. (Rik) Faith + * \author Gareth Hughes + */ + +/* + * Copyright 1999 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * VA LINUX SYSTEMS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#include "drmP.h" +#include +#include + +#if __OS_HAS_AGP + +#include + +/** + * Get AGP information. + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg pointer to a (output) drm_agp_info structure. + * \return zero on success or a negative number on failure. + * + * Verifies the AGP device has been initialized and acquired and fills in the + * drm_agp_info structure with the information in drm_agp_head::agp_info. + */ +int drm_agp_info(struct drm_device *dev, struct drm_agp_info *info) +{ + DRM_AGP_KERN *kern; + + if (!dev->agp || !dev->agp->acquired) + return -EINVAL; + + kern = &dev->agp->agp_info; + info->agp_version_major = kern->version.major; + info->agp_version_minor = kern->version.minor; + info->mode = kern->mode; + info->aperture_base = kern->aper_base; + info->aperture_size = kern->aper_size * 1024 * 1024; + info->memory_allowed = kern->max_memory << PAGE_SHIFT; + info->memory_used = kern->current_memory << PAGE_SHIFT; + info->id_vendor = kern->device->vendor; + info->id_device = kern->device->device; + + return 0; +} + +EXPORT_SYMBOL(drm_agp_info); + +int drm_agp_info_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_agp_info *info = data; + int err; + + err = drm_agp_info(dev, info); + if (err) + return err; + + return 0; +} + +/** + * Acquire the AGP device. + * + * \param dev DRM device that is to acquire AGP. + * \return zero on success or a negative number on failure. + * + * Verifies the AGP device hasn't been acquired before and calls + * \c agp_backend_acquire. + */ +int drm_agp_acquire(struct drm_device * dev) +{ + if (!dev->agp) + return -ENODEV; + if (dev->agp->acquired) + return -EBUSY; + if (!(dev->agp->bridge = agp_backend_acquire(dev->pdev))) + return -ENODEV; + dev->agp->acquired = 1; + return 0; +} + +EXPORT_SYMBOL(drm_agp_acquire); + +/** + * Acquire the AGP device (ioctl). + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg user argument. + * \return zero on success or a negative number on failure. + * + * Verifies the AGP device hasn't been acquired before and calls + * \c agp_backend_acquire. + */ +int drm_agp_acquire_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + return drm_agp_acquire((struct drm_device *) file_priv->minor->dev); +} + +/** + * Release the AGP device. + * + * \param dev DRM device that is to release AGP. + * \return zero on success or a negative number on failure. + * + * Verifies the AGP device has been acquired and calls \c agp_backend_release. + */ +int drm_agp_release(struct drm_device * dev) +{ + if (!dev->agp || !dev->agp->acquired) + return -EINVAL; + agp_backend_release(dev->agp->bridge); + dev->agp->acquired = 0; + return 0; +} +EXPORT_SYMBOL(drm_agp_release); + +int drm_agp_release_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + return drm_agp_release(dev); +} + +/** + * Enable the AGP bus. + * + * \param dev DRM device that has previously acquired AGP. + * \param mode Requested AGP mode. + * \return zero on success or a negative number on failure. + * + * Verifies the AGP device has been acquired but not enabled, and calls + * \c agp_enable. + */ +int drm_agp_enable(struct drm_device * dev, struct drm_agp_mode mode) +{ + if (!dev->agp || !dev->agp->acquired) + return -EINVAL; + + dev->agp->mode = mode.mode; + agp_enable(dev->agp->bridge, mode.mode); + dev->agp->enabled = 1; + return 0; +} + +EXPORT_SYMBOL(drm_agp_enable); + +int drm_agp_enable_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_agp_mode *mode = data; + + return drm_agp_enable(dev, *mode); +} + +/** + * Allocate AGP memory. + * + * \param inode device inode. + * \param file_priv file private pointer. + * \param cmd command. + * \param arg pointer to a drm_agp_buffer structure. + * \return zero on success or a negative number on failure. + * + * Verifies the AGP device is present and has been acquired, allocates the + * memory via agp_allocate_memory() and creates a drm_agp_mem entry for it. + */ +int drm_agp_alloc(struct drm_device *dev, struct drm_agp_buffer *request) +{ + struct drm_agp_mem *entry; + DRM_AGP_MEM *memory; + unsigned long pages; + u32 type; + + if (!dev->agp || !dev->agp->acquired) + return -EINVAL; + if (!(entry = kmalloc(sizeof(*entry), GFP_KERNEL))) + return -ENOMEM; + + memset(entry, 0, sizeof(*entry)); + + pages = (request->size + PAGE_SIZE - 1) / PAGE_SIZE; + type = (u32) request->type; + if (!(memory = agp_allocate_memory(dev->agp->bridge, pages, type))) { + kfree(entry); + return -ENOMEM; + } + + entry->handle = (unsigned long)memory->key + 1; + entry->memory = memory; + entry->bound = 0; + entry->pages = pages; + list_add(&entry->head, &dev->agp->memory); + + request->handle = entry->handle; + request->physical = memory->physical; + + return 0; +} +EXPORT_SYMBOL(drm_agp_alloc); + + +int drm_agp_alloc_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_agp_buffer *request = data; + + return drm_agp_alloc(dev, request); +} + +/** + * Search for the AGP memory entry associated with a handle. + * + * \param dev DRM device structure. + * \param handle AGP memory handle. + * \return pointer to the drm_agp_mem structure associated with \p handle. + * + * Walks through drm_agp_head::memory until finding a matching handle. + */ +static struct drm_agp_mem *drm_agp_lookup_entry(struct drm_device * dev, + unsigned long handle) +{ + struct drm_agp_mem *entry; + + list_for_each_entry(entry, &dev->agp->memory, head) { + if (entry->handle == handle) + return entry; + } + return NULL; +} + +/** + * Unbind AGP memory from the GATT (ioctl). + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg pointer to a drm_agp_binding structure. + * \return zero on success or a negative number on failure. + * + * Verifies the AGP device is present and acquired, looks-up the AGP memory + * entry and passes it to the unbind_agp() function. + */ +int drm_agp_unbind(struct drm_device *dev, struct drm_agp_binding *request) +{ + struct drm_agp_mem *entry; + int ret; + + if (!dev->agp || !dev->agp->acquired) + return -EINVAL; + if (!(entry = drm_agp_lookup_entry(dev, request->handle))) + return -EINVAL; + if (!entry->bound) + return -EINVAL; + ret = drm_unbind_agp(entry->memory); + if (ret == 0) + entry->bound = 0; + return ret; +} +EXPORT_SYMBOL(drm_agp_unbind); + + +int drm_agp_unbind_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_agp_binding *request = data; + + return drm_agp_unbind(dev, request); +} + +/** + * Bind AGP memory into the GATT (ioctl) + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg pointer to a drm_agp_binding structure. + * \return zero on success or a negative number on failure. + * + * Verifies the AGP device is present and has been acquired and that no memory + * is currently bound into the GATT. Looks-up the AGP memory entry and passes + * it to bind_agp() function. + */ +int drm_agp_bind(struct drm_device *dev, struct drm_agp_binding *request) +{ + struct drm_agp_mem *entry; + int retcode; + int page; + + if (!dev->agp || !dev->agp->acquired) + return -EINVAL; + if (!(entry = drm_agp_lookup_entry(dev, request->handle))) + return -EINVAL; + if (entry->bound) + return -EINVAL; + page = (request->offset + PAGE_SIZE - 1) / PAGE_SIZE; + if ((retcode = drm_bind_agp(entry->memory, page))) + return retcode; + entry->bound = dev->agp->base + (page << PAGE_SHIFT); + DRM_DEBUG("base = 0x%lx entry->bound = 0x%lx\n", + dev->agp->base, entry->bound); + return 0; +} +EXPORT_SYMBOL(drm_agp_bind); + + +int drm_agp_bind_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_agp_binding *request = data; + + return drm_agp_bind(dev, request); +} + +/** + * Free AGP memory (ioctl). + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg pointer to a drm_agp_buffer structure. + * \return zero on success or a negative number on failure. + * + * Verifies the AGP device is present and has been acquired and looks up the + * AGP memory entry. If the memory it's currently bound, unbind it via + * unbind_agp(). Frees it via free_agp() as well as the entry itself + * and unlinks from the doubly linked list it's inserted in. + */ +int drm_agp_free(struct drm_device *dev, struct drm_agp_buffer *request) +{ + struct drm_agp_mem *entry; + + if (!dev->agp || !dev->agp->acquired) + return -EINVAL; + if (!(entry = drm_agp_lookup_entry(dev, request->handle))) + return -EINVAL; + if (entry->bound) + drm_unbind_agp(entry->memory); + + list_del(&entry->head); + + drm_free_agp(entry->memory, entry->pages); + kfree(entry); + return 0; +} +EXPORT_SYMBOL(drm_agp_free); + + + +int drm_agp_free_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_agp_buffer *request = data; + + return drm_agp_free(dev, request); +} + +/** + * Initialize the AGP resources. + * + * \return pointer to a drm_agp_head structure. + * + * Gets the drm_agp_t structure which is made available by the agpgart module + * via the inter_module_* functions. Creates and initializes a drm_agp_head + * structure. + */ +struct drm_agp_head *drm_agp_init(struct drm_device *dev) +{ + struct drm_agp_head *head = NULL; + + if (!(head = kmalloc(sizeof(*head), GFP_KERNEL))) + return NULL; + memset((void *)head, 0, sizeof(*head)); + head->bridge = agp_find_bridge(dev->pdev); + if (!head->bridge) { + if (!(head->bridge = agp_backend_acquire(dev->pdev))) { + kfree(head); + return NULL; + } + agp_copy_info(head->bridge, &head->agp_info); + agp_backend_release(head->bridge); + } else { + agp_copy_info(head->bridge, &head->agp_info); + } + if (head->agp_info.chipset == NOT_SUPPORTED) { + kfree(head); + return NULL; + } + INIT_LIST_HEAD(&head->memory); + head->cant_use_aperture = head->agp_info.cant_use_aperture; + head->page_mask = head->agp_info.page_mask; + head->base = head->agp_info.aper_base; + return head; +} + +/** + * Binds a collection of pages into AGP memory at the given offset, returning + * the AGP memory structure containing them. + * + * No reference is held on the pages during this time -- it is up to the + * caller to handle that. + */ +DRM_AGP_MEM * +drm_agp_bind_pages(struct drm_device *dev, + struct page **pages, + unsigned long num_pages, + uint32_t gtt_offset, + u32 type) +{ + DRM_AGP_MEM *mem; + int ret, i; + + DRM_DEBUG("\n"); + + mem = agp_allocate_memory(dev->agp->bridge, num_pages, + type); + if (mem == NULL) { + DRM_ERROR("Failed to allocate memory for %ld pages\n", + num_pages); + return NULL; + } + + for (i = 0; i < num_pages; i++) + mem->pages[i] = pages[i]; + mem->page_count = num_pages; + + mem->is_flushed = true; + ret = agp_bind_memory(mem, gtt_offset / PAGE_SIZE); + if (ret != 0) { + DRM_ERROR("Failed to bind AGP memory: %d\n", ret); + agp_free_memory(mem); + return NULL; + } + + return mem; +} +EXPORT_SYMBOL(drm_agp_bind_pages); + +#endif /* __OS_HAS_AGP */ diff --git a/drivers/gpu/drm/drm_auth.c b/drivers/gpu/drm/drm_auth.c new file mode 100644 index 00000000..ba237904 --- /dev/null +++ b/drivers/gpu/drm/drm_auth.c @@ -0,0 +1,194 @@ +/** + * \file drm_auth.c + * IOCTLs for authentication + * + * \author Rickard E. (Rik) Faith + * \author Gareth Hughes + */ + +/* + * Created: Tue Feb 2 08:37:54 1999 by faith@valinux.com + * + * Copyright 1999 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * VA LINUX SYSTEMS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#include "drmP.h" + +/** + * Find the file with the given magic number. + * + * \param dev DRM device. + * \param magic magic number. + * + * Searches in drm_device::magiclist within all files with the same hash key + * the one with matching magic number, while holding the drm_device::struct_mutex + * lock. + */ +static struct drm_file *drm_find_file(struct drm_master *master, drm_magic_t magic) +{ + struct drm_file *retval = NULL; + struct drm_magic_entry *pt; + struct drm_hash_item *hash; + struct drm_device *dev = master->minor->dev; + + mutex_lock(&dev->struct_mutex); + if (!drm_ht_find_item(&master->magiclist, (unsigned long)magic, &hash)) { + pt = drm_hash_entry(hash, struct drm_magic_entry, hash_item); + retval = pt->priv; + } + mutex_unlock(&dev->struct_mutex); + return retval; +} + +/** + * Adds a magic number. + * + * \param dev DRM device. + * \param priv file private data. + * \param magic magic number. + * + * Creates a drm_magic_entry structure and appends to the linked list + * associated the magic number hash key in drm_device::magiclist, while holding + * the drm_device::struct_mutex lock. + */ +static int drm_add_magic(struct drm_master *master, struct drm_file *priv, + drm_magic_t magic) +{ + struct drm_magic_entry *entry; + struct drm_device *dev = master->minor->dev; + DRM_DEBUG("%d\n", magic); + + entry = kzalloc(sizeof(*entry), GFP_KERNEL); + if (!entry) + return -ENOMEM; + entry->priv = priv; + entry->hash_item.key = (unsigned long)magic; + mutex_lock(&dev->struct_mutex); + drm_ht_insert_item(&master->magiclist, &entry->hash_item); + list_add_tail(&entry->head, &master->magicfree); + mutex_unlock(&dev->struct_mutex); + + return 0; +} + +/** + * Remove a magic number. + * + * \param dev DRM device. + * \param magic magic number. + * + * Searches and unlinks the entry in drm_device::magiclist with the magic + * number hash key, while holding the drm_device::struct_mutex lock. + */ +int drm_remove_magic(struct drm_master *master, drm_magic_t magic) +{ + struct drm_magic_entry *pt; + struct drm_hash_item *hash; + struct drm_device *dev = master->minor->dev; + + DRM_DEBUG("%d\n", magic); + + mutex_lock(&dev->struct_mutex); + if (drm_ht_find_item(&master->magiclist, (unsigned long)magic, &hash)) { + mutex_unlock(&dev->struct_mutex); + return -EINVAL; + } + pt = drm_hash_entry(hash, struct drm_magic_entry, hash_item); + drm_ht_remove_item(&master->magiclist, hash); + list_del(&pt->head); + mutex_unlock(&dev->struct_mutex); + + kfree(pt); + + return 0; +} + +/** + * Get a unique magic number (ioctl). + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg pointer to a resulting drm_auth structure. + * \return zero on success, or a negative number on failure. + * + * If there is a magic number in drm_file::magic then use it, otherwise + * searches an unique non-zero magic number and add it associating it with \p + * file_priv. + * This ioctl needs protection by the drm_global_mutex, which protects + * struct drm_file::magic and struct drm_magic_entry::priv. + */ +int drm_getmagic(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + static drm_magic_t sequence = 0; + static DEFINE_SPINLOCK(lock); + struct drm_auth *auth = data; + + /* Find unique magic */ + if (file_priv->magic) { + auth->magic = file_priv->magic; + } else { + do { + spin_lock(&lock); + if (!sequence) + ++sequence; /* reserve 0 */ + auth->magic = sequence++; + spin_unlock(&lock); + } while (drm_find_file(file_priv->master, auth->magic)); + file_priv->magic = auth->magic; + drm_add_magic(file_priv->master, file_priv, auth->magic); + } + + DRM_DEBUG("%u\n", auth->magic); + + return 0; +} + +/** + * Authenticate with a magic. + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg pointer to a drm_auth structure. + * \return zero if authentication successed, or a negative number otherwise. + * + * Checks if \p file_priv is associated with the magic number passed in \arg. + * This ioctl needs protection by the drm_global_mutex, which protects + * struct drm_file::magic and struct drm_magic_entry::priv. + */ +int drm_authmagic(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_auth *auth = data; + struct drm_file *file; + + DRM_DEBUG("%u\n", auth->magic); + if ((file = drm_find_file(file_priv->master, auth->magic))) { + file->authenticated = 1; + drm_remove_magic(file_priv->master, auth->magic); + return 0; + } + return -EINVAL; +} diff --git a/drivers/gpu/drm/drm_buffer.c b/drivers/gpu/drm/drm_buffer.c new file mode 100644 index 00000000..529a0dbe --- /dev/null +++ b/drivers/gpu/drm/drm_buffer.c @@ -0,0 +1,184 @@ +/************************************************************************** + * + * Copyright 2010 Pauli Nieminen. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. + * + * + **************************************************************************/ +/* + * Multipart buffer for coping data which is larger than the page size. + * + * Authors: + * Pauli Nieminen + */ + +#include "drm_buffer.h" + +/** + * Allocate the drm buffer object. + * + * buf: Pointer to a pointer where the object is stored. + * size: The number of bytes to allocate. + */ +int drm_buffer_alloc(struct drm_buffer **buf, int size) +{ + int nr_pages = size / PAGE_SIZE + 1; + int idx; + + /* Allocating pointer table to end of structure makes drm_buffer + * variable sized */ + *buf = kzalloc(sizeof(struct drm_buffer) + nr_pages*sizeof(char *), + GFP_KERNEL); + + if (*buf == NULL) { + DRM_ERROR("Failed to allocate drm buffer object to hold" + " %d bytes in %d pages.\n", + size, nr_pages); + return -ENOMEM; + } + + (*buf)->size = size; + + for (idx = 0; idx < nr_pages; ++idx) { + + (*buf)->data[idx] = + kmalloc(min(PAGE_SIZE, size - idx * PAGE_SIZE), + GFP_KERNEL); + + + if ((*buf)->data[idx] == NULL) { + DRM_ERROR("Failed to allocate %dth page for drm" + " buffer with %d bytes and %d pages.\n", + idx + 1, size, nr_pages); + goto error_out; + } + + } + + return 0; + +error_out: + + /* Only last element can be null pointer so check for it first. */ + if ((*buf)->data[idx]) + kfree((*buf)->data[idx]); + + for (--idx; idx >= 0; --idx) + kfree((*buf)->data[idx]); + + kfree(*buf); + return -ENOMEM; +} +EXPORT_SYMBOL(drm_buffer_alloc); + +/** + * Copy the user data to the begin of the buffer and reset the processing + * iterator. + * + * user_data: A pointer the data that is copied to the buffer. + * size: The Number of bytes to copy. + */ +int drm_buffer_copy_from_user(struct drm_buffer *buf, + void __user *user_data, int size) +{ + int nr_pages = size / PAGE_SIZE + 1; + int idx; + + if (size > buf->size) { + DRM_ERROR("Requesting to copy %d bytes to a drm buffer with" + " %d bytes space\n", + size, buf->size); + return -EFAULT; + } + + for (idx = 0; idx < nr_pages; ++idx) { + + if (DRM_COPY_FROM_USER(buf->data[idx], + user_data + idx * PAGE_SIZE, + min(PAGE_SIZE, size - idx * PAGE_SIZE))) { + DRM_ERROR("Failed to copy user data (%p) to drm buffer" + " (%p) %dth page.\n", + user_data, buf, idx); + return -EFAULT; + + } + } + buf->iterator = 0; + return 0; +} +EXPORT_SYMBOL(drm_buffer_copy_from_user); + +/** + * Free the drm buffer object + */ +void drm_buffer_free(struct drm_buffer *buf) +{ + + if (buf != NULL) { + + int nr_pages = buf->size / PAGE_SIZE + 1; + int idx; + for (idx = 0; idx < nr_pages; ++idx) + kfree(buf->data[idx]); + + kfree(buf); + } +} +EXPORT_SYMBOL(drm_buffer_free); + +/** + * Read an object from buffer that may be split to multiple parts. If object + * is not split function just returns the pointer to object in buffer. But in + * case of split object data is copied to given stack object that is suplied + * by caller. + * + * The processing location of the buffer is also advanced to the next byte + * after the object. + * + * objsize: The size of the objet in bytes. + * stack_obj: A pointer to a memory location where object can be copied. + */ +void *drm_buffer_read_object(struct drm_buffer *buf, + int objsize, void *stack_obj) +{ + int idx = drm_buffer_index(buf); + int page = drm_buffer_page(buf); + void *obj = NULL; + + if (idx + objsize <= PAGE_SIZE) { + obj = &buf->data[page][idx]; + } else { + /* The object is split which forces copy to temporary object.*/ + int beginsz = PAGE_SIZE - idx; + memcpy(stack_obj, &buf->data[page][idx], beginsz); + + memcpy(stack_obj + beginsz, &buf->data[page + 1][0], + objsize - beginsz); + + obj = stack_obj; + } + + drm_buffer_advance(buf, objsize); + return obj; +} +EXPORT_SYMBOL(drm_buffer_read_object); diff --git a/drivers/gpu/drm/drm_bufs.c b/drivers/gpu/drm/drm_bufs.c new file mode 100644 index 00000000..61e1ef90 --- /dev/null +++ b/drivers/gpu/drm/drm_bufs.c @@ -0,0 +1,1629 @@ +/** + * \file drm_bufs.c + * Generic buffer template + * + * \author Rickard E. (Rik) Faith + * \author Gareth Hughes + */ + +/* + * Created: Thu Nov 23 03:10:50 2000 by gareth@valinux.com + * + * Copyright 1999, 2000 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * VA LINUX SYSTEMS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#include +#include +#include +#include +#include "drmP.h" + +static struct drm_map_list *drm_find_matching_map(struct drm_device *dev, + struct drm_local_map *map) +{ + struct drm_map_list *entry; + list_for_each_entry(entry, &dev->maplist, head) { + /* + * Because the kernel-userspace ABI is fixed at a 32-bit offset + * while PCI resources may live above that, we only compare the + * lower 32 bits of the map offset for maps of type + * _DRM_FRAMEBUFFER or _DRM_REGISTERS. + * It is assumed that if a driver have more than one resource + * of each type, the lower 32 bits are different. + */ + if (!entry->map || + map->type != entry->map->type || + entry->master != dev->primary->master) + continue; + switch (map->type) { + case _DRM_SHM: + if (map->flags != _DRM_CONTAINS_LOCK) + break; + return entry; + case _DRM_REGISTERS: + case _DRM_FRAME_BUFFER: + if ((entry->map->offset & 0xffffffff) == + (map->offset & 0xffffffff)) + return entry; + default: /* Make gcc happy */ + ; + } + if (entry->map->offset == map->offset) + return entry; + } + + return NULL; +} + +static int drm_map_handle(struct drm_device *dev, struct drm_hash_item *hash, + unsigned long user_token, int hashed_handle, int shm) +{ + int use_hashed_handle, shift; + unsigned long add; + +#if (BITS_PER_LONG == 64) + use_hashed_handle = ((user_token & 0xFFFFFFFF00000000UL) || hashed_handle); +#elif (BITS_PER_LONG == 32) + use_hashed_handle = hashed_handle; +#else +#error Unsupported long size. Neither 64 nor 32 bits. +#endif + + if (!use_hashed_handle) { + int ret; + hash->key = user_token >> PAGE_SHIFT; + ret = drm_ht_insert_item(&dev->map_hash, hash); + if (ret != -EINVAL) + return ret; + } + + shift = 0; + add = DRM_MAP_HASH_OFFSET >> PAGE_SHIFT; + if (shm && (SHMLBA > PAGE_SIZE)) { + int bits = ilog2(SHMLBA >> PAGE_SHIFT) + 1; + + /* For shared memory, we have to preserve the SHMLBA + * bits of the eventual vma->vm_pgoff value during + * mmap(). Otherwise we run into cache aliasing problems + * on some platforms. On these platforms, the pgoff of + * a mmap() request is used to pick a suitable virtual + * address for the mmap() region such that it will not + * cause cache aliasing problems. + * + * Therefore, make sure the SHMLBA relevant bits of the + * hash value we use are equal to those in the original + * kernel virtual address. + */ + shift = bits; + add |= ((user_token >> PAGE_SHIFT) & ((1UL << bits) - 1UL)); + } + + return drm_ht_just_insert_please(&dev->map_hash, hash, + user_token, 32 - PAGE_SHIFT - 3, + shift, add); +} + +/** + * Core function to create a range of memory available for mapping by a + * non-root process. + * + * Adjusts the memory offset to its absolute value according to the mapping + * type. Adds the map to the map list drm_device::maplist. Adds MTRR's where + * applicable and if supported by the kernel. + */ +static int drm_addmap_core(struct drm_device * dev, resource_size_t offset, + unsigned int size, enum drm_map_type type, + enum drm_map_flags flags, + struct drm_map_list ** maplist) +{ + struct drm_local_map *map; + struct drm_map_list *list; + drm_dma_handle_t *dmah; + unsigned long user_token; + int ret; + + map = kmalloc(sizeof(*map), GFP_KERNEL); + if (!map) + return -ENOMEM; + + map->offset = offset; + map->size = size; + map->flags = flags; + map->type = type; + + /* Only allow shared memory to be removable since we only keep enough + * book keeping information about shared memory to allow for removal + * when processes fork. + */ + if ((map->flags & _DRM_REMOVABLE) && map->type != _DRM_SHM) { + kfree(map); + return -EINVAL; + } + DRM_DEBUG("offset = 0x%08llx, size = 0x%08lx, type = %d\n", + (unsigned long long)map->offset, map->size, map->type); + + /* page-align _DRM_SHM maps. They are allocated here so there is no security + * hole created by that and it works around various broken drivers that use + * a non-aligned quantity to map the SAREA. --BenH + */ + if (map->type == _DRM_SHM) + map->size = PAGE_ALIGN(map->size); + + if ((map->offset & (~(resource_size_t)PAGE_MASK)) || (map->size & (~PAGE_MASK))) { + kfree(map); + return -EINVAL; + } + map->mtrr = -1; + map->handle = NULL; + + switch (map->type) { + case _DRM_REGISTERS: + case _DRM_FRAME_BUFFER: +#if !defined(__sparc__) && !defined(__alpha__) && !defined(__ia64__) && !defined(__powerpc64__) && !defined(__x86_64__) && !defined(__arm__) + if (map->offset + (map->size-1) < map->offset || + map->offset < virt_to_phys(high_memory)) { + kfree(map); + return -EINVAL; + } +#endif + /* Some drivers preinitialize some maps, without the X Server + * needing to be aware of it. Therefore, we just return success + * when the server tries to create a duplicate map. + */ + list = drm_find_matching_map(dev, map); + if (list != NULL) { + if (list->map->size != map->size) { + DRM_DEBUG("Matching maps of type %d with " + "mismatched sizes, (%ld vs %ld)\n", + map->type, map->size, + list->map->size); + list->map->size = map->size; + } + + kfree(map); + *maplist = list; + return 0; + } + + if (drm_core_has_MTRR(dev)) { + if (map->type == _DRM_FRAME_BUFFER || + (map->flags & _DRM_WRITE_COMBINING)) { + map->mtrr = mtrr_add(map->offset, map->size, + MTRR_TYPE_WRCOMB, 1); + } + } + if (map->type == _DRM_REGISTERS) { + map->handle = ioremap(map->offset, map->size); + if (!map->handle) { + kfree(map); + return -ENOMEM; + } + } + + break; + case _DRM_SHM: + list = drm_find_matching_map(dev, map); + if (list != NULL) { + if(list->map->size != map->size) { + DRM_DEBUG("Matching maps of type %d with " + "mismatched sizes, (%ld vs %ld)\n", + map->type, map->size, list->map->size); + list->map->size = map->size; + } + + kfree(map); + *maplist = list; + return 0; + } + map->handle = vmalloc_user(map->size); + DRM_DEBUG("%lu %d %p\n", + map->size, drm_order(map->size), map->handle); + if (!map->handle) { + kfree(map); + return -ENOMEM; + } + map->offset = (unsigned long)map->handle; + if (map->flags & _DRM_CONTAINS_LOCK) { + /* Prevent a 2nd X Server from creating a 2nd lock */ + if (dev->primary->master->lock.hw_lock != NULL) { + vfree(map->handle); + kfree(map); + return -EBUSY; + } + dev->sigdata.lock = dev->primary->master->lock.hw_lock = map->handle; /* Pointer to lock */ + } + break; + case _DRM_AGP: { + struct drm_agp_mem *entry; + int valid = 0; + + if (!drm_core_has_AGP(dev)) { + kfree(map); + return -EINVAL; + } +#ifdef __alpha__ + map->offset += dev->hose->mem_space->start; +#endif + /* In some cases (i810 driver), user space may have already + * added the AGP base itself, because dev->agp->base previously + * only got set during AGP enable. So, only add the base + * address if the map's offset isn't already within the + * aperture. + */ + if (map->offset < dev->agp->base || + map->offset > dev->agp->base + + dev->agp->agp_info.aper_size * 1024 * 1024 - 1) { + map->offset += dev->agp->base; + } + map->mtrr = dev->agp->agp_mtrr; /* for getmap */ + + /* This assumes the DRM is in total control of AGP space. + * It's not always the case as AGP can be in the control + * of user space (i.e. i810 driver). So this loop will get + * skipped and we double check that dev->agp->memory is + * actually set as well as being invalid before EPERM'ing + */ + list_for_each_entry(entry, &dev->agp->memory, head) { + if ((map->offset >= entry->bound) && + (map->offset + map->size <= entry->bound + entry->pages * PAGE_SIZE)) { + valid = 1; + break; + } + } + if (!list_empty(&dev->agp->memory) && !valid) { + kfree(map); + return -EPERM; + } + DRM_DEBUG("AGP offset = 0x%08llx, size = 0x%08lx\n", + (unsigned long long)map->offset, map->size); + + break; + } + case _DRM_GEM: + DRM_ERROR("tried to addmap GEM object\n"); + break; + case _DRM_SCATTER_GATHER: + if (!dev->sg) { + kfree(map); + return -EINVAL; + } + map->offset += (unsigned long)dev->sg->virtual; + break; + case _DRM_CONSISTENT: + /* dma_addr_t is 64bit on i386 with CONFIG_HIGHMEM64G, + * As we're limiting the address to 2^32-1 (or less), + * casting it down to 32 bits is no problem, but we + * need to point to a 64bit variable first. */ + dmah = drm_pci_alloc(dev, map->size, map->size); + if (!dmah) { + kfree(map); + return -ENOMEM; + } + map->handle = dmah->vaddr; + map->offset = (unsigned long)dmah->busaddr; + kfree(dmah); + break; + default: + kfree(map); + return -EINVAL; + } + + list = kzalloc(sizeof(*list), GFP_KERNEL); + if (!list) { + if (map->type == _DRM_REGISTERS) + iounmap(map->handle); + kfree(map); + return -EINVAL; + } + list->map = map; + + mutex_lock(&dev->struct_mutex); + list_add(&list->head, &dev->maplist); + + /* Assign a 32-bit handle */ + /* We do it here so that dev->struct_mutex protects the increment */ + user_token = (map->type == _DRM_SHM) ? (unsigned long)map->handle : + map->offset; + ret = drm_map_handle(dev, &list->hash, user_token, 0, + (map->type == _DRM_SHM)); + if (ret) { + if (map->type == _DRM_REGISTERS) + iounmap(map->handle); + kfree(map); + kfree(list); + mutex_unlock(&dev->struct_mutex); + return ret; + } + + list->user_token = list->hash.key << PAGE_SHIFT; + mutex_unlock(&dev->struct_mutex); + + if (!(map->flags & _DRM_DRIVER)) + list->master = dev->primary->master; + *maplist = list; + return 0; + } + +int drm_addmap(struct drm_device * dev, resource_size_t offset, + unsigned int size, enum drm_map_type type, + enum drm_map_flags flags, struct drm_local_map ** map_ptr) +{ + struct drm_map_list *list; + int rc; + + rc = drm_addmap_core(dev, offset, size, type, flags, &list); + if (!rc) + *map_ptr = list->map; + return rc; +} + +EXPORT_SYMBOL(drm_addmap); + +/** + * Ioctl to specify a range of memory that is available for mapping by a + * non-root process. + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg pointer to a drm_map structure. + * \return zero on success or a negative value on error. + * + */ +int drm_addmap_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_map *map = data; + struct drm_map_list *maplist; + int err; + + if (!(capable(CAP_SYS_ADMIN) || map->type == _DRM_AGP || map->type == _DRM_SHM)) + return -EPERM; + + err = drm_addmap_core(dev, map->offset, map->size, map->type, + map->flags, &maplist); + + if (err) + return err; + + /* avoid a warning on 64-bit, this casting isn't very nice, but the API is set so too late */ + map->handle = (void *)(unsigned long)maplist->user_token; + return 0; +} + +/** + * Remove a map private from list and deallocate resources if the mapping + * isn't in use. + * + * Searches the map on drm_device::maplist, removes it from the list, see if + * its being used, and free any associate resource (such as MTRR's) if it's not + * being on use. + * + * \sa drm_addmap + */ +int drm_rmmap_locked(struct drm_device *dev, struct drm_local_map *map) +{ + struct drm_map_list *r_list = NULL, *list_t; + drm_dma_handle_t dmah; + int found = 0; + struct drm_master *master; + + /* Find the list entry for the map and remove it */ + list_for_each_entry_safe(r_list, list_t, &dev->maplist, head) { + if (r_list->map == map) { + master = r_list->master; + list_del(&r_list->head); + drm_ht_remove_key(&dev->map_hash, + r_list->user_token >> PAGE_SHIFT); + kfree(r_list); + found = 1; + break; + } + } + + if (!found) + return -EINVAL; + + switch (map->type) { + case _DRM_REGISTERS: + iounmap(map->handle); + /* FALLTHROUGH */ + case _DRM_FRAME_BUFFER: + if (drm_core_has_MTRR(dev) && map->mtrr >= 0) { + int retcode; + retcode = mtrr_del(map->mtrr, map->offset, map->size); + DRM_DEBUG("mtrr_del=%d\n", retcode); + } + break; + case _DRM_SHM: + vfree(map->handle); + if (master) { + if (dev->sigdata.lock == master->lock.hw_lock) + dev->sigdata.lock = NULL; + master->lock.hw_lock = NULL; /* SHM removed */ + master->lock.file_priv = NULL; + wake_up_interruptible_all(&master->lock.lock_queue); + } + break; + case _DRM_AGP: + case _DRM_SCATTER_GATHER: + break; + case _DRM_CONSISTENT: + dmah.vaddr = map->handle; + dmah.busaddr = map->offset; + dmah.size = map->size; + __drm_pci_free(dev, &dmah); + break; + case _DRM_GEM: + DRM_ERROR("tried to rmmap GEM object\n"); + break; + } + kfree(map); + + return 0; +} +EXPORT_SYMBOL(drm_rmmap_locked); + +int drm_rmmap(struct drm_device *dev, struct drm_local_map *map) +{ + int ret; + + mutex_lock(&dev->struct_mutex); + ret = drm_rmmap_locked(dev, map); + mutex_unlock(&dev->struct_mutex); + + return ret; +} +EXPORT_SYMBOL(drm_rmmap); + +/* The rmmap ioctl appears to be unnecessary. All mappings are torn down on + * the last close of the device, and this is necessary for cleanup when things + * exit uncleanly. Therefore, having userland manually remove mappings seems + * like a pointless exercise since they're going away anyway. + * + * One use case might be after addmap is allowed for normal users for SHM and + * gets used by drivers that the server doesn't need to care about. This seems + * unlikely. + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg pointer to a struct drm_map structure. + * \return zero on success or a negative value on error. + */ +int drm_rmmap_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_map *request = data; + struct drm_local_map *map = NULL; + struct drm_map_list *r_list; + int ret; + + mutex_lock(&dev->struct_mutex); + list_for_each_entry(r_list, &dev->maplist, head) { + if (r_list->map && + r_list->user_token == (unsigned long)request->handle && + r_list->map->flags & _DRM_REMOVABLE) { + map = r_list->map; + break; + } + } + + /* List has wrapped around to the head pointer, or its empty we didn't + * find anything. + */ + if (list_empty(&dev->maplist) || !map) { + mutex_unlock(&dev->struct_mutex); + return -EINVAL; + } + + /* Register and framebuffer maps are permanent */ + if ((map->type == _DRM_REGISTERS) || (map->type == _DRM_FRAME_BUFFER)) { + mutex_unlock(&dev->struct_mutex); + return 0; + } + + ret = drm_rmmap_locked(dev, map); + + mutex_unlock(&dev->struct_mutex); + + return ret; +} + +/** + * Cleanup after an error on one of the addbufs() functions. + * + * \param dev DRM device. + * \param entry buffer entry where the error occurred. + * + * Frees any pages and buffers associated with the given entry. + */ +static void drm_cleanup_buf_error(struct drm_device * dev, + struct drm_buf_entry * entry) +{ + int i; + + if (entry->seg_count) { + for (i = 0; i < entry->seg_count; i++) { + if (entry->seglist[i]) { + drm_pci_free(dev, entry->seglist[i]); + } + } + kfree(entry->seglist); + + entry->seg_count = 0; + } + + if (entry->buf_count) { + for (i = 0; i < entry->buf_count; i++) { + kfree(entry->buflist[i].dev_private); + } + kfree(entry->buflist); + + entry->buf_count = 0; + } +} + +#if __OS_HAS_AGP +/** + * Add AGP buffers for DMA transfers. + * + * \param dev struct drm_device to which the buffers are to be added. + * \param request pointer to a struct drm_buf_desc describing the request. + * \return zero on success or a negative number on failure. + * + * After some sanity checks creates a drm_buf structure for each buffer and + * reallocates the buffer list of the same size order to accommodate the new + * buffers. + */ +int drm_addbufs_agp(struct drm_device * dev, struct drm_buf_desc * request) +{ + struct drm_device_dma *dma = dev->dma; + struct drm_buf_entry *entry; + struct drm_agp_mem *agp_entry; + struct drm_buf *buf; + unsigned long offset; + unsigned long agp_offset; + int count; + int order; + int size; + int alignment; + int page_order; + int total; + int byte_count; + int i, valid; + struct drm_buf **temp_buflist; + + if (!dma) + return -EINVAL; + + count = request->count; + order = drm_order(request->size); + size = 1 << order; + + alignment = (request->flags & _DRM_PAGE_ALIGN) + ? PAGE_ALIGN(size) : size; + page_order = order - PAGE_SHIFT > 0 ? order - PAGE_SHIFT : 0; + total = PAGE_SIZE << page_order; + + byte_count = 0; + agp_offset = dev->agp->base + request->agp_start; + + DRM_DEBUG("count: %d\n", count); + DRM_DEBUG("order: %d\n", order); + DRM_DEBUG("size: %d\n", size); + DRM_DEBUG("agp_offset: %lx\n", agp_offset); + DRM_DEBUG("alignment: %d\n", alignment); + DRM_DEBUG("page_order: %d\n", page_order); + DRM_DEBUG("total: %d\n", total); + + if (order < DRM_MIN_ORDER || order > DRM_MAX_ORDER) + return -EINVAL; + if (dev->queue_count) + return -EBUSY; /* Not while in use */ + + /* Make sure buffers are located in AGP memory that we own */ + valid = 0; + list_for_each_entry(agp_entry, &dev->agp->memory, head) { + if ((agp_offset >= agp_entry->bound) && + (agp_offset + total * count <= agp_entry->bound + agp_entry->pages * PAGE_SIZE)) { + valid = 1; + break; + } + } + if (!list_empty(&dev->agp->memory) && !valid) { + DRM_DEBUG("zone invalid\n"); + return -EINVAL; + } + spin_lock(&dev->count_lock); + if (dev->buf_use) { + spin_unlock(&dev->count_lock); + return -EBUSY; + } + atomic_inc(&dev->buf_alloc); + spin_unlock(&dev->count_lock); + + mutex_lock(&dev->struct_mutex); + entry = &dma->bufs[order]; + if (entry->buf_count) { + mutex_unlock(&dev->struct_mutex); + atomic_dec(&dev->buf_alloc); + return -ENOMEM; /* May only call once for each order */ + } + + if (count < 0 || count > 4096) { + mutex_unlock(&dev->struct_mutex); + atomic_dec(&dev->buf_alloc); + return -EINVAL; + } + + entry->buflist = kzalloc(count * sizeof(*entry->buflist), GFP_KERNEL); + if (!entry->buflist) { + mutex_unlock(&dev->struct_mutex); + atomic_dec(&dev->buf_alloc); + return -ENOMEM; + } + + entry->buf_size = size; + entry->page_order = page_order; + + offset = 0; + + while (entry->buf_count < count) { + buf = &entry->buflist[entry->buf_count]; + buf->idx = dma->buf_count + entry->buf_count; + buf->total = alignment; + buf->order = order; + buf->used = 0; + + buf->offset = (dma->byte_count + offset); + buf->bus_address = agp_offset + offset; + buf->address = (void *)(agp_offset + offset); + buf->next = NULL; + buf->waiting = 0; + buf->pending = 0; + init_waitqueue_head(&buf->dma_wait); + buf->file_priv = NULL; + + buf->dev_priv_size = dev->driver->dev_priv_size; + buf->dev_private = kzalloc(buf->dev_priv_size, GFP_KERNEL); + if (!buf->dev_private) { + /* Set count correctly so we free the proper amount. */ + entry->buf_count = count; + drm_cleanup_buf_error(dev, entry); + mutex_unlock(&dev->struct_mutex); + atomic_dec(&dev->buf_alloc); + return -ENOMEM; + } + + DRM_DEBUG("buffer %d @ %p\n", entry->buf_count, buf->address); + + offset += alignment; + entry->buf_count++; + byte_count += PAGE_SIZE << page_order; + } + + DRM_DEBUG("byte_count: %d\n", byte_count); + + temp_buflist = krealloc(dma->buflist, + (dma->buf_count + entry->buf_count) * + sizeof(*dma->buflist), GFP_KERNEL); + if (!temp_buflist) { + /* Free the entry because it isn't valid */ + drm_cleanup_buf_error(dev, entry); + mutex_unlock(&dev->struct_mutex); + atomic_dec(&dev->buf_alloc); + return -ENOMEM; + } + dma->buflist = temp_buflist; + + for (i = 0; i < entry->buf_count; i++) { + dma->buflist[i + dma->buf_count] = &entry->buflist[i]; + } + + dma->buf_count += entry->buf_count; + dma->seg_count += entry->seg_count; + dma->page_count += byte_count >> PAGE_SHIFT; + dma->byte_count += byte_count; + + DRM_DEBUG("dma->buf_count : %d\n", dma->buf_count); + DRM_DEBUG("entry->buf_count : %d\n", entry->buf_count); + + mutex_unlock(&dev->struct_mutex); + + request->count = entry->buf_count; + request->size = size; + + dma->flags = _DRM_DMA_USE_AGP; + + atomic_dec(&dev->buf_alloc); + return 0; +} +EXPORT_SYMBOL(drm_addbufs_agp); +#endif /* __OS_HAS_AGP */ + +int drm_addbufs_pci(struct drm_device * dev, struct drm_buf_desc * request) +{ + struct drm_device_dma *dma = dev->dma; + int count; + int order; + int size; + int total; + int page_order; + struct drm_buf_entry *entry; + drm_dma_handle_t *dmah; + struct drm_buf *buf; + int alignment; + unsigned long offset; + int i; + int byte_count; + int page_count; + unsigned long *temp_pagelist; + struct drm_buf **temp_buflist; + + if (!drm_core_check_feature(dev, DRIVER_PCI_DMA)) + return -EINVAL; + + if (!dma) + return -EINVAL; + + if (!capable(CAP_SYS_ADMIN)) + return -EPERM; + + count = request->count; + order = drm_order(request->size); + size = 1 << order; + + DRM_DEBUG("count=%d, size=%d (%d), order=%d, queue_count=%d\n", + request->count, request->size, size, order, dev->queue_count); + + if (order < DRM_MIN_ORDER || order > DRM_MAX_ORDER) + return -EINVAL; + if (dev->queue_count) + return -EBUSY; /* Not while in use */ + + alignment = (request->flags & _DRM_PAGE_ALIGN) + ? PAGE_ALIGN(size) : size; + page_order = order - PAGE_SHIFT > 0 ? order - PAGE_SHIFT : 0; + total = PAGE_SIZE << page_order; + + spin_lock(&dev->count_lock); + if (dev->buf_use) { + spin_unlock(&dev->count_lock); + return -EBUSY; + } + atomic_inc(&dev->buf_alloc); + spin_unlock(&dev->count_lock); + + mutex_lock(&dev->struct_mutex); + entry = &dma->bufs[order]; + if (entry->buf_count) { + mutex_unlock(&dev->struct_mutex); + atomic_dec(&dev->buf_alloc); + return -ENOMEM; /* May only call once for each order */ + } + + if (count < 0 || count > 4096) { + mutex_unlock(&dev->struct_mutex); + atomic_dec(&dev->buf_alloc); + return -EINVAL; + } + + entry->buflist = kzalloc(count * sizeof(*entry->buflist), GFP_KERNEL); + if (!entry->buflist) { + mutex_unlock(&dev->struct_mutex); + atomic_dec(&dev->buf_alloc); + return -ENOMEM; + } + + entry->seglist = kzalloc(count * sizeof(*entry->seglist), GFP_KERNEL); + if (!entry->seglist) { + kfree(entry->buflist); + mutex_unlock(&dev->struct_mutex); + atomic_dec(&dev->buf_alloc); + return -ENOMEM; + } + + /* Keep the original pagelist until we know all the allocations + * have succeeded + */ + temp_pagelist = kmalloc((dma->page_count + (count << page_order)) * + sizeof(*dma->pagelist), GFP_KERNEL); + if (!temp_pagelist) { + kfree(entry->buflist); + kfree(entry->seglist); + mutex_unlock(&dev->struct_mutex); + atomic_dec(&dev->buf_alloc); + return -ENOMEM; + } + memcpy(temp_pagelist, + dma->pagelist, dma->page_count * sizeof(*dma->pagelist)); + DRM_DEBUG("pagelist: %d entries\n", + dma->page_count + (count << page_order)); + + entry->buf_size = size; + entry->page_order = page_order; + byte_count = 0; + page_count = 0; + + while (entry->buf_count < count) { + + dmah = drm_pci_alloc(dev, PAGE_SIZE << page_order, 0x1000); + + if (!dmah) { + /* Set count correctly so we free the proper amount. */ + entry->buf_count = count; + entry->seg_count = count; + drm_cleanup_buf_error(dev, entry); + kfree(temp_pagelist); + mutex_unlock(&dev->struct_mutex); + atomic_dec(&dev->buf_alloc); + return -ENOMEM; + } + entry->seglist[entry->seg_count++] = dmah; + for (i = 0; i < (1 << page_order); i++) { + DRM_DEBUG("page %d @ 0x%08lx\n", + dma->page_count + page_count, + (unsigned long)dmah->vaddr + PAGE_SIZE * i); + temp_pagelist[dma->page_count + page_count++] + = (unsigned long)dmah->vaddr + PAGE_SIZE * i; + } + for (offset = 0; + offset + size <= total && entry->buf_count < count; + offset += alignment, ++entry->buf_count) { + buf = &entry->buflist[entry->buf_count]; + buf->idx = dma->buf_count + entry->buf_count; + buf->total = alignment; + buf->order = order; + buf->used = 0; + buf->offset = (dma->byte_count + byte_count + offset); + buf->address = (void *)(dmah->vaddr + offset); + buf->bus_address = dmah->busaddr + offset; + buf->next = NULL; + buf->waiting = 0; + buf->pending = 0; + init_waitqueue_head(&buf->dma_wait); + buf->file_priv = NULL; + + buf->dev_priv_size = dev->driver->dev_priv_size; + buf->dev_private = kzalloc(buf->dev_priv_size, + GFP_KERNEL); + if (!buf->dev_private) { + /* Set count correctly so we free the proper amount. */ + entry->buf_count = count; + entry->seg_count = count; + drm_cleanup_buf_error(dev, entry); + kfree(temp_pagelist); + mutex_unlock(&dev->struct_mutex); + atomic_dec(&dev->buf_alloc); + return -ENOMEM; + } + + DRM_DEBUG("buffer %d @ %p\n", + entry->buf_count, buf->address); + } + byte_count += PAGE_SIZE << page_order; + } + + temp_buflist = krealloc(dma->buflist, + (dma->buf_count + entry->buf_count) * + sizeof(*dma->buflist), GFP_KERNEL); + if (!temp_buflist) { + /* Free the entry because it isn't valid */ + drm_cleanup_buf_error(dev, entry); + kfree(temp_pagelist); + mutex_unlock(&dev->struct_mutex); + atomic_dec(&dev->buf_alloc); + return -ENOMEM; + } + dma->buflist = temp_buflist; + + for (i = 0; i < entry->buf_count; i++) { + dma->buflist[i + dma->buf_count] = &entry->buflist[i]; + } + + /* No allocations failed, so now we can replace the original pagelist + * with the new one. + */ + if (dma->page_count) { + kfree(dma->pagelist); + } + dma->pagelist = temp_pagelist; + + dma->buf_count += entry->buf_count; + dma->seg_count += entry->seg_count; + dma->page_count += entry->seg_count << page_order; + dma->byte_count += PAGE_SIZE * (entry->seg_count << page_order); + + mutex_unlock(&dev->struct_mutex); + + request->count = entry->buf_count; + request->size = size; + + if (request->flags & _DRM_PCI_BUFFER_RO) + dma->flags = _DRM_DMA_USE_PCI_RO; + + atomic_dec(&dev->buf_alloc); + return 0; + +} +EXPORT_SYMBOL(drm_addbufs_pci); + +static int drm_addbufs_sg(struct drm_device * dev, struct drm_buf_desc * request) +{ + struct drm_device_dma *dma = dev->dma; + struct drm_buf_entry *entry; + struct drm_buf *buf; + unsigned long offset; + unsigned long agp_offset; + int count; + int order; + int size; + int alignment; + int page_order; + int total; + int byte_count; + int i; + struct drm_buf **temp_buflist; + + if (!drm_core_check_feature(dev, DRIVER_SG)) + return -EINVAL; + + if (!dma) + return -EINVAL; + + if (!capable(CAP_SYS_ADMIN)) + return -EPERM; + + count = request->count; + order = drm_order(request->size); + size = 1 << order; + + alignment = (request->flags & _DRM_PAGE_ALIGN) + ? PAGE_ALIGN(size) : size; + page_order = order - PAGE_SHIFT > 0 ? order - PAGE_SHIFT : 0; + total = PAGE_SIZE << page_order; + + byte_count = 0; + agp_offset = request->agp_start; + + DRM_DEBUG("count: %d\n", count); + DRM_DEBUG("order: %d\n", order); + DRM_DEBUG("size: %d\n", size); + DRM_DEBUG("agp_offset: %lu\n", agp_offset); + DRM_DEBUG("alignment: %d\n", alignment); + DRM_DEBUG("page_order: %d\n", page_order); + DRM_DEBUG("total: %d\n", total); + + if (order < DRM_MIN_ORDER || order > DRM_MAX_ORDER) + return -EINVAL; + if (dev->queue_count) + return -EBUSY; /* Not while in use */ + + spin_lock(&dev->count_lock); + if (dev->buf_use) { + spin_unlock(&dev->count_lock); + return -EBUSY; + } + atomic_inc(&dev->buf_alloc); + spin_unlock(&dev->count_lock); + + mutex_lock(&dev->struct_mutex); + entry = &dma->bufs[order]; + if (entry->buf_count) { + mutex_unlock(&dev->struct_mutex); + atomic_dec(&dev->buf_alloc); + return -ENOMEM; /* May only call once for each order */ + } + + if (count < 0 || count > 4096) { + mutex_unlock(&dev->struct_mutex); + atomic_dec(&dev->buf_alloc); + return -EINVAL; + } + + entry->buflist = kzalloc(count * sizeof(*entry->buflist), + GFP_KERNEL); + if (!entry->buflist) { + mutex_unlock(&dev->struct_mutex); + atomic_dec(&dev->buf_alloc); + return -ENOMEM; + } + + entry->buf_size = size; + entry->page_order = page_order; + + offset = 0; + + while (entry->buf_count < count) { + buf = &entry->buflist[entry->buf_count]; + buf->idx = dma->buf_count + entry->buf_count; + buf->total = alignment; + buf->order = order; + buf->used = 0; + + buf->offset = (dma->byte_count + offset); + buf->bus_address = agp_offset + offset; + buf->address = (void *)(agp_offset + offset + + (unsigned long)dev->sg->virtual); + buf->next = NULL; + buf->waiting = 0; + buf->pending = 0; + init_waitqueue_head(&buf->dma_wait); + buf->file_priv = NULL; + + buf->dev_priv_size = dev->driver->dev_priv_size; + buf->dev_private = kzalloc(buf->dev_priv_size, GFP_KERNEL); + if (!buf->dev_private) { + /* Set count correctly so we free the proper amount. */ + entry->buf_count = count; + drm_cleanup_buf_error(dev, entry); + mutex_unlock(&dev->struct_mutex); + atomic_dec(&dev->buf_alloc); + return -ENOMEM; + } + + DRM_DEBUG("buffer %d @ %p\n", entry->buf_count, buf->address); + + offset += alignment; + entry->buf_count++; + byte_count += PAGE_SIZE << page_order; + } + + DRM_DEBUG("byte_count: %d\n", byte_count); + + temp_buflist = krealloc(dma->buflist, + (dma->buf_count + entry->buf_count) * + sizeof(*dma->buflist), GFP_KERNEL); + if (!temp_buflist) { + /* Free the entry because it isn't valid */ + drm_cleanup_buf_error(dev, entry); + mutex_unlock(&dev->struct_mutex); + atomic_dec(&dev->buf_alloc); + return -ENOMEM; + } + dma->buflist = temp_buflist; + + for (i = 0; i < entry->buf_count; i++) { + dma->buflist[i + dma->buf_count] = &entry->buflist[i]; + } + + dma->buf_count += entry->buf_count; + dma->seg_count += entry->seg_count; + dma->page_count += byte_count >> PAGE_SHIFT; + dma->byte_count += byte_count; + + DRM_DEBUG("dma->buf_count : %d\n", dma->buf_count); + DRM_DEBUG("entry->buf_count : %d\n", entry->buf_count); + + mutex_unlock(&dev->struct_mutex); + + request->count = entry->buf_count; + request->size = size; + + dma->flags = _DRM_DMA_USE_SG; + + atomic_dec(&dev->buf_alloc); + return 0; +} + +static int drm_addbufs_fb(struct drm_device * dev, struct drm_buf_desc * request) +{ + struct drm_device_dma *dma = dev->dma; + struct drm_buf_entry *entry; + struct drm_buf *buf; + unsigned long offset; + unsigned long agp_offset; + int count; + int order; + int size; + int alignment; + int page_order; + int total; + int byte_count; + int i; + struct drm_buf **temp_buflist; + + if (!drm_core_check_feature(dev, DRIVER_FB_DMA)) + return -EINVAL; + + if (!dma) + return -EINVAL; + + if (!capable(CAP_SYS_ADMIN)) + return -EPERM; + + count = request->count; + order = drm_order(request->size); + size = 1 << order; + + alignment = (request->flags & _DRM_PAGE_ALIGN) + ? PAGE_ALIGN(size) : size; + page_order = order - PAGE_SHIFT > 0 ? order - PAGE_SHIFT : 0; + total = PAGE_SIZE << page_order; + + byte_count = 0; + agp_offset = request->agp_start; + + DRM_DEBUG("count: %d\n", count); + DRM_DEBUG("order: %d\n", order); + DRM_DEBUG("size: %d\n", size); + DRM_DEBUG("agp_offset: %lu\n", agp_offset); + DRM_DEBUG("alignment: %d\n", alignment); + DRM_DEBUG("page_order: %d\n", page_order); + DRM_DEBUG("total: %d\n", total); + + if (order < DRM_MIN_ORDER || order > DRM_MAX_ORDER) + return -EINVAL; + if (dev->queue_count) + return -EBUSY; /* Not while in use */ + + spin_lock(&dev->count_lock); + if (dev->buf_use) { + spin_unlock(&dev->count_lock); + return -EBUSY; + } + atomic_inc(&dev->buf_alloc); + spin_unlock(&dev->count_lock); + + mutex_lock(&dev->struct_mutex); + entry = &dma->bufs[order]; + if (entry->buf_count) { + mutex_unlock(&dev->struct_mutex); + atomic_dec(&dev->buf_alloc); + return -ENOMEM; /* May only call once for each order */ + } + + if (count < 0 || count > 4096) { + mutex_unlock(&dev->struct_mutex); + atomic_dec(&dev->buf_alloc); + return -EINVAL; + } + + entry->buflist = kzalloc(count * sizeof(*entry->buflist), + GFP_KERNEL); + if (!entry->buflist) { + mutex_unlock(&dev->struct_mutex); + atomic_dec(&dev->buf_alloc); + return -ENOMEM; + } + + entry->buf_size = size; + entry->page_order = page_order; + + offset = 0; + + while (entry->buf_count < count) { + buf = &entry->buflist[entry->buf_count]; + buf->idx = dma->buf_count + entry->buf_count; + buf->total = alignment; + buf->order = order; + buf->used = 0; + + buf->offset = (dma->byte_count + offset); + buf->bus_address = agp_offset + offset; + buf->address = (void *)(agp_offset + offset); + buf->next = NULL; + buf->waiting = 0; + buf->pending = 0; + init_waitqueue_head(&buf->dma_wait); + buf->file_priv = NULL; + + buf->dev_priv_size = dev->driver->dev_priv_size; + buf->dev_private = kzalloc(buf->dev_priv_size, GFP_KERNEL); + if (!buf->dev_private) { + /* Set count correctly so we free the proper amount. */ + entry->buf_count = count; + drm_cleanup_buf_error(dev, entry); + mutex_unlock(&dev->struct_mutex); + atomic_dec(&dev->buf_alloc); + return -ENOMEM; + } + + DRM_DEBUG("buffer %d @ %p\n", entry->buf_count, buf->address); + + offset += alignment; + entry->buf_count++; + byte_count += PAGE_SIZE << page_order; + } + + DRM_DEBUG("byte_count: %d\n", byte_count); + + temp_buflist = krealloc(dma->buflist, + (dma->buf_count + entry->buf_count) * + sizeof(*dma->buflist), GFP_KERNEL); + if (!temp_buflist) { + /* Free the entry because it isn't valid */ + drm_cleanup_buf_error(dev, entry); + mutex_unlock(&dev->struct_mutex); + atomic_dec(&dev->buf_alloc); + return -ENOMEM; + } + dma->buflist = temp_buflist; + + for (i = 0; i < entry->buf_count; i++) { + dma->buflist[i + dma->buf_count] = &entry->buflist[i]; + } + + dma->buf_count += entry->buf_count; + dma->seg_count += entry->seg_count; + dma->page_count += byte_count >> PAGE_SHIFT; + dma->byte_count += byte_count; + + DRM_DEBUG("dma->buf_count : %d\n", dma->buf_count); + DRM_DEBUG("entry->buf_count : %d\n", entry->buf_count); + + mutex_unlock(&dev->struct_mutex); + + request->count = entry->buf_count; + request->size = size; + + dma->flags = _DRM_DMA_USE_FB; + + atomic_dec(&dev->buf_alloc); + return 0; +} + + +/** + * Add buffers for DMA transfers (ioctl). + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg pointer to a struct drm_buf_desc request. + * \return zero on success or a negative number on failure. + * + * According with the memory type specified in drm_buf_desc::flags and the + * build options, it dispatches the call either to addbufs_agp(), + * addbufs_sg() or addbufs_pci() for AGP, scatter-gather or consistent + * PCI memory respectively. + */ +int drm_addbufs(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_buf_desc *request = data; + int ret; + + if (!drm_core_check_feature(dev, DRIVER_HAVE_DMA)) + return -EINVAL; + +#if __OS_HAS_AGP + if (request->flags & _DRM_AGP_BUFFER) + ret = drm_addbufs_agp(dev, request); + else +#endif + if (request->flags & _DRM_SG_BUFFER) + ret = drm_addbufs_sg(dev, request); + else if (request->flags & _DRM_FB_BUFFER) + ret = drm_addbufs_fb(dev, request); + else + ret = drm_addbufs_pci(dev, request); + + return ret; +} + +/** + * Get information about the buffer mappings. + * + * This was originally mean for debugging purposes, or by a sophisticated + * client library to determine how best to use the available buffers (e.g., + * large buffers can be used for image transfer). + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg pointer to a drm_buf_info structure. + * \return zero on success or a negative number on failure. + * + * Increments drm_device::buf_use while holding the drm_device::count_lock + * lock, preventing of allocating more buffers after this call. Information + * about each requested buffer is then copied into user space. + */ +int drm_infobufs(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_device_dma *dma = dev->dma; + struct drm_buf_info *request = data; + int i; + int count; + + if (!drm_core_check_feature(dev, DRIVER_HAVE_DMA)) + return -EINVAL; + + if (!dma) + return -EINVAL; + + spin_lock(&dev->count_lock); + if (atomic_read(&dev->buf_alloc)) { + spin_unlock(&dev->count_lock); + return -EBUSY; + } + ++dev->buf_use; /* Can't allocate more after this call */ + spin_unlock(&dev->count_lock); + + for (i = 0, count = 0; i < DRM_MAX_ORDER + 1; i++) { + if (dma->bufs[i].buf_count) + ++count; + } + + DRM_DEBUG("count = %d\n", count); + + if (request->count >= count) { + for (i = 0, count = 0; i < DRM_MAX_ORDER + 1; i++) { + if (dma->bufs[i].buf_count) { + struct drm_buf_desc __user *to = + &request->list[count]; + struct drm_buf_entry *from = &dma->bufs[i]; + struct drm_freelist *list = &dma->bufs[i].freelist; + if (copy_to_user(&to->count, + &from->buf_count, + sizeof(from->buf_count)) || + copy_to_user(&to->size, + &from->buf_size, + sizeof(from->buf_size)) || + copy_to_user(&to->low_mark, + &list->low_mark, + sizeof(list->low_mark)) || + copy_to_user(&to->high_mark, + &list->high_mark, + sizeof(list->high_mark))) + return -EFAULT; + + DRM_DEBUG("%d %d %d %d %d\n", + i, + dma->bufs[i].buf_count, + dma->bufs[i].buf_size, + dma->bufs[i].freelist.low_mark, + dma->bufs[i].freelist.high_mark); + ++count; + } + } + } + request->count = count; + + return 0; +} + +/** + * Specifies a low and high water mark for buffer allocation + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg a pointer to a drm_buf_desc structure. + * \return zero on success or a negative number on failure. + * + * Verifies that the size order is bounded between the admissible orders and + * updates the respective drm_device_dma::bufs entry low and high water mark. + * + * \note This ioctl is deprecated and mostly never used. + */ +int drm_markbufs(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_device_dma *dma = dev->dma; + struct drm_buf_desc *request = data; + int order; + struct drm_buf_entry *entry; + + if (!drm_core_check_feature(dev, DRIVER_HAVE_DMA)) + return -EINVAL; + + if (!dma) + return -EINVAL; + + DRM_DEBUG("%d, %d, %d\n", + request->size, request->low_mark, request->high_mark); + order = drm_order(request->size); + if (order < DRM_MIN_ORDER || order > DRM_MAX_ORDER) + return -EINVAL; + entry = &dma->bufs[order]; + + if (request->low_mark < 0 || request->low_mark > entry->buf_count) + return -EINVAL; + if (request->high_mark < 0 || request->high_mark > entry->buf_count) + return -EINVAL; + + entry->freelist.low_mark = request->low_mark; + entry->freelist.high_mark = request->high_mark; + + return 0; +} + +/** + * Unreserve the buffers in list, previously reserved using drmDMA. + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg pointer to a drm_buf_free structure. + * \return zero on success or a negative number on failure. + * + * Calls free_buffer() for each used buffer. + * This function is primarily used for debugging. + */ +int drm_freebufs(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_device_dma *dma = dev->dma; + struct drm_buf_free *request = data; + int i; + int idx; + struct drm_buf *buf; + + if (!drm_core_check_feature(dev, DRIVER_HAVE_DMA)) + return -EINVAL; + + if (!dma) + return -EINVAL; + + DRM_DEBUG("%d\n", request->count); + for (i = 0; i < request->count; i++) { + if (copy_from_user(&idx, &request->list[i], sizeof(idx))) + return -EFAULT; + if (idx < 0 || idx >= dma->buf_count) { + DRM_ERROR("Index %d (of %d max)\n", + idx, dma->buf_count - 1); + return -EINVAL; + } + buf = dma->buflist[idx]; + if (buf->file_priv != file_priv) { + DRM_ERROR("Process %d freeing buffer not owned\n", + task_pid_nr(current)); + return -EINVAL; + } + drm_free_buffer(dev, buf); + } + + return 0; +} + +/** + * Maps all of the DMA buffers into client-virtual space (ioctl). + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg pointer to a drm_buf_map structure. + * \return zero on success or a negative number on failure. + * + * Maps the AGP, SG or PCI buffer region with do_mmap(), and copies information + * about each buffer into user space. For PCI buffers, it calls do_mmap() with + * offset equal to 0, which drm_mmap() interpretes as PCI buffers and calls + * drm_mmap_dma(). + */ +int drm_mapbufs(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_device_dma *dma = dev->dma; + int retcode = 0; + const int zero = 0; + unsigned long virtual; + unsigned long address; + struct drm_buf_map *request = data; + int i; + + if (!drm_core_check_feature(dev, DRIVER_HAVE_DMA)) + return -EINVAL; + + if (!dma) + return -EINVAL; + + spin_lock(&dev->count_lock); + if (atomic_read(&dev->buf_alloc)) { + spin_unlock(&dev->count_lock); + return -EBUSY; + } + dev->buf_use++; /* Can't allocate more after this call */ + spin_unlock(&dev->count_lock); + + if (request->count >= dma->buf_count) { + if ((drm_core_has_AGP(dev) && (dma->flags & _DRM_DMA_USE_AGP)) + || (drm_core_check_feature(dev, DRIVER_SG) + && (dma->flags & _DRM_DMA_USE_SG)) + || (drm_core_check_feature(dev, DRIVER_FB_DMA) + && (dma->flags & _DRM_DMA_USE_FB))) { + struct drm_local_map *map = dev->agp_buffer_map; + unsigned long token = dev->agp_buffer_token; + + if (!map) { + retcode = -EINVAL; + goto done; + } + down_write(¤t->mm->mmap_sem); + virtual = do_mmap(file_priv->filp, 0, map->size, + PROT_READ | PROT_WRITE, + MAP_SHARED, + token); + up_write(¤t->mm->mmap_sem); + } else { + down_write(¤t->mm->mmap_sem); + virtual = do_mmap(file_priv->filp, 0, dma->byte_count, + PROT_READ | PROT_WRITE, + MAP_SHARED, 0); + up_write(¤t->mm->mmap_sem); + } + if (virtual > -1024UL) { + /* Real error */ + retcode = (signed long)virtual; + goto done; + } + request->virtual = (void __user *)virtual; + + for (i = 0; i < dma->buf_count; i++) { + if (copy_to_user(&request->list[i].idx, + &dma->buflist[i]->idx, + sizeof(request->list[0].idx))) { + retcode = -EFAULT; + goto done; + } + if (copy_to_user(&request->list[i].total, + &dma->buflist[i]->total, + sizeof(request->list[0].total))) { + retcode = -EFAULT; + goto done; + } + if (copy_to_user(&request->list[i].used, + &zero, sizeof(zero))) { + retcode = -EFAULT; + goto done; + } + address = virtual + dma->buflist[i]->offset; /* *** */ + if (copy_to_user(&request->list[i].address, + &address, sizeof(address))) { + retcode = -EFAULT; + goto done; + } + } + } + done: + request->count = dma->buf_count; + DRM_DEBUG("%d buffers, retcode = %d\n", request->count, retcode); + + return retcode; +} + +/** + * Compute size order. Returns the exponent of the smaller power of two which + * is greater or equal to given number. + * + * \param size size. + * \return order. + * + * \todo Can be made faster. + */ +int drm_order(unsigned long size) +{ + int order; + unsigned long tmp; + + for (order = 0, tmp = size >> 1; tmp; tmp >>= 1, order++) ; + + if (size & (size - 1)) + ++order; + + return order; +} +EXPORT_SYMBOL(drm_order); diff --git a/drivers/gpu/drm/drm_cache.c b/drivers/gpu/drm/drm_cache.c new file mode 100644 index 00000000..0e3bd5b5 --- /dev/null +++ b/drivers/gpu/drm/drm_cache.c @@ -0,0 +1,99 @@ +/************************************************************************** + * + * Copyright (c) 2006-2007 Tungsten Graphics, Inc., Cedar Park, TX., USA + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ +/* + * Authors: Thomas Hellström + */ + +#include "drmP.h" + +#if defined(CONFIG_X86) +static void +drm_clflush_page(struct page *page) +{ + uint8_t *page_virtual; + unsigned int i; + + if (unlikely(page == NULL)) + return; + + page_virtual = kmap_atomic(page, KM_USER0); + for (i = 0; i < PAGE_SIZE; i += boot_cpu_data.x86_clflush_size) + clflush(page_virtual + i); + kunmap_atomic(page_virtual, KM_USER0); +} + +static void drm_cache_flush_clflush(struct page *pages[], + unsigned long num_pages) +{ + unsigned long i; + + mb(); + for (i = 0; i < num_pages; i++) + drm_clflush_page(*pages++); + mb(); +} + +static void +drm_clflush_ipi_handler(void *null) +{ + wbinvd(); +} +#endif + +void +drm_clflush_pages(struct page *pages[], unsigned long num_pages) +{ + +#if defined(CONFIG_X86) + if (cpu_has_clflush) { + drm_cache_flush_clflush(pages, num_pages); + return; + } + + if (on_each_cpu(drm_clflush_ipi_handler, NULL, 1) != 0) + printk(KERN_ERR "Timed out waiting for cache flush.\n"); + +#elif defined(__powerpc__) + unsigned long i; + for (i = 0; i < num_pages; i++) { + struct page *page = pages[i]; + void *page_virtual; + + if (unlikely(page == NULL)) + continue; + + page_virtual = kmap_atomic(page, KM_USER0); + flush_dcache_range((unsigned long)page_virtual, + (unsigned long)page_virtual + PAGE_SIZE); + kunmap_atomic(page_virtual, KM_USER0); + } +#else + printk(KERN_ERR "Architecture has no drm_cache.c support\n"); + WARN_ON_ONCE(1); +#endif +} +EXPORT_SYMBOL(drm_clflush_pages); diff --git a/drivers/gpu/drm/drm_context.c b/drivers/gpu/drm/drm_context.c new file mode 100644 index 00000000..6d440fb8 --- /dev/null +++ b/drivers/gpu/drm/drm_context.c @@ -0,0 +1,461 @@ +/** + * \file drm_context.c + * IOCTLs for generic contexts + * + * \author Rickard E. (Rik) Faith + * \author Gareth Hughes + */ + +/* + * Created: Fri Nov 24 18:31:37 2000 by gareth@valinux.com + * + * Copyright 1999, 2000 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * VA LINUX SYSTEMS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +/* + * ChangeLog: + * 2001-11-16 Torsten Duwe + * added context constructor/destructor hooks, + * needed by SiS driver's memory management. + */ + +#include "drmP.h" + +/******************************************************************/ +/** \name Context bitmap support */ +/*@{*/ + +/** + * Free a handle from the context bitmap. + * + * \param dev DRM device. + * \param ctx_handle context handle. + * + * Clears the bit specified by \p ctx_handle in drm_device::ctx_bitmap and the entry + * in drm_device::ctx_idr, while holding the drm_device::struct_mutex + * lock. + */ +void drm_ctxbitmap_free(struct drm_device * dev, int ctx_handle) +{ + mutex_lock(&dev->struct_mutex); + idr_remove(&dev->ctx_idr, ctx_handle); + mutex_unlock(&dev->struct_mutex); +} + +/** + * Context bitmap allocation. + * + * \param dev DRM device. + * \return (non-negative) context handle on success or a negative number on failure. + * + * Allocate a new idr from drm_device::ctx_idr while holding the + * drm_device::struct_mutex lock. + */ +static int drm_ctxbitmap_next(struct drm_device * dev) +{ + int new_id; + int ret; + +again: + if (idr_pre_get(&dev->ctx_idr, GFP_KERNEL) == 0) { + DRM_ERROR("Out of memory expanding drawable idr\n"); + return -ENOMEM; + } + mutex_lock(&dev->struct_mutex); + ret = idr_get_new_above(&dev->ctx_idr, NULL, + DRM_RESERVED_CONTEXTS, &new_id); + if (ret == -EAGAIN) { + mutex_unlock(&dev->struct_mutex); + goto again; + } + mutex_unlock(&dev->struct_mutex); + return new_id; +} + +/** + * Context bitmap initialization. + * + * \param dev DRM device. + * + * Initialise the drm_device::ctx_idr + */ +int drm_ctxbitmap_init(struct drm_device * dev) +{ + idr_init(&dev->ctx_idr); + return 0; +} + +/** + * Context bitmap cleanup. + * + * \param dev DRM device. + * + * Free all idr members using drm_ctx_sarea_free helper function + * while holding the drm_device::struct_mutex lock. + */ +void drm_ctxbitmap_cleanup(struct drm_device * dev) +{ + mutex_lock(&dev->struct_mutex); + idr_remove_all(&dev->ctx_idr); + mutex_unlock(&dev->struct_mutex); +} + +/*@}*/ + +/******************************************************************/ +/** \name Per Context SAREA Support */ +/*@{*/ + +/** + * Get per-context SAREA. + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg user argument pointing to a drm_ctx_priv_map structure. + * \return zero on success or a negative number on failure. + * + * Gets the map from drm_device::ctx_idr with the handle specified and + * returns its handle. + */ +int drm_getsareactx(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_ctx_priv_map *request = data; + struct drm_local_map *map; + struct drm_map_list *_entry; + + mutex_lock(&dev->struct_mutex); + + map = idr_find(&dev->ctx_idr, request->ctx_id); + if (!map) { + mutex_unlock(&dev->struct_mutex); + return -EINVAL; + } + + mutex_unlock(&dev->struct_mutex); + + request->handle = NULL; + list_for_each_entry(_entry, &dev->maplist, head) { + if (_entry->map == map) { + request->handle = + (void *)(unsigned long)_entry->user_token; + break; + } + } + if (request->handle == NULL) + return -EINVAL; + + return 0; +} + +/** + * Set per-context SAREA. + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg user argument pointing to a drm_ctx_priv_map structure. + * \return zero on success or a negative number on failure. + * + * Searches the mapping specified in \p arg and update the entry in + * drm_device::ctx_idr with it. + */ +int drm_setsareactx(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_ctx_priv_map *request = data; + struct drm_local_map *map = NULL; + struct drm_map_list *r_list = NULL; + + mutex_lock(&dev->struct_mutex); + list_for_each_entry(r_list, &dev->maplist, head) { + if (r_list->map + && r_list->user_token == (unsigned long) request->handle) + goto found; + } + bad: + mutex_unlock(&dev->struct_mutex); + return -EINVAL; + + found: + map = r_list->map; + if (!map) + goto bad; + + if (IS_ERR(idr_replace(&dev->ctx_idr, map, request->ctx_id))) + goto bad; + + mutex_unlock(&dev->struct_mutex); + + return 0; +} + +/*@}*/ + +/******************************************************************/ +/** \name The actual DRM context handling routines */ +/*@{*/ + +/** + * Switch context. + * + * \param dev DRM device. + * \param old old context handle. + * \param new new context handle. + * \return zero on success or a negative number on failure. + * + * Attempt to set drm_device::context_flag. + */ +static int drm_context_switch(struct drm_device * dev, int old, int new) +{ + if (test_and_set_bit(0, &dev->context_flag)) { + DRM_ERROR("Reentering -- FIXME\n"); + return -EBUSY; + } + + DRM_DEBUG("Context switch from %d to %d\n", old, new); + + if (new == dev->last_context) { + clear_bit(0, &dev->context_flag); + return 0; + } + + return 0; +} + +/** + * Complete context switch. + * + * \param dev DRM device. + * \param new new context handle. + * \return zero on success or a negative number on failure. + * + * Updates drm_device::last_context and drm_device::last_switch. Verifies the + * hardware lock is held, clears the drm_device::context_flag and wakes up + * drm_device::context_wait. + */ +static int drm_context_switch_complete(struct drm_device *dev, + struct drm_file *file_priv, int new) +{ + dev->last_context = new; /* PRE/POST: This is the _only_ writer. */ + dev->last_switch = jiffies; + + if (!_DRM_LOCK_IS_HELD(file_priv->master->lock.hw_lock->lock)) { + DRM_ERROR("Lock isn't held after context switch\n"); + } + + /* If a context switch is ever initiated + when the kernel holds the lock, release + that lock here. */ + clear_bit(0, &dev->context_flag); + wake_up(&dev->context_wait); + + return 0; +} + +/** + * Reserve contexts. + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg user argument pointing to a drm_ctx_res structure. + * \return zero on success or a negative number on failure. + */ +int drm_resctx(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_ctx_res *res = data; + struct drm_ctx ctx; + int i; + + if (res->count >= DRM_RESERVED_CONTEXTS) { + memset(&ctx, 0, sizeof(ctx)); + for (i = 0; i < DRM_RESERVED_CONTEXTS; i++) { + ctx.handle = i; + if (copy_to_user(&res->contexts[i], &ctx, sizeof(ctx))) + return -EFAULT; + } + } + res->count = DRM_RESERVED_CONTEXTS; + + return 0; +} + +/** + * Add context. + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg user argument pointing to a drm_ctx structure. + * \return zero on success or a negative number on failure. + * + * Get a new handle for the context and copy to userspace. + */ +int drm_addctx(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_ctx_list *ctx_entry; + struct drm_ctx *ctx = data; + + ctx->handle = drm_ctxbitmap_next(dev); + if (ctx->handle == DRM_KERNEL_CONTEXT) { + /* Skip kernel's context and get a new one. */ + ctx->handle = drm_ctxbitmap_next(dev); + } + DRM_DEBUG("%d\n", ctx->handle); + if (ctx->handle == -1) { + DRM_DEBUG("Not enough free contexts.\n"); + /* Should this return -EBUSY instead? */ + return -ENOMEM; + } + + ctx_entry = kmalloc(sizeof(*ctx_entry), GFP_KERNEL); + if (!ctx_entry) { + DRM_DEBUG("out of memory\n"); + return -ENOMEM; + } + + INIT_LIST_HEAD(&ctx_entry->head); + ctx_entry->handle = ctx->handle; + ctx_entry->tag = file_priv; + + mutex_lock(&dev->ctxlist_mutex); + list_add(&ctx_entry->head, &dev->ctxlist); + ++dev->ctx_count; + mutex_unlock(&dev->ctxlist_mutex); + + return 0; +} + +int drm_modctx(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + /* This does nothing */ + return 0; +} + +/** + * Get context. + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg user argument pointing to a drm_ctx structure. + * \return zero on success or a negative number on failure. + */ +int drm_getctx(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + struct drm_ctx *ctx = data; + + /* This is 0, because we don't handle any context flags */ + ctx->flags = 0; + + return 0; +} + +/** + * Switch context. + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg user argument pointing to a drm_ctx structure. + * \return zero on success or a negative number on failure. + * + * Calls context_switch(). + */ +int drm_switchctx(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_ctx *ctx = data; + + DRM_DEBUG("%d\n", ctx->handle); + return drm_context_switch(dev, dev->last_context, ctx->handle); +} + +/** + * New context. + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg user argument pointing to a drm_ctx structure. + * \return zero on success or a negative number on failure. + * + * Calls context_switch_complete(). + */ +int drm_newctx(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_ctx *ctx = data; + + DRM_DEBUG("%d\n", ctx->handle); + drm_context_switch_complete(dev, file_priv, ctx->handle); + + return 0; +} + +/** + * Remove context. + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg user argument pointing to a drm_ctx structure. + * \return zero on success or a negative number on failure. + * + * If not the special kernel context, calls ctxbitmap_free() to free the specified context. + */ +int drm_rmctx(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_ctx *ctx = data; + + DRM_DEBUG("%d\n", ctx->handle); + if (ctx->handle != DRM_KERNEL_CONTEXT) { + if (dev->driver->context_dtor) + dev->driver->context_dtor(dev, ctx->handle); + drm_ctxbitmap_free(dev, ctx->handle); + } + + mutex_lock(&dev->ctxlist_mutex); + if (!list_empty(&dev->ctxlist)) { + struct drm_ctx_list *pos, *n; + + list_for_each_entry_safe(pos, n, &dev->ctxlist, head) { + if (pos->handle == ctx->handle) { + list_del(&pos->head); + kfree(pos); + --dev->ctx_count; + } + } + } + mutex_unlock(&dev->ctxlist_mutex); + + return 0; +} + +/*@}*/ diff --git a/drivers/gpu/drm/drm_crtc.c b/drivers/gpu/drm/drm_crtc.c new file mode 100644 index 00000000..1367ced8 --- /dev/null +++ b/drivers/gpu/drm/drm_crtc.c @@ -0,0 +1,2781 @@ +/* + * Copyright (c) 2006-2008 Intel Corporation + * Copyright (c) 2007 Dave Airlie + * Copyright (c) 2008 Red Hat Inc. + * + * DRM core CRTC related functions + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that copyright + * notice and this permission notice appear in supporting documentation, and + * that the name of the copyright holders not be used in advertising or + * publicity pertaining to distribution of the software without specific, + * written prior permission. The copyright holders make no representations + * about the suitability of this software for any purpose. It is provided "as + * is" without express or implied warranty. + * + * THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, + * INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO + * EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY SPECIAL, INDIRECT OR + * CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, + * DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER + * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE + * OF THIS SOFTWARE. + * + * Authors: + * Keith Packard + * Eric Anholt + * Dave Airlie + * Jesse Barnes + */ +#include +#include +#include "drm.h" +#include "drmP.h" +#include "drm_crtc.h" +#include "drm_edid.h" + +struct drm_prop_enum_list { + int type; + char *name; +}; + +/* Avoid boilerplate. I'm tired of typing. */ +#define DRM_ENUM_NAME_FN(fnname, list) \ + char *fnname(int val) \ + { \ + int i; \ + for (i = 0; i < ARRAY_SIZE(list); i++) { \ + if (list[i].type == val) \ + return list[i].name; \ + } \ + return "(unknown)"; \ + } + +/* + * Global properties + */ +static struct drm_prop_enum_list drm_dpms_enum_list[] = +{ { DRM_MODE_DPMS_ON, "On" }, + { DRM_MODE_DPMS_STANDBY, "Standby" }, + { DRM_MODE_DPMS_SUSPEND, "Suspend" }, + { DRM_MODE_DPMS_OFF, "Off" } +}; + +DRM_ENUM_NAME_FN(drm_get_dpms_name, drm_dpms_enum_list) + +/* + * Optional properties + */ +static struct drm_prop_enum_list drm_scaling_mode_enum_list[] = +{ + { DRM_MODE_SCALE_NONE, "None" }, + { DRM_MODE_SCALE_FULLSCREEN, "Full" }, + { DRM_MODE_SCALE_CENTER, "Center" }, + { DRM_MODE_SCALE_ASPECT, "Full aspect" }, +}; + +static struct drm_prop_enum_list drm_dithering_mode_enum_list[] = +{ + { DRM_MODE_DITHERING_OFF, "Off" }, + { DRM_MODE_DITHERING_ON, "On" }, + { DRM_MODE_DITHERING_AUTO, "Automatic" }, +}; + +/* + * Non-global properties, but "required" for certain connectors. + */ +static struct drm_prop_enum_list drm_dvi_i_select_enum_list[] = +{ + { DRM_MODE_SUBCONNECTOR_Automatic, "Automatic" }, /* DVI-I and TV-out */ + { DRM_MODE_SUBCONNECTOR_DVID, "DVI-D" }, /* DVI-I */ + { DRM_MODE_SUBCONNECTOR_DVIA, "DVI-A" }, /* DVI-I */ +}; + +DRM_ENUM_NAME_FN(drm_get_dvi_i_select_name, drm_dvi_i_select_enum_list) + +static struct drm_prop_enum_list drm_dvi_i_subconnector_enum_list[] = +{ + { DRM_MODE_SUBCONNECTOR_Unknown, "Unknown" }, /* DVI-I and TV-out */ + { DRM_MODE_SUBCONNECTOR_DVID, "DVI-D" }, /* DVI-I */ + { DRM_MODE_SUBCONNECTOR_DVIA, "DVI-A" }, /* DVI-I */ +}; + +DRM_ENUM_NAME_FN(drm_get_dvi_i_subconnector_name, + drm_dvi_i_subconnector_enum_list) + +static struct drm_prop_enum_list drm_tv_select_enum_list[] = +{ + { DRM_MODE_SUBCONNECTOR_Automatic, "Automatic" }, /* DVI-I and TV-out */ + { DRM_MODE_SUBCONNECTOR_Composite, "Composite" }, /* TV-out */ + { DRM_MODE_SUBCONNECTOR_SVIDEO, "SVIDEO" }, /* TV-out */ + { DRM_MODE_SUBCONNECTOR_Component, "Component" }, /* TV-out */ + { DRM_MODE_SUBCONNECTOR_SCART, "SCART" }, /* TV-out */ +}; + +DRM_ENUM_NAME_FN(drm_get_tv_select_name, drm_tv_select_enum_list) + +static struct drm_prop_enum_list drm_tv_subconnector_enum_list[] = +{ + { DRM_MODE_SUBCONNECTOR_Unknown, "Unknown" }, /* DVI-I and TV-out */ + { DRM_MODE_SUBCONNECTOR_Composite, "Composite" }, /* TV-out */ + { DRM_MODE_SUBCONNECTOR_SVIDEO, "SVIDEO" }, /* TV-out */ + { DRM_MODE_SUBCONNECTOR_Component, "Component" }, /* TV-out */ + { DRM_MODE_SUBCONNECTOR_SCART, "SCART" }, /* TV-out */ +}; + +DRM_ENUM_NAME_FN(drm_get_tv_subconnector_name, + drm_tv_subconnector_enum_list) + +static struct drm_prop_enum_list drm_dirty_info_enum_list[] = { + { DRM_MODE_DIRTY_OFF, "Off" }, + { DRM_MODE_DIRTY_ON, "On" }, + { DRM_MODE_DIRTY_ANNOTATE, "Annotate" }, +}; + +DRM_ENUM_NAME_FN(drm_get_dirty_info_name, + drm_dirty_info_enum_list) + +struct drm_conn_prop_enum_list { + int type; + char *name; + int count; +}; + +/* + * Connector and encoder types. + */ +static struct drm_conn_prop_enum_list drm_connector_enum_list[] = +{ { DRM_MODE_CONNECTOR_Unknown, "Unknown", 0 }, + { DRM_MODE_CONNECTOR_VGA, "VGA", 0 }, + { DRM_MODE_CONNECTOR_DVII, "DVI-I", 0 }, + { DRM_MODE_CONNECTOR_DVID, "DVI-D", 0 }, + { DRM_MODE_CONNECTOR_DVIA, "DVI-A", 0 }, + { DRM_MODE_CONNECTOR_Composite, "Composite", 0 }, + { DRM_MODE_CONNECTOR_SVIDEO, "SVIDEO", 0 }, + { DRM_MODE_CONNECTOR_LVDS, "LVDS", 0 }, + { DRM_MODE_CONNECTOR_Component, "Component", 0 }, + { DRM_MODE_CONNECTOR_9PinDIN, "DIN", 0 }, + { DRM_MODE_CONNECTOR_DisplayPort, "DP", 0 }, + { DRM_MODE_CONNECTOR_HDMIA, "HDMI-A", 0 }, + { DRM_MODE_CONNECTOR_HDMIB, "HDMI-B", 0 }, + { DRM_MODE_CONNECTOR_TV, "TV", 0 }, + { DRM_MODE_CONNECTOR_eDP, "eDP", 0 }, +}; + +static struct drm_prop_enum_list drm_encoder_enum_list[] = +{ { DRM_MODE_ENCODER_NONE, "None" }, + { DRM_MODE_ENCODER_DAC, "DAC" }, + { DRM_MODE_ENCODER_TMDS, "TMDS" }, + { DRM_MODE_ENCODER_LVDS, "LVDS" }, + { DRM_MODE_ENCODER_TVDAC, "TV" }, +}; + +char *drm_get_encoder_name(struct drm_encoder *encoder) +{ + static char buf[32]; + + snprintf(buf, 32, "%s-%d", + drm_encoder_enum_list[encoder->encoder_type].name, + encoder->base.id); + return buf; +} +EXPORT_SYMBOL(drm_get_encoder_name); + +char *drm_get_connector_name(struct drm_connector *connector) +{ + static char buf[32]; + + snprintf(buf, 32, "%s-%d", + drm_connector_enum_list[connector->connector_type].name, + connector->connector_type_id); + return buf; +} +EXPORT_SYMBOL(drm_get_connector_name); + +char *drm_get_connector_status_name(enum drm_connector_status status) +{ + if (status == connector_status_connected) + return "connected"; + else if (status == connector_status_disconnected) + return "disconnected"; + else + return "unknown"; +} + +/** + * drm_mode_object_get - allocate a new identifier + * @dev: DRM device + * @ptr: object pointer, used to generate unique ID + * @type: object type + * + * LOCKING: + * + * Create a unique identifier based on @ptr in @dev's identifier space. Used + * for tracking modes, CRTCs and connectors. + * + * RETURNS: + * New unique (relative to other objects in @dev) integer identifier for the + * object. + */ +static int drm_mode_object_get(struct drm_device *dev, + struct drm_mode_object *obj, uint32_t obj_type) +{ + int new_id = 0; + int ret; + +again: + if (idr_pre_get(&dev->mode_config.crtc_idr, GFP_KERNEL) == 0) { + DRM_ERROR("Ran out memory getting a mode number\n"); + return -EINVAL; + } + + mutex_lock(&dev->mode_config.idr_mutex); + ret = idr_get_new_above(&dev->mode_config.crtc_idr, obj, 1, &new_id); + mutex_unlock(&dev->mode_config.idr_mutex); + if (ret == -EAGAIN) + goto again; + + obj->id = new_id; + obj->type = obj_type; + return 0; +} + +/** + * drm_mode_object_put - free an identifer + * @dev: DRM device + * @id: ID to free + * + * LOCKING: + * Caller must hold DRM mode_config lock. + * + * Free @id from @dev's unique identifier pool. + */ +static void drm_mode_object_put(struct drm_device *dev, + struct drm_mode_object *object) +{ + mutex_lock(&dev->mode_config.idr_mutex); + idr_remove(&dev->mode_config.crtc_idr, object->id); + mutex_unlock(&dev->mode_config.idr_mutex); +} + +struct drm_mode_object *drm_mode_object_find(struct drm_device *dev, + uint32_t id, uint32_t type) +{ + struct drm_mode_object *obj = NULL; + + mutex_lock(&dev->mode_config.idr_mutex); + obj = idr_find(&dev->mode_config.crtc_idr, id); + if (!obj || (obj->type != type) || (obj->id != id)) + obj = NULL; + mutex_unlock(&dev->mode_config.idr_mutex); + + return obj; +} +EXPORT_SYMBOL(drm_mode_object_find); + +/** + * drm_framebuffer_init - initialize a framebuffer + * @dev: DRM device + * + * LOCKING: + * Caller must hold mode config lock. + * + * Allocates an ID for the framebuffer's parent mode object, sets its mode + * functions & device file and adds it to the master fd list. + * + * RETURNS: + * Zero on success, error code on failure. + */ +int drm_framebuffer_init(struct drm_device *dev, struct drm_framebuffer *fb, + const struct drm_framebuffer_funcs *funcs) +{ + int ret; + + ret = drm_mode_object_get(dev, &fb->base, DRM_MODE_OBJECT_FB); + if (ret) { + return ret; + } + + fb->dev = dev; + fb->funcs = funcs; + dev->mode_config.num_fb++; + list_add(&fb->head, &dev->mode_config.fb_list); + + return 0; +} +EXPORT_SYMBOL(drm_framebuffer_init); + +/** + * drm_framebuffer_cleanup - remove a framebuffer object + * @fb: framebuffer to remove + * + * LOCKING: + * Caller must hold mode config lock. + * + * Scans all the CRTCs in @dev's mode_config. If they're using @fb, removes + * it, setting it to NULL. + */ +void drm_framebuffer_cleanup(struct drm_framebuffer *fb) +{ + struct drm_device *dev = fb->dev; + struct drm_crtc *crtc; + struct drm_mode_set set; + int ret; + + /* remove from any CRTC */ + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { + if (crtc->fb == fb) { + /* should turn off the crtc */ + memset(&set, 0, sizeof(struct drm_mode_set)); + set.crtc = crtc; + set.fb = NULL; + ret = crtc->funcs->set_config(&set); + if (ret) + DRM_ERROR("failed to reset crtc %p when fb was deleted\n", crtc); + } + } + + drm_mode_object_put(dev, &fb->base); + list_del(&fb->head); + dev->mode_config.num_fb--; +} +EXPORT_SYMBOL(drm_framebuffer_cleanup); + +/** + * drm_crtc_init - Initialise a new CRTC object + * @dev: DRM device + * @crtc: CRTC object to init + * @funcs: callbacks for the new CRTC + * + * LOCKING: + * Caller must hold mode config lock. + * + * Inits a new object created as base part of an driver crtc object. + */ +void drm_crtc_init(struct drm_device *dev, struct drm_crtc *crtc, + const struct drm_crtc_funcs *funcs) +{ + crtc->dev = dev; + crtc->funcs = funcs; + + mutex_lock(&dev->mode_config.mutex); + drm_mode_object_get(dev, &crtc->base, DRM_MODE_OBJECT_CRTC); + + list_add_tail(&crtc->head, &dev->mode_config.crtc_list); + dev->mode_config.num_crtc++; + mutex_unlock(&dev->mode_config.mutex); +} +EXPORT_SYMBOL(drm_crtc_init); + +/** + * drm_crtc_cleanup - Cleans up the core crtc usage. + * @crtc: CRTC to cleanup + * + * LOCKING: + * Caller must hold mode config lock. + * + * Cleanup @crtc. Removes from drm modesetting space + * does NOT free object, caller does that. + */ +void drm_crtc_cleanup(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + + if (crtc->gamma_store) { + kfree(crtc->gamma_store); + crtc->gamma_store = NULL; + } + + drm_mode_object_put(dev, &crtc->base); + list_del(&crtc->head); + dev->mode_config.num_crtc--; +} +EXPORT_SYMBOL(drm_crtc_cleanup); + +/** + * drm_mode_probed_add - add a mode to a connector's probed mode list + * @connector: connector the new mode + * @mode: mode data + * + * LOCKING: + * Caller must hold mode config lock. + * + * Add @mode to @connector's mode list for later use. + */ +void drm_mode_probed_add(struct drm_connector *connector, + struct drm_display_mode *mode) +{ + list_add(&mode->head, &connector->probed_modes); +} +EXPORT_SYMBOL(drm_mode_probed_add); + +/** + * drm_mode_remove - remove and free a mode + * @connector: connector list to modify + * @mode: mode to remove + * + * LOCKING: + * Caller must hold mode config lock. + * + * Remove @mode from @connector's mode list, then free it. + */ +void drm_mode_remove(struct drm_connector *connector, + struct drm_display_mode *mode) +{ + list_del(&mode->head); + kfree(mode); +} +EXPORT_SYMBOL(drm_mode_remove); + +/** + * drm_connector_init - Init a preallocated connector + * @dev: DRM device + * @connector: the connector to init + * @funcs: callbacks for this connector + * @name: user visible name of the connector + * + * LOCKING: + * Caller must hold @dev's mode_config lock. + * + * Initialises a preallocated connector. Connectors should be + * subclassed as part of driver connector objects. + */ +void drm_connector_init(struct drm_device *dev, + struct drm_connector *connector, + const struct drm_connector_funcs *funcs, + int connector_type) +{ + mutex_lock(&dev->mode_config.mutex); + + connector->dev = dev; + connector->funcs = funcs; + drm_mode_object_get(dev, &connector->base, DRM_MODE_OBJECT_CONNECTOR); + connector->connector_type = connector_type; + connector->connector_type_id = + ++drm_connector_enum_list[connector_type].count; /* TODO */ + INIT_LIST_HEAD(&connector->user_modes); + INIT_LIST_HEAD(&connector->probed_modes); + INIT_LIST_HEAD(&connector->modes); + connector->edid_blob_ptr = NULL; + + list_add_tail(&connector->head, &dev->mode_config.connector_list); + dev->mode_config.num_connector++; + + drm_connector_attach_property(connector, + dev->mode_config.edid_property, 0); + + drm_connector_attach_property(connector, + dev->mode_config.dpms_property, 0); + + mutex_unlock(&dev->mode_config.mutex); +} +EXPORT_SYMBOL(drm_connector_init); + +/** + * drm_connector_cleanup - cleans up an initialised connector + * @connector: connector to cleanup + * + * LOCKING: + * Caller must hold @dev's mode_config lock. + * + * Cleans up the connector but doesn't free the object. + */ +void drm_connector_cleanup(struct drm_connector *connector) +{ + struct drm_device *dev = connector->dev; + struct drm_display_mode *mode, *t; + + list_for_each_entry_safe(mode, t, &connector->probed_modes, head) + drm_mode_remove(connector, mode); + + list_for_each_entry_safe(mode, t, &connector->modes, head) + drm_mode_remove(connector, mode); + + list_for_each_entry_safe(mode, t, &connector->user_modes, head) + drm_mode_remove(connector, mode); + + mutex_lock(&dev->mode_config.mutex); + drm_mode_object_put(dev, &connector->base); + list_del(&connector->head); + mutex_unlock(&dev->mode_config.mutex); +} +EXPORT_SYMBOL(drm_connector_cleanup); + +void drm_encoder_init(struct drm_device *dev, + struct drm_encoder *encoder, + const struct drm_encoder_funcs *funcs, + int encoder_type) +{ + mutex_lock(&dev->mode_config.mutex); + + encoder->dev = dev; + + drm_mode_object_get(dev, &encoder->base, DRM_MODE_OBJECT_ENCODER); + encoder->encoder_type = encoder_type; + encoder->funcs = funcs; + + list_add_tail(&encoder->head, &dev->mode_config.encoder_list); + dev->mode_config.num_encoder++; + + mutex_unlock(&dev->mode_config.mutex); +} +EXPORT_SYMBOL(drm_encoder_init); + +void drm_encoder_cleanup(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + mutex_lock(&dev->mode_config.mutex); + drm_mode_object_put(dev, &encoder->base); + list_del(&encoder->head); + mutex_unlock(&dev->mode_config.mutex); +} +EXPORT_SYMBOL(drm_encoder_cleanup); + +/** + * drm_mode_create - create a new display mode + * @dev: DRM device + * + * LOCKING: + * Caller must hold DRM mode_config lock. + * + * Create a new drm_display_mode, give it an ID, and return it. + * + * RETURNS: + * Pointer to new mode on success, NULL on error. + */ +struct drm_display_mode *drm_mode_create(struct drm_device *dev) +{ + struct drm_display_mode *nmode; + + nmode = kzalloc(sizeof(struct drm_display_mode), GFP_KERNEL); + if (!nmode) + return NULL; + + drm_mode_object_get(dev, &nmode->base, DRM_MODE_OBJECT_MODE); + return nmode; +} +EXPORT_SYMBOL(drm_mode_create); + +/** + * drm_mode_destroy - remove a mode + * @dev: DRM device + * @mode: mode to remove + * + * LOCKING: + * Caller must hold mode config lock. + * + * Free @mode's unique identifier, then free it. + */ +void drm_mode_destroy(struct drm_device *dev, struct drm_display_mode *mode) +{ + drm_mode_object_put(dev, &mode->base); + + kfree(mode); +} +EXPORT_SYMBOL(drm_mode_destroy); + +static int drm_mode_create_standard_connector_properties(struct drm_device *dev) +{ + struct drm_property *edid; + struct drm_property *dpms; + int i; + + /* + * Standard properties (apply to all connectors) + */ + edid = drm_property_create(dev, DRM_MODE_PROP_BLOB | + DRM_MODE_PROP_IMMUTABLE, + "EDID", 0); + dev->mode_config.edid_property = edid; + + dpms = drm_property_create(dev, DRM_MODE_PROP_ENUM, + "DPMS", ARRAY_SIZE(drm_dpms_enum_list)); + for (i = 0; i < ARRAY_SIZE(drm_dpms_enum_list); i++) + drm_property_add_enum(dpms, i, drm_dpms_enum_list[i].type, + drm_dpms_enum_list[i].name); + dev->mode_config.dpms_property = dpms; + + return 0; +} + +/** + * drm_mode_create_dvi_i_properties - create DVI-I specific connector properties + * @dev: DRM device + * + * Called by a driver the first time a DVI-I connector is made. + */ +int drm_mode_create_dvi_i_properties(struct drm_device *dev) +{ + struct drm_property *dvi_i_selector; + struct drm_property *dvi_i_subconnector; + int i; + + if (dev->mode_config.dvi_i_select_subconnector_property) + return 0; + + dvi_i_selector = + drm_property_create(dev, DRM_MODE_PROP_ENUM, + "select subconnector", + ARRAY_SIZE(drm_dvi_i_select_enum_list)); + for (i = 0; i < ARRAY_SIZE(drm_dvi_i_select_enum_list); i++) + drm_property_add_enum(dvi_i_selector, i, + drm_dvi_i_select_enum_list[i].type, + drm_dvi_i_select_enum_list[i].name); + dev->mode_config.dvi_i_select_subconnector_property = dvi_i_selector; + + dvi_i_subconnector = + drm_property_create(dev, DRM_MODE_PROP_ENUM | + DRM_MODE_PROP_IMMUTABLE, + "subconnector", + ARRAY_SIZE(drm_dvi_i_subconnector_enum_list)); + for (i = 0; i < ARRAY_SIZE(drm_dvi_i_subconnector_enum_list); i++) + drm_property_add_enum(dvi_i_subconnector, i, + drm_dvi_i_subconnector_enum_list[i].type, + drm_dvi_i_subconnector_enum_list[i].name); + dev->mode_config.dvi_i_subconnector_property = dvi_i_subconnector; + + return 0; +} +EXPORT_SYMBOL(drm_mode_create_dvi_i_properties); + +/** + * drm_create_tv_properties - create TV specific connector properties + * @dev: DRM device + * @num_modes: number of different TV formats (modes) supported + * @modes: array of pointers to strings containing name of each format + * + * Called by a driver's TV initialization routine, this function creates + * the TV specific connector properties for a given device. Caller is + * responsible for allocating a list of format names and passing them to + * this routine. + */ +int drm_mode_create_tv_properties(struct drm_device *dev, int num_modes, + char *modes[]) +{ + struct drm_property *tv_selector; + struct drm_property *tv_subconnector; + int i; + + if (dev->mode_config.tv_select_subconnector_property) + return 0; + + /* + * Basic connector properties + */ + tv_selector = drm_property_create(dev, DRM_MODE_PROP_ENUM, + "select subconnector", + ARRAY_SIZE(drm_tv_select_enum_list)); + for (i = 0; i < ARRAY_SIZE(drm_tv_select_enum_list); i++) + drm_property_add_enum(tv_selector, i, + drm_tv_select_enum_list[i].type, + drm_tv_select_enum_list[i].name); + dev->mode_config.tv_select_subconnector_property = tv_selector; + + tv_subconnector = + drm_property_create(dev, DRM_MODE_PROP_ENUM | + DRM_MODE_PROP_IMMUTABLE, "subconnector", + ARRAY_SIZE(drm_tv_subconnector_enum_list)); + for (i = 0; i < ARRAY_SIZE(drm_tv_subconnector_enum_list); i++) + drm_property_add_enum(tv_subconnector, i, + drm_tv_subconnector_enum_list[i].type, + drm_tv_subconnector_enum_list[i].name); + dev->mode_config.tv_subconnector_property = tv_subconnector; + + /* + * Other, TV specific properties: margins & TV modes. + */ + dev->mode_config.tv_left_margin_property = + drm_property_create(dev, DRM_MODE_PROP_RANGE, + "left margin", 2); + dev->mode_config.tv_left_margin_property->values[0] = 0; + dev->mode_config.tv_left_margin_property->values[1] = 100; + + dev->mode_config.tv_right_margin_property = + drm_property_create(dev, DRM_MODE_PROP_RANGE, + "right margin", 2); + dev->mode_config.tv_right_margin_property->values[0] = 0; + dev->mode_config.tv_right_margin_property->values[1] = 100; + + dev->mode_config.tv_top_margin_property = + drm_property_create(dev, DRM_MODE_PROP_RANGE, + "top margin", 2); + dev->mode_config.tv_top_margin_property->values[0] = 0; + dev->mode_config.tv_top_margin_property->values[1] = 100; + + dev->mode_config.tv_bottom_margin_property = + drm_property_create(dev, DRM_MODE_PROP_RANGE, + "bottom margin", 2); + dev->mode_config.tv_bottom_margin_property->values[0] = 0; + dev->mode_config.tv_bottom_margin_property->values[1] = 100; + + dev->mode_config.tv_mode_property = + drm_property_create(dev, DRM_MODE_PROP_ENUM, + "mode", num_modes); + for (i = 0; i < num_modes; i++) + drm_property_add_enum(dev->mode_config.tv_mode_property, i, + i, modes[i]); + + dev->mode_config.tv_brightness_property = + drm_property_create(dev, DRM_MODE_PROP_RANGE, + "brightness", 2); + dev->mode_config.tv_brightness_property->values[0] = 0; + dev->mode_config.tv_brightness_property->values[1] = 100; + + dev->mode_config.tv_contrast_property = + drm_property_create(dev, DRM_MODE_PROP_RANGE, + "contrast", 2); + dev->mode_config.tv_contrast_property->values[0] = 0; + dev->mode_config.tv_contrast_property->values[1] = 100; + + dev->mode_config.tv_flicker_reduction_property = + drm_property_create(dev, DRM_MODE_PROP_RANGE, + "flicker reduction", 2); + dev->mode_config.tv_flicker_reduction_property->values[0] = 0; + dev->mode_config.tv_flicker_reduction_property->values[1] = 100; + + dev->mode_config.tv_overscan_property = + drm_property_create(dev, DRM_MODE_PROP_RANGE, + "overscan", 2); + dev->mode_config.tv_overscan_property->values[0] = 0; + dev->mode_config.tv_overscan_property->values[1] = 100; + + dev->mode_config.tv_saturation_property = + drm_property_create(dev, DRM_MODE_PROP_RANGE, + "saturation", 2); + dev->mode_config.tv_saturation_property->values[0] = 0; + dev->mode_config.tv_saturation_property->values[1] = 100; + + dev->mode_config.tv_hue_property = + drm_property_create(dev, DRM_MODE_PROP_RANGE, + "hue", 2); + dev->mode_config.tv_hue_property->values[0] = 0; + dev->mode_config.tv_hue_property->values[1] = 100; + + return 0; +} +EXPORT_SYMBOL(drm_mode_create_tv_properties); + +/** + * drm_mode_create_scaling_mode_property - create scaling mode property + * @dev: DRM device + * + * Called by a driver the first time it's needed, must be attached to desired + * connectors. + */ +int drm_mode_create_scaling_mode_property(struct drm_device *dev) +{ + struct drm_property *scaling_mode; + int i; + + if (dev->mode_config.scaling_mode_property) + return 0; + + scaling_mode = + drm_property_create(dev, DRM_MODE_PROP_ENUM, "scaling mode", + ARRAY_SIZE(drm_scaling_mode_enum_list)); + for (i = 0; i < ARRAY_SIZE(drm_scaling_mode_enum_list); i++) + drm_property_add_enum(scaling_mode, i, + drm_scaling_mode_enum_list[i].type, + drm_scaling_mode_enum_list[i].name); + + dev->mode_config.scaling_mode_property = scaling_mode; + + return 0; +} +EXPORT_SYMBOL(drm_mode_create_scaling_mode_property); + +/** + * drm_mode_create_dithering_property - create dithering property + * @dev: DRM device + * + * Called by a driver the first time it's needed, must be attached to desired + * connectors. + */ +int drm_mode_create_dithering_property(struct drm_device *dev) +{ + struct drm_property *dithering_mode; + int i; + + if (dev->mode_config.dithering_mode_property) + return 0; + + dithering_mode = + drm_property_create(dev, DRM_MODE_PROP_ENUM, "dithering", + ARRAY_SIZE(drm_dithering_mode_enum_list)); + for (i = 0; i < ARRAY_SIZE(drm_dithering_mode_enum_list); i++) + drm_property_add_enum(dithering_mode, i, + drm_dithering_mode_enum_list[i].type, + drm_dithering_mode_enum_list[i].name); + dev->mode_config.dithering_mode_property = dithering_mode; + + return 0; +} +EXPORT_SYMBOL(drm_mode_create_dithering_property); + +/** + * drm_mode_create_dirty_property - create dirty property + * @dev: DRM device + * + * Called by a driver the first time it's needed, must be attached to desired + * connectors. + */ +int drm_mode_create_dirty_info_property(struct drm_device *dev) +{ + struct drm_property *dirty_info; + int i; + + if (dev->mode_config.dirty_info_property) + return 0; + + dirty_info = + drm_property_create(dev, DRM_MODE_PROP_ENUM | + DRM_MODE_PROP_IMMUTABLE, + "dirty", + ARRAY_SIZE(drm_dirty_info_enum_list)); + for (i = 0; i < ARRAY_SIZE(drm_dirty_info_enum_list); i++) + drm_property_add_enum(dirty_info, i, + drm_dirty_info_enum_list[i].type, + drm_dirty_info_enum_list[i].name); + dev->mode_config.dirty_info_property = dirty_info; + + return 0; +} +EXPORT_SYMBOL(drm_mode_create_dirty_info_property); + +/** + * drm_mode_config_init - initialize DRM mode_configuration structure + * @dev: DRM device + * + * LOCKING: + * None, should happen single threaded at init time. + * + * Initialize @dev's mode_config structure, used for tracking the graphics + * configuration of @dev. + */ +void drm_mode_config_init(struct drm_device *dev) +{ + mutex_init(&dev->mode_config.mutex); + mutex_init(&dev->mode_config.idr_mutex); + INIT_LIST_HEAD(&dev->mode_config.fb_list); + INIT_LIST_HEAD(&dev->mode_config.crtc_list); + INIT_LIST_HEAD(&dev->mode_config.connector_list); + INIT_LIST_HEAD(&dev->mode_config.encoder_list); + INIT_LIST_HEAD(&dev->mode_config.property_list); + INIT_LIST_HEAD(&dev->mode_config.property_blob_list); + idr_init(&dev->mode_config.crtc_idr); + + mutex_lock(&dev->mode_config.mutex); + drm_mode_create_standard_connector_properties(dev); + mutex_unlock(&dev->mode_config.mutex); + + /* Just to be sure */ + dev->mode_config.num_fb = 0; + dev->mode_config.num_connector = 0; + dev->mode_config.num_crtc = 0; + dev->mode_config.num_encoder = 0; +} +EXPORT_SYMBOL(drm_mode_config_init); + +int drm_mode_group_init(struct drm_device *dev, struct drm_mode_group *group) +{ + uint32_t total_objects = 0; + + total_objects += dev->mode_config.num_crtc; + total_objects += dev->mode_config.num_connector; + total_objects += dev->mode_config.num_encoder; + + group->id_list = kzalloc(total_objects * sizeof(uint32_t), GFP_KERNEL); + if (!group->id_list) + return -ENOMEM; + + group->num_crtcs = 0; + group->num_connectors = 0; + group->num_encoders = 0; + return 0; +} + +int drm_mode_group_init_legacy_group(struct drm_device *dev, + struct drm_mode_group *group) +{ + struct drm_crtc *crtc; + struct drm_encoder *encoder; + struct drm_connector *connector; + int ret; + + if ((ret = drm_mode_group_init(dev, group))) + return ret; + + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) + group->id_list[group->num_crtcs++] = crtc->base.id; + + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) + group->id_list[group->num_crtcs + group->num_encoders++] = + encoder->base.id; + + list_for_each_entry(connector, &dev->mode_config.connector_list, head) + group->id_list[group->num_crtcs + group->num_encoders + + group->num_connectors++] = connector->base.id; + + return 0; +} + +/** + * drm_mode_config_cleanup - free up DRM mode_config info + * @dev: DRM device + * + * LOCKING: + * Caller must hold mode config lock. + * + * Free up all the connectors and CRTCs associated with this DRM device, then + * free up the framebuffers and associated buffer objects. + * + * FIXME: cleanup any dangling user buffer objects too + */ +void drm_mode_config_cleanup(struct drm_device *dev) +{ + struct drm_connector *connector, *ot; + struct drm_crtc *crtc, *ct; + struct drm_encoder *encoder, *enct; + struct drm_framebuffer *fb, *fbt; + struct drm_property *property, *pt; + + list_for_each_entry_safe(encoder, enct, &dev->mode_config.encoder_list, + head) { + encoder->funcs->destroy(encoder); + } + + list_for_each_entry_safe(connector, ot, + &dev->mode_config.connector_list, head) { + connector->funcs->destroy(connector); + } + + list_for_each_entry_safe(property, pt, &dev->mode_config.property_list, + head) { + drm_property_destroy(dev, property); + } + + list_for_each_entry_safe(fb, fbt, &dev->mode_config.fb_list, head) { + fb->funcs->destroy(fb); + } + + list_for_each_entry_safe(crtc, ct, &dev->mode_config.crtc_list, head) { + crtc->funcs->destroy(crtc); + } + +} +EXPORT_SYMBOL(drm_mode_config_cleanup); + +/** + * drm_crtc_convert_to_umode - convert a drm_display_mode into a modeinfo + * @out: drm_mode_modeinfo struct to return to the user + * @in: drm_display_mode to use + * + * LOCKING: + * None. + * + * Convert a drm_display_mode into a drm_mode_modeinfo structure to return to + * the user. + */ +void drm_crtc_convert_to_umode(struct drm_mode_modeinfo *out, + struct drm_display_mode *in) +{ + out->clock = in->clock; + out->hdisplay = in->hdisplay; + out->hsync_start = in->hsync_start; + out->hsync_end = in->hsync_end; + out->htotal = in->htotal; + out->hskew = in->hskew; + out->vdisplay = in->vdisplay; + out->vsync_start = in->vsync_start; + out->vsync_end = in->vsync_end; + out->vtotal = in->vtotal; + out->vscan = in->vscan; + out->vrefresh = in->vrefresh; + out->flags = in->flags; + out->type = in->type; + strncpy(out->name, in->name, DRM_DISPLAY_MODE_LEN); + out->name[DRM_DISPLAY_MODE_LEN-1] = 0; +} + +/** + * drm_crtc_convert_to_umode - convert a modeinfo into a drm_display_mode + * @out: drm_display_mode to return to the user + * @in: drm_mode_modeinfo to use + * + * LOCKING: + * None. + * + * Convert a drm_mode_modeinfo into a drm_display_mode structure to return to + * the caller. + */ +void drm_crtc_convert_umode(struct drm_display_mode *out, + struct drm_mode_modeinfo *in) +{ + out->clock = in->clock; + out->hdisplay = in->hdisplay; + out->hsync_start = in->hsync_start; + out->hsync_end = in->hsync_end; + out->htotal = in->htotal; + out->hskew = in->hskew; + out->vdisplay = in->vdisplay; + out->vsync_start = in->vsync_start; + out->vsync_end = in->vsync_end; + out->vtotal = in->vtotal; + out->vscan = in->vscan; + out->vrefresh = in->vrefresh; + out->flags = in->flags; + out->type = in->type; + strncpy(out->name, in->name, DRM_DISPLAY_MODE_LEN); + out->name[DRM_DISPLAY_MODE_LEN-1] = 0; +} + +/** + * drm_mode_getresources - get graphics configuration + * @inode: inode from the ioctl + * @filp: file * from the ioctl + * @cmd: cmd from ioctl + * @arg: arg from ioctl + * + * LOCKING: + * Takes mode config lock. + * + * Construct a set of configuration description structures and return + * them to the user, including CRTC, connector and framebuffer configuration. + * + * Called by the user via ioctl. + * + * RETURNS: + * Zero on success, errno on failure. + */ +int drm_mode_getresources(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_mode_card_res *card_res = data; + struct list_head *lh; + struct drm_framebuffer *fb; + struct drm_connector *connector; + struct drm_crtc *crtc; + struct drm_encoder *encoder; + int ret = 0; + int connector_count = 0; + int crtc_count = 0; + int fb_count = 0; + int encoder_count = 0; + int copied = 0, i; + uint32_t __user *fb_id; + uint32_t __user *crtc_id; + uint32_t __user *connector_id; + uint32_t __user *encoder_id; + struct drm_mode_group *mode_group; + + if (!drm_core_check_feature(dev, DRIVER_MODESET)) + return -EINVAL; + + mutex_lock(&dev->mode_config.mutex); + + /* + * For the non-control nodes we need to limit the list of resources + * by IDs in the group list for this node + */ + list_for_each(lh, &file_priv->fbs) + fb_count++; + + mode_group = &file_priv->master->minor->mode_group; + if (file_priv->master->minor->type == DRM_MINOR_CONTROL) { + + list_for_each(lh, &dev->mode_config.crtc_list) + crtc_count++; + + list_for_each(lh, &dev->mode_config.connector_list) + connector_count++; + + list_for_each(lh, &dev->mode_config.encoder_list) + encoder_count++; + } else { + + crtc_count = mode_group->num_crtcs; + connector_count = mode_group->num_connectors; + encoder_count = mode_group->num_encoders; + } + + card_res->max_height = dev->mode_config.max_height; + card_res->min_height = dev->mode_config.min_height; + card_res->max_width = dev->mode_config.max_width; + card_res->min_width = dev->mode_config.min_width; + + /* handle this in 4 parts */ + /* FBs */ + if (card_res->count_fbs >= fb_count) { + copied = 0; + fb_id = (uint32_t __user *)(unsigned long)card_res->fb_id_ptr; + list_for_each_entry(fb, &file_priv->fbs, filp_head) { + if (put_user(fb->base.id, fb_id + copied)) { + ret = -EFAULT; + goto out; + } + copied++; + } + } + card_res->count_fbs = fb_count; + + /* CRTCs */ + if (card_res->count_crtcs >= crtc_count) { + copied = 0; + crtc_id = (uint32_t __user *)(unsigned long)card_res->crtc_id_ptr; + if (file_priv->master->minor->type == DRM_MINOR_CONTROL) { + list_for_each_entry(crtc, &dev->mode_config.crtc_list, + head) { + DRM_DEBUG_KMS("[CRTC:%d]\n", crtc->base.id); + if (put_user(crtc->base.id, crtc_id + copied)) { + ret = -EFAULT; + goto out; + } + copied++; + } + } else { + for (i = 0; i < mode_group->num_crtcs; i++) { + if (put_user(mode_group->id_list[i], + crtc_id + copied)) { + ret = -EFAULT; + goto out; + } + copied++; + } + } + } + card_res->count_crtcs = crtc_count; + + /* Encoders */ + if (card_res->count_encoders >= encoder_count) { + copied = 0; + encoder_id = (uint32_t __user *)(unsigned long)card_res->encoder_id_ptr; + if (file_priv->master->minor->type == DRM_MINOR_CONTROL) { + list_for_each_entry(encoder, + &dev->mode_config.encoder_list, + head) { + DRM_DEBUG_KMS("[ENCODER:%d:%s]\n", encoder->base.id, + drm_get_encoder_name(encoder)); + if (put_user(encoder->base.id, encoder_id + + copied)) { + ret = -EFAULT; + goto out; + } + copied++; + } + } else { + for (i = mode_group->num_crtcs; i < mode_group->num_crtcs + mode_group->num_encoders; i++) { + if (put_user(mode_group->id_list[i], + encoder_id + copied)) { + ret = -EFAULT; + goto out; + } + copied++; + } + + } + } + card_res->count_encoders = encoder_count; + + /* Connectors */ + if (card_res->count_connectors >= connector_count) { + copied = 0; + connector_id = (uint32_t __user *)(unsigned long)card_res->connector_id_ptr; + if (file_priv->master->minor->type == DRM_MINOR_CONTROL) { + list_for_each_entry(connector, + &dev->mode_config.connector_list, + head) { + DRM_DEBUG_KMS("[CONNECTOR:%d:%s]\n", + connector->base.id, + drm_get_connector_name(connector)); + if (put_user(connector->base.id, + connector_id + copied)) { + ret = -EFAULT; + goto out; + } + copied++; + } + } else { + int start = mode_group->num_crtcs + + mode_group->num_encoders; + for (i = start; i < start + mode_group->num_connectors; i++) { + if (put_user(mode_group->id_list[i], + connector_id + copied)) { + ret = -EFAULT; + goto out; + } + copied++; + } + } + } + card_res->count_connectors = connector_count; + + DRM_DEBUG_KMS("CRTC[%d] CONNECTORS[%d] ENCODERS[%d]\n", card_res->count_crtcs, + card_res->count_connectors, card_res->count_encoders); + +out: + mutex_unlock(&dev->mode_config.mutex); + return ret; +} + +/** + * drm_mode_getcrtc - get CRTC configuration + * @inode: inode from the ioctl + * @filp: file * from the ioctl + * @cmd: cmd from ioctl + * @arg: arg from ioctl + * + * LOCKING: + * Caller? (FIXME) + * + * Construct a CRTC configuration structure to return to the user. + * + * Called by the user via ioctl. + * + * RETURNS: + * Zero on success, errno on failure. + */ +int drm_mode_getcrtc(struct drm_device *dev, + void *data, struct drm_file *file_priv) +{ + struct drm_mode_crtc *crtc_resp = data; + struct drm_crtc *crtc; + struct drm_mode_object *obj; + int ret = 0; + + if (!drm_core_check_feature(dev, DRIVER_MODESET)) + return -EINVAL; + + mutex_lock(&dev->mode_config.mutex); + + obj = drm_mode_object_find(dev, crtc_resp->crtc_id, + DRM_MODE_OBJECT_CRTC); + if (!obj) { + ret = -EINVAL; + goto out; + } + crtc = obj_to_crtc(obj); + + crtc_resp->x = crtc->x; + crtc_resp->y = crtc->y; + crtc_resp->gamma_size = crtc->gamma_size; + if (crtc->fb) + crtc_resp->fb_id = crtc->fb->base.id; + else + crtc_resp->fb_id = 0; + + if (crtc->enabled) { + + drm_crtc_convert_to_umode(&crtc_resp->mode, &crtc->mode); + crtc_resp->mode_valid = 1; + + } else { + crtc_resp->mode_valid = 0; + } + +out: + mutex_unlock(&dev->mode_config.mutex); + return ret; +} + +/** + * drm_mode_getconnector - get connector configuration + * @inode: inode from the ioctl + * @filp: file * from the ioctl + * @cmd: cmd from ioctl + * @arg: arg from ioctl + * + * LOCKING: + * Caller? (FIXME) + * + * Construct a connector configuration structure to return to the user. + * + * Called by the user via ioctl. + * + * RETURNS: + * Zero on success, errno on failure. + */ +int drm_mode_getconnector(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_mode_get_connector *out_resp = data; + struct drm_mode_object *obj; + struct drm_connector *connector; + struct drm_display_mode *mode; + int mode_count = 0; + int props_count = 0; + int encoders_count = 0; + int ret = 0; + int copied = 0; + int i; + struct drm_mode_modeinfo u_mode; + struct drm_mode_modeinfo __user *mode_ptr; + uint32_t __user *prop_ptr; + uint64_t __user *prop_values; + uint32_t __user *encoder_ptr; + + if (!drm_core_check_feature(dev, DRIVER_MODESET)) + return -EINVAL; + + memset(&u_mode, 0, sizeof(struct drm_mode_modeinfo)); + + DRM_DEBUG_KMS("[CONNECTOR:%d:?]\n", out_resp->connector_id); + + mutex_lock(&dev->mode_config.mutex); + + obj = drm_mode_object_find(dev, out_resp->connector_id, + DRM_MODE_OBJECT_CONNECTOR); + if (!obj) { + ret = -EINVAL; + goto out; + } + connector = obj_to_connector(obj); + + for (i = 0; i < DRM_CONNECTOR_MAX_PROPERTY; i++) { + if (connector->property_ids[i] != 0) { + props_count++; + } + } + + for (i = 0; i < DRM_CONNECTOR_MAX_ENCODER; i++) { + if (connector->encoder_ids[i] != 0) { + encoders_count++; + } + } + + if (out_resp->count_modes == 0) { + connector->funcs->fill_modes(connector, + dev->mode_config.max_width, + dev->mode_config.max_height); + } + + /* delayed so we get modes regardless of pre-fill_modes state */ + list_for_each_entry(mode, &connector->modes, head) + mode_count++; + + out_resp->connector_id = connector->base.id; + out_resp->connector_type = connector->connector_type; + out_resp->connector_type_id = connector->connector_type_id; + out_resp->mm_width = connector->display_info.width_mm; + out_resp->mm_height = connector->display_info.height_mm; + out_resp->subpixel = connector->display_info.subpixel_order; + out_resp->connection = connector->status; + if (connector->encoder) + out_resp->encoder_id = connector->encoder->base.id; + else + out_resp->encoder_id = 0; + + /* + * This ioctl is called twice, once to determine how much space is + * needed, and the 2nd time to fill it. + */ + if ((out_resp->count_modes >= mode_count) && mode_count) { + copied = 0; + mode_ptr = (struct drm_mode_modeinfo *)(unsigned long)out_resp->modes_ptr; + list_for_each_entry(mode, &connector->modes, head) { + drm_crtc_convert_to_umode(&u_mode, mode); + if (copy_to_user(mode_ptr + copied, + &u_mode, sizeof(u_mode))) { + ret = -EFAULT; + goto out; + } + copied++; + } + } + out_resp->count_modes = mode_count; + + if ((out_resp->count_props >= props_count) && props_count) { + copied = 0; + prop_ptr = (uint32_t *)(unsigned long)(out_resp->props_ptr); + prop_values = (uint64_t *)(unsigned long)(out_resp->prop_values_ptr); + for (i = 0; i < DRM_CONNECTOR_MAX_PROPERTY; i++) { + if (connector->property_ids[i] != 0) { + if (put_user(connector->property_ids[i], + prop_ptr + copied)) { + ret = -EFAULT; + goto out; + } + + if (put_user(connector->property_values[i], + prop_values + copied)) { + ret = -EFAULT; + goto out; + } + copied++; + } + } + } + out_resp->count_props = props_count; + + if ((out_resp->count_encoders >= encoders_count) && encoders_count) { + copied = 0; + encoder_ptr = (uint32_t *)(unsigned long)(out_resp->encoders_ptr); + for (i = 0; i < DRM_CONNECTOR_MAX_ENCODER; i++) { + if (connector->encoder_ids[i] != 0) { + if (put_user(connector->encoder_ids[i], + encoder_ptr + copied)) { + ret = -EFAULT; + goto out; + } + copied++; + } + } + } + out_resp->count_encoders = encoders_count; + +out: + mutex_unlock(&dev->mode_config.mutex); + return ret; +} + +int drm_mode_getencoder(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_mode_get_encoder *enc_resp = data; + struct drm_mode_object *obj; + struct drm_encoder *encoder; + int ret = 0; + + if (!drm_core_check_feature(dev, DRIVER_MODESET)) + return -EINVAL; + + mutex_lock(&dev->mode_config.mutex); + obj = drm_mode_object_find(dev, enc_resp->encoder_id, + DRM_MODE_OBJECT_ENCODER); + if (!obj) { + ret = -EINVAL; + goto out; + } + encoder = obj_to_encoder(obj); + + if (encoder->crtc) + enc_resp->crtc_id = encoder->crtc->base.id; + else + enc_resp->crtc_id = 0; + enc_resp->encoder_type = encoder->encoder_type; + enc_resp->encoder_id = encoder->base.id; + enc_resp->possible_crtcs = encoder->possible_crtcs; + enc_resp->possible_clones = encoder->possible_clones; + +out: + mutex_unlock(&dev->mode_config.mutex); + return ret; +} + +/** + * drm_mode_setcrtc - set CRTC configuration + * @inode: inode from the ioctl + * @filp: file * from the ioctl + * @cmd: cmd from ioctl + * @arg: arg from ioctl + * + * LOCKING: + * Caller? (FIXME) + * + * Build a new CRTC configuration based on user request. + * + * Called by the user via ioctl. + * + * RETURNS: + * Zero on success, errno on failure. + */ +int drm_mode_setcrtc(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_mode_config *config = &dev->mode_config; + struct drm_mode_crtc *crtc_req = data; + struct drm_mode_object *obj; + struct drm_crtc *crtc, *crtcfb; + struct drm_connector **connector_set = NULL, *connector; + struct drm_framebuffer *fb = NULL; + struct drm_display_mode *mode = NULL; + struct drm_mode_set set; + uint32_t __user *set_connectors_ptr; + int ret = 0; + int i; + + if (!drm_core_check_feature(dev, DRIVER_MODESET)) + return -EINVAL; + + mutex_lock(&dev->mode_config.mutex); + obj = drm_mode_object_find(dev, crtc_req->crtc_id, + DRM_MODE_OBJECT_CRTC); + if (!obj) { + DRM_DEBUG_KMS("Unknown CRTC ID %d\n", crtc_req->crtc_id); + ret = -EINVAL; + goto out; + } + crtc = obj_to_crtc(obj); + DRM_DEBUG_KMS("[CRTC:%d]\n", crtc->base.id); + + if (crtc_req->mode_valid) { + /* If we have a mode we need a framebuffer. */ + /* If we pass -1, set the mode with the currently bound fb */ + if (crtc_req->fb_id == -1) { + list_for_each_entry(crtcfb, + &dev->mode_config.crtc_list, head) { + if (crtcfb == crtc) { + DRM_DEBUG_KMS("Using current fb for " + "setmode\n"); + fb = crtc->fb; + } + } + } else { + obj = drm_mode_object_find(dev, crtc_req->fb_id, + DRM_MODE_OBJECT_FB); + if (!obj) { + DRM_DEBUG_KMS("Unknown FB ID%d\n", + crtc_req->fb_id); + ret = -EINVAL; + goto out; + } + fb = obj_to_fb(obj); + } + + mode = drm_mode_create(dev); + drm_crtc_convert_umode(mode, &crtc_req->mode); + drm_mode_set_crtcinfo(mode, CRTC_INTERLACE_HALVE_V); + } + + if (crtc_req->count_connectors == 0 && mode) { + DRM_DEBUG_KMS("Count connectors is 0 but mode set\n"); + ret = -EINVAL; + goto out; + } + + if (crtc_req->count_connectors > 0 && (!mode || !fb)) { + DRM_DEBUG_KMS("Count connectors is %d but no mode or fb set\n", + crtc_req->count_connectors); + ret = -EINVAL; + goto out; + } + + if (crtc_req->count_connectors > 0) { + u32 out_id; + + /* Avoid unbounded kernel memory allocation */ + if (crtc_req->count_connectors > config->num_connector) { + ret = -EINVAL; + goto out; + } + + connector_set = kmalloc(crtc_req->count_connectors * + sizeof(struct drm_connector *), + GFP_KERNEL); + if (!connector_set) { + ret = -ENOMEM; + goto out; + } + + for (i = 0; i < crtc_req->count_connectors; i++) { + set_connectors_ptr = (uint32_t *)(unsigned long)crtc_req->set_connectors_ptr; + if (get_user(out_id, &set_connectors_ptr[i])) { + ret = -EFAULT; + goto out; + } + + obj = drm_mode_object_find(dev, out_id, + DRM_MODE_OBJECT_CONNECTOR); + if (!obj) { + DRM_DEBUG_KMS("Connector id %d unknown\n", + out_id); + ret = -EINVAL; + goto out; + } + connector = obj_to_connector(obj); + DRM_DEBUG_KMS("[CONNECTOR:%d:%s]\n", + connector->base.id, + drm_get_connector_name(connector)); + + connector_set[i] = connector; + } + } + + set.crtc = crtc; + set.x = crtc_req->x; + set.y = crtc_req->y; + set.mode = mode; + set.connectors = connector_set; + set.num_connectors = crtc_req->count_connectors; + set.fb = fb; + ret = crtc->funcs->set_config(&set); + +out: + kfree(connector_set); + mutex_unlock(&dev->mode_config.mutex); + return ret; +} + +int drm_mode_cursor_ioctl(struct drm_device *dev, + void *data, struct drm_file *file_priv) +{ + struct drm_mode_cursor *req = data; + struct drm_mode_object *obj; + struct drm_crtc *crtc; + int ret = 0; + + if (!drm_core_check_feature(dev, DRIVER_MODESET)) + return -EINVAL; + + if (!req->flags) { + DRM_ERROR("no operation set\n"); + return -EINVAL; + } + + mutex_lock(&dev->mode_config.mutex); + obj = drm_mode_object_find(dev, req->crtc_id, DRM_MODE_OBJECT_CRTC); + if (!obj) { + DRM_DEBUG_KMS("Unknown CRTC ID %d\n", req->crtc_id); + ret = -EINVAL; + goto out; + } + crtc = obj_to_crtc(obj); + + if (req->flags & DRM_MODE_CURSOR_BO) { + if (!crtc->funcs->cursor_set) { + DRM_ERROR("crtc does not support cursor\n"); + ret = -ENXIO; + goto out; + } + /* Turns off the cursor if handle is 0 */ + ret = crtc->funcs->cursor_set(crtc, file_priv, req->handle, + req->width, req->height); + } + + if (req->flags & DRM_MODE_CURSOR_MOVE) { + if (crtc->funcs->cursor_move) { + ret = crtc->funcs->cursor_move(crtc, req->x, req->y); + } else { + DRM_ERROR("crtc does not support cursor\n"); + ret = -EFAULT; + goto out; + } + } +out: + mutex_unlock(&dev->mode_config.mutex); + return ret; +} + +/** + * drm_mode_addfb - add an FB to the graphics configuration + * @inode: inode from the ioctl + * @filp: file * from the ioctl + * @cmd: cmd from ioctl + * @arg: arg from ioctl + * + * LOCKING: + * Takes mode config lock. + * + * Add a new FB to the specified CRTC, given a user request. + * + * Called by the user via ioctl. + * + * RETURNS: + * Zero on success, errno on failure. + */ +int drm_mode_addfb(struct drm_device *dev, + void *data, struct drm_file *file_priv) +{ + struct drm_mode_fb_cmd *r = data; + struct drm_mode_config *config = &dev->mode_config; + struct drm_framebuffer *fb; + int ret = 0; + + if (!drm_core_check_feature(dev, DRIVER_MODESET)) + return -EINVAL; + + if ((config->min_width > r->width) || (r->width > config->max_width)) { + DRM_ERROR("mode new framebuffer width not within limits\n"); + return -EINVAL; + } + if ((config->min_height > r->height) || (r->height > config->max_height)) { + DRM_ERROR("mode new framebuffer height not within limits\n"); + return -EINVAL; + } + + mutex_lock(&dev->mode_config.mutex); + + /* TODO check buffer is sufficiently large */ + /* TODO setup destructor callback */ + + fb = dev->mode_config.funcs->fb_create(dev, file_priv, r); + if (IS_ERR(fb)) { + DRM_ERROR("could not create framebuffer\n"); + ret = PTR_ERR(fb); + goto out; + } + + r->fb_id = fb->base.id; + list_add(&fb->filp_head, &file_priv->fbs); + DRM_DEBUG_KMS("[FB:%d]\n", fb->base.id); + +out: + mutex_unlock(&dev->mode_config.mutex); + return ret; +} + +/** + * drm_mode_rmfb - remove an FB from the configuration + * @inode: inode from the ioctl + * @filp: file * from the ioctl + * @cmd: cmd from ioctl + * @arg: arg from ioctl + * + * LOCKING: + * Takes mode config lock. + * + * Remove the FB specified by the user. + * + * Called by the user via ioctl. + * + * RETURNS: + * Zero on success, errno on failure. + */ +int drm_mode_rmfb(struct drm_device *dev, + void *data, struct drm_file *file_priv) +{ + struct drm_mode_object *obj; + struct drm_framebuffer *fb = NULL; + struct drm_framebuffer *fbl = NULL; + uint32_t *id = data; + int ret = 0; + int found = 0; + + if (!drm_core_check_feature(dev, DRIVER_MODESET)) + return -EINVAL; + + mutex_lock(&dev->mode_config.mutex); + obj = drm_mode_object_find(dev, *id, DRM_MODE_OBJECT_FB); + /* TODO check that we really get a framebuffer back. */ + if (!obj) { + DRM_ERROR("mode invalid framebuffer id\n"); + ret = -EINVAL; + goto out; + } + fb = obj_to_fb(obj); + + list_for_each_entry(fbl, &file_priv->fbs, filp_head) + if (fb == fbl) + found = 1; + + if (!found) { + DRM_ERROR("tried to remove a fb that we didn't own\n"); + ret = -EINVAL; + goto out; + } + + /* TODO release all crtc connected to the framebuffer */ + /* TODO unhock the destructor from the buffer object */ + + list_del(&fb->filp_head); + fb->funcs->destroy(fb); + +out: + mutex_unlock(&dev->mode_config.mutex); + return ret; +} + +/** + * drm_mode_getfb - get FB info + * @inode: inode from the ioctl + * @filp: file * from the ioctl + * @cmd: cmd from ioctl + * @arg: arg from ioctl + * + * LOCKING: + * Caller? (FIXME) + * + * Lookup the FB given its ID and return info about it. + * + * Called by the user via ioctl. + * + * RETURNS: + * Zero on success, errno on failure. + */ +int drm_mode_getfb(struct drm_device *dev, + void *data, struct drm_file *file_priv) +{ + struct drm_mode_fb_cmd *r = data; + struct drm_mode_object *obj; + struct drm_framebuffer *fb; + int ret = 0; + + if (!drm_core_check_feature(dev, DRIVER_MODESET)) + return -EINVAL; + + mutex_lock(&dev->mode_config.mutex); + obj = drm_mode_object_find(dev, r->fb_id, DRM_MODE_OBJECT_FB); + if (!obj) { + DRM_ERROR("invalid framebuffer id\n"); + ret = -EINVAL; + goto out; + } + fb = obj_to_fb(obj); + + r->height = fb->height; + r->width = fb->width; + r->depth = fb->depth; + r->bpp = fb->bits_per_pixel; + r->pitch = fb->pitch; + fb->funcs->create_handle(fb, file_priv, &r->handle); + +out: + mutex_unlock(&dev->mode_config.mutex); + return ret; +} + +int drm_mode_dirtyfb_ioctl(struct drm_device *dev, + void *data, struct drm_file *file_priv) +{ + struct drm_clip_rect __user *clips_ptr; + struct drm_clip_rect *clips = NULL; + struct drm_mode_fb_dirty_cmd *r = data; + struct drm_mode_object *obj; + struct drm_framebuffer *fb; + unsigned flags; + int num_clips; + int ret = 0; + + if (!drm_core_check_feature(dev, DRIVER_MODESET)) + return -EINVAL; + + mutex_lock(&dev->mode_config.mutex); + obj = drm_mode_object_find(dev, r->fb_id, DRM_MODE_OBJECT_FB); + if (!obj) { + DRM_ERROR("invalid framebuffer id\n"); + ret = -EINVAL; + goto out_err1; + } + fb = obj_to_fb(obj); + + num_clips = r->num_clips; + clips_ptr = (struct drm_clip_rect *)(unsigned long)r->clips_ptr; + + if (!num_clips != !clips_ptr) { + ret = -EINVAL; + goto out_err1; + } + + flags = DRM_MODE_FB_DIRTY_FLAGS & r->flags; + + /* If userspace annotates copy, clips must come in pairs */ + if (flags & DRM_MODE_FB_DIRTY_ANNOTATE_COPY && (num_clips % 2)) { + ret = -EINVAL; + goto out_err1; + } + + if (num_clips && clips_ptr) { + if (num_clips < 0 || num_clips > DRM_MODE_FB_DIRTY_MAX_CLIPS) { + ret = -EINVAL; + goto out_err1; + } + clips = kzalloc(num_clips * sizeof(*clips), GFP_KERNEL); + if (!clips) { + ret = -ENOMEM; + goto out_err1; + } + + ret = copy_from_user(clips, clips_ptr, + num_clips * sizeof(*clips)); + if (ret) { + ret = -EFAULT; + goto out_err2; + } + } + + if (fb->funcs->dirty) { + ret = fb->funcs->dirty(fb, file_priv, flags, r->color, + clips, num_clips); + } else { + ret = -ENOSYS; + goto out_err2; + } + +out_err2: + kfree(clips); +out_err1: + mutex_unlock(&dev->mode_config.mutex); + return ret; +} + + +/** + * drm_fb_release - remove and free the FBs on this file + * @filp: file * from the ioctl + * + * LOCKING: + * Takes mode config lock. + * + * Destroy all the FBs associated with @filp. + * + * Called by the user via ioctl. + * + * RETURNS: + * Zero on success, errno on failure. + */ +void drm_fb_release(struct drm_file *priv) +{ + struct drm_device *dev = priv->minor->dev; + struct drm_framebuffer *fb, *tfb; + + mutex_lock(&dev->mode_config.mutex); + list_for_each_entry_safe(fb, tfb, &priv->fbs, filp_head) { + list_del(&fb->filp_head); + fb->funcs->destroy(fb); + } + mutex_unlock(&dev->mode_config.mutex); +} + +/** + * drm_mode_attachmode - add a mode to the user mode list + * @dev: DRM device + * @connector: connector to add the mode to + * @mode: mode to add + * + * Add @mode to @connector's user mode list. + */ +static int drm_mode_attachmode(struct drm_device *dev, + struct drm_connector *connector, + struct drm_display_mode *mode) +{ + int ret = 0; + + list_add_tail(&mode->head, &connector->user_modes); + return ret; +} + +int drm_mode_attachmode_crtc(struct drm_device *dev, struct drm_crtc *crtc, + struct drm_display_mode *mode) +{ + struct drm_connector *connector; + int ret = 0; + struct drm_display_mode *dup_mode; + int need_dup = 0; + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + if (!connector->encoder) + break; + if (connector->encoder->crtc == crtc) { + if (need_dup) + dup_mode = drm_mode_duplicate(dev, mode); + else + dup_mode = mode; + ret = drm_mode_attachmode(dev, connector, dup_mode); + if (ret) + return ret; + need_dup = 1; + } + } + return 0; +} +EXPORT_SYMBOL(drm_mode_attachmode_crtc); + +static int drm_mode_detachmode(struct drm_device *dev, + struct drm_connector *connector, + struct drm_display_mode *mode) +{ + int found = 0; + int ret = 0; + struct drm_display_mode *match_mode, *t; + + list_for_each_entry_safe(match_mode, t, &connector->user_modes, head) { + if (drm_mode_equal(match_mode, mode)) { + list_del(&match_mode->head); + drm_mode_destroy(dev, match_mode); + found = 1; + break; + } + } + + if (!found) + ret = -EINVAL; + + return ret; +} + +int drm_mode_detachmode_crtc(struct drm_device *dev, struct drm_display_mode *mode) +{ + struct drm_connector *connector; + + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + drm_mode_detachmode(dev, connector, mode); + } + return 0; +} +EXPORT_SYMBOL(drm_mode_detachmode_crtc); + +/** + * drm_fb_attachmode - Attach a user mode to an connector + * @inode: inode from the ioctl + * @filp: file * from the ioctl + * @cmd: cmd from ioctl + * @arg: arg from ioctl + * + * This attaches a user specified mode to an connector. + * Called by the user via ioctl. + * + * RETURNS: + * Zero on success, errno on failure. + */ +int drm_mode_attachmode_ioctl(struct drm_device *dev, + void *data, struct drm_file *file_priv) +{ + struct drm_mode_mode_cmd *mode_cmd = data; + struct drm_connector *connector; + struct drm_display_mode *mode; + struct drm_mode_object *obj; + struct drm_mode_modeinfo *umode = &mode_cmd->mode; + int ret = 0; + + if (!drm_core_check_feature(dev, DRIVER_MODESET)) + return -EINVAL; + + mutex_lock(&dev->mode_config.mutex); + + obj = drm_mode_object_find(dev, mode_cmd->connector_id, DRM_MODE_OBJECT_CONNECTOR); + if (!obj) { + ret = -EINVAL; + goto out; + } + connector = obj_to_connector(obj); + + mode = drm_mode_create(dev); + if (!mode) { + ret = -ENOMEM; + goto out; + } + + drm_crtc_convert_umode(mode, umode); + + ret = drm_mode_attachmode(dev, connector, mode); +out: + mutex_unlock(&dev->mode_config.mutex); + return ret; +} + + +/** + * drm_fb_detachmode - Detach a user specified mode from an connector + * @inode: inode from the ioctl + * @filp: file * from the ioctl + * @cmd: cmd from ioctl + * @arg: arg from ioctl + * + * Called by the user via ioctl. + * + * RETURNS: + * Zero on success, errno on failure. + */ +int drm_mode_detachmode_ioctl(struct drm_device *dev, + void *data, struct drm_file *file_priv) +{ + struct drm_mode_object *obj; + struct drm_mode_mode_cmd *mode_cmd = data; + struct drm_connector *connector; + struct drm_display_mode mode; + struct drm_mode_modeinfo *umode = &mode_cmd->mode; + int ret = 0; + + if (!drm_core_check_feature(dev, DRIVER_MODESET)) + return -EINVAL; + + mutex_lock(&dev->mode_config.mutex); + + obj = drm_mode_object_find(dev, mode_cmd->connector_id, DRM_MODE_OBJECT_CONNECTOR); + if (!obj) { + ret = -EINVAL; + goto out; + } + connector = obj_to_connector(obj); + + drm_crtc_convert_umode(&mode, umode); + ret = drm_mode_detachmode(dev, connector, &mode); +out: + mutex_unlock(&dev->mode_config.mutex); + return ret; +} + +struct drm_property *drm_property_create(struct drm_device *dev, int flags, + const char *name, int num_values) +{ + struct drm_property *property = NULL; + + property = kzalloc(sizeof(struct drm_property), GFP_KERNEL); + if (!property) + return NULL; + + if (num_values) { + property->values = kzalloc(sizeof(uint64_t)*num_values, GFP_KERNEL); + if (!property->values) + goto fail; + } + + drm_mode_object_get(dev, &property->base, DRM_MODE_OBJECT_PROPERTY); + property->flags = flags; + property->num_values = num_values; + INIT_LIST_HEAD(&property->enum_blob_list); + + if (name) + strncpy(property->name, name, DRM_PROP_NAME_LEN); + + list_add_tail(&property->head, &dev->mode_config.property_list); + return property; +fail: + kfree(property); + return NULL; +} +EXPORT_SYMBOL(drm_property_create); + +int drm_property_add_enum(struct drm_property *property, int index, + uint64_t value, const char *name) +{ + struct drm_property_enum *prop_enum; + + if (!(property->flags & DRM_MODE_PROP_ENUM)) + return -EINVAL; + + if (!list_empty(&property->enum_blob_list)) { + list_for_each_entry(prop_enum, &property->enum_blob_list, head) { + if (prop_enum->value == value) { + strncpy(prop_enum->name, name, DRM_PROP_NAME_LEN); + prop_enum->name[DRM_PROP_NAME_LEN-1] = '\0'; + return 0; + } + } + } + + prop_enum = kzalloc(sizeof(struct drm_property_enum), GFP_KERNEL); + if (!prop_enum) + return -ENOMEM; + + strncpy(prop_enum->name, name, DRM_PROP_NAME_LEN); + prop_enum->name[DRM_PROP_NAME_LEN-1] = '\0'; + prop_enum->value = value; + + property->values[index] = value; + list_add_tail(&prop_enum->head, &property->enum_blob_list); + return 0; +} +EXPORT_SYMBOL(drm_property_add_enum); + +void drm_property_destroy(struct drm_device *dev, struct drm_property *property) +{ + struct drm_property_enum *prop_enum, *pt; + + list_for_each_entry_safe(prop_enum, pt, &property->enum_blob_list, head) { + list_del(&prop_enum->head); + kfree(prop_enum); + } + + if (property->num_values) + kfree(property->values); + drm_mode_object_put(dev, &property->base); + list_del(&property->head); + kfree(property); +} +EXPORT_SYMBOL(drm_property_destroy); + +int drm_connector_attach_property(struct drm_connector *connector, + struct drm_property *property, uint64_t init_val) +{ + int i; + + for (i = 0; i < DRM_CONNECTOR_MAX_PROPERTY; i++) { + if (connector->property_ids[i] == 0) { + connector->property_ids[i] = property->base.id; + connector->property_values[i] = init_val; + break; + } + } + + if (i == DRM_CONNECTOR_MAX_PROPERTY) + return -EINVAL; + return 0; +} +EXPORT_SYMBOL(drm_connector_attach_property); + +int drm_connector_property_set_value(struct drm_connector *connector, + struct drm_property *property, uint64_t value) +{ + int i; + + for (i = 0; i < DRM_CONNECTOR_MAX_PROPERTY; i++) { + if (connector->property_ids[i] == property->base.id) { + connector->property_values[i] = value; + break; + } + } + + if (i == DRM_CONNECTOR_MAX_PROPERTY) + return -EINVAL; + return 0; +} +EXPORT_SYMBOL(drm_connector_property_set_value); + +int drm_connector_property_get_value(struct drm_connector *connector, + struct drm_property *property, uint64_t *val) +{ + int i; + + for (i = 0; i < DRM_CONNECTOR_MAX_PROPERTY; i++) { + if (connector->property_ids[i] == property->base.id) { + *val = connector->property_values[i]; + break; + } + } + + if (i == DRM_CONNECTOR_MAX_PROPERTY) + return -EINVAL; + return 0; +} +EXPORT_SYMBOL(drm_connector_property_get_value); + +int drm_mode_getproperty_ioctl(struct drm_device *dev, + void *data, struct drm_file *file_priv) +{ + struct drm_mode_object *obj; + struct drm_mode_get_property *out_resp = data; + struct drm_property *property; + int enum_count = 0; + int blob_count = 0; + int value_count = 0; + int ret = 0, i; + int copied; + struct drm_property_enum *prop_enum; + struct drm_mode_property_enum __user *enum_ptr; + struct drm_property_blob *prop_blob; + uint32_t *blob_id_ptr; + uint64_t __user *values_ptr; + uint32_t __user *blob_length_ptr; + + if (!drm_core_check_feature(dev, DRIVER_MODESET)) + return -EINVAL; + + mutex_lock(&dev->mode_config.mutex); + obj = drm_mode_object_find(dev, out_resp->prop_id, DRM_MODE_OBJECT_PROPERTY); + if (!obj) { + ret = -EINVAL; + goto done; + } + property = obj_to_property(obj); + + if (property->flags & DRM_MODE_PROP_ENUM) { + list_for_each_entry(prop_enum, &property->enum_blob_list, head) + enum_count++; + } else if (property->flags & DRM_MODE_PROP_BLOB) { + list_for_each_entry(prop_blob, &property->enum_blob_list, head) + blob_count++; + } + + value_count = property->num_values; + + strncpy(out_resp->name, property->name, DRM_PROP_NAME_LEN); + out_resp->name[DRM_PROP_NAME_LEN-1] = 0; + out_resp->flags = property->flags; + + if ((out_resp->count_values >= value_count) && value_count) { + values_ptr = (uint64_t *)(unsigned long)out_resp->values_ptr; + for (i = 0; i < value_count; i++) { + if (copy_to_user(values_ptr + i, &property->values[i], sizeof(uint64_t))) { + ret = -EFAULT; + goto done; + } + } + } + out_resp->count_values = value_count; + + if (property->flags & DRM_MODE_PROP_ENUM) { + if ((out_resp->count_enum_blobs >= enum_count) && enum_count) { + copied = 0; + enum_ptr = (struct drm_mode_property_enum *)(unsigned long)out_resp->enum_blob_ptr; + list_for_each_entry(prop_enum, &property->enum_blob_list, head) { + + if (copy_to_user(&enum_ptr[copied].value, &prop_enum->value, sizeof(uint64_t))) { + ret = -EFAULT; + goto done; + } + + if (copy_to_user(&enum_ptr[copied].name, + &prop_enum->name, DRM_PROP_NAME_LEN)) { + ret = -EFAULT; + goto done; + } + copied++; + } + } + out_resp->count_enum_blobs = enum_count; + } + + if (property->flags & DRM_MODE_PROP_BLOB) { + if ((out_resp->count_enum_blobs >= blob_count) && blob_count) { + copied = 0; + blob_id_ptr = (uint32_t *)(unsigned long)out_resp->enum_blob_ptr; + blob_length_ptr = (uint32_t *)(unsigned long)out_resp->values_ptr; + + list_for_each_entry(prop_blob, &property->enum_blob_list, head) { + if (put_user(prop_blob->base.id, blob_id_ptr + copied)) { + ret = -EFAULT; + goto done; + } + + if (put_user(prop_blob->length, blob_length_ptr + copied)) { + ret = -EFAULT; + goto done; + } + + copied++; + } + } + out_resp->count_enum_blobs = blob_count; + } +done: + mutex_unlock(&dev->mode_config.mutex); + return ret; +} + +static struct drm_property_blob *drm_property_create_blob(struct drm_device *dev, int length, + void *data) +{ + struct drm_property_blob *blob; + + if (!length || !data) + return NULL; + + blob = kzalloc(sizeof(struct drm_property_blob)+length, GFP_KERNEL); + if (!blob) + return NULL; + + blob->data = (void *)((char *)blob + sizeof(struct drm_property_blob)); + blob->length = length; + + memcpy(blob->data, data, length); + + drm_mode_object_get(dev, &blob->base, DRM_MODE_OBJECT_BLOB); + + list_add_tail(&blob->head, &dev->mode_config.property_blob_list); + return blob; +} + +static void drm_property_destroy_blob(struct drm_device *dev, + struct drm_property_blob *blob) +{ + drm_mode_object_put(dev, &blob->base); + list_del(&blob->head); + kfree(blob); +} + +int drm_mode_getblob_ioctl(struct drm_device *dev, + void *data, struct drm_file *file_priv) +{ + struct drm_mode_object *obj; + struct drm_mode_get_blob *out_resp = data; + struct drm_property_blob *blob; + int ret = 0; + void *blob_ptr; + + if (!drm_core_check_feature(dev, DRIVER_MODESET)) + return -EINVAL; + + mutex_lock(&dev->mode_config.mutex); + obj = drm_mode_object_find(dev, out_resp->blob_id, DRM_MODE_OBJECT_BLOB); + if (!obj) { + ret = -EINVAL; + goto done; + } + blob = obj_to_blob(obj); + + if (out_resp->length == blob->length) { + blob_ptr = (void *)(unsigned long)out_resp->data; + if (copy_to_user(blob_ptr, blob->data, blob->length)){ + ret = -EFAULT; + goto done; + } + } + out_resp->length = blob->length; + +done: + mutex_unlock(&dev->mode_config.mutex); + return ret; +} + +int drm_mode_connector_update_edid_property(struct drm_connector *connector, + struct edid *edid) +{ + struct drm_device *dev = connector->dev; + int ret = 0, size; + + if (connector->edid_blob_ptr) + drm_property_destroy_blob(dev, connector->edid_blob_ptr); + + /* Delete edid, when there is none. */ + if (!edid) { + connector->edid_blob_ptr = NULL; + ret = drm_connector_property_set_value(connector, dev->mode_config.edid_property, 0); + return ret; + } + + size = EDID_LENGTH * (1 + edid->extensions); + connector->edid_blob_ptr = drm_property_create_blob(connector->dev, + size, edid); + + ret = drm_connector_property_set_value(connector, + dev->mode_config.edid_property, + connector->edid_blob_ptr->base.id); + + return ret; +} +EXPORT_SYMBOL(drm_mode_connector_update_edid_property); + +int drm_mode_connector_property_set_ioctl(struct drm_device *dev, + void *data, struct drm_file *file_priv) +{ + struct drm_mode_connector_set_property *out_resp = data; + struct drm_mode_object *obj; + struct drm_property *property; + struct drm_connector *connector; + int ret = -EINVAL; + int i; + + if (!drm_core_check_feature(dev, DRIVER_MODESET)) + return -EINVAL; + + mutex_lock(&dev->mode_config.mutex); + + obj = drm_mode_object_find(dev, out_resp->connector_id, DRM_MODE_OBJECT_CONNECTOR); + if (!obj) { + goto out; + } + connector = obj_to_connector(obj); + + for (i = 0; i < DRM_CONNECTOR_MAX_PROPERTY; i++) { + if (connector->property_ids[i] == out_resp->prop_id) + break; + } + + if (i == DRM_CONNECTOR_MAX_PROPERTY) { + goto out; + } + + obj = drm_mode_object_find(dev, out_resp->prop_id, DRM_MODE_OBJECT_PROPERTY); + if (!obj) { + goto out; + } + property = obj_to_property(obj); + + if (property->flags & DRM_MODE_PROP_IMMUTABLE) + goto out; + + if (property->flags & DRM_MODE_PROP_RANGE) { + if (out_resp->value < property->values[0]) + goto out; + + if (out_resp->value > property->values[1]) + goto out; + } else { + int found = 0; + for (i = 0; i < property->num_values; i++) { + if (property->values[i] == out_resp->value) { + found = 1; + break; + } + } + if (!found) { + goto out; + } + } + + /* Do DPMS ourselves */ + if (property == connector->dev->mode_config.dpms_property) { + if (connector->funcs->dpms) + (*connector->funcs->dpms)(connector, (int) out_resp->value); + ret = 0; + } else if (connector->funcs->set_property) + ret = connector->funcs->set_property(connector, property, out_resp->value); + + /* store the property value if successful */ + if (!ret) + drm_connector_property_set_value(connector, property, out_resp->value); +out: + mutex_unlock(&dev->mode_config.mutex); + return ret; +} + +int drm_mode_connector_attach_encoder(struct drm_connector *connector, + struct drm_encoder *encoder) +{ + int i; + + for (i = 0; i < DRM_CONNECTOR_MAX_ENCODER; i++) { + if (connector->encoder_ids[i] == 0) { + connector->encoder_ids[i] = encoder->base.id; + return 0; + } + } + return -ENOMEM; +} +EXPORT_SYMBOL(drm_mode_connector_attach_encoder); + +void drm_mode_connector_detach_encoder(struct drm_connector *connector, + struct drm_encoder *encoder) +{ + int i; + for (i = 0; i < DRM_CONNECTOR_MAX_ENCODER; i++) { + if (connector->encoder_ids[i] == encoder->base.id) { + connector->encoder_ids[i] = 0; + if (connector->encoder == encoder) + connector->encoder = NULL; + break; + } + } +} +EXPORT_SYMBOL(drm_mode_connector_detach_encoder); + +bool drm_mode_crtc_set_gamma_size(struct drm_crtc *crtc, + int gamma_size) +{ + crtc->gamma_size = gamma_size; + + crtc->gamma_store = kzalloc(gamma_size * sizeof(uint16_t) * 3, GFP_KERNEL); + if (!crtc->gamma_store) { + crtc->gamma_size = 0; + return false; + } + + return true; +} +EXPORT_SYMBOL(drm_mode_crtc_set_gamma_size); + +int drm_mode_gamma_set_ioctl(struct drm_device *dev, + void *data, struct drm_file *file_priv) +{ + struct drm_mode_crtc_lut *crtc_lut = data; + struct drm_mode_object *obj; + struct drm_crtc *crtc; + void *r_base, *g_base, *b_base; + int size; + int ret = 0; + + if (!drm_core_check_feature(dev, DRIVER_MODESET)) + return -EINVAL; + + mutex_lock(&dev->mode_config.mutex); + obj = drm_mode_object_find(dev, crtc_lut->crtc_id, DRM_MODE_OBJECT_CRTC); + if (!obj) { + ret = -EINVAL; + goto out; + } + crtc = obj_to_crtc(obj); + + /* memcpy into gamma store */ + if (crtc_lut->gamma_size != crtc->gamma_size) { + ret = -EINVAL; + goto out; + } + + size = crtc_lut->gamma_size * (sizeof(uint16_t)); + r_base = crtc->gamma_store; + if (copy_from_user(r_base, (void __user *)(unsigned long)crtc_lut->red, size)) { + ret = -EFAULT; + goto out; + } + + g_base = r_base + size; + if (copy_from_user(g_base, (void __user *)(unsigned long)crtc_lut->green, size)) { + ret = -EFAULT; + goto out; + } + + b_base = g_base + size; + if (copy_from_user(b_base, (void __user *)(unsigned long)crtc_lut->blue, size)) { + ret = -EFAULT; + goto out; + } + + crtc->funcs->gamma_set(crtc, r_base, g_base, b_base, 0, crtc->gamma_size); + +out: + mutex_unlock(&dev->mode_config.mutex); + return ret; + +} + +int drm_mode_gamma_get_ioctl(struct drm_device *dev, + void *data, struct drm_file *file_priv) +{ + struct drm_mode_crtc_lut *crtc_lut = data; + struct drm_mode_object *obj; + struct drm_crtc *crtc; + void *r_base, *g_base, *b_base; + int size; + int ret = 0; + + if (!drm_core_check_feature(dev, DRIVER_MODESET)) + return -EINVAL; + + mutex_lock(&dev->mode_config.mutex); + obj = drm_mode_object_find(dev, crtc_lut->crtc_id, DRM_MODE_OBJECT_CRTC); + if (!obj) { + ret = -EINVAL; + goto out; + } + crtc = obj_to_crtc(obj); + + /* memcpy into gamma store */ + if (crtc_lut->gamma_size != crtc->gamma_size) { + ret = -EINVAL; + goto out; + } + + size = crtc_lut->gamma_size * (sizeof(uint16_t)); + r_base = crtc->gamma_store; + if (copy_to_user((void __user *)(unsigned long)crtc_lut->red, r_base, size)) { + ret = -EFAULT; + goto out; + } + + g_base = r_base + size; + if (copy_to_user((void __user *)(unsigned long)crtc_lut->green, g_base, size)) { + ret = -EFAULT; + goto out; + } + + b_base = g_base + size; + if (copy_to_user((void __user *)(unsigned long)crtc_lut->blue, b_base, size)) { + ret = -EFAULT; + goto out; + } +out: + mutex_unlock(&dev->mode_config.mutex); + return ret; +} + +int drm_mode_page_flip_ioctl(struct drm_device *dev, + void *data, struct drm_file *file_priv) +{ + struct drm_mode_crtc_page_flip *page_flip = data; + struct drm_mode_object *obj; + struct drm_crtc *crtc; + struct drm_framebuffer *fb; + struct drm_pending_vblank_event *e = NULL; + unsigned long flags; + int ret = -EINVAL; + + if (page_flip->flags & ~DRM_MODE_PAGE_FLIP_FLAGS || + page_flip->reserved != 0) + return -EINVAL; + + mutex_lock(&dev->mode_config.mutex); + obj = drm_mode_object_find(dev, page_flip->crtc_id, DRM_MODE_OBJECT_CRTC); + if (!obj) + goto out; + crtc = obj_to_crtc(obj); + + if (crtc->fb == NULL) { + /* The framebuffer is currently unbound, presumably + * due to a hotplug event, that userspace has not + * yet discovered. + */ + ret = -EBUSY; + goto out; + } + + if (crtc->funcs->page_flip == NULL) + goto out; + + obj = drm_mode_object_find(dev, page_flip->fb_id, DRM_MODE_OBJECT_FB); + if (!obj) + goto out; + fb = obj_to_fb(obj); + + if (page_flip->flags & DRM_MODE_PAGE_FLIP_EVENT) { + ret = -ENOMEM; + spin_lock_irqsave(&dev->event_lock, flags); + if (file_priv->event_space < sizeof e->event) { + spin_unlock_irqrestore(&dev->event_lock, flags); + goto out; + } + file_priv->event_space -= sizeof e->event; + spin_unlock_irqrestore(&dev->event_lock, flags); + + e = kzalloc(sizeof *e, GFP_KERNEL); + if (e == NULL) { + spin_lock_irqsave(&dev->event_lock, flags); + file_priv->event_space += sizeof e->event; + spin_unlock_irqrestore(&dev->event_lock, flags); + goto out; + } + + e->event.base.type = DRM_EVENT_FLIP_COMPLETE; + e->event.base.length = sizeof e->event; + e->event.user_data = page_flip->user_data; + e->base.event = &e->event.base; + e->base.file_priv = file_priv; + e->base.destroy = + (void (*) (struct drm_pending_event *)) kfree; + } + + ret = crtc->funcs->page_flip(crtc, fb, e); + if (ret) { + spin_lock_irqsave(&dev->event_lock, flags); + file_priv->event_space += sizeof e->event; + spin_unlock_irqrestore(&dev->event_lock, flags); + kfree(e); + } + +out: + mutex_unlock(&dev->mode_config.mutex); + return ret; +} + +void drm_mode_config_reset(struct drm_device *dev) +{ + struct drm_crtc *crtc; + struct drm_encoder *encoder; + struct drm_connector *connector; + + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) + if (crtc->funcs->reset) + crtc->funcs->reset(crtc); + + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) + if (encoder->funcs->reset) + encoder->funcs->reset(encoder); + + list_for_each_entry(connector, &dev->mode_config.connector_list, head) + if (connector->funcs->reset) + connector->funcs->reset(connector); +} +EXPORT_SYMBOL(drm_mode_config_reset); + +int drm_mode_create_dumb_ioctl(struct drm_device *dev, + void *data, struct drm_file *file_priv) +{ + struct drm_mode_create_dumb *args = data; + + if (!dev->driver->dumb_create) + return -ENOSYS; + return dev->driver->dumb_create(file_priv, dev, args); +} + +int drm_mode_mmap_dumb_ioctl(struct drm_device *dev, + void *data, struct drm_file *file_priv) +{ + struct drm_mode_map_dumb *args = data; + + /* call driver ioctl to get mmap offset */ + if (!dev->driver->dumb_map_offset) + return -ENOSYS; + + return dev->driver->dumb_map_offset(file_priv, dev, args->handle, &args->offset); +} + +int drm_mode_destroy_dumb_ioctl(struct drm_device *dev, + void *data, struct drm_file *file_priv) +{ + struct drm_mode_destroy_dumb *args = data; + + if (!dev->driver->dumb_destroy) + return -ENOSYS; + + return dev->driver->dumb_destroy(file_priv, dev, args->handle); +} diff --git a/drivers/gpu/drm/drm_crtc_helper.c b/drivers/gpu/drm/drm_crtc_helper.c new file mode 100644 index 00000000..92369655 --- /dev/null +++ b/drivers/gpu/drm/drm_crtc_helper.c @@ -0,0 +1,964 @@ +/* + * Copyright (c) 2006-2008 Intel Corporation + * Copyright (c) 2007 Dave Airlie + * + * DRM core CRTC related functions + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that copyright + * notice and this permission notice appear in supporting documentation, and + * that the name of the copyright holders not be used in advertising or + * publicity pertaining to distribution of the software without specific, + * written prior permission. The copyright holders make no representations + * about the suitability of this software for any purpose. It is provided "as + * is" without express or implied warranty. + * + * THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, + * INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO + * EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY SPECIAL, INDIRECT OR + * CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, + * DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER + * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE + * OF THIS SOFTWARE. + * + * Authors: + * Keith Packard + * Eric Anholt + * Dave Airlie + * Jesse Barnes + */ + +#include "drmP.h" +#include "drm_crtc.h" +#include "drm_crtc_helper.h" +#include "drm_fb_helper.h" + +static bool drm_kms_helper_poll = true; +module_param_named(poll, drm_kms_helper_poll, bool, 0600); + +static void drm_mode_validate_flag(struct drm_connector *connector, + int flags) +{ + struct drm_display_mode *mode, *t; + + if (flags == (DRM_MODE_FLAG_DBLSCAN | DRM_MODE_FLAG_INTERLACE)) + return; + + list_for_each_entry_safe(mode, t, &connector->modes, head) { + if ((mode->flags & DRM_MODE_FLAG_INTERLACE) && + !(flags & DRM_MODE_FLAG_INTERLACE)) + mode->status = MODE_NO_INTERLACE; + if ((mode->flags & DRM_MODE_FLAG_DBLSCAN) && + !(flags & DRM_MODE_FLAG_DBLSCAN)) + mode->status = MODE_NO_DBLESCAN; + } + + return; +} + +/** + * drm_helper_probe_single_connector_modes - get complete set of display modes + * @dev: DRM device + * @maxX: max width for modes + * @maxY: max height for modes + * + * LOCKING: + * Caller must hold mode config lock. + * + * Based on @dev's mode_config layout, scan all the connectors and try to detect + * modes on them. Modes will first be added to the connector's probed_modes + * list, then culled (based on validity and the @maxX, @maxY parameters) and + * put into the normal modes list. + * + * Intended to be used either at bootup time or when major configuration + * changes have occurred. + * + * FIXME: take into account monitor limits + * + * RETURNS: + * Number of modes found on @connector. + */ +int drm_helper_probe_single_connector_modes(struct drm_connector *connector, + uint32_t maxX, uint32_t maxY) +{ + struct drm_device *dev = connector->dev; + struct drm_display_mode *mode, *t; + struct drm_connector_helper_funcs *connector_funcs = + connector->helper_private; + int count = 0; + int mode_flags = 0; + + DRM_DEBUG_KMS("[CONNECTOR:%d:%s]\n", connector->base.id, + drm_get_connector_name(connector)); + /* set all modes to the unverified state */ + list_for_each_entry_safe(mode, t, &connector->modes, head) + mode->status = MODE_UNVERIFIED; + + if (connector->force) { + if (connector->force == DRM_FORCE_ON) + connector->status = connector_status_connected; + else + connector->status = connector_status_disconnected; + if (connector->funcs->force) + connector->funcs->force(connector); + } else { + connector->status = connector->funcs->detect(connector, true); + drm_kms_helper_poll_enable(dev); + } + + if (connector->status == connector_status_disconnected) { + DRM_DEBUG_KMS("[CONNECTOR:%d:%s] disconnected\n", + connector->base.id, drm_get_connector_name(connector)); + drm_mode_connector_update_edid_property(connector, NULL); + goto prune; + } + + count = (*connector_funcs->get_modes)(connector); + if (count == 0 && connector->status == connector_status_connected) + count = drm_add_modes_noedid(connector, 1024, 768); + if (count == 0) + goto prune; + + drm_mode_connector_list_update(connector); + + if (maxX && maxY) + drm_mode_validate_size(dev, &connector->modes, maxX, + maxY, 0); + + if (connector->interlace_allowed) + mode_flags |= DRM_MODE_FLAG_INTERLACE; + if (connector->doublescan_allowed) + mode_flags |= DRM_MODE_FLAG_DBLSCAN; + drm_mode_validate_flag(connector, mode_flags); + + list_for_each_entry_safe(mode, t, &connector->modes, head) { + if (mode->status == MODE_OK) + mode->status = connector_funcs->mode_valid(connector, + mode); + } + +prune: + drm_mode_prune_invalid(dev, &connector->modes, true); + + if (list_empty(&connector->modes)) + return 0; + + drm_mode_sort(&connector->modes); + + DRM_DEBUG_KMS("[CONNECTOR:%d:%s] probed modes :\n", connector->base.id, + drm_get_connector_name(connector)); + list_for_each_entry_safe(mode, t, &connector->modes, head) { + mode->vrefresh = drm_mode_vrefresh(mode); + + drm_mode_set_crtcinfo(mode, CRTC_INTERLACE_HALVE_V); + drm_mode_debug_printmodeline(mode); + } + + return count; +} +EXPORT_SYMBOL(drm_helper_probe_single_connector_modes); + +/** + * drm_helper_encoder_in_use - check if a given encoder is in use + * @encoder: encoder to check + * + * LOCKING: + * Caller must hold mode config lock. + * + * Walk @encoders's DRM device's mode_config and see if it's in use. + * + * RETURNS: + * True if @encoder is part of the mode_config, false otherwise. + */ +bool drm_helper_encoder_in_use(struct drm_encoder *encoder) +{ + struct drm_connector *connector; + struct drm_device *dev = encoder->dev; + list_for_each_entry(connector, &dev->mode_config.connector_list, head) + if (connector->encoder == encoder) + return true; + return false; +} +EXPORT_SYMBOL(drm_helper_encoder_in_use); + +/** + * drm_helper_crtc_in_use - check if a given CRTC is in a mode_config + * @crtc: CRTC to check + * + * LOCKING: + * Caller must hold mode config lock. + * + * Walk @crtc's DRM device's mode_config and see if it's in use. + * + * RETURNS: + * True if @crtc is part of the mode_config, false otherwise. + */ +bool drm_helper_crtc_in_use(struct drm_crtc *crtc) +{ + struct drm_encoder *encoder; + struct drm_device *dev = crtc->dev; + /* FIXME: Locking around list access? */ + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) + if (encoder->crtc == crtc && drm_helper_encoder_in_use(encoder)) + return true; + return false; +} +EXPORT_SYMBOL(drm_helper_crtc_in_use); + +static void +drm_encoder_disable(struct drm_encoder *encoder) +{ + struct drm_encoder_helper_funcs *encoder_funcs = encoder->helper_private; + + if (encoder_funcs->disable) + (*encoder_funcs->disable)(encoder); + else + (*encoder_funcs->dpms)(encoder, DRM_MODE_DPMS_OFF); +} + +/** + * drm_helper_disable_unused_functions - disable unused objects + * @dev: DRM device + * + * LOCKING: + * Caller must hold mode config lock. + * + * If an connector or CRTC isn't part of @dev's mode_config, it can be disabled + * by calling its dpms function, which should power it off. + */ +void drm_helper_disable_unused_functions(struct drm_device *dev) +{ + struct drm_encoder *encoder; + struct drm_connector *connector; + struct drm_crtc *crtc; + + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + if (!connector->encoder) + continue; + if (connector->status == connector_status_disconnected) + connector->encoder = NULL; + } + + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + if (!drm_helper_encoder_in_use(encoder)) { + drm_encoder_disable(encoder); + /* disconnector encoder from any connector */ + encoder->crtc = NULL; + } + } + + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { + struct drm_crtc_helper_funcs *crtc_funcs = crtc->helper_private; + crtc->enabled = drm_helper_crtc_in_use(crtc); + if (!crtc->enabled) { + if (crtc_funcs->disable) + (*crtc_funcs->disable)(crtc); + else + (*crtc_funcs->dpms)(crtc, DRM_MODE_DPMS_OFF); + crtc->fb = NULL; + } + } +} +EXPORT_SYMBOL(drm_helper_disable_unused_functions); + +/** + * drm_encoder_crtc_ok - can a given crtc drive a given encoder? + * @encoder: encoder to test + * @crtc: crtc to test + * + * Return false if @encoder can't be driven by @crtc, true otherwise. + */ +static bool drm_encoder_crtc_ok(struct drm_encoder *encoder, + struct drm_crtc *crtc) +{ + struct drm_device *dev; + struct drm_crtc *tmp; + int crtc_mask = 1; + + WARN(!crtc, "checking null crtc?\n"); + + dev = crtc->dev; + + list_for_each_entry(tmp, &dev->mode_config.crtc_list, head) { + if (tmp == crtc) + break; + crtc_mask <<= 1; + } + + if (encoder->possible_crtcs & crtc_mask) + return true; + return false; +} + +/* + * Check the CRTC we're going to map each output to vs. its current + * CRTC. If they don't match, we have to disable the output and the CRTC + * since the driver will have to re-route things. + */ +static void +drm_crtc_prepare_encoders(struct drm_device *dev) +{ + struct drm_encoder_helper_funcs *encoder_funcs; + struct drm_encoder *encoder; + + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + encoder_funcs = encoder->helper_private; + /* Disable unused encoders */ + if (encoder->crtc == NULL) + drm_encoder_disable(encoder); + /* Disable encoders whose CRTC is about to change */ + if (encoder_funcs->get_crtc && + encoder->crtc != (*encoder_funcs->get_crtc)(encoder)) + drm_encoder_disable(encoder); + } +} + +/** + * drm_crtc_set_mode - set a mode + * @crtc: CRTC to program + * @mode: mode to use + * @x: width of mode + * @y: height of mode + * + * LOCKING: + * Caller must hold mode config lock. + * + * Try to set @mode on @crtc. Give @crtc and its associated connectors a chance + * to fixup or reject the mode prior to trying to set it. + * + * RETURNS: + * True if the mode was set successfully, or false otherwise. + */ +bool drm_crtc_helper_set_mode(struct drm_crtc *crtc, + struct drm_display_mode *mode, + int x, int y, + struct drm_framebuffer *old_fb) +{ + struct drm_device *dev = crtc->dev; + struct drm_display_mode *adjusted_mode, saved_mode, saved_hwmode; + struct drm_crtc_helper_funcs *crtc_funcs = crtc->helper_private; + struct drm_encoder_helper_funcs *encoder_funcs; + int saved_x, saved_y; + struct drm_encoder *encoder; + bool ret = true; + + crtc->enabled = drm_helper_crtc_in_use(crtc); + if (!crtc->enabled) + return true; + + adjusted_mode = drm_mode_duplicate(dev, mode); + + saved_hwmode = crtc->hwmode; + saved_mode = crtc->mode; + saved_x = crtc->x; + saved_y = crtc->y; + + /* Update crtc values up front so the driver can rely on them for mode + * setting. + */ + crtc->mode = *mode; + crtc->x = x; + crtc->y = y; + + /* Pass our mode to the connectors and the CRTC to give them a chance to + * adjust it according to limitations or connector properties, and also + * a chance to reject the mode entirely. + */ + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + + if (encoder->crtc != crtc) + continue; + encoder_funcs = encoder->helper_private; + if (!(ret = encoder_funcs->mode_fixup(encoder, mode, + adjusted_mode))) { + goto done; + } + } + + if (!(ret = crtc_funcs->mode_fixup(crtc, mode, adjusted_mode))) { + goto done; + } + DRM_DEBUG_KMS("[CRTC:%d]\n", crtc->base.id); + + /* Prepare the encoders and CRTCs before setting the mode. */ + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + + if (encoder->crtc != crtc) + continue; + encoder_funcs = encoder->helper_private; + /* Disable the encoders as the first thing we do. */ + encoder_funcs->prepare(encoder); + } + + drm_crtc_prepare_encoders(dev); + + crtc_funcs->prepare(crtc); + + /* Set up the DPLL and any encoders state that needs to adjust or depend + * on the DPLL. + */ + ret = !crtc_funcs->mode_set(crtc, mode, adjusted_mode, x, y, old_fb); + if (!ret) + goto done; + + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + + if (encoder->crtc != crtc) + continue; + + DRM_DEBUG_KMS("[ENCODER:%d:%s] set [MODE:%d:%s]\n", + encoder->base.id, drm_get_encoder_name(encoder), + mode->base.id, mode->name); + encoder_funcs = encoder->helper_private; + encoder_funcs->mode_set(encoder, mode, adjusted_mode); + } + + /* Now enable the clocks, plane, pipe, and connectors that we set up. */ + crtc_funcs->commit(crtc); + + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + + if (encoder->crtc != crtc) + continue; + + encoder_funcs = encoder->helper_private; + encoder_funcs->commit(encoder); + + } + + /* Store real post-adjustment hardware mode. */ + crtc->hwmode = *adjusted_mode; + + /* Calculate and store various constants which + * are later needed by vblank and swap-completion + * timestamping. They are derived from true hwmode. + */ + drm_calc_timestamping_constants(crtc); + + /* FIXME: add subpixel order */ +done: + drm_mode_destroy(dev, adjusted_mode); + if (!ret) { + crtc->hwmode = saved_hwmode; + crtc->mode = saved_mode; + crtc->x = saved_x; + crtc->y = saved_y; + } + + return ret; +} +EXPORT_SYMBOL(drm_crtc_helper_set_mode); + + +/** + * drm_crtc_helper_set_config - set a new config from userspace + * @crtc: CRTC to setup + * @crtc_info: user provided configuration + * @new_mode: new mode to set + * @connector_set: set of connectors for the new config + * @fb: new framebuffer + * + * LOCKING: + * Caller must hold mode config lock. + * + * Setup a new configuration, provided by the user in @crtc_info, and enable + * it. + * + * RETURNS: + * Zero. (FIXME) + */ +int drm_crtc_helper_set_config(struct drm_mode_set *set) +{ + struct drm_device *dev; + struct drm_crtc *save_crtcs, *new_crtc, *crtc; + struct drm_encoder *save_encoders, *new_encoder, *encoder; + struct drm_framebuffer *old_fb = NULL; + bool mode_changed = false; /* if true do a full mode set */ + bool fb_changed = false; /* if true and !mode_changed just do a flip */ + struct drm_connector *save_connectors, *connector; + int count = 0, ro, fail = 0; + struct drm_crtc_helper_funcs *crtc_funcs; + int ret = 0; + int i; + + DRM_DEBUG_KMS("\n"); + + if (!set) + return -EINVAL; + + if (!set->crtc) + return -EINVAL; + + if (!set->crtc->helper_private) + return -EINVAL; + + crtc_funcs = set->crtc->helper_private; + + if (!set->mode) + set->fb = NULL; + + if (set->fb) { + DRM_DEBUG_KMS("[CRTC:%d] [FB:%d] #connectors=%d (x y) (%i %i)\n", + set->crtc->base.id, set->fb->base.id, + (int)set->num_connectors, set->x, set->y); + } else { + DRM_DEBUG_KMS("[CRTC:%d] [NOFB]\n", set->crtc->base.id); + set->mode = NULL; + set->num_connectors = 0; + } + + dev = set->crtc->dev; + + /* Allocate space for the backup of all (non-pointer) crtc, encoder and + * connector data. */ + save_crtcs = kzalloc(dev->mode_config.num_crtc * + sizeof(struct drm_crtc), GFP_KERNEL); + if (!save_crtcs) + return -ENOMEM; + + save_encoders = kzalloc(dev->mode_config.num_encoder * + sizeof(struct drm_encoder), GFP_KERNEL); + if (!save_encoders) { + kfree(save_crtcs); + return -ENOMEM; + } + + save_connectors = kzalloc(dev->mode_config.num_connector * + sizeof(struct drm_connector), GFP_KERNEL); + if (!save_connectors) { + kfree(save_crtcs); + kfree(save_encoders); + return -ENOMEM; + } + + /* Copy data. Note that driver private data is not affected. + * Should anything bad happen only the expected state is + * restored, not the drivers personal bookkeeping. + */ + count = 0; + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { + save_crtcs[count++] = *crtc; + } + + count = 0; + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + save_encoders[count++] = *encoder; + } + + count = 0; + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + save_connectors[count++] = *connector; + } + + /* We should be able to check here if the fb has the same properties + * and then just flip_or_move it */ + if (set->crtc->fb != set->fb) { + /* If we have no fb then treat it as a full mode set */ + if (set->crtc->fb == NULL) { + DRM_DEBUG_KMS("crtc has no fb, full mode set\n"); + mode_changed = true; + } else if (set->fb == NULL) { + mode_changed = true; + } else + fb_changed = true; + } + + if (set->x != set->crtc->x || set->y != set->crtc->y) + fb_changed = true; + + if (set->mode && !drm_mode_equal(set->mode, &set->crtc->mode)) { + DRM_DEBUG_KMS("modes are different, full mode set\n"); + drm_mode_debug_printmodeline(&set->crtc->mode); + drm_mode_debug_printmodeline(set->mode); + mode_changed = true; + } + + /* a) traverse passed in connector list and get encoders for them */ + count = 0; + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + struct drm_connector_helper_funcs *connector_funcs = + connector->helper_private; + new_encoder = connector->encoder; + for (ro = 0; ro < set->num_connectors; ro++) { + if (set->connectors[ro] == connector) { + new_encoder = connector_funcs->best_encoder(connector); + /* if we can't get an encoder for a connector + we are setting now - then fail */ + if (new_encoder == NULL) + /* don't break so fail path works correct */ + fail = 1; + break; + } + } + + if (new_encoder != connector->encoder) { + DRM_DEBUG_KMS("encoder changed, full mode switch\n"); + mode_changed = true; + /* If the encoder is reused for another connector, then + * the appropriate crtc will be set later. + */ + if (connector->encoder) + connector->encoder->crtc = NULL; + connector->encoder = new_encoder; + } + } + + if (fail) { + ret = -EINVAL; + goto fail; + } + + count = 0; + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + if (!connector->encoder) + continue; + + if (connector->encoder->crtc == set->crtc) + new_crtc = NULL; + else + new_crtc = connector->encoder->crtc; + + for (ro = 0; ro < set->num_connectors; ro++) { + if (set->connectors[ro] == connector) + new_crtc = set->crtc; + } + + /* Make sure the new CRTC will work with the encoder */ + if (new_crtc && + !drm_encoder_crtc_ok(connector->encoder, new_crtc)) { + ret = -EINVAL; + goto fail; + } + if (new_crtc != connector->encoder->crtc) { + DRM_DEBUG_KMS("crtc changed, full mode switch\n"); + mode_changed = true; + connector->encoder->crtc = new_crtc; + } + if (new_crtc) { + DRM_DEBUG_KMS("[CONNECTOR:%d:%s] to [CRTC:%d]\n", + connector->base.id, drm_get_connector_name(connector), + new_crtc->base.id); + } else { + DRM_DEBUG_KMS("[CONNECTOR:%d:%s] to [NOCRTC]\n", + connector->base.id, drm_get_connector_name(connector)); + } + } + + /* mode_set_base is not a required function */ + if (fb_changed && !crtc_funcs->mode_set_base) + mode_changed = true; + + if (mode_changed) { + set->crtc->enabled = drm_helper_crtc_in_use(set->crtc); + if (set->crtc->enabled) { + DRM_DEBUG_KMS("attempting to set mode from" + " userspace\n"); + drm_mode_debug_printmodeline(set->mode); + old_fb = set->crtc->fb; + set->crtc->fb = set->fb; + if (!drm_crtc_helper_set_mode(set->crtc, set->mode, + set->x, set->y, + old_fb)) { + DRM_ERROR("failed to set mode on [CRTC:%d]\n", + set->crtc->base.id); + set->crtc->fb = old_fb; + ret = -EINVAL; + goto fail; + } + DRM_DEBUG_KMS("Setting connector DPMS state to on\n"); + for (i = 0; i < set->num_connectors; i++) { + DRM_DEBUG_KMS("\t[CONNECTOR:%d:%s] set DPMS on\n", set->connectors[i]->base.id, + drm_get_connector_name(set->connectors[i])); + set->connectors[i]->dpms = DRM_MODE_DPMS_ON; + } + } + drm_helper_disable_unused_functions(dev); + } else if (fb_changed) { + set->crtc->x = set->x; + set->crtc->y = set->y; + + old_fb = set->crtc->fb; + if (set->crtc->fb != set->fb) + set->crtc->fb = set->fb; + ret = crtc_funcs->mode_set_base(set->crtc, + set->x, set->y, old_fb); + if (ret != 0) { + set->crtc->fb = old_fb; + goto fail; + } + } + + kfree(save_connectors); + kfree(save_encoders); + kfree(save_crtcs); + return 0; + +fail: + /* Restore all previous data. */ + count = 0; + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { + *crtc = save_crtcs[count++]; + } + + count = 0; + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + *encoder = save_encoders[count++]; + } + + count = 0; + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + *connector = save_connectors[count++]; + } + + kfree(save_connectors); + kfree(save_encoders); + kfree(save_crtcs); + return ret; +} +EXPORT_SYMBOL(drm_crtc_helper_set_config); + +static int drm_helper_choose_encoder_dpms(struct drm_encoder *encoder) +{ + int dpms = DRM_MODE_DPMS_OFF; + struct drm_connector *connector; + struct drm_device *dev = encoder->dev; + + list_for_each_entry(connector, &dev->mode_config.connector_list, head) + if (connector->encoder == encoder) + if (connector->dpms < dpms) + dpms = connector->dpms; + return dpms; +} + +static int drm_helper_choose_crtc_dpms(struct drm_crtc *crtc) +{ + int dpms = DRM_MODE_DPMS_OFF; + struct drm_connector *connector; + struct drm_device *dev = crtc->dev; + + list_for_each_entry(connector, &dev->mode_config.connector_list, head) + if (connector->encoder && connector->encoder->crtc == crtc) + if (connector->dpms < dpms) + dpms = connector->dpms; + return dpms; +} + +/** + * drm_helper_connector_dpms + * @connector affected connector + * @mode DPMS mode + * + * Calls the low-level connector DPMS function, then + * calls appropriate encoder and crtc DPMS functions as well + */ +void drm_helper_connector_dpms(struct drm_connector *connector, int mode) +{ + struct drm_encoder *encoder = connector->encoder; + struct drm_crtc *crtc = encoder ? encoder->crtc : NULL; + int old_dpms; + + if (mode == connector->dpms) + return; + + old_dpms = connector->dpms; + connector->dpms = mode; + + /* from off to on, do crtc then encoder */ + if (mode < old_dpms) { + if (crtc) { + struct drm_crtc_helper_funcs *crtc_funcs = crtc->helper_private; + if (crtc_funcs->dpms) + (*crtc_funcs->dpms) (crtc, + drm_helper_choose_crtc_dpms(crtc)); + } + if (encoder) { + struct drm_encoder_helper_funcs *encoder_funcs = encoder->helper_private; + if (encoder_funcs->dpms) + (*encoder_funcs->dpms) (encoder, + drm_helper_choose_encoder_dpms(encoder)); + } + } + + /* from on to off, do encoder then crtc */ + if (mode > old_dpms) { + if (encoder) { + struct drm_encoder_helper_funcs *encoder_funcs = encoder->helper_private; + if (encoder_funcs->dpms) + (*encoder_funcs->dpms) (encoder, + drm_helper_choose_encoder_dpms(encoder)); + } + if (crtc) { + struct drm_crtc_helper_funcs *crtc_funcs = crtc->helper_private; + if (crtc_funcs->dpms) + (*crtc_funcs->dpms) (crtc, + drm_helper_choose_crtc_dpms(crtc)); + } + } + + return; +} +EXPORT_SYMBOL(drm_helper_connector_dpms); + +int drm_helper_mode_fill_fb_struct(struct drm_framebuffer *fb, + struct drm_mode_fb_cmd *mode_cmd) +{ + fb->width = mode_cmd->width; + fb->height = mode_cmd->height; + fb->pitch = mode_cmd->pitch; + fb->bits_per_pixel = mode_cmd->bpp; + fb->depth = mode_cmd->depth; + + return 0; +} +EXPORT_SYMBOL(drm_helper_mode_fill_fb_struct); + +int drm_helper_resume_force_mode(struct drm_device *dev) +{ + struct drm_crtc *crtc; + struct drm_encoder *encoder; + struct drm_encoder_helper_funcs *encoder_funcs; + struct drm_crtc_helper_funcs *crtc_funcs; + int ret; + + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { + + if (!crtc->enabled) + continue; + + ret = drm_crtc_helper_set_mode(crtc, &crtc->mode, + crtc->x, crtc->y, crtc->fb); + + if (ret == false) + DRM_ERROR("failed to set mode on crtc %p\n", crtc); + + /* Turn off outputs that were already powered off */ + if (drm_helper_choose_crtc_dpms(crtc)) { + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + + if(encoder->crtc != crtc) + continue; + + encoder_funcs = encoder->helper_private; + if (encoder_funcs->dpms) + (*encoder_funcs->dpms) (encoder, + drm_helper_choose_encoder_dpms(encoder)); + } + + crtc_funcs = crtc->helper_private; + if (crtc_funcs->dpms) + (*crtc_funcs->dpms) (crtc, + drm_helper_choose_crtc_dpms(crtc)); + } + } + /* disable the unused connectors while restoring the modesetting */ + drm_helper_disable_unused_functions(dev); + return 0; +} +EXPORT_SYMBOL(drm_helper_resume_force_mode); + +#define DRM_OUTPUT_POLL_PERIOD (10*HZ) +static void output_poll_execute(struct work_struct *work) +{ + struct delayed_work *delayed_work = to_delayed_work(work); + struct drm_device *dev = container_of(delayed_work, struct drm_device, mode_config.output_poll_work); + struct drm_connector *connector; + enum drm_connector_status old_status; + bool repoll = false, changed = false; + + if (!drm_kms_helper_poll) + return; + + mutex_lock(&dev->mode_config.mutex); + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + + /* if this is HPD or polled don't check it - + TV out for instance */ + if (!connector->polled) + continue; + + else if (connector->polled & (DRM_CONNECTOR_POLL_CONNECT | DRM_CONNECTOR_POLL_DISCONNECT)) + repoll = true; + + old_status = connector->status; + /* if we are connected and don't want to poll for disconnect + skip it */ + if (old_status == connector_status_connected && + !(connector->polled & DRM_CONNECTOR_POLL_DISCONNECT) && + !(connector->polled & DRM_CONNECTOR_POLL_HPD)) + continue; + + connector->status = connector->funcs->detect(connector, false); + DRM_DEBUG_KMS("[CONNECTOR:%d:%s] status updated from %d to %d\n", + connector->base.id, + drm_get_connector_name(connector), + old_status, connector->status); + if (old_status != connector->status) + changed = true; + } + + mutex_unlock(&dev->mode_config.mutex); + + if (changed) { + /* send a uevent + call fbdev */ + drm_sysfs_hotplug_event(dev); + if (dev->mode_config.funcs->output_poll_changed) + dev->mode_config.funcs->output_poll_changed(dev); + } + + if (repoll) + queue_delayed_work(system_nrt_wq, delayed_work, DRM_OUTPUT_POLL_PERIOD); +} + +void drm_kms_helper_poll_disable(struct drm_device *dev) +{ + if (!dev->mode_config.poll_enabled) + return; + cancel_delayed_work_sync(&dev->mode_config.output_poll_work); +} +EXPORT_SYMBOL(drm_kms_helper_poll_disable); + +void drm_kms_helper_poll_enable(struct drm_device *dev) +{ + bool poll = false; + struct drm_connector *connector; + + if (!dev->mode_config.poll_enabled || !drm_kms_helper_poll) + return; + + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + if (connector->polled) + poll = true; + } + + if (poll) + queue_delayed_work(system_nrt_wq, &dev->mode_config.output_poll_work, DRM_OUTPUT_POLL_PERIOD); +} +EXPORT_SYMBOL(drm_kms_helper_poll_enable); + +void drm_kms_helper_poll_init(struct drm_device *dev) +{ + INIT_DELAYED_WORK(&dev->mode_config.output_poll_work, output_poll_execute); + dev->mode_config.poll_enabled = true; + + drm_kms_helper_poll_enable(dev); +} +EXPORT_SYMBOL(drm_kms_helper_poll_init); + +void drm_kms_helper_poll_fini(struct drm_device *dev) +{ + drm_kms_helper_poll_disable(dev); +} +EXPORT_SYMBOL(drm_kms_helper_poll_fini); + +void drm_helper_hpd_irq_event(struct drm_device *dev) +{ + if (!dev->mode_config.poll_enabled) + return; + + /* kill timer and schedule immediate execution, this doesn't block */ + cancel_delayed_work(&dev->mode_config.output_poll_work); + if (drm_kms_helper_poll) + queue_delayed_work(system_nrt_wq, &dev->mode_config.output_poll_work, 0); +} +EXPORT_SYMBOL(drm_helper_hpd_irq_event); diff --git a/drivers/gpu/drm/drm_debugfs.c b/drivers/gpu/drm/drm_debugfs.c new file mode 100644 index 00000000..9d8c892d --- /dev/null +++ b/drivers/gpu/drm/drm_debugfs.c @@ -0,0 +1,236 @@ +/** + * \file drm_debugfs.c + * debugfs support for DRM + * + * \author Ben Gamari + */ + +/* + * Created: Sun Dec 21 13:08:50 2008 by bgamari@gmail.com + * + * Copyright 2008 Ben Gamari + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * VA LINUX SYSTEMS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#include +#include +#include +#include "drmP.h" + +#if defined(CONFIG_DEBUG_FS) + +/*************************************************** + * Initialization, etc. + **************************************************/ + +static struct drm_info_list drm_debugfs_list[] = { + {"name", drm_name_info, 0}, + {"vm", drm_vm_info, 0}, + {"clients", drm_clients_info, 0}, + {"queues", drm_queues_info, 0}, + {"bufs", drm_bufs_info, 0}, + {"gem_names", drm_gem_name_info, DRIVER_GEM}, +#if DRM_DEBUG_CODE + {"vma", drm_vma_info, 0}, +#endif +}; +#define DRM_DEBUGFS_ENTRIES ARRAY_SIZE(drm_debugfs_list) + + +static int drm_debugfs_open(struct inode *inode, struct file *file) +{ + struct drm_info_node *node = inode->i_private; + + return single_open(file, node->info_ent->show, node); +} + + +static const struct file_operations drm_debugfs_fops = { + .owner = THIS_MODULE, + .open = drm_debugfs_open, + .read = seq_read, + .llseek = seq_lseek, + .release = single_release, +}; + + +/** + * Initialize a given set of debugfs files for a device + * + * \param files The array of files to create + * \param count The number of files given + * \param root DRI debugfs dir entry. + * \param minor device minor number + * \return Zero on success, non-zero on failure + * + * Create a given set of debugfs files represented by an array of + * gdm_debugfs_lists in the given root directory. + */ +int drm_debugfs_create_files(struct drm_info_list *files, int count, + struct dentry *root, struct drm_minor *minor) +{ + struct drm_device *dev = minor->dev; + struct dentry *ent; + struct drm_info_node *tmp; + char name[64]; + int i, ret; + + for (i = 0; i < count; i++) { + u32 features = files[i].driver_features; + + if (features != 0 && + (dev->driver->driver_features & features) != features) + continue; + + tmp = kmalloc(sizeof(struct drm_info_node), GFP_KERNEL); + if (tmp == NULL) { + ret = -1; + goto fail; + } + ent = debugfs_create_file(files[i].name, S_IFREG | S_IRUGO, + root, tmp, &drm_debugfs_fops); + if (!ent) { + DRM_ERROR("Cannot create /sys/kernel/debug/dri/%s/%s\n", + name, files[i].name); + kfree(tmp); + ret = -1; + goto fail; + } + + tmp->minor = minor; + tmp->dent = ent; + tmp->info_ent = &files[i]; + list_add(&(tmp->list), &(minor->debugfs_nodes.list)); + } + return 0; + +fail: + drm_debugfs_remove_files(files, count, minor); + return ret; +} +EXPORT_SYMBOL(drm_debugfs_create_files); + +/** + * Initialize the DRI debugfs filesystem for a device + * + * \param dev DRM device + * \param minor device minor number + * \param root DRI debugfs dir entry. + * + * Create the DRI debugfs root entry "/sys/kernel/debug/dri", the device debugfs root entry + * "/sys/kernel/debug/dri/%minor%/", and each entry in debugfs_list as + * "/sys/kernel/debug/dri/%minor%/%name%". + */ +int drm_debugfs_init(struct drm_minor *minor, int minor_id, + struct dentry *root) +{ + struct drm_device *dev = minor->dev; + char name[64]; + int ret; + + INIT_LIST_HEAD(&minor->debugfs_nodes.list); + sprintf(name, "%d", minor_id); + minor->debugfs_root = debugfs_create_dir(name, root); + if (!minor->debugfs_root) { + DRM_ERROR("Cannot create /sys/kernel/debug/dri/%s\n", name); + return -1; + } + + ret = drm_debugfs_create_files(drm_debugfs_list, DRM_DEBUGFS_ENTRIES, + minor->debugfs_root, minor); + if (ret) { + debugfs_remove(minor->debugfs_root); + minor->debugfs_root = NULL; + DRM_ERROR("Failed to create core drm debugfs files\n"); + return ret; + } + + if (dev->driver->debugfs_init) { + ret = dev->driver->debugfs_init(minor); + if (ret) { + DRM_ERROR("DRM: Driver failed to initialize " + "/sys/kernel/debug/dri.\n"); + return ret; + } + } + return 0; +} + + +/** + * Remove a list of debugfs files + * + * \param files The list of files + * \param count The number of files + * \param minor The minor of which we should remove the files + * \return always zero. + * + * Remove all debugfs entries created by debugfs_init(). + */ +int drm_debugfs_remove_files(struct drm_info_list *files, int count, + struct drm_minor *minor) +{ + struct list_head *pos, *q; + struct drm_info_node *tmp; + int i; + + for (i = 0; i < count; i++) { + list_for_each_safe(pos, q, &minor->debugfs_nodes.list) { + tmp = list_entry(pos, struct drm_info_node, list); + if (tmp->info_ent == &files[i]) { + debugfs_remove(tmp->dent); + list_del(pos); + kfree(tmp); + } + } + } + return 0; +} +EXPORT_SYMBOL(drm_debugfs_remove_files); + +/** + * Cleanup the debugfs filesystem resources. + * + * \param minor device minor number. + * \return always zero. + * + * Remove all debugfs entries created by debugfs_init(). + */ +int drm_debugfs_cleanup(struct drm_minor *minor) +{ + struct drm_device *dev = minor->dev; + + if (!minor->debugfs_root) + return 0; + + if (dev->driver->debugfs_cleanup) + dev->driver->debugfs_cleanup(minor); + + drm_debugfs_remove_files(drm_debugfs_list, DRM_DEBUGFS_ENTRIES, minor); + + debugfs_remove(minor->debugfs_root); + minor->debugfs_root = NULL; + + return 0; +} + +#endif /* CONFIG_DEBUG_FS */ + diff --git a/drivers/gpu/drm/drm_dma.c b/drivers/gpu/drm/drm_dma.c new file mode 100644 index 00000000..252cbd74 --- /dev/null +++ b/drivers/gpu/drm/drm_dma.c @@ -0,0 +1,161 @@ +/** + * \file drm_dma.c + * DMA IOCTL and function support + * + * \author Rickard E. (Rik) Faith + * \author Gareth Hughes + */ + +/* + * Created: Fri Mar 19 14:30:16 1999 by faith@valinux.com + * + * Copyright 1999, 2000 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * VA LINUX SYSTEMS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#include "drmP.h" + +/** + * Initialize the DMA data. + * + * \param dev DRM device. + * \return zero on success or a negative value on failure. + * + * Allocate and initialize a drm_device_dma structure. + */ +int drm_dma_setup(struct drm_device *dev) +{ + int i; + + dev->dma = kzalloc(sizeof(*dev->dma), GFP_KERNEL); + if (!dev->dma) + return -ENOMEM; + + for (i = 0; i <= DRM_MAX_ORDER; i++) + memset(&dev->dma->bufs[i], 0, sizeof(dev->dma->bufs[0])); + + return 0; +} + +/** + * Cleanup the DMA resources. + * + * \param dev DRM device. + * + * Free all pages associated with DMA buffers, the buffers and pages lists, and + * finally the drm_device::dma structure itself. + */ +void drm_dma_takedown(struct drm_device *dev) +{ + struct drm_device_dma *dma = dev->dma; + int i, j; + + if (!dma) + return; + + /* Clear dma buffers */ + for (i = 0; i <= DRM_MAX_ORDER; i++) { + if (dma->bufs[i].seg_count) { + DRM_DEBUG("order %d: buf_count = %d," + " seg_count = %d\n", + i, + dma->bufs[i].buf_count, + dma->bufs[i].seg_count); + for (j = 0; j < dma->bufs[i].seg_count; j++) { + if (dma->bufs[i].seglist[j]) { + drm_pci_free(dev, dma->bufs[i].seglist[j]); + } + } + kfree(dma->bufs[i].seglist); + } + if (dma->bufs[i].buf_count) { + for (j = 0; j < dma->bufs[i].buf_count; j++) { + kfree(dma->bufs[i].buflist[j].dev_private); + } + kfree(dma->bufs[i].buflist); + } + } + + kfree(dma->buflist); + kfree(dma->pagelist); + kfree(dev->dma); + dev->dma = NULL; +} + +/** + * Free a buffer. + * + * \param dev DRM device. + * \param buf buffer to free. + * + * Resets the fields of \p buf. + */ +void drm_free_buffer(struct drm_device *dev, struct drm_buf * buf) +{ + if (!buf) + return; + + buf->waiting = 0; + buf->pending = 0; + buf->file_priv = NULL; + buf->used = 0; + + if (drm_core_check_feature(dev, DRIVER_DMA_QUEUE) + && waitqueue_active(&buf->dma_wait)) { + wake_up_interruptible(&buf->dma_wait); + } +} + +/** + * Reclaim the buffers. + * + * \param file_priv DRM file private. + * + * Frees each buffer associated with \p file_priv not already on the hardware. + */ +void drm_core_reclaim_buffers(struct drm_device *dev, + struct drm_file *file_priv) +{ + struct drm_device_dma *dma = dev->dma; + int i; + + if (!dma) + return; + for (i = 0; i < dma->buf_count; i++) { + if (dma->buflist[i]->file_priv == file_priv) { + switch (dma->buflist[i]->list) { + case DRM_LIST_NONE: + drm_free_buffer(dev, dma->buflist[i]); + break; + case DRM_LIST_WAIT: + dma->buflist[i]->list = DRM_LIST_RECLAIM; + break; + default: + /* Buffer already on hardware. */ + break; + } + } + } +} + +EXPORT_SYMBOL(drm_core_reclaim_buffers); diff --git a/drivers/gpu/drm/drm_dp_i2c_helper.c b/drivers/gpu/drm/drm_dp_i2c_helper.c new file mode 100644 index 00000000..f7eba0a0 --- /dev/null +++ b/drivers/gpu/drm/drm_dp_i2c_helper.c @@ -0,0 +1,208 @@ +/* + * Copyright © 2009 Keith Packard + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that copyright + * notice and this permission notice appear in supporting documentation, and + * that the name of the copyright holders not be used in advertising or + * publicity pertaining to distribution of the software without specific, + * written prior permission. The copyright holders make no representations + * about the suitability of this software for any purpose. It is provided "as + * is" without express or implied warranty. + * + * THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, + * INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO + * EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY SPECIAL, INDIRECT OR + * CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, + * DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER + * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE + * OF THIS SOFTWARE. + */ + +#include +#include +#include +#include +#include +#include +#include +#include "drm_dp_helper.h" +#include "drmP.h" + +/* Run a single AUX_CH I2C transaction, writing/reading data as necessary */ +static int +i2c_algo_dp_aux_transaction(struct i2c_adapter *adapter, int mode, + uint8_t write_byte, uint8_t *read_byte) +{ + struct i2c_algo_dp_aux_data *algo_data = adapter->algo_data; + int ret; + + ret = (*algo_data->aux_ch)(adapter, mode, + write_byte, read_byte); + return ret; +} + +/* + * I2C over AUX CH + */ + +/* + * Send the address. If the I2C link is running, this 'restarts' + * the connection with the new address, this is used for doing + * a write followed by a read (as needed for DDC) + */ +static int +i2c_algo_dp_aux_address(struct i2c_adapter *adapter, u16 address, bool reading) +{ + struct i2c_algo_dp_aux_data *algo_data = adapter->algo_data; + int mode = MODE_I2C_START; + int ret; + + if (reading) + mode |= MODE_I2C_READ; + else + mode |= MODE_I2C_WRITE; + algo_data->address = address; + algo_data->running = true; + ret = i2c_algo_dp_aux_transaction(adapter, mode, 0, NULL); + return ret; +} + +/* + * Stop the I2C transaction. This closes out the link, sending + * a bare address packet with the MOT bit turned off + */ +static void +i2c_algo_dp_aux_stop(struct i2c_adapter *adapter, bool reading) +{ + struct i2c_algo_dp_aux_data *algo_data = adapter->algo_data; + int mode = MODE_I2C_STOP; + + if (reading) + mode |= MODE_I2C_READ; + else + mode |= MODE_I2C_WRITE; + if (algo_data->running) { + (void) i2c_algo_dp_aux_transaction(adapter, mode, 0, NULL); + algo_data->running = false; + } +} + +/* + * Write a single byte to the current I2C address, the + * the I2C link must be running or this returns -EIO + */ +static int +i2c_algo_dp_aux_put_byte(struct i2c_adapter *adapter, u8 byte) +{ + struct i2c_algo_dp_aux_data *algo_data = adapter->algo_data; + int ret; + + if (!algo_data->running) + return -EIO; + + ret = i2c_algo_dp_aux_transaction(adapter, MODE_I2C_WRITE, byte, NULL); + return ret; +} + +/* + * Read a single byte from the current I2C address, the + * I2C link must be running or this returns -EIO + */ +static int +i2c_algo_dp_aux_get_byte(struct i2c_adapter *adapter, u8 *byte_ret) +{ + struct i2c_algo_dp_aux_data *algo_data = adapter->algo_data; + int ret; + + if (!algo_data->running) + return -EIO; + + ret = i2c_algo_dp_aux_transaction(adapter, MODE_I2C_READ, 0, byte_ret); + return ret; +} + +static int +i2c_algo_dp_aux_xfer(struct i2c_adapter *adapter, + struct i2c_msg *msgs, + int num) +{ + int ret = 0; + bool reading = false; + int m; + int b; + + for (m = 0; m < num; m++) { + u16 len = msgs[m].len; + u8 *buf = msgs[m].buf; + reading = (msgs[m].flags & I2C_M_RD) != 0; + ret = i2c_algo_dp_aux_address(adapter, msgs[m].addr, reading); + if (ret < 0) + break; + if (reading) { + for (b = 0; b < len; b++) { + ret = i2c_algo_dp_aux_get_byte(adapter, &buf[b]); + if (ret < 0) + break; + } + } else { + for (b = 0; b < len; b++) { + ret = i2c_algo_dp_aux_put_byte(adapter, buf[b]); + if (ret < 0) + break; + } + } + if (ret < 0) + break; + } + if (ret >= 0) + ret = num; + i2c_algo_dp_aux_stop(adapter, reading); + DRM_DEBUG_KMS("dp_aux_xfer return %d\n", ret); + return ret; +} + +static u32 +i2c_algo_dp_aux_functionality(struct i2c_adapter *adapter) +{ + return I2C_FUNC_I2C | I2C_FUNC_SMBUS_EMUL | + I2C_FUNC_SMBUS_READ_BLOCK_DATA | + I2C_FUNC_SMBUS_BLOCK_PROC_CALL | + I2C_FUNC_10BIT_ADDR; +} + +static const struct i2c_algorithm i2c_dp_aux_algo = { + .master_xfer = i2c_algo_dp_aux_xfer, + .functionality = i2c_algo_dp_aux_functionality, +}; + +static void +i2c_dp_aux_reset_bus(struct i2c_adapter *adapter) +{ + (void) i2c_algo_dp_aux_address(adapter, 0, false); + (void) i2c_algo_dp_aux_stop(adapter, false); + +} + +static int +i2c_dp_aux_prepare_bus(struct i2c_adapter *adapter) +{ + adapter->algo = &i2c_dp_aux_algo; + adapter->retries = 3; + i2c_dp_aux_reset_bus(adapter); + return 0; +} + +int +i2c_dp_aux_add_bus(struct i2c_adapter *adapter) +{ + int error; + + error = i2c_dp_aux_prepare_bus(adapter); + if (error) + return error; + error = i2c_add_adapter(adapter); + return error; +} +EXPORT_SYMBOL(i2c_dp_aux_add_bus); diff --git a/drivers/gpu/drm/drm_drv.c b/drivers/gpu/drm/drm_drv.c new file mode 100644 index 00000000..93a112d4 --- /dev/null +++ b/drivers/gpu/drm/drm_drv.c @@ -0,0 +1,490 @@ +/** + * \file drm_drv.c + * Generic driver template + * + * \author Rickard E. (Rik) Faith + * \author Gareth Hughes + * + * To use this template, you must at least define the following (samples + * given for the MGA driver): + * + * \code + * #define DRIVER_AUTHOR "VA Linux Systems, Inc." + * + * #define DRIVER_NAME "mga" + * #define DRIVER_DESC "Matrox G200/G400" + * #define DRIVER_DATE "20001127" + * + * #define drm_x mga_##x + * \endcode + */ + +/* + * Created: Thu Nov 23 03:10:50 2000 by gareth@valinux.com + * + * Copyright 1999, 2000 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * VA LINUX SYSTEMS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#include +#include +#include "drmP.h" +#include "drm_core.h" + + +static int drm_version(struct drm_device *dev, void *data, + struct drm_file *file_priv); + +#define DRM_IOCTL_DEF(ioctl, _func, _flags) \ + [DRM_IOCTL_NR(ioctl)] = {.cmd = ioctl, .func = _func, .flags = _flags, .cmd_drv = 0} + +/** Ioctl table */ +static struct drm_ioctl_desc drm_ioctls[] = { + DRM_IOCTL_DEF(DRM_IOCTL_VERSION, drm_version, 0), + DRM_IOCTL_DEF(DRM_IOCTL_GET_UNIQUE, drm_getunique, 0), + DRM_IOCTL_DEF(DRM_IOCTL_GET_MAGIC, drm_getmagic, 0), + DRM_IOCTL_DEF(DRM_IOCTL_IRQ_BUSID, drm_irq_by_busid, DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF(DRM_IOCTL_GET_MAP, drm_getmap, 0), + DRM_IOCTL_DEF(DRM_IOCTL_GET_CLIENT, drm_getclient, 0), + DRM_IOCTL_DEF(DRM_IOCTL_GET_STATS, drm_getstats, 0), + DRM_IOCTL_DEF(DRM_IOCTL_GET_CAP, drm_getcap, 0), + DRM_IOCTL_DEF(DRM_IOCTL_SET_VERSION, drm_setversion, DRM_MASTER), + + DRM_IOCTL_DEF(DRM_IOCTL_SET_UNIQUE, drm_setunique, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF(DRM_IOCTL_BLOCK, drm_noop, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF(DRM_IOCTL_UNBLOCK, drm_noop, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF(DRM_IOCTL_AUTH_MAGIC, drm_authmagic, DRM_AUTH|DRM_MASTER), + + DRM_IOCTL_DEF(DRM_IOCTL_ADD_MAP, drm_addmap_ioctl, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF(DRM_IOCTL_RM_MAP, drm_rmmap_ioctl, DRM_AUTH), + + DRM_IOCTL_DEF(DRM_IOCTL_SET_SAREA_CTX, drm_setsareactx, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF(DRM_IOCTL_GET_SAREA_CTX, drm_getsareactx, DRM_AUTH), + + DRM_IOCTL_DEF(DRM_IOCTL_SET_MASTER, drm_setmaster_ioctl, DRM_ROOT_ONLY), + DRM_IOCTL_DEF(DRM_IOCTL_DROP_MASTER, drm_dropmaster_ioctl, DRM_ROOT_ONLY), + + DRM_IOCTL_DEF(DRM_IOCTL_ADD_CTX, drm_addctx, DRM_AUTH|DRM_ROOT_ONLY), + DRM_IOCTL_DEF(DRM_IOCTL_RM_CTX, drm_rmctx, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF(DRM_IOCTL_MOD_CTX, drm_modctx, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF(DRM_IOCTL_GET_CTX, drm_getctx, DRM_AUTH), + DRM_IOCTL_DEF(DRM_IOCTL_SWITCH_CTX, drm_switchctx, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF(DRM_IOCTL_NEW_CTX, drm_newctx, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF(DRM_IOCTL_RES_CTX, drm_resctx, DRM_AUTH), + + DRM_IOCTL_DEF(DRM_IOCTL_ADD_DRAW, drm_noop, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF(DRM_IOCTL_RM_DRAW, drm_noop, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + + DRM_IOCTL_DEF(DRM_IOCTL_LOCK, drm_lock, DRM_AUTH), + DRM_IOCTL_DEF(DRM_IOCTL_UNLOCK, drm_unlock, DRM_AUTH), + + DRM_IOCTL_DEF(DRM_IOCTL_FINISH, drm_noop, DRM_AUTH), + + DRM_IOCTL_DEF(DRM_IOCTL_ADD_BUFS, drm_addbufs, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF(DRM_IOCTL_MARK_BUFS, drm_markbufs, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF(DRM_IOCTL_INFO_BUFS, drm_infobufs, DRM_AUTH), + DRM_IOCTL_DEF(DRM_IOCTL_MAP_BUFS, drm_mapbufs, DRM_AUTH), + DRM_IOCTL_DEF(DRM_IOCTL_FREE_BUFS, drm_freebufs, DRM_AUTH), + /* The DRM_IOCTL_DMA ioctl should be defined by the driver. */ + DRM_IOCTL_DEF(DRM_IOCTL_DMA, NULL, DRM_AUTH), + + DRM_IOCTL_DEF(DRM_IOCTL_CONTROL, drm_control, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + +#if __OS_HAS_AGP + DRM_IOCTL_DEF(DRM_IOCTL_AGP_ACQUIRE, drm_agp_acquire_ioctl, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF(DRM_IOCTL_AGP_RELEASE, drm_agp_release_ioctl, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF(DRM_IOCTL_AGP_ENABLE, drm_agp_enable_ioctl, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF(DRM_IOCTL_AGP_INFO, drm_agp_info_ioctl, DRM_AUTH), + DRM_IOCTL_DEF(DRM_IOCTL_AGP_ALLOC, drm_agp_alloc_ioctl, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF(DRM_IOCTL_AGP_FREE, drm_agp_free_ioctl, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF(DRM_IOCTL_AGP_BIND, drm_agp_bind_ioctl, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF(DRM_IOCTL_AGP_UNBIND, drm_agp_unbind_ioctl, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), +#endif + + DRM_IOCTL_DEF(DRM_IOCTL_SG_ALLOC, drm_sg_alloc_ioctl, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF(DRM_IOCTL_SG_FREE, drm_sg_free, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + + DRM_IOCTL_DEF(DRM_IOCTL_WAIT_VBLANK, drm_wait_vblank, 0), + + DRM_IOCTL_DEF(DRM_IOCTL_MODESET_CTL, drm_modeset_ctl, 0), + + DRM_IOCTL_DEF(DRM_IOCTL_UPDATE_DRAW, drm_noop, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + + DRM_IOCTL_DEF(DRM_IOCTL_GEM_CLOSE, drm_gem_close_ioctl, DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_GEM_FLINK, drm_gem_flink_ioctl, DRM_AUTH|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_GEM_OPEN, drm_gem_open_ioctl, DRM_AUTH|DRM_UNLOCKED), + + DRM_IOCTL_DEF(DRM_IOCTL_MODE_GETRESOURCES, drm_mode_getresources, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_GETCRTC, drm_mode_getcrtc, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_SETCRTC, drm_mode_setcrtc, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_CURSOR, drm_mode_cursor_ioctl, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_GETGAMMA, drm_mode_gamma_get_ioctl, DRM_MASTER|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_SETGAMMA, drm_mode_gamma_set_ioctl, DRM_MASTER|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_GETENCODER, drm_mode_getencoder, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_GETCONNECTOR, drm_mode_getconnector, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_ATTACHMODE, drm_mode_attachmode_ioctl, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_DETACHMODE, drm_mode_detachmode_ioctl, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_GETPROPERTY, drm_mode_getproperty_ioctl, DRM_MASTER | DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_SETPROPERTY, drm_mode_connector_property_set_ioctl, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_GETPROPBLOB, drm_mode_getblob_ioctl, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_GETFB, drm_mode_getfb, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_ADDFB, drm_mode_addfb, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_RMFB, drm_mode_rmfb, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_PAGE_FLIP, drm_mode_page_flip_ioctl, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_DIRTYFB, drm_mode_dirtyfb_ioctl, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_CREATE_DUMB, drm_mode_create_dumb_ioctl, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_MAP_DUMB, drm_mode_mmap_dumb_ioctl, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF(DRM_IOCTL_MODE_DESTROY_DUMB, drm_mode_destroy_dumb_ioctl, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED) +}; + +#define DRM_CORE_IOCTL_COUNT ARRAY_SIZE( drm_ioctls ) + +/** + * Take down the DRM device. + * + * \param dev DRM device structure. + * + * Frees every resource in \p dev. + * + * \sa drm_device + */ +int drm_lastclose(struct drm_device * dev) +{ + struct drm_vma_entry *vma, *vma_temp; + int i; + + DRM_DEBUG("\n"); + + if (dev->driver->lastclose) + dev->driver->lastclose(dev); + DRM_DEBUG("driver lastclose completed\n"); + + if (dev->irq_enabled && !drm_core_check_feature(dev, DRIVER_MODESET)) + drm_irq_uninstall(dev); + + mutex_lock(&dev->struct_mutex); + + /* Clear AGP information */ + if (drm_core_has_AGP(dev) && dev->agp && + !drm_core_check_feature(dev, DRIVER_MODESET)) { + struct drm_agp_mem *entry, *tempe; + + /* Remove AGP resources, but leave dev->agp + intact until drv_cleanup is called. */ + list_for_each_entry_safe(entry, tempe, &dev->agp->memory, head) { + if (entry->bound) + drm_unbind_agp(entry->memory); + drm_free_agp(entry->memory, entry->pages); + kfree(entry); + } + INIT_LIST_HEAD(&dev->agp->memory); + + if (dev->agp->acquired) + drm_agp_release(dev); + + dev->agp->acquired = 0; + dev->agp->enabled = 0; + } + if (drm_core_check_feature(dev, DRIVER_SG) && dev->sg && + !drm_core_check_feature(dev, DRIVER_MODESET)) { + drm_sg_cleanup(dev->sg); + dev->sg = NULL; + } + + /* Clear vma list (only built for debugging) */ + list_for_each_entry_safe(vma, vma_temp, &dev->vmalist, head) { + list_del(&vma->head); + kfree(vma); + } + + if (drm_core_check_feature(dev, DRIVER_DMA_QUEUE) && dev->queuelist) { + for (i = 0; i < dev->queue_count; i++) { + kfree(dev->queuelist[i]); + dev->queuelist[i] = NULL; + } + kfree(dev->queuelist); + dev->queuelist = NULL; + } + dev->queue_count = 0; + + if (drm_core_check_feature(dev, DRIVER_HAVE_DMA) && + !drm_core_check_feature(dev, DRIVER_MODESET)) + drm_dma_takedown(dev); + + dev->dev_mapping = NULL; + mutex_unlock(&dev->struct_mutex); + + DRM_DEBUG("lastclose completed\n"); + return 0; +} + +/** File operations structure */ +static const struct file_operations drm_stub_fops = { + .owner = THIS_MODULE, + .open = drm_stub_open, + .llseek = noop_llseek, +}; + +static int __init drm_core_init(void) +{ + int ret = -ENOMEM; + + drm_global_init(); + idr_init(&drm_minors_idr); + + if (register_chrdev(DRM_MAJOR, "drm", &drm_stub_fops)) + goto err_p1; + + drm_class = drm_sysfs_create(THIS_MODULE, "drm"); + if (IS_ERR(drm_class)) { + printk(KERN_ERR "DRM: Error creating drm class.\n"); + ret = PTR_ERR(drm_class); + goto err_p2; + } + + drm_proc_root = proc_mkdir("dri", NULL); + if (!drm_proc_root) { + DRM_ERROR("Cannot create /proc/dri\n"); + ret = -1; + goto err_p3; + } + + drm_debugfs_root = debugfs_create_dir("dri", NULL); + if (!drm_debugfs_root) { + DRM_ERROR("Cannot create /sys/kernel/debug/dri\n"); + ret = -1; + goto err_p3; + } + + DRM_INFO("Initialized %s %d.%d.%d %s\n", + CORE_NAME, CORE_MAJOR, CORE_MINOR, CORE_PATCHLEVEL, CORE_DATE); + return 0; +err_p3: + drm_sysfs_destroy(); +err_p2: + unregister_chrdev(DRM_MAJOR, "drm"); + + idr_destroy(&drm_minors_idr); +err_p1: + return ret; +} + +static void __exit drm_core_exit(void) +{ + remove_proc_entry("dri", NULL); + debugfs_remove(drm_debugfs_root); + drm_sysfs_destroy(); + + unregister_chrdev(DRM_MAJOR, "drm"); + + idr_remove_all(&drm_minors_idr); + idr_destroy(&drm_minors_idr); +} + +module_init(drm_core_init); +module_exit(drm_core_exit); + +/** + * Copy and IOCTL return string to user space + */ +static int drm_copy_field(char *buf, size_t *buf_len, const char *value) +{ + int len; + + /* don't overflow userbuf */ + len = strlen(value); + if (len > *buf_len) + len = *buf_len; + + /* let userspace know exact length of driver value (which could be + * larger than the userspace-supplied buffer) */ + *buf_len = strlen(value); + + /* finally, try filling in the userbuf */ + if (len && buf) + if (copy_to_user(buf, value, len)) + return -EFAULT; + return 0; +} + +/** + * Get version information + * + * \param inode device inode. + * \param filp file pointer. + * \param cmd command. + * \param arg user argument, pointing to a drm_version structure. + * \return zero on success or negative number on failure. + * + * Fills in the version information in \p arg. + */ +static int drm_version(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_version *version = data; + int err; + + version->version_major = dev->driver->major; + version->version_minor = dev->driver->minor; + version->version_patchlevel = dev->driver->patchlevel; + err = drm_copy_field(version->name, &version->name_len, + dev->driver->name); + if (!err) + err = drm_copy_field(version->date, &version->date_len, + dev->driver->date); + if (!err) + err = drm_copy_field(version->desc, &version->desc_len, + dev->driver->desc); + + return err; +} + +/** + * Called whenever a process performs an ioctl on /dev/drm. + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg user argument. + * \return zero on success or negative number on failure. + * + * Looks up the ioctl function in the ::ioctls table, checking for root + * previleges if so required, and dispatches to the respective function. + */ +long drm_ioctl(struct file *filp, + unsigned int cmd, unsigned long arg) +{ + struct drm_file *file_priv = filp->private_data; + struct drm_device *dev; + struct drm_ioctl_desc *ioctl; + drm_ioctl_t *func; + unsigned int nr = DRM_IOCTL_NR(cmd); + int retcode = -EINVAL; + char stack_kdata[128]; + char *kdata = NULL; + unsigned int usize, asize; + + dev = file_priv->minor->dev; + atomic_inc(&dev->ioctl_count); + atomic_inc(&dev->counts[_DRM_STAT_IOCTLS]); + ++file_priv->ioctl_count; + + DRM_DEBUG("pid=%d, cmd=0x%02x, nr=0x%02x, dev 0x%lx, auth=%d\n", + task_pid_nr(current), cmd, nr, + (long)old_encode_dev(file_priv->minor->device), + file_priv->authenticated); + + if ((nr >= DRM_CORE_IOCTL_COUNT) && + ((nr < DRM_COMMAND_BASE) || (nr >= DRM_COMMAND_END))) + goto err_i1; + if ((nr >= DRM_COMMAND_BASE) && (nr < DRM_COMMAND_END) && + (nr < DRM_COMMAND_BASE + dev->driver->num_ioctls)) { + u32 drv_size; + ioctl = &dev->driver->ioctls[nr - DRM_COMMAND_BASE]; + drv_size = _IOC_SIZE(ioctl->cmd_drv); + usize = asize = _IOC_SIZE(cmd); + if (drv_size > asize) + asize = drv_size; + } + else if ((nr >= DRM_COMMAND_END) || (nr < DRM_COMMAND_BASE)) { + ioctl = &drm_ioctls[nr]; + cmd = ioctl->cmd; + usize = asize = _IOC_SIZE(cmd); + } else + goto err_i1; + + /* Do not trust userspace, use our own definition */ + func = ioctl->func; + /* is there a local override? */ + if ((nr == DRM_IOCTL_NR(DRM_IOCTL_DMA)) && dev->driver->dma_ioctl) + func = dev->driver->dma_ioctl; + + if (!func) { + DRM_DEBUG("no function\n"); + retcode = -EINVAL; + } else if (((ioctl->flags & DRM_ROOT_ONLY) && !capable(CAP_SYS_ADMIN)) || + ((ioctl->flags & DRM_AUTH) && !file_priv->authenticated) || + ((ioctl->flags & DRM_MASTER) && !file_priv->is_master) || + (!(ioctl->flags & DRM_CONTROL_ALLOW) && (file_priv->minor->type == DRM_MINOR_CONTROL))) { + retcode = -EACCES; + } else { + if (cmd & (IOC_IN | IOC_OUT)) { + if (asize <= sizeof(stack_kdata)) { + kdata = stack_kdata; + } else { + kdata = kmalloc(asize, GFP_KERNEL); + if (!kdata) { + retcode = -ENOMEM; + goto err_i1; + } + } + } + + if (cmd & IOC_IN) { + if (copy_from_user(kdata, (void __user *)arg, + usize) != 0) { + retcode = -EFAULT; + goto err_i1; + } + } else + memset(kdata, 0, usize); + + if (ioctl->flags & DRM_UNLOCKED) + retcode = func(dev, kdata, file_priv); + else { + mutex_lock(&drm_global_mutex); + retcode = func(dev, kdata, file_priv); + mutex_unlock(&drm_global_mutex); + } + + if (cmd & IOC_OUT) { + if (copy_to_user((void __user *)arg, kdata, + usize) != 0) + retcode = -EFAULT; + } + } + + err_i1: + if (kdata != stack_kdata) + kfree(kdata); + atomic_dec(&dev->ioctl_count); + if (retcode) + DRM_DEBUG("ret = %x\n", retcode); + return retcode; +} + +EXPORT_SYMBOL(drm_ioctl); + +struct drm_local_map *drm_getsarea(struct drm_device *dev) +{ + struct drm_map_list *entry; + + list_for_each_entry(entry, &dev->maplist, head) { + if (entry->map && entry->map->type == _DRM_SHM && + (entry->map->flags & _DRM_CONTAINS_LOCK)) { + return entry->map; + } + } + return NULL; +} +EXPORT_SYMBOL(drm_getsarea); diff --git a/drivers/gpu/drm/drm_edid.c b/drivers/gpu/drm/drm_edid.c new file mode 100644 index 00000000..1bbb85b9 --- /dev/null +++ b/drivers/gpu/drm/drm_edid.c @@ -0,0 +1,1600 @@ +/* + * Copyright (c) 2006 Luc Verhaegen (quirks list) + * Copyright (c) 2007-2008 Intel Corporation + * Jesse Barnes + * Copyright 2010 Red Hat, Inc. + * + * DDC probing routines (drm_ddc_read & drm_do_probe_ddc_edid) originally from + * FB layer. + * Copyright (C) 2006 Dennis Munsie + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sub license, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + */ +#include +#include +#include +#include "drmP.h" +#include "drm_edid.h" +#include "drm_edid_modes.h" + +#define version_greater(edid, maj, min) \ + (((edid)->version > (maj)) || \ + ((edid)->version == (maj) && (edid)->revision > (min))) + +#define EDID_EST_TIMINGS 16 +#define EDID_STD_TIMINGS 8 +#define EDID_DETAILED_TIMINGS 4 + +/* + * EDID blocks out in the wild have a variety of bugs, try to collect + * them here (note that userspace may work around broken monitors first, + * but fixes should make their way here so that the kernel "just works" + * on as many displays as possible). + */ + +/* First detailed mode wrong, use largest 60Hz mode */ +#define EDID_QUIRK_PREFER_LARGE_60 (1 << 0) +/* Reported 135MHz pixel clock is too high, needs adjustment */ +#define EDID_QUIRK_135_CLOCK_TOO_HIGH (1 << 1) +/* Prefer the largest mode at 75 Hz */ +#define EDID_QUIRK_PREFER_LARGE_75 (1 << 2) +/* Detail timing is in cm not mm */ +#define EDID_QUIRK_DETAILED_IN_CM (1 << 3) +/* Detailed timing descriptors have bogus size values, so just take the + * maximum size and use that. + */ +#define EDID_QUIRK_DETAILED_USE_MAXIMUM_SIZE (1 << 4) +/* Monitor forgot to set the first detailed is preferred bit. */ +#define EDID_QUIRK_FIRST_DETAILED_PREFERRED (1 << 5) +/* use +hsync +vsync for detailed mode */ +#define EDID_QUIRK_DETAILED_SYNC_PP (1 << 6) + +struct detailed_mode_closure { + struct drm_connector *connector; + struct edid *edid; + bool preferred; + u32 quirks; + int modes; +}; + +#define LEVEL_DMT 0 +#define LEVEL_GTF 1 +#define LEVEL_GTF2 2 +#define LEVEL_CVT 3 + +static struct edid_quirk { + char *vendor; + int product_id; + u32 quirks; +} edid_quirk_list[] = { + /* Acer AL1706 */ + { "ACR", 44358, EDID_QUIRK_PREFER_LARGE_60 }, + /* Acer F51 */ + { "API", 0x7602, EDID_QUIRK_PREFER_LARGE_60 }, + /* Unknown Acer */ + { "ACR", 2423, EDID_QUIRK_FIRST_DETAILED_PREFERRED }, + + /* Belinea 10 15 55 */ + { "MAX", 1516, EDID_QUIRK_PREFER_LARGE_60 }, + { "MAX", 0x77e, EDID_QUIRK_PREFER_LARGE_60 }, + + /* Envision Peripherals, Inc. EN-7100e */ + { "EPI", 59264, EDID_QUIRK_135_CLOCK_TOO_HIGH }, + /* Envision EN2028 */ + { "EPI", 8232, EDID_QUIRK_PREFER_LARGE_60 }, + + /* Funai Electronics PM36B */ + { "FCM", 13600, EDID_QUIRK_PREFER_LARGE_75 | + EDID_QUIRK_DETAILED_IN_CM }, + + /* LG Philips LCD LP154W01-A5 */ + { "LPL", 0, EDID_QUIRK_DETAILED_USE_MAXIMUM_SIZE }, + { "LPL", 0x2a00, EDID_QUIRK_DETAILED_USE_MAXIMUM_SIZE }, + + /* Philips 107p5 CRT */ + { "PHL", 57364, EDID_QUIRK_FIRST_DETAILED_PREFERRED }, + + /* Proview AY765C */ + { "PTS", 765, EDID_QUIRK_FIRST_DETAILED_PREFERRED }, + + /* Samsung SyncMaster 205BW. Note: irony */ + { "SAM", 541, EDID_QUIRK_DETAILED_SYNC_PP }, + /* Samsung SyncMaster 22[5-6]BW */ + { "SAM", 596, EDID_QUIRK_PREFER_LARGE_60 }, + { "SAM", 638, EDID_QUIRK_PREFER_LARGE_60 }, +}; + +/*** DDC fetch and block validation ***/ + +static const u8 edid_header[] = { + 0x00, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0x00 +}; + + /* + * Sanity check the header of the base EDID block. Return 8 if the header + * is perfect, down to 0 if it's totally wrong. + */ +int drm_edid_header_is_valid(const u8 *raw_edid) +{ + int i, score = 0; + + for (i = 0; i < sizeof(edid_header); i++) + if (raw_edid[i] == edid_header[i]) + score++; + + return score; +} +EXPORT_SYMBOL(drm_edid_header_is_valid); + + +/* + * Sanity check the EDID block (base or extension). Return 0 if the block + * doesn't check out, or 1 if it's valid. + */ +static bool +drm_edid_block_valid(u8 *raw_edid) +{ + int i; + u8 csum = 0; + struct edid *edid = (struct edid *)raw_edid; + + if (raw_edid[0] == 0x00) { + int score = drm_edid_header_is_valid(raw_edid); + if (score == 8) ; + else if (score >= 6) { + DRM_DEBUG("Fixing EDID header, your hardware may be failing\n"); + memcpy(raw_edid, edid_header, sizeof(edid_header)); + } else { + goto bad; + } + } + + for (i = 0; i < EDID_LENGTH; i++) + csum += raw_edid[i]; + if (csum) { + DRM_ERROR("EDID checksum is invalid, remainder is %d\n", csum); + + /* allow CEA to slide through, switches mangle this */ + if (raw_edid[0] != 0x02) + goto bad; + } + + /* per-block-type checks */ + switch (raw_edid[0]) { + case 0: /* base */ + if (edid->version != 1) { + DRM_ERROR("EDID has major version %d, instead of 1\n", edid->version); + goto bad; + } + + if (edid->revision > 4) + DRM_DEBUG("EDID minor > 4, assuming backward compatibility\n"); + break; + + default: + break; + } + + return 1; + +bad: + if (raw_edid) { + printk(KERN_ERR "Raw EDID:\n"); + print_hex_dump_bytes(KERN_ERR, DUMP_PREFIX_NONE, raw_edid, EDID_LENGTH); + printk(KERN_ERR "\n"); + } + return 0; +} + +/** + * drm_edid_is_valid - sanity check EDID data + * @edid: EDID data + * + * Sanity-check an entire EDID record (including extensions) + */ +bool drm_edid_is_valid(struct edid *edid) +{ + int i; + u8 *raw = (u8 *)edid; + + if (!edid) + return false; + + for (i = 0; i <= edid->extensions; i++) + if (!drm_edid_block_valid(raw + i * EDID_LENGTH)) + return false; + + return true; +} +EXPORT_SYMBOL(drm_edid_is_valid); + +#define DDC_ADDR 0x50 +#define DDC_SEGMENT_ADDR 0x30 +/** + * Get EDID information via I2C. + * + * \param adapter : i2c device adaptor + * \param buf : EDID data buffer to be filled + * \param len : EDID data buffer length + * \return 0 on success or -1 on failure. + * + * Try to fetch EDID information by calling i2c driver function. + */ +static int +drm_do_probe_ddc_edid(struct i2c_adapter *adapter, unsigned char *buf, + int block, int len) +{ + unsigned char start = block * EDID_LENGTH; + int ret, retries = 5; + + /* The core i2c driver will automatically retry the transfer if the + * adapter reports EAGAIN. However, we find that bit-banging transfers + * are susceptible to errors under a heavily loaded machine and + * generate spurious NAKs and timeouts. Retrying the transfer + * of the individual block a few times seems to overcome this. + */ + do { + struct i2c_msg msgs[] = { + { + .addr = DDC_ADDR, + .flags = 0, + .len = 1, + .buf = &start, + }, { + .addr = DDC_ADDR, + .flags = I2C_M_RD, + .len = len, + .buf = buf, + } + }; + ret = i2c_transfer(adapter, msgs, 2); + } while (ret != 2 && --retries); + + return ret == 2 ? 0 : -1; +} + +static bool drm_edid_is_zero(u8 *in_edid, int length) +{ + int i; + u32 *raw_edid = (u32 *)in_edid; + + for (i = 0; i < length / 4; i++) + if (*(raw_edid + i) != 0) + return false; + return true; +} + +static u8 * +drm_do_get_edid(struct drm_connector *connector, struct i2c_adapter *adapter) +{ + int i, j = 0, valid_extensions = 0; + u8 *block, *new; + + if ((block = kmalloc(EDID_LENGTH, GFP_KERNEL)) == NULL) + return NULL; + + /* base block fetch */ + for (i = 0; i < 4; i++) { + if (drm_do_probe_ddc_edid(adapter, block, 0, EDID_LENGTH)) + goto out; + if (drm_edid_block_valid(block)) + break; + if (i == 0 && drm_edid_is_zero(block, EDID_LENGTH)) { + connector->null_edid_counter++; + goto carp; + } + } + if (i == 4) + goto carp; + + /* if there's no extensions, we're done */ + if (block[0x7e] == 0) + return block; + + new = krealloc(block, (block[0x7e] + 1) * EDID_LENGTH, GFP_KERNEL); + if (!new) + goto out; + block = new; + + for (j = 1; j <= block[0x7e]; j++) { + for (i = 0; i < 4; i++) { + if (drm_do_probe_ddc_edid(adapter, + block + (valid_extensions + 1) * EDID_LENGTH, + j, EDID_LENGTH)) + goto out; + if (drm_edid_block_valid(block + (valid_extensions + 1) * EDID_LENGTH)) { + valid_extensions++; + break; + } + } + if (i == 4) + dev_warn(connector->dev->dev, + "%s: Ignoring invalid EDID block %d.\n", + drm_get_connector_name(connector), j); + } + + if (valid_extensions != block[0x7e]) { + block[EDID_LENGTH-1] += block[0x7e] - valid_extensions; + block[0x7e] = valid_extensions; + new = krealloc(block, (valid_extensions + 1) * EDID_LENGTH, GFP_KERNEL); + if (!new) + goto out; + block = new; + } + + return block; + +carp: + dev_warn(connector->dev->dev, "%s: EDID block %d invalid.\n", + drm_get_connector_name(connector), j); + +out: + kfree(block); + return NULL; +} + +/** + * Probe DDC presence. + * + * \param adapter : i2c device adaptor + * \return 1 on success + */ +static bool +drm_probe_ddc(struct i2c_adapter *adapter) +{ + unsigned char out; + + return (drm_do_probe_ddc_edid(adapter, &out, 0, 1) == 0); +} + +/** + * drm_get_edid - get EDID data, if available + * @connector: connector we're probing + * @adapter: i2c adapter to use for DDC + * + * Poke the given i2c channel to grab EDID data if possible. If found, + * attach it to the connector. + * + * Return edid data or NULL if we couldn't find any. + */ +struct edid *drm_get_edid(struct drm_connector *connector, + struct i2c_adapter *adapter) +{ + struct edid *edid = NULL; + + if (drm_probe_ddc(adapter)) + edid = (struct edid *)drm_do_get_edid(connector, adapter); + + connector->display_info.raw_edid = (char *)edid; + + return edid; + +} +EXPORT_SYMBOL(drm_get_edid); + +/*** EDID parsing ***/ + +/** + * edid_vendor - match a string against EDID's obfuscated vendor field + * @edid: EDID to match + * @vendor: vendor string + * + * Returns true if @vendor is in @edid, false otherwise + */ +static bool edid_vendor(struct edid *edid, char *vendor) +{ + char edid_vendor[3]; + + edid_vendor[0] = ((edid->mfg_id[0] & 0x7c) >> 2) + '@'; + edid_vendor[1] = (((edid->mfg_id[0] & 0x3) << 3) | + ((edid->mfg_id[1] & 0xe0) >> 5)) + '@'; + edid_vendor[2] = (edid->mfg_id[1] & 0x1f) + '@'; + + return !strncmp(edid_vendor, vendor, 3); +} + +/** + * edid_get_quirks - return quirk flags for a given EDID + * @edid: EDID to process + * + * This tells subsequent routines what fixes they need to apply. + */ +static u32 edid_get_quirks(struct edid *edid) +{ + struct edid_quirk *quirk; + int i; + + for (i = 0; i < ARRAY_SIZE(edid_quirk_list); i++) { + quirk = &edid_quirk_list[i]; + + if (edid_vendor(edid, quirk->vendor) && + (EDID_PRODUCT_ID(edid) == quirk->product_id)) + return quirk->quirks; + } + + return 0; +} + +#define MODE_SIZE(m) ((m)->hdisplay * (m)->vdisplay) +#define MODE_REFRESH_DIFF(m,r) (abs((m)->vrefresh - target_refresh)) + +/** + * edid_fixup_preferred - set preferred modes based on quirk list + * @connector: has mode list to fix up + * @quirks: quirks list + * + * Walk the mode list for @connector, clearing the preferred status + * on existing modes and setting it anew for the right mode ala @quirks. + */ +static void edid_fixup_preferred(struct drm_connector *connector, + u32 quirks) +{ + struct drm_display_mode *t, *cur_mode, *preferred_mode; + int target_refresh = 0; + + if (list_empty(&connector->probed_modes)) + return; + + if (quirks & EDID_QUIRK_PREFER_LARGE_60) + target_refresh = 60; + if (quirks & EDID_QUIRK_PREFER_LARGE_75) + target_refresh = 75; + + preferred_mode = list_first_entry(&connector->probed_modes, + struct drm_display_mode, head); + + list_for_each_entry_safe(cur_mode, t, &connector->probed_modes, head) { + cur_mode->type &= ~DRM_MODE_TYPE_PREFERRED; + + if (cur_mode == preferred_mode) + continue; + + /* Largest mode is preferred */ + if (MODE_SIZE(cur_mode) > MODE_SIZE(preferred_mode)) + preferred_mode = cur_mode; + + /* At a given size, try to get closest to target refresh */ + if ((MODE_SIZE(cur_mode) == MODE_SIZE(preferred_mode)) && + MODE_REFRESH_DIFF(cur_mode, target_refresh) < + MODE_REFRESH_DIFF(preferred_mode, target_refresh)) { + preferred_mode = cur_mode; + } + } + + preferred_mode->type |= DRM_MODE_TYPE_PREFERRED; +} + +struct drm_display_mode *drm_mode_find_dmt(struct drm_device *dev, + int hsize, int vsize, int fresh) +{ + struct drm_display_mode *mode = NULL; + int i; + + for (i = 0; i < drm_num_dmt_modes; i++) { + const struct drm_display_mode *ptr = &drm_dmt_modes[i]; + if (hsize == ptr->hdisplay && + vsize == ptr->vdisplay && + fresh == drm_mode_vrefresh(ptr)) { + /* get the expected default mode */ + mode = drm_mode_duplicate(dev, ptr); + break; + } + } + return mode; +} +EXPORT_SYMBOL(drm_mode_find_dmt); + +typedef void detailed_cb(struct detailed_timing *timing, void *closure); + +static void +cea_for_each_detailed_block(u8 *ext, detailed_cb *cb, void *closure) +{ + int i, n = 0; + u8 rev = ext[0x01], d = ext[0x02]; + u8 *det_base = ext + d; + + switch (rev) { + case 0: + /* can't happen */ + return; + case 1: + /* have to infer how many blocks we have, check pixel clock */ + for (i = 0; i < 6; i++) + if (det_base[18*i] || det_base[18*i+1]) + n++; + break; + default: + /* explicit count */ + n = min(ext[0x03] & 0x0f, 6); + break; + } + + for (i = 0; i < n; i++) + cb((struct detailed_timing *)(det_base + 18 * i), closure); +} + +static void +vtb_for_each_detailed_block(u8 *ext, detailed_cb *cb, void *closure) +{ + unsigned int i, n = min((int)ext[0x02], 6); + u8 *det_base = ext + 5; + + if (ext[0x01] != 1) + return; /* unknown version */ + + for (i = 0; i < n; i++) + cb((struct detailed_timing *)(det_base + 18 * i), closure); +} + +static void +drm_for_each_detailed_block(u8 *raw_edid, detailed_cb *cb, void *closure) +{ + int i; + struct edid *edid = (struct edid *)raw_edid; + + if (edid == NULL) + return; + + for (i = 0; i < EDID_DETAILED_TIMINGS; i++) + cb(&(edid->detailed_timings[i]), closure); + + for (i = 1; i <= raw_edid[0x7e]; i++) { + u8 *ext = raw_edid + (i * EDID_LENGTH); + switch (*ext) { + case CEA_EXT: + cea_for_each_detailed_block(ext, cb, closure); + break; + case VTB_EXT: + vtb_for_each_detailed_block(ext, cb, closure); + break; + default: + break; + } + } +} + +static void +is_rb(struct detailed_timing *t, void *data) +{ + u8 *r = (u8 *)t; + if (r[3] == EDID_DETAIL_MONITOR_RANGE) + if (r[15] & 0x10) + *(bool *)data = true; +} + +/* EDID 1.4 defines this explicitly. For EDID 1.3, we guess, badly. */ +static bool +drm_monitor_supports_rb(struct edid *edid) +{ + if (edid->revision >= 4) { + bool ret; + drm_for_each_detailed_block((u8 *)edid, is_rb, &ret); + return ret; + } + + return ((edid->input & DRM_EDID_INPUT_DIGITAL) != 0); +} + +static void +find_gtf2(struct detailed_timing *t, void *data) +{ + u8 *r = (u8 *)t; + if (r[3] == EDID_DETAIL_MONITOR_RANGE && r[10] == 0x02) + *(u8 **)data = r; +} + +/* Secondary GTF curve kicks in above some break frequency */ +static int +drm_gtf2_hbreak(struct edid *edid) +{ + u8 *r = NULL; + drm_for_each_detailed_block((u8 *)edid, find_gtf2, &r); + return r ? (r[12] * 2) : 0; +} + +static int +drm_gtf2_2c(struct edid *edid) +{ + u8 *r = NULL; + drm_for_each_detailed_block((u8 *)edid, find_gtf2, &r); + return r ? r[13] : 0; +} + +static int +drm_gtf2_m(struct edid *edid) +{ + u8 *r = NULL; + drm_for_each_detailed_block((u8 *)edid, find_gtf2, &r); + return r ? (r[15] << 8) + r[14] : 0; +} + +static int +drm_gtf2_k(struct edid *edid) +{ + u8 *r = NULL; + drm_for_each_detailed_block((u8 *)edid, find_gtf2, &r); + return r ? r[16] : 0; +} + +static int +drm_gtf2_2j(struct edid *edid) +{ + u8 *r = NULL; + drm_for_each_detailed_block((u8 *)edid, find_gtf2, &r); + return r ? r[17] : 0; +} + +/** + * standard_timing_level - get std. timing level(CVT/GTF/DMT) + * @edid: EDID block to scan + */ +static int standard_timing_level(struct edid *edid) +{ + if (edid->revision >= 2) { + if (edid->revision >= 4 && (edid->features & DRM_EDID_FEATURE_DEFAULT_GTF)) + return LEVEL_CVT; + if (drm_gtf2_hbreak(edid)) + return LEVEL_GTF2; + return LEVEL_GTF; + } + return LEVEL_DMT; +} + +/* + * 0 is reserved. The spec says 0x01 fill for unused timings. Some old + * monitors fill with ascii space (0x20) instead. + */ +static int +bad_std_timing(u8 a, u8 b) +{ + return (a == 0x00 && b == 0x00) || + (a == 0x01 && b == 0x01) || + (a == 0x20 && b == 0x20); +} + +/** + * drm_mode_std - convert standard mode info (width, height, refresh) into mode + * @t: standard timing params + * @timing_level: standard timing level + * + * Take the standard timing params (in this case width, aspect, and refresh) + * and convert them into a real mode using CVT/GTF/DMT. + */ +static struct drm_display_mode * +drm_mode_std(struct drm_connector *connector, struct edid *edid, + struct std_timing *t, int revision) +{ + struct drm_device *dev = connector->dev; + struct drm_display_mode *m, *mode = NULL; + int hsize, vsize; + int vrefresh_rate; + unsigned aspect_ratio = (t->vfreq_aspect & EDID_TIMING_ASPECT_MASK) + >> EDID_TIMING_ASPECT_SHIFT; + unsigned vfreq = (t->vfreq_aspect & EDID_TIMING_VFREQ_MASK) + >> EDID_TIMING_VFREQ_SHIFT; + int timing_level = standard_timing_level(edid); + + if (bad_std_timing(t->hsize, t->vfreq_aspect)) + return NULL; + + /* According to the EDID spec, the hdisplay = hsize * 8 + 248 */ + hsize = t->hsize * 8 + 248; + /* vrefresh_rate = vfreq + 60 */ + vrefresh_rate = vfreq + 60; + /* the vdisplay is calculated based on the aspect ratio */ + if (aspect_ratio == 0) { + if (revision < 3) + vsize = hsize; + else + vsize = (hsize * 10) / 16; + } else if (aspect_ratio == 1) + vsize = (hsize * 3) / 4; + else if (aspect_ratio == 2) + vsize = (hsize * 4) / 5; + else + vsize = (hsize * 9) / 16; + + /* HDTV hack, part 1 */ + if (vrefresh_rate == 60 && + ((hsize == 1360 && vsize == 765) || + (hsize == 1368 && vsize == 769))) { + hsize = 1366; + vsize = 768; + } + + /* + * If this connector already has a mode for this size and refresh + * rate (because it came from detailed or CVT info), use that + * instead. This way we don't have to guess at interlace or + * reduced blanking. + */ + list_for_each_entry(m, &connector->probed_modes, head) + if (m->hdisplay == hsize && m->vdisplay == vsize && + drm_mode_vrefresh(m) == vrefresh_rate) + return NULL; + + /* HDTV hack, part 2 */ + if (hsize == 1366 && vsize == 768 && vrefresh_rate == 60) { + mode = drm_cvt_mode(dev, 1366, 768, vrefresh_rate, 0, 0, + false); + mode->hdisplay = 1366; + mode->hsync_start = mode->hsync_start - 1; + mode->hsync_end = mode->hsync_end - 1; + return mode; + } + + /* check whether it can be found in default mode table */ + mode = drm_mode_find_dmt(dev, hsize, vsize, vrefresh_rate); + if (mode) + return mode; + + switch (timing_level) { + case LEVEL_DMT: + break; + case LEVEL_GTF: + mode = drm_gtf_mode(dev, hsize, vsize, vrefresh_rate, 0, 0); + break; + case LEVEL_GTF2: + /* + * This is potentially wrong if there's ever a monitor with + * more than one ranges section, each claiming a different + * secondary GTF curve. Please don't do that. + */ + mode = drm_gtf_mode(dev, hsize, vsize, vrefresh_rate, 0, 0); + if (drm_mode_hsync(mode) > drm_gtf2_hbreak(edid)) { + kfree(mode); + mode = drm_gtf_mode_complex(dev, hsize, vsize, + vrefresh_rate, 0, 0, + drm_gtf2_m(edid), + drm_gtf2_2c(edid), + drm_gtf2_k(edid), + drm_gtf2_2j(edid)); + } + break; + case LEVEL_CVT: + mode = drm_cvt_mode(dev, hsize, vsize, vrefresh_rate, 0, 0, + false); + break; + } + return mode; +} + +/* + * EDID is delightfully ambiguous about how interlaced modes are to be + * encoded. Our internal representation is of frame height, but some + * HDTV detailed timings are encoded as field height. + * + * The format list here is from CEA, in frame size. Technically we + * should be checking refresh rate too. Whatever. + */ +static void +drm_mode_do_interlace_quirk(struct drm_display_mode *mode, + struct detailed_pixel_timing *pt) +{ + int i; + static const struct { + int w, h; + } cea_interlaced[] = { + { 1920, 1080 }, + { 720, 480 }, + { 1440, 480 }, + { 2880, 480 }, + { 720, 576 }, + { 1440, 576 }, + { 2880, 576 }, + }; + + if (!(pt->misc & DRM_EDID_PT_INTERLACED)) + return; + + for (i = 0; i < ARRAY_SIZE(cea_interlaced); i++) { + if ((mode->hdisplay == cea_interlaced[i].w) && + (mode->vdisplay == cea_interlaced[i].h / 2)) { + mode->vdisplay *= 2; + mode->vsync_start *= 2; + mode->vsync_end *= 2; + mode->vtotal *= 2; + mode->vtotal |= 1; + } + } + + mode->flags |= DRM_MODE_FLAG_INTERLACE; +} + +/** + * drm_mode_detailed - create a new mode from an EDID detailed timing section + * @dev: DRM device (needed to create new mode) + * @edid: EDID block + * @timing: EDID detailed timing info + * @quirks: quirks to apply + * + * An EDID detailed timing block contains enough info for us to create and + * return a new struct drm_display_mode. + */ +static struct drm_display_mode *drm_mode_detailed(struct drm_device *dev, + struct edid *edid, + struct detailed_timing *timing, + u32 quirks) +{ + struct drm_display_mode *mode; + struct detailed_pixel_timing *pt = &timing->data.pixel_data; + unsigned hactive = (pt->hactive_hblank_hi & 0xf0) << 4 | pt->hactive_lo; + unsigned vactive = (pt->vactive_vblank_hi & 0xf0) << 4 | pt->vactive_lo; + unsigned hblank = (pt->hactive_hblank_hi & 0xf) << 8 | pt->hblank_lo; + unsigned vblank = (pt->vactive_vblank_hi & 0xf) << 8 | pt->vblank_lo; + unsigned hsync_offset = (pt->hsync_vsync_offset_pulse_width_hi & 0xc0) << 2 | pt->hsync_offset_lo; + unsigned hsync_pulse_width = (pt->hsync_vsync_offset_pulse_width_hi & 0x30) << 4 | pt->hsync_pulse_width_lo; + unsigned vsync_offset = (pt->hsync_vsync_offset_pulse_width_hi & 0xc) >> 2 | pt->vsync_offset_pulse_width_lo >> 4; + unsigned vsync_pulse_width = (pt->hsync_vsync_offset_pulse_width_hi & 0x3) << 4 | (pt->vsync_offset_pulse_width_lo & 0xf); + + /* ignore tiny modes */ + if (hactive < 64 || vactive < 64) + return NULL; + + if (pt->misc & DRM_EDID_PT_STEREO) { + printk(KERN_WARNING "stereo mode not supported\n"); + return NULL; + } + if (!(pt->misc & DRM_EDID_PT_SEPARATE_SYNC)) { + printk(KERN_WARNING "composite sync not supported\n"); + } + + /* it is incorrect if hsync/vsync width is zero */ + if (!hsync_pulse_width || !vsync_pulse_width) { + DRM_DEBUG_KMS("Incorrect Detailed timing. " + "Wrong Hsync/Vsync pulse width\n"); + return NULL; + } + mode = drm_mode_create(dev); + if (!mode) + return NULL; + + mode->type = DRM_MODE_TYPE_DRIVER; + + if (quirks & EDID_QUIRK_135_CLOCK_TOO_HIGH) + timing->pixel_clock = cpu_to_le16(1088); + + mode->clock = le16_to_cpu(timing->pixel_clock) * 10; + + mode->hdisplay = hactive; + mode->hsync_start = mode->hdisplay + hsync_offset; + mode->hsync_end = mode->hsync_start + hsync_pulse_width; + mode->htotal = mode->hdisplay + hblank; + + mode->vdisplay = vactive; + mode->vsync_start = mode->vdisplay + vsync_offset; + mode->vsync_end = mode->vsync_start + vsync_pulse_width; + mode->vtotal = mode->vdisplay + vblank; + + /* Some EDIDs have bogus h/vtotal values */ + if (mode->hsync_end > mode->htotal) + mode->htotal = mode->hsync_end + 1; + if (mode->vsync_end > mode->vtotal) + mode->vtotal = mode->vsync_end + 1; + + drm_mode_do_interlace_quirk(mode, pt); + + drm_mode_set_name(mode); + + if (quirks & EDID_QUIRK_DETAILED_SYNC_PP) { + pt->misc |= DRM_EDID_PT_HSYNC_POSITIVE | DRM_EDID_PT_VSYNC_POSITIVE; + } + + mode->flags |= (pt->misc & DRM_EDID_PT_HSYNC_POSITIVE) ? + DRM_MODE_FLAG_PHSYNC : DRM_MODE_FLAG_NHSYNC; + mode->flags |= (pt->misc & DRM_EDID_PT_VSYNC_POSITIVE) ? + DRM_MODE_FLAG_PVSYNC : DRM_MODE_FLAG_NVSYNC; + + mode->width_mm = pt->width_mm_lo | (pt->width_height_mm_hi & 0xf0) << 4; + mode->height_mm = pt->height_mm_lo | (pt->width_height_mm_hi & 0xf) << 8; + + if (quirks & EDID_QUIRK_DETAILED_IN_CM) { + mode->width_mm *= 10; + mode->height_mm *= 10; + } + + if (quirks & EDID_QUIRK_DETAILED_USE_MAXIMUM_SIZE) { + mode->width_mm = edid->width_cm * 10; + mode->height_mm = edid->height_cm * 10; + } + + return mode; +} + +static bool +mode_is_rb(const struct drm_display_mode *mode) +{ + return (mode->htotal - mode->hdisplay == 160) && + (mode->hsync_end - mode->hdisplay == 80) && + (mode->hsync_end - mode->hsync_start == 32) && + (mode->vsync_start - mode->vdisplay == 3); +} + +static bool +mode_in_hsync_range(const struct drm_display_mode *mode, + struct edid *edid, u8 *t) +{ + int hsync, hmin, hmax; + + hmin = t[7]; + if (edid->revision >= 4) + hmin += ((t[4] & 0x04) ? 255 : 0); + hmax = t[8]; + if (edid->revision >= 4) + hmax += ((t[4] & 0x08) ? 255 : 0); + hsync = drm_mode_hsync(mode); + + return (hsync <= hmax && hsync >= hmin); +} + +static bool +mode_in_vsync_range(const struct drm_display_mode *mode, + struct edid *edid, u8 *t) +{ + int vsync, vmin, vmax; + + vmin = t[5]; + if (edid->revision >= 4) + vmin += ((t[4] & 0x01) ? 255 : 0); + vmax = t[6]; + if (edid->revision >= 4) + vmax += ((t[4] & 0x02) ? 255 : 0); + vsync = drm_mode_vrefresh(mode); + + return (vsync <= vmax && vsync >= vmin); +} + +static u32 +range_pixel_clock(struct edid *edid, u8 *t) +{ + /* unspecified */ + if (t[9] == 0 || t[9] == 255) + return 0; + + /* 1.4 with CVT support gives us real precision, yay */ + if (edid->revision >= 4 && t[10] == 0x04) + return (t[9] * 10000) - ((t[12] >> 2) * 250); + + /* 1.3 is pathetic, so fuzz up a bit */ + return t[9] * 10000 + 5001; +} + +static bool +mode_in_range(const struct drm_display_mode *mode, struct edid *edid, + struct detailed_timing *timing) +{ + u32 max_clock; + u8 *t = (u8 *)timing; + + if (!mode_in_hsync_range(mode, edid, t)) + return false; + + if (!mode_in_vsync_range(mode, edid, t)) + return false; + + if ((max_clock = range_pixel_clock(edid, t))) + if (mode->clock > max_clock) + return false; + + /* 1.4 max horizontal check */ + if (edid->revision >= 4 && t[10] == 0x04) + if (t[13] && mode->hdisplay > 8 * (t[13] + (256 * (t[12]&0x3)))) + return false; + + if (mode_is_rb(mode) && !drm_monitor_supports_rb(edid)) + return false; + + return true; +} + +/* + * XXX If drm_dmt_modes ever regrows the CVT-R modes (and it will) this will + * need to account for them. + */ +static int +drm_gtf_modes_for_range(struct drm_connector *connector, struct edid *edid, + struct detailed_timing *timing) +{ + int i, modes = 0; + struct drm_display_mode *newmode; + struct drm_device *dev = connector->dev; + + for (i = 0; i < drm_num_dmt_modes; i++) { + if (mode_in_range(drm_dmt_modes + i, edid, timing)) { + newmode = drm_mode_duplicate(dev, &drm_dmt_modes[i]); + if (newmode) { + drm_mode_probed_add(connector, newmode); + modes++; + } + } + } + + return modes; +} + +static void +do_inferred_modes(struct detailed_timing *timing, void *c) +{ + struct detailed_mode_closure *closure = c; + struct detailed_non_pixel *data = &timing->data.other_data; + int gtf = (closure->edid->features & DRM_EDID_FEATURE_DEFAULT_GTF); + + if (gtf && data->type == EDID_DETAIL_MONITOR_RANGE) + closure->modes += drm_gtf_modes_for_range(closure->connector, + closure->edid, + timing); +} + +static int +add_inferred_modes(struct drm_connector *connector, struct edid *edid) +{ + struct detailed_mode_closure closure = { + connector, edid, 0, 0, 0 + }; + + if (version_greater(edid, 1, 0)) + drm_for_each_detailed_block((u8 *)edid, do_inferred_modes, + &closure); + + return closure.modes; +} + +static int +drm_est3_modes(struct drm_connector *connector, struct detailed_timing *timing) +{ + int i, j, m, modes = 0; + struct drm_display_mode *mode; + u8 *est = ((u8 *)timing) + 5; + + for (i = 0; i < 6; i++) { + for (j = 7; j > 0; j--) { + m = (i * 8) + (7 - j); + if (m >= ARRAY_SIZE(est3_modes)) + break; + if (est[i] & (1 << j)) { + mode = drm_mode_find_dmt(connector->dev, + est3_modes[m].w, + est3_modes[m].h, + est3_modes[m].r + /*, est3_modes[m].rb */); + if (mode) { + drm_mode_probed_add(connector, mode); + modes++; + } + } + } + } + + return modes; +} + +static void +do_established_modes(struct detailed_timing *timing, void *c) +{ + struct detailed_mode_closure *closure = c; + struct detailed_non_pixel *data = &timing->data.other_data; + + if (data->type == EDID_DETAIL_EST_TIMINGS) + closure->modes += drm_est3_modes(closure->connector, timing); +} + +/** + * add_established_modes - get est. modes from EDID and add them + * @edid: EDID block to scan + * + * Each EDID block contains a bitmap of the supported "established modes" list + * (defined above). Tease them out and add them to the global modes list. + */ +static int +add_established_modes(struct drm_connector *connector, struct edid *edid) +{ + struct drm_device *dev = connector->dev; + unsigned long est_bits = edid->established_timings.t1 | + (edid->established_timings.t2 << 8) | + ((edid->established_timings.mfg_rsvd & 0x80) << 9); + int i, modes = 0; + struct detailed_mode_closure closure = { + connector, edid, 0, 0, 0 + }; + + for (i = 0; i <= EDID_EST_TIMINGS; i++) { + if (est_bits & (1<data.other_data; + struct drm_connector *connector = closure->connector; + struct edid *edid = closure->edid; + + if (data->type == EDID_DETAIL_STD_MODES) { + int i; + for (i = 0; i < 6; i++) { + struct std_timing *std; + struct drm_display_mode *newmode; + + std = &data->data.timings[i]; + newmode = drm_mode_std(connector, edid, std, + edid->revision); + if (newmode) { + drm_mode_probed_add(connector, newmode); + closure->modes++; + } + } + } +} + +/** + * add_standard_modes - get std. modes from EDID and add them + * @edid: EDID block to scan + * + * Standard modes can be calculated using the appropriate standard (DMT, + * GTF or CVT. Grab them from @edid and add them to the list. + */ +static int +add_standard_modes(struct drm_connector *connector, struct edid *edid) +{ + int i, modes = 0; + struct detailed_mode_closure closure = { + connector, edid, 0, 0, 0 + }; + + for (i = 0; i < EDID_STD_TIMINGS; i++) { + struct drm_display_mode *newmode; + + newmode = drm_mode_std(connector, edid, + &edid->standard_timings[i], + edid->revision); + if (newmode) { + drm_mode_probed_add(connector, newmode); + modes++; + } + } + + if (version_greater(edid, 1, 0)) + drm_for_each_detailed_block((u8 *)edid, do_standard_modes, + &closure); + + /* XXX should also look for standard codes in VTB blocks */ + + return modes + closure.modes; +} + +static int drm_cvt_modes(struct drm_connector *connector, + struct detailed_timing *timing) +{ + int i, j, modes = 0; + struct drm_display_mode *newmode; + struct drm_device *dev = connector->dev; + struct cvt_timing *cvt; + const int rates[] = { 60, 85, 75, 60, 50 }; + const u8 empty[3] = { 0, 0, 0 }; + + for (i = 0; i < 4; i++) { + int uninitialized_var(width), height; + cvt = &(timing->data.other_data.data.cvt[i]); + + if (!memcmp(cvt->code, empty, 3)) + continue; + + height = (cvt->code[0] + ((cvt->code[1] & 0xf0) << 4) + 1) * 2; + switch (cvt->code[1] & 0x0c) { + case 0x00: + width = height * 4 / 3; + break; + case 0x04: + width = height * 16 / 9; + break; + case 0x08: + width = height * 16 / 10; + break; + case 0x0c: + width = height * 15 / 9; + break; + } + + for (j = 1; j < 5; j++) { + if (cvt->code[2] & (1 << j)) { + newmode = drm_cvt_mode(dev, width, height, + rates[j], j == 0, + false, false); + if (newmode) { + drm_mode_probed_add(connector, newmode); + modes++; + } + } + } + } + + return modes; +} + +static void +do_cvt_mode(struct detailed_timing *timing, void *c) +{ + struct detailed_mode_closure *closure = c; + struct detailed_non_pixel *data = &timing->data.other_data; + + if (data->type == EDID_DETAIL_CVT_3BYTE) + closure->modes += drm_cvt_modes(closure->connector, timing); +} + +static int +add_cvt_modes(struct drm_connector *connector, struct edid *edid) +{ + struct detailed_mode_closure closure = { + connector, edid, 0, 0, 0 + }; + + if (version_greater(edid, 1, 2)) + drm_for_each_detailed_block((u8 *)edid, do_cvt_mode, &closure); + + /* XXX should also look for CVT codes in VTB blocks */ + + return closure.modes; +} + +static void +do_detailed_mode(struct detailed_timing *timing, void *c) +{ + struct detailed_mode_closure *closure = c; + struct drm_display_mode *newmode; + + if (timing->pixel_clock) { + newmode = drm_mode_detailed(closure->connector->dev, + closure->edid, timing, + closure->quirks); + if (!newmode) + return; + + if (closure->preferred) + newmode->type |= DRM_MODE_TYPE_PREFERRED; + + drm_mode_probed_add(closure->connector, newmode); + closure->modes++; + closure->preferred = 0; + } +} + +/* + * add_detailed_modes - Add modes from detailed timings + * @connector: attached connector + * @edid: EDID block to scan + * @quirks: quirks to apply + */ +static int +add_detailed_modes(struct drm_connector *connector, struct edid *edid, + u32 quirks) +{ + struct detailed_mode_closure closure = { + connector, + edid, + 1, + quirks, + 0 + }; + + if (closure.preferred && !version_greater(edid, 1, 3)) + closure.preferred = + (edid->features & DRM_EDID_FEATURE_PREFERRED_TIMING); + + drm_for_each_detailed_block((u8 *)edid, do_detailed_mode, &closure); + + return closure.modes; +} + +#define HDMI_IDENTIFIER 0x000C03 +#define AUDIO_BLOCK 0x01 +#define VENDOR_BLOCK 0x03 +#define EDID_BASIC_AUDIO (1 << 6) + +/** + * Search EDID for CEA extension block. + */ +u8 *drm_find_cea_extension(struct edid *edid) +{ + u8 *edid_ext = NULL; + int i; + + /* No EDID or EDID extensions */ + if (edid == NULL || edid->extensions == 0) + return NULL; + + /* Find CEA extension */ + for (i = 0; i < edid->extensions; i++) { + edid_ext = (u8 *)edid + EDID_LENGTH * (i + 1); + if (edid_ext[0] == CEA_EXT) + break; + } + + if (i == edid->extensions) + return NULL; + + return edid_ext; +} +EXPORT_SYMBOL(drm_find_cea_extension); + +/** + * drm_detect_hdmi_monitor - detect whether monitor is hdmi. + * @edid: monitor EDID information + * + * Parse the CEA extension according to CEA-861-B. + * Return true if HDMI, false if not or unknown. + */ +bool drm_detect_hdmi_monitor(struct edid *edid) +{ + u8 *edid_ext; + int i, hdmi_id; + int start_offset, end_offset; + bool is_hdmi = false; + + edid_ext = drm_find_cea_extension(edid); + if (!edid_ext) + goto end; + + /* Data block offset in CEA extension block */ + start_offset = 4; + end_offset = edid_ext[2]; + + /* + * Because HDMI identifier is in Vendor Specific Block, + * search it from all data blocks of CEA extension. + */ + for (i = start_offset; i < end_offset; + /* Increased by data block len */ + i += ((edid_ext[i] & 0x1f) + 1)) { + /* Find vendor specific block */ + if ((edid_ext[i] >> 5) == VENDOR_BLOCK) { + hdmi_id = edid_ext[i + 1] | (edid_ext[i + 2] << 8) | + edid_ext[i + 3] << 16; + /* Find HDMI identifier */ + if (hdmi_id == HDMI_IDENTIFIER) + is_hdmi = true; + break; + } + } + +end: + return is_hdmi; +} +EXPORT_SYMBOL(drm_detect_hdmi_monitor); + +/** + * drm_detect_monitor_audio - check monitor audio capability + * + * Monitor should have CEA extension block. + * If monitor has 'basic audio', but no CEA audio blocks, it's 'basic + * audio' only. If there is any audio extension block and supported + * audio format, assume at least 'basic audio' support, even if 'basic + * audio' is not defined in EDID. + * + */ +bool drm_detect_monitor_audio(struct edid *edid) +{ + u8 *edid_ext; + int i, j; + bool has_audio = false; + int start_offset, end_offset; + + edid_ext = drm_find_cea_extension(edid); + if (!edid_ext) + goto end; + + has_audio = ((edid_ext[3] & EDID_BASIC_AUDIO) != 0); + + if (has_audio) { + DRM_DEBUG_KMS("Monitor has basic audio support\n"); + goto end; + } + + /* Data block offset in CEA extension block */ + start_offset = 4; + end_offset = edid_ext[2]; + + for (i = start_offset; i < end_offset; + i += ((edid_ext[i] & 0x1f) + 1)) { + if ((edid_ext[i] >> 5) == AUDIO_BLOCK) { + has_audio = true; + for (j = 1; j < (edid_ext[i] & 0x1f); j += 3) + DRM_DEBUG_KMS("CEA audio format %d\n", + (edid_ext[i + j] >> 3) & 0xf); + goto end; + } + } +end: + return has_audio; +} +EXPORT_SYMBOL(drm_detect_monitor_audio); + +/** + * drm_add_display_info - pull display info out if present + * @edid: EDID data + * @info: display info (attached to connector) + * + * Grab any available display info and stuff it into the drm_display_info + * structure that's part of the connector. Useful for tracking bpp and + * color spaces. + */ +static void drm_add_display_info(struct edid *edid, + struct drm_display_info *info) +{ + info->width_mm = edid->width_cm * 10; + info->height_mm = edid->height_cm * 10; + + /* driver figures it out in this case */ + info->bpc = 0; + info->color_formats = 0; + + /* Only defined for 1.4 with digital displays */ + if (edid->revision < 4) + return; + + if (!(edid->input & DRM_EDID_INPUT_DIGITAL)) + return; + + switch (edid->input & DRM_EDID_DIGITAL_DEPTH_MASK) { + case DRM_EDID_DIGITAL_DEPTH_6: + info->bpc = 6; + break; + case DRM_EDID_DIGITAL_DEPTH_8: + info->bpc = 8; + break; + case DRM_EDID_DIGITAL_DEPTH_10: + info->bpc = 10; + break; + case DRM_EDID_DIGITAL_DEPTH_12: + info->bpc = 12; + break; + case DRM_EDID_DIGITAL_DEPTH_14: + info->bpc = 14; + break; + case DRM_EDID_DIGITAL_DEPTH_16: + info->bpc = 16; + break; + case DRM_EDID_DIGITAL_DEPTH_UNDEF: + default: + info->bpc = 0; + break; + } + + info->color_formats = DRM_COLOR_FORMAT_RGB444; + if (info->color_formats & DRM_EDID_FEATURE_RGB_YCRCB444) + info->color_formats = DRM_COLOR_FORMAT_YCRCB444; + if (info->color_formats & DRM_EDID_FEATURE_RGB_YCRCB422) + info->color_formats = DRM_COLOR_FORMAT_YCRCB422; +} + +/** + * drm_add_edid_modes - add modes from EDID data, if available + * @connector: connector we're probing + * @edid: edid data + * + * Add the specified modes to the connector's mode list. + * + * Return number of modes added or 0 if we couldn't find any. + */ +int drm_add_edid_modes(struct drm_connector *connector, struct edid *edid) +{ + int num_modes = 0; + u32 quirks; + + if (edid == NULL) { + return 0; + } + if (!drm_edid_is_valid(edid)) { + dev_warn(connector->dev->dev, "%s: EDID invalid.\n", + drm_get_connector_name(connector)); + return 0; + } + + quirks = edid_get_quirks(edid); + + /* + * EDID spec says modes should be preferred in this order: + * - preferred detailed mode + * - other detailed modes from base block + * - detailed modes from extension blocks + * - CVT 3-byte code modes + * - standard timing codes + * - established timing codes + * - modes inferred from GTF or CVT range information + * + * We get this pretty much right. + * + * XXX order for additional mode types in extension blocks? + */ + num_modes += add_detailed_modes(connector, edid, quirks); + num_modes += add_cvt_modes(connector, edid); + num_modes += add_standard_modes(connector, edid); + num_modes += add_established_modes(connector, edid); + num_modes += add_inferred_modes(connector, edid); + + if (quirks & (EDID_QUIRK_PREFER_LARGE_60 | EDID_QUIRK_PREFER_LARGE_75)) + edid_fixup_preferred(connector, quirks); + + drm_add_display_info(edid, &connector->display_info); + + return num_modes; +} +EXPORT_SYMBOL(drm_add_edid_modes); + +/** + * drm_add_modes_noedid - add modes for the connectors without EDID + * @connector: connector we're probing + * @hdisplay: the horizontal display limit + * @vdisplay: the vertical display limit + * + * Add the specified modes to the connector's mode list. Only when the + * hdisplay/vdisplay is not beyond the given limit, it will be added. + * + * Return number of modes added or 0 if we couldn't find any. + */ +int drm_add_modes_noedid(struct drm_connector *connector, + int hdisplay, int vdisplay) +{ + int i, count, num_modes = 0; + struct drm_display_mode *mode; + struct drm_device *dev = connector->dev; + + count = sizeof(drm_dmt_modes) / sizeof(struct drm_display_mode); + if (hdisplay < 0) + hdisplay = 0; + if (vdisplay < 0) + vdisplay = 0; + + for (i = 0; i < count; i++) { + const struct drm_display_mode *ptr = &drm_dmt_modes[i]; + if (hdisplay && vdisplay) { + /* + * Only when two are valid, they will be used to check + * whether the mode should be added to the mode list of + * the connector. + */ + if (ptr->hdisplay > hdisplay || + ptr->vdisplay > vdisplay) + continue; + } + if (drm_mode_vrefresh(ptr) > 61) + continue; + mode = drm_mode_duplicate(dev, ptr); + if (mode) { + drm_mode_probed_add(connector, mode); + num_modes++; + } + } + return num_modes; +} +EXPORT_SYMBOL(drm_add_modes_noedid); diff --git a/drivers/gpu/drm/drm_edid_modes.h b/drivers/gpu/drm/drm_edid_modes.h new file mode 100644 index 00000000..5f206448 --- /dev/null +++ b/drivers/gpu/drm/drm_edid_modes.h @@ -0,0 +1,380 @@ +/* + * Copyright (c) 2007-2008 Intel Corporation + * Jesse Barnes + * Copyright 2010 Red Hat, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sub license, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + */ + +#include +#include "drmP.h" +#include "drm_edid.h" + +/* + * Autogenerated from the DMT spec. + * This table is copied from xfree86/modes/xf86EdidModes.c. + * But the mode with Reduced blank feature is deleted. + */ +static const struct drm_display_mode drm_dmt_modes[] = { + /* 640x350@85Hz */ + { DRM_MODE("640x350", DRM_MODE_TYPE_DRIVER, 31500, 640, 672, + 736, 832, 0, 350, 382, 385, 445, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_NVSYNC) }, + /* 640x400@85Hz */ + { DRM_MODE("640x400", DRM_MODE_TYPE_DRIVER, 31500, 640, 672, + 736, 832, 0, 400, 401, 404, 445, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 720x400@85Hz */ + { DRM_MODE("720x400", DRM_MODE_TYPE_DRIVER, 35500, 720, 756, + 828, 936, 0, 400, 401, 404, 446, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 640x480@60Hz */ + { DRM_MODE("640x480", DRM_MODE_TYPE_DRIVER, 25175, 640, 656, + 752, 800, 0, 480, 489, 492, 525, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_NVSYNC) }, + /* 640x480@72Hz */ + { DRM_MODE("640x480", DRM_MODE_TYPE_DRIVER, 31500, 640, 664, + 704, 832, 0, 480, 489, 492, 520, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_NVSYNC) }, + /* 640x480@75Hz */ + { DRM_MODE("640x480", DRM_MODE_TYPE_DRIVER, 31500, 640, 656, + 720, 840, 0, 480, 481, 484, 500, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_NVSYNC) }, + /* 640x480@85Hz */ + { DRM_MODE("640x480", DRM_MODE_TYPE_DRIVER, 36000, 640, 696, + 752, 832, 0, 480, 481, 484, 509, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_NVSYNC) }, + /* 800x600@56Hz */ + { DRM_MODE("800x600", DRM_MODE_TYPE_DRIVER, 36000, 800, 824, + 896, 1024, 0, 600, 601, 603, 625, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 800x600@60Hz */ + { DRM_MODE("800x600", DRM_MODE_TYPE_DRIVER, 40000, 800, 840, + 968, 1056, 0, 600, 601, 605, 628, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 800x600@72Hz */ + { DRM_MODE("800x600", DRM_MODE_TYPE_DRIVER, 50000, 800, 856, + 976, 1040, 0, 600, 637, 643, 666, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 800x600@75Hz */ + { DRM_MODE("800x600", DRM_MODE_TYPE_DRIVER, 49500, 800, 816, + 896, 1056, 0, 600, 601, 604, 625, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 800x600@85Hz */ + { DRM_MODE("800x600", DRM_MODE_TYPE_DRIVER, 56250, 800, 832, + 896, 1048, 0, 600, 601, 604, 631, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 848x480@60Hz */ + { DRM_MODE("848x480", DRM_MODE_TYPE_DRIVER, 33750, 848, 864, + 976, 1088, 0, 480, 486, 494, 517, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1024x768@43Hz, interlace */ + { DRM_MODE("1024x768", DRM_MODE_TYPE_DRIVER, 44900, 1024, 1032, + 1208, 1264, 0, 768, 768, 772, 817, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC | + DRM_MODE_FLAG_INTERLACE) }, + /* 1024x768@60Hz */ + { DRM_MODE("1024x768", DRM_MODE_TYPE_DRIVER, 65000, 1024, 1048, + 1184, 1344, 0, 768, 771, 777, 806, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_NVSYNC) }, + /* 1024x768@70Hz */ + { DRM_MODE("1024x768", DRM_MODE_TYPE_DRIVER, 75000, 1024, 1048, + 1184, 1328, 0, 768, 771, 777, 806, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_NVSYNC) }, + /* 1024x768@75Hz */ + { DRM_MODE("1024x768", DRM_MODE_TYPE_DRIVER, 78750, 1024, 1040, + 1136, 1312, 0, 768, 769, 772, 800, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1024x768@85Hz */ + { DRM_MODE("1024x768", DRM_MODE_TYPE_DRIVER, 94500, 1024, 1072, + 1168, 1376, 0, 768, 769, 772, 808, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1152x864@75Hz */ + { DRM_MODE("1152x864", DRM_MODE_TYPE_DRIVER, 108000, 1152, 1216, + 1344, 1600, 0, 864, 865, 868, 900, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1280x768@60Hz */ + { DRM_MODE("1280x768", DRM_MODE_TYPE_DRIVER, 79500, 1280, 1344, + 1472, 1664, 0, 768, 771, 778, 798, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1280x768@75Hz */ + { DRM_MODE("1280x768", DRM_MODE_TYPE_DRIVER, 102250, 1280, 1360, + 1488, 1696, 0, 768, 771, 778, 805, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_NVSYNC) }, + /* 1280x768@85Hz */ + { DRM_MODE("1280x768", DRM_MODE_TYPE_DRIVER, 117500, 1280, 1360, + 1496, 1712, 0, 768, 771, 778, 809, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1280x800@60Hz */ + { DRM_MODE("1280x800", DRM_MODE_TYPE_DRIVER, 83500, 1280, 1352, + 1480, 1680, 0, 800, 803, 809, 831, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_NVSYNC) }, + /* 1280x800@75Hz */ + { DRM_MODE("1280x800", DRM_MODE_TYPE_DRIVER, 106500, 1280, 1360, + 1488, 1696, 0, 800, 803, 809, 838, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1280x800@85Hz */ + { DRM_MODE("1280x800", DRM_MODE_TYPE_DRIVER, 122500, 1280, 1360, + 1496, 1712, 0, 800, 803, 809, 843, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1280x960@60Hz */ + { DRM_MODE("1280x960", DRM_MODE_TYPE_DRIVER, 108000, 1280, 1376, + 1488, 1800, 0, 960, 961, 964, 1000, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1280x960@85Hz */ + { DRM_MODE("1280x960", DRM_MODE_TYPE_DRIVER, 148500, 1280, 1344, + 1504, 1728, 0, 960, 961, 964, 1011, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1280x1024@60Hz */ + { DRM_MODE("1280x1024", DRM_MODE_TYPE_DRIVER, 108000, 1280, 1328, + 1440, 1688, 0, 1024, 1025, 1028, 1066, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1280x1024@75Hz */ + { DRM_MODE("1280x1024", DRM_MODE_TYPE_DRIVER, 135000, 1280, 1296, + 1440, 1688, 0, 1024, 1025, 1028, 1066, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1280x1024@85Hz */ + { DRM_MODE("1280x1024", DRM_MODE_TYPE_DRIVER, 157500, 1280, 1344, + 1504, 1728, 0, 1024, 1025, 1028, 1072, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1360x768@60Hz */ + { DRM_MODE("1360x768", DRM_MODE_TYPE_DRIVER, 85500, 1360, 1424, + 1536, 1792, 0, 768, 771, 777, 795, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1440x1050@60Hz */ + { DRM_MODE("1400x1050", DRM_MODE_TYPE_DRIVER, 121750, 1400, 1488, + 1632, 1864, 0, 1050, 1053, 1057, 1089, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1440x1050@75Hz */ + { DRM_MODE("1400x1050", DRM_MODE_TYPE_DRIVER, 156000, 1400, 1504, + 1648, 1896, 0, 1050, 1053, 1057, 1099, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1440x1050@85Hz */ + { DRM_MODE("1400x1050", DRM_MODE_TYPE_DRIVER, 179500, 1400, 1504, + 1656, 1912, 0, 1050, 1053, 1057, 1105, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1440x900@60Hz */ + { DRM_MODE("1440x900", DRM_MODE_TYPE_DRIVER, 106500, 1440, 1520, + 1672, 1904, 0, 900, 903, 909, 934, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1440x900@75Hz */ + { DRM_MODE("1440x900", DRM_MODE_TYPE_DRIVER, 136750, 1440, 1536, + 1688, 1936, 0, 900, 903, 909, 942, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1440x900@85Hz */ + { DRM_MODE("1440x900", DRM_MODE_TYPE_DRIVER, 157000, 1440, 1544, + 1696, 1952, 0, 900, 903, 909, 948, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1600x1200@60Hz */ + { DRM_MODE("1600x1200", DRM_MODE_TYPE_DRIVER, 162000, 1600, 1664, + 1856, 2160, 0, 1200, 1201, 1204, 1250, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1600x1200@65Hz */ + { DRM_MODE("1600x1200", DRM_MODE_TYPE_DRIVER, 175500, 1600, 1664, + 1856, 2160, 0, 1200, 1201, 1204, 1250, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1600x1200@70Hz */ + { DRM_MODE("1600x1200", DRM_MODE_TYPE_DRIVER, 189000, 1600, 1664, + 1856, 2160, 0, 1200, 1201, 1204, 1250, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1600x1200@75Hz */ + { DRM_MODE("1600x1200", DRM_MODE_TYPE_DRIVER, 202500, 1600, 1664, + 1856, 2160, 0, 1200, 1201, 1204, 1250, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1600x1200@85Hz */ + { DRM_MODE("1600x1200", DRM_MODE_TYPE_DRIVER, 229500, 1600, 1664, + 1856, 2160, 0, 1200, 1201, 1204, 1250, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1680x1050@60Hz */ + { DRM_MODE("1680x1050", DRM_MODE_TYPE_DRIVER, 146250, 1680, 1784, + 1960, 2240, 0, 1050, 1053, 1059, 1089, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1680x1050@75Hz */ + { DRM_MODE("1680x1050", DRM_MODE_TYPE_DRIVER, 187000, 1680, 1800, + 1976, 2272, 0, 1050, 1053, 1059, 1099, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1680x1050@85Hz */ + { DRM_MODE("1680x1050", DRM_MODE_TYPE_DRIVER, 214750, 1680, 1808, + 1984, 2288, 0, 1050, 1053, 1059, 1105, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1792x1344@60Hz */ + { DRM_MODE("1792x1344", DRM_MODE_TYPE_DRIVER, 204750, 1792, 1920, + 2120, 2448, 0, 1344, 1345, 1348, 1394, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1729x1344@75Hz */ + { DRM_MODE("1792x1344", DRM_MODE_TYPE_DRIVER, 261000, 1792, 1888, + 2104, 2456, 0, 1344, 1345, 1348, 1417, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1853x1392@60Hz */ + { DRM_MODE("1856x1392", DRM_MODE_TYPE_DRIVER, 218250, 1856, 1952, + 2176, 2528, 0, 1392, 1393, 1396, 1439, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1856x1392@75Hz */ + { DRM_MODE("1856x1392", DRM_MODE_TYPE_DRIVER, 288000, 1856, 1984, + 2208, 2560, 0, 1392, 1395, 1399, 1500, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1920x1200@60Hz */ + { DRM_MODE("1920x1200", DRM_MODE_TYPE_DRIVER, 193250, 1920, 2056, + 2256, 2592, 0, 1200, 1203, 1209, 1245, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1920x1200@75Hz */ + { DRM_MODE("1920x1200", DRM_MODE_TYPE_DRIVER, 245250, 1920, 2056, + 2264, 2608, 0, 1200, 1203, 1209, 1255, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1920x1200@85Hz */ + { DRM_MODE("1920x1200", DRM_MODE_TYPE_DRIVER, 281250, 1920, 2064, + 2272, 2624, 0, 1200, 1203, 1209, 1262, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1920x1440@60Hz */ + { DRM_MODE("1920x1440", DRM_MODE_TYPE_DRIVER, 234000, 1920, 2048, + 2256, 2600, 0, 1440, 1441, 1444, 1500, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 1920x1440@75Hz */ + { DRM_MODE("1920x1440", DRM_MODE_TYPE_DRIVER, 297000, 1920, 2064, + 2288, 2640, 0, 1440, 1441, 1444, 1500, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 2560x1600@60Hz */ + { DRM_MODE("2560x1600", DRM_MODE_TYPE_DRIVER, 348500, 2560, 2752, + 3032, 3504, 0, 1600, 1603, 1609, 1658, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 2560x1600@75HZ */ + { DRM_MODE("2560x1600", DRM_MODE_TYPE_DRIVER, 443250, 2560, 2768, + 3048, 3536, 0, 1600, 1603, 1609, 1672, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_PVSYNC) }, + /* 2560x1600@85HZ */ + { DRM_MODE("2560x1600", DRM_MODE_TYPE_DRIVER, 505250, 2560, 2768, + 3048, 3536, 0, 1600, 1603, 1609, 1682, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_PVSYNC) }, +}; +static const int drm_num_dmt_modes = + sizeof(drm_dmt_modes) / sizeof(struct drm_display_mode); + +static const struct drm_display_mode edid_est_modes[] = { + { DRM_MODE("800x600", DRM_MODE_TYPE_DRIVER, 40000, 800, 840, + 968, 1056, 0, 600, 601, 605, 628, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, /* 800x600@60Hz */ + { DRM_MODE("800x600", DRM_MODE_TYPE_DRIVER, 36000, 800, 824, + 896, 1024, 0, 600, 601, 603, 625, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, /* 800x600@56Hz */ + { DRM_MODE("640x480", DRM_MODE_TYPE_DRIVER, 31500, 640, 656, + 720, 840, 0, 480, 481, 484, 500, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_NVSYNC) }, /* 640x480@75Hz */ + { DRM_MODE("640x480", DRM_MODE_TYPE_DRIVER, 31500, 640, 664, + 704, 832, 0, 480, 489, 491, 520, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_NVSYNC) }, /* 640x480@72Hz */ + { DRM_MODE("640x480", DRM_MODE_TYPE_DRIVER, 30240, 640, 704, + 768, 864, 0, 480, 483, 486, 525, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_NVSYNC) }, /* 640x480@67Hz */ + { DRM_MODE("640x480", DRM_MODE_TYPE_DRIVER, 25200, 640, 656, + 752, 800, 0, 480, 490, 492, 525, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_NVSYNC) }, /* 640x480@60Hz */ + { DRM_MODE("720x400", DRM_MODE_TYPE_DRIVER, 35500, 720, 738, + 846, 900, 0, 400, 421, 423, 449, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_NVSYNC) }, /* 720x400@88Hz */ + { DRM_MODE("720x400", DRM_MODE_TYPE_DRIVER, 28320, 720, 738, + 846, 900, 0, 400, 412, 414, 449, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_PVSYNC) }, /* 720x400@70Hz */ + { DRM_MODE("1280x1024", DRM_MODE_TYPE_DRIVER, 135000, 1280, 1296, + 1440, 1688, 0, 1024, 1025, 1028, 1066, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, /* 1280x1024@75Hz */ + { DRM_MODE("1024x768", DRM_MODE_TYPE_DRIVER, 78800, 1024, 1040, + 1136, 1312, 0, 768, 769, 772, 800, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, /* 1024x768@75Hz */ + { DRM_MODE("1024x768", DRM_MODE_TYPE_DRIVER, 75000, 1024, 1048, + 1184, 1328, 0, 768, 771, 777, 806, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_NVSYNC) }, /* 1024x768@70Hz */ + { DRM_MODE("1024x768", DRM_MODE_TYPE_DRIVER, 65000, 1024, 1048, + 1184, 1344, 0, 768, 771, 777, 806, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_NVSYNC) }, /* 1024x768@60Hz */ + { DRM_MODE("1024x768", DRM_MODE_TYPE_DRIVER,44900, 1024, 1032, + 1208, 1264, 0, 768, 768, 776, 817, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC | DRM_MODE_FLAG_INTERLACE) }, /* 1024x768@43Hz */ + { DRM_MODE("832x624", DRM_MODE_TYPE_DRIVER, 57284, 832, 864, + 928, 1152, 0, 624, 625, 628, 667, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_NVSYNC) }, /* 832x624@75Hz */ + { DRM_MODE("800x600", DRM_MODE_TYPE_DRIVER, 49500, 800, 816, + 896, 1056, 0, 600, 601, 604, 625, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, /* 800x600@75Hz */ + { DRM_MODE("800x600", DRM_MODE_TYPE_DRIVER, 50000, 800, 856, + 976, 1040, 0, 600, 637, 643, 666, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, /* 800x600@72Hz */ + { DRM_MODE("1152x864", DRM_MODE_TYPE_DRIVER, 108000, 1152, 1216, + 1344, 1600, 0, 864, 865, 868, 900, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, /* 1152x864@75Hz */ +}; + +static const struct { + short w; + short h; + short r; + short rb; +} est3_modes[] = { + /* byte 6 */ + { 640, 350, 85, 0 }, + { 640, 400, 85, 0 }, + { 720, 400, 85, 0 }, + { 640, 480, 85, 0 }, + { 848, 480, 60, 0 }, + { 800, 600, 85, 0 }, + { 1024, 768, 85, 0 }, + { 1152, 864, 75, 0 }, + /* byte 7 */ + { 1280, 768, 60, 1 }, + { 1280, 768, 60, 0 }, + { 1280, 768, 75, 0 }, + { 1280, 768, 85, 0 }, + { 1280, 960, 60, 0 }, + { 1280, 960, 85, 0 }, + { 1280, 1024, 60, 0 }, + { 1280, 1024, 85, 0 }, + /* byte 8 */ + { 1360, 768, 60, 0 }, + { 1440, 900, 60, 1 }, + { 1440, 900, 60, 0 }, + { 1440, 900, 75, 0 }, + { 1440, 900, 85, 0 }, + { 1400, 1050, 60, 1 }, + { 1400, 1050, 60, 0 }, + { 1400, 1050, 75, 0 }, + /* byte 9 */ + { 1400, 1050, 85, 0 }, + { 1680, 1050, 60, 1 }, + { 1680, 1050, 60, 0 }, + { 1680, 1050, 75, 0 }, + { 1680, 1050, 85, 0 }, + { 1600, 1200, 60, 0 }, + { 1600, 1200, 65, 0 }, + { 1600, 1200, 70, 0 }, + /* byte 10 */ + { 1600, 1200, 75, 0 }, + { 1600, 1200, 85, 0 }, + { 1792, 1344, 60, 0 }, + { 1792, 1344, 85, 0 }, + { 1856, 1392, 60, 0 }, + { 1856, 1392, 75, 0 }, + { 1920, 1200, 60, 1 }, + { 1920, 1200, 60, 0 }, + /* byte 11 */ + { 1920, 1200, 75, 0 }, + { 1920, 1200, 85, 0 }, + { 1920, 1440, 60, 0 }, + { 1920, 1440, 75, 0 }, +}; +static const int num_est3_modes = sizeof(est3_modes) / sizeof(est3_modes[0]); diff --git a/drivers/gpu/drm/drm_encoder_slave.c b/drivers/gpu/drm/drm_encoder_slave.c new file mode 100644 index 00000000..d62c064f --- /dev/null +++ b/drivers/gpu/drm/drm_encoder_slave.c @@ -0,0 +1,123 @@ +/* + * Copyright (C) 2009 Francisco Jerez. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "drm_encoder_slave.h" + +/** + * drm_i2c_encoder_init - Initialize an I2C slave encoder + * @dev: DRM device. + * @encoder: Encoder to be attached to the I2C device. You aren't + * required to have called drm_encoder_init() before. + * @adap: I2C adapter that will be used to communicate with + * the device. + * @info: Information that will be used to create the I2C device. + * Required fields are @addr and @type. + * + * Create an I2C device on the specified bus (the module containing its + * driver is transparently loaded) and attach it to the specified + * &drm_encoder_slave. The @slave_funcs field will be initialized with + * the hooks provided by the slave driver. + * + * If @info->platform_data is non-NULL it will be used as the initial + * slave config. + * + * Returns 0 on success or a negative errno on failure, in particular, + * -ENODEV is returned when no matching driver is found. + */ +int drm_i2c_encoder_init(struct drm_device *dev, + struct drm_encoder_slave *encoder, + struct i2c_adapter *adap, + const struct i2c_board_info *info) +{ + char modalias[sizeof(I2C_MODULE_PREFIX) + + I2C_NAME_SIZE]; + struct module *module = NULL; + struct i2c_client *client; + struct drm_i2c_encoder_driver *encoder_drv; + int err = 0; + + snprintf(modalias, sizeof(modalias), + "%s%s", I2C_MODULE_PREFIX, info->type); + request_module(modalias); + + client = i2c_new_device(adap, info); + if (!client) { + err = -ENOMEM; + goto fail; + } + + if (!client->driver) { + err = -ENODEV; + goto fail_unregister; + } + + module = client->driver->driver.owner; + if (!try_module_get(module)) { + err = -ENODEV; + goto fail_unregister; + } + + encoder->bus_priv = client; + + encoder_drv = to_drm_i2c_encoder_driver(client->driver); + + err = encoder_drv->encoder_init(client, dev, encoder); + if (err) + goto fail_unregister; + + if (info->platform_data) + encoder->slave_funcs->set_config(&encoder->base, + info->platform_data); + + return 0; + +fail_unregister: + i2c_unregister_device(client); + module_put(module); +fail: + return err; +} +EXPORT_SYMBOL(drm_i2c_encoder_init); + +/** + * drm_i2c_encoder_destroy - Unregister the I2C device backing an encoder + * @drm_encoder: Encoder to be unregistered. + * + * This should be called from the @destroy method of an I2C slave + * encoder driver once I2C access is no longer needed. + */ +void drm_i2c_encoder_destroy(struct drm_encoder *drm_encoder) +{ + struct drm_encoder_slave *encoder = to_encoder_slave(drm_encoder); + struct i2c_client *client = drm_i2c_encoder_get_client(drm_encoder); + struct module *module = client->driver->driver.owner; + + i2c_unregister_device(client); + encoder->bus_priv = NULL; + + module_put(module); +} +EXPORT_SYMBOL(drm_i2c_encoder_destroy); diff --git a/drivers/gpu/drm/drm_fb_helper.c b/drivers/gpu/drm/drm_fb_helper.c new file mode 100644 index 00000000..a9dcdc7d --- /dev/null +++ b/drivers/gpu/drm/drm_fb_helper.c @@ -0,0 +1,1461 @@ +/* + * Copyright (c) 2006-2009 Red Hat Inc. + * Copyright (c) 2006-2008 Intel Corporation + * Copyright (c) 2007 Dave Airlie + * + * DRM framebuffer helper functions + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that copyright + * notice and this permission notice appear in supporting documentation, and + * that the name of the copyright holders not be used in advertising or + * publicity pertaining to distribution of the software without specific, + * written prior permission. The copyright holders make no representations + * about the suitability of this software for any purpose. It is provided "as + * is" without express or implied warranty. + * + * THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, + * INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO + * EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY SPECIAL, INDIRECT OR + * CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, + * DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER + * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE + * OF THIS SOFTWARE. + * + * Authors: + * Dave Airlie + * Jesse Barnes + */ +#include +#include +#include +#include +#include "drmP.h" +#include "drm_crtc.h" +#include "drm_fb_helper.h" +#include "drm_crtc_helper.h" + +MODULE_AUTHOR("David Airlie, Jesse Barnes"); +MODULE_DESCRIPTION("DRM KMS helper"); +MODULE_LICENSE("GPL and additional rights"); + +static LIST_HEAD(kernel_fb_helper_list); + +/* simple single crtc case helper function */ +int drm_fb_helper_single_add_all_connectors(struct drm_fb_helper *fb_helper) +{ + struct drm_device *dev = fb_helper->dev; + struct drm_connector *connector; + int i; + + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + struct drm_fb_helper_connector *fb_helper_connector; + + fb_helper_connector = kzalloc(sizeof(struct drm_fb_helper_connector), GFP_KERNEL); + if (!fb_helper_connector) + goto fail; + + fb_helper_connector->connector = connector; + fb_helper->connector_info[fb_helper->connector_count++] = fb_helper_connector; + } + return 0; +fail: + for (i = 0; i < fb_helper->connector_count; i++) { + kfree(fb_helper->connector_info[i]); + fb_helper->connector_info[i] = NULL; + } + fb_helper->connector_count = 0; + return -ENOMEM; +} +EXPORT_SYMBOL(drm_fb_helper_single_add_all_connectors); + +static int drm_fb_helper_parse_command_line(struct drm_fb_helper *fb_helper) +{ + struct drm_fb_helper_connector *fb_helper_conn; + int i; + + for (i = 0; i < fb_helper->connector_count; i++) { + struct drm_cmdline_mode *mode; + struct drm_connector *connector; + char *option = NULL; + + fb_helper_conn = fb_helper->connector_info[i]; + connector = fb_helper_conn->connector; + mode = &fb_helper_conn->cmdline_mode; + + /* do something on return - turn off connector maybe */ + if (fb_get_options(drm_get_connector_name(connector), &option)) + continue; + + if (drm_mode_parse_command_line_for_connector(option, + connector, + mode)) { + if (mode->force) { + const char *s; + switch (mode->force) { + case DRM_FORCE_OFF: s = "OFF"; break; + case DRM_FORCE_ON_DIGITAL: s = "ON - dig"; break; + default: + case DRM_FORCE_ON: s = "ON"; break; + } + + DRM_INFO("forcing %s connector %s\n", + drm_get_connector_name(connector), s); + connector->force = mode->force; + } + + DRM_DEBUG_KMS("cmdline mode for connector %s %dx%d@%dHz%s%s%s\n", + drm_get_connector_name(connector), + mode->xres, mode->yres, + mode->refresh_specified ? mode->refresh : 60, + mode->rb ? " reduced blanking" : "", + mode->margins ? " with margins" : "", + mode->interlace ? " interlaced" : ""); + } + + } + return 0; +} + +static void drm_fb_helper_save_lut_atomic(struct drm_crtc *crtc, struct drm_fb_helper *helper) +{ + uint16_t *r_base, *g_base, *b_base; + int i; + + r_base = crtc->gamma_store; + g_base = r_base + crtc->gamma_size; + b_base = g_base + crtc->gamma_size; + + for (i = 0; i < crtc->gamma_size; i++) + helper->funcs->gamma_get(crtc, &r_base[i], &g_base[i], &b_base[i], i); +} + +static void drm_fb_helper_restore_lut_atomic(struct drm_crtc *crtc) +{ + uint16_t *r_base, *g_base, *b_base; + + r_base = crtc->gamma_store; + g_base = r_base + crtc->gamma_size; + b_base = g_base + crtc->gamma_size; + + crtc->funcs->gamma_set(crtc, r_base, g_base, b_base, 0, crtc->gamma_size); +} + +int drm_fb_helper_debug_enter(struct fb_info *info) +{ + struct drm_fb_helper *helper = info->par; + struct drm_crtc_helper_funcs *funcs; + int i; + + if (list_empty(&kernel_fb_helper_list)) + return false; + + list_for_each_entry(helper, &kernel_fb_helper_list, kernel_fb_list) { + for (i = 0; i < helper->crtc_count; i++) { + struct drm_mode_set *mode_set = + &helper->crtc_info[i].mode_set; + + if (!mode_set->crtc->enabled) + continue; + + funcs = mode_set->crtc->helper_private; + drm_fb_helper_save_lut_atomic(mode_set->crtc, helper); + funcs->mode_set_base_atomic(mode_set->crtc, + mode_set->fb, + mode_set->x, + mode_set->y, + ENTER_ATOMIC_MODE_SET); + } + } + + return 0; +} +EXPORT_SYMBOL(drm_fb_helper_debug_enter); + +/* Find the real fb for a given fb helper CRTC */ +static struct drm_framebuffer *drm_mode_config_fb(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + struct drm_crtc *c; + + list_for_each_entry(c, &dev->mode_config.crtc_list, head) { + if (crtc->base.id == c->base.id) + return c->fb; + } + + return NULL; +} + +int drm_fb_helper_debug_leave(struct fb_info *info) +{ + struct drm_fb_helper *helper = info->par; + struct drm_crtc *crtc; + struct drm_crtc_helper_funcs *funcs; + struct drm_framebuffer *fb; + int i; + + for (i = 0; i < helper->crtc_count; i++) { + struct drm_mode_set *mode_set = &helper->crtc_info[i].mode_set; + crtc = mode_set->crtc; + funcs = crtc->helper_private; + fb = drm_mode_config_fb(crtc); + + if (!crtc->enabled) + continue; + + if (!fb) { + DRM_ERROR("no fb to restore??\n"); + continue; + } + + drm_fb_helper_restore_lut_atomic(mode_set->crtc); + funcs->mode_set_base_atomic(mode_set->crtc, fb, crtc->x, + crtc->y, LEAVE_ATOMIC_MODE_SET); + } + + return 0; +} +EXPORT_SYMBOL(drm_fb_helper_debug_leave); + +bool drm_fb_helper_restore_fbdev_mode(struct drm_fb_helper *fb_helper) +{ + bool error = false; + int i, ret; + for (i = 0; i < fb_helper->crtc_count; i++) { + struct drm_mode_set *mode_set = &fb_helper->crtc_info[i].mode_set; + ret = drm_crtc_helper_set_config(mode_set); + if (ret) + error = true; + } + return error; +} +EXPORT_SYMBOL(drm_fb_helper_restore_fbdev_mode); + +bool drm_fb_helper_force_kernel_mode(void) +{ + bool ret, error = false; + struct drm_fb_helper *helper; + + if (list_empty(&kernel_fb_helper_list)) + return false; + + list_for_each_entry(helper, &kernel_fb_helper_list, kernel_fb_list) { + if (helper->dev->switch_power_state == DRM_SWITCH_POWER_OFF) + continue; + + ret = drm_fb_helper_restore_fbdev_mode(helper); + if (ret) + error = true; + } + return error; +} + +int drm_fb_helper_panic(struct notifier_block *n, unsigned long ununsed, + void *panic_str) +{ + printk(KERN_ERR "panic occurred, switching back to text console\n"); + return drm_fb_helper_force_kernel_mode(); + return 0; +} +EXPORT_SYMBOL(drm_fb_helper_panic); + +static struct notifier_block paniced = { + .notifier_call = drm_fb_helper_panic, +}; + +/** + * drm_fb_helper_restore - restore the framebuffer console (kernel) config + * + * Restore's the kernel's fbcon mode, used for lastclose & panic paths. + */ +void drm_fb_helper_restore(void) +{ + bool ret; + ret = drm_fb_helper_force_kernel_mode(); + if (ret == true) + DRM_ERROR("Failed to restore crtc configuration\n"); +} +EXPORT_SYMBOL(drm_fb_helper_restore); + +#ifdef CONFIG_MAGIC_SYSRQ +static void drm_fb_helper_restore_work_fn(struct work_struct *ignored) +{ + drm_fb_helper_restore(); +} +static DECLARE_WORK(drm_fb_helper_restore_work, drm_fb_helper_restore_work_fn); + +static void drm_fb_helper_sysrq(int dummy1) +{ + schedule_work(&drm_fb_helper_restore_work); +} + +static struct sysrq_key_op sysrq_drm_fb_helper_restore_op = { + .handler = drm_fb_helper_sysrq, + .help_msg = "force-fb(V)", + .action_msg = "Restore framebuffer console", +}; +#else +static struct sysrq_key_op sysrq_drm_fb_helper_restore_op = { }; +#endif + +static void drm_fb_helper_on(struct fb_info *info) +{ + struct drm_fb_helper *fb_helper = info->par; + struct drm_device *dev = fb_helper->dev; + struct drm_crtc *crtc; + struct drm_crtc_helper_funcs *crtc_funcs; + struct drm_connector *connector; + struct drm_encoder *encoder; + int i, j; + + /* + * For each CRTC in this fb, turn the crtc on then, + * find all associated encoders and turn them on. + */ + mutex_lock(&dev->mode_config.mutex); + for (i = 0; i < fb_helper->crtc_count; i++) { + crtc = fb_helper->crtc_info[i].mode_set.crtc; + crtc_funcs = crtc->helper_private; + + if (!crtc->enabled) + continue; + + crtc_funcs->dpms(crtc, DRM_MODE_DPMS_ON); + + /* Walk the connectors & encoders on this fb turning them on */ + for (j = 0; j < fb_helper->connector_count; j++) { + connector = fb_helper->connector_info[j]->connector; + connector->dpms = DRM_MODE_DPMS_ON; + drm_connector_property_set_value(connector, + dev->mode_config.dpms_property, + DRM_MODE_DPMS_ON); + } + /* Found a CRTC on this fb, now find encoders */ + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + if (encoder->crtc == crtc) { + struct drm_encoder_helper_funcs *encoder_funcs; + + encoder_funcs = encoder->helper_private; + encoder_funcs->dpms(encoder, DRM_MODE_DPMS_ON); + } + } + } + mutex_unlock(&dev->mode_config.mutex); +} + +static void drm_fb_helper_off(struct fb_info *info, int dpms_mode) +{ + struct drm_fb_helper *fb_helper = info->par; + struct drm_device *dev = fb_helper->dev; + struct drm_crtc *crtc; + struct drm_crtc_helper_funcs *crtc_funcs; + struct drm_connector *connector; + struct drm_encoder *encoder; + int i, j; + + /* + * For each CRTC in this fb, find all associated encoders + * and turn them off, then turn off the CRTC. + */ + mutex_lock(&dev->mode_config.mutex); + for (i = 0; i < fb_helper->crtc_count; i++) { + crtc = fb_helper->crtc_info[i].mode_set.crtc; + crtc_funcs = crtc->helper_private; + + if (!crtc->enabled) + continue; + + /* Walk the connectors on this fb and mark them off */ + for (j = 0; j < fb_helper->connector_count; j++) { + connector = fb_helper->connector_info[j]->connector; + connector->dpms = dpms_mode; + drm_connector_property_set_value(connector, + dev->mode_config.dpms_property, + dpms_mode); + } + /* Found a CRTC on this fb, now find encoders */ + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + if (encoder->crtc == crtc) { + struct drm_encoder_helper_funcs *encoder_funcs; + + encoder_funcs = encoder->helper_private; + encoder_funcs->dpms(encoder, dpms_mode); + } + } + crtc_funcs->dpms(crtc, DRM_MODE_DPMS_OFF); + } + mutex_unlock(&dev->mode_config.mutex); +} + +int drm_fb_helper_blank(int blank, struct fb_info *info) +{ + switch (blank) { + /* Display: On; HSync: On, VSync: On */ + case FB_BLANK_UNBLANK: + drm_fb_helper_on(info); + break; + /* Display: Off; HSync: On, VSync: On */ + case FB_BLANK_NORMAL: + drm_fb_helper_off(info, DRM_MODE_DPMS_STANDBY); + break; + /* Display: Off; HSync: Off, VSync: On */ + case FB_BLANK_HSYNC_SUSPEND: + drm_fb_helper_off(info, DRM_MODE_DPMS_STANDBY); + break; + /* Display: Off; HSync: On, VSync: Off */ + case FB_BLANK_VSYNC_SUSPEND: + drm_fb_helper_off(info, DRM_MODE_DPMS_SUSPEND); + break; + /* Display: Off; HSync: Off, VSync: Off */ + case FB_BLANK_POWERDOWN: + drm_fb_helper_off(info, DRM_MODE_DPMS_OFF); + break; + } + return 0; +} +EXPORT_SYMBOL(drm_fb_helper_blank); + +static void drm_fb_helper_crtc_free(struct drm_fb_helper *helper) +{ + int i; + + for (i = 0; i < helper->connector_count; i++) + kfree(helper->connector_info[i]); + kfree(helper->connector_info); + for (i = 0; i < helper->crtc_count; i++) + kfree(helper->crtc_info[i].mode_set.connectors); + kfree(helper->crtc_info); +} + +int drm_fb_helper_init(struct drm_device *dev, + struct drm_fb_helper *fb_helper, + int crtc_count, int max_conn_count) +{ + struct drm_crtc *crtc; + int ret = 0; + int i; + + fb_helper->dev = dev; + + INIT_LIST_HEAD(&fb_helper->kernel_fb_list); + + fb_helper->crtc_info = kcalloc(crtc_count, sizeof(struct drm_fb_helper_crtc), GFP_KERNEL); + if (!fb_helper->crtc_info) + return -ENOMEM; + + fb_helper->crtc_count = crtc_count; + fb_helper->connector_info = kcalloc(dev->mode_config.num_connector, sizeof(struct drm_fb_helper_connector *), GFP_KERNEL); + if (!fb_helper->connector_info) { + kfree(fb_helper->crtc_info); + return -ENOMEM; + } + fb_helper->connector_count = 0; + + for (i = 0; i < crtc_count; i++) { + fb_helper->crtc_info[i].mode_set.connectors = + kcalloc(max_conn_count, + sizeof(struct drm_connector *), + GFP_KERNEL); + + if (!fb_helper->crtc_info[i].mode_set.connectors) { + ret = -ENOMEM; + goto out_free; + } + fb_helper->crtc_info[i].mode_set.num_connectors = 0; + } + + i = 0; + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { + fb_helper->crtc_info[i].crtc_id = crtc->base.id; + fb_helper->crtc_info[i].mode_set.crtc = crtc; + i++; + } + fb_helper->conn_limit = max_conn_count; + return 0; +out_free: + drm_fb_helper_crtc_free(fb_helper); + return -ENOMEM; +} +EXPORT_SYMBOL(drm_fb_helper_init); + +void drm_fb_helper_fini(struct drm_fb_helper *fb_helper) +{ + if (!list_empty(&fb_helper->kernel_fb_list)) { + list_del(&fb_helper->kernel_fb_list); + if (list_empty(&kernel_fb_helper_list)) { + printk(KERN_INFO "drm: unregistered panic notifier\n"); + atomic_notifier_chain_unregister(&panic_notifier_list, + &paniced); + unregister_sysrq_key('v', &sysrq_drm_fb_helper_restore_op); + } + } + + drm_fb_helper_crtc_free(fb_helper); + +} +EXPORT_SYMBOL(drm_fb_helper_fini); + +static int setcolreg(struct drm_crtc *crtc, u16 red, u16 green, + u16 blue, u16 regno, struct fb_info *info) +{ + struct drm_fb_helper *fb_helper = info->par; + struct drm_framebuffer *fb = fb_helper->fb; + int pindex; + + if (info->fix.visual == FB_VISUAL_TRUECOLOR) { + u32 *palette; + u32 value; + /* place color in psuedopalette */ + if (regno > 16) + return -EINVAL; + palette = (u32 *)info->pseudo_palette; + red >>= (16 - info->var.red.length); + green >>= (16 - info->var.green.length); + blue >>= (16 - info->var.blue.length); + value = (red << info->var.red.offset) | + (green << info->var.green.offset) | + (blue << info->var.blue.offset); + if (info->var.transp.length > 0) { + u32 mask = (1 << info->var.transp.length) - 1; + mask <<= info->var.transp.offset; + value |= mask; + } + palette[regno] = value; + return 0; + } + + pindex = regno; + + if (fb->bits_per_pixel == 16) { + pindex = regno << 3; + + if (fb->depth == 16 && regno > 63) + return -EINVAL; + if (fb->depth == 15 && regno > 31) + return -EINVAL; + + if (fb->depth == 16) { + u16 r, g, b; + int i; + if (regno < 32) { + for (i = 0; i < 8; i++) + fb_helper->funcs->gamma_set(crtc, red, + green, blue, pindex + i); + } + + fb_helper->funcs->gamma_get(crtc, &r, + &g, &b, + pindex >> 1); + + for (i = 0; i < 4; i++) + fb_helper->funcs->gamma_set(crtc, r, + green, b, + (pindex >> 1) + i); + } + } + + if (fb->depth != 16) + fb_helper->funcs->gamma_set(crtc, red, green, blue, pindex); + return 0; +} + +int drm_fb_helper_setcmap(struct fb_cmap *cmap, struct fb_info *info) +{ + struct drm_fb_helper *fb_helper = info->par; + struct drm_crtc_helper_funcs *crtc_funcs; + u16 *red, *green, *blue, *transp; + struct drm_crtc *crtc; + int i, j, rc = 0; + int start; + + for (i = 0; i < fb_helper->crtc_count; i++) { + crtc = fb_helper->crtc_info[i].mode_set.crtc; + crtc_funcs = crtc->helper_private; + + red = cmap->red; + green = cmap->green; + blue = cmap->blue; + transp = cmap->transp; + start = cmap->start; + + for (j = 0; j < cmap->len; j++) { + u16 hred, hgreen, hblue, htransp = 0xffff; + + hred = *red++; + hgreen = *green++; + hblue = *blue++; + + if (transp) + htransp = *transp++; + + rc = setcolreg(crtc, hred, hgreen, hblue, start++, info); + if (rc) + return rc; + } + crtc_funcs->load_lut(crtc); + } + return rc; +} +EXPORT_SYMBOL(drm_fb_helper_setcmap); + +int drm_fb_helper_check_var(struct fb_var_screeninfo *var, + struct fb_info *info) +{ + struct drm_fb_helper *fb_helper = info->par; + struct drm_framebuffer *fb = fb_helper->fb; + int depth; + + if (var->pixclock != 0 || in_dbg_master()) + return -EINVAL; + + /* Need to resize the fb object !!! */ + if (var->bits_per_pixel > fb->bits_per_pixel || + var->xres > fb->width || var->yres > fb->height || + var->xres_virtual > fb->width || var->yres_virtual > fb->height) { + DRM_DEBUG("fb userspace requested width/height/bpp is greater than current fb " + "request %dx%d-%d (virtual %dx%d) > %dx%d-%d\n", + var->xres, var->yres, var->bits_per_pixel, + var->xres_virtual, var->yres_virtual, + fb->width, fb->height, fb->bits_per_pixel); + return -EINVAL; + } + + switch (var->bits_per_pixel) { + case 16: + depth = (var->green.length == 6) ? 16 : 15; + break; + case 32: + depth = (var->transp.length > 0) ? 32 : 24; + break; + default: + depth = var->bits_per_pixel; + break; + } + + switch (depth) { + case 8: + var->red.offset = 0; + var->green.offset = 0; + var->blue.offset = 0; + var->red.length = 8; + var->green.length = 8; + var->blue.length = 8; + var->transp.length = 0; + var->transp.offset = 0; + break; + case 15: + var->red.offset = 10; + var->green.offset = 5; + var->blue.offset = 0; + var->red.length = 5; + var->green.length = 5; + var->blue.length = 5; + var->transp.length = 1; + var->transp.offset = 15; + break; + case 16: + var->red.offset = 11; + var->green.offset = 5; + var->blue.offset = 0; + var->red.length = 5; + var->green.length = 6; + var->blue.length = 5; + var->transp.length = 0; + var->transp.offset = 0; + break; + case 24: + var->red.offset = 16; + var->green.offset = 8; + var->blue.offset = 0; + var->red.length = 8; + var->green.length = 8; + var->blue.length = 8; + var->transp.length = 0; + var->transp.offset = 0; + break; + case 32: + var->red.offset = 16; + var->green.offset = 8; + var->blue.offset = 0; + var->red.length = 8; + var->green.length = 8; + var->blue.length = 8; + var->transp.length = 8; + var->transp.offset = 24; + break; + default: + return -EINVAL; + } + return 0; +} +EXPORT_SYMBOL(drm_fb_helper_check_var); + +/* this will let fbcon do the mode init */ +int drm_fb_helper_set_par(struct fb_info *info) +{ + struct drm_fb_helper *fb_helper = info->par; + struct drm_device *dev = fb_helper->dev; + struct fb_var_screeninfo *var = &info->var; + struct drm_crtc *crtc; + int ret; + int i; + + if (var->pixclock != 0) { + DRM_ERROR("PIXEL CLOCK SET\n"); + return -EINVAL; + } + + mutex_lock(&dev->mode_config.mutex); + for (i = 0; i < fb_helper->crtc_count; i++) { + crtc = fb_helper->crtc_info[i].mode_set.crtc; + ret = crtc->funcs->set_config(&fb_helper->crtc_info[i].mode_set); + if (ret) { + mutex_unlock(&dev->mode_config.mutex); + return ret; + } + } + mutex_unlock(&dev->mode_config.mutex); + + if (fb_helper->delayed_hotplug) { + fb_helper->delayed_hotplug = false; + drm_fb_helper_hotplug_event(fb_helper); + } + return 0; +} +EXPORT_SYMBOL(drm_fb_helper_set_par); + +int drm_fb_helper_pan_display(struct fb_var_screeninfo *var, + struct fb_info *info) +{ + struct drm_fb_helper *fb_helper = info->par; + struct drm_device *dev = fb_helper->dev; + struct drm_mode_set *modeset; + struct drm_crtc *crtc; + int ret = 0; + int i; + + mutex_lock(&dev->mode_config.mutex); + for (i = 0; i < fb_helper->crtc_count; i++) { + crtc = fb_helper->crtc_info[i].mode_set.crtc; + + modeset = &fb_helper->crtc_info[i].mode_set; + + modeset->x = var->xoffset; + modeset->y = var->yoffset; + + if (modeset->num_connectors) { + ret = crtc->funcs->set_config(modeset); + if (!ret) { + info->var.xoffset = var->xoffset; + info->var.yoffset = var->yoffset; + } + } + } + mutex_unlock(&dev->mode_config.mutex); + return ret; +} +EXPORT_SYMBOL(drm_fb_helper_pan_display); + +int drm_fb_helper_single_fb_probe(struct drm_fb_helper *fb_helper, + int preferred_bpp) +{ + int new_fb = 0; + int crtc_count = 0; + int i; + struct fb_info *info; + struct drm_fb_helper_surface_size sizes; + int gamma_size = 0; + + memset(&sizes, 0, sizeof(struct drm_fb_helper_surface_size)); + sizes.surface_depth = 24; + sizes.surface_bpp = 32; + sizes.fb_width = (unsigned)-1; + sizes.fb_height = (unsigned)-1; + + /* if driver picks 8 or 16 by default use that + for both depth/bpp */ + if (preferred_bpp != sizes.surface_bpp) { + sizes.surface_depth = sizes.surface_bpp = preferred_bpp; + } + /* first up get a count of crtcs now in use and new min/maxes width/heights */ + for (i = 0; i < fb_helper->connector_count; i++) { + struct drm_fb_helper_connector *fb_helper_conn = fb_helper->connector_info[i]; + struct drm_cmdline_mode *cmdline_mode; + + cmdline_mode = &fb_helper_conn->cmdline_mode; + + if (cmdline_mode->bpp_specified) { + switch (cmdline_mode->bpp) { + case 8: + sizes.surface_depth = sizes.surface_bpp = 8; + break; + case 15: + sizes.surface_depth = 15; + sizes.surface_bpp = 16; + break; + case 16: + sizes.surface_depth = sizes.surface_bpp = 16; + break; + case 24: + sizes.surface_depth = sizes.surface_bpp = 24; + break; + case 32: + sizes.surface_depth = 24; + sizes.surface_bpp = 32; + break; + } + break; + } + } + + crtc_count = 0; + for (i = 0; i < fb_helper->crtc_count; i++) { + struct drm_display_mode *desired_mode; + desired_mode = fb_helper->crtc_info[i].desired_mode; + + if (desired_mode) { + if (gamma_size == 0) + gamma_size = fb_helper->crtc_info[i].mode_set.crtc->gamma_size; + if (desired_mode->hdisplay < sizes.fb_width) + sizes.fb_width = desired_mode->hdisplay; + if (desired_mode->vdisplay < sizes.fb_height) + sizes.fb_height = desired_mode->vdisplay; + if (desired_mode->hdisplay > sizes.surface_width) + sizes.surface_width = desired_mode->hdisplay; + if (desired_mode->vdisplay > sizes.surface_height) + sizes.surface_height = desired_mode->vdisplay; + crtc_count++; + } + } + + if (crtc_count == 0 || sizes.fb_width == -1 || sizes.fb_height == -1) { + /* hmm everyone went away - assume VGA cable just fell out + and will come back later. */ + DRM_INFO("Cannot find any crtc or sizes - going 1024x768\n"); + sizes.fb_width = sizes.surface_width = 1024; + sizes.fb_height = sizes.surface_height = 768; + } + + /* push down into drivers */ + new_fb = (*fb_helper->funcs->fb_probe)(fb_helper, &sizes); + if (new_fb < 0) + return new_fb; + + info = fb_helper->fbdev; + + /* set the fb pointer */ + for (i = 0; i < fb_helper->crtc_count; i++) { + fb_helper->crtc_info[i].mode_set.fb = fb_helper->fb; + } + + if (new_fb) { + info->var.pixclock = 0; + if (register_framebuffer(info) < 0) { + return -EINVAL; + } + + printk(KERN_INFO "fb%d: %s frame buffer device\n", info->node, + info->fix.id); + + } else { + drm_fb_helper_set_par(info); + } + + /* Switch back to kernel console on panic */ + /* multi card linked list maybe */ + if (list_empty(&kernel_fb_helper_list)) { + printk(KERN_INFO "drm: registered panic notifier\n"); + atomic_notifier_chain_register(&panic_notifier_list, + &paniced); + register_sysrq_key('v', &sysrq_drm_fb_helper_restore_op); + } + if (new_fb) + list_add(&fb_helper->kernel_fb_list, &kernel_fb_helper_list); + + return 0; +} +EXPORT_SYMBOL(drm_fb_helper_single_fb_probe); + +void drm_fb_helper_fill_fix(struct fb_info *info, uint32_t pitch, + uint32_t depth) +{ + info->fix.type = FB_TYPE_PACKED_PIXELS; + info->fix.visual = depth == 8 ? FB_VISUAL_PSEUDOCOLOR : + FB_VISUAL_TRUECOLOR; + info->fix.mmio_start = 0; + info->fix.mmio_len = 0; + info->fix.type_aux = 0; + info->fix.xpanstep = 1; /* doing it in hw */ + info->fix.ypanstep = 1; /* doing it in hw */ + info->fix.ywrapstep = 0; + info->fix.accel = FB_ACCEL_NONE; + info->fix.type_aux = 0; + + info->fix.line_length = pitch; + return; +} +EXPORT_SYMBOL(drm_fb_helper_fill_fix); + +void drm_fb_helper_fill_var(struct fb_info *info, struct drm_fb_helper *fb_helper, + uint32_t fb_width, uint32_t fb_height) +{ + struct drm_framebuffer *fb = fb_helper->fb; + info->pseudo_palette = fb_helper->pseudo_palette; + info->var.xres_virtual = fb->width; + info->var.yres_virtual = fb->height; + info->var.bits_per_pixel = fb->bits_per_pixel; + info->var.accel_flags = FB_ACCELF_TEXT; + info->var.xoffset = 0; + info->var.yoffset = 0; + info->var.activate = FB_ACTIVATE_NOW; + info->var.height = -1; + info->var.width = -1; + + switch (fb->depth) { + case 8: + info->var.red.offset = 0; + info->var.green.offset = 0; + info->var.blue.offset = 0; + info->var.red.length = 8; /* 8bit DAC */ + info->var.green.length = 8; + info->var.blue.length = 8; + info->var.transp.offset = 0; + info->var.transp.length = 0; + break; + case 15: + info->var.red.offset = 10; + info->var.green.offset = 5; + info->var.blue.offset = 0; + info->var.red.length = 5; + info->var.green.length = 5; + info->var.blue.length = 5; + info->var.transp.offset = 15; + info->var.transp.length = 1; + break; + case 16: + info->var.red.offset = 11; + info->var.green.offset = 5; + info->var.blue.offset = 0; + info->var.red.length = 5; + info->var.green.length = 6; + info->var.blue.length = 5; + info->var.transp.offset = 0; + break; + case 24: + info->var.red.offset = 16; + info->var.green.offset = 8; + info->var.blue.offset = 0; + info->var.red.length = 8; + info->var.green.length = 8; + info->var.blue.length = 8; + info->var.transp.offset = 0; + info->var.transp.length = 0; + break; + case 32: + info->var.red.offset = 16; + info->var.green.offset = 8; + info->var.blue.offset = 0; + info->var.red.length = 8; + info->var.green.length = 8; + info->var.blue.length = 8; + info->var.transp.offset = 24; + info->var.transp.length = 8; + break; + default: + break; + } + + info->var.xres = fb_width; + info->var.yres = fb_height; +} +EXPORT_SYMBOL(drm_fb_helper_fill_var); + +static int drm_fb_helper_probe_connector_modes(struct drm_fb_helper *fb_helper, + uint32_t maxX, + uint32_t maxY) +{ + struct drm_connector *connector; + int count = 0; + int i; + + for (i = 0; i < fb_helper->connector_count; i++) { + connector = fb_helper->connector_info[i]->connector; + count += connector->funcs->fill_modes(connector, maxX, maxY); + } + + return count; +} + +static struct drm_display_mode *drm_has_preferred_mode(struct drm_fb_helper_connector *fb_connector, int width, int height) +{ + struct drm_display_mode *mode; + + list_for_each_entry(mode, &fb_connector->connector->modes, head) { + if (drm_mode_width(mode) > width || + drm_mode_height(mode) > height) + continue; + if (mode->type & DRM_MODE_TYPE_PREFERRED) + return mode; + } + return NULL; +} + +static bool drm_has_cmdline_mode(struct drm_fb_helper_connector *fb_connector) +{ + struct drm_cmdline_mode *cmdline_mode; + cmdline_mode = &fb_connector->cmdline_mode; + return cmdline_mode->specified; +} + +static struct drm_display_mode *drm_pick_cmdline_mode(struct drm_fb_helper_connector *fb_helper_conn, + int width, int height) +{ + struct drm_cmdline_mode *cmdline_mode; + struct drm_display_mode *mode = NULL; + + cmdline_mode = &fb_helper_conn->cmdline_mode; + if (cmdline_mode->specified == false) + return mode; + + /* attempt to find a matching mode in the list of modes + * we have gotten so far, if not add a CVT mode that conforms + */ + if (cmdline_mode->rb || cmdline_mode->margins) + goto create_mode; + + list_for_each_entry(mode, &fb_helper_conn->connector->modes, head) { + /* check width/height */ + if (mode->hdisplay != cmdline_mode->xres || + mode->vdisplay != cmdline_mode->yres) + continue; + + if (cmdline_mode->refresh_specified) { + if (mode->vrefresh != cmdline_mode->refresh) + continue; + } + + if (cmdline_mode->interlace) { + if (!(mode->flags & DRM_MODE_FLAG_INTERLACE)) + continue; + } + return mode; + } + +create_mode: + mode = drm_mode_create_from_cmdline_mode(fb_helper_conn->connector->dev, + cmdline_mode); + list_add(&mode->head, &fb_helper_conn->connector->modes); + return mode; +} + +static bool drm_connector_enabled(struct drm_connector *connector, bool strict) +{ + bool enable; + + if (strict) { + enable = connector->status == connector_status_connected; + } else { + enable = connector->status != connector_status_disconnected; + } + return enable; +} + +static void drm_enable_connectors(struct drm_fb_helper *fb_helper, + bool *enabled) +{ + bool any_enabled = false; + struct drm_connector *connector; + int i = 0; + + for (i = 0; i < fb_helper->connector_count; i++) { + connector = fb_helper->connector_info[i]->connector; + enabled[i] = drm_connector_enabled(connector, true); + DRM_DEBUG_KMS("connector %d enabled? %s\n", connector->base.id, + enabled[i] ? "yes" : "no"); + any_enabled |= enabled[i]; + } + + if (any_enabled) + return; + + for (i = 0; i < fb_helper->connector_count; i++) { + connector = fb_helper->connector_info[i]->connector; + enabled[i] = drm_connector_enabled(connector, false); + } +} + +static bool drm_target_cloned(struct drm_fb_helper *fb_helper, + struct drm_display_mode **modes, + bool *enabled, int width, int height) +{ + int count, i, j; + bool can_clone = false; + struct drm_fb_helper_connector *fb_helper_conn; + struct drm_display_mode *dmt_mode, *mode; + + /* only contemplate cloning in the single crtc case */ + if (fb_helper->crtc_count > 1) + return false; + + count = 0; + for (i = 0; i < fb_helper->connector_count; i++) { + if (enabled[i]) + count++; + } + + /* only contemplate cloning if more than one connector is enabled */ + if (count <= 1) + return false; + + /* check the command line or if nothing common pick 1024x768 */ + can_clone = true; + for (i = 0; i < fb_helper->connector_count; i++) { + if (!enabled[i]) + continue; + fb_helper_conn = fb_helper->connector_info[i]; + modes[i] = drm_pick_cmdline_mode(fb_helper_conn, width, height); + if (!modes[i]) { + can_clone = false; + break; + } + for (j = 0; j < i; j++) { + if (!enabled[j]) + continue; + if (!drm_mode_equal(modes[j], modes[i])) + can_clone = false; + } + } + + if (can_clone) { + DRM_DEBUG_KMS("can clone using command line\n"); + return true; + } + + /* try and find a 1024x768 mode on each connector */ + can_clone = true; + dmt_mode = drm_mode_find_dmt(fb_helper->dev, 1024, 768, 60); + + for (i = 0; i < fb_helper->connector_count; i++) { + + if (!enabled[i]) + continue; + + fb_helper_conn = fb_helper->connector_info[i]; + list_for_each_entry(mode, &fb_helper_conn->connector->modes, head) { + if (drm_mode_equal(mode, dmt_mode)) + modes[i] = mode; + } + if (!modes[i]) + can_clone = false; + } + + if (can_clone) { + DRM_DEBUG_KMS("can clone using 1024x768\n"); + return true; + } + DRM_INFO("kms: can't enable cloning when we probably wanted to.\n"); + return false; +} + +static bool drm_target_preferred(struct drm_fb_helper *fb_helper, + struct drm_display_mode **modes, + bool *enabled, int width, int height) +{ + struct drm_fb_helper_connector *fb_helper_conn; + int i; + + for (i = 0; i < fb_helper->connector_count; i++) { + fb_helper_conn = fb_helper->connector_info[i]; + + if (enabled[i] == false) + continue; + + DRM_DEBUG_KMS("looking for cmdline mode on connector %d\n", + fb_helper_conn->connector->base.id); + + /* got for command line mode first */ + modes[i] = drm_pick_cmdline_mode(fb_helper_conn, width, height); + if (!modes[i]) { + DRM_DEBUG_KMS("looking for preferred mode on connector %d\n", + fb_helper_conn->connector->base.id); + modes[i] = drm_has_preferred_mode(fb_helper_conn, width, height); + } + /* No preferred modes, pick one off the list */ + if (!modes[i] && !list_empty(&fb_helper_conn->connector->modes)) { + list_for_each_entry(modes[i], &fb_helper_conn->connector->modes, head) + break; + } + DRM_DEBUG_KMS("found mode %s\n", modes[i] ? modes[i]->name : + "none"); + } + return true; +} + +static int drm_pick_crtcs(struct drm_fb_helper *fb_helper, + struct drm_fb_helper_crtc **best_crtcs, + struct drm_display_mode **modes, + int n, int width, int height) +{ + int c, o; + struct drm_device *dev = fb_helper->dev; + struct drm_connector *connector; + struct drm_connector_helper_funcs *connector_funcs; + struct drm_encoder *encoder; + struct drm_fb_helper_crtc *best_crtc; + int my_score, best_score, score; + struct drm_fb_helper_crtc **crtcs, *crtc; + struct drm_fb_helper_connector *fb_helper_conn; + + if (n == fb_helper->connector_count) + return 0; + + fb_helper_conn = fb_helper->connector_info[n]; + connector = fb_helper_conn->connector; + + best_crtcs[n] = NULL; + best_crtc = NULL; + best_score = drm_pick_crtcs(fb_helper, best_crtcs, modes, n+1, width, height); + if (modes[n] == NULL) + return best_score; + + crtcs = kzalloc(dev->mode_config.num_connector * + sizeof(struct drm_fb_helper_crtc *), GFP_KERNEL); + if (!crtcs) + return best_score; + + my_score = 1; + if (connector->status == connector_status_connected) + my_score++; + if (drm_has_cmdline_mode(fb_helper_conn)) + my_score++; + if (drm_has_preferred_mode(fb_helper_conn, width, height)) + my_score++; + + connector_funcs = connector->helper_private; + encoder = connector_funcs->best_encoder(connector); + if (!encoder) + goto out; + + /* select a crtc for this connector and then attempt to configure + remaining connectors */ + for (c = 0; c < fb_helper->crtc_count; c++) { + crtc = &fb_helper->crtc_info[c]; + + if ((encoder->possible_crtcs & (1 << c)) == 0) { + continue; + } + + for (o = 0; o < n; o++) + if (best_crtcs[o] == crtc) + break; + + if (o < n) { + /* ignore cloning unless only a single crtc */ + if (fb_helper->crtc_count > 1) + continue; + + if (!drm_mode_equal(modes[o], modes[n])) + continue; + } + + crtcs[n] = crtc; + memcpy(crtcs, best_crtcs, n * sizeof(struct drm_fb_helper_crtc *)); + score = my_score + drm_pick_crtcs(fb_helper, crtcs, modes, n + 1, + width, height); + if (score > best_score) { + best_crtc = crtc; + best_score = score; + memcpy(best_crtcs, crtcs, + dev->mode_config.num_connector * + sizeof(struct drm_fb_helper_crtc *)); + } + } +out: + kfree(crtcs); + return best_score; +} + +static void drm_setup_crtcs(struct drm_fb_helper *fb_helper) +{ + struct drm_device *dev = fb_helper->dev; + struct drm_fb_helper_crtc **crtcs; + struct drm_display_mode **modes; + struct drm_encoder *encoder; + struct drm_mode_set *modeset; + bool *enabled; + int width, height; + int i, ret; + + DRM_DEBUG_KMS("\n"); + + width = dev->mode_config.max_width; + height = dev->mode_config.max_height; + + /* clean out all the encoder/crtc combos */ + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + encoder->crtc = NULL; + } + + crtcs = kcalloc(dev->mode_config.num_connector, + sizeof(struct drm_fb_helper_crtc *), GFP_KERNEL); + modes = kcalloc(dev->mode_config.num_connector, + sizeof(struct drm_display_mode *), GFP_KERNEL); + enabled = kcalloc(dev->mode_config.num_connector, + sizeof(bool), GFP_KERNEL); + + drm_enable_connectors(fb_helper, enabled); + + ret = drm_target_cloned(fb_helper, modes, enabled, width, height); + if (!ret) { + ret = drm_target_preferred(fb_helper, modes, enabled, width, height); + if (!ret) + DRM_ERROR("Unable to find initial modes\n"); + } + + DRM_DEBUG_KMS("picking CRTCs for %dx%d config\n", width, height); + + drm_pick_crtcs(fb_helper, crtcs, modes, 0, width, height); + + /* need to set the modesets up here for use later */ + /* fill out the connector<->crtc mappings into the modesets */ + for (i = 0; i < fb_helper->crtc_count; i++) { + modeset = &fb_helper->crtc_info[i].mode_set; + modeset->num_connectors = 0; + } + + for (i = 0; i < fb_helper->connector_count; i++) { + struct drm_display_mode *mode = modes[i]; + struct drm_fb_helper_crtc *fb_crtc = crtcs[i]; + modeset = &fb_crtc->mode_set; + + if (mode && fb_crtc) { + DRM_DEBUG_KMS("desired mode %s set on crtc %d\n", + mode->name, fb_crtc->mode_set.crtc->base.id); + fb_crtc->desired_mode = mode; + if (modeset->mode) + drm_mode_destroy(dev, modeset->mode); + modeset->mode = drm_mode_duplicate(dev, + fb_crtc->desired_mode); + modeset->connectors[modeset->num_connectors++] = fb_helper->connector_info[i]->connector; + } + } + + kfree(crtcs); + kfree(modes); + kfree(enabled); +} + +/** + * drm_helper_initial_config - setup a sane initial connector configuration + * @dev: DRM device + * + * LOCKING: + * Called at init time, must take mode config lock. + * + * Scan the CRTCs and connectors and try to put together an initial setup. + * At the moment, this is a cloned configuration across all heads with + * a new framebuffer object as the backing store. + * + * RETURNS: + * Zero if everything went ok, nonzero otherwise. + */ +bool drm_fb_helper_initial_config(struct drm_fb_helper *fb_helper, int bpp_sel) +{ + struct drm_device *dev = fb_helper->dev; + int count = 0; + + /* disable all the possible outputs/crtcs before entering KMS mode */ + drm_helper_disable_unused_functions(fb_helper->dev); + + drm_fb_helper_parse_command_line(fb_helper); + + count = drm_fb_helper_probe_connector_modes(fb_helper, + dev->mode_config.max_width, + dev->mode_config.max_height); + /* + * we shouldn't end up with no modes here. + */ + if (count == 0) { + printk(KERN_INFO "No connectors reported connected with modes\n"); + } + drm_setup_crtcs(fb_helper); + + return drm_fb_helper_single_fb_probe(fb_helper, bpp_sel); +} +EXPORT_SYMBOL(drm_fb_helper_initial_config); + +/** + * drm_fb_helper_hotplug_event - respond to a hotplug notification by + * probing all the outputs attached to the fb. + * @fb_helper: the drm_fb_helper + * + * LOCKING: + * Called at runtime, must take mode config lock. + * + * Scan the connectors attached to the fb_helper and try to put together a + * setup after *notification of a change in output configuration. + * + * RETURNS: + * 0 on success and a non-zero error code otherwise. + */ +int drm_fb_helper_hotplug_event(struct drm_fb_helper *fb_helper) +{ + struct drm_device *dev = fb_helper->dev; + int count = 0; + u32 max_width, max_height, bpp_sel; + bool bound = false, crtcs_bound = false; + struct drm_crtc *crtc; + + if (!fb_helper->fb) + return 0; + + mutex_lock(&dev->mode_config.mutex); + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { + if (crtc->fb) + crtcs_bound = true; + if (crtc->fb == fb_helper->fb) + bound = true; + } + + if (!bound && crtcs_bound) { + fb_helper->delayed_hotplug = true; + mutex_unlock(&dev->mode_config.mutex); + return 0; + } + DRM_DEBUG_KMS("\n"); + + max_width = fb_helper->fb->width; + max_height = fb_helper->fb->height; + bpp_sel = fb_helper->fb->bits_per_pixel; + + count = drm_fb_helper_probe_connector_modes(fb_helper, max_width, + max_height); + drm_setup_crtcs(fb_helper); + mutex_unlock(&dev->mode_config.mutex); + + return drm_fb_helper_single_fb_probe(fb_helper, bpp_sel); +} +EXPORT_SYMBOL(drm_fb_helper_hotplug_event); + +/* The Kconfig DRM_KMS_HELPER selects FRAMEBUFFER_CONSOLE (if !EXPERT) + * but the module doesn't depend on any fb console symbols. At least + * attempt to load fbcon to avoid leaving the system without a usable console. + */ +#if defined(CONFIG_FRAMEBUFFER_CONSOLE_MODULE) && !defined(CONFIG_EXPERT) +static int __init drm_fb_helper_modinit(void) +{ + const char *name = "fbcon"; + struct module *fbcon; + + mutex_lock(&module_mutex); + fbcon = find_module(name); + mutex_unlock(&module_mutex); + + if (!fbcon) + request_module_nowait(name); + return 0; +} + +module_init(drm_fb_helper_modinit); +#endif diff --git a/drivers/gpu/drm/drm_fops.c b/drivers/gpu/drm/drm_fops.c new file mode 100644 index 00000000..c42e12cc --- /dev/null +++ b/drivers/gpu/drm/drm_fops.c @@ -0,0 +1,658 @@ +/** + * \file drm_fops.c + * File operations for DRM + * + * \author Rickard E. (Rik) Faith + * \author Daryll Strauss + * \author Gareth Hughes + */ + +/* + * Created: Mon Jan 4 08:58:31 1999 by faith@valinux.com + * + * Copyright 1999 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * VA LINUX SYSTEMS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#include "drmP.h" +#include +#include + +/* from BKL pushdown: note that nothing else serializes idr_find() */ +DEFINE_MUTEX(drm_global_mutex); +EXPORT_SYMBOL(drm_global_mutex); + +static int drm_open_helper(struct inode *inode, struct file *filp, + struct drm_device * dev); + +static int drm_setup(struct drm_device * dev) +{ + int i; + int ret; + + if (dev->driver->firstopen) { + ret = dev->driver->firstopen(dev); + if (ret != 0) + return ret; + } + + atomic_set(&dev->ioctl_count, 0); + atomic_set(&dev->vma_count, 0); + + if (drm_core_check_feature(dev, DRIVER_HAVE_DMA) && + !drm_core_check_feature(dev, DRIVER_MODESET)) { + dev->buf_use = 0; + atomic_set(&dev->buf_alloc, 0); + + i = drm_dma_setup(dev); + if (i < 0) + return i; + } + + for (i = 0; i < ARRAY_SIZE(dev->counts); i++) + atomic_set(&dev->counts[i], 0); + + dev->sigdata.lock = NULL; + + dev->queue_count = 0; + dev->queue_reserved = 0; + dev->queue_slots = 0; + dev->queuelist = NULL; + dev->context_flag = 0; + dev->interrupt_flag = 0; + dev->dma_flag = 0; + dev->last_context = 0; + dev->last_switch = 0; + dev->last_checked = 0; + init_waitqueue_head(&dev->context_wait); + dev->if_version = 0; + + dev->ctx_start = 0; + dev->lck_start = 0; + + dev->buf_async = NULL; + init_waitqueue_head(&dev->buf_readers); + init_waitqueue_head(&dev->buf_writers); + + DRM_DEBUG("\n"); + + /* + * The kernel's context could be created here, but is now created + * in drm_dma_enqueue. This is more resource-efficient for + * hardware that does not do DMA, but may mean that + * drm_select_queue fails between the time the interrupt is + * initialized and the time the queues are initialized. + */ + + return 0; +} + +/** + * Open file. + * + * \param inode device inode + * \param filp file pointer. + * \return zero on success or a negative number on failure. + * + * Searches the DRM device with the same minor number, calls open_helper(), and + * increments the device open count. If the open count was previous at zero, + * i.e., it's the first that the device is open, then calls setup(). + */ +int drm_open(struct inode *inode, struct file *filp) +{ + struct drm_device *dev = NULL; + int minor_id = iminor(inode); + struct drm_minor *minor; + int retcode = 0; + + minor = idr_find(&drm_minors_idr, minor_id); + if (!minor) + return -ENODEV; + + if (!(dev = minor->dev)) + return -ENODEV; + + retcode = drm_open_helper(inode, filp, dev); + if (!retcode) { + atomic_inc(&dev->counts[_DRM_STAT_OPENS]); + if (!dev->open_count++) + retcode = drm_setup(dev); + } + if (!retcode) { + mutex_lock(&dev->struct_mutex); + if (minor->type == DRM_MINOR_LEGACY) { + if (dev->dev_mapping == NULL) + dev->dev_mapping = inode->i_mapping; + else if (dev->dev_mapping != inode->i_mapping) + retcode = -ENODEV; + } + mutex_unlock(&dev->struct_mutex); + } + + return retcode; +} +EXPORT_SYMBOL(drm_open); + +/** + * File \c open operation. + * + * \param inode device inode. + * \param filp file pointer. + * + * Puts the dev->fops corresponding to the device minor number into + * \p filp, call the \c open method, and restore the file operations. + */ +int drm_stub_open(struct inode *inode, struct file *filp) +{ + struct drm_device *dev = NULL; + struct drm_minor *minor; + int minor_id = iminor(inode); + int err = -ENODEV; + const struct file_operations *old_fops; + + DRM_DEBUG("\n"); + + mutex_lock(&drm_global_mutex); + minor = idr_find(&drm_minors_idr, minor_id); + if (!minor) + goto out; + + if (!(dev = minor->dev)) + goto out; + + old_fops = filp->f_op; + filp->f_op = fops_get(&dev->driver->fops); + if (filp->f_op == NULL) { + filp->f_op = old_fops; + goto out; + } + if (filp->f_op->open && (err = filp->f_op->open(inode, filp))) { + fops_put(filp->f_op); + filp->f_op = fops_get(old_fops); + } + fops_put(old_fops); + +out: + mutex_unlock(&drm_global_mutex); + return err; +} + +/** + * Check whether DRI will run on this CPU. + * + * \return non-zero if the DRI will run on this CPU, or zero otherwise. + */ +static int drm_cpu_valid(void) +{ +#if defined(__i386__) + if (boot_cpu_data.x86 == 3) + return 0; /* No cmpxchg on a 386 */ +#endif +#if defined(__sparc__) && !defined(__sparc_v9__) + return 0; /* No cmpxchg before v9 sparc. */ +#endif + return 1; +} + +/** + * Called whenever a process opens /dev/drm. + * + * \param inode device inode. + * \param filp file pointer. + * \param dev device. + * \return zero on success or a negative number on failure. + * + * Creates and initializes a drm_file structure for the file private data in \p + * filp and add it into the double linked list in \p dev. + */ +static int drm_open_helper(struct inode *inode, struct file *filp, + struct drm_device * dev) +{ + int minor_id = iminor(inode); + struct drm_file *priv; + int ret; + + if (filp->f_flags & O_EXCL) + return -EBUSY; /* No exclusive opens */ + if (!drm_cpu_valid()) + return -EINVAL; + if (dev->switch_power_state != DRM_SWITCH_POWER_ON) + return -EINVAL; + + DRM_DEBUG("pid = %d, minor = %d\n", task_pid_nr(current), minor_id); + + priv = kzalloc(sizeof(*priv), GFP_KERNEL); + if (!priv) + return -ENOMEM; + + filp->private_data = priv; + priv->filp = filp; + priv->uid = current_euid(); + priv->pid = task_pid_nr(current); + priv->minor = idr_find(&drm_minors_idr, minor_id); + priv->ioctl_count = 0; + /* for compatibility root is always authenticated */ + priv->authenticated = capable(CAP_SYS_ADMIN); + priv->lock_count = 0; + + INIT_LIST_HEAD(&priv->lhead); + INIT_LIST_HEAD(&priv->fbs); + INIT_LIST_HEAD(&priv->event_list); + init_waitqueue_head(&priv->event_wait); + priv->event_space = 4096; /* set aside 4k for event buffer */ + + if (dev->driver->driver_features & DRIVER_GEM) + drm_gem_open(dev, priv); + + if (dev->driver->open) { + ret = dev->driver->open(dev, priv); + if (ret < 0) + goto out_free; + } + + + /* if there is no current master make this fd it */ + mutex_lock(&dev->struct_mutex); + if (!priv->minor->master) { + /* create a new master */ + priv->minor->master = drm_master_create(priv->minor); + if (!priv->minor->master) { + mutex_unlock(&dev->struct_mutex); + ret = -ENOMEM; + goto out_free; + } + + priv->is_master = 1; + /* take another reference for the copy in the local file priv */ + priv->master = drm_master_get(priv->minor->master); + + priv->authenticated = 1; + + mutex_unlock(&dev->struct_mutex); + if (dev->driver->master_create) { + ret = dev->driver->master_create(dev, priv->master); + if (ret) { + mutex_lock(&dev->struct_mutex); + /* drop both references if this fails */ + drm_master_put(&priv->minor->master); + drm_master_put(&priv->master); + mutex_unlock(&dev->struct_mutex); + goto out_free; + } + } + mutex_lock(&dev->struct_mutex); + if (dev->driver->master_set) { + ret = dev->driver->master_set(dev, priv, true); + if (ret) { + /* drop both references if this fails */ + drm_master_put(&priv->minor->master); + drm_master_put(&priv->master); + mutex_unlock(&dev->struct_mutex); + goto out_free; + } + } + mutex_unlock(&dev->struct_mutex); + } else { + /* get a reference to the master */ + priv->master = drm_master_get(priv->minor->master); + mutex_unlock(&dev->struct_mutex); + } + + mutex_lock(&dev->struct_mutex); + list_add(&priv->lhead, &dev->filelist); + mutex_unlock(&dev->struct_mutex); + +#ifdef __alpha__ + /* + * Default the hose + */ + if (!dev->hose) { + struct pci_dev *pci_dev; + pci_dev = pci_get_class(PCI_CLASS_DISPLAY_VGA << 8, NULL); + if (pci_dev) { + dev->hose = pci_dev->sysdata; + pci_dev_put(pci_dev); + } + if (!dev->hose) { + struct pci_bus *b = pci_bus_b(pci_root_buses.next); + if (b) + dev->hose = b->sysdata; + } + } +#endif + + return 0; + out_free: + kfree(priv); + filp->private_data = NULL; + return ret; +} + +/** No-op. */ +int drm_fasync(int fd, struct file *filp, int on) +{ + struct drm_file *priv = filp->private_data; + struct drm_device *dev = priv->minor->dev; + + DRM_DEBUG("fd = %d, device = 0x%lx\n", fd, + (long)old_encode_dev(priv->minor->device)); + return fasync_helper(fd, filp, on, &dev->buf_async); +} +EXPORT_SYMBOL(drm_fasync); + +/* + * Reclaim locked buffers; note that this may be a bad idea if the current + * context doesn't have the hw lock... + */ +static void drm_reclaim_locked_buffers(struct drm_device *dev, struct file *f) +{ + struct drm_file *file_priv = f->private_data; + + if (drm_i_have_hw_lock(dev, file_priv)) { + dev->driver->reclaim_buffers_locked(dev, file_priv); + } else { + unsigned long _end = jiffies + 3 * DRM_HZ; + int locked = 0; + + drm_idlelock_take(&file_priv->master->lock); + + /* + * Wait for a while. + */ + do { + spin_lock_bh(&file_priv->master->lock.spinlock); + locked = file_priv->master->lock.idle_has_lock; + spin_unlock_bh(&file_priv->master->lock.spinlock); + if (locked) + break; + schedule(); + } while (!time_after_eq(jiffies, _end)); + + if (!locked) { + DRM_ERROR("reclaim_buffers_locked() deadlock. Please rework this\n" + "\tdriver to use reclaim_buffers_idlelocked() instead.\n" + "\tI will go on reclaiming the buffers anyway.\n"); + } + + dev->driver->reclaim_buffers_locked(dev, file_priv); + drm_idlelock_release(&file_priv->master->lock); + } +} + +static void drm_master_release(struct drm_device *dev, struct file *filp) +{ + struct drm_file *file_priv = filp->private_data; + + if (dev->driver->reclaim_buffers_locked && + file_priv->master->lock.hw_lock) + drm_reclaim_locked_buffers(dev, filp); + + if (dev->driver->reclaim_buffers_idlelocked && + file_priv->master->lock.hw_lock) { + drm_idlelock_take(&file_priv->master->lock); + dev->driver->reclaim_buffers_idlelocked(dev, file_priv); + drm_idlelock_release(&file_priv->master->lock); + } + + + if (drm_i_have_hw_lock(dev, file_priv)) { + DRM_DEBUG("File %p released, freeing lock for context %d\n", + filp, _DRM_LOCKING_CONTEXT(file_priv->master->lock.hw_lock->lock)); + drm_lock_free(&file_priv->master->lock, + _DRM_LOCKING_CONTEXT(file_priv->master->lock.hw_lock->lock)); + } + + if (drm_core_check_feature(dev, DRIVER_HAVE_DMA) && + !dev->driver->reclaim_buffers_locked) { + dev->driver->reclaim_buffers(dev, file_priv); + } +} + +static void drm_events_release(struct drm_file *file_priv) +{ + struct drm_device *dev = file_priv->minor->dev; + struct drm_pending_event *e, *et; + struct drm_pending_vblank_event *v, *vt; + unsigned long flags; + + spin_lock_irqsave(&dev->event_lock, flags); + + /* Remove pending flips */ + list_for_each_entry_safe(v, vt, &dev->vblank_event_list, base.link) + if (v->base.file_priv == file_priv) { + list_del(&v->base.link); + drm_vblank_put(dev, v->pipe); + v->base.destroy(&v->base); + } + + /* Remove unconsumed events */ + list_for_each_entry_safe(e, et, &file_priv->event_list, link) + e->destroy(e); + + spin_unlock_irqrestore(&dev->event_lock, flags); +} + +/** + * Release file. + * + * \param inode device inode + * \param file_priv DRM file private. + * \return zero on success or a negative number on failure. + * + * If the hardware lock is held then free it, and take it again for the kernel + * context since it's necessary to reclaim buffers. Unlink the file private + * data from its list and free it. Decreases the open count and if it reaches + * zero calls drm_lastclose(). + */ +int drm_release(struct inode *inode, struct file *filp) +{ + struct drm_file *file_priv = filp->private_data; + struct drm_device *dev = file_priv->minor->dev; + int retcode = 0; + + mutex_lock(&drm_global_mutex); + + DRM_DEBUG("open_count = %d\n", dev->open_count); + + if (dev->driver->preclose) + dev->driver->preclose(dev, file_priv); + + /* ======================================================== + * Begin inline drm_release + */ + + DRM_DEBUG("pid = %d, device = 0x%lx, open_count = %d\n", + task_pid_nr(current), + (long)old_encode_dev(file_priv->minor->device), + dev->open_count); + + /* Release any auth tokens that might point to this file_priv, + (do that under the drm_global_mutex) */ + if (file_priv->magic) + (void) drm_remove_magic(file_priv->master, file_priv->magic); + + /* if the master has gone away we can't do anything with the lock */ + if (file_priv->minor->master) + drm_master_release(dev, filp); + + drm_events_release(file_priv); + + if (dev->driver->driver_features & DRIVER_GEM) + drm_gem_release(dev, file_priv); + + if (dev->driver->driver_features & DRIVER_MODESET) + drm_fb_release(file_priv); + + mutex_lock(&dev->ctxlist_mutex); + if (!list_empty(&dev->ctxlist)) { + struct drm_ctx_list *pos, *n; + + list_for_each_entry_safe(pos, n, &dev->ctxlist, head) { + if (pos->tag == file_priv && + pos->handle != DRM_KERNEL_CONTEXT) { + if (dev->driver->context_dtor) + dev->driver->context_dtor(dev, + pos->handle); + + drm_ctxbitmap_free(dev, pos->handle); + + list_del(&pos->head); + kfree(pos); + --dev->ctx_count; + } + } + } + mutex_unlock(&dev->ctxlist_mutex); + + mutex_lock(&dev->struct_mutex); + + if (file_priv->is_master) { + struct drm_master *master = file_priv->master; + struct drm_file *temp; + list_for_each_entry(temp, &dev->filelist, lhead) { + if ((temp->master == file_priv->master) && + (temp != file_priv)) + temp->authenticated = 0; + } + + /** + * Since the master is disappearing, so is the + * possibility to lock. + */ + + if (master->lock.hw_lock) { + if (dev->sigdata.lock == master->lock.hw_lock) + dev->sigdata.lock = NULL; + master->lock.hw_lock = NULL; + master->lock.file_priv = NULL; + wake_up_interruptible_all(&master->lock.lock_queue); + } + + if (file_priv->minor->master == file_priv->master) { + /* drop the reference held my the minor */ + if (dev->driver->master_drop) + dev->driver->master_drop(dev, file_priv, true); + drm_master_put(&file_priv->minor->master); + } + } + + /* drop the reference held my the file priv */ + drm_master_put(&file_priv->master); + file_priv->is_master = 0; + list_del(&file_priv->lhead); + mutex_unlock(&dev->struct_mutex); + + if (dev->driver->postclose) + dev->driver->postclose(dev, file_priv); + kfree(file_priv); + + /* ======================================================== + * End inline drm_release + */ + + atomic_inc(&dev->counts[_DRM_STAT_CLOSES]); + if (!--dev->open_count) { + if (atomic_read(&dev->ioctl_count)) { + DRM_ERROR("Device busy: %d\n", + atomic_read(&dev->ioctl_count)); + retcode = -EBUSY; + } else + retcode = drm_lastclose(dev); + } + mutex_unlock(&drm_global_mutex); + + return retcode; +} +EXPORT_SYMBOL(drm_release); + +static bool +drm_dequeue_event(struct drm_file *file_priv, + size_t total, size_t max, struct drm_pending_event **out) +{ + struct drm_device *dev = file_priv->minor->dev; + struct drm_pending_event *e; + unsigned long flags; + bool ret = false; + + spin_lock_irqsave(&dev->event_lock, flags); + + *out = NULL; + if (list_empty(&file_priv->event_list)) + goto out; + e = list_first_entry(&file_priv->event_list, + struct drm_pending_event, link); + if (e->event->length + total > max) + goto out; + + file_priv->event_space += e->event->length; + list_del(&e->link); + *out = e; + ret = true; + +out: + spin_unlock_irqrestore(&dev->event_lock, flags); + return ret; +} + +ssize_t drm_read(struct file *filp, char __user *buffer, + size_t count, loff_t *offset) +{ + struct drm_file *file_priv = filp->private_data; + struct drm_pending_event *e; + size_t total; + ssize_t ret; + + ret = wait_event_interruptible(file_priv->event_wait, + !list_empty(&file_priv->event_list)); + if (ret < 0) + return ret; + + total = 0; + while (drm_dequeue_event(file_priv, total, count, &e)) { + if (copy_to_user(buffer + total, + e->event, e->event->length)) { + total = -EFAULT; + break; + } + + total += e->event->length; + e->destroy(e); + } + + return total; +} +EXPORT_SYMBOL(drm_read); + +unsigned int drm_poll(struct file *filp, struct poll_table_struct *wait) +{ + struct drm_file *file_priv = filp->private_data; + unsigned int mask = 0; + + poll_wait(filp, &file_priv->event_wait, wait); + + if (!list_empty(&file_priv->event_list)) + mask |= POLLIN | POLLRDNORM; + + return mask; +} +EXPORT_SYMBOL(drm_poll); diff --git a/drivers/gpu/drm/drm_gem.c b/drivers/gpu/drm/drm_gem.c new file mode 100644 index 00000000..4012fe42 --- /dev/null +++ b/drivers/gpu/drm/drm_gem.c @@ -0,0 +1,584 @@ +/* + * Copyright © 2008 Intel Corporation + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS + * IN THE SOFTWARE. + * + * Authors: + * Eric Anholt + * + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include "drmP.h" + +/** @file drm_gem.c + * + * This file provides some of the base ioctls and library routines for + * the graphics memory manager implemented by each device driver. + * + * Because various devices have different requirements in terms of + * synchronization and migration strategies, implementing that is left up to + * the driver, and all that the general API provides should be generic -- + * allocating objects, reading/writing data with the cpu, freeing objects. + * Even there, platform-dependent optimizations for reading/writing data with + * the CPU mean we'll likely hook those out to driver-specific calls. However, + * the DRI2 implementation wants to have at least allocate/mmap be generic. + * + * The goal was to have swap-backed object allocation managed through + * struct file. However, file descriptors as handles to a struct file have + * two major failings: + * - Process limits prevent more than 1024 or so being used at a time by + * default. + * - Inability to allocate high fds will aggravate the X Server's select() + * handling, and likely that of many GL client applications as well. + * + * This led to a plan of using our own integer IDs (called handles, following + * DRM terminology) to mimic fds, and implement the fd syscalls we need as + * ioctls. The objects themselves will still include the struct file so + * that we can transition to fds if the required kernel infrastructure shows + * up at a later date, and as our interface with shmfs for memory allocation. + */ + +/* + * We make up offsets for buffer objects so we can recognize them at + * mmap time. + */ + +/* pgoff in mmap is an unsigned long, so we need to make sure that + * the faked up offset will fit + */ + +#if BITS_PER_LONG == 64 +#define DRM_FILE_PAGE_OFFSET_START ((0xFFFFFFFFUL >> PAGE_SHIFT) + 1) +#define DRM_FILE_PAGE_OFFSET_SIZE ((0xFFFFFFFFUL >> PAGE_SHIFT) * 16) +#else +#define DRM_FILE_PAGE_OFFSET_START ((0xFFFFFFFUL >> PAGE_SHIFT) + 1) +#define DRM_FILE_PAGE_OFFSET_SIZE ((0xFFFFFFFUL >> PAGE_SHIFT) * 16) +#endif + +/** + * Initialize the GEM device fields + */ + +int +drm_gem_init(struct drm_device *dev) +{ + struct drm_gem_mm *mm; + + spin_lock_init(&dev->object_name_lock); + idr_init(&dev->object_name_idr); + + mm = kzalloc(sizeof(struct drm_gem_mm), GFP_KERNEL); + if (!mm) { + DRM_ERROR("out of memory\n"); + return -ENOMEM; + } + + dev->mm_private = mm; + + if (drm_ht_create(&mm->offset_hash, 12)) { + kfree(mm); + return -ENOMEM; + } + + if (drm_mm_init(&mm->offset_manager, DRM_FILE_PAGE_OFFSET_START, + DRM_FILE_PAGE_OFFSET_SIZE)) { + drm_ht_remove(&mm->offset_hash); + kfree(mm); + return -ENOMEM; + } + + return 0; +} + +void +drm_gem_destroy(struct drm_device *dev) +{ + struct drm_gem_mm *mm = dev->mm_private; + + drm_mm_takedown(&mm->offset_manager); + drm_ht_remove(&mm->offset_hash); + kfree(mm); + dev->mm_private = NULL; +} + +/** + * Initialize an already allocate GEM object of the specified size with + * shmfs backing store. + */ +int drm_gem_object_init(struct drm_device *dev, + struct drm_gem_object *obj, size_t size) +{ + BUG_ON((size & (PAGE_SIZE - 1)) != 0); + + obj->dev = dev; + obj->filp = shmem_file_setup("drm mm object", size, VM_NORESERVE); + if (IS_ERR(obj->filp)) + return -ENOMEM; + + kref_init(&obj->refcount); + atomic_set(&obj->handle_count, 0); + obj->size = size; + + return 0; +} +EXPORT_SYMBOL(drm_gem_object_init); + +/** + * Allocate a GEM object of the specified size with shmfs backing store + */ +struct drm_gem_object * +drm_gem_object_alloc(struct drm_device *dev, size_t size) +{ + struct drm_gem_object *obj; + + obj = kzalloc(sizeof(*obj), GFP_KERNEL); + if (!obj) + goto free; + + if (drm_gem_object_init(dev, obj, size) != 0) + goto free; + + if (dev->driver->gem_init_object != NULL && + dev->driver->gem_init_object(obj) != 0) { + goto fput; + } + return obj; +fput: + /* Object_init mangles the global counters - readjust them. */ + fput(obj->filp); +free: + kfree(obj); + return NULL; +} +EXPORT_SYMBOL(drm_gem_object_alloc); + +/** + * Removes the mapping from handle to filp for this object. + */ +int +drm_gem_handle_delete(struct drm_file *filp, u32 handle) +{ + struct drm_device *dev; + struct drm_gem_object *obj; + + /* This is gross. The idr system doesn't let us try a delete and + * return an error code. It just spews if you fail at deleting. + * So, we have to grab a lock around finding the object and then + * doing the delete on it and dropping the refcount, or the user + * could race us to double-decrement the refcount and cause a + * use-after-free later. Given the frequency of our handle lookups, + * we may want to use ida for number allocation and a hash table + * for the pointers, anyway. + */ + spin_lock(&filp->table_lock); + + /* Check if we currently have a reference on the object */ + obj = idr_find(&filp->object_idr, handle); + if (obj == NULL) { + spin_unlock(&filp->table_lock); + return -EINVAL; + } + dev = obj->dev; + + /* Release reference and decrement refcount. */ + idr_remove(&filp->object_idr, handle); + spin_unlock(&filp->table_lock); + + drm_gem_object_handle_unreference_unlocked(obj); + + return 0; +} +EXPORT_SYMBOL(drm_gem_handle_delete); + +/** + * Create a handle for this object. This adds a handle reference + * to the object, which includes a regular reference count. Callers + * will likely want to dereference the object afterwards. + */ +int +drm_gem_handle_create(struct drm_file *file_priv, + struct drm_gem_object *obj, + u32 *handlep) +{ + int ret; + + /* + * Get the user-visible handle using idr. + */ +again: + /* ensure there is space available to allocate a handle */ + if (idr_pre_get(&file_priv->object_idr, GFP_KERNEL) == 0) + return -ENOMEM; + + /* do the allocation under our spinlock */ + spin_lock(&file_priv->table_lock); + ret = idr_get_new_above(&file_priv->object_idr, obj, 1, (int *)handlep); + spin_unlock(&file_priv->table_lock); + if (ret == -EAGAIN) + goto again; + + if (ret != 0) + return ret; + + drm_gem_object_handle_reference(obj); + return 0; +} +EXPORT_SYMBOL(drm_gem_handle_create); + +/** Returns a reference to the object named by the handle. */ +struct drm_gem_object * +drm_gem_object_lookup(struct drm_device *dev, struct drm_file *filp, + u32 handle) +{ + struct drm_gem_object *obj; + + spin_lock(&filp->table_lock); + + /* Check if we currently have a reference on the object */ + obj = idr_find(&filp->object_idr, handle); + if (obj == NULL) { + spin_unlock(&filp->table_lock); + return NULL; + } + + drm_gem_object_reference(obj); + + spin_unlock(&filp->table_lock); + + return obj; +} +EXPORT_SYMBOL(drm_gem_object_lookup); + +/** + * Releases the handle to an mm object. + */ +int +drm_gem_close_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_gem_close *args = data; + int ret; + + if (!(dev->driver->driver_features & DRIVER_GEM)) + return -ENODEV; + + ret = drm_gem_handle_delete(file_priv, args->handle); + + return ret; +} + +/** + * Create a global name for an object, returning the name. + * + * Note that the name does not hold a reference; when the object + * is freed, the name goes away. + */ +int +drm_gem_flink_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_gem_flink *args = data; + struct drm_gem_object *obj; + int ret; + + if (!(dev->driver->driver_features & DRIVER_GEM)) + return -ENODEV; + + obj = drm_gem_object_lookup(dev, file_priv, args->handle); + if (obj == NULL) + return -ENOENT; + +again: + if (idr_pre_get(&dev->object_name_idr, GFP_KERNEL) == 0) { + ret = -ENOMEM; + goto err; + } + + spin_lock(&dev->object_name_lock); + if (!obj->name) { + ret = idr_get_new_above(&dev->object_name_idr, obj, 1, + &obj->name); + args->name = (uint64_t) obj->name; + spin_unlock(&dev->object_name_lock); + + if (ret == -EAGAIN) + goto again; + + if (ret != 0) + goto err; + + /* Allocate a reference for the name table. */ + drm_gem_object_reference(obj); + } else { + args->name = (uint64_t) obj->name; + spin_unlock(&dev->object_name_lock); + ret = 0; + } + +err: + drm_gem_object_unreference_unlocked(obj); + return ret; +} + +/** + * Open an object using the global name, returning a handle and the size. + * + * This handle (of course) holds a reference to the object, so the object + * will not go away until the handle is deleted. + */ +int +drm_gem_open_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_gem_open *args = data; + struct drm_gem_object *obj; + int ret; + u32 handle; + + if (!(dev->driver->driver_features & DRIVER_GEM)) + return -ENODEV; + + spin_lock(&dev->object_name_lock); + obj = idr_find(&dev->object_name_idr, (int) args->name); + if (obj) + drm_gem_object_reference(obj); + spin_unlock(&dev->object_name_lock); + if (!obj) + return -ENOENT; + + ret = drm_gem_handle_create(file_priv, obj, &handle); + drm_gem_object_unreference_unlocked(obj); + if (ret) + return ret; + + args->handle = handle; + args->size = obj->size; + + return 0; +} + +/** + * Called at device open time, sets up the structure for handling refcounting + * of mm objects. + */ +void +drm_gem_open(struct drm_device *dev, struct drm_file *file_private) +{ + idr_init(&file_private->object_idr); + spin_lock_init(&file_private->table_lock); +} + +/** + * Called at device close to release the file's + * handle references on objects. + */ +static int +drm_gem_object_release_handle(int id, void *ptr, void *data) +{ + struct drm_gem_object *obj = ptr; + + drm_gem_object_handle_unreference_unlocked(obj); + + return 0; +} + +/** + * Called at close time when the filp is going away. + * + * Releases any remaining references on objects by this filp. + */ +void +drm_gem_release(struct drm_device *dev, struct drm_file *file_private) +{ + idr_for_each(&file_private->object_idr, + &drm_gem_object_release_handle, NULL); + + idr_remove_all(&file_private->object_idr); + idr_destroy(&file_private->object_idr); +} + +void +drm_gem_object_release(struct drm_gem_object *obj) +{ + fput(obj->filp); +} +EXPORT_SYMBOL(drm_gem_object_release); + +/** + * Called after the last reference to the object has been lost. + * Must be called holding struct_ mutex + * + * Frees the object + */ +void +drm_gem_object_free(struct kref *kref) +{ + struct drm_gem_object *obj = (struct drm_gem_object *) kref; + struct drm_device *dev = obj->dev; + + BUG_ON(!mutex_is_locked(&dev->struct_mutex)); + + if (dev->driver->gem_free_object != NULL) + dev->driver->gem_free_object(obj); +} +EXPORT_SYMBOL(drm_gem_object_free); + +static void drm_gem_object_ref_bug(struct kref *list_kref) +{ + BUG(); +} + +/** + * Called after the last handle to the object has been closed + * + * Removes any name for the object. Note that this must be + * called before drm_gem_object_free or we'll be touching + * freed memory + */ +void drm_gem_object_handle_free(struct drm_gem_object *obj) +{ + struct drm_device *dev = obj->dev; + + /* Remove any name for this object */ + spin_lock(&dev->object_name_lock); + if (obj->name) { + idr_remove(&dev->object_name_idr, obj->name); + obj->name = 0; + spin_unlock(&dev->object_name_lock); + /* + * The object name held a reference to this object, drop + * that now. + * + * This cannot be the last reference, since the handle holds one too. + */ + kref_put(&obj->refcount, drm_gem_object_ref_bug); + } else + spin_unlock(&dev->object_name_lock); + +} +EXPORT_SYMBOL(drm_gem_object_handle_free); + +void drm_gem_vm_open(struct vm_area_struct *vma) +{ + struct drm_gem_object *obj = vma->vm_private_data; + + drm_gem_object_reference(obj); + + mutex_lock(&obj->dev->struct_mutex); + drm_vm_open_locked(vma); + mutex_unlock(&obj->dev->struct_mutex); +} +EXPORT_SYMBOL(drm_gem_vm_open); + +void drm_gem_vm_close(struct vm_area_struct *vma) +{ + struct drm_gem_object *obj = vma->vm_private_data; + struct drm_device *dev = obj->dev; + + mutex_lock(&dev->struct_mutex); + drm_vm_close_locked(vma); + drm_gem_object_unreference(obj); + mutex_unlock(&dev->struct_mutex); +} +EXPORT_SYMBOL(drm_gem_vm_close); + + +/** + * drm_gem_mmap - memory map routine for GEM objects + * @filp: DRM file pointer + * @vma: VMA for the area to be mapped + * + * If a driver supports GEM object mapping, mmap calls on the DRM file + * descriptor will end up here. + * + * If we find the object based on the offset passed in (vma->vm_pgoff will + * contain the fake offset we created when the GTT map ioctl was called on + * the object), we set up the driver fault handler so that any accesses + * to the object can be trapped, to perform migration, GTT binding, surface + * register allocation, or performance monitoring. + */ +int drm_gem_mmap(struct file *filp, struct vm_area_struct *vma) +{ + struct drm_file *priv = filp->private_data; + struct drm_device *dev = priv->minor->dev; + struct drm_gem_mm *mm = dev->mm_private; + struct drm_local_map *map = NULL; + struct drm_gem_object *obj; + struct drm_hash_item *hash; + int ret = 0; + + mutex_lock(&dev->struct_mutex); + + if (drm_ht_find_item(&mm->offset_hash, vma->vm_pgoff, &hash)) { + mutex_unlock(&dev->struct_mutex); + return drm_mmap(filp, vma); + } + + map = drm_hash_entry(hash, struct drm_map_list, hash)->map; + if (!map || + ((map->flags & _DRM_RESTRICTED) && !capable(CAP_SYS_ADMIN))) { + ret = -EPERM; + goto out_unlock; + } + + /* Check for valid size. */ + if (map->size < vma->vm_end - vma->vm_start) { + ret = -EINVAL; + goto out_unlock; + } + + obj = map->handle; + if (!obj->dev->driver->gem_vm_ops) { + ret = -EINVAL; + goto out_unlock; + } + + vma->vm_flags |= VM_RESERVED | VM_IO | VM_PFNMAP | VM_DONTEXPAND; + vma->vm_ops = obj->dev->driver->gem_vm_ops; + vma->vm_private_data = map->handle; + vma->vm_page_prot = pgprot_writecombine(vm_get_page_prot(vma->vm_flags)); + + /* Take a ref for this mapping of the object, so that the fault + * handler can dereference the mmap offset's pointer to the object. + * This reference is cleaned up by the corresponding vm_close + * (which should happen whether the vma was created by this call, or + * by a vm_open due to mremap or partial unmap or whatever). + */ + drm_gem_object_reference(obj); + + vma->vm_file = filp; /* Needed for drm_vm_open() */ + drm_vm_open_locked(vma); + +out_unlock: + mutex_unlock(&dev->struct_mutex); + + return ret; +} +EXPORT_SYMBOL(drm_gem_mmap); diff --git a/drivers/gpu/drm/drm_global.c b/drivers/gpu/drm/drm_global.c new file mode 100644 index 00000000..c87dc964 --- /dev/null +++ b/drivers/gpu/drm/drm_global.c @@ -0,0 +1,112 @@ +/************************************************************************** + * + * Copyright 2008-2009 VMware, Inc., Palo Alto, CA., USA + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ +/* + * Authors: Thomas Hellstrom + */ + +#include +#include +#include +#include "drm_global.h" + +struct drm_global_item { + struct mutex mutex; + void *object; + int refcount; +}; + +static struct drm_global_item glob[DRM_GLOBAL_NUM]; + +void drm_global_init(void) +{ + int i; + + for (i = 0; i < DRM_GLOBAL_NUM; ++i) { + struct drm_global_item *item = &glob[i]; + mutex_init(&item->mutex); + item->object = NULL; + item->refcount = 0; + } +} + +void drm_global_release(void) +{ + int i; + for (i = 0; i < DRM_GLOBAL_NUM; ++i) { + struct drm_global_item *item = &glob[i]; + BUG_ON(item->object != NULL); + BUG_ON(item->refcount != 0); + } +} + +int drm_global_item_ref(struct drm_global_reference *ref) +{ + int ret; + struct drm_global_item *item = &glob[ref->global_type]; + void *object; + + mutex_lock(&item->mutex); + if (item->refcount == 0) { + item->object = kzalloc(ref->size, GFP_KERNEL); + if (unlikely(item->object == NULL)) { + ret = -ENOMEM; + goto out_err; + } + + ref->object = item->object; + ret = ref->init(ref); + if (unlikely(ret != 0)) + goto out_err; + + } + ++item->refcount; + ref->object = item->object; + object = item->object; + mutex_unlock(&item->mutex); + return 0; +out_err: + mutex_unlock(&item->mutex); + item->object = NULL; + return ret; +} +EXPORT_SYMBOL(drm_global_item_ref); + +void drm_global_item_unref(struct drm_global_reference *ref) +{ + struct drm_global_item *item = &glob[ref->global_type]; + + mutex_lock(&item->mutex); + BUG_ON(item->refcount == 0); + BUG_ON(ref->object != item->object); + if (--item->refcount == 0) { + ref->release(ref); + item->object = NULL; + } + mutex_unlock(&item->mutex); +} +EXPORT_SYMBOL(drm_global_item_unref); + diff --git a/drivers/gpu/drm/drm_hashtab.c b/drivers/gpu/drm/drm_hashtab.c new file mode 100644 index 00000000..e3a75688 --- /dev/null +++ b/drivers/gpu/drm/drm_hashtab.c @@ -0,0 +1,196 @@ +/************************************************************************** + * + * Copyright 2006 Tungsten Graphics, Inc., Bismarck, ND. USA. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. + * + * + **************************************************************************/ +/* + * Simple open hash tab implementation. + * + * Authors: + * Thomas Hellström + */ + +#include "drmP.h" +#include "drm_hashtab.h" +#include +#include + +int drm_ht_create(struct drm_open_hash *ht, unsigned int order) +{ + unsigned int size = 1 << order; + + ht->order = order; + ht->table = NULL; + if (size <= PAGE_SIZE / sizeof(*ht->table)) + ht->table = kcalloc(size, sizeof(*ht->table), GFP_KERNEL); + else + ht->table = vzalloc(size*sizeof(*ht->table)); + if (!ht->table) { + DRM_ERROR("Out of memory for hash table\n"); + return -ENOMEM; + } + return 0; +} +EXPORT_SYMBOL(drm_ht_create); + +void drm_ht_verbose_list(struct drm_open_hash *ht, unsigned long key) +{ + struct drm_hash_item *entry; + struct hlist_head *h_list; + struct hlist_node *list; + unsigned int hashed_key; + int count = 0; + + hashed_key = hash_long(key, ht->order); + DRM_DEBUG("Key is 0x%08lx, Hashed key is 0x%08x\n", key, hashed_key); + h_list = &ht->table[hashed_key]; + hlist_for_each(list, h_list) { + entry = hlist_entry(list, struct drm_hash_item, head); + DRM_DEBUG("count %d, key: 0x%08lx\n", count++, entry->key); + } +} + +static struct hlist_node *drm_ht_find_key(struct drm_open_hash *ht, + unsigned long key) +{ + struct drm_hash_item *entry; + struct hlist_head *h_list; + struct hlist_node *list; + unsigned int hashed_key; + + hashed_key = hash_long(key, ht->order); + h_list = &ht->table[hashed_key]; + hlist_for_each(list, h_list) { + entry = hlist_entry(list, struct drm_hash_item, head); + if (entry->key == key) + return list; + if (entry->key > key) + break; + } + return NULL; +} + + +int drm_ht_insert_item(struct drm_open_hash *ht, struct drm_hash_item *item) +{ + struct drm_hash_item *entry; + struct hlist_head *h_list; + struct hlist_node *list, *parent; + unsigned int hashed_key; + unsigned long key = item->key; + + hashed_key = hash_long(key, ht->order); + h_list = &ht->table[hashed_key]; + parent = NULL; + hlist_for_each(list, h_list) { + entry = hlist_entry(list, struct drm_hash_item, head); + if (entry->key == key) + return -EINVAL; + if (entry->key > key) + break; + parent = list; + } + if (parent) { + hlist_add_after(parent, &item->head); + } else { + hlist_add_head(&item->head, h_list); + } + return 0; +} +EXPORT_SYMBOL(drm_ht_insert_item); + +/* + * Just insert an item and return any "bits" bit key that hasn't been + * used before. + */ +int drm_ht_just_insert_please(struct drm_open_hash *ht, struct drm_hash_item *item, + unsigned long seed, int bits, int shift, + unsigned long add) +{ + int ret; + unsigned long mask = (1 << bits) - 1; + unsigned long first, unshifted_key; + + unshifted_key = hash_long(seed, bits); + first = unshifted_key; + do { + item->key = (unshifted_key << shift) + add; + ret = drm_ht_insert_item(ht, item); + if (ret) + unshifted_key = (unshifted_key + 1) & mask; + } while(ret && (unshifted_key != first)); + + if (ret) { + DRM_ERROR("Available key bit space exhausted\n"); + return -EINVAL; + } + return 0; +} +EXPORT_SYMBOL(drm_ht_just_insert_please); + +int drm_ht_find_item(struct drm_open_hash *ht, unsigned long key, + struct drm_hash_item **item) +{ + struct hlist_node *list; + + list = drm_ht_find_key(ht, key); + if (!list) + return -EINVAL; + + *item = hlist_entry(list, struct drm_hash_item, head); + return 0; +} +EXPORT_SYMBOL(drm_ht_find_item); + +int drm_ht_remove_key(struct drm_open_hash *ht, unsigned long key) +{ + struct hlist_node *list; + + list = drm_ht_find_key(ht, key); + if (list) { + hlist_del_init(list); + return 0; + } + return -EINVAL; +} + +int drm_ht_remove_item(struct drm_open_hash *ht, struct drm_hash_item *item) +{ + hlist_del_init(&item->head); + return 0; +} +EXPORT_SYMBOL(drm_ht_remove_item); + +void drm_ht_remove(struct drm_open_hash *ht) +{ + if (ht->table) { + if ((PAGE_SIZE / sizeof(*ht->table)) >> ht->order) + kfree(ht->table); + else + vfree(ht->table); + ht->table = NULL; + } +} +EXPORT_SYMBOL(drm_ht_remove); diff --git a/drivers/gpu/drm/drm_info.c b/drivers/gpu/drm/drm_info.c new file mode 100644 index 00000000..ab1162da --- /dev/null +++ b/drivers/gpu/drm/drm_info.c @@ -0,0 +1,315 @@ +/** + * \file drm_info.c + * DRM info file implementations + * + * \author Ben Gamari + */ + +/* + * Created: Sun Dec 21 13:09:50 2008 by bgamari@gmail.com + * + * Copyright 1999 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California. + * Copyright 2008 Ben Gamari + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * VA LINUX SYSTEMS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#include +#include "drmP.h" + +/** + * Called when "/proc/dri/.../name" is read. + * + * Prints the device name together with the bus id if available. + */ +int drm_name_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_minor *minor = node->minor; + struct drm_device *dev = minor->dev; + struct drm_master *master = minor->master; + const char *bus_name; + if (!master) + return 0; + + bus_name = dev->driver->bus->get_name(dev); + if (master->unique) { + seq_printf(m, "%s %s %s\n", + bus_name, + dev_name(dev->dev), master->unique); + } else { + seq_printf(m, "%s %s\n", + bus_name, dev_name(dev->dev)); + } + return 0; +} + +/** + * Called when "/proc/dri/.../vm" is read. + * + * Prints information about all mappings in drm_device::maplist. + */ +int drm_vm_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + struct drm_local_map *map; + struct drm_map_list *r_list; + + /* Hardcoded from _DRM_FRAME_BUFFER, + _DRM_REGISTERS, _DRM_SHM, _DRM_AGP, and + _DRM_SCATTER_GATHER and _DRM_CONSISTENT */ + const char *types[] = { "FB", "REG", "SHM", "AGP", "SG", "PCI" }; + const char *type; + int i; + + mutex_lock(&dev->struct_mutex); + seq_printf(m, "slot offset size type flags address mtrr\n\n"); + i = 0; + list_for_each_entry(r_list, &dev->maplist, head) { + map = r_list->map; + if (!map) + continue; + if (map->type < 0 || map->type > 5) + type = "??"; + else + type = types[map->type]; + + seq_printf(m, "%4d 0x%016llx 0x%08lx %4.4s 0x%02x 0x%08lx ", + i, + (unsigned long long)map->offset, + map->size, type, map->flags, + (unsigned long) r_list->user_token); + if (map->mtrr < 0) + seq_printf(m, "none\n"); + else + seq_printf(m, "%4d\n", map->mtrr); + i++; + } + mutex_unlock(&dev->struct_mutex); + return 0; +} + +/** + * Called when "/proc/dri/.../queues" is read. + */ +int drm_queues_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + int i; + struct drm_queue *q; + + mutex_lock(&dev->struct_mutex); + seq_printf(m, " ctx/flags use fin" + " blk/rw/rwf wait flushed queued" + " locks\n\n"); + for (i = 0; i < dev->queue_count; i++) { + q = dev->queuelist[i]; + atomic_inc(&q->use_count); + seq_printf(m, "%5d/0x%03x %5d %5d" + " %5d/%c%c/%c%c%c %5Zd\n", + i, + q->flags, + atomic_read(&q->use_count), + atomic_read(&q->finalization), + atomic_read(&q->block_count), + atomic_read(&q->block_read) ? 'r' : '-', + atomic_read(&q->block_write) ? 'w' : '-', + waitqueue_active(&q->read_queue) ? 'r' : '-', + waitqueue_active(&q->write_queue) ? 'w' : '-', + waitqueue_active(&q->flush_queue) ? 'f' : '-', + DRM_BUFCOUNT(&q->waitlist)); + atomic_dec(&q->use_count); + } + mutex_unlock(&dev->struct_mutex); + return 0; +} + +/** + * Called when "/proc/dri/.../bufs" is read. + */ +int drm_bufs_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + struct drm_device_dma *dma; + int i, seg_pages; + + mutex_lock(&dev->struct_mutex); + dma = dev->dma; + if (!dma) { + mutex_unlock(&dev->struct_mutex); + return 0; + } + + seq_printf(m, " o size count free segs pages kB\n\n"); + for (i = 0; i <= DRM_MAX_ORDER; i++) { + if (dma->bufs[i].buf_count) { + seg_pages = dma->bufs[i].seg_count * (1 << dma->bufs[i].page_order); + seq_printf(m, "%2d %8d %5d %5d %5d %5d %5ld\n", + i, + dma->bufs[i].buf_size, + dma->bufs[i].buf_count, + atomic_read(&dma->bufs[i].freelist.count), + dma->bufs[i].seg_count, + seg_pages, + seg_pages * PAGE_SIZE / 1024); + } + } + seq_printf(m, "\n"); + for (i = 0; i < dma->buf_count; i++) { + if (i && !(i % 32)) + seq_printf(m, "\n"); + seq_printf(m, " %d", dma->buflist[i]->list); + } + seq_printf(m, "\n"); + mutex_unlock(&dev->struct_mutex); + return 0; +} + +/** + * Called when "/proc/dri/.../vblank" is read. + */ +int drm_vblank_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + int crtc; + + mutex_lock(&dev->struct_mutex); + for (crtc = 0; crtc < dev->num_crtcs; crtc++) { + seq_printf(m, "CRTC %d enable: %d\n", + crtc, atomic_read(&dev->vblank_refcount[crtc])); + seq_printf(m, "CRTC %d counter: %d\n", + crtc, drm_vblank_count(dev, crtc)); + seq_printf(m, "CRTC %d last wait: %d\n", + crtc, dev->last_vblank_wait[crtc]); + seq_printf(m, "CRTC %d in modeset: %d\n", + crtc, dev->vblank_inmodeset[crtc]); + } + mutex_unlock(&dev->struct_mutex); + return 0; +} + +/** + * Called when "/proc/dri/.../clients" is read. + * + */ +int drm_clients_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + struct drm_file *priv; + + mutex_lock(&dev->struct_mutex); + seq_printf(m, "a dev pid uid magic ioctls\n\n"); + list_for_each_entry(priv, &dev->filelist, lhead) { + seq_printf(m, "%c %3d %5d %5d %10u %10lu\n", + priv->authenticated ? 'y' : 'n', + priv->minor->index, + priv->pid, + priv->uid, priv->magic, priv->ioctl_count); + } + mutex_unlock(&dev->struct_mutex); + return 0; +} + + +int drm_gem_one_name_info(int id, void *ptr, void *data) +{ + struct drm_gem_object *obj = ptr; + struct seq_file *m = data; + + seq_printf(m, "name %d size %zd\n", obj->name, obj->size); + + seq_printf(m, "%6d %8zd %7d %8d\n", + obj->name, obj->size, + atomic_read(&obj->handle_count), + atomic_read(&obj->refcount.refcount)); + return 0; +} + +int drm_gem_name_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + + seq_printf(m, " name size handles refcount\n"); + idr_for_each(&dev->object_name_idr, drm_gem_one_name_info, m); + return 0; +} + +#if DRM_DEBUG_CODE + +int drm_vma_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + struct drm_vma_entry *pt; + struct vm_area_struct *vma; +#if defined(__i386__) + unsigned int pgprot; +#endif + + mutex_lock(&dev->struct_mutex); + seq_printf(m, "vma use count: %d, high_memory = %pK, 0x%pK\n", + atomic_read(&dev->vma_count), + high_memory, (void *)virt_to_phys(high_memory)); + + list_for_each_entry(pt, &dev->vmalist, head) { + vma = pt->vma; + if (!vma) + continue; + seq_printf(m, + "\n%5d 0x%pK-0x%pK %c%c%c%c%c%c 0x%08lx000", + pt->pid, + (void *)vma->vm_start, (void *)vma->vm_end, + vma->vm_flags & VM_READ ? 'r' : '-', + vma->vm_flags & VM_WRITE ? 'w' : '-', + vma->vm_flags & VM_EXEC ? 'x' : '-', + vma->vm_flags & VM_MAYSHARE ? 's' : 'p', + vma->vm_flags & VM_LOCKED ? 'l' : '-', + vma->vm_flags & VM_IO ? 'i' : '-', + vma->vm_pgoff); + +#if defined(__i386__) + pgprot = pgprot_val(vma->vm_page_prot); + seq_printf(m, " %c%c%c%c%c%c%c%c%c", + pgprot & _PAGE_PRESENT ? 'p' : '-', + pgprot & _PAGE_RW ? 'w' : 'r', + pgprot & _PAGE_USER ? 'u' : 's', + pgprot & _PAGE_PWT ? 't' : 'b', + pgprot & _PAGE_PCD ? 'u' : 'c', + pgprot & _PAGE_ACCESSED ? 'a' : '-', + pgprot & _PAGE_DIRTY ? 'd' : '-', + pgprot & _PAGE_PSE ? 'm' : 'k', + pgprot & _PAGE_GLOBAL ? 'g' : 'l'); +#endif + seq_printf(m, "\n"); + } + mutex_unlock(&dev->struct_mutex); + return 0; +} + +#endif + diff --git a/drivers/gpu/drm/drm_ioc32.c b/drivers/gpu/drm/drm_ioc32.c new file mode 100644 index 00000000..4a058c7a --- /dev/null +++ b/drivers/gpu/drm/drm_ioc32.c @@ -0,0 +1,1083 @@ +/** + * \file drm_ioc32.c + * + * 32-bit ioctl compatibility routines for the DRM. + * + * \author Paul Mackerras + * + * Copyright (C) Paul Mackerras 2005. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHOR BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS + * IN THE SOFTWARE. + */ +#include +#include + +#include "drmP.h" +#include "drm_core.h" + +#define DRM_IOCTL_VERSION32 DRM_IOWR(0x00, drm_version32_t) +#define DRM_IOCTL_GET_UNIQUE32 DRM_IOWR(0x01, drm_unique32_t) +#define DRM_IOCTL_GET_MAP32 DRM_IOWR(0x04, drm_map32_t) +#define DRM_IOCTL_GET_CLIENT32 DRM_IOWR(0x05, drm_client32_t) +#define DRM_IOCTL_GET_STATS32 DRM_IOR( 0x06, drm_stats32_t) + +#define DRM_IOCTL_SET_UNIQUE32 DRM_IOW( 0x10, drm_unique32_t) +#define DRM_IOCTL_ADD_MAP32 DRM_IOWR(0x15, drm_map32_t) +#define DRM_IOCTL_ADD_BUFS32 DRM_IOWR(0x16, drm_buf_desc32_t) +#define DRM_IOCTL_MARK_BUFS32 DRM_IOW( 0x17, drm_buf_desc32_t) +#define DRM_IOCTL_INFO_BUFS32 DRM_IOWR(0x18, drm_buf_info32_t) +#define DRM_IOCTL_MAP_BUFS32 DRM_IOWR(0x19, drm_buf_map32_t) +#define DRM_IOCTL_FREE_BUFS32 DRM_IOW( 0x1a, drm_buf_free32_t) + +#define DRM_IOCTL_RM_MAP32 DRM_IOW( 0x1b, drm_map32_t) + +#define DRM_IOCTL_SET_SAREA_CTX32 DRM_IOW( 0x1c, drm_ctx_priv_map32_t) +#define DRM_IOCTL_GET_SAREA_CTX32 DRM_IOWR(0x1d, drm_ctx_priv_map32_t) + +#define DRM_IOCTL_RES_CTX32 DRM_IOWR(0x26, drm_ctx_res32_t) +#define DRM_IOCTL_DMA32 DRM_IOWR(0x29, drm_dma32_t) + +#define DRM_IOCTL_AGP_ENABLE32 DRM_IOW( 0x32, drm_agp_mode32_t) +#define DRM_IOCTL_AGP_INFO32 DRM_IOR( 0x33, drm_agp_info32_t) +#define DRM_IOCTL_AGP_ALLOC32 DRM_IOWR(0x34, drm_agp_buffer32_t) +#define DRM_IOCTL_AGP_FREE32 DRM_IOW( 0x35, drm_agp_buffer32_t) +#define DRM_IOCTL_AGP_BIND32 DRM_IOW( 0x36, drm_agp_binding32_t) +#define DRM_IOCTL_AGP_UNBIND32 DRM_IOW( 0x37, drm_agp_binding32_t) + +#define DRM_IOCTL_SG_ALLOC32 DRM_IOW( 0x38, drm_scatter_gather32_t) +#define DRM_IOCTL_SG_FREE32 DRM_IOW( 0x39, drm_scatter_gather32_t) + +#define DRM_IOCTL_UPDATE_DRAW32 DRM_IOW( 0x3f, drm_update_draw32_t) + +#define DRM_IOCTL_WAIT_VBLANK32 DRM_IOWR(0x3a, drm_wait_vblank32_t) + +typedef struct drm_version_32 { + int version_major; /**< Major version */ + int version_minor; /**< Minor version */ + int version_patchlevel; /**< Patch level */ + u32 name_len; /**< Length of name buffer */ + u32 name; /**< Name of driver */ + u32 date_len; /**< Length of date buffer */ + u32 date; /**< User-space buffer to hold date */ + u32 desc_len; /**< Length of desc buffer */ + u32 desc; /**< User-space buffer to hold desc */ +} drm_version32_t; + +static int compat_drm_version(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_version32_t v32; + struct drm_version __user *version; + int err; + + if (copy_from_user(&v32, (void __user *)arg, sizeof(v32))) + return -EFAULT; + + version = compat_alloc_user_space(sizeof(*version)); + if (!access_ok(VERIFY_WRITE, version, sizeof(*version))) + return -EFAULT; + if (__put_user(v32.name_len, &version->name_len) + || __put_user((void __user *)(unsigned long)v32.name, + &version->name) + || __put_user(v32.date_len, &version->date_len) + || __put_user((void __user *)(unsigned long)v32.date, + &version->date) + || __put_user(v32.desc_len, &version->desc_len) + || __put_user((void __user *)(unsigned long)v32.desc, + &version->desc)) + return -EFAULT; + + err = drm_ioctl(file, + DRM_IOCTL_VERSION, (unsigned long)version); + if (err) + return err; + + if (__get_user(v32.version_major, &version->version_major) + || __get_user(v32.version_minor, &version->version_minor) + || __get_user(v32.version_patchlevel, &version->version_patchlevel) + || __get_user(v32.name_len, &version->name_len) + || __get_user(v32.date_len, &version->date_len) + || __get_user(v32.desc_len, &version->desc_len)) + return -EFAULT; + + if (copy_to_user((void __user *)arg, &v32, sizeof(v32))) + return -EFAULT; + return 0; +} + +typedef struct drm_unique32 { + u32 unique_len; /**< Length of unique */ + u32 unique; /**< Unique name for driver instantiation */ +} drm_unique32_t; + +static int compat_drm_getunique(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_unique32_t uq32; + struct drm_unique __user *u; + int err; + + if (copy_from_user(&uq32, (void __user *)arg, sizeof(uq32))) + return -EFAULT; + + u = compat_alloc_user_space(sizeof(*u)); + if (!access_ok(VERIFY_WRITE, u, sizeof(*u))) + return -EFAULT; + if (__put_user(uq32.unique_len, &u->unique_len) + || __put_user((void __user *)(unsigned long)uq32.unique, + &u->unique)) + return -EFAULT; + + err = drm_ioctl(file, DRM_IOCTL_GET_UNIQUE, (unsigned long)u); + if (err) + return err; + + if (__get_user(uq32.unique_len, &u->unique_len)) + return -EFAULT; + if (copy_to_user((void __user *)arg, &uq32, sizeof(uq32))) + return -EFAULT; + return 0; +} + +static int compat_drm_setunique(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_unique32_t uq32; + struct drm_unique __user *u; + + if (copy_from_user(&uq32, (void __user *)arg, sizeof(uq32))) + return -EFAULT; + + u = compat_alloc_user_space(sizeof(*u)); + if (!access_ok(VERIFY_WRITE, u, sizeof(*u))) + return -EFAULT; + if (__put_user(uq32.unique_len, &u->unique_len) + || __put_user((void __user *)(unsigned long)uq32.unique, + &u->unique)) + return -EFAULT; + + return drm_ioctl(file, DRM_IOCTL_SET_UNIQUE, (unsigned long)u); +} + +typedef struct drm_map32 { + u32 offset; /**< Requested physical address (0 for SAREA)*/ + u32 size; /**< Requested physical size (bytes) */ + enum drm_map_type type; /**< Type of memory to map */ + enum drm_map_flags flags; /**< Flags */ + u32 handle; /**< User-space: "Handle" to pass to mmap() */ + int mtrr; /**< MTRR slot used */ +} drm_map32_t; + +static int compat_drm_getmap(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_map32_t __user *argp = (void __user *)arg; + drm_map32_t m32; + struct drm_map __user *map; + int idx, err; + void *handle; + + if (get_user(idx, &argp->offset)) + return -EFAULT; + + map = compat_alloc_user_space(sizeof(*map)); + if (!access_ok(VERIFY_WRITE, map, sizeof(*map))) + return -EFAULT; + if (__put_user(idx, &map->offset)) + return -EFAULT; + + err = drm_ioctl(file, DRM_IOCTL_GET_MAP, (unsigned long)map); + if (err) + return err; + + if (__get_user(m32.offset, &map->offset) + || __get_user(m32.size, &map->size) + || __get_user(m32.type, &map->type) + || __get_user(m32.flags, &map->flags) + || __get_user(handle, &map->handle) + || __get_user(m32.mtrr, &map->mtrr)) + return -EFAULT; + + m32.handle = (unsigned long)handle; + if (copy_to_user(argp, &m32, sizeof(m32))) + return -EFAULT; + return 0; + +} + +static int compat_drm_addmap(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_map32_t __user *argp = (void __user *)arg; + drm_map32_t m32; + struct drm_map __user *map; + int err; + void *handle; + + if (copy_from_user(&m32, argp, sizeof(m32))) + return -EFAULT; + + map = compat_alloc_user_space(sizeof(*map)); + if (!access_ok(VERIFY_WRITE, map, sizeof(*map))) + return -EFAULT; + if (__put_user(m32.offset, &map->offset) + || __put_user(m32.size, &map->size) + || __put_user(m32.type, &map->type) + || __put_user(m32.flags, &map->flags)) + return -EFAULT; + + err = drm_ioctl(file, DRM_IOCTL_ADD_MAP, (unsigned long)map); + if (err) + return err; + + if (__get_user(m32.offset, &map->offset) + || __get_user(m32.mtrr, &map->mtrr) + || __get_user(handle, &map->handle)) + return -EFAULT; + + m32.handle = (unsigned long)handle; + if (m32.handle != (unsigned long)handle) + printk_ratelimited(KERN_ERR "compat_drm_addmap truncated handle" + " %p for type %d offset %x\n", + handle, m32.type, m32.offset); + + if (copy_to_user(argp, &m32, sizeof(m32))) + return -EFAULT; + + return 0; +} + +static int compat_drm_rmmap(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_map32_t __user *argp = (void __user *)arg; + struct drm_map __user *map; + u32 handle; + + if (get_user(handle, &argp->handle)) + return -EFAULT; + + map = compat_alloc_user_space(sizeof(*map)); + if (!access_ok(VERIFY_WRITE, map, sizeof(*map))) + return -EFAULT; + if (__put_user((void *)(unsigned long)handle, &map->handle)) + return -EFAULT; + + return drm_ioctl(file, DRM_IOCTL_RM_MAP, (unsigned long)map); +} + +typedef struct drm_client32 { + int idx; /**< Which client desired? */ + int auth; /**< Is client authenticated? */ + u32 pid; /**< Process ID */ + u32 uid; /**< User ID */ + u32 magic; /**< Magic */ + u32 iocs; /**< Ioctl count */ +} drm_client32_t; + +static int compat_drm_getclient(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_client32_t c32; + drm_client32_t __user *argp = (void __user *)arg; + struct drm_client __user *client; + int idx, err; + + if (get_user(idx, &argp->idx)) + return -EFAULT; + + client = compat_alloc_user_space(sizeof(*client)); + if (!access_ok(VERIFY_WRITE, client, sizeof(*client))) + return -EFAULT; + if (__put_user(idx, &client->idx)) + return -EFAULT; + + err = drm_ioctl(file, DRM_IOCTL_GET_CLIENT, (unsigned long)client); + if (err) + return err; + + if (__get_user(c32.auth, &client->auth) + || __get_user(c32.pid, &client->pid) + || __get_user(c32.uid, &client->uid) + || __get_user(c32.magic, &client->magic) + || __get_user(c32.iocs, &client->iocs)) + return -EFAULT; + + if (copy_to_user(argp, &c32, sizeof(c32))) + return -EFAULT; + return 0; +} + +typedef struct drm_stats32 { + u32 count; + struct { + u32 value; + enum drm_stat_type type; + } data[15]; +} drm_stats32_t; + +static int compat_drm_getstats(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_stats32_t s32; + drm_stats32_t __user *argp = (void __user *)arg; + struct drm_stats __user *stats; + int i, err; + + stats = compat_alloc_user_space(sizeof(*stats)); + if (!access_ok(VERIFY_WRITE, stats, sizeof(*stats))) + return -EFAULT; + + err = drm_ioctl(file, DRM_IOCTL_GET_STATS, (unsigned long)stats); + if (err) + return err; + + if (__get_user(s32.count, &stats->count)) + return -EFAULT; + for (i = 0; i < 15; ++i) + if (__get_user(s32.data[i].value, &stats->data[i].value) + || __get_user(s32.data[i].type, &stats->data[i].type)) + return -EFAULT; + + if (copy_to_user(argp, &s32, sizeof(s32))) + return -EFAULT; + return 0; +} + +typedef struct drm_buf_desc32 { + int count; /**< Number of buffers of this size */ + int size; /**< Size in bytes */ + int low_mark; /**< Low water mark */ + int high_mark; /**< High water mark */ + int flags; + u32 agp_start; /**< Start address in the AGP aperture */ +} drm_buf_desc32_t; + +static int compat_drm_addbufs(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_buf_desc32_t __user *argp = (void __user *)arg; + struct drm_buf_desc __user *buf; + int err; + unsigned long agp_start; + + buf = compat_alloc_user_space(sizeof(*buf)); + if (!access_ok(VERIFY_WRITE, buf, sizeof(*buf)) + || !access_ok(VERIFY_WRITE, argp, sizeof(*argp))) + return -EFAULT; + + if (__copy_in_user(buf, argp, offsetof(drm_buf_desc32_t, agp_start)) + || __get_user(agp_start, &argp->agp_start) + || __put_user(agp_start, &buf->agp_start)) + return -EFAULT; + + err = drm_ioctl(file, DRM_IOCTL_ADD_BUFS, (unsigned long)buf); + if (err) + return err; + + if (__copy_in_user(argp, buf, offsetof(drm_buf_desc32_t, agp_start)) + || __get_user(agp_start, &buf->agp_start) + || __put_user(agp_start, &argp->agp_start)) + return -EFAULT; + + return 0; +} + +static int compat_drm_markbufs(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_buf_desc32_t b32; + drm_buf_desc32_t __user *argp = (void __user *)arg; + struct drm_buf_desc __user *buf; + + if (copy_from_user(&b32, argp, sizeof(b32))) + return -EFAULT; + + buf = compat_alloc_user_space(sizeof(*buf)); + if (!access_ok(VERIFY_WRITE, buf, sizeof(*buf))) + return -EFAULT; + + if (__put_user(b32.size, &buf->size) + || __put_user(b32.low_mark, &buf->low_mark) + || __put_user(b32.high_mark, &buf->high_mark)) + return -EFAULT; + + return drm_ioctl(file, DRM_IOCTL_MARK_BUFS, (unsigned long)buf); +} + +typedef struct drm_buf_info32 { + int count; /**< Entries in list */ + u32 list; +} drm_buf_info32_t; + +static int compat_drm_infobufs(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_buf_info32_t req32; + drm_buf_info32_t __user *argp = (void __user *)arg; + drm_buf_desc32_t __user *to; + struct drm_buf_info __user *request; + struct drm_buf_desc __user *list; + size_t nbytes; + int i, err; + int count, actual; + + if (copy_from_user(&req32, argp, sizeof(req32))) + return -EFAULT; + + count = req32.count; + to = (drm_buf_desc32_t __user *) (unsigned long)req32.list; + if (count < 0) + count = 0; + if (count > 0 + && !access_ok(VERIFY_WRITE, to, count * sizeof(drm_buf_desc32_t))) + return -EFAULT; + + nbytes = sizeof(*request) + count * sizeof(struct drm_buf_desc); + request = compat_alloc_user_space(nbytes); + if (!access_ok(VERIFY_WRITE, request, nbytes)) + return -EFAULT; + list = (struct drm_buf_desc *) (request + 1); + + if (__put_user(count, &request->count) + || __put_user(list, &request->list)) + return -EFAULT; + + err = drm_ioctl(file, DRM_IOCTL_INFO_BUFS, (unsigned long)request); + if (err) + return err; + + if (__get_user(actual, &request->count)) + return -EFAULT; + if (count >= actual) + for (i = 0; i < actual; ++i) + if (__copy_in_user(&to[i], &list[i], + offsetof(struct drm_buf_desc, flags))) + return -EFAULT; + + if (__put_user(actual, &argp->count)) + return -EFAULT; + + return 0; +} + +typedef struct drm_buf_pub32 { + int idx; /**< Index into the master buffer list */ + int total; /**< Buffer size */ + int used; /**< Amount of buffer in use (for DMA) */ + u32 address; /**< Address of buffer */ +} drm_buf_pub32_t; + +typedef struct drm_buf_map32 { + int count; /**< Length of the buffer list */ + u32 virtual; /**< Mmap'd area in user-virtual */ + u32 list; /**< Buffer information */ +} drm_buf_map32_t; + +static int compat_drm_mapbufs(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_buf_map32_t __user *argp = (void __user *)arg; + drm_buf_map32_t req32; + drm_buf_pub32_t __user *list32; + struct drm_buf_map __user *request; + struct drm_buf_pub __user *list; + int i, err; + int count, actual; + size_t nbytes; + void __user *addr; + + if (copy_from_user(&req32, argp, sizeof(req32))) + return -EFAULT; + count = req32.count; + list32 = (void __user *)(unsigned long)req32.list; + + if (count < 0) + return -EINVAL; + nbytes = sizeof(*request) + count * sizeof(struct drm_buf_pub); + request = compat_alloc_user_space(nbytes); + if (!access_ok(VERIFY_WRITE, request, nbytes)) + return -EFAULT; + list = (struct drm_buf_pub *) (request + 1); + + if (__put_user(count, &request->count) + || __put_user(list, &request->list)) + return -EFAULT; + + err = drm_ioctl(file, DRM_IOCTL_MAP_BUFS, (unsigned long)request); + if (err) + return err; + + if (__get_user(actual, &request->count)) + return -EFAULT; + if (count >= actual) + for (i = 0; i < actual; ++i) + if (__copy_in_user(&list32[i], &list[i], + offsetof(struct drm_buf_pub, address)) + || __get_user(addr, &list[i].address) + || __put_user((unsigned long)addr, + &list32[i].address)) + return -EFAULT; + + if (__put_user(actual, &argp->count) + || __get_user(addr, &request->virtual) + || __put_user((unsigned long)addr, &argp->virtual)) + return -EFAULT; + + return 0; +} + +typedef struct drm_buf_free32 { + int count; + u32 list; +} drm_buf_free32_t; + +static int compat_drm_freebufs(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_buf_free32_t req32; + struct drm_buf_free __user *request; + drm_buf_free32_t __user *argp = (void __user *)arg; + + if (copy_from_user(&req32, argp, sizeof(req32))) + return -EFAULT; + + request = compat_alloc_user_space(sizeof(*request)); + if (!access_ok(VERIFY_WRITE, request, sizeof(*request))) + return -EFAULT; + if (__put_user(req32.count, &request->count) + || __put_user((int __user *)(unsigned long)req32.list, + &request->list)) + return -EFAULT; + + return drm_ioctl(file, DRM_IOCTL_FREE_BUFS, (unsigned long)request); +} + +typedef struct drm_ctx_priv_map32 { + unsigned int ctx_id; /**< Context requesting private mapping */ + u32 handle; /**< Handle of map */ +} drm_ctx_priv_map32_t; + +static int compat_drm_setsareactx(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_ctx_priv_map32_t req32; + struct drm_ctx_priv_map __user *request; + drm_ctx_priv_map32_t __user *argp = (void __user *)arg; + + if (copy_from_user(&req32, argp, sizeof(req32))) + return -EFAULT; + + request = compat_alloc_user_space(sizeof(*request)); + if (!access_ok(VERIFY_WRITE, request, sizeof(*request))) + return -EFAULT; + if (__put_user(req32.ctx_id, &request->ctx_id) + || __put_user((void *)(unsigned long)req32.handle, + &request->handle)) + return -EFAULT; + + return drm_ioctl(file, DRM_IOCTL_SET_SAREA_CTX, (unsigned long)request); +} + +static int compat_drm_getsareactx(struct file *file, unsigned int cmd, + unsigned long arg) +{ + struct drm_ctx_priv_map __user *request; + drm_ctx_priv_map32_t __user *argp = (void __user *)arg; + int err; + unsigned int ctx_id; + void *handle; + + if (!access_ok(VERIFY_WRITE, argp, sizeof(*argp)) + || __get_user(ctx_id, &argp->ctx_id)) + return -EFAULT; + + request = compat_alloc_user_space(sizeof(*request)); + if (!access_ok(VERIFY_WRITE, request, sizeof(*request))) + return -EFAULT; + if (__put_user(ctx_id, &request->ctx_id)) + return -EFAULT; + + err = drm_ioctl(file, DRM_IOCTL_GET_SAREA_CTX, (unsigned long)request); + if (err) + return err; + + if (__get_user(handle, &request->handle) + || __put_user((unsigned long)handle, &argp->handle)) + return -EFAULT; + + return 0; +} + +typedef struct drm_ctx_res32 { + int count; + u32 contexts; +} drm_ctx_res32_t; + +static int compat_drm_resctx(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_ctx_res32_t __user *argp = (void __user *)arg; + drm_ctx_res32_t res32; + struct drm_ctx_res __user *res; + int err; + + if (copy_from_user(&res32, argp, sizeof(res32))) + return -EFAULT; + + res = compat_alloc_user_space(sizeof(*res)); + if (!access_ok(VERIFY_WRITE, res, sizeof(*res))) + return -EFAULT; + if (__put_user(res32.count, &res->count) + || __put_user((struct drm_ctx __user *) (unsigned long)res32.contexts, + &res->contexts)) + return -EFAULT; + + err = drm_ioctl(file, DRM_IOCTL_RES_CTX, (unsigned long)res); + if (err) + return err; + + if (__get_user(res32.count, &res->count) + || __put_user(res32.count, &argp->count)) + return -EFAULT; + + return 0; +} + +typedef struct drm_dma32 { + int context; /**< Context handle */ + int send_count; /**< Number of buffers to send */ + u32 send_indices; /**< List of handles to buffers */ + u32 send_sizes; /**< Lengths of data to send */ + enum drm_dma_flags flags; /**< Flags */ + int request_count; /**< Number of buffers requested */ + int request_size; /**< Desired size for buffers */ + u32 request_indices; /**< Buffer information */ + u32 request_sizes; + int granted_count; /**< Number of buffers granted */ +} drm_dma32_t; + +static int compat_drm_dma(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_dma32_t d32; + drm_dma32_t __user *argp = (void __user *)arg; + struct drm_dma __user *d; + int err; + + if (copy_from_user(&d32, argp, sizeof(d32))) + return -EFAULT; + + d = compat_alloc_user_space(sizeof(*d)); + if (!access_ok(VERIFY_WRITE, d, sizeof(*d))) + return -EFAULT; + + if (__put_user(d32.context, &d->context) + || __put_user(d32.send_count, &d->send_count) + || __put_user((int __user *)(unsigned long)d32.send_indices, + &d->send_indices) + || __put_user((int __user *)(unsigned long)d32.send_sizes, + &d->send_sizes) + || __put_user(d32.flags, &d->flags) + || __put_user(d32.request_count, &d->request_count) + || __put_user((int __user *)(unsigned long)d32.request_indices, + &d->request_indices) + || __put_user((int __user *)(unsigned long)d32.request_sizes, + &d->request_sizes)) + return -EFAULT; + + err = drm_ioctl(file, DRM_IOCTL_DMA, (unsigned long)d); + if (err) + return err; + + if (__get_user(d32.request_size, &d->request_size) + || __get_user(d32.granted_count, &d->granted_count) + || __put_user(d32.request_size, &argp->request_size) + || __put_user(d32.granted_count, &argp->granted_count)) + return -EFAULT; + + return 0; +} + +#if __OS_HAS_AGP +typedef struct drm_agp_mode32 { + u32 mode; /**< AGP mode */ +} drm_agp_mode32_t; + +static int compat_drm_agp_enable(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_agp_mode32_t __user *argp = (void __user *)arg; + drm_agp_mode32_t m32; + struct drm_agp_mode __user *mode; + + if (get_user(m32.mode, &argp->mode)) + return -EFAULT; + + mode = compat_alloc_user_space(sizeof(*mode)); + if (put_user(m32.mode, &mode->mode)) + return -EFAULT; + + return drm_ioctl(file, DRM_IOCTL_AGP_ENABLE, (unsigned long)mode); +} + +typedef struct drm_agp_info32 { + int agp_version_major; + int agp_version_minor; + u32 mode; + u32 aperture_base; /* physical address */ + u32 aperture_size; /* bytes */ + u32 memory_allowed; /* bytes */ + u32 memory_used; + + /* PCI information */ + unsigned short id_vendor; + unsigned short id_device; +} drm_agp_info32_t; + +static int compat_drm_agp_info(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_agp_info32_t __user *argp = (void __user *)arg; + drm_agp_info32_t i32; + struct drm_agp_info __user *info; + int err; + + info = compat_alloc_user_space(sizeof(*info)); + if (!access_ok(VERIFY_WRITE, info, sizeof(*info))) + return -EFAULT; + + err = drm_ioctl(file, DRM_IOCTL_AGP_INFO, (unsigned long)info); + if (err) + return err; + + if (__get_user(i32.agp_version_major, &info->agp_version_major) + || __get_user(i32.agp_version_minor, &info->agp_version_minor) + || __get_user(i32.mode, &info->mode) + || __get_user(i32.aperture_base, &info->aperture_base) + || __get_user(i32.aperture_size, &info->aperture_size) + || __get_user(i32.memory_allowed, &info->memory_allowed) + || __get_user(i32.memory_used, &info->memory_used) + || __get_user(i32.id_vendor, &info->id_vendor) + || __get_user(i32.id_device, &info->id_device)) + return -EFAULT; + + if (copy_to_user(argp, &i32, sizeof(i32))) + return -EFAULT; + + return 0; +} + +typedef struct drm_agp_buffer32 { + u32 size; /**< In bytes -- will round to page boundary */ + u32 handle; /**< Used for binding / unbinding */ + u32 type; /**< Type of memory to allocate */ + u32 physical; /**< Physical used by i810 */ +} drm_agp_buffer32_t; + +static int compat_drm_agp_alloc(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_agp_buffer32_t __user *argp = (void __user *)arg; + drm_agp_buffer32_t req32; + struct drm_agp_buffer __user *request; + int err; + + if (copy_from_user(&req32, argp, sizeof(req32))) + return -EFAULT; + + request = compat_alloc_user_space(sizeof(*request)); + if (!access_ok(VERIFY_WRITE, request, sizeof(*request)) + || __put_user(req32.size, &request->size) + || __put_user(req32.type, &request->type)) + return -EFAULT; + + err = drm_ioctl(file, DRM_IOCTL_AGP_ALLOC, (unsigned long)request); + if (err) + return err; + + if (__get_user(req32.handle, &request->handle) + || __get_user(req32.physical, &request->physical) + || copy_to_user(argp, &req32, sizeof(req32))) { + drm_ioctl(file, DRM_IOCTL_AGP_FREE, (unsigned long)request); + return -EFAULT; + } + + return 0; +} + +static int compat_drm_agp_free(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_agp_buffer32_t __user *argp = (void __user *)arg; + struct drm_agp_buffer __user *request; + u32 handle; + + request = compat_alloc_user_space(sizeof(*request)); + if (!access_ok(VERIFY_WRITE, request, sizeof(*request)) + || get_user(handle, &argp->handle) + || __put_user(handle, &request->handle)) + return -EFAULT; + + return drm_ioctl(file, DRM_IOCTL_AGP_FREE, (unsigned long)request); +} + +typedef struct drm_agp_binding32 { + u32 handle; /**< From drm_agp_buffer */ + u32 offset; /**< In bytes -- will round to page boundary */ +} drm_agp_binding32_t; + +static int compat_drm_agp_bind(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_agp_binding32_t __user *argp = (void __user *)arg; + drm_agp_binding32_t req32; + struct drm_agp_binding __user *request; + + if (copy_from_user(&req32, argp, sizeof(req32))) + return -EFAULT; + + request = compat_alloc_user_space(sizeof(*request)); + if (!access_ok(VERIFY_WRITE, request, sizeof(*request)) + || __put_user(req32.handle, &request->handle) + || __put_user(req32.offset, &request->offset)) + return -EFAULT; + + return drm_ioctl(file, DRM_IOCTL_AGP_BIND, (unsigned long)request); +} + +static int compat_drm_agp_unbind(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_agp_binding32_t __user *argp = (void __user *)arg; + struct drm_agp_binding __user *request; + u32 handle; + + request = compat_alloc_user_space(sizeof(*request)); + if (!access_ok(VERIFY_WRITE, request, sizeof(*request)) + || get_user(handle, &argp->handle) + || __put_user(handle, &request->handle)) + return -EFAULT; + + return drm_ioctl(file, DRM_IOCTL_AGP_UNBIND, (unsigned long)request); +} +#endif /* __OS_HAS_AGP */ + +typedef struct drm_scatter_gather32 { + u32 size; /**< In bytes -- will round to page boundary */ + u32 handle; /**< Used for mapping / unmapping */ +} drm_scatter_gather32_t; + +static int compat_drm_sg_alloc(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_scatter_gather32_t __user *argp = (void __user *)arg; + struct drm_scatter_gather __user *request; + int err; + unsigned long x; + + request = compat_alloc_user_space(sizeof(*request)); + if (!access_ok(VERIFY_WRITE, request, sizeof(*request)) + || !access_ok(VERIFY_WRITE, argp, sizeof(*argp)) + || __get_user(x, &argp->size) + || __put_user(x, &request->size)) + return -EFAULT; + + err = drm_ioctl(file, DRM_IOCTL_SG_ALLOC, (unsigned long)request); + if (err) + return err; + + /* XXX not sure about the handle conversion here... */ + if (__get_user(x, &request->handle) + || __put_user(x >> PAGE_SHIFT, &argp->handle)) + return -EFAULT; + + return 0; +} + +static int compat_drm_sg_free(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_scatter_gather32_t __user *argp = (void __user *)arg; + struct drm_scatter_gather __user *request; + unsigned long x; + + request = compat_alloc_user_space(sizeof(*request)); + if (!access_ok(VERIFY_WRITE, request, sizeof(*request)) + || !access_ok(VERIFY_WRITE, argp, sizeof(*argp)) + || __get_user(x, &argp->handle) + || __put_user(x << PAGE_SHIFT, &request->handle)) + return -EFAULT; + + return drm_ioctl(file, DRM_IOCTL_SG_FREE, (unsigned long)request); +} + +#if defined(CONFIG_X86) || defined(CONFIG_IA64) +typedef struct drm_update_draw32 { + drm_drawable_t handle; + unsigned int type; + unsigned int num; + /* 64-bit version has a 32-bit pad here */ + u64 data; /**< Pointer */ +} __attribute__((packed)) drm_update_draw32_t; + +static int compat_drm_update_draw(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_update_draw32_t update32; + struct drm_update_draw __user *request; + int err; + + if (copy_from_user(&update32, (void __user *)arg, sizeof(update32))) + return -EFAULT; + + request = compat_alloc_user_space(sizeof(*request)); + if (!access_ok(VERIFY_WRITE, request, sizeof(*request)) || + __put_user(update32.handle, &request->handle) || + __put_user(update32.type, &request->type) || + __put_user(update32.num, &request->num) || + __put_user(update32.data, &request->data)) + return -EFAULT; + + err = drm_ioctl(file, DRM_IOCTL_UPDATE_DRAW, (unsigned long)request); + return err; +} +#endif + +struct drm_wait_vblank_request32 { + enum drm_vblank_seq_type type; + unsigned int sequence; + u32 signal; +}; + +struct drm_wait_vblank_reply32 { + enum drm_vblank_seq_type type; + unsigned int sequence; + s32 tval_sec; + s32 tval_usec; +}; + +typedef union drm_wait_vblank32 { + struct drm_wait_vblank_request32 request; + struct drm_wait_vblank_reply32 reply; +} drm_wait_vblank32_t; + +static int compat_drm_wait_vblank(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_wait_vblank32_t __user *argp = (void __user *)arg; + drm_wait_vblank32_t req32; + union drm_wait_vblank __user *request; + int err; + + if (copy_from_user(&req32, argp, sizeof(req32))) + return -EFAULT; + + request = compat_alloc_user_space(sizeof(*request)); + if (!access_ok(VERIFY_WRITE, request, sizeof(*request)) + || __put_user(req32.request.type, &request->request.type) + || __put_user(req32.request.sequence, &request->request.sequence) + || __put_user(req32.request.signal, &request->request.signal)) + return -EFAULT; + + err = drm_ioctl(file, DRM_IOCTL_WAIT_VBLANK, (unsigned long)request); + if (err) + return err; + + if (__get_user(req32.reply.type, &request->reply.type) + || __get_user(req32.reply.sequence, &request->reply.sequence) + || __get_user(req32.reply.tval_sec, &request->reply.tval_sec) + || __get_user(req32.reply.tval_usec, &request->reply.tval_usec)) + return -EFAULT; + + if (copy_to_user(argp, &req32, sizeof(req32))) + return -EFAULT; + + return 0; +} + +drm_ioctl_compat_t *drm_compat_ioctls[] = { + [DRM_IOCTL_NR(DRM_IOCTL_VERSION32)] = compat_drm_version, + [DRM_IOCTL_NR(DRM_IOCTL_GET_UNIQUE32)] = compat_drm_getunique, + [DRM_IOCTL_NR(DRM_IOCTL_GET_MAP32)] = compat_drm_getmap, + [DRM_IOCTL_NR(DRM_IOCTL_GET_CLIENT32)] = compat_drm_getclient, + [DRM_IOCTL_NR(DRM_IOCTL_GET_STATS32)] = compat_drm_getstats, + [DRM_IOCTL_NR(DRM_IOCTL_SET_UNIQUE32)] = compat_drm_setunique, + [DRM_IOCTL_NR(DRM_IOCTL_ADD_MAP32)] = compat_drm_addmap, + [DRM_IOCTL_NR(DRM_IOCTL_ADD_BUFS32)] = compat_drm_addbufs, + [DRM_IOCTL_NR(DRM_IOCTL_MARK_BUFS32)] = compat_drm_markbufs, + [DRM_IOCTL_NR(DRM_IOCTL_INFO_BUFS32)] = compat_drm_infobufs, + [DRM_IOCTL_NR(DRM_IOCTL_MAP_BUFS32)] = compat_drm_mapbufs, + [DRM_IOCTL_NR(DRM_IOCTL_FREE_BUFS32)] = compat_drm_freebufs, + [DRM_IOCTL_NR(DRM_IOCTL_RM_MAP32)] = compat_drm_rmmap, + [DRM_IOCTL_NR(DRM_IOCTL_SET_SAREA_CTX32)] = compat_drm_setsareactx, + [DRM_IOCTL_NR(DRM_IOCTL_GET_SAREA_CTX32)] = compat_drm_getsareactx, + [DRM_IOCTL_NR(DRM_IOCTL_RES_CTX32)] = compat_drm_resctx, + [DRM_IOCTL_NR(DRM_IOCTL_DMA32)] = compat_drm_dma, +#if __OS_HAS_AGP + [DRM_IOCTL_NR(DRM_IOCTL_AGP_ENABLE32)] = compat_drm_agp_enable, + [DRM_IOCTL_NR(DRM_IOCTL_AGP_INFO32)] = compat_drm_agp_info, + [DRM_IOCTL_NR(DRM_IOCTL_AGP_ALLOC32)] = compat_drm_agp_alloc, + [DRM_IOCTL_NR(DRM_IOCTL_AGP_FREE32)] = compat_drm_agp_free, + [DRM_IOCTL_NR(DRM_IOCTL_AGP_BIND32)] = compat_drm_agp_bind, + [DRM_IOCTL_NR(DRM_IOCTL_AGP_UNBIND32)] = compat_drm_agp_unbind, +#endif + [DRM_IOCTL_NR(DRM_IOCTL_SG_ALLOC32)] = compat_drm_sg_alloc, + [DRM_IOCTL_NR(DRM_IOCTL_SG_FREE32)] = compat_drm_sg_free, +#if defined(CONFIG_X86) || defined(CONFIG_IA64) + [DRM_IOCTL_NR(DRM_IOCTL_UPDATE_DRAW32)] = compat_drm_update_draw, +#endif + [DRM_IOCTL_NR(DRM_IOCTL_WAIT_VBLANK32)] = compat_drm_wait_vblank, +}; + +/** + * Called whenever a 32-bit process running under a 64-bit kernel + * performs an ioctl on /dev/drm. + * + * \param file_priv DRM file private. + * \param cmd command. + * \param arg user argument. + * \return zero on success or negative number on failure. + */ +long drm_compat_ioctl(struct file *filp, unsigned int cmd, unsigned long arg) +{ + unsigned int nr = DRM_IOCTL_NR(cmd); + drm_ioctl_compat_t *fn; + int ret; + + /* Assume that ioctls without an explicit compat routine will just + * work. This may not always be a good assumption, but it's better + * than always failing. + */ + if (nr >= ARRAY_SIZE(drm_compat_ioctls)) + return drm_ioctl(filp, cmd, arg); + + fn = drm_compat_ioctls[nr]; + + if (fn != NULL) + ret = (*fn) (filp, cmd, arg); + else + ret = drm_ioctl(filp, cmd, arg); + + return ret; +} + +EXPORT_SYMBOL(drm_compat_ioctl); diff --git a/drivers/gpu/drm/drm_ioctl.c b/drivers/gpu/drm/drm_ioctl.c new file mode 100644 index 00000000..904d7e9c --- /dev/null +++ b/drivers/gpu/drm/drm_ioctl.c @@ -0,0 +1,355 @@ +/** + * \file drm_ioctl.c + * IOCTL processing for DRM + * + * \author Rickard E. (Rik) Faith + * \author Gareth Hughes + */ + +/* + * Created: Fri Jan 8 09:01:26 1999 by faith@valinux.com + * + * Copyright 1999 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * VA LINUX SYSTEMS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#include "drmP.h" +#include "drm_core.h" + +#include "linux/pci.h" + +/** + * Get the bus id. + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg user argument, pointing to a drm_unique structure. + * \return zero on success or a negative number on failure. + * + * Copies the bus id from drm_device::unique into user space. + */ +int drm_getunique(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_unique *u = data; + struct drm_master *master = file_priv->master; + + if (u->unique_len >= master->unique_len) { + if (copy_to_user(u->unique, master->unique, master->unique_len)) + return -EFAULT; + } + u->unique_len = master->unique_len; + + return 0; +} + +static void +drm_unset_busid(struct drm_device *dev, + struct drm_master *master) +{ + kfree(dev->devname); + dev->devname = NULL; + + kfree(master->unique); + master->unique = NULL; + master->unique_len = 0; + master->unique_size = 0; +} + +/** + * Set the bus id. + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg user argument, pointing to a drm_unique structure. + * \return zero on success or a negative number on failure. + * + * Copies the bus id from userspace into drm_device::unique, and verifies that + * it matches the device this DRM is attached to (EINVAL otherwise). Deprecated + * in interface version 1.1 and will return EBUSY when setversion has requested + * version 1.1 or greater. + */ +int drm_setunique(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_unique *u = data; + struct drm_master *master = file_priv->master; + int ret; + + if (master->unique_len || master->unique) + return -EBUSY; + + if (!u->unique_len || u->unique_len > 1024) + return -EINVAL; + + if (!dev->driver->bus->set_unique) + return -EINVAL; + + ret = dev->driver->bus->set_unique(dev, master, u); + if (ret) + goto err; + + return 0; + +err: + drm_unset_busid(dev, master); + return ret; +} + +static int drm_set_busid(struct drm_device *dev, struct drm_file *file_priv) +{ + struct drm_master *master = file_priv->master; + int ret; + + if (master->unique != NULL) + drm_unset_busid(dev, master); + + ret = dev->driver->bus->set_busid(dev, master); + if (ret) + goto err; + return 0; +err: + drm_unset_busid(dev, master); + return ret; +} + +/** + * Get a mapping information. + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg user argument, pointing to a drm_map structure. + * + * \return zero on success or a negative number on failure. + * + * Searches for the mapping with the specified offset and copies its information + * into userspace + */ +int drm_getmap(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_map *map = data; + struct drm_map_list *r_list = NULL; + struct list_head *list; + int idx; + int i; + + idx = map->offset; + + mutex_lock(&dev->struct_mutex); + if (idx < 0) { + mutex_unlock(&dev->struct_mutex); + return -EINVAL; + } + + i = 0; + list_for_each(list, &dev->maplist) { + if (i == idx) { + r_list = list_entry(list, struct drm_map_list, head); + break; + } + i++; + } + if (!r_list || !r_list->map) { + mutex_unlock(&dev->struct_mutex); + return -EINVAL; + } + + map->offset = r_list->map->offset; + map->size = r_list->map->size; + map->type = r_list->map->type; + map->flags = r_list->map->flags; + map->handle = (void *)(unsigned long) r_list->user_token; + map->mtrr = r_list->map->mtrr; + mutex_unlock(&dev->struct_mutex); + + return 0; +} + +/** + * Get client information. + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg user argument, pointing to a drm_client structure. + * + * \return zero on success or a negative number on failure. + * + * Searches for the client with the specified index and copies its information + * into userspace + */ +int drm_getclient(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_client *client = data; + struct drm_file *pt; + int idx; + int i; + + idx = client->idx; + mutex_lock(&dev->struct_mutex); + + i = 0; + list_for_each_entry(pt, &dev->filelist, lhead) { + if (i++ >= idx) { + client->auth = pt->authenticated; + client->pid = pt->pid; + client->uid = pt->uid; + client->magic = pt->magic; + client->iocs = pt->ioctl_count; + mutex_unlock(&dev->struct_mutex); + + return 0; + } + } + mutex_unlock(&dev->struct_mutex); + + return -EINVAL; +} + +/** + * Get statistics information. + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg user argument, pointing to a drm_stats structure. + * + * \return zero on success or a negative number on failure. + */ +int drm_getstats(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_stats *stats = data; + int i; + + memset(stats, 0, sizeof(*stats)); + + mutex_lock(&dev->struct_mutex); + + for (i = 0; i < dev->counters; i++) { + if (dev->types[i] == _DRM_STAT_LOCK) + stats->data[i].value = + (file_priv->master->lock.hw_lock ? file_priv->master->lock.hw_lock->lock : 0); + else + stats->data[i].value = atomic_read(&dev->counts[i]); + stats->data[i].type = dev->types[i]; + } + + stats->count = dev->counters; + + mutex_unlock(&dev->struct_mutex); + + return 0; +} + +/** + * Get device/driver capabilities + */ +int drm_getcap(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + struct drm_get_cap *req = data; + + req->value = 0; + switch (req->capability) { + case DRM_CAP_DUMB_BUFFER: + if (dev->driver->dumb_create) + req->value = 1; + break; + case DRM_CAP_VBLANK_HIGH_CRTC: + req->value = 1; + break; + default: + return -EINVAL; + } + return 0; +} + +/** + * Setversion ioctl. + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg user argument, pointing to a drm_lock structure. + * \return zero on success or negative number on failure. + * + * Sets the requested interface version + */ +int drm_setversion(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + struct drm_set_version *sv = data; + int if_version, retcode = 0; + + if (sv->drm_di_major != -1) { + if (sv->drm_di_major != DRM_IF_MAJOR || + sv->drm_di_minor < 0 || sv->drm_di_minor > DRM_IF_MINOR) { + retcode = -EINVAL; + goto done; + } + if_version = DRM_IF_VERSION(sv->drm_di_major, + sv->drm_di_minor); + dev->if_version = max(if_version, dev->if_version); + if (sv->drm_di_minor >= 1) { + /* + * Version 1.1 includes tying of DRM to specific device + * Version 1.4 has proper PCI domain support + */ + retcode = drm_set_busid(dev, file_priv); + if (retcode) + goto done; + } + } + + if (sv->drm_dd_major != -1) { + if (sv->drm_dd_major != dev->driver->major || + sv->drm_dd_minor < 0 || sv->drm_dd_minor > + dev->driver->minor) { + retcode = -EINVAL; + goto done; + } + + if (dev->driver->set_version) + dev->driver->set_version(dev, sv); + } + +done: + sv->drm_di_major = DRM_IF_MAJOR; + sv->drm_di_minor = DRM_IF_MINOR; + sv->drm_dd_major = dev->driver->major; + sv->drm_dd_minor = dev->driver->minor; + + return retcode; +} + +/** No-op ioctl. */ +int drm_noop(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + DRM_DEBUG("\n"); + return 0; +} diff --git a/drivers/gpu/drm/drm_irq.c b/drivers/gpu/drm/drm_irq.c new file mode 100644 index 00000000..2022a5c9 --- /dev/null +++ b/drivers/gpu/drm/drm_irq.c @@ -0,0 +1,1338 @@ +/** + * \file drm_irq.c + * IRQ support + * + * \author Rickard E. (Rik) Faith + * \author Gareth Hughes + */ + +/* + * Created: Fri Mar 19 14:30:16 1999 by faith@valinux.com + * + * Copyright 1999, 2000 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * VA LINUX SYSTEMS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#include "drmP.h" +#include "drm_trace.h" + +#include /* For task queue support */ +#include + +#include + +/* Access macro for slots in vblank timestamp ringbuffer. */ +#define vblanktimestamp(dev, crtc, count) ( \ + (dev)->_vblank_time[(crtc) * DRM_VBLANKTIME_RBSIZE + \ + ((count) % DRM_VBLANKTIME_RBSIZE)]) + +/* Retry timestamp calculation up to 3 times to satisfy + * drm_timestamp_precision before giving up. + */ +#define DRM_TIMESTAMP_MAXRETRIES 3 + +/* Threshold in nanoseconds for detection of redundant + * vblank irq in drm_handle_vblank(). 1 msec should be ok. + */ +#define DRM_REDUNDANT_VBLIRQ_THRESH_NS 1000000 + +/** + * Get interrupt from bus id. + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg user argument, pointing to a drm_irq_busid structure. + * \return zero on success or a negative number on failure. + * + * Finds the PCI device with the specified bus id and gets its IRQ number. + * This IOCTL is deprecated, and will now return EINVAL for any busid not equal + * to that of the device that this DRM instance attached to. + */ +int drm_irq_by_busid(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_irq_busid *p = data; + + if (!dev->driver->bus->irq_by_busid) + return -EINVAL; + + if (!drm_core_check_feature(dev, DRIVER_HAVE_IRQ)) + return -EINVAL; + + return dev->driver->bus->irq_by_busid(dev, p); +} + +/* + * Clear vblank timestamp buffer for a crtc. + */ +static void clear_vblank_timestamps(struct drm_device *dev, int crtc) +{ + memset(&dev->_vblank_time[crtc * DRM_VBLANKTIME_RBSIZE], 0, + DRM_VBLANKTIME_RBSIZE * sizeof(struct timeval)); +} + +/* + * Disable vblank irq's on crtc, make sure that last vblank count + * of hardware and corresponding consistent software vblank counter + * are preserved, even if there are any spurious vblank irq's after + * disable. + */ +static void vblank_disable_and_save(struct drm_device *dev, int crtc) +{ + unsigned long irqflags; + u32 vblcount; + s64 diff_ns; + int vblrc; + struct timeval tvblank; + + /* Prevent vblank irq processing while disabling vblank irqs, + * so no updates of timestamps or count can happen after we've + * disabled. Needed to prevent races in case of delayed irq's. + * Disable preemption, so vblank_time_lock is held as short as + * possible, even under a kernel with PREEMPT_RT patches. + */ + preempt_disable(); + spin_lock_irqsave(&dev->vblank_time_lock, irqflags); + + dev->driver->disable_vblank(dev, crtc); + dev->vblank_enabled[crtc] = 0; + + /* No further vblank irq's will be processed after + * this point. Get current hardware vblank count and + * vblank timestamp, repeat until they are consistent. + * + * FIXME: There is still a race condition here and in + * drm_update_vblank_count() which can cause off-by-one + * reinitialization of software vblank counter. If gpu + * vblank counter doesn't increment exactly at the leading + * edge of a vblank interval, then we can lose 1 count if + * we happen to execute between start of vblank and the + * delayed gpu counter increment. + */ + do { + dev->last_vblank[crtc] = dev->driver->get_vblank_counter(dev, crtc); + vblrc = drm_get_last_vbltimestamp(dev, crtc, &tvblank, 0); + } while (dev->last_vblank[crtc] != dev->driver->get_vblank_counter(dev, crtc)); + + /* Compute time difference to stored timestamp of last vblank + * as updated by last invocation of drm_handle_vblank() in vblank irq. + */ + vblcount = atomic_read(&dev->_vblank_count[crtc]); + diff_ns = timeval_to_ns(&tvblank) - + timeval_to_ns(&vblanktimestamp(dev, crtc, vblcount)); + + /* If there is at least 1 msec difference between the last stored + * timestamp and tvblank, then we are currently executing our + * disable inside a new vblank interval, the tvblank timestamp + * corresponds to this new vblank interval and the irq handler + * for this vblank didn't run yet and won't run due to our disable. + * Therefore we need to do the job of drm_handle_vblank() and + * increment the vblank counter by one to account for this vblank. + * + * Skip this step if there isn't any high precision timestamp + * available. In that case we can't account for this and just + * hope for the best. + */ + if ((vblrc > 0) && (abs64(diff_ns) > 1000000)) { + atomic_inc(&dev->_vblank_count[crtc]); + smp_mb__after_atomic_inc(); + } + + /* Invalidate all timestamps while vblank irq's are off. */ + clear_vblank_timestamps(dev, crtc); + + spin_unlock_irqrestore(&dev->vblank_time_lock, irqflags); + preempt_enable(); +} + +static void vblank_disable_fn(unsigned long arg) +{ + struct drm_device *dev = (struct drm_device *)arg; + unsigned long irqflags; + int i; + + if (!dev->vblank_disable_allowed) + return; + + for (i = 0; i < dev->num_crtcs; i++) { + spin_lock_irqsave(&dev->vbl_lock, irqflags); + if (atomic_read(&dev->vblank_refcount[i]) == 0 && + dev->vblank_enabled[i]) { + DRM_DEBUG("disabling vblank on crtc %d\n", i); + vblank_disable_and_save(dev, i); + } + spin_unlock_irqrestore(&dev->vbl_lock, irqflags); + } +} + +void drm_vblank_cleanup(struct drm_device *dev) +{ + /* Bail if the driver didn't call drm_vblank_init() */ + if (dev->num_crtcs == 0) + return; + + del_timer(&dev->vblank_disable_timer); + + vblank_disable_fn((unsigned long)dev); + + kfree(dev->vbl_queue); + kfree(dev->_vblank_count); + kfree(dev->vblank_refcount); + kfree(dev->vblank_enabled); + kfree(dev->last_vblank); + kfree(dev->last_vblank_wait); + kfree(dev->vblank_inmodeset); + kfree(dev->_vblank_time); + + dev->num_crtcs = 0; +} +EXPORT_SYMBOL(drm_vblank_cleanup); + +int drm_vblank_init(struct drm_device *dev, int num_crtcs) +{ + int i, ret = -ENOMEM; + + setup_timer(&dev->vblank_disable_timer, vblank_disable_fn, + (unsigned long)dev); + spin_lock_init(&dev->vbl_lock); + spin_lock_init(&dev->vblank_time_lock); + + dev->num_crtcs = num_crtcs; + + dev->vbl_queue = kmalloc(sizeof(wait_queue_head_t) * num_crtcs, + GFP_KERNEL); + if (!dev->vbl_queue) + goto err; + + dev->_vblank_count = kmalloc(sizeof(atomic_t) * num_crtcs, GFP_KERNEL); + if (!dev->_vblank_count) + goto err; + + dev->vblank_refcount = kmalloc(sizeof(atomic_t) * num_crtcs, + GFP_KERNEL); + if (!dev->vblank_refcount) + goto err; + + dev->vblank_enabled = kcalloc(num_crtcs, sizeof(int), GFP_KERNEL); + if (!dev->vblank_enabled) + goto err; + + dev->last_vblank = kcalloc(num_crtcs, sizeof(u32), GFP_KERNEL); + if (!dev->last_vblank) + goto err; + + dev->last_vblank_wait = kcalloc(num_crtcs, sizeof(u32), GFP_KERNEL); + if (!dev->last_vblank_wait) + goto err; + + dev->vblank_inmodeset = kcalloc(num_crtcs, sizeof(int), GFP_KERNEL); + if (!dev->vblank_inmodeset) + goto err; + + dev->_vblank_time = kcalloc(num_crtcs * DRM_VBLANKTIME_RBSIZE, + sizeof(struct timeval), GFP_KERNEL); + if (!dev->_vblank_time) + goto err; + + DRM_INFO("Supports vblank timestamp caching Rev 1 (10.10.2010).\n"); + + /* Driver specific high-precision vblank timestamping supported? */ + if (dev->driver->get_vblank_timestamp) + DRM_INFO("Driver supports precise vblank timestamp query.\n"); + else + DRM_INFO("No driver support for vblank timestamp query.\n"); + + /* Zero per-crtc vblank stuff */ + for (i = 0; i < num_crtcs; i++) { + init_waitqueue_head(&dev->vbl_queue[i]); + atomic_set(&dev->_vblank_count[i], 0); + atomic_set(&dev->vblank_refcount[i], 0); + } + + dev->vblank_disable_allowed = 0; + return 0; + +err: + drm_vblank_cleanup(dev); + return ret; +} +EXPORT_SYMBOL(drm_vblank_init); + +static void drm_irq_vgaarb_nokms(void *cookie, bool state) +{ + struct drm_device *dev = cookie; + + if (dev->driver->vgaarb_irq) { + dev->driver->vgaarb_irq(dev, state); + return; + } + + if (!dev->irq_enabled) + return; + + if (state) + dev->driver->irq_uninstall(dev); + else { + dev->driver->irq_preinstall(dev); + dev->driver->irq_postinstall(dev); + } +} + +/** + * Install IRQ handler. + * + * \param dev DRM device. + * + * Initializes the IRQ related data. Installs the handler, calling the driver + * \c drm_driver_irq_preinstall() and \c drm_driver_irq_postinstall() functions + * before and after the installation. + */ +int drm_irq_install(struct drm_device *dev) +{ + int ret = 0; + unsigned long sh_flags = 0; + char *irqname; + + if (!drm_core_check_feature(dev, DRIVER_HAVE_IRQ)) + return -EINVAL; + + if (drm_dev_to_irq(dev) == 0) + return -EINVAL; + + mutex_lock(&dev->struct_mutex); + + /* Driver must have been initialized */ + if (!dev->dev_private) { + mutex_unlock(&dev->struct_mutex); + return -EINVAL; + } + + if (dev->irq_enabled) { + mutex_unlock(&dev->struct_mutex); + return -EBUSY; + } + dev->irq_enabled = 1; + mutex_unlock(&dev->struct_mutex); + + DRM_DEBUG("irq=%d\n", drm_dev_to_irq(dev)); + + /* Before installing handler */ + dev->driver->irq_preinstall(dev); + + /* Install handler */ + if (drm_core_check_feature(dev, DRIVER_IRQ_SHARED)) + sh_flags = IRQF_SHARED; + + if (dev->devname) + irqname = dev->devname; + else + irqname = dev->driver->name; + + ret = request_irq(drm_dev_to_irq(dev), dev->driver->irq_handler, + sh_flags, irqname, dev); + + if (ret < 0) { + mutex_lock(&dev->struct_mutex); + dev->irq_enabled = 0; + mutex_unlock(&dev->struct_mutex); + return ret; + } + + if (!drm_core_check_feature(dev, DRIVER_MODESET)) + vga_client_register(dev->pdev, (void *)dev, drm_irq_vgaarb_nokms, NULL); + + /* After installing handler */ + ret = dev->driver->irq_postinstall(dev); + if (ret < 0) { + mutex_lock(&dev->struct_mutex); + dev->irq_enabled = 0; + mutex_unlock(&dev->struct_mutex); + } + + return ret; +} +EXPORT_SYMBOL(drm_irq_install); + +/** + * Uninstall the IRQ handler. + * + * \param dev DRM device. + * + * Calls the driver's \c drm_driver_irq_uninstall() function, and stops the irq. + */ +int drm_irq_uninstall(struct drm_device *dev) +{ + unsigned long irqflags; + int irq_enabled, i; + + if (!drm_core_check_feature(dev, DRIVER_HAVE_IRQ)) + return -EINVAL; + + mutex_lock(&dev->struct_mutex); + irq_enabled = dev->irq_enabled; + dev->irq_enabled = 0; + mutex_unlock(&dev->struct_mutex); + + /* + * Wake up any waiters so they don't hang. + */ + spin_lock_irqsave(&dev->vbl_lock, irqflags); + for (i = 0; i < dev->num_crtcs; i++) { + DRM_WAKEUP(&dev->vbl_queue[i]); + dev->vblank_enabled[i] = 0; + dev->last_vblank[i] = dev->driver->get_vblank_counter(dev, i); + } + spin_unlock_irqrestore(&dev->vbl_lock, irqflags); + + if (!irq_enabled) + return -EINVAL; + + DRM_DEBUG("irq=%d\n", drm_dev_to_irq(dev)); + + if (!drm_core_check_feature(dev, DRIVER_MODESET)) + vga_client_register(dev->pdev, NULL, NULL, NULL); + + dev->driver->irq_uninstall(dev); + + free_irq(drm_dev_to_irq(dev), dev); + + return 0; +} +EXPORT_SYMBOL(drm_irq_uninstall); + +/** + * IRQ control ioctl. + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg user argument, pointing to a drm_control structure. + * \return zero on success or a negative number on failure. + * + * Calls irq_install() or irq_uninstall() according to \p arg. + */ +int drm_control(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_control *ctl = data; + + /* if we haven't irq we fallback for compatibility reasons - + * this used to be a separate function in drm_dma.h + */ + + + switch (ctl->func) { + case DRM_INST_HANDLER: + if (!drm_core_check_feature(dev, DRIVER_HAVE_IRQ)) + return 0; + if (drm_core_check_feature(dev, DRIVER_MODESET)) + return 0; + if (dev->if_version < DRM_IF_VERSION(1, 2) && + ctl->irq != drm_dev_to_irq(dev)) + return -EINVAL; + return drm_irq_install(dev); + case DRM_UNINST_HANDLER: + if (!drm_core_check_feature(dev, DRIVER_HAVE_IRQ)) + return 0; + if (drm_core_check_feature(dev, DRIVER_MODESET)) + return 0; + return drm_irq_uninstall(dev); + default: + return -EINVAL; + } +} + +/** + * drm_calc_timestamping_constants - Calculate and + * store various constants which are later needed by + * vblank and swap-completion timestamping, e.g, by + * drm_calc_vbltimestamp_from_scanoutpos(). + * They are derived from crtc's true scanout timing, + * so they take things like panel scaling or other + * adjustments into account. + * + * @crtc drm_crtc whose timestamp constants should be updated. + * + */ +void drm_calc_timestamping_constants(struct drm_crtc *crtc) +{ + s64 linedur_ns = 0, pixeldur_ns = 0, framedur_ns = 0; + u64 dotclock; + + /* Dot clock in Hz: */ + dotclock = (u64) crtc->hwmode.clock * 1000; + + /* Fields of interlaced scanout modes are only halve a frame duration. + * Double the dotclock to get halve the frame-/line-/pixelduration. + */ + if (crtc->hwmode.flags & DRM_MODE_FLAG_INTERLACE) + dotclock *= 2; + + /* Valid dotclock? */ + if (dotclock > 0) { + /* Convert scanline length in pixels and video dot clock to + * line duration, frame duration and pixel duration in + * nanoseconds: + */ + pixeldur_ns = (s64) div64_u64(1000000000, dotclock); + linedur_ns = (s64) div64_u64(((u64) crtc->hwmode.crtc_htotal * + 1000000000), dotclock); + framedur_ns = (s64) crtc->hwmode.crtc_vtotal * linedur_ns; + } else + DRM_ERROR("crtc %d: Can't calculate constants, dotclock = 0!\n", + crtc->base.id); + + crtc->pixeldur_ns = pixeldur_ns; + crtc->linedur_ns = linedur_ns; + crtc->framedur_ns = framedur_ns; + + DRM_DEBUG("crtc %d: hwmode: htotal %d, vtotal %d, vdisplay %d\n", + crtc->base.id, crtc->hwmode.crtc_htotal, + crtc->hwmode.crtc_vtotal, crtc->hwmode.crtc_vdisplay); + DRM_DEBUG("crtc %d: clock %d kHz framedur %d linedur %d, pixeldur %d\n", + crtc->base.id, (int) dotclock/1000, (int) framedur_ns, + (int) linedur_ns, (int) pixeldur_ns); +} +EXPORT_SYMBOL(drm_calc_timestamping_constants); + +/** + * drm_calc_vbltimestamp_from_scanoutpos - helper routine for kms + * drivers. Implements calculation of exact vblank timestamps from + * given drm_display_mode timings and current video scanout position + * of a crtc. This can be called from within get_vblank_timestamp() + * implementation of a kms driver to implement the actual timestamping. + * + * Should return timestamps conforming to the OML_sync_control OpenML + * extension specification. The timestamp corresponds to the end of + * the vblank interval, aka start of scanout of topmost-leftmost display + * pixel in the following video frame. + * + * Requires support for optional dev->driver->get_scanout_position() + * in kms driver, plus a bit of setup code to provide a drm_display_mode + * that corresponds to the true scanout timing. + * + * The current implementation only handles standard video modes. It + * returns as no operation if a doublescan or interlaced video mode is + * active. Higher level code is expected to handle this. + * + * @dev: DRM device. + * @crtc: Which crtc's vblank timestamp to retrieve. + * @max_error: Desired maximum allowable error in timestamps (nanosecs). + * On return contains true maximum error of timestamp. + * @vblank_time: Pointer to struct timeval which should receive the timestamp. + * @flags: Flags to pass to driver: + * 0 = Default. + * DRM_CALLED_FROM_VBLIRQ = If function is called from vbl irq handler. + * @refcrtc: drm_crtc* of crtc which defines scanout timing. + * + * Returns negative value on error, failure or if not supported in current + * video mode: + * + * -EINVAL - Invalid crtc. + * -EAGAIN - Temporary unavailable, e.g., called before initial modeset. + * -ENOTSUPP - Function not supported in current display mode. + * -EIO - Failed, e.g., due to failed scanout position query. + * + * Returns or'ed positive status flags on success: + * + * DRM_VBLANKTIME_SCANOUTPOS_METHOD - Signal this method used for timestamping. + * DRM_VBLANKTIME_INVBL - Timestamp taken while scanout was in vblank interval. + * + */ +int drm_calc_vbltimestamp_from_scanoutpos(struct drm_device *dev, int crtc, + int *max_error, + struct timeval *vblank_time, + unsigned flags, + struct drm_crtc *refcrtc) +{ + struct timeval stime, raw_time; + struct drm_display_mode *mode; + int vbl_status, vtotal, vdisplay; + int vpos, hpos, i; + s64 framedur_ns, linedur_ns, pixeldur_ns, delta_ns, duration_ns; + bool invbl; + + if (crtc < 0 || crtc >= dev->num_crtcs) { + DRM_ERROR("Invalid crtc %d\n", crtc); + return -EINVAL; + } + + /* Scanout position query not supported? Should not happen. */ + if (!dev->driver->get_scanout_position) { + DRM_ERROR("Called from driver w/o get_scanout_position()!?\n"); + return -EIO; + } + + mode = &refcrtc->hwmode; + vtotal = mode->crtc_vtotal; + vdisplay = mode->crtc_vdisplay; + + /* Durations of frames, lines, pixels in nanoseconds. */ + framedur_ns = refcrtc->framedur_ns; + linedur_ns = refcrtc->linedur_ns; + pixeldur_ns = refcrtc->pixeldur_ns; + + /* If mode timing undefined, just return as no-op: + * Happens during initial modesetting of a crtc. + */ + if (vtotal <= 0 || vdisplay <= 0 || framedur_ns == 0) { + DRM_DEBUG("crtc %d: Noop due to uninitialized mode.\n", crtc); + return -EAGAIN; + } + + /* Get current scanout position with system timestamp. + * Repeat query up to DRM_TIMESTAMP_MAXRETRIES times + * if single query takes longer than max_error nanoseconds. + * + * This guarantees a tight bound on maximum error if + * code gets preempted or delayed for some reason. + */ + for (i = 0; i < DRM_TIMESTAMP_MAXRETRIES; i++) { + /* Disable preemption to make it very likely to + * succeed in the first iteration even on PREEMPT_RT kernel. + */ + preempt_disable(); + + /* Get system timestamp before query. */ + do_gettimeofday(&stime); + + /* Get vertical and horizontal scanout pos. vpos, hpos. */ + vbl_status = dev->driver->get_scanout_position(dev, crtc, &vpos, &hpos); + + /* Get system timestamp after query. */ + do_gettimeofday(&raw_time); + + preempt_enable(); + + /* Return as no-op if scanout query unsupported or failed. */ + if (!(vbl_status & DRM_SCANOUTPOS_VALID)) { + DRM_DEBUG("crtc %d : scanoutpos query failed [%d].\n", + crtc, vbl_status); + return -EIO; + } + + duration_ns = timeval_to_ns(&raw_time) - timeval_to_ns(&stime); + + /* Accept result with < max_error nsecs timing uncertainty. */ + if (duration_ns <= (s64) *max_error) + break; + } + + /* Noisy system timing? */ + if (i == DRM_TIMESTAMP_MAXRETRIES) { + DRM_DEBUG("crtc %d: Noisy timestamp %d us > %d us [%d reps].\n", + crtc, (int) duration_ns/1000, *max_error/1000, i); + } + + /* Return upper bound of timestamp precision error. */ + *max_error = (int) duration_ns; + + /* Check if in vblank area: + * vpos is >=0 in video scanout area, but negative + * within vblank area, counting down the number of lines until + * start of scanout. + */ + invbl = vbl_status & DRM_SCANOUTPOS_INVBL; + + /* Convert scanout position into elapsed time at raw_time query + * since start of scanout at first display scanline. delta_ns + * can be negative if start of scanout hasn't happened yet. + */ + delta_ns = (s64) vpos * linedur_ns + (s64) hpos * pixeldur_ns; + + /* Is vpos outside nominal vblank area, but less than + * 1/100 of a frame height away from start of vblank? + * If so, assume this isn't a massively delayed vblank + * interrupt, but a vblank interrupt that fired a few + * microseconds before true start of vblank. Compensate + * by adding a full frame duration to the final timestamp. + * Happens, e.g., on ATI R500, R600. + * + * We only do this if DRM_CALLED_FROM_VBLIRQ. + */ + if ((flags & DRM_CALLED_FROM_VBLIRQ) && !invbl && + ((vdisplay - vpos) < vtotal / 100)) { + delta_ns = delta_ns - framedur_ns; + + /* Signal this correction as "applied". */ + vbl_status |= 0x8; + } + + /* Subtract time delta from raw timestamp to get final + * vblank_time timestamp for end of vblank. + */ + *vblank_time = ns_to_timeval(timeval_to_ns(&raw_time) - delta_ns); + + DRM_DEBUG("crtc %d : v %d p(%d,%d)@ %ld.%ld -> %ld.%ld [e %d us, %d rep]\n", + crtc, (int)vbl_status, hpos, vpos, + (long)raw_time.tv_sec, (long)raw_time.tv_usec, + (long)vblank_time->tv_sec, (long)vblank_time->tv_usec, + (int)duration_ns/1000, i); + + vbl_status = DRM_VBLANKTIME_SCANOUTPOS_METHOD; + if (invbl) + vbl_status |= DRM_VBLANKTIME_INVBL; + + return vbl_status; +} +EXPORT_SYMBOL(drm_calc_vbltimestamp_from_scanoutpos); + +/** + * drm_get_last_vbltimestamp - retrieve raw timestamp for the most recent + * vblank interval. + * + * @dev: DRM device + * @crtc: which crtc's vblank timestamp to retrieve + * @tvblank: Pointer to target struct timeval which should receive the timestamp + * @flags: Flags to pass to driver: + * 0 = Default. + * DRM_CALLED_FROM_VBLIRQ = If function is called from vbl irq handler. + * + * Fetches the system timestamp corresponding to the time of the most recent + * vblank interval on specified crtc. May call into kms-driver to + * compute the timestamp with a high-precision GPU specific method. + * + * Returns zero if timestamp originates from uncorrected do_gettimeofday() + * call, i.e., it isn't very precisely locked to the true vblank. + * + * Returns non-zero if timestamp is considered to be very precise. + */ +u32 drm_get_last_vbltimestamp(struct drm_device *dev, int crtc, + struct timeval *tvblank, unsigned flags) +{ + int ret = 0; + + /* Define requested maximum error on timestamps (nanoseconds). */ + int max_error = (int) drm_timestamp_precision * 1000; + + /* Query driver if possible and precision timestamping enabled. */ + if (dev->driver->get_vblank_timestamp && (max_error > 0)) { + ret = dev->driver->get_vblank_timestamp(dev, crtc, &max_error, + tvblank, flags); + if (ret > 0) + return (u32) ret; + } + + /* GPU high precision timestamp query unsupported or failed. + * Return gettimeofday timestamp as best estimate. + */ + do_gettimeofday(tvblank); + + return 0; +} +EXPORT_SYMBOL(drm_get_last_vbltimestamp); + +/** + * drm_vblank_count - retrieve "cooked" vblank counter value + * @dev: DRM device + * @crtc: which counter to retrieve + * + * Fetches the "cooked" vblank count value that represents the number of + * vblank events since the system was booted, including lost events due to + * modesetting activity. + */ +u32 drm_vblank_count(struct drm_device *dev, int crtc) +{ + return atomic_read(&dev->_vblank_count[crtc]); +} +EXPORT_SYMBOL(drm_vblank_count); + +/** + * drm_vblank_count_and_time - retrieve "cooked" vblank counter value + * and the system timestamp corresponding to that vblank counter value. + * + * @dev: DRM device + * @crtc: which counter to retrieve + * @vblanktime: Pointer to struct timeval to receive the vblank timestamp. + * + * Fetches the "cooked" vblank count value that represents the number of + * vblank events since the system was booted, including lost events due to + * modesetting activity. Returns corresponding system timestamp of the time + * of the vblank interval that corresponds to the current value vblank counter + * value. + */ +u32 drm_vblank_count_and_time(struct drm_device *dev, int crtc, + struct timeval *vblanktime) +{ + u32 cur_vblank; + + /* Read timestamp from slot of _vblank_time ringbuffer + * that corresponds to current vblank count. Retry if + * count has incremented during readout. This works like + * a seqlock. + */ + do { + cur_vblank = atomic_read(&dev->_vblank_count[crtc]); + *vblanktime = vblanktimestamp(dev, crtc, cur_vblank); + smp_rmb(); + } while (cur_vblank != atomic_read(&dev->_vblank_count[crtc])); + + return cur_vblank; +} +EXPORT_SYMBOL(drm_vblank_count_and_time); + +/** + * drm_update_vblank_count - update the master vblank counter + * @dev: DRM device + * @crtc: counter to update + * + * Call back into the driver to update the appropriate vblank counter + * (specified by @crtc). Deal with wraparound, if it occurred, and + * update the last read value so we can deal with wraparound on the next + * call if necessary. + * + * Only necessary when going from off->on, to account for frames we + * didn't get an interrupt for. + * + * Note: caller must hold dev->vbl_lock since this reads & writes + * device vblank fields. + */ +static void drm_update_vblank_count(struct drm_device *dev, int crtc) +{ + u32 cur_vblank, diff, tslot, rc; + struct timeval t_vblank; + + /* + * Interrupts were disabled prior to this call, so deal with counter + * wrap if needed. + * NOTE! It's possible we lost a full dev->max_vblank_count events + * here if the register is small or we had vblank interrupts off for + * a long time. + * + * We repeat the hardware vblank counter & timestamp query until + * we get consistent results. This to prevent races between gpu + * updating its hardware counter while we are retrieving the + * corresponding vblank timestamp. + */ + do { + cur_vblank = dev->driver->get_vblank_counter(dev, crtc); + rc = drm_get_last_vbltimestamp(dev, crtc, &t_vblank, 0); + } while (cur_vblank != dev->driver->get_vblank_counter(dev, crtc)); + + /* Deal with counter wrap */ + diff = cur_vblank - dev->last_vblank[crtc]; + if (cur_vblank < dev->last_vblank[crtc]) { + diff += dev->max_vblank_count; + + DRM_DEBUG("last_vblank[%d]=0x%x, cur_vblank=0x%x => diff=0x%x\n", + crtc, dev->last_vblank[crtc], cur_vblank, diff); + } + + DRM_DEBUG("enabling vblank interrupts on crtc %d, missed %d\n", + crtc, diff); + + /* Reinitialize corresponding vblank timestamp if high-precision query + * available. Skip this step if query unsupported or failed. Will + * reinitialize delayed at next vblank interrupt in that case. + */ + if (rc) { + tslot = atomic_read(&dev->_vblank_count[crtc]) + diff; + vblanktimestamp(dev, crtc, tslot) = t_vblank; + } + + smp_mb__before_atomic_inc(); + atomic_add(diff, &dev->_vblank_count[crtc]); + smp_mb__after_atomic_inc(); +} + +/** + * drm_vblank_get - get a reference count on vblank events + * @dev: DRM device + * @crtc: which CRTC to own + * + * Acquire a reference count on vblank events to avoid having them disabled + * while in use. + * + * RETURNS + * Zero on success, nonzero on failure. + */ +int drm_vblank_get(struct drm_device *dev, int crtc) +{ + unsigned long irqflags, irqflags2; + int ret = 0; + + spin_lock_irqsave(&dev->vbl_lock, irqflags); + /* Going from 0->1 means we have to enable interrupts again */ + if (atomic_add_return(1, &dev->vblank_refcount[crtc]) == 1) { + /* Disable preemption while holding vblank_time_lock. Do + * it explicitely to guard against PREEMPT_RT kernel. + */ + preempt_disable(); + spin_lock_irqsave(&dev->vblank_time_lock, irqflags2); + if (!dev->vblank_enabled[crtc]) { + /* Enable vblank irqs under vblank_time_lock protection. + * All vblank count & timestamp updates are held off + * until we are done reinitializing master counter and + * timestamps. Filtercode in drm_handle_vblank() will + * prevent double-accounting of same vblank interval. + */ + ret = dev->driver->enable_vblank(dev, crtc); + DRM_DEBUG("enabling vblank on crtc %d, ret: %d\n", + crtc, ret); + if (ret) + atomic_dec(&dev->vblank_refcount[crtc]); + else { + dev->vblank_enabled[crtc] = 1; + drm_update_vblank_count(dev, crtc); + } + } + spin_unlock_irqrestore(&dev->vblank_time_lock, irqflags2); + preempt_enable(); + } else { + if (!dev->vblank_enabled[crtc]) { + atomic_dec(&dev->vblank_refcount[crtc]); + ret = -EINVAL; + } + } + spin_unlock_irqrestore(&dev->vbl_lock, irqflags); + + return ret; +} +EXPORT_SYMBOL(drm_vblank_get); + +/** + * drm_vblank_put - give up ownership of vblank events + * @dev: DRM device + * @crtc: which counter to give up + * + * Release ownership of a given vblank counter, turning off interrupts + * if possible. Disable interrupts after drm_vblank_offdelay milliseconds. + */ +void drm_vblank_put(struct drm_device *dev, int crtc) +{ + BUG_ON(atomic_read(&dev->vblank_refcount[crtc]) == 0); + + /* Last user schedules interrupt disable */ + if (atomic_dec_and_test(&dev->vblank_refcount[crtc]) && + (drm_vblank_offdelay > 0)) + mod_timer(&dev->vblank_disable_timer, + jiffies + ((drm_vblank_offdelay * DRM_HZ)/1000)); +} +EXPORT_SYMBOL(drm_vblank_put); + +void drm_vblank_off(struct drm_device *dev, int crtc) +{ + struct drm_pending_vblank_event *e, *t; + struct timeval now; + unsigned long irqflags; + unsigned int seq; + + spin_lock_irqsave(&dev->vbl_lock, irqflags); + vblank_disable_and_save(dev, crtc); + DRM_WAKEUP(&dev->vbl_queue[crtc]); + + /* Send any queued vblank events, lest the natives grow disquiet */ + seq = drm_vblank_count_and_time(dev, crtc, &now); + list_for_each_entry_safe(e, t, &dev->vblank_event_list, base.link) { + if (e->pipe != crtc) + continue; + DRM_DEBUG("Sending premature vblank event on disable: \ + wanted %d, current %d\n", + e->event.sequence, seq); + + e->event.sequence = seq; + e->event.tv_sec = now.tv_sec; + e->event.tv_usec = now.tv_usec; + drm_vblank_put(dev, e->pipe); + list_move_tail(&e->base.link, &e->base.file_priv->event_list); + wake_up_interruptible(&e->base.file_priv->event_wait); + trace_drm_vblank_event_delivered(e->base.pid, e->pipe, + e->event.sequence); + } + + spin_unlock_irqrestore(&dev->vbl_lock, irqflags); +} +EXPORT_SYMBOL(drm_vblank_off); + +/** + * drm_vblank_pre_modeset - account for vblanks across mode sets + * @dev: DRM device + * @crtc: CRTC in question + * @post: post or pre mode set? + * + * Account for vblank events across mode setting events, which will likely + * reset the hardware frame counter. + */ +void drm_vblank_pre_modeset(struct drm_device *dev, int crtc) +{ + /* vblank is not initialized (IRQ not installed ?) */ + if (!dev->num_crtcs) + return; + /* + * To avoid all the problems that might happen if interrupts + * were enabled/disabled around or between these calls, we just + * have the kernel take a reference on the CRTC (just once though + * to avoid corrupting the count if multiple, mismatch calls occur), + * so that interrupts remain enabled in the interim. + */ + if (!dev->vblank_inmodeset[crtc]) { + dev->vblank_inmodeset[crtc] = 0x1; + if (drm_vblank_get(dev, crtc) == 0) + dev->vblank_inmodeset[crtc] |= 0x2; + } +} +EXPORT_SYMBOL(drm_vblank_pre_modeset); + +void drm_vblank_post_modeset(struct drm_device *dev, int crtc) +{ + unsigned long irqflags; + + if (dev->vblank_inmodeset[crtc]) { + spin_lock_irqsave(&dev->vbl_lock, irqflags); + dev->vblank_disable_allowed = 1; + spin_unlock_irqrestore(&dev->vbl_lock, irqflags); + + if (dev->vblank_inmodeset[crtc] & 0x2) + drm_vblank_put(dev, crtc); + + dev->vblank_inmodeset[crtc] = 0; + } +} +EXPORT_SYMBOL(drm_vblank_post_modeset); + +/** + * drm_modeset_ctl - handle vblank event counter changes across mode switch + * @DRM_IOCTL_ARGS: standard ioctl arguments + * + * Applications should call the %_DRM_PRE_MODESET and %_DRM_POST_MODESET + * ioctls around modesetting so that any lost vblank events are accounted for. + * + * Generally the counter will reset across mode sets. If interrupts are + * enabled around this call, we don't have to do anything since the counter + * will have already been incremented. + */ +int drm_modeset_ctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_modeset_ctl *modeset = data; + int ret = 0; + unsigned int crtc; + + /* If drm_vblank_init() hasn't been called yet, just no-op */ + if (!dev->num_crtcs) + goto out; + + crtc = modeset->crtc; + if (crtc >= dev->num_crtcs) { + ret = -EINVAL; + goto out; + } + + switch (modeset->cmd) { + case _DRM_PRE_MODESET: + drm_vblank_pre_modeset(dev, crtc); + break; + case _DRM_POST_MODESET: + drm_vblank_post_modeset(dev, crtc); + break; + default: + ret = -EINVAL; + break; + } + +out: + return ret; +} + +static int drm_queue_vblank_event(struct drm_device *dev, int pipe, + union drm_wait_vblank *vblwait, + struct drm_file *file_priv) +{ + struct drm_pending_vblank_event *e; + struct timeval now; + unsigned long flags; + unsigned int seq; + int ret; + + e = kzalloc(sizeof *e, GFP_KERNEL); + if (e == NULL) { + ret = -ENOMEM; + goto err_put; + } + + e->pipe = pipe; + e->base.pid = current->pid; + e->event.base.type = DRM_EVENT_VBLANK; + e->event.base.length = sizeof e->event; + e->event.user_data = vblwait->request.signal; + e->base.event = &e->event.base; + e->base.file_priv = file_priv; + e->base.destroy = (void (*) (struct drm_pending_event *)) kfree; + + spin_lock_irqsave(&dev->event_lock, flags); + + if (file_priv->event_space < sizeof e->event) { + ret = -EBUSY; + goto err_unlock; + } + + file_priv->event_space -= sizeof e->event; + seq = drm_vblank_count_and_time(dev, pipe, &now); + + if ((vblwait->request.type & _DRM_VBLANK_NEXTONMISS) && + (seq - vblwait->request.sequence) <= (1 << 23)) { + vblwait->request.sequence = seq + 1; + vblwait->reply.sequence = vblwait->request.sequence; + } + + DRM_DEBUG("event on vblank count %d, current %d, crtc %d\n", + vblwait->request.sequence, seq, pipe); + + trace_drm_vblank_event_queued(current->pid, pipe, + vblwait->request.sequence); + + e->event.sequence = vblwait->request.sequence; + if ((seq - vblwait->request.sequence) <= (1 << 23)) { + e->event.sequence = seq; + e->event.tv_sec = now.tv_sec; + e->event.tv_usec = now.tv_usec; + drm_vblank_put(dev, pipe); + list_add_tail(&e->base.link, &e->base.file_priv->event_list); + wake_up_interruptible(&e->base.file_priv->event_wait); + vblwait->reply.sequence = seq; + trace_drm_vblank_event_delivered(current->pid, pipe, + vblwait->request.sequence); + } else { + list_add_tail(&e->base.link, &dev->vblank_event_list); + vblwait->reply.sequence = vblwait->request.sequence; + } + + spin_unlock_irqrestore(&dev->event_lock, flags); + + return 0; + +err_unlock: + spin_unlock_irqrestore(&dev->event_lock, flags); + kfree(e); +err_put: + drm_vblank_put(dev, pipe); + return ret; +} + +/** + * Wait for VBLANK. + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param data user argument, pointing to a drm_wait_vblank structure. + * \return zero on success or a negative number on failure. + * + * This function enables the vblank interrupt on the pipe requested, then + * sleeps waiting for the requested sequence number to occur, and drops + * the vblank interrupt refcount afterwards. (vblank irq disable follows that + * after a timeout with no further vblank waits scheduled). + */ +int drm_wait_vblank(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + union drm_wait_vblank *vblwait = data; + int ret = 0; + unsigned int flags, seq, crtc, high_crtc; + + if ((!drm_dev_to_irq(dev)) || (!dev->irq_enabled)) + return -EINVAL; + + if (vblwait->request.type & _DRM_VBLANK_SIGNAL) + return -EINVAL; + + if (vblwait->request.type & + ~(_DRM_VBLANK_TYPES_MASK | _DRM_VBLANK_FLAGS_MASK | + _DRM_VBLANK_HIGH_CRTC_MASK)) { + DRM_ERROR("Unsupported type value 0x%x, supported mask 0x%x\n", + vblwait->request.type, + (_DRM_VBLANK_TYPES_MASK | _DRM_VBLANK_FLAGS_MASK | + _DRM_VBLANK_HIGH_CRTC_MASK)); + return -EINVAL; + } + + flags = vblwait->request.type & _DRM_VBLANK_FLAGS_MASK; + high_crtc = (vblwait->request.type & _DRM_VBLANK_HIGH_CRTC_MASK); + if (high_crtc) + crtc = high_crtc >> _DRM_VBLANK_HIGH_CRTC_SHIFT; + else + crtc = flags & _DRM_VBLANK_SECONDARY ? 1 : 0; + if (crtc >= dev->num_crtcs) + return -EINVAL; + + ret = drm_vblank_get(dev, crtc); + if (ret) { + DRM_DEBUG("failed to acquire vblank counter, %d\n", ret); + return ret; + } + seq = drm_vblank_count(dev, crtc); + + switch (vblwait->request.type & _DRM_VBLANK_TYPES_MASK) { + case _DRM_VBLANK_RELATIVE: + vblwait->request.sequence += seq; + vblwait->request.type &= ~_DRM_VBLANK_RELATIVE; + case _DRM_VBLANK_ABSOLUTE: + break; + default: + ret = -EINVAL; + goto done; + } + + if (flags & _DRM_VBLANK_EVENT) + return drm_queue_vblank_event(dev, crtc, vblwait, file_priv); + + if ((flags & _DRM_VBLANK_NEXTONMISS) && + (seq - vblwait->request.sequence) <= (1<<23)) { + vblwait->request.sequence = seq + 1; + } + + DRM_DEBUG("waiting on vblank count %d, crtc %d\n", + vblwait->request.sequence, crtc); + dev->last_vblank_wait[crtc] = vblwait->request.sequence; + DRM_WAIT_ON(ret, dev->vbl_queue[crtc], 3 * DRM_HZ, + (((drm_vblank_count(dev, crtc) - + vblwait->request.sequence) <= (1 << 23)) || + !dev->irq_enabled)); + + if (ret != -EINTR) { + struct timeval now; + + vblwait->reply.sequence = drm_vblank_count_and_time(dev, crtc, &now); + vblwait->reply.tval_sec = now.tv_sec; + vblwait->reply.tval_usec = now.tv_usec; + + DRM_DEBUG("returning %d to client\n", + vblwait->reply.sequence); + } else { + DRM_DEBUG("vblank wait interrupted by signal\n"); + } + +done: + drm_vblank_put(dev, crtc); + return ret; +} + +void drm_handle_vblank_events(struct drm_device *dev, int crtc) +{ + struct drm_pending_vblank_event *e, *t; + struct timeval now; + unsigned long flags; + unsigned int seq; + + seq = drm_vblank_count_and_time(dev, crtc, &now); + + spin_lock_irqsave(&dev->event_lock, flags); + + list_for_each_entry_safe(e, t, &dev->vblank_event_list, base.link) { + if (e->pipe != crtc) + continue; + if ((seq - e->event.sequence) > (1<<23)) + continue; + + DRM_DEBUG("vblank event on %d, current %d\n", + e->event.sequence, seq); + + e->event.sequence = seq; + e->event.tv_sec = now.tv_sec; + e->event.tv_usec = now.tv_usec; + drm_vblank_put(dev, e->pipe); + list_move_tail(&e->base.link, &e->base.file_priv->event_list); + wake_up_interruptible(&e->base.file_priv->event_wait); + trace_drm_vblank_event_delivered(e->base.pid, e->pipe, + e->event.sequence); + } + + spin_unlock_irqrestore(&dev->event_lock, flags); + + trace_drm_vblank_event(crtc, seq); +} + +/** + * drm_handle_vblank - handle a vblank event + * @dev: DRM device + * @crtc: where this event occurred + * + * Drivers should call this routine in their vblank interrupt handlers to + * update the vblank counter and send any signals that may be pending. + */ +bool drm_handle_vblank(struct drm_device *dev, int crtc) +{ + u32 vblcount; + s64 diff_ns; + struct timeval tvblank; + unsigned long irqflags; + + if (!dev->num_crtcs) + return false; + + /* Need timestamp lock to prevent concurrent execution with + * vblank enable/disable, as this would cause inconsistent + * or corrupted timestamps and vblank counts. + */ + spin_lock_irqsave(&dev->vblank_time_lock, irqflags); + + /* Vblank irq handling disabled. Nothing to do. */ + if (!dev->vblank_enabled[crtc]) { + spin_unlock_irqrestore(&dev->vblank_time_lock, irqflags); + return false; + } + + /* Fetch corresponding timestamp for this vblank interval from + * driver and store it in proper slot of timestamp ringbuffer. + */ + + /* Get current timestamp and count. */ + vblcount = atomic_read(&dev->_vblank_count[crtc]); + drm_get_last_vbltimestamp(dev, crtc, &tvblank, DRM_CALLED_FROM_VBLIRQ); + + /* Compute time difference to timestamp of last vblank */ + diff_ns = timeval_to_ns(&tvblank) - + timeval_to_ns(&vblanktimestamp(dev, crtc, vblcount)); + + /* Update vblank timestamp and count if at least + * DRM_REDUNDANT_VBLIRQ_THRESH_NS nanoseconds + * difference between last stored timestamp and current + * timestamp. A smaller difference means basically + * identical timestamps. Happens if this vblank has + * been already processed and this is a redundant call, + * e.g., due to spurious vblank interrupts. We need to + * ignore those for accounting. + */ + if (abs64(diff_ns) > DRM_REDUNDANT_VBLIRQ_THRESH_NS) { + /* Store new timestamp in ringbuffer. */ + vblanktimestamp(dev, crtc, vblcount + 1) = tvblank; + + /* Increment cooked vblank count. This also atomically commits + * the timestamp computed above. + */ + smp_mb__before_atomic_inc(); + atomic_inc(&dev->_vblank_count[crtc]); + smp_mb__after_atomic_inc(); + } else { + DRM_DEBUG("crtc %d: Redundant vblirq ignored. diff_ns = %d\n", + crtc, (int) diff_ns); + } + + DRM_WAKEUP(&dev->vbl_queue[crtc]); + drm_handle_vblank_events(dev, crtc); + + spin_unlock_irqrestore(&dev->vblank_time_lock, irqflags); + return true; +} +EXPORT_SYMBOL(drm_handle_vblank); diff --git a/drivers/gpu/drm/drm_lock.c b/drivers/gpu/drm/drm_lock.c new file mode 100644 index 00000000..632ae243 --- /dev/null +++ b/drivers/gpu/drm/drm_lock.c @@ -0,0 +1,374 @@ +/** + * \file drm_lock.c + * IOCTLs for locking + * + * \author Rickard E. (Rik) Faith + * \author Gareth Hughes + */ + +/* + * Created: Tue Feb 2 08:37:54 1999 by faith@valinux.com + * + * Copyright 1999 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * VA LINUX SYSTEMS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#include "drmP.h" + +static int drm_notifier(void *priv); + +static int drm_lock_take(struct drm_lock_data *lock_data, unsigned int context); + +/** + * Lock ioctl. + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg user argument, pointing to a drm_lock structure. + * \return zero on success or negative number on failure. + * + * Add the current task to the lock wait queue, and attempt to take to lock. + */ +int drm_lock(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + DECLARE_WAITQUEUE(entry, current); + struct drm_lock *lock = data; + struct drm_master *master = file_priv->master; + int ret = 0; + + ++file_priv->lock_count; + + if (lock->context == DRM_KERNEL_CONTEXT) { + DRM_ERROR("Process %d using kernel context %d\n", + task_pid_nr(current), lock->context); + return -EINVAL; + } + + DRM_DEBUG("%d (pid %d) requests lock (0x%08x), flags = 0x%08x\n", + lock->context, task_pid_nr(current), + master->lock.hw_lock->lock, lock->flags); + + if (drm_core_check_feature(dev, DRIVER_DMA_QUEUE)) + if (lock->context < 0) + return -EINVAL; + + add_wait_queue(&master->lock.lock_queue, &entry); + spin_lock_bh(&master->lock.spinlock); + master->lock.user_waiters++; + spin_unlock_bh(&master->lock.spinlock); + + for (;;) { + __set_current_state(TASK_INTERRUPTIBLE); + if (!master->lock.hw_lock) { + /* Device has been unregistered */ + send_sig(SIGTERM, current, 0); + ret = -EINTR; + break; + } + if (drm_lock_take(&master->lock, lock->context)) { + master->lock.file_priv = file_priv; + master->lock.lock_time = jiffies; + atomic_inc(&dev->counts[_DRM_STAT_LOCKS]); + break; /* Got lock */ + } + + /* Contention */ + mutex_unlock(&drm_global_mutex); + schedule(); + mutex_lock(&drm_global_mutex); + if (signal_pending(current)) { + ret = -EINTR; + break; + } + } + spin_lock_bh(&master->lock.spinlock); + master->lock.user_waiters--; + spin_unlock_bh(&master->lock.spinlock); + __set_current_state(TASK_RUNNING); + remove_wait_queue(&master->lock.lock_queue, &entry); + + DRM_DEBUG("%d %s\n", lock->context, + ret ? "interrupted" : "has lock"); + if (ret) return ret; + + /* don't set the block all signals on the master process for now + * really probably not the correct answer but lets us debug xkb + * xserver for now */ + if (!file_priv->is_master) { + sigemptyset(&dev->sigmask); + sigaddset(&dev->sigmask, SIGSTOP); + sigaddset(&dev->sigmask, SIGTSTP); + sigaddset(&dev->sigmask, SIGTTIN); + sigaddset(&dev->sigmask, SIGTTOU); + dev->sigdata.context = lock->context; + dev->sigdata.lock = master->lock.hw_lock; + block_all_signals(drm_notifier, &dev->sigdata, &dev->sigmask); + } + + if (dev->driver->dma_quiescent && (lock->flags & _DRM_LOCK_QUIESCENT)) + { + if (dev->driver->dma_quiescent(dev)) { + DRM_DEBUG("%d waiting for DMA quiescent\n", + lock->context); + return -EBUSY; + } + } + + return 0; +} + +/** + * Unlock ioctl. + * + * \param inode device inode. + * \param file_priv DRM file private. + * \param cmd command. + * \param arg user argument, pointing to a drm_lock structure. + * \return zero on success or negative number on failure. + * + * Transfer and free the lock. + */ +int drm_unlock(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + struct drm_lock *lock = data; + struct drm_master *master = file_priv->master; + + if (lock->context == DRM_KERNEL_CONTEXT) { + DRM_ERROR("Process %d using kernel context %d\n", + task_pid_nr(current), lock->context); + return -EINVAL; + } + + atomic_inc(&dev->counts[_DRM_STAT_UNLOCKS]); + + if (drm_lock_free(&master->lock, lock->context)) { + /* FIXME: Should really bail out here. */ + } + + unblock_all_signals(); + return 0; +} + +/** + * Take the heavyweight lock. + * + * \param lock lock pointer. + * \param context locking context. + * \return one if the lock is held, or zero otherwise. + * + * Attempt to mark the lock as held by the given context, via the \p cmpxchg instruction. + */ +static +int drm_lock_take(struct drm_lock_data *lock_data, + unsigned int context) +{ + unsigned int old, new, prev; + volatile unsigned int *lock = &lock_data->hw_lock->lock; + + spin_lock_bh(&lock_data->spinlock); + do { + old = *lock; + if (old & _DRM_LOCK_HELD) + new = old | _DRM_LOCK_CONT; + else { + new = context | _DRM_LOCK_HELD | + ((lock_data->user_waiters + lock_data->kernel_waiters > 1) ? + _DRM_LOCK_CONT : 0); + } + prev = cmpxchg(lock, old, new); + } while (prev != old); + spin_unlock_bh(&lock_data->spinlock); + + if (_DRM_LOCKING_CONTEXT(old) == context) { + if (old & _DRM_LOCK_HELD) { + if (context != DRM_KERNEL_CONTEXT) { + DRM_ERROR("%d holds heavyweight lock\n", + context); + } + return 0; + } + } + + if ((_DRM_LOCKING_CONTEXT(new)) == context && (new & _DRM_LOCK_HELD)) { + /* Have lock */ + return 1; + } + return 0; +} + +/** + * This takes a lock forcibly and hands it to context. Should ONLY be used + * inside *_unlock to give lock to kernel before calling *_dma_schedule. + * + * \param dev DRM device. + * \param lock lock pointer. + * \param context locking context. + * \return always one. + * + * Resets the lock file pointer. + * Marks the lock as held by the given context, via the \p cmpxchg instruction. + */ +static int drm_lock_transfer(struct drm_lock_data *lock_data, + unsigned int context) +{ + unsigned int old, new, prev; + volatile unsigned int *lock = &lock_data->hw_lock->lock; + + lock_data->file_priv = NULL; + do { + old = *lock; + new = context | _DRM_LOCK_HELD; + prev = cmpxchg(lock, old, new); + } while (prev != old); + return 1; +} + +/** + * Free lock. + * + * \param dev DRM device. + * \param lock lock. + * \param context context. + * + * Resets the lock file pointer. + * Marks the lock as not held, via the \p cmpxchg instruction. Wakes any task + * waiting on the lock queue. + */ +int drm_lock_free(struct drm_lock_data *lock_data, unsigned int context) +{ + unsigned int old, new, prev; + volatile unsigned int *lock = &lock_data->hw_lock->lock; + + spin_lock_bh(&lock_data->spinlock); + if (lock_data->kernel_waiters != 0) { + drm_lock_transfer(lock_data, 0); + lock_data->idle_has_lock = 1; + spin_unlock_bh(&lock_data->spinlock); + return 1; + } + spin_unlock_bh(&lock_data->spinlock); + + do { + old = *lock; + new = _DRM_LOCKING_CONTEXT(old); + prev = cmpxchg(lock, old, new); + } while (prev != old); + + if (_DRM_LOCK_IS_HELD(old) && _DRM_LOCKING_CONTEXT(old) != context) { + DRM_ERROR("%d freed heavyweight lock held by %d\n", + context, _DRM_LOCKING_CONTEXT(old)); + return 1; + } + wake_up_interruptible(&lock_data->lock_queue); + return 0; +} + +/** + * If we get here, it means that the process has called DRM_IOCTL_LOCK + * without calling DRM_IOCTL_UNLOCK. + * + * If the lock is not held, then let the signal proceed as usual. If the lock + * is held, then set the contended flag and keep the signal blocked. + * + * \param priv pointer to a drm_sigdata structure. + * \return one if the signal should be delivered normally, or zero if the + * signal should be blocked. + */ +static int drm_notifier(void *priv) +{ + struct drm_sigdata *s = (struct drm_sigdata *) priv; + unsigned int old, new, prev; + + /* Allow signal delivery if lock isn't held */ + if (!s->lock || !_DRM_LOCK_IS_HELD(s->lock->lock) + || _DRM_LOCKING_CONTEXT(s->lock->lock) != s->context) + return 1; + + /* Otherwise, set flag to force call to + drmUnlock */ + do { + old = s->lock->lock; + new = old | _DRM_LOCK_CONT; + prev = cmpxchg(&s->lock->lock, old, new); + } while (prev != old); + return 0; +} + +/** + * This function returns immediately and takes the hw lock + * with the kernel context if it is free, otherwise it gets the highest priority when and if + * it is eventually released. + * + * This guarantees that the kernel will _eventually_ have the lock _unless_ it is held + * by a blocked process. (In the latter case an explicit wait for the hardware lock would cause + * a deadlock, which is why the "idlelock" was invented). + * + * This should be sufficient to wait for GPU idle without + * having to worry about starvation. + */ + +void drm_idlelock_take(struct drm_lock_data *lock_data) +{ + int ret = 0; + + spin_lock_bh(&lock_data->spinlock); + lock_data->kernel_waiters++; + if (!lock_data->idle_has_lock) { + + spin_unlock_bh(&lock_data->spinlock); + ret = drm_lock_take(lock_data, DRM_KERNEL_CONTEXT); + spin_lock_bh(&lock_data->spinlock); + + if (ret == 1) + lock_data->idle_has_lock = 1; + } + spin_unlock_bh(&lock_data->spinlock); +} + +void drm_idlelock_release(struct drm_lock_data *lock_data) +{ + unsigned int old, prev; + volatile unsigned int *lock = &lock_data->hw_lock->lock; + + spin_lock_bh(&lock_data->spinlock); + if (--lock_data->kernel_waiters == 0) { + if (lock_data->idle_has_lock) { + do { + old = *lock; + prev = cmpxchg(lock, old, DRM_KERNEL_CONTEXT); + } while (prev != old); + wake_up_interruptible(&lock_data->lock_queue); + lock_data->idle_has_lock = 0; + } + } + spin_unlock_bh(&lock_data->spinlock); +} + +int drm_i_have_hw_lock(struct drm_device *dev, struct drm_file *file_priv) +{ + struct drm_master *master = file_priv->master; + return (file_priv->lock_count && master->lock.hw_lock && + _DRM_LOCK_IS_HELD(master->lock.hw_lock->lock) && + master->lock.file_priv == file_priv); +} diff --git a/drivers/gpu/drm/drm_memory.c b/drivers/gpu/drm/drm_memory.c new file mode 100644 index 00000000..c9b80500 --- /dev/null +++ b/drivers/gpu/drm/drm_memory.c @@ -0,0 +1,162 @@ +/** + * \file drm_memory.c + * Memory management wrappers for DRM + * + * \author Rickard E. (Rik) Faith + * \author Gareth Hughes + */ + +/* + * Created: Thu Feb 4 14:00:34 1999 by faith@valinux.com + * + * Copyright 1999 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * VA LINUX SYSTEMS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#include +#include "drmP.h" + +/** + * Called when "/proc/dri/%dev%/mem" is read. + * + * \param buf output buffer. + * \param start start of output data. + * \param offset requested start offset. + * \param len requested number of bytes. + * \param eof whether there is no more data to return. + * \param data private data. + * \return number of written bytes. + * + * No-op. + */ +int drm_mem_info(char *buf, char **start, off_t offset, + int len, int *eof, void *data) +{ + return 0; +} + +#if __OS_HAS_AGP +static void *agp_remap(unsigned long offset, unsigned long size, + struct drm_device * dev) +{ + unsigned long i, num_pages = + PAGE_ALIGN(size) / PAGE_SIZE; + struct drm_agp_mem *agpmem; + struct page **page_map; + struct page **phys_page_map; + void *addr; + + size = PAGE_ALIGN(size); + +#ifdef __alpha__ + offset -= dev->hose->mem_space->start; +#endif + + list_for_each_entry(agpmem, &dev->agp->memory, head) + if (agpmem->bound <= offset + && (agpmem->bound + (agpmem->pages << PAGE_SHIFT)) >= + (offset + size)) + break; + if (&agpmem->head == &dev->agp->memory) + return NULL; + + /* + * OK, we're mapping AGP space on a chipset/platform on which memory accesses by + * the CPU do not get remapped by the GART. We fix this by using the kernel's + * page-table instead (that's probably faster anyhow...). + */ + /* note: use vmalloc() because num_pages could be large... */ + page_map = vmalloc(num_pages * sizeof(struct page *)); + if (!page_map) + return NULL; + + phys_page_map = (agpmem->memory->pages + (offset - agpmem->bound) / PAGE_SIZE); + for (i = 0; i < num_pages; ++i) + page_map[i] = phys_page_map[i]; + addr = vmap(page_map, num_pages, VM_IOREMAP, PAGE_AGP); + vfree(page_map); + + return addr; +} + +/** Wrapper around agp_free_memory() */ +void drm_free_agp(DRM_AGP_MEM * handle, int pages) +{ + agp_free_memory(handle); +} +EXPORT_SYMBOL(drm_free_agp); + +/** Wrapper around agp_bind_memory() */ +int drm_bind_agp(DRM_AGP_MEM * handle, unsigned int start) +{ + return agp_bind_memory(handle, start); +} + +/** Wrapper around agp_unbind_memory() */ +int drm_unbind_agp(DRM_AGP_MEM * handle) +{ + return agp_unbind_memory(handle); +} +EXPORT_SYMBOL(drm_unbind_agp); + +#else /* __OS_HAS_AGP */ +static inline void *agp_remap(unsigned long offset, unsigned long size, + struct drm_device * dev) +{ + return NULL; +} + +#endif /* agp */ + +void drm_core_ioremap(struct drm_local_map *map, struct drm_device *dev) +{ + if (drm_core_has_AGP(dev) && + dev->agp && dev->agp->cant_use_aperture && map->type == _DRM_AGP) + map->handle = agp_remap(map->offset, map->size, dev); + else + map->handle = ioremap(map->offset, map->size); +} +EXPORT_SYMBOL(drm_core_ioremap); + +void drm_core_ioremap_wc(struct drm_local_map *map, struct drm_device *dev) +{ + if (drm_core_has_AGP(dev) && + dev->agp && dev->agp->cant_use_aperture && map->type == _DRM_AGP) + map->handle = agp_remap(map->offset, map->size, dev); + else + map->handle = ioremap_wc(map->offset, map->size); +} +EXPORT_SYMBOL(drm_core_ioremap_wc); + +void drm_core_ioremapfree(struct drm_local_map *map, struct drm_device *dev) +{ + if (!map->handle || !map->size) + return; + + if (drm_core_has_AGP(dev) && + dev->agp && dev->agp->cant_use_aperture && map->type == _DRM_AGP) + vunmap(map->handle); + else + iounmap(map->handle); +} +EXPORT_SYMBOL(drm_core_ioremapfree); diff --git a/drivers/gpu/drm/drm_mm.c b/drivers/gpu/drm/drm_mm.c new file mode 100644 index 00000000..959186cb --- /dev/null +++ b/drivers/gpu/drm/drm_mm.c @@ -0,0 +1,716 @@ +/************************************************************************** + * + * Copyright 2006 Tungsten Graphics, Inc., Bismarck, ND., USA. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. + * + * + **************************************************************************/ + +/* + * Generic simple memory manager implementation. Intended to be used as a base + * class implementation for more advanced memory managers. + * + * Note that the algorithm used is quite simple and there might be substantial + * performance gains if a smarter free list is implemented. Currently it is just an + * unordered stack of free regions. This could easily be improved if an RB-tree + * is used instead. At least if we expect heavy fragmentation. + * + * Aligned allocations can also see improvement. + * + * Authors: + * Thomas Hellström + */ + +#include "drmP.h" +#include "drm_mm.h" +#include +#include + +#define MM_UNUSED_TARGET 4 + +static struct drm_mm_node *drm_mm_kmalloc(struct drm_mm *mm, int atomic) +{ + struct drm_mm_node *child; + + if (atomic) + child = kzalloc(sizeof(*child), GFP_ATOMIC); + else + child = kzalloc(sizeof(*child), GFP_KERNEL); + + if (unlikely(child == NULL)) { + spin_lock(&mm->unused_lock); + if (list_empty(&mm->unused_nodes)) + child = NULL; + else { + child = + list_entry(mm->unused_nodes.next, + struct drm_mm_node, node_list); + list_del(&child->node_list); + --mm->num_unused; + } + spin_unlock(&mm->unused_lock); + } + return child; +} + +/* drm_mm_pre_get() - pre allocate drm_mm_node structure + * drm_mm: memory manager struct we are pre-allocating for + * + * Returns 0 on success or -ENOMEM if allocation fails. + */ +int drm_mm_pre_get(struct drm_mm *mm) +{ + struct drm_mm_node *node; + + spin_lock(&mm->unused_lock); + while (mm->num_unused < MM_UNUSED_TARGET) { + spin_unlock(&mm->unused_lock); + node = kzalloc(sizeof(*node), GFP_KERNEL); + spin_lock(&mm->unused_lock); + + if (unlikely(node == NULL)) { + int ret = (mm->num_unused < 2) ? -ENOMEM : 0; + spin_unlock(&mm->unused_lock); + return ret; + } + ++mm->num_unused; + list_add_tail(&node->node_list, &mm->unused_nodes); + } + spin_unlock(&mm->unused_lock); + return 0; +} +EXPORT_SYMBOL(drm_mm_pre_get); + +static inline unsigned long drm_mm_hole_node_start(struct drm_mm_node *hole_node) +{ + return hole_node->start + hole_node->size; +} + +static inline unsigned long drm_mm_hole_node_end(struct drm_mm_node *hole_node) +{ + struct drm_mm_node *next_node = + list_entry(hole_node->node_list.next, struct drm_mm_node, + node_list); + + return next_node->start; +} + +static void drm_mm_insert_helper(struct drm_mm_node *hole_node, + struct drm_mm_node *node, + unsigned long size, unsigned alignment) +{ + struct drm_mm *mm = hole_node->mm; + unsigned long tmp = 0, wasted = 0; + unsigned long hole_start = drm_mm_hole_node_start(hole_node); + unsigned long hole_end = drm_mm_hole_node_end(hole_node); + + BUG_ON(!hole_node->hole_follows || node->allocated); + + if (alignment) + tmp = hole_start % alignment; + + if (!tmp) { + hole_node->hole_follows = 0; + list_del_init(&hole_node->hole_stack); + } else + wasted = alignment - tmp; + + node->start = hole_start + wasted; + node->size = size; + node->mm = mm; + node->allocated = 1; + + INIT_LIST_HEAD(&node->hole_stack); + list_add(&node->node_list, &hole_node->node_list); + + BUG_ON(node->start + node->size > hole_end); + + if (node->start + node->size < hole_end) { + list_add(&node->hole_stack, &mm->hole_stack); + node->hole_follows = 1; + } else { + node->hole_follows = 0; + } +} + +struct drm_mm_node *drm_mm_get_block_generic(struct drm_mm_node *hole_node, + unsigned long size, + unsigned alignment, + int atomic) +{ + struct drm_mm_node *node; + + node = drm_mm_kmalloc(hole_node->mm, atomic); + if (unlikely(node == NULL)) + return NULL; + + drm_mm_insert_helper(hole_node, node, size, alignment); + + return node; +} +EXPORT_SYMBOL(drm_mm_get_block_generic); + +/** + * Search for free space and insert a preallocated memory node. Returns + * -ENOSPC if no suitable free area is available. The preallocated memory node + * must be cleared. + */ +int drm_mm_insert_node(struct drm_mm *mm, struct drm_mm_node *node, + unsigned long size, unsigned alignment) +{ + struct drm_mm_node *hole_node; + + hole_node = drm_mm_search_free(mm, size, alignment, 0); + if (!hole_node) + return -ENOSPC; + + drm_mm_insert_helper(hole_node, node, size, alignment); + + return 0; +} +EXPORT_SYMBOL(drm_mm_insert_node); + +static void drm_mm_insert_helper_range(struct drm_mm_node *hole_node, + struct drm_mm_node *node, + unsigned long size, unsigned alignment, + unsigned long start, unsigned long end) +{ + struct drm_mm *mm = hole_node->mm; + unsigned long tmp = 0, wasted = 0; + unsigned long hole_start = drm_mm_hole_node_start(hole_node); + unsigned long hole_end = drm_mm_hole_node_end(hole_node); + + BUG_ON(!hole_node->hole_follows || node->allocated); + + if (hole_start < start) + wasted += start - hole_start; + if (alignment) + tmp = (hole_start + wasted) % alignment; + + if (tmp) + wasted += alignment - tmp; + + if (!wasted) { + hole_node->hole_follows = 0; + list_del_init(&hole_node->hole_stack); + } + + node->start = hole_start + wasted; + node->size = size; + node->mm = mm; + node->allocated = 1; + + INIT_LIST_HEAD(&node->hole_stack); + list_add(&node->node_list, &hole_node->node_list); + + BUG_ON(node->start + node->size > hole_end); + BUG_ON(node->start + node->size > end); + + if (node->start + node->size < hole_end) { + list_add(&node->hole_stack, &mm->hole_stack); + node->hole_follows = 1; + } else { + node->hole_follows = 0; + } +} + +struct drm_mm_node *drm_mm_get_block_range_generic(struct drm_mm_node *hole_node, + unsigned long size, + unsigned alignment, + unsigned long start, + unsigned long end, + int atomic) +{ + struct drm_mm_node *node; + + node = drm_mm_kmalloc(hole_node->mm, atomic); + if (unlikely(node == NULL)) + return NULL; + + drm_mm_insert_helper_range(hole_node, node, size, alignment, + start, end); + + return node; +} +EXPORT_SYMBOL(drm_mm_get_block_range_generic); + +/** + * Search for free space and insert a preallocated memory node. Returns + * -ENOSPC if no suitable free area is available. This is for range + * restricted allocations. The preallocated memory node must be cleared. + */ +int drm_mm_insert_node_in_range(struct drm_mm *mm, struct drm_mm_node *node, + unsigned long size, unsigned alignment, + unsigned long start, unsigned long end) +{ + struct drm_mm_node *hole_node; + + hole_node = drm_mm_search_free_in_range(mm, size, alignment, + start, end, 0); + if (!hole_node) + return -ENOSPC; + + drm_mm_insert_helper_range(hole_node, node, size, alignment, + start, end); + + return 0; +} +EXPORT_SYMBOL(drm_mm_insert_node_in_range); + +/** + * Remove a memory node from the allocator. + */ +void drm_mm_remove_node(struct drm_mm_node *node) +{ + struct drm_mm *mm = node->mm; + struct drm_mm_node *prev_node; + + BUG_ON(node->scanned_block || node->scanned_prev_free + || node->scanned_next_free); + + prev_node = + list_entry(node->node_list.prev, struct drm_mm_node, node_list); + + if (node->hole_follows) { + BUG_ON(drm_mm_hole_node_start(node) + == drm_mm_hole_node_end(node)); + list_del(&node->hole_stack); + } else + BUG_ON(drm_mm_hole_node_start(node) + != drm_mm_hole_node_end(node)); + + if (!prev_node->hole_follows) { + prev_node->hole_follows = 1; + list_add(&prev_node->hole_stack, &mm->hole_stack); + } else + list_move(&prev_node->hole_stack, &mm->hole_stack); + + list_del(&node->node_list); + node->allocated = 0; +} +EXPORT_SYMBOL(drm_mm_remove_node); + +/* + * Remove a memory node from the allocator and free the allocated struct + * drm_mm_node. Only to be used on a struct drm_mm_node obtained by one of the + * drm_mm_get_block functions. + */ +void drm_mm_put_block(struct drm_mm_node *node) +{ + + struct drm_mm *mm = node->mm; + + drm_mm_remove_node(node); + + spin_lock(&mm->unused_lock); + if (mm->num_unused < MM_UNUSED_TARGET) { + list_add(&node->node_list, &mm->unused_nodes); + ++mm->num_unused; + } else + kfree(node); + spin_unlock(&mm->unused_lock); +} +EXPORT_SYMBOL(drm_mm_put_block); + +static int check_free_hole(unsigned long start, unsigned long end, + unsigned long size, unsigned alignment) +{ + unsigned wasted = 0; + + if (end - start < size) + return 0; + + if (alignment) { + unsigned tmp = start % alignment; + if (tmp) + wasted = alignment - tmp; + } + + if (end >= start + size + wasted) { + return 1; + } + + return 0; +} + +struct drm_mm_node *drm_mm_search_free(const struct drm_mm *mm, + unsigned long size, + unsigned alignment, int best_match) +{ + struct drm_mm_node *entry; + struct drm_mm_node *best; + unsigned long best_size; + + BUG_ON(mm->scanned_blocks); + + best = NULL; + best_size = ~0UL; + + list_for_each_entry(entry, &mm->hole_stack, hole_stack) { + BUG_ON(!entry->hole_follows); + if (!check_free_hole(drm_mm_hole_node_start(entry), + drm_mm_hole_node_end(entry), + size, alignment)) + continue; + + if (!best_match) + return entry; + + if (entry->size < best_size) { + best = entry; + best_size = entry->size; + } + } + + return best; +} +EXPORT_SYMBOL(drm_mm_search_free); + +struct drm_mm_node *drm_mm_search_free_in_range(const struct drm_mm *mm, + unsigned long size, + unsigned alignment, + unsigned long start, + unsigned long end, + int best_match) +{ + struct drm_mm_node *entry; + struct drm_mm_node *best; + unsigned long best_size; + + BUG_ON(mm->scanned_blocks); + + best = NULL; + best_size = ~0UL; + + list_for_each_entry(entry, &mm->hole_stack, hole_stack) { + unsigned long adj_start = drm_mm_hole_node_start(entry) < start ? + start : drm_mm_hole_node_start(entry); + unsigned long adj_end = drm_mm_hole_node_end(entry) > end ? + end : drm_mm_hole_node_end(entry); + + BUG_ON(!entry->hole_follows); + if (!check_free_hole(adj_start, adj_end, size, alignment)) + continue; + + if (!best_match) + return entry; + + if (entry->size < best_size) { + best = entry; + best_size = entry->size; + } + } + + return best; +} +EXPORT_SYMBOL(drm_mm_search_free_in_range); + +/** + * Moves an allocation. To be used with embedded struct drm_mm_node. + */ +void drm_mm_replace_node(struct drm_mm_node *old, struct drm_mm_node *new) +{ + list_replace(&old->node_list, &new->node_list); + list_replace(&old->hole_stack, &new->hole_stack); + new->hole_follows = old->hole_follows; + new->mm = old->mm; + new->start = old->start; + new->size = old->size; + + old->allocated = 0; + new->allocated = 1; +} +EXPORT_SYMBOL(drm_mm_replace_node); + +/** + * Initializa lru scanning. + * + * This simply sets up the scanning routines with the parameters for the desired + * hole. + * + * Warning: As long as the scan list is non-empty, no other operations than + * adding/removing nodes to/from the scan list are allowed. + */ +void drm_mm_init_scan(struct drm_mm *mm, unsigned long size, + unsigned alignment) +{ + mm->scan_alignment = alignment; + mm->scan_size = size; + mm->scanned_blocks = 0; + mm->scan_hit_start = 0; + mm->scan_hit_size = 0; + mm->scan_check_range = 0; + mm->prev_scanned_node = NULL; +} +EXPORT_SYMBOL(drm_mm_init_scan); + +/** + * Initializa lru scanning. + * + * This simply sets up the scanning routines with the parameters for the desired + * hole. This version is for range-restricted scans. + * + * Warning: As long as the scan list is non-empty, no other operations than + * adding/removing nodes to/from the scan list are allowed. + */ +void drm_mm_init_scan_with_range(struct drm_mm *mm, unsigned long size, + unsigned alignment, + unsigned long start, + unsigned long end) +{ + mm->scan_alignment = alignment; + mm->scan_size = size; + mm->scanned_blocks = 0; + mm->scan_hit_start = 0; + mm->scan_hit_size = 0; + mm->scan_start = start; + mm->scan_end = end; + mm->scan_check_range = 1; + mm->prev_scanned_node = NULL; +} +EXPORT_SYMBOL(drm_mm_init_scan_with_range); + +/** + * Add a node to the scan list that might be freed to make space for the desired + * hole. + * + * Returns non-zero, if a hole has been found, zero otherwise. + */ +int drm_mm_scan_add_block(struct drm_mm_node *node) +{ + struct drm_mm *mm = node->mm; + struct drm_mm_node *prev_node; + unsigned long hole_start, hole_end; + unsigned long adj_start; + unsigned long adj_end; + + mm->scanned_blocks++; + + BUG_ON(node->scanned_block); + node->scanned_block = 1; + + prev_node = list_entry(node->node_list.prev, struct drm_mm_node, + node_list); + + node->scanned_preceeds_hole = prev_node->hole_follows; + prev_node->hole_follows = 1; + list_del(&node->node_list); + node->node_list.prev = &prev_node->node_list; + node->node_list.next = &mm->prev_scanned_node->node_list; + mm->prev_scanned_node = node; + + hole_start = drm_mm_hole_node_start(prev_node); + hole_end = drm_mm_hole_node_end(prev_node); + if (mm->scan_check_range) { + adj_start = hole_start < mm->scan_start ? + mm->scan_start : hole_start; + adj_end = hole_end > mm->scan_end ? + mm->scan_end : hole_end; + } else { + adj_start = hole_start; + adj_end = hole_end; + } + + if (check_free_hole(adj_start , adj_end, + mm->scan_size, mm->scan_alignment)) { + mm->scan_hit_start = hole_start; + mm->scan_hit_size = hole_end; + + return 1; + } + + return 0; +} +EXPORT_SYMBOL(drm_mm_scan_add_block); + +/** + * Remove a node from the scan list. + * + * Nodes _must_ be removed in the exact same order from the scan list as they + * have been added, otherwise the internal state of the memory manager will be + * corrupted. + * + * When the scan list is empty, the selected memory nodes can be freed. An + * immediately following drm_mm_search_free with best_match = 0 will then return + * the just freed block (because its at the top of the free_stack list). + * + * Returns one if this block should be evicted, zero otherwise. Will always + * return zero when no hole has been found. + */ +int drm_mm_scan_remove_block(struct drm_mm_node *node) +{ + struct drm_mm *mm = node->mm; + struct drm_mm_node *prev_node; + + mm->scanned_blocks--; + + BUG_ON(!node->scanned_block); + node->scanned_block = 0; + + prev_node = list_entry(node->node_list.prev, struct drm_mm_node, + node_list); + + prev_node->hole_follows = node->scanned_preceeds_hole; + INIT_LIST_HEAD(&node->node_list); + list_add(&node->node_list, &prev_node->node_list); + + /* Only need to check for containement because start&size for the + * complete resulting free block (not just the desired part) is + * stored. */ + if (node->start >= mm->scan_hit_start && + node->start + node->size + <= mm->scan_hit_start + mm->scan_hit_size) { + return 1; + } + + return 0; +} +EXPORT_SYMBOL(drm_mm_scan_remove_block); + +int drm_mm_clean(struct drm_mm * mm) +{ + struct list_head *head = &mm->head_node.node_list; + + return (head->next->next == head); +} +EXPORT_SYMBOL(drm_mm_clean); + +int drm_mm_init(struct drm_mm * mm, unsigned long start, unsigned long size) +{ + INIT_LIST_HEAD(&mm->hole_stack); + INIT_LIST_HEAD(&mm->unused_nodes); + mm->num_unused = 0; + mm->scanned_blocks = 0; + spin_lock_init(&mm->unused_lock); + + /* Clever trick to avoid a special case in the free hole tracking. */ + INIT_LIST_HEAD(&mm->head_node.node_list); + INIT_LIST_HEAD(&mm->head_node.hole_stack); + mm->head_node.hole_follows = 1; + mm->head_node.scanned_block = 0; + mm->head_node.scanned_prev_free = 0; + mm->head_node.scanned_next_free = 0; + mm->head_node.mm = mm; + mm->head_node.start = start + size; + mm->head_node.size = start - mm->head_node.start; + list_add_tail(&mm->head_node.hole_stack, &mm->hole_stack); + + return 0; +} +EXPORT_SYMBOL(drm_mm_init); + +void drm_mm_takedown(struct drm_mm * mm) +{ + struct drm_mm_node *entry, *next; + + if (!list_empty(&mm->head_node.node_list)) { + DRM_ERROR("Memory manager not clean. Delaying takedown\n"); + return; + } + + spin_lock(&mm->unused_lock); + list_for_each_entry_safe(entry, next, &mm->unused_nodes, node_list) { + list_del(&entry->node_list); + kfree(entry); + --mm->num_unused; + } + spin_unlock(&mm->unused_lock); + + BUG_ON(mm->num_unused != 0); +} +EXPORT_SYMBOL(drm_mm_takedown); + +void drm_mm_debug_table(struct drm_mm *mm, const char *prefix) +{ + struct drm_mm_node *entry; + unsigned long total_used = 0, total_free = 0, total = 0; + unsigned long hole_start, hole_end, hole_size; + + hole_start = drm_mm_hole_node_start(&mm->head_node); + hole_end = drm_mm_hole_node_end(&mm->head_node); + hole_size = hole_end - hole_start; + if (hole_size) + printk(KERN_DEBUG "%s 0x%08lx-0x%08lx: %8lu: free\n", + prefix, hole_start, hole_end, + hole_size); + total_free += hole_size; + + drm_mm_for_each_node(entry, mm) { + printk(KERN_DEBUG "%s 0x%08lx-0x%08lx: %8lu: used\n", + prefix, entry->start, entry->start + entry->size, + entry->size); + total_used += entry->size; + + if (entry->hole_follows) { + hole_start = drm_mm_hole_node_start(entry); + hole_end = drm_mm_hole_node_end(entry); + hole_size = hole_end - hole_start; + printk(KERN_DEBUG "%s 0x%08lx-0x%08lx: %8lu: free\n", + prefix, hole_start, hole_end, + hole_size); + total_free += hole_size; + } + } + total = total_free + total_used; + + printk(KERN_DEBUG "%s total: %lu, used %lu free %lu\n", prefix, total, + total_used, total_free); +} +EXPORT_SYMBOL(drm_mm_debug_table); + +#if defined(CONFIG_DEBUG_FS) +int drm_mm_dump_table(struct seq_file *m, struct drm_mm *mm) +{ + struct drm_mm_node *entry; + unsigned long total_used = 0, total_free = 0, total = 0; + unsigned long hole_start, hole_end, hole_size; + + hole_start = drm_mm_hole_node_start(&mm->head_node); + hole_end = drm_mm_hole_node_end(&mm->head_node); + hole_size = hole_end - hole_start; + if (hole_size) + seq_printf(m, "0x%08lx-0x%08lx: 0x%08lx: free\n", + hole_start, hole_end, hole_size); + total_free += hole_size; + + drm_mm_for_each_node(entry, mm) { + seq_printf(m, "0x%08lx-0x%08lx: 0x%08lx: used\n", + entry->start, entry->start + entry->size, + entry->size); + total_used += entry->size; + if (entry->hole_follows) { + hole_start = drm_mm_hole_node_start(entry); + hole_end = drm_mm_hole_node_end(entry); + hole_size = hole_end - hole_start; + seq_printf(m, "0x%08lx-0x%08lx: 0x%08lx: free\n", + hole_start, hole_end, hole_size); + total_free += hole_size; + } + } + total = total_free + total_used; + + seq_printf(m, "total: %lu, used %lu free %lu\n", total, total_used, total_free); + return 0; +} +EXPORT_SYMBOL(drm_mm_dump_table); +#endif diff --git a/drivers/gpu/drm/drm_modes.c b/drivers/gpu/drm/drm_modes.c new file mode 100644 index 00000000..c2d32f20 --- /dev/null +++ b/drivers/gpu/drm/drm_modes.c @@ -0,0 +1,1132 @@ +/* + * Copyright © 1997-2003 by The XFree86 Project, Inc. + * Copyright © 2007 Dave Airlie + * Copyright © 2007-2008 Intel Corporation + * Jesse Barnes + * Copyright 2005-2006 Luc Verhaegen + * Copyright (c) 2001, Andy Ritger aritger@nvidia.com + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Except as contained in this notice, the name of the copyright holder(s) + * and author(s) shall not be used in advertising or otherwise to promote + * the sale, use or other dealings in this Software without prior written + * authorization from the copyright holder(s) and author(s). + */ + +#include +#include +#include "drmP.h" +#include "drm.h" +#include "drm_crtc.h" + +/** + * drm_mode_debug_printmodeline - debug print a mode + * @dev: DRM device + * @mode: mode to print + * + * LOCKING: + * None. + * + * Describe @mode using DRM_DEBUG. + */ +void drm_mode_debug_printmodeline(struct drm_display_mode *mode) +{ + DRM_DEBUG_KMS("Modeline %d:\"%s\" %d %d %d %d %d %d %d %d %d %d " + "0x%x 0x%x\n", + mode->base.id, mode->name, mode->vrefresh, mode->clock, + mode->hdisplay, mode->hsync_start, + mode->hsync_end, mode->htotal, + mode->vdisplay, mode->vsync_start, + mode->vsync_end, mode->vtotal, mode->type, mode->flags); +} +EXPORT_SYMBOL(drm_mode_debug_printmodeline); + +/** + * drm_cvt_mode -create a modeline based on CVT algorithm + * @dev: DRM device + * @hdisplay: hdisplay size + * @vdisplay: vdisplay size + * @vrefresh : vrefresh rate + * @reduced : Whether the GTF calculation is simplified + * @interlaced:Whether the interlace is supported + * + * LOCKING: + * none. + * + * return the modeline based on CVT algorithm + * + * This function is called to generate the modeline based on CVT algorithm + * according to the hdisplay, vdisplay, vrefresh. + * It is based from the VESA(TM) Coordinated Video Timing Generator by + * Graham Loveridge April 9, 2003 available at + * http://www.elo.utfsm.cl/~elo212/docs/CVTd6r1.xls + * + * And it is copied from xf86CVTmode in xserver/hw/xfree86/modes/xf86cvt.c. + * What I have done is to translate it by using integer calculation. + */ +#define HV_FACTOR 1000 +struct drm_display_mode *drm_cvt_mode(struct drm_device *dev, int hdisplay, + int vdisplay, int vrefresh, + bool reduced, bool interlaced, bool margins) +{ + /* 1) top/bottom margin size (% of height) - default: 1.8, */ +#define CVT_MARGIN_PERCENTAGE 18 + /* 2) character cell horizontal granularity (pixels) - default 8 */ +#define CVT_H_GRANULARITY 8 + /* 3) Minimum vertical porch (lines) - default 3 */ +#define CVT_MIN_V_PORCH 3 + /* 4) Minimum number of vertical back porch lines - default 6 */ +#define CVT_MIN_V_BPORCH 6 + /* Pixel Clock step (kHz) */ +#define CVT_CLOCK_STEP 250 + struct drm_display_mode *drm_mode; + unsigned int vfieldrate, hperiod; + int hdisplay_rnd, hmargin, vdisplay_rnd, vmargin, vsync; + int interlace; + + /* allocate the drm_display_mode structure. If failure, we will + * return directly + */ + drm_mode = drm_mode_create(dev); + if (!drm_mode) + return NULL; + + /* the CVT default refresh rate is 60Hz */ + if (!vrefresh) + vrefresh = 60; + + /* the required field fresh rate */ + if (interlaced) + vfieldrate = vrefresh * 2; + else + vfieldrate = vrefresh; + + /* horizontal pixels */ + hdisplay_rnd = hdisplay - (hdisplay % CVT_H_GRANULARITY); + + /* determine the left&right borders */ + hmargin = 0; + if (margins) { + hmargin = hdisplay_rnd * CVT_MARGIN_PERCENTAGE / 1000; + hmargin -= hmargin % CVT_H_GRANULARITY; + } + /* find the total active pixels */ + drm_mode->hdisplay = hdisplay_rnd + 2 * hmargin; + + /* find the number of lines per field */ + if (interlaced) + vdisplay_rnd = vdisplay / 2; + else + vdisplay_rnd = vdisplay; + + /* find the top & bottom borders */ + vmargin = 0; + if (margins) + vmargin = vdisplay_rnd * CVT_MARGIN_PERCENTAGE / 1000; + + drm_mode->vdisplay = vdisplay + 2 * vmargin; + + /* Interlaced */ + if (interlaced) + interlace = 1; + else + interlace = 0; + + /* Determine VSync Width from aspect ratio */ + if (!(vdisplay % 3) && ((vdisplay * 4 / 3) == hdisplay)) + vsync = 4; + else if (!(vdisplay % 9) && ((vdisplay * 16 / 9) == hdisplay)) + vsync = 5; + else if (!(vdisplay % 10) && ((vdisplay * 16 / 10) == hdisplay)) + vsync = 6; + else if (!(vdisplay % 4) && ((vdisplay * 5 / 4) == hdisplay)) + vsync = 7; + else if (!(vdisplay % 9) && ((vdisplay * 15 / 9) == hdisplay)) + vsync = 7; + else /* custom */ + vsync = 10; + + if (!reduced) { + /* simplify the GTF calculation */ + /* 4) Minimum time of vertical sync + back porch interval (µs) + * default 550.0 + */ + int tmp1, tmp2; +#define CVT_MIN_VSYNC_BP 550 + /* 3) Nominal HSync width (% of line period) - default 8 */ +#define CVT_HSYNC_PERCENTAGE 8 + unsigned int hblank_percentage; + int vsyncandback_porch, vback_porch, hblank; + + /* estimated the horizontal period */ + tmp1 = HV_FACTOR * 1000000 - + CVT_MIN_VSYNC_BP * HV_FACTOR * vfieldrate; + tmp2 = (vdisplay_rnd + 2 * vmargin + CVT_MIN_V_PORCH) * 2 + + interlace; + hperiod = tmp1 * 2 / (tmp2 * vfieldrate); + + tmp1 = CVT_MIN_VSYNC_BP * HV_FACTOR / hperiod + 1; + /* 9. Find number of lines in sync + backporch */ + if (tmp1 < (vsync + CVT_MIN_V_PORCH)) + vsyncandback_porch = vsync + CVT_MIN_V_PORCH; + else + vsyncandback_porch = tmp1; + /* 10. Find number of lines in back porch */ + vback_porch = vsyncandback_porch - vsync; + drm_mode->vtotal = vdisplay_rnd + 2 * vmargin + + vsyncandback_porch + CVT_MIN_V_PORCH; + /* 5) Definition of Horizontal blanking time limitation */ + /* Gradient (%/kHz) - default 600 */ +#define CVT_M_FACTOR 600 + /* Offset (%) - default 40 */ +#define CVT_C_FACTOR 40 + /* Blanking time scaling factor - default 128 */ +#define CVT_K_FACTOR 128 + /* Scaling factor weighting - default 20 */ +#define CVT_J_FACTOR 20 +#define CVT_M_PRIME (CVT_M_FACTOR * CVT_K_FACTOR / 256) +#define CVT_C_PRIME ((CVT_C_FACTOR - CVT_J_FACTOR) * CVT_K_FACTOR / 256 + \ + CVT_J_FACTOR) + /* 12. Find ideal blanking duty cycle from formula */ + hblank_percentage = CVT_C_PRIME * HV_FACTOR - CVT_M_PRIME * + hperiod / 1000; + /* 13. Blanking time */ + if (hblank_percentage < 20 * HV_FACTOR) + hblank_percentage = 20 * HV_FACTOR; + hblank = drm_mode->hdisplay * hblank_percentage / + (100 * HV_FACTOR - hblank_percentage); + hblank -= hblank % (2 * CVT_H_GRANULARITY); + /* 14. find the total pixes per line */ + drm_mode->htotal = drm_mode->hdisplay + hblank; + drm_mode->hsync_end = drm_mode->hdisplay + hblank / 2; + drm_mode->hsync_start = drm_mode->hsync_end - + (drm_mode->htotal * CVT_HSYNC_PERCENTAGE) / 100; + drm_mode->hsync_start += CVT_H_GRANULARITY - + drm_mode->hsync_start % CVT_H_GRANULARITY; + /* fill the Vsync values */ + drm_mode->vsync_start = drm_mode->vdisplay + CVT_MIN_V_PORCH; + drm_mode->vsync_end = drm_mode->vsync_start + vsync; + } else { + /* Reduced blanking */ + /* Minimum vertical blanking interval time (µs)- default 460 */ +#define CVT_RB_MIN_VBLANK 460 + /* Fixed number of clocks for horizontal sync */ +#define CVT_RB_H_SYNC 32 + /* Fixed number of clocks for horizontal blanking */ +#define CVT_RB_H_BLANK 160 + /* Fixed number of lines for vertical front porch - default 3*/ +#define CVT_RB_VFPORCH 3 + int vbilines; + int tmp1, tmp2; + /* 8. Estimate Horizontal period. */ + tmp1 = HV_FACTOR * 1000000 - + CVT_RB_MIN_VBLANK * HV_FACTOR * vfieldrate; + tmp2 = vdisplay_rnd + 2 * vmargin; + hperiod = tmp1 / (tmp2 * vfieldrate); + /* 9. Find number of lines in vertical blanking */ + vbilines = CVT_RB_MIN_VBLANK * HV_FACTOR / hperiod + 1; + /* 10. Check if vertical blanking is sufficient */ + if (vbilines < (CVT_RB_VFPORCH + vsync + CVT_MIN_V_BPORCH)) + vbilines = CVT_RB_VFPORCH + vsync + CVT_MIN_V_BPORCH; + /* 11. Find total number of lines in vertical field */ + drm_mode->vtotal = vdisplay_rnd + 2 * vmargin + vbilines; + /* 12. Find total number of pixels in a line */ + drm_mode->htotal = drm_mode->hdisplay + CVT_RB_H_BLANK; + /* Fill in HSync values */ + drm_mode->hsync_end = drm_mode->hdisplay + CVT_RB_H_BLANK / 2; + drm_mode->hsync_start = drm_mode->hsync_end - CVT_RB_H_SYNC; + /* Fill in VSync values */ + drm_mode->vsync_start = drm_mode->vdisplay + CVT_RB_VFPORCH; + drm_mode->vsync_end = drm_mode->vsync_start + vsync; + } + /* 15/13. Find pixel clock frequency (kHz for xf86) */ + drm_mode->clock = drm_mode->htotal * HV_FACTOR * 1000 / hperiod; + drm_mode->clock -= drm_mode->clock % CVT_CLOCK_STEP; + /* 18/16. Find actual vertical frame frequency */ + /* ignore - just set the mode flag for interlaced */ + if (interlaced) { + drm_mode->vtotal *= 2; + drm_mode->flags |= DRM_MODE_FLAG_INTERLACE; + } + /* Fill the mode line name */ + drm_mode_set_name(drm_mode); + if (reduced) + drm_mode->flags |= (DRM_MODE_FLAG_PHSYNC | + DRM_MODE_FLAG_NVSYNC); + else + drm_mode->flags |= (DRM_MODE_FLAG_PVSYNC | + DRM_MODE_FLAG_NHSYNC); + + return drm_mode; +} +EXPORT_SYMBOL(drm_cvt_mode); + +/** + * drm_gtf_mode_complex - create the modeline based on full GTF algorithm + * + * @dev :drm device + * @hdisplay :hdisplay size + * @vdisplay :vdisplay size + * @vrefresh :vrefresh rate. + * @interlaced :whether the interlace is supported + * @margins :desired margin size + * @GTF_[MCKJ] :extended GTF formula parameters + * + * LOCKING. + * none. + * + * return the modeline based on full GTF algorithm. + * + * GTF feature blocks specify C and J in multiples of 0.5, so we pass them + * in here multiplied by two. For a C of 40, pass in 80. + */ +struct drm_display_mode * +drm_gtf_mode_complex(struct drm_device *dev, int hdisplay, int vdisplay, + int vrefresh, bool interlaced, int margins, + int GTF_M, int GTF_2C, int GTF_K, int GTF_2J) +{ /* 1) top/bottom margin size (% of height) - default: 1.8, */ +#define GTF_MARGIN_PERCENTAGE 18 + /* 2) character cell horizontal granularity (pixels) - default 8 */ +#define GTF_CELL_GRAN 8 + /* 3) Minimum vertical porch (lines) - default 3 */ +#define GTF_MIN_V_PORCH 1 + /* width of vsync in lines */ +#define V_SYNC_RQD 3 + /* width of hsync as % of total line */ +#define H_SYNC_PERCENT 8 + /* min time of vsync + back porch (microsec) */ +#define MIN_VSYNC_PLUS_BP 550 + /* C' and M' are part of the Blanking Duty Cycle computation */ +#define GTF_C_PRIME ((((GTF_2C - GTF_2J) * GTF_K / 256) + GTF_2J) / 2) +#define GTF_M_PRIME (GTF_K * GTF_M / 256) + struct drm_display_mode *drm_mode; + unsigned int hdisplay_rnd, vdisplay_rnd, vfieldrate_rqd; + int top_margin, bottom_margin; + int interlace; + unsigned int hfreq_est; + int vsync_plus_bp, vback_porch; + unsigned int vtotal_lines, vfieldrate_est, hperiod; + unsigned int vfield_rate, vframe_rate; + int left_margin, right_margin; + unsigned int total_active_pixels, ideal_duty_cycle; + unsigned int hblank, total_pixels, pixel_freq; + int hsync, hfront_porch, vodd_front_porch_lines; + unsigned int tmp1, tmp2; + + drm_mode = drm_mode_create(dev); + if (!drm_mode) + return NULL; + + /* 1. In order to give correct results, the number of horizontal + * pixels requested is first processed to ensure that it is divisible + * by the character size, by rounding it to the nearest character + * cell boundary: + */ + hdisplay_rnd = (hdisplay + GTF_CELL_GRAN / 2) / GTF_CELL_GRAN; + hdisplay_rnd = hdisplay_rnd * GTF_CELL_GRAN; + + /* 2. If interlace is requested, the number of vertical lines assumed + * by the calculation must be halved, as the computation calculates + * the number of vertical lines per field. + */ + if (interlaced) + vdisplay_rnd = vdisplay / 2; + else + vdisplay_rnd = vdisplay; + + /* 3. Find the frame rate required: */ + if (interlaced) + vfieldrate_rqd = vrefresh * 2; + else + vfieldrate_rqd = vrefresh; + + /* 4. Find number of lines in Top margin: */ + top_margin = 0; + if (margins) + top_margin = (vdisplay_rnd * GTF_MARGIN_PERCENTAGE + 500) / + 1000; + /* 5. Find number of lines in bottom margin: */ + bottom_margin = top_margin; + + /* 6. If interlace is required, then set variable interlace: */ + if (interlaced) + interlace = 1; + else + interlace = 0; + + /* 7. Estimate the Horizontal frequency */ + { + tmp1 = (1000000 - MIN_VSYNC_PLUS_BP * vfieldrate_rqd) / 500; + tmp2 = (vdisplay_rnd + 2 * top_margin + GTF_MIN_V_PORCH) * + 2 + interlace; + hfreq_est = (tmp2 * 1000 * vfieldrate_rqd) / tmp1; + } + + /* 8. Find the number of lines in V sync + back porch */ + /* [V SYNC+BP] = RINT(([MIN VSYNC+BP] * hfreq_est / 1000000)) */ + vsync_plus_bp = MIN_VSYNC_PLUS_BP * hfreq_est / 1000; + vsync_plus_bp = (vsync_plus_bp + 500) / 1000; + /* 9. Find the number of lines in V back porch alone: */ + vback_porch = vsync_plus_bp - V_SYNC_RQD; + /* 10. Find the total number of lines in Vertical field period: */ + vtotal_lines = vdisplay_rnd + top_margin + bottom_margin + + vsync_plus_bp + GTF_MIN_V_PORCH; + /* 11. Estimate the Vertical field frequency: */ + vfieldrate_est = hfreq_est / vtotal_lines; + /* 12. Find the actual horizontal period: */ + hperiod = 1000000 / (vfieldrate_rqd * vtotal_lines); + + /* 13. Find the actual Vertical field frequency: */ + vfield_rate = hfreq_est / vtotal_lines; + /* 14. Find the Vertical frame frequency: */ + if (interlaced) + vframe_rate = vfield_rate / 2; + else + vframe_rate = vfield_rate; + /* 15. Find number of pixels in left margin: */ + if (margins) + left_margin = (hdisplay_rnd * GTF_MARGIN_PERCENTAGE + 500) / + 1000; + else + left_margin = 0; + + /* 16.Find number of pixels in right margin: */ + right_margin = left_margin; + /* 17.Find total number of active pixels in image and left and right */ + total_active_pixels = hdisplay_rnd + left_margin + right_margin; + /* 18.Find the ideal blanking duty cycle from blanking duty cycle */ + ideal_duty_cycle = GTF_C_PRIME * 1000 - + (GTF_M_PRIME * 1000000 / hfreq_est); + /* 19.Find the number of pixels in the blanking time to the nearest + * double character cell: */ + hblank = total_active_pixels * ideal_duty_cycle / + (100000 - ideal_duty_cycle); + hblank = (hblank + GTF_CELL_GRAN) / (2 * GTF_CELL_GRAN); + hblank = hblank * 2 * GTF_CELL_GRAN; + /* 20.Find total number of pixels: */ + total_pixels = total_active_pixels + hblank; + /* 21.Find pixel clock frequency: */ + pixel_freq = total_pixels * hfreq_est / 1000; + /* Stage 1 computations are now complete; I should really pass + * the results to another function and do the Stage 2 computations, + * but I only need a few more values so I'll just append the + * computations here for now */ + /* 17. Find the number of pixels in the horizontal sync period: */ + hsync = H_SYNC_PERCENT * total_pixels / 100; + hsync = (hsync + GTF_CELL_GRAN / 2) / GTF_CELL_GRAN; + hsync = hsync * GTF_CELL_GRAN; + /* 18. Find the number of pixels in horizontal front porch period */ + hfront_porch = hblank / 2 - hsync; + /* 36. Find the number of lines in the odd front porch period: */ + vodd_front_porch_lines = GTF_MIN_V_PORCH ; + + /* finally, pack the results in the mode struct */ + drm_mode->hdisplay = hdisplay_rnd; + drm_mode->hsync_start = hdisplay_rnd + hfront_porch; + drm_mode->hsync_end = drm_mode->hsync_start + hsync; + drm_mode->htotal = total_pixels; + drm_mode->vdisplay = vdisplay_rnd; + drm_mode->vsync_start = vdisplay_rnd + vodd_front_porch_lines; + drm_mode->vsync_end = drm_mode->vsync_start + V_SYNC_RQD; + drm_mode->vtotal = vtotal_lines; + + drm_mode->clock = pixel_freq; + + if (interlaced) { + drm_mode->vtotal *= 2; + drm_mode->flags |= DRM_MODE_FLAG_INTERLACE; + } + + drm_mode_set_name(drm_mode); + if (GTF_M == 600 && GTF_2C == 80 && GTF_K == 128 && GTF_2J == 40) + drm_mode->flags = DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_PVSYNC; + else + drm_mode->flags = DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_NVSYNC; + + return drm_mode; +} +EXPORT_SYMBOL(drm_gtf_mode_complex); + +/** + * drm_gtf_mode - create the modeline based on GTF algorithm + * + * @dev :drm device + * @hdisplay :hdisplay size + * @vdisplay :vdisplay size + * @vrefresh :vrefresh rate. + * @interlaced :whether the interlace is supported + * @margins :whether the margin is supported + * + * LOCKING. + * none. + * + * return the modeline based on GTF algorithm + * + * This function is to create the modeline based on the GTF algorithm. + * Generalized Timing Formula is derived from: + * GTF Spreadsheet by Andy Morrish (1/5/97) + * available at http://www.vesa.org + * + * And it is copied from the file of xserver/hw/xfree86/modes/xf86gtf.c. + * What I have done is to translate it by using integer calculation. + * I also refer to the function of fb_get_mode in the file of + * drivers/video/fbmon.c + * + * Standard GTF parameters: + * M = 600 + * C = 40 + * K = 128 + * J = 20 + */ +struct drm_display_mode * +drm_gtf_mode(struct drm_device *dev, int hdisplay, int vdisplay, int vrefresh, + bool lace, int margins) +{ + return drm_gtf_mode_complex(dev, hdisplay, vdisplay, vrefresh, lace, + margins, 600, 40 * 2, 128, 20 * 2); +} +EXPORT_SYMBOL(drm_gtf_mode); + +/** + * drm_mode_set_name - set the name on a mode + * @mode: name will be set in this mode + * + * LOCKING: + * None. + * + * Set the name of @mode to a standard format. + */ +void drm_mode_set_name(struct drm_display_mode *mode) +{ + bool interlaced = !!(mode->flags & DRM_MODE_FLAG_INTERLACE); + + snprintf(mode->name, DRM_DISPLAY_MODE_LEN, "%dx%d%s", + mode->hdisplay, mode->vdisplay, + interlaced ? "i" : ""); +} +EXPORT_SYMBOL(drm_mode_set_name); + +/** + * drm_mode_list_concat - move modes from one list to another + * @head: source list + * @new: dst list + * + * LOCKING: + * Caller must ensure both lists are locked. + * + * Move all the modes from @head to @new. + */ +void drm_mode_list_concat(struct list_head *head, struct list_head *new) +{ + + struct list_head *entry, *tmp; + + list_for_each_safe(entry, tmp, head) { + list_move_tail(entry, new); + } +} +EXPORT_SYMBOL(drm_mode_list_concat); + +/** + * drm_mode_width - get the width of a mode + * @mode: mode + * + * LOCKING: + * None. + * + * Return @mode's width (hdisplay) value. + * + * FIXME: is this needed? + * + * RETURNS: + * @mode->hdisplay + */ +int drm_mode_width(struct drm_display_mode *mode) +{ + return mode->hdisplay; + +} +EXPORT_SYMBOL(drm_mode_width); + +/** + * drm_mode_height - get the height of a mode + * @mode: mode + * + * LOCKING: + * None. + * + * Return @mode's height (vdisplay) value. + * + * FIXME: is this needed? + * + * RETURNS: + * @mode->vdisplay + */ +int drm_mode_height(struct drm_display_mode *mode) +{ + return mode->vdisplay; +} +EXPORT_SYMBOL(drm_mode_height); + +/** drm_mode_hsync - get the hsync of a mode + * @mode: mode + * + * LOCKING: + * None. + * + * Return @modes's hsync rate in kHz, rounded to the nearest int. + */ +int drm_mode_hsync(const struct drm_display_mode *mode) +{ + unsigned int calc_val; + + if (mode->hsync) + return mode->hsync; + + if (mode->htotal < 0) + return 0; + + calc_val = (mode->clock * 1000) / mode->htotal; /* hsync in Hz */ + calc_val += 500; /* round to 1000Hz */ + calc_val /= 1000; /* truncate to kHz */ + + return calc_val; +} +EXPORT_SYMBOL(drm_mode_hsync); + +/** + * drm_mode_vrefresh - get the vrefresh of a mode + * @mode: mode + * + * LOCKING: + * None. + * + * Return @mode's vrefresh rate in Hz or calculate it if necessary. + * + * FIXME: why is this needed? shouldn't vrefresh be set already? + * + * RETURNS: + * Vertical refresh rate. It will be the result of actual value plus 0.5. + * If it is 70.288, it will return 70Hz. + * If it is 59.6, it will return 60Hz. + */ +int drm_mode_vrefresh(const struct drm_display_mode *mode) +{ + int refresh = 0; + unsigned int calc_val; + + if (mode->vrefresh > 0) + refresh = mode->vrefresh; + else if (mode->htotal > 0 && mode->vtotal > 0) { + int vtotal; + vtotal = mode->vtotal; + /* work out vrefresh the value will be x1000 */ + calc_val = (mode->clock * 1000); + calc_val /= mode->htotal; + refresh = (calc_val + vtotal / 2) / vtotal; + + if (mode->flags & DRM_MODE_FLAG_INTERLACE) + refresh *= 2; + if (mode->flags & DRM_MODE_FLAG_DBLSCAN) + refresh /= 2; + if (mode->vscan > 1) + refresh /= mode->vscan; + } + return refresh; +} +EXPORT_SYMBOL(drm_mode_vrefresh); + +/** + * drm_mode_set_crtcinfo - set CRTC modesetting parameters + * @p: mode + * @adjust_flags: unused? (FIXME) + * + * LOCKING: + * None. + * + * Setup the CRTC modesetting parameters for @p, adjusting if necessary. + */ +void drm_mode_set_crtcinfo(struct drm_display_mode *p, int adjust_flags) +{ + if ((p == NULL) || ((p->type & DRM_MODE_TYPE_CRTC_C) == DRM_MODE_TYPE_BUILTIN)) + return; + + p->crtc_hdisplay = p->hdisplay; + p->crtc_hsync_start = p->hsync_start; + p->crtc_hsync_end = p->hsync_end; + p->crtc_htotal = p->htotal; + p->crtc_hskew = p->hskew; + p->crtc_vdisplay = p->vdisplay; + p->crtc_vsync_start = p->vsync_start; + p->crtc_vsync_end = p->vsync_end; + p->crtc_vtotal = p->vtotal; + + if (p->flags & DRM_MODE_FLAG_INTERLACE) { + if (adjust_flags & CRTC_INTERLACE_HALVE_V) { + p->crtc_vdisplay /= 2; + p->crtc_vsync_start /= 2; + p->crtc_vsync_end /= 2; + p->crtc_vtotal /= 2; + } + + p->crtc_vtotal |= 1; + } + + if (p->flags & DRM_MODE_FLAG_DBLSCAN) { + p->crtc_vdisplay *= 2; + p->crtc_vsync_start *= 2; + p->crtc_vsync_end *= 2; + p->crtc_vtotal *= 2; + } + + if (p->vscan > 1) { + p->crtc_vdisplay *= p->vscan; + p->crtc_vsync_start *= p->vscan; + p->crtc_vsync_end *= p->vscan; + p->crtc_vtotal *= p->vscan; + } + + p->crtc_vblank_start = min(p->crtc_vsync_start, p->crtc_vdisplay); + p->crtc_vblank_end = max(p->crtc_vsync_end, p->crtc_vtotal); + p->crtc_hblank_start = min(p->crtc_hsync_start, p->crtc_hdisplay); + p->crtc_hblank_end = max(p->crtc_hsync_end, p->crtc_htotal); + + p->crtc_hadjusted = false; + p->crtc_vadjusted = false; +} +EXPORT_SYMBOL(drm_mode_set_crtcinfo); + + +/** + * drm_mode_duplicate - allocate and duplicate an existing mode + * @m: mode to duplicate + * + * LOCKING: + * None. + * + * Just allocate a new mode, copy the existing mode into it, and return + * a pointer to it. Used to create new instances of established modes. + */ +struct drm_display_mode *drm_mode_duplicate(struct drm_device *dev, + const struct drm_display_mode *mode) +{ + struct drm_display_mode *nmode; + int new_id; + + nmode = drm_mode_create(dev); + if (!nmode) + return NULL; + + new_id = nmode->base.id; + *nmode = *mode; + nmode->base.id = new_id; + INIT_LIST_HEAD(&nmode->head); + return nmode; +} +EXPORT_SYMBOL(drm_mode_duplicate); + +/** + * drm_mode_equal - test modes for equality + * @mode1: first mode + * @mode2: second mode + * + * LOCKING: + * None. + * + * Check to see if @mode1 and @mode2 are equivalent. + * + * RETURNS: + * True if the modes are equal, false otherwise. + */ +bool drm_mode_equal(struct drm_display_mode *mode1, struct drm_display_mode *mode2) +{ + /* do clock check convert to PICOS so fb modes get matched + * the same */ + if (mode1->clock && mode2->clock) { + if (KHZ2PICOS(mode1->clock) != KHZ2PICOS(mode2->clock)) + return false; + } else if (mode1->clock != mode2->clock) + return false; + + if (mode1->hdisplay == mode2->hdisplay && + mode1->hsync_start == mode2->hsync_start && + mode1->hsync_end == mode2->hsync_end && + mode1->htotal == mode2->htotal && + mode1->hskew == mode2->hskew && + mode1->vdisplay == mode2->vdisplay && + mode1->vsync_start == mode2->vsync_start && + mode1->vsync_end == mode2->vsync_end && + mode1->vtotal == mode2->vtotal && + mode1->vscan == mode2->vscan && + mode1->flags == mode2->flags) + return true; + + return false; +} +EXPORT_SYMBOL(drm_mode_equal); + +/** + * drm_mode_validate_size - make sure modes adhere to size constraints + * @dev: DRM device + * @mode_list: list of modes to check + * @maxX: maximum width + * @maxY: maximum height + * @maxPitch: max pitch + * + * LOCKING: + * Caller must hold a lock protecting @mode_list. + * + * The DRM device (@dev) has size and pitch limits. Here we validate the + * modes we probed for @dev against those limits and set their status as + * necessary. + */ +void drm_mode_validate_size(struct drm_device *dev, + struct list_head *mode_list, + int maxX, int maxY, int maxPitch) +{ + struct drm_display_mode *mode; + + list_for_each_entry(mode, mode_list, head) { + if (maxPitch > 0 && mode->hdisplay > maxPitch) + mode->status = MODE_BAD_WIDTH; + + if (maxX > 0 && mode->hdisplay > maxX) + mode->status = MODE_VIRTUAL_X; + + if (maxY > 0 && mode->vdisplay > maxY) + mode->status = MODE_VIRTUAL_Y; + } +} +EXPORT_SYMBOL(drm_mode_validate_size); + +/** + * drm_mode_validate_clocks - validate modes against clock limits + * @dev: DRM device + * @mode_list: list of modes to check + * @min: minimum clock rate array + * @max: maximum clock rate array + * @n_ranges: number of clock ranges (size of arrays) + * + * LOCKING: + * Caller must hold a lock protecting @mode_list. + * + * Some code may need to check a mode list against the clock limits of the + * device in question. This function walks the mode list, testing to make + * sure each mode falls within a given range (defined by @min and @max + * arrays) and sets @mode->status as needed. + */ +void drm_mode_validate_clocks(struct drm_device *dev, + struct list_head *mode_list, + int *min, int *max, int n_ranges) +{ + struct drm_display_mode *mode; + int i; + + list_for_each_entry(mode, mode_list, head) { + bool good = false; + for (i = 0; i < n_ranges; i++) { + if (mode->clock >= min[i] && mode->clock <= max[i]) { + good = true; + break; + } + } + if (!good) + mode->status = MODE_CLOCK_RANGE; + } +} +EXPORT_SYMBOL(drm_mode_validate_clocks); + +/** + * drm_mode_prune_invalid - remove invalid modes from mode list + * @dev: DRM device + * @mode_list: list of modes to check + * @verbose: be verbose about it + * + * LOCKING: + * Caller must hold a lock protecting @mode_list. + * + * Once mode list generation is complete, a caller can use this routine to + * remove invalid modes from a mode list. If any of the modes have a + * status other than %MODE_OK, they are removed from @mode_list and freed. + */ +void drm_mode_prune_invalid(struct drm_device *dev, + struct list_head *mode_list, bool verbose) +{ + struct drm_display_mode *mode, *t; + + list_for_each_entry_safe(mode, t, mode_list, head) { + if (mode->status != MODE_OK) { + list_del(&mode->head); + if (verbose) { + drm_mode_debug_printmodeline(mode); + DRM_DEBUG_KMS("Not using %s mode %d\n", + mode->name, mode->status); + } + drm_mode_destroy(dev, mode); + } + } +} +EXPORT_SYMBOL(drm_mode_prune_invalid); + +/** + * drm_mode_compare - compare modes for favorability + * @priv: unused + * @lh_a: list_head for first mode + * @lh_b: list_head for second mode + * + * LOCKING: + * None. + * + * Compare two modes, given by @lh_a and @lh_b, returning a value indicating + * which is better. + * + * RETURNS: + * Negative if @lh_a is better than @lh_b, zero if they're equivalent, or + * positive if @lh_b is better than @lh_a. + */ +static int drm_mode_compare(void *priv, struct list_head *lh_a, struct list_head *lh_b) +{ + struct drm_display_mode *a = list_entry(lh_a, struct drm_display_mode, head); + struct drm_display_mode *b = list_entry(lh_b, struct drm_display_mode, head); + int diff; + + diff = ((b->type & DRM_MODE_TYPE_PREFERRED) != 0) - + ((a->type & DRM_MODE_TYPE_PREFERRED) != 0); + if (diff) + return diff; + diff = b->hdisplay * b->vdisplay - a->hdisplay * a->vdisplay; + if (diff) + return diff; + diff = b->clock - a->clock; + return diff; +} + +/** + * drm_mode_sort - sort mode list + * @mode_list: list to sort + * + * LOCKING: + * Caller must hold a lock protecting @mode_list. + * + * Sort @mode_list by favorability, putting good modes first. + */ +void drm_mode_sort(struct list_head *mode_list) +{ + list_sort(NULL, mode_list, drm_mode_compare); +} +EXPORT_SYMBOL(drm_mode_sort); + +/** + * drm_mode_connector_list_update - update the mode list for the connector + * @connector: the connector to update + * + * LOCKING: + * Caller must hold a lock protecting @mode_list. + * + * This moves the modes from the @connector probed_modes list + * to the actual mode list. It compares the probed mode against the current + * list and only adds different modes. All modes unverified after this point + * will be removed by the prune invalid modes. + */ +void drm_mode_connector_list_update(struct drm_connector *connector) +{ + struct drm_display_mode *mode; + struct drm_display_mode *pmode, *pt; + int found_it; + + list_for_each_entry_safe(pmode, pt, &connector->probed_modes, + head) { + found_it = 0; + /* go through current modes checking for the new probed mode */ + list_for_each_entry(mode, &connector->modes, head) { + if (drm_mode_equal(pmode, mode)) { + found_it = 1; + /* if equal delete the probed mode */ + mode->status = pmode->status; + /* Merge type bits together */ + mode->type |= pmode->type; + list_del(&pmode->head); + drm_mode_destroy(connector->dev, pmode); + break; + } + } + + if (!found_it) { + list_move_tail(&pmode->head, &connector->modes); + } + } +} +EXPORT_SYMBOL(drm_mode_connector_list_update); + +/** + * drm_mode_parse_command_line_for_connector - parse command line for connector + * @mode_option - per connector mode option + * @connector - connector to parse line for + * + * This parses the connector specific then generic command lines for + * modes and options to configure the connector. + * + * This uses the same parameters as the fb modedb.c, except for extra + * x[M][R][-][@][i][m][eDd] + * + * enable/enable Digital/disable bit at the end + */ +bool drm_mode_parse_command_line_for_connector(const char *mode_option, + struct drm_connector *connector, + struct drm_cmdline_mode *mode) +{ + const char *name; + unsigned int namelen; + int res_specified = 0, bpp_specified = 0, refresh_specified = 0; + unsigned int xres = 0, yres = 0, bpp = 32, refresh = 0; + int yres_specified = 0, cvt = 0, rb = 0, interlace = 0, margins = 0; + int i; + enum drm_connector_force force = DRM_FORCE_UNSPECIFIED; + +#ifdef CONFIG_FB + if (!mode_option) + mode_option = fb_mode_option; +#endif + + if (!mode_option) { + mode->specified = false; + return false; + } + + name = mode_option; + namelen = strlen(name); + for (i = namelen-1; i >= 0; i--) { + switch (name[i]) { + case '@': + namelen = i; + if (!refresh_specified && !bpp_specified && + !yres_specified) { + refresh = simple_strtol(&name[i+1], NULL, 10); + refresh_specified = 1; + if (cvt || rb) + cvt = 0; + } else + goto done; + break; + case '-': + namelen = i; + if (!bpp_specified && !yres_specified) { + bpp = simple_strtol(&name[i+1], NULL, 10); + bpp_specified = 1; + if (cvt || rb) + cvt = 0; + } else + goto done; + break; + case 'x': + if (!yres_specified) { + yres = simple_strtol(&name[i+1], NULL, 10); + yres_specified = 1; + } else + goto done; + case '0' ... '9': + break; + case 'M': + if (!yres_specified) + cvt = 1; + break; + case 'R': + if (cvt) + rb = 1; + break; + case 'm': + if (!cvt) + margins = 1; + break; + case 'i': + if (!cvt) + interlace = 1; + break; + case 'e': + force = DRM_FORCE_ON; + break; + case 'D': + if ((connector->connector_type != DRM_MODE_CONNECTOR_DVII) && + (connector->connector_type != DRM_MODE_CONNECTOR_HDMIB)) + force = DRM_FORCE_ON; + else + force = DRM_FORCE_ON_DIGITAL; + break; + case 'd': + force = DRM_FORCE_OFF; + break; + default: + goto done; + } + } + if (i < 0 && yres_specified) { + xres = simple_strtol(name, NULL, 10); + res_specified = 1; + } +done: + if (res_specified) { + mode->specified = true; + mode->xres = xres; + mode->yres = yres; + } + + if (refresh_specified) { + mode->refresh_specified = true; + mode->refresh = refresh; + } + + if (bpp_specified) { + mode->bpp_specified = true; + mode->bpp = bpp; + } + mode->rb = rb ? true : false; + mode->cvt = cvt ? true : false; + mode->interlace = interlace ? true : false; + mode->force = force; + + return true; +} +EXPORT_SYMBOL(drm_mode_parse_command_line_for_connector); + +struct drm_display_mode * +drm_mode_create_from_cmdline_mode(struct drm_device *dev, + struct drm_cmdline_mode *cmd) +{ + struct drm_display_mode *mode; + + if (cmd->cvt) + mode = drm_cvt_mode(dev, + cmd->xres, cmd->yres, + cmd->refresh_specified ? cmd->refresh : 60, + cmd->rb, cmd->interlace, + cmd->margins); + else + mode = drm_gtf_mode(dev, + cmd->xres, cmd->yres, + cmd->refresh_specified ? cmd->refresh : 60, + cmd->interlace, + cmd->margins); + if (!mode) + return NULL; + + drm_mode_set_crtcinfo(mode, CRTC_INTERLACE_HALVE_V); + return mode; +} +EXPORT_SYMBOL(drm_mode_create_from_cmdline_mode); diff --git a/drivers/gpu/drm/drm_pci.c b/drivers/gpu/drm/drm_pci.c new file mode 100644 index 00000000..b6a19cb0 --- /dev/null +++ b/drivers/gpu/drm/drm_pci.c @@ -0,0 +1,468 @@ +/* drm_pci.h -- PCI DMA memory management wrappers for DRM -*- linux-c -*- */ +/** + * \file drm_pci.c + * \brief Functions and ioctls to manage PCI memory + * + * \warning These interfaces aren't stable yet. + * + * \todo Implement the remaining ioctl's for the PCI pools. + * \todo The wrappers here are so thin that they would be better off inlined.. + * + * \author José Fonseca + * \author Leif Delgass + */ + +/* + * Copyright 2003 José Fonseca. + * Copyright 2003 Leif Delgass. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN + * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + */ + +#include +#include +#include +#include "drmP.h" + +/**********************************************************************/ +/** \name PCI memory */ +/*@{*/ + +/** + * \brief Allocate a PCI consistent memory block, for DMA. + */ +drm_dma_handle_t *drm_pci_alloc(struct drm_device * dev, size_t size, size_t align) +{ + drm_dma_handle_t *dmah; +#if 1 + unsigned long addr; + size_t sz; +#endif + + /* pci_alloc_consistent only guarantees alignment to the smallest + * PAGE_SIZE order which is greater than or equal to the requested size. + * Return NULL here for now to make sure nobody tries for larger alignment + */ + if (align > size) + return NULL; + + dmah = kmalloc(sizeof(drm_dma_handle_t), GFP_KERNEL); + if (!dmah) + return NULL; + + dmah->size = size; + dmah->vaddr = dma_alloc_coherent(&dev->pdev->dev, size, &dmah->busaddr, GFP_KERNEL | __GFP_COMP); + + if (dmah->vaddr == NULL) { + kfree(dmah); + return NULL; + } + + memset(dmah->vaddr, 0, size); + + /* XXX - Is virt_to_page() legal for consistent mem? */ + /* Reserve */ + for (addr = (unsigned long)dmah->vaddr, sz = size; + sz > 0; addr += PAGE_SIZE, sz -= PAGE_SIZE) { + SetPageReserved(virt_to_page(addr)); + } + + return dmah; +} + +EXPORT_SYMBOL(drm_pci_alloc); + +/** + * \brief Free a PCI consistent memory block without freeing its descriptor. + * + * This function is for internal use in the Linux-specific DRM core code. + */ +void __drm_pci_free(struct drm_device * dev, drm_dma_handle_t * dmah) +{ +#if 1 + unsigned long addr; + size_t sz; +#endif + + if (dmah->vaddr) { + /* XXX - Is virt_to_page() legal for consistent mem? */ + /* Unreserve */ + for (addr = (unsigned long)dmah->vaddr, sz = dmah->size; + sz > 0; addr += PAGE_SIZE, sz -= PAGE_SIZE) { + ClearPageReserved(virt_to_page(addr)); + } + dma_free_coherent(&dev->pdev->dev, dmah->size, dmah->vaddr, + dmah->busaddr); + } +} + +/** + * \brief Free a PCI consistent memory block + */ +void drm_pci_free(struct drm_device * dev, drm_dma_handle_t * dmah) +{ + __drm_pci_free(dev, dmah); + kfree(dmah); +} + +EXPORT_SYMBOL(drm_pci_free); + +#ifdef CONFIG_PCI + +static int drm_get_pci_domain(struct drm_device *dev) +{ +#ifndef __alpha__ + /* For historical reasons, drm_get_pci_domain() is busticated + * on most archs and has to remain so for userspace interface + * < 1.4, except on alpha which was right from the beginning + */ + if (dev->if_version < 0x10004) + return 0; +#endif /* __alpha__ */ + + return pci_domain_nr(dev->pdev->bus); +} + +static int drm_pci_get_irq(struct drm_device *dev) +{ + return dev->pdev->irq; +} + +static const char *drm_pci_get_name(struct drm_device *dev) +{ + struct pci_driver *pdriver = dev->driver->kdriver.pci; + return pdriver->name; +} + +int drm_pci_set_busid(struct drm_device *dev, struct drm_master *master) +{ + int len, ret; + struct pci_driver *pdriver = dev->driver->kdriver.pci; + master->unique_len = 40; + master->unique_size = master->unique_len; + master->unique = kmalloc(master->unique_size, GFP_KERNEL); + if (master->unique == NULL) + return -ENOMEM; + + + len = snprintf(master->unique, master->unique_len, + "pci:%04x:%02x:%02x.%d", + drm_get_pci_domain(dev), + dev->pdev->bus->number, + PCI_SLOT(dev->pdev->devfn), + PCI_FUNC(dev->pdev->devfn)); + + if (len >= master->unique_len) { + DRM_ERROR("buffer overflow"); + ret = -EINVAL; + goto err; + } else + master->unique_len = len; + + dev->devname = + kmalloc(strlen(pdriver->name) + + master->unique_len + 2, GFP_KERNEL); + + if (dev->devname == NULL) { + ret = -ENOMEM; + goto err; + } + + sprintf(dev->devname, "%s@%s", pdriver->name, + master->unique); + + return 0; +err: + return ret; +} + +int drm_pci_set_unique(struct drm_device *dev, + struct drm_master *master, + struct drm_unique *u) +{ + int domain, bus, slot, func, ret; + const char *bus_name; + + master->unique_len = u->unique_len; + master->unique_size = u->unique_len + 1; + master->unique = kmalloc(master->unique_size, GFP_KERNEL); + if (!master->unique) { + ret = -ENOMEM; + goto err; + } + + if (copy_from_user(master->unique, u->unique, master->unique_len)) { + ret = -EFAULT; + goto err; + } + + master->unique[master->unique_len] = '\0'; + + bus_name = dev->driver->bus->get_name(dev); + dev->devname = kmalloc(strlen(bus_name) + + strlen(master->unique) + 2, GFP_KERNEL); + if (!dev->devname) { + ret = -ENOMEM; + goto err; + } + + sprintf(dev->devname, "%s@%s", bus_name, + master->unique); + + /* Return error if the busid submitted doesn't match the device's actual + * busid. + */ + ret = sscanf(master->unique, "PCI:%d:%d:%d", &bus, &slot, &func); + if (ret != 3) { + ret = -EINVAL; + goto err; + } + + domain = bus >> 8; + bus &= 0xff; + + if ((domain != drm_get_pci_domain(dev)) || + (bus != dev->pdev->bus->number) || + (slot != PCI_SLOT(dev->pdev->devfn)) || + (func != PCI_FUNC(dev->pdev->devfn))) { + ret = -EINVAL; + goto err; + } + return 0; +err: + return ret; +} + + +static int drm_pci_irq_by_busid(struct drm_device *dev, struct drm_irq_busid *p) +{ + if ((p->busnum >> 8) != drm_get_pci_domain(dev) || + (p->busnum & 0xff) != dev->pdev->bus->number || + p->devnum != PCI_SLOT(dev->pdev->devfn) || p->funcnum != PCI_FUNC(dev->pdev->devfn)) + return -EINVAL; + + p->irq = dev->pdev->irq; + + DRM_DEBUG("%d:%d:%d => IRQ %d\n", p->busnum, p->devnum, p->funcnum, + p->irq); + return 0; +} + +int drm_pci_agp_init(struct drm_device *dev) +{ + if (drm_core_has_AGP(dev)) { + if (drm_pci_device_is_agp(dev)) + dev->agp = drm_agp_init(dev); + if (drm_core_check_feature(dev, DRIVER_REQUIRE_AGP) + && (dev->agp == NULL)) { + DRM_ERROR("Cannot initialize the agpgart module.\n"); + return -EINVAL; + } + if (drm_core_has_MTRR(dev)) { + if (dev->agp) + dev->agp->agp_mtrr = + mtrr_add(dev->agp->agp_info.aper_base, + dev->agp->agp_info.aper_size * + 1024 * 1024, MTRR_TYPE_WRCOMB, 1); + } + } + return 0; +} + +static struct drm_bus drm_pci_bus = { + .bus_type = DRIVER_BUS_PCI, + .get_irq = drm_pci_get_irq, + .get_name = drm_pci_get_name, + .set_busid = drm_pci_set_busid, + .set_unique = drm_pci_set_unique, + .irq_by_busid = drm_pci_irq_by_busid, + .agp_init = drm_pci_agp_init, +}; + +/** + * Register. + * + * \param pdev - PCI device structure + * \param ent entry from the PCI ID table with device type flags + * \return zero on success or a negative number on failure. + * + * Attempt to gets inter module "drm" information. If we are first + * then register the character device and inter module information. + * Try and register, if we fail to register, backout previous work. + */ +int drm_get_pci_dev(struct pci_dev *pdev, const struct pci_device_id *ent, + struct drm_driver *driver) +{ + struct drm_device *dev; + int ret; + + DRM_DEBUG("\n"); + + dev = kzalloc(sizeof(*dev), GFP_KERNEL); + if (!dev) + return -ENOMEM; + + ret = pci_enable_device(pdev); + if (ret) + goto err_g1; + + pci_set_master(pdev); + + dev->pdev = pdev; + dev->dev = &pdev->dev; + + dev->pci_device = pdev->device; + dev->pci_vendor = pdev->vendor; + +#ifdef __alpha__ + dev->hose = pdev->sysdata; +#endif + + mutex_lock(&drm_global_mutex); + + if ((ret = drm_fill_in_dev(dev, ent, driver))) { + printk(KERN_ERR "DRM: Fill_in_dev failed.\n"); + goto err_g2; + } + + if (drm_core_check_feature(dev, DRIVER_MODESET)) { + pci_set_drvdata(pdev, dev); + ret = drm_get_minor(dev, &dev->control, DRM_MINOR_CONTROL); + if (ret) + goto err_g2; + } + + if ((ret = drm_get_minor(dev, &dev->primary, DRM_MINOR_LEGACY))) + goto err_g3; + + if (dev->driver->load) { + ret = dev->driver->load(dev, ent->driver_data); + if (ret) + goto err_g4; + } + + /* setup the grouping for the legacy output */ + if (drm_core_check_feature(dev, DRIVER_MODESET)) { + ret = drm_mode_group_init_legacy_group(dev, + &dev->primary->mode_group); + if (ret) + goto err_g4; + } + + list_add_tail(&dev->driver_item, &driver->device_list); + + DRM_INFO("Initialized %s %d.%d.%d %s for %s on minor %d\n", + driver->name, driver->major, driver->minor, driver->patchlevel, + driver->date, pci_name(pdev), dev->primary->index); + + mutex_unlock(&drm_global_mutex); + return 0; + +err_g4: + drm_put_minor(&dev->primary); +err_g3: + if (drm_core_check_feature(dev, DRIVER_MODESET)) + drm_put_minor(&dev->control); +err_g2: + pci_disable_device(pdev); +err_g1: + kfree(dev); + mutex_unlock(&drm_global_mutex); + return ret; +} +EXPORT_SYMBOL(drm_get_pci_dev); + +/** + * PCI device initialization. Called direct from modules at load time. + * + * \return zero on success or a negative number on failure. + * + * Initializes a drm_device structures,registering the + * stubs and initializing the AGP device. + * + * Expands the \c DRIVER_PREINIT and \c DRIVER_POST_INIT macros before and + * after the initialization for driver customization. + */ +int drm_pci_init(struct drm_driver *driver, struct pci_driver *pdriver) +{ + struct pci_dev *pdev = NULL; + const struct pci_device_id *pid; + int i; + + DRM_DEBUG("\n"); + + INIT_LIST_HEAD(&driver->device_list); + driver->kdriver.pci = pdriver; + driver->bus = &drm_pci_bus; + + if (driver->driver_features & DRIVER_MODESET) + return pci_register_driver(pdriver); + + /* If not using KMS, fall back to stealth mode manual scanning. */ + for (i = 0; pdriver->id_table[i].vendor != 0; i++) { + pid = &pdriver->id_table[i]; + + /* Loop around setting up a DRM device for each PCI device + * matching our ID and device class. If we had the internal + * function that pci_get_subsys and pci_get_class used, we'd + * be able to just pass pid in instead of doing a two-stage + * thing. + */ + pdev = NULL; + while ((pdev = + pci_get_subsys(pid->vendor, pid->device, pid->subvendor, + pid->subdevice, pdev)) != NULL) { + if ((pdev->class & pid->class_mask) != pid->class) + continue; + + /* stealth mode requires a manual probe */ + pci_dev_get(pdev); + drm_get_pci_dev(pdev, pid, driver); + } + } + return 0; +} + +#else + +int drm_pci_init(struct drm_driver *driver, struct pci_driver *pdriver) +{ + return -1; +} + +#endif + +EXPORT_SYMBOL(drm_pci_init); + +/*@}*/ +void drm_pci_exit(struct drm_driver *driver, struct pci_driver *pdriver) +{ + struct drm_device *dev, *tmp; + DRM_DEBUG("\n"); + + if (driver->driver_features & DRIVER_MODESET) { + pci_unregister_driver(pdriver); + } else { + list_for_each_entry_safe(dev, tmp, &driver->device_list, driver_item) + drm_put_dev(dev); + } + DRM_INFO("Module unloaded\n"); +} +EXPORT_SYMBOL(drm_pci_exit); diff --git a/drivers/gpu/drm/drm_platform.c b/drivers/gpu/drm/drm_platform.c new file mode 100644 index 00000000..7223f06d --- /dev/null +++ b/drivers/gpu/drm/drm_platform.c @@ -0,0 +1,196 @@ +/* + * Derived from drm_pci.c + * + * Copyright 2003 José Fonseca. + * Copyright 2003 Leif Delgass. + * Copyright (c) 2009, Code Aurora Forum. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN + * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + */ + +#include "drmP.h" + +/** + * Register. + * + * \param platdev - Platform device struture + * \return zero on success or a negative number on failure. + * + * Attempt to gets inter module "drm" information. If we are first + * then register the character device and inter module information. + * Try and register, if we fail to register, backout previous work. + */ + +int drm_get_platform_dev(struct platform_device *platdev, + struct drm_driver *driver) +{ + struct drm_device *dev; + int ret; + + DRM_DEBUG("\n"); + + dev = kzalloc(sizeof(*dev), GFP_KERNEL); + if (!dev) + return -ENOMEM; + + dev->platformdev = platdev; + dev->dev = &platdev->dev; + + mutex_lock(&drm_global_mutex); + + ret = drm_fill_in_dev(dev, NULL, driver); + + if (ret) { + printk(KERN_ERR "DRM: Fill_in_dev failed.\n"); + goto err_g1; + } + + if (drm_core_check_feature(dev, DRIVER_MODESET)) { + dev_set_drvdata(&platdev->dev, dev); + ret = drm_get_minor(dev, &dev->control, DRM_MINOR_CONTROL); + if (ret) + goto err_g1; + } + + ret = drm_get_minor(dev, &dev->primary, DRM_MINOR_LEGACY); + if (ret) + goto err_g2; + + if (dev->driver->load) { + ret = dev->driver->load(dev, 0); + if (ret) + goto err_g3; + } + + /* setup the grouping for the legacy output */ + if (drm_core_check_feature(dev, DRIVER_MODESET)) { + ret = drm_mode_group_init_legacy_group(dev, + &dev->primary->mode_group); + if (ret) + goto err_g3; + } + + list_add_tail(&dev->driver_item, &driver->device_list); + + mutex_unlock(&drm_global_mutex); + + DRM_INFO("Initialized %s %d.%d.%d %s on minor %d\n", + driver->name, driver->major, driver->minor, driver->patchlevel, + driver->date, dev->primary->index); + + return 0; + +err_g3: + drm_put_minor(&dev->primary); +err_g2: + if (drm_core_check_feature(dev, DRIVER_MODESET)) + drm_put_minor(&dev->control); +err_g1: + kfree(dev); + mutex_unlock(&drm_global_mutex); + return ret; +} +EXPORT_SYMBOL(drm_get_platform_dev); + +static int drm_platform_get_irq(struct drm_device *dev) +{ + return platform_get_irq(dev->platformdev, 0); +} + +static const char *drm_platform_get_name(struct drm_device *dev) +{ + return dev->platformdev->name; +} + +static int drm_platform_set_busid(struct drm_device *dev, struct drm_master *master) +{ + int len, ret; + + master->unique_len = 10 + strlen(dev->platformdev->name); + master->unique = kmalloc(master->unique_len + 1, GFP_KERNEL); + + if (master->unique == NULL) + return -ENOMEM; + + len = snprintf(master->unique, master->unique_len, + "platform:%s", dev->platformdev->name); + + if (len > master->unique_len) { + DRM_ERROR("Unique buffer overflowed\n"); + ret = -EINVAL; + goto err; + } + + dev->devname = + kmalloc(strlen(dev->platformdev->name) + + master->unique_len + 2, GFP_KERNEL); + + if (dev->devname == NULL) { + ret = -ENOMEM; + goto err; + } + + sprintf(dev->devname, "%s@%s", dev->platformdev->name, + master->unique); + return 0; +err: + return ret; +} + +static struct drm_bus drm_platform_bus = { + .bus_type = DRIVER_BUS_PLATFORM, + .get_irq = drm_platform_get_irq, + .get_name = drm_platform_get_name, + .set_busid = drm_platform_set_busid, +}; + +/** + * Platform device initialization. Called direct from modules. + * + * \return zero on success or a negative number on failure. + * + * Initializes a drm_device structures,registering the + * stubs + * + * Expands the \c DRIVER_PREINIT and \c DRIVER_POST_INIT macros before and + * after the initialization for driver customization. + */ + +int drm_platform_init(struct drm_driver *driver, struct platform_device *platform_device) +{ + DRM_DEBUG("\n"); + + driver->kdriver.platform_device = platform_device; + driver->bus = &drm_platform_bus; + INIT_LIST_HEAD(&driver->device_list); + return drm_get_platform_dev(platform_device, driver); +} +EXPORT_SYMBOL(drm_platform_init); + +void drm_platform_exit(struct drm_driver *driver, struct platform_device *platform_device) +{ + struct drm_device *dev, *tmp; + DRM_DEBUG("\n"); + + list_for_each_entry_safe(dev, tmp, &driver->device_list, driver_item) + drm_put_dev(dev); + DRM_INFO("Module unloaded\n"); +} +EXPORT_SYMBOL(drm_platform_exit); diff --git a/drivers/gpu/drm/drm_proc.c b/drivers/gpu/drm/drm_proc.c new file mode 100644 index 00000000..9e5b07ef --- /dev/null +++ b/drivers/gpu/drm/drm_proc.c @@ -0,0 +1,221 @@ +/** + * \file drm_proc.c + * /proc support for DRM + * + * \author Rickard E. (Rik) Faith + * \author Gareth Hughes + * + * \par Acknowledgements: + * Matthew J Sottek sent in a patch to fix + * the problem with the proc files not outputting all their information. + */ + +/* + * Created: Mon Jan 11 09:48:47 1999 by faith@valinux.com + * + * Copyright 1999 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * VA LINUX SYSTEMS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#include +#include +#include "drmP.h" + +/*************************************************** + * Initialization, etc. + **************************************************/ + +/** + * Proc file list. + */ +static struct drm_info_list drm_proc_list[] = { + {"name", drm_name_info, 0}, + {"vm", drm_vm_info, 0}, + {"clients", drm_clients_info, 0}, + {"queues", drm_queues_info, 0}, + {"bufs", drm_bufs_info, 0}, + {"gem_names", drm_gem_name_info, DRIVER_GEM}, +#if DRM_DEBUG_CODE + {"vma", drm_vma_info, 0}, +#endif +}; +#define DRM_PROC_ENTRIES ARRAY_SIZE(drm_proc_list) + +static int drm_proc_open(struct inode *inode, struct file *file) +{ + struct drm_info_node* node = PDE(inode)->data; + + return single_open(file, node->info_ent->show, node); +} + +static const struct file_operations drm_proc_fops = { + .owner = THIS_MODULE, + .open = drm_proc_open, + .read = seq_read, + .llseek = seq_lseek, + .release = single_release, +}; + + +/** + * Initialize a given set of proc files for a device + * + * \param files The array of files to create + * \param count The number of files given + * \param root DRI proc dir entry. + * \param minor device minor number + * \return Zero on success, non-zero on failure + * + * Create a given set of proc files represented by an array of + * gdm_proc_lists in the given root directory. + */ +int drm_proc_create_files(struct drm_info_list *files, int count, + struct proc_dir_entry *root, struct drm_minor *minor) +{ + struct drm_device *dev = minor->dev; + struct proc_dir_entry *ent; + struct drm_info_node *tmp; + char name[64]; + int i, ret; + + for (i = 0; i < count; i++) { + u32 features = files[i].driver_features; + + if (features != 0 && + (dev->driver->driver_features & features) != features) + continue; + + tmp = kmalloc(sizeof(struct drm_info_node), GFP_KERNEL); + if (tmp == NULL) { + ret = -1; + goto fail; + } + tmp->minor = minor; + tmp->info_ent = &files[i]; + list_add(&tmp->list, &minor->proc_nodes.list); + + ent = proc_create_data(files[i].name, S_IRUGO, root, + &drm_proc_fops, tmp); + if (!ent) { + DRM_ERROR("Cannot create /proc/dri/%s/%s\n", + name, files[i].name); + list_del(&tmp->list); + kfree(tmp); + ret = -1; + goto fail; + } + + } + return 0; + +fail: + for (i = 0; i < count; i++) + remove_proc_entry(drm_proc_list[i].name, minor->proc_root); + return ret; +} + +/** + * Initialize the DRI proc filesystem for a device + * + * \param dev DRM device + * \param minor device minor number + * \param root DRI proc dir entry. + * \param dev_root resulting DRI device proc dir entry. + * \return root entry pointer on success, or NULL on failure. + * + * Create the DRI proc root entry "/proc/dri", the device proc root entry + * "/proc/dri/%minor%/", and each entry in proc_list as + * "/proc/dri/%minor%/%name%". + */ +int drm_proc_init(struct drm_minor *minor, int minor_id, + struct proc_dir_entry *root) +{ + char name[64]; + int ret; + + INIT_LIST_HEAD(&minor->proc_nodes.list); + sprintf(name, "%d", minor_id); + minor->proc_root = proc_mkdir(name, root); + if (!minor->proc_root) { + DRM_ERROR("Cannot create /proc/dri/%s\n", name); + return -1; + } + + ret = drm_proc_create_files(drm_proc_list, DRM_PROC_ENTRIES, + minor->proc_root, minor); + if (ret) { + remove_proc_entry(name, root); + minor->proc_root = NULL; + DRM_ERROR("Failed to create core drm proc files\n"); + return ret; + } + + return 0; +} + +int drm_proc_remove_files(struct drm_info_list *files, int count, + struct drm_minor *minor) +{ + struct list_head *pos, *q; + struct drm_info_node *tmp; + int i; + + for (i = 0; i < count; i++) { + list_for_each_safe(pos, q, &minor->proc_nodes.list) { + tmp = list_entry(pos, struct drm_info_node, list); + if (tmp->info_ent == &files[i]) { + remove_proc_entry(files[i].name, + minor->proc_root); + list_del(pos); + kfree(tmp); + } + } + } + return 0; +} + +/** + * Cleanup the proc filesystem resources. + * + * \param minor device minor number. + * \param root DRI proc dir entry. + * \param dev_root DRI device proc dir entry. + * \return always zero. + * + * Remove all proc entries created by proc_init(). + */ +int drm_proc_cleanup(struct drm_minor *minor, struct proc_dir_entry *root) +{ + char name[64]; + + if (!root || !minor->proc_root) + return 0; + + drm_proc_remove_files(drm_proc_list, DRM_PROC_ENTRIES, minor); + + sprintf(name, "%d", minor->index); + remove_proc_entry(name, root); + + return 0; +} + diff --git a/drivers/gpu/drm/drm_scatter.c b/drivers/gpu/drm/drm_scatter.c new file mode 100644 index 00000000..d15e09b0 --- /dev/null +++ b/drivers/gpu/drm/drm_scatter.c @@ -0,0 +1,217 @@ +/** + * \file drm_scatter.c + * IOCTLs to manage scatter/gather memory + * + * \author Gareth Hughes + */ + +/* + * Created: Mon Dec 18 23:20:54 2000 by gareth@valinux.com + * + * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + */ + +#include +#include +#include "drmP.h" + +#define DEBUG_SCATTER 0 + +static inline void *drm_vmalloc_dma(unsigned long size) +{ +#if defined(__powerpc__) && defined(CONFIG_NOT_COHERENT_CACHE) + return __vmalloc(size, GFP_KERNEL, PAGE_KERNEL | _PAGE_NO_CACHE); +#else + return vmalloc_32(size); +#endif +} + +void drm_sg_cleanup(struct drm_sg_mem * entry) +{ + struct page *page; + int i; + + for (i = 0; i < entry->pages; i++) { + page = entry->pagelist[i]; + if (page) + ClearPageReserved(page); + } + + vfree(entry->virtual); + + kfree(entry->busaddr); + kfree(entry->pagelist); + kfree(entry); +} + +#ifdef _LP64 +# define ScatterHandle(x) (unsigned int)((x >> 32) + (x & ((1L << 32) - 1))) +#else +# define ScatterHandle(x) (unsigned int)(x) +#endif + +int drm_sg_alloc(struct drm_device *dev, struct drm_scatter_gather * request) +{ + struct drm_sg_mem *entry; + unsigned long pages, i, j; + + DRM_DEBUG("\n"); + + if (!drm_core_check_feature(dev, DRIVER_SG)) + return -EINVAL; + + if (dev->sg) + return -EINVAL; + + entry = kmalloc(sizeof(*entry), GFP_KERNEL); + if (!entry) + return -ENOMEM; + + memset(entry, 0, sizeof(*entry)); + pages = (request->size + PAGE_SIZE - 1) / PAGE_SIZE; + DRM_DEBUG("size=%ld pages=%ld\n", request->size, pages); + + entry->pages = pages; + entry->pagelist = kmalloc(pages * sizeof(*entry->pagelist), GFP_KERNEL); + if (!entry->pagelist) { + kfree(entry); + return -ENOMEM; + } + + memset(entry->pagelist, 0, pages * sizeof(*entry->pagelist)); + + entry->busaddr = kmalloc(pages * sizeof(*entry->busaddr), GFP_KERNEL); + if (!entry->busaddr) { + kfree(entry->pagelist); + kfree(entry); + return -ENOMEM; + } + memset((void *)entry->busaddr, 0, pages * sizeof(*entry->busaddr)); + + entry->virtual = drm_vmalloc_dma(pages << PAGE_SHIFT); + if (!entry->virtual) { + kfree(entry->busaddr); + kfree(entry->pagelist); + kfree(entry); + return -ENOMEM; + } + + /* This also forces the mapping of COW pages, so our page list + * will be valid. Please don't remove it... + */ + memset(entry->virtual, 0, pages << PAGE_SHIFT); + + entry->handle = ScatterHandle((unsigned long)entry->virtual); + + DRM_DEBUG("handle = %08lx\n", entry->handle); + DRM_DEBUG("virtual = %p\n", entry->virtual); + + for (i = (unsigned long)entry->virtual, j = 0; j < pages; + i += PAGE_SIZE, j++) { + entry->pagelist[j] = vmalloc_to_page((void *)i); + if (!entry->pagelist[j]) + goto failed; + SetPageReserved(entry->pagelist[j]); + } + + request->handle = entry->handle; + + dev->sg = entry; + +#if DEBUG_SCATTER + /* Verify that each page points to its virtual address, and vice + * versa. + */ + { + int error = 0; + + for (i = 0; i < pages; i++) { + unsigned long *tmp; + + tmp = page_address(entry->pagelist[i]); + for (j = 0; + j < PAGE_SIZE / sizeof(unsigned long); + j++, tmp++) { + *tmp = 0xcafebabe; + } + tmp = (unsigned long *)((u8 *) entry->virtual + + (PAGE_SIZE * i)); + for (j = 0; + j < PAGE_SIZE / sizeof(unsigned long); + j++, tmp++) { + if (*tmp != 0xcafebabe && error == 0) { + error = 1; + DRM_ERROR("Scatter allocation error, " + "pagelist does not match " + "virtual mapping\n"); + } + } + tmp = page_address(entry->pagelist[i]); + for (j = 0; + j < PAGE_SIZE / sizeof(unsigned long); + j++, tmp++) { + *tmp = 0; + } + } + if (error == 0) + DRM_ERROR("Scatter allocation matches pagelist\n"); + } +#endif + + return 0; + + failed: + drm_sg_cleanup(entry); + return -ENOMEM; +} + +int drm_sg_alloc_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_scatter_gather *request = data; + + return drm_sg_alloc(dev, request); + +} + +int drm_sg_free(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_scatter_gather *request = data; + struct drm_sg_mem *entry; + + if (!drm_core_check_feature(dev, DRIVER_SG)) + return -EINVAL; + + entry = dev->sg; + dev->sg = NULL; + + if (!entry || entry->handle != request->handle) + return -EINVAL; + + DRM_DEBUG("virtual = %p\n", entry->virtual); + + drm_sg_cleanup(entry); + + return 0; +} diff --git a/drivers/gpu/drm/drm_sman.c b/drivers/gpu/drm/drm_sman.c new file mode 100644 index 00000000..34664587 --- /dev/null +++ b/drivers/gpu/drm/drm_sman.c @@ -0,0 +1,350 @@ +/************************************************************************** + * + * Copyright 2006 Tungsten Graphics, Inc., Bismarck., ND., USA. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * + **************************************************************************/ +/* + * Simple memory manager interface that keeps track on allocate regions on a + * per "owner" basis. All regions associated with an "owner" can be released + * with a simple call. Typically if the "owner" exists. The owner is any + * "unsigned long" identifier. Can typically be a pointer to a file private + * struct or a context identifier. + * + * Authors: + * Thomas Hellström + */ + +#include "drm_sman.h" + +struct drm_owner_item { + struct drm_hash_item owner_hash; + struct list_head sman_list; + struct list_head mem_blocks; +}; + +void drm_sman_takedown(struct drm_sman * sman) +{ + drm_ht_remove(&sman->user_hash_tab); + drm_ht_remove(&sman->owner_hash_tab); + kfree(sman->mm); +} + +EXPORT_SYMBOL(drm_sman_takedown); + +int +drm_sman_init(struct drm_sman * sman, unsigned int num_managers, + unsigned int user_order, unsigned int owner_order) +{ + int ret = 0; + + sman->mm = kcalloc(num_managers, sizeof(*sman->mm), GFP_KERNEL); + if (!sman->mm) { + ret = -ENOMEM; + goto out; + } + sman->num_managers = num_managers; + INIT_LIST_HEAD(&sman->owner_items); + ret = drm_ht_create(&sman->owner_hash_tab, owner_order); + if (ret) + goto out1; + ret = drm_ht_create(&sman->user_hash_tab, user_order); + if (!ret) + goto out; + + drm_ht_remove(&sman->owner_hash_tab); +out1: + kfree(sman->mm); +out: + return ret; +} + +EXPORT_SYMBOL(drm_sman_init); + +static void *drm_sman_mm_allocate(void *private, unsigned long size, + unsigned alignment) +{ + struct drm_mm *mm = (struct drm_mm *) private; + struct drm_mm_node *tmp; + + tmp = drm_mm_search_free(mm, size, alignment, 1); + if (!tmp) { + return NULL; + } + tmp = drm_mm_get_block(tmp, size, alignment); + return tmp; +} + +static void drm_sman_mm_free(void *private, void *ref) +{ + struct drm_mm_node *node = (struct drm_mm_node *) ref; + + drm_mm_put_block(node); +} + +static void drm_sman_mm_destroy(void *private) +{ + struct drm_mm *mm = (struct drm_mm *) private; + drm_mm_takedown(mm); + kfree(mm); +} + +static unsigned long drm_sman_mm_offset(void *private, void *ref) +{ + struct drm_mm_node *node = (struct drm_mm_node *) ref; + return node->start; +} + +int +drm_sman_set_range(struct drm_sman * sman, unsigned int manager, + unsigned long start, unsigned long size) +{ + struct drm_sman_mm *sman_mm; + struct drm_mm *mm; + int ret; + + BUG_ON(manager >= sman->num_managers); + + sman_mm = &sman->mm[manager]; + mm = kzalloc(sizeof(*mm), GFP_KERNEL); + if (!mm) { + return -ENOMEM; + } + sman_mm->private = mm; + ret = drm_mm_init(mm, start, size); + + if (ret) { + kfree(mm); + return ret; + } + + sman_mm->allocate = drm_sman_mm_allocate; + sman_mm->free = drm_sman_mm_free; + sman_mm->destroy = drm_sman_mm_destroy; + sman_mm->offset = drm_sman_mm_offset; + + return 0; +} + +EXPORT_SYMBOL(drm_sman_set_range); + +int +drm_sman_set_manager(struct drm_sman * sman, unsigned int manager, + struct drm_sman_mm * allocator) +{ + BUG_ON(manager >= sman->num_managers); + sman->mm[manager] = *allocator; + + return 0; +} +EXPORT_SYMBOL(drm_sman_set_manager); + +static struct drm_owner_item *drm_sman_get_owner_item(struct drm_sman * sman, + unsigned long owner) +{ + int ret; + struct drm_hash_item *owner_hash_item; + struct drm_owner_item *owner_item; + + ret = drm_ht_find_item(&sman->owner_hash_tab, owner, &owner_hash_item); + if (!ret) { + return drm_hash_entry(owner_hash_item, struct drm_owner_item, + owner_hash); + } + + owner_item = kzalloc(sizeof(*owner_item), GFP_KERNEL); + if (!owner_item) + goto out; + + INIT_LIST_HEAD(&owner_item->mem_blocks); + owner_item->owner_hash.key = owner; + if (drm_ht_insert_item(&sman->owner_hash_tab, &owner_item->owner_hash)) + goto out1; + + list_add_tail(&owner_item->sman_list, &sman->owner_items); + return owner_item; + +out1: + kfree(owner_item); +out: + return NULL; +} + +struct drm_memblock_item *drm_sman_alloc(struct drm_sman *sman, unsigned int manager, + unsigned long size, unsigned alignment, + unsigned long owner) +{ + void *tmp; + struct drm_sman_mm *sman_mm; + struct drm_owner_item *owner_item; + struct drm_memblock_item *memblock; + + BUG_ON(manager >= sman->num_managers); + + sman_mm = &sman->mm[manager]; + tmp = sman_mm->allocate(sman_mm->private, size, alignment); + + if (!tmp) { + return NULL; + } + + memblock = kzalloc(sizeof(*memblock), GFP_KERNEL); + + if (!memblock) + goto out; + + memblock->mm_info = tmp; + memblock->mm = sman_mm; + memblock->sman = sman; + + if (drm_ht_just_insert_please + (&sman->user_hash_tab, &memblock->user_hash, + (unsigned long)memblock, 32, 0, 0)) + goto out1; + + owner_item = drm_sman_get_owner_item(sman, owner); + if (!owner_item) + goto out2; + + list_add_tail(&memblock->owner_list, &owner_item->mem_blocks); + + return memblock; + +out2: + drm_ht_remove_item(&sman->user_hash_tab, &memblock->user_hash); +out1: + kfree(memblock); +out: + sman_mm->free(sman_mm->private, tmp); + + return NULL; +} + +EXPORT_SYMBOL(drm_sman_alloc); + +static void drm_sman_free(struct drm_memblock_item *item) +{ + struct drm_sman *sman = item->sman; + + list_del(&item->owner_list); + drm_ht_remove_item(&sman->user_hash_tab, &item->user_hash); + item->mm->free(item->mm->private, item->mm_info); + kfree(item); +} + +int drm_sman_free_key(struct drm_sman *sman, unsigned int key) +{ + struct drm_hash_item *hash_item; + struct drm_memblock_item *memblock_item; + + if (drm_ht_find_item(&sman->user_hash_tab, key, &hash_item)) + return -EINVAL; + + memblock_item = drm_hash_entry(hash_item, struct drm_memblock_item, + user_hash); + drm_sman_free(memblock_item); + return 0; +} + +EXPORT_SYMBOL(drm_sman_free_key); + +static void drm_sman_remove_owner(struct drm_sman *sman, + struct drm_owner_item *owner_item) +{ + list_del(&owner_item->sman_list); + drm_ht_remove_item(&sman->owner_hash_tab, &owner_item->owner_hash); + kfree(owner_item); +} + +int drm_sman_owner_clean(struct drm_sman *sman, unsigned long owner) +{ + + struct drm_hash_item *hash_item; + struct drm_owner_item *owner_item; + + if (drm_ht_find_item(&sman->owner_hash_tab, owner, &hash_item)) { + return -1; + } + + owner_item = drm_hash_entry(hash_item, struct drm_owner_item, owner_hash); + if (owner_item->mem_blocks.next == &owner_item->mem_blocks) { + drm_sman_remove_owner(sman, owner_item); + return -1; + } + + return 0; +} + +EXPORT_SYMBOL(drm_sman_owner_clean); + +static void drm_sman_do_owner_cleanup(struct drm_sman *sman, + struct drm_owner_item *owner_item) +{ + struct drm_memblock_item *entry, *next; + + list_for_each_entry_safe(entry, next, &owner_item->mem_blocks, + owner_list) { + drm_sman_free(entry); + } + drm_sman_remove_owner(sman, owner_item); +} + +void drm_sman_owner_cleanup(struct drm_sman *sman, unsigned long owner) +{ + + struct drm_hash_item *hash_item; + struct drm_owner_item *owner_item; + + if (drm_ht_find_item(&sman->owner_hash_tab, owner, &hash_item)) { + + return; + } + + owner_item = drm_hash_entry(hash_item, struct drm_owner_item, owner_hash); + drm_sman_do_owner_cleanup(sman, owner_item); +} + +EXPORT_SYMBOL(drm_sman_owner_cleanup); + +void drm_sman_cleanup(struct drm_sman *sman) +{ + struct drm_owner_item *entry, *next; + unsigned int i; + struct drm_sman_mm *sman_mm; + + list_for_each_entry_safe(entry, next, &sman->owner_items, sman_list) { + drm_sman_do_owner_cleanup(sman, entry); + } + if (sman->mm) { + for (i = 0; i < sman->num_managers; ++i) { + sman_mm = &sman->mm[i]; + if (sman_mm->private) { + sman_mm->destroy(sman_mm->private); + sman_mm->private = NULL; + } + } + } +} + +EXPORT_SYMBOL(drm_sman_cleanup); diff --git a/drivers/gpu/drm/drm_stub.c b/drivers/gpu/drm/drm_stub.c new file mode 100644 index 00000000..6d7b083c --- /dev/null +++ b/drivers/gpu/drm/drm_stub.c @@ -0,0 +1,494 @@ +/** + * \file drm_stub.h + * Stub support + * + * \author Rickard E. (Rik) Faith + */ + +/* + * Created: Fri Jan 19 10:48:35 2001 by faith@acm.org + * + * Copyright 2001 VA Linux Systems, Inc., Sunnyvale, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + */ + +#include +#include +#include +#include "drmP.h" +#include "drm_core.h" + +unsigned int drm_debug = 0; /* 1 to enable debug output */ +EXPORT_SYMBOL(drm_debug); + +unsigned int drm_vblank_offdelay = 5000; /* Default to 5000 msecs. */ +EXPORT_SYMBOL(drm_vblank_offdelay); + +unsigned int drm_timestamp_precision = 20; /* Default to 20 usecs. */ +EXPORT_SYMBOL(drm_timestamp_precision); + +MODULE_AUTHOR(CORE_AUTHOR); +MODULE_DESCRIPTION(CORE_DESC); +MODULE_LICENSE("GPL and additional rights"); +MODULE_PARM_DESC(debug, "Enable debug output"); +MODULE_PARM_DESC(vblankoffdelay, "Delay until vblank irq auto-disable [msecs]"); +MODULE_PARM_DESC(timestamp_precision_usec, "Max. error on timestamps [usecs]"); + +module_param_named(debug, drm_debug, int, 0600); +module_param_named(vblankoffdelay, drm_vblank_offdelay, int, 0600); +module_param_named(timestamp_precision_usec, drm_timestamp_precision, int, 0600); + +struct idr drm_minors_idr; + +struct class *drm_class; +struct proc_dir_entry *drm_proc_root; +struct dentry *drm_debugfs_root; + +int drm_err(const char *func, const char *format, ...) +{ + struct va_format vaf; + va_list args; + int r; + + va_start(args, format); + + vaf.fmt = format; + vaf.va = &args; + + r = printk(KERN_ERR "[" DRM_NAME ":%s] *ERROR* %pV", func, &vaf); + + va_end(args); + + return r; +} +EXPORT_SYMBOL(drm_err); + +void drm_ut_debug_printk(unsigned int request_level, + const char *prefix, + const char *function_name, + const char *format, ...) +{ + va_list args; + + if (drm_debug & request_level) { + if (function_name) + printk(KERN_DEBUG "[%s:%s], ", prefix, function_name); + va_start(args, format); + vprintk(format, args); + va_end(args); + } +} +EXPORT_SYMBOL(drm_ut_debug_printk); + +static int drm_minor_get_id(struct drm_device *dev, int type) +{ + int new_id; + int ret; + int base = 0, limit = 63; + + if (type == DRM_MINOR_CONTROL) { + base += 64; + limit = base + 127; + } else if (type == DRM_MINOR_RENDER) { + base += 128; + limit = base + 255; + } + +again: + if (idr_pre_get(&drm_minors_idr, GFP_KERNEL) == 0) { + DRM_ERROR("Out of memory expanding drawable idr\n"); + return -ENOMEM; + } + mutex_lock(&dev->struct_mutex); + ret = idr_get_new_above(&drm_minors_idr, NULL, + base, &new_id); + mutex_unlock(&dev->struct_mutex); + if (ret == -EAGAIN) { + goto again; + } else if (ret) { + return ret; + } + + if (new_id >= limit) { + idr_remove(&drm_minors_idr, new_id); + return -EINVAL; + } + return new_id; +} + +struct drm_master *drm_master_create(struct drm_minor *minor) +{ + struct drm_master *master; + + master = kzalloc(sizeof(*master), GFP_KERNEL); + if (!master) + return NULL; + + kref_init(&master->refcount); + spin_lock_init(&master->lock.spinlock); + init_waitqueue_head(&master->lock.lock_queue); + drm_ht_create(&master->magiclist, DRM_MAGIC_HASH_ORDER); + INIT_LIST_HEAD(&master->magicfree); + master->minor = minor; + + list_add_tail(&master->head, &minor->master_list); + + return master; +} + +struct drm_master *drm_master_get(struct drm_master *master) +{ + kref_get(&master->refcount); + return master; +} +EXPORT_SYMBOL(drm_master_get); + +static void drm_master_destroy(struct kref *kref) +{ + struct drm_master *master = container_of(kref, struct drm_master, refcount); + struct drm_magic_entry *pt, *next; + struct drm_device *dev = master->minor->dev; + struct drm_map_list *r_list, *list_temp; + + list_del(&master->head); + + if (dev->driver->master_destroy) + dev->driver->master_destroy(dev, master); + + list_for_each_entry_safe(r_list, list_temp, &dev->maplist, head) { + if (r_list->master == master) { + drm_rmmap_locked(dev, r_list->map); + r_list = NULL; + } + } + + if (master->unique) { + kfree(master->unique); + master->unique = NULL; + master->unique_len = 0; + } + + kfree(dev->devname); + dev->devname = NULL; + + list_for_each_entry_safe(pt, next, &master->magicfree, head) { + list_del(&pt->head); + drm_ht_remove_item(&master->magiclist, &pt->hash_item); + kfree(pt); + } + + drm_ht_remove(&master->magiclist); + + kfree(master); +} + +void drm_master_put(struct drm_master **master) +{ + kref_put(&(*master)->refcount, drm_master_destroy); + *master = NULL; +} +EXPORT_SYMBOL(drm_master_put); + +int drm_setmaster_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + int ret = 0; + + if (file_priv->is_master) + return 0; + + if (file_priv->minor->master && file_priv->minor->master != file_priv->master) + return -EINVAL; + + if (!file_priv->master) + return -EINVAL; + + if (!file_priv->minor->master && + file_priv->minor->master != file_priv->master) { + mutex_lock(&dev->struct_mutex); + file_priv->minor->master = drm_master_get(file_priv->master); + file_priv->is_master = 1; + if (dev->driver->master_set) { + ret = dev->driver->master_set(dev, file_priv, false); + if (unlikely(ret != 0)) { + file_priv->is_master = 0; + drm_master_put(&file_priv->minor->master); + } + } + mutex_unlock(&dev->struct_mutex); + } + + return 0; +} + +int drm_dropmaster_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + if (!file_priv->is_master) + return -EINVAL; + + if (!file_priv->minor->master) + return -EINVAL; + + mutex_lock(&dev->struct_mutex); + if (dev->driver->master_drop) + dev->driver->master_drop(dev, file_priv, false); + drm_master_put(&file_priv->minor->master); + file_priv->is_master = 0; + mutex_unlock(&dev->struct_mutex); + return 0; +} + +int drm_fill_in_dev(struct drm_device *dev, + const struct pci_device_id *ent, + struct drm_driver *driver) +{ + int retcode; + + INIT_LIST_HEAD(&dev->filelist); + INIT_LIST_HEAD(&dev->ctxlist); + INIT_LIST_HEAD(&dev->vmalist); + INIT_LIST_HEAD(&dev->maplist); + INIT_LIST_HEAD(&dev->vblank_event_list); + + spin_lock_init(&dev->count_lock); + spin_lock_init(&dev->event_lock); + mutex_init(&dev->struct_mutex); + mutex_init(&dev->ctxlist_mutex); + + if (drm_ht_create(&dev->map_hash, 12)) { + return -ENOMEM; + } + + /* the DRM has 6 basic counters */ + dev->counters = 6; + dev->types[0] = _DRM_STAT_LOCK; + dev->types[1] = _DRM_STAT_OPENS; + dev->types[2] = _DRM_STAT_CLOSES; + dev->types[3] = _DRM_STAT_IOCTLS; + dev->types[4] = _DRM_STAT_LOCKS; + dev->types[5] = _DRM_STAT_UNLOCKS; + + dev->driver = driver; + + if (dev->driver->bus->agp_init) { + retcode = dev->driver->bus->agp_init(dev); + if (retcode) + goto error_out_unreg; + } + + + + retcode = drm_ctxbitmap_init(dev); + if (retcode) { + DRM_ERROR("Cannot allocate memory for context bitmap.\n"); + goto error_out_unreg; + } + + if (driver->driver_features & DRIVER_GEM) { + retcode = drm_gem_init(dev); + if (retcode) { + DRM_ERROR("Cannot initialize graphics execution " + "manager (GEM)\n"); + goto error_out_unreg; + } + } + + return 0; + + error_out_unreg: + drm_lastclose(dev); + return retcode; +} + + +/** + * Get a secondary minor number. + * + * \param dev device data structure + * \param sec-minor structure to hold the assigned minor + * \return negative number on failure. + * + * Search an empty entry and initialize it to the given parameters, and + * create the proc init entry via proc_init(). This routines assigns + * minor numbers to secondary heads of multi-headed cards + */ +int drm_get_minor(struct drm_device *dev, struct drm_minor **minor, int type) +{ + struct drm_minor *new_minor; + int ret; + int minor_id; + + DRM_DEBUG("\n"); + + minor_id = drm_minor_get_id(dev, type); + if (minor_id < 0) + return minor_id; + + new_minor = kzalloc(sizeof(struct drm_minor), GFP_KERNEL); + if (!new_minor) { + ret = -ENOMEM; + goto err_idr; + } + + new_minor->type = type; + new_minor->device = MKDEV(DRM_MAJOR, minor_id); + new_minor->dev = dev; + new_minor->index = minor_id; + INIT_LIST_HEAD(&new_minor->master_list); + + idr_replace(&drm_minors_idr, new_minor, minor_id); + + if (type == DRM_MINOR_LEGACY) { + ret = drm_proc_init(new_minor, minor_id, drm_proc_root); + if (ret) { + DRM_ERROR("DRM: Failed to initialize /proc/dri.\n"); + goto err_mem; + } + } else + new_minor->proc_root = NULL; + +#if defined(CONFIG_DEBUG_FS) + ret = drm_debugfs_init(new_minor, minor_id, drm_debugfs_root); + if (ret) { + DRM_ERROR("DRM: Failed to initialize /sys/kernel/debug/dri.\n"); + goto err_g2; + } +#endif + + ret = drm_sysfs_device_add(new_minor); + if (ret) { + printk(KERN_ERR + "DRM: Error sysfs_device_add.\n"); + goto err_g2; + } + *minor = new_minor; + + DRM_DEBUG("new minor assigned %d\n", minor_id); + return 0; + + +err_g2: + if (new_minor->type == DRM_MINOR_LEGACY) + drm_proc_cleanup(new_minor, drm_proc_root); +err_mem: + kfree(new_minor); +err_idr: + idr_remove(&drm_minors_idr, minor_id); + *minor = NULL; + return ret; +} + +/** + * Put a secondary minor number. + * + * \param sec_minor - structure to be released + * \return always zero + * + * Cleans up the proc resources. Not legal for this to be the + * last minor released. + * + */ +int drm_put_minor(struct drm_minor **minor_p) +{ + struct drm_minor *minor = *minor_p; + + DRM_DEBUG("release secondary minor %d\n", minor->index); + + if (minor->type == DRM_MINOR_LEGACY) + drm_proc_cleanup(minor, drm_proc_root); +#if defined(CONFIG_DEBUG_FS) + drm_debugfs_cleanup(minor); +#endif + + drm_sysfs_device_remove(minor); + + idr_remove(&drm_minors_idr, minor->index); + + kfree(minor); + *minor_p = NULL; + return 0; +} + +/** + * Called via drm_exit() at module unload time or when pci device is + * unplugged. + * + * Cleans up all DRM device, calling drm_lastclose(). + * + */ +void drm_put_dev(struct drm_device *dev) +{ + struct drm_driver *driver; + struct drm_map_list *r_list, *list_temp; + + DRM_DEBUG("\n"); + + if (!dev) { + DRM_ERROR("cleanup called no dev\n"); + return; + } + driver = dev->driver; + + drm_lastclose(dev); + + if (drm_core_has_MTRR(dev) && drm_core_has_AGP(dev) && + dev->agp && dev->agp->agp_mtrr >= 0) { + int retval; + retval = mtrr_del(dev->agp->agp_mtrr, + dev->agp->agp_info.aper_base, + dev->agp->agp_info.aper_size * 1024 * 1024); + DRM_DEBUG("mtrr_del=%d\n", retval); + } + + if (dev->driver->unload) + dev->driver->unload(dev); + + if (drm_core_has_AGP(dev) && dev->agp) { + kfree(dev->agp); + dev->agp = NULL; + } + + drm_vblank_cleanup(dev); + + list_for_each_entry_safe(r_list, list_temp, &dev->maplist, head) + drm_rmmap(dev, r_list->map); + drm_ht_remove(&dev->map_hash); + + drm_ctxbitmap_cleanup(dev); + + if (drm_core_check_feature(dev, DRIVER_MODESET)) + drm_put_minor(&dev->control); + + if (driver->driver_features & DRIVER_GEM) + drm_gem_destroy(dev); + + drm_put_minor(&dev->primary); + + list_del(&dev->driver_item); + if (dev->devname) { + kfree(dev->devname); + dev->devname = NULL; + } + kfree(dev); +} +EXPORT_SYMBOL(drm_put_dev); diff --git a/drivers/gpu/drm/drm_sysfs.c b/drivers/gpu/drm/drm_sysfs.c new file mode 100644 index 00000000..b17949d9 --- /dev/null +++ b/drivers/gpu/drm/drm_sysfs.c @@ -0,0 +1,561 @@ + +/* + * drm_sysfs.c - Modifications to drm_sysfs_class.c to support + * extra sysfs attribute from DRM. Normal drm_sysfs_class + * does not allow adding attributes. + * + * Copyright (c) 2004 Jon Smirl + * Copyright (c) 2003-2004 Greg Kroah-Hartman + * Copyright (c) 2003-2004 IBM Corp. + * + * This file is released under the GPLv2 + * + */ + +#include +#include +#include +#include + +#include "drm_sysfs.h" +#include "drm_core.h" +#include "drmP.h" + +#define to_drm_minor(d) container_of(d, struct drm_minor, kdev) +#define to_drm_connector(d) container_of(d, struct drm_connector, kdev) + +static struct device_type drm_sysfs_device_minor = { + .name = "drm_minor" +}; + +/** + * drm_class_suspend - DRM class suspend hook + * @dev: Linux device to suspend + * @state: power state to enter + * + * Just figures out what the actual struct drm_device associated with + * @dev is and calls its suspend hook, if present. + */ +static int drm_class_suspend(struct device *dev, pm_message_t state) +{ + if (dev->type == &drm_sysfs_device_minor) { + struct drm_minor *drm_minor = to_drm_minor(dev); + struct drm_device *drm_dev = drm_minor->dev; + + if (drm_minor->type == DRM_MINOR_LEGACY && + !drm_core_check_feature(drm_dev, DRIVER_MODESET) && + drm_dev->driver->suspend) + return drm_dev->driver->suspend(drm_dev, state); + } + return 0; +} + +/** + * drm_class_resume - DRM class resume hook + * @dev: Linux device to resume + * + * Just figures out what the actual struct drm_device associated with + * @dev is and calls its resume hook, if present. + */ +static int drm_class_resume(struct device *dev) +{ + if (dev->type == &drm_sysfs_device_minor) { + struct drm_minor *drm_minor = to_drm_minor(dev); + struct drm_device *drm_dev = drm_minor->dev; + + if (drm_minor->type == DRM_MINOR_LEGACY && + !drm_core_check_feature(drm_dev, DRIVER_MODESET) && + drm_dev->driver->resume) + return drm_dev->driver->resume(drm_dev); + } + return 0; +} + +static char *drm_devnode(struct device *dev, mode_t *mode) +{ + return kasprintf(GFP_KERNEL, "dri/%s", dev_name(dev)); +} + +static CLASS_ATTR_STRING(version, S_IRUGO, + CORE_NAME " " + __stringify(CORE_MAJOR) "." + __stringify(CORE_MINOR) "." + __stringify(CORE_PATCHLEVEL) " " + CORE_DATE); + +/** + * drm_sysfs_create - create a struct drm_sysfs_class structure + * @owner: pointer to the module that is to "own" this struct drm_sysfs_class + * @name: pointer to a string for the name of this class. + * + * This is used to create DRM class pointer that can then be used + * in calls to drm_sysfs_device_add(). + * + * Note, the pointer created here is to be destroyed when finished by making a + * call to drm_sysfs_destroy(). + */ +struct class *drm_sysfs_create(struct module *owner, char *name) +{ + struct class *class; + int err; + + class = class_create(owner, name); + if (IS_ERR(class)) { + err = PTR_ERR(class); + goto err_out; + } + + class->suspend = drm_class_suspend; + class->resume = drm_class_resume; + + err = class_create_file(class, &class_attr_version.attr); + if (err) + goto err_out_class; + + class->devnode = drm_devnode; + + return class; + +err_out_class: + class_destroy(class); +err_out: + return ERR_PTR(err); +} + +/** + * drm_sysfs_destroy - destroys DRM class + * + * Destroy the DRM device class. + */ +void drm_sysfs_destroy(void) +{ + if ((drm_class == NULL) || (IS_ERR(drm_class))) + return; + class_remove_file(drm_class, &class_attr_version.attr); + class_destroy(drm_class); +} + +/** + * drm_sysfs_device_release - do nothing + * @dev: Linux device + * + * Normally, this would free the DRM device associated with @dev, along + * with cleaning up any other stuff. But we do that in the DRM core, so + * this function can just return and hope that the core does its job. + */ +static void drm_sysfs_device_release(struct device *dev) +{ + memset(dev, 0, sizeof(struct device)); + return; +} + +/* + * Connector properties + */ +static ssize_t status_show(struct device *device, + struct device_attribute *attr, + char *buf) +{ + struct drm_connector *connector = to_drm_connector(device); + enum drm_connector_status status; + int ret; + + ret = mutex_lock_interruptible(&connector->dev->mode_config.mutex); + if (ret) + return ret; + + status = connector->funcs->detect(connector, true); + mutex_unlock(&connector->dev->mode_config.mutex); + + return snprintf(buf, PAGE_SIZE, "%s\n", + drm_get_connector_status_name(status)); +} + +static ssize_t dpms_show(struct device *device, + struct device_attribute *attr, + char *buf) +{ + struct drm_connector *connector = to_drm_connector(device); + struct drm_device *dev = connector->dev; + uint64_t dpms_status; + int ret; + + ret = drm_connector_property_get_value(connector, + dev->mode_config.dpms_property, + &dpms_status); + if (ret) + return 0; + + return snprintf(buf, PAGE_SIZE, "%s\n", + drm_get_dpms_name((int)dpms_status)); +} + +static ssize_t enabled_show(struct device *device, + struct device_attribute *attr, + char *buf) +{ + struct drm_connector *connector = to_drm_connector(device); + + return snprintf(buf, PAGE_SIZE, "%s\n", connector->encoder ? "enabled" : + "disabled"); +} + +static ssize_t edid_show(struct file *filp, struct kobject *kobj, + struct bin_attribute *attr, char *buf, loff_t off, + size_t count) +{ + struct device *connector_dev = container_of(kobj, struct device, kobj); + struct drm_connector *connector = to_drm_connector(connector_dev); + unsigned char *edid; + size_t size; + + if (!connector->edid_blob_ptr) + return 0; + + edid = connector->edid_blob_ptr->data; + size = connector->edid_blob_ptr->length; + if (!edid) + return 0; + + if (off >= size) + return 0; + + if (off + count > size) + count = size - off; + memcpy(buf, edid + off, count); + + return count; +} + +static ssize_t modes_show(struct device *device, + struct device_attribute *attr, + char *buf) +{ + struct drm_connector *connector = to_drm_connector(device); + struct drm_display_mode *mode; + int written = 0; + + list_for_each_entry(mode, &connector->modes, head) { + written += snprintf(buf + written, PAGE_SIZE - written, "%s\n", + mode->name); + } + + return written; +} + +static ssize_t subconnector_show(struct device *device, + struct device_attribute *attr, + char *buf) +{ + struct drm_connector *connector = to_drm_connector(device); + struct drm_device *dev = connector->dev; + struct drm_property *prop = NULL; + uint64_t subconnector; + int is_tv = 0; + int ret; + + switch (connector->connector_type) { + case DRM_MODE_CONNECTOR_DVII: + prop = dev->mode_config.dvi_i_subconnector_property; + break; + case DRM_MODE_CONNECTOR_Composite: + case DRM_MODE_CONNECTOR_SVIDEO: + case DRM_MODE_CONNECTOR_Component: + case DRM_MODE_CONNECTOR_TV: + prop = dev->mode_config.tv_subconnector_property; + is_tv = 1; + break; + default: + DRM_ERROR("Wrong connector type for this property\n"); + return 0; + } + + if (!prop) { + DRM_ERROR("Unable to find subconnector property\n"); + return 0; + } + + ret = drm_connector_property_get_value(connector, prop, &subconnector); + if (ret) + return 0; + + return snprintf(buf, PAGE_SIZE, "%s", is_tv ? + drm_get_tv_subconnector_name((int)subconnector) : + drm_get_dvi_i_subconnector_name((int)subconnector)); +} + +static ssize_t select_subconnector_show(struct device *device, + struct device_attribute *attr, + char *buf) +{ + struct drm_connector *connector = to_drm_connector(device); + struct drm_device *dev = connector->dev; + struct drm_property *prop = NULL; + uint64_t subconnector; + int is_tv = 0; + int ret; + + switch (connector->connector_type) { + case DRM_MODE_CONNECTOR_DVII: + prop = dev->mode_config.dvi_i_select_subconnector_property; + break; + case DRM_MODE_CONNECTOR_Composite: + case DRM_MODE_CONNECTOR_SVIDEO: + case DRM_MODE_CONNECTOR_Component: + case DRM_MODE_CONNECTOR_TV: + prop = dev->mode_config.tv_select_subconnector_property; + is_tv = 1; + break; + default: + DRM_ERROR("Wrong connector type for this property\n"); + return 0; + } + + if (!prop) { + DRM_ERROR("Unable to find select subconnector property\n"); + return 0; + } + + ret = drm_connector_property_get_value(connector, prop, &subconnector); + if (ret) + return 0; + + return snprintf(buf, PAGE_SIZE, "%s", is_tv ? + drm_get_tv_select_name((int)subconnector) : + drm_get_dvi_i_select_name((int)subconnector)); +} + +static struct device_attribute connector_attrs[] = { + __ATTR_RO(status), + __ATTR_RO(enabled), + __ATTR_RO(dpms), + __ATTR_RO(modes), +}; + +/* These attributes are for both DVI-I connectors and all types of tv-out. */ +static struct device_attribute connector_attrs_opt1[] = { + __ATTR_RO(subconnector), + __ATTR_RO(select_subconnector), +}; + +static struct bin_attribute edid_attr = { + .attr.name = "edid", + .attr.mode = 0444, + .size = 0, + .read = edid_show, +}; + +/** + * drm_sysfs_connector_add - add an connector to sysfs + * @connector: connector to add + * + * Create an connector device in sysfs, along with its associated connector + * properties (so far, connection status, dpms, mode list & edid) and + * generate a hotplug event so userspace knows there's a new connector + * available. + * + * Note: + * This routine should only be called *once* for each DRM minor registered. + * A second call for an already registered device will trigger the BUG_ON + * below. + */ +int drm_sysfs_connector_add(struct drm_connector *connector) +{ + struct drm_device *dev = connector->dev; + int attr_cnt = 0; + int opt_cnt = 0; + int i; + int ret = 0; + + /* We shouldn't get called more than once for the same connector */ + BUG_ON(device_is_registered(&connector->kdev)); + + connector->kdev.parent = &dev->primary->kdev; + connector->kdev.class = drm_class; + connector->kdev.release = drm_sysfs_device_release; + + DRM_DEBUG("adding \"%s\" to sysfs\n", + drm_get_connector_name(connector)); + + dev_set_name(&connector->kdev, "card%d-%s", + dev->primary->index, drm_get_connector_name(connector)); + ret = device_register(&connector->kdev); + + if (ret) { + DRM_ERROR("failed to register connector device: %d\n", ret); + goto out; + } + + /* Standard attributes */ + + for (attr_cnt = 0; attr_cnt < ARRAY_SIZE(connector_attrs); attr_cnt++) { + ret = device_create_file(&connector->kdev, &connector_attrs[attr_cnt]); + if (ret) + goto err_out_files; + } + + /* Optional attributes */ + /* + * In the long run it maybe a good idea to make one set of + * optionals per connector type. + */ + switch (connector->connector_type) { + case DRM_MODE_CONNECTOR_DVII: + case DRM_MODE_CONNECTOR_Composite: + case DRM_MODE_CONNECTOR_SVIDEO: + case DRM_MODE_CONNECTOR_Component: + case DRM_MODE_CONNECTOR_TV: + for (opt_cnt = 0; opt_cnt < ARRAY_SIZE(connector_attrs_opt1); opt_cnt++) { + ret = device_create_file(&connector->kdev, &connector_attrs_opt1[opt_cnt]); + if (ret) + goto err_out_files; + } + break; + default: + break; + } + + ret = sysfs_create_bin_file(&connector->kdev.kobj, &edid_attr); + if (ret) + goto err_out_files; + + /* Let userspace know we have a new connector */ + drm_sysfs_hotplug_event(dev); + + return 0; + +err_out_files: + for (i = 0; i < opt_cnt; i++) + device_remove_file(&connector->kdev, &connector_attrs_opt1[i]); + for (i = 0; i < attr_cnt; i++) + device_remove_file(&connector->kdev, &connector_attrs[i]); + device_unregister(&connector->kdev); + +out: + return ret; +} +EXPORT_SYMBOL(drm_sysfs_connector_add); + +/** + * drm_sysfs_connector_remove - remove an connector device from sysfs + * @connector: connector to remove + * + * Remove @connector and its associated attributes from sysfs. Note that + * the device model core will take care of sending the "remove" uevent + * at this time, so we don't need to do it. + * + * Note: + * This routine should only be called if the connector was previously + * successfully registered. If @connector hasn't been registered yet, + * you'll likely see a panic somewhere deep in sysfs code when called. + */ +void drm_sysfs_connector_remove(struct drm_connector *connector) +{ + int i; + + DRM_DEBUG("removing \"%s\" from sysfs\n", + drm_get_connector_name(connector)); + + for (i = 0; i < ARRAY_SIZE(connector_attrs); i++) + device_remove_file(&connector->kdev, &connector_attrs[i]); + sysfs_remove_bin_file(&connector->kdev.kobj, &edid_attr); + device_unregister(&connector->kdev); +} +EXPORT_SYMBOL(drm_sysfs_connector_remove); + +/** + * drm_sysfs_hotplug_event - generate a DRM uevent + * @dev: DRM device + * + * Send a uevent for the DRM device specified by @dev. Currently we only + * set HOTPLUG=1 in the uevent environment, but this could be expanded to + * deal with other types of events. + */ +void drm_sysfs_hotplug_event(struct drm_device *dev) +{ + char *event_string = "HOTPLUG=1"; + char *envp[] = { event_string, NULL }; + + DRM_DEBUG("generating hotplug event\n"); + + kobject_uevent_env(&dev->primary->kdev.kobj, KOBJ_CHANGE, envp); +} +EXPORT_SYMBOL(drm_sysfs_hotplug_event); + +/** + * drm_sysfs_device_add - adds a class device to sysfs for a character driver + * @dev: DRM device to be added + * @head: DRM head in question + * + * Add a DRM device to the DRM's device model class. We use @dev's PCI device + * as the parent for the Linux device, and make sure it has a file containing + * the driver we're using (for userspace compatibility). + */ +int drm_sysfs_device_add(struct drm_minor *minor) +{ + int err; + char *minor_str; + + /* minor->kdev.parent = minor->dev->dev; */ + minor->kdev.parent = NULL; + + minor->kdev.class = drm_class; + minor->kdev.release = drm_sysfs_device_release; + minor->kdev.devt = minor->device; + minor->kdev.type = &drm_sysfs_device_minor; + if (minor->type == DRM_MINOR_CONTROL) + minor_str = "controlD%d"; + else if (minor->type == DRM_MINOR_RENDER) + minor_str = "renderD%d"; + else + minor_str = "card%d"; + + dev_set_name(&minor->kdev, minor_str, minor->index); + + err = device_register(&minor->kdev); + if (err) { + DRM_ERROR("device add failed: %d\n", err); + goto err_out; + } + + return 0; + +err_out: + return err; +} + +/** + * drm_sysfs_device_remove - remove DRM device + * @dev: DRM device to remove + * + * This call unregisters and cleans up a class device that was created with a + * call to drm_sysfs_device_add() + */ +void drm_sysfs_device_remove(struct drm_minor *minor) +{ + device_unregister(&minor->kdev); +} + + +/** + * drm_class_device_register - Register a struct device in the drm class. + * + * @dev: pointer to struct device to register. + * + * @dev should have all relevant members pre-filled with the exception + * of the class member. In particular, the device_type member must + * be set. + */ + +int drm_class_device_register(struct device *dev) +{ + dev->class = drm_class; + return device_register(dev); +} +EXPORT_SYMBOL_GPL(drm_class_device_register); + +void drm_class_device_unregister(struct device *dev) +{ + return device_unregister(dev); +} +EXPORT_SYMBOL_GPL(drm_class_device_unregister); diff --git a/drivers/gpu/drm/drm_trace.h b/drivers/gpu/drm/drm_trace.h new file mode 100644 index 00000000..03ea964a --- /dev/null +++ b/drivers/gpu/drm/drm_trace.h @@ -0,0 +1,66 @@ +#if !defined(_DRM_TRACE_H_) || defined(TRACE_HEADER_MULTI_READ) +#define _DRM_TRACE_H_ + +#include +#include +#include + +#undef TRACE_SYSTEM +#define TRACE_SYSTEM drm +#define TRACE_SYSTEM_STRING __stringify(TRACE_SYSTEM) +#define TRACE_INCLUDE_FILE drm_trace + +TRACE_EVENT(drm_vblank_event, + TP_PROTO(int crtc, unsigned int seq), + TP_ARGS(crtc, seq), + TP_STRUCT__entry( + __field(int, crtc) + __field(unsigned int, seq) + ), + TP_fast_assign( + __entry->crtc = crtc; + __entry->seq = seq; + ), + TP_printk("crtc=%d, seq=%d", __entry->crtc, __entry->seq) +); + +TRACE_EVENT(drm_vblank_event_queued, + TP_PROTO(pid_t pid, int crtc, unsigned int seq), + TP_ARGS(pid, crtc, seq), + TP_STRUCT__entry( + __field(pid_t, pid) + __field(int, crtc) + __field(unsigned int, seq) + ), + TP_fast_assign( + __entry->pid = pid; + __entry->crtc = crtc; + __entry->seq = seq; + ), + TP_printk("pid=%d, crtc=%d, seq=%d", __entry->pid, __entry->crtc, \ + __entry->seq) +); + +TRACE_EVENT(drm_vblank_event_delivered, + TP_PROTO(pid_t pid, int crtc, unsigned int seq), + TP_ARGS(pid, crtc, seq), + TP_STRUCT__entry( + __field(pid_t, pid) + __field(int, crtc) + __field(unsigned int, seq) + ), + TP_fast_assign( + __entry->pid = pid; + __entry->crtc = crtc; + __entry->seq = seq; + ), + TP_printk("pid=%d, crtc=%d, seq=%d", __entry->pid, __entry->crtc, \ + __entry->seq) +); + +#endif /* _DRM_TRACE_H_ */ + +/* This part must be outside protection */ +#undef TRACE_INCLUDE_PATH +#define TRACE_INCLUDE_PATH . +#include diff --git a/drivers/gpu/drm/drm_trace_points.c b/drivers/gpu/drm/drm_trace_points.c new file mode 100644 index 00000000..0d0eb908 --- /dev/null +++ b/drivers/gpu/drm/drm_trace_points.c @@ -0,0 +1,4 @@ +#include "drmP.h" + +#define CREATE_TRACE_POINTS +#include "drm_trace.h" diff --git a/drivers/gpu/drm/drm_usb.c b/drivers/gpu/drm/drm_usb.c new file mode 100644 index 00000000..206d2300 --- /dev/null +++ b/drivers/gpu/drm/drm_usb.c @@ -0,0 +1,117 @@ +#include "drmP.h" +#include + +#ifdef CONFIG_USB +int drm_get_usb_dev(struct usb_interface *interface, + const struct usb_device_id *id, + struct drm_driver *driver) +{ + struct drm_device *dev; + struct usb_device *usbdev; + int ret; + + DRM_DEBUG("\n"); + + dev = kzalloc(sizeof(*dev), GFP_KERNEL); + if (!dev) + return -ENOMEM; + + usbdev = interface_to_usbdev(interface); + dev->usbdev = usbdev; + dev->dev = &usbdev->dev; + + mutex_lock(&drm_global_mutex); + + ret = drm_fill_in_dev(dev, NULL, driver); + if (ret) { + printk(KERN_ERR "DRM: Fill_in_dev failed.\n"); + goto err_g1; + } + + usb_set_intfdata(interface, dev); + ret = drm_get_minor(dev, &dev->control, DRM_MINOR_CONTROL); + if (ret) + goto err_g1; + + ret = drm_get_minor(dev, &dev->primary, DRM_MINOR_LEGACY); + if (ret) + goto err_g2; + + if (dev->driver->load) { + ret = dev->driver->load(dev, 0); + if (ret) + goto err_g3; + } + + /* setup the grouping for the legacy output */ + ret = drm_mode_group_init_legacy_group(dev, + &dev->primary->mode_group); + if (ret) + goto err_g3; + + list_add_tail(&dev->driver_item, &driver->device_list); + + mutex_unlock(&drm_global_mutex); + + DRM_INFO("Initialized %s %d.%d.%d %s on minor %d\n", + driver->name, driver->major, driver->minor, driver->patchlevel, + driver->date, dev->primary->index); + + return 0; + +err_g3: + drm_put_minor(&dev->primary); +err_g2: + drm_put_minor(&dev->control); +err_g1: + kfree(dev); + mutex_unlock(&drm_global_mutex); + return ret; + +} +EXPORT_SYMBOL(drm_get_usb_dev); + +static int drm_usb_get_irq(struct drm_device *dev) +{ + return 0; +} + +static const char *drm_usb_get_name(struct drm_device *dev) +{ + return "USB"; +} + +static int drm_usb_set_busid(struct drm_device *dev, + struct drm_master *master) +{ + return 0; +} + +static struct drm_bus drm_usb_bus = { + .bus_type = DRIVER_BUS_USB, + .get_irq = drm_usb_get_irq, + .get_name = drm_usb_get_name, + .set_busid = drm_usb_set_busid, +}; + +int drm_usb_init(struct drm_driver *driver, struct usb_driver *udriver) +{ + int res; + DRM_DEBUG("\n"); + + INIT_LIST_HEAD(&driver->device_list); + driver->kdriver.usb = udriver; + driver->bus = &drm_usb_bus; + + res = usb_register(udriver); + return res; +} +EXPORT_SYMBOL(drm_usb_init); + +void drm_usb_exit(struct drm_driver *driver, + struct usb_driver *udriver) +{ + usb_deregister(udriver); +} +EXPORT_SYMBOL(drm_usb_exit); +#endif diff --git a/drivers/gpu/drm/drm_vm.c b/drivers/gpu/drm/drm_vm.c new file mode 100644 index 00000000..5db96d45 --- /dev/null +++ b/drivers/gpu/drm/drm_vm.c @@ -0,0 +1,690 @@ +/** + * \file drm_vm.c + * Memory mapping for DRM + * + * \author Rickard E. (Rik) Faith + * \author Gareth Hughes + */ + +/* + * Created: Mon Jan 4 08:58:31 1999 by faith@valinux.com + * + * Copyright 1999 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * VA LINUX SYSTEMS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#include "drmP.h" +#if defined(__ia64__) +#include +#include +#endif + +static void drm_vm_open(struct vm_area_struct *vma); +static void drm_vm_close(struct vm_area_struct *vma); + +static pgprot_t drm_io_prot(uint32_t map_type, struct vm_area_struct *vma) +{ + pgprot_t tmp = vm_get_page_prot(vma->vm_flags); + +#if defined(__i386__) || defined(__x86_64__) + if (boot_cpu_data.x86 > 3 && map_type != _DRM_AGP) { + pgprot_val(tmp) |= _PAGE_PCD; + pgprot_val(tmp) &= ~_PAGE_PWT; + } +#elif defined(__powerpc__) + pgprot_val(tmp) |= _PAGE_NO_CACHE; + if (map_type == _DRM_REGISTERS) + pgprot_val(tmp) |= _PAGE_GUARDED; +#elif defined(__ia64__) + if (efi_range_is_wc(vma->vm_start, vma->vm_end - + vma->vm_start)) + tmp = pgprot_writecombine(tmp); + else + tmp = pgprot_noncached(tmp); +#elif defined(__sparc__) || defined(__arm__) + tmp = pgprot_noncached(tmp); +#endif + return tmp; +} + +static pgprot_t drm_dma_prot(uint32_t map_type, struct vm_area_struct *vma) +{ + pgprot_t tmp = vm_get_page_prot(vma->vm_flags); + +#if defined(__powerpc__) && defined(CONFIG_NOT_COHERENT_CACHE) + tmp |= _PAGE_NO_CACHE; +#endif + return tmp; +} + +/** + * \c fault method for AGP virtual memory. + * + * \param vma virtual memory area. + * \param address access address. + * \return pointer to the page structure. + * + * Find the right map and if it's AGP memory find the real physical page to + * map, get the page, increment the use count and return it. + */ +#if __OS_HAS_AGP +static int drm_do_vm_fault(struct vm_area_struct *vma, struct vm_fault *vmf) +{ + struct drm_file *priv = vma->vm_file->private_data; + struct drm_device *dev = priv->minor->dev; + struct drm_local_map *map = NULL; + struct drm_map_list *r_list; + struct drm_hash_item *hash; + + /* + * Find the right map + */ + if (!drm_core_has_AGP(dev)) + goto vm_fault_error; + + if (!dev->agp || !dev->agp->cant_use_aperture) + goto vm_fault_error; + + if (drm_ht_find_item(&dev->map_hash, vma->vm_pgoff, &hash)) + goto vm_fault_error; + + r_list = drm_hash_entry(hash, struct drm_map_list, hash); + map = r_list->map; + + if (map && map->type == _DRM_AGP) { + /* + * Using vm_pgoff as a selector forces us to use this unusual + * addressing scheme. + */ + resource_size_t offset = (unsigned long)vmf->virtual_address - + vma->vm_start; + resource_size_t baddr = map->offset + offset; + struct drm_agp_mem *agpmem; + struct page *page; + +#ifdef __alpha__ + /* + * Adjust to a bus-relative address + */ + baddr -= dev->hose->mem_space->start; +#endif + + /* + * It's AGP memory - find the real physical page to map + */ + list_for_each_entry(agpmem, &dev->agp->memory, head) { + if (agpmem->bound <= baddr && + agpmem->bound + agpmem->pages * PAGE_SIZE > baddr) + break; + } + + if (&agpmem->head == &dev->agp->memory) + goto vm_fault_error; + + /* + * Get the page, inc the use count, and return it + */ + offset = (baddr - agpmem->bound) >> PAGE_SHIFT; + page = agpmem->memory->pages[offset]; + get_page(page); + vmf->page = page; + + DRM_DEBUG + ("baddr = 0x%llx page = 0x%p, offset = 0x%llx, count=%d\n", + (unsigned long long)baddr, + agpmem->memory->pages[offset], + (unsigned long long)offset, + page_count(page)); + return 0; + } +vm_fault_error: + return VM_FAULT_SIGBUS; /* Disallow mremap */ +} +#else /* __OS_HAS_AGP */ +static int drm_do_vm_fault(struct vm_area_struct *vma, struct vm_fault *vmf) +{ + return VM_FAULT_SIGBUS; +} +#endif /* __OS_HAS_AGP */ + +/** + * \c nopage method for shared virtual memory. + * + * \param vma virtual memory area. + * \param address access address. + * \return pointer to the page structure. + * + * Get the mapping, find the real physical page to map, get the page, and + * return it. + */ +static int drm_do_vm_shm_fault(struct vm_area_struct *vma, struct vm_fault *vmf) +{ + struct drm_local_map *map = vma->vm_private_data; + unsigned long offset; + unsigned long i; + struct page *page; + + if (!map) + return VM_FAULT_SIGBUS; /* Nothing allocated */ + + offset = (unsigned long)vmf->virtual_address - vma->vm_start; + i = (unsigned long)map->handle + offset; + page = vmalloc_to_page((void *)i); + if (!page) + return VM_FAULT_SIGBUS; + get_page(page); + vmf->page = page; + + DRM_DEBUG("shm_fault 0x%lx\n", offset); + return 0; +} + +/** + * \c close method for shared virtual memory. + * + * \param vma virtual memory area. + * + * Deletes map information if we are the last + * person to close a mapping and it's not in the global maplist. + */ +static void drm_vm_shm_close(struct vm_area_struct *vma) +{ + struct drm_file *priv = vma->vm_file->private_data; + struct drm_device *dev = priv->minor->dev; + struct drm_vma_entry *pt, *temp; + struct drm_local_map *map; + struct drm_map_list *r_list; + int found_maps = 0; + + DRM_DEBUG("0x%08lx,0x%08lx\n", + vma->vm_start, vma->vm_end - vma->vm_start); + atomic_dec(&dev->vma_count); + + map = vma->vm_private_data; + + mutex_lock(&dev->struct_mutex); + list_for_each_entry_safe(pt, temp, &dev->vmalist, head) { + if (pt->vma->vm_private_data == map) + found_maps++; + if (pt->vma == vma) { + list_del(&pt->head); + kfree(pt); + } + } + + /* We were the only map that was found */ + if (found_maps == 1 && map->flags & _DRM_REMOVABLE) { + /* Check to see if we are in the maplist, if we are not, then + * we delete this mappings information. + */ + found_maps = 0; + list_for_each_entry(r_list, &dev->maplist, head) { + if (r_list->map == map) + found_maps++; + } + + if (!found_maps) { + drm_dma_handle_t dmah; + + switch (map->type) { + case _DRM_REGISTERS: + case _DRM_FRAME_BUFFER: + if (drm_core_has_MTRR(dev) && map->mtrr >= 0) { + int retcode; + retcode = mtrr_del(map->mtrr, + map->offset, + map->size); + DRM_DEBUG("mtrr_del = %d\n", retcode); + } + iounmap(map->handle); + break; + case _DRM_SHM: + vfree(map->handle); + break; + case _DRM_AGP: + case _DRM_SCATTER_GATHER: + break; + case _DRM_CONSISTENT: + dmah.vaddr = map->handle; + dmah.busaddr = map->offset; + dmah.size = map->size; + __drm_pci_free(dev, &dmah); + break; + case _DRM_GEM: + DRM_ERROR("tried to rmmap GEM object\n"); + break; + } + kfree(map); + } + } + mutex_unlock(&dev->struct_mutex); +} + +/** + * \c fault method for DMA virtual memory. + * + * \param vma virtual memory area. + * \param address access address. + * \return pointer to the page structure. + * + * Determine the page number from the page offset and get it from drm_device_dma::pagelist. + */ +static int drm_do_vm_dma_fault(struct vm_area_struct *vma, struct vm_fault *vmf) +{ + struct drm_file *priv = vma->vm_file->private_data; + struct drm_device *dev = priv->minor->dev; + struct drm_device_dma *dma = dev->dma; + unsigned long offset; + unsigned long page_nr; + struct page *page; + + if (!dma) + return VM_FAULT_SIGBUS; /* Error */ + if (!dma->pagelist) + return VM_FAULT_SIGBUS; /* Nothing allocated */ + + offset = (unsigned long)vmf->virtual_address - vma->vm_start; /* vm_[pg]off[set] should be 0 */ + page_nr = offset >> PAGE_SHIFT; /* page_nr could just be vmf->pgoff */ + page = virt_to_page((dma->pagelist[page_nr] + (offset & (~PAGE_MASK)))); + + get_page(page); + vmf->page = page; + + DRM_DEBUG("dma_fault 0x%lx (page %lu)\n", offset, page_nr); + return 0; +} + +/** + * \c fault method for scatter-gather virtual memory. + * + * \param vma virtual memory area. + * \param address access address. + * \return pointer to the page structure. + * + * Determine the map offset from the page offset and get it from drm_sg_mem::pagelist. + */ +static int drm_do_vm_sg_fault(struct vm_area_struct *vma, struct vm_fault *vmf) +{ + struct drm_local_map *map = vma->vm_private_data; + struct drm_file *priv = vma->vm_file->private_data; + struct drm_device *dev = priv->minor->dev; + struct drm_sg_mem *entry = dev->sg; + unsigned long offset; + unsigned long map_offset; + unsigned long page_offset; + struct page *page; + + if (!entry) + return VM_FAULT_SIGBUS; /* Error */ + if (!entry->pagelist) + return VM_FAULT_SIGBUS; /* Nothing allocated */ + + offset = (unsigned long)vmf->virtual_address - vma->vm_start; + map_offset = map->offset - (unsigned long)dev->sg->virtual; + page_offset = (offset >> PAGE_SHIFT) + (map_offset >> PAGE_SHIFT); + page = entry->pagelist[page_offset]; + get_page(page); + vmf->page = page; + + return 0; +} + +static int drm_vm_fault(struct vm_area_struct *vma, struct vm_fault *vmf) +{ + return drm_do_vm_fault(vma, vmf); +} + +static int drm_vm_shm_fault(struct vm_area_struct *vma, struct vm_fault *vmf) +{ + return drm_do_vm_shm_fault(vma, vmf); +} + +static int drm_vm_dma_fault(struct vm_area_struct *vma, struct vm_fault *vmf) +{ + return drm_do_vm_dma_fault(vma, vmf); +} + +static int drm_vm_sg_fault(struct vm_area_struct *vma, struct vm_fault *vmf) +{ + return drm_do_vm_sg_fault(vma, vmf); +} + +/** AGP virtual memory operations */ +static const struct vm_operations_struct drm_vm_ops = { + .fault = drm_vm_fault, + .open = drm_vm_open, + .close = drm_vm_close, +}; + +/** Shared virtual memory operations */ +static const struct vm_operations_struct drm_vm_shm_ops = { + .fault = drm_vm_shm_fault, + .open = drm_vm_open, + .close = drm_vm_shm_close, +}; + +/** DMA virtual memory operations */ +static const struct vm_operations_struct drm_vm_dma_ops = { + .fault = drm_vm_dma_fault, + .open = drm_vm_open, + .close = drm_vm_close, +}; + +/** Scatter-gather virtual memory operations */ +static const struct vm_operations_struct drm_vm_sg_ops = { + .fault = drm_vm_sg_fault, + .open = drm_vm_open, + .close = drm_vm_close, +}; + +/** + * \c open method for shared virtual memory. + * + * \param vma virtual memory area. + * + * Create a new drm_vma_entry structure as the \p vma private data entry and + * add it to drm_device::vmalist. + */ +void drm_vm_open_locked(struct vm_area_struct *vma) +{ + struct drm_file *priv = vma->vm_file->private_data; + struct drm_device *dev = priv->minor->dev; + struct drm_vma_entry *vma_entry; + + DRM_DEBUG("0x%08lx,0x%08lx\n", + vma->vm_start, vma->vm_end - vma->vm_start); + atomic_inc(&dev->vma_count); + + vma_entry = kmalloc(sizeof(*vma_entry), GFP_KERNEL); + if (vma_entry) { + vma_entry->vma = vma; + vma_entry->pid = current->pid; + list_add(&vma_entry->head, &dev->vmalist); + } +} + +static void drm_vm_open(struct vm_area_struct *vma) +{ + struct drm_file *priv = vma->vm_file->private_data; + struct drm_device *dev = priv->minor->dev; + + mutex_lock(&dev->struct_mutex); + drm_vm_open_locked(vma); + mutex_unlock(&dev->struct_mutex); +} + +void drm_vm_close_locked(struct vm_area_struct *vma) +{ + struct drm_file *priv = vma->vm_file->private_data; + struct drm_device *dev = priv->minor->dev; + struct drm_vma_entry *pt, *temp; + + DRM_DEBUG("0x%08lx,0x%08lx\n", + vma->vm_start, vma->vm_end - vma->vm_start); + atomic_dec(&dev->vma_count); + + list_for_each_entry_safe(pt, temp, &dev->vmalist, head) { + if (pt->vma == vma) { + list_del(&pt->head); + kfree(pt); + break; + } + } +} + +/** + * \c close method for all virtual memory types. + * + * \param vma virtual memory area. + * + * Search the \p vma private data entry in drm_device::vmalist, unlink it, and + * free it. + */ +static void drm_vm_close(struct vm_area_struct *vma) +{ + struct drm_file *priv = vma->vm_file->private_data; + struct drm_device *dev = priv->minor->dev; + + mutex_lock(&dev->struct_mutex); + drm_vm_close_locked(vma); + mutex_unlock(&dev->struct_mutex); +} + +/** + * mmap DMA memory. + * + * \param file_priv DRM file private. + * \param vma virtual memory area. + * \return zero on success or a negative number on failure. + * + * Sets the virtual memory area operations structure to vm_dma_ops, the file + * pointer, and calls vm_open(). + */ +static int drm_mmap_dma(struct file *filp, struct vm_area_struct *vma) +{ + struct drm_file *priv = filp->private_data; + struct drm_device *dev; + struct drm_device_dma *dma; + unsigned long length = vma->vm_end - vma->vm_start; + + dev = priv->minor->dev; + dma = dev->dma; + DRM_DEBUG("start = 0x%lx, end = 0x%lx, page offset = 0x%lx\n", + vma->vm_start, vma->vm_end, vma->vm_pgoff); + + /* Length must match exact page count */ + if (!dma || (length >> PAGE_SHIFT) != dma->page_count) { + return -EINVAL; + } + + if (!capable(CAP_SYS_ADMIN) && + (dma->flags & _DRM_DMA_USE_PCI_RO)) { + vma->vm_flags &= ~(VM_WRITE | VM_MAYWRITE); +#if defined(__i386__) || defined(__x86_64__) + pgprot_val(vma->vm_page_prot) &= ~_PAGE_RW; +#else + /* Ye gads this is ugly. With more thought + we could move this up higher and use + `protection_map' instead. */ + vma->vm_page_prot = + __pgprot(pte_val + (pte_wrprotect + (__pte(pgprot_val(vma->vm_page_prot))))); +#endif + } + + vma->vm_ops = &drm_vm_dma_ops; + + vma->vm_flags |= VM_RESERVED; /* Don't swap */ + vma->vm_flags |= VM_DONTEXPAND; + + vma->vm_file = filp; /* Needed for drm_vm_open() */ + drm_vm_open_locked(vma); + return 0; +} + +static resource_size_t drm_core_get_reg_ofs(struct drm_device *dev) +{ +#ifdef __alpha__ + return dev->hose->dense_mem_base; +#else + return 0; +#endif +} + +/** + * mmap DMA memory. + * + * \param file_priv DRM file private. + * \param vma virtual memory area. + * \return zero on success or a negative number on failure. + * + * If the virtual memory area has no offset associated with it then it's a DMA + * area, so calls mmap_dma(). Otherwise searches the map in drm_device::maplist, + * checks that the restricted flag is not set, sets the virtual memory operations + * according to the mapping type and remaps the pages. Finally sets the file + * pointer and calls vm_open(). + */ +int drm_mmap_locked(struct file *filp, struct vm_area_struct *vma) +{ + struct drm_file *priv = filp->private_data; + struct drm_device *dev = priv->minor->dev; + struct drm_local_map *map = NULL; + resource_size_t offset = 0; + struct drm_hash_item *hash; + + DRM_DEBUG("start = 0x%lx, end = 0x%lx, page offset = 0x%lx\n", + vma->vm_start, vma->vm_end, vma->vm_pgoff); + + if (!priv->authenticated) + return -EACCES; + + /* We check for "dma". On Apple's UniNorth, it's valid to have + * the AGP mapped at physical address 0 + * --BenH. + */ + if (!vma->vm_pgoff +#if __OS_HAS_AGP + && (!dev->agp + || dev->agp->agp_info.device->vendor != PCI_VENDOR_ID_APPLE) +#endif + ) + return drm_mmap_dma(filp, vma); + + if (drm_ht_find_item(&dev->map_hash, vma->vm_pgoff, &hash)) { + DRM_ERROR("Could not find map\n"); + return -EINVAL; + } + + map = drm_hash_entry(hash, struct drm_map_list, hash)->map; + if (!map || ((map->flags & _DRM_RESTRICTED) && !capable(CAP_SYS_ADMIN))) + return -EPERM; + + /* Check for valid size. */ + if (map->size < vma->vm_end - vma->vm_start) + return -EINVAL; + + if (!capable(CAP_SYS_ADMIN) && (map->flags & _DRM_READ_ONLY)) { + vma->vm_flags &= ~(VM_WRITE | VM_MAYWRITE); +#if defined(__i386__) || defined(__x86_64__) + pgprot_val(vma->vm_page_prot) &= ~_PAGE_RW; +#else + /* Ye gads this is ugly. With more thought + we could move this up higher and use + `protection_map' instead. */ + vma->vm_page_prot = + __pgprot(pte_val + (pte_wrprotect + (__pte(pgprot_val(vma->vm_page_prot))))); +#endif + } + + switch (map->type) { +#if !defined(__arm__) + case _DRM_AGP: + if (drm_core_has_AGP(dev) && dev->agp->cant_use_aperture) { + /* + * On some platforms we can't talk to bus dma address from the CPU, so for + * memory of type DRM_AGP, we'll deal with sorting out the real physical + * pages and mappings in fault() + */ +#if defined(__powerpc__) + pgprot_val(vma->vm_page_prot) |= _PAGE_NO_CACHE; +#endif + vma->vm_ops = &drm_vm_ops; + break; + } + /* fall through to _DRM_FRAME_BUFFER... */ +#endif + case _DRM_FRAME_BUFFER: + case _DRM_REGISTERS: + offset = drm_core_get_reg_ofs(dev); + vma->vm_flags |= VM_IO; /* not in core dump */ + vma->vm_page_prot = drm_io_prot(map->type, vma); +#if !defined(__arm__) + if (io_remap_pfn_range(vma, vma->vm_start, + (map->offset + offset) >> PAGE_SHIFT, + vma->vm_end - vma->vm_start, + vma->vm_page_prot)) + return -EAGAIN; +#else + if (remap_pfn_range(vma, vma->vm_start, + (map->offset + offset) >> PAGE_SHIFT, + vma->vm_end - vma->vm_start, + vma->vm_page_prot)) + return -EAGAIN; +#endif + + DRM_DEBUG(" Type = %d; start = 0x%lx, end = 0x%lx," + " offset = 0x%llx\n", + map->type, + vma->vm_start, vma->vm_end, (unsigned long long)(map->offset + offset)); + + vma->vm_ops = &drm_vm_ops; + break; + case _DRM_CONSISTENT: + /* Consistent memory is really like shared memory. But + * it's allocated in a different way, so avoid fault */ + if (remap_pfn_range(vma, vma->vm_start, + page_to_pfn(virt_to_page(map->handle)), + vma->vm_end - vma->vm_start, vma->vm_page_prot)) + return -EAGAIN; + vma->vm_page_prot = drm_dma_prot(map->type, vma); + /* fall through to _DRM_SHM */ + case _DRM_SHM: + vma->vm_ops = &drm_vm_shm_ops; + vma->vm_private_data = (void *)map; + /* Don't let this area swap. Change when + DRM_KERNEL advisory is supported. */ + vma->vm_flags |= VM_RESERVED; + break; + case _DRM_SCATTER_GATHER: + vma->vm_ops = &drm_vm_sg_ops; + vma->vm_private_data = (void *)map; + vma->vm_flags |= VM_RESERVED; + vma->vm_page_prot = drm_dma_prot(map->type, vma); + break; + default: + return -EINVAL; /* This should never happen. */ + } + vma->vm_flags |= VM_RESERVED; /* Don't swap */ + vma->vm_flags |= VM_DONTEXPAND; + + vma->vm_file = filp; /* Needed for drm_vm_open() */ + drm_vm_open_locked(vma); + return 0; +} + +int drm_mmap(struct file *filp, struct vm_area_struct *vma) +{ + struct drm_file *priv = filp->private_data; + struct drm_device *dev = priv->minor->dev; + int ret; + + mutex_lock(&dev->struct_mutex); + ret = drm_mmap_locked(filp, vma); + mutex_unlock(&dev->struct_mutex); + + return ret; +} +EXPORT_SYMBOL(drm_mmap); diff --git a/drivers/gpu/drm/i2c/Makefile b/drivers/gpu/drm/i2c/Makefile new file mode 100644 index 00000000..92862563 --- /dev/null +++ b/drivers/gpu/drm/i2c/Makefile @@ -0,0 +1,7 @@ +ccflags-y := -Iinclude/drm + +ch7006-y := ch7006_drv.o ch7006_mode.o +obj-$(CONFIG_DRM_I2C_CH7006) += ch7006.o + +sil164-y := sil164_drv.o +obj-$(CONFIG_DRM_I2C_SIL164) += sil164.o diff --git a/drivers/gpu/drm/i2c/ch7006_drv.c b/drivers/gpu/drm/i2c/ch7006_drv.c new file mode 100644 index 00000000..08792a74 --- /dev/null +++ b/drivers/gpu/drm/i2c/ch7006_drv.c @@ -0,0 +1,555 @@ +/* + * Copyright (C) 2009 Francisco Jerez. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "ch7006_priv.h" + +/* DRM encoder functions */ + +static void ch7006_encoder_set_config(struct drm_encoder *encoder, + void *params) +{ + struct ch7006_priv *priv = to_ch7006_priv(encoder); + + priv->params = *(struct ch7006_encoder_params *)params; +} + +static void ch7006_encoder_destroy(struct drm_encoder *encoder) +{ + struct ch7006_priv *priv = to_ch7006_priv(encoder); + + drm_property_destroy(encoder->dev, priv->scale_property); + + kfree(priv); + to_encoder_slave(encoder)->slave_priv = NULL; + + drm_i2c_encoder_destroy(encoder); +} + +static void ch7006_encoder_dpms(struct drm_encoder *encoder, int mode) +{ + struct i2c_client *client = drm_i2c_encoder_get_client(encoder); + struct ch7006_priv *priv = to_ch7006_priv(encoder); + struct ch7006_state *state = &priv->state; + + ch7006_dbg(client, "\n"); + + if (mode == priv->last_dpms) + return; + priv->last_dpms = mode; + + ch7006_setup_power_state(encoder); + + ch7006_load_reg(client, state, CH7006_POWER); +} + +static void ch7006_encoder_save(struct drm_encoder *encoder) +{ + struct i2c_client *client = drm_i2c_encoder_get_client(encoder); + struct ch7006_priv *priv = to_ch7006_priv(encoder); + + ch7006_dbg(client, "\n"); + + ch7006_state_save(client, &priv->saved_state); +} + +static void ch7006_encoder_restore(struct drm_encoder *encoder) +{ + struct i2c_client *client = drm_i2c_encoder_get_client(encoder); + struct ch7006_priv *priv = to_ch7006_priv(encoder); + + ch7006_dbg(client, "\n"); + + ch7006_state_load(client, &priv->saved_state); +} + +static bool ch7006_encoder_mode_fixup(struct drm_encoder *encoder, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct ch7006_priv *priv = to_ch7006_priv(encoder); + + /* The ch7006 is painfully picky with the input timings so no + * custom modes for now... */ + + priv->mode = ch7006_lookup_mode(encoder, mode); + + return !!priv->mode; +} + +static int ch7006_encoder_mode_valid(struct drm_encoder *encoder, + struct drm_display_mode *mode) +{ + if (ch7006_lookup_mode(encoder, mode)) + return MODE_OK; + else + return MODE_BAD; +} + +static void ch7006_encoder_mode_set(struct drm_encoder *encoder, + struct drm_display_mode *drm_mode, + struct drm_display_mode *adjusted_mode) +{ + struct i2c_client *client = drm_i2c_encoder_get_client(encoder); + struct ch7006_priv *priv = to_ch7006_priv(encoder); + struct ch7006_encoder_params *params = &priv->params; + struct ch7006_state *state = &priv->state; + uint8_t *regs = state->regs; + struct ch7006_mode *mode = priv->mode; + struct ch7006_tv_norm_info *norm = &ch7006_tv_norms[priv->norm]; + int start_active; + + ch7006_dbg(client, "\n"); + + regs[CH7006_DISPMODE] = norm->dispmode | mode->dispmode; + regs[CH7006_BWIDTH] = 0; + regs[CH7006_INPUT_FORMAT] = bitf(CH7006_INPUT_FORMAT_FORMAT, + params->input_format); + + regs[CH7006_CLKMODE] = CH7006_CLKMODE_SUBC_LOCK + | bitf(CH7006_CLKMODE_XCM, params->xcm) + | bitf(CH7006_CLKMODE_PCM, params->pcm); + if (params->clock_mode) + regs[CH7006_CLKMODE] |= CH7006_CLKMODE_MASTER; + if (params->clock_edge) + regs[CH7006_CLKMODE] |= CH7006_CLKMODE_POS_EDGE; + + start_active = (drm_mode->htotal & ~0x7) - (drm_mode->hsync_start & ~0x7); + regs[CH7006_POV] = bitf(CH7006_POV_START_ACTIVE_8, start_active); + regs[CH7006_START_ACTIVE] = bitf(CH7006_START_ACTIVE_0, start_active); + + regs[CH7006_INPUT_SYNC] = 0; + if (params->sync_direction) + regs[CH7006_INPUT_SYNC] |= CH7006_INPUT_SYNC_OUTPUT; + if (params->sync_encoding) + regs[CH7006_INPUT_SYNC] |= CH7006_INPUT_SYNC_EMBEDDED; + if (drm_mode->flags & DRM_MODE_FLAG_PVSYNC) + regs[CH7006_INPUT_SYNC] |= CH7006_INPUT_SYNC_PVSYNC; + if (drm_mode->flags & DRM_MODE_FLAG_PHSYNC) + regs[CH7006_INPUT_SYNC] |= CH7006_INPUT_SYNC_PHSYNC; + + regs[CH7006_DETECT] = 0; + regs[CH7006_BCLKOUT] = 0; + + regs[CH7006_SUBC_INC3] = 0; + if (params->pout_level) + regs[CH7006_SUBC_INC3] |= CH7006_SUBC_INC3_POUT_3_3V; + + regs[CH7006_SUBC_INC4] = 0; + if (params->active_detect) + regs[CH7006_SUBC_INC4] |= CH7006_SUBC_INC4_DS_INPUT; + + regs[CH7006_PLL_CONTROL] = priv->saved_state.regs[CH7006_PLL_CONTROL]; + + ch7006_setup_levels(encoder); + ch7006_setup_subcarrier(encoder); + ch7006_setup_pll(encoder); + ch7006_setup_power_state(encoder); + ch7006_setup_properties(encoder); + + ch7006_state_load(client, state); +} + +static enum drm_connector_status ch7006_encoder_detect(struct drm_encoder *encoder, + struct drm_connector *connector) +{ + struct i2c_client *client = drm_i2c_encoder_get_client(encoder); + struct ch7006_priv *priv = to_ch7006_priv(encoder); + struct ch7006_state *state = &priv->state; + int det; + + ch7006_dbg(client, "\n"); + + ch7006_save_reg(client, state, CH7006_DETECT); + ch7006_save_reg(client, state, CH7006_POWER); + ch7006_save_reg(client, state, CH7006_CLKMODE); + + ch7006_write(client, CH7006_POWER, CH7006_POWER_RESET | + bitfs(CH7006_POWER_LEVEL, NORMAL)); + ch7006_write(client, CH7006_CLKMODE, CH7006_CLKMODE_MASTER); + + ch7006_write(client, CH7006_DETECT, CH7006_DETECT_SENSE); + + ch7006_write(client, CH7006_DETECT, 0); + + det = ch7006_read(client, CH7006_DETECT); + + ch7006_load_reg(client, state, CH7006_CLKMODE); + ch7006_load_reg(client, state, CH7006_POWER); + ch7006_load_reg(client, state, CH7006_DETECT); + + if ((det & (CH7006_DETECT_SVIDEO_Y_TEST| + CH7006_DETECT_SVIDEO_C_TEST| + CH7006_DETECT_CVBS_TEST)) == 0) + priv->subconnector = DRM_MODE_SUBCONNECTOR_SCART; + else if ((det & (CH7006_DETECT_SVIDEO_Y_TEST| + CH7006_DETECT_SVIDEO_C_TEST)) == 0) + priv->subconnector = DRM_MODE_SUBCONNECTOR_SVIDEO; + else if ((det & CH7006_DETECT_CVBS_TEST) == 0) + priv->subconnector = DRM_MODE_SUBCONNECTOR_Composite; + else + priv->subconnector = DRM_MODE_SUBCONNECTOR_Unknown; + + drm_connector_property_set_value(connector, + encoder->dev->mode_config.tv_subconnector_property, + priv->subconnector); + + return priv->subconnector ? connector_status_connected : + connector_status_disconnected; +} + +static int ch7006_encoder_get_modes(struct drm_encoder *encoder, + struct drm_connector *connector) +{ + struct ch7006_priv *priv = to_ch7006_priv(encoder); + struct ch7006_mode *mode; + int n = 0; + + for (mode = ch7006_modes; mode->mode.clock; mode++) { + if (~mode->valid_scales & 1<scale || + ~mode->valid_norms & 1<norm) + continue; + + drm_mode_probed_add(connector, + drm_mode_duplicate(encoder->dev, &mode->mode)); + + n++; + } + + return n; +} + +static int ch7006_encoder_create_resources(struct drm_encoder *encoder, + struct drm_connector *connector) +{ + struct ch7006_priv *priv = to_ch7006_priv(encoder); + struct drm_device *dev = encoder->dev; + struct drm_mode_config *conf = &dev->mode_config; + + drm_mode_create_tv_properties(dev, NUM_TV_NORMS, ch7006_tv_norm_names); + + priv->scale_property = drm_property_create(dev, DRM_MODE_PROP_RANGE, + "scale", 2); + priv->scale_property->values[0] = 0; + priv->scale_property->values[1] = 2; + + drm_connector_attach_property(connector, conf->tv_select_subconnector_property, + priv->select_subconnector); + drm_connector_attach_property(connector, conf->tv_subconnector_property, + priv->subconnector); + drm_connector_attach_property(connector, conf->tv_left_margin_property, + priv->hmargin); + drm_connector_attach_property(connector, conf->tv_bottom_margin_property, + priv->vmargin); + drm_connector_attach_property(connector, conf->tv_mode_property, + priv->norm); + drm_connector_attach_property(connector, conf->tv_brightness_property, + priv->brightness); + drm_connector_attach_property(connector, conf->tv_contrast_property, + priv->contrast); + drm_connector_attach_property(connector, conf->tv_flicker_reduction_property, + priv->flicker); + drm_connector_attach_property(connector, priv->scale_property, + priv->scale); + + return 0; +} + +static int ch7006_encoder_set_property(struct drm_encoder *encoder, + struct drm_connector *connector, + struct drm_property *property, + uint64_t val) +{ + struct i2c_client *client = drm_i2c_encoder_get_client(encoder); + struct ch7006_priv *priv = to_ch7006_priv(encoder); + struct ch7006_state *state = &priv->state; + struct drm_mode_config *conf = &encoder->dev->mode_config; + struct drm_crtc *crtc = encoder->crtc; + bool modes_changed = false; + + ch7006_dbg(client, "\n"); + + if (property == conf->tv_select_subconnector_property) { + priv->select_subconnector = val; + + ch7006_setup_power_state(encoder); + + ch7006_load_reg(client, state, CH7006_POWER); + + } else if (property == conf->tv_left_margin_property) { + priv->hmargin = val; + + ch7006_setup_properties(encoder); + + ch7006_load_reg(client, state, CH7006_POV); + ch7006_load_reg(client, state, CH7006_HPOS); + + } else if (property == conf->tv_bottom_margin_property) { + priv->vmargin = val; + + ch7006_setup_properties(encoder); + + ch7006_load_reg(client, state, CH7006_POV); + ch7006_load_reg(client, state, CH7006_VPOS); + + } else if (property == conf->tv_mode_property) { + if (connector->dpms != DRM_MODE_DPMS_OFF) + return -EINVAL; + + priv->norm = val; + + modes_changed = true; + + } else if (property == conf->tv_brightness_property) { + priv->brightness = val; + + ch7006_setup_levels(encoder); + + ch7006_load_reg(client, state, CH7006_BLACK_LEVEL); + + } else if (property == conf->tv_contrast_property) { + priv->contrast = val; + + ch7006_setup_properties(encoder); + + ch7006_load_reg(client, state, CH7006_CONTRAST); + + } else if (property == conf->tv_flicker_reduction_property) { + priv->flicker = val; + + ch7006_setup_properties(encoder); + + ch7006_load_reg(client, state, CH7006_FFILTER); + + } else if (property == priv->scale_property) { + if (connector->dpms != DRM_MODE_DPMS_OFF) + return -EINVAL; + + priv->scale = val; + + modes_changed = true; + + } else { + return -EINVAL; + } + + if (modes_changed) { + drm_helper_probe_single_connector_modes(connector, 0, 0); + + /* Disable the crtc to ensure a full modeset is + * performed whenever it's turned on again. */ + if (crtc) { + struct drm_mode_set modeset = { + .crtc = crtc, + }; + + crtc->funcs->set_config(&modeset); + } + } + + return 0; +} + +static struct drm_encoder_slave_funcs ch7006_encoder_funcs = { + .set_config = ch7006_encoder_set_config, + .destroy = ch7006_encoder_destroy, + .dpms = ch7006_encoder_dpms, + .save = ch7006_encoder_save, + .restore = ch7006_encoder_restore, + .mode_fixup = ch7006_encoder_mode_fixup, + .mode_valid = ch7006_encoder_mode_valid, + .mode_set = ch7006_encoder_mode_set, + .detect = ch7006_encoder_detect, + .get_modes = ch7006_encoder_get_modes, + .create_resources = ch7006_encoder_create_resources, + .set_property = ch7006_encoder_set_property, +}; + + +/* I2C driver functions */ + +static int ch7006_probe(struct i2c_client *client, const struct i2c_device_id *id) +{ + uint8_t addr = CH7006_VERSION_ID; + uint8_t val; + int ret; + + ch7006_dbg(client, "\n"); + + ret = i2c_master_send(client, &addr, sizeof(addr)); + if (ret < 0) + goto fail; + + ret = i2c_master_recv(client, &val, sizeof(val)); + if (ret < 0) + goto fail; + + ch7006_info(client, "Detected version ID: %x\n", val); + + /* I don't know what this is for, but otherwise I get no + * signal. + */ + ch7006_write(client, 0x3d, 0x0); + + return 0; + +fail: + ch7006_err(client, "Error %d reading version ID\n", ret); + + return -ENODEV; +} + +static int ch7006_remove(struct i2c_client *client) +{ + ch7006_dbg(client, "\n"); + + return 0; +} + +static int ch7006_suspend(struct i2c_client *client, pm_message_t mesg) +{ + ch7006_dbg(client, "\n"); + + return 0; +} + +static int ch7006_resume(struct i2c_client *client) +{ + ch7006_dbg(client, "\n"); + + ch7006_write(client, 0x3d, 0x0); + + return 0; +} + +static int ch7006_encoder_init(struct i2c_client *client, + struct drm_device *dev, + struct drm_encoder_slave *encoder) +{ + struct ch7006_priv *priv; + int i; + + ch7006_dbg(client, "\n"); + + priv = kzalloc(sizeof(*priv), GFP_KERNEL); + if (!priv) + return -ENOMEM; + + encoder->slave_priv = priv; + encoder->slave_funcs = &ch7006_encoder_funcs; + + priv->norm = TV_NORM_PAL; + priv->select_subconnector = DRM_MODE_SUBCONNECTOR_Automatic; + priv->subconnector = DRM_MODE_SUBCONNECTOR_Unknown; + priv->scale = 1; + priv->contrast = 50; + priv->brightness = 50; + priv->flicker = 50; + priv->hmargin = 50; + priv->vmargin = 50; + priv->last_dpms = -1; + priv->chip_version = ch7006_read(client, CH7006_VERSION_ID); + + if (ch7006_tv_norm) { + for (i = 0; i < NUM_TV_NORMS; i++) { + if (!strcmp(ch7006_tv_norm_names[i], ch7006_tv_norm)) { + priv->norm = i; + break; + } + } + + if (i == NUM_TV_NORMS) + ch7006_err(client, "Invalid TV norm setting \"%s\".\n", + ch7006_tv_norm); + } + + if (ch7006_scale >= 0 && ch7006_scale <= 2) + priv->scale = ch7006_scale; + else + ch7006_err(client, "Invalid scale setting \"%d\".\n", + ch7006_scale); + + return 0; +} + +static struct i2c_device_id ch7006_ids[] = { + { "ch7006", 0 }, + { } +}; +MODULE_DEVICE_TABLE(i2c, ch7006_ids); + +static struct drm_i2c_encoder_driver ch7006_driver = { + .i2c_driver = { + .probe = ch7006_probe, + .remove = ch7006_remove, + .suspend = ch7006_suspend, + .resume = ch7006_resume, + + .driver = { + .name = "ch7006", + }, + + .id_table = ch7006_ids, + }, + + .encoder_init = ch7006_encoder_init, +}; + + +/* Module initialization */ + +static int __init ch7006_init(void) +{ + return drm_i2c_encoder_register(THIS_MODULE, &ch7006_driver); +} + +static void __exit ch7006_exit(void) +{ + drm_i2c_encoder_unregister(&ch7006_driver); +} + +int ch7006_debug; +module_param_named(debug, ch7006_debug, int, 0600); +MODULE_PARM_DESC(debug, "Enable debug output."); + +char *ch7006_tv_norm; +module_param_named(tv_norm, ch7006_tv_norm, charp, 0600); +MODULE_PARM_DESC(tv_norm, "Default TV norm.\n" + "\t\tSupported: PAL, PAL-M, PAL-N, PAL-Nc, PAL-60, NTSC-M, NTSC-J.\n" + "\t\tDefault: PAL"); + +int ch7006_scale = 1; +module_param_named(scale, ch7006_scale, int, 0600); +MODULE_PARM_DESC(scale, "Default scale.\n" + "\t\tSupported: 0 -> Select video modes with a higher blanking ratio.\n" + "\t\t\t1 -> Select default video modes.\n" + "\t\t\t2 -> Select video modes with a lower blanking ratio."); + +MODULE_AUTHOR("Francisco Jerez "); +MODULE_DESCRIPTION("Chrontel ch7006 TV encoder driver"); +MODULE_LICENSE("GPL and additional rights"); + +module_init(ch7006_init); +module_exit(ch7006_exit); diff --git a/drivers/gpu/drm/i2c/ch7006_mode.c b/drivers/gpu/drm/i2c/ch7006_mode.c new file mode 100644 index 00000000..c860f24a --- /dev/null +++ b/drivers/gpu/drm/i2c/ch7006_mode.c @@ -0,0 +1,471 @@ +/* + * Copyright (C) 2009 Francisco Jerez. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "ch7006_priv.h" + +char *ch7006_tv_norm_names[] = { + [TV_NORM_PAL] = "PAL", + [TV_NORM_PAL_M] = "PAL-M", + [TV_NORM_PAL_N] = "PAL-N", + [TV_NORM_PAL_NC] = "PAL-Nc", + [TV_NORM_PAL_60] = "PAL-60", + [TV_NORM_NTSC_M] = "NTSC-M", + [TV_NORM_NTSC_J] = "NTSC-J", +}; + +#define NTSC_LIKE_TIMINGS .vrefresh = 60 * fixed1/1.001, \ + .vdisplay = 480, \ + .vtotal = 525, \ + .hvirtual = 660 + +#define PAL_LIKE_TIMINGS .vrefresh = 50 * fixed1, \ + .vdisplay = 576, \ + .vtotal = 625, \ + .hvirtual = 810 + +struct ch7006_tv_norm_info ch7006_tv_norms[] = { + [TV_NORM_NTSC_M] = { + NTSC_LIKE_TIMINGS, + .black_level = 0.339 * fixed1, + .subc_freq = 3579545 * fixed1, + .dispmode = bitfs(CH7006_DISPMODE_OUTPUT_STD, NTSC), + .voffset = 0, + }, + [TV_NORM_NTSC_J] = { + NTSC_LIKE_TIMINGS, + .black_level = 0.286 * fixed1, + .subc_freq = 3579545 * fixed1, + .dispmode = bitfs(CH7006_DISPMODE_OUTPUT_STD, NTSC_J), + .voffset = 0, + }, + [TV_NORM_PAL] = { + PAL_LIKE_TIMINGS, + .black_level = 0.3 * fixed1, + .subc_freq = 4433618.75 * fixed1, + .dispmode = bitfs(CH7006_DISPMODE_OUTPUT_STD, PAL), + .voffset = 0, + }, + [TV_NORM_PAL_M] = { + NTSC_LIKE_TIMINGS, + .black_level = 0.339 * fixed1, + .subc_freq = 3575611.433 * fixed1, + .dispmode = bitfs(CH7006_DISPMODE_OUTPUT_STD, PAL_M), + .voffset = 16, + }, + + /* The following modes seem to work right but they're + * undocumented */ + + [TV_NORM_PAL_N] = { + PAL_LIKE_TIMINGS, + .black_level = 0.339 * fixed1, + .subc_freq = 4433618.75 * fixed1, + .dispmode = bitfs(CH7006_DISPMODE_OUTPUT_STD, PAL), + .voffset = 0, + }, + [TV_NORM_PAL_NC] = { + PAL_LIKE_TIMINGS, + .black_level = 0.3 * fixed1, + .subc_freq = 3582056.25 * fixed1, + .dispmode = bitfs(CH7006_DISPMODE_OUTPUT_STD, PAL), + .voffset = 0, + }, + [TV_NORM_PAL_60] = { + NTSC_LIKE_TIMINGS, + .black_level = 0.3 * fixed1, + .subc_freq = 4433618.75 * fixed1, + .dispmode = bitfs(CH7006_DISPMODE_OUTPUT_STD, PAL_M), + .voffset = 16, + }, +}; + +#define __MODE(f, hd, vd, ht, vt, hsynp, vsynp, \ + subc, scale, scale_mask, norm_mask, e_hd, e_vd) { \ + .mode = { \ + .name = #hd "x" #vd, \ + .status = 0, \ + .type = DRM_MODE_TYPE_DRIVER, \ + .clock = f, \ + .hdisplay = hd, \ + .hsync_start = e_hd + 16, \ + .hsync_end = e_hd + 80, \ + .htotal = ht, \ + .hskew = 0, \ + .vdisplay = vd, \ + .vsync_start = vd + 10, \ + .vsync_end = vd + 26, \ + .vtotal = vt, \ + .vscan = 0, \ + .flags = DRM_MODE_FLAG_##hsynp##HSYNC | \ + DRM_MODE_FLAG_##vsynp##VSYNC, \ + .vrefresh = 0, \ + }, \ + .enc_hdisp = e_hd, \ + .enc_vdisp = e_vd, \ + .subc_coeff = subc * fixed1, \ + .dispmode = bitfs(CH7006_DISPMODE_SCALING_RATIO, scale) | \ + bitfs(CH7006_DISPMODE_INPUT_RES, e_hd##x##e_vd), \ + .valid_scales = scale_mask, \ + .valid_norms = norm_mask \ + } + +#define MODE(f, hd, vd, ht, vt, hsynp, vsynp, \ + subc, scale, scale_mask, norm_mask) \ + __MODE(f, hd, vd, ht, vt, hsynp, vsynp, subc, scale, \ + scale_mask, norm_mask, hd, vd) + +#define NTSC_LIKE (1 << TV_NORM_NTSC_M | 1 << TV_NORM_NTSC_J | \ + 1 << TV_NORM_PAL_M | 1 << TV_NORM_PAL_60) + +#define PAL_LIKE (1 << TV_NORM_PAL | 1 << TV_NORM_PAL_N | 1 << TV_NORM_PAL_NC) + +struct ch7006_mode ch7006_modes[] = { + MODE(21000, 512, 384, 840, 500, N, N, 181.797557582, 5_4, 0x6, PAL_LIKE), + MODE(26250, 512, 384, 840, 625, N, N, 145.438046066, 1_1, 0x1, PAL_LIKE), + MODE(20140, 512, 384, 800, 420, N, N, 213.257083791, 5_4, 0x4, NTSC_LIKE), + MODE(24671, 512, 384, 784, 525, N, N, 174.0874153, 1_1, 0x3, NTSC_LIKE), + MODE(28125, 720, 400, 1125, 500, N, N, 135.742176298, 5_4, 0x6, PAL_LIKE), + MODE(34875, 720, 400, 1116, 625, N, N, 109.469496898, 1_1, 0x1, PAL_LIKE), + MODE(23790, 720, 400, 945, 420, N, N, 160.475642016, 5_4, 0x4, NTSC_LIKE), + MODE(29455, 720, 400, 936, 525, N, N, 129.614941843, 1_1, 0x3, NTSC_LIKE), + MODE(25000, 640, 400, 1000, 500, N, N, 152.709948279, 5_4, 0x6, PAL_LIKE), + MODE(31500, 640, 400, 1008, 625, N, N, 121.198371646, 1_1, 0x1, PAL_LIKE), + MODE(21147, 640, 400, 840, 420, N, N, 180.535097338, 5_4, 0x4, NTSC_LIKE), + MODE(26434, 640, 400, 840, 525, N, N, 144.42807787, 1_1, 0x2, NTSC_LIKE), + MODE(30210, 640, 400, 840, 600, N, N, 126.374568276, 7_8, 0x1, NTSC_LIKE), + MODE(21000, 640, 480, 840, 500, N, N, 181.797557582, 5_4, 0x4, PAL_LIKE), + MODE(26250, 640, 480, 840, 625, N, N, 145.438046066, 1_1, 0x2, PAL_LIKE), + MODE(31500, 640, 480, 840, 750, N, N, 121.198371646, 5_6, 0x1, PAL_LIKE), + MODE(24671, 640, 480, 784, 525, N, N, 174.0874153, 1_1, 0x4, NTSC_LIKE), + MODE(28196, 640, 480, 784, 600, N, N, 152.326488422, 7_8, 0x2, NTSC_LIKE), + MODE(30210, 640, 480, 800, 630, N, N, 142.171389101, 5_6, 0x1, NTSC_LIKE), + __MODE(29500, 720, 576, 944, 625, P, P, 145.592111636, 1_1, 0x7, PAL_LIKE, 800, 600), + MODE(36000, 800, 600, 960, 750, P, P, 119.304647022, 5_6, 0x6, PAL_LIKE), + MODE(39000, 800, 600, 936, 836, P, P, 110.127366499, 3_4, 0x1, PAL_LIKE), + MODE(39273, 800, 600, 1040, 630, P, P, 145.816809399, 5_6, 0x4, NTSC_LIKE), + MODE(43636, 800, 600, 1040, 700, P, P, 131.235128487, 3_4, 0x2, NTSC_LIKE), + MODE(47832, 800, 600, 1064, 750, P, P, 119.723275165, 7_10, 0x1, NTSC_LIKE), + {} +}; + +struct ch7006_mode *ch7006_lookup_mode(struct drm_encoder *encoder, + struct drm_display_mode *drm_mode) +{ + struct ch7006_priv *priv = to_ch7006_priv(encoder); + struct ch7006_mode *mode; + + for (mode = ch7006_modes; mode->mode.clock; mode++) { + + if (~mode->valid_norms & 1<norm) + continue; + + if (mode->mode.hdisplay != drm_mode->hdisplay || + mode->mode.vdisplay != drm_mode->vdisplay || + mode->mode.vtotal != drm_mode->vtotal || + mode->mode.htotal != drm_mode->htotal || + mode->mode.clock != drm_mode->clock) + continue; + + return mode; + } + + return NULL; +} + +/* Some common HW state calculation code */ + +void ch7006_setup_levels(struct drm_encoder *encoder) +{ + struct i2c_client *client = drm_i2c_encoder_get_client(encoder); + struct ch7006_priv *priv = to_ch7006_priv(encoder); + uint8_t *regs = priv->state.regs; + struct ch7006_tv_norm_info *norm = &ch7006_tv_norms[priv->norm]; + int gain; + int black_level; + + /* Set DAC_GAIN if the voltage drop between white and black is + * high enough. */ + if (norm->black_level < 339*fixed1/1000) { + gain = 76; + + regs[CH7006_INPUT_FORMAT] |= CH7006_INPUT_FORMAT_DAC_GAIN; + } else { + gain = 71; + + regs[CH7006_INPUT_FORMAT] &= ~CH7006_INPUT_FORMAT_DAC_GAIN; + } + + black_level = round_fixed(norm->black_level*26625)/gain; + + /* Correct it with the specified brightness. */ + black_level = interpolate(90, black_level, 208, priv->brightness); + + regs[CH7006_BLACK_LEVEL] = bitf(CH7006_BLACK_LEVEL_0, black_level); + + ch7006_dbg(client, "black level: %d\n", black_level); +} + +void ch7006_setup_subcarrier(struct drm_encoder *encoder) +{ + struct i2c_client *client = drm_i2c_encoder_get_client(encoder); + struct ch7006_priv *priv = to_ch7006_priv(encoder); + struct ch7006_state *state = &priv->state; + struct ch7006_tv_norm_info *norm = &ch7006_tv_norms[priv->norm]; + struct ch7006_mode *mode = priv->mode; + uint32_t subc_inc; + + subc_inc = round_fixed((mode->subc_coeff >> 8) + * (norm->subc_freq >> 24)); + + setbitf(state, CH7006_SUBC_INC0, 28, subc_inc); + setbitf(state, CH7006_SUBC_INC1, 24, subc_inc); + setbitf(state, CH7006_SUBC_INC2, 20, subc_inc); + setbitf(state, CH7006_SUBC_INC3, 16, subc_inc); + setbitf(state, CH7006_SUBC_INC4, 12, subc_inc); + setbitf(state, CH7006_SUBC_INC5, 8, subc_inc); + setbitf(state, CH7006_SUBC_INC6, 4, subc_inc); + setbitf(state, CH7006_SUBC_INC7, 0, subc_inc); + + ch7006_dbg(client, "subcarrier inc: %u\n", subc_inc); +} + +void ch7006_setup_pll(struct drm_encoder *encoder) +{ + struct i2c_client *client = drm_i2c_encoder_get_client(encoder); + struct ch7006_priv *priv = to_ch7006_priv(encoder); + uint8_t *regs = priv->state.regs; + struct ch7006_mode *mode = priv->mode; + int n, best_n = 0; + int m, best_m = 0; + int freq, best_freq = 0; + + for (n = 0; n < CH7006_MAXN; n++) { + for (m = 0; m < CH7006_MAXM; m++) { + freq = CH7006_FREQ0*(n+2)/(m+2); + + if (abs(freq - mode->mode.clock) < + abs(best_freq - mode->mode.clock)) { + best_freq = freq; + best_n = n; + best_m = m; + } + } + } + + regs[CH7006_PLLOV] = bitf(CH7006_PLLOV_N_8, best_n) | + bitf(CH7006_PLLOV_M_8, best_m); + + regs[CH7006_PLLM] = bitf(CH7006_PLLM_0, best_m); + regs[CH7006_PLLN] = bitf(CH7006_PLLN_0, best_n); + + if (best_n < 108) + regs[CH7006_PLL_CONTROL] |= CH7006_PLL_CONTROL_CAPACITOR; + else + regs[CH7006_PLL_CONTROL] &= ~CH7006_PLL_CONTROL_CAPACITOR; + + ch7006_dbg(client, "n=%d m=%d f=%d c=%d\n", + best_n, best_m, best_freq, best_n < 108); +} + +void ch7006_setup_power_state(struct drm_encoder *encoder) +{ + struct ch7006_priv *priv = to_ch7006_priv(encoder); + uint8_t *power = &priv->state.regs[CH7006_POWER]; + int subconnector; + + subconnector = priv->select_subconnector ? priv->select_subconnector : + priv->subconnector; + + *power = CH7006_POWER_RESET; + + if (priv->last_dpms == DRM_MODE_DPMS_ON) { + switch (subconnector) { + case DRM_MODE_SUBCONNECTOR_SVIDEO: + *power |= bitfs(CH7006_POWER_LEVEL, CVBS_OFF); + break; + case DRM_MODE_SUBCONNECTOR_Composite: + *power |= bitfs(CH7006_POWER_LEVEL, SVIDEO_OFF); + break; + case DRM_MODE_SUBCONNECTOR_SCART: + *power |= bitfs(CH7006_POWER_LEVEL, NORMAL) | + CH7006_POWER_SCART; + break; + } + + } else { + if (priv->chip_version >= 0x20) + *power |= bitfs(CH7006_POWER_LEVEL, FULL_POWER_OFF); + else + *power |= bitfs(CH7006_POWER_LEVEL, POWER_OFF); + } +} + +void ch7006_setup_properties(struct drm_encoder *encoder) +{ + struct i2c_client *client = drm_i2c_encoder_get_client(encoder); + struct ch7006_priv *priv = to_ch7006_priv(encoder); + struct ch7006_state *state = &priv->state; + struct ch7006_tv_norm_info *norm = &ch7006_tv_norms[priv->norm]; + struct ch7006_mode *ch_mode = priv->mode; + struct drm_display_mode *mode = &ch_mode->mode; + uint8_t *regs = state->regs; + int flicker, contrast, hpos, vpos; + uint64_t scale, aspect; + + flicker = interpolate(0, 2, 3, priv->flicker); + regs[CH7006_FFILTER] = bitf(CH7006_FFILTER_TEXT, flicker) | + bitf(CH7006_FFILTER_LUMA, flicker) | + bitf(CH7006_FFILTER_CHROMA, 1); + + contrast = interpolate(0, 5, 7, priv->contrast); + regs[CH7006_CONTRAST] = bitf(CH7006_CONTRAST_0, contrast); + + scale = norm->vtotal*fixed1; + do_div(scale, mode->vtotal); + + aspect = ch_mode->enc_hdisp*fixed1; + do_div(aspect, ch_mode->enc_vdisp); + + hpos = round_fixed((norm->hvirtual * aspect - mode->hdisplay * scale) + * priv->hmargin * mode->vtotal) / norm->vtotal / 100 / 4; + + setbitf(state, CH7006_POV, HPOS_8, hpos); + setbitf(state, CH7006_HPOS, 0, hpos); + + vpos = max(0, norm->vdisplay - round_fixed(mode->vdisplay*scale) + + norm->voffset) * priv->vmargin / 100 / 2; + + setbitf(state, CH7006_POV, VPOS_8, vpos); + setbitf(state, CH7006_VPOS, 0, vpos); + + ch7006_dbg(client, "hpos: %d, vpos: %d\n", hpos, vpos); +} + +/* HW access functions */ + +void ch7006_write(struct i2c_client *client, uint8_t addr, uint8_t val) +{ + uint8_t buf[] = {addr, val}; + int ret; + + ret = i2c_master_send(client, buf, ARRAY_SIZE(buf)); + if (ret < 0) + ch7006_err(client, "Error %d writing to subaddress 0x%x\n", + ret, addr); +} + +uint8_t ch7006_read(struct i2c_client *client, uint8_t addr) +{ + uint8_t val; + int ret; + + ret = i2c_master_send(client, &addr, sizeof(addr)); + if (ret < 0) + goto fail; + + ret = i2c_master_recv(client, &val, sizeof(val)); + if (ret < 0) + goto fail; + + return val; + +fail: + ch7006_err(client, "Error %d reading from subaddress 0x%x\n", + ret, addr); + return 0; +} + +void ch7006_state_load(struct i2c_client *client, + struct ch7006_state *state) +{ + ch7006_load_reg(client, state, CH7006_POWER); + + ch7006_load_reg(client, state, CH7006_DISPMODE); + ch7006_load_reg(client, state, CH7006_FFILTER); + ch7006_load_reg(client, state, CH7006_BWIDTH); + ch7006_load_reg(client, state, CH7006_INPUT_FORMAT); + ch7006_load_reg(client, state, CH7006_CLKMODE); + ch7006_load_reg(client, state, CH7006_START_ACTIVE); + ch7006_load_reg(client, state, CH7006_POV); + ch7006_load_reg(client, state, CH7006_BLACK_LEVEL); + ch7006_load_reg(client, state, CH7006_HPOS); + ch7006_load_reg(client, state, CH7006_VPOS); + ch7006_load_reg(client, state, CH7006_INPUT_SYNC); + ch7006_load_reg(client, state, CH7006_DETECT); + ch7006_load_reg(client, state, CH7006_CONTRAST); + ch7006_load_reg(client, state, CH7006_PLLOV); + ch7006_load_reg(client, state, CH7006_PLLM); + ch7006_load_reg(client, state, CH7006_PLLN); + ch7006_load_reg(client, state, CH7006_BCLKOUT); + ch7006_load_reg(client, state, CH7006_SUBC_INC0); + ch7006_load_reg(client, state, CH7006_SUBC_INC1); + ch7006_load_reg(client, state, CH7006_SUBC_INC2); + ch7006_load_reg(client, state, CH7006_SUBC_INC3); + ch7006_load_reg(client, state, CH7006_SUBC_INC4); + ch7006_load_reg(client, state, CH7006_SUBC_INC5); + ch7006_load_reg(client, state, CH7006_SUBC_INC6); + ch7006_load_reg(client, state, CH7006_SUBC_INC7); + ch7006_load_reg(client, state, CH7006_PLL_CONTROL); + ch7006_load_reg(client, state, CH7006_CALC_SUBC_INC0); +} + +void ch7006_state_save(struct i2c_client *client, + struct ch7006_state *state) +{ + ch7006_save_reg(client, state, CH7006_POWER); + + ch7006_save_reg(client, state, CH7006_DISPMODE); + ch7006_save_reg(client, state, CH7006_FFILTER); + ch7006_save_reg(client, state, CH7006_BWIDTH); + ch7006_save_reg(client, state, CH7006_INPUT_FORMAT); + ch7006_save_reg(client, state, CH7006_CLKMODE); + ch7006_save_reg(client, state, CH7006_START_ACTIVE); + ch7006_save_reg(client, state, CH7006_POV); + ch7006_save_reg(client, state, CH7006_BLACK_LEVEL); + ch7006_save_reg(client, state, CH7006_HPOS); + ch7006_save_reg(client, state, CH7006_VPOS); + ch7006_save_reg(client, state, CH7006_INPUT_SYNC); + ch7006_save_reg(client, state, CH7006_DETECT); + ch7006_save_reg(client, state, CH7006_CONTRAST); + ch7006_save_reg(client, state, CH7006_PLLOV); + ch7006_save_reg(client, state, CH7006_PLLM); + ch7006_save_reg(client, state, CH7006_PLLN); + ch7006_save_reg(client, state, CH7006_BCLKOUT); + ch7006_save_reg(client, state, CH7006_SUBC_INC0); + ch7006_save_reg(client, state, CH7006_SUBC_INC1); + ch7006_save_reg(client, state, CH7006_SUBC_INC2); + ch7006_save_reg(client, state, CH7006_SUBC_INC3); + ch7006_save_reg(client, state, CH7006_SUBC_INC4); + ch7006_save_reg(client, state, CH7006_SUBC_INC5); + ch7006_save_reg(client, state, CH7006_SUBC_INC6); + ch7006_save_reg(client, state, CH7006_SUBC_INC7); + ch7006_save_reg(client, state, CH7006_PLL_CONTROL); + ch7006_save_reg(client, state, CH7006_CALC_SUBC_INC0); + + state->regs[CH7006_FFILTER] = (state->regs[CH7006_FFILTER] & 0xf0) | + (state->regs[CH7006_FFILTER] & 0x0c) >> 2 | + (state->regs[CH7006_FFILTER] & 0x03) << 2; +} diff --git a/drivers/gpu/drm/i2c/ch7006_priv.h b/drivers/gpu/drm/i2c/ch7006_priv.h new file mode 100644 index 00000000..17667b7d --- /dev/null +++ b/drivers/gpu/drm/i2c/ch7006_priv.h @@ -0,0 +1,345 @@ +/* + * Copyright (C) 2009 Francisco Jerez. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __DRM_I2C_CH7006_PRIV_H__ +#define __DRM_I2C_CH7006_PRIV_H__ + +#include "drmP.h" +#include "drm_crtc_helper.h" +#include "drm_encoder_slave.h" +#include "i2c/ch7006.h" + +typedef int64_t fixed; +#define fixed1 (1LL << 32) + +enum ch7006_tv_norm { + TV_NORM_PAL, + TV_NORM_PAL_M, + TV_NORM_PAL_N, + TV_NORM_PAL_NC, + TV_NORM_PAL_60, + TV_NORM_NTSC_M, + TV_NORM_NTSC_J, + NUM_TV_NORMS +}; + +struct ch7006_tv_norm_info { + fixed vrefresh; + int vdisplay; + int vtotal; + int hvirtual; + + fixed subc_freq; + fixed black_level; + + uint32_t dispmode; + int voffset; +}; + +struct ch7006_mode { + struct drm_display_mode mode; + + int enc_hdisp; + int enc_vdisp; + + fixed subc_coeff; + uint32_t dispmode; + + uint32_t valid_scales; + uint32_t valid_norms; +}; + +struct ch7006_state { + uint8_t regs[0x26]; +}; + +struct ch7006_priv { + struct ch7006_encoder_params params; + struct ch7006_mode *mode; + + struct ch7006_state state; + struct ch7006_state saved_state; + + struct drm_property *scale_property; + + int select_subconnector; + int subconnector; + int hmargin; + int vmargin; + enum ch7006_tv_norm norm; + int brightness; + int contrast; + int flicker; + int scale; + + int chip_version; + int last_dpms; +}; + +#define to_ch7006_priv(x) \ + ((struct ch7006_priv *)to_encoder_slave(x)->slave_priv) + +extern int ch7006_debug; +extern char *ch7006_tv_norm; +extern int ch7006_scale; + +extern char *ch7006_tv_norm_names[]; +extern struct ch7006_tv_norm_info ch7006_tv_norms[]; +extern struct ch7006_mode ch7006_modes[]; + +struct ch7006_mode *ch7006_lookup_mode(struct drm_encoder *encoder, + struct drm_display_mode *drm_mode); + +void ch7006_setup_levels(struct drm_encoder *encoder); +void ch7006_setup_subcarrier(struct drm_encoder *encoder); +void ch7006_setup_pll(struct drm_encoder *encoder); +void ch7006_setup_power_state(struct drm_encoder *encoder); +void ch7006_setup_properties(struct drm_encoder *encoder); + +void ch7006_write(struct i2c_client *client, uint8_t addr, uint8_t val); +uint8_t ch7006_read(struct i2c_client *client, uint8_t addr); + +void ch7006_state_load(struct i2c_client *client, + struct ch7006_state *state); +void ch7006_state_save(struct i2c_client *client, + struct ch7006_state *state); + +/* Some helper macros */ + +#define ch7006_dbg(client, format, ...) do { \ + if (ch7006_debug) \ + dev_printk(KERN_DEBUG, &client->dev, \ + "%s: " format, __func__, ## __VA_ARGS__); \ + } while (0) +#define ch7006_info(client, format, ...) \ + dev_info(&client->dev, format, __VA_ARGS__) +#define ch7006_err(client, format, ...) \ + dev_err(&client->dev, format, __VA_ARGS__) + +#define __mask(src, bitfield) \ + (((2 << (1 ? bitfield)) - 1) & ~((1 << (0 ? bitfield)) - 1)) +#define mask(bitfield) __mask(bitfield) + +#define __bitf(src, bitfield, x) \ + (((x) >> (src) << (0 ? bitfield)) & __mask(src, bitfield)) +#define bitf(bitfield, x) __bitf(bitfield, x) +#define bitfs(bitfield, s) __bitf(bitfield, bitfield##_##s) +#define setbitf(state, reg, bitfield, x) \ + state->regs[reg] = (state->regs[reg] & ~mask(reg##_##bitfield)) \ + | bitf(reg##_##bitfield, x) + +#define __unbitf(src, bitfield, x) \ + ((x & __mask(src, bitfield)) >> (0 ? bitfield) << (src)) +#define unbitf(bitfield, x) __unbitf(bitfield, x) + +static inline int interpolate(int y0, int y1, int y2, int x) +{ + return y1 + (x < 50 ? y1 - y0 : y2 - y1) * (x - 50) / 50; +} + +static inline int32_t round_fixed(fixed x) +{ + return (x + fixed1/2) >> 32; +} + +#define ch7006_load_reg(client, state, reg) ch7006_write(client, reg, state->regs[reg]) +#define ch7006_save_reg(client, state, reg) state->regs[reg] = ch7006_read(client, reg) + +/* Fixed hardware specs */ + +#define CH7006_FREQ0 14318 +#define CH7006_MAXN 650 +#define CH7006_MAXM 315 + +/* Register definitions */ + +#define CH7006_DISPMODE 0x00 +#define CH7006_DISPMODE_INPUT_RES 0, 7:5 +#define CH7006_DISPMODE_INPUT_RES_512x384 0x0 +#define CH7006_DISPMODE_INPUT_RES_720x400 0x1 +#define CH7006_DISPMODE_INPUT_RES_640x400 0x2 +#define CH7006_DISPMODE_INPUT_RES_640x480 0x3 +#define CH7006_DISPMODE_INPUT_RES_800x600 0x4 +#define CH7006_DISPMODE_INPUT_RES_NATIVE 0x5 +#define CH7006_DISPMODE_OUTPUT_STD 0, 4:3 +#define CH7006_DISPMODE_OUTPUT_STD_PAL 0x0 +#define CH7006_DISPMODE_OUTPUT_STD_NTSC 0x1 +#define CH7006_DISPMODE_OUTPUT_STD_PAL_M 0x2 +#define CH7006_DISPMODE_OUTPUT_STD_NTSC_J 0x3 +#define CH7006_DISPMODE_SCALING_RATIO 0, 2:0 +#define CH7006_DISPMODE_SCALING_RATIO_5_4 0x0 +#define CH7006_DISPMODE_SCALING_RATIO_1_1 0x1 +#define CH7006_DISPMODE_SCALING_RATIO_7_8 0x2 +#define CH7006_DISPMODE_SCALING_RATIO_5_6 0x3 +#define CH7006_DISPMODE_SCALING_RATIO_3_4 0x4 +#define CH7006_DISPMODE_SCALING_RATIO_7_10 0x5 + +#define CH7006_FFILTER 0x01 +#define CH7006_FFILTER_TEXT 0, 5:4 +#define CH7006_FFILTER_LUMA 0, 3:2 +#define CH7006_FFILTER_CHROMA 0, 1:0 +#define CH7006_FFILTER_CHROMA_NO_DCRAWL 0x3 + +#define CH7006_BWIDTH 0x03 +#define CH7006_BWIDTH_5L_FFILER (1 << 7) +#define CH7006_BWIDTH_CVBS_NO_CHROMA (1 << 6) +#define CH7006_BWIDTH_CHROMA 0, 5:4 +#define CH7006_BWIDTH_SVIDEO_YPEAK (1 << 3) +#define CH7006_BWIDTH_SVIDEO_LUMA 0, 2:1 +#define CH7006_BWIDTH_CVBS_LUMA 0, 0:0 + +#define CH7006_INPUT_FORMAT 0x04 +#define CH7006_INPUT_FORMAT_DAC_GAIN (1 << 6) +#define CH7006_INPUT_FORMAT_RGB_PASS_THROUGH (1 << 5) +#define CH7006_INPUT_FORMAT_FORMAT 0, 3:0 +#define CH7006_INPUT_FORMAT_FORMAT_RGB16 0x0 +#define CH7006_INPUT_FORMAT_FORMAT_YCrCb24m16 0x1 +#define CH7006_INPUT_FORMAT_FORMAT_RGB24m16 0x2 +#define CH7006_INPUT_FORMAT_FORMAT_RGB15 0x3 +#define CH7006_INPUT_FORMAT_FORMAT_RGB24m12C 0x4 +#define CH7006_INPUT_FORMAT_FORMAT_RGB24m12I 0x5 +#define CH7006_INPUT_FORMAT_FORMAT_RGB24m8 0x6 +#define CH7006_INPUT_FORMAT_FORMAT_RGB16m8 0x7 +#define CH7006_INPUT_FORMAT_FORMAT_RGB15m8 0x8 +#define CH7006_INPUT_FORMAT_FORMAT_YCrCb24m8 0x9 + +#define CH7006_CLKMODE 0x06 +#define CH7006_CLKMODE_SUBC_LOCK (1 << 7) +#define CH7006_CLKMODE_MASTER (1 << 6) +#define CH7006_CLKMODE_POS_EDGE (1 << 4) +#define CH7006_CLKMODE_XCM 0, 3:2 +#define CH7006_CLKMODE_PCM 0, 1:0 + +#define CH7006_START_ACTIVE 0x07 +#define CH7006_START_ACTIVE_0 0, 7:0 + +#define CH7006_POV 0x08 +#define CH7006_POV_START_ACTIVE_8 8, 2:2 +#define CH7006_POV_HPOS_8 8, 1:1 +#define CH7006_POV_VPOS_8 8, 0:0 + +#define CH7006_BLACK_LEVEL 0x09 +#define CH7006_BLACK_LEVEL_0 0, 7:0 + +#define CH7006_HPOS 0x0a +#define CH7006_HPOS_0 0, 7:0 + +#define CH7006_VPOS 0x0b +#define CH7006_VPOS_0 0, 7:0 + +#define CH7006_INPUT_SYNC 0x0d +#define CH7006_INPUT_SYNC_EMBEDDED (1 << 3) +#define CH7006_INPUT_SYNC_OUTPUT (1 << 2) +#define CH7006_INPUT_SYNC_PVSYNC (1 << 1) +#define CH7006_INPUT_SYNC_PHSYNC (1 << 0) + +#define CH7006_POWER 0x0e +#define CH7006_POWER_SCART (1 << 4) +#define CH7006_POWER_RESET (1 << 3) +#define CH7006_POWER_LEVEL 0, 2:0 +#define CH7006_POWER_LEVEL_CVBS_OFF 0x0 +#define CH7006_POWER_LEVEL_POWER_OFF 0x1 +#define CH7006_POWER_LEVEL_SVIDEO_OFF 0x2 +#define CH7006_POWER_LEVEL_NORMAL 0x3 +#define CH7006_POWER_LEVEL_FULL_POWER_OFF 0x4 + +#define CH7006_DETECT 0x10 +#define CH7006_DETECT_SVIDEO_Y_TEST (1 << 3) +#define CH7006_DETECT_SVIDEO_C_TEST (1 << 2) +#define CH7006_DETECT_CVBS_TEST (1 << 1) +#define CH7006_DETECT_SENSE (1 << 0) + +#define CH7006_CONTRAST 0x11 +#define CH7006_CONTRAST_0 0, 2:0 + +#define CH7006_PLLOV 0x13 +#define CH7006_PLLOV_N_8 8, 2:1 +#define CH7006_PLLOV_M_8 8, 0:0 + +#define CH7006_PLLM 0x14 +#define CH7006_PLLM_0 0, 7:0 + +#define CH7006_PLLN 0x15 +#define CH7006_PLLN_0 0, 7:0 + +#define CH7006_BCLKOUT 0x17 + +#define CH7006_SUBC_INC0 0x18 +#define CH7006_SUBC_INC0_28 28, 3:0 + +#define CH7006_SUBC_INC1 0x19 +#define CH7006_SUBC_INC1_24 24, 3:0 + +#define CH7006_SUBC_INC2 0x1a +#define CH7006_SUBC_INC2_20 20, 3:0 + +#define CH7006_SUBC_INC3 0x1b +#define CH7006_SUBC_INC3_GPIO1_VAL (1 << 7) +#define CH7006_SUBC_INC3_GPIO0_VAL (1 << 6) +#define CH7006_SUBC_INC3_POUT_3_3V (1 << 5) +#define CH7006_SUBC_INC3_POUT_INV (1 << 4) +#define CH7006_SUBC_INC3_16 16, 3:0 + +#define CH7006_SUBC_INC4 0x1c +#define CH7006_SUBC_INC4_GPIO1_IN (1 << 7) +#define CH7006_SUBC_INC4_GPIO0_IN (1 << 6) +#define CH7006_SUBC_INC4_DS_INPUT (1 << 4) +#define CH7006_SUBC_INC4_12 12, 3:0 + +#define CH7006_SUBC_INC5 0x1d +#define CH7006_SUBC_INC5_8 8, 3:0 + +#define CH7006_SUBC_INC6 0x1e +#define CH7006_SUBC_INC6_4 4, 3:0 + +#define CH7006_SUBC_INC7 0x1f +#define CH7006_SUBC_INC7_0 0, 3:0 + +#define CH7006_PLL_CONTROL 0x20 +#define CH7006_PLL_CONTROL_CPI (1 << 5) +#define CH7006_PLL_CONTROL_CAPACITOR (1 << 4) +#define CH7006_PLL_CONTROL_7STAGES (1 << 3) +#define CH7006_PLL_CONTROL_DIGITAL_5V (1 << 2) +#define CH7006_PLL_CONTROL_ANALOG_5V (1 << 1) +#define CH7006_PLL_CONTROL_MEMORY_5V (1 << 0) + +#define CH7006_CALC_SUBC_INC0 0x21 +#define CH7006_CALC_SUBC_INC0_24 24, 4:3 +#define CH7006_CALC_SUBC_INC0_HYST 0, 2:1 +#define CH7006_CALC_SUBC_INC0_AUTO (1 << 0) + +#define CH7006_CALC_SUBC_INC1 0x22 +#define CH7006_CALC_SUBC_INC1_16 16, 7:0 + +#define CH7006_CALC_SUBC_INC2 0x23 +#define CH7006_CALC_SUBC_INC2_8 8, 7:0 + +#define CH7006_CALC_SUBC_INC3 0x24 +#define CH7006_CALC_SUBC_INC3_0 0, 7:0 + +#define CH7006_VERSION_ID 0x25 + +#endif diff --git a/drivers/gpu/drm/i2c/sil164_drv.c b/drivers/gpu/drm/i2c/sil164_drv.c new file mode 100644 index 00000000..0b677329 --- /dev/null +++ b/drivers/gpu/drm/i2c/sil164_drv.c @@ -0,0 +1,462 @@ +/* + * Copyright (C) 2010 Francisco Jerez. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "drmP.h" +#include "drm_crtc_helper.h" +#include "drm_encoder_slave.h" +#include "i2c/sil164.h" + +struct sil164_priv { + struct sil164_encoder_params config; + struct i2c_client *duallink_slave; + + uint8_t saved_state[0x10]; + uint8_t saved_slave_state[0x10]; +}; + +#define to_sil164_priv(x) \ + ((struct sil164_priv *)to_encoder_slave(x)->slave_priv) + +#define sil164_dbg(client, format, ...) do { \ + if (drm_debug & DRM_UT_KMS) \ + dev_printk(KERN_DEBUG, &client->dev, \ + "%s: " format, __func__, ## __VA_ARGS__); \ + } while (0) +#define sil164_info(client, format, ...) \ + dev_info(&client->dev, format, __VA_ARGS__) +#define sil164_err(client, format, ...) \ + dev_err(&client->dev, format, __VA_ARGS__) + +#define SIL164_I2C_ADDR_MASTER 0x38 +#define SIL164_I2C_ADDR_SLAVE 0x39 + +/* HW register definitions */ + +#define SIL164_VENDOR_LO 0x0 +#define SIL164_VENDOR_HI 0x1 +#define SIL164_DEVICE_LO 0x2 +#define SIL164_DEVICE_HI 0x3 +#define SIL164_REVISION 0x4 +#define SIL164_FREQ_MIN 0x6 +#define SIL164_FREQ_MAX 0x7 +#define SIL164_CONTROL0 0x8 +# define SIL164_CONTROL0_POWER_ON 0x01 +# define SIL164_CONTROL0_EDGE_RISING 0x02 +# define SIL164_CONTROL0_INPUT_24BIT 0x04 +# define SIL164_CONTROL0_DUAL_EDGE 0x08 +# define SIL164_CONTROL0_HSYNC_ON 0x10 +# define SIL164_CONTROL0_VSYNC_ON 0x20 +#define SIL164_DETECT 0x9 +# define SIL164_DETECT_INTR_STAT 0x01 +# define SIL164_DETECT_HOTPLUG_STAT 0x02 +# define SIL164_DETECT_RECEIVER_STAT 0x04 +# define SIL164_DETECT_INTR_MODE_RECEIVER 0x00 +# define SIL164_DETECT_INTR_MODE_HOTPLUG 0x08 +# define SIL164_DETECT_OUT_MODE_HIGH 0x00 +# define SIL164_DETECT_OUT_MODE_INTR 0x10 +# define SIL164_DETECT_OUT_MODE_RECEIVER 0x20 +# define SIL164_DETECT_OUT_MODE_HOTPLUG 0x30 +# define SIL164_DETECT_VSWING_STAT 0x80 +#define SIL164_CONTROL1 0xa +# define SIL164_CONTROL1_DESKEW_ENABLE 0x10 +# define SIL164_CONTROL1_DESKEW_INCR_SHIFT 5 +#define SIL164_GPIO 0xb +#define SIL164_CONTROL2 0xc +# define SIL164_CONTROL2_FILTER_ENABLE 0x01 +# define SIL164_CONTROL2_FILTER_SETTING_SHIFT 1 +# define SIL164_CONTROL2_DUALLINK_MASTER 0x40 +# define SIL164_CONTROL2_SYNC_CONT 0x80 +#define SIL164_DUALLINK 0xd +# define SIL164_DUALLINK_ENABLE 0x10 +# define SIL164_DUALLINK_SKEW_SHIFT 5 +#define SIL164_PLLZONE 0xe +# define SIL164_PLLZONE_STAT 0x08 +# define SIL164_PLLZONE_FORCE_ON 0x10 +# define SIL164_PLLZONE_FORCE_HIGH 0x20 + +/* HW access functions */ + +static void +sil164_write(struct i2c_client *client, uint8_t addr, uint8_t val) +{ + uint8_t buf[] = {addr, val}; + int ret; + + ret = i2c_master_send(client, buf, ARRAY_SIZE(buf)); + if (ret < 0) + sil164_err(client, "Error %d writing to subaddress 0x%x\n", + ret, addr); +} + +static uint8_t +sil164_read(struct i2c_client *client, uint8_t addr) +{ + uint8_t val; + int ret; + + ret = i2c_master_send(client, &addr, sizeof(addr)); + if (ret < 0) + goto fail; + + ret = i2c_master_recv(client, &val, sizeof(val)); + if (ret < 0) + goto fail; + + return val; + +fail: + sil164_err(client, "Error %d reading from subaddress 0x%x\n", + ret, addr); + return 0; +} + +static void +sil164_save_state(struct i2c_client *client, uint8_t *state) +{ + int i; + + for (i = 0x8; i <= 0xe; i++) + state[i] = sil164_read(client, i); +} + +static void +sil164_restore_state(struct i2c_client *client, uint8_t *state) +{ + int i; + + for (i = 0x8; i <= 0xe; i++) + sil164_write(client, i, state[i]); +} + +static void +sil164_set_power_state(struct i2c_client *client, bool on) +{ + uint8_t control0 = sil164_read(client, SIL164_CONTROL0); + + if (on) + control0 |= SIL164_CONTROL0_POWER_ON; + else + control0 &= ~SIL164_CONTROL0_POWER_ON; + + sil164_write(client, SIL164_CONTROL0, control0); +} + +static void +sil164_init_state(struct i2c_client *client, + struct sil164_encoder_params *config, + bool duallink) +{ + sil164_write(client, SIL164_CONTROL0, + SIL164_CONTROL0_HSYNC_ON | + SIL164_CONTROL0_VSYNC_ON | + (config->input_edge ? SIL164_CONTROL0_EDGE_RISING : 0) | + (config->input_width ? SIL164_CONTROL0_INPUT_24BIT : 0) | + (config->input_dual ? SIL164_CONTROL0_DUAL_EDGE : 0)); + + sil164_write(client, SIL164_DETECT, + SIL164_DETECT_INTR_STAT | + SIL164_DETECT_OUT_MODE_RECEIVER); + + sil164_write(client, SIL164_CONTROL1, + (config->input_skew ? SIL164_CONTROL1_DESKEW_ENABLE : 0) | + (((config->input_skew + 4) & 0x7) + << SIL164_CONTROL1_DESKEW_INCR_SHIFT)); + + sil164_write(client, SIL164_CONTROL2, + SIL164_CONTROL2_SYNC_CONT | + (config->pll_filter ? 0 : SIL164_CONTROL2_FILTER_ENABLE) | + (4 << SIL164_CONTROL2_FILTER_SETTING_SHIFT)); + + sil164_write(client, SIL164_PLLZONE, 0); + + if (duallink) + sil164_write(client, SIL164_DUALLINK, + SIL164_DUALLINK_ENABLE | + (((config->duallink_skew + 4) & 0x7) + << SIL164_DUALLINK_SKEW_SHIFT)); + else + sil164_write(client, SIL164_DUALLINK, 0); +} + +/* DRM encoder functions */ + +static void +sil164_encoder_set_config(struct drm_encoder *encoder, void *params) +{ + struct sil164_priv *priv = to_sil164_priv(encoder); + + priv->config = *(struct sil164_encoder_params *)params; +} + +static void +sil164_encoder_dpms(struct drm_encoder *encoder, int mode) +{ + struct sil164_priv *priv = to_sil164_priv(encoder); + bool on = (mode == DRM_MODE_DPMS_ON); + bool duallink = (on && encoder->crtc->mode.clock > 165000); + + sil164_set_power_state(drm_i2c_encoder_get_client(encoder), on); + + if (priv->duallink_slave) + sil164_set_power_state(priv->duallink_slave, duallink); +} + +static void +sil164_encoder_save(struct drm_encoder *encoder) +{ + struct sil164_priv *priv = to_sil164_priv(encoder); + + sil164_save_state(drm_i2c_encoder_get_client(encoder), + priv->saved_state); + + if (priv->duallink_slave) + sil164_save_state(priv->duallink_slave, + priv->saved_slave_state); +} + +static void +sil164_encoder_restore(struct drm_encoder *encoder) +{ + struct sil164_priv *priv = to_sil164_priv(encoder); + + sil164_restore_state(drm_i2c_encoder_get_client(encoder), + priv->saved_state); + + if (priv->duallink_slave) + sil164_restore_state(priv->duallink_slave, + priv->saved_slave_state); +} + +static bool +sil164_encoder_mode_fixup(struct drm_encoder *encoder, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + return true; +} + +static int +sil164_encoder_mode_valid(struct drm_encoder *encoder, + struct drm_display_mode *mode) +{ + struct sil164_priv *priv = to_sil164_priv(encoder); + + if (mode->clock < 32000) + return MODE_CLOCK_LOW; + + if (mode->clock > 330000 || + (mode->clock > 165000 && !priv->duallink_slave)) + return MODE_CLOCK_HIGH; + + return MODE_OK; +} + +static void +sil164_encoder_mode_set(struct drm_encoder *encoder, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct sil164_priv *priv = to_sil164_priv(encoder); + bool duallink = adjusted_mode->clock > 165000; + + sil164_init_state(drm_i2c_encoder_get_client(encoder), + &priv->config, duallink); + + if (priv->duallink_slave) + sil164_init_state(priv->duallink_slave, + &priv->config, duallink); + + sil164_encoder_dpms(encoder, DRM_MODE_DPMS_ON); +} + +static enum drm_connector_status +sil164_encoder_detect(struct drm_encoder *encoder, + struct drm_connector *connector) +{ + struct i2c_client *client = drm_i2c_encoder_get_client(encoder); + + if (sil164_read(client, SIL164_DETECT) & SIL164_DETECT_HOTPLUG_STAT) + return connector_status_connected; + else + return connector_status_disconnected; +} + +static int +sil164_encoder_get_modes(struct drm_encoder *encoder, + struct drm_connector *connector) +{ + return 0; +} + +static int +sil164_encoder_create_resources(struct drm_encoder *encoder, + struct drm_connector *connector) +{ + return 0; +} + +static int +sil164_encoder_set_property(struct drm_encoder *encoder, + struct drm_connector *connector, + struct drm_property *property, + uint64_t val) +{ + return 0; +} + +static void +sil164_encoder_destroy(struct drm_encoder *encoder) +{ + struct sil164_priv *priv = to_sil164_priv(encoder); + + if (priv->duallink_slave) + i2c_unregister_device(priv->duallink_slave); + + kfree(priv); + drm_i2c_encoder_destroy(encoder); +} + +static struct drm_encoder_slave_funcs sil164_encoder_funcs = { + .set_config = sil164_encoder_set_config, + .destroy = sil164_encoder_destroy, + .dpms = sil164_encoder_dpms, + .save = sil164_encoder_save, + .restore = sil164_encoder_restore, + .mode_fixup = sil164_encoder_mode_fixup, + .mode_valid = sil164_encoder_mode_valid, + .mode_set = sil164_encoder_mode_set, + .detect = sil164_encoder_detect, + .get_modes = sil164_encoder_get_modes, + .create_resources = sil164_encoder_create_resources, + .set_property = sil164_encoder_set_property, +}; + +/* I2C driver functions */ + +static int +sil164_probe(struct i2c_client *client, const struct i2c_device_id *id) +{ + int vendor = sil164_read(client, SIL164_VENDOR_HI) << 8 | + sil164_read(client, SIL164_VENDOR_LO); + int device = sil164_read(client, SIL164_DEVICE_HI) << 8 | + sil164_read(client, SIL164_DEVICE_LO); + int rev = sil164_read(client, SIL164_REVISION); + + if (vendor != 0x1 || device != 0x6) { + sil164_dbg(client, "Unknown device %x:%x.%x\n", + vendor, device, rev); + return -ENODEV; + } + + sil164_info(client, "Detected device %x:%x.%x\n", + vendor, device, rev); + + return 0; +} + +static int +sil164_remove(struct i2c_client *client) +{ + return 0; +} + +static struct i2c_client * +sil164_detect_slave(struct i2c_client *client) +{ + struct i2c_adapter *adap = client->adapter; + struct i2c_msg msg = { + .addr = SIL164_I2C_ADDR_SLAVE, + .len = 0, + }; + const struct i2c_board_info info = { + I2C_BOARD_INFO("sil164", SIL164_I2C_ADDR_SLAVE) + }; + + if (i2c_transfer(adap, &msg, 1) != 1) { + sil164_dbg(adap, "No dual-link slave found."); + return NULL; + } + + return i2c_new_device(adap, &info); +} + +static int +sil164_encoder_init(struct i2c_client *client, + struct drm_device *dev, + struct drm_encoder_slave *encoder) +{ + struct sil164_priv *priv; + + priv = kzalloc(sizeof(*priv), GFP_KERNEL); + if (!priv) + return -ENOMEM; + + encoder->slave_priv = priv; + encoder->slave_funcs = &sil164_encoder_funcs; + + priv->duallink_slave = sil164_detect_slave(client); + + return 0; +} + +static struct i2c_device_id sil164_ids[] = { + { "sil164", 0 }, + { } +}; +MODULE_DEVICE_TABLE(i2c, sil164_ids); + +static struct drm_i2c_encoder_driver sil164_driver = { + .i2c_driver = { + .probe = sil164_probe, + .remove = sil164_remove, + .driver = { + .name = "sil164", + }, + .id_table = sil164_ids, + }, + .encoder_init = sil164_encoder_init, +}; + +/* Module initialization */ + +static int __init +sil164_init(void) +{ + return drm_i2c_encoder_register(THIS_MODULE, &sil164_driver); +} + +static void __exit +sil164_exit(void) +{ + drm_i2c_encoder_unregister(&sil164_driver); +} + +MODULE_AUTHOR("Francisco Jerez "); +MODULE_DESCRIPTION("Silicon Image sil164 TMDS transmitter driver"); +MODULE_LICENSE("GPL and additional rights"); + +module_init(sil164_init); +module_exit(sil164_exit); diff --git a/drivers/gpu/drm/i810/Makefile b/drivers/gpu/drm/i810/Makefile new file mode 100644 index 00000000..43844eca --- /dev/null +++ b/drivers/gpu/drm/i810/Makefile @@ -0,0 +1,8 @@ +# +# Makefile for the drm device driver. This driver provides support for the +# Direct Rendering Infrastructure (DRI) in XFree86 4.1.0 and higher. + +ccflags-y := -Iinclude/drm +i810-y := i810_drv.o i810_dma.o + +obj-$(CONFIG_DRM_I810) += i810.o diff --git a/drivers/gpu/drm/i810/i810_dma.c b/drivers/gpu/drm/i810/i810_dma.c new file mode 100644 index 00000000..8f371e8d --- /dev/null +++ b/drivers/gpu/drm/i810/i810_dma.c @@ -0,0 +1,1276 @@ +/* i810_dma.c -- DMA support for the i810 -*- linux-c -*- + * Created: Mon Dec 13 01:50:01 1999 by jhartmann@precisioninsight.com + * + * Copyright 1999 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: Rickard E. (Rik) Faith + * Jeff Hartmann + * Keith Whitwell + * + */ + +#include "drmP.h" +#include "drm.h" +#include "i810_drm.h" +#include "i810_drv.h" +#include /* For task queue support */ +#include +#include +#include + +#define I810_BUF_FREE 2 +#define I810_BUF_CLIENT 1 +#define I810_BUF_HARDWARE 0 + +#define I810_BUF_UNMAPPED 0 +#define I810_BUF_MAPPED 1 + +static struct drm_buf *i810_freelist_get(struct drm_device * dev) +{ + struct drm_device_dma *dma = dev->dma; + int i; + int used; + + /* Linear search might not be the best solution */ + + for (i = 0; i < dma->buf_count; i++) { + struct drm_buf *buf = dma->buflist[i]; + drm_i810_buf_priv_t *buf_priv = buf->dev_private; + /* In use is already a pointer */ + used = cmpxchg(buf_priv->in_use, I810_BUF_FREE, + I810_BUF_CLIENT); + if (used == I810_BUF_FREE) + return buf; + } + return NULL; +} + +/* This should only be called if the buffer is not sent to the hardware + * yet, the hardware updates in use for us once its on the ring buffer. + */ + +static int i810_freelist_put(struct drm_device *dev, struct drm_buf *buf) +{ + drm_i810_buf_priv_t *buf_priv = buf->dev_private; + int used; + + /* In use is already a pointer */ + used = cmpxchg(buf_priv->in_use, I810_BUF_CLIENT, I810_BUF_FREE); + if (used != I810_BUF_CLIENT) { + DRM_ERROR("Freeing buffer thats not in use : %d\n", buf->idx); + return -EINVAL; + } + + return 0; +} + +static int i810_mmap_buffers(struct file *filp, struct vm_area_struct *vma) +{ + struct drm_file *priv = filp->private_data; + struct drm_device *dev; + drm_i810_private_t *dev_priv; + struct drm_buf *buf; + drm_i810_buf_priv_t *buf_priv; + + dev = priv->minor->dev; + dev_priv = dev->dev_private; + buf = dev_priv->mmap_buffer; + buf_priv = buf->dev_private; + + vma->vm_flags |= (VM_IO | VM_DONTCOPY); + vma->vm_file = filp; + + buf_priv->currently_mapped = I810_BUF_MAPPED; + + if (io_remap_pfn_range(vma, vma->vm_start, + vma->vm_pgoff, + vma->vm_end - vma->vm_start, vma->vm_page_prot)) + return -EAGAIN; + return 0; +} + +static const struct file_operations i810_buffer_fops = { + .open = drm_open, + .release = drm_release, + .unlocked_ioctl = drm_ioctl, + .mmap = i810_mmap_buffers, + .fasync = drm_fasync, + .llseek = noop_llseek, +}; + +static int i810_map_buffer(struct drm_buf *buf, struct drm_file *file_priv) +{ + struct drm_device *dev = file_priv->minor->dev; + drm_i810_buf_priv_t *buf_priv = buf->dev_private; + drm_i810_private_t *dev_priv = dev->dev_private; + const struct file_operations *old_fops; + int retcode = 0; + + if (buf_priv->currently_mapped == I810_BUF_MAPPED) + return -EINVAL; + + down_write(¤t->mm->mmap_sem); + old_fops = file_priv->filp->f_op; + file_priv->filp->f_op = &i810_buffer_fops; + dev_priv->mmap_buffer = buf; + buf_priv->virtual = (void *)do_mmap(file_priv->filp, 0, buf->total, + PROT_READ | PROT_WRITE, + MAP_SHARED, buf->bus_address); + dev_priv->mmap_buffer = NULL; + file_priv->filp->f_op = old_fops; + if (IS_ERR(buf_priv->virtual)) { + /* Real error */ + DRM_ERROR("mmap error\n"); + retcode = PTR_ERR(buf_priv->virtual); + buf_priv->virtual = NULL; + } + up_write(¤t->mm->mmap_sem); + + return retcode; +} + +static int i810_unmap_buffer(struct drm_buf *buf) +{ + drm_i810_buf_priv_t *buf_priv = buf->dev_private; + int retcode = 0; + + if (buf_priv->currently_mapped != I810_BUF_MAPPED) + return -EINVAL; + + down_write(¤t->mm->mmap_sem); + retcode = do_munmap(current->mm, + (unsigned long)buf_priv->virtual, + (size_t) buf->total); + up_write(¤t->mm->mmap_sem); + + buf_priv->currently_mapped = I810_BUF_UNMAPPED; + buf_priv->virtual = NULL; + + return retcode; +} + +static int i810_dma_get_buffer(struct drm_device *dev, drm_i810_dma_t *d, + struct drm_file *file_priv) +{ + struct drm_buf *buf; + drm_i810_buf_priv_t *buf_priv; + int retcode = 0; + + buf = i810_freelist_get(dev); + if (!buf) { + retcode = -ENOMEM; + DRM_DEBUG("retcode=%d\n", retcode); + return retcode; + } + + retcode = i810_map_buffer(buf, file_priv); + if (retcode) { + i810_freelist_put(dev, buf); + DRM_ERROR("mapbuf failed, retcode %d\n", retcode); + return retcode; + } + buf->file_priv = file_priv; + buf_priv = buf->dev_private; + d->granted = 1; + d->request_idx = buf->idx; + d->request_size = buf->total; + d->virtual = buf_priv->virtual; + + return retcode; +} + +static int i810_dma_cleanup(struct drm_device *dev) +{ + struct drm_device_dma *dma = dev->dma; + + /* Make sure interrupts are disabled here because the uninstall ioctl + * may not have been called from userspace and after dev_private + * is freed, it's too late. + */ + if (drm_core_check_feature(dev, DRIVER_HAVE_IRQ) && dev->irq_enabled) + drm_irq_uninstall(dev); + + if (dev->dev_private) { + int i; + drm_i810_private_t *dev_priv = + (drm_i810_private_t *) dev->dev_private; + + if (dev_priv->ring.virtual_start) + drm_core_ioremapfree(&dev_priv->ring.map, dev); + if (dev_priv->hw_status_page) { + pci_free_consistent(dev->pdev, PAGE_SIZE, + dev_priv->hw_status_page, + dev_priv->dma_status_page); + /* Need to rewrite hardware status page */ + I810_WRITE(0x02080, 0x1ffff000); + } + kfree(dev->dev_private); + dev->dev_private = NULL; + + for (i = 0; i < dma->buf_count; i++) { + struct drm_buf *buf = dma->buflist[i]; + drm_i810_buf_priv_t *buf_priv = buf->dev_private; + + if (buf_priv->kernel_virtual && buf->total) + drm_core_ioremapfree(&buf_priv->map, dev); + } + } + return 0; +} + +static int i810_wait_ring(struct drm_device *dev, int n) +{ + drm_i810_private_t *dev_priv = dev->dev_private; + drm_i810_ring_buffer_t *ring = &(dev_priv->ring); + int iters = 0; + unsigned long end; + unsigned int last_head = I810_READ(LP_RING + RING_HEAD) & HEAD_ADDR; + + end = jiffies + (HZ * 3); + while (ring->space < n) { + ring->head = I810_READ(LP_RING + RING_HEAD) & HEAD_ADDR; + ring->space = ring->head - (ring->tail + 8); + if (ring->space < 0) + ring->space += ring->Size; + + if (ring->head != last_head) { + end = jiffies + (HZ * 3); + last_head = ring->head; + } + + iters++; + if (time_before(end, jiffies)) { + DRM_ERROR("space: %d wanted %d\n", ring->space, n); + DRM_ERROR("lockup\n"); + goto out_wait_ring; + } + udelay(1); + } + +out_wait_ring: + return iters; +} + +static void i810_kernel_lost_context(struct drm_device *dev) +{ + drm_i810_private_t *dev_priv = dev->dev_private; + drm_i810_ring_buffer_t *ring = &(dev_priv->ring); + + ring->head = I810_READ(LP_RING + RING_HEAD) & HEAD_ADDR; + ring->tail = I810_READ(LP_RING + RING_TAIL); + ring->space = ring->head - (ring->tail + 8); + if (ring->space < 0) + ring->space += ring->Size; +} + +static int i810_freelist_init(struct drm_device *dev, drm_i810_private_t *dev_priv) +{ + struct drm_device_dma *dma = dev->dma; + int my_idx = 24; + u32 *hw_status = (u32 *) (dev_priv->hw_status_page + my_idx); + int i; + + if (dma->buf_count > 1019) { + /* Not enough space in the status page for the freelist */ + return -EINVAL; + } + + for (i = 0; i < dma->buf_count; i++) { + struct drm_buf *buf = dma->buflist[i]; + drm_i810_buf_priv_t *buf_priv = buf->dev_private; + + buf_priv->in_use = hw_status++; + buf_priv->my_use_idx = my_idx; + my_idx += 4; + + *buf_priv->in_use = I810_BUF_FREE; + + buf_priv->map.offset = buf->bus_address; + buf_priv->map.size = buf->total; + buf_priv->map.type = _DRM_AGP; + buf_priv->map.flags = 0; + buf_priv->map.mtrr = 0; + + drm_core_ioremap(&buf_priv->map, dev); + buf_priv->kernel_virtual = buf_priv->map.handle; + + } + return 0; +} + +static int i810_dma_initialize(struct drm_device *dev, + drm_i810_private_t *dev_priv, + drm_i810_init_t *init) +{ + struct drm_map_list *r_list; + memset(dev_priv, 0, sizeof(drm_i810_private_t)); + + list_for_each_entry(r_list, &dev->maplist, head) { + if (r_list->map && + r_list->map->type == _DRM_SHM && + r_list->map->flags & _DRM_CONTAINS_LOCK) { + dev_priv->sarea_map = r_list->map; + break; + } + } + if (!dev_priv->sarea_map) { + dev->dev_private = (void *)dev_priv; + i810_dma_cleanup(dev); + DRM_ERROR("can not find sarea!\n"); + return -EINVAL; + } + dev_priv->mmio_map = drm_core_findmap(dev, init->mmio_offset); + if (!dev_priv->mmio_map) { + dev->dev_private = (void *)dev_priv; + i810_dma_cleanup(dev); + DRM_ERROR("can not find mmio map!\n"); + return -EINVAL; + } + dev->agp_buffer_token = init->buffers_offset; + dev->agp_buffer_map = drm_core_findmap(dev, init->buffers_offset); + if (!dev->agp_buffer_map) { + dev->dev_private = (void *)dev_priv; + i810_dma_cleanup(dev); + DRM_ERROR("can not find dma buffer map!\n"); + return -EINVAL; + } + + dev_priv->sarea_priv = (drm_i810_sarea_t *) + ((u8 *) dev_priv->sarea_map->handle + init->sarea_priv_offset); + + dev_priv->ring.Start = init->ring_start; + dev_priv->ring.End = init->ring_end; + dev_priv->ring.Size = init->ring_size; + + dev_priv->ring.map.offset = dev->agp->base + init->ring_start; + dev_priv->ring.map.size = init->ring_size; + dev_priv->ring.map.type = _DRM_AGP; + dev_priv->ring.map.flags = 0; + dev_priv->ring.map.mtrr = 0; + + drm_core_ioremap(&dev_priv->ring.map, dev); + + if (dev_priv->ring.map.handle == NULL) { + dev->dev_private = (void *)dev_priv; + i810_dma_cleanup(dev); + DRM_ERROR("can not ioremap virtual address for" + " ring buffer\n"); + return -ENOMEM; + } + + dev_priv->ring.virtual_start = dev_priv->ring.map.handle; + + dev_priv->ring.tail_mask = dev_priv->ring.Size - 1; + + dev_priv->w = init->w; + dev_priv->h = init->h; + dev_priv->pitch = init->pitch; + dev_priv->back_offset = init->back_offset; + dev_priv->depth_offset = init->depth_offset; + dev_priv->front_offset = init->front_offset; + + dev_priv->overlay_offset = init->overlay_offset; + dev_priv->overlay_physical = init->overlay_physical; + + dev_priv->front_di1 = init->front_offset | init->pitch_bits; + dev_priv->back_di1 = init->back_offset | init->pitch_bits; + dev_priv->zi1 = init->depth_offset | init->pitch_bits; + + /* Program Hardware Status Page */ + dev_priv->hw_status_page = + pci_alloc_consistent(dev->pdev, PAGE_SIZE, + &dev_priv->dma_status_page); + if (!dev_priv->hw_status_page) { + dev->dev_private = (void *)dev_priv; + i810_dma_cleanup(dev); + DRM_ERROR("Can not allocate hardware status page\n"); + return -ENOMEM; + } + memset(dev_priv->hw_status_page, 0, PAGE_SIZE); + DRM_DEBUG("hw status page @ %p\n", dev_priv->hw_status_page); + + I810_WRITE(0x02080, dev_priv->dma_status_page); + DRM_DEBUG("Enabled hardware status page\n"); + + /* Now we need to init our freelist */ + if (i810_freelist_init(dev, dev_priv) != 0) { + dev->dev_private = (void *)dev_priv; + i810_dma_cleanup(dev); + DRM_ERROR("Not enough space in the status page for" + " the freelist\n"); + return -ENOMEM; + } + dev->dev_private = (void *)dev_priv; + + return 0; +} + +static int i810_dma_init(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + drm_i810_private_t *dev_priv; + drm_i810_init_t *init = data; + int retcode = 0; + + switch (init->func) { + case I810_INIT_DMA_1_4: + DRM_INFO("Using v1.4 init.\n"); + dev_priv = kmalloc(sizeof(drm_i810_private_t), GFP_KERNEL); + if (dev_priv == NULL) + return -ENOMEM; + retcode = i810_dma_initialize(dev, dev_priv, init); + break; + + case I810_CLEANUP_DMA: + DRM_INFO("DMA Cleanup\n"); + retcode = i810_dma_cleanup(dev); + break; + default: + return -EINVAL; + } + + return retcode; +} + +/* Most efficient way to verify state for the i810 is as it is + * emitted. Non-conformant state is silently dropped. + * + * Use 'volatile' & local var tmp to force the emitted values to be + * identical to the verified ones. + */ +static void i810EmitContextVerified(struct drm_device *dev, + volatile unsigned int *code) +{ + drm_i810_private_t *dev_priv = dev->dev_private; + int i, j = 0; + unsigned int tmp; + RING_LOCALS; + + BEGIN_LP_RING(I810_CTX_SETUP_SIZE); + + OUT_RING(GFX_OP_COLOR_FACTOR); + OUT_RING(code[I810_CTXREG_CF1]); + + OUT_RING(GFX_OP_STIPPLE); + OUT_RING(code[I810_CTXREG_ST1]); + + for (i = 4; i < I810_CTX_SETUP_SIZE; i++) { + tmp = code[i]; + + if ((tmp & (7 << 29)) == (3 << 29) && + (tmp & (0x1f << 24)) < (0x1d << 24)) { + OUT_RING(tmp); + j++; + } else + printk("constext state dropped!!!\n"); + } + + if (j & 1) + OUT_RING(0); + + ADVANCE_LP_RING(); +} + +static void i810EmitTexVerified(struct drm_device *dev, volatile unsigned int *code) +{ + drm_i810_private_t *dev_priv = dev->dev_private; + int i, j = 0; + unsigned int tmp; + RING_LOCALS; + + BEGIN_LP_RING(I810_TEX_SETUP_SIZE); + + OUT_RING(GFX_OP_MAP_INFO); + OUT_RING(code[I810_TEXREG_MI1]); + OUT_RING(code[I810_TEXREG_MI2]); + OUT_RING(code[I810_TEXREG_MI3]); + + for (i = 4; i < I810_TEX_SETUP_SIZE; i++) { + tmp = code[i]; + + if ((tmp & (7 << 29)) == (3 << 29) && + (tmp & (0x1f << 24)) < (0x1d << 24)) { + OUT_RING(tmp); + j++; + } else + printk("texture state dropped!!!\n"); + } + + if (j & 1) + OUT_RING(0); + + ADVANCE_LP_RING(); +} + +/* Need to do some additional checking when setting the dest buffer. + */ +static void i810EmitDestVerified(struct drm_device *dev, + volatile unsigned int *code) +{ + drm_i810_private_t *dev_priv = dev->dev_private; + unsigned int tmp; + RING_LOCALS; + + BEGIN_LP_RING(I810_DEST_SETUP_SIZE + 2); + + tmp = code[I810_DESTREG_DI1]; + if (tmp == dev_priv->front_di1 || tmp == dev_priv->back_di1) { + OUT_RING(CMD_OP_DESTBUFFER_INFO); + OUT_RING(tmp); + } else + DRM_DEBUG("bad di1 %x (allow %x or %x)\n", + tmp, dev_priv->front_di1, dev_priv->back_di1); + + /* invarient: + */ + OUT_RING(CMD_OP_Z_BUFFER_INFO); + OUT_RING(dev_priv->zi1); + + OUT_RING(GFX_OP_DESTBUFFER_VARS); + OUT_RING(code[I810_DESTREG_DV1]); + + OUT_RING(GFX_OP_DRAWRECT_INFO); + OUT_RING(code[I810_DESTREG_DR1]); + OUT_RING(code[I810_DESTREG_DR2]); + OUT_RING(code[I810_DESTREG_DR3]); + OUT_RING(code[I810_DESTREG_DR4]); + OUT_RING(0); + + ADVANCE_LP_RING(); +} + +static void i810EmitState(struct drm_device *dev) +{ + drm_i810_private_t *dev_priv = dev->dev_private; + drm_i810_sarea_t *sarea_priv = dev_priv->sarea_priv; + unsigned int dirty = sarea_priv->dirty; + + DRM_DEBUG("%x\n", dirty); + + if (dirty & I810_UPLOAD_BUFFERS) { + i810EmitDestVerified(dev, sarea_priv->BufferState); + sarea_priv->dirty &= ~I810_UPLOAD_BUFFERS; + } + + if (dirty & I810_UPLOAD_CTX) { + i810EmitContextVerified(dev, sarea_priv->ContextState); + sarea_priv->dirty &= ~I810_UPLOAD_CTX; + } + + if (dirty & I810_UPLOAD_TEX0) { + i810EmitTexVerified(dev, sarea_priv->TexState[0]); + sarea_priv->dirty &= ~I810_UPLOAD_TEX0; + } + + if (dirty & I810_UPLOAD_TEX1) { + i810EmitTexVerified(dev, sarea_priv->TexState[1]); + sarea_priv->dirty &= ~I810_UPLOAD_TEX1; + } +} + +/* need to verify + */ +static void i810_dma_dispatch_clear(struct drm_device *dev, int flags, + unsigned int clear_color, + unsigned int clear_zval) +{ + drm_i810_private_t *dev_priv = dev->dev_private; + drm_i810_sarea_t *sarea_priv = dev_priv->sarea_priv; + int nbox = sarea_priv->nbox; + struct drm_clip_rect *pbox = sarea_priv->boxes; + int pitch = dev_priv->pitch; + int cpp = 2; + int i; + RING_LOCALS; + + if (dev_priv->current_page == 1) { + unsigned int tmp = flags; + + flags &= ~(I810_FRONT | I810_BACK); + if (tmp & I810_FRONT) + flags |= I810_BACK; + if (tmp & I810_BACK) + flags |= I810_FRONT; + } + + i810_kernel_lost_context(dev); + + if (nbox > I810_NR_SAREA_CLIPRECTS) + nbox = I810_NR_SAREA_CLIPRECTS; + + for (i = 0; i < nbox; i++, pbox++) { + unsigned int x = pbox->x1; + unsigned int y = pbox->y1; + unsigned int width = (pbox->x2 - x) * cpp; + unsigned int height = pbox->y2 - y; + unsigned int start = y * pitch + x * cpp; + + if (pbox->x1 > pbox->x2 || + pbox->y1 > pbox->y2 || + pbox->x2 > dev_priv->w || pbox->y2 > dev_priv->h) + continue; + + if (flags & I810_FRONT) { + BEGIN_LP_RING(6); + OUT_RING(BR00_BITBLT_CLIENT | BR00_OP_COLOR_BLT | 0x3); + OUT_RING(BR13_SOLID_PATTERN | (0xF0 << 16) | pitch); + OUT_RING((height << 16) | width); + OUT_RING(start); + OUT_RING(clear_color); + OUT_RING(0); + ADVANCE_LP_RING(); + } + + if (flags & I810_BACK) { + BEGIN_LP_RING(6); + OUT_RING(BR00_BITBLT_CLIENT | BR00_OP_COLOR_BLT | 0x3); + OUT_RING(BR13_SOLID_PATTERN | (0xF0 << 16) | pitch); + OUT_RING((height << 16) | width); + OUT_RING(dev_priv->back_offset + start); + OUT_RING(clear_color); + OUT_RING(0); + ADVANCE_LP_RING(); + } + + if (flags & I810_DEPTH) { + BEGIN_LP_RING(6); + OUT_RING(BR00_BITBLT_CLIENT | BR00_OP_COLOR_BLT | 0x3); + OUT_RING(BR13_SOLID_PATTERN | (0xF0 << 16) | pitch); + OUT_RING((height << 16) | width); + OUT_RING(dev_priv->depth_offset + start); + OUT_RING(clear_zval); + OUT_RING(0); + ADVANCE_LP_RING(); + } + } +} + +static void i810_dma_dispatch_swap(struct drm_device *dev) +{ + drm_i810_private_t *dev_priv = dev->dev_private; + drm_i810_sarea_t *sarea_priv = dev_priv->sarea_priv; + int nbox = sarea_priv->nbox; + struct drm_clip_rect *pbox = sarea_priv->boxes; + int pitch = dev_priv->pitch; + int cpp = 2; + int i; + RING_LOCALS; + + DRM_DEBUG("swapbuffers\n"); + + i810_kernel_lost_context(dev); + + if (nbox > I810_NR_SAREA_CLIPRECTS) + nbox = I810_NR_SAREA_CLIPRECTS; + + for (i = 0; i < nbox; i++, pbox++) { + unsigned int w = pbox->x2 - pbox->x1; + unsigned int h = pbox->y2 - pbox->y1; + unsigned int dst = pbox->x1 * cpp + pbox->y1 * pitch; + unsigned int start = dst; + + if (pbox->x1 > pbox->x2 || + pbox->y1 > pbox->y2 || + pbox->x2 > dev_priv->w || pbox->y2 > dev_priv->h) + continue; + + BEGIN_LP_RING(6); + OUT_RING(BR00_BITBLT_CLIENT | BR00_OP_SRC_COPY_BLT | 0x4); + OUT_RING(pitch | (0xCC << 16)); + OUT_RING((h << 16) | (w * cpp)); + if (dev_priv->current_page == 0) + OUT_RING(dev_priv->front_offset + start); + else + OUT_RING(dev_priv->back_offset + start); + OUT_RING(pitch); + if (dev_priv->current_page == 0) + OUT_RING(dev_priv->back_offset + start); + else + OUT_RING(dev_priv->front_offset + start); + ADVANCE_LP_RING(); + } +} + +static void i810_dma_dispatch_vertex(struct drm_device *dev, + struct drm_buf *buf, int discard, int used) +{ + drm_i810_private_t *dev_priv = dev->dev_private; + drm_i810_buf_priv_t *buf_priv = buf->dev_private; + drm_i810_sarea_t *sarea_priv = dev_priv->sarea_priv; + struct drm_clip_rect *box = sarea_priv->boxes; + int nbox = sarea_priv->nbox; + unsigned long address = (unsigned long)buf->bus_address; + unsigned long start = address - dev->agp->base; + int i = 0; + RING_LOCALS; + + i810_kernel_lost_context(dev); + + if (nbox > I810_NR_SAREA_CLIPRECTS) + nbox = I810_NR_SAREA_CLIPRECTS; + + if (used > 4 * 1024) + used = 0; + + if (sarea_priv->dirty) + i810EmitState(dev); + + if (buf_priv->currently_mapped == I810_BUF_MAPPED) { + unsigned int prim = (sarea_priv->vertex_prim & PR_MASK); + + *(u32 *) buf_priv->kernel_virtual = + ((GFX_OP_PRIMITIVE | prim | ((used / 4) - 2))); + + if (used & 4) { + *(u32 *) ((char *) buf_priv->kernel_virtual + used) = 0; + used += 4; + } + + i810_unmap_buffer(buf); + } + + if (used) { + do { + if (i < nbox) { + BEGIN_LP_RING(4); + OUT_RING(GFX_OP_SCISSOR | SC_UPDATE_SCISSOR | + SC_ENABLE); + OUT_RING(GFX_OP_SCISSOR_INFO); + OUT_RING(box[i].x1 | (box[i].y1 << 16)); + OUT_RING((box[i].x2 - + 1) | ((box[i].y2 - 1) << 16)); + ADVANCE_LP_RING(); + } + + BEGIN_LP_RING(4); + OUT_RING(CMD_OP_BATCH_BUFFER); + OUT_RING(start | BB1_PROTECTED); + OUT_RING(start + used - 4); + OUT_RING(0); + ADVANCE_LP_RING(); + + } while (++i < nbox); + } + + if (discard) { + dev_priv->counter++; + + (void)cmpxchg(buf_priv->in_use, I810_BUF_CLIENT, + I810_BUF_HARDWARE); + + BEGIN_LP_RING(8); + OUT_RING(CMD_STORE_DWORD_IDX); + OUT_RING(20); + OUT_RING(dev_priv->counter); + OUT_RING(CMD_STORE_DWORD_IDX); + OUT_RING(buf_priv->my_use_idx); + OUT_RING(I810_BUF_FREE); + OUT_RING(CMD_REPORT_HEAD); + OUT_RING(0); + ADVANCE_LP_RING(); + } +} + +static void i810_dma_dispatch_flip(struct drm_device *dev) +{ + drm_i810_private_t *dev_priv = dev->dev_private; + int pitch = dev_priv->pitch; + RING_LOCALS; + + DRM_DEBUG("page=%d pfCurrentPage=%d\n", + dev_priv->current_page, + dev_priv->sarea_priv->pf_current_page); + + i810_kernel_lost_context(dev); + + BEGIN_LP_RING(2); + OUT_RING(INST_PARSER_CLIENT | INST_OP_FLUSH | INST_FLUSH_MAP_CACHE); + OUT_RING(0); + ADVANCE_LP_RING(); + + BEGIN_LP_RING(I810_DEST_SETUP_SIZE + 2); + /* On i815 at least ASYNC is buggy */ + /* pitch<<5 is from 11.2.8 p158, + its the pitch / 8 then left shifted 8, + so (pitch >> 3) << 8 */ + OUT_RING(CMD_OP_FRONTBUFFER_INFO | (pitch << 5) /*| ASYNC_FLIP */ ); + if (dev_priv->current_page == 0) { + OUT_RING(dev_priv->back_offset); + dev_priv->current_page = 1; + } else { + OUT_RING(dev_priv->front_offset); + dev_priv->current_page = 0; + } + OUT_RING(0); + ADVANCE_LP_RING(); + + BEGIN_LP_RING(2); + OUT_RING(CMD_OP_WAIT_FOR_EVENT | WAIT_FOR_PLANE_A_FLIP); + OUT_RING(0); + ADVANCE_LP_RING(); + + /* Increment the frame counter. The client-side 3D driver must + * throttle the framerate by waiting for this value before + * performing the swapbuffer ioctl. + */ + dev_priv->sarea_priv->pf_current_page = dev_priv->current_page; + +} + +static void i810_dma_quiescent(struct drm_device *dev) +{ + drm_i810_private_t *dev_priv = dev->dev_private; + RING_LOCALS; + + i810_kernel_lost_context(dev); + + BEGIN_LP_RING(4); + OUT_RING(INST_PARSER_CLIENT | INST_OP_FLUSH | INST_FLUSH_MAP_CACHE); + OUT_RING(CMD_REPORT_HEAD); + OUT_RING(0); + OUT_RING(0); + ADVANCE_LP_RING(); + + i810_wait_ring(dev, dev_priv->ring.Size - 8); +} + +static int i810_flush_queue(struct drm_device *dev) +{ + drm_i810_private_t *dev_priv = dev->dev_private; + struct drm_device_dma *dma = dev->dma; + int i, ret = 0; + RING_LOCALS; + + i810_kernel_lost_context(dev); + + BEGIN_LP_RING(2); + OUT_RING(CMD_REPORT_HEAD); + OUT_RING(0); + ADVANCE_LP_RING(); + + i810_wait_ring(dev, dev_priv->ring.Size - 8); + + for (i = 0; i < dma->buf_count; i++) { + struct drm_buf *buf = dma->buflist[i]; + drm_i810_buf_priv_t *buf_priv = buf->dev_private; + + int used = cmpxchg(buf_priv->in_use, I810_BUF_HARDWARE, + I810_BUF_FREE); + + if (used == I810_BUF_HARDWARE) + DRM_DEBUG("reclaimed from HARDWARE\n"); + if (used == I810_BUF_CLIENT) + DRM_DEBUG("still on client\n"); + } + + return ret; +} + +/* Must be called with the lock held */ +static void i810_reclaim_buffers(struct drm_device *dev, + struct drm_file *file_priv) +{ + struct drm_device_dma *dma = dev->dma; + int i; + + if (!dma) + return; + if (!dev->dev_private) + return; + if (!dma->buflist) + return; + + i810_flush_queue(dev); + + for (i = 0; i < dma->buf_count; i++) { + struct drm_buf *buf = dma->buflist[i]; + drm_i810_buf_priv_t *buf_priv = buf->dev_private; + + if (buf->file_priv == file_priv && buf_priv) { + int used = cmpxchg(buf_priv->in_use, I810_BUF_CLIENT, + I810_BUF_FREE); + + if (used == I810_BUF_CLIENT) + DRM_DEBUG("reclaimed from client\n"); + if (buf_priv->currently_mapped == I810_BUF_MAPPED) + buf_priv->currently_mapped = I810_BUF_UNMAPPED; + } + } +} + +static int i810_flush_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + LOCK_TEST_WITH_RETURN(dev, file_priv); + + i810_flush_queue(dev); + return 0; +} + +static int i810_dma_vertex(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_device_dma *dma = dev->dma; + drm_i810_private_t *dev_priv = (drm_i810_private_t *) dev->dev_private; + u32 *hw_status = dev_priv->hw_status_page; + drm_i810_sarea_t *sarea_priv = (drm_i810_sarea_t *) + dev_priv->sarea_priv; + drm_i810_vertex_t *vertex = data; + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + DRM_DEBUG("idx %d used %d discard %d\n", + vertex->idx, vertex->used, vertex->discard); + + if (vertex->idx < 0 || vertex->idx > dma->buf_count) + return -EINVAL; + + i810_dma_dispatch_vertex(dev, + dma->buflist[vertex->idx], + vertex->discard, vertex->used); + + atomic_add(vertex->used, &dev->counts[_DRM_STAT_SECONDARY]); + atomic_inc(&dev->counts[_DRM_STAT_DMA]); + sarea_priv->last_enqueue = dev_priv->counter - 1; + sarea_priv->last_dispatch = (int)hw_status[5]; + + return 0; +} + +static int i810_clear_bufs(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + drm_i810_clear_t *clear = data; + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + /* GH: Someone's doing nasty things... */ + if (!dev->dev_private) + return -EINVAL; + + i810_dma_dispatch_clear(dev, clear->flags, + clear->clear_color, clear->clear_depth); + return 0; +} + +static int i810_swap_bufs(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + DRM_DEBUG("\n"); + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + i810_dma_dispatch_swap(dev); + return 0; +} + +static int i810_getage(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + drm_i810_private_t *dev_priv = (drm_i810_private_t *) dev->dev_private; + u32 *hw_status = dev_priv->hw_status_page; + drm_i810_sarea_t *sarea_priv = (drm_i810_sarea_t *) + dev_priv->sarea_priv; + + sarea_priv->last_dispatch = (int)hw_status[5]; + return 0; +} + +static int i810_getbuf(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + int retcode = 0; + drm_i810_dma_t *d = data; + drm_i810_private_t *dev_priv = (drm_i810_private_t *) dev->dev_private; + u32 *hw_status = dev_priv->hw_status_page; + drm_i810_sarea_t *sarea_priv = (drm_i810_sarea_t *) + dev_priv->sarea_priv; + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + d->granted = 0; + + retcode = i810_dma_get_buffer(dev, d, file_priv); + + DRM_DEBUG("i810_dma: %d returning %d, granted = %d\n", + task_pid_nr(current), retcode, d->granted); + + sarea_priv->last_dispatch = (int)hw_status[5]; + + return retcode; +} + +static int i810_copybuf(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + /* Never copy - 2.4.x doesn't need it */ + return 0; +} + +static int i810_docopy(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + /* Never copy - 2.4.x doesn't need it */ + return 0; +} + +static void i810_dma_dispatch_mc(struct drm_device *dev, struct drm_buf *buf, int used, + unsigned int last_render) +{ + drm_i810_private_t *dev_priv = dev->dev_private; + drm_i810_buf_priv_t *buf_priv = buf->dev_private; + drm_i810_sarea_t *sarea_priv = dev_priv->sarea_priv; + unsigned long address = (unsigned long)buf->bus_address; + unsigned long start = address - dev->agp->base; + int u; + RING_LOCALS; + + i810_kernel_lost_context(dev); + + u = cmpxchg(buf_priv->in_use, I810_BUF_CLIENT, I810_BUF_HARDWARE); + if (u != I810_BUF_CLIENT) + DRM_DEBUG("MC found buffer that isn't mine!\n"); + + if (used > 4 * 1024) + used = 0; + + sarea_priv->dirty = 0x7f; + + DRM_DEBUG("addr 0x%lx, used 0x%x\n", address, used); + + dev_priv->counter++; + DRM_DEBUG("dispatch counter : %ld\n", dev_priv->counter); + DRM_DEBUG("start : %lx\n", start); + DRM_DEBUG("used : %d\n", used); + DRM_DEBUG("start + used - 4 : %ld\n", start + used - 4); + + if (buf_priv->currently_mapped == I810_BUF_MAPPED) { + if (used & 4) { + *(u32 *) ((char *) buf_priv->virtual + used) = 0; + used += 4; + } + + i810_unmap_buffer(buf); + } + BEGIN_LP_RING(4); + OUT_RING(CMD_OP_BATCH_BUFFER); + OUT_RING(start | BB1_PROTECTED); + OUT_RING(start + used - 4); + OUT_RING(0); + ADVANCE_LP_RING(); + + BEGIN_LP_RING(8); + OUT_RING(CMD_STORE_DWORD_IDX); + OUT_RING(buf_priv->my_use_idx); + OUT_RING(I810_BUF_FREE); + OUT_RING(0); + + OUT_RING(CMD_STORE_DWORD_IDX); + OUT_RING(16); + OUT_RING(last_render); + OUT_RING(0); + ADVANCE_LP_RING(); +} + +static int i810_dma_mc(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_device_dma *dma = dev->dma; + drm_i810_private_t *dev_priv = (drm_i810_private_t *) dev->dev_private; + u32 *hw_status = dev_priv->hw_status_page; + drm_i810_sarea_t *sarea_priv = (drm_i810_sarea_t *) + dev_priv->sarea_priv; + drm_i810_mc_t *mc = data; + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + if (mc->idx >= dma->buf_count || mc->idx < 0) + return -EINVAL; + + i810_dma_dispatch_mc(dev, dma->buflist[mc->idx], mc->used, + mc->last_render); + + atomic_add(mc->used, &dev->counts[_DRM_STAT_SECONDARY]); + atomic_inc(&dev->counts[_DRM_STAT_DMA]); + sarea_priv->last_enqueue = dev_priv->counter - 1; + sarea_priv->last_dispatch = (int)hw_status[5]; + + return 0; +} + +static int i810_rstatus(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + drm_i810_private_t *dev_priv = (drm_i810_private_t *) dev->dev_private; + + return (int)(((u32 *) (dev_priv->hw_status_page))[4]); +} + +static int i810_ov0_info(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + drm_i810_private_t *dev_priv = (drm_i810_private_t *) dev->dev_private; + drm_i810_overlay_t *ov = data; + + ov->offset = dev_priv->overlay_offset; + ov->physical = dev_priv->overlay_physical; + + return 0; +} + +static int i810_fstatus(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + drm_i810_private_t *dev_priv = (drm_i810_private_t *) dev->dev_private; + + LOCK_TEST_WITH_RETURN(dev, file_priv); + return I810_READ(0x30008); +} + +static int i810_ov0_flip(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + drm_i810_private_t *dev_priv = (drm_i810_private_t *) dev->dev_private; + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + /* Tell the overlay to update */ + I810_WRITE(0x30000, dev_priv->overlay_physical | 0x80000000); + + return 0; +} + +/* Not sure why this isn't set all the time: + */ +static void i810_do_init_pageflip(struct drm_device *dev) +{ + drm_i810_private_t *dev_priv = dev->dev_private; + + DRM_DEBUG("\n"); + dev_priv->page_flipping = 1; + dev_priv->current_page = 0; + dev_priv->sarea_priv->pf_current_page = dev_priv->current_page; +} + +static int i810_do_cleanup_pageflip(struct drm_device *dev) +{ + drm_i810_private_t *dev_priv = dev->dev_private; + + DRM_DEBUG("\n"); + if (dev_priv->current_page != 0) + i810_dma_dispatch_flip(dev); + + dev_priv->page_flipping = 0; + return 0; +} + +static int i810_flip_bufs(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + drm_i810_private_t *dev_priv = dev->dev_private; + + DRM_DEBUG("\n"); + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + if (!dev_priv->page_flipping) + i810_do_init_pageflip(dev); + + i810_dma_dispatch_flip(dev); + return 0; +} + +int i810_driver_load(struct drm_device *dev, unsigned long flags) +{ + /* i810 has 4 more counters */ + dev->counters += 4; + dev->types[6] = _DRM_STAT_IRQ; + dev->types[7] = _DRM_STAT_PRIMARY; + dev->types[8] = _DRM_STAT_SECONDARY; + dev->types[9] = _DRM_STAT_DMA; + + return 0; +} + +void i810_driver_lastclose(struct drm_device *dev) +{ + i810_dma_cleanup(dev); +} + +void i810_driver_preclose(struct drm_device *dev, struct drm_file *file_priv) +{ + if (dev->dev_private) { + drm_i810_private_t *dev_priv = dev->dev_private; + if (dev_priv->page_flipping) + i810_do_cleanup_pageflip(dev); + } +} + +void i810_driver_reclaim_buffers_locked(struct drm_device *dev, + struct drm_file *file_priv) +{ + i810_reclaim_buffers(dev, file_priv); +} + +int i810_driver_dma_quiescent(struct drm_device *dev) +{ + i810_dma_quiescent(dev); + return 0; +} + +struct drm_ioctl_desc i810_ioctls[] = { + DRM_IOCTL_DEF_DRV(I810_INIT, i810_dma_init, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I810_VERTEX, i810_dma_vertex, DRM_AUTH|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I810_CLEAR, i810_clear_bufs, DRM_AUTH|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I810_FLUSH, i810_flush_ioctl, DRM_AUTH|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I810_GETAGE, i810_getage, DRM_AUTH|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I810_GETBUF, i810_getbuf, DRM_AUTH|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I810_SWAP, i810_swap_bufs, DRM_AUTH|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I810_COPY, i810_copybuf, DRM_AUTH|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I810_DOCOPY, i810_docopy, DRM_AUTH|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I810_OV0INFO, i810_ov0_info, DRM_AUTH|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I810_FSTATUS, i810_fstatus, DRM_AUTH|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I810_OV0FLIP, i810_ov0_flip, DRM_AUTH|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I810_MC, i810_dma_mc, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I810_RSTATUS, i810_rstatus, DRM_AUTH|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I810_FLIP, i810_flip_bufs, DRM_AUTH|DRM_UNLOCKED), +}; + +int i810_max_ioctl = DRM_ARRAY_SIZE(i810_ioctls); + +/** + * Determine if the device really is AGP or not. + * + * All Intel graphics chipsets are treated as AGP, even if they are really + * PCI-e. + * + * \param dev The device to be tested. + * + * \returns + * A value of 1 is always retured to indictate every i810 is AGP. + */ +int i810_driver_device_is_agp(struct drm_device *dev) +{ + return 1; +} diff --git a/drivers/gpu/drm/i810/i810_drv.c b/drivers/gpu/drm/i810/i810_drv.c new file mode 100644 index 00000000..6f98d059 --- /dev/null +++ b/drivers/gpu/drm/i810/i810_drv.c @@ -0,0 +1,100 @@ +/* i810_drv.c -- I810 driver -*- linux-c -*- + * Created: Mon Dec 13 01:56:22 1999 by jhartmann@precisioninsight.com + * + * Copyright 1999 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * VA LINUX SYSTEMS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: + * Rickard E. (Rik) Faith + * Jeff Hartmann + * Gareth Hughes + */ + +#include "drmP.h" +#include "drm.h" +#include "i810_drm.h" +#include "i810_drv.h" + +#include "drm_pciids.h" + +static struct pci_device_id pciidlist[] = { + i810_PCI_IDS +}; + +static struct drm_driver driver = { + .driver_features = + DRIVER_USE_AGP | DRIVER_REQUIRE_AGP | DRIVER_USE_MTRR | + DRIVER_HAVE_DMA | DRIVER_DMA_QUEUE, + .dev_priv_size = sizeof(drm_i810_buf_priv_t), + .load = i810_driver_load, + .lastclose = i810_driver_lastclose, + .preclose = i810_driver_preclose, + .device_is_agp = i810_driver_device_is_agp, + .reclaim_buffers_locked = i810_driver_reclaim_buffers_locked, + .dma_quiescent = i810_driver_dma_quiescent, + .ioctls = i810_ioctls, + .fops = { + .owner = THIS_MODULE, + .open = drm_open, + .release = drm_release, + .unlocked_ioctl = drm_ioctl, + .mmap = drm_mmap, + .poll = drm_poll, + .fasync = drm_fasync, + .llseek = noop_llseek, + }, + + .name = DRIVER_NAME, + .desc = DRIVER_DESC, + .date = DRIVER_DATE, + .major = DRIVER_MAJOR, + .minor = DRIVER_MINOR, + .patchlevel = DRIVER_PATCHLEVEL, +}; + +static struct pci_driver i810_pci_driver = { + .name = DRIVER_NAME, + .id_table = pciidlist, +}; + +static int __init i810_init(void) +{ + if (num_possible_cpus() > 1) { + pr_err("drm/i810 does not support SMP\n"); + return -EINVAL; + } + driver.num_ioctls = i810_max_ioctl; + return drm_pci_init(&driver, &i810_pci_driver); +} + +static void __exit i810_exit(void) +{ + drm_pci_exit(&driver, &i810_pci_driver); +} + +module_init(i810_init); +module_exit(i810_exit); + +MODULE_AUTHOR(DRIVER_AUTHOR); +MODULE_DESCRIPTION(DRIVER_DESC); +MODULE_LICENSE("GPL and additional rights"); diff --git a/drivers/gpu/drm/i810/i810_drv.h b/drivers/gpu/drm/i810/i810_drv.h new file mode 100644 index 00000000..c9339f48 --- /dev/null +++ b/drivers/gpu/drm/i810/i810_drv.h @@ -0,0 +1,245 @@ +/* i810_drv.h -- Private header for the Matrox g200/g400 driver -*- linux-c -*- + * Created: Mon Dec 13 01:50:01 1999 by jhartmann@precisioninsight.com + * + * Copyright 1999 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California. + * All rights reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: Rickard E. (Rik) Faith + * Jeff Hartmann + * + */ + +#ifndef _I810_DRV_H_ +#define _I810_DRV_H_ + +/* General customization: + */ + +#define DRIVER_AUTHOR "VA Linux Systems Inc." + +#define DRIVER_NAME "i810" +#define DRIVER_DESC "Intel i810" +#define DRIVER_DATE "20030605" + +/* Interface history + * + * 1.1 - XFree86 4.1 + * 1.2 - XvMC interfaces + * - XFree86 4.2 + * 1.2.1 - Disable copying code (leave stub ioctls for backwards compatibility) + * - Remove requirement for interrupt (leave stubs again) + * 1.3 - Add page flipping. + * 1.4 - fix DRM interface + */ +#define DRIVER_MAJOR 1 +#define DRIVER_MINOR 4 +#define DRIVER_PATCHLEVEL 0 + +typedef struct drm_i810_buf_priv { + u32 *in_use; + int my_use_idx; + int currently_mapped; + void *virtual; + void *kernel_virtual; + drm_local_map_t map; +} drm_i810_buf_priv_t; + +typedef struct _drm_i810_ring_buffer { + int tail_mask; + unsigned long Start; + unsigned long End; + unsigned long Size; + u8 *virtual_start; + int head; + int tail; + int space; + drm_local_map_t map; +} drm_i810_ring_buffer_t; + +typedef struct drm_i810_private { + struct drm_local_map *sarea_map; + struct drm_local_map *mmio_map; + + drm_i810_sarea_t *sarea_priv; + drm_i810_ring_buffer_t ring; + + void *hw_status_page; + unsigned long counter; + + dma_addr_t dma_status_page; + + struct drm_buf *mmap_buffer; + + u32 front_di1, back_di1, zi1; + + int back_offset; + int depth_offset; + int overlay_offset; + int overlay_physical; + int w, h; + int pitch; + int back_pitch; + int depth_pitch; + + int do_boxes; + int dma_used; + + int current_page; + int page_flipping; + + wait_queue_head_t irq_queue; + atomic_t irq_received; + atomic_t irq_emitted; + + int front_offset; +} drm_i810_private_t; + + /* i810_dma.c */ +extern int i810_driver_dma_quiescent(struct drm_device *dev); +extern void i810_driver_reclaim_buffers_locked(struct drm_device *dev, + struct drm_file *file_priv); +extern int i810_driver_load(struct drm_device *, unsigned long flags); +extern void i810_driver_lastclose(struct drm_device *dev); +extern void i810_driver_preclose(struct drm_device *dev, + struct drm_file *file_priv); +extern void i810_driver_reclaim_buffers_locked(struct drm_device *dev, + struct drm_file *file_priv); +extern int i810_driver_device_is_agp(struct drm_device *dev); + +extern long i810_ioctl(struct file *file, unsigned int cmd, unsigned long arg); +extern struct drm_ioctl_desc i810_ioctls[]; +extern int i810_max_ioctl; + +#define I810_BASE(reg) ((unsigned long) \ + dev_priv->mmio_map->handle) +#define I810_ADDR(reg) (I810_BASE(reg) + reg) +#define I810_DEREF(reg) (*(__volatile__ int *)I810_ADDR(reg)) +#define I810_READ(reg) I810_DEREF(reg) +#define I810_WRITE(reg, val) do { I810_DEREF(reg) = val; } while (0) +#define I810_DEREF16(reg) (*(__volatile__ u16 *)I810_ADDR(reg)) +#define I810_READ16(reg) I810_DEREF16(reg) +#define I810_WRITE16(reg, val) do { I810_DEREF16(reg) = val; } while (0) + +#define I810_VERBOSE 0 +#define RING_LOCALS unsigned int outring, ringmask; \ + volatile char *virt; + +#define BEGIN_LP_RING(n) do { \ + if (I810_VERBOSE) \ + DRM_DEBUG("BEGIN_LP_RING(%d)\n", n); \ + if (dev_priv->ring.space < n*4) \ + i810_wait_ring(dev, n*4); \ + dev_priv->ring.space -= n*4; \ + outring = dev_priv->ring.tail; \ + ringmask = dev_priv->ring.tail_mask; \ + virt = dev_priv->ring.virtual_start; \ +} while (0) + +#define ADVANCE_LP_RING() do { \ + if (I810_VERBOSE) \ + DRM_DEBUG("ADVANCE_LP_RING\n"); \ + dev_priv->ring.tail = outring; \ + I810_WRITE(LP_RING + RING_TAIL, outring); \ +} while (0) + +#define OUT_RING(n) do { \ + if (I810_VERBOSE) \ + DRM_DEBUG(" OUT_RING %x\n", (int)(n)); \ + *(volatile unsigned int *)(virt + outring) = n; \ + outring += 4; \ + outring &= ringmask; \ +} while (0) + +#define GFX_OP_USER_INTERRUPT ((0<<29)|(2<<23)) +#define GFX_OP_BREAKPOINT_INTERRUPT ((0<<29)|(1<<23)) +#define CMD_REPORT_HEAD (7<<23) +#define CMD_STORE_DWORD_IDX ((0x21<<23) | 0x1) +#define CMD_OP_BATCH_BUFFER ((0x0<<29)|(0x30<<23)|0x1) + +#define INST_PARSER_CLIENT 0x00000000 +#define INST_OP_FLUSH 0x02000000 +#define INST_FLUSH_MAP_CACHE 0x00000001 + +#define BB1_START_ADDR_MASK (~0x7) +#define BB1_PROTECTED (1<<0) +#define BB1_UNPROTECTED (0<<0) +#define BB2_END_ADDR_MASK (~0x7) + +#define I810REG_HWSTAM 0x02098 +#define I810REG_INT_IDENTITY_R 0x020a4 +#define I810REG_INT_MASK_R 0x020a8 +#define I810REG_INT_ENABLE_R 0x020a0 + +#define LP_RING 0x2030 +#define HP_RING 0x2040 +#define RING_TAIL 0x00 +#define TAIL_ADDR 0x000FFFF8 +#define RING_HEAD 0x04 +#define HEAD_WRAP_COUNT 0xFFE00000 +#define HEAD_WRAP_ONE 0x00200000 +#define HEAD_ADDR 0x001FFFFC +#define RING_START 0x08 +#define START_ADDR 0x00FFFFF8 +#define RING_LEN 0x0C +#define RING_NR_PAGES 0x000FF000 +#define RING_REPORT_MASK 0x00000006 +#define RING_REPORT_64K 0x00000002 +#define RING_REPORT_128K 0x00000004 +#define RING_NO_REPORT 0x00000000 +#define RING_VALID_MASK 0x00000001 +#define RING_VALID 0x00000001 +#define RING_INVALID 0x00000000 + +#define GFX_OP_SCISSOR ((0x3<<29)|(0x1c<<24)|(0x10<<19)) +#define SC_UPDATE_SCISSOR (0x1<<1) +#define SC_ENABLE_MASK (0x1<<0) +#define SC_ENABLE (0x1<<0) + +#define GFX_OP_SCISSOR_INFO ((0x3<<29)|(0x1d<<24)|(0x81<<16)|(0x1)) +#define SCI_YMIN_MASK (0xffff<<16) +#define SCI_XMIN_MASK (0xffff<<0) +#define SCI_YMAX_MASK (0xffff<<16) +#define SCI_XMAX_MASK (0xffff<<0) + +#define GFX_OP_COLOR_FACTOR ((0x3<<29)|(0x1d<<24)|(0x1<<16)|0x0) +#define GFX_OP_STIPPLE ((0x3<<29)|(0x1d<<24)|(0x83<<16)) +#define GFX_OP_MAP_INFO ((0x3<<29)|(0x1d<<24)|0x2) +#define GFX_OP_DESTBUFFER_VARS ((0x3<<29)|(0x1d<<24)|(0x85<<16)|0x0) +#define GFX_OP_DRAWRECT_INFO ((0x3<<29)|(0x1d<<24)|(0x80<<16)|(0x3)) +#define GFX_OP_PRIMITIVE ((0x3<<29)|(0x1f<<24)) + +#define CMD_OP_Z_BUFFER_INFO ((0x0<<29)|(0x16<<23)) +#define CMD_OP_DESTBUFFER_INFO ((0x0<<29)|(0x15<<23)) +#define CMD_OP_FRONTBUFFER_INFO ((0x0<<29)|(0x14<<23)) +#define CMD_OP_WAIT_FOR_EVENT ((0x0<<29)|(0x03<<23)) + +#define BR00_BITBLT_CLIENT 0x40000000 +#define BR00_OP_COLOR_BLT 0x10000000 +#define BR00_OP_SRC_COPY_BLT 0x10C00000 +#define BR13_SOLID_PATTERN 0x80000000 + +#define WAIT_FOR_PLANE_A_SCANLINES (1<<1) +#define WAIT_FOR_PLANE_A_FLIP (1<<2) +#define WAIT_FOR_VBLANK (1<<3) + +#endif diff --git a/drivers/gpu/drm/i915/Makefile b/drivers/gpu/drm/i915/Makefile new file mode 100644 index 00000000..0ae6a7c5 --- /dev/null +++ b/drivers/gpu/drm/i915/Makefile @@ -0,0 +1,44 @@ +# +# Makefile for the drm device driver. This driver provides support for the +# Direct Rendering Infrastructure (DRI) in XFree86 4.1.0 and higher. + +ccflags-y := -Iinclude/drm +i915-y := i915_drv.o i915_dma.o i915_irq.o i915_mem.o \ + i915_debugfs.o \ + i915_suspend.o \ + i915_gem.o \ + i915_gem_debug.o \ + i915_gem_evict.o \ + i915_gem_execbuffer.o \ + i915_gem_gtt.o \ + i915_gem_tiling.o \ + i915_trace_points.o \ + intel_display.o \ + intel_crt.o \ + intel_lvds.o \ + intel_bios.o \ + intel_dp.o \ + intel_hdmi.o \ + intel_sdvo.o \ + intel_modes.o \ + intel_panel.o \ + intel_i2c.o \ + intel_fb.o \ + intel_tv.o \ + intel_dvo.o \ + intel_ringbuffer.o \ + intel_overlay.o \ + intel_opregion.o \ + dvo_ch7xxx.o \ + dvo_ch7017.o \ + dvo_ivch.o \ + dvo_tfp410.o \ + dvo_sil164.o + +i915-$(CONFIG_COMPAT) += i915_ioc32.o + +i915-$(CONFIG_ACPI) += intel_acpi.o + +obj-$(CONFIG_DRM_I915) += i915.o + +CFLAGS_i915_trace_points.o := -I$(src) diff --git a/drivers/gpu/drm/i915/dvo.h b/drivers/gpu/drm/i915/dvo.h new file mode 100644 index 00000000..8c2ad014 --- /dev/null +++ b/drivers/gpu/drm/i915/dvo.h @@ -0,0 +1,144 @@ +/* + * Copyright © 2006 Eric Anholt + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that copyright + * notice and this permission notice appear in supporting documentation, and + * that the name of the copyright holders not be used in advertising or + * publicity pertaining to distribution of the software without specific, + * written prior permission. The copyright holders make no representations + * about the suitability of this software for any purpose. It is provided "as + * is" without express or implied warranty. + * + * THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, + * INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO + * EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY SPECIAL, INDIRECT OR + * CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, + * DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER + * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE + * OF THIS SOFTWARE. + */ + +#ifndef _INTEL_DVO_H +#define _INTEL_DVO_H + +#include +#include "drmP.h" +#include "drm.h" +#include "drm_crtc.h" +#include "intel_drv.h" + +struct intel_dvo_device { + const char *name; + int type; + /* DVOA/B/C output register */ + u32 dvo_reg; + /* GPIO register used for i2c bus to control this device */ + u32 gpio; + int slave_addr; + + const struct intel_dvo_dev_ops *dev_ops; + void *dev_priv; + struct i2c_adapter *i2c_bus; +}; + +struct intel_dvo_dev_ops { + /* + * Initialize the device at startup time. + * Returns NULL if the device does not exist. + */ + bool (*init)(struct intel_dvo_device *dvo, + struct i2c_adapter *i2cbus); + + /* + * Called to allow the output a chance to create properties after the + * RandR objects have been created. + */ + void (*create_resources)(struct intel_dvo_device *dvo); + + /* + * Turn on/off output or set intermediate power levels if available. + * + * Unsupported intermediate modes drop to the lower power setting. + * If the mode is DPMSModeOff, the output must be disabled, + * as the DPLL may be disabled afterwards. + */ + void (*dpms)(struct intel_dvo_device *dvo, int mode); + + /* + * Callback for testing a video mode for a given output. + * + * This function should only check for cases where a mode can't + * be supported on the output specifically, and not represent + * generic CRTC limitations. + * + * \return MODE_OK if the mode is valid, or another MODE_* otherwise. + */ + int (*mode_valid)(struct intel_dvo_device *dvo, + struct drm_display_mode *mode); + + /* + * Callback to adjust the mode to be set in the CRTC. + * + * This allows an output to adjust the clock or even the entire set of + * timings, which is used for panels with fixed timings or for + * buses with clock limitations. + */ + bool (*mode_fixup)(struct intel_dvo_device *dvo, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode); + + /* + * Callback for preparing mode changes on an output + */ + void (*prepare)(struct intel_dvo_device *dvo); + + /* + * Callback for committing mode changes on an output + */ + void (*commit)(struct intel_dvo_device *dvo); + + /* + * Callback for setting up a video mode after fixups have been made. + * + * This is only called while the output is disabled. The dpms callback + * must be all that's necessary for the output, to turn the output on + * after this function is called. + */ + void (*mode_set)(struct intel_dvo_device *dvo, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode); + + /* + * Probe for a connected output, and return detect_status. + */ + enum drm_connector_status (*detect)(struct intel_dvo_device *dvo); + + /** + * Query the device for the modes it provides. + * + * This function may also update MonInfo, mm_width, and mm_height. + * + * \return singly-linked list of modes or NULL if no modes found. + */ + struct drm_display_mode *(*get_modes)(struct intel_dvo_device *dvo); + + /** + * Clean up driver-specific bits of the output + */ + void (*destroy) (struct intel_dvo_device *dvo); + + /** + * Debugging hook to dump device registers to log file + */ + void (*dump_regs)(struct intel_dvo_device *dvo); +}; + +extern struct intel_dvo_dev_ops sil164_ops; +extern struct intel_dvo_dev_ops ch7xxx_ops; +extern struct intel_dvo_dev_ops ivch_ops; +extern struct intel_dvo_dev_ops tfp410_ops; +extern struct intel_dvo_dev_ops ch7017_ops; + +#endif /* _INTEL_DVO_H */ diff --git a/drivers/gpu/drm/i915/dvo_ch7017.c b/drivers/gpu/drm/i915/dvo_ch7017.c new file mode 100644 index 00000000..d3e8c540 --- /dev/null +++ b/drivers/gpu/drm/i915/dvo_ch7017.c @@ -0,0 +1,401 @@ +/* + * Copyright © 2006 Intel Corporation + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Eric Anholt + * + */ + +#include "dvo.h" + +#define CH7017_TV_DISPLAY_MODE 0x00 +#define CH7017_FLICKER_FILTER 0x01 +#define CH7017_VIDEO_BANDWIDTH 0x02 +#define CH7017_TEXT_ENHANCEMENT 0x03 +#define CH7017_START_ACTIVE_VIDEO 0x04 +#define CH7017_HORIZONTAL_POSITION 0x05 +#define CH7017_VERTICAL_POSITION 0x06 +#define CH7017_BLACK_LEVEL 0x07 +#define CH7017_CONTRAST_ENHANCEMENT 0x08 +#define CH7017_TV_PLL 0x09 +#define CH7017_TV_PLL_M 0x0a +#define CH7017_TV_PLL_N 0x0b +#define CH7017_SUB_CARRIER_0 0x0c +#define CH7017_CIV_CONTROL 0x10 +#define CH7017_CIV_0 0x11 +#define CH7017_CHROMA_BOOST 0x14 +#define CH7017_CLOCK_MODE 0x1c +#define CH7017_INPUT_CLOCK 0x1d +#define CH7017_GPIO_CONTROL 0x1e +#define CH7017_INPUT_DATA_FORMAT 0x1f +#define CH7017_CONNECTION_DETECT 0x20 +#define CH7017_DAC_CONTROL 0x21 +#define CH7017_BUFFERED_CLOCK_OUTPUT 0x22 +#define CH7017_DEFEAT_VSYNC 0x47 +#define CH7017_TEST_PATTERN 0x48 + +#define CH7017_POWER_MANAGEMENT 0x49 +/** Enables the TV output path. */ +#define CH7017_TV_EN (1 << 0) +#define CH7017_DAC0_POWER_DOWN (1 << 1) +#define CH7017_DAC1_POWER_DOWN (1 << 2) +#define CH7017_DAC2_POWER_DOWN (1 << 3) +#define CH7017_DAC3_POWER_DOWN (1 << 4) +/** Powers down the TV out block, and DAC0-3 */ +#define CH7017_TV_POWER_DOWN_EN (1 << 5) + +#define CH7017_VERSION_ID 0x4a + +#define CH7017_DEVICE_ID 0x4b +#define CH7017_DEVICE_ID_VALUE 0x1b +#define CH7018_DEVICE_ID_VALUE 0x1a +#define CH7019_DEVICE_ID_VALUE 0x19 + +#define CH7017_XCLK_D2_ADJUST 0x53 +#define CH7017_UP_SCALER_COEFF_0 0x55 +#define CH7017_UP_SCALER_COEFF_1 0x56 +#define CH7017_UP_SCALER_COEFF_2 0x57 +#define CH7017_UP_SCALER_COEFF_3 0x58 +#define CH7017_UP_SCALER_COEFF_4 0x59 +#define CH7017_UP_SCALER_VERTICAL_INC_0 0x5a +#define CH7017_UP_SCALER_VERTICAL_INC_1 0x5b +#define CH7017_GPIO_INVERT 0x5c +#define CH7017_UP_SCALER_HORIZONTAL_INC_0 0x5d +#define CH7017_UP_SCALER_HORIZONTAL_INC_1 0x5e + +#define CH7017_HORIZONTAL_ACTIVE_PIXEL_INPUT 0x5f +/**< Low bits of horizontal active pixel input */ + +#define CH7017_ACTIVE_INPUT_LINE_OUTPUT 0x60 +/** High bits of horizontal active pixel input */ +#define CH7017_LVDS_HAP_INPUT_MASK (0x7 << 0) +/** High bits of vertical active line output */ +#define CH7017_LVDS_VAL_HIGH_MASK (0x7 << 3) + +#define CH7017_VERTICAL_ACTIVE_LINE_OUTPUT 0x61 +/**< Low bits of vertical active line output */ + +#define CH7017_HORIZONTAL_ACTIVE_PIXEL_OUTPUT 0x62 +/**< Low bits of horizontal active pixel output */ + +#define CH7017_LVDS_POWER_DOWN 0x63 +/** High bits of horizontal active pixel output */ +#define CH7017_LVDS_HAP_HIGH_MASK (0x7 << 0) +/** Enables the LVDS power down state transition */ +#define CH7017_LVDS_POWER_DOWN_EN (1 << 6) +/** Enables the LVDS upscaler */ +#define CH7017_LVDS_UPSCALER_EN (1 << 7) +#define CH7017_LVDS_POWER_DOWN_DEFAULT_RESERVED 0x08 + +#define CH7017_LVDS_ENCODING 0x64 +#define CH7017_LVDS_DITHER_2D (1 << 2) +#define CH7017_LVDS_DITHER_DIS (1 << 3) +#define CH7017_LVDS_DUAL_CHANNEL_EN (1 << 4) +#define CH7017_LVDS_24_BIT (1 << 5) + +#define CH7017_LVDS_ENCODING_2 0x65 + +#define CH7017_LVDS_PLL_CONTROL 0x66 +/** Enables the LVDS panel output path */ +#define CH7017_LVDS_PANEN (1 << 0) +/** Enables the LVDS panel backlight */ +#define CH7017_LVDS_BKLEN (1 << 3) + +#define CH7017_POWER_SEQUENCING_T1 0x67 +#define CH7017_POWER_SEQUENCING_T2 0x68 +#define CH7017_POWER_SEQUENCING_T3 0x69 +#define CH7017_POWER_SEQUENCING_T4 0x6a +#define CH7017_POWER_SEQUENCING_T5 0x6b +#define CH7017_GPIO_DRIVER_TYPE 0x6c +#define CH7017_GPIO_DATA 0x6d +#define CH7017_GPIO_DIRECTION_CONTROL 0x6e + +#define CH7017_LVDS_PLL_FEEDBACK_DIV 0x71 +# define CH7017_LVDS_PLL_FEED_BACK_DIVIDER_SHIFT 4 +# define CH7017_LVDS_PLL_FEED_FORWARD_DIVIDER_SHIFT 0 +# define CH7017_LVDS_PLL_FEEDBACK_DEFAULT_RESERVED 0x80 + +#define CH7017_LVDS_PLL_VCO_CONTROL 0x72 +# define CH7017_LVDS_PLL_VCO_DEFAULT_RESERVED 0x80 +# define CH7017_LVDS_PLL_VCO_SHIFT 4 +# define CH7017_LVDS_PLL_POST_SCALE_DIV_SHIFT 0 + +#define CH7017_OUTPUTS_ENABLE 0x73 +# define CH7017_CHARGE_PUMP_LOW 0x0 +# define CH7017_CHARGE_PUMP_HIGH 0x3 +# define CH7017_LVDS_CHANNEL_A (1 << 3) +# define CH7017_LVDS_CHANNEL_B (1 << 4) +# define CH7017_TV_DAC_A (1 << 5) +# define CH7017_TV_DAC_B (1 << 6) +# define CH7017_DDC_SELECT_DC2 (1 << 7) + +#define CH7017_LVDS_OUTPUT_AMPLITUDE 0x74 +#define CH7017_LVDS_PLL_EMI_REDUCTION 0x75 +#define CH7017_LVDS_POWER_DOWN_FLICKER 0x76 + +#define CH7017_LVDS_CONTROL_2 0x78 +# define CH7017_LOOP_FILTER_SHIFT 5 +# define CH7017_PHASE_DETECTOR_SHIFT 0 + +#define CH7017_BANG_LIMIT_CONTROL 0x7f + +struct ch7017_priv { + uint8_t dummy; +}; + +static void ch7017_dump_regs(struct intel_dvo_device *dvo); +static void ch7017_dpms(struct intel_dvo_device *dvo, int mode); + +static bool ch7017_read(struct intel_dvo_device *dvo, u8 addr, u8 *val) +{ + struct i2c_msg msgs[] = { + { + .addr = dvo->slave_addr, + .flags = 0, + .len = 1, + .buf = &addr, + }, + { + .addr = dvo->slave_addr, + .flags = I2C_M_RD, + .len = 1, + .buf = val, + } + }; + return i2c_transfer(dvo->i2c_bus, msgs, 2) == 2; +} + +static bool ch7017_write(struct intel_dvo_device *dvo, u8 addr, u8 val) +{ + uint8_t buf[2] = { addr, val }; + struct i2c_msg msg = { + .addr = dvo->slave_addr, + .flags = 0, + .len = 2, + .buf = buf, + }; + return i2c_transfer(dvo->i2c_bus, &msg, 1) == 1; +} + +/** Probes for a CH7017 on the given bus and slave address. */ +static bool ch7017_init(struct intel_dvo_device *dvo, + struct i2c_adapter *adapter) +{ + struct ch7017_priv *priv; + const char *str; + u8 val; + + priv = kzalloc(sizeof(struct ch7017_priv), GFP_KERNEL); + if (priv == NULL) + return false; + + dvo->i2c_bus = adapter; + dvo->dev_priv = priv; + + if (!ch7017_read(dvo, CH7017_DEVICE_ID, &val)) + goto fail; + + switch (val) { + case CH7017_DEVICE_ID_VALUE: + str = "ch7017"; + break; + case CH7018_DEVICE_ID_VALUE: + str = "ch7018"; + break; + case CH7019_DEVICE_ID_VALUE: + str = "ch7019"; + break; + default: + DRM_DEBUG_KMS("ch701x not detected, got %d: from %s " + "slave %d.\n", + val, adapter->name,dvo->slave_addr); + goto fail; + } + + DRM_DEBUG_KMS("%s detected on %s, addr %d\n", + str, adapter->name, dvo->slave_addr); + return true; + +fail: + kfree(priv); + return false; +} + +static enum drm_connector_status ch7017_detect(struct intel_dvo_device *dvo) +{ + return connector_status_connected; +} + +static enum drm_mode_status ch7017_mode_valid(struct intel_dvo_device *dvo, + struct drm_display_mode *mode) +{ + if (mode->clock > 160000) + return MODE_CLOCK_HIGH; + + return MODE_OK; +} + +static void ch7017_mode_set(struct intel_dvo_device *dvo, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + uint8_t lvds_pll_feedback_div, lvds_pll_vco_control; + uint8_t outputs_enable, lvds_control_2, lvds_power_down; + uint8_t horizontal_active_pixel_input; + uint8_t horizontal_active_pixel_output, vertical_active_line_output; + uint8_t active_input_line_output; + + DRM_DEBUG_KMS("Registers before mode setting\n"); + ch7017_dump_regs(dvo); + + /* LVDS PLL settings from page 75 of 7017-7017ds.pdf*/ + if (mode->clock < 100000) { + outputs_enable = CH7017_LVDS_CHANNEL_A | CH7017_CHARGE_PUMP_LOW; + lvds_pll_feedback_div = CH7017_LVDS_PLL_FEEDBACK_DEFAULT_RESERVED | + (2 << CH7017_LVDS_PLL_FEED_BACK_DIVIDER_SHIFT) | + (13 << CH7017_LVDS_PLL_FEED_FORWARD_DIVIDER_SHIFT); + lvds_pll_vco_control = CH7017_LVDS_PLL_VCO_DEFAULT_RESERVED | + (2 << CH7017_LVDS_PLL_VCO_SHIFT) | + (3 << CH7017_LVDS_PLL_POST_SCALE_DIV_SHIFT); + lvds_control_2 = (1 << CH7017_LOOP_FILTER_SHIFT) | + (0 << CH7017_PHASE_DETECTOR_SHIFT); + } else { + outputs_enable = CH7017_LVDS_CHANNEL_A | CH7017_CHARGE_PUMP_HIGH; + lvds_pll_feedback_div = CH7017_LVDS_PLL_FEEDBACK_DEFAULT_RESERVED | + (2 << CH7017_LVDS_PLL_FEED_BACK_DIVIDER_SHIFT) | + (3 << CH7017_LVDS_PLL_FEED_FORWARD_DIVIDER_SHIFT); + lvds_pll_feedback_div = 35; + lvds_control_2 = (3 << CH7017_LOOP_FILTER_SHIFT) | + (0 << CH7017_PHASE_DETECTOR_SHIFT); + if (1) { /* XXX: dual channel panel detection. Assume yes for now. */ + outputs_enable |= CH7017_LVDS_CHANNEL_B; + lvds_pll_vco_control = CH7017_LVDS_PLL_VCO_DEFAULT_RESERVED | + (2 << CH7017_LVDS_PLL_VCO_SHIFT) | + (13 << CH7017_LVDS_PLL_POST_SCALE_DIV_SHIFT); + } else { + lvds_pll_vco_control = CH7017_LVDS_PLL_VCO_DEFAULT_RESERVED | + (1 << CH7017_LVDS_PLL_VCO_SHIFT) | + (13 << CH7017_LVDS_PLL_POST_SCALE_DIV_SHIFT); + } + } + + horizontal_active_pixel_input = mode->hdisplay & 0x00ff; + + vertical_active_line_output = mode->vdisplay & 0x00ff; + horizontal_active_pixel_output = mode->hdisplay & 0x00ff; + + active_input_line_output = ((mode->hdisplay & 0x0700) >> 8) | + (((mode->vdisplay & 0x0700) >> 8) << 3); + + lvds_power_down = CH7017_LVDS_POWER_DOWN_DEFAULT_RESERVED | + (mode->hdisplay & 0x0700) >> 8; + + ch7017_dpms(dvo, DRM_MODE_DPMS_OFF); + ch7017_write(dvo, CH7017_HORIZONTAL_ACTIVE_PIXEL_INPUT, + horizontal_active_pixel_input); + ch7017_write(dvo, CH7017_HORIZONTAL_ACTIVE_PIXEL_OUTPUT, + horizontal_active_pixel_output); + ch7017_write(dvo, CH7017_VERTICAL_ACTIVE_LINE_OUTPUT, + vertical_active_line_output); + ch7017_write(dvo, CH7017_ACTIVE_INPUT_LINE_OUTPUT, + active_input_line_output); + ch7017_write(dvo, CH7017_LVDS_PLL_VCO_CONTROL, lvds_pll_vco_control); + ch7017_write(dvo, CH7017_LVDS_PLL_FEEDBACK_DIV, lvds_pll_feedback_div); + ch7017_write(dvo, CH7017_LVDS_CONTROL_2, lvds_control_2); + ch7017_write(dvo, CH7017_OUTPUTS_ENABLE, outputs_enable); + + /* Turn the LVDS back on with new settings. */ + ch7017_write(dvo, CH7017_LVDS_POWER_DOWN, lvds_power_down); + + DRM_DEBUG_KMS("Registers after mode setting\n"); + ch7017_dump_regs(dvo); +} + +/* set the CH7017 power state */ +static void ch7017_dpms(struct intel_dvo_device *dvo, int mode) +{ + uint8_t val; + + ch7017_read(dvo, CH7017_LVDS_POWER_DOWN, &val); + + /* Turn off TV/VGA, and never turn it on since we don't support it. */ + ch7017_write(dvo, CH7017_POWER_MANAGEMENT, + CH7017_DAC0_POWER_DOWN | + CH7017_DAC1_POWER_DOWN | + CH7017_DAC2_POWER_DOWN | + CH7017_DAC3_POWER_DOWN | + CH7017_TV_POWER_DOWN_EN); + + if (mode == DRM_MODE_DPMS_ON) { + /* Turn on the LVDS */ + ch7017_write(dvo, CH7017_LVDS_POWER_DOWN, + val & ~CH7017_LVDS_POWER_DOWN_EN); + } else { + /* Turn off the LVDS */ + ch7017_write(dvo, CH7017_LVDS_POWER_DOWN, + val | CH7017_LVDS_POWER_DOWN_EN); + } + + /* XXX: Should actually wait for update power status somehow */ + msleep(20); +} + +static void ch7017_dump_regs(struct intel_dvo_device *dvo) +{ + uint8_t val; + +#define DUMP(reg) \ +do { \ + ch7017_read(dvo, reg, &val); \ + DRM_DEBUG_KMS(#reg ": %02x\n", val); \ +} while (0) + + DUMP(CH7017_HORIZONTAL_ACTIVE_PIXEL_INPUT); + DUMP(CH7017_HORIZONTAL_ACTIVE_PIXEL_OUTPUT); + DUMP(CH7017_VERTICAL_ACTIVE_LINE_OUTPUT); + DUMP(CH7017_ACTIVE_INPUT_LINE_OUTPUT); + DUMP(CH7017_LVDS_PLL_VCO_CONTROL); + DUMP(CH7017_LVDS_PLL_FEEDBACK_DIV); + DUMP(CH7017_LVDS_CONTROL_2); + DUMP(CH7017_OUTPUTS_ENABLE); + DUMP(CH7017_LVDS_POWER_DOWN); +} + +static void ch7017_destroy(struct intel_dvo_device *dvo) +{ + struct ch7017_priv *priv = dvo->dev_priv; + + if (priv) { + kfree(priv); + dvo->dev_priv = NULL; + } +} + +struct intel_dvo_dev_ops ch7017_ops = { + .init = ch7017_init, + .detect = ch7017_detect, + .mode_valid = ch7017_mode_valid, + .mode_set = ch7017_mode_set, + .dpms = ch7017_dpms, + .dump_regs = ch7017_dump_regs, + .destroy = ch7017_destroy, +}; diff --git a/drivers/gpu/drm/i915/dvo_ch7xxx.c b/drivers/gpu/drm/i915/dvo_ch7xxx.c new file mode 100644 index 00000000..7eaa94e4 --- /dev/null +++ b/drivers/gpu/drm/i915/dvo_ch7xxx.c @@ -0,0 +1,331 @@ +/************************************************************************** + +Copyright © 2006 Dave Airlie + +All Rights Reserved. + +Permission is hereby granted, free of charge, to any person obtaining a +copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sub license, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial portions +of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS +OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. +IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR +ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, +TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE +SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + +**************************************************************************/ + +#include "dvo.h" + +#define CH7xxx_REG_VID 0x4a +#define CH7xxx_REG_DID 0x4b + +#define CH7011_VID 0x83 /* 7010 as well */ +#define CH7009A_VID 0x84 +#define CH7009B_VID 0x85 +#define CH7301_VID 0x95 + +#define CH7xxx_VID 0x84 +#define CH7xxx_DID 0x17 + +#define CH7xxx_NUM_REGS 0x4c + +#define CH7xxx_CM 0x1c +#define CH7xxx_CM_XCM (1<<0) +#define CH7xxx_CM_MCP (1<<2) +#define CH7xxx_INPUT_CLOCK 0x1d +#define CH7xxx_GPIO 0x1e +#define CH7xxx_GPIO_HPIR (1<<3) +#define CH7xxx_IDF 0x1f + +#define CH7xxx_IDF_HSP (1<<3) +#define CH7xxx_IDF_VSP (1<<4) + +#define CH7xxx_CONNECTION_DETECT 0x20 +#define CH7xxx_CDET_DVI (1<<5) + +#define CH7301_DAC_CNTL 0x21 +#define CH7301_HOTPLUG 0x23 +#define CH7xxx_TCTL 0x31 +#define CH7xxx_TVCO 0x32 +#define CH7xxx_TPCP 0x33 +#define CH7xxx_TPD 0x34 +#define CH7xxx_TPVT 0x35 +#define CH7xxx_TLPF 0x36 +#define CH7xxx_TCT 0x37 +#define CH7301_TEST_PATTERN 0x48 + +#define CH7xxx_PM 0x49 +#define CH7xxx_PM_FPD (1<<0) +#define CH7301_PM_DACPD0 (1<<1) +#define CH7301_PM_DACPD1 (1<<2) +#define CH7301_PM_DACPD2 (1<<3) +#define CH7xxx_PM_DVIL (1<<6) +#define CH7xxx_PM_DVIP (1<<7) + +#define CH7301_SYNC_POLARITY 0x56 +#define CH7301_SYNC_RGB_YUV (1<<0) +#define CH7301_SYNC_POL_DVI (1<<5) + +/** @file + * driver for the Chrontel 7xxx DVI chip over DVO. + */ + +static struct ch7xxx_id_struct { + uint8_t vid; + char *name; +} ch7xxx_ids[] = { + { CH7011_VID, "CH7011" }, + { CH7009A_VID, "CH7009A" }, + { CH7009B_VID, "CH7009B" }, + { CH7301_VID, "CH7301" }, +}; + +struct ch7xxx_priv { + bool quiet; +}; + +static char *ch7xxx_get_id(uint8_t vid) +{ + int i; + + for (i = 0; i < ARRAY_SIZE(ch7xxx_ids); i++) { + if (ch7xxx_ids[i].vid == vid) + return ch7xxx_ids[i].name; + } + + return NULL; +} + +/** Reads an 8 bit register */ +static bool ch7xxx_readb(struct intel_dvo_device *dvo, int addr, uint8_t *ch) +{ + struct ch7xxx_priv *ch7xxx= dvo->dev_priv; + struct i2c_adapter *adapter = dvo->i2c_bus; + u8 out_buf[2]; + u8 in_buf[2]; + + struct i2c_msg msgs[] = { + { + .addr = dvo->slave_addr, + .flags = 0, + .len = 1, + .buf = out_buf, + }, + { + .addr = dvo->slave_addr, + .flags = I2C_M_RD, + .len = 1, + .buf = in_buf, + } + }; + + out_buf[0] = addr; + out_buf[1] = 0; + + if (i2c_transfer(adapter, msgs, 2) == 2) { + *ch = in_buf[0]; + return true; + }; + + if (!ch7xxx->quiet) { + DRM_DEBUG_KMS("Unable to read register 0x%02x from %s:%02x.\n", + addr, adapter->name, dvo->slave_addr); + } + return false; +} + +/** Writes an 8 bit register */ +static bool ch7xxx_writeb(struct intel_dvo_device *dvo, int addr, uint8_t ch) +{ + struct ch7xxx_priv *ch7xxx = dvo->dev_priv; + struct i2c_adapter *adapter = dvo->i2c_bus; + uint8_t out_buf[2]; + struct i2c_msg msg = { + .addr = dvo->slave_addr, + .flags = 0, + .len = 2, + .buf = out_buf, + }; + + out_buf[0] = addr; + out_buf[1] = ch; + + if (i2c_transfer(adapter, &msg, 1) == 1) + return true; + + if (!ch7xxx->quiet) { + DRM_DEBUG_KMS("Unable to write register 0x%02x to %s:%d.\n", + addr, adapter->name, dvo->slave_addr); + } + + return false; +} + +static bool ch7xxx_init(struct intel_dvo_device *dvo, + struct i2c_adapter *adapter) +{ + /* this will detect the CH7xxx chip on the specified i2c bus */ + struct ch7xxx_priv *ch7xxx; + uint8_t vendor, device; + char *name; + + ch7xxx = kzalloc(sizeof(struct ch7xxx_priv), GFP_KERNEL); + if (ch7xxx == NULL) + return false; + + dvo->i2c_bus = adapter; + dvo->dev_priv = ch7xxx; + ch7xxx->quiet = true; + + if (!ch7xxx_readb(dvo, CH7xxx_REG_VID, &vendor)) + goto out; + + name = ch7xxx_get_id(vendor); + if (!name) { + DRM_DEBUG_KMS("ch7xxx not detected; got 0x%02x from %s " + "slave %d.\n", + vendor, adapter->name, dvo->slave_addr); + goto out; + } + + + if (!ch7xxx_readb(dvo, CH7xxx_REG_DID, &device)) + goto out; + + if (device != CH7xxx_DID) { + DRM_DEBUG_KMS("ch7xxx not detected; got 0x%02x from %s " + "slave %d.\n", + vendor, adapter->name, dvo->slave_addr); + goto out; + } + + ch7xxx->quiet = false; + DRM_DEBUG_KMS("Detected %s chipset, vendor/device ID 0x%02x/0x%02x\n", + name, vendor, device); + return true; +out: + kfree(ch7xxx); + return false; +} + +static enum drm_connector_status ch7xxx_detect(struct intel_dvo_device *dvo) +{ + uint8_t cdet, orig_pm, pm; + + ch7xxx_readb(dvo, CH7xxx_PM, &orig_pm); + + pm = orig_pm; + pm &= ~CH7xxx_PM_FPD; + pm |= CH7xxx_PM_DVIL | CH7xxx_PM_DVIP; + + ch7xxx_writeb(dvo, CH7xxx_PM, pm); + + ch7xxx_readb(dvo, CH7xxx_CONNECTION_DETECT, &cdet); + + ch7xxx_writeb(dvo, CH7xxx_PM, orig_pm); + + if (cdet & CH7xxx_CDET_DVI) + return connector_status_connected; + return connector_status_disconnected; +} + +static enum drm_mode_status ch7xxx_mode_valid(struct intel_dvo_device *dvo, + struct drm_display_mode *mode) +{ + if (mode->clock > 165000) + return MODE_CLOCK_HIGH; + + return MODE_OK; +} + +static void ch7xxx_mode_set(struct intel_dvo_device *dvo, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + uint8_t tvco, tpcp, tpd, tlpf, idf; + + if (mode->clock <= 65000) { + tvco = 0x23; + tpcp = 0x08; + tpd = 0x16; + tlpf = 0x60; + } else { + tvco = 0x2d; + tpcp = 0x06; + tpd = 0x26; + tlpf = 0xa0; + } + + ch7xxx_writeb(dvo, CH7xxx_TCTL, 0x00); + ch7xxx_writeb(dvo, CH7xxx_TVCO, tvco); + ch7xxx_writeb(dvo, CH7xxx_TPCP, tpcp); + ch7xxx_writeb(dvo, CH7xxx_TPD, tpd); + ch7xxx_writeb(dvo, CH7xxx_TPVT, 0x30); + ch7xxx_writeb(dvo, CH7xxx_TLPF, tlpf); + ch7xxx_writeb(dvo, CH7xxx_TCT, 0x00); + + ch7xxx_readb(dvo, CH7xxx_IDF, &idf); + + idf &= ~(CH7xxx_IDF_HSP | CH7xxx_IDF_VSP); + if (mode->flags & DRM_MODE_FLAG_PHSYNC) + idf |= CH7xxx_IDF_HSP; + + if (mode->flags & DRM_MODE_FLAG_PVSYNC) + idf |= CH7xxx_IDF_HSP; + + ch7xxx_writeb(dvo, CH7xxx_IDF, idf); +} + +/* set the CH7xxx power state */ +static void ch7xxx_dpms(struct intel_dvo_device *dvo, int mode) +{ + if (mode == DRM_MODE_DPMS_ON) + ch7xxx_writeb(dvo, CH7xxx_PM, CH7xxx_PM_DVIL | CH7xxx_PM_DVIP); + else + ch7xxx_writeb(dvo, CH7xxx_PM, CH7xxx_PM_FPD); +} + +static void ch7xxx_dump_regs(struct intel_dvo_device *dvo) +{ + int i; + + for (i = 0; i < CH7xxx_NUM_REGS; i++) { + uint8_t val; + if ((i % 8) == 0 ) + DRM_LOG_KMS("\n %02X: ", i); + ch7xxx_readb(dvo, i, &val); + DRM_LOG_KMS("%02X ", val); + } +} + +static void ch7xxx_destroy(struct intel_dvo_device *dvo) +{ + struct ch7xxx_priv *ch7xxx = dvo->dev_priv; + + if (ch7xxx) { + kfree(ch7xxx); + dvo->dev_priv = NULL; + } +} + +struct intel_dvo_dev_ops ch7xxx_ops = { + .init = ch7xxx_init, + .detect = ch7xxx_detect, + .mode_valid = ch7xxx_mode_valid, + .mode_set = ch7xxx_mode_set, + .dpms = ch7xxx_dpms, + .dump_regs = ch7xxx_dump_regs, + .destroy = ch7xxx_destroy, +}; diff --git a/drivers/gpu/drm/i915/dvo_ivch.c b/drivers/gpu/drm/i915/dvo_ivch.c new file mode 100644 index 00000000..a12ed941 --- /dev/null +++ b/drivers/gpu/drm/i915/dvo_ivch.c @@ -0,0 +1,421 @@ +/* + * Copyright © 2006 Intel Corporation + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Eric Anholt + * + */ + +#include "dvo.h" + +/* + * register definitions for the i82807aa. + * + * Documentation on this chipset can be found in datasheet #29069001 at + * intel.com. + */ + +/* + * VCH Revision & GMBus Base Addr + */ +#define VR00 0x00 +# define VR00_BASE_ADDRESS_MASK 0x007f + +/* + * Functionality Enable + */ +#define VR01 0x01 + +/* + * Enable the panel fitter + */ +# define VR01_PANEL_FIT_ENABLE (1 << 3) +/* + * Enables the LCD display. + * + * This must not be set while VR01_DVO_BYPASS_ENABLE is set. + */ +# define VR01_LCD_ENABLE (1 << 2) +/** Enables the DVO repeater. */ +# define VR01_DVO_BYPASS_ENABLE (1 << 1) +/** Enables the DVO clock */ +# define VR01_DVO_ENABLE (1 << 0) + +/* + * LCD Interface Format + */ +#define VR10 0x10 +/** Enables LVDS output instead of CMOS */ +# define VR10_LVDS_ENABLE (1 << 4) +/** Enables 18-bit LVDS output. */ +# define VR10_INTERFACE_1X18 (0 << 2) +/** Enables 24-bit LVDS or CMOS output */ +# define VR10_INTERFACE_1X24 (1 << 2) +/** Enables 2x18-bit LVDS or CMOS output. */ +# define VR10_INTERFACE_2X18 (2 << 2) +/** Enables 2x24-bit LVDS output */ +# define VR10_INTERFACE_2X24 (3 << 2) + +/* + * VR20 LCD Horizontal Display Size + */ +#define VR20 0x20 + +/* + * LCD Vertical Display Size + */ +#define VR21 0x20 + +/* + * Panel power down status + */ +#define VR30 0x30 +/** Read only bit indicating that the panel is not in a safe poweroff state. */ +# define VR30_PANEL_ON (1 << 15) + +#define VR40 0x40 +# define VR40_STALL_ENABLE (1 << 13) +# define VR40_VERTICAL_INTERP_ENABLE (1 << 12) +# define VR40_ENHANCED_PANEL_FITTING (1 << 11) +# define VR40_HORIZONTAL_INTERP_ENABLE (1 << 10) +# define VR40_AUTO_RATIO_ENABLE (1 << 9) +# define VR40_CLOCK_GATING_ENABLE (1 << 8) + +/* + * Panel Fitting Vertical Ratio + * (((image_height - 1) << 16) / ((panel_height - 1))) >> 2 + */ +#define VR41 0x41 + +/* + * Panel Fitting Horizontal Ratio + * (((image_width - 1) << 16) / ((panel_width - 1))) >> 2 + */ +#define VR42 0x42 + +/* + * Horizontal Image Size + */ +#define VR43 0x43 + +/* VR80 GPIO 0 + */ +#define VR80 0x80 +#define VR81 0x81 +#define VR82 0x82 +#define VR83 0x83 +#define VR84 0x84 +#define VR85 0x85 +#define VR86 0x86 +#define VR87 0x87 + +/* VR88 GPIO 8 + */ +#define VR88 0x88 + +/* Graphics BIOS scratch 0 + */ +#define VR8E 0x8E +# define VR8E_PANEL_TYPE_MASK (0xf << 0) +# define VR8E_PANEL_INTERFACE_CMOS (0 << 4) +# define VR8E_PANEL_INTERFACE_LVDS (1 << 4) +# define VR8E_FORCE_DEFAULT_PANEL (1 << 5) + +/* Graphics BIOS scratch 1 + */ +#define VR8F 0x8F +# define VR8F_VCH_PRESENT (1 << 0) +# define VR8F_DISPLAY_CONN (1 << 1) +# define VR8F_POWER_MASK (0x3c) +# define VR8F_POWER_POS (2) + + +struct ivch_priv { + bool quiet; + + uint16_t width, height; +}; + + +static void ivch_dump_regs(struct intel_dvo_device *dvo); + +/** + * Reads a register on the ivch. + * + * Each of the 256 registers are 16 bits long. + */ +static bool ivch_read(struct intel_dvo_device *dvo, int addr, uint16_t *data) +{ + struct ivch_priv *priv = dvo->dev_priv; + struct i2c_adapter *adapter = dvo->i2c_bus; + u8 out_buf[1]; + u8 in_buf[2]; + + struct i2c_msg msgs[] = { + { + .addr = dvo->slave_addr, + .flags = I2C_M_RD, + .len = 0, + }, + { + .addr = 0, + .flags = I2C_M_NOSTART, + .len = 1, + .buf = out_buf, + }, + { + .addr = dvo->slave_addr, + .flags = I2C_M_RD | I2C_M_NOSTART, + .len = 2, + .buf = in_buf, + } + }; + + out_buf[0] = addr; + + if (i2c_transfer(adapter, msgs, 3) == 3) { + *data = (in_buf[1] << 8) | in_buf[0]; + return true; + }; + + if (!priv->quiet) { + DRM_DEBUG_KMS("Unable to read register 0x%02x from " + "%s:%02x.\n", + addr, adapter->name, dvo->slave_addr); + } + return false; +} + +/** Writes a 16-bit register on the ivch */ +static bool ivch_write(struct intel_dvo_device *dvo, int addr, uint16_t data) +{ + struct ivch_priv *priv = dvo->dev_priv; + struct i2c_adapter *adapter = dvo->i2c_bus; + u8 out_buf[3]; + struct i2c_msg msg = { + .addr = dvo->slave_addr, + .flags = 0, + .len = 3, + .buf = out_buf, + }; + + out_buf[0] = addr; + out_buf[1] = data & 0xff; + out_buf[2] = data >> 8; + + if (i2c_transfer(adapter, &msg, 1) == 1) + return true; + + if (!priv->quiet) { + DRM_DEBUG_KMS("Unable to write register 0x%02x to %s:%d.\n", + addr, adapter->name, dvo->slave_addr); + } + + return false; +} + +/** Probes the given bus and slave address for an ivch */ +static bool ivch_init(struct intel_dvo_device *dvo, + struct i2c_adapter *adapter) +{ + struct ivch_priv *priv; + uint16_t temp; + + priv = kzalloc(sizeof(struct ivch_priv), GFP_KERNEL); + if (priv == NULL) + return false; + + dvo->i2c_bus = adapter; + dvo->dev_priv = priv; + priv->quiet = true; + + if (!ivch_read(dvo, VR00, &temp)) + goto out; + priv->quiet = false; + + /* Since the identification bits are probably zeroes, which doesn't seem + * very unique, check that the value in the base address field matches + * the address it's responding on. + */ + if ((temp & VR00_BASE_ADDRESS_MASK) != dvo->slave_addr) { + DRM_DEBUG_KMS("ivch detect failed due to address mismatch " + "(%d vs %d)\n", + (temp & VR00_BASE_ADDRESS_MASK), dvo->slave_addr); + goto out; + } + + ivch_read(dvo, VR20, &priv->width); + ivch_read(dvo, VR21, &priv->height); + + return true; + +out: + kfree(priv); + return false; +} + +static enum drm_connector_status ivch_detect(struct intel_dvo_device *dvo) +{ + return connector_status_connected; +} + +static enum drm_mode_status ivch_mode_valid(struct intel_dvo_device *dvo, + struct drm_display_mode *mode) +{ + if (mode->clock > 112000) + return MODE_CLOCK_HIGH; + + return MODE_OK; +} + +/** Sets the power state of the panel connected to the ivch */ +static void ivch_dpms(struct intel_dvo_device *dvo, int mode) +{ + int i; + uint16_t vr01, vr30, backlight; + + /* Set the new power state of the panel. */ + if (!ivch_read(dvo, VR01, &vr01)) + return; + + if (mode == DRM_MODE_DPMS_ON) + backlight = 1; + else + backlight = 0; + ivch_write(dvo, VR80, backlight); + + if (mode == DRM_MODE_DPMS_ON) + vr01 |= VR01_LCD_ENABLE | VR01_DVO_ENABLE; + else + vr01 &= ~(VR01_LCD_ENABLE | VR01_DVO_ENABLE); + + ivch_write(dvo, VR01, vr01); + + /* Wait for the panel to make its state transition */ + for (i = 0; i < 100; i++) { + if (!ivch_read(dvo, VR30, &vr30)) + break; + + if (((vr30 & VR30_PANEL_ON) != 0) == (mode == DRM_MODE_DPMS_ON)) + break; + udelay(1000); + } + /* wait some more; vch may fail to resync sometimes without this */ + udelay(16 * 1000); +} + +static void ivch_mode_set(struct intel_dvo_device *dvo, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + uint16_t vr40 = 0; + uint16_t vr01; + + vr01 = 0; + vr40 = (VR40_STALL_ENABLE | VR40_VERTICAL_INTERP_ENABLE | + VR40_HORIZONTAL_INTERP_ENABLE); + + if (mode->hdisplay != adjusted_mode->hdisplay || + mode->vdisplay != adjusted_mode->vdisplay) { + uint16_t x_ratio, y_ratio; + + vr01 |= VR01_PANEL_FIT_ENABLE; + vr40 |= VR40_CLOCK_GATING_ENABLE; + x_ratio = (((mode->hdisplay - 1) << 16) / + (adjusted_mode->hdisplay - 1)) >> 2; + y_ratio = (((mode->vdisplay - 1) << 16) / + (adjusted_mode->vdisplay - 1)) >> 2; + ivch_write (dvo, VR42, x_ratio); + ivch_write (dvo, VR41, y_ratio); + } else { + vr01 &= ~VR01_PANEL_FIT_ENABLE; + vr40 &= ~VR40_CLOCK_GATING_ENABLE; + } + vr40 &= ~VR40_AUTO_RATIO_ENABLE; + + ivch_write(dvo, VR01, vr01); + ivch_write(dvo, VR40, vr40); + + ivch_dump_regs(dvo); +} + +static void ivch_dump_regs(struct intel_dvo_device *dvo) +{ + uint16_t val; + + ivch_read(dvo, VR00, &val); + DRM_LOG_KMS("VR00: 0x%04x\n", val); + ivch_read(dvo, VR01, &val); + DRM_LOG_KMS("VR01: 0x%04x\n", val); + ivch_read(dvo, VR30, &val); + DRM_LOG_KMS("VR30: 0x%04x\n", val); + ivch_read(dvo, VR40, &val); + DRM_LOG_KMS("VR40: 0x%04x\n", val); + + /* GPIO registers */ + ivch_read(dvo, VR80, &val); + DRM_LOG_KMS("VR80: 0x%04x\n", val); + ivch_read(dvo, VR81, &val); + DRM_LOG_KMS("VR81: 0x%04x\n", val); + ivch_read(dvo, VR82, &val); + DRM_LOG_KMS("VR82: 0x%04x\n", val); + ivch_read(dvo, VR83, &val); + DRM_LOG_KMS("VR83: 0x%04x\n", val); + ivch_read(dvo, VR84, &val); + DRM_LOG_KMS("VR84: 0x%04x\n", val); + ivch_read(dvo, VR85, &val); + DRM_LOG_KMS("VR85: 0x%04x\n", val); + ivch_read(dvo, VR86, &val); + DRM_LOG_KMS("VR86: 0x%04x\n", val); + ivch_read(dvo, VR87, &val); + DRM_LOG_KMS("VR87: 0x%04x\n", val); + ivch_read(dvo, VR88, &val); + DRM_LOG_KMS("VR88: 0x%04x\n", val); + + /* Scratch register 0 - AIM Panel type */ + ivch_read(dvo, VR8E, &val); + DRM_LOG_KMS("VR8E: 0x%04x\n", val); + + /* Scratch register 1 - Status register */ + ivch_read(dvo, VR8F, &val); + DRM_LOG_KMS("VR8F: 0x%04x\n", val); +} + +static void ivch_destroy(struct intel_dvo_device *dvo) +{ + struct ivch_priv *priv = dvo->dev_priv; + + if (priv) { + kfree(priv); + dvo->dev_priv = NULL; + } +} + +struct intel_dvo_dev_ops ivch_ops= { + .init = ivch_init, + .dpms = ivch_dpms, + .mode_valid = ivch_mode_valid, + .mode_set = ivch_mode_set, + .detect = ivch_detect, + .dump_regs = ivch_dump_regs, + .destroy = ivch_destroy, +}; diff --git a/drivers/gpu/drm/i915/dvo_sil164.c b/drivers/gpu/drm/i915/dvo_sil164.c new file mode 100644 index 00000000..e4b4091d --- /dev/null +++ b/drivers/gpu/drm/i915/dvo_sil164.c @@ -0,0 +1,263 @@ +/************************************************************************** + +Copyright © 2006 Dave Airlie + +All Rights Reserved. + +Permission is hereby granted, free of charge, to any person obtaining a +copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sub license, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial portions +of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS +OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. +IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR +ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, +TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE +SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + +**************************************************************************/ + +#include "dvo.h" + +#define SIL164_VID 0x0001 +#define SIL164_DID 0x0006 + +#define SIL164_VID_LO 0x00 +#define SIL164_VID_HI 0x01 +#define SIL164_DID_LO 0x02 +#define SIL164_DID_HI 0x03 +#define SIL164_REV 0x04 +#define SIL164_RSVD 0x05 +#define SIL164_FREQ_LO 0x06 +#define SIL164_FREQ_HI 0x07 + +#define SIL164_REG8 0x08 +#define SIL164_8_VEN (1<<5) +#define SIL164_8_HEN (1<<4) +#define SIL164_8_DSEL (1<<3) +#define SIL164_8_BSEL (1<<2) +#define SIL164_8_EDGE (1<<1) +#define SIL164_8_PD (1<<0) + +#define SIL164_REG9 0x09 +#define SIL164_9_VLOW (1<<7) +#define SIL164_9_MSEL_MASK (0x7<<4) +#define SIL164_9_TSEL (1<<3) +#define SIL164_9_RSEN (1<<2) +#define SIL164_9_HTPLG (1<<1) +#define SIL164_9_MDI (1<<0) + +#define SIL164_REGC 0x0c + +struct sil164_priv { + //I2CDevRec d; + bool quiet; +}; + +#define SILPTR(d) ((SIL164Ptr)(d->DriverPrivate.ptr)) + +static bool sil164_readb(struct intel_dvo_device *dvo, int addr, uint8_t *ch) +{ + struct sil164_priv *sil = dvo->dev_priv; + struct i2c_adapter *adapter = dvo->i2c_bus; + u8 out_buf[2]; + u8 in_buf[2]; + + struct i2c_msg msgs[] = { + { + .addr = dvo->slave_addr, + .flags = 0, + .len = 1, + .buf = out_buf, + }, + { + .addr = dvo->slave_addr, + .flags = I2C_M_RD, + .len = 1, + .buf = in_buf, + } + }; + + out_buf[0] = addr; + out_buf[1] = 0; + + if (i2c_transfer(adapter, msgs, 2) == 2) { + *ch = in_buf[0]; + return true; + }; + + if (!sil->quiet) { + DRM_DEBUG_KMS("Unable to read register 0x%02x from %s:%02x.\n", + addr, adapter->name, dvo->slave_addr); + } + return false; +} + +static bool sil164_writeb(struct intel_dvo_device *dvo, int addr, uint8_t ch) +{ + struct sil164_priv *sil= dvo->dev_priv; + struct i2c_adapter *adapter = dvo->i2c_bus; + uint8_t out_buf[2]; + struct i2c_msg msg = { + .addr = dvo->slave_addr, + .flags = 0, + .len = 2, + .buf = out_buf, + }; + + out_buf[0] = addr; + out_buf[1] = ch; + + if (i2c_transfer(adapter, &msg, 1) == 1) + return true; + + if (!sil->quiet) { + DRM_DEBUG_KMS("Unable to write register 0x%02x to %s:%d.\n", + addr, adapter->name, dvo->slave_addr); + } + + return false; +} + +/* Silicon Image 164 driver for chip on i2c bus */ +static bool sil164_init(struct intel_dvo_device *dvo, + struct i2c_adapter *adapter) +{ + /* this will detect the SIL164 chip on the specified i2c bus */ + struct sil164_priv *sil; + unsigned char ch; + + sil = kzalloc(sizeof(struct sil164_priv), GFP_KERNEL); + if (sil == NULL) + return false; + + dvo->i2c_bus = adapter; + dvo->dev_priv = sil; + sil->quiet = true; + + if (!sil164_readb(dvo, SIL164_VID_LO, &ch)) + goto out; + + if (ch != (SIL164_VID & 0xff)) { + DRM_DEBUG_KMS("sil164 not detected got %d: from %s Slave %d.\n", + ch, adapter->name, dvo->slave_addr); + goto out; + } + + if (!sil164_readb(dvo, SIL164_DID_LO, &ch)) + goto out; + + if (ch != (SIL164_DID & 0xff)) { + DRM_DEBUG_KMS("sil164 not detected got %d: from %s Slave %d.\n", + ch, adapter->name, dvo->slave_addr); + goto out; + } + sil->quiet = false; + + DRM_DEBUG_KMS("init sil164 dvo controller successfully!\n"); + return true; + +out: + kfree(sil); + return false; +} + +static enum drm_connector_status sil164_detect(struct intel_dvo_device *dvo) +{ + uint8_t reg9; + + sil164_readb(dvo, SIL164_REG9, ®9); + + if (reg9 & SIL164_9_HTPLG) + return connector_status_connected; + else + return connector_status_disconnected; +} + +static enum drm_mode_status sil164_mode_valid(struct intel_dvo_device *dvo, + struct drm_display_mode *mode) +{ + return MODE_OK; +} + +static void sil164_mode_set(struct intel_dvo_device *dvo, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + /* As long as the basics are set up, since we don't have clock + * dependencies in the mode setup, we can just leave the + * registers alone and everything will work fine. + */ + /* recommended programming sequence from doc */ + /*sil164_writeb(sil, 0x08, 0x30); + sil164_writeb(sil, 0x09, 0x00); + sil164_writeb(sil, 0x0a, 0x90); + sil164_writeb(sil, 0x0c, 0x89); + sil164_writeb(sil, 0x08, 0x31);*/ + /* don't do much */ + return; +} + +/* set the SIL164 power state */ +static void sil164_dpms(struct intel_dvo_device *dvo, int mode) +{ + int ret; + unsigned char ch; + + ret = sil164_readb(dvo, SIL164_REG8, &ch); + if (ret == false) + return; + + if (mode == DRM_MODE_DPMS_ON) + ch |= SIL164_8_PD; + else + ch &= ~SIL164_8_PD; + + sil164_writeb(dvo, SIL164_REG8, ch); + return; +} + +static void sil164_dump_regs(struct intel_dvo_device *dvo) +{ + uint8_t val; + + sil164_readb(dvo, SIL164_FREQ_LO, &val); + DRM_LOG_KMS("SIL164_FREQ_LO: 0x%02x\n", val); + sil164_readb(dvo, SIL164_FREQ_HI, &val); + DRM_LOG_KMS("SIL164_FREQ_HI: 0x%02x\n", val); + sil164_readb(dvo, SIL164_REG8, &val); + DRM_LOG_KMS("SIL164_REG8: 0x%02x\n", val); + sil164_readb(dvo, SIL164_REG9, &val); + DRM_LOG_KMS("SIL164_REG9: 0x%02x\n", val); + sil164_readb(dvo, SIL164_REGC, &val); + DRM_LOG_KMS("SIL164_REGC: 0x%02x\n", val); +} + +static void sil164_destroy(struct intel_dvo_device *dvo) +{ + struct sil164_priv *sil = dvo->dev_priv; + + if (sil) { + kfree(sil); + dvo->dev_priv = NULL; + } +} + +struct intel_dvo_dev_ops sil164_ops = { + .init = sil164_init, + .detect = sil164_detect, + .mode_valid = sil164_mode_valid, + .mode_set = sil164_mode_set, + .dpms = sil164_dpms, + .dump_regs = sil164_dump_regs, + .destroy = sil164_destroy, +}; diff --git a/drivers/gpu/drm/i915/dvo_tfp410.c b/drivers/gpu/drm/i915/dvo_tfp410.c new file mode 100644 index 00000000..8ab2855b --- /dev/null +++ b/drivers/gpu/drm/i915/dvo_tfp410.c @@ -0,0 +1,304 @@ +/* + * Copyright © 2007 Dave Mueller + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS + * IN THE SOFTWARE. + * + * Authors: + * Dave Mueller + * + */ + +#include "dvo.h" + +/* register definitions according to the TFP410 data sheet */ +#define TFP410_VID 0x014C +#define TFP410_DID 0x0410 + +#define TFP410_VID_LO 0x00 +#define TFP410_VID_HI 0x01 +#define TFP410_DID_LO 0x02 +#define TFP410_DID_HI 0x03 +#define TFP410_REV 0x04 + +#define TFP410_CTL_1 0x08 +#define TFP410_CTL_1_TDIS (1<<6) +#define TFP410_CTL_1_VEN (1<<5) +#define TFP410_CTL_1_HEN (1<<4) +#define TFP410_CTL_1_DSEL (1<<3) +#define TFP410_CTL_1_BSEL (1<<2) +#define TFP410_CTL_1_EDGE (1<<1) +#define TFP410_CTL_1_PD (1<<0) + +#define TFP410_CTL_2 0x09 +#define TFP410_CTL_2_VLOW (1<<7) +#define TFP410_CTL_2_MSEL_MASK (0x7<<4) +#define TFP410_CTL_2_MSEL (1<<4) +#define TFP410_CTL_2_TSEL (1<<3) +#define TFP410_CTL_2_RSEN (1<<2) +#define TFP410_CTL_2_HTPLG (1<<1) +#define TFP410_CTL_2_MDI (1<<0) + +#define TFP410_CTL_3 0x0A +#define TFP410_CTL_3_DK_MASK (0x7<<5) +#define TFP410_CTL_3_DK (1<<5) +#define TFP410_CTL_3_DKEN (1<<4) +#define TFP410_CTL_3_CTL_MASK (0x7<<1) +#define TFP410_CTL_3_CTL (1<<1) + +#define TFP410_USERCFG 0x0B + +#define TFP410_DE_DLY 0x32 + +#define TFP410_DE_CTL 0x33 +#define TFP410_DE_CTL_DEGEN (1<<6) +#define TFP410_DE_CTL_VSPOL (1<<5) +#define TFP410_DE_CTL_HSPOL (1<<4) +#define TFP410_DE_CTL_DEDLY8 (1<<0) + +#define TFP410_DE_TOP 0x34 + +#define TFP410_DE_CNT_LO 0x36 +#define TFP410_DE_CNT_HI 0x37 + +#define TFP410_DE_LIN_LO 0x38 +#define TFP410_DE_LIN_HI 0x39 + +#define TFP410_H_RES_LO 0x3A +#define TFP410_H_RES_HI 0x3B + +#define TFP410_V_RES_LO 0x3C +#define TFP410_V_RES_HI 0x3D + +struct tfp410_priv { + bool quiet; +}; + +static bool tfp410_readb(struct intel_dvo_device *dvo, int addr, uint8_t *ch) +{ + struct tfp410_priv *tfp = dvo->dev_priv; + struct i2c_adapter *adapter = dvo->i2c_bus; + u8 out_buf[2]; + u8 in_buf[2]; + + struct i2c_msg msgs[] = { + { + .addr = dvo->slave_addr, + .flags = 0, + .len = 1, + .buf = out_buf, + }, + { + .addr = dvo->slave_addr, + .flags = I2C_M_RD, + .len = 1, + .buf = in_buf, + } + }; + + out_buf[0] = addr; + out_buf[1] = 0; + + if (i2c_transfer(adapter, msgs, 2) == 2) { + *ch = in_buf[0]; + return true; + }; + + if (!tfp->quiet) { + DRM_DEBUG_KMS("Unable to read register 0x%02x from %s:%02x.\n", + addr, adapter->name, dvo->slave_addr); + } + return false; +} + +static bool tfp410_writeb(struct intel_dvo_device *dvo, int addr, uint8_t ch) +{ + struct tfp410_priv *tfp = dvo->dev_priv; + struct i2c_adapter *adapter = dvo->i2c_bus; + uint8_t out_buf[2]; + struct i2c_msg msg = { + .addr = dvo->slave_addr, + .flags = 0, + .len = 2, + .buf = out_buf, + }; + + out_buf[0] = addr; + out_buf[1] = ch; + + if (i2c_transfer(adapter, &msg, 1) == 1) + return true; + + if (!tfp->quiet) { + DRM_DEBUG_KMS("Unable to write register 0x%02x to %s:%d.\n", + addr, adapter->name, dvo->slave_addr); + } + + return false; +} + +static int tfp410_getid(struct intel_dvo_device *dvo, int addr) +{ + uint8_t ch1, ch2; + + if (tfp410_readb(dvo, addr+0, &ch1) && + tfp410_readb(dvo, addr+1, &ch2)) + return ((ch2 << 8) & 0xFF00) | (ch1 & 0x00FF); + + return -1; +} + +/* Ti TFP410 driver for chip on i2c bus */ +static bool tfp410_init(struct intel_dvo_device *dvo, + struct i2c_adapter *adapter) +{ + /* this will detect the tfp410 chip on the specified i2c bus */ + struct tfp410_priv *tfp; + int id; + + tfp = kzalloc(sizeof(struct tfp410_priv), GFP_KERNEL); + if (tfp == NULL) + return false; + + dvo->i2c_bus = adapter; + dvo->dev_priv = tfp; + tfp->quiet = true; + + if ((id = tfp410_getid(dvo, TFP410_VID_LO)) != TFP410_VID) { + DRM_DEBUG_KMS("tfp410 not detected got VID %X: from %s " + "Slave %d.\n", + id, adapter->name, dvo->slave_addr); + goto out; + } + + if ((id = tfp410_getid(dvo, TFP410_DID_LO)) != TFP410_DID) { + DRM_DEBUG_KMS("tfp410 not detected got DID %X: from %s " + "Slave %d.\n", + id, adapter->name, dvo->slave_addr); + goto out; + } + tfp->quiet = false; + return true; +out: + kfree(tfp); + return false; +} + +static enum drm_connector_status tfp410_detect(struct intel_dvo_device *dvo) +{ + enum drm_connector_status ret = connector_status_disconnected; + uint8_t ctl2; + + if (tfp410_readb(dvo, TFP410_CTL_2, &ctl2)) { + if (ctl2 & TFP410_CTL_2_RSEN) + ret = connector_status_connected; + else + ret = connector_status_disconnected; + } + + return ret; +} + +static enum drm_mode_status tfp410_mode_valid(struct intel_dvo_device *dvo, + struct drm_display_mode *mode) +{ + return MODE_OK; +} + +static void tfp410_mode_set(struct intel_dvo_device *dvo, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + /* As long as the basics are set up, since we don't have clock dependencies + * in the mode setup, we can just leave the registers alone and everything + * will work fine. + */ + /* don't do much */ + return; +} + +/* set the tfp410 power state */ +static void tfp410_dpms(struct intel_dvo_device *dvo, int mode) +{ + uint8_t ctl1; + + if (!tfp410_readb(dvo, TFP410_CTL_1, &ctl1)) + return; + + if (mode == DRM_MODE_DPMS_ON) + ctl1 |= TFP410_CTL_1_PD; + else + ctl1 &= ~TFP410_CTL_1_PD; + + tfp410_writeb(dvo, TFP410_CTL_1, ctl1); +} + +static void tfp410_dump_regs(struct intel_dvo_device *dvo) +{ + uint8_t val, val2; + + tfp410_readb(dvo, TFP410_REV, &val); + DRM_LOG_KMS("TFP410_REV: 0x%02X\n", val); + tfp410_readb(dvo, TFP410_CTL_1, &val); + DRM_LOG_KMS("TFP410_CTL1: 0x%02X\n", val); + tfp410_readb(dvo, TFP410_CTL_2, &val); + DRM_LOG_KMS("TFP410_CTL2: 0x%02X\n", val); + tfp410_readb(dvo, TFP410_CTL_3, &val); + DRM_LOG_KMS("TFP410_CTL3: 0x%02X\n", val); + tfp410_readb(dvo, TFP410_USERCFG, &val); + DRM_LOG_KMS("TFP410_USERCFG: 0x%02X\n", val); + tfp410_readb(dvo, TFP410_DE_DLY, &val); + DRM_LOG_KMS("TFP410_DE_DLY: 0x%02X\n", val); + tfp410_readb(dvo, TFP410_DE_CTL, &val); + DRM_LOG_KMS("TFP410_DE_CTL: 0x%02X\n", val); + tfp410_readb(dvo, TFP410_DE_TOP, &val); + DRM_LOG_KMS("TFP410_DE_TOP: 0x%02X\n", val); + tfp410_readb(dvo, TFP410_DE_CNT_LO, &val); + tfp410_readb(dvo, TFP410_DE_CNT_HI, &val2); + DRM_LOG_KMS("TFP410_DE_CNT: 0x%02X%02X\n", val2, val); + tfp410_readb(dvo, TFP410_DE_LIN_LO, &val); + tfp410_readb(dvo, TFP410_DE_LIN_HI, &val2); + DRM_LOG_KMS("TFP410_DE_LIN: 0x%02X%02X\n", val2, val); + tfp410_readb(dvo, TFP410_H_RES_LO, &val); + tfp410_readb(dvo, TFP410_H_RES_HI, &val2); + DRM_LOG_KMS("TFP410_H_RES: 0x%02X%02X\n", val2, val); + tfp410_readb(dvo, TFP410_V_RES_LO, &val); + tfp410_readb(dvo, TFP410_V_RES_HI, &val2); + DRM_LOG_KMS("TFP410_V_RES: 0x%02X%02X\n", val2, val); +} + +static void tfp410_destroy(struct intel_dvo_device *dvo) +{ + struct tfp410_priv *tfp = dvo->dev_priv; + + if (tfp) { + kfree(tfp); + dvo->dev_priv = NULL; + } +} + +struct intel_dvo_dev_ops tfp410_ops = { + .init = tfp410_init, + .detect = tfp410_detect, + .mode_valid = tfp410_mode_valid, + .mode_set = tfp410_mode_set, + .dpms = tfp410_dpms, + .dump_regs = tfp410_dump_regs, + .destroy = tfp410_destroy, +}; diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c new file mode 100644 index 00000000..e36efdc6 --- /dev/null +++ b/drivers/gpu/drm/i915/i915_debugfs.c @@ -0,0 +1,1470 @@ +/* + * Copyright © 2008 Intel Corporation + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS + * IN THE SOFTWARE. + * + * Authors: + * Eric Anholt + * Keith Packard + * + */ + +#include +#include +#include +#include "drmP.h" +#include "drm.h" +#include "intel_drv.h" +#include "intel_ringbuffer.h" +#include "i915_drm.h" +#include "i915_drv.h" + +#define DRM_I915_RING_DEBUG 1 + + +#if defined(CONFIG_DEBUG_FS) + +enum { + ACTIVE_LIST, + FLUSHING_LIST, + INACTIVE_LIST, + PINNED_LIST, + DEFERRED_FREE_LIST, +}; + +static const char *yesno(int v) +{ + return v ? "yes" : "no"; +} + +static int i915_capabilities(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + const struct intel_device_info *info = INTEL_INFO(dev); + + seq_printf(m, "gen: %d\n", info->gen); +#define B(x) seq_printf(m, #x ": %s\n", yesno(info->x)) + B(is_mobile); + B(is_i85x); + B(is_i915g); + B(is_i945gm); + B(is_g33); + B(need_gfx_hws); + B(is_g4x); + B(is_pineview); + B(is_broadwater); + B(is_crestline); + B(has_fbc); + B(has_pipe_cxsr); + B(has_hotplug); + B(cursor_needs_physical); + B(has_overlay); + B(overlay_needs_physical); + B(supports_tv); + B(has_bsd_ring); + B(has_blt_ring); +#undef B + + return 0; +} + +static const char *get_pin_flag(struct drm_i915_gem_object *obj) +{ + if (obj->user_pin_count > 0) + return "P"; + else if (obj->pin_count > 0) + return "p"; + else + return " "; +} + +static const char *get_tiling_flag(struct drm_i915_gem_object *obj) +{ + switch (obj->tiling_mode) { + default: + case I915_TILING_NONE: return " "; + case I915_TILING_X: return "X"; + case I915_TILING_Y: return "Y"; + } +} + +static const char *cache_level_str(int type) +{ + switch (type) { + case I915_CACHE_NONE: return " uncached"; + case I915_CACHE_LLC: return " snooped (LLC)"; + case I915_CACHE_LLC_MLC: return " snooped (LLC+MLC)"; + default: return ""; + } +} + +static void +describe_obj(struct seq_file *m, struct drm_i915_gem_object *obj) +{ + seq_printf(m, "%p: %s%s %8zd %04x %04x %d %d%s%s%s", + &obj->base, + get_pin_flag(obj), + get_tiling_flag(obj), + obj->base.size, + obj->base.read_domains, + obj->base.write_domain, + obj->last_rendering_seqno, + obj->last_fenced_seqno, + cache_level_str(obj->cache_level), + obj->dirty ? " dirty" : "", + obj->madv == I915_MADV_DONTNEED ? " purgeable" : ""); + if (obj->base.name) + seq_printf(m, " (name: %d)", obj->base.name); + if (obj->fence_reg != I915_FENCE_REG_NONE) + seq_printf(m, " (fence: %d)", obj->fence_reg); + if (obj->gtt_space != NULL) + seq_printf(m, " (gtt offset: %08x, size: %08x)", + obj->gtt_offset, (unsigned int)obj->gtt_space->size); + if (obj->pin_mappable || obj->fault_mappable) { + char s[3], *t = s; + if (obj->pin_mappable) + *t++ = 'p'; + if (obj->fault_mappable) + *t++ = 'f'; + *t = '\0'; + seq_printf(m, " (%s mappable)", s); + } + if (obj->ring != NULL) + seq_printf(m, " (%s)", obj->ring->name); +} + +static int i915_gem_object_list_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + uintptr_t list = (uintptr_t) node->info_ent->data; + struct list_head *head; + struct drm_device *dev = node->minor->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_i915_gem_object *obj; + size_t total_obj_size, total_gtt_size; + int count, ret; + + ret = mutex_lock_interruptible(&dev->struct_mutex); + if (ret) + return ret; + + switch (list) { + case ACTIVE_LIST: + seq_printf(m, "Active:\n"); + head = &dev_priv->mm.active_list; + break; + case INACTIVE_LIST: + seq_printf(m, "Inactive:\n"); + head = &dev_priv->mm.inactive_list; + break; + case PINNED_LIST: + seq_printf(m, "Pinned:\n"); + head = &dev_priv->mm.pinned_list; + break; + case FLUSHING_LIST: + seq_printf(m, "Flushing:\n"); + head = &dev_priv->mm.flushing_list; + break; + case DEFERRED_FREE_LIST: + seq_printf(m, "Deferred free:\n"); + head = &dev_priv->mm.deferred_free_list; + break; + default: + mutex_unlock(&dev->struct_mutex); + return -EINVAL; + } + + total_obj_size = total_gtt_size = count = 0; + list_for_each_entry(obj, head, mm_list) { + seq_printf(m, " "); + describe_obj(m, obj); + seq_printf(m, "\n"); + total_obj_size += obj->base.size; + total_gtt_size += obj->gtt_space->size; + count++; + } + mutex_unlock(&dev->struct_mutex); + + seq_printf(m, "Total %d objects, %zu bytes, %zu GTT size\n", + count, total_obj_size, total_gtt_size); + return 0; +} + +#define count_objects(list, member) do { \ + list_for_each_entry(obj, list, member) { \ + size += obj->gtt_space->size; \ + ++count; \ + if (obj->map_and_fenceable) { \ + mappable_size += obj->gtt_space->size; \ + ++mappable_count; \ + } \ + } \ +} while(0) + +static int i915_gem_object_info(struct seq_file *m, void* data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + u32 count, mappable_count; + size_t size, mappable_size; + struct drm_i915_gem_object *obj; + int ret; + + ret = mutex_lock_interruptible(&dev->struct_mutex); + if (ret) + return ret; + + seq_printf(m, "%u objects, %zu bytes\n", + dev_priv->mm.object_count, + dev_priv->mm.object_memory); + + size = count = mappable_size = mappable_count = 0; + count_objects(&dev_priv->mm.gtt_list, gtt_list); + seq_printf(m, "%u [%u] objects, %zu [%zu] bytes in gtt\n", + count, mappable_count, size, mappable_size); + + size = count = mappable_size = mappable_count = 0; + count_objects(&dev_priv->mm.active_list, mm_list); + count_objects(&dev_priv->mm.flushing_list, mm_list); + seq_printf(m, " %u [%u] active objects, %zu [%zu] bytes\n", + count, mappable_count, size, mappable_size); + + size = count = mappable_size = mappable_count = 0; + count_objects(&dev_priv->mm.pinned_list, mm_list); + seq_printf(m, " %u [%u] pinned objects, %zu [%zu] bytes\n", + count, mappable_count, size, mappable_size); + + size = count = mappable_size = mappable_count = 0; + count_objects(&dev_priv->mm.inactive_list, mm_list); + seq_printf(m, " %u [%u] inactive objects, %zu [%zu] bytes\n", + count, mappable_count, size, mappable_size); + + size = count = mappable_size = mappable_count = 0; + count_objects(&dev_priv->mm.deferred_free_list, mm_list); + seq_printf(m, " %u [%u] freed objects, %zu [%zu] bytes\n", + count, mappable_count, size, mappable_size); + + size = count = mappable_size = mappable_count = 0; + list_for_each_entry(obj, &dev_priv->mm.gtt_list, gtt_list) { + if (obj->fault_mappable) { + size += obj->gtt_space->size; + ++count; + } + if (obj->pin_mappable) { + mappable_size += obj->gtt_space->size; + ++mappable_count; + } + } + seq_printf(m, "%u pinned mappable objects, %zu bytes\n", + mappable_count, mappable_size); + seq_printf(m, "%u fault mappable objects, %zu bytes\n", + count, size); + + seq_printf(m, "%zu [%zu] gtt total\n", + dev_priv->mm.gtt_total, dev_priv->mm.mappable_gtt_total); + + mutex_unlock(&dev->struct_mutex); + + return 0; +} + +static int i915_gem_gtt_info(struct seq_file *m, void* data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_i915_gem_object *obj; + size_t total_obj_size, total_gtt_size; + int count, ret; + + ret = mutex_lock_interruptible(&dev->struct_mutex); + if (ret) + return ret; + + total_obj_size = total_gtt_size = count = 0; + list_for_each_entry(obj, &dev_priv->mm.gtt_list, gtt_list) { + seq_printf(m, " "); + describe_obj(m, obj); + seq_printf(m, "\n"); + total_obj_size += obj->base.size; + total_gtt_size += obj->gtt_space->size; + count++; + } + + mutex_unlock(&dev->struct_mutex); + + seq_printf(m, "Total %d objects, %zu bytes, %zu GTT size\n", + count, total_obj_size, total_gtt_size); + + return 0; +} + + +static int i915_gem_pageflip_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + unsigned long flags; + struct intel_crtc *crtc; + + list_for_each_entry(crtc, &dev->mode_config.crtc_list, base.head) { + const char pipe = pipe_name(crtc->pipe); + const char plane = plane_name(crtc->plane); + struct intel_unpin_work *work; + + spin_lock_irqsave(&dev->event_lock, flags); + work = crtc->unpin_work; + if (work == NULL) { + seq_printf(m, "No flip due on pipe %c (plane %c)\n", + pipe, plane); + } else { + if (!work->pending) { + seq_printf(m, "Flip queued on pipe %c (plane %c)\n", + pipe, plane); + } else { + seq_printf(m, "Flip pending (waiting for vsync) on pipe %c (plane %c)\n", + pipe, plane); + } + if (work->enable_stall_check) + seq_printf(m, "Stall check enabled, "); + else + seq_printf(m, "Stall check waiting for page flip ioctl, "); + seq_printf(m, "%d prepares\n", work->pending); + + if (work->old_fb_obj) { + struct drm_i915_gem_object *obj = work->old_fb_obj; + if (obj) + seq_printf(m, "Old framebuffer gtt_offset 0x%08x\n", obj->gtt_offset); + } + if (work->pending_flip_obj) { + struct drm_i915_gem_object *obj = work->pending_flip_obj; + if (obj) + seq_printf(m, "New framebuffer gtt_offset 0x%08x\n", obj->gtt_offset); + } + } + spin_unlock_irqrestore(&dev->event_lock, flags); + } + + return 0; +} + +static int i915_gem_request_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_i915_gem_request *gem_request; + int ret, count; + + ret = mutex_lock_interruptible(&dev->struct_mutex); + if (ret) + return ret; + + count = 0; + if (!list_empty(&dev_priv->ring[RCS].request_list)) { + seq_printf(m, "Render requests:\n"); + list_for_each_entry(gem_request, + &dev_priv->ring[RCS].request_list, + list) { + seq_printf(m, " %d @ %d\n", + gem_request->seqno, + (int) (jiffies - gem_request->emitted_jiffies)); + } + count++; + } + if (!list_empty(&dev_priv->ring[VCS].request_list)) { + seq_printf(m, "BSD requests:\n"); + list_for_each_entry(gem_request, + &dev_priv->ring[VCS].request_list, + list) { + seq_printf(m, " %d @ %d\n", + gem_request->seqno, + (int) (jiffies - gem_request->emitted_jiffies)); + } + count++; + } + if (!list_empty(&dev_priv->ring[BCS].request_list)) { + seq_printf(m, "BLT requests:\n"); + list_for_each_entry(gem_request, + &dev_priv->ring[BCS].request_list, + list) { + seq_printf(m, " %d @ %d\n", + gem_request->seqno, + (int) (jiffies - gem_request->emitted_jiffies)); + } + count++; + } + mutex_unlock(&dev->struct_mutex); + + if (count == 0) + seq_printf(m, "No requests\n"); + + return 0; +} + +static void i915_ring_seqno_info(struct seq_file *m, + struct intel_ring_buffer *ring) +{ + if (ring->get_seqno) { + seq_printf(m, "Current sequence (%s): %d\n", + ring->name, ring->get_seqno(ring)); + seq_printf(m, "Waiter sequence (%s): %d\n", + ring->name, ring->waiting_seqno); + seq_printf(m, "IRQ sequence (%s): %d\n", + ring->name, ring->irq_seqno); + } +} + +static int i915_gem_seqno_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + int ret, i; + + ret = mutex_lock_interruptible(&dev->struct_mutex); + if (ret) + return ret; + + for (i = 0; i < I915_NUM_RINGS; i++) + i915_ring_seqno_info(m, &dev_priv->ring[i]); + + mutex_unlock(&dev->struct_mutex); + + return 0; +} + + +static int i915_interrupt_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + int ret, i, pipe; + + ret = mutex_lock_interruptible(&dev->struct_mutex); + if (ret) + return ret; + + if (!HAS_PCH_SPLIT(dev)) { + seq_printf(m, "Interrupt enable: %08x\n", + I915_READ(IER)); + seq_printf(m, "Interrupt identity: %08x\n", + I915_READ(IIR)); + seq_printf(m, "Interrupt mask: %08x\n", + I915_READ(IMR)); + for_each_pipe(pipe) + seq_printf(m, "Pipe %c stat: %08x\n", + pipe_name(pipe), + I915_READ(PIPESTAT(pipe))); + } else { + seq_printf(m, "North Display Interrupt enable: %08x\n", + I915_READ(DEIER)); + seq_printf(m, "North Display Interrupt identity: %08x\n", + I915_READ(DEIIR)); + seq_printf(m, "North Display Interrupt mask: %08x\n", + I915_READ(DEIMR)); + seq_printf(m, "South Display Interrupt enable: %08x\n", + I915_READ(SDEIER)); + seq_printf(m, "South Display Interrupt identity: %08x\n", + I915_READ(SDEIIR)); + seq_printf(m, "South Display Interrupt mask: %08x\n", + I915_READ(SDEIMR)); + seq_printf(m, "Graphics Interrupt enable: %08x\n", + I915_READ(GTIER)); + seq_printf(m, "Graphics Interrupt identity: %08x\n", + I915_READ(GTIIR)); + seq_printf(m, "Graphics Interrupt mask: %08x\n", + I915_READ(GTIMR)); + } + seq_printf(m, "Interrupts received: %d\n", + atomic_read(&dev_priv->irq_received)); + for (i = 0; i < I915_NUM_RINGS; i++) { + if (IS_GEN6(dev)) { + seq_printf(m, "Graphics Interrupt mask (%s): %08x\n", + dev_priv->ring[i].name, + I915_READ_IMR(&dev_priv->ring[i])); + } + i915_ring_seqno_info(m, &dev_priv->ring[i]); + } + mutex_unlock(&dev->struct_mutex); + + return 0; +} + +static int i915_gem_fence_regs_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + int i, ret; + + ret = mutex_lock_interruptible(&dev->struct_mutex); + if (ret) + return ret; + + seq_printf(m, "Reserved fences = %d\n", dev_priv->fence_reg_start); + seq_printf(m, "Total fences = %d\n", dev_priv->num_fence_regs); + for (i = 0; i < dev_priv->num_fence_regs; i++) { + struct drm_i915_gem_object *obj = dev_priv->fence_regs[i].obj; + + seq_printf(m, "Fenced object[%2d] = ", i); + if (obj == NULL) + seq_printf(m, "unused"); + else + describe_obj(m, obj); + seq_printf(m, "\n"); + } + + mutex_unlock(&dev->struct_mutex); + return 0; +} + +static int i915_hws_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + struct intel_ring_buffer *ring; + const volatile u32 __iomem *hws; + int i; + + ring = &dev_priv->ring[(uintptr_t)node->info_ent->data]; + hws = (volatile u32 __iomem *)ring->status_page.page_addr; + if (hws == NULL) + return 0; + + for (i = 0; i < 4096 / sizeof(u32) / 4; i += 4) { + seq_printf(m, "0x%08x: 0x%08x 0x%08x 0x%08x 0x%08x\n", + i * 4, + hws[i], hws[i + 1], hws[i + 2], hws[i + 3]); + } + return 0; +} + +static void i915_dump_object(struct seq_file *m, + struct io_mapping *mapping, + struct drm_i915_gem_object *obj) +{ + int page, page_count, i; + + page_count = obj->base.size / PAGE_SIZE; + for (page = 0; page < page_count; page++) { + u32 *mem = io_mapping_map_wc(mapping, + obj->gtt_offset + page * PAGE_SIZE); + for (i = 0; i < PAGE_SIZE; i += 4) + seq_printf(m, "%08x : %08x\n", i, mem[i / 4]); + io_mapping_unmap(mem); + } +} + +static int i915_batchbuffer_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_i915_gem_object *obj; + int ret; + + ret = mutex_lock_interruptible(&dev->struct_mutex); + if (ret) + return ret; + + list_for_each_entry(obj, &dev_priv->mm.active_list, mm_list) { + if (obj->base.read_domains & I915_GEM_DOMAIN_COMMAND) { + seq_printf(m, "--- gtt_offset = 0x%08x\n", obj->gtt_offset); + i915_dump_object(m, dev_priv->mm.gtt_mapping, obj); + } + } + + mutex_unlock(&dev->struct_mutex); + return 0; +} + +static int i915_ringbuffer_data(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + struct intel_ring_buffer *ring; + int ret; + + ret = mutex_lock_interruptible(&dev->struct_mutex); + if (ret) + return ret; + + ring = &dev_priv->ring[(uintptr_t)node->info_ent->data]; + if (!ring->obj) { + seq_printf(m, "No ringbuffer setup\n"); + } else { + const u8 __iomem *virt = ring->virtual_start; + uint32_t off; + + for (off = 0; off < ring->size; off += 4) { + uint32_t *ptr = (uint32_t *)(virt + off); + seq_printf(m, "%08x : %08x\n", off, *ptr); + } + } + mutex_unlock(&dev->struct_mutex); + + return 0; +} + +static int i915_ringbuffer_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + struct intel_ring_buffer *ring; + + ring = &dev_priv->ring[(uintptr_t)node->info_ent->data]; + if (ring->size == 0) + return 0; + + seq_printf(m, "Ring %s:\n", ring->name); + seq_printf(m, " Head : %08x\n", I915_READ_HEAD(ring) & HEAD_ADDR); + seq_printf(m, " Tail : %08x\n", I915_READ_TAIL(ring) & TAIL_ADDR); + seq_printf(m, " Size : %08x\n", ring->size); + seq_printf(m, " Active : %08x\n", intel_ring_get_active_head(ring)); + seq_printf(m, " NOPID : %08x\n", I915_READ_NOPID(ring)); + if (IS_GEN6(dev)) { + seq_printf(m, " Sync 0 : %08x\n", I915_READ_SYNC_0(ring)); + seq_printf(m, " Sync 1 : %08x\n", I915_READ_SYNC_1(ring)); + } + seq_printf(m, " Control : %08x\n", I915_READ_CTL(ring)); + seq_printf(m, " Start : %08x\n", I915_READ_START(ring)); + + return 0; +} + +static const char *ring_str(int ring) +{ + switch (ring) { + case RING_RENDER: return " render"; + case RING_BSD: return " bsd"; + case RING_BLT: return " blt"; + default: return ""; + } +} + +static const char *pin_flag(int pinned) +{ + if (pinned > 0) + return " P"; + else if (pinned < 0) + return " p"; + else + return ""; +} + +static const char *tiling_flag(int tiling) +{ + switch (tiling) { + default: + case I915_TILING_NONE: return ""; + case I915_TILING_X: return " X"; + case I915_TILING_Y: return " Y"; + } +} + +static const char *dirty_flag(int dirty) +{ + return dirty ? " dirty" : ""; +} + +static const char *purgeable_flag(int purgeable) +{ + return purgeable ? " purgeable" : ""; +} + +static void print_error_buffers(struct seq_file *m, + const char *name, + struct drm_i915_error_buffer *err, + int count) +{ + seq_printf(m, "%s [%d]:\n", name, count); + + while (count--) { + seq_printf(m, " %08x %8u %04x %04x %08x%s%s%s%s%s%s", + err->gtt_offset, + err->size, + err->read_domains, + err->write_domain, + err->seqno, + pin_flag(err->pinned), + tiling_flag(err->tiling), + dirty_flag(err->dirty), + purgeable_flag(err->purgeable), + ring_str(err->ring), + cache_level_str(err->cache_level)); + + if (err->name) + seq_printf(m, " (name: %d)", err->name); + if (err->fence_reg != I915_FENCE_REG_NONE) + seq_printf(m, " (fence: %d)", err->fence_reg); + + seq_printf(m, "\n"); + err++; + } +} + +static int i915_error_state(struct seq_file *m, void *unused) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_i915_error_state *error; + unsigned long flags; + int i, page, offset, elt; + + spin_lock_irqsave(&dev_priv->error_lock, flags); + if (!dev_priv->first_error) { + seq_printf(m, "no error state collected\n"); + goto out; + } + + error = dev_priv->first_error; + + seq_printf(m, "Time: %ld s %ld us\n", error->time.tv_sec, + error->time.tv_usec); + seq_printf(m, "PCI ID: 0x%04x\n", dev->pci_device); + seq_printf(m, "EIR: 0x%08x\n", error->eir); + seq_printf(m, "PGTBL_ER: 0x%08x\n", error->pgtbl_er); + if (INTEL_INFO(dev)->gen >= 6) { + seq_printf(m, "ERROR: 0x%08x\n", error->error); + seq_printf(m, "Blitter command stream:\n"); + seq_printf(m, " ACTHD: 0x%08x\n", error->bcs_acthd); + seq_printf(m, " IPEIR: 0x%08x\n", error->bcs_ipeir); + seq_printf(m, " IPEHR: 0x%08x\n", error->bcs_ipehr); + seq_printf(m, " INSTDONE: 0x%08x\n", error->bcs_instdone); + seq_printf(m, " seqno: 0x%08x\n", error->bcs_seqno); + seq_printf(m, "Video (BSD) command stream:\n"); + seq_printf(m, " ACTHD: 0x%08x\n", error->vcs_acthd); + seq_printf(m, " IPEIR: 0x%08x\n", error->vcs_ipeir); + seq_printf(m, " IPEHR: 0x%08x\n", error->vcs_ipehr); + seq_printf(m, " INSTDONE: 0x%08x\n", error->vcs_instdone); + seq_printf(m, " seqno: 0x%08x\n", error->vcs_seqno); + } + seq_printf(m, "Render command stream:\n"); + seq_printf(m, " ACTHD: 0x%08x\n", error->acthd); + seq_printf(m, " IPEIR: 0x%08x\n", error->ipeir); + seq_printf(m, " IPEHR: 0x%08x\n", error->ipehr); + seq_printf(m, " INSTDONE: 0x%08x\n", error->instdone); + if (INTEL_INFO(dev)->gen >= 4) { + seq_printf(m, " INSTDONE1: 0x%08x\n", error->instdone1); + seq_printf(m, " INSTPS: 0x%08x\n", error->instps); + } + seq_printf(m, " INSTPM: 0x%08x\n", error->instpm); + seq_printf(m, " seqno: 0x%08x\n", error->seqno); + + for (i = 0; i < dev_priv->num_fence_regs; i++) + seq_printf(m, " fence[%d] = %08llx\n", i, error->fence[i]); + + if (error->active_bo) + print_error_buffers(m, "Active", + error->active_bo, + error->active_bo_count); + + if (error->pinned_bo) + print_error_buffers(m, "Pinned", + error->pinned_bo, + error->pinned_bo_count); + + for (i = 0; i < ARRAY_SIZE(error->batchbuffer); i++) { + if (error->batchbuffer[i]) { + struct drm_i915_error_object *obj = error->batchbuffer[i]; + + seq_printf(m, "%s --- gtt_offset = 0x%08x\n", + dev_priv->ring[i].name, + obj->gtt_offset); + offset = 0; + for (page = 0; page < obj->page_count; page++) { + for (elt = 0; elt < PAGE_SIZE/4; elt++) { + seq_printf(m, "%08x : %08x\n", offset, obj->pages[page][elt]); + offset += 4; + } + } + } + } + + for (i = 0; i < ARRAY_SIZE(error->ringbuffer); i++) { + if (error->ringbuffer[i]) { + struct drm_i915_error_object *obj = error->ringbuffer[i]; + seq_printf(m, "%s --- ringbuffer = 0x%08x\n", + dev_priv->ring[i].name, + obj->gtt_offset); + offset = 0; + for (page = 0; page < obj->page_count; page++) { + for (elt = 0; elt < PAGE_SIZE/4; elt++) { + seq_printf(m, "%08x : %08x\n", + offset, + obj->pages[page][elt]); + offset += 4; + } + } + } + } + + if (error->overlay) + intel_overlay_print_error_state(m, error->overlay); + + if (error->display) + intel_display_print_error_state(m, dev, error->display); + +out: + spin_unlock_irqrestore(&dev_priv->error_lock, flags); + + return 0; +} + +static int i915_rstdby_delays(struct seq_file *m, void *unused) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + u16 crstanddelay = I915_READ16(CRSTANDVID); + + seq_printf(m, "w/ctx: %d, w/o ctx: %d\n", (crstanddelay >> 8) & 0x3f, (crstanddelay & 0x3f)); + + return 0; +} + +static int i915_cur_delayinfo(struct seq_file *m, void *unused) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + int ret; + + if (IS_GEN5(dev)) { + u16 rgvswctl = I915_READ16(MEMSWCTL); + u16 rgvstat = I915_READ16(MEMSTAT_ILK); + + seq_printf(m, "Requested P-state: %d\n", (rgvswctl >> 8) & 0xf); + seq_printf(m, "Requested VID: %d\n", rgvswctl & 0x3f); + seq_printf(m, "Current VID: %d\n", (rgvstat & MEMSTAT_VID_MASK) >> + MEMSTAT_VID_SHIFT); + seq_printf(m, "Current P-state: %d\n", + (rgvstat & MEMSTAT_PSTATE_MASK) >> MEMSTAT_PSTATE_SHIFT); + } else if (IS_GEN6(dev) || IS_GEN7(dev)) { + u32 gt_perf_status = I915_READ(GEN6_GT_PERF_STATUS); + u32 rp_state_limits = I915_READ(GEN6_RP_STATE_LIMITS); + u32 rp_state_cap = I915_READ(GEN6_RP_STATE_CAP); + u32 rpstat; + u32 rpupei, rpcurup, rpprevup; + u32 rpdownei, rpcurdown, rpprevdown; + int max_freq; + + /* RPSTAT1 is in the GT power well */ + ret = mutex_lock_interruptible(&dev->struct_mutex); + if (ret) + return ret; + + gen6_gt_force_wake_get(dev_priv); + + rpstat = I915_READ(GEN6_RPSTAT1); + rpupei = I915_READ(GEN6_RP_CUR_UP_EI); + rpcurup = I915_READ(GEN6_RP_CUR_UP); + rpprevup = I915_READ(GEN6_RP_PREV_UP); + rpdownei = I915_READ(GEN6_RP_CUR_DOWN_EI); + rpcurdown = I915_READ(GEN6_RP_CUR_DOWN); + rpprevdown = I915_READ(GEN6_RP_PREV_DOWN); + + gen6_gt_force_wake_put(dev_priv); + mutex_unlock(&dev->struct_mutex); + + seq_printf(m, "GT_PERF_STATUS: 0x%08x\n", gt_perf_status); + seq_printf(m, "RPSTAT1: 0x%08x\n", rpstat); + seq_printf(m, "Render p-state ratio: %d\n", + (gt_perf_status & 0xff00) >> 8); + seq_printf(m, "Render p-state VID: %d\n", + gt_perf_status & 0xff); + seq_printf(m, "Render p-state limit: %d\n", + rp_state_limits & 0xff); + seq_printf(m, "CAGF: %dMHz\n", ((rpstat & GEN6_CAGF_MASK) >> + GEN6_CAGF_SHIFT) * 50); + seq_printf(m, "RP CUR UP EI: %dus\n", rpupei & + GEN6_CURICONT_MASK); + seq_printf(m, "RP CUR UP: %dus\n", rpcurup & + GEN6_CURBSYTAVG_MASK); + seq_printf(m, "RP PREV UP: %dus\n", rpprevup & + GEN6_CURBSYTAVG_MASK); + seq_printf(m, "RP CUR DOWN EI: %dus\n", rpdownei & + GEN6_CURIAVG_MASK); + seq_printf(m, "RP CUR DOWN: %dus\n", rpcurdown & + GEN6_CURBSYTAVG_MASK); + seq_printf(m, "RP PREV DOWN: %dus\n", rpprevdown & + GEN6_CURBSYTAVG_MASK); + + max_freq = (rp_state_cap & 0xff0000) >> 16; + seq_printf(m, "Lowest (RPN) frequency: %dMHz\n", + max_freq * 50); + + max_freq = (rp_state_cap & 0xff00) >> 8; + seq_printf(m, "Nominal (RP1) frequency: %dMHz\n", + max_freq * 50); + + max_freq = rp_state_cap & 0xff; + seq_printf(m, "Max non-overclocked (RP0) frequency: %dMHz\n", + max_freq * 50); + } else { + seq_printf(m, "no P-state info available\n"); + } + + return 0; +} + +static int i915_delayfreq_table(struct seq_file *m, void *unused) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + u32 delayfreq; + int i; + + for (i = 0; i < 16; i++) { + delayfreq = I915_READ(PXVFREQ_BASE + i * 4); + seq_printf(m, "P%02dVIDFREQ: 0x%08x (VID: %d)\n", i, delayfreq, + (delayfreq & PXVFREQ_PX_MASK) >> PXVFREQ_PX_SHIFT); + } + + return 0; +} + +static inline int MAP_TO_MV(int map) +{ + return 1250 - (map * 25); +} + +static int i915_inttoext_table(struct seq_file *m, void *unused) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + u32 inttoext; + int i; + + for (i = 1; i <= 32; i++) { + inttoext = I915_READ(INTTOEXT_BASE_ILK + i * 4); + seq_printf(m, "INTTOEXT%02d: 0x%08x\n", i, inttoext); + } + + return 0; +} + +static int i915_drpc_info(struct seq_file *m, void *unused) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + u32 rgvmodectl = I915_READ(MEMMODECTL); + u32 rstdbyctl = I915_READ(RSTDBYCTL); + u16 crstandvid = I915_READ16(CRSTANDVID); + + seq_printf(m, "HD boost: %s\n", (rgvmodectl & MEMMODE_BOOST_EN) ? + "yes" : "no"); + seq_printf(m, "Boost freq: %d\n", + (rgvmodectl & MEMMODE_BOOST_FREQ_MASK) >> + MEMMODE_BOOST_FREQ_SHIFT); + seq_printf(m, "HW control enabled: %s\n", + rgvmodectl & MEMMODE_HWIDLE_EN ? "yes" : "no"); + seq_printf(m, "SW control enabled: %s\n", + rgvmodectl & MEMMODE_SWMODE_EN ? "yes" : "no"); + seq_printf(m, "Gated voltage change: %s\n", + rgvmodectl & MEMMODE_RCLK_GATE ? "yes" : "no"); + seq_printf(m, "Starting frequency: P%d\n", + (rgvmodectl & MEMMODE_FSTART_MASK) >> MEMMODE_FSTART_SHIFT); + seq_printf(m, "Max P-state: P%d\n", + (rgvmodectl & MEMMODE_FMAX_MASK) >> MEMMODE_FMAX_SHIFT); + seq_printf(m, "Min P-state: P%d\n", (rgvmodectl & MEMMODE_FMIN_MASK)); + seq_printf(m, "RS1 VID: %d\n", (crstandvid & 0x3f)); + seq_printf(m, "RS2 VID: %d\n", ((crstandvid >> 8) & 0x3f)); + seq_printf(m, "Render standby enabled: %s\n", + (rstdbyctl & RCX_SW_EXIT) ? "no" : "yes"); + seq_printf(m, "Current RS state: "); + switch (rstdbyctl & RSX_STATUS_MASK) { + case RSX_STATUS_ON: + seq_printf(m, "on\n"); + break; + case RSX_STATUS_RC1: + seq_printf(m, "RC1\n"); + break; + case RSX_STATUS_RC1E: + seq_printf(m, "RC1E\n"); + break; + case RSX_STATUS_RS1: + seq_printf(m, "RS1\n"); + break; + case RSX_STATUS_RS2: + seq_printf(m, "RS2 (RC6)\n"); + break; + case RSX_STATUS_RS3: + seq_printf(m, "RC3 (RC6+)\n"); + break; + default: + seq_printf(m, "unknown\n"); + break; + } + + return 0; +} + +static int i915_fbc_status(struct seq_file *m, void *unused) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + + if (!I915_HAS_FBC(dev)) { + seq_printf(m, "FBC unsupported on this chipset\n"); + return 0; + } + + if (intel_fbc_enabled(dev)) { + seq_printf(m, "FBC enabled\n"); + } else { + seq_printf(m, "FBC disabled: "); + switch (dev_priv->no_fbc_reason) { + case FBC_NO_OUTPUT: + seq_printf(m, "no outputs"); + break; + case FBC_STOLEN_TOO_SMALL: + seq_printf(m, "not enough stolen memory"); + break; + case FBC_UNSUPPORTED_MODE: + seq_printf(m, "mode not supported"); + break; + case FBC_MODE_TOO_LARGE: + seq_printf(m, "mode too large"); + break; + case FBC_BAD_PLANE: + seq_printf(m, "FBC unsupported on plane"); + break; + case FBC_NOT_TILED: + seq_printf(m, "scanout buffer not tiled"); + break; + case FBC_MULTIPLE_PIPES: + seq_printf(m, "multiple pipes are enabled"); + break; + case FBC_MODULE_PARAM: + seq_printf(m, "disabled per module param (default off)"); + break; + default: + seq_printf(m, "unknown reason"); + } + seq_printf(m, "\n"); + } + return 0; +} + +static int i915_sr_status(struct seq_file *m, void *unused) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + bool sr_enabled = false; + + if (HAS_PCH_SPLIT(dev)) + sr_enabled = I915_READ(WM1_LP_ILK) & WM1_LP_SR_EN; + else if (IS_CRESTLINE(dev) || IS_I945G(dev) || IS_I945GM(dev)) + sr_enabled = I915_READ(FW_BLC_SELF) & FW_BLC_SELF_EN; + else if (IS_I915GM(dev)) + sr_enabled = I915_READ(INSTPM) & INSTPM_SELF_EN; + else if (IS_PINEVIEW(dev)) + sr_enabled = I915_READ(DSPFW3) & PINEVIEW_SELF_REFRESH_EN; + + seq_printf(m, "self-refresh: %s\n", + sr_enabled ? "enabled" : "disabled"); + + return 0; +} + +static int i915_emon_status(struct seq_file *m, void *unused) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + unsigned long temp, chipset, gfx; + int ret; + + ret = mutex_lock_interruptible(&dev->struct_mutex); + if (ret) + return ret; + + temp = i915_mch_val(dev_priv); + chipset = i915_chipset_val(dev_priv); + gfx = i915_gfx_val(dev_priv); + mutex_unlock(&dev->struct_mutex); + + seq_printf(m, "GMCH temp: %ld\n", temp); + seq_printf(m, "Chipset power: %ld\n", chipset); + seq_printf(m, "GFX power: %ld\n", gfx); + seq_printf(m, "Total power: %ld\n", chipset + gfx); + + return 0; +} + +static int i915_gfxec(struct seq_file *m, void *unused) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + + seq_printf(m, "GFXEC: %ld\n", (unsigned long)I915_READ(0x112f4)); + + return 0; +} + +static int i915_opregion(struct seq_file *m, void *unused) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + struct intel_opregion *opregion = &dev_priv->opregion; + int ret; + + ret = mutex_lock_interruptible(&dev->struct_mutex); + if (ret) + return ret; + + if (opregion->header) + seq_write(m, opregion->header, OPREGION_SIZE); + + mutex_unlock(&dev->struct_mutex); + + return 0; +} + +static int i915_gem_framebuffer_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + struct intel_fbdev *ifbdev; + struct intel_framebuffer *fb; + int ret; + + ret = mutex_lock_interruptible(&dev->mode_config.mutex); + if (ret) + return ret; + + ifbdev = dev_priv->fbdev; + fb = to_intel_framebuffer(ifbdev->helper.fb); + + seq_printf(m, "fbcon size: %d x %d, depth %d, %d bpp, obj ", + fb->base.width, + fb->base.height, + fb->base.depth, + fb->base.bits_per_pixel); + describe_obj(m, fb->obj); + seq_printf(m, "\n"); + + list_for_each_entry(fb, &dev->mode_config.fb_list, base.head) { + if (&fb->base == ifbdev->helper.fb) + continue; + + seq_printf(m, "user size: %d x %d, depth %d, %d bpp, obj ", + fb->base.width, + fb->base.height, + fb->base.depth, + fb->base.bits_per_pixel); + describe_obj(m, fb->obj); + seq_printf(m, "\n"); + } + + mutex_unlock(&dev->mode_config.mutex); + + return 0; +} + +static int i915_context_status(struct seq_file *m, void *unused) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + int ret; + + ret = mutex_lock_interruptible(&dev->mode_config.mutex); + if (ret) + return ret; + + if (dev_priv->pwrctx) { + seq_printf(m, "power context "); + describe_obj(m, dev_priv->pwrctx); + seq_printf(m, "\n"); + } + + if (dev_priv->renderctx) { + seq_printf(m, "render context "); + describe_obj(m, dev_priv->renderctx); + seq_printf(m, "\n"); + } + + mutex_unlock(&dev->mode_config.mutex); + + return 0; +} + +static int i915_gen6_forcewake_count_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + + seq_printf(m, "forcewake count = %d\n", + atomic_read(&dev_priv->forcewake_count)); + + return 0; +} + +static int +i915_wedged_open(struct inode *inode, + struct file *filp) +{ + filp->private_data = inode->i_private; + return 0; +} + +static ssize_t +i915_wedged_read(struct file *filp, + char __user *ubuf, + size_t max, + loff_t *ppos) +{ + struct drm_device *dev = filp->private_data; + drm_i915_private_t *dev_priv = dev->dev_private; + char buf[80]; + int len; + + len = snprintf(buf, sizeof (buf), + "wedged : %d\n", + atomic_read(&dev_priv->mm.wedged)); + + if (len > sizeof (buf)) + len = sizeof (buf); + + return simple_read_from_buffer(ubuf, max, ppos, buf, len); +} + +static ssize_t +i915_wedged_write(struct file *filp, + const char __user *ubuf, + size_t cnt, + loff_t *ppos) +{ + struct drm_device *dev = filp->private_data; + char buf[20]; + int val = 1; + + if (cnt > 0) { + if (cnt > sizeof (buf) - 1) + return -EINVAL; + + if (copy_from_user(buf, ubuf, cnt)) + return -EFAULT; + buf[cnt] = 0; + + val = simple_strtoul(buf, NULL, 0); + } + + DRM_INFO("Manually setting wedged to %d\n", val); + i915_handle_error(dev, val); + + return cnt; +} + +static const struct file_operations i915_wedged_fops = { + .owner = THIS_MODULE, + .open = i915_wedged_open, + .read = i915_wedged_read, + .write = i915_wedged_write, + .llseek = default_llseek, +}; + +/* As the drm_debugfs_init() routines are called before dev->dev_private is + * allocated we need to hook into the minor for release. */ +static int +drm_add_fake_info_node(struct drm_minor *minor, + struct dentry *ent, + const void *key) +{ + struct drm_info_node *node; + + node = kmalloc(sizeof(struct drm_info_node), GFP_KERNEL); + if (node == NULL) { + debugfs_remove(ent); + return -ENOMEM; + } + + node->minor = minor; + node->dent = ent; + node->info_ent = (void *) key; + list_add(&node->list, &minor->debugfs_nodes.list); + + return 0; +} + +static int i915_wedged_create(struct dentry *root, struct drm_minor *minor) +{ + struct drm_device *dev = minor->dev; + struct dentry *ent; + + ent = debugfs_create_file("i915_wedged", + S_IRUGO | S_IWUSR, + root, dev, + &i915_wedged_fops); + if (IS_ERR(ent)) + return PTR_ERR(ent); + + return drm_add_fake_info_node(minor, ent, &i915_wedged_fops); +} + +static int i915_forcewake_open(struct inode *inode, struct file *file) +{ + struct drm_device *dev = inode->i_private; + struct drm_i915_private *dev_priv = dev->dev_private; + int ret; + + if (!IS_GEN6(dev)) + return 0; + + ret = mutex_lock_interruptible(&dev->struct_mutex); + if (ret) + return ret; + gen6_gt_force_wake_get(dev_priv); + mutex_unlock(&dev->struct_mutex); + + return 0; +} + +int i915_forcewake_release(struct inode *inode, struct file *file) +{ + struct drm_device *dev = inode->i_private; + struct drm_i915_private *dev_priv = dev->dev_private; + + if (!IS_GEN6(dev)) + return 0; + + /* + * It's bad that we can potentially hang userspace if struct_mutex gets + * forever stuck. However, if we cannot acquire this lock it means that + * almost certainly the driver has hung, is not unload-able. Therefore + * hanging here is probably a minor inconvenience not to be seen my + * almost every user. + */ + mutex_lock(&dev->struct_mutex); + gen6_gt_force_wake_put(dev_priv); + mutex_unlock(&dev->struct_mutex); + + return 0; +} + +static const struct file_operations i915_forcewake_fops = { + .owner = THIS_MODULE, + .open = i915_forcewake_open, + .release = i915_forcewake_release, +}; + +static int i915_forcewake_create(struct dentry *root, struct drm_minor *minor) +{ + struct drm_device *dev = minor->dev; + struct dentry *ent; + + ent = debugfs_create_file("i915_forcewake_user", + S_IRUSR, + root, dev, + &i915_forcewake_fops); + if (IS_ERR(ent)) + return PTR_ERR(ent); + + return drm_add_fake_info_node(minor, ent, &i915_forcewake_fops); +} + +static struct drm_info_list i915_debugfs_list[] = { + {"i915_capabilities", i915_capabilities, 0}, + {"i915_gem_objects", i915_gem_object_info, 0}, + {"i915_gem_gtt", i915_gem_gtt_info, 0}, + {"i915_gem_active", i915_gem_object_list_info, 0, (void *) ACTIVE_LIST}, + {"i915_gem_flushing", i915_gem_object_list_info, 0, (void *) FLUSHING_LIST}, + {"i915_gem_inactive", i915_gem_object_list_info, 0, (void *) INACTIVE_LIST}, + {"i915_gem_pinned", i915_gem_object_list_info, 0, (void *) PINNED_LIST}, + {"i915_gem_deferred_free", i915_gem_object_list_info, 0, (void *) DEFERRED_FREE_LIST}, + {"i915_gem_pageflip", i915_gem_pageflip_info, 0}, + {"i915_gem_request", i915_gem_request_info, 0}, + {"i915_gem_seqno", i915_gem_seqno_info, 0}, + {"i915_gem_fence_regs", i915_gem_fence_regs_info, 0}, + {"i915_gem_interrupt", i915_interrupt_info, 0}, + {"i915_gem_hws", i915_hws_info, 0, (void *)RCS}, + {"i915_gem_hws_blt", i915_hws_info, 0, (void *)BCS}, + {"i915_gem_hws_bsd", i915_hws_info, 0, (void *)VCS}, + {"i915_ringbuffer_data", i915_ringbuffer_data, 0, (void *)RCS}, + {"i915_ringbuffer_info", i915_ringbuffer_info, 0, (void *)RCS}, + {"i915_bsd_ringbuffer_data", i915_ringbuffer_data, 0, (void *)VCS}, + {"i915_bsd_ringbuffer_info", i915_ringbuffer_info, 0, (void *)VCS}, + {"i915_blt_ringbuffer_data", i915_ringbuffer_data, 0, (void *)BCS}, + {"i915_blt_ringbuffer_info", i915_ringbuffer_info, 0, (void *)BCS}, + {"i915_batchbuffers", i915_batchbuffer_info, 0}, + {"i915_error_state", i915_error_state, 0}, + {"i915_rstdby_delays", i915_rstdby_delays, 0}, + {"i915_cur_delayinfo", i915_cur_delayinfo, 0}, + {"i915_delayfreq_table", i915_delayfreq_table, 0}, + {"i915_inttoext_table", i915_inttoext_table, 0}, + {"i915_drpc_info", i915_drpc_info, 0}, + {"i915_emon_status", i915_emon_status, 0}, + {"i915_gfxec", i915_gfxec, 0}, + {"i915_fbc_status", i915_fbc_status, 0}, + {"i915_sr_status", i915_sr_status, 0}, + {"i915_opregion", i915_opregion, 0}, + {"i915_gem_framebuffer", i915_gem_framebuffer_info, 0}, + {"i915_context_status", i915_context_status, 0}, + {"i915_gen6_forcewake_count", i915_gen6_forcewake_count_info, 0}, +}; +#define I915_DEBUGFS_ENTRIES ARRAY_SIZE(i915_debugfs_list) + +int i915_debugfs_init(struct drm_minor *minor) +{ + int ret; + + ret = i915_wedged_create(minor->debugfs_root, minor); + if (ret) + return ret; + + ret = i915_forcewake_create(minor->debugfs_root, minor); + if (ret) + return ret; + + return drm_debugfs_create_files(i915_debugfs_list, + I915_DEBUGFS_ENTRIES, + minor->debugfs_root, minor); +} + +void i915_debugfs_cleanup(struct drm_minor *minor) +{ + drm_debugfs_remove_files(i915_debugfs_list, + I915_DEBUGFS_ENTRIES, minor); + drm_debugfs_remove_files((struct drm_info_list *) &i915_forcewake_fops, + 1, minor); + drm_debugfs_remove_files((struct drm_info_list *) &i915_wedged_fops, + 1, minor); +} + +#endif /* CONFIG_DEBUG_FS */ diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c new file mode 100644 index 00000000..ef164432 --- /dev/null +++ b/drivers/gpu/drm/i915/i915_dma.c @@ -0,0 +1,2321 @@ +/* i915_dma.c -- DMA support for the I915 -*- linux-c -*- + */ +/* + * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "drmP.h" +#include "drm.h" +#include "drm_crtc_helper.h" +#include "drm_fb_helper.h" +#include "intel_drv.h" +#include "i915_drm.h" +#include "i915_drv.h" +#include "i915_trace.h" +#include "../../../platform/x86/intel_ips.h" +#include +#include +#include +#include +#include +#include +#include + +static void i915_write_hws_pga(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + u32 addr; + + addr = dev_priv->status_page_dmah->busaddr; + if (INTEL_INFO(dev)->gen >= 4) + addr |= (dev_priv->status_page_dmah->busaddr >> 28) & 0xf0; + I915_WRITE(HWS_PGA, addr); +} + +/** + * Sets up the hardware status page for devices that need a physical address + * in the register. + */ +static int i915_init_phys_hws(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + + /* Program Hardware Status Page */ + dev_priv->status_page_dmah = + drm_pci_alloc(dev, PAGE_SIZE, PAGE_SIZE); + + if (!dev_priv->status_page_dmah) { + DRM_ERROR("Can not allocate hardware status page\n"); + return -ENOMEM; + } + + memset_io((void __force __iomem *)dev_priv->status_page_dmah->vaddr, + 0, PAGE_SIZE); + + i915_write_hws_pga(dev); + + DRM_DEBUG_DRIVER("Enabled hardware status page\n"); + return 0; +} + +/** + * Frees the hardware status page, whether it's a physical address or a virtual + * address set up by the X Server. + */ +static void i915_free_hws(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct intel_ring_buffer *ring = LP_RING(dev_priv); + + if (dev_priv->status_page_dmah) { + drm_pci_free(dev, dev_priv->status_page_dmah); + dev_priv->status_page_dmah = NULL; + } + + if (ring->status_page.gfx_addr) { + ring->status_page.gfx_addr = 0; + drm_core_ioremapfree(&dev_priv->hws_map, dev); + } + + /* Need to rewrite hardware status page */ + I915_WRITE(HWS_PGA, 0x1ffff000); +} + +void i915_kernel_lost_context(struct drm_device * dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_i915_master_private *master_priv; + struct intel_ring_buffer *ring = LP_RING(dev_priv); + + /* + * We should never lose context on the ring with modesetting + * as we don't expose it to userspace + */ + if (drm_core_check_feature(dev, DRIVER_MODESET)) + return; + + ring->head = I915_READ_HEAD(ring) & HEAD_ADDR; + ring->tail = I915_READ_TAIL(ring) & TAIL_ADDR; + ring->space = ring->head - (ring->tail + 8); + if (ring->space < 0) + ring->space += ring->size; + + if (!dev->primary->master) + return; + + master_priv = dev->primary->master->driver_priv; + if (ring->head == ring->tail && master_priv->sarea_priv) + master_priv->sarea_priv->perf_boxes |= I915_BOX_RING_EMPTY; +} + +static int i915_dma_cleanup(struct drm_device * dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + int i; + + /* Make sure interrupts are disabled here because the uninstall ioctl + * may not have been called from userspace and after dev_private + * is freed, it's too late. + */ + if (dev->irq_enabled) + drm_irq_uninstall(dev); + + mutex_lock(&dev->struct_mutex); + for (i = 0; i < I915_NUM_RINGS; i++) + intel_cleanup_ring_buffer(&dev_priv->ring[i]); + mutex_unlock(&dev->struct_mutex); + + /* Clear the HWS virtual address at teardown */ + if (I915_NEED_GFX_HWS(dev)) + i915_free_hws(dev); + + return 0; +} + +static int i915_initialize(struct drm_device * dev, drm_i915_init_t * init) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_i915_master_private *master_priv = dev->primary->master->driver_priv; + int ret; + + master_priv->sarea = drm_getsarea(dev); + if (master_priv->sarea) { + master_priv->sarea_priv = (drm_i915_sarea_t *) + ((u8 *)master_priv->sarea->handle + init->sarea_priv_offset); + } else { + DRM_DEBUG_DRIVER("sarea not found assuming DRI2 userspace\n"); + } + + if (init->ring_size != 0) { + if (LP_RING(dev_priv)->obj != NULL) { + i915_dma_cleanup(dev); + DRM_ERROR("Client tried to initialize ringbuffer in " + "GEM mode\n"); + return -EINVAL; + } + + ret = intel_render_ring_init_dri(dev, + init->ring_start, + init->ring_size); + if (ret) { + i915_dma_cleanup(dev); + return ret; + } + } + + dev_priv->cpp = init->cpp; + dev_priv->back_offset = init->back_offset; + dev_priv->front_offset = init->front_offset; + dev_priv->current_page = 0; + if (master_priv->sarea_priv) + master_priv->sarea_priv->pf_current_page = 0; + + /* Allow hardware batchbuffers unless told otherwise. + */ + dev_priv->allow_batchbuffer = 1; + + return 0; +} + +static int i915_dma_resume(struct drm_device * dev) +{ + drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + struct intel_ring_buffer *ring = LP_RING(dev_priv); + + DRM_DEBUG_DRIVER("%s\n", __func__); + + if (ring->map.handle == NULL) { + DRM_ERROR("can not ioremap virtual address for" + " ring buffer\n"); + return -ENOMEM; + } + + /* Program Hardware Status Page */ + if (!ring->status_page.page_addr) { + DRM_ERROR("Can not find hardware status page\n"); + return -EINVAL; + } + DRM_DEBUG_DRIVER("hw status page @ %p\n", + ring->status_page.page_addr); + if (ring->status_page.gfx_addr != 0) + intel_ring_setup_status_page(ring); + else + i915_write_hws_pga(dev); + + DRM_DEBUG_DRIVER("Enabled hardware status page\n"); + + return 0; +} + +static int i915_dma_init(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + drm_i915_init_t *init = data; + int retcode = 0; + + switch (init->func) { + case I915_INIT_DMA: + retcode = i915_initialize(dev, init); + break; + case I915_CLEANUP_DMA: + retcode = i915_dma_cleanup(dev); + break; + case I915_RESUME_DMA: + retcode = i915_dma_resume(dev); + break; + default: + retcode = -EINVAL; + break; + } + + return retcode; +} + +/* Implement basically the same security restrictions as hardware does + * for MI_BATCH_NON_SECURE. These can be made stricter at any time. + * + * Most of the calculations below involve calculating the size of a + * particular instruction. It's important to get the size right as + * that tells us where the next instruction to check is. Any illegal + * instruction detected will be given a size of zero, which is a + * signal to abort the rest of the buffer. + */ +static int validate_cmd(int cmd) +{ + switch (((cmd >> 29) & 0x7)) { + case 0x0: + switch ((cmd >> 23) & 0x3f) { + case 0x0: + return 1; /* MI_NOOP */ + case 0x4: + return 1; /* MI_FLUSH */ + default: + return 0; /* disallow everything else */ + } + break; + case 0x1: + return 0; /* reserved */ + case 0x2: + return (cmd & 0xff) + 2; /* 2d commands */ + case 0x3: + if (((cmd >> 24) & 0x1f) <= 0x18) + return 1; + + switch ((cmd >> 24) & 0x1f) { + case 0x1c: + return 1; + case 0x1d: + switch ((cmd >> 16) & 0xff) { + case 0x3: + return (cmd & 0x1f) + 2; + case 0x4: + return (cmd & 0xf) + 2; + default: + return (cmd & 0xffff) + 2; + } + case 0x1e: + if (cmd & (1 << 23)) + return (cmd & 0xffff) + 1; + else + return 1; + case 0x1f: + if ((cmd & (1 << 23)) == 0) /* inline vertices */ + return (cmd & 0x1ffff) + 2; + else if (cmd & (1 << 17)) /* indirect random */ + if ((cmd & 0xffff) == 0) + return 0; /* unknown length, too hard */ + else + return (((cmd & 0xffff) + 1) / 2) + 1; + else + return 2; /* indirect sequential */ + default: + return 0; + } + default: + return 0; + } + + return 0; +} + +static int i915_emit_cmds(struct drm_device * dev, int *buffer, int dwords) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + int i, ret; + + if ((dwords+1) * sizeof(int) >= LP_RING(dev_priv)->size - 8) + return -EINVAL; + + for (i = 0; i < dwords;) { + int sz = validate_cmd(buffer[i]); + if (sz == 0 || i + sz > dwords) + return -EINVAL; + i += sz; + } + + ret = BEGIN_LP_RING((dwords+1)&~1); + if (ret) + return ret; + + for (i = 0; i < dwords; i++) + OUT_RING(buffer[i]); + if (dwords & 1) + OUT_RING(0); + + ADVANCE_LP_RING(); + + return 0; +} + +int +i915_emit_box(struct drm_device *dev, + struct drm_clip_rect *box, + int DR1, int DR4) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + int ret; + + if (box->y2 <= box->y1 || box->x2 <= box->x1 || + box->y2 <= 0 || box->x2 <= 0) { + DRM_ERROR("Bad box %d,%d..%d,%d\n", + box->x1, box->y1, box->x2, box->y2); + return -EINVAL; + } + + if (INTEL_INFO(dev)->gen >= 4) { + ret = BEGIN_LP_RING(4); + if (ret) + return ret; + + OUT_RING(GFX_OP_DRAWRECT_INFO_I965); + OUT_RING((box->x1 & 0xffff) | (box->y1 << 16)); + OUT_RING(((box->x2 - 1) & 0xffff) | ((box->y2 - 1) << 16)); + OUT_RING(DR4); + } else { + ret = BEGIN_LP_RING(6); + if (ret) + return ret; + + OUT_RING(GFX_OP_DRAWRECT_INFO); + OUT_RING(DR1); + OUT_RING((box->x1 & 0xffff) | (box->y1 << 16)); + OUT_RING(((box->x2 - 1) & 0xffff) | ((box->y2 - 1) << 16)); + OUT_RING(DR4); + OUT_RING(0); + } + ADVANCE_LP_RING(); + + return 0; +} + +/* XXX: Emitting the counter should really be moved to part of the IRQ + * emit. For now, do it in both places: + */ + +static void i915_emit_breadcrumb(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_i915_master_private *master_priv = dev->primary->master->driver_priv; + + dev_priv->counter++; + if (dev_priv->counter > 0x7FFFFFFFUL) + dev_priv->counter = 0; + if (master_priv->sarea_priv) + master_priv->sarea_priv->last_enqueue = dev_priv->counter; + + if (BEGIN_LP_RING(4) == 0) { + OUT_RING(MI_STORE_DWORD_INDEX); + OUT_RING(I915_BREADCRUMB_INDEX << MI_STORE_DWORD_INDEX_SHIFT); + OUT_RING(dev_priv->counter); + OUT_RING(0); + ADVANCE_LP_RING(); + } +} + +static int i915_dispatch_cmdbuffer(struct drm_device * dev, + drm_i915_cmdbuffer_t *cmd, + struct drm_clip_rect *cliprects, + void *cmdbuf) +{ + int nbox = cmd->num_cliprects; + int i = 0, count, ret; + + if (cmd->sz & 0x3) { + DRM_ERROR("alignment"); + return -EINVAL; + } + + i915_kernel_lost_context(dev); + + count = nbox ? nbox : 1; + + for (i = 0; i < count; i++) { + if (i < nbox) { + ret = i915_emit_box(dev, &cliprects[i], + cmd->DR1, cmd->DR4); + if (ret) + return ret; + } + + ret = i915_emit_cmds(dev, cmdbuf, cmd->sz / 4); + if (ret) + return ret; + } + + i915_emit_breadcrumb(dev); + return 0; +} + +static int i915_dispatch_batchbuffer(struct drm_device * dev, + drm_i915_batchbuffer_t * batch, + struct drm_clip_rect *cliprects) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + int nbox = batch->num_cliprects; + int i, count, ret; + + if ((batch->start | batch->used) & 0x7) { + DRM_ERROR("alignment"); + return -EINVAL; + } + + i915_kernel_lost_context(dev); + + count = nbox ? nbox : 1; + for (i = 0; i < count; i++) { + if (i < nbox) { + ret = i915_emit_box(dev, &cliprects[i], + batch->DR1, batch->DR4); + if (ret) + return ret; + } + + if (!IS_I830(dev) && !IS_845G(dev)) { + ret = BEGIN_LP_RING(2); + if (ret) + return ret; + + if (INTEL_INFO(dev)->gen >= 4) { + OUT_RING(MI_BATCH_BUFFER_START | (2 << 6) | MI_BATCH_NON_SECURE_I965); + OUT_RING(batch->start); + } else { + OUT_RING(MI_BATCH_BUFFER_START | (2 << 6)); + OUT_RING(batch->start | MI_BATCH_NON_SECURE); + } + } else { + ret = BEGIN_LP_RING(4); + if (ret) + return ret; + + OUT_RING(MI_BATCH_BUFFER); + OUT_RING(batch->start | MI_BATCH_NON_SECURE); + OUT_RING(batch->start + batch->used - 4); + OUT_RING(0); + } + ADVANCE_LP_RING(); + } + + + if (IS_G4X(dev) || IS_GEN5(dev)) { + if (BEGIN_LP_RING(2) == 0) { + OUT_RING(MI_FLUSH | MI_NO_WRITE_FLUSH | MI_INVALIDATE_ISP); + OUT_RING(MI_NOOP); + ADVANCE_LP_RING(); + } + } + + i915_emit_breadcrumb(dev); + return 0; +} + +static int i915_dispatch_flip(struct drm_device * dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_i915_master_private *master_priv = + dev->primary->master->driver_priv; + int ret; + + if (!master_priv->sarea_priv) + return -EINVAL; + + DRM_DEBUG_DRIVER("%s: page=%d pfCurrentPage=%d\n", + __func__, + dev_priv->current_page, + master_priv->sarea_priv->pf_current_page); + + i915_kernel_lost_context(dev); + + ret = BEGIN_LP_RING(10); + if (ret) + return ret; + + OUT_RING(MI_FLUSH | MI_READ_FLUSH); + OUT_RING(0); + + OUT_RING(CMD_OP_DISPLAYBUFFER_INFO | ASYNC_FLIP); + OUT_RING(0); + if (dev_priv->current_page == 0) { + OUT_RING(dev_priv->back_offset); + dev_priv->current_page = 1; + } else { + OUT_RING(dev_priv->front_offset); + dev_priv->current_page = 0; + } + OUT_RING(0); + + OUT_RING(MI_WAIT_FOR_EVENT | MI_WAIT_FOR_PLANE_A_FLIP); + OUT_RING(0); + + ADVANCE_LP_RING(); + + master_priv->sarea_priv->last_enqueue = dev_priv->counter++; + + if (BEGIN_LP_RING(4) == 0) { + OUT_RING(MI_STORE_DWORD_INDEX); + OUT_RING(I915_BREADCRUMB_INDEX << MI_STORE_DWORD_INDEX_SHIFT); + OUT_RING(dev_priv->counter); + OUT_RING(0); + ADVANCE_LP_RING(); + } + + master_priv->sarea_priv->pf_current_page = dev_priv->current_page; + return 0; +} + +static int i915_quiescent(struct drm_device *dev) +{ + struct intel_ring_buffer *ring = LP_RING(dev->dev_private); + + i915_kernel_lost_context(dev); + return intel_wait_ring_idle(ring); +} + +static int i915_flush_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + int ret; + + RING_LOCK_TEST_WITH_RETURN(dev, file_priv); + + mutex_lock(&dev->struct_mutex); + ret = i915_quiescent(dev); + mutex_unlock(&dev->struct_mutex); + + return ret; +} + +static int i915_batchbuffer(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + struct drm_i915_master_private *master_priv = dev->primary->master->driver_priv; + drm_i915_sarea_t *sarea_priv = (drm_i915_sarea_t *) + master_priv->sarea_priv; + drm_i915_batchbuffer_t *batch = data; + int ret; + struct drm_clip_rect *cliprects = NULL; + + if (!dev_priv->allow_batchbuffer) { + DRM_ERROR("Batchbuffer ioctl disabled\n"); + return -EINVAL; + } + + DRM_DEBUG_DRIVER("i915 batchbuffer, start %x used %d cliprects %d\n", + batch->start, batch->used, batch->num_cliprects); + + RING_LOCK_TEST_WITH_RETURN(dev, file_priv); + + if (batch->num_cliprects < 0) + return -EINVAL; + + if (batch->num_cliprects) { + cliprects = kcalloc(batch->num_cliprects, + sizeof(struct drm_clip_rect), + GFP_KERNEL); + if (cliprects == NULL) + return -ENOMEM; + + ret = copy_from_user(cliprects, batch->cliprects, + batch->num_cliprects * + sizeof(struct drm_clip_rect)); + if (ret != 0) { + ret = -EFAULT; + goto fail_free; + } + } + + mutex_lock(&dev->struct_mutex); + ret = i915_dispatch_batchbuffer(dev, batch, cliprects); + mutex_unlock(&dev->struct_mutex); + + if (sarea_priv) + sarea_priv->last_dispatch = READ_BREADCRUMB(dev_priv); + +fail_free: + kfree(cliprects); + + return ret; +} + +static int i915_cmdbuffer(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + struct drm_i915_master_private *master_priv = dev->primary->master->driver_priv; + drm_i915_sarea_t *sarea_priv = (drm_i915_sarea_t *) + master_priv->sarea_priv; + drm_i915_cmdbuffer_t *cmdbuf = data; + struct drm_clip_rect *cliprects = NULL; + void *batch_data; + int ret; + + DRM_DEBUG_DRIVER("i915 cmdbuffer, buf %p sz %d cliprects %d\n", + cmdbuf->buf, cmdbuf->sz, cmdbuf->num_cliprects); + + RING_LOCK_TEST_WITH_RETURN(dev, file_priv); + + if (cmdbuf->num_cliprects < 0) + return -EINVAL; + + batch_data = kmalloc(cmdbuf->sz, GFP_KERNEL); + if (batch_data == NULL) + return -ENOMEM; + + ret = copy_from_user(batch_data, cmdbuf->buf, cmdbuf->sz); + if (ret != 0) { + ret = -EFAULT; + goto fail_batch_free; + } + + if (cmdbuf->num_cliprects) { + cliprects = kcalloc(cmdbuf->num_cliprects, + sizeof(struct drm_clip_rect), GFP_KERNEL); + if (cliprects == NULL) { + ret = -ENOMEM; + goto fail_batch_free; + } + + ret = copy_from_user(cliprects, cmdbuf->cliprects, + cmdbuf->num_cliprects * + sizeof(struct drm_clip_rect)); + if (ret != 0) { + ret = -EFAULT; + goto fail_clip_free; + } + } + + mutex_lock(&dev->struct_mutex); + ret = i915_dispatch_cmdbuffer(dev, cmdbuf, cliprects, batch_data); + mutex_unlock(&dev->struct_mutex); + if (ret) { + DRM_ERROR("i915_dispatch_cmdbuffer failed\n"); + goto fail_clip_free; + } + + if (sarea_priv) + sarea_priv->last_dispatch = READ_BREADCRUMB(dev_priv); + +fail_clip_free: + kfree(cliprects); +fail_batch_free: + kfree(batch_data); + + return ret; +} + +static int i915_flip_bufs(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + int ret; + + DRM_DEBUG_DRIVER("%s\n", __func__); + + RING_LOCK_TEST_WITH_RETURN(dev, file_priv); + + mutex_lock(&dev->struct_mutex); + ret = i915_dispatch_flip(dev); + mutex_unlock(&dev->struct_mutex); + + return ret; +} + +static int i915_getparam(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + drm_i915_getparam_t *param = data; + int value; + + if (!dev_priv) { + DRM_ERROR("called with no initialization\n"); + return -EINVAL; + } + + switch (param->param) { + case I915_PARAM_IRQ_ACTIVE: + value = dev->pdev->irq ? 1 : 0; + break; + case I915_PARAM_ALLOW_BATCHBUFFER: + value = dev_priv->allow_batchbuffer ? 1 : 0; + break; + case I915_PARAM_LAST_DISPATCH: + value = READ_BREADCRUMB(dev_priv); + break; + case I915_PARAM_CHIPSET_ID: + value = dev->pci_device; + break; + case I915_PARAM_HAS_GEM: + value = dev_priv->has_gem; + break; + case I915_PARAM_NUM_FENCES_AVAIL: + value = dev_priv->num_fence_regs - dev_priv->fence_reg_start; + break; + case I915_PARAM_HAS_OVERLAY: + value = dev_priv->overlay ? 1 : 0; + break; + case I915_PARAM_HAS_PAGEFLIPPING: + value = 1; + break; + case I915_PARAM_HAS_EXECBUF2: + /* depends on GEM */ + value = dev_priv->has_gem; + break; + case I915_PARAM_HAS_BSD: + value = HAS_BSD(dev); + break; + case I915_PARAM_HAS_BLT: + value = HAS_BLT(dev); + break; + case I915_PARAM_HAS_RELAXED_FENCING: + value = 1; + break; + case I915_PARAM_HAS_COHERENT_RINGS: + value = 1; + break; + case I915_PARAM_HAS_EXEC_CONSTANTS: + value = INTEL_INFO(dev)->gen >= 4; + break; + case I915_PARAM_HAS_RELAXED_DELTA: + value = 1; + break; + default: + DRM_DEBUG_DRIVER("Unknown parameter %d\n", + param->param); + return -EINVAL; + } + + if (DRM_COPY_TO_USER(param->value, &value, sizeof(int))) { + DRM_ERROR("DRM_COPY_TO_USER failed\n"); + return -EFAULT; + } + + return 0; +} + +static int i915_setparam(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + drm_i915_setparam_t *param = data; + + if (!dev_priv) { + DRM_ERROR("called with no initialization\n"); + return -EINVAL; + } + + switch (param->param) { + case I915_SETPARAM_USE_MI_BATCHBUFFER_START: + break; + case I915_SETPARAM_TEX_LRU_LOG_GRANULARITY: + dev_priv->tex_lru_log_granularity = param->value; + break; + case I915_SETPARAM_ALLOW_BATCHBUFFER: + dev_priv->allow_batchbuffer = param->value; + break; + case I915_SETPARAM_NUM_USED_FENCES: + if (param->value > dev_priv->num_fence_regs || + param->value < 0) + return -EINVAL; + /* Userspace can use first N regs */ + dev_priv->fence_reg_start = param->value; + break; + default: + DRM_DEBUG_DRIVER("unknown parameter %d\n", + param->param); + return -EINVAL; + } + + return 0; +} + +static int i915_set_status_page(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + drm_i915_hws_addr_t *hws = data; + struct intel_ring_buffer *ring = LP_RING(dev_priv); + + if (!I915_NEED_GFX_HWS(dev)) + return -EINVAL; + + if (!dev_priv) { + DRM_ERROR("called with no initialization\n"); + return -EINVAL; + } + + if (drm_core_check_feature(dev, DRIVER_MODESET)) { + WARN(1, "tried to set status page when mode setting active\n"); + return 0; + } + + DRM_DEBUG_DRIVER("set status page addr 0x%08x\n", (u32)hws->addr); + + ring->status_page.gfx_addr = hws->addr & (0x1ffff<<12); + + dev_priv->hws_map.offset = dev->agp->base + hws->addr; + dev_priv->hws_map.size = 4*1024; + dev_priv->hws_map.type = 0; + dev_priv->hws_map.flags = 0; + dev_priv->hws_map.mtrr = 0; + + drm_core_ioremap_wc(&dev_priv->hws_map, dev); + if (dev_priv->hws_map.handle == NULL) { + i915_dma_cleanup(dev); + ring->status_page.gfx_addr = 0; + DRM_ERROR("can not ioremap virtual address for" + " G33 hw status page\n"); + return -ENOMEM; + } + ring->status_page.page_addr = + (void __force __iomem *)dev_priv->hws_map.handle; + memset_io(ring->status_page.page_addr, 0, PAGE_SIZE); + I915_WRITE(HWS_PGA, ring->status_page.gfx_addr); + + DRM_DEBUG_DRIVER("load hws HWS_PGA with gfx mem 0x%x\n", + ring->status_page.gfx_addr); + DRM_DEBUG_DRIVER("load hws at %p\n", + ring->status_page.page_addr); + return 0; +} + +static int i915_get_bridge_dev(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + dev_priv->bridge_dev = pci_get_bus_and_slot(0, PCI_DEVFN(0,0)); + if (!dev_priv->bridge_dev) { + DRM_ERROR("bridge device not found\n"); + return -1; + } + return 0; +} + +#define MCHBAR_I915 0x44 +#define MCHBAR_I965 0x48 +#define MCHBAR_SIZE (4*4096) + +#define DEVEN_REG 0x54 +#define DEVEN_MCHBAR_EN (1 << 28) + +/* Allocate space for the MCH regs if needed, return nonzero on error */ +static int +intel_alloc_mchbar_resource(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + int reg = INTEL_INFO(dev)->gen >= 4 ? MCHBAR_I965 : MCHBAR_I915; + u32 temp_lo, temp_hi = 0; + u64 mchbar_addr; + int ret; + + if (INTEL_INFO(dev)->gen >= 4) + pci_read_config_dword(dev_priv->bridge_dev, reg + 4, &temp_hi); + pci_read_config_dword(dev_priv->bridge_dev, reg, &temp_lo); + mchbar_addr = ((u64)temp_hi << 32) | temp_lo; + + /* If ACPI doesn't have it, assume we need to allocate it ourselves */ +#ifdef CONFIG_PNP + if (mchbar_addr && + pnp_range_reserved(mchbar_addr, mchbar_addr + MCHBAR_SIZE)) + return 0; +#endif + + /* Get some space for it */ + dev_priv->mch_res.name = "i915 MCHBAR"; + dev_priv->mch_res.flags = IORESOURCE_MEM; + ret = pci_bus_alloc_resource(dev_priv->bridge_dev->bus, + &dev_priv->mch_res, + MCHBAR_SIZE, MCHBAR_SIZE, + PCIBIOS_MIN_MEM, + 0, pcibios_align_resource, + dev_priv->bridge_dev); + if (ret) { + DRM_DEBUG_DRIVER("failed bus alloc: %d\n", ret); + dev_priv->mch_res.start = 0; + return ret; + } + + if (INTEL_INFO(dev)->gen >= 4) + pci_write_config_dword(dev_priv->bridge_dev, reg + 4, + upper_32_bits(dev_priv->mch_res.start)); + + pci_write_config_dword(dev_priv->bridge_dev, reg, + lower_32_bits(dev_priv->mch_res.start)); + return 0; +} + +/* Setup MCHBAR if possible, return true if we should disable it again */ +static void +intel_setup_mchbar(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + int mchbar_reg = INTEL_INFO(dev)->gen >= 4 ? MCHBAR_I965 : MCHBAR_I915; + u32 temp; + bool enabled; + + dev_priv->mchbar_need_disable = false; + + if (IS_I915G(dev) || IS_I915GM(dev)) { + pci_read_config_dword(dev_priv->bridge_dev, DEVEN_REG, &temp); + enabled = !!(temp & DEVEN_MCHBAR_EN); + } else { + pci_read_config_dword(dev_priv->bridge_dev, mchbar_reg, &temp); + enabled = temp & 1; + } + + /* If it's already enabled, don't have to do anything */ + if (enabled) + return; + + if (intel_alloc_mchbar_resource(dev)) + return; + + dev_priv->mchbar_need_disable = true; + + /* Space is allocated or reserved, so enable it. */ + if (IS_I915G(dev) || IS_I915GM(dev)) { + pci_write_config_dword(dev_priv->bridge_dev, DEVEN_REG, + temp | DEVEN_MCHBAR_EN); + } else { + pci_read_config_dword(dev_priv->bridge_dev, mchbar_reg, &temp); + pci_write_config_dword(dev_priv->bridge_dev, mchbar_reg, temp | 1); + } +} + +static void +intel_teardown_mchbar(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + int mchbar_reg = INTEL_INFO(dev)->gen >= 4 ? MCHBAR_I965 : MCHBAR_I915; + u32 temp; + + if (dev_priv->mchbar_need_disable) { + if (IS_I915G(dev) || IS_I915GM(dev)) { + pci_read_config_dword(dev_priv->bridge_dev, DEVEN_REG, &temp); + temp &= ~DEVEN_MCHBAR_EN; + pci_write_config_dword(dev_priv->bridge_dev, DEVEN_REG, temp); + } else { + pci_read_config_dword(dev_priv->bridge_dev, mchbar_reg, &temp); + temp &= ~1; + pci_write_config_dword(dev_priv->bridge_dev, mchbar_reg, temp); + } + } + + if (dev_priv->mch_res.start) + release_resource(&dev_priv->mch_res); +} + +#define PTE_ADDRESS_MASK 0xfffff000 +#define PTE_ADDRESS_MASK_HIGH 0x000000f0 /* i915+ */ +#define PTE_MAPPING_TYPE_UNCACHED (0 << 1) +#define PTE_MAPPING_TYPE_DCACHE (1 << 1) /* i830 only */ +#define PTE_MAPPING_TYPE_CACHED (3 << 1) +#define PTE_MAPPING_TYPE_MASK (3 << 1) +#define PTE_VALID (1 << 0) + +/** + * i915_stolen_to_phys - take an offset into stolen memory and turn it into + * a physical one + * @dev: drm device + * @offset: address to translate + * + * Some chip functions require allocations from stolen space and need the + * physical address of the memory in question. + */ +static unsigned long i915_stolen_to_phys(struct drm_device *dev, u32 offset) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct pci_dev *pdev = dev_priv->bridge_dev; + u32 base; + +#if 0 + /* On the machines I have tested the Graphics Base of Stolen Memory + * is unreliable, so compute the base by subtracting the stolen memory + * from the Top of Low Usable DRAM which is where the BIOS places + * the graphics stolen memory. + */ + if (INTEL_INFO(dev)->gen > 3 || IS_G33(dev)) { + /* top 32bits are reserved = 0 */ + pci_read_config_dword(pdev, 0xA4, &base); + } else { + /* XXX presume 8xx is the same as i915 */ + pci_bus_read_config_dword(pdev->bus, 2, 0x5C, &base); + } +#else + if (INTEL_INFO(dev)->gen > 3 || IS_G33(dev)) { + u16 val; + pci_read_config_word(pdev, 0xb0, &val); + base = val >> 4 << 20; + } else { + u8 val; + pci_read_config_byte(pdev, 0x9c, &val); + base = val >> 3 << 27; + } + base -= dev_priv->mm.gtt->stolen_size; +#endif + + return base + offset; +} + +static void i915_warn_stolen(struct drm_device *dev) +{ + DRM_ERROR("not enough stolen space for compressed buffer, disabling\n"); + DRM_ERROR("hint: you may be able to increase stolen memory size in the BIOS to avoid this\n"); +} + +static void i915_setup_compression(struct drm_device *dev, int size) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_mm_node *compressed_fb, *uninitialized_var(compressed_llb); + unsigned long cfb_base; + unsigned long ll_base = 0; + + compressed_fb = drm_mm_search_free(&dev_priv->mm.stolen, size, 4096, 0); + if (compressed_fb) + compressed_fb = drm_mm_get_block(compressed_fb, size, 4096); + if (!compressed_fb) + goto err; + + cfb_base = i915_stolen_to_phys(dev, compressed_fb->start); + if (!cfb_base) + goto err_fb; + + if (!(IS_GM45(dev) || HAS_PCH_SPLIT(dev))) { + compressed_llb = drm_mm_search_free(&dev_priv->mm.stolen, + 4096, 4096, 0); + if (compressed_llb) + compressed_llb = drm_mm_get_block(compressed_llb, + 4096, 4096); + if (!compressed_llb) + goto err_fb; + + ll_base = i915_stolen_to_phys(dev, compressed_llb->start); + if (!ll_base) + goto err_llb; + } + + dev_priv->cfb_size = size; + + intel_disable_fbc(dev); + dev_priv->compressed_fb = compressed_fb; + if (HAS_PCH_SPLIT(dev)) + I915_WRITE(ILK_DPFC_CB_BASE, compressed_fb->start); + else if (IS_GM45(dev)) { + I915_WRITE(DPFC_CB_BASE, compressed_fb->start); + } else { + I915_WRITE(FBC_CFB_BASE, cfb_base); + I915_WRITE(FBC_LL_BASE, ll_base); + dev_priv->compressed_llb = compressed_llb; + } + + DRM_DEBUG_KMS("FBC base 0x%08lx, ll base 0x%08lx, size %dM\n", + cfb_base, ll_base, size >> 20); + return; + +err_llb: + drm_mm_put_block(compressed_llb); +err_fb: + drm_mm_put_block(compressed_fb); +err: + dev_priv->no_fbc_reason = FBC_STOLEN_TOO_SMALL; + i915_warn_stolen(dev); +} + +static void i915_cleanup_compression(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + drm_mm_put_block(dev_priv->compressed_fb); + if (dev_priv->compressed_llb) + drm_mm_put_block(dev_priv->compressed_llb); +} + +/* true = enable decode, false = disable decoder */ +static unsigned int i915_vga_set_decode(void *cookie, bool state) +{ + struct drm_device *dev = cookie; + + intel_modeset_vga_set_state(dev, state); + if (state) + return VGA_RSRC_LEGACY_IO | VGA_RSRC_LEGACY_MEM | + VGA_RSRC_NORMAL_IO | VGA_RSRC_NORMAL_MEM; + else + return VGA_RSRC_NORMAL_IO | VGA_RSRC_NORMAL_MEM; +} + +static void i915_switcheroo_set_state(struct pci_dev *pdev, enum vga_switcheroo_state state) +{ + struct drm_device *dev = pci_get_drvdata(pdev); + pm_message_t pmm = { .event = PM_EVENT_SUSPEND }; + if (state == VGA_SWITCHEROO_ON) { + printk(KERN_INFO "i915: switched on\n"); + dev->switch_power_state = DRM_SWITCH_POWER_CHANGING; + /* i915 resume handler doesn't set to D0 */ + pci_set_power_state(dev->pdev, PCI_D0); + i915_resume(dev); + dev->switch_power_state = DRM_SWITCH_POWER_ON; + } else { + printk(KERN_ERR "i915: switched off\n"); + dev->switch_power_state = DRM_SWITCH_POWER_CHANGING; + i915_suspend(dev, pmm); + dev->switch_power_state = DRM_SWITCH_POWER_OFF; + } +} + +static bool i915_switcheroo_can_switch(struct pci_dev *pdev) +{ + struct drm_device *dev = pci_get_drvdata(pdev); + bool can_switch; + + spin_lock(&dev->count_lock); + can_switch = (dev->open_count == 0); + spin_unlock(&dev->count_lock); + return can_switch; +} + +static int i915_load_gem_init(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + unsigned long prealloc_size, gtt_size, mappable_size; + int ret; + + prealloc_size = dev_priv->mm.gtt->stolen_size; + gtt_size = dev_priv->mm.gtt->gtt_total_entries << PAGE_SHIFT; + mappable_size = dev_priv->mm.gtt->gtt_mappable_entries << PAGE_SHIFT; + + /* Basic memrange allocator for stolen space */ + drm_mm_init(&dev_priv->mm.stolen, 0, prealloc_size); + + /* Let GEM Manage all of the aperture. + * + * However, leave one page at the end still bound to the scratch page. + * There are a number of places where the hardware apparently + * prefetches past the end of the object, and we've seen multiple + * hangs with the GPU head pointer stuck in a batchbuffer bound + * at the last page of the aperture. One page should be enough to + * keep any prefetching inside of the aperture. + */ + i915_gem_do_init(dev, 0, mappable_size, gtt_size - PAGE_SIZE); + + mutex_lock(&dev->struct_mutex); + ret = i915_gem_init_ringbuffer(dev); + mutex_unlock(&dev->struct_mutex); + if (ret) + return ret; + + /* Try to set up FBC with a reasonable compressed buffer size */ + if (I915_HAS_FBC(dev) && i915_powersave) { + int cfb_size; + + /* Leave 1M for line length buffer & misc. */ + + /* Try to get a 32M buffer... */ + if (prealloc_size > (36*1024*1024)) + cfb_size = 32*1024*1024; + else /* fall back to 7/8 of the stolen space */ + cfb_size = prealloc_size * 7 / 8; + i915_setup_compression(dev, cfb_size); + } + + /* Allow hardware batchbuffers unless told otherwise. */ + dev_priv->allow_batchbuffer = 1; + return 0; +} + +static int i915_load_modeset_init(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + int ret; + + ret = intel_parse_bios(dev); + if (ret) + DRM_INFO("failed to find VBIOS tables\n"); + + /* If we have > 1 VGA cards, then we need to arbitrate access + * to the common VGA resources. + * + * If we are a secondary display controller (!PCI_DISPLAY_CLASS_VGA), + * then we do not take part in VGA arbitration and the + * vga_client_register() fails with -ENODEV. + */ + ret = vga_client_register(dev->pdev, dev, NULL, i915_vga_set_decode); + if (ret && ret != -ENODEV) + goto out; + + intel_register_dsm_handler(); + + ret = vga_switcheroo_register_client(dev->pdev, + i915_switcheroo_set_state, + NULL, + i915_switcheroo_can_switch); + if (ret) + goto cleanup_vga_client; + + /* IIR "flip pending" bit means done if this bit is set */ + if (IS_GEN3(dev) && (I915_READ(ECOSKPD) & ECO_FLIP_DONE)) + dev_priv->flip_pending_is_done = true; + + intel_modeset_init(dev); + + ret = i915_load_gem_init(dev); + if (ret) + goto cleanup_vga_switcheroo; + + intel_modeset_gem_init(dev); + + ret = drm_irq_install(dev); + if (ret) + goto cleanup_gem; + + /* Always safe in the mode setting case. */ + /* FIXME: do pre/post-mode set stuff in core KMS code */ + dev->vblank_disable_allowed = 1; + + ret = intel_fbdev_init(dev); + if (ret) + goto cleanup_irq; + + drm_kms_helper_poll_init(dev); + + /* We're off and running w/KMS */ + dev_priv->mm.suspended = 0; + + return 0; + +cleanup_irq: + drm_irq_uninstall(dev); +cleanup_gem: + mutex_lock(&dev->struct_mutex); + i915_gem_cleanup_ringbuffer(dev); + mutex_unlock(&dev->struct_mutex); +cleanup_vga_switcheroo: + vga_switcheroo_unregister_client(dev->pdev); +cleanup_vga_client: + vga_client_register(dev->pdev, NULL, NULL, NULL); +out: + return ret; +} + +int i915_master_create(struct drm_device *dev, struct drm_master *master) +{ + struct drm_i915_master_private *master_priv; + + master_priv = kzalloc(sizeof(*master_priv), GFP_KERNEL); + if (!master_priv) + return -ENOMEM; + + master->driver_priv = master_priv; + return 0; +} + +void i915_master_destroy(struct drm_device *dev, struct drm_master *master) +{ + struct drm_i915_master_private *master_priv = master->driver_priv; + + if (!master_priv) + return; + + kfree(master_priv); + + master->driver_priv = NULL; +} + +static void i915_pineview_get_mem_freq(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + u32 tmp; + + tmp = I915_READ(CLKCFG); + + switch (tmp & CLKCFG_FSB_MASK) { + case CLKCFG_FSB_533: + dev_priv->fsb_freq = 533; /* 133*4 */ + break; + case CLKCFG_FSB_800: + dev_priv->fsb_freq = 800; /* 200*4 */ + break; + case CLKCFG_FSB_667: + dev_priv->fsb_freq = 667; /* 167*4 */ + break; + case CLKCFG_FSB_400: + dev_priv->fsb_freq = 400; /* 100*4 */ + break; + } + + switch (tmp & CLKCFG_MEM_MASK) { + case CLKCFG_MEM_533: + dev_priv->mem_freq = 533; + break; + case CLKCFG_MEM_667: + dev_priv->mem_freq = 667; + break; + case CLKCFG_MEM_800: + dev_priv->mem_freq = 800; + break; + } + + /* detect pineview DDR3 setting */ + tmp = I915_READ(CSHRDDR3CTL); + dev_priv->is_ddr3 = (tmp & CSHRDDR3CTL_DDR3) ? 1 : 0; +} + +static void i915_ironlake_get_mem_freq(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + u16 ddrpll, csipll; + + ddrpll = I915_READ16(DDRMPLL1); + csipll = I915_READ16(CSIPLL0); + + switch (ddrpll & 0xff) { + case 0xc: + dev_priv->mem_freq = 800; + break; + case 0x10: + dev_priv->mem_freq = 1066; + break; + case 0x14: + dev_priv->mem_freq = 1333; + break; + case 0x18: + dev_priv->mem_freq = 1600; + break; + default: + DRM_DEBUG_DRIVER("unknown memory frequency 0x%02x\n", + ddrpll & 0xff); + dev_priv->mem_freq = 0; + break; + } + + dev_priv->r_t = dev_priv->mem_freq; + + switch (csipll & 0x3ff) { + case 0x00c: + dev_priv->fsb_freq = 3200; + break; + case 0x00e: + dev_priv->fsb_freq = 3733; + break; + case 0x010: + dev_priv->fsb_freq = 4266; + break; + case 0x012: + dev_priv->fsb_freq = 4800; + break; + case 0x014: + dev_priv->fsb_freq = 5333; + break; + case 0x016: + dev_priv->fsb_freq = 5866; + break; + case 0x018: + dev_priv->fsb_freq = 6400; + break; + default: + DRM_DEBUG_DRIVER("unknown fsb frequency 0x%04x\n", + csipll & 0x3ff); + dev_priv->fsb_freq = 0; + break; + } + + if (dev_priv->fsb_freq == 3200) { + dev_priv->c_m = 0; + } else if (dev_priv->fsb_freq > 3200 && dev_priv->fsb_freq <= 4800) { + dev_priv->c_m = 1; + } else { + dev_priv->c_m = 2; + } +} + +static const struct cparams { + u16 i; + u16 t; + u16 m; + u16 c; +} cparams[] = { + { 1, 1333, 301, 28664 }, + { 1, 1066, 294, 24460 }, + { 1, 800, 294, 25192 }, + { 0, 1333, 276, 27605 }, + { 0, 1066, 276, 27605 }, + { 0, 800, 231, 23784 }, +}; + +unsigned long i915_chipset_val(struct drm_i915_private *dev_priv) +{ + u64 total_count, diff, ret; + u32 count1, count2, count3, m = 0, c = 0; + unsigned long now = jiffies_to_msecs(jiffies), diff1; + int i; + + diff1 = now - dev_priv->last_time1; + + /* Prevent division-by-zero if we are asking too fast. + * Also, we don't get interesting results if we are polling + * faster than once in 10ms, so just return the saved value + * in such cases. + */ + if (diff1 <= 10) + return dev_priv->chipset_power; + + count1 = I915_READ(DMIEC); + count2 = I915_READ(DDREC); + count3 = I915_READ(CSIEC); + + total_count = count1 + count2 + count3; + + /* FIXME: handle per-counter overflow */ + if (total_count < dev_priv->last_count1) { + diff = ~0UL - dev_priv->last_count1; + diff += total_count; + } else { + diff = total_count - dev_priv->last_count1; + } + + for (i = 0; i < ARRAY_SIZE(cparams); i++) { + if (cparams[i].i == dev_priv->c_m && + cparams[i].t == dev_priv->r_t) { + m = cparams[i].m; + c = cparams[i].c; + break; + } + } + + diff = div_u64(diff, diff1); + ret = ((m * diff) + c); + ret = div_u64(ret, 10); + + dev_priv->last_count1 = total_count; + dev_priv->last_time1 = now; + + dev_priv->chipset_power = ret; + + return ret; +} + +unsigned long i915_mch_val(struct drm_i915_private *dev_priv) +{ + unsigned long m, x, b; + u32 tsfs; + + tsfs = I915_READ(TSFS); + + m = ((tsfs & TSFS_SLOPE_MASK) >> TSFS_SLOPE_SHIFT); + x = I915_READ8(TR1); + + b = tsfs & TSFS_INTR_MASK; + + return ((m * x) / 127) - b; +} + +static u16 pvid_to_extvid(struct drm_i915_private *dev_priv, u8 pxvid) +{ + static const struct v_table { + u16 vd; /* in .1 mil */ + u16 vm; /* in .1 mil */ + } v_table[] = { + { 0, 0, }, + { 375, 0, }, + { 500, 0, }, + { 625, 0, }, + { 750, 0, }, + { 875, 0, }, + { 1000, 0, }, + { 1125, 0, }, + { 4125, 3000, }, + { 4125, 3000, }, + { 4125, 3000, }, + { 4125, 3000, }, + { 4125, 3000, }, + { 4125, 3000, }, + { 4125, 3000, }, + { 4125, 3000, }, + { 4125, 3000, }, + { 4125, 3000, }, + { 4125, 3000, }, + { 4125, 3000, }, + { 4125, 3000, }, + { 4125, 3000, }, + { 4125, 3000, }, + { 4125, 3000, }, + { 4125, 3000, }, + { 4125, 3000, }, + { 4125, 3000, }, + { 4125, 3000, }, + { 4125, 3000, }, + { 4125, 3000, }, + { 4125, 3000, }, + { 4125, 3000, }, + { 4250, 3125, }, + { 4375, 3250, }, + { 4500, 3375, }, + { 4625, 3500, }, + { 4750, 3625, }, + { 4875, 3750, }, + { 5000, 3875, }, + { 5125, 4000, }, + { 5250, 4125, }, + { 5375, 4250, }, + { 5500, 4375, }, + { 5625, 4500, }, + { 5750, 4625, }, + { 5875, 4750, }, + { 6000, 4875, }, + { 6125, 5000, }, + { 6250, 5125, }, + { 6375, 5250, }, + { 6500, 5375, }, + { 6625, 5500, }, + { 6750, 5625, }, + { 6875, 5750, }, + { 7000, 5875, }, + { 7125, 6000, }, + { 7250, 6125, }, + { 7375, 6250, }, + { 7500, 6375, }, + { 7625, 6500, }, + { 7750, 6625, }, + { 7875, 6750, }, + { 8000, 6875, }, + { 8125, 7000, }, + { 8250, 7125, }, + { 8375, 7250, }, + { 8500, 7375, }, + { 8625, 7500, }, + { 8750, 7625, }, + { 8875, 7750, }, + { 9000, 7875, }, + { 9125, 8000, }, + { 9250, 8125, }, + { 9375, 8250, }, + { 9500, 8375, }, + { 9625, 8500, }, + { 9750, 8625, }, + { 9875, 8750, }, + { 10000, 8875, }, + { 10125, 9000, }, + { 10250, 9125, }, + { 10375, 9250, }, + { 10500, 9375, }, + { 10625, 9500, }, + { 10750, 9625, }, + { 10875, 9750, }, + { 11000, 9875, }, + { 11125, 10000, }, + { 11250, 10125, }, + { 11375, 10250, }, + { 11500, 10375, }, + { 11625, 10500, }, + { 11750, 10625, }, + { 11875, 10750, }, + { 12000, 10875, }, + { 12125, 11000, }, + { 12250, 11125, }, + { 12375, 11250, }, + { 12500, 11375, }, + { 12625, 11500, }, + { 12750, 11625, }, + { 12875, 11750, }, + { 13000, 11875, }, + { 13125, 12000, }, + { 13250, 12125, }, + { 13375, 12250, }, + { 13500, 12375, }, + { 13625, 12500, }, + { 13750, 12625, }, + { 13875, 12750, }, + { 14000, 12875, }, + { 14125, 13000, }, + { 14250, 13125, }, + { 14375, 13250, }, + { 14500, 13375, }, + { 14625, 13500, }, + { 14750, 13625, }, + { 14875, 13750, }, + { 15000, 13875, }, + { 15125, 14000, }, + { 15250, 14125, }, + { 15375, 14250, }, + { 15500, 14375, }, + { 15625, 14500, }, + { 15750, 14625, }, + { 15875, 14750, }, + { 16000, 14875, }, + { 16125, 15000, }, + }; + if (dev_priv->info->is_mobile) + return v_table[pxvid].vm; + else + return v_table[pxvid].vd; +} + +void i915_update_gfx_val(struct drm_i915_private *dev_priv) +{ + struct timespec now, diff1; + u64 diff; + unsigned long diffms; + u32 count; + + getrawmonotonic(&now); + diff1 = timespec_sub(now, dev_priv->last_time2); + + /* Don't divide by 0 */ + diffms = diff1.tv_sec * 1000 + diff1.tv_nsec / 1000000; + if (!diffms) + return; + + count = I915_READ(GFXEC); + + if (count < dev_priv->last_count2) { + diff = ~0UL - dev_priv->last_count2; + diff += count; + } else { + diff = count - dev_priv->last_count2; + } + + dev_priv->last_count2 = count; + dev_priv->last_time2 = now; + + /* More magic constants... */ + diff = diff * 1181; + diff = div_u64(diff, diffms * 10); + dev_priv->gfx_power = diff; +} + +unsigned long i915_gfx_val(struct drm_i915_private *dev_priv) +{ + unsigned long t, corr, state1, corr2, state2; + u32 pxvid, ext_v; + + pxvid = I915_READ(PXVFREQ_BASE + (dev_priv->cur_delay * 4)); + pxvid = (pxvid >> 24) & 0x7f; + ext_v = pvid_to_extvid(dev_priv, pxvid); + + state1 = ext_v; + + t = i915_mch_val(dev_priv); + + /* Revel in the empirically derived constants */ + + /* Correction factor in 1/100000 units */ + if (t > 80) + corr = ((t * 2349) + 135940); + else if (t >= 50) + corr = ((t * 964) + 29317); + else /* < 50 */ + corr = ((t * 301) + 1004); + + corr = corr * ((150142 * state1) / 10000 - 78642); + corr /= 100000; + corr2 = (corr * dev_priv->corr); + + state2 = (corr2 * state1) / 10000; + state2 /= 100; /* convert to mW */ + + i915_update_gfx_val(dev_priv); + + return dev_priv->gfx_power + state2; +} + +/* Global for IPS driver to get at the current i915 device */ +static struct drm_i915_private *i915_mch_dev; +/* + * Lock protecting IPS related data structures + * - i915_mch_dev + * - dev_priv->max_delay + * - dev_priv->min_delay + * - dev_priv->fmax + * - dev_priv->gpu_busy + */ +static DEFINE_SPINLOCK(mchdev_lock); + +/** + * i915_read_mch_val - return value for IPS use + * + * Calculate and return a value for the IPS driver to use when deciding whether + * we have thermal and power headroom to increase CPU or GPU power budget. + */ +unsigned long i915_read_mch_val(void) +{ + struct drm_i915_private *dev_priv; + unsigned long chipset_val, graphics_val, ret = 0; + + spin_lock(&mchdev_lock); + if (!i915_mch_dev) + goto out_unlock; + dev_priv = i915_mch_dev; + + chipset_val = i915_chipset_val(dev_priv); + graphics_val = i915_gfx_val(dev_priv); + + ret = chipset_val + graphics_val; + +out_unlock: + spin_unlock(&mchdev_lock); + + return ret; +} +EXPORT_SYMBOL_GPL(i915_read_mch_val); + +/** + * i915_gpu_raise - raise GPU frequency limit + * + * Raise the limit; IPS indicates we have thermal headroom. + */ +bool i915_gpu_raise(void) +{ + struct drm_i915_private *dev_priv; + bool ret = true; + + spin_lock(&mchdev_lock); + if (!i915_mch_dev) { + ret = false; + goto out_unlock; + } + dev_priv = i915_mch_dev; + + if (dev_priv->max_delay > dev_priv->fmax) + dev_priv->max_delay--; + +out_unlock: + spin_unlock(&mchdev_lock); + + return ret; +} +EXPORT_SYMBOL_GPL(i915_gpu_raise); + +/** + * i915_gpu_lower - lower GPU frequency limit + * + * IPS indicates we're close to a thermal limit, so throttle back the GPU + * frequency maximum. + */ +bool i915_gpu_lower(void) +{ + struct drm_i915_private *dev_priv; + bool ret = true; + + spin_lock(&mchdev_lock); + if (!i915_mch_dev) { + ret = false; + goto out_unlock; + } + dev_priv = i915_mch_dev; + + if (dev_priv->max_delay < dev_priv->min_delay) + dev_priv->max_delay++; + +out_unlock: + spin_unlock(&mchdev_lock); + + return ret; +} +EXPORT_SYMBOL_GPL(i915_gpu_lower); + +/** + * i915_gpu_busy - indicate GPU business to IPS + * + * Tell the IPS driver whether or not the GPU is busy. + */ +bool i915_gpu_busy(void) +{ + struct drm_i915_private *dev_priv; + bool ret = false; + + spin_lock(&mchdev_lock); + if (!i915_mch_dev) + goto out_unlock; + dev_priv = i915_mch_dev; + + ret = dev_priv->busy; + +out_unlock: + spin_unlock(&mchdev_lock); + + return ret; +} +EXPORT_SYMBOL_GPL(i915_gpu_busy); + +/** + * i915_gpu_turbo_disable - disable graphics turbo + * + * Disable graphics turbo by resetting the max frequency and setting the + * current frequency to the default. + */ +bool i915_gpu_turbo_disable(void) +{ + struct drm_i915_private *dev_priv; + bool ret = true; + + spin_lock(&mchdev_lock); + if (!i915_mch_dev) { + ret = false; + goto out_unlock; + } + dev_priv = i915_mch_dev; + + dev_priv->max_delay = dev_priv->fstart; + + if (!ironlake_set_drps(dev_priv->dev, dev_priv->fstart)) + ret = false; + +out_unlock: + spin_unlock(&mchdev_lock); + + return ret; +} +EXPORT_SYMBOL_GPL(i915_gpu_turbo_disable); + +/** + * Tells the intel_ips driver that the i915 driver is now loaded, if + * IPS got loaded first. + * + * This awkward dance is so that neither module has to depend on the + * other in order for IPS to do the appropriate communication of + * GPU turbo limits to i915. + */ +static void +ips_ping_for_i915_load(void) +{ + void (*link)(void); + + link = symbol_get(ips_link_to_i915_driver); + if (link) { + link(); + symbol_put(ips_link_to_i915_driver); + } +} + +/** + * i915_driver_load - setup chip and create an initial config + * @dev: DRM device + * @flags: startup flags + * + * The driver load routine has to do several things: + * - drive output discovery via intel_modeset_init() + * - initialize the memory manager + * - allocate initial config memory + * - setup the DRM framebuffer with the allocated memory + */ +int i915_driver_load(struct drm_device *dev, unsigned long flags) +{ + struct drm_i915_private *dev_priv; + int ret = 0, mmio_bar; + uint32_t agp_size; + + /* i915 has 4 more counters */ + dev->counters += 4; + dev->types[6] = _DRM_STAT_IRQ; + dev->types[7] = _DRM_STAT_PRIMARY; + dev->types[8] = _DRM_STAT_SECONDARY; + dev->types[9] = _DRM_STAT_DMA; + + dev_priv = kzalloc(sizeof(drm_i915_private_t), GFP_KERNEL); + if (dev_priv == NULL) + return -ENOMEM; + + dev->dev_private = (void *)dev_priv; + dev_priv->dev = dev; + dev_priv->info = (struct intel_device_info *) flags; + + if (i915_get_bridge_dev(dev)) { + ret = -EIO; + goto free_priv; + } + + /* overlay on gen2 is broken and can't address above 1G */ + if (IS_GEN2(dev)) + dma_set_coherent_mask(&dev->pdev->dev, DMA_BIT_MASK(30)); + + /* 965GM sometimes incorrectly writes to hardware status page (HWS) + * using 32bit addressing, overwriting memory if HWS is located + * above 4GB. + * + * The documentation also mentions an issue with undefined + * behaviour if any general state is accessed within a page above 4GB, + * which also needs to be handled carefully. + */ + if (IS_BROADWATER(dev) || IS_CRESTLINE(dev)) + dma_set_coherent_mask(&dev->pdev->dev, DMA_BIT_MASK(32)); + + mmio_bar = IS_GEN2(dev) ? 1 : 0; + dev_priv->regs = pci_iomap(dev->pdev, mmio_bar, 0); + if (!dev_priv->regs) { + DRM_ERROR("failed to map registers\n"); + ret = -EIO; + goto put_bridge; + } + + dev_priv->mm.gtt = intel_gtt_get(); + if (!dev_priv->mm.gtt) { + DRM_ERROR("Failed to initialize GTT\n"); + ret = -ENODEV; + goto out_rmmap; + } + + agp_size = dev_priv->mm.gtt->gtt_mappable_entries << PAGE_SHIFT; + + dev_priv->mm.gtt_mapping = + io_mapping_create_wc(dev->agp->base, agp_size); + if (dev_priv->mm.gtt_mapping == NULL) { + ret = -EIO; + goto out_rmmap; + } + + /* Set up a WC MTRR for non-PAT systems. This is more common than + * one would think, because the kernel disables PAT on first + * generation Core chips because WC PAT gets overridden by a UC + * MTRR if present. Even if a UC MTRR isn't present. + */ + dev_priv->mm.gtt_mtrr = mtrr_add(dev->agp->base, + agp_size, + MTRR_TYPE_WRCOMB, 1); + if (dev_priv->mm.gtt_mtrr < 0) { + DRM_INFO("MTRR allocation failed. Graphics " + "performance may suffer.\n"); + } + + /* The i915 workqueue is primarily used for batched retirement of + * requests (and thus managing bo) once the task has been completed + * by the GPU. i915_gem_retire_requests() is called directly when we + * need high-priority retirement, such as waiting for an explicit + * bo. + * + * It is also used for periodic low-priority events, such as + * idle-timers and recording error state. + * + * All tasks on the workqueue are expected to acquire the dev mutex + * so there is no point in running more than one instance of the + * workqueue at any time: max_active = 1 and NON_REENTRANT. + */ + dev_priv->wq = alloc_workqueue("i915", + WQ_UNBOUND | WQ_NON_REENTRANT, + 1); + if (dev_priv->wq == NULL) { + DRM_ERROR("Failed to create our workqueue.\n"); + ret = -ENOMEM; + goto out_mtrrfree; + } + + /* enable GEM by default */ + dev_priv->has_gem = 1; + + intel_irq_init(dev); + + /* Try to make sure MCHBAR is enabled before poking at it */ + intel_setup_mchbar(dev); + intel_setup_gmbus(dev); + intel_opregion_setup(dev); + + /* Make sure the bios did its job and set up vital registers */ + intel_setup_bios(dev); + + i915_gem_load(dev); + + /* Init HWS */ + if (!I915_NEED_GFX_HWS(dev)) { + ret = i915_init_phys_hws(dev); + if (ret) + goto out_gem_unload; + } + + if (IS_PINEVIEW(dev)) + i915_pineview_get_mem_freq(dev); + else if (IS_GEN5(dev)) + i915_ironlake_get_mem_freq(dev); + + /* On the 945G/GM, the chipset reports the MSI capability on the + * integrated graphics even though the support isn't actually there + * according to the published specs. It doesn't appear to function + * correctly in testing on 945G. + * This may be a side effect of MSI having been made available for PEG + * and the registers being closely associated. + * + * According to chipset errata, on the 965GM, MSI interrupts may + * be lost or delayed, but we use them anyways to avoid + * stuck interrupts on some machines. + */ + if (!IS_I945G(dev) && !IS_I945GM(dev)) + pci_enable_msi(dev->pdev); + + spin_lock_init(&dev_priv->irq_lock); + spin_lock_init(&dev_priv->error_lock); + spin_lock_init(&dev_priv->rps_lock); + + if (IS_MOBILE(dev) || !IS_GEN2(dev)) + dev_priv->num_pipe = 2; + else + dev_priv->num_pipe = 1; + + ret = drm_vblank_init(dev, dev_priv->num_pipe); + if (ret) + goto out_gem_unload; + + /* Start out suspended */ + dev_priv->mm.suspended = 1; + + intel_detect_pch(dev); + + if (drm_core_check_feature(dev, DRIVER_MODESET)) { + ret = i915_load_modeset_init(dev); + if (ret < 0) { + DRM_ERROR("failed to init modeset\n"); + goto out_gem_unload; + } + } + + /* Must be done after probing outputs */ + intel_opregion_init(dev); + acpi_video_register(); + + setup_timer(&dev_priv->hangcheck_timer, i915_hangcheck_elapsed, + (unsigned long) dev); + + spin_lock(&mchdev_lock); + i915_mch_dev = dev_priv; + dev_priv->mchdev_lock = &mchdev_lock; + spin_unlock(&mchdev_lock); + + ips_ping_for_i915_load(); + + return 0; + +out_gem_unload: + if (dev_priv->mm.inactive_shrinker.shrink) + unregister_shrinker(&dev_priv->mm.inactive_shrinker); + + if (dev->pdev->msi_enabled) + pci_disable_msi(dev->pdev); + + intel_teardown_gmbus(dev); + intel_teardown_mchbar(dev); + destroy_workqueue(dev_priv->wq); +out_mtrrfree: + if (dev_priv->mm.gtt_mtrr >= 0) { + mtrr_del(dev_priv->mm.gtt_mtrr, dev->agp->base, + dev->agp->agp_info.aper_size * 1024 * 1024); + dev_priv->mm.gtt_mtrr = -1; + } + io_mapping_free(dev_priv->mm.gtt_mapping); +out_rmmap: + pci_iounmap(dev->pdev, dev_priv->regs); +put_bridge: + pci_dev_put(dev_priv->bridge_dev); +free_priv: + kfree(dev_priv); + return ret; +} + +int i915_driver_unload(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + int ret; + + spin_lock(&mchdev_lock); + i915_mch_dev = NULL; + spin_unlock(&mchdev_lock); + + if (dev_priv->mm.inactive_shrinker.shrink) + unregister_shrinker(&dev_priv->mm.inactive_shrinker); + + mutex_lock(&dev->struct_mutex); + ret = i915_gpu_idle(dev); + if (ret) + DRM_ERROR("failed to idle hardware: %d\n", ret); + mutex_unlock(&dev->struct_mutex); + + /* Cancel the retire work handler, which should be idle now. */ + cancel_delayed_work_sync(&dev_priv->mm.retire_work); + + io_mapping_free(dev_priv->mm.gtt_mapping); + if (dev_priv->mm.gtt_mtrr >= 0) { + mtrr_del(dev_priv->mm.gtt_mtrr, dev->agp->base, + dev->agp->agp_info.aper_size * 1024 * 1024); + dev_priv->mm.gtt_mtrr = -1; + } + + acpi_video_unregister(); + + if (drm_core_check_feature(dev, DRIVER_MODESET)) { + intel_fbdev_fini(dev); + intel_modeset_cleanup(dev); + + /* + * free the memory space allocated for the child device + * config parsed from VBT + */ + if (dev_priv->child_dev && dev_priv->child_dev_num) { + kfree(dev_priv->child_dev); + dev_priv->child_dev = NULL; + dev_priv->child_dev_num = 0; + } + + vga_switcheroo_unregister_client(dev->pdev); + vga_client_register(dev->pdev, NULL, NULL, NULL); + } + + /* Free error state after interrupts are fully disabled. */ + del_timer_sync(&dev_priv->hangcheck_timer); + cancel_work_sync(&dev_priv->error_work); + i915_destroy_error_state(dev); + + if (dev->pdev->msi_enabled) + pci_disable_msi(dev->pdev); + + intel_opregion_fini(dev); + + if (drm_core_check_feature(dev, DRIVER_MODESET)) { + /* Flush any outstanding unpin_work. */ + flush_workqueue(dev_priv->wq); + + mutex_lock(&dev->struct_mutex); + i915_gem_free_all_phys_object(dev); + i915_gem_cleanup_ringbuffer(dev); + mutex_unlock(&dev->struct_mutex); + if (I915_HAS_FBC(dev) && i915_powersave) + i915_cleanup_compression(dev); + drm_mm_takedown(&dev_priv->mm.stolen); + + intel_cleanup_overlay(dev); + + if (!I915_NEED_GFX_HWS(dev)) + i915_free_hws(dev); + } + + if (dev_priv->regs != NULL) + pci_iounmap(dev->pdev, dev_priv->regs); + + intel_teardown_gmbus(dev); + intel_teardown_mchbar(dev); + + destroy_workqueue(dev_priv->wq); + + pci_dev_put(dev_priv->bridge_dev); + kfree(dev->dev_private); + + return 0; +} + +int i915_driver_open(struct drm_device *dev, struct drm_file *file) +{ + struct drm_i915_file_private *file_priv; + + DRM_DEBUG_DRIVER("\n"); + file_priv = kmalloc(sizeof(*file_priv), GFP_KERNEL); + if (!file_priv) + return -ENOMEM; + + file->driver_priv = file_priv; + + spin_lock_init(&file_priv->mm.lock); + INIT_LIST_HEAD(&file_priv->mm.request_list); + + return 0; +} + +/** + * i915_driver_lastclose - clean up after all DRM clients have exited + * @dev: DRM device + * + * Take care of cleaning up after all DRM clients have exited. In the + * mode setting case, we want to restore the kernel's initial mode (just + * in case the last client left us in a bad state). + * + * Additionally, in the non-mode setting case, we'll tear down the AGP + * and DMA structures, since the kernel won't be using them, and clea + * up any GEM state. + */ +void i915_driver_lastclose(struct drm_device * dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + + if (!dev_priv || drm_core_check_feature(dev, DRIVER_MODESET)) { + intel_fb_restore_mode(dev); + vga_switcheroo_process_delayed_switch(); + return; + } + + i915_gem_lastclose(dev); + + if (dev_priv->agp_heap) + i915_mem_takedown(&(dev_priv->agp_heap)); + + i915_dma_cleanup(dev); +} + +void i915_driver_preclose(struct drm_device * dev, struct drm_file *file_priv) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + i915_gem_release(dev, file_priv); + if (!drm_core_check_feature(dev, DRIVER_MODESET)) + i915_mem_release(dev, file_priv, dev_priv->agp_heap); +} + +void i915_driver_postclose(struct drm_device *dev, struct drm_file *file) +{ + struct drm_i915_file_private *file_priv = file->driver_priv; + + kfree(file_priv); +} + +struct drm_ioctl_desc i915_ioctls[] = { + DRM_IOCTL_DEF_DRV(I915_INIT, i915_dma_init, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF_DRV(I915_FLUSH, i915_flush_ioctl, DRM_AUTH), + DRM_IOCTL_DEF_DRV(I915_FLIP, i915_flip_bufs, DRM_AUTH), + DRM_IOCTL_DEF_DRV(I915_BATCHBUFFER, i915_batchbuffer, DRM_AUTH), + DRM_IOCTL_DEF_DRV(I915_IRQ_EMIT, i915_irq_emit, DRM_AUTH), + DRM_IOCTL_DEF_DRV(I915_IRQ_WAIT, i915_irq_wait, DRM_AUTH), + DRM_IOCTL_DEF_DRV(I915_GETPARAM, i915_getparam, DRM_AUTH), + DRM_IOCTL_DEF_DRV(I915_SETPARAM, i915_setparam, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF_DRV(I915_ALLOC, i915_mem_alloc, DRM_AUTH), + DRM_IOCTL_DEF_DRV(I915_FREE, i915_mem_free, DRM_AUTH), + DRM_IOCTL_DEF_DRV(I915_INIT_HEAP, i915_mem_init_heap, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF_DRV(I915_CMDBUFFER, i915_cmdbuffer, DRM_AUTH), + DRM_IOCTL_DEF_DRV(I915_DESTROY_HEAP, i915_mem_destroy_heap, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF_DRV(I915_SET_VBLANK_PIPE, i915_vblank_pipe_set, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF_DRV(I915_GET_VBLANK_PIPE, i915_vblank_pipe_get, DRM_AUTH), + DRM_IOCTL_DEF_DRV(I915_VBLANK_SWAP, i915_vblank_swap, DRM_AUTH), + DRM_IOCTL_DEF_DRV(I915_HWS_ADDR, i915_set_status_page, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF_DRV(I915_GEM_INIT, i915_gem_init_ioctl, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I915_GEM_EXECBUFFER, i915_gem_execbuffer, DRM_AUTH|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I915_GEM_EXECBUFFER2, i915_gem_execbuffer2, DRM_AUTH|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I915_GEM_PIN, i915_gem_pin_ioctl, DRM_AUTH|DRM_ROOT_ONLY|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I915_GEM_UNPIN, i915_gem_unpin_ioctl, DRM_AUTH|DRM_ROOT_ONLY|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I915_GEM_BUSY, i915_gem_busy_ioctl, DRM_AUTH|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I915_GEM_THROTTLE, i915_gem_throttle_ioctl, DRM_AUTH|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I915_GEM_ENTERVT, i915_gem_entervt_ioctl, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I915_GEM_LEAVEVT, i915_gem_leavevt_ioctl, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I915_GEM_CREATE, i915_gem_create_ioctl, DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I915_GEM_PREAD, i915_gem_pread_ioctl, DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I915_GEM_PWRITE, i915_gem_pwrite_ioctl, DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I915_GEM_MMAP, i915_gem_mmap_ioctl, DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I915_GEM_MMAP_GTT, i915_gem_mmap_gtt_ioctl, DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I915_GEM_SET_DOMAIN, i915_gem_set_domain_ioctl, DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I915_GEM_SW_FINISH, i915_gem_sw_finish_ioctl, DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I915_GEM_SET_TILING, i915_gem_set_tiling, DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I915_GEM_GET_TILING, i915_gem_get_tiling, DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I915_GEM_GET_APERTURE, i915_gem_get_aperture_ioctl, DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I915_GET_PIPE_FROM_CRTC_ID, intel_get_pipe_from_crtc_id, DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I915_GEM_MADVISE, i915_gem_madvise_ioctl, DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I915_OVERLAY_PUT_IMAGE, intel_overlay_put_image, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(I915_OVERLAY_ATTRS, intel_overlay_attrs, DRM_MASTER|DRM_CONTROL_ALLOW|DRM_UNLOCKED), +}; + +int i915_max_ioctl = DRM_ARRAY_SIZE(i915_ioctls); + +/** + * Determine if the device really is AGP or not. + * + * All Intel graphics chipsets are treated as AGP, even if they are really + * PCI-e. + * + * \param dev The device to be tested. + * + * \returns + * A value of 1 is always retured to indictate every i9x5 is AGP. + */ +int i915_driver_device_is_agp(struct drm_device * dev) +{ + return 1; +} diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c new file mode 100644 index 00000000..111686ad --- /dev/null +++ b/drivers/gpu/drm/i915/i915_drv.c @@ -0,0 +1,869 @@ +/* i915_drv.c -- i830,i845,i855,i865,i915 driver -*- linux-c -*- + */ +/* + * + * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include +#include "drmP.h" +#include "drm.h" +#include "i915_drm.h" +#include "i915_drv.h" +#include "intel_drv.h" + +#include +#include "drm_crtc_helper.h" + +static int i915_modeset = -1; +module_param_named(modeset, i915_modeset, int, 0400); + +unsigned int i915_fbpercrtc = 0; +module_param_named(fbpercrtc, i915_fbpercrtc, int, 0400); + +int i915_panel_ignore_lid = 0; +module_param_named(panel_ignore_lid, i915_panel_ignore_lid, int, 0600); + +unsigned int i915_powersave = 1; +module_param_named(powersave, i915_powersave, int, 0600); + +unsigned int i915_semaphores = 0; +module_param_named(semaphores, i915_semaphores, int, 0600); + +unsigned int i915_enable_rc6 = 0; +module_param_named(i915_enable_rc6, i915_enable_rc6, int, 0600); + +unsigned int i915_enable_fbc = 0; +module_param_named(i915_enable_fbc, i915_enable_fbc, int, 0600); + +unsigned int i915_lvds_downclock = 0; +module_param_named(lvds_downclock, i915_lvds_downclock, int, 0400); + +unsigned int i915_panel_use_ssc = 1; +module_param_named(lvds_use_ssc, i915_panel_use_ssc, int, 0600); + +int i915_vbt_sdvo_panel_type = -1; +module_param_named(vbt_sdvo_panel_type, i915_vbt_sdvo_panel_type, int, 0600); + +static bool i915_try_reset = true; +module_param_named(reset, i915_try_reset, bool, 0600); + +static struct drm_driver driver; +extern int intel_agp_enabled; + +#define INTEL_VGA_DEVICE(id, info) { \ + .class = PCI_CLASS_DISPLAY_VGA << 8, \ + .class_mask = 0xff0000, \ + .vendor = 0x8086, \ + .device = id, \ + .subvendor = PCI_ANY_ID, \ + .subdevice = PCI_ANY_ID, \ + .driver_data = (unsigned long) info } + +static const struct intel_device_info intel_i830_info = { + .gen = 2, .is_mobile = 1, .cursor_needs_physical = 1, + .has_overlay = 1, .overlay_needs_physical = 1, +}; + +static const struct intel_device_info intel_845g_info = { + .gen = 2, + .has_overlay = 1, .overlay_needs_physical = 1, +}; + +static const struct intel_device_info intel_i85x_info = { + .gen = 2, .is_i85x = 1, .is_mobile = 1, + .cursor_needs_physical = 1, + .has_overlay = 1, .overlay_needs_physical = 1, +}; + +static const struct intel_device_info intel_i865g_info = { + .gen = 2, + .has_overlay = 1, .overlay_needs_physical = 1, +}; + +static const struct intel_device_info intel_i915g_info = { + .gen = 3, .is_i915g = 1, .cursor_needs_physical = 1, + .has_overlay = 1, .overlay_needs_physical = 1, +}; +static const struct intel_device_info intel_i915gm_info = { + .gen = 3, .is_mobile = 1, + .cursor_needs_physical = 1, + .has_overlay = 1, .overlay_needs_physical = 1, + .supports_tv = 1, +}; +static const struct intel_device_info intel_i945g_info = { + .gen = 3, .has_hotplug = 1, .cursor_needs_physical = 1, + .has_overlay = 1, .overlay_needs_physical = 1, +}; +static const struct intel_device_info intel_i945gm_info = { + .gen = 3, .is_i945gm = 1, .is_mobile = 1, + .has_hotplug = 1, .cursor_needs_physical = 1, + .has_overlay = 1, .overlay_needs_physical = 1, + .supports_tv = 1, +}; + +static const struct intel_device_info intel_i965g_info = { + .gen = 4, .is_broadwater = 1, + .has_hotplug = 1, + .has_overlay = 1, +}; + +static const struct intel_device_info intel_i965gm_info = { + .gen = 4, .is_crestline = 1, + .is_mobile = 1, .has_fbc = 1, .has_hotplug = 1, + .has_overlay = 1, + .supports_tv = 1, +}; + +static const struct intel_device_info intel_g33_info = { + .gen = 3, .is_g33 = 1, + .need_gfx_hws = 1, .has_hotplug = 1, + .has_overlay = 1, +}; + +static const struct intel_device_info intel_g45_info = { + .gen = 4, .is_g4x = 1, .need_gfx_hws = 1, + .has_pipe_cxsr = 1, .has_hotplug = 1, + .has_bsd_ring = 1, +}; + +static const struct intel_device_info intel_gm45_info = { + .gen = 4, .is_g4x = 1, + .is_mobile = 1, .need_gfx_hws = 1, .has_fbc = 1, + .has_pipe_cxsr = 1, .has_hotplug = 1, + .supports_tv = 1, + .has_bsd_ring = 1, +}; + +static const struct intel_device_info intel_pineview_info = { + .gen = 3, .is_g33 = 1, .is_pineview = 1, .is_mobile = 1, + .need_gfx_hws = 1, .has_hotplug = 1, + .has_overlay = 1, +}; + +static const struct intel_device_info intel_ironlake_d_info = { + .gen = 5, + .need_gfx_hws = 1, .has_pipe_cxsr = 1, .has_hotplug = 1, + .has_bsd_ring = 1, +}; + +static const struct intel_device_info intel_ironlake_m_info = { + .gen = 5, .is_mobile = 1, + .need_gfx_hws = 1, .has_hotplug = 1, + .has_fbc = 1, + .has_bsd_ring = 1, +}; + +static const struct intel_device_info intel_sandybridge_d_info = { + .gen = 6, + .need_gfx_hws = 1, .has_hotplug = 1, + .has_bsd_ring = 1, + .has_blt_ring = 1, +}; + +static const struct intel_device_info intel_sandybridge_m_info = { + .gen = 6, .is_mobile = 1, + .need_gfx_hws = 1, .has_hotplug = 1, + .has_fbc = 1, + .has_bsd_ring = 1, + .has_blt_ring = 1, +}; + +static const struct intel_device_info intel_ivybridge_d_info = { + .is_ivybridge = 1, .gen = 7, + .need_gfx_hws = 1, .has_hotplug = 1, + .has_bsd_ring = 1, + .has_blt_ring = 1, +}; + +static const struct intel_device_info intel_ivybridge_m_info = { + .is_ivybridge = 1, .gen = 7, .is_mobile = 1, + .need_gfx_hws = 1, .has_hotplug = 1, + .has_fbc = 0, /* FBC is not enabled on Ivybridge mobile yet */ + .has_bsd_ring = 1, + .has_blt_ring = 1, +}; + +static const struct pci_device_id pciidlist[] = { /* aka */ + INTEL_VGA_DEVICE(0x3577, &intel_i830_info), /* I830_M */ + INTEL_VGA_DEVICE(0x2562, &intel_845g_info), /* 845_G */ + INTEL_VGA_DEVICE(0x3582, &intel_i85x_info), /* I855_GM */ + INTEL_VGA_DEVICE(0x358e, &intel_i85x_info), + INTEL_VGA_DEVICE(0x2572, &intel_i865g_info), /* I865_G */ + INTEL_VGA_DEVICE(0x2582, &intel_i915g_info), /* I915_G */ + INTEL_VGA_DEVICE(0x258a, &intel_i915g_info), /* E7221_G */ + INTEL_VGA_DEVICE(0x2592, &intel_i915gm_info), /* I915_GM */ + INTEL_VGA_DEVICE(0x2772, &intel_i945g_info), /* I945_G */ + INTEL_VGA_DEVICE(0x27a2, &intel_i945gm_info), /* I945_GM */ + INTEL_VGA_DEVICE(0x27ae, &intel_i945gm_info), /* I945_GME */ + INTEL_VGA_DEVICE(0x2972, &intel_i965g_info), /* I946_GZ */ + INTEL_VGA_DEVICE(0x2982, &intel_i965g_info), /* G35_G */ + INTEL_VGA_DEVICE(0x2992, &intel_i965g_info), /* I965_Q */ + INTEL_VGA_DEVICE(0x29a2, &intel_i965g_info), /* I965_G */ + INTEL_VGA_DEVICE(0x29b2, &intel_g33_info), /* Q35_G */ + INTEL_VGA_DEVICE(0x29c2, &intel_g33_info), /* G33_G */ + INTEL_VGA_DEVICE(0x29d2, &intel_g33_info), /* Q33_G */ + INTEL_VGA_DEVICE(0x2a02, &intel_i965gm_info), /* I965_GM */ + INTEL_VGA_DEVICE(0x2a12, &intel_i965gm_info), /* I965_GME */ + INTEL_VGA_DEVICE(0x2a42, &intel_gm45_info), /* GM45_G */ + INTEL_VGA_DEVICE(0x2e02, &intel_g45_info), /* IGD_E_G */ + INTEL_VGA_DEVICE(0x2e12, &intel_g45_info), /* Q45_G */ + INTEL_VGA_DEVICE(0x2e22, &intel_g45_info), /* G45_G */ + INTEL_VGA_DEVICE(0x2e32, &intel_g45_info), /* G41_G */ + INTEL_VGA_DEVICE(0x2e42, &intel_g45_info), /* B43_G */ + INTEL_VGA_DEVICE(0x2e92, &intel_g45_info), /* B43_G.1 */ + INTEL_VGA_DEVICE(0xa001, &intel_pineview_info), + INTEL_VGA_DEVICE(0xa011, &intel_pineview_info), + INTEL_VGA_DEVICE(0x0042, &intel_ironlake_d_info), + INTEL_VGA_DEVICE(0x0046, &intel_ironlake_m_info), + INTEL_VGA_DEVICE(0x0102, &intel_sandybridge_d_info), + INTEL_VGA_DEVICE(0x0112, &intel_sandybridge_d_info), + INTEL_VGA_DEVICE(0x0122, &intel_sandybridge_d_info), + INTEL_VGA_DEVICE(0x0106, &intel_sandybridge_m_info), + INTEL_VGA_DEVICE(0x0116, &intel_sandybridge_m_info), + INTEL_VGA_DEVICE(0x0126, &intel_sandybridge_m_info), + INTEL_VGA_DEVICE(0x010A, &intel_sandybridge_d_info), + INTEL_VGA_DEVICE(0x0156, &intel_ivybridge_m_info), /* GT1 mobile */ + INTEL_VGA_DEVICE(0x0166, &intel_ivybridge_m_info), /* GT2 mobile */ + INTEL_VGA_DEVICE(0x0152, &intel_ivybridge_d_info), /* GT1 desktop */ + INTEL_VGA_DEVICE(0x0162, &intel_ivybridge_d_info), /* GT2 desktop */ + INTEL_VGA_DEVICE(0x015a, &intel_ivybridge_d_info), /* GT1 server */ + {0, 0, 0} +}; + +#if defined(CONFIG_DRM_I915_KMS) +MODULE_DEVICE_TABLE(pci, pciidlist); +#endif + +#define INTEL_PCH_DEVICE_ID_MASK 0xff00 +#define INTEL_PCH_IBX_DEVICE_ID_TYPE 0x3b00 +#define INTEL_PCH_CPT_DEVICE_ID_TYPE 0x1c00 +#define INTEL_PCH_PPT_DEVICE_ID_TYPE 0x1e00 + +void intel_detect_pch (struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct pci_dev *pch; + + /* + * The reason to probe ISA bridge instead of Dev31:Fun0 is to + * make graphics device passthrough work easy for VMM, that only + * need to expose ISA bridge to let driver know the real hardware + * underneath. This is a requirement from virtualization team. + */ + pch = pci_get_class(PCI_CLASS_BRIDGE_ISA << 8, NULL); + if (pch) { + if (pch->vendor == PCI_VENDOR_ID_INTEL) { + int id; + id = pch->device & INTEL_PCH_DEVICE_ID_MASK; + + if (id == INTEL_PCH_IBX_DEVICE_ID_TYPE) { + dev_priv->pch_type = PCH_IBX; + DRM_DEBUG_KMS("Found Ibex Peak PCH\n"); + } else if (id == INTEL_PCH_CPT_DEVICE_ID_TYPE) { + dev_priv->pch_type = PCH_CPT; + DRM_DEBUG_KMS("Found CougarPoint PCH\n"); + } else if (id == INTEL_PCH_PPT_DEVICE_ID_TYPE) { + /* PantherPoint is CPT compatible */ + dev_priv->pch_type = PCH_CPT; + DRM_DEBUG_KMS("Found PatherPoint PCH\n"); + } + } + pci_dev_put(pch); + } +} + +static void __gen6_gt_force_wake_get(struct drm_i915_private *dev_priv) +{ + int count; + + count = 0; + while (count++ < 50 && (I915_READ_NOTRACE(FORCEWAKE_ACK) & 1)) + udelay(10); + + I915_WRITE_NOTRACE(FORCEWAKE, 1); + POSTING_READ(FORCEWAKE); + + count = 0; + while (count++ < 50 && (I915_READ_NOTRACE(FORCEWAKE_ACK) & 1) == 0) + udelay(10); +} + +/* + * Generally this is called implicitly by the register read function. However, + * if some sequence requires the GT to not power down then this function should + * be called at the beginning of the sequence followed by a call to + * gen6_gt_force_wake_put() at the end of the sequence. + */ +void gen6_gt_force_wake_get(struct drm_i915_private *dev_priv) +{ + WARN_ON(!mutex_is_locked(&dev_priv->dev->struct_mutex)); + + /* Forcewake is atomic in case we get in here without the lock */ + if (atomic_add_return(1, &dev_priv->forcewake_count) == 1) + __gen6_gt_force_wake_get(dev_priv); +} + +static void __gen6_gt_force_wake_put(struct drm_i915_private *dev_priv) +{ + I915_WRITE_NOTRACE(FORCEWAKE, 0); + POSTING_READ(FORCEWAKE); +} + +/* + * see gen6_gt_force_wake_get() + */ +void gen6_gt_force_wake_put(struct drm_i915_private *dev_priv) +{ + WARN_ON(!mutex_is_locked(&dev_priv->dev->struct_mutex)); + + if (atomic_dec_and_test(&dev_priv->forcewake_count)) + __gen6_gt_force_wake_put(dev_priv); +} + +void __gen6_gt_wait_for_fifo(struct drm_i915_private *dev_priv) +{ + int loop = 500; + u32 fifo = I915_READ_NOTRACE(GT_FIFO_FREE_ENTRIES); + while (fifo < 20 && loop--) { + udelay(10); + fifo = I915_READ_NOTRACE(GT_FIFO_FREE_ENTRIES); + } +} + +static int i915_drm_freeze(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + drm_kms_helper_poll_disable(dev); + + pci_save_state(dev->pdev); + + /* If KMS is active, we do the leavevt stuff here */ + if (drm_core_check_feature(dev, DRIVER_MODESET)) { + int error = i915_gem_idle(dev); + if (error) { + dev_err(&dev->pdev->dev, + "GEM idle failed, resume might fail\n"); + return error; + } + drm_irq_uninstall(dev); + } + + i915_save_state(dev); + + intel_opregion_fini(dev); + + /* Modeset on resume, not lid events */ + dev_priv->modeset_on_lid = 0; + + console_lock(); + intel_fbdev_set_suspend(dev, 1); + console_unlock(); + + return 0; +} + +int i915_suspend(struct drm_device *dev, pm_message_t state) +{ + int error; + + if (!dev || !dev->dev_private) { + DRM_ERROR("dev: %p\n", dev); + DRM_ERROR("DRM not initialized, aborting suspend.\n"); + return -ENODEV; + } + + if (state.event == PM_EVENT_PRETHAW) + return 0; + + + if (dev->switch_power_state == DRM_SWITCH_POWER_OFF) + return 0; + + error = i915_drm_freeze(dev); + if (error) + return error; + + if (state.event == PM_EVENT_SUSPEND) { + /* Shut down the device */ + pci_disable_device(dev->pdev); + pci_set_power_state(dev->pdev, PCI_D3hot); + } + + return 0; +} + +static int i915_drm_thaw(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + int error = 0; + + if (drm_core_check_feature(dev, DRIVER_MODESET)) { + mutex_lock(&dev->struct_mutex); + i915_gem_restore_gtt_mappings(dev); + mutex_unlock(&dev->struct_mutex); + } + + i915_restore_state(dev); + intel_opregion_setup(dev); + + /* KMS EnterVT equivalent */ + if (drm_core_check_feature(dev, DRIVER_MODESET)) { + mutex_lock(&dev->struct_mutex); + dev_priv->mm.suspended = 0; + + error = i915_gem_init_ringbuffer(dev); + mutex_unlock(&dev->struct_mutex); + + drm_mode_config_reset(dev); + drm_irq_install(dev); + + /* Resume the modeset for every activated CRTC */ + mutex_lock(&dev->mode_config.mutex); + drm_helper_resume_force_mode(dev); + mutex_unlock(&dev->mode_config.mutex); + + if (IS_IRONLAKE_M(dev)) + ironlake_enable_rc6(dev); + } + + intel_opregion_init(dev); + + dev_priv->modeset_on_lid = 0; + + console_lock(); + intel_fbdev_set_suspend(dev, 0); + console_unlock(); + return error; +} + +int i915_resume(struct drm_device *dev) +{ + int ret; + + if (dev->switch_power_state == DRM_SWITCH_POWER_OFF) + return 0; + + if (pci_enable_device(dev->pdev)) + return -EIO; + + pci_set_master(dev->pdev); + + ret = i915_drm_thaw(dev); + if (ret) + return ret; + + drm_kms_helper_poll_enable(dev); + return 0; +} + +static int i8xx_do_reset(struct drm_device *dev, u8 flags) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + if (IS_I85X(dev)) + return -ENODEV; + + I915_WRITE(D_STATE, I915_READ(D_STATE) | DSTATE_GFX_RESET_I830); + POSTING_READ(D_STATE); + + if (IS_I830(dev) || IS_845G(dev)) { + I915_WRITE(DEBUG_RESET_I830, + DEBUG_RESET_DISPLAY | + DEBUG_RESET_RENDER | + DEBUG_RESET_FULL); + POSTING_READ(DEBUG_RESET_I830); + msleep(1); + + I915_WRITE(DEBUG_RESET_I830, 0); + POSTING_READ(DEBUG_RESET_I830); + } + + msleep(1); + + I915_WRITE(D_STATE, I915_READ(D_STATE) & ~DSTATE_GFX_RESET_I830); + POSTING_READ(D_STATE); + + return 0; +} + +static int i965_reset_complete(struct drm_device *dev) +{ + u8 gdrst; + pci_read_config_byte(dev->pdev, I965_GDRST, &gdrst); + return gdrst & 0x1; +} + +static int i965_do_reset(struct drm_device *dev, u8 flags) +{ + u8 gdrst; + + /* + * Set the domains we want to reset (GRDOM/bits 2 and 3) as + * well as the reset bit (GR/bit 0). Setting the GR bit + * triggers the reset; when done, the hardware will clear it. + */ + pci_read_config_byte(dev->pdev, I965_GDRST, &gdrst); + pci_write_config_byte(dev->pdev, I965_GDRST, gdrst | flags | 0x1); + + return wait_for(i965_reset_complete(dev), 500); +} + +static int ironlake_do_reset(struct drm_device *dev, u8 flags) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u32 gdrst = I915_READ(MCHBAR_MIRROR_BASE + ILK_GDSR); + I915_WRITE(MCHBAR_MIRROR_BASE + ILK_GDSR, gdrst | flags | 0x1); + return wait_for(I915_READ(MCHBAR_MIRROR_BASE + ILK_GDSR) & 0x1, 500); +} + +static int gen6_do_reset(struct drm_device *dev, u8 flags) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + I915_WRITE(GEN6_GDRST, GEN6_GRDOM_FULL); + return wait_for((I915_READ(GEN6_GDRST) & GEN6_GRDOM_FULL) == 0, 500); +} + +/** + * i965_reset - reset chip after a hang + * @dev: drm device to reset + * @flags: reset domains + * + * Reset the chip. Useful if a hang is detected. Returns zero on successful + * reset or otherwise an error code. + * + * Procedure is fairly simple: + * - reset the chip using the reset reg + * - re-init context state + * - re-init hardware status page + * - re-init ring buffer + * - re-init interrupt state + * - re-init display + */ +int i915_reset(struct drm_device *dev, u8 flags) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + /* + * We really should only reset the display subsystem if we actually + * need to + */ + bool need_display = true; + int ret; + + if (!i915_try_reset) + return 0; + + if (!mutex_trylock(&dev->struct_mutex)) + return -EBUSY; + + i915_gem_reset(dev); + + ret = -ENODEV; + if (get_seconds() - dev_priv->last_gpu_reset < 5) { + DRM_ERROR("GPU hanging too fast, declaring wedged!\n"); + } else switch (INTEL_INFO(dev)->gen) { + case 7: + case 6: + ret = gen6_do_reset(dev, flags); + /* If reset with a user forcewake, try to restore */ + if (atomic_read(&dev_priv->forcewake_count)) + __gen6_gt_force_wake_get(dev_priv); + break; + case 5: + ret = ironlake_do_reset(dev, flags); + break; + case 4: + ret = i965_do_reset(dev, flags); + break; + case 2: + ret = i8xx_do_reset(dev, flags); + break; + } + dev_priv->last_gpu_reset = get_seconds(); + if (ret) { + DRM_ERROR("Failed to reset chip.\n"); + mutex_unlock(&dev->struct_mutex); + return ret; + } + + /* Ok, now get things going again... */ + + /* + * Everything depends on having the GTT running, so we need to start + * there. Fortunately we don't need to do this unless we reset the + * chip at a PCI level. + * + * Next we need to restore the context, but we don't use those + * yet either... + * + * Ring buffer needs to be re-initialized in the KMS case, or if X + * was running at the time of the reset (i.e. we weren't VT + * switched away). + */ + if (drm_core_check_feature(dev, DRIVER_MODESET) || + !dev_priv->mm.suspended) { + dev_priv->mm.suspended = 0; + + dev_priv->ring[RCS].init(&dev_priv->ring[RCS]); + if (HAS_BSD(dev)) + dev_priv->ring[VCS].init(&dev_priv->ring[VCS]); + if (HAS_BLT(dev)) + dev_priv->ring[BCS].init(&dev_priv->ring[BCS]); + + mutex_unlock(&dev->struct_mutex); + drm_irq_uninstall(dev); + drm_mode_config_reset(dev); + drm_irq_install(dev); + mutex_lock(&dev->struct_mutex); + } + + mutex_unlock(&dev->struct_mutex); + + /* + * Perform a full modeset as on later generations, e.g. Ironlake, we may + * need to retrain the display link and cannot just restore the register + * values. + */ + if (need_display) { + mutex_lock(&dev->mode_config.mutex); + drm_helper_resume_force_mode(dev); + mutex_unlock(&dev->mode_config.mutex); + } + + return 0; +} + + +static int __devinit +i915_pci_probe(struct pci_dev *pdev, const struct pci_device_id *ent) +{ + /* Only bind to function 0 of the device. Early generations + * used function 1 as a placeholder for multi-head. This causes + * us confusion instead, especially on the systems where both + * functions have the same PCI-ID! + */ + if (PCI_FUNC(pdev->devfn)) + return -ENODEV; + + return drm_get_pci_dev(pdev, ent, &driver); +} + +static void +i915_pci_remove(struct pci_dev *pdev) +{ + struct drm_device *dev = pci_get_drvdata(pdev); + + drm_put_dev(dev); +} + +static int i915_pm_suspend(struct device *dev) +{ + struct pci_dev *pdev = to_pci_dev(dev); + struct drm_device *drm_dev = pci_get_drvdata(pdev); + int error; + + if (!drm_dev || !drm_dev->dev_private) { + dev_err(dev, "DRM not initialized, aborting suspend.\n"); + return -ENODEV; + } + + if (drm_dev->switch_power_state == DRM_SWITCH_POWER_OFF) + return 0; + + error = i915_drm_freeze(drm_dev); + if (error) + return error; + + pci_disable_device(pdev); + pci_set_power_state(pdev, PCI_D3hot); + + return 0; +} + +static int i915_pm_resume(struct device *dev) +{ + struct pci_dev *pdev = to_pci_dev(dev); + struct drm_device *drm_dev = pci_get_drvdata(pdev); + + return i915_resume(drm_dev); +} + +static int i915_pm_freeze(struct device *dev) +{ + struct pci_dev *pdev = to_pci_dev(dev); + struct drm_device *drm_dev = pci_get_drvdata(pdev); + + if (!drm_dev || !drm_dev->dev_private) { + dev_err(dev, "DRM not initialized, aborting suspend.\n"); + return -ENODEV; + } + + return i915_drm_freeze(drm_dev); +} + +static int i915_pm_thaw(struct device *dev) +{ + struct pci_dev *pdev = to_pci_dev(dev); + struct drm_device *drm_dev = pci_get_drvdata(pdev); + + return i915_drm_thaw(drm_dev); +} + +static int i915_pm_poweroff(struct device *dev) +{ + struct pci_dev *pdev = to_pci_dev(dev); + struct drm_device *drm_dev = pci_get_drvdata(pdev); + + return i915_drm_freeze(drm_dev); +} + +static const struct dev_pm_ops i915_pm_ops = { + .suspend = i915_pm_suspend, + .resume = i915_pm_resume, + .freeze = i915_pm_freeze, + .thaw = i915_pm_thaw, + .poweroff = i915_pm_poweroff, + .restore = i915_pm_resume, +}; + +static struct vm_operations_struct i915_gem_vm_ops = { + .fault = i915_gem_fault, + .open = drm_gem_vm_open, + .close = drm_gem_vm_close, +}; + +static struct drm_driver driver = { + /* don't use mtrr's here, the Xserver or user space app should + * deal with them for intel hardware. + */ + .driver_features = + DRIVER_USE_AGP | DRIVER_REQUIRE_AGP | /* DRIVER_USE_MTRR |*/ + DRIVER_HAVE_IRQ | DRIVER_IRQ_SHARED | DRIVER_GEM, + .load = i915_driver_load, + .unload = i915_driver_unload, + .open = i915_driver_open, + .lastclose = i915_driver_lastclose, + .preclose = i915_driver_preclose, + .postclose = i915_driver_postclose, + + /* Used in place of i915_pm_ops for non-DRIVER_MODESET */ + .suspend = i915_suspend, + .resume = i915_resume, + + .device_is_agp = i915_driver_device_is_agp, + .reclaim_buffers = drm_core_reclaim_buffers, + .master_create = i915_master_create, + .master_destroy = i915_master_destroy, +#if defined(CONFIG_DEBUG_FS) + .debugfs_init = i915_debugfs_init, + .debugfs_cleanup = i915_debugfs_cleanup, +#endif + .gem_init_object = i915_gem_init_object, + .gem_free_object = i915_gem_free_object, + .gem_vm_ops = &i915_gem_vm_ops, + .dumb_create = i915_gem_dumb_create, + .dumb_map_offset = i915_gem_mmap_gtt, + .dumb_destroy = i915_gem_dumb_destroy, + .ioctls = i915_ioctls, + .fops = { + .owner = THIS_MODULE, + .open = drm_open, + .release = drm_release, + .unlocked_ioctl = drm_ioctl, + .mmap = drm_gem_mmap, + .poll = drm_poll, + .fasync = drm_fasync, + .read = drm_read, +#ifdef CONFIG_COMPAT + .compat_ioctl = i915_compat_ioctl, +#endif + .llseek = noop_llseek, + }, + + .name = DRIVER_NAME, + .desc = DRIVER_DESC, + .date = DRIVER_DATE, + .major = DRIVER_MAJOR, + .minor = DRIVER_MINOR, + .patchlevel = DRIVER_PATCHLEVEL, +}; + +static struct pci_driver i915_pci_driver = { + .name = DRIVER_NAME, + .id_table = pciidlist, + .probe = i915_pci_probe, + .remove = i915_pci_remove, + .driver.pm = &i915_pm_ops, +}; + +static int __init i915_init(void) +{ + if (!intel_agp_enabled) { + DRM_ERROR("drm/i915 can't work without intel_agp module!\n"); + return -ENODEV; + } + + driver.num_ioctls = i915_max_ioctl; + + /* + * If CONFIG_DRM_I915_KMS is set, default to KMS unless + * explicitly disabled with the module pararmeter. + * + * Otherwise, just follow the parameter (defaulting to off). + * + * Allow optional vga_text_mode_force boot option to override + * the default behavior. + */ +#if defined(CONFIG_DRM_I915_KMS) + if (i915_modeset != 0) + driver.driver_features |= DRIVER_MODESET; +#endif + if (i915_modeset == 1) + driver.driver_features |= DRIVER_MODESET; + +#ifdef CONFIG_VGA_CONSOLE + if (vgacon_text_force() && i915_modeset == -1) + driver.driver_features &= ~DRIVER_MODESET; +#endif + + if (!(driver.driver_features & DRIVER_MODESET)) + driver.get_vblank_timestamp = NULL; + + return drm_pci_init(&driver, &i915_pci_driver); +} + +static void __exit i915_exit(void) +{ + drm_pci_exit(&driver, &i915_pci_driver); +} + +module_init(i915_init); +module_exit(i915_exit); + +MODULE_AUTHOR(DRIVER_AUTHOR); +MODULE_DESCRIPTION(DRIVER_DESC); +MODULE_LICENSE("GPL and additional rights"); diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h new file mode 100644 index 00000000..b570415c --- /dev/null +++ b/drivers/gpu/drm/i915/i915_drv.h @@ -0,0 +1,1403 @@ +/* i915_drv.h -- Private header for the I915 driver -*- linux-c -*- + */ +/* + * + * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef _I915_DRV_H_ +#define _I915_DRV_H_ + +#include "i915_reg.h" +#include "intel_bios.h" +#include "intel_ringbuffer.h" +#include +#include +#include + +/* General customization: + */ + +#define DRIVER_AUTHOR "Tungsten Graphics, Inc." + +#define DRIVER_NAME "i915" +#define DRIVER_DESC "Intel Graphics" +#define DRIVER_DATE "20080730" + +enum pipe { + PIPE_A = 0, + PIPE_B, + PIPE_C, + I915_MAX_PIPES +}; +#define pipe_name(p) ((p) + 'A') + +enum plane { + PLANE_A = 0, + PLANE_B, + PLANE_C, +}; +#define plane_name(p) ((p) + 'A') + +#define I915_GEM_GPU_DOMAINS (~(I915_GEM_DOMAIN_CPU | I915_GEM_DOMAIN_GTT)) + +#define for_each_pipe(p) for ((p) = 0; (p) < dev_priv->num_pipe; (p)++) + +/* Interface history: + * + * 1.1: Original. + * 1.2: Add Power Management + * 1.3: Add vblank support + * 1.4: Fix cmdbuffer path, add heap destroy + * 1.5: Add vblank pipe configuration + * 1.6: - New ioctl for scheduling buffer swaps on vertical blank + * - Support vertical blank on secondary display pipe + */ +#define DRIVER_MAJOR 1 +#define DRIVER_MINOR 6 +#define DRIVER_PATCHLEVEL 0 + +#define WATCH_COHERENCY 0 +#define WATCH_LISTS 0 + +#define I915_GEM_PHYS_CURSOR_0 1 +#define I915_GEM_PHYS_CURSOR_1 2 +#define I915_GEM_PHYS_OVERLAY_REGS 3 +#define I915_MAX_PHYS_OBJECT (I915_GEM_PHYS_OVERLAY_REGS) + +struct drm_i915_gem_phys_object { + int id; + struct page **page_list; + drm_dma_handle_t *handle; + struct drm_i915_gem_object *cur_obj; +}; + +struct mem_block { + struct mem_block *next; + struct mem_block *prev; + int start; + int size; + struct drm_file *file_priv; /* NULL: free, -1: heap, other: real files */ +}; + +struct opregion_header; +struct opregion_acpi; +struct opregion_swsci; +struct opregion_asle; + +struct intel_opregion { + struct opregion_header *header; + struct opregion_acpi *acpi; + struct opregion_swsci *swsci; + struct opregion_asle *asle; + void *vbt; + u32 __iomem *lid_state; +}; +#define OPREGION_SIZE (8*1024) + +struct intel_overlay; +struct intel_overlay_error_state; + +struct drm_i915_master_private { + drm_local_map_t *sarea; + struct _drm_i915_sarea *sarea_priv; +}; +#define I915_FENCE_REG_NONE -1 + +struct drm_i915_fence_reg { + struct list_head lru_list; + struct drm_i915_gem_object *obj; + uint32_t setup_seqno; +}; + +struct sdvo_device_mapping { + u8 initialized; + u8 dvo_port; + u8 slave_addr; + u8 dvo_wiring; + u8 i2c_pin; + u8 i2c_speed; + u8 ddc_pin; +}; + +struct intel_display_error_state; + +struct drm_i915_error_state { + u32 eir; + u32 pgtbl_er; + u32 pipestat[I915_MAX_PIPES]; + u32 ipeir; + u32 ipehr; + u32 instdone; + u32 acthd; + u32 error; /* gen6+ */ + u32 bcs_acthd; /* gen6+ blt engine */ + u32 bcs_ipehr; + u32 bcs_ipeir; + u32 bcs_instdone; + u32 bcs_seqno; + u32 vcs_acthd; /* gen6+ bsd engine */ + u32 vcs_ipehr; + u32 vcs_ipeir; + u32 vcs_instdone; + u32 vcs_seqno; + u32 instpm; + u32 instps; + u32 instdone1; + u32 seqno; + u64 bbaddr; + u64 fence[16]; + struct timeval time; + struct drm_i915_error_object { + int page_count; + u32 gtt_offset; + u32 *pages[0]; + } *ringbuffer[I915_NUM_RINGS], *batchbuffer[I915_NUM_RINGS]; + struct drm_i915_error_buffer { + u32 size; + u32 name; + u32 seqno; + u32 gtt_offset; + u32 read_domains; + u32 write_domain; + s32 fence_reg:5; + s32 pinned:2; + u32 tiling:2; + u32 dirty:1; + u32 purgeable:1; + u32 ring:4; + u32 cache_level:2; + } *active_bo, *pinned_bo; + u32 active_bo_count, pinned_bo_count; + struct intel_overlay_error_state *overlay; + struct intel_display_error_state *display; +}; + +struct drm_i915_display_funcs { + void (*dpms)(struct drm_crtc *crtc, int mode); + bool (*fbc_enabled)(struct drm_device *dev); + void (*enable_fbc)(struct drm_crtc *crtc, unsigned long interval); + void (*disable_fbc)(struct drm_device *dev); + int (*get_display_clock_speed)(struct drm_device *dev); + int (*get_fifo_size)(struct drm_device *dev, int plane); + void (*update_wm)(struct drm_device *dev); + int (*crtc_mode_set)(struct drm_crtc *crtc, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode, + int x, int y, + struct drm_framebuffer *old_fb); + void (*fdi_link_train)(struct drm_crtc *crtc); + void (*init_clock_gating)(struct drm_device *dev); + void (*init_pch_clock_gating)(struct drm_device *dev); + int (*queue_flip)(struct drm_device *dev, struct drm_crtc *crtc, + struct drm_framebuffer *fb, + struct drm_i915_gem_object *obj); + /* clock updates for mode set */ + /* cursor updates */ + /* render clock increase/decrease */ + /* display clock increase/decrease */ + /* pll clock increase/decrease */ +}; + +struct intel_device_info { + u8 gen; + u8 is_mobile : 1; + u8 is_i85x : 1; + u8 is_i915g : 1; + u8 is_i945gm : 1; + u8 is_g33 : 1; + u8 need_gfx_hws : 1; + u8 is_g4x : 1; + u8 is_pineview : 1; + u8 is_broadwater : 1; + u8 is_crestline : 1; + u8 is_ivybridge : 1; + u8 has_fbc : 1; + u8 has_pipe_cxsr : 1; + u8 has_hotplug : 1; + u8 cursor_needs_physical : 1; + u8 has_overlay : 1; + u8 overlay_needs_physical : 1; + u8 supports_tv : 1; + u8 has_bsd_ring : 1; + u8 has_blt_ring : 1; +}; + +enum no_fbc_reason { + FBC_NO_OUTPUT, /* no outputs enabled to compress */ + FBC_STOLEN_TOO_SMALL, /* not enough space to hold compressed buffers */ + FBC_UNSUPPORTED_MODE, /* interlace or doublescanned mode */ + FBC_MODE_TOO_LARGE, /* mode too large for compression */ + FBC_BAD_PLANE, /* fbc not supported on plane */ + FBC_NOT_TILED, /* buffer not tiled */ + FBC_MULTIPLE_PIPES, /* more than one pipe active */ + FBC_MODULE_PARAM, +}; + +enum intel_pch { + PCH_IBX, /* Ibexpeak PCH */ + PCH_CPT, /* Cougarpoint PCH */ +}; + +#define QUIRK_PIPEA_FORCE (1<<0) +#define QUIRK_LVDS_SSC_DISABLE (1<<1) + +struct intel_fbdev; + +typedef struct drm_i915_private { + struct drm_device *dev; + + const struct intel_device_info *info; + + int has_gem; + int relative_constants_mode; + + void __iomem *regs; + + struct intel_gmbus { + struct i2c_adapter adapter; + struct i2c_adapter *force_bit; + u32 reg0; + } *gmbus; + + struct pci_dev *bridge_dev; + struct intel_ring_buffer ring[I915_NUM_RINGS]; + uint32_t next_seqno; + + drm_dma_handle_t *status_page_dmah; + uint32_t counter; + drm_local_map_t hws_map; + struct drm_i915_gem_object *pwrctx; + struct drm_i915_gem_object *renderctx; + + struct resource mch_res; + + unsigned int cpp; + int back_offset; + int front_offset; + int current_page; + int page_flipping; + + atomic_t irq_received; + + /* protects the irq masks */ + spinlock_t irq_lock; + /** Cached value of IMR to avoid reads in updating the bitfield */ + u32 pipestat[2]; + u32 irq_mask; + u32 gt_irq_mask; + u32 pch_irq_mask; + + u32 hotplug_supported_mask; + struct work_struct hotplug_work; + + int tex_lru_log_granularity; + int allow_batchbuffer; + struct mem_block *agp_heap; + unsigned int sr01, adpa, ppcr, dvob, dvoc, lvds; + int vblank_pipe; + int num_pipe; + + /* For hangcheck timer */ +#define DRM_I915_HANGCHECK_PERIOD 1500 /* in ms */ + struct timer_list hangcheck_timer; + int hangcheck_count; + uint32_t last_acthd; + uint32_t last_acthd_bsd; + uint32_t last_acthd_blt; + uint32_t last_instdone; + uint32_t last_instdone1; + + unsigned long cfb_size; + unsigned long cfb_pitch; + unsigned long cfb_offset; + int cfb_fence; + int cfb_plane; + int cfb_y; + + struct intel_opregion opregion; + + /* overlay */ + struct intel_overlay *overlay; + + /* LVDS info */ + int backlight_level; /* restore backlight to this value */ + bool backlight_enabled; + struct drm_display_mode *panel_fixed_mode; + struct drm_display_mode *lfp_lvds_vbt_mode; /* if any */ + struct drm_display_mode *sdvo_lvds_vbt_mode; /* if any */ + + /* Feature bits from the VBIOS */ + unsigned int int_tv_support:1; + unsigned int lvds_dither:1; + unsigned int lvds_vbt:1; + unsigned int int_crt_support:1; + unsigned int lvds_use_ssc:1; + int lvds_ssc_freq; + struct { + int rate; + int lanes; + int preemphasis; + int vswing; + + bool initialized; + bool support; + int bpp; + struct edp_power_seq pps; + } edp; + bool no_aux_handshake; + + struct notifier_block lid_notifier; + + int crt_ddc_pin; + struct drm_i915_fence_reg fence_regs[16]; /* assume 965 */ + int fence_reg_start; /* 4 if userland hasn't ioctl'd us yet */ + int num_fence_regs; /* 8 on pre-965, 16 otherwise */ + + unsigned int fsb_freq, mem_freq, is_ddr3; + + spinlock_t error_lock; + struct drm_i915_error_state *first_error; + struct work_struct error_work; + struct completion error_completion; + struct workqueue_struct *wq; + + /* Display functions */ + struct drm_i915_display_funcs display; + + /* PCH chipset type */ + enum intel_pch pch_type; + + unsigned long quirks; + + /* Register state */ + bool modeset_on_lid; + u8 saveLBB; + u32 saveDSPACNTR; + u32 saveDSPBCNTR; + u32 saveDSPARB; + u32 saveHWS; + u32 savePIPEACONF; + u32 savePIPEBCONF; + u32 savePIPEASRC; + u32 savePIPEBSRC; + u32 saveFPA0; + u32 saveFPA1; + u32 saveDPLL_A; + u32 saveDPLL_A_MD; + u32 saveHTOTAL_A; + u32 saveHBLANK_A; + u32 saveHSYNC_A; + u32 saveVTOTAL_A; + u32 saveVBLANK_A; + u32 saveVSYNC_A; + u32 saveBCLRPAT_A; + u32 saveTRANSACONF; + u32 saveTRANS_HTOTAL_A; + u32 saveTRANS_HBLANK_A; + u32 saveTRANS_HSYNC_A; + u32 saveTRANS_VTOTAL_A; + u32 saveTRANS_VBLANK_A; + u32 saveTRANS_VSYNC_A; + u32 savePIPEASTAT; + u32 saveDSPASTRIDE; + u32 saveDSPASIZE; + u32 saveDSPAPOS; + u32 saveDSPAADDR; + u32 saveDSPASURF; + u32 saveDSPATILEOFF; + u32 savePFIT_PGM_RATIOS; + u32 saveBLC_HIST_CTL; + u32 saveBLC_PWM_CTL; + u32 saveBLC_PWM_CTL2; + u32 saveBLC_CPU_PWM_CTL; + u32 saveBLC_CPU_PWM_CTL2; + u32 saveFPB0; + u32 saveFPB1; + u32 saveDPLL_B; + u32 saveDPLL_B_MD; + u32 saveHTOTAL_B; + u32 saveHBLANK_B; + u32 saveHSYNC_B; + u32 saveVTOTAL_B; + u32 saveVBLANK_B; + u32 saveVSYNC_B; + u32 saveBCLRPAT_B; + u32 saveTRANSBCONF; + u32 saveTRANS_HTOTAL_B; + u32 saveTRANS_HBLANK_B; + u32 saveTRANS_HSYNC_B; + u32 saveTRANS_VTOTAL_B; + u32 saveTRANS_VBLANK_B; + u32 saveTRANS_VSYNC_B; + u32 savePIPEBSTAT; + u32 saveDSPBSTRIDE; + u32 saveDSPBSIZE; + u32 saveDSPBPOS; + u32 saveDSPBADDR; + u32 saveDSPBSURF; + u32 saveDSPBTILEOFF; + u32 saveVGA0; + u32 saveVGA1; + u32 saveVGA_PD; + u32 saveVGACNTRL; + u32 saveADPA; + u32 saveLVDS; + u32 savePP_ON_DELAYS; + u32 savePP_OFF_DELAYS; + u32 saveDVOA; + u32 saveDVOB; + u32 saveDVOC; + u32 savePP_ON; + u32 savePP_OFF; + u32 savePP_CONTROL; + u32 savePP_DIVISOR; + u32 savePFIT_CONTROL; + u32 save_palette_a[256]; + u32 save_palette_b[256]; + u32 saveDPFC_CB_BASE; + u32 saveFBC_CFB_BASE; + u32 saveFBC_LL_BASE; + u32 saveFBC_CONTROL; + u32 saveFBC_CONTROL2; + u32 saveIER; + u32 saveIIR; + u32 saveIMR; + u32 saveDEIER; + u32 saveDEIMR; + u32 saveGTIER; + u32 saveGTIMR; + u32 saveFDI_RXA_IMR; + u32 saveFDI_RXB_IMR; + u32 saveCACHE_MODE_0; + u32 saveMI_ARB_STATE; + u32 saveSWF0[16]; + u32 saveSWF1[16]; + u32 saveSWF2[3]; + u8 saveMSR; + u8 saveSR[8]; + u8 saveGR[25]; + u8 saveAR_INDEX; + u8 saveAR[21]; + u8 saveDACMASK; + u8 saveCR[37]; + uint64_t saveFENCE[16]; + u32 saveCURACNTR; + u32 saveCURAPOS; + u32 saveCURABASE; + u32 saveCURBCNTR; + u32 saveCURBPOS; + u32 saveCURBBASE; + u32 saveCURSIZE; + u32 saveDP_B; + u32 saveDP_C; + u32 saveDP_D; + u32 savePIPEA_GMCH_DATA_M; + u32 savePIPEB_GMCH_DATA_M; + u32 savePIPEA_GMCH_DATA_N; + u32 savePIPEB_GMCH_DATA_N; + u32 savePIPEA_DP_LINK_M; + u32 savePIPEB_DP_LINK_M; + u32 savePIPEA_DP_LINK_N; + u32 savePIPEB_DP_LINK_N; + u32 saveFDI_RXA_CTL; + u32 saveFDI_TXA_CTL; + u32 saveFDI_RXB_CTL; + u32 saveFDI_TXB_CTL; + u32 savePFA_CTL_1; + u32 savePFB_CTL_1; + u32 savePFA_WIN_SZ; + u32 savePFB_WIN_SZ; + u32 savePFA_WIN_POS; + u32 savePFB_WIN_POS; + u32 savePCH_DREF_CONTROL; + u32 saveDISP_ARB_CTL; + u32 savePIPEA_DATA_M1; + u32 savePIPEA_DATA_N1; + u32 savePIPEA_LINK_M1; + u32 savePIPEA_LINK_N1; + u32 savePIPEB_DATA_M1; + u32 savePIPEB_DATA_N1; + u32 savePIPEB_LINK_M1; + u32 savePIPEB_LINK_N1; + u32 saveMCHBAR_RENDER_STANDBY; + u32 savePCH_PORT_HOTPLUG; + + struct { + /** Bridge to intel-gtt-ko */ + const struct intel_gtt *gtt; + /** Memory allocator for GTT stolen memory */ + struct drm_mm stolen; + /** Memory allocator for GTT */ + struct drm_mm gtt_space; + /** List of all objects in gtt_space. Used to restore gtt + * mappings on resume */ + struct list_head gtt_list; + + /** Usable portion of the GTT for GEM */ + unsigned long gtt_start; + unsigned long gtt_mappable_end; + unsigned long gtt_end; + + struct io_mapping *gtt_mapping; + int gtt_mtrr; + + struct shrinker inactive_shrinker; + + /** + * List of objects currently involved in rendering. + * + * Includes buffers having the contents of their GPU caches + * flushed, not necessarily primitives. last_rendering_seqno + * represents when the rendering involved will be completed. + * + * A reference is held on the buffer while on this list. + */ + struct list_head active_list; + + /** + * List of objects which are not in the ringbuffer but which + * still have a write_domain which needs to be flushed before + * unbinding. + * + * last_rendering_seqno is 0 while an object is in this list. + * + * A reference is held on the buffer while on this list. + */ + struct list_head flushing_list; + + /** + * LRU list of objects which are not in the ringbuffer and + * are ready to unbind, but are still in the GTT. + * + * last_rendering_seqno is 0 while an object is in this list. + * + * A reference is not held on the buffer while on this list, + * as merely being GTT-bound shouldn't prevent its being + * freed, and we'll pull it off the list in the free path. + */ + struct list_head inactive_list; + + /** + * LRU list of objects which are not in the ringbuffer but + * are still pinned in the GTT. + */ + struct list_head pinned_list; + + /** LRU list of objects with fence regs on them. */ + struct list_head fence_list; + + /** + * List of objects currently pending being freed. + * + * These objects are no longer in use, but due to a signal + * we were prevented from freeing them at the appointed time. + */ + struct list_head deferred_free_list; + + /** + * We leave the user IRQ off as much as possible, + * but this means that requests will finish and never + * be retired once the system goes idle. Set a timer to + * fire periodically while the ring is running. When it + * fires, go retire requests. + */ + struct delayed_work retire_work; + + /** + * Are we in a non-interruptible section of code like + * modesetting? + */ + bool interruptible; + + /** + * Flag if the X Server, and thus DRM, is not currently in + * control of the device. + * + * This is set between LeaveVT and EnterVT. It needs to be + * replaced with a semaphore. It also needs to be + * transitioned away from for kernel modesetting. + */ + int suspended; + + /** + * Flag if the hardware appears to be wedged. + * + * This is set when attempts to idle the device timeout. + * It prevents command submission from occurring and makes + * every pending request fail + */ + atomic_t wedged; + + /** Bit 6 swizzling required for X tiling */ + uint32_t bit_6_swizzle_x; + /** Bit 6 swizzling required for Y tiling */ + uint32_t bit_6_swizzle_y; + + /* storage for physical objects */ + struct drm_i915_gem_phys_object *phys_objs[I915_MAX_PHYS_OBJECT]; + + /* accounting, useful for userland debugging */ + size_t gtt_total; + size_t mappable_gtt_total; + size_t object_memory; + u32 object_count; + } mm; + struct sdvo_device_mapping sdvo_mappings[2]; + /* indicate whether the LVDS_BORDER should be enabled or not */ + unsigned int lvds_border_bits; + /* Panel fitter placement and size for Ironlake+ */ + u32 pch_pf_pos, pch_pf_size; + int panel_t3, panel_t12; + + struct drm_crtc *plane_to_crtc_mapping[2]; + struct drm_crtc *pipe_to_crtc_mapping[2]; + wait_queue_head_t pending_flip_queue; + bool flip_pending_is_done; + + /* Reclocking support */ + bool render_reclock_avail; + bool lvds_downclock_avail; + /* indicates the reduced downclock for LVDS*/ + int lvds_downclock; + struct work_struct idle_work; + struct timer_list idle_timer; + bool busy; + u16 orig_clock; + int child_dev_num; + struct child_device_config *child_dev; + struct drm_connector *int_lvds_connector; + + bool mchbar_need_disable; + + struct work_struct rps_work; + spinlock_t rps_lock; + u32 pm_iir; + + u8 cur_delay; + u8 min_delay; + u8 max_delay; + u8 fmax; + u8 fstart; + + u64 last_count1; + unsigned long last_time1; + unsigned long chipset_power; + u64 last_count2; + struct timespec last_time2; + unsigned long gfx_power; + int c_m; + int r_t; + u8 corr; + spinlock_t *mchdev_lock; + + enum no_fbc_reason no_fbc_reason; + + struct drm_mm_node *compressed_fb; + struct drm_mm_node *compressed_llb; + + unsigned long last_gpu_reset; + + /* list of fbdev register on this device */ + struct intel_fbdev *fbdev; + + struct drm_property *broadcast_rgb_property; + struct drm_property *force_audio_property; + + atomic_t forcewake_count; +} drm_i915_private_t; + +enum i915_cache_level { + I915_CACHE_NONE, + I915_CACHE_LLC, + I915_CACHE_LLC_MLC, /* gen6+ */ +}; + +struct drm_i915_gem_object { + struct drm_gem_object base; + + /** Current space allocated to this object in the GTT, if any. */ + struct drm_mm_node *gtt_space; + struct list_head gtt_list; + + /** This object's place on the active/flushing/inactive lists */ + struct list_head ring_list; + struct list_head mm_list; + /** This object's place on GPU write list */ + struct list_head gpu_write_list; + /** This object's place in the batchbuffer or on the eviction list */ + struct list_head exec_list; + + /** + * This is set if the object is on the active or flushing lists + * (has pending rendering), and is not set if it's on inactive (ready + * to be unbound). + */ + unsigned int active : 1; + + /** + * This is set if the object has been written to since last bound + * to the GTT + */ + unsigned int dirty : 1; + + /** + * This is set if the object has been written to since the last + * GPU flush. + */ + unsigned int pending_gpu_write : 1; + + /** + * Fence register bits (if any) for this object. Will be set + * as needed when mapped into the GTT. + * Protected by dev->struct_mutex. + * + * Size: 4 bits for 16 fences + sign (for FENCE_REG_NONE) + */ + signed int fence_reg : 5; + + /** + * Advice: are the backing pages purgeable? + */ + unsigned int madv : 2; + + /** + * Current tiling mode for the object. + */ + unsigned int tiling_mode : 2; + unsigned int tiling_changed : 1; + + /** How many users have pinned this object in GTT space. The following + * users can each hold at most one reference: pwrite/pread, pin_ioctl + * (via user_pin_count), execbuffer (objects are not allowed multiple + * times for the same batchbuffer), and the framebuffer code. When + * switching/pageflipping, the framebuffer code has at most two buffers + * pinned per crtc. + * + * In the worst case this is 1 + 1 + 1 + 2*2 = 7. That would fit into 3 + * bits with absolutely no headroom. So use 4 bits. */ + unsigned int pin_count : 4; +#define DRM_I915_GEM_OBJECT_MAX_PIN_COUNT 0xf + + /** + * Is the object at the current location in the gtt mappable and + * fenceable? Used to avoid costly recalculations. + */ + unsigned int map_and_fenceable : 1; + + /** + * Whether the current gtt mapping needs to be mappable (and isn't just + * mappable by accident). Track pin and fault separate for a more + * accurate mappable working set. + */ + unsigned int fault_mappable : 1; + unsigned int pin_mappable : 1; + + /* + * Is the GPU currently using a fence to access this buffer, + */ + unsigned int pending_fenced_gpu_access:1; + unsigned int fenced_gpu_access:1; + + unsigned int cache_level:2; + + struct page **pages; + + /** + * DMAR support + */ + struct scatterlist *sg_list; + int num_sg; + + /** + * Used for performing relocations during execbuffer insertion. + */ + struct hlist_node exec_node; + unsigned long exec_handle; + struct drm_i915_gem_exec_object2 *exec_entry; + + /** + * Current offset of the object in GTT space. + * + * This is the same as gtt_space->start + */ + uint32_t gtt_offset; + + /** Breadcrumb of last rendering to the buffer. */ + uint32_t last_rendering_seqno; + struct intel_ring_buffer *ring; + + /** Breadcrumb of last fenced GPU access to the buffer. */ + uint32_t last_fenced_seqno; + struct intel_ring_buffer *last_fenced_ring; + + /** Current tiling stride for the object, if it's tiled. */ + uint32_t stride; + + /** Record of address bit 17 of each page at last unbind. */ + unsigned long *bit_17; + + + /** + * If present, while GEM_DOMAIN_CPU is in the read domain this array + * flags which individual pages are valid. + */ + uint8_t *page_cpu_valid; + + /** User space pin count and filp owning the pin */ + uint32_t user_pin_count; + struct drm_file *pin_filp; + + /** for phy allocated objects */ + struct drm_i915_gem_phys_object *phys_obj; + + /** + * Number of crtcs where this object is currently the fb, but + * will be page flipped away on the next vblank. When it + * reaches 0, dev_priv->pending_flip_queue will be woken up. + */ + atomic_t pending_flip; +}; + +#define to_intel_bo(x) container_of(x, struct drm_i915_gem_object, base) + +/** + * Request queue structure. + * + * The request queue allows us to note sequence numbers that have been emitted + * and may be associated with active buffers to be retired. + * + * By keeping this list, we can avoid having to do questionable + * sequence-number comparisons on buffer last_rendering_seqnos, and associate + * an emission time with seqnos for tracking how far ahead of the GPU we are. + */ +struct drm_i915_gem_request { + /** On Which ring this request was generated */ + struct intel_ring_buffer *ring; + + /** GEM sequence number associated with this request. */ + uint32_t seqno; + + /** Time at which this request was emitted, in jiffies. */ + unsigned long emitted_jiffies; + + /** global list entry for this request */ + struct list_head list; + + struct drm_i915_file_private *file_priv; + /** file_priv list entry for this request */ + struct list_head client_list; +}; + +struct drm_i915_file_private { + struct { + struct spinlock lock; + struct list_head request_list; + } mm; +}; + +#define INTEL_INFO(dev) (((struct drm_i915_private *) (dev)->dev_private)->info) + +#define IS_I830(dev) ((dev)->pci_device == 0x3577) +#define IS_845G(dev) ((dev)->pci_device == 0x2562) +#define IS_I85X(dev) (INTEL_INFO(dev)->is_i85x) +#define IS_I865G(dev) ((dev)->pci_device == 0x2572) +#define IS_I915G(dev) (INTEL_INFO(dev)->is_i915g) +#define IS_I915GM(dev) ((dev)->pci_device == 0x2592) +#define IS_I945G(dev) ((dev)->pci_device == 0x2772) +#define IS_I945GM(dev) (INTEL_INFO(dev)->is_i945gm) +#define IS_BROADWATER(dev) (INTEL_INFO(dev)->is_broadwater) +#define IS_CRESTLINE(dev) (INTEL_INFO(dev)->is_crestline) +#define IS_GM45(dev) ((dev)->pci_device == 0x2A42) +#define IS_G4X(dev) (INTEL_INFO(dev)->is_g4x) +#define IS_PINEVIEW_G(dev) ((dev)->pci_device == 0xa001) +#define IS_PINEVIEW_M(dev) ((dev)->pci_device == 0xa011) +#define IS_PINEVIEW(dev) (INTEL_INFO(dev)->is_pineview) +#define IS_G33(dev) (INTEL_INFO(dev)->is_g33) +#define IS_IRONLAKE_D(dev) ((dev)->pci_device == 0x0042) +#define IS_IRONLAKE_M(dev) ((dev)->pci_device == 0x0046) +#define IS_IVYBRIDGE(dev) (INTEL_INFO(dev)->is_ivybridge) +#define IS_MOBILE(dev) (INTEL_INFO(dev)->is_mobile) + +/* + * The genX designation typically refers to the render engine, so render + * capability related checks should use IS_GEN, while display and other checks + * have their own (e.g. HAS_PCH_SPLIT for ILK+ display, IS_foo for particular + * chips, etc.). + */ +#define IS_GEN2(dev) (INTEL_INFO(dev)->gen == 2) +#define IS_GEN3(dev) (INTEL_INFO(dev)->gen == 3) +#define IS_GEN4(dev) (INTEL_INFO(dev)->gen == 4) +#define IS_GEN5(dev) (INTEL_INFO(dev)->gen == 5) +#define IS_GEN6(dev) (INTEL_INFO(dev)->gen == 6) +#define IS_GEN7(dev) (INTEL_INFO(dev)->gen == 7) + +#define HAS_BSD(dev) (INTEL_INFO(dev)->has_bsd_ring) +#define HAS_BLT(dev) (INTEL_INFO(dev)->has_blt_ring) +#define I915_NEED_GFX_HWS(dev) (INTEL_INFO(dev)->need_gfx_hws) + +#define HAS_OVERLAY(dev) (INTEL_INFO(dev)->has_overlay) +#define OVERLAY_NEEDS_PHYSICAL(dev) (INTEL_INFO(dev)->overlay_needs_physical) + +/* With the 945 and later, Y tiling got adjusted so that it was 32 128-byte + * rows, which changed the alignment requirements and fence programming. + */ +#define HAS_128_BYTE_Y_TILING(dev) (!IS_GEN2(dev) && !(IS_I915G(dev) || \ + IS_I915GM(dev))) +#define SUPPORTS_DIGITAL_OUTPUTS(dev) (!IS_GEN2(dev) && !IS_PINEVIEW(dev)) +#define SUPPORTS_INTEGRATED_HDMI(dev) (IS_G4X(dev) || IS_GEN5(dev)) +#define SUPPORTS_INTEGRATED_DP(dev) (IS_G4X(dev) || IS_GEN5(dev)) +#define SUPPORTS_EDP(dev) (IS_IRONLAKE_M(dev)) +#define SUPPORTS_TV(dev) (INTEL_INFO(dev)->supports_tv) +#define I915_HAS_HOTPLUG(dev) (INTEL_INFO(dev)->has_hotplug) +/* dsparb controlled by hw only */ +#define DSPARB_HWCONTROL(dev) (IS_G4X(dev) || IS_IRONLAKE(dev)) + +#define HAS_FW_BLC(dev) (INTEL_INFO(dev)->gen > 2) +#define HAS_PIPE_CXSR(dev) (INTEL_INFO(dev)->has_pipe_cxsr) +#define I915_HAS_FBC(dev) (INTEL_INFO(dev)->has_fbc) + +#define HAS_PCH_SPLIT(dev) (IS_GEN5(dev) || IS_GEN6(dev) || IS_IVYBRIDGE(dev)) +#define HAS_PIPE_CONTROL(dev) (INTEL_INFO(dev)->gen >= 5) + +#define INTEL_PCH_TYPE(dev) (((struct drm_i915_private *)(dev)->dev_private)->pch_type) +#define HAS_PCH_CPT(dev) (INTEL_PCH_TYPE(dev) == PCH_CPT) +#define HAS_PCH_IBX(dev) (INTEL_PCH_TYPE(dev) == PCH_IBX) + +#include "i915_trace.h" + +extern struct drm_ioctl_desc i915_ioctls[]; +extern int i915_max_ioctl; +extern unsigned int i915_fbpercrtc; +extern int i915_panel_ignore_lid; +extern unsigned int i915_powersave; +extern unsigned int i915_semaphores; +extern unsigned int i915_lvds_downclock; +extern unsigned int i915_panel_use_ssc; +extern int i915_vbt_sdvo_panel_type; +extern unsigned int i915_enable_rc6; +extern unsigned int i915_enable_fbc; + +extern int i915_suspend(struct drm_device *dev, pm_message_t state); +extern int i915_resume(struct drm_device *dev); +extern int i915_master_create(struct drm_device *dev, struct drm_master *master); +extern void i915_master_destroy(struct drm_device *dev, struct drm_master *master); + + /* i915_dma.c */ +extern void i915_kernel_lost_context(struct drm_device * dev); +extern int i915_driver_load(struct drm_device *, unsigned long flags); +extern int i915_driver_unload(struct drm_device *); +extern int i915_driver_open(struct drm_device *dev, struct drm_file *file_priv); +extern void i915_driver_lastclose(struct drm_device * dev); +extern void i915_driver_preclose(struct drm_device *dev, + struct drm_file *file_priv); +extern void i915_driver_postclose(struct drm_device *dev, + struct drm_file *file_priv); +extern int i915_driver_device_is_agp(struct drm_device * dev); +extern long i915_compat_ioctl(struct file *filp, unsigned int cmd, + unsigned long arg); +extern int i915_emit_box(struct drm_device *dev, + struct drm_clip_rect *box, + int DR1, int DR4); +extern int i915_reset(struct drm_device *dev, u8 flags); +extern unsigned long i915_chipset_val(struct drm_i915_private *dev_priv); +extern unsigned long i915_mch_val(struct drm_i915_private *dev_priv); +extern unsigned long i915_gfx_val(struct drm_i915_private *dev_priv); +extern void i915_update_gfx_val(struct drm_i915_private *dev_priv); + + +/* i915_irq.c */ +void i915_hangcheck_elapsed(unsigned long data); +void i915_handle_error(struct drm_device *dev, bool wedged); +extern int i915_irq_emit(struct drm_device *dev, void *data, + struct drm_file *file_priv); +extern int i915_irq_wait(struct drm_device *dev, void *data, + struct drm_file *file_priv); + +extern void intel_irq_init(struct drm_device *dev); + +extern int i915_vblank_pipe_set(struct drm_device *dev, void *data, + struct drm_file *file_priv); +extern int i915_vblank_pipe_get(struct drm_device *dev, void *data, + struct drm_file *file_priv); +extern int i915_vblank_swap(struct drm_device *dev, void *data, + struct drm_file *file_priv); + +void +i915_enable_pipestat(drm_i915_private_t *dev_priv, int pipe, u32 mask); + +void +i915_disable_pipestat(drm_i915_private_t *dev_priv, int pipe, u32 mask); + +void intel_enable_asle (struct drm_device *dev); + +#ifdef CONFIG_DEBUG_FS +extern void i915_destroy_error_state(struct drm_device *dev); +#else +#define i915_destroy_error_state(x) +#endif + + +/* i915_mem.c */ +extern int i915_mem_alloc(struct drm_device *dev, void *data, + struct drm_file *file_priv); +extern int i915_mem_free(struct drm_device *dev, void *data, + struct drm_file *file_priv); +extern int i915_mem_init_heap(struct drm_device *dev, void *data, + struct drm_file *file_priv); +extern int i915_mem_destroy_heap(struct drm_device *dev, void *data, + struct drm_file *file_priv); +extern void i915_mem_takedown(struct mem_block **heap); +extern void i915_mem_release(struct drm_device * dev, + struct drm_file *file_priv, struct mem_block *heap); +/* i915_gem.c */ +int i915_gem_init_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv); +int i915_gem_create_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv); +int i915_gem_pread_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv); +int i915_gem_pwrite_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv); +int i915_gem_mmap_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv); +int i915_gem_mmap_gtt_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv); +int i915_gem_set_domain_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv); +int i915_gem_sw_finish_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv); +int i915_gem_execbuffer(struct drm_device *dev, void *data, + struct drm_file *file_priv); +int i915_gem_execbuffer2(struct drm_device *dev, void *data, + struct drm_file *file_priv); +int i915_gem_pin_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv); +int i915_gem_unpin_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv); +int i915_gem_busy_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv); +int i915_gem_throttle_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv); +int i915_gem_madvise_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv); +int i915_gem_entervt_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv); +int i915_gem_leavevt_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv); +int i915_gem_set_tiling(struct drm_device *dev, void *data, + struct drm_file *file_priv); +int i915_gem_get_tiling(struct drm_device *dev, void *data, + struct drm_file *file_priv); +int i915_gem_get_aperture_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv); +void i915_gem_load(struct drm_device *dev); +int i915_gem_init_object(struct drm_gem_object *obj); +int __must_check i915_gem_flush_ring(struct intel_ring_buffer *ring, + uint32_t invalidate_domains, + uint32_t flush_domains); +struct drm_i915_gem_object *i915_gem_alloc_object(struct drm_device *dev, + size_t size); +void i915_gem_free_object(struct drm_gem_object *obj); +int __must_check i915_gem_object_pin(struct drm_i915_gem_object *obj, + uint32_t alignment, + bool map_and_fenceable); +void i915_gem_object_unpin(struct drm_i915_gem_object *obj); +int __must_check i915_gem_object_unbind(struct drm_i915_gem_object *obj); +void i915_gem_release_mmap(struct drm_i915_gem_object *obj); +void i915_gem_lastclose(struct drm_device *dev); + +int __must_check i915_mutex_lock_interruptible(struct drm_device *dev); +int __must_check i915_gem_object_wait_rendering(struct drm_i915_gem_object *obj); +void i915_gem_object_move_to_active(struct drm_i915_gem_object *obj, + struct intel_ring_buffer *ring, + u32 seqno); + +int i915_gem_dumb_create(struct drm_file *file_priv, + struct drm_device *dev, + struct drm_mode_create_dumb *args); +int i915_gem_mmap_gtt(struct drm_file *file_priv, struct drm_device *dev, + uint32_t handle, uint64_t *offset); +int i915_gem_dumb_destroy(struct drm_file *file_priv, struct drm_device *dev, + uint32_t handle); +/** + * Returns true if seq1 is later than seq2. + */ +static inline bool +i915_seqno_passed(uint32_t seq1, uint32_t seq2) +{ + return (int32_t)(seq1 - seq2) >= 0; +} + +static inline u32 +i915_gem_next_request_seqno(struct intel_ring_buffer *ring) +{ + drm_i915_private_t *dev_priv = ring->dev->dev_private; + return ring->outstanding_lazy_request = dev_priv->next_seqno; +} + +int __must_check i915_gem_object_get_fence(struct drm_i915_gem_object *obj, + struct intel_ring_buffer *pipelined); +int __must_check i915_gem_object_put_fence(struct drm_i915_gem_object *obj); + +void i915_gem_retire_requests(struct drm_device *dev); +void i915_gem_reset(struct drm_device *dev); +void i915_gem_clflush_object(struct drm_i915_gem_object *obj); +int __must_check i915_gem_object_set_domain(struct drm_i915_gem_object *obj, + uint32_t read_domains, + uint32_t write_domain); +int __must_check i915_gem_object_flush_gpu(struct drm_i915_gem_object *obj); +int __must_check i915_gem_init_ringbuffer(struct drm_device *dev); +void i915_gem_cleanup_ringbuffer(struct drm_device *dev); +void i915_gem_do_init(struct drm_device *dev, + unsigned long start, + unsigned long mappable_end, + unsigned long end); +int __must_check i915_gpu_idle(struct drm_device *dev); +int __must_check i915_gem_idle(struct drm_device *dev); +int __must_check i915_add_request(struct intel_ring_buffer *ring, + struct drm_file *file, + struct drm_i915_gem_request *request); +int __must_check i915_wait_request(struct intel_ring_buffer *ring, + uint32_t seqno); +int i915_gem_fault(struct vm_area_struct *vma, struct vm_fault *vmf); +int __must_check +i915_gem_object_set_to_gtt_domain(struct drm_i915_gem_object *obj, + bool write); +int __must_check +i915_gem_object_set_to_display_plane(struct drm_i915_gem_object *obj, + struct intel_ring_buffer *pipelined); +int i915_gem_attach_phys_object(struct drm_device *dev, + struct drm_i915_gem_object *obj, + int id, + int align); +void i915_gem_detach_phys_object(struct drm_device *dev, + struct drm_i915_gem_object *obj); +void i915_gem_free_all_phys_object(struct drm_device *dev); +void i915_gem_release(struct drm_device *dev, struct drm_file *file); + +uint32_t +i915_gem_get_unfenced_gtt_alignment(struct drm_device *dev, + uint32_t size, + int tiling_mode); + +/* i915_gem_gtt.c */ +void i915_gem_restore_gtt_mappings(struct drm_device *dev); +int __must_check i915_gem_gtt_bind_object(struct drm_i915_gem_object *obj); +void i915_gem_gtt_unbind_object(struct drm_i915_gem_object *obj); + +/* i915_gem_evict.c */ +int __must_check i915_gem_evict_something(struct drm_device *dev, int min_size, + unsigned alignment, bool mappable); +int __must_check i915_gem_evict_everything(struct drm_device *dev, + bool purgeable_only); +int __must_check i915_gem_evict_inactive(struct drm_device *dev, + bool purgeable_only); + +/* i915_gem_tiling.c */ +void i915_gem_detect_bit_6_swizzle(struct drm_device *dev); +void i915_gem_object_do_bit_17_swizzle(struct drm_i915_gem_object *obj); +void i915_gem_object_save_bit_17_swizzle(struct drm_i915_gem_object *obj); + +/* i915_gem_debug.c */ +void i915_gem_dump_object(struct drm_i915_gem_object *obj, int len, + const char *where, uint32_t mark); +#if WATCH_LISTS +int i915_verify_lists(struct drm_device *dev); +#else +#define i915_verify_lists(dev) 0 +#endif +void i915_gem_object_check_coherency(struct drm_i915_gem_object *obj, + int handle); +void i915_gem_dump_object(struct drm_i915_gem_object *obj, int len, + const char *where, uint32_t mark); + +/* i915_debugfs.c */ +int i915_debugfs_init(struct drm_minor *minor); +void i915_debugfs_cleanup(struct drm_minor *minor); + +/* i915_suspend.c */ +extern int i915_save_state(struct drm_device *dev); +extern int i915_restore_state(struct drm_device *dev); + +/* i915_suspend.c */ +extern int i915_save_state(struct drm_device *dev); +extern int i915_restore_state(struct drm_device *dev); + +/* intel_i2c.c */ +extern int intel_setup_gmbus(struct drm_device *dev); +extern void intel_teardown_gmbus(struct drm_device *dev); +extern void intel_gmbus_set_speed(struct i2c_adapter *adapter, int speed); +extern void intel_gmbus_force_bit(struct i2c_adapter *adapter, bool force_bit); +extern inline bool intel_gmbus_is_forced_bit(struct i2c_adapter *adapter) +{ + return container_of(adapter, struct intel_gmbus, adapter)->force_bit; +} +extern void intel_i2c_reset(struct drm_device *dev); + +/* intel_opregion.c */ +extern int intel_opregion_setup(struct drm_device *dev); +#ifdef CONFIG_ACPI +extern void intel_opregion_init(struct drm_device *dev); +extern void intel_opregion_fini(struct drm_device *dev); +extern void intel_opregion_asle_intr(struct drm_device *dev); +extern void intel_opregion_gse_intr(struct drm_device *dev); +extern void intel_opregion_enable_asle(struct drm_device *dev); +#else +static inline void intel_opregion_init(struct drm_device *dev) { return; } +static inline void intel_opregion_fini(struct drm_device *dev) { return; } +static inline void intel_opregion_asle_intr(struct drm_device *dev) { return; } +static inline void intel_opregion_gse_intr(struct drm_device *dev) { return; } +static inline void intel_opregion_enable_asle(struct drm_device *dev) { return; } +#endif + +/* intel_acpi.c */ +#ifdef CONFIG_ACPI +extern void intel_register_dsm_handler(void); +extern void intel_unregister_dsm_handler(void); +#else +static inline void intel_register_dsm_handler(void) { return; } +static inline void intel_unregister_dsm_handler(void) { return; } +#endif /* CONFIG_ACPI */ + +/* modesetting */ +extern void intel_modeset_init(struct drm_device *dev); +extern void intel_modeset_gem_init(struct drm_device *dev); +extern void intel_modeset_cleanup(struct drm_device *dev); +extern int intel_modeset_vga_set_state(struct drm_device *dev, bool state); +extern void i8xx_disable_fbc(struct drm_device *dev); +extern void g4x_disable_fbc(struct drm_device *dev); +extern void ironlake_disable_fbc(struct drm_device *dev); +extern void intel_disable_fbc(struct drm_device *dev); +extern void intel_enable_fbc(struct drm_crtc *crtc, unsigned long interval); +extern bool intel_fbc_enabled(struct drm_device *dev); +extern bool ironlake_set_drps(struct drm_device *dev, u8 val); +extern void ironlake_enable_rc6(struct drm_device *dev); +extern void gen6_set_rps(struct drm_device *dev, u8 val); +extern void intel_detect_pch (struct drm_device *dev); +extern int intel_trans_dp_port_sel (struct drm_crtc *crtc); + +/* overlay */ +#ifdef CONFIG_DEBUG_FS +extern struct intel_overlay_error_state *intel_overlay_capture_error_state(struct drm_device *dev); +extern void intel_overlay_print_error_state(struct seq_file *m, struct intel_overlay_error_state *error); + +extern struct intel_display_error_state *intel_display_capture_error_state(struct drm_device *dev); +extern void intel_display_print_error_state(struct seq_file *m, + struct drm_device *dev, + struct intel_display_error_state *error); +#endif + +#define LP_RING(d) (&((struct drm_i915_private *)(d))->ring[RCS]) + +#define BEGIN_LP_RING(n) \ + intel_ring_begin(LP_RING(dev_priv), (n)) + +#define OUT_RING(x) \ + intel_ring_emit(LP_RING(dev_priv), x) + +#define ADVANCE_LP_RING() \ + intel_ring_advance(LP_RING(dev_priv)) + +/** + * Lock test for when it's just for synchronization of ring access. + * + * In that case, we don't need to do it when GEM is initialized as nobody else + * has access to the ring. + */ +#define RING_LOCK_TEST_WITH_RETURN(dev, file) do { \ + if (LP_RING(dev->dev_private)->obj == NULL) \ + LOCK_TEST_WITH_RETURN(dev, file); \ +} while (0) + +/* On SNB platform, before reading ring registers forcewake bit + * must be set to prevent GT core from power down and stale values being + * returned. + */ +void gen6_gt_force_wake_get(struct drm_i915_private *dev_priv); +void gen6_gt_force_wake_put(struct drm_i915_private *dev_priv); +void __gen6_gt_wait_for_fifo(struct drm_i915_private *dev_priv); + +/* We give fast paths for the really cool registers */ +#define NEEDS_FORCE_WAKE(dev_priv, reg) \ + (((dev_priv)->info->gen >= 6) && \ + ((reg) < 0x40000) && \ + ((reg) != FORCEWAKE)) + +#define __i915_read(x, y) \ +static inline u##x i915_read##x(struct drm_i915_private *dev_priv, u32 reg) { \ + u##x val = 0; \ + if (NEEDS_FORCE_WAKE((dev_priv), (reg))) { \ + gen6_gt_force_wake_get(dev_priv); \ + val = read##y(dev_priv->regs + reg); \ + gen6_gt_force_wake_put(dev_priv); \ + } else { \ + val = read##y(dev_priv->regs + reg); \ + } \ + trace_i915_reg_rw(false, reg, val, sizeof(val)); \ + return val; \ +} + +__i915_read(8, b) +__i915_read(16, w) +__i915_read(32, l) +__i915_read(64, q) +#undef __i915_read + +#define __i915_write(x, y) \ +static inline void i915_write##x(struct drm_i915_private *dev_priv, u32 reg, u##x val) { \ + trace_i915_reg_rw(true, reg, val, sizeof(val)); \ + if (NEEDS_FORCE_WAKE((dev_priv), (reg))) { \ + __gen6_gt_wait_for_fifo(dev_priv); \ + } \ + write##y(val, dev_priv->regs + reg); \ +} +__i915_write(8, b) +__i915_write(16, w) +__i915_write(32, l) +__i915_write(64, q) +#undef __i915_write + +#define I915_READ8(reg) i915_read8(dev_priv, (reg)) +#define I915_WRITE8(reg, val) i915_write8(dev_priv, (reg), (val)) + +#define I915_READ16(reg) i915_read16(dev_priv, (reg)) +#define I915_WRITE16(reg, val) i915_write16(dev_priv, (reg), (val)) +#define I915_READ16_NOTRACE(reg) readw(dev_priv->regs + (reg)) +#define I915_WRITE16_NOTRACE(reg, val) writew(val, dev_priv->regs + (reg)) + +#define I915_READ(reg) i915_read32(dev_priv, (reg)) +#define I915_WRITE(reg, val) i915_write32(dev_priv, (reg), (val)) +#define I915_READ_NOTRACE(reg) readl(dev_priv->regs + (reg)) +#define I915_WRITE_NOTRACE(reg, val) writel(val, dev_priv->regs + (reg)) + +#define I915_WRITE64(reg, val) i915_write64(dev_priv, (reg), (val)) +#define I915_READ64(reg) i915_read64(dev_priv, (reg)) + +#define POSTING_READ(reg) (void)I915_READ_NOTRACE(reg) +#define POSTING_READ16(reg) (void)I915_READ16_NOTRACE(reg) + + +#endif diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c new file mode 100644 index 00000000..55485930 --- /dev/null +++ b/drivers/gpu/drm/i915/i915_gem.c @@ -0,0 +1,4149 @@ +/* + * Copyright © 2008 Intel Corporation + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS + * IN THE SOFTWARE. + * + * Authors: + * Eric Anholt + * + */ + +#include "drmP.h" +#include "drm.h" +#include "i915_drm.h" +#include "i915_drv.h" +#include "i915_trace.h" +#include "intel_drv.h" +#include +#include +#include +#include + +static __must_check int i915_gem_object_flush_gpu_write_domain(struct drm_i915_gem_object *obj); +static void i915_gem_object_flush_gtt_write_domain(struct drm_i915_gem_object *obj); +static void i915_gem_object_flush_cpu_write_domain(struct drm_i915_gem_object *obj); +static __must_check int i915_gem_object_set_to_cpu_domain(struct drm_i915_gem_object *obj, + bool write); +static __must_check int i915_gem_object_set_cpu_read_domain_range(struct drm_i915_gem_object *obj, + uint64_t offset, + uint64_t size); +static void i915_gem_object_set_to_full_cpu_read_domain(struct drm_i915_gem_object *obj); +static __must_check int i915_gem_object_bind_to_gtt(struct drm_i915_gem_object *obj, + unsigned alignment, + bool map_and_fenceable); +static void i915_gem_clear_fence_reg(struct drm_device *dev, + struct drm_i915_fence_reg *reg); +static int i915_gem_phys_pwrite(struct drm_device *dev, + struct drm_i915_gem_object *obj, + struct drm_i915_gem_pwrite *args, + struct drm_file *file); +static void i915_gem_free_object_tail(struct drm_i915_gem_object *obj); + +static int i915_gem_inactive_shrink(struct shrinker *shrinker, + struct shrink_control *sc); + +/* some bookkeeping */ +static void i915_gem_info_add_obj(struct drm_i915_private *dev_priv, + size_t size) +{ + dev_priv->mm.object_count++; + dev_priv->mm.object_memory += size; +} + +static void i915_gem_info_remove_obj(struct drm_i915_private *dev_priv, + size_t size) +{ + dev_priv->mm.object_count--; + dev_priv->mm.object_memory -= size; +} + +static int +i915_gem_wait_for_error(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct completion *x = &dev_priv->error_completion; + unsigned long flags; + int ret; + + if (!atomic_read(&dev_priv->mm.wedged)) + return 0; + + ret = wait_for_completion_interruptible(x); + if (ret) + return ret; + + if (atomic_read(&dev_priv->mm.wedged)) { + /* GPU is hung, bump the completion count to account for + * the token we just consumed so that we never hit zero and + * end up waiting upon a subsequent completion event that + * will never happen. + */ + spin_lock_irqsave(&x->wait.lock, flags); + x->done++; + spin_unlock_irqrestore(&x->wait.lock, flags); + } + return 0; +} + +int i915_mutex_lock_interruptible(struct drm_device *dev) +{ + int ret; + + ret = i915_gem_wait_for_error(dev); + if (ret) + return ret; + + ret = mutex_lock_interruptible(&dev->struct_mutex); + if (ret) + return ret; + + WARN_ON(i915_verify_lists(dev)); + return 0; +} + +static inline bool +i915_gem_object_is_inactive(struct drm_i915_gem_object *obj) +{ + return obj->gtt_space && !obj->active && obj->pin_count == 0; +} + +void i915_gem_do_init(struct drm_device *dev, + unsigned long start, + unsigned long mappable_end, + unsigned long end) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + + drm_mm_init(&dev_priv->mm.gtt_space, start, end - start); + + dev_priv->mm.gtt_start = start; + dev_priv->mm.gtt_mappable_end = mappable_end; + dev_priv->mm.gtt_end = end; + dev_priv->mm.gtt_total = end - start; + dev_priv->mm.mappable_gtt_total = min(end, mappable_end) - start; + + /* Take over this portion of the GTT */ + intel_gtt_clear_range(start / PAGE_SIZE, (end-start) / PAGE_SIZE); +} + +int +i915_gem_init_ioctl(struct drm_device *dev, void *data, + struct drm_file *file) +{ + struct drm_i915_gem_init *args = data; + + if (args->gtt_start >= args->gtt_end || + (args->gtt_end | args->gtt_start) & (PAGE_SIZE - 1)) + return -EINVAL; + + mutex_lock(&dev->struct_mutex); + i915_gem_do_init(dev, args->gtt_start, args->gtt_end, args->gtt_end); + mutex_unlock(&dev->struct_mutex); + + return 0; +} + +int +i915_gem_get_aperture_ioctl(struct drm_device *dev, void *data, + struct drm_file *file) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_i915_gem_get_aperture *args = data; + struct drm_i915_gem_object *obj; + size_t pinned; + + if (!(dev->driver->driver_features & DRIVER_GEM)) + return -ENODEV; + + pinned = 0; + mutex_lock(&dev->struct_mutex); + list_for_each_entry(obj, &dev_priv->mm.pinned_list, mm_list) + pinned += obj->gtt_space->size; + mutex_unlock(&dev->struct_mutex); + + args->aper_size = dev_priv->mm.gtt_total; + args->aper_available_size = args->aper_size -pinned; + + return 0; +} + +static int +i915_gem_create(struct drm_file *file, + struct drm_device *dev, + uint64_t size, + uint32_t *handle_p) +{ + struct drm_i915_gem_object *obj; + int ret; + u32 handle; + + size = roundup(size, PAGE_SIZE); + + /* Allocate the new object */ + obj = i915_gem_alloc_object(dev, size); + if (obj == NULL) + return -ENOMEM; + + ret = drm_gem_handle_create(file, &obj->base, &handle); + if (ret) { + drm_gem_object_release(&obj->base); + i915_gem_info_remove_obj(dev->dev_private, obj->base.size); + kfree(obj); + return ret; + } + + /* drop reference from allocate - handle holds it now */ + drm_gem_object_unreference(&obj->base); + trace_i915_gem_object_create(obj); + + *handle_p = handle; + return 0; +} + +int +i915_gem_dumb_create(struct drm_file *file, + struct drm_device *dev, + struct drm_mode_create_dumb *args) +{ + /* have to work out size/pitch and return them */ + args->pitch = ALIGN(args->width * ((args->bpp + 7) / 8), 64); + args->size = args->pitch * args->height; + return i915_gem_create(file, dev, + args->size, &args->handle); +} + +int i915_gem_dumb_destroy(struct drm_file *file, + struct drm_device *dev, + uint32_t handle) +{ + return drm_gem_handle_delete(file, handle); +} + +/** + * Creates a new mm object and returns a handle to it. + */ +int +i915_gem_create_ioctl(struct drm_device *dev, void *data, + struct drm_file *file) +{ + struct drm_i915_gem_create *args = data; + return i915_gem_create(file, dev, + args->size, &args->handle); +} + +static int i915_gem_object_needs_bit17_swizzle(struct drm_i915_gem_object *obj) +{ + drm_i915_private_t *dev_priv = obj->base.dev->dev_private; + + return dev_priv->mm.bit_6_swizzle_x == I915_BIT_6_SWIZZLE_9_10_17 && + obj->tiling_mode != I915_TILING_NONE; +} + +static inline void +slow_shmem_copy(struct page *dst_page, + int dst_offset, + struct page *src_page, + int src_offset, + int length) +{ + char *dst_vaddr, *src_vaddr; + + dst_vaddr = kmap(dst_page); + src_vaddr = kmap(src_page); + + memcpy(dst_vaddr + dst_offset, src_vaddr + src_offset, length); + + kunmap(src_page); + kunmap(dst_page); +} + +static inline void +slow_shmem_bit17_copy(struct page *gpu_page, + int gpu_offset, + struct page *cpu_page, + int cpu_offset, + int length, + int is_read) +{ + char *gpu_vaddr, *cpu_vaddr; + + /* Use the unswizzled path if this page isn't affected. */ + if ((page_to_phys(gpu_page) & (1 << 17)) == 0) { + if (is_read) + return slow_shmem_copy(cpu_page, cpu_offset, + gpu_page, gpu_offset, length); + else + return slow_shmem_copy(gpu_page, gpu_offset, + cpu_page, cpu_offset, length); + } + + gpu_vaddr = kmap(gpu_page); + cpu_vaddr = kmap(cpu_page); + + /* Copy the data, XORing A6 with A17 (1). The user already knows he's + * XORing with the other bits (A9 for Y, A9 and A10 for X) + */ + while (length > 0) { + int cacheline_end = ALIGN(gpu_offset + 1, 64); + int this_length = min(cacheline_end - gpu_offset, length); + int swizzled_gpu_offset = gpu_offset ^ 64; + + if (is_read) { + memcpy(cpu_vaddr + cpu_offset, + gpu_vaddr + swizzled_gpu_offset, + this_length); + } else { + memcpy(gpu_vaddr + swizzled_gpu_offset, + cpu_vaddr + cpu_offset, + this_length); + } + cpu_offset += this_length; + gpu_offset += this_length; + length -= this_length; + } + + kunmap(cpu_page); + kunmap(gpu_page); +} + +/** + * This is the fast shmem pread path, which attempts to copy_from_user directly + * from the backing pages of the object to the user's address space. On a + * fault, it fails so we can fall back to i915_gem_shmem_pwrite_slow(). + */ +static int +i915_gem_shmem_pread_fast(struct drm_device *dev, + struct drm_i915_gem_object *obj, + struct drm_i915_gem_pread *args, + struct drm_file *file) +{ + struct address_space *mapping = obj->base.filp->f_path.dentry->d_inode->i_mapping; + ssize_t remain; + loff_t offset; + char __user *user_data; + int page_offset, page_length; + + user_data = (char __user *) (uintptr_t) args->data_ptr; + remain = args->size; + + offset = args->offset; + + while (remain > 0) { + struct page *page; + char *vaddr; + int ret; + + /* Operation in this page + * + * page_offset = offset within page + * page_length = bytes to copy for this page + */ + page_offset = offset_in_page(offset); + page_length = remain; + if ((page_offset + remain) > PAGE_SIZE) + page_length = PAGE_SIZE - page_offset; + + page = shmem_read_mapping_page(mapping, offset >> PAGE_SHIFT); + if (IS_ERR(page)) + return PTR_ERR(page); + + vaddr = kmap_atomic(page); + ret = __copy_to_user_inatomic(user_data, + vaddr + page_offset, + page_length); + kunmap_atomic(vaddr); + + mark_page_accessed(page); + page_cache_release(page); + if (ret) + return -EFAULT; + + remain -= page_length; + user_data += page_length; + offset += page_length; + } + + return 0; +} + +/** + * This is the fallback shmem pread path, which allocates temporary storage + * in kernel space to copy_to_user into outside of the struct_mutex, so we + * can copy out of the object's backing pages while holding the struct mutex + * and not take page faults. + */ +static int +i915_gem_shmem_pread_slow(struct drm_device *dev, + struct drm_i915_gem_object *obj, + struct drm_i915_gem_pread *args, + struct drm_file *file) +{ + struct address_space *mapping = obj->base.filp->f_path.dentry->d_inode->i_mapping; + struct mm_struct *mm = current->mm; + struct page **user_pages; + ssize_t remain; + loff_t offset, pinned_pages, i; + loff_t first_data_page, last_data_page, num_pages; + int shmem_page_offset; + int data_page_index, data_page_offset; + int page_length; + int ret; + uint64_t data_ptr = args->data_ptr; + int do_bit17_swizzling; + + remain = args->size; + + /* Pin the user pages containing the data. We can't fault while + * holding the struct mutex, yet we want to hold it while + * dereferencing the user data. + */ + first_data_page = data_ptr / PAGE_SIZE; + last_data_page = (data_ptr + args->size - 1) / PAGE_SIZE; + num_pages = last_data_page - first_data_page + 1; + + user_pages = drm_malloc_ab(num_pages, sizeof(struct page *)); + if (user_pages == NULL) + return -ENOMEM; + + mutex_unlock(&dev->struct_mutex); + down_read(&mm->mmap_sem); + pinned_pages = get_user_pages(current, mm, (uintptr_t)args->data_ptr, + num_pages, 1, 0, user_pages, NULL); + up_read(&mm->mmap_sem); + mutex_lock(&dev->struct_mutex); + if (pinned_pages < num_pages) { + ret = -EFAULT; + goto out; + } + + ret = i915_gem_object_set_cpu_read_domain_range(obj, + args->offset, + args->size); + if (ret) + goto out; + + do_bit17_swizzling = i915_gem_object_needs_bit17_swizzle(obj); + + offset = args->offset; + + while (remain > 0) { + struct page *page; + + /* Operation in this page + * + * shmem_page_offset = offset within page in shmem file + * data_page_index = page number in get_user_pages return + * data_page_offset = offset with data_page_index page. + * page_length = bytes to copy for this page + */ + shmem_page_offset = offset_in_page(offset); + data_page_index = data_ptr / PAGE_SIZE - first_data_page; + data_page_offset = offset_in_page(data_ptr); + + page_length = remain; + if ((shmem_page_offset + page_length) > PAGE_SIZE) + page_length = PAGE_SIZE - shmem_page_offset; + if ((data_page_offset + page_length) > PAGE_SIZE) + page_length = PAGE_SIZE - data_page_offset; + + page = shmem_read_mapping_page(mapping, offset >> PAGE_SHIFT); + if (IS_ERR(page)) { + ret = PTR_ERR(page); + goto out; + } + + if (do_bit17_swizzling) { + slow_shmem_bit17_copy(page, + shmem_page_offset, + user_pages[data_page_index], + data_page_offset, + page_length, + 1); + } else { + slow_shmem_copy(user_pages[data_page_index], + data_page_offset, + page, + shmem_page_offset, + page_length); + } + + mark_page_accessed(page); + page_cache_release(page); + + remain -= page_length; + data_ptr += page_length; + offset += page_length; + } + +out: + for (i = 0; i < pinned_pages; i++) { + SetPageDirty(user_pages[i]); + mark_page_accessed(user_pages[i]); + page_cache_release(user_pages[i]); + } + drm_free_large(user_pages); + + return ret; +} + +/** + * Reads data from the object referenced by handle. + * + * On error, the contents of *data are undefined. + */ +int +i915_gem_pread_ioctl(struct drm_device *dev, void *data, + struct drm_file *file) +{ + struct drm_i915_gem_pread *args = data; + struct drm_i915_gem_object *obj; + int ret = 0; + + if (args->size == 0) + return 0; + + if (!access_ok(VERIFY_WRITE, + (char __user *)(uintptr_t)args->data_ptr, + args->size)) + return -EFAULT; + + ret = fault_in_pages_writeable((char __user *)(uintptr_t)args->data_ptr, + args->size); + if (ret) + return -EFAULT; + + ret = i915_mutex_lock_interruptible(dev); + if (ret) + return ret; + + obj = to_intel_bo(drm_gem_object_lookup(dev, file, args->handle)); + if (&obj->base == NULL) { + ret = -ENOENT; + goto unlock; + } + + /* Bounds check source. */ + if (args->offset > obj->base.size || + args->size > obj->base.size - args->offset) { + ret = -EINVAL; + goto out; + } + + trace_i915_gem_object_pread(obj, args->offset, args->size); + + ret = i915_gem_object_set_cpu_read_domain_range(obj, + args->offset, + args->size); + if (ret) + goto out; + + ret = -EFAULT; + if (!i915_gem_object_needs_bit17_swizzle(obj)) + ret = i915_gem_shmem_pread_fast(dev, obj, args, file); + if (ret == -EFAULT) + ret = i915_gem_shmem_pread_slow(dev, obj, args, file); + +out: + drm_gem_object_unreference(&obj->base); +unlock: + mutex_unlock(&dev->struct_mutex); + return ret; +} + +/* This is the fast write path which cannot handle + * page faults in the source data + */ + +static inline int +fast_user_write(struct io_mapping *mapping, + loff_t page_base, int page_offset, + char __user *user_data, + int length) +{ + char *vaddr_atomic; + unsigned long unwritten; + + vaddr_atomic = io_mapping_map_atomic_wc(mapping, page_base); + unwritten = __copy_from_user_inatomic_nocache(vaddr_atomic + page_offset, + user_data, length); + io_mapping_unmap_atomic(vaddr_atomic); + return unwritten; +} + +/* Here's the write path which can sleep for + * page faults + */ + +static inline void +slow_kernel_write(struct io_mapping *mapping, + loff_t gtt_base, int gtt_offset, + struct page *user_page, int user_offset, + int length) +{ + char __iomem *dst_vaddr; + char *src_vaddr; + + dst_vaddr = io_mapping_map_wc(mapping, gtt_base); + src_vaddr = kmap(user_page); + + memcpy_toio(dst_vaddr + gtt_offset, + src_vaddr + user_offset, + length); + + kunmap(user_page); + io_mapping_unmap(dst_vaddr); +} + +/** + * This is the fast pwrite path, where we copy the data directly from the + * user into the GTT, uncached. + */ +static int +i915_gem_gtt_pwrite_fast(struct drm_device *dev, + struct drm_i915_gem_object *obj, + struct drm_i915_gem_pwrite *args, + struct drm_file *file) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + ssize_t remain; + loff_t offset, page_base; + char __user *user_data; + int page_offset, page_length; + + user_data = (char __user *) (uintptr_t) args->data_ptr; + remain = args->size; + + offset = obj->gtt_offset + args->offset; + + while (remain > 0) { + /* Operation in this page + * + * page_base = page offset within aperture + * page_offset = offset within page + * page_length = bytes to copy for this page + */ + page_base = offset & PAGE_MASK; + page_offset = offset_in_page(offset); + page_length = remain; + if ((page_offset + remain) > PAGE_SIZE) + page_length = PAGE_SIZE - page_offset; + + /* If we get a fault while copying data, then (presumably) our + * source page isn't available. Return the error and we'll + * retry in the slow path. + */ + if (fast_user_write(dev_priv->mm.gtt_mapping, page_base, + page_offset, user_data, page_length)) + return -EFAULT; + + remain -= page_length; + user_data += page_length; + offset += page_length; + } + + return 0; +} + +/** + * This is the fallback GTT pwrite path, which uses get_user_pages to pin + * the memory and maps it using kmap_atomic for copying. + * + * This code resulted in x11perf -rgb10text consuming about 10% more CPU + * than using i915_gem_gtt_pwrite_fast on a G45 (32-bit). + */ +static int +i915_gem_gtt_pwrite_slow(struct drm_device *dev, + struct drm_i915_gem_object *obj, + struct drm_i915_gem_pwrite *args, + struct drm_file *file) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + ssize_t remain; + loff_t gtt_page_base, offset; + loff_t first_data_page, last_data_page, num_pages; + loff_t pinned_pages, i; + struct page **user_pages; + struct mm_struct *mm = current->mm; + int gtt_page_offset, data_page_offset, data_page_index, page_length; + int ret; + uint64_t data_ptr = args->data_ptr; + + remain = args->size; + + /* Pin the user pages containing the data. We can't fault while + * holding the struct mutex, and all of the pwrite implementations + * want to hold it while dereferencing the user data. + */ + first_data_page = data_ptr / PAGE_SIZE; + last_data_page = (data_ptr + args->size - 1) / PAGE_SIZE; + num_pages = last_data_page - first_data_page + 1; + + user_pages = drm_malloc_ab(num_pages, sizeof(struct page *)); + if (user_pages == NULL) + return -ENOMEM; + + mutex_unlock(&dev->struct_mutex); + down_read(&mm->mmap_sem); + pinned_pages = get_user_pages(current, mm, (uintptr_t)args->data_ptr, + num_pages, 0, 0, user_pages, NULL); + up_read(&mm->mmap_sem); + mutex_lock(&dev->struct_mutex); + if (pinned_pages < num_pages) { + ret = -EFAULT; + goto out_unpin_pages; + } + + ret = i915_gem_object_set_to_gtt_domain(obj, true); + if (ret) + goto out_unpin_pages; + + ret = i915_gem_object_put_fence(obj); + if (ret) + goto out_unpin_pages; + + offset = obj->gtt_offset + args->offset; + + while (remain > 0) { + /* Operation in this page + * + * gtt_page_base = page offset within aperture + * gtt_page_offset = offset within page in aperture + * data_page_index = page number in get_user_pages return + * data_page_offset = offset with data_page_index page. + * page_length = bytes to copy for this page + */ + gtt_page_base = offset & PAGE_MASK; + gtt_page_offset = offset_in_page(offset); + data_page_index = data_ptr / PAGE_SIZE - first_data_page; + data_page_offset = offset_in_page(data_ptr); + + page_length = remain; + if ((gtt_page_offset + page_length) > PAGE_SIZE) + page_length = PAGE_SIZE - gtt_page_offset; + if ((data_page_offset + page_length) > PAGE_SIZE) + page_length = PAGE_SIZE - data_page_offset; + + slow_kernel_write(dev_priv->mm.gtt_mapping, + gtt_page_base, gtt_page_offset, + user_pages[data_page_index], + data_page_offset, + page_length); + + remain -= page_length; + offset += page_length; + data_ptr += page_length; + } + +out_unpin_pages: + for (i = 0; i < pinned_pages; i++) + page_cache_release(user_pages[i]); + drm_free_large(user_pages); + + return ret; +} + +/** + * This is the fast shmem pwrite path, which attempts to directly + * copy_from_user into the kmapped pages backing the object. + */ +static int +i915_gem_shmem_pwrite_fast(struct drm_device *dev, + struct drm_i915_gem_object *obj, + struct drm_i915_gem_pwrite *args, + struct drm_file *file) +{ + struct address_space *mapping = obj->base.filp->f_path.dentry->d_inode->i_mapping; + ssize_t remain; + loff_t offset; + char __user *user_data; + int page_offset, page_length; + + user_data = (char __user *) (uintptr_t) args->data_ptr; + remain = args->size; + + offset = args->offset; + obj->dirty = 1; + + while (remain > 0) { + struct page *page; + char *vaddr; + int ret; + + /* Operation in this page + * + * page_offset = offset within page + * page_length = bytes to copy for this page + */ + page_offset = offset_in_page(offset); + page_length = remain; + if ((page_offset + remain) > PAGE_SIZE) + page_length = PAGE_SIZE - page_offset; + + page = shmem_read_mapping_page(mapping, offset >> PAGE_SHIFT); + if (IS_ERR(page)) + return PTR_ERR(page); + + vaddr = kmap_atomic(page, KM_USER0); + ret = __copy_from_user_inatomic(vaddr + page_offset, + user_data, + page_length); + kunmap_atomic(vaddr, KM_USER0); + + set_page_dirty(page); + mark_page_accessed(page); + page_cache_release(page); + + /* If we get a fault while copying data, then (presumably) our + * source page isn't available. Return the error and we'll + * retry in the slow path. + */ + if (ret) + return -EFAULT; + + remain -= page_length; + user_data += page_length; + offset += page_length; + } + + return 0; +} + +/** + * This is the fallback shmem pwrite path, which uses get_user_pages to pin + * the memory and maps it using kmap_atomic for copying. + * + * This avoids taking mmap_sem for faulting on the user's address while the + * struct_mutex is held. + */ +static int +i915_gem_shmem_pwrite_slow(struct drm_device *dev, + struct drm_i915_gem_object *obj, + struct drm_i915_gem_pwrite *args, + struct drm_file *file) +{ + struct address_space *mapping = obj->base.filp->f_path.dentry->d_inode->i_mapping; + struct mm_struct *mm = current->mm; + struct page **user_pages; + ssize_t remain; + loff_t offset, pinned_pages, i; + loff_t first_data_page, last_data_page, num_pages; + int shmem_page_offset; + int data_page_index, data_page_offset; + int page_length; + int ret; + uint64_t data_ptr = args->data_ptr; + int do_bit17_swizzling; + + remain = args->size; + + /* Pin the user pages containing the data. We can't fault while + * holding the struct mutex, and all of the pwrite implementations + * want to hold it while dereferencing the user data. + */ + first_data_page = data_ptr / PAGE_SIZE; + last_data_page = (data_ptr + args->size - 1) / PAGE_SIZE; + num_pages = last_data_page - first_data_page + 1; + + user_pages = drm_malloc_ab(num_pages, sizeof(struct page *)); + if (user_pages == NULL) + return -ENOMEM; + + mutex_unlock(&dev->struct_mutex); + down_read(&mm->mmap_sem); + pinned_pages = get_user_pages(current, mm, (uintptr_t)args->data_ptr, + num_pages, 0, 0, user_pages, NULL); + up_read(&mm->mmap_sem); + mutex_lock(&dev->struct_mutex); + if (pinned_pages < num_pages) { + ret = -EFAULT; + goto out; + } + + ret = i915_gem_object_set_to_cpu_domain(obj, 1); + if (ret) + goto out; + + do_bit17_swizzling = i915_gem_object_needs_bit17_swizzle(obj); + + offset = args->offset; + obj->dirty = 1; + + while (remain > 0) { + struct page *page; + + /* Operation in this page + * + * shmem_page_offset = offset within page in shmem file + * data_page_index = page number in get_user_pages return + * data_page_offset = offset with data_page_index page. + * page_length = bytes to copy for this page + */ + shmem_page_offset = offset_in_page(offset); + data_page_index = data_ptr / PAGE_SIZE - first_data_page; + data_page_offset = offset_in_page(data_ptr); + + page_length = remain; + if ((shmem_page_offset + page_length) > PAGE_SIZE) + page_length = PAGE_SIZE - shmem_page_offset; + if ((data_page_offset + page_length) > PAGE_SIZE) + page_length = PAGE_SIZE - data_page_offset; + + page = shmem_read_mapping_page(mapping, offset >> PAGE_SHIFT); + if (IS_ERR(page)) { + ret = PTR_ERR(page); + goto out; + } + + if (do_bit17_swizzling) { + slow_shmem_bit17_copy(page, + shmem_page_offset, + user_pages[data_page_index], + data_page_offset, + page_length, + 0); + } else { + slow_shmem_copy(page, + shmem_page_offset, + user_pages[data_page_index], + data_page_offset, + page_length); + } + + set_page_dirty(page); + mark_page_accessed(page); + page_cache_release(page); + + remain -= page_length; + data_ptr += page_length; + offset += page_length; + } + +out: + for (i = 0; i < pinned_pages; i++) + page_cache_release(user_pages[i]); + drm_free_large(user_pages); + + return ret; +} + +/** + * Writes data to the object referenced by handle. + * + * On error, the contents of the buffer that were to be modified are undefined. + */ +int +i915_gem_pwrite_ioctl(struct drm_device *dev, void *data, + struct drm_file *file) +{ + struct drm_i915_gem_pwrite *args = data; + struct drm_i915_gem_object *obj; + int ret; + + if (args->size == 0) + return 0; + + if (!access_ok(VERIFY_READ, + (char __user *)(uintptr_t)args->data_ptr, + args->size)) + return -EFAULT; + + ret = fault_in_pages_readable((char __user *)(uintptr_t)args->data_ptr, + args->size); + if (ret) + return -EFAULT; + + ret = i915_mutex_lock_interruptible(dev); + if (ret) + return ret; + + obj = to_intel_bo(drm_gem_object_lookup(dev, file, args->handle)); + if (&obj->base == NULL) { + ret = -ENOENT; + goto unlock; + } + + /* Bounds check destination. */ + if (args->offset > obj->base.size || + args->size > obj->base.size - args->offset) { + ret = -EINVAL; + goto out; + } + + trace_i915_gem_object_pwrite(obj, args->offset, args->size); + + /* We can only do the GTT pwrite on untiled buffers, as otherwise + * it would end up going through the fenced access, and we'll get + * different detiling behavior between reading and writing. + * pread/pwrite currently are reading and writing from the CPU + * perspective, requiring manual detiling by the client. + */ + if (obj->phys_obj) + ret = i915_gem_phys_pwrite(dev, obj, args, file); + else if (obj->gtt_space && + obj->base.write_domain != I915_GEM_DOMAIN_CPU) { + ret = i915_gem_object_pin(obj, 0, true); + if (ret) + goto out; + + ret = i915_gem_object_set_to_gtt_domain(obj, true); + if (ret) + goto out_unpin; + + ret = i915_gem_object_put_fence(obj); + if (ret) + goto out_unpin; + + ret = i915_gem_gtt_pwrite_fast(dev, obj, args, file); + if (ret == -EFAULT) + ret = i915_gem_gtt_pwrite_slow(dev, obj, args, file); + +out_unpin: + i915_gem_object_unpin(obj); + } else { + ret = i915_gem_object_set_to_cpu_domain(obj, 1); + if (ret) + goto out; + + ret = -EFAULT; + if (!i915_gem_object_needs_bit17_swizzle(obj)) + ret = i915_gem_shmem_pwrite_fast(dev, obj, args, file); + if (ret == -EFAULT) + ret = i915_gem_shmem_pwrite_slow(dev, obj, args, file); + } + +out: + drm_gem_object_unreference(&obj->base); +unlock: + mutex_unlock(&dev->struct_mutex); + return ret; +} + +/** + * Called when user space prepares to use an object with the CPU, either + * through the mmap ioctl's mapping or a GTT mapping. + */ +int +i915_gem_set_domain_ioctl(struct drm_device *dev, void *data, + struct drm_file *file) +{ + struct drm_i915_gem_set_domain *args = data; + struct drm_i915_gem_object *obj; + uint32_t read_domains = args->read_domains; + uint32_t write_domain = args->write_domain; + int ret; + + if (!(dev->driver->driver_features & DRIVER_GEM)) + return -ENODEV; + + /* Only handle setting domains to types used by the CPU. */ + if (write_domain & I915_GEM_GPU_DOMAINS) + return -EINVAL; + + if (read_domains & I915_GEM_GPU_DOMAINS) + return -EINVAL; + + /* Having something in the write domain implies it's in the read + * domain, and only that read domain. Enforce that in the request. + */ + if (write_domain != 0 && read_domains != write_domain) + return -EINVAL; + + ret = i915_mutex_lock_interruptible(dev); + if (ret) + return ret; + + obj = to_intel_bo(drm_gem_object_lookup(dev, file, args->handle)); + if (&obj->base == NULL) { + ret = -ENOENT; + goto unlock; + } + + if (read_domains & I915_GEM_DOMAIN_GTT) { + ret = i915_gem_object_set_to_gtt_domain(obj, write_domain != 0); + + /* Silently promote "you're not bound, there was nothing to do" + * to success, since the client was just asking us to + * make sure everything was done. + */ + if (ret == -EINVAL) + ret = 0; + } else { + ret = i915_gem_object_set_to_cpu_domain(obj, write_domain != 0); + } + + drm_gem_object_unreference(&obj->base); +unlock: + mutex_unlock(&dev->struct_mutex); + return ret; +} + +/** + * Called when user space has done writes to this buffer + */ +int +i915_gem_sw_finish_ioctl(struct drm_device *dev, void *data, + struct drm_file *file) +{ + struct drm_i915_gem_sw_finish *args = data; + struct drm_i915_gem_object *obj; + int ret = 0; + + if (!(dev->driver->driver_features & DRIVER_GEM)) + return -ENODEV; + + ret = i915_mutex_lock_interruptible(dev); + if (ret) + return ret; + + obj = to_intel_bo(drm_gem_object_lookup(dev, file, args->handle)); + if (&obj->base == NULL) { + ret = -ENOENT; + goto unlock; + } + + /* Pinned buffers may be scanout, so flush the cache */ + if (obj->pin_count) + i915_gem_object_flush_cpu_write_domain(obj); + + drm_gem_object_unreference(&obj->base); +unlock: + mutex_unlock(&dev->struct_mutex); + return ret; +} + +/** + * Maps the contents of an object, returning the address it is mapped + * into. + * + * While the mapping holds a reference on the contents of the object, it doesn't + * imply a ref on the object itself. + */ +int +i915_gem_mmap_ioctl(struct drm_device *dev, void *data, + struct drm_file *file) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_i915_gem_mmap *args = data; + struct drm_gem_object *obj; + unsigned long addr; + + if (!(dev->driver->driver_features & DRIVER_GEM)) + return -ENODEV; + + obj = drm_gem_object_lookup(dev, file, args->handle); + if (obj == NULL) + return -ENOENT; + + if (obj->size > dev_priv->mm.gtt_mappable_end) { + drm_gem_object_unreference_unlocked(obj); + return -E2BIG; + } + + down_write(¤t->mm->mmap_sem); + addr = do_mmap(obj->filp, 0, args->size, + PROT_READ | PROT_WRITE, MAP_SHARED, + args->offset); + up_write(¤t->mm->mmap_sem); + drm_gem_object_unreference_unlocked(obj); + if (IS_ERR((void *)addr)) + return addr; + + args->addr_ptr = (uint64_t) addr; + + return 0; +} + +/** + * i915_gem_fault - fault a page into the GTT + * vma: VMA in question + * vmf: fault info + * + * The fault handler is set up by drm_gem_mmap() when a object is GTT mapped + * from userspace. The fault handler takes care of binding the object to + * the GTT (if needed), allocating and programming a fence register (again, + * only if needed based on whether the old reg is still valid or the object + * is tiled) and inserting a new PTE into the faulting process. + * + * Note that the faulting process may involve evicting existing objects + * from the GTT and/or fence registers to make room. So performance may + * suffer if the GTT working set is large or there are few fence registers + * left. + */ +int i915_gem_fault(struct vm_area_struct *vma, struct vm_fault *vmf) +{ + struct drm_i915_gem_object *obj = to_intel_bo(vma->vm_private_data); + struct drm_device *dev = obj->base.dev; + drm_i915_private_t *dev_priv = dev->dev_private; + pgoff_t page_offset; + unsigned long pfn; + int ret = 0; + bool write = !!(vmf->flags & FAULT_FLAG_WRITE); + + /* We don't use vmf->pgoff since that has the fake offset */ + page_offset = ((unsigned long)vmf->virtual_address - vma->vm_start) >> + PAGE_SHIFT; + + ret = i915_mutex_lock_interruptible(dev); + if (ret) + goto out; + + trace_i915_gem_object_fault(obj, page_offset, true, write); + + /* Now bind it into the GTT if needed */ + if (!obj->map_and_fenceable) { + ret = i915_gem_object_unbind(obj); + if (ret) + goto unlock; + } + if (!obj->gtt_space) { + ret = i915_gem_object_bind_to_gtt(obj, 0, true); + if (ret) + goto unlock; + + ret = i915_gem_object_set_to_gtt_domain(obj, write); + if (ret) + goto unlock; + } + + if (obj->tiling_mode == I915_TILING_NONE) + ret = i915_gem_object_put_fence(obj); + else + ret = i915_gem_object_get_fence(obj, NULL); + if (ret) + goto unlock; + + if (i915_gem_object_is_inactive(obj)) + list_move_tail(&obj->mm_list, &dev_priv->mm.inactive_list); + + obj->fault_mappable = true; + + pfn = ((dev->agp->base + obj->gtt_offset) >> PAGE_SHIFT) + + page_offset; + + /* Finally, remap it using the new GTT offset */ + ret = vm_insert_pfn(vma, (unsigned long)vmf->virtual_address, pfn); +unlock: + mutex_unlock(&dev->struct_mutex); +out: + switch (ret) { + case -EIO: + case -EAGAIN: + /* Give the error handler a chance to run and move the + * objects off the GPU active list. Next time we service the + * fault, we should be able to transition the page into the + * GTT without touching the GPU (and so avoid further + * EIO/EGAIN). If the GPU is wedged, then there is no issue + * with coherency, just lost writes. + */ + set_need_resched(); + case 0: + case -ERESTARTSYS: + case -EINTR: + return VM_FAULT_NOPAGE; + case -ENOMEM: + return VM_FAULT_OOM; + default: + return VM_FAULT_SIGBUS; + } +} + +/** + * i915_gem_create_mmap_offset - create a fake mmap offset for an object + * @obj: obj in question + * + * GEM memory mapping works by handing back to userspace a fake mmap offset + * it can use in a subsequent mmap(2) call. The DRM core code then looks + * up the object based on the offset and sets up the various memory mapping + * structures. + * + * This routine allocates and attaches a fake offset for @obj. + */ +static int +i915_gem_create_mmap_offset(struct drm_i915_gem_object *obj) +{ + struct drm_device *dev = obj->base.dev; + struct drm_gem_mm *mm = dev->mm_private; + struct drm_map_list *list; + struct drm_local_map *map; + int ret = 0; + + /* Set the object up for mmap'ing */ + list = &obj->base.map_list; + list->map = kzalloc(sizeof(struct drm_map_list), GFP_KERNEL); + if (!list->map) + return -ENOMEM; + + map = list->map; + map->type = _DRM_GEM; + map->size = obj->base.size; + map->handle = obj; + + /* Get a DRM GEM mmap offset allocated... */ + list->file_offset_node = drm_mm_search_free(&mm->offset_manager, + obj->base.size / PAGE_SIZE, + 0, 0); + if (!list->file_offset_node) { + DRM_ERROR("failed to allocate offset for bo %d\n", + obj->base.name); + ret = -ENOSPC; + goto out_free_list; + } + + list->file_offset_node = drm_mm_get_block(list->file_offset_node, + obj->base.size / PAGE_SIZE, + 0); + if (!list->file_offset_node) { + ret = -ENOMEM; + goto out_free_list; + } + + list->hash.key = list->file_offset_node->start; + ret = drm_ht_insert_item(&mm->offset_hash, &list->hash); + if (ret) { + DRM_ERROR("failed to add to map hash\n"); + goto out_free_mm; + } + + return 0; + +out_free_mm: + drm_mm_put_block(list->file_offset_node); +out_free_list: + kfree(list->map); + list->map = NULL; + + return ret; +} + +/** + * i915_gem_release_mmap - remove physical page mappings + * @obj: obj in question + * + * Preserve the reservation of the mmapping with the DRM core code, but + * relinquish ownership of the pages back to the system. + * + * It is vital that we remove the page mapping if we have mapped a tiled + * object through the GTT and then lose the fence register due to + * resource pressure. Similarly if the object has been moved out of the + * aperture, than pages mapped into userspace must be revoked. Removing the + * mapping will then trigger a page fault on the next user access, allowing + * fixup by i915_gem_fault(). + */ +void +i915_gem_release_mmap(struct drm_i915_gem_object *obj) +{ + if (!obj->fault_mappable) + return; + + if (obj->base.dev->dev_mapping) + unmap_mapping_range(obj->base.dev->dev_mapping, + (loff_t)obj->base.map_list.hash.key<base.size, 1); + + obj->fault_mappable = false; +} + +static void +i915_gem_free_mmap_offset(struct drm_i915_gem_object *obj) +{ + struct drm_device *dev = obj->base.dev; + struct drm_gem_mm *mm = dev->mm_private; + struct drm_map_list *list = &obj->base.map_list; + + drm_ht_remove_item(&mm->offset_hash, &list->hash); + drm_mm_put_block(list->file_offset_node); + kfree(list->map); + list->map = NULL; +} + +static uint32_t +i915_gem_get_gtt_size(struct drm_device *dev, uint32_t size, int tiling_mode) +{ + uint32_t gtt_size; + + if (INTEL_INFO(dev)->gen >= 4 || + tiling_mode == I915_TILING_NONE) + return size; + + /* Previous chips need a power-of-two fence region when tiling */ + if (INTEL_INFO(dev)->gen == 3) + gtt_size = 1024*1024; + else + gtt_size = 512*1024; + + while (gtt_size < size) + gtt_size <<= 1; + + return gtt_size; +} + +/** + * i915_gem_get_gtt_alignment - return required GTT alignment for an object + * @obj: object to check + * + * Return the required GTT alignment for an object, taking into account + * potential fence register mapping. + */ +static uint32_t +i915_gem_get_gtt_alignment(struct drm_device *dev, + uint32_t size, + int tiling_mode) +{ + /* + * Minimum alignment is 4k (GTT page size), but might be greater + * if a fence register is needed for the object. + */ + if (INTEL_INFO(dev)->gen >= 4 || + tiling_mode == I915_TILING_NONE) + return 4096; + + /* + * Previous chips need to be aligned to the size of the smallest + * fence register that can contain the object. + */ + return i915_gem_get_gtt_size(dev, size, tiling_mode); +} + +/** + * i915_gem_get_unfenced_gtt_alignment - return required GTT alignment for an + * unfenced object + * @dev: the device + * @size: size of the object + * @tiling_mode: tiling mode of the object + * + * Return the required GTT alignment for an object, only taking into account + * unfenced tiled surface requirements. + */ +uint32_t +i915_gem_get_unfenced_gtt_alignment(struct drm_device *dev, + uint32_t size, + int tiling_mode) +{ + /* + * Minimum alignment is 4k (GTT page size) for sane hw. + */ + if (INTEL_INFO(dev)->gen >= 4 || IS_G33(dev) || + tiling_mode == I915_TILING_NONE) + return 4096; + + /* Previous hardware however needs to be aligned to a power-of-two + * tile height. The simplest method for determining this is to reuse + * the power-of-tile object size. + */ + return i915_gem_get_gtt_size(dev, size, tiling_mode); +} + +int +i915_gem_mmap_gtt(struct drm_file *file, + struct drm_device *dev, + uint32_t handle, + uint64_t *offset) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_i915_gem_object *obj; + int ret; + + if (!(dev->driver->driver_features & DRIVER_GEM)) + return -ENODEV; + + ret = i915_mutex_lock_interruptible(dev); + if (ret) + return ret; + + obj = to_intel_bo(drm_gem_object_lookup(dev, file, handle)); + if (&obj->base == NULL) { + ret = -ENOENT; + goto unlock; + } + + if (obj->base.size > dev_priv->mm.gtt_mappable_end) { + ret = -E2BIG; + goto out; + } + + if (obj->madv != I915_MADV_WILLNEED) { + DRM_ERROR("Attempting to mmap a purgeable buffer\n"); + ret = -EINVAL; + goto out; + } + + if (!obj->base.map_list.map) { + ret = i915_gem_create_mmap_offset(obj); + if (ret) + goto out; + } + + *offset = (u64)obj->base.map_list.hash.key << PAGE_SHIFT; + +out: + drm_gem_object_unreference(&obj->base); +unlock: + mutex_unlock(&dev->struct_mutex); + return ret; +} + +/** + * i915_gem_mmap_gtt_ioctl - prepare an object for GTT mmap'ing + * @dev: DRM device + * @data: GTT mapping ioctl data + * @file: GEM object info + * + * Simply returns the fake offset to userspace so it can mmap it. + * The mmap call will end up in drm_gem_mmap(), which will set things + * up so we can get faults in the handler above. + * + * The fault handler will take care of binding the object into the GTT + * (since it may have been evicted to make room for something), allocating + * a fence register, and mapping the appropriate aperture address into + * userspace. + */ +int +i915_gem_mmap_gtt_ioctl(struct drm_device *dev, void *data, + struct drm_file *file) +{ + struct drm_i915_gem_mmap_gtt *args = data; + + if (!(dev->driver->driver_features & DRIVER_GEM)) + return -ENODEV; + + return i915_gem_mmap_gtt(file, dev, args->handle, &args->offset); +} + + +static int +i915_gem_object_get_pages_gtt(struct drm_i915_gem_object *obj, + gfp_t gfpmask) +{ + int page_count, i; + struct address_space *mapping; + struct inode *inode; + struct page *page; + + /* Get the list of pages out of our struct file. They'll be pinned + * at this point until we release them. + */ + page_count = obj->base.size / PAGE_SIZE; + BUG_ON(obj->pages != NULL); + obj->pages = drm_malloc_ab(page_count, sizeof(struct page *)); + if (obj->pages == NULL) + return -ENOMEM; + + inode = obj->base.filp->f_path.dentry->d_inode; + mapping = inode->i_mapping; + gfpmask |= mapping_gfp_mask(mapping); + + for (i = 0; i < page_count; i++) { + page = shmem_read_mapping_page_gfp(mapping, i, gfpmask); + if (IS_ERR(page)) + goto err_pages; + + obj->pages[i] = page; + } + + if (obj->tiling_mode != I915_TILING_NONE) + i915_gem_object_do_bit_17_swizzle(obj); + + return 0; + +err_pages: + while (i--) + page_cache_release(obj->pages[i]); + + drm_free_large(obj->pages); + obj->pages = NULL; + return PTR_ERR(page); +} + +static void +i915_gem_object_put_pages_gtt(struct drm_i915_gem_object *obj) +{ + int page_count = obj->base.size / PAGE_SIZE; + int i; + + BUG_ON(obj->madv == __I915_MADV_PURGED); + + if (obj->tiling_mode != I915_TILING_NONE) + i915_gem_object_save_bit_17_swizzle(obj); + + if (obj->madv == I915_MADV_DONTNEED) + obj->dirty = 0; + + for (i = 0; i < page_count; i++) { + if (obj->dirty) + set_page_dirty(obj->pages[i]); + + if (obj->madv == I915_MADV_WILLNEED) + mark_page_accessed(obj->pages[i]); + + page_cache_release(obj->pages[i]); + } + obj->dirty = 0; + + drm_free_large(obj->pages); + obj->pages = NULL; +} + +void +i915_gem_object_move_to_active(struct drm_i915_gem_object *obj, + struct intel_ring_buffer *ring, + u32 seqno) +{ + struct drm_device *dev = obj->base.dev; + struct drm_i915_private *dev_priv = dev->dev_private; + + BUG_ON(ring == NULL); + obj->ring = ring; + + /* Add a reference if we're newly entering the active list. */ + if (!obj->active) { + drm_gem_object_reference(&obj->base); + obj->active = 1; + } + + /* Move from whatever list we were on to the tail of execution. */ + list_move_tail(&obj->mm_list, &dev_priv->mm.active_list); + list_move_tail(&obj->ring_list, &ring->active_list); + + obj->last_rendering_seqno = seqno; + if (obj->fenced_gpu_access) { + struct drm_i915_fence_reg *reg; + + BUG_ON(obj->fence_reg == I915_FENCE_REG_NONE); + + obj->last_fenced_seqno = seqno; + obj->last_fenced_ring = ring; + + reg = &dev_priv->fence_regs[obj->fence_reg]; + list_move_tail(®->lru_list, &dev_priv->mm.fence_list); + } +} + +static void +i915_gem_object_move_off_active(struct drm_i915_gem_object *obj) +{ + list_del_init(&obj->ring_list); + obj->last_rendering_seqno = 0; +} + +static void +i915_gem_object_move_to_flushing(struct drm_i915_gem_object *obj) +{ + struct drm_device *dev = obj->base.dev; + drm_i915_private_t *dev_priv = dev->dev_private; + + BUG_ON(!obj->active); + list_move_tail(&obj->mm_list, &dev_priv->mm.flushing_list); + + i915_gem_object_move_off_active(obj); +} + +static void +i915_gem_object_move_to_inactive(struct drm_i915_gem_object *obj) +{ + struct drm_device *dev = obj->base.dev; + struct drm_i915_private *dev_priv = dev->dev_private; + + if (obj->pin_count != 0) + list_move_tail(&obj->mm_list, &dev_priv->mm.pinned_list); + else + list_move_tail(&obj->mm_list, &dev_priv->mm.inactive_list); + + BUG_ON(!list_empty(&obj->gpu_write_list)); + BUG_ON(!obj->active); + obj->ring = NULL; + + i915_gem_object_move_off_active(obj); + obj->fenced_gpu_access = false; + + obj->active = 0; + obj->pending_gpu_write = false; + drm_gem_object_unreference(&obj->base); + + WARN_ON(i915_verify_lists(dev)); +} + +/* Immediately discard the backing storage */ +static void +i915_gem_object_truncate(struct drm_i915_gem_object *obj) +{ + struct inode *inode; + + /* Our goal here is to return as much of the memory as + * is possible back to the system as we are called from OOM. + * To do this we must instruct the shmfs to drop all of its + * backing pages, *now*. + */ + inode = obj->base.filp->f_path.dentry->d_inode; + shmem_truncate_range(inode, 0, (loff_t)-1); + + obj->madv = __I915_MADV_PURGED; +} + +static inline int +i915_gem_object_is_purgeable(struct drm_i915_gem_object *obj) +{ + return obj->madv == I915_MADV_DONTNEED; +} + +static void +i915_gem_process_flushing_list(struct intel_ring_buffer *ring, + uint32_t flush_domains) +{ + struct drm_i915_gem_object *obj, *next; + + list_for_each_entry_safe(obj, next, + &ring->gpu_write_list, + gpu_write_list) { + if (obj->base.write_domain & flush_domains) { + uint32_t old_write_domain = obj->base.write_domain; + + obj->base.write_domain = 0; + list_del_init(&obj->gpu_write_list); + i915_gem_object_move_to_active(obj, ring, + i915_gem_next_request_seqno(ring)); + + trace_i915_gem_object_change_domain(obj, + obj->base.read_domains, + old_write_domain); + } + } +} + +int +i915_add_request(struct intel_ring_buffer *ring, + struct drm_file *file, + struct drm_i915_gem_request *request) +{ + drm_i915_private_t *dev_priv = ring->dev->dev_private; + uint32_t seqno; + int was_empty; + int ret; + + BUG_ON(request == NULL); + + ret = ring->add_request(ring, &seqno); + if (ret) + return ret; + + trace_i915_gem_request_add(ring, seqno); + + request->seqno = seqno; + request->ring = ring; + request->emitted_jiffies = jiffies; + was_empty = list_empty(&ring->request_list); + list_add_tail(&request->list, &ring->request_list); + + if (file) { + struct drm_i915_file_private *file_priv = file->driver_priv; + + spin_lock(&file_priv->mm.lock); + request->file_priv = file_priv; + list_add_tail(&request->client_list, + &file_priv->mm.request_list); + spin_unlock(&file_priv->mm.lock); + } + + ring->outstanding_lazy_request = false; + + if (!dev_priv->mm.suspended) { + mod_timer(&dev_priv->hangcheck_timer, + jiffies + msecs_to_jiffies(DRM_I915_HANGCHECK_PERIOD)); + if (was_empty) + queue_delayed_work(dev_priv->wq, + &dev_priv->mm.retire_work, HZ); + } + return 0; +} + +static inline void +i915_gem_request_remove_from_client(struct drm_i915_gem_request *request) +{ + struct drm_i915_file_private *file_priv = request->file_priv; + + if (!file_priv) + return; + + spin_lock(&file_priv->mm.lock); + if (request->file_priv) { + list_del(&request->client_list); + request->file_priv = NULL; + } + spin_unlock(&file_priv->mm.lock); +} + +static void i915_gem_reset_ring_lists(struct drm_i915_private *dev_priv, + struct intel_ring_buffer *ring) +{ + while (!list_empty(&ring->request_list)) { + struct drm_i915_gem_request *request; + + request = list_first_entry(&ring->request_list, + struct drm_i915_gem_request, + list); + + list_del(&request->list); + i915_gem_request_remove_from_client(request); + kfree(request); + } + + while (!list_empty(&ring->active_list)) { + struct drm_i915_gem_object *obj; + + obj = list_first_entry(&ring->active_list, + struct drm_i915_gem_object, + ring_list); + + obj->base.write_domain = 0; + list_del_init(&obj->gpu_write_list); + i915_gem_object_move_to_inactive(obj); + } +} + +static void i915_gem_reset_fences(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + int i; + + for (i = 0; i < 16; i++) { + struct drm_i915_fence_reg *reg = &dev_priv->fence_regs[i]; + struct drm_i915_gem_object *obj = reg->obj; + + if (!obj) + continue; + + if (obj->tiling_mode) + i915_gem_release_mmap(obj); + + reg->obj->fence_reg = I915_FENCE_REG_NONE; + reg->obj->fenced_gpu_access = false; + reg->obj->last_fenced_seqno = 0; + reg->obj->last_fenced_ring = NULL; + i915_gem_clear_fence_reg(dev, reg); + } +} + +void i915_gem_reset(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_i915_gem_object *obj; + int i; + + for (i = 0; i < I915_NUM_RINGS; i++) + i915_gem_reset_ring_lists(dev_priv, &dev_priv->ring[i]); + + /* Remove anything from the flushing lists. The GPU cache is likely + * to be lost on reset along with the data, so simply move the + * lost bo to the inactive list. + */ + while (!list_empty(&dev_priv->mm.flushing_list)) { + obj= list_first_entry(&dev_priv->mm.flushing_list, + struct drm_i915_gem_object, + mm_list); + + obj->base.write_domain = 0; + list_del_init(&obj->gpu_write_list); + i915_gem_object_move_to_inactive(obj); + } + + /* Move everything out of the GPU domains to ensure we do any + * necessary invalidation upon reuse. + */ + list_for_each_entry(obj, + &dev_priv->mm.inactive_list, + mm_list) + { + obj->base.read_domains &= ~I915_GEM_GPU_DOMAINS; + } + + /* The fence registers are invalidated so clear them out */ + i915_gem_reset_fences(dev); +} + +/** + * This function clears the request list as sequence numbers are passed. + */ +static void +i915_gem_retire_requests_ring(struct intel_ring_buffer *ring) +{ + uint32_t seqno; + int i; + + if (list_empty(&ring->request_list)) + return; + + WARN_ON(i915_verify_lists(ring->dev)); + + seqno = ring->get_seqno(ring); + + for (i = 0; i < ARRAY_SIZE(ring->sync_seqno); i++) + if (seqno >= ring->sync_seqno[i]) + ring->sync_seqno[i] = 0; + + while (!list_empty(&ring->request_list)) { + struct drm_i915_gem_request *request; + + request = list_first_entry(&ring->request_list, + struct drm_i915_gem_request, + list); + + if (!i915_seqno_passed(seqno, request->seqno)) + break; + + trace_i915_gem_request_retire(ring, request->seqno); + + list_del(&request->list); + i915_gem_request_remove_from_client(request); + kfree(request); + } + + /* Move any buffers on the active list that are no longer referenced + * by the ringbuffer to the flushing/inactive lists as appropriate. + */ + while (!list_empty(&ring->active_list)) { + struct drm_i915_gem_object *obj; + + obj= list_first_entry(&ring->active_list, + struct drm_i915_gem_object, + ring_list); + + if (!i915_seqno_passed(seqno, obj->last_rendering_seqno)) + break; + + if (obj->base.write_domain != 0) + i915_gem_object_move_to_flushing(obj); + else + i915_gem_object_move_to_inactive(obj); + } + + if (unlikely(ring->trace_irq_seqno && + i915_seqno_passed(seqno, ring->trace_irq_seqno))) { + ring->irq_put(ring); + ring->trace_irq_seqno = 0; + } + + WARN_ON(i915_verify_lists(ring->dev)); +} + +void +i915_gem_retire_requests(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + int i; + + if (!list_empty(&dev_priv->mm.deferred_free_list)) { + struct drm_i915_gem_object *obj, *next; + + /* We must be careful that during unbind() we do not + * accidentally infinitely recurse into retire requests. + * Currently: + * retire -> free -> unbind -> wait -> retire_ring + */ + list_for_each_entry_safe(obj, next, + &dev_priv->mm.deferred_free_list, + mm_list) + i915_gem_free_object_tail(obj); + } + + for (i = 0; i < I915_NUM_RINGS; i++) + i915_gem_retire_requests_ring(&dev_priv->ring[i]); +} + +static void +i915_gem_retire_work_handler(struct work_struct *work) +{ + drm_i915_private_t *dev_priv; + struct drm_device *dev; + bool idle; + int i; + + dev_priv = container_of(work, drm_i915_private_t, + mm.retire_work.work); + dev = dev_priv->dev; + + /* Come back later if the device is busy... */ + if (!mutex_trylock(&dev->struct_mutex)) { + queue_delayed_work(dev_priv->wq, &dev_priv->mm.retire_work, HZ); + return; + } + + i915_gem_retire_requests(dev); + + /* Send a periodic flush down the ring so we don't hold onto GEM + * objects indefinitely. + */ + idle = true; + for (i = 0; i < I915_NUM_RINGS; i++) { + struct intel_ring_buffer *ring = &dev_priv->ring[i]; + + if (!list_empty(&ring->gpu_write_list)) { + struct drm_i915_gem_request *request; + int ret; + + ret = i915_gem_flush_ring(ring, + 0, I915_GEM_GPU_DOMAINS); + request = kzalloc(sizeof(*request), GFP_KERNEL); + if (ret || request == NULL || + i915_add_request(ring, NULL, request)) + kfree(request); + } + + idle &= list_empty(&ring->request_list); + } + + if (!dev_priv->mm.suspended && !idle) + queue_delayed_work(dev_priv->wq, &dev_priv->mm.retire_work, HZ); + + mutex_unlock(&dev->struct_mutex); +} + +/** + * Waits for a sequence number to be signaled, and cleans up the + * request and object lists appropriately for that event. + */ +int +i915_wait_request(struct intel_ring_buffer *ring, + uint32_t seqno) +{ + drm_i915_private_t *dev_priv = ring->dev->dev_private; + u32 ier; + int ret = 0; + + BUG_ON(seqno == 0); + + if (atomic_read(&dev_priv->mm.wedged)) { + struct completion *x = &dev_priv->error_completion; + bool recovery_complete; + unsigned long flags; + + /* Give the error handler a chance to run. */ + spin_lock_irqsave(&x->wait.lock, flags); + recovery_complete = x->done > 0; + spin_unlock_irqrestore(&x->wait.lock, flags); + + return recovery_complete ? -EIO : -EAGAIN; + } + + if (seqno == ring->outstanding_lazy_request) { + struct drm_i915_gem_request *request; + + request = kzalloc(sizeof(*request), GFP_KERNEL); + if (request == NULL) + return -ENOMEM; + + ret = i915_add_request(ring, NULL, request); + if (ret) { + kfree(request); + return ret; + } + + seqno = request->seqno; + } + + if (!i915_seqno_passed(ring->get_seqno(ring), seqno)) { + if (HAS_PCH_SPLIT(ring->dev)) + ier = I915_READ(DEIER) | I915_READ(GTIER); + else + ier = I915_READ(IER); + if (!ier) { + DRM_ERROR("something (likely vbetool) disabled " + "interrupts, re-enabling\n"); + ring->dev->driver->irq_preinstall(ring->dev); + ring->dev->driver->irq_postinstall(ring->dev); + } + + trace_i915_gem_request_wait_begin(ring, seqno); + + ring->waiting_seqno = seqno; + if (ring->irq_get(ring)) { + if (dev_priv->mm.interruptible) + ret = wait_event_interruptible(ring->irq_queue, + i915_seqno_passed(ring->get_seqno(ring), seqno) + || atomic_read(&dev_priv->mm.wedged)); + else + wait_event(ring->irq_queue, + i915_seqno_passed(ring->get_seqno(ring), seqno) + || atomic_read(&dev_priv->mm.wedged)); + + ring->irq_put(ring); + } else if (wait_for(i915_seqno_passed(ring->get_seqno(ring), + seqno) || + atomic_read(&dev_priv->mm.wedged), 3000)) + ret = -EBUSY; + ring->waiting_seqno = 0; + + trace_i915_gem_request_wait_end(ring, seqno); + } + if (atomic_read(&dev_priv->mm.wedged)) + ret = -EAGAIN; + + if (ret && ret != -ERESTARTSYS) + DRM_ERROR("%s returns %d (awaiting %d at %d, next %d)\n", + __func__, ret, seqno, ring->get_seqno(ring), + dev_priv->next_seqno); + + /* Directly dispatch request retiring. While we have the work queue + * to handle this, the waiter on a request often wants an associated + * buffer to have made it to the inactive list, and we would need + * a separate wait queue to handle that. + */ + if (ret == 0) + i915_gem_retire_requests_ring(ring); + + return ret; +} + +/** + * Ensures that all rendering to the object has completed and the object is + * safe to unbind from the GTT or access from the CPU. + */ +int +i915_gem_object_wait_rendering(struct drm_i915_gem_object *obj) +{ + int ret; + + /* This function only exists to support waiting for existing rendering, + * not for emitting required flushes. + */ + BUG_ON((obj->base.write_domain & I915_GEM_GPU_DOMAINS) != 0); + + /* If there is rendering queued on the buffer being evicted, wait for + * it. + */ + if (obj->active) { + ret = i915_wait_request(obj->ring, obj->last_rendering_seqno); + if (ret) + return ret; + } + + return 0; +} + +/** + * Unbinds an object from the GTT aperture. + */ +int +i915_gem_object_unbind(struct drm_i915_gem_object *obj) +{ + int ret = 0; + + if (obj->gtt_space == NULL) + return 0; + + if (obj->pin_count != 0) { + DRM_ERROR("Attempting to unbind pinned buffer\n"); + return -EINVAL; + } + + /* blow away mappings if mapped through GTT */ + i915_gem_release_mmap(obj); + + /* Move the object to the CPU domain to ensure that + * any possible CPU writes while it's not in the GTT + * are flushed when we go to remap it. This will + * also ensure that all pending GPU writes are finished + * before we unbind. + */ + ret = i915_gem_object_set_to_cpu_domain(obj, 1); + if (ret == -ERESTARTSYS) + return ret; + /* Continue on if we fail due to EIO, the GPU is hung so we + * should be safe and we need to cleanup or else we might + * cause memory corruption through use-after-free. + */ + if (ret) { + i915_gem_clflush_object(obj); + obj->base.read_domains = obj->base.write_domain = I915_GEM_DOMAIN_CPU; + } + + /* release the fence reg _after_ flushing */ + ret = i915_gem_object_put_fence(obj); + if (ret == -ERESTARTSYS) + return ret; + + trace_i915_gem_object_unbind(obj); + + i915_gem_gtt_unbind_object(obj); + i915_gem_object_put_pages_gtt(obj); + + list_del_init(&obj->gtt_list); + list_del_init(&obj->mm_list); + /* Avoid an unnecessary call to unbind on rebind. */ + obj->map_and_fenceable = true; + + drm_mm_put_block(obj->gtt_space); + obj->gtt_space = NULL; + obj->gtt_offset = 0; + + if (i915_gem_object_is_purgeable(obj)) + i915_gem_object_truncate(obj); + + return ret; +} + +int +i915_gem_flush_ring(struct intel_ring_buffer *ring, + uint32_t invalidate_domains, + uint32_t flush_domains) +{ + int ret; + + if (((invalidate_domains | flush_domains) & I915_GEM_GPU_DOMAINS) == 0) + return 0; + + trace_i915_gem_ring_flush(ring, invalidate_domains, flush_domains); + + ret = ring->flush(ring, invalidate_domains, flush_domains); + if (ret) + return ret; + + if (flush_domains & I915_GEM_GPU_DOMAINS) + i915_gem_process_flushing_list(ring, flush_domains); + + return 0; +} + +static int i915_ring_idle(struct intel_ring_buffer *ring) +{ + int ret; + + if (list_empty(&ring->gpu_write_list) && list_empty(&ring->active_list)) + return 0; + + if (!list_empty(&ring->gpu_write_list)) { + ret = i915_gem_flush_ring(ring, + I915_GEM_GPU_DOMAINS, I915_GEM_GPU_DOMAINS); + if (ret) + return ret; + } + + return i915_wait_request(ring, i915_gem_next_request_seqno(ring)); +} + +int +i915_gpu_idle(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + bool lists_empty; + int ret, i; + + lists_empty = (list_empty(&dev_priv->mm.flushing_list) && + list_empty(&dev_priv->mm.active_list)); + if (lists_empty) + return 0; + + /* Flush everything onto the inactive list. */ + for (i = 0; i < I915_NUM_RINGS; i++) { + ret = i915_ring_idle(&dev_priv->ring[i]); + if (ret) + return ret; + } + + return 0; +} + +static int sandybridge_write_fence_reg(struct drm_i915_gem_object *obj, + struct intel_ring_buffer *pipelined) +{ + struct drm_device *dev = obj->base.dev; + drm_i915_private_t *dev_priv = dev->dev_private; + u32 size = obj->gtt_space->size; + int regnum = obj->fence_reg; + uint64_t val; + + val = (uint64_t)((obj->gtt_offset + size - 4096) & + 0xfffff000) << 32; + val |= obj->gtt_offset & 0xfffff000; + val |= (uint64_t)((obj->stride / 128) - 1) << + SANDYBRIDGE_FENCE_PITCH_SHIFT; + + if (obj->tiling_mode == I915_TILING_Y) + val |= 1 << I965_FENCE_TILING_Y_SHIFT; + val |= I965_FENCE_REG_VALID; + + if (pipelined) { + int ret = intel_ring_begin(pipelined, 6); + if (ret) + return ret; + + intel_ring_emit(pipelined, MI_NOOP); + intel_ring_emit(pipelined, MI_LOAD_REGISTER_IMM(2)); + intel_ring_emit(pipelined, FENCE_REG_SANDYBRIDGE_0 + regnum*8); + intel_ring_emit(pipelined, (u32)val); + intel_ring_emit(pipelined, FENCE_REG_SANDYBRIDGE_0 + regnum*8 + 4); + intel_ring_emit(pipelined, (u32)(val >> 32)); + intel_ring_advance(pipelined); + } else + I915_WRITE64(FENCE_REG_SANDYBRIDGE_0 + regnum * 8, val); + + return 0; +} + +static int i965_write_fence_reg(struct drm_i915_gem_object *obj, + struct intel_ring_buffer *pipelined) +{ + struct drm_device *dev = obj->base.dev; + drm_i915_private_t *dev_priv = dev->dev_private; + u32 size = obj->gtt_space->size; + int regnum = obj->fence_reg; + uint64_t val; + + val = (uint64_t)((obj->gtt_offset + size - 4096) & + 0xfffff000) << 32; + val |= obj->gtt_offset & 0xfffff000; + val |= ((obj->stride / 128) - 1) << I965_FENCE_PITCH_SHIFT; + if (obj->tiling_mode == I915_TILING_Y) + val |= 1 << I965_FENCE_TILING_Y_SHIFT; + val |= I965_FENCE_REG_VALID; + + if (pipelined) { + int ret = intel_ring_begin(pipelined, 6); + if (ret) + return ret; + + intel_ring_emit(pipelined, MI_NOOP); + intel_ring_emit(pipelined, MI_LOAD_REGISTER_IMM(2)); + intel_ring_emit(pipelined, FENCE_REG_965_0 + regnum*8); + intel_ring_emit(pipelined, (u32)val); + intel_ring_emit(pipelined, FENCE_REG_965_0 + regnum*8 + 4); + intel_ring_emit(pipelined, (u32)(val >> 32)); + intel_ring_advance(pipelined); + } else + I915_WRITE64(FENCE_REG_965_0 + regnum * 8, val); + + return 0; +} + +static int i915_write_fence_reg(struct drm_i915_gem_object *obj, + struct intel_ring_buffer *pipelined) +{ + struct drm_device *dev = obj->base.dev; + drm_i915_private_t *dev_priv = dev->dev_private; + u32 size = obj->gtt_space->size; + u32 fence_reg, val, pitch_val; + int tile_width; + + if (WARN((obj->gtt_offset & ~I915_FENCE_START_MASK) || + (size & -size) != size || + (obj->gtt_offset & (size - 1)), + "object 0x%08x [fenceable? %d] not 1M or pot-size (0x%08x) aligned\n", + obj->gtt_offset, obj->map_and_fenceable, size)) + return -EINVAL; + + if (obj->tiling_mode == I915_TILING_Y && HAS_128_BYTE_Y_TILING(dev)) + tile_width = 128; + else + tile_width = 512; + + /* Note: pitch better be a power of two tile widths */ + pitch_val = obj->stride / tile_width; + pitch_val = ffs(pitch_val) - 1; + + val = obj->gtt_offset; + if (obj->tiling_mode == I915_TILING_Y) + val |= 1 << I830_FENCE_TILING_Y_SHIFT; + val |= I915_FENCE_SIZE_BITS(size); + val |= pitch_val << I830_FENCE_PITCH_SHIFT; + val |= I830_FENCE_REG_VALID; + + fence_reg = obj->fence_reg; + if (fence_reg < 8) + fence_reg = FENCE_REG_830_0 + fence_reg * 4; + else + fence_reg = FENCE_REG_945_8 + (fence_reg - 8) * 4; + + if (pipelined) { + int ret = intel_ring_begin(pipelined, 4); + if (ret) + return ret; + + intel_ring_emit(pipelined, MI_NOOP); + intel_ring_emit(pipelined, MI_LOAD_REGISTER_IMM(1)); + intel_ring_emit(pipelined, fence_reg); + intel_ring_emit(pipelined, val); + intel_ring_advance(pipelined); + } else + I915_WRITE(fence_reg, val); + + return 0; +} + +static int i830_write_fence_reg(struct drm_i915_gem_object *obj, + struct intel_ring_buffer *pipelined) +{ + struct drm_device *dev = obj->base.dev; + drm_i915_private_t *dev_priv = dev->dev_private; + u32 size = obj->gtt_space->size; + int regnum = obj->fence_reg; + uint32_t val; + uint32_t pitch_val; + + if (WARN((obj->gtt_offset & ~I830_FENCE_START_MASK) || + (size & -size) != size || + (obj->gtt_offset & (size - 1)), + "object 0x%08x not 512K or pot-size 0x%08x aligned\n", + obj->gtt_offset, size)) + return -EINVAL; + + pitch_val = obj->stride / 128; + pitch_val = ffs(pitch_val) - 1; + + val = obj->gtt_offset; + if (obj->tiling_mode == I915_TILING_Y) + val |= 1 << I830_FENCE_TILING_Y_SHIFT; + val |= I830_FENCE_SIZE_BITS(size); + val |= pitch_val << I830_FENCE_PITCH_SHIFT; + val |= I830_FENCE_REG_VALID; + + if (pipelined) { + int ret = intel_ring_begin(pipelined, 4); + if (ret) + return ret; + + intel_ring_emit(pipelined, MI_NOOP); + intel_ring_emit(pipelined, MI_LOAD_REGISTER_IMM(1)); + intel_ring_emit(pipelined, FENCE_REG_830_0 + regnum*4); + intel_ring_emit(pipelined, val); + intel_ring_advance(pipelined); + } else + I915_WRITE(FENCE_REG_830_0 + regnum * 4, val); + + return 0; +} + +static bool ring_passed_seqno(struct intel_ring_buffer *ring, u32 seqno) +{ + return i915_seqno_passed(ring->get_seqno(ring), seqno); +} + +static int +i915_gem_object_flush_fence(struct drm_i915_gem_object *obj, + struct intel_ring_buffer *pipelined) +{ + int ret; + + if (obj->fenced_gpu_access) { + if (obj->base.write_domain & I915_GEM_GPU_DOMAINS) { + ret = i915_gem_flush_ring(obj->last_fenced_ring, + 0, obj->base.write_domain); + if (ret) + return ret; + } + + obj->fenced_gpu_access = false; + } + + if (obj->last_fenced_seqno && pipelined != obj->last_fenced_ring) { + if (!ring_passed_seqno(obj->last_fenced_ring, + obj->last_fenced_seqno)) { + ret = i915_wait_request(obj->last_fenced_ring, + obj->last_fenced_seqno); + if (ret) + return ret; + } + + obj->last_fenced_seqno = 0; + obj->last_fenced_ring = NULL; + } + + /* Ensure that all CPU reads are completed before installing a fence + * and all writes before removing the fence. + */ + if (obj->base.read_domains & I915_GEM_DOMAIN_GTT) + mb(); + + return 0; +} + +int +i915_gem_object_put_fence(struct drm_i915_gem_object *obj) +{ + int ret; + + if (obj->tiling_mode) + i915_gem_release_mmap(obj); + + ret = i915_gem_object_flush_fence(obj, NULL); + if (ret) + return ret; + + if (obj->fence_reg != I915_FENCE_REG_NONE) { + struct drm_i915_private *dev_priv = obj->base.dev->dev_private; + i915_gem_clear_fence_reg(obj->base.dev, + &dev_priv->fence_regs[obj->fence_reg]); + + obj->fence_reg = I915_FENCE_REG_NONE; + } + + return 0; +} + +static struct drm_i915_fence_reg * +i915_find_fence_reg(struct drm_device *dev, + struct intel_ring_buffer *pipelined) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_i915_fence_reg *reg, *first, *avail; + int i; + + /* First try to find a free reg */ + avail = NULL; + for (i = dev_priv->fence_reg_start; i < dev_priv->num_fence_regs; i++) { + reg = &dev_priv->fence_regs[i]; + if (!reg->obj) + return reg; + + if (!reg->obj->pin_count) + avail = reg; + } + + if (avail == NULL) + return NULL; + + /* None available, try to steal one or wait for a user to finish */ + avail = first = NULL; + list_for_each_entry(reg, &dev_priv->mm.fence_list, lru_list) { + if (reg->obj->pin_count) + continue; + + if (first == NULL) + first = reg; + + if (!pipelined || + !reg->obj->last_fenced_ring || + reg->obj->last_fenced_ring == pipelined) { + avail = reg; + break; + } + } + + if (avail == NULL) + avail = first; + + return avail; +} + +/** + * i915_gem_object_get_fence - set up a fence reg for an object + * @obj: object to map through a fence reg + * @pipelined: ring on which to queue the change, or NULL for CPU access + * @interruptible: must we wait uninterruptibly for the register to retire? + * + * When mapping objects through the GTT, userspace wants to be able to write + * to them without having to worry about swizzling if the object is tiled. + * + * This function walks the fence regs looking for a free one for @obj, + * stealing one if it can't find any. + * + * It then sets up the reg based on the object's properties: address, pitch + * and tiling format. + */ +int +i915_gem_object_get_fence(struct drm_i915_gem_object *obj, + struct intel_ring_buffer *pipelined) +{ + struct drm_device *dev = obj->base.dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_i915_fence_reg *reg; + int ret; + + /* XXX disable pipelining. There are bugs. Shocking. */ + pipelined = NULL; + + /* Just update our place in the LRU if our fence is getting reused. */ + if (obj->fence_reg != I915_FENCE_REG_NONE) { + reg = &dev_priv->fence_regs[obj->fence_reg]; + list_move_tail(®->lru_list, &dev_priv->mm.fence_list); + + if (obj->tiling_changed) { + ret = i915_gem_object_flush_fence(obj, pipelined); + if (ret) + return ret; + + if (!obj->fenced_gpu_access && !obj->last_fenced_seqno) + pipelined = NULL; + + if (pipelined) { + reg->setup_seqno = + i915_gem_next_request_seqno(pipelined); + obj->last_fenced_seqno = reg->setup_seqno; + obj->last_fenced_ring = pipelined; + } + + goto update; + } + + if (!pipelined) { + if (reg->setup_seqno) { + if (!ring_passed_seqno(obj->last_fenced_ring, + reg->setup_seqno)) { + ret = i915_wait_request(obj->last_fenced_ring, + reg->setup_seqno); + if (ret) + return ret; + } + + reg->setup_seqno = 0; + } + } else if (obj->last_fenced_ring && + obj->last_fenced_ring != pipelined) { + ret = i915_gem_object_flush_fence(obj, pipelined); + if (ret) + return ret; + } + + return 0; + } + + reg = i915_find_fence_reg(dev, pipelined); + if (reg == NULL) + return -ENOSPC; + + ret = i915_gem_object_flush_fence(obj, pipelined); + if (ret) + return ret; + + if (reg->obj) { + struct drm_i915_gem_object *old = reg->obj; + + drm_gem_object_reference(&old->base); + + if (old->tiling_mode) + i915_gem_release_mmap(old); + + ret = i915_gem_object_flush_fence(old, pipelined); + if (ret) { + drm_gem_object_unreference(&old->base); + return ret; + } + + if (old->last_fenced_seqno == 0 && obj->last_fenced_seqno == 0) + pipelined = NULL; + + old->fence_reg = I915_FENCE_REG_NONE; + old->last_fenced_ring = pipelined; + old->last_fenced_seqno = + pipelined ? i915_gem_next_request_seqno(pipelined) : 0; + + drm_gem_object_unreference(&old->base); + } else if (obj->last_fenced_seqno == 0) + pipelined = NULL; + + reg->obj = obj; + list_move_tail(®->lru_list, &dev_priv->mm.fence_list); + obj->fence_reg = reg - dev_priv->fence_regs; + obj->last_fenced_ring = pipelined; + + reg->setup_seqno = + pipelined ? i915_gem_next_request_seqno(pipelined) : 0; + obj->last_fenced_seqno = reg->setup_seqno; + +update: + obj->tiling_changed = false; + switch (INTEL_INFO(dev)->gen) { + case 7: + case 6: + ret = sandybridge_write_fence_reg(obj, pipelined); + break; + case 5: + case 4: + ret = i965_write_fence_reg(obj, pipelined); + break; + case 3: + ret = i915_write_fence_reg(obj, pipelined); + break; + case 2: + ret = i830_write_fence_reg(obj, pipelined); + break; + } + + return ret; +} + +/** + * i915_gem_clear_fence_reg - clear out fence register info + * @obj: object to clear + * + * Zeroes out the fence register itself and clears out the associated + * data structures in dev_priv and obj. + */ +static void +i915_gem_clear_fence_reg(struct drm_device *dev, + struct drm_i915_fence_reg *reg) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + uint32_t fence_reg = reg - dev_priv->fence_regs; + + switch (INTEL_INFO(dev)->gen) { + case 7: + case 6: + I915_WRITE64(FENCE_REG_SANDYBRIDGE_0 + fence_reg*8, 0); + break; + case 5: + case 4: + I915_WRITE64(FENCE_REG_965_0 + fence_reg*8, 0); + break; + case 3: + if (fence_reg >= 8) + fence_reg = FENCE_REG_945_8 + (fence_reg - 8) * 4; + else + case 2: + fence_reg = FENCE_REG_830_0 + fence_reg * 4; + + I915_WRITE(fence_reg, 0); + break; + } + + list_del_init(®->lru_list); + reg->obj = NULL; + reg->setup_seqno = 0; +} + +/** + * Finds free space in the GTT aperture and binds the object there. + */ +static int +i915_gem_object_bind_to_gtt(struct drm_i915_gem_object *obj, + unsigned alignment, + bool map_and_fenceable) +{ + struct drm_device *dev = obj->base.dev; + drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_mm_node *free_space; + gfp_t gfpmask = __GFP_NORETRY | __GFP_NOWARN; + u32 size, fence_size, fence_alignment, unfenced_alignment; + bool mappable, fenceable; + int ret; + + if (obj->madv != I915_MADV_WILLNEED) { + DRM_ERROR("Attempting to bind a purgeable object\n"); + return -EINVAL; + } + + fence_size = i915_gem_get_gtt_size(dev, + obj->base.size, + obj->tiling_mode); + fence_alignment = i915_gem_get_gtt_alignment(dev, + obj->base.size, + obj->tiling_mode); + unfenced_alignment = + i915_gem_get_unfenced_gtt_alignment(dev, + obj->base.size, + obj->tiling_mode); + + if (alignment == 0) + alignment = map_and_fenceable ? fence_alignment : + unfenced_alignment; + if (map_and_fenceable && alignment & (fence_alignment - 1)) { + DRM_ERROR("Invalid object alignment requested %u\n", alignment); + return -EINVAL; + } + + size = map_and_fenceable ? fence_size : obj->base.size; + + /* If the object is bigger than the entire aperture, reject it early + * before evicting everything in a vain attempt to find space. + */ + if (obj->base.size > + (map_and_fenceable ? dev_priv->mm.gtt_mappable_end : dev_priv->mm.gtt_total)) { + DRM_ERROR("Attempting to bind an object larger than the aperture\n"); + return -E2BIG; + } + + search_free: + if (map_and_fenceable) + free_space = + drm_mm_search_free_in_range(&dev_priv->mm.gtt_space, + size, alignment, 0, + dev_priv->mm.gtt_mappable_end, + 0); + else + free_space = drm_mm_search_free(&dev_priv->mm.gtt_space, + size, alignment, 0); + + if (free_space != NULL) { + if (map_and_fenceable) + obj->gtt_space = + drm_mm_get_block_range_generic(free_space, + size, alignment, 0, + dev_priv->mm.gtt_mappable_end, + 0); + else + obj->gtt_space = + drm_mm_get_block(free_space, size, alignment); + } + if (obj->gtt_space == NULL) { + /* If the gtt is empty and we're still having trouble + * fitting our object in, we're out of memory. + */ + ret = i915_gem_evict_something(dev, size, alignment, + map_and_fenceable); + if (ret) + return ret; + + goto search_free; + } + + ret = i915_gem_object_get_pages_gtt(obj, gfpmask); + if (ret) { + drm_mm_put_block(obj->gtt_space); + obj->gtt_space = NULL; + + if (ret == -ENOMEM) { + /* first try to reclaim some memory by clearing the GTT */ + ret = i915_gem_evict_everything(dev, false); + if (ret) { + /* now try to shrink everyone else */ + if (gfpmask) { + gfpmask = 0; + goto search_free; + } + + return -ENOMEM; + } + + goto search_free; + } + + return ret; + } + + ret = i915_gem_gtt_bind_object(obj); + if (ret) { + i915_gem_object_put_pages_gtt(obj); + drm_mm_put_block(obj->gtt_space); + obj->gtt_space = NULL; + + if (i915_gem_evict_everything(dev, false)) + return ret; + + goto search_free; + } + + list_add_tail(&obj->gtt_list, &dev_priv->mm.gtt_list); + list_add_tail(&obj->mm_list, &dev_priv->mm.inactive_list); + + /* Assert that the object is not currently in any GPU domain. As it + * wasn't in the GTT, there shouldn't be any way it could have been in + * a GPU cache + */ + BUG_ON(obj->base.read_domains & I915_GEM_GPU_DOMAINS); + BUG_ON(obj->base.write_domain & I915_GEM_GPU_DOMAINS); + + obj->gtt_offset = obj->gtt_space->start; + + fenceable = + obj->gtt_space->size == fence_size && + (obj->gtt_space->start & (fence_alignment -1)) == 0; + + mappable = + obj->gtt_offset + obj->base.size <= dev_priv->mm.gtt_mappable_end; + + obj->map_and_fenceable = mappable && fenceable; + + trace_i915_gem_object_bind(obj, map_and_fenceable); + return 0; +} + +void +i915_gem_clflush_object(struct drm_i915_gem_object *obj) +{ + /* If we don't have a page list set up, then we're not pinned + * to GPU, and we can ignore the cache flush because it'll happen + * again at bind time. + */ + if (obj->pages == NULL) + return; + + /* If the GPU is snooping the contents of the CPU cache, + * we do not need to manually clear the CPU cache lines. However, + * the caches are only snooped when the render cache is + * flushed/invalidated. As we always have to emit invalidations + * and flushes when moving into and out of the RENDER domain, correct + * snooping behaviour occurs naturally as the result of our domain + * tracking. + */ + if (obj->cache_level != I915_CACHE_NONE) + return; + + trace_i915_gem_object_clflush(obj); + + drm_clflush_pages(obj->pages, obj->base.size / PAGE_SIZE); +} + +/** Flushes any GPU write domain for the object if it's dirty. */ +static int +i915_gem_object_flush_gpu_write_domain(struct drm_i915_gem_object *obj) +{ + if ((obj->base.write_domain & I915_GEM_GPU_DOMAINS) == 0) + return 0; + + /* Queue the GPU write cache flushing we need. */ + return i915_gem_flush_ring(obj->ring, 0, obj->base.write_domain); +} + +/** Flushes the GTT write domain for the object if it's dirty. */ +static void +i915_gem_object_flush_gtt_write_domain(struct drm_i915_gem_object *obj) +{ + uint32_t old_write_domain; + + if (obj->base.write_domain != I915_GEM_DOMAIN_GTT) + return; + + /* No actual flushing is required for the GTT write domain. Writes + * to it immediately go to main memory as far as we know, so there's + * no chipset flush. It also doesn't land in render cache. + * + * However, we do have to enforce the order so that all writes through + * the GTT land before any writes to the device, such as updates to + * the GATT itself. + */ + wmb(); + + old_write_domain = obj->base.write_domain; + obj->base.write_domain = 0; + + trace_i915_gem_object_change_domain(obj, + obj->base.read_domains, + old_write_domain); +} + +/** Flushes the CPU write domain for the object if it's dirty. */ +static void +i915_gem_object_flush_cpu_write_domain(struct drm_i915_gem_object *obj) +{ + uint32_t old_write_domain; + + if (obj->base.write_domain != I915_GEM_DOMAIN_CPU) + return; + + i915_gem_clflush_object(obj); + intel_gtt_chipset_flush(); + old_write_domain = obj->base.write_domain; + obj->base.write_domain = 0; + + trace_i915_gem_object_change_domain(obj, + obj->base.read_domains, + old_write_domain); +} + +/** + * Moves a single object to the GTT read, and possibly write domain. + * + * This function returns when the move is complete, including waiting on + * flushes to occur. + */ +int +i915_gem_object_set_to_gtt_domain(struct drm_i915_gem_object *obj, bool write) +{ + uint32_t old_write_domain, old_read_domains; + int ret; + + /* Not valid to be called on unbound objects. */ + if (obj->gtt_space == NULL) + return -EINVAL; + + if (obj->base.write_domain == I915_GEM_DOMAIN_GTT) + return 0; + + ret = i915_gem_object_flush_gpu_write_domain(obj); + if (ret) + return ret; + + if (obj->pending_gpu_write || write) { + ret = i915_gem_object_wait_rendering(obj); + if (ret) + return ret; + } + + i915_gem_object_flush_cpu_write_domain(obj); + + old_write_domain = obj->base.write_domain; + old_read_domains = obj->base.read_domains; + + /* It should now be out of any other write domains, and we can update + * the domain values for our changes. + */ + BUG_ON((obj->base.write_domain & ~I915_GEM_DOMAIN_GTT) != 0); + obj->base.read_domains |= I915_GEM_DOMAIN_GTT; + if (write) { + obj->base.read_domains = I915_GEM_DOMAIN_GTT; + obj->base.write_domain = I915_GEM_DOMAIN_GTT; + obj->dirty = 1; + } + + trace_i915_gem_object_change_domain(obj, + old_read_domains, + old_write_domain); + + return 0; +} + +/* + * Prepare buffer for display plane. Use uninterruptible for possible flush + * wait, as in modesetting process we're not supposed to be interrupted. + */ +int +i915_gem_object_set_to_display_plane(struct drm_i915_gem_object *obj, + struct intel_ring_buffer *pipelined) +{ + uint32_t old_read_domains; + int ret; + + /* Not valid to be called on unbound objects. */ + if (obj->gtt_space == NULL) + return -EINVAL; + + ret = i915_gem_object_flush_gpu_write_domain(obj); + if (ret) + return ret; + + + /* Currently, we are always called from an non-interruptible context. */ + if (pipelined != obj->ring) { + ret = i915_gem_object_wait_rendering(obj); + if (ret) + return ret; + } + + i915_gem_object_flush_cpu_write_domain(obj); + + old_read_domains = obj->base.read_domains; + obj->base.read_domains |= I915_GEM_DOMAIN_GTT; + + trace_i915_gem_object_change_domain(obj, + old_read_domains, + obj->base.write_domain); + + return 0; +} + +int +i915_gem_object_flush_gpu(struct drm_i915_gem_object *obj) +{ + int ret; + + if (!obj->active) + return 0; + + if (obj->base.write_domain & I915_GEM_GPU_DOMAINS) { + ret = i915_gem_flush_ring(obj->ring, 0, obj->base.write_domain); + if (ret) + return ret; + } + + return i915_gem_object_wait_rendering(obj); +} + +/** + * Moves a single object to the CPU read, and possibly write domain. + * + * This function returns when the move is complete, including waiting on + * flushes to occur. + */ +static int +i915_gem_object_set_to_cpu_domain(struct drm_i915_gem_object *obj, bool write) +{ + uint32_t old_write_domain, old_read_domains; + int ret; + + if (obj->base.write_domain == I915_GEM_DOMAIN_CPU) + return 0; + + ret = i915_gem_object_flush_gpu_write_domain(obj); + if (ret) + return ret; + + ret = i915_gem_object_wait_rendering(obj); + if (ret) + return ret; + + i915_gem_object_flush_gtt_write_domain(obj); + + /* If we have a partially-valid cache of the object in the CPU, + * finish invalidating it and free the per-page flags. + */ + i915_gem_object_set_to_full_cpu_read_domain(obj); + + old_write_domain = obj->base.write_domain; + old_read_domains = obj->base.read_domains; + + /* Flush the CPU cache if it's still invalid. */ + if ((obj->base.read_domains & I915_GEM_DOMAIN_CPU) == 0) { + i915_gem_clflush_object(obj); + + obj->base.read_domains |= I915_GEM_DOMAIN_CPU; + } + + /* It should now be out of any other write domains, and we can update + * the domain values for our changes. + */ + BUG_ON((obj->base.write_domain & ~I915_GEM_DOMAIN_CPU) != 0); + + /* If we're writing through the CPU, then the GPU read domains will + * need to be invalidated at next use. + */ + if (write) { + obj->base.read_domains = I915_GEM_DOMAIN_CPU; + obj->base.write_domain = I915_GEM_DOMAIN_CPU; + } + + trace_i915_gem_object_change_domain(obj, + old_read_domains, + old_write_domain); + + return 0; +} + +/** + * Moves the object from a partially CPU read to a full one. + * + * Note that this only resolves i915_gem_object_set_cpu_read_domain_range(), + * and doesn't handle transitioning from !(read_domains & I915_GEM_DOMAIN_CPU). + */ +static void +i915_gem_object_set_to_full_cpu_read_domain(struct drm_i915_gem_object *obj) +{ + if (!obj->page_cpu_valid) + return; + + /* If we're partially in the CPU read domain, finish moving it in. + */ + if (obj->base.read_domains & I915_GEM_DOMAIN_CPU) { + int i; + + for (i = 0; i <= (obj->base.size - 1) / PAGE_SIZE; i++) { + if (obj->page_cpu_valid[i]) + continue; + drm_clflush_pages(obj->pages + i, 1); + } + } + + /* Free the page_cpu_valid mappings which are now stale, whether + * or not we've got I915_GEM_DOMAIN_CPU. + */ + kfree(obj->page_cpu_valid); + obj->page_cpu_valid = NULL; +} + +/** + * Set the CPU read domain on a range of the object. + * + * The object ends up with I915_GEM_DOMAIN_CPU in its read flags although it's + * not entirely valid. The page_cpu_valid member of the object flags which + * pages have been flushed, and will be respected by + * i915_gem_object_set_to_cpu_domain() if it's called on to get a valid mapping + * of the whole object. + * + * This function returns when the move is complete, including waiting on + * flushes to occur. + */ +static int +i915_gem_object_set_cpu_read_domain_range(struct drm_i915_gem_object *obj, + uint64_t offset, uint64_t size) +{ + uint32_t old_read_domains; + int i, ret; + + if (offset == 0 && size == obj->base.size) + return i915_gem_object_set_to_cpu_domain(obj, 0); + + ret = i915_gem_object_flush_gpu_write_domain(obj); + if (ret) + return ret; + + ret = i915_gem_object_wait_rendering(obj); + if (ret) + return ret; + + i915_gem_object_flush_gtt_write_domain(obj); + + /* If we're already fully in the CPU read domain, we're done. */ + if (obj->page_cpu_valid == NULL && + (obj->base.read_domains & I915_GEM_DOMAIN_CPU) != 0) + return 0; + + /* Otherwise, create/clear the per-page CPU read domain flag if we're + * newly adding I915_GEM_DOMAIN_CPU + */ + if (obj->page_cpu_valid == NULL) { + obj->page_cpu_valid = kzalloc(obj->base.size / PAGE_SIZE, + GFP_KERNEL); + if (obj->page_cpu_valid == NULL) + return -ENOMEM; + } else if ((obj->base.read_domains & I915_GEM_DOMAIN_CPU) == 0) + memset(obj->page_cpu_valid, 0, obj->base.size / PAGE_SIZE); + + /* Flush the cache on any pages that are still invalid from the CPU's + * perspective. + */ + for (i = offset / PAGE_SIZE; i <= (offset + size - 1) / PAGE_SIZE; + i++) { + if (obj->page_cpu_valid[i]) + continue; + + drm_clflush_pages(obj->pages + i, 1); + + obj->page_cpu_valid[i] = 1; + } + + /* It should now be out of any other write domains, and we can update + * the domain values for our changes. + */ + BUG_ON((obj->base.write_domain & ~I915_GEM_DOMAIN_CPU) != 0); + + old_read_domains = obj->base.read_domains; + obj->base.read_domains |= I915_GEM_DOMAIN_CPU; + + trace_i915_gem_object_change_domain(obj, + old_read_domains, + obj->base.write_domain); + + return 0; +} + +/* Throttle our rendering by waiting until the ring has completed our requests + * emitted over 20 msec ago. + * + * Note that if we were to use the current jiffies each time around the loop, + * we wouldn't escape the function with any frames outstanding if the time to + * render a frame was over 20ms. + * + * This should get us reasonable parallelism between CPU and GPU but also + * relatively low latency when blocking on a particular request to finish. + */ +static int +i915_gem_ring_throttle(struct drm_device *dev, struct drm_file *file) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_i915_file_private *file_priv = file->driver_priv; + unsigned long recent_enough = jiffies - msecs_to_jiffies(20); + struct drm_i915_gem_request *request; + struct intel_ring_buffer *ring = NULL; + u32 seqno = 0; + int ret; + + if (atomic_read(&dev_priv->mm.wedged)) + return -EIO; + + spin_lock(&file_priv->mm.lock); + list_for_each_entry(request, &file_priv->mm.request_list, client_list) { + if (time_after_eq(request->emitted_jiffies, recent_enough)) + break; + + ring = request->ring; + seqno = request->seqno; + } + spin_unlock(&file_priv->mm.lock); + + if (seqno == 0) + return 0; + + ret = 0; + if (!i915_seqno_passed(ring->get_seqno(ring), seqno)) { + /* And wait for the seqno passing without holding any locks and + * causing extra latency for others. This is safe as the irq + * generation is designed to be run atomically and so is + * lockless. + */ + if (ring->irq_get(ring)) { + ret = wait_event_interruptible(ring->irq_queue, + i915_seqno_passed(ring->get_seqno(ring), seqno) + || atomic_read(&dev_priv->mm.wedged)); + ring->irq_put(ring); + + if (ret == 0 && atomic_read(&dev_priv->mm.wedged)) + ret = -EIO; + } + } + + if (ret == 0) + queue_delayed_work(dev_priv->wq, &dev_priv->mm.retire_work, 0); + + return ret; +} + +int +i915_gem_object_pin(struct drm_i915_gem_object *obj, + uint32_t alignment, + bool map_and_fenceable) +{ + struct drm_device *dev = obj->base.dev; + struct drm_i915_private *dev_priv = dev->dev_private; + int ret; + + BUG_ON(obj->pin_count == DRM_I915_GEM_OBJECT_MAX_PIN_COUNT); + WARN_ON(i915_verify_lists(dev)); + + if (obj->gtt_space != NULL) { + if ((alignment && obj->gtt_offset & (alignment - 1)) || + (map_and_fenceable && !obj->map_and_fenceable)) { + WARN(obj->pin_count, + "bo is already pinned with incorrect alignment:" + " offset=%x, req.alignment=%x, req.map_and_fenceable=%d," + " obj->map_and_fenceable=%d\n", + obj->gtt_offset, alignment, + map_and_fenceable, + obj->map_and_fenceable); + ret = i915_gem_object_unbind(obj); + if (ret) + return ret; + } + } + + if (obj->gtt_space == NULL) { + ret = i915_gem_object_bind_to_gtt(obj, alignment, + map_and_fenceable); + if (ret) + return ret; + } + + if (obj->pin_count++ == 0) { + if (!obj->active) + list_move_tail(&obj->mm_list, + &dev_priv->mm.pinned_list); + } + obj->pin_mappable |= map_and_fenceable; + + WARN_ON(i915_verify_lists(dev)); + return 0; +} + +void +i915_gem_object_unpin(struct drm_i915_gem_object *obj) +{ + struct drm_device *dev = obj->base.dev; + drm_i915_private_t *dev_priv = dev->dev_private; + + WARN_ON(i915_verify_lists(dev)); + BUG_ON(obj->pin_count == 0); + BUG_ON(obj->gtt_space == NULL); + + if (--obj->pin_count == 0) { + if (!obj->active) + list_move_tail(&obj->mm_list, + &dev_priv->mm.inactive_list); + obj->pin_mappable = false; + } + WARN_ON(i915_verify_lists(dev)); +} + +int +i915_gem_pin_ioctl(struct drm_device *dev, void *data, + struct drm_file *file) +{ + struct drm_i915_gem_pin *args = data; + struct drm_i915_gem_object *obj; + int ret; + + ret = i915_mutex_lock_interruptible(dev); + if (ret) + return ret; + + obj = to_intel_bo(drm_gem_object_lookup(dev, file, args->handle)); + if (&obj->base == NULL) { + ret = -ENOENT; + goto unlock; + } + + if (obj->madv != I915_MADV_WILLNEED) { + DRM_ERROR("Attempting to pin a purgeable buffer\n"); + ret = -EINVAL; + goto out; + } + + if (obj->pin_filp != NULL && obj->pin_filp != file) { + DRM_ERROR("Already pinned in i915_gem_pin_ioctl(): %d\n", + args->handle); + ret = -EINVAL; + goto out; + } + + obj->user_pin_count++; + obj->pin_filp = file; + if (obj->user_pin_count == 1) { + ret = i915_gem_object_pin(obj, args->alignment, true); + if (ret) + goto out; + } + + /* XXX - flush the CPU caches for pinned objects + * as the X server doesn't manage domains yet + */ + i915_gem_object_flush_cpu_write_domain(obj); + args->offset = obj->gtt_offset; +out: + drm_gem_object_unreference(&obj->base); +unlock: + mutex_unlock(&dev->struct_mutex); + return ret; +} + +int +i915_gem_unpin_ioctl(struct drm_device *dev, void *data, + struct drm_file *file) +{ + struct drm_i915_gem_pin *args = data; + struct drm_i915_gem_object *obj; + int ret; + + ret = i915_mutex_lock_interruptible(dev); + if (ret) + return ret; + + obj = to_intel_bo(drm_gem_object_lookup(dev, file, args->handle)); + if (&obj->base == NULL) { + ret = -ENOENT; + goto unlock; + } + + if (obj->pin_filp != file) { + DRM_ERROR("Not pinned by caller in i915_gem_pin_ioctl(): %d\n", + args->handle); + ret = -EINVAL; + goto out; + } + obj->user_pin_count--; + if (obj->user_pin_count == 0) { + obj->pin_filp = NULL; + i915_gem_object_unpin(obj); + } + +out: + drm_gem_object_unreference(&obj->base); +unlock: + mutex_unlock(&dev->struct_mutex); + return ret; +} + +int +i915_gem_busy_ioctl(struct drm_device *dev, void *data, + struct drm_file *file) +{ + struct drm_i915_gem_busy *args = data; + struct drm_i915_gem_object *obj; + int ret; + + ret = i915_mutex_lock_interruptible(dev); + if (ret) + return ret; + + obj = to_intel_bo(drm_gem_object_lookup(dev, file, args->handle)); + if (&obj->base == NULL) { + ret = -ENOENT; + goto unlock; + } + + /* Count all active objects as busy, even if they are currently not used + * by the gpu. Users of this interface expect objects to eventually + * become non-busy without any further actions, therefore emit any + * necessary flushes here. + */ + args->busy = obj->active; + if (args->busy) { + /* Unconditionally flush objects, even when the gpu still uses this + * object. Userspace calling this function indicates that it wants to + * use this buffer rather sooner than later, so issuing the required + * flush earlier is beneficial. + */ + if (obj->base.write_domain & I915_GEM_GPU_DOMAINS) { + ret = i915_gem_flush_ring(obj->ring, + 0, obj->base.write_domain); + } else if (obj->ring->outstanding_lazy_request == + obj->last_rendering_seqno) { + struct drm_i915_gem_request *request; + + /* This ring is not being cleared by active usage, + * so emit a request to do so. + */ + request = kzalloc(sizeof(*request), GFP_KERNEL); + if (request) + ret = i915_add_request(obj->ring, NULL,request); + else + ret = -ENOMEM; + } + + /* Update the active list for the hardware's current position. + * Otherwise this only updates on a delayed timer or when irqs + * are actually unmasked, and our working set ends up being + * larger than required. + */ + i915_gem_retire_requests_ring(obj->ring); + + args->busy = obj->active; + } + + drm_gem_object_unreference(&obj->base); +unlock: + mutex_unlock(&dev->struct_mutex); + return ret; +} + +int +i915_gem_throttle_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + return i915_gem_ring_throttle(dev, file_priv); +} + +int +i915_gem_madvise_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_i915_gem_madvise *args = data; + struct drm_i915_gem_object *obj; + int ret; + + switch (args->madv) { + case I915_MADV_DONTNEED: + case I915_MADV_WILLNEED: + break; + default: + return -EINVAL; + } + + ret = i915_mutex_lock_interruptible(dev); + if (ret) + return ret; + + obj = to_intel_bo(drm_gem_object_lookup(dev, file_priv, args->handle)); + if (&obj->base == NULL) { + ret = -ENOENT; + goto unlock; + } + + if (obj->pin_count) { + ret = -EINVAL; + goto out; + } + + if (obj->madv != __I915_MADV_PURGED) + obj->madv = args->madv; + + /* if the object is no longer bound, discard its backing storage */ + if (i915_gem_object_is_purgeable(obj) && + obj->gtt_space == NULL) + i915_gem_object_truncate(obj); + + args->retained = obj->madv != __I915_MADV_PURGED; + +out: + drm_gem_object_unreference(&obj->base); +unlock: + mutex_unlock(&dev->struct_mutex); + return ret; +} + +struct drm_i915_gem_object *i915_gem_alloc_object(struct drm_device *dev, + size_t size) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_i915_gem_object *obj; + struct address_space *mapping; + + obj = kzalloc(sizeof(*obj), GFP_KERNEL); + if (obj == NULL) + return NULL; + + if (drm_gem_object_init(dev, &obj->base, size) != 0) { + kfree(obj); + return NULL; + } + + mapping = obj->base.filp->f_path.dentry->d_inode->i_mapping; + mapping_set_gfp_mask(mapping, GFP_HIGHUSER | __GFP_RECLAIMABLE); + + i915_gem_info_add_obj(dev_priv, size); + + obj->base.write_domain = I915_GEM_DOMAIN_CPU; + obj->base.read_domains = I915_GEM_DOMAIN_CPU; + + obj->cache_level = I915_CACHE_NONE; + obj->base.driver_private = NULL; + obj->fence_reg = I915_FENCE_REG_NONE; + INIT_LIST_HEAD(&obj->mm_list); + INIT_LIST_HEAD(&obj->gtt_list); + INIT_LIST_HEAD(&obj->ring_list); + INIT_LIST_HEAD(&obj->exec_list); + INIT_LIST_HEAD(&obj->gpu_write_list); + obj->madv = I915_MADV_WILLNEED; + /* Avoid an unnecessary call to unbind on the first bind. */ + obj->map_and_fenceable = true; + + return obj; +} + +int i915_gem_init_object(struct drm_gem_object *obj) +{ + BUG(); + + return 0; +} + +static void i915_gem_free_object_tail(struct drm_i915_gem_object *obj) +{ + struct drm_device *dev = obj->base.dev; + drm_i915_private_t *dev_priv = dev->dev_private; + int ret; + + ret = i915_gem_object_unbind(obj); + if (ret == -ERESTARTSYS) { + list_move(&obj->mm_list, + &dev_priv->mm.deferred_free_list); + return; + } + + trace_i915_gem_object_destroy(obj); + + if (obj->base.map_list.map) + i915_gem_free_mmap_offset(obj); + + drm_gem_object_release(&obj->base); + i915_gem_info_remove_obj(dev_priv, obj->base.size); + + kfree(obj->page_cpu_valid); + kfree(obj->bit_17); + kfree(obj); +} + +void i915_gem_free_object(struct drm_gem_object *gem_obj) +{ + struct drm_i915_gem_object *obj = to_intel_bo(gem_obj); + struct drm_device *dev = obj->base.dev; + + while (obj->pin_count > 0) + i915_gem_object_unpin(obj); + + if (obj->phys_obj) + i915_gem_detach_phys_object(dev, obj); + + i915_gem_free_object_tail(obj); +} + +int +i915_gem_idle(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + int ret; + + mutex_lock(&dev->struct_mutex); + + if (dev_priv->mm.suspended) { + mutex_unlock(&dev->struct_mutex); + return 0; + } + + ret = i915_gpu_idle(dev); + if (ret) { + mutex_unlock(&dev->struct_mutex); + return ret; + } + + /* Under UMS, be paranoid and evict. */ + if (!drm_core_check_feature(dev, DRIVER_MODESET)) { + ret = i915_gem_evict_inactive(dev, false); + if (ret) { + mutex_unlock(&dev->struct_mutex); + return ret; + } + } + + i915_gem_reset_fences(dev); + + /* Hack! Don't let anybody do execbuf while we don't control the chip. + * We need to replace this with a semaphore, or something. + * And not confound mm.suspended! + */ + dev_priv->mm.suspended = 1; + del_timer_sync(&dev_priv->hangcheck_timer); + + i915_kernel_lost_context(dev); + i915_gem_cleanup_ringbuffer(dev); + + mutex_unlock(&dev->struct_mutex); + + /* Cancel the retire work handler, which should be idle now. */ + cancel_delayed_work_sync(&dev_priv->mm.retire_work); + + return 0; +} + +int +i915_gem_init_ringbuffer(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + int ret; + + ret = intel_init_render_ring_buffer(dev); + if (ret) + return ret; + + if (HAS_BSD(dev)) { + ret = intel_init_bsd_ring_buffer(dev); + if (ret) + goto cleanup_render_ring; + } + + if (HAS_BLT(dev)) { + ret = intel_init_blt_ring_buffer(dev); + if (ret) + goto cleanup_bsd_ring; + } + + dev_priv->next_seqno = 1; + + return 0; + +cleanup_bsd_ring: + intel_cleanup_ring_buffer(&dev_priv->ring[VCS]); +cleanup_render_ring: + intel_cleanup_ring_buffer(&dev_priv->ring[RCS]); + return ret; +} + +void +i915_gem_cleanup_ringbuffer(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + int i; + + for (i = 0; i < I915_NUM_RINGS; i++) + intel_cleanup_ring_buffer(&dev_priv->ring[i]); +} + +int +i915_gem_entervt_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + int ret, i; + + if (drm_core_check_feature(dev, DRIVER_MODESET)) + return 0; + + if (atomic_read(&dev_priv->mm.wedged)) { + DRM_ERROR("Reenabling wedged hardware, good luck\n"); + atomic_set(&dev_priv->mm.wedged, 0); + } + + mutex_lock(&dev->struct_mutex); + dev_priv->mm.suspended = 0; + + ret = i915_gem_init_ringbuffer(dev); + if (ret != 0) { + mutex_unlock(&dev->struct_mutex); + return ret; + } + + BUG_ON(!list_empty(&dev_priv->mm.active_list)); + BUG_ON(!list_empty(&dev_priv->mm.flushing_list)); + BUG_ON(!list_empty(&dev_priv->mm.inactive_list)); + for (i = 0; i < I915_NUM_RINGS; i++) { + BUG_ON(!list_empty(&dev_priv->ring[i].active_list)); + BUG_ON(!list_empty(&dev_priv->ring[i].request_list)); + } + mutex_unlock(&dev->struct_mutex); + + ret = drm_irq_install(dev); + if (ret) + goto cleanup_ringbuffer; + + return 0; + +cleanup_ringbuffer: + mutex_lock(&dev->struct_mutex); + i915_gem_cleanup_ringbuffer(dev); + dev_priv->mm.suspended = 1; + mutex_unlock(&dev->struct_mutex); + + return ret; +} + +int +i915_gem_leavevt_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + if (drm_core_check_feature(dev, DRIVER_MODESET)) + return 0; + + drm_irq_uninstall(dev); + return i915_gem_idle(dev); +} + +void +i915_gem_lastclose(struct drm_device *dev) +{ + int ret; + + if (drm_core_check_feature(dev, DRIVER_MODESET)) + return; + + ret = i915_gem_idle(dev); + if (ret) + DRM_ERROR("failed to idle hardware: %d\n", ret); +} + +static void +init_ring_lists(struct intel_ring_buffer *ring) +{ + INIT_LIST_HEAD(&ring->active_list); + INIT_LIST_HEAD(&ring->request_list); + INIT_LIST_HEAD(&ring->gpu_write_list); +} + +void +i915_gem_load(struct drm_device *dev) +{ + int i; + drm_i915_private_t *dev_priv = dev->dev_private; + + INIT_LIST_HEAD(&dev_priv->mm.active_list); + INIT_LIST_HEAD(&dev_priv->mm.flushing_list); + INIT_LIST_HEAD(&dev_priv->mm.inactive_list); + INIT_LIST_HEAD(&dev_priv->mm.pinned_list); + INIT_LIST_HEAD(&dev_priv->mm.fence_list); + INIT_LIST_HEAD(&dev_priv->mm.deferred_free_list); + INIT_LIST_HEAD(&dev_priv->mm.gtt_list); + for (i = 0; i < I915_NUM_RINGS; i++) + init_ring_lists(&dev_priv->ring[i]); + for (i = 0; i < 16; i++) + INIT_LIST_HEAD(&dev_priv->fence_regs[i].lru_list); + INIT_DELAYED_WORK(&dev_priv->mm.retire_work, + i915_gem_retire_work_handler); + init_completion(&dev_priv->error_completion); + + /* On GEN3 we really need to make sure the ARB C3 LP bit is set */ + if (IS_GEN3(dev)) { + u32 tmp = I915_READ(MI_ARB_STATE); + if (!(tmp & MI_ARB_C3_LP_WRITE_ENABLE)) { + /* arb state is a masked write, so set bit + bit in mask */ + tmp = MI_ARB_C3_LP_WRITE_ENABLE | (MI_ARB_C3_LP_WRITE_ENABLE << MI_ARB_MASK_SHIFT); + I915_WRITE(MI_ARB_STATE, tmp); + } + } + + dev_priv->relative_constants_mode = I915_EXEC_CONSTANTS_REL_GENERAL; + + /* Old X drivers will take 0-2 for front, back, depth buffers */ + if (!drm_core_check_feature(dev, DRIVER_MODESET)) + dev_priv->fence_reg_start = 3; + + if (INTEL_INFO(dev)->gen >= 4 || IS_I945G(dev) || IS_I945GM(dev) || IS_G33(dev)) + dev_priv->num_fence_regs = 16; + else + dev_priv->num_fence_regs = 8; + + /* Initialize fence registers to zero */ + for (i = 0; i < dev_priv->num_fence_regs; i++) { + i915_gem_clear_fence_reg(dev, &dev_priv->fence_regs[i]); + } + + i915_gem_detect_bit_6_swizzle(dev); + init_waitqueue_head(&dev_priv->pending_flip_queue); + + dev_priv->mm.interruptible = true; + + dev_priv->mm.inactive_shrinker.shrink = i915_gem_inactive_shrink; + dev_priv->mm.inactive_shrinker.seeks = DEFAULT_SEEKS; + register_shrinker(&dev_priv->mm.inactive_shrinker); +} + +/* + * Create a physically contiguous memory object for this object + * e.g. for cursor + overlay regs + */ +static int i915_gem_init_phys_object(struct drm_device *dev, + int id, int size, int align) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_i915_gem_phys_object *phys_obj; + int ret; + + if (dev_priv->mm.phys_objs[id - 1] || !size) + return 0; + + phys_obj = kzalloc(sizeof(struct drm_i915_gem_phys_object), GFP_KERNEL); + if (!phys_obj) + return -ENOMEM; + + phys_obj->id = id; + + phys_obj->handle = drm_pci_alloc(dev, size, align); + if (!phys_obj->handle) { + ret = -ENOMEM; + goto kfree_obj; + } +#ifdef CONFIG_X86 + set_memory_wc((unsigned long)phys_obj->handle->vaddr, phys_obj->handle->size / PAGE_SIZE); +#endif + + dev_priv->mm.phys_objs[id - 1] = phys_obj; + + return 0; +kfree_obj: + kfree(phys_obj); + return ret; +} + +static void i915_gem_free_phys_object(struct drm_device *dev, int id) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_i915_gem_phys_object *phys_obj; + + if (!dev_priv->mm.phys_objs[id - 1]) + return; + + phys_obj = dev_priv->mm.phys_objs[id - 1]; + if (phys_obj->cur_obj) { + i915_gem_detach_phys_object(dev, phys_obj->cur_obj); + } + +#ifdef CONFIG_X86 + set_memory_wb((unsigned long)phys_obj->handle->vaddr, phys_obj->handle->size / PAGE_SIZE); +#endif + drm_pci_free(dev, phys_obj->handle); + kfree(phys_obj); + dev_priv->mm.phys_objs[id - 1] = NULL; +} + +void i915_gem_free_all_phys_object(struct drm_device *dev) +{ + int i; + + for (i = I915_GEM_PHYS_CURSOR_0; i <= I915_MAX_PHYS_OBJECT; i++) + i915_gem_free_phys_object(dev, i); +} + +void i915_gem_detach_phys_object(struct drm_device *dev, + struct drm_i915_gem_object *obj) +{ + struct address_space *mapping = obj->base.filp->f_path.dentry->d_inode->i_mapping; + char *vaddr; + int i; + int page_count; + + if (!obj->phys_obj) + return; + vaddr = obj->phys_obj->handle->vaddr; + + page_count = obj->base.size / PAGE_SIZE; + for (i = 0; i < page_count; i++) { + struct page *page = shmem_read_mapping_page(mapping, i); + if (!IS_ERR(page)) { + char *dst = kmap_atomic(page); + memcpy(dst, vaddr + i*PAGE_SIZE, PAGE_SIZE); + kunmap_atomic(dst); + + drm_clflush_pages(&page, 1); + + set_page_dirty(page); + mark_page_accessed(page); + page_cache_release(page); + } + } + intel_gtt_chipset_flush(); + + obj->phys_obj->cur_obj = NULL; + obj->phys_obj = NULL; +} + +int +i915_gem_attach_phys_object(struct drm_device *dev, + struct drm_i915_gem_object *obj, + int id, + int align) +{ + struct address_space *mapping = obj->base.filp->f_path.dentry->d_inode->i_mapping; + drm_i915_private_t *dev_priv = dev->dev_private; + int ret = 0; + int page_count; + int i; + + if (id > I915_MAX_PHYS_OBJECT) + return -EINVAL; + + if (obj->phys_obj) { + if (obj->phys_obj->id == id) + return 0; + i915_gem_detach_phys_object(dev, obj); + } + + /* create a new object */ + if (!dev_priv->mm.phys_objs[id - 1]) { + ret = i915_gem_init_phys_object(dev, id, + obj->base.size, align); + if (ret) { + DRM_ERROR("failed to init phys object %d size: %zu\n", + id, obj->base.size); + return ret; + } + } + + /* bind to the object */ + obj->phys_obj = dev_priv->mm.phys_objs[id - 1]; + obj->phys_obj->cur_obj = obj; + + page_count = obj->base.size / PAGE_SIZE; + + for (i = 0; i < page_count; i++) { + struct page *page; + char *dst, *src; + + page = shmem_read_mapping_page(mapping, i); + if (IS_ERR(page)) + return PTR_ERR(page); + + src = kmap_atomic(page); + dst = obj->phys_obj->handle->vaddr + (i * PAGE_SIZE); + memcpy(dst, src, PAGE_SIZE); + kunmap_atomic(src); + + mark_page_accessed(page); + page_cache_release(page); + } + + return 0; +} + +static int +i915_gem_phys_pwrite(struct drm_device *dev, + struct drm_i915_gem_object *obj, + struct drm_i915_gem_pwrite *args, + struct drm_file *file_priv) +{ + void *vaddr = obj->phys_obj->handle->vaddr + args->offset; + char __user *user_data = (char __user *) (uintptr_t) args->data_ptr; + + if (__copy_from_user_inatomic_nocache(vaddr, user_data, args->size)) { + unsigned long unwritten; + + /* The physical object once assigned is fixed for the lifetime + * of the obj, so we can safely drop the lock and continue + * to access vaddr. + */ + mutex_unlock(&dev->struct_mutex); + unwritten = copy_from_user(vaddr, user_data, args->size); + mutex_lock(&dev->struct_mutex); + if (unwritten) + return -EFAULT; + } + + intel_gtt_chipset_flush(); + return 0; +} + +void i915_gem_release(struct drm_device *dev, struct drm_file *file) +{ + struct drm_i915_file_private *file_priv = file->driver_priv; + + /* Clean up our request list when the client is going away, so that + * later retire_requests won't dereference our soon-to-be-gone + * file_priv. + */ + spin_lock(&file_priv->mm.lock); + while (!list_empty(&file_priv->mm.request_list)) { + struct drm_i915_gem_request *request; + + request = list_first_entry(&file_priv->mm.request_list, + struct drm_i915_gem_request, + client_list); + list_del(&request->client_list); + request->file_priv = NULL; + } + spin_unlock(&file_priv->mm.lock); +} + +static int +i915_gpu_is_active(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + int lists_empty; + + lists_empty = list_empty(&dev_priv->mm.flushing_list) && + list_empty(&dev_priv->mm.active_list); + + return !lists_empty; +} + +static int +i915_gem_inactive_shrink(struct shrinker *shrinker, struct shrink_control *sc) +{ + struct drm_i915_private *dev_priv = + container_of(shrinker, + struct drm_i915_private, + mm.inactive_shrinker); + struct drm_device *dev = dev_priv->dev; + struct drm_i915_gem_object *obj, *next; + int nr_to_scan = sc->nr_to_scan; + int cnt; + + if (!mutex_trylock(&dev->struct_mutex)) + return 0; + + /* "fast-path" to count number of available objects */ + if (nr_to_scan == 0) { + cnt = 0; + list_for_each_entry(obj, + &dev_priv->mm.inactive_list, + mm_list) + cnt++; + mutex_unlock(&dev->struct_mutex); + return cnt / 100 * sysctl_vfs_cache_pressure; + } + +rescan: + /* first scan for clean buffers */ + i915_gem_retire_requests(dev); + + list_for_each_entry_safe(obj, next, + &dev_priv->mm.inactive_list, + mm_list) { + if (i915_gem_object_is_purgeable(obj)) { + if (i915_gem_object_unbind(obj) == 0 && + --nr_to_scan == 0) + break; + } + } + + /* second pass, evict/count anything still on the inactive list */ + cnt = 0; + list_for_each_entry_safe(obj, next, + &dev_priv->mm.inactive_list, + mm_list) { + if (nr_to_scan && + i915_gem_object_unbind(obj) == 0) + nr_to_scan--; + else + cnt++; + } + + if (nr_to_scan && i915_gpu_is_active(dev)) { + /* + * We are desperate for pages, so as a last resort, wait + * for the GPU to finish and discard whatever we can. + * This has a dramatic impact to reduce the number of + * OOM-killer events whilst running the GPU aggressively. + */ + if (i915_gpu_idle(dev) == 0) + goto rescan; + } + mutex_unlock(&dev->struct_mutex); + return cnt / 100 * sysctl_vfs_cache_pressure; +} diff --git a/drivers/gpu/drm/i915/i915_gem_debug.c b/drivers/gpu/drm/i915/i915_gem_debug.c new file mode 100644 index 00000000..8da1899b --- /dev/null +++ b/drivers/gpu/drm/i915/i915_gem_debug.c @@ -0,0 +1,203 @@ +/* + * Copyright © 2008 Intel Corporation + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS + * IN THE SOFTWARE. + * + * Authors: + * Keith Packard + * + */ + +#include "drmP.h" +#include "drm.h" +#include "i915_drm.h" +#include "i915_drv.h" + +#if WATCH_LISTS +int +i915_verify_lists(struct drm_device *dev) +{ + static int warned; + drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_i915_gem_object *obj; + int err = 0; + + if (warned) + return 0; + + list_for_each_entry(obj, &dev_priv->render_ring.active_list, list) { + if (obj->base.dev != dev || + !atomic_read(&obj->base.refcount.refcount)) { + DRM_ERROR("freed render active %p\n", obj); + err++; + break; + } else if (!obj->active || + (obj->base.read_domains & I915_GEM_GPU_DOMAINS) == 0) { + DRM_ERROR("invalid render active %p (a %d r %x)\n", + obj, + obj->active, + obj->base.read_domains); + err++; + } else if (obj->base.write_domain && list_empty(&obj->gpu_write_list)) { + DRM_ERROR("invalid render active %p (w %x, gwl %d)\n", + obj, + obj->base.write_domain, + !list_empty(&obj->gpu_write_list)); + err++; + } + } + + list_for_each_entry(obj, &dev_priv->mm.flushing_list, list) { + if (obj->base.dev != dev || + !atomic_read(&obj->base.refcount.refcount)) { + DRM_ERROR("freed flushing %p\n", obj); + err++; + break; + } else if (!obj->active || + (obj->base.write_domain & I915_GEM_GPU_DOMAINS) == 0 || + list_empty(&obj->gpu_write_list)){ + DRM_ERROR("invalid flushing %p (a %d w %x gwl %d)\n", + obj, + obj->active, + obj->base.write_domain, + !list_empty(&obj->gpu_write_list)); + err++; + } + } + + list_for_each_entry(obj, &dev_priv->mm.gpu_write_list, gpu_write_list) { + if (obj->base.dev != dev || + !atomic_read(&obj->base.refcount.refcount)) { + DRM_ERROR("freed gpu write %p\n", obj); + err++; + break; + } else if (!obj->active || + (obj->base.write_domain & I915_GEM_GPU_DOMAINS) == 0) { + DRM_ERROR("invalid gpu write %p (a %d w %x)\n", + obj, + obj->active, + obj->base.write_domain); + err++; + } + } + + list_for_each_entry(obj, &dev_priv->mm.inactive_list, list) { + if (obj->base.dev != dev || + !atomic_read(&obj->base.refcount.refcount)) { + DRM_ERROR("freed inactive %p\n", obj); + err++; + break; + } else if (obj->pin_count || obj->active || + (obj->base.write_domain & I915_GEM_GPU_DOMAINS)) { + DRM_ERROR("invalid inactive %p (p %d a %d w %x)\n", + obj, + obj->pin_count, obj->active, + obj->base.write_domain); + err++; + } + } + + list_for_each_entry(obj, &dev_priv->mm.pinned_list, list) { + if (obj->base.dev != dev || + !atomic_read(&obj->base.refcount.refcount)) { + DRM_ERROR("freed pinned %p\n", obj); + err++; + break; + } else if (!obj->pin_count || obj->active || + (obj->base.write_domain & I915_GEM_GPU_DOMAINS)) { + DRM_ERROR("invalid pinned %p (p %d a %d w %x)\n", + obj, + obj->pin_count, obj->active, + obj->base.write_domain); + err++; + } + } + + return warned = err; +} +#endif /* WATCH_INACTIVE */ + +#if WATCH_COHERENCY +void +i915_gem_object_check_coherency(struct drm_i915_gem_object *obj, int handle) +{ + struct drm_device *dev = obj->base.dev; + int page; + uint32_t *gtt_mapping; + uint32_t *backing_map = NULL; + int bad_count = 0; + + DRM_INFO("%s: checking coherency of object %p@0x%08x (%d, %zdkb):\n", + __func__, obj, obj->gtt_offset, handle, + obj->size / 1024); + + gtt_mapping = ioremap(dev->agp->base + obj->gtt_offset, obj->base.size); + if (gtt_mapping == NULL) { + DRM_ERROR("failed to map GTT space\n"); + return; + } + + for (page = 0; page < obj->size / PAGE_SIZE; page++) { + int i; + + backing_map = kmap_atomic(obj->pages[page], KM_USER0); + + if (backing_map == NULL) { + DRM_ERROR("failed to map backing page\n"); + goto out; + } + + for (i = 0; i < PAGE_SIZE / 4; i++) { + uint32_t cpuval = backing_map[i]; + uint32_t gttval = readl(gtt_mapping + + page * 1024 + i); + + if (cpuval != gttval) { + DRM_INFO("incoherent CPU vs GPU at 0x%08x: " + "0x%08x vs 0x%08x\n", + (int)(obj->gtt_offset + + page * PAGE_SIZE + i * 4), + cpuval, gttval); + if (bad_count++ >= 8) { + DRM_INFO("...\n"); + goto out; + } + } + } + kunmap_atomic(backing_map, KM_USER0); + backing_map = NULL; + } + + out: + if (backing_map != NULL) + kunmap_atomic(backing_map, KM_USER0); + iounmap(gtt_mapping); + + /* give syslog time to catch up */ + msleep(1); + + /* Directly flush the object, since we just loaded values with the CPU + * from the backing pages and we don't want to disturb the cache + * management that we're trying to observe. + */ + + i915_gem_clflush_object(obj); +} +#endif diff --git a/drivers/gpu/drm/i915/i915_gem_evict.c b/drivers/gpu/drm/i915/i915_gem_evict.c new file mode 100644 index 00000000..da05a269 --- /dev/null +++ b/drivers/gpu/drm/i915/i915_gem_evict.c @@ -0,0 +1,224 @@ +/* + * Copyright © 2008-2010 Intel Corporation + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS + * IN THE SOFTWARE. + * + * Authors: + * Eric Anholt + * Chris Wilson + * + */ + +#include "drmP.h" +#include "drm.h" +#include "i915_drv.h" +#include "i915_drm.h" +#include "i915_trace.h" + +static bool +mark_free(struct drm_i915_gem_object *obj, struct list_head *unwind) +{ + list_add(&obj->exec_list, unwind); + drm_gem_object_reference(&obj->base); + return drm_mm_scan_add_block(obj->gtt_space); +} + +int +i915_gem_evict_something(struct drm_device *dev, int min_size, + unsigned alignment, bool mappable) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct list_head eviction_list, unwind_list; + struct drm_i915_gem_object *obj; + int ret = 0; + + i915_gem_retire_requests(dev); + + /* Re-check for free space after retiring requests */ + if (mappable) { + if (drm_mm_search_free_in_range(&dev_priv->mm.gtt_space, + min_size, alignment, 0, + dev_priv->mm.gtt_mappable_end, + 0)) + return 0; + } else { + if (drm_mm_search_free(&dev_priv->mm.gtt_space, + min_size, alignment, 0)) + return 0; + } + + trace_i915_gem_evict(dev, min_size, alignment, mappable); + + /* + * The goal is to evict objects and amalgamate space in LRU order. + * The oldest idle objects reside on the inactive list, which is in + * retirement order. The next objects to retire are those on the (per + * ring) active list that do not have an outstanding flush. Once the + * hardware reports completion (the seqno is updated after the + * batchbuffer has been finished) the clean buffer objects would + * be retired to the inactive list. Any dirty objects would be added + * to the tail of the flushing list. So after processing the clean + * active objects we need to emit a MI_FLUSH to retire the flushing + * list, hence the retirement order of the flushing list is in + * advance of the dirty objects on the active lists. + * + * The retirement sequence is thus: + * 1. Inactive objects (already retired) + * 2. Clean active objects + * 3. Flushing list + * 4. Dirty active objects. + * + * On each list, the oldest objects lie at the HEAD with the freshest + * object on the TAIL. + */ + + INIT_LIST_HEAD(&unwind_list); + if (mappable) + drm_mm_init_scan_with_range(&dev_priv->mm.gtt_space, min_size, + alignment, 0, + dev_priv->mm.gtt_mappable_end); + else + drm_mm_init_scan(&dev_priv->mm.gtt_space, min_size, alignment); + + /* First see if there is a large enough contiguous idle region... */ + list_for_each_entry(obj, &dev_priv->mm.inactive_list, mm_list) { + if (mark_free(obj, &unwind_list)) + goto found; + } + + /* Now merge in the soon-to-be-expired objects... */ + list_for_each_entry(obj, &dev_priv->mm.active_list, mm_list) { + /* Does the object require an outstanding flush? */ + if (obj->base.write_domain || obj->pin_count) + continue; + + if (mark_free(obj, &unwind_list)) + goto found; + } + + /* Finally add anything with a pending flush (in order of retirement) */ + list_for_each_entry(obj, &dev_priv->mm.flushing_list, mm_list) { + if (obj->pin_count) + continue; + + if (mark_free(obj, &unwind_list)) + goto found; + } + list_for_each_entry(obj, &dev_priv->mm.active_list, mm_list) { + if (! obj->base.write_domain || obj->pin_count) + continue; + + if (mark_free(obj, &unwind_list)) + goto found; + } + + /* Nothing found, clean up and bail out! */ + while (!list_empty(&unwind_list)) { + obj = list_first_entry(&unwind_list, + struct drm_i915_gem_object, + exec_list); + + ret = drm_mm_scan_remove_block(obj->gtt_space); + BUG_ON(ret); + + list_del_init(&obj->exec_list); + drm_gem_object_unreference(&obj->base); + } + + /* We expect the caller to unpin, evict all and try again, or give up. + * So calling i915_gem_evict_everything() is unnecessary. + */ + return -ENOSPC; + +found: + /* drm_mm doesn't allow any other other operations while + * scanning, therefore store to be evicted objects on a + * temporary list. */ + INIT_LIST_HEAD(&eviction_list); + while (!list_empty(&unwind_list)) { + obj = list_first_entry(&unwind_list, + struct drm_i915_gem_object, + exec_list); + if (drm_mm_scan_remove_block(obj->gtt_space)) { + list_move(&obj->exec_list, &eviction_list); + continue; + } + list_del_init(&obj->exec_list); + drm_gem_object_unreference(&obj->base); + } + + /* Unbinding will emit any required flushes */ + while (!list_empty(&eviction_list)) { + obj = list_first_entry(&eviction_list, + struct drm_i915_gem_object, + exec_list); + if (ret == 0) + ret = i915_gem_object_unbind(obj); + + list_del_init(&obj->exec_list); + drm_gem_object_unreference(&obj->base); + } + + return ret; +} + +int +i915_gem_evict_everything(struct drm_device *dev, bool purgeable_only) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + int ret; + bool lists_empty; + + lists_empty = (list_empty(&dev_priv->mm.inactive_list) && + list_empty(&dev_priv->mm.flushing_list) && + list_empty(&dev_priv->mm.active_list)); + if (lists_empty) + return -ENOSPC; + + trace_i915_gem_evict_everything(dev, purgeable_only); + + /* Flush everything (on to the inactive lists) and evict */ + ret = i915_gpu_idle(dev); + if (ret) + return ret; + + BUG_ON(!list_empty(&dev_priv->mm.flushing_list)); + + return i915_gem_evict_inactive(dev, purgeable_only); +} + +/** Unbinds all inactive objects. */ +int +i915_gem_evict_inactive(struct drm_device *dev, bool purgeable_only) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_i915_gem_object *obj, *next; + + list_for_each_entry_safe(obj, next, + &dev_priv->mm.inactive_list, mm_list) { + if (!purgeable_only || obj->madv != I915_MADV_WILLNEED) { + int ret = i915_gem_object_unbind(obj); + if (ret) + return ret; + } + } + + return 0; +} diff --git a/drivers/gpu/drm/i915/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/i915_gem_execbuffer.c new file mode 100644 index 00000000..bc927ae3 --- /dev/null +++ b/drivers/gpu/drm/i915/i915_gem_execbuffer.c @@ -0,0 +1,1348 @@ +/* + * Copyright © 2008,2010 Intel Corporation + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS + * IN THE SOFTWARE. + * + * Authors: + * Eric Anholt + * Chris Wilson + * + */ + +#include "drmP.h" +#include "drm.h" +#include "i915_drm.h" +#include "i915_drv.h" +#include "i915_trace.h" +#include "intel_drv.h" + +struct change_domains { + uint32_t invalidate_domains; + uint32_t flush_domains; + uint32_t flush_rings; + uint32_t flips; +}; + +/* + * Set the next domain for the specified object. This + * may not actually perform the necessary flushing/invaliding though, + * as that may want to be batched with other set_domain operations + * + * This is (we hope) the only really tricky part of gem. The goal + * is fairly simple -- track which caches hold bits of the object + * and make sure they remain coherent. A few concrete examples may + * help to explain how it works. For shorthand, we use the notation + * (read_domains, write_domain), e.g. (CPU, CPU) to indicate the + * a pair of read and write domain masks. + * + * Case 1: the batch buffer + * + * 1. Allocated + * 2. Written by CPU + * 3. Mapped to GTT + * 4. Read by GPU + * 5. Unmapped from GTT + * 6. Freed + * + * Let's take these a step at a time + * + * 1. Allocated + * Pages allocated from the kernel may still have + * cache contents, so we set them to (CPU, CPU) always. + * 2. Written by CPU (using pwrite) + * The pwrite function calls set_domain (CPU, CPU) and + * this function does nothing (as nothing changes) + * 3. Mapped by GTT + * This function asserts that the object is not + * currently in any GPU-based read or write domains + * 4. Read by GPU + * i915_gem_execbuffer calls set_domain (COMMAND, 0). + * As write_domain is zero, this function adds in the + * current read domains (CPU+COMMAND, 0). + * flush_domains is set to CPU. + * invalidate_domains is set to COMMAND + * clflush is run to get data out of the CPU caches + * then i915_dev_set_domain calls i915_gem_flush to + * emit an MI_FLUSH and drm_agp_chipset_flush + * 5. Unmapped from GTT + * i915_gem_object_unbind calls set_domain (CPU, CPU) + * flush_domains and invalidate_domains end up both zero + * so no flushing/invalidating happens + * 6. Freed + * yay, done + * + * Case 2: The shared render buffer + * + * 1. Allocated + * 2. Mapped to GTT + * 3. Read/written by GPU + * 4. set_domain to (CPU,CPU) + * 5. Read/written by CPU + * 6. Read/written by GPU + * + * 1. Allocated + * Same as last example, (CPU, CPU) + * 2. Mapped to GTT + * Nothing changes (assertions find that it is not in the GPU) + * 3. Read/written by GPU + * execbuffer calls set_domain (RENDER, RENDER) + * flush_domains gets CPU + * invalidate_domains gets GPU + * clflush (obj) + * MI_FLUSH and drm_agp_chipset_flush + * 4. set_domain (CPU, CPU) + * flush_domains gets GPU + * invalidate_domains gets CPU + * wait_rendering (obj) to make sure all drawing is complete. + * This will include an MI_FLUSH to get the data from GPU + * to memory + * clflush (obj) to invalidate the CPU cache + * Another MI_FLUSH in i915_gem_flush (eliminate this somehow?) + * 5. Read/written by CPU + * cache lines are loaded and dirtied + * 6. Read written by GPU + * Same as last GPU access + * + * Case 3: The constant buffer + * + * 1. Allocated + * 2. Written by CPU + * 3. Read by GPU + * 4. Updated (written) by CPU again + * 5. Read by GPU + * + * 1. Allocated + * (CPU, CPU) + * 2. Written by CPU + * (CPU, CPU) + * 3. Read by GPU + * (CPU+RENDER, 0) + * flush_domains = CPU + * invalidate_domains = RENDER + * clflush (obj) + * MI_FLUSH + * drm_agp_chipset_flush + * 4. Updated (written) by CPU again + * (CPU, CPU) + * flush_domains = 0 (no previous write domain) + * invalidate_domains = 0 (no new read domains) + * 5. Read by GPU + * (CPU+RENDER, 0) + * flush_domains = CPU + * invalidate_domains = RENDER + * clflush (obj) + * MI_FLUSH + * drm_agp_chipset_flush + */ +static void +i915_gem_object_set_to_gpu_domain(struct drm_i915_gem_object *obj, + struct intel_ring_buffer *ring, + struct change_domains *cd) +{ + uint32_t invalidate_domains = 0, flush_domains = 0; + + /* + * If the object isn't moving to a new write domain, + * let the object stay in multiple read domains + */ + if (obj->base.pending_write_domain == 0) + obj->base.pending_read_domains |= obj->base.read_domains; + + /* + * Flush the current write domain if + * the new read domains don't match. Invalidate + * any read domains which differ from the old + * write domain + */ + if (obj->base.write_domain && + (((obj->base.write_domain != obj->base.pending_read_domains || + obj->ring != ring)) || + (obj->fenced_gpu_access && !obj->pending_fenced_gpu_access))) { + flush_domains |= obj->base.write_domain; + invalidate_domains |= + obj->base.pending_read_domains & ~obj->base.write_domain; + } + /* + * Invalidate any read caches which may have + * stale data. That is, any new read domains. + */ + invalidate_domains |= obj->base.pending_read_domains & ~obj->base.read_domains; + if ((flush_domains | invalidate_domains) & I915_GEM_DOMAIN_CPU) + i915_gem_clflush_object(obj); + + if (obj->base.pending_write_domain) + cd->flips |= atomic_read(&obj->pending_flip); + + /* The actual obj->write_domain will be updated with + * pending_write_domain after we emit the accumulated flush for all + * of our domain changes in execbuffers (which clears objects' + * write_domains). So if we have a current write domain that we + * aren't changing, set pending_write_domain to that. + */ + if (flush_domains == 0 && obj->base.pending_write_domain == 0) + obj->base.pending_write_domain = obj->base.write_domain; + + cd->invalidate_domains |= invalidate_domains; + cd->flush_domains |= flush_domains; + if (flush_domains & I915_GEM_GPU_DOMAINS) + cd->flush_rings |= obj->ring->id; + if (invalidate_domains & I915_GEM_GPU_DOMAINS) + cd->flush_rings |= ring->id; +} + +struct eb_objects { + int and; + struct hlist_head buckets[0]; +}; + +static struct eb_objects * +eb_create(int size) +{ + struct eb_objects *eb; + int count = PAGE_SIZE / sizeof(struct hlist_head) / 2; + while (count > size) + count >>= 1; + eb = kzalloc(count*sizeof(struct hlist_head) + + sizeof(struct eb_objects), + GFP_KERNEL); + if (eb == NULL) + return eb; + + eb->and = count - 1; + return eb; +} + +static void +eb_reset(struct eb_objects *eb) +{ + memset(eb->buckets, 0, (eb->and+1)*sizeof(struct hlist_head)); +} + +static void +eb_add_object(struct eb_objects *eb, struct drm_i915_gem_object *obj) +{ + hlist_add_head(&obj->exec_node, + &eb->buckets[obj->exec_handle & eb->and]); +} + +static struct drm_i915_gem_object * +eb_get_object(struct eb_objects *eb, unsigned long handle) +{ + struct hlist_head *head; + struct hlist_node *node; + struct drm_i915_gem_object *obj; + + head = &eb->buckets[handle & eb->and]; + hlist_for_each(node, head) { + obj = hlist_entry(node, struct drm_i915_gem_object, exec_node); + if (obj->exec_handle == handle) + return obj; + } + + return NULL; +} + +static void +eb_destroy(struct eb_objects *eb) +{ + kfree(eb); +} + +static int +i915_gem_execbuffer_relocate_entry(struct drm_i915_gem_object *obj, + struct eb_objects *eb, + struct drm_i915_gem_relocation_entry *reloc) +{ + struct drm_device *dev = obj->base.dev; + struct drm_gem_object *target_obj; + uint32_t target_offset; + int ret = -EINVAL; + + /* we've already hold a reference to all valid objects */ + target_obj = &eb_get_object(eb, reloc->target_handle)->base; + if (unlikely(target_obj == NULL)) + return -ENOENT; + + target_offset = to_intel_bo(target_obj)->gtt_offset; + + /* The target buffer should have appeared before us in the + * exec_object list, so it should have a GTT space bound by now. + */ + if (unlikely(target_offset == 0)) { + DRM_ERROR("No GTT space found for object %d\n", + reloc->target_handle); + return ret; + } + + /* Validate that the target is in a valid r/w GPU domain */ + if (unlikely(reloc->write_domain & (reloc->write_domain - 1))) { + DRM_ERROR("reloc with multiple write domains: " + "obj %p target %d offset %d " + "read %08x write %08x", + obj, reloc->target_handle, + (int) reloc->offset, + reloc->read_domains, + reloc->write_domain); + return ret; + } + if (unlikely((reloc->write_domain | reloc->read_domains) & I915_GEM_DOMAIN_CPU)) { + DRM_ERROR("reloc with read/write CPU domains: " + "obj %p target %d offset %d " + "read %08x write %08x", + obj, reloc->target_handle, + (int) reloc->offset, + reloc->read_domains, + reloc->write_domain); + return ret; + } + if (unlikely(reloc->write_domain && target_obj->pending_write_domain && + reloc->write_domain != target_obj->pending_write_domain)) { + DRM_ERROR("Write domain conflict: " + "obj %p target %d offset %d " + "new %08x old %08x\n", + obj, reloc->target_handle, + (int) reloc->offset, + reloc->write_domain, + target_obj->pending_write_domain); + return ret; + } + + target_obj->pending_read_domains |= reloc->read_domains; + target_obj->pending_write_domain |= reloc->write_domain; + + /* If the relocation already has the right value in it, no + * more work needs to be done. + */ + if (target_offset == reloc->presumed_offset) + return 0; + + /* Check that the relocation address is valid... */ + if (unlikely(reloc->offset > obj->base.size - 4)) { + DRM_ERROR("Relocation beyond object bounds: " + "obj %p target %d offset %d size %d.\n", + obj, reloc->target_handle, + (int) reloc->offset, + (int) obj->base.size); + return ret; + } + if (unlikely(reloc->offset & 3)) { + DRM_ERROR("Relocation not 4-byte aligned: " + "obj %p target %d offset %d.\n", + obj, reloc->target_handle, + (int) reloc->offset); + return ret; + } + + reloc->delta += target_offset; + if (obj->base.write_domain == I915_GEM_DOMAIN_CPU) { + uint32_t page_offset = reloc->offset & ~PAGE_MASK; + char *vaddr; + + vaddr = kmap_atomic(obj->pages[reloc->offset >> PAGE_SHIFT]); + *(uint32_t *)(vaddr + page_offset) = reloc->delta; + kunmap_atomic(vaddr); + } else { + struct drm_i915_private *dev_priv = dev->dev_private; + uint32_t __iomem *reloc_entry; + void __iomem *reloc_page; + + /* We can't wait for rendering with pagefaults disabled */ + if (obj->active && in_atomic()) + return -EFAULT; + + ret = i915_gem_object_set_to_gtt_domain(obj, 1); + if (ret) + return ret; + + /* Map the page containing the relocation we're going to perform. */ + reloc->offset += obj->gtt_offset; + reloc_page = io_mapping_map_atomic_wc(dev_priv->mm.gtt_mapping, + reloc->offset & PAGE_MASK); + reloc_entry = (uint32_t __iomem *) + (reloc_page + (reloc->offset & ~PAGE_MASK)); + iowrite32(reloc->delta, reloc_entry); + io_mapping_unmap_atomic(reloc_page); + } + + /* and update the user's relocation entry */ + reloc->presumed_offset = target_offset; + + return 0; +} + +static int +i915_gem_execbuffer_relocate_object(struct drm_i915_gem_object *obj, + struct eb_objects *eb) +{ + struct drm_i915_gem_relocation_entry __user *user_relocs; + struct drm_i915_gem_exec_object2 *entry = obj->exec_entry; + int i, ret; + + user_relocs = (void __user *)(uintptr_t)entry->relocs_ptr; + for (i = 0; i < entry->relocation_count; i++) { + struct drm_i915_gem_relocation_entry reloc; + + if (__copy_from_user_inatomic(&reloc, + user_relocs+i, + sizeof(reloc))) + return -EFAULT; + + ret = i915_gem_execbuffer_relocate_entry(obj, eb, &reloc); + if (ret) + return ret; + + if (__copy_to_user_inatomic(&user_relocs[i].presumed_offset, + &reloc.presumed_offset, + sizeof(reloc.presumed_offset))) + return -EFAULT; + } + + return 0; +} + +static int +i915_gem_execbuffer_relocate_object_slow(struct drm_i915_gem_object *obj, + struct eb_objects *eb, + struct drm_i915_gem_relocation_entry *relocs) +{ + const struct drm_i915_gem_exec_object2 *entry = obj->exec_entry; + int i, ret; + + for (i = 0; i < entry->relocation_count; i++) { + ret = i915_gem_execbuffer_relocate_entry(obj, eb, &relocs[i]); + if (ret) + return ret; + } + + return 0; +} + +static int +i915_gem_execbuffer_relocate(struct drm_device *dev, + struct eb_objects *eb, + struct list_head *objects) +{ + struct drm_i915_gem_object *obj; + int ret = 0; + + /* This is the fast path and we cannot handle a pagefault whilst + * holding the struct mutex lest the user pass in the relocations + * contained within a mmaped bo. For in such a case we, the page + * fault handler would call i915_gem_fault() and we would try to + * acquire the struct mutex again. Obviously this is bad and so + * lockdep complains vehemently. + */ + pagefault_disable(); + list_for_each_entry(obj, objects, exec_list) { + ret = i915_gem_execbuffer_relocate_object(obj, eb); + if (ret) + break; + } + pagefault_enable(); + + return ret; +} + +static int +i915_gem_execbuffer_reserve(struct intel_ring_buffer *ring, + struct drm_file *file, + struct list_head *objects) +{ + struct drm_i915_gem_object *obj; + int ret, retry; + bool has_fenced_gpu_access = INTEL_INFO(ring->dev)->gen < 4; + struct list_head ordered_objects; + + INIT_LIST_HEAD(&ordered_objects); + while (!list_empty(objects)) { + struct drm_i915_gem_exec_object2 *entry; + bool need_fence, need_mappable; + + obj = list_first_entry(objects, + struct drm_i915_gem_object, + exec_list); + entry = obj->exec_entry; + + need_fence = + has_fenced_gpu_access && + entry->flags & EXEC_OBJECT_NEEDS_FENCE && + obj->tiling_mode != I915_TILING_NONE; + need_mappable = + entry->relocation_count ? true : need_fence; + + if (need_mappable) + list_move(&obj->exec_list, &ordered_objects); + else + list_move_tail(&obj->exec_list, &ordered_objects); + + obj->base.pending_read_domains = 0; + obj->base.pending_write_domain = 0; + } + list_splice(&ordered_objects, objects); + + /* Attempt to pin all of the buffers into the GTT. + * This is done in 3 phases: + * + * 1a. Unbind all objects that do not match the GTT constraints for + * the execbuffer (fenceable, mappable, alignment etc). + * 1b. Increment pin count for already bound objects. + * 2. Bind new objects. + * 3. Decrement pin count. + * + * This avoid unnecessary unbinding of later objects in order to makr + * room for the earlier objects *unless* we need to defragment. + */ + retry = 0; + do { + ret = 0; + + /* Unbind any ill-fitting objects or pin. */ + list_for_each_entry(obj, objects, exec_list) { + struct drm_i915_gem_exec_object2 *entry = obj->exec_entry; + bool need_fence, need_mappable; + if (!obj->gtt_space) + continue; + + need_fence = + has_fenced_gpu_access && + entry->flags & EXEC_OBJECT_NEEDS_FENCE && + obj->tiling_mode != I915_TILING_NONE; + need_mappable = + entry->relocation_count ? true : need_fence; + + if ((entry->alignment && obj->gtt_offset & (entry->alignment - 1)) || + (need_mappable && !obj->map_and_fenceable)) + ret = i915_gem_object_unbind(obj); + else + ret = i915_gem_object_pin(obj, + entry->alignment, + need_mappable); + if (ret) + goto err; + + entry++; + } + + /* Bind fresh objects */ + list_for_each_entry(obj, objects, exec_list) { + struct drm_i915_gem_exec_object2 *entry = obj->exec_entry; + bool need_fence; + + need_fence = + has_fenced_gpu_access && + entry->flags & EXEC_OBJECT_NEEDS_FENCE && + obj->tiling_mode != I915_TILING_NONE; + + if (!obj->gtt_space) { + bool need_mappable = + entry->relocation_count ? true : need_fence; + + ret = i915_gem_object_pin(obj, + entry->alignment, + need_mappable); + if (ret) + break; + } + + if (has_fenced_gpu_access) { + if (need_fence) { + ret = i915_gem_object_get_fence(obj, ring); + if (ret) + break; + } else if (entry->flags & EXEC_OBJECT_NEEDS_FENCE && + obj->tiling_mode == I915_TILING_NONE) { + /* XXX pipelined! */ + ret = i915_gem_object_put_fence(obj); + if (ret) + break; + } + obj->pending_fenced_gpu_access = need_fence; + } + + entry->offset = obj->gtt_offset; + } + + /* Decrement pin count for bound objects */ + list_for_each_entry(obj, objects, exec_list) { + if (obj->gtt_space) + i915_gem_object_unpin(obj); + } + + if (ret != -ENOSPC || retry > 1) + return ret; + + /* First attempt, just clear anything that is purgeable. + * Second attempt, clear the entire GTT. + */ + ret = i915_gem_evict_everything(ring->dev, retry == 0); + if (ret) + return ret; + + retry++; + } while (1); + +err: + obj = list_entry(obj->exec_list.prev, + struct drm_i915_gem_object, + exec_list); + while (objects != &obj->exec_list) { + if (obj->gtt_space) + i915_gem_object_unpin(obj); + + obj = list_entry(obj->exec_list.prev, + struct drm_i915_gem_object, + exec_list); + } + + return ret; +} + +static int +i915_gem_execbuffer_relocate_slow(struct drm_device *dev, + struct drm_file *file, + struct intel_ring_buffer *ring, + struct list_head *objects, + struct eb_objects *eb, + struct drm_i915_gem_exec_object2 *exec, + int count) +{ + struct drm_i915_gem_relocation_entry *reloc; + struct drm_i915_gem_object *obj; + int *reloc_offset; + int i, total, ret; + + /* We may process another execbuffer during the unlock... */ + while (!list_empty(objects)) { + obj = list_first_entry(objects, + struct drm_i915_gem_object, + exec_list); + list_del_init(&obj->exec_list); + drm_gem_object_unreference(&obj->base); + } + + mutex_unlock(&dev->struct_mutex); + + total = 0; + for (i = 0; i < count; i++) + total += exec[i].relocation_count; + + reloc_offset = drm_malloc_ab(count, sizeof(*reloc_offset)); + reloc = drm_malloc_ab(total, sizeof(*reloc)); + if (reloc == NULL || reloc_offset == NULL) { + drm_free_large(reloc); + drm_free_large(reloc_offset); + mutex_lock(&dev->struct_mutex); + return -ENOMEM; + } + + total = 0; + for (i = 0; i < count; i++) { + struct drm_i915_gem_relocation_entry __user *user_relocs; + + user_relocs = (void __user *)(uintptr_t)exec[i].relocs_ptr; + + if (copy_from_user(reloc+total, user_relocs, + exec[i].relocation_count * sizeof(*reloc))) { + ret = -EFAULT; + mutex_lock(&dev->struct_mutex); + goto err; + } + + reloc_offset[i] = total; + total += exec[i].relocation_count; + } + + ret = i915_mutex_lock_interruptible(dev); + if (ret) { + mutex_lock(&dev->struct_mutex); + goto err; + } + + /* reacquire the objects */ + eb_reset(eb); + for (i = 0; i < count; i++) { + obj = to_intel_bo(drm_gem_object_lookup(dev, file, + exec[i].handle)); + if (&obj->base == NULL) { + DRM_ERROR("Invalid object handle %d at index %d\n", + exec[i].handle, i); + ret = -ENOENT; + goto err; + } + + list_add_tail(&obj->exec_list, objects); + obj->exec_handle = exec[i].handle; + obj->exec_entry = &exec[i]; + eb_add_object(eb, obj); + } + + ret = i915_gem_execbuffer_reserve(ring, file, objects); + if (ret) + goto err; + + list_for_each_entry(obj, objects, exec_list) { + int offset = obj->exec_entry - exec; + ret = i915_gem_execbuffer_relocate_object_slow(obj, eb, + reloc + reloc_offset[offset]); + if (ret) + goto err; + } + + /* Leave the user relocations as are, this is the painfully slow path, + * and we want to avoid the complication of dropping the lock whilst + * having buffers reserved in the aperture and so causing spurious + * ENOSPC for random operations. + */ + +err: + drm_free_large(reloc); + drm_free_large(reloc_offset); + return ret; +} + +static int +i915_gem_execbuffer_flush(struct drm_device *dev, + uint32_t invalidate_domains, + uint32_t flush_domains, + uint32_t flush_rings) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + int i, ret; + + if (flush_domains & I915_GEM_DOMAIN_CPU) + intel_gtt_chipset_flush(); + + if (flush_domains & I915_GEM_DOMAIN_GTT) + wmb(); + + if ((flush_domains | invalidate_domains) & I915_GEM_GPU_DOMAINS) { + for (i = 0; i < I915_NUM_RINGS; i++) + if (flush_rings & (1 << i)) { + ret = i915_gem_flush_ring(&dev_priv->ring[i], + invalidate_domains, + flush_domains); + if (ret) + return ret; + } + } + + return 0; +} + +static int +i915_gem_execbuffer_sync_rings(struct drm_i915_gem_object *obj, + struct intel_ring_buffer *to) +{ + struct intel_ring_buffer *from = obj->ring; + u32 seqno; + int ret, idx; + + if (from == NULL || to == from) + return 0; + + /* XXX gpu semaphores are implicated in various hard hangs on SNB */ + if (INTEL_INFO(obj->base.dev)->gen < 6 || !i915_semaphores) + return i915_gem_object_wait_rendering(obj); + + idx = intel_ring_sync_index(from, to); + + seqno = obj->last_rendering_seqno; + if (seqno <= from->sync_seqno[idx]) + return 0; + + if (seqno == from->outstanding_lazy_request) { + struct drm_i915_gem_request *request; + + request = kzalloc(sizeof(*request), GFP_KERNEL); + if (request == NULL) + return -ENOMEM; + + ret = i915_add_request(from, NULL, request); + if (ret) { + kfree(request); + return ret; + } + + seqno = request->seqno; + } + + from->sync_seqno[idx] = seqno; + return intel_ring_sync(to, from, seqno - 1); +} + +static int +i915_gem_execbuffer_wait_for_flips(struct intel_ring_buffer *ring, u32 flips) +{ + u32 plane, flip_mask; + int ret; + + /* Check for any pending flips. As we only maintain a flip queue depth + * of 1, we can simply insert a WAIT for the next display flip prior + * to executing the batch and avoid stalling the CPU. + */ + + for (plane = 0; flips >> plane; plane++) { + if (((flips >> plane) & 1) == 0) + continue; + + if (plane) + flip_mask = MI_WAIT_FOR_PLANE_B_FLIP; + else + flip_mask = MI_WAIT_FOR_PLANE_A_FLIP; + + ret = intel_ring_begin(ring, 2); + if (ret) + return ret; + + intel_ring_emit(ring, MI_WAIT_FOR_EVENT | flip_mask); + intel_ring_emit(ring, MI_NOOP); + intel_ring_advance(ring); + } + + return 0; +} + + +static int +i915_gem_execbuffer_move_to_gpu(struct intel_ring_buffer *ring, + struct list_head *objects) +{ + struct drm_i915_gem_object *obj; + struct change_domains cd; + int ret; + + memset(&cd, 0, sizeof(cd)); + list_for_each_entry(obj, objects, exec_list) + i915_gem_object_set_to_gpu_domain(obj, ring, &cd); + + if (cd.invalidate_domains | cd.flush_domains) { + ret = i915_gem_execbuffer_flush(ring->dev, + cd.invalidate_domains, + cd.flush_domains, + cd.flush_rings); + if (ret) + return ret; + } + + if (cd.flips) { + ret = i915_gem_execbuffer_wait_for_flips(ring, cd.flips); + if (ret) + return ret; + } + + list_for_each_entry(obj, objects, exec_list) { + ret = i915_gem_execbuffer_sync_rings(obj, ring); + if (ret) + return ret; + } + + return 0; +} + +static bool +i915_gem_check_execbuffer(struct drm_i915_gem_execbuffer2 *exec) +{ + return ((exec->batch_start_offset | exec->batch_len) & 0x7) == 0; +} + +static int +validate_exec_list(struct drm_i915_gem_exec_object2 *exec, + int count) +{ + int i; + + for (i = 0; i < count; i++) { + char __user *ptr = (char __user *)(uintptr_t)exec[i].relocs_ptr; + int length; /* limited by fault_in_pages_readable() */ + + /* First check for malicious input causing overflow */ + if (exec[i].relocation_count > + INT_MAX / sizeof(struct drm_i915_gem_relocation_entry)) + return -EINVAL; + + length = exec[i].relocation_count * + sizeof(struct drm_i915_gem_relocation_entry); + if (!access_ok(VERIFY_READ, ptr, length)) + return -EFAULT; + + /* we may also need to update the presumed offsets */ + if (!access_ok(VERIFY_WRITE, ptr, length)) + return -EFAULT; + + if (fault_in_pages_readable(ptr, length)) + return -EFAULT; + } + + return 0; +} + +static void +i915_gem_execbuffer_move_to_active(struct list_head *objects, + struct intel_ring_buffer *ring, + u32 seqno) +{ + struct drm_i915_gem_object *obj; + + list_for_each_entry(obj, objects, exec_list) { + u32 old_read = obj->base.read_domains; + u32 old_write = obj->base.write_domain; + + + obj->base.read_domains = obj->base.pending_read_domains; + obj->base.write_domain = obj->base.pending_write_domain; + obj->fenced_gpu_access = obj->pending_fenced_gpu_access; + + i915_gem_object_move_to_active(obj, ring, seqno); + if (obj->base.write_domain) { + obj->dirty = 1; + obj->pending_gpu_write = true; + list_move_tail(&obj->gpu_write_list, + &ring->gpu_write_list); + intel_mark_busy(ring->dev, obj); + } + + trace_i915_gem_object_change_domain(obj, old_read, old_write); + } +} + +static void +i915_gem_execbuffer_retire_commands(struct drm_device *dev, + struct drm_file *file, + struct intel_ring_buffer *ring) +{ + struct drm_i915_gem_request *request; + u32 invalidate; + + /* + * Ensure that the commands in the batch buffer are + * finished before the interrupt fires. + * + * The sampler always gets flushed on i965 (sigh). + */ + invalidate = I915_GEM_DOMAIN_COMMAND; + if (INTEL_INFO(dev)->gen >= 4) + invalidate |= I915_GEM_DOMAIN_SAMPLER; + if (ring->flush(ring, invalidate, 0)) { + i915_gem_next_request_seqno(ring); + return; + } + + /* Add a breadcrumb for the completion of the batch buffer */ + request = kzalloc(sizeof(*request), GFP_KERNEL); + if (request == NULL || i915_add_request(ring, file, request)) { + i915_gem_next_request_seqno(ring); + kfree(request); + } +} + +static int +i915_gem_do_execbuffer(struct drm_device *dev, void *data, + struct drm_file *file, + struct drm_i915_gem_execbuffer2 *args, + struct drm_i915_gem_exec_object2 *exec) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct list_head objects; + struct eb_objects *eb; + struct drm_i915_gem_object *batch_obj; + struct drm_clip_rect *cliprects = NULL; + struct intel_ring_buffer *ring; + u32 exec_start, exec_len; + u32 seqno; + int ret, mode, i; + + if (!i915_gem_check_execbuffer(args)) { + DRM_ERROR("execbuf with invalid offset/length\n"); + return -EINVAL; + } + + ret = validate_exec_list(exec, args->buffer_count); + if (ret) + return ret; + + switch (args->flags & I915_EXEC_RING_MASK) { + case I915_EXEC_DEFAULT: + case I915_EXEC_RENDER: + ring = &dev_priv->ring[RCS]; + break; + case I915_EXEC_BSD: + if (!HAS_BSD(dev)) { + DRM_ERROR("execbuf with invalid ring (BSD)\n"); + return -EINVAL; + } + ring = &dev_priv->ring[VCS]; + break; + case I915_EXEC_BLT: + if (!HAS_BLT(dev)) { + DRM_ERROR("execbuf with invalid ring (BLT)\n"); + return -EINVAL; + } + ring = &dev_priv->ring[BCS]; + break; + default: + DRM_ERROR("execbuf with unknown ring: %d\n", + (int)(args->flags & I915_EXEC_RING_MASK)); + return -EINVAL; + } + + mode = args->flags & I915_EXEC_CONSTANTS_MASK; + switch (mode) { + case I915_EXEC_CONSTANTS_REL_GENERAL: + case I915_EXEC_CONSTANTS_ABSOLUTE: + case I915_EXEC_CONSTANTS_REL_SURFACE: + if (ring == &dev_priv->ring[RCS] && + mode != dev_priv->relative_constants_mode) { + if (INTEL_INFO(dev)->gen < 4) + return -EINVAL; + + if (INTEL_INFO(dev)->gen > 5 && + mode == I915_EXEC_CONSTANTS_REL_SURFACE) + return -EINVAL; + + ret = intel_ring_begin(ring, 4); + if (ret) + return ret; + + intel_ring_emit(ring, MI_NOOP); + intel_ring_emit(ring, MI_LOAD_REGISTER_IMM(1)); + intel_ring_emit(ring, INSTPM); + intel_ring_emit(ring, + I915_EXEC_CONSTANTS_MASK << 16 | mode); + intel_ring_advance(ring); + + dev_priv->relative_constants_mode = mode; + } + break; + default: + DRM_ERROR("execbuf with unknown constants: %d\n", mode); + return -EINVAL; + } + + if (args->buffer_count < 1) { + DRM_ERROR("execbuf with %d buffers\n", args->buffer_count); + return -EINVAL; + } + + if (args->num_cliprects != 0) { + if (ring != &dev_priv->ring[RCS]) { + DRM_ERROR("clip rectangles are only valid with the render ring\n"); + return -EINVAL; + } + + if (args->num_cliprects > UINT_MAX / sizeof(*cliprects)) { + DRM_DEBUG("execbuf with %u cliprects\n", + args->num_cliprects); + return -EINVAL; + } + cliprects = kmalloc(args->num_cliprects * sizeof(*cliprects), + GFP_KERNEL); + if (cliprects == NULL) { + ret = -ENOMEM; + goto pre_mutex_err; + } + + if (copy_from_user(cliprects, + (struct drm_clip_rect __user *)(uintptr_t) + args->cliprects_ptr, + sizeof(*cliprects)*args->num_cliprects)) { + ret = -EFAULT; + goto pre_mutex_err; + } + } + + ret = i915_mutex_lock_interruptible(dev); + if (ret) + goto pre_mutex_err; + + if (dev_priv->mm.suspended) { + mutex_unlock(&dev->struct_mutex); + ret = -EBUSY; + goto pre_mutex_err; + } + + eb = eb_create(args->buffer_count); + if (eb == NULL) { + mutex_unlock(&dev->struct_mutex); + ret = -ENOMEM; + goto pre_mutex_err; + } + + /* Look up object handles */ + INIT_LIST_HEAD(&objects); + for (i = 0; i < args->buffer_count; i++) { + struct drm_i915_gem_object *obj; + + obj = to_intel_bo(drm_gem_object_lookup(dev, file, + exec[i].handle)); + if (&obj->base == NULL) { + DRM_ERROR("Invalid object handle %d at index %d\n", + exec[i].handle, i); + /* prevent error path from reading uninitialized data */ + ret = -ENOENT; + goto err; + } + + if (!list_empty(&obj->exec_list)) { + DRM_ERROR("Object %p [handle %d, index %d] appears more than once in object list\n", + obj, exec[i].handle, i); + ret = -EINVAL; + goto err; + } + + list_add_tail(&obj->exec_list, &objects); + obj->exec_handle = exec[i].handle; + obj->exec_entry = &exec[i]; + eb_add_object(eb, obj); + } + + /* take note of the batch buffer before we might reorder the lists */ + batch_obj = list_entry(objects.prev, + struct drm_i915_gem_object, + exec_list); + + /* Move the objects en-masse into the GTT, evicting if necessary. */ + ret = i915_gem_execbuffer_reserve(ring, file, &objects); + if (ret) + goto err; + + /* The objects are in their final locations, apply the relocations. */ + ret = i915_gem_execbuffer_relocate(dev, eb, &objects); + if (ret) { + if (ret == -EFAULT) { + ret = i915_gem_execbuffer_relocate_slow(dev, file, ring, + &objects, eb, + exec, + args->buffer_count); + BUG_ON(!mutex_is_locked(&dev->struct_mutex)); + } + if (ret) + goto err; + } + + /* Set the pending read domains for the batch buffer to COMMAND */ + if (batch_obj->base.pending_write_domain) { + DRM_ERROR("Attempting to use self-modifying batch buffer\n"); + ret = -EINVAL; + goto err; + } + batch_obj->base.pending_read_domains |= I915_GEM_DOMAIN_COMMAND; + + ret = i915_gem_execbuffer_move_to_gpu(ring, &objects); + if (ret) + goto err; + + seqno = i915_gem_next_request_seqno(ring); + for (i = 0; i < ARRAY_SIZE(ring->sync_seqno); i++) { + if (seqno < ring->sync_seqno[i]) { + /* The GPU can not handle its semaphore value wrapping, + * so every billion or so execbuffers, we need to stall + * the GPU in order to reset the counters. + */ + ret = i915_gpu_idle(dev); + if (ret) + goto err; + + BUG_ON(ring->sync_seqno[i]); + } + } + + trace_i915_gem_ring_dispatch(ring, seqno); + + exec_start = batch_obj->gtt_offset + args->batch_start_offset; + exec_len = args->batch_len; + if (cliprects) { + for (i = 0; i < args->num_cliprects; i++) { + ret = i915_emit_box(dev, &cliprects[i], + args->DR1, args->DR4); + if (ret) + goto err; + + ret = ring->dispatch_execbuffer(ring, + exec_start, exec_len); + if (ret) + goto err; + } + } else { + ret = ring->dispatch_execbuffer(ring, exec_start, exec_len); + if (ret) + goto err; + } + + i915_gem_execbuffer_move_to_active(&objects, ring, seqno); + i915_gem_execbuffer_retire_commands(dev, file, ring); + +err: + eb_destroy(eb); + while (!list_empty(&objects)) { + struct drm_i915_gem_object *obj; + + obj = list_first_entry(&objects, + struct drm_i915_gem_object, + exec_list); + list_del_init(&obj->exec_list); + drm_gem_object_unreference(&obj->base); + } + + mutex_unlock(&dev->struct_mutex); + +pre_mutex_err: + kfree(cliprects); + return ret; +} + +/* + * Legacy execbuffer just creates an exec2 list from the original exec object + * list array and passes it to the real function. + */ +int +i915_gem_execbuffer(struct drm_device *dev, void *data, + struct drm_file *file) +{ + struct drm_i915_gem_execbuffer *args = data; + struct drm_i915_gem_execbuffer2 exec2; + struct drm_i915_gem_exec_object *exec_list = NULL; + struct drm_i915_gem_exec_object2 *exec2_list = NULL; + int ret, i; + + if (args->buffer_count < 1) { + DRM_ERROR("execbuf with %d buffers\n", args->buffer_count); + return -EINVAL; + } + + /* Copy in the exec list from userland */ + exec_list = drm_malloc_ab(sizeof(*exec_list), args->buffer_count); + exec2_list = drm_malloc_ab(sizeof(*exec2_list), args->buffer_count); + if (exec_list == NULL || exec2_list == NULL) { + DRM_ERROR("Failed to allocate exec list for %d buffers\n", + args->buffer_count); + drm_free_large(exec_list); + drm_free_large(exec2_list); + return -ENOMEM; + } + ret = copy_from_user(exec_list, + (struct drm_i915_relocation_entry __user *) + (uintptr_t) args->buffers_ptr, + sizeof(*exec_list) * args->buffer_count); + if (ret != 0) { + DRM_ERROR("copy %d exec entries failed %d\n", + args->buffer_count, ret); + drm_free_large(exec_list); + drm_free_large(exec2_list); + return -EFAULT; + } + + for (i = 0; i < args->buffer_count; i++) { + exec2_list[i].handle = exec_list[i].handle; + exec2_list[i].relocation_count = exec_list[i].relocation_count; + exec2_list[i].relocs_ptr = exec_list[i].relocs_ptr; + exec2_list[i].alignment = exec_list[i].alignment; + exec2_list[i].offset = exec_list[i].offset; + if (INTEL_INFO(dev)->gen < 4) + exec2_list[i].flags = EXEC_OBJECT_NEEDS_FENCE; + else + exec2_list[i].flags = 0; + } + + exec2.buffers_ptr = args->buffers_ptr; + exec2.buffer_count = args->buffer_count; + exec2.batch_start_offset = args->batch_start_offset; + exec2.batch_len = args->batch_len; + exec2.DR1 = args->DR1; + exec2.DR4 = args->DR4; + exec2.num_cliprects = args->num_cliprects; + exec2.cliprects_ptr = args->cliprects_ptr; + exec2.flags = I915_EXEC_RENDER; + + ret = i915_gem_do_execbuffer(dev, data, file, &exec2, exec2_list); + if (!ret) { + /* Copy the new buffer offsets back to the user's exec list. */ + for (i = 0; i < args->buffer_count; i++) + exec_list[i].offset = exec2_list[i].offset; + /* ... and back out to userspace */ + ret = copy_to_user((struct drm_i915_relocation_entry __user *) + (uintptr_t) args->buffers_ptr, + exec_list, + sizeof(*exec_list) * args->buffer_count); + if (ret) { + ret = -EFAULT; + DRM_ERROR("failed to copy %d exec entries " + "back to user (%d)\n", + args->buffer_count, ret); + } + } + + drm_free_large(exec_list); + drm_free_large(exec2_list); + return ret; +} + +int +i915_gem_execbuffer2(struct drm_device *dev, void *data, + struct drm_file *file) +{ + struct drm_i915_gem_execbuffer2 *args = data; + struct drm_i915_gem_exec_object2 *exec2_list = NULL; + int ret; + + if (args->buffer_count < 1 || + args->buffer_count > UINT_MAX / sizeof(*exec2_list)) { + DRM_ERROR("execbuf2 with %d buffers\n", args->buffer_count); + return -EINVAL; + } + + exec2_list = kmalloc(sizeof(*exec2_list)*args->buffer_count, + GFP_KERNEL | __GFP_NOWARN | __GFP_NORETRY); + if (exec2_list == NULL) + exec2_list = drm_malloc_ab(sizeof(*exec2_list), + args->buffer_count); + if (exec2_list == NULL) { + DRM_ERROR("Failed to allocate exec list for %d buffers\n", + args->buffer_count); + return -ENOMEM; + } + ret = copy_from_user(exec2_list, + (struct drm_i915_relocation_entry __user *) + (uintptr_t) args->buffers_ptr, + sizeof(*exec2_list) * args->buffer_count); + if (ret != 0) { + DRM_ERROR("copy %d exec entries failed %d\n", + args->buffer_count, ret); + drm_free_large(exec2_list); + return -EFAULT; + } + + ret = i915_gem_do_execbuffer(dev, data, file, args, exec2_list); + if (!ret) { + /* Copy the new buffer offsets back to the user's exec list. */ + ret = copy_to_user((struct drm_i915_relocation_entry __user *) + (uintptr_t) args->buffers_ptr, + exec2_list, + sizeof(*exec2_list) * args->buffer_count); + if (ret) { + ret = -EFAULT; + DRM_ERROR("failed to copy %d exec entries " + "back to user (%d)\n", + args->buffer_count, ret); + } + } + + drm_free_large(exec2_list); + return ret; +} diff --git a/drivers/gpu/drm/i915/i915_gem_gtt.c b/drivers/gpu/drm/i915/i915_gem_gtt.c new file mode 100644 index 00000000..e46b6457 --- /dev/null +++ b/drivers/gpu/drm/i915/i915_gem_gtt.c @@ -0,0 +1,122 @@ +/* + * Copyright © 2010 Daniel Vetter + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS + * IN THE SOFTWARE. + * + */ + +#include "drmP.h" +#include "drm.h" +#include "i915_drm.h" +#include "i915_drv.h" +#include "i915_trace.h" +#include "intel_drv.h" + +/* XXX kill agp_type! */ +static unsigned int cache_level_to_agp_type(struct drm_device *dev, + enum i915_cache_level cache_level) +{ + switch (cache_level) { + case I915_CACHE_LLC_MLC: + if (INTEL_INFO(dev)->gen >= 6) + return AGP_USER_CACHED_MEMORY_LLC_MLC; + /* Older chipsets do not have this extra level of CPU + * cacheing, so fallthrough and request the PTE simply + * as cached. + */ + case I915_CACHE_LLC: + return AGP_USER_CACHED_MEMORY; + default: + case I915_CACHE_NONE: + return AGP_USER_MEMORY; + } +} + +void i915_gem_restore_gtt_mappings(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_i915_gem_object *obj; + + /* First fill our portion of the GTT with scratch pages */ + intel_gtt_clear_range(dev_priv->mm.gtt_start / PAGE_SIZE, + (dev_priv->mm.gtt_end - dev_priv->mm.gtt_start) / PAGE_SIZE); + + list_for_each_entry(obj, &dev_priv->mm.gtt_list, gtt_list) { + unsigned int agp_type = + cache_level_to_agp_type(dev, obj->cache_level); + + i915_gem_clflush_object(obj); + + if (dev_priv->mm.gtt->needs_dmar) { + BUG_ON(!obj->sg_list); + + intel_gtt_insert_sg_entries(obj->sg_list, + obj->num_sg, + obj->gtt_space->start >> PAGE_SHIFT, + agp_type); + } else + intel_gtt_insert_pages(obj->gtt_space->start + >> PAGE_SHIFT, + obj->base.size >> PAGE_SHIFT, + obj->pages, + agp_type); + } + + intel_gtt_chipset_flush(); +} + +int i915_gem_gtt_bind_object(struct drm_i915_gem_object *obj) +{ + struct drm_device *dev = obj->base.dev; + struct drm_i915_private *dev_priv = dev->dev_private; + unsigned int agp_type = cache_level_to_agp_type(dev, obj->cache_level); + int ret; + + if (dev_priv->mm.gtt->needs_dmar) { + ret = intel_gtt_map_memory(obj->pages, + obj->base.size >> PAGE_SHIFT, + &obj->sg_list, + &obj->num_sg); + if (ret != 0) + return ret; + + intel_gtt_insert_sg_entries(obj->sg_list, + obj->num_sg, + obj->gtt_space->start >> PAGE_SHIFT, + agp_type); + } else + intel_gtt_insert_pages(obj->gtt_space->start >> PAGE_SHIFT, + obj->base.size >> PAGE_SHIFT, + obj->pages, + agp_type); + + return 0; +} + +void i915_gem_gtt_unbind_object(struct drm_i915_gem_object *obj) +{ + intel_gtt_clear_range(obj->gtt_space->start >> PAGE_SHIFT, + obj->base.size >> PAGE_SHIFT); + + if (obj->sg_list) { + intel_gtt_unmap_memory(obj->sg_list, obj->num_sg); + obj->sg_list = NULL; + } +} diff --git a/drivers/gpu/drm/i915/i915_gem_tiling.c b/drivers/gpu/drm/i915/i915_gem_tiling.c new file mode 100644 index 00000000..99c4faa5 --- /dev/null +++ b/drivers/gpu/drm/i915/i915_gem_tiling.c @@ -0,0 +1,491 @@ +/* + * Copyright © 2008 Intel Corporation + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS + * IN THE SOFTWARE. + * + * Authors: + * Eric Anholt + * + */ + +#include "linux/string.h" +#include "linux/bitops.h" +#include "drmP.h" +#include "drm.h" +#include "i915_drm.h" +#include "i915_drv.h" + +/** @file i915_gem_tiling.c + * + * Support for managing tiling state of buffer objects. + * + * The idea behind tiling is to increase cache hit rates by rearranging + * pixel data so that a group of pixel accesses are in the same cacheline. + * Performance improvement from doing this on the back/depth buffer are on + * the order of 30%. + * + * Intel architectures make this somewhat more complicated, though, by + * adjustments made to addressing of data when the memory is in interleaved + * mode (matched pairs of DIMMS) to improve memory bandwidth. + * For interleaved memory, the CPU sends every sequential 64 bytes + * to an alternate memory channel so it can get the bandwidth from both. + * + * The GPU also rearranges its accesses for increased bandwidth to interleaved + * memory, and it matches what the CPU does for non-tiled. However, when tiled + * it does it a little differently, since one walks addresses not just in the + * X direction but also Y. So, along with alternating channels when bit + * 6 of the address flips, it also alternates when other bits flip -- Bits 9 + * (every 512 bytes, an X tile scanline) and 10 (every two X tile scanlines) + * are common to both the 915 and 965-class hardware. + * + * The CPU also sometimes XORs in higher bits as well, to improve + * bandwidth doing strided access like we do so frequently in graphics. This + * is called "Channel XOR Randomization" in the MCH documentation. The result + * is that the CPU is XORing in either bit 11 or bit 17 to bit 6 of its address + * decode. + * + * All of this bit 6 XORing has an effect on our memory management, + * as we need to make sure that the 3d driver can correctly address object + * contents. + * + * If we don't have interleaved memory, all tiling is safe and no swizzling is + * required. + * + * When bit 17 is XORed in, we simply refuse to tile at all. Bit + * 17 is not just a page offset, so as we page an objet out and back in, + * individual pages in it will have different bit 17 addresses, resulting in + * each 64 bytes being swapped with its neighbor! + * + * Otherwise, if interleaved, we have to tell the 3d driver what the address + * swizzling it needs to do is, since it's writing with the CPU to the pages + * (bit 6 and potentially bit 11 XORed in), and the GPU is reading from the + * pages (bit 6, 9, and 10 XORed in), resulting in a cumulative bit swizzling + * required by the CPU of XORing in bit 6, 9, 10, and potentially 11, in order + * to match what the GPU expects. + */ + +/** + * Detects bit 6 swizzling of address lookup between IGD access and CPU + * access through main memory. + */ +void +i915_gem_detect_bit_6_swizzle(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + uint32_t swizzle_x = I915_BIT_6_SWIZZLE_UNKNOWN; + uint32_t swizzle_y = I915_BIT_6_SWIZZLE_UNKNOWN; + + if (INTEL_INFO(dev)->gen >= 5) { + /* On Ironlake whatever DRAM config, GPU always do + * same swizzling setup. + */ + swizzle_x = I915_BIT_6_SWIZZLE_9_10; + swizzle_y = I915_BIT_6_SWIZZLE_9; + } else if (IS_GEN2(dev)) { + /* As far as we know, the 865 doesn't have these bit 6 + * swizzling issues. + */ + swizzle_x = I915_BIT_6_SWIZZLE_NONE; + swizzle_y = I915_BIT_6_SWIZZLE_NONE; + } else if (IS_MOBILE(dev)) { + uint32_t dcc; + + /* On mobile 9xx chipsets, channel interleave by the CPU is + * determined by DCC. For single-channel, neither the CPU + * nor the GPU do swizzling. For dual channel interleaved, + * the GPU's interleave is bit 9 and 10 for X tiled, and bit + * 9 for Y tiled. The CPU's interleave is independent, and + * can be based on either bit 11 (haven't seen this yet) or + * bit 17 (common). + */ + dcc = I915_READ(DCC); + switch (dcc & DCC_ADDRESSING_MODE_MASK) { + case DCC_ADDRESSING_MODE_SINGLE_CHANNEL: + case DCC_ADDRESSING_MODE_DUAL_CHANNEL_ASYMMETRIC: + swizzle_x = I915_BIT_6_SWIZZLE_NONE; + swizzle_y = I915_BIT_6_SWIZZLE_NONE; + break; + case DCC_ADDRESSING_MODE_DUAL_CHANNEL_INTERLEAVED: + if (dcc & DCC_CHANNEL_XOR_DISABLE) { + /* This is the base swizzling by the GPU for + * tiled buffers. + */ + swizzle_x = I915_BIT_6_SWIZZLE_9_10; + swizzle_y = I915_BIT_6_SWIZZLE_9; + } else if ((dcc & DCC_CHANNEL_XOR_BIT_17) == 0) { + /* Bit 11 swizzling by the CPU in addition. */ + swizzle_x = I915_BIT_6_SWIZZLE_9_10_11; + swizzle_y = I915_BIT_6_SWIZZLE_9_11; + } else { + /* Bit 17 swizzling by the CPU in addition. */ + swizzle_x = I915_BIT_6_SWIZZLE_9_10_17; + swizzle_y = I915_BIT_6_SWIZZLE_9_17; + } + break; + } + if (dcc == 0xffffffff) { + DRM_ERROR("Couldn't read from MCHBAR. " + "Disabling tiling.\n"); + swizzle_x = I915_BIT_6_SWIZZLE_UNKNOWN; + swizzle_y = I915_BIT_6_SWIZZLE_UNKNOWN; + } + } else { + /* The 965, G33, and newer, have a very flexible memory + * configuration. It will enable dual-channel mode + * (interleaving) on as much memory as it can, and the GPU + * will additionally sometimes enable different bit 6 + * swizzling for tiled objects from the CPU. + * + * Here's what I found on the G965: + * slot fill memory size swizzling + * 0A 0B 1A 1B 1-ch 2-ch + * 512 0 0 0 512 0 O + * 512 0 512 0 16 1008 X + * 512 0 0 512 16 1008 X + * 0 512 0 512 16 1008 X + * 1024 1024 1024 0 2048 1024 O + * + * We could probably detect this based on either the DRB + * matching, which was the case for the swizzling required in + * the table above, or from the 1-ch value being less than + * the minimum size of a rank. + */ + if (I915_READ16(C0DRB3) != I915_READ16(C1DRB3)) { + swizzle_x = I915_BIT_6_SWIZZLE_NONE; + swizzle_y = I915_BIT_6_SWIZZLE_NONE; + } else { + swizzle_x = I915_BIT_6_SWIZZLE_9_10; + swizzle_y = I915_BIT_6_SWIZZLE_9; + } + } + + dev_priv->mm.bit_6_swizzle_x = swizzle_x; + dev_priv->mm.bit_6_swizzle_y = swizzle_y; +} + +/* Check pitch constriants for all chips & tiling formats */ +static bool +i915_tiling_ok(struct drm_device *dev, int stride, int size, int tiling_mode) +{ + int tile_width; + + /* Linear is always fine */ + if (tiling_mode == I915_TILING_NONE) + return true; + + if (IS_GEN2(dev) || + (tiling_mode == I915_TILING_Y && HAS_128_BYTE_Y_TILING(dev))) + tile_width = 128; + else + tile_width = 512; + + /* check maximum stride & object size */ + if (INTEL_INFO(dev)->gen >= 4) { + /* i965 stores the end address of the gtt mapping in the fence + * reg, so dont bother to check the size */ + if (stride / 128 > I965_FENCE_MAX_PITCH_VAL) + return false; + } else { + if (stride > 8192) + return false; + + if (IS_GEN3(dev)) { + if (size > I830_FENCE_MAX_SIZE_VAL << 20) + return false; + } else { + if (size > I830_FENCE_MAX_SIZE_VAL << 19) + return false; + } + } + + /* 965+ just needs multiples of tile width */ + if (INTEL_INFO(dev)->gen >= 4) { + if (stride & (tile_width - 1)) + return false; + return true; + } + + /* Pre-965 needs power of two tile widths */ + if (stride < tile_width) + return false; + + if (stride & (stride - 1)) + return false; + + return true; +} + +/* Is the current GTT allocation valid for the change in tiling? */ +static bool +i915_gem_object_fence_ok(struct drm_i915_gem_object *obj, int tiling_mode) +{ + u32 size; + + if (tiling_mode == I915_TILING_NONE) + return true; + + if (INTEL_INFO(obj->base.dev)->gen >= 4) + return true; + + if (INTEL_INFO(obj->base.dev)->gen == 3) { + if (obj->gtt_offset & ~I915_FENCE_START_MASK) + return false; + } else { + if (obj->gtt_offset & ~I830_FENCE_START_MASK) + return false; + } + + /* + * Previous chips need to be aligned to the size of the smallest + * fence register that can contain the object. + */ + if (INTEL_INFO(obj->base.dev)->gen == 3) + size = 1024*1024; + else + size = 512*1024; + + while (size < obj->base.size) + size <<= 1; + + if (obj->gtt_space->size != size) + return false; + + if (obj->gtt_offset & (size - 1)) + return false; + + return true; +} + +/** + * Sets the tiling mode of an object, returning the required swizzling of + * bit 6 of addresses in the object. + */ +int +i915_gem_set_tiling(struct drm_device *dev, void *data, + struct drm_file *file) +{ + struct drm_i915_gem_set_tiling *args = data; + drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_i915_gem_object *obj; + int ret = 0; + + obj = to_intel_bo(drm_gem_object_lookup(dev, file, args->handle)); + if (&obj->base == NULL) + return -ENOENT; + + if (!i915_tiling_ok(dev, + args->stride, obj->base.size, args->tiling_mode)) { + drm_gem_object_unreference_unlocked(&obj->base); + return -EINVAL; + } + + if (obj->pin_count) { + drm_gem_object_unreference_unlocked(&obj->base); + return -EBUSY; + } + + if (args->tiling_mode == I915_TILING_NONE) { + args->swizzle_mode = I915_BIT_6_SWIZZLE_NONE; + args->stride = 0; + } else { + if (args->tiling_mode == I915_TILING_X) + args->swizzle_mode = dev_priv->mm.bit_6_swizzle_x; + else + args->swizzle_mode = dev_priv->mm.bit_6_swizzle_y; + + /* Hide bit 17 swizzling from the user. This prevents old Mesa + * from aborting the application on sw fallbacks to bit 17, + * and we use the pread/pwrite bit17 paths to swizzle for it. + * If there was a user that was relying on the swizzle + * information for drm_intel_bo_map()ed reads/writes this would + * break it, but we don't have any of those. + */ + if (args->swizzle_mode == I915_BIT_6_SWIZZLE_9_17) + args->swizzle_mode = I915_BIT_6_SWIZZLE_9; + if (args->swizzle_mode == I915_BIT_6_SWIZZLE_9_10_17) + args->swizzle_mode = I915_BIT_6_SWIZZLE_9_10; + + /* If we can't handle the swizzling, make it untiled. */ + if (args->swizzle_mode == I915_BIT_6_SWIZZLE_UNKNOWN) { + args->tiling_mode = I915_TILING_NONE; + args->swizzle_mode = I915_BIT_6_SWIZZLE_NONE; + args->stride = 0; + } + } + + mutex_lock(&dev->struct_mutex); + if (args->tiling_mode != obj->tiling_mode || + args->stride != obj->stride) { + /* We need to rebind the object if its current allocation + * no longer meets the alignment restrictions for its new + * tiling mode. Otherwise we can just leave it alone, but + * need to ensure that any fence register is cleared. + */ + i915_gem_release_mmap(obj); + + obj->map_and_fenceable = + obj->gtt_space == NULL || + (obj->gtt_offset + obj->base.size <= dev_priv->mm.gtt_mappable_end && + i915_gem_object_fence_ok(obj, args->tiling_mode)); + + /* Rebind if we need a change of alignment */ + if (!obj->map_and_fenceable) { + u32 unfenced_alignment = + i915_gem_get_unfenced_gtt_alignment(dev, + obj->base.size, + args->tiling_mode); + if (obj->gtt_offset & (unfenced_alignment - 1)) + ret = i915_gem_object_unbind(obj); + } + + if (ret == 0) { + obj->tiling_changed = true; + obj->tiling_mode = args->tiling_mode; + obj->stride = args->stride; + } + } + /* we have to maintain this existing ABI... */ + args->stride = obj->stride; + args->tiling_mode = obj->tiling_mode; + drm_gem_object_unreference(&obj->base); + mutex_unlock(&dev->struct_mutex); + + return ret; +} + +/** + * Returns the current tiling mode and required bit 6 swizzling for the object. + */ +int +i915_gem_get_tiling(struct drm_device *dev, void *data, + struct drm_file *file) +{ + struct drm_i915_gem_get_tiling *args = data; + drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_i915_gem_object *obj; + + obj = to_intel_bo(drm_gem_object_lookup(dev, file, args->handle)); + if (&obj->base == NULL) + return -ENOENT; + + mutex_lock(&dev->struct_mutex); + + args->tiling_mode = obj->tiling_mode; + switch (obj->tiling_mode) { + case I915_TILING_X: + args->swizzle_mode = dev_priv->mm.bit_6_swizzle_x; + break; + case I915_TILING_Y: + args->swizzle_mode = dev_priv->mm.bit_6_swizzle_y; + break; + case I915_TILING_NONE: + args->swizzle_mode = I915_BIT_6_SWIZZLE_NONE; + break; + default: + DRM_ERROR("unknown tiling mode\n"); + } + + /* Hide bit 17 from the user -- see comment in i915_gem_set_tiling */ + if (args->swizzle_mode == I915_BIT_6_SWIZZLE_9_17) + args->swizzle_mode = I915_BIT_6_SWIZZLE_9; + if (args->swizzle_mode == I915_BIT_6_SWIZZLE_9_10_17) + args->swizzle_mode = I915_BIT_6_SWIZZLE_9_10; + + drm_gem_object_unreference(&obj->base); + mutex_unlock(&dev->struct_mutex); + + return 0; +} + +/** + * Swap every 64 bytes of this page around, to account for it having a new + * bit 17 of its physical address and therefore being interpreted differently + * by the GPU. + */ +static void +i915_gem_swizzle_page(struct page *page) +{ + char temp[64]; + char *vaddr; + int i; + + vaddr = kmap(page); + + for (i = 0; i < PAGE_SIZE; i += 128) { + memcpy(temp, &vaddr[i], 64); + memcpy(&vaddr[i], &vaddr[i + 64], 64); + memcpy(&vaddr[i + 64], temp, 64); + } + + kunmap(page); +} + +void +i915_gem_object_do_bit_17_swizzle(struct drm_i915_gem_object *obj) +{ + struct drm_device *dev = obj->base.dev; + drm_i915_private_t *dev_priv = dev->dev_private; + int page_count = obj->base.size >> PAGE_SHIFT; + int i; + + if (dev_priv->mm.bit_6_swizzle_x != I915_BIT_6_SWIZZLE_9_10_17) + return; + + if (obj->bit_17 == NULL) + return; + + for (i = 0; i < page_count; i++) { + char new_bit_17 = page_to_phys(obj->pages[i]) >> 17; + if ((new_bit_17 & 0x1) != + (test_bit(i, obj->bit_17) != 0)) { + i915_gem_swizzle_page(obj->pages[i]); + set_page_dirty(obj->pages[i]); + } + } +} + +void +i915_gem_object_save_bit_17_swizzle(struct drm_i915_gem_object *obj) +{ + struct drm_device *dev = obj->base.dev; + drm_i915_private_t *dev_priv = dev->dev_private; + int page_count = obj->base.size >> PAGE_SHIFT; + int i; + + if (dev_priv->mm.bit_6_swizzle_x != I915_BIT_6_SWIZZLE_9_10_17) + return; + + if (obj->bit_17 == NULL) { + obj->bit_17 = kmalloc(BITS_TO_LONGS(page_count) * + sizeof(long), GFP_KERNEL); + if (obj->bit_17 == NULL) { + DRM_ERROR("Failed to allocate memory for bit 17 " + "record\n"); + return; + } + } + + for (i = 0; i < page_count; i++) { + if (page_to_phys(obj->pages[i]) & (1 << 17)) + __set_bit(i, obj->bit_17); + else + __clear_bit(i, obj->bit_17); + } +} diff --git a/drivers/gpu/drm/i915/i915_ioc32.c b/drivers/gpu/drm/i915/i915_ioc32.c new file mode 100644 index 00000000..13b02899 --- /dev/null +++ b/drivers/gpu/drm/i915/i915_ioc32.c @@ -0,0 +1,219 @@ +/** + * \file i915_ioc32.c + * + * 32-bit ioctl compatibility routines for the i915 DRM. + * + * \author Alan Hourihane + * + * + * Copyright (C) Paul Mackerras 2005 + * Copyright (C) Alan Hourihane 2005 + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHOR BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS + * IN THE SOFTWARE. + */ +#include + +#include "drmP.h" +#include "drm.h" +#include "i915_drm.h" + +typedef struct _drm_i915_batchbuffer32 { + int start; /* agp offset */ + int used; /* nr bytes in use */ + int DR1; /* hw flags for GFX_OP_DRAWRECT_INFO */ + int DR4; /* window origin for GFX_OP_DRAWRECT_INFO */ + int num_cliprects; /* mulitpass with multiple cliprects? */ + u32 cliprects; /* pointer to userspace cliprects */ +} drm_i915_batchbuffer32_t; + +static int compat_i915_batchbuffer(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_i915_batchbuffer32_t batchbuffer32; + drm_i915_batchbuffer_t __user *batchbuffer; + + if (copy_from_user + (&batchbuffer32, (void __user *)arg, sizeof(batchbuffer32))) + return -EFAULT; + + batchbuffer = compat_alloc_user_space(sizeof(*batchbuffer)); + if (!access_ok(VERIFY_WRITE, batchbuffer, sizeof(*batchbuffer)) + || __put_user(batchbuffer32.start, &batchbuffer->start) + || __put_user(batchbuffer32.used, &batchbuffer->used) + || __put_user(batchbuffer32.DR1, &batchbuffer->DR1) + || __put_user(batchbuffer32.DR4, &batchbuffer->DR4) + || __put_user(batchbuffer32.num_cliprects, + &batchbuffer->num_cliprects) + || __put_user((int __user *)(unsigned long)batchbuffer32.cliprects, + &batchbuffer->cliprects)) + return -EFAULT; + + return drm_ioctl(file, DRM_IOCTL_I915_BATCHBUFFER, + (unsigned long)batchbuffer); +} + +typedef struct _drm_i915_cmdbuffer32 { + u32 buf; /* pointer to userspace command buffer */ + int sz; /* nr bytes in buf */ + int DR1; /* hw flags for GFX_OP_DRAWRECT_INFO */ + int DR4; /* window origin for GFX_OP_DRAWRECT_INFO */ + int num_cliprects; /* mulitpass with multiple cliprects? */ + u32 cliprects; /* pointer to userspace cliprects */ +} drm_i915_cmdbuffer32_t; + +static int compat_i915_cmdbuffer(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_i915_cmdbuffer32_t cmdbuffer32; + drm_i915_cmdbuffer_t __user *cmdbuffer; + + if (copy_from_user + (&cmdbuffer32, (void __user *)arg, sizeof(cmdbuffer32))) + return -EFAULT; + + cmdbuffer = compat_alloc_user_space(sizeof(*cmdbuffer)); + if (!access_ok(VERIFY_WRITE, cmdbuffer, sizeof(*cmdbuffer)) + || __put_user((int __user *)(unsigned long)cmdbuffer32.buf, + &cmdbuffer->buf) + || __put_user(cmdbuffer32.sz, &cmdbuffer->sz) + || __put_user(cmdbuffer32.DR1, &cmdbuffer->DR1) + || __put_user(cmdbuffer32.DR4, &cmdbuffer->DR4) + || __put_user(cmdbuffer32.num_cliprects, &cmdbuffer->num_cliprects) + || __put_user((int __user *)(unsigned long)cmdbuffer32.cliprects, + &cmdbuffer->cliprects)) + return -EFAULT; + + return drm_ioctl(file, DRM_IOCTL_I915_CMDBUFFER, + (unsigned long)cmdbuffer); +} + +typedef struct drm_i915_irq_emit32 { + u32 irq_seq; +} drm_i915_irq_emit32_t; + +static int compat_i915_irq_emit(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_i915_irq_emit32_t req32; + drm_i915_irq_emit_t __user *request; + + if (copy_from_user(&req32, (void __user *)arg, sizeof(req32))) + return -EFAULT; + + request = compat_alloc_user_space(sizeof(*request)); + if (!access_ok(VERIFY_WRITE, request, sizeof(*request)) + || __put_user((int __user *)(unsigned long)req32.irq_seq, + &request->irq_seq)) + return -EFAULT; + + return drm_ioctl(file, DRM_IOCTL_I915_IRQ_EMIT, + (unsigned long)request); +} +typedef struct drm_i915_getparam32 { + int param; + u32 value; +} drm_i915_getparam32_t; + +static int compat_i915_getparam(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_i915_getparam32_t req32; + drm_i915_getparam_t __user *request; + + if (copy_from_user(&req32, (void __user *)arg, sizeof(req32))) + return -EFAULT; + + request = compat_alloc_user_space(sizeof(*request)); + if (!access_ok(VERIFY_WRITE, request, sizeof(*request)) + || __put_user(req32.param, &request->param) + || __put_user((void __user *)(unsigned long)req32.value, + &request->value)) + return -EFAULT; + + return drm_ioctl(file, DRM_IOCTL_I915_GETPARAM, + (unsigned long)request); +} + +typedef struct drm_i915_mem_alloc32 { + int region; + int alignment; + int size; + u32 region_offset; /* offset from start of fb or agp */ +} drm_i915_mem_alloc32_t; + +static int compat_i915_alloc(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_i915_mem_alloc32_t req32; + drm_i915_mem_alloc_t __user *request; + + if (copy_from_user(&req32, (void __user *)arg, sizeof(req32))) + return -EFAULT; + + request = compat_alloc_user_space(sizeof(*request)); + if (!access_ok(VERIFY_WRITE, request, sizeof(*request)) + || __put_user(req32.region, &request->region) + || __put_user(req32.alignment, &request->alignment) + || __put_user(req32.size, &request->size) + || __put_user((void __user *)(unsigned long)req32.region_offset, + &request->region_offset)) + return -EFAULT; + + return drm_ioctl(file, DRM_IOCTL_I915_ALLOC, + (unsigned long)request); +} + +drm_ioctl_compat_t *i915_compat_ioctls[] = { + [DRM_I915_BATCHBUFFER] = compat_i915_batchbuffer, + [DRM_I915_CMDBUFFER] = compat_i915_cmdbuffer, + [DRM_I915_GETPARAM] = compat_i915_getparam, + [DRM_I915_IRQ_EMIT] = compat_i915_irq_emit, + [DRM_I915_ALLOC] = compat_i915_alloc +}; + +/** + * Called whenever a 32-bit process running under a 64-bit kernel + * performs an ioctl on /dev/dri/card. + * + * \param filp file pointer. + * \param cmd command. + * \param arg user argument. + * \return zero on success or negative number on failure. + */ +long i915_compat_ioctl(struct file *filp, unsigned int cmd, unsigned long arg) +{ + unsigned int nr = DRM_IOCTL_NR(cmd); + drm_ioctl_compat_t *fn = NULL; + int ret; + + if (nr < DRM_COMMAND_BASE) + return drm_compat_ioctl(filp, cmd, arg); + + if (nr < DRM_COMMAND_BASE + DRM_ARRAY_SIZE(i915_compat_ioctls)) + fn = i915_compat_ioctls[nr - DRM_COMMAND_BASE]; + + if (fn != NULL) + ret = (*fn) (filp, cmd, arg); + else + ret = drm_ioctl(filp, cmd, arg); + + return ret; +} diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c new file mode 100644 index 00000000..d05f03c6 --- /dev/null +++ b/drivers/gpu/drm/i915/i915_irq.c @@ -0,0 +1,2088 @@ +/* i915_irq.c -- IRQ support for the I915 -*- linux-c -*- + */ +/* + * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include +#include +#include "drmP.h" +#include "drm.h" +#include "i915_drm.h" +#include "i915_drv.h" +#include "i915_trace.h" +#include "intel_drv.h" + +#define MAX_NOPID ((u32)~0) + +/** + * Interrupts that are always left unmasked. + * + * Since pipe events are edge-triggered from the PIPESTAT register to IIR, + * we leave them always unmasked in IMR and then control enabling them through + * PIPESTAT alone. + */ +#define I915_INTERRUPT_ENABLE_FIX \ + (I915_ASLE_INTERRUPT | \ + I915_DISPLAY_PIPE_A_EVENT_INTERRUPT | \ + I915_DISPLAY_PIPE_B_EVENT_INTERRUPT | \ + I915_DISPLAY_PLANE_A_FLIP_PENDING_INTERRUPT | \ + I915_DISPLAY_PLANE_B_FLIP_PENDING_INTERRUPT | \ + I915_RENDER_COMMAND_PARSER_ERROR_INTERRUPT) + +/** Interrupts that we mask and unmask at runtime. */ +#define I915_INTERRUPT_ENABLE_VAR (I915_USER_INTERRUPT | I915_BSD_USER_INTERRUPT) + +#define I915_PIPE_VBLANK_STATUS (PIPE_START_VBLANK_INTERRUPT_STATUS |\ + PIPE_VBLANK_INTERRUPT_STATUS) + +#define I915_PIPE_VBLANK_ENABLE (PIPE_START_VBLANK_INTERRUPT_ENABLE |\ + PIPE_VBLANK_INTERRUPT_ENABLE) + +#define DRM_I915_VBLANK_PIPE_ALL (DRM_I915_VBLANK_PIPE_A | \ + DRM_I915_VBLANK_PIPE_B) + +/* For display hotplug interrupt */ +static void +ironlake_enable_display_irq(drm_i915_private_t *dev_priv, u32 mask) +{ + if ((dev_priv->irq_mask & mask) != 0) { + dev_priv->irq_mask &= ~mask; + I915_WRITE(DEIMR, dev_priv->irq_mask); + POSTING_READ(DEIMR); + } +} + +static inline void +ironlake_disable_display_irq(drm_i915_private_t *dev_priv, u32 mask) +{ + if ((dev_priv->irq_mask & mask) != mask) { + dev_priv->irq_mask |= mask; + I915_WRITE(DEIMR, dev_priv->irq_mask); + POSTING_READ(DEIMR); + } +} + +void +i915_enable_pipestat(drm_i915_private_t *dev_priv, int pipe, u32 mask) +{ + if ((dev_priv->pipestat[pipe] & mask) != mask) { + u32 reg = PIPESTAT(pipe); + + dev_priv->pipestat[pipe] |= mask; + /* Enable the interrupt, clear any pending status */ + I915_WRITE(reg, dev_priv->pipestat[pipe] | (mask >> 16)); + POSTING_READ(reg); + } +} + +void +i915_disable_pipestat(drm_i915_private_t *dev_priv, int pipe, u32 mask) +{ + if ((dev_priv->pipestat[pipe] & mask) != 0) { + u32 reg = PIPESTAT(pipe); + + dev_priv->pipestat[pipe] &= ~mask; + I915_WRITE(reg, dev_priv->pipestat[pipe]); + POSTING_READ(reg); + } +} + +/** + * intel_enable_asle - enable ASLE interrupt for OpRegion + */ +void intel_enable_asle(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + unsigned long irqflags; + + spin_lock_irqsave(&dev_priv->irq_lock, irqflags); + + if (HAS_PCH_SPLIT(dev)) + ironlake_enable_display_irq(dev_priv, DE_GSE); + else { + i915_enable_pipestat(dev_priv, 1, + PIPE_LEGACY_BLC_EVENT_ENABLE); + if (INTEL_INFO(dev)->gen >= 4) + i915_enable_pipestat(dev_priv, 0, + PIPE_LEGACY_BLC_EVENT_ENABLE); + } + + spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags); +} + +/** + * i915_pipe_enabled - check if a pipe is enabled + * @dev: DRM device + * @pipe: pipe to check + * + * Reading certain registers when the pipe is disabled can hang the chip. + * Use this routine to make sure the PLL is running and the pipe is active + * before reading such registers if unsure. + */ +static int +i915_pipe_enabled(struct drm_device *dev, int pipe) +{ + drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + return I915_READ(PIPECONF(pipe)) & PIPECONF_ENABLE; +} + +/* Called from drm generic code, passed a 'crtc', which + * we use as a pipe index + */ +static u32 i915_get_vblank_counter(struct drm_device *dev, int pipe) +{ + drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + unsigned long high_frame; + unsigned long low_frame; + u32 high1, high2, low; + + if (!i915_pipe_enabled(dev, pipe)) { + DRM_DEBUG_DRIVER("trying to get vblank count for disabled " + "pipe %c\n", pipe_name(pipe)); + return 0; + } + + high_frame = PIPEFRAME(pipe); + low_frame = PIPEFRAMEPIXEL(pipe); + + /* + * High & low register fields aren't synchronized, so make sure + * we get a low value that's stable across two reads of the high + * register. + */ + do { + high1 = I915_READ(high_frame) & PIPE_FRAME_HIGH_MASK; + low = I915_READ(low_frame) & PIPE_FRAME_LOW_MASK; + high2 = I915_READ(high_frame) & PIPE_FRAME_HIGH_MASK; + } while (high1 != high2); + + high1 >>= PIPE_FRAME_HIGH_SHIFT; + low >>= PIPE_FRAME_LOW_SHIFT; + return (high1 << 8) | low; +} + +static u32 gm45_get_vblank_counter(struct drm_device *dev, int pipe) +{ + drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + int reg = PIPE_FRMCOUNT_GM45(pipe); + + if (!i915_pipe_enabled(dev, pipe)) { + DRM_DEBUG_DRIVER("trying to get vblank count for disabled " + "pipe %c\n", pipe_name(pipe)); + return 0; + } + + return I915_READ(reg); +} + +static int i915_get_crtc_scanoutpos(struct drm_device *dev, int pipe, + int *vpos, int *hpos) +{ + drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + u32 vbl = 0, position = 0; + int vbl_start, vbl_end, htotal, vtotal; + bool in_vbl = true; + int ret = 0; + + if (!i915_pipe_enabled(dev, pipe)) { + DRM_DEBUG_DRIVER("trying to get scanoutpos for disabled " + "pipe %c\n", pipe_name(pipe)); + return 0; + } + + /* Get vtotal. */ + vtotal = 1 + ((I915_READ(VTOTAL(pipe)) >> 16) & 0x1fff); + + if (INTEL_INFO(dev)->gen >= 4) { + /* No obvious pixelcount register. Only query vertical + * scanout position from Display scan line register. + */ + position = I915_READ(PIPEDSL(pipe)); + + /* Decode into vertical scanout position. Don't have + * horizontal scanout position. + */ + *vpos = position & 0x1fff; + *hpos = 0; + } else { + /* Have access to pixelcount since start of frame. + * We can split this into vertical and horizontal + * scanout position. + */ + position = (I915_READ(PIPEFRAMEPIXEL(pipe)) & PIPE_PIXEL_MASK) >> PIPE_PIXEL_SHIFT; + + htotal = 1 + ((I915_READ(HTOTAL(pipe)) >> 16) & 0x1fff); + *vpos = position / htotal; + *hpos = position - (*vpos * htotal); + } + + /* Query vblank area. */ + vbl = I915_READ(VBLANK(pipe)); + + /* Test position against vblank region. */ + vbl_start = vbl & 0x1fff; + vbl_end = (vbl >> 16) & 0x1fff; + + if ((*vpos < vbl_start) || (*vpos > vbl_end)) + in_vbl = false; + + /* Inside "upper part" of vblank area? Apply corrective offset: */ + if (in_vbl && (*vpos >= vbl_start)) + *vpos = *vpos - vtotal; + + /* Readouts valid? */ + if (vbl > 0) + ret |= DRM_SCANOUTPOS_VALID | DRM_SCANOUTPOS_ACCURATE; + + /* In vblank? */ + if (in_vbl) + ret |= DRM_SCANOUTPOS_INVBL; + + return ret; +} + +static int i915_get_vblank_timestamp(struct drm_device *dev, int pipe, + int *max_error, + struct timeval *vblank_time, + unsigned flags) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_crtc *crtc; + + if (pipe < 0 || pipe >= dev_priv->num_pipe) { + DRM_ERROR("Invalid crtc %d\n", pipe); + return -EINVAL; + } + + /* Get drm_crtc to timestamp: */ + crtc = intel_get_crtc_for_pipe(dev, pipe); + if (crtc == NULL) { + DRM_ERROR("Invalid crtc %d\n", pipe); + return -EINVAL; + } + + if (!crtc->enabled) { + DRM_DEBUG_KMS("crtc %d is disabled\n", pipe); + return -EBUSY; + } + + /* Helper routine in DRM core does all the work: */ + return drm_calc_vbltimestamp_from_scanoutpos(dev, pipe, max_error, + vblank_time, flags, + crtc); +} + +/* + * Handle hotplug events outside the interrupt handler proper. + */ +static void i915_hotplug_work_func(struct work_struct *work) +{ + drm_i915_private_t *dev_priv = container_of(work, drm_i915_private_t, + hotplug_work); + struct drm_device *dev = dev_priv->dev; + struct drm_mode_config *mode_config = &dev->mode_config; + struct intel_encoder *encoder; + + mutex_lock(&mode_config->mutex); + DRM_DEBUG_KMS("running encoder hotplug functions\n"); + + list_for_each_entry(encoder, &mode_config->encoder_list, base.head) + if (encoder->hot_plug) + encoder->hot_plug(encoder); + + mutex_unlock(&mode_config->mutex); + + /* Just fire off a uevent and let userspace tell us what to do */ + drm_helper_hpd_irq_event(dev); +} + +static void i915_handle_rps_change(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + u32 busy_up, busy_down, max_avg, min_avg; + u8 new_delay = dev_priv->cur_delay; + + I915_WRITE16(MEMINTRSTS, MEMINT_EVAL_CHG); + busy_up = I915_READ(RCPREVBSYTUPAVG); + busy_down = I915_READ(RCPREVBSYTDNAVG); + max_avg = I915_READ(RCBMAXAVG); + min_avg = I915_READ(RCBMINAVG); + + /* Handle RCS change request from hw */ + if (busy_up > max_avg) { + if (dev_priv->cur_delay != dev_priv->max_delay) + new_delay = dev_priv->cur_delay - 1; + if (new_delay < dev_priv->max_delay) + new_delay = dev_priv->max_delay; + } else if (busy_down < min_avg) { + if (dev_priv->cur_delay != dev_priv->min_delay) + new_delay = dev_priv->cur_delay + 1; + if (new_delay > dev_priv->min_delay) + new_delay = dev_priv->min_delay; + } + + if (ironlake_set_drps(dev, new_delay)) + dev_priv->cur_delay = new_delay; + + return; +} + +static void notify_ring(struct drm_device *dev, + struct intel_ring_buffer *ring) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u32 seqno; + + if (ring->obj == NULL) + return; + + seqno = ring->get_seqno(ring); + trace_i915_gem_request_complete(ring, seqno); + + ring->irq_seqno = seqno; + wake_up_all(&ring->irq_queue); + + dev_priv->hangcheck_count = 0; + mod_timer(&dev_priv->hangcheck_timer, + jiffies + msecs_to_jiffies(DRM_I915_HANGCHECK_PERIOD)); +} + +static void gen6_pm_rps_work(struct work_struct *work) +{ + drm_i915_private_t *dev_priv = container_of(work, drm_i915_private_t, + rps_work); + u8 new_delay = dev_priv->cur_delay; + u32 pm_iir, pm_imr; + + spin_lock_irq(&dev_priv->rps_lock); + pm_iir = dev_priv->pm_iir; + dev_priv->pm_iir = 0; + pm_imr = I915_READ(GEN6_PMIMR); + spin_unlock_irq(&dev_priv->rps_lock); + + if (!pm_iir) + return; + + mutex_lock(&dev_priv->dev->struct_mutex); + if (pm_iir & GEN6_PM_RP_UP_THRESHOLD) { + if (dev_priv->cur_delay != dev_priv->max_delay) + new_delay = dev_priv->cur_delay + 1; + if (new_delay > dev_priv->max_delay) + new_delay = dev_priv->max_delay; + } else if (pm_iir & (GEN6_PM_RP_DOWN_THRESHOLD | GEN6_PM_RP_DOWN_TIMEOUT)) { + gen6_gt_force_wake_get(dev_priv); + if (dev_priv->cur_delay != dev_priv->min_delay) + new_delay = dev_priv->cur_delay - 1; + if (new_delay < dev_priv->min_delay) { + new_delay = dev_priv->min_delay; + I915_WRITE(GEN6_RP_INTERRUPT_LIMITS, + I915_READ(GEN6_RP_INTERRUPT_LIMITS) | + ((new_delay << 16) & 0x3f0000)); + } else { + /* Make sure we continue to get down interrupts + * until we hit the minimum frequency */ + I915_WRITE(GEN6_RP_INTERRUPT_LIMITS, + I915_READ(GEN6_RP_INTERRUPT_LIMITS) & ~0x3f0000); + } + gen6_gt_force_wake_put(dev_priv); + } + + gen6_set_rps(dev_priv->dev, new_delay); + dev_priv->cur_delay = new_delay; + + /* + * rps_lock not held here because clearing is non-destructive. There is + * an *extremely* unlikely race with gen6_rps_enable() that is prevented + * by holding struct_mutex for the duration of the write. + */ + I915_WRITE(GEN6_PMIMR, pm_imr & ~pm_iir); + mutex_unlock(&dev_priv->dev->struct_mutex); +} + +static void pch_irq_handler(struct drm_device *dev, u32 pch_iir) +{ + drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + int pipe; + + if (pch_iir & SDE_AUDIO_POWER_MASK) + DRM_DEBUG_DRIVER("PCH audio power change on port %d\n", + (pch_iir & SDE_AUDIO_POWER_MASK) >> + SDE_AUDIO_POWER_SHIFT); + + if (pch_iir & SDE_GMBUS) + DRM_DEBUG_DRIVER("PCH GMBUS interrupt\n"); + + if (pch_iir & SDE_AUDIO_HDCP_MASK) + DRM_DEBUG_DRIVER("PCH HDCP audio interrupt\n"); + + if (pch_iir & SDE_AUDIO_TRANS_MASK) + DRM_DEBUG_DRIVER("PCH transcoder audio interrupt\n"); + + if (pch_iir & SDE_POISON) + DRM_ERROR("PCH poison interrupt\n"); + + if (pch_iir & SDE_FDI_MASK) + for_each_pipe(pipe) + DRM_DEBUG_DRIVER(" pipe %c FDI IIR: 0x%08x\n", + pipe_name(pipe), + I915_READ(FDI_RX_IIR(pipe))); + + if (pch_iir & (SDE_TRANSB_CRC_DONE | SDE_TRANSA_CRC_DONE)) + DRM_DEBUG_DRIVER("PCH transcoder CRC done interrupt\n"); + + if (pch_iir & (SDE_TRANSB_CRC_ERR | SDE_TRANSA_CRC_ERR)) + DRM_DEBUG_DRIVER("PCH transcoder CRC error interrupt\n"); + + if (pch_iir & SDE_TRANSB_FIFO_UNDER) + DRM_DEBUG_DRIVER("PCH transcoder B underrun interrupt\n"); + if (pch_iir & SDE_TRANSA_FIFO_UNDER) + DRM_DEBUG_DRIVER("PCH transcoder A underrun interrupt\n"); +} + +static irqreturn_t ivybridge_irq_handler(DRM_IRQ_ARGS) +{ + struct drm_device *dev = (struct drm_device *) arg; + drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + int ret = IRQ_NONE; + u32 de_iir, gt_iir, de_ier, pch_iir, pm_iir; + struct drm_i915_master_private *master_priv; + + atomic_inc(&dev_priv->irq_received); + + /* disable master interrupt before clearing iir */ + de_ier = I915_READ(DEIER); + I915_WRITE(DEIER, de_ier & ~DE_MASTER_IRQ_CONTROL); + POSTING_READ(DEIER); + + de_iir = I915_READ(DEIIR); + gt_iir = I915_READ(GTIIR); + pch_iir = I915_READ(SDEIIR); + pm_iir = I915_READ(GEN6_PMIIR); + + if (de_iir == 0 && gt_iir == 0 && pch_iir == 0 && pm_iir == 0) + goto done; + + ret = IRQ_HANDLED; + + if (dev->primary->master) { + master_priv = dev->primary->master->driver_priv; + if (master_priv->sarea_priv) + master_priv->sarea_priv->last_dispatch = + READ_BREADCRUMB(dev_priv); + } + + if (gt_iir & (GT_USER_INTERRUPT | GT_PIPE_NOTIFY)) + notify_ring(dev, &dev_priv->ring[RCS]); + if (gt_iir & GT_GEN6_BSD_USER_INTERRUPT) + notify_ring(dev, &dev_priv->ring[VCS]); + if (gt_iir & GT_BLT_USER_INTERRUPT) + notify_ring(dev, &dev_priv->ring[BCS]); + + if (de_iir & DE_GSE_IVB) + intel_opregion_gse_intr(dev); + + if (de_iir & DE_PLANEA_FLIP_DONE_IVB) { + intel_prepare_page_flip(dev, 0); + intel_finish_page_flip_plane(dev, 0); + } + + if (de_iir & DE_PLANEB_FLIP_DONE_IVB) { + intel_prepare_page_flip(dev, 1); + intel_finish_page_flip_plane(dev, 1); + } + + if (de_iir & DE_PIPEA_VBLANK_IVB) + drm_handle_vblank(dev, 0); + + if (de_iir & DE_PIPEB_VBLANK_IVB) + drm_handle_vblank(dev, 1); + + /* check event from PCH */ + if (de_iir & DE_PCH_EVENT_IVB) { + if (pch_iir & SDE_HOTPLUG_MASK_CPT) + queue_work(dev_priv->wq, &dev_priv->hotplug_work); + pch_irq_handler(dev, pch_iir); + } + + if (pm_iir & GEN6_PM_DEFERRED_EVENTS) { + unsigned long flags; + spin_lock_irqsave(&dev_priv->rps_lock, flags); + WARN(dev_priv->pm_iir & pm_iir, "Missed a PM interrupt\n"); + I915_WRITE(GEN6_PMIMR, pm_iir); + dev_priv->pm_iir |= pm_iir; + spin_unlock_irqrestore(&dev_priv->rps_lock, flags); + queue_work(dev_priv->wq, &dev_priv->rps_work); + } + + /* should clear PCH hotplug event before clear CPU irq */ + I915_WRITE(SDEIIR, pch_iir); + I915_WRITE(GTIIR, gt_iir); + I915_WRITE(DEIIR, de_iir); + I915_WRITE(GEN6_PMIIR, pm_iir); + +done: + I915_WRITE(DEIER, de_ier); + POSTING_READ(DEIER); + + return ret; +} + +static irqreturn_t ironlake_irq_handler(DRM_IRQ_ARGS) +{ + struct drm_device *dev = (struct drm_device *) arg; + drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + int ret = IRQ_NONE; + u32 de_iir, gt_iir, de_ier, pch_iir, pm_iir; + u32 hotplug_mask; + struct drm_i915_master_private *master_priv; + u32 bsd_usr_interrupt = GT_BSD_USER_INTERRUPT; + + atomic_inc(&dev_priv->irq_received); + + if (IS_GEN6(dev)) + bsd_usr_interrupt = GT_GEN6_BSD_USER_INTERRUPT; + + /* disable master interrupt before clearing iir */ + de_ier = I915_READ(DEIER); + I915_WRITE(DEIER, de_ier & ~DE_MASTER_IRQ_CONTROL); + POSTING_READ(DEIER); + + de_iir = I915_READ(DEIIR); + gt_iir = I915_READ(GTIIR); + pch_iir = I915_READ(SDEIIR); + pm_iir = I915_READ(GEN6_PMIIR); + + if (de_iir == 0 && gt_iir == 0 && pch_iir == 0 && + (!IS_GEN6(dev) || pm_iir == 0)) + goto done; + + if (HAS_PCH_CPT(dev)) + hotplug_mask = SDE_HOTPLUG_MASK_CPT; + else + hotplug_mask = SDE_HOTPLUG_MASK; + + ret = IRQ_HANDLED; + + if (dev->primary->master) { + master_priv = dev->primary->master->driver_priv; + if (master_priv->sarea_priv) + master_priv->sarea_priv->last_dispatch = + READ_BREADCRUMB(dev_priv); + } + + if (gt_iir & (GT_USER_INTERRUPT | GT_PIPE_NOTIFY)) + notify_ring(dev, &dev_priv->ring[RCS]); + if (gt_iir & bsd_usr_interrupt) + notify_ring(dev, &dev_priv->ring[VCS]); + if (gt_iir & GT_BLT_USER_INTERRUPT) + notify_ring(dev, &dev_priv->ring[BCS]); + + if (de_iir & DE_GSE) + intel_opregion_gse_intr(dev); + + if (de_iir & DE_PLANEA_FLIP_DONE) { + intel_prepare_page_flip(dev, 0); + intel_finish_page_flip_plane(dev, 0); + } + + if (de_iir & DE_PLANEB_FLIP_DONE) { + intel_prepare_page_flip(dev, 1); + intel_finish_page_flip_plane(dev, 1); + } + + if (de_iir & DE_PIPEA_VBLANK) + drm_handle_vblank(dev, 0); + + if (de_iir & DE_PIPEB_VBLANK) + drm_handle_vblank(dev, 1); + + /* check event from PCH */ + if (de_iir & DE_PCH_EVENT) { + if (pch_iir & hotplug_mask) + queue_work(dev_priv->wq, &dev_priv->hotplug_work); + pch_irq_handler(dev, pch_iir); + } + + if (de_iir & DE_PCU_EVENT) { + I915_WRITE16(MEMINTRSTS, I915_READ(MEMINTRSTS)); + i915_handle_rps_change(dev); + } + + if (IS_GEN6(dev) && pm_iir & GEN6_PM_DEFERRED_EVENTS) { + /* + * IIR bits should never already be set because IMR should + * prevent an interrupt from being shown in IIR. The warning + * displays a case where we've unsafely cleared + * dev_priv->pm_iir. Although missing an interrupt of the same + * type is not a problem, it displays a problem in the logic. + * + * The mask bit in IMR is cleared by rps_work. + */ + unsigned long flags; + spin_lock_irqsave(&dev_priv->rps_lock, flags); + WARN(dev_priv->pm_iir & pm_iir, "Missed a PM interrupt\n"); + I915_WRITE(GEN6_PMIMR, pm_iir); + dev_priv->pm_iir |= pm_iir; + spin_unlock_irqrestore(&dev_priv->rps_lock, flags); + queue_work(dev_priv->wq, &dev_priv->rps_work); + } + + /* should clear PCH hotplug event before clear CPU irq */ + I915_WRITE(SDEIIR, pch_iir); + I915_WRITE(GTIIR, gt_iir); + I915_WRITE(DEIIR, de_iir); + I915_WRITE(GEN6_PMIIR, pm_iir); + +done: + I915_WRITE(DEIER, de_ier); + POSTING_READ(DEIER); + + return ret; +} + +/** + * i915_error_work_func - do process context error handling work + * @work: work struct + * + * Fire an error uevent so userspace can see that a hang or error + * was detected. + */ +static void i915_error_work_func(struct work_struct *work) +{ + drm_i915_private_t *dev_priv = container_of(work, drm_i915_private_t, + error_work); + struct drm_device *dev = dev_priv->dev; + char *error_event[] = { "ERROR=1", NULL }; + char *reset_event[] = { "RESET=1", NULL }; + char *reset_done_event[] = { "ERROR=0", NULL }; + + kobject_uevent_env(&dev->primary->kdev.kobj, KOBJ_CHANGE, error_event); + + if (atomic_read(&dev_priv->mm.wedged)) { + DRM_DEBUG_DRIVER("resetting chip\n"); + kobject_uevent_env(&dev->primary->kdev.kobj, KOBJ_CHANGE, reset_event); + if (!i915_reset(dev, GRDOM_RENDER)) { + atomic_set(&dev_priv->mm.wedged, 0); + kobject_uevent_env(&dev->primary->kdev.kobj, KOBJ_CHANGE, reset_done_event); + } + complete_all(&dev_priv->error_completion); + } +} + +#ifdef CONFIG_DEBUG_FS +static struct drm_i915_error_object * +i915_error_object_create(struct drm_i915_private *dev_priv, + struct drm_i915_gem_object *src) +{ + struct drm_i915_error_object *dst; + int page, page_count; + u32 reloc_offset; + + if (src == NULL || src->pages == NULL) + return NULL; + + page_count = src->base.size / PAGE_SIZE; + + dst = kmalloc(sizeof(*dst) + page_count * sizeof (u32 *), GFP_ATOMIC); + if (dst == NULL) + return NULL; + + reloc_offset = src->gtt_offset; + for (page = 0; page < page_count; page++) { + unsigned long flags; + void __iomem *s; + void *d; + + d = kmalloc(PAGE_SIZE, GFP_ATOMIC); + if (d == NULL) + goto unwind; + + local_irq_save(flags); + s = io_mapping_map_atomic_wc(dev_priv->mm.gtt_mapping, + reloc_offset); + memcpy_fromio(d, s, PAGE_SIZE); + io_mapping_unmap_atomic(s); + local_irq_restore(flags); + + dst->pages[page] = d; + + reloc_offset += PAGE_SIZE; + } + dst->page_count = page_count; + dst->gtt_offset = src->gtt_offset; + + return dst; + +unwind: + while (page--) + kfree(dst->pages[page]); + kfree(dst); + return NULL; +} + +static void +i915_error_object_free(struct drm_i915_error_object *obj) +{ + int page; + + if (obj == NULL) + return; + + for (page = 0; page < obj->page_count; page++) + kfree(obj->pages[page]); + + kfree(obj); +} + +static void +i915_error_state_free(struct drm_device *dev, + struct drm_i915_error_state *error) +{ + int i; + + for (i = 0; i < ARRAY_SIZE(error->batchbuffer); i++) + i915_error_object_free(error->batchbuffer[i]); + + for (i = 0; i < ARRAY_SIZE(error->ringbuffer); i++) + i915_error_object_free(error->ringbuffer[i]); + + kfree(error->active_bo); + kfree(error->overlay); + kfree(error); +} + +static u32 capture_bo_list(struct drm_i915_error_buffer *err, + int count, + struct list_head *head) +{ + struct drm_i915_gem_object *obj; + int i = 0; + + list_for_each_entry(obj, head, mm_list) { + err->size = obj->base.size; + err->name = obj->base.name; + err->seqno = obj->last_rendering_seqno; + err->gtt_offset = obj->gtt_offset; + err->read_domains = obj->base.read_domains; + err->write_domain = obj->base.write_domain; + err->fence_reg = obj->fence_reg; + err->pinned = 0; + if (obj->pin_count > 0) + err->pinned = 1; + if (obj->user_pin_count > 0) + err->pinned = -1; + err->tiling = obj->tiling_mode; + err->dirty = obj->dirty; + err->purgeable = obj->madv != I915_MADV_WILLNEED; + err->ring = obj->ring ? obj->ring->id : 0; + err->cache_level = obj->cache_level; + + if (++i == count) + break; + + err++; + } + + return i; +} + +static void i915_gem_record_fences(struct drm_device *dev, + struct drm_i915_error_state *error) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + int i; + + /* Fences */ + switch (INTEL_INFO(dev)->gen) { + case 7: + case 6: + for (i = 0; i < 16; i++) + error->fence[i] = I915_READ64(FENCE_REG_SANDYBRIDGE_0 + (i * 8)); + break; + case 5: + case 4: + for (i = 0; i < 16; i++) + error->fence[i] = I915_READ64(FENCE_REG_965_0 + (i * 8)); + break; + case 3: + if (IS_I945G(dev) || IS_I945GM(dev) || IS_G33(dev)) + for (i = 0; i < 8; i++) + error->fence[i+8] = I915_READ(FENCE_REG_945_8 + (i * 4)); + case 2: + for (i = 0; i < 8; i++) + error->fence[i] = I915_READ(FENCE_REG_830_0 + (i * 4)); + break; + + } +} + +static struct drm_i915_error_object * +i915_error_first_batchbuffer(struct drm_i915_private *dev_priv, + struct intel_ring_buffer *ring) +{ + struct drm_i915_gem_object *obj; + u32 seqno; + + if (!ring->get_seqno) + return NULL; + + seqno = ring->get_seqno(ring); + list_for_each_entry(obj, &dev_priv->mm.active_list, mm_list) { + if (obj->ring != ring) + continue; + + if (i915_seqno_passed(seqno, obj->last_rendering_seqno)) + continue; + + if ((obj->base.read_domains & I915_GEM_DOMAIN_COMMAND) == 0) + continue; + + /* We need to copy these to an anonymous buffer as the simplest + * method to avoid being overwritten by userspace. + */ + return i915_error_object_create(dev_priv, obj); + } + + return NULL; +} + +/** + * i915_capture_error_state - capture an error record for later analysis + * @dev: drm device + * + * Should be called when an error is detected (either a hang or an error + * interrupt) to capture error state from the time of the error. Fills + * out a structure which becomes available in debugfs for user level tools + * to pick up. + */ +static void i915_capture_error_state(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_i915_gem_object *obj; + struct drm_i915_error_state *error; + unsigned long flags; + int i, pipe; + + spin_lock_irqsave(&dev_priv->error_lock, flags); + error = dev_priv->first_error; + spin_unlock_irqrestore(&dev_priv->error_lock, flags); + if (error) + return; + + /* Account for pipe specific data like PIPE*STAT */ + error = kmalloc(sizeof(*error), GFP_ATOMIC); + if (!error) { + DRM_DEBUG_DRIVER("out of memory, not capturing error state\n"); + return; + } + + DRM_INFO("capturing error event; look for more information in /debug/dri/%d/i915_error_state\n", + dev->primary->index); + + error->seqno = dev_priv->ring[RCS].get_seqno(&dev_priv->ring[RCS]); + error->eir = I915_READ(EIR); + error->pgtbl_er = I915_READ(PGTBL_ER); + for_each_pipe(pipe) + error->pipestat[pipe] = I915_READ(PIPESTAT(pipe)); + error->instpm = I915_READ(INSTPM); + error->error = 0; + if (INTEL_INFO(dev)->gen >= 6) { + error->error = I915_READ(ERROR_GEN6); + + error->bcs_acthd = I915_READ(BCS_ACTHD); + error->bcs_ipehr = I915_READ(BCS_IPEHR); + error->bcs_ipeir = I915_READ(BCS_IPEIR); + error->bcs_instdone = I915_READ(BCS_INSTDONE); + error->bcs_seqno = 0; + if (dev_priv->ring[BCS].get_seqno) + error->bcs_seqno = dev_priv->ring[BCS].get_seqno(&dev_priv->ring[BCS]); + + error->vcs_acthd = I915_READ(VCS_ACTHD); + error->vcs_ipehr = I915_READ(VCS_IPEHR); + error->vcs_ipeir = I915_READ(VCS_IPEIR); + error->vcs_instdone = I915_READ(VCS_INSTDONE); + error->vcs_seqno = 0; + if (dev_priv->ring[VCS].get_seqno) + error->vcs_seqno = dev_priv->ring[VCS].get_seqno(&dev_priv->ring[VCS]); + } + if (INTEL_INFO(dev)->gen >= 4) { + error->ipeir = I915_READ(IPEIR_I965); + error->ipehr = I915_READ(IPEHR_I965); + error->instdone = I915_READ(INSTDONE_I965); + error->instps = I915_READ(INSTPS); + error->instdone1 = I915_READ(INSTDONE1); + error->acthd = I915_READ(ACTHD_I965); + error->bbaddr = I915_READ64(BB_ADDR); + } else { + error->ipeir = I915_READ(IPEIR); + error->ipehr = I915_READ(IPEHR); + error->instdone = I915_READ(INSTDONE); + error->acthd = I915_READ(ACTHD); + error->bbaddr = 0; + } + i915_gem_record_fences(dev, error); + + /* Record the active batch and ring buffers */ + for (i = 0; i < I915_NUM_RINGS; i++) { + error->batchbuffer[i] = + i915_error_first_batchbuffer(dev_priv, + &dev_priv->ring[i]); + + error->ringbuffer[i] = + i915_error_object_create(dev_priv, + dev_priv->ring[i].obj); + } + + /* Record buffers on the active and pinned lists. */ + error->active_bo = NULL; + error->pinned_bo = NULL; + + i = 0; + list_for_each_entry(obj, &dev_priv->mm.active_list, mm_list) + i++; + error->active_bo_count = i; + list_for_each_entry(obj, &dev_priv->mm.pinned_list, mm_list) + i++; + error->pinned_bo_count = i - error->active_bo_count; + + error->active_bo = NULL; + error->pinned_bo = NULL; + if (i) { + error->active_bo = kmalloc(sizeof(*error->active_bo)*i, + GFP_ATOMIC); + if (error->active_bo) + error->pinned_bo = + error->active_bo + error->active_bo_count; + } + + if (error->active_bo) + error->active_bo_count = + capture_bo_list(error->active_bo, + error->active_bo_count, + &dev_priv->mm.active_list); + + if (error->pinned_bo) + error->pinned_bo_count = + capture_bo_list(error->pinned_bo, + error->pinned_bo_count, + &dev_priv->mm.pinned_list); + + do_gettimeofday(&error->time); + + error->overlay = intel_overlay_capture_error_state(dev); + error->display = intel_display_capture_error_state(dev); + + spin_lock_irqsave(&dev_priv->error_lock, flags); + if (dev_priv->first_error == NULL) { + dev_priv->first_error = error; + error = NULL; + } + spin_unlock_irqrestore(&dev_priv->error_lock, flags); + + if (error) + i915_error_state_free(dev, error); +} + +void i915_destroy_error_state(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_i915_error_state *error; + + spin_lock(&dev_priv->error_lock); + error = dev_priv->first_error; + dev_priv->first_error = NULL; + spin_unlock(&dev_priv->error_lock); + + if (error) + i915_error_state_free(dev, error); +} +#else +#define i915_capture_error_state(x) +#endif + +static void i915_report_and_clear_eir(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u32 eir = I915_READ(EIR); + int pipe; + + if (!eir) + return; + + printk(KERN_ERR "render error detected, EIR: 0x%08x\n", + eir); + + if (IS_G4X(dev)) { + if (eir & (GM45_ERROR_MEM_PRIV | GM45_ERROR_CP_PRIV)) { + u32 ipeir = I915_READ(IPEIR_I965); + + printk(KERN_ERR " IPEIR: 0x%08x\n", + I915_READ(IPEIR_I965)); + printk(KERN_ERR " IPEHR: 0x%08x\n", + I915_READ(IPEHR_I965)); + printk(KERN_ERR " INSTDONE: 0x%08x\n", + I915_READ(INSTDONE_I965)); + printk(KERN_ERR " INSTPS: 0x%08x\n", + I915_READ(INSTPS)); + printk(KERN_ERR " INSTDONE1: 0x%08x\n", + I915_READ(INSTDONE1)); + printk(KERN_ERR " ACTHD: 0x%08x\n", + I915_READ(ACTHD_I965)); + I915_WRITE(IPEIR_I965, ipeir); + POSTING_READ(IPEIR_I965); + } + if (eir & GM45_ERROR_PAGE_TABLE) { + u32 pgtbl_err = I915_READ(PGTBL_ER); + printk(KERN_ERR "page table error\n"); + printk(KERN_ERR " PGTBL_ER: 0x%08x\n", + pgtbl_err); + I915_WRITE(PGTBL_ER, pgtbl_err); + POSTING_READ(PGTBL_ER); + } + } + + if (!IS_GEN2(dev)) { + if (eir & I915_ERROR_PAGE_TABLE) { + u32 pgtbl_err = I915_READ(PGTBL_ER); + printk(KERN_ERR "page table error\n"); + printk(KERN_ERR " PGTBL_ER: 0x%08x\n", + pgtbl_err); + I915_WRITE(PGTBL_ER, pgtbl_err); + POSTING_READ(PGTBL_ER); + } + } + + if (eir & I915_ERROR_MEMORY_REFRESH) { + printk(KERN_ERR "memory refresh error:\n"); + for_each_pipe(pipe) + printk(KERN_ERR "pipe %c stat: 0x%08x\n", + pipe_name(pipe), I915_READ(PIPESTAT(pipe))); + /* pipestat has already been acked */ + } + if (eir & I915_ERROR_INSTRUCTION) { + printk(KERN_ERR "instruction error\n"); + printk(KERN_ERR " INSTPM: 0x%08x\n", + I915_READ(INSTPM)); + if (INTEL_INFO(dev)->gen < 4) { + u32 ipeir = I915_READ(IPEIR); + + printk(KERN_ERR " IPEIR: 0x%08x\n", + I915_READ(IPEIR)); + printk(KERN_ERR " IPEHR: 0x%08x\n", + I915_READ(IPEHR)); + printk(KERN_ERR " INSTDONE: 0x%08x\n", + I915_READ(INSTDONE)); + printk(KERN_ERR " ACTHD: 0x%08x\n", + I915_READ(ACTHD)); + I915_WRITE(IPEIR, ipeir); + POSTING_READ(IPEIR); + } else { + u32 ipeir = I915_READ(IPEIR_I965); + + printk(KERN_ERR " IPEIR: 0x%08x\n", + I915_READ(IPEIR_I965)); + printk(KERN_ERR " IPEHR: 0x%08x\n", + I915_READ(IPEHR_I965)); + printk(KERN_ERR " INSTDONE: 0x%08x\n", + I915_READ(INSTDONE_I965)); + printk(KERN_ERR " INSTPS: 0x%08x\n", + I915_READ(INSTPS)); + printk(KERN_ERR " INSTDONE1: 0x%08x\n", + I915_READ(INSTDONE1)); + printk(KERN_ERR " ACTHD: 0x%08x\n", + I915_READ(ACTHD_I965)); + I915_WRITE(IPEIR_I965, ipeir); + POSTING_READ(IPEIR_I965); + } + } + + I915_WRITE(EIR, eir); + POSTING_READ(EIR); + eir = I915_READ(EIR); + if (eir) { + /* + * some errors might have become stuck, + * mask them. + */ + DRM_ERROR("EIR stuck: 0x%08x, masking\n", eir); + I915_WRITE(EMR, I915_READ(EMR) | eir); + I915_WRITE(IIR, I915_RENDER_COMMAND_PARSER_ERROR_INTERRUPT); + } +} + +/** + * i915_handle_error - handle an error interrupt + * @dev: drm device + * + * Do some basic checking of regsiter state at error interrupt time and + * dump it to the syslog. Also call i915_capture_error_state() to make + * sure we get a record and make it available in debugfs. Fire a uevent + * so userspace knows something bad happened (should trigger collection + * of a ring dump etc.). + */ +void i915_handle_error(struct drm_device *dev, bool wedged) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + i915_capture_error_state(dev); + i915_report_and_clear_eir(dev); + + if (wedged) { + INIT_COMPLETION(dev_priv->error_completion); + atomic_set(&dev_priv->mm.wedged, 1); + + /* + * Wakeup waiting processes so they don't hang + */ + wake_up_all(&dev_priv->ring[RCS].irq_queue); + if (HAS_BSD(dev)) + wake_up_all(&dev_priv->ring[VCS].irq_queue); + if (HAS_BLT(dev)) + wake_up_all(&dev_priv->ring[BCS].irq_queue); + } + + queue_work(dev_priv->wq, &dev_priv->error_work); +} + +static void i915_pageflip_stall_check(struct drm_device *dev, int pipe) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_crtc *crtc = dev_priv->pipe_to_crtc_mapping[pipe]; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + struct drm_i915_gem_object *obj; + struct intel_unpin_work *work; + unsigned long flags; + bool stall_detected; + + /* Ignore early vblank irqs */ + if (intel_crtc == NULL) + return; + + spin_lock_irqsave(&dev->event_lock, flags); + work = intel_crtc->unpin_work; + + if (work == NULL || work->pending || !work->enable_stall_check) { + /* Either the pending flip IRQ arrived, or we're too early. Don't check */ + spin_unlock_irqrestore(&dev->event_lock, flags); + return; + } + + /* Potential stall - if we see that the flip has happened, assume a missed interrupt */ + obj = work->pending_flip_obj; + if (INTEL_INFO(dev)->gen >= 4) { + int dspsurf = DSPSURF(intel_crtc->plane); + stall_detected = I915_READ(dspsurf) == obj->gtt_offset; + } else { + int dspaddr = DSPADDR(intel_crtc->plane); + stall_detected = I915_READ(dspaddr) == (obj->gtt_offset + + crtc->y * crtc->fb->pitch + + crtc->x * crtc->fb->bits_per_pixel/8); + } + + spin_unlock_irqrestore(&dev->event_lock, flags); + + if (stall_detected) { + DRM_DEBUG_DRIVER("Pageflip stall detected\n"); + intel_prepare_page_flip(dev, intel_crtc->plane); + } +} + +static irqreturn_t i915_driver_irq_handler(DRM_IRQ_ARGS) +{ + struct drm_device *dev = (struct drm_device *) arg; + drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + struct drm_i915_master_private *master_priv; + u32 iir, new_iir; + u32 pipe_stats[I915_MAX_PIPES]; + u32 vblank_status; + int vblank = 0; + unsigned long irqflags; + int irq_received; + int ret = IRQ_NONE, pipe; + bool blc_event = false; + + atomic_inc(&dev_priv->irq_received); + + iir = I915_READ(IIR); + + if (INTEL_INFO(dev)->gen >= 4) + vblank_status = PIPE_START_VBLANK_INTERRUPT_STATUS; + else + vblank_status = PIPE_VBLANK_INTERRUPT_STATUS; + + for (;;) { + irq_received = iir != 0; + + /* Can't rely on pipestat interrupt bit in iir as it might + * have been cleared after the pipestat interrupt was received. + * It doesn't set the bit in iir again, but it still produces + * interrupts (for non-MSI). + */ + spin_lock_irqsave(&dev_priv->irq_lock, irqflags); + if (iir & I915_RENDER_COMMAND_PARSER_ERROR_INTERRUPT) + i915_handle_error(dev, false); + + for_each_pipe(pipe) { + int reg = PIPESTAT(pipe); + pipe_stats[pipe] = I915_READ(reg); + + /* + * Clear the PIPE*STAT regs before the IIR + */ + if (pipe_stats[pipe] & 0x8000ffff) { + if (pipe_stats[pipe] & PIPE_FIFO_UNDERRUN_STATUS) + DRM_DEBUG_DRIVER("pipe %c underrun\n", + pipe_name(pipe)); + I915_WRITE(reg, pipe_stats[pipe]); + irq_received = 1; + } + } + spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags); + + if (!irq_received) + break; + + ret = IRQ_HANDLED; + + /* Consume port. Then clear IIR or we'll miss events */ + if ((I915_HAS_HOTPLUG(dev)) && + (iir & I915_DISPLAY_PORT_INTERRUPT)) { + u32 hotplug_status = I915_READ(PORT_HOTPLUG_STAT); + + DRM_DEBUG_DRIVER("hotplug event received, stat 0x%08x\n", + hotplug_status); + if (hotplug_status & dev_priv->hotplug_supported_mask) + queue_work(dev_priv->wq, + &dev_priv->hotplug_work); + + I915_WRITE(PORT_HOTPLUG_STAT, hotplug_status); + I915_READ(PORT_HOTPLUG_STAT); + } + + I915_WRITE(IIR, iir); + new_iir = I915_READ(IIR); /* Flush posted writes */ + + if (dev->primary->master) { + master_priv = dev->primary->master->driver_priv; + if (master_priv->sarea_priv) + master_priv->sarea_priv->last_dispatch = + READ_BREADCRUMB(dev_priv); + } + + if (iir & I915_USER_INTERRUPT) + notify_ring(dev, &dev_priv->ring[RCS]); + if (iir & I915_BSD_USER_INTERRUPT) + notify_ring(dev, &dev_priv->ring[VCS]); + + if (iir & I915_DISPLAY_PLANE_A_FLIP_PENDING_INTERRUPT) { + intel_prepare_page_flip(dev, 0); + if (dev_priv->flip_pending_is_done) + intel_finish_page_flip_plane(dev, 0); + } + + if (iir & I915_DISPLAY_PLANE_B_FLIP_PENDING_INTERRUPT) { + intel_prepare_page_flip(dev, 1); + if (dev_priv->flip_pending_is_done) + intel_finish_page_flip_plane(dev, 1); + } + + for_each_pipe(pipe) { + if (pipe_stats[pipe] & vblank_status && + drm_handle_vblank(dev, pipe)) { + vblank++; + if (!dev_priv->flip_pending_is_done) { + i915_pageflip_stall_check(dev, pipe); + intel_finish_page_flip(dev, pipe); + } + } + + if (pipe_stats[pipe] & PIPE_LEGACY_BLC_EVENT_STATUS) + blc_event = true; + } + + + if (blc_event || (iir & I915_ASLE_INTERRUPT)) + intel_opregion_asle_intr(dev); + + /* With MSI, interrupts are only generated when iir + * transitions from zero to nonzero. If another bit got + * set while we were handling the existing iir bits, then + * we would never get another interrupt. + * + * This is fine on non-MSI as well, as if we hit this path + * we avoid exiting the interrupt handler only to generate + * another one. + * + * Note that for MSI this could cause a stray interrupt report + * if an interrupt landed in the time between writing IIR and + * the posting read. This should be rare enough to never + * trigger the 99% of 100,000 interrupts test for disabling + * stray interrupts. + */ + iir = new_iir; + } + + return ret; +} + +static int i915_emit_irq(struct drm_device * dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_i915_master_private *master_priv = dev->primary->master->driver_priv; + + i915_kernel_lost_context(dev); + + DRM_DEBUG_DRIVER("\n"); + + dev_priv->counter++; + if (dev_priv->counter > 0x7FFFFFFFUL) + dev_priv->counter = 1; + if (master_priv->sarea_priv) + master_priv->sarea_priv->last_enqueue = dev_priv->counter; + + if (BEGIN_LP_RING(4) == 0) { + OUT_RING(MI_STORE_DWORD_INDEX); + OUT_RING(I915_BREADCRUMB_INDEX << MI_STORE_DWORD_INDEX_SHIFT); + OUT_RING(dev_priv->counter); + OUT_RING(MI_USER_INTERRUPT); + ADVANCE_LP_RING(); + } + + return dev_priv->counter; +} + +static int i915_wait_irq(struct drm_device * dev, int irq_nr) +{ + drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + struct drm_i915_master_private *master_priv = dev->primary->master->driver_priv; + int ret = 0; + struct intel_ring_buffer *ring = LP_RING(dev_priv); + + DRM_DEBUG_DRIVER("irq_nr=%d breadcrumb=%d\n", irq_nr, + READ_BREADCRUMB(dev_priv)); + + if (READ_BREADCRUMB(dev_priv) >= irq_nr) { + if (master_priv->sarea_priv) + master_priv->sarea_priv->last_dispatch = READ_BREADCRUMB(dev_priv); + return 0; + } + + if (master_priv->sarea_priv) + master_priv->sarea_priv->perf_boxes |= I915_BOX_WAIT; + + if (ring->irq_get(ring)) { + DRM_WAIT_ON(ret, ring->irq_queue, 3 * DRM_HZ, + READ_BREADCRUMB(dev_priv) >= irq_nr); + ring->irq_put(ring); + } else if (wait_for(READ_BREADCRUMB(dev_priv) >= irq_nr, 3000)) + ret = -EBUSY; + + if (ret == -EBUSY) { + DRM_ERROR("EBUSY -- rec: %d emitted: %d\n", + READ_BREADCRUMB(dev_priv), (int)dev_priv->counter); + } + + return ret; +} + +/* Needs the lock as it touches the ring. + */ +int i915_irq_emit(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + drm_i915_irq_emit_t *emit = data; + int result; + + if (!dev_priv || !LP_RING(dev_priv)->virtual_start) { + DRM_ERROR("called with no initialization\n"); + return -EINVAL; + } + + RING_LOCK_TEST_WITH_RETURN(dev, file_priv); + + mutex_lock(&dev->struct_mutex); + result = i915_emit_irq(dev); + mutex_unlock(&dev->struct_mutex); + + if (DRM_COPY_TO_USER(emit->irq_seq, &result, sizeof(int))) { + DRM_ERROR("copy_to_user\n"); + return -EFAULT; + } + + return 0; +} + +/* Doesn't need the hardware lock. + */ +int i915_irq_wait(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + drm_i915_irq_wait_t *irqwait = data; + + if (!dev_priv) { + DRM_ERROR("called with no initialization\n"); + return -EINVAL; + } + + return i915_wait_irq(dev, irqwait->irq_seq); +} + +/* Called from drm generic code, passed 'crtc' which + * we use as a pipe index + */ +static int i915_enable_vblank(struct drm_device *dev, int pipe) +{ + drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + unsigned long irqflags; + + if (!i915_pipe_enabled(dev, pipe)) + return -EINVAL; + + spin_lock_irqsave(&dev_priv->irq_lock, irqflags); + if (INTEL_INFO(dev)->gen >= 4) + i915_enable_pipestat(dev_priv, pipe, + PIPE_START_VBLANK_INTERRUPT_ENABLE); + else + i915_enable_pipestat(dev_priv, pipe, + PIPE_VBLANK_INTERRUPT_ENABLE); + + /* maintain vblank delivery even in deep C-states */ + if (dev_priv->info->gen == 3) + I915_WRITE(INSTPM, INSTPM_AGPBUSY_DIS << 16); + spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags); + + return 0; +} + +static int ironlake_enable_vblank(struct drm_device *dev, int pipe) +{ + drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + unsigned long irqflags; + + if (!i915_pipe_enabled(dev, pipe)) + return -EINVAL; + + spin_lock_irqsave(&dev_priv->irq_lock, irqflags); + ironlake_enable_display_irq(dev_priv, (pipe == 0) ? + DE_PIPEA_VBLANK: DE_PIPEB_VBLANK); + spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags); + + return 0; +} + +static int ivybridge_enable_vblank(struct drm_device *dev, int pipe) +{ + drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + unsigned long irqflags; + + if (!i915_pipe_enabled(dev, pipe)) + return -EINVAL; + + spin_lock_irqsave(&dev_priv->irq_lock, irqflags); + ironlake_enable_display_irq(dev_priv, (pipe == 0) ? + DE_PIPEA_VBLANK_IVB : DE_PIPEB_VBLANK_IVB); + spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags); + + return 0; +} + +/* Called from drm generic code, passed 'crtc' which + * we use as a pipe index + */ +static void i915_disable_vblank(struct drm_device *dev, int pipe) +{ + drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + unsigned long irqflags; + + spin_lock_irqsave(&dev_priv->irq_lock, irqflags); + if (dev_priv->info->gen == 3) + I915_WRITE(INSTPM, + INSTPM_AGPBUSY_DIS << 16 | INSTPM_AGPBUSY_DIS); + + i915_disable_pipestat(dev_priv, pipe, + PIPE_VBLANK_INTERRUPT_ENABLE | + PIPE_START_VBLANK_INTERRUPT_ENABLE); + spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags); +} + +static void ironlake_disable_vblank(struct drm_device *dev, int pipe) +{ + drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + unsigned long irqflags; + + spin_lock_irqsave(&dev_priv->irq_lock, irqflags); + ironlake_disable_display_irq(dev_priv, (pipe == 0) ? + DE_PIPEA_VBLANK: DE_PIPEB_VBLANK); + spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags); +} + +static void ivybridge_disable_vblank(struct drm_device *dev, int pipe) +{ + drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + unsigned long irqflags; + + spin_lock_irqsave(&dev_priv->irq_lock, irqflags); + ironlake_disable_display_irq(dev_priv, (pipe == 0) ? + DE_PIPEA_VBLANK_IVB : DE_PIPEB_VBLANK_IVB); + spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags); +} + +/* Set the vblank monitor pipe + */ +int i915_vblank_pipe_set(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + + if (!dev_priv) { + DRM_ERROR("called with no initialization\n"); + return -EINVAL; + } + + return 0; +} + +int i915_vblank_pipe_get(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + drm_i915_vblank_pipe_t *pipe = data; + + if (!dev_priv) { + DRM_ERROR("called with no initialization\n"); + return -EINVAL; + } + + pipe->pipe = DRM_I915_VBLANK_PIPE_A | DRM_I915_VBLANK_PIPE_B; + + return 0; +} + +/** + * Schedule buffer swap at given vertical blank. + */ +int i915_vblank_swap(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + /* The delayed swap mechanism was fundamentally racy, and has been + * removed. The model was that the client requested a delayed flip/swap + * from the kernel, then waited for vblank before continuing to perform + * rendering. The problem was that the kernel might wake the client + * up before it dispatched the vblank swap (since the lock has to be + * held while touching the ringbuffer), in which case the client would + * clear and start the next frame before the swap occurred, and + * flicker would occur in addition to likely missing the vblank. + * + * In the absence of this ioctl, userland falls back to a correct path + * of waiting for a vblank, then dispatching the swap on its own. + * Context switching to userland and back is plenty fast enough for + * meeting the requirements of vblank swapping. + */ + return -EINVAL; +} + +static u32 +ring_last_seqno(struct intel_ring_buffer *ring) +{ + return list_entry(ring->request_list.prev, + struct drm_i915_gem_request, list)->seqno; +} + +static bool i915_hangcheck_ring_idle(struct intel_ring_buffer *ring, bool *err) +{ + if (list_empty(&ring->request_list) || + i915_seqno_passed(ring->get_seqno(ring), ring_last_seqno(ring))) { + /* Issue a wake-up to catch stuck h/w. */ + if (ring->waiting_seqno && waitqueue_active(&ring->irq_queue)) { + DRM_ERROR("Hangcheck timer elapsed... %s idle [waiting on %d, at %d], missed IRQ?\n", + ring->name, + ring->waiting_seqno, + ring->get_seqno(ring)); + wake_up_all(&ring->irq_queue); + *err = true; + } + return true; + } + return false; +} + +static bool kick_ring(struct intel_ring_buffer *ring) +{ + struct drm_device *dev = ring->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + u32 tmp = I915_READ_CTL(ring); + if (tmp & RING_WAIT) { + DRM_ERROR("Kicking stuck wait on %s\n", + ring->name); + I915_WRITE_CTL(ring, tmp); + return true; + } + if (IS_GEN6(dev) && + (tmp & RING_WAIT_SEMAPHORE)) { + DRM_ERROR("Kicking stuck semaphore on %s\n", + ring->name); + I915_WRITE_CTL(ring, tmp); + return true; + } + return false; +} + +/** + * This is called when the chip hasn't reported back with completed + * batchbuffers in a long time. The first time this is called we simply record + * ACTHD. If ACTHD hasn't changed by the time the hangcheck timer elapses + * again, we assume the chip is wedged and try to fix it. + */ +void i915_hangcheck_elapsed(unsigned long data) +{ + struct drm_device *dev = (struct drm_device *)data; + drm_i915_private_t *dev_priv = dev->dev_private; + uint32_t acthd, instdone, instdone1, acthd_bsd, acthd_blt; + bool err = false; + + /* If all work is done then ACTHD clearly hasn't advanced. */ + if (i915_hangcheck_ring_idle(&dev_priv->ring[RCS], &err) && + i915_hangcheck_ring_idle(&dev_priv->ring[VCS], &err) && + i915_hangcheck_ring_idle(&dev_priv->ring[BCS], &err)) { + dev_priv->hangcheck_count = 0; + if (err) + goto repeat; + return; + } + + if (INTEL_INFO(dev)->gen < 4) { + instdone = I915_READ(INSTDONE); + instdone1 = 0; + } else { + instdone = I915_READ(INSTDONE_I965); + instdone1 = I915_READ(INSTDONE1); + } + acthd = intel_ring_get_active_head(&dev_priv->ring[RCS]); + acthd_bsd = HAS_BSD(dev) ? + intel_ring_get_active_head(&dev_priv->ring[VCS]) : 0; + acthd_blt = HAS_BLT(dev) ? + intel_ring_get_active_head(&dev_priv->ring[BCS]) : 0; + + if (dev_priv->last_acthd == acthd && + dev_priv->last_acthd_bsd == acthd_bsd && + dev_priv->last_acthd_blt == acthd_blt && + dev_priv->last_instdone == instdone && + dev_priv->last_instdone1 == instdone1) { + if (dev_priv->hangcheck_count++ > 1) { + DRM_ERROR("Hangcheck timer elapsed... GPU hung\n"); + + if (!IS_GEN2(dev)) { + /* Is the chip hanging on a WAIT_FOR_EVENT? + * If so we can simply poke the RB_WAIT bit + * and break the hang. This should work on + * all but the second generation chipsets. + */ + + if (kick_ring(&dev_priv->ring[RCS])) + goto repeat; + + if (HAS_BSD(dev) && + kick_ring(&dev_priv->ring[VCS])) + goto repeat; + + if (HAS_BLT(dev) && + kick_ring(&dev_priv->ring[BCS])) + goto repeat; + } + + i915_handle_error(dev, true); + return; + } + } else { + dev_priv->hangcheck_count = 0; + + dev_priv->last_acthd = acthd; + dev_priv->last_acthd_bsd = acthd_bsd; + dev_priv->last_acthd_blt = acthd_blt; + dev_priv->last_instdone = instdone; + dev_priv->last_instdone1 = instdone1; + } + +repeat: + /* Reset timer case chip hangs without another request being added */ + mod_timer(&dev_priv->hangcheck_timer, + jiffies + msecs_to_jiffies(DRM_I915_HANGCHECK_PERIOD)); +} + +/* drm_dma.h hooks +*/ +static void ironlake_irq_preinstall(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + + atomic_set(&dev_priv->irq_received, 0); + + INIT_WORK(&dev_priv->hotplug_work, i915_hotplug_work_func); + INIT_WORK(&dev_priv->error_work, i915_error_work_func); + if (IS_GEN6(dev) || IS_IVYBRIDGE(dev)) + INIT_WORK(&dev_priv->rps_work, gen6_pm_rps_work); + + I915_WRITE(HWSTAM, 0xeffe); + if (IS_GEN6(dev) || IS_GEN7(dev)) { + /* Workaround stalls observed on Sandy Bridge GPUs by + * making the blitter command streamer generate a + * write to the Hardware Status Page for + * MI_USER_INTERRUPT. This appears to serialize the + * previous seqno write out before the interrupt + * happens. + */ + I915_WRITE(GEN6_BLITTER_HWSTAM, ~GEN6_BLITTER_USER_INTERRUPT); + I915_WRITE(GEN6_BSD_HWSTAM, ~GEN6_BSD_USER_INTERRUPT); + } + + /* XXX hotplug from PCH */ + + I915_WRITE(DEIMR, 0xffffffff); + I915_WRITE(DEIER, 0x0); + POSTING_READ(DEIER); + + /* and GT */ + I915_WRITE(GTIMR, 0xffffffff); + I915_WRITE(GTIER, 0x0); + POSTING_READ(GTIER); + + /* south display irq */ + I915_WRITE(SDEIMR, 0xffffffff); + I915_WRITE(SDEIER, 0x0); + POSTING_READ(SDEIER); +} + +static int ironlake_irq_postinstall(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + /* enable kind of interrupts always enabled */ + u32 display_mask = DE_MASTER_IRQ_CONTROL | DE_GSE | DE_PCH_EVENT | + DE_PLANEA_FLIP_DONE | DE_PLANEB_FLIP_DONE; + u32 render_irqs; + u32 hotplug_mask; + + DRM_INIT_WAITQUEUE(&dev_priv->ring[RCS].irq_queue); + if (HAS_BSD(dev)) + DRM_INIT_WAITQUEUE(&dev_priv->ring[VCS].irq_queue); + if (HAS_BLT(dev)) + DRM_INIT_WAITQUEUE(&dev_priv->ring[BCS].irq_queue); + + dev_priv->vblank_pipe = DRM_I915_VBLANK_PIPE_A | DRM_I915_VBLANK_PIPE_B; + dev_priv->irq_mask = ~display_mask; + + /* should always can generate irq */ + I915_WRITE(DEIIR, I915_READ(DEIIR)); + I915_WRITE(DEIMR, dev_priv->irq_mask); + I915_WRITE(DEIER, display_mask | DE_PIPEA_VBLANK | DE_PIPEB_VBLANK); + POSTING_READ(DEIER); + + dev_priv->gt_irq_mask = ~0; + + I915_WRITE(GTIIR, I915_READ(GTIIR)); + I915_WRITE(GTIMR, dev_priv->gt_irq_mask); + + if (IS_GEN6(dev)) + render_irqs = + GT_USER_INTERRUPT | + GT_GEN6_BSD_USER_INTERRUPT | + GT_BLT_USER_INTERRUPT; + else + render_irqs = + GT_USER_INTERRUPT | + GT_PIPE_NOTIFY | + GT_BSD_USER_INTERRUPT; + I915_WRITE(GTIER, render_irqs); + POSTING_READ(GTIER); + + if (HAS_PCH_CPT(dev)) { + hotplug_mask = (SDE_CRT_HOTPLUG_CPT | + SDE_PORTB_HOTPLUG_CPT | + SDE_PORTC_HOTPLUG_CPT | + SDE_PORTD_HOTPLUG_CPT); + } else { + hotplug_mask = (SDE_CRT_HOTPLUG | + SDE_PORTB_HOTPLUG | + SDE_PORTC_HOTPLUG | + SDE_PORTD_HOTPLUG | + SDE_AUX_MASK); + } + + dev_priv->pch_irq_mask = ~hotplug_mask; + + I915_WRITE(SDEIIR, I915_READ(SDEIIR)); + I915_WRITE(SDEIMR, dev_priv->pch_irq_mask); + I915_WRITE(SDEIER, hotplug_mask); + POSTING_READ(SDEIER); + + if (IS_IRONLAKE_M(dev)) { + /* Clear & enable PCU event interrupts */ + I915_WRITE(DEIIR, DE_PCU_EVENT); + I915_WRITE(DEIER, I915_READ(DEIER) | DE_PCU_EVENT); + ironlake_enable_display_irq(dev_priv, DE_PCU_EVENT); + } + + return 0; +} + +static int ivybridge_irq_postinstall(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + /* enable kind of interrupts always enabled */ + u32 display_mask = DE_MASTER_IRQ_CONTROL | DE_GSE_IVB | + DE_PCH_EVENT_IVB | DE_PLANEA_FLIP_DONE_IVB | + DE_PLANEB_FLIP_DONE_IVB; + u32 render_irqs; + u32 hotplug_mask; + + DRM_INIT_WAITQUEUE(&dev_priv->ring[RCS].irq_queue); + if (HAS_BSD(dev)) + DRM_INIT_WAITQUEUE(&dev_priv->ring[VCS].irq_queue); + if (HAS_BLT(dev)) + DRM_INIT_WAITQUEUE(&dev_priv->ring[BCS].irq_queue); + + dev_priv->vblank_pipe = DRM_I915_VBLANK_PIPE_A | DRM_I915_VBLANK_PIPE_B; + dev_priv->irq_mask = ~display_mask; + + /* should always can generate irq */ + I915_WRITE(DEIIR, I915_READ(DEIIR)); + I915_WRITE(DEIMR, dev_priv->irq_mask); + I915_WRITE(DEIER, display_mask | DE_PIPEA_VBLANK_IVB | + DE_PIPEB_VBLANK_IVB); + POSTING_READ(DEIER); + + dev_priv->gt_irq_mask = ~0; + + I915_WRITE(GTIIR, I915_READ(GTIIR)); + I915_WRITE(GTIMR, dev_priv->gt_irq_mask); + + render_irqs = GT_USER_INTERRUPT | GT_GEN6_BSD_USER_INTERRUPT | + GT_BLT_USER_INTERRUPT; + I915_WRITE(GTIER, render_irqs); + POSTING_READ(GTIER); + + hotplug_mask = (SDE_CRT_HOTPLUG_CPT | + SDE_PORTB_HOTPLUG_CPT | + SDE_PORTC_HOTPLUG_CPT | + SDE_PORTD_HOTPLUG_CPT); + dev_priv->pch_irq_mask = ~hotplug_mask; + + I915_WRITE(SDEIIR, I915_READ(SDEIIR)); + I915_WRITE(SDEIMR, dev_priv->pch_irq_mask); + I915_WRITE(SDEIER, hotplug_mask); + POSTING_READ(SDEIER); + + return 0; +} + +static void i915_driver_irq_preinstall(struct drm_device * dev) +{ + drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + int pipe; + + atomic_set(&dev_priv->irq_received, 0); + + INIT_WORK(&dev_priv->hotplug_work, i915_hotplug_work_func); + INIT_WORK(&dev_priv->error_work, i915_error_work_func); + + if (I915_HAS_HOTPLUG(dev)) { + I915_WRITE(PORT_HOTPLUG_EN, 0); + I915_WRITE(PORT_HOTPLUG_STAT, I915_READ(PORT_HOTPLUG_STAT)); + } + + I915_WRITE(HWSTAM, 0xeffe); + for_each_pipe(pipe) + I915_WRITE(PIPESTAT(pipe), 0); + I915_WRITE(IMR, 0xffffffff); + I915_WRITE(IER, 0x0); + POSTING_READ(IER); +} + +/* + * Must be called after intel_modeset_init or hotplug interrupts won't be + * enabled correctly. + */ +static int i915_driver_irq_postinstall(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + u32 enable_mask = I915_INTERRUPT_ENABLE_FIX | I915_INTERRUPT_ENABLE_VAR; + u32 error_mask; + + dev_priv->vblank_pipe = DRM_I915_VBLANK_PIPE_A | DRM_I915_VBLANK_PIPE_B; + + /* Unmask the interrupts that we always want on. */ + dev_priv->irq_mask = ~I915_INTERRUPT_ENABLE_FIX; + + dev_priv->pipestat[0] = 0; + dev_priv->pipestat[1] = 0; + + if (I915_HAS_HOTPLUG(dev)) { + /* Enable in IER... */ + enable_mask |= I915_DISPLAY_PORT_INTERRUPT; + /* and unmask in IMR */ + dev_priv->irq_mask &= ~I915_DISPLAY_PORT_INTERRUPT; + } + + /* + * Enable some error detection, note the instruction error mask + * bit is reserved, so we leave it masked. + */ + if (IS_G4X(dev)) { + error_mask = ~(GM45_ERROR_PAGE_TABLE | + GM45_ERROR_MEM_PRIV | + GM45_ERROR_CP_PRIV | + I915_ERROR_MEMORY_REFRESH); + } else { + error_mask = ~(I915_ERROR_PAGE_TABLE | + I915_ERROR_MEMORY_REFRESH); + } + I915_WRITE(EMR, error_mask); + + I915_WRITE(IMR, dev_priv->irq_mask); + I915_WRITE(IER, enable_mask); + POSTING_READ(IER); + + if (I915_HAS_HOTPLUG(dev)) { + u32 hotplug_en = I915_READ(PORT_HOTPLUG_EN); + + /* Note HDMI and DP share bits */ + if (dev_priv->hotplug_supported_mask & HDMIB_HOTPLUG_INT_STATUS) + hotplug_en |= HDMIB_HOTPLUG_INT_EN; + if (dev_priv->hotplug_supported_mask & HDMIC_HOTPLUG_INT_STATUS) + hotplug_en |= HDMIC_HOTPLUG_INT_EN; + if (dev_priv->hotplug_supported_mask & HDMID_HOTPLUG_INT_STATUS) + hotplug_en |= HDMID_HOTPLUG_INT_EN; + if (dev_priv->hotplug_supported_mask & SDVOC_HOTPLUG_INT_STATUS) + hotplug_en |= SDVOC_HOTPLUG_INT_EN; + if (dev_priv->hotplug_supported_mask & SDVOB_HOTPLUG_INT_STATUS) + hotplug_en |= SDVOB_HOTPLUG_INT_EN; + if (dev_priv->hotplug_supported_mask & CRT_HOTPLUG_INT_STATUS) { + hotplug_en |= CRT_HOTPLUG_INT_EN; + + /* Programming the CRT detection parameters tends + to generate a spurious hotplug event about three + seconds later. So just do it once. + */ + if (IS_G4X(dev)) + hotplug_en |= CRT_HOTPLUG_ACTIVATION_PERIOD_64; + hotplug_en |= CRT_HOTPLUG_VOLTAGE_COMPARE_50; + } + + /* Ignore TV since it's buggy */ + + I915_WRITE(PORT_HOTPLUG_EN, hotplug_en); + } + + intel_opregion_enable_asle(dev); + + return 0; +} + +static void ironlake_irq_uninstall(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + + if (!dev_priv) + return; + + dev_priv->vblank_pipe = 0; + + I915_WRITE(HWSTAM, 0xffffffff); + + I915_WRITE(DEIMR, 0xffffffff); + I915_WRITE(DEIER, 0x0); + I915_WRITE(DEIIR, I915_READ(DEIIR)); + + I915_WRITE(GTIMR, 0xffffffff); + I915_WRITE(GTIER, 0x0); + I915_WRITE(GTIIR, I915_READ(GTIIR)); +} + +static void i915_driver_irq_uninstall(struct drm_device * dev) +{ + drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + int pipe; + + if (!dev_priv) + return; + + dev_priv->vblank_pipe = 0; + + if (I915_HAS_HOTPLUG(dev)) { + I915_WRITE(PORT_HOTPLUG_EN, 0); + I915_WRITE(PORT_HOTPLUG_STAT, I915_READ(PORT_HOTPLUG_STAT)); + } + + I915_WRITE(HWSTAM, 0xffffffff); + for_each_pipe(pipe) + I915_WRITE(PIPESTAT(pipe), 0); + I915_WRITE(IMR, 0xffffffff); + I915_WRITE(IER, 0x0); + + for_each_pipe(pipe) + I915_WRITE(PIPESTAT(pipe), + I915_READ(PIPESTAT(pipe)) & 0x8000ffff); + I915_WRITE(IIR, I915_READ(IIR)); +} + +void intel_irq_init(struct drm_device *dev) +{ + dev->driver->get_vblank_counter = i915_get_vblank_counter; + dev->max_vblank_count = 0xffffff; /* only 24 bits of frame count */ + if (IS_G4X(dev) || IS_GEN5(dev) || IS_GEN6(dev) || IS_IVYBRIDGE(dev)) { + dev->max_vblank_count = 0xffffffff; /* full 32 bit counter */ + dev->driver->get_vblank_counter = gm45_get_vblank_counter; + } + + + dev->driver->get_vblank_timestamp = i915_get_vblank_timestamp; + dev->driver->get_scanout_position = i915_get_crtc_scanoutpos; + + if (IS_IVYBRIDGE(dev)) { + /* Share pre & uninstall handlers with ILK/SNB */ + dev->driver->irq_handler = ivybridge_irq_handler; + dev->driver->irq_preinstall = ironlake_irq_preinstall; + dev->driver->irq_postinstall = ivybridge_irq_postinstall; + dev->driver->irq_uninstall = ironlake_irq_uninstall; + dev->driver->enable_vblank = ivybridge_enable_vblank; + dev->driver->disable_vblank = ivybridge_disable_vblank; + } else if (HAS_PCH_SPLIT(dev)) { + dev->driver->irq_handler = ironlake_irq_handler; + dev->driver->irq_preinstall = ironlake_irq_preinstall; + dev->driver->irq_postinstall = ironlake_irq_postinstall; + dev->driver->irq_uninstall = ironlake_irq_uninstall; + dev->driver->enable_vblank = ironlake_enable_vblank; + dev->driver->disable_vblank = ironlake_disable_vblank; + } else { + dev->driver->irq_preinstall = i915_driver_irq_preinstall; + dev->driver->irq_postinstall = i915_driver_irq_postinstall; + dev->driver->irq_uninstall = i915_driver_irq_uninstall; + dev->driver->irq_handler = i915_driver_irq_handler; + dev->driver->enable_vblank = i915_enable_vblank; + dev->driver->disable_vblank = i915_disable_vblank; + } +} diff --git a/drivers/gpu/drm/i915/i915_mem.c b/drivers/gpu/drm/i915/i915_mem.c new file mode 100644 index 00000000..83b7b81b --- /dev/null +++ b/drivers/gpu/drm/i915/i915_mem.c @@ -0,0 +1,387 @@ +/* i915_mem.c -- Simple agp/fb memory manager for i915 -*- linux-c -*- + */ +/* + * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "drmP.h" +#include "drm.h" +#include "i915_drm.h" +#include "i915_drv.h" + +/* This memory manager is integrated into the global/local lru + * mechanisms used by the clients. Specifically, it operates by + * setting the 'in_use' fields of the global LRU to indicate whether + * this region is privately allocated to a client. + * + * This does require the client to actually respect that field. + * + * Currently no effort is made to allocate 'private' memory in any + * clever way - the LRU information isn't used to determine which + * block to allocate, and the ring is drained prior to allocations -- + * in other words allocation is expensive. + */ +static void mark_block(struct drm_device * dev, struct mem_block *p, int in_use) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_i915_master_private *master_priv = dev->primary->master->driver_priv; + drm_i915_sarea_t *sarea_priv = master_priv->sarea_priv; + struct drm_tex_region *list; + unsigned shift, nr; + unsigned start; + unsigned end; + unsigned i; + int age; + + shift = dev_priv->tex_lru_log_granularity; + nr = I915_NR_TEX_REGIONS; + + start = p->start >> shift; + end = (p->start + p->size - 1) >> shift; + + age = ++sarea_priv->texAge; + list = sarea_priv->texList; + + /* Mark the regions with the new flag and update their age. Move + * them to head of list to preserve LRU semantics. + */ + for (i = start; i <= end; i++) { + list[i].in_use = in_use; + list[i].age = age; + + /* remove_from_list(i) + */ + list[(unsigned)list[i].next].prev = list[i].prev; + list[(unsigned)list[i].prev].next = list[i].next; + + /* insert_at_head(list, i) + */ + list[i].prev = nr; + list[i].next = list[nr].next; + list[(unsigned)list[nr].next].prev = i; + list[nr].next = i; + } +} + +/* Very simple allocator for agp memory, working on a static range + * already mapped into each client's address space. + */ + +static struct mem_block *split_block(struct mem_block *p, int start, int size, + struct drm_file *file_priv) +{ + /* Maybe cut off the start of an existing block */ + if (start > p->start) { + struct mem_block *newblock = kmalloc(sizeof(*newblock), + GFP_KERNEL); + if (!newblock) + goto out; + newblock->start = start; + newblock->size = p->size - (start - p->start); + newblock->file_priv = NULL; + newblock->next = p->next; + newblock->prev = p; + p->next->prev = newblock; + p->next = newblock; + p->size -= newblock->size; + p = newblock; + } + + /* Maybe cut off the end of an existing block */ + if (size < p->size) { + struct mem_block *newblock = kmalloc(sizeof(*newblock), + GFP_KERNEL); + if (!newblock) + goto out; + newblock->start = start + size; + newblock->size = p->size - size; + newblock->file_priv = NULL; + newblock->next = p->next; + newblock->prev = p; + p->next->prev = newblock; + p->next = newblock; + p->size = size; + } + + out: + /* Our block is in the middle */ + p->file_priv = file_priv; + return p; +} + +static struct mem_block *alloc_block(struct mem_block *heap, int size, + int align2, struct drm_file *file_priv) +{ + struct mem_block *p; + int mask = (1 << align2) - 1; + + for (p = heap->next; p != heap; p = p->next) { + int start = (p->start + mask) & ~mask; + if (p->file_priv == NULL && start + size <= p->start + p->size) + return split_block(p, start, size, file_priv); + } + + return NULL; +} + +static struct mem_block *find_block(struct mem_block *heap, int start) +{ + struct mem_block *p; + + for (p = heap->next; p != heap; p = p->next) + if (p->start == start) + return p; + + return NULL; +} + +static void free_block(struct mem_block *p) +{ + p->file_priv = NULL; + + /* Assumes a single contiguous range. Needs a special file_priv in + * 'heap' to stop it being subsumed. + */ + if (p->next->file_priv == NULL) { + struct mem_block *q = p->next; + p->size += q->size; + p->next = q->next; + p->next->prev = p; + kfree(q); + } + + if (p->prev->file_priv == NULL) { + struct mem_block *q = p->prev; + q->size += p->size; + q->next = p->next; + q->next->prev = q; + kfree(p); + } +} + +/* Initialize. How to check for an uninitialized heap? + */ +static int init_heap(struct mem_block **heap, int start, int size) +{ + struct mem_block *blocks = kmalloc(sizeof(*blocks), GFP_KERNEL); + + if (!blocks) + return -ENOMEM; + + *heap = kmalloc(sizeof(**heap), GFP_KERNEL); + if (!*heap) { + kfree(blocks); + return -ENOMEM; + } + + blocks->start = start; + blocks->size = size; + blocks->file_priv = NULL; + blocks->next = blocks->prev = *heap; + + memset(*heap, 0, sizeof(**heap)); + (*heap)->file_priv = (struct drm_file *) - 1; + (*heap)->next = (*heap)->prev = blocks; + return 0; +} + +/* Free all blocks associated with the releasing file. + */ +void i915_mem_release(struct drm_device * dev, struct drm_file *file_priv, + struct mem_block *heap) +{ + struct mem_block *p; + + if (!heap || !heap->next) + return; + + for (p = heap->next; p != heap; p = p->next) { + if (p->file_priv == file_priv) { + p->file_priv = NULL; + mark_block(dev, p, 0); + } + } + + /* Assumes a single contiguous range. Needs a special file_priv in + * 'heap' to stop it being subsumed. + */ + for (p = heap->next; p != heap; p = p->next) { + while (p->file_priv == NULL && p->next->file_priv == NULL) { + struct mem_block *q = p->next; + p->size += q->size; + p->next = q->next; + p->next->prev = p; + kfree(q); + } + } +} + +/* Shutdown. + */ +void i915_mem_takedown(struct mem_block **heap) +{ + struct mem_block *p; + + if (!*heap) + return; + + for (p = (*heap)->next; p != *heap;) { + struct mem_block *q = p; + p = p->next; + kfree(q); + } + + kfree(*heap); + *heap = NULL; +} + +static struct mem_block **get_heap(drm_i915_private_t * dev_priv, int region) +{ + switch (region) { + case I915_MEM_REGION_AGP: + return &dev_priv->agp_heap; + default: + return NULL; + } +} + +/* IOCTL HANDLERS */ + +int i915_mem_alloc(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + drm_i915_mem_alloc_t *alloc = data; + struct mem_block *block, **heap; + + if (!dev_priv) { + DRM_ERROR("called with no initialization\n"); + return -EINVAL; + } + + heap = get_heap(dev_priv, alloc->region); + if (!heap || !*heap) + return -EFAULT; + + /* Make things easier on ourselves: all allocations at least + * 4k aligned. + */ + if (alloc->alignment < 12) + alloc->alignment = 12; + + block = alloc_block(*heap, alloc->size, alloc->alignment, file_priv); + + if (!block) + return -ENOMEM; + + mark_block(dev, block, 1); + + if (DRM_COPY_TO_USER(alloc->region_offset, &block->start, + sizeof(int))) { + DRM_ERROR("copy_to_user\n"); + return -EFAULT; + } + + return 0; +} + +int i915_mem_free(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + drm_i915_mem_free_t *memfree = data; + struct mem_block *block, **heap; + + if (!dev_priv) { + DRM_ERROR("called with no initialization\n"); + return -EINVAL; + } + + heap = get_heap(dev_priv, memfree->region); + if (!heap || !*heap) + return -EFAULT; + + block = find_block(*heap, memfree->region_offset); + if (!block) + return -EFAULT; + + if (block->file_priv != file_priv) + return -EPERM; + + mark_block(dev, block, 0); + free_block(block); + return 0; +} + +int i915_mem_init_heap(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + drm_i915_mem_init_heap_t *initheap = data; + struct mem_block **heap; + + if (!dev_priv) { + DRM_ERROR("called with no initialization\n"); + return -EINVAL; + } + + heap = get_heap(dev_priv, initheap->region); + if (!heap) + return -EFAULT; + + if (*heap) { + DRM_ERROR("heap already initialized?"); + return -EFAULT; + } + + return init_heap(heap, initheap->start, initheap->size); +} + +int i915_mem_destroy_heap( struct drm_device *dev, void *data, + struct drm_file *file_priv ) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + drm_i915_mem_destroy_heap_t *destroyheap = data; + struct mem_block **heap; + + if ( !dev_priv ) { + DRM_ERROR( "called with no initialization\n" ); + return -EINVAL; + } + + heap = get_heap( dev_priv, destroyheap->region ); + if (!heap) { + DRM_ERROR("get_heap failed"); + return -EFAULT; + } + + if (!*heap) { + DRM_ERROR("heap not initialized?"); + return -EFAULT; + } + + i915_mem_takedown( heap ); + return 0; +} diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h new file mode 100644 index 00000000..387b2b34 --- /dev/null +++ b/drivers/gpu/drm/i915/i915_reg.h @@ -0,0 +1,3484 @@ +/* Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + */ + +#ifndef _I915_REG_H_ +#define _I915_REG_H_ + +#define _PIPE(pipe, a, b) ((a) + (pipe)*((b)-(a))) + +/* + * The Bridge device's PCI config space has information about the + * fb aperture size and the amount of pre-reserved memory. + * This is all handled in the intel-gtt.ko module. i915.ko only + * cares about the vga bit for the vga rbiter. + */ +#define INTEL_GMCH_CTRL 0x52 +#define INTEL_GMCH_VGA_DISABLE (1 << 1) + +/* PCI config space */ + +#define HPLLCC 0xc0 /* 855 only */ +#define GC_CLOCK_CONTROL_MASK (0xf << 0) +#define GC_CLOCK_133_200 (0 << 0) +#define GC_CLOCK_100_200 (1 << 0) +#define GC_CLOCK_100_133 (2 << 0) +#define GC_CLOCK_166_250 (3 << 0) +#define GCFGC2 0xda +#define GCFGC 0xf0 /* 915+ only */ +#define GC_LOW_FREQUENCY_ENABLE (1 << 7) +#define GC_DISPLAY_CLOCK_190_200_MHZ (0 << 4) +#define GC_DISPLAY_CLOCK_333_MHZ (4 << 4) +#define GC_DISPLAY_CLOCK_MASK (7 << 4) +#define GM45_GC_RENDER_CLOCK_MASK (0xf << 0) +#define GM45_GC_RENDER_CLOCK_266_MHZ (8 << 0) +#define GM45_GC_RENDER_CLOCK_320_MHZ (9 << 0) +#define GM45_GC_RENDER_CLOCK_400_MHZ (0xb << 0) +#define GM45_GC_RENDER_CLOCK_533_MHZ (0xc << 0) +#define I965_GC_RENDER_CLOCK_MASK (0xf << 0) +#define I965_GC_RENDER_CLOCK_267_MHZ (2 << 0) +#define I965_GC_RENDER_CLOCK_333_MHZ (3 << 0) +#define I965_GC_RENDER_CLOCK_444_MHZ (4 << 0) +#define I965_GC_RENDER_CLOCK_533_MHZ (5 << 0) +#define I945_GC_RENDER_CLOCK_MASK (7 << 0) +#define I945_GC_RENDER_CLOCK_166_MHZ (0 << 0) +#define I945_GC_RENDER_CLOCK_200_MHZ (1 << 0) +#define I945_GC_RENDER_CLOCK_250_MHZ (3 << 0) +#define I945_GC_RENDER_CLOCK_400_MHZ (5 << 0) +#define I915_GC_RENDER_CLOCK_MASK (7 << 0) +#define I915_GC_RENDER_CLOCK_166_MHZ (0 << 0) +#define I915_GC_RENDER_CLOCK_200_MHZ (1 << 0) +#define I915_GC_RENDER_CLOCK_333_MHZ (4 << 0) +#define LBB 0xf4 + +/* Graphics reset regs */ +#define I965_GDRST 0xc0 /* PCI config register */ +#define ILK_GDSR 0x2ca4 /* MCHBAR offset */ +#define GRDOM_FULL (0<<2) +#define GRDOM_RENDER (1<<2) +#define GRDOM_MEDIA (3<<2) + +#define GEN6_GDRST 0x941c +#define GEN6_GRDOM_FULL (1 << 0) +#define GEN6_GRDOM_RENDER (1 << 1) +#define GEN6_GRDOM_MEDIA (1 << 2) +#define GEN6_GRDOM_BLT (1 << 3) + +/* VGA stuff */ + +#define VGA_ST01_MDA 0x3ba +#define VGA_ST01_CGA 0x3da + +#define VGA_MSR_WRITE 0x3c2 +#define VGA_MSR_READ 0x3cc +#define VGA_MSR_MEM_EN (1<<1) +#define VGA_MSR_CGA_MODE (1<<0) + +#define VGA_SR_INDEX 0x3c4 +#define VGA_SR_DATA 0x3c5 + +#define VGA_AR_INDEX 0x3c0 +#define VGA_AR_VID_EN (1<<5) +#define VGA_AR_DATA_WRITE 0x3c0 +#define VGA_AR_DATA_READ 0x3c1 + +#define VGA_GR_INDEX 0x3ce +#define VGA_GR_DATA 0x3cf +/* GR05 */ +#define VGA_GR_MEM_READ_MODE_SHIFT 3 +#define VGA_GR_MEM_READ_MODE_PLANE 1 +/* GR06 */ +#define VGA_GR_MEM_MODE_MASK 0xc +#define VGA_GR_MEM_MODE_SHIFT 2 +#define VGA_GR_MEM_A0000_AFFFF 0 +#define VGA_GR_MEM_A0000_BFFFF 1 +#define VGA_GR_MEM_B0000_B7FFF 2 +#define VGA_GR_MEM_B0000_BFFFF 3 + +#define VGA_DACMASK 0x3c6 +#define VGA_DACRX 0x3c7 +#define VGA_DACWX 0x3c8 +#define VGA_DACDATA 0x3c9 + +#define VGA_CR_INDEX_MDA 0x3b4 +#define VGA_CR_DATA_MDA 0x3b5 +#define VGA_CR_INDEX_CGA 0x3d4 +#define VGA_CR_DATA_CGA 0x3d5 + +/* + * Memory interface instructions used by the kernel + */ +#define MI_INSTR(opcode, flags) (((opcode) << 23) | (flags)) + +#define MI_NOOP MI_INSTR(0, 0) +#define MI_USER_INTERRUPT MI_INSTR(0x02, 0) +#define MI_WAIT_FOR_EVENT MI_INSTR(0x03, 0) +#define MI_WAIT_FOR_OVERLAY_FLIP (1<<16) +#define MI_WAIT_FOR_PLANE_B_FLIP (1<<6) +#define MI_WAIT_FOR_PLANE_A_FLIP (1<<2) +#define MI_WAIT_FOR_PLANE_A_SCANLINES (1<<1) +#define MI_FLUSH MI_INSTR(0x04, 0) +#define MI_READ_FLUSH (1 << 0) +#define MI_EXE_FLUSH (1 << 1) +#define MI_NO_WRITE_FLUSH (1 << 2) +#define MI_SCENE_COUNT (1 << 3) /* just increment scene count */ +#define MI_END_SCENE (1 << 4) /* flush binner and incr scene count */ +#define MI_INVALIDATE_ISP (1 << 5) /* invalidate indirect state pointers */ +#define MI_BATCH_BUFFER_END MI_INSTR(0x0a, 0) +#define MI_SUSPEND_FLUSH MI_INSTR(0x0b, 0) +#define MI_SUSPEND_FLUSH_EN (1<<0) +#define MI_REPORT_HEAD MI_INSTR(0x07, 0) +#define MI_OVERLAY_FLIP MI_INSTR(0x11,0) +#define MI_OVERLAY_CONTINUE (0x0<<21) +#define MI_OVERLAY_ON (0x1<<21) +#define MI_OVERLAY_OFF (0x2<<21) +#define MI_LOAD_SCAN_LINES_INCL MI_INSTR(0x12, 0) +#define MI_DISPLAY_FLIP MI_INSTR(0x14, 2) +#define MI_DISPLAY_FLIP_I915 MI_INSTR(0x14, 1) +#define MI_DISPLAY_FLIP_PLANE(n) ((n) << 20) +#define MI_SET_CONTEXT MI_INSTR(0x18, 0) +#define MI_MM_SPACE_GTT (1<<8) +#define MI_MM_SPACE_PHYSICAL (0<<8) +#define MI_SAVE_EXT_STATE_EN (1<<3) +#define MI_RESTORE_EXT_STATE_EN (1<<2) +#define MI_FORCE_RESTORE (1<<1) +#define MI_RESTORE_INHIBIT (1<<0) +#define MI_STORE_DWORD_IMM MI_INSTR(0x20, 1) +#define MI_MEM_VIRTUAL (1 << 22) /* 965+ only */ +#define MI_STORE_DWORD_INDEX MI_INSTR(0x21, 1) +#define MI_STORE_DWORD_INDEX_SHIFT 2 +/* Official intel docs are somewhat sloppy concerning MI_LOAD_REGISTER_IMM: + * - Always issue a MI_NOOP _before_ the MI_LOAD_REGISTER_IMM - otherwise hw + * simply ignores the register load under certain conditions. + * - One can actually load arbitrary many arbitrary registers: Simply issue x + * address/value pairs. Don't overdue it, though, x <= 2^4 must hold! + */ +#define MI_LOAD_REGISTER_IMM(x) MI_INSTR(0x22, 2*x-1) +#define MI_FLUSH_DW MI_INSTR(0x26, 1) /* for GEN6 */ +#define MI_INVALIDATE_TLB (1<<18) +#define MI_INVALIDATE_BSD (1<<7) +#define MI_BATCH_BUFFER MI_INSTR(0x30, 1) +#define MI_BATCH_NON_SECURE (1) +#define MI_BATCH_NON_SECURE_I965 (1<<8) +#define MI_BATCH_BUFFER_START MI_INSTR(0x31, 0) +#define MI_SEMAPHORE_MBOX MI_INSTR(0x16, 1) /* gen6+ */ +#define MI_SEMAPHORE_GLOBAL_GTT (1<<22) +#define MI_SEMAPHORE_UPDATE (1<<21) +#define MI_SEMAPHORE_COMPARE (1<<20) +#define MI_SEMAPHORE_REGISTER (1<<18) +/* + * 3D instructions used by the kernel + */ +#define GFX_INSTR(opcode, flags) ((0x3 << 29) | ((opcode) << 24) | (flags)) + +#define GFX_OP_RASTER_RULES ((0x3<<29)|(0x7<<24)) +#define GFX_OP_SCISSOR ((0x3<<29)|(0x1c<<24)|(0x10<<19)) +#define SC_UPDATE_SCISSOR (0x1<<1) +#define SC_ENABLE_MASK (0x1<<0) +#define SC_ENABLE (0x1<<0) +#define GFX_OP_LOAD_INDIRECT ((0x3<<29)|(0x1d<<24)|(0x7<<16)) +#define GFX_OP_SCISSOR_INFO ((0x3<<29)|(0x1d<<24)|(0x81<<16)|(0x1)) +#define SCI_YMIN_MASK (0xffff<<16) +#define SCI_XMIN_MASK (0xffff<<0) +#define SCI_YMAX_MASK (0xffff<<16) +#define SCI_XMAX_MASK (0xffff<<0) +#define GFX_OP_SCISSOR_ENABLE ((0x3<<29)|(0x1c<<24)|(0x10<<19)) +#define GFX_OP_SCISSOR_RECT ((0x3<<29)|(0x1d<<24)|(0x81<<16)|1) +#define GFX_OP_COLOR_FACTOR ((0x3<<29)|(0x1d<<24)|(0x1<<16)|0x0) +#define GFX_OP_STIPPLE ((0x3<<29)|(0x1d<<24)|(0x83<<16)) +#define GFX_OP_MAP_INFO ((0x3<<29)|(0x1d<<24)|0x4) +#define GFX_OP_DESTBUFFER_VARS ((0x3<<29)|(0x1d<<24)|(0x85<<16)|0x0) +#define GFX_OP_DESTBUFFER_INFO ((0x3<<29)|(0x1d<<24)|(0x8e<<16)|1) +#define GFX_OP_DRAWRECT_INFO ((0x3<<29)|(0x1d<<24)|(0x80<<16)|(0x3)) +#define GFX_OP_DRAWRECT_INFO_I965 ((0x7900<<16)|0x2) +#define SRC_COPY_BLT_CMD ((2<<29)|(0x43<<22)|4) +#define XY_SRC_COPY_BLT_CMD ((2<<29)|(0x53<<22)|6) +#define XY_MONO_SRC_COPY_IMM_BLT ((2<<29)|(0x71<<22)|5) +#define XY_SRC_COPY_BLT_WRITE_ALPHA (1<<21) +#define XY_SRC_COPY_BLT_WRITE_RGB (1<<20) +#define BLT_DEPTH_8 (0<<24) +#define BLT_DEPTH_16_565 (1<<24) +#define BLT_DEPTH_16_1555 (2<<24) +#define BLT_DEPTH_32 (3<<24) +#define BLT_ROP_GXCOPY (0xcc<<16) +#define XY_SRC_COPY_BLT_SRC_TILED (1<<15) /* 965+ only */ +#define XY_SRC_COPY_BLT_DST_TILED (1<<11) /* 965+ only */ +#define CMD_OP_DISPLAYBUFFER_INFO ((0x0<<29)|(0x14<<23)|2) +#define ASYNC_FLIP (1<<22) +#define DISPLAY_PLANE_A (0<<20) +#define DISPLAY_PLANE_B (1<<20) +#define GFX_OP_PIPE_CONTROL ((0x3<<29)|(0x3<<27)|(0x2<<24)|2) +#define PIPE_CONTROL_QW_WRITE (1<<14) +#define PIPE_CONTROL_DEPTH_STALL (1<<13) +#define PIPE_CONTROL_WC_FLUSH (1<<12) +#define PIPE_CONTROL_IS_FLUSH (1<<11) /* MBZ on Ironlake */ +#define PIPE_CONTROL_TC_FLUSH (1<<10) /* GM45+ only */ +#define PIPE_CONTROL_ISP_DIS (1<<9) +#define PIPE_CONTROL_NOTIFY (1<<8) +#define PIPE_CONTROL_GLOBAL_GTT (1<<2) /* in addr dword */ +#define PIPE_CONTROL_STALL_EN (1<<1) /* in addr word, Ironlake+ only */ + + +/* + * Reset registers + */ +#define DEBUG_RESET_I830 0x6070 +#define DEBUG_RESET_FULL (1<<7) +#define DEBUG_RESET_RENDER (1<<8) +#define DEBUG_RESET_DISPLAY (1<<9) + + +/* + * Fence registers + */ +#define FENCE_REG_830_0 0x2000 +#define FENCE_REG_945_8 0x3000 +#define I830_FENCE_START_MASK 0x07f80000 +#define I830_FENCE_TILING_Y_SHIFT 12 +#define I830_FENCE_SIZE_BITS(size) ((ffs((size) >> 19) - 1) << 8) +#define I830_FENCE_PITCH_SHIFT 4 +#define I830_FENCE_REG_VALID (1<<0) +#define I915_FENCE_MAX_PITCH_VAL 4 +#define I830_FENCE_MAX_PITCH_VAL 6 +#define I830_FENCE_MAX_SIZE_VAL (1<<8) + +#define I915_FENCE_START_MASK 0x0ff00000 +#define I915_FENCE_SIZE_BITS(size) ((ffs((size) >> 20) - 1) << 8) + +#define FENCE_REG_965_0 0x03000 +#define I965_FENCE_PITCH_SHIFT 2 +#define I965_FENCE_TILING_Y_SHIFT 1 +#define I965_FENCE_REG_VALID (1<<0) +#define I965_FENCE_MAX_PITCH_VAL 0x0400 + +#define FENCE_REG_SANDYBRIDGE_0 0x100000 +#define SANDYBRIDGE_FENCE_PITCH_SHIFT 32 + +/* + * Instruction and interrupt control regs + */ +#define PGTBL_ER 0x02024 +#define RENDER_RING_BASE 0x02000 +#define BSD_RING_BASE 0x04000 +#define GEN6_BSD_RING_BASE 0x12000 +#define BLT_RING_BASE 0x22000 +#define RING_TAIL(base) ((base)+0x30) +#define RING_HEAD(base) ((base)+0x34) +#define RING_START(base) ((base)+0x38) +#define RING_CTL(base) ((base)+0x3c) +#define RING_SYNC_0(base) ((base)+0x40) +#define RING_SYNC_1(base) ((base)+0x44) +#define RING_MAX_IDLE(base) ((base)+0x54) +#define RING_HWS_PGA(base) ((base)+0x80) +#define RING_HWS_PGA_GEN6(base) ((base)+0x2080) +#define RENDER_HWS_PGA_GEN7 (0x04080) +#define BSD_HWS_PGA_GEN7 (0x04180) +#define BLT_HWS_PGA_GEN7 (0x04280) +#define RING_ACTHD(base) ((base)+0x74) +#define RING_NOPID(base) ((base)+0x94) +#define RING_IMR(base) ((base)+0xa8) +#define TAIL_ADDR 0x001FFFF8 +#define HEAD_WRAP_COUNT 0xFFE00000 +#define HEAD_WRAP_ONE 0x00200000 +#define HEAD_ADDR 0x001FFFFC +#define RING_NR_PAGES 0x001FF000 +#define RING_REPORT_MASK 0x00000006 +#define RING_REPORT_64K 0x00000002 +#define RING_REPORT_128K 0x00000004 +#define RING_NO_REPORT 0x00000000 +#define RING_VALID_MASK 0x00000001 +#define RING_VALID 0x00000001 +#define RING_INVALID 0x00000000 +#define RING_WAIT_I8XX (1<<0) /* gen2, PRBx_HEAD */ +#define RING_WAIT (1<<11) /* gen3+, PRBx_CTL */ +#define RING_WAIT_SEMAPHORE (1<<10) /* gen6+ */ +#if 0 +#define PRB0_TAIL 0x02030 +#define PRB0_HEAD 0x02034 +#define PRB0_START 0x02038 +#define PRB0_CTL 0x0203c +#define PRB1_TAIL 0x02040 /* 915+ only */ +#define PRB1_HEAD 0x02044 /* 915+ only */ +#define PRB1_START 0x02048 /* 915+ only */ +#define PRB1_CTL 0x0204c /* 915+ only */ +#endif +#define IPEIR_I965 0x02064 +#define IPEHR_I965 0x02068 +#define INSTDONE_I965 0x0206c +#define INSTPS 0x02070 /* 965+ only */ +#define INSTDONE1 0x0207c /* 965+ only */ +#define ACTHD_I965 0x02074 +#define HWS_PGA 0x02080 +#define HWS_ADDRESS_MASK 0xfffff000 +#define HWS_START_ADDRESS_SHIFT 4 +#define PWRCTXA 0x2088 /* 965GM+ only */ +#define PWRCTX_EN (1<<0) +#define IPEIR 0x02088 +#define IPEHR 0x0208c +#define INSTDONE 0x02090 +#define NOPID 0x02094 +#define HWSTAM 0x02098 +#define VCS_INSTDONE 0x1206C +#define VCS_IPEIR 0x12064 +#define VCS_IPEHR 0x12068 +#define VCS_ACTHD 0x12074 +#define BCS_INSTDONE 0x2206C +#define BCS_IPEIR 0x22064 +#define BCS_IPEHR 0x22068 +#define BCS_ACTHD 0x22074 + +#define ERROR_GEN6 0x040a0 + +/* GM45+ chicken bits -- debug workaround bits that may be required + * for various sorts of correct behavior. The top 16 bits of each are + * the enables for writing to the corresponding low bit. + */ +#define _3D_CHICKEN 0x02084 +#define _3D_CHICKEN2 0x0208c +/* Disables pipelining of read flushes past the SF-WIZ interface. + * Required on all Ironlake steppings according to the B-Spec, but the + * particular danger of not doing so is not specified. + */ +# define _3D_CHICKEN2_WM_READ_PIPELINED (1 << 14) +#define _3D_CHICKEN3 0x02090 + +#define MI_MODE 0x0209c +# define VS_TIMER_DISPATCH (1 << 6) +# define MI_FLUSH_ENABLE (1 << 11) + +#define GFX_MODE 0x02520 +#define GFX_RUN_LIST_ENABLE (1<<15) +#define GFX_TLB_INVALIDATE_ALWAYS (1<<13) +#define GFX_SURFACE_FAULT_ENABLE (1<<12) +#define GFX_REPLAY_MODE (1<<11) +#define GFX_PSMI_GRANULARITY (1<<10) +#define GFX_PPGTT_ENABLE (1<<9) + +#define SCPD0 0x0209c /* 915+ only */ +#define IER 0x020a0 +#define IIR 0x020a4 +#define IMR 0x020a8 +#define ISR 0x020ac +#define I915_PIPE_CONTROL_NOTIFY_INTERRUPT (1<<18) +#define I915_DISPLAY_PORT_INTERRUPT (1<<17) +#define I915_RENDER_COMMAND_PARSER_ERROR_INTERRUPT (1<<15) +#define I915_GMCH_THERMAL_SENSOR_EVENT_INTERRUPT (1<<14) /* p-state */ +#define I915_HWB_OOM_INTERRUPT (1<<13) +#define I915_SYNC_STATUS_INTERRUPT (1<<12) +#define I915_DISPLAY_PLANE_A_FLIP_PENDING_INTERRUPT (1<<11) +#define I915_DISPLAY_PLANE_B_FLIP_PENDING_INTERRUPT (1<<10) +#define I915_OVERLAY_PLANE_FLIP_PENDING_INTERRUPT (1<<9) +#define I915_DISPLAY_PLANE_C_FLIP_PENDING_INTERRUPT (1<<8) +#define I915_DISPLAY_PIPE_A_VBLANK_INTERRUPT (1<<7) +#define I915_DISPLAY_PIPE_A_EVENT_INTERRUPT (1<<6) +#define I915_DISPLAY_PIPE_B_VBLANK_INTERRUPT (1<<5) +#define I915_DISPLAY_PIPE_B_EVENT_INTERRUPT (1<<4) +#define I915_DEBUG_INTERRUPT (1<<2) +#define I915_USER_INTERRUPT (1<<1) +#define I915_ASLE_INTERRUPT (1<<0) +#define I915_BSD_USER_INTERRUPT (1<<25) +#define EIR 0x020b0 +#define EMR 0x020b4 +#define ESR 0x020b8 +#define GM45_ERROR_PAGE_TABLE (1<<5) +#define GM45_ERROR_MEM_PRIV (1<<4) +#define I915_ERROR_PAGE_TABLE (1<<4) +#define GM45_ERROR_CP_PRIV (1<<3) +#define I915_ERROR_MEMORY_REFRESH (1<<1) +#define I915_ERROR_INSTRUCTION (1<<0) +#define INSTPM 0x020c0 +#define INSTPM_SELF_EN (1<<12) /* 915GM only */ +#define INSTPM_AGPBUSY_DIS (1<<11) /* gen3: when disabled, pending interrupts + will not assert AGPBUSY# and will only + be delivered when out of C3. */ +#define ACTHD 0x020c8 +#define FW_BLC 0x020d8 +#define FW_BLC2 0x020dc +#define FW_BLC_SELF 0x020e0 /* 915+ only */ +#define FW_BLC_SELF_EN_MASK (1<<31) +#define FW_BLC_SELF_FIFO_MASK (1<<16) /* 945 only */ +#define FW_BLC_SELF_EN (1<<15) /* 945 only */ +#define MM_BURST_LENGTH 0x00700000 +#define MM_FIFO_WATERMARK 0x0001F000 +#define LM_BURST_LENGTH 0x00000700 +#define LM_FIFO_WATERMARK 0x0000001F +#define MI_ARB_STATE 0x020e4 /* 915+ only */ +#define MI_ARB_MASK_SHIFT 16 /* shift for enable bits */ + +/* Make render/texture TLB fetches lower priorty than associated data + * fetches. This is not turned on by default + */ +#define MI_ARB_RENDER_TLB_LOW_PRIORITY (1 << 15) + +/* Isoch request wait on GTT enable (Display A/B/C streams). + * Make isoch requests stall on the TLB update. May cause + * display underruns (test mode only) + */ +#define MI_ARB_ISOCH_WAIT_GTT (1 << 14) + +/* Block grant count for isoch requests when block count is + * set to a finite value. + */ +#define MI_ARB_BLOCK_GRANT_MASK (3 << 12) +#define MI_ARB_BLOCK_GRANT_8 (0 << 12) /* for 3 display planes */ +#define MI_ARB_BLOCK_GRANT_4 (1 << 12) /* for 2 display planes */ +#define MI_ARB_BLOCK_GRANT_2 (2 << 12) /* for 1 display plane */ +#define MI_ARB_BLOCK_GRANT_0 (3 << 12) /* don't use */ + +/* Enable render writes to complete in C2/C3/C4 power states. + * If this isn't enabled, render writes are prevented in low + * power states. That seems bad to me. + */ +#define MI_ARB_C3_LP_WRITE_ENABLE (1 << 11) + +/* This acknowledges an async flip immediately instead + * of waiting for 2TLB fetches. + */ +#define MI_ARB_ASYNC_FLIP_ACK_IMMEDIATE (1 << 10) + +/* Enables non-sequential data reads through arbiter + */ +#define MI_ARB_DUAL_DATA_PHASE_DISABLE (1 << 9) + +/* Disable FSB snooping of cacheable write cycles from binner/render + * command stream + */ +#define MI_ARB_CACHE_SNOOP_DISABLE (1 << 8) + +/* Arbiter time slice for non-isoch streams */ +#define MI_ARB_TIME_SLICE_MASK (7 << 5) +#define MI_ARB_TIME_SLICE_1 (0 << 5) +#define MI_ARB_TIME_SLICE_2 (1 << 5) +#define MI_ARB_TIME_SLICE_4 (2 << 5) +#define MI_ARB_TIME_SLICE_6 (3 << 5) +#define MI_ARB_TIME_SLICE_8 (4 << 5) +#define MI_ARB_TIME_SLICE_10 (5 << 5) +#define MI_ARB_TIME_SLICE_14 (6 << 5) +#define MI_ARB_TIME_SLICE_16 (7 << 5) + +/* Low priority grace period page size */ +#define MI_ARB_LOW_PRIORITY_GRACE_4KB (0 << 4) /* default */ +#define MI_ARB_LOW_PRIORITY_GRACE_8KB (1 << 4) + +/* Disable display A/B trickle feed */ +#define MI_ARB_DISPLAY_TRICKLE_FEED_DISABLE (1 << 2) + +/* Set display plane priority */ +#define MI_ARB_DISPLAY_PRIORITY_A_B (0 << 0) /* display A > display B */ +#define MI_ARB_DISPLAY_PRIORITY_B_A (1 << 0) /* display B > display A */ + +#define CACHE_MODE_0 0x02120 /* 915+ only */ +#define CM0_MASK_SHIFT 16 +#define CM0_IZ_OPT_DISABLE (1<<6) +#define CM0_ZR_OPT_DISABLE (1<<5) +#define CM0_DEPTH_EVICT_DISABLE (1<<4) +#define CM0_COLOR_EVICT_DISABLE (1<<3) +#define CM0_DEPTH_WRITE_DISABLE (1<<1) +#define CM0_RC_OP_FLUSH_DISABLE (1<<0) +#define BB_ADDR 0x02140 /* 8 bytes */ +#define GFX_FLSH_CNTL 0x02170 /* 915+ only */ +#define ECOSKPD 0x021d0 +#define ECO_GATING_CX_ONLY (1<<3) +#define ECO_FLIP_DONE (1<<0) + +/* GEN6 interrupt control */ +#define GEN6_RENDER_HWSTAM 0x2098 +#define GEN6_RENDER_IMR 0x20a8 +#define GEN6_RENDER_CONTEXT_SWITCH_INTERRUPT (1 << 8) +#define GEN6_RENDER_PPGTT_PAGE_FAULT (1 << 7) +#define GEN6_RENDER_TIMEOUT_COUNTER_EXPIRED (1 << 6) +#define GEN6_RENDER_L3_PARITY_ERROR (1 << 5) +#define GEN6_RENDER_PIPE_CONTROL_NOTIFY_INTERRUPT (1 << 4) +#define GEN6_RENDER_COMMAND_PARSER_MASTER_ERROR (1 << 3) +#define GEN6_RENDER_SYNC_STATUS (1 << 2) +#define GEN6_RENDER_DEBUG_INTERRUPT (1 << 1) +#define GEN6_RENDER_USER_INTERRUPT (1 << 0) + +#define GEN6_BLITTER_HWSTAM 0x22098 +#define GEN6_BLITTER_IMR 0x220a8 +#define GEN6_BLITTER_MI_FLUSH_DW_NOTIFY_INTERRUPT (1 << 26) +#define GEN6_BLITTER_COMMAND_PARSER_MASTER_ERROR (1 << 25) +#define GEN6_BLITTER_SYNC_STATUS (1 << 24) +#define GEN6_BLITTER_USER_INTERRUPT (1 << 22) + +#define GEN6_BLITTER_ECOSKPD 0x221d0 +#define GEN6_BLITTER_LOCK_SHIFT 16 +#define GEN6_BLITTER_FBC_NOTIFY (1<<3) + +#define GEN6_BSD_SLEEP_PSMI_CONTROL 0x12050 +#define GEN6_BSD_SLEEP_PSMI_CONTROL_RC_ILDL_MESSAGE_MODIFY_MASK (1 << 16) +#define GEN6_BSD_SLEEP_PSMI_CONTROL_RC_ILDL_MESSAGE_DISABLE (1 << 0) +#define GEN6_BSD_SLEEP_PSMI_CONTROL_RC_ILDL_MESSAGE_ENABLE 0 +#define GEN6_BSD_SLEEP_PSMI_CONTROL_IDLE_INDICATOR (1 << 3) + +#define GEN6_BSD_HWSTAM 0x12098 +#define GEN6_BSD_IMR 0x120a8 +#define GEN6_BSD_USER_INTERRUPT (1 << 12) + +#define GEN6_BSD_RNCID 0x12198 + +#define GEN7_FF_THREAD_MODE 0x20a0 +#define GEN7_FF_SCHED_MASK 0x0077070 +#define GEN7_FF_TS_SCHED_HS1 (0x5<<16) +#define GEN7_FF_TS_SCHED_HS0 (0x3<<16) +#define GEN7_FF_TS_SCHED_LOAD_BALANCE (0x1<<16) +#define GEN7_FF_TS_SCHED_HW (0x0<<16) /* Default */ +#define GEN7_FF_VS_SCHED_HS1 (0x5<<12) +#define GEN7_FF_VS_SCHED_HS0 (0x3<<12) +#define GEN7_FF_VS_SCHED_LOAD_BALANCE (0x1<<12) /* Default */ +#define GEN7_FF_VS_SCHED_HW (0x0<<12) +#define GEN7_FF_DS_SCHED_HS1 (0x5<<4) +#define GEN7_FF_DS_SCHED_HS0 (0x3<<4) +#define GEN7_FF_DS_SCHED_LOAD_BALANCE (0x1<<4) /* Default */ +#define GEN7_FF_DS_SCHED_HW (0x0<<4) + +/* + * Framebuffer compression (915+ only) + */ + +#define FBC_CFB_BASE 0x03200 /* 4k page aligned */ +#define FBC_LL_BASE 0x03204 /* 4k page aligned */ +#define FBC_CONTROL 0x03208 +#define FBC_CTL_EN (1<<31) +#define FBC_CTL_PERIODIC (1<<30) +#define FBC_CTL_INTERVAL_SHIFT (16) +#define FBC_CTL_UNCOMPRESSIBLE (1<<14) +#define FBC_CTL_C3_IDLE (1<<13) +#define FBC_CTL_STRIDE_SHIFT (5) +#define FBC_CTL_FENCENO (1<<0) +#define FBC_COMMAND 0x0320c +#define FBC_CMD_COMPRESS (1<<0) +#define FBC_STATUS 0x03210 +#define FBC_STAT_COMPRESSING (1<<31) +#define FBC_STAT_COMPRESSED (1<<30) +#define FBC_STAT_MODIFIED (1<<29) +#define FBC_STAT_CURRENT_LINE (1<<0) +#define FBC_CONTROL2 0x03214 +#define FBC_CTL_FENCE_DBL (0<<4) +#define FBC_CTL_IDLE_IMM (0<<2) +#define FBC_CTL_IDLE_FULL (1<<2) +#define FBC_CTL_IDLE_LINE (2<<2) +#define FBC_CTL_IDLE_DEBUG (3<<2) +#define FBC_CTL_CPU_FENCE (1<<1) +#define FBC_CTL_PLANEA (0<<0) +#define FBC_CTL_PLANEB (1<<0) +#define FBC_FENCE_OFF 0x0321b +#define FBC_TAG 0x03300 + +#define FBC_LL_SIZE (1536) + +/* Framebuffer compression for GM45+ */ +#define DPFC_CB_BASE 0x3200 +#define DPFC_CONTROL 0x3208 +#define DPFC_CTL_EN (1<<31) +#define DPFC_CTL_PLANEA (0<<30) +#define DPFC_CTL_PLANEB (1<<30) +#define DPFC_CTL_FENCE_EN (1<<29) +#define DPFC_SR_EN (1<<10) +#define DPFC_CTL_LIMIT_1X (0<<6) +#define DPFC_CTL_LIMIT_2X (1<<6) +#define DPFC_CTL_LIMIT_4X (2<<6) +#define DPFC_RECOMP_CTL 0x320c +#define DPFC_RECOMP_STALL_EN (1<<27) +#define DPFC_RECOMP_STALL_WM_SHIFT (16) +#define DPFC_RECOMP_STALL_WM_MASK (0x07ff0000) +#define DPFC_RECOMP_TIMER_COUNT_SHIFT (0) +#define DPFC_RECOMP_TIMER_COUNT_MASK (0x0000003f) +#define DPFC_STATUS 0x3210 +#define DPFC_INVAL_SEG_SHIFT (16) +#define DPFC_INVAL_SEG_MASK (0x07ff0000) +#define DPFC_COMP_SEG_SHIFT (0) +#define DPFC_COMP_SEG_MASK (0x000003ff) +#define DPFC_STATUS2 0x3214 +#define DPFC_FENCE_YOFF 0x3218 +#define DPFC_CHICKEN 0x3224 +#define DPFC_HT_MODIFY (1<<31) + +/* Framebuffer compression for Ironlake */ +#define ILK_DPFC_CB_BASE 0x43200 +#define ILK_DPFC_CONTROL 0x43208 +/* The bit 28-8 is reserved */ +#define DPFC_RESERVED (0x1FFFFF00) +#define ILK_DPFC_RECOMP_CTL 0x4320c +#define ILK_DPFC_STATUS 0x43210 +#define ILK_DPFC_FENCE_YOFF 0x43218 +#define ILK_DPFC_CHICKEN 0x43224 +#define ILK_FBC_RT_BASE 0x2128 +#define ILK_FBC_RT_VALID (1<<0) + +#define ILK_DISPLAY_CHICKEN1 0x42000 +#define ILK_FBCQ_DIS (1<<22) +#define ILK_PABSTRETCH_DIS (1<<21) + + +/* + * Framebuffer compression for Sandybridge + * + * The following two registers are of type GTTMMADR + */ +#define SNB_DPFC_CTL_SA 0x100100 +#define SNB_CPU_FENCE_ENABLE (1<<29) +#define DPFC_CPU_FENCE_OFFSET 0x100104 + + +/* + * GPIO regs + */ +#define GPIOA 0x5010 +#define GPIOB 0x5014 +#define GPIOC 0x5018 +#define GPIOD 0x501c +#define GPIOE 0x5020 +#define GPIOF 0x5024 +#define GPIOG 0x5028 +#define GPIOH 0x502c +# define GPIO_CLOCK_DIR_MASK (1 << 0) +# define GPIO_CLOCK_DIR_IN (0 << 1) +# define GPIO_CLOCK_DIR_OUT (1 << 1) +# define GPIO_CLOCK_VAL_MASK (1 << 2) +# define GPIO_CLOCK_VAL_OUT (1 << 3) +# define GPIO_CLOCK_VAL_IN (1 << 4) +# define GPIO_CLOCK_PULLUP_DISABLE (1 << 5) +# define GPIO_DATA_DIR_MASK (1 << 8) +# define GPIO_DATA_DIR_IN (0 << 9) +# define GPIO_DATA_DIR_OUT (1 << 9) +# define GPIO_DATA_VAL_MASK (1 << 10) +# define GPIO_DATA_VAL_OUT (1 << 11) +# define GPIO_DATA_VAL_IN (1 << 12) +# define GPIO_DATA_PULLUP_DISABLE (1 << 13) + +#define GMBUS0 0x5100 /* clock/port select */ +#define GMBUS_RATE_100KHZ (0<<8) +#define GMBUS_RATE_50KHZ (1<<8) +#define GMBUS_RATE_400KHZ (2<<8) /* reserved on Pineview */ +#define GMBUS_RATE_1MHZ (3<<8) /* reserved on Pineview */ +#define GMBUS_HOLD_EXT (1<<7) /* 300ns hold time, rsvd on Pineview */ +#define GMBUS_PORT_DISABLED 0 +#define GMBUS_PORT_SSC 1 +#define GMBUS_PORT_VGADDC 2 +#define GMBUS_PORT_PANEL 3 +#define GMBUS_PORT_DPC 4 /* HDMIC */ +#define GMBUS_PORT_DPB 5 /* SDVO, HDMIB */ + /* 6 reserved */ +#define GMBUS_PORT_DPD 7 /* HDMID */ +#define GMBUS_NUM_PORTS 8 +#define GMBUS1 0x5104 /* command/status */ +#define GMBUS_SW_CLR_INT (1<<31) +#define GMBUS_SW_RDY (1<<30) +#define GMBUS_ENT (1<<29) /* enable timeout */ +#define GMBUS_CYCLE_NONE (0<<25) +#define GMBUS_CYCLE_WAIT (1<<25) +#define GMBUS_CYCLE_INDEX (2<<25) +#define GMBUS_CYCLE_STOP (4<<25) +#define GMBUS_BYTE_COUNT_SHIFT 16 +#define GMBUS_SLAVE_INDEX_SHIFT 8 +#define GMBUS_SLAVE_ADDR_SHIFT 1 +#define GMBUS_SLAVE_READ (1<<0) +#define GMBUS_SLAVE_WRITE (0<<0) +#define GMBUS2 0x5108 /* status */ +#define GMBUS_INUSE (1<<15) +#define GMBUS_HW_WAIT_PHASE (1<<14) +#define GMBUS_STALL_TIMEOUT (1<<13) +#define GMBUS_INT (1<<12) +#define GMBUS_HW_RDY (1<<11) +#define GMBUS_SATOER (1<<10) +#define GMBUS_ACTIVE (1<<9) +#define GMBUS3 0x510c /* data buffer bytes 3-0 */ +#define GMBUS4 0x5110 /* interrupt mask (Pineview+) */ +#define GMBUS_SLAVE_TIMEOUT_EN (1<<4) +#define GMBUS_NAK_EN (1<<3) +#define GMBUS_IDLE_EN (1<<2) +#define GMBUS_HW_WAIT_EN (1<<1) +#define GMBUS_HW_RDY_EN (1<<0) +#define GMBUS5 0x5120 /* byte index */ +#define GMBUS_2BYTE_INDEX_EN (1<<31) + +/* + * Clock control & power management + */ + +#define VGA0 0x6000 +#define VGA1 0x6004 +#define VGA_PD 0x6010 +#define VGA0_PD_P2_DIV_4 (1 << 7) +#define VGA0_PD_P1_DIV_2 (1 << 5) +#define VGA0_PD_P1_SHIFT 0 +#define VGA0_PD_P1_MASK (0x1f << 0) +#define VGA1_PD_P2_DIV_4 (1 << 15) +#define VGA1_PD_P1_DIV_2 (1 << 13) +#define VGA1_PD_P1_SHIFT 8 +#define VGA1_PD_P1_MASK (0x1f << 8) +#define _DPLL_A 0x06014 +#define _DPLL_B 0x06018 +#define DPLL(pipe) _PIPE(pipe, _DPLL_A, _DPLL_B) +#define DPLL_VCO_ENABLE (1 << 31) +#define DPLL_DVO_HIGH_SPEED (1 << 30) +#define DPLL_SYNCLOCK_ENABLE (1 << 29) +#define DPLL_VGA_MODE_DIS (1 << 28) +#define DPLLB_MODE_DAC_SERIAL (1 << 26) /* i915 */ +#define DPLLB_MODE_LVDS (2 << 26) /* i915 */ +#define DPLL_MODE_MASK (3 << 26) +#define DPLL_DAC_SERIAL_P2_CLOCK_DIV_10 (0 << 24) /* i915 */ +#define DPLL_DAC_SERIAL_P2_CLOCK_DIV_5 (1 << 24) /* i915 */ +#define DPLLB_LVDS_P2_CLOCK_DIV_14 (0 << 24) /* i915 */ +#define DPLLB_LVDS_P2_CLOCK_DIV_7 (1 << 24) /* i915 */ +#define DPLL_P2_CLOCK_DIV_MASK 0x03000000 /* i915 */ +#define DPLL_FPA01_P1_POST_DIV_MASK 0x00ff0000 /* i915 */ +#define DPLL_FPA01_P1_POST_DIV_MASK_PINEVIEW 0x00ff8000 /* Pineview */ + +#define SRX_INDEX 0x3c4 +#define SRX_DATA 0x3c5 +#define SR01 1 +#define SR01_SCREEN_OFF (1<<5) + +#define PPCR 0x61204 +#define PPCR_ON (1<<0) + +#define DVOB 0x61140 +#define DVOB_ON (1<<31) +#define DVOC 0x61160 +#define DVOC_ON (1<<31) +#define LVDS 0x61180 +#define LVDS_ON (1<<31) + +/* Scratch pad debug 0 reg: + */ +#define DPLL_FPA01_P1_POST_DIV_MASK_I830 0x001f0000 +/* + * The i830 generation, in LVDS mode, defines P1 as the bit number set within + * this field (only one bit may be set). + */ +#define DPLL_FPA01_P1_POST_DIV_MASK_I830_LVDS 0x003f0000 +#define DPLL_FPA01_P1_POST_DIV_SHIFT 16 +#define DPLL_FPA01_P1_POST_DIV_SHIFT_PINEVIEW 15 +/* i830, required in DVO non-gang */ +#define PLL_P2_DIVIDE_BY_4 (1 << 23) +#define PLL_P1_DIVIDE_BY_TWO (1 << 21) /* i830 */ +#define PLL_REF_INPUT_DREFCLK (0 << 13) +#define PLL_REF_INPUT_TVCLKINA (1 << 13) /* i830 */ +#define PLL_REF_INPUT_TVCLKINBC (2 << 13) /* SDVO TVCLKIN */ +#define PLLB_REF_INPUT_SPREADSPECTRUMIN (3 << 13) +#define PLL_REF_INPUT_MASK (3 << 13) +#define PLL_LOAD_PULSE_PHASE_SHIFT 9 +/* Ironlake */ +# define PLL_REF_SDVO_HDMI_MULTIPLIER_SHIFT 9 +# define PLL_REF_SDVO_HDMI_MULTIPLIER_MASK (7 << 9) +# define PLL_REF_SDVO_HDMI_MULTIPLIER(x) (((x)-1) << 9) +# define DPLL_FPA1_P1_POST_DIV_SHIFT 0 +# define DPLL_FPA1_P1_POST_DIV_MASK 0xff + +/* + * Parallel to Serial Load Pulse phase selection. + * Selects the phase for the 10X DPLL clock for the PCIe + * digital display port. The range is 4 to 13; 10 or more + * is just a flip delay. The default is 6 + */ +#define PLL_LOAD_PULSE_PHASE_MASK (0xf << PLL_LOAD_PULSE_PHASE_SHIFT) +#define DISPLAY_RATE_SELECT_FPA1 (1 << 8) +/* + * SDVO multiplier for 945G/GM. Not used on 965. + */ +#define SDVO_MULTIPLIER_MASK 0x000000ff +#define SDVO_MULTIPLIER_SHIFT_HIRES 4 +#define SDVO_MULTIPLIER_SHIFT_VGA 0 +#define _DPLL_A_MD 0x0601c /* 965+ only */ +/* + * UDI pixel divider, controlling how many pixels are stuffed into a packet. + * + * Value is pixels minus 1. Must be set to 1 pixel for SDVO. + */ +#define DPLL_MD_UDI_DIVIDER_MASK 0x3f000000 +#define DPLL_MD_UDI_DIVIDER_SHIFT 24 +/* UDI pixel divider for VGA, same as DPLL_MD_UDI_DIVIDER_MASK. */ +#define DPLL_MD_VGA_UDI_DIVIDER_MASK 0x003f0000 +#define DPLL_MD_VGA_UDI_DIVIDER_SHIFT 16 +/* + * SDVO/UDI pixel multiplier. + * + * SDVO requires that the bus clock rate be between 1 and 2 Ghz, and the bus + * clock rate is 10 times the DPLL clock. At low resolution/refresh rate + * modes, the bus rate would be below the limits, so SDVO allows for stuffing + * dummy bytes in the datastream at an increased clock rate, with both sides of + * the link knowing how many bytes are fill. + * + * So, for a mode with a dotclock of 65Mhz, we would want to double the clock + * rate to 130Mhz to get a bus rate of 1.30Ghz. The DPLL clock rate would be + * set to 130Mhz, and the SDVO multiplier set to 2x in this register and + * through an SDVO command. + * + * This register field has values of multiplication factor minus 1, with + * a maximum multiplier of 5 for SDVO. + */ +#define DPLL_MD_UDI_MULTIPLIER_MASK 0x00003f00 +#define DPLL_MD_UDI_MULTIPLIER_SHIFT 8 +/* + * SDVO/UDI pixel multiplier for VGA, same as DPLL_MD_UDI_MULTIPLIER_MASK. + * This best be set to the default value (3) or the CRT won't work. No, + * I don't entirely understand what this does... + */ +#define DPLL_MD_VGA_UDI_MULTIPLIER_MASK 0x0000003f +#define DPLL_MD_VGA_UDI_MULTIPLIER_SHIFT 0 +#define _DPLL_B_MD 0x06020 /* 965+ only */ +#define DPLL_MD(pipe) _PIPE(pipe, _DPLL_A_MD, _DPLL_B_MD) +#define _FPA0 0x06040 +#define _FPA1 0x06044 +#define _FPB0 0x06048 +#define _FPB1 0x0604c +#define FP0(pipe) _PIPE(pipe, _FPA0, _FPB0) +#define FP1(pipe) _PIPE(pipe, _FPA1, _FPB1) +#define FP_N_DIV_MASK 0x003f0000 +#define FP_N_PINEVIEW_DIV_MASK 0x00ff0000 +#define FP_N_DIV_SHIFT 16 +#define FP_M1_DIV_MASK 0x00003f00 +#define FP_M1_DIV_SHIFT 8 +#define FP_M2_DIV_MASK 0x0000003f +#define FP_M2_PINEVIEW_DIV_MASK 0x000000ff +#define FP_M2_DIV_SHIFT 0 +#define DPLL_TEST 0x606c +#define DPLLB_TEST_SDVO_DIV_1 (0 << 22) +#define DPLLB_TEST_SDVO_DIV_2 (1 << 22) +#define DPLLB_TEST_SDVO_DIV_4 (2 << 22) +#define DPLLB_TEST_SDVO_DIV_MASK (3 << 22) +#define DPLLB_TEST_N_BYPASS (1 << 19) +#define DPLLB_TEST_M_BYPASS (1 << 18) +#define DPLLB_INPUT_BUFFER_ENABLE (1 << 16) +#define DPLLA_TEST_N_BYPASS (1 << 3) +#define DPLLA_TEST_M_BYPASS (1 << 2) +#define DPLLA_INPUT_BUFFER_ENABLE (1 << 0) +#define D_STATE 0x6104 +#define DSTATE_GFX_RESET_I830 (1<<6) +#define DSTATE_PLL_D3_OFF (1<<3) +#define DSTATE_GFX_CLOCK_GATING (1<<1) +#define DSTATE_DOT_CLOCK_GATING (1<<0) +#define DSPCLK_GATE_D 0x6200 +# define DPUNIT_B_CLOCK_GATE_DISABLE (1 << 30) /* 965 */ +# define VSUNIT_CLOCK_GATE_DISABLE (1 << 29) /* 965 */ +# define VRHUNIT_CLOCK_GATE_DISABLE (1 << 28) /* 965 */ +# define VRDUNIT_CLOCK_GATE_DISABLE (1 << 27) /* 965 */ +# define AUDUNIT_CLOCK_GATE_DISABLE (1 << 26) /* 965 */ +# define DPUNIT_A_CLOCK_GATE_DISABLE (1 << 25) /* 965 */ +# define DPCUNIT_CLOCK_GATE_DISABLE (1 << 24) /* 965 */ +# define TVRUNIT_CLOCK_GATE_DISABLE (1 << 23) /* 915-945 */ +# define TVCUNIT_CLOCK_GATE_DISABLE (1 << 22) /* 915-945 */ +# define TVFUNIT_CLOCK_GATE_DISABLE (1 << 21) /* 915-945 */ +# define TVEUNIT_CLOCK_GATE_DISABLE (1 << 20) /* 915-945 */ +# define DVSUNIT_CLOCK_GATE_DISABLE (1 << 19) /* 915-945 */ +# define DSSUNIT_CLOCK_GATE_DISABLE (1 << 18) /* 915-945 */ +# define DDBUNIT_CLOCK_GATE_DISABLE (1 << 17) /* 915-945 */ +# define DPRUNIT_CLOCK_GATE_DISABLE (1 << 16) /* 915-945 */ +# define DPFUNIT_CLOCK_GATE_DISABLE (1 << 15) /* 915-945 */ +# define DPBMUNIT_CLOCK_GATE_DISABLE (1 << 14) /* 915-945 */ +# define DPLSUNIT_CLOCK_GATE_DISABLE (1 << 13) /* 915-945 */ +# define DPLUNIT_CLOCK_GATE_DISABLE (1 << 12) /* 915-945 */ +# define DPOUNIT_CLOCK_GATE_DISABLE (1 << 11) +# define DPBUNIT_CLOCK_GATE_DISABLE (1 << 10) +# define DCUNIT_CLOCK_GATE_DISABLE (1 << 9) +# define DPUNIT_CLOCK_GATE_DISABLE (1 << 8) +# define VRUNIT_CLOCK_GATE_DISABLE (1 << 7) /* 915+: reserved */ +# define OVHUNIT_CLOCK_GATE_DISABLE (1 << 6) /* 830-865 */ +# define DPIOUNIT_CLOCK_GATE_DISABLE (1 << 6) /* 915-945 */ +# define OVFUNIT_CLOCK_GATE_DISABLE (1 << 5) +# define OVBUNIT_CLOCK_GATE_DISABLE (1 << 4) +/** + * This bit must be set on the 830 to prevent hangs when turning off the + * overlay scaler. + */ +# define OVRUNIT_CLOCK_GATE_DISABLE (1 << 3) +# define OVCUNIT_CLOCK_GATE_DISABLE (1 << 2) +# define OVUUNIT_CLOCK_GATE_DISABLE (1 << 1) +# define ZVUNIT_CLOCK_GATE_DISABLE (1 << 0) /* 830 */ +# define OVLUNIT_CLOCK_GATE_DISABLE (1 << 0) /* 845,865 */ + +#define RENCLK_GATE_D1 0x6204 +# define BLITTER_CLOCK_GATE_DISABLE (1 << 13) /* 945GM only */ +# define MPEG_CLOCK_GATE_DISABLE (1 << 12) /* 945GM only */ +# define PC_FE_CLOCK_GATE_DISABLE (1 << 11) +# define PC_BE_CLOCK_GATE_DISABLE (1 << 10) +# define WINDOWER_CLOCK_GATE_DISABLE (1 << 9) +# define INTERPOLATOR_CLOCK_GATE_DISABLE (1 << 8) +# define COLOR_CALCULATOR_CLOCK_GATE_DISABLE (1 << 7) +# define MOTION_COMP_CLOCK_GATE_DISABLE (1 << 6) +# define MAG_CLOCK_GATE_DISABLE (1 << 5) +/** This bit must be unset on 855,865 */ +# define MECI_CLOCK_GATE_DISABLE (1 << 4) +# define DCMP_CLOCK_GATE_DISABLE (1 << 3) +# define MEC_CLOCK_GATE_DISABLE (1 << 2) +# define MECO_CLOCK_GATE_DISABLE (1 << 1) +/** This bit must be set on 855,865. */ +# define SV_CLOCK_GATE_DISABLE (1 << 0) +# define I915_MPEG_CLOCK_GATE_DISABLE (1 << 16) +# define I915_VLD_IP_PR_CLOCK_GATE_DISABLE (1 << 15) +# define I915_MOTION_COMP_CLOCK_GATE_DISABLE (1 << 14) +# define I915_BD_BF_CLOCK_GATE_DISABLE (1 << 13) +# define I915_SF_SE_CLOCK_GATE_DISABLE (1 << 12) +# define I915_WM_CLOCK_GATE_DISABLE (1 << 11) +# define I915_IZ_CLOCK_GATE_DISABLE (1 << 10) +# define I915_PI_CLOCK_GATE_DISABLE (1 << 9) +# define I915_DI_CLOCK_GATE_DISABLE (1 << 8) +# define I915_SH_SV_CLOCK_GATE_DISABLE (1 << 7) +# define I915_PL_DG_QC_FT_CLOCK_GATE_DISABLE (1 << 6) +# define I915_SC_CLOCK_GATE_DISABLE (1 << 5) +# define I915_FL_CLOCK_GATE_DISABLE (1 << 4) +# define I915_DM_CLOCK_GATE_DISABLE (1 << 3) +# define I915_PS_CLOCK_GATE_DISABLE (1 << 2) +# define I915_CC_CLOCK_GATE_DISABLE (1 << 1) +# define I915_BY_CLOCK_GATE_DISABLE (1 << 0) + +# define I965_RCZ_CLOCK_GATE_DISABLE (1 << 30) +/** This bit must always be set on 965G/965GM */ +# define I965_RCC_CLOCK_GATE_DISABLE (1 << 29) +# define I965_RCPB_CLOCK_GATE_DISABLE (1 << 28) +# define I965_DAP_CLOCK_GATE_DISABLE (1 << 27) +# define I965_ROC_CLOCK_GATE_DISABLE (1 << 26) +# define I965_GW_CLOCK_GATE_DISABLE (1 << 25) +# define I965_TD_CLOCK_GATE_DISABLE (1 << 24) +/** This bit must always be set on 965G */ +# define I965_ISC_CLOCK_GATE_DISABLE (1 << 23) +# define I965_IC_CLOCK_GATE_DISABLE (1 << 22) +# define I965_EU_CLOCK_GATE_DISABLE (1 << 21) +# define I965_IF_CLOCK_GATE_DISABLE (1 << 20) +# define I965_TC_CLOCK_GATE_DISABLE (1 << 19) +# define I965_SO_CLOCK_GATE_DISABLE (1 << 17) +# define I965_FBC_CLOCK_GATE_DISABLE (1 << 16) +# define I965_MARI_CLOCK_GATE_DISABLE (1 << 15) +# define I965_MASF_CLOCK_GATE_DISABLE (1 << 14) +# define I965_MAWB_CLOCK_GATE_DISABLE (1 << 13) +# define I965_EM_CLOCK_GATE_DISABLE (1 << 12) +# define I965_UC_CLOCK_GATE_DISABLE (1 << 11) +# define I965_SI_CLOCK_GATE_DISABLE (1 << 6) +# define I965_MT_CLOCK_GATE_DISABLE (1 << 5) +# define I965_PL_CLOCK_GATE_DISABLE (1 << 4) +# define I965_DG_CLOCK_GATE_DISABLE (1 << 3) +# define I965_QC_CLOCK_GATE_DISABLE (1 << 2) +# define I965_FT_CLOCK_GATE_DISABLE (1 << 1) +# define I965_DM_CLOCK_GATE_DISABLE (1 << 0) + +#define RENCLK_GATE_D2 0x6208 +#define VF_UNIT_CLOCK_GATE_DISABLE (1 << 9) +#define GS_UNIT_CLOCK_GATE_DISABLE (1 << 7) +#define CL_UNIT_CLOCK_GATE_DISABLE (1 << 6) +#define RAMCLK_GATE_D 0x6210 /* CRL only */ +#define DEUC 0x6214 /* CRL only */ + +/* + * Palette regs + */ + +#define _PALETTE_A 0x0a000 +#define _PALETTE_B 0x0a800 +#define PALETTE(pipe) _PIPE(pipe, _PALETTE_A, _PALETTE_B) + +/* MCH MMIO space */ + +/* + * MCHBAR mirror. + * + * This mirrors the MCHBAR MMIO space whose location is determined by + * device 0 function 0's pci config register 0x44 or 0x48 and matches it in + * every way. It is not accessible from the CP register read instructions. + * + */ +#define MCHBAR_MIRROR_BASE 0x10000 + +#define MCHBAR_MIRROR_BASE_SNB 0x140000 + +/** 915-945 and GM965 MCH register controlling DRAM channel access */ +#define DCC 0x10200 +#define DCC_ADDRESSING_MODE_SINGLE_CHANNEL (0 << 0) +#define DCC_ADDRESSING_MODE_DUAL_CHANNEL_ASYMMETRIC (1 << 0) +#define DCC_ADDRESSING_MODE_DUAL_CHANNEL_INTERLEAVED (2 << 0) +#define DCC_ADDRESSING_MODE_MASK (3 << 0) +#define DCC_CHANNEL_XOR_DISABLE (1 << 10) +#define DCC_CHANNEL_XOR_BIT_17 (1 << 9) + +/** Pineview MCH register contains DDR3 setting */ +#define CSHRDDR3CTL 0x101a8 +#define CSHRDDR3CTL_DDR3 (1 << 2) + +/** 965 MCH register controlling DRAM channel configuration */ +#define C0DRB3 0x10206 +#define C1DRB3 0x10606 + +/* Clocking configuration register */ +#define CLKCFG 0x10c00 +#define CLKCFG_FSB_400 (5 << 0) /* hrawclk 100 */ +#define CLKCFG_FSB_533 (1 << 0) /* hrawclk 133 */ +#define CLKCFG_FSB_667 (3 << 0) /* hrawclk 166 */ +#define CLKCFG_FSB_800 (2 << 0) /* hrawclk 200 */ +#define CLKCFG_FSB_1067 (6 << 0) /* hrawclk 266 */ +#define CLKCFG_FSB_1333 (7 << 0) /* hrawclk 333 */ +/* Note, below two are guess */ +#define CLKCFG_FSB_1600 (4 << 0) /* hrawclk 400 */ +#define CLKCFG_FSB_1600_ALT (0 << 0) /* hrawclk 400 */ +#define CLKCFG_FSB_MASK (7 << 0) +#define CLKCFG_MEM_533 (1 << 4) +#define CLKCFG_MEM_667 (2 << 4) +#define CLKCFG_MEM_800 (3 << 4) +#define CLKCFG_MEM_MASK (7 << 4) + +#define TSC1 0x11001 +#define TSE (1<<0) +#define TR1 0x11006 +#define TSFS 0x11020 +#define TSFS_SLOPE_MASK 0x0000ff00 +#define TSFS_SLOPE_SHIFT 8 +#define TSFS_INTR_MASK 0x000000ff + +#define CRSTANDVID 0x11100 +#define PXVFREQ_BASE 0x11110 /* P[0-15]VIDFREQ (0x1114c) (Ironlake) */ +#define PXVFREQ_PX_MASK 0x7f000000 +#define PXVFREQ_PX_SHIFT 24 +#define VIDFREQ_BASE 0x11110 +#define VIDFREQ1 0x11110 /* VIDFREQ1-4 (0x1111c) (Cantiga) */ +#define VIDFREQ2 0x11114 +#define VIDFREQ3 0x11118 +#define VIDFREQ4 0x1111c +#define VIDFREQ_P0_MASK 0x1f000000 +#define VIDFREQ_P0_SHIFT 24 +#define VIDFREQ_P0_CSCLK_MASK 0x00f00000 +#define VIDFREQ_P0_CSCLK_SHIFT 20 +#define VIDFREQ_P0_CRCLK_MASK 0x000f0000 +#define VIDFREQ_P0_CRCLK_SHIFT 16 +#define VIDFREQ_P1_MASK 0x00001f00 +#define VIDFREQ_P1_SHIFT 8 +#define VIDFREQ_P1_CSCLK_MASK 0x000000f0 +#define VIDFREQ_P1_CSCLK_SHIFT 4 +#define VIDFREQ_P1_CRCLK_MASK 0x0000000f +#define INTTOEXT_BASE_ILK 0x11300 +#define INTTOEXT_BASE 0x11120 /* INTTOEXT1-8 (0x1113c) */ +#define INTTOEXT_MAP3_SHIFT 24 +#define INTTOEXT_MAP3_MASK (0x1f << INTTOEXT_MAP3_SHIFT) +#define INTTOEXT_MAP2_SHIFT 16 +#define INTTOEXT_MAP2_MASK (0x1f << INTTOEXT_MAP2_SHIFT) +#define INTTOEXT_MAP1_SHIFT 8 +#define INTTOEXT_MAP1_MASK (0x1f << INTTOEXT_MAP1_SHIFT) +#define INTTOEXT_MAP0_SHIFT 0 +#define INTTOEXT_MAP0_MASK (0x1f << INTTOEXT_MAP0_SHIFT) +#define MEMSWCTL 0x11170 /* Ironlake only */ +#define MEMCTL_CMD_MASK 0xe000 +#define MEMCTL_CMD_SHIFT 13 +#define MEMCTL_CMD_RCLK_OFF 0 +#define MEMCTL_CMD_RCLK_ON 1 +#define MEMCTL_CMD_CHFREQ 2 +#define MEMCTL_CMD_CHVID 3 +#define MEMCTL_CMD_VMMOFF 4 +#define MEMCTL_CMD_VMMON 5 +#define MEMCTL_CMD_STS (1<<12) /* write 1 triggers command, clears + when command complete */ +#define MEMCTL_FREQ_MASK 0x0f00 /* jitter, from 0-15 */ +#define MEMCTL_FREQ_SHIFT 8 +#define MEMCTL_SFCAVM (1<<7) +#define MEMCTL_TGT_VID_MASK 0x007f +#define MEMIHYST 0x1117c +#define MEMINTREN 0x11180 /* 16 bits */ +#define MEMINT_RSEXIT_EN (1<<8) +#define MEMINT_CX_SUPR_EN (1<<7) +#define MEMINT_CONT_BUSY_EN (1<<6) +#define MEMINT_AVG_BUSY_EN (1<<5) +#define MEMINT_EVAL_CHG_EN (1<<4) +#define MEMINT_MON_IDLE_EN (1<<3) +#define MEMINT_UP_EVAL_EN (1<<2) +#define MEMINT_DOWN_EVAL_EN (1<<1) +#define MEMINT_SW_CMD_EN (1<<0) +#define MEMINTRSTR 0x11182 /* 16 bits */ +#define MEM_RSEXIT_MASK 0xc000 +#define MEM_RSEXIT_SHIFT 14 +#define MEM_CONT_BUSY_MASK 0x3000 +#define MEM_CONT_BUSY_SHIFT 12 +#define MEM_AVG_BUSY_MASK 0x0c00 +#define MEM_AVG_BUSY_SHIFT 10 +#define MEM_EVAL_CHG_MASK 0x0300 +#define MEM_EVAL_BUSY_SHIFT 8 +#define MEM_MON_IDLE_MASK 0x00c0 +#define MEM_MON_IDLE_SHIFT 6 +#define MEM_UP_EVAL_MASK 0x0030 +#define MEM_UP_EVAL_SHIFT 4 +#define MEM_DOWN_EVAL_MASK 0x000c +#define MEM_DOWN_EVAL_SHIFT 2 +#define MEM_SW_CMD_MASK 0x0003 +#define MEM_INT_STEER_GFX 0 +#define MEM_INT_STEER_CMR 1 +#define MEM_INT_STEER_SMI 2 +#define MEM_INT_STEER_SCI 3 +#define MEMINTRSTS 0x11184 +#define MEMINT_RSEXIT (1<<7) +#define MEMINT_CONT_BUSY (1<<6) +#define MEMINT_AVG_BUSY (1<<5) +#define MEMINT_EVAL_CHG (1<<4) +#define MEMINT_MON_IDLE (1<<3) +#define MEMINT_UP_EVAL (1<<2) +#define MEMINT_DOWN_EVAL (1<<1) +#define MEMINT_SW_CMD (1<<0) +#define MEMMODECTL 0x11190 +#define MEMMODE_BOOST_EN (1<<31) +#define MEMMODE_BOOST_FREQ_MASK 0x0f000000 /* jitter for boost, 0-15 */ +#define MEMMODE_BOOST_FREQ_SHIFT 24 +#define MEMMODE_IDLE_MODE_MASK 0x00030000 +#define MEMMODE_IDLE_MODE_SHIFT 16 +#define MEMMODE_IDLE_MODE_EVAL 0 +#define MEMMODE_IDLE_MODE_CONT 1 +#define MEMMODE_HWIDLE_EN (1<<15) +#define MEMMODE_SWMODE_EN (1<<14) +#define MEMMODE_RCLK_GATE (1<<13) +#define MEMMODE_HW_UPDATE (1<<12) +#define MEMMODE_FSTART_MASK 0x00000f00 /* starting jitter, 0-15 */ +#define MEMMODE_FSTART_SHIFT 8 +#define MEMMODE_FMAX_MASK 0x000000f0 /* max jitter, 0-15 */ +#define MEMMODE_FMAX_SHIFT 4 +#define MEMMODE_FMIN_MASK 0x0000000f /* min jitter, 0-15 */ +#define RCBMAXAVG 0x1119c +#define MEMSWCTL2 0x1119e /* Cantiga only */ +#define SWMEMCMD_RENDER_OFF (0 << 13) +#define SWMEMCMD_RENDER_ON (1 << 13) +#define SWMEMCMD_SWFREQ (2 << 13) +#define SWMEMCMD_TARVID (3 << 13) +#define SWMEMCMD_VRM_OFF (4 << 13) +#define SWMEMCMD_VRM_ON (5 << 13) +#define CMDSTS (1<<12) +#define SFCAVM (1<<11) +#define SWFREQ_MASK 0x0380 /* P0-7 */ +#define SWFREQ_SHIFT 7 +#define TARVID_MASK 0x001f +#define MEMSTAT_CTG 0x111a0 +#define RCBMINAVG 0x111a0 +#define RCUPEI 0x111b0 +#define RCDNEI 0x111b4 +#define RSTDBYCTL 0x111b8 +#define RS1EN (1<<31) +#define RS2EN (1<<30) +#define RS3EN (1<<29) +#define D3RS3EN (1<<28) /* Display D3 imlies RS3 */ +#define SWPROMORSX (1<<27) /* RSx promotion timers ignored */ +#define RCWAKERW (1<<26) /* Resetwarn from PCH causes wakeup */ +#define DPRSLPVREN (1<<25) /* Fast voltage ramp enable */ +#define GFXTGHYST (1<<24) /* Hysteresis to allow trunk gating */ +#define RCX_SW_EXIT (1<<23) /* Leave RSx and prevent re-entry */ +#define RSX_STATUS_MASK (7<<20) +#define RSX_STATUS_ON (0<<20) +#define RSX_STATUS_RC1 (1<<20) +#define RSX_STATUS_RC1E (2<<20) +#define RSX_STATUS_RS1 (3<<20) +#define RSX_STATUS_RS2 (4<<20) /* aka rc6 */ +#define RSX_STATUS_RSVD (5<<20) /* deep rc6 unsupported on ilk */ +#define RSX_STATUS_RS3 (6<<20) /* rs3 unsupported on ilk */ +#define RSX_STATUS_RSVD2 (7<<20) +#define UWRCRSXE (1<<19) /* wake counter limit prevents rsx */ +#define RSCRP (1<<18) /* rs requests control on rs1/2 reqs */ +#define JRSC (1<<17) /* rsx coupled to cpu c-state */ +#define RS2INC0 (1<<16) /* allow rs2 in cpu c0 */ +#define RS1CONTSAV_MASK (3<<14) +#define RS1CONTSAV_NO_RS1 (0<<14) /* rs1 doesn't save/restore context */ +#define RS1CONTSAV_RSVD (1<<14) +#define RS1CONTSAV_SAVE_RS1 (2<<14) /* rs1 saves context */ +#define RS1CONTSAV_FULL_RS1 (3<<14) /* rs1 saves and restores context */ +#define NORMSLEXLAT_MASK (3<<12) +#define SLOW_RS123 (0<<12) +#define SLOW_RS23 (1<<12) +#define SLOW_RS3 (2<<12) +#define NORMAL_RS123 (3<<12) +#define RCMODE_TIMEOUT (1<<11) /* 0 is eval interval method */ +#define IMPROMOEN (1<<10) /* promo is immediate or delayed until next idle interval (only for timeout method above) */ +#define RCENTSYNC (1<<9) /* rs coupled to cpu c-state (3/6/7) */ +#define STATELOCK (1<<7) /* locked to rs_cstate if 0 */ +#define RS_CSTATE_MASK (3<<4) +#define RS_CSTATE_C367_RS1 (0<<4) +#define RS_CSTATE_C36_RS1_C7_RS2 (1<<4) +#define RS_CSTATE_RSVD (2<<4) +#define RS_CSTATE_C367_RS2 (3<<4) +#define REDSAVES (1<<3) /* no context save if was idle during rs0 */ +#define REDRESTORES (1<<2) /* no restore if was idle during rs0 */ +#define VIDCTL 0x111c0 +#define VIDSTS 0x111c8 +#define VIDSTART 0x111cc /* 8 bits */ +#define MEMSTAT_ILK 0x111f8 +#define MEMSTAT_VID_MASK 0x7f00 +#define MEMSTAT_VID_SHIFT 8 +#define MEMSTAT_PSTATE_MASK 0x00f8 +#define MEMSTAT_PSTATE_SHIFT 3 +#define MEMSTAT_MON_ACTV (1<<2) +#define MEMSTAT_SRC_CTL_MASK 0x0003 +#define MEMSTAT_SRC_CTL_CORE 0 +#define MEMSTAT_SRC_CTL_TRB 1 +#define MEMSTAT_SRC_CTL_THM 2 +#define MEMSTAT_SRC_CTL_STDBY 3 +#define RCPREVBSYTUPAVG 0x113b8 +#define RCPREVBSYTDNAVG 0x113bc +#define PMMISC 0x11214 +#define MCPPCE_EN (1<<0) /* enable PM_MSG from PCH->MPC */ +#define SDEW 0x1124c +#define CSIEW0 0x11250 +#define CSIEW1 0x11254 +#define CSIEW2 0x11258 +#define PEW 0x1125c +#define DEW 0x11270 +#define MCHAFE 0x112c0 +#define CSIEC 0x112e0 +#define DMIEC 0x112e4 +#define DDREC 0x112e8 +#define PEG0EC 0x112ec +#define PEG1EC 0x112f0 +#define GFXEC 0x112f4 +#define RPPREVBSYTUPAVG 0x113b8 +#define RPPREVBSYTDNAVG 0x113bc +#define ECR 0x11600 +#define ECR_GPFE (1<<31) +#define ECR_IMONE (1<<30) +#define ECR_CAP_MASK 0x0000001f /* Event range, 0-31 */ +#define OGW0 0x11608 +#define OGW1 0x1160c +#define EG0 0x11610 +#define EG1 0x11614 +#define EG2 0x11618 +#define EG3 0x1161c +#define EG4 0x11620 +#define EG5 0x11624 +#define EG6 0x11628 +#define EG7 0x1162c +#define PXW 0x11664 +#define PXWL 0x11680 +#define LCFUSE02 0x116c0 +#define LCFUSE_HIV_MASK 0x000000ff +#define CSIPLL0 0x12c10 +#define DDRMPLL1 0X12c20 +#define PEG_BAND_GAP_DATA 0x14d68 + +#define GEN6_GT_PERF_STATUS 0x145948 +#define GEN6_RP_STATE_LIMITS 0x145994 +#define GEN6_RP_STATE_CAP 0x145998 + +/* + * Logical Context regs + */ +#define CCID 0x2180 +#define CCID_EN (1<<0) +/* + * Overlay regs + */ + +#define OVADD 0x30000 +#define DOVSTA 0x30008 +#define OC_BUF (0x3<<20) +#define OGAMC5 0x30010 +#define OGAMC4 0x30014 +#define OGAMC3 0x30018 +#define OGAMC2 0x3001c +#define OGAMC1 0x30020 +#define OGAMC0 0x30024 + +/* + * Display engine regs + */ + +/* Pipe A timing regs */ +#define _HTOTAL_A 0x60000 +#define _HBLANK_A 0x60004 +#define _HSYNC_A 0x60008 +#define _VTOTAL_A 0x6000c +#define _VBLANK_A 0x60010 +#define _VSYNC_A 0x60014 +#define _PIPEASRC 0x6001c +#define _BCLRPAT_A 0x60020 + +/* Pipe B timing regs */ +#define _HTOTAL_B 0x61000 +#define _HBLANK_B 0x61004 +#define _HSYNC_B 0x61008 +#define _VTOTAL_B 0x6100c +#define _VBLANK_B 0x61010 +#define _VSYNC_B 0x61014 +#define _PIPEBSRC 0x6101c +#define _BCLRPAT_B 0x61020 + +#define HTOTAL(pipe) _PIPE(pipe, _HTOTAL_A, _HTOTAL_B) +#define HBLANK(pipe) _PIPE(pipe, _HBLANK_A, _HBLANK_B) +#define HSYNC(pipe) _PIPE(pipe, _HSYNC_A, _HSYNC_B) +#define VTOTAL(pipe) _PIPE(pipe, _VTOTAL_A, _VTOTAL_B) +#define VBLANK(pipe) _PIPE(pipe, _VBLANK_A, _VBLANK_B) +#define VSYNC(pipe) _PIPE(pipe, _VSYNC_A, _VSYNC_B) +#define BCLRPAT(pipe) _PIPE(pipe, _BCLRPAT_A, _BCLRPAT_B) + +/* VGA port control */ +#define ADPA 0x61100 +#define ADPA_DAC_ENABLE (1<<31) +#define ADPA_DAC_DISABLE 0 +#define ADPA_PIPE_SELECT_MASK (1<<30) +#define ADPA_PIPE_A_SELECT 0 +#define ADPA_PIPE_B_SELECT (1<<30) +#define ADPA_USE_VGA_HVPOLARITY (1<<15) +#define ADPA_SETS_HVPOLARITY 0 +#define ADPA_VSYNC_CNTL_DISABLE (1<<11) +#define ADPA_VSYNC_CNTL_ENABLE 0 +#define ADPA_HSYNC_CNTL_DISABLE (1<<10) +#define ADPA_HSYNC_CNTL_ENABLE 0 +#define ADPA_VSYNC_ACTIVE_HIGH (1<<4) +#define ADPA_VSYNC_ACTIVE_LOW 0 +#define ADPA_HSYNC_ACTIVE_HIGH (1<<3) +#define ADPA_HSYNC_ACTIVE_LOW 0 +#define ADPA_DPMS_MASK (~(3<<10)) +#define ADPA_DPMS_ON (0<<10) +#define ADPA_DPMS_SUSPEND (1<<10) +#define ADPA_DPMS_STANDBY (2<<10) +#define ADPA_DPMS_OFF (3<<10) + + +/* Hotplug control (945+ only) */ +#define PORT_HOTPLUG_EN 0x61110 +#define HDMIB_HOTPLUG_INT_EN (1 << 29) +#define DPB_HOTPLUG_INT_EN (1 << 29) +#define HDMIC_HOTPLUG_INT_EN (1 << 28) +#define DPC_HOTPLUG_INT_EN (1 << 28) +#define HDMID_HOTPLUG_INT_EN (1 << 27) +#define DPD_HOTPLUG_INT_EN (1 << 27) +#define SDVOB_HOTPLUG_INT_EN (1 << 26) +#define SDVOC_HOTPLUG_INT_EN (1 << 25) +#define TV_HOTPLUG_INT_EN (1 << 18) +#define CRT_HOTPLUG_INT_EN (1 << 9) +#define CRT_HOTPLUG_FORCE_DETECT (1 << 3) +#define CRT_HOTPLUG_ACTIVATION_PERIOD_32 (0 << 8) +/* must use period 64 on GM45 according to docs */ +#define CRT_HOTPLUG_ACTIVATION_PERIOD_64 (1 << 8) +#define CRT_HOTPLUG_DAC_ON_TIME_2M (0 << 7) +#define CRT_HOTPLUG_DAC_ON_TIME_4M (1 << 7) +#define CRT_HOTPLUG_VOLTAGE_COMPARE_40 (0 << 5) +#define CRT_HOTPLUG_VOLTAGE_COMPARE_50 (1 << 5) +#define CRT_HOTPLUG_VOLTAGE_COMPARE_60 (2 << 5) +#define CRT_HOTPLUG_VOLTAGE_COMPARE_70 (3 << 5) +#define CRT_HOTPLUG_VOLTAGE_COMPARE_MASK (3 << 5) +#define CRT_HOTPLUG_DETECT_DELAY_1G (0 << 4) +#define CRT_HOTPLUG_DETECT_DELAY_2G (1 << 4) +#define CRT_HOTPLUG_DETECT_VOLTAGE_325MV (0 << 2) +#define CRT_HOTPLUG_DETECT_VOLTAGE_475MV (1 << 2) + +#define PORT_HOTPLUG_STAT 0x61114 +#define HDMIB_HOTPLUG_INT_STATUS (1 << 29) +#define DPB_HOTPLUG_INT_STATUS (1 << 29) +#define HDMIC_HOTPLUG_INT_STATUS (1 << 28) +#define DPC_HOTPLUG_INT_STATUS (1 << 28) +#define HDMID_HOTPLUG_INT_STATUS (1 << 27) +#define DPD_HOTPLUG_INT_STATUS (1 << 27) +#define CRT_HOTPLUG_INT_STATUS (1 << 11) +#define TV_HOTPLUG_INT_STATUS (1 << 10) +#define CRT_HOTPLUG_MONITOR_MASK (3 << 8) +#define CRT_HOTPLUG_MONITOR_COLOR (3 << 8) +#define CRT_HOTPLUG_MONITOR_MONO (2 << 8) +#define CRT_HOTPLUG_MONITOR_NONE (0 << 8) +#define SDVOC_HOTPLUG_INT_STATUS (1 << 7) +#define SDVOB_HOTPLUG_INT_STATUS (1 << 6) + +/* SDVO port control */ +#define SDVOB 0x61140 +#define SDVOC 0x61160 +#define SDVO_ENABLE (1 << 31) +#define SDVO_PIPE_B_SELECT (1 << 30) +#define SDVO_STALL_SELECT (1 << 29) +#define SDVO_INTERRUPT_ENABLE (1 << 26) +/** + * 915G/GM SDVO pixel multiplier. + * + * Programmed value is multiplier - 1, up to 5x. + * + * \sa DPLL_MD_UDI_MULTIPLIER_MASK + */ +#define SDVO_PORT_MULTIPLY_MASK (7 << 23) +#define SDVO_PORT_MULTIPLY_SHIFT 23 +#define SDVO_PHASE_SELECT_MASK (15 << 19) +#define SDVO_PHASE_SELECT_DEFAULT (6 << 19) +#define SDVO_CLOCK_OUTPUT_INVERT (1 << 18) +#define SDVOC_GANG_MODE (1 << 16) +#define SDVO_ENCODING_SDVO (0x0 << 10) +#define SDVO_ENCODING_HDMI (0x2 << 10) +/** Requird for HDMI operation */ +#define SDVO_NULL_PACKETS_DURING_VSYNC (1 << 9) +#define SDVO_COLOR_RANGE_16_235 (1 << 8) +#define SDVO_BORDER_ENABLE (1 << 7) +#define SDVO_AUDIO_ENABLE (1 << 6) +/** New with 965, default is to be set */ +#define SDVO_VSYNC_ACTIVE_HIGH (1 << 4) +/** New with 965, default is to be set */ +#define SDVO_HSYNC_ACTIVE_HIGH (1 << 3) +#define SDVOB_PCIE_CONCURRENCY (1 << 3) +#define SDVO_DETECTED (1 << 2) +/* Bits to be preserved when writing */ +#define SDVOB_PRESERVE_MASK ((1 << 17) | (1 << 16) | (1 << 14) | (1 << 26)) +#define SDVOC_PRESERVE_MASK ((1 << 17) | (1 << 26)) + +/* DVO port control */ +#define DVOA 0x61120 +#define DVOB 0x61140 +#define DVOC 0x61160 +#define DVO_ENABLE (1 << 31) +#define DVO_PIPE_B_SELECT (1 << 30) +#define DVO_PIPE_STALL_UNUSED (0 << 28) +#define DVO_PIPE_STALL (1 << 28) +#define DVO_PIPE_STALL_TV (2 << 28) +#define DVO_PIPE_STALL_MASK (3 << 28) +#define DVO_USE_VGA_SYNC (1 << 15) +#define DVO_DATA_ORDER_I740 (0 << 14) +#define DVO_DATA_ORDER_FP (1 << 14) +#define DVO_VSYNC_DISABLE (1 << 11) +#define DVO_HSYNC_DISABLE (1 << 10) +#define DVO_VSYNC_TRISTATE (1 << 9) +#define DVO_HSYNC_TRISTATE (1 << 8) +#define DVO_BORDER_ENABLE (1 << 7) +#define DVO_DATA_ORDER_GBRG (1 << 6) +#define DVO_DATA_ORDER_RGGB (0 << 6) +#define DVO_DATA_ORDER_GBRG_ERRATA (0 << 6) +#define DVO_DATA_ORDER_RGGB_ERRATA (1 << 6) +#define DVO_VSYNC_ACTIVE_HIGH (1 << 4) +#define DVO_HSYNC_ACTIVE_HIGH (1 << 3) +#define DVO_BLANK_ACTIVE_HIGH (1 << 2) +#define DVO_OUTPUT_CSTATE_PIXELS (1 << 1) /* SDG only */ +#define DVO_OUTPUT_SOURCE_SIZE_PIXELS (1 << 0) /* SDG only */ +#define DVO_PRESERVE_MASK (0x7<<24) +#define DVOA_SRCDIM 0x61124 +#define DVOB_SRCDIM 0x61144 +#define DVOC_SRCDIM 0x61164 +#define DVO_SRCDIM_HORIZONTAL_SHIFT 12 +#define DVO_SRCDIM_VERTICAL_SHIFT 0 + +/* LVDS port control */ +#define LVDS 0x61180 +/* + * Enables the LVDS port. This bit must be set before DPLLs are enabled, as + * the DPLL semantics change when the LVDS is assigned to that pipe. + */ +#define LVDS_PORT_EN (1 << 31) +/* Selects pipe B for LVDS data. Must be set on pre-965. */ +#define LVDS_PIPEB_SELECT (1 << 30) +#define LVDS_PIPE_MASK (1 << 30) +/* LVDS dithering flag on 965/g4x platform */ +#define LVDS_ENABLE_DITHER (1 << 25) +/* LVDS sync polarity flags. Set to invert (i.e. negative) */ +#define LVDS_VSYNC_POLARITY (1 << 21) +#define LVDS_HSYNC_POLARITY (1 << 20) + +/* Enable border for unscaled (or aspect-scaled) display */ +#define LVDS_BORDER_ENABLE (1 << 15) +/* + * Enables the A0-A2 data pairs and CLKA, containing 18 bits of color data per + * pixel. + */ +#define LVDS_A0A2_CLKA_POWER_MASK (3 << 8) +#define LVDS_A0A2_CLKA_POWER_DOWN (0 << 8) +#define LVDS_A0A2_CLKA_POWER_UP (3 << 8) +/* + * Controls the A3 data pair, which contains the additional LSBs for 24 bit + * mode. Only enabled if LVDS_A0A2_CLKA_POWER_UP also indicates it should be + * on. + */ +#define LVDS_A3_POWER_MASK (3 << 6) +#define LVDS_A3_POWER_DOWN (0 << 6) +#define LVDS_A3_POWER_UP (3 << 6) +/* + * Controls the CLKB pair. This should only be set when LVDS_B0B3_POWER_UP + * is set. + */ +#define LVDS_CLKB_POWER_MASK (3 << 4) +#define LVDS_CLKB_POWER_DOWN (0 << 4) +#define LVDS_CLKB_POWER_UP (3 << 4) +/* + * Controls the B0-B3 data pairs. This must be set to match the DPLL p2 + * setting for whether we are in dual-channel mode. The B3 pair will + * additionally only be powered up when LVDS_A3_POWER_UP is set. + */ +#define LVDS_B0B3_POWER_MASK (3 << 2) +#define LVDS_B0B3_POWER_DOWN (0 << 2) +#define LVDS_B0B3_POWER_UP (3 << 2) + +#define LVDS_PIPE_ENABLED(V, P) \ + (((V) & (LVDS_PIPE_MASK | LVDS_PORT_EN)) == ((P) << 30 | LVDS_PORT_EN)) + +/* Video Data Island Packet control */ +#define VIDEO_DIP_DATA 0x61178 +#define VIDEO_DIP_CTL 0x61170 +#define VIDEO_DIP_ENABLE (1 << 31) +#define VIDEO_DIP_PORT_B (1 << 29) +#define VIDEO_DIP_PORT_C (2 << 29) +#define VIDEO_DIP_ENABLE_AVI (1 << 21) +#define VIDEO_DIP_ENABLE_VENDOR (2 << 21) +#define VIDEO_DIP_ENABLE_SPD (8 << 21) +#define VIDEO_DIP_SELECT_AVI (0 << 19) +#define VIDEO_DIP_SELECT_VENDOR (1 << 19) +#define VIDEO_DIP_SELECT_SPD (3 << 19) +#define VIDEO_DIP_FREQ_ONCE (0 << 16) +#define VIDEO_DIP_FREQ_VSYNC (1 << 16) +#define VIDEO_DIP_FREQ_2VSYNC (2 << 16) + +/* Panel power sequencing */ +#define PP_STATUS 0x61200 +#define PP_ON (1 << 31) +/* + * Indicates that all dependencies of the panel are on: + * + * - PLL enabled + * - pipe enabled + * - LVDS/DVOB/DVOC on + */ +#define PP_READY (1 << 30) +#define PP_SEQUENCE_NONE (0 << 28) +#define PP_SEQUENCE_ON (1 << 28) +#define PP_SEQUENCE_OFF (2 << 28) +#define PP_SEQUENCE_MASK 0x30000000 +#define PP_CYCLE_DELAY_ACTIVE (1 << 27) +#define PP_SEQUENCE_STATE_ON_IDLE (1 << 3) +#define PP_SEQUENCE_STATE_MASK 0x0000000f +#define PP_CONTROL 0x61204 +#define POWER_TARGET_ON (1 << 0) +#define PP_ON_DELAYS 0x61208 +#define PP_OFF_DELAYS 0x6120c +#define PP_DIVISOR 0x61210 + +/* Panel fitting */ +#define PFIT_CONTROL 0x61230 +#define PFIT_ENABLE (1 << 31) +#define PFIT_PIPE_MASK (3 << 29) +#define PFIT_PIPE_SHIFT 29 +#define VERT_INTERP_DISABLE (0 << 10) +#define VERT_INTERP_BILINEAR (1 << 10) +#define VERT_INTERP_MASK (3 << 10) +#define VERT_AUTO_SCALE (1 << 9) +#define HORIZ_INTERP_DISABLE (0 << 6) +#define HORIZ_INTERP_BILINEAR (1 << 6) +#define HORIZ_INTERP_MASK (3 << 6) +#define HORIZ_AUTO_SCALE (1 << 5) +#define PANEL_8TO6_DITHER_ENABLE (1 << 3) +#define PFIT_FILTER_FUZZY (0 << 24) +#define PFIT_SCALING_AUTO (0 << 26) +#define PFIT_SCALING_PROGRAMMED (1 << 26) +#define PFIT_SCALING_PILLAR (2 << 26) +#define PFIT_SCALING_LETTER (3 << 26) +#define PFIT_PGM_RATIOS 0x61234 +#define PFIT_VERT_SCALE_MASK 0xfff00000 +#define PFIT_HORIZ_SCALE_MASK 0x0000fff0 +/* Pre-965 */ +#define PFIT_VERT_SCALE_SHIFT 20 +#define PFIT_VERT_SCALE_MASK 0xfff00000 +#define PFIT_HORIZ_SCALE_SHIFT 4 +#define PFIT_HORIZ_SCALE_MASK 0x0000fff0 +/* 965+ */ +#define PFIT_VERT_SCALE_SHIFT_965 16 +#define PFIT_VERT_SCALE_MASK_965 0x1fff0000 +#define PFIT_HORIZ_SCALE_SHIFT_965 0 +#define PFIT_HORIZ_SCALE_MASK_965 0x00001fff + +#define PFIT_AUTO_RATIOS 0x61238 + +/* Backlight control */ +#define BLC_PWM_CTL 0x61254 +#define BACKLIGHT_MODULATION_FREQ_SHIFT (17) +#define BLC_PWM_CTL2 0x61250 /* 965+ only */ +#define BLM_COMBINATION_MODE (1 << 30) +/* + * This is the most significant 15 bits of the number of backlight cycles in a + * complete cycle of the modulated backlight control. + * + * The actual value is this field multiplied by two. + */ +#define BACKLIGHT_MODULATION_FREQ_MASK (0x7fff << 17) +#define BLM_LEGACY_MODE (1 << 16) +/* + * This is the number of cycles out of the backlight modulation cycle for which + * the backlight is on. + * + * This field must be no greater than the number of cycles in the complete + * backlight modulation cycle. + */ +#define BACKLIGHT_DUTY_CYCLE_SHIFT (0) +#define BACKLIGHT_DUTY_CYCLE_MASK (0xffff) + +#define BLC_HIST_CTL 0x61260 + +/* TV port control */ +#define TV_CTL 0x68000 +/** Enables the TV encoder */ +# define TV_ENC_ENABLE (1 << 31) +/** Sources the TV encoder input from pipe B instead of A. */ +# define TV_ENC_PIPEB_SELECT (1 << 30) +/** Outputs composite video (DAC A only) */ +# define TV_ENC_OUTPUT_COMPOSITE (0 << 28) +/** Outputs SVideo video (DAC B/C) */ +# define TV_ENC_OUTPUT_SVIDEO (1 << 28) +/** Outputs Component video (DAC A/B/C) */ +# define TV_ENC_OUTPUT_COMPONENT (2 << 28) +/** Outputs Composite and SVideo (DAC A/B/C) */ +# define TV_ENC_OUTPUT_SVIDEO_COMPOSITE (3 << 28) +# define TV_TRILEVEL_SYNC (1 << 21) +/** Enables slow sync generation (945GM only) */ +# define TV_SLOW_SYNC (1 << 20) +/** Selects 4x oversampling for 480i and 576p */ +# define TV_OVERSAMPLE_4X (0 << 18) +/** Selects 2x oversampling for 720p and 1080i */ +# define TV_OVERSAMPLE_2X (1 << 18) +/** Selects no oversampling for 1080p */ +# define TV_OVERSAMPLE_NONE (2 << 18) +/** Selects 8x oversampling */ +# define TV_OVERSAMPLE_8X (3 << 18) +/** Selects progressive mode rather than interlaced */ +# define TV_PROGRESSIVE (1 << 17) +/** Sets the colorburst to PAL mode. Required for non-M PAL modes. */ +# define TV_PAL_BURST (1 << 16) +/** Field for setting delay of Y compared to C */ +# define TV_YC_SKEW_MASK (7 << 12) +/** Enables a fix for 480p/576p standard definition modes on the 915GM only */ +# define TV_ENC_SDP_FIX (1 << 11) +/** + * Enables a fix for the 915GM only. + * + * Not sure what it does. + */ +# define TV_ENC_C0_FIX (1 << 10) +/** Bits that must be preserved by software */ +# define TV_CTL_SAVE ((1 << 11) | (3 << 9) | (7 << 6) | 0xf) +# define TV_FUSE_STATE_MASK (3 << 4) +/** Read-only state that reports all features enabled */ +# define TV_FUSE_STATE_ENABLED (0 << 4) +/** Read-only state that reports that Macrovision is disabled in hardware*/ +# define TV_FUSE_STATE_NO_MACROVISION (1 << 4) +/** Read-only state that reports that TV-out is disabled in hardware. */ +# define TV_FUSE_STATE_DISABLED (2 << 4) +/** Normal operation */ +# define TV_TEST_MODE_NORMAL (0 << 0) +/** Encoder test pattern 1 - combo pattern */ +# define TV_TEST_MODE_PATTERN_1 (1 << 0) +/** Encoder test pattern 2 - full screen vertical 75% color bars */ +# define TV_TEST_MODE_PATTERN_2 (2 << 0) +/** Encoder test pattern 3 - full screen horizontal 75% color bars */ +# define TV_TEST_MODE_PATTERN_3 (3 << 0) +/** Encoder test pattern 4 - random noise */ +# define TV_TEST_MODE_PATTERN_4 (4 << 0) +/** Encoder test pattern 5 - linear color ramps */ +# define TV_TEST_MODE_PATTERN_5 (5 << 0) +/** + * This test mode forces the DACs to 50% of full output. + * + * This is used for load detection in combination with TVDAC_SENSE_MASK + */ +# define TV_TEST_MODE_MONITOR_DETECT (7 << 0) +# define TV_TEST_MODE_MASK (7 << 0) + +#define TV_DAC 0x68004 +# define TV_DAC_SAVE 0x00ffff00 +/** + * Reports that DAC state change logic has reported change (RO). + * + * This gets cleared when TV_DAC_STATE_EN is cleared +*/ +# define TVDAC_STATE_CHG (1 << 31) +# define TVDAC_SENSE_MASK (7 << 28) +/** Reports that DAC A voltage is above the detect threshold */ +# define TVDAC_A_SENSE (1 << 30) +/** Reports that DAC B voltage is above the detect threshold */ +# define TVDAC_B_SENSE (1 << 29) +/** Reports that DAC C voltage is above the detect threshold */ +# define TVDAC_C_SENSE (1 << 28) +/** + * Enables DAC state detection logic, for load-based TV detection. + * + * The PLL of the chosen pipe (in TV_CTL) must be running, and the encoder set + * to off, for load detection to work. + */ +# define TVDAC_STATE_CHG_EN (1 << 27) +/** Sets the DAC A sense value to high */ +# define TVDAC_A_SENSE_CTL (1 << 26) +/** Sets the DAC B sense value to high */ +# define TVDAC_B_SENSE_CTL (1 << 25) +/** Sets the DAC C sense value to high */ +# define TVDAC_C_SENSE_CTL (1 << 24) +/** Overrides the ENC_ENABLE and DAC voltage levels */ +# define DAC_CTL_OVERRIDE (1 << 7) +/** Sets the slew rate. Must be preserved in software */ +# define ENC_TVDAC_SLEW_FAST (1 << 6) +# define DAC_A_1_3_V (0 << 4) +# define DAC_A_1_1_V (1 << 4) +# define DAC_A_0_7_V (2 << 4) +# define DAC_A_MASK (3 << 4) +# define DAC_B_1_3_V (0 << 2) +# define DAC_B_1_1_V (1 << 2) +# define DAC_B_0_7_V (2 << 2) +# define DAC_B_MASK (3 << 2) +# define DAC_C_1_3_V (0 << 0) +# define DAC_C_1_1_V (1 << 0) +# define DAC_C_0_7_V (2 << 0) +# define DAC_C_MASK (3 << 0) + +/** + * CSC coefficients are stored in a floating point format with 9 bits of + * mantissa and 2 or 3 bits of exponent. The exponent is represented as 2**-n, + * where 2-bit exponents are unsigned n, and 3-bit exponents are signed n with + * -1 (0x3) being the only legal negative value. + */ +#define TV_CSC_Y 0x68010 +# define TV_RY_MASK 0x07ff0000 +# define TV_RY_SHIFT 16 +# define TV_GY_MASK 0x00000fff +# define TV_GY_SHIFT 0 + +#define TV_CSC_Y2 0x68014 +# define TV_BY_MASK 0x07ff0000 +# define TV_BY_SHIFT 16 +/** + * Y attenuation for component video. + * + * Stored in 1.9 fixed point. + */ +# define TV_AY_MASK 0x000003ff +# define TV_AY_SHIFT 0 + +#define TV_CSC_U 0x68018 +# define TV_RU_MASK 0x07ff0000 +# define TV_RU_SHIFT 16 +# define TV_GU_MASK 0x000007ff +# define TV_GU_SHIFT 0 + +#define TV_CSC_U2 0x6801c +# define TV_BU_MASK 0x07ff0000 +# define TV_BU_SHIFT 16 +/** + * U attenuation for component video. + * + * Stored in 1.9 fixed point. + */ +# define TV_AU_MASK 0x000003ff +# define TV_AU_SHIFT 0 + +#define TV_CSC_V 0x68020 +# define TV_RV_MASK 0x0fff0000 +# define TV_RV_SHIFT 16 +# define TV_GV_MASK 0x000007ff +# define TV_GV_SHIFT 0 + +#define TV_CSC_V2 0x68024 +# define TV_BV_MASK 0x07ff0000 +# define TV_BV_SHIFT 16 +/** + * V attenuation for component video. + * + * Stored in 1.9 fixed point. + */ +# define TV_AV_MASK 0x000007ff +# define TV_AV_SHIFT 0 + +#define TV_CLR_KNOBS 0x68028 +/** 2s-complement brightness adjustment */ +# define TV_BRIGHTNESS_MASK 0xff000000 +# define TV_BRIGHTNESS_SHIFT 24 +/** Contrast adjustment, as a 2.6 unsigned floating point number */ +# define TV_CONTRAST_MASK 0x00ff0000 +# define TV_CONTRAST_SHIFT 16 +/** Saturation adjustment, as a 2.6 unsigned floating point number */ +# define TV_SATURATION_MASK 0x0000ff00 +# define TV_SATURATION_SHIFT 8 +/** Hue adjustment, as an integer phase angle in degrees */ +# define TV_HUE_MASK 0x000000ff +# define TV_HUE_SHIFT 0 + +#define TV_CLR_LEVEL 0x6802c +/** Controls the DAC level for black */ +# define TV_BLACK_LEVEL_MASK 0x01ff0000 +# define TV_BLACK_LEVEL_SHIFT 16 +/** Controls the DAC level for blanking */ +# define TV_BLANK_LEVEL_MASK 0x000001ff +# define TV_BLANK_LEVEL_SHIFT 0 + +#define TV_H_CTL_1 0x68030 +/** Number of pixels in the hsync. */ +# define TV_HSYNC_END_MASK 0x1fff0000 +# define TV_HSYNC_END_SHIFT 16 +/** Total number of pixels minus one in the line (display and blanking). */ +# define TV_HTOTAL_MASK 0x00001fff +# define TV_HTOTAL_SHIFT 0 + +#define TV_H_CTL_2 0x68034 +/** Enables the colorburst (needed for non-component color) */ +# define TV_BURST_ENA (1 << 31) +/** Offset of the colorburst from the start of hsync, in pixels minus one. */ +# define TV_HBURST_START_SHIFT 16 +# define TV_HBURST_START_MASK 0x1fff0000 +/** Length of the colorburst */ +# define TV_HBURST_LEN_SHIFT 0 +# define TV_HBURST_LEN_MASK 0x0001fff + +#define TV_H_CTL_3 0x68038 +/** End of hblank, measured in pixels minus one from start of hsync */ +# define TV_HBLANK_END_SHIFT 16 +# define TV_HBLANK_END_MASK 0x1fff0000 +/** Start of hblank, measured in pixels minus one from start of hsync */ +# define TV_HBLANK_START_SHIFT 0 +# define TV_HBLANK_START_MASK 0x0001fff + +#define TV_V_CTL_1 0x6803c +/** XXX */ +# define TV_NBR_END_SHIFT 16 +# define TV_NBR_END_MASK 0x07ff0000 +/** XXX */ +# define TV_VI_END_F1_SHIFT 8 +# define TV_VI_END_F1_MASK 0x00003f00 +/** XXX */ +# define TV_VI_END_F2_SHIFT 0 +# define TV_VI_END_F2_MASK 0x0000003f + +#define TV_V_CTL_2 0x68040 +/** Length of vsync, in half lines */ +# define TV_VSYNC_LEN_MASK 0x07ff0000 +# define TV_VSYNC_LEN_SHIFT 16 +/** Offset of the start of vsync in field 1, measured in one less than the + * number of half lines. + */ +# define TV_VSYNC_START_F1_MASK 0x00007f00 +# define TV_VSYNC_START_F1_SHIFT 8 +/** + * Offset of the start of vsync in field 2, measured in one less than the + * number of half lines. + */ +# define TV_VSYNC_START_F2_MASK 0x0000007f +# define TV_VSYNC_START_F2_SHIFT 0 + +#define TV_V_CTL_3 0x68044 +/** Enables generation of the equalization signal */ +# define TV_EQUAL_ENA (1 << 31) +/** Length of vsync, in half lines */ +# define TV_VEQ_LEN_MASK 0x007f0000 +# define TV_VEQ_LEN_SHIFT 16 +/** Offset of the start of equalization in field 1, measured in one less than + * the number of half lines. + */ +# define TV_VEQ_START_F1_MASK 0x0007f00 +# define TV_VEQ_START_F1_SHIFT 8 +/** + * Offset of the start of equalization in field 2, measured in one less than + * the number of half lines. + */ +# define TV_VEQ_START_F2_MASK 0x000007f +# define TV_VEQ_START_F2_SHIFT 0 + +#define TV_V_CTL_4 0x68048 +/** + * Offset to start of vertical colorburst, measured in one less than the + * number of lines from vertical start. + */ +# define TV_VBURST_START_F1_MASK 0x003f0000 +# define TV_VBURST_START_F1_SHIFT 16 +/** + * Offset to the end of vertical colorburst, measured in one less than the + * number of lines from the start of NBR. + */ +# define TV_VBURST_END_F1_MASK 0x000000ff +# define TV_VBURST_END_F1_SHIFT 0 + +#define TV_V_CTL_5 0x6804c +/** + * Offset to start of vertical colorburst, measured in one less than the + * number of lines from vertical start. + */ +# define TV_VBURST_START_F2_MASK 0x003f0000 +# define TV_VBURST_START_F2_SHIFT 16 +/** + * Offset to the end of vertical colorburst, measured in one less than the + * number of lines from the start of NBR. + */ +# define TV_VBURST_END_F2_MASK 0x000000ff +# define TV_VBURST_END_F2_SHIFT 0 + +#define TV_V_CTL_6 0x68050 +/** + * Offset to start of vertical colorburst, measured in one less than the + * number of lines from vertical start. + */ +# define TV_VBURST_START_F3_MASK 0x003f0000 +# define TV_VBURST_START_F3_SHIFT 16 +/** + * Offset to the end of vertical colorburst, measured in one less than the + * number of lines from the start of NBR. + */ +# define TV_VBURST_END_F3_MASK 0x000000ff +# define TV_VBURST_END_F3_SHIFT 0 + +#define TV_V_CTL_7 0x68054 +/** + * Offset to start of vertical colorburst, measured in one less than the + * number of lines from vertical start. + */ +# define TV_VBURST_START_F4_MASK 0x003f0000 +# define TV_VBURST_START_F4_SHIFT 16 +/** + * Offset to the end of vertical colorburst, measured in one less than the + * number of lines from the start of NBR. + */ +# define TV_VBURST_END_F4_MASK 0x000000ff +# define TV_VBURST_END_F4_SHIFT 0 + +#define TV_SC_CTL_1 0x68060 +/** Turns on the first subcarrier phase generation DDA */ +# define TV_SC_DDA1_EN (1 << 31) +/** Turns on the first subcarrier phase generation DDA */ +# define TV_SC_DDA2_EN (1 << 30) +/** Turns on the first subcarrier phase generation DDA */ +# define TV_SC_DDA3_EN (1 << 29) +/** Sets the subcarrier DDA to reset frequency every other field */ +# define TV_SC_RESET_EVERY_2 (0 << 24) +/** Sets the subcarrier DDA to reset frequency every fourth field */ +# define TV_SC_RESET_EVERY_4 (1 << 24) +/** Sets the subcarrier DDA to reset frequency every eighth field */ +# define TV_SC_RESET_EVERY_8 (2 << 24) +/** Sets the subcarrier DDA to never reset the frequency */ +# define TV_SC_RESET_NEVER (3 << 24) +/** Sets the peak amplitude of the colorburst.*/ +# define TV_BURST_LEVEL_MASK 0x00ff0000 +# define TV_BURST_LEVEL_SHIFT 16 +/** Sets the increment of the first subcarrier phase generation DDA */ +# define TV_SCDDA1_INC_MASK 0x00000fff +# define TV_SCDDA1_INC_SHIFT 0 + +#define TV_SC_CTL_2 0x68064 +/** Sets the rollover for the second subcarrier phase generation DDA */ +# define TV_SCDDA2_SIZE_MASK 0x7fff0000 +# define TV_SCDDA2_SIZE_SHIFT 16 +/** Sets the increent of the second subcarrier phase generation DDA */ +# define TV_SCDDA2_INC_MASK 0x00007fff +# define TV_SCDDA2_INC_SHIFT 0 + +#define TV_SC_CTL_3 0x68068 +/** Sets the rollover for the third subcarrier phase generation DDA */ +# define TV_SCDDA3_SIZE_MASK 0x7fff0000 +# define TV_SCDDA3_SIZE_SHIFT 16 +/** Sets the increent of the third subcarrier phase generation DDA */ +# define TV_SCDDA3_INC_MASK 0x00007fff +# define TV_SCDDA3_INC_SHIFT 0 + +#define TV_WIN_POS 0x68070 +/** X coordinate of the display from the start of horizontal active */ +# define TV_XPOS_MASK 0x1fff0000 +# define TV_XPOS_SHIFT 16 +/** Y coordinate of the display from the start of vertical active (NBR) */ +# define TV_YPOS_MASK 0x00000fff +# define TV_YPOS_SHIFT 0 + +#define TV_WIN_SIZE 0x68074 +/** Horizontal size of the display window, measured in pixels*/ +# define TV_XSIZE_MASK 0x1fff0000 +# define TV_XSIZE_SHIFT 16 +/** + * Vertical size of the display window, measured in pixels. + * + * Must be even for interlaced modes. + */ +# define TV_YSIZE_MASK 0x00000fff +# define TV_YSIZE_SHIFT 0 + +#define TV_FILTER_CTL_1 0x68080 +/** + * Enables automatic scaling calculation. + * + * If set, the rest of the registers are ignored, and the calculated values can + * be read back from the register. + */ +# define TV_AUTO_SCALE (1 << 31) +/** + * Disables the vertical filter. + * + * This is required on modes more than 1024 pixels wide */ +# define TV_V_FILTER_BYPASS (1 << 29) +/** Enables adaptive vertical filtering */ +# define TV_VADAPT (1 << 28) +# define TV_VADAPT_MODE_MASK (3 << 26) +/** Selects the least adaptive vertical filtering mode */ +# define TV_VADAPT_MODE_LEAST (0 << 26) +/** Selects the moderately adaptive vertical filtering mode */ +# define TV_VADAPT_MODE_MODERATE (1 << 26) +/** Selects the most adaptive vertical filtering mode */ +# define TV_VADAPT_MODE_MOST (3 << 26) +/** + * Sets the horizontal scaling factor. + * + * This should be the fractional part of the horizontal scaling factor divided + * by the oversampling rate. TV_HSCALE should be less than 1, and set to: + * + * (src width - 1) / ((oversample * dest width) - 1) + */ +# define TV_HSCALE_FRAC_MASK 0x00003fff +# define TV_HSCALE_FRAC_SHIFT 0 + +#define TV_FILTER_CTL_2 0x68084 +/** + * Sets the integer part of the 3.15 fixed-point vertical scaling factor. + * + * TV_VSCALE should be (src height - 1) / ((interlace * dest height) - 1) + */ +# define TV_VSCALE_INT_MASK 0x00038000 +# define TV_VSCALE_INT_SHIFT 15 +/** + * Sets the fractional part of the 3.15 fixed-point vertical scaling factor. + * + * \sa TV_VSCALE_INT_MASK + */ +# define TV_VSCALE_FRAC_MASK 0x00007fff +# define TV_VSCALE_FRAC_SHIFT 0 + +#define TV_FILTER_CTL_3 0x68088 +/** + * Sets the integer part of the 3.15 fixed-point vertical scaling factor. + * + * TV_VSCALE should be (src height - 1) / (1/4 * (dest height - 1)) + * + * For progressive modes, TV_VSCALE_IP_INT should be set to zeroes. + */ +# define TV_VSCALE_IP_INT_MASK 0x00038000 +# define TV_VSCALE_IP_INT_SHIFT 15 +/** + * Sets the fractional part of the 3.15 fixed-point vertical scaling factor. + * + * For progressive modes, TV_VSCALE_IP_INT should be set to zeroes. + * + * \sa TV_VSCALE_IP_INT_MASK + */ +# define TV_VSCALE_IP_FRAC_MASK 0x00007fff +# define TV_VSCALE_IP_FRAC_SHIFT 0 + +#define TV_CC_CONTROL 0x68090 +# define TV_CC_ENABLE (1 << 31) +/** + * Specifies which field to send the CC data in. + * + * CC data is usually sent in field 0. + */ +# define TV_CC_FID_MASK (1 << 27) +# define TV_CC_FID_SHIFT 27 +/** Sets the horizontal position of the CC data. Usually 135. */ +# define TV_CC_HOFF_MASK 0x03ff0000 +# define TV_CC_HOFF_SHIFT 16 +/** Sets the vertical position of the CC data. Usually 21 */ +# define TV_CC_LINE_MASK 0x0000003f +# define TV_CC_LINE_SHIFT 0 + +#define TV_CC_DATA 0x68094 +# define TV_CC_RDY (1 << 31) +/** Second word of CC data to be transmitted. */ +# define TV_CC_DATA_2_MASK 0x007f0000 +# define TV_CC_DATA_2_SHIFT 16 +/** First word of CC data to be transmitted. */ +# define TV_CC_DATA_1_MASK 0x0000007f +# define TV_CC_DATA_1_SHIFT 0 + +#define TV_H_LUMA_0 0x68100 +#define TV_H_LUMA_59 0x681ec +#define TV_H_CHROMA_0 0x68200 +#define TV_H_CHROMA_59 0x682ec +#define TV_V_LUMA_0 0x68300 +#define TV_V_LUMA_42 0x683a8 +#define TV_V_CHROMA_0 0x68400 +#define TV_V_CHROMA_42 0x684a8 + +/* Display Port */ +#define DP_A 0x64000 /* eDP */ +#define DP_B 0x64100 +#define DP_C 0x64200 +#define DP_D 0x64300 + +#define DP_PORT_EN (1 << 31) +#define DP_PIPEB_SELECT (1 << 30) +#define DP_PIPE_MASK (1 << 30) + +#define DP_PIPE_ENABLED(V, P) \ + (((V) & (DP_PIPE_MASK | DP_PORT_EN)) == ((P) << 30 | DP_PORT_EN)) + +/* Link training mode - select a suitable mode for each stage */ +#define DP_LINK_TRAIN_PAT_1 (0 << 28) +#define DP_LINK_TRAIN_PAT_2 (1 << 28) +#define DP_LINK_TRAIN_PAT_IDLE (2 << 28) +#define DP_LINK_TRAIN_OFF (3 << 28) +#define DP_LINK_TRAIN_MASK (3 << 28) +#define DP_LINK_TRAIN_SHIFT 28 + +/* CPT Link training mode */ +#define DP_LINK_TRAIN_PAT_1_CPT (0 << 8) +#define DP_LINK_TRAIN_PAT_2_CPT (1 << 8) +#define DP_LINK_TRAIN_PAT_IDLE_CPT (2 << 8) +#define DP_LINK_TRAIN_OFF_CPT (3 << 8) +#define DP_LINK_TRAIN_MASK_CPT (7 << 8) +#define DP_LINK_TRAIN_SHIFT_CPT 8 + +/* Signal voltages. These are mostly controlled by the other end */ +#define DP_VOLTAGE_0_4 (0 << 25) +#define DP_VOLTAGE_0_6 (1 << 25) +#define DP_VOLTAGE_0_8 (2 << 25) +#define DP_VOLTAGE_1_2 (3 << 25) +#define DP_VOLTAGE_MASK (7 << 25) +#define DP_VOLTAGE_SHIFT 25 + +/* Signal pre-emphasis levels, like voltages, the other end tells us what + * they want + */ +#define DP_PRE_EMPHASIS_0 (0 << 22) +#define DP_PRE_EMPHASIS_3_5 (1 << 22) +#define DP_PRE_EMPHASIS_6 (2 << 22) +#define DP_PRE_EMPHASIS_9_5 (3 << 22) +#define DP_PRE_EMPHASIS_MASK (7 << 22) +#define DP_PRE_EMPHASIS_SHIFT 22 + +/* How many wires to use. I guess 3 was too hard */ +#define DP_PORT_WIDTH_1 (0 << 19) +#define DP_PORT_WIDTH_2 (1 << 19) +#define DP_PORT_WIDTH_4 (3 << 19) +#define DP_PORT_WIDTH_MASK (7 << 19) + +/* Mystic DPCD version 1.1 special mode */ +#define DP_ENHANCED_FRAMING (1 << 18) + +/* eDP */ +#define DP_PLL_FREQ_270MHZ (0 << 16) +#define DP_PLL_FREQ_160MHZ (1 << 16) +#define DP_PLL_FREQ_MASK (3 << 16) + +/** locked once port is enabled */ +#define DP_PORT_REVERSAL (1 << 15) + +/* eDP */ +#define DP_PLL_ENABLE (1 << 14) + +/** sends the clock on lane 15 of the PEG for debug */ +#define DP_CLOCK_OUTPUT_ENABLE (1 << 13) + +#define DP_SCRAMBLING_DISABLE (1 << 12) +#define DP_SCRAMBLING_DISABLE_IRONLAKE (1 << 7) + +/** limit RGB values to avoid confusing TVs */ +#define DP_COLOR_RANGE_16_235 (1 << 8) + +/** Turn on the audio link */ +#define DP_AUDIO_OUTPUT_ENABLE (1 << 6) + +/** vs and hs sync polarity */ +#define DP_SYNC_VS_HIGH (1 << 4) +#define DP_SYNC_HS_HIGH (1 << 3) + +/** A fantasy */ +#define DP_DETECTED (1 << 2) + +/** The aux channel provides a way to talk to the + * signal sink for DDC etc. Max packet size supported + * is 20 bytes in each direction, hence the 5 fixed + * data registers + */ +#define DPA_AUX_CH_CTL 0x64010 +#define DPA_AUX_CH_DATA1 0x64014 +#define DPA_AUX_CH_DATA2 0x64018 +#define DPA_AUX_CH_DATA3 0x6401c +#define DPA_AUX_CH_DATA4 0x64020 +#define DPA_AUX_CH_DATA5 0x64024 + +#define DPB_AUX_CH_CTL 0x64110 +#define DPB_AUX_CH_DATA1 0x64114 +#define DPB_AUX_CH_DATA2 0x64118 +#define DPB_AUX_CH_DATA3 0x6411c +#define DPB_AUX_CH_DATA4 0x64120 +#define DPB_AUX_CH_DATA5 0x64124 + +#define DPC_AUX_CH_CTL 0x64210 +#define DPC_AUX_CH_DATA1 0x64214 +#define DPC_AUX_CH_DATA2 0x64218 +#define DPC_AUX_CH_DATA3 0x6421c +#define DPC_AUX_CH_DATA4 0x64220 +#define DPC_AUX_CH_DATA5 0x64224 + +#define DPD_AUX_CH_CTL 0x64310 +#define DPD_AUX_CH_DATA1 0x64314 +#define DPD_AUX_CH_DATA2 0x64318 +#define DPD_AUX_CH_DATA3 0x6431c +#define DPD_AUX_CH_DATA4 0x64320 +#define DPD_AUX_CH_DATA5 0x64324 + +#define DP_AUX_CH_CTL_SEND_BUSY (1 << 31) +#define DP_AUX_CH_CTL_DONE (1 << 30) +#define DP_AUX_CH_CTL_INTERRUPT (1 << 29) +#define DP_AUX_CH_CTL_TIME_OUT_ERROR (1 << 28) +#define DP_AUX_CH_CTL_TIME_OUT_400us (0 << 26) +#define DP_AUX_CH_CTL_TIME_OUT_600us (1 << 26) +#define DP_AUX_CH_CTL_TIME_OUT_800us (2 << 26) +#define DP_AUX_CH_CTL_TIME_OUT_1600us (3 << 26) +#define DP_AUX_CH_CTL_TIME_OUT_MASK (3 << 26) +#define DP_AUX_CH_CTL_RECEIVE_ERROR (1 << 25) +#define DP_AUX_CH_CTL_MESSAGE_SIZE_MASK (0x1f << 20) +#define DP_AUX_CH_CTL_MESSAGE_SIZE_SHIFT 20 +#define DP_AUX_CH_CTL_PRECHARGE_2US_MASK (0xf << 16) +#define DP_AUX_CH_CTL_PRECHARGE_2US_SHIFT 16 +#define DP_AUX_CH_CTL_AUX_AKSV_SELECT (1 << 15) +#define DP_AUX_CH_CTL_MANCHESTER_TEST (1 << 14) +#define DP_AUX_CH_CTL_SYNC_TEST (1 << 13) +#define DP_AUX_CH_CTL_DEGLITCH_TEST (1 << 12) +#define DP_AUX_CH_CTL_PRECHARGE_TEST (1 << 11) +#define DP_AUX_CH_CTL_BIT_CLOCK_2X_MASK (0x7ff) +#define DP_AUX_CH_CTL_BIT_CLOCK_2X_SHIFT 0 + +/* + * Computing GMCH M and N values for the Display Port link + * + * GMCH M/N = dot clock * bytes per pixel / ls_clk * # of lanes + * + * ls_clk (we assume) is the DP link clock (1.62 or 2.7 GHz) + * + * The GMCH value is used internally + * + * bytes_per_pixel is the number of bytes coming out of the plane, + * which is after the LUTs, so we want the bytes for our color format. + * For our current usage, this is always 3, one byte for R, G and B. + */ +#define _PIPEA_GMCH_DATA_M 0x70050 +#define _PIPEB_GMCH_DATA_M 0x71050 + +/* Transfer unit size for display port - 1, default is 0x3f (for TU size 64) */ +#define PIPE_GMCH_DATA_M_TU_SIZE_MASK (0x3f << 25) +#define PIPE_GMCH_DATA_M_TU_SIZE_SHIFT 25 + +#define PIPE_GMCH_DATA_M_MASK (0xffffff) + +#define _PIPEA_GMCH_DATA_N 0x70054 +#define _PIPEB_GMCH_DATA_N 0x71054 +#define PIPE_GMCH_DATA_N_MASK (0xffffff) + +/* + * Computing Link M and N values for the Display Port link + * + * Link M / N = pixel_clock / ls_clk + * + * (the DP spec calls pixel_clock the 'strm_clk') + * + * The Link value is transmitted in the Main Stream + * Attributes and VB-ID. + */ + +#define _PIPEA_DP_LINK_M 0x70060 +#define _PIPEB_DP_LINK_M 0x71060 +#define PIPEA_DP_LINK_M_MASK (0xffffff) + +#define _PIPEA_DP_LINK_N 0x70064 +#define _PIPEB_DP_LINK_N 0x71064 +#define PIPEA_DP_LINK_N_MASK (0xffffff) + +#define PIPE_GMCH_DATA_M(pipe) _PIPE(pipe, _PIPEA_GMCH_DATA_M, _PIPEB_GMCH_DATA_M) +#define PIPE_GMCH_DATA_N(pipe) _PIPE(pipe, _PIPEA_GMCH_DATA_N, _PIPEB_GMCH_DATA_N) +#define PIPE_DP_LINK_M(pipe) _PIPE(pipe, _PIPEA_DP_LINK_M, _PIPEB_DP_LINK_M) +#define PIPE_DP_LINK_N(pipe) _PIPE(pipe, _PIPEA_DP_LINK_N, _PIPEB_DP_LINK_N) + +/* Display & cursor control */ + +/* Pipe A */ +#define _PIPEADSL 0x70000 +#define DSL_LINEMASK 0x00000fff +#define _PIPEACONF 0x70008 +#define PIPECONF_ENABLE (1<<31) +#define PIPECONF_DISABLE 0 +#define PIPECONF_DOUBLE_WIDE (1<<30) +#define I965_PIPECONF_ACTIVE (1<<30) +#define PIPECONF_FRAME_START_DELAY_MASK (3<<27) +#define PIPECONF_SINGLE_WIDE 0 +#define PIPECONF_PIPE_UNLOCKED 0 +#define PIPECONF_PIPE_LOCKED (1<<25) +#define PIPECONF_PALETTE 0 +#define PIPECONF_GAMMA (1<<24) +#define PIPECONF_FORCE_BORDER (1<<25) +#define PIPECONF_PROGRESSIVE (0 << 21) +#define PIPECONF_INTERLACE_W_FIELD_INDICATION (6 << 21) +#define PIPECONF_INTERLACE_FIELD_0_ONLY (7 << 21) +#define PIPECONF_CXSR_DOWNCLOCK (1<<16) +#define PIPECONF_BPP_MASK (0x000000e0) +#define PIPECONF_BPP_8 (0<<5) +#define PIPECONF_BPP_10 (1<<5) +#define PIPECONF_BPP_6 (2<<5) +#define PIPECONF_BPP_12 (3<<5) +#define PIPECONF_DITHER_EN (1<<4) +#define PIPECONF_DITHER_TYPE_MASK (0x0000000c) +#define PIPECONF_DITHER_TYPE_SP (0<<2) +#define PIPECONF_DITHER_TYPE_ST1 (1<<2) +#define PIPECONF_DITHER_TYPE_ST2 (2<<2) +#define PIPECONF_DITHER_TYPE_TEMP (3<<2) +#define _PIPEASTAT 0x70024 +#define PIPE_FIFO_UNDERRUN_STATUS (1UL<<31) +#define PIPE_CRC_ERROR_ENABLE (1UL<<29) +#define PIPE_CRC_DONE_ENABLE (1UL<<28) +#define PIPE_GMBUS_EVENT_ENABLE (1UL<<27) +#define PIPE_HOTPLUG_INTERRUPT_ENABLE (1UL<<26) +#define PIPE_VSYNC_INTERRUPT_ENABLE (1UL<<25) +#define PIPE_DISPLAY_LINE_COMPARE_ENABLE (1UL<<24) +#define PIPE_DPST_EVENT_ENABLE (1UL<<23) +#define PIPE_LEGACY_BLC_EVENT_ENABLE (1UL<<22) +#define PIPE_ODD_FIELD_INTERRUPT_ENABLE (1UL<<21) +#define PIPE_EVEN_FIELD_INTERRUPT_ENABLE (1UL<<20) +#define PIPE_HOTPLUG_TV_INTERRUPT_ENABLE (1UL<<18) /* pre-965 */ +#define PIPE_START_VBLANK_INTERRUPT_ENABLE (1UL<<18) /* 965 or later */ +#define PIPE_VBLANK_INTERRUPT_ENABLE (1UL<<17) +#define PIPE_OVERLAY_UPDATED_ENABLE (1UL<<16) +#define PIPE_CRC_ERROR_INTERRUPT_STATUS (1UL<<13) +#define PIPE_CRC_DONE_INTERRUPT_STATUS (1UL<<12) +#define PIPE_GMBUS_INTERRUPT_STATUS (1UL<<11) +#define PIPE_HOTPLUG_INTERRUPT_STATUS (1UL<<10) +#define PIPE_VSYNC_INTERRUPT_STATUS (1UL<<9) +#define PIPE_DISPLAY_LINE_COMPARE_STATUS (1UL<<8) +#define PIPE_DPST_EVENT_STATUS (1UL<<7) +#define PIPE_LEGACY_BLC_EVENT_STATUS (1UL<<6) +#define PIPE_ODD_FIELD_INTERRUPT_STATUS (1UL<<5) +#define PIPE_EVEN_FIELD_INTERRUPT_STATUS (1UL<<4) +#define PIPE_HOTPLUG_TV_INTERRUPT_STATUS (1UL<<2) /* pre-965 */ +#define PIPE_START_VBLANK_INTERRUPT_STATUS (1UL<<2) /* 965 or later */ +#define PIPE_VBLANK_INTERRUPT_STATUS (1UL<<1) +#define PIPE_OVERLAY_UPDATED_STATUS (1UL<<0) +#define PIPE_BPC_MASK (7 << 5) /* Ironlake */ +#define PIPE_8BPC (0 << 5) +#define PIPE_10BPC (1 << 5) +#define PIPE_6BPC (2 << 5) +#define PIPE_12BPC (3 << 5) + +#define PIPESRC(pipe) _PIPE(pipe, _PIPEASRC, _PIPEBSRC) +#define PIPECONF(pipe) _PIPE(pipe, _PIPEACONF, _PIPEBCONF) +#define PIPEDSL(pipe) _PIPE(pipe, _PIPEADSL, _PIPEBDSL) +#define PIPEFRAME(pipe) _PIPE(pipe, _PIPEAFRAMEHIGH, _PIPEBFRAMEHIGH) +#define PIPEFRAMEPIXEL(pipe) _PIPE(pipe, _PIPEAFRAMEPIXEL, _PIPEBFRAMEPIXEL) +#define PIPESTAT(pipe) _PIPE(pipe, _PIPEASTAT, _PIPEBSTAT) + +#define DSPARB 0x70030 +#define DSPARB_CSTART_MASK (0x7f << 7) +#define DSPARB_CSTART_SHIFT 7 +#define DSPARB_BSTART_MASK (0x7f) +#define DSPARB_BSTART_SHIFT 0 +#define DSPARB_BEND_SHIFT 9 /* on 855 */ +#define DSPARB_AEND_SHIFT 0 + +#define DSPFW1 0x70034 +#define DSPFW_SR_SHIFT 23 +#define DSPFW_SR_MASK (0x1ff<<23) +#define DSPFW_CURSORB_SHIFT 16 +#define DSPFW_CURSORB_MASK (0x3f<<16) +#define DSPFW_PLANEB_SHIFT 8 +#define DSPFW_PLANEB_MASK (0x7f<<8) +#define DSPFW_PLANEA_MASK (0x7f) +#define DSPFW2 0x70038 +#define DSPFW_CURSORA_MASK 0x00003f00 +#define DSPFW_CURSORA_SHIFT 8 +#define DSPFW_PLANEC_MASK (0x7f) +#define DSPFW3 0x7003c +#define DSPFW_HPLL_SR_EN (1<<31) +#define DSPFW_CURSOR_SR_SHIFT 24 +#define PINEVIEW_SELF_REFRESH_EN (1<<30) +#define DSPFW_CURSOR_SR_MASK (0x3f<<24) +#define DSPFW_HPLL_CURSOR_SHIFT 16 +#define DSPFW_HPLL_CURSOR_MASK (0x3f<<16) +#define DSPFW_HPLL_SR_MASK (0x1ff) + +/* FIFO watermark sizes etc */ +#define G4X_FIFO_LINE_SIZE 64 +#define I915_FIFO_LINE_SIZE 64 +#define I830_FIFO_LINE_SIZE 32 + +#define G4X_FIFO_SIZE 127 +#define I965_FIFO_SIZE 512 +#define I945_FIFO_SIZE 127 +#define I915_FIFO_SIZE 95 +#define I855GM_FIFO_SIZE 127 /* In cachelines */ +#define I830_FIFO_SIZE 95 + +#define G4X_MAX_WM 0x3f +#define I915_MAX_WM 0x3f + +#define PINEVIEW_DISPLAY_FIFO 512 /* in 64byte unit */ +#define PINEVIEW_FIFO_LINE_SIZE 64 +#define PINEVIEW_MAX_WM 0x1ff +#define PINEVIEW_DFT_WM 0x3f +#define PINEVIEW_DFT_HPLLOFF_WM 0 +#define PINEVIEW_GUARD_WM 10 +#define PINEVIEW_CURSOR_FIFO 64 +#define PINEVIEW_CURSOR_MAX_WM 0x3f +#define PINEVIEW_CURSOR_DFT_WM 0 +#define PINEVIEW_CURSOR_GUARD_WM 5 + +#define I965_CURSOR_FIFO 64 +#define I965_CURSOR_MAX_WM 32 +#define I965_CURSOR_DFT_WM 8 + +/* define the Watermark register on Ironlake */ +#define WM0_PIPEA_ILK 0x45100 +#define WM0_PIPE_PLANE_MASK (0x7f<<16) +#define WM0_PIPE_PLANE_SHIFT 16 +#define WM0_PIPE_SPRITE_MASK (0x3f<<8) +#define WM0_PIPE_SPRITE_SHIFT 8 +#define WM0_PIPE_CURSOR_MASK (0x1f) + +#define WM0_PIPEB_ILK 0x45104 +#define WM1_LP_ILK 0x45108 +#define WM1_LP_SR_EN (1<<31) +#define WM1_LP_LATENCY_SHIFT 24 +#define WM1_LP_LATENCY_MASK (0x7f<<24) +#define WM1_LP_FBC_MASK (0xf<<20) +#define WM1_LP_FBC_SHIFT 20 +#define WM1_LP_SR_MASK (0x1ff<<8) +#define WM1_LP_SR_SHIFT 8 +#define WM1_LP_CURSOR_MASK (0x3f) +#define WM2_LP_ILK 0x4510c +#define WM2_LP_EN (1<<31) +#define WM3_LP_ILK 0x45110 +#define WM3_LP_EN (1<<31) +#define WM1S_LP_ILK 0x45120 +#define WM1S_LP_EN (1<<31) + +/* Memory latency timer register */ +#define MLTR_ILK 0x11222 +#define MLTR_WM1_SHIFT 0 +#define MLTR_WM2_SHIFT 8 +/* the unit of memory self-refresh latency time is 0.5us */ +#define ILK_SRLT_MASK 0x3f +#define ILK_LATENCY(shift) (I915_READ(MLTR_ILK) >> (shift) & ILK_SRLT_MASK) +#define ILK_READ_WM1_LATENCY() ILK_LATENCY(MLTR_WM1_SHIFT) +#define ILK_READ_WM2_LATENCY() ILK_LATENCY(MLTR_WM2_SHIFT) + +/* define the fifo size on Ironlake */ +#define ILK_DISPLAY_FIFO 128 +#define ILK_DISPLAY_MAXWM 64 +#define ILK_DISPLAY_DFTWM 8 +#define ILK_CURSOR_FIFO 32 +#define ILK_CURSOR_MAXWM 16 +#define ILK_CURSOR_DFTWM 8 + +#define ILK_DISPLAY_SR_FIFO 512 +#define ILK_DISPLAY_MAX_SRWM 0x1ff +#define ILK_DISPLAY_DFT_SRWM 0x3f +#define ILK_CURSOR_SR_FIFO 64 +#define ILK_CURSOR_MAX_SRWM 0x3f +#define ILK_CURSOR_DFT_SRWM 8 + +#define ILK_FIFO_LINE_SIZE 64 + +/* define the WM info on Sandybridge */ +#define SNB_DISPLAY_FIFO 128 +#define SNB_DISPLAY_MAXWM 0x7f /* bit 16:22 */ +#define SNB_DISPLAY_DFTWM 8 +#define SNB_CURSOR_FIFO 32 +#define SNB_CURSOR_MAXWM 0x1f /* bit 4:0 */ +#define SNB_CURSOR_DFTWM 8 + +#define SNB_DISPLAY_SR_FIFO 512 +#define SNB_DISPLAY_MAX_SRWM 0x1ff /* bit 16:8 */ +#define SNB_DISPLAY_DFT_SRWM 0x3f +#define SNB_CURSOR_SR_FIFO 64 +#define SNB_CURSOR_MAX_SRWM 0x3f /* bit 5:0 */ +#define SNB_CURSOR_DFT_SRWM 8 + +#define SNB_FBC_MAX_SRWM 0xf /* bit 23:20 */ + +#define SNB_FIFO_LINE_SIZE 64 + + +/* the address where we get all kinds of latency value */ +#define SSKPD 0x5d10 +#define SSKPD_WM_MASK 0x3f +#define SSKPD_WM0_SHIFT 0 +#define SSKPD_WM1_SHIFT 8 +#define SSKPD_WM2_SHIFT 16 +#define SSKPD_WM3_SHIFT 24 + +#define SNB_LATENCY(shift) (I915_READ(MCHBAR_MIRROR_BASE_SNB + SSKPD) >> (shift) & SSKPD_WM_MASK) +#define SNB_READ_WM0_LATENCY() SNB_LATENCY(SSKPD_WM0_SHIFT) +#define SNB_READ_WM1_LATENCY() SNB_LATENCY(SSKPD_WM1_SHIFT) +#define SNB_READ_WM2_LATENCY() SNB_LATENCY(SSKPD_WM2_SHIFT) +#define SNB_READ_WM3_LATENCY() SNB_LATENCY(SSKPD_WM3_SHIFT) + +/* + * The two pipe frame counter registers are not synchronized, so + * reading a stable value is somewhat tricky. The following code + * should work: + * + * do { + * high1 = ((INREG(PIPEAFRAMEHIGH) & PIPE_FRAME_HIGH_MASK) >> + * PIPE_FRAME_HIGH_SHIFT; + * low1 = ((INREG(PIPEAFRAMEPIXEL) & PIPE_FRAME_LOW_MASK) >> + * PIPE_FRAME_LOW_SHIFT); + * high2 = ((INREG(PIPEAFRAMEHIGH) & PIPE_FRAME_HIGH_MASK) >> + * PIPE_FRAME_HIGH_SHIFT); + * } while (high1 != high2); + * frame = (high1 << 8) | low1; + */ +#define _PIPEAFRAMEHIGH 0x70040 +#define PIPE_FRAME_HIGH_MASK 0x0000ffff +#define PIPE_FRAME_HIGH_SHIFT 0 +#define _PIPEAFRAMEPIXEL 0x70044 +#define PIPE_FRAME_LOW_MASK 0xff000000 +#define PIPE_FRAME_LOW_SHIFT 24 +#define PIPE_PIXEL_MASK 0x00ffffff +#define PIPE_PIXEL_SHIFT 0 +/* GM45+ just has to be different */ +#define _PIPEA_FRMCOUNT_GM45 0x70040 +#define _PIPEA_FLIPCOUNT_GM45 0x70044 +#define PIPE_FRMCOUNT_GM45(pipe) _PIPE(pipe, _PIPEA_FRMCOUNT_GM45, _PIPEB_FRMCOUNT_GM45) + +/* Cursor A & B regs */ +#define _CURACNTR 0x70080 +/* Old style CUR*CNTR flags (desktop 8xx) */ +#define CURSOR_ENABLE 0x80000000 +#define CURSOR_GAMMA_ENABLE 0x40000000 +#define CURSOR_STRIDE_MASK 0x30000000 +#define CURSOR_FORMAT_SHIFT 24 +#define CURSOR_FORMAT_MASK (0x07 << CURSOR_FORMAT_SHIFT) +#define CURSOR_FORMAT_2C (0x00 << CURSOR_FORMAT_SHIFT) +#define CURSOR_FORMAT_3C (0x01 << CURSOR_FORMAT_SHIFT) +#define CURSOR_FORMAT_4C (0x02 << CURSOR_FORMAT_SHIFT) +#define CURSOR_FORMAT_ARGB (0x04 << CURSOR_FORMAT_SHIFT) +#define CURSOR_FORMAT_XRGB (0x05 << CURSOR_FORMAT_SHIFT) +/* New style CUR*CNTR flags */ +#define CURSOR_MODE 0x27 +#define CURSOR_MODE_DISABLE 0x00 +#define CURSOR_MODE_64_32B_AX 0x07 +#define CURSOR_MODE_64_ARGB_AX ((1 << 5) | CURSOR_MODE_64_32B_AX) +#define MCURSOR_PIPE_SELECT (1 << 28) +#define MCURSOR_PIPE_A 0x00 +#define MCURSOR_PIPE_B (1 << 28) +#define MCURSOR_GAMMA_ENABLE (1 << 26) +#define _CURABASE 0x70084 +#define _CURAPOS 0x70088 +#define CURSOR_POS_MASK 0x007FF +#define CURSOR_POS_SIGN 0x8000 +#define CURSOR_X_SHIFT 0 +#define CURSOR_Y_SHIFT 16 +#define CURSIZE 0x700a0 +#define _CURBCNTR 0x700c0 +#define _CURBBASE 0x700c4 +#define _CURBPOS 0x700c8 + +#define _CURBCNTR_IVB 0x71080 +#define _CURBBASE_IVB 0x71084 +#define _CURBPOS_IVB 0x71088 + +#define CURCNTR(pipe) _PIPE(pipe, _CURACNTR, _CURBCNTR) +#define CURBASE(pipe) _PIPE(pipe, _CURABASE, _CURBBASE) +#define CURPOS(pipe) _PIPE(pipe, _CURAPOS, _CURBPOS) + +#define CURCNTR_IVB(pipe) _PIPE(pipe, _CURACNTR, _CURBCNTR_IVB) +#define CURBASE_IVB(pipe) _PIPE(pipe, _CURABASE, _CURBBASE_IVB) +#define CURPOS_IVB(pipe) _PIPE(pipe, _CURAPOS, _CURBPOS_IVB) + +/* Display A control */ +#define _DSPACNTR 0x70180 +#define DISPLAY_PLANE_ENABLE (1<<31) +#define DISPLAY_PLANE_DISABLE 0 +#define DISPPLANE_GAMMA_ENABLE (1<<30) +#define DISPPLANE_GAMMA_DISABLE 0 +#define DISPPLANE_PIXFORMAT_MASK (0xf<<26) +#define DISPPLANE_8BPP (0x2<<26) +#define DISPPLANE_15_16BPP (0x4<<26) +#define DISPPLANE_16BPP (0x5<<26) +#define DISPPLANE_32BPP_NO_ALPHA (0x6<<26) +#define DISPPLANE_32BPP (0x7<<26) +#define DISPPLANE_32BPP_30BIT_NO_ALPHA (0xa<<26) +#define DISPPLANE_STEREO_ENABLE (1<<25) +#define DISPPLANE_STEREO_DISABLE 0 +#define DISPPLANE_SEL_PIPE_SHIFT 24 +#define DISPPLANE_SEL_PIPE_MASK (3< + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + */ + +#include "drmP.h" +#include "drm.h" +#include "i915_drm.h" +#include "intel_drv.h" + +static bool i915_pipe_enabled(struct drm_device *dev, enum pipe pipe) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u32 dpll_reg; + + /* On IVB, 3rd pipe shares PLL with another one */ + if (pipe > 1) + return false; + + if (HAS_PCH_SPLIT(dev)) + dpll_reg = (pipe == PIPE_A) ? _PCH_DPLL_A : _PCH_DPLL_B; + else + dpll_reg = (pipe == PIPE_A) ? _DPLL_A : _DPLL_B; + + return (I915_READ(dpll_reg) & DPLL_VCO_ENABLE); +} + +static void i915_save_palette(struct drm_device *dev, enum pipe pipe) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + unsigned long reg = (pipe == PIPE_A ? _PALETTE_A : _PALETTE_B); + u32 *array; + int i; + + if (!i915_pipe_enabled(dev, pipe)) + return; + + if (HAS_PCH_SPLIT(dev)) + reg = (pipe == PIPE_A) ? _LGC_PALETTE_A : _LGC_PALETTE_B; + + if (pipe == PIPE_A) + array = dev_priv->save_palette_a; + else + array = dev_priv->save_palette_b; + + for(i = 0; i < 256; i++) + array[i] = I915_READ(reg + (i << 2)); +} + +static void i915_restore_palette(struct drm_device *dev, enum pipe pipe) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + unsigned long reg = (pipe == PIPE_A ? _PALETTE_A : _PALETTE_B); + u32 *array; + int i; + + if (!i915_pipe_enabled(dev, pipe)) + return; + + if (HAS_PCH_SPLIT(dev)) + reg = (pipe == PIPE_A) ? _LGC_PALETTE_A : _LGC_PALETTE_B; + + if (pipe == PIPE_A) + array = dev_priv->save_palette_a; + else + array = dev_priv->save_palette_b; + + for(i = 0; i < 256; i++) + I915_WRITE(reg + (i << 2), array[i]); +} + +static u8 i915_read_indexed(struct drm_device *dev, u16 index_port, u16 data_port, u8 reg) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + I915_WRITE8(index_port, reg); + return I915_READ8(data_port); +} + +static u8 i915_read_ar(struct drm_device *dev, u16 st01, u8 reg, u16 palette_enable) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + I915_READ8(st01); + I915_WRITE8(VGA_AR_INDEX, palette_enable | reg); + return I915_READ8(VGA_AR_DATA_READ); +} + +static void i915_write_ar(struct drm_device *dev, u16 st01, u8 reg, u8 val, u16 palette_enable) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + I915_READ8(st01); + I915_WRITE8(VGA_AR_INDEX, palette_enable | reg); + I915_WRITE8(VGA_AR_DATA_WRITE, val); +} + +static void i915_write_indexed(struct drm_device *dev, u16 index_port, u16 data_port, u8 reg, u8 val) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + I915_WRITE8(index_port, reg); + I915_WRITE8(data_port, val); +} + +static void i915_save_vga(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + int i; + u16 cr_index, cr_data, st01; + + /* VGA color palette registers */ + dev_priv->saveDACMASK = I915_READ8(VGA_DACMASK); + + /* MSR bits */ + dev_priv->saveMSR = I915_READ8(VGA_MSR_READ); + if (dev_priv->saveMSR & VGA_MSR_CGA_MODE) { + cr_index = VGA_CR_INDEX_CGA; + cr_data = VGA_CR_DATA_CGA; + st01 = VGA_ST01_CGA; + } else { + cr_index = VGA_CR_INDEX_MDA; + cr_data = VGA_CR_DATA_MDA; + st01 = VGA_ST01_MDA; + } + + /* CRT controller regs */ + i915_write_indexed(dev, cr_index, cr_data, 0x11, + i915_read_indexed(dev, cr_index, cr_data, 0x11) & + (~0x80)); + for (i = 0; i <= 0x24; i++) + dev_priv->saveCR[i] = + i915_read_indexed(dev, cr_index, cr_data, i); + /* Make sure we don't turn off CR group 0 writes */ + dev_priv->saveCR[0x11] &= ~0x80; + + /* Attribute controller registers */ + I915_READ8(st01); + dev_priv->saveAR_INDEX = I915_READ8(VGA_AR_INDEX); + for (i = 0; i <= 0x14; i++) + dev_priv->saveAR[i] = i915_read_ar(dev, st01, i, 0); + I915_READ8(st01); + I915_WRITE8(VGA_AR_INDEX, dev_priv->saveAR_INDEX); + I915_READ8(st01); + + /* Graphics controller registers */ + for (i = 0; i < 9; i++) + dev_priv->saveGR[i] = + i915_read_indexed(dev, VGA_GR_INDEX, VGA_GR_DATA, i); + + dev_priv->saveGR[0x10] = + i915_read_indexed(dev, VGA_GR_INDEX, VGA_GR_DATA, 0x10); + dev_priv->saveGR[0x11] = + i915_read_indexed(dev, VGA_GR_INDEX, VGA_GR_DATA, 0x11); + dev_priv->saveGR[0x18] = + i915_read_indexed(dev, VGA_GR_INDEX, VGA_GR_DATA, 0x18); + + /* Sequencer registers */ + for (i = 0; i < 8; i++) + dev_priv->saveSR[i] = + i915_read_indexed(dev, VGA_SR_INDEX, VGA_SR_DATA, i); +} + +static void i915_restore_vga(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + int i; + u16 cr_index, cr_data, st01; + + /* MSR bits */ + I915_WRITE8(VGA_MSR_WRITE, dev_priv->saveMSR); + if (dev_priv->saveMSR & VGA_MSR_CGA_MODE) { + cr_index = VGA_CR_INDEX_CGA; + cr_data = VGA_CR_DATA_CGA; + st01 = VGA_ST01_CGA; + } else { + cr_index = VGA_CR_INDEX_MDA; + cr_data = VGA_CR_DATA_MDA; + st01 = VGA_ST01_MDA; + } + + /* Sequencer registers, don't write SR07 */ + for (i = 0; i < 7; i++) + i915_write_indexed(dev, VGA_SR_INDEX, VGA_SR_DATA, i, + dev_priv->saveSR[i]); + + /* CRT controller regs */ + /* Enable CR group 0 writes */ + i915_write_indexed(dev, cr_index, cr_data, 0x11, dev_priv->saveCR[0x11]); + for (i = 0; i <= 0x24; i++) + i915_write_indexed(dev, cr_index, cr_data, i, dev_priv->saveCR[i]); + + /* Graphics controller regs */ + for (i = 0; i < 9; i++) + i915_write_indexed(dev, VGA_GR_INDEX, VGA_GR_DATA, i, + dev_priv->saveGR[i]); + + i915_write_indexed(dev, VGA_GR_INDEX, VGA_GR_DATA, 0x10, + dev_priv->saveGR[0x10]); + i915_write_indexed(dev, VGA_GR_INDEX, VGA_GR_DATA, 0x11, + dev_priv->saveGR[0x11]); + i915_write_indexed(dev, VGA_GR_INDEX, VGA_GR_DATA, 0x18, + dev_priv->saveGR[0x18]); + + /* Attribute controller registers */ + I915_READ8(st01); /* switch back to index mode */ + for (i = 0; i <= 0x14; i++) + i915_write_ar(dev, st01, i, dev_priv->saveAR[i], 0); + I915_READ8(st01); /* switch back to index mode */ + I915_WRITE8(VGA_AR_INDEX, dev_priv->saveAR_INDEX | 0x20); + I915_READ8(st01); + + /* VGA color palette registers */ + I915_WRITE8(VGA_DACMASK, dev_priv->saveDACMASK); +} + +static void i915_save_modeset_reg(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + int i; + + if (drm_core_check_feature(dev, DRIVER_MODESET)) + return; + + /* Cursor state */ + dev_priv->saveCURACNTR = I915_READ(_CURACNTR); + dev_priv->saveCURAPOS = I915_READ(_CURAPOS); + dev_priv->saveCURABASE = I915_READ(_CURABASE); + dev_priv->saveCURBCNTR = I915_READ(_CURBCNTR); + dev_priv->saveCURBPOS = I915_READ(_CURBPOS); + dev_priv->saveCURBBASE = I915_READ(_CURBBASE); + if (IS_GEN2(dev)) + dev_priv->saveCURSIZE = I915_READ(CURSIZE); + + if (HAS_PCH_SPLIT(dev)) { + dev_priv->savePCH_DREF_CONTROL = I915_READ(PCH_DREF_CONTROL); + dev_priv->saveDISP_ARB_CTL = I915_READ(DISP_ARB_CTL); + } + + /* Pipe & plane A info */ + dev_priv->savePIPEACONF = I915_READ(_PIPEACONF); + dev_priv->savePIPEASRC = I915_READ(_PIPEASRC); + if (HAS_PCH_SPLIT(dev)) { + dev_priv->saveFPA0 = I915_READ(_PCH_FPA0); + dev_priv->saveFPA1 = I915_READ(_PCH_FPA1); + dev_priv->saveDPLL_A = I915_READ(_PCH_DPLL_A); + } else { + dev_priv->saveFPA0 = I915_READ(_FPA0); + dev_priv->saveFPA1 = I915_READ(_FPA1); + dev_priv->saveDPLL_A = I915_READ(_DPLL_A); + } + if (INTEL_INFO(dev)->gen >= 4 && !HAS_PCH_SPLIT(dev)) + dev_priv->saveDPLL_A_MD = I915_READ(_DPLL_A_MD); + dev_priv->saveHTOTAL_A = I915_READ(_HTOTAL_A); + dev_priv->saveHBLANK_A = I915_READ(_HBLANK_A); + dev_priv->saveHSYNC_A = I915_READ(_HSYNC_A); + dev_priv->saveVTOTAL_A = I915_READ(_VTOTAL_A); + dev_priv->saveVBLANK_A = I915_READ(_VBLANK_A); + dev_priv->saveVSYNC_A = I915_READ(_VSYNC_A); + if (!HAS_PCH_SPLIT(dev)) + dev_priv->saveBCLRPAT_A = I915_READ(_BCLRPAT_A); + + if (HAS_PCH_SPLIT(dev)) { + dev_priv->savePIPEA_DATA_M1 = I915_READ(_PIPEA_DATA_M1); + dev_priv->savePIPEA_DATA_N1 = I915_READ(_PIPEA_DATA_N1); + dev_priv->savePIPEA_LINK_M1 = I915_READ(_PIPEA_LINK_M1); + dev_priv->savePIPEA_LINK_N1 = I915_READ(_PIPEA_LINK_N1); + + dev_priv->saveFDI_TXA_CTL = I915_READ(_FDI_TXA_CTL); + dev_priv->saveFDI_RXA_CTL = I915_READ(_FDI_RXA_CTL); + + dev_priv->savePFA_CTL_1 = I915_READ(_PFA_CTL_1); + dev_priv->savePFA_WIN_SZ = I915_READ(_PFA_WIN_SZ); + dev_priv->savePFA_WIN_POS = I915_READ(_PFA_WIN_POS); + + dev_priv->saveTRANSACONF = I915_READ(_TRANSACONF); + dev_priv->saveTRANS_HTOTAL_A = I915_READ(_TRANS_HTOTAL_A); + dev_priv->saveTRANS_HBLANK_A = I915_READ(_TRANS_HBLANK_A); + dev_priv->saveTRANS_HSYNC_A = I915_READ(_TRANS_HSYNC_A); + dev_priv->saveTRANS_VTOTAL_A = I915_READ(_TRANS_VTOTAL_A); + dev_priv->saveTRANS_VBLANK_A = I915_READ(_TRANS_VBLANK_A); + dev_priv->saveTRANS_VSYNC_A = I915_READ(_TRANS_VSYNC_A); + } + + dev_priv->saveDSPACNTR = I915_READ(_DSPACNTR); + dev_priv->saveDSPASTRIDE = I915_READ(_DSPASTRIDE); + dev_priv->saveDSPASIZE = I915_READ(_DSPASIZE); + dev_priv->saveDSPAPOS = I915_READ(_DSPAPOS); + dev_priv->saveDSPAADDR = I915_READ(_DSPAADDR); + if (INTEL_INFO(dev)->gen >= 4) { + dev_priv->saveDSPASURF = I915_READ(_DSPASURF); + dev_priv->saveDSPATILEOFF = I915_READ(_DSPATILEOFF); + } + i915_save_palette(dev, PIPE_A); + dev_priv->savePIPEASTAT = I915_READ(_PIPEASTAT); + + /* Pipe & plane B info */ + dev_priv->savePIPEBCONF = I915_READ(_PIPEBCONF); + dev_priv->savePIPEBSRC = I915_READ(_PIPEBSRC); + if (HAS_PCH_SPLIT(dev)) { + dev_priv->saveFPB0 = I915_READ(_PCH_FPB0); + dev_priv->saveFPB1 = I915_READ(_PCH_FPB1); + dev_priv->saveDPLL_B = I915_READ(_PCH_DPLL_B); + } else { + dev_priv->saveFPB0 = I915_READ(_FPB0); + dev_priv->saveFPB1 = I915_READ(_FPB1); + dev_priv->saveDPLL_B = I915_READ(_DPLL_B); + } + if (INTEL_INFO(dev)->gen >= 4 && !HAS_PCH_SPLIT(dev)) + dev_priv->saveDPLL_B_MD = I915_READ(_DPLL_B_MD); + dev_priv->saveHTOTAL_B = I915_READ(_HTOTAL_B); + dev_priv->saveHBLANK_B = I915_READ(_HBLANK_B); + dev_priv->saveHSYNC_B = I915_READ(_HSYNC_B); + dev_priv->saveVTOTAL_B = I915_READ(_VTOTAL_B); + dev_priv->saveVBLANK_B = I915_READ(_VBLANK_B); + dev_priv->saveVSYNC_B = I915_READ(_VSYNC_B); + if (!HAS_PCH_SPLIT(dev)) + dev_priv->saveBCLRPAT_B = I915_READ(_BCLRPAT_B); + + if (HAS_PCH_SPLIT(dev)) { + dev_priv->savePIPEB_DATA_M1 = I915_READ(_PIPEB_DATA_M1); + dev_priv->savePIPEB_DATA_N1 = I915_READ(_PIPEB_DATA_N1); + dev_priv->savePIPEB_LINK_M1 = I915_READ(_PIPEB_LINK_M1); + dev_priv->savePIPEB_LINK_N1 = I915_READ(_PIPEB_LINK_N1); + + dev_priv->saveFDI_TXB_CTL = I915_READ(_FDI_TXB_CTL); + dev_priv->saveFDI_RXB_CTL = I915_READ(_FDI_RXB_CTL); + + dev_priv->savePFB_CTL_1 = I915_READ(_PFB_CTL_1); + dev_priv->savePFB_WIN_SZ = I915_READ(_PFB_WIN_SZ); + dev_priv->savePFB_WIN_POS = I915_READ(_PFB_WIN_POS); + + dev_priv->saveTRANSBCONF = I915_READ(_TRANSBCONF); + dev_priv->saveTRANS_HTOTAL_B = I915_READ(_TRANS_HTOTAL_B); + dev_priv->saveTRANS_HBLANK_B = I915_READ(_TRANS_HBLANK_B); + dev_priv->saveTRANS_HSYNC_B = I915_READ(_TRANS_HSYNC_B); + dev_priv->saveTRANS_VTOTAL_B = I915_READ(_TRANS_VTOTAL_B); + dev_priv->saveTRANS_VBLANK_B = I915_READ(_TRANS_VBLANK_B); + dev_priv->saveTRANS_VSYNC_B = I915_READ(_TRANS_VSYNC_B); + } + + dev_priv->saveDSPBCNTR = I915_READ(_DSPBCNTR); + dev_priv->saveDSPBSTRIDE = I915_READ(_DSPBSTRIDE); + dev_priv->saveDSPBSIZE = I915_READ(_DSPBSIZE); + dev_priv->saveDSPBPOS = I915_READ(_DSPBPOS); + dev_priv->saveDSPBADDR = I915_READ(_DSPBADDR); + if (INTEL_INFO(dev)->gen >= 4) { + dev_priv->saveDSPBSURF = I915_READ(_DSPBSURF); + dev_priv->saveDSPBTILEOFF = I915_READ(_DSPBTILEOFF); + } + i915_save_palette(dev, PIPE_B); + dev_priv->savePIPEBSTAT = I915_READ(_PIPEBSTAT); + + /* Fences */ + switch (INTEL_INFO(dev)->gen) { + case 7: + case 6: + for (i = 0; i < 16; i++) + dev_priv->saveFENCE[i] = I915_READ64(FENCE_REG_SANDYBRIDGE_0 + (i * 8)); + break; + case 5: + case 4: + for (i = 0; i < 16; i++) + dev_priv->saveFENCE[i] = I915_READ64(FENCE_REG_965_0 + (i * 8)); + break; + case 3: + if (IS_I945G(dev) || IS_I945GM(dev) || IS_G33(dev)) + for (i = 0; i < 8; i++) + dev_priv->saveFENCE[i+8] = I915_READ(FENCE_REG_945_8 + (i * 4)); + case 2: + for (i = 0; i < 8; i++) + dev_priv->saveFENCE[i] = I915_READ(FENCE_REG_830_0 + (i * 4)); + break; + } + + return; +} + +static void i915_restore_modeset_reg(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + int dpll_a_reg, fpa0_reg, fpa1_reg; + int dpll_b_reg, fpb0_reg, fpb1_reg; + int i; + + if (drm_core_check_feature(dev, DRIVER_MODESET)) + return; + + /* Fences */ + switch (INTEL_INFO(dev)->gen) { + case 7: + case 6: + for (i = 0; i < 16; i++) + I915_WRITE64(FENCE_REG_SANDYBRIDGE_0 + (i * 8), dev_priv->saveFENCE[i]); + break; + case 5: + case 4: + for (i = 0; i < 16; i++) + I915_WRITE64(FENCE_REG_965_0 + (i * 8), dev_priv->saveFENCE[i]); + break; + case 3: + case 2: + if (IS_I945G(dev) || IS_I945GM(dev) || IS_G33(dev)) + for (i = 0; i < 8; i++) + I915_WRITE(FENCE_REG_945_8 + (i * 4), dev_priv->saveFENCE[i+8]); + for (i = 0; i < 8; i++) + I915_WRITE(FENCE_REG_830_0 + (i * 4), dev_priv->saveFENCE[i]); + break; + } + + + if (HAS_PCH_SPLIT(dev)) { + dpll_a_reg = _PCH_DPLL_A; + dpll_b_reg = _PCH_DPLL_B; + fpa0_reg = _PCH_FPA0; + fpb0_reg = _PCH_FPB0; + fpa1_reg = _PCH_FPA1; + fpb1_reg = _PCH_FPB1; + } else { + dpll_a_reg = _DPLL_A; + dpll_b_reg = _DPLL_B; + fpa0_reg = _FPA0; + fpb0_reg = _FPB0; + fpa1_reg = _FPA1; + fpb1_reg = _FPB1; + } + + if (HAS_PCH_SPLIT(dev)) { + I915_WRITE(PCH_DREF_CONTROL, dev_priv->savePCH_DREF_CONTROL); + I915_WRITE(DISP_ARB_CTL, dev_priv->saveDISP_ARB_CTL); + } + + /* Pipe & plane A info */ + /* Prime the clock */ + if (dev_priv->saveDPLL_A & DPLL_VCO_ENABLE) { + I915_WRITE(dpll_a_reg, dev_priv->saveDPLL_A & + ~DPLL_VCO_ENABLE); + POSTING_READ(dpll_a_reg); + udelay(150); + } + I915_WRITE(fpa0_reg, dev_priv->saveFPA0); + I915_WRITE(fpa1_reg, dev_priv->saveFPA1); + /* Actually enable it */ + I915_WRITE(dpll_a_reg, dev_priv->saveDPLL_A); + POSTING_READ(dpll_a_reg); + udelay(150); + if (INTEL_INFO(dev)->gen >= 4 && !HAS_PCH_SPLIT(dev)) { + I915_WRITE(_DPLL_A_MD, dev_priv->saveDPLL_A_MD); + POSTING_READ(_DPLL_A_MD); + } + udelay(150); + + /* Restore mode */ + I915_WRITE(_HTOTAL_A, dev_priv->saveHTOTAL_A); + I915_WRITE(_HBLANK_A, dev_priv->saveHBLANK_A); + I915_WRITE(_HSYNC_A, dev_priv->saveHSYNC_A); + I915_WRITE(_VTOTAL_A, dev_priv->saveVTOTAL_A); + I915_WRITE(_VBLANK_A, dev_priv->saveVBLANK_A); + I915_WRITE(_VSYNC_A, dev_priv->saveVSYNC_A); + if (!HAS_PCH_SPLIT(dev)) + I915_WRITE(_BCLRPAT_A, dev_priv->saveBCLRPAT_A); + + if (HAS_PCH_SPLIT(dev)) { + I915_WRITE(_PIPEA_DATA_M1, dev_priv->savePIPEA_DATA_M1); + I915_WRITE(_PIPEA_DATA_N1, dev_priv->savePIPEA_DATA_N1); + I915_WRITE(_PIPEA_LINK_M1, dev_priv->savePIPEA_LINK_M1); + I915_WRITE(_PIPEA_LINK_N1, dev_priv->savePIPEA_LINK_N1); + + I915_WRITE(_FDI_RXA_CTL, dev_priv->saveFDI_RXA_CTL); + I915_WRITE(_FDI_TXA_CTL, dev_priv->saveFDI_TXA_CTL); + + I915_WRITE(_PFA_CTL_1, dev_priv->savePFA_CTL_1); + I915_WRITE(_PFA_WIN_SZ, dev_priv->savePFA_WIN_SZ); + I915_WRITE(_PFA_WIN_POS, dev_priv->savePFA_WIN_POS); + + I915_WRITE(_TRANSACONF, dev_priv->saveTRANSACONF); + I915_WRITE(_TRANS_HTOTAL_A, dev_priv->saveTRANS_HTOTAL_A); + I915_WRITE(_TRANS_HBLANK_A, dev_priv->saveTRANS_HBLANK_A); + I915_WRITE(_TRANS_HSYNC_A, dev_priv->saveTRANS_HSYNC_A); + I915_WRITE(_TRANS_VTOTAL_A, dev_priv->saveTRANS_VTOTAL_A); + I915_WRITE(_TRANS_VBLANK_A, dev_priv->saveTRANS_VBLANK_A); + I915_WRITE(_TRANS_VSYNC_A, dev_priv->saveTRANS_VSYNC_A); + } + + /* Restore plane info */ + I915_WRITE(_DSPASIZE, dev_priv->saveDSPASIZE); + I915_WRITE(_DSPAPOS, dev_priv->saveDSPAPOS); + I915_WRITE(_PIPEASRC, dev_priv->savePIPEASRC); + I915_WRITE(_DSPAADDR, dev_priv->saveDSPAADDR); + I915_WRITE(_DSPASTRIDE, dev_priv->saveDSPASTRIDE); + if (INTEL_INFO(dev)->gen >= 4) { + I915_WRITE(_DSPASURF, dev_priv->saveDSPASURF); + I915_WRITE(_DSPATILEOFF, dev_priv->saveDSPATILEOFF); + } + + I915_WRITE(_PIPEACONF, dev_priv->savePIPEACONF); + + i915_restore_palette(dev, PIPE_A); + /* Enable the plane */ + I915_WRITE(_DSPACNTR, dev_priv->saveDSPACNTR); + I915_WRITE(_DSPAADDR, I915_READ(_DSPAADDR)); + + /* Pipe & plane B info */ + if (dev_priv->saveDPLL_B & DPLL_VCO_ENABLE) { + I915_WRITE(dpll_b_reg, dev_priv->saveDPLL_B & + ~DPLL_VCO_ENABLE); + POSTING_READ(dpll_b_reg); + udelay(150); + } + I915_WRITE(fpb0_reg, dev_priv->saveFPB0); + I915_WRITE(fpb1_reg, dev_priv->saveFPB1); + /* Actually enable it */ + I915_WRITE(dpll_b_reg, dev_priv->saveDPLL_B); + POSTING_READ(dpll_b_reg); + udelay(150); + if (INTEL_INFO(dev)->gen >= 4 && !HAS_PCH_SPLIT(dev)) { + I915_WRITE(_DPLL_B_MD, dev_priv->saveDPLL_B_MD); + POSTING_READ(_DPLL_B_MD); + } + udelay(150); + + /* Restore mode */ + I915_WRITE(_HTOTAL_B, dev_priv->saveHTOTAL_B); + I915_WRITE(_HBLANK_B, dev_priv->saveHBLANK_B); + I915_WRITE(_HSYNC_B, dev_priv->saveHSYNC_B); + I915_WRITE(_VTOTAL_B, dev_priv->saveVTOTAL_B); + I915_WRITE(_VBLANK_B, dev_priv->saveVBLANK_B); + I915_WRITE(_VSYNC_B, dev_priv->saveVSYNC_B); + if (!HAS_PCH_SPLIT(dev)) + I915_WRITE(_BCLRPAT_B, dev_priv->saveBCLRPAT_B); + + if (HAS_PCH_SPLIT(dev)) { + I915_WRITE(_PIPEB_DATA_M1, dev_priv->savePIPEB_DATA_M1); + I915_WRITE(_PIPEB_DATA_N1, dev_priv->savePIPEB_DATA_N1); + I915_WRITE(_PIPEB_LINK_M1, dev_priv->savePIPEB_LINK_M1); + I915_WRITE(_PIPEB_LINK_N1, dev_priv->savePIPEB_LINK_N1); + + I915_WRITE(_FDI_RXB_CTL, dev_priv->saveFDI_RXB_CTL); + I915_WRITE(_FDI_TXB_CTL, dev_priv->saveFDI_TXB_CTL); + + I915_WRITE(_PFB_CTL_1, dev_priv->savePFB_CTL_1); + I915_WRITE(_PFB_WIN_SZ, dev_priv->savePFB_WIN_SZ); + I915_WRITE(_PFB_WIN_POS, dev_priv->savePFB_WIN_POS); + + I915_WRITE(_TRANSBCONF, dev_priv->saveTRANSBCONF); + I915_WRITE(_TRANS_HTOTAL_B, dev_priv->saveTRANS_HTOTAL_B); + I915_WRITE(_TRANS_HBLANK_B, dev_priv->saveTRANS_HBLANK_B); + I915_WRITE(_TRANS_HSYNC_B, dev_priv->saveTRANS_HSYNC_B); + I915_WRITE(_TRANS_VTOTAL_B, dev_priv->saveTRANS_VTOTAL_B); + I915_WRITE(_TRANS_VBLANK_B, dev_priv->saveTRANS_VBLANK_B); + I915_WRITE(_TRANS_VSYNC_B, dev_priv->saveTRANS_VSYNC_B); + } + + /* Restore plane info */ + I915_WRITE(_DSPBSIZE, dev_priv->saveDSPBSIZE); + I915_WRITE(_DSPBPOS, dev_priv->saveDSPBPOS); + I915_WRITE(_PIPEBSRC, dev_priv->savePIPEBSRC); + I915_WRITE(_DSPBADDR, dev_priv->saveDSPBADDR); + I915_WRITE(_DSPBSTRIDE, dev_priv->saveDSPBSTRIDE); + if (INTEL_INFO(dev)->gen >= 4) { + I915_WRITE(_DSPBSURF, dev_priv->saveDSPBSURF); + I915_WRITE(_DSPBTILEOFF, dev_priv->saveDSPBTILEOFF); + } + + I915_WRITE(_PIPEBCONF, dev_priv->savePIPEBCONF); + + i915_restore_palette(dev, PIPE_B); + /* Enable the plane */ + I915_WRITE(_DSPBCNTR, dev_priv->saveDSPBCNTR); + I915_WRITE(_DSPBADDR, I915_READ(_DSPBADDR)); + + /* Cursor state */ + I915_WRITE(_CURAPOS, dev_priv->saveCURAPOS); + I915_WRITE(_CURACNTR, dev_priv->saveCURACNTR); + I915_WRITE(_CURABASE, dev_priv->saveCURABASE); + I915_WRITE(_CURBPOS, dev_priv->saveCURBPOS); + I915_WRITE(_CURBCNTR, dev_priv->saveCURBCNTR); + I915_WRITE(_CURBBASE, dev_priv->saveCURBBASE); + if (IS_GEN2(dev)) + I915_WRITE(CURSIZE, dev_priv->saveCURSIZE); + + return; +} + +static void i915_save_display(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + /* Display arbitration control */ + dev_priv->saveDSPARB = I915_READ(DSPARB); + + /* This is only meaningful in non-KMS mode */ + /* Don't save them in KMS mode */ + i915_save_modeset_reg(dev); + + /* CRT state */ + if (HAS_PCH_SPLIT(dev)) { + dev_priv->saveADPA = I915_READ(PCH_ADPA); + } else { + dev_priv->saveADPA = I915_READ(ADPA); + } + + /* LVDS state */ + if (HAS_PCH_SPLIT(dev)) { + dev_priv->savePP_CONTROL = I915_READ(PCH_PP_CONTROL); + dev_priv->saveBLC_PWM_CTL = I915_READ(BLC_PWM_PCH_CTL1); + dev_priv->saveBLC_PWM_CTL2 = I915_READ(BLC_PWM_PCH_CTL2); + dev_priv->saveBLC_CPU_PWM_CTL = I915_READ(BLC_PWM_CPU_CTL); + dev_priv->saveBLC_CPU_PWM_CTL2 = I915_READ(BLC_PWM_CPU_CTL2); + dev_priv->saveLVDS = I915_READ(PCH_LVDS); + } else { + dev_priv->savePP_CONTROL = I915_READ(PP_CONTROL); + dev_priv->savePFIT_PGM_RATIOS = I915_READ(PFIT_PGM_RATIOS); + dev_priv->saveBLC_PWM_CTL = I915_READ(BLC_PWM_CTL); + dev_priv->saveBLC_HIST_CTL = I915_READ(BLC_HIST_CTL); + if (INTEL_INFO(dev)->gen >= 4) + dev_priv->saveBLC_PWM_CTL2 = I915_READ(BLC_PWM_CTL2); + if (IS_MOBILE(dev) && !IS_I830(dev)) + dev_priv->saveLVDS = I915_READ(LVDS); + } + + if (!IS_I830(dev) && !IS_845G(dev) && !HAS_PCH_SPLIT(dev)) + dev_priv->savePFIT_CONTROL = I915_READ(PFIT_CONTROL); + + if (HAS_PCH_SPLIT(dev)) { + dev_priv->savePP_ON_DELAYS = I915_READ(PCH_PP_ON_DELAYS); + dev_priv->savePP_OFF_DELAYS = I915_READ(PCH_PP_OFF_DELAYS); + dev_priv->savePP_DIVISOR = I915_READ(PCH_PP_DIVISOR); + } else { + dev_priv->savePP_ON_DELAYS = I915_READ(PP_ON_DELAYS); + dev_priv->savePP_OFF_DELAYS = I915_READ(PP_OFF_DELAYS); + dev_priv->savePP_DIVISOR = I915_READ(PP_DIVISOR); + } + + /* Display Port state */ + if (SUPPORTS_INTEGRATED_DP(dev)) { + dev_priv->saveDP_B = I915_READ(DP_B); + dev_priv->saveDP_C = I915_READ(DP_C); + dev_priv->saveDP_D = I915_READ(DP_D); + dev_priv->savePIPEA_GMCH_DATA_M = I915_READ(_PIPEA_GMCH_DATA_M); + dev_priv->savePIPEB_GMCH_DATA_M = I915_READ(_PIPEB_GMCH_DATA_M); + dev_priv->savePIPEA_GMCH_DATA_N = I915_READ(_PIPEA_GMCH_DATA_N); + dev_priv->savePIPEB_GMCH_DATA_N = I915_READ(_PIPEB_GMCH_DATA_N); + dev_priv->savePIPEA_DP_LINK_M = I915_READ(_PIPEA_DP_LINK_M); + dev_priv->savePIPEB_DP_LINK_M = I915_READ(_PIPEB_DP_LINK_M); + dev_priv->savePIPEA_DP_LINK_N = I915_READ(_PIPEA_DP_LINK_N); + dev_priv->savePIPEB_DP_LINK_N = I915_READ(_PIPEB_DP_LINK_N); + } + /* FIXME: save TV & SDVO state */ + + /* Only save FBC state on the platform that supports FBC */ + if (I915_HAS_FBC(dev)) { + if (HAS_PCH_SPLIT(dev)) { + dev_priv->saveDPFC_CB_BASE = I915_READ(ILK_DPFC_CB_BASE); + } else if (IS_GM45(dev)) { + dev_priv->saveDPFC_CB_BASE = I915_READ(DPFC_CB_BASE); + } else { + dev_priv->saveFBC_CFB_BASE = I915_READ(FBC_CFB_BASE); + dev_priv->saveFBC_LL_BASE = I915_READ(FBC_LL_BASE); + dev_priv->saveFBC_CONTROL2 = I915_READ(FBC_CONTROL2); + dev_priv->saveFBC_CONTROL = I915_READ(FBC_CONTROL); + } + } + + /* VGA state */ + dev_priv->saveVGA0 = I915_READ(VGA0); + dev_priv->saveVGA1 = I915_READ(VGA1); + dev_priv->saveVGA_PD = I915_READ(VGA_PD); + if (HAS_PCH_SPLIT(dev)) + dev_priv->saveVGACNTRL = I915_READ(CPU_VGACNTRL); + else + dev_priv->saveVGACNTRL = I915_READ(VGACNTRL); + + i915_save_vga(dev); +} + +static void i915_restore_display(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + /* Display arbitration */ + I915_WRITE(DSPARB, dev_priv->saveDSPARB); + + /* Display port ratios (must be done before clock is set) */ + if (SUPPORTS_INTEGRATED_DP(dev)) { + I915_WRITE(_PIPEA_GMCH_DATA_M, dev_priv->savePIPEA_GMCH_DATA_M); + I915_WRITE(_PIPEB_GMCH_DATA_M, dev_priv->savePIPEB_GMCH_DATA_M); + I915_WRITE(_PIPEA_GMCH_DATA_N, dev_priv->savePIPEA_GMCH_DATA_N); + I915_WRITE(_PIPEB_GMCH_DATA_N, dev_priv->savePIPEB_GMCH_DATA_N); + I915_WRITE(_PIPEA_DP_LINK_M, dev_priv->savePIPEA_DP_LINK_M); + I915_WRITE(_PIPEB_DP_LINK_M, dev_priv->savePIPEB_DP_LINK_M); + I915_WRITE(_PIPEA_DP_LINK_N, dev_priv->savePIPEA_DP_LINK_N); + I915_WRITE(_PIPEB_DP_LINK_N, dev_priv->savePIPEB_DP_LINK_N); + } + + /* This is only meaningful in non-KMS mode */ + /* Don't restore them in KMS mode */ + i915_restore_modeset_reg(dev); + + /* CRT state */ + if (HAS_PCH_SPLIT(dev)) + I915_WRITE(PCH_ADPA, dev_priv->saveADPA); + else + I915_WRITE(ADPA, dev_priv->saveADPA); + + /* LVDS state */ + if (INTEL_INFO(dev)->gen >= 4 && !HAS_PCH_SPLIT(dev)) + I915_WRITE(BLC_PWM_CTL2, dev_priv->saveBLC_PWM_CTL2); + + if (HAS_PCH_SPLIT(dev)) { + I915_WRITE(PCH_LVDS, dev_priv->saveLVDS); + } else if (IS_MOBILE(dev) && !IS_I830(dev)) + I915_WRITE(LVDS, dev_priv->saveLVDS); + + if (!IS_I830(dev) && !IS_845G(dev) && !HAS_PCH_SPLIT(dev)) + I915_WRITE(PFIT_CONTROL, dev_priv->savePFIT_CONTROL); + + if (HAS_PCH_SPLIT(dev)) { + I915_WRITE(BLC_PWM_PCH_CTL1, dev_priv->saveBLC_PWM_CTL); + I915_WRITE(BLC_PWM_PCH_CTL2, dev_priv->saveBLC_PWM_CTL2); + I915_WRITE(BLC_PWM_CPU_CTL, dev_priv->saveBLC_CPU_PWM_CTL); + I915_WRITE(BLC_PWM_CPU_CTL2, dev_priv->saveBLC_CPU_PWM_CTL2); + I915_WRITE(PCH_PP_ON_DELAYS, dev_priv->savePP_ON_DELAYS); + I915_WRITE(PCH_PP_OFF_DELAYS, dev_priv->savePP_OFF_DELAYS); + I915_WRITE(PCH_PP_DIVISOR, dev_priv->savePP_DIVISOR); + I915_WRITE(PCH_PP_CONTROL, dev_priv->savePP_CONTROL); + I915_WRITE(RSTDBYCTL, + dev_priv->saveMCHBAR_RENDER_STANDBY); + } else { + I915_WRITE(PFIT_PGM_RATIOS, dev_priv->savePFIT_PGM_RATIOS); + I915_WRITE(BLC_PWM_CTL, dev_priv->saveBLC_PWM_CTL); + I915_WRITE(BLC_HIST_CTL, dev_priv->saveBLC_HIST_CTL); + I915_WRITE(PP_ON_DELAYS, dev_priv->savePP_ON_DELAYS); + I915_WRITE(PP_OFF_DELAYS, dev_priv->savePP_OFF_DELAYS); + I915_WRITE(PP_DIVISOR, dev_priv->savePP_DIVISOR); + I915_WRITE(PP_CONTROL, dev_priv->savePP_CONTROL); + } + + /* Display Port state */ + if (SUPPORTS_INTEGRATED_DP(dev)) { + I915_WRITE(DP_B, dev_priv->saveDP_B); + I915_WRITE(DP_C, dev_priv->saveDP_C); + I915_WRITE(DP_D, dev_priv->saveDP_D); + } + /* FIXME: restore TV & SDVO state */ + + /* only restore FBC info on the platform that supports FBC*/ + if (I915_HAS_FBC(dev)) { + if (HAS_PCH_SPLIT(dev)) { + ironlake_disable_fbc(dev); + I915_WRITE(ILK_DPFC_CB_BASE, dev_priv->saveDPFC_CB_BASE); + } else if (IS_GM45(dev)) { + g4x_disable_fbc(dev); + I915_WRITE(DPFC_CB_BASE, dev_priv->saveDPFC_CB_BASE); + } else { + i8xx_disable_fbc(dev); + I915_WRITE(FBC_CFB_BASE, dev_priv->saveFBC_CFB_BASE); + I915_WRITE(FBC_LL_BASE, dev_priv->saveFBC_LL_BASE); + I915_WRITE(FBC_CONTROL2, dev_priv->saveFBC_CONTROL2); + I915_WRITE(FBC_CONTROL, dev_priv->saveFBC_CONTROL); + } + } + /* VGA state */ + if (HAS_PCH_SPLIT(dev)) + I915_WRITE(CPU_VGACNTRL, dev_priv->saveVGACNTRL); + else + I915_WRITE(VGACNTRL, dev_priv->saveVGACNTRL); + + I915_WRITE(VGA0, dev_priv->saveVGA0); + I915_WRITE(VGA1, dev_priv->saveVGA1); + I915_WRITE(VGA_PD, dev_priv->saveVGA_PD); + POSTING_READ(VGA_PD); + udelay(150); + + i915_restore_vga(dev); +} + +int i915_save_state(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + int i; + + pci_read_config_byte(dev->pdev, LBB, &dev_priv->saveLBB); + + mutex_lock(&dev->struct_mutex); + + /* Hardware status page */ + dev_priv->saveHWS = I915_READ(HWS_PGA); + + i915_save_display(dev); + + /* Interrupt state */ + if (HAS_PCH_SPLIT(dev)) { + dev_priv->saveDEIER = I915_READ(DEIER); + dev_priv->saveDEIMR = I915_READ(DEIMR); + dev_priv->saveGTIER = I915_READ(GTIER); + dev_priv->saveGTIMR = I915_READ(GTIMR); + dev_priv->saveFDI_RXA_IMR = I915_READ(_FDI_RXA_IMR); + dev_priv->saveFDI_RXB_IMR = I915_READ(_FDI_RXB_IMR); + dev_priv->saveMCHBAR_RENDER_STANDBY = + I915_READ(RSTDBYCTL); + dev_priv->savePCH_PORT_HOTPLUG = I915_READ(PCH_PORT_HOTPLUG); + } else { + dev_priv->saveIER = I915_READ(IER); + dev_priv->saveIMR = I915_READ(IMR); + } + + if (IS_IRONLAKE_M(dev)) + ironlake_disable_drps(dev); + if (IS_GEN6(dev)) + gen6_disable_rps(dev); + + /* Cache mode state */ + dev_priv->saveCACHE_MODE_0 = I915_READ(CACHE_MODE_0); + + /* Memory Arbitration state */ + dev_priv->saveMI_ARB_STATE = I915_READ(MI_ARB_STATE); + + /* Scratch space */ + for (i = 0; i < 16; i++) { + dev_priv->saveSWF0[i] = I915_READ(SWF00 + (i << 2)); + dev_priv->saveSWF1[i] = I915_READ(SWF10 + (i << 2)); + } + for (i = 0; i < 3; i++) + dev_priv->saveSWF2[i] = I915_READ(SWF30 + (i << 2)); + + mutex_unlock(&dev->struct_mutex); + + return 0; +} + +int i915_restore_state(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + int i; + + pci_write_config_byte(dev->pdev, LBB, dev_priv->saveLBB); + + mutex_lock(&dev->struct_mutex); + + /* Hardware status page */ + I915_WRITE(HWS_PGA, dev_priv->saveHWS); + + i915_restore_display(dev); + + /* Interrupt state */ + if (HAS_PCH_SPLIT(dev)) { + I915_WRITE(DEIER, dev_priv->saveDEIER); + I915_WRITE(DEIMR, dev_priv->saveDEIMR); + I915_WRITE(GTIER, dev_priv->saveGTIER); + I915_WRITE(GTIMR, dev_priv->saveGTIMR); + I915_WRITE(_FDI_RXA_IMR, dev_priv->saveFDI_RXA_IMR); + I915_WRITE(_FDI_RXB_IMR, dev_priv->saveFDI_RXB_IMR); + I915_WRITE(PCH_PORT_HOTPLUG, dev_priv->savePCH_PORT_HOTPLUG); + } else { + I915_WRITE(IER, dev_priv->saveIER); + I915_WRITE(IMR, dev_priv->saveIMR); + } + mutex_unlock(&dev->struct_mutex); + + intel_init_clock_gating(dev); + + if (IS_IRONLAKE_M(dev)) { + ironlake_enable_drps(dev); + intel_init_emon(dev); + } + + if (IS_GEN6(dev)) + gen6_enable_rps(dev_priv); + + mutex_lock(&dev->struct_mutex); + + /* Cache mode state */ + I915_WRITE (CACHE_MODE_0, dev_priv->saveCACHE_MODE_0 | 0xffff0000); + + /* Memory arbitration state */ + I915_WRITE (MI_ARB_STATE, dev_priv->saveMI_ARB_STATE | 0xffff0000); + + for (i = 0; i < 16; i++) { + I915_WRITE(SWF00 + (i << 2), dev_priv->saveSWF0[i]); + I915_WRITE(SWF10 + (i << 2), dev_priv->saveSWF1[i]); + } + for (i = 0; i < 3; i++) + I915_WRITE(SWF30 + (i << 2), dev_priv->saveSWF2[i]); + + mutex_unlock(&dev->struct_mutex); + + intel_i2c_reset(dev); + + return 0; +} diff --git a/drivers/gpu/drm/i915/i915_trace.h b/drivers/gpu/drm/i915/i915_trace.h new file mode 100644 index 00000000..d623fefb --- /dev/null +++ b/drivers/gpu/drm/i915/i915_trace.h @@ -0,0 +1,418 @@ +#if !defined(_I915_TRACE_H_) || defined(TRACE_HEADER_MULTI_READ) +#define _I915_TRACE_H_ + +#include +#include +#include + +#include +#include "i915_drv.h" +#include "intel_ringbuffer.h" + +#undef TRACE_SYSTEM +#define TRACE_SYSTEM i915 +#define TRACE_SYSTEM_STRING __stringify(TRACE_SYSTEM) +#define TRACE_INCLUDE_FILE i915_trace + +/* object tracking */ + +TRACE_EVENT(i915_gem_object_create, + TP_PROTO(struct drm_i915_gem_object *obj), + TP_ARGS(obj), + + TP_STRUCT__entry( + __field(struct drm_i915_gem_object *, obj) + __field(u32, size) + ), + + TP_fast_assign( + __entry->obj = obj; + __entry->size = obj->base.size; + ), + + TP_printk("obj=%p, size=%u", __entry->obj, __entry->size) +); + +TRACE_EVENT(i915_gem_object_bind, + TP_PROTO(struct drm_i915_gem_object *obj, bool mappable), + TP_ARGS(obj, mappable), + + TP_STRUCT__entry( + __field(struct drm_i915_gem_object *, obj) + __field(u32, offset) + __field(u32, size) + __field(bool, mappable) + ), + + TP_fast_assign( + __entry->obj = obj; + __entry->offset = obj->gtt_space->start; + __entry->size = obj->gtt_space->size; + __entry->mappable = mappable; + ), + + TP_printk("obj=%p, offset=%08x size=%x%s", + __entry->obj, __entry->offset, __entry->size, + __entry->mappable ? ", mappable" : "") +); + +TRACE_EVENT(i915_gem_object_unbind, + TP_PROTO(struct drm_i915_gem_object *obj), + TP_ARGS(obj), + + TP_STRUCT__entry( + __field(struct drm_i915_gem_object *, obj) + __field(u32, offset) + __field(u32, size) + ), + + TP_fast_assign( + __entry->obj = obj; + __entry->offset = obj->gtt_space->start; + __entry->size = obj->gtt_space->size; + ), + + TP_printk("obj=%p, offset=%08x size=%x", + __entry->obj, __entry->offset, __entry->size) +); + +TRACE_EVENT(i915_gem_object_change_domain, + TP_PROTO(struct drm_i915_gem_object *obj, u32 old_read, u32 old_write), + TP_ARGS(obj, old_read, old_write), + + TP_STRUCT__entry( + __field(struct drm_i915_gem_object *, obj) + __field(u32, read_domains) + __field(u32, write_domain) + ), + + TP_fast_assign( + __entry->obj = obj; + __entry->read_domains = obj->base.read_domains | (old_read << 16); + __entry->write_domain = obj->base.write_domain | (old_write << 16); + ), + + TP_printk("obj=%p, read=%02x=>%02x, write=%02x=>%02x", + __entry->obj, + __entry->read_domains >> 16, + __entry->read_domains & 0xffff, + __entry->write_domain >> 16, + __entry->write_domain & 0xffff) +); + +TRACE_EVENT(i915_gem_object_pwrite, + TP_PROTO(struct drm_i915_gem_object *obj, u32 offset, u32 len), + TP_ARGS(obj, offset, len), + + TP_STRUCT__entry( + __field(struct drm_i915_gem_object *, obj) + __field(u32, offset) + __field(u32, len) + ), + + TP_fast_assign( + __entry->obj = obj; + __entry->offset = offset; + __entry->len = len; + ), + + TP_printk("obj=%p, offset=%u, len=%u", + __entry->obj, __entry->offset, __entry->len) +); + +TRACE_EVENT(i915_gem_object_pread, + TP_PROTO(struct drm_i915_gem_object *obj, u32 offset, u32 len), + TP_ARGS(obj, offset, len), + + TP_STRUCT__entry( + __field(struct drm_i915_gem_object *, obj) + __field(u32, offset) + __field(u32, len) + ), + + TP_fast_assign( + __entry->obj = obj; + __entry->offset = offset; + __entry->len = len; + ), + + TP_printk("obj=%p, offset=%u, len=%u", + __entry->obj, __entry->offset, __entry->len) +); + +TRACE_EVENT(i915_gem_object_fault, + TP_PROTO(struct drm_i915_gem_object *obj, u32 index, bool gtt, bool write), + TP_ARGS(obj, index, gtt, write), + + TP_STRUCT__entry( + __field(struct drm_i915_gem_object *, obj) + __field(u32, index) + __field(bool, gtt) + __field(bool, write) + ), + + TP_fast_assign( + __entry->obj = obj; + __entry->index = index; + __entry->gtt = gtt; + __entry->write = write; + ), + + TP_printk("obj=%p, %s index=%u %s", + __entry->obj, + __entry->gtt ? "GTT" : "CPU", + __entry->index, + __entry->write ? ", writable" : "") +); + +DECLARE_EVENT_CLASS(i915_gem_object, + TP_PROTO(struct drm_i915_gem_object *obj), + TP_ARGS(obj), + + TP_STRUCT__entry( + __field(struct drm_i915_gem_object *, obj) + ), + + TP_fast_assign( + __entry->obj = obj; + ), + + TP_printk("obj=%p", __entry->obj) +); + +DEFINE_EVENT(i915_gem_object, i915_gem_object_clflush, + TP_PROTO(struct drm_i915_gem_object *obj), + TP_ARGS(obj) +); + +DEFINE_EVENT(i915_gem_object, i915_gem_object_destroy, + TP_PROTO(struct drm_i915_gem_object *obj), + TP_ARGS(obj) +); + +TRACE_EVENT(i915_gem_evict, + TP_PROTO(struct drm_device *dev, u32 size, u32 align, bool mappable), + TP_ARGS(dev, size, align, mappable), + + TP_STRUCT__entry( + __field(u32, dev) + __field(u32, size) + __field(u32, align) + __field(bool, mappable) + ), + + TP_fast_assign( + __entry->dev = dev->primary->index; + __entry->size = size; + __entry->align = align; + __entry->mappable = mappable; + ), + + TP_printk("dev=%d, size=%d, align=%d %s", + __entry->dev, __entry->size, __entry->align, + __entry->mappable ? ", mappable" : "") +); + +TRACE_EVENT(i915_gem_evict_everything, + TP_PROTO(struct drm_device *dev, bool purgeable), + TP_ARGS(dev, purgeable), + + TP_STRUCT__entry( + __field(u32, dev) + __field(bool, purgeable) + ), + + TP_fast_assign( + __entry->dev = dev->primary->index; + __entry->purgeable = purgeable; + ), + + TP_printk("dev=%d%s", + __entry->dev, + __entry->purgeable ? ", purgeable only" : "") +); + +TRACE_EVENT(i915_gem_ring_dispatch, + TP_PROTO(struct intel_ring_buffer *ring, u32 seqno), + TP_ARGS(ring, seqno), + + TP_STRUCT__entry( + __field(u32, dev) + __field(u32, ring) + __field(u32, seqno) + ), + + TP_fast_assign( + __entry->dev = ring->dev->primary->index; + __entry->ring = ring->id; + __entry->seqno = seqno; + i915_trace_irq_get(ring, seqno); + ), + + TP_printk("dev=%u, ring=%u, seqno=%u", + __entry->dev, __entry->ring, __entry->seqno) +); + +TRACE_EVENT(i915_gem_ring_flush, + TP_PROTO(struct intel_ring_buffer *ring, u32 invalidate, u32 flush), + TP_ARGS(ring, invalidate, flush), + + TP_STRUCT__entry( + __field(u32, dev) + __field(u32, ring) + __field(u32, invalidate) + __field(u32, flush) + ), + + TP_fast_assign( + __entry->dev = ring->dev->primary->index; + __entry->ring = ring->id; + __entry->invalidate = invalidate; + __entry->flush = flush; + ), + + TP_printk("dev=%u, ring=%x, invalidate=%04x, flush=%04x", + __entry->dev, __entry->ring, + __entry->invalidate, __entry->flush) +); + +DECLARE_EVENT_CLASS(i915_gem_request, + TP_PROTO(struct intel_ring_buffer *ring, u32 seqno), + TP_ARGS(ring, seqno), + + TP_STRUCT__entry( + __field(u32, dev) + __field(u32, ring) + __field(u32, seqno) + ), + + TP_fast_assign( + __entry->dev = ring->dev->primary->index; + __entry->ring = ring->id; + __entry->seqno = seqno; + ), + + TP_printk("dev=%u, ring=%u, seqno=%u", + __entry->dev, __entry->ring, __entry->seqno) +); + +DEFINE_EVENT(i915_gem_request, i915_gem_request_add, + TP_PROTO(struct intel_ring_buffer *ring, u32 seqno), + TP_ARGS(ring, seqno) +); + +DEFINE_EVENT(i915_gem_request, i915_gem_request_complete, + TP_PROTO(struct intel_ring_buffer *ring, u32 seqno), + TP_ARGS(ring, seqno) +); + +DEFINE_EVENT(i915_gem_request, i915_gem_request_retire, + TP_PROTO(struct intel_ring_buffer *ring, u32 seqno), + TP_ARGS(ring, seqno) +); + +DEFINE_EVENT(i915_gem_request, i915_gem_request_wait_begin, + TP_PROTO(struct intel_ring_buffer *ring, u32 seqno), + TP_ARGS(ring, seqno) +); + +DEFINE_EVENT(i915_gem_request, i915_gem_request_wait_end, + TP_PROTO(struct intel_ring_buffer *ring, u32 seqno), + TP_ARGS(ring, seqno) +); + +DECLARE_EVENT_CLASS(i915_ring, + TP_PROTO(struct intel_ring_buffer *ring), + TP_ARGS(ring), + + TP_STRUCT__entry( + __field(u32, dev) + __field(u32, ring) + ), + + TP_fast_assign( + __entry->dev = ring->dev->primary->index; + __entry->ring = ring->id; + ), + + TP_printk("dev=%u, ring=%u", __entry->dev, __entry->ring) +); + +DEFINE_EVENT(i915_ring, i915_ring_wait_begin, + TP_PROTO(struct intel_ring_buffer *ring), + TP_ARGS(ring) +); + +DEFINE_EVENT(i915_ring, i915_ring_wait_end, + TP_PROTO(struct intel_ring_buffer *ring), + TP_ARGS(ring) +); + +TRACE_EVENT(i915_flip_request, + TP_PROTO(int plane, struct drm_i915_gem_object *obj), + + TP_ARGS(plane, obj), + + TP_STRUCT__entry( + __field(int, plane) + __field(struct drm_i915_gem_object *, obj) + ), + + TP_fast_assign( + __entry->plane = plane; + __entry->obj = obj; + ), + + TP_printk("plane=%d, obj=%p", __entry->plane, __entry->obj) +); + +TRACE_EVENT(i915_flip_complete, + TP_PROTO(int plane, struct drm_i915_gem_object *obj), + + TP_ARGS(plane, obj), + + TP_STRUCT__entry( + __field(int, plane) + __field(struct drm_i915_gem_object *, obj) + ), + + TP_fast_assign( + __entry->plane = plane; + __entry->obj = obj; + ), + + TP_printk("plane=%d, obj=%p", __entry->plane, __entry->obj) +); + +TRACE_EVENT(i915_reg_rw, + TP_PROTO(bool write, u32 reg, u64 val, int len), + + TP_ARGS(write, reg, val, len), + + TP_STRUCT__entry( + __field(u64, val) + __field(u32, reg) + __field(u16, write) + __field(u16, len) + ), + + TP_fast_assign( + __entry->val = (u64)val; + __entry->reg = reg; + __entry->write = write; + __entry->len = len; + ), + + TP_printk("%s reg=0x%x, len=%d, val=(0x%x, 0x%x)", + __entry->write ? "write" : "read", + __entry->reg, __entry->len, + (u32)(__entry->val & 0xffffffff), + (u32)(__entry->val >> 32)) +); + +#endif /* _I915_TRACE_H_ */ + +/* This part must be outside protection */ +#undef TRACE_INCLUDE_PATH +#define TRACE_INCLUDE_PATH . +#include diff --git a/drivers/gpu/drm/i915/i915_trace_points.c b/drivers/gpu/drm/i915/i915_trace_points.c new file mode 100644 index 00000000..ead876eb --- /dev/null +++ b/drivers/gpu/drm/i915/i915_trace_points.c @@ -0,0 +1,11 @@ +/* + * Copyright © 2009 Intel Corporation + * + * Authors: + * Chris Wilson + */ + +#include "i915_drv.h" + +#define CREATE_TRACE_POINTS +#include "i915_trace.h" diff --git a/drivers/gpu/drm/i915/intel_acpi.c b/drivers/gpu/drm/i915/intel_acpi.c new file mode 100644 index 00000000..2cb8e0b9 --- /dev/null +++ b/drivers/gpu/drm/i915/intel_acpi.c @@ -0,0 +1,252 @@ +/* + * Intel ACPI functions + * + * _DSM related code stolen from nouveau_acpi.c. + */ +#include +#include +#include +#include + +#include "drmP.h" + +#define INTEL_DSM_REVISION_ID 1 /* For Calpella anyway... */ + +#define INTEL_DSM_FN_SUPPORTED_FUNCTIONS 0 /* No args */ +#define INTEL_DSM_FN_PLATFORM_MUX_INFO 1 /* No args */ + +static struct intel_dsm_priv { + acpi_handle dhandle; +} intel_dsm_priv; + +static const u8 intel_dsm_guid[] = { + 0xd3, 0x73, 0xd8, 0x7e, + 0xd0, 0xc2, + 0x4f, 0x4e, + 0xa8, 0x54, + 0x0f, 0x13, 0x17, 0xb0, 0x1c, 0x2c +}; + +static int intel_dsm(acpi_handle handle, int func, int arg) +{ + struct acpi_buffer output = { ACPI_ALLOCATE_BUFFER, NULL }; + struct acpi_object_list input; + union acpi_object params[4]; + union acpi_object *obj; + u32 result; + int ret = 0; + + input.count = 4; + input.pointer = params; + params[0].type = ACPI_TYPE_BUFFER; + params[0].buffer.length = sizeof(intel_dsm_guid); + params[0].buffer.pointer = (char *)intel_dsm_guid; + params[1].type = ACPI_TYPE_INTEGER; + params[1].integer.value = INTEL_DSM_REVISION_ID; + params[2].type = ACPI_TYPE_INTEGER; + params[2].integer.value = func; + params[3].type = ACPI_TYPE_INTEGER; + params[3].integer.value = arg; + + ret = acpi_evaluate_object(handle, "_DSM", &input, &output); + if (ret) { + DRM_DEBUG_DRIVER("failed to evaluate _DSM: %d\n", ret); + return ret; + } + + obj = (union acpi_object *)output.pointer; + + result = 0; + switch (obj->type) { + case ACPI_TYPE_INTEGER: + result = obj->integer.value; + break; + + case ACPI_TYPE_BUFFER: + if (obj->buffer.length == 4) { + result =(obj->buffer.pointer[0] | + (obj->buffer.pointer[1] << 8) | + (obj->buffer.pointer[2] << 16) | + (obj->buffer.pointer[3] << 24)); + break; + } + default: + ret = -EINVAL; + break; + } + if (result == 0x80000002) + ret = -ENODEV; + + kfree(output.pointer); + return ret; +} + +static char *intel_dsm_port_name(u8 id) +{ + switch (id) { + case 0: + return "Reserved"; + case 1: + return "Analog VGA"; + case 2: + return "LVDS"; + case 3: + return "Reserved"; + case 4: + return "HDMI/DVI_B"; + case 5: + return "HDMI/DVI_C"; + case 6: + return "HDMI/DVI_D"; + case 7: + return "DisplayPort_A"; + case 8: + return "DisplayPort_B"; + case 9: + return "DisplayPort_C"; + case 0xa: + return "DisplayPort_D"; + case 0xb: + case 0xc: + case 0xd: + return "Reserved"; + case 0xe: + return "WiDi"; + default: + return "bad type"; + } +} + +static char *intel_dsm_mux_type(u8 type) +{ + switch (type) { + case 0: + return "unknown"; + case 1: + return "No MUX, iGPU only"; + case 2: + return "No MUX, dGPU only"; + case 3: + return "MUXed between iGPU and dGPU"; + default: + return "bad type"; + } +} + +static void intel_dsm_platform_mux_info(void) +{ + struct acpi_buffer output = { ACPI_ALLOCATE_BUFFER, NULL }; + struct acpi_object_list input; + union acpi_object params[4]; + union acpi_object *pkg; + int i, ret; + + input.count = 4; + input.pointer = params; + params[0].type = ACPI_TYPE_BUFFER; + params[0].buffer.length = sizeof(intel_dsm_guid); + params[0].buffer.pointer = (char *)intel_dsm_guid; + params[1].type = ACPI_TYPE_INTEGER; + params[1].integer.value = INTEL_DSM_REVISION_ID; + params[2].type = ACPI_TYPE_INTEGER; + params[2].integer.value = INTEL_DSM_FN_PLATFORM_MUX_INFO; + params[3].type = ACPI_TYPE_INTEGER; + params[3].integer.value = 0; + + ret = acpi_evaluate_object(intel_dsm_priv.dhandle, "_DSM", &input, + &output); + if (ret) { + DRM_DEBUG_DRIVER("failed to evaluate _DSM: %d\n", ret); + goto out; + } + + pkg = (union acpi_object *)output.pointer; + + if (pkg->type == ACPI_TYPE_PACKAGE) { + union acpi_object *connector_count = &pkg->package.elements[0]; + DRM_DEBUG_DRIVER("MUX info connectors: %lld\n", + (unsigned long long)connector_count->integer.value); + for (i = 1; i < pkg->package.count; i++) { + union acpi_object *obj = &pkg->package.elements[i]; + union acpi_object *connector_id = + &obj->package.elements[0]; + union acpi_object *info = &obj->package.elements[1]; + DRM_DEBUG_DRIVER("Connector id: 0x%016llx\n", + (unsigned long long)connector_id->integer.value); + DRM_DEBUG_DRIVER(" port id: %s\n", + intel_dsm_port_name(info->buffer.pointer[0])); + DRM_DEBUG_DRIVER(" display mux info: %s\n", + intel_dsm_mux_type(info->buffer.pointer[1])); + DRM_DEBUG_DRIVER(" aux/dc mux info: %s\n", + intel_dsm_mux_type(info->buffer.pointer[2])); + DRM_DEBUG_DRIVER(" hpd mux info: %s\n", + intel_dsm_mux_type(info->buffer.pointer[3])); + } + } else { + DRM_ERROR("MUX INFO call failed\n"); + } + +out: + kfree(output.pointer); +} + +static bool intel_dsm_pci_probe(struct pci_dev *pdev) +{ + acpi_handle dhandle, intel_handle; + acpi_status status; + int ret; + + dhandle = DEVICE_ACPI_HANDLE(&pdev->dev); + if (!dhandle) + return false; + + status = acpi_get_handle(dhandle, "_DSM", &intel_handle); + if (ACPI_FAILURE(status)) { + DRM_DEBUG_KMS("no _DSM method for intel device\n"); + return false; + } + + ret = intel_dsm(dhandle, INTEL_DSM_FN_SUPPORTED_FUNCTIONS, 0); + if (ret < 0) { + DRM_ERROR("failed to get supported _DSM functions\n"); + return false; + } + + intel_dsm_priv.dhandle = dhandle; + + intel_dsm_platform_mux_info(); + return true; +} + +static bool intel_dsm_detect(void) +{ + char acpi_method_name[255] = { 0 }; + struct acpi_buffer buffer = {sizeof(acpi_method_name), acpi_method_name}; + struct pci_dev *pdev = NULL; + bool has_dsm = false; + int vga_count = 0; + + while ((pdev = pci_get_class(PCI_CLASS_DISPLAY_VGA << 8, pdev)) != NULL) { + vga_count++; + has_dsm |= intel_dsm_pci_probe(pdev); + } + + if (vga_count == 2 && has_dsm) { + acpi_get_name(intel_dsm_priv.dhandle, ACPI_FULL_PATHNAME, &buffer); + DRM_DEBUG_DRIVER("VGA switcheroo: detected DSM switching method %s handle\n", + acpi_method_name); + return true; + } + + return false; +} + +void intel_register_dsm_handler(void) +{ + if (!intel_dsm_detect()) + return; +} + +void intel_unregister_dsm_handler(void) +{ +} diff --git a/drivers/gpu/drm/i915/intel_bios.c b/drivers/gpu/drm/i915/intel_bios.c new file mode 100644 index 00000000..e5fa074b --- /dev/null +++ b/drivers/gpu/drm/i915/intel_bios.c @@ -0,0 +1,703 @@ +/* + * Copyright © 2006 Intel Corporation + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + * + * Authors: + * Eric Anholt + * + */ +#include +#include +#include "drmP.h" +#include "drm.h" +#include "i915_drm.h" +#include "i915_drv.h" +#include "intel_bios.h" + +#define SLAVE_ADDR1 0x70 +#define SLAVE_ADDR2 0x72 + +static int panel_type; + +static void * +find_section(struct bdb_header *bdb, int section_id) +{ + u8 *base = (u8 *)bdb; + int index = 0; + u16 total, current_size; + u8 current_id; + + /* skip to first section */ + index += bdb->header_size; + total = bdb->bdb_size; + + /* walk the sections looking for section_id */ + while (index < total) { + current_id = *(base + index); + index++; + current_size = *((u16 *)(base + index)); + index += 2; + if (current_id == section_id) + return base + index; + index += current_size; + } + + return NULL; +} + +static u16 +get_blocksize(void *p) +{ + u16 *block_ptr, block_size; + + block_ptr = (u16 *)((char *)p - 2); + block_size = *block_ptr; + return block_size; +} + +static void +fill_detail_timing_data(struct drm_display_mode *panel_fixed_mode, + struct lvds_dvo_timing *dvo_timing) +{ + panel_fixed_mode->hdisplay = (dvo_timing->hactive_hi << 8) | + dvo_timing->hactive_lo; + panel_fixed_mode->hsync_start = panel_fixed_mode->hdisplay + + ((dvo_timing->hsync_off_hi << 8) | dvo_timing->hsync_off_lo); + panel_fixed_mode->hsync_end = panel_fixed_mode->hsync_start + + dvo_timing->hsync_pulse_width; + panel_fixed_mode->htotal = panel_fixed_mode->hdisplay + + ((dvo_timing->hblank_hi << 8) | dvo_timing->hblank_lo); + + panel_fixed_mode->vdisplay = (dvo_timing->vactive_hi << 8) | + dvo_timing->vactive_lo; + panel_fixed_mode->vsync_start = panel_fixed_mode->vdisplay + + dvo_timing->vsync_off; + panel_fixed_mode->vsync_end = panel_fixed_mode->vsync_start + + dvo_timing->vsync_pulse_width; + panel_fixed_mode->vtotal = panel_fixed_mode->vdisplay + + ((dvo_timing->vblank_hi << 8) | dvo_timing->vblank_lo); + panel_fixed_mode->clock = dvo_timing->clock * 10; + panel_fixed_mode->type = DRM_MODE_TYPE_PREFERRED; + + if (dvo_timing->hsync_positive) + panel_fixed_mode->flags |= DRM_MODE_FLAG_PHSYNC; + else + panel_fixed_mode->flags |= DRM_MODE_FLAG_NHSYNC; + + if (dvo_timing->vsync_positive) + panel_fixed_mode->flags |= DRM_MODE_FLAG_PVSYNC; + else + panel_fixed_mode->flags |= DRM_MODE_FLAG_NVSYNC; + + /* Some VBTs have bogus h/vtotal values */ + if (panel_fixed_mode->hsync_end > panel_fixed_mode->htotal) + panel_fixed_mode->htotal = panel_fixed_mode->hsync_end + 1; + if (panel_fixed_mode->vsync_end > panel_fixed_mode->vtotal) + panel_fixed_mode->vtotal = panel_fixed_mode->vsync_end + 1; + + drm_mode_set_name(panel_fixed_mode); +} + +/* Try to find integrated panel data */ +static void +parse_lfp_panel_data(struct drm_i915_private *dev_priv, + struct bdb_header *bdb) +{ + struct bdb_lvds_options *lvds_options; + struct bdb_lvds_lfp_data *lvds_lfp_data; + struct bdb_lvds_lfp_data_ptrs *lvds_lfp_data_ptrs; + struct bdb_lvds_lfp_data_entry *entry; + struct lvds_dvo_timing *dvo_timing; + struct drm_display_mode *panel_fixed_mode; + int lfp_data_size, dvo_timing_offset; + int i, temp_downclock; + struct drm_display_mode *temp_mode; + + lvds_options = find_section(bdb, BDB_LVDS_OPTIONS); + if (!lvds_options) + return; + + dev_priv->lvds_dither = lvds_options->pixel_dither; + if (lvds_options->panel_type == 0xff) + return; + + panel_type = lvds_options->panel_type; + + lvds_lfp_data = find_section(bdb, BDB_LVDS_LFP_DATA); + if (!lvds_lfp_data) + return; + + lvds_lfp_data_ptrs = find_section(bdb, BDB_LVDS_LFP_DATA_PTRS); + if (!lvds_lfp_data_ptrs) + return; + + dev_priv->lvds_vbt = 1; + + lfp_data_size = lvds_lfp_data_ptrs->ptr[1].dvo_timing_offset - + lvds_lfp_data_ptrs->ptr[0].dvo_timing_offset; + entry = (struct bdb_lvds_lfp_data_entry *) + ((uint8_t *)lvds_lfp_data->data + (lfp_data_size * + lvds_options->panel_type)); + dvo_timing_offset = lvds_lfp_data_ptrs->ptr[0].dvo_timing_offset - + lvds_lfp_data_ptrs->ptr[0].fp_timing_offset; + + /* + * the size of fp_timing varies on the different platform. + * So calculate the DVO timing relative offset in LVDS data + * entry to get the DVO timing entry + */ + dvo_timing = (struct lvds_dvo_timing *) + ((unsigned char *)entry + dvo_timing_offset); + + panel_fixed_mode = kzalloc(sizeof(*panel_fixed_mode), GFP_KERNEL); + if (!panel_fixed_mode) + return; + + fill_detail_timing_data(panel_fixed_mode, dvo_timing); + + dev_priv->lfp_lvds_vbt_mode = panel_fixed_mode; + + DRM_DEBUG_KMS("Found panel mode in BIOS VBT tables:\n"); + drm_mode_debug_printmodeline(panel_fixed_mode); + + temp_mode = kzalloc(sizeof(*temp_mode), GFP_KERNEL); + temp_downclock = panel_fixed_mode->clock; + /* + * enumerate the LVDS panel timing info entry in VBT to check whether + * the LVDS downclock is found. + */ + for (i = 0; i < 16; i++) { + entry = (struct bdb_lvds_lfp_data_entry *) + ((uint8_t *)lvds_lfp_data->data + (lfp_data_size * i)); + dvo_timing = (struct lvds_dvo_timing *) + ((unsigned char *)entry + dvo_timing_offset); + + fill_detail_timing_data(temp_mode, dvo_timing); + + if (temp_mode->hdisplay == panel_fixed_mode->hdisplay && + temp_mode->hsync_start == panel_fixed_mode->hsync_start && + temp_mode->hsync_end == panel_fixed_mode->hsync_end && + temp_mode->htotal == panel_fixed_mode->htotal && + temp_mode->vdisplay == panel_fixed_mode->vdisplay && + temp_mode->vsync_start == panel_fixed_mode->vsync_start && + temp_mode->vsync_end == panel_fixed_mode->vsync_end && + temp_mode->vtotal == panel_fixed_mode->vtotal && + temp_mode->clock < temp_downclock) { + /* + * downclock is already found. But we expect + * to find the lower downclock. + */ + temp_downclock = temp_mode->clock; + } + /* clear it to zero */ + memset(temp_mode, 0, sizeof(*temp_mode)); + } + kfree(temp_mode); + if (temp_downclock < panel_fixed_mode->clock && + i915_lvds_downclock) { + dev_priv->lvds_downclock_avail = 1; + dev_priv->lvds_downclock = temp_downclock; + DRM_DEBUG_KMS("LVDS downclock is found in VBT. " + "Normal Clock %dKHz, downclock %dKHz\n", + temp_downclock, panel_fixed_mode->clock); + } + return; +} + +/* Try to find sdvo panel data */ +static void +parse_sdvo_panel_data(struct drm_i915_private *dev_priv, + struct bdb_header *bdb) +{ + struct lvds_dvo_timing *dvo_timing; + struct drm_display_mode *panel_fixed_mode; + int index; + + index = i915_vbt_sdvo_panel_type; + if (index == -1) { + struct bdb_sdvo_lvds_options *sdvo_lvds_options; + + sdvo_lvds_options = find_section(bdb, BDB_SDVO_LVDS_OPTIONS); + if (!sdvo_lvds_options) + return; + + index = sdvo_lvds_options->panel_type; + } + + dvo_timing = find_section(bdb, BDB_SDVO_PANEL_DTDS); + if (!dvo_timing) + return; + + panel_fixed_mode = kzalloc(sizeof(*panel_fixed_mode), GFP_KERNEL); + if (!panel_fixed_mode) + return; + + fill_detail_timing_data(panel_fixed_mode, dvo_timing + index); + + dev_priv->sdvo_lvds_vbt_mode = panel_fixed_mode; + + DRM_DEBUG_KMS("Found SDVO panel mode in BIOS VBT tables:\n"); + drm_mode_debug_printmodeline(panel_fixed_mode); +} + +static int intel_bios_ssc_frequency(struct drm_device *dev, + bool alternate) +{ + switch (INTEL_INFO(dev)->gen) { + case 2: + return alternate ? 66 : 48; + case 3: + case 4: + return alternate ? 100 : 96; + default: + return alternate ? 100 : 120; + } +} + +static void +parse_general_features(struct drm_i915_private *dev_priv, + struct bdb_header *bdb) +{ + struct drm_device *dev = dev_priv->dev; + struct bdb_general_features *general; + + general = find_section(bdb, BDB_GENERAL_FEATURES); + if (general) { + dev_priv->int_tv_support = general->int_tv_support; + dev_priv->int_crt_support = general->int_crt_support; + dev_priv->lvds_use_ssc = general->enable_ssc; + dev_priv->lvds_ssc_freq = + intel_bios_ssc_frequency(dev, general->ssc_freq); + } +} + +static void +parse_general_definitions(struct drm_i915_private *dev_priv, + struct bdb_header *bdb) +{ + struct bdb_general_definitions *general; + + general = find_section(bdb, BDB_GENERAL_DEFINITIONS); + if (general) { + u16 block_size = get_blocksize(general); + if (block_size >= sizeof(*general)) { + int bus_pin = general->crt_ddc_gmbus_pin; + DRM_DEBUG_KMS("crt_ddc_bus_pin: %d\n", bus_pin); + if (bus_pin >= 1 && bus_pin <= 6) + dev_priv->crt_ddc_pin = bus_pin; + } else { + DRM_DEBUG_KMS("BDB_GD too small (%d). Invalid.\n", + block_size); + } + } +} + +static void +parse_sdvo_device_mapping(struct drm_i915_private *dev_priv, + struct bdb_header *bdb) +{ + struct sdvo_device_mapping *p_mapping; + struct bdb_general_definitions *p_defs; + struct child_device_config *p_child; + int i, child_device_num, count; + u16 block_size; + + p_defs = find_section(bdb, BDB_GENERAL_DEFINITIONS); + if (!p_defs) { + DRM_DEBUG_KMS("No general definition block is found, unable to construct sdvo mapping.\n"); + return; + } + /* judge whether the size of child device meets the requirements. + * If the child device size obtained from general definition block + * is different with sizeof(struct child_device_config), skip the + * parsing of sdvo device info + */ + if (p_defs->child_dev_size != sizeof(*p_child)) { + /* different child dev size . Ignore it */ + DRM_DEBUG_KMS("different child size is found. Invalid.\n"); + return; + } + /* get the block size of general definitions */ + block_size = get_blocksize(p_defs); + /* get the number of child device */ + child_device_num = (block_size - sizeof(*p_defs)) / + sizeof(*p_child); + count = 0; + for (i = 0; i < child_device_num; i++) { + p_child = &(p_defs->devices[i]); + if (!p_child->device_type) { + /* skip the device block if device type is invalid */ + continue; + } + if (p_child->slave_addr != SLAVE_ADDR1 && + p_child->slave_addr != SLAVE_ADDR2) { + /* + * If the slave address is neither 0x70 nor 0x72, + * it is not a SDVO device. Skip it. + */ + continue; + } + if (p_child->dvo_port != DEVICE_PORT_DVOB && + p_child->dvo_port != DEVICE_PORT_DVOC) { + /* skip the incorrect SDVO port */ + DRM_DEBUG_KMS("Incorrect SDVO port. Skip it \n"); + continue; + } + DRM_DEBUG_KMS("the SDVO device with slave addr %2x is found on" + " %s port\n", + p_child->slave_addr, + (p_child->dvo_port == DEVICE_PORT_DVOB) ? + "SDVOB" : "SDVOC"); + p_mapping = &(dev_priv->sdvo_mappings[p_child->dvo_port - 1]); + if (!p_mapping->initialized) { + p_mapping->dvo_port = p_child->dvo_port; + p_mapping->slave_addr = p_child->slave_addr; + p_mapping->dvo_wiring = p_child->dvo_wiring; + p_mapping->ddc_pin = p_child->ddc_pin; + p_mapping->i2c_pin = p_child->i2c_pin; + p_mapping->i2c_speed = p_child->i2c_speed; + p_mapping->initialized = 1; + DRM_DEBUG_KMS("SDVO device: dvo=%x, addr=%x, wiring=%d, ddc_pin=%d, i2c_pin=%d, i2c_speed=%d\n", + p_mapping->dvo_port, + p_mapping->slave_addr, + p_mapping->dvo_wiring, + p_mapping->ddc_pin, + p_mapping->i2c_pin, + p_mapping->i2c_speed); + } else { + DRM_DEBUG_KMS("Maybe one SDVO port is shared by " + "two SDVO device.\n"); + } + if (p_child->slave2_addr) { + /* Maybe this is a SDVO device with multiple inputs */ + /* And the mapping info is not added */ + DRM_DEBUG_KMS("there exists the slave2_addr. Maybe this" + " is a SDVO device with multiple inputs.\n"); + } + count++; + } + + if (!count) { + /* No SDVO device info is found */ + DRM_DEBUG_KMS("No SDVO device info is found in VBT\n"); + } + return; +} + +static void +parse_driver_features(struct drm_i915_private *dev_priv, + struct bdb_header *bdb) +{ + struct drm_device *dev = dev_priv->dev; + struct bdb_driver_features *driver; + + driver = find_section(bdb, BDB_DRIVER_FEATURES); + if (!driver) + return; + + if (SUPPORTS_EDP(dev) && + driver->lvds_config == BDB_DRIVER_FEATURE_EDP) + dev_priv->edp.support = 1; + + if (driver->dual_frequency) + dev_priv->render_reclock_avail = true; +} + +static void +parse_edp(struct drm_i915_private *dev_priv, struct bdb_header *bdb) +{ + struct bdb_edp *edp; + struct edp_power_seq *edp_pps; + struct edp_link_params *edp_link_params; + + edp = find_section(bdb, BDB_EDP); + if (!edp) { + if (SUPPORTS_EDP(dev_priv->dev) && dev_priv->edp.support) { + DRM_DEBUG_KMS("No eDP BDB found but eDP panel " + "supported, assume %dbpp panel color " + "depth.\n", + dev_priv->edp.bpp); + } + return; + } + + switch ((edp->color_depth >> (panel_type * 2)) & 3) { + case EDP_18BPP: + dev_priv->edp.bpp = 18; + break; + case EDP_24BPP: + dev_priv->edp.bpp = 24; + break; + case EDP_30BPP: + dev_priv->edp.bpp = 30; + break; + } + + /* Get the eDP sequencing and link info */ + edp_pps = &edp->power_seqs[panel_type]; + edp_link_params = &edp->link_params[panel_type]; + + dev_priv->edp.pps = *edp_pps; + + dev_priv->edp.rate = edp_link_params->rate ? DP_LINK_BW_2_7 : + DP_LINK_BW_1_62; + switch (edp_link_params->lanes) { + case 0: + dev_priv->edp.lanes = 1; + break; + case 1: + dev_priv->edp.lanes = 2; + break; + case 3: + default: + dev_priv->edp.lanes = 4; + break; + } + switch (edp_link_params->preemphasis) { + case 0: + dev_priv->edp.preemphasis = DP_TRAIN_PRE_EMPHASIS_0; + break; + case 1: + dev_priv->edp.preemphasis = DP_TRAIN_PRE_EMPHASIS_3_5; + break; + case 2: + dev_priv->edp.preemphasis = DP_TRAIN_PRE_EMPHASIS_6; + break; + case 3: + dev_priv->edp.preemphasis = DP_TRAIN_PRE_EMPHASIS_9_5; + break; + } + switch (edp_link_params->vswing) { + case 0: + dev_priv->edp.vswing = DP_TRAIN_VOLTAGE_SWING_400; + break; + case 1: + dev_priv->edp.vswing = DP_TRAIN_VOLTAGE_SWING_600; + break; + case 2: + dev_priv->edp.vswing = DP_TRAIN_VOLTAGE_SWING_800; + break; + case 3: + dev_priv->edp.vswing = DP_TRAIN_VOLTAGE_SWING_1200; + break; + } +} + +static void +parse_device_mapping(struct drm_i915_private *dev_priv, + struct bdb_header *bdb) +{ + struct bdb_general_definitions *p_defs; + struct child_device_config *p_child, *child_dev_ptr; + int i, child_device_num, count; + u16 block_size; + + p_defs = find_section(bdb, BDB_GENERAL_DEFINITIONS); + if (!p_defs) { + DRM_DEBUG_KMS("No general definition block is found, no devices defined.\n"); + return; + } + /* judge whether the size of child device meets the requirements. + * If the child device size obtained from general definition block + * is different with sizeof(struct child_device_config), skip the + * parsing of sdvo device info + */ + if (p_defs->child_dev_size != sizeof(*p_child)) { + /* different child dev size . Ignore it */ + DRM_DEBUG_KMS("different child size is found. Invalid.\n"); + return; + } + /* get the block size of general definitions */ + block_size = get_blocksize(p_defs); + /* get the number of child device */ + child_device_num = (block_size - sizeof(*p_defs)) / + sizeof(*p_child); + count = 0; + /* get the number of child device that is present */ + for (i = 0; i < child_device_num; i++) { + p_child = &(p_defs->devices[i]); + if (!p_child->device_type) { + /* skip the device block if device type is invalid */ + continue; + } + count++; + } + if (!count) { + DRM_DEBUG_KMS("no child dev is parsed from VBT \n"); + return; + } + dev_priv->child_dev = kzalloc(sizeof(*p_child) * count, GFP_KERNEL); + if (!dev_priv->child_dev) { + DRM_DEBUG_KMS("No memory space for child device\n"); + return; + } + + dev_priv->child_dev_num = count; + count = 0; + for (i = 0; i < child_device_num; i++) { + p_child = &(p_defs->devices[i]); + if (!p_child->device_type) { + /* skip the device block if device type is invalid */ + continue; + } + child_dev_ptr = dev_priv->child_dev + count; + count++; + memcpy((void *)child_dev_ptr, (void *)p_child, + sizeof(*p_child)); + } + return; +} + +static void +init_vbt_defaults(struct drm_i915_private *dev_priv) +{ + struct drm_device *dev = dev_priv->dev; + + dev_priv->crt_ddc_pin = GMBUS_PORT_VGADDC; + + /* LFP panel data */ + dev_priv->lvds_dither = 1; + dev_priv->lvds_vbt = 0; + + /* SDVO panel data */ + dev_priv->sdvo_lvds_vbt_mode = NULL; + + /* general features */ + dev_priv->int_tv_support = 1; + dev_priv->int_crt_support = 1; + + /* Default to using SSC */ + dev_priv->lvds_use_ssc = 1; + dev_priv->lvds_ssc_freq = intel_bios_ssc_frequency(dev, 1); + DRM_DEBUG("Set default to SSC at %dMHz\n", dev_priv->lvds_ssc_freq); + + /* eDP data */ + dev_priv->edp.bpp = 18; +} + +static int __init intel_no_opregion_vbt_callback(const struct dmi_system_id *id) +{ + DRM_DEBUG_KMS("Falling back to manually reading VBT from " + "VBIOS ROM for %s\n", + id->ident); + return 1; +} + +static const struct dmi_system_id intel_no_opregion_vbt[] = { + { + .callback = intel_no_opregion_vbt_callback, + .ident = "ThinkCentre A57", + .matches = { + DMI_MATCH(DMI_SYS_VENDOR, "LENOVO"), + DMI_MATCH(DMI_PRODUCT_NAME, "97027RG"), + }, + }, + { } +}; + +/** + * intel_parse_bios - find VBT and initialize settings from the BIOS + * @dev: DRM device + * + * Loads the Video BIOS and checks that the VBT exists. Sets scratch registers + * to appropriate values. + * + * Returns 0 on success, nonzero on failure. + */ +bool +intel_parse_bios(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct pci_dev *pdev = dev->pdev; + struct bdb_header *bdb = NULL; + u8 __iomem *bios = NULL; + + init_vbt_defaults(dev_priv); + + /* XXX Should this validation be moved to intel_opregion.c? */ + if (!dmi_check_system(intel_no_opregion_vbt) && dev_priv->opregion.vbt) { + struct vbt_header *vbt = dev_priv->opregion.vbt; + if (memcmp(vbt->signature, "$VBT", 4) == 0) { + DRM_DEBUG_DRIVER("Using VBT from OpRegion: %20s\n", + vbt->signature); + bdb = (struct bdb_header *)((char *)vbt + vbt->bdb_offset); + } else + dev_priv->opregion.vbt = NULL; + } + + if (bdb == NULL) { + struct vbt_header *vbt = NULL; + size_t size; + int i; + + bios = pci_map_rom(pdev, &size); + if (!bios) + return -1; + + /* Scour memory looking for the VBT signature */ + for (i = 0; i + 4 < size; i++) { + if (!memcmp(bios + i, "$VBT", 4)) { + vbt = (struct vbt_header *)(bios + i); + break; + } + } + + if (!vbt) { + DRM_ERROR("VBT signature missing\n"); + pci_unmap_rom(pdev, bios); + return -1; + } + + bdb = (struct bdb_header *)(bios + i + vbt->bdb_offset); + } + + /* Grab useful general definitions */ + parse_general_features(dev_priv, bdb); + parse_general_definitions(dev_priv, bdb); + parse_lfp_panel_data(dev_priv, bdb); + parse_sdvo_panel_data(dev_priv, bdb); + parse_sdvo_device_mapping(dev_priv, bdb); + parse_device_mapping(dev_priv, bdb); + parse_driver_features(dev_priv, bdb); + parse_edp(dev_priv, bdb); + + if (bios) + pci_unmap_rom(pdev, bios); + + return 0; +} + +/* Ensure that vital registers have been initialised, even if the BIOS + * is absent or just failing to do its job. + */ +void intel_setup_bios(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + /* Set the Panel Power On/Off timings if uninitialized. */ + if ((I915_READ(PP_ON_DELAYS) == 0) && (I915_READ(PP_OFF_DELAYS) == 0)) { + /* Set T2 to 40ms and T5 to 200ms */ + I915_WRITE(PP_ON_DELAYS, 0x019007d0); + + /* Set T3 to 35ms and Tx to 200ms */ + I915_WRITE(PP_OFF_DELAYS, 0x015e07d0); + } +} diff --git a/drivers/gpu/drm/i915/intel_bios.h b/drivers/gpu/drm/i915/intel_bios.h new file mode 100644 index 00000000..5f8e4edc --- /dev/null +++ b/drivers/gpu/drm/i915/intel_bios.h @@ -0,0 +1,611 @@ +/* + * Copyright © 2006 Intel Corporation + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + * + * Authors: + * Eric Anholt + * + */ + +#ifndef _I830_BIOS_H_ +#define _I830_BIOS_H_ + +#include "drmP.h" + +struct vbt_header { + u8 signature[20]; /**< Always starts with 'VBT$' */ + u16 version; /**< decimal */ + u16 header_size; /**< in bytes */ + u16 vbt_size; /**< in bytes */ + u8 vbt_checksum; + u8 reserved0; + u32 bdb_offset; /**< from beginning of VBT */ + u32 aim_offset[4]; /**< from beginning of VBT */ +} __attribute__((packed)); + +struct bdb_header { + u8 signature[16]; /**< Always 'BIOS_DATA_BLOCK' */ + u16 version; /**< decimal */ + u16 header_size; /**< in bytes */ + u16 bdb_size; /**< in bytes */ +}; + +/* strictly speaking, this is a "skip" block, but it has interesting info */ +struct vbios_data { + u8 type; /* 0 == desktop, 1 == mobile */ + u8 relstage; + u8 chipset; + u8 lvds_present:1; + u8 tv_present:1; + u8 rsvd2:6; /* finish byte */ + u8 rsvd3[4]; + u8 signon[155]; + u8 copyright[61]; + u16 code_segment; + u8 dos_boot_mode; + u8 bandwidth_percent; + u8 rsvd4; /* popup memory size */ + u8 resize_pci_bios; + u8 rsvd5; /* is crt already on ddc2 */ +} __attribute__((packed)); + +/* + * There are several types of BIOS data blocks (BDBs), each block has + * an ID and size in the first 3 bytes (ID in first, size in next 2). + * Known types are listed below. + */ +#define BDB_GENERAL_FEATURES 1 +#define BDB_GENERAL_DEFINITIONS 2 +#define BDB_OLD_TOGGLE_LIST 3 +#define BDB_MODE_SUPPORT_LIST 4 +#define BDB_GENERIC_MODE_TABLE 5 +#define BDB_EXT_MMIO_REGS 6 +#define BDB_SWF_IO 7 +#define BDB_SWF_MMIO 8 +#define BDB_DOT_CLOCK_TABLE 9 +#define BDB_MODE_REMOVAL_TABLE 10 +#define BDB_CHILD_DEVICE_TABLE 11 +#define BDB_DRIVER_FEATURES 12 +#define BDB_DRIVER_PERSISTENCE 13 +#define BDB_EXT_TABLE_PTRS 14 +#define BDB_DOT_CLOCK_OVERRIDE 15 +#define BDB_DISPLAY_SELECT 16 +/* 17 rsvd */ +#define BDB_DRIVER_ROTATION 18 +#define BDB_DISPLAY_REMOVE 19 +#define BDB_OEM_CUSTOM 20 +#define BDB_EFP_LIST 21 /* workarounds for VGA hsync/vsync */ +#define BDB_SDVO_LVDS_OPTIONS 22 +#define BDB_SDVO_PANEL_DTDS 23 +#define BDB_SDVO_LVDS_PNP_IDS 24 +#define BDB_SDVO_LVDS_POWER_SEQ 25 +#define BDB_TV_OPTIONS 26 +#define BDB_EDP 27 +#define BDB_LVDS_OPTIONS 40 +#define BDB_LVDS_LFP_DATA_PTRS 41 +#define BDB_LVDS_LFP_DATA 42 +#define BDB_LVDS_BACKLIGHT 43 +#define BDB_LVDS_POWER 44 +#define BDB_SKIP 254 /* VBIOS private block, ignore */ + +struct bdb_general_features { + /* bits 1 */ + u8 panel_fitting:2; + u8 flexaim:1; + u8 msg_enable:1; + u8 clear_screen:3; + u8 color_flip:1; + + /* bits 2 */ + u8 download_ext_vbt:1; + u8 enable_ssc:1; + u8 ssc_freq:1; + u8 enable_lfp_on_override:1; + u8 disable_ssc_ddt:1; + u8 rsvd8:3; /* finish byte */ + + /* bits 3 */ + u8 disable_smooth_vision:1; + u8 single_dvi:1; + u8 rsvd9:6; /* finish byte */ + + /* bits 4 */ + u8 legacy_monitor_detect; + + /* bits 5 */ + u8 int_crt_support:1; + u8 int_tv_support:1; + u8 rsvd11:6; /* finish byte */ +} __attribute__((packed)); + +/* pre-915 */ +#define GPIO_PIN_DVI_LVDS 0x03 /* "DVI/LVDS DDC GPIO pins" */ +#define GPIO_PIN_ADD_I2C 0x05 /* "ADDCARD I2C GPIO pins" */ +#define GPIO_PIN_ADD_DDC 0x04 /* "ADDCARD DDC GPIO pins" */ +#define GPIO_PIN_ADD_DDC_I2C 0x06 /* "ADDCARD DDC/I2C GPIO pins" */ + +/* Pre 915 */ +#define DEVICE_TYPE_NONE 0x00 +#define DEVICE_TYPE_CRT 0x01 +#define DEVICE_TYPE_TV 0x09 +#define DEVICE_TYPE_EFP 0x12 +#define DEVICE_TYPE_LFP 0x22 +/* On 915+ */ +#define DEVICE_TYPE_CRT_DPMS 0x6001 +#define DEVICE_TYPE_CRT_DPMS_HOTPLUG 0x4001 +#define DEVICE_TYPE_TV_COMPOSITE 0x0209 +#define DEVICE_TYPE_TV_MACROVISION 0x0289 +#define DEVICE_TYPE_TV_RF_COMPOSITE 0x020c +#define DEVICE_TYPE_TV_SVIDEO_COMPOSITE 0x0609 +#define DEVICE_TYPE_TV_SCART 0x0209 +#define DEVICE_TYPE_TV_CODEC_HOTPLUG_PWR 0x6009 +#define DEVICE_TYPE_EFP_HOTPLUG_PWR 0x6012 +#define DEVICE_TYPE_EFP_DVI_HOTPLUG_PWR 0x6052 +#define DEVICE_TYPE_EFP_DVI_I 0x6053 +#define DEVICE_TYPE_EFP_DVI_D_DUAL 0x6152 +#define DEVICE_TYPE_EFP_DVI_D_HDCP 0x60d2 +#define DEVICE_TYPE_OPENLDI_HOTPLUG_PWR 0x6062 +#define DEVICE_TYPE_OPENLDI_DUALPIX 0x6162 +#define DEVICE_TYPE_LFP_PANELLINK 0x5012 +#define DEVICE_TYPE_LFP_CMOS_PWR 0x5042 +#define DEVICE_TYPE_LFP_LVDS_PWR 0x5062 +#define DEVICE_TYPE_LFP_LVDS_DUAL 0x5162 +#define DEVICE_TYPE_LFP_LVDS_DUAL_HDCP 0x51e2 + +#define DEVICE_CFG_NONE 0x00 +#define DEVICE_CFG_12BIT_DVOB 0x01 +#define DEVICE_CFG_12BIT_DVOC 0x02 +#define DEVICE_CFG_24BIT_DVOBC 0x09 +#define DEVICE_CFG_24BIT_DVOCB 0x0a +#define DEVICE_CFG_DUAL_DVOB 0x11 +#define DEVICE_CFG_DUAL_DVOC 0x12 +#define DEVICE_CFG_DUAL_DVOBC 0x13 +#define DEVICE_CFG_DUAL_LINK_DVOBC 0x19 +#define DEVICE_CFG_DUAL_LINK_DVOCB 0x1a + +#define DEVICE_WIRE_NONE 0x00 +#define DEVICE_WIRE_DVOB 0x01 +#define DEVICE_WIRE_DVOC 0x02 +#define DEVICE_WIRE_DVOBC 0x03 +#define DEVICE_WIRE_DVOBB 0x05 +#define DEVICE_WIRE_DVOCC 0x06 +#define DEVICE_WIRE_DVOB_MASTER 0x0d +#define DEVICE_WIRE_DVOC_MASTER 0x0e + +#define DEVICE_PORT_DVOA 0x00 /* none on 845+ */ +#define DEVICE_PORT_DVOB 0x01 +#define DEVICE_PORT_DVOC 0x02 + +struct child_device_config { + u16 handle; + u16 device_type; + u8 i2c_speed; + u8 rsvd[9]; + u16 addin_offset; + u8 dvo_port; /* See Device_PORT_* above */ + u8 i2c_pin; + u8 slave_addr; + u8 ddc_pin; + u16 edid_ptr; + u8 dvo_cfg; /* See DEVICE_CFG_* above */ + u8 dvo2_port; + u8 i2c2_pin; + u8 slave2_addr; + u8 ddc2_pin; + u8 capabilities; + u8 dvo_wiring;/* See DEVICE_WIRE_* above */ + u8 dvo2_wiring; + u16 extended_type; + u8 dvo_function; +} __attribute__((packed)); + +struct bdb_general_definitions { + /* DDC GPIO */ + u8 crt_ddc_gmbus_pin; + + /* DPMS bits */ + u8 dpms_acpi:1; + u8 skip_boot_crt_detect:1; + u8 dpms_aim:1; + u8 rsvd1:5; /* finish byte */ + + /* boot device bits */ + u8 boot_display[2]; + u8 child_dev_size; + + /* + * Device info: + * If TV is present, it'll be at devices[0]. + * LVDS will be next, either devices[0] or [1], if present. + * On some platforms the number of device is 6. But could be as few as + * 4 if both TV and LVDS are missing. + * And the device num is related with the size of general definition + * block. It is obtained by using the following formula: + * number = (block_size - sizeof(bdb_general_definitions))/ + * sizeof(child_device_config); + */ + struct child_device_config devices[0]; +} __attribute__((packed)); + +struct bdb_lvds_options { + u8 panel_type; + u8 rsvd1; + /* LVDS capabilities, stored in a dword */ + u8 pfit_mode:2; + u8 pfit_text_mode_enhanced:1; + u8 pfit_gfx_mode_enhanced:1; + u8 pfit_ratio_auto:1; + u8 pixel_dither:1; + u8 lvds_edid:1; + u8 rsvd2:1; + u8 rsvd4; +} __attribute__((packed)); + +/* LFP pointer table contains entries to the struct below */ +struct bdb_lvds_lfp_data_ptr { + u16 fp_timing_offset; /* offsets are from start of bdb */ + u8 fp_table_size; + u16 dvo_timing_offset; + u8 dvo_table_size; + u16 panel_pnp_id_offset; + u8 pnp_table_size; +} __attribute__((packed)); + +struct bdb_lvds_lfp_data_ptrs { + u8 lvds_entries; /* followed by one or more lvds_data_ptr structs */ + struct bdb_lvds_lfp_data_ptr ptr[16]; +} __attribute__((packed)); + +/* LFP data has 3 blocks per entry */ +struct lvds_fp_timing { + u16 x_res; + u16 y_res; + u32 lvds_reg; + u32 lvds_reg_val; + u32 pp_on_reg; + u32 pp_on_reg_val; + u32 pp_off_reg; + u32 pp_off_reg_val; + u32 pp_cycle_reg; + u32 pp_cycle_reg_val; + u32 pfit_reg; + u32 pfit_reg_val; + u16 terminator; +} __attribute__((packed)); + +struct lvds_dvo_timing { + u16 clock; /**< In 10khz */ + u8 hactive_lo; + u8 hblank_lo; + u8 hblank_hi:4; + u8 hactive_hi:4; + u8 vactive_lo; + u8 vblank_lo; + u8 vblank_hi:4; + u8 vactive_hi:4; + u8 hsync_off_lo; + u8 hsync_pulse_width; + u8 vsync_pulse_width:4; + u8 vsync_off:4; + u8 rsvd0:6; + u8 hsync_off_hi:2; + u8 h_image; + u8 v_image; + u8 max_hv; + u8 h_border; + u8 v_border; + u8 rsvd1:3; + u8 digital:2; + u8 vsync_positive:1; + u8 hsync_positive:1; + u8 rsvd2:1; +} __attribute__((packed)); + +struct lvds_pnp_id { + u16 mfg_name; + u16 product_code; + u32 serial; + u8 mfg_week; + u8 mfg_year; +} __attribute__((packed)); + +struct bdb_lvds_lfp_data_entry { + struct lvds_fp_timing fp_timing; + struct lvds_dvo_timing dvo_timing; + struct lvds_pnp_id pnp_id; +} __attribute__((packed)); + +struct bdb_lvds_lfp_data { + struct bdb_lvds_lfp_data_entry data[16]; +} __attribute__((packed)); + +struct aimdb_header { + char signature[16]; + char oem_device[20]; + u16 aimdb_version; + u16 aimdb_header_size; + u16 aimdb_size; +} __attribute__((packed)); + +struct aimdb_block { + u8 aimdb_id; + u16 aimdb_size; +} __attribute__((packed)); + +struct vch_panel_data { + u16 fp_timing_offset; + u8 fp_timing_size; + u16 dvo_timing_offset; + u8 dvo_timing_size; + u16 text_fitting_offset; + u8 text_fitting_size; + u16 graphics_fitting_offset; + u8 graphics_fitting_size; +} __attribute__((packed)); + +struct vch_bdb_22 { + struct aimdb_block aimdb_block; + struct vch_panel_data panels[16]; +} __attribute__((packed)); + +struct bdb_sdvo_lvds_options { + u8 panel_backlight; + u8 h40_set_panel_type; + u8 panel_type; + u8 ssc_clk_freq; + u16 als_low_trip; + u16 als_high_trip; + u8 sclalarcoeff_tab_row_num; + u8 sclalarcoeff_tab_row_size; + u8 coefficient[8]; + u8 panel_misc_bits_1; + u8 panel_misc_bits_2; + u8 panel_misc_bits_3; + u8 panel_misc_bits_4; +} __attribute__((packed)); + + +#define BDB_DRIVER_FEATURE_NO_LVDS 0 +#define BDB_DRIVER_FEATURE_INT_LVDS 1 +#define BDB_DRIVER_FEATURE_SDVO_LVDS 2 +#define BDB_DRIVER_FEATURE_EDP 3 + +struct bdb_driver_features { + u8 boot_dev_algorithm:1; + u8 block_display_switch:1; + u8 allow_display_switch:1; + u8 hotplug_dvo:1; + u8 dual_view_zoom:1; + u8 int15h_hook:1; + u8 sprite_in_clone:1; + u8 primary_lfp_id:1; + + u16 boot_mode_x; + u16 boot_mode_y; + u8 boot_mode_bpp; + u8 boot_mode_refresh; + + u16 enable_lfp_primary:1; + u16 selective_mode_pruning:1; + u16 dual_frequency:1; + u16 render_clock_freq:1; /* 0: high freq; 1: low freq */ + u16 nt_clone_support:1; + u16 power_scheme_ui:1; /* 0: CUI; 1: 3rd party */ + u16 sprite_display_assign:1; /* 0: secondary; 1: primary */ + u16 cui_aspect_scaling:1; + u16 preserve_aspect_ratio:1; + u16 sdvo_device_power_down:1; + u16 crt_hotplug:1; + u16 lvds_config:2; + u16 tv_hotplug:1; + u16 hdmi_config:2; + + u8 static_display:1; + u8 reserved2:7; + u16 legacy_crt_max_x; + u16 legacy_crt_max_y; + u8 legacy_crt_max_refresh; + + u8 hdmi_termination; + u8 custom_vbt_version; +} __attribute__((packed)); + +#define EDP_18BPP 0 +#define EDP_24BPP 1 +#define EDP_30BPP 2 +#define EDP_RATE_1_62 0 +#define EDP_RATE_2_7 1 +#define EDP_LANE_1 0 +#define EDP_LANE_2 1 +#define EDP_LANE_4 3 +#define EDP_PREEMPHASIS_NONE 0 +#define EDP_PREEMPHASIS_3_5dB 1 +#define EDP_PREEMPHASIS_6dB 2 +#define EDP_PREEMPHASIS_9_5dB 3 +#define EDP_VSWING_0_4V 0 +#define EDP_VSWING_0_6V 1 +#define EDP_VSWING_0_8V 2 +#define EDP_VSWING_1_2V 3 + +struct edp_power_seq { + u16 t3; + u16 t7; + u16 t9; + u16 t10; + u16 t12; +} __attribute__ ((packed)); + +struct edp_link_params { + u8 rate:4; + u8 lanes:4; + u8 preemphasis:4; + u8 vswing:4; +} __attribute__ ((packed)); + +struct bdb_edp { + struct edp_power_seq power_seqs[16]; + u32 color_depth; + u32 sdrrs_msa_timing_delay; + struct edp_link_params link_params[16]; +} __attribute__ ((packed)); + +void intel_setup_bios(struct drm_device *dev); +bool intel_parse_bios(struct drm_device *dev); + +/* + * Driver<->VBIOS interaction occurs through scratch bits in + * GR18 & SWF*. + */ + +/* GR18 bits are set on display switch and hotkey events */ +#define GR18_DRIVER_SWITCH_EN (1<<7) /* 0: VBIOS control, 1: driver control */ +#define GR18_HOTKEY_MASK 0x78 /* See also SWF4 15:0 */ +#define GR18_HK_NONE (0x0<<3) +#define GR18_HK_LFP_STRETCH (0x1<<3) +#define GR18_HK_TOGGLE_DISP (0x2<<3) +#define GR18_HK_DISP_SWITCH (0x4<<3) /* see SWF14 15:0 for what to enable */ +#define GR18_HK_POPUP_DISABLED (0x6<<3) +#define GR18_HK_POPUP_ENABLED (0x7<<3) +#define GR18_HK_PFIT (0x8<<3) +#define GR18_HK_APM_CHANGE (0xa<<3) +#define GR18_HK_MULTIPLE (0xc<<3) +#define GR18_USER_INT_EN (1<<2) +#define GR18_A0000_FLUSH_EN (1<<1) +#define GR18_SMM_EN (1<<0) + +/* Set by driver, cleared by VBIOS */ +#define SWF00_YRES_SHIFT 16 +#define SWF00_XRES_SHIFT 0 +#define SWF00_RES_MASK 0xffff + +/* Set by VBIOS at boot time and driver at runtime */ +#define SWF01_TV2_FORMAT_SHIFT 8 +#define SWF01_TV1_FORMAT_SHIFT 0 +#define SWF01_TV_FORMAT_MASK 0xffff + +#define SWF10_VBIOS_BLC_I2C_EN (1<<29) +#define SWF10_GTT_OVERRIDE_EN (1<<28) +#define SWF10_LFP_DPMS_OVR (1<<27) /* override DPMS on display switch */ +#define SWF10_ACTIVE_TOGGLE_LIST_MASK (7<<24) +#define SWF10_OLD_TOGGLE 0x0 +#define SWF10_TOGGLE_LIST_1 0x1 +#define SWF10_TOGGLE_LIST_2 0x2 +#define SWF10_TOGGLE_LIST_3 0x3 +#define SWF10_TOGGLE_LIST_4 0x4 +#define SWF10_PANNING_EN (1<<23) +#define SWF10_DRIVER_LOADED (1<<22) +#define SWF10_EXTENDED_DESKTOP (1<<21) +#define SWF10_EXCLUSIVE_MODE (1<<20) +#define SWF10_OVERLAY_EN (1<<19) +#define SWF10_PLANEB_HOLDOFF (1<<18) +#define SWF10_PLANEA_HOLDOFF (1<<17) +#define SWF10_VGA_HOLDOFF (1<<16) +#define SWF10_ACTIVE_DISP_MASK 0xffff +#define SWF10_PIPEB_LFP2 (1<<15) +#define SWF10_PIPEB_EFP2 (1<<14) +#define SWF10_PIPEB_TV2 (1<<13) +#define SWF10_PIPEB_CRT2 (1<<12) +#define SWF10_PIPEB_LFP (1<<11) +#define SWF10_PIPEB_EFP (1<<10) +#define SWF10_PIPEB_TV (1<<9) +#define SWF10_PIPEB_CRT (1<<8) +#define SWF10_PIPEA_LFP2 (1<<7) +#define SWF10_PIPEA_EFP2 (1<<6) +#define SWF10_PIPEA_TV2 (1<<5) +#define SWF10_PIPEA_CRT2 (1<<4) +#define SWF10_PIPEA_LFP (1<<3) +#define SWF10_PIPEA_EFP (1<<2) +#define SWF10_PIPEA_TV (1<<1) +#define SWF10_PIPEA_CRT (1<<0) + +#define SWF11_MEMORY_SIZE_SHIFT 16 +#define SWF11_SV_TEST_EN (1<<15) +#define SWF11_IS_AGP (1<<14) +#define SWF11_DISPLAY_HOLDOFF (1<<13) +#define SWF11_DPMS_REDUCED (1<<12) +#define SWF11_IS_VBE_MODE (1<<11) +#define SWF11_PIPEB_ACCESS (1<<10) /* 0 here means pipe a */ +#define SWF11_DPMS_MASK 0x07 +#define SWF11_DPMS_OFF (1<<2) +#define SWF11_DPMS_SUSPEND (1<<1) +#define SWF11_DPMS_STANDBY (1<<0) +#define SWF11_DPMS_ON 0 + +#define SWF14_GFX_PFIT_EN (1<<31) +#define SWF14_TEXT_PFIT_EN (1<<30) +#define SWF14_LID_STATUS_CLOSED (1<<29) /* 0 here means open */ +#define SWF14_POPUP_EN (1<<28) +#define SWF14_DISPLAY_HOLDOFF (1<<27) +#define SWF14_DISP_DETECT_EN (1<<26) +#define SWF14_DOCKING_STATUS_DOCKED (1<<25) /* 0 here means undocked */ +#define SWF14_DRIVER_STATUS (1<<24) +#define SWF14_OS_TYPE_WIN9X (1<<23) +#define SWF14_OS_TYPE_WINNT (1<<22) +/* 21:19 rsvd */ +#define SWF14_PM_TYPE_MASK 0x00070000 +#define SWF14_PM_ACPI_VIDEO (0x4 << 16) +#define SWF14_PM_ACPI (0x3 << 16) +#define SWF14_PM_APM_12 (0x2 << 16) +#define SWF14_PM_APM_11 (0x1 << 16) +#define SWF14_HK_REQUEST_MASK 0x0000ffff /* see GR18 6:3 for event type */ + /* if GR18 indicates a display switch */ +#define SWF14_DS_PIPEB_LFP2_EN (1<<15) +#define SWF14_DS_PIPEB_EFP2_EN (1<<14) +#define SWF14_DS_PIPEB_TV2_EN (1<<13) +#define SWF14_DS_PIPEB_CRT2_EN (1<<12) +#define SWF14_DS_PIPEB_LFP_EN (1<<11) +#define SWF14_DS_PIPEB_EFP_EN (1<<10) +#define SWF14_DS_PIPEB_TV_EN (1<<9) +#define SWF14_DS_PIPEB_CRT_EN (1<<8) +#define SWF14_DS_PIPEA_LFP2_EN (1<<7) +#define SWF14_DS_PIPEA_EFP2_EN (1<<6) +#define SWF14_DS_PIPEA_TV2_EN (1<<5) +#define SWF14_DS_PIPEA_CRT2_EN (1<<4) +#define SWF14_DS_PIPEA_LFP_EN (1<<3) +#define SWF14_DS_PIPEA_EFP_EN (1<<2) +#define SWF14_DS_PIPEA_TV_EN (1<<1) +#define SWF14_DS_PIPEA_CRT_EN (1<<0) + /* if GR18 indicates a panel fitting request */ +#define SWF14_PFIT_EN (1<<0) /* 0 means disable */ + /* if GR18 indicates an APM change request */ +#define SWF14_APM_HIBERNATE 0x4 +#define SWF14_APM_SUSPEND 0x3 +#define SWF14_APM_STANDBY 0x1 +#define SWF14_APM_RESTORE 0x0 + +/* Add the device class for LFP, TV, HDMI */ +#define DEVICE_TYPE_INT_LFP 0x1022 +#define DEVICE_TYPE_INT_TV 0x1009 +#define DEVICE_TYPE_HDMI 0x60D2 +#define DEVICE_TYPE_DP 0x68C6 +#define DEVICE_TYPE_eDP 0x78C6 + +/* define the DVO port for HDMI output type */ +#define DVO_B 1 +#define DVO_C 2 +#define DVO_D 3 + +/* define the PORT for DP output type */ +#define PORT_IDPB 7 +#define PORT_IDPC 8 +#define PORT_IDPD 9 + +#endif /* _I830_BIOS_H_ */ diff --git a/drivers/gpu/drm/i915/intel_crt.c b/drivers/gpu/drm/i915/intel_crt.c new file mode 100644 index 00000000..0979d887 --- /dev/null +++ b/drivers/gpu/drm/i915/intel_crt.c @@ -0,0 +1,609 @@ +/* + * Copyright © 2006-2007 Intel Corporation + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Eric Anholt + */ + +#include +#include +#include "drmP.h" +#include "drm.h" +#include "drm_crtc.h" +#include "drm_crtc_helper.h" +#include "drm_edid.h" +#include "intel_drv.h" +#include "i915_drm.h" +#include "i915_drv.h" + +/* Here's the desired hotplug mode */ +#define ADPA_HOTPLUG_BITS (ADPA_CRT_HOTPLUG_PERIOD_128 | \ + ADPA_CRT_HOTPLUG_WARMUP_10MS | \ + ADPA_CRT_HOTPLUG_SAMPLE_4S | \ + ADPA_CRT_HOTPLUG_VOLTAGE_50 | \ + ADPA_CRT_HOTPLUG_VOLREF_325MV | \ + ADPA_CRT_HOTPLUG_ENABLE) + +struct intel_crt { + struct intel_encoder base; + bool force_hotplug_required; +}; + +static struct intel_crt *intel_attached_crt(struct drm_connector *connector) +{ + return container_of(intel_attached_encoder(connector), + struct intel_crt, base); +} + +static void intel_crt_dpms(struct drm_encoder *encoder, int mode) +{ + struct drm_device *dev = encoder->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + u32 temp, reg; + + if (HAS_PCH_SPLIT(dev)) + reg = PCH_ADPA; + else + reg = ADPA; + + temp = I915_READ(reg); + temp &= ~(ADPA_HSYNC_CNTL_DISABLE | ADPA_VSYNC_CNTL_DISABLE); + temp &= ~ADPA_DAC_ENABLE; + + switch(mode) { + case DRM_MODE_DPMS_ON: + temp |= ADPA_DAC_ENABLE; + break; + case DRM_MODE_DPMS_STANDBY: + temp |= ADPA_DAC_ENABLE | ADPA_HSYNC_CNTL_DISABLE; + break; + case DRM_MODE_DPMS_SUSPEND: + temp |= ADPA_DAC_ENABLE | ADPA_VSYNC_CNTL_DISABLE; + break; + case DRM_MODE_DPMS_OFF: + temp |= ADPA_HSYNC_CNTL_DISABLE | ADPA_VSYNC_CNTL_DISABLE; + break; + } + + I915_WRITE(reg, temp); +} + +static int intel_crt_mode_valid(struct drm_connector *connector, + struct drm_display_mode *mode) +{ + struct drm_device *dev = connector->dev; + + int max_clock = 0; + if (mode->flags & DRM_MODE_FLAG_DBLSCAN) + return MODE_NO_DBLESCAN; + + if (mode->clock < 25000) + return MODE_CLOCK_LOW; + + if (IS_GEN2(dev)) + max_clock = 350000; + else + max_clock = 400000; + if (mode->clock > max_clock) + return MODE_CLOCK_HIGH; + + return MODE_OK; +} + +static bool intel_crt_mode_fixup(struct drm_encoder *encoder, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + return true; +} + +static void intel_crt_mode_set(struct drm_encoder *encoder, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + + struct drm_device *dev = encoder->dev; + struct drm_crtc *crtc = encoder->crtc; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + struct drm_i915_private *dev_priv = dev->dev_private; + int dpll_md_reg; + u32 adpa, dpll_md; + u32 adpa_reg; + + dpll_md_reg = DPLL_MD(intel_crtc->pipe); + + if (HAS_PCH_SPLIT(dev)) + adpa_reg = PCH_ADPA; + else + adpa_reg = ADPA; + + /* + * Disable separate mode multiplier used when cloning SDVO to CRT + * XXX this needs to be adjusted when we really are cloning + */ + if (INTEL_INFO(dev)->gen >= 4 && !HAS_PCH_SPLIT(dev)) { + dpll_md = I915_READ(dpll_md_reg); + I915_WRITE(dpll_md_reg, + dpll_md & ~DPLL_MD_UDI_MULTIPLIER_MASK); + } + + adpa = ADPA_HOTPLUG_BITS; + if (adjusted_mode->flags & DRM_MODE_FLAG_PHSYNC) + adpa |= ADPA_HSYNC_ACTIVE_HIGH; + if (adjusted_mode->flags & DRM_MODE_FLAG_PVSYNC) + adpa |= ADPA_VSYNC_ACTIVE_HIGH; + + if (intel_crtc->pipe == 0) { + if (HAS_PCH_CPT(dev)) + adpa |= PORT_TRANS_A_SEL_CPT; + else + adpa |= ADPA_PIPE_A_SELECT; + } else { + if (HAS_PCH_CPT(dev)) + adpa |= PORT_TRANS_B_SEL_CPT; + else + adpa |= ADPA_PIPE_B_SELECT; + } + + if (!HAS_PCH_SPLIT(dev)) + I915_WRITE(BCLRPAT(intel_crtc->pipe), 0); + + I915_WRITE(adpa_reg, adpa); +} + +static bool intel_ironlake_crt_detect_hotplug(struct drm_connector *connector) +{ + struct drm_device *dev = connector->dev; + struct intel_crt *crt = intel_attached_crt(connector); + struct drm_i915_private *dev_priv = dev->dev_private; + u32 adpa; + bool ret; + + /* The first time through, trigger an explicit detection cycle */ + if (crt->force_hotplug_required) { + bool turn_off_dac = HAS_PCH_SPLIT(dev); + u32 save_adpa; + + crt->force_hotplug_required = 0; + + save_adpa = adpa = I915_READ(PCH_ADPA); + DRM_DEBUG_KMS("trigger hotplug detect cycle: adpa=0x%x\n", adpa); + + adpa |= ADPA_CRT_HOTPLUG_FORCE_TRIGGER; + if (turn_off_dac) + adpa &= ~ADPA_DAC_ENABLE; + + I915_WRITE(PCH_ADPA, adpa); + + if (wait_for((I915_READ(PCH_ADPA) & ADPA_CRT_HOTPLUG_FORCE_TRIGGER) == 0, + 1000)) + DRM_DEBUG_KMS("timed out waiting for FORCE_TRIGGER"); + + if (turn_off_dac) { + I915_WRITE(PCH_ADPA, save_adpa); + POSTING_READ(PCH_ADPA); + } + } + + /* Check the status to see if both blue and green are on now */ + adpa = I915_READ(PCH_ADPA); + if ((adpa & ADPA_CRT_HOTPLUG_MONITOR_MASK) != 0) + ret = true; + else + ret = false; + DRM_DEBUG_KMS("ironlake hotplug adpa=0x%x, result %d\n", adpa, ret); + + return ret; +} + +/** + * Uses CRT_HOTPLUG_EN and CRT_HOTPLUG_STAT to detect CRT presence. + * + * Not for i915G/i915GM + * + * \return true if CRT is connected. + * \return false if CRT is disconnected. + */ +static bool intel_crt_detect_hotplug(struct drm_connector *connector) +{ + struct drm_device *dev = connector->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + u32 hotplug_en, orig, stat; + bool ret = false; + int i, tries = 0; + + if (HAS_PCH_SPLIT(dev)) + return intel_ironlake_crt_detect_hotplug(connector); + + /* + * On 4 series desktop, CRT detect sequence need to be done twice + * to get a reliable result. + */ + + if (IS_G4X(dev) && !IS_GM45(dev)) + tries = 2; + else + tries = 1; + hotplug_en = orig = I915_READ(PORT_HOTPLUG_EN); + hotplug_en |= CRT_HOTPLUG_FORCE_DETECT; + + for (i = 0; i < tries ; i++) { + /* turn on the FORCE_DETECT */ + I915_WRITE(PORT_HOTPLUG_EN, hotplug_en); + /* wait for FORCE_DETECT to go off */ + if (wait_for((I915_READ(PORT_HOTPLUG_EN) & + CRT_HOTPLUG_FORCE_DETECT) == 0, + 1000)) + DRM_DEBUG_KMS("timed out waiting for FORCE_DETECT to go off"); + } + + stat = I915_READ(PORT_HOTPLUG_STAT); + if ((stat & CRT_HOTPLUG_MONITOR_MASK) != CRT_HOTPLUG_MONITOR_NONE) + ret = true; + + /* clear the interrupt we just generated, if any */ + I915_WRITE(PORT_HOTPLUG_STAT, CRT_HOTPLUG_INT_STATUS); + + /* and put the bits back */ + I915_WRITE(PORT_HOTPLUG_EN, orig); + + return ret; +} + +static bool intel_crt_detect_ddc(struct drm_connector *connector) +{ + struct intel_crt *crt = intel_attached_crt(connector); + struct drm_i915_private *dev_priv = crt->base.base.dev->dev_private; + + /* CRT should always be at 0, but check anyway */ + if (crt->base.type != INTEL_OUTPUT_ANALOG) + return false; + + if (intel_ddc_probe(&crt->base, dev_priv->crt_ddc_pin)) { + struct edid *edid; + bool is_digital = false; + + edid = drm_get_edid(connector, + &dev_priv->gmbus[dev_priv->crt_ddc_pin].adapter); + /* + * This may be a DVI-I connector with a shared DDC + * link between analog and digital outputs, so we + * have to check the EDID input spec of the attached device. + * + * On the other hand, what should we do if it is a broken EDID? + */ + if (edid != NULL) { + is_digital = edid->input & DRM_EDID_INPUT_DIGITAL; + connector->display_info.raw_edid = NULL; + kfree(edid); + } + + if (!is_digital) { + DRM_DEBUG_KMS("CRT detected via DDC:0x50 [EDID]\n"); + return true; + } else { + DRM_DEBUG_KMS("CRT not detected via DDC:0x50 [EDID reports a digital panel]\n"); + } + } + + return false; +} + +static enum drm_connector_status +intel_crt_load_detect(struct intel_crt *crt) +{ + struct drm_device *dev = crt->base.base.dev; + struct drm_i915_private *dev_priv = dev->dev_private; + uint32_t pipe = to_intel_crtc(crt->base.base.crtc)->pipe; + uint32_t save_bclrpat; + uint32_t save_vtotal; + uint32_t vtotal, vactive; + uint32_t vsample; + uint32_t vblank, vblank_start, vblank_end; + uint32_t dsl; + uint32_t bclrpat_reg; + uint32_t vtotal_reg; + uint32_t vblank_reg; + uint32_t vsync_reg; + uint32_t pipeconf_reg; + uint32_t pipe_dsl_reg; + uint8_t st00; + enum drm_connector_status status; + + DRM_DEBUG_KMS("starting load-detect on CRT\n"); + + bclrpat_reg = BCLRPAT(pipe); + vtotal_reg = VTOTAL(pipe); + vblank_reg = VBLANK(pipe); + vsync_reg = VSYNC(pipe); + pipeconf_reg = PIPECONF(pipe); + pipe_dsl_reg = PIPEDSL(pipe); + + save_bclrpat = I915_READ(bclrpat_reg); + save_vtotal = I915_READ(vtotal_reg); + vblank = I915_READ(vblank_reg); + + vtotal = ((save_vtotal >> 16) & 0xfff) + 1; + vactive = (save_vtotal & 0x7ff) + 1; + + vblank_start = (vblank & 0xfff) + 1; + vblank_end = ((vblank >> 16) & 0xfff) + 1; + + /* Set the border color to purple. */ + I915_WRITE(bclrpat_reg, 0x500050); + + if (!IS_GEN2(dev)) { + uint32_t pipeconf = I915_READ(pipeconf_reg); + I915_WRITE(pipeconf_reg, pipeconf | PIPECONF_FORCE_BORDER); + POSTING_READ(pipeconf_reg); + /* Wait for next Vblank to substitue + * border color for Color info */ + intel_wait_for_vblank(dev, pipe); + st00 = I915_READ8(VGA_MSR_WRITE); + status = ((st00 & (1 << 4)) != 0) ? + connector_status_connected : + connector_status_disconnected; + + I915_WRITE(pipeconf_reg, pipeconf); + } else { + bool restore_vblank = false; + int count, detect; + + /* + * If there isn't any border, add some. + * Yes, this will flicker + */ + if (vblank_start <= vactive && vblank_end >= vtotal) { + uint32_t vsync = I915_READ(vsync_reg); + uint32_t vsync_start = (vsync & 0xffff) + 1; + + vblank_start = vsync_start; + I915_WRITE(vblank_reg, + (vblank_start - 1) | + ((vblank_end - 1) << 16)); + restore_vblank = true; + } + /* sample in the vertical border, selecting the larger one */ + if (vblank_start - vactive >= vtotal - vblank_end) + vsample = (vblank_start + vactive) >> 1; + else + vsample = (vtotal + vblank_end) >> 1; + + /* + * Wait for the border to be displayed + */ + while (I915_READ(pipe_dsl_reg) >= vactive) + ; + while ((dsl = I915_READ(pipe_dsl_reg)) <= vsample) + ; + /* + * Watch ST00 for an entire scanline + */ + detect = 0; + count = 0; + do { + count++; + /* Read the ST00 VGA status register */ + st00 = I915_READ8(VGA_MSR_WRITE); + if (st00 & (1 << 4)) + detect++; + } while ((I915_READ(pipe_dsl_reg) == dsl)); + + /* restore vblank if necessary */ + if (restore_vblank) + I915_WRITE(vblank_reg, vblank); + /* + * If more than 3/4 of the scanline detected a monitor, + * then it is assumed to be present. This works even on i830, + * where there isn't any way to force the border color across + * the screen + */ + status = detect * 4 > count * 3 ? + connector_status_connected : + connector_status_disconnected; + } + + /* Restore previous settings */ + I915_WRITE(bclrpat_reg, save_bclrpat); + + return status; +} + +static enum drm_connector_status +intel_crt_detect(struct drm_connector *connector, bool force) +{ + struct drm_device *dev = connector->dev; + struct intel_crt *crt = intel_attached_crt(connector); + struct drm_crtc *crtc; + enum drm_connector_status status; + + if (I915_HAS_HOTPLUG(dev)) { + if (intel_crt_detect_hotplug(connector)) { + DRM_DEBUG_KMS("CRT detected via hotplug\n"); + return connector_status_connected; + } else { + DRM_DEBUG_KMS("CRT not detected via hotplug\n"); + return connector_status_disconnected; + } + } + + if (intel_crt_detect_ddc(connector)) + return connector_status_connected; + + if (!force) + return connector->status; + + /* for pre-945g platforms use load detect */ + crtc = crt->base.base.crtc; + if (crtc && crtc->enabled) { + status = intel_crt_load_detect(crt); + } else { + struct intel_load_detect_pipe tmp; + + if (intel_get_load_detect_pipe(&crt->base, connector, NULL, + &tmp)) { + if (intel_crt_detect_ddc(connector)) + status = connector_status_connected; + else + status = intel_crt_load_detect(crt); + intel_release_load_detect_pipe(&crt->base, connector, + &tmp); + } else + status = connector_status_unknown; + } + + return status; +} + +static void intel_crt_destroy(struct drm_connector *connector) +{ + drm_sysfs_connector_remove(connector); + drm_connector_cleanup(connector); + kfree(connector); +} + +static int intel_crt_get_modes(struct drm_connector *connector) +{ + struct drm_device *dev = connector->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + int ret; + + ret = intel_ddc_get_modes(connector, + &dev_priv->gmbus[dev_priv->crt_ddc_pin].adapter); + if (ret || !IS_G4X(dev)) + return ret; + + /* Try to probe digital port for output in DVI-I -> VGA mode. */ + return intel_ddc_get_modes(connector, + &dev_priv->gmbus[GMBUS_PORT_DPB].adapter); +} + +static int intel_crt_set_property(struct drm_connector *connector, + struct drm_property *property, + uint64_t value) +{ + return 0; +} + +static void intel_crt_reset(struct drm_connector *connector) +{ + struct drm_device *dev = connector->dev; + struct intel_crt *crt = intel_attached_crt(connector); + + if (HAS_PCH_SPLIT(dev)) + crt->force_hotplug_required = 1; +} + +/* + * Routines for controlling stuff on the analog port + */ + +static const struct drm_encoder_helper_funcs intel_crt_helper_funcs = { + .dpms = intel_crt_dpms, + .mode_fixup = intel_crt_mode_fixup, + .prepare = intel_encoder_prepare, + .commit = intel_encoder_commit, + .mode_set = intel_crt_mode_set, +}; + +static const struct drm_connector_funcs intel_crt_connector_funcs = { + .reset = intel_crt_reset, + .dpms = drm_helper_connector_dpms, + .detect = intel_crt_detect, + .fill_modes = drm_helper_probe_single_connector_modes, + .destroy = intel_crt_destroy, + .set_property = intel_crt_set_property, +}; + +static const struct drm_connector_helper_funcs intel_crt_connector_helper_funcs = { + .mode_valid = intel_crt_mode_valid, + .get_modes = intel_crt_get_modes, + .best_encoder = intel_best_encoder, +}; + +static const struct drm_encoder_funcs intel_crt_enc_funcs = { + .destroy = intel_encoder_destroy, +}; + +void intel_crt_init(struct drm_device *dev) +{ + struct drm_connector *connector; + struct intel_crt *crt; + struct intel_connector *intel_connector; + struct drm_i915_private *dev_priv = dev->dev_private; + + crt = kzalloc(sizeof(struct intel_crt), GFP_KERNEL); + if (!crt) + return; + + intel_connector = kzalloc(sizeof(struct intel_connector), GFP_KERNEL); + if (!intel_connector) { + kfree(crt); + return; + } + + connector = &intel_connector->base; + drm_connector_init(dev, &intel_connector->base, + &intel_crt_connector_funcs, DRM_MODE_CONNECTOR_VGA); + + drm_encoder_init(dev, &crt->base.base, &intel_crt_enc_funcs, + DRM_MODE_ENCODER_DAC); + + intel_connector_attach_encoder(intel_connector, &crt->base); + + crt->base.type = INTEL_OUTPUT_ANALOG; + crt->base.clone_mask = (1 << INTEL_SDVO_NON_TV_CLONE_BIT | + 1 << INTEL_ANALOG_CLONE_BIT | + 1 << INTEL_SDVO_LVDS_CLONE_BIT); + crt->base.crtc_mask = (1 << 0) | (1 << 1); + connector->interlace_allowed = 1; + connector->doublescan_allowed = 0; + + drm_encoder_helper_add(&crt->base.base, &intel_crt_helper_funcs); + drm_connector_helper_add(connector, &intel_crt_connector_helper_funcs); + + drm_sysfs_connector_add(connector); + + if (I915_HAS_HOTPLUG(dev)) + connector->polled = DRM_CONNECTOR_POLL_HPD; + else + connector->polled = DRM_CONNECTOR_POLL_CONNECT; + + /* + * Configure the automatic hotplug detection stuff + */ + crt->force_hotplug_required = 0; + if (HAS_PCH_SPLIT(dev)) { + u32 adpa; + + adpa = I915_READ(PCH_ADPA); + adpa &= ~ADPA_CRT_HOTPLUG_MASK; + adpa |= ADPA_HOTPLUG_BITS; + I915_WRITE(PCH_ADPA, adpa); + POSTING_READ(PCH_ADPA); + + DRM_DEBUG_KMS("pch crt adpa set to 0x%x\n", adpa); + crt->force_hotplug_required = 1; + } + + dev_priv->hotplug_supported_mask |= CRT_HOTPLUG_INT_STATUS; +} diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c new file mode 100644 index 00000000..4b8e2355 --- /dev/null +++ b/drivers/gpu/drm/i915/intel_display.c @@ -0,0 +1,8240 @@ +/* + * Copyright © 2006-2007 Intel Corporation + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Eric Anholt + */ + +#include +#include +#include +#include +#include +#include +#include "drmP.h" +#include "intel_drv.h" +#include "i915_drm.h" +#include "i915_drv.h" +#include "i915_trace.h" +#include "drm_dp_helper.h" + +#include "drm_crtc_helper.h" + +#define HAS_eDP (intel_pipe_has_type(crtc, INTEL_OUTPUT_EDP)) + +bool intel_pipe_has_type (struct drm_crtc *crtc, int type); +static void intel_update_watermarks(struct drm_device *dev); +static void intel_increase_pllclock(struct drm_crtc *crtc); +static void intel_crtc_update_cursor(struct drm_crtc *crtc, bool on); + +typedef struct { + /* given values */ + int n; + int m1, m2; + int p1, p2; + /* derived values */ + int dot; + int vco; + int m; + int p; +} intel_clock_t; + +typedef struct { + int min, max; +} intel_range_t; + +typedef struct { + int dot_limit; + int p2_slow, p2_fast; +} intel_p2_t; + +#define INTEL_P2_NUM 2 +typedef struct intel_limit intel_limit_t; +struct intel_limit { + intel_range_t dot, vco, n, m, m1, m2, p, p1; + intel_p2_t p2; + bool (* find_pll)(const intel_limit_t *, struct drm_crtc *, + int, int, intel_clock_t *); +}; + +/* FDI */ +#define IRONLAKE_FDI_FREQ 2700000 /* in kHz for mode->clock */ + +static bool +intel_find_best_PLL(const intel_limit_t *limit, struct drm_crtc *crtc, + int target, int refclk, intel_clock_t *best_clock); +static bool +intel_g4x_find_best_PLL(const intel_limit_t *limit, struct drm_crtc *crtc, + int target, int refclk, intel_clock_t *best_clock); + +static bool +intel_find_pll_g4x_dp(const intel_limit_t *, struct drm_crtc *crtc, + int target, int refclk, intel_clock_t *best_clock); +static bool +intel_find_pll_ironlake_dp(const intel_limit_t *, struct drm_crtc *crtc, + int target, int refclk, intel_clock_t *best_clock); + +static inline u32 /* units of 100MHz */ +intel_fdi_link_freq(struct drm_device *dev) +{ + if (IS_GEN5(dev)) { + struct drm_i915_private *dev_priv = dev->dev_private; + return (I915_READ(FDI_PLL_BIOS_0) & FDI_PLL_FB_CLOCK_MASK) + 2; + } else + return 27; +} + +static const intel_limit_t intel_limits_i8xx_dvo = { + .dot = { .min = 25000, .max = 350000 }, + .vco = { .min = 930000, .max = 1400000 }, + .n = { .min = 3, .max = 16 }, + .m = { .min = 96, .max = 140 }, + .m1 = { .min = 18, .max = 26 }, + .m2 = { .min = 6, .max = 16 }, + .p = { .min = 4, .max = 128 }, + .p1 = { .min = 2, .max = 33 }, + .p2 = { .dot_limit = 165000, + .p2_slow = 4, .p2_fast = 2 }, + .find_pll = intel_find_best_PLL, +}; + +static const intel_limit_t intel_limits_i8xx_lvds = { + .dot = { .min = 25000, .max = 350000 }, + .vco = { .min = 930000, .max = 1400000 }, + .n = { .min = 3, .max = 16 }, + .m = { .min = 96, .max = 140 }, + .m1 = { .min = 18, .max = 26 }, + .m2 = { .min = 6, .max = 16 }, + .p = { .min = 4, .max = 128 }, + .p1 = { .min = 1, .max = 6 }, + .p2 = { .dot_limit = 165000, + .p2_slow = 14, .p2_fast = 7 }, + .find_pll = intel_find_best_PLL, +}; + +static const intel_limit_t intel_limits_i9xx_sdvo = { + .dot = { .min = 20000, .max = 400000 }, + .vco = { .min = 1400000, .max = 2800000 }, + .n = { .min = 1, .max = 6 }, + .m = { .min = 70, .max = 120 }, + .m1 = { .min = 10, .max = 22 }, + .m2 = { .min = 5, .max = 9 }, + .p = { .min = 5, .max = 80 }, + .p1 = { .min = 1, .max = 8 }, + .p2 = { .dot_limit = 200000, + .p2_slow = 10, .p2_fast = 5 }, + .find_pll = intel_find_best_PLL, +}; + +static const intel_limit_t intel_limits_i9xx_lvds = { + .dot = { .min = 20000, .max = 400000 }, + .vco = { .min = 1400000, .max = 2800000 }, + .n = { .min = 1, .max = 6 }, + .m = { .min = 70, .max = 120 }, + .m1 = { .min = 10, .max = 22 }, + .m2 = { .min = 5, .max = 9 }, + .p = { .min = 7, .max = 98 }, + .p1 = { .min = 1, .max = 8 }, + .p2 = { .dot_limit = 112000, + .p2_slow = 14, .p2_fast = 7 }, + .find_pll = intel_find_best_PLL, +}; + + +static const intel_limit_t intel_limits_g4x_sdvo = { + .dot = { .min = 25000, .max = 270000 }, + .vco = { .min = 1750000, .max = 3500000}, + .n = { .min = 1, .max = 4 }, + .m = { .min = 104, .max = 138 }, + .m1 = { .min = 17, .max = 23 }, + .m2 = { .min = 5, .max = 11 }, + .p = { .min = 10, .max = 30 }, + .p1 = { .min = 1, .max = 3}, + .p2 = { .dot_limit = 270000, + .p2_slow = 10, + .p2_fast = 10 + }, + .find_pll = intel_g4x_find_best_PLL, +}; + +static const intel_limit_t intel_limits_g4x_hdmi = { + .dot = { .min = 22000, .max = 400000 }, + .vco = { .min = 1750000, .max = 3500000}, + .n = { .min = 1, .max = 4 }, + .m = { .min = 104, .max = 138 }, + .m1 = { .min = 16, .max = 23 }, + .m2 = { .min = 5, .max = 11 }, + .p = { .min = 5, .max = 80 }, + .p1 = { .min = 1, .max = 8}, + .p2 = { .dot_limit = 165000, + .p2_slow = 10, .p2_fast = 5 }, + .find_pll = intel_g4x_find_best_PLL, +}; + +static const intel_limit_t intel_limits_g4x_single_channel_lvds = { + .dot = { .min = 20000, .max = 115000 }, + .vco = { .min = 1750000, .max = 3500000 }, + .n = { .min = 1, .max = 3 }, + .m = { .min = 104, .max = 138 }, + .m1 = { .min = 17, .max = 23 }, + .m2 = { .min = 5, .max = 11 }, + .p = { .min = 28, .max = 112 }, + .p1 = { .min = 2, .max = 8 }, + .p2 = { .dot_limit = 0, + .p2_slow = 14, .p2_fast = 14 + }, + .find_pll = intel_g4x_find_best_PLL, +}; + +static const intel_limit_t intel_limits_g4x_dual_channel_lvds = { + .dot = { .min = 80000, .max = 224000 }, + .vco = { .min = 1750000, .max = 3500000 }, + .n = { .min = 1, .max = 3 }, + .m = { .min = 104, .max = 138 }, + .m1 = { .min = 17, .max = 23 }, + .m2 = { .min = 5, .max = 11 }, + .p = { .min = 14, .max = 42 }, + .p1 = { .min = 2, .max = 6 }, + .p2 = { .dot_limit = 0, + .p2_slow = 7, .p2_fast = 7 + }, + .find_pll = intel_g4x_find_best_PLL, +}; + +static const intel_limit_t intel_limits_g4x_display_port = { + .dot = { .min = 161670, .max = 227000 }, + .vco = { .min = 1750000, .max = 3500000}, + .n = { .min = 1, .max = 2 }, + .m = { .min = 97, .max = 108 }, + .m1 = { .min = 0x10, .max = 0x12 }, + .m2 = { .min = 0x05, .max = 0x06 }, + .p = { .min = 10, .max = 20 }, + .p1 = { .min = 1, .max = 2}, + .p2 = { .dot_limit = 0, + .p2_slow = 10, .p2_fast = 10 }, + .find_pll = intel_find_pll_g4x_dp, +}; + +static const intel_limit_t intel_limits_pineview_sdvo = { + .dot = { .min = 20000, .max = 400000}, + .vco = { .min = 1700000, .max = 3500000 }, + /* Pineview's Ncounter is a ring counter */ + .n = { .min = 3, .max = 6 }, + .m = { .min = 2, .max = 256 }, + /* Pineview only has one combined m divider, which we treat as m2. */ + .m1 = { .min = 0, .max = 0 }, + .m2 = { .min = 0, .max = 254 }, + .p = { .min = 5, .max = 80 }, + .p1 = { .min = 1, .max = 8 }, + .p2 = { .dot_limit = 200000, + .p2_slow = 10, .p2_fast = 5 }, + .find_pll = intel_find_best_PLL, +}; + +static const intel_limit_t intel_limits_pineview_lvds = { + .dot = { .min = 20000, .max = 400000 }, + .vco = { .min = 1700000, .max = 3500000 }, + .n = { .min = 3, .max = 6 }, + .m = { .min = 2, .max = 256 }, + .m1 = { .min = 0, .max = 0 }, + .m2 = { .min = 0, .max = 254 }, + .p = { .min = 7, .max = 112 }, + .p1 = { .min = 1, .max = 8 }, + .p2 = { .dot_limit = 112000, + .p2_slow = 14, .p2_fast = 14 }, + .find_pll = intel_find_best_PLL, +}; + +/* Ironlake / Sandybridge + * + * We calculate clock using (register_value + 2) for N/M1/M2, so here + * the range value for them is (actual_value - 2). + */ +static const intel_limit_t intel_limits_ironlake_dac = { + .dot = { .min = 25000, .max = 350000 }, + .vco = { .min = 1760000, .max = 3510000 }, + .n = { .min = 1, .max = 5 }, + .m = { .min = 79, .max = 127 }, + .m1 = { .min = 12, .max = 22 }, + .m2 = { .min = 5, .max = 9 }, + .p = { .min = 5, .max = 80 }, + .p1 = { .min = 1, .max = 8 }, + .p2 = { .dot_limit = 225000, + .p2_slow = 10, .p2_fast = 5 }, + .find_pll = intel_g4x_find_best_PLL, +}; + +static const intel_limit_t intel_limits_ironlake_single_lvds = { + .dot = { .min = 25000, .max = 350000 }, + .vco = { .min = 1760000, .max = 3510000 }, + .n = { .min = 1, .max = 3 }, + .m = { .min = 79, .max = 118 }, + .m1 = { .min = 12, .max = 22 }, + .m2 = { .min = 5, .max = 9 }, + .p = { .min = 28, .max = 112 }, + .p1 = { .min = 2, .max = 8 }, + .p2 = { .dot_limit = 225000, + .p2_slow = 14, .p2_fast = 14 }, + .find_pll = intel_g4x_find_best_PLL, +}; + +static const intel_limit_t intel_limits_ironlake_dual_lvds = { + .dot = { .min = 25000, .max = 350000 }, + .vco = { .min = 1760000, .max = 3510000 }, + .n = { .min = 1, .max = 3 }, + .m = { .min = 79, .max = 127 }, + .m1 = { .min = 12, .max = 22 }, + .m2 = { .min = 5, .max = 9 }, + .p = { .min = 14, .max = 56 }, + .p1 = { .min = 2, .max = 8 }, + .p2 = { .dot_limit = 225000, + .p2_slow = 7, .p2_fast = 7 }, + .find_pll = intel_g4x_find_best_PLL, +}; + +/* LVDS 100mhz refclk limits. */ +static const intel_limit_t intel_limits_ironlake_single_lvds_100m = { + .dot = { .min = 25000, .max = 350000 }, + .vco = { .min = 1760000, .max = 3510000 }, + .n = { .min = 1, .max = 2 }, + .m = { .min = 79, .max = 126 }, + .m1 = { .min = 12, .max = 22 }, + .m2 = { .min = 5, .max = 9 }, + .p = { .min = 28, .max = 112 }, + .p1 = { .min = 2,.max = 8 }, + .p2 = { .dot_limit = 225000, + .p2_slow = 14, .p2_fast = 14 }, + .find_pll = intel_g4x_find_best_PLL, +}; + +static const intel_limit_t intel_limits_ironlake_dual_lvds_100m = { + .dot = { .min = 25000, .max = 350000 }, + .vco = { .min = 1760000, .max = 3510000 }, + .n = { .min = 1, .max = 3 }, + .m = { .min = 79, .max = 126 }, + .m1 = { .min = 12, .max = 22 }, + .m2 = { .min = 5, .max = 9 }, + .p = { .min = 14, .max = 42 }, + .p1 = { .min = 2,.max = 6 }, + .p2 = { .dot_limit = 225000, + .p2_slow = 7, .p2_fast = 7 }, + .find_pll = intel_g4x_find_best_PLL, +}; + +static const intel_limit_t intel_limits_ironlake_display_port = { + .dot = { .min = 25000, .max = 350000 }, + .vco = { .min = 1760000, .max = 3510000}, + .n = { .min = 1, .max = 2 }, + .m = { .min = 81, .max = 90 }, + .m1 = { .min = 12, .max = 22 }, + .m2 = { .min = 5, .max = 9 }, + .p = { .min = 10, .max = 20 }, + .p1 = { .min = 1, .max = 2}, + .p2 = { .dot_limit = 0, + .p2_slow = 10, .p2_fast = 10 }, + .find_pll = intel_find_pll_ironlake_dp, +}; + +static const intel_limit_t *intel_ironlake_limit(struct drm_crtc *crtc, + int refclk) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + const intel_limit_t *limit; + + if (intel_pipe_has_type(crtc, INTEL_OUTPUT_LVDS)) { + if ((I915_READ(PCH_LVDS) & LVDS_CLKB_POWER_MASK) == + LVDS_CLKB_POWER_UP) { + /* LVDS dual channel */ + if (refclk == 100000) + limit = &intel_limits_ironlake_dual_lvds_100m; + else + limit = &intel_limits_ironlake_dual_lvds; + } else { + if (refclk == 100000) + limit = &intel_limits_ironlake_single_lvds_100m; + else + limit = &intel_limits_ironlake_single_lvds; + } + } else if (intel_pipe_has_type(crtc, INTEL_OUTPUT_DISPLAYPORT) || + HAS_eDP) + limit = &intel_limits_ironlake_display_port; + else + limit = &intel_limits_ironlake_dac; + + return limit; +} + +static const intel_limit_t *intel_g4x_limit(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + const intel_limit_t *limit; + + if (intel_pipe_has_type(crtc, INTEL_OUTPUT_LVDS)) { + if ((I915_READ(LVDS) & LVDS_CLKB_POWER_MASK) == + LVDS_CLKB_POWER_UP) + /* LVDS with dual channel */ + limit = &intel_limits_g4x_dual_channel_lvds; + else + /* LVDS with dual channel */ + limit = &intel_limits_g4x_single_channel_lvds; + } else if (intel_pipe_has_type(crtc, INTEL_OUTPUT_HDMI) || + intel_pipe_has_type(crtc, INTEL_OUTPUT_ANALOG)) { + limit = &intel_limits_g4x_hdmi; + } else if (intel_pipe_has_type(crtc, INTEL_OUTPUT_SDVO)) { + limit = &intel_limits_g4x_sdvo; + } else if (intel_pipe_has_type (crtc, INTEL_OUTPUT_DISPLAYPORT)) { + limit = &intel_limits_g4x_display_port; + } else /* The option is for other outputs */ + limit = &intel_limits_i9xx_sdvo; + + return limit; +} + +static const intel_limit_t *intel_limit(struct drm_crtc *crtc, int refclk) +{ + struct drm_device *dev = crtc->dev; + const intel_limit_t *limit; + + if (HAS_PCH_SPLIT(dev)) + limit = intel_ironlake_limit(crtc, refclk); + else if (IS_G4X(dev)) { + limit = intel_g4x_limit(crtc); + } else if (IS_PINEVIEW(dev)) { + if (intel_pipe_has_type(crtc, INTEL_OUTPUT_LVDS)) + limit = &intel_limits_pineview_lvds; + else + limit = &intel_limits_pineview_sdvo; + } else if (!IS_GEN2(dev)) { + if (intel_pipe_has_type(crtc, INTEL_OUTPUT_LVDS)) + limit = &intel_limits_i9xx_lvds; + else + limit = &intel_limits_i9xx_sdvo; + } else { + if (intel_pipe_has_type(crtc, INTEL_OUTPUT_LVDS)) + limit = &intel_limits_i8xx_lvds; + else + limit = &intel_limits_i8xx_dvo; + } + return limit; +} + +/* m1 is reserved as 0 in Pineview, n is a ring counter */ +static void pineview_clock(int refclk, intel_clock_t *clock) +{ + clock->m = clock->m2 + 2; + clock->p = clock->p1 * clock->p2; + clock->vco = refclk * clock->m / clock->n; + clock->dot = clock->vco / clock->p; +} + +static void intel_clock(struct drm_device *dev, int refclk, intel_clock_t *clock) +{ + if (IS_PINEVIEW(dev)) { + pineview_clock(refclk, clock); + return; + } + clock->m = 5 * (clock->m1 + 2) + (clock->m2 + 2); + clock->p = clock->p1 * clock->p2; + clock->vco = refclk * clock->m / (clock->n + 2); + clock->dot = clock->vco / clock->p; +} + +/** + * Returns whether any output on the specified pipe is of the specified type + */ +bool intel_pipe_has_type(struct drm_crtc *crtc, int type) +{ + struct drm_device *dev = crtc->dev; + struct drm_mode_config *mode_config = &dev->mode_config; + struct intel_encoder *encoder; + + list_for_each_entry(encoder, &mode_config->encoder_list, base.head) + if (encoder->base.crtc == crtc && encoder->type == type) + return true; + + return false; +} + +#define INTELPllInvalid(s) do { /* DRM_DEBUG(s); */ return false; } while (0) +/** + * Returns whether the given set of divisors are valid for a given refclk with + * the given connectors. + */ + +static bool intel_PLL_is_valid(struct drm_device *dev, + const intel_limit_t *limit, + const intel_clock_t *clock) +{ + if (clock->p1 < limit->p1.min || limit->p1.max < clock->p1) + INTELPllInvalid ("p1 out of range\n"); + if (clock->p < limit->p.min || limit->p.max < clock->p) + INTELPllInvalid ("p out of range\n"); + if (clock->m2 < limit->m2.min || limit->m2.max < clock->m2) + INTELPllInvalid ("m2 out of range\n"); + if (clock->m1 < limit->m1.min || limit->m1.max < clock->m1) + INTELPllInvalid ("m1 out of range\n"); + if (clock->m1 <= clock->m2 && !IS_PINEVIEW(dev)) + INTELPllInvalid ("m1 <= m2\n"); + if (clock->m < limit->m.min || limit->m.max < clock->m) + INTELPllInvalid ("m out of range\n"); + if (clock->n < limit->n.min || limit->n.max < clock->n) + INTELPllInvalid ("n out of range\n"); + if (clock->vco < limit->vco.min || limit->vco.max < clock->vco) + INTELPllInvalid ("vco out of range\n"); + /* XXX: We may need to be checking "Dot clock" depending on the multiplier, + * connector, etc., rather than just a single range. + */ + if (clock->dot < limit->dot.min || limit->dot.max < clock->dot) + INTELPllInvalid ("dot out of range\n"); + + return true; +} + +static bool +intel_find_best_PLL(const intel_limit_t *limit, struct drm_crtc *crtc, + int target, int refclk, intel_clock_t *best_clock) + +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + intel_clock_t clock; + int err = target; + + if (intel_pipe_has_type(crtc, INTEL_OUTPUT_LVDS) && + (I915_READ(LVDS)) != 0) { + /* + * For LVDS, if the panel is on, just rely on its current + * settings for dual-channel. We haven't figured out how to + * reliably set up different single/dual channel state, if we + * even can. + */ + if ((I915_READ(LVDS) & LVDS_CLKB_POWER_MASK) == + LVDS_CLKB_POWER_UP) + clock.p2 = limit->p2.p2_fast; + else + clock.p2 = limit->p2.p2_slow; + } else { + if (target < limit->p2.dot_limit) + clock.p2 = limit->p2.p2_slow; + else + clock.p2 = limit->p2.p2_fast; + } + + memset (best_clock, 0, sizeof (*best_clock)); + + for (clock.m1 = limit->m1.min; clock.m1 <= limit->m1.max; + clock.m1++) { + for (clock.m2 = limit->m2.min; + clock.m2 <= limit->m2.max; clock.m2++) { + /* m1 is always 0 in Pineview */ + if (clock.m2 >= clock.m1 && !IS_PINEVIEW(dev)) + break; + for (clock.n = limit->n.min; + clock.n <= limit->n.max; clock.n++) { + for (clock.p1 = limit->p1.min; + clock.p1 <= limit->p1.max; clock.p1++) { + int this_err; + + intel_clock(dev, refclk, &clock); + if (!intel_PLL_is_valid(dev, limit, + &clock)) + continue; + + this_err = abs(clock.dot - target); + if (this_err < err) { + *best_clock = clock; + err = this_err; + } + } + } + } + } + + return (err != target); +} + +static bool +intel_g4x_find_best_PLL(const intel_limit_t *limit, struct drm_crtc *crtc, + int target, int refclk, intel_clock_t *best_clock) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + intel_clock_t clock; + int max_n; + bool found; + /* approximately equals target * 0.00585 */ + int err_most = (target >> 8) + (target >> 9); + found = false; + + if (intel_pipe_has_type(crtc, INTEL_OUTPUT_LVDS)) { + int lvds_reg; + + if (HAS_PCH_SPLIT(dev)) + lvds_reg = PCH_LVDS; + else + lvds_reg = LVDS; + if ((I915_READ(lvds_reg) & LVDS_CLKB_POWER_MASK) == + LVDS_CLKB_POWER_UP) + clock.p2 = limit->p2.p2_fast; + else + clock.p2 = limit->p2.p2_slow; + } else { + if (target < limit->p2.dot_limit) + clock.p2 = limit->p2.p2_slow; + else + clock.p2 = limit->p2.p2_fast; + } + + memset(best_clock, 0, sizeof(*best_clock)); + max_n = limit->n.max; + /* based on hardware requirement, prefer smaller n to precision */ + for (clock.n = limit->n.min; clock.n <= max_n; clock.n++) { + /* based on hardware requirement, prefere larger m1,m2 */ + for (clock.m1 = limit->m1.max; + clock.m1 >= limit->m1.min; clock.m1--) { + for (clock.m2 = limit->m2.max; + clock.m2 >= limit->m2.min; clock.m2--) { + for (clock.p1 = limit->p1.max; + clock.p1 >= limit->p1.min; clock.p1--) { + int this_err; + + intel_clock(dev, refclk, &clock); + if (!intel_PLL_is_valid(dev, limit, + &clock)) + continue; + + this_err = abs(clock.dot - target); + if (this_err < err_most) { + *best_clock = clock; + err_most = this_err; + max_n = clock.n; + found = true; + } + } + } + } + } + return found; +} + +static bool +intel_find_pll_ironlake_dp(const intel_limit_t *limit, struct drm_crtc *crtc, + int target, int refclk, intel_clock_t *best_clock) +{ + struct drm_device *dev = crtc->dev; + intel_clock_t clock; + + if (target < 200000) { + clock.n = 1; + clock.p1 = 2; + clock.p2 = 10; + clock.m1 = 12; + clock.m2 = 9; + } else { + clock.n = 2; + clock.p1 = 1; + clock.p2 = 10; + clock.m1 = 14; + clock.m2 = 8; + } + intel_clock(dev, refclk, &clock); + memcpy(best_clock, &clock, sizeof(intel_clock_t)); + return true; +} + +/* DisplayPort has only two frequencies, 162MHz and 270MHz */ +static bool +intel_find_pll_g4x_dp(const intel_limit_t *limit, struct drm_crtc *crtc, + int target, int refclk, intel_clock_t *best_clock) +{ + intel_clock_t clock; + if (target < 200000) { + clock.p1 = 2; + clock.p2 = 10; + clock.n = 2; + clock.m1 = 23; + clock.m2 = 8; + } else { + clock.p1 = 1; + clock.p2 = 10; + clock.n = 1; + clock.m1 = 14; + clock.m2 = 2; + } + clock.m = 5 * (clock.m1 + 2) + (clock.m2 + 2); + clock.p = (clock.p1 * clock.p2); + clock.dot = 96000 * clock.m / (clock.n + 2) / clock.p; + clock.vco = 0; + memcpy(best_clock, &clock, sizeof(intel_clock_t)); + return true; +} + +/** + * intel_wait_for_vblank - wait for vblank on a given pipe + * @dev: drm device + * @pipe: pipe to wait for + * + * Wait for vblank to occur on a given pipe. Needed for various bits of + * mode setting code. + */ +void intel_wait_for_vblank(struct drm_device *dev, int pipe) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + int pipestat_reg = PIPESTAT(pipe); + + /* Clear existing vblank status. Note this will clear any other + * sticky status fields as well. + * + * This races with i915_driver_irq_handler() with the result + * that either function could miss a vblank event. Here it is not + * fatal, as we will either wait upon the next vblank interrupt or + * timeout. Generally speaking intel_wait_for_vblank() is only + * called during modeset at which time the GPU should be idle and + * should *not* be performing page flips and thus not waiting on + * vblanks... + * Currently, the result of us stealing a vblank from the irq + * handler is that a single frame will be skipped during swapbuffers. + */ + I915_WRITE(pipestat_reg, + I915_READ(pipestat_reg) | PIPE_VBLANK_INTERRUPT_STATUS); + + /* Wait for vblank interrupt bit to set */ + if (wait_for(I915_READ(pipestat_reg) & + PIPE_VBLANK_INTERRUPT_STATUS, + 50)) + DRM_DEBUG_KMS("vblank wait timed out\n"); +} + +/* + * intel_wait_for_pipe_off - wait for pipe to turn off + * @dev: drm device + * @pipe: pipe to wait for + * + * After disabling a pipe, we can't wait for vblank in the usual way, + * spinning on the vblank interrupt status bit, since we won't actually + * see an interrupt when the pipe is disabled. + * + * On Gen4 and above: + * wait for the pipe register state bit to turn off + * + * Otherwise: + * wait for the display line value to settle (it usually + * ends up stopping at the start of the next frame). + * + */ +void intel_wait_for_pipe_off(struct drm_device *dev, int pipe) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + if (INTEL_INFO(dev)->gen >= 4) { + int reg = PIPECONF(pipe); + + /* Wait for the Pipe State to go off */ + if (wait_for((I915_READ(reg) & I965_PIPECONF_ACTIVE) == 0, + 100)) + DRM_DEBUG_KMS("pipe_off wait timed out\n"); + } else { + u32 last_line; + int reg = PIPEDSL(pipe); + unsigned long timeout = jiffies + msecs_to_jiffies(100); + + /* Wait for the display line to settle */ + do { + last_line = I915_READ(reg) & DSL_LINEMASK; + mdelay(5); + } while (((I915_READ(reg) & DSL_LINEMASK) != last_line) && + time_after(timeout, jiffies)); + if (time_after(jiffies, timeout)) + DRM_DEBUG_KMS("pipe_off wait timed out\n"); + } +} + +static const char *state_string(bool enabled) +{ + return enabled ? "on" : "off"; +} + +/* Only for pre-ILK configs */ +static void assert_pll(struct drm_i915_private *dev_priv, + enum pipe pipe, bool state) +{ + int reg; + u32 val; + bool cur_state; + + reg = DPLL(pipe); + val = I915_READ(reg); + cur_state = !!(val & DPLL_VCO_ENABLE); + WARN(cur_state != state, + "PLL state assertion failure (expected %s, current %s)\n", + state_string(state), state_string(cur_state)); +} +#define assert_pll_enabled(d, p) assert_pll(d, p, true) +#define assert_pll_disabled(d, p) assert_pll(d, p, false) + +/* For ILK+ */ +static void assert_pch_pll(struct drm_i915_private *dev_priv, + enum pipe pipe, bool state) +{ + int reg; + u32 val; + bool cur_state; + + reg = PCH_DPLL(pipe); + val = I915_READ(reg); + cur_state = !!(val & DPLL_VCO_ENABLE); + WARN(cur_state != state, + "PCH PLL state assertion failure (expected %s, current %s)\n", + state_string(state), state_string(cur_state)); +} +#define assert_pch_pll_enabled(d, p) assert_pch_pll(d, p, true) +#define assert_pch_pll_disabled(d, p) assert_pch_pll(d, p, false) + +static void assert_fdi_tx(struct drm_i915_private *dev_priv, + enum pipe pipe, bool state) +{ + int reg; + u32 val; + bool cur_state; + + reg = FDI_TX_CTL(pipe); + val = I915_READ(reg); + cur_state = !!(val & FDI_TX_ENABLE); + WARN(cur_state != state, + "FDI TX state assertion failure (expected %s, current %s)\n", + state_string(state), state_string(cur_state)); +} +#define assert_fdi_tx_enabled(d, p) assert_fdi_tx(d, p, true) +#define assert_fdi_tx_disabled(d, p) assert_fdi_tx(d, p, false) + +static void assert_fdi_rx(struct drm_i915_private *dev_priv, + enum pipe pipe, bool state) +{ + int reg; + u32 val; + bool cur_state; + + reg = FDI_RX_CTL(pipe); + val = I915_READ(reg); + cur_state = !!(val & FDI_RX_ENABLE); + WARN(cur_state != state, + "FDI RX state assertion failure (expected %s, current %s)\n", + state_string(state), state_string(cur_state)); +} +#define assert_fdi_rx_enabled(d, p) assert_fdi_rx(d, p, true) +#define assert_fdi_rx_disabled(d, p) assert_fdi_rx(d, p, false) + +static void assert_fdi_tx_pll_enabled(struct drm_i915_private *dev_priv, + enum pipe pipe) +{ + int reg; + u32 val; + + /* ILK FDI PLL is always enabled */ + if (dev_priv->info->gen == 5) + return; + + reg = FDI_TX_CTL(pipe); + val = I915_READ(reg); + WARN(!(val & FDI_TX_PLL_ENABLE), "FDI TX PLL assertion failure, should be active but is disabled\n"); +} + +static void assert_fdi_rx_pll_enabled(struct drm_i915_private *dev_priv, + enum pipe pipe) +{ + int reg; + u32 val; + + reg = FDI_RX_CTL(pipe); + val = I915_READ(reg); + WARN(!(val & FDI_RX_PLL_ENABLE), "FDI RX PLL assertion failure, should be active but is disabled\n"); +} + +static void assert_panel_unlocked(struct drm_i915_private *dev_priv, + enum pipe pipe) +{ + int pp_reg, lvds_reg; + u32 val; + enum pipe panel_pipe = PIPE_A; + bool locked = locked; + + if (HAS_PCH_SPLIT(dev_priv->dev)) { + pp_reg = PCH_PP_CONTROL; + lvds_reg = PCH_LVDS; + } else { + pp_reg = PP_CONTROL; + lvds_reg = LVDS; + } + + val = I915_READ(pp_reg); + if (!(val & PANEL_POWER_ON) || + ((val & PANEL_UNLOCK_REGS) == PANEL_UNLOCK_REGS)) + locked = false; + + if (I915_READ(lvds_reg) & LVDS_PIPEB_SELECT) + panel_pipe = PIPE_B; + + WARN(panel_pipe == pipe && locked, + "panel assertion failure, pipe %c regs locked\n", + pipe_name(pipe)); +} + +static void assert_pipe(struct drm_i915_private *dev_priv, + enum pipe pipe, bool state) +{ + int reg; + u32 val; + bool cur_state; + + reg = PIPECONF(pipe); + val = I915_READ(reg); + cur_state = !!(val & PIPECONF_ENABLE); + WARN(cur_state != state, + "pipe %c assertion failure (expected %s, current %s)\n", + pipe_name(pipe), state_string(state), state_string(cur_state)); +} +#define assert_pipe_enabled(d, p) assert_pipe(d, p, true) +#define assert_pipe_disabled(d, p) assert_pipe(d, p, false) + +static void assert_plane_enabled(struct drm_i915_private *dev_priv, + enum plane plane) +{ + int reg; + u32 val; + + reg = DSPCNTR(plane); + val = I915_READ(reg); + WARN(!(val & DISPLAY_PLANE_ENABLE), + "plane %c assertion failure, should be active but is disabled\n", + plane_name(plane)); +} + +static void assert_planes_disabled(struct drm_i915_private *dev_priv, + enum pipe pipe) +{ + int reg, i; + u32 val; + int cur_pipe; + + /* Planes are fixed to pipes on ILK+ */ + if (HAS_PCH_SPLIT(dev_priv->dev)) + return; + + /* Need to check both planes against the pipe */ + for (i = 0; i < 2; i++) { + reg = DSPCNTR(i); + val = I915_READ(reg); + cur_pipe = (val & DISPPLANE_SEL_PIPE_MASK) >> + DISPPLANE_SEL_PIPE_SHIFT; + WARN((val & DISPLAY_PLANE_ENABLE) && pipe == cur_pipe, + "plane %c assertion failure, should be off on pipe %c but is still active\n", + plane_name(i), pipe_name(pipe)); + } +} + +static void assert_pch_refclk_enabled(struct drm_i915_private *dev_priv) +{ + u32 val; + bool enabled; + + val = I915_READ(PCH_DREF_CONTROL); + enabled = !!(val & (DREF_SSC_SOURCE_MASK | DREF_NONSPREAD_SOURCE_MASK | + DREF_SUPERSPREAD_SOURCE_MASK)); + WARN(!enabled, "PCH refclk assertion failure, should be active but is disabled\n"); +} + +static void assert_transcoder_disabled(struct drm_i915_private *dev_priv, + enum pipe pipe) +{ + int reg; + u32 val; + bool enabled; + + reg = TRANSCONF(pipe); + val = I915_READ(reg); + enabled = !!(val & TRANS_ENABLE); + WARN(enabled, + "transcoder assertion failed, should be off on pipe %c but is still active\n", + pipe_name(pipe)); +} + +static void assert_pch_dp_disabled(struct drm_i915_private *dev_priv, + enum pipe pipe, int reg) +{ + u32 val = I915_READ(reg); + WARN(DP_PIPE_ENABLED(val, pipe), + "PCH DP (0x%08x) enabled on transcoder %c, should be disabled\n", + reg, pipe_name(pipe)); +} + +static void assert_pch_hdmi_disabled(struct drm_i915_private *dev_priv, + enum pipe pipe, int reg) +{ + u32 val = I915_READ(reg); + WARN(HDMI_PIPE_ENABLED(val, pipe), + "PCH DP (0x%08x) enabled on transcoder %c, should be disabled\n", + reg, pipe_name(pipe)); +} + +static void assert_pch_ports_disabled(struct drm_i915_private *dev_priv, + enum pipe pipe) +{ + int reg; + u32 val; + + assert_pch_dp_disabled(dev_priv, pipe, PCH_DP_B); + assert_pch_dp_disabled(dev_priv, pipe, PCH_DP_C); + assert_pch_dp_disabled(dev_priv, pipe, PCH_DP_D); + + reg = PCH_ADPA; + val = I915_READ(reg); + WARN(ADPA_PIPE_ENABLED(val, pipe), + "PCH VGA enabled on transcoder %c, should be disabled\n", + pipe_name(pipe)); + + reg = PCH_LVDS; + val = I915_READ(reg); + WARN(LVDS_PIPE_ENABLED(val, pipe), + "PCH LVDS enabled on transcoder %c, should be disabled\n", + pipe_name(pipe)); + + assert_pch_hdmi_disabled(dev_priv, pipe, HDMIB); + assert_pch_hdmi_disabled(dev_priv, pipe, HDMIC); + assert_pch_hdmi_disabled(dev_priv, pipe, HDMID); +} + +/** + * intel_enable_pll - enable a PLL + * @dev_priv: i915 private structure + * @pipe: pipe PLL to enable + * + * Enable @pipe's PLL so we can start pumping pixels from a plane. Check to + * make sure the PLL reg is writable first though, since the panel write + * protect mechanism may be enabled. + * + * Note! This is for pre-ILK only. + */ +static void intel_enable_pll(struct drm_i915_private *dev_priv, enum pipe pipe) +{ + int reg; + u32 val; + + /* No really, not for ILK+ */ + BUG_ON(dev_priv->info->gen >= 5); + + /* PLL is protected by panel, make sure we can write it */ + if (IS_MOBILE(dev_priv->dev) && !IS_I830(dev_priv->dev)) + assert_panel_unlocked(dev_priv, pipe); + + reg = DPLL(pipe); + val = I915_READ(reg); + val |= DPLL_VCO_ENABLE; + + /* We do this three times for luck */ + I915_WRITE(reg, val); + POSTING_READ(reg); + udelay(150); /* wait for warmup */ + I915_WRITE(reg, val); + POSTING_READ(reg); + udelay(150); /* wait for warmup */ + I915_WRITE(reg, val); + POSTING_READ(reg); + udelay(150); /* wait for warmup */ +} + +/** + * intel_disable_pll - disable a PLL + * @dev_priv: i915 private structure + * @pipe: pipe PLL to disable + * + * Disable the PLL for @pipe, making sure the pipe is off first. + * + * Note! This is for pre-ILK only. + */ +static void intel_disable_pll(struct drm_i915_private *dev_priv, enum pipe pipe) +{ + int reg; + u32 val; + + /* Don't disable pipe A or pipe A PLLs if needed */ + if (pipe == PIPE_A && (dev_priv->quirks & QUIRK_PIPEA_FORCE)) + return; + + /* Make sure the pipe isn't still relying on us */ + assert_pipe_disabled(dev_priv, pipe); + + reg = DPLL(pipe); + val = I915_READ(reg); + val &= ~DPLL_VCO_ENABLE; + I915_WRITE(reg, val); + POSTING_READ(reg); +} + +/** + * intel_enable_pch_pll - enable PCH PLL + * @dev_priv: i915 private structure + * @pipe: pipe PLL to enable + * + * The PCH PLL needs to be enabled before the PCH transcoder, since it + * drives the transcoder clock. + */ +static void intel_enable_pch_pll(struct drm_i915_private *dev_priv, + enum pipe pipe) +{ + int reg; + u32 val; + + /* PCH only available on ILK+ */ + BUG_ON(dev_priv->info->gen < 5); + + /* PCH refclock must be enabled first */ + assert_pch_refclk_enabled(dev_priv); + + reg = PCH_DPLL(pipe); + val = I915_READ(reg); + val |= DPLL_VCO_ENABLE; + I915_WRITE(reg, val); + POSTING_READ(reg); + udelay(200); +} + +static void intel_disable_pch_pll(struct drm_i915_private *dev_priv, + enum pipe pipe) +{ + int reg; + u32 val; + + /* PCH only available on ILK+ */ + BUG_ON(dev_priv->info->gen < 5); + + /* Make sure transcoder isn't still depending on us */ + assert_transcoder_disabled(dev_priv, pipe); + + reg = PCH_DPLL(pipe); + val = I915_READ(reg); + val &= ~DPLL_VCO_ENABLE; + I915_WRITE(reg, val); + POSTING_READ(reg); + udelay(200); +} + +static void intel_enable_transcoder(struct drm_i915_private *dev_priv, + enum pipe pipe) +{ + int reg; + u32 val; + + /* PCH only available on ILK+ */ + BUG_ON(dev_priv->info->gen < 5); + + /* Make sure PCH DPLL is enabled */ + assert_pch_pll_enabled(dev_priv, pipe); + + /* FDI must be feeding us bits for PCH ports */ + assert_fdi_tx_enabled(dev_priv, pipe); + assert_fdi_rx_enabled(dev_priv, pipe); + + reg = TRANSCONF(pipe); + val = I915_READ(reg); + /* + * make the BPC in transcoder be consistent with + * that in pipeconf reg. + */ + val &= ~PIPE_BPC_MASK; + val |= I915_READ(PIPECONF(pipe)) & PIPE_BPC_MASK; + I915_WRITE(reg, val | TRANS_ENABLE); + if (wait_for(I915_READ(reg) & TRANS_STATE_ENABLE, 100)) + DRM_ERROR("failed to enable transcoder %d\n", pipe); +} + +static void intel_disable_transcoder(struct drm_i915_private *dev_priv, + enum pipe pipe) +{ + int reg; + u32 val; + + /* FDI relies on the transcoder */ + assert_fdi_tx_disabled(dev_priv, pipe); + assert_fdi_rx_disabled(dev_priv, pipe); + + /* Ports must be off as well */ + assert_pch_ports_disabled(dev_priv, pipe); + + reg = TRANSCONF(pipe); + val = I915_READ(reg); + val &= ~TRANS_ENABLE; + I915_WRITE(reg, val); + /* wait for PCH transcoder off, transcoder state */ + if (wait_for((I915_READ(reg) & TRANS_STATE_ENABLE) == 0, 50)) + DRM_ERROR("failed to disable transcoder\n"); +} + +/** + * intel_enable_pipe - enable a pipe, asserting requirements + * @dev_priv: i915 private structure + * @pipe: pipe to enable + * @pch_port: on ILK+, is this pipe driving a PCH port or not + * + * Enable @pipe, making sure that various hardware specific requirements + * are met, if applicable, e.g. PLL enabled, LVDS pairs enabled, etc. + * + * @pipe should be %PIPE_A or %PIPE_B. + * + * Will wait until the pipe is actually running (i.e. first vblank) before + * returning. + */ +static void intel_enable_pipe(struct drm_i915_private *dev_priv, enum pipe pipe, + bool pch_port) +{ + int reg; + u32 val; + + /* + * A pipe without a PLL won't actually be able to drive bits from + * a plane. On ILK+ the pipe PLLs are integrated, so we don't + * need the check. + */ + if (!HAS_PCH_SPLIT(dev_priv->dev)) + assert_pll_enabled(dev_priv, pipe); + else { + if (pch_port) { + /* if driving the PCH, we need FDI enabled */ + assert_fdi_rx_pll_enabled(dev_priv, pipe); + assert_fdi_tx_pll_enabled(dev_priv, pipe); + } + /* FIXME: assert CPU port conditions for SNB+ */ + } + + reg = PIPECONF(pipe); + val = I915_READ(reg); + if (val & PIPECONF_ENABLE) + return; + + I915_WRITE(reg, val | PIPECONF_ENABLE); + intel_wait_for_vblank(dev_priv->dev, pipe); +} + +/** + * intel_disable_pipe - disable a pipe, asserting requirements + * @dev_priv: i915 private structure + * @pipe: pipe to disable + * + * Disable @pipe, making sure that various hardware specific requirements + * are met, if applicable, e.g. plane disabled, panel fitter off, etc. + * + * @pipe should be %PIPE_A or %PIPE_B. + * + * Will wait until the pipe has shut down before returning. + */ +static void intel_disable_pipe(struct drm_i915_private *dev_priv, + enum pipe pipe) +{ + int reg; + u32 val; + + /* + * Make sure planes won't keep trying to pump pixels to us, + * or we might hang the display. + */ + assert_planes_disabled(dev_priv, pipe); + + /* Don't disable pipe A or pipe A PLLs if needed */ + if (pipe == PIPE_A && (dev_priv->quirks & QUIRK_PIPEA_FORCE)) + return; + + reg = PIPECONF(pipe); + val = I915_READ(reg); + if ((val & PIPECONF_ENABLE) == 0) + return; + + I915_WRITE(reg, val & ~PIPECONF_ENABLE); + intel_wait_for_pipe_off(dev_priv->dev, pipe); +} + +/** + * intel_enable_plane - enable a display plane on a given pipe + * @dev_priv: i915 private structure + * @plane: plane to enable + * @pipe: pipe being fed + * + * Enable @plane on @pipe, making sure that @pipe is running first. + */ +static void intel_enable_plane(struct drm_i915_private *dev_priv, + enum plane plane, enum pipe pipe) +{ + int reg; + u32 val; + + /* If the pipe isn't enabled, we can't pump pixels and may hang */ + assert_pipe_enabled(dev_priv, pipe); + + reg = DSPCNTR(plane); + val = I915_READ(reg); + if (val & DISPLAY_PLANE_ENABLE) + return; + + I915_WRITE(reg, val | DISPLAY_PLANE_ENABLE); + intel_wait_for_vblank(dev_priv->dev, pipe); +} + +/* + * Plane regs are double buffered, going from enabled->disabled needs a + * trigger in order to latch. The display address reg provides this. + */ +static void intel_flush_display_plane(struct drm_i915_private *dev_priv, + enum plane plane) +{ + u32 reg = DSPADDR(plane); + I915_WRITE(reg, I915_READ(reg)); +} + +/** + * intel_disable_plane - disable a display plane + * @dev_priv: i915 private structure + * @plane: plane to disable + * @pipe: pipe consuming the data + * + * Disable @plane; should be an independent operation. + */ +static void intel_disable_plane(struct drm_i915_private *dev_priv, + enum plane plane, enum pipe pipe) +{ + int reg; + u32 val; + + reg = DSPCNTR(plane); + val = I915_READ(reg); + if ((val & DISPLAY_PLANE_ENABLE) == 0) + return; + + I915_WRITE(reg, val & ~DISPLAY_PLANE_ENABLE); + intel_flush_display_plane(dev_priv, plane); + intel_wait_for_vblank(dev_priv->dev, pipe); +} + +static void disable_pch_dp(struct drm_i915_private *dev_priv, + enum pipe pipe, int reg) +{ + u32 val = I915_READ(reg); + if (DP_PIPE_ENABLED(val, pipe)) + I915_WRITE(reg, val & ~DP_PORT_EN); +} + +static void disable_pch_hdmi(struct drm_i915_private *dev_priv, + enum pipe pipe, int reg) +{ + u32 val = I915_READ(reg); + if (HDMI_PIPE_ENABLED(val, pipe)) + I915_WRITE(reg, val & ~PORT_ENABLE); +} + +/* Disable any ports connected to this transcoder */ +static void intel_disable_pch_ports(struct drm_i915_private *dev_priv, + enum pipe pipe) +{ + u32 reg, val; + + val = I915_READ(PCH_PP_CONTROL); + I915_WRITE(PCH_PP_CONTROL, val | PANEL_UNLOCK_REGS); + + disable_pch_dp(dev_priv, pipe, PCH_DP_B); + disable_pch_dp(dev_priv, pipe, PCH_DP_C); + disable_pch_dp(dev_priv, pipe, PCH_DP_D); + + reg = PCH_ADPA; + val = I915_READ(reg); + if (ADPA_PIPE_ENABLED(val, pipe)) + I915_WRITE(reg, val & ~ADPA_DAC_ENABLE); + + reg = PCH_LVDS; + val = I915_READ(reg); + if (LVDS_PIPE_ENABLED(val, pipe)) { + I915_WRITE(reg, val & ~LVDS_PORT_EN); + POSTING_READ(reg); + udelay(100); + } + + disable_pch_hdmi(dev_priv, pipe, HDMIB); + disable_pch_hdmi(dev_priv, pipe, HDMIC); + disable_pch_hdmi(dev_priv, pipe, HDMID); +} + +static void i8xx_enable_fbc(struct drm_crtc *crtc, unsigned long interval) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_framebuffer *fb = crtc->fb; + struct intel_framebuffer *intel_fb = to_intel_framebuffer(fb); + struct drm_i915_gem_object *obj = intel_fb->obj; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int plane, i; + u32 fbc_ctl, fbc_ctl2; + + if (fb->pitch == dev_priv->cfb_pitch && + obj->fence_reg == dev_priv->cfb_fence && + intel_crtc->plane == dev_priv->cfb_plane && + I915_READ(FBC_CONTROL) & FBC_CTL_EN) + return; + + i8xx_disable_fbc(dev); + + dev_priv->cfb_pitch = dev_priv->cfb_size / FBC_LL_SIZE; + + if (fb->pitch < dev_priv->cfb_pitch) + dev_priv->cfb_pitch = fb->pitch; + + /* FBC_CTL wants 64B units */ + dev_priv->cfb_pitch = (dev_priv->cfb_pitch / 64) - 1; + dev_priv->cfb_fence = obj->fence_reg; + dev_priv->cfb_plane = intel_crtc->plane; + plane = dev_priv->cfb_plane == 0 ? FBC_CTL_PLANEA : FBC_CTL_PLANEB; + + /* Clear old tags */ + for (i = 0; i < (FBC_LL_SIZE / 32) + 1; i++) + I915_WRITE(FBC_TAG + (i * 4), 0); + + /* Set it up... */ + fbc_ctl2 = FBC_CTL_FENCE_DBL | FBC_CTL_IDLE_IMM | plane; + if (obj->tiling_mode != I915_TILING_NONE) + fbc_ctl2 |= FBC_CTL_CPU_FENCE; + I915_WRITE(FBC_CONTROL2, fbc_ctl2); + I915_WRITE(FBC_FENCE_OFF, crtc->y); + + /* enable it... */ + fbc_ctl = FBC_CTL_EN | FBC_CTL_PERIODIC; + if (IS_I945GM(dev)) + fbc_ctl |= FBC_CTL_C3_IDLE; /* 945 needs special SR handling */ + fbc_ctl |= (dev_priv->cfb_pitch & 0xff) << FBC_CTL_STRIDE_SHIFT; + fbc_ctl |= (interval & 0x2fff) << FBC_CTL_INTERVAL_SHIFT; + if (obj->tiling_mode != I915_TILING_NONE) + fbc_ctl |= dev_priv->cfb_fence; + I915_WRITE(FBC_CONTROL, fbc_ctl); + + DRM_DEBUG_KMS("enabled FBC, pitch %ld, yoff %d, plane %d, ", + dev_priv->cfb_pitch, crtc->y, dev_priv->cfb_plane); +} + +void i8xx_disable_fbc(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u32 fbc_ctl; + + /* Disable compression */ + fbc_ctl = I915_READ(FBC_CONTROL); + if ((fbc_ctl & FBC_CTL_EN) == 0) + return; + + fbc_ctl &= ~FBC_CTL_EN; + I915_WRITE(FBC_CONTROL, fbc_ctl); + + /* Wait for compressing bit to clear */ + if (wait_for((I915_READ(FBC_STATUS) & FBC_STAT_COMPRESSING) == 0, 10)) { + DRM_DEBUG_KMS("FBC idle timed out\n"); + return; + } + + DRM_DEBUG_KMS("disabled FBC\n"); +} + +static bool i8xx_fbc_enabled(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + return I915_READ(FBC_CONTROL) & FBC_CTL_EN; +} + +static void g4x_enable_fbc(struct drm_crtc *crtc, unsigned long interval) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_framebuffer *fb = crtc->fb; + struct intel_framebuffer *intel_fb = to_intel_framebuffer(fb); + struct drm_i915_gem_object *obj = intel_fb->obj; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int plane = intel_crtc->plane == 0 ? DPFC_CTL_PLANEA : DPFC_CTL_PLANEB; + unsigned long stall_watermark = 200; + u32 dpfc_ctl; + + dpfc_ctl = I915_READ(DPFC_CONTROL); + if (dpfc_ctl & DPFC_CTL_EN) { + if (dev_priv->cfb_pitch == dev_priv->cfb_pitch / 64 - 1 && + dev_priv->cfb_fence == obj->fence_reg && + dev_priv->cfb_plane == intel_crtc->plane && + dev_priv->cfb_y == crtc->y) + return; + + I915_WRITE(DPFC_CONTROL, dpfc_ctl & ~DPFC_CTL_EN); + intel_wait_for_vblank(dev, intel_crtc->pipe); + } + + dev_priv->cfb_pitch = (dev_priv->cfb_pitch / 64) - 1; + dev_priv->cfb_fence = obj->fence_reg; + dev_priv->cfb_plane = intel_crtc->plane; + dev_priv->cfb_y = crtc->y; + + dpfc_ctl = plane | DPFC_SR_EN | DPFC_CTL_LIMIT_1X; + if (obj->tiling_mode != I915_TILING_NONE) { + dpfc_ctl |= DPFC_CTL_FENCE_EN | dev_priv->cfb_fence; + I915_WRITE(DPFC_CHICKEN, DPFC_HT_MODIFY); + } else { + I915_WRITE(DPFC_CHICKEN, ~DPFC_HT_MODIFY); + } + + I915_WRITE(DPFC_RECOMP_CTL, DPFC_RECOMP_STALL_EN | + (stall_watermark << DPFC_RECOMP_STALL_WM_SHIFT) | + (interval << DPFC_RECOMP_TIMER_COUNT_SHIFT)); + I915_WRITE(DPFC_FENCE_YOFF, crtc->y); + + /* enable it... */ + I915_WRITE(DPFC_CONTROL, I915_READ(DPFC_CONTROL) | DPFC_CTL_EN); + + DRM_DEBUG_KMS("enabled fbc on plane %d\n", intel_crtc->plane); +} + +void g4x_disable_fbc(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u32 dpfc_ctl; + + /* Disable compression */ + dpfc_ctl = I915_READ(DPFC_CONTROL); + if (dpfc_ctl & DPFC_CTL_EN) { + dpfc_ctl &= ~DPFC_CTL_EN; + I915_WRITE(DPFC_CONTROL, dpfc_ctl); + + DRM_DEBUG_KMS("disabled FBC\n"); + } +} + +static bool g4x_fbc_enabled(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + return I915_READ(DPFC_CONTROL) & DPFC_CTL_EN; +} + +static void sandybridge_blit_fbc_update(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u32 blt_ecoskpd; + + /* Make sure blitter notifies FBC of writes */ + gen6_gt_force_wake_get(dev_priv); + blt_ecoskpd = I915_READ(GEN6_BLITTER_ECOSKPD); + blt_ecoskpd |= GEN6_BLITTER_FBC_NOTIFY << + GEN6_BLITTER_LOCK_SHIFT; + I915_WRITE(GEN6_BLITTER_ECOSKPD, blt_ecoskpd); + blt_ecoskpd |= GEN6_BLITTER_FBC_NOTIFY; + I915_WRITE(GEN6_BLITTER_ECOSKPD, blt_ecoskpd); + blt_ecoskpd &= ~(GEN6_BLITTER_FBC_NOTIFY << + GEN6_BLITTER_LOCK_SHIFT); + I915_WRITE(GEN6_BLITTER_ECOSKPD, blt_ecoskpd); + POSTING_READ(GEN6_BLITTER_ECOSKPD); + gen6_gt_force_wake_put(dev_priv); +} + +static void ironlake_enable_fbc(struct drm_crtc *crtc, unsigned long interval) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_framebuffer *fb = crtc->fb; + struct intel_framebuffer *intel_fb = to_intel_framebuffer(fb); + struct drm_i915_gem_object *obj = intel_fb->obj; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int plane = intel_crtc->plane == 0 ? DPFC_CTL_PLANEA : DPFC_CTL_PLANEB; + unsigned long stall_watermark = 200; + u32 dpfc_ctl; + + dpfc_ctl = I915_READ(ILK_DPFC_CONTROL); + if (dpfc_ctl & DPFC_CTL_EN) { + if (dev_priv->cfb_pitch == dev_priv->cfb_pitch / 64 - 1 && + dev_priv->cfb_fence == obj->fence_reg && + dev_priv->cfb_plane == intel_crtc->plane && + dev_priv->cfb_offset == obj->gtt_offset && + dev_priv->cfb_y == crtc->y) + return; + + I915_WRITE(ILK_DPFC_CONTROL, dpfc_ctl & ~DPFC_CTL_EN); + intel_wait_for_vblank(dev, intel_crtc->pipe); + } + + dev_priv->cfb_pitch = (dev_priv->cfb_pitch / 64) - 1; + dev_priv->cfb_fence = obj->fence_reg; + dev_priv->cfb_plane = intel_crtc->plane; + dev_priv->cfb_offset = obj->gtt_offset; + dev_priv->cfb_y = crtc->y; + + dpfc_ctl &= DPFC_RESERVED; + dpfc_ctl |= (plane | DPFC_CTL_LIMIT_1X); + if (obj->tiling_mode != I915_TILING_NONE) { + dpfc_ctl |= (DPFC_CTL_FENCE_EN | dev_priv->cfb_fence); + I915_WRITE(ILK_DPFC_CHICKEN, DPFC_HT_MODIFY); + } else { + I915_WRITE(ILK_DPFC_CHICKEN, ~DPFC_HT_MODIFY); + } + + I915_WRITE(ILK_DPFC_RECOMP_CTL, DPFC_RECOMP_STALL_EN | + (stall_watermark << DPFC_RECOMP_STALL_WM_SHIFT) | + (interval << DPFC_RECOMP_TIMER_COUNT_SHIFT)); + I915_WRITE(ILK_DPFC_FENCE_YOFF, crtc->y); + I915_WRITE(ILK_FBC_RT_BASE, obj->gtt_offset | ILK_FBC_RT_VALID); + /* enable it... */ + I915_WRITE(ILK_DPFC_CONTROL, dpfc_ctl | DPFC_CTL_EN); + + if (IS_GEN6(dev)) { + I915_WRITE(SNB_DPFC_CTL_SA, + SNB_CPU_FENCE_ENABLE | dev_priv->cfb_fence); + I915_WRITE(DPFC_CPU_FENCE_OFFSET, crtc->y); + sandybridge_blit_fbc_update(dev); + } + + DRM_DEBUG_KMS("enabled fbc on plane %d\n", intel_crtc->plane); +} + +void ironlake_disable_fbc(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u32 dpfc_ctl; + + /* Disable compression */ + dpfc_ctl = I915_READ(ILK_DPFC_CONTROL); + if (dpfc_ctl & DPFC_CTL_EN) { + dpfc_ctl &= ~DPFC_CTL_EN; + I915_WRITE(ILK_DPFC_CONTROL, dpfc_ctl); + + DRM_DEBUG_KMS("disabled FBC\n"); + } +} + +static bool ironlake_fbc_enabled(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + return I915_READ(ILK_DPFC_CONTROL) & DPFC_CTL_EN; +} + +bool intel_fbc_enabled(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + if (!dev_priv->display.fbc_enabled) + return false; + + return dev_priv->display.fbc_enabled(dev); +} + +void intel_enable_fbc(struct drm_crtc *crtc, unsigned long interval) +{ + struct drm_i915_private *dev_priv = crtc->dev->dev_private; + + if (!dev_priv->display.enable_fbc) + return; + + dev_priv->display.enable_fbc(crtc, interval); +} + +void intel_disable_fbc(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + if (!dev_priv->display.disable_fbc) + return; + + dev_priv->display.disable_fbc(dev); +} + +/** + * intel_update_fbc - enable/disable FBC as needed + * @dev: the drm_device + * + * Set up the framebuffer compression hardware at mode set time. We + * enable it if possible: + * - plane A only (on pre-965) + * - no pixel mulitply/line duplication + * - no alpha buffer discard + * - no dual wide + * - framebuffer <= 2048 in width, 1536 in height + * + * We can't assume that any compression will take place (worst case), + * so the compressed buffer has to be the same size as the uncompressed + * one. It also must reside (along with the line length buffer) in + * stolen memory. + * + * We need to enable/disable FBC on a global basis. + */ +static void intel_update_fbc(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_crtc *crtc = NULL, *tmp_crtc; + struct intel_crtc *intel_crtc; + struct drm_framebuffer *fb; + struct intel_framebuffer *intel_fb; + struct drm_i915_gem_object *obj; + + DRM_DEBUG_KMS("\n"); + + if (!i915_powersave) + return; + + if (!I915_HAS_FBC(dev)) + return; + + /* + * If FBC is already on, we just have to verify that we can + * keep it that way... + * Need to disable if: + * - more than one pipe is active + * - changing FBC params (stride, fence, mode) + * - new fb is too large to fit in compressed buffer + * - going to an unsupported config (interlace, pixel multiply, etc.) + */ + list_for_each_entry(tmp_crtc, &dev->mode_config.crtc_list, head) { + if (tmp_crtc->enabled && tmp_crtc->fb) { + if (crtc) { + DRM_DEBUG_KMS("more than one pipe active, disabling compression\n"); + dev_priv->no_fbc_reason = FBC_MULTIPLE_PIPES; + goto out_disable; + } + crtc = tmp_crtc; + } + } + + if (!crtc || crtc->fb == NULL) { + DRM_DEBUG_KMS("no output, disabling\n"); + dev_priv->no_fbc_reason = FBC_NO_OUTPUT; + goto out_disable; + } + + intel_crtc = to_intel_crtc(crtc); + fb = crtc->fb; + intel_fb = to_intel_framebuffer(fb); + obj = intel_fb->obj; + + if (!i915_enable_fbc) { + DRM_DEBUG_KMS("fbc disabled per module param (default off)\n"); + dev_priv->no_fbc_reason = FBC_MODULE_PARAM; + goto out_disable; + } + if (intel_fb->obj->base.size > dev_priv->cfb_size) { + DRM_DEBUG_KMS("framebuffer too large, disabling " + "compression\n"); + dev_priv->no_fbc_reason = FBC_STOLEN_TOO_SMALL; + goto out_disable; + } + if ((crtc->mode.flags & DRM_MODE_FLAG_INTERLACE) || + (crtc->mode.flags & DRM_MODE_FLAG_DBLSCAN)) { + DRM_DEBUG_KMS("mode incompatible with compression, " + "disabling\n"); + dev_priv->no_fbc_reason = FBC_UNSUPPORTED_MODE; + goto out_disable; + } + if ((crtc->mode.hdisplay > 2048) || + (crtc->mode.vdisplay > 1536)) { + DRM_DEBUG_KMS("mode too large for compression, disabling\n"); + dev_priv->no_fbc_reason = FBC_MODE_TOO_LARGE; + goto out_disable; + } + if ((IS_I915GM(dev) || IS_I945GM(dev)) && intel_crtc->plane != 0) { + DRM_DEBUG_KMS("plane not 0, disabling compression\n"); + dev_priv->no_fbc_reason = FBC_BAD_PLANE; + goto out_disable; + } + if (obj->tiling_mode != I915_TILING_X) { + DRM_DEBUG_KMS("framebuffer not tiled, disabling compression\n"); + dev_priv->no_fbc_reason = FBC_NOT_TILED; + goto out_disable; + } + + /* If the kernel debugger is active, always disable compression */ + if (in_dbg_master()) + goto out_disable; + + intel_enable_fbc(crtc, 500); + return; + +out_disable: + /* Multiple disables should be harmless */ + if (intel_fbc_enabled(dev)) { + DRM_DEBUG_KMS("unsupported config, disabling FBC\n"); + intel_disable_fbc(dev); + } +} + +int +intel_pin_and_fence_fb_obj(struct drm_device *dev, + struct drm_i915_gem_object *obj, + struct intel_ring_buffer *pipelined) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u32 alignment; + int ret; + + switch (obj->tiling_mode) { + case I915_TILING_NONE: + if (IS_BROADWATER(dev) || IS_CRESTLINE(dev)) + alignment = 128 * 1024; + else if (INTEL_INFO(dev)->gen >= 4) + alignment = 4 * 1024; + else + alignment = 64 * 1024; + break; + case I915_TILING_X: + /* pin() will align the object as required by fence */ + alignment = 0; + break; + case I915_TILING_Y: + /* FIXME: Is this true? */ + DRM_ERROR("Y tiled not allowed for scan out buffers\n"); + return -EINVAL; + default: + BUG(); + } + + dev_priv->mm.interruptible = false; + ret = i915_gem_object_pin(obj, alignment, true); + if (ret) + goto err_interruptible; + + ret = i915_gem_object_set_to_display_plane(obj, pipelined); + if (ret) + goto err_unpin; + + /* Install a fence for tiled scan-out. Pre-i965 always needs a + * fence, whereas 965+ only requires a fence if using + * framebuffer compression. For simplicity, we always install + * a fence as the cost is not that onerous. + */ + if (obj->tiling_mode != I915_TILING_NONE) { + ret = i915_gem_object_get_fence(obj, pipelined); + if (ret) + goto err_unpin; + } + + dev_priv->mm.interruptible = true; + return 0; + +err_unpin: + i915_gem_object_unpin(obj); +err_interruptible: + dev_priv->mm.interruptible = true; + return ret; +} + +/* Assume fb object is pinned & idle & fenced and just update base pointers */ +static int +intel_pipe_set_base_atomic(struct drm_crtc *crtc, struct drm_framebuffer *fb, + int x, int y, enum mode_set_atomic state) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + struct intel_framebuffer *intel_fb; + struct drm_i915_gem_object *obj; + int plane = intel_crtc->plane; + unsigned long Start, Offset; + u32 dspcntr; + u32 reg; + + switch (plane) { + case 0: + case 1: + break; + default: + DRM_ERROR("Can't update plane %d in SAREA\n", plane); + return -EINVAL; + } + + intel_fb = to_intel_framebuffer(fb); + obj = intel_fb->obj; + + reg = DSPCNTR(plane); + dspcntr = I915_READ(reg); + /* Mask out pixel format bits in case we change it */ + dspcntr &= ~DISPPLANE_PIXFORMAT_MASK; + switch (fb->bits_per_pixel) { + case 8: + dspcntr |= DISPPLANE_8BPP; + break; + case 16: + if (fb->depth == 15) + dspcntr |= DISPPLANE_15_16BPP; + else + dspcntr |= DISPPLANE_16BPP; + break; + case 24: + case 32: + dspcntr |= DISPPLANE_32BPP_NO_ALPHA; + break; + default: + DRM_ERROR("Unknown color depth\n"); + return -EINVAL; + } + if (INTEL_INFO(dev)->gen >= 4) { + if (obj->tiling_mode != I915_TILING_NONE) + dspcntr |= DISPPLANE_TILED; + else + dspcntr &= ~DISPPLANE_TILED; + } + + if (HAS_PCH_SPLIT(dev)) + /* must disable */ + dspcntr |= DISPPLANE_TRICKLE_FEED_DISABLE; + + I915_WRITE(reg, dspcntr); + + Start = obj->gtt_offset; + Offset = y * fb->pitch + x * (fb->bits_per_pixel / 8); + + DRM_DEBUG_KMS("Writing base %08lX %08lX %d %d %d\n", + Start, Offset, x, y, fb->pitch); + I915_WRITE(DSPSTRIDE(plane), fb->pitch); + if (INTEL_INFO(dev)->gen >= 4) { + I915_WRITE(DSPSURF(plane), Start); + I915_WRITE(DSPTILEOFF(plane), (y << 16) | x); + I915_WRITE(DSPADDR(plane), Offset); + } else + I915_WRITE(DSPADDR(plane), Start + Offset); + POSTING_READ(reg); + + intel_update_fbc(dev); + intel_increase_pllclock(crtc); + + return 0; +} + +static int +intel_pipe_set_base(struct drm_crtc *crtc, int x, int y, + struct drm_framebuffer *old_fb) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_master_private *master_priv; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int ret; + + /* no fb bound */ + if (!crtc->fb) { + DRM_DEBUG_KMS("No FB bound\n"); + return 0; + } + + switch (intel_crtc->plane) { + case 0: + case 1: + break; + default: + return -EINVAL; + } + + mutex_lock(&dev->struct_mutex); + ret = intel_pin_and_fence_fb_obj(dev, + to_intel_framebuffer(crtc->fb)->obj, + NULL); + if (ret != 0) { + mutex_unlock(&dev->struct_mutex); + return ret; + } + + if (old_fb) { + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_i915_gem_object *obj = to_intel_framebuffer(old_fb)->obj; + + wait_event(dev_priv->pending_flip_queue, + atomic_read(&dev_priv->mm.wedged) || + atomic_read(&obj->pending_flip) == 0); + + /* Big Hammer, we also need to ensure that any pending + * MI_WAIT_FOR_EVENT inside a user batch buffer on the + * current scanout is retired before unpinning the old + * framebuffer. + * + * This should only fail upon a hung GPU, in which case we + * can safely continue. + */ + ret = i915_gem_object_flush_gpu(obj); + (void) ret; + } + + ret = intel_pipe_set_base_atomic(crtc, crtc->fb, x, y, + LEAVE_ATOMIC_MODE_SET); + if (ret) { + i915_gem_object_unpin(to_intel_framebuffer(crtc->fb)->obj); + mutex_unlock(&dev->struct_mutex); + return ret; + } + + if (old_fb) { + intel_wait_for_vblank(dev, intel_crtc->pipe); + i915_gem_object_unpin(to_intel_framebuffer(old_fb)->obj); + } + + mutex_unlock(&dev->struct_mutex); + + if (!dev->primary->master) + return 0; + + master_priv = dev->primary->master->driver_priv; + if (!master_priv->sarea_priv) + return 0; + + if (intel_crtc->pipe) { + master_priv->sarea_priv->pipeB_x = x; + master_priv->sarea_priv->pipeB_y = y; + } else { + master_priv->sarea_priv->pipeA_x = x; + master_priv->sarea_priv->pipeA_y = y; + } + + return 0; +} + +static void ironlake_set_pll_edp(struct drm_crtc *crtc, int clock) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + u32 dpa_ctl; + + DRM_DEBUG_KMS("eDP PLL enable for clock %d\n", clock); + dpa_ctl = I915_READ(DP_A); + dpa_ctl &= ~DP_PLL_FREQ_MASK; + + if (clock < 200000) { + u32 temp; + dpa_ctl |= DP_PLL_FREQ_160MHZ; + /* workaround for 160Mhz: + 1) program 0x4600c bits 15:0 = 0x8124 + 2) program 0x46010 bit 0 = 1 + 3) program 0x46034 bit 24 = 1 + 4) program 0x64000 bit 14 = 1 + */ + temp = I915_READ(0x4600c); + temp &= 0xffff0000; + I915_WRITE(0x4600c, temp | 0x8124); + + temp = I915_READ(0x46010); + I915_WRITE(0x46010, temp | 1); + + temp = I915_READ(0x46034); + I915_WRITE(0x46034, temp | (1 << 24)); + } else { + dpa_ctl |= DP_PLL_FREQ_270MHZ; + } + I915_WRITE(DP_A, dpa_ctl); + + POSTING_READ(DP_A); + udelay(500); +} + +static void intel_fdi_normal_train(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int pipe = intel_crtc->pipe; + u32 reg, temp; + + /* enable normal train */ + reg = FDI_TX_CTL(pipe); + temp = I915_READ(reg); + if (IS_IVYBRIDGE(dev)) { + temp &= ~FDI_LINK_TRAIN_NONE_IVB; + temp |= FDI_LINK_TRAIN_NONE_IVB | FDI_TX_ENHANCE_FRAME_ENABLE; + } else { + temp &= ~FDI_LINK_TRAIN_NONE; + temp |= FDI_LINK_TRAIN_NONE | FDI_TX_ENHANCE_FRAME_ENABLE; + } + I915_WRITE(reg, temp); + + reg = FDI_RX_CTL(pipe); + temp = I915_READ(reg); + if (HAS_PCH_CPT(dev)) { + temp &= ~FDI_LINK_TRAIN_PATTERN_MASK_CPT; + temp |= FDI_LINK_TRAIN_NORMAL_CPT; + } else { + temp &= ~FDI_LINK_TRAIN_NONE; + temp |= FDI_LINK_TRAIN_NONE; + } + I915_WRITE(reg, temp | FDI_RX_ENHANCE_FRAME_ENABLE); + + /* wait one idle pattern time */ + POSTING_READ(reg); + udelay(1000); + + /* IVB wants error correction enabled */ + if (IS_IVYBRIDGE(dev)) + I915_WRITE(reg, I915_READ(reg) | FDI_FS_ERRC_ENABLE | + FDI_FE_ERRC_ENABLE); +} + +/* The FDI link training functions for ILK/Ibexpeak. */ +static void ironlake_fdi_link_train(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int pipe = intel_crtc->pipe; + int plane = intel_crtc->plane; + u32 reg, temp, tries; + + /* FDI needs bits from pipe & plane first */ + assert_pipe_enabled(dev_priv, pipe); + assert_plane_enabled(dev_priv, plane); + + /* Train 1: umask FDI RX Interrupt symbol_lock and bit_lock bit + for train result */ + reg = FDI_RX_IMR(pipe); + temp = I915_READ(reg); + temp &= ~FDI_RX_SYMBOL_LOCK; + temp &= ~FDI_RX_BIT_LOCK; + I915_WRITE(reg, temp); + I915_READ(reg); + udelay(150); + + /* enable CPU FDI TX and PCH FDI RX */ + reg = FDI_TX_CTL(pipe); + temp = I915_READ(reg); + temp &= ~(7 << 19); + temp |= (intel_crtc->fdi_lanes - 1) << 19; + temp &= ~FDI_LINK_TRAIN_NONE; + temp |= FDI_LINK_TRAIN_PATTERN_1; + I915_WRITE(reg, temp | FDI_TX_ENABLE); + + reg = FDI_RX_CTL(pipe); + temp = I915_READ(reg); + temp &= ~FDI_LINK_TRAIN_NONE; + temp |= FDI_LINK_TRAIN_PATTERN_1; + I915_WRITE(reg, temp | FDI_RX_ENABLE); + + POSTING_READ(reg); + udelay(150); + + /* Ironlake workaround, enable clock pointer after FDI enable*/ + if (HAS_PCH_IBX(dev)) { + I915_WRITE(FDI_RX_CHICKEN(pipe), FDI_RX_PHASE_SYNC_POINTER_OVR); + I915_WRITE(FDI_RX_CHICKEN(pipe), FDI_RX_PHASE_SYNC_POINTER_OVR | + FDI_RX_PHASE_SYNC_POINTER_EN); + } + + reg = FDI_RX_IIR(pipe); + for (tries = 0; tries < 5; tries++) { + temp = I915_READ(reg); + DRM_DEBUG_KMS("FDI_RX_IIR 0x%x\n", temp); + + if ((temp & FDI_RX_BIT_LOCK)) { + DRM_DEBUG_KMS("FDI train 1 done.\n"); + I915_WRITE(reg, temp | FDI_RX_BIT_LOCK); + break; + } + } + if (tries == 5) + DRM_ERROR("FDI train 1 fail!\n"); + + /* Train 2 */ + reg = FDI_TX_CTL(pipe); + temp = I915_READ(reg); + temp &= ~FDI_LINK_TRAIN_NONE; + temp |= FDI_LINK_TRAIN_PATTERN_2; + I915_WRITE(reg, temp); + + reg = FDI_RX_CTL(pipe); + temp = I915_READ(reg); + temp &= ~FDI_LINK_TRAIN_NONE; + temp |= FDI_LINK_TRAIN_PATTERN_2; + I915_WRITE(reg, temp); + + POSTING_READ(reg); + udelay(150); + + reg = FDI_RX_IIR(pipe); + for (tries = 0; tries < 5; tries++) { + temp = I915_READ(reg); + DRM_DEBUG_KMS("FDI_RX_IIR 0x%x\n", temp); + + if (temp & FDI_RX_SYMBOL_LOCK) { + I915_WRITE(reg, temp | FDI_RX_SYMBOL_LOCK); + DRM_DEBUG_KMS("FDI train 2 done.\n"); + break; + } + } + if (tries == 5) + DRM_ERROR("FDI train 2 fail!\n"); + + DRM_DEBUG_KMS("FDI train done\n"); + +} + +static const int snb_b_fdi_train_param [] = { + FDI_LINK_TRAIN_400MV_0DB_SNB_B, + FDI_LINK_TRAIN_400MV_6DB_SNB_B, + FDI_LINK_TRAIN_600MV_3_5DB_SNB_B, + FDI_LINK_TRAIN_800MV_0DB_SNB_B, +}; + +/* The FDI link training functions for SNB/Cougarpoint. */ +static void gen6_fdi_link_train(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int pipe = intel_crtc->pipe; + u32 reg, temp, i; + + /* Train 1: umask FDI RX Interrupt symbol_lock and bit_lock bit + for train result */ + reg = FDI_RX_IMR(pipe); + temp = I915_READ(reg); + temp &= ~FDI_RX_SYMBOL_LOCK; + temp &= ~FDI_RX_BIT_LOCK; + I915_WRITE(reg, temp); + + POSTING_READ(reg); + udelay(150); + + /* enable CPU FDI TX and PCH FDI RX */ + reg = FDI_TX_CTL(pipe); + temp = I915_READ(reg); + temp &= ~(7 << 19); + temp |= (intel_crtc->fdi_lanes - 1) << 19; + temp &= ~FDI_LINK_TRAIN_NONE; + temp |= FDI_LINK_TRAIN_PATTERN_1; + temp &= ~FDI_LINK_TRAIN_VOL_EMP_MASK; + /* SNB-B */ + temp |= FDI_LINK_TRAIN_400MV_0DB_SNB_B; + I915_WRITE(reg, temp | FDI_TX_ENABLE); + + reg = FDI_RX_CTL(pipe); + temp = I915_READ(reg); + if (HAS_PCH_CPT(dev)) { + temp &= ~FDI_LINK_TRAIN_PATTERN_MASK_CPT; + temp |= FDI_LINK_TRAIN_PATTERN_1_CPT; + } else { + temp &= ~FDI_LINK_TRAIN_NONE; + temp |= FDI_LINK_TRAIN_PATTERN_1; + } + I915_WRITE(reg, temp | FDI_RX_ENABLE); + + POSTING_READ(reg); + udelay(150); + + for (i = 0; i < 4; i++ ) { + reg = FDI_TX_CTL(pipe); + temp = I915_READ(reg); + temp &= ~FDI_LINK_TRAIN_VOL_EMP_MASK; + temp |= snb_b_fdi_train_param[i]; + I915_WRITE(reg, temp); + + POSTING_READ(reg); + udelay(500); + + reg = FDI_RX_IIR(pipe); + temp = I915_READ(reg); + DRM_DEBUG_KMS("FDI_RX_IIR 0x%x\n", temp); + + if (temp & FDI_RX_BIT_LOCK) { + I915_WRITE(reg, temp | FDI_RX_BIT_LOCK); + DRM_DEBUG_KMS("FDI train 1 done.\n"); + break; + } + } + if (i == 4) + DRM_ERROR("FDI train 1 fail!\n"); + + /* Train 2 */ + reg = FDI_TX_CTL(pipe); + temp = I915_READ(reg); + temp &= ~FDI_LINK_TRAIN_NONE; + temp |= FDI_LINK_TRAIN_PATTERN_2; + if (IS_GEN6(dev)) { + temp &= ~FDI_LINK_TRAIN_VOL_EMP_MASK; + /* SNB-B */ + temp |= FDI_LINK_TRAIN_400MV_0DB_SNB_B; + } + I915_WRITE(reg, temp); + + reg = FDI_RX_CTL(pipe); + temp = I915_READ(reg); + if (HAS_PCH_CPT(dev)) { + temp &= ~FDI_LINK_TRAIN_PATTERN_MASK_CPT; + temp |= FDI_LINK_TRAIN_PATTERN_2_CPT; + } else { + temp &= ~FDI_LINK_TRAIN_NONE; + temp |= FDI_LINK_TRAIN_PATTERN_2; + } + I915_WRITE(reg, temp); + + POSTING_READ(reg); + udelay(150); + + for (i = 0; i < 4; i++ ) { + reg = FDI_TX_CTL(pipe); + temp = I915_READ(reg); + temp &= ~FDI_LINK_TRAIN_VOL_EMP_MASK; + temp |= snb_b_fdi_train_param[i]; + I915_WRITE(reg, temp); + + POSTING_READ(reg); + udelay(500); + + reg = FDI_RX_IIR(pipe); + temp = I915_READ(reg); + DRM_DEBUG_KMS("FDI_RX_IIR 0x%x\n", temp); + + if (temp & FDI_RX_SYMBOL_LOCK) { + I915_WRITE(reg, temp | FDI_RX_SYMBOL_LOCK); + DRM_DEBUG_KMS("FDI train 2 done.\n"); + break; + } + } + if (i == 4) + DRM_ERROR("FDI train 2 fail!\n"); + + DRM_DEBUG_KMS("FDI train done.\n"); +} + +/* Manual link training for Ivy Bridge A0 parts */ +static void ivb_manual_fdi_link_train(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int pipe = intel_crtc->pipe; + u32 reg, temp, i; + + /* Train 1: umask FDI RX Interrupt symbol_lock and bit_lock bit + for train result */ + reg = FDI_RX_IMR(pipe); + temp = I915_READ(reg); + temp &= ~FDI_RX_SYMBOL_LOCK; + temp &= ~FDI_RX_BIT_LOCK; + I915_WRITE(reg, temp); + + POSTING_READ(reg); + udelay(150); + + /* enable CPU FDI TX and PCH FDI RX */ + reg = FDI_TX_CTL(pipe); + temp = I915_READ(reg); + temp &= ~(7 << 19); + temp |= (intel_crtc->fdi_lanes - 1) << 19; + temp &= ~(FDI_LINK_TRAIN_AUTO | FDI_LINK_TRAIN_NONE_IVB); + temp |= FDI_LINK_TRAIN_PATTERN_1_IVB; + temp &= ~FDI_LINK_TRAIN_VOL_EMP_MASK; + temp |= FDI_LINK_TRAIN_400MV_0DB_SNB_B; + temp |= FDI_COMPOSITE_SYNC; + I915_WRITE(reg, temp | FDI_TX_ENABLE); + + reg = FDI_RX_CTL(pipe); + temp = I915_READ(reg); + temp &= ~FDI_LINK_TRAIN_AUTO; + temp &= ~FDI_LINK_TRAIN_PATTERN_MASK_CPT; + temp |= FDI_LINK_TRAIN_PATTERN_1_CPT; + temp |= FDI_COMPOSITE_SYNC; + I915_WRITE(reg, temp | FDI_RX_ENABLE); + + POSTING_READ(reg); + udelay(150); + + for (i = 0; i < 4; i++ ) { + reg = FDI_TX_CTL(pipe); + temp = I915_READ(reg); + temp &= ~FDI_LINK_TRAIN_VOL_EMP_MASK; + temp |= snb_b_fdi_train_param[i]; + I915_WRITE(reg, temp); + + POSTING_READ(reg); + udelay(500); + + reg = FDI_RX_IIR(pipe); + temp = I915_READ(reg); + DRM_DEBUG_KMS("FDI_RX_IIR 0x%x\n", temp); + + if (temp & FDI_RX_BIT_LOCK || + (I915_READ(reg) & FDI_RX_BIT_LOCK)) { + I915_WRITE(reg, temp | FDI_RX_BIT_LOCK); + DRM_DEBUG_KMS("FDI train 1 done.\n"); + break; + } + } + if (i == 4) + DRM_ERROR("FDI train 1 fail!\n"); + + /* Train 2 */ + reg = FDI_TX_CTL(pipe); + temp = I915_READ(reg); + temp &= ~FDI_LINK_TRAIN_NONE_IVB; + temp |= FDI_LINK_TRAIN_PATTERN_2_IVB; + temp &= ~FDI_LINK_TRAIN_VOL_EMP_MASK; + temp |= FDI_LINK_TRAIN_400MV_0DB_SNB_B; + I915_WRITE(reg, temp); + + reg = FDI_RX_CTL(pipe); + temp = I915_READ(reg); + temp &= ~FDI_LINK_TRAIN_PATTERN_MASK_CPT; + temp |= FDI_LINK_TRAIN_PATTERN_2_CPT; + I915_WRITE(reg, temp); + + POSTING_READ(reg); + udelay(150); + + for (i = 0; i < 4; i++ ) { + reg = FDI_TX_CTL(pipe); + temp = I915_READ(reg); + temp &= ~FDI_LINK_TRAIN_VOL_EMP_MASK; + temp |= snb_b_fdi_train_param[i]; + I915_WRITE(reg, temp); + + POSTING_READ(reg); + udelay(500); + + reg = FDI_RX_IIR(pipe); + temp = I915_READ(reg); + DRM_DEBUG_KMS("FDI_RX_IIR 0x%x\n", temp); + + if (temp & FDI_RX_SYMBOL_LOCK) { + I915_WRITE(reg, temp | FDI_RX_SYMBOL_LOCK); + DRM_DEBUG_KMS("FDI train 2 done.\n"); + break; + } + } + if (i == 4) + DRM_ERROR("FDI train 2 fail!\n"); + + DRM_DEBUG_KMS("FDI train done.\n"); +} + +static void ironlake_fdi_pll_enable(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int pipe = intel_crtc->pipe; + u32 reg, temp; + + /* Write the TU size bits so error detection works */ + I915_WRITE(FDI_RX_TUSIZE1(pipe), + I915_READ(PIPE_DATA_M1(pipe)) & TU_SIZE_MASK); + + /* enable PCH FDI RX PLL, wait warmup plus DMI latency */ + reg = FDI_RX_CTL(pipe); + temp = I915_READ(reg); + temp &= ~((0x7 << 19) | (0x7 << 16)); + temp |= (intel_crtc->fdi_lanes - 1) << 19; + temp |= (I915_READ(PIPECONF(pipe)) & PIPE_BPC_MASK) << 11; + I915_WRITE(reg, temp | FDI_RX_PLL_ENABLE); + + POSTING_READ(reg); + udelay(200); + + /* Switch from Rawclk to PCDclk */ + temp = I915_READ(reg); + I915_WRITE(reg, temp | FDI_PCDCLK); + + POSTING_READ(reg); + udelay(200); + + /* Enable CPU FDI TX PLL, always on for Ironlake */ + reg = FDI_TX_CTL(pipe); + temp = I915_READ(reg); + if ((temp & FDI_TX_PLL_ENABLE) == 0) { + I915_WRITE(reg, temp | FDI_TX_PLL_ENABLE); + + POSTING_READ(reg); + udelay(100); + } +} + +static void ironlake_fdi_disable(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int pipe = intel_crtc->pipe; + u32 reg, temp; + + /* disable CPU FDI tx and PCH FDI rx */ + reg = FDI_TX_CTL(pipe); + temp = I915_READ(reg); + I915_WRITE(reg, temp & ~FDI_TX_ENABLE); + POSTING_READ(reg); + + reg = FDI_RX_CTL(pipe); + temp = I915_READ(reg); + temp &= ~(0x7 << 16); + temp |= (I915_READ(PIPECONF(pipe)) & PIPE_BPC_MASK) << 11; + I915_WRITE(reg, temp & ~FDI_RX_ENABLE); + + POSTING_READ(reg); + udelay(100); + + /* Ironlake workaround, disable clock pointer after downing FDI */ + if (HAS_PCH_IBX(dev)) { + I915_WRITE(FDI_RX_CHICKEN(pipe), FDI_RX_PHASE_SYNC_POINTER_OVR); + I915_WRITE(FDI_RX_CHICKEN(pipe), + I915_READ(FDI_RX_CHICKEN(pipe) & + ~FDI_RX_PHASE_SYNC_POINTER_EN)); + } + + /* still set train pattern 1 */ + reg = FDI_TX_CTL(pipe); + temp = I915_READ(reg); + temp &= ~FDI_LINK_TRAIN_NONE; + temp |= FDI_LINK_TRAIN_PATTERN_1; + I915_WRITE(reg, temp); + + reg = FDI_RX_CTL(pipe); + temp = I915_READ(reg); + if (HAS_PCH_CPT(dev)) { + temp &= ~FDI_LINK_TRAIN_PATTERN_MASK_CPT; + temp |= FDI_LINK_TRAIN_PATTERN_1_CPT; + } else { + temp &= ~FDI_LINK_TRAIN_NONE; + temp |= FDI_LINK_TRAIN_PATTERN_1; + } + /* BPC in FDI rx is consistent with that in PIPECONF */ + temp &= ~(0x07 << 16); + temp |= (I915_READ(PIPECONF(pipe)) & PIPE_BPC_MASK) << 11; + I915_WRITE(reg, temp); + + POSTING_READ(reg); + udelay(100); +} + +/* + * When we disable a pipe, we need to clear any pending scanline wait events + * to avoid hanging the ring, which we assume we are waiting on. + */ +static void intel_clear_scanline_wait(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_ring_buffer *ring; + u32 tmp; + + if (IS_GEN2(dev)) + /* Can't break the hang on i8xx */ + return; + + ring = LP_RING(dev_priv); + tmp = I915_READ_CTL(ring); + if (tmp & RING_WAIT) + I915_WRITE_CTL(ring, tmp); +} + +static void intel_crtc_wait_for_pending_flips(struct drm_crtc *crtc) +{ + struct drm_i915_gem_object *obj; + struct drm_i915_private *dev_priv; + + if (crtc->fb == NULL) + return; + + obj = to_intel_framebuffer(crtc->fb)->obj; + dev_priv = crtc->dev->dev_private; + wait_event(dev_priv->pending_flip_queue, + atomic_read(&obj->pending_flip) == 0); +} + +static bool intel_crtc_driving_pch(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + struct drm_mode_config *mode_config = &dev->mode_config; + struct intel_encoder *encoder; + + /* + * If there's a non-PCH eDP on this crtc, it must be DP_A, and that + * must be driven by its own crtc; no sharing is possible. + */ + list_for_each_entry(encoder, &mode_config->encoder_list, base.head) { + if (encoder->base.crtc != crtc) + continue; + + switch (encoder->type) { + case INTEL_OUTPUT_EDP: + if (!intel_encoder_is_pch_edp(&encoder->base)) + return false; + continue; + } + } + + return true; +} + +/* + * Enable PCH resources required for PCH ports: + * - PCH PLLs + * - FDI training & RX/TX + * - update transcoder timings + * - DP transcoding bits + * - transcoder + */ +static void ironlake_pch_enable(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int pipe = intel_crtc->pipe; + u32 reg, temp; + + /* For PCH output, training FDI link */ + dev_priv->display.fdi_link_train(crtc); + + intel_enable_pch_pll(dev_priv, pipe); + + if (HAS_PCH_CPT(dev)) { + /* Be sure PCH DPLL SEL is set */ + temp = I915_READ(PCH_DPLL_SEL); + if (pipe == 0 && (temp & TRANSA_DPLL_ENABLE) == 0) + temp |= (TRANSA_DPLL_ENABLE | TRANSA_DPLLA_SEL); + else if (pipe == 1 && (temp & TRANSB_DPLL_ENABLE) == 0) + temp |= (TRANSB_DPLL_ENABLE | TRANSB_DPLLB_SEL); + I915_WRITE(PCH_DPLL_SEL, temp); + } + + /* set transcoder timing, panel must allow it */ + assert_panel_unlocked(dev_priv, pipe); + I915_WRITE(TRANS_HTOTAL(pipe), I915_READ(HTOTAL(pipe))); + I915_WRITE(TRANS_HBLANK(pipe), I915_READ(HBLANK(pipe))); + I915_WRITE(TRANS_HSYNC(pipe), I915_READ(HSYNC(pipe))); + + I915_WRITE(TRANS_VTOTAL(pipe), I915_READ(VTOTAL(pipe))); + I915_WRITE(TRANS_VBLANK(pipe), I915_READ(VBLANK(pipe))); + I915_WRITE(TRANS_VSYNC(pipe), I915_READ(VSYNC(pipe))); + + intel_fdi_normal_train(crtc); + + /* For PCH DP, enable TRANS_DP_CTL */ + if (HAS_PCH_CPT(dev) && + intel_pipe_has_type(crtc, INTEL_OUTPUT_DISPLAYPORT)) { + reg = TRANS_DP_CTL(pipe); + temp = I915_READ(reg); + temp &= ~(TRANS_DP_PORT_SEL_MASK | + TRANS_DP_SYNC_MASK | + TRANS_DP_BPC_MASK); + temp |= (TRANS_DP_OUTPUT_ENABLE | + TRANS_DP_ENH_FRAMING); + temp |= TRANS_DP_8BPC; + + if (crtc->mode.flags & DRM_MODE_FLAG_PHSYNC) + temp |= TRANS_DP_HSYNC_ACTIVE_HIGH; + if (crtc->mode.flags & DRM_MODE_FLAG_PVSYNC) + temp |= TRANS_DP_VSYNC_ACTIVE_HIGH; + + switch (intel_trans_dp_port_sel(crtc)) { + case PCH_DP_B: + temp |= TRANS_DP_PORT_SEL_B; + break; + case PCH_DP_C: + temp |= TRANS_DP_PORT_SEL_C; + break; + case PCH_DP_D: + temp |= TRANS_DP_PORT_SEL_D; + break; + default: + DRM_DEBUG_KMS("Wrong PCH DP port return. Guess port B\n"); + temp |= TRANS_DP_PORT_SEL_B; + break; + } + + I915_WRITE(reg, temp); + } + + intel_enable_transcoder(dev_priv, pipe); +} + +static void ironlake_crtc_enable(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int pipe = intel_crtc->pipe; + int plane = intel_crtc->plane; + u32 temp; + bool is_pch_port; + + if (intel_crtc->active) + return; + + intel_crtc->active = true; + intel_update_watermarks(dev); + + if (intel_pipe_has_type(crtc, INTEL_OUTPUT_LVDS)) { + temp = I915_READ(PCH_LVDS); + if ((temp & LVDS_PORT_EN) == 0) + I915_WRITE(PCH_LVDS, temp | LVDS_PORT_EN); + } + + is_pch_port = intel_crtc_driving_pch(crtc); + + if (is_pch_port) + ironlake_fdi_pll_enable(crtc); + else + ironlake_fdi_disable(crtc); + + /* Enable panel fitting for LVDS */ + if (dev_priv->pch_pf_size && + (intel_pipe_has_type(crtc, INTEL_OUTPUT_LVDS) || HAS_eDP)) { + /* Force use of hard-coded filter coefficients + * as some pre-programmed values are broken, + * e.g. x201. + */ + I915_WRITE(PF_CTL(pipe), PF_ENABLE | PF_FILTER_MED_3x3); + I915_WRITE(PF_WIN_POS(pipe), dev_priv->pch_pf_pos); + I915_WRITE(PF_WIN_SZ(pipe), dev_priv->pch_pf_size); + } + + /* + * On ILK+ LUT must be loaded before the pipe is running but with + * clocks enabled + */ + intel_crtc_load_lut(crtc); + + intel_enable_pipe(dev_priv, pipe, is_pch_port); + intel_enable_plane(dev_priv, plane, pipe); + + if (is_pch_port) + ironlake_pch_enable(crtc); + + mutex_lock(&dev->struct_mutex); + intel_update_fbc(dev); + mutex_unlock(&dev->struct_mutex); + + intel_crtc_update_cursor(crtc, true); +} + +static void ironlake_crtc_disable(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int pipe = intel_crtc->pipe; + int plane = intel_crtc->plane; + u32 reg, temp; + + if (!intel_crtc->active) + return; + + intel_crtc_wait_for_pending_flips(crtc); + drm_vblank_off(dev, pipe); + intel_crtc_update_cursor(crtc, false); + + intel_disable_plane(dev_priv, plane, pipe); + + if (dev_priv->cfb_plane == plane && + dev_priv->display.disable_fbc) + dev_priv->display.disable_fbc(dev); + + intel_disable_pipe(dev_priv, pipe); + + /* Disable PF */ + I915_WRITE(PF_CTL(pipe), 0); + I915_WRITE(PF_WIN_SZ(pipe), 0); + + ironlake_fdi_disable(crtc); + + /* This is a horrible layering violation; we should be doing this in + * the connector/encoder ->prepare instead, but we don't always have + * enough information there about the config to know whether it will + * actually be necessary or just cause undesired flicker. + */ + intel_disable_pch_ports(dev_priv, pipe); + + intel_disable_transcoder(dev_priv, pipe); + + if (HAS_PCH_CPT(dev)) { + /* disable TRANS_DP_CTL */ + reg = TRANS_DP_CTL(pipe); + temp = I915_READ(reg); + temp &= ~(TRANS_DP_OUTPUT_ENABLE | TRANS_DP_PORT_SEL_MASK); + temp |= TRANS_DP_PORT_SEL_NONE; + I915_WRITE(reg, temp); + + /* disable DPLL_SEL */ + temp = I915_READ(PCH_DPLL_SEL); + switch (pipe) { + case 0: + temp &= ~(TRANSA_DPLL_ENABLE | TRANSA_DPLLA_SEL); + break; + case 1: + temp &= ~(TRANSB_DPLL_ENABLE | TRANSB_DPLLB_SEL); + break; + case 2: + /* FIXME: manage transcoder PLLs? */ + temp &= ~(TRANSC_DPLL_ENABLE | TRANSC_DPLLB_SEL); + break; + default: + BUG(); /* wtf */ + } + I915_WRITE(PCH_DPLL_SEL, temp); + } + + /* disable PCH DPLL */ + intel_disable_pch_pll(dev_priv, pipe); + + /* Switch from PCDclk to Rawclk */ + reg = FDI_RX_CTL(pipe); + temp = I915_READ(reg); + I915_WRITE(reg, temp & ~FDI_PCDCLK); + + /* Disable CPU FDI TX PLL */ + reg = FDI_TX_CTL(pipe); + temp = I915_READ(reg); + I915_WRITE(reg, temp & ~FDI_TX_PLL_ENABLE); + + POSTING_READ(reg); + udelay(100); + + reg = FDI_RX_CTL(pipe); + temp = I915_READ(reg); + I915_WRITE(reg, temp & ~FDI_RX_PLL_ENABLE); + + /* Wait for the clocks to turn off. */ + POSTING_READ(reg); + udelay(100); + + intel_crtc->active = false; + intel_update_watermarks(dev); + + mutex_lock(&dev->struct_mutex); + intel_update_fbc(dev); + intel_clear_scanline_wait(dev); + mutex_unlock(&dev->struct_mutex); +} + +static void ironlake_crtc_dpms(struct drm_crtc *crtc, int mode) +{ + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int pipe = intel_crtc->pipe; + int plane = intel_crtc->plane; + + /* XXX: When our outputs are all unaware of DPMS modes other than off + * and on, we should map those modes to DRM_MODE_DPMS_OFF in the CRTC. + */ + switch (mode) { + case DRM_MODE_DPMS_ON: + case DRM_MODE_DPMS_STANDBY: + case DRM_MODE_DPMS_SUSPEND: + DRM_DEBUG_KMS("crtc %d/%d dpms on\n", pipe, plane); + ironlake_crtc_enable(crtc); + break; + + case DRM_MODE_DPMS_OFF: + DRM_DEBUG_KMS("crtc %d/%d dpms off\n", pipe, plane); + ironlake_crtc_disable(crtc); + break; + } +} + +static void intel_crtc_dpms_overlay(struct intel_crtc *intel_crtc, bool enable) +{ + if (!enable && intel_crtc->overlay) { + struct drm_device *dev = intel_crtc->base.dev; + struct drm_i915_private *dev_priv = dev->dev_private; + + mutex_lock(&dev->struct_mutex); + dev_priv->mm.interruptible = false; + (void) intel_overlay_switch_off(intel_crtc->overlay); + dev_priv->mm.interruptible = true; + mutex_unlock(&dev->struct_mutex); + } + + /* Let userspace switch the overlay on again. In most cases userspace + * has to recompute where to put it anyway. + */ +} + +static void i9xx_crtc_enable(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int pipe = intel_crtc->pipe; + int plane = intel_crtc->plane; + + if (intel_crtc->active) + return; + + intel_crtc->active = true; + intel_update_watermarks(dev); + + intel_enable_pll(dev_priv, pipe); + intel_enable_pipe(dev_priv, pipe, false); + intel_enable_plane(dev_priv, plane, pipe); + + intel_crtc_load_lut(crtc); + intel_update_fbc(dev); + + /* Give the overlay scaler a chance to enable if it's on this pipe */ + intel_crtc_dpms_overlay(intel_crtc, true); + intel_crtc_update_cursor(crtc, true); +} + +static void i9xx_crtc_disable(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int pipe = intel_crtc->pipe; + int plane = intel_crtc->plane; + + if (!intel_crtc->active) + return; + + /* Give the overlay scaler a chance to disable if it's on this pipe */ + intel_crtc_wait_for_pending_flips(crtc); + drm_vblank_off(dev, pipe); + intel_crtc_dpms_overlay(intel_crtc, false); + intel_crtc_update_cursor(crtc, false); + + if (dev_priv->cfb_plane == plane && + dev_priv->display.disable_fbc) + dev_priv->display.disable_fbc(dev); + + intel_disable_plane(dev_priv, plane, pipe); + intel_disable_pipe(dev_priv, pipe); + intel_disable_pll(dev_priv, pipe); + + intel_crtc->active = false; + intel_update_fbc(dev); + intel_update_watermarks(dev); + intel_clear_scanline_wait(dev); +} + +static void i9xx_crtc_dpms(struct drm_crtc *crtc, int mode) +{ + /* XXX: When our outputs are all unaware of DPMS modes other than off + * and on, we should map those modes to DRM_MODE_DPMS_OFF in the CRTC. + */ + switch (mode) { + case DRM_MODE_DPMS_ON: + case DRM_MODE_DPMS_STANDBY: + case DRM_MODE_DPMS_SUSPEND: + i9xx_crtc_enable(crtc); + break; + case DRM_MODE_DPMS_OFF: + i9xx_crtc_disable(crtc); + break; + } +} + +/** + * Sets the power management mode of the pipe and plane. + */ +static void intel_crtc_dpms(struct drm_crtc *crtc, int mode) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_i915_master_private *master_priv; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int pipe = intel_crtc->pipe; + bool enabled; + + if (intel_crtc->dpms_mode == mode) + return; + + intel_crtc->dpms_mode = mode; + + dev_priv->display.dpms(crtc, mode); + + if (!dev->primary->master) + return; + + master_priv = dev->primary->master->driver_priv; + if (!master_priv->sarea_priv) + return; + + enabled = crtc->enabled && mode != DRM_MODE_DPMS_OFF; + + switch (pipe) { + case 0: + master_priv->sarea_priv->pipeA_w = enabled ? crtc->mode.hdisplay : 0; + master_priv->sarea_priv->pipeA_h = enabled ? crtc->mode.vdisplay : 0; + break; + case 1: + master_priv->sarea_priv->pipeB_w = enabled ? crtc->mode.hdisplay : 0; + master_priv->sarea_priv->pipeB_h = enabled ? crtc->mode.vdisplay : 0; + break; + default: + DRM_ERROR("Can't update pipe %c in SAREA\n", pipe_name(pipe)); + break; + } +} + +static void intel_crtc_disable(struct drm_crtc *crtc) +{ + struct drm_crtc_helper_funcs *crtc_funcs = crtc->helper_private; + struct drm_device *dev = crtc->dev; + + crtc_funcs->dpms(crtc, DRM_MODE_DPMS_OFF); + + if (crtc->fb) { + mutex_lock(&dev->struct_mutex); + i915_gem_object_unpin(to_intel_framebuffer(crtc->fb)->obj); + mutex_unlock(&dev->struct_mutex); + } +} + +/* Prepare for a mode set. + * + * Note we could be a lot smarter here. We need to figure out which outputs + * will be enabled, which disabled (in short, how the config will changes) + * and perform the minimum necessary steps to accomplish that, e.g. updating + * watermarks, FBC configuration, making sure PLLs are programmed correctly, + * panel fitting is in the proper state, etc. + */ +static void i9xx_crtc_prepare(struct drm_crtc *crtc) +{ + i9xx_crtc_disable(crtc); +} + +static void i9xx_crtc_commit(struct drm_crtc *crtc) +{ + i9xx_crtc_enable(crtc); +} + +static void ironlake_crtc_prepare(struct drm_crtc *crtc) +{ + ironlake_crtc_disable(crtc); +} + +static void ironlake_crtc_commit(struct drm_crtc *crtc) +{ + ironlake_crtc_enable(crtc); +} + +void intel_encoder_prepare (struct drm_encoder *encoder) +{ + struct drm_encoder_helper_funcs *encoder_funcs = encoder->helper_private; + /* lvds has its own version of prepare see intel_lvds_prepare */ + encoder_funcs->dpms(encoder, DRM_MODE_DPMS_OFF); +} + +void intel_encoder_commit (struct drm_encoder *encoder) +{ + struct drm_encoder_helper_funcs *encoder_funcs = encoder->helper_private; + /* lvds has its own version of commit see intel_lvds_commit */ + encoder_funcs->dpms(encoder, DRM_MODE_DPMS_ON); +} + +void intel_encoder_destroy(struct drm_encoder *encoder) +{ + struct intel_encoder *intel_encoder = to_intel_encoder(encoder); + + drm_encoder_cleanup(encoder); + kfree(intel_encoder); +} + +static bool intel_crtc_mode_fixup(struct drm_crtc *crtc, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct drm_device *dev = crtc->dev; + + if (HAS_PCH_SPLIT(dev)) { + /* FDI link clock is fixed at 2.7G */ + if (mode->clock * 3 > IRONLAKE_FDI_FREQ * 4) + return false; + } + + /* XXX some encoders set the crtcinfo, others don't. + * Obviously we need some form of conflict resolution here... + */ + if (adjusted_mode->crtc_htotal == 0) + drm_mode_set_crtcinfo(adjusted_mode, 0); + + return true; +} + +static int i945_get_display_clock_speed(struct drm_device *dev) +{ + return 400000; +} + +static int i915_get_display_clock_speed(struct drm_device *dev) +{ + return 333000; +} + +static int i9xx_misc_get_display_clock_speed(struct drm_device *dev) +{ + return 200000; +} + +static int i915gm_get_display_clock_speed(struct drm_device *dev) +{ + u16 gcfgc = 0; + + pci_read_config_word(dev->pdev, GCFGC, &gcfgc); + + if (gcfgc & GC_LOW_FREQUENCY_ENABLE) + return 133000; + else { + switch (gcfgc & GC_DISPLAY_CLOCK_MASK) { + case GC_DISPLAY_CLOCK_333_MHZ: + return 333000; + default: + case GC_DISPLAY_CLOCK_190_200_MHZ: + return 190000; + } + } +} + +static int i865_get_display_clock_speed(struct drm_device *dev) +{ + return 266000; +} + +static int i855_get_display_clock_speed(struct drm_device *dev) +{ + u16 hpllcc = 0; + /* Assume that the hardware is in the high speed state. This + * should be the default. + */ + switch (hpllcc & GC_CLOCK_CONTROL_MASK) { + case GC_CLOCK_133_200: + case GC_CLOCK_100_200: + return 200000; + case GC_CLOCK_166_250: + return 250000; + case GC_CLOCK_100_133: + return 133000; + } + + /* Shouldn't happen */ + return 0; +} + +static int i830_get_display_clock_speed(struct drm_device *dev) +{ + return 133000; +} + +struct fdi_m_n { + u32 tu; + u32 gmch_m; + u32 gmch_n; + u32 link_m; + u32 link_n; +}; + +static void +fdi_reduce_ratio(u32 *num, u32 *den) +{ + while (*num > 0xffffff || *den > 0xffffff) { + *num >>= 1; + *den >>= 1; + } +} + +static void +ironlake_compute_m_n(int bits_per_pixel, int nlanes, int pixel_clock, + int link_clock, struct fdi_m_n *m_n) +{ + m_n->tu = 64; /* default size */ + + /* BUG_ON(pixel_clock > INT_MAX / 36); */ + m_n->gmch_m = bits_per_pixel * pixel_clock; + m_n->gmch_n = link_clock * nlanes * 8; + fdi_reduce_ratio(&m_n->gmch_m, &m_n->gmch_n); + + m_n->link_m = pixel_clock; + m_n->link_n = link_clock; + fdi_reduce_ratio(&m_n->link_m, &m_n->link_n); +} + + +struct intel_watermark_params { + unsigned long fifo_size; + unsigned long max_wm; + unsigned long default_wm; + unsigned long guard_size; + unsigned long cacheline_size; +}; + +/* Pineview has different values for various configs */ +static const struct intel_watermark_params pineview_display_wm = { + PINEVIEW_DISPLAY_FIFO, + PINEVIEW_MAX_WM, + PINEVIEW_DFT_WM, + PINEVIEW_GUARD_WM, + PINEVIEW_FIFO_LINE_SIZE +}; +static const struct intel_watermark_params pineview_display_hplloff_wm = { + PINEVIEW_DISPLAY_FIFO, + PINEVIEW_MAX_WM, + PINEVIEW_DFT_HPLLOFF_WM, + PINEVIEW_GUARD_WM, + PINEVIEW_FIFO_LINE_SIZE +}; +static const struct intel_watermark_params pineview_cursor_wm = { + PINEVIEW_CURSOR_FIFO, + PINEVIEW_CURSOR_MAX_WM, + PINEVIEW_CURSOR_DFT_WM, + PINEVIEW_CURSOR_GUARD_WM, + PINEVIEW_FIFO_LINE_SIZE, +}; +static const struct intel_watermark_params pineview_cursor_hplloff_wm = { + PINEVIEW_CURSOR_FIFO, + PINEVIEW_CURSOR_MAX_WM, + PINEVIEW_CURSOR_DFT_WM, + PINEVIEW_CURSOR_GUARD_WM, + PINEVIEW_FIFO_LINE_SIZE +}; +static const struct intel_watermark_params g4x_wm_info = { + G4X_FIFO_SIZE, + G4X_MAX_WM, + G4X_MAX_WM, + 2, + G4X_FIFO_LINE_SIZE, +}; +static const struct intel_watermark_params g4x_cursor_wm_info = { + I965_CURSOR_FIFO, + I965_CURSOR_MAX_WM, + I965_CURSOR_DFT_WM, + 2, + G4X_FIFO_LINE_SIZE, +}; +static const struct intel_watermark_params i965_cursor_wm_info = { + I965_CURSOR_FIFO, + I965_CURSOR_MAX_WM, + I965_CURSOR_DFT_WM, + 2, + I915_FIFO_LINE_SIZE, +}; +static const struct intel_watermark_params i945_wm_info = { + I945_FIFO_SIZE, + I915_MAX_WM, + 1, + 2, + I915_FIFO_LINE_SIZE +}; +static const struct intel_watermark_params i915_wm_info = { + I915_FIFO_SIZE, + I915_MAX_WM, + 1, + 2, + I915_FIFO_LINE_SIZE +}; +static const struct intel_watermark_params i855_wm_info = { + I855GM_FIFO_SIZE, + I915_MAX_WM, + 1, + 2, + I830_FIFO_LINE_SIZE +}; +static const struct intel_watermark_params i830_wm_info = { + I830_FIFO_SIZE, + I915_MAX_WM, + 1, + 2, + I830_FIFO_LINE_SIZE +}; + +static const struct intel_watermark_params ironlake_display_wm_info = { + ILK_DISPLAY_FIFO, + ILK_DISPLAY_MAXWM, + ILK_DISPLAY_DFTWM, + 2, + ILK_FIFO_LINE_SIZE +}; +static const struct intel_watermark_params ironlake_cursor_wm_info = { + ILK_CURSOR_FIFO, + ILK_CURSOR_MAXWM, + ILK_CURSOR_DFTWM, + 2, + ILK_FIFO_LINE_SIZE +}; +static const struct intel_watermark_params ironlake_display_srwm_info = { + ILK_DISPLAY_SR_FIFO, + ILK_DISPLAY_MAX_SRWM, + ILK_DISPLAY_DFT_SRWM, + 2, + ILK_FIFO_LINE_SIZE +}; +static const struct intel_watermark_params ironlake_cursor_srwm_info = { + ILK_CURSOR_SR_FIFO, + ILK_CURSOR_MAX_SRWM, + ILK_CURSOR_DFT_SRWM, + 2, + ILK_FIFO_LINE_SIZE +}; + +static const struct intel_watermark_params sandybridge_display_wm_info = { + SNB_DISPLAY_FIFO, + SNB_DISPLAY_MAXWM, + SNB_DISPLAY_DFTWM, + 2, + SNB_FIFO_LINE_SIZE +}; +static const struct intel_watermark_params sandybridge_cursor_wm_info = { + SNB_CURSOR_FIFO, + SNB_CURSOR_MAXWM, + SNB_CURSOR_DFTWM, + 2, + SNB_FIFO_LINE_SIZE +}; +static const struct intel_watermark_params sandybridge_display_srwm_info = { + SNB_DISPLAY_SR_FIFO, + SNB_DISPLAY_MAX_SRWM, + SNB_DISPLAY_DFT_SRWM, + 2, + SNB_FIFO_LINE_SIZE +}; +static const struct intel_watermark_params sandybridge_cursor_srwm_info = { + SNB_CURSOR_SR_FIFO, + SNB_CURSOR_MAX_SRWM, + SNB_CURSOR_DFT_SRWM, + 2, + SNB_FIFO_LINE_SIZE +}; + + +/** + * intel_calculate_wm - calculate watermark level + * @clock_in_khz: pixel clock + * @wm: chip FIFO params + * @pixel_size: display pixel size + * @latency_ns: memory latency for the platform + * + * Calculate the watermark level (the level at which the display plane will + * start fetching from memory again). Each chip has a different display + * FIFO size and allocation, so the caller needs to figure that out and pass + * in the correct intel_watermark_params structure. + * + * As the pixel clock runs, the FIFO will be drained at a rate that depends + * on the pixel size. When it reaches the watermark level, it'll start + * fetching FIFO line sized based chunks from memory until the FIFO fills + * past the watermark point. If the FIFO drains completely, a FIFO underrun + * will occur, and a display engine hang could result. + */ +static unsigned long intel_calculate_wm(unsigned long clock_in_khz, + const struct intel_watermark_params *wm, + int fifo_size, + int pixel_size, + unsigned long latency_ns) +{ + long entries_required, wm_size; + + /* + * Note: we need to make sure we don't overflow for various clock & + * latency values. + * clocks go from a few thousand to several hundred thousand. + * latency is usually a few thousand + */ + entries_required = ((clock_in_khz / 1000) * pixel_size * latency_ns) / + 1000; + entries_required = DIV_ROUND_UP(entries_required, wm->cacheline_size); + + DRM_DEBUG_KMS("FIFO entries required for mode: %ld\n", entries_required); + + wm_size = fifo_size - (entries_required + wm->guard_size); + + DRM_DEBUG_KMS("FIFO watermark level: %ld\n", wm_size); + + /* Don't promote wm_size to unsigned... */ + if (wm_size > (long)wm->max_wm) + wm_size = wm->max_wm; + if (wm_size <= 0) + wm_size = wm->default_wm; + return wm_size; +} + +struct cxsr_latency { + int is_desktop; + int is_ddr3; + unsigned long fsb_freq; + unsigned long mem_freq; + unsigned long display_sr; + unsigned long display_hpll_disable; + unsigned long cursor_sr; + unsigned long cursor_hpll_disable; +}; + +static const struct cxsr_latency cxsr_latency_table[] = { + {1, 0, 800, 400, 3382, 33382, 3983, 33983}, /* DDR2-400 SC */ + {1, 0, 800, 667, 3354, 33354, 3807, 33807}, /* DDR2-667 SC */ + {1, 0, 800, 800, 3347, 33347, 3763, 33763}, /* DDR2-800 SC */ + {1, 1, 800, 667, 6420, 36420, 6873, 36873}, /* DDR3-667 SC */ + {1, 1, 800, 800, 5902, 35902, 6318, 36318}, /* DDR3-800 SC */ + + {1, 0, 667, 400, 3400, 33400, 4021, 34021}, /* DDR2-400 SC */ + {1, 0, 667, 667, 3372, 33372, 3845, 33845}, /* DDR2-667 SC */ + {1, 0, 667, 800, 3386, 33386, 3822, 33822}, /* DDR2-800 SC */ + {1, 1, 667, 667, 6438, 36438, 6911, 36911}, /* DDR3-667 SC */ + {1, 1, 667, 800, 5941, 35941, 6377, 36377}, /* DDR3-800 SC */ + + {1, 0, 400, 400, 3472, 33472, 4173, 34173}, /* DDR2-400 SC */ + {1, 0, 400, 667, 3443, 33443, 3996, 33996}, /* DDR2-667 SC */ + {1, 0, 400, 800, 3430, 33430, 3946, 33946}, /* DDR2-800 SC */ + {1, 1, 400, 667, 6509, 36509, 7062, 37062}, /* DDR3-667 SC */ + {1, 1, 400, 800, 5985, 35985, 6501, 36501}, /* DDR3-800 SC */ + + {0, 0, 800, 400, 3438, 33438, 4065, 34065}, /* DDR2-400 SC */ + {0, 0, 800, 667, 3410, 33410, 3889, 33889}, /* DDR2-667 SC */ + {0, 0, 800, 800, 3403, 33403, 3845, 33845}, /* DDR2-800 SC */ + {0, 1, 800, 667, 6476, 36476, 6955, 36955}, /* DDR3-667 SC */ + {0, 1, 800, 800, 5958, 35958, 6400, 36400}, /* DDR3-800 SC */ + + {0, 0, 667, 400, 3456, 33456, 4103, 34106}, /* DDR2-400 SC */ + {0, 0, 667, 667, 3428, 33428, 3927, 33927}, /* DDR2-667 SC */ + {0, 0, 667, 800, 3443, 33443, 3905, 33905}, /* DDR2-800 SC */ + {0, 1, 667, 667, 6494, 36494, 6993, 36993}, /* DDR3-667 SC */ + {0, 1, 667, 800, 5998, 35998, 6460, 36460}, /* DDR3-800 SC */ + + {0, 0, 400, 400, 3528, 33528, 4255, 34255}, /* DDR2-400 SC */ + {0, 0, 400, 667, 3500, 33500, 4079, 34079}, /* DDR2-667 SC */ + {0, 0, 400, 800, 3487, 33487, 4029, 34029}, /* DDR2-800 SC */ + {0, 1, 400, 667, 6566, 36566, 7145, 37145}, /* DDR3-667 SC */ + {0, 1, 400, 800, 6042, 36042, 6584, 36584}, /* DDR3-800 SC */ +}; + +static const struct cxsr_latency *intel_get_cxsr_latency(int is_desktop, + int is_ddr3, + int fsb, + int mem) +{ + const struct cxsr_latency *latency; + int i; + + if (fsb == 0 || mem == 0) + return NULL; + + for (i = 0; i < ARRAY_SIZE(cxsr_latency_table); i++) { + latency = &cxsr_latency_table[i]; + if (is_desktop == latency->is_desktop && + is_ddr3 == latency->is_ddr3 && + fsb == latency->fsb_freq && mem == latency->mem_freq) + return latency; + } + + DRM_DEBUG_KMS("Unknown FSB/MEM found, disable CxSR\n"); + + return NULL; +} + +static void pineview_disable_cxsr(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + /* deactivate cxsr */ + I915_WRITE(DSPFW3, I915_READ(DSPFW3) & ~PINEVIEW_SELF_REFRESH_EN); +} + +/* + * Latency for FIFO fetches is dependent on several factors: + * - memory configuration (speed, channels) + * - chipset + * - current MCH state + * It can be fairly high in some situations, so here we assume a fairly + * pessimal value. It's a tradeoff between extra memory fetches (if we + * set this value too high, the FIFO will fetch frequently to stay full) + * and power consumption (set it too low to save power and we might see + * FIFO underruns and display "flicker"). + * + * A value of 5us seems to be a good balance; safe for very low end + * platforms but not overly aggressive on lower latency configs. + */ +static const int latency_ns = 5000; + +static int i9xx_get_fifo_size(struct drm_device *dev, int plane) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + uint32_t dsparb = I915_READ(DSPARB); + int size; + + size = dsparb & 0x7f; + if (plane) + size = ((dsparb >> DSPARB_CSTART_SHIFT) & 0x7f) - size; + + DRM_DEBUG_KMS("FIFO size - (0x%08x) %s: %d\n", dsparb, + plane ? "B" : "A", size); + + return size; +} + +static int i85x_get_fifo_size(struct drm_device *dev, int plane) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + uint32_t dsparb = I915_READ(DSPARB); + int size; + + size = dsparb & 0x1ff; + if (plane) + size = ((dsparb >> DSPARB_BEND_SHIFT) & 0x1ff) - size; + size >>= 1; /* Convert to cachelines */ + + DRM_DEBUG_KMS("FIFO size - (0x%08x) %s: %d\n", dsparb, + plane ? "B" : "A", size); + + return size; +} + +static int i845_get_fifo_size(struct drm_device *dev, int plane) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + uint32_t dsparb = I915_READ(DSPARB); + int size; + + size = dsparb & 0x7f; + size >>= 2; /* Convert to cachelines */ + + DRM_DEBUG_KMS("FIFO size - (0x%08x) %s: %d\n", dsparb, + plane ? "B" : "A", + size); + + return size; +} + +static int i830_get_fifo_size(struct drm_device *dev, int plane) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + uint32_t dsparb = I915_READ(DSPARB); + int size; + + size = dsparb & 0x7f; + size >>= 1; /* Convert to cachelines */ + + DRM_DEBUG_KMS("FIFO size - (0x%08x) %s: %d\n", dsparb, + plane ? "B" : "A", size); + + return size; +} + +static struct drm_crtc *single_enabled_crtc(struct drm_device *dev) +{ + struct drm_crtc *crtc, *enabled = NULL; + + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { + if (crtc->enabled && crtc->fb) { + if (enabled) + return NULL; + enabled = crtc; + } + } + + return enabled; +} + +static void pineview_update_wm(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_crtc *crtc; + const struct cxsr_latency *latency; + u32 reg; + unsigned long wm; + + latency = intel_get_cxsr_latency(IS_PINEVIEW_G(dev), dev_priv->is_ddr3, + dev_priv->fsb_freq, dev_priv->mem_freq); + if (!latency) { + DRM_DEBUG_KMS("Unknown FSB/MEM found, disable CxSR\n"); + pineview_disable_cxsr(dev); + return; + } + + crtc = single_enabled_crtc(dev); + if (crtc) { + int clock = crtc->mode.clock; + int pixel_size = crtc->fb->bits_per_pixel / 8; + + /* Display SR */ + wm = intel_calculate_wm(clock, &pineview_display_wm, + pineview_display_wm.fifo_size, + pixel_size, latency->display_sr); + reg = I915_READ(DSPFW1); + reg &= ~DSPFW_SR_MASK; + reg |= wm << DSPFW_SR_SHIFT; + I915_WRITE(DSPFW1, reg); + DRM_DEBUG_KMS("DSPFW1 register is %x\n", reg); + + /* cursor SR */ + wm = intel_calculate_wm(clock, &pineview_cursor_wm, + pineview_display_wm.fifo_size, + pixel_size, latency->cursor_sr); + reg = I915_READ(DSPFW3); + reg &= ~DSPFW_CURSOR_SR_MASK; + reg |= (wm & 0x3f) << DSPFW_CURSOR_SR_SHIFT; + I915_WRITE(DSPFW3, reg); + + /* Display HPLL off SR */ + wm = intel_calculate_wm(clock, &pineview_display_hplloff_wm, + pineview_display_hplloff_wm.fifo_size, + pixel_size, latency->display_hpll_disable); + reg = I915_READ(DSPFW3); + reg &= ~DSPFW_HPLL_SR_MASK; + reg |= wm & DSPFW_HPLL_SR_MASK; + I915_WRITE(DSPFW3, reg); + + /* cursor HPLL off SR */ + wm = intel_calculate_wm(clock, &pineview_cursor_hplloff_wm, + pineview_display_hplloff_wm.fifo_size, + pixel_size, latency->cursor_hpll_disable); + reg = I915_READ(DSPFW3); + reg &= ~DSPFW_HPLL_CURSOR_MASK; + reg |= (wm & 0x3f) << DSPFW_HPLL_CURSOR_SHIFT; + I915_WRITE(DSPFW3, reg); + DRM_DEBUG_KMS("DSPFW3 register is %x\n", reg); + + /* activate cxsr */ + I915_WRITE(DSPFW3, + I915_READ(DSPFW3) | PINEVIEW_SELF_REFRESH_EN); + DRM_DEBUG_KMS("Self-refresh is enabled\n"); + } else { + pineview_disable_cxsr(dev); + DRM_DEBUG_KMS("Self-refresh is disabled\n"); + } +} + +static bool g4x_compute_wm0(struct drm_device *dev, + int plane, + const struct intel_watermark_params *display, + int display_latency_ns, + const struct intel_watermark_params *cursor, + int cursor_latency_ns, + int *plane_wm, + int *cursor_wm) +{ + struct drm_crtc *crtc; + int htotal, hdisplay, clock, pixel_size; + int line_time_us, line_count; + int entries, tlb_miss; + + crtc = intel_get_crtc_for_plane(dev, plane); + if (crtc->fb == NULL || !crtc->enabled) { + *cursor_wm = cursor->guard_size; + *plane_wm = display->guard_size; + return false; + } + + htotal = crtc->mode.htotal; + hdisplay = crtc->mode.hdisplay; + clock = crtc->mode.clock; + pixel_size = crtc->fb->bits_per_pixel / 8; + + /* Use the small buffer method to calculate plane watermark */ + entries = ((clock * pixel_size / 1000) * display_latency_ns) / 1000; + tlb_miss = display->fifo_size*display->cacheline_size - hdisplay * 8; + if (tlb_miss > 0) + entries += tlb_miss; + entries = DIV_ROUND_UP(entries, display->cacheline_size); + *plane_wm = entries + display->guard_size; + if (*plane_wm > (int)display->max_wm) + *plane_wm = display->max_wm; + + /* Use the large buffer method to calculate cursor watermark */ + line_time_us = ((htotal * 1000) / clock); + line_count = (cursor_latency_ns / line_time_us + 1000) / 1000; + entries = line_count * 64 * pixel_size; + tlb_miss = cursor->fifo_size*cursor->cacheline_size - hdisplay * 8; + if (tlb_miss > 0) + entries += tlb_miss; + entries = DIV_ROUND_UP(entries, cursor->cacheline_size); + *cursor_wm = entries + cursor->guard_size; + if (*cursor_wm > (int)cursor->max_wm) + *cursor_wm = (int)cursor->max_wm; + + return true; +} + +/* + * Check the wm result. + * + * If any calculated watermark values is larger than the maximum value that + * can be programmed into the associated watermark register, that watermark + * must be disabled. + */ +static bool g4x_check_srwm(struct drm_device *dev, + int display_wm, int cursor_wm, + const struct intel_watermark_params *display, + const struct intel_watermark_params *cursor) +{ + DRM_DEBUG_KMS("SR watermark: display plane %d, cursor %d\n", + display_wm, cursor_wm); + + if (display_wm > display->max_wm) { + DRM_DEBUG_KMS("display watermark is too large(%d/%ld), disabling\n", + display_wm, display->max_wm); + return false; + } + + if (cursor_wm > cursor->max_wm) { + DRM_DEBUG_KMS("cursor watermark is too large(%d/%ld), disabling\n", + cursor_wm, cursor->max_wm); + return false; + } + + if (!(display_wm || cursor_wm)) { + DRM_DEBUG_KMS("SR latency is 0, disabling\n"); + return false; + } + + return true; +} + +static bool g4x_compute_srwm(struct drm_device *dev, + int plane, + int latency_ns, + const struct intel_watermark_params *display, + const struct intel_watermark_params *cursor, + int *display_wm, int *cursor_wm) +{ + struct drm_crtc *crtc; + int hdisplay, htotal, pixel_size, clock; + unsigned long line_time_us; + int line_count, line_size; + int small, large; + int entries; + + if (!latency_ns) { + *display_wm = *cursor_wm = 0; + return false; + } + + crtc = intel_get_crtc_for_plane(dev, plane); + hdisplay = crtc->mode.hdisplay; + htotal = crtc->mode.htotal; + clock = crtc->mode.clock; + pixel_size = crtc->fb->bits_per_pixel / 8; + + line_time_us = (htotal * 1000) / clock; + line_count = (latency_ns / line_time_us + 1000) / 1000; + line_size = hdisplay * pixel_size; + + /* Use the minimum of the small and large buffer method for primary */ + small = ((clock * pixel_size / 1000) * latency_ns) / 1000; + large = line_count * line_size; + + entries = DIV_ROUND_UP(min(small, large), display->cacheline_size); + *display_wm = entries + display->guard_size; + + /* calculate the self-refresh watermark for display cursor */ + entries = line_count * pixel_size * 64; + entries = DIV_ROUND_UP(entries, cursor->cacheline_size); + *cursor_wm = entries + cursor->guard_size; + + return g4x_check_srwm(dev, + *display_wm, *cursor_wm, + display, cursor); +} + +#define single_plane_enabled(mask) is_power_of_2(mask) + +static void g4x_update_wm(struct drm_device *dev) +{ + static const int sr_latency_ns = 12000; + struct drm_i915_private *dev_priv = dev->dev_private; + int planea_wm, planeb_wm, cursora_wm, cursorb_wm; + int plane_sr, cursor_sr; + unsigned int enabled = 0; + + if (g4x_compute_wm0(dev, 0, + &g4x_wm_info, latency_ns, + &g4x_cursor_wm_info, latency_ns, + &planea_wm, &cursora_wm)) + enabled |= 1; + + if (g4x_compute_wm0(dev, 1, + &g4x_wm_info, latency_ns, + &g4x_cursor_wm_info, latency_ns, + &planeb_wm, &cursorb_wm)) + enabled |= 2; + + plane_sr = cursor_sr = 0; + if (single_plane_enabled(enabled) && + g4x_compute_srwm(dev, ffs(enabled) - 1, + sr_latency_ns, + &g4x_wm_info, + &g4x_cursor_wm_info, + &plane_sr, &cursor_sr)) + I915_WRITE(FW_BLC_SELF, FW_BLC_SELF_EN); + else + I915_WRITE(FW_BLC_SELF, + I915_READ(FW_BLC_SELF) & ~FW_BLC_SELF_EN); + + DRM_DEBUG_KMS("Setting FIFO watermarks - A: plane=%d, cursor=%d, B: plane=%d, cursor=%d, SR: plane=%d, cursor=%d\n", + planea_wm, cursora_wm, + planeb_wm, cursorb_wm, + plane_sr, cursor_sr); + + I915_WRITE(DSPFW1, + (plane_sr << DSPFW_SR_SHIFT) | + (cursorb_wm << DSPFW_CURSORB_SHIFT) | + (planeb_wm << DSPFW_PLANEB_SHIFT) | + planea_wm); + I915_WRITE(DSPFW2, + (I915_READ(DSPFW2) & DSPFW_CURSORA_MASK) | + (cursora_wm << DSPFW_CURSORA_SHIFT)); + /* HPLL off in SR has some issues on G4x... disable it */ + I915_WRITE(DSPFW3, + (I915_READ(DSPFW3) & ~DSPFW_HPLL_SR_EN) | + (cursor_sr << DSPFW_CURSOR_SR_SHIFT)); +} + +static void i965_update_wm(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_crtc *crtc; + int srwm = 1; + int cursor_sr = 16; + + /* Calc sr entries for one plane configs */ + crtc = single_enabled_crtc(dev); + if (crtc) { + /* self-refresh has much higher latency */ + static const int sr_latency_ns = 12000; + int clock = crtc->mode.clock; + int htotal = crtc->mode.htotal; + int hdisplay = crtc->mode.hdisplay; + int pixel_size = crtc->fb->bits_per_pixel / 8; + unsigned long line_time_us; + int entries; + + line_time_us = ((htotal * 1000) / clock); + + /* Use ns/us then divide to preserve precision */ + entries = (((sr_latency_ns / line_time_us) + 1000) / 1000) * + pixel_size * hdisplay; + entries = DIV_ROUND_UP(entries, I915_FIFO_LINE_SIZE); + srwm = I965_FIFO_SIZE - entries; + if (srwm < 0) + srwm = 1; + srwm &= 0x1ff; + DRM_DEBUG_KMS("self-refresh entries: %d, wm: %d\n", + entries, srwm); + + entries = (((sr_latency_ns / line_time_us) + 1000) / 1000) * + pixel_size * 64; + entries = DIV_ROUND_UP(entries, + i965_cursor_wm_info.cacheline_size); + cursor_sr = i965_cursor_wm_info.fifo_size - + (entries + i965_cursor_wm_info.guard_size); + + if (cursor_sr > i965_cursor_wm_info.max_wm) + cursor_sr = i965_cursor_wm_info.max_wm; + + DRM_DEBUG_KMS("self-refresh watermark: display plane %d " + "cursor %d\n", srwm, cursor_sr); + + if (IS_CRESTLINE(dev)) + I915_WRITE(FW_BLC_SELF, FW_BLC_SELF_EN); + } else { + /* Turn off self refresh if both pipes are enabled */ + if (IS_CRESTLINE(dev)) + I915_WRITE(FW_BLC_SELF, I915_READ(FW_BLC_SELF) + & ~FW_BLC_SELF_EN); + } + + DRM_DEBUG_KMS("Setting FIFO watermarks - A: 8, B: 8, C: 8, SR %d\n", + srwm); + + /* 965 has limitations... */ + I915_WRITE(DSPFW1, (srwm << DSPFW_SR_SHIFT) | + (8 << 16) | (8 << 8) | (8 << 0)); + I915_WRITE(DSPFW2, (8 << 8) | (8 << 0)); + /* update cursor SR watermark */ + I915_WRITE(DSPFW3, (cursor_sr << DSPFW_CURSOR_SR_SHIFT)); +} + +static void i9xx_update_wm(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + const struct intel_watermark_params *wm_info; + uint32_t fwater_lo; + uint32_t fwater_hi; + int cwm, srwm = 1; + int fifo_size; + int planea_wm, planeb_wm; + struct drm_crtc *crtc, *enabled = NULL; + + if (IS_I945GM(dev)) + wm_info = &i945_wm_info; + else if (!IS_GEN2(dev)) + wm_info = &i915_wm_info; + else + wm_info = &i855_wm_info; + + fifo_size = dev_priv->display.get_fifo_size(dev, 0); + crtc = intel_get_crtc_for_plane(dev, 0); + if (crtc->enabled && crtc->fb) { + planea_wm = intel_calculate_wm(crtc->mode.clock, + wm_info, fifo_size, + crtc->fb->bits_per_pixel / 8, + latency_ns); + enabled = crtc; + } else + planea_wm = fifo_size - wm_info->guard_size; + + fifo_size = dev_priv->display.get_fifo_size(dev, 1); + crtc = intel_get_crtc_for_plane(dev, 1); + if (crtc->enabled && crtc->fb) { + planeb_wm = intel_calculate_wm(crtc->mode.clock, + wm_info, fifo_size, + crtc->fb->bits_per_pixel / 8, + latency_ns); + if (enabled == NULL) + enabled = crtc; + else + enabled = NULL; + } else + planeb_wm = fifo_size - wm_info->guard_size; + + DRM_DEBUG_KMS("FIFO watermarks - A: %d, B: %d\n", planea_wm, planeb_wm); + + /* + * Overlay gets an aggressive default since video jitter is bad. + */ + cwm = 2; + + /* Play safe and disable self-refresh before adjusting watermarks. */ + if (IS_I945G(dev) || IS_I945GM(dev)) + I915_WRITE(FW_BLC_SELF, FW_BLC_SELF_EN_MASK | 0); + else if (IS_I915GM(dev)) + I915_WRITE(INSTPM, I915_READ(INSTPM) & ~INSTPM_SELF_EN); + + /* Calc sr entries for one plane configs */ + if (HAS_FW_BLC(dev) && enabled) { + /* self-refresh has much higher latency */ + static const int sr_latency_ns = 6000; + int clock = enabled->mode.clock; + int htotal = enabled->mode.htotal; + int hdisplay = enabled->mode.hdisplay; + int pixel_size = enabled->fb->bits_per_pixel / 8; + unsigned long line_time_us; + int entries; + + line_time_us = (htotal * 1000) / clock; + + /* Use ns/us then divide to preserve precision */ + entries = (((sr_latency_ns / line_time_us) + 1000) / 1000) * + pixel_size * hdisplay; + entries = DIV_ROUND_UP(entries, wm_info->cacheline_size); + DRM_DEBUG_KMS("self-refresh entries: %d\n", entries); + srwm = wm_info->fifo_size - entries; + if (srwm < 0) + srwm = 1; + + if (IS_I945G(dev) || IS_I945GM(dev)) + I915_WRITE(FW_BLC_SELF, + FW_BLC_SELF_FIFO_MASK | (srwm & 0xff)); + else if (IS_I915GM(dev)) + I915_WRITE(FW_BLC_SELF, srwm & 0x3f); + } + + DRM_DEBUG_KMS("Setting FIFO watermarks - A: %d, B: %d, C: %d, SR %d\n", + planea_wm, planeb_wm, cwm, srwm); + + fwater_lo = ((planeb_wm & 0x3f) << 16) | (planea_wm & 0x3f); + fwater_hi = (cwm & 0x1f); + + /* Set request length to 8 cachelines per fetch */ + fwater_lo = fwater_lo | (1 << 24) | (1 << 8); + fwater_hi = fwater_hi | (1 << 8); + + I915_WRITE(FW_BLC, fwater_lo); + I915_WRITE(FW_BLC2, fwater_hi); + + if (HAS_FW_BLC(dev)) { + if (enabled) { + if (IS_I945G(dev) || IS_I945GM(dev)) + I915_WRITE(FW_BLC_SELF, + FW_BLC_SELF_EN_MASK | FW_BLC_SELF_EN); + else if (IS_I915GM(dev)) + I915_WRITE(INSTPM, I915_READ(INSTPM) | INSTPM_SELF_EN); + DRM_DEBUG_KMS("memory self refresh enabled\n"); + } else + DRM_DEBUG_KMS("memory self refresh disabled\n"); + } +} + +static void i830_update_wm(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_crtc *crtc; + uint32_t fwater_lo; + int planea_wm; + + crtc = single_enabled_crtc(dev); + if (crtc == NULL) + return; + + planea_wm = intel_calculate_wm(crtc->mode.clock, &i830_wm_info, + dev_priv->display.get_fifo_size(dev, 0), + crtc->fb->bits_per_pixel / 8, + latency_ns); + fwater_lo = I915_READ(FW_BLC) & ~0xfff; + fwater_lo |= (3<<8) | planea_wm; + + DRM_DEBUG_KMS("Setting FIFO watermarks - A: %d\n", planea_wm); + + I915_WRITE(FW_BLC, fwater_lo); +} + +#define ILK_LP0_PLANE_LATENCY 700 +#define ILK_LP0_CURSOR_LATENCY 1300 + +/* + * Check the wm result. + * + * If any calculated watermark values is larger than the maximum value that + * can be programmed into the associated watermark register, that watermark + * must be disabled. + */ +static bool ironlake_check_srwm(struct drm_device *dev, int level, + int fbc_wm, int display_wm, int cursor_wm, + const struct intel_watermark_params *display, + const struct intel_watermark_params *cursor) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + DRM_DEBUG_KMS("watermark %d: display plane %d, fbc lines %d," + " cursor %d\n", level, display_wm, fbc_wm, cursor_wm); + + if (fbc_wm > SNB_FBC_MAX_SRWM) { + DRM_DEBUG_KMS("fbc watermark(%d) is too large(%d), disabling wm%d+\n", + fbc_wm, SNB_FBC_MAX_SRWM, level); + + /* fbc has it's own way to disable FBC WM */ + I915_WRITE(DISP_ARB_CTL, + I915_READ(DISP_ARB_CTL) | DISP_FBC_WM_DIS); + return false; + } + + if (display_wm > display->max_wm) { + DRM_DEBUG_KMS("display watermark(%d) is too large(%d), disabling wm%d+\n", + display_wm, SNB_DISPLAY_MAX_SRWM, level); + return false; + } + + if (cursor_wm > cursor->max_wm) { + DRM_DEBUG_KMS("cursor watermark(%d) is too large(%d), disabling wm%d+\n", + cursor_wm, SNB_CURSOR_MAX_SRWM, level); + return false; + } + + if (!(fbc_wm || display_wm || cursor_wm)) { + DRM_DEBUG_KMS("latency %d is 0, disabling wm%d+\n", level, level); + return false; + } + + return true; +} + +/* + * Compute watermark values of WM[1-3], + */ +static bool ironlake_compute_srwm(struct drm_device *dev, int level, int plane, + int latency_ns, + const struct intel_watermark_params *display, + const struct intel_watermark_params *cursor, + int *fbc_wm, int *display_wm, int *cursor_wm) +{ + struct drm_crtc *crtc; + unsigned long line_time_us; + int hdisplay, htotal, pixel_size, clock; + int line_count, line_size; + int small, large; + int entries; + + if (!latency_ns) { + *fbc_wm = *display_wm = *cursor_wm = 0; + return false; + } + + crtc = intel_get_crtc_for_plane(dev, plane); + hdisplay = crtc->mode.hdisplay; + htotal = crtc->mode.htotal; + clock = crtc->mode.clock; + pixel_size = crtc->fb->bits_per_pixel / 8; + + line_time_us = (htotal * 1000) / clock; + line_count = (latency_ns / line_time_us + 1000) / 1000; + line_size = hdisplay * pixel_size; + + /* Use the minimum of the small and large buffer method for primary */ + small = ((clock * pixel_size / 1000) * latency_ns) / 1000; + large = line_count * line_size; + + entries = DIV_ROUND_UP(min(small, large), display->cacheline_size); + *display_wm = entries + display->guard_size; + + /* + * Spec says: + * FBC WM = ((Final Primary WM * 64) / number of bytes per line) + 2 + */ + *fbc_wm = DIV_ROUND_UP(*display_wm * 64, line_size) + 2; + + /* calculate the self-refresh watermark for display cursor */ + entries = line_count * pixel_size * 64; + entries = DIV_ROUND_UP(entries, cursor->cacheline_size); + *cursor_wm = entries + cursor->guard_size; + + return ironlake_check_srwm(dev, level, + *fbc_wm, *display_wm, *cursor_wm, + display, cursor); +} + +static void ironlake_update_wm(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + int fbc_wm, plane_wm, cursor_wm; + unsigned int enabled; + + enabled = 0; + if (g4x_compute_wm0(dev, 0, + &ironlake_display_wm_info, + ILK_LP0_PLANE_LATENCY, + &ironlake_cursor_wm_info, + ILK_LP0_CURSOR_LATENCY, + &plane_wm, &cursor_wm)) { + I915_WRITE(WM0_PIPEA_ILK, + (plane_wm << WM0_PIPE_PLANE_SHIFT) | cursor_wm); + DRM_DEBUG_KMS("FIFO watermarks For pipe A -" + " plane %d, " "cursor: %d\n", + plane_wm, cursor_wm); + enabled |= 1; + } + + if (g4x_compute_wm0(dev, 1, + &ironlake_display_wm_info, + ILK_LP0_PLANE_LATENCY, + &ironlake_cursor_wm_info, + ILK_LP0_CURSOR_LATENCY, + &plane_wm, &cursor_wm)) { + I915_WRITE(WM0_PIPEB_ILK, + (plane_wm << WM0_PIPE_PLANE_SHIFT) | cursor_wm); + DRM_DEBUG_KMS("FIFO watermarks For pipe B -" + " plane %d, cursor: %d\n", + plane_wm, cursor_wm); + enabled |= 2; + } + + /* + * Calculate and update the self-refresh watermark only when one + * display plane is used. + */ + I915_WRITE(WM3_LP_ILK, 0); + I915_WRITE(WM2_LP_ILK, 0); + I915_WRITE(WM1_LP_ILK, 0); + + if (!single_plane_enabled(enabled)) + return; + enabled = ffs(enabled) - 1; + + /* WM1 */ + if (!ironlake_compute_srwm(dev, 1, enabled, + ILK_READ_WM1_LATENCY() * 500, + &ironlake_display_srwm_info, + &ironlake_cursor_srwm_info, + &fbc_wm, &plane_wm, &cursor_wm)) + return; + + I915_WRITE(WM1_LP_ILK, + WM1_LP_SR_EN | + (ILK_READ_WM1_LATENCY() << WM1_LP_LATENCY_SHIFT) | + (fbc_wm << WM1_LP_FBC_SHIFT) | + (plane_wm << WM1_LP_SR_SHIFT) | + cursor_wm); + + /* WM2 */ + if (!ironlake_compute_srwm(dev, 2, enabled, + ILK_READ_WM2_LATENCY() * 500, + &ironlake_display_srwm_info, + &ironlake_cursor_srwm_info, + &fbc_wm, &plane_wm, &cursor_wm)) + return; + + I915_WRITE(WM2_LP_ILK, + WM2_LP_EN | + (ILK_READ_WM2_LATENCY() << WM1_LP_LATENCY_SHIFT) | + (fbc_wm << WM1_LP_FBC_SHIFT) | + (plane_wm << WM1_LP_SR_SHIFT) | + cursor_wm); + + /* + * WM3 is unsupported on ILK, probably because we don't have latency + * data for that power state + */ +} + +static void sandybridge_update_wm(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + int latency = SNB_READ_WM0_LATENCY() * 100; /* In unit 0.1us */ + int fbc_wm, plane_wm, cursor_wm; + unsigned int enabled; + + enabled = 0; + if (g4x_compute_wm0(dev, 0, + &sandybridge_display_wm_info, latency, + &sandybridge_cursor_wm_info, latency, + &plane_wm, &cursor_wm)) { + I915_WRITE(WM0_PIPEA_ILK, + (plane_wm << WM0_PIPE_PLANE_SHIFT) | cursor_wm); + DRM_DEBUG_KMS("FIFO watermarks For pipe A -" + " plane %d, " "cursor: %d\n", + plane_wm, cursor_wm); + enabled |= 1; + } + + if (g4x_compute_wm0(dev, 1, + &sandybridge_display_wm_info, latency, + &sandybridge_cursor_wm_info, latency, + &plane_wm, &cursor_wm)) { + I915_WRITE(WM0_PIPEB_ILK, + (plane_wm << WM0_PIPE_PLANE_SHIFT) | cursor_wm); + DRM_DEBUG_KMS("FIFO watermarks For pipe B -" + " plane %d, cursor: %d\n", + plane_wm, cursor_wm); + enabled |= 2; + } + + /* + * Calculate and update the self-refresh watermark only when one + * display plane is used. + * + * SNB support 3 levels of watermark. + * + * WM1/WM2/WM2 watermarks have to be enabled in the ascending order, + * and disabled in the descending order + * + */ + I915_WRITE(WM3_LP_ILK, 0); + I915_WRITE(WM2_LP_ILK, 0); + I915_WRITE(WM1_LP_ILK, 0); + + if (!single_plane_enabled(enabled)) + return; + enabled = ffs(enabled) - 1; + + /* WM1 */ + if (!ironlake_compute_srwm(dev, 1, enabled, + SNB_READ_WM1_LATENCY() * 500, + &sandybridge_display_srwm_info, + &sandybridge_cursor_srwm_info, + &fbc_wm, &plane_wm, &cursor_wm)) + return; + + I915_WRITE(WM1_LP_ILK, + WM1_LP_SR_EN | + (SNB_READ_WM1_LATENCY() << WM1_LP_LATENCY_SHIFT) | + (fbc_wm << WM1_LP_FBC_SHIFT) | + (plane_wm << WM1_LP_SR_SHIFT) | + cursor_wm); + + /* WM2 */ + if (!ironlake_compute_srwm(dev, 2, enabled, + SNB_READ_WM2_LATENCY() * 500, + &sandybridge_display_srwm_info, + &sandybridge_cursor_srwm_info, + &fbc_wm, &plane_wm, &cursor_wm)) + return; + + I915_WRITE(WM2_LP_ILK, + WM2_LP_EN | + (SNB_READ_WM2_LATENCY() << WM1_LP_LATENCY_SHIFT) | + (fbc_wm << WM1_LP_FBC_SHIFT) | + (plane_wm << WM1_LP_SR_SHIFT) | + cursor_wm); + + /* WM3 */ + if (!ironlake_compute_srwm(dev, 3, enabled, + SNB_READ_WM3_LATENCY() * 500, + &sandybridge_display_srwm_info, + &sandybridge_cursor_srwm_info, + &fbc_wm, &plane_wm, &cursor_wm)) + return; + + I915_WRITE(WM3_LP_ILK, + WM3_LP_EN | + (SNB_READ_WM3_LATENCY() << WM1_LP_LATENCY_SHIFT) | + (fbc_wm << WM1_LP_FBC_SHIFT) | + (plane_wm << WM1_LP_SR_SHIFT) | + cursor_wm); +} + +/** + * intel_update_watermarks - update FIFO watermark values based on current modes + * + * Calculate watermark values for the various WM regs based on current mode + * and plane configuration. + * + * There are several cases to deal with here: + * - normal (i.e. non-self-refresh) + * - self-refresh (SR) mode + * - lines are large relative to FIFO size (buffer can hold up to 2) + * - lines are small relative to FIFO size (buffer can hold more than 2 + * lines), so need to account for TLB latency + * + * The normal calculation is: + * watermark = dotclock * bytes per pixel * latency + * where latency is platform & configuration dependent (we assume pessimal + * values here). + * + * The SR calculation is: + * watermark = (trunc(latency/line time)+1) * surface width * + * bytes per pixel + * where + * line time = htotal / dotclock + * surface width = hdisplay for normal plane and 64 for cursor + * and latency is assumed to be high, as above. + * + * The final value programmed to the register should always be rounded up, + * and include an extra 2 entries to account for clock crossings. + * + * We don't use the sprite, so we can ignore that. And on Crestline we have + * to set the non-SR watermarks to 8. + */ +static void intel_update_watermarks(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + if (dev_priv->display.update_wm) + dev_priv->display.update_wm(dev); +} + +static inline bool intel_panel_use_ssc(struct drm_i915_private *dev_priv) +{ + return dev_priv->lvds_use_ssc && i915_panel_use_ssc + && !(dev_priv->quirks & QUIRK_LVDS_SSC_DISABLE); +} + +static int i9xx_crtc_mode_set(struct drm_crtc *crtc, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode, + int x, int y, + struct drm_framebuffer *old_fb) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int pipe = intel_crtc->pipe; + int plane = intel_crtc->plane; + int refclk, num_connectors = 0; + intel_clock_t clock, reduced_clock; + u32 dpll, fp = 0, fp2 = 0, dspcntr, pipeconf; + bool ok, has_reduced_clock = false, is_sdvo = false, is_dvo = false; + bool is_crt = false, is_lvds = false, is_tv = false, is_dp = false; + struct drm_mode_config *mode_config = &dev->mode_config; + struct intel_encoder *encoder; + const intel_limit_t *limit; + int ret; + u32 temp; + u32 lvds_sync = 0; + + list_for_each_entry(encoder, &mode_config->encoder_list, base.head) { + if (encoder->base.crtc != crtc) + continue; + + switch (encoder->type) { + case INTEL_OUTPUT_LVDS: + is_lvds = true; + break; + case INTEL_OUTPUT_SDVO: + case INTEL_OUTPUT_HDMI: + is_sdvo = true; + if (encoder->needs_tv_clock) + is_tv = true; + break; + case INTEL_OUTPUT_DVO: + is_dvo = true; + break; + case INTEL_OUTPUT_TVOUT: + is_tv = true; + break; + case INTEL_OUTPUT_ANALOG: + is_crt = true; + break; + case INTEL_OUTPUT_DISPLAYPORT: + is_dp = true; + break; + } + + num_connectors++; + } + + if (is_lvds && intel_panel_use_ssc(dev_priv) && num_connectors < 2) { + refclk = dev_priv->lvds_ssc_freq * 1000; + DRM_DEBUG_KMS("using SSC reference clock of %d MHz\n", + refclk / 1000); + } else if (!IS_GEN2(dev)) { + refclk = 96000; + } else { + refclk = 48000; + } + + /* + * Returns a set of divisors for the desired target clock with the given + * refclk, or FALSE. The returned values represent the clock equation: + * reflck * (5 * (m1 + 2) + (m2 + 2)) / (n + 2) / p1 / p2. + */ + limit = intel_limit(crtc, refclk); + ok = limit->find_pll(limit, crtc, adjusted_mode->clock, refclk, &clock); + if (!ok) { + DRM_ERROR("Couldn't find PLL settings for mode!\n"); + return -EINVAL; + } + + /* Ensure that the cursor is valid for the new mode before changing... */ + intel_crtc_update_cursor(crtc, true); + + if (is_lvds && dev_priv->lvds_downclock_avail) { + has_reduced_clock = limit->find_pll(limit, crtc, + dev_priv->lvds_downclock, + refclk, + &reduced_clock); + if (has_reduced_clock && (clock.p != reduced_clock.p)) { + /* + * If the different P is found, it means that we can't + * switch the display clock by using the FP0/FP1. + * In such case we will disable the LVDS downclock + * feature. + */ + DRM_DEBUG_KMS("Different P is found for " + "LVDS clock/downclock\n"); + has_reduced_clock = 0; + } + } + /* SDVO TV has fixed PLL values depend on its clock range, + this mirrors vbios setting. */ + if (is_sdvo && is_tv) { + if (adjusted_mode->clock >= 100000 + && adjusted_mode->clock < 140500) { + clock.p1 = 2; + clock.p2 = 10; + clock.n = 3; + clock.m1 = 16; + clock.m2 = 8; + } else if (adjusted_mode->clock >= 140500 + && adjusted_mode->clock <= 200000) { + clock.p1 = 1; + clock.p2 = 10; + clock.n = 6; + clock.m1 = 12; + clock.m2 = 8; + } + } + + if (IS_PINEVIEW(dev)) { + fp = (1 << clock.n) << 16 | clock.m1 << 8 | clock.m2; + if (has_reduced_clock) + fp2 = (1 << reduced_clock.n) << 16 | + reduced_clock.m1 << 8 | reduced_clock.m2; + } else { + fp = clock.n << 16 | clock.m1 << 8 | clock.m2; + if (has_reduced_clock) + fp2 = reduced_clock.n << 16 | reduced_clock.m1 << 8 | + reduced_clock.m2; + } + + dpll = DPLL_VGA_MODE_DIS; + + if (!IS_GEN2(dev)) { + if (is_lvds) + dpll |= DPLLB_MODE_LVDS; + else + dpll |= DPLLB_MODE_DAC_SERIAL; + if (is_sdvo) { + int pixel_multiplier = intel_mode_get_pixel_multiplier(adjusted_mode); + if (pixel_multiplier > 1) { + if (IS_I945G(dev) || IS_I945GM(dev) || IS_G33(dev)) + dpll |= (pixel_multiplier - 1) << SDVO_MULTIPLIER_SHIFT_HIRES; + } + dpll |= DPLL_DVO_HIGH_SPEED; + } + if (is_dp) + dpll |= DPLL_DVO_HIGH_SPEED; + + /* compute bitmask from p1 value */ + if (IS_PINEVIEW(dev)) + dpll |= (1 << (clock.p1 - 1)) << DPLL_FPA01_P1_POST_DIV_SHIFT_PINEVIEW; + else { + dpll |= (1 << (clock.p1 - 1)) << DPLL_FPA01_P1_POST_DIV_SHIFT; + if (IS_G4X(dev) && has_reduced_clock) + dpll |= (1 << (reduced_clock.p1 - 1)) << DPLL_FPA1_P1_POST_DIV_SHIFT; + } + switch (clock.p2) { + case 5: + dpll |= DPLL_DAC_SERIAL_P2_CLOCK_DIV_5; + break; + case 7: + dpll |= DPLLB_LVDS_P2_CLOCK_DIV_7; + break; + case 10: + dpll |= DPLL_DAC_SERIAL_P2_CLOCK_DIV_10; + break; + case 14: + dpll |= DPLLB_LVDS_P2_CLOCK_DIV_14; + break; + } + if (INTEL_INFO(dev)->gen >= 4) + dpll |= (6 << PLL_LOAD_PULSE_PHASE_SHIFT); + } else { + if (is_lvds) { + dpll |= (1 << (clock.p1 - 1)) << DPLL_FPA01_P1_POST_DIV_SHIFT; + } else { + if (clock.p1 == 2) + dpll |= PLL_P1_DIVIDE_BY_TWO; + else + dpll |= (clock.p1 - 2) << DPLL_FPA01_P1_POST_DIV_SHIFT; + if (clock.p2 == 4) + dpll |= PLL_P2_DIVIDE_BY_4; + } + } + + if (is_sdvo && is_tv) + dpll |= PLL_REF_INPUT_TVCLKINBC; + else if (is_tv) + /* XXX: just matching BIOS for now */ + /* dpll |= PLL_REF_INPUT_TVCLKINBC; */ + dpll |= 3; + else if (is_lvds && intel_panel_use_ssc(dev_priv) && num_connectors < 2) + dpll |= PLLB_REF_INPUT_SPREADSPECTRUMIN; + else + dpll |= PLL_REF_INPUT_DREFCLK; + + /* setup pipeconf */ + pipeconf = I915_READ(PIPECONF(pipe)); + + /* Set up the display plane register */ + dspcntr = DISPPLANE_GAMMA_ENABLE; + + /* Ironlake's plane is forced to pipe, bit 24 is to + enable color space conversion */ + if (pipe == 0) + dspcntr &= ~DISPPLANE_SEL_PIPE_MASK; + else + dspcntr |= DISPPLANE_SEL_PIPE_B; + + if (pipe == 0 && INTEL_INFO(dev)->gen < 4) { + /* Enable pixel doubling when the dot clock is > 90% of the (display) + * core speed. + * + * XXX: No double-wide on 915GM pipe B. Is that the only reason for the + * pipe == 0 check? + */ + if (mode->clock > + dev_priv->display.get_display_clock_speed(dev) * 9 / 10) + pipeconf |= PIPECONF_DOUBLE_WIDE; + else + pipeconf &= ~PIPECONF_DOUBLE_WIDE; + } + + dpll |= DPLL_VCO_ENABLE; + + DRM_DEBUG_KMS("Mode for pipe %c:\n", pipe == 0 ? 'A' : 'B'); + drm_mode_debug_printmodeline(mode); + + I915_WRITE(FP0(pipe), fp); + I915_WRITE(DPLL(pipe), dpll & ~DPLL_VCO_ENABLE); + + POSTING_READ(DPLL(pipe)); + udelay(150); + + /* The LVDS pin pair needs to be on before the DPLLs are enabled. + * This is an exception to the general rule that mode_set doesn't turn + * things on. + */ + if (is_lvds) { + temp = I915_READ(LVDS); + temp |= LVDS_PORT_EN | LVDS_A0A2_CLKA_POWER_UP; + if (pipe == 1) { + temp |= LVDS_PIPEB_SELECT; + } else { + temp &= ~LVDS_PIPEB_SELECT; + } + /* set the corresponsding LVDS_BORDER bit */ + temp |= dev_priv->lvds_border_bits; + /* Set the B0-B3 data pairs corresponding to whether we're going to + * set the DPLLs for dual-channel mode or not. + */ + if (clock.p2 == 7) + temp |= LVDS_B0B3_POWER_UP | LVDS_CLKB_POWER_UP; + else + temp &= ~(LVDS_B0B3_POWER_UP | LVDS_CLKB_POWER_UP); + + /* It would be nice to set 24 vs 18-bit mode (LVDS_A3_POWER_UP) + * appropriately here, but we need to look more thoroughly into how + * panels behave in the two modes. + */ + /* set the dithering flag on LVDS as needed */ + if (INTEL_INFO(dev)->gen >= 4) { + if (dev_priv->lvds_dither) + temp |= LVDS_ENABLE_DITHER; + else + temp &= ~LVDS_ENABLE_DITHER; + } + if (adjusted_mode->flags & DRM_MODE_FLAG_NHSYNC) + lvds_sync |= LVDS_HSYNC_POLARITY; + if (adjusted_mode->flags & DRM_MODE_FLAG_NVSYNC) + lvds_sync |= LVDS_VSYNC_POLARITY; + if ((temp & (LVDS_HSYNC_POLARITY | LVDS_VSYNC_POLARITY)) + != lvds_sync) { + char flags[2] = "-+"; + DRM_INFO("Changing LVDS panel from " + "(%chsync, %cvsync) to (%chsync, %cvsync)\n", + flags[!(temp & LVDS_HSYNC_POLARITY)], + flags[!(temp & LVDS_VSYNC_POLARITY)], + flags[!(lvds_sync & LVDS_HSYNC_POLARITY)], + flags[!(lvds_sync & LVDS_VSYNC_POLARITY)]); + temp &= ~(LVDS_HSYNC_POLARITY | LVDS_VSYNC_POLARITY); + temp |= lvds_sync; + } + I915_WRITE(LVDS, temp); + } + + if (is_dp) { + intel_dp_set_m_n(crtc, mode, adjusted_mode); + } + + I915_WRITE(DPLL(pipe), dpll); + + /* Wait for the clocks to stabilize. */ + POSTING_READ(DPLL(pipe)); + udelay(150); + + if (INTEL_INFO(dev)->gen >= 4) { + temp = 0; + if (is_sdvo) { + temp = intel_mode_get_pixel_multiplier(adjusted_mode); + if (temp > 1) + temp = (temp - 1) << DPLL_MD_UDI_MULTIPLIER_SHIFT; + else + temp = 0; + } + I915_WRITE(DPLL_MD(pipe), temp); + } else { + /* The pixel multiplier can only be updated once the + * DPLL is enabled and the clocks are stable. + * + * So write it again. + */ + I915_WRITE(DPLL(pipe), dpll); + } + + intel_crtc->lowfreq_avail = false; + if (is_lvds && has_reduced_clock && i915_powersave) { + I915_WRITE(FP1(pipe), fp2); + intel_crtc->lowfreq_avail = true; + if (HAS_PIPE_CXSR(dev)) { + DRM_DEBUG_KMS("enabling CxSR downclocking\n"); + pipeconf |= PIPECONF_CXSR_DOWNCLOCK; + } + } else { + I915_WRITE(FP1(pipe), fp); + if (HAS_PIPE_CXSR(dev)) { + DRM_DEBUG_KMS("disabling CxSR downclocking\n"); + pipeconf &= ~PIPECONF_CXSR_DOWNCLOCK; + } + } + + if (adjusted_mode->flags & DRM_MODE_FLAG_INTERLACE) { + pipeconf |= PIPECONF_INTERLACE_W_FIELD_INDICATION; + /* the chip adds 2 halflines automatically */ + adjusted_mode->crtc_vdisplay -= 1; + adjusted_mode->crtc_vtotal -= 1; + adjusted_mode->crtc_vblank_start -= 1; + adjusted_mode->crtc_vblank_end -= 1; + adjusted_mode->crtc_vsync_end -= 1; + adjusted_mode->crtc_vsync_start -= 1; + } else + pipeconf &= ~PIPECONF_INTERLACE_W_FIELD_INDICATION; /* progressive */ + + I915_WRITE(HTOTAL(pipe), + (adjusted_mode->crtc_hdisplay - 1) | + ((adjusted_mode->crtc_htotal - 1) << 16)); + I915_WRITE(HBLANK(pipe), + (adjusted_mode->crtc_hblank_start - 1) | + ((adjusted_mode->crtc_hblank_end - 1) << 16)); + I915_WRITE(HSYNC(pipe), + (adjusted_mode->crtc_hsync_start - 1) | + ((adjusted_mode->crtc_hsync_end - 1) << 16)); + + I915_WRITE(VTOTAL(pipe), + (adjusted_mode->crtc_vdisplay - 1) | + ((adjusted_mode->crtc_vtotal - 1) << 16)); + I915_WRITE(VBLANK(pipe), + (adjusted_mode->crtc_vblank_start - 1) | + ((adjusted_mode->crtc_vblank_end - 1) << 16)); + I915_WRITE(VSYNC(pipe), + (adjusted_mode->crtc_vsync_start - 1) | + ((adjusted_mode->crtc_vsync_end - 1) << 16)); + + /* pipesrc and dspsize control the size that is scaled from, + * which should always be the user's requested size. + */ + I915_WRITE(DSPSIZE(plane), + ((mode->vdisplay - 1) << 16) | + (mode->hdisplay - 1)); + I915_WRITE(DSPPOS(plane), 0); + I915_WRITE(PIPESRC(pipe), + ((mode->hdisplay - 1) << 16) | (mode->vdisplay - 1)); + + I915_WRITE(PIPECONF(pipe), pipeconf); + POSTING_READ(PIPECONF(pipe)); + intel_enable_pipe(dev_priv, pipe, false); + + intel_wait_for_vblank(dev, pipe); + + I915_WRITE(DSPCNTR(plane), dspcntr); + POSTING_READ(DSPCNTR(plane)); + intel_enable_plane(dev_priv, plane, pipe); + + ret = intel_pipe_set_base(crtc, x, y, old_fb); + + intel_update_watermarks(dev); + + return ret; +} + +static int ironlake_crtc_mode_set(struct drm_crtc *crtc, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode, + int x, int y, + struct drm_framebuffer *old_fb) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int pipe = intel_crtc->pipe; + int plane = intel_crtc->plane; + int refclk, num_connectors = 0; + intel_clock_t clock, reduced_clock; + u32 dpll, fp = 0, fp2 = 0, dspcntr, pipeconf; + bool ok, has_reduced_clock = false, is_sdvo = false; + bool is_crt = false, is_lvds = false, is_tv = false, is_dp = false; + struct intel_encoder *has_edp_encoder = NULL; + struct drm_mode_config *mode_config = &dev->mode_config; + struct intel_encoder *encoder; + const intel_limit_t *limit; + int ret; + struct fdi_m_n m_n = {0}; + u32 temp; + u32 lvds_sync = 0; + int target_clock, pixel_multiplier, lane, link_bw, bpp, factor; + + list_for_each_entry(encoder, &mode_config->encoder_list, base.head) { + if (encoder->base.crtc != crtc) + continue; + + switch (encoder->type) { + case INTEL_OUTPUT_LVDS: + is_lvds = true; + break; + case INTEL_OUTPUT_SDVO: + case INTEL_OUTPUT_HDMI: + is_sdvo = true; + if (encoder->needs_tv_clock) + is_tv = true; + break; + case INTEL_OUTPUT_TVOUT: + is_tv = true; + break; + case INTEL_OUTPUT_ANALOG: + is_crt = true; + break; + case INTEL_OUTPUT_DISPLAYPORT: + is_dp = true; + break; + case INTEL_OUTPUT_EDP: + has_edp_encoder = encoder; + break; + } + + num_connectors++; + } + + if (is_lvds && intel_panel_use_ssc(dev_priv) && num_connectors < 2) { + refclk = dev_priv->lvds_ssc_freq * 1000; + DRM_DEBUG_KMS("using SSC reference clock of %d MHz\n", + refclk / 1000); + } else { + refclk = 96000; + if (!has_edp_encoder || + intel_encoder_is_pch_edp(&has_edp_encoder->base)) + refclk = 120000; /* 120Mhz refclk */ + } + + /* + * Returns a set of divisors for the desired target clock with the given + * refclk, or FALSE. The returned values represent the clock equation: + * reflck * (5 * (m1 + 2) + (m2 + 2)) / (n + 2) / p1 / p2. + */ + limit = intel_limit(crtc, refclk); + ok = limit->find_pll(limit, crtc, adjusted_mode->clock, refclk, &clock); + if (!ok) { + DRM_ERROR("Couldn't find PLL settings for mode!\n"); + return -EINVAL; + } + + /* Ensure that the cursor is valid for the new mode before changing... */ + intel_crtc_update_cursor(crtc, true); + + if (is_lvds && dev_priv->lvds_downclock_avail) { + has_reduced_clock = limit->find_pll(limit, crtc, + dev_priv->lvds_downclock, + refclk, + &reduced_clock); + if (has_reduced_clock && (clock.p != reduced_clock.p)) { + /* + * If the different P is found, it means that we can't + * switch the display clock by using the FP0/FP1. + * In such case we will disable the LVDS downclock + * feature. + */ + DRM_DEBUG_KMS("Different P is found for " + "LVDS clock/downclock\n"); + has_reduced_clock = 0; + } + } + /* SDVO TV has fixed PLL values depend on its clock range, + this mirrors vbios setting. */ + if (is_sdvo && is_tv) { + if (adjusted_mode->clock >= 100000 + && adjusted_mode->clock < 140500) { + clock.p1 = 2; + clock.p2 = 10; + clock.n = 3; + clock.m1 = 16; + clock.m2 = 8; + } else if (adjusted_mode->clock >= 140500 + && adjusted_mode->clock <= 200000) { + clock.p1 = 1; + clock.p2 = 10; + clock.n = 6; + clock.m1 = 12; + clock.m2 = 8; + } + } + + /* FDI link */ + pixel_multiplier = intel_mode_get_pixel_multiplier(adjusted_mode); + lane = 0; + /* CPU eDP doesn't require FDI link, so just set DP M/N + according to current link config */ + if (has_edp_encoder && + !intel_encoder_is_pch_edp(&has_edp_encoder->base)) { + target_clock = mode->clock; + intel_edp_link_config(has_edp_encoder, + &lane, &link_bw); + } else { + /* [e]DP over FDI requires target mode clock + instead of link clock */ + if (is_dp || intel_encoder_is_pch_edp(&has_edp_encoder->base)) + target_clock = mode->clock; + else + target_clock = adjusted_mode->clock; + + /* FDI is a binary signal running at ~2.7GHz, encoding + * each output octet as 10 bits. The actual frequency + * is stored as a divider into a 100MHz clock, and the + * mode pixel clock is stored in units of 1KHz. + * Hence the bw of each lane in terms of the mode signal + * is: + */ + link_bw = intel_fdi_link_freq(dev) * MHz(100)/KHz(1)/10; + } + + /* determine panel color depth */ + temp = I915_READ(PIPECONF(pipe)); + temp &= ~PIPE_BPC_MASK; + if (is_lvds) { + /* the BPC will be 6 if it is 18-bit LVDS panel */ + if ((I915_READ(PCH_LVDS) & LVDS_A3_POWER_MASK) == LVDS_A3_POWER_UP) + temp |= PIPE_8BPC; + else + temp |= PIPE_6BPC; + } else if (has_edp_encoder) { + switch (dev_priv->edp.bpp/3) { + case 8: + temp |= PIPE_8BPC; + break; + case 10: + temp |= PIPE_10BPC; + break; + case 6: + temp |= PIPE_6BPC; + break; + case 12: + temp |= PIPE_12BPC; + break; + } + } else + temp |= PIPE_8BPC; + I915_WRITE(PIPECONF(pipe), temp); + + switch (temp & PIPE_BPC_MASK) { + case PIPE_8BPC: + bpp = 24; + break; + case PIPE_10BPC: + bpp = 30; + break; + case PIPE_6BPC: + bpp = 18; + break; + case PIPE_12BPC: + bpp = 36; + break; + default: + DRM_ERROR("unknown pipe bpc value\n"); + bpp = 24; + } + + if (!lane) { + /* + * Account for spread spectrum to avoid + * oversubscribing the link. Max center spread + * is 2.5%; use 5% for safety's sake. + */ + u32 bps = target_clock * bpp * 21 / 20; + lane = bps / (link_bw * 8) + 1; + } + + intel_crtc->fdi_lanes = lane; + + if (pixel_multiplier > 1) + link_bw *= pixel_multiplier; + ironlake_compute_m_n(bpp, lane, target_clock, link_bw, &m_n); + + /* Ironlake: try to setup display ref clock before DPLL + * enabling. This is only under driver's control after + * PCH B stepping, previous chipset stepping should be + * ignoring this setting. + */ + temp = I915_READ(PCH_DREF_CONTROL); + /* Always enable nonspread source */ + temp &= ~DREF_NONSPREAD_SOURCE_MASK; + temp |= DREF_NONSPREAD_SOURCE_ENABLE; + temp &= ~DREF_SSC_SOURCE_MASK; + temp |= DREF_SSC_SOURCE_ENABLE; + I915_WRITE(PCH_DREF_CONTROL, temp); + + POSTING_READ(PCH_DREF_CONTROL); + udelay(200); + + if (has_edp_encoder) { + if (intel_panel_use_ssc(dev_priv)) { + temp |= DREF_SSC1_ENABLE; + I915_WRITE(PCH_DREF_CONTROL, temp); + + POSTING_READ(PCH_DREF_CONTROL); + udelay(200); + } + temp &= ~DREF_CPU_SOURCE_OUTPUT_MASK; + + /* Enable CPU source on CPU attached eDP */ + if (!intel_encoder_is_pch_edp(&has_edp_encoder->base)) { + if (intel_panel_use_ssc(dev_priv)) + temp |= DREF_CPU_SOURCE_OUTPUT_DOWNSPREAD; + else + temp |= DREF_CPU_SOURCE_OUTPUT_NONSPREAD; + } else { + /* Enable SSC on PCH eDP if needed */ + if (intel_panel_use_ssc(dev_priv)) { + DRM_ERROR("enabling SSC on PCH\n"); + temp |= DREF_SUPERSPREAD_SOURCE_ENABLE; + } + } + I915_WRITE(PCH_DREF_CONTROL, temp); + POSTING_READ(PCH_DREF_CONTROL); + udelay(200); + } + + fp = clock.n << 16 | clock.m1 << 8 | clock.m2; + if (has_reduced_clock) + fp2 = reduced_clock.n << 16 | reduced_clock.m1 << 8 | + reduced_clock.m2; + + /* Enable autotuning of the PLL clock (if permissible) */ + factor = 21; + if (is_lvds) { + if ((intel_panel_use_ssc(dev_priv) && + dev_priv->lvds_ssc_freq == 100) || + (I915_READ(PCH_LVDS) & LVDS_CLKB_POWER_MASK) == LVDS_CLKB_POWER_UP) + factor = 25; + } else if (is_sdvo && is_tv) + factor = 20; + + if (clock.m < factor * clock.n) + fp |= FP_CB_TUNE; + + dpll = 0; + + if (is_lvds) + dpll |= DPLLB_MODE_LVDS; + else + dpll |= DPLLB_MODE_DAC_SERIAL; + if (is_sdvo) { + int pixel_multiplier = intel_mode_get_pixel_multiplier(adjusted_mode); + if (pixel_multiplier > 1) { + dpll |= (pixel_multiplier - 1) << PLL_REF_SDVO_HDMI_MULTIPLIER_SHIFT; + } + dpll |= DPLL_DVO_HIGH_SPEED; + } + if (is_dp || intel_encoder_is_pch_edp(&has_edp_encoder->base)) + dpll |= DPLL_DVO_HIGH_SPEED; + + /* compute bitmask from p1 value */ + dpll |= (1 << (clock.p1 - 1)) << DPLL_FPA01_P1_POST_DIV_SHIFT; + /* also FPA1 */ + dpll |= (1 << (clock.p1 - 1)) << DPLL_FPA1_P1_POST_DIV_SHIFT; + + switch (clock.p2) { + case 5: + dpll |= DPLL_DAC_SERIAL_P2_CLOCK_DIV_5; + break; + case 7: + dpll |= DPLLB_LVDS_P2_CLOCK_DIV_7; + break; + case 10: + dpll |= DPLL_DAC_SERIAL_P2_CLOCK_DIV_10; + break; + case 14: + dpll |= DPLLB_LVDS_P2_CLOCK_DIV_14; + break; + } + + if (is_sdvo && is_tv) + dpll |= PLL_REF_INPUT_TVCLKINBC; + else if (is_tv) + /* XXX: just matching BIOS for now */ + /* dpll |= PLL_REF_INPUT_TVCLKINBC; */ + dpll |= 3; + else if (is_lvds && intel_panel_use_ssc(dev_priv) && num_connectors < 2) + dpll |= PLLB_REF_INPUT_SPREADSPECTRUMIN; + else + dpll |= PLL_REF_INPUT_DREFCLK; + + /* setup pipeconf */ + pipeconf = I915_READ(PIPECONF(pipe)); + + /* Set up the display plane register */ + dspcntr = DISPPLANE_GAMMA_ENABLE; + + DRM_DEBUG_KMS("Mode for pipe %c:\n", pipe == 0 ? 'A' : 'B'); + drm_mode_debug_printmodeline(mode); + + /* PCH eDP needs FDI, but CPU eDP does not */ + if (!has_edp_encoder || intel_encoder_is_pch_edp(&has_edp_encoder->base)) { + I915_WRITE(PCH_FP0(pipe), fp); + I915_WRITE(PCH_DPLL(pipe), dpll & ~DPLL_VCO_ENABLE); + + POSTING_READ(PCH_DPLL(pipe)); + udelay(150); + } + + /* enable transcoder DPLL */ + if (HAS_PCH_CPT(dev)) { + temp = I915_READ(PCH_DPLL_SEL); + switch (pipe) { + case 0: + temp |= TRANSA_DPLL_ENABLE | TRANSA_DPLLA_SEL; + break; + case 1: + temp |= TRANSB_DPLL_ENABLE | TRANSB_DPLLB_SEL; + break; + case 2: + /* FIXME: manage transcoder PLLs? */ + temp |= TRANSC_DPLL_ENABLE | TRANSC_DPLLB_SEL; + break; + default: + BUG(); + } + I915_WRITE(PCH_DPLL_SEL, temp); + + POSTING_READ(PCH_DPLL_SEL); + udelay(150); + } + + /* The LVDS pin pair needs to be on before the DPLLs are enabled. + * This is an exception to the general rule that mode_set doesn't turn + * things on. + */ + if (is_lvds) { + temp = I915_READ(PCH_LVDS); + temp |= LVDS_PORT_EN | LVDS_A0A2_CLKA_POWER_UP; + if (pipe == 1) { + if (HAS_PCH_CPT(dev)) + temp |= PORT_TRANS_B_SEL_CPT; + else + temp |= LVDS_PIPEB_SELECT; + } else { + if (HAS_PCH_CPT(dev)) + temp &= ~PORT_TRANS_SEL_MASK; + else + temp &= ~LVDS_PIPEB_SELECT; + } + /* set the corresponsding LVDS_BORDER bit */ + temp |= dev_priv->lvds_border_bits; + /* Set the B0-B3 data pairs corresponding to whether we're going to + * set the DPLLs for dual-channel mode or not. + */ + if (clock.p2 == 7) + temp |= LVDS_B0B3_POWER_UP | LVDS_CLKB_POWER_UP; + else + temp &= ~(LVDS_B0B3_POWER_UP | LVDS_CLKB_POWER_UP); + + /* It would be nice to set 24 vs 18-bit mode (LVDS_A3_POWER_UP) + * appropriately here, but we need to look more thoroughly into how + * panels behave in the two modes. + */ + if (adjusted_mode->flags & DRM_MODE_FLAG_NHSYNC) + lvds_sync |= LVDS_HSYNC_POLARITY; + if (adjusted_mode->flags & DRM_MODE_FLAG_NVSYNC) + lvds_sync |= LVDS_VSYNC_POLARITY; + if ((temp & (LVDS_HSYNC_POLARITY | LVDS_VSYNC_POLARITY)) + != lvds_sync) { + char flags[2] = "-+"; + DRM_INFO("Changing LVDS panel from " + "(%chsync, %cvsync) to (%chsync, %cvsync)\n", + flags[!(temp & LVDS_HSYNC_POLARITY)], + flags[!(temp & LVDS_VSYNC_POLARITY)], + flags[!(lvds_sync & LVDS_HSYNC_POLARITY)], + flags[!(lvds_sync & LVDS_VSYNC_POLARITY)]); + temp &= ~(LVDS_HSYNC_POLARITY | LVDS_VSYNC_POLARITY); + temp |= lvds_sync; + } + I915_WRITE(PCH_LVDS, temp); + } + + /* set the dithering flag and clear for anything other than a panel. */ + pipeconf &= ~PIPECONF_DITHER_EN; + pipeconf &= ~PIPECONF_DITHER_TYPE_MASK; + if (dev_priv->lvds_dither && (is_lvds || has_edp_encoder)) { + pipeconf |= PIPECONF_DITHER_EN; + pipeconf |= PIPECONF_DITHER_TYPE_ST1; + } + + if (is_dp || intel_encoder_is_pch_edp(&has_edp_encoder->base)) { + intel_dp_set_m_n(crtc, mode, adjusted_mode); + } else { + /* For non-DP output, clear any trans DP clock recovery setting.*/ + I915_WRITE(TRANSDATA_M1(pipe), 0); + I915_WRITE(TRANSDATA_N1(pipe), 0); + I915_WRITE(TRANSDPLINK_M1(pipe), 0); + I915_WRITE(TRANSDPLINK_N1(pipe), 0); + } + + if (!has_edp_encoder || + intel_encoder_is_pch_edp(&has_edp_encoder->base)) { + I915_WRITE(PCH_DPLL(pipe), dpll); + + /* Wait for the clocks to stabilize. */ + POSTING_READ(PCH_DPLL(pipe)); + udelay(150); + + /* The pixel multiplier can only be updated once the + * DPLL is enabled and the clocks are stable. + * + * So write it again. + */ + I915_WRITE(PCH_DPLL(pipe), dpll); + } + + intel_crtc->lowfreq_avail = false; + if (is_lvds && has_reduced_clock && i915_powersave) { + I915_WRITE(PCH_FP1(pipe), fp2); + intel_crtc->lowfreq_avail = true; + if (HAS_PIPE_CXSR(dev)) { + DRM_DEBUG_KMS("enabling CxSR downclocking\n"); + pipeconf |= PIPECONF_CXSR_DOWNCLOCK; + } + } else { + I915_WRITE(PCH_FP1(pipe), fp); + if (HAS_PIPE_CXSR(dev)) { + DRM_DEBUG_KMS("disabling CxSR downclocking\n"); + pipeconf &= ~PIPECONF_CXSR_DOWNCLOCK; + } + } + + if (adjusted_mode->flags & DRM_MODE_FLAG_INTERLACE) { + pipeconf |= PIPECONF_INTERLACE_W_FIELD_INDICATION; + /* the chip adds 2 halflines automatically */ + adjusted_mode->crtc_vdisplay -= 1; + adjusted_mode->crtc_vtotal -= 1; + adjusted_mode->crtc_vblank_start -= 1; + adjusted_mode->crtc_vblank_end -= 1; + adjusted_mode->crtc_vsync_end -= 1; + adjusted_mode->crtc_vsync_start -= 1; + } else + pipeconf &= ~PIPECONF_INTERLACE_W_FIELD_INDICATION; /* progressive */ + + I915_WRITE(HTOTAL(pipe), + (adjusted_mode->crtc_hdisplay - 1) | + ((adjusted_mode->crtc_htotal - 1) << 16)); + I915_WRITE(HBLANK(pipe), + (adjusted_mode->crtc_hblank_start - 1) | + ((adjusted_mode->crtc_hblank_end - 1) << 16)); + I915_WRITE(HSYNC(pipe), + (adjusted_mode->crtc_hsync_start - 1) | + ((adjusted_mode->crtc_hsync_end - 1) << 16)); + + I915_WRITE(VTOTAL(pipe), + (adjusted_mode->crtc_vdisplay - 1) | + ((adjusted_mode->crtc_vtotal - 1) << 16)); + I915_WRITE(VBLANK(pipe), + (adjusted_mode->crtc_vblank_start - 1) | + ((adjusted_mode->crtc_vblank_end - 1) << 16)); + I915_WRITE(VSYNC(pipe), + (adjusted_mode->crtc_vsync_start - 1) | + ((adjusted_mode->crtc_vsync_end - 1) << 16)); + + /* pipesrc controls the size that is scaled from, which should + * always be the user's requested size. + */ + I915_WRITE(PIPESRC(pipe), + ((mode->hdisplay - 1) << 16) | (mode->vdisplay - 1)); + + I915_WRITE(PIPE_DATA_M1(pipe), TU_SIZE(m_n.tu) | m_n.gmch_m); + I915_WRITE(PIPE_DATA_N1(pipe), m_n.gmch_n); + I915_WRITE(PIPE_LINK_M1(pipe), m_n.link_m); + I915_WRITE(PIPE_LINK_N1(pipe), m_n.link_n); + + if (has_edp_encoder && + !intel_encoder_is_pch_edp(&has_edp_encoder->base)) { + ironlake_set_pll_edp(crtc, adjusted_mode->clock); + } + + I915_WRITE(PIPECONF(pipe), pipeconf); + POSTING_READ(PIPECONF(pipe)); + + intel_wait_for_vblank(dev, pipe); + + if (IS_GEN5(dev)) { + /* enable address swizzle for tiling buffer */ + temp = I915_READ(DISP_ARB_CTL); + I915_WRITE(DISP_ARB_CTL, temp | DISP_TILE_SURFACE_SWIZZLING); + } + + I915_WRITE(DSPCNTR(plane), dspcntr); + POSTING_READ(DSPCNTR(plane)); + + ret = intel_pipe_set_base(crtc, x, y, old_fb); + + intel_update_watermarks(dev); + + return ret; +} + +static int intel_crtc_mode_set(struct drm_crtc *crtc, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode, + int x, int y, + struct drm_framebuffer *old_fb) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int pipe = intel_crtc->pipe; + int ret; + + drm_vblank_pre_modeset(dev, pipe); + + ret = dev_priv->display.crtc_mode_set(crtc, mode, adjusted_mode, + x, y, old_fb); + + drm_vblank_post_modeset(dev, pipe); + + return ret; +} + +/** Loads the palette/gamma unit for the CRTC with the prepared values */ +void intel_crtc_load_lut(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int palreg = PALETTE(intel_crtc->pipe); + int i; + + /* The clocks have to be on to load the palette. */ + if (!crtc->enabled || !intel_crtc->active) + return; + + /* use legacy palette for Ironlake */ + if (HAS_PCH_SPLIT(dev)) + palreg = LGC_PALETTE(intel_crtc->pipe); + + for (i = 0; i < 256; i++) { + I915_WRITE(palreg + 4 * i, + (intel_crtc->lut_r[i] << 16) | + (intel_crtc->lut_g[i] << 8) | + intel_crtc->lut_b[i]); + } +} + +static void i845_update_cursor(struct drm_crtc *crtc, u32 base) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + bool visible = base != 0; + u32 cntl; + + if (intel_crtc->cursor_visible == visible) + return; + + cntl = I915_READ(_CURACNTR); + if (visible) { + /* On these chipsets we can only modify the base whilst + * the cursor is disabled. + */ + I915_WRITE(_CURABASE, base); + + cntl &= ~(CURSOR_FORMAT_MASK); + /* XXX width must be 64, stride 256 => 0x00 << 28 */ + cntl |= CURSOR_ENABLE | + CURSOR_GAMMA_ENABLE | + CURSOR_FORMAT_ARGB; + } else + cntl &= ~(CURSOR_ENABLE | CURSOR_GAMMA_ENABLE); + I915_WRITE(_CURACNTR, cntl); + + intel_crtc->cursor_visible = visible; +} + +static void i9xx_update_cursor(struct drm_crtc *crtc, u32 base) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int pipe = intel_crtc->pipe; + bool visible = base != 0; + + if (intel_crtc->cursor_visible != visible) { + uint32_t cntl = I915_READ(CURCNTR(pipe)); + if (base) { + cntl &= ~(CURSOR_MODE | MCURSOR_PIPE_SELECT); + cntl |= CURSOR_MODE_64_ARGB_AX | MCURSOR_GAMMA_ENABLE; + cntl |= pipe << 28; /* Connect to correct pipe */ + } else { + cntl &= ~(CURSOR_MODE | MCURSOR_GAMMA_ENABLE); + cntl |= CURSOR_MODE_DISABLE; + } + I915_WRITE(CURCNTR(pipe), cntl); + + intel_crtc->cursor_visible = visible; + } + /* and commit changes on next vblank */ + I915_WRITE(CURBASE(pipe), base); +} + +static void ivb_update_cursor(struct drm_crtc *crtc, u32 base) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int pipe = intel_crtc->pipe; + bool visible = base != 0; + + if (intel_crtc->cursor_visible != visible) { + uint32_t cntl = I915_READ(CURCNTR_IVB(pipe)); + if (base) { + cntl &= ~CURSOR_MODE; + cntl |= CURSOR_MODE_64_ARGB_AX | MCURSOR_GAMMA_ENABLE; + } else { + cntl &= ~(CURSOR_MODE | MCURSOR_GAMMA_ENABLE); + cntl |= CURSOR_MODE_DISABLE; + } + I915_WRITE(CURCNTR_IVB(pipe), cntl); + + intel_crtc->cursor_visible = visible; + } + /* and commit changes on next vblank */ + I915_WRITE(CURBASE_IVB(pipe), base); +} + +/* If no-part of the cursor is visible on the framebuffer, then the GPU may hang... */ +static void intel_crtc_update_cursor(struct drm_crtc *crtc, + bool on) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int pipe = intel_crtc->pipe; + int x = intel_crtc->cursor_x; + int y = intel_crtc->cursor_y; + u32 base, pos; + bool visible; + + pos = 0; + + if (on && crtc->enabled && crtc->fb) { + base = intel_crtc->cursor_addr; + if (x > (int) crtc->fb->width) + base = 0; + + if (y > (int) crtc->fb->height) + base = 0; + } else + base = 0; + + if (x < 0) { + if (x + intel_crtc->cursor_width < 0) + base = 0; + + pos |= CURSOR_POS_SIGN << CURSOR_X_SHIFT; + x = -x; + } + pos |= x << CURSOR_X_SHIFT; + + if (y < 0) { + if (y + intel_crtc->cursor_height < 0) + base = 0; + + pos |= CURSOR_POS_SIGN << CURSOR_Y_SHIFT; + y = -y; + } + pos |= y << CURSOR_Y_SHIFT; + + visible = base != 0; + if (!visible && !intel_crtc->cursor_visible) + return; + + if (IS_IVYBRIDGE(dev)) { + I915_WRITE(CURPOS_IVB(pipe), pos); + ivb_update_cursor(crtc, base); + } else { + I915_WRITE(CURPOS(pipe), pos); + if (IS_845G(dev) || IS_I865G(dev)) + i845_update_cursor(crtc, base); + else + i9xx_update_cursor(crtc, base); + } + + if (visible) + intel_mark_busy(dev, to_intel_framebuffer(crtc->fb)->obj); +} + +static int intel_crtc_cursor_set(struct drm_crtc *crtc, + struct drm_file *file, + uint32_t handle, + uint32_t width, uint32_t height) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + struct drm_i915_gem_object *obj; + uint32_t addr; + int ret; + + DRM_DEBUG_KMS("\n"); + + /* if we want to turn off the cursor ignore width and height */ + if (!handle) { + DRM_DEBUG_KMS("cursor off\n"); + addr = 0; + obj = NULL; + mutex_lock(&dev->struct_mutex); + goto finish; + } + + /* Currently we only support 64x64 cursors */ + if (width != 64 || height != 64) { + DRM_ERROR("we currently only support 64x64 cursors\n"); + return -EINVAL; + } + + obj = to_intel_bo(drm_gem_object_lookup(dev, file, handle)); + if (&obj->base == NULL) + return -ENOENT; + + if (obj->base.size < width * height * 4) { + DRM_ERROR("buffer is to small\n"); + ret = -ENOMEM; + goto fail; + } + + /* we only need to pin inside GTT if cursor is non-phy */ + mutex_lock(&dev->struct_mutex); + if (!dev_priv->info->cursor_needs_physical) { + if (obj->tiling_mode) { + DRM_ERROR("cursor cannot be tiled\n"); + ret = -EINVAL; + goto fail_locked; + } + + ret = i915_gem_object_pin(obj, PAGE_SIZE, true); + if (ret) { + DRM_ERROR("failed to pin cursor bo\n"); + goto fail_locked; + } + + ret = i915_gem_object_set_to_gtt_domain(obj, 0); + if (ret) { + DRM_ERROR("failed to move cursor bo into the GTT\n"); + goto fail_unpin; + } + + ret = i915_gem_object_put_fence(obj); + if (ret) { + DRM_ERROR("failed to move cursor bo into the GTT\n"); + goto fail_unpin; + } + + addr = obj->gtt_offset; + } else { + int align = IS_I830(dev) ? 16 * 1024 : 256; + ret = i915_gem_attach_phys_object(dev, obj, + (intel_crtc->pipe == 0) ? I915_GEM_PHYS_CURSOR_0 : I915_GEM_PHYS_CURSOR_1, + align); + if (ret) { + DRM_ERROR("failed to attach phys object\n"); + goto fail_locked; + } + addr = obj->phys_obj->handle->busaddr; + } + + if (IS_GEN2(dev)) + I915_WRITE(CURSIZE, (height << 12) | width); + + finish: + if (intel_crtc->cursor_bo) { + if (dev_priv->info->cursor_needs_physical) { + if (intel_crtc->cursor_bo != obj) + i915_gem_detach_phys_object(dev, intel_crtc->cursor_bo); + } else + i915_gem_object_unpin(intel_crtc->cursor_bo); + drm_gem_object_unreference(&intel_crtc->cursor_bo->base); + } + + mutex_unlock(&dev->struct_mutex); + + intel_crtc->cursor_addr = addr; + intel_crtc->cursor_bo = obj; + intel_crtc->cursor_width = width; + intel_crtc->cursor_height = height; + + intel_crtc_update_cursor(crtc, true); + + return 0; +fail_unpin: + i915_gem_object_unpin(obj); +fail_locked: + mutex_unlock(&dev->struct_mutex); +fail: + drm_gem_object_unreference_unlocked(&obj->base); + return ret; +} + +static int intel_crtc_cursor_move(struct drm_crtc *crtc, int x, int y) +{ + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + + intel_crtc->cursor_x = x; + intel_crtc->cursor_y = y; + + intel_crtc_update_cursor(crtc, true); + + return 0; +} + +/** Sets the color ramps on behalf of RandR */ +void intel_crtc_fb_gamma_set(struct drm_crtc *crtc, u16 red, u16 green, + u16 blue, int regno) +{ + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + + intel_crtc->lut_r[regno] = red >> 8; + intel_crtc->lut_g[regno] = green >> 8; + intel_crtc->lut_b[regno] = blue >> 8; +} + +void intel_crtc_fb_gamma_get(struct drm_crtc *crtc, u16 *red, u16 *green, + u16 *blue, int regno) +{ + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + + *red = intel_crtc->lut_r[regno] << 8; + *green = intel_crtc->lut_g[regno] << 8; + *blue = intel_crtc->lut_b[regno] << 8; +} + +static void intel_crtc_gamma_set(struct drm_crtc *crtc, u16 *red, u16 *green, + u16 *blue, uint32_t start, uint32_t size) +{ + int end = (start + size > 256) ? 256 : start + size, i; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + + for (i = start; i < end; i++) { + intel_crtc->lut_r[i] = red[i] >> 8; + intel_crtc->lut_g[i] = green[i] >> 8; + intel_crtc->lut_b[i] = blue[i] >> 8; + } + + intel_crtc_load_lut(crtc); +} + +/** + * Get a pipe with a simple mode set on it for doing load-based monitor + * detection. + * + * It will be up to the load-detect code to adjust the pipe as appropriate for + * its requirements. The pipe will be connected to no other encoders. + * + * Currently this code will only succeed if there is a pipe with no encoders + * configured for it. In the future, it could choose to temporarily disable + * some outputs to free up a pipe for its use. + * + * \return crtc, or NULL if no pipes are available. + */ + +/* VESA 640x480x72Hz mode to set on the pipe */ +static struct drm_display_mode load_detect_mode = { + DRM_MODE("640x480", DRM_MODE_TYPE_DEFAULT, 31500, 640, 664, + 704, 832, 0, 480, 489, 491, 520, 0, DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_NVSYNC), +}; + +static struct drm_framebuffer * +intel_framebuffer_create(struct drm_device *dev, + struct drm_mode_fb_cmd *mode_cmd, + struct drm_i915_gem_object *obj) +{ + struct intel_framebuffer *intel_fb; + int ret; + + intel_fb = kzalloc(sizeof(*intel_fb), GFP_KERNEL); + if (!intel_fb) { + drm_gem_object_unreference_unlocked(&obj->base); + return ERR_PTR(-ENOMEM); + } + + ret = intel_framebuffer_init(dev, intel_fb, mode_cmd, obj); + if (ret) { + drm_gem_object_unreference_unlocked(&obj->base); + kfree(intel_fb); + return ERR_PTR(ret); + } + + return &intel_fb->base; +} + +static u32 +intel_framebuffer_pitch_for_width(int width, int bpp) +{ + u32 pitch = DIV_ROUND_UP(width * bpp, 8); + return ALIGN(pitch, 64); +} + +static u32 +intel_framebuffer_size_for_mode(struct drm_display_mode *mode, int bpp) +{ + u32 pitch = intel_framebuffer_pitch_for_width(mode->hdisplay, bpp); + return ALIGN(pitch * mode->vdisplay, PAGE_SIZE); +} + +static struct drm_framebuffer * +intel_framebuffer_create_for_mode(struct drm_device *dev, + struct drm_display_mode *mode, + int depth, int bpp) +{ + struct drm_i915_gem_object *obj; + struct drm_mode_fb_cmd mode_cmd; + + obj = i915_gem_alloc_object(dev, + intel_framebuffer_size_for_mode(mode, bpp)); + if (obj == NULL) + return ERR_PTR(-ENOMEM); + + mode_cmd.width = mode->hdisplay; + mode_cmd.height = mode->vdisplay; + mode_cmd.depth = depth; + mode_cmd.bpp = bpp; + mode_cmd.pitch = intel_framebuffer_pitch_for_width(mode_cmd.width, bpp); + + return intel_framebuffer_create(dev, &mode_cmd, obj); +} + +static struct drm_framebuffer * +mode_fits_in_fbdev(struct drm_device *dev, + struct drm_display_mode *mode) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_i915_gem_object *obj; + struct drm_framebuffer *fb; + + if (dev_priv->fbdev == NULL) + return NULL; + + obj = dev_priv->fbdev->ifb.obj; + if (obj == NULL) + return NULL; + + fb = &dev_priv->fbdev->ifb.base; + if (fb->pitch < intel_framebuffer_pitch_for_width(mode->hdisplay, + fb->bits_per_pixel)) + return NULL; + + if (obj->base.size < mode->vdisplay * fb->pitch) + return NULL; + + return fb; +} + +bool intel_get_load_detect_pipe(struct intel_encoder *intel_encoder, + struct drm_connector *connector, + struct drm_display_mode *mode, + struct intel_load_detect_pipe *old) +{ + struct intel_crtc *intel_crtc; + struct drm_crtc *possible_crtc; + struct drm_encoder *encoder = &intel_encoder->base; + struct drm_crtc *crtc = NULL; + struct drm_device *dev = encoder->dev; + struct drm_framebuffer *old_fb; + int i = -1; + + DRM_DEBUG_KMS("[CONNECTOR:%d:%s], [ENCODER:%d:%s]\n", + connector->base.id, drm_get_connector_name(connector), + encoder->base.id, drm_get_encoder_name(encoder)); + + /* + * Algorithm gets a little messy: + * + * - if the connector already has an assigned crtc, use it (but make + * sure it's on first) + * + * - try to find the first unused crtc that can drive this connector, + * and use that if we find one + */ + + /* See if we already have a CRTC for this connector */ + if (encoder->crtc) { + crtc = encoder->crtc; + + intel_crtc = to_intel_crtc(crtc); + old->dpms_mode = intel_crtc->dpms_mode; + old->load_detect_temp = false; + + /* Make sure the crtc and connector are running */ + if (intel_crtc->dpms_mode != DRM_MODE_DPMS_ON) { + struct drm_encoder_helper_funcs *encoder_funcs; + struct drm_crtc_helper_funcs *crtc_funcs; + + crtc_funcs = crtc->helper_private; + crtc_funcs->dpms(crtc, DRM_MODE_DPMS_ON); + + encoder_funcs = encoder->helper_private; + encoder_funcs->dpms(encoder, DRM_MODE_DPMS_ON); + } + + return true; + } + + /* Find an unused one (if possible) */ + list_for_each_entry(possible_crtc, &dev->mode_config.crtc_list, head) { + i++; + if (!(encoder->possible_crtcs & (1 << i))) + continue; + if (!possible_crtc->enabled) { + crtc = possible_crtc; + break; + } + } + + /* + * If we didn't find an unused CRTC, don't use any. + */ + if (!crtc) { + DRM_DEBUG_KMS("no pipe available for load-detect\n"); + return false; + } + + encoder->crtc = crtc; + connector->encoder = encoder; + + intel_crtc = to_intel_crtc(crtc); + old->dpms_mode = intel_crtc->dpms_mode; + old->load_detect_temp = true; + old->release_fb = NULL; + + if (!mode) + mode = &load_detect_mode; + + old_fb = crtc->fb; + + /* We need a framebuffer large enough to accommodate all accesses + * that the plane may generate whilst we perform load detection. + * We can not rely on the fbcon either being present (we get called + * during its initialisation to detect all boot displays, or it may + * not even exist) or that it is large enough to satisfy the + * requested mode. + */ + crtc->fb = mode_fits_in_fbdev(dev, mode); + if (crtc->fb == NULL) { + DRM_DEBUG_KMS("creating tmp fb for load-detection\n"); + crtc->fb = intel_framebuffer_create_for_mode(dev, mode, 24, 32); + old->release_fb = crtc->fb; + } else + DRM_DEBUG_KMS("reusing fbdev for load-detection framebuffer\n"); + if (IS_ERR(crtc->fb)) { + DRM_DEBUG_KMS("failed to allocate framebuffer for load-detection\n"); + crtc->fb = old_fb; + return false; + } + + if (!drm_crtc_helper_set_mode(crtc, mode, 0, 0, old_fb)) { + DRM_DEBUG_KMS("failed to set mode on load-detect pipe\n"); + if (old->release_fb) + old->release_fb->funcs->destroy(old->release_fb); + crtc->fb = old_fb; + return false; + } + + /* let the connector get through one full cycle before testing */ + intel_wait_for_vblank(dev, intel_crtc->pipe); + + return true; +} + +void intel_release_load_detect_pipe(struct intel_encoder *intel_encoder, + struct drm_connector *connector, + struct intel_load_detect_pipe *old) +{ + struct drm_encoder *encoder = &intel_encoder->base; + struct drm_device *dev = encoder->dev; + struct drm_crtc *crtc = encoder->crtc; + struct drm_encoder_helper_funcs *encoder_funcs = encoder->helper_private; + struct drm_crtc_helper_funcs *crtc_funcs = crtc->helper_private; + + DRM_DEBUG_KMS("[CONNECTOR:%d:%s], [ENCODER:%d:%s]\n", + connector->base.id, drm_get_connector_name(connector), + encoder->base.id, drm_get_encoder_name(encoder)); + + if (old->load_detect_temp) { + connector->encoder = NULL; + drm_helper_disable_unused_functions(dev); + + if (old->release_fb) + old->release_fb->funcs->destroy(old->release_fb); + + return; + } + + /* Switch crtc and encoder back off if necessary */ + if (old->dpms_mode != DRM_MODE_DPMS_ON) { + encoder_funcs->dpms(encoder, old->dpms_mode); + crtc_funcs->dpms(crtc, old->dpms_mode); + } +} + +/* Returns the clock of the currently programmed mode of the given pipe. */ +static int intel_crtc_clock_get(struct drm_device *dev, struct drm_crtc *crtc) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int pipe = intel_crtc->pipe; + u32 dpll = I915_READ(DPLL(pipe)); + u32 fp; + intel_clock_t clock; + + if ((dpll & DISPLAY_RATE_SELECT_FPA1) == 0) + fp = I915_READ(FP0(pipe)); + else + fp = I915_READ(FP1(pipe)); + + clock.m1 = (fp & FP_M1_DIV_MASK) >> FP_M1_DIV_SHIFT; + if (IS_PINEVIEW(dev)) { + clock.n = ffs((fp & FP_N_PINEVIEW_DIV_MASK) >> FP_N_DIV_SHIFT) - 1; + clock.m2 = (fp & FP_M2_PINEVIEW_DIV_MASK) >> FP_M2_DIV_SHIFT; + } else { + clock.n = (fp & FP_N_DIV_MASK) >> FP_N_DIV_SHIFT; + clock.m2 = (fp & FP_M2_DIV_MASK) >> FP_M2_DIV_SHIFT; + } + + if (!IS_GEN2(dev)) { + if (IS_PINEVIEW(dev)) + clock.p1 = ffs((dpll & DPLL_FPA01_P1_POST_DIV_MASK_PINEVIEW) >> + DPLL_FPA01_P1_POST_DIV_SHIFT_PINEVIEW); + else + clock.p1 = ffs((dpll & DPLL_FPA01_P1_POST_DIV_MASK) >> + DPLL_FPA01_P1_POST_DIV_SHIFT); + + switch (dpll & DPLL_MODE_MASK) { + case DPLLB_MODE_DAC_SERIAL: + clock.p2 = dpll & DPLL_DAC_SERIAL_P2_CLOCK_DIV_5 ? + 5 : 10; + break; + case DPLLB_MODE_LVDS: + clock.p2 = dpll & DPLLB_LVDS_P2_CLOCK_DIV_7 ? + 7 : 14; + break; + default: + DRM_DEBUG_KMS("Unknown DPLL mode %08x in programmed " + "mode\n", (int)(dpll & DPLL_MODE_MASK)); + return 0; + } + + /* XXX: Handle the 100Mhz refclk */ + intel_clock(dev, 96000, &clock); + } else { + bool is_lvds = (pipe == 1) && (I915_READ(LVDS) & LVDS_PORT_EN); + + if (is_lvds) { + clock.p1 = ffs((dpll & DPLL_FPA01_P1_POST_DIV_MASK_I830_LVDS) >> + DPLL_FPA01_P1_POST_DIV_SHIFT); + clock.p2 = 14; + + if ((dpll & PLL_REF_INPUT_MASK) == + PLLB_REF_INPUT_SPREADSPECTRUMIN) { + /* XXX: might not be 66MHz */ + intel_clock(dev, 66000, &clock); + } else + intel_clock(dev, 48000, &clock); + } else { + if (dpll & PLL_P1_DIVIDE_BY_TWO) + clock.p1 = 2; + else { + clock.p1 = ((dpll & DPLL_FPA01_P1_POST_DIV_MASK_I830) >> + DPLL_FPA01_P1_POST_DIV_SHIFT) + 2; + } + if (dpll & PLL_P2_DIVIDE_BY_4) + clock.p2 = 4; + else + clock.p2 = 2; + + intel_clock(dev, 48000, &clock); + } + } + + /* XXX: It would be nice to validate the clocks, but we can't reuse + * i830PllIsValid() because it relies on the xf86_config connector + * configuration being accurate, which it isn't necessarily. + */ + + return clock.dot; +} + +/** Returns the currently programmed mode of the given pipe. */ +struct drm_display_mode *intel_crtc_mode_get(struct drm_device *dev, + struct drm_crtc *crtc) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int pipe = intel_crtc->pipe; + struct drm_display_mode *mode; + int htot = I915_READ(HTOTAL(pipe)); + int hsync = I915_READ(HSYNC(pipe)); + int vtot = I915_READ(VTOTAL(pipe)); + int vsync = I915_READ(VSYNC(pipe)); + + mode = kzalloc(sizeof(*mode), GFP_KERNEL); + if (!mode) + return NULL; + + mode->clock = intel_crtc_clock_get(dev, crtc); + mode->hdisplay = (htot & 0xffff) + 1; + mode->htotal = ((htot & 0xffff0000) >> 16) + 1; + mode->hsync_start = (hsync & 0xffff) + 1; + mode->hsync_end = ((hsync & 0xffff0000) >> 16) + 1; + mode->vdisplay = (vtot & 0xffff) + 1; + mode->vtotal = ((vtot & 0xffff0000) >> 16) + 1; + mode->vsync_start = (vsync & 0xffff) + 1; + mode->vsync_end = ((vsync & 0xffff0000) >> 16) + 1; + + drm_mode_set_name(mode); + drm_mode_set_crtcinfo(mode, 0); + + return mode; +} + +#define GPU_IDLE_TIMEOUT 500 /* ms */ + +/* When this timer fires, we've been idle for awhile */ +static void intel_gpu_idle_timer(unsigned long arg) +{ + struct drm_device *dev = (struct drm_device *)arg; + drm_i915_private_t *dev_priv = dev->dev_private; + + if (!list_empty(&dev_priv->mm.active_list)) { + /* Still processing requests, so just re-arm the timer. */ + mod_timer(&dev_priv->idle_timer, jiffies + + msecs_to_jiffies(GPU_IDLE_TIMEOUT)); + return; + } + + dev_priv->busy = false; + queue_work(dev_priv->wq, &dev_priv->idle_work); +} + +#define CRTC_IDLE_TIMEOUT 1000 /* ms */ + +static void intel_crtc_idle_timer(unsigned long arg) +{ + struct intel_crtc *intel_crtc = (struct intel_crtc *)arg; + struct drm_crtc *crtc = &intel_crtc->base; + drm_i915_private_t *dev_priv = crtc->dev->dev_private; + struct intel_framebuffer *intel_fb; + + intel_fb = to_intel_framebuffer(crtc->fb); + if (intel_fb && intel_fb->obj->active) { + /* The framebuffer is still being accessed by the GPU. */ + mod_timer(&intel_crtc->idle_timer, jiffies + + msecs_to_jiffies(CRTC_IDLE_TIMEOUT)); + return; + } + + intel_crtc->busy = false; + queue_work(dev_priv->wq, &dev_priv->idle_work); +} + +static void intel_increase_pllclock(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int pipe = intel_crtc->pipe; + int dpll_reg = DPLL(pipe); + int dpll; + + if (HAS_PCH_SPLIT(dev)) + return; + + if (!dev_priv->lvds_downclock_avail) + return; + + dpll = I915_READ(dpll_reg); + if (!HAS_PIPE_CXSR(dev) && (dpll & DISPLAY_RATE_SELECT_FPA1)) { + DRM_DEBUG_DRIVER("upclocking LVDS\n"); + + /* Unlock panel regs */ + I915_WRITE(PP_CONTROL, + I915_READ(PP_CONTROL) | PANEL_UNLOCK_REGS); + + dpll &= ~DISPLAY_RATE_SELECT_FPA1; + I915_WRITE(dpll_reg, dpll); + intel_wait_for_vblank(dev, pipe); + + dpll = I915_READ(dpll_reg); + if (dpll & DISPLAY_RATE_SELECT_FPA1) + DRM_DEBUG_DRIVER("failed to upclock LVDS!\n"); + + /* ...and lock them again */ + I915_WRITE(PP_CONTROL, I915_READ(PP_CONTROL) & 0x3); + } + + /* Schedule downclock */ + mod_timer(&intel_crtc->idle_timer, jiffies + + msecs_to_jiffies(CRTC_IDLE_TIMEOUT)); +} + +static void intel_decrease_pllclock(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int pipe = intel_crtc->pipe; + int dpll_reg = DPLL(pipe); + int dpll = I915_READ(dpll_reg); + + if (HAS_PCH_SPLIT(dev)) + return; + + if (!dev_priv->lvds_downclock_avail) + return; + + /* + * Since this is called by a timer, we should never get here in + * the manual case. + */ + if (!HAS_PIPE_CXSR(dev) && intel_crtc->lowfreq_avail) { + DRM_DEBUG_DRIVER("downclocking LVDS\n"); + + /* Unlock panel regs */ + I915_WRITE(PP_CONTROL, I915_READ(PP_CONTROL) | + PANEL_UNLOCK_REGS); + + dpll |= DISPLAY_RATE_SELECT_FPA1; + I915_WRITE(dpll_reg, dpll); + intel_wait_for_vblank(dev, pipe); + dpll = I915_READ(dpll_reg); + if (!(dpll & DISPLAY_RATE_SELECT_FPA1)) + DRM_DEBUG_DRIVER("failed to downclock LVDS!\n"); + + /* ...and lock them again */ + I915_WRITE(PP_CONTROL, I915_READ(PP_CONTROL) & 0x3); + } + +} + +/** + * intel_idle_update - adjust clocks for idleness + * @work: work struct + * + * Either the GPU or display (or both) went idle. Check the busy status + * here and adjust the CRTC and GPU clocks as necessary. + */ +static void intel_idle_update(struct work_struct *work) +{ + drm_i915_private_t *dev_priv = container_of(work, drm_i915_private_t, + idle_work); + struct drm_device *dev = dev_priv->dev; + struct drm_crtc *crtc; + struct intel_crtc *intel_crtc; + + if (!i915_powersave) + return; + + mutex_lock(&dev->struct_mutex); + + i915_update_gfx_val(dev_priv); + + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { + /* Skip inactive CRTCs */ + if (!crtc->fb) + continue; + + intel_crtc = to_intel_crtc(crtc); + if (!intel_crtc->busy) + intel_decrease_pllclock(crtc); + } + + + mutex_unlock(&dev->struct_mutex); +} + +/** + * intel_mark_busy - mark the GPU and possibly the display busy + * @dev: drm device + * @obj: object we're operating on + * + * Callers can use this function to indicate that the GPU is busy processing + * commands. If @obj matches one of the CRTC objects (i.e. it's a scanout + * buffer), we'll also mark the display as busy, so we know to increase its + * clock frequency. + */ +void intel_mark_busy(struct drm_device *dev, struct drm_i915_gem_object *obj) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_crtc *crtc = NULL; + struct intel_framebuffer *intel_fb; + struct intel_crtc *intel_crtc; + + if (!drm_core_check_feature(dev, DRIVER_MODESET)) + return; + + if (!dev_priv->busy) + dev_priv->busy = true; + else + mod_timer(&dev_priv->idle_timer, jiffies + + msecs_to_jiffies(GPU_IDLE_TIMEOUT)); + + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { + if (!crtc->fb) + continue; + + intel_crtc = to_intel_crtc(crtc); + intel_fb = to_intel_framebuffer(crtc->fb); + if (intel_fb->obj == obj) { + if (!intel_crtc->busy) { + /* Non-busy -> busy, upclock */ + intel_increase_pllclock(crtc); + intel_crtc->busy = true; + } else { + /* Busy -> busy, put off timer */ + mod_timer(&intel_crtc->idle_timer, jiffies + + msecs_to_jiffies(CRTC_IDLE_TIMEOUT)); + } + } + } +} + +static void intel_crtc_destroy(struct drm_crtc *crtc) +{ + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + struct drm_device *dev = crtc->dev; + struct intel_unpin_work *work; + unsigned long flags; + + spin_lock_irqsave(&dev->event_lock, flags); + work = intel_crtc->unpin_work; + intel_crtc->unpin_work = NULL; + spin_unlock_irqrestore(&dev->event_lock, flags); + + if (work) { + cancel_work_sync(&work->work); + kfree(work); + } + + drm_crtc_cleanup(crtc); + + kfree(intel_crtc); +} + +static void intel_unpin_work_fn(struct work_struct *__work) +{ + struct intel_unpin_work *work = + container_of(__work, struct intel_unpin_work, work); + + mutex_lock(&work->dev->struct_mutex); + i915_gem_object_unpin(work->old_fb_obj); + drm_gem_object_unreference(&work->pending_flip_obj->base); + drm_gem_object_unreference(&work->old_fb_obj->base); + + mutex_unlock(&work->dev->struct_mutex); + kfree(work); +} + +static void do_intel_finish_page_flip(struct drm_device *dev, + struct drm_crtc *crtc) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + struct intel_unpin_work *work; + struct drm_i915_gem_object *obj; + struct drm_pending_vblank_event *e; + struct timeval tnow, tvbl; + unsigned long flags; + + /* Ignore early vblank irqs */ + if (intel_crtc == NULL) + return; + + do_gettimeofday(&tnow); + + spin_lock_irqsave(&dev->event_lock, flags); + work = intel_crtc->unpin_work; + if (work == NULL || !work->pending) { + spin_unlock_irqrestore(&dev->event_lock, flags); + return; + } + + intel_crtc->unpin_work = NULL; + + if (work->event) { + e = work->event; + e->event.sequence = drm_vblank_count_and_time(dev, intel_crtc->pipe, &tvbl); + + /* Called before vblank count and timestamps have + * been updated for the vblank interval of flip + * completion? Need to increment vblank count and + * add one videorefresh duration to returned timestamp + * to account for this. We assume this happened if we + * get called over 0.9 frame durations after the last + * timestamped vblank. + * + * This calculation can not be used with vrefresh rates + * below 5Hz (10Hz to be on the safe side) without + * promoting to 64 integers. + */ + if (10 * (timeval_to_ns(&tnow) - timeval_to_ns(&tvbl)) > + 9 * crtc->framedur_ns) { + e->event.sequence++; + tvbl = ns_to_timeval(timeval_to_ns(&tvbl) + + crtc->framedur_ns); + } + + e->event.tv_sec = tvbl.tv_sec; + e->event.tv_usec = tvbl.tv_usec; + + list_add_tail(&e->base.link, + &e->base.file_priv->event_list); + wake_up_interruptible(&e->base.file_priv->event_wait); + } + + drm_vblank_put(dev, intel_crtc->pipe); + + spin_unlock_irqrestore(&dev->event_lock, flags); + + obj = work->old_fb_obj; + + atomic_clear_mask(1 << intel_crtc->plane, + &obj->pending_flip.counter); + if (atomic_read(&obj->pending_flip) == 0) + wake_up(&dev_priv->pending_flip_queue); + + schedule_work(&work->work); + + trace_i915_flip_complete(intel_crtc->plane, work->pending_flip_obj); +} + +void intel_finish_page_flip(struct drm_device *dev, int pipe) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_crtc *crtc = dev_priv->pipe_to_crtc_mapping[pipe]; + + do_intel_finish_page_flip(dev, crtc); +} + +void intel_finish_page_flip_plane(struct drm_device *dev, int plane) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_crtc *crtc = dev_priv->plane_to_crtc_mapping[plane]; + + do_intel_finish_page_flip(dev, crtc); +} + +void intel_prepare_page_flip(struct drm_device *dev, int plane) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = + to_intel_crtc(dev_priv->plane_to_crtc_mapping[plane]); + unsigned long flags; + + spin_lock_irqsave(&dev->event_lock, flags); + if (intel_crtc->unpin_work) { + if ((++intel_crtc->unpin_work->pending) > 1) + DRM_ERROR("Prepared flip multiple times\n"); + } else { + DRM_DEBUG_DRIVER("preparing flip with no unpin work?\n"); + } + spin_unlock_irqrestore(&dev->event_lock, flags); +} + +static int intel_gen2_queue_flip(struct drm_device *dev, + struct drm_crtc *crtc, + struct drm_framebuffer *fb, + struct drm_i915_gem_object *obj) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + unsigned long offset; + u32 flip_mask; + int ret; + + ret = intel_pin_and_fence_fb_obj(dev, obj, LP_RING(dev_priv)); + if (ret) + goto out; + + /* Offset into the new buffer for cases of shared fbs between CRTCs */ + offset = crtc->y * fb->pitch + crtc->x * fb->bits_per_pixel/8; + + ret = BEGIN_LP_RING(6); + if (ret) + goto out; + + /* Can't queue multiple flips, so wait for the previous + * one to finish before executing the next. + */ + if (intel_crtc->plane) + flip_mask = MI_WAIT_FOR_PLANE_B_FLIP; + else + flip_mask = MI_WAIT_FOR_PLANE_A_FLIP; + OUT_RING(MI_WAIT_FOR_EVENT | flip_mask); + OUT_RING(MI_NOOP); + OUT_RING(MI_DISPLAY_FLIP | + MI_DISPLAY_FLIP_PLANE(intel_crtc->plane)); + OUT_RING(fb->pitch); + OUT_RING(obj->gtt_offset + offset); + OUT_RING(MI_NOOP); + ADVANCE_LP_RING(); +out: + return ret; +} + +static int intel_gen3_queue_flip(struct drm_device *dev, + struct drm_crtc *crtc, + struct drm_framebuffer *fb, + struct drm_i915_gem_object *obj) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + unsigned long offset; + u32 flip_mask; + int ret; + + ret = intel_pin_and_fence_fb_obj(dev, obj, LP_RING(dev_priv)); + if (ret) + goto out; + + /* Offset into the new buffer for cases of shared fbs between CRTCs */ + offset = crtc->y * fb->pitch + crtc->x * fb->bits_per_pixel/8; + + ret = BEGIN_LP_RING(6); + if (ret) + goto out; + + if (intel_crtc->plane) + flip_mask = MI_WAIT_FOR_PLANE_B_FLIP; + else + flip_mask = MI_WAIT_FOR_PLANE_A_FLIP; + OUT_RING(MI_WAIT_FOR_EVENT | flip_mask); + OUT_RING(MI_NOOP); + OUT_RING(MI_DISPLAY_FLIP_I915 | + MI_DISPLAY_FLIP_PLANE(intel_crtc->plane)); + OUT_RING(fb->pitch); + OUT_RING(obj->gtt_offset + offset); + OUT_RING(MI_NOOP); + + ADVANCE_LP_RING(); +out: + return ret; +} + +static int intel_gen4_queue_flip(struct drm_device *dev, + struct drm_crtc *crtc, + struct drm_framebuffer *fb, + struct drm_i915_gem_object *obj) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + uint32_t pf, pipesrc; + int ret; + + ret = intel_pin_and_fence_fb_obj(dev, obj, LP_RING(dev_priv)); + if (ret) + goto out; + + ret = BEGIN_LP_RING(4); + if (ret) + goto out; + + /* i965+ uses the linear or tiled offsets from the + * Display Registers (which do not change across a page-flip) + * so we need only reprogram the base address. + */ + OUT_RING(MI_DISPLAY_FLIP | + MI_DISPLAY_FLIP_PLANE(intel_crtc->plane)); + OUT_RING(fb->pitch); + OUT_RING(obj->gtt_offset | obj->tiling_mode); + + /* XXX Enabling the panel-fitter across page-flip is so far + * untested on non-native modes, so ignore it for now. + * pf = I915_READ(pipe == 0 ? PFA_CTL_1 : PFB_CTL_1) & PF_ENABLE; + */ + pf = 0; + pipesrc = I915_READ(PIPESRC(intel_crtc->pipe)) & 0x0fff0fff; + OUT_RING(pf | pipesrc); + ADVANCE_LP_RING(); +out: + return ret; +} + +static int intel_gen6_queue_flip(struct drm_device *dev, + struct drm_crtc *crtc, + struct drm_framebuffer *fb, + struct drm_i915_gem_object *obj) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + uint32_t pf, pipesrc; + int ret; + + ret = intel_pin_and_fence_fb_obj(dev, obj, LP_RING(dev_priv)); + if (ret) + goto out; + + ret = BEGIN_LP_RING(4); + if (ret) + goto out; + + OUT_RING(MI_DISPLAY_FLIP | + MI_DISPLAY_FLIP_PLANE(intel_crtc->plane)); + OUT_RING(fb->pitch | obj->tiling_mode); + OUT_RING(obj->gtt_offset); + + pf = I915_READ(PF_CTL(intel_crtc->pipe)) & PF_ENABLE; + pipesrc = I915_READ(PIPESRC(intel_crtc->pipe)) & 0x0fff0fff; + OUT_RING(pf | pipesrc); + ADVANCE_LP_RING(); +out: + return ret; +} + +/* + * On gen7 we currently use the blit ring because (in early silicon at least) + * the render ring doesn't give us interrpts for page flip completion, which + * means clients will hang after the first flip is queued. Fortunately the + * blit ring generates interrupts properly, so use it instead. + */ +static int intel_gen7_queue_flip(struct drm_device *dev, + struct drm_crtc *crtc, + struct drm_framebuffer *fb, + struct drm_i915_gem_object *obj) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + struct intel_ring_buffer *ring = &dev_priv->ring[BCS]; + int ret; + + ret = intel_pin_and_fence_fb_obj(dev, obj, ring); + if (ret) + goto out; + + ret = intel_ring_begin(ring, 4); + if (ret) + goto out; + + intel_ring_emit(ring, MI_DISPLAY_FLIP_I915 | (intel_crtc->plane << 19)); + intel_ring_emit(ring, (fb->pitch | obj->tiling_mode)); + intel_ring_emit(ring, (obj->gtt_offset)); + intel_ring_emit(ring, (MI_NOOP)); + intel_ring_advance(ring); +out: + return ret; +} + +static int intel_default_queue_flip(struct drm_device *dev, + struct drm_crtc *crtc, + struct drm_framebuffer *fb, + struct drm_i915_gem_object *obj) +{ + return -ENODEV; +} + +static int intel_crtc_page_flip(struct drm_crtc *crtc, + struct drm_framebuffer *fb, + struct drm_pending_vblank_event *event) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_framebuffer *intel_fb; + struct drm_i915_gem_object *obj; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + struct intel_unpin_work *work; + unsigned long flags; + int ret; + + work = kzalloc(sizeof *work, GFP_KERNEL); + if (work == NULL) + return -ENOMEM; + + work->event = event; + work->dev = crtc->dev; + intel_fb = to_intel_framebuffer(crtc->fb); + work->old_fb_obj = intel_fb->obj; + INIT_WORK(&work->work, intel_unpin_work_fn); + + /* We borrow the event spin lock for protecting unpin_work */ + spin_lock_irqsave(&dev->event_lock, flags); + if (intel_crtc->unpin_work) { + spin_unlock_irqrestore(&dev->event_lock, flags); + kfree(work); + + DRM_DEBUG_DRIVER("flip queue: crtc already busy\n"); + return -EBUSY; + } + intel_crtc->unpin_work = work; + spin_unlock_irqrestore(&dev->event_lock, flags); + + intel_fb = to_intel_framebuffer(fb); + obj = intel_fb->obj; + + mutex_lock(&dev->struct_mutex); + + /* Reference the objects for the scheduled work. */ + drm_gem_object_reference(&work->old_fb_obj->base); + drm_gem_object_reference(&obj->base); + + crtc->fb = fb; + + ret = drm_vblank_get(dev, intel_crtc->pipe); + if (ret) + goto cleanup_objs; + + work->pending_flip_obj = obj; + + work->enable_stall_check = true; + + /* Block clients from rendering to the new back buffer until + * the flip occurs and the object is no longer visible. + */ + atomic_add(1 << intel_crtc->plane, &work->old_fb_obj->pending_flip); + + ret = dev_priv->display.queue_flip(dev, crtc, fb, obj); + if (ret) + goto cleanup_pending; + + mutex_unlock(&dev->struct_mutex); + + trace_i915_flip_request(intel_crtc->plane, obj); + + return 0; + +cleanup_pending: + atomic_sub(1 << intel_crtc->plane, &work->old_fb_obj->pending_flip); +cleanup_objs: + drm_gem_object_unreference(&work->old_fb_obj->base); + drm_gem_object_unreference(&obj->base); + mutex_unlock(&dev->struct_mutex); + + spin_lock_irqsave(&dev->event_lock, flags); + intel_crtc->unpin_work = NULL; + spin_unlock_irqrestore(&dev->event_lock, flags); + + kfree(work); + + return ret; +} + +static void intel_sanitize_modesetting(struct drm_device *dev, + int pipe, int plane) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u32 reg, val; + int i; + + /* Clear any frame start delays used for debugging left by the BIOS */ + for_each_pipe(i) { + reg = PIPECONF(i); + I915_WRITE(reg, I915_READ(reg) & ~PIPECONF_FRAME_START_DELAY_MASK); + } + + if (HAS_PCH_SPLIT(dev)) + return; + + /* Who knows what state these registers were left in by the BIOS or + * grub? + * + * If we leave the registers in a conflicting state (e.g. with the + * display plane reading from the other pipe than the one we intend + * to use) then when we attempt to teardown the active mode, we will + * not disable the pipes and planes in the correct order -- leaving + * a plane reading from a disabled pipe and possibly leading to + * undefined behaviour. + */ + + reg = DSPCNTR(plane); + val = I915_READ(reg); + + if ((val & DISPLAY_PLANE_ENABLE) == 0) + return; + if (!!(val & DISPPLANE_SEL_PIPE_MASK) == pipe) + return; + + /* This display plane is active and attached to the other CPU pipe. */ + pipe = !pipe; + + /* Disable the plane and wait for it to stop reading from the pipe. */ + intel_disable_plane(dev_priv, plane, pipe); + intel_disable_pipe(dev_priv, pipe); +} + +static void intel_crtc_reset(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + + /* Reset flags back to the 'unknown' status so that they + * will be correctly set on the initial modeset. + */ + intel_crtc->dpms_mode = -1; + + /* We need to fix up any BIOS configuration that conflicts with + * our expectations. + */ + intel_sanitize_modesetting(dev, intel_crtc->pipe, intel_crtc->plane); +} + +static struct drm_crtc_helper_funcs intel_helper_funcs = { + .dpms = intel_crtc_dpms, + .mode_fixup = intel_crtc_mode_fixup, + .mode_set = intel_crtc_mode_set, + .mode_set_base = intel_pipe_set_base, + .mode_set_base_atomic = intel_pipe_set_base_atomic, + .load_lut = intel_crtc_load_lut, + .disable = intel_crtc_disable, +}; + +static const struct drm_crtc_funcs intel_crtc_funcs = { + .reset = intel_crtc_reset, + .cursor_set = intel_crtc_cursor_set, + .cursor_move = intel_crtc_cursor_move, + .gamma_set = intel_crtc_gamma_set, + .set_config = drm_crtc_helper_set_config, + .destroy = intel_crtc_destroy, + .page_flip = intel_crtc_page_flip, +}; + +static void intel_crtc_init(struct drm_device *dev, int pipe) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc; + int i; + + intel_crtc = kzalloc(sizeof(struct intel_crtc) + (INTELFB_CONN_LIMIT * sizeof(struct drm_connector *)), GFP_KERNEL); + if (intel_crtc == NULL) + return; + + drm_crtc_init(dev, &intel_crtc->base, &intel_crtc_funcs); + + drm_mode_crtc_set_gamma_size(&intel_crtc->base, 256); + for (i = 0; i < 256; i++) { + intel_crtc->lut_r[i] = i; + intel_crtc->lut_g[i] = i; + intel_crtc->lut_b[i] = i; + } + + /* Swap pipes & planes for FBC on pre-965 */ + intel_crtc->pipe = pipe; + intel_crtc->plane = pipe; + if (IS_MOBILE(dev) && IS_GEN3(dev)) { + DRM_DEBUG_KMS("swapping pipes & planes for FBC\n"); + intel_crtc->plane = !pipe; + } + + BUG_ON(pipe >= ARRAY_SIZE(dev_priv->plane_to_crtc_mapping) || + dev_priv->plane_to_crtc_mapping[intel_crtc->plane] != NULL); + dev_priv->plane_to_crtc_mapping[intel_crtc->plane] = &intel_crtc->base; + dev_priv->pipe_to_crtc_mapping[intel_crtc->pipe] = &intel_crtc->base; + + intel_crtc_reset(&intel_crtc->base); + intel_crtc->active = true; /* force the pipe off on setup_init_config */ + + if (HAS_PCH_SPLIT(dev)) { + intel_helper_funcs.prepare = ironlake_crtc_prepare; + intel_helper_funcs.commit = ironlake_crtc_commit; + } else { + intel_helper_funcs.prepare = i9xx_crtc_prepare; + intel_helper_funcs.commit = i9xx_crtc_commit; + } + + drm_crtc_helper_add(&intel_crtc->base, &intel_helper_funcs); + + intel_crtc->busy = false; + + setup_timer(&intel_crtc->idle_timer, intel_crtc_idle_timer, + (unsigned long)intel_crtc); +} + +int intel_get_pipe_from_crtc_id(struct drm_device *dev, void *data, + struct drm_file *file) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_i915_get_pipe_from_crtc_id *pipe_from_crtc_id = data; + struct drm_mode_object *drmmode_obj; + struct intel_crtc *crtc; + + if (!dev_priv) { + DRM_ERROR("called with no initialization\n"); + return -EINVAL; + } + + drmmode_obj = drm_mode_object_find(dev, pipe_from_crtc_id->crtc_id, + DRM_MODE_OBJECT_CRTC); + + if (!drmmode_obj) { + DRM_ERROR("no such CRTC id\n"); + return -EINVAL; + } + + crtc = to_intel_crtc(obj_to_crtc(drmmode_obj)); + pipe_from_crtc_id->pipe = crtc->pipe; + + return 0; +} + +static int intel_encoder_clones(struct drm_device *dev, int type_mask) +{ + struct intel_encoder *encoder; + int index_mask = 0; + int entry = 0; + + list_for_each_entry(encoder, &dev->mode_config.encoder_list, base.head) { + if (type_mask & encoder->clone_mask) + index_mask |= (1 << entry); + entry++; + } + + return index_mask; +} + +static bool has_edp_a(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + if (!IS_MOBILE(dev)) + return false; + + if ((I915_READ(DP_A) & DP_DETECTED) == 0) + return false; + + if (IS_GEN5(dev) && + (I915_READ(ILK_DISPLAY_CHICKEN_FUSES) & ILK_eDP_A_DISABLE)) + return false; + + return true; +} + +static void intel_setup_outputs(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_encoder *encoder; + bool dpd_is_edp = false; + bool has_lvds = false; + + if (IS_MOBILE(dev) && !IS_I830(dev)) + has_lvds = intel_lvds_init(dev); + if (!has_lvds && !HAS_PCH_SPLIT(dev)) { + /* disable the panel fitter on everything but LVDS */ + I915_WRITE(PFIT_CONTROL, 0); + } + + if (HAS_PCH_SPLIT(dev)) { + dpd_is_edp = intel_dpd_is_edp(dev); + + if (has_edp_a(dev)) + intel_dp_init(dev, DP_A); + + if (dpd_is_edp && (I915_READ(PCH_DP_D) & DP_DETECTED)) + intel_dp_init(dev, PCH_DP_D); + } + + intel_crt_init(dev); + + if (HAS_PCH_SPLIT(dev)) { + int found; + + if (I915_READ(HDMIB) & PORT_DETECTED) { + /* PCH SDVOB multiplex with HDMIB */ + found = intel_sdvo_init(dev, PCH_SDVOB); + if (!found) + intel_hdmi_init(dev, HDMIB); + if (!found && (I915_READ(PCH_DP_B) & DP_DETECTED)) + intel_dp_init(dev, PCH_DP_B); + } + + if (I915_READ(HDMIC) & PORT_DETECTED) + intel_hdmi_init(dev, HDMIC); + + if (I915_READ(HDMID) & PORT_DETECTED) + intel_hdmi_init(dev, HDMID); + + if (I915_READ(PCH_DP_C) & DP_DETECTED) + intel_dp_init(dev, PCH_DP_C); + + if (!dpd_is_edp && (I915_READ(PCH_DP_D) & DP_DETECTED)) + intel_dp_init(dev, PCH_DP_D); + + } else if (SUPPORTS_DIGITAL_OUTPUTS(dev)) { + bool found = false; + + if (I915_READ(SDVOB) & SDVO_DETECTED) { + DRM_DEBUG_KMS("probing SDVOB\n"); + found = intel_sdvo_init(dev, SDVOB); + if (!found && SUPPORTS_INTEGRATED_HDMI(dev)) { + DRM_DEBUG_KMS("probing HDMI on SDVOB\n"); + intel_hdmi_init(dev, SDVOB); + } + + if (!found && SUPPORTS_INTEGRATED_DP(dev)) { + DRM_DEBUG_KMS("probing DP_B\n"); + intel_dp_init(dev, DP_B); + } + } + + /* Before G4X SDVOC doesn't have its own detect register */ + + if (I915_READ(SDVOB) & SDVO_DETECTED) { + DRM_DEBUG_KMS("probing SDVOC\n"); + found = intel_sdvo_init(dev, SDVOC); + } + + if (!found && (I915_READ(SDVOC) & SDVO_DETECTED)) { + + if (SUPPORTS_INTEGRATED_HDMI(dev)) { + DRM_DEBUG_KMS("probing HDMI on SDVOC\n"); + intel_hdmi_init(dev, SDVOC); + } + if (SUPPORTS_INTEGRATED_DP(dev)) { + DRM_DEBUG_KMS("probing DP_C\n"); + intel_dp_init(dev, DP_C); + } + } + + if (SUPPORTS_INTEGRATED_DP(dev) && + (I915_READ(DP_D) & DP_DETECTED)) { + DRM_DEBUG_KMS("probing DP_D\n"); + intel_dp_init(dev, DP_D); + } + } else if (IS_GEN2(dev)) + intel_dvo_init(dev); + + if (SUPPORTS_TV(dev)) + intel_tv_init(dev); + + list_for_each_entry(encoder, &dev->mode_config.encoder_list, base.head) { + encoder->base.possible_crtcs = encoder->crtc_mask; + encoder->base.possible_clones = + intel_encoder_clones(dev, encoder->clone_mask); + } + + intel_panel_setup_backlight(dev); + + /* disable all the possible outputs/crtcs before entering KMS mode */ + drm_helper_disable_unused_functions(dev); +} + +static void intel_user_framebuffer_destroy(struct drm_framebuffer *fb) +{ + struct intel_framebuffer *intel_fb = to_intel_framebuffer(fb); + + drm_framebuffer_cleanup(fb); + drm_gem_object_unreference_unlocked(&intel_fb->obj->base); + + kfree(intel_fb); +} + +static int intel_user_framebuffer_create_handle(struct drm_framebuffer *fb, + struct drm_file *file, + unsigned int *handle) +{ + struct intel_framebuffer *intel_fb = to_intel_framebuffer(fb); + struct drm_i915_gem_object *obj = intel_fb->obj; + + return drm_gem_handle_create(file, &obj->base, handle); +} + +static const struct drm_framebuffer_funcs intel_fb_funcs = { + .destroy = intel_user_framebuffer_destroy, + .create_handle = intel_user_framebuffer_create_handle, +}; + +int intel_framebuffer_init(struct drm_device *dev, + struct intel_framebuffer *intel_fb, + struct drm_mode_fb_cmd *mode_cmd, + struct drm_i915_gem_object *obj) +{ + int ret; + + if (obj->tiling_mode == I915_TILING_Y) + return -EINVAL; + + if (mode_cmd->pitch & 63) + return -EINVAL; + + switch (mode_cmd->bpp) { + case 8: + case 16: + case 24: + case 32: + break; + default: + return -EINVAL; + } + + ret = drm_framebuffer_init(dev, &intel_fb->base, &intel_fb_funcs); + if (ret) { + DRM_ERROR("framebuffer init failed %d\n", ret); + return ret; + } + + drm_helper_mode_fill_fb_struct(&intel_fb->base, mode_cmd); + intel_fb->obj = obj; + return 0; +} + +static struct drm_framebuffer * +intel_user_framebuffer_create(struct drm_device *dev, + struct drm_file *filp, + struct drm_mode_fb_cmd *mode_cmd) +{ + struct drm_i915_gem_object *obj; + + obj = to_intel_bo(drm_gem_object_lookup(dev, filp, mode_cmd->handle)); + if (&obj->base == NULL) + return ERR_PTR(-ENOENT); + + return intel_framebuffer_create(dev, mode_cmd, obj); +} + +static const struct drm_mode_config_funcs intel_mode_funcs = { + .fb_create = intel_user_framebuffer_create, + .output_poll_changed = intel_fb_output_poll_changed, +}; + +static struct drm_i915_gem_object * +intel_alloc_context_page(struct drm_device *dev) +{ + struct drm_i915_gem_object *ctx; + int ret; + + WARN_ON(!mutex_is_locked(&dev->struct_mutex)); + + ctx = i915_gem_alloc_object(dev, 4096); + if (!ctx) { + DRM_DEBUG("failed to alloc power context, RC6 disabled\n"); + return NULL; + } + + ret = i915_gem_object_pin(ctx, 4096, true); + if (ret) { + DRM_ERROR("failed to pin power context: %d\n", ret); + goto err_unref; + } + + ret = i915_gem_object_set_to_gtt_domain(ctx, 1); + if (ret) { + DRM_ERROR("failed to set-domain on power context: %d\n", ret); + goto err_unpin; + } + + return ctx; + +err_unpin: + i915_gem_object_unpin(ctx); +err_unref: + drm_gem_object_unreference(&ctx->base); + mutex_unlock(&dev->struct_mutex); + return NULL; +} + +bool ironlake_set_drps(struct drm_device *dev, u8 val) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u16 rgvswctl; + + rgvswctl = I915_READ16(MEMSWCTL); + if (rgvswctl & MEMCTL_CMD_STS) { + DRM_DEBUG("gpu busy, RCS change rejected\n"); + return false; /* still busy with another command */ + } + + rgvswctl = (MEMCTL_CMD_CHFREQ << MEMCTL_CMD_SHIFT) | + (val << MEMCTL_FREQ_SHIFT) | MEMCTL_SFCAVM; + I915_WRITE16(MEMSWCTL, rgvswctl); + POSTING_READ16(MEMSWCTL); + + rgvswctl |= MEMCTL_CMD_STS; + I915_WRITE16(MEMSWCTL, rgvswctl); + + return true; +} + +void ironlake_enable_drps(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u32 rgvmodectl = I915_READ(MEMMODECTL); + u8 fmax, fmin, fstart, vstart; + + /* Enable temp reporting */ + I915_WRITE16(PMMISC, I915_READ(PMMISC) | MCPPCE_EN); + I915_WRITE16(TSC1, I915_READ(TSC1) | TSE); + + /* 100ms RC evaluation intervals */ + I915_WRITE(RCUPEI, 100000); + I915_WRITE(RCDNEI, 100000); + + /* Set max/min thresholds to 90ms and 80ms respectively */ + I915_WRITE(RCBMAXAVG, 90000); + I915_WRITE(RCBMINAVG, 80000); + + I915_WRITE(MEMIHYST, 1); + + /* Set up min, max, and cur for interrupt handling */ + fmax = (rgvmodectl & MEMMODE_FMAX_MASK) >> MEMMODE_FMAX_SHIFT; + fmin = (rgvmodectl & MEMMODE_FMIN_MASK); + fstart = (rgvmodectl & MEMMODE_FSTART_MASK) >> + MEMMODE_FSTART_SHIFT; + + vstart = (I915_READ(PXVFREQ_BASE + (fstart * 4)) & PXVFREQ_PX_MASK) >> + PXVFREQ_PX_SHIFT; + + dev_priv->fmax = fmax; /* IPS callback will increase this */ + dev_priv->fstart = fstart; + + dev_priv->max_delay = fstart; + dev_priv->min_delay = fmin; + dev_priv->cur_delay = fstart; + + DRM_DEBUG_DRIVER("fmax: %d, fmin: %d, fstart: %d\n", + fmax, fmin, fstart); + + I915_WRITE(MEMINTREN, MEMINT_CX_SUPR_EN | MEMINT_EVAL_CHG_EN); + + /* + * Interrupts will be enabled in ironlake_irq_postinstall + */ + + I915_WRITE(VIDSTART, vstart); + POSTING_READ(VIDSTART); + + rgvmodectl |= MEMMODE_SWMODE_EN; + I915_WRITE(MEMMODECTL, rgvmodectl); + + if (wait_for((I915_READ(MEMSWCTL) & MEMCTL_CMD_STS) == 0, 10)) + DRM_ERROR("stuck trying to change perf mode\n"); + msleep(1); + + ironlake_set_drps(dev, fstart); + + dev_priv->last_count1 = I915_READ(0x112e4) + I915_READ(0x112e8) + + I915_READ(0x112e0); + dev_priv->last_time1 = jiffies_to_msecs(jiffies); + dev_priv->last_count2 = I915_READ(0x112f4); + getrawmonotonic(&dev_priv->last_time2); +} + +void ironlake_disable_drps(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u16 rgvswctl = I915_READ16(MEMSWCTL); + + /* Ack interrupts, disable EFC interrupt */ + I915_WRITE(MEMINTREN, I915_READ(MEMINTREN) & ~MEMINT_EVAL_CHG_EN); + I915_WRITE(MEMINTRSTS, MEMINT_EVAL_CHG); + I915_WRITE(DEIER, I915_READ(DEIER) & ~DE_PCU_EVENT); + I915_WRITE(DEIIR, DE_PCU_EVENT); + I915_WRITE(DEIMR, I915_READ(DEIMR) | DE_PCU_EVENT); + + /* Go back to the starting frequency */ + ironlake_set_drps(dev, dev_priv->fstart); + msleep(1); + rgvswctl |= MEMCTL_CMD_STS; + I915_WRITE(MEMSWCTL, rgvswctl); + msleep(1); + +} + +void gen6_set_rps(struct drm_device *dev, u8 val) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u32 swreq; + + swreq = (val & 0x3ff) << 25; + I915_WRITE(GEN6_RPNSWREQ, swreq); +} + +void gen6_disable_rps(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + I915_WRITE(GEN6_RPNSWREQ, 1 << 31); + I915_WRITE(GEN6_PMINTRMSK, 0xffffffff); + I915_WRITE(GEN6_PMIER, 0); + + spin_lock_irq(&dev_priv->rps_lock); + dev_priv->pm_iir = 0; + spin_unlock_irq(&dev_priv->rps_lock); + + I915_WRITE(GEN6_PMIIR, I915_READ(GEN6_PMIIR)); +} + +static unsigned long intel_pxfreq(u32 vidfreq) +{ + unsigned long freq; + int div = (vidfreq & 0x3f0000) >> 16; + int post = (vidfreq & 0x3000) >> 12; + int pre = (vidfreq & 0x7); + + if (!pre) + return 0; + + freq = ((div * 133333) / ((1<dev_private; + u32 lcfuse; + u8 pxw[16]; + int i; + + /* Disable to program */ + I915_WRITE(ECR, 0); + POSTING_READ(ECR); + + /* Program energy weights for various events */ + I915_WRITE(SDEW, 0x15040d00); + I915_WRITE(CSIEW0, 0x007f0000); + I915_WRITE(CSIEW1, 0x1e220004); + I915_WRITE(CSIEW2, 0x04000004); + + for (i = 0; i < 5; i++) + I915_WRITE(PEW + (i * 4), 0); + for (i = 0; i < 3; i++) + I915_WRITE(DEW + (i * 4), 0); + + /* Program P-state weights to account for frequency power adjustment */ + for (i = 0; i < 16; i++) { + u32 pxvidfreq = I915_READ(PXVFREQ_BASE + (i * 4)); + unsigned long freq = intel_pxfreq(pxvidfreq); + unsigned long vid = (pxvidfreq & PXVFREQ_PX_MASK) >> + PXVFREQ_PX_SHIFT; + unsigned long val; + + val = vid * vid; + val *= (freq / 1000); + val *= 255; + val /= (127*127*900); + if (val > 0xff) + DRM_ERROR("bad pxval: %ld\n", val); + pxw[i] = val; + } + /* Render standby states get 0 weight */ + pxw[14] = 0; + pxw[15] = 0; + + for (i = 0; i < 4; i++) { + u32 val = (pxw[i*4] << 24) | (pxw[(i*4)+1] << 16) | + (pxw[(i*4)+2] << 8) | (pxw[(i*4)+3]); + I915_WRITE(PXW + (i * 4), val); + } + + /* Adjust magic regs to magic values (more experimental results) */ + I915_WRITE(OGW0, 0); + I915_WRITE(OGW1, 0); + I915_WRITE(EG0, 0x00007f00); + I915_WRITE(EG1, 0x0000000e); + I915_WRITE(EG2, 0x000e0000); + I915_WRITE(EG3, 0x68000300); + I915_WRITE(EG4, 0x42000000); + I915_WRITE(EG5, 0x00140031); + I915_WRITE(EG6, 0); + I915_WRITE(EG7, 0); + + for (i = 0; i < 8; i++) + I915_WRITE(PXWL + (i * 4), 0); + + /* Enable PMON + select events */ + I915_WRITE(ECR, 0x80000019); + + lcfuse = I915_READ(LCFUSE02); + + dev_priv->corr = (lcfuse & LCFUSE_HIV_MASK); +} + +void gen6_enable_rps(struct drm_i915_private *dev_priv) +{ + u32 rp_state_cap = I915_READ(GEN6_RP_STATE_CAP); + u32 gt_perf_status = I915_READ(GEN6_GT_PERF_STATUS); + u32 pcu_mbox, rc6_mask = 0; + int cur_freq, min_freq, max_freq; + int i; + + /* Here begins a magic sequence of register writes to enable + * auto-downclocking. + * + * Perhaps there might be some value in exposing these to + * userspace... + */ + I915_WRITE(GEN6_RC_STATE, 0); + mutex_lock(&dev_priv->dev->struct_mutex); + gen6_gt_force_wake_get(dev_priv); + + /* disable the counters and set deterministic thresholds */ + I915_WRITE(GEN6_RC_CONTROL, 0); + + I915_WRITE(GEN6_RC1_WAKE_RATE_LIMIT, 1000 << 16); + I915_WRITE(GEN6_RC6_WAKE_RATE_LIMIT, 40 << 16 | 30); + I915_WRITE(GEN6_RC6pp_WAKE_RATE_LIMIT, 30); + I915_WRITE(GEN6_RC_EVALUATION_INTERVAL, 125000); + I915_WRITE(GEN6_RC_IDLE_HYSTERSIS, 25); + + for (i = 0; i < I915_NUM_RINGS; i++) + I915_WRITE(RING_MAX_IDLE(dev_priv->ring[i].mmio_base), 10); + + I915_WRITE(GEN6_RC_SLEEP, 0); + I915_WRITE(GEN6_RC1e_THRESHOLD, 1000); + I915_WRITE(GEN6_RC6_THRESHOLD, 50000); + I915_WRITE(GEN6_RC6p_THRESHOLD, 100000); + I915_WRITE(GEN6_RC6pp_THRESHOLD, 64000); /* unused */ + + if (i915_enable_rc6) + rc6_mask = GEN6_RC_CTL_RC6p_ENABLE | + GEN6_RC_CTL_RC6_ENABLE; + + I915_WRITE(GEN6_RC_CONTROL, + rc6_mask | + GEN6_RC_CTL_EI_MODE(1) | + GEN6_RC_CTL_HW_ENABLE); + + I915_WRITE(GEN6_RPNSWREQ, + GEN6_FREQUENCY(10) | + GEN6_OFFSET(0) | + GEN6_AGGRESSIVE_TURBO); + I915_WRITE(GEN6_RC_VIDEO_FREQ, + GEN6_FREQUENCY(12)); + + I915_WRITE(GEN6_RP_DOWN_TIMEOUT, 1000000); + I915_WRITE(GEN6_RP_INTERRUPT_LIMITS, + 18 << 24 | + 6 << 16); + I915_WRITE(GEN6_RP_UP_THRESHOLD, 10000); + I915_WRITE(GEN6_RP_DOWN_THRESHOLD, 1000000); + I915_WRITE(GEN6_RP_UP_EI, 100000); + I915_WRITE(GEN6_RP_DOWN_EI, 5000000); + I915_WRITE(GEN6_RP_IDLE_HYSTERSIS, 10); + I915_WRITE(GEN6_RP_CONTROL, + GEN6_RP_MEDIA_TURBO | + GEN6_RP_USE_NORMAL_FREQ | + GEN6_RP_MEDIA_IS_GFX | + GEN6_RP_ENABLE | + GEN6_RP_UP_BUSY_AVG | + GEN6_RP_DOWN_IDLE_CONT); + + if (wait_for((I915_READ(GEN6_PCODE_MAILBOX) & GEN6_PCODE_READY) == 0, + 500)) + DRM_ERROR("timeout waiting for pcode mailbox to become idle\n"); + + I915_WRITE(GEN6_PCODE_DATA, 0); + I915_WRITE(GEN6_PCODE_MAILBOX, + GEN6_PCODE_READY | + GEN6_PCODE_WRITE_MIN_FREQ_TABLE); + if (wait_for((I915_READ(GEN6_PCODE_MAILBOX) & GEN6_PCODE_READY) == 0, + 500)) + DRM_ERROR("timeout waiting for pcode mailbox to finish\n"); + + min_freq = (rp_state_cap & 0xff0000) >> 16; + max_freq = rp_state_cap & 0xff; + cur_freq = (gt_perf_status & 0xff00) >> 8; + + /* Check for overclock support */ + if (wait_for((I915_READ(GEN6_PCODE_MAILBOX) & GEN6_PCODE_READY) == 0, + 500)) + DRM_ERROR("timeout waiting for pcode mailbox to become idle\n"); + I915_WRITE(GEN6_PCODE_MAILBOX, GEN6_READ_OC_PARAMS); + pcu_mbox = I915_READ(GEN6_PCODE_DATA); + if (wait_for((I915_READ(GEN6_PCODE_MAILBOX) & GEN6_PCODE_READY) == 0, + 500)) + DRM_ERROR("timeout waiting for pcode mailbox to finish\n"); + if (pcu_mbox & (1<<31)) { /* OC supported */ + max_freq = pcu_mbox & 0xff; + DRM_DEBUG_DRIVER("overclocking supported, adjusting frequency max to %dMHz\n", pcu_mbox * 50); + } + + /* In units of 100MHz */ + dev_priv->max_delay = max_freq; + dev_priv->min_delay = min_freq; + dev_priv->cur_delay = cur_freq; + + /* requires MSI enabled */ + I915_WRITE(GEN6_PMIER, + GEN6_PM_MBOX_EVENT | + GEN6_PM_THERMAL_EVENT | + GEN6_PM_RP_DOWN_TIMEOUT | + GEN6_PM_RP_UP_THRESHOLD | + GEN6_PM_RP_DOWN_THRESHOLD | + GEN6_PM_RP_UP_EI_EXPIRED | + GEN6_PM_RP_DOWN_EI_EXPIRED); + spin_lock_irq(&dev_priv->rps_lock); + WARN_ON(dev_priv->pm_iir != 0); + I915_WRITE(GEN6_PMIMR, 0); + spin_unlock_irq(&dev_priv->rps_lock); + /* enable all PM interrupts */ + I915_WRITE(GEN6_PMINTRMSK, 0); + + gen6_gt_force_wake_put(dev_priv); + mutex_unlock(&dev_priv->dev->struct_mutex); +} + +static void ironlake_init_clock_gating(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + uint32_t dspclk_gate = VRHUNIT_CLOCK_GATE_DISABLE; + + /* Required for FBC */ + dspclk_gate |= DPFCUNIT_CLOCK_GATE_DISABLE | + DPFCRUNIT_CLOCK_GATE_DISABLE | + DPFDUNIT_CLOCK_GATE_DISABLE; + /* Required for CxSR */ + dspclk_gate |= DPARBUNIT_CLOCK_GATE_DISABLE; + + I915_WRITE(PCH_3DCGDIS0, + MARIUNIT_CLOCK_GATE_DISABLE | + SVSMUNIT_CLOCK_GATE_DISABLE); + I915_WRITE(PCH_3DCGDIS1, + VFMUNIT_CLOCK_GATE_DISABLE); + + I915_WRITE(PCH_DSPCLK_GATE_D, dspclk_gate); + + /* + * According to the spec the following bits should be set in + * order to enable memory self-refresh + * The bit 22/21 of 0x42004 + * The bit 5 of 0x42020 + * The bit 15 of 0x45000 + */ + I915_WRITE(ILK_DISPLAY_CHICKEN2, + (I915_READ(ILK_DISPLAY_CHICKEN2) | + ILK_DPARB_GATE | ILK_VSDPFD_FULL)); + I915_WRITE(ILK_DSPCLK_GATE, + (I915_READ(ILK_DSPCLK_GATE) | + ILK_DPARB_CLK_GATE)); + I915_WRITE(DISP_ARB_CTL, + (I915_READ(DISP_ARB_CTL) | + DISP_FBC_WM_DIS)); + I915_WRITE(WM3_LP_ILK, 0); + I915_WRITE(WM2_LP_ILK, 0); + I915_WRITE(WM1_LP_ILK, 0); + + /* + * Based on the document from hardware guys the following bits + * should be set unconditionally in order to enable FBC. + * The bit 22 of 0x42000 + * The bit 22 of 0x42004 + * The bit 7,8,9 of 0x42020. + */ + if (IS_IRONLAKE_M(dev)) { + I915_WRITE(ILK_DISPLAY_CHICKEN1, + I915_READ(ILK_DISPLAY_CHICKEN1) | + ILK_FBCQ_DIS); + I915_WRITE(ILK_DISPLAY_CHICKEN2, + I915_READ(ILK_DISPLAY_CHICKEN2) | + ILK_DPARB_GATE); + I915_WRITE(ILK_DSPCLK_GATE, + I915_READ(ILK_DSPCLK_GATE) | + ILK_DPFC_DIS1 | + ILK_DPFC_DIS2 | + ILK_CLK_FBC); + } + + I915_WRITE(ILK_DISPLAY_CHICKEN2, + I915_READ(ILK_DISPLAY_CHICKEN2) | + ILK_ELPIN_409_SELECT); + I915_WRITE(_3D_CHICKEN2, + _3D_CHICKEN2_WM_READ_PIPELINED << 16 | + _3D_CHICKEN2_WM_READ_PIPELINED); +} + +static void gen6_init_clock_gating(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + int pipe; + uint32_t dspclk_gate = VRHUNIT_CLOCK_GATE_DISABLE; + + I915_WRITE(PCH_DSPCLK_GATE_D, dspclk_gate); + + I915_WRITE(ILK_DISPLAY_CHICKEN2, + I915_READ(ILK_DISPLAY_CHICKEN2) | + ILK_ELPIN_409_SELECT); + + I915_WRITE(WM3_LP_ILK, 0); + I915_WRITE(WM2_LP_ILK, 0); + I915_WRITE(WM1_LP_ILK, 0); + + /* According to the BSpec vol1g, bit 12 (RCPBUNIT) clock + * gating disable must be set. Failure to set it results in + * flickering pixels due to Z write ordering failures after + * some amount of runtime in the Mesa "fire" demo, and Unigine + * Sanctuary and Tropics, and apparently anything else with + * alpha test or pixel discard. + * + * According to the spec, bit 11 (RCCUNIT) must also be set, + * but we didn't debug actual testcases to find it out. + */ + I915_WRITE(GEN6_UCGCTL2, + GEN6_RCPBUNIT_CLOCK_GATE_DISABLE | + GEN6_RCCUNIT_CLOCK_GATE_DISABLE); + + /* + * According to the spec the following bits should be + * set in order to enable memory self-refresh and fbc: + * The bit21 and bit22 of 0x42000 + * The bit21 and bit22 of 0x42004 + * The bit5 and bit7 of 0x42020 + * The bit14 of 0x70180 + * The bit14 of 0x71180 + */ + I915_WRITE(ILK_DISPLAY_CHICKEN1, + I915_READ(ILK_DISPLAY_CHICKEN1) | + ILK_FBCQ_DIS | ILK_PABSTRETCH_DIS); + I915_WRITE(ILK_DISPLAY_CHICKEN2, + I915_READ(ILK_DISPLAY_CHICKEN2) | + ILK_DPARB_GATE | ILK_VSDPFD_FULL); + I915_WRITE(ILK_DSPCLK_GATE, + I915_READ(ILK_DSPCLK_GATE) | + ILK_DPARB_CLK_GATE | + ILK_DPFD_CLK_GATE); + + for_each_pipe(pipe) + I915_WRITE(DSPCNTR(pipe), + I915_READ(DSPCNTR(pipe)) | + DISPPLANE_TRICKLE_FEED_DISABLE); +} + +static void gen7_setup_fixed_func_scheduler(struct drm_i915_private *dev_priv) +{ + uint32_t reg = I915_READ(GEN7_FF_THREAD_MODE); + + reg &= ~GEN7_FF_SCHED_MASK; + reg |= GEN7_FF_TS_SCHED_HW; + reg |= GEN7_FF_VS_SCHED_HW; + reg |= GEN7_FF_DS_SCHED_HW; + + I915_WRITE(GEN7_FF_THREAD_MODE, reg); +} + +static void ivybridge_init_clock_gating(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + int pipe; + uint32_t dspclk_gate = VRHUNIT_CLOCK_GATE_DISABLE; + + I915_WRITE(PCH_DSPCLK_GATE_D, dspclk_gate); + + I915_WRITE(WM3_LP_ILK, 0); + I915_WRITE(WM2_LP_ILK, 0); + I915_WRITE(WM1_LP_ILK, 0); + + /* According to the spec, bit 13 (RCZUNIT) must be set on IVB. + * This implements the WaDisableRCZUnitClockGating workaround. + */ + I915_WRITE(GEN6_UCGCTL2, GEN6_RCZUNIT_CLOCK_GATE_DISABLE); + + I915_WRITE(ILK_DSPCLK_GATE, IVB_VRHUNIT_CLK_GATE); + + /* Apply the WaDisableRHWOOptimizationForRenderHang workaround. */ + I915_WRITE(GEN7_COMMON_SLICE_CHICKEN1, + GEN7_CSC1_RHWO_OPT_DISABLE_IN_RCC); + + /* WaApplyL3ControlAndL3ChickenMode requires those two on Ivy Bridge */ + I915_WRITE(GEN7_L3CNTLREG1, + GEN7_WA_FOR_GEN7_L3_CONTROL); + I915_WRITE(GEN7_L3_CHICKEN_MODE_REGISTER, + GEN7_WA_L3_CHICKEN_MODE); + + /* This is required by WaCatErrorRejectionIssue */ + I915_WRITE(GEN7_SQ_CHICKEN_MBCUNIT_CONFIG, + I915_READ(GEN7_SQ_CHICKEN_MBCUNIT_CONFIG) | + GEN7_SQ_CHICKEN_MBCUNIT_SQINTMOB); + + for_each_pipe(pipe) + I915_WRITE(DSPCNTR(pipe), + I915_READ(DSPCNTR(pipe)) | + DISPPLANE_TRICKLE_FEED_DISABLE); +} + +static void g4x_init_clock_gating(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + uint32_t dspclk_gate; + + I915_WRITE(RENCLK_GATE_D1, 0); + I915_WRITE(RENCLK_GATE_D2, VF_UNIT_CLOCK_GATE_DISABLE | + GS_UNIT_CLOCK_GATE_DISABLE | + CL_UNIT_CLOCK_GATE_DISABLE); + I915_WRITE(RAMCLK_GATE_D, 0); + dspclk_gate = VRHUNIT_CLOCK_GATE_DISABLE | + OVRUNIT_CLOCK_GATE_DISABLE | + OVCUNIT_CLOCK_GATE_DISABLE; + if (IS_GM45(dev)) + dspclk_gate |= DSSUNIT_CLOCK_GATE_DISABLE; + I915_WRITE(DSPCLK_GATE_D, dspclk_gate); +} + +static void crestline_init_clock_gating(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + I915_WRITE(RENCLK_GATE_D1, I965_RCC_CLOCK_GATE_DISABLE); + I915_WRITE(RENCLK_GATE_D2, 0); + I915_WRITE(DSPCLK_GATE_D, 0); + I915_WRITE(RAMCLK_GATE_D, 0); + I915_WRITE16(DEUC, 0); +} + +static void broadwater_init_clock_gating(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + I915_WRITE(RENCLK_GATE_D1, I965_RCZ_CLOCK_GATE_DISABLE | + I965_RCC_CLOCK_GATE_DISABLE | + I965_RCPB_CLOCK_GATE_DISABLE | + I965_ISC_CLOCK_GATE_DISABLE | + I965_FBC_CLOCK_GATE_DISABLE); + I915_WRITE(RENCLK_GATE_D2, 0); +} + +static void gen3_init_clock_gating(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u32 dstate = I915_READ(D_STATE); + + dstate |= DSTATE_PLL_D3_OFF | DSTATE_GFX_CLOCK_GATING | + DSTATE_DOT_CLOCK_GATING; + I915_WRITE(D_STATE, dstate); +} + +static void i85x_init_clock_gating(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + I915_WRITE(RENCLK_GATE_D1, SV_CLOCK_GATE_DISABLE); +} + +static void i830_init_clock_gating(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + I915_WRITE(DSPCLK_GATE_D, OVRUNIT_CLOCK_GATE_DISABLE); +} + +static void ibx_init_clock_gating(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + /* + * On Ibex Peak and Cougar Point, we need to disable clock + * gating for the panel power sequencer or it will fail to + * start up when no ports are active. + */ + I915_WRITE(SOUTH_DSPCLK_GATE_D, PCH_DPLSUNIT_CLOCK_GATE_DISABLE); +} + +static void cpt_init_clock_gating(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + /* + * On Ibex Peak and Cougar Point, we need to disable clock + * gating for the panel power sequencer or it will fail to + * start up when no ports are active. + */ + I915_WRITE(SOUTH_DSPCLK_GATE_D, PCH_DPLSUNIT_CLOCK_GATE_DISABLE); + I915_WRITE(SOUTH_CHICKEN2, I915_READ(SOUTH_CHICKEN2) | + DPLS_EDP_PPS_FIX_DIS); +} + +static void ironlake_teardown_rc6(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + if (dev_priv->renderctx) { + i915_gem_object_unpin(dev_priv->renderctx); + drm_gem_object_unreference(&dev_priv->renderctx->base); + dev_priv->renderctx = NULL; + } + + if (dev_priv->pwrctx) { + i915_gem_object_unpin(dev_priv->pwrctx); + drm_gem_object_unreference(&dev_priv->pwrctx->base); + dev_priv->pwrctx = NULL; + } + + gen7_setup_fixed_func_scheduler(dev_priv); +} + +static void ironlake_disable_rc6(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + if (I915_READ(PWRCTXA)) { + /* Wake the GPU, prevent RC6, then restore RSTDBYCTL */ + I915_WRITE(RSTDBYCTL, I915_READ(RSTDBYCTL) | RCX_SW_EXIT); + wait_for(((I915_READ(RSTDBYCTL) & RSX_STATUS_MASK) == RSX_STATUS_ON), + 50); + + I915_WRITE(PWRCTXA, 0); + POSTING_READ(PWRCTXA); + + I915_WRITE(RSTDBYCTL, I915_READ(RSTDBYCTL) & ~RCX_SW_EXIT); + POSTING_READ(RSTDBYCTL); + } + + ironlake_teardown_rc6(dev); +} + +static int ironlake_setup_rc6(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + if (dev_priv->renderctx == NULL) + dev_priv->renderctx = intel_alloc_context_page(dev); + if (!dev_priv->renderctx) + return -ENOMEM; + + if (dev_priv->pwrctx == NULL) + dev_priv->pwrctx = intel_alloc_context_page(dev); + if (!dev_priv->pwrctx) { + ironlake_teardown_rc6(dev); + return -ENOMEM; + } + + return 0; +} + +void ironlake_enable_rc6(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + int ret; + + /* rc6 disabled by default due to repeated reports of hanging during + * boot and resume. + */ + if (!i915_enable_rc6) + return; + + mutex_lock(&dev->struct_mutex); + ret = ironlake_setup_rc6(dev); + if (ret) { + mutex_unlock(&dev->struct_mutex); + return; + } + + /* + * GPU can automatically power down the render unit if given a page + * to save state. + */ + ret = BEGIN_LP_RING(6); + if (ret) { + ironlake_teardown_rc6(dev); + mutex_unlock(&dev->struct_mutex); + return; + } + + OUT_RING(MI_SUSPEND_FLUSH | MI_SUSPEND_FLUSH_EN); + OUT_RING(MI_SET_CONTEXT); + OUT_RING(dev_priv->renderctx->gtt_offset | + MI_MM_SPACE_GTT | + MI_SAVE_EXT_STATE_EN | + MI_RESTORE_EXT_STATE_EN | + MI_RESTORE_INHIBIT); + OUT_RING(MI_SUSPEND_FLUSH); + OUT_RING(MI_NOOP); + OUT_RING(MI_FLUSH); + ADVANCE_LP_RING(); + + /* + * Wait for the command parser to advance past MI_SET_CONTEXT. The HW + * does an implicit flush, combined with MI_FLUSH above, it should be + * safe to assume that renderctx is valid + */ + ret = intel_wait_ring_idle(LP_RING(dev_priv)); + if (ret) { + DRM_ERROR("failed to enable ironlake power power savings\n"); + ironlake_teardown_rc6(dev); + mutex_unlock(&dev->struct_mutex); + return; + } + + I915_WRITE(PWRCTXA, dev_priv->pwrctx->gtt_offset | PWRCTX_EN); + I915_WRITE(RSTDBYCTL, I915_READ(RSTDBYCTL) & ~RCX_SW_EXIT); + mutex_unlock(&dev->struct_mutex); +} + +void intel_init_clock_gating(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + dev_priv->display.init_clock_gating(dev); + + if (dev_priv->display.init_pch_clock_gating) + dev_priv->display.init_pch_clock_gating(dev); +} + +/* Set up chip specific display functions */ +static void intel_init_display(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + /* We always want a DPMS function */ + if (HAS_PCH_SPLIT(dev)) { + dev_priv->display.dpms = ironlake_crtc_dpms; + dev_priv->display.crtc_mode_set = ironlake_crtc_mode_set; + } else { + dev_priv->display.dpms = i9xx_crtc_dpms; + dev_priv->display.crtc_mode_set = i9xx_crtc_mode_set; + } + + if (I915_HAS_FBC(dev)) { + if (HAS_PCH_SPLIT(dev)) { + dev_priv->display.fbc_enabled = ironlake_fbc_enabled; + dev_priv->display.enable_fbc = ironlake_enable_fbc; + dev_priv->display.disable_fbc = ironlake_disable_fbc; + } else if (IS_GM45(dev)) { + dev_priv->display.fbc_enabled = g4x_fbc_enabled; + dev_priv->display.enable_fbc = g4x_enable_fbc; + dev_priv->display.disable_fbc = g4x_disable_fbc; + } else if (IS_CRESTLINE(dev)) { + dev_priv->display.fbc_enabled = i8xx_fbc_enabled; + dev_priv->display.enable_fbc = i8xx_enable_fbc; + dev_priv->display.disable_fbc = i8xx_disable_fbc; + } + /* 855GM needs testing */ + } + + /* Returns the core display clock speed */ + if (IS_I945G(dev) || (IS_G33(dev) && ! IS_PINEVIEW_M(dev))) + dev_priv->display.get_display_clock_speed = + i945_get_display_clock_speed; + else if (IS_I915G(dev)) + dev_priv->display.get_display_clock_speed = + i915_get_display_clock_speed; + else if (IS_I945GM(dev) || IS_845G(dev) || IS_PINEVIEW_M(dev)) + dev_priv->display.get_display_clock_speed = + i9xx_misc_get_display_clock_speed; + else if (IS_I915GM(dev)) + dev_priv->display.get_display_clock_speed = + i915gm_get_display_clock_speed; + else if (IS_I865G(dev)) + dev_priv->display.get_display_clock_speed = + i865_get_display_clock_speed; + else if (IS_I85X(dev)) + dev_priv->display.get_display_clock_speed = + i855_get_display_clock_speed; + else /* 852, 830 */ + dev_priv->display.get_display_clock_speed = + i830_get_display_clock_speed; + + /* For FIFO watermark updates */ + if (HAS_PCH_SPLIT(dev)) { + if (HAS_PCH_IBX(dev)) + dev_priv->display.init_pch_clock_gating = ibx_init_clock_gating; + else if (HAS_PCH_CPT(dev)) + dev_priv->display.init_pch_clock_gating = cpt_init_clock_gating; + + if (IS_GEN5(dev)) { + if (I915_READ(MLTR_ILK) & ILK_SRLT_MASK) + dev_priv->display.update_wm = ironlake_update_wm; + else { + DRM_DEBUG_KMS("Failed to get proper latency. " + "Disable CxSR\n"); + dev_priv->display.update_wm = NULL; + } + dev_priv->display.fdi_link_train = ironlake_fdi_link_train; + dev_priv->display.init_clock_gating = ironlake_init_clock_gating; + } else if (IS_GEN6(dev)) { + if (SNB_READ_WM0_LATENCY()) { + dev_priv->display.update_wm = sandybridge_update_wm; + } else { + DRM_DEBUG_KMS("Failed to read display plane latency. " + "Disable CxSR\n"); + dev_priv->display.update_wm = NULL; + } + dev_priv->display.fdi_link_train = gen6_fdi_link_train; + dev_priv->display.init_clock_gating = gen6_init_clock_gating; + } else if (IS_IVYBRIDGE(dev)) { + /* FIXME: detect B0+ stepping and use auto training */ + dev_priv->display.fdi_link_train = ivb_manual_fdi_link_train; + if (SNB_READ_WM0_LATENCY()) { + dev_priv->display.update_wm = sandybridge_update_wm; + } else { + DRM_DEBUG_KMS("Failed to read display plane latency. " + "Disable CxSR\n"); + dev_priv->display.update_wm = NULL; + } + dev_priv->display.init_clock_gating = ivybridge_init_clock_gating; + + } else + dev_priv->display.update_wm = NULL; + } else if (IS_PINEVIEW(dev)) { + if (!intel_get_cxsr_latency(IS_PINEVIEW_G(dev), + dev_priv->is_ddr3, + dev_priv->fsb_freq, + dev_priv->mem_freq)) { + DRM_INFO("failed to find known CxSR latency " + "(found ddr%s fsb freq %d, mem freq %d), " + "disabling CxSR\n", + (dev_priv->is_ddr3 == 1) ? "3": "2", + dev_priv->fsb_freq, dev_priv->mem_freq); + /* Disable CxSR and never update its watermark again */ + pineview_disable_cxsr(dev); + dev_priv->display.update_wm = NULL; + } else + dev_priv->display.update_wm = pineview_update_wm; + dev_priv->display.init_clock_gating = gen3_init_clock_gating; + } else if (IS_G4X(dev)) { + dev_priv->display.update_wm = g4x_update_wm; + dev_priv->display.init_clock_gating = g4x_init_clock_gating; + } else if (IS_GEN4(dev)) { + dev_priv->display.update_wm = i965_update_wm; + if (IS_CRESTLINE(dev)) + dev_priv->display.init_clock_gating = crestline_init_clock_gating; + else if (IS_BROADWATER(dev)) + dev_priv->display.init_clock_gating = broadwater_init_clock_gating; + } else if (IS_GEN3(dev)) { + dev_priv->display.update_wm = i9xx_update_wm; + dev_priv->display.get_fifo_size = i9xx_get_fifo_size; + dev_priv->display.init_clock_gating = gen3_init_clock_gating; + } else if (IS_I865G(dev)) { + dev_priv->display.update_wm = i830_update_wm; + dev_priv->display.init_clock_gating = i85x_init_clock_gating; + dev_priv->display.get_fifo_size = i830_get_fifo_size; + } else if (IS_I85X(dev)) { + dev_priv->display.update_wm = i9xx_update_wm; + dev_priv->display.get_fifo_size = i85x_get_fifo_size; + dev_priv->display.init_clock_gating = i85x_init_clock_gating; + } else { + dev_priv->display.update_wm = i830_update_wm; + dev_priv->display.init_clock_gating = i830_init_clock_gating; + if (IS_845G(dev)) + dev_priv->display.get_fifo_size = i845_get_fifo_size; + else + dev_priv->display.get_fifo_size = i830_get_fifo_size; + } + + /* Default just returns -ENODEV to indicate unsupported */ + dev_priv->display.queue_flip = intel_default_queue_flip; + + switch (INTEL_INFO(dev)->gen) { + case 2: + dev_priv->display.queue_flip = intel_gen2_queue_flip; + break; + + case 3: + dev_priv->display.queue_flip = intel_gen3_queue_flip; + break; + + case 4: + case 5: + dev_priv->display.queue_flip = intel_gen4_queue_flip; + break; + + case 6: + dev_priv->display.queue_flip = intel_gen6_queue_flip; + break; + case 7: + dev_priv->display.queue_flip = intel_gen7_queue_flip; + break; + } +} + +/* + * Some BIOSes insist on assuming the GPU's pipe A is enabled at suspend, + * resume, or other times. This quirk makes sure that's the case for + * affected systems. + */ +static void quirk_pipea_force (struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + dev_priv->quirks |= QUIRK_PIPEA_FORCE; + DRM_DEBUG_DRIVER("applying pipe a force quirk\n"); +} + +/* + * Some machines (Lenovo U160) do not work with SSC on LVDS for some reason + */ +static void quirk_ssc_force_disable(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + dev_priv->quirks |= QUIRK_LVDS_SSC_DISABLE; +} + +struct intel_quirk { + int device; + int subsystem_vendor; + int subsystem_device; + void (*hook)(struct drm_device *dev); +}; + +struct intel_quirk intel_quirks[] = { + /* HP Compaq 2730p needs pipe A force quirk (LP: #291555) */ + { 0x2a42, 0x103c, 0x30eb, quirk_pipea_force }, + /* HP Mini needs pipe A force quirk (LP: #322104) */ + { 0x27ae,0x103c, 0x361a, quirk_pipea_force }, + + /* Thinkpad R31 needs pipe A force quirk */ + { 0x3577, 0x1014, 0x0505, quirk_pipea_force }, + /* Toshiba Protege R-205, S-209 needs pipe A force quirk */ + { 0x2592, 0x1179, 0x0001, quirk_pipea_force }, + + /* ThinkPad X30 needs pipe A force quirk (LP: #304614) */ + { 0x3577, 0x1014, 0x0513, quirk_pipea_force }, + /* ThinkPad X40 needs pipe A force quirk */ + + /* ThinkPad T60 needs pipe A force quirk (bug #16494) */ + { 0x2782, 0x17aa, 0x201a, quirk_pipea_force }, + + /* 855 & before need to leave pipe A & dpll A up */ + { 0x3582, PCI_ANY_ID, PCI_ANY_ID, quirk_pipea_force }, + { 0x2562, PCI_ANY_ID, PCI_ANY_ID, quirk_pipea_force }, + + /* Lenovo U160 cannot use SSC on LVDS */ + { 0x0046, 0x17aa, 0x3920, quirk_ssc_force_disable }, +}; + +static void intel_init_quirks(struct drm_device *dev) +{ + struct pci_dev *d = dev->pdev; + int i; + + for (i = 0; i < ARRAY_SIZE(intel_quirks); i++) { + struct intel_quirk *q = &intel_quirks[i]; + + if (d->device == q->device && + (d->subsystem_vendor == q->subsystem_vendor || + q->subsystem_vendor == PCI_ANY_ID) && + (d->subsystem_device == q->subsystem_device || + q->subsystem_device == PCI_ANY_ID)) + q->hook(dev); + } +} + +/* Disable the VGA plane that we never use */ +static void i915_disable_vga(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u8 sr1; + u32 vga_reg; + + if (HAS_PCH_SPLIT(dev)) + vga_reg = CPU_VGACNTRL; + else + vga_reg = VGACNTRL; + + vga_get_uninterruptible(dev->pdev, VGA_RSRC_LEGACY_IO); + outb(1, VGA_SR_INDEX); + sr1 = inb(VGA_SR_DATA); + outb(sr1 | 1<<5, VGA_SR_DATA); + vga_put(dev->pdev, VGA_RSRC_LEGACY_IO); + udelay(300); + + I915_WRITE(vga_reg, VGA_DISP_DISABLE); + POSTING_READ(vga_reg); +} + +void intel_modeset_init(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + int i; + + drm_mode_config_init(dev); + + dev->mode_config.min_width = 0; + dev->mode_config.min_height = 0; + + dev->mode_config.funcs = (void *)&intel_mode_funcs; + + intel_init_quirks(dev); + + intel_init_display(dev); + + if (IS_GEN2(dev)) { + dev->mode_config.max_width = 2048; + dev->mode_config.max_height = 2048; + } else if (IS_GEN3(dev)) { + dev->mode_config.max_width = 4096; + dev->mode_config.max_height = 4096; + } else { + dev->mode_config.max_width = 8192; + dev->mode_config.max_height = 8192; + } + dev->mode_config.fb_base = dev->agp->base; + + DRM_DEBUG_KMS("%d display pipe%s available.\n", + dev_priv->num_pipe, dev_priv->num_pipe > 1 ? "s" : ""); + + for (i = 0; i < dev_priv->num_pipe; i++) { + intel_crtc_init(dev, i); + } + + /* Just disable it once at startup */ + i915_disable_vga(dev); + intel_setup_outputs(dev); + + intel_init_clock_gating(dev); + + if (IS_IRONLAKE_M(dev)) { + ironlake_enable_drps(dev); + intel_init_emon(dev); + } + + if (IS_GEN6(dev) || IS_GEN7(dev)) + gen6_enable_rps(dev_priv); + + INIT_WORK(&dev_priv->idle_work, intel_idle_update); + setup_timer(&dev_priv->idle_timer, intel_gpu_idle_timer, + (unsigned long)dev); +} + +void intel_modeset_gem_init(struct drm_device *dev) +{ + if (IS_IRONLAKE_M(dev)) + ironlake_enable_rc6(dev); + + intel_setup_overlay(dev); +} + +void intel_modeset_cleanup(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_crtc *crtc; + struct intel_crtc *intel_crtc; + + drm_kms_helper_poll_fini(dev); + mutex_lock(&dev->struct_mutex); + + intel_unregister_dsm_handler(); + + + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { + /* Skip inactive CRTCs */ + if (!crtc->fb) + continue; + + intel_crtc = to_intel_crtc(crtc); + intel_increase_pllclock(crtc); + } + + if (dev_priv->display.disable_fbc) + dev_priv->display.disable_fbc(dev); + + if (IS_IRONLAKE_M(dev)) + ironlake_disable_drps(dev); + if (IS_GEN6(dev) || IS_GEN7(dev)) + gen6_disable_rps(dev); + + if (IS_IRONLAKE_M(dev)) + ironlake_disable_rc6(dev); + + mutex_unlock(&dev->struct_mutex); + + /* Disable the irq before mode object teardown, for the irq might + * enqueue unpin/hotplug work. */ + drm_irq_uninstall(dev); + cancel_work_sync(&dev_priv->hotplug_work); + + /* Shut off idle work before the crtcs get freed. */ + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { + intel_crtc = to_intel_crtc(crtc); + del_timer_sync(&intel_crtc->idle_timer); + } + del_timer_sync(&dev_priv->idle_timer); + cancel_work_sync(&dev_priv->idle_work); + + drm_mode_config_cleanup(dev); +} + +/* + * Return which encoder is currently attached for connector. + */ +struct drm_encoder *intel_best_encoder(struct drm_connector *connector) +{ + return &intel_attached_encoder(connector)->base; +} + +void intel_connector_attach_encoder(struct intel_connector *connector, + struct intel_encoder *encoder) +{ + connector->encoder = encoder; + drm_mode_connector_attach_encoder(&connector->base, + &encoder->base); +} + +/* + * set vga decode state - true == enable VGA decode + */ +int intel_modeset_vga_set_state(struct drm_device *dev, bool state) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u16 gmch_ctrl; + + pci_read_config_word(dev_priv->bridge_dev, INTEL_GMCH_CTRL, &gmch_ctrl); + if (state) + gmch_ctrl &= ~INTEL_GMCH_VGA_DISABLE; + else + gmch_ctrl |= INTEL_GMCH_VGA_DISABLE; + pci_write_config_word(dev_priv->bridge_dev, INTEL_GMCH_CTRL, gmch_ctrl); + return 0; +} + +#ifdef CONFIG_DEBUG_FS +#include + +struct intel_display_error_state { + struct intel_cursor_error_state { + u32 control; + u32 position; + u32 base; + u32 size; + } cursor[2]; + + struct intel_pipe_error_state { + u32 conf; + u32 source; + + u32 htotal; + u32 hblank; + u32 hsync; + u32 vtotal; + u32 vblank; + u32 vsync; + } pipe[2]; + + struct intel_plane_error_state { + u32 control; + u32 stride; + u32 size; + u32 pos; + u32 addr; + u32 surface; + u32 tile_offset; + } plane[2]; +}; + +struct intel_display_error_state * +intel_display_capture_error_state(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct intel_display_error_state *error; + int i; + + error = kmalloc(sizeof(*error), GFP_ATOMIC); + if (error == NULL) + return NULL; + + for (i = 0; i < 2; i++) { + error->cursor[i].control = I915_READ(CURCNTR(i)); + error->cursor[i].position = I915_READ(CURPOS(i)); + error->cursor[i].base = I915_READ(CURBASE(i)); + + error->plane[i].control = I915_READ(DSPCNTR(i)); + error->plane[i].stride = I915_READ(DSPSTRIDE(i)); + error->plane[i].size = I915_READ(DSPSIZE(i)); + error->plane[i].pos= I915_READ(DSPPOS(i)); + error->plane[i].addr = I915_READ(DSPADDR(i)); + if (INTEL_INFO(dev)->gen >= 4) { + error->plane[i].surface = I915_READ(DSPSURF(i)); + error->plane[i].tile_offset = I915_READ(DSPTILEOFF(i)); + } + + error->pipe[i].conf = I915_READ(PIPECONF(i)); + error->pipe[i].source = I915_READ(PIPESRC(i)); + error->pipe[i].htotal = I915_READ(HTOTAL(i)); + error->pipe[i].hblank = I915_READ(HBLANK(i)); + error->pipe[i].hsync = I915_READ(HSYNC(i)); + error->pipe[i].vtotal = I915_READ(VTOTAL(i)); + error->pipe[i].vblank = I915_READ(VBLANK(i)); + error->pipe[i].vsync = I915_READ(VSYNC(i)); + } + + return error; +} + +void +intel_display_print_error_state(struct seq_file *m, + struct drm_device *dev, + struct intel_display_error_state *error) +{ + int i; + + for (i = 0; i < 2; i++) { + seq_printf(m, "Pipe [%d]:\n", i); + seq_printf(m, " CONF: %08x\n", error->pipe[i].conf); + seq_printf(m, " SRC: %08x\n", error->pipe[i].source); + seq_printf(m, " HTOTAL: %08x\n", error->pipe[i].htotal); + seq_printf(m, " HBLANK: %08x\n", error->pipe[i].hblank); + seq_printf(m, " HSYNC: %08x\n", error->pipe[i].hsync); + seq_printf(m, " VTOTAL: %08x\n", error->pipe[i].vtotal); + seq_printf(m, " VBLANK: %08x\n", error->pipe[i].vblank); + seq_printf(m, " VSYNC: %08x\n", error->pipe[i].vsync); + + seq_printf(m, "Plane [%d]:\n", i); + seq_printf(m, " CNTR: %08x\n", error->plane[i].control); + seq_printf(m, " STRIDE: %08x\n", error->plane[i].stride); + seq_printf(m, " SIZE: %08x\n", error->plane[i].size); + seq_printf(m, " POS: %08x\n", error->plane[i].pos); + seq_printf(m, " ADDR: %08x\n", error->plane[i].addr); + if (INTEL_INFO(dev)->gen >= 4) { + seq_printf(m, " SURF: %08x\n", error->plane[i].surface); + seq_printf(m, " TILEOFF: %08x\n", error->plane[i].tile_offset); + } + + seq_printf(m, "Cursor [%d]:\n", i); + seq_printf(m, " CNTR: %08x\n", error->cursor[i].control); + seq_printf(m, " POS: %08x\n", error->cursor[i].position); + seq_printf(m, " BASE: %08x\n", error->cursor[i].base); + } +} +#endif diff --git a/drivers/gpu/drm/i915/intel_dp.c b/drivers/gpu/drm/i915/intel_dp.c new file mode 100644 index 00000000..bf9fea94 --- /dev/null +++ b/drivers/gpu/drm/i915/intel_dp.c @@ -0,0 +1,2083 @@ +/* + * Copyright © 2008 Intel Corporation + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS + * IN THE SOFTWARE. + * + * Authors: + * Keith Packard + * + */ + +#include +#include +#include "drmP.h" +#include "drm.h" +#include "drm_crtc.h" +#include "drm_crtc_helper.h" +#include "intel_drv.h" +#include "i915_drm.h" +#include "i915_drv.h" +#include "drm_dp_helper.h" + + +#define DP_LINK_STATUS_SIZE 6 +#define DP_LINK_CHECK_TIMEOUT (10 * 1000) + +#define DP_LINK_CONFIGURATION_SIZE 9 + +struct intel_dp { + struct intel_encoder base; + uint32_t output_reg; + uint32_t DP; + uint8_t link_configuration[DP_LINK_CONFIGURATION_SIZE]; + bool has_audio; + int force_audio; + uint32_t color_range; + uint8_t link_bw; + uint8_t lane_count; + uint8_t dpcd[4]; + struct i2c_adapter adapter; + struct i2c_algo_dp_aux_data algo; + bool is_pch_edp; + uint8_t train_set[4]; + uint8_t link_status[DP_LINK_STATUS_SIZE]; +}; + +/** + * is_edp - is the given port attached to an eDP panel (either CPU or PCH) + * @intel_dp: DP struct + * + * If a CPU or PCH DP output is attached to an eDP panel, this function + * will return true, and false otherwise. + */ +static bool is_edp(struct intel_dp *intel_dp) +{ + return intel_dp->base.type == INTEL_OUTPUT_EDP; +} + +/** + * is_pch_edp - is the port on the PCH and attached to an eDP panel? + * @intel_dp: DP struct + * + * Returns true if the given DP struct corresponds to a PCH DP port attached + * to an eDP panel, false otherwise. Helpful for determining whether we + * may need FDI resources for a given DP output or not. + */ +static bool is_pch_edp(struct intel_dp *intel_dp) +{ + return intel_dp->is_pch_edp; +} + +static struct intel_dp *enc_to_intel_dp(struct drm_encoder *encoder) +{ + return container_of(encoder, struct intel_dp, base.base); +} + +static struct intel_dp *intel_attached_dp(struct drm_connector *connector) +{ + return container_of(intel_attached_encoder(connector), + struct intel_dp, base); +} + +/** + * intel_encoder_is_pch_edp - is the given encoder a PCH attached eDP? + * @encoder: DRM encoder + * + * Return true if @encoder corresponds to a PCH attached eDP panel. Needed + * by intel_display.c. + */ +bool intel_encoder_is_pch_edp(struct drm_encoder *encoder) +{ + struct intel_dp *intel_dp; + + if (!encoder) + return false; + + intel_dp = enc_to_intel_dp(encoder); + + return is_pch_edp(intel_dp); +} + +static void intel_dp_start_link_train(struct intel_dp *intel_dp); +static void intel_dp_complete_link_train(struct intel_dp *intel_dp); +static void intel_dp_link_down(struct intel_dp *intel_dp); + +void +intel_edp_link_config (struct intel_encoder *intel_encoder, + int *lane_num, int *link_bw) +{ + struct intel_dp *intel_dp = container_of(intel_encoder, struct intel_dp, base); + + *lane_num = intel_dp->lane_count; + if (intel_dp->link_bw == DP_LINK_BW_1_62) + *link_bw = 162000; + else if (intel_dp->link_bw == DP_LINK_BW_2_7) + *link_bw = 270000; +} + +static int +intel_dp_max_lane_count(struct intel_dp *intel_dp) +{ + int max_lane_count = 4; + + if (intel_dp->dpcd[DP_DPCD_REV] >= 0x11) { + max_lane_count = intel_dp->dpcd[DP_MAX_LANE_COUNT] & 0x1f; + switch (max_lane_count) { + case 1: case 2: case 4: + break; + default: + max_lane_count = 4; + } + } + return max_lane_count; +} + +static int +intel_dp_max_link_bw(struct intel_dp *intel_dp) +{ + int max_link_bw = intel_dp->dpcd[DP_MAX_LINK_RATE]; + + switch (max_link_bw) { + case DP_LINK_BW_1_62: + case DP_LINK_BW_2_7: + break; + default: + max_link_bw = DP_LINK_BW_1_62; + break; + } + return max_link_bw; +} + +static int +intel_dp_link_clock(uint8_t link_bw) +{ + if (link_bw == DP_LINK_BW_2_7) + return 270000; + else + return 162000; +} + +/* I think this is a fiction */ +static int +intel_dp_link_required(struct drm_device *dev, struct intel_dp *intel_dp, int pixel_clock) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + if (is_edp(intel_dp)) + return (pixel_clock * dev_priv->edp.bpp + 7) / 8; + else + return pixel_clock * 3; +} + +static int +intel_dp_max_data_rate(int max_link_clock, int max_lanes) +{ + return (max_link_clock * max_lanes * 8) / 10; +} + +static int +intel_dp_mode_valid(struct drm_connector *connector, + struct drm_display_mode *mode) +{ + struct intel_dp *intel_dp = intel_attached_dp(connector); + struct drm_device *dev = connector->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + int max_link_clock = intel_dp_link_clock(intel_dp_max_link_bw(intel_dp)); + int max_lanes = intel_dp_max_lane_count(intel_dp); + + if (is_edp(intel_dp) && dev_priv->panel_fixed_mode) { + if (mode->hdisplay > dev_priv->panel_fixed_mode->hdisplay) + return MODE_PANEL; + + if (mode->vdisplay > dev_priv->panel_fixed_mode->vdisplay) + return MODE_PANEL; + } + + /* only refuse the mode on non eDP since we have seen some weird eDP panels + which are outside spec tolerances but somehow work by magic */ + if (!is_edp(intel_dp) && + (intel_dp_link_required(connector->dev, intel_dp, mode->clock) + > intel_dp_max_data_rate(max_link_clock, max_lanes))) + return MODE_CLOCK_HIGH; + + if (mode->clock < 10000) + return MODE_CLOCK_LOW; + + return MODE_OK; +} + +static uint32_t +pack_aux(uint8_t *src, int src_bytes) +{ + int i; + uint32_t v = 0; + + if (src_bytes > 4) + src_bytes = 4; + for (i = 0; i < src_bytes; i++) + v |= ((uint32_t) src[i]) << ((3-i) * 8); + return v; +} + +static void +unpack_aux(uint32_t src, uint8_t *dst, int dst_bytes) +{ + int i; + if (dst_bytes > 4) + dst_bytes = 4; + for (i = 0; i < dst_bytes; i++) + dst[i] = src >> ((3-i) * 8); +} + +/* hrawclock is 1/4 the FSB frequency */ +static int +intel_hrawclk(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + uint32_t clkcfg; + + clkcfg = I915_READ(CLKCFG); + switch (clkcfg & CLKCFG_FSB_MASK) { + case CLKCFG_FSB_400: + return 100; + case CLKCFG_FSB_533: + return 133; + case CLKCFG_FSB_667: + return 166; + case CLKCFG_FSB_800: + return 200; + case CLKCFG_FSB_1067: + return 266; + case CLKCFG_FSB_1333: + return 333; + /* these two are just a guess; one of them might be right */ + case CLKCFG_FSB_1600: + case CLKCFG_FSB_1600_ALT: + return 400; + default: + return 133; + } +} + +static int +intel_dp_aux_ch(struct intel_dp *intel_dp, + uint8_t *send, int send_bytes, + uint8_t *recv, int recv_size) +{ + uint32_t output_reg = intel_dp->output_reg; + struct drm_device *dev = intel_dp->base.base.dev; + struct drm_i915_private *dev_priv = dev->dev_private; + uint32_t ch_ctl = output_reg + 0x10; + uint32_t ch_data = ch_ctl + 4; + int i; + int recv_bytes; + uint32_t status; + uint32_t aux_clock_divider; + int try, precharge; + + /* The clock divider is based off the hrawclk, + * and would like to run at 2MHz. So, take the + * hrawclk value and divide by 2 and use that + * + * Note that PCH attached eDP panels should use a 125MHz input + * clock divider. + */ + if (is_edp(intel_dp) && !is_pch_edp(intel_dp)) { + if (IS_GEN6(dev)) + aux_clock_divider = 200; /* SNB eDP input clock at 400Mhz */ + else + aux_clock_divider = 225; /* eDP input clock at 450Mhz */ + } else if (HAS_PCH_SPLIT(dev)) + aux_clock_divider = 62; /* IRL input clock fixed at 125Mhz */ + else + aux_clock_divider = intel_hrawclk(dev) / 2; + + if (IS_GEN6(dev)) + precharge = 3; + else + precharge = 5; + + if (I915_READ(ch_ctl) & DP_AUX_CH_CTL_SEND_BUSY) { + DRM_ERROR("dp_aux_ch not started status 0x%08x\n", + I915_READ(ch_ctl)); + return -EBUSY; + } + + /* Must try at least 3 times according to DP spec */ + for (try = 0; try < 5; try++) { + /* Load the send data into the aux channel data registers */ + for (i = 0; i < send_bytes; i += 4) + I915_WRITE(ch_data + i, + pack_aux(send + i, send_bytes - i)); + + /* Send the command and wait for it to complete */ + I915_WRITE(ch_ctl, + DP_AUX_CH_CTL_SEND_BUSY | + DP_AUX_CH_CTL_TIME_OUT_400us | + (send_bytes << DP_AUX_CH_CTL_MESSAGE_SIZE_SHIFT) | + (precharge << DP_AUX_CH_CTL_PRECHARGE_2US_SHIFT) | + (aux_clock_divider << DP_AUX_CH_CTL_BIT_CLOCK_2X_SHIFT) | + DP_AUX_CH_CTL_DONE | + DP_AUX_CH_CTL_TIME_OUT_ERROR | + DP_AUX_CH_CTL_RECEIVE_ERROR); + for (;;) { + status = I915_READ(ch_ctl); + if ((status & DP_AUX_CH_CTL_SEND_BUSY) == 0) + break; + udelay(100); + } + + /* Clear done status and any errors */ + I915_WRITE(ch_ctl, + status | + DP_AUX_CH_CTL_DONE | + DP_AUX_CH_CTL_TIME_OUT_ERROR | + DP_AUX_CH_CTL_RECEIVE_ERROR); + if (status & DP_AUX_CH_CTL_DONE) + break; + } + + if ((status & DP_AUX_CH_CTL_DONE) == 0) { + DRM_ERROR("dp_aux_ch not done status 0x%08x\n", status); + return -EBUSY; + } + + /* Check for timeout or receive error. + * Timeouts occur when the sink is not connected + */ + if (status & DP_AUX_CH_CTL_RECEIVE_ERROR) { + DRM_ERROR("dp_aux_ch receive error status 0x%08x\n", status); + return -EIO; + } + + /* Timeouts occur when the device isn't connected, so they're + * "normal" -- don't fill the kernel log with these */ + if (status & DP_AUX_CH_CTL_TIME_OUT_ERROR) { + DRM_DEBUG_KMS("dp_aux_ch timeout status 0x%08x\n", status); + return -ETIMEDOUT; + } + + /* Unload any bytes sent back from the other side */ + recv_bytes = ((status & DP_AUX_CH_CTL_MESSAGE_SIZE_MASK) >> + DP_AUX_CH_CTL_MESSAGE_SIZE_SHIFT); + if (recv_bytes > recv_size) + recv_bytes = recv_size; + + for (i = 0; i < recv_bytes; i += 4) + unpack_aux(I915_READ(ch_data + i), + recv + i, recv_bytes - i); + + return recv_bytes; +} + +/* Write data to the aux channel in native mode */ +static int +intel_dp_aux_native_write(struct intel_dp *intel_dp, + uint16_t address, uint8_t *send, int send_bytes) +{ + int ret; + uint8_t msg[20]; + int msg_bytes; + uint8_t ack; + + if (send_bytes > 16) + return -1; + msg[0] = AUX_NATIVE_WRITE << 4; + msg[1] = address >> 8; + msg[2] = address & 0xff; + msg[3] = send_bytes - 1; + memcpy(&msg[4], send, send_bytes); + msg_bytes = send_bytes + 4; + for (;;) { + ret = intel_dp_aux_ch(intel_dp, msg, msg_bytes, &ack, 1); + if (ret < 0) + return ret; + if ((ack & AUX_NATIVE_REPLY_MASK) == AUX_NATIVE_REPLY_ACK) + break; + else if ((ack & AUX_NATIVE_REPLY_MASK) == AUX_NATIVE_REPLY_DEFER) + udelay(100); + else + return -EIO; + } + return send_bytes; +} + +/* Write a single byte to the aux channel in native mode */ +static int +intel_dp_aux_native_write_1(struct intel_dp *intel_dp, + uint16_t address, uint8_t byte) +{ + return intel_dp_aux_native_write(intel_dp, address, &byte, 1); +} + +/* read bytes from a native aux channel */ +static int +intel_dp_aux_native_read(struct intel_dp *intel_dp, + uint16_t address, uint8_t *recv, int recv_bytes) +{ + uint8_t msg[4]; + int msg_bytes; + uint8_t reply[20]; + int reply_bytes; + uint8_t ack; + int ret; + + msg[0] = AUX_NATIVE_READ << 4; + msg[1] = address >> 8; + msg[2] = address & 0xff; + msg[3] = recv_bytes - 1; + + msg_bytes = 4; + reply_bytes = recv_bytes + 1; + + for (;;) { + ret = intel_dp_aux_ch(intel_dp, msg, msg_bytes, + reply, reply_bytes); + if (ret == 0) + return -EPROTO; + if (ret < 0) + return ret; + ack = reply[0]; + if ((ack & AUX_NATIVE_REPLY_MASK) == AUX_NATIVE_REPLY_ACK) { + memcpy(recv, reply + 1, ret - 1); + return ret - 1; + } + else if ((ack & AUX_NATIVE_REPLY_MASK) == AUX_NATIVE_REPLY_DEFER) + udelay(100); + else + return -EIO; + } +} + +static int +intel_dp_i2c_aux_ch(struct i2c_adapter *adapter, int mode, + uint8_t write_byte, uint8_t *read_byte) +{ + struct i2c_algo_dp_aux_data *algo_data = adapter->algo_data; + struct intel_dp *intel_dp = container_of(adapter, + struct intel_dp, + adapter); + uint16_t address = algo_data->address; + uint8_t msg[5]; + uint8_t reply[2]; + unsigned retry; + int msg_bytes; + int reply_bytes; + int ret; + + /* Set up the command byte */ + if (mode & MODE_I2C_READ) + msg[0] = AUX_I2C_READ << 4; + else + msg[0] = AUX_I2C_WRITE << 4; + + if (!(mode & MODE_I2C_STOP)) + msg[0] |= AUX_I2C_MOT << 4; + + msg[1] = address >> 8; + msg[2] = address; + + switch (mode) { + case MODE_I2C_WRITE: + msg[3] = 0; + msg[4] = write_byte; + msg_bytes = 5; + reply_bytes = 1; + break; + case MODE_I2C_READ: + msg[3] = 0; + msg_bytes = 4; + reply_bytes = 2; + break; + default: + msg_bytes = 3; + reply_bytes = 1; + break; + } + + for (retry = 0; retry < 5; retry++) { + ret = intel_dp_aux_ch(intel_dp, + msg, msg_bytes, + reply, reply_bytes); + if (ret < 0) { + DRM_DEBUG_KMS("aux_ch failed %d\n", ret); + return ret; + } + + switch (reply[0] & AUX_NATIVE_REPLY_MASK) { + case AUX_NATIVE_REPLY_ACK: + /* I2C-over-AUX Reply field is only valid + * when paired with AUX ACK. + */ + break; + case AUX_NATIVE_REPLY_NACK: + DRM_DEBUG_KMS("aux_ch native nack\n"); + return -EREMOTEIO; + case AUX_NATIVE_REPLY_DEFER: + udelay(100); + continue; + default: + DRM_ERROR("aux_ch invalid native reply 0x%02x\n", + reply[0]); + return -EREMOTEIO; + } + + switch (reply[0] & AUX_I2C_REPLY_MASK) { + case AUX_I2C_REPLY_ACK: + if (mode == MODE_I2C_READ) { + *read_byte = reply[1]; + } + return reply_bytes - 1; + case AUX_I2C_REPLY_NACK: + DRM_DEBUG_KMS("aux_i2c nack\n"); + return -EREMOTEIO; + case AUX_I2C_REPLY_DEFER: + DRM_DEBUG_KMS("aux_i2c defer\n"); + udelay(100); + break; + default: + DRM_ERROR("aux_i2c invalid reply 0x%02x\n", reply[0]); + return -EREMOTEIO; + } + } + + DRM_ERROR("too many retries, giving up\n"); + return -EREMOTEIO; +} + +static int +intel_dp_i2c_init(struct intel_dp *intel_dp, + struct intel_connector *intel_connector, const char *name) +{ + DRM_DEBUG_KMS("i2c_init %s\n", name); + intel_dp->algo.running = false; + intel_dp->algo.address = 0; + intel_dp->algo.aux_ch = intel_dp_i2c_aux_ch; + + memset(&intel_dp->adapter, '\0', sizeof (intel_dp->adapter)); + intel_dp->adapter.owner = THIS_MODULE; + intel_dp->adapter.class = I2C_CLASS_DDC; + strncpy (intel_dp->adapter.name, name, sizeof(intel_dp->adapter.name) - 1); + intel_dp->adapter.name[sizeof(intel_dp->adapter.name) - 1] = '\0'; + intel_dp->adapter.algo_data = &intel_dp->algo; + intel_dp->adapter.dev.parent = &intel_connector->base.kdev; + + return i2c_dp_aux_add_bus(&intel_dp->adapter); +} + +static bool +intel_dp_mode_fixup(struct drm_encoder *encoder, struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct drm_device *dev = encoder->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_dp *intel_dp = enc_to_intel_dp(encoder); + int lane_count, clock; + int max_lane_count = intel_dp_max_lane_count(intel_dp); + int max_clock = intel_dp_max_link_bw(intel_dp) == DP_LINK_BW_2_7 ? 1 : 0; + static int bws[2] = { DP_LINK_BW_1_62, DP_LINK_BW_2_7 }; + + if (is_edp(intel_dp) && dev_priv->panel_fixed_mode) { + intel_fixed_panel_mode(dev_priv->panel_fixed_mode, adjusted_mode); + intel_pch_panel_fitting(dev, DRM_MODE_SCALE_FULLSCREEN, + mode, adjusted_mode); + /* + * the mode->clock is used to calculate the Data&Link M/N + * of the pipe. For the eDP the fixed clock should be used. + */ + mode->clock = dev_priv->panel_fixed_mode->clock; + } + + for (lane_count = 1; lane_count <= max_lane_count; lane_count <<= 1) { + for (clock = 0; clock <= max_clock; clock++) { + int link_avail = intel_dp_max_data_rate(intel_dp_link_clock(bws[clock]), lane_count); + + if (intel_dp_link_required(encoder->dev, intel_dp, mode->clock) + <= link_avail) { + intel_dp->link_bw = bws[clock]; + intel_dp->lane_count = lane_count; + adjusted_mode->clock = intel_dp_link_clock(intel_dp->link_bw); + DRM_DEBUG_KMS("Display port link bw %02x lane " + "count %d clock %d\n", + intel_dp->link_bw, intel_dp->lane_count, + adjusted_mode->clock); + return true; + } + } + } + + if (is_edp(intel_dp)) { + /* okay we failed just pick the highest */ + intel_dp->lane_count = max_lane_count; + intel_dp->link_bw = bws[max_clock]; + adjusted_mode->clock = intel_dp_link_clock(intel_dp->link_bw); + DRM_DEBUG_KMS("Force picking display port link bw %02x lane " + "count %d clock %d\n", + intel_dp->link_bw, intel_dp->lane_count, + adjusted_mode->clock); + + return true; + } + + return false; +} + +struct intel_dp_m_n { + uint32_t tu; + uint32_t gmch_m; + uint32_t gmch_n; + uint32_t link_m; + uint32_t link_n; +}; + +static void +intel_reduce_ratio(uint32_t *num, uint32_t *den) +{ + while (*num > 0xffffff || *den > 0xffffff) { + *num >>= 1; + *den >>= 1; + } +} + +static void +intel_dp_compute_m_n(int bpp, + int nlanes, + int pixel_clock, + int link_clock, + struct intel_dp_m_n *m_n) +{ + m_n->tu = 64; + m_n->gmch_m = (pixel_clock * bpp) >> 3; + m_n->gmch_n = link_clock * nlanes; + intel_reduce_ratio(&m_n->gmch_m, &m_n->gmch_n); + m_n->link_m = pixel_clock; + m_n->link_n = link_clock; + intel_reduce_ratio(&m_n->link_m, &m_n->link_n); +} + +void +intel_dp_set_m_n(struct drm_crtc *crtc, struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct drm_device *dev = crtc->dev; + struct drm_mode_config *mode_config = &dev->mode_config; + struct drm_encoder *encoder; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int lane_count = 4, bpp = 24; + struct intel_dp_m_n m_n; + int pipe = intel_crtc->pipe; + + /* + * Find the lane count in the intel_encoder private + */ + list_for_each_entry(encoder, &mode_config->encoder_list, head) { + struct intel_dp *intel_dp; + + if (encoder->crtc != crtc) + continue; + + intel_dp = enc_to_intel_dp(encoder); + if (intel_dp->base.type == INTEL_OUTPUT_DISPLAYPORT) { + lane_count = intel_dp->lane_count; + break; + } else if (is_edp(intel_dp)) { + lane_count = dev_priv->edp.lanes; + bpp = dev_priv->edp.bpp; + break; + } + } + + /* + * Compute the GMCH and Link ratios. The '3' here is + * the number of bytes_per_pixel post-LUT, which we always + * set up for 8-bits of R/G/B, or 3 bytes total. + */ + intel_dp_compute_m_n(bpp, lane_count, + mode->clock, adjusted_mode->clock, &m_n); + + if (HAS_PCH_SPLIT(dev)) { + I915_WRITE(TRANSDATA_M1(pipe), + ((m_n.tu - 1) << PIPE_GMCH_DATA_M_TU_SIZE_SHIFT) | + m_n.gmch_m); + I915_WRITE(TRANSDATA_N1(pipe), m_n.gmch_n); + I915_WRITE(TRANSDPLINK_M1(pipe), m_n.link_m); + I915_WRITE(TRANSDPLINK_N1(pipe), m_n.link_n); + } else { + I915_WRITE(PIPE_GMCH_DATA_M(pipe), + ((m_n.tu - 1) << PIPE_GMCH_DATA_M_TU_SIZE_SHIFT) | + m_n.gmch_m); + I915_WRITE(PIPE_GMCH_DATA_N(pipe), m_n.gmch_n); + I915_WRITE(PIPE_DP_LINK_M(pipe), m_n.link_m); + I915_WRITE(PIPE_DP_LINK_N(pipe), m_n.link_n); + } +} + +static void +intel_dp_mode_set(struct drm_encoder *encoder, struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct drm_device *dev = encoder->dev; + struct intel_dp *intel_dp = enc_to_intel_dp(encoder); + struct drm_crtc *crtc = intel_dp->base.base.crtc; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + + intel_dp->DP = DP_VOLTAGE_0_4 | DP_PRE_EMPHASIS_0; + intel_dp->DP |= intel_dp->color_range; + + if (adjusted_mode->flags & DRM_MODE_FLAG_PHSYNC) + intel_dp->DP |= DP_SYNC_HS_HIGH; + if (adjusted_mode->flags & DRM_MODE_FLAG_PVSYNC) + intel_dp->DP |= DP_SYNC_VS_HIGH; + + if (HAS_PCH_CPT(dev) && !is_edp(intel_dp)) + intel_dp->DP |= DP_LINK_TRAIN_OFF_CPT; + else + intel_dp->DP |= DP_LINK_TRAIN_OFF; + + switch (intel_dp->lane_count) { + case 1: + intel_dp->DP |= DP_PORT_WIDTH_1; + break; + case 2: + intel_dp->DP |= DP_PORT_WIDTH_2; + break; + case 4: + intel_dp->DP |= DP_PORT_WIDTH_4; + break; + } + if (intel_dp->has_audio) + intel_dp->DP |= DP_AUDIO_OUTPUT_ENABLE; + + memset(intel_dp->link_configuration, 0, DP_LINK_CONFIGURATION_SIZE); + intel_dp->link_configuration[0] = intel_dp->link_bw; + intel_dp->link_configuration[1] = intel_dp->lane_count; + + /* + * Check for DPCD version > 1.1 and enhanced framing support + */ + if (intel_dp->dpcd[DP_DPCD_REV] >= 0x11 && + (intel_dp->dpcd[DP_MAX_LANE_COUNT] & DP_ENHANCED_FRAME_CAP)) { + intel_dp->link_configuration[1] |= DP_LANE_COUNT_ENHANCED_FRAME_EN; + intel_dp->DP |= DP_ENHANCED_FRAMING; + } + + /* CPT DP's pipe select is decided in TRANS_DP_CTL */ + if (intel_crtc->pipe == 1 && !HAS_PCH_CPT(dev)) + intel_dp->DP |= DP_PIPEB_SELECT; + + if (is_edp(intel_dp) && !is_pch_edp(intel_dp)) { + /* don't miss out required setting for eDP */ + intel_dp->DP |= DP_PLL_ENABLE; + if (adjusted_mode->clock < 200000) + intel_dp->DP |= DP_PLL_FREQ_160MHZ; + else + intel_dp->DP |= DP_PLL_FREQ_270MHZ; + } +} + +static void ironlake_edp_panel_vdd_on(struct intel_dp *intel_dp) +{ + struct drm_device *dev = intel_dp->base.base.dev; + struct drm_i915_private *dev_priv = dev->dev_private; + u32 pp; + + /* + * If the panel wasn't on, make sure there's not a currently + * active PP sequence before enabling AUX VDD. + */ + if (!(I915_READ(PCH_PP_STATUS) & PP_ON)) + msleep(dev_priv->panel_t3); + + pp = I915_READ(PCH_PP_CONTROL); + pp |= EDP_FORCE_VDD; + I915_WRITE(PCH_PP_CONTROL, pp); + POSTING_READ(PCH_PP_CONTROL); +} + +static void ironlake_edp_panel_vdd_off(struct intel_dp *intel_dp) +{ + struct drm_device *dev = intel_dp->base.base.dev; + struct drm_i915_private *dev_priv = dev->dev_private; + u32 pp; + + pp = I915_READ(PCH_PP_CONTROL); + pp &= ~EDP_FORCE_VDD; + I915_WRITE(PCH_PP_CONTROL, pp); + POSTING_READ(PCH_PP_CONTROL); + + /* Make sure sequencer is idle before allowing subsequent activity */ + msleep(dev_priv->panel_t12); +} + +/* Returns true if the panel was already on when called */ +static bool ironlake_edp_panel_on (struct intel_dp *intel_dp) +{ + struct drm_device *dev = intel_dp->base.base.dev; + struct drm_i915_private *dev_priv = dev->dev_private; + u32 pp, idle_on_mask = PP_ON | PP_SEQUENCE_STATE_ON_IDLE; + + if (I915_READ(PCH_PP_STATUS) & PP_ON) + return true; + + pp = I915_READ(PCH_PP_CONTROL); + + /* ILK workaround: disable reset around power sequence */ + pp &= ~PANEL_POWER_RESET; + I915_WRITE(PCH_PP_CONTROL, pp); + POSTING_READ(PCH_PP_CONTROL); + + pp |= PANEL_UNLOCK_REGS | POWER_TARGET_ON; + I915_WRITE(PCH_PP_CONTROL, pp); + POSTING_READ(PCH_PP_CONTROL); + + if (wait_for((I915_READ(PCH_PP_STATUS) & idle_on_mask) == idle_on_mask, + 5000)) + DRM_ERROR("panel on wait timed out: 0x%08x\n", + I915_READ(PCH_PP_STATUS)); + + pp |= PANEL_POWER_RESET; /* restore panel reset bit */ + I915_WRITE(PCH_PP_CONTROL, pp); + POSTING_READ(PCH_PP_CONTROL); + + return false; +} + +static void ironlake_edp_panel_off (struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u32 pp, idle_off_mask = PP_ON | PP_SEQUENCE_MASK | + PP_CYCLE_DELAY_ACTIVE | PP_SEQUENCE_STATE_MASK; + + pp = I915_READ(PCH_PP_CONTROL); + + /* ILK workaround: disable reset around power sequence */ + pp &= ~PANEL_POWER_RESET; + I915_WRITE(PCH_PP_CONTROL, pp); + POSTING_READ(PCH_PP_CONTROL); + + pp &= ~POWER_TARGET_ON; + I915_WRITE(PCH_PP_CONTROL, pp); + POSTING_READ(PCH_PP_CONTROL); + + if (wait_for((I915_READ(PCH_PP_STATUS) & idle_off_mask) == 0, 5000)) + DRM_ERROR("panel off wait timed out: 0x%08x\n", + I915_READ(PCH_PP_STATUS)); + + pp |= PANEL_POWER_RESET; /* restore panel reset bit */ + I915_WRITE(PCH_PP_CONTROL, pp); + POSTING_READ(PCH_PP_CONTROL); +} + +static void ironlake_edp_backlight_on (struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u32 pp; + + DRM_DEBUG_KMS("\n"); + /* + * If we enable the backlight right away following a panel power + * on, we may see slight flicker as the panel syncs with the eDP + * link. So delay a bit to make sure the image is solid before + * allowing it to appear. + */ + msleep(300); + pp = I915_READ(PCH_PP_CONTROL); + pp |= EDP_BLC_ENABLE; + I915_WRITE(PCH_PP_CONTROL, pp); +} + +static void ironlake_edp_backlight_off (struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u32 pp; + + DRM_DEBUG_KMS("\n"); + pp = I915_READ(PCH_PP_CONTROL); + pp &= ~EDP_BLC_ENABLE; + I915_WRITE(PCH_PP_CONTROL, pp); +} + +static void ironlake_edp_pll_on(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + u32 dpa_ctl; + + DRM_DEBUG_KMS("\n"); + dpa_ctl = I915_READ(DP_A); + dpa_ctl |= DP_PLL_ENABLE; + I915_WRITE(DP_A, dpa_ctl); + POSTING_READ(DP_A); + udelay(200); +} + +static void ironlake_edp_pll_off(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + u32 dpa_ctl; + + dpa_ctl = I915_READ(DP_A); + dpa_ctl &= ~DP_PLL_ENABLE; + I915_WRITE(DP_A, dpa_ctl); + POSTING_READ(DP_A); + udelay(200); +} + +/* If the sink supports it, try to set the power state appropriately */ +static void intel_dp_sink_dpms(struct intel_dp *intel_dp, int mode) +{ + int ret, i; + + /* Should have a valid DPCD by this point */ + if (intel_dp->dpcd[DP_DPCD_REV] < 0x11) + return; + + if (mode != DRM_MODE_DPMS_ON) { + ret = intel_dp_aux_native_write_1(intel_dp, DP_SET_POWER, + DP_SET_POWER_D3); + if (ret != 1) + DRM_DEBUG_DRIVER("failed to write sink power state\n"); + } else { + /* + * When turning on, we need to retry for 1ms to give the sink + * time to wake up. + */ + for (i = 0; i < 3; i++) { + ret = intel_dp_aux_native_write_1(intel_dp, + DP_SET_POWER, + DP_SET_POWER_D0); + if (ret == 1) + break; + msleep(1); + } + } +} + +static void intel_dp_prepare(struct drm_encoder *encoder) +{ + struct intel_dp *intel_dp = enc_to_intel_dp(encoder); + struct drm_device *dev = encoder->dev; + + /* Wake up the sink first */ + intel_dp_sink_dpms(intel_dp, DRM_MODE_DPMS_ON); + + if (is_edp(intel_dp)) { + ironlake_edp_backlight_off(dev); + ironlake_edp_panel_off(dev); + if (!is_pch_edp(intel_dp)) + ironlake_edp_pll_on(encoder); + else + ironlake_edp_pll_off(encoder); + } + intel_dp_link_down(intel_dp); +} + +static void intel_dp_commit(struct drm_encoder *encoder) +{ + struct intel_dp *intel_dp = enc_to_intel_dp(encoder); + struct drm_device *dev = encoder->dev; + + if (is_edp(intel_dp)) + ironlake_edp_panel_vdd_on(intel_dp); + + intel_dp_start_link_train(intel_dp); + + if (is_edp(intel_dp)) { + ironlake_edp_panel_on(intel_dp); + ironlake_edp_panel_vdd_off(intel_dp); + } + + intel_dp_complete_link_train(intel_dp); + + if (is_edp(intel_dp)) + ironlake_edp_backlight_on(dev); +} + +static void +intel_dp_dpms(struct drm_encoder *encoder, int mode) +{ + struct intel_dp *intel_dp = enc_to_intel_dp(encoder); + struct drm_device *dev = encoder->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + uint32_t dp_reg = I915_READ(intel_dp->output_reg); + + if (mode != DRM_MODE_DPMS_ON) { + if (is_edp(intel_dp)) + ironlake_edp_backlight_off(dev); + intel_dp_sink_dpms(intel_dp, mode); + intel_dp_link_down(intel_dp); + if (is_edp(intel_dp)) + ironlake_edp_panel_off(dev); + if (is_edp(intel_dp) && !is_pch_edp(intel_dp)) + ironlake_edp_pll_off(encoder); + } else { + if (is_edp(intel_dp)) + ironlake_edp_panel_vdd_on(intel_dp); + intel_dp_sink_dpms(intel_dp, mode); + if (!(dp_reg & DP_PORT_EN)) { + intel_dp_start_link_train(intel_dp); + if (is_edp(intel_dp)) { + ironlake_edp_panel_on(intel_dp); + ironlake_edp_panel_vdd_off(intel_dp); + } + intel_dp_complete_link_train(intel_dp); + } + if (is_edp(intel_dp)) + ironlake_edp_backlight_on(dev); + } +} + +/* + * Native read with retry for link status and receiver capability reads for + * cases where the sink may still be asleep. + */ +static bool +intel_dp_aux_native_read_retry(struct intel_dp *intel_dp, uint16_t address, + uint8_t *recv, int recv_bytes) +{ + int ret, i; + + /* + * Sinks are *supposed* to come up within 1ms from an off state, + * but we're also supposed to retry 3 times per the spec. + */ + for (i = 0; i < 3; i++) { + ret = intel_dp_aux_native_read(intel_dp, address, recv, + recv_bytes); + if (ret == recv_bytes) + return true; + msleep(1); + } + + return false; +} + +/* + * Fetch AUX CH registers 0x202 - 0x207 which contain + * link status information + */ +static bool +intel_dp_get_link_status(struct intel_dp *intel_dp) +{ + return intel_dp_aux_native_read_retry(intel_dp, + DP_LANE0_1_STATUS, + intel_dp->link_status, + DP_LINK_STATUS_SIZE); +} + +static uint8_t +intel_dp_link_status(uint8_t link_status[DP_LINK_STATUS_SIZE], + int r) +{ + return link_status[r - DP_LANE0_1_STATUS]; +} + +static uint8_t +intel_get_adjust_request_voltage(uint8_t link_status[DP_LINK_STATUS_SIZE], + int lane) +{ + int i = DP_ADJUST_REQUEST_LANE0_1 + (lane >> 1); + int s = ((lane & 1) ? + DP_ADJUST_VOLTAGE_SWING_LANE1_SHIFT : + DP_ADJUST_VOLTAGE_SWING_LANE0_SHIFT); + uint8_t l = intel_dp_link_status(link_status, i); + + return ((l >> s) & 3) << DP_TRAIN_VOLTAGE_SWING_SHIFT; +} + +static uint8_t +intel_get_adjust_request_pre_emphasis(uint8_t link_status[DP_LINK_STATUS_SIZE], + int lane) +{ + int i = DP_ADJUST_REQUEST_LANE0_1 + (lane >> 1); + int s = ((lane & 1) ? + DP_ADJUST_PRE_EMPHASIS_LANE1_SHIFT : + DP_ADJUST_PRE_EMPHASIS_LANE0_SHIFT); + uint8_t l = intel_dp_link_status(link_status, i); + + return ((l >> s) & 3) << DP_TRAIN_PRE_EMPHASIS_SHIFT; +} + + +#if 0 +static char *voltage_names[] = { + "0.4V", "0.6V", "0.8V", "1.2V" +}; +static char *pre_emph_names[] = { + "0dB", "3.5dB", "6dB", "9.5dB" +}; +static char *link_train_names[] = { + "pattern 1", "pattern 2", "idle", "off" +}; +#endif + +/* + * These are source-specific values; current Intel hardware supports + * a maximum voltage of 800mV and a maximum pre-emphasis of 6dB + */ +#define I830_DP_VOLTAGE_MAX DP_TRAIN_VOLTAGE_SWING_800 + +static uint8_t +intel_dp_pre_emphasis_max(uint8_t voltage_swing) +{ + switch (voltage_swing & DP_TRAIN_VOLTAGE_SWING_MASK) { + case DP_TRAIN_VOLTAGE_SWING_400: + return DP_TRAIN_PRE_EMPHASIS_6; + case DP_TRAIN_VOLTAGE_SWING_600: + return DP_TRAIN_PRE_EMPHASIS_6; + case DP_TRAIN_VOLTAGE_SWING_800: + return DP_TRAIN_PRE_EMPHASIS_3_5; + case DP_TRAIN_VOLTAGE_SWING_1200: + default: + return DP_TRAIN_PRE_EMPHASIS_0; + } +} + +static void +intel_get_adjust_train(struct intel_dp *intel_dp) +{ + uint8_t v = 0; + uint8_t p = 0; + int lane; + + for (lane = 0; lane < intel_dp->lane_count; lane++) { + uint8_t this_v = intel_get_adjust_request_voltage(intel_dp->link_status, lane); + uint8_t this_p = intel_get_adjust_request_pre_emphasis(intel_dp->link_status, lane); + + if (this_v > v) + v = this_v; + if (this_p > p) + p = this_p; + } + + if (v >= I830_DP_VOLTAGE_MAX) + v = I830_DP_VOLTAGE_MAX | DP_TRAIN_MAX_SWING_REACHED; + + if (p >= intel_dp_pre_emphasis_max(v)) + p = intel_dp_pre_emphasis_max(v) | DP_TRAIN_MAX_PRE_EMPHASIS_REACHED; + + for (lane = 0; lane < 4; lane++) + intel_dp->train_set[lane] = v | p; +} + +static uint32_t +intel_dp_signal_levels(uint8_t train_set, int lane_count) +{ + uint32_t signal_levels = 0; + + switch (train_set & DP_TRAIN_VOLTAGE_SWING_MASK) { + case DP_TRAIN_VOLTAGE_SWING_400: + default: + signal_levels |= DP_VOLTAGE_0_4; + break; + case DP_TRAIN_VOLTAGE_SWING_600: + signal_levels |= DP_VOLTAGE_0_6; + break; + case DP_TRAIN_VOLTAGE_SWING_800: + signal_levels |= DP_VOLTAGE_0_8; + break; + case DP_TRAIN_VOLTAGE_SWING_1200: + signal_levels |= DP_VOLTAGE_1_2; + break; + } + switch (train_set & DP_TRAIN_PRE_EMPHASIS_MASK) { + case DP_TRAIN_PRE_EMPHASIS_0: + default: + signal_levels |= DP_PRE_EMPHASIS_0; + break; + case DP_TRAIN_PRE_EMPHASIS_3_5: + signal_levels |= DP_PRE_EMPHASIS_3_5; + break; + case DP_TRAIN_PRE_EMPHASIS_6: + signal_levels |= DP_PRE_EMPHASIS_6; + break; + case DP_TRAIN_PRE_EMPHASIS_9_5: + signal_levels |= DP_PRE_EMPHASIS_9_5; + break; + } + return signal_levels; +} + +/* Gen6's DP voltage swing and pre-emphasis control */ +static uint32_t +intel_gen6_edp_signal_levels(uint8_t train_set) +{ + int signal_levels = train_set & (DP_TRAIN_VOLTAGE_SWING_MASK | + DP_TRAIN_PRE_EMPHASIS_MASK); + switch (signal_levels) { + case DP_TRAIN_VOLTAGE_SWING_400 | DP_TRAIN_PRE_EMPHASIS_0: + case DP_TRAIN_VOLTAGE_SWING_600 | DP_TRAIN_PRE_EMPHASIS_0: + return EDP_LINK_TRAIN_400_600MV_0DB_SNB_B; + case DP_TRAIN_VOLTAGE_SWING_400 | DP_TRAIN_PRE_EMPHASIS_3_5: + return EDP_LINK_TRAIN_400MV_3_5DB_SNB_B; + case DP_TRAIN_VOLTAGE_SWING_400 | DP_TRAIN_PRE_EMPHASIS_6: + case DP_TRAIN_VOLTAGE_SWING_600 | DP_TRAIN_PRE_EMPHASIS_6: + return EDP_LINK_TRAIN_400_600MV_6DB_SNB_B; + case DP_TRAIN_VOLTAGE_SWING_600 | DP_TRAIN_PRE_EMPHASIS_3_5: + case DP_TRAIN_VOLTAGE_SWING_800 | DP_TRAIN_PRE_EMPHASIS_3_5: + return EDP_LINK_TRAIN_600_800MV_3_5DB_SNB_B; + case DP_TRAIN_VOLTAGE_SWING_800 | DP_TRAIN_PRE_EMPHASIS_0: + case DP_TRAIN_VOLTAGE_SWING_1200 | DP_TRAIN_PRE_EMPHASIS_0: + return EDP_LINK_TRAIN_800_1200MV_0DB_SNB_B; + default: + DRM_DEBUG_KMS("Unsupported voltage swing/pre-emphasis level:" + "0x%x\n", signal_levels); + return EDP_LINK_TRAIN_400_600MV_0DB_SNB_B; + } +} + +static uint8_t +intel_get_lane_status(uint8_t link_status[DP_LINK_STATUS_SIZE], + int lane) +{ + int i = DP_LANE0_1_STATUS + (lane >> 1); + int s = (lane & 1) * 4; + uint8_t l = intel_dp_link_status(link_status, i); + + return (l >> s) & 0xf; +} + +/* Check for clock recovery is done on all channels */ +static bool +intel_clock_recovery_ok(uint8_t link_status[DP_LINK_STATUS_SIZE], int lane_count) +{ + int lane; + uint8_t lane_status; + + for (lane = 0; lane < lane_count; lane++) { + lane_status = intel_get_lane_status(link_status, lane); + if ((lane_status & DP_LANE_CR_DONE) == 0) + return false; + } + return true; +} + +/* Check to see if channel eq is done on all channels */ +#define CHANNEL_EQ_BITS (DP_LANE_CR_DONE|\ + DP_LANE_CHANNEL_EQ_DONE|\ + DP_LANE_SYMBOL_LOCKED) +static bool +intel_channel_eq_ok(struct intel_dp *intel_dp) +{ + uint8_t lane_align; + uint8_t lane_status; + int lane; + + lane_align = intel_dp_link_status(intel_dp->link_status, + DP_LANE_ALIGN_STATUS_UPDATED); + if ((lane_align & DP_INTERLANE_ALIGN_DONE) == 0) + return false; + for (lane = 0; lane < intel_dp->lane_count; lane++) { + lane_status = intel_get_lane_status(intel_dp->link_status, lane); + if ((lane_status & CHANNEL_EQ_BITS) != CHANNEL_EQ_BITS) + return false; + } + return true; +} + +static bool +intel_dp_set_link_train(struct intel_dp *intel_dp, + uint32_t dp_reg_value, + uint8_t dp_train_pat) +{ + struct drm_device *dev = intel_dp->base.base.dev; + struct drm_i915_private *dev_priv = dev->dev_private; + int ret; + + I915_WRITE(intel_dp->output_reg, dp_reg_value); + POSTING_READ(intel_dp->output_reg); + + intel_dp_aux_native_write_1(intel_dp, + DP_TRAINING_PATTERN_SET, + dp_train_pat); + + ret = intel_dp_aux_native_write(intel_dp, + DP_TRAINING_LANE0_SET, + intel_dp->train_set, 4); + if (ret != 4) + return false; + + return true; +} + +/* Enable corresponding port and start training pattern 1 */ +static void +intel_dp_start_link_train(struct intel_dp *intel_dp) +{ + struct drm_device *dev = intel_dp->base.base.dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(intel_dp->base.base.crtc); + int i; + uint8_t voltage; + bool clock_recovery = false; + int tries; + u32 reg; + uint32_t DP = intel_dp->DP; + + /* Enable output, wait for it to become active */ + I915_WRITE(intel_dp->output_reg, intel_dp->DP); + POSTING_READ(intel_dp->output_reg); + intel_wait_for_vblank(dev, intel_crtc->pipe); + + /* Write the link configuration data */ + intel_dp_aux_native_write(intel_dp, DP_LINK_BW_SET, + intel_dp->link_configuration, + DP_LINK_CONFIGURATION_SIZE); + + DP |= DP_PORT_EN; + if (HAS_PCH_CPT(dev) && !is_edp(intel_dp)) + DP &= ~DP_LINK_TRAIN_MASK_CPT; + else + DP &= ~DP_LINK_TRAIN_MASK; + memset(intel_dp->train_set, 0, 4); + voltage = 0xff; + tries = 0; + clock_recovery = false; + for (;;) { + /* Use intel_dp->train_set[0] to set the voltage and pre emphasis values */ + uint32_t signal_levels; + if (IS_GEN6(dev) && is_edp(intel_dp)) { + signal_levels = intel_gen6_edp_signal_levels(intel_dp->train_set[0]); + DP = (DP & ~EDP_LINK_TRAIN_VOL_EMP_MASK_SNB) | signal_levels; + } else { + signal_levels = intel_dp_signal_levels(intel_dp->train_set[0], intel_dp->lane_count); + DP = (DP & ~(DP_VOLTAGE_MASK|DP_PRE_EMPHASIS_MASK)) | signal_levels; + } + + if (HAS_PCH_CPT(dev) && !is_edp(intel_dp)) + reg = DP | DP_LINK_TRAIN_PAT_1_CPT; + else + reg = DP | DP_LINK_TRAIN_PAT_1; + + if (!intel_dp_set_link_train(intel_dp, reg, + DP_TRAINING_PATTERN_1)) + break; + /* Set training pattern 1 */ + + udelay(100); + if (!intel_dp_get_link_status(intel_dp)) + break; + + if (intel_clock_recovery_ok(intel_dp->link_status, intel_dp->lane_count)) { + clock_recovery = true; + break; + } + + /* Check to see if we've tried the max voltage */ + for (i = 0; i < intel_dp->lane_count; i++) + if ((intel_dp->train_set[i] & DP_TRAIN_MAX_SWING_REACHED) == 0) + break; + if (i == intel_dp->lane_count) + break; + + /* Check to see if we've tried the same voltage 5 times */ + if ((intel_dp->train_set[0] & DP_TRAIN_VOLTAGE_SWING_MASK) == voltage) { + ++tries; + if (tries == 5) + break; + } else + tries = 0; + voltage = intel_dp->train_set[0] & DP_TRAIN_VOLTAGE_SWING_MASK; + + /* Compute new intel_dp->train_set as requested by target */ + intel_get_adjust_train(intel_dp); + } + + intel_dp->DP = DP; +} + +static void +intel_dp_complete_link_train(struct intel_dp *intel_dp) +{ + struct drm_device *dev = intel_dp->base.base.dev; + struct drm_i915_private *dev_priv = dev->dev_private; + bool channel_eq = false; + int tries, cr_tries; + u32 reg; + uint32_t DP = intel_dp->DP; + + /* channel equalization */ + tries = 0; + cr_tries = 0; + channel_eq = false; + for (;;) { + /* Use intel_dp->train_set[0] to set the voltage and pre emphasis values */ + uint32_t signal_levels; + + if (cr_tries > 5) { + DRM_ERROR("failed to train DP, aborting\n"); + intel_dp_link_down(intel_dp); + break; + } + + if (IS_GEN6(dev) && is_edp(intel_dp)) { + signal_levels = intel_gen6_edp_signal_levels(intel_dp->train_set[0]); + DP = (DP & ~EDP_LINK_TRAIN_VOL_EMP_MASK_SNB) | signal_levels; + } else { + signal_levels = intel_dp_signal_levels(intel_dp->train_set[0], intel_dp->lane_count); + DP = (DP & ~(DP_VOLTAGE_MASK|DP_PRE_EMPHASIS_MASK)) | signal_levels; + } + + if (HAS_PCH_CPT(dev) && !is_edp(intel_dp)) + reg = DP | DP_LINK_TRAIN_PAT_2_CPT; + else + reg = DP | DP_LINK_TRAIN_PAT_2; + + /* channel eq pattern */ + if (!intel_dp_set_link_train(intel_dp, reg, + DP_TRAINING_PATTERN_2)) + break; + + udelay(400); + if (!intel_dp_get_link_status(intel_dp)) + break; + + /* Make sure clock is still ok */ + if (!intel_clock_recovery_ok(intel_dp->link_status, intel_dp->lane_count)) { + intel_dp_start_link_train(intel_dp); + cr_tries++; + continue; + } + + if (intel_channel_eq_ok(intel_dp)) { + channel_eq = true; + break; + } + + /* Try 5 times, then try clock recovery if that fails */ + if (tries > 5) { + intel_dp_link_down(intel_dp); + intel_dp_start_link_train(intel_dp); + tries = 0; + cr_tries++; + continue; + } + + /* Compute new intel_dp->train_set as requested by target */ + intel_get_adjust_train(intel_dp); + ++tries; + } + + if (HAS_PCH_CPT(dev) && !is_edp(intel_dp)) + reg = DP | DP_LINK_TRAIN_OFF_CPT; + else + reg = DP | DP_LINK_TRAIN_OFF; + + I915_WRITE(intel_dp->output_reg, reg); + POSTING_READ(intel_dp->output_reg); + intel_dp_aux_native_write_1(intel_dp, + DP_TRAINING_PATTERN_SET, DP_TRAINING_PATTERN_DISABLE); +} + +static void +intel_dp_link_down(struct intel_dp *intel_dp) +{ + struct drm_device *dev = intel_dp->base.base.dev; + struct drm_i915_private *dev_priv = dev->dev_private; + uint32_t DP = intel_dp->DP; + + if ((I915_READ(intel_dp->output_reg) & DP_PORT_EN) == 0) + return; + + DRM_DEBUG_KMS("\n"); + + if (is_edp(intel_dp)) { + DP &= ~DP_PLL_ENABLE; + I915_WRITE(intel_dp->output_reg, DP); + POSTING_READ(intel_dp->output_reg); + udelay(100); + } + + if (HAS_PCH_CPT(dev) && !is_edp(intel_dp)) { + DP &= ~DP_LINK_TRAIN_MASK_CPT; + I915_WRITE(intel_dp->output_reg, DP | DP_LINK_TRAIN_PAT_IDLE_CPT); + } else { + DP &= ~DP_LINK_TRAIN_MASK; + I915_WRITE(intel_dp->output_reg, DP | DP_LINK_TRAIN_PAT_IDLE); + } + POSTING_READ(intel_dp->output_reg); + + msleep(17); + + if (is_edp(intel_dp)) + DP |= DP_LINK_TRAIN_OFF; + + if (!HAS_PCH_CPT(dev) && + I915_READ(intel_dp->output_reg) & DP_PIPEB_SELECT) { + struct drm_crtc *crtc = intel_dp->base.base.crtc; + + /* Hardware workaround: leaving our transcoder select + * set to transcoder B while it's off will prevent the + * corresponding HDMI output on transcoder A. + * + * Combine this with another hardware workaround: + * transcoder select bit can only be cleared while the + * port is enabled. + */ + DP &= ~DP_PIPEB_SELECT; + I915_WRITE(intel_dp->output_reg, DP); + + /* Changes to enable or select take place the vblank + * after being written. + */ + if (crtc == NULL) { + /* We can arrive here never having been attached + * to a CRTC, for instance, due to inheriting + * random state from the BIOS. + * + * If the pipe is not running, play safe and + * wait for the clocks to stabilise before + * continuing. + */ + POSTING_READ(intel_dp->output_reg); + msleep(50); + } else + intel_wait_for_vblank(dev, to_intel_crtc(crtc)->pipe); + } + + DP &= ~DP_AUDIO_OUTPUT_ENABLE; + I915_WRITE(intel_dp->output_reg, DP & ~DP_PORT_EN); + POSTING_READ(intel_dp->output_reg); +} + +/* + * According to DP spec + * 5.1.2: + * 1. Read DPCD + * 2. Configure link according to Receiver Capabilities + * 3. Use Link Training from 2.5.3.3 and 3.5.1.3 + * 4. Check link status on receipt of hot-plug interrupt + */ + +static void +intel_dp_check_link_status(struct intel_dp *intel_dp) +{ + int ret; + + if (!intel_dp->base.base.crtc) + return; + + if (!intel_dp_get_link_status(intel_dp)) { + intel_dp_link_down(intel_dp); + return; + } + + /* Try to read receiver status if the link appears to be up */ + ret = intel_dp_aux_native_read(intel_dp, + 0x000, intel_dp->dpcd, + sizeof (intel_dp->dpcd)); + if (ret != sizeof(intel_dp->dpcd)) { + intel_dp_link_down(intel_dp); + return; + } + + if (!intel_channel_eq_ok(intel_dp)) { + intel_dp_start_link_train(intel_dp); + intel_dp_complete_link_train(intel_dp); + } +} + +static enum drm_connector_status +ironlake_dp_detect(struct intel_dp *intel_dp) +{ + enum drm_connector_status status; + bool ret; + + /* Can't disconnect eDP, but you can close the lid... */ + if (is_edp(intel_dp)) { + status = intel_panel_detect(intel_dp->base.base.dev); + if (status == connector_status_unknown) + status = connector_status_connected; + return status; + } + + status = connector_status_disconnected; + ret = intel_dp_aux_native_read_retry(intel_dp, + 0x000, intel_dp->dpcd, + sizeof (intel_dp->dpcd)); + if (ret && intel_dp->dpcd[DP_DPCD_REV] != 0) + status = connector_status_connected; + DRM_DEBUG_KMS("DPCD: %hx%hx%hx%hx\n", intel_dp->dpcd[0], + intel_dp->dpcd[1], intel_dp->dpcd[2], intel_dp->dpcd[3]); + return status; +} + +static enum drm_connector_status +g4x_dp_detect(struct intel_dp *intel_dp) +{ + struct drm_device *dev = intel_dp->base.base.dev; + struct drm_i915_private *dev_priv = dev->dev_private; + enum drm_connector_status status; + uint32_t temp, bit; + + switch (intel_dp->output_reg) { + case DP_B: + bit = DPB_HOTPLUG_INT_STATUS; + break; + case DP_C: + bit = DPC_HOTPLUG_INT_STATUS; + break; + case DP_D: + bit = DPD_HOTPLUG_INT_STATUS; + break; + default: + return connector_status_unknown; + } + + temp = I915_READ(PORT_HOTPLUG_STAT); + + if ((temp & bit) == 0) + return connector_status_disconnected; + + status = connector_status_disconnected; + if (intel_dp_aux_native_read(intel_dp, 0x000, intel_dp->dpcd, + sizeof (intel_dp->dpcd)) == sizeof (intel_dp->dpcd)) + { + if (intel_dp->dpcd[DP_DPCD_REV] != 0) + status = connector_status_connected; + } + + return status; +} + +static struct edid * +intel_dp_get_edid(struct drm_connector *connector, struct i2c_adapter *adapter) +{ + struct intel_dp *intel_dp = intel_attached_dp(connector); + struct edid *edid; + + ironlake_edp_panel_vdd_on(intel_dp); + edid = drm_get_edid(connector, adapter); + ironlake_edp_panel_vdd_off(intel_dp); + return edid; +} + +static int +intel_dp_get_edid_modes(struct drm_connector *connector, struct i2c_adapter *adapter) +{ + struct intel_dp *intel_dp = intel_attached_dp(connector); + int ret; + + ironlake_edp_panel_vdd_on(intel_dp); + ret = intel_ddc_get_modes(connector, adapter); + ironlake_edp_panel_vdd_off(intel_dp); + return ret; +} + + +/** + * Uses CRT_HOTPLUG_EN and CRT_HOTPLUG_STAT to detect DP connection. + * + * \return true if DP port is connected. + * \return false if DP port is disconnected. + */ +static enum drm_connector_status +intel_dp_detect(struct drm_connector *connector, bool force) +{ + struct intel_dp *intel_dp = intel_attached_dp(connector); + struct drm_device *dev = intel_dp->base.base.dev; + enum drm_connector_status status; + struct edid *edid = NULL; + + intel_dp->has_audio = false; + + if (HAS_PCH_SPLIT(dev)) + status = ironlake_dp_detect(intel_dp); + else + status = g4x_dp_detect(intel_dp); + if (status != connector_status_connected) + return status; + + if (intel_dp->force_audio) { + intel_dp->has_audio = intel_dp->force_audio > 0; + } else { + edid = intel_dp_get_edid(connector, &intel_dp->adapter); + if (edid) { + intel_dp->has_audio = drm_detect_monitor_audio(edid); + connector->display_info.raw_edid = NULL; + kfree(edid); + } + } + + return connector_status_connected; +} + +static int intel_dp_get_modes(struct drm_connector *connector) +{ + struct intel_dp *intel_dp = intel_attached_dp(connector); + struct drm_device *dev = intel_dp->base.base.dev; + struct drm_i915_private *dev_priv = dev->dev_private; + int ret; + + /* We should parse the EDID data and find out if it has an audio sink + */ + + ret = intel_dp_get_edid_modes(connector, &intel_dp->adapter); + if (ret) { + if (is_edp(intel_dp) && !dev_priv->panel_fixed_mode) { + struct drm_display_mode *newmode; + list_for_each_entry(newmode, &connector->probed_modes, + head) { + if (newmode->type & DRM_MODE_TYPE_PREFERRED) { + dev_priv->panel_fixed_mode = + drm_mode_duplicate(dev, newmode); + break; + } + } + } + + return ret; + } + + /* if eDP has no EDID, try to use fixed panel mode from VBT */ + if (is_edp(intel_dp)) { + if (dev_priv->panel_fixed_mode != NULL) { + struct drm_display_mode *mode; + mode = drm_mode_duplicate(dev, dev_priv->panel_fixed_mode); + drm_mode_probed_add(connector, mode); + return 1; + } + } + return 0; +} + +static bool +intel_dp_detect_audio(struct drm_connector *connector) +{ + struct intel_dp *intel_dp = intel_attached_dp(connector); + struct edid *edid; + bool has_audio = false; + + edid = intel_dp_get_edid(connector, &intel_dp->adapter); + if (edid) { + has_audio = drm_detect_monitor_audio(edid); + + connector->display_info.raw_edid = NULL; + kfree(edid); + } + + return has_audio; +} + +static int +intel_dp_set_property(struct drm_connector *connector, + struct drm_property *property, + uint64_t val) +{ + struct drm_i915_private *dev_priv = connector->dev->dev_private; + struct intel_dp *intel_dp = intel_attached_dp(connector); + int ret; + + ret = drm_connector_property_set_value(connector, property, val); + if (ret) + return ret; + + if (property == dev_priv->force_audio_property) { + int i = val; + bool has_audio; + + if (i == intel_dp->force_audio) + return 0; + + intel_dp->force_audio = i; + + if (i == 0) + has_audio = intel_dp_detect_audio(connector); + else + has_audio = i > 0; + + if (has_audio == intel_dp->has_audio) + return 0; + + intel_dp->has_audio = has_audio; + goto done; + } + + if (property == dev_priv->broadcast_rgb_property) { + if (val == !!intel_dp->color_range) + return 0; + + intel_dp->color_range = val ? DP_COLOR_RANGE_16_235 : 0; + goto done; + } + + return -EINVAL; + +done: + if (intel_dp->base.base.crtc) { + struct drm_crtc *crtc = intel_dp->base.base.crtc; + drm_crtc_helper_set_mode(crtc, &crtc->mode, + crtc->x, crtc->y, + crtc->fb); + } + + return 0; +} + +static void +intel_dp_destroy (struct drm_connector *connector) +{ + drm_sysfs_connector_remove(connector); + drm_connector_cleanup(connector); + kfree(connector); +} + +static void intel_dp_encoder_destroy(struct drm_encoder *encoder) +{ + struct intel_dp *intel_dp = enc_to_intel_dp(encoder); + + i2c_del_adapter(&intel_dp->adapter); + drm_encoder_cleanup(encoder); + kfree(intel_dp); +} + +static const struct drm_encoder_helper_funcs intel_dp_helper_funcs = { + .dpms = intel_dp_dpms, + .mode_fixup = intel_dp_mode_fixup, + .prepare = intel_dp_prepare, + .mode_set = intel_dp_mode_set, + .commit = intel_dp_commit, +}; + +static const struct drm_connector_funcs intel_dp_connector_funcs = { + .dpms = drm_helper_connector_dpms, + .detect = intel_dp_detect, + .fill_modes = drm_helper_probe_single_connector_modes, + .set_property = intel_dp_set_property, + .destroy = intel_dp_destroy, +}; + +static const struct drm_connector_helper_funcs intel_dp_connector_helper_funcs = { + .get_modes = intel_dp_get_modes, + .mode_valid = intel_dp_mode_valid, + .best_encoder = intel_best_encoder, +}; + +static const struct drm_encoder_funcs intel_dp_enc_funcs = { + .destroy = intel_dp_encoder_destroy, +}; + +static void +intel_dp_hot_plug(struct intel_encoder *intel_encoder) +{ + struct intel_dp *intel_dp = container_of(intel_encoder, struct intel_dp, base); + + intel_dp_check_link_status(intel_dp); +} + +/* Return which DP Port should be selected for Transcoder DP control */ +int +intel_trans_dp_port_sel (struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + struct drm_mode_config *mode_config = &dev->mode_config; + struct drm_encoder *encoder; + + list_for_each_entry(encoder, &mode_config->encoder_list, head) { + struct intel_dp *intel_dp; + + if (encoder->crtc != crtc) + continue; + + intel_dp = enc_to_intel_dp(encoder); + if (intel_dp->base.type == INTEL_OUTPUT_DISPLAYPORT) + return intel_dp->output_reg; + } + + return -1; +} + +/* check the VBT to see whether the eDP is on DP-D port */ +bool intel_dpd_is_edp(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct child_device_config *p_child; + int i; + + if (!dev_priv->child_dev_num) + return false; + + for (i = 0; i < dev_priv->child_dev_num; i++) { + p_child = dev_priv->child_dev + i; + + if (p_child->dvo_port == PORT_IDPD && + p_child->device_type == DEVICE_TYPE_eDP) + return true; + } + return false; +} + +static void +intel_dp_add_properties(struct intel_dp *intel_dp, struct drm_connector *connector) +{ + intel_attach_force_audio_property(connector); + intel_attach_broadcast_rgb_property(connector); +} + +void +intel_dp_init(struct drm_device *dev, int output_reg) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_connector *connector; + struct intel_dp *intel_dp; + struct intel_encoder *intel_encoder; + struct intel_connector *intel_connector; + const char *name = NULL; + int type; + + intel_dp = kzalloc(sizeof(struct intel_dp), GFP_KERNEL); + if (!intel_dp) + return; + + intel_dp->output_reg = output_reg; + + intel_connector = kzalloc(sizeof(struct intel_connector), GFP_KERNEL); + if (!intel_connector) { + kfree(intel_dp); + return; + } + intel_encoder = &intel_dp->base; + + if (HAS_PCH_SPLIT(dev) && output_reg == PCH_DP_D) + if (intel_dpd_is_edp(dev)) + intel_dp->is_pch_edp = true; + + if (output_reg == DP_A || is_pch_edp(intel_dp)) { + type = DRM_MODE_CONNECTOR_eDP; + intel_encoder->type = INTEL_OUTPUT_EDP; + } else { + type = DRM_MODE_CONNECTOR_DisplayPort; + intel_encoder->type = INTEL_OUTPUT_DISPLAYPORT; + } + + connector = &intel_connector->base; + drm_connector_init(dev, connector, &intel_dp_connector_funcs, type); + drm_connector_helper_add(connector, &intel_dp_connector_helper_funcs); + + connector->polled = DRM_CONNECTOR_POLL_HPD; + + if (output_reg == DP_B || output_reg == PCH_DP_B) + intel_encoder->clone_mask = (1 << INTEL_DP_B_CLONE_BIT); + else if (output_reg == DP_C || output_reg == PCH_DP_C) + intel_encoder->clone_mask = (1 << INTEL_DP_C_CLONE_BIT); + else if (output_reg == DP_D || output_reg == PCH_DP_D) + intel_encoder->clone_mask = (1 << INTEL_DP_D_CLONE_BIT); + + if (is_edp(intel_dp)) + intel_encoder->clone_mask = (1 << INTEL_EDP_CLONE_BIT); + + intel_encoder->crtc_mask = (1 << 0) | (1 << 1); + connector->interlace_allowed = true; + connector->doublescan_allowed = 0; + + drm_encoder_init(dev, &intel_encoder->base, &intel_dp_enc_funcs, + DRM_MODE_ENCODER_TMDS); + drm_encoder_helper_add(&intel_encoder->base, &intel_dp_helper_funcs); + + intel_connector_attach_encoder(intel_connector, intel_encoder); + drm_sysfs_connector_add(connector); + + /* Set up the DDC bus. */ + switch (output_reg) { + case DP_A: + name = "DPDDC-A"; + break; + case DP_B: + case PCH_DP_B: + dev_priv->hotplug_supported_mask |= + HDMIB_HOTPLUG_INT_STATUS; + name = "DPDDC-B"; + break; + case DP_C: + case PCH_DP_C: + dev_priv->hotplug_supported_mask |= + HDMIC_HOTPLUG_INT_STATUS; + name = "DPDDC-C"; + break; + case DP_D: + case PCH_DP_D: + dev_priv->hotplug_supported_mask |= + HDMID_HOTPLUG_INT_STATUS; + name = "DPDDC-D"; + break; + } + + intel_dp_i2c_init(intel_dp, intel_connector, name); + + /* Cache some DPCD data in the eDP case */ + if (is_edp(intel_dp)) { + int ret; + u32 pp_on, pp_div; + + pp_on = I915_READ(PCH_PP_ON_DELAYS); + pp_div = I915_READ(PCH_PP_DIVISOR); + + /* Get T3 & T12 values (note: VESA not bspec terminology) */ + dev_priv->panel_t3 = (pp_on & 0x1fff0000) >> 16; + dev_priv->panel_t3 /= 10; /* t3 in 100us units */ + dev_priv->panel_t12 = pp_div & 0xf; + dev_priv->panel_t12 *= 100; /* t12 in 100ms units */ + + ironlake_edp_panel_vdd_on(intel_dp); + ret = intel_dp_aux_native_read(intel_dp, DP_DPCD_REV, + intel_dp->dpcd, + sizeof(intel_dp->dpcd)); + ironlake_edp_panel_vdd_off(intel_dp); + if (ret == sizeof(intel_dp->dpcd)) { + if (intel_dp->dpcd[DP_DPCD_REV] >= 0x11) + dev_priv->no_aux_handshake = + intel_dp->dpcd[DP_MAX_DOWNSPREAD] & + DP_NO_AUX_HANDSHAKE_LINK_TRAINING; + } else { + /* if this fails, presume the device is a ghost */ + DRM_INFO("failed to retrieve link info, disabling eDP\n"); + intel_dp_encoder_destroy(&intel_dp->base.base); + intel_dp_destroy(&intel_connector->base); + return; + } + } + + intel_encoder->hot_plug = intel_dp_hot_plug; + + if (is_edp(intel_dp)) { + /* initialize panel mode from VBT if available for eDP */ + if (dev_priv->lfp_lvds_vbt_mode) { + dev_priv->panel_fixed_mode = + drm_mode_duplicate(dev, dev_priv->lfp_lvds_vbt_mode); + if (dev_priv->panel_fixed_mode) { + dev_priv->panel_fixed_mode->type |= + DRM_MODE_TYPE_PREFERRED; + } + } + } + + intel_dp_add_properties(intel_dp, connector); + + /* For G4X desktop chip, PEG_BAND_GAP_DATA 3:0 must first be written + * 0xd. Failure to do so will result in spurious interrupts being + * generated on the port when a cable is not attached. + */ + if (IS_G4X(dev) && !IS_GM45(dev)) { + u32 temp = I915_READ(PEG_BAND_GAP_DATA); + I915_WRITE(PEG_BAND_GAP_DATA, (temp & ~0xf) | 0xd); + } +} diff --git a/drivers/gpu/drm/i915/intel_drv.h b/drivers/gpu/drm/i915/intel_drv.h new file mode 100644 index 00000000..47bf5e15 --- /dev/null +++ b/drivers/gpu/drm/i915/intel_drv.h @@ -0,0 +1,350 @@ +/* + * Copyright (c) 2006 Dave Airlie + * Copyright (c) 2007-2008 Intel Corporation + * Jesse Barnes + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS + * IN THE SOFTWARE. + */ +#ifndef __INTEL_DRV_H__ +#define __INTEL_DRV_H__ + +#include +#include "i915_drv.h" +#include "drm_crtc.h" +#include "drm_crtc_helper.h" +#include "drm_fb_helper.h" + +#define _wait_for(COND, MS, W) ({ \ + unsigned long timeout__ = jiffies + msecs_to_jiffies(MS); \ + int ret__ = 0; \ + while (! (COND)) { \ + if (time_after(jiffies, timeout__)) { \ + ret__ = -ETIMEDOUT; \ + break; \ + } \ + if (W && !(in_atomic() || in_dbg_master())) msleep(W); \ + } \ + ret__; \ +}) + +#define wait_for(COND, MS) _wait_for(COND, MS, 1) +#define wait_for_atomic(COND, MS) _wait_for(COND, MS, 0) + +#define MSLEEP(x) do { \ + if (in_dbg_master()) \ + mdelay(x); \ + else \ + msleep(x); \ +} while(0) + +#define KHz(x) (1000*x) +#define MHz(x) KHz(1000*x) + +/* + * Display related stuff + */ + +/* store information about an Ixxx DVO */ +/* The i830->i865 use multiple DVOs with multiple i2cs */ +/* the i915, i945 have a single sDVO i2c bus - which is different */ +#define MAX_OUTPUTS 6 +/* maximum connectors per crtcs in the mode set */ +#define INTELFB_CONN_LIMIT 4 + +#define INTEL_I2C_BUS_DVO 1 +#define INTEL_I2C_BUS_SDVO 2 + +/* these are outputs from the chip - integrated only + external chips are via DVO or SDVO output */ +#define INTEL_OUTPUT_UNUSED 0 +#define INTEL_OUTPUT_ANALOG 1 +#define INTEL_OUTPUT_DVO 2 +#define INTEL_OUTPUT_SDVO 3 +#define INTEL_OUTPUT_LVDS 4 +#define INTEL_OUTPUT_TVOUT 5 +#define INTEL_OUTPUT_HDMI 6 +#define INTEL_OUTPUT_DISPLAYPORT 7 +#define INTEL_OUTPUT_EDP 8 + +/* Intel Pipe Clone Bit */ +#define INTEL_HDMIB_CLONE_BIT 1 +#define INTEL_HDMIC_CLONE_BIT 2 +#define INTEL_HDMID_CLONE_BIT 3 +#define INTEL_HDMIE_CLONE_BIT 4 +#define INTEL_HDMIF_CLONE_BIT 5 +#define INTEL_SDVO_NON_TV_CLONE_BIT 6 +#define INTEL_SDVO_TV_CLONE_BIT 7 +#define INTEL_SDVO_LVDS_CLONE_BIT 8 +#define INTEL_ANALOG_CLONE_BIT 9 +#define INTEL_TV_CLONE_BIT 10 +#define INTEL_DP_B_CLONE_BIT 11 +#define INTEL_DP_C_CLONE_BIT 12 +#define INTEL_DP_D_CLONE_BIT 13 +#define INTEL_LVDS_CLONE_BIT 14 +#define INTEL_DVO_TMDS_CLONE_BIT 15 +#define INTEL_DVO_LVDS_CLONE_BIT 16 +#define INTEL_EDP_CLONE_BIT 17 + +#define INTEL_DVO_CHIP_NONE 0 +#define INTEL_DVO_CHIP_LVDS 1 +#define INTEL_DVO_CHIP_TMDS 2 +#define INTEL_DVO_CHIP_TVOUT 4 + +/* drm_display_mode->private_flags */ +#define INTEL_MODE_PIXEL_MULTIPLIER_SHIFT (0x0) +#define INTEL_MODE_PIXEL_MULTIPLIER_MASK (0xf << INTEL_MODE_PIXEL_MULTIPLIER_SHIFT) + +static inline void +intel_mode_set_pixel_multiplier(struct drm_display_mode *mode, + int multiplier) +{ + mode->clock *= multiplier; + mode->private_flags |= multiplier; +} + +static inline int +intel_mode_get_pixel_multiplier(const struct drm_display_mode *mode) +{ + return (mode->private_flags & INTEL_MODE_PIXEL_MULTIPLIER_MASK) >> INTEL_MODE_PIXEL_MULTIPLIER_SHIFT; +} + +struct intel_framebuffer { + struct drm_framebuffer base; + struct drm_i915_gem_object *obj; +}; + +struct intel_fbdev { + struct drm_fb_helper helper; + struct intel_framebuffer ifb; + struct list_head fbdev_list; + struct drm_display_mode *our_mode; +}; + +struct intel_encoder { + struct drm_encoder base; + int type; + bool needs_tv_clock; + void (*hot_plug)(struct intel_encoder *); + int crtc_mask; + int clone_mask; +}; + +struct intel_connector { + struct drm_connector base; + struct intel_encoder *encoder; +}; + +struct intel_crtc { + struct drm_crtc base; + enum pipe pipe; + enum plane plane; + u8 lut_r[256], lut_g[256], lut_b[256]; + int dpms_mode; + bool active; /* is the crtc on? independent of the dpms mode */ + bool busy; /* is scanout buffer being updated frequently? */ + struct timer_list idle_timer; + bool lowfreq_avail; + struct intel_overlay *overlay; + struct intel_unpin_work *unpin_work; + int fdi_lanes; + + struct drm_i915_gem_object *cursor_bo; + uint32_t cursor_addr; + int16_t cursor_x, cursor_y; + int16_t cursor_width, cursor_height; + bool cursor_visible; +}; + +#define to_intel_crtc(x) container_of(x, struct intel_crtc, base) +#define to_intel_connector(x) container_of(x, struct intel_connector, base) +#define to_intel_encoder(x) container_of(x, struct intel_encoder, base) +#define to_intel_framebuffer(x) container_of(x, struct intel_framebuffer, base) + +#define DIP_TYPE_AVI 0x82 +#define DIP_VERSION_AVI 0x2 +#define DIP_LEN_AVI 13 + +struct dip_infoframe { + uint8_t type; /* HB0 */ + uint8_t ver; /* HB1 */ + uint8_t len; /* HB2 - body len, not including checksum */ + uint8_t ecc; /* Header ECC */ + uint8_t checksum; /* PB0 */ + union { + struct { + /* PB1 - Y 6:5, A 4:4, B 3:2, S 1:0 */ + uint8_t Y_A_B_S; + /* PB2 - C 7:6, M 5:4, R 3:0 */ + uint8_t C_M_R; + /* PB3 - ITC 7:7, EC 6:4, Q 3:2, SC 1:0 */ + uint8_t ITC_EC_Q_SC; + /* PB4 - VIC 6:0 */ + uint8_t VIC; + /* PB5 - PR 3:0 */ + uint8_t PR; + /* PB6 to PB13 */ + uint16_t top_bar_end; + uint16_t bottom_bar_start; + uint16_t left_bar_end; + uint16_t right_bar_start; + } avi; + uint8_t payload[27]; + } __attribute__ ((packed)) body; +} __attribute__((packed)); + +static inline struct drm_crtc * +intel_get_crtc_for_pipe(struct drm_device *dev, int pipe) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + return dev_priv->pipe_to_crtc_mapping[pipe]; +} + +static inline struct drm_crtc * +intel_get_crtc_for_plane(struct drm_device *dev, int plane) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + return dev_priv->plane_to_crtc_mapping[plane]; +} + +struct intel_unpin_work { + struct work_struct work; + struct drm_device *dev; + struct drm_i915_gem_object *old_fb_obj; + struct drm_i915_gem_object *pending_flip_obj; + struct drm_pending_vblank_event *event; + int pending; + bool enable_stall_check; +}; + +int intel_ddc_get_modes(struct drm_connector *c, struct i2c_adapter *adapter); +extern bool intel_ddc_probe(struct intel_encoder *intel_encoder, int ddc_bus); + +extern void intel_attach_force_audio_property(struct drm_connector *connector); +extern void intel_attach_broadcast_rgb_property(struct drm_connector *connector); + +extern void intel_crt_init(struct drm_device *dev); +extern void intel_hdmi_init(struct drm_device *dev, int sdvox_reg); +void intel_dip_infoframe_csum(struct dip_infoframe *avi_if); +extern bool intel_sdvo_init(struct drm_device *dev, int output_device); +extern void intel_dvo_init(struct drm_device *dev); +extern void intel_tv_init(struct drm_device *dev); +extern void intel_mark_busy(struct drm_device *dev, + struct drm_i915_gem_object *obj); +extern bool intel_lvds_init(struct drm_device *dev); +extern void intel_dp_init(struct drm_device *dev, int dp_reg); +void +intel_dp_set_m_n(struct drm_crtc *crtc, struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode); +extern bool intel_dpd_is_edp(struct drm_device *dev); +extern void intel_edp_link_config (struct intel_encoder *, int *, int *); +extern bool intel_encoder_is_pch_edp(struct drm_encoder *encoder); + +/* intel_panel.c */ +extern void intel_fixed_panel_mode(struct drm_display_mode *fixed_mode, + struct drm_display_mode *adjusted_mode); +extern void intel_pch_panel_fitting(struct drm_device *dev, + int fitting_mode, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode); +extern u32 intel_panel_get_max_backlight(struct drm_device *dev); +extern u32 intel_panel_get_backlight(struct drm_device *dev); +extern void intel_panel_set_backlight(struct drm_device *dev, u32 level); +extern void intel_panel_setup_backlight(struct drm_device *dev); +extern void intel_panel_enable_backlight(struct drm_device *dev); +extern void intel_panel_disable_backlight(struct drm_device *dev); +extern enum drm_connector_status intel_panel_detect(struct drm_device *dev); + +extern void intel_crtc_load_lut(struct drm_crtc *crtc); +extern void intel_encoder_prepare (struct drm_encoder *encoder); +extern void intel_encoder_commit (struct drm_encoder *encoder); +extern void intel_encoder_destroy(struct drm_encoder *encoder); + +static inline struct intel_encoder *intel_attached_encoder(struct drm_connector *connector) +{ + return to_intel_connector(connector)->encoder; +} + +extern void intel_connector_attach_encoder(struct intel_connector *connector, + struct intel_encoder *encoder); +extern struct drm_encoder *intel_best_encoder(struct drm_connector *connector); + +extern struct drm_display_mode *intel_crtc_mode_get(struct drm_device *dev, + struct drm_crtc *crtc); +int intel_get_pipe_from_crtc_id(struct drm_device *dev, void *data, + struct drm_file *file_priv); +extern void intel_wait_for_vblank(struct drm_device *dev, int pipe); +extern void intel_wait_for_pipe_off(struct drm_device *dev, int pipe); + +struct intel_load_detect_pipe { + struct drm_framebuffer *release_fb; + bool load_detect_temp; + int dpms_mode; +}; +extern bool intel_get_load_detect_pipe(struct intel_encoder *intel_encoder, + struct drm_connector *connector, + struct drm_display_mode *mode, + struct intel_load_detect_pipe *old); +extern void intel_release_load_detect_pipe(struct intel_encoder *intel_encoder, + struct drm_connector *connector, + struct intel_load_detect_pipe *old); + +extern struct drm_connector* intel_sdvo_find(struct drm_device *dev, int sdvoB); +extern int intel_sdvo_supports_hotplug(struct drm_connector *connector); +extern void intel_sdvo_set_hotplug(struct drm_connector *connector, int enable); +extern void intelfb_restore(void); +extern void intel_crtc_fb_gamma_set(struct drm_crtc *crtc, u16 red, u16 green, + u16 blue, int regno); +extern void intel_crtc_fb_gamma_get(struct drm_crtc *crtc, u16 *red, u16 *green, + u16 *blue, int regno); +extern void intel_enable_clock_gating(struct drm_device *dev); +extern void ironlake_enable_drps(struct drm_device *dev); +extern void ironlake_disable_drps(struct drm_device *dev); +extern void gen6_enable_rps(struct drm_i915_private *dev_priv); +extern void gen6_disable_rps(struct drm_device *dev); +extern void intel_init_emon(struct drm_device *dev); + +extern int intel_pin_and_fence_fb_obj(struct drm_device *dev, + struct drm_i915_gem_object *obj, + struct intel_ring_buffer *pipelined); + +extern int intel_framebuffer_init(struct drm_device *dev, + struct intel_framebuffer *ifb, + struct drm_mode_fb_cmd *mode_cmd, + struct drm_i915_gem_object *obj); +extern int intel_fbdev_init(struct drm_device *dev); +extern void intel_fbdev_fini(struct drm_device *dev); +extern void intel_fbdev_set_suspend(struct drm_device *dev, int state); +extern void intel_prepare_page_flip(struct drm_device *dev, int plane); +extern void intel_finish_page_flip(struct drm_device *dev, int pipe); +extern void intel_finish_page_flip_plane(struct drm_device *dev, int plane); + +extern void intel_setup_overlay(struct drm_device *dev); +extern void intel_cleanup_overlay(struct drm_device *dev); +extern int intel_overlay_switch_off(struct intel_overlay *overlay); +extern int intel_overlay_put_image(struct drm_device *dev, void *data, + struct drm_file *file_priv); +extern int intel_overlay_attrs(struct drm_device *dev, void *data, + struct drm_file *file_priv); + +extern void intel_fb_output_poll_changed(struct drm_device *dev); +extern void intel_fb_restore_mode(struct drm_device *dev); + +extern void intel_init_clock_gating(struct drm_device *dev); +#endif /* __INTEL_DRV_H__ */ diff --git a/drivers/gpu/drm/i915/intel_dvo.c b/drivers/gpu/drm/i915/intel_dvo.c new file mode 100644 index 00000000..6eda1b51 --- /dev/null +++ b/drivers/gpu/drm/i915/intel_dvo.c @@ -0,0 +1,448 @@ +/* + * Copyright 2006 Dave Airlie + * Copyright © 2006-2007 Intel Corporation + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Eric Anholt + */ +#include +#include +#include "drmP.h" +#include "drm.h" +#include "drm_crtc.h" +#include "intel_drv.h" +#include "i915_drm.h" +#include "i915_drv.h" +#include "dvo.h" + +#define SIL164_ADDR 0x38 +#define CH7xxx_ADDR 0x76 +#define TFP410_ADDR 0x38 + +static const struct intel_dvo_device intel_dvo_devices[] = { + { + .type = INTEL_DVO_CHIP_TMDS, + .name = "sil164", + .dvo_reg = DVOC, + .slave_addr = SIL164_ADDR, + .dev_ops = &sil164_ops, + }, + { + .type = INTEL_DVO_CHIP_TMDS, + .name = "ch7xxx", + .dvo_reg = DVOC, + .slave_addr = CH7xxx_ADDR, + .dev_ops = &ch7xxx_ops, + }, + { + .type = INTEL_DVO_CHIP_LVDS, + .name = "ivch", + .dvo_reg = DVOA, + .slave_addr = 0x02, /* Might also be 0x44, 0x84, 0xc4 */ + .dev_ops = &ivch_ops, + }, + { + .type = INTEL_DVO_CHIP_TMDS, + .name = "tfp410", + .dvo_reg = DVOC, + .slave_addr = TFP410_ADDR, + .dev_ops = &tfp410_ops, + }, + { + .type = INTEL_DVO_CHIP_LVDS, + .name = "ch7017", + .dvo_reg = DVOC, + .slave_addr = 0x75, + .gpio = GMBUS_PORT_DPB, + .dev_ops = &ch7017_ops, + } +}; + +struct intel_dvo { + struct intel_encoder base; + + struct intel_dvo_device dev; + + struct drm_display_mode *panel_fixed_mode; + bool panel_wants_dither; +}; + +static struct intel_dvo *enc_to_intel_dvo(struct drm_encoder *encoder) +{ + return container_of(encoder, struct intel_dvo, base.base); +} + +static struct intel_dvo *intel_attached_dvo(struct drm_connector *connector) +{ + return container_of(intel_attached_encoder(connector), + struct intel_dvo, base); +} + +static void intel_dvo_dpms(struct drm_encoder *encoder, int mode) +{ + struct drm_i915_private *dev_priv = encoder->dev->dev_private; + struct intel_dvo *intel_dvo = enc_to_intel_dvo(encoder); + u32 dvo_reg = intel_dvo->dev.dvo_reg; + u32 temp = I915_READ(dvo_reg); + + if (mode == DRM_MODE_DPMS_ON) { + I915_WRITE(dvo_reg, temp | DVO_ENABLE); + I915_READ(dvo_reg); + intel_dvo->dev.dev_ops->dpms(&intel_dvo->dev, mode); + } else { + intel_dvo->dev.dev_ops->dpms(&intel_dvo->dev, mode); + I915_WRITE(dvo_reg, temp & ~DVO_ENABLE); + I915_READ(dvo_reg); + } +} + +static int intel_dvo_mode_valid(struct drm_connector *connector, + struct drm_display_mode *mode) +{ + struct intel_dvo *intel_dvo = intel_attached_dvo(connector); + + if (mode->flags & DRM_MODE_FLAG_DBLSCAN) + return MODE_NO_DBLESCAN; + + /* XXX: Validate clock range */ + + if (intel_dvo->panel_fixed_mode) { + if (mode->hdisplay > intel_dvo->panel_fixed_mode->hdisplay) + return MODE_PANEL; + if (mode->vdisplay > intel_dvo->panel_fixed_mode->vdisplay) + return MODE_PANEL; + } + + return intel_dvo->dev.dev_ops->mode_valid(&intel_dvo->dev, mode); +} + +static bool intel_dvo_mode_fixup(struct drm_encoder *encoder, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct intel_dvo *intel_dvo = enc_to_intel_dvo(encoder); + + /* If we have timings from the BIOS for the panel, put them in + * to the adjusted mode. The CRTC will be set up for this mode, + * with the panel scaling set up to source from the H/VDisplay + * of the original mode. + */ + if (intel_dvo->panel_fixed_mode != NULL) { +#define C(x) adjusted_mode->x = intel_dvo->panel_fixed_mode->x + C(hdisplay); + C(hsync_start); + C(hsync_end); + C(htotal); + C(vdisplay); + C(vsync_start); + C(vsync_end); + C(vtotal); + C(clock); + drm_mode_set_crtcinfo(adjusted_mode, CRTC_INTERLACE_HALVE_V); +#undef C + } + + if (intel_dvo->dev.dev_ops->mode_fixup) + return intel_dvo->dev.dev_ops->mode_fixup(&intel_dvo->dev, mode, adjusted_mode); + + return true; +} + +static void intel_dvo_mode_set(struct drm_encoder *encoder, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct drm_device *dev = encoder->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(encoder->crtc); + struct intel_dvo *intel_dvo = enc_to_intel_dvo(encoder); + int pipe = intel_crtc->pipe; + u32 dvo_val; + u32 dvo_reg = intel_dvo->dev.dvo_reg, dvo_srcdim_reg; + int dpll_reg = DPLL(pipe); + + switch (dvo_reg) { + case DVOA: + default: + dvo_srcdim_reg = DVOA_SRCDIM; + break; + case DVOB: + dvo_srcdim_reg = DVOB_SRCDIM; + break; + case DVOC: + dvo_srcdim_reg = DVOC_SRCDIM; + break; + } + + intel_dvo->dev.dev_ops->mode_set(&intel_dvo->dev, mode, adjusted_mode); + + /* Save the data order, since I don't know what it should be set to. */ + dvo_val = I915_READ(dvo_reg) & + (DVO_PRESERVE_MASK | DVO_DATA_ORDER_GBRG); + dvo_val |= DVO_DATA_ORDER_FP | DVO_BORDER_ENABLE | + DVO_BLANK_ACTIVE_HIGH; + + if (pipe == 1) + dvo_val |= DVO_PIPE_B_SELECT; + dvo_val |= DVO_PIPE_STALL; + if (adjusted_mode->flags & DRM_MODE_FLAG_PHSYNC) + dvo_val |= DVO_HSYNC_ACTIVE_HIGH; + if (adjusted_mode->flags & DRM_MODE_FLAG_PVSYNC) + dvo_val |= DVO_VSYNC_ACTIVE_HIGH; + + I915_WRITE(dpll_reg, I915_READ(dpll_reg) | DPLL_DVO_HIGH_SPEED); + + /*I915_WRITE(DVOB_SRCDIM, + (adjusted_mode->hdisplay << DVO_SRCDIM_HORIZONTAL_SHIFT) | + (adjusted_mode->VDisplay << DVO_SRCDIM_VERTICAL_SHIFT));*/ + I915_WRITE(dvo_srcdim_reg, + (adjusted_mode->hdisplay << DVO_SRCDIM_HORIZONTAL_SHIFT) | + (adjusted_mode->vdisplay << DVO_SRCDIM_VERTICAL_SHIFT)); + /*I915_WRITE(DVOB, dvo_val);*/ + I915_WRITE(dvo_reg, dvo_val); +} + +/** + * Detect the output connection on our DVO device. + * + * Unimplemented. + */ +static enum drm_connector_status +intel_dvo_detect(struct drm_connector *connector, bool force) +{ + struct intel_dvo *intel_dvo = intel_attached_dvo(connector); + return intel_dvo->dev.dev_ops->detect(&intel_dvo->dev); +} + +static int intel_dvo_get_modes(struct drm_connector *connector) +{ + struct intel_dvo *intel_dvo = intel_attached_dvo(connector); + struct drm_i915_private *dev_priv = connector->dev->dev_private; + + /* We should probably have an i2c driver get_modes function for those + * devices which will have a fixed set of modes determined by the chip + * (TV-out, for example), but for now with just TMDS and LVDS, + * that's not the case. + */ + intel_ddc_get_modes(connector, + &dev_priv->gmbus[GMBUS_PORT_DPC].adapter); + if (!list_empty(&connector->probed_modes)) + return 1; + + if (intel_dvo->panel_fixed_mode != NULL) { + struct drm_display_mode *mode; + mode = drm_mode_duplicate(connector->dev, intel_dvo->panel_fixed_mode); + if (mode) { + drm_mode_probed_add(connector, mode); + return 1; + } + } + + return 0; +} + +static void intel_dvo_destroy(struct drm_connector *connector) +{ + drm_sysfs_connector_remove(connector); + drm_connector_cleanup(connector); + kfree(connector); +} + +static const struct drm_encoder_helper_funcs intel_dvo_helper_funcs = { + .dpms = intel_dvo_dpms, + .mode_fixup = intel_dvo_mode_fixup, + .prepare = intel_encoder_prepare, + .mode_set = intel_dvo_mode_set, + .commit = intel_encoder_commit, +}; + +static const struct drm_connector_funcs intel_dvo_connector_funcs = { + .dpms = drm_helper_connector_dpms, + .detect = intel_dvo_detect, + .destroy = intel_dvo_destroy, + .fill_modes = drm_helper_probe_single_connector_modes, +}; + +static const struct drm_connector_helper_funcs intel_dvo_connector_helper_funcs = { + .mode_valid = intel_dvo_mode_valid, + .get_modes = intel_dvo_get_modes, + .best_encoder = intel_best_encoder, +}; + +static void intel_dvo_enc_destroy(struct drm_encoder *encoder) +{ + struct intel_dvo *intel_dvo = enc_to_intel_dvo(encoder); + + if (intel_dvo->dev.dev_ops->destroy) + intel_dvo->dev.dev_ops->destroy(&intel_dvo->dev); + + kfree(intel_dvo->panel_fixed_mode); + + intel_encoder_destroy(encoder); +} + +static const struct drm_encoder_funcs intel_dvo_enc_funcs = { + .destroy = intel_dvo_enc_destroy, +}; + +/** + * Attempts to get a fixed panel timing for LVDS (currently only the i830). + * + * Other chips with DVO LVDS will need to extend this to deal with the LVDS + * chip being on DVOB/C and having multiple pipes. + */ +static struct drm_display_mode * +intel_dvo_get_current_mode(struct drm_connector *connector) +{ + struct drm_device *dev = connector->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_dvo *intel_dvo = intel_attached_dvo(connector); + uint32_t dvo_val = I915_READ(intel_dvo->dev.dvo_reg); + struct drm_display_mode *mode = NULL; + + /* If the DVO port is active, that'll be the LVDS, so we can pull out + * its timings to get how the BIOS set up the panel. + */ + if (dvo_val & DVO_ENABLE) { + struct drm_crtc *crtc; + int pipe = (dvo_val & DVO_PIPE_B_SELECT) ? 1 : 0; + + crtc = intel_get_crtc_for_pipe(dev, pipe); + if (crtc) { + mode = intel_crtc_mode_get(dev, crtc); + if (mode) { + mode->type |= DRM_MODE_TYPE_PREFERRED; + if (dvo_val & DVO_HSYNC_ACTIVE_HIGH) + mode->flags |= DRM_MODE_FLAG_PHSYNC; + if (dvo_val & DVO_VSYNC_ACTIVE_HIGH) + mode->flags |= DRM_MODE_FLAG_PVSYNC; + } + } + } + + return mode; +} + +void intel_dvo_init(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_encoder *intel_encoder; + struct intel_dvo *intel_dvo; + struct intel_connector *intel_connector; + int i; + int encoder_type = DRM_MODE_ENCODER_NONE; + + intel_dvo = kzalloc(sizeof(struct intel_dvo), GFP_KERNEL); + if (!intel_dvo) + return; + + intel_connector = kzalloc(sizeof(struct intel_connector), GFP_KERNEL); + if (!intel_connector) { + kfree(intel_dvo); + return; + } + + intel_encoder = &intel_dvo->base; + drm_encoder_init(dev, &intel_encoder->base, + &intel_dvo_enc_funcs, encoder_type); + + /* Now, try to find a controller */ + for (i = 0; i < ARRAY_SIZE(intel_dvo_devices); i++) { + struct drm_connector *connector = &intel_connector->base; + const struct intel_dvo_device *dvo = &intel_dvo_devices[i]; + struct i2c_adapter *i2c; + int gpio; + + /* Allow the I2C driver info to specify the GPIO to be used in + * special cases, but otherwise default to what's defined + * in the spec. + */ + if (dvo->gpio != 0) + gpio = dvo->gpio; + else if (dvo->type == INTEL_DVO_CHIP_LVDS) + gpio = GMBUS_PORT_SSC; + else + gpio = GMBUS_PORT_DPB; + + /* Set up the I2C bus necessary for the chip we're probing. + * It appears that everything is on GPIOE except for panels + * on i830 laptops, which are on GPIOB (DVOA). + */ + i2c = &dev_priv->gmbus[gpio].adapter; + + intel_dvo->dev = *dvo; + if (!dvo->dev_ops->init(&intel_dvo->dev, i2c)) + continue; + + intel_encoder->type = INTEL_OUTPUT_DVO; + intel_encoder->crtc_mask = (1 << 0) | (1 << 1); + switch (dvo->type) { + case INTEL_DVO_CHIP_TMDS: + intel_encoder->clone_mask = + (1 << INTEL_DVO_TMDS_CLONE_BIT) | + (1 << INTEL_ANALOG_CLONE_BIT); + drm_connector_init(dev, connector, + &intel_dvo_connector_funcs, + DRM_MODE_CONNECTOR_DVII); + encoder_type = DRM_MODE_ENCODER_TMDS; + break; + case INTEL_DVO_CHIP_LVDS: + intel_encoder->clone_mask = + (1 << INTEL_DVO_LVDS_CLONE_BIT); + drm_connector_init(dev, connector, + &intel_dvo_connector_funcs, + DRM_MODE_CONNECTOR_LVDS); + encoder_type = DRM_MODE_ENCODER_LVDS; + break; + } + + drm_connector_helper_add(connector, + &intel_dvo_connector_helper_funcs); + connector->display_info.subpixel_order = SubPixelHorizontalRGB; + connector->interlace_allowed = false; + connector->doublescan_allowed = false; + + drm_encoder_helper_add(&intel_encoder->base, + &intel_dvo_helper_funcs); + + intel_connector_attach_encoder(intel_connector, intel_encoder); + if (dvo->type == INTEL_DVO_CHIP_LVDS) { + /* For our LVDS chipsets, we should hopefully be able + * to dig the fixed panel mode out of the BIOS data. + * However, it's in a different format from the BIOS + * data on chipsets with integrated LVDS (stored in AIM + * headers, likely), so for now, just get the current + * mode being output through DVO. + */ + intel_dvo->panel_fixed_mode = + intel_dvo_get_current_mode(connector); + intel_dvo->panel_wants_dither = true; + } + + drm_sysfs_connector_add(connector); + return; + } + + drm_encoder_cleanup(&intel_encoder->base); + kfree(intel_dvo); + kfree(intel_connector); +} diff --git a/drivers/gpu/drm/i915/intel_fb.c b/drivers/gpu/drm/i915/intel_fb.c new file mode 100644 index 00000000..d0ce34b7 --- /dev/null +++ b/drivers/gpu/drm/i915/intel_fb.c @@ -0,0 +1,286 @@ +/* + * Copyright © 2007 David Airlie + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * David Airlie + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include "drmP.h" +#include "drm.h" +#include "drm_crtc.h" +#include "drm_fb_helper.h" +#include "intel_drv.h" +#include "i915_drm.h" +#include "i915_drv.h" + +static struct fb_ops intelfb_ops = { + .owner = THIS_MODULE, + .fb_check_var = drm_fb_helper_check_var, + .fb_set_par = drm_fb_helper_set_par, + .fb_fillrect = cfb_fillrect, + .fb_copyarea = cfb_copyarea, + .fb_imageblit = cfb_imageblit, + .fb_pan_display = drm_fb_helper_pan_display, + .fb_blank = drm_fb_helper_blank, + .fb_setcmap = drm_fb_helper_setcmap, + .fb_debug_enter = drm_fb_helper_debug_enter, + .fb_debug_leave = drm_fb_helper_debug_leave, +}; + +static int intelfb_create(struct intel_fbdev *ifbdev, + struct drm_fb_helper_surface_size *sizes) +{ + struct drm_device *dev = ifbdev->helper.dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct fb_info *info; + struct drm_framebuffer *fb; + struct drm_mode_fb_cmd mode_cmd; + struct drm_i915_gem_object *obj; + struct device *device = &dev->pdev->dev; + int size, ret; + + /* we don't do packed 24bpp */ + if (sizes->surface_bpp == 24) + sizes->surface_bpp = 32; + + mode_cmd.width = sizes->surface_width; + mode_cmd.height = sizes->surface_height; + + mode_cmd.bpp = sizes->surface_bpp; + mode_cmd.pitch = ALIGN(mode_cmd.width * ((mode_cmd.bpp + 7) / 8), 64); + mode_cmd.depth = sizes->surface_depth; + + size = mode_cmd.pitch * mode_cmd.height; + size = ALIGN(size, PAGE_SIZE); + obj = i915_gem_alloc_object(dev, size); + if (!obj) { + DRM_ERROR("failed to allocate framebuffer\n"); + ret = -ENOMEM; + goto out; + } + + mutex_lock(&dev->struct_mutex); + + /* Flush everything out, we'll be doing GTT only from now on */ + ret = intel_pin_and_fence_fb_obj(dev, obj, false); + if (ret) { + DRM_ERROR("failed to pin fb: %d\n", ret); + goto out_unref; + } + + info = framebuffer_alloc(0, device); + if (!info) { + ret = -ENOMEM; + goto out_unpin; + } + + info->par = ifbdev; + + ret = intel_framebuffer_init(dev, &ifbdev->ifb, &mode_cmd, obj); + if (ret) + goto out_unpin; + + fb = &ifbdev->ifb.base; + + ifbdev->helper.fb = fb; + ifbdev->helper.fbdev = info; + + strcpy(info->fix.id, "inteldrmfb"); + + info->flags = FBINFO_DEFAULT | FBINFO_CAN_FORCE_OUTPUT; + info->fbops = &intelfb_ops; + + ret = fb_alloc_cmap(&info->cmap, 256, 0); + if (ret) { + ret = -ENOMEM; + goto out_unpin; + } + /* setup aperture base/size for vesafb takeover */ + info->apertures = alloc_apertures(1); + if (!info->apertures) { + ret = -ENOMEM; + goto out_unpin; + } + info->apertures->ranges[0].base = dev->mode_config.fb_base; + info->apertures->ranges[0].size = + dev_priv->mm.gtt->gtt_mappable_entries << PAGE_SHIFT; + + info->fix.smem_start = dev->mode_config.fb_base + obj->gtt_offset; + info->fix.smem_len = size; + + info->screen_base = ioremap_wc(dev->agp->base + obj->gtt_offset, size); + if (!info->screen_base) { + ret = -ENOSPC; + goto out_unpin; + } + info->screen_size = size; + +// memset(info->screen_base, 0, size); + + drm_fb_helper_fill_fix(info, fb->pitch, fb->depth); + drm_fb_helper_fill_var(info, &ifbdev->helper, sizes->fb_width, sizes->fb_height); + + info->pixmap.size = 64*1024; + info->pixmap.buf_align = 8; + info->pixmap.access_align = 32; + info->pixmap.flags = FB_PIXMAP_SYSTEM; + info->pixmap.scan_align = 1; + + DRM_DEBUG_KMS("allocated %dx%d fb: 0x%08x, bo %p\n", + fb->width, fb->height, + obj->gtt_offset, obj); + + + mutex_unlock(&dev->struct_mutex); + vga_switcheroo_client_fb_set(dev->pdev, info); + return 0; + +out_unpin: + i915_gem_object_unpin(obj); +out_unref: + drm_gem_object_unreference(&obj->base); + mutex_unlock(&dev->struct_mutex); +out: + return ret; +} + +static int intel_fb_find_or_create_single(struct drm_fb_helper *helper, + struct drm_fb_helper_surface_size *sizes) +{ + struct intel_fbdev *ifbdev = (struct intel_fbdev *)helper; + int new_fb = 0; + int ret; + + if (!helper->fb) { + ret = intelfb_create(ifbdev, sizes); + if (ret) + return ret; + new_fb = 1; + } + return new_fb; +} + +static struct drm_fb_helper_funcs intel_fb_helper_funcs = { + .gamma_set = intel_crtc_fb_gamma_set, + .gamma_get = intel_crtc_fb_gamma_get, + .fb_probe = intel_fb_find_or_create_single, +}; + +static void intel_fbdev_destroy(struct drm_device *dev, + struct intel_fbdev *ifbdev) +{ + struct fb_info *info; + struct intel_framebuffer *ifb = &ifbdev->ifb; + + if (ifbdev->helper.fbdev) { + info = ifbdev->helper.fbdev; + unregister_framebuffer(info); + iounmap(info->screen_base); + if (info->cmap.len) + fb_dealloc_cmap(&info->cmap); + framebuffer_release(info); + } + + drm_fb_helper_fini(&ifbdev->helper); + + drm_framebuffer_cleanup(&ifb->base); + if (ifb->obj) { + drm_gem_object_unreference_unlocked(&ifb->obj->base); + ifb->obj = NULL; + } +} + +int intel_fbdev_init(struct drm_device *dev) +{ + struct intel_fbdev *ifbdev; + drm_i915_private_t *dev_priv = dev->dev_private; + int ret; + + ifbdev = kzalloc(sizeof(struct intel_fbdev), GFP_KERNEL); + if (!ifbdev) + return -ENOMEM; + + dev_priv->fbdev = ifbdev; + ifbdev->helper.funcs = &intel_fb_helper_funcs; + + ret = drm_fb_helper_init(dev, &ifbdev->helper, + dev_priv->num_pipe, + INTELFB_CONN_LIMIT); + if (ret) { + kfree(ifbdev); + return ret; + } + + drm_fb_helper_single_add_all_connectors(&ifbdev->helper); + drm_fb_helper_initial_config(&ifbdev->helper, 32); + return 0; +} + +void intel_fbdev_fini(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + if (!dev_priv->fbdev) + return; + + intel_fbdev_destroy(dev, dev_priv->fbdev); + kfree(dev_priv->fbdev); + dev_priv->fbdev = NULL; +} + +void intel_fbdev_set_suspend(struct drm_device *dev, int state) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + if (!dev_priv->fbdev) + return; + + fb_set_suspend(dev_priv->fbdev->helper.fbdev, state); +} + +MODULE_LICENSE("GPL and additional rights"); + +void intel_fb_output_poll_changed(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + drm_fb_helper_hotplug_event(&dev_priv->fbdev->helper); +} + +void intel_fb_restore_mode(struct drm_device *dev) +{ + int ret; + drm_i915_private_t *dev_priv = dev->dev_private; + + ret = drm_fb_helper_restore_fbdev_mode(&dev_priv->fbdev->helper); + if (ret) + DRM_DEBUG("failed to restore crtc mode\n"); +} diff --git a/drivers/gpu/drm/i915/intel_hdmi.c b/drivers/gpu/drm/i915/intel_hdmi.c new file mode 100644 index 00000000..236bbe09 --- /dev/null +++ b/drivers/gpu/drm/i915/intel_hdmi.c @@ -0,0 +1,449 @@ +/* + * Copyright 2006 Dave Airlie + * Copyright © 2006-2009 Intel Corporation + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Eric Anholt + * Jesse Barnes + */ + +#include +#include +#include +#include "drmP.h" +#include "drm.h" +#include "drm_crtc.h" +#include "drm_edid.h" +#include "intel_drv.h" +#include "i915_drm.h" +#include "i915_drv.h" + +struct intel_hdmi { + struct intel_encoder base; + u32 sdvox_reg; + int ddc_bus; + uint32_t color_range; + bool has_hdmi_sink; + bool has_audio; + int force_audio; +}; + +static struct intel_hdmi *enc_to_intel_hdmi(struct drm_encoder *encoder) +{ + return container_of(encoder, struct intel_hdmi, base.base); +} + +static struct intel_hdmi *intel_attached_hdmi(struct drm_connector *connector) +{ + return container_of(intel_attached_encoder(connector), + struct intel_hdmi, base); +} + +void intel_dip_infoframe_csum(struct dip_infoframe *avi_if) +{ + uint8_t *data = (uint8_t *)avi_if; + uint8_t sum = 0; + unsigned i; + + avi_if->checksum = 0; + avi_if->ecc = 0; + + for (i = 0; i < sizeof(*avi_if); i++) + sum += data[i]; + + avi_if->checksum = 0x100 - sum; +} + +static void intel_hdmi_set_avi_infoframe(struct drm_encoder *encoder) +{ + struct dip_infoframe avi_if = { + .type = DIP_TYPE_AVI, + .ver = DIP_VERSION_AVI, + .len = DIP_LEN_AVI, + }; + uint32_t *data = (uint32_t *)&avi_if; + struct drm_device *dev = encoder->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_hdmi *intel_hdmi = enc_to_intel_hdmi(encoder); + u32 port; + unsigned i; + + if (!intel_hdmi->has_hdmi_sink) + return; + + /* XXX first guess at handling video port, is this corrent? */ + if (intel_hdmi->sdvox_reg == SDVOB) + port = VIDEO_DIP_PORT_B; + else if (intel_hdmi->sdvox_reg == SDVOC) + port = VIDEO_DIP_PORT_C; + else + return; + + I915_WRITE(VIDEO_DIP_CTL, VIDEO_DIP_ENABLE | port | + VIDEO_DIP_SELECT_AVI | VIDEO_DIP_FREQ_VSYNC); + + intel_dip_infoframe_csum(&avi_if); + for (i = 0; i < sizeof(avi_if); i += 4) { + I915_WRITE(VIDEO_DIP_DATA, *data); + data++; + } + + I915_WRITE(VIDEO_DIP_CTL, VIDEO_DIP_ENABLE | port | + VIDEO_DIP_SELECT_AVI | VIDEO_DIP_FREQ_VSYNC | + VIDEO_DIP_ENABLE_AVI); +} + +static void intel_hdmi_mode_set(struct drm_encoder *encoder, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct drm_device *dev = encoder->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_crtc *crtc = encoder->crtc; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + struct intel_hdmi *intel_hdmi = enc_to_intel_hdmi(encoder); + u32 sdvox; + + sdvox = SDVO_ENCODING_HDMI | SDVO_BORDER_ENABLE; + sdvox |= intel_hdmi->color_range; + if (adjusted_mode->flags & DRM_MODE_FLAG_PVSYNC) + sdvox |= SDVO_VSYNC_ACTIVE_HIGH; + if (adjusted_mode->flags & DRM_MODE_FLAG_PHSYNC) + sdvox |= SDVO_HSYNC_ACTIVE_HIGH; + + /* Required on CPT */ + if (intel_hdmi->has_hdmi_sink && HAS_PCH_CPT(dev)) + sdvox |= HDMI_MODE_SELECT; + + if (intel_hdmi->has_audio) { + sdvox |= SDVO_AUDIO_ENABLE; + sdvox |= SDVO_NULL_PACKETS_DURING_VSYNC; + } + + if (intel_crtc->pipe == 1) { + if (HAS_PCH_CPT(dev)) + sdvox |= PORT_TRANS_B_SEL_CPT; + else + sdvox |= SDVO_PIPE_B_SELECT; + } + + I915_WRITE(intel_hdmi->sdvox_reg, sdvox); + POSTING_READ(intel_hdmi->sdvox_reg); + + intel_hdmi_set_avi_infoframe(encoder); +} + +static void intel_hdmi_dpms(struct drm_encoder *encoder, int mode) +{ + struct drm_device *dev = encoder->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_hdmi *intel_hdmi = enc_to_intel_hdmi(encoder); + u32 temp; + u32 enable_bits = SDVO_ENABLE; + + if (intel_hdmi->has_audio) + enable_bits |= SDVO_AUDIO_ENABLE; + + temp = I915_READ(intel_hdmi->sdvox_reg); + + /* HW workaround, need to toggle enable bit off and on for 12bpc, but + * we do this anyway which shows more stable in testing. + */ + if (HAS_PCH_SPLIT(dev)) { + I915_WRITE(intel_hdmi->sdvox_reg, temp & ~SDVO_ENABLE); + POSTING_READ(intel_hdmi->sdvox_reg); + } + + if (mode != DRM_MODE_DPMS_ON) { + temp &= ~enable_bits; + } else { + temp |= enable_bits; + } + + I915_WRITE(intel_hdmi->sdvox_reg, temp); + POSTING_READ(intel_hdmi->sdvox_reg); + + /* HW workaround, need to write this twice for issue that may result + * in first write getting masked. + */ + if (HAS_PCH_SPLIT(dev)) { + I915_WRITE(intel_hdmi->sdvox_reg, temp); + POSTING_READ(intel_hdmi->sdvox_reg); + } +} + +static int intel_hdmi_mode_valid(struct drm_connector *connector, + struct drm_display_mode *mode) +{ + if (mode->clock > 165000) + return MODE_CLOCK_HIGH; + if (mode->clock < 20000) + return MODE_CLOCK_LOW; + + if (mode->flags & DRM_MODE_FLAG_DBLSCAN) + return MODE_NO_DBLESCAN; + + return MODE_OK; +} + +static bool intel_hdmi_mode_fixup(struct drm_encoder *encoder, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + return true; +} + +static enum drm_connector_status +intel_hdmi_detect(struct drm_connector *connector, bool force) +{ + struct intel_hdmi *intel_hdmi = intel_attached_hdmi(connector); + struct drm_i915_private *dev_priv = connector->dev->dev_private; + struct edid *edid; + enum drm_connector_status status = connector_status_disconnected; + + intel_hdmi->has_hdmi_sink = false; + intel_hdmi->has_audio = false; + edid = drm_get_edid(connector, + &dev_priv->gmbus[intel_hdmi->ddc_bus].adapter); + + if (edid) { + if (edid->input & DRM_EDID_INPUT_DIGITAL) { + status = connector_status_connected; + intel_hdmi->has_hdmi_sink = drm_detect_hdmi_monitor(edid); + intel_hdmi->has_audio = drm_detect_monitor_audio(edid); + } + connector->display_info.raw_edid = NULL; + kfree(edid); + } + + if (status == connector_status_connected) { + if (intel_hdmi->force_audio) + intel_hdmi->has_audio = intel_hdmi->force_audio > 0; + } + + return status; +} + +static int intel_hdmi_get_modes(struct drm_connector *connector) +{ + struct intel_hdmi *intel_hdmi = intel_attached_hdmi(connector); + struct drm_i915_private *dev_priv = connector->dev->dev_private; + + /* We should parse the EDID data and find out if it's an HDMI sink so + * we can send audio to it. + */ + + return intel_ddc_get_modes(connector, + &dev_priv->gmbus[intel_hdmi->ddc_bus].adapter); +} + +static bool +intel_hdmi_detect_audio(struct drm_connector *connector) +{ + struct intel_hdmi *intel_hdmi = intel_attached_hdmi(connector); + struct drm_i915_private *dev_priv = connector->dev->dev_private; + struct edid *edid; + bool has_audio = false; + + edid = drm_get_edid(connector, + &dev_priv->gmbus[intel_hdmi->ddc_bus].adapter); + if (edid) { + if (edid->input & DRM_EDID_INPUT_DIGITAL) + has_audio = drm_detect_monitor_audio(edid); + + connector->display_info.raw_edid = NULL; + kfree(edid); + } + + return has_audio; +} + +static int +intel_hdmi_set_property(struct drm_connector *connector, + struct drm_property *property, + uint64_t val) +{ + struct intel_hdmi *intel_hdmi = intel_attached_hdmi(connector); + struct drm_i915_private *dev_priv = connector->dev->dev_private; + int ret; + + ret = drm_connector_property_set_value(connector, property, val); + if (ret) + return ret; + + if (property == dev_priv->force_audio_property) { + int i = val; + bool has_audio; + + if (i == intel_hdmi->force_audio) + return 0; + + intel_hdmi->force_audio = i; + + if (i == 0) + has_audio = intel_hdmi_detect_audio(connector); + else + has_audio = i > 0; + + if (has_audio == intel_hdmi->has_audio) + return 0; + + intel_hdmi->has_audio = has_audio; + goto done; + } + + if (property == dev_priv->broadcast_rgb_property) { + if (val == !!intel_hdmi->color_range) + return 0; + + intel_hdmi->color_range = val ? SDVO_COLOR_RANGE_16_235 : 0; + goto done; + } + + return -EINVAL; + +done: + if (intel_hdmi->base.base.crtc) { + struct drm_crtc *crtc = intel_hdmi->base.base.crtc; + drm_crtc_helper_set_mode(crtc, &crtc->mode, + crtc->x, crtc->y, + crtc->fb); + } + + return 0; +} + +static void intel_hdmi_destroy(struct drm_connector *connector) +{ + drm_sysfs_connector_remove(connector); + drm_connector_cleanup(connector); + kfree(connector); +} + +static const struct drm_encoder_helper_funcs intel_hdmi_helper_funcs = { + .dpms = intel_hdmi_dpms, + .mode_fixup = intel_hdmi_mode_fixup, + .prepare = intel_encoder_prepare, + .mode_set = intel_hdmi_mode_set, + .commit = intel_encoder_commit, +}; + +static const struct drm_connector_funcs intel_hdmi_connector_funcs = { + .dpms = drm_helper_connector_dpms, + .detect = intel_hdmi_detect, + .fill_modes = drm_helper_probe_single_connector_modes, + .set_property = intel_hdmi_set_property, + .destroy = intel_hdmi_destroy, +}; + +static const struct drm_connector_helper_funcs intel_hdmi_connector_helper_funcs = { + .get_modes = intel_hdmi_get_modes, + .mode_valid = intel_hdmi_mode_valid, + .best_encoder = intel_best_encoder, +}; + +static const struct drm_encoder_funcs intel_hdmi_enc_funcs = { + .destroy = intel_encoder_destroy, +}; + +static void +intel_hdmi_add_properties(struct intel_hdmi *intel_hdmi, struct drm_connector *connector) +{ + intel_attach_force_audio_property(connector); + intel_attach_broadcast_rgb_property(connector); +} + +void intel_hdmi_init(struct drm_device *dev, int sdvox_reg) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_connector *connector; + struct intel_encoder *intel_encoder; + struct intel_connector *intel_connector; + struct intel_hdmi *intel_hdmi; + + intel_hdmi = kzalloc(sizeof(struct intel_hdmi), GFP_KERNEL); + if (!intel_hdmi) + return; + + intel_connector = kzalloc(sizeof(struct intel_connector), GFP_KERNEL); + if (!intel_connector) { + kfree(intel_hdmi); + return; + } + + intel_encoder = &intel_hdmi->base; + drm_encoder_init(dev, &intel_encoder->base, &intel_hdmi_enc_funcs, + DRM_MODE_ENCODER_TMDS); + + connector = &intel_connector->base; + drm_connector_init(dev, connector, &intel_hdmi_connector_funcs, + DRM_MODE_CONNECTOR_HDMIA); + drm_connector_helper_add(connector, &intel_hdmi_connector_helper_funcs); + + intel_encoder->type = INTEL_OUTPUT_HDMI; + + connector->polled = DRM_CONNECTOR_POLL_HPD; + connector->interlace_allowed = 0; + connector->doublescan_allowed = 0; + intel_encoder->crtc_mask = (1 << 0) | (1 << 1); + + /* Set up the DDC bus. */ + if (sdvox_reg == SDVOB) { + intel_encoder->clone_mask = (1 << INTEL_HDMIB_CLONE_BIT); + intel_hdmi->ddc_bus = GMBUS_PORT_DPB; + dev_priv->hotplug_supported_mask |= HDMIB_HOTPLUG_INT_STATUS; + } else if (sdvox_reg == SDVOC) { + intel_encoder->clone_mask = (1 << INTEL_HDMIC_CLONE_BIT); + intel_hdmi->ddc_bus = GMBUS_PORT_DPC; + dev_priv->hotplug_supported_mask |= HDMIC_HOTPLUG_INT_STATUS; + } else if (sdvox_reg == HDMIB) { + intel_encoder->clone_mask = (1 << INTEL_HDMID_CLONE_BIT); + intel_hdmi->ddc_bus = GMBUS_PORT_DPB; + dev_priv->hotplug_supported_mask |= HDMIB_HOTPLUG_INT_STATUS; + } else if (sdvox_reg == HDMIC) { + intel_encoder->clone_mask = (1 << INTEL_HDMIE_CLONE_BIT); + intel_hdmi->ddc_bus = GMBUS_PORT_DPC; + dev_priv->hotplug_supported_mask |= HDMIC_HOTPLUG_INT_STATUS; + } else if (sdvox_reg == HDMID) { + intel_encoder->clone_mask = (1 << INTEL_HDMIF_CLONE_BIT); + intel_hdmi->ddc_bus = GMBUS_PORT_DPD; + dev_priv->hotplug_supported_mask |= HDMID_HOTPLUG_INT_STATUS; + } + + intel_hdmi->sdvox_reg = sdvox_reg; + + drm_encoder_helper_add(&intel_encoder->base, &intel_hdmi_helper_funcs); + + intel_hdmi_add_properties(intel_hdmi, connector); + + intel_connector_attach_encoder(intel_connector, intel_encoder); + drm_sysfs_connector_add(connector); + + /* For G4X desktop chip, PEG_BAND_GAP_DATA 3:0 must first be written + * 0xd. Failure to do so will result in spurious interrupts being + * generated on the port when a cable is not attached. + */ + if (IS_G4X(dev) && !IS_GM45(dev)) { + u32 temp = I915_READ(PEG_BAND_GAP_DATA); + I915_WRITE(PEG_BAND_GAP_DATA, (temp & ~0xf) | 0xd); + } +} diff --git a/drivers/gpu/drm/i915/intel_i2c.c b/drivers/gpu/drm/i915/intel_i2c.c new file mode 100644 index 00000000..d98cee60 --- /dev/null +++ b/drivers/gpu/drm/i915/intel_i2c.c @@ -0,0 +1,472 @@ +/* + * Copyright (c) 2006 Dave Airlie + * Copyright © 2006-2008,2010 Intel Corporation + * Jesse Barnes + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Eric Anholt + * Chris Wilson + */ +#include +#include +#include "drmP.h" +#include "drm.h" +#include "intel_drv.h" +#include "i915_drm.h" +#include "i915_drv.h" + +/* Intel GPIO access functions */ + +#define I2C_RISEFALL_TIME 20 + +static inline struct intel_gmbus * +to_intel_gmbus(struct i2c_adapter *i2c) +{ + return container_of(i2c, struct intel_gmbus, adapter); +} + +struct intel_gpio { + struct i2c_adapter adapter; + struct i2c_algo_bit_data algo; + struct drm_i915_private *dev_priv; + u32 reg; +}; + +void +intel_i2c_reset(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + if (HAS_PCH_SPLIT(dev)) + I915_WRITE(PCH_GMBUS0, 0); + else + I915_WRITE(GMBUS0, 0); +} + +static void intel_i2c_quirk_set(struct drm_i915_private *dev_priv, bool enable) +{ + u32 val; + + /* When using bit bashing for I2C, this bit needs to be set to 1 */ + if (!IS_PINEVIEW(dev_priv->dev)) + return; + + val = I915_READ(DSPCLK_GATE_D); + if (enable) + val |= DPCUNIT_CLOCK_GATE_DISABLE; + else + val &= ~DPCUNIT_CLOCK_GATE_DISABLE; + I915_WRITE(DSPCLK_GATE_D, val); +} + +static u32 get_reserved(struct intel_gpio *gpio) +{ + struct drm_i915_private *dev_priv = gpio->dev_priv; + struct drm_device *dev = dev_priv->dev; + u32 reserved = 0; + + /* On most chips, these bits must be preserved in software. */ + if (!IS_I830(dev) && !IS_845G(dev)) + reserved = I915_READ_NOTRACE(gpio->reg) & + (GPIO_DATA_PULLUP_DISABLE | + GPIO_CLOCK_PULLUP_DISABLE); + + return reserved; +} + +static int get_clock(void *data) +{ + struct intel_gpio *gpio = data; + struct drm_i915_private *dev_priv = gpio->dev_priv; + u32 reserved = get_reserved(gpio); + I915_WRITE_NOTRACE(gpio->reg, reserved | GPIO_CLOCK_DIR_MASK); + I915_WRITE_NOTRACE(gpio->reg, reserved); + return (I915_READ_NOTRACE(gpio->reg) & GPIO_CLOCK_VAL_IN) != 0; +} + +static int get_data(void *data) +{ + struct intel_gpio *gpio = data; + struct drm_i915_private *dev_priv = gpio->dev_priv; + u32 reserved = get_reserved(gpio); + I915_WRITE_NOTRACE(gpio->reg, reserved | GPIO_DATA_DIR_MASK); + I915_WRITE_NOTRACE(gpio->reg, reserved); + return (I915_READ_NOTRACE(gpio->reg) & GPIO_DATA_VAL_IN) != 0; +} + +static void set_clock(void *data, int state_high) +{ + struct intel_gpio *gpio = data; + struct drm_i915_private *dev_priv = gpio->dev_priv; + u32 reserved = get_reserved(gpio); + u32 clock_bits; + + if (state_high) + clock_bits = GPIO_CLOCK_DIR_IN | GPIO_CLOCK_DIR_MASK; + else + clock_bits = GPIO_CLOCK_DIR_OUT | GPIO_CLOCK_DIR_MASK | + GPIO_CLOCK_VAL_MASK; + + I915_WRITE_NOTRACE(gpio->reg, reserved | clock_bits); + POSTING_READ(gpio->reg); +} + +static void set_data(void *data, int state_high) +{ + struct intel_gpio *gpio = data; + struct drm_i915_private *dev_priv = gpio->dev_priv; + u32 reserved = get_reserved(gpio); + u32 data_bits; + + if (state_high) + data_bits = GPIO_DATA_DIR_IN | GPIO_DATA_DIR_MASK; + else + data_bits = GPIO_DATA_DIR_OUT | GPIO_DATA_DIR_MASK | + GPIO_DATA_VAL_MASK; + + I915_WRITE_NOTRACE(gpio->reg, reserved | data_bits); + POSTING_READ(gpio->reg); +} + +static struct i2c_adapter * +intel_gpio_create(struct drm_i915_private *dev_priv, u32 pin) +{ + static const int map_pin_to_reg[] = { + 0, + GPIOB, + GPIOA, + GPIOC, + GPIOD, + GPIOE, + 0, + GPIOF, + }; + struct intel_gpio *gpio; + + if (pin >= ARRAY_SIZE(map_pin_to_reg) || !map_pin_to_reg[pin]) + return NULL; + + gpio = kzalloc(sizeof(struct intel_gpio), GFP_KERNEL); + if (gpio == NULL) + return NULL; + + gpio->reg = map_pin_to_reg[pin]; + if (HAS_PCH_SPLIT(dev_priv->dev)) + gpio->reg += PCH_GPIOA - GPIOA; + gpio->dev_priv = dev_priv; + + snprintf(gpio->adapter.name, sizeof(gpio->adapter.name), + "i915 GPIO%c", "?BACDE?F"[pin]); + gpio->adapter.owner = THIS_MODULE; + gpio->adapter.algo_data = &gpio->algo; + gpio->adapter.dev.parent = &dev_priv->dev->pdev->dev; + gpio->algo.setsda = set_data; + gpio->algo.setscl = set_clock; + gpio->algo.getsda = get_data; + gpio->algo.getscl = get_clock; + gpio->algo.udelay = I2C_RISEFALL_TIME; + gpio->algo.timeout = usecs_to_jiffies(2200); + gpio->algo.data = gpio; + + if (i2c_bit_add_bus(&gpio->adapter)) + goto out_free; + + return &gpio->adapter; + +out_free: + kfree(gpio); + return NULL; +} + +static int +intel_i2c_quirk_xfer(struct drm_i915_private *dev_priv, + struct i2c_adapter *adapter, + struct i2c_msg *msgs, + int num) +{ + struct intel_gpio *gpio = container_of(adapter, + struct intel_gpio, + adapter); + int ret; + + intel_i2c_reset(dev_priv->dev); + + intel_i2c_quirk_set(dev_priv, true); + set_data(gpio, 1); + set_clock(gpio, 1); + udelay(I2C_RISEFALL_TIME); + + ret = adapter->algo->master_xfer(adapter, msgs, num); + + set_data(gpio, 1); + set_clock(gpio, 1); + intel_i2c_quirk_set(dev_priv, false); + + return ret; +} + +static int +gmbus_xfer(struct i2c_adapter *adapter, + struct i2c_msg *msgs, + int num) +{ + struct intel_gmbus *bus = container_of(adapter, + struct intel_gmbus, + adapter); + struct drm_i915_private *dev_priv = adapter->algo_data; + int i, reg_offset; + + if (bus->force_bit) + return intel_i2c_quirk_xfer(dev_priv, + bus->force_bit, msgs, num); + + reg_offset = HAS_PCH_SPLIT(dev_priv->dev) ? PCH_GMBUS0 - GMBUS0 : 0; + + I915_WRITE(GMBUS0 + reg_offset, bus->reg0); + + for (i = 0; i < num; i++) { + u16 len = msgs[i].len; + u8 *buf = msgs[i].buf; + + if (msgs[i].flags & I2C_M_RD) { + I915_WRITE(GMBUS1 + reg_offset, + GMBUS_CYCLE_WAIT | (i + 1 == num ? GMBUS_CYCLE_STOP : 0) | + (len << GMBUS_BYTE_COUNT_SHIFT) | + (msgs[i].addr << GMBUS_SLAVE_ADDR_SHIFT) | + GMBUS_SLAVE_READ | GMBUS_SW_RDY); + POSTING_READ(GMBUS2+reg_offset); + do { + u32 val, loop = 0; + + if (wait_for(I915_READ(GMBUS2 + reg_offset) & (GMBUS_SATOER | GMBUS_HW_RDY), 50)) + goto timeout; + if (I915_READ(GMBUS2 + reg_offset) & GMBUS_SATOER) + goto clear_err; + + val = I915_READ(GMBUS3 + reg_offset); + do { + *buf++ = val & 0xff; + val >>= 8; + } while (--len && ++loop < 4); + } while (len); + } else { + u32 val, loop; + + val = loop = 0; + do { + val |= *buf++ << (8 * loop); + } while (--len && ++loop < 4); + + I915_WRITE(GMBUS3 + reg_offset, val); + I915_WRITE(GMBUS1 + reg_offset, + (i + 1 == num ? GMBUS_CYCLE_STOP : GMBUS_CYCLE_WAIT) | + (msgs[i].len << GMBUS_BYTE_COUNT_SHIFT) | + (msgs[i].addr << GMBUS_SLAVE_ADDR_SHIFT) | + GMBUS_SLAVE_WRITE | GMBUS_SW_RDY); + POSTING_READ(GMBUS2+reg_offset); + + while (len) { + if (wait_for(I915_READ(GMBUS2 + reg_offset) & (GMBUS_SATOER | GMBUS_HW_RDY), 50)) + goto timeout; + if (I915_READ(GMBUS2 + reg_offset) & GMBUS_SATOER) + goto clear_err; + + val = loop = 0; + do { + val |= *buf++ << (8 * loop); + } while (--len && ++loop < 4); + + I915_WRITE(GMBUS3 + reg_offset, val); + POSTING_READ(GMBUS2+reg_offset); + } + } + + if (i + 1 < num && wait_for(I915_READ(GMBUS2 + reg_offset) & (GMBUS_SATOER | GMBUS_HW_WAIT_PHASE), 50)) + goto timeout; + if (I915_READ(GMBUS2 + reg_offset) & GMBUS_SATOER) + goto clear_err; + } + + goto done; + +clear_err: + /* Toggle the Software Clear Interrupt bit. This has the effect + * of resetting the GMBUS controller and so clearing the + * BUS_ERROR raised by the slave's NAK. + */ + I915_WRITE(GMBUS1 + reg_offset, GMBUS_SW_CLR_INT); + I915_WRITE(GMBUS1 + reg_offset, 0); + +done: + /* Mark the GMBUS interface as disabled. We will re-enable it at the + * start of the next xfer, till then let it sleep. + */ + I915_WRITE(GMBUS0 + reg_offset, 0); + return i; + +timeout: + DRM_INFO("GMBUS timed out, falling back to bit banging on pin %d [%s]\n", + bus->reg0 & 0xff, bus->adapter.name); + I915_WRITE(GMBUS0 + reg_offset, 0); + + /* Hardware may not support GMBUS over these pins? Try GPIO bitbanging instead. */ + bus->force_bit = intel_gpio_create(dev_priv, bus->reg0 & 0xff); + if (!bus->force_bit) + return -ENOMEM; + + return intel_i2c_quirk_xfer(dev_priv, bus->force_bit, msgs, num); +} + +static u32 gmbus_func(struct i2c_adapter *adapter) +{ + struct intel_gmbus *bus = container_of(adapter, + struct intel_gmbus, + adapter); + + if (bus->force_bit) + bus->force_bit->algo->functionality(bus->force_bit); + + return (I2C_FUNC_I2C | I2C_FUNC_SMBUS_EMUL | + /* I2C_FUNC_10BIT_ADDR | */ + I2C_FUNC_SMBUS_READ_BLOCK_DATA | + I2C_FUNC_SMBUS_BLOCK_PROC_CALL); +} + +static const struct i2c_algorithm gmbus_algorithm = { + .master_xfer = gmbus_xfer, + .functionality = gmbus_func +}; + +/** + * intel_gmbus_setup - instantiate all Intel i2c GMBuses + * @dev: DRM device + */ +int intel_setup_gmbus(struct drm_device *dev) +{ + static const char *names[GMBUS_NUM_PORTS] = { + "disabled", + "ssc", + "vga", + "panel", + "dpc", + "dpb", + "reserved", + "dpd", + }; + struct drm_i915_private *dev_priv = dev->dev_private; + int ret, i; + + dev_priv->gmbus = kcalloc(sizeof(struct intel_gmbus), GMBUS_NUM_PORTS, + GFP_KERNEL); + if (dev_priv->gmbus == NULL) + return -ENOMEM; + + for (i = 0; i < GMBUS_NUM_PORTS; i++) { + struct intel_gmbus *bus = &dev_priv->gmbus[i]; + + bus->adapter.owner = THIS_MODULE; + bus->adapter.class = I2C_CLASS_DDC; + snprintf(bus->adapter.name, + sizeof(bus->adapter.name), + "i915 gmbus %s", + names[i]); + + bus->adapter.dev.parent = &dev->pdev->dev; + bus->adapter.algo_data = dev_priv; + + bus->adapter.algo = &gmbus_algorithm; + ret = i2c_add_adapter(&bus->adapter); + if (ret) + goto err; + + /* By default use a conservative clock rate */ + bus->reg0 = i | GMBUS_RATE_100KHZ; + + /* XXX force bit banging until GMBUS is fully debugged */ + bus->force_bit = intel_gpio_create(dev_priv, i); + } + + intel_i2c_reset(dev_priv->dev); + + return 0; + +err: + while (--i) { + struct intel_gmbus *bus = &dev_priv->gmbus[i]; + i2c_del_adapter(&bus->adapter); + } + kfree(dev_priv->gmbus); + dev_priv->gmbus = NULL; + return ret; +} + +void intel_gmbus_set_speed(struct i2c_adapter *adapter, int speed) +{ + struct intel_gmbus *bus = to_intel_gmbus(adapter); + + /* speed: + * 0x0 = 100 KHz + * 0x1 = 50 KHz + * 0x2 = 400 KHz + * 0x3 = 1000 Khz + */ + bus->reg0 = (bus->reg0 & ~(0x3 << 8)) | (speed << 8); +} + +void intel_gmbus_force_bit(struct i2c_adapter *adapter, bool force_bit) +{ + struct intel_gmbus *bus = to_intel_gmbus(adapter); + + if (force_bit) { + if (bus->force_bit == NULL) { + struct drm_i915_private *dev_priv = adapter->algo_data; + bus->force_bit = intel_gpio_create(dev_priv, + bus->reg0 & 0xff); + } + } else { + if (bus->force_bit) { + i2c_del_adapter(bus->force_bit); + kfree(bus->force_bit); + bus->force_bit = NULL; + } + } +} + +void intel_teardown_gmbus(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + int i; + + if (dev_priv->gmbus == NULL) + return; + + for (i = 0; i < GMBUS_NUM_PORTS; i++) { + struct intel_gmbus *bus = &dev_priv->gmbus[i]; + if (bus->force_bit) { + i2c_del_adapter(bus->force_bit); + kfree(bus->force_bit); + } + i2c_del_adapter(&bus->adapter); + } + + kfree(dev_priv->gmbus); + dev_priv->gmbus = NULL; +} diff --git a/drivers/gpu/drm/i915/intel_lvds.c b/drivers/gpu/drm/i915/intel_lvds.c new file mode 100644 index 00000000..ff85a91e --- /dev/null +++ b/drivers/gpu/drm/i915/intel_lvds.c @@ -0,0 +1,1069 @@ +/* + * Copyright © 2006-2007 Intel Corporation + * Copyright (c) 2006 Dave Airlie + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Eric Anholt + * Dave Airlie + * Jesse Barnes + */ + +#include +#include +#include +#include +#include "drmP.h" +#include "drm.h" +#include "drm_crtc.h" +#include "drm_edid.h" +#include "intel_drv.h" +#include "i915_drm.h" +#include "i915_drv.h" +#include + +/* Private structure for the integrated LVDS support */ +struct intel_lvds { + struct intel_encoder base; + + struct edid *edid; + + int fitting_mode; + u32 pfit_control; + u32 pfit_pgm_ratios; + bool pfit_dirty; + + struct drm_display_mode *fixed_mode; +}; + +static struct intel_lvds *to_intel_lvds(struct drm_encoder *encoder) +{ + return container_of(encoder, struct intel_lvds, base.base); +} + +static struct intel_lvds *intel_attached_lvds(struct drm_connector *connector) +{ + return container_of(intel_attached_encoder(connector), + struct intel_lvds, base); +} + +/** + * Sets the power state for the panel. + */ +static void intel_lvds_enable(struct intel_lvds *intel_lvds) +{ + struct drm_device *dev = intel_lvds->base.base.dev; + struct drm_i915_private *dev_priv = dev->dev_private; + u32 ctl_reg, lvds_reg; + + if (HAS_PCH_SPLIT(dev)) { + ctl_reg = PCH_PP_CONTROL; + lvds_reg = PCH_LVDS; + } else { + ctl_reg = PP_CONTROL; + lvds_reg = LVDS; + } + + I915_WRITE(lvds_reg, I915_READ(lvds_reg) | LVDS_PORT_EN); + + if (intel_lvds->pfit_dirty) { + /* + * Enable automatic panel scaling so that non-native modes + * fill the screen. The panel fitter should only be + * adjusted whilst the pipe is disabled, according to + * register description and PRM. + */ + DRM_DEBUG_KMS("applying panel-fitter: %x, %x\n", + intel_lvds->pfit_control, + intel_lvds->pfit_pgm_ratios); + if (wait_for((I915_READ(PP_STATUS) & PP_ON) == 0, 1000)) { + DRM_ERROR("timed out waiting for panel to power off\n"); + } else { + I915_WRITE(PFIT_PGM_RATIOS, intel_lvds->pfit_pgm_ratios); + I915_WRITE(PFIT_CONTROL, intel_lvds->pfit_control); + intel_lvds->pfit_dirty = false; + } + } + + I915_WRITE(ctl_reg, I915_READ(ctl_reg) | POWER_TARGET_ON); + POSTING_READ(lvds_reg); + + intel_panel_enable_backlight(dev); +} + +static void intel_lvds_disable(struct intel_lvds *intel_lvds) +{ + struct drm_device *dev = intel_lvds->base.base.dev; + struct drm_i915_private *dev_priv = dev->dev_private; + u32 ctl_reg, lvds_reg; + + if (HAS_PCH_SPLIT(dev)) { + ctl_reg = PCH_PP_CONTROL; + lvds_reg = PCH_LVDS; + } else { + ctl_reg = PP_CONTROL; + lvds_reg = LVDS; + } + + intel_panel_disable_backlight(dev); + + I915_WRITE(ctl_reg, I915_READ(ctl_reg) & ~POWER_TARGET_ON); + + if (intel_lvds->pfit_control) { + if (wait_for((I915_READ(PP_STATUS) & PP_ON) == 0, 1000)) + DRM_ERROR("timed out waiting for panel to power off\n"); + + I915_WRITE(PFIT_CONTROL, 0); + intel_lvds->pfit_dirty = true; + } + + I915_WRITE(lvds_reg, I915_READ(lvds_reg) & ~LVDS_PORT_EN); + POSTING_READ(lvds_reg); +} + +static void intel_lvds_dpms(struct drm_encoder *encoder, int mode) +{ + struct intel_lvds *intel_lvds = to_intel_lvds(encoder); + + if (mode == DRM_MODE_DPMS_ON) + intel_lvds_enable(intel_lvds); + else + intel_lvds_disable(intel_lvds); + + /* XXX: We never power down the LVDS pairs. */ +} + +static int intel_lvds_mode_valid(struct drm_connector *connector, + struct drm_display_mode *mode) +{ + struct intel_lvds *intel_lvds = intel_attached_lvds(connector); + struct drm_display_mode *fixed_mode = intel_lvds->fixed_mode; + + if (mode->hdisplay > fixed_mode->hdisplay) + return MODE_PANEL; + if (mode->vdisplay > fixed_mode->vdisplay) + return MODE_PANEL; + + return MODE_OK; +} + +static void +centre_horizontally(struct drm_display_mode *mode, + int width) +{ + u32 border, sync_pos, blank_width, sync_width; + + /* keep the hsync and hblank widths constant */ + sync_width = mode->crtc_hsync_end - mode->crtc_hsync_start; + blank_width = mode->crtc_hblank_end - mode->crtc_hblank_start; + sync_pos = (blank_width - sync_width + 1) / 2; + + border = (mode->hdisplay - width + 1) / 2; + border += border & 1; /* make the border even */ + + mode->crtc_hdisplay = width; + mode->crtc_hblank_start = width + border; + mode->crtc_hblank_end = mode->crtc_hblank_start + blank_width; + + mode->crtc_hsync_start = mode->crtc_hblank_start + sync_pos; + mode->crtc_hsync_end = mode->crtc_hsync_start + sync_width; +} + +static void +centre_vertically(struct drm_display_mode *mode, + int height) +{ + u32 border, sync_pos, blank_width, sync_width; + + /* keep the vsync and vblank widths constant */ + sync_width = mode->crtc_vsync_end - mode->crtc_vsync_start; + blank_width = mode->crtc_vblank_end - mode->crtc_vblank_start; + sync_pos = (blank_width - sync_width + 1) / 2; + + border = (mode->vdisplay - height + 1) / 2; + + mode->crtc_vdisplay = height; + mode->crtc_vblank_start = height + border; + mode->crtc_vblank_end = mode->crtc_vblank_start + blank_width; + + mode->crtc_vsync_start = mode->crtc_vblank_start + sync_pos; + mode->crtc_vsync_end = mode->crtc_vsync_start + sync_width; +} + +static inline u32 panel_fitter_scaling(u32 source, u32 target) +{ + /* + * Floating point operation is not supported. So the FACTOR + * is defined, which can avoid the floating point computation + * when calculating the panel ratio. + */ +#define ACCURACY 12 +#define FACTOR (1 << ACCURACY) + u32 ratio = source * FACTOR / target; + return (FACTOR * ratio + FACTOR/2) / FACTOR; +} + +static bool intel_lvds_mode_fixup(struct drm_encoder *encoder, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct drm_device *dev = encoder->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(encoder->crtc); + struct intel_lvds *intel_lvds = to_intel_lvds(encoder); + struct drm_encoder *tmp_encoder; + u32 pfit_control = 0, pfit_pgm_ratios = 0, border = 0; + int pipe; + + /* Should never happen!! */ + if (INTEL_INFO(dev)->gen < 4 && intel_crtc->pipe == 0) { + DRM_ERROR("Can't support LVDS on pipe A\n"); + return false; + } + + /* Should never happen!! */ + list_for_each_entry(tmp_encoder, &dev->mode_config.encoder_list, head) { + if (tmp_encoder != encoder && tmp_encoder->crtc == encoder->crtc) { + DRM_ERROR("Can't enable LVDS and another " + "encoder on the same pipe\n"); + return false; + } + } + + /* + * We have timings from the BIOS for the panel, put them in + * to the adjusted mode. The CRTC will be set up for this mode, + * with the panel scaling set up to source from the H/VDisplay + * of the original mode. + */ + intel_fixed_panel_mode(intel_lvds->fixed_mode, adjusted_mode); + + if (HAS_PCH_SPLIT(dev)) { + intel_pch_panel_fitting(dev, intel_lvds->fitting_mode, + mode, adjusted_mode); + return true; + } + + /* Native modes don't need fitting */ + if (adjusted_mode->hdisplay == mode->hdisplay && + adjusted_mode->vdisplay == mode->vdisplay) + goto out; + + /* 965+ wants fuzzy fitting */ + if (INTEL_INFO(dev)->gen >= 4) + pfit_control |= ((intel_crtc->pipe << PFIT_PIPE_SHIFT) | + PFIT_FILTER_FUZZY); + + /* + * Enable automatic panel scaling for non-native modes so that they fill + * the screen. Should be enabled before the pipe is enabled, according + * to register description and PRM. + * Change the value here to see the borders for debugging + */ + for_each_pipe(pipe) + I915_WRITE(BCLRPAT(pipe), 0); + + switch (intel_lvds->fitting_mode) { + case DRM_MODE_SCALE_CENTER: + /* + * For centered modes, we have to calculate border widths & + * heights and modify the values programmed into the CRTC. + */ + centre_horizontally(adjusted_mode, mode->hdisplay); + centre_vertically(adjusted_mode, mode->vdisplay); + border = LVDS_BORDER_ENABLE; + break; + + case DRM_MODE_SCALE_ASPECT: + /* Scale but preserve the aspect ratio */ + if (INTEL_INFO(dev)->gen >= 4) { + u32 scaled_width = adjusted_mode->hdisplay * mode->vdisplay; + u32 scaled_height = mode->hdisplay * adjusted_mode->vdisplay; + + /* 965+ is easy, it does everything in hw */ + if (scaled_width > scaled_height) + pfit_control |= PFIT_ENABLE | PFIT_SCALING_PILLAR; + else if (scaled_width < scaled_height) + pfit_control |= PFIT_ENABLE | PFIT_SCALING_LETTER; + else if (adjusted_mode->hdisplay != mode->hdisplay) + pfit_control |= PFIT_ENABLE | PFIT_SCALING_AUTO; + } else { + u32 scaled_width = adjusted_mode->hdisplay * mode->vdisplay; + u32 scaled_height = mode->hdisplay * adjusted_mode->vdisplay; + /* + * For earlier chips we have to calculate the scaling + * ratio by hand and program it into the + * PFIT_PGM_RATIO register + */ + if (scaled_width > scaled_height) { /* pillar */ + centre_horizontally(adjusted_mode, scaled_height / mode->vdisplay); + + border = LVDS_BORDER_ENABLE; + if (mode->vdisplay != adjusted_mode->vdisplay) { + u32 bits = panel_fitter_scaling(mode->vdisplay, adjusted_mode->vdisplay); + pfit_pgm_ratios |= (bits << PFIT_HORIZ_SCALE_SHIFT | + bits << PFIT_VERT_SCALE_SHIFT); + pfit_control |= (PFIT_ENABLE | + VERT_INTERP_BILINEAR | + HORIZ_INTERP_BILINEAR); + } + } else if (scaled_width < scaled_height) { /* letter */ + centre_vertically(adjusted_mode, scaled_width / mode->hdisplay); + + border = LVDS_BORDER_ENABLE; + if (mode->hdisplay != adjusted_mode->hdisplay) { + u32 bits = panel_fitter_scaling(mode->hdisplay, adjusted_mode->hdisplay); + pfit_pgm_ratios |= (bits << PFIT_HORIZ_SCALE_SHIFT | + bits << PFIT_VERT_SCALE_SHIFT); + pfit_control |= (PFIT_ENABLE | + VERT_INTERP_BILINEAR | + HORIZ_INTERP_BILINEAR); + } + } else + /* Aspects match, Let hw scale both directions */ + pfit_control |= (PFIT_ENABLE | + VERT_AUTO_SCALE | HORIZ_AUTO_SCALE | + VERT_INTERP_BILINEAR | + HORIZ_INTERP_BILINEAR); + } + break; + + case DRM_MODE_SCALE_FULLSCREEN: + /* + * Full scaling, even if it changes the aspect ratio. + * Fortunately this is all done for us in hw. + */ + if (mode->vdisplay != adjusted_mode->vdisplay || + mode->hdisplay != adjusted_mode->hdisplay) { + pfit_control |= PFIT_ENABLE; + if (INTEL_INFO(dev)->gen >= 4) + pfit_control |= PFIT_SCALING_AUTO; + else + pfit_control |= (VERT_AUTO_SCALE | + VERT_INTERP_BILINEAR | + HORIZ_AUTO_SCALE | + HORIZ_INTERP_BILINEAR); + } + break; + + default: + break; + } + +out: + /* If not enabling scaling, be consistent and always use 0. */ + if ((pfit_control & PFIT_ENABLE) == 0) { + pfit_control = 0; + pfit_pgm_ratios = 0; + } + + /* Make sure pre-965 set dither correctly */ + if (INTEL_INFO(dev)->gen < 4 && dev_priv->lvds_dither) + pfit_control |= PANEL_8TO6_DITHER_ENABLE; + + if (pfit_control != intel_lvds->pfit_control || + pfit_pgm_ratios != intel_lvds->pfit_pgm_ratios) { + intel_lvds->pfit_control = pfit_control; + intel_lvds->pfit_pgm_ratios = pfit_pgm_ratios; + intel_lvds->pfit_dirty = true; + } + dev_priv->lvds_border_bits = border; + + /* + * XXX: It would be nice to support lower refresh rates on the + * panels to reduce power consumption, and perhaps match the + * user's requested refresh rate. + */ + + return true; +} + +static void intel_lvds_prepare(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_lvds *intel_lvds = to_intel_lvds(encoder); + + /* We try to do the minimum that is necessary in order to unlock + * the registers for mode setting. + * + * On Ironlake, this is quite simple as we just set the unlock key + * and ignore all subtleties. (This may cause some issues...) + * + * Prior to Ironlake, we must disable the pipe if we want to adjust + * the panel fitter. However at all other times we can just reset + * the registers regardless. + */ + + if (HAS_PCH_SPLIT(dev)) { + I915_WRITE(PCH_PP_CONTROL, + I915_READ(PCH_PP_CONTROL) | PANEL_UNLOCK_REGS); + } else if (intel_lvds->pfit_dirty) { + I915_WRITE(PP_CONTROL, + (I915_READ(PP_CONTROL) | PANEL_UNLOCK_REGS) + & ~POWER_TARGET_ON); + } else { + I915_WRITE(PP_CONTROL, + I915_READ(PP_CONTROL) | PANEL_UNLOCK_REGS); + } +} + +static void intel_lvds_commit(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_lvds *intel_lvds = to_intel_lvds(encoder); + + /* Undo any unlocking done in prepare to prevent accidental + * adjustment of the registers. + */ + if (HAS_PCH_SPLIT(dev)) { + u32 val = I915_READ(PCH_PP_CONTROL); + if ((val & PANEL_UNLOCK_REGS) == PANEL_UNLOCK_REGS) + I915_WRITE(PCH_PP_CONTROL, val & 0x3); + } else { + u32 val = I915_READ(PP_CONTROL); + if ((val & PANEL_UNLOCK_REGS) == PANEL_UNLOCK_REGS) + I915_WRITE(PP_CONTROL, val & 0x3); + } + + /* Always do a full power on as we do not know what state + * we were left in. + */ + intel_lvds_enable(intel_lvds); +} + +static void intel_lvds_mode_set(struct drm_encoder *encoder, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + /* + * The LVDS pin pair will already have been turned on in the + * intel_crtc_mode_set since it has a large impact on the DPLL + * settings. + */ +} + +/** + * Detect the LVDS connection. + * + * Since LVDS doesn't have hotlug, we use the lid as a proxy. Open means + * connected and closed means disconnected. We also send hotplug events as + * needed, using lid status notification from the input layer. + */ +static enum drm_connector_status +intel_lvds_detect(struct drm_connector *connector, bool force) +{ + struct drm_device *dev = connector->dev; + enum drm_connector_status status; + + status = intel_panel_detect(dev); + if (status != connector_status_unknown) + return status; + + return connector_status_connected; +} + +/** + * Return the list of DDC modes if available, or the BIOS fixed mode otherwise. + */ +static int intel_lvds_get_modes(struct drm_connector *connector) +{ + struct intel_lvds *intel_lvds = intel_attached_lvds(connector); + struct drm_device *dev = connector->dev; + struct drm_display_mode *mode; + + if (intel_lvds->edid) + return drm_add_edid_modes(connector, intel_lvds->edid); + + mode = drm_mode_duplicate(dev, intel_lvds->fixed_mode); + if (mode == NULL) + return 0; + + drm_mode_probed_add(connector, mode); + return 1; +} + +static int intel_no_modeset_on_lid_dmi_callback(const struct dmi_system_id *id) +{ + DRM_DEBUG_KMS("Skipping forced modeset for %s\n", id->ident); + return 1; +} + +/* The GPU hangs up on these systems if modeset is performed on LID open */ +static const struct dmi_system_id intel_no_modeset_on_lid[] = { + { + .callback = intel_no_modeset_on_lid_dmi_callback, + .ident = "Toshiba Tecra A11", + .matches = { + DMI_MATCH(DMI_SYS_VENDOR, "TOSHIBA"), + DMI_MATCH(DMI_PRODUCT_NAME, "TECRA A11"), + }, + }, + + { } /* terminating entry */ +}; + +/* + * Lid events. Note the use of 'modeset_on_lid': + * - we set it on lid close, and reset it on open + * - we use it as a "only once" bit (ie we ignore + * duplicate events where it was already properly + * set/reset) + * - the suspend/resume paths will also set it to + * zero, since they restore the mode ("lid open"). + */ +static int intel_lid_notify(struct notifier_block *nb, unsigned long val, + void *unused) +{ + struct drm_i915_private *dev_priv = + container_of(nb, struct drm_i915_private, lid_notifier); + struct drm_device *dev = dev_priv->dev; + struct drm_connector *connector = dev_priv->int_lvds_connector; + + if (dev->switch_power_state != DRM_SWITCH_POWER_ON) + return NOTIFY_OK; + + /* + * check and update the status of LVDS connector after receiving + * the LID nofication event. + */ + if (connector) + connector->status = connector->funcs->detect(connector, + false); + + /* Don't force modeset on machines where it causes a GPU lockup */ + if (dmi_check_system(intel_no_modeset_on_lid)) + return NOTIFY_OK; + if (!acpi_lid_open()) { + dev_priv->modeset_on_lid = 1; + return NOTIFY_OK; + } + + if (!dev_priv->modeset_on_lid) + return NOTIFY_OK; + + dev_priv->modeset_on_lid = 0; + + mutex_lock(&dev->mode_config.mutex); + drm_helper_resume_force_mode(dev); + mutex_unlock(&dev->mode_config.mutex); + + return NOTIFY_OK; +} + +/** + * intel_lvds_destroy - unregister and free LVDS structures + * @connector: connector to free + * + * Unregister the DDC bus for this connector then free the driver private + * structure. + */ +static void intel_lvds_destroy(struct drm_connector *connector) +{ + struct drm_device *dev = connector->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + + if (dev_priv->lid_notifier.notifier_call) + acpi_lid_notifier_unregister(&dev_priv->lid_notifier); + drm_sysfs_connector_remove(connector); + drm_connector_cleanup(connector); + kfree(connector); +} + +static int intel_lvds_set_property(struct drm_connector *connector, + struct drm_property *property, + uint64_t value) +{ + struct intel_lvds *intel_lvds = intel_attached_lvds(connector); + struct drm_device *dev = connector->dev; + + if (property == dev->mode_config.scaling_mode_property) { + struct drm_crtc *crtc = intel_lvds->base.base.crtc; + + if (value == DRM_MODE_SCALE_NONE) { + DRM_DEBUG_KMS("no scaling not supported\n"); + return -EINVAL; + } + + if (intel_lvds->fitting_mode == value) { + /* the LVDS scaling property is not changed */ + return 0; + } + intel_lvds->fitting_mode = value; + if (crtc && crtc->enabled) { + /* + * If the CRTC is enabled, the display will be changed + * according to the new panel fitting mode. + */ + drm_crtc_helper_set_mode(crtc, &crtc->mode, + crtc->x, crtc->y, crtc->fb); + } + } + + return 0; +} + +static const struct drm_encoder_helper_funcs intel_lvds_helper_funcs = { + .dpms = intel_lvds_dpms, + .mode_fixup = intel_lvds_mode_fixup, + .prepare = intel_lvds_prepare, + .mode_set = intel_lvds_mode_set, + .commit = intel_lvds_commit, +}; + +static const struct drm_connector_helper_funcs intel_lvds_connector_helper_funcs = { + .get_modes = intel_lvds_get_modes, + .mode_valid = intel_lvds_mode_valid, + .best_encoder = intel_best_encoder, +}; + +static const struct drm_connector_funcs intel_lvds_connector_funcs = { + .dpms = drm_helper_connector_dpms, + .detect = intel_lvds_detect, + .fill_modes = drm_helper_probe_single_connector_modes, + .set_property = intel_lvds_set_property, + .destroy = intel_lvds_destroy, +}; + +static const struct drm_encoder_funcs intel_lvds_enc_funcs = { + .destroy = intel_encoder_destroy, +}; + +static int __init intel_no_lvds_dmi_callback(const struct dmi_system_id *id) +{ + DRM_DEBUG_KMS("Skipping LVDS initialization for %s\n", id->ident); + return 1; +} + +/* These systems claim to have LVDS, but really don't */ +static const struct dmi_system_id intel_no_lvds[] = { + { + .callback = intel_no_lvds_dmi_callback, + .ident = "Apple Mac Mini (Core series)", + .matches = { + DMI_MATCH(DMI_SYS_VENDOR, "Apple"), + DMI_MATCH(DMI_PRODUCT_NAME, "Macmini1,1"), + }, + }, + { + .callback = intel_no_lvds_dmi_callback, + .ident = "Apple Mac Mini (Core 2 series)", + .matches = { + DMI_MATCH(DMI_SYS_VENDOR, "Apple"), + DMI_MATCH(DMI_PRODUCT_NAME, "Macmini2,1"), + }, + }, + { + .callback = intel_no_lvds_dmi_callback, + .ident = "MSI IM-945GSE-A", + .matches = { + DMI_MATCH(DMI_SYS_VENDOR, "MSI"), + DMI_MATCH(DMI_PRODUCT_NAME, "A9830IMS"), + }, + }, + { + .callback = intel_no_lvds_dmi_callback, + .ident = "Dell Studio Hybrid", + .matches = { + DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."), + DMI_MATCH(DMI_PRODUCT_NAME, "Studio Hybrid 140g"), + }, + }, + { + .callback = intel_no_lvds_dmi_callback, + .ident = "AOpen Mini PC", + .matches = { + DMI_MATCH(DMI_SYS_VENDOR, "AOpen"), + DMI_MATCH(DMI_PRODUCT_NAME, "i965GMx-IF"), + }, + }, + { + .callback = intel_no_lvds_dmi_callback, + .ident = "AOpen Mini PC MP915", + .matches = { + DMI_MATCH(DMI_BOARD_VENDOR, "AOpen"), + DMI_MATCH(DMI_BOARD_NAME, "i915GMx-F"), + }, + }, + { + .callback = intel_no_lvds_dmi_callback, + .ident = "AOpen i915GMm-HFS", + .matches = { + DMI_MATCH(DMI_BOARD_VENDOR, "AOpen"), + DMI_MATCH(DMI_BOARD_NAME, "i915GMm-HFS"), + }, + }, + { + .callback = intel_no_lvds_dmi_callback, + .ident = "AOpen i45GMx-I", + .matches = { + DMI_MATCH(DMI_BOARD_VENDOR, "AOpen"), + DMI_MATCH(DMI_BOARD_NAME, "i45GMx-I"), + }, + }, + { + .callback = intel_no_lvds_dmi_callback, + .ident = "Aopen i945GTt-VFA", + .matches = { + DMI_MATCH(DMI_PRODUCT_VERSION, "AO00001JW"), + }, + }, + { + .callback = intel_no_lvds_dmi_callback, + .ident = "Clientron U800", + .matches = { + DMI_MATCH(DMI_SYS_VENDOR, "Clientron"), + DMI_MATCH(DMI_PRODUCT_NAME, "U800"), + }, + }, + { + .callback = intel_no_lvds_dmi_callback, + .ident = "Asus EeeBox PC EB1007", + .matches = { + DMI_MATCH(DMI_SYS_VENDOR, "ASUSTeK Computer INC."), + DMI_MATCH(DMI_PRODUCT_NAME, "EB1007"), + }, + }, + { + .callback = intel_no_lvds_dmi_callback, + .ident = "MSI Wind Box DC500", + .matches = { + DMI_MATCH(DMI_BOARD_VENDOR, "MICRO-STAR INTERNATIONAL CO., LTD"), + DMI_MATCH(DMI_BOARD_NAME, "MS-7469"), + }, + }, + + { } /* terminating entry */ +}; + +/** + * intel_find_lvds_downclock - find the reduced downclock for LVDS in EDID + * @dev: drm device + * @connector: LVDS connector + * + * Find the reduced downclock for LVDS in EDID. + */ +static void intel_find_lvds_downclock(struct drm_device *dev, + struct drm_display_mode *fixed_mode, + struct drm_connector *connector) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_display_mode *scan; + int temp_downclock; + + temp_downclock = fixed_mode->clock; + list_for_each_entry(scan, &connector->probed_modes, head) { + /* + * If one mode has the same resolution with the fixed_panel + * mode while they have the different refresh rate, it means + * that the reduced downclock is found for the LVDS. In such + * case we can set the different FPx0/1 to dynamically select + * between low and high frequency. + */ + if (scan->hdisplay == fixed_mode->hdisplay && + scan->hsync_start == fixed_mode->hsync_start && + scan->hsync_end == fixed_mode->hsync_end && + scan->htotal == fixed_mode->htotal && + scan->vdisplay == fixed_mode->vdisplay && + scan->vsync_start == fixed_mode->vsync_start && + scan->vsync_end == fixed_mode->vsync_end && + scan->vtotal == fixed_mode->vtotal) { + if (scan->clock < temp_downclock) { + /* + * The downclock is already found. But we + * expect to find the lower downclock. + */ + temp_downclock = scan->clock; + } + } + } + if (temp_downclock < fixed_mode->clock && i915_lvds_downclock) { + /* We found the downclock for LVDS. */ + dev_priv->lvds_downclock_avail = 1; + dev_priv->lvds_downclock = temp_downclock; + DRM_DEBUG_KMS("LVDS downclock is found in EDID. " + "Normal clock %dKhz, downclock %dKhz\n", + fixed_mode->clock, temp_downclock); + } +} + +/* + * Enumerate the child dev array parsed from VBT to check whether + * the LVDS is present. + * If it is present, return 1. + * If it is not present, return false. + * If no child dev is parsed from VBT, it assumes that the LVDS is present. + */ +static bool lvds_is_present_in_vbt(struct drm_device *dev, + u8 *i2c_pin) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + int i; + + if (!dev_priv->child_dev_num) + return true; + + for (i = 0; i < dev_priv->child_dev_num; i++) { + struct child_device_config *child = dev_priv->child_dev + i; + + /* If the device type is not LFP, continue. + * We have to check both the new identifiers as well as the + * old for compatibility with some BIOSes. + */ + if (child->device_type != DEVICE_TYPE_INT_LFP && + child->device_type != DEVICE_TYPE_LFP) + continue; + + if (child->i2c_pin) + *i2c_pin = child->i2c_pin; + + /* However, we cannot trust the BIOS writers to populate + * the VBT correctly. Since LVDS requires additional + * information from AIM blocks, a non-zero addin offset is + * a good indicator that the LVDS is actually present. + */ + if (child->addin_offset) + return true; + + /* But even then some BIOS writers perform some black magic + * and instantiate the device without reference to any + * additional data. Trust that if the VBT was written into + * the OpRegion then they have validated the LVDS's existence. + */ + if (dev_priv->opregion.vbt) + return true; + } + + return false; +} + +/** + * intel_lvds_init - setup LVDS connectors on this device + * @dev: drm device + * + * Create the connector, register the LVDS DDC bus, and try to figure out what + * modes we can display on the LVDS panel (if present). + */ +bool intel_lvds_init(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_lvds *intel_lvds; + struct intel_encoder *intel_encoder; + struct intel_connector *intel_connector; + struct drm_connector *connector; + struct drm_encoder *encoder; + struct drm_display_mode *scan; /* *modes, *bios_mode; */ + struct drm_crtc *crtc; + u32 lvds; + int pipe; + u8 pin; + + /* Skip init on machines we know falsely report LVDS */ + if (dmi_check_system(intel_no_lvds)) + return false; + + pin = GMBUS_PORT_PANEL; + if (!lvds_is_present_in_vbt(dev, &pin)) { + DRM_DEBUG_KMS("LVDS is not present in VBT\n"); + return false; + } + + if (HAS_PCH_SPLIT(dev)) { + if ((I915_READ(PCH_LVDS) & LVDS_DETECTED) == 0) + return false; + if (dev_priv->edp.support) { + DRM_DEBUG_KMS("disable LVDS for eDP support\n"); + return false; + } + } + + intel_lvds = kzalloc(sizeof(struct intel_lvds), GFP_KERNEL); + if (!intel_lvds) { + return false; + } + + intel_connector = kzalloc(sizeof(struct intel_connector), GFP_KERNEL); + if (!intel_connector) { + kfree(intel_lvds); + return false; + } + + if (!HAS_PCH_SPLIT(dev)) { + intel_lvds->pfit_control = I915_READ(PFIT_CONTROL); + } + + intel_encoder = &intel_lvds->base; + encoder = &intel_encoder->base; + connector = &intel_connector->base; + drm_connector_init(dev, &intel_connector->base, &intel_lvds_connector_funcs, + DRM_MODE_CONNECTOR_LVDS); + + drm_encoder_init(dev, &intel_encoder->base, &intel_lvds_enc_funcs, + DRM_MODE_ENCODER_LVDS); + + intel_connector_attach_encoder(intel_connector, intel_encoder); + intel_encoder->type = INTEL_OUTPUT_LVDS; + + intel_encoder->clone_mask = (1 << INTEL_LVDS_CLONE_BIT); + intel_encoder->crtc_mask = (1 << 1); + if (INTEL_INFO(dev)->gen >= 5) + intel_encoder->crtc_mask |= (1 << 0); + drm_encoder_helper_add(encoder, &intel_lvds_helper_funcs); + drm_connector_helper_add(connector, &intel_lvds_connector_helper_funcs); + connector->display_info.subpixel_order = SubPixelHorizontalRGB; + connector->interlace_allowed = false; + connector->doublescan_allowed = false; + + /* create the scaling mode property */ + drm_mode_create_scaling_mode_property(dev); + /* + * the initial panel fitting mode will be FULL_SCREEN. + */ + + drm_connector_attach_property(&intel_connector->base, + dev->mode_config.scaling_mode_property, + DRM_MODE_SCALE_ASPECT); + intel_lvds->fitting_mode = DRM_MODE_SCALE_ASPECT; + /* + * LVDS discovery: + * 1) check for EDID on DDC + * 2) check for VBT data + * 3) check to see if LVDS is already on + * if none of the above, no panel + * 4) make sure lid is open + * if closed, act like it's not there for now + */ + + /* + * Attempt to get the fixed panel mode from DDC. Assume that the + * preferred mode is the right one. + */ + intel_lvds->edid = drm_get_edid(connector, + &dev_priv->gmbus[pin].adapter); + if (intel_lvds->edid) { + if (drm_add_edid_modes(connector, + intel_lvds->edid)) { + drm_mode_connector_update_edid_property(connector, + intel_lvds->edid); + } else { + kfree(intel_lvds->edid); + intel_lvds->edid = NULL; + } + } + if (!intel_lvds->edid) { + /* Didn't get an EDID, so + * Set wide sync ranges so we get all modes + * handed to valid_mode for checking + */ + connector->display_info.min_vfreq = 0; + connector->display_info.max_vfreq = 200; + connector->display_info.min_hfreq = 0; + connector->display_info.max_hfreq = 200; + } + + list_for_each_entry(scan, &connector->probed_modes, head) { + if (scan->type & DRM_MODE_TYPE_PREFERRED) { + intel_lvds->fixed_mode = + drm_mode_duplicate(dev, scan); + intel_find_lvds_downclock(dev, + intel_lvds->fixed_mode, + connector); + goto out; + } + } + + /* Failed to get EDID, what about VBT? */ + if (dev_priv->lfp_lvds_vbt_mode) { + intel_lvds->fixed_mode = + drm_mode_duplicate(dev, dev_priv->lfp_lvds_vbt_mode); + if (intel_lvds->fixed_mode) { + intel_lvds->fixed_mode->type |= + DRM_MODE_TYPE_PREFERRED; + goto out; + } + } + + /* + * If we didn't get EDID, try checking if the panel is already turned + * on. If so, assume that whatever is currently programmed is the + * correct mode. + */ + + /* Ironlake: FIXME if still fail, not try pipe mode now */ + if (HAS_PCH_SPLIT(dev)) + goto failed; + + lvds = I915_READ(LVDS); + pipe = (lvds & LVDS_PIPEB_SELECT) ? 1 : 0; + crtc = intel_get_crtc_for_pipe(dev, pipe); + + if (crtc && (lvds & LVDS_PORT_EN)) { + intel_lvds->fixed_mode = intel_crtc_mode_get(dev, crtc); + if (intel_lvds->fixed_mode) { + intel_lvds->fixed_mode->type |= + DRM_MODE_TYPE_PREFERRED; + goto out; + } + } + + /* If we still don't have a mode after all that, give up. */ + if (!intel_lvds->fixed_mode) + goto failed; + +out: + if (HAS_PCH_SPLIT(dev)) { + u32 pwm; + + pipe = (I915_READ(PCH_LVDS) & LVDS_PIPEB_SELECT) ? 1 : 0; + + /* make sure PWM is enabled and locked to the LVDS pipe */ + pwm = I915_READ(BLC_PWM_CPU_CTL2); + if (pipe == 0 && (pwm & PWM_PIPE_B)) + I915_WRITE(BLC_PWM_CPU_CTL2, pwm & ~PWM_ENABLE); + if (pipe) + pwm |= PWM_PIPE_B; + else + pwm &= ~PWM_PIPE_B; + I915_WRITE(BLC_PWM_CPU_CTL2, pwm | PWM_ENABLE); + + pwm = I915_READ(BLC_PWM_PCH_CTL1); + pwm |= PWM_PCH_ENABLE; + I915_WRITE(BLC_PWM_PCH_CTL1, pwm); + } + dev_priv->lid_notifier.notifier_call = intel_lid_notify; + if (acpi_lid_notifier_register(&dev_priv->lid_notifier)) { + DRM_DEBUG_KMS("lid notifier registration failed\n"); + dev_priv->lid_notifier.notifier_call = NULL; + } + /* keep the LVDS connector */ + dev_priv->int_lvds_connector = connector; + drm_sysfs_connector_add(connector); + return true; + +failed: + DRM_DEBUG_KMS("No LVDS modes found, disabling.\n"); + drm_connector_cleanup(connector); + drm_encoder_cleanup(encoder); + kfree(intel_lvds); + kfree(intel_connector); + return false; +} diff --git a/drivers/gpu/drm/i915/intel_modes.c b/drivers/gpu/drm/i915/intel_modes.c new file mode 100644 index 00000000..3b26a3ba --- /dev/null +++ b/drivers/gpu/drm/i915/intel_modes.c @@ -0,0 +1,142 @@ +/* + * Copyright (c) 2007 Dave Airlie + * Copyright (c) 2007, 2010 Intel Corporation + * Jesse Barnes + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + */ + +#include +#include +#include +#include "drmP.h" +#include "intel_drv.h" +#include "i915_drv.h" + +/** + * intel_ddc_probe + * + */ +bool intel_ddc_probe(struct intel_encoder *intel_encoder, int ddc_bus) +{ + struct drm_i915_private *dev_priv = intel_encoder->base.dev->dev_private; + u8 out_buf[] = { 0x0, 0x0}; + u8 buf[2]; + struct i2c_msg msgs[] = { + { + .addr = 0x50, + .flags = 0, + .len = 1, + .buf = out_buf, + }, + { + .addr = 0x50, + .flags = I2C_M_RD, + .len = 1, + .buf = buf, + } + }; + + return i2c_transfer(&dev_priv->gmbus[ddc_bus].adapter, msgs, 2) == 2; +} + +/** + * intel_ddc_get_modes - get modelist from monitor + * @connector: DRM connector device to use + * @adapter: i2c adapter + * + * Fetch the EDID information from @connector using the DDC bus. + */ +int intel_ddc_get_modes(struct drm_connector *connector, + struct i2c_adapter *adapter) +{ + struct edid *edid; + int ret = 0; + + edid = drm_get_edid(connector, adapter); + if (edid) { + drm_mode_connector_update_edid_property(connector, edid); + ret = drm_add_edid_modes(connector, edid); + connector->display_info.raw_edid = NULL; + kfree(edid); + } + + return ret; +} + +static const char *force_audio_names[] = { + "off", + "auto", + "on", +}; + +void +intel_attach_force_audio_property(struct drm_connector *connector) +{ + struct drm_device *dev = connector->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_property *prop; + int i; + + prop = dev_priv->force_audio_property; + if (prop == NULL) { + prop = drm_property_create(dev, DRM_MODE_PROP_ENUM, + "audio", + ARRAY_SIZE(force_audio_names)); + if (prop == NULL) + return; + + for (i = 0; i < ARRAY_SIZE(force_audio_names); i++) + drm_property_add_enum(prop, i, i-1, force_audio_names[i]); + + dev_priv->force_audio_property = prop; + } + drm_connector_attach_property(connector, prop, 0); +} + +static const char *broadcast_rgb_names[] = { + "Full", + "Limited 16:235", +}; + +void +intel_attach_broadcast_rgb_property(struct drm_connector *connector) +{ + struct drm_device *dev = connector->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_property *prop; + int i; + + prop = dev_priv->broadcast_rgb_property; + if (prop == NULL) { + prop = drm_property_create(dev, DRM_MODE_PROP_ENUM, + "Broadcast RGB", + ARRAY_SIZE(broadcast_rgb_names)); + if (prop == NULL) + return; + + for (i = 0; i < ARRAY_SIZE(broadcast_rgb_names); i++) + drm_property_add_enum(prop, i, i, broadcast_rgb_names[i]); + + dev_priv->broadcast_rgb_property = prop; + } + + drm_connector_attach_property(connector, prop, 0); +} diff --git a/drivers/gpu/drm/i915/intel_opregion.c b/drivers/gpu/drm/i915/intel_opregion.c new file mode 100644 index 00000000..d2c71042 --- /dev/null +++ b/drivers/gpu/drm/i915/intel_opregion.c @@ -0,0 +1,516 @@ +/* + * Copyright 2008 Intel Corporation + * Copyright 2008 Red Hat + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NON-INFRINGEMENT. IN NO EVENT SHALL INTEL AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN + * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + * + */ + +#include +#include +#include + +#include "drmP.h" +#include "i915_drm.h" +#include "i915_drv.h" +#include "intel_drv.h" + +#define PCI_ASLE 0xe4 +#define PCI_ASLS 0xfc + +#define OPREGION_HEADER_OFFSET 0 +#define OPREGION_ACPI_OFFSET 0x100 +#define ACPI_CLID 0x01ac /* current lid state indicator */ +#define ACPI_CDCK 0x01b0 /* current docking state indicator */ +#define OPREGION_SWSCI_OFFSET 0x200 +#define OPREGION_ASLE_OFFSET 0x300 +#define OPREGION_VBT_OFFSET 0x400 + +#define OPREGION_SIGNATURE "IntelGraphicsMem" +#define MBOX_ACPI (1<<0) +#define MBOX_SWSCI (1<<1) +#define MBOX_ASLE (1<<2) + +struct opregion_header { + u8 signature[16]; + u32 size; + u32 opregion_ver; + u8 bios_ver[32]; + u8 vbios_ver[16]; + u8 driver_ver[16]; + u32 mboxes; + u8 reserved[164]; +} __attribute__((packed)); + +/* OpRegion mailbox #1: public ACPI methods */ +struct opregion_acpi { + u32 drdy; /* driver readiness */ + u32 csts; /* notification status */ + u32 cevt; /* current event */ + u8 rsvd1[20]; + u32 didl[8]; /* supported display devices ID list */ + u32 cpdl[8]; /* currently presented display list */ + u32 cadl[8]; /* currently active display list */ + u32 nadl[8]; /* next active devices list */ + u32 aslp; /* ASL sleep time-out */ + u32 tidx; /* toggle table index */ + u32 chpd; /* current hotplug enable indicator */ + u32 clid; /* current lid state*/ + u32 cdck; /* current docking state */ + u32 sxsw; /* Sx state resume */ + u32 evts; /* ASL supported events */ + u32 cnot; /* current OS notification */ + u32 nrdy; /* driver status */ + u8 rsvd2[60]; +} __attribute__((packed)); + +/* OpRegion mailbox #2: SWSCI */ +struct opregion_swsci { + u32 scic; /* SWSCI command|status|data */ + u32 parm; /* command parameters */ + u32 dslp; /* driver sleep time-out */ + u8 rsvd[244]; +} __attribute__((packed)); + +/* OpRegion mailbox #3: ASLE */ +struct opregion_asle { + u32 ardy; /* driver readiness */ + u32 aslc; /* ASLE interrupt command */ + u32 tche; /* technology enabled indicator */ + u32 alsi; /* current ALS illuminance reading */ + u32 bclp; /* backlight brightness to set */ + u32 pfit; /* panel fitting state */ + u32 cblv; /* current brightness level */ + u16 bclm[20]; /* backlight level duty cycle mapping table */ + u32 cpfm; /* current panel fitting mode */ + u32 epfm; /* enabled panel fitting modes */ + u8 plut[74]; /* panel LUT and identifier */ + u32 pfmb; /* PWM freq and min brightness */ + u8 rsvd[102]; +} __attribute__((packed)); + +/* ASLE irq request bits */ +#define ASLE_SET_ALS_ILLUM (1 << 0) +#define ASLE_SET_BACKLIGHT (1 << 1) +#define ASLE_SET_PFIT (1 << 2) +#define ASLE_SET_PWM_FREQ (1 << 3) +#define ASLE_REQ_MSK 0xf + +/* response bits of ASLE irq request */ +#define ASLE_ALS_ILLUM_FAILED (1<<10) +#define ASLE_BACKLIGHT_FAILED (1<<12) +#define ASLE_PFIT_FAILED (1<<14) +#define ASLE_PWM_FREQ_FAILED (1<<16) + +/* ASLE backlight brightness to set */ +#define ASLE_BCLP_VALID (1<<31) +#define ASLE_BCLP_MSK (~(1<<31)) + +/* ASLE panel fitting request */ +#define ASLE_PFIT_VALID (1<<31) +#define ASLE_PFIT_CENTER (1<<0) +#define ASLE_PFIT_STRETCH_TEXT (1<<1) +#define ASLE_PFIT_STRETCH_GFX (1<<2) + +/* PWM frequency and minimum brightness */ +#define ASLE_PFMB_BRIGHTNESS_MASK (0xff) +#define ASLE_PFMB_BRIGHTNESS_VALID (1<<8) +#define ASLE_PFMB_PWM_MASK (0x7ffffe00) +#define ASLE_PFMB_PWM_VALID (1<<31) + +#define ASLE_CBLV_VALID (1<<31) + +#define ACPI_OTHER_OUTPUT (0<<8) +#define ACPI_VGA_OUTPUT (1<<8) +#define ACPI_TV_OUTPUT (2<<8) +#define ACPI_DIGITAL_OUTPUT (3<<8) +#define ACPI_LVDS_OUTPUT (4<<8) + +#ifdef CONFIG_ACPI +static u32 asle_set_backlight(struct drm_device *dev, u32 bclp) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct opregion_asle *asle = dev_priv->opregion.asle; + u32 max; + + if (!(bclp & ASLE_BCLP_VALID)) + return ASLE_BACKLIGHT_FAILED; + + bclp &= ASLE_BCLP_MSK; + if (bclp > 255) + return ASLE_BACKLIGHT_FAILED; + + max = intel_panel_get_max_backlight(dev); + intel_panel_set_backlight(dev, bclp * max / 255); + asle->cblv = (bclp*0x64)/0xff | ASLE_CBLV_VALID; + + return 0; +} + +static u32 asle_set_als_illum(struct drm_device *dev, u32 alsi) +{ + /* alsi is the current ALS reading in lux. 0 indicates below sensor + range, 0xffff indicates above sensor range. 1-0xfffe are valid */ + return 0; +} + +static u32 asle_set_pwm_freq(struct drm_device *dev, u32 pfmb) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + if (pfmb & ASLE_PFMB_PWM_VALID) { + u32 blc_pwm_ctl = I915_READ(BLC_PWM_CTL); + u32 pwm = pfmb & ASLE_PFMB_PWM_MASK; + blc_pwm_ctl &= BACKLIGHT_DUTY_CYCLE_MASK; + pwm = pwm >> 9; + /* FIXME - what do we do with the PWM? */ + } + return 0; +} + +static u32 asle_set_pfit(struct drm_device *dev, u32 pfit) +{ + /* Panel fitting is currently controlled by the X code, so this is a + noop until modesetting support works fully */ + if (!(pfit & ASLE_PFIT_VALID)) + return ASLE_PFIT_FAILED; + return 0; +} + +void intel_opregion_asle_intr(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct opregion_asle *asle = dev_priv->opregion.asle; + u32 asle_stat = 0; + u32 asle_req; + + if (!asle) + return; + + asle_req = asle->aslc & ASLE_REQ_MSK; + + if (!asle_req) { + DRM_DEBUG_DRIVER("non asle set request??\n"); + return; + } + + if (asle_req & ASLE_SET_ALS_ILLUM) + asle_stat |= asle_set_als_illum(dev, asle->alsi); + + if (asle_req & ASLE_SET_BACKLIGHT) + asle_stat |= asle_set_backlight(dev, asle->bclp); + + if (asle_req & ASLE_SET_PFIT) + asle_stat |= asle_set_pfit(dev, asle->pfit); + + if (asle_req & ASLE_SET_PWM_FREQ) + asle_stat |= asle_set_pwm_freq(dev, asle->pfmb); + + asle->aslc = asle_stat; +} + +/* Only present on Ironlake+ */ +void intel_opregion_gse_intr(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct opregion_asle *asle = dev_priv->opregion.asle; + u32 asle_stat = 0; + u32 asle_req; + + if (!asle) + return; + + asle_req = asle->aslc & ASLE_REQ_MSK; + + if (!asle_req) { + DRM_DEBUG_DRIVER("non asle set request??\n"); + return; + } + + if (asle_req & ASLE_SET_ALS_ILLUM) { + DRM_DEBUG_DRIVER("Illum is not supported\n"); + asle_stat |= ASLE_ALS_ILLUM_FAILED; + } + + if (asle_req & ASLE_SET_BACKLIGHT) + asle_stat |= asle_set_backlight(dev, asle->bclp); + + if (asle_req & ASLE_SET_PFIT) { + DRM_DEBUG_DRIVER("Pfit is not supported\n"); + asle_stat |= ASLE_PFIT_FAILED; + } + + if (asle_req & ASLE_SET_PWM_FREQ) { + DRM_DEBUG_DRIVER("PWM freq is not supported\n"); + asle_stat |= ASLE_PWM_FREQ_FAILED; + } + + asle->aslc = asle_stat; +} +#define ASLE_ALS_EN (1<<0) +#define ASLE_BLC_EN (1<<1) +#define ASLE_PFIT_EN (1<<2) +#define ASLE_PFMB_EN (1<<3) + +void intel_opregion_enable_asle(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct opregion_asle *asle = dev_priv->opregion.asle; + + if (asle) { + if (IS_MOBILE(dev)) + intel_enable_asle(dev); + + asle->tche = ASLE_ALS_EN | ASLE_BLC_EN | ASLE_PFIT_EN | + ASLE_PFMB_EN; + asle->ardy = 1; + } +} + +#define ACPI_EV_DISPLAY_SWITCH (1<<0) +#define ACPI_EV_LID (1<<1) +#define ACPI_EV_DOCK (1<<2) + +static struct intel_opregion *system_opregion; + +static int intel_opregion_video_event(struct notifier_block *nb, + unsigned long val, void *data) +{ + /* The only video events relevant to opregion are 0x80. These indicate + either a docking event, lid switch or display switch request. In + Linux, these are handled by the dock, button and video drivers. + We might want to fix the video driver to be opregion-aware in + future, but right now we just indicate to the firmware that the + request has been handled */ + + struct opregion_acpi *acpi; + + if (!system_opregion) + return NOTIFY_DONE; + + acpi = system_opregion->acpi; + acpi->csts = 0; + + return NOTIFY_OK; +} + +static struct notifier_block intel_opregion_notifier = { + .notifier_call = intel_opregion_video_event, +}; + +/* + * Initialise the DIDL field in opregion. This passes a list of devices to + * the firmware. Values are defined by section B.4.2 of the ACPI specification + * (version 3) + */ + +static void intel_didl_outputs(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_opregion *opregion = &dev_priv->opregion; + struct drm_connector *connector; + acpi_handle handle; + struct acpi_device *acpi_dev, *acpi_cdev, *acpi_video_bus = NULL; + unsigned long long device_id; + acpi_status status; + int i = 0; + + handle = DEVICE_ACPI_HANDLE(&dev->pdev->dev); + if (!handle || ACPI_FAILURE(acpi_bus_get_device(handle, &acpi_dev))) + return; + + if (acpi_is_video_device(acpi_dev)) + acpi_video_bus = acpi_dev; + else { + list_for_each_entry(acpi_cdev, &acpi_dev->children, node) { + if (acpi_is_video_device(acpi_cdev)) { + acpi_video_bus = acpi_cdev; + break; + } + } + } + + if (!acpi_video_bus) { + printk(KERN_WARNING "No ACPI video bus found\n"); + return; + } + + list_for_each_entry(acpi_cdev, &acpi_video_bus->children, node) { + if (i >= 8) { + dev_printk (KERN_ERR, &dev->pdev->dev, + "More than 8 outputs detected\n"); + return; + } + status = + acpi_evaluate_integer(acpi_cdev->handle, "_ADR", + NULL, &device_id); + if (ACPI_SUCCESS(status)) { + if (!device_id) + goto blind_set; + opregion->acpi->didl[i] = (u32)(device_id & 0x0f0f); + i++; + } + } + +end: + /* If fewer than 8 outputs, the list must be null terminated */ + if (i < 8) + opregion->acpi->didl[i] = 0; + return; + +blind_set: + i = 0; + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + int output_type = ACPI_OTHER_OUTPUT; + if (i >= 8) { + dev_printk (KERN_ERR, &dev->pdev->dev, + "More than 8 outputs detected\n"); + return; + } + switch (connector->connector_type) { + case DRM_MODE_CONNECTOR_VGA: + case DRM_MODE_CONNECTOR_DVIA: + output_type = ACPI_VGA_OUTPUT; + break; + case DRM_MODE_CONNECTOR_Composite: + case DRM_MODE_CONNECTOR_SVIDEO: + case DRM_MODE_CONNECTOR_Component: + case DRM_MODE_CONNECTOR_9PinDIN: + output_type = ACPI_TV_OUTPUT; + break; + case DRM_MODE_CONNECTOR_DVII: + case DRM_MODE_CONNECTOR_DVID: + case DRM_MODE_CONNECTOR_DisplayPort: + case DRM_MODE_CONNECTOR_HDMIA: + case DRM_MODE_CONNECTOR_HDMIB: + output_type = ACPI_DIGITAL_OUTPUT; + break; + case DRM_MODE_CONNECTOR_LVDS: + output_type = ACPI_LVDS_OUTPUT; + break; + } + opregion->acpi->didl[i] |= (1<<31) | output_type | i; + i++; + } + goto end; +} + +void intel_opregion_init(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_opregion *opregion = &dev_priv->opregion; + + if (!opregion->header) + return; + + if (opregion->acpi) { + if (drm_core_check_feature(dev, DRIVER_MODESET)) + intel_didl_outputs(dev); + + /* Notify BIOS we are ready to handle ACPI video ext notifs. + * Right now, all the events are handled by the ACPI video module. + * We don't actually need to do anything with them. */ + opregion->acpi->csts = 0; + opregion->acpi->drdy = 1; + + system_opregion = opregion; + register_acpi_notifier(&intel_opregion_notifier); + } + + if (opregion->asle) + intel_opregion_enable_asle(dev); +} + +void intel_opregion_fini(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_opregion *opregion = &dev_priv->opregion; + + if (!opregion->header) + return; + + if (opregion->acpi) { + opregion->acpi->drdy = 0; + + system_opregion = NULL; + unregister_acpi_notifier(&intel_opregion_notifier); + } + + /* just clear all opregion memory pointers now */ + iounmap(opregion->header); + opregion->header = NULL; + opregion->acpi = NULL; + opregion->swsci = NULL; + opregion->asle = NULL; + opregion->vbt = NULL; +} +#endif + +int intel_opregion_setup(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_opregion *opregion = &dev_priv->opregion; + void *base; + u32 asls, mboxes; + int err = 0; + + pci_read_config_dword(dev->pdev, PCI_ASLS, &asls); + DRM_DEBUG_DRIVER("graphic opregion physical addr: 0x%x\n", asls); + if (asls == 0) { + DRM_DEBUG_DRIVER("ACPI OpRegion not supported!\n"); + return -ENOTSUPP; + } + + base = acpi_os_ioremap(asls, OPREGION_SIZE); + if (!base) + return -ENOMEM; + + if (memcmp(base, OPREGION_SIGNATURE, 16)) { + DRM_DEBUG_DRIVER("opregion signature mismatch\n"); + err = -EINVAL; + goto err_out; + } + opregion->header = base; + opregion->vbt = base + OPREGION_VBT_OFFSET; + + opregion->lid_state = base + ACPI_CLID; + + mboxes = opregion->header->mboxes; + if (mboxes & MBOX_ACPI) { + DRM_DEBUG_DRIVER("Public ACPI methods supported\n"); + opregion->acpi = base + OPREGION_ACPI_OFFSET; + } + + if (mboxes & MBOX_SWSCI) { + DRM_DEBUG_DRIVER("SWSCI supported\n"); + opregion->swsci = base + OPREGION_SWSCI_OFFSET; + } + if (mboxes & MBOX_ASLE) { + DRM_DEBUG_DRIVER("ASLE supported\n"); + opregion->asle = base + OPREGION_ASLE_OFFSET; + } + + return 0; + +err_out: + iounmap(base); + return err; +} diff --git a/drivers/gpu/drm/i915/intel_overlay.c b/drivers/gpu/drm/i915/intel_overlay.c new file mode 100644 index 00000000..9e2959bc --- /dev/null +++ b/drivers/gpu/drm/i915/intel_overlay.c @@ -0,0 +1,1617 @@ +/* + * Copyright © 2009 + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + * + * Authors: + * Daniel Vetter + * + * Derived from Xorg ddx, xf86-video-intel, src/i830_video.c + */ + +#include +#include "drmP.h" +#include "drm.h" +#include "i915_drm.h" +#include "i915_drv.h" +#include "i915_reg.h" +#include "intel_drv.h" + +/* Limits for overlay size. According to intel doc, the real limits are: + * Y width: 4095, UV width (planar): 2047, Y height: 2047, + * UV width (planar): * 1023. But the xorg thinks 2048 for height and width. Use + * the mininum of both. */ +#define IMAGE_MAX_WIDTH 2048 +#define IMAGE_MAX_HEIGHT 2046 /* 2 * 1023 */ +/* on 830 and 845 these large limits result in the card hanging */ +#define IMAGE_MAX_WIDTH_LEGACY 1024 +#define IMAGE_MAX_HEIGHT_LEGACY 1088 + +/* overlay register definitions */ +/* OCMD register */ +#define OCMD_TILED_SURFACE (0x1<<19) +#define OCMD_MIRROR_MASK (0x3<<17) +#define OCMD_MIRROR_MODE (0x3<<17) +#define OCMD_MIRROR_HORIZONTAL (0x1<<17) +#define OCMD_MIRROR_VERTICAL (0x2<<17) +#define OCMD_MIRROR_BOTH (0x3<<17) +#define OCMD_BYTEORDER_MASK (0x3<<14) /* zero for YUYV or FOURCC YUY2 */ +#define OCMD_UV_SWAP (0x1<<14) /* YVYU */ +#define OCMD_Y_SWAP (0x2<<14) /* UYVY or FOURCC UYVY */ +#define OCMD_Y_AND_UV_SWAP (0x3<<14) /* VYUY */ +#define OCMD_SOURCE_FORMAT_MASK (0xf<<10) +#define OCMD_RGB_888 (0x1<<10) /* not in i965 Intel docs */ +#define OCMD_RGB_555 (0x2<<10) /* not in i965 Intel docs */ +#define OCMD_RGB_565 (0x3<<10) /* not in i965 Intel docs */ +#define OCMD_YUV_422_PACKED (0x8<<10) +#define OCMD_YUV_411_PACKED (0x9<<10) /* not in i965 Intel docs */ +#define OCMD_YUV_420_PLANAR (0xc<<10) +#define OCMD_YUV_422_PLANAR (0xd<<10) +#define OCMD_YUV_410_PLANAR (0xe<<10) /* also 411 */ +#define OCMD_TVSYNCFLIP_PARITY (0x1<<9) +#define OCMD_TVSYNCFLIP_ENABLE (0x1<<7) +#define OCMD_BUF_TYPE_MASK (0x1<<5) +#define OCMD_BUF_TYPE_FRAME (0x0<<5) +#define OCMD_BUF_TYPE_FIELD (0x1<<5) +#define OCMD_TEST_MODE (0x1<<4) +#define OCMD_BUFFER_SELECT (0x3<<2) +#define OCMD_BUFFER0 (0x0<<2) +#define OCMD_BUFFER1 (0x1<<2) +#define OCMD_FIELD_SELECT (0x1<<2) +#define OCMD_FIELD0 (0x0<<1) +#define OCMD_FIELD1 (0x1<<1) +#define OCMD_ENABLE (0x1<<0) + +/* OCONFIG register */ +#define OCONF_PIPE_MASK (0x1<<18) +#define OCONF_PIPE_A (0x0<<18) +#define OCONF_PIPE_B (0x1<<18) +#define OCONF_GAMMA2_ENABLE (0x1<<16) +#define OCONF_CSC_MODE_BT601 (0x0<<5) +#define OCONF_CSC_MODE_BT709 (0x1<<5) +#define OCONF_CSC_BYPASS (0x1<<4) +#define OCONF_CC_OUT_8BIT (0x1<<3) +#define OCONF_TEST_MODE (0x1<<2) +#define OCONF_THREE_LINE_BUFFER (0x1<<0) +#define OCONF_TWO_LINE_BUFFER (0x0<<0) + +/* DCLRKM (dst-key) register */ +#define DST_KEY_ENABLE (0x1<<31) +#define CLK_RGB24_MASK 0x0 +#define CLK_RGB16_MASK 0x070307 +#define CLK_RGB15_MASK 0x070707 +#define CLK_RGB8I_MASK 0xffffff + +#define RGB16_TO_COLORKEY(c) \ + (((c & 0xF800) << 8) | ((c & 0x07E0) << 5) | ((c & 0x001F) << 3)) +#define RGB15_TO_COLORKEY(c) \ + (((c & 0x7c00) << 9) | ((c & 0x03E0) << 6) | ((c & 0x001F) << 3)) + +/* overlay flip addr flag */ +#define OFC_UPDATE 0x1 + +/* polyphase filter coefficients */ +#define N_HORIZ_Y_TAPS 5 +#define N_VERT_Y_TAPS 3 +#define N_HORIZ_UV_TAPS 3 +#define N_VERT_UV_TAPS 3 +#define N_PHASES 17 +#define MAX_TAPS 5 + +/* memory bufferd overlay registers */ +struct overlay_registers { + u32 OBUF_0Y; + u32 OBUF_1Y; + u32 OBUF_0U; + u32 OBUF_0V; + u32 OBUF_1U; + u32 OBUF_1V; + u32 OSTRIDE; + u32 YRGB_VPH; + u32 UV_VPH; + u32 HORZ_PH; + u32 INIT_PHS; + u32 DWINPOS; + u32 DWINSZ; + u32 SWIDTH; + u32 SWIDTHSW; + u32 SHEIGHT; + u32 YRGBSCALE; + u32 UVSCALE; + u32 OCLRC0; + u32 OCLRC1; + u32 DCLRKV; + u32 DCLRKM; + u32 SCLRKVH; + u32 SCLRKVL; + u32 SCLRKEN; + u32 OCONFIG; + u32 OCMD; + u32 RESERVED1; /* 0x6C */ + u32 OSTART_0Y; + u32 OSTART_1Y; + u32 OSTART_0U; + u32 OSTART_0V; + u32 OSTART_1U; + u32 OSTART_1V; + u32 OTILEOFF_0Y; + u32 OTILEOFF_1Y; + u32 OTILEOFF_0U; + u32 OTILEOFF_0V; + u32 OTILEOFF_1U; + u32 OTILEOFF_1V; + u32 FASTHSCALE; /* 0xA0 */ + u32 UVSCALEV; /* 0xA4 */ + u32 RESERVEDC[(0x200 - 0xA8) / 4]; /* 0xA8 - 0x1FC */ + u16 Y_VCOEFS[N_VERT_Y_TAPS * N_PHASES]; /* 0x200 */ + u16 RESERVEDD[0x100 / 2 - N_VERT_Y_TAPS * N_PHASES]; + u16 Y_HCOEFS[N_HORIZ_Y_TAPS * N_PHASES]; /* 0x300 */ + u16 RESERVEDE[0x200 / 2 - N_HORIZ_Y_TAPS * N_PHASES]; + u16 UV_VCOEFS[N_VERT_UV_TAPS * N_PHASES]; /* 0x500 */ + u16 RESERVEDF[0x100 / 2 - N_VERT_UV_TAPS * N_PHASES]; + u16 UV_HCOEFS[N_HORIZ_UV_TAPS * N_PHASES]; /* 0x600 */ + u16 RESERVEDG[0x100 / 2 - N_HORIZ_UV_TAPS * N_PHASES]; +}; + +struct intel_overlay { + struct drm_device *dev; + struct intel_crtc *crtc; + struct drm_i915_gem_object *vid_bo; + struct drm_i915_gem_object *old_vid_bo; + int active; + int pfit_active; + u32 pfit_vscale_ratio; /* shifted-point number, (1<<12) == 1.0 */ + u32 color_key; + u32 brightness, contrast, saturation; + u32 old_xscale, old_yscale; + /* register access */ + u32 flip_addr; + struct drm_i915_gem_object *reg_bo; + /* flip handling */ + uint32_t last_flip_req; + void (*flip_tail)(struct intel_overlay *); +}; + +static struct overlay_registers * +intel_overlay_map_regs(struct intel_overlay *overlay) +{ + drm_i915_private_t *dev_priv = overlay->dev->dev_private; + struct overlay_registers *regs; + + if (OVERLAY_NEEDS_PHYSICAL(overlay->dev)) + regs = overlay->reg_bo->phys_obj->handle->vaddr; + else + regs = io_mapping_map_wc(dev_priv->mm.gtt_mapping, + overlay->reg_bo->gtt_offset); + + return regs; +} + +static void intel_overlay_unmap_regs(struct intel_overlay *overlay, + struct overlay_registers *regs) +{ + if (!OVERLAY_NEEDS_PHYSICAL(overlay->dev)) + io_mapping_unmap(regs); +} + +static int intel_overlay_do_wait_request(struct intel_overlay *overlay, + struct drm_i915_gem_request *request, + void (*tail)(struct intel_overlay *)) +{ + struct drm_device *dev = overlay->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + int ret; + + BUG_ON(overlay->last_flip_req); + ret = i915_add_request(LP_RING(dev_priv), NULL, request); + if (ret) { + kfree(request); + return ret; + } + overlay->last_flip_req = request->seqno; + overlay->flip_tail = tail; + ret = i915_wait_request(LP_RING(dev_priv), overlay->last_flip_req); + if (ret) + return ret; + + overlay->last_flip_req = 0; + return 0; +} + +/* Workaround for i830 bug where pipe a must be enable to change control regs */ +static int +i830_activate_pipe_a(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct intel_crtc *crtc; + struct drm_crtc_helper_funcs *crtc_funcs; + struct drm_display_mode vesa_640x480 = { + DRM_MODE("640x480", DRM_MODE_TYPE_DRIVER, 25175, 640, 656, + 752, 800, 0, 480, 489, 492, 525, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_NVSYNC) + }, *mode; + + crtc = to_intel_crtc(dev_priv->pipe_to_crtc_mapping[0]); + if (crtc->dpms_mode == DRM_MODE_DPMS_ON) + return 0; + + /* most i8xx have pipe a forced on, so don't trust dpms mode */ + if (I915_READ(_PIPEACONF) & PIPECONF_ENABLE) + return 0; + + crtc_funcs = crtc->base.helper_private; + if (crtc_funcs->dpms == NULL) + return 0; + + DRM_DEBUG_DRIVER("Enabling pipe A in order to enable overlay\n"); + + mode = drm_mode_duplicate(dev, &vesa_640x480); + drm_mode_set_crtcinfo(mode, CRTC_INTERLACE_HALVE_V); + if(!drm_crtc_helper_set_mode(&crtc->base, mode, + crtc->base.x, crtc->base.y, + crtc->base.fb)) + return 0; + + crtc_funcs->dpms(&crtc->base, DRM_MODE_DPMS_ON); + return 1; +} + +static void +i830_deactivate_pipe_a(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_crtc *crtc = dev_priv->pipe_to_crtc_mapping[0]; + struct drm_crtc_helper_funcs *crtc_funcs = crtc->helper_private; + + crtc_funcs->dpms(crtc, DRM_MODE_DPMS_OFF); +} + +/* overlay needs to be disable in OCMD reg */ +static int intel_overlay_on(struct intel_overlay *overlay) +{ + struct drm_device *dev = overlay->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_i915_gem_request *request; + int pipe_a_quirk = 0; + int ret; + + BUG_ON(overlay->active); + overlay->active = 1; + + if (IS_I830(dev)) { + pipe_a_quirk = i830_activate_pipe_a(dev); + if (pipe_a_quirk < 0) + return pipe_a_quirk; + } + + request = kzalloc(sizeof(*request), GFP_KERNEL); + if (request == NULL) { + ret = -ENOMEM; + goto out; + } + + ret = BEGIN_LP_RING(4); + if (ret) { + kfree(request); + goto out; + } + + OUT_RING(MI_OVERLAY_FLIP | MI_OVERLAY_ON); + OUT_RING(overlay->flip_addr | OFC_UPDATE); + OUT_RING(MI_WAIT_FOR_EVENT | MI_WAIT_FOR_OVERLAY_FLIP); + OUT_RING(MI_NOOP); + ADVANCE_LP_RING(); + + ret = intel_overlay_do_wait_request(overlay, request, NULL); +out: + if (pipe_a_quirk) + i830_deactivate_pipe_a(dev); + + return ret; +} + +/* overlay needs to be enabled in OCMD reg */ +static int intel_overlay_continue(struct intel_overlay *overlay, + bool load_polyphase_filter) +{ + struct drm_device *dev = overlay->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_i915_gem_request *request; + u32 flip_addr = overlay->flip_addr; + u32 tmp; + int ret; + + BUG_ON(!overlay->active); + + request = kzalloc(sizeof(*request), GFP_KERNEL); + if (request == NULL) + return -ENOMEM; + + if (load_polyphase_filter) + flip_addr |= OFC_UPDATE; + + /* check for underruns */ + tmp = I915_READ(DOVSTA); + if (tmp & (1 << 17)) + DRM_DEBUG("overlay underrun, DOVSTA: %x\n", tmp); + + ret = BEGIN_LP_RING(2); + if (ret) { + kfree(request); + return ret; + } + OUT_RING(MI_OVERLAY_FLIP | MI_OVERLAY_CONTINUE); + OUT_RING(flip_addr); + ADVANCE_LP_RING(); + + ret = i915_add_request(LP_RING(dev_priv), NULL, request); + if (ret) { + kfree(request); + return ret; + } + + overlay->last_flip_req = request->seqno; + return 0; +} + +static void intel_overlay_release_old_vid_tail(struct intel_overlay *overlay) +{ + struct drm_i915_gem_object *obj = overlay->old_vid_bo; + + i915_gem_object_unpin(obj); + drm_gem_object_unreference(&obj->base); + + overlay->old_vid_bo = NULL; +} + +static void intel_overlay_off_tail(struct intel_overlay *overlay) +{ + struct drm_i915_gem_object *obj = overlay->vid_bo; + + /* never have the overlay hw on without showing a frame */ + BUG_ON(!overlay->vid_bo); + + i915_gem_object_unpin(obj); + drm_gem_object_unreference(&obj->base); + overlay->vid_bo = NULL; + + overlay->crtc->overlay = NULL; + overlay->crtc = NULL; + overlay->active = 0; +} + +/* overlay needs to be disabled in OCMD reg */ +static int intel_overlay_off(struct intel_overlay *overlay) +{ + struct drm_device *dev = overlay->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + u32 flip_addr = overlay->flip_addr; + struct drm_i915_gem_request *request; + int ret; + + BUG_ON(!overlay->active); + + request = kzalloc(sizeof(*request), GFP_KERNEL); + if (request == NULL) + return -ENOMEM; + + /* According to intel docs the overlay hw may hang (when switching + * off) without loading the filter coeffs. It is however unclear whether + * this applies to the disabling of the overlay or to the switching off + * of the hw. Do it in both cases */ + flip_addr |= OFC_UPDATE; + + ret = BEGIN_LP_RING(6); + if (ret) { + kfree(request); + return ret; + } + /* wait for overlay to go idle */ + OUT_RING(MI_OVERLAY_FLIP | MI_OVERLAY_CONTINUE); + OUT_RING(flip_addr); + OUT_RING(MI_WAIT_FOR_EVENT | MI_WAIT_FOR_OVERLAY_FLIP); + /* turn overlay off */ + OUT_RING(MI_OVERLAY_FLIP | MI_OVERLAY_OFF); + OUT_RING(flip_addr); + OUT_RING(MI_WAIT_FOR_EVENT | MI_WAIT_FOR_OVERLAY_FLIP); + ADVANCE_LP_RING(); + + return intel_overlay_do_wait_request(overlay, request, + intel_overlay_off_tail); +} + +/* recover from an interruption due to a signal + * We have to be careful not to repeat work forever an make forward progess. */ +static int intel_overlay_recover_from_interrupt(struct intel_overlay *overlay) +{ + struct drm_device *dev = overlay->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + int ret; + + if (overlay->last_flip_req == 0) + return 0; + + ret = i915_wait_request(LP_RING(dev_priv), overlay->last_flip_req); + if (ret) + return ret; + + if (overlay->flip_tail) + overlay->flip_tail(overlay); + + overlay->last_flip_req = 0; + return 0; +} + +/* Wait for pending overlay flip and release old frame. + * Needs to be called before the overlay register are changed + * via intel_overlay_(un)map_regs + */ +static int intel_overlay_release_old_vid(struct intel_overlay *overlay) +{ + struct drm_device *dev = overlay->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + int ret; + + /* Only wait if there is actually an old frame to release to + * guarantee forward progress. + */ + if (!overlay->old_vid_bo) + return 0; + + if (I915_READ(ISR) & I915_OVERLAY_PLANE_FLIP_PENDING_INTERRUPT) { + struct drm_i915_gem_request *request; + + /* synchronous slowpath */ + request = kzalloc(sizeof(*request), GFP_KERNEL); + if (request == NULL) + return -ENOMEM; + + ret = BEGIN_LP_RING(2); + if (ret) { + kfree(request); + return ret; + } + + OUT_RING(MI_WAIT_FOR_EVENT | MI_WAIT_FOR_OVERLAY_FLIP); + OUT_RING(MI_NOOP); + ADVANCE_LP_RING(); + + ret = intel_overlay_do_wait_request(overlay, request, + intel_overlay_release_old_vid_tail); + if (ret) + return ret; + } + + intel_overlay_release_old_vid_tail(overlay); + return 0; +} + +struct put_image_params { + int format; + short dst_x; + short dst_y; + short dst_w; + short dst_h; + short src_w; + short src_scan_h; + short src_scan_w; + short src_h; + short stride_Y; + short stride_UV; + int offset_Y; + int offset_U; + int offset_V; +}; + +static int packed_depth_bytes(u32 format) +{ + switch (format & I915_OVERLAY_DEPTH_MASK) { + case I915_OVERLAY_YUV422: + return 4; + case I915_OVERLAY_YUV411: + /* return 6; not implemented */ + default: + return -EINVAL; + } +} + +static int packed_width_bytes(u32 format, short width) +{ + switch (format & I915_OVERLAY_DEPTH_MASK) { + case I915_OVERLAY_YUV422: + return width << 1; + default: + return -EINVAL; + } +} + +static int uv_hsubsampling(u32 format) +{ + switch (format & I915_OVERLAY_DEPTH_MASK) { + case I915_OVERLAY_YUV422: + case I915_OVERLAY_YUV420: + return 2; + case I915_OVERLAY_YUV411: + case I915_OVERLAY_YUV410: + return 4; + default: + return -EINVAL; + } +} + +static int uv_vsubsampling(u32 format) +{ + switch (format & I915_OVERLAY_DEPTH_MASK) { + case I915_OVERLAY_YUV420: + case I915_OVERLAY_YUV410: + return 2; + case I915_OVERLAY_YUV422: + case I915_OVERLAY_YUV411: + return 1; + default: + return -EINVAL; + } +} + +static u32 calc_swidthsw(struct drm_device *dev, u32 offset, u32 width) +{ + u32 mask, shift, ret; + if (IS_GEN2(dev)) { + mask = 0x1f; + shift = 5; + } else { + mask = 0x3f; + shift = 6; + } + ret = ((offset + width + mask) >> shift) - (offset >> shift); + if (!IS_GEN2(dev)) + ret <<= 1; + ret -=1; + return ret << 2; +} + +static const u16 y_static_hcoeffs[N_HORIZ_Y_TAPS * N_PHASES] = { + 0x3000, 0xb4a0, 0x1930, 0x1920, 0xb4a0, + 0x3000, 0xb500, 0x19d0, 0x1880, 0xb440, + 0x3000, 0xb540, 0x1a88, 0x2f80, 0xb3e0, + 0x3000, 0xb580, 0x1b30, 0x2e20, 0xb380, + 0x3000, 0xb5c0, 0x1bd8, 0x2cc0, 0xb320, + 0x3020, 0xb5e0, 0x1c60, 0x2b80, 0xb2c0, + 0x3020, 0xb5e0, 0x1cf8, 0x2a20, 0xb260, + 0x3020, 0xb5e0, 0x1d80, 0x28e0, 0xb200, + 0x3020, 0xb5c0, 0x1e08, 0x3f40, 0xb1c0, + 0x3020, 0xb580, 0x1e78, 0x3ce0, 0xb160, + 0x3040, 0xb520, 0x1ed8, 0x3aa0, 0xb120, + 0x3040, 0xb4a0, 0x1f30, 0x3880, 0xb0e0, + 0x3040, 0xb400, 0x1f78, 0x3680, 0xb0a0, + 0x3020, 0xb340, 0x1fb8, 0x34a0, 0xb060, + 0x3020, 0xb240, 0x1fe0, 0x32e0, 0xb040, + 0x3020, 0xb140, 0x1ff8, 0x3160, 0xb020, + 0xb000, 0x3000, 0x0800, 0x3000, 0xb000 +}; + +static const u16 uv_static_hcoeffs[N_HORIZ_UV_TAPS * N_PHASES] = { + 0x3000, 0x1800, 0x1800, 0xb000, 0x18d0, 0x2e60, + 0xb000, 0x1990, 0x2ce0, 0xb020, 0x1a68, 0x2b40, + 0xb040, 0x1b20, 0x29e0, 0xb060, 0x1bd8, 0x2880, + 0xb080, 0x1c88, 0x3e60, 0xb0a0, 0x1d28, 0x3c00, + 0xb0c0, 0x1db8, 0x39e0, 0xb0e0, 0x1e40, 0x37e0, + 0xb100, 0x1eb8, 0x3620, 0xb100, 0x1f18, 0x34a0, + 0xb100, 0x1f68, 0x3360, 0xb0e0, 0x1fa8, 0x3240, + 0xb0c0, 0x1fe0, 0x3140, 0xb060, 0x1ff0, 0x30a0, + 0x3000, 0x0800, 0x3000 +}; + +static void update_polyphase_filter(struct overlay_registers *regs) +{ + memcpy(regs->Y_HCOEFS, y_static_hcoeffs, sizeof(y_static_hcoeffs)); + memcpy(regs->UV_HCOEFS, uv_static_hcoeffs, sizeof(uv_static_hcoeffs)); +} + +static bool update_scaling_factors(struct intel_overlay *overlay, + struct overlay_registers *regs, + struct put_image_params *params) +{ + /* fixed point with a 12 bit shift */ + u32 xscale, yscale, xscale_UV, yscale_UV; +#define FP_SHIFT 12 +#define FRACT_MASK 0xfff + bool scale_changed = false; + int uv_hscale = uv_hsubsampling(params->format); + int uv_vscale = uv_vsubsampling(params->format); + + if (params->dst_w > 1) + xscale = ((params->src_scan_w - 1) << FP_SHIFT) + /(params->dst_w); + else + xscale = 1 << FP_SHIFT; + + if (params->dst_h > 1) + yscale = ((params->src_scan_h - 1) << FP_SHIFT) + /(params->dst_h); + else + yscale = 1 << FP_SHIFT; + + /*if (params->format & I915_OVERLAY_YUV_PLANAR) {*/ + xscale_UV = xscale/uv_hscale; + yscale_UV = yscale/uv_vscale; + /* make the Y scale to UV scale ratio an exact multiply */ + xscale = xscale_UV * uv_hscale; + yscale = yscale_UV * uv_vscale; + /*} else { + xscale_UV = 0; + yscale_UV = 0; + }*/ + + if (xscale != overlay->old_xscale || yscale != overlay->old_yscale) + scale_changed = true; + overlay->old_xscale = xscale; + overlay->old_yscale = yscale; + + regs->YRGBSCALE = (((yscale & FRACT_MASK) << 20) | + ((xscale >> FP_SHIFT) << 16) | + ((xscale & FRACT_MASK) << 3)); + + regs->UVSCALE = (((yscale_UV & FRACT_MASK) << 20) | + ((xscale_UV >> FP_SHIFT) << 16) | + ((xscale_UV & FRACT_MASK) << 3)); + + regs->UVSCALEV = ((((yscale >> FP_SHIFT) << 16) | + ((yscale_UV >> FP_SHIFT) << 0))); + + if (scale_changed) + update_polyphase_filter(regs); + + return scale_changed; +} + +static void update_colorkey(struct intel_overlay *overlay, + struct overlay_registers *regs) +{ + u32 key = overlay->color_key; + + switch (overlay->crtc->base.fb->bits_per_pixel) { + case 8: + regs->DCLRKV = 0; + regs->DCLRKM = CLK_RGB8I_MASK | DST_KEY_ENABLE; + break; + + case 16: + if (overlay->crtc->base.fb->depth == 15) { + regs->DCLRKV = RGB15_TO_COLORKEY(key); + regs->DCLRKM = CLK_RGB15_MASK | DST_KEY_ENABLE; + } else { + regs->DCLRKV = RGB16_TO_COLORKEY(key); + regs->DCLRKM = CLK_RGB16_MASK | DST_KEY_ENABLE; + } + break; + + case 24: + case 32: + regs->DCLRKV = key; + regs->DCLRKM = CLK_RGB24_MASK | DST_KEY_ENABLE; + break; + } +} + +static u32 overlay_cmd_reg(struct put_image_params *params) +{ + u32 cmd = OCMD_ENABLE | OCMD_BUF_TYPE_FRAME | OCMD_BUFFER0; + + if (params->format & I915_OVERLAY_YUV_PLANAR) { + switch (params->format & I915_OVERLAY_DEPTH_MASK) { + case I915_OVERLAY_YUV422: + cmd |= OCMD_YUV_422_PLANAR; + break; + case I915_OVERLAY_YUV420: + cmd |= OCMD_YUV_420_PLANAR; + break; + case I915_OVERLAY_YUV411: + case I915_OVERLAY_YUV410: + cmd |= OCMD_YUV_410_PLANAR; + break; + } + } else { /* YUV packed */ + switch (params->format & I915_OVERLAY_DEPTH_MASK) { + case I915_OVERLAY_YUV422: + cmd |= OCMD_YUV_422_PACKED; + break; + case I915_OVERLAY_YUV411: + cmd |= OCMD_YUV_411_PACKED; + break; + } + + switch (params->format & I915_OVERLAY_SWAP_MASK) { + case I915_OVERLAY_NO_SWAP: + break; + case I915_OVERLAY_UV_SWAP: + cmd |= OCMD_UV_SWAP; + break; + case I915_OVERLAY_Y_SWAP: + cmd |= OCMD_Y_SWAP; + break; + case I915_OVERLAY_Y_AND_UV_SWAP: + cmd |= OCMD_Y_AND_UV_SWAP; + break; + } + } + + return cmd; +} + +static int intel_overlay_do_put_image(struct intel_overlay *overlay, + struct drm_i915_gem_object *new_bo, + struct put_image_params *params) +{ + int ret, tmp_width; + struct overlay_registers *regs; + bool scale_changed = false; + struct drm_device *dev = overlay->dev; + + BUG_ON(!mutex_is_locked(&dev->struct_mutex)); + BUG_ON(!mutex_is_locked(&dev->mode_config.mutex)); + BUG_ON(!overlay); + + ret = intel_overlay_release_old_vid(overlay); + if (ret != 0) + return ret; + + ret = i915_gem_object_pin(new_bo, PAGE_SIZE, true); + if (ret != 0) + return ret; + + ret = i915_gem_object_set_to_gtt_domain(new_bo, 0); + if (ret != 0) + goto out_unpin; + + ret = i915_gem_object_put_fence(new_bo); + if (ret) + goto out_unpin; + + if (!overlay->active) { + regs = intel_overlay_map_regs(overlay); + if (!regs) { + ret = -ENOMEM; + goto out_unpin; + } + regs->OCONFIG = OCONF_CC_OUT_8BIT; + if (IS_GEN4(overlay->dev)) + regs->OCONFIG |= OCONF_CSC_MODE_BT709; + regs->OCONFIG |= overlay->crtc->pipe == 0 ? + OCONF_PIPE_A : OCONF_PIPE_B; + intel_overlay_unmap_regs(overlay, regs); + + ret = intel_overlay_on(overlay); + if (ret != 0) + goto out_unpin; + } + + regs = intel_overlay_map_regs(overlay); + if (!regs) { + ret = -ENOMEM; + goto out_unpin; + } + + regs->DWINPOS = (params->dst_y << 16) | params->dst_x; + regs->DWINSZ = (params->dst_h << 16) | params->dst_w; + + if (params->format & I915_OVERLAY_YUV_PACKED) + tmp_width = packed_width_bytes(params->format, params->src_w); + else + tmp_width = params->src_w; + + regs->SWIDTH = params->src_w; + regs->SWIDTHSW = calc_swidthsw(overlay->dev, + params->offset_Y, tmp_width); + regs->SHEIGHT = params->src_h; + regs->OBUF_0Y = new_bo->gtt_offset + params-> offset_Y; + regs->OSTRIDE = params->stride_Y; + + if (params->format & I915_OVERLAY_YUV_PLANAR) { + int uv_hscale = uv_hsubsampling(params->format); + int uv_vscale = uv_vsubsampling(params->format); + u32 tmp_U, tmp_V; + regs->SWIDTH |= (params->src_w/uv_hscale) << 16; + tmp_U = calc_swidthsw(overlay->dev, params->offset_U, + params->src_w/uv_hscale); + tmp_V = calc_swidthsw(overlay->dev, params->offset_V, + params->src_w/uv_hscale); + regs->SWIDTHSW |= max_t(u32, tmp_U, tmp_V) << 16; + regs->SHEIGHT |= (params->src_h/uv_vscale) << 16; + regs->OBUF_0U = new_bo->gtt_offset + params->offset_U; + regs->OBUF_0V = new_bo->gtt_offset + params->offset_V; + regs->OSTRIDE |= params->stride_UV << 16; + } + + scale_changed = update_scaling_factors(overlay, regs, params); + + update_colorkey(overlay, regs); + + regs->OCMD = overlay_cmd_reg(params); + + intel_overlay_unmap_regs(overlay, regs); + + ret = intel_overlay_continue(overlay, scale_changed); + if (ret) + goto out_unpin; + + overlay->old_vid_bo = overlay->vid_bo; + overlay->vid_bo = new_bo; + + return 0; + +out_unpin: + i915_gem_object_unpin(new_bo); + return ret; +} + +int intel_overlay_switch_off(struct intel_overlay *overlay) +{ + struct overlay_registers *regs; + struct drm_device *dev = overlay->dev; + int ret; + + BUG_ON(!mutex_is_locked(&dev->struct_mutex)); + BUG_ON(!mutex_is_locked(&dev->mode_config.mutex)); + + ret = intel_overlay_recover_from_interrupt(overlay); + if (ret != 0) + return ret; + + if (!overlay->active) + return 0; + + ret = intel_overlay_release_old_vid(overlay); + if (ret != 0) + return ret; + + regs = intel_overlay_map_regs(overlay); + regs->OCMD = 0; + intel_overlay_unmap_regs(overlay, regs); + + ret = intel_overlay_off(overlay); + if (ret != 0) + return ret; + + intel_overlay_off_tail(overlay); + return 0; +} + +static int check_overlay_possible_on_crtc(struct intel_overlay *overlay, + struct intel_crtc *crtc) +{ + drm_i915_private_t *dev_priv = overlay->dev->dev_private; + + if (!crtc->active) + return -EINVAL; + + /* can't use the overlay with double wide pipe */ + if (INTEL_INFO(overlay->dev)->gen < 4 && + (I915_READ(PIPECONF(crtc->pipe)) & (PIPECONF_DOUBLE_WIDE | PIPECONF_ENABLE)) != PIPECONF_ENABLE) + return -EINVAL; + + return 0; +} + +static void update_pfit_vscale_ratio(struct intel_overlay *overlay) +{ + struct drm_device *dev = overlay->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + u32 pfit_control = I915_READ(PFIT_CONTROL); + u32 ratio; + + /* XXX: This is not the same logic as in the xorg driver, but more in + * line with the intel documentation for the i965 + */ + if (INTEL_INFO(dev)->gen >= 4) { + /* on i965 use the PGM reg to read out the autoscaler values */ + ratio = I915_READ(PFIT_PGM_RATIOS) >> PFIT_VERT_SCALE_SHIFT_965; + } else { + if (pfit_control & VERT_AUTO_SCALE) + ratio = I915_READ(PFIT_AUTO_RATIOS); + else + ratio = I915_READ(PFIT_PGM_RATIOS); + ratio >>= PFIT_VERT_SCALE_SHIFT; + } + + overlay->pfit_vscale_ratio = ratio; +} + +static int check_overlay_dst(struct intel_overlay *overlay, + struct drm_intel_overlay_put_image *rec) +{ + struct drm_display_mode *mode = &overlay->crtc->base.mode; + + if (rec->dst_x < mode->crtc_hdisplay && + rec->dst_x + rec->dst_width <= mode->crtc_hdisplay && + rec->dst_y < mode->crtc_vdisplay && + rec->dst_y + rec->dst_height <= mode->crtc_vdisplay) + return 0; + else + return -EINVAL; +} + +static int check_overlay_scaling(struct put_image_params *rec) +{ + u32 tmp; + + /* downscaling limit is 8.0 */ + tmp = ((rec->src_scan_h << 16) / rec->dst_h) >> 16; + if (tmp > 7) + return -EINVAL; + tmp = ((rec->src_scan_w << 16) / rec->dst_w) >> 16; + if (tmp > 7) + return -EINVAL; + + return 0; +} + +static int check_overlay_src(struct drm_device *dev, + struct drm_intel_overlay_put_image *rec, + struct drm_i915_gem_object *new_bo) +{ + int uv_hscale = uv_hsubsampling(rec->flags); + int uv_vscale = uv_vsubsampling(rec->flags); + u32 stride_mask; + int depth; + u32 tmp; + + /* check src dimensions */ + if (IS_845G(dev) || IS_I830(dev)) { + if (rec->src_height > IMAGE_MAX_HEIGHT_LEGACY || + rec->src_width > IMAGE_MAX_WIDTH_LEGACY) + return -EINVAL; + } else { + if (rec->src_height > IMAGE_MAX_HEIGHT || + rec->src_width > IMAGE_MAX_WIDTH) + return -EINVAL; + } + + /* better safe than sorry, use 4 as the maximal subsampling ratio */ + if (rec->src_height < N_VERT_Y_TAPS*4 || + rec->src_width < N_HORIZ_Y_TAPS*4) + return -EINVAL; + + /* check alignment constraints */ + switch (rec->flags & I915_OVERLAY_TYPE_MASK) { + case I915_OVERLAY_RGB: + /* not implemented */ + return -EINVAL; + + case I915_OVERLAY_YUV_PACKED: + if (uv_vscale != 1) + return -EINVAL; + + depth = packed_depth_bytes(rec->flags); + if (depth < 0) + return depth; + + /* ignore UV planes */ + rec->stride_UV = 0; + rec->offset_U = 0; + rec->offset_V = 0; + /* check pixel alignment */ + if (rec->offset_Y % depth) + return -EINVAL; + break; + + case I915_OVERLAY_YUV_PLANAR: + if (uv_vscale < 0 || uv_hscale < 0) + return -EINVAL; + /* no offset restrictions for planar formats */ + break; + + default: + return -EINVAL; + } + + if (rec->src_width % uv_hscale) + return -EINVAL; + + /* stride checking */ + if (IS_I830(dev) || IS_845G(dev)) + stride_mask = 255; + else + stride_mask = 63; + + if (rec->stride_Y & stride_mask || rec->stride_UV & stride_mask) + return -EINVAL; + if (IS_GEN4(dev) && rec->stride_Y < 512) + return -EINVAL; + + tmp = (rec->flags & I915_OVERLAY_TYPE_MASK) == I915_OVERLAY_YUV_PLANAR ? + 4096 : 8192; + if (rec->stride_Y > tmp || rec->stride_UV > 2*1024) + return -EINVAL; + + /* check buffer dimensions */ + switch (rec->flags & I915_OVERLAY_TYPE_MASK) { + case I915_OVERLAY_RGB: + case I915_OVERLAY_YUV_PACKED: + /* always 4 Y values per depth pixels */ + if (packed_width_bytes(rec->flags, rec->src_width) > rec->stride_Y) + return -EINVAL; + + tmp = rec->stride_Y*rec->src_height; + if (rec->offset_Y + tmp > new_bo->base.size) + return -EINVAL; + break; + + case I915_OVERLAY_YUV_PLANAR: + if (rec->src_width > rec->stride_Y) + return -EINVAL; + if (rec->src_width/uv_hscale > rec->stride_UV) + return -EINVAL; + + tmp = rec->stride_Y * rec->src_height; + if (rec->offset_Y + tmp > new_bo->base.size) + return -EINVAL; + + tmp = rec->stride_UV * (rec->src_height / uv_vscale); + if (rec->offset_U + tmp > new_bo->base.size || + rec->offset_V + tmp > new_bo->base.size) + return -EINVAL; + break; + } + + return 0; +} + +/** + * Return the pipe currently connected to the panel fitter, + * or -1 if the panel fitter is not present or not in use + */ +static int intel_panel_fitter_pipe(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u32 pfit_control; + + /* i830 doesn't have a panel fitter */ + if (IS_I830(dev)) + return -1; + + pfit_control = I915_READ(PFIT_CONTROL); + + /* See if the panel fitter is in use */ + if ((pfit_control & PFIT_ENABLE) == 0) + return -1; + + /* 965 can place panel fitter on either pipe */ + if (IS_GEN4(dev)) + return (pfit_control >> 29) & 0x3; + + /* older chips can only use pipe 1 */ + return 1; +} + +int intel_overlay_put_image(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_intel_overlay_put_image *put_image_rec = data; + drm_i915_private_t *dev_priv = dev->dev_private; + struct intel_overlay *overlay; + struct drm_mode_object *drmmode_obj; + struct intel_crtc *crtc; + struct drm_i915_gem_object *new_bo; + struct put_image_params *params; + int ret; + + if (!dev_priv) { + DRM_ERROR("called with no initialization\n"); + return -EINVAL; + } + + overlay = dev_priv->overlay; + if (!overlay) { + DRM_DEBUG("userspace bug: no overlay\n"); + return -ENODEV; + } + + if (!(put_image_rec->flags & I915_OVERLAY_ENABLE)) { + mutex_lock(&dev->mode_config.mutex); + mutex_lock(&dev->struct_mutex); + + ret = intel_overlay_switch_off(overlay); + + mutex_unlock(&dev->struct_mutex); + mutex_unlock(&dev->mode_config.mutex); + + return ret; + } + + params = kmalloc(sizeof(struct put_image_params), GFP_KERNEL); + if (!params) + return -ENOMEM; + + drmmode_obj = drm_mode_object_find(dev, put_image_rec->crtc_id, + DRM_MODE_OBJECT_CRTC); + if (!drmmode_obj) { + ret = -ENOENT; + goto out_free; + } + crtc = to_intel_crtc(obj_to_crtc(drmmode_obj)); + + new_bo = to_intel_bo(drm_gem_object_lookup(dev, file_priv, + put_image_rec->bo_handle)); + if (&new_bo->base == NULL) { + ret = -ENOENT; + goto out_free; + } + + mutex_lock(&dev->mode_config.mutex); + mutex_lock(&dev->struct_mutex); + + if (new_bo->tiling_mode) { + DRM_ERROR("buffer used for overlay image can not be tiled\n"); + ret = -EINVAL; + goto out_unlock; + } + + ret = intel_overlay_recover_from_interrupt(overlay); + if (ret != 0) + goto out_unlock; + + if (overlay->crtc != crtc) { + struct drm_display_mode *mode = &crtc->base.mode; + ret = intel_overlay_switch_off(overlay); + if (ret != 0) + goto out_unlock; + + ret = check_overlay_possible_on_crtc(overlay, crtc); + if (ret != 0) + goto out_unlock; + + overlay->crtc = crtc; + crtc->overlay = overlay; + + /* line too wide, i.e. one-line-mode */ + if (mode->hdisplay > 1024 && + intel_panel_fitter_pipe(dev) == crtc->pipe) { + overlay->pfit_active = 1; + update_pfit_vscale_ratio(overlay); + } else + overlay->pfit_active = 0; + } + + ret = check_overlay_dst(overlay, put_image_rec); + if (ret != 0) + goto out_unlock; + + if (overlay->pfit_active) { + params->dst_y = ((((u32)put_image_rec->dst_y) << 12) / + overlay->pfit_vscale_ratio); + /* shifting right rounds downwards, so add 1 */ + params->dst_h = ((((u32)put_image_rec->dst_height) << 12) / + overlay->pfit_vscale_ratio) + 1; + } else { + params->dst_y = put_image_rec->dst_y; + params->dst_h = put_image_rec->dst_height; + } + params->dst_x = put_image_rec->dst_x; + params->dst_w = put_image_rec->dst_width; + + params->src_w = put_image_rec->src_width; + params->src_h = put_image_rec->src_height; + params->src_scan_w = put_image_rec->src_scan_width; + params->src_scan_h = put_image_rec->src_scan_height; + if (params->src_scan_h > params->src_h || + params->src_scan_w > params->src_w) { + ret = -EINVAL; + goto out_unlock; + } + + ret = check_overlay_src(dev, put_image_rec, new_bo); + if (ret != 0) + goto out_unlock; + params->format = put_image_rec->flags & ~I915_OVERLAY_FLAGS_MASK; + params->stride_Y = put_image_rec->stride_Y; + params->stride_UV = put_image_rec->stride_UV; + params->offset_Y = put_image_rec->offset_Y; + params->offset_U = put_image_rec->offset_U; + params->offset_V = put_image_rec->offset_V; + + /* Check scaling after src size to prevent a divide-by-zero. */ + ret = check_overlay_scaling(params); + if (ret != 0) + goto out_unlock; + + ret = intel_overlay_do_put_image(overlay, new_bo, params); + if (ret != 0) + goto out_unlock; + + mutex_unlock(&dev->struct_mutex); + mutex_unlock(&dev->mode_config.mutex); + + kfree(params); + + return 0; + +out_unlock: + mutex_unlock(&dev->struct_mutex); + mutex_unlock(&dev->mode_config.mutex); + drm_gem_object_unreference_unlocked(&new_bo->base); +out_free: + kfree(params); + + return ret; +} + +static void update_reg_attrs(struct intel_overlay *overlay, + struct overlay_registers *regs) +{ + regs->OCLRC0 = (overlay->contrast << 18) | (overlay->brightness & 0xff); + regs->OCLRC1 = overlay->saturation; +} + +static bool check_gamma_bounds(u32 gamma1, u32 gamma2) +{ + int i; + + if (gamma1 & 0xff000000 || gamma2 & 0xff000000) + return false; + + for (i = 0; i < 3; i++) { + if (((gamma1 >> i*8) & 0xff) >= ((gamma2 >> i*8) & 0xff)) + return false; + } + + return true; +} + +static bool check_gamma5_errata(u32 gamma5) +{ + int i; + + for (i = 0; i < 3; i++) { + if (((gamma5 >> i*8) & 0xff) == 0x80) + return false; + } + + return true; +} + +static int check_gamma(struct drm_intel_overlay_attrs *attrs) +{ + if (!check_gamma_bounds(0, attrs->gamma0) || + !check_gamma_bounds(attrs->gamma0, attrs->gamma1) || + !check_gamma_bounds(attrs->gamma1, attrs->gamma2) || + !check_gamma_bounds(attrs->gamma2, attrs->gamma3) || + !check_gamma_bounds(attrs->gamma3, attrs->gamma4) || + !check_gamma_bounds(attrs->gamma4, attrs->gamma5) || + !check_gamma_bounds(attrs->gamma5, 0x00ffffff)) + return -EINVAL; + + if (!check_gamma5_errata(attrs->gamma5)) + return -EINVAL; + + return 0; +} + +int intel_overlay_attrs(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_intel_overlay_attrs *attrs = data; + drm_i915_private_t *dev_priv = dev->dev_private; + struct intel_overlay *overlay; + struct overlay_registers *regs; + int ret; + + if (!dev_priv) { + DRM_ERROR("called with no initialization\n"); + return -EINVAL; + } + + overlay = dev_priv->overlay; + if (!overlay) { + DRM_DEBUG("userspace bug: no overlay\n"); + return -ENODEV; + } + + mutex_lock(&dev->mode_config.mutex); + mutex_lock(&dev->struct_mutex); + + ret = -EINVAL; + if (!(attrs->flags & I915_OVERLAY_UPDATE_ATTRS)) { + attrs->color_key = overlay->color_key; + attrs->brightness = overlay->brightness; + attrs->contrast = overlay->contrast; + attrs->saturation = overlay->saturation; + + if (!IS_GEN2(dev)) { + attrs->gamma0 = I915_READ(OGAMC0); + attrs->gamma1 = I915_READ(OGAMC1); + attrs->gamma2 = I915_READ(OGAMC2); + attrs->gamma3 = I915_READ(OGAMC3); + attrs->gamma4 = I915_READ(OGAMC4); + attrs->gamma5 = I915_READ(OGAMC5); + } + } else { + if (attrs->brightness < -128 || attrs->brightness > 127) + goto out_unlock; + if (attrs->contrast > 255) + goto out_unlock; + if (attrs->saturation > 1023) + goto out_unlock; + + overlay->color_key = attrs->color_key; + overlay->brightness = attrs->brightness; + overlay->contrast = attrs->contrast; + overlay->saturation = attrs->saturation; + + regs = intel_overlay_map_regs(overlay); + if (!regs) { + ret = -ENOMEM; + goto out_unlock; + } + + update_reg_attrs(overlay, regs); + + intel_overlay_unmap_regs(overlay, regs); + + if (attrs->flags & I915_OVERLAY_UPDATE_GAMMA) { + if (IS_GEN2(dev)) + goto out_unlock; + + if (overlay->active) { + ret = -EBUSY; + goto out_unlock; + } + + ret = check_gamma(attrs); + if (ret) + goto out_unlock; + + I915_WRITE(OGAMC0, attrs->gamma0); + I915_WRITE(OGAMC1, attrs->gamma1); + I915_WRITE(OGAMC2, attrs->gamma2); + I915_WRITE(OGAMC3, attrs->gamma3); + I915_WRITE(OGAMC4, attrs->gamma4); + I915_WRITE(OGAMC5, attrs->gamma5); + } + } + + ret = 0; +out_unlock: + mutex_unlock(&dev->struct_mutex); + mutex_unlock(&dev->mode_config.mutex); + + return ret; +} + +void intel_setup_overlay(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct intel_overlay *overlay; + struct drm_i915_gem_object *reg_bo; + struct overlay_registers *regs; + int ret; + + if (!HAS_OVERLAY(dev)) + return; + + overlay = kzalloc(sizeof(struct intel_overlay), GFP_KERNEL); + if (!overlay) + return; + + mutex_lock(&dev->struct_mutex); + if (WARN_ON(dev_priv->overlay)) + goto out_free; + + overlay->dev = dev; + + reg_bo = i915_gem_alloc_object(dev, PAGE_SIZE); + if (!reg_bo) + goto out_free; + overlay->reg_bo = reg_bo; + + if (OVERLAY_NEEDS_PHYSICAL(dev)) { + ret = i915_gem_attach_phys_object(dev, reg_bo, + I915_GEM_PHYS_OVERLAY_REGS, + PAGE_SIZE); + if (ret) { + DRM_ERROR("failed to attach phys overlay regs\n"); + goto out_free_bo; + } + overlay->flip_addr = reg_bo->phys_obj->handle->busaddr; + } else { + ret = i915_gem_object_pin(reg_bo, PAGE_SIZE, true); + if (ret) { + DRM_ERROR("failed to pin overlay register bo\n"); + goto out_free_bo; + } + overlay->flip_addr = reg_bo->gtt_offset; + + ret = i915_gem_object_set_to_gtt_domain(reg_bo, true); + if (ret) { + DRM_ERROR("failed to move overlay register bo into the GTT\n"); + goto out_unpin_bo; + } + } + + /* init all values */ + overlay->color_key = 0x0101fe; + overlay->brightness = -19; + overlay->contrast = 75; + overlay->saturation = 146; + + regs = intel_overlay_map_regs(overlay); + if (!regs) + goto out_unpin_bo; + + memset(regs, 0, sizeof(struct overlay_registers)); + update_polyphase_filter(regs); + update_reg_attrs(overlay, regs); + + intel_overlay_unmap_regs(overlay, regs); + + dev_priv->overlay = overlay; + mutex_unlock(&dev->struct_mutex); + DRM_INFO("initialized overlay support\n"); + return; + +out_unpin_bo: + if (!OVERLAY_NEEDS_PHYSICAL(dev)) + i915_gem_object_unpin(reg_bo); +out_free_bo: + drm_gem_object_unreference(®_bo->base); +out_free: + mutex_unlock(&dev->struct_mutex); + kfree(overlay); + return; +} + +void intel_cleanup_overlay(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + + if (!dev_priv->overlay) + return; + + /* The bo's should be free'd by the generic code already. + * Furthermore modesetting teardown happens beforehand so the + * hardware should be off already */ + BUG_ON(dev_priv->overlay->active); + + drm_gem_object_unreference_unlocked(&dev_priv->overlay->reg_bo->base); + kfree(dev_priv->overlay); +} + +#ifdef CONFIG_DEBUG_FS +#include + +struct intel_overlay_error_state { + struct overlay_registers regs; + unsigned long base; + u32 dovsta; + u32 isr; +}; + +static struct overlay_registers * +intel_overlay_map_regs_atomic(struct intel_overlay *overlay) +{ + drm_i915_private_t *dev_priv = overlay->dev->dev_private; + struct overlay_registers *regs; + + if (OVERLAY_NEEDS_PHYSICAL(overlay->dev)) + regs = overlay->reg_bo->phys_obj->handle->vaddr; + else + regs = io_mapping_map_atomic_wc(dev_priv->mm.gtt_mapping, + overlay->reg_bo->gtt_offset); + + return regs; +} + +static void intel_overlay_unmap_regs_atomic(struct intel_overlay *overlay, + struct overlay_registers *regs) +{ + if (!OVERLAY_NEEDS_PHYSICAL(overlay->dev)) + io_mapping_unmap_atomic(regs); +} + + +struct intel_overlay_error_state * +intel_overlay_capture_error_state(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct intel_overlay *overlay = dev_priv->overlay; + struct intel_overlay_error_state *error; + struct overlay_registers __iomem *regs; + + if (!overlay || !overlay->active) + return NULL; + + error = kmalloc(sizeof(*error), GFP_ATOMIC); + if (error == NULL) + return NULL; + + error->dovsta = I915_READ(DOVSTA); + error->isr = I915_READ(ISR); + if (OVERLAY_NEEDS_PHYSICAL(overlay->dev)) + error->base = (long) overlay->reg_bo->phys_obj->handle->vaddr; + else + error->base = (long) overlay->reg_bo->gtt_offset; + + regs = intel_overlay_map_regs_atomic(overlay); + if (!regs) + goto err; + + memcpy_fromio(&error->regs, regs, sizeof(struct overlay_registers)); + intel_overlay_unmap_regs_atomic(overlay, regs); + + return error; + +err: + kfree(error); + return NULL; +} + +void +intel_overlay_print_error_state(struct seq_file *m, struct intel_overlay_error_state *error) +{ + seq_printf(m, "Overlay, status: 0x%08x, interrupt: 0x%08x\n", + error->dovsta, error->isr); + seq_printf(m, " Register file at 0x%08lx:\n", + error->base); + +#define P(x) seq_printf(m, " " #x ": 0x%08x\n", error->regs.x) + P(OBUF_0Y); + P(OBUF_1Y); + P(OBUF_0U); + P(OBUF_0V); + P(OBUF_1U); + P(OBUF_1V); + P(OSTRIDE); + P(YRGB_VPH); + P(UV_VPH); + P(HORZ_PH); + P(INIT_PHS); + P(DWINPOS); + P(DWINSZ); + P(SWIDTH); + P(SWIDTHSW); + P(SHEIGHT); + P(YRGBSCALE); + P(UVSCALE); + P(OCLRC0); + P(OCLRC1); + P(DCLRKV); + P(DCLRKM); + P(SCLRKVH); + P(SCLRKVL); + P(SCLRKEN); + P(OCONFIG); + P(OCMD); + P(OSTART_0Y); + P(OSTART_1Y); + P(OSTART_0U); + P(OSTART_0V); + P(OSTART_1U); + P(OSTART_1V); + P(OTILEOFF_0Y); + P(OTILEOFF_1Y); + P(OTILEOFF_0U); + P(OTILEOFF_0V); + P(OTILEOFF_1U); + P(OTILEOFF_1V); + P(FASTHSCALE); + P(UVSCALEV); +#undef P +} +#endif diff --git a/drivers/gpu/drm/i915/intel_panel.c b/drivers/gpu/drm/i915/intel_panel.c new file mode 100644 index 00000000..f8aa8211 --- /dev/null +++ b/drivers/gpu/drm/i915/intel_panel.c @@ -0,0 +1,316 @@ +/* + * Copyright © 2006-2010 Intel Corporation + * Copyright (c) 2006 Dave Airlie + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Eric Anholt + * Dave Airlie + * Jesse Barnes + * Chris Wilson + */ + +#include "intel_drv.h" + +#define PCI_LBPC 0xf4 /* legacy/combination backlight modes */ + +void +intel_fixed_panel_mode(struct drm_display_mode *fixed_mode, + struct drm_display_mode *adjusted_mode) +{ + adjusted_mode->hdisplay = fixed_mode->hdisplay; + adjusted_mode->hsync_start = fixed_mode->hsync_start; + adjusted_mode->hsync_end = fixed_mode->hsync_end; + adjusted_mode->htotal = fixed_mode->htotal; + + adjusted_mode->vdisplay = fixed_mode->vdisplay; + adjusted_mode->vsync_start = fixed_mode->vsync_start; + adjusted_mode->vsync_end = fixed_mode->vsync_end; + adjusted_mode->vtotal = fixed_mode->vtotal; + + adjusted_mode->clock = fixed_mode->clock; + + drm_mode_set_crtcinfo(adjusted_mode, CRTC_INTERLACE_HALVE_V); +} + +/* adjusted_mode has been preset to be the panel's fixed mode */ +void +intel_pch_panel_fitting(struct drm_device *dev, + int fitting_mode, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + int x, y, width, height; + + x = y = width = height = 0; + + /* Native modes don't need fitting */ + if (adjusted_mode->hdisplay == mode->hdisplay && + adjusted_mode->vdisplay == mode->vdisplay) + goto done; + + switch (fitting_mode) { + case DRM_MODE_SCALE_CENTER: + width = mode->hdisplay; + height = mode->vdisplay; + x = (adjusted_mode->hdisplay - width + 1)/2; + y = (adjusted_mode->vdisplay - height + 1)/2; + break; + + case DRM_MODE_SCALE_ASPECT: + /* Scale but preserve the aspect ratio */ + { + u32 scaled_width = adjusted_mode->hdisplay * mode->vdisplay; + u32 scaled_height = mode->hdisplay * adjusted_mode->vdisplay; + if (scaled_width > scaled_height) { /* pillar */ + width = scaled_height / mode->vdisplay; + if (width & 1) + width++; + x = (adjusted_mode->hdisplay - width + 1) / 2; + y = 0; + height = adjusted_mode->vdisplay; + } else if (scaled_width < scaled_height) { /* letter */ + height = scaled_width / mode->hdisplay; + if (height & 1) + height++; + y = (adjusted_mode->vdisplay - height + 1) / 2; + x = 0; + width = adjusted_mode->hdisplay; + } else { + x = y = 0; + width = adjusted_mode->hdisplay; + height = adjusted_mode->vdisplay; + } + } + break; + + default: + case DRM_MODE_SCALE_FULLSCREEN: + x = y = 0; + width = adjusted_mode->hdisplay; + height = adjusted_mode->vdisplay; + break; + } + +done: + dev_priv->pch_pf_pos = (x << 16) | y; + dev_priv->pch_pf_size = (width << 16) | height; +} + +static int is_backlight_combination_mode(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + if (INTEL_INFO(dev)->gen >= 4) + return I915_READ(BLC_PWM_CTL2) & BLM_COMBINATION_MODE; + + if (IS_GEN2(dev)) + return I915_READ(BLC_PWM_CTL) & BLM_LEGACY_MODE; + + return 0; +} + +static u32 i915_read_blc_pwm_ctl(struct drm_i915_private *dev_priv) +{ + u32 val; + + /* Restore the CTL value if it lost, e.g. GPU reset */ + + if (HAS_PCH_SPLIT(dev_priv->dev)) { + val = I915_READ(BLC_PWM_PCH_CTL2); + if (dev_priv->saveBLC_PWM_CTL2 == 0) { + dev_priv->saveBLC_PWM_CTL2 = val; + } else if (val == 0) { + I915_WRITE(BLC_PWM_PCH_CTL2, + dev_priv->saveBLC_PWM_CTL); + val = dev_priv->saveBLC_PWM_CTL; + } + } else { + val = I915_READ(BLC_PWM_CTL); + if (dev_priv->saveBLC_PWM_CTL == 0) { + dev_priv->saveBLC_PWM_CTL = val; + dev_priv->saveBLC_PWM_CTL2 = I915_READ(BLC_PWM_CTL2); + } else if (val == 0) { + I915_WRITE(BLC_PWM_CTL, + dev_priv->saveBLC_PWM_CTL); + I915_WRITE(BLC_PWM_CTL2, + dev_priv->saveBLC_PWM_CTL2); + val = dev_priv->saveBLC_PWM_CTL; + } + } + + return val; +} + +u32 intel_panel_get_max_backlight(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u32 max; + + max = i915_read_blc_pwm_ctl(dev_priv); + if (max == 0) { + /* XXX add code here to query mode clock or hardware clock + * and program max PWM appropriately. + */ + printk_once(KERN_WARNING "fixme: max PWM is zero.\n"); + return 1; + } + + if (HAS_PCH_SPLIT(dev)) { + max >>= 16; + } else { + if (IS_PINEVIEW(dev)) { + max >>= 17; + } else { + max >>= 16; + if (INTEL_INFO(dev)->gen < 4) + max &= ~1; + } + + if (is_backlight_combination_mode(dev)) + max *= 0xff; + } + + DRM_DEBUG_DRIVER("max backlight PWM = %d\n", max); + return max; +} + +u32 intel_panel_get_backlight(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u32 val; + + if (HAS_PCH_SPLIT(dev)) { + val = I915_READ(BLC_PWM_CPU_CTL) & BACKLIGHT_DUTY_CYCLE_MASK; + } else { + val = I915_READ(BLC_PWM_CTL) & BACKLIGHT_DUTY_CYCLE_MASK; + if (IS_PINEVIEW(dev)) + val >>= 1; + + if (is_backlight_combination_mode(dev)){ + u8 lbpc; + + val &= ~1; + pci_read_config_byte(dev->pdev, PCI_LBPC, &lbpc); + val *= lbpc; + } + } + + DRM_DEBUG_DRIVER("get backlight PWM = %d\n", val); + return val; +} + +static void intel_pch_panel_set_backlight(struct drm_device *dev, u32 level) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u32 val = I915_READ(BLC_PWM_CPU_CTL) & ~BACKLIGHT_DUTY_CYCLE_MASK; + I915_WRITE(BLC_PWM_CPU_CTL, val | level); +} + +static void intel_panel_actually_set_backlight(struct drm_device *dev, u32 level) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u32 tmp; + + DRM_DEBUG_DRIVER("set backlight PWM = %d\n", level); + + if (HAS_PCH_SPLIT(dev)) + return intel_pch_panel_set_backlight(dev, level); + + if (is_backlight_combination_mode(dev)){ + u32 max = intel_panel_get_max_backlight(dev); + u8 lbpc; + + lbpc = level * 0xfe / max + 1; + level /= lbpc; + pci_write_config_byte(dev->pdev, PCI_LBPC, lbpc); + } + + tmp = I915_READ(BLC_PWM_CTL); + if (IS_PINEVIEW(dev)) { + tmp &= ~(BACKLIGHT_DUTY_CYCLE_MASK - 1); + level <<= 1; + } else + tmp &= ~BACKLIGHT_DUTY_CYCLE_MASK; + I915_WRITE(BLC_PWM_CTL, tmp | level); +} + +void intel_panel_set_backlight(struct drm_device *dev, u32 level) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + dev_priv->backlight_level = level; + if (dev_priv->backlight_enabled) + intel_panel_actually_set_backlight(dev, level); +} + +void intel_panel_disable_backlight(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + dev_priv->backlight_enabled = false; + intel_panel_actually_set_backlight(dev, 0); +} + +void intel_panel_enable_backlight(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + if (dev_priv->backlight_level == 0) + dev_priv->backlight_level = intel_panel_get_max_backlight(dev); + + dev_priv->backlight_enabled = true; + intel_panel_actually_set_backlight(dev, dev_priv->backlight_level); +} + +void intel_panel_setup_backlight(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + dev_priv->backlight_level = intel_panel_get_backlight(dev); + dev_priv->backlight_enabled = dev_priv->backlight_level != 0; +} + +enum drm_connector_status +intel_panel_detect(struct drm_device *dev) +{ +#if 0 + struct drm_i915_private *dev_priv = dev->dev_private; +#endif + + if (i915_panel_ignore_lid) + return i915_panel_ignore_lid > 0 ? + connector_status_connected : + connector_status_disconnected; + + /* opregion lid state on HP 2540p is wrong at boot up, + * appears to be either the BIOS or Linux ACPI fault */ +#if 0 + /* Assume that the BIOS does not lie through the OpRegion... */ + if (dev_priv->opregion.lid_state) + return ioread32(dev_priv->opregion.lid_state) & 0x1 ? + connector_status_connected : + connector_status_disconnected; +#endif + + return connector_status_unknown; +} diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c new file mode 100644 index 00000000..2d6039b4 --- /dev/null +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c @@ -0,0 +1,1373 @@ +/* + * Copyright © 2008-2010 Intel Corporation + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS + * IN THE SOFTWARE. + * + * Authors: + * Eric Anholt + * Zou Nan hai + * Xiang Hai hao + * + */ + +#include "drmP.h" +#include "drm.h" +#include "i915_drv.h" +#include "i915_drm.h" +#include "i915_trace.h" +#include "intel_drv.h" + +static inline int ring_space(struct intel_ring_buffer *ring) +{ + int space = (ring->head & HEAD_ADDR) - (ring->tail + 8); + if (space < 0) + space += ring->size; + return space; +} + +static u32 i915_gem_get_seqno(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + u32 seqno; + + seqno = dev_priv->next_seqno; + + /* reserve 0 for non-seqno */ + if (++dev_priv->next_seqno == 0) + dev_priv->next_seqno = 1; + + return seqno; +} + +static int +render_ring_flush(struct intel_ring_buffer *ring, + u32 invalidate_domains, + u32 flush_domains) +{ + struct drm_device *dev = ring->dev; + u32 cmd; + int ret; + + /* + * read/write caches: + * + * I915_GEM_DOMAIN_RENDER is always invalidated, but is + * only flushed if MI_NO_WRITE_FLUSH is unset. On 965, it is + * also flushed at 2d versus 3d pipeline switches. + * + * read-only caches: + * + * I915_GEM_DOMAIN_SAMPLER is flushed on pre-965 if + * MI_READ_FLUSH is set, and is always flushed on 965. + * + * I915_GEM_DOMAIN_COMMAND may not exist? + * + * I915_GEM_DOMAIN_INSTRUCTION, which exists on 965, is + * invalidated when MI_EXE_FLUSH is set. + * + * I915_GEM_DOMAIN_VERTEX, which exists on 965, is + * invalidated with every MI_FLUSH. + * + * TLBs: + * + * On 965, TLBs associated with I915_GEM_DOMAIN_COMMAND + * and I915_GEM_DOMAIN_CPU in are invalidated at PTE write and + * I915_GEM_DOMAIN_RENDER and I915_GEM_DOMAIN_SAMPLER + * are flushed at any MI_FLUSH. + */ + + cmd = MI_FLUSH | MI_NO_WRITE_FLUSH; + if ((invalidate_domains|flush_domains) & + I915_GEM_DOMAIN_RENDER) + cmd &= ~MI_NO_WRITE_FLUSH; + if (INTEL_INFO(dev)->gen < 4) { + /* + * On the 965, the sampler cache always gets flushed + * and this bit is reserved. + */ + if (invalidate_domains & I915_GEM_DOMAIN_SAMPLER) + cmd |= MI_READ_FLUSH; + } + if (invalidate_domains & I915_GEM_DOMAIN_INSTRUCTION) + cmd |= MI_EXE_FLUSH; + + if (invalidate_domains & I915_GEM_DOMAIN_COMMAND && + (IS_G4X(dev) || IS_GEN5(dev))) + cmd |= MI_INVALIDATE_ISP; + + ret = intel_ring_begin(ring, 2); + if (ret) + return ret; + + intel_ring_emit(ring, cmd); + intel_ring_emit(ring, MI_NOOP); + intel_ring_advance(ring); + + return 0; +} + +static void ring_write_tail(struct intel_ring_buffer *ring, + u32 value) +{ + drm_i915_private_t *dev_priv = ring->dev->dev_private; + I915_WRITE_TAIL(ring, value); +} + +u32 intel_ring_get_active_head(struct intel_ring_buffer *ring) +{ + drm_i915_private_t *dev_priv = ring->dev->dev_private; + u32 acthd_reg = INTEL_INFO(ring->dev)->gen >= 4 ? + RING_ACTHD(ring->mmio_base) : ACTHD; + + return I915_READ(acthd_reg); +} + +static int init_ring_common(struct intel_ring_buffer *ring) +{ + drm_i915_private_t *dev_priv = ring->dev->dev_private; + struct drm_i915_gem_object *obj = ring->obj; + u32 head; + + /* Stop the ring if it's running. */ + I915_WRITE_CTL(ring, 0); + I915_WRITE_HEAD(ring, 0); + ring->write_tail(ring, 0); + + /* Initialize the ring. */ + I915_WRITE_START(ring, obj->gtt_offset); + head = I915_READ_HEAD(ring) & HEAD_ADDR; + + /* G45 ring initialization fails to reset head to zero */ + if (head != 0) { + DRM_DEBUG_KMS("%s head not reset to zero " + "ctl %08x head %08x tail %08x start %08x\n", + ring->name, + I915_READ_CTL(ring), + I915_READ_HEAD(ring), + I915_READ_TAIL(ring), + I915_READ_START(ring)); + + I915_WRITE_HEAD(ring, 0); + + if (I915_READ_HEAD(ring) & HEAD_ADDR) { + DRM_ERROR("failed to set %s head to zero " + "ctl %08x head %08x tail %08x start %08x\n", + ring->name, + I915_READ_CTL(ring), + I915_READ_HEAD(ring), + I915_READ_TAIL(ring), + I915_READ_START(ring)); + } + } + + I915_WRITE_CTL(ring, + ((ring->size - PAGE_SIZE) & RING_NR_PAGES) + | RING_REPORT_64K | RING_VALID); + + /* If the head is still not zero, the ring is dead */ + if ((I915_READ_CTL(ring) & RING_VALID) == 0 || + I915_READ_START(ring) != obj->gtt_offset || + (I915_READ_HEAD(ring) & HEAD_ADDR) != 0) { + DRM_ERROR("%s initialization failed " + "ctl %08x head %08x tail %08x start %08x\n", + ring->name, + I915_READ_CTL(ring), + I915_READ_HEAD(ring), + I915_READ_TAIL(ring), + I915_READ_START(ring)); + return -EIO; + } + + if (!drm_core_check_feature(ring->dev, DRIVER_MODESET)) + i915_kernel_lost_context(ring->dev); + else { + ring->head = I915_READ_HEAD(ring); + ring->tail = I915_READ_TAIL(ring) & TAIL_ADDR; + ring->space = ring_space(ring); + } + + return 0; +} + +/* + * 965+ support PIPE_CONTROL commands, which provide finer grained control + * over cache flushing. + */ +struct pipe_control { + struct drm_i915_gem_object *obj; + volatile u32 *cpu_page; + u32 gtt_offset; +}; + +static int +init_pipe_control(struct intel_ring_buffer *ring) +{ + struct pipe_control *pc; + struct drm_i915_gem_object *obj; + int ret; + + if (ring->private) + return 0; + + pc = kmalloc(sizeof(*pc), GFP_KERNEL); + if (!pc) + return -ENOMEM; + + obj = i915_gem_alloc_object(ring->dev, 4096); + if (obj == NULL) { + DRM_ERROR("Failed to allocate seqno page\n"); + ret = -ENOMEM; + goto err; + } + obj->cache_level = I915_CACHE_LLC; + + ret = i915_gem_object_pin(obj, 4096, true); + if (ret) + goto err_unref; + + pc->gtt_offset = obj->gtt_offset; + pc->cpu_page = kmap(obj->pages[0]); + if (pc->cpu_page == NULL) + goto err_unpin; + + pc->obj = obj; + ring->private = pc; + return 0; + +err_unpin: + i915_gem_object_unpin(obj); +err_unref: + drm_gem_object_unreference(&obj->base); +err: + kfree(pc); + return ret; +} + +static void +cleanup_pipe_control(struct intel_ring_buffer *ring) +{ + struct pipe_control *pc = ring->private; + struct drm_i915_gem_object *obj; + + if (!ring->private) + return; + + obj = pc->obj; + kunmap(obj->pages[0]); + i915_gem_object_unpin(obj); + drm_gem_object_unreference(&obj->base); + + kfree(pc); + ring->private = NULL; +} + +static int init_render_ring(struct intel_ring_buffer *ring) +{ + struct drm_device *dev = ring->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + int ret = init_ring_common(ring); + + if (INTEL_INFO(dev)->gen > 3) { + int mode = VS_TIMER_DISPATCH << 16 | VS_TIMER_DISPATCH; + if (IS_GEN6(dev) || IS_GEN7(dev)) + mode |= MI_FLUSH_ENABLE << 16 | MI_FLUSH_ENABLE; + I915_WRITE(MI_MODE, mode); + } + + if (INTEL_INFO(dev)->gen >= 6) { + } else if (IS_GEN5(dev)) { + ret = init_pipe_control(ring); + if (ret) + return ret; + } + + return ret; +} + +static void render_ring_cleanup(struct intel_ring_buffer *ring) +{ + if (!ring->private) + return; + + cleanup_pipe_control(ring); +} + +static void +update_semaphore(struct intel_ring_buffer *ring, int i, u32 seqno) +{ + struct drm_device *dev = ring->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + int id; + + /* + * cs -> 1 = vcs, 0 = bcs + * vcs -> 1 = bcs, 0 = cs, + * bcs -> 1 = cs, 0 = vcs. + */ + id = ring - dev_priv->ring; + id += 2 - i; + id %= 3; + + intel_ring_emit(ring, + MI_SEMAPHORE_MBOX | + MI_SEMAPHORE_REGISTER | + MI_SEMAPHORE_UPDATE); + intel_ring_emit(ring, seqno); + intel_ring_emit(ring, + RING_SYNC_0(dev_priv->ring[id].mmio_base) + 4*i); +} + +static int +gen6_add_request(struct intel_ring_buffer *ring, + u32 *result) +{ + u32 seqno; + int ret; + + ret = intel_ring_begin(ring, 10); + if (ret) + return ret; + + seqno = i915_gem_get_seqno(ring->dev); + update_semaphore(ring, 0, seqno); + update_semaphore(ring, 1, seqno); + + intel_ring_emit(ring, MI_STORE_DWORD_INDEX); + intel_ring_emit(ring, I915_GEM_HWS_INDEX << MI_STORE_DWORD_INDEX_SHIFT); + intel_ring_emit(ring, seqno); + intel_ring_emit(ring, MI_USER_INTERRUPT); + intel_ring_advance(ring); + + *result = seqno; + return 0; +} + +int +intel_ring_sync(struct intel_ring_buffer *ring, + struct intel_ring_buffer *to, + u32 seqno) +{ + int ret; + + ret = intel_ring_begin(ring, 4); + if (ret) + return ret; + + intel_ring_emit(ring, + MI_SEMAPHORE_MBOX | + MI_SEMAPHORE_REGISTER | + intel_ring_sync_index(ring, to) << 17 | + MI_SEMAPHORE_COMPARE); + intel_ring_emit(ring, seqno); + intel_ring_emit(ring, 0); + intel_ring_emit(ring, MI_NOOP); + intel_ring_advance(ring); + + return 0; +} + +#define PIPE_CONTROL_FLUSH(ring__, addr__) \ +do { \ + intel_ring_emit(ring__, GFX_OP_PIPE_CONTROL | PIPE_CONTROL_QW_WRITE | \ + PIPE_CONTROL_DEPTH_STALL | 2); \ + intel_ring_emit(ring__, (addr__) | PIPE_CONTROL_GLOBAL_GTT); \ + intel_ring_emit(ring__, 0); \ + intel_ring_emit(ring__, 0); \ +} while (0) + +static int +pc_render_add_request(struct intel_ring_buffer *ring, + u32 *result) +{ + struct drm_device *dev = ring->dev; + u32 seqno = i915_gem_get_seqno(dev); + struct pipe_control *pc = ring->private; + u32 scratch_addr = pc->gtt_offset + 128; + int ret; + + /* For Ironlake, MI_USER_INTERRUPT was deprecated and apparently + * incoherent with writes to memory, i.e. completely fubar, + * so we need to use PIPE_NOTIFY instead. + * + * However, we also need to workaround the qword write + * incoherence by flushing the 6 PIPE_NOTIFY buffers out to + * memory before requesting an interrupt. + */ + ret = intel_ring_begin(ring, 32); + if (ret) + return ret; + + intel_ring_emit(ring, GFX_OP_PIPE_CONTROL | PIPE_CONTROL_QW_WRITE | + PIPE_CONTROL_WC_FLUSH | PIPE_CONTROL_TC_FLUSH); + intel_ring_emit(ring, pc->gtt_offset | PIPE_CONTROL_GLOBAL_GTT); + intel_ring_emit(ring, seqno); + intel_ring_emit(ring, 0); + PIPE_CONTROL_FLUSH(ring, scratch_addr); + scratch_addr += 128; /* write to separate cachelines */ + PIPE_CONTROL_FLUSH(ring, scratch_addr); + scratch_addr += 128; + PIPE_CONTROL_FLUSH(ring, scratch_addr); + scratch_addr += 128; + PIPE_CONTROL_FLUSH(ring, scratch_addr); + scratch_addr += 128; + PIPE_CONTROL_FLUSH(ring, scratch_addr); + scratch_addr += 128; + PIPE_CONTROL_FLUSH(ring, scratch_addr); + intel_ring_emit(ring, GFX_OP_PIPE_CONTROL | PIPE_CONTROL_QW_WRITE | + PIPE_CONTROL_WC_FLUSH | PIPE_CONTROL_TC_FLUSH | + PIPE_CONTROL_NOTIFY); + intel_ring_emit(ring, pc->gtt_offset | PIPE_CONTROL_GLOBAL_GTT); + intel_ring_emit(ring, seqno); + intel_ring_emit(ring, 0); + intel_ring_advance(ring); + + *result = seqno; + return 0; +} + +static int +render_ring_add_request(struct intel_ring_buffer *ring, + u32 *result) +{ + struct drm_device *dev = ring->dev; + u32 seqno = i915_gem_get_seqno(dev); + int ret; + + ret = intel_ring_begin(ring, 4); + if (ret) + return ret; + + intel_ring_emit(ring, MI_STORE_DWORD_INDEX); + intel_ring_emit(ring, I915_GEM_HWS_INDEX << MI_STORE_DWORD_INDEX_SHIFT); + intel_ring_emit(ring, seqno); + intel_ring_emit(ring, MI_USER_INTERRUPT); + intel_ring_advance(ring); + + *result = seqno; + return 0; +} + +static u32 +ring_get_seqno(struct intel_ring_buffer *ring) +{ + return intel_read_status_page(ring, I915_GEM_HWS_INDEX); +} + +static u32 +pc_render_get_seqno(struct intel_ring_buffer *ring) +{ + struct pipe_control *pc = ring->private; + return pc->cpu_page[0]; +} + +static void +ironlake_enable_irq(drm_i915_private_t *dev_priv, u32 mask) +{ + dev_priv->gt_irq_mask &= ~mask; + I915_WRITE(GTIMR, dev_priv->gt_irq_mask); + POSTING_READ(GTIMR); +} + +static void +ironlake_disable_irq(drm_i915_private_t *dev_priv, u32 mask) +{ + dev_priv->gt_irq_mask |= mask; + I915_WRITE(GTIMR, dev_priv->gt_irq_mask); + POSTING_READ(GTIMR); +} + +static void +i915_enable_irq(drm_i915_private_t *dev_priv, u32 mask) +{ + dev_priv->irq_mask &= ~mask; + I915_WRITE(IMR, dev_priv->irq_mask); + POSTING_READ(IMR); +} + +static void +i915_disable_irq(drm_i915_private_t *dev_priv, u32 mask) +{ + dev_priv->irq_mask |= mask; + I915_WRITE(IMR, dev_priv->irq_mask); + POSTING_READ(IMR); +} + +static bool +render_ring_get_irq(struct intel_ring_buffer *ring) +{ + struct drm_device *dev = ring->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + + if (!dev->irq_enabled) + return false; + + spin_lock(&ring->irq_lock); + if (ring->irq_refcount++ == 0) { + if (HAS_PCH_SPLIT(dev)) + ironlake_enable_irq(dev_priv, + GT_PIPE_NOTIFY | GT_USER_INTERRUPT); + else + i915_enable_irq(dev_priv, I915_USER_INTERRUPT); + } + spin_unlock(&ring->irq_lock); + + return true; +} + +static void +render_ring_put_irq(struct intel_ring_buffer *ring) +{ + struct drm_device *dev = ring->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + + spin_lock(&ring->irq_lock); + if (--ring->irq_refcount == 0) { + if (HAS_PCH_SPLIT(dev)) + ironlake_disable_irq(dev_priv, + GT_USER_INTERRUPT | + GT_PIPE_NOTIFY); + else + i915_disable_irq(dev_priv, I915_USER_INTERRUPT); + } + spin_unlock(&ring->irq_lock); +} + +void intel_ring_setup_status_page(struct intel_ring_buffer *ring) +{ + struct drm_device *dev = ring->dev; + drm_i915_private_t *dev_priv = ring->dev->dev_private; + u32 mmio = 0; + + /* The ring status page addresses are no longer next to the rest of + * the ring registers as of gen7. + */ + if (IS_GEN7(dev)) { + switch (ring->id) { + case RING_RENDER: + mmio = RENDER_HWS_PGA_GEN7; + break; + case RING_BLT: + mmio = BLT_HWS_PGA_GEN7; + break; + case RING_BSD: + mmio = BSD_HWS_PGA_GEN7; + break; + } + } else if (IS_GEN6(ring->dev)) { + mmio = RING_HWS_PGA_GEN6(ring->mmio_base); + } else { + mmio = RING_HWS_PGA(ring->mmio_base); + } + + I915_WRITE(mmio, (u32)ring->status_page.gfx_addr); + POSTING_READ(mmio); +} + +static int +bsd_ring_flush(struct intel_ring_buffer *ring, + u32 invalidate_domains, + u32 flush_domains) +{ + int ret; + + ret = intel_ring_begin(ring, 2); + if (ret) + return ret; + + intel_ring_emit(ring, MI_FLUSH); + intel_ring_emit(ring, MI_NOOP); + intel_ring_advance(ring); + return 0; +} + +static int +ring_add_request(struct intel_ring_buffer *ring, + u32 *result) +{ + u32 seqno; + int ret; + + ret = intel_ring_begin(ring, 4); + if (ret) + return ret; + + seqno = i915_gem_get_seqno(ring->dev); + + intel_ring_emit(ring, MI_STORE_DWORD_INDEX); + intel_ring_emit(ring, I915_GEM_HWS_INDEX << MI_STORE_DWORD_INDEX_SHIFT); + intel_ring_emit(ring, seqno); + intel_ring_emit(ring, MI_USER_INTERRUPT); + intel_ring_advance(ring); + + *result = seqno; + return 0; +} + +static bool +gen6_ring_get_irq(struct intel_ring_buffer *ring, u32 gflag, u32 rflag) +{ + struct drm_device *dev = ring->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + + if (!dev->irq_enabled) + return false; + + spin_lock(&ring->irq_lock); + if (ring->irq_refcount++ == 0) { + ring->irq_mask &= ~rflag; + I915_WRITE_IMR(ring, ring->irq_mask); + ironlake_enable_irq(dev_priv, gflag); + } + spin_unlock(&ring->irq_lock); + + return true; +} + +static void +gen6_ring_put_irq(struct intel_ring_buffer *ring, u32 gflag, u32 rflag) +{ + struct drm_device *dev = ring->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + + spin_lock(&ring->irq_lock); + if (--ring->irq_refcount == 0) { + ring->irq_mask |= rflag; + I915_WRITE_IMR(ring, ring->irq_mask); + ironlake_disable_irq(dev_priv, gflag); + } + spin_unlock(&ring->irq_lock); +} + +static bool +bsd_ring_get_irq(struct intel_ring_buffer *ring) +{ + struct drm_device *dev = ring->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + + if (!dev->irq_enabled) + return false; + + spin_lock(&ring->irq_lock); + if (ring->irq_refcount++ == 0) { + if (IS_G4X(dev)) + i915_enable_irq(dev_priv, I915_BSD_USER_INTERRUPT); + else + ironlake_enable_irq(dev_priv, GT_BSD_USER_INTERRUPT); + } + spin_unlock(&ring->irq_lock); + + return true; +} +static void +bsd_ring_put_irq(struct intel_ring_buffer *ring) +{ + struct drm_device *dev = ring->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + + spin_lock(&ring->irq_lock); + if (--ring->irq_refcount == 0) { + if (IS_G4X(dev)) + i915_disable_irq(dev_priv, I915_BSD_USER_INTERRUPT); + else + ironlake_disable_irq(dev_priv, GT_BSD_USER_INTERRUPT); + } + spin_unlock(&ring->irq_lock); +} + +static int +ring_dispatch_execbuffer(struct intel_ring_buffer *ring, u32 offset, u32 length) +{ + int ret; + + ret = intel_ring_begin(ring, 2); + if (ret) + return ret; + + intel_ring_emit(ring, + MI_BATCH_BUFFER_START | (2 << 6) | + MI_BATCH_NON_SECURE_I965); + intel_ring_emit(ring, offset); + intel_ring_advance(ring); + + return 0; +} + +static int +render_ring_dispatch_execbuffer(struct intel_ring_buffer *ring, + u32 offset, u32 len) +{ + struct drm_device *dev = ring->dev; + int ret; + + if (IS_I830(dev) || IS_845G(dev)) { + ret = intel_ring_begin(ring, 4); + if (ret) + return ret; + + intel_ring_emit(ring, MI_BATCH_BUFFER); + intel_ring_emit(ring, offset | MI_BATCH_NON_SECURE); + intel_ring_emit(ring, offset + len - 8); + intel_ring_emit(ring, 0); + } else { + ret = intel_ring_begin(ring, 2); + if (ret) + return ret; + + if (INTEL_INFO(dev)->gen >= 4) { + intel_ring_emit(ring, + MI_BATCH_BUFFER_START | (2 << 6) | + MI_BATCH_NON_SECURE_I965); + intel_ring_emit(ring, offset); + } else { + intel_ring_emit(ring, + MI_BATCH_BUFFER_START | (2 << 6)); + intel_ring_emit(ring, offset | MI_BATCH_NON_SECURE); + } + } + intel_ring_advance(ring); + + return 0; +} + +static void cleanup_status_page(struct intel_ring_buffer *ring) +{ + drm_i915_private_t *dev_priv = ring->dev->dev_private; + struct drm_i915_gem_object *obj; + + obj = ring->status_page.obj; + if (obj == NULL) + return; + + kunmap(obj->pages[0]); + i915_gem_object_unpin(obj); + drm_gem_object_unreference(&obj->base); + ring->status_page.obj = NULL; + + memset(&dev_priv->hws_map, 0, sizeof(dev_priv->hws_map)); +} + +static int init_status_page(struct intel_ring_buffer *ring) +{ + struct drm_device *dev = ring->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_i915_gem_object *obj; + int ret; + + obj = i915_gem_alloc_object(dev, 4096); + if (obj == NULL) { + DRM_ERROR("Failed to allocate status page\n"); + ret = -ENOMEM; + goto err; + } + obj->cache_level = I915_CACHE_LLC; + + ret = i915_gem_object_pin(obj, 4096, true); + if (ret != 0) { + goto err_unref; + } + + ring->status_page.gfx_addr = obj->gtt_offset; + ring->status_page.page_addr = kmap(obj->pages[0]); + if (ring->status_page.page_addr == NULL) { + memset(&dev_priv->hws_map, 0, sizeof(dev_priv->hws_map)); + goto err_unpin; + } + ring->status_page.obj = obj; + memset(ring->status_page.page_addr, 0, PAGE_SIZE); + + intel_ring_setup_status_page(ring); + DRM_DEBUG_DRIVER("%s hws offset: 0x%08x\n", + ring->name, ring->status_page.gfx_addr); + + return 0; + +err_unpin: + i915_gem_object_unpin(obj); +err_unref: + drm_gem_object_unreference(&obj->base); +err: + return ret; +} + +int intel_init_ring_buffer(struct drm_device *dev, + struct intel_ring_buffer *ring) +{ + struct drm_i915_gem_object *obj; + int ret; + + ring->dev = dev; + INIT_LIST_HEAD(&ring->active_list); + INIT_LIST_HEAD(&ring->request_list); + INIT_LIST_HEAD(&ring->gpu_write_list); + + init_waitqueue_head(&ring->irq_queue); + spin_lock_init(&ring->irq_lock); + ring->irq_mask = ~0; + + if (I915_NEED_GFX_HWS(dev)) { + ret = init_status_page(ring); + if (ret) + return ret; + } + + obj = i915_gem_alloc_object(dev, ring->size); + if (obj == NULL) { + DRM_ERROR("Failed to allocate ringbuffer\n"); + ret = -ENOMEM; + goto err_hws; + } + + ring->obj = obj; + + ret = i915_gem_object_pin(obj, PAGE_SIZE, true); + if (ret) + goto err_unref; + + ring->map.size = ring->size; + ring->map.offset = dev->agp->base + obj->gtt_offset; + ring->map.type = 0; + ring->map.flags = 0; + ring->map.mtrr = 0; + + drm_core_ioremap_wc(&ring->map, dev); + if (ring->map.handle == NULL) { + DRM_ERROR("Failed to map ringbuffer.\n"); + ret = -EINVAL; + goto err_unpin; + } + + ring->virtual_start = ring->map.handle; + ret = ring->init(ring); + if (ret) + goto err_unmap; + + /* Workaround an erratum on the i830 which causes a hang if + * the TAIL pointer points to within the last 2 cachelines + * of the buffer. + */ + ring->effective_size = ring->size; + if (IS_I830(ring->dev) || IS_845G(ring->dev)) + ring->effective_size -= 128; + + return 0; + +err_unmap: + drm_core_ioremapfree(&ring->map, dev); +err_unpin: + i915_gem_object_unpin(obj); +err_unref: + drm_gem_object_unreference(&obj->base); + ring->obj = NULL; +err_hws: + cleanup_status_page(ring); + return ret; +} + +void intel_cleanup_ring_buffer(struct intel_ring_buffer *ring) +{ + struct drm_i915_private *dev_priv; + int ret; + + if (ring->obj == NULL) + return; + + /* Disable the ring buffer. The ring must be idle at this point */ + dev_priv = ring->dev->dev_private; + ret = intel_wait_ring_idle(ring); + if (ret) + DRM_ERROR("failed to quiesce %s whilst cleaning up: %d\n", + ring->name, ret); + + I915_WRITE_CTL(ring, 0); + + drm_core_ioremapfree(&ring->map, ring->dev); + + i915_gem_object_unpin(ring->obj); + drm_gem_object_unreference(&ring->obj->base); + ring->obj = NULL; + + if (ring->cleanup) + ring->cleanup(ring); + + cleanup_status_page(ring); +} + +static int intel_wrap_ring_buffer(struct intel_ring_buffer *ring) +{ + unsigned int *virt; + int rem = ring->size - ring->tail; + + if (ring->space < rem) { + int ret = intel_wait_ring_buffer(ring, rem); + if (ret) + return ret; + } + + virt = (unsigned int *)(ring->virtual_start + ring->tail); + rem /= 8; + while (rem--) { + *virt++ = MI_NOOP; + *virt++ = MI_NOOP; + } + + ring->tail = 0; + ring->space = ring_space(ring); + + return 0; +} + +int intel_wait_ring_buffer(struct intel_ring_buffer *ring, int n) +{ + struct drm_device *dev = ring->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + unsigned long end; + u32 head; + + /* If the reported head position has wrapped or hasn't advanced, + * fallback to the slow and accurate path. + */ + head = intel_read_status_page(ring, 4); + if (head > ring->head) { + ring->head = head; + ring->space = ring_space(ring); + if (ring->space >= n) + return 0; + } + + trace_i915_ring_wait_begin(ring); + end = jiffies + 3 * HZ; + do { + ring->head = I915_READ_HEAD(ring); + ring->space = ring_space(ring); + if (ring->space >= n) { + trace_i915_ring_wait_end(ring); + return 0; + } + + if (dev->primary->master) { + struct drm_i915_master_private *master_priv = dev->primary->master->driver_priv; + if (master_priv->sarea_priv) + master_priv->sarea_priv->perf_boxes |= I915_BOX_WAIT; + } + + msleep(1); + if (atomic_read(&dev_priv->mm.wedged)) + return -EAGAIN; + } while (!time_after(jiffies, end)); + trace_i915_ring_wait_end(ring); + return -EBUSY; +} + +int intel_ring_begin(struct intel_ring_buffer *ring, + int num_dwords) +{ + struct drm_i915_private *dev_priv = ring->dev->dev_private; + int n = 4*num_dwords; + int ret; + + if (unlikely(atomic_read(&dev_priv->mm.wedged))) + return -EIO; + + if (unlikely(ring->tail + n > ring->effective_size)) { + ret = intel_wrap_ring_buffer(ring); + if (unlikely(ret)) + return ret; + } + + if (unlikely(ring->space < n)) { + ret = intel_wait_ring_buffer(ring, n); + if (unlikely(ret)) + return ret; + } + + ring->space -= n; + return 0; +} + +void intel_ring_advance(struct intel_ring_buffer *ring) +{ + ring->tail &= ring->size - 1; + ring->write_tail(ring, ring->tail); +} + +static const struct intel_ring_buffer render_ring = { + .name = "render ring", + .id = RING_RENDER, + .mmio_base = RENDER_RING_BASE, + .size = 32 * PAGE_SIZE, + .init = init_render_ring, + .write_tail = ring_write_tail, + .flush = render_ring_flush, + .add_request = render_ring_add_request, + .get_seqno = ring_get_seqno, + .irq_get = render_ring_get_irq, + .irq_put = render_ring_put_irq, + .dispatch_execbuffer = render_ring_dispatch_execbuffer, + .cleanup = render_ring_cleanup, +}; + +/* ring buffer for bit-stream decoder */ + +static const struct intel_ring_buffer bsd_ring = { + .name = "bsd ring", + .id = RING_BSD, + .mmio_base = BSD_RING_BASE, + .size = 32 * PAGE_SIZE, + .init = init_ring_common, + .write_tail = ring_write_tail, + .flush = bsd_ring_flush, + .add_request = ring_add_request, + .get_seqno = ring_get_seqno, + .irq_get = bsd_ring_get_irq, + .irq_put = bsd_ring_put_irq, + .dispatch_execbuffer = ring_dispatch_execbuffer, +}; + + +static void gen6_bsd_ring_write_tail(struct intel_ring_buffer *ring, + u32 value) +{ + drm_i915_private_t *dev_priv = ring->dev->dev_private; + + /* Every tail move must follow the sequence below */ + I915_WRITE(GEN6_BSD_SLEEP_PSMI_CONTROL, + GEN6_BSD_SLEEP_PSMI_CONTROL_RC_ILDL_MESSAGE_MODIFY_MASK | + GEN6_BSD_SLEEP_PSMI_CONTROL_RC_ILDL_MESSAGE_DISABLE); + I915_WRITE(GEN6_BSD_RNCID, 0x0); + + if (wait_for((I915_READ(GEN6_BSD_SLEEP_PSMI_CONTROL) & + GEN6_BSD_SLEEP_PSMI_CONTROL_IDLE_INDICATOR) == 0, + 50)) + DRM_ERROR("timed out waiting for IDLE Indicator\n"); + + I915_WRITE_TAIL(ring, value); + I915_WRITE(GEN6_BSD_SLEEP_PSMI_CONTROL, + GEN6_BSD_SLEEP_PSMI_CONTROL_RC_ILDL_MESSAGE_MODIFY_MASK | + GEN6_BSD_SLEEP_PSMI_CONTROL_RC_ILDL_MESSAGE_ENABLE); +} + +static int gen6_ring_flush(struct intel_ring_buffer *ring, + u32 invalidate, u32 flush) +{ + uint32_t cmd; + int ret; + + ret = intel_ring_begin(ring, 4); + if (ret) + return ret; + + cmd = MI_FLUSH_DW; + if (invalidate & I915_GEM_GPU_DOMAINS) + cmd |= MI_INVALIDATE_TLB | MI_INVALIDATE_BSD; + intel_ring_emit(ring, cmd); + intel_ring_emit(ring, 0); + intel_ring_emit(ring, 0); + intel_ring_emit(ring, MI_NOOP); + intel_ring_advance(ring); + return 0; +} + +static int +gen6_ring_dispatch_execbuffer(struct intel_ring_buffer *ring, + u32 offset, u32 len) +{ + int ret; + + ret = intel_ring_begin(ring, 2); + if (ret) + return ret; + + intel_ring_emit(ring, MI_BATCH_BUFFER_START | MI_BATCH_NON_SECURE_I965); + /* bit0-7 is the length on GEN6+ */ + intel_ring_emit(ring, offset); + intel_ring_advance(ring); + + return 0; +} + +static bool +gen6_render_ring_get_irq(struct intel_ring_buffer *ring) +{ + return gen6_ring_get_irq(ring, + GT_USER_INTERRUPT, + GEN6_RENDER_USER_INTERRUPT); +} + +static void +gen6_render_ring_put_irq(struct intel_ring_buffer *ring) +{ + return gen6_ring_put_irq(ring, + GT_USER_INTERRUPT, + GEN6_RENDER_USER_INTERRUPT); +} + +static bool +gen6_bsd_ring_get_irq(struct intel_ring_buffer *ring) +{ + return gen6_ring_get_irq(ring, + GT_GEN6_BSD_USER_INTERRUPT, + GEN6_BSD_USER_INTERRUPT); +} + +static void +gen6_bsd_ring_put_irq(struct intel_ring_buffer *ring) +{ + return gen6_ring_put_irq(ring, + GT_GEN6_BSD_USER_INTERRUPT, + GEN6_BSD_USER_INTERRUPT); +} + +/* ring buffer for Video Codec for Gen6+ */ +static const struct intel_ring_buffer gen6_bsd_ring = { + .name = "gen6 bsd ring", + .id = RING_BSD, + .mmio_base = GEN6_BSD_RING_BASE, + .size = 32 * PAGE_SIZE, + .init = init_ring_common, + .write_tail = gen6_bsd_ring_write_tail, + .flush = gen6_ring_flush, + .add_request = gen6_add_request, + .get_seqno = ring_get_seqno, + .irq_get = gen6_bsd_ring_get_irq, + .irq_put = gen6_bsd_ring_put_irq, + .dispatch_execbuffer = gen6_ring_dispatch_execbuffer, +}; + +/* Blitter support (SandyBridge+) */ + +static bool +blt_ring_get_irq(struct intel_ring_buffer *ring) +{ + return gen6_ring_get_irq(ring, + GT_BLT_USER_INTERRUPT, + GEN6_BLITTER_USER_INTERRUPT); +} + +static void +blt_ring_put_irq(struct intel_ring_buffer *ring) +{ + gen6_ring_put_irq(ring, + GT_BLT_USER_INTERRUPT, + GEN6_BLITTER_USER_INTERRUPT); +} + + +/* Workaround for some stepping of SNB, + * each time when BLT engine ring tail moved, + * the first command in the ring to be parsed + * should be MI_BATCH_BUFFER_START + */ +#define NEED_BLT_WORKAROUND(dev) \ + (IS_GEN6(dev) && (dev->pdev->revision < 8)) + +static inline struct drm_i915_gem_object * +to_blt_workaround(struct intel_ring_buffer *ring) +{ + return ring->private; +} + +static int blt_ring_init(struct intel_ring_buffer *ring) +{ + if (NEED_BLT_WORKAROUND(ring->dev)) { + struct drm_i915_gem_object *obj; + u32 *ptr; + int ret; + + obj = i915_gem_alloc_object(ring->dev, 4096); + if (obj == NULL) + return -ENOMEM; + + ret = i915_gem_object_pin(obj, 4096, true); + if (ret) { + drm_gem_object_unreference(&obj->base); + return ret; + } + + ptr = kmap(obj->pages[0]); + *ptr++ = MI_BATCH_BUFFER_END; + *ptr++ = MI_NOOP; + kunmap(obj->pages[0]); + + ret = i915_gem_object_set_to_gtt_domain(obj, false); + if (ret) { + i915_gem_object_unpin(obj); + drm_gem_object_unreference(&obj->base); + return ret; + } + + ring->private = obj; + } + + return init_ring_common(ring); +} + +static int blt_ring_begin(struct intel_ring_buffer *ring, + int num_dwords) +{ + if (ring->private) { + int ret = intel_ring_begin(ring, num_dwords+2); + if (ret) + return ret; + + intel_ring_emit(ring, MI_BATCH_BUFFER_START); + intel_ring_emit(ring, to_blt_workaround(ring)->gtt_offset); + + return 0; + } else + return intel_ring_begin(ring, 4); +} + +static int blt_ring_flush(struct intel_ring_buffer *ring, + u32 invalidate, u32 flush) +{ + uint32_t cmd; + int ret; + + ret = blt_ring_begin(ring, 4); + if (ret) + return ret; + + cmd = MI_FLUSH_DW; + if (invalidate & I915_GEM_DOMAIN_RENDER) + cmd |= MI_INVALIDATE_TLB; + intel_ring_emit(ring, cmd); + intel_ring_emit(ring, 0); + intel_ring_emit(ring, 0); + intel_ring_emit(ring, MI_NOOP); + intel_ring_advance(ring); + return 0; +} + +static void blt_ring_cleanup(struct intel_ring_buffer *ring) +{ + if (!ring->private) + return; + + i915_gem_object_unpin(ring->private); + drm_gem_object_unreference(ring->private); + ring->private = NULL; +} + +static const struct intel_ring_buffer gen6_blt_ring = { + .name = "blt ring", + .id = RING_BLT, + .mmio_base = BLT_RING_BASE, + .size = 32 * PAGE_SIZE, + .init = blt_ring_init, + .write_tail = ring_write_tail, + .flush = blt_ring_flush, + .add_request = gen6_add_request, + .get_seqno = ring_get_seqno, + .irq_get = blt_ring_get_irq, + .irq_put = blt_ring_put_irq, + .dispatch_execbuffer = gen6_ring_dispatch_execbuffer, + .cleanup = blt_ring_cleanup, +}; + +int intel_init_render_ring_buffer(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct intel_ring_buffer *ring = &dev_priv->ring[RCS]; + + *ring = render_ring; + if (INTEL_INFO(dev)->gen >= 6) { + ring->add_request = gen6_add_request; + ring->irq_get = gen6_render_ring_get_irq; + ring->irq_put = gen6_render_ring_put_irq; + } else if (IS_GEN5(dev)) { + ring->add_request = pc_render_add_request; + ring->get_seqno = pc_render_get_seqno; + } + + if (!I915_NEED_GFX_HWS(dev)) { + ring->status_page.page_addr = dev_priv->status_page_dmah->vaddr; + memset(ring->status_page.page_addr, 0, PAGE_SIZE); + } + + return intel_init_ring_buffer(dev, ring); +} + +int intel_render_ring_init_dri(struct drm_device *dev, u64 start, u32 size) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct intel_ring_buffer *ring = &dev_priv->ring[RCS]; + + *ring = render_ring; + if (INTEL_INFO(dev)->gen >= 6) { + ring->add_request = gen6_add_request; + ring->irq_get = gen6_render_ring_get_irq; + ring->irq_put = gen6_render_ring_put_irq; + } else if (IS_GEN5(dev)) { + ring->add_request = pc_render_add_request; + ring->get_seqno = pc_render_get_seqno; + } + + if (!I915_NEED_GFX_HWS(dev)) + ring->status_page.page_addr = dev_priv->status_page_dmah->vaddr; + + ring->dev = dev; + INIT_LIST_HEAD(&ring->active_list); + INIT_LIST_HEAD(&ring->request_list); + INIT_LIST_HEAD(&ring->gpu_write_list); + + ring->size = size; + ring->effective_size = ring->size; + if (IS_I830(ring->dev)) + ring->effective_size -= 128; + + ring->map.offset = start; + ring->map.size = size; + ring->map.type = 0; + ring->map.flags = 0; + ring->map.mtrr = 0; + + drm_core_ioremap_wc(&ring->map, dev); + if (ring->map.handle == NULL) { + DRM_ERROR("can not ioremap virtual address for" + " ring buffer\n"); + return -ENOMEM; + } + + ring->virtual_start = (void __force __iomem *)ring->map.handle; + return 0; +} + +int intel_init_bsd_ring_buffer(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct intel_ring_buffer *ring = &dev_priv->ring[VCS]; + + if (IS_GEN6(dev) || IS_GEN7(dev)) + *ring = gen6_bsd_ring; + else + *ring = bsd_ring; + + return intel_init_ring_buffer(dev, ring); +} + +int intel_init_blt_ring_buffer(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct intel_ring_buffer *ring = &dev_priv->ring[BCS]; + + *ring = gen6_blt_ring; + + return intel_init_ring_buffer(dev, ring); +} diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.h b/drivers/gpu/drm/i915/intel_ringbuffer.h new file mode 100644 index 00000000..39ac2b63 --- /dev/null +++ b/drivers/gpu/drm/i915/intel_ringbuffer.h @@ -0,0 +1,203 @@ +#ifndef _INTEL_RINGBUFFER_H_ +#define _INTEL_RINGBUFFER_H_ + +enum { + RCS = 0x0, + VCS, + BCS, + I915_NUM_RINGS, +}; + +struct intel_hw_status_page { + u32 __iomem *page_addr; + unsigned int gfx_addr; + struct drm_i915_gem_object *obj; +}; + +#define I915_READ_TAIL(ring) I915_READ(RING_TAIL((ring)->mmio_base)) +#define I915_WRITE_TAIL(ring, val) I915_WRITE(RING_TAIL((ring)->mmio_base), val) + +#define I915_READ_START(ring) I915_READ(RING_START((ring)->mmio_base)) +#define I915_WRITE_START(ring, val) I915_WRITE(RING_START((ring)->mmio_base), val) + +#define I915_READ_HEAD(ring) I915_READ(RING_HEAD((ring)->mmio_base)) +#define I915_WRITE_HEAD(ring, val) I915_WRITE(RING_HEAD((ring)->mmio_base), val) + +#define I915_READ_CTL(ring) I915_READ(RING_CTL((ring)->mmio_base)) +#define I915_WRITE_CTL(ring, val) I915_WRITE(RING_CTL((ring)->mmio_base), val) + +#define I915_READ_IMR(ring) I915_READ(RING_IMR((ring)->mmio_base)) +#define I915_WRITE_IMR(ring, val) I915_WRITE(RING_IMR((ring)->mmio_base), val) + +#define I915_READ_NOPID(ring) I915_READ(RING_NOPID((ring)->mmio_base)) +#define I915_READ_SYNC_0(ring) I915_READ(RING_SYNC_0((ring)->mmio_base)) +#define I915_READ_SYNC_1(ring) I915_READ(RING_SYNC_1((ring)->mmio_base)) + +struct intel_ring_buffer { + const char *name; + enum intel_ring_id { + RING_RENDER = 0x1, + RING_BSD = 0x2, + RING_BLT = 0x4, + } id; + u32 mmio_base; + void __iomem *virtual_start; + struct drm_device *dev; + struct drm_i915_gem_object *obj; + + u32 head; + u32 tail; + int space; + int size; + int effective_size; + struct intel_hw_status_page status_page; + + spinlock_t irq_lock; + u32 irq_refcount; + u32 irq_mask; + u32 irq_seqno; /* last seq seem at irq time */ + u32 trace_irq_seqno; + u32 waiting_seqno; + u32 sync_seqno[I915_NUM_RINGS-1]; + bool __must_check (*irq_get)(struct intel_ring_buffer *ring); + void (*irq_put)(struct intel_ring_buffer *ring); + + int (*init)(struct intel_ring_buffer *ring); + + void (*write_tail)(struct intel_ring_buffer *ring, + u32 value); + int __must_check (*flush)(struct intel_ring_buffer *ring, + u32 invalidate_domains, + u32 flush_domains); + int (*add_request)(struct intel_ring_buffer *ring, + u32 *seqno); + u32 (*get_seqno)(struct intel_ring_buffer *ring); + int (*dispatch_execbuffer)(struct intel_ring_buffer *ring, + u32 offset, u32 length); + void (*cleanup)(struct intel_ring_buffer *ring); + + /** + * List of objects currently involved in rendering from the + * ringbuffer. + * + * Includes buffers having the contents of their GPU caches + * flushed, not necessarily primitives. last_rendering_seqno + * represents when the rendering involved will be completed. + * + * A reference is held on the buffer while on this list. + */ + struct list_head active_list; + + /** + * List of breadcrumbs associated with GPU requests currently + * outstanding. + */ + struct list_head request_list; + + /** + * List of objects currently pending a GPU write flush. + * + * All elements on this list will belong to either the + * active_list or flushing_list, last_rendering_seqno can + * be used to differentiate between the two elements. + */ + struct list_head gpu_write_list; + + /** + * Do we have some not yet emitted requests outstanding? + */ + u32 outstanding_lazy_request; + + wait_queue_head_t irq_queue; + drm_local_map_t map; + + void *private; +}; + +static inline u32 +intel_ring_sync_index(struct intel_ring_buffer *ring, + struct intel_ring_buffer *other) +{ + int idx; + + /* + * cs -> 0 = vcs, 1 = bcs + * vcs -> 0 = bcs, 1 = cs, + * bcs -> 0 = cs, 1 = vcs. + */ + + idx = (other - ring) - 1; + if (idx < 0) + idx += I915_NUM_RINGS; + + return idx; +} + +static inline u32 +intel_read_status_page(struct intel_ring_buffer *ring, + int reg) +{ + return ioread32(ring->status_page.page_addr + reg); +} + +/** + * Reads a dword out of the status page, which is written to from the command + * queue by automatic updates, MI_REPORT_HEAD, MI_STORE_DATA_INDEX, or + * MI_STORE_DATA_IMM. + * + * The following dwords have a reserved meaning: + * 0x00: ISR copy, updated when an ISR bit not set in the HWSTAM changes. + * 0x04: ring 0 head pointer + * 0x05: ring 1 head pointer (915-class) + * 0x06: ring 2 head pointer (915-class) + * 0x10-0x1b: Context status DWords (GM45) + * 0x1f: Last written status offset. (GM45) + * + * The area from dword 0x20 to 0x3ff is available for driver usage. + */ +#define READ_HWSP(dev_priv, reg) intel_read_status_page(LP_RING(dev_priv), reg) +#define READ_BREADCRUMB(dev_priv) READ_HWSP(dev_priv, I915_BREADCRUMB_INDEX) +#define I915_GEM_HWS_INDEX 0x20 +#define I915_BREADCRUMB_INDEX 0x21 + +void intel_cleanup_ring_buffer(struct intel_ring_buffer *ring); + +int __must_check intel_wait_ring_buffer(struct intel_ring_buffer *ring, int n); +static inline int intel_wait_ring_idle(struct intel_ring_buffer *ring) +{ + return intel_wait_ring_buffer(ring, ring->size - 8); +} + +int __must_check intel_ring_begin(struct intel_ring_buffer *ring, int n); + +static inline void intel_ring_emit(struct intel_ring_buffer *ring, + u32 data) +{ + iowrite32(data, ring->virtual_start + ring->tail); + ring->tail += 4; +} + +void intel_ring_advance(struct intel_ring_buffer *ring); + +u32 intel_ring_get_seqno(struct intel_ring_buffer *ring); +int intel_ring_sync(struct intel_ring_buffer *ring, + struct intel_ring_buffer *to, + u32 seqno); + +int intel_init_render_ring_buffer(struct drm_device *dev); +int intel_init_bsd_ring_buffer(struct drm_device *dev); +int intel_init_blt_ring_buffer(struct drm_device *dev); + +u32 intel_ring_get_active_head(struct intel_ring_buffer *ring); +void intel_ring_setup_status_page(struct intel_ring_buffer *ring); + +static inline void i915_trace_irq_get(struct intel_ring_buffer *ring, u32 seqno) +{ + if (ring->trace_irq_seqno == 0 && ring->irq_get(ring)) + ring->trace_irq_seqno = seqno; +} + +/* DRI warts */ +int intel_render_ring_init_dri(struct drm_device *dev, u64 start, u32 size); + +#endif /* _INTEL_RINGBUFFER_H_ */ diff --git a/drivers/gpu/drm/i915/intel_sdvo.c b/drivers/gpu/drm/i915/intel_sdvo.c new file mode 100644 index 00000000..c9010607 --- /dev/null +++ b/drivers/gpu/drm/i915/intel_sdvo.c @@ -0,0 +1,2618 @@ +/* + * Copyright 2006 Dave Airlie + * Copyright © 2006-2007 Intel Corporation + * Jesse Barnes + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Eric Anholt + */ +#include +#include +#include +#include "drmP.h" +#include "drm.h" +#include "drm_crtc.h" +#include "drm_edid.h" +#include "intel_drv.h" +#include "i915_drm.h" +#include "i915_drv.h" +#include "intel_sdvo_regs.h" + +#define SDVO_TMDS_MASK (SDVO_OUTPUT_TMDS0 | SDVO_OUTPUT_TMDS1) +#define SDVO_RGB_MASK (SDVO_OUTPUT_RGB0 | SDVO_OUTPUT_RGB1) +#define SDVO_LVDS_MASK (SDVO_OUTPUT_LVDS0 | SDVO_OUTPUT_LVDS1) +#define SDVO_TV_MASK (SDVO_OUTPUT_CVBS0 | SDVO_OUTPUT_SVID0) + +#define SDVO_OUTPUT_MASK (SDVO_TMDS_MASK | SDVO_RGB_MASK | SDVO_LVDS_MASK |\ + SDVO_TV_MASK) + +#define IS_TV(c) (c->output_flag & SDVO_TV_MASK) +#define IS_TMDS(c) (c->output_flag & SDVO_TMDS_MASK) +#define IS_LVDS(c) (c->output_flag & SDVO_LVDS_MASK) +#define IS_TV_OR_LVDS(c) (c->output_flag & (SDVO_TV_MASK | SDVO_LVDS_MASK)) + + +static const char *tv_format_names[] = { + "NTSC_M" , "NTSC_J" , "NTSC_443", + "PAL_B" , "PAL_D" , "PAL_G" , + "PAL_H" , "PAL_I" , "PAL_M" , + "PAL_N" , "PAL_NC" , "PAL_60" , + "SECAM_B" , "SECAM_D" , "SECAM_G" , + "SECAM_K" , "SECAM_K1", "SECAM_L" , + "SECAM_60" +}; + +#define TV_FORMAT_NUM (sizeof(tv_format_names) / sizeof(*tv_format_names)) + +struct intel_sdvo { + struct intel_encoder base; + + struct i2c_adapter *i2c; + u8 slave_addr; + + struct i2c_adapter ddc; + + /* Register for the SDVO device: SDVOB or SDVOC */ + int sdvo_reg; + + /* Active outputs controlled by this SDVO output */ + uint16_t controlled_output; + + /* + * Capabilities of the SDVO device returned by + * i830_sdvo_get_capabilities() + */ + struct intel_sdvo_caps caps; + + /* Pixel clock limitations reported by the SDVO device, in kHz */ + int pixel_clock_min, pixel_clock_max; + + /* + * For multiple function SDVO device, + * this is for current attached outputs. + */ + uint16_t attached_output; + + /** + * This is used to select the color range of RBG outputs in HDMI mode. + * It is only valid when using TMDS encoding and 8 bit per color mode. + */ + uint32_t color_range; + + /** + * This is set if we're going to treat the device as TV-out. + * + * While we have these nice friendly flags for output types that ought + * to decide this for us, the S-Video output on our HDMI+S-Video card + * shows up as RGB1 (VGA). + */ + bool is_tv; + + /* This is for current tv format name */ + int tv_format_index; + + /** + * This is set if we treat the device as HDMI, instead of DVI. + */ + bool is_hdmi; + bool has_hdmi_monitor; + bool has_hdmi_audio; + + /** + * This is set if we detect output of sdvo device as LVDS and + * have a valid fixed mode to use with the panel. + */ + bool is_lvds; + + /** + * This is sdvo fixed pannel mode pointer + */ + struct drm_display_mode *sdvo_lvds_fixed_mode; + + /* DDC bus used by this SDVO encoder */ + uint8_t ddc_bus; + + /* Input timings for adjusted_mode */ + struct intel_sdvo_dtd input_dtd; +}; + +struct intel_sdvo_connector { + struct intel_connector base; + + /* Mark the type of connector */ + uint16_t output_flag; + + int force_audio; + + /* This contains all current supported TV format */ + u8 tv_format_supported[TV_FORMAT_NUM]; + int format_supported_num; + struct drm_property *tv_format; + + /* add the property for the SDVO-TV */ + struct drm_property *left; + struct drm_property *right; + struct drm_property *top; + struct drm_property *bottom; + struct drm_property *hpos; + struct drm_property *vpos; + struct drm_property *contrast; + struct drm_property *saturation; + struct drm_property *hue; + struct drm_property *sharpness; + struct drm_property *flicker_filter; + struct drm_property *flicker_filter_adaptive; + struct drm_property *flicker_filter_2d; + struct drm_property *tv_chroma_filter; + struct drm_property *tv_luma_filter; + struct drm_property *dot_crawl; + + /* add the property for the SDVO-TV/LVDS */ + struct drm_property *brightness; + + /* Add variable to record current setting for the above property */ + u32 left_margin, right_margin, top_margin, bottom_margin; + + /* this is to get the range of margin.*/ + u32 max_hscan, max_vscan; + u32 max_hpos, cur_hpos; + u32 max_vpos, cur_vpos; + u32 cur_brightness, max_brightness; + u32 cur_contrast, max_contrast; + u32 cur_saturation, max_saturation; + u32 cur_hue, max_hue; + u32 cur_sharpness, max_sharpness; + u32 cur_flicker_filter, max_flicker_filter; + u32 cur_flicker_filter_adaptive, max_flicker_filter_adaptive; + u32 cur_flicker_filter_2d, max_flicker_filter_2d; + u32 cur_tv_chroma_filter, max_tv_chroma_filter; + u32 cur_tv_luma_filter, max_tv_luma_filter; + u32 cur_dot_crawl, max_dot_crawl; +}; + +static struct intel_sdvo *to_intel_sdvo(struct drm_encoder *encoder) +{ + return container_of(encoder, struct intel_sdvo, base.base); +} + +static struct intel_sdvo *intel_attached_sdvo(struct drm_connector *connector) +{ + return container_of(intel_attached_encoder(connector), + struct intel_sdvo, base); +} + +static struct intel_sdvo_connector *to_intel_sdvo_connector(struct drm_connector *connector) +{ + return container_of(to_intel_connector(connector), struct intel_sdvo_connector, base); +} + +static bool +intel_sdvo_output_setup(struct intel_sdvo *intel_sdvo, uint16_t flags); +static bool +intel_sdvo_tv_create_property(struct intel_sdvo *intel_sdvo, + struct intel_sdvo_connector *intel_sdvo_connector, + int type); +static bool +intel_sdvo_create_enhance_property(struct intel_sdvo *intel_sdvo, + struct intel_sdvo_connector *intel_sdvo_connector); + +/** + * Writes the SDVOB or SDVOC with the given value, but always writes both + * SDVOB and SDVOC to work around apparent hardware issues (according to + * comments in the BIOS). + */ +static void intel_sdvo_write_sdvox(struct intel_sdvo *intel_sdvo, u32 val) +{ + struct drm_device *dev = intel_sdvo->base.base.dev; + struct drm_i915_private *dev_priv = dev->dev_private; + u32 bval = val, cval = val; + int i; + + if (intel_sdvo->sdvo_reg == PCH_SDVOB) { + I915_WRITE(intel_sdvo->sdvo_reg, val); + I915_READ(intel_sdvo->sdvo_reg); + return; + } + + if (intel_sdvo->sdvo_reg == SDVOB) { + cval = I915_READ(SDVOC); + } else { + bval = I915_READ(SDVOB); + } + /* + * Write the registers twice for luck. Sometimes, + * writing them only once doesn't appear to 'stick'. + * The BIOS does this too. Yay, magic + */ + for (i = 0; i < 2; i++) + { + I915_WRITE(SDVOB, bval); + I915_READ(SDVOB); + I915_WRITE(SDVOC, cval); + I915_READ(SDVOC); + } +} + +static bool intel_sdvo_read_byte(struct intel_sdvo *intel_sdvo, u8 addr, u8 *ch) +{ + struct i2c_msg msgs[] = { + { + .addr = intel_sdvo->slave_addr, + .flags = 0, + .len = 1, + .buf = &addr, + }, + { + .addr = intel_sdvo->slave_addr, + .flags = I2C_M_RD, + .len = 1, + .buf = ch, + } + }; + int ret; + + if ((ret = i2c_transfer(intel_sdvo->i2c, msgs, 2)) == 2) + return true; + + DRM_DEBUG_KMS("i2c transfer returned %d\n", ret); + return false; +} + +#define SDVO_CMD_NAME_ENTRY(cmd) {cmd, #cmd} +/** Mapping of command numbers to names, for debug output */ +static const struct _sdvo_cmd_name { + u8 cmd; + const char *name; +} sdvo_cmd_names[] = { + SDVO_CMD_NAME_ENTRY(SDVO_CMD_RESET), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_DEVICE_CAPS), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_FIRMWARE_REV), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_TRAINED_INPUTS), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_ACTIVE_OUTPUTS), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_ACTIVE_OUTPUTS), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_IN_OUT_MAP), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_IN_OUT_MAP), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_ATTACHED_DISPLAYS), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_HOT_PLUG_SUPPORT), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_ACTIVE_HOT_PLUG), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_ACTIVE_HOT_PLUG), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_INTERRUPT_EVENT_SOURCE), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_TARGET_INPUT), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_TARGET_OUTPUT), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_INPUT_TIMINGS_PART1), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_INPUT_TIMINGS_PART2), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_INPUT_TIMINGS_PART1), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_INPUT_TIMINGS_PART2), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_INPUT_TIMINGS_PART1), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_OUTPUT_TIMINGS_PART1), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_OUTPUT_TIMINGS_PART2), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_OUTPUT_TIMINGS_PART1), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_OUTPUT_TIMINGS_PART2), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_CREATE_PREFERRED_INPUT_TIMING), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_PREFERRED_INPUT_TIMING_PART1), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_PREFERRED_INPUT_TIMING_PART2), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_INPUT_PIXEL_CLOCK_RANGE), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_OUTPUT_PIXEL_CLOCK_RANGE), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_SUPPORTED_CLOCK_RATE_MULTS), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_CLOCK_RATE_MULT), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_CLOCK_RATE_MULT), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_SUPPORTED_TV_FORMATS), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_TV_FORMAT), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_TV_FORMAT), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_SUPPORTED_POWER_STATES), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_POWER_STATE), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_ENCODER_POWER_STATE), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_DISPLAY_POWER_STATE), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_CONTROL_BUS_SWITCH), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_SDTV_RESOLUTION_SUPPORT), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_SCALED_HDTV_RESOLUTION_SUPPORT), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_SUPPORTED_ENHANCEMENTS), + + /* Add the op code for SDVO enhancements */ + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_MAX_HPOS), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_HPOS), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_HPOS), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_MAX_VPOS), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_VPOS), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_VPOS), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_MAX_SATURATION), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_SATURATION), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_SATURATION), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_MAX_HUE), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_HUE), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_HUE), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_MAX_CONTRAST), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_CONTRAST), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_CONTRAST), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_MAX_BRIGHTNESS), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_BRIGHTNESS), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_BRIGHTNESS), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_MAX_OVERSCAN_H), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_OVERSCAN_H), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_OVERSCAN_H), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_MAX_OVERSCAN_V), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_OVERSCAN_V), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_OVERSCAN_V), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_MAX_FLICKER_FILTER), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_FLICKER_FILTER), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_FLICKER_FILTER), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_MAX_FLICKER_FILTER_ADAPTIVE), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_FLICKER_FILTER_ADAPTIVE), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_FLICKER_FILTER_ADAPTIVE), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_MAX_FLICKER_FILTER_2D), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_FLICKER_FILTER_2D), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_FLICKER_FILTER_2D), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_MAX_SHARPNESS), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_SHARPNESS), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_SHARPNESS), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_DOT_CRAWL), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_DOT_CRAWL), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_MAX_TV_CHROMA_FILTER), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_TV_CHROMA_FILTER), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_TV_CHROMA_FILTER), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_MAX_TV_LUMA_FILTER), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_TV_LUMA_FILTER), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_TV_LUMA_FILTER), + + /* HDMI op code */ + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_SUPP_ENCODE), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_ENCODE), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_ENCODE), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_PIXEL_REPLI), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_PIXEL_REPLI), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_COLORIMETRY_CAP), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_COLORIMETRY), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_COLORIMETRY), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_AUDIO_ENCRYPT_PREFER), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_AUDIO_STAT), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_AUDIO_STAT), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_HBUF_INDEX), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_HBUF_INDEX), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_HBUF_INFO), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_HBUF_AV_SPLIT), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_HBUF_AV_SPLIT), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_HBUF_TXRATE), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_HBUF_TXRATE), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_HBUF_DATA), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_HBUF_DATA), +}; + +#define IS_SDVOB(reg) (reg == SDVOB || reg == PCH_SDVOB) +#define SDVO_NAME(svdo) (IS_SDVOB((svdo)->sdvo_reg) ? "SDVOB" : "SDVOC") + +static void intel_sdvo_debug_write(struct intel_sdvo *intel_sdvo, u8 cmd, + const void *args, int args_len) +{ + int i; + + DRM_DEBUG_KMS("%s: W: %02X ", + SDVO_NAME(intel_sdvo), cmd); + for (i = 0; i < args_len; i++) + DRM_LOG_KMS("%02X ", ((u8 *)args)[i]); + for (; i < 8; i++) + DRM_LOG_KMS(" "); + for (i = 0; i < ARRAY_SIZE(sdvo_cmd_names); i++) { + if (cmd == sdvo_cmd_names[i].cmd) { + DRM_LOG_KMS("(%s)", sdvo_cmd_names[i].name); + break; + } + } + if (i == ARRAY_SIZE(sdvo_cmd_names)) + DRM_LOG_KMS("(%02X)", cmd); + DRM_LOG_KMS("\n"); +} + +static const char *cmd_status_names[] = { + "Power on", + "Success", + "Not supported", + "Invalid arg", + "Pending", + "Target not specified", + "Scaling not supported" +}; + +static bool intel_sdvo_write_cmd(struct intel_sdvo *intel_sdvo, u8 cmd, + const void *args, int args_len) +{ + u8 buf[args_len*2 + 2], status; + struct i2c_msg msgs[args_len + 3]; + int i, ret; + + intel_sdvo_debug_write(intel_sdvo, cmd, args, args_len); + + for (i = 0; i < args_len; i++) { + msgs[i].addr = intel_sdvo->slave_addr; + msgs[i].flags = 0; + msgs[i].len = 2; + msgs[i].buf = buf + 2 *i; + buf[2*i + 0] = SDVO_I2C_ARG_0 - i; + buf[2*i + 1] = ((u8*)args)[i]; + } + msgs[i].addr = intel_sdvo->slave_addr; + msgs[i].flags = 0; + msgs[i].len = 2; + msgs[i].buf = buf + 2*i; + buf[2*i + 0] = SDVO_I2C_OPCODE; + buf[2*i + 1] = cmd; + + /* the following two are to read the response */ + status = SDVO_I2C_CMD_STATUS; + msgs[i+1].addr = intel_sdvo->slave_addr; + msgs[i+1].flags = 0; + msgs[i+1].len = 1; + msgs[i+1].buf = &status; + + msgs[i+2].addr = intel_sdvo->slave_addr; + msgs[i+2].flags = I2C_M_RD; + msgs[i+2].len = 1; + msgs[i+2].buf = &status; + + ret = i2c_transfer(intel_sdvo->i2c, msgs, i+3); + if (ret < 0) { + DRM_DEBUG_KMS("I2c transfer returned %d\n", ret); + return false; + } + if (ret != i+3) { + /* failure in I2C transfer */ + DRM_DEBUG_KMS("I2c transfer returned %d/%d\n", ret, i+3); + return false; + } + + return true; +} + +static bool intel_sdvo_read_response(struct intel_sdvo *intel_sdvo, + void *response, int response_len) +{ + u8 retry = 5; + u8 status; + int i; + + DRM_DEBUG_KMS("%s: R: ", SDVO_NAME(intel_sdvo)); + + /* + * The documentation states that all commands will be + * processed within 15µs, and that we need only poll + * the status byte a maximum of 3 times in order for the + * command to be complete. + * + * Check 5 times in case the hardware failed to read the docs. + */ + if (!intel_sdvo_read_byte(intel_sdvo, + SDVO_I2C_CMD_STATUS, + &status)) + goto log_fail; + + while (status == SDVO_CMD_STATUS_PENDING && retry--) { + udelay(15); + if (!intel_sdvo_read_byte(intel_sdvo, + SDVO_I2C_CMD_STATUS, + &status)) + goto log_fail; + } + + if (status <= SDVO_CMD_STATUS_SCALING_NOT_SUPP) + DRM_LOG_KMS("(%s)", cmd_status_names[status]); + else + DRM_LOG_KMS("(??? %d)", status); + + if (status != SDVO_CMD_STATUS_SUCCESS) + goto log_fail; + + /* Read the command response */ + for (i = 0; i < response_len; i++) { + if (!intel_sdvo_read_byte(intel_sdvo, + SDVO_I2C_RETURN_0 + i, + &((u8 *)response)[i])) + goto log_fail; + DRM_LOG_KMS(" %02X", ((u8 *)response)[i]); + } + DRM_LOG_KMS("\n"); + return true; + +log_fail: + DRM_LOG_KMS("... failed\n"); + return false; +} + +static int intel_sdvo_get_pixel_multiplier(struct drm_display_mode *mode) +{ + if (mode->clock >= 100000) + return 1; + else if (mode->clock >= 50000) + return 2; + else + return 4; +} + +static bool intel_sdvo_set_control_bus_switch(struct intel_sdvo *intel_sdvo, + u8 ddc_bus) +{ + /* This must be the immediately preceding write before the i2c xfer */ + return intel_sdvo_write_cmd(intel_sdvo, + SDVO_CMD_SET_CONTROL_BUS_SWITCH, + &ddc_bus, 1); +} + +static bool intel_sdvo_set_value(struct intel_sdvo *intel_sdvo, u8 cmd, const void *data, int len) +{ + if (!intel_sdvo_write_cmd(intel_sdvo, cmd, data, len)) + return false; + + return intel_sdvo_read_response(intel_sdvo, NULL, 0); +} + +static bool +intel_sdvo_get_value(struct intel_sdvo *intel_sdvo, u8 cmd, void *value, int len) +{ + if (!intel_sdvo_write_cmd(intel_sdvo, cmd, NULL, 0)) + return false; + + return intel_sdvo_read_response(intel_sdvo, value, len); +} + +static bool intel_sdvo_set_target_input(struct intel_sdvo *intel_sdvo) +{ + struct intel_sdvo_set_target_input_args targets = {0}; + return intel_sdvo_set_value(intel_sdvo, + SDVO_CMD_SET_TARGET_INPUT, + &targets, sizeof(targets)); +} + +/** + * Return whether each input is trained. + * + * This function is making an assumption about the layout of the response, + * which should be checked against the docs. + */ +static bool intel_sdvo_get_trained_inputs(struct intel_sdvo *intel_sdvo, bool *input_1, bool *input_2) +{ + struct intel_sdvo_get_trained_inputs_response response; + + BUILD_BUG_ON(sizeof(response) != 1); + if (!intel_sdvo_get_value(intel_sdvo, SDVO_CMD_GET_TRAINED_INPUTS, + &response, sizeof(response))) + return false; + + *input_1 = response.input0_trained; + *input_2 = response.input1_trained; + return true; +} + +static bool intel_sdvo_set_active_outputs(struct intel_sdvo *intel_sdvo, + u16 outputs) +{ + return intel_sdvo_set_value(intel_sdvo, + SDVO_CMD_SET_ACTIVE_OUTPUTS, + &outputs, sizeof(outputs)); +} + +static bool intel_sdvo_set_encoder_power_state(struct intel_sdvo *intel_sdvo, + int mode) +{ + u8 state = SDVO_ENCODER_STATE_ON; + + switch (mode) { + case DRM_MODE_DPMS_ON: + state = SDVO_ENCODER_STATE_ON; + break; + case DRM_MODE_DPMS_STANDBY: + state = SDVO_ENCODER_STATE_STANDBY; + break; + case DRM_MODE_DPMS_SUSPEND: + state = SDVO_ENCODER_STATE_SUSPEND; + break; + case DRM_MODE_DPMS_OFF: + state = SDVO_ENCODER_STATE_OFF; + break; + } + + return intel_sdvo_set_value(intel_sdvo, + SDVO_CMD_SET_ENCODER_POWER_STATE, &state, sizeof(state)); +} + +static bool intel_sdvo_get_input_pixel_clock_range(struct intel_sdvo *intel_sdvo, + int *clock_min, + int *clock_max) +{ + struct intel_sdvo_pixel_clock_range clocks; + + BUILD_BUG_ON(sizeof(clocks) != 4); + if (!intel_sdvo_get_value(intel_sdvo, + SDVO_CMD_GET_INPUT_PIXEL_CLOCK_RANGE, + &clocks, sizeof(clocks))) + return false; + + /* Convert the values from units of 10 kHz to kHz. */ + *clock_min = clocks.min * 10; + *clock_max = clocks.max * 10; + return true; +} + +static bool intel_sdvo_set_target_output(struct intel_sdvo *intel_sdvo, + u16 outputs) +{ + return intel_sdvo_set_value(intel_sdvo, + SDVO_CMD_SET_TARGET_OUTPUT, + &outputs, sizeof(outputs)); +} + +static bool intel_sdvo_set_timing(struct intel_sdvo *intel_sdvo, u8 cmd, + struct intel_sdvo_dtd *dtd) +{ + return intel_sdvo_set_value(intel_sdvo, cmd, &dtd->part1, sizeof(dtd->part1)) && + intel_sdvo_set_value(intel_sdvo, cmd + 1, &dtd->part2, sizeof(dtd->part2)); +} + +static bool intel_sdvo_set_input_timing(struct intel_sdvo *intel_sdvo, + struct intel_sdvo_dtd *dtd) +{ + return intel_sdvo_set_timing(intel_sdvo, + SDVO_CMD_SET_INPUT_TIMINGS_PART1, dtd); +} + +static bool intel_sdvo_set_output_timing(struct intel_sdvo *intel_sdvo, + struct intel_sdvo_dtd *dtd) +{ + return intel_sdvo_set_timing(intel_sdvo, + SDVO_CMD_SET_OUTPUT_TIMINGS_PART1, dtd); +} + +static bool +intel_sdvo_create_preferred_input_timing(struct intel_sdvo *intel_sdvo, + uint16_t clock, + uint16_t width, + uint16_t height) +{ + struct intel_sdvo_preferred_input_timing_args args; + + memset(&args, 0, sizeof(args)); + args.clock = clock; + args.width = width; + args.height = height; + args.interlace = 0; + + if (intel_sdvo->is_lvds && + (intel_sdvo->sdvo_lvds_fixed_mode->hdisplay != width || + intel_sdvo->sdvo_lvds_fixed_mode->vdisplay != height)) + args.scaled = 1; + + return intel_sdvo_set_value(intel_sdvo, + SDVO_CMD_CREATE_PREFERRED_INPUT_TIMING, + &args, sizeof(args)); +} + +static bool intel_sdvo_get_preferred_input_timing(struct intel_sdvo *intel_sdvo, + struct intel_sdvo_dtd *dtd) +{ + BUILD_BUG_ON(sizeof(dtd->part1) != 8); + BUILD_BUG_ON(sizeof(dtd->part2) != 8); + return intel_sdvo_get_value(intel_sdvo, SDVO_CMD_GET_PREFERRED_INPUT_TIMING_PART1, + &dtd->part1, sizeof(dtd->part1)) && + intel_sdvo_get_value(intel_sdvo, SDVO_CMD_GET_PREFERRED_INPUT_TIMING_PART2, + &dtd->part2, sizeof(dtd->part2)); +} + +static bool intel_sdvo_set_clock_rate_mult(struct intel_sdvo *intel_sdvo, u8 val) +{ + return intel_sdvo_set_value(intel_sdvo, SDVO_CMD_SET_CLOCK_RATE_MULT, &val, 1); +} + +static void intel_sdvo_get_dtd_from_mode(struct intel_sdvo_dtd *dtd, + const struct drm_display_mode *mode) +{ + uint16_t width, height; + uint16_t h_blank_len, h_sync_len, v_blank_len, v_sync_len; + uint16_t h_sync_offset, v_sync_offset; + int mode_clock; + + width = mode->crtc_hdisplay; + height = mode->crtc_vdisplay; + + /* do some mode translations */ + h_blank_len = mode->crtc_hblank_end - mode->crtc_hblank_start; + h_sync_len = mode->crtc_hsync_end - mode->crtc_hsync_start; + + v_blank_len = mode->crtc_vblank_end - mode->crtc_vblank_start; + v_sync_len = mode->crtc_vsync_end - mode->crtc_vsync_start; + + h_sync_offset = mode->crtc_hsync_start - mode->crtc_hblank_start; + v_sync_offset = mode->crtc_vsync_start - mode->crtc_vblank_start; + + mode_clock = mode->clock; + mode_clock /= intel_mode_get_pixel_multiplier(mode) ?: 1; + mode_clock /= 10; + dtd->part1.clock = mode_clock; + + dtd->part1.h_active = width & 0xff; + dtd->part1.h_blank = h_blank_len & 0xff; + dtd->part1.h_high = (((width >> 8) & 0xf) << 4) | + ((h_blank_len >> 8) & 0xf); + dtd->part1.v_active = height & 0xff; + dtd->part1.v_blank = v_blank_len & 0xff; + dtd->part1.v_high = (((height >> 8) & 0xf) << 4) | + ((v_blank_len >> 8) & 0xf); + + dtd->part2.h_sync_off = h_sync_offset & 0xff; + dtd->part2.h_sync_width = h_sync_len & 0xff; + dtd->part2.v_sync_off_width = (v_sync_offset & 0xf) << 4 | + (v_sync_len & 0xf); + dtd->part2.sync_off_width_high = ((h_sync_offset & 0x300) >> 2) | + ((h_sync_len & 0x300) >> 4) | ((v_sync_offset & 0x30) >> 2) | + ((v_sync_len & 0x30) >> 4); + + dtd->part2.dtd_flags = 0x18; + if (mode->flags & DRM_MODE_FLAG_INTERLACE) + dtd->part2.dtd_flags |= DTD_FLAG_INTERLACE; + if (mode->flags & DRM_MODE_FLAG_PHSYNC) + dtd->part2.dtd_flags |= DTD_FLAG_HSYNC_POSITIVE; + if (mode->flags & DRM_MODE_FLAG_PVSYNC) + dtd->part2.dtd_flags |= DTD_FLAG_VSYNC_POSITIVE; + + dtd->part2.sdvo_flags = 0; + dtd->part2.v_sync_off_high = v_sync_offset & 0xc0; + dtd->part2.reserved = 0; +} + +static void intel_sdvo_get_mode_from_dtd(struct drm_display_mode * mode, + const struct intel_sdvo_dtd *dtd) +{ + mode->hdisplay = dtd->part1.h_active; + mode->hdisplay += ((dtd->part1.h_high >> 4) & 0x0f) << 8; + mode->hsync_start = mode->hdisplay + dtd->part2.h_sync_off; + mode->hsync_start += (dtd->part2.sync_off_width_high & 0xc0) << 2; + mode->hsync_end = mode->hsync_start + dtd->part2.h_sync_width; + mode->hsync_end += (dtd->part2.sync_off_width_high & 0x30) << 4; + mode->htotal = mode->hdisplay + dtd->part1.h_blank; + mode->htotal += (dtd->part1.h_high & 0xf) << 8; + + mode->vdisplay = dtd->part1.v_active; + mode->vdisplay += ((dtd->part1.v_high >> 4) & 0x0f) << 8; + mode->vsync_start = mode->vdisplay; + mode->vsync_start += (dtd->part2.v_sync_off_width >> 4) & 0xf; + mode->vsync_start += (dtd->part2.sync_off_width_high & 0x0c) << 2; + mode->vsync_start += dtd->part2.v_sync_off_high & 0xc0; + mode->vsync_end = mode->vsync_start + + (dtd->part2.v_sync_off_width & 0xf); + mode->vsync_end += (dtd->part2.sync_off_width_high & 0x3) << 4; + mode->vtotal = mode->vdisplay + dtd->part1.v_blank; + mode->vtotal += (dtd->part1.v_high & 0xf) << 8; + + mode->clock = dtd->part1.clock * 10; + + mode->flags &= ~(DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC); + if (dtd->part2.dtd_flags & DTD_FLAG_INTERLACE) + mode->flags |= DRM_MODE_FLAG_INTERLACE; + if (dtd->part2.dtd_flags & DTD_FLAG_HSYNC_POSITIVE) + mode->flags |= DRM_MODE_FLAG_PHSYNC; + if (dtd->part2.dtd_flags & DTD_FLAG_VSYNC_POSITIVE) + mode->flags |= DRM_MODE_FLAG_PVSYNC; +} + +static bool intel_sdvo_check_supp_encode(struct intel_sdvo *intel_sdvo) +{ + struct intel_sdvo_encode encode; + + BUILD_BUG_ON(sizeof(encode) != 2); + return intel_sdvo_get_value(intel_sdvo, + SDVO_CMD_GET_SUPP_ENCODE, + &encode, sizeof(encode)); +} + +static bool intel_sdvo_set_encode(struct intel_sdvo *intel_sdvo, + uint8_t mode) +{ + return intel_sdvo_set_value(intel_sdvo, SDVO_CMD_SET_ENCODE, &mode, 1); +} + +static bool intel_sdvo_set_colorimetry(struct intel_sdvo *intel_sdvo, + uint8_t mode) +{ + return intel_sdvo_set_value(intel_sdvo, SDVO_CMD_SET_COLORIMETRY, &mode, 1); +} + +#if 0 +static void intel_sdvo_dump_hdmi_buf(struct intel_sdvo *intel_sdvo) +{ + int i, j; + uint8_t set_buf_index[2]; + uint8_t av_split; + uint8_t buf_size; + uint8_t buf[48]; + uint8_t *pos; + + intel_sdvo_get_value(encoder, SDVO_CMD_GET_HBUF_AV_SPLIT, &av_split, 1); + + for (i = 0; i <= av_split; i++) { + set_buf_index[0] = i; set_buf_index[1] = 0; + intel_sdvo_write_cmd(encoder, SDVO_CMD_SET_HBUF_INDEX, + set_buf_index, 2); + intel_sdvo_write_cmd(encoder, SDVO_CMD_GET_HBUF_INFO, NULL, 0); + intel_sdvo_read_response(encoder, &buf_size, 1); + + pos = buf; + for (j = 0; j <= buf_size; j += 8) { + intel_sdvo_write_cmd(encoder, SDVO_CMD_GET_HBUF_DATA, + NULL, 0); + intel_sdvo_read_response(encoder, pos, 8); + pos += 8; + } + } +} +#endif + +static bool intel_sdvo_set_avi_infoframe(struct intel_sdvo *intel_sdvo) +{ + struct dip_infoframe avi_if = { + .type = DIP_TYPE_AVI, + .ver = DIP_VERSION_AVI, + .len = DIP_LEN_AVI, + }; + uint8_t tx_rate = SDVO_HBUF_TX_VSYNC; + uint8_t set_buf_index[2] = { 1, 0 }; + uint64_t *data = (uint64_t *)&avi_if; + unsigned i; + + intel_dip_infoframe_csum(&avi_if); + + if (!intel_sdvo_set_value(intel_sdvo, + SDVO_CMD_SET_HBUF_INDEX, + set_buf_index, 2)) + return false; + + for (i = 0; i < sizeof(avi_if); i += 8) { + if (!intel_sdvo_set_value(intel_sdvo, + SDVO_CMD_SET_HBUF_DATA, + data, 8)) + return false; + data++; + } + + return intel_sdvo_set_value(intel_sdvo, + SDVO_CMD_SET_HBUF_TXRATE, + &tx_rate, 1); +} + +static bool intel_sdvo_set_tv_format(struct intel_sdvo *intel_sdvo) +{ + struct intel_sdvo_tv_format format; + uint32_t format_map; + + format_map = 1 << intel_sdvo->tv_format_index; + memset(&format, 0, sizeof(format)); + memcpy(&format, &format_map, min(sizeof(format), sizeof(format_map))); + + BUILD_BUG_ON(sizeof(format) != 6); + return intel_sdvo_set_value(intel_sdvo, + SDVO_CMD_SET_TV_FORMAT, + &format, sizeof(format)); +} + +static bool +intel_sdvo_set_output_timings_from_mode(struct intel_sdvo *intel_sdvo, + struct drm_display_mode *mode) +{ + struct intel_sdvo_dtd output_dtd; + + if (!intel_sdvo_set_target_output(intel_sdvo, + intel_sdvo->attached_output)) + return false; + + intel_sdvo_get_dtd_from_mode(&output_dtd, mode); + if (!intel_sdvo_set_output_timing(intel_sdvo, &output_dtd)) + return false; + + return true; +} + +static bool +intel_sdvo_set_input_timings_for_mode(struct intel_sdvo *intel_sdvo, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + /* Reset the input timing to the screen. Assume always input 0. */ + if (!intel_sdvo_set_target_input(intel_sdvo)) + return false; + + if (!intel_sdvo_create_preferred_input_timing(intel_sdvo, + mode->clock / 10, + mode->hdisplay, + mode->vdisplay)) + return false; + + if (!intel_sdvo_get_preferred_input_timing(intel_sdvo, + &intel_sdvo->input_dtd)) + return false; + + intel_sdvo_get_mode_from_dtd(adjusted_mode, &intel_sdvo->input_dtd); + + drm_mode_set_crtcinfo(adjusted_mode, 0); + return true; +} + +static bool intel_sdvo_mode_fixup(struct drm_encoder *encoder, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct intel_sdvo *intel_sdvo = to_intel_sdvo(encoder); + int multiplier; + + /* We need to construct preferred input timings based on our + * output timings. To do that, we have to set the output + * timings, even though this isn't really the right place in + * the sequence to do it. Oh well. + */ + if (intel_sdvo->is_tv) { + if (!intel_sdvo_set_output_timings_from_mode(intel_sdvo, mode)) + return false; + + (void) intel_sdvo_set_input_timings_for_mode(intel_sdvo, + mode, + adjusted_mode); + } else if (intel_sdvo->is_lvds) { + if (!intel_sdvo_set_output_timings_from_mode(intel_sdvo, + intel_sdvo->sdvo_lvds_fixed_mode)) + return false; + + (void) intel_sdvo_set_input_timings_for_mode(intel_sdvo, + mode, + adjusted_mode); + } + + /* Make the CRTC code factor in the SDVO pixel multiplier. The + * SDVO device will factor out the multiplier during mode_set. + */ + multiplier = intel_sdvo_get_pixel_multiplier(adjusted_mode); + intel_mode_set_pixel_multiplier(adjusted_mode, multiplier); + + return true; +} + +static void intel_sdvo_mode_set(struct drm_encoder *encoder, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct drm_device *dev = encoder->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_crtc *crtc = encoder->crtc; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + struct intel_sdvo *intel_sdvo = to_intel_sdvo(encoder); + u32 sdvox; + struct intel_sdvo_in_out_map in_out; + struct intel_sdvo_dtd input_dtd, output_dtd; + int pixel_multiplier = intel_mode_get_pixel_multiplier(adjusted_mode); + int rate; + + if (!mode) + return; + + /* First, set the input mapping for the first input to our controlled + * output. This is only correct if we're a single-input device, in + * which case the first input is the output from the appropriate SDVO + * channel on the motherboard. In a two-input device, the first input + * will be SDVOB and the second SDVOC. + */ + in_out.in0 = intel_sdvo->attached_output; + in_out.in1 = 0; + + intel_sdvo_set_value(intel_sdvo, + SDVO_CMD_SET_IN_OUT_MAP, + &in_out, sizeof(in_out)); + + /* Set the output timings to the screen */ + if (!intel_sdvo_set_target_output(intel_sdvo, + intel_sdvo->attached_output)) + return; + + /* lvds has a special fixed output timing. */ + if (intel_sdvo->is_lvds) + intel_sdvo_get_dtd_from_mode(&output_dtd, + intel_sdvo->sdvo_lvds_fixed_mode); + else + intel_sdvo_get_dtd_from_mode(&output_dtd, mode); + (void) intel_sdvo_set_output_timing(intel_sdvo, &output_dtd); + + /* Set the input timing to the screen. Assume always input 0. */ + if (!intel_sdvo_set_target_input(intel_sdvo)) + return; + + if (intel_sdvo->has_hdmi_monitor) { + intel_sdvo_set_encode(intel_sdvo, SDVO_ENCODE_HDMI); + intel_sdvo_set_colorimetry(intel_sdvo, + SDVO_COLORIMETRY_RGB256); + intel_sdvo_set_avi_infoframe(intel_sdvo); + } else + intel_sdvo_set_encode(intel_sdvo, SDVO_ENCODE_DVI); + + if (intel_sdvo->is_tv && + !intel_sdvo_set_tv_format(intel_sdvo)) + return; + + /* We have tried to get input timing in mode_fixup, and filled into + * adjusted_mode. + */ + intel_sdvo_get_dtd_from_mode(&input_dtd, adjusted_mode); + (void) intel_sdvo_set_input_timing(intel_sdvo, &input_dtd); + + switch (pixel_multiplier) { + default: + case 1: rate = SDVO_CLOCK_RATE_MULT_1X; break; + case 2: rate = SDVO_CLOCK_RATE_MULT_2X; break; + case 4: rate = SDVO_CLOCK_RATE_MULT_4X; break; + } + if (!intel_sdvo_set_clock_rate_mult(intel_sdvo, rate)) + return; + + /* Set the SDVO control regs. */ + if (INTEL_INFO(dev)->gen >= 4) { + /* The real mode polarity is set by the SDVO commands, using + * struct intel_sdvo_dtd. */ + sdvox = SDVO_VSYNC_ACTIVE_HIGH | SDVO_HSYNC_ACTIVE_HIGH; + if (intel_sdvo->is_hdmi) + sdvox |= intel_sdvo->color_range; + if (INTEL_INFO(dev)->gen < 5) + sdvox |= SDVO_BORDER_ENABLE; + } else { + sdvox = I915_READ(intel_sdvo->sdvo_reg); + switch (intel_sdvo->sdvo_reg) { + case SDVOB: + sdvox &= SDVOB_PRESERVE_MASK; + break; + case SDVOC: + sdvox &= SDVOC_PRESERVE_MASK; + break; + } + sdvox |= (9 << 19) | SDVO_BORDER_ENABLE; + } + if (intel_crtc->pipe == 1) + sdvox |= SDVO_PIPE_B_SELECT; + if (intel_sdvo->has_hdmi_audio) + sdvox |= SDVO_AUDIO_ENABLE; + + if (INTEL_INFO(dev)->gen >= 4) { + /* done in crtc_mode_set as the dpll_md reg must be written early */ + } else if (IS_I945G(dev) || IS_I945GM(dev) || IS_G33(dev)) { + /* done in crtc_mode_set as it lives inside the dpll register */ + } else { + sdvox |= (pixel_multiplier - 1) << SDVO_PORT_MULTIPLY_SHIFT; + } + + if (input_dtd.part2.sdvo_flags & SDVO_NEED_TO_STALL && + INTEL_INFO(dev)->gen < 5) + sdvox |= SDVO_STALL_SELECT; + intel_sdvo_write_sdvox(intel_sdvo, sdvox); +} + +static void intel_sdvo_dpms(struct drm_encoder *encoder, int mode) +{ + struct drm_device *dev = encoder->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_sdvo *intel_sdvo = to_intel_sdvo(encoder); + struct intel_crtc *intel_crtc = to_intel_crtc(encoder->crtc); + u32 temp; + + if (mode != DRM_MODE_DPMS_ON) { + intel_sdvo_set_active_outputs(intel_sdvo, 0); + if (0) + intel_sdvo_set_encoder_power_state(intel_sdvo, mode); + + if (mode == DRM_MODE_DPMS_OFF) { + temp = I915_READ(intel_sdvo->sdvo_reg); + if ((temp & SDVO_ENABLE) != 0) { + intel_sdvo_write_sdvox(intel_sdvo, temp & ~SDVO_ENABLE); + } + } + } else { + bool input1, input2; + int i; + u8 status; + + temp = I915_READ(intel_sdvo->sdvo_reg); + if ((temp & SDVO_ENABLE) == 0) + intel_sdvo_write_sdvox(intel_sdvo, temp | SDVO_ENABLE); + for (i = 0; i < 2; i++) + intel_wait_for_vblank(dev, intel_crtc->pipe); + + status = intel_sdvo_get_trained_inputs(intel_sdvo, &input1, &input2); + /* Warn if the device reported failure to sync. + * A lot of SDVO devices fail to notify of sync, but it's + * a given it the status is a success, we succeeded. + */ + if (status == SDVO_CMD_STATUS_SUCCESS && !input1) { + DRM_DEBUG_KMS("First %s output reported failure to " + "sync\n", SDVO_NAME(intel_sdvo)); + } + + if (0) + intel_sdvo_set_encoder_power_state(intel_sdvo, mode); + intel_sdvo_set_active_outputs(intel_sdvo, intel_sdvo->attached_output); + } + return; +} + +static int intel_sdvo_mode_valid(struct drm_connector *connector, + struct drm_display_mode *mode) +{ + struct intel_sdvo *intel_sdvo = intel_attached_sdvo(connector); + + if (mode->flags & DRM_MODE_FLAG_DBLSCAN) + return MODE_NO_DBLESCAN; + + if (intel_sdvo->pixel_clock_min > mode->clock) + return MODE_CLOCK_LOW; + + if (intel_sdvo->pixel_clock_max < mode->clock) + return MODE_CLOCK_HIGH; + + if (intel_sdvo->is_lvds) { + if (mode->hdisplay > intel_sdvo->sdvo_lvds_fixed_mode->hdisplay) + return MODE_PANEL; + + if (mode->vdisplay > intel_sdvo->sdvo_lvds_fixed_mode->vdisplay) + return MODE_PANEL; + } + + return MODE_OK; +} + +static bool intel_sdvo_get_capabilities(struct intel_sdvo *intel_sdvo, struct intel_sdvo_caps *caps) +{ + BUILD_BUG_ON(sizeof(*caps) != 8); + if (!intel_sdvo_get_value(intel_sdvo, + SDVO_CMD_GET_DEVICE_CAPS, + caps, sizeof(*caps))) + return false; + + DRM_DEBUG_KMS("SDVO capabilities:\n" + " vendor_id: %d\n" + " device_id: %d\n" + " device_rev_id: %d\n" + " sdvo_version_major: %d\n" + " sdvo_version_minor: %d\n" + " sdvo_inputs_mask: %d\n" + " smooth_scaling: %d\n" + " sharp_scaling: %d\n" + " up_scaling: %d\n" + " down_scaling: %d\n" + " stall_support: %d\n" + " output_flags: %d\n", + caps->vendor_id, + caps->device_id, + caps->device_rev_id, + caps->sdvo_version_major, + caps->sdvo_version_minor, + caps->sdvo_inputs_mask, + caps->smooth_scaling, + caps->sharp_scaling, + caps->up_scaling, + caps->down_scaling, + caps->stall_support, + caps->output_flags); + + return true; +} + +/* No use! */ +#if 0 +struct drm_connector* intel_sdvo_find(struct drm_device *dev, int sdvoB) +{ + struct drm_connector *connector = NULL; + struct intel_sdvo *iout = NULL; + struct intel_sdvo *sdvo; + + /* find the sdvo connector */ + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + iout = to_intel_sdvo(connector); + + if (iout->type != INTEL_OUTPUT_SDVO) + continue; + + sdvo = iout->dev_priv; + + if (sdvo->sdvo_reg == SDVOB && sdvoB) + return connector; + + if (sdvo->sdvo_reg == SDVOC && !sdvoB) + return connector; + + } + + return NULL; +} + +int intel_sdvo_supports_hotplug(struct drm_connector *connector) +{ + u8 response[2]; + u8 status; + struct intel_sdvo *intel_sdvo; + DRM_DEBUG_KMS("\n"); + + if (!connector) + return 0; + + intel_sdvo = to_intel_sdvo(connector); + + return intel_sdvo_get_value(intel_sdvo, SDVO_CMD_GET_HOT_PLUG_SUPPORT, + &response, 2) && response[0]; +} + +void intel_sdvo_set_hotplug(struct drm_connector *connector, int on) +{ + u8 response[2]; + u8 status; + struct intel_sdvo *intel_sdvo = to_intel_sdvo(connector); + + intel_sdvo_write_cmd(intel_sdvo, SDVO_CMD_GET_ACTIVE_HOT_PLUG, NULL, 0); + intel_sdvo_read_response(intel_sdvo, &response, 2); + + if (on) { + intel_sdvo_write_cmd(intel_sdvo, SDVO_CMD_GET_HOT_PLUG_SUPPORT, NULL, 0); + status = intel_sdvo_read_response(intel_sdvo, &response, 2); + + intel_sdvo_write_cmd(intel_sdvo, SDVO_CMD_SET_ACTIVE_HOT_PLUG, &response, 2); + } else { + response[0] = 0; + response[1] = 0; + intel_sdvo_write_cmd(intel_sdvo, SDVO_CMD_SET_ACTIVE_HOT_PLUG, &response, 2); + } + + intel_sdvo_write_cmd(intel_sdvo, SDVO_CMD_GET_ACTIVE_HOT_PLUG, NULL, 0); + intel_sdvo_read_response(intel_sdvo, &response, 2); +} +#endif + +static bool +intel_sdvo_multifunc_encoder(struct intel_sdvo *intel_sdvo) +{ + /* Is there more than one type of output? */ + int caps = intel_sdvo->caps.output_flags & 0xf; + return caps & -caps; +} + +static struct edid * +intel_sdvo_get_edid(struct drm_connector *connector) +{ + struct intel_sdvo *sdvo = intel_attached_sdvo(connector); + return drm_get_edid(connector, &sdvo->ddc); +} + +/* Mac mini hack -- use the same DDC as the analog connector */ +static struct edid * +intel_sdvo_get_analog_edid(struct drm_connector *connector) +{ + struct drm_i915_private *dev_priv = connector->dev->dev_private; + + return drm_get_edid(connector, + &dev_priv->gmbus[dev_priv->crt_ddc_pin].adapter); +} + +enum drm_connector_status +intel_sdvo_hdmi_sink_detect(struct drm_connector *connector) +{ + struct intel_sdvo *intel_sdvo = intel_attached_sdvo(connector); + enum drm_connector_status status; + struct edid *edid; + + edid = intel_sdvo_get_edid(connector); + + if (edid == NULL && intel_sdvo_multifunc_encoder(intel_sdvo)) { + u8 ddc, saved_ddc = intel_sdvo->ddc_bus; + + /* + * Don't use the 1 as the argument of DDC bus switch to get + * the EDID. It is used for SDVO SPD ROM. + */ + for (ddc = intel_sdvo->ddc_bus >> 1; ddc > 1; ddc >>= 1) { + intel_sdvo->ddc_bus = ddc; + edid = intel_sdvo_get_edid(connector); + if (edid) + break; + } + /* + * If we found the EDID on the other bus, + * assume that is the correct DDC bus. + */ + if (edid == NULL) + intel_sdvo->ddc_bus = saved_ddc; + } + + /* + * When there is no edid and no monitor is connected with VGA + * port, try to use the CRT ddc to read the EDID for DVI-connector. + */ + if (edid == NULL) + edid = intel_sdvo_get_analog_edid(connector); + + status = connector_status_unknown; + if (edid != NULL) { + /* DDC bus is shared, match EDID to connector type */ + if (edid->input & DRM_EDID_INPUT_DIGITAL) { + status = connector_status_connected; + if (intel_sdvo->is_hdmi) { + intel_sdvo->has_hdmi_monitor = drm_detect_hdmi_monitor(edid); + intel_sdvo->has_hdmi_audio = drm_detect_monitor_audio(edid); + } + } else + status = connector_status_disconnected; + connector->display_info.raw_edid = NULL; + kfree(edid); + } + + if (status == connector_status_connected) { + struct intel_sdvo_connector *intel_sdvo_connector = to_intel_sdvo_connector(connector); + if (intel_sdvo_connector->force_audio) + intel_sdvo->has_hdmi_audio = intel_sdvo_connector->force_audio > 0; + } + + return status; +} + +static enum drm_connector_status +intel_sdvo_detect(struct drm_connector *connector, bool force) +{ + uint16_t response; + struct intel_sdvo *intel_sdvo = intel_attached_sdvo(connector); + struct intel_sdvo_connector *intel_sdvo_connector = to_intel_sdvo_connector(connector); + enum drm_connector_status ret; + + if (!intel_sdvo_write_cmd(intel_sdvo, + SDVO_CMD_GET_ATTACHED_DISPLAYS, NULL, 0)) + return connector_status_unknown; + + /* add 30ms delay when the output type might be TV */ + if (intel_sdvo->caps.output_flags & + (SDVO_OUTPUT_SVID0 | SDVO_OUTPUT_CVBS0)) + mdelay(30); + + if (!intel_sdvo_read_response(intel_sdvo, &response, 2)) + return connector_status_unknown; + + DRM_DEBUG_KMS("SDVO response %d %d [%x]\n", + response & 0xff, response >> 8, + intel_sdvo_connector->output_flag); + + if (response == 0) + return connector_status_disconnected; + + intel_sdvo->attached_output = response; + + intel_sdvo->has_hdmi_monitor = false; + intel_sdvo->has_hdmi_audio = false; + + if ((intel_sdvo_connector->output_flag & response) == 0) + ret = connector_status_disconnected; + else if (IS_TMDS(intel_sdvo_connector)) + ret = intel_sdvo_hdmi_sink_detect(connector); + else { + struct edid *edid; + + /* if we have an edid check it matches the connection */ + edid = intel_sdvo_get_edid(connector); + if (edid == NULL) + edid = intel_sdvo_get_analog_edid(connector); + if (edid != NULL) { + if (edid->input & DRM_EDID_INPUT_DIGITAL) + ret = connector_status_disconnected; + else + ret = connector_status_connected; + connector->display_info.raw_edid = NULL; + kfree(edid); + } else + ret = connector_status_connected; + } + + /* May update encoder flag for like clock for SDVO TV, etc.*/ + if (ret == connector_status_connected) { + intel_sdvo->is_tv = false; + intel_sdvo->is_lvds = false; + intel_sdvo->base.needs_tv_clock = false; + + if (response & SDVO_TV_MASK) { + intel_sdvo->is_tv = true; + intel_sdvo->base.needs_tv_clock = true; + } + if (response & SDVO_LVDS_MASK) + intel_sdvo->is_lvds = intel_sdvo->sdvo_lvds_fixed_mode != NULL; + } + + return ret; +} + +static void intel_sdvo_get_ddc_modes(struct drm_connector *connector) +{ + struct edid *edid; + + /* set the bus switch and get the modes */ + edid = intel_sdvo_get_edid(connector); + + /* + * Mac mini hack. On this device, the DVI-I connector shares one DDC + * link between analog and digital outputs. So, if the regular SDVO + * DDC fails, check to see if the analog output is disconnected, in + * which case we'll look there for the digital DDC data. + */ + if (edid == NULL) + edid = intel_sdvo_get_analog_edid(connector); + + if (edid != NULL) { + struct intel_sdvo_connector *intel_sdvo_connector = to_intel_sdvo_connector(connector); + bool monitor_is_digital = !!(edid->input & DRM_EDID_INPUT_DIGITAL); + bool connector_is_digital = !!IS_TMDS(intel_sdvo_connector); + + if (connector_is_digital == monitor_is_digital) { + drm_mode_connector_update_edid_property(connector, edid); + drm_add_edid_modes(connector, edid); + } + + connector->display_info.raw_edid = NULL; + kfree(edid); + } +} + +/* + * Set of SDVO TV modes. + * Note! This is in reply order (see loop in get_tv_modes). + * XXX: all 60Hz refresh? + */ +static const struct drm_display_mode sdvo_tv_modes[] = { + { DRM_MODE("320x200", DRM_MODE_TYPE_DRIVER, 5815, 320, 321, 384, + 416, 0, 200, 201, 232, 233, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + { DRM_MODE("320x240", DRM_MODE_TYPE_DRIVER, 6814, 320, 321, 384, + 416, 0, 240, 241, 272, 273, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + { DRM_MODE("400x300", DRM_MODE_TYPE_DRIVER, 9910, 400, 401, 464, + 496, 0, 300, 301, 332, 333, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + { DRM_MODE("640x350", DRM_MODE_TYPE_DRIVER, 16913, 640, 641, 704, + 736, 0, 350, 351, 382, 383, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + { DRM_MODE("640x400", DRM_MODE_TYPE_DRIVER, 19121, 640, 641, 704, + 736, 0, 400, 401, 432, 433, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + { DRM_MODE("640x480", DRM_MODE_TYPE_DRIVER, 22654, 640, 641, 704, + 736, 0, 480, 481, 512, 513, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + { DRM_MODE("704x480", DRM_MODE_TYPE_DRIVER, 24624, 704, 705, 768, + 800, 0, 480, 481, 512, 513, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + { DRM_MODE("704x576", DRM_MODE_TYPE_DRIVER, 29232, 704, 705, 768, + 800, 0, 576, 577, 608, 609, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + { DRM_MODE("720x350", DRM_MODE_TYPE_DRIVER, 18751, 720, 721, 784, + 816, 0, 350, 351, 382, 383, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + { DRM_MODE("720x400", DRM_MODE_TYPE_DRIVER, 21199, 720, 721, 784, + 816, 0, 400, 401, 432, 433, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + { DRM_MODE("720x480", DRM_MODE_TYPE_DRIVER, 25116, 720, 721, 784, + 816, 0, 480, 481, 512, 513, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + { DRM_MODE("720x540", DRM_MODE_TYPE_DRIVER, 28054, 720, 721, 784, + 816, 0, 540, 541, 572, 573, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + { DRM_MODE("720x576", DRM_MODE_TYPE_DRIVER, 29816, 720, 721, 784, + 816, 0, 576, 577, 608, 609, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + { DRM_MODE("768x576", DRM_MODE_TYPE_DRIVER, 31570, 768, 769, 832, + 864, 0, 576, 577, 608, 609, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + { DRM_MODE("800x600", DRM_MODE_TYPE_DRIVER, 34030, 800, 801, 864, + 896, 0, 600, 601, 632, 633, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + { DRM_MODE("832x624", DRM_MODE_TYPE_DRIVER, 36581, 832, 833, 896, + 928, 0, 624, 625, 656, 657, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + { DRM_MODE("920x766", DRM_MODE_TYPE_DRIVER, 48707, 920, 921, 984, + 1016, 0, 766, 767, 798, 799, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + { DRM_MODE("1024x768", DRM_MODE_TYPE_DRIVER, 53827, 1024, 1025, 1088, + 1120, 0, 768, 769, 800, 801, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + { DRM_MODE("1280x1024", DRM_MODE_TYPE_DRIVER, 87265, 1280, 1281, 1344, + 1376, 0, 1024, 1025, 1056, 1057, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, +}; + +static void intel_sdvo_get_tv_modes(struct drm_connector *connector) +{ + struct intel_sdvo *intel_sdvo = intel_attached_sdvo(connector); + struct intel_sdvo_sdtv_resolution_request tv_res; + uint32_t reply = 0, format_map = 0; + int i; + + /* Read the list of supported input resolutions for the selected TV + * format. + */ + format_map = 1 << intel_sdvo->tv_format_index; + memcpy(&tv_res, &format_map, + min(sizeof(format_map), sizeof(struct intel_sdvo_sdtv_resolution_request))); + + if (!intel_sdvo_set_target_output(intel_sdvo, intel_sdvo->attached_output)) + return; + + BUILD_BUG_ON(sizeof(tv_res) != 3); + if (!intel_sdvo_write_cmd(intel_sdvo, + SDVO_CMD_GET_SDTV_RESOLUTION_SUPPORT, + &tv_res, sizeof(tv_res))) + return; + if (!intel_sdvo_read_response(intel_sdvo, &reply, 3)) + return; + + for (i = 0; i < ARRAY_SIZE(sdvo_tv_modes); i++) + if (reply & (1 << i)) { + struct drm_display_mode *nmode; + nmode = drm_mode_duplicate(connector->dev, + &sdvo_tv_modes[i]); + if (nmode) + drm_mode_probed_add(connector, nmode); + } +} + +static void intel_sdvo_get_lvds_modes(struct drm_connector *connector) +{ + struct intel_sdvo *intel_sdvo = intel_attached_sdvo(connector); + struct drm_i915_private *dev_priv = connector->dev->dev_private; + struct drm_display_mode *newmode; + + /* + * Attempt to get the mode list from DDC. + * Assume that the preferred modes are + * arranged in priority order. + */ + intel_ddc_get_modes(connector, intel_sdvo->i2c); + if (list_empty(&connector->probed_modes) == false) + goto end; + + /* Fetch modes from VBT */ + if (dev_priv->sdvo_lvds_vbt_mode != NULL) { + newmode = drm_mode_duplicate(connector->dev, + dev_priv->sdvo_lvds_vbt_mode); + if (newmode != NULL) { + /* Guarantee the mode is preferred */ + newmode->type = (DRM_MODE_TYPE_PREFERRED | + DRM_MODE_TYPE_DRIVER); + drm_mode_probed_add(connector, newmode); + } + } + +end: + list_for_each_entry(newmode, &connector->probed_modes, head) { + if (newmode->type & DRM_MODE_TYPE_PREFERRED) { + intel_sdvo->sdvo_lvds_fixed_mode = + drm_mode_duplicate(connector->dev, newmode); + + drm_mode_set_crtcinfo(intel_sdvo->sdvo_lvds_fixed_mode, + 0); + + intel_sdvo->is_lvds = true; + break; + } + } + +} + +static int intel_sdvo_get_modes(struct drm_connector *connector) +{ + struct intel_sdvo_connector *intel_sdvo_connector = to_intel_sdvo_connector(connector); + + if (IS_TV(intel_sdvo_connector)) + intel_sdvo_get_tv_modes(connector); + else if (IS_LVDS(intel_sdvo_connector)) + intel_sdvo_get_lvds_modes(connector); + else + intel_sdvo_get_ddc_modes(connector); + + return !list_empty(&connector->probed_modes); +} + +static void +intel_sdvo_destroy_enhance_property(struct drm_connector *connector) +{ + struct intel_sdvo_connector *intel_sdvo_connector = to_intel_sdvo_connector(connector); + struct drm_device *dev = connector->dev; + + if (intel_sdvo_connector->left) + drm_property_destroy(dev, intel_sdvo_connector->left); + if (intel_sdvo_connector->right) + drm_property_destroy(dev, intel_sdvo_connector->right); + if (intel_sdvo_connector->top) + drm_property_destroy(dev, intel_sdvo_connector->top); + if (intel_sdvo_connector->bottom) + drm_property_destroy(dev, intel_sdvo_connector->bottom); + if (intel_sdvo_connector->hpos) + drm_property_destroy(dev, intel_sdvo_connector->hpos); + if (intel_sdvo_connector->vpos) + drm_property_destroy(dev, intel_sdvo_connector->vpos); + if (intel_sdvo_connector->saturation) + drm_property_destroy(dev, intel_sdvo_connector->saturation); + if (intel_sdvo_connector->contrast) + drm_property_destroy(dev, intel_sdvo_connector->contrast); + if (intel_sdvo_connector->hue) + drm_property_destroy(dev, intel_sdvo_connector->hue); + if (intel_sdvo_connector->sharpness) + drm_property_destroy(dev, intel_sdvo_connector->sharpness); + if (intel_sdvo_connector->flicker_filter) + drm_property_destroy(dev, intel_sdvo_connector->flicker_filter); + if (intel_sdvo_connector->flicker_filter_2d) + drm_property_destroy(dev, intel_sdvo_connector->flicker_filter_2d); + if (intel_sdvo_connector->flicker_filter_adaptive) + drm_property_destroy(dev, intel_sdvo_connector->flicker_filter_adaptive); + if (intel_sdvo_connector->tv_luma_filter) + drm_property_destroy(dev, intel_sdvo_connector->tv_luma_filter); + if (intel_sdvo_connector->tv_chroma_filter) + drm_property_destroy(dev, intel_sdvo_connector->tv_chroma_filter); + if (intel_sdvo_connector->dot_crawl) + drm_property_destroy(dev, intel_sdvo_connector->dot_crawl); + if (intel_sdvo_connector->brightness) + drm_property_destroy(dev, intel_sdvo_connector->brightness); +} + +static void intel_sdvo_destroy(struct drm_connector *connector) +{ + struct intel_sdvo_connector *intel_sdvo_connector = to_intel_sdvo_connector(connector); + + if (intel_sdvo_connector->tv_format) + drm_property_destroy(connector->dev, + intel_sdvo_connector->tv_format); + + intel_sdvo_destroy_enhance_property(connector); + drm_sysfs_connector_remove(connector); + drm_connector_cleanup(connector); + kfree(connector); +} + +static bool intel_sdvo_detect_hdmi_audio(struct drm_connector *connector) +{ + struct intel_sdvo *intel_sdvo = intel_attached_sdvo(connector); + struct edid *edid; + bool has_audio = false; + + if (!intel_sdvo->is_hdmi) + return false; + + edid = intel_sdvo_get_edid(connector); + if (edid != NULL && edid->input & DRM_EDID_INPUT_DIGITAL) + has_audio = drm_detect_monitor_audio(edid); + + return has_audio; +} + +static int +intel_sdvo_set_property(struct drm_connector *connector, + struct drm_property *property, + uint64_t val) +{ + struct intel_sdvo *intel_sdvo = intel_attached_sdvo(connector); + struct intel_sdvo_connector *intel_sdvo_connector = to_intel_sdvo_connector(connector); + struct drm_i915_private *dev_priv = connector->dev->dev_private; + uint16_t temp_value; + uint8_t cmd; + int ret; + + ret = drm_connector_property_set_value(connector, property, val); + if (ret) + return ret; + + if (property == dev_priv->force_audio_property) { + int i = val; + bool has_audio; + + if (i == intel_sdvo_connector->force_audio) + return 0; + + intel_sdvo_connector->force_audio = i; + + if (i == 0) + has_audio = intel_sdvo_detect_hdmi_audio(connector); + else + has_audio = i > 0; + + if (has_audio == intel_sdvo->has_hdmi_audio) + return 0; + + intel_sdvo->has_hdmi_audio = has_audio; + goto done; + } + + if (property == dev_priv->broadcast_rgb_property) { + if (val == !!intel_sdvo->color_range) + return 0; + + intel_sdvo->color_range = val ? SDVO_COLOR_RANGE_16_235 : 0; + goto done; + } + +#define CHECK_PROPERTY(name, NAME) \ + if (intel_sdvo_connector->name == property) { \ + if (intel_sdvo_connector->cur_##name == temp_value) return 0; \ + if (intel_sdvo_connector->max_##name < temp_value) return -EINVAL; \ + cmd = SDVO_CMD_SET_##NAME; \ + intel_sdvo_connector->cur_##name = temp_value; \ + goto set_value; \ + } + + if (property == intel_sdvo_connector->tv_format) { + if (val >= TV_FORMAT_NUM) + return -EINVAL; + + if (intel_sdvo->tv_format_index == + intel_sdvo_connector->tv_format_supported[val]) + return 0; + + intel_sdvo->tv_format_index = intel_sdvo_connector->tv_format_supported[val]; + goto done; + } else if (IS_TV_OR_LVDS(intel_sdvo_connector)) { + temp_value = val; + if (intel_sdvo_connector->left == property) { + drm_connector_property_set_value(connector, + intel_sdvo_connector->right, val); + if (intel_sdvo_connector->left_margin == temp_value) + return 0; + + intel_sdvo_connector->left_margin = temp_value; + intel_sdvo_connector->right_margin = temp_value; + temp_value = intel_sdvo_connector->max_hscan - + intel_sdvo_connector->left_margin; + cmd = SDVO_CMD_SET_OVERSCAN_H; + goto set_value; + } else if (intel_sdvo_connector->right == property) { + drm_connector_property_set_value(connector, + intel_sdvo_connector->left, val); + if (intel_sdvo_connector->right_margin == temp_value) + return 0; + + intel_sdvo_connector->left_margin = temp_value; + intel_sdvo_connector->right_margin = temp_value; + temp_value = intel_sdvo_connector->max_hscan - + intel_sdvo_connector->left_margin; + cmd = SDVO_CMD_SET_OVERSCAN_H; + goto set_value; + } else if (intel_sdvo_connector->top == property) { + drm_connector_property_set_value(connector, + intel_sdvo_connector->bottom, val); + if (intel_sdvo_connector->top_margin == temp_value) + return 0; + + intel_sdvo_connector->top_margin = temp_value; + intel_sdvo_connector->bottom_margin = temp_value; + temp_value = intel_sdvo_connector->max_vscan - + intel_sdvo_connector->top_margin; + cmd = SDVO_CMD_SET_OVERSCAN_V; + goto set_value; + } else if (intel_sdvo_connector->bottom == property) { + drm_connector_property_set_value(connector, + intel_sdvo_connector->top, val); + if (intel_sdvo_connector->bottom_margin == temp_value) + return 0; + + intel_sdvo_connector->top_margin = temp_value; + intel_sdvo_connector->bottom_margin = temp_value; + temp_value = intel_sdvo_connector->max_vscan - + intel_sdvo_connector->top_margin; + cmd = SDVO_CMD_SET_OVERSCAN_V; + goto set_value; + } + CHECK_PROPERTY(hpos, HPOS) + CHECK_PROPERTY(vpos, VPOS) + CHECK_PROPERTY(saturation, SATURATION) + CHECK_PROPERTY(contrast, CONTRAST) + CHECK_PROPERTY(hue, HUE) + CHECK_PROPERTY(brightness, BRIGHTNESS) + CHECK_PROPERTY(sharpness, SHARPNESS) + CHECK_PROPERTY(flicker_filter, FLICKER_FILTER) + CHECK_PROPERTY(flicker_filter_2d, FLICKER_FILTER_2D) + CHECK_PROPERTY(flicker_filter_adaptive, FLICKER_FILTER_ADAPTIVE) + CHECK_PROPERTY(tv_chroma_filter, TV_CHROMA_FILTER) + CHECK_PROPERTY(tv_luma_filter, TV_LUMA_FILTER) + CHECK_PROPERTY(dot_crawl, DOT_CRAWL) + } + + return -EINVAL; /* unknown property */ + +set_value: + if (!intel_sdvo_set_value(intel_sdvo, cmd, &temp_value, 2)) + return -EIO; + + +done: + if (intel_sdvo->base.base.crtc) { + struct drm_crtc *crtc = intel_sdvo->base.base.crtc; + drm_crtc_helper_set_mode(crtc, &crtc->mode, crtc->x, + crtc->y, crtc->fb); + } + + return 0; +#undef CHECK_PROPERTY +} + +static const struct drm_encoder_helper_funcs intel_sdvo_helper_funcs = { + .dpms = intel_sdvo_dpms, + .mode_fixup = intel_sdvo_mode_fixup, + .prepare = intel_encoder_prepare, + .mode_set = intel_sdvo_mode_set, + .commit = intel_encoder_commit, +}; + +static const struct drm_connector_funcs intel_sdvo_connector_funcs = { + .dpms = drm_helper_connector_dpms, + .detect = intel_sdvo_detect, + .fill_modes = drm_helper_probe_single_connector_modes, + .set_property = intel_sdvo_set_property, + .destroy = intel_sdvo_destroy, +}; + +static const struct drm_connector_helper_funcs intel_sdvo_connector_helper_funcs = { + .get_modes = intel_sdvo_get_modes, + .mode_valid = intel_sdvo_mode_valid, + .best_encoder = intel_best_encoder, +}; + +static void intel_sdvo_enc_destroy(struct drm_encoder *encoder) +{ + struct intel_sdvo *intel_sdvo = to_intel_sdvo(encoder); + + if (intel_sdvo->sdvo_lvds_fixed_mode != NULL) + drm_mode_destroy(encoder->dev, + intel_sdvo->sdvo_lvds_fixed_mode); + + i2c_del_adapter(&intel_sdvo->ddc); + intel_encoder_destroy(encoder); +} + +static const struct drm_encoder_funcs intel_sdvo_enc_funcs = { + .destroy = intel_sdvo_enc_destroy, +}; + +static void +intel_sdvo_guess_ddc_bus(struct intel_sdvo *sdvo) +{ + uint16_t mask = 0; + unsigned int num_bits; + + /* Make a mask of outputs less than or equal to our own priority in the + * list. + */ + switch (sdvo->controlled_output) { + case SDVO_OUTPUT_LVDS1: + mask |= SDVO_OUTPUT_LVDS1; + case SDVO_OUTPUT_LVDS0: + mask |= SDVO_OUTPUT_LVDS0; + case SDVO_OUTPUT_TMDS1: + mask |= SDVO_OUTPUT_TMDS1; + case SDVO_OUTPUT_TMDS0: + mask |= SDVO_OUTPUT_TMDS0; + case SDVO_OUTPUT_RGB1: + mask |= SDVO_OUTPUT_RGB1; + case SDVO_OUTPUT_RGB0: + mask |= SDVO_OUTPUT_RGB0; + break; + } + + /* Count bits to find what number we are in the priority list. */ + mask &= sdvo->caps.output_flags; + num_bits = hweight16(mask); + /* If more than 3 outputs, default to DDC bus 3 for now. */ + if (num_bits > 3) + num_bits = 3; + + /* Corresponds to SDVO_CONTROL_BUS_DDCx */ + sdvo->ddc_bus = 1 << num_bits; +} + +/** + * Choose the appropriate DDC bus for control bus switch command for this + * SDVO output based on the controlled output. + * + * DDC bus number assignment is in a priority order of RGB outputs, then TMDS + * outputs, then LVDS outputs. + */ +static void +intel_sdvo_select_ddc_bus(struct drm_i915_private *dev_priv, + struct intel_sdvo *sdvo, u32 reg) +{ + struct sdvo_device_mapping *mapping; + + if (IS_SDVOB(reg)) + mapping = &(dev_priv->sdvo_mappings[0]); + else + mapping = &(dev_priv->sdvo_mappings[1]); + + if (mapping->initialized) + sdvo->ddc_bus = 1 << ((mapping->ddc_pin & 0xf0) >> 4); + else + intel_sdvo_guess_ddc_bus(sdvo); +} + +static void +intel_sdvo_select_i2c_bus(struct drm_i915_private *dev_priv, + struct intel_sdvo *sdvo, u32 reg) +{ + struct sdvo_device_mapping *mapping; + u8 pin, speed; + + if (IS_SDVOB(reg)) + mapping = &dev_priv->sdvo_mappings[0]; + else + mapping = &dev_priv->sdvo_mappings[1]; + + pin = GMBUS_PORT_DPB; + speed = GMBUS_RATE_1MHZ >> 8; + if (mapping->initialized) { + pin = mapping->i2c_pin; + speed = mapping->i2c_speed; + } + + if (pin < GMBUS_NUM_PORTS) { + sdvo->i2c = &dev_priv->gmbus[pin].adapter; + intel_gmbus_set_speed(sdvo->i2c, speed); + intel_gmbus_force_bit(sdvo->i2c, true); + } else + sdvo->i2c = &dev_priv->gmbus[GMBUS_PORT_DPB].adapter; +} + +static bool +intel_sdvo_is_hdmi_connector(struct intel_sdvo *intel_sdvo, int device) +{ + return intel_sdvo_check_supp_encode(intel_sdvo); +} + +static u8 +intel_sdvo_get_slave_addr(struct drm_device *dev, int sdvo_reg) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct sdvo_device_mapping *my_mapping, *other_mapping; + + if (IS_SDVOB(sdvo_reg)) { + my_mapping = &dev_priv->sdvo_mappings[0]; + other_mapping = &dev_priv->sdvo_mappings[1]; + } else { + my_mapping = &dev_priv->sdvo_mappings[1]; + other_mapping = &dev_priv->sdvo_mappings[0]; + } + + /* If the BIOS described our SDVO device, take advantage of it. */ + if (my_mapping->slave_addr) + return my_mapping->slave_addr; + + /* If the BIOS only described a different SDVO device, use the + * address that it isn't using. + */ + if (other_mapping->slave_addr) { + if (other_mapping->slave_addr == 0x70) + return 0x72; + else + return 0x70; + } + + /* No SDVO device info is found for another DVO port, + * so use mapping assumption we had before BIOS parsing. + */ + if (IS_SDVOB(sdvo_reg)) + return 0x70; + else + return 0x72; +} + +static void +intel_sdvo_connector_init(struct intel_sdvo_connector *connector, + struct intel_sdvo *encoder) +{ + drm_connector_init(encoder->base.base.dev, + &connector->base.base, + &intel_sdvo_connector_funcs, + connector->base.base.connector_type); + + drm_connector_helper_add(&connector->base.base, + &intel_sdvo_connector_helper_funcs); + + connector->base.base.interlace_allowed = 0; + connector->base.base.doublescan_allowed = 0; + connector->base.base.display_info.subpixel_order = SubPixelHorizontalRGB; + + intel_connector_attach_encoder(&connector->base, &encoder->base); + drm_sysfs_connector_add(&connector->base.base); +} + +static void +intel_sdvo_add_hdmi_properties(struct intel_sdvo_connector *connector) +{ + struct drm_device *dev = connector->base.base.dev; + + intel_attach_force_audio_property(&connector->base.base); + if (INTEL_INFO(dev)->gen >= 4 && IS_MOBILE(dev)) + intel_attach_broadcast_rgb_property(&connector->base.base); +} + +static bool +intel_sdvo_dvi_init(struct intel_sdvo *intel_sdvo, int device) +{ + struct drm_encoder *encoder = &intel_sdvo->base.base; + struct drm_connector *connector; + struct intel_connector *intel_connector; + struct intel_sdvo_connector *intel_sdvo_connector; + + intel_sdvo_connector = kzalloc(sizeof(struct intel_sdvo_connector), GFP_KERNEL); + if (!intel_sdvo_connector) + return false; + + if (device == 0) { + intel_sdvo->controlled_output |= SDVO_OUTPUT_TMDS0; + intel_sdvo_connector->output_flag = SDVO_OUTPUT_TMDS0; + } else if (device == 1) { + intel_sdvo->controlled_output |= SDVO_OUTPUT_TMDS1; + intel_sdvo_connector->output_flag = SDVO_OUTPUT_TMDS1; + } + + intel_connector = &intel_sdvo_connector->base; + connector = &intel_connector->base; + connector->polled = DRM_CONNECTOR_POLL_CONNECT | DRM_CONNECTOR_POLL_DISCONNECT; + encoder->encoder_type = DRM_MODE_ENCODER_TMDS; + connector->connector_type = DRM_MODE_CONNECTOR_DVID; + + if (intel_sdvo_is_hdmi_connector(intel_sdvo, device)) { + connector->connector_type = DRM_MODE_CONNECTOR_HDMIA; + intel_sdvo->is_hdmi = true; + } + intel_sdvo->base.clone_mask = ((1 << INTEL_SDVO_NON_TV_CLONE_BIT) | + (1 << INTEL_ANALOG_CLONE_BIT)); + + intel_sdvo_connector_init(intel_sdvo_connector, intel_sdvo); + if (intel_sdvo->is_hdmi) + intel_sdvo_add_hdmi_properties(intel_sdvo_connector); + + return true; +} + +static bool +intel_sdvo_tv_init(struct intel_sdvo *intel_sdvo, int type) +{ + struct drm_encoder *encoder = &intel_sdvo->base.base; + struct drm_connector *connector; + struct intel_connector *intel_connector; + struct intel_sdvo_connector *intel_sdvo_connector; + + intel_sdvo_connector = kzalloc(sizeof(struct intel_sdvo_connector), GFP_KERNEL); + if (!intel_sdvo_connector) + return false; + + intel_connector = &intel_sdvo_connector->base; + connector = &intel_connector->base; + encoder->encoder_type = DRM_MODE_ENCODER_TVDAC; + connector->connector_type = DRM_MODE_CONNECTOR_SVIDEO; + + intel_sdvo->controlled_output |= type; + intel_sdvo_connector->output_flag = type; + + intel_sdvo->is_tv = true; + intel_sdvo->base.needs_tv_clock = true; + intel_sdvo->base.clone_mask = 1 << INTEL_SDVO_TV_CLONE_BIT; + + intel_sdvo_connector_init(intel_sdvo_connector, intel_sdvo); + + if (!intel_sdvo_tv_create_property(intel_sdvo, intel_sdvo_connector, type)) + goto err; + + if (!intel_sdvo_create_enhance_property(intel_sdvo, intel_sdvo_connector)) + goto err; + + return true; + +err: + intel_sdvo_destroy(connector); + return false; +} + +static bool +intel_sdvo_analog_init(struct intel_sdvo *intel_sdvo, int device) +{ + struct drm_encoder *encoder = &intel_sdvo->base.base; + struct drm_connector *connector; + struct intel_connector *intel_connector; + struct intel_sdvo_connector *intel_sdvo_connector; + + intel_sdvo_connector = kzalloc(sizeof(struct intel_sdvo_connector), GFP_KERNEL); + if (!intel_sdvo_connector) + return false; + + intel_connector = &intel_sdvo_connector->base; + connector = &intel_connector->base; + connector->polled = DRM_CONNECTOR_POLL_CONNECT; + encoder->encoder_type = DRM_MODE_ENCODER_DAC; + connector->connector_type = DRM_MODE_CONNECTOR_VGA; + + if (device == 0) { + intel_sdvo->controlled_output |= SDVO_OUTPUT_RGB0; + intel_sdvo_connector->output_flag = SDVO_OUTPUT_RGB0; + } else if (device == 1) { + intel_sdvo->controlled_output |= SDVO_OUTPUT_RGB1; + intel_sdvo_connector->output_flag = SDVO_OUTPUT_RGB1; + } + + intel_sdvo->base.clone_mask = ((1 << INTEL_SDVO_NON_TV_CLONE_BIT) | + (1 << INTEL_ANALOG_CLONE_BIT)); + + intel_sdvo_connector_init(intel_sdvo_connector, + intel_sdvo); + return true; +} + +static bool +intel_sdvo_lvds_init(struct intel_sdvo *intel_sdvo, int device) +{ + struct drm_encoder *encoder = &intel_sdvo->base.base; + struct drm_connector *connector; + struct intel_connector *intel_connector; + struct intel_sdvo_connector *intel_sdvo_connector; + + intel_sdvo_connector = kzalloc(sizeof(struct intel_sdvo_connector), GFP_KERNEL); + if (!intel_sdvo_connector) + return false; + + intel_connector = &intel_sdvo_connector->base; + connector = &intel_connector->base; + encoder->encoder_type = DRM_MODE_ENCODER_LVDS; + connector->connector_type = DRM_MODE_CONNECTOR_LVDS; + + if (device == 0) { + intel_sdvo->controlled_output |= SDVO_OUTPUT_LVDS0; + intel_sdvo_connector->output_flag = SDVO_OUTPUT_LVDS0; + } else if (device == 1) { + intel_sdvo->controlled_output |= SDVO_OUTPUT_LVDS1; + intel_sdvo_connector->output_flag = SDVO_OUTPUT_LVDS1; + } + + intel_sdvo->base.clone_mask = ((1 << INTEL_ANALOG_CLONE_BIT) | + (1 << INTEL_SDVO_LVDS_CLONE_BIT)); + + intel_sdvo_connector_init(intel_sdvo_connector, intel_sdvo); + if (!intel_sdvo_create_enhance_property(intel_sdvo, intel_sdvo_connector)) + goto err; + + return true; + +err: + intel_sdvo_destroy(connector); + return false; +} + +static bool +intel_sdvo_output_setup(struct intel_sdvo *intel_sdvo, uint16_t flags) +{ + intel_sdvo->is_tv = false; + intel_sdvo->base.needs_tv_clock = false; + intel_sdvo->is_lvds = false; + + /* SDVO requires XXX1 function may not exist unless it has XXX0 function.*/ + + if (flags & SDVO_OUTPUT_TMDS0) + if (!intel_sdvo_dvi_init(intel_sdvo, 0)) + return false; + + if ((flags & SDVO_TMDS_MASK) == SDVO_TMDS_MASK) + if (!intel_sdvo_dvi_init(intel_sdvo, 1)) + return false; + + /* TV has no XXX1 function block */ + if (flags & SDVO_OUTPUT_SVID0) + if (!intel_sdvo_tv_init(intel_sdvo, SDVO_OUTPUT_SVID0)) + return false; + + if (flags & SDVO_OUTPUT_CVBS0) + if (!intel_sdvo_tv_init(intel_sdvo, SDVO_OUTPUT_CVBS0)) + return false; + + if (flags & SDVO_OUTPUT_RGB0) + if (!intel_sdvo_analog_init(intel_sdvo, 0)) + return false; + + if ((flags & SDVO_RGB_MASK) == SDVO_RGB_MASK) + if (!intel_sdvo_analog_init(intel_sdvo, 1)) + return false; + + if (flags & SDVO_OUTPUT_LVDS0) + if (!intel_sdvo_lvds_init(intel_sdvo, 0)) + return false; + + if ((flags & SDVO_LVDS_MASK) == SDVO_LVDS_MASK) + if (!intel_sdvo_lvds_init(intel_sdvo, 1)) + return false; + + if ((flags & SDVO_OUTPUT_MASK) == 0) { + unsigned char bytes[2]; + + intel_sdvo->controlled_output = 0; + memcpy(bytes, &intel_sdvo->caps.output_flags, 2); + DRM_DEBUG_KMS("%s: Unknown SDVO output type (0x%02x%02x)\n", + SDVO_NAME(intel_sdvo), + bytes[0], bytes[1]); + return false; + } + intel_sdvo->base.crtc_mask = (1 << 0) | (1 << 1); + + return true; +} + +static bool intel_sdvo_tv_create_property(struct intel_sdvo *intel_sdvo, + struct intel_sdvo_connector *intel_sdvo_connector, + int type) +{ + struct drm_device *dev = intel_sdvo->base.base.dev; + struct intel_sdvo_tv_format format; + uint32_t format_map, i; + + if (!intel_sdvo_set_target_output(intel_sdvo, type)) + return false; + + BUILD_BUG_ON(sizeof(format) != 6); + if (!intel_sdvo_get_value(intel_sdvo, + SDVO_CMD_GET_SUPPORTED_TV_FORMATS, + &format, sizeof(format))) + return false; + + memcpy(&format_map, &format, min(sizeof(format_map), sizeof(format))); + + if (format_map == 0) + return false; + + intel_sdvo_connector->format_supported_num = 0; + for (i = 0 ; i < TV_FORMAT_NUM; i++) + if (format_map & (1 << i)) + intel_sdvo_connector->tv_format_supported[intel_sdvo_connector->format_supported_num++] = i; + + + intel_sdvo_connector->tv_format = + drm_property_create(dev, DRM_MODE_PROP_ENUM, + "mode", intel_sdvo_connector->format_supported_num); + if (!intel_sdvo_connector->tv_format) + return false; + + for (i = 0; i < intel_sdvo_connector->format_supported_num; i++) + drm_property_add_enum( + intel_sdvo_connector->tv_format, i, + i, tv_format_names[intel_sdvo_connector->tv_format_supported[i]]); + + intel_sdvo->tv_format_index = intel_sdvo_connector->tv_format_supported[0]; + drm_connector_attach_property(&intel_sdvo_connector->base.base, + intel_sdvo_connector->tv_format, 0); + return true; + +} + +#define ENHANCEMENT(name, NAME) do { \ + if (enhancements.name) { \ + if (!intel_sdvo_get_value(intel_sdvo, SDVO_CMD_GET_MAX_##NAME, &data_value, 4) || \ + !intel_sdvo_get_value(intel_sdvo, SDVO_CMD_GET_##NAME, &response, 2)) \ + return false; \ + intel_sdvo_connector->max_##name = data_value[0]; \ + intel_sdvo_connector->cur_##name = response; \ + intel_sdvo_connector->name = \ + drm_property_create(dev, DRM_MODE_PROP_RANGE, #name, 2); \ + if (!intel_sdvo_connector->name) return false; \ + intel_sdvo_connector->name->values[0] = 0; \ + intel_sdvo_connector->name->values[1] = data_value[0]; \ + drm_connector_attach_property(connector, \ + intel_sdvo_connector->name, \ + intel_sdvo_connector->cur_##name); \ + DRM_DEBUG_KMS(#name ": max %d, default %d, current %d\n", \ + data_value[0], data_value[1], response); \ + } \ +} while(0) + +static bool +intel_sdvo_create_enhance_property_tv(struct intel_sdvo *intel_sdvo, + struct intel_sdvo_connector *intel_sdvo_connector, + struct intel_sdvo_enhancements_reply enhancements) +{ + struct drm_device *dev = intel_sdvo->base.base.dev; + struct drm_connector *connector = &intel_sdvo_connector->base.base; + uint16_t response, data_value[2]; + + /* when horizontal overscan is supported, Add the left/right property */ + if (enhancements.overscan_h) { + if (!intel_sdvo_get_value(intel_sdvo, + SDVO_CMD_GET_MAX_OVERSCAN_H, + &data_value, 4)) + return false; + + if (!intel_sdvo_get_value(intel_sdvo, + SDVO_CMD_GET_OVERSCAN_H, + &response, 2)) + return false; + + intel_sdvo_connector->max_hscan = data_value[0]; + intel_sdvo_connector->left_margin = data_value[0] - response; + intel_sdvo_connector->right_margin = intel_sdvo_connector->left_margin; + intel_sdvo_connector->left = + drm_property_create(dev, DRM_MODE_PROP_RANGE, + "left_margin", 2); + if (!intel_sdvo_connector->left) + return false; + + intel_sdvo_connector->left->values[0] = 0; + intel_sdvo_connector->left->values[1] = data_value[0]; + drm_connector_attach_property(connector, + intel_sdvo_connector->left, + intel_sdvo_connector->left_margin); + + intel_sdvo_connector->right = + drm_property_create(dev, DRM_MODE_PROP_RANGE, + "right_margin", 2); + if (!intel_sdvo_connector->right) + return false; + + intel_sdvo_connector->right->values[0] = 0; + intel_sdvo_connector->right->values[1] = data_value[0]; + drm_connector_attach_property(connector, + intel_sdvo_connector->right, + intel_sdvo_connector->right_margin); + DRM_DEBUG_KMS("h_overscan: max %d, " + "default %d, current %d\n", + data_value[0], data_value[1], response); + } + + if (enhancements.overscan_v) { + if (!intel_sdvo_get_value(intel_sdvo, + SDVO_CMD_GET_MAX_OVERSCAN_V, + &data_value, 4)) + return false; + + if (!intel_sdvo_get_value(intel_sdvo, + SDVO_CMD_GET_OVERSCAN_V, + &response, 2)) + return false; + + intel_sdvo_connector->max_vscan = data_value[0]; + intel_sdvo_connector->top_margin = data_value[0] - response; + intel_sdvo_connector->bottom_margin = intel_sdvo_connector->top_margin; + intel_sdvo_connector->top = + drm_property_create(dev, DRM_MODE_PROP_RANGE, + "top_margin", 2); + if (!intel_sdvo_connector->top) + return false; + + intel_sdvo_connector->top->values[0] = 0; + intel_sdvo_connector->top->values[1] = data_value[0]; + drm_connector_attach_property(connector, + intel_sdvo_connector->top, + intel_sdvo_connector->top_margin); + + intel_sdvo_connector->bottom = + drm_property_create(dev, DRM_MODE_PROP_RANGE, + "bottom_margin", 2); + if (!intel_sdvo_connector->bottom) + return false; + + intel_sdvo_connector->bottom->values[0] = 0; + intel_sdvo_connector->bottom->values[1] = data_value[0]; + drm_connector_attach_property(connector, + intel_sdvo_connector->bottom, + intel_sdvo_connector->bottom_margin); + DRM_DEBUG_KMS("v_overscan: max %d, " + "default %d, current %d\n", + data_value[0], data_value[1], response); + } + + ENHANCEMENT(hpos, HPOS); + ENHANCEMENT(vpos, VPOS); + ENHANCEMENT(saturation, SATURATION); + ENHANCEMENT(contrast, CONTRAST); + ENHANCEMENT(hue, HUE); + ENHANCEMENT(sharpness, SHARPNESS); + ENHANCEMENT(brightness, BRIGHTNESS); + ENHANCEMENT(flicker_filter, FLICKER_FILTER); + ENHANCEMENT(flicker_filter_adaptive, FLICKER_FILTER_ADAPTIVE); + ENHANCEMENT(flicker_filter_2d, FLICKER_FILTER_2D); + ENHANCEMENT(tv_chroma_filter, TV_CHROMA_FILTER); + ENHANCEMENT(tv_luma_filter, TV_LUMA_FILTER); + + if (enhancements.dot_crawl) { + if (!intel_sdvo_get_value(intel_sdvo, SDVO_CMD_GET_DOT_CRAWL, &response, 2)) + return false; + + intel_sdvo_connector->max_dot_crawl = 1; + intel_sdvo_connector->cur_dot_crawl = response & 0x1; + intel_sdvo_connector->dot_crawl = + drm_property_create(dev, DRM_MODE_PROP_RANGE, "dot_crawl", 2); + if (!intel_sdvo_connector->dot_crawl) + return false; + + intel_sdvo_connector->dot_crawl->values[0] = 0; + intel_sdvo_connector->dot_crawl->values[1] = 1; + drm_connector_attach_property(connector, + intel_sdvo_connector->dot_crawl, + intel_sdvo_connector->cur_dot_crawl); + DRM_DEBUG_KMS("dot crawl: current %d\n", response); + } + + return true; +} + +static bool +intel_sdvo_create_enhance_property_lvds(struct intel_sdvo *intel_sdvo, + struct intel_sdvo_connector *intel_sdvo_connector, + struct intel_sdvo_enhancements_reply enhancements) +{ + struct drm_device *dev = intel_sdvo->base.base.dev; + struct drm_connector *connector = &intel_sdvo_connector->base.base; + uint16_t response, data_value[2]; + + ENHANCEMENT(brightness, BRIGHTNESS); + + return true; +} +#undef ENHANCEMENT + +static bool intel_sdvo_create_enhance_property(struct intel_sdvo *intel_sdvo, + struct intel_sdvo_connector *intel_sdvo_connector) +{ + union { + struct intel_sdvo_enhancements_reply reply; + uint16_t response; + } enhancements; + + BUILD_BUG_ON(sizeof(enhancements) != 2); + + enhancements.response = 0; + intel_sdvo_get_value(intel_sdvo, + SDVO_CMD_GET_SUPPORTED_ENHANCEMENTS, + &enhancements, sizeof(enhancements)); + if (enhancements.response == 0) { + DRM_DEBUG_KMS("No enhancement is supported\n"); + return true; + } + + if (IS_TV(intel_sdvo_connector)) + return intel_sdvo_create_enhance_property_tv(intel_sdvo, intel_sdvo_connector, enhancements.reply); + else if(IS_LVDS(intel_sdvo_connector)) + return intel_sdvo_create_enhance_property_lvds(intel_sdvo, intel_sdvo_connector, enhancements.reply); + else + return true; +} + +static int intel_sdvo_ddc_proxy_xfer(struct i2c_adapter *adapter, + struct i2c_msg *msgs, + int num) +{ + struct intel_sdvo *sdvo = adapter->algo_data; + + if (!intel_sdvo_set_control_bus_switch(sdvo, sdvo->ddc_bus)) + return -EIO; + + return sdvo->i2c->algo->master_xfer(sdvo->i2c, msgs, num); +} + +static u32 intel_sdvo_ddc_proxy_func(struct i2c_adapter *adapter) +{ + struct intel_sdvo *sdvo = adapter->algo_data; + return sdvo->i2c->algo->functionality(sdvo->i2c); +} + +static const struct i2c_algorithm intel_sdvo_ddc_proxy = { + .master_xfer = intel_sdvo_ddc_proxy_xfer, + .functionality = intel_sdvo_ddc_proxy_func +}; + +static bool +intel_sdvo_init_ddc_proxy(struct intel_sdvo *sdvo, + struct drm_device *dev) +{ + sdvo->ddc.owner = THIS_MODULE; + sdvo->ddc.class = I2C_CLASS_DDC; + snprintf(sdvo->ddc.name, I2C_NAME_SIZE, "SDVO DDC proxy"); + sdvo->ddc.dev.parent = &dev->pdev->dev; + sdvo->ddc.algo_data = sdvo; + sdvo->ddc.algo = &intel_sdvo_ddc_proxy; + + return i2c_add_adapter(&sdvo->ddc) == 0; +} + +bool intel_sdvo_init(struct drm_device *dev, int sdvo_reg) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_encoder *intel_encoder; + struct intel_sdvo *intel_sdvo; + int i; + + intel_sdvo = kzalloc(sizeof(struct intel_sdvo), GFP_KERNEL); + if (!intel_sdvo) + return false; + + intel_sdvo->sdvo_reg = sdvo_reg; + intel_sdvo->slave_addr = intel_sdvo_get_slave_addr(dev, sdvo_reg) >> 1; + intel_sdvo_select_i2c_bus(dev_priv, intel_sdvo, sdvo_reg); + if (!intel_sdvo_init_ddc_proxy(intel_sdvo, dev)) { + kfree(intel_sdvo); + return false; + } + + /* encoder type will be decided later */ + intel_encoder = &intel_sdvo->base; + intel_encoder->type = INTEL_OUTPUT_SDVO; + drm_encoder_init(dev, &intel_encoder->base, &intel_sdvo_enc_funcs, 0); + + /* Read the regs to test if we can talk to the device */ + for (i = 0; i < 0x40; i++) { + u8 byte; + + if (!intel_sdvo_read_byte(intel_sdvo, i, &byte)) { + DRM_DEBUG_KMS("No SDVO device found on SDVO%c\n", + IS_SDVOB(sdvo_reg) ? 'B' : 'C'); + goto err; + } + } + + if (IS_SDVOB(sdvo_reg)) + dev_priv->hotplug_supported_mask |= SDVOB_HOTPLUG_INT_STATUS; + else + dev_priv->hotplug_supported_mask |= SDVOC_HOTPLUG_INT_STATUS; + + drm_encoder_helper_add(&intel_encoder->base, &intel_sdvo_helper_funcs); + + /* In default case sdvo lvds is false */ + if (!intel_sdvo_get_capabilities(intel_sdvo, &intel_sdvo->caps)) + goto err; + + if (intel_sdvo_output_setup(intel_sdvo, + intel_sdvo->caps.output_flags) != true) { + DRM_DEBUG_KMS("SDVO output failed to setup on SDVO%c\n", + IS_SDVOB(sdvo_reg) ? 'B' : 'C'); + goto err; + } + + intel_sdvo_select_ddc_bus(dev_priv, intel_sdvo, sdvo_reg); + + /* Set the input timing to the screen. Assume always input 0. */ + if (!intel_sdvo_set_target_input(intel_sdvo)) + goto err; + + if (!intel_sdvo_get_input_pixel_clock_range(intel_sdvo, + &intel_sdvo->pixel_clock_min, + &intel_sdvo->pixel_clock_max)) + goto err; + + DRM_DEBUG_KMS("%s device VID/DID: %02X:%02X.%02X, " + "clock range %dMHz - %dMHz, " + "input 1: %c, input 2: %c, " + "output 1: %c, output 2: %c\n", + SDVO_NAME(intel_sdvo), + intel_sdvo->caps.vendor_id, intel_sdvo->caps.device_id, + intel_sdvo->caps.device_rev_id, + intel_sdvo->pixel_clock_min / 1000, + intel_sdvo->pixel_clock_max / 1000, + (intel_sdvo->caps.sdvo_inputs_mask & 0x1) ? 'Y' : 'N', + (intel_sdvo->caps.sdvo_inputs_mask & 0x2) ? 'Y' : 'N', + /* check currently supported outputs */ + intel_sdvo->caps.output_flags & + (SDVO_OUTPUT_TMDS0 | SDVO_OUTPUT_RGB0) ? 'Y' : 'N', + intel_sdvo->caps.output_flags & + (SDVO_OUTPUT_TMDS1 | SDVO_OUTPUT_RGB1) ? 'Y' : 'N'); + return true; + +err: + drm_encoder_cleanup(&intel_encoder->base); + i2c_del_adapter(&intel_sdvo->ddc); + kfree(intel_sdvo); + + return false; +} diff --git a/drivers/gpu/drm/i915/intel_sdvo_regs.h b/drivers/gpu/drm/i915/intel_sdvo_regs.h new file mode 100644 index 00000000..c5c8ddf2 --- /dev/null +++ b/drivers/gpu/drm/i915/intel_sdvo_regs.h @@ -0,0 +1,728 @@ +/* + * Copyright © 2006-2007 Intel Corporation + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Eric Anholt + */ + +/** + * @file SDVO command definitions and structures. + */ + +#define SDVO_OUTPUT_FIRST (0) +#define SDVO_OUTPUT_TMDS0 (1 << 0) +#define SDVO_OUTPUT_RGB0 (1 << 1) +#define SDVO_OUTPUT_CVBS0 (1 << 2) +#define SDVO_OUTPUT_SVID0 (1 << 3) +#define SDVO_OUTPUT_YPRPB0 (1 << 4) +#define SDVO_OUTPUT_SCART0 (1 << 5) +#define SDVO_OUTPUT_LVDS0 (1 << 6) +#define SDVO_OUTPUT_TMDS1 (1 << 8) +#define SDVO_OUTPUT_RGB1 (1 << 9) +#define SDVO_OUTPUT_CVBS1 (1 << 10) +#define SDVO_OUTPUT_SVID1 (1 << 11) +#define SDVO_OUTPUT_YPRPB1 (1 << 12) +#define SDVO_OUTPUT_SCART1 (1 << 13) +#define SDVO_OUTPUT_LVDS1 (1 << 14) +#define SDVO_OUTPUT_LAST (14) + +struct intel_sdvo_caps { + u8 vendor_id; + u8 device_id; + u8 device_rev_id; + u8 sdvo_version_major; + u8 sdvo_version_minor; + unsigned int sdvo_inputs_mask:2; + unsigned int smooth_scaling:1; + unsigned int sharp_scaling:1; + unsigned int up_scaling:1; + unsigned int down_scaling:1; + unsigned int stall_support:1; + unsigned int pad:1; + u16 output_flags; +} __attribute__((packed)); + +/* Note: SDVO detailed timing flags match EDID misc flags. */ +#define DTD_FLAG_HSYNC_POSITIVE (1 << 1) +#define DTD_FLAG_VSYNC_POSITIVE (1 << 2) +#define DTD_FLAG_INTERLACE (1 << 7) + +/** This matches the EDID DTD structure, more or less */ +struct intel_sdvo_dtd { + struct { + u16 clock; /**< pixel clock, in 10kHz units */ + u8 h_active; /**< lower 8 bits (pixels) */ + u8 h_blank; /**< lower 8 bits (pixels) */ + u8 h_high; /**< upper 4 bits each h_active, h_blank */ + u8 v_active; /**< lower 8 bits (lines) */ + u8 v_blank; /**< lower 8 bits (lines) */ + u8 v_high; /**< upper 4 bits each v_active, v_blank */ + } part1; + + struct { + u8 h_sync_off; /**< lower 8 bits, from hblank start */ + u8 h_sync_width; /**< lower 8 bits (pixels) */ + /** lower 4 bits each vsync offset, vsync width */ + u8 v_sync_off_width; + /** + * 2 high bits of hsync offset, 2 high bits of hsync width, + * bits 4-5 of vsync offset, and 2 high bits of vsync width. + */ + u8 sync_off_width_high; + u8 dtd_flags; + u8 sdvo_flags; + /** bits 6-7 of vsync offset at bits 6-7 */ + u8 v_sync_off_high; + u8 reserved; + } part2; +} __attribute__((packed)); + +struct intel_sdvo_pixel_clock_range { + u16 min; /**< pixel clock, in 10kHz units */ + u16 max; /**< pixel clock, in 10kHz units */ +} __attribute__((packed)); + +struct intel_sdvo_preferred_input_timing_args { + u16 clock; + u16 width; + u16 height; + u8 interlace:1; + u8 scaled:1; + u8 pad:6; +} __attribute__((packed)); + +/* I2C registers for SDVO */ +#define SDVO_I2C_ARG_0 0x07 +#define SDVO_I2C_ARG_1 0x06 +#define SDVO_I2C_ARG_2 0x05 +#define SDVO_I2C_ARG_3 0x04 +#define SDVO_I2C_ARG_4 0x03 +#define SDVO_I2C_ARG_5 0x02 +#define SDVO_I2C_ARG_6 0x01 +#define SDVO_I2C_ARG_7 0x00 +#define SDVO_I2C_OPCODE 0x08 +#define SDVO_I2C_CMD_STATUS 0x09 +#define SDVO_I2C_RETURN_0 0x0a +#define SDVO_I2C_RETURN_1 0x0b +#define SDVO_I2C_RETURN_2 0x0c +#define SDVO_I2C_RETURN_3 0x0d +#define SDVO_I2C_RETURN_4 0x0e +#define SDVO_I2C_RETURN_5 0x0f +#define SDVO_I2C_RETURN_6 0x10 +#define SDVO_I2C_RETURN_7 0x11 +#define SDVO_I2C_VENDOR_BEGIN 0x20 + +/* Status results */ +#define SDVO_CMD_STATUS_POWER_ON 0x0 +#define SDVO_CMD_STATUS_SUCCESS 0x1 +#define SDVO_CMD_STATUS_NOTSUPP 0x2 +#define SDVO_CMD_STATUS_INVALID_ARG 0x3 +#define SDVO_CMD_STATUS_PENDING 0x4 +#define SDVO_CMD_STATUS_TARGET_NOT_SPECIFIED 0x5 +#define SDVO_CMD_STATUS_SCALING_NOT_SUPP 0x6 + +/* SDVO commands, argument/result registers */ + +#define SDVO_CMD_RESET 0x01 + +/** Returns a struct intel_sdvo_caps */ +#define SDVO_CMD_GET_DEVICE_CAPS 0x02 + +#define SDVO_CMD_GET_FIRMWARE_REV 0x86 +# define SDVO_DEVICE_FIRMWARE_MINOR SDVO_I2C_RETURN_0 +# define SDVO_DEVICE_FIRMWARE_MAJOR SDVO_I2C_RETURN_1 +# define SDVO_DEVICE_FIRMWARE_PATCH SDVO_I2C_RETURN_2 + +/** + * Reports which inputs are trained (managed to sync). + * + * Devices must have trained within 2 vsyncs of a mode change. + */ +#define SDVO_CMD_GET_TRAINED_INPUTS 0x03 +struct intel_sdvo_get_trained_inputs_response { + unsigned int input0_trained:1; + unsigned int input1_trained:1; + unsigned int pad:6; +} __attribute__((packed)); + +/** Returns a struct intel_sdvo_output_flags of active outputs. */ +#define SDVO_CMD_GET_ACTIVE_OUTPUTS 0x04 + +/** + * Sets the current set of active outputs. + * + * Takes a struct intel_sdvo_output_flags. Must be preceded by a SET_IN_OUT_MAP + * on multi-output devices. + */ +#define SDVO_CMD_SET_ACTIVE_OUTPUTS 0x05 + +/** + * Returns the current mapping of SDVO inputs to outputs on the device. + * + * Returns two struct intel_sdvo_output_flags structures. + */ +#define SDVO_CMD_GET_IN_OUT_MAP 0x06 +struct intel_sdvo_in_out_map { + u16 in0, in1; +}; + +/** + * Sets the current mapping of SDVO inputs to outputs on the device. + * + * Takes two struct i380_sdvo_output_flags structures. + */ +#define SDVO_CMD_SET_IN_OUT_MAP 0x07 + +/** + * Returns a struct intel_sdvo_output_flags of attached displays. + */ +#define SDVO_CMD_GET_ATTACHED_DISPLAYS 0x0b + +/** + * Returns a struct intel_sdvo_ouptut_flags of displays supporting hot plugging. + */ +#define SDVO_CMD_GET_HOT_PLUG_SUPPORT 0x0c + +/** + * Takes a struct intel_sdvo_output_flags. + */ +#define SDVO_CMD_SET_ACTIVE_HOT_PLUG 0x0d + +/** + * Returns a struct intel_sdvo_output_flags of displays with hot plug + * interrupts enabled. + */ +#define SDVO_CMD_GET_ACTIVE_HOT_PLUG 0x0e + +#define SDVO_CMD_GET_INTERRUPT_EVENT_SOURCE 0x0f +struct intel_sdvo_get_interrupt_event_source_response { + u16 interrupt_status; + unsigned int ambient_light_interrupt:1; + unsigned int hdmi_audio_encrypt_change:1; + unsigned int pad:6; +} __attribute__((packed)); + +/** + * Selects which input is affected by future input commands. + * + * Commands affected include SET_INPUT_TIMINGS_PART[12], + * GET_INPUT_TIMINGS_PART[12], GET_PREFERRED_INPUT_TIMINGS_PART[12], + * GET_INPUT_PIXEL_CLOCK_RANGE, and CREATE_PREFERRED_INPUT_TIMINGS. + */ +#define SDVO_CMD_SET_TARGET_INPUT 0x10 +struct intel_sdvo_set_target_input_args { + unsigned int target_1:1; + unsigned int pad:7; +} __attribute__((packed)); + +/** + * Takes a struct intel_sdvo_output_flags of which outputs are targeted by + * future output commands. + * + * Affected commands inclue SET_OUTPUT_TIMINGS_PART[12], + * GET_OUTPUT_TIMINGS_PART[12], and GET_OUTPUT_PIXEL_CLOCK_RANGE. + */ +#define SDVO_CMD_SET_TARGET_OUTPUT 0x11 + +#define SDVO_CMD_GET_INPUT_TIMINGS_PART1 0x12 +#define SDVO_CMD_GET_INPUT_TIMINGS_PART2 0x13 +#define SDVO_CMD_SET_INPUT_TIMINGS_PART1 0x14 +#define SDVO_CMD_SET_INPUT_TIMINGS_PART2 0x15 +#define SDVO_CMD_SET_OUTPUT_TIMINGS_PART1 0x16 +#define SDVO_CMD_SET_OUTPUT_TIMINGS_PART2 0x17 +#define SDVO_CMD_GET_OUTPUT_TIMINGS_PART1 0x18 +#define SDVO_CMD_GET_OUTPUT_TIMINGS_PART2 0x19 +/* Part 1 */ +# define SDVO_DTD_CLOCK_LOW SDVO_I2C_ARG_0 +# define SDVO_DTD_CLOCK_HIGH SDVO_I2C_ARG_1 +# define SDVO_DTD_H_ACTIVE SDVO_I2C_ARG_2 +# define SDVO_DTD_H_BLANK SDVO_I2C_ARG_3 +# define SDVO_DTD_H_HIGH SDVO_I2C_ARG_4 +# define SDVO_DTD_V_ACTIVE SDVO_I2C_ARG_5 +# define SDVO_DTD_V_BLANK SDVO_I2C_ARG_6 +# define SDVO_DTD_V_HIGH SDVO_I2C_ARG_7 +/* Part 2 */ +# define SDVO_DTD_HSYNC_OFF SDVO_I2C_ARG_0 +# define SDVO_DTD_HSYNC_WIDTH SDVO_I2C_ARG_1 +# define SDVO_DTD_VSYNC_OFF_WIDTH SDVO_I2C_ARG_2 +# define SDVO_DTD_SYNC_OFF_WIDTH_HIGH SDVO_I2C_ARG_3 +# define SDVO_DTD_DTD_FLAGS SDVO_I2C_ARG_4 +# define SDVO_DTD_DTD_FLAG_INTERLACED (1 << 7) +# define SDVO_DTD_DTD_FLAG_STEREO_MASK (3 << 5) +# define SDVO_DTD_DTD_FLAG_INPUT_MASK (3 << 3) +# define SDVO_DTD_DTD_FLAG_SYNC_MASK (3 << 1) +# define SDVO_DTD_SDVO_FLAS SDVO_I2C_ARG_5 +# define SDVO_DTD_SDVO_FLAG_STALL (1 << 7) +# define SDVO_DTD_SDVO_FLAG_CENTERED (0 << 6) +# define SDVO_DTD_SDVO_FLAG_UPPER_LEFT (1 << 6) +# define SDVO_DTD_SDVO_FLAG_SCALING_MASK (3 << 4) +# define SDVO_DTD_SDVO_FLAG_SCALING_NONE (0 << 4) +# define SDVO_DTD_SDVO_FLAG_SCALING_SHARP (1 << 4) +# define SDVO_DTD_SDVO_FLAG_SCALING_SMOOTH (2 << 4) +# define SDVO_DTD_VSYNC_OFF_HIGH SDVO_I2C_ARG_6 + +/** + * Generates a DTD based on the given width, height, and flags. + * + * This will be supported by any device supporting scaling or interlaced + * modes. + */ +#define SDVO_CMD_CREATE_PREFERRED_INPUT_TIMING 0x1a +# define SDVO_PREFERRED_INPUT_TIMING_CLOCK_LOW SDVO_I2C_ARG_0 +# define SDVO_PREFERRED_INPUT_TIMING_CLOCK_HIGH SDVO_I2C_ARG_1 +# define SDVO_PREFERRED_INPUT_TIMING_WIDTH_LOW SDVO_I2C_ARG_2 +# define SDVO_PREFERRED_INPUT_TIMING_WIDTH_HIGH SDVO_I2C_ARG_3 +# define SDVO_PREFERRED_INPUT_TIMING_HEIGHT_LOW SDVO_I2C_ARG_4 +# define SDVO_PREFERRED_INPUT_TIMING_HEIGHT_HIGH SDVO_I2C_ARG_5 +# define SDVO_PREFERRED_INPUT_TIMING_FLAGS SDVO_I2C_ARG_6 +# define SDVO_PREFERRED_INPUT_TIMING_FLAGS_INTERLACED (1 << 0) +# define SDVO_PREFERRED_INPUT_TIMING_FLAGS_SCALED (1 << 1) + +#define SDVO_CMD_GET_PREFERRED_INPUT_TIMING_PART1 0x1b +#define SDVO_CMD_GET_PREFERRED_INPUT_TIMING_PART2 0x1c + +/** Returns a struct intel_sdvo_pixel_clock_range */ +#define SDVO_CMD_GET_INPUT_PIXEL_CLOCK_RANGE 0x1d +/** Returns a struct intel_sdvo_pixel_clock_range */ +#define SDVO_CMD_GET_OUTPUT_PIXEL_CLOCK_RANGE 0x1e + +/** Returns a byte bitfield containing SDVO_CLOCK_RATE_MULT_* flags */ +#define SDVO_CMD_GET_SUPPORTED_CLOCK_RATE_MULTS 0x1f + +/** Returns a byte containing a SDVO_CLOCK_RATE_MULT_* flag */ +#define SDVO_CMD_GET_CLOCK_RATE_MULT 0x20 +/** Takes a byte containing a SDVO_CLOCK_RATE_MULT_* flag */ +#define SDVO_CMD_SET_CLOCK_RATE_MULT 0x21 +# define SDVO_CLOCK_RATE_MULT_1X (1 << 0) +# define SDVO_CLOCK_RATE_MULT_2X (1 << 1) +# define SDVO_CLOCK_RATE_MULT_4X (1 << 3) + +#define SDVO_CMD_GET_SUPPORTED_TV_FORMATS 0x27 +/** 6 bytes of bit flags for TV formats shared by all TV format functions */ +struct intel_sdvo_tv_format { + unsigned int ntsc_m:1; + unsigned int ntsc_j:1; + unsigned int ntsc_443:1; + unsigned int pal_b:1; + unsigned int pal_d:1; + unsigned int pal_g:1; + unsigned int pal_h:1; + unsigned int pal_i:1; + + unsigned int pal_m:1; + unsigned int pal_n:1; + unsigned int pal_nc:1; + unsigned int pal_60:1; + unsigned int secam_b:1; + unsigned int secam_d:1; + unsigned int secam_g:1; + unsigned int secam_k:1; + + unsigned int secam_k1:1; + unsigned int secam_l:1; + unsigned int secam_60:1; + unsigned int hdtv_std_smpte_240m_1080i_59:1; + unsigned int hdtv_std_smpte_240m_1080i_60:1; + unsigned int hdtv_std_smpte_260m_1080i_59:1; + unsigned int hdtv_std_smpte_260m_1080i_60:1; + unsigned int hdtv_std_smpte_274m_1080i_50:1; + + unsigned int hdtv_std_smpte_274m_1080i_59:1; + unsigned int hdtv_std_smpte_274m_1080i_60:1; + unsigned int hdtv_std_smpte_274m_1080p_23:1; + unsigned int hdtv_std_smpte_274m_1080p_24:1; + unsigned int hdtv_std_smpte_274m_1080p_25:1; + unsigned int hdtv_std_smpte_274m_1080p_29:1; + unsigned int hdtv_std_smpte_274m_1080p_30:1; + unsigned int hdtv_std_smpte_274m_1080p_50:1; + + unsigned int hdtv_std_smpte_274m_1080p_59:1; + unsigned int hdtv_std_smpte_274m_1080p_60:1; + unsigned int hdtv_std_smpte_295m_1080i_50:1; + unsigned int hdtv_std_smpte_295m_1080p_50:1; + unsigned int hdtv_std_smpte_296m_720p_59:1; + unsigned int hdtv_std_smpte_296m_720p_60:1; + unsigned int hdtv_std_smpte_296m_720p_50:1; + unsigned int hdtv_std_smpte_293m_480p_59:1; + + unsigned int hdtv_std_smpte_170m_480i_59:1; + unsigned int hdtv_std_iturbt601_576i_50:1; + unsigned int hdtv_std_iturbt601_576p_50:1; + unsigned int hdtv_std_eia_7702a_480i_60:1; + unsigned int hdtv_std_eia_7702a_480p_60:1; + unsigned int pad:3; +} __attribute__((packed)); + +#define SDVO_CMD_GET_TV_FORMAT 0x28 + +#define SDVO_CMD_SET_TV_FORMAT 0x29 + +/** Returns the resolutiosn that can be used with the given TV format */ +#define SDVO_CMD_GET_SDTV_RESOLUTION_SUPPORT 0x83 +struct intel_sdvo_sdtv_resolution_request { + unsigned int ntsc_m:1; + unsigned int ntsc_j:1; + unsigned int ntsc_443:1; + unsigned int pal_b:1; + unsigned int pal_d:1; + unsigned int pal_g:1; + unsigned int pal_h:1; + unsigned int pal_i:1; + + unsigned int pal_m:1; + unsigned int pal_n:1; + unsigned int pal_nc:1; + unsigned int pal_60:1; + unsigned int secam_b:1; + unsigned int secam_d:1; + unsigned int secam_g:1; + unsigned int secam_k:1; + + unsigned int secam_k1:1; + unsigned int secam_l:1; + unsigned int secam_60:1; + unsigned int pad:5; +} __attribute__((packed)); + +struct intel_sdvo_sdtv_resolution_reply { + unsigned int res_320x200:1; + unsigned int res_320x240:1; + unsigned int res_400x300:1; + unsigned int res_640x350:1; + unsigned int res_640x400:1; + unsigned int res_640x480:1; + unsigned int res_704x480:1; + unsigned int res_704x576:1; + + unsigned int res_720x350:1; + unsigned int res_720x400:1; + unsigned int res_720x480:1; + unsigned int res_720x540:1; + unsigned int res_720x576:1; + unsigned int res_768x576:1; + unsigned int res_800x600:1; + unsigned int res_832x624:1; + + unsigned int res_920x766:1; + unsigned int res_1024x768:1; + unsigned int res_1280x1024:1; + unsigned int pad:5; +} __attribute__((packed)); + +/* Get supported resolution with squire pixel aspect ratio that can be + scaled for the requested HDTV format */ +#define SDVO_CMD_GET_SCALED_HDTV_RESOLUTION_SUPPORT 0x85 + +struct intel_sdvo_hdtv_resolution_request { + unsigned int hdtv_std_smpte_240m_1080i_59:1; + unsigned int hdtv_std_smpte_240m_1080i_60:1; + unsigned int hdtv_std_smpte_260m_1080i_59:1; + unsigned int hdtv_std_smpte_260m_1080i_60:1; + unsigned int hdtv_std_smpte_274m_1080i_50:1; + unsigned int hdtv_std_smpte_274m_1080i_59:1; + unsigned int hdtv_std_smpte_274m_1080i_60:1; + unsigned int hdtv_std_smpte_274m_1080p_23:1; + + unsigned int hdtv_std_smpte_274m_1080p_24:1; + unsigned int hdtv_std_smpte_274m_1080p_25:1; + unsigned int hdtv_std_smpte_274m_1080p_29:1; + unsigned int hdtv_std_smpte_274m_1080p_30:1; + unsigned int hdtv_std_smpte_274m_1080p_50:1; + unsigned int hdtv_std_smpte_274m_1080p_59:1; + unsigned int hdtv_std_smpte_274m_1080p_60:1; + unsigned int hdtv_std_smpte_295m_1080i_50:1; + + unsigned int hdtv_std_smpte_295m_1080p_50:1; + unsigned int hdtv_std_smpte_296m_720p_59:1; + unsigned int hdtv_std_smpte_296m_720p_60:1; + unsigned int hdtv_std_smpte_296m_720p_50:1; + unsigned int hdtv_std_smpte_293m_480p_59:1; + unsigned int hdtv_std_smpte_170m_480i_59:1; + unsigned int hdtv_std_iturbt601_576i_50:1; + unsigned int hdtv_std_iturbt601_576p_50:1; + + unsigned int hdtv_std_eia_7702a_480i_60:1; + unsigned int hdtv_std_eia_7702a_480p_60:1; + unsigned int pad:6; +} __attribute__((packed)); + +struct intel_sdvo_hdtv_resolution_reply { + unsigned int res_640x480:1; + unsigned int res_800x600:1; + unsigned int res_1024x768:1; + unsigned int res_1280x960:1; + unsigned int res_1400x1050:1; + unsigned int res_1600x1200:1; + unsigned int res_1920x1440:1; + unsigned int res_2048x1536:1; + + unsigned int res_2560x1920:1; + unsigned int res_3200x2400:1; + unsigned int res_3840x2880:1; + unsigned int pad1:5; + + unsigned int res_848x480:1; + unsigned int res_1064x600:1; + unsigned int res_1280x720:1; + unsigned int res_1360x768:1; + unsigned int res_1704x960:1; + unsigned int res_1864x1050:1; + unsigned int res_1920x1080:1; + unsigned int res_2128x1200:1; + + unsigned int res_2560x1400:1; + unsigned int res_2728x1536:1; + unsigned int res_3408x1920:1; + unsigned int res_4264x2400:1; + unsigned int res_5120x2880:1; + unsigned int pad2:3; + + unsigned int res_768x480:1; + unsigned int res_960x600:1; + unsigned int res_1152x720:1; + unsigned int res_1124x768:1; + unsigned int res_1536x960:1; + unsigned int res_1680x1050:1; + unsigned int res_1728x1080:1; + unsigned int res_1920x1200:1; + + unsigned int res_2304x1440:1; + unsigned int res_2456x1536:1; + unsigned int res_3072x1920:1; + unsigned int res_3840x2400:1; + unsigned int res_4608x2880:1; + unsigned int pad3:3; + + unsigned int res_1280x1024:1; + unsigned int pad4:7; + + unsigned int res_1280x768:1; + unsigned int pad5:7; +} __attribute__((packed)); + +/* Get supported power state returns info for encoder and monitor, rely on + last SetTargetInput and SetTargetOutput calls */ +#define SDVO_CMD_GET_SUPPORTED_POWER_STATES 0x2a +/* Get power state returns info for encoder and monitor, rely on last + SetTargetInput and SetTargetOutput calls */ +#define SDVO_CMD_GET_POWER_STATE 0x2b +#define SDVO_CMD_GET_ENCODER_POWER_STATE 0x2b +#define SDVO_CMD_SET_ENCODER_POWER_STATE 0x2c +# define SDVO_ENCODER_STATE_ON (1 << 0) +# define SDVO_ENCODER_STATE_STANDBY (1 << 1) +# define SDVO_ENCODER_STATE_SUSPEND (1 << 2) +# define SDVO_ENCODER_STATE_OFF (1 << 3) +# define SDVO_MONITOR_STATE_ON (1 << 4) +# define SDVO_MONITOR_STATE_STANDBY (1 << 5) +# define SDVO_MONITOR_STATE_SUSPEND (1 << 6) +# define SDVO_MONITOR_STATE_OFF (1 << 7) + +#define SDVO_CMD_GET_MAX_PANEL_POWER_SEQUENCING 0x2d +#define SDVO_CMD_GET_PANEL_POWER_SEQUENCING 0x2e +#define SDVO_CMD_SET_PANEL_POWER_SEQUENCING 0x2f +/** + * The panel power sequencing parameters are in units of milliseconds. + * The high fields are bits 8:9 of the 10-bit values. + */ +struct sdvo_panel_power_sequencing { + u8 t0; + u8 t1; + u8 t2; + u8 t3; + u8 t4; + + unsigned int t0_high:2; + unsigned int t1_high:2; + unsigned int t2_high:2; + unsigned int t3_high:2; + + unsigned int t4_high:2; + unsigned int pad:6; +} __attribute__((packed)); + +#define SDVO_CMD_GET_MAX_BACKLIGHT_LEVEL 0x30 +struct sdvo_max_backlight_reply { + u8 max_value; + u8 default_value; +} __attribute__((packed)); + +#define SDVO_CMD_GET_BACKLIGHT_LEVEL 0x31 +#define SDVO_CMD_SET_BACKLIGHT_LEVEL 0x32 + +#define SDVO_CMD_GET_AMBIENT_LIGHT 0x33 +struct sdvo_get_ambient_light_reply { + u16 trip_low; + u16 trip_high; + u16 value; +} __attribute__((packed)); +#define SDVO_CMD_SET_AMBIENT_LIGHT 0x34 +struct sdvo_set_ambient_light_reply { + u16 trip_low; + u16 trip_high; + unsigned int enable:1; + unsigned int pad:7; +} __attribute__((packed)); + +/* Set display power state */ +#define SDVO_CMD_SET_DISPLAY_POWER_STATE 0x7d +# define SDVO_DISPLAY_STATE_ON (1 << 0) +# define SDVO_DISPLAY_STATE_STANDBY (1 << 1) +# define SDVO_DISPLAY_STATE_SUSPEND (1 << 2) +# define SDVO_DISPLAY_STATE_OFF (1 << 3) + +#define SDVO_CMD_GET_SUPPORTED_ENHANCEMENTS 0x84 +struct intel_sdvo_enhancements_reply { + unsigned int flicker_filter:1; + unsigned int flicker_filter_adaptive:1; + unsigned int flicker_filter_2d:1; + unsigned int saturation:1; + unsigned int hue:1; + unsigned int brightness:1; + unsigned int contrast:1; + unsigned int overscan_h:1; + + unsigned int overscan_v:1; + unsigned int hpos:1; + unsigned int vpos:1; + unsigned int sharpness:1; + unsigned int dot_crawl:1; + unsigned int dither:1; + unsigned int tv_chroma_filter:1; + unsigned int tv_luma_filter:1; +} __attribute__((packed)); + +/* Picture enhancement limits below are dependent on the current TV format, + * and thus need to be queried and set after it. + */ +#define SDVO_CMD_GET_MAX_FLICKER_FILTER 0x4d +#define SDVO_CMD_GET_MAX_FLICKER_FILTER_ADAPTIVE 0x7b +#define SDVO_CMD_GET_MAX_FLICKER_FILTER_2D 0x52 +#define SDVO_CMD_GET_MAX_SATURATION 0x55 +#define SDVO_CMD_GET_MAX_HUE 0x58 +#define SDVO_CMD_GET_MAX_BRIGHTNESS 0x5b +#define SDVO_CMD_GET_MAX_CONTRAST 0x5e +#define SDVO_CMD_GET_MAX_OVERSCAN_H 0x61 +#define SDVO_CMD_GET_MAX_OVERSCAN_V 0x64 +#define SDVO_CMD_GET_MAX_HPOS 0x67 +#define SDVO_CMD_GET_MAX_VPOS 0x6a +#define SDVO_CMD_GET_MAX_SHARPNESS 0x6d +#define SDVO_CMD_GET_MAX_TV_CHROMA_FILTER 0x74 +#define SDVO_CMD_GET_MAX_TV_LUMA_FILTER 0x77 +struct intel_sdvo_enhancement_limits_reply { + u16 max_value; + u16 default_value; +} __attribute__((packed)); + +#define SDVO_CMD_GET_LVDS_PANEL_INFORMATION 0x7f +#define SDVO_CMD_SET_LVDS_PANEL_INFORMATION 0x80 +# define SDVO_LVDS_COLOR_DEPTH_18 (0 << 0) +# define SDVO_LVDS_COLOR_DEPTH_24 (1 << 0) +# define SDVO_LVDS_CONNECTOR_SPWG (0 << 2) +# define SDVO_LVDS_CONNECTOR_OPENLDI (1 << 2) +# define SDVO_LVDS_SINGLE_CHANNEL (0 << 4) +# define SDVO_LVDS_DUAL_CHANNEL (1 << 4) + +#define SDVO_CMD_GET_FLICKER_FILTER 0x4e +#define SDVO_CMD_SET_FLICKER_FILTER 0x4f +#define SDVO_CMD_GET_FLICKER_FILTER_ADAPTIVE 0x50 +#define SDVO_CMD_SET_FLICKER_FILTER_ADAPTIVE 0x51 +#define SDVO_CMD_GET_FLICKER_FILTER_2D 0x53 +#define SDVO_CMD_SET_FLICKER_FILTER_2D 0x54 +#define SDVO_CMD_GET_SATURATION 0x56 +#define SDVO_CMD_SET_SATURATION 0x57 +#define SDVO_CMD_GET_HUE 0x59 +#define SDVO_CMD_SET_HUE 0x5a +#define SDVO_CMD_GET_BRIGHTNESS 0x5c +#define SDVO_CMD_SET_BRIGHTNESS 0x5d +#define SDVO_CMD_GET_CONTRAST 0x5f +#define SDVO_CMD_SET_CONTRAST 0x60 +#define SDVO_CMD_GET_OVERSCAN_H 0x62 +#define SDVO_CMD_SET_OVERSCAN_H 0x63 +#define SDVO_CMD_GET_OVERSCAN_V 0x65 +#define SDVO_CMD_SET_OVERSCAN_V 0x66 +#define SDVO_CMD_GET_HPOS 0x68 +#define SDVO_CMD_SET_HPOS 0x69 +#define SDVO_CMD_GET_VPOS 0x6b +#define SDVO_CMD_SET_VPOS 0x6c +#define SDVO_CMD_GET_SHARPNESS 0x6e +#define SDVO_CMD_SET_SHARPNESS 0x6f +#define SDVO_CMD_GET_TV_CHROMA_FILTER 0x75 +#define SDVO_CMD_SET_TV_CHROMA_FILTER 0x76 +#define SDVO_CMD_GET_TV_LUMA_FILTER 0x78 +#define SDVO_CMD_SET_TV_LUMA_FILTER 0x79 +struct intel_sdvo_enhancements_arg { + u16 value; +}__attribute__((packed)); + +#define SDVO_CMD_GET_DOT_CRAWL 0x70 +#define SDVO_CMD_SET_DOT_CRAWL 0x71 +# define SDVO_DOT_CRAWL_ON (1 << 0) +# define SDVO_DOT_CRAWL_DEFAULT_ON (1 << 1) + +#define SDVO_CMD_GET_DITHER 0x72 +#define SDVO_CMD_SET_DITHER 0x73 +# define SDVO_DITHER_ON (1 << 0) +# define SDVO_DITHER_DEFAULT_ON (1 << 1) + +#define SDVO_CMD_SET_CONTROL_BUS_SWITCH 0x7a +# define SDVO_CONTROL_BUS_PROM (1 << 0) +# define SDVO_CONTROL_BUS_DDC1 (1 << 1) +# define SDVO_CONTROL_BUS_DDC2 (1 << 2) +# define SDVO_CONTROL_BUS_DDC3 (1 << 3) + +/* HDMI op codes */ +#define SDVO_CMD_GET_SUPP_ENCODE 0x9d +#define SDVO_CMD_GET_ENCODE 0x9e +#define SDVO_CMD_SET_ENCODE 0x9f + #define SDVO_ENCODE_DVI 0x0 + #define SDVO_ENCODE_HDMI 0x1 +#define SDVO_CMD_SET_PIXEL_REPLI 0x8b +#define SDVO_CMD_GET_PIXEL_REPLI 0x8c +#define SDVO_CMD_GET_COLORIMETRY_CAP 0x8d +#define SDVO_CMD_SET_COLORIMETRY 0x8e + #define SDVO_COLORIMETRY_RGB256 0x0 + #define SDVO_COLORIMETRY_RGB220 0x1 + #define SDVO_COLORIMETRY_YCrCb422 0x3 + #define SDVO_COLORIMETRY_YCrCb444 0x4 +#define SDVO_CMD_GET_COLORIMETRY 0x8f +#define SDVO_CMD_GET_AUDIO_ENCRYPT_PREFER 0x90 +#define SDVO_CMD_SET_AUDIO_STAT 0x91 +#define SDVO_CMD_GET_AUDIO_STAT 0x92 +#define SDVO_CMD_SET_HBUF_INDEX 0x93 +#define SDVO_CMD_GET_HBUF_INDEX 0x94 +#define SDVO_CMD_GET_HBUF_INFO 0x95 +#define SDVO_CMD_SET_HBUF_AV_SPLIT 0x96 +#define SDVO_CMD_GET_HBUF_AV_SPLIT 0x97 +#define SDVO_CMD_SET_HBUF_DATA 0x98 +#define SDVO_CMD_GET_HBUF_DATA 0x99 +#define SDVO_CMD_SET_HBUF_TXRATE 0x9a +#define SDVO_CMD_GET_HBUF_TXRATE 0x9b + #define SDVO_HBUF_TX_DISABLED (0 << 6) + #define SDVO_HBUF_TX_ONCE (2 << 6) + #define SDVO_HBUF_TX_VSYNC (3 << 6) +#define SDVO_CMD_GET_AUDIO_TX_INFO 0x9c +#define SDVO_NEED_TO_STALL (1 << 7) + +struct intel_sdvo_encode{ + u8 dvi_rev; + u8 hdmi_rev; +} __attribute__ ((packed)); diff --git a/drivers/gpu/drm/i915/intel_tv.c b/drivers/gpu/drm/i915/intel_tv.c new file mode 100644 index 00000000..2136e6bc --- /dev/null +++ b/drivers/gpu/drm/i915/intel_tv.c @@ -0,0 +1,1730 @@ +/* + * Copyright © 2006-2008 Intel Corporation + * Jesse Barnes + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Eric Anholt + * + */ + +/** @file + * Integrated TV-out support for the 915GM and 945GM. + */ + +#include "drmP.h" +#include "drm.h" +#include "drm_crtc.h" +#include "drm_edid.h" +#include "intel_drv.h" +#include "i915_drm.h" +#include "i915_drv.h" + +enum tv_margin { + TV_MARGIN_LEFT, TV_MARGIN_TOP, + TV_MARGIN_RIGHT, TV_MARGIN_BOTTOM +}; + +/** Private structure for the integrated TV support */ +struct intel_tv { + struct intel_encoder base; + + int type; + const char *tv_format; + int margin[4]; + u32 save_TV_H_CTL_1; + u32 save_TV_H_CTL_2; + u32 save_TV_H_CTL_3; + u32 save_TV_V_CTL_1; + u32 save_TV_V_CTL_2; + u32 save_TV_V_CTL_3; + u32 save_TV_V_CTL_4; + u32 save_TV_V_CTL_5; + u32 save_TV_V_CTL_6; + u32 save_TV_V_CTL_7; + u32 save_TV_SC_CTL_1, save_TV_SC_CTL_2, save_TV_SC_CTL_3; + + u32 save_TV_CSC_Y; + u32 save_TV_CSC_Y2; + u32 save_TV_CSC_U; + u32 save_TV_CSC_U2; + u32 save_TV_CSC_V; + u32 save_TV_CSC_V2; + u32 save_TV_CLR_KNOBS; + u32 save_TV_CLR_LEVEL; + u32 save_TV_WIN_POS; + u32 save_TV_WIN_SIZE; + u32 save_TV_FILTER_CTL_1; + u32 save_TV_FILTER_CTL_2; + u32 save_TV_FILTER_CTL_3; + + u32 save_TV_H_LUMA[60]; + u32 save_TV_H_CHROMA[60]; + u32 save_TV_V_LUMA[43]; + u32 save_TV_V_CHROMA[43]; + + u32 save_TV_DAC; + u32 save_TV_CTL; +}; + +struct video_levels { + int blank, black, burst; +}; + +struct color_conversion { + u16 ry, gy, by, ay; + u16 ru, gu, bu, au; + u16 rv, gv, bv, av; +}; + +static const u32 filter_table[] = { + 0xB1403000, 0x2E203500, 0x35002E20, 0x3000B140, + 0x35A0B160, 0x2DC02E80, 0xB1403480, 0xB1603000, + 0x2EA03640, 0x34002D80, 0x3000B120, 0x36E0B160, + 0x2D202EF0, 0xB1203380, 0xB1603000, 0x2F303780, + 0x33002CC0, 0x3000B100, 0x3820B160, 0x2C802F50, + 0xB10032A0, 0xB1603000, 0x2F9038C0, 0x32202C20, + 0x3000B0E0, 0x3980B160, 0x2BC02FC0, 0xB0E031C0, + 0xB1603000, 0x2FF03A20, 0x31602B60, 0xB020B0C0, + 0x3AE0B160, 0x2B001810, 0xB0C03120, 0xB140B020, + 0x18283BA0, 0x30C02A80, 0xB020B0A0, 0x3C60B140, + 0x2A201838, 0xB0A03080, 0xB120B020, 0x18383D20, + 0x304029C0, 0xB040B080, 0x3DE0B100, 0x29601848, + 0xB0803000, 0xB100B040, 0x18483EC0, 0xB0402900, + 0xB040B060, 0x3F80B0C0, 0x28801858, 0xB060B080, + 0xB0A0B060, 0x18602820, 0xB0A02820, 0x0000B060, + 0xB1403000, 0x2E203500, 0x35002E20, 0x3000B140, + 0x35A0B160, 0x2DC02E80, 0xB1403480, 0xB1603000, + 0x2EA03640, 0x34002D80, 0x3000B120, 0x36E0B160, + 0x2D202EF0, 0xB1203380, 0xB1603000, 0x2F303780, + 0x33002CC0, 0x3000B100, 0x3820B160, 0x2C802F50, + 0xB10032A0, 0xB1603000, 0x2F9038C0, 0x32202C20, + 0x3000B0E0, 0x3980B160, 0x2BC02FC0, 0xB0E031C0, + 0xB1603000, 0x2FF03A20, 0x31602B60, 0xB020B0C0, + 0x3AE0B160, 0x2B001810, 0xB0C03120, 0xB140B020, + 0x18283BA0, 0x30C02A80, 0xB020B0A0, 0x3C60B140, + 0x2A201838, 0xB0A03080, 0xB120B020, 0x18383D20, + 0x304029C0, 0xB040B080, 0x3DE0B100, 0x29601848, + 0xB0803000, 0xB100B040, 0x18483EC0, 0xB0402900, + 0xB040B060, 0x3F80B0C0, 0x28801858, 0xB060B080, + 0xB0A0B060, 0x18602820, 0xB0A02820, 0x0000B060, + 0x36403000, 0x2D002CC0, 0x30003640, 0x2D0036C0, + 0x35C02CC0, 0x37403000, 0x2C802D40, 0x30003540, + 0x2D8037C0, 0x34C02C40, 0x38403000, 0x2BC02E00, + 0x30003440, 0x2E2038C0, 0x34002B80, 0x39803000, + 0x2B402E40, 0x30003380, 0x2E603A00, 0x33402B00, + 0x3A803040, 0x2A802EA0, 0x30403300, 0x2EC03B40, + 0x32802A40, 0x3C003040, 0x2A002EC0, 0x30803240, + 0x2EC03C80, 0x320029C0, 0x3D403080, 0x29402F00, + 0x308031C0, 0x2F203DC0, 0x31802900, 0x3E8030C0, + 0x28802F40, 0x30C03140, 0x2F203F40, 0x31402840, + 0x28003100, 0x28002F00, 0x00003100, 0x36403000, + 0x2D002CC0, 0x30003640, 0x2D0036C0, + 0x35C02CC0, 0x37403000, 0x2C802D40, 0x30003540, + 0x2D8037C0, 0x34C02C40, 0x38403000, 0x2BC02E00, + 0x30003440, 0x2E2038C0, 0x34002B80, 0x39803000, + 0x2B402E40, 0x30003380, 0x2E603A00, 0x33402B00, + 0x3A803040, 0x2A802EA0, 0x30403300, 0x2EC03B40, + 0x32802A40, 0x3C003040, 0x2A002EC0, 0x30803240, + 0x2EC03C80, 0x320029C0, 0x3D403080, 0x29402F00, + 0x308031C0, 0x2F203DC0, 0x31802900, 0x3E8030C0, + 0x28802F40, 0x30C03140, 0x2F203F40, 0x31402840, + 0x28003100, 0x28002F00, 0x00003100, +}; + +/* + * Color conversion values have 3 separate fixed point formats: + * + * 10 bit fields (ay, au) + * 1.9 fixed point (b.bbbbbbbbb) + * 11 bit fields (ry, by, ru, gu, gv) + * exp.mantissa (ee.mmmmmmmmm) + * ee = 00 = 10^-1 (0.mmmmmmmmm) + * ee = 01 = 10^-2 (0.0mmmmmmmmm) + * ee = 10 = 10^-3 (0.00mmmmmmmmm) + * ee = 11 = 10^-4 (0.000mmmmmmmmm) + * 12 bit fields (gy, rv, bu) + * exp.mantissa (eee.mmmmmmmmm) + * eee = 000 = 10^-1 (0.mmmmmmmmm) + * eee = 001 = 10^-2 (0.0mmmmmmmmm) + * eee = 010 = 10^-3 (0.00mmmmmmmmm) + * eee = 011 = 10^-4 (0.000mmmmmmmmm) + * eee = 100 = reserved + * eee = 101 = reserved + * eee = 110 = reserved + * eee = 111 = 10^0 (m.mmmmmmmm) (only usable for 1.0 representation) + * + * Saturation and contrast are 8 bits, with their own representation: + * 8 bit field (saturation, contrast) + * exp.mantissa (ee.mmmmmm) + * ee = 00 = 10^-1 (0.mmmmmm) + * ee = 01 = 10^0 (m.mmmmm) + * ee = 10 = 10^1 (mm.mmmm) + * ee = 11 = 10^2 (mmm.mmm) + * + * Simple conversion function: + * + * static u32 + * float_to_csc_11(float f) + * { + * u32 exp; + * u32 mant; + * u32 ret; + * + * if (f < 0) + * f = -f; + * + * if (f >= 1) { + * exp = 0x7; + * mant = 1 << 8; + * } else { + * for (exp = 0; exp < 3 && f < 0.5; exp++) + * f *= 2.0; + * mant = (f * (1 << 9) + 0.5); + * if (mant >= (1 << 9)) + * mant = (1 << 9) - 1; + * } + * ret = (exp << 9) | mant; + * return ret; + * } + */ + +/* + * Behold, magic numbers! If we plant them they might grow a big + * s-video cable to the sky... or something. + * + * Pre-converted to appropriate hex value. + */ + +/* + * PAL & NTSC values for composite & s-video connections + */ +static const struct color_conversion ntsc_m_csc_composite = { + .ry = 0x0332, .gy = 0x012d, .by = 0x07d3, .ay = 0x0104, + .ru = 0x0733, .gu = 0x052d, .bu = 0x05c7, .au = 0x0200, + .rv = 0x0340, .gv = 0x030c, .bv = 0x06d0, .av = 0x0200, +}; + +static const struct video_levels ntsc_m_levels_composite = { + .blank = 225, .black = 267, .burst = 113, +}; + +static const struct color_conversion ntsc_m_csc_svideo = { + .ry = 0x0332, .gy = 0x012d, .by = 0x07d3, .ay = 0x0133, + .ru = 0x076a, .gu = 0x0564, .bu = 0x030d, .au = 0x0200, + .rv = 0x037a, .gv = 0x033d, .bv = 0x06f6, .av = 0x0200, +}; + +static const struct video_levels ntsc_m_levels_svideo = { + .blank = 266, .black = 316, .burst = 133, +}; + +static const struct color_conversion ntsc_j_csc_composite = { + .ry = 0x0332, .gy = 0x012d, .by = 0x07d3, .ay = 0x0119, + .ru = 0x074c, .gu = 0x0546, .bu = 0x05ec, .au = 0x0200, + .rv = 0x035a, .gv = 0x0322, .bv = 0x06e1, .av = 0x0200, +}; + +static const struct video_levels ntsc_j_levels_composite = { + .blank = 225, .black = 225, .burst = 113, +}; + +static const struct color_conversion ntsc_j_csc_svideo = { + .ry = 0x0332, .gy = 0x012d, .by = 0x07d3, .ay = 0x014c, + .ru = 0x0788, .gu = 0x0581, .bu = 0x0322, .au = 0x0200, + .rv = 0x0399, .gv = 0x0356, .bv = 0x070a, .av = 0x0200, +}; + +static const struct video_levels ntsc_j_levels_svideo = { + .blank = 266, .black = 266, .burst = 133, +}; + +static const struct color_conversion pal_csc_composite = { + .ry = 0x0332, .gy = 0x012d, .by = 0x07d3, .ay = 0x0113, + .ru = 0x0745, .gu = 0x053f, .bu = 0x05e1, .au = 0x0200, + .rv = 0x0353, .gv = 0x031c, .bv = 0x06dc, .av = 0x0200, +}; + +static const struct video_levels pal_levels_composite = { + .blank = 237, .black = 237, .burst = 118, +}; + +static const struct color_conversion pal_csc_svideo = { + .ry = 0x0332, .gy = 0x012d, .by = 0x07d3, .ay = 0x0145, + .ru = 0x0780, .gu = 0x0579, .bu = 0x031c, .au = 0x0200, + .rv = 0x0390, .gv = 0x034f, .bv = 0x0705, .av = 0x0200, +}; + +static const struct video_levels pal_levels_svideo = { + .blank = 280, .black = 280, .burst = 139, +}; + +static const struct color_conversion pal_m_csc_composite = { + .ry = 0x0332, .gy = 0x012d, .by = 0x07d3, .ay = 0x0104, + .ru = 0x0733, .gu = 0x052d, .bu = 0x05c7, .au = 0x0200, + .rv = 0x0340, .gv = 0x030c, .bv = 0x06d0, .av = 0x0200, +}; + +static const struct video_levels pal_m_levels_composite = { + .blank = 225, .black = 267, .burst = 113, +}; + +static const struct color_conversion pal_m_csc_svideo = { + .ry = 0x0332, .gy = 0x012d, .by = 0x07d3, .ay = 0x0133, + .ru = 0x076a, .gu = 0x0564, .bu = 0x030d, .au = 0x0200, + .rv = 0x037a, .gv = 0x033d, .bv = 0x06f6, .av = 0x0200, +}; + +static const struct video_levels pal_m_levels_svideo = { + .blank = 266, .black = 316, .burst = 133, +}; + +static const struct color_conversion pal_n_csc_composite = { + .ry = 0x0332, .gy = 0x012d, .by = 0x07d3, .ay = 0x0104, + .ru = 0x0733, .gu = 0x052d, .bu = 0x05c7, .au = 0x0200, + .rv = 0x0340, .gv = 0x030c, .bv = 0x06d0, .av = 0x0200, +}; + +static const struct video_levels pal_n_levels_composite = { + .blank = 225, .black = 267, .burst = 118, +}; + +static const struct color_conversion pal_n_csc_svideo = { + .ry = 0x0332, .gy = 0x012d, .by = 0x07d3, .ay = 0x0133, + .ru = 0x076a, .gu = 0x0564, .bu = 0x030d, .au = 0x0200, + .rv = 0x037a, .gv = 0x033d, .bv = 0x06f6, .av = 0x0200, +}; + +static const struct video_levels pal_n_levels_svideo = { + .blank = 266, .black = 316, .burst = 139, +}; + +/* + * Component connections + */ +static const struct color_conversion sdtv_csc_yprpb = { + .ry = 0x0332, .gy = 0x012d, .by = 0x07d3, .ay = 0x0145, + .ru = 0x0559, .gu = 0x0353, .bu = 0x0100, .au = 0x0200, + .rv = 0x0100, .gv = 0x03ad, .bv = 0x074d, .av = 0x0200, +}; + +static const struct color_conversion sdtv_csc_rgb = { + .ry = 0x0000, .gy = 0x0f00, .by = 0x0000, .ay = 0x0166, + .ru = 0x0000, .gu = 0x0000, .bu = 0x0f00, .au = 0x0166, + .rv = 0x0f00, .gv = 0x0000, .bv = 0x0000, .av = 0x0166, +}; + +static const struct color_conversion hdtv_csc_yprpb = { + .ry = 0x05b3, .gy = 0x016e, .by = 0x0728, .ay = 0x0145, + .ru = 0x07d5, .gu = 0x038b, .bu = 0x0100, .au = 0x0200, + .rv = 0x0100, .gv = 0x03d1, .bv = 0x06bc, .av = 0x0200, +}; + +static const struct color_conversion hdtv_csc_rgb = { + .ry = 0x0000, .gy = 0x0f00, .by = 0x0000, .ay = 0x0166, + .ru = 0x0000, .gu = 0x0000, .bu = 0x0f00, .au = 0x0166, + .rv = 0x0f00, .gv = 0x0000, .bv = 0x0000, .av = 0x0166, +}; + +static const struct video_levels component_levels = { + .blank = 279, .black = 279, .burst = 0, +}; + + +struct tv_mode { + const char *name; + int clock; + int refresh; /* in millihertz (for precision) */ + u32 oversample; + int hsync_end, hblank_start, hblank_end, htotal; + bool progressive, trilevel_sync, component_only; + int vsync_start_f1, vsync_start_f2, vsync_len; + bool veq_ena; + int veq_start_f1, veq_start_f2, veq_len; + int vi_end_f1, vi_end_f2, nbr_end; + bool burst_ena; + int hburst_start, hburst_len; + int vburst_start_f1, vburst_end_f1; + int vburst_start_f2, vburst_end_f2; + int vburst_start_f3, vburst_end_f3; + int vburst_start_f4, vburst_end_f4; + /* + * subcarrier programming + */ + int dda2_size, dda3_size, dda1_inc, dda2_inc, dda3_inc; + u32 sc_reset; + bool pal_burst; + /* + * blank/black levels + */ + const struct video_levels *composite_levels, *svideo_levels; + const struct color_conversion *composite_color, *svideo_color; + const u32 *filter_table; + int max_srcw; +}; + + +/* + * Sub carrier DDA + * + * I think this works as follows: + * + * subcarrier freq = pixel_clock * (dda1_inc + dda2_inc / dda2_size) / 4096 + * + * Presumably, when dda3 is added in, it gets to adjust the dda2_inc value + * + * So, + * dda1_ideal = subcarrier/pixel * 4096 + * dda1_inc = floor (dda1_ideal) + * dda2 = dda1_ideal - dda1_inc + * + * then pick a ratio for dda2 that gives the closest approximation. If + * you can't get close enough, you can play with dda3 as well. This + * seems likely to happen when dda2 is small as the jumps would be larger + * + * To invert this, + * + * pixel_clock = subcarrier * 4096 / (dda1_inc + dda2_inc / dda2_size) + * + * The constants below were all computed using a 107.520MHz clock + */ + +/** + * Register programming values for TV modes. + * + * These values account for -1s required. + */ + +static const struct tv_mode tv_modes[] = { + { + .name = "NTSC-M", + .clock = 108000, + .refresh = 59940, + .oversample = TV_OVERSAMPLE_8X, + .component_only = 0, + /* 525 Lines, 60 Fields, 15.734KHz line, Sub-Carrier 3.580MHz */ + + .hsync_end = 64, .hblank_end = 124, + .hblank_start = 836, .htotal = 857, + + .progressive = false, .trilevel_sync = false, + + .vsync_start_f1 = 6, .vsync_start_f2 = 7, + .vsync_len = 6, + + .veq_ena = true, .veq_start_f1 = 0, + .veq_start_f2 = 1, .veq_len = 18, + + .vi_end_f1 = 20, .vi_end_f2 = 21, + .nbr_end = 240, + + .burst_ena = true, + .hburst_start = 72, .hburst_len = 34, + .vburst_start_f1 = 9, .vburst_end_f1 = 240, + .vburst_start_f2 = 10, .vburst_end_f2 = 240, + .vburst_start_f3 = 9, .vburst_end_f3 = 240, + .vburst_start_f4 = 10, .vburst_end_f4 = 240, + + /* desired 3.5800000 actual 3.5800000 clock 107.52 */ + .dda1_inc = 135, + .dda2_inc = 20800, .dda2_size = 27456, + .dda3_inc = 0, .dda3_size = 0, + .sc_reset = TV_SC_RESET_EVERY_4, + .pal_burst = false, + + .composite_levels = &ntsc_m_levels_composite, + .composite_color = &ntsc_m_csc_composite, + .svideo_levels = &ntsc_m_levels_svideo, + .svideo_color = &ntsc_m_csc_svideo, + + .filter_table = filter_table, + }, + { + .name = "NTSC-443", + .clock = 108000, + .refresh = 59940, + .oversample = TV_OVERSAMPLE_8X, + .component_only = 0, + /* 525 Lines, 60 Fields, 15.734KHz line, Sub-Carrier 4.43MHz */ + .hsync_end = 64, .hblank_end = 124, + .hblank_start = 836, .htotal = 857, + + .progressive = false, .trilevel_sync = false, + + .vsync_start_f1 = 6, .vsync_start_f2 = 7, + .vsync_len = 6, + + .veq_ena = true, .veq_start_f1 = 0, + .veq_start_f2 = 1, .veq_len = 18, + + .vi_end_f1 = 20, .vi_end_f2 = 21, + .nbr_end = 240, + + .burst_ena = true, + .hburst_start = 72, .hburst_len = 34, + .vburst_start_f1 = 9, .vburst_end_f1 = 240, + .vburst_start_f2 = 10, .vburst_end_f2 = 240, + .vburst_start_f3 = 9, .vburst_end_f3 = 240, + .vburst_start_f4 = 10, .vburst_end_f4 = 240, + + /* desired 4.4336180 actual 4.4336180 clock 107.52 */ + .dda1_inc = 168, + .dda2_inc = 4093, .dda2_size = 27456, + .dda3_inc = 310, .dda3_size = 525, + .sc_reset = TV_SC_RESET_NEVER, + .pal_burst = false, + + .composite_levels = &ntsc_m_levels_composite, + .composite_color = &ntsc_m_csc_composite, + .svideo_levels = &ntsc_m_levels_svideo, + .svideo_color = &ntsc_m_csc_svideo, + + .filter_table = filter_table, + }, + { + .name = "NTSC-J", + .clock = 108000, + .refresh = 59940, + .oversample = TV_OVERSAMPLE_8X, + .component_only = 0, + + /* 525 Lines, 60 Fields, 15.734KHz line, Sub-Carrier 3.580MHz */ + .hsync_end = 64, .hblank_end = 124, + .hblank_start = 836, .htotal = 857, + + .progressive = false, .trilevel_sync = false, + + .vsync_start_f1 = 6, .vsync_start_f2 = 7, + .vsync_len = 6, + + .veq_ena = true, .veq_start_f1 = 0, + .veq_start_f2 = 1, .veq_len = 18, + + .vi_end_f1 = 20, .vi_end_f2 = 21, + .nbr_end = 240, + + .burst_ena = true, + .hburst_start = 72, .hburst_len = 34, + .vburst_start_f1 = 9, .vburst_end_f1 = 240, + .vburst_start_f2 = 10, .vburst_end_f2 = 240, + .vburst_start_f3 = 9, .vburst_end_f3 = 240, + .vburst_start_f4 = 10, .vburst_end_f4 = 240, + + /* desired 3.5800000 actual 3.5800000 clock 107.52 */ + .dda1_inc = 135, + .dda2_inc = 20800, .dda2_size = 27456, + .dda3_inc = 0, .dda3_size = 0, + .sc_reset = TV_SC_RESET_EVERY_4, + .pal_burst = false, + + .composite_levels = &ntsc_j_levels_composite, + .composite_color = &ntsc_j_csc_composite, + .svideo_levels = &ntsc_j_levels_svideo, + .svideo_color = &ntsc_j_csc_svideo, + + .filter_table = filter_table, + }, + { + .name = "PAL-M", + .clock = 108000, + .refresh = 59940, + .oversample = TV_OVERSAMPLE_8X, + .component_only = 0, + + /* 525 Lines, 60 Fields, 15.734KHz line, Sub-Carrier 3.580MHz */ + .hsync_end = 64, .hblank_end = 124, + .hblank_start = 836, .htotal = 857, + + .progressive = false, .trilevel_sync = false, + + .vsync_start_f1 = 6, .vsync_start_f2 = 7, + .vsync_len = 6, + + .veq_ena = true, .veq_start_f1 = 0, + .veq_start_f2 = 1, .veq_len = 18, + + .vi_end_f1 = 20, .vi_end_f2 = 21, + .nbr_end = 240, + + .burst_ena = true, + .hburst_start = 72, .hburst_len = 34, + .vburst_start_f1 = 9, .vburst_end_f1 = 240, + .vburst_start_f2 = 10, .vburst_end_f2 = 240, + .vburst_start_f3 = 9, .vburst_end_f3 = 240, + .vburst_start_f4 = 10, .vburst_end_f4 = 240, + + /* desired 3.5800000 actual 3.5800000 clock 107.52 */ + .dda1_inc = 135, + .dda2_inc = 16704, .dda2_size = 27456, + .dda3_inc = 0, .dda3_size = 0, + .sc_reset = TV_SC_RESET_EVERY_8, + .pal_burst = true, + + .composite_levels = &pal_m_levels_composite, + .composite_color = &pal_m_csc_composite, + .svideo_levels = &pal_m_levels_svideo, + .svideo_color = &pal_m_csc_svideo, + + .filter_table = filter_table, + }, + { + /* 625 Lines, 50 Fields, 15.625KHz line, Sub-Carrier 4.434MHz */ + .name = "PAL-N", + .clock = 108000, + .refresh = 50000, + .oversample = TV_OVERSAMPLE_8X, + .component_only = 0, + + .hsync_end = 64, .hblank_end = 128, + .hblank_start = 844, .htotal = 863, + + .progressive = false, .trilevel_sync = false, + + + .vsync_start_f1 = 6, .vsync_start_f2 = 7, + .vsync_len = 6, + + .veq_ena = true, .veq_start_f1 = 0, + .veq_start_f2 = 1, .veq_len = 18, + + .vi_end_f1 = 24, .vi_end_f2 = 25, + .nbr_end = 286, + + .burst_ena = true, + .hburst_start = 73, .hburst_len = 34, + .vburst_start_f1 = 8, .vburst_end_f1 = 285, + .vburst_start_f2 = 8, .vburst_end_f2 = 286, + .vburst_start_f3 = 9, .vburst_end_f3 = 286, + .vburst_start_f4 = 9, .vburst_end_f4 = 285, + + + /* desired 4.4336180 actual 4.4336180 clock 107.52 */ + .dda1_inc = 135, + .dda2_inc = 23578, .dda2_size = 27648, + .dda3_inc = 134, .dda3_size = 625, + .sc_reset = TV_SC_RESET_EVERY_8, + .pal_burst = true, + + .composite_levels = &pal_n_levels_composite, + .composite_color = &pal_n_csc_composite, + .svideo_levels = &pal_n_levels_svideo, + .svideo_color = &pal_n_csc_svideo, + + .filter_table = filter_table, + }, + { + /* 625 Lines, 50 Fields, 15.625KHz line, Sub-Carrier 4.434MHz */ + .name = "PAL", + .clock = 108000, + .refresh = 50000, + .oversample = TV_OVERSAMPLE_8X, + .component_only = 0, + + .hsync_end = 64, .hblank_end = 142, + .hblank_start = 844, .htotal = 863, + + .progressive = false, .trilevel_sync = false, + + .vsync_start_f1 = 5, .vsync_start_f2 = 6, + .vsync_len = 5, + + .veq_ena = true, .veq_start_f1 = 0, + .veq_start_f2 = 1, .veq_len = 15, + + .vi_end_f1 = 24, .vi_end_f2 = 25, + .nbr_end = 286, + + .burst_ena = true, + .hburst_start = 73, .hburst_len = 32, + .vburst_start_f1 = 8, .vburst_end_f1 = 285, + .vburst_start_f2 = 8, .vburst_end_f2 = 286, + .vburst_start_f3 = 9, .vburst_end_f3 = 286, + .vburst_start_f4 = 9, .vburst_end_f4 = 285, + + /* desired 4.4336180 actual 4.4336180 clock 107.52 */ + .dda1_inc = 168, + .dda2_inc = 4122, .dda2_size = 27648, + .dda3_inc = 67, .dda3_size = 625, + .sc_reset = TV_SC_RESET_EVERY_8, + .pal_burst = true, + + .composite_levels = &pal_levels_composite, + .composite_color = &pal_csc_composite, + .svideo_levels = &pal_levels_svideo, + .svideo_color = &pal_csc_svideo, + + .filter_table = filter_table, + }, + { + .name = "480p@59.94Hz", + .clock = 107520, + .refresh = 59940, + .oversample = TV_OVERSAMPLE_4X, + .component_only = 1, + + .hsync_end = 64, .hblank_end = 122, + .hblank_start = 842, .htotal = 857, + + .progressive = true,.trilevel_sync = false, + + .vsync_start_f1 = 12, .vsync_start_f2 = 12, + .vsync_len = 12, + + .veq_ena = false, + + .vi_end_f1 = 44, .vi_end_f2 = 44, + .nbr_end = 479, + + .burst_ena = false, + + .filter_table = filter_table, + }, + { + .name = "480p@60Hz", + .clock = 107520, + .refresh = 60000, + .oversample = TV_OVERSAMPLE_4X, + .component_only = 1, + + .hsync_end = 64, .hblank_end = 122, + .hblank_start = 842, .htotal = 856, + + .progressive = true,.trilevel_sync = false, + + .vsync_start_f1 = 12, .vsync_start_f2 = 12, + .vsync_len = 12, + + .veq_ena = false, + + .vi_end_f1 = 44, .vi_end_f2 = 44, + .nbr_end = 479, + + .burst_ena = false, + + .filter_table = filter_table, + }, + { + .name = "576p", + .clock = 107520, + .refresh = 50000, + .oversample = TV_OVERSAMPLE_4X, + .component_only = 1, + + .hsync_end = 64, .hblank_end = 139, + .hblank_start = 859, .htotal = 863, + + .progressive = true, .trilevel_sync = false, + + .vsync_start_f1 = 10, .vsync_start_f2 = 10, + .vsync_len = 10, + + .veq_ena = false, + + .vi_end_f1 = 48, .vi_end_f2 = 48, + .nbr_end = 575, + + .burst_ena = false, + + .filter_table = filter_table, + }, + { + .name = "720p@60Hz", + .clock = 148800, + .refresh = 60000, + .oversample = TV_OVERSAMPLE_2X, + .component_only = 1, + + .hsync_end = 80, .hblank_end = 300, + .hblank_start = 1580, .htotal = 1649, + + .progressive = true, .trilevel_sync = true, + + .vsync_start_f1 = 10, .vsync_start_f2 = 10, + .vsync_len = 10, + + .veq_ena = false, + + .vi_end_f1 = 29, .vi_end_f2 = 29, + .nbr_end = 719, + + .burst_ena = false, + + .filter_table = filter_table, + }, + { + .name = "720p@59.94Hz", + .clock = 148800, + .refresh = 59940, + .oversample = TV_OVERSAMPLE_2X, + .component_only = 1, + + .hsync_end = 80, .hblank_end = 300, + .hblank_start = 1580, .htotal = 1651, + + .progressive = true, .trilevel_sync = true, + + .vsync_start_f1 = 10, .vsync_start_f2 = 10, + .vsync_len = 10, + + .veq_ena = false, + + .vi_end_f1 = 29, .vi_end_f2 = 29, + .nbr_end = 719, + + .burst_ena = false, + + .filter_table = filter_table, + }, + { + .name = "720p@50Hz", + .clock = 148800, + .refresh = 50000, + .oversample = TV_OVERSAMPLE_2X, + .component_only = 1, + + .hsync_end = 80, .hblank_end = 300, + .hblank_start = 1580, .htotal = 1979, + + .progressive = true, .trilevel_sync = true, + + .vsync_start_f1 = 10, .vsync_start_f2 = 10, + .vsync_len = 10, + + .veq_ena = false, + + .vi_end_f1 = 29, .vi_end_f2 = 29, + .nbr_end = 719, + + .burst_ena = false, + + .filter_table = filter_table, + .max_srcw = 800 + }, + { + .name = "1080i@50Hz", + .clock = 148800, + .refresh = 50000, + .oversample = TV_OVERSAMPLE_2X, + .component_only = 1, + + .hsync_end = 88, .hblank_end = 235, + .hblank_start = 2155, .htotal = 2639, + + .progressive = false, .trilevel_sync = true, + + .vsync_start_f1 = 4, .vsync_start_f2 = 5, + .vsync_len = 10, + + .veq_ena = true, .veq_start_f1 = 4, + .veq_start_f2 = 4, .veq_len = 10, + + + .vi_end_f1 = 21, .vi_end_f2 = 22, + .nbr_end = 539, + + .burst_ena = false, + + .filter_table = filter_table, + }, + { + .name = "1080i@60Hz", + .clock = 148800, + .refresh = 60000, + .oversample = TV_OVERSAMPLE_2X, + .component_only = 1, + + .hsync_end = 88, .hblank_end = 235, + .hblank_start = 2155, .htotal = 2199, + + .progressive = false, .trilevel_sync = true, + + .vsync_start_f1 = 4, .vsync_start_f2 = 5, + .vsync_len = 10, + + .veq_ena = true, .veq_start_f1 = 4, + .veq_start_f2 = 4, .veq_len = 10, + + + .vi_end_f1 = 21, .vi_end_f2 = 22, + .nbr_end = 539, + + .burst_ena = false, + + .filter_table = filter_table, + }, + { + .name = "1080i@59.94Hz", + .clock = 148800, + .refresh = 29970, + .oversample = TV_OVERSAMPLE_2X, + .component_only = 1, + + .hsync_end = 88, .hblank_end = 235, + .hblank_start = 2155, .htotal = 2201, + + .progressive = false, .trilevel_sync = true, + + .vsync_start_f1 = 4, .vsync_start_f2 = 5, + .vsync_len = 10, + + .veq_ena = true, .veq_start_f1 = 4, + .veq_start_f2 = 4, .veq_len = 10, + + + .vi_end_f1 = 21, .vi_end_f2 = 22, + .nbr_end = 539, + + .burst_ena = false, + + .filter_table = filter_table, + }, +}; + +static struct intel_tv *enc_to_intel_tv(struct drm_encoder *encoder) +{ + return container_of(encoder, struct intel_tv, base.base); +} + +static struct intel_tv *intel_attached_tv(struct drm_connector *connector) +{ + return container_of(intel_attached_encoder(connector), + struct intel_tv, + base); +} + +static void +intel_tv_dpms(struct drm_encoder *encoder, int mode) +{ + struct drm_device *dev = encoder->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + + switch(mode) { + case DRM_MODE_DPMS_ON: + I915_WRITE(TV_CTL, I915_READ(TV_CTL) | TV_ENC_ENABLE); + break; + case DRM_MODE_DPMS_STANDBY: + case DRM_MODE_DPMS_SUSPEND: + case DRM_MODE_DPMS_OFF: + I915_WRITE(TV_CTL, I915_READ(TV_CTL) & ~TV_ENC_ENABLE); + break; + } +} + +static const struct tv_mode * +intel_tv_mode_lookup(const char *tv_format) +{ + int i; + + for (i = 0; i < sizeof(tv_modes) / sizeof (tv_modes[0]); i++) { + const struct tv_mode *tv_mode = &tv_modes[i]; + + if (!strcmp(tv_format, tv_mode->name)) + return tv_mode; + } + return NULL; +} + +static const struct tv_mode * +intel_tv_mode_find(struct intel_tv *intel_tv) +{ + return intel_tv_mode_lookup(intel_tv->tv_format); +} + +static enum drm_mode_status +intel_tv_mode_valid(struct drm_connector *connector, + struct drm_display_mode *mode) +{ + struct intel_tv *intel_tv = intel_attached_tv(connector); + const struct tv_mode *tv_mode = intel_tv_mode_find(intel_tv); + + /* Ensure TV refresh is close to desired refresh */ + if (tv_mode && abs(tv_mode->refresh - drm_mode_vrefresh(mode) * 1000) + < 1000) + return MODE_OK; + + return MODE_CLOCK_RANGE; +} + + +static bool +intel_tv_mode_fixup(struct drm_encoder *encoder, struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct drm_device *dev = encoder->dev; + struct drm_mode_config *drm_config = &dev->mode_config; + struct intel_tv *intel_tv = enc_to_intel_tv(encoder); + const struct tv_mode *tv_mode = intel_tv_mode_find(intel_tv); + struct drm_encoder *other_encoder; + + if (!tv_mode) + return false; + + /* FIXME: lock encoder list */ + list_for_each_entry(other_encoder, &drm_config->encoder_list, head) { + if (other_encoder != encoder && + other_encoder->crtc == encoder->crtc) + return false; + } + + adjusted_mode->clock = tv_mode->clock; + return true; +} + +static void +intel_tv_mode_set(struct drm_encoder *encoder, struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct drm_device *dev = encoder->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_crtc *crtc = encoder->crtc; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + struct intel_tv *intel_tv = enc_to_intel_tv(encoder); + const struct tv_mode *tv_mode = intel_tv_mode_find(intel_tv); + u32 tv_ctl; + u32 hctl1, hctl2, hctl3; + u32 vctl1, vctl2, vctl3, vctl4, vctl5, vctl6, vctl7; + u32 scctl1, scctl2, scctl3; + int i, j; + const struct video_levels *video_levels; + const struct color_conversion *color_conversion; + bool burst_ena; + int pipe = intel_crtc->pipe; + + if (!tv_mode) + return; /* can't happen (mode_prepare prevents this) */ + + tv_ctl = I915_READ(TV_CTL); + tv_ctl &= TV_CTL_SAVE; + + switch (intel_tv->type) { + default: + case DRM_MODE_CONNECTOR_Unknown: + case DRM_MODE_CONNECTOR_Composite: + tv_ctl |= TV_ENC_OUTPUT_COMPOSITE; + video_levels = tv_mode->composite_levels; + color_conversion = tv_mode->composite_color; + burst_ena = tv_mode->burst_ena; + break; + case DRM_MODE_CONNECTOR_Component: + tv_ctl |= TV_ENC_OUTPUT_COMPONENT; + video_levels = &component_levels; + if (tv_mode->burst_ena) + color_conversion = &sdtv_csc_yprpb; + else + color_conversion = &hdtv_csc_yprpb; + burst_ena = false; + break; + case DRM_MODE_CONNECTOR_SVIDEO: + tv_ctl |= TV_ENC_OUTPUT_SVIDEO; + video_levels = tv_mode->svideo_levels; + color_conversion = tv_mode->svideo_color; + burst_ena = tv_mode->burst_ena; + break; + } + hctl1 = (tv_mode->hsync_end << TV_HSYNC_END_SHIFT) | + (tv_mode->htotal << TV_HTOTAL_SHIFT); + + hctl2 = (tv_mode->hburst_start << 16) | + (tv_mode->hburst_len << TV_HBURST_LEN_SHIFT); + + if (burst_ena) + hctl2 |= TV_BURST_ENA; + + hctl3 = (tv_mode->hblank_start << TV_HBLANK_START_SHIFT) | + (tv_mode->hblank_end << TV_HBLANK_END_SHIFT); + + vctl1 = (tv_mode->nbr_end << TV_NBR_END_SHIFT) | + (tv_mode->vi_end_f1 << TV_VI_END_F1_SHIFT) | + (tv_mode->vi_end_f2 << TV_VI_END_F2_SHIFT); + + vctl2 = (tv_mode->vsync_len << TV_VSYNC_LEN_SHIFT) | + (tv_mode->vsync_start_f1 << TV_VSYNC_START_F1_SHIFT) | + (tv_mode->vsync_start_f2 << TV_VSYNC_START_F2_SHIFT); + + vctl3 = (tv_mode->veq_len << TV_VEQ_LEN_SHIFT) | + (tv_mode->veq_start_f1 << TV_VEQ_START_F1_SHIFT) | + (tv_mode->veq_start_f2 << TV_VEQ_START_F2_SHIFT); + + if (tv_mode->veq_ena) + vctl3 |= TV_EQUAL_ENA; + + vctl4 = (tv_mode->vburst_start_f1 << TV_VBURST_START_F1_SHIFT) | + (tv_mode->vburst_end_f1 << TV_VBURST_END_F1_SHIFT); + + vctl5 = (tv_mode->vburst_start_f2 << TV_VBURST_START_F2_SHIFT) | + (tv_mode->vburst_end_f2 << TV_VBURST_END_F2_SHIFT); + + vctl6 = (tv_mode->vburst_start_f3 << TV_VBURST_START_F3_SHIFT) | + (tv_mode->vburst_end_f3 << TV_VBURST_END_F3_SHIFT); + + vctl7 = (tv_mode->vburst_start_f4 << TV_VBURST_START_F4_SHIFT) | + (tv_mode->vburst_end_f4 << TV_VBURST_END_F4_SHIFT); + + if (intel_crtc->pipe == 1) + tv_ctl |= TV_ENC_PIPEB_SELECT; + tv_ctl |= tv_mode->oversample; + + if (tv_mode->progressive) + tv_ctl |= TV_PROGRESSIVE; + if (tv_mode->trilevel_sync) + tv_ctl |= TV_TRILEVEL_SYNC; + if (tv_mode->pal_burst) + tv_ctl |= TV_PAL_BURST; + + scctl1 = 0; + if (tv_mode->dda1_inc) + scctl1 |= TV_SC_DDA1_EN; + if (tv_mode->dda2_inc) + scctl1 |= TV_SC_DDA2_EN; + if (tv_mode->dda3_inc) + scctl1 |= TV_SC_DDA3_EN; + scctl1 |= tv_mode->sc_reset; + if (video_levels) + scctl1 |= video_levels->burst << TV_BURST_LEVEL_SHIFT; + scctl1 |= tv_mode->dda1_inc << TV_SCDDA1_INC_SHIFT; + + scctl2 = tv_mode->dda2_size << TV_SCDDA2_SIZE_SHIFT | + tv_mode->dda2_inc << TV_SCDDA2_INC_SHIFT; + + scctl3 = tv_mode->dda3_size << TV_SCDDA3_SIZE_SHIFT | + tv_mode->dda3_inc << TV_SCDDA3_INC_SHIFT; + + /* Enable two fixes for the chips that need them. */ + if (dev->pci_device < 0x2772) + tv_ctl |= TV_ENC_C0_FIX | TV_ENC_SDP_FIX; + + I915_WRITE(TV_H_CTL_1, hctl1); + I915_WRITE(TV_H_CTL_2, hctl2); + I915_WRITE(TV_H_CTL_3, hctl3); + I915_WRITE(TV_V_CTL_1, vctl1); + I915_WRITE(TV_V_CTL_2, vctl2); + I915_WRITE(TV_V_CTL_3, vctl3); + I915_WRITE(TV_V_CTL_4, vctl4); + I915_WRITE(TV_V_CTL_5, vctl5); + I915_WRITE(TV_V_CTL_6, vctl6); + I915_WRITE(TV_V_CTL_7, vctl7); + I915_WRITE(TV_SC_CTL_1, scctl1); + I915_WRITE(TV_SC_CTL_2, scctl2); + I915_WRITE(TV_SC_CTL_3, scctl3); + + if (color_conversion) { + I915_WRITE(TV_CSC_Y, (color_conversion->ry << 16) | + color_conversion->gy); + I915_WRITE(TV_CSC_Y2,(color_conversion->by << 16) | + color_conversion->ay); + I915_WRITE(TV_CSC_U, (color_conversion->ru << 16) | + color_conversion->gu); + I915_WRITE(TV_CSC_U2, (color_conversion->bu << 16) | + color_conversion->au); + I915_WRITE(TV_CSC_V, (color_conversion->rv << 16) | + color_conversion->gv); + I915_WRITE(TV_CSC_V2, (color_conversion->bv << 16) | + color_conversion->av); + } + + if (INTEL_INFO(dev)->gen >= 4) + I915_WRITE(TV_CLR_KNOBS, 0x00404000); + else + I915_WRITE(TV_CLR_KNOBS, 0x00606000); + + if (video_levels) + I915_WRITE(TV_CLR_LEVEL, + ((video_levels->black << TV_BLACK_LEVEL_SHIFT) | + (video_levels->blank << TV_BLANK_LEVEL_SHIFT))); + { + int pipeconf_reg = PIPECONF(pipe); + int dspcntr_reg = DSPCNTR(intel_crtc->plane); + int pipeconf = I915_READ(pipeconf_reg); + int dspcntr = I915_READ(dspcntr_reg); + int dspbase_reg = DSPADDR(intel_crtc->plane); + int xpos = 0x0, ypos = 0x0; + unsigned int xsize, ysize; + /* Pipe must be off here */ + I915_WRITE(dspcntr_reg, dspcntr & ~DISPLAY_PLANE_ENABLE); + /* Flush the plane changes */ + I915_WRITE(dspbase_reg, I915_READ(dspbase_reg)); + + /* Wait for vblank for the disable to take effect */ + if (IS_GEN2(dev)) + intel_wait_for_vblank(dev, intel_crtc->pipe); + + I915_WRITE(pipeconf_reg, pipeconf & ~PIPECONF_ENABLE); + /* Wait for vblank for the disable to take effect. */ + intel_wait_for_pipe_off(dev, intel_crtc->pipe); + + /* Filter ctl must be set before TV_WIN_SIZE */ + I915_WRITE(TV_FILTER_CTL_1, TV_AUTO_SCALE); + xsize = tv_mode->hblank_start - tv_mode->hblank_end; + if (tv_mode->progressive) + ysize = tv_mode->nbr_end + 1; + else + ysize = 2*tv_mode->nbr_end + 1; + + xpos += intel_tv->margin[TV_MARGIN_LEFT]; + ypos += intel_tv->margin[TV_MARGIN_TOP]; + xsize -= (intel_tv->margin[TV_MARGIN_LEFT] + + intel_tv->margin[TV_MARGIN_RIGHT]); + ysize -= (intel_tv->margin[TV_MARGIN_TOP] + + intel_tv->margin[TV_MARGIN_BOTTOM]); + I915_WRITE(TV_WIN_POS, (xpos<<16)|ypos); + I915_WRITE(TV_WIN_SIZE, (xsize<<16)|ysize); + + I915_WRITE(pipeconf_reg, pipeconf); + I915_WRITE(dspcntr_reg, dspcntr); + /* Flush the plane changes */ + I915_WRITE(dspbase_reg, I915_READ(dspbase_reg)); + } + + j = 0; + for (i = 0; i < 60; i++) + I915_WRITE(TV_H_LUMA_0 + (i<<2), tv_mode->filter_table[j++]); + for (i = 0; i < 60; i++) + I915_WRITE(TV_H_CHROMA_0 + (i<<2), tv_mode->filter_table[j++]); + for (i = 0; i < 43; i++) + I915_WRITE(TV_V_LUMA_0 + (i<<2), tv_mode->filter_table[j++]); + for (i = 0; i < 43; i++) + I915_WRITE(TV_V_CHROMA_0 + (i<<2), tv_mode->filter_table[j++]); + I915_WRITE(TV_DAC, I915_READ(TV_DAC) & TV_DAC_SAVE); + I915_WRITE(TV_CTL, tv_ctl); +} + +static const struct drm_display_mode reported_modes[] = { + { + .name = "NTSC 480i", + .clock = 107520, + .hdisplay = 1280, + .hsync_start = 1368, + .hsync_end = 1496, + .htotal = 1712, + + .vdisplay = 1024, + .vsync_start = 1027, + .vsync_end = 1034, + .vtotal = 1104, + .type = DRM_MODE_TYPE_DRIVER, + }, +}; + +/** + * Detects TV presence by checking for load. + * + * Requires that the current pipe's DPLL is active. + + * \return true if TV is connected. + * \return false if TV is disconnected. + */ +static int +intel_tv_detect_type (struct intel_tv *intel_tv, + struct drm_connector *connector) +{ + struct drm_encoder *encoder = &intel_tv->base.base; + struct drm_device *dev = encoder->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + unsigned long irqflags; + u32 tv_ctl, save_tv_ctl; + u32 tv_dac, save_tv_dac; + int type; + + /* Disable TV interrupts around load detect or we'll recurse */ + if (connector->polled & DRM_CONNECTOR_POLL_HPD) { + spin_lock_irqsave(&dev_priv->irq_lock, irqflags); + i915_disable_pipestat(dev_priv, 0, + PIPE_HOTPLUG_INTERRUPT_ENABLE | + PIPE_HOTPLUG_TV_INTERRUPT_ENABLE); + spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags); + } + + save_tv_dac = tv_dac = I915_READ(TV_DAC); + save_tv_ctl = tv_ctl = I915_READ(TV_CTL); + + /* Poll for TV detection */ + tv_ctl &= ~(TV_ENC_ENABLE | TV_TEST_MODE_MASK); + tv_ctl |= TV_TEST_MODE_MONITOR_DETECT; + + tv_dac &= ~(TVDAC_SENSE_MASK | DAC_A_MASK | DAC_B_MASK | DAC_C_MASK); + tv_dac |= (TVDAC_STATE_CHG_EN | + TVDAC_A_SENSE_CTL | + TVDAC_B_SENSE_CTL | + TVDAC_C_SENSE_CTL | + DAC_CTL_OVERRIDE | + DAC_A_0_7_V | + DAC_B_0_7_V | + DAC_C_0_7_V); + + I915_WRITE(TV_CTL, tv_ctl); + I915_WRITE(TV_DAC, tv_dac); + POSTING_READ(TV_DAC); + + intel_wait_for_vblank(intel_tv->base.base.dev, + to_intel_crtc(intel_tv->base.base.crtc)->pipe); + + type = -1; + if (wait_for((tv_dac = I915_READ(TV_DAC)) & TVDAC_STATE_CHG, 20) == 0) { + DRM_DEBUG_KMS("TV detected: %x, %x\n", tv_ctl, tv_dac); + /* + * A B C + * 0 1 1 Composite + * 1 0 X svideo + * 0 0 0 Component + */ + if ((tv_dac & TVDAC_SENSE_MASK) == (TVDAC_B_SENSE | TVDAC_C_SENSE)) { + DRM_DEBUG_KMS("Detected Composite TV connection\n"); + type = DRM_MODE_CONNECTOR_Composite; + } else if ((tv_dac & (TVDAC_A_SENSE|TVDAC_B_SENSE)) == TVDAC_A_SENSE) { + DRM_DEBUG_KMS("Detected S-Video TV connection\n"); + type = DRM_MODE_CONNECTOR_SVIDEO; + } else if ((tv_dac & TVDAC_SENSE_MASK) == 0) { + DRM_DEBUG_KMS("Detected Component TV connection\n"); + type = DRM_MODE_CONNECTOR_Component; + } else { + DRM_DEBUG_KMS("Unrecognised TV connection\n"); + } + } + + I915_WRITE(TV_DAC, save_tv_dac & ~TVDAC_STATE_CHG_EN); + I915_WRITE(TV_CTL, save_tv_ctl); + POSTING_READ(TV_CTL); + + /* For unknown reasons the hw barfs if we don't do this vblank wait. */ + intel_wait_for_vblank(intel_tv->base.base.dev, + to_intel_crtc(intel_tv->base.base.crtc)->pipe); + + /* Restore interrupt config */ + if (connector->polled & DRM_CONNECTOR_POLL_HPD) { + spin_lock_irqsave(&dev_priv->irq_lock, irqflags); + i915_enable_pipestat(dev_priv, 0, + PIPE_HOTPLUG_INTERRUPT_ENABLE | + PIPE_HOTPLUG_TV_INTERRUPT_ENABLE); + spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags); + } + + return type; +} + +/* + * Here we set accurate tv format according to connector type + * i.e Component TV should not be assigned by NTSC or PAL + */ +static void intel_tv_find_better_format(struct drm_connector *connector) +{ + struct intel_tv *intel_tv = intel_attached_tv(connector); + const struct tv_mode *tv_mode = intel_tv_mode_find(intel_tv); + int i; + + if ((intel_tv->type == DRM_MODE_CONNECTOR_Component) == + tv_mode->component_only) + return; + + + for (i = 0; i < sizeof(tv_modes) / sizeof(*tv_modes); i++) { + tv_mode = tv_modes + i; + + if ((intel_tv->type == DRM_MODE_CONNECTOR_Component) == + tv_mode->component_only) + break; + } + + intel_tv->tv_format = tv_mode->name; + drm_connector_property_set_value(connector, + connector->dev->mode_config.tv_mode_property, i); +} + +/** + * Detect the TV connection. + * + * Currently this always returns CONNECTOR_STATUS_UNKNOWN, as we need to be sure + * we have a pipe programmed in order to probe the TV. + */ +static enum drm_connector_status +intel_tv_detect(struct drm_connector *connector, bool force) +{ + struct drm_display_mode mode; + struct intel_tv *intel_tv = intel_attached_tv(connector); + int type; + + mode = reported_modes[0]; + drm_mode_set_crtcinfo(&mode, CRTC_INTERLACE_HALVE_V); + + if (intel_tv->base.base.crtc && intel_tv->base.base.crtc->enabled) { + type = intel_tv_detect_type(intel_tv, connector); + } else if (force) { + struct intel_load_detect_pipe tmp; + + if (intel_get_load_detect_pipe(&intel_tv->base, connector, + &mode, &tmp)) { + type = intel_tv_detect_type(intel_tv, connector); + intel_release_load_detect_pipe(&intel_tv->base, + connector, + &tmp); + } else + return connector_status_unknown; + } else + return connector->status; + + if (type < 0) + return connector_status_disconnected; + + intel_tv->type = type; + intel_tv_find_better_format(connector); + + return connector_status_connected; +} + +static const struct input_res { + const char *name; + int w, h; +} input_res_table[] = { + {"640x480", 640, 480}, + {"800x600", 800, 600}, + {"1024x768", 1024, 768}, + {"1280x1024", 1280, 1024}, + {"848x480", 848, 480}, + {"1280x720", 1280, 720}, + {"1920x1080", 1920, 1080}, +}; + +/* + * Chose preferred mode according to line number of TV format + */ +static void +intel_tv_chose_preferred_modes(struct drm_connector *connector, + struct drm_display_mode *mode_ptr) +{ + struct intel_tv *intel_tv = intel_attached_tv(connector); + const struct tv_mode *tv_mode = intel_tv_mode_find(intel_tv); + + if (tv_mode->nbr_end < 480 && mode_ptr->vdisplay == 480) + mode_ptr->type |= DRM_MODE_TYPE_PREFERRED; + else if (tv_mode->nbr_end > 480) { + if (tv_mode->progressive == true && tv_mode->nbr_end < 720) { + if (mode_ptr->vdisplay == 720) + mode_ptr->type |= DRM_MODE_TYPE_PREFERRED; + } else if (mode_ptr->vdisplay == 1080) + mode_ptr->type |= DRM_MODE_TYPE_PREFERRED; + } +} + +/** + * Stub get_modes function. + * + * This should probably return a set of fixed modes, unless we can figure out + * how to probe modes off of TV connections. + */ + +static int +intel_tv_get_modes(struct drm_connector *connector) +{ + struct drm_display_mode *mode_ptr; + struct intel_tv *intel_tv = intel_attached_tv(connector); + const struct tv_mode *tv_mode = intel_tv_mode_find(intel_tv); + int j, count = 0; + u64 tmp; + + for (j = 0; j < ARRAY_SIZE(input_res_table); + j++) { + const struct input_res *input = &input_res_table[j]; + unsigned int hactive_s = input->w; + unsigned int vactive_s = input->h; + + if (tv_mode->max_srcw && input->w > tv_mode->max_srcw) + continue; + + if (input->w > 1024 && (!tv_mode->progressive + && !tv_mode->component_only)) + continue; + + mode_ptr = drm_mode_create(connector->dev); + if (!mode_ptr) + continue; + strncpy(mode_ptr->name, input->name, DRM_DISPLAY_MODE_LEN); + + mode_ptr->hdisplay = hactive_s; + mode_ptr->hsync_start = hactive_s + 1; + mode_ptr->hsync_end = hactive_s + 64; + if (mode_ptr->hsync_end <= mode_ptr->hsync_start) + mode_ptr->hsync_end = mode_ptr->hsync_start + 1; + mode_ptr->htotal = hactive_s + 96; + + mode_ptr->vdisplay = vactive_s; + mode_ptr->vsync_start = vactive_s + 1; + mode_ptr->vsync_end = vactive_s + 32; + if (mode_ptr->vsync_end <= mode_ptr->vsync_start) + mode_ptr->vsync_end = mode_ptr->vsync_start + 1; + mode_ptr->vtotal = vactive_s + 33; + + tmp = (u64) tv_mode->refresh * mode_ptr->vtotal; + tmp *= mode_ptr->htotal; + tmp = div_u64(tmp, 1000000); + mode_ptr->clock = (int) tmp; + + mode_ptr->type = DRM_MODE_TYPE_DRIVER; + intel_tv_chose_preferred_modes(connector, mode_ptr); + drm_mode_probed_add(connector, mode_ptr); + count++; + } + + return count; +} + +static void +intel_tv_destroy (struct drm_connector *connector) +{ + drm_sysfs_connector_remove(connector); + drm_connector_cleanup(connector); + kfree(connector); +} + + +static int +intel_tv_set_property(struct drm_connector *connector, struct drm_property *property, + uint64_t val) +{ + struct drm_device *dev = connector->dev; + struct intel_tv *intel_tv = intel_attached_tv(connector); + struct drm_crtc *crtc = intel_tv->base.base.crtc; + int ret = 0; + bool changed = false; + + ret = drm_connector_property_set_value(connector, property, val); + if (ret < 0) + goto out; + + if (property == dev->mode_config.tv_left_margin_property && + intel_tv->margin[TV_MARGIN_LEFT] != val) { + intel_tv->margin[TV_MARGIN_LEFT] = val; + changed = true; + } else if (property == dev->mode_config.tv_right_margin_property && + intel_tv->margin[TV_MARGIN_RIGHT] != val) { + intel_tv->margin[TV_MARGIN_RIGHT] = val; + changed = true; + } else if (property == dev->mode_config.tv_top_margin_property && + intel_tv->margin[TV_MARGIN_TOP] != val) { + intel_tv->margin[TV_MARGIN_TOP] = val; + changed = true; + } else if (property == dev->mode_config.tv_bottom_margin_property && + intel_tv->margin[TV_MARGIN_BOTTOM] != val) { + intel_tv->margin[TV_MARGIN_BOTTOM] = val; + changed = true; + } else if (property == dev->mode_config.tv_mode_property) { + if (val >= ARRAY_SIZE(tv_modes)) { + ret = -EINVAL; + goto out; + } + if (!strcmp(intel_tv->tv_format, tv_modes[val].name)) + goto out; + + intel_tv->tv_format = tv_modes[val].name; + changed = true; + } else { + ret = -EINVAL; + goto out; + } + + if (changed && crtc) + drm_crtc_helper_set_mode(crtc, &crtc->mode, crtc->x, + crtc->y, crtc->fb); +out: + return ret; +} + +static const struct drm_encoder_helper_funcs intel_tv_helper_funcs = { + .dpms = intel_tv_dpms, + .mode_fixup = intel_tv_mode_fixup, + .prepare = intel_encoder_prepare, + .mode_set = intel_tv_mode_set, + .commit = intel_encoder_commit, +}; + +static const struct drm_connector_funcs intel_tv_connector_funcs = { + .dpms = drm_helper_connector_dpms, + .detect = intel_tv_detect, + .destroy = intel_tv_destroy, + .set_property = intel_tv_set_property, + .fill_modes = drm_helper_probe_single_connector_modes, +}; + +static const struct drm_connector_helper_funcs intel_tv_connector_helper_funcs = { + .mode_valid = intel_tv_mode_valid, + .get_modes = intel_tv_get_modes, + .best_encoder = intel_best_encoder, +}; + +static const struct drm_encoder_funcs intel_tv_enc_funcs = { + .destroy = intel_encoder_destroy, +}; + +/* + * Enumerate the child dev array parsed from VBT to check whether + * the integrated TV is present. + * If it is present, return 1. + * If it is not present, return false. + * If no child dev is parsed from VBT, it assumes that the TV is present. + */ +static int tv_is_present_in_vbt(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct child_device_config *p_child; + int i, ret; + + if (!dev_priv->child_dev_num) + return 1; + + ret = 0; + for (i = 0; i < dev_priv->child_dev_num; i++) { + p_child = dev_priv->child_dev + i; + /* + * If the device type is not TV, continue. + */ + if (p_child->device_type != DEVICE_TYPE_INT_TV && + p_child->device_type != DEVICE_TYPE_TV) + continue; + /* Only when the addin_offset is non-zero, it is regarded + * as present. + */ + if (p_child->addin_offset) { + ret = 1; + break; + } + } + return ret; +} + +void +intel_tv_init(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_connector *connector; + struct intel_tv *intel_tv; + struct intel_encoder *intel_encoder; + struct intel_connector *intel_connector; + u32 tv_dac_on, tv_dac_off, save_tv_dac; + char *tv_format_names[ARRAY_SIZE(tv_modes)]; + int i, initial_mode = 0; + + if ((I915_READ(TV_CTL) & TV_FUSE_STATE_MASK) == TV_FUSE_STATE_DISABLED) + return; + + if (!tv_is_present_in_vbt(dev)) { + DRM_DEBUG_KMS("Integrated TV is not present.\n"); + return; + } + /* Even if we have an encoder we may not have a connector */ + if (!dev_priv->int_tv_support) + return; + + /* + * Sanity check the TV output by checking to see if the + * DAC register holds a value + */ + save_tv_dac = I915_READ(TV_DAC); + + I915_WRITE(TV_DAC, save_tv_dac | TVDAC_STATE_CHG_EN); + tv_dac_on = I915_READ(TV_DAC); + + I915_WRITE(TV_DAC, save_tv_dac & ~TVDAC_STATE_CHG_EN); + tv_dac_off = I915_READ(TV_DAC); + + I915_WRITE(TV_DAC, save_tv_dac); + + /* + * If the register does not hold the state change enable + * bit, (either as a 0 or a 1), assume it doesn't really + * exist + */ + if ((tv_dac_on & TVDAC_STATE_CHG_EN) == 0 || + (tv_dac_off & TVDAC_STATE_CHG_EN) != 0) + return; + + intel_tv = kzalloc(sizeof(struct intel_tv), GFP_KERNEL); + if (!intel_tv) { + return; + } + + intel_connector = kzalloc(sizeof(struct intel_connector), GFP_KERNEL); + if (!intel_connector) { + kfree(intel_tv); + return; + } + + intel_encoder = &intel_tv->base; + connector = &intel_connector->base; + + /* The documentation, for the older chipsets at least, recommend + * using a polling method rather than hotplug detection for TVs. + * This is because in order to perform the hotplug detection, the PLLs + * for the TV must be kept alive increasing power drain and starving + * bandwidth from other encoders. Notably for instance, it causes + * pipe underruns on Crestline when this encoder is supposedly idle. + * + * More recent chipsets favour HDMI rather than integrated S-Video. + */ + connector->polled = DRM_CONNECTOR_POLL_CONNECT; + + drm_connector_init(dev, connector, &intel_tv_connector_funcs, + DRM_MODE_CONNECTOR_SVIDEO); + + drm_encoder_init(dev, &intel_encoder->base, &intel_tv_enc_funcs, + DRM_MODE_ENCODER_TVDAC); + + intel_connector_attach_encoder(intel_connector, intel_encoder); + intel_encoder->type = INTEL_OUTPUT_TVOUT; + intel_encoder->crtc_mask = (1 << 0) | (1 << 1); + intel_encoder->clone_mask = (1 << INTEL_TV_CLONE_BIT); + intel_encoder->base.possible_crtcs = ((1 << 0) | (1 << 1)); + intel_encoder->base.possible_clones = (1 << INTEL_OUTPUT_TVOUT); + intel_tv->type = DRM_MODE_CONNECTOR_Unknown; + + /* BIOS margin values */ + intel_tv->margin[TV_MARGIN_LEFT] = 54; + intel_tv->margin[TV_MARGIN_TOP] = 36; + intel_tv->margin[TV_MARGIN_RIGHT] = 46; + intel_tv->margin[TV_MARGIN_BOTTOM] = 37; + + intel_tv->tv_format = tv_modes[initial_mode].name; + + drm_encoder_helper_add(&intel_encoder->base, &intel_tv_helper_funcs); + drm_connector_helper_add(connector, &intel_tv_connector_helper_funcs); + connector->interlace_allowed = false; + connector->doublescan_allowed = false; + + /* Create TV properties then attach current values */ + for (i = 0; i < ARRAY_SIZE(tv_modes); i++) + tv_format_names[i] = (char *)tv_modes[i].name; + drm_mode_create_tv_properties(dev, + ARRAY_SIZE(tv_modes), + tv_format_names); + + drm_connector_attach_property(connector, dev->mode_config.tv_mode_property, + initial_mode); + drm_connector_attach_property(connector, + dev->mode_config.tv_left_margin_property, + intel_tv->margin[TV_MARGIN_LEFT]); + drm_connector_attach_property(connector, + dev->mode_config.tv_top_margin_property, + intel_tv->margin[TV_MARGIN_TOP]); + drm_connector_attach_property(connector, + dev->mode_config.tv_right_margin_property, + intel_tv->margin[TV_MARGIN_RIGHT]); + drm_connector_attach_property(connector, + dev->mode_config.tv_bottom_margin_property, + intel_tv->margin[TV_MARGIN_BOTTOM]); + drm_sysfs_connector_add(connector); +} diff --git a/drivers/gpu/drm/mga/Makefile b/drivers/gpu/drm/mga/Makefile new file mode 100644 index 00000000..60684785 --- /dev/null +++ b/drivers/gpu/drm/mga/Makefile @@ -0,0 +1,11 @@ +# +# Makefile for the drm device driver. This driver provides support for the +# Direct Rendering Infrastructure (DRI) in XFree86 4.1.0 and higher. + +ccflags-y := -Iinclude/drm +mga-y := mga_drv.o mga_dma.o mga_state.o mga_warp.o mga_irq.o + +mga-$(CONFIG_COMPAT) += mga_ioc32.o + +obj-$(CONFIG_DRM_MGA) += mga.o + diff --git a/drivers/gpu/drm/mga/mga_dma.c b/drivers/gpu/drm/mga/mga_dma.c new file mode 100644 index 00000000..5ccb65de --- /dev/null +++ b/drivers/gpu/drm/mga/mga_dma.c @@ -0,0 +1,1154 @@ +/* mga_dma.c -- DMA support for mga g200/g400 -*- linux-c -*- + * Created: Mon Dec 13 01:50:01 1999 by jhartmann@precisioninsight.com + * + * Copyright 1999 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + */ + +/** + * \file mga_dma.c + * DMA support for MGA G200 / G400. + * + * \author Rickard E. (Rik) Faith + * \author Jeff Hartmann + * \author Keith Whitwell + * \author Gareth Hughes + */ + +#include "drmP.h" +#include "drm.h" +#include "drm_sarea.h" +#include "mga_drm.h" +#include "mga_drv.h" + +#define MGA_DEFAULT_USEC_TIMEOUT 10000 +#define MGA_FREELIST_DEBUG 0 + +#define MINIMAL_CLEANUP 0 +#define FULL_CLEANUP 1 +static int mga_do_cleanup_dma(struct drm_device *dev, int full_cleanup); + +/* ================================================================ + * Engine control + */ + +int mga_do_wait_for_idle(drm_mga_private_t *dev_priv) +{ + u32 status = 0; + int i; + DRM_DEBUG("\n"); + + for (i = 0; i < dev_priv->usec_timeout; i++) { + status = MGA_READ(MGA_STATUS) & MGA_ENGINE_IDLE_MASK; + if (status == MGA_ENDPRDMASTS) { + MGA_WRITE8(MGA_CRTC_INDEX, 0); + return 0; + } + DRM_UDELAY(1); + } + +#if MGA_DMA_DEBUG + DRM_ERROR("failed!\n"); + DRM_INFO(" status=0x%08x\n", status); +#endif + return -EBUSY; +} + +static int mga_do_dma_reset(drm_mga_private_t *dev_priv) +{ + drm_mga_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_mga_primary_buffer_t *primary = &dev_priv->prim; + + DRM_DEBUG("\n"); + + /* The primary DMA stream should look like new right about now. + */ + primary->tail = 0; + primary->space = primary->size; + primary->last_flush = 0; + + sarea_priv->last_wrap = 0; + + /* FIXME: Reset counters, buffer ages etc... + */ + + /* FIXME: What else do we need to reinitialize? WARP stuff? + */ + + return 0; +} + +/* ================================================================ + * Primary DMA stream + */ + +void mga_do_dma_flush(drm_mga_private_t *dev_priv) +{ + drm_mga_primary_buffer_t *primary = &dev_priv->prim; + u32 head, tail; + u32 status = 0; + int i; + DMA_LOCALS; + DRM_DEBUG("\n"); + + /* We need to wait so that we can do an safe flush */ + for (i = 0; i < dev_priv->usec_timeout; i++) { + status = MGA_READ(MGA_STATUS) & MGA_ENGINE_IDLE_MASK; + if (status == MGA_ENDPRDMASTS) + break; + DRM_UDELAY(1); + } + + if (primary->tail == primary->last_flush) { + DRM_DEBUG(" bailing out...\n"); + return; + } + + tail = primary->tail + dev_priv->primary->offset; + + /* We need to pad the stream between flushes, as the card + * actually (partially?) reads the first of these commands. + * See page 4-16 in the G400 manual, middle of the page or so. + */ + BEGIN_DMA(1); + + DMA_BLOCK(MGA_DMAPAD, 0x00000000, + MGA_DMAPAD, 0x00000000, + MGA_DMAPAD, 0x00000000, MGA_DMAPAD, 0x00000000); + + ADVANCE_DMA(); + + primary->last_flush = primary->tail; + + head = MGA_READ(MGA_PRIMADDRESS); + + if (head <= tail) + primary->space = primary->size - primary->tail; + else + primary->space = head - tail; + + DRM_DEBUG(" head = 0x%06lx\n", (unsigned long)(head - dev_priv->primary->offset)); + DRM_DEBUG(" tail = 0x%06lx\n", (unsigned long)(tail - dev_priv->primary->offset)); + DRM_DEBUG(" space = 0x%06x\n", primary->space); + + mga_flush_write_combine(); + MGA_WRITE(MGA_PRIMEND, tail | dev_priv->dma_access); + + DRM_DEBUG("done.\n"); +} + +void mga_do_dma_wrap_start(drm_mga_private_t *dev_priv) +{ + drm_mga_primary_buffer_t *primary = &dev_priv->prim; + u32 head, tail; + DMA_LOCALS; + DRM_DEBUG("\n"); + + BEGIN_DMA_WRAP(); + + DMA_BLOCK(MGA_DMAPAD, 0x00000000, + MGA_DMAPAD, 0x00000000, + MGA_DMAPAD, 0x00000000, MGA_DMAPAD, 0x00000000); + + ADVANCE_DMA(); + + tail = primary->tail + dev_priv->primary->offset; + + primary->tail = 0; + primary->last_flush = 0; + primary->last_wrap++; + + head = MGA_READ(MGA_PRIMADDRESS); + + if (head == dev_priv->primary->offset) + primary->space = primary->size; + else + primary->space = head - dev_priv->primary->offset; + + DRM_DEBUG(" head = 0x%06lx\n", (unsigned long)(head - dev_priv->primary->offset)); + DRM_DEBUG(" tail = 0x%06x\n", primary->tail); + DRM_DEBUG(" wrap = %d\n", primary->last_wrap); + DRM_DEBUG(" space = 0x%06x\n", primary->space); + + mga_flush_write_combine(); + MGA_WRITE(MGA_PRIMEND, tail | dev_priv->dma_access); + + set_bit(0, &primary->wrapped); + DRM_DEBUG("done.\n"); +} + +void mga_do_dma_wrap_end(drm_mga_private_t *dev_priv) +{ + drm_mga_primary_buffer_t *primary = &dev_priv->prim; + drm_mga_sarea_t *sarea_priv = dev_priv->sarea_priv; + u32 head = dev_priv->primary->offset; + DRM_DEBUG("\n"); + + sarea_priv->last_wrap++; + DRM_DEBUG(" wrap = %d\n", sarea_priv->last_wrap); + + mga_flush_write_combine(); + MGA_WRITE(MGA_PRIMADDRESS, head | MGA_DMA_GENERAL); + + clear_bit(0, &primary->wrapped); + DRM_DEBUG("done.\n"); +} + +/* ================================================================ + * Freelist management + */ + +#define MGA_BUFFER_USED (~0) +#define MGA_BUFFER_FREE 0 + +#if MGA_FREELIST_DEBUG +static void mga_freelist_print(struct drm_device *dev) +{ + drm_mga_private_t *dev_priv = dev->dev_private; + drm_mga_freelist_t *entry; + + DRM_INFO("\n"); + DRM_INFO("current dispatch: last=0x%x done=0x%x\n", + dev_priv->sarea_priv->last_dispatch, + (unsigned int)(MGA_READ(MGA_PRIMADDRESS) - + dev_priv->primary->offset)); + DRM_INFO("current freelist:\n"); + + for (entry = dev_priv->head->next; entry; entry = entry->next) { + DRM_INFO(" %p idx=%2d age=0x%x 0x%06lx\n", + entry, entry->buf->idx, entry->age.head, + (unsigned long)(entry->age.head - dev_priv->primary->offset)); + } + DRM_INFO("\n"); +} +#endif + +static int mga_freelist_init(struct drm_device *dev, drm_mga_private_t *dev_priv) +{ + struct drm_device_dma *dma = dev->dma; + struct drm_buf *buf; + drm_mga_buf_priv_t *buf_priv; + drm_mga_freelist_t *entry; + int i; + DRM_DEBUG("count=%d\n", dma->buf_count); + + dev_priv->head = kzalloc(sizeof(drm_mga_freelist_t), GFP_KERNEL); + if (dev_priv->head == NULL) + return -ENOMEM; + + SET_AGE(&dev_priv->head->age, MGA_BUFFER_USED, 0); + + for (i = 0; i < dma->buf_count; i++) { + buf = dma->buflist[i]; + buf_priv = buf->dev_private; + + entry = kzalloc(sizeof(drm_mga_freelist_t), GFP_KERNEL); + if (entry == NULL) + return -ENOMEM; + + entry->next = dev_priv->head->next; + entry->prev = dev_priv->head; + SET_AGE(&entry->age, MGA_BUFFER_FREE, 0); + entry->buf = buf; + + if (dev_priv->head->next != NULL) + dev_priv->head->next->prev = entry; + if (entry->next == NULL) + dev_priv->tail = entry; + + buf_priv->list_entry = entry; + buf_priv->discard = 0; + buf_priv->dispatched = 0; + + dev_priv->head->next = entry; + } + + return 0; +} + +static void mga_freelist_cleanup(struct drm_device *dev) +{ + drm_mga_private_t *dev_priv = dev->dev_private; + drm_mga_freelist_t *entry; + drm_mga_freelist_t *next; + DRM_DEBUG("\n"); + + entry = dev_priv->head; + while (entry) { + next = entry->next; + kfree(entry); + entry = next; + } + + dev_priv->head = dev_priv->tail = NULL; +} + +#if 0 +/* FIXME: Still needed? + */ +static void mga_freelist_reset(struct drm_device *dev) +{ + struct drm_device_dma *dma = dev->dma; + struct drm_buf *buf; + drm_mga_buf_priv_t *buf_priv; + int i; + + for (i = 0; i < dma->buf_count; i++) { + buf = dma->buflist[i]; + buf_priv = buf->dev_private; + SET_AGE(&buf_priv->list_entry->age, MGA_BUFFER_FREE, 0); + } +} +#endif + +static struct drm_buf *mga_freelist_get(struct drm_device * dev) +{ + drm_mga_private_t *dev_priv = dev->dev_private; + drm_mga_freelist_t *next; + drm_mga_freelist_t *prev; + drm_mga_freelist_t *tail = dev_priv->tail; + u32 head, wrap; + DRM_DEBUG("\n"); + + head = MGA_READ(MGA_PRIMADDRESS); + wrap = dev_priv->sarea_priv->last_wrap; + + DRM_DEBUG(" tail=0x%06lx %d\n", + tail->age.head ? + (unsigned long)(tail->age.head - dev_priv->primary->offset) : 0, + tail->age.wrap); + DRM_DEBUG(" head=0x%06lx %d\n", + (unsigned long)(head - dev_priv->primary->offset), wrap); + + if (TEST_AGE(&tail->age, head, wrap)) { + prev = dev_priv->tail->prev; + next = dev_priv->tail; + prev->next = NULL; + next->prev = next->next = NULL; + dev_priv->tail = prev; + SET_AGE(&next->age, MGA_BUFFER_USED, 0); + return next->buf; + } + + DRM_DEBUG("returning NULL!\n"); + return NULL; +} + +int mga_freelist_put(struct drm_device *dev, struct drm_buf *buf) +{ + drm_mga_private_t *dev_priv = dev->dev_private; + drm_mga_buf_priv_t *buf_priv = buf->dev_private; + drm_mga_freelist_t *head, *entry, *prev; + + DRM_DEBUG("age=0x%06lx wrap=%d\n", + (unsigned long)(buf_priv->list_entry->age.head - + dev_priv->primary->offset), + buf_priv->list_entry->age.wrap); + + entry = buf_priv->list_entry; + head = dev_priv->head; + + if (buf_priv->list_entry->age.head == MGA_BUFFER_USED) { + SET_AGE(&entry->age, MGA_BUFFER_FREE, 0); + prev = dev_priv->tail; + prev->next = entry; + entry->prev = prev; + entry->next = NULL; + } else { + prev = head->next; + head->next = entry; + prev->prev = entry; + entry->prev = head; + entry->next = prev; + } + + return 0; +} + +/* ================================================================ + * DMA initialization, cleanup + */ + +int mga_driver_load(struct drm_device *dev, unsigned long flags) +{ + drm_mga_private_t *dev_priv; + int ret; + + dev_priv = kzalloc(sizeof(drm_mga_private_t), GFP_KERNEL); + if (!dev_priv) + return -ENOMEM; + + dev->dev_private = (void *)dev_priv; + + dev_priv->usec_timeout = MGA_DEFAULT_USEC_TIMEOUT; + dev_priv->chipset = flags; + + dev_priv->mmio_base = pci_resource_start(dev->pdev, 1); + dev_priv->mmio_size = pci_resource_len(dev->pdev, 1); + + dev->counters += 3; + dev->types[6] = _DRM_STAT_IRQ; + dev->types[7] = _DRM_STAT_PRIMARY; + dev->types[8] = _DRM_STAT_SECONDARY; + + ret = drm_vblank_init(dev, 1); + + if (ret) { + (void) mga_driver_unload(dev); + return ret; + } + + return 0; +} + +#if __OS_HAS_AGP +/** + * Bootstrap the driver for AGP DMA. + * + * \todo + * Investigate whether there is any benefit to storing the WARP microcode in + * AGP memory. If not, the microcode may as well always be put in PCI + * memory. + * + * \todo + * This routine needs to set dma_bs->agp_mode to the mode actually configured + * in the hardware. Looking just at the Linux AGP driver code, I don't see + * an easy way to determine this. + * + * \sa mga_do_dma_bootstrap, mga_do_pci_dma_bootstrap + */ +static int mga_do_agp_dma_bootstrap(struct drm_device *dev, + drm_mga_dma_bootstrap_t *dma_bs) +{ + drm_mga_private_t *const dev_priv = + (drm_mga_private_t *) dev->dev_private; + unsigned int warp_size = MGA_WARP_UCODE_SIZE; + int err; + unsigned offset; + const unsigned secondary_size = dma_bs->secondary_bin_count + * dma_bs->secondary_bin_size; + const unsigned agp_size = (dma_bs->agp_size << 20); + struct drm_buf_desc req; + struct drm_agp_mode mode; + struct drm_agp_info info; + struct drm_agp_buffer agp_req; + struct drm_agp_binding bind_req; + + /* Acquire AGP. */ + err = drm_agp_acquire(dev); + if (err) { + DRM_ERROR("Unable to acquire AGP: %d\n", err); + return err; + } + + err = drm_agp_info(dev, &info); + if (err) { + DRM_ERROR("Unable to get AGP info: %d\n", err); + return err; + } + + mode.mode = (info.mode & ~0x07) | dma_bs->agp_mode; + err = drm_agp_enable(dev, mode); + if (err) { + DRM_ERROR("Unable to enable AGP (mode = 0x%lx)\n", mode.mode); + return err; + } + + /* In addition to the usual AGP mode configuration, the G200 AGP cards + * need to have the AGP mode "manually" set. + */ + + if (dev_priv->chipset == MGA_CARD_TYPE_G200) { + if (mode.mode & 0x02) + MGA_WRITE(MGA_AGP_PLL, MGA_AGP2XPLL_ENABLE); + else + MGA_WRITE(MGA_AGP_PLL, MGA_AGP2XPLL_DISABLE); + } + + /* Allocate and bind AGP memory. */ + agp_req.size = agp_size; + agp_req.type = 0; + err = drm_agp_alloc(dev, &agp_req); + if (err) { + dev_priv->agp_size = 0; + DRM_ERROR("Unable to allocate %uMB AGP memory\n", + dma_bs->agp_size); + return err; + } + + dev_priv->agp_size = agp_size; + dev_priv->agp_handle = agp_req.handle; + + bind_req.handle = agp_req.handle; + bind_req.offset = 0; + err = drm_agp_bind(dev, &bind_req); + if (err) { + DRM_ERROR("Unable to bind AGP memory: %d\n", err); + return err; + } + + /* Make drm_addbufs happy by not trying to create a mapping for less + * than a page. + */ + if (warp_size < PAGE_SIZE) + warp_size = PAGE_SIZE; + + offset = 0; + err = drm_addmap(dev, offset, warp_size, + _DRM_AGP, _DRM_READ_ONLY, &dev_priv->warp); + if (err) { + DRM_ERROR("Unable to map WARP microcode: %d\n", err); + return err; + } + + offset += warp_size; + err = drm_addmap(dev, offset, dma_bs->primary_size, + _DRM_AGP, _DRM_READ_ONLY, &dev_priv->primary); + if (err) { + DRM_ERROR("Unable to map primary DMA region: %d\n", err); + return err; + } + + offset += dma_bs->primary_size; + err = drm_addmap(dev, offset, secondary_size, + _DRM_AGP, 0, &dev->agp_buffer_map); + if (err) { + DRM_ERROR("Unable to map secondary DMA region: %d\n", err); + return err; + } + + (void)memset(&req, 0, sizeof(req)); + req.count = dma_bs->secondary_bin_count; + req.size = dma_bs->secondary_bin_size; + req.flags = _DRM_AGP_BUFFER; + req.agp_start = offset; + + err = drm_addbufs_agp(dev, &req); + if (err) { + DRM_ERROR("Unable to add secondary DMA buffers: %d\n", err); + return err; + } + + { + struct drm_map_list *_entry; + unsigned long agp_token = 0; + + list_for_each_entry(_entry, &dev->maplist, head) { + if (_entry->map == dev->agp_buffer_map) + agp_token = _entry->user_token; + } + if (!agp_token) + return -EFAULT; + + dev->agp_buffer_token = agp_token; + } + + offset += secondary_size; + err = drm_addmap(dev, offset, agp_size - offset, + _DRM_AGP, 0, &dev_priv->agp_textures); + if (err) { + DRM_ERROR("Unable to map AGP texture region %d\n", err); + return err; + } + + drm_core_ioremap(dev_priv->warp, dev); + drm_core_ioremap(dev_priv->primary, dev); + drm_core_ioremap(dev->agp_buffer_map, dev); + + if (!dev_priv->warp->handle || + !dev_priv->primary->handle || !dev->agp_buffer_map->handle) { + DRM_ERROR("failed to ioremap agp regions! (%p, %p, %p)\n", + dev_priv->warp->handle, dev_priv->primary->handle, + dev->agp_buffer_map->handle); + return -ENOMEM; + } + + dev_priv->dma_access = MGA_PAGPXFER; + dev_priv->wagp_enable = MGA_WAGP_ENABLE; + + DRM_INFO("Initialized card for AGP DMA.\n"); + return 0; +} +#else +static int mga_do_agp_dma_bootstrap(struct drm_device *dev, + drm_mga_dma_bootstrap_t *dma_bs) +{ + return -EINVAL; +} +#endif + +/** + * Bootstrap the driver for PCI DMA. + * + * \todo + * The algorithm for decreasing the size of the primary DMA buffer could be + * better. The size should be rounded up to the nearest page size, then + * decrease the request size by a single page each pass through the loop. + * + * \todo + * Determine whether the maximum address passed to drm_pci_alloc is correct. + * The same goes for drm_addbufs_pci. + * + * \sa mga_do_dma_bootstrap, mga_do_agp_dma_bootstrap + */ +static int mga_do_pci_dma_bootstrap(struct drm_device *dev, + drm_mga_dma_bootstrap_t *dma_bs) +{ + drm_mga_private_t *const dev_priv = + (drm_mga_private_t *) dev->dev_private; + unsigned int warp_size = MGA_WARP_UCODE_SIZE; + unsigned int primary_size; + unsigned int bin_count; + int err; + struct drm_buf_desc req; + + if (dev->dma == NULL) { + DRM_ERROR("dev->dma is NULL\n"); + return -EFAULT; + } + + /* Make drm_addbufs happy by not trying to create a mapping for less + * than a page. + */ + if (warp_size < PAGE_SIZE) + warp_size = PAGE_SIZE; + + /* The proper alignment is 0x100 for this mapping */ + err = drm_addmap(dev, 0, warp_size, _DRM_CONSISTENT, + _DRM_READ_ONLY, &dev_priv->warp); + if (err != 0) { + DRM_ERROR("Unable to create mapping for WARP microcode: %d\n", + err); + return err; + } + + /* Other than the bottom two bits being used to encode other + * information, there don't appear to be any restrictions on the + * alignment of the primary or secondary DMA buffers. + */ + + for (primary_size = dma_bs->primary_size; primary_size != 0; + primary_size >>= 1) { + /* The proper alignment for this mapping is 0x04 */ + err = drm_addmap(dev, 0, primary_size, _DRM_CONSISTENT, + _DRM_READ_ONLY, &dev_priv->primary); + if (!err) + break; + } + + if (err != 0) { + DRM_ERROR("Unable to allocate primary DMA region: %d\n", err); + return -ENOMEM; + } + + if (dev_priv->primary->size != dma_bs->primary_size) { + DRM_INFO("Primary DMA buffer size reduced from %u to %u.\n", + dma_bs->primary_size, + (unsigned)dev_priv->primary->size); + dma_bs->primary_size = dev_priv->primary->size; + } + + for (bin_count = dma_bs->secondary_bin_count; bin_count > 0; + bin_count--) { + (void)memset(&req, 0, sizeof(req)); + req.count = bin_count; + req.size = dma_bs->secondary_bin_size; + + err = drm_addbufs_pci(dev, &req); + if (!err) + break; + } + + if (bin_count == 0) { + DRM_ERROR("Unable to add secondary DMA buffers: %d\n", err); + return err; + } + + if (bin_count != dma_bs->secondary_bin_count) { + DRM_INFO("Secondary PCI DMA buffer bin count reduced from %u " + "to %u.\n", dma_bs->secondary_bin_count, bin_count); + + dma_bs->secondary_bin_count = bin_count; + } + + dev_priv->dma_access = 0; + dev_priv->wagp_enable = 0; + + dma_bs->agp_mode = 0; + + DRM_INFO("Initialized card for PCI DMA.\n"); + return 0; +} + +static int mga_do_dma_bootstrap(struct drm_device *dev, + drm_mga_dma_bootstrap_t *dma_bs) +{ + const int is_agp = (dma_bs->agp_mode != 0) && drm_pci_device_is_agp(dev); + int err; + drm_mga_private_t *const dev_priv = + (drm_mga_private_t *) dev->dev_private; + + dev_priv->used_new_dma_init = 1; + + /* The first steps are the same for both PCI and AGP based DMA. Map + * the cards MMIO registers and map a status page. + */ + err = drm_addmap(dev, dev_priv->mmio_base, dev_priv->mmio_size, + _DRM_REGISTERS, _DRM_READ_ONLY, &dev_priv->mmio); + if (err) { + DRM_ERROR("Unable to map MMIO region: %d\n", err); + return err; + } + + err = drm_addmap(dev, 0, SAREA_MAX, _DRM_SHM, + _DRM_READ_ONLY | _DRM_LOCKED | _DRM_KERNEL, + &dev_priv->status); + if (err) { + DRM_ERROR("Unable to map status region: %d\n", err); + return err; + } + + /* The DMA initialization procedure is slightly different for PCI and + * AGP cards. AGP cards just allocate a large block of AGP memory and + * carve off portions of it for internal uses. The remaining memory + * is returned to user-mode to be used for AGP textures. + */ + if (is_agp) + err = mga_do_agp_dma_bootstrap(dev, dma_bs); + + /* If we attempted to initialize the card for AGP DMA but failed, + * clean-up any mess that may have been created. + */ + + if (err) + mga_do_cleanup_dma(dev, MINIMAL_CLEANUP); + + /* Not only do we want to try and initialized PCI cards for PCI DMA, + * but we also try to initialized AGP cards that could not be + * initialized for AGP DMA. This covers the case where we have an AGP + * card in a system with an unsupported AGP chipset. In that case the + * card will be detected as AGP, but we won't be able to allocate any + * AGP memory, etc. + */ + + if (!is_agp || err) + err = mga_do_pci_dma_bootstrap(dev, dma_bs); + + return err; +} + +int mga_dma_bootstrap(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + drm_mga_dma_bootstrap_t *bootstrap = data; + int err; + static const int modes[] = { 0, 1, 2, 2, 4, 4, 4, 4 }; + const drm_mga_private_t *const dev_priv = + (drm_mga_private_t *) dev->dev_private; + + err = mga_do_dma_bootstrap(dev, bootstrap); + if (err) { + mga_do_cleanup_dma(dev, FULL_CLEANUP); + return err; + } + + if (dev_priv->agp_textures != NULL) { + bootstrap->texture_handle = dev_priv->agp_textures->offset; + bootstrap->texture_size = dev_priv->agp_textures->size; + } else { + bootstrap->texture_handle = 0; + bootstrap->texture_size = 0; + } + + bootstrap->agp_mode = modes[bootstrap->agp_mode & 0x07]; + + return err; +} + +static int mga_do_init_dma(struct drm_device *dev, drm_mga_init_t *init) +{ + drm_mga_private_t *dev_priv; + int ret; + DRM_DEBUG("\n"); + + dev_priv = dev->dev_private; + + if (init->sgram) + dev_priv->clear_cmd = MGA_DWGCTL_CLEAR | MGA_ATYPE_BLK; + else + dev_priv->clear_cmd = MGA_DWGCTL_CLEAR | MGA_ATYPE_RSTR; + dev_priv->maccess = init->maccess; + + dev_priv->fb_cpp = init->fb_cpp; + dev_priv->front_offset = init->front_offset; + dev_priv->front_pitch = init->front_pitch; + dev_priv->back_offset = init->back_offset; + dev_priv->back_pitch = init->back_pitch; + + dev_priv->depth_cpp = init->depth_cpp; + dev_priv->depth_offset = init->depth_offset; + dev_priv->depth_pitch = init->depth_pitch; + + /* FIXME: Need to support AGP textures... + */ + dev_priv->texture_offset = init->texture_offset[0]; + dev_priv->texture_size = init->texture_size[0]; + + dev_priv->sarea = drm_getsarea(dev); + if (!dev_priv->sarea) { + DRM_ERROR("failed to find sarea!\n"); + return -EINVAL; + } + + if (!dev_priv->used_new_dma_init) { + + dev_priv->dma_access = MGA_PAGPXFER; + dev_priv->wagp_enable = MGA_WAGP_ENABLE; + + dev_priv->status = drm_core_findmap(dev, init->status_offset); + if (!dev_priv->status) { + DRM_ERROR("failed to find status page!\n"); + return -EINVAL; + } + dev_priv->mmio = drm_core_findmap(dev, init->mmio_offset); + if (!dev_priv->mmio) { + DRM_ERROR("failed to find mmio region!\n"); + return -EINVAL; + } + dev_priv->warp = drm_core_findmap(dev, init->warp_offset); + if (!dev_priv->warp) { + DRM_ERROR("failed to find warp microcode region!\n"); + return -EINVAL; + } + dev_priv->primary = drm_core_findmap(dev, init->primary_offset); + if (!dev_priv->primary) { + DRM_ERROR("failed to find primary dma region!\n"); + return -EINVAL; + } + dev->agp_buffer_token = init->buffers_offset; + dev->agp_buffer_map = + drm_core_findmap(dev, init->buffers_offset); + if (!dev->agp_buffer_map) { + DRM_ERROR("failed to find dma buffer region!\n"); + return -EINVAL; + } + + drm_core_ioremap(dev_priv->warp, dev); + drm_core_ioremap(dev_priv->primary, dev); + drm_core_ioremap(dev->agp_buffer_map, dev); + } + + dev_priv->sarea_priv = + (drm_mga_sarea_t *) ((u8 *) dev_priv->sarea->handle + + init->sarea_priv_offset); + + if (!dev_priv->warp->handle || + !dev_priv->primary->handle || + ((dev_priv->dma_access != 0) && + ((dev->agp_buffer_map == NULL) || + (dev->agp_buffer_map->handle == NULL)))) { + DRM_ERROR("failed to ioremap agp regions!\n"); + return -ENOMEM; + } + + ret = mga_warp_install_microcode(dev_priv); + if (ret < 0) { + DRM_ERROR("failed to install WARP ucode!: %d\n", ret); + return ret; + } + + ret = mga_warp_init(dev_priv); + if (ret < 0) { + DRM_ERROR("failed to init WARP engine!: %d\n", ret); + return ret; + } + + dev_priv->prim.status = (u32 *) dev_priv->status->handle; + + mga_do_wait_for_idle(dev_priv); + + /* Init the primary DMA registers. + */ + MGA_WRITE(MGA_PRIMADDRESS, dev_priv->primary->offset | MGA_DMA_GENERAL); +#if 0 + MGA_WRITE(MGA_PRIMPTR, virt_to_bus((void *)dev_priv->prim.status) | MGA_PRIMPTREN0 | /* Soft trap, SECEND, SETUPEND */ + MGA_PRIMPTREN1); /* DWGSYNC */ +#endif + + dev_priv->prim.start = (u8 *) dev_priv->primary->handle; + dev_priv->prim.end = ((u8 *) dev_priv->primary->handle + + dev_priv->primary->size); + dev_priv->prim.size = dev_priv->primary->size; + + dev_priv->prim.tail = 0; + dev_priv->prim.space = dev_priv->prim.size; + dev_priv->prim.wrapped = 0; + + dev_priv->prim.last_flush = 0; + dev_priv->prim.last_wrap = 0; + + dev_priv->prim.high_mark = 256 * DMA_BLOCK_SIZE; + + dev_priv->prim.status[0] = dev_priv->primary->offset; + dev_priv->prim.status[1] = 0; + + dev_priv->sarea_priv->last_wrap = 0; + dev_priv->sarea_priv->last_frame.head = 0; + dev_priv->sarea_priv->last_frame.wrap = 0; + + if (mga_freelist_init(dev, dev_priv) < 0) { + DRM_ERROR("could not initialize freelist\n"); + return -ENOMEM; + } + + return 0; +} + +static int mga_do_cleanup_dma(struct drm_device *dev, int full_cleanup) +{ + int err = 0; + DRM_DEBUG("\n"); + + /* Make sure interrupts are disabled here because the uninstall ioctl + * may not have been called from userspace and after dev_private + * is freed, it's too late. + */ + if (dev->irq_enabled) + drm_irq_uninstall(dev); + + if (dev->dev_private) { + drm_mga_private_t *dev_priv = dev->dev_private; + + if ((dev_priv->warp != NULL) + && (dev_priv->warp->type != _DRM_CONSISTENT)) + drm_core_ioremapfree(dev_priv->warp, dev); + + if ((dev_priv->primary != NULL) + && (dev_priv->primary->type != _DRM_CONSISTENT)) + drm_core_ioremapfree(dev_priv->primary, dev); + + if (dev->agp_buffer_map != NULL) + drm_core_ioremapfree(dev->agp_buffer_map, dev); + + if (dev_priv->used_new_dma_init) { +#if __OS_HAS_AGP + if (dev_priv->agp_handle != 0) { + struct drm_agp_binding unbind_req; + struct drm_agp_buffer free_req; + + unbind_req.handle = dev_priv->agp_handle; + drm_agp_unbind(dev, &unbind_req); + + free_req.handle = dev_priv->agp_handle; + drm_agp_free(dev, &free_req); + + dev_priv->agp_textures = NULL; + dev_priv->agp_size = 0; + dev_priv->agp_handle = 0; + } + + if ((dev->agp != NULL) && dev->agp->acquired) + err = drm_agp_release(dev); +#endif + } + + dev_priv->warp = NULL; + dev_priv->primary = NULL; + dev_priv->sarea = NULL; + dev_priv->sarea_priv = NULL; + dev->agp_buffer_map = NULL; + + if (full_cleanup) { + dev_priv->mmio = NULL; + dev_priv->status = NULL; + dev_priv->used_new_dma_init = 0; + } + + memset(&dev_priv->prim, 0, sizeof(dev_priv->prim)); + dev_priv->warp_pipe = 0; + memset(dev_priv->warp_pipe_phys, 0, + sizeof(dev_priv->warp_pipe_phys)); + + if (dev_priv->head != NULL) + mga_freelist_cleanup(dev); + } + + return err; +} + +int mga_dma_init(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + drm_mga_init_t *init = data; + int err; + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + switch (init->func) { + case MGA_INIT_DMA: + err = mga_do_init_dma(dev, init); + if (err) + (void)mga_do_cleanup_dma(dev, FULL_CLEANUP); + return err; + case MGA_CLEANUP_DMA: + return mga_do_cleanup_dma(dev, FULL_CLEANUP); + } + + return -EINVAL; +} + +/* ================================================================ + * Primary DMA stream management + */ + +int mga_dma_flush(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + drm_mga_private_t *dev_priv = (drm_mga_private_t *) dev->dev_private; + struct drm_lock *lock = data; + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + DRM_DEBUG("%s%s%s\n", + (lock->flags & _DRM_LOCK_FLUSH) ? "flush, " : "", + (lock->flags & _DRM_LOCK_FLUSH_ALL) ? "flush all, " : "", + (lock->flags & _DRM_LOCK_QUIESCENT) ? "idle, " : ""); + + WRAP_WAIT_WITH_RETURN(dev_priv); + + if (lock->flags & (_DRM_LOCK_FLUSH | _DRM_LOCK_FLUSH_ALL)) + mga_do_dma_flush(dev_priv); + + if (lock->flags & _DRM_LOCK_QUIESCENT) { +#if MGA_DMA_DEBUG + int ret = mga_do_wait_for_idle(dev_priv); + if (ret < 0) + DRM_INFO("-EBUSY\n"); + return ret; +#else + return mga_do_wait_for_idle(dev_priv); +#endif + } else { + return 0; + } +} + +int mga_dma_reset(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + drm_mga_private_t *dev_priv = (drm_mga_private_t *) dev->dev_private; + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + return mga_do_dma_reset(dev_priv); +} + +/* ================================================================ + * DMA buffer management + */ + +static int mga_dma_get_buffers(struct drm_device *dev, + struct drm_file *file_priv, struct drm_dma *d) +{ + struct drm_buf *buf; + int i; + + for (i = d->granted_count; i < d->request_count; i++) { + buf = mga_freelist_get(dev); + if (!buf) + return -EAGAIN; + + buf->file_priv = file_priv; + + if (DRM_COPY_TO_USER(&d->request_indices[i], + &buf->idx, sizeof(buf->idx))) + return -EFAULT; + if (DRM_COPY_TO_USER(&d->request_sizes[i], + &buf->total, sizeof(buf->total))) + return -EFAULT; + + d->granted_count++; + } + return 0; +} + +int mga_dma_buffers(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_device_dma *dma = dev->dma; + drm_mga_private_t *dev_priv = (drm_mga_private_t *) dev->dev_private; + struct drm_dma *d = data; + int ret = 0; + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + /* Please don't send us buffers. + */ + if (d->send_count != 0) { + DRM_ERROR("Process %d trying to send %d buffers via drmDMA\n", + DRM_CURRENTPID, d->send_count); + return -EINVAL; + } + + /* We'll send you buffers. + */ + if (d->request_count < 0 || d->request_count > dma->buf_count) { + DRM_ERROR("Process %d trying to get %d buffers (of %d max)\n", + DRM_CURRENTPID, d->request_count, dma->buf_count); + return -EINVAL; + } + + WRAP_TEST_WITH_RETURN(dev_priv); + + d->granted_count = 0; + + if (d->request_count) + ret = mga_dma_get_buffers(dev, file_priv, d); + + return ret; +} + +/** + * Called just before the module is unloaded. + */ +int mga_driver_unload(struct drm_device *dev) +{ + kfree(dev->dev_private); + dev->dev_private = NULL; + + return 0; +} + +/** + * Called when the last opener of the device is closed. + */ +void mga_driver_lastclose(struct drm_device *dev) +{ + mga_do_cleanup_dma(dev, FULL_CLEANUP); +} + +int mga_driver_dma_quiescent(struct drm_device *dev) +{ + drm_mga_private_t *dev_priv = dev->dev_private; + return mga_do_wait_for_idle(dev_priv); +} diff --git a/drivers/gpu/drm/mga/mga_drv.c b/drivers/gpu/drm/mga/mga_drv.c new file mode 100644 index 00000000..42d31874 --- /dev/null +++ b/drivers/gpu/drm/mga/mga_drv.c @@ -0,0 +1,142 @@ +/* mga_drv.c -- Matrox G200/G400 driver -*- linux-c -*- + * Created: Mon Dec 13 01:56:22 1999 by jhartmann@precisioninsight.com + * + * Copyright 1999 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * VA LINUX SYSTEMS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: + * Rickard E. (Rik) Faith + * Gareth Hughes + */ + +#include "drmP.h" +#include "drm.h" +#include "mga_drm.h" +#include "mga_drv.h" + +#include "drm_pciids.h" + +static int mga_driver_device_is_agp(struct drm_device *dev); + +static struct pci_device_id pciidlist[] = { + mga_PCI_IDS +}; + +static struct drm_driver driver = { + .driver_features = + DRIVER_USE_AGP | DRIVER_USE_MTRR | DRIVER_PCI_DMA | + DRIVER_HAVE_DMA | DRIVER_HAVE_IRQ | DRIVER_IRQ_SHARED, + .dev_priv_size = sizeof(drm_mga_buf_priv_t), + .load = mga_driver_load, + .unload = mga_driver_unload, + .lastclose = mga_driver_lastclose, + .dma_quiescent = mga_driver_dma_quiescent, + .device_is_agp = mga_driver_device_is_agp, + .get_vblank_counter = mga_get_vblank_counter, + .enable_vblank = mga_enable_vblank, + .disable_vblank = mga_disable_vblank, + .irq_preinstall = mga_driver_irq_preinstall, + .irq_postinstall = mga_driver_irq_postinstall, + .irq_uninstall = mga_driver_irq_uninstall, + .irq_handler = mga_driver_irq_handler, + .reclaim_buffers = drm_core_reclaim_buffers, + .ioctls = mga_ioctls, + .dma_ioctl = mga_dma_buffers, + .fops = { + .owner = THIS_MODULE, + .open = drm_open, + .release = drm_release, + .unlocked_ioctl = drm_ioctl, + .mmap = drm_mmap, + .poll = drm_poll, + .fasync = drm_fasync, +#ifdef CONFIG_COMPAT + .compat_ioctl = mga_compat_ioctl, +#endif + .llseek = noop_llseek, + }, + + .name = DRIVER_NAME, + .desc = DRIVER_DESC, + .date = DRIVER_DATE, + .major = DRIVER_MAJOR, + .minor = DRIVER_MINOR, + .patchlevel = DRIVER_PATCHLEVEL, +}; + +static struct pci_driver mga_pci_driver = { + .name = DRIVER_NAME, + .id_table = pciidlist, +}; + +static int __init mga_init(void) +{ + driver.num_ioctls = mga_max_ioctl; + return drm_pci_init(&driver, &mga_pci_driver); +} + +static void __exit mga_exit(void) +{ + drm_pci_exit(&driver, &mga_pci_driver); +} + +module_init(mga_init); +module_exit(mga_exit); + +MODULE_AUTHOR(DRIVER_AUTHOR); +MODULE_DESCRIPTION(DRIVER_DESC); +MODULE_LICENSE("GPL and additional rights"); + +/** + * Determine if the device really is AGP or not. + * + * In addition to the usual tests performed by \c drm_device_is_agp, this + * function detects PCI G450 cards that appear to the system exactly like + * AGP G450 cards. + * + * \param dev The device to be tested. + * + * \returns + * If the device is a PCI G450, zero is returned. Otherwise 2 is returned. + */ +static int mga_driver_device_is_agp(struct drm_device *dev) +{ + const struct pci_dev *const pdev = dev->pdev; + + /* There are PCI versions of the G450. These cards have the + * same PCI ID as the AGP G450, but have an additional PCI-to-PCI + * bridge chip. We detect these cards, which are not currently + * supported by this driver, by looking at the device ID of the + * bus the "card" is on. If vendor is 0x3388 (Hint Corp) and the + * device is 0x0021 (HB6 Universal PCI-PCI bridge), we reject the + * device. + */ + + if ((pdev->device == 0x0525) && pdev->bus->self + && (pdev->bus->self->vendor == 0x3388) + && (pdev->bus->self->device == 0x0021)) { + return 0; + } + + return 2; +} diff --git a/drivers/gpu/drm/mga/mga_drv.h b/drivers/gpu/drm/mga/mga_drv.h new file mode 100644 index 00000000..54558a01 --- /dev/null +++ b/drivers/gpu/drm/mga/mga_drv.h @@ -0,0 +1,666 @@ +/* mga_drv.h -- Private header for the Matrox G200/G400 driver -*- linux-c -*- + * Created: Mon Dec 13 01:50:01 1999 by jhartmann@precisioninsight.com + * + * Copyright 1999 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California. + * All rights reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * VA LINUX SYSTEMS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: + * Gareth Hughes + */ + +#ifndef __MGA_DRV_H__ +#define __MGA_DRV_H__ + +/* General customization: + */ + +#define DRIVER_AUTHOR "Gareth Hughes, VA Linux Systems Inc." + +#define DRIVER_NAME "mga" +#define DRIVER_DESC "Matrox G200/G400" +#define DRIVER_DATE "20051102" + +#define DRIVER_MAJOR 3 +#define DRIVER_MINOR 2 +#define DRIVER_PATCHLEVEL 1 + +typedef struct drm_mga_primary_buffer { + u8 *start; + u8 *end; + int size; + + u32 tail; + int space; + volatile long wrapped; + + volatile u32 *status; + + u32 last_flush; + u32 last_wrap; + + u32 high_mark; +} drm_mga_primary_buffer_t; + +typedef struct drm_mga_freelist { + struct drm_mga_freelist *next; + struct drm_mga_freelist *prev; + drm_mga_age_t age; + struct drm_buf *buf; +} drm_mga_freelist_t; + +typedef struct { + drm_mga_freelist_t *list_entry; + int discard; + int dispatched; +} drm_mga_buf_priv_t; + +typedef struct drm_mga_private { + drm_mga_primary_buffer_t prim; + drm_mga_sarea_t *sarea_priv; + + drm_mga_freelist_t *head; + drm_mga_freelist_t *tail; + + unsigned int warp_pipe; + unsigned long warp_pipe_phys[MGA_MAX_WARP_PIPES]; + + int chipset; + int usec_timeout; + + /** + * If set, the new DMA initialization sequence was used. This is + * primarilly used to select how the driver should uninitialized its + * internal DMA structures. + */ + int used_new_dma_init; + + /** + * If AGP memory is used for DMA buffers, this will be the value + * \c MGA_PAGPXFER. Otherwise, it will be zero (for a PCI transfer). + */ + u32 dma_access; + + /** + * If AGP memory is used for DMA buffers, this will be the value + * \c MGA_WAGP_ENABLE. Otherwise, it will be zero (for a PCI + * transfer). + */ + u32 wagp_enable; + + /** + * \name MMIO region parameters. + * + * \sa drm_mga_private_t::mmio + */ + /*@{ */ + resource_size_t mmio_base; /**< Bus address of base of MMIO. */ + resource_size_t mmio_size; /**< Size of the MMIO region. */ + /*@} */ + + u32 clear_cmd; + u32 maccess; + + atomic_t vbl_received; /**< Number of vblanks received. */ + wait_queue_head_t fence_queue; + atomic_t last_fence_retired; + u32 next_fence_to_post; + + unsigned int fb_cpp; + unsigned int front_offset; + unsigned int front_pitch; + unsigned int back_offset; + unsigned int back_pitch; + + unsigned int depth_cpp; + unsigned int depth_offset; + unsigned int depth_pitch; + + unsigned int texture_offset; + unsigned int texture_size; + + drm_local_map_t *sarea; + drm_local_map_t *mmio; + drm_local_map_t *status; + drm_local_map_t *warp; + drm_local_map_t *primary; + drm_local_map_t *agp_textures; + + unsigned long agp_handle; + unsigned int agp_size; +} drm_mga_private_t; + +extern struct drm_ioctl_desc mga_ioctls[]; +extern int mga_max_ioctl; + + /* mga_dma.c */ +extern int mga_dma_bootstrap(struct drm_device *dev, void *data, + struct drm_file *file_priv); +extern int mga_dma_init(struct drm_device *dev, void *data, + struct drm_file *file_priv); +extern int mga_dma_flush(struct drm_device *dev, void *data, + struct drm_file *file_priv); +extern int mga_dma_reset(struct drm_device *dev, void *data, + struct drm_file *file_priv); +extern int mga_dma_buffers(struct drm_device *dev, void *data, + struct drm_file *file_priv); +extern int mga_driver_load(struct drm_device *dev, unsigned long flags); +extern int mga_driver_unload(struct drm_device *dev); +extern void mga_driver_lastclose(struct drm_device *dev); +extern int mga_driver_dma_quiescent(struct drm_device *dev); + +extern int mga_do_wait_for_idle(drm_mga_private_t *dev_priv); + +extern void mga_do_dma_flush(drm_mga_private_t *dev_priv); +extern void mga_do_dma_wrap_start(drm_mga_private_t *dev_priv); +extern void mga_do_dma_wrap_end(drm_mga_private_t *dev_priv); + +extern int mga_freelist_put(struct drm_device *dev, struct drm_buf *buf); + + /* mga_warp.c */ +extern int mga_warp_install_microcode(drm_mga_private_t *dev_priv); +extern int mga_warp_init(drm_mga_private_t *dev_priv); + + /* mga_irq.c */ +extern int mga_enable_vblank(struct drm_device *dev, int crtc); +extern void mga_disable_vblank(struct drm_device *dev, int crtc); +extern u32 mga_get_vblank_counter(struct drm_device *dev, int crtc); +extern int mga_driver_fence_wait(struct drm_device *dev, unsigned int *sequence); +extern int mga_driver_vblank_wait(struct drm_device *dev, unsigned int *sequence); +extern irqreturn_t mga_driver_irq_handler(DRM_IRQ_ARGS); +extern void mga_driver_irq_preinstall(struct drm_device *dev); +extern int mga_driver_irq_postinstall(struct drm_device *dev); +extern void mga_driver_irq_uninstall(struct drm_device *dev); +extern long mga_compat_ioctl(struct file *filp, unsigned int cmd, + unsigned long arg); + +#define mga_flush_write_combine() DRM_WRITEMEMORYBARRIER() + +#define MGA_READ8(reg) DRM_READ8(dev_priv->mmio, (reg)) +#define MGA_READ(reg) DRM_READ32(dev_priv->mmio, (reg)) +#define MGA_WRITE8(reg, val) DRM_WRITE8(dev_priv->mmio, (reg), (val)) +#define MGA_WRITE(reg, val) DRM_WRITE32(dev_priv->mmio, (reg), (val)) + +#define DWGREG0 0x1c00 +#define DWGREG0_END 0x1dff +#define DWGREG1 0x2c00 +#define DWGREG1_END 0x2dff + +#define ISREG0(r) (r >= DWGREG0 && r <= DWGREG0_END) +#define DMAREG0(r) (u8)((r - DWGREG0) >> 2) +#define DMAREG1(r) (u8)(((r - DWGREG1) >> 2) | 0x80) +#define DMAREG(r) (ISREG0(r) ? DMAREG0(r) : DMAREG1(r)) + +/* ================================================================ + * Helper macross... + */ + +#define MGA_EMIT_STATE(dev_priv, dirty) \ +do { \ + if ((dirty) & ~MGA_UPLOAD_CLIPRECTS) { \ + if (dev_priv->chipset >= MGA_CARD_TYPE_G400) \ + mga_g400_emit_state(dev_priv); \ + else \ + mga_g200_emit_state(dev_priv); \ + } \ +} while (0) + +#define WRAP_TEST_WITH_RETURN(dev_priv) \ +do { \ + if (test_bit(0, &dev_priv->prim.wrapped)) { \ + if (mga_is_idle(dev_priv)) { \ + mga_do_dma_wrap_end(dev_priv); \ + } else if (dev_priv->prim.space < \ + dev_priv->prim.high_mark) { \ + if (MGA_DMA_DEBUG) \ + DRM_INFO("wrap...\n"); \ + return -EBUSY; \ + } \ + } \ +} while (0) + +#define WRAP_WAIT_WITH_RETURN(dev_priv) \ +do { \ + if (test_bit(0, &dev_priv->prim.wrapped)) { \ + if (mga_do_wait_for_idle(dev_priv) < 0) { \ + if (MGA_DMA_DEBUG) \ + DRM_INFO("wrap...\n"); \ + return -EBUSY; \ + } \ + mga_do_dma_wrap_end(dev_priv); \ + } \ +} while (0) + +/* ================================================================ + * Primary DMA command stream + */ + +#define MGA_VERBOSE 0 + +#define DMA_LOCALS unsigned int write; volatile u8 *prim; + +#define DMA_BLOCK_SIZE (5 * sizeof(u32)) + +#define BEGIN_DMA(n) \ +do { \ + if (MGA_VERBOSE) { \ + DRM_INFO("BEGIN_DMA(%d)\n", (n)); \ + DRM_INFO(" space=0x%x req=0x%Zx\n", \ + dev_priv->prim.space, (n) * DMA_BLOCK_SIZE); \ + } \ + prim = dev_priv->prim.start; \ + write = dev_priv->prim.tail; \ +} while (0) + +#define BEGIN_DMA_WRAP() \ +do { \ + if (MGA_VERBOSE) { \ + DRM_INFO("BEGIN_DMA()\n"); \ + DRM_INFO(" space=0x%x\n", dev_priv->prim.space); \ + } \ + prim = dev_priv->prim.start; \ + write = dev_priv->prim.tail; \ +} while (0) + +#define ADVANCE_DMA() \ +do { \ + dev_priv->prim.tail = write; \ + if (MGA_VERBOSE) \ + DRM_INFO("ADVANCE_DMA() tail=0x%05x sp=0x%x\n", \ + write, dev_priv->prim.space); \ +} while (0) + +#define FLUSH_DMA() \ +do { \ + if (0) { \ + DRM_INFO("\n"); \ + DRM_INFO(" tail=0x%06x head=0x%06lx\n", \ + dev_priv->prim.tail, \ + (unsigned long)(MGA_READ(MGA_PRIMADDRESS) - \ + dev_priv->primary->offset)); \ + } \ + if (!test_bit(0, &dev_priv->prim.wrapped)) { \ + if (dev_priv->prim.space < dev_priv->prim.high_mark) \ + mga_do_dma_wrap_start(dev_priv); \ + else \ + mga_do_dma_flush(dev_priv); \ + } \ +} while (0) + +/* Never use this, always use DMA_BLOCK(...) for primary DMA output. + */ +#define DMA_WRITE(offset, val) \ +do { \ + if (MGA_VERBOSE) \ + DRM_INFO(" DMA_WRITE( 0x%08x ) at 0x%04Zx\n", \ + (u32)(val), write + (offset) * sizeof(u32)); \ + *(volatile u32 *)(prim + write + (offset) * sizeof(u32)) = val; \ +} while (0) + +#define DMA_BLOCK(reg0, val0, reg1, val1, reg2, val2, reg3, val3) \ +do { \ + DMA_WRITE(0, ((DMAREG(reg0) << 0) | \ + (DMAREG(reg1) << 8) | \ + (DMAREG(reg2) << 16) | \ + (DMAREG(reg3) << 24))); \ + DMA_WRITE(1, val0); \ + DMA_WRITE(2, val1); \ + DMA_WRITE(3, val2); \ + DMA_WRITE(4, val3); \ + write += DMA_BLOCK_SIZE; \ +} while (0) + +/* Buffer aging via primary DMA stream head pointer. + */ + +#define SET_AGE(age, h, w) \ +do { \ + (age)->head = h; \ + (age)->wrap = w; \ +} while (0) + +#define TEST_AGE(age, h, w) ((age)->wrap < w || \ + ((age)->wrap == w && \ + (age)->head < h)) + +#define AGE_BUFFER(buf_priv) \ +do { \ + drm_mga_freelist_t *entry = (buf_priv)->list_entry; \ + if ((buf_priv)->dispatched) { \ + entry->age.head = (dev_priv->prim.tail + \ + dev_priv->primary->offset); \ + entry->age.wrap = dev_priv->sarea_priv->last_wrap; \ + } else { \ + entry->age.head = 0; \ + entry->age.wrap = 0; \ + } \ +} while (0) + +#define MGA_ENGINE_IDLE_MASK (MGA_SOFTRAPEN | \ + MGA_DWGENGSTS | \ + MGA_ENDPRDMASTS) +#define MGA_DMA_IDLE_MASK (MGA_SOFTRAPEN | \ + MGA_ENDPRDMASTS) + +#define MGA_DMA_DEBUG 0 + +/* A reduced set of the mga registers. + */ +#define MGA_CRTC_INDEX 0x1fd4 +#define MGA_CRTC_DATA 0x1fd5 + +/* CRTC11 */ +#define MGA_VINTCLR (1 << 4) +#define MGA_VINTEN (1 << 5) + +#define MGA_ALPHACTRL 0x2c7c +#define MGA_AR0 0x1c60 +#define MGA_AR1 0x1c64 +#define MGA_AR2 0x1c68 +#define MGA_AR3 0x1c6c +#define MGA_AR4 0x1c70 +#define MGA_AR5 0x1c74 +#define MGA_AR6 0x1c78 + +#define MGA_CXBNDRY 0x1c80 +#define MGA_CXLEFT 0x1ca0 +#define MGA_CXRIGHT 0x1ca4 + +#define MGA_DMAPAD 0x1c54 +#define MGA_DSTORG 0x2cb8 +#define MGA_DWGCTL 0x1c00 +# define MGA_OPCOD_MASK (15 << 0) +# define MGA_OPCOD_TRAP (4 << 0) +# define MGA_OPCOD_TEXTURE_TRAP (6 << 0) +# define MGA_OPCOD_BITBLT (8 << 0) +# define MGA_OPCOD_ILOAD (9 << 0) +# define MGA_ATYPE_MASK (7 << 4) +# define MGA_ATYPE_RPL (0 << 4) +# define MGA_ATYPE_RSTR (1 << 4) +# define MGA_ATYPE_ZI (3 << 4) +# define MGA_ATYPE_BLK (4 << 4) +# define MGA_ATYPE_I (7 << 4) +# define MGA_LINEAR (1 << 7) +# define MGA_ZMODE_MASK (7 << 8) +# define MGA_ZMODE_NOZCMP (0 << 8) +# define MGA_ZMODE_ZE (2 << 8) +# define MGA_ZMODE_ZNE (3 << 8) +# define MGA_ZMODE_ZLT (4 << 8) +# define MGA_ZMODE_ZLTE (5 << 8) +# define MGA_ZMODE_ZGT (6 << 8) +# define MGA_ZMODE_ZGTE (7 << 8) +# define MGA_SOLID (1 << 11) +# define MGA_ARZERO (1 << 12) +# define MGA_SGNZERO (1 << 13) +# define MGA_SHIFTZERO (1 << 14) +# define MGA_BOP_MASK (15 << 16) +# define MGA_BOP_ZERO (0 << 16) +# define MGA_BOP_DST (10 << 16) +# define MGA_BOP_SRC (12 << 16) +# define MGA_BOP_ONE (15 << 16) +# define MGA_TRANS_SHIFT 20 +# define MGA_TRANS_MASK (15 << 20) +# define MGA_BLTMOD_MASK (15 << 25) +# define MGA_BLTMOD_BMONOLEF (0 << 25) +# define MGA_BLTMOD_BMONOWF (4 << 25) +# define MGA_BLTMOD_PLAN (1 << 25) +# define MGA_BLTMOD_BFCOL (2 << 25) +# define MGA_BLTMOD_BU32BGR (3 << 25) +# define MGA_BLTMOD_BU32RGB (7 << 25) +# define MGA_BLTMOD_BU24BGR (11 << 25) +# define MGA_BLTMOD_BU24RGB (15 << 25) +# define MGA_PATTERN (1 << 29) +# define MGA_TRANSC (1 << 30) +# define MGA_CLIPDIS (1 << 31) +#define MGA_DWGSYNC 0x2c4c + +#define MGA_FCOL 0x1c24 +#define MGA_FIFOSTATUS 0x1e10 +#define MGA_FOGCOL 0x1cf4 +#define MGA_FXBNDRY 0x1c84 +#define MGA_FXLEFT 0x1ca8 +#define MGA_FXRIGHT 0x1cac + +#define MGA_ICLEAR 0x1e18 +# define MGA_SOFTRAPICLR (1 << 0) +# define MGA_VLINEICLR (1 << 5) +#define MGA_IEN 0x1e1c +# define MGA_SOFTRAPIEN (1 << 0) +# define MGA_VLINEIEN (1 << 5) + +#define MGA_LEN 0x1c5c + +#define MGA_MACCESS 0x1c04 + +#define MGA_PITCH 0x1c8c +#define MGA_PLNWT 0x1c1c +#define MGA_PRIMADDRESS 0x1e58 +# define MGA_DMA_GENERAL (0 << 0) +# define MGA_DMA_BLIT (1 << 0) +# define MGA_DMA_VECTOR (2 << 0) +# define MGA_DMA_VERTEX (3 << 0) +#define MGA_PRIMEND 0x1e5c +# define MGA_PRIMNOSTART (1 << 0) +# define MGA_PAGPXFER (1 << 1) +#define MGA_PRIMPTR 0x1e50 +# define MGA_PRIMPTREN0 (1 << 0) +# define MGA_PRIMPTREN1 (1 << 1) + +#define MGA_RST 0x1e40 +# define MGA_SOFTRESET (1 << 0) +# define MGA_SOFTEXTRST (1 << 1) + +#define MGA_SECADDRESS 0x2c40 +#define MGA_SECEND 0x2c44 +#define MGA_SETUPADDRESS 0x2cd0 +#define MGA_SETUPEND 0x2cd4 +#define MGA_SGN 0x1c58 +#define MGA_SOFTRAP 0x2c48 +#define MGA_SRCORG 0x2cb4 +# define MGA_SRMMAP_MASK (1 << 0) +# define MGA_SRCMAP_FB (0 << 0) +# define MGA_SRCMAP_SYSMEM (1 << 0) +# define MGA_SRCACC_MASK (1 << 1) +# define MGA_SRCACC_PCI (0 << 1) +# define MGA_SRCACC_AGP (1 << 1) +#define MGA_STATUS 0x1e14 +# define MGA_SOFTRAPEN (1 << 0) +# define MGA_VSYNCPEN (1 << 4) +# define MGA_VLINEPEN (1 << 5) +# define MGA_DWGENGSTS (1 << 16) +# define MGA_ENDPRDMASTS (1 << 17) +#define MGA_STENCIL 0x2cc8 +#define MGA_STENCILCTL 0x2ccc + +#define MGA_TDUALSTAGE0 0x2cf8 +#define MGA_TDUALSTAGE1 0x2cfc +#define MGA_TEXBORDERCOL 0x2c5c +#define MGA_TEXCTL 0x2c30 +#define MGA_TEXCTL2 0x2c3c +# define MGA_DUALTEX (1 << 7) +# define MGA_G400_TC2_MAGIC (1 << 15) +# define MGA_MAP1_ENABLE (1 << 31) +#define MGA_TEXFILTER 0x2c58 +#define MGA_TEXHEIGHT 0x2c2c +#define MGA_TEXORG 0x2c24 +# define MGA_TEXORGMAP_MASK (1 << 0) +# define MGA_TEXORGMAP_FB (0 << 0) +# define MGA_TEXORGMAP_SYSMEM (1 << 0) +# define MGA_TEXORGACC_MASK (1 << 1) +# define MGA_TEXORGACC_PCI (0 << 1) +# define MGA_TEXORGACC_AGP (1 << 1) +#define MGA_TEXORG1 0x2ca4 +#define MGA_TEXORG2 0x2ca8 +#define MGA_TEXORG3 0x2cac +#define MGA_TEXORG4 0x2cb0 +#define MGA_TEXTRANS 0x2c34 +#define MGA_TEXTRANSHIGH 0x2c38 +#define MGA_TEXWIDTH 0x2c28 + +#define MGA_WACCEPTSEQ 0x1dd4 +#define MGA_WCODEADDR 0x1e6c +#define MGA_WFLAG 0x1dc4 +#define MGA_WFLAG1 0x1de0 +#define MGA_WFLAGNB 0x1e64 +#define MGA_WFLAGNB1 0x1e08 +#define MGA_WGETMSB 0x1dc8 +#define MGA_WIADDR 0x1dc0 +#define MGA_WIADDR2 0x1dd8 +# define MGA_WMODE_SUSPEND (0 << 0) +# define MGA_WMODE_RESUME (1 << 0) +# define MGA_WMODE_JUMP (2 << 0) +# define MGA_WMODE_START (3 << 0) +# define MGA_WAGP_ENABLE (1 << 2) +#define MGA_WMISC 0x1e70 +# define MGA_WUCODECACHE_ENABLE (1 << 0) +# define MGA_WMASTER_ENABLE (1 << 1) +# define MGA_WCACHEFLUSH_ENABLE (1 << 3) +#define MGA_WVRTXSZ 0x1dcc + +#define MGA_YBOT 0x1c9c +#define MGA_YDST 0x1c90 +#define MGA_YDSTLEN 0x1c88 +#define MGA_YDSTORG 0x1c94 +#define MGA_YTOP 0x1c98 + +#define MGA_ZORG 0x1c0c + +/* This finishes the current batch of commands + */ +#define MGA_EXEC 0x0100 + +/* AGP PLL encoding (for G200 only). + */ +#define MGA_AGP_PLL 0x1e4c +# define MGA_AGP2XPLL_DISABLE (0 << 0) +# define MGA_AGP2XPLL_ENABLE (1 << 0) + +/* Warp registers + */ +#define MGA_WR0 0x2d00 +#define MGA_WR1 0x2d04 +#define MGA_WR2 0x2d08 +#define MGA_WR3 0x2d0c +#define MGA_WR4 0x2d10 +#define MGA_WR5 0x2d14 +#define MGA_WR6 0x2d18 +#define MGA_WR7 0x2d1c +#define MGA_WR8 0x2d20 +#define MGA_WR9 0x2d24 +#define MGA_WR10 0x2d28 +#define MGA_WR11 0x2d2c +#define MGA_WR12 0x2d30 +#define MGA_WR13 0x2d34 +#define MGA_WR14 0x2d38 +#define MGA_WR15 0x2d3c +#define MGA_WR16 0x2d40 +#define MGA_WR17 0x2d44 +#define MGA_WR18 0x2d48 +#define MGA_WR19 0x2d4c +#define MGA_WR20 0x2d50 +#define MGA_WR21 0x2d54 +#define MGA_WR22 0x2d58 +#define MGA_WR23 0x2d5c +#define MGA_WR24 0x2d60 +#define MGA_WR25 0x2d64 +#define MGA_WR26 0x2d68 +#define MGA_WR27 0x2d6c +#define MGA_WR28 0x2d70 +#define MGA_WR29 0x2d74 +#define MGA_WR30 0x2d78 +#define MGA_WR31 0x2d7c +#define MGA_WR32 0x2d80 +#define MGA_WR33 0x2d84 +#define MGA_WR34 0x2d88 +#define MGA_WR35 0x2d8c +#define MGA_WR36 0x2d90 +#define MGA_WR37 0x2d94 +#define MGA_WR38 0x2d98 +#define MGA_WR39 0x2d9c +#define MGA_WR40 0x2da0 +#define MGA_WR41 0x2da4 +#define MGA_WR42 0x2da8 +#define MGA_WR43 0x2dac +#define MGA_WR44 0x2db0 +#define MGA_WR45 0x2db4 +#define MGA_WR46 0x2db8 +#define MGA_WR47 0x2dbc +#define MGA_WR48 0x2dc0 +#define MGA_WR49 0x2dc4 +#define MGA_WR50 0x2dc8 +#define MGA_WR51 0x2dcc +#define MGA_WR52 0x2dd0 +#define MGA_WR53 0x2dd4 +#define MGA_WR54 0x2dd8 +#define MGA_WR55 0x2ddc +#define MGA_WR56 0x2de0 +#define MGA_WR57 0x2de4 +#define MGA_WR58 0x2de8 +#define MGA_WR59 0x2dec +#define MGA_WR60 0x2df0 +#define MGA_WR61 0x2df4 +#define MGA_WR62 0x2df8 +#define MGA_WR63 0x2dfc +# define MGA_G400_WR_MAGIC (1 << 6) +# define MGA_G400_WR56_MAGIC 0x46480000 /* 12800.0f */ + +#define MGA_ILOAD_ALIGN 64 +#define MGA_ILOAD_MASK (MGA_ILOAD_ALIGN - 1) + +#define MGA_DWGCTL_FLUSH (MGA_OPCOD_TEXTURE_TRAP | \ + MGA_ATYPE_I | \ + MGA_ZMODE_NOZCMP | \ + MGA_ARZERO | \ + MGA_SGNZERO | \ + MGA_BOP_SRC | \ + (15 << MGA_TRANS_SHIFT)) + +#define MGA_DWGCTL_CLEAR (MGA_OPCOD_TRAP | \ + MGA_ZMODE_NOZCMP | \ + MGA_SOLID | \ + MGA_ARZERO | \ + MGA_SGNZERO | \ + MGA_SHIFTZERO | \ + MGA_BOP_SRC | \ + (0 << MGA_TRANS_SHIFT) | \ + MGA_BLTMOD_BMONOLEF | \ + MGA_TRANSC | \ + MGA_CLIPDIS) + +#define MGA_DWGCTL_COPY (MGA_OPCOD_BITBLT | \ + MGA_ATYPE_RPL | \ + MGA_SGNZERO | \ + MGA_SHIFTZERO | \ + MGA_BOP_SRC | \ + (0 << MGA_TRANS_SHIFT) | \ + MGA_BLTMOD_BFCOL | \ + MGA_CLIPDIS) + +/* Simple idle test. + */ +static __inline__ int mga_is_idle(drm_mga_private_t *dev_priv) +{ + u32 status = MGA_READ(MGA_STATUS) & MGA_ENGINE_IDLE_MASK; + return (status == MGA_ENDPRDMASTS); +} + +#endif diff --git a/drivers/gpu/drm/mga/mga_ioc32.c b/drivers/gpu/drm/mga/mga_ioc32.c new file mode 100644 index 00000000..c1f877b7 --- /dev/null +++ b/drivers/gpu/drm/mga/mga_ioc32.c @@ -0,0 +1,226 @@ +/** + * \file mga_ioc32.c + * + * 32-bit ioctl compatibility routines for the MGA DRM. + * + * \author Dave Airlie with code from patches by Egbert Eich + * + * + * Copyright (C) Paul Mackerras 2005 + * Copyright (C) Egbert Eich 2003,2004 + * Copyright (C) Dave Airlie 2005 + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHOR BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS + * IN THE SOFTWARE. + */ +#include + +#include "drmP.h" +#include "drm.h" +#include "mga_drm.h" + +typedef struct drm32_mga_init { + int func; + u32 sarea_priv_offset; + int chipset; + int sgram; + unsigned int maccess; + unsigned int fb_cpp; + unsigned int front_offset, front_pitch; + unsigned int back_offset, back_pitch; + unsigned int depth_cpp; + unsigned int depth_offset, depth_pitch; + unsigned int texture_offset[MGA_NR_TEX_HEAPS]; + unsigned int texture_size[MGA_NR_TEX_HEAPS]; + u32 fb_offset; + u32 mmio_offset; + u32 status_offset; + u32 warp_offset; + u32 primary_offset; + u32 buffers_offset; +} drm_mga_init32_t; + +static int compat_mga_init(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_mga_init32_t init32; + drm_mga_init_t __user *init; + int err = 0, i; + + if (copy_from_user(&init32, (void __user *)arg, sizeof(init32))) + return -EFAULT; + + init = compat_alloc_user_space(sizeof(*init)); + if (!access_ok(VERIFY_WRITE, init, sizeof(*init)) + || __put_user(init32.func, &init->func) + || __put_user(init32.sarea_priv_offset, &init->sarea_priv_offset) + || __put_user(init32.chipset, &init->chipset) + || __put_user(init32.sgram, &init->sgram) + || __put_user(init32.maccess, &init->maccess) + || __put_user(init32.fb_cpp, &init->fb_cpp) + || __put_user(init32.front_offset, &init->front_offset) + || __put_user(init32.front_pitch, &init->front_pitch) + || __put_user(init32.back_offset, &init->back_offset) + || __put_user(init32.back_pitch, &init->back_pitch) + || __put_user(init32.depth_cpp, &init->depth_cpp) + || __put_user(init32.depth_offset, &init->depth_offset) + || __put_user(init32.depth_pitch, &init->depth_pitch) + || __put_user(init32.fb_offset, &init->fb_offset) + || __put_user(init32.mmio_offset, &init->mmio_offset) + || __put_user(init32.status_offset, &init->status_offset) + || __put_user(init32.warp_offset, &init->warp_offset) + || __put_user(init32.primary_offset, &init->primary_offset) + || __put_user(init32.buffers_offset, &init->buffers_offset)) + return -EFAULT; + + for (i = 0; i < MGA_NR_TEX_HEAPS; i++) { + err |= + __put_user(init32.texture_offset[i], + &init->texture_offset[i]); + err |= + __put_user(init32.texture_size[i], &init->texture_size[i]); + } + if (err) + return -EFAULT; + + return drm_ioctl(file, DRM_IOCTL_MGA_INIT, (unsigned long)init); +} + +typedef struct drm_mga_getparam32 { + int param; + u32 value; +} drm_mga_getparam32_t; + +static int compat_mga_getparam(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_mga_getparam32_t getparam32; + drm_mga_getparam_t __user *getparam; + + if (copy_from_user(&getparam32, (void __user *)arg, sizeof(getparam32))) + return -EFAULT; + + getparam = compat_alloc_user_space(sizeof(*getparam)); + if (!access_ok(VERIFY_WRITE, getparam, sizeof(*getparam)) + || __put_user(getparam32.param, &getparam->param) + || __put_user((void __user *)(unsigned long)getparam32.value, + &getparam->value)) + return -EFAULT; + + return drm_ioctl(file, DRM_IOCTL_MGA_GETPARAM, (unsigned long)getparam); +} + +typedef struct drm_mga_drm_bootstrap32 { + u32 texture_handle; + u32 texture_size; + u32 primary_size; + u32 secondary_bin_count; + u32 secondary_bin_size; + u32 agp_mode; + u8 agp_size; +} drm_mga_dma_bootstrap32_t; + +static int compat_mga_dma_bootstrap(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_mga_dma_bootstrap32_t dma_bootstrap32; + drm_mga_dma_bootstrap_t __user *dma_bootstrap; + int err; + + if (copy_from_user(&dma_bootstrap32, (void __user *)arg, + sizeof(dma_bootstrap32))) + return -EFAULT; + + dma_bootstrap = compat_alloc_user_space(sizeof(*dma_bootstrap)); + if (!access_ok(VERIFY_WRITE, dma_bootstrap, sizeof(*dma_bootstrap)) + || __put_user(dma_bootstrap32.texture_handle, + &dma_bootstrap->texture_handle) + || __put_user(dma_bootstrap32.texture_size, + &dma_bootstrap->texture_size) + || __put_user(dma_bootstrap32.primary_size, + &dma_bootstrap->primary_size) + || __put_user(dma_bootstrap32.secondary_bin_count, + &dma_bootstrap->secondary_bin_count) + || __put_user(dma_bootstrap32.secondary_bin_size, + &dma_bootstrap->secondary_bin_size) + || __put_user(dma_bootstrap32.agp_mode, &dma_bootstrap->agp_mode) + || __put_user(dma_bootstrap32.agp_size, &dma_bootstrap->agp_size)) + return -EFAULT; + + err = drm_ioctl(file, DRM_IOCTL_MGA_DMA_BOOTSTRAP, + (unsigned long)dma_bootstrap); + if (err) + return err; + + if (__get_user(dma_bootstrap32.texture_handle, + &dma_bootstrap->texture_handle) + || __get_user(dma_bootstrap32.texture_size, + &dma_bootstrap->texture_size) + || __get_user(dma_bootstrap32.primary_size, + &dma_bootstrap->primary_size) + || __get_user(dma_bootstrap32.secondary_bin_count, + &dma_bootstrap->secondary_bin_count) + || __get_user(dma_bootstrap32.secondary_bin_size, + &dma_bootstrap->secondary_bin_size) + || __get_user(dma_bootstrap32.agp_mode, &dma_bootstrap->agp_mode) + || __get_user(dma_bootstrap32.agp_size, &dma_bootstrap->agp_size)) + return -EFAULT; + + if (copy_to_user((void __user *)arg, &dma_bootstrap32, + sizeof(dma_bootstrap32))) + return -EFAULT; + + return 0; +} + +drm_ioctl_compat_t *mga_compat_ioctls[] = { + [DRM_MGA_INIT] = compat_mga_init, + [DRM_MGA_GETPARAM] = compat_mga_getparam, + [DRM_MGA_DMA_BOOTSTRAP] = compat_mga_dma_bootstrap, +}; + +/** + * Called whenever a 32-bit process running under a 64-bit kernel + * performs an ioctl on /dev/dri/card. + * + * \param filp file pointer. + * \param cmd command. + * \param arg user argument. + * \return zero on success or negative number on failure. + */ +long mga_compat_ioctl(struct file *filp, unsigned int cmd, unsigned long arg) +{ + unsigned int nr = DRM_IOCTL_NR(cmd); + drm_ioctl_compat_t *fn = NULL; + int ret; + + if (nr < DRM_COMMAND_BASE) + return drm_compat_ioctl(filp, cmd, arg); + + if (nr < DRM_COMMAND_BASE + DRM_ARRAY_SIZE(mga_compat_ioctls)) + fn = mga_compat_ioctls[nr - DRM_COMMAND_BASE]; + + if (fn != NULL) + ret = (*fn) (filp, cmd, arg); + else + ret = drm_ioctl(filp, cmd, arg); + + return ret; +} diff --git a/drivers/gpu/drm/mga/mga_irq.c b/drivers/gpu/drm/mga/mga_irq.c new file mode 100644 index 00000000..25812022 --- /dev/null +++ b/drivers/gpu/drm/mga/mga_irq.c @@ -0,0 +1,174 @@ +/* mga_irq.c -- IRQ handling for radeon -*- linux-c -*- + */ +/* + * Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + * + * The Weather Channel (TM) funded Tungsten Graphics to develop the + * initial release of the Radeon 8500 driver under the XFree86 license. + * This notice must be preserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Keith Whitwell + * Eric Anholt + */ + +#include "drmP.h" +#include "drm.h" +#include "mga_drm.h" +#include "mga_drv.h" + +u32 mga_get_vblank_counter(struct drm_device *dev, int crtc) +{ + const drm_mga_private_t *const dev_priv = + (drm_mga_private_t *) dev->dev_private; + + if (crtc != 0) + return 0; + + return atomic_read(&dev_priv->vbl_received); +} + + +irqreturn_t mga_driver_irq_handler(DRM_IRQ_ARGS) +{ + struct drm_device *dev = (struct drm_device *) arg; + drm_mga_private_t *dev_priv = (drm_mga_private_t *) dev->dev_private; + int status; + int handled = 0; + + status = MGA_READ(MGA_STATUS); + + /* VBLANK interrupt */ + if (status & MGA_VLINEPEN) { + MGA_WRITE(MGA_ICLEAR, MGA_VLINEICLR); + atomic_inc(&dev_priv->vbl_received); + drm_handle_vblank(dev, 0); + handled = 1; + } + + /* SOFTRAP interrupt */ + if (status & MGA_SOFTRAPEN) { + const u32 prim_start = MGA_READ(MGA_PRIMADDRESS); + const u32 prim_end = MGA_READ(MGA_PRIMEND); + + + MGA_WRITE(MGA_ICLEAR, MGA_SOFTRAPICLR); + + /* In addition to clearing the interrupt-pending bit, we + * have to write to MGA_PRIMEND to re-start the DMA operation. + */ + if ((prim_start & ~0x03) != (prim_end & ~0x03)) + MGA_WRITE(MGA_PRIMEND, prim_end); + + atomic_inc(&dev_priv->last_fence_retired); + DRM_WAKEUP(&dev_priv->fence_queue); + handled = 1; + } + + if (handled) + return IRQ_HANDLED; + return IRQ_NONE; +} + +int mga_enable_vblank(struct drm_device *dev, int crtc) +{ + drm_mga_private_t *dev_priv = (drm_mga_private_t *) dev->dev_private; + + if (crtc != 0) { + DRM_ERROR("tried to enable vblank on non-existent crtc %d\n", + crtc); + return 0; + } + + MGA_WRITE(MGA_IEN, MGA_VLINEIEN | MGA_SOFTRAPEN); + return 0; +} + + +void mga_disable_vblank(struct drm_device *dev, int crtc) +{ + if (crtc != 0) { + DRM_ERROR("tried to disable vblank on non-existent crtc %d\n", + crtc); + } + + /* Do *NOT* disable the vertical refresh interrupt. MGA doesn't have + * a nice hardware counter that tracks the number of refreshes when + * the interrupt is disabled, and the kernel doesn't know the refresh + * rate to calculate an estimate. + */ + /* MGA_WRITE(MGA_IEN, MGA_VLINEIEN | MGA_SOFTRAPEN); */ +} + +int mga_driver_fence_wait(struct drm_device *dev, unsigned int *sequence) +{ + drm_mga_private_t *dev_priv = (drm_mga_private_t *) dev->dev_private; + unsigned int cur_fence; + int ret = 0; + + /* Assume that the user has missed the current sequence number + * by about a day rather than she wants to wait for years + * using fences. + */ + DRM_WAIT_ON(ret, dev_priv->fence_queue, 3 * DRM_HZ, + (((cur_fence = atomic_read(&dev_priv->last_fence_retired)) + - *sequence) <= (1 << 23))); + + *sequence = cur_fence; + + return ret; +} + +void mga_driver_irq_preinstall(struct drm_device *dev) +{ + drm_mga_private_t *dev_priv = (drm_mga_private_t *) dev->dev_private; + + /* Disable *all* interrupts */ + MGA_WRITE(MGA_IEN, 0); + /* Clear bits if they're already high */ + MGA_WRITE(MGA_ICLEAR, ~0); +} + +int mga_driver_irq_postinstall(struct drm_device *dev) +{ + drm_mga_private_t *dev_priv = (drm_mga_private_t *) dev->dev_private; + + DRM_INIT_WAITQUEUE(&dev_priv->fence_queue); + + /* Turn on soft trap interrupt. Vertical blank interrupts are enabled + * in mga_enable_vblank. + */ + MGA_WRITE(MGA_IEN, MGA_SOFTRAPEN); + return 0; +} + +void mga_driver_irq_uninstall(struct drm_device *dev) +{ + drm_mga_private_t *dev_priv = (drm_mga_private_t *) dev->dev_private; + if (!dev_priv) + return; + + /* Disable *all* interrupts */ + MGA_WRITE(MGA_IEN, 0); + + dev->irq_enabled = 0; +} diff --git a/drivers/gpu/drm/mga/mga_state.c b/drivers/gpu/drm/mga/mga_state.c new file mode 100644 index 00000000..9ce2827f --- /dev/null +++ b/drivers/gpu/drm/mga/mga_state.c @@ -0,0 +1,1103 @@ +/* mga_state.c -- State support for MGA G200/G400 -*- linux-c -*- + * Created: Thu Jan 27 02:53:43 2000 by jhartmann@precisioninsight.com + * + * Copyright 1999 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * VA LINUX SYSTEMS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: + * Jeff Hartmann + * Keith Whitwell + * + * Rewritten by: + * Gareth Hughes + */ + +#include "drmP.h" +#include "drm.h" +#include "mga_drm.h" +#include "mga_drv.h" + +/* ================================================================ + * DMA hardware state programming functions + */ + +static void mga_emit_clip_rect(drm_mga_private_t *dev_priv, + struct drm_clip_rect *box) +{ + drm_mga_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_mga_context_regs_t *ctx = &sarea_priv->context_state; + unsigned int pitch = dev_priv->front_pitch; + DMA_LOCALS; + + BEGIN_DMA(2); + + /* Force reset of DWGCTL on G400 (eliminates clip disable bit). + */ + if (dev_priv->chipset >= MGA_CARD_TYPE_G400) { + DMA_BLOCK(MGA_DWGCTL, ctx->dwgctl, + MGA_LEN + MGA_EXEC, 0x80000000, + MGA_DWGCTL, ctx->dwgctl, + MGA_LEN + MGA_EXEC, 0x80000000); + } + DMA_BLOCK(MGA_DMAPAD, 0x00000000, + MGA_CXBNDRY, ((box->x2 - 1) << 16) | box->x1, + MGA_YTOP, box->y1 * pitch, MGA_YBOT, (box->y2 - 1) * pitch); + + ADVANCE_DMA(); +} + +static __inline__ void mga_g200_emit_context(drm_mga_private_t *dev_priv) +{ + drm_mga_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_mga_context_regs_t *ctx = &sarea_priv->context_state; + DMA_LOCALS; + + BEGIN_DMA(3); + + DMA_BLOCK(MGA_DSTORG, ctx->dstorg, + MGA_MACCESS, ctx->maccess, + MGA_PLNWT, ctx->plnwt, MGA_DWGCTL, ctx->dwgctl); + + DMA_BLOCK(MGA_ALPHACTRL, ctx->alphactrl, + MGA_FOGCOL, ctx->fogcolor, + MGA_WFLAG, ctx->wflag, MGA_ZORG, dev_priv->depth_offset); + + DMA_BLOCK(MGA_FCOL, ctx->fcol, + MGA_DMAPAD, 0x00000000, + MGA_DMAPAD, 0x00000000, MGA_DMAPAD, 0x00000000); + + ADVANCE_DMA(); +} + +static __inline__ void mga_g400_emit_context(drm_mga_private_t *dev_priv) +{ + drm_mga_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_mga_context_regs_t *ctx = &sarea_priv->context_state; + DMA_LOCALS; + + BEGIN_DMA(4); + + DMA_BLOCK(MGA_DSTORG, ctx->dstorg, + MGA_MACCESS, ctx->maccess, + MGA_PLNWT, ctx->plnwt, MGA_DWGCTL, ctx->dwgctl); + + DMA_BLOCK(MGA_ALPHACTRL, ctx->alphactrl, + MGA_FOGCOL, ctx->fogcolor, + MGA_WFLAG, ctx->wflag, MGA_ZORG, dev_priv->depth_offset); + + DMA_BLOCK(MGA_WFLAG1, ctx->wflag, + MGA_TDUALSTAGE0, ctx->tdualstage0, + MGA_TDUALSTAGE1, ctx->tdualstage1, MGA_FCOL, ctx->fcol); + + DMA_BLOCK(MGA_STENCIL, ctx->stencil, + MGA_STENCILCTL, ctx->stencilctl, + MGA_DMAPAD, 0x00000000, MGA_DMAPAD, 0x00000000); + + ADVANCE_DMA(); +} + +static __inline__ void mga_g200_emit_tex0(drm_mga_private_t *dev_priv) +{ + drm_mga_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_mga_texture_regs_t *tex = &sarea_priv->tex_state[0]; + DMA_LOCALS; + + BEGIN_DMA(4); + + DMA_BLOCK(MGA_TEXCTL2, tex->texctl2, + MGA_TEXCTL, tex->texctl, + MGA_TEXFILTER, tex->texfilter, + MGA_TEXBORDERCOL, tex->texbordercol); + + DMA_BLOCK(MGA_TEXORG, tex->texorg, + MGA_TEXORG1, tex->texorg1, + MGA_TEXORG2, tex->texorg2, MGA_TEXORG3, tex->texorg3); + + DMA_BLOCK(MGA_TEXORG4, tex->texorg4, + MGA_TEXWIDTH, tex->texwidth, + MGA_TEXHEIGHT, tex->texheight, MGA_WR24, tex->texwidth); + + DMA_BLOCK(MGA_WR34, tex->texheight, + MGA_TEXTRANS, 0x0000ffff, + MGA_TEXTRANSHIGH, 0x0000ffff, MGA_DMAPAD, 0x00000000); + + ADVANCE_DMA(); +} + +static __inline__ void mga_g400_emit_tex0(drm_mga_private_t *dev_priv) +{ + drm_mga_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_mga_texture_regs_t *tex = &sarea_priv->tex_state[0]; + DMA_LOCALS; + +/* printk("mga_g400_emit_tex0 %x %x %x\n", tex->texorg, */ +/* tex->texctl, tex->texctl2); */ + + BEGIN_DMA(6); + + DMA_BLOCK(MGA_TEXCTL2, tex->texctl2 | MGA_G400_TC2_MAGIC, + MGA_TEXCTL, tex->texctl, + MGA_TEXFILTER, tex->texfilter, + MGA_TEXBORDERCOL, tex->texbordercol); + + DMA_BLOCK(MGA_TEXORG, tex->texorg, + MGA_TEXORG1, tex->texorg1, + MGA_TEXORG2, tex->texorg2, MGA_TEXORG3, tex->texorg3); + + DMA_BLOCK(MGA_TEXORG4, tex->texorg4, + MGA_TEXWIDTH, tex->texwidth, + MGA_TEXHEIGHT, tex->texheight, MGA_WR49, 0x00000000); + + DMA_BLOCK(MGA_WR57, 0x00000000, + MGA_WR53, 0x00000000, + MGA_WR61, 0x00000000, MGA_WR52, MGA_G400_WR_MAGIC); + + DMA_BLOCK(MGA_WR60, MGA_G400_WR_MAGIC, + MGA_WR54, tex->texwidth | MGA_G400_WR_MAGIC, + MGA_WR62, tex->texheight | MGA_G400_WR_MAGIC, + MGA_DMAPAD, 0x00000000); + + DMA_BLOCK(MGA_DMAPAD, 0x00000000, + MGA_DMAPAD, 0x00000000, + MGA_TEXTRANS, 0x0000ffff, MGA_TEXTRANSHIGH, 0x0000ffff); + + ADVANCE_DMA(); +} + +static __inline__ void mga_g400_emit_tex1(drm_mga_private_t *dev_priv) +{ + drm_mga_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_mga_texture_regs_t *tex = &sarea_priv->tex_state[1]; + DMA_LOCALS; + +/* printk("mga_g400_emit_tex1 %x %x %x\n", tex->texorg, */ +/* tex->texctl, tex->texctl2); */ + + BEGIN_DMA(5); + + DMA_BLOCK(MGA_TEXCTL2, (tex->texctl2 | + MGA_MAP1_ENABLE | + MGA_G400_TC2_MAGIC), + MGA_TEXCTL, tex->texctl, + MGA_TEXFILTER, tex->texfilter, + MGA_TEXBORDERCOL, tex->texbordercol); + + DMA_BLOCK(MGA_TEXORG, tex->texorg, + MGA_TEXORG1, tex->texorg1, + MGA_TEXORG2, tex->texorg2, MGA_TEXORG3, tex->texorg3); + + DMA_BLOCK(MGA_TEXORG4, tex->texorg4, + MGA_TEXWIDTH, tex->texwidth, + MGA_TEXHEIGHT, tex->texheight, MGA_WR49, 0x00000000); + + DMA_BLOCK(MGA_WR57, 0x00000000, + MGA_WR53, 0x00000000, + MGA_WR61, 0x00000000, + MGA_WR52, tex->texwidth | MGA_G400_WR_MAGIC); + + DMA_BLOCK(MGA_WR60, tex->texheight | MGA_G400_WR_MAGIC, + MGA_TEXTRANS, 0x0000ffff, + MGA_TEXTRANSHIGH, 0x0000ffff, + MGA_TEXCTL2, tex->texctl2 | MGA_G400_TC2_MAGIC); + + ADVANCE_DMA(); +} + +static __inline__ void mga_g200_emit_pipe(drm_mga_private_t *dev_priv) +{ + drm_mga_sarea_t *sarea_priv = dev_priv->sarea_priv; + unsigned int pipe = sarea_priv->warp_pipe; + DMA_LOCALS; + + BEGIN_DMA(3); + + DMA_BLOCK(MGA_WIADDR, MGA_WMODE_SUSPEND, + MGA_WVRTXSZ, 0x00000007, + MGA_WFLAG, 0x00000000, MGA_WR24, 0x00000000); + + DMA_BLOCK(MGA_WR25, 0x00000100, + MGA_WR34, 0x00000000, + MGA_WR42, 0x0000ffff, MGA_WR60, 0x0000ffff); + + /* Padding required due to hardware bug. + */ + DMA_BLOCK(MGA_DMAPAD, 0xffffffff, + MGA_DMAPAD, 0xffffffff, + MGA_DMAPAD, 0xffffffff, + MGA_WIADDR, (dev_priv->warp_pipe_phys[pipe] | + MGA_WMODE_START | dev_priv->wagp_enable)); + + ADVANCE_DMA(); +} + +static __inline__ void mga_g400_emit_pipe(drm_mga_private_t *dev_priv) +{ + drm_mga_sarea_t *sarea_priv = dev_priv->sarea_priv; + unsigned int pipe = sarea_priv->warp_pipe; + DMA_LOCALS; + +/* printk("mga_g400_emit_pipe %x\n", pipe); */ + + BEGIN_DMA(10); + + DMA_BLOCK(MGA_WIADDR2, MGA_WMODE_SUSPEND, + MGA_DMAPAD, 0x00000000, + MGA_DMAPAD, 0x00000000, MGA_DMAPAD, 0x00000000); + + if (pipe & MGA_T2) { + DMA_BLOCK(MGA_WVRTXSZ, 0x00001e09, + MGA_DMAPAD, 0x00000000, + MGA_DMAPAD, 0x00000000, MGA_DMAPAD, 0x00000000); + + DMA_BLOCK(MGA_WACCEPTSEQ, 0x00000000, + MGA_WACCEPTSEQ, 0x00000000, + MGA_WACCEPTSEQ, 0x00000000, + MGA_WACCEPTSEQ, 0x1e000000); + } else { + if (dev_priv->warp_pipe & MGA_T2) { + /* Flush the WARP pipe */ + DMA_BLOCK(MGA_YDST, 0x00000000, + MGA_FXLEFT, 0x00000000, + MGA_FXRIGHT, 0x00000001, + MGA_DWGCTL, MGA_DWGCTL_FLUSH); + + DMA_BLOCK(MGA_LEN + MGA_EXEC, 0x00000001, + MGA_DWGSYNC, 0x00007000, + MGA_TEXCTL2, MGA_G400_TC2_MAGIC, + MGA_LEN + MGA_EXEC, 0x00000000); + + DMA_BLOCK(MGA_TEXCTL2, (MGA_DUALTEX | + MGA_G400_TC2_MAGIC), + MGA_LEN + MGA_EXEC, 0x00000000, + MGA_TEXCTL2, MGA_G400_TC2_MAGIC, + MGA_DMAPAD, 0x00000000); + } + + DMA_BLOCK(MGA_WVRTXSZ, 0x00001807, + MGA_DMAPAD, 0x00000000, + MGA_DMAPAD, 0x00000000, MGA_DMAPAD, 0x00000000); + + DMA_BLOCK(MGA_WACCEPTSEQ, 0x00000000, + MGA_WACCEPTSEQ, 0x00000000, + MGA_WACCEPTSEQ, 0x00000000, + MGA_WACCEPTSEQ, 0x18000000); + } + + DMA_BLOCK(MGA_WFLAG, 0x00000000, + MGA_WFLAG1, 0x00000000, + MGA_WR56, MGA_G400_WR56_MAGIC, MGA_DMAPAD, 0x00000000); + + DMA_BLOCK(MGA_WR49, 0x00000000, /* tex0 */ + MGA_WR57, 0x00000000, /* tex0 */ + MGA_WR53, 0x00000000, /* tex1 */ + MGA_WR61, 0x00000000); /* tex1 */ + + DMA_BLOCK(MGA_WR54, MGA_G400_WR_MAGIC, /* tex0 width */ + MGA_WR62, MGA_G400_WR_MAGIC, /* tex0 height */ + MGA_WR52, MGA_G400_WR_MAGIC, /* tex1 width */ + MGA_WR60, MGA_G400_WR_MAGIC); /* tex1 height */ + + /* Padding required due to hardware bug */ + DMA_BLOCK(MGA_DMAPAD, 0xffffffff, + MGA_DMAPAD, 0xffffffff, + MGA_DMAPAD, 0xffffffff, + MGA_WIADDR2, (dev_priv->warp_pipe_phys[pipe] | + MGA_WMODE_START | dev_priv->wagp_enable)); + + ADVANCE_DMA(); +} + +static void mga_g200_emit_state(drm_mga_private_t *dev_priv) +{ + drm_mga_sarea_t *sarea_priv = dev_priv->sarea_priv; + unsigned int dirty = sarea_priv->dirty; + + if (sarea_priv->warp_pipe != dev_priv->warp_pipe) { + mga_g200_emit_pipe(dev_priv); + dev_priv->warp_pipe = sarea_priv->warp_pipe; + } + + if (dirty & MGA_UPLOAD_CONTEXT) { + mga_g200_emit_context(dev_priv); + sarea_priv->dirty &= ~MGA_UPLOAD_CONTEXT; + } + + if (dirty & MGA_UPLOAD_TEX0) { + mga_g200_emit_tex0(dev_priv); + sarea_priv->dirty &= ~MGA_UPLOAD_TEX0; + } +} + +static void mga_g400_emit_state(drm_mga_private_t *dev_priv) +{ + drm_mga_sarea_t *sarea_priv = dev_priv->sarea_priv; + unsigned int dirty = sarea_priv->dirty; + int multitex = sarea_priv->warp_pipe & MGA_T2; + + if (sarea_priv->warp_pipe != dev_priv->warp_pipe) { + mga_g400_emit_pipe(dev_priv); + dev_priv->warp_pipe = sarea_priv->warp_pipe; + } + + if (dirty & MGA_UPLOAD_CONTEXT) { + mga_g400_emit_context(dev_priv); + sarea_priv->dirty &= ~MGA_UPLOAD_CONTEXT; + } + + if (dirty & MGA_UPLOAD_TEX0) { + mga_g400_emit_tex0(dev_priv); + sarea_priv->dirty &= ~MGA_UPLOAD_TEX0; + } + + if ((dirty & MGA_UPLOAD_TEX1) && multitex) { + mga_g400_emit_tex1(dev_priv); + sarea_priv->dirty &= ~MGA_UPLOAD_TEX1; + } +} + +/* ================================================================ + * SAREA state verification + */ + +/* Disallow all write destinations except the front and backbuffer. + */ +static int mga_verify_context(drm_mga_private_t *dev_priv) +{ + drm_mga_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_mga_context_regs_t *ctx = &sarea_priv->context_state; + + if (ctx->dstorg != dev_priv->front_offset && + ctx->dstorg != dev_priv->back_offset) { + DRM_ERROR("*** bad DSTORG: %x (front %x, back %x)\n\n", + ctx->dstorg, dev_priv->front_offset, + dev_priv->back_offset); + ctx->dstorg = 0; + return -EINVAL; + } + + return 0; +} + +/* Disallow texture reads from PCI space. + */ +static int mga_verify_tex(drm_mga_private_t *dev_priv, int unit) +{ + drm_mga_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_mga_texture_regs_t *tex = &sarea_priv->tex_state[unit]; + unsigned int org; + + org = tex->texorg & (MGA_TEXORGMAP_MASK | MGA_TEXORGACC_MASK); + + if (org == (MGA_TEXORGMAP_SYSMEM | MGA_TEXORGACC_PCI)) { + DRM_ERROR("*** bad TEXORG: 0x%x, unit %d\n", tex->texorg, unit); + tex->texorg = 0; + return -EINVAL; + } + + return 0; +} + +static int mga_verify_state(drm_mga_private_t *dev_priv) +{ + drm_mga_sarea_t *sarea_priv = dev_priv->sarea_priv; + unsigned int dirty = sarea_priv->dirty; + int ret = 0; + + if (sarea_priv->nbox > MGA_NR_SAREA_CLIPRECTS) + sarea_priv->nbox = MGA_NR_SAREA_CLIPRECTS; + + if (dirty & MGA_UPLOAD_CONTEXT) + ret |= mga_verify_context(dev_priv); + + if (dirty & MGA_UPLOAD_TEX0) + ret |= mga_verify_tex(dev_priv, 0); + + if (dev_priv->chipset >= MGA_CARD_TYPE_G400) { + if (dirty & MGA_UPLOAD_TEX1) + ret |= mga_verify_tex(dev_priv, 1); + + if (dirty & MGA_UPLOAD_PIPE) + ret |= (sarea_priv->warp_pipe > MGA_MAX_G400_PIPES); + } else { + if (dirty & MGA_UPLOAD_PIPE) + ret |= (sarea_priv->warp_pipe > MGA_MAX_G200_PIPES); + } + + return (ret == 0); +} + +static int mga_verify_iload(drm_mga_private_t *dev_priv, + unsigned int dstorg, unsigned int length) +{ + if (dstorg < dev_priv->texture_offset || + dstorg + length > (dev_priv->texture_offset + + dev_priv->texture_size)) { + DRM_ERROR("*** bad iload DSTORG: 0x%x\n", dstorg); + return -EINVAL; + } + + if (length & MGA_ILOAD_MASK) { + DRM_ERROR("*** bad iload length: 0x%x\n", + length & MGA_ILOAD_MASK); + return -EINVAL; + } + + return 0; +} + +static int mga_verify_blit(drm_mga_private_t *dev_priv, + unsigned int srcorg, unsigned int dstorg) +{ + if ((srcorg & 0x3) == (MGA_SRCACC_PCI | MGA_SRCMAP_SYSMEM) || + (dstorg & 0x3) == (MGA_SRCACC_PCI | MGA_SRCMAP_SYSMEM)) { + DRM_ERROR("*** bad blit: src=0x%x dst=0x%x\n", srcorg, dstorg); + return -EINVAL; + } + return 0; +} + +/* ================================================================ + * + */ + +static void mga_dma_dispatch_clear(struct drm_device *dev, drm_mga_clear_t *clear) +{ + drm_mga_private_t *dev_priv = dev->dev_private; + drm_mga_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_mga_context_regs_t *ctx = &sarea_priv->context_state; + struct drm_clip_rect *pbox = sarea_priv->boxes; + int nbox = sarea_priv->nbox; + int i; + DMA_LOCALS; + DRM_DEBUG("\n"); + + BEGIN_DMA(1); + + DMA_BLOCK(MGA_DMAPAD, 0x00000000, + MGA_DMAPAD, 0x00000000, + MGA_DWGSYNC, 0x00007100, MGA_DWGSYNC, 0x00007000); + + ADVANCE_DMA(); + + for (i = 0; i < nbox; i++) { + struct drm_clip_rect *box = &pbox[i]; + u32 height = box->y2 - box->y1; + + DRM_DEBUG(" from=%d,%d to=%d,%d\n", + box->x1, box->y1, box->x2, box->y2); + + if (clear->flags & MGA_FRONT) { + BEGIN_DMA(2); + + DMA_BLOCK(MGA_DMAPAD, 0x00000000, + MGA_PLNWT, clear->color_mask, + MGA_YDSTLEN, (box->y1 << 16) | height, + MGA_FXBNDRY, (box->x2 << 16) | box->x1); + + DMA_BLOCK(MGA_DMAPAD, 0x00000000, + MGA_FCOL, clear->clear_color, + MGA_DSTORG, dev_priv->front_offset, + MGA_DWGCTL + MGA_EXEC, dev_priv->clear_cmd); + + ADVANCE_DMA(); + } + + if (clear->flags & MGA_BACK) { + BEGIN_DMA(2); + + DMA_BLOCK(MGA_DMAPAD, 0x00000000, + MGA_PLNWT, clear->color_mask, + MGA_YDSTLEN, (box->y1 << 16) | height, + MGA_FXBNDRY, (box->x2 << 16) | box->x1); + + DMA_BLOCK(MGA_DMAPAD, 0x00000000, + MGA_FCOL, clear->clear_color, + MGA_DSTORG, dev_priv->back_offset, + MGA_DWGCTL + MGA_EXEC, dev_priv->clear_cmd); + + ADVANCE_DMA(); + } + + if (clear->flags & MGA_DEPTH) { + BEGIN_DMA(2); + + DMA_BLOCK(MGA_DMAPAD, 0x00000000, + MGA_PLNWT, clear->depth_mask, + MGA_YDSTLEN, (box->y1 << 16) | height, + MGA_FXBNDRY, (box->x2 << 16) | box->x1); + + DMA_BLOCK(MGA_DMAPAD, 0x00000000, + MGA_FCOL, clear->clear_depth, + MGA_DSTORG, dev_priv->depth_offset, + MGA_DWGCTL + MGA_EXEC, dev_priv->clear_cmd); + + ADVANCE_DMA(); + } + + } + + BEGIN_DMA(1); + + /* Force reset of DWGCTL */ + DMA_BLOCK(MGA_DMAPAD, 0x00000000, + MGA_DMAPAD, 0x00000000, + MGA_PLNWT, ctx->plnwt, MGA_DWGCTL, ctx->dwgctl); + + ADVANCE_DMA(); + + FLUSH_DMA(); +} + +static void mga_dma_dispatch_swap(struct drm_device *dev) +{ + drm_mga_private_t *dev_priv = dev->dev_private; + drm_mga_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_mga_context_regs_t *ctx = &sarea_priv->context_state; + struct drm_clip_rect *pbox = sarea_priv->boxes; + int nbox = sarea_priv->nbox; + int i; + DMA_LOCALS; + DRM_DEBUG("\n"); + + sarea_priv->last_frame.head = dev_priv->prim.tail; + sarea_priv->last_frame.wrap = dev_priv->prim.last_wrap; + + BEGIN_DMA(4 + nbox); + + DMA_BLOCK(MGA_DMAPAD, 0x00000000, + MGA_DMAPAD, 0x00000000, + MGA_DWGSYNC, 0x00007100, MGA_DWGSYNC, 0x00007000); + + DMA_BLOCK(MGA_DSTORG, dev_priv->front_offset, + MGA_MACCESS, dev_priv->maccess, + MGA_SRCORG, dev_priv->back_offset, + MGA_AR5, dev_priv->front_pitch); + + DMA_BLOCK(MGA_DMAPAD, 0x00000000, + MGA_DMAPAD, 0x00000000, + MGA_PLNWT, 0xffffffff, MGA_DWGCTL, MGA_DWGCTL_COPY); + + for (i = 0; i < nbox; i++) { + struct drm_clip_rect *box = &pbox[i]; + u32 height = box->y2 - box->y1; + u32 start = box->y1 * dev_priv->front_pitch; + + DRM_DEBUG(" from=%d,%d to=%d,%d\n", + box->x1, box->y1, box->x2, box->y2); + + DMA_BLOCK(MGA_AR0, start + box->x2 - 1, + MGA_AR3, start + box->x1, + MGA_FXBNDRY, ((box->x2 - 1) << 16) | box->x1, + MGA_YDSTLEN + MGA_EXEC, (box->y1 << 16) | height); + } + + DMA_BLOCK(MGA_DMAPAD, 0x00000000, + MGA_PLNWT, ctx->plnwt, + MGA_SRCORG, dev_priv->front_offset, MGA_DWGCTL, ctx->dwgctl); + + ADVANCE_DMA(); + + FLUSH_DMA(); + + DRM_DEBUG("... done.\n"); +} + +static void mga_dma_dispatch_vertex(struct drm_device *dev, struct drm_buf *buf) +{ + drm_mga_private_t *dev_priv = dev->dev_private; + drm_mga_buf_priv_t *buf_priv = buf->dev_private; + drm_mga_sarea_t *sarea_priv = dev_priv->sarea_priv; + u32 address = (u32) buf->bus_address; + u32 length = (u32) buf->used; + int i = 0; + DMA_LOCALS; + DRM_DEBUG("buf=%d used=%d\n", buf->idx, buf->used); + + if (buf->used) { + buf_priv->dispatched = 1; + + MGA_EMIT_STATE(dev_priv, sarea_priv->dirty); + + do { + if (i < sarea_priv->nbox) { + mga_emit_clip_rect(dev_priv, + &sarea_priv->boxes[i]); + } + + BEGIN_DMA(1); + + DMA_BLOCK(MGA_DMAPAD, 0x00000000, + MGA_DMAPAD, 0x00000000, + MGA_SECADDRESS, (address | + MGA_DMA_VERTEX), + MGA_SECEND, ((address + length) | + dev_priv->dma_access)); + + ADVANCE_DMA(); + } while (++i < sarea_priv->nbox); + } + + if (buf_priv->discard) { + AGE_BUFFER(buf_priv); + buf->pending = 0; + buf->used = 0; + buf_priv->dispatched = 0; + + mga_freelist_put(dev, buf); + } + + FLUSH_DMA(); +} + +static void mga_dma_dispatch_indices(struct drm_device *dev, struct drm_buf *buf, + unsigned int start, unsigned int end) +{ + drm_mga_private_t *dev_priv = dev->dev_private; + drm_mga_buf_priv_t *buf_priv = buf->dev_private; + drm_mga_sarea_t *sarea_priv = dev_priv->sarea_priv; + u32 address = (u32) buf->bus_address; + int i = 0; + DMA_LOCALS; + DRM_DEBUG("buf=%d start=%d end=%d\n", buf->idx, start, end); + + if (start != end) { + buf_priv->dispatched = 1; + + MGA_EMIT_STATE(dev_priv, sarea_priv->dirty); + + do { + if (i < sarea_priv->nbox) { + mga_emit_clip_rect(dev_priv, + &sarea_priv->boxes[i]); + } + + BEGIN_DMA(1); + + DMA_BLOCK(MGA_DMAPAD, 0x00000000, + MGA_DMAPAD, 0x00000000, + MGA_SETUPADDRESS, address + start, + MGA_SETUPEND, ((address + end) | + dev_priv->dma_access)); + + ADVANCE_DMA(); + } while (++i < sarea_priv->nbox); + } + + if (buf_priv->discard) { + AGE_BUFFER(buf_priv); + buf->pending = 0; + buf->used = 0; + buf_priv->dispatched = 0; + + mga_freelist_put(dev, buf); + } + + FLUSH_DMA(); +} + +/* This copies a 64 byte aligned agp region to the frambuffer with a + * standard blit, the ioctl needs to do checking. + */ +static void mga_dma_dispatch_iload(struct drm_device *dev, struct drm_buf *buf, + unsigned int dstorg, unsigned int length) +{ + drm_mga_private_t *dev_priv = dev->dev_private; + drm_mga_buf_priv_t *buf_priv = buf->dev_private; + drm_mga_context_regs_t *ctx = &dev_priv->sarea_priv->context_state; + u32 srcorg = + buf->bus_address | dev_priv->dma_access | MGA_SRCMAP_SYSMEM; + u32 y2; + DMA_LOCALS; + DRM_DEBUG("buf=%d used=%d\n", buf->idx, buf->used); + + y2 = length / 64; + + BEGIN_DMA(5); + + DMA_BLOCK(MGA_DMAPAD, 0x00000000, + MGA_DMAPAD, 0x00000000, + MGA_DWGSYNC, 0x00007100, MGA_DWGSYNC, 0x00007000); + + DMA_BLOCK(MGA_DSTORG, dstorg, + MGA_MACCESS, 0x00000000, MGA_SRCORG, srcorg, MGA_AR5, 64); + + DMA_BLOCK(MGA_PITCH, 64, + MGA_PLNWT, 0xffffffff, + MGA_DMAPAD, 0x00000000, MGA_DWGCTL, MGA_DWGCTL_COPY); + + DMA_BLOCK(MGA_AR0, 63, + MGA_AR3, 0, + MGA_FXBNDRY, (63 << 16) | 0, MGA_YDSTLEN + MGA_EXEC, y2); + + DMA_BLOCK(MGA_PLNWT, ctx->plnwt, + MGA_SRCORG, dev_priv->front_offset, + MGA_PITCH, dev_priv->front_pitch, MGA_DWGSYNC, 0x00007000); + + ADVANCE_DMA(); + + AGE_BUFFER(buf_priv); + + buf->pending = 0; + buf->used = 0; + buf_priv->dispatched = 0; + + mga_freelist_put(dev, buf); + + FLUSH_DMA(); +} + +static void mga_dma_dispatch_blit(struct drm_device *dev, drm_mga_blit_t *blit) +{ + drm_mga_private_t *dev_priv = dev->dev_private; + drm_mga_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_mga_context_regs_t *ctx = &sarea_priv->context_state; + struct drm_clip_rect *pbox = sarea_priv->boxes; + int nbox = sarea_priv->nbox; + u32 scandir = 0, i; + DMA_LOCALS; + DRM_DEBUG("\n"); + + BEGIN_DMA(4 + nbox); + + DMA_BLOCK(MGA_DMAPAD, 0x00000000, + MGA_DMAPAD, 0x00000000, + MGA_DWGSYNC, 0x00007100, MGA_DWGSYNC, 0x00007000); + + DMA_BLOCK(MGA_DWGCTL, MGA_DWGCTL_COPY, + MGA_PLNWT, blit->planemask, + MGA_SRCORG, blit->srcorg, MGA_DSTORG, blit->dstorg); + + DMA_BLOCK(MGA_SGN, scandir, + MGA_MACCESS, dev_priv->maccess, + MGA_AR5, blit->ydir * blit->src_pitch, + MGA_PITCH, blit->dst_pitch); + + for (i = 0; i < nbox; i++) { + int srcx = pbox[i].x1 + blit->delta_sx; + int srcy = pbox[i].y1 + blit->delta_sy; + int dstx = pbox[i].x1 + blit->delta_dx; + int dsty = pbox[i].y1 + blit->delta_dy; + int h = pbox[i].y2 - pbox[i].y1; + int w = pbox[i].x2 - pbox[i].x1 - 1; + int start; + + if (blit->ydir == -1) + srcy = blit->height - srcy - 1; + + start = srcy * blit->src_pitch + srcx; + + DMA_BLOCK(MGA_AR0, start + w, + MGA_AR3, start, + MGA_FXBNDRY, ((dstx + w) << 16) | (dstx & 0xffff), + MGA_YDSTLEN + MGA_EXEC, (dsty << 16) | h); + } + + /* Do something to flush AGP? + */ + + /* Force reset of DWGCTL */ + DMA_BLOCK(MGA_DMAPAD, 0x00000000, + MGA_PLNWT, ctx->plnwt, + MGA_PITCH, dev_priv->front_pitch, MGA_DWGCTL, ctx->dwgctl); + + ADVANCE_DMA(); +} + +/* ================================================================ + * + */ + +static int mga_dma_clear(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_mga_private_t *dev_priv = dev->dev_private; + drm_mga_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_mga_clear_t *clear = data; + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + if (sarea_priv->nbox > MGA_NR_SAREA_CLIPRECTS) + sarea_priv->nbox = MGA_NR_SAREA_CLIPRECTS; + + WRAP_TEST_WITH_RETURN(dev_priv); + + mga_dma_dispatch_clear(dev, clear); + + /* Make sure we restore the 3D state next time. + */ + dev_priv->sarea_priv->dirty |= MGA_UPLOAD_CONTEXT; + + return 0; +} + +static int mga_dma_swap(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_mga_private_t *dev_priv = dev->dev_private; + drm_mga_sarea_t *sarea_priv = dev_priv->sarea_priv; + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + if (sarea_priv->nbox > MGA_NR_SAREA_CLIPRECTS) + sarea_priv->nbox = MGA_NR_SAREA_CLIPRECTS; + + WRAP_TEST_WITH_RETURN(dev_priv); + + mga_dma_dispatch_swap(dev); + + /* Make sure we restore the 3D state next time. + */ + dev_priv->sarea_priv->dirty |= MGA_UPLOAD_CONTEXT; + + return 0; +} + +static int mga_dma_vertex(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_mga_private_t *dev_priv = dev->dev_private; + struct drm_device_dma *dma = dev->dma; + struct drm_buf *buf; + drm_mga_buf_priv_t *buf_priv; + drm_mga_vertex_t *vertex = data; + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + if (vertex->idx < 0 || vertex->idx > dma->buf_count) + return -EINVAL; + buf = dma->buflist[vertex->idx]; + buf_priv = buf->dev_private; + + buf->used = vertex->used; + buf_priv->discard = vertex->discard; + + if (!mga_verify_state(dev_priv)) { + if (vertex->discard) { + if (buf_priv->dispatched == 1) + AGE_BUFFER(buf_priv); + buf_priv->dispatched = 0; + mga_freelist_put(dev, buf); + } + return -EINVAL; + } + + WRAP_TEST_WITH_RETURN(dev_priv); + + mga_dma_dispatch_vertex(dev, buf); + + return 0; +} + +static int mga_dma_indices(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_mga_private_t *dev_priv = dev->dev_private; + struct drm_device_dma *dma = dev->dma; + struct drm_buf *buf; + drm_mga_buf_priv_t *buf_priv; + drm_mga_indices_t *indices = data; + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + if (indices->idx < 0 || indices->idx > dma->buf_count) + return -EINVAL; + + buf = dma->buflist[indices->idx]; + buf_priv = buf->dev_private; + + buf_priv->discard = indices->discard; + + if (!mga_verify_state(dev_priv)) { + if (indices->discard) { + if (buf_priv->dispatched == 1) + AGE_BUFFER(buf_priv); + buf_priv->dispatched = 0; + mga_freelist_put(dev, buf); + } + return -EINVAL; + } + + WRAP_TEST_WITH_RETURN(dev_priv); + + mga_dma_dispatch_indices(dev, buf, indices->start, indices->end); + + return 0; +} + +static int mga_dma_iload(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + struct drm_device_dma *dma = dev->dma; + drm_mga_private_t *dev_priv = dev->dev_private; + struct drm_buf *buf; + drm_mga_buf_priv_t *buf_priv; + drm_mga_iload_t *iload = data; + DRM_DEBUG("\n"); + + LOCK_TEST_WITH_RETURN(dev, file_priv); + +#if 0 + if (mga_do_wait_for_idle(dev_priv) < 0) { + if (MGA_DMA_DEBUG) + DRM_INFO("-EBUSY\n"); + return -EBUSY; + } +#endif + if (iload->idx < 0 || iload->idx > dma->buf_count) + return -EINVAL; + + buf = dma->buflist[iload->idx]; + buf_priv = buf->dev_private; + + if (mga_verify_iload(dev_priv, iload->dstorg, iload->length)) { + mga_freelist_put(dev, buf); + return -EINVAL; + } + + WRAP_TEST_WITH_RETURN(dev_priv); + + mga_dma_dispatch_iload(dev, buf, iload->dstorg, iload->length); + + /* Make sure we restore the 3D state next time. + */ + dev_priv->sarea_priv->dirty |= MGA_UPLOAD_CONTEXT; + + return 0; +} + +static int mga_dma_blit(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_mga_private_t *dev_priv = dev->dev_private; + drm_mga_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_mga_blit_t *blit = data; + DRM_DEBUG("\n"); + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + if (sarea_priv->nbox > MGA_NR_SAREA_CLIPRECTS) + sarea_priv->nbox = MGA_NR_SAREA_CLIPRECTS; + + if (mga_verify_blit(dev_priv, blit->srcorg, blit->dstorg)) + return -EINVAL; + + WRAP_TEST_WITH_RETURN(dev_priv); + + mga_dma_dispatch_blit(dev, blit); + + /* Make sure we restore the 3D state next time. + */ + dev_priv->sarea_priv->dirty |= MGA_UPLOAD_CONTEXT; + + return 0; +} + +static int mga_getparam(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_mga_private_t *dev_priv = dev->dev_private; + drm_mga_getparam_t *param = data; + int value; + + if (!dev_priv) { + DRM_ERROR("called with no initialization\n"); + return -EINVAL; + } + + DRM_DEBUG("pid=%d\n", DRM_CURRENTPID); + + switch (param->param) { + case MGA_PARAM_IRQ_NR: + value = drm_dev_to_irq(dev); + break; + case MGA_PARAM_CARD_TYPE: + value = dev_priv->chipset; + break; + default: + return -EINVAL; + } + + if (DRM_COPY_TO_USER(param->value, &value, sizeof(int))) { + DRM_ERROR("copy_to_user\n"); + return -EFAULT; + } + + return 0; +} + +static int mga_set_fence(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_mga_private_t *dev_priv = dev->dev_private; + u32 *fence = data; + DMA_LOCALS; + + if (!dev_priv) { + DRM_ERROR("called with no initialization\n"); + return -EINVAL; + } + + DRM_DEBUG("pid=%d\n", DRM_CURRENTPID); + + /* I would normal do this assignment in the declaration of fence, + * but dev_priv may be NULL. + */ + + *fence = dev_priv->next_fence_to_post; + dev_priv->next_fence_to_post++; + + BEGIN_DMA(1); + DMA_BLOCK(MGA_DMAPAD, 0x00000000, + MGA_DMAPAD, 0x00000000, + MGA_DMAPAD, 0x00000000, MGA_SOFTRAP, 0x00000000); + ADVANCE_DMA(); + + return 0; +} + +static int mga_wait_fence(struct drm_device *dev, void *data, struct drm_file * +file_priv) +{ + drm_mga_private_t *dev_priv = dev->dev_private; + u32 *fence = data; + + if (!dev_priv) { + DRM_ERROR("called with no initialization\n"); + return -EINVAL; + } + + DRM_DEBUG("pid=%d\n", DRM_CURRENTPID); + + mga_driver_fence_wait(dev, fence); + return 0; +} + +struct drm_ioctl_desc mga_ioctls[] = { + DRM_IOCTL_DEF_DRV(MGA_INIT, mga_dma_init, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF_DRV(MGA_FLUSH, mga_dma_flush, DRM_AUTH), + DRM_IOCTL_DEF_DRV(MGA_RESET, mga_dma_reset, DRM_AUTH), + DRM_IOCTL_DEF_DRV(MGA_SWAP, mga_dma_swap, DRM_AUTH), + DRM_IOCTL_DEF_DRV(MGA_CLEAR, mga_dma_clear, DRM_AUTH), + DRM_IOCTL_DEF_DRV(MGA_VERTEX, mga_dma_vertex, DRM_AUTH), + DRM_IOCTL_DEF_DRV(MGA_INDICES, mga_dma_indices, DRM_AUTH), + DRM_IOCTL_DEF_DRV(MGA_ILOAD, mga_dma_iload, DRM_AUTH), + DRM_IOCTL_DEF_DRV(MGA_BLIT, mga_dma_blit, DRM_AUTH), + DRM_IOCTL_DEF_DRV(MGA_GETPARAM, mga_getparam, DRM_AUTH), + DRM_IOCTL_DEF_DRV(MGA_SET_FENCE, mga_set_fence, DRM_AUTH), + DRM_IOCTL_DEF_DRV(MGA_WAIT_FENCE, mga_wait_fence, DRM_AUTH), + DRM_IOCTL_DEF_DRV(MGA_DMA_BOOTSTRAP, mga_dma_bootstrap, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), +}; + +int mga_max_ioctl = DRM_ARRAY_SIZE(mga_ioctls); diff --git a/drivers/gpu/drm/mga/mga_warp.c b/drivers/gpu/drm/mga/mga_warp.c new file mode 100644 index 00000000..f172bd5c --- /dev/null +++ b/drivers/gpu/drm/mga/mga_warp.c @@ -0,0 +1,169 @@ +/* mga_warp.c -- Matrox G200/G400 WARP engine management -*- linux-c -*- + * Created: Thu Jan 11 21:29:32 2001 by gareth@valinux.com + * + * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * VA LINUX SYSTEMS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: + * Gareth Hughes + */ + +#include +#include +#include + +#include "drmP.h" +#include "drm.h" +#include "mga_drm.h" +#include "mga_drv.h" + +#define FIRMWARE_G200 "matrox/g200_warp.fw" +#define FIRMWARE_G400 "matrox/g400_warp.fw" + +MODULE_FIRMWARE(FIRMWARE_G200); +MODULE_FIRMWARE(FIRMWARE_G400); + +#define MGA_WARP_CODE_ALIGN 256 /* in bytes */ + +#define WARP_UCODE_SIZE(size) ALIGN(size, MGA_WARP_CODE_ALIGN) + +int mga_warp_install_microcode(drm_mga_private_t *dev_priv) +{ + unsigned char *vcbase = dev_priv->warp->handle; + unsigned long pcbase = dev_priv->warp->offset; + const char *firmware_name; + struct platform_device *pdev; + const struct firmware *fw = NULL; + const struct ihex_binrec *rec; + unsigned int size; + int n_pipes, where; + int rc = 0; + + switch (dev_priv->chipset) { + case MGA_CARD_TYPE_G400: + case MGA_CARD_TYPE_G550: + firmware_name = FIRMWARE_G400; + n_pipes = MGA_MAX_G400_PIPES; + break; + case MGA_CARD_TYPE_G200: + firmware_name = FIRMWARE_G200; + n_pipes = MGA_MAX_G200_PIPES; + break; + default: + return -EINVAL; + } + + pdev = platform_device_register_simple("mga_warp", 0, NULL, 0); + if (IS_ERR(pdev)) { + DRM_ERROR("mga: Failed to register microcode\n"); + return PTR_ERR(pdev); + } + rc = request_ihex_firmware(&fw, firmware_name, &pdev->dev); + platform_device_unregister(pdev); + if (rc) { + DRM_ERROR("mga: Failed to load microcode \"%s\"\n", + firmware_name); + return rc; + } + + size = 0; + where = 0; + for (rec = (const struct ihex_binrec *)fw->data; + rec; + rec = ihex_next_binrec(rec)) { + size += WARP_UCODE_SIZE(be16_to_cpu(rec->len)); + where++; + } + + if (where != n_pipes) { + DRM_ERROR("mga: Invalid microcode \"%s\"\n", firmware_name); + rc = -EINVAL; + goto out; + } + size = PAGE_ALIGN(size); + DRM_DEBUG("MGA ucode size = %d bytes\n", size); + if (size > dev_priv->warp->size) { + DRM_ERROR("microcode too large! (%u > %lu)\n", + size, dev_priv->warp->size); + rc = -ENOMEM; + goto out; + } + + memset(dev_priv->warp_pipe_phys, 0, sizeof(dev_priv->warp_pipe_phys)); + + where = 0; + for (rec = (const struct ihex_binrec *)fw->data; + rec; + rec = ihex_next_binrec(rec)) { + unsigned int src_size, dst_size; + + DRM_DEBUG(" pcbase = 0x%08lx vcbase = %p\n", pcbase, vcbase); + dev_priv->warp_pipe_phys[where] = pcbase; + src_size = be16_to_cpu(rec->len); + dst_size = WARP_UCODE_SIZE(src_size); + memcpy(vcbase, rec->data, src_size); + pcbase += dst_size; + vcbase += dst_size; + where++; + } + +out: + release_firmware(fw); + return rc; +} + +#define WMISC_EXPECTED (MGA_WUCODECACHE_ENABLE | MGA_WMASTER_ENABLE) + +int mga_warp_init(drm_mga_private_t *dev_priv) +{ + u32 wmisc; + + /* FIXME: Get rid of these damned magic numbers... + */ + switch (dev_priv->chipset) { + case MGA_CARD_TYPE_G400: + case MGA_CARD_TYPE_G550: + MGA_WRITE(MGA_WIADDR2, MGA_WMODE_SUSPEND); + MGA_WRITE(MGA_WGETMSB, 0x00000E00); + MGA_WRITE(MGA_WVRTXSZ, 0x00001807); + MGA_WRITE(MGA_WACCEPTSEQ, 0x18000000); + break; + case MGA_CARD_TYPE_G200: + MGA_WRITE(MGA_WIADDR, MGA_WMODE_SUSPEND); + MGA_WRITE(MGA_WGETMSB, 0x1606); + MGA_WRITE(MGA_WVRTXSZ, 7); + break; + default: + return -EINVAL; + } + + MGA_WRITE(MGA_WMISC, (MGA_WUCODECACHE_ENABLE | + MGA_WMASTER_ENABLE | MGA_WCACHEFLUSH_ENABLE)); + wmisc = MGA_READ(MGA_WMISC); + if (wmisc != WMISC_EXPECTED) { + DRM_ERROR("WARP engine config failed! 0x%x != 0x%x\n", + wmisc, WMISC_EXPECTED); + return -EINVAL; + } + + return 0; +} diff --git a/drivers/gpu/drm/nouveau/Kconfig b/drivers/gpu/drm/nouveau/Kconfig new file mode 100644 index 00000000..ca163991 --- /dev/null +++ b/drivers/gpu/drm/nouveau/Kconfig @@ -0,0 +1,56 @@ +config DRM_NOUVEAU + tristate "Nouveau (nVidia) cards" + depends on DRM && PCI + select FW_LOADER + select DRM_KMS_HELPER + select DRM_TTM + select FB_CFB_FILLRECT + select FB_CFB_COPYAREA + select FB_CFB_IMAGEBLIT + select FB + select FRAMEBUFFER_CONSOLE if !EXPERT + select FB_BACKLIGHT if DRM_NOUVEAU_BACKLIGHT + select ACPI_VIDEO if ACPI && X86 && BACKLIGHT_CLASS_DEVICE && VIDEO_OUTPUT_CONTROL && INPUT + select ACPI_WMI if ACPI + select MXM_WMI if ACPI + help + Choose this option for open-source nVidia support. + +config DRM_NOUVEAU_BACKLIGHT + bool "Support for backlight control" + depends on DRM_NOUVEAU + default y + help + Say Y here if you want to control the backlight of your display + (e.g. a laptop panel). + +config DRM_NOUVEAU_DEBUG + bool "Build in Nouveau's debugfs support" + depends on DRM_NOUVEAU && DEBUG_FS + default y + help + Say Y here if you want Nouveau to output debugging information + via debugfs. + +menu "I2C encoder or helper chips" + depends on DRM && DRM_KMS_HELPER && I2C + +config DRM_I2C_CH7006 + tristate "Chrontel ch7006 TV encoder" + default m if DRM_NOUVEAU + help + Support for Chrontel ch7006 and similar TV encoders, found + on some nVidia video cards. + + This driver is currently only useful if you're also using + the nouveau driver. + +config DRM_I2C_SIL164 + tristate "Silicon Image sil164 TMDS transmitter" + default m if DRM_NOUVEAU + help + Support for sil164 and similar single-link (or dual-link + when used in pairs) TMDS transmitters, used in some nVidia + video cards. + +endmenu diff --git a/drivers/gpu/drm/nouveau/Makefile b/drivers/gpu/drm/nouveau/Makefile new file mode 100644 index 00000000..0583677e --- /dev/null +++ b/drivers/gpu/drm/nouveau/Makefile @@ -0,0 +1,42 @@ +# +# Makefile for the drm device driver. This driver provides support for the +# Direct Rendering Infrastructure (DRI) in XFree86 4.1.0 and higher. + +ccflags-y := -Iinclude/drm +nouveau-y := nouveau_drv.o nouveau_state.o nouveau_channel.o nouveau_mem.o \ + nouveau_object.o nouveau_irq.o nouveau_notifier.o \ + nouveau_sgdma.o nouveau_dma.o nouveau_util.o \ + nouveau_bo.o nouveau_fence.o nouveau_gem.o nouveau_ttm.o \ + nouveau_hw.o nouveau_calc.o nouveau_bios.o nouveau_i2c.o \ + nouveau_display.o nouveau_connector.o nouveau_fbcon.o \ + nouveau_dp.o nouveau_ramht.o \ + nouveau_pm.o nouveau_volt.o nouveau_perf.o nouveau_temp.o \ + nouveau_mm.o nouveau_vm.o \ + nv04_timer.o \ + nv04_mc.o nv40_mc.o nv50_mc.o \ + nv04_fb.o nv10_fb.o nv30_fb.o nv40_fb.o nv50_fb.o nvc0_fb.o \ + nv04_fifo.o nv10_fifo.o nv40_fifo.o nv50_fifo.o nvc0_fifo.o \ + nv04_graph.o nv10_graph.o nv20_graph.o \ + nv40_graph.o nv50_graph.o nvc0_graph.o \ + nv40_grctx.o nv50_grctx.o nvc0_grctx.o \ + nv84_crypt.o \ + nva3_copy.o nvc0_copy.o \ + nv40_mpeg.o nv50_mpeg.o \ + nv04_instmem.o nv50_instmem.o nvc0_instmem.o \ + nv50_evo.o nv50_crtc.o nv50_dac.o nv50_sor.o \ + nv50_cursor.o nv50_display.o \ + nv04_dac.o nv04_dfp.o nv04_tv.o nv17_tv.o nv17_tv_modes.o \ + nv04_crtc.o nv04_display.o nv04_cursor.o \ + nv04_fbcon.o nv50_fbcon.o nvc0_fbcon.o \ + nv10_gpio.o nv50_gpio.o \ + nv50_calc.o \ + nv04_pm.o nv50_pm.o nva3_pm.o \ + nv50_vram.o nvc0_vram.o \ + nv50_vm.o nvc0_vm.o + +nouveau-$(CONFIG_DRM_NOUVEAU_DEBUG) += nouveau_debugfs.o +nouveau-$(CONFIG_COMPAT) += nouveau_ioc32.o +nouveau-$(CONFIG_DRM_NOUVEAU_BACKLIGHT) += nouveau_backlight.o +nouveau-$(CONFIG_ACPI) += nouveau_acpi.o + +obj-$(CONFIG_DRM_NOUVEAU)+= nouveau.o diff --git a/drivers/gpu/drm/nouveau/nouveau_acpi.c b/drivers/gpu/drm/nouveau/nouveau_acpi.c new file mode 100644 index 00000000..525744d5 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_acpi.c @@ -0,0 +1,387 @@ +#include +#include +#include +#include +#include +#include +#include +#include + +#include "drmP.h" +#include "drm.h" +#include "drm_sarea.h" +#include "drm_crtc_helper.h" +#include "nouveau_drv.h" +#include "nouveau_drm.h" +#include "nv50_display.h" +#include "nouveau_connector.h" + +#include + +#define NOUVEAU_DSM_SUPPORTED 0x00 +#define NOUVEAU_DSM_SUPPORTED_FUNCTIONS 0x00 + +#define NOUVEAU_DSM_ACTIVE 0x01 +#define NOUVEAU_DSM_ACTIVE_QUERY 0x00 + +#define NOUVEAU_DSM_LED 0x02 +#define NOUVEAU_DSM_LED_STATE 0x00 +#define NOUVEAU_DSM_LED_OFF 0x10 +#define NOUVEAU_DSM_LED_STAMINA 0x11 +#define NOUVEAU_DSM_LED_SPEED 0x12 + +#define NOUVEAU_DSM_POWER 0x03 +#define NOUVEAU_DSM_POWER_STATE 0x00 +#define NOUVEAU_DSM_POWER_SPEED 0x01 +#define NOUVEAU_DSM_POWER_STAMINA 0x02 + +static struct nouveau_dsm_priv { + bool dsm_detected; + bool optimus_detected; + acpi_handle dhandle; + acpi_handle rom_handle; +} nouveau_dsm_priv; + +#define NOUVEAU_DSM_HAS_MUX 0x1 +#define NOUVEAU_DSM_HAS_OPT 0x2 + +static const char nouveau_dsm_muid[] = { + 0xA0, 0xA0, 0x95, 0x9D, 0x60, 0x00, 0x48, 0x4D, + 0xB3, 0x4D, 0x7E, 0x5F, 0xEA, 0x12, 0x9F, 0xD4, +}; + +static const char nouveau_op_dsm_muid[] = { + 0xF8, 0xD8, 0x86, 0xA4, 0xDA, 0x0B, 0x1B, 0x47, + 0xA7, 0x2B, 0x60, 0x42, 0xA6, 0xB5, 0xBE, 0xE0, +}; + +static int nouveau_optimus_dsm(acpi_handle handle, int func, int arg, uint32_t *result) +{ + struct acpi_buffer output = { ACPI_ALLOCATE_BUFFER, NULL }; + struct acpi_object_list input; + union acpi_object params[4]; + union acpi_object *obj; + int err; + + input.count = 4; + input.pointer = params; + params[0].type = ACPI_TYPE_BUFFER; + params[0].buffer.length = sizeof(nouveau_op_dsm_muid); + params[0].buffer.pointer = (char *)nouveau_op_dsm_muid; + params[1].type = ACPI_TYPE_INTEGER; + params[1].integer.value = 0x00000100; + params[2].type = ACPI_TYPE_INTEGER; + params[2].integer.value = func; + params[3].type = ACPI_TYPE_BUFFER; + params[3].buffer.length = 0; + + err = acpi_evaluate_object(handle, "_DSM", &input, &output); + if (err) { + printk(KERN_INFO "failed to evaluate _DSM: %d\n", err); + return err; + } + + obj = (union acpi_object *)output.pointer; + + if (obj->type == ACPI_TYPE_INTEGER) + if (obj->integer.value == 0x80000002) { + return -ENODEV; + } + + if (obj->type == ACPI_TYPE_BUFFER) { + if (obj->buffer.length == 4 && result) { + *result = 0; + *result |= obj->buffer.pointer[0]; + *result |= (obj->buffer.pointer[1] << 8); + *result |= (obj->buffer.pointer[2] << 16); + *result |= (obj->buffer.pointer[3] << 24); + } + } + + kfree(output.pointer); + return 0; +} + +static int nouveau_dsm(acpi_handle handle, int func, int arg, uint32_t *result) +{ + struct acpi_buffer output = { ACPI_ALLOCATE_BUFFER, NULL }; + struct acpi_object_list input; + union acpi_object params[4]; + union acpi_object *obj; + int err; + + input.count = 4; + input.pointer = params; + params[0].type = ACPI_TYPE_BUFFER; + params[0].buffer.length = sizeof(nouveau_dsm_muid); + params[0].buffer.pointer = (char *)nouveau_dsm_muid; + params[1].type = ACPI_TYPE_INTEGER; + params[1].integer.value = 0x00000102; + params[2].type = ACPI_TYPE_INTEGER; + params[2].integer.value = func; + params[3].type = ACPI_TYPE_INTEGER; + params[3].integer.value = arg; + + err = acpi_evaluate_object(handle, "_DSM", &input, &output); + if (err) { + printk(KERN_INFO "failed to evaluate _DSM: %d\n", err); + return err; + } + + obj = (union acpi_object *)output.pointer; + + if (obj->type == ACPI_TYPE_INTEGER) + if (obj->integer.value == 0x80000002) + return -ENODEV; + + if (obj->type == ACPI_TYPE_BUFFER) { + if (obj->buffer.length == 4 && result) { + *result = 0; + *result |= obj->buffer.pointer[0]; + *result |= (obj->buffer.pointer[1] << 8); + *result |= (obj->buffer.pointer[2] << 16); + *result |= (obj->buffer.pointer[3] << 24); + } + } + + kfree(output.pointer); + return 0; +} + +static int nouveau_dsm_switch_mux(acpi_handle handle, int mux_id) +{ + mxm_wmi_call_mxmx(mux_id == NOUVEAU_DSM_LED_STAMINA ? MXM_MXDS_ADAPTER_IGD : MXM_MXDS_ADAPTER_0); + mxm_wmi_call_mxds(mux_id == NOUVEAU_DSM_LED_STAMINA ? MXM_MXDS_ADAPTER_IGD : MXM_MXDS_ADAPTER_0); + return nouveau_dsm(handle, NOUVEAU_DSM_LED, mux_id, NULL); +} + +static int nouveau_dsm_set_discrete_state(acpi_handle handle, enum vga_switcheroo_state state) +{ + int arg; + if (state == VGA_SWITCHEROO_ON) + arg = NOUVEAU_DSM_POWER_SPEED; + else + arg = NOUVEAU_DSM_POWER_STAMINA; + nouveau_dsm(handle, NOUVEAU_DSM_POWER, arg, NULL); + return 0; +} + +static int nouveau_dsm_switchto(enum vga_switcheroo_client_id id) +{ + if (id == VGA_SWITCHEROO_IGD) + return nouveau_dsm_switch_mux(nouveau_dsm_priv.dhandle, NOUVEAU_DSM_LED_STAMINA); + else + return nouveau_dsm_switch_mux(nouveau_dsm_priv.dhandle, NOUVEAU_DSM_LED_SPEED); +} + +static int nouveau_dsm_power_state(enum vga_switcheroo_client_id id, + enum vga_switcheroo_state state) +{ + if (id == VGA_SWITCHEROO_IGD) + return 0; + + return nouveau_dsm_set_discrete_state(nouveau_dsm_priv.dhandle, state); +} + +static int nouveau_dsm_init(void) +{ + return 0; +} + +static int nouveau_dsm_get_client_id(struct pci_dev *pdev) +{ + /* easy option one - intel vendor ID means Integrated */ + if (pdev->vendor == PCI_VENDOR_ID_INTEL) + return VGA_SWITCHEROO_IGD; + + /* is this device on Bus 0? - this may need improving */ + if (pdev->bus->number == 0) + return VGA_SWITCHEROO_IGD; + + return VGA_SWITCHEROO_DIS; +} + +static struct vga_switcheroo_handler nouveau_dsm_handler = { + .switchto = nouveau_dsm_switchto, + .power_state = nouveau_dsm_power_state, + .init = nouveau_dsm_init, + .get_client_id = nouveau_dsm_get_client_id, +}; + +static int nouveau_dsm_pci_probe(struct pci_dev *pdev) +{ + acpi_handle dhandle, nvidia_handle; + acpi_status status; + int ret, retval = 0; + uint32_t result; + + dhandle = DEVICE_ACPI_HANDLE(&pdev->dev); + if (!dhandle) + return false; + + status = acpi_get_handle(dhandle, "_DSM", &nvidia_handle); + if (ACPI_FAILURE(status)) { + return false; + } + + ret = nouveau_dsm(dhandle, NOUVEAU_DSM_SUPPORTED, + NOUVEAU_DSM_SUPPORTED_FUNCTIONS, &result); + if (ret == 0) + retval |= NOUVEAU_DSM_HAS_MUX; + + ret = nouveau_optimus_dsm(dhandle, 0, 0, &result); + if (ret == 0) + retval |= NOUVEAU_DSM_HAS_OPT; + + if (retval) + nouveau_dsm_priv.dhandle = dhandle; + + return retval; +} + +static bool nouveau_dsm_detect(void) +{ + char acpi_method_name[255] = { 0 }; + struct acpi_buffer buffer = {sizeof(acpi_method_name), acpi_method_name}; + struct pci_dev *pdev = NULL; + int has_dsm = 0; + int has_optimus; + int vga_count = 0; + bool guid_valid; + int retval; + bool ret = false; + + /* lookup the MXM GUID */ + guid_valid = mxm_wmi_supported(); + + if (guid_valid) + printk("MXM: GUID detected in BIOS\n"); + + /* now do DSM detection */ + while ((pdev = pci_get_class(PCI_CLASS_DISPLAY_VGA << 8, pdev)) != NULL) { + vga_count++; + + retval = nouveau_dsm_pci_probe(pdev); + if (retval & NOUVEAU_DSM_HAS_MUX) + has_dsm |= 1; + if (retval & NOUVEAU_DSM_HAS_OPT) + has_optimus = 1; + } + + if (vga_count == 2 && has_dsm && guid_valid) { + acpi_get_name(nouveau_dsm_priv.dhandle, ACPI_FULL_PATHNAME, &buffer); + printk(KERN_INFO "VGA switcheroo: detected DSM switching method %s handle\n", + acpi_method_name); + nouveau_dsm_priv.dsm_detected = true; + ret = true; + } + + if (has_optimus == 1) + nouveau_dsm_priv.optimus_detected = true; + + return ret; +} + +void nouveau_register_dsm_handler(void) +{ + bool r; + + r = nouveau_dsm_detect(); + if (!r) + return; + + vga_switcheroo_register_handler(&nouveau_dsm_handler); +} + +void nouveau_unregister_dsm_handler(void) +{ + vga_switcheroo_unregister_handler(); +} + +/* retrieve the ROM in 4k blocks */ +static int nouveau_rom_call(acpi_handle rom_handle, uint8_t *bios, + int offset, int len) +{ + acpi_status status; + union acpi_object rom_arg_elements[2], *obj; + struct acpi_object_list rom_arg; + struct acpi_buffer buffer = { ACPI_ALLOCATE_BUFFER, NULL}; + + rom_arg.count = 2; + rom_arg.pointer = &rom_arg_elements[0]; + + rom_arg_elements[0].type = ACPI_TYPE_INTEGER; + rom_arg_elements[0].integer.value = offset; + + rom_arg_elements[1].type = ACPI_TYPE_INTEGER; + rom_arg_elements[1].integer.value = len; + + status = acpi_evaluate_object(rom_handle, NULL, &rom_arg, &buffer); + if (ACPI_FAILURE(status)) { + printk(KERN_INFO "failed to evaluate ROM got %s\n", acpi_format_exception(status)); + return -ENODEV; + } + obj = (union acpi_object *)buffer.pointer; + memcpy(bios+offset, obj->buffer.pointer, len); + kfree(buffer.pointer); + return len; +} + +bool nouveau_acpi_rom_supported(struct pci_dev *pdev) +{ + acpi_status status; + acpi_handle dhandle, rom_handle; + + if (!nouveau_dsm_priv.dsm_detected && !nouveau_dsm_priv.optimus_detected) + return false; + + dhandle = DEVICE_ACPI_HANDLE(&pdev->dev); + if (!dhandle) + return false; + + status = acpi_get_handle(dhandle, "_ROM", &rom_handle); + if (ACPI_FAILURE(status)) + return false; + + nouveau_dsm_priv.rom_handle = rom_handle; + return true; +} + +int nouveau_acpi_get_bios_chunk(uint8_t *bios, int offset, int len) +{ + return nouveau_rom_call(nouveau_dsm_priv.rom_handle, bios, offset, len); +} + +int +nouveau_acpi_edid(struct drm_device *dev, struct drm_connector *connector) +{ + struct nouveau_connector *nv_connector = nouveau_connector(connector); + struct acpi_device *acpidev; + acpi_handle handle; + int type, ret; + void *edid; + + switch (connector->connector_type) { + case DRM_MODE_CONNECTOR_LVDS: + case DRM_MODE_CONNECTOR_eDP: + type = ACPI_VIDEO_DISPLAY_LCD; + break; + default: + return -EINVAL; + } + + handle = DEVICE_ACPI_HANDLE(&dev->pdev->dev); + if (!handle) + return -ENODEV; + + ret = acpi_bus_get_device(handle, &acpidev); + if (ret) + return -ENODEV; + + ret = acpi_video_get_edid(acpidev, type, -1, &edid); + if (ret < 0) + return ret; + + nv_connector->edid = kmemdup(edid, EDID_LENGTH, GFP_KERNEL); + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nouveau_backlight.c b/drivers/gpu/drm/nouveau/nouveau_backlight.c new file mode 100644 index 00000000..00a55dfd --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_backlight.c @@ -0,0 +1,174 @@ +/* + * Copyright (C) 2009 Red Hat + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +/* + * Authors: + * Matthew Garrett + * + * Register locations derived from NVClock by Roderick Colenbrander + */ + +#include +#include + +#include "drmP.h" +#include "nouveau_drv.h" +#include "nouveau_drm.h" +#include "nouveau_reg.h" + +static int nv40_get_intensity(struct backlight_device *bd) +{ + struct drm_device *dev = bl_get_data(bd); + int val = (nv_rd32(dev, NV40_PMC_BACKLIGHT) & NV40_PMC_BACKLIGHT_MASK) + >> 16; + + return val; +} + +static int nv40_set_intensity(struct backlight_device *bd) +{ + struct drm_device *dev = bl_get_data(bd); + int val = bd->props.brightness; + int reg = nv_rd32(dev, NV40_PMC_BACKLIGHT); + + nv_wr32(dev, NV40_PMC_BACKLIGHT, + (val << 16) | (reg & ~NV40_PMC_BACKLIGHT_MASK)); + + return 0; +} + +static const struct backlight_ops nv40_bl_ops = { + .options = BL_CORE_SUSPENDRESUME, + .get_brightness = nv40_get_intensity, + .update_status = nv40_set_intensity, +}; + +static int nv50_get_intensity(struct backlight_device *bd) +{ + struct drm_device *dev = bl_get_data(bd); + + return nv_rd32(dev, NV50_PDISPLAY_SOR_BACKLIGHT); +} + +static int nv50_set_intensity(struct backlight_device *bd) +{ + struct drm_device *dev = bl_get_data(bd); + int val = bd->props.brightness; + + nv_wr32(dev, NV50_PDISPLAY_SOR_BACKLIGHT, + val | NV50_PDISPLAY_SOR_BACKLIGHT_ENABLE); + return 0; +} + +static const struct backlight_ops nv50_bl_ops = { + .options = BL_CORE_SUSPENDRESUME, + .get_brightness = nv50_get_intensity, + .update_status = nv50_set_intensity, +}; + +static int nouveau_nv40_backlight_init(struct drm_connector *connector) +{ + struct drm_device *dev = connector->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct backlight_properties props; + struct backlight_device *bd; + + if (!(nv_rd32(dev, NV40_PMC_BACKLIGHT) & NV40_PMC_BACKLIGHT_MASK)) + return 0; + + memset(&props, 0, sizeof(struct backlight_properties)); + props.type = BACKLIGHT_RAW; + props.max_brightness = 31; + bd = backlight_device_register("nv_backlight", &connector->kdev, dev, + &nv40_bl_ops, &props); + if (IS_ERR(bd)) + return PTR_ERR(bd); + + dev_priv->backlight = bd; + bd->props.brightness = nv40_get_intensity(bd); + backlight_update_status(bd); + + return 0; +} + +static int nouveau_nv50_backlight_init(struct drm_connector *connector) +{ + struct drm_device *dev = connector->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct backlight_properties props; + struct backlight_device *bd; + + if (!nv_rd32(dev, NV50_PDISPLAY_SOR_BACKLIGHT)) + return 0; + + memset(&props, 0, sizeof(struct backlight_properties)); + props.type = BACKLIGHT_RAW; + props.max_brightness = 1025; + bd = backlight_device_register("nv_backlight", &connector->kdev, dev, + &nv50_bl_ops, &props); + if (IS_ERR(bd)) + return PTR_ERR(bd); + + dev_priv->backlight = bd; + bd->props.brightness = nv50_get_intensity(bd); + backlight_update_status(bd); + return 0; +} + +int nouveau_backlight_init(struct drm_connector *connector) +{ + struct drm_device *dev = connector->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + +#ifdef CONFIG_ACPI + if (acpi_video_backlight_support()) { + NV_INFO(dev, "ACPI backlight interface available, " + "not registering our own\n"); + return 0; + } +#endif + + switch (dev_priv->card_type) { + case NV_40: + return nouveau_nv40_backlight_init(connector); + case NV_50: + return nouveau_nv50_backlight_init(connector); + default: + break; + } + + return 0; +} + +void nouveau_backlight_exit(struct drm_connector *connector) +{ + struct drm_device *dev = connector->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + + if (dev_priv->backlight) { + backlight_device_unregister(dev_priv->backlight); + dev_priv->backlight = NULL; + } +} diff --git a/drivers/gpu/drm/nouveau/nouveau_bios.c b/drivers/gpu/drm/nouveau/nouveau_bios.c new file mode 100644 index 00000000..729d5fd7 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_bios.c @@ -0,0 +1,6930 @@ +/* + * Copyright 2005-2006 Erik Waling + * Copyright 2006 Stephane Marchesin + * Copyright 2007-2009 Stuart Bennett + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#include "drmP.h" +#define NV_DEBUG_NOTRACE +#include "nouveau_drv.h" +#include "nouveau_hw.h" +#include "nouveau_encoder.h" + +#include + +/* these defines are made up */ +#define NV_CIO_CRE_44_HEADA 0x0 +#define NV_CIO_CRE_44_HEADB 0x3 +#define FEATURE_MOBILE 0x10 /* also FEATURE_QUADRO for BMP */ +#define LEGACY_I2C_CRT 0x80 +#define LEGACY_I2C_PANEL 0x81 +#define LEGACY_I2C_TV 0x82 + +#define EDID1_LEN 128 + +#define BIOSLOG(sip, fmt, arg...) NV_DEBUG(sip->dev, fmt, ##arg) +#define LOG_OLD_VALUE(x) + +struct init_exec { + bool execute; + bool repeat; +}; + +static bool nv_cksum(const uint8_t *data, unsigned int length) +{ + /* + * There's a few checksums in the BIOS, so here's a generic checking + * function. + */ + int i; + uint8_t sum = 0; + + for (i = 0; i < length; i++) + sum += data[i]; + + if (sum) + return true; + + return false; +} + +static int +score_vbios(struct drm_device *dev, const uint8_t *data, const bool writeable) +{ + if (!(data[0] == 0x55 && data[1] == 0xAA)) { + NV_TRACEWARN(dev, "... BIOS signature not found\n"); + return 0; + } + + if (nv_cksum(data, data[2] * 512)) { + NV_TRACEWARN(dev, "... BIOS checksum invalid\n"); + /* if a ro image is somewhat bad, it's probably all rubbish */ + return writeable ? 2 : 1; + } else + NV_TRACE(dev, "... appears to be valid\n"); + + return 3; +} + +static void load_vbios_prom(struct drm_device *dev, uint8_t *data) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + uint32_t pci_nv_20, save_pci_nv_20; + int pcir_ptr; + int i; + + if (dev_priv->card_type >= NV_50) + pci_nv_20 = 0x88050; + else + pci_nv_20 = NV_PBUS_PCI_NV_20; + + /* enable ROM access */ + save_pci_nv_20 = nvReadMC(dev, pci_nv_20); + nvWriteMC(dev, pci_nv_20, + save_pci_nv_20 & ~NV_PBUS_PCI_NV_20_ROM_SHADOW_ENABLED); + + /* bail if no rom signature */ + if (nv_rd08(dev, NV_PROM_OFFSET) != 0x55 || + nv_rd08(dev, NV_PROM_OFFSET + 1) != 0xaa) + goto out; + + /* additional check (see note below) - read PCI record header */ + pcir_ptr = nv_rd08(dev, NV_PROM_OFFSET + 0x18) | + nv_rd08(dev, NV_PROM_OFFSET + 0x19) << 8; + if (nv_rd08(dev, NV_PROM_OFFSET + pcir_ptr) != 'P' || + nv_rd08(dev, NV_PROM_OFFSET + pcir_ptr + 1) != 'C' || + nv_rd08(dev, NV_PROM_OFFSET + pcir_ptr + 2) != 'I' || + nv_rd08(dev, NV_PROM_OFFSET + pcir_ptr + 3) != 'R') + goto out; + + /* on some 6600GT/6800LE prom reads are messed up. nvclock alleges a + * a good read may be obtained by waiting or re-reading (cargocult: 5x) + * each byte. we'll hope pramin has something usable instead + */ + for (i = 0; i < NV_PROM_SIZE; i++) + data[i] = nv_rd08(dev, NV_PROM_OFFSET + i); + +out: + /* disable ROM access */ + nvWriteMC(dev, pci_nv_20, + save_pci_nv_20 | NV_PBUS_PCI_NV_20_ROM_SHADOW_ENABLED); +} + +static void load_vbios_pramin(struct drm_device *dev, uint8_t *data) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + uint32_t old_bar0_pramin = 0; + int i; + + if (dev_priv->card_type >= NV_50) { + uint32_t vbios_vram = (nv_rd32(dev, 0x619f04) & ~0xff) << 8; + + if (!vbios_vram) + vbios_vram = (nv_rd32(dev, 0x1700) << 16) + 0xf0000; + + old_bar0_pramin = nv_rd32(dev, 0x1700); + nv_wr32(dev, 0x1700, vbios_vram >> 16); + } + + /* bail if no rom signature */ + if (nv_rd08(dev, NV_PRAMIN_OFFSET) != 0x55 || + nv_rd08(dev, NV_PRAMIN_OFFSET + 1) != 0xaa) + goto out; + + for (i = 0; i < NV_PROM_SIZE; i++) + data[i] = nv_rd08(dev, NV_PRAMIN_OFFSET + i); + +out: + if (dev_priv->card_type >= NV_50) + nv_wr32(dev, 0x1700, old_bar0_pramin); +} + +static void load_vbios_pci(struct drm_device *dev, uint8_t *data) +{ + void __iomem *rom = NULL; + size_t rom_len; + int ret; + + ret = pci_enable_rom(dev->pdev); + if (ret) + return; + + rom = pci_map_rom(dev->pdev, &rom_len); + if (!rom) + goto out; + memcpy_fromio(data, rom, rom_len); + pci_unmap_rom(dev->pdev, rom); + +out: + pci_disable_rom(dev->pdev); +} + +static void load_vbios_acpi(struct drm_device *dev, uint8_t *data) +{ + int i; + int ret; + int size = 64 * 1024; + + if (!nouveau_acpi_rom_supported(dev->pdev)) + return; + + for (i = 0; i < (size / ROM_BIOS_PAGE); i++) { + ret = nouveau_acpi_get_bios_chunk(data, + (i * ROM_BIOS_PAGE), + ROM_BIOS_PAGE); + if (ret <= 0) + break; + } + return; +} + +struct methods { + const char desc[8]; + void (*loadbios)(struct drm_device *, uint8_t *); + const bool rw; +}; + +static struct methods shadow_methods[] = { + { "PRAMIN", load_vbios_pramin, true }, + { "PROM", load_vbios_prom, false }, + { "PCIROM", load_vbios_pci, true }, + { "ACPI", load_vbios_acpi, true }, +}; +#define NUM_SHADOW_METHODS ARRAY_SIZE(shadow_methods) + +static bool NVShadowVBIOS(struct drm_device *dev, uint8_t *data) +{ + struct methods *methods = shadow_methods; + int testscore = 3; + int scores[NUM_SHADOW_METHODS], i; + + if (nouveau_vbios) { + for (i = 0; i < NUM_SHADOW_METHODS; i++) + if (!strcasecmp(nouveau_vbios, methods[i].desc)) + break; + + if (i < NUM_SHADOW_METHODS) { + NV_INFO(dev, "Attempting to use BIOS image from %s\n", + methods[i].desc); + + methods[i].loadbios(dev, data); + if (score_vbios(dev, data, methods[i].rw)) + return true; + } + + NV_ERROR(dev, "VBIOS source \'%s\' invalid\n", nouveau_vbios); + } + + for (i = 0; i < NUM_SHADOW_METHODS; i++) { + NV_TRACE(dev, "Attempting to load BIOS image from %s\n", + methods[i].desc); + data[0] = data[1] = 0; /* avoid reuse of previous image */ + methods[i].loadbios(dev, data); + scores[i] = score_vbios(dev, data, methods[i].rw); + if (scores[i] == testscore) + return true; + } + + while (--testscore > 0) { + for (i = 0; i < NUM_SHADOW_METHODS; i++) { + if (scores[i] == testscore) { + NV_TRACE(dev, "Using BIOS image from %s\n", + methods[i].desc); + methods[i].loadbios(dev, data); + return true; + } + } + } + + NV_ERROR(dev, "No valid BIOS image found\n"); + return false; +} + +struct init_tbl_entry { + char *name; + uint8_t id; + /* Return: + * > 0: success, length of opcode + * 0: success, but abort further parsing of table (INIT_DONE etc) + * < 0: failure, table parsing will be aborted + */ + int (*handler)(struct nvbios *, uint16_t, struct init_exec *); +}; + +static int parse_init_table(struct nvbios *, uint16_t, struct init_exec *); + +#define MACRO_INDEX_SIZE 2 +#define MACRO_SIZE 8 +#define CONDITION_SIZE 12 +#define IO_FLAG_CONDITION_SIZE 9 +#define IO_CONDITION_SIZE 5 +#define MEM_INIT_SIZE 66 + +static void still_alive(void) +{ +#if 0 + sync(); + mdelay(2); +#endif +} + +static uint32_t +munge_reg(struct nvbios *bios, uint32_t reg) +{ + struct drm_nouveau_private *dev_priv = bios->dev->dev_private; + struct dcb_entry *dcbent = bios->display.output; + + if (dev_priv->card_type < NV_50) + return reg; + + if (reg & 0x40000000) { + BUG_ON(!dcbent); + + reg += (ffs(dcbent->or) - 1) * 0x800; + if ((reg & 0x20000000) && !(dcbent->sorconf.link & 1)) + reg += 0x00000080; + } + + reg &= ~0x60000000; + return reg; +} + +static int +valid_reg(struct nvbios *bios, uint32_t reg) +{ + struct drm_nouveau_private *dev_priv = bios->dev->dev_private; + struct drm_device *dev = bios->dev; + + /* C51 has misaligned regs on purpose. Marvellous */ + if (reg & 0x2 || + (reg & 0x1 && dev_priv->vbios.chip_version != 0x51)) + NV_ERROR(dev, "======= misaligned reg 0x%08X =======\n", reg); + + /* warn on C51 regs that haven't been verified accessible in tracing */ + if (reg & 0x1 && dev_priv->vbios.chip_version == 0x51 && + reg != 0x130d && reg != 0x1311 && reg != 0x60081d) + NV_WARN(dev, "=== C51 misaligned reg 0x%08X not verified ===\n", + reg); + + if (reg >= (8*1024*1024)) { + NV_ERROR(dev, "=== reg 0x%08x out of mapped bounds ===\n", reg); + return 0; + } + + return 1; +} + +static bool +valid_idx_port(struct nvbios *bios, uint16_t port) +{ + struct drm_nouveau_private *dev_priv = bios->dev->dev_private; + struct drm_device *dev = bios->dev; + + /* + * If adding more ports here, the read/write functions below will need + * updating so that the correct mmio range (PRMCIO, PRMDIO, PRMVIO) is + * used for the port in question + */ + if (dev_priv->card_type < NV_50) { + if (port == NV_CIO_CRX__COLOR) + return true; + if (port == NV_VIO_SRX) + return true; + } else { + if (port == NV_CIO_CRX__COLOR) + return true; + } + + NV_ERROR(dev, "========== unknown indexed io port 0x%04X ==========\n", + port); + + return false; +} + +static bool +valid_port(struct nvbios *bios, uint16_t port) +{ + struct drm_device *dev = bios->dev; + + /* + * If adding more ports here, the read/write functions below will need + * updating so that the correct mmio range (PRMCIO, PRMDIO, PRMVIO) is + * used for the port in question + */ + if (port == NV_VIO_VSE2) + return true; + + NV_ERROR(dev, "========== unknown io port 0x%04X ==========\n", port); + + return false; +} + +static uint32_t +bios_rd32(struct nvbios *bios, uint32_t reg) +{ + uint32_t data; + + reg = munge_reg(bios, reg); + if (!valid_reg(bios, reg)) + return 0; + + /* + * C51 sometimes uses regs with bit0 set in the address. For these + * cases there should exist a translation in a BIOS table to an IO + * port address which the BIOS uses for accessing the reg + * + * These only seem to appear for the power control regs to a flat panel, + * and the GPIO regs at 0x60081*. In C51 mmio traces the normal regs + * for 0x1308 and 0x1310 are used - hence the mask below. An S3 + * suspend-resume mmio trace from a C51 will be required to see if this + * is true for the power microcode in 0x14.., or whether the direct IO + * port access method is needed + */ + if (reg & 0x1) + reg &= ~0x1; + + data = nv_rd32(bios->dev, reg); + + BIOSLOG(bios, " Read: Reg: 0x%08X, Data: 0x%08X\n", reg, data); + + return data; +} + +static void +bios_wr32(struct nvbios *bios, uint32_t reg, uint32_t data) +{ + struct drm_nouveau_private *dev_priv = bios->dev->dev_private; + + reg = munge_reg(bios, reg); + if (!valid_reg(bios, reg)) + return; + + /* see note in bios_rd32 */ + if (reg & 0x1) + reg &= 0xfffffffe; + + LOG_OLD_VALUE(bios_rd32(bios, reg)); + BIOSLOG(bios, " Write: Reg: 0x%08X, Data: 0x%08X\n", reg, data); + + if (dev_priv->vbios.execute) { + still_alive(); + nv_wr32(bios->dev, reg, data); + } +} + +static uint8_t +bios_idxprt_rd(struct nvbios *bios, uint16_t port, uint8_t index) +{ + struct drm_nouveau_private *dev_priv = bios->dev->dev_private; + struct drm_device *dev = bios->dev; + uint8_t data; + + if (!valid_idx_port(bios, port)) + return 0; + + if (dev_priv->card_type < NV_50) { + if (port == NV_VIO_SRX) + data = NVReadVgaSeq(dev, bios->state.crtchead, index); + else /* assume NV_CIO_CRX__COLOR */ + data = NVReadVgaCrtc(dev, bios->state.crtchead, index); + } else { + uint32_t data32; + + data32 = bios_rd32(bios, NV50_PDISPLAY_VGACRTC(index & ~3)); + data = (data32 >> ((index & 3) << 3)) & 0xff; + } + + BIOSLOG(bios, " Indexed IO read: Port: 0x%04X, Index: 0x%02X, " + "Head: 0x%02X, Data: 0x%02X\n", + port, index, bios->state.crtchead, data); + return data; +} + +static void +bios_idxprt_wr(struct nvbios *bios, uint16_t port, uint8_t index, uint8_t data) +{ + struct drm_nouveau_private *dev_priv = bios->dev->dev_private; + struct drm_device *dev = bios->dev; + + if (!valid_idx_port(bios, port)) + return; + + /* + * The current head is maintained in the nvbios member state.crtchead. + * We trap changes to CR44 and update the head variable and hence the + * register set written. + * As CR44 only exists on CRTC0, we update crtchead to head0 in advance + * of the write, and to head1 after the write + */ + if (port == NV_CIO_CRX__COLOR && index == NV_CIO_CRE_44 && + data != NV_CIO_CRE_44_HEADB) + bios->state.crtchead = 0; + + LOG_OLD_VALUE(bios_idxprt_rd(bios, port, index)); + BIOSLOG(bios, " Indexed IO write: Port: 0x%04X, Index: 0x%02X, " + "Head: 0x%02X, Data: 0x%02X\n", + port, index, bios->state.crtchead, data); + + if (bios->execute && dev_priv->card_type < NV_50) { + still_alive(); + if (port == NV_VIO_SRX) + NVWriteVgaSeq(dev, bios->state.crtchead, index, data); + else /* assume NV_CIO_CRX__COLOR */ + NVWriteVgaCrtc(dev, bios->state.crtchead, index, data); + } else + if (bios->execute) { + uint32_t data32, shift = (index & 3) << 3; + + still_alive(); + + data32 = bios_rd32(bios, NV50_PDISPLAY_VGACRTC(index & ~3)); + data32 &= ~(0xff << shift); + data32 |= (data << shift); + bios_wr32(bios, NV50_PDISPLAY_VGACRTC(index & ~3), data32); + } + + if (port == NV_CIO_CRX__COLOR && + index == NV_CIO_CRE_44 && data == NV_CIO_CRE_44_HEADB) + bios->state.crtchead = 1; +} + +static uint8_t +bios_port_rd(struct nvbios *bios, uint16_t port) +{ + uint8_t data, head = bios->state.crtchead; + + if (!valid_port(bios, port)) + return 0; + + data = NVReadPRMVIO(bios->dev, head, NV_PRMVIO0_OFFSET + port); + + BIOSLOG(bios, " IO read: Port: 0x%04X, Head: 0x%02X, Data: 0x%02X\n", + port, head, data); + + return data; +} + +static void +bios_port_wr(struct nvbios *bios, uint16_t port, uint8_t data) +{ + int head = bios->state.crtchead; + + if (!valid_port(bios, port)) + return; + + LOG_OLD_VALUE(bios_port_rd(bios, port)); + BIOSLOG(bios, " IO write: Port: 0x%04X, Head: 0x%02X, Data: 0x%02X\n", + port, head, data); + + if (!bios->execute) + return; + + still_alive(); + NVWritePRMVIO(bios->dev, head, NV_PRMVIO0_OFFSET + port, data); +} + +static bool +io_flag_condition_met(struct nvbios *bios, uint16_t offset, uint8_t cond) +{ + /* + * The IO flag condition entry has 2 bytes for the CRTC port; 1 byte + * for the CRTC index; 1 byte for the mask to apply to the value + * retrieved from the CRTC; 1 byte for the shift right to apply to the + * masked CRTC value; 2 bytes for the offset to the flag array, to + * which the shifted value is added; 1 byte for the mask applied to the + * value read from the flag array; and 1 byte for the value to compare + * against the masked byte from the flag table. + */ + + uint16_t condptr = bios->io_flag_condition_tbl_ptr + cond * IO_FLAG_CONDITION_SIZE; + uint16_t crtcport = ROM16(bios->data[condptr]); + uint8_t crtcindex = bios->data[condptr + 2]; + uint8_t mask = bios->data[condptr + 3]; + uint8_t shift = bios->data[condptr + 4]; + uint16_t flagarray = ROM16(bios->data[condptr + 5]); + uint8_t flagarraymask = bios->data[condptr + 7]; + uint8_t cmpval = bios->data[condptr + 8]; + uint8_t data; + + BIOSLOG(bios, "0x%04X: Port: 0x%04X, Index: 0x%02X, Mask: 0x%02X, " + "Shift: 0x%02X, FlagArray: 0x%04X, FAMask: 0x%02X, " + "Cmpval: 0x%02X\n", + offset, crtcport, crtcindex, mask, shift, flagarray, flagarraymask, cmpval); + + data = bios_idxprt_rd(bios, crtcport, crtcindex); + + data = bios->data[flagarray + ((data & mask) >> shift)]; + data &= flagarraymask; + + BIOSLOG(bios, "0x%04X: Checking if 0x%02X equals 0x%02X\n", + offset, data, cmpval); + + return (data == cmpval); +} + +static bool +bios_condition_met(struct nvbios *bios, uint16_t offset, uint8_t cond) +{ + /* + * The condition table entry has 4 bytes for the address of the + * register to check, 4 bytes for a mask to apply to the register and + * 4 for a test comparison value + */ + + uint16_t condptr = bios->condition_tbl_ptr + cond * CONDITION_SIZE; + uint32_t reg = ROM32(bios->data[condptr]); + uint32_t mask = ROM32(bios->data[condptr + 4]); + uint32_t cmpval = ROM32(bios->data[condptr + 8]); + uint32_t data; + + BIOSLOG(bios, "0x%04X: Cond: 0x%02X, Reg: 0x%08X, Mask: 0x%08X\n", + offset, cond, reg, mask); + + data = bios_rd32(bios, reg) & mask; + + BIOSLOG(bios, "0x%04X: Checking if 0x%08X equals 0x%08X\n", + offset, data, cmpval); + + return (data == cmpval); +} + +static bool +io_condition_met(struct nvbios *bios, uint16_t offset, uint8_t cond) +{ + /* + * The IO condition entry has 2 bytes for the IO port address; 1 byte + * for the index to write to io_port; 1 byte for the mask to apply to + * the byte read from io_port+1; and 1 byte for the value to compare + * against the masked byte. + */ + + uint16_t condptr = bios->io_condition_tbl_ptr + cond * IO_CONDITION_SIZE; + uint16_t io_port = ROM16(bios->data[condptr]); + uint8_t port_index = bios->data[condptr + 2]; + uint8_t mask = bios->data[condptr + 3]; + uint8_t cmpval = bios->data[condptr + 4]; + + uint8_t data = bios_idxprt_rd(bios, io_port, port_index) & mask; + + BIOSLOG(bios, "0x%04X: Checking if 0x%02X equals 0x%02X\n", + offset, data, cmpval); + + return (data == cmpval); +} + +static int +nv50_pll_set(struct drm_device *dev, uint32_t reg, uint32_t clk) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + uint32_t reg0 = nv_rd32(dev, reg + 0); + uint32_t reg1 = nv_rd32(dev, reg + 4); + struct nouveau_pll_vals pll; + struct pll_lims pll_limits; + int ret; + + ret = get_pll_limits(dev, reg, &pll_limits); + if (ret) + return ret; + + clk = nouveau_calc_pll_mnp(dev, &pll_limits, clk, &pll); + if (!clk) + return -ERANGE; + + reg0 = (reg0 & 0xfff8ffff) | (pll.log2P << 16); + reg1 = (reg1 & 0xffff0000) | (pll.N1 << 8) | pll.M1; + + if (dev_priv->vbios.execute) { + still_alive(); + nv_wr32(dev, reg + 4, reg1); + nv_wr32(dev, reg + 0, reg0); + } + + return 0; +} + +static int +setPLL(struct nvbios *bios, uint32_t reg, uint32_t clk) +{ + struct drm_device *dev = bios->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + /* clk in kHz */ + struct pll_lims pll_lim; + struct nouveau_pll_vals pllvals; + int ret; + + if (dev_priv->card_type >= NV_50) + return nv50_pll_set(dev, reg, clk); + + /* high regs (such as in the mac g5 table) are not -= 4 */ + ret = get_pll_limits(dev, reg > 0x405c ? reg : reg - 4, &pll_lim); + if (ret) + return ret; + + clk = nouveau_calc_pll_mnp(dev, &pll_lim, clk, &pllvals); + if (!clk) + return -ERANGE; + + if (bios->execute) { + still_alive(); + nouveau_hw_setpll(dev, reg, &pllvals); + } + + return 0; +} + +static int dcb_entry_idx_from_crtchead(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nvbios *bios = &dev_priv->vbios; + + /* + * For the results of this function to be correct, CR44 must have been + * set (using bios_idxprt_wr to set crtchead), CR58 set for CR57 = 0, + * and the DCB table parsed, before the script calling the function is + * run. run_digital_op_script is example of how to do such setup + */ + + uint8_t dcb_entry = NVReadVgaCrtc5758(dev, bios->state.crtchead, 0); + + if (dcb_entry > bios->dcb.entries) { + NV_ERROR(dev, "CR58 doesn't have a valid DCB entry currently " + "(%02X)\n", dcb_entry); + dcb_entry = 0x7f; /* unused / invalid marker */ + } + + return dcb_entry; +} + +static int +read_dcb_i2c_entry(struct drm_device *dev, int dcb_version, uint8_t *i2ctable, int index, struct dcb_i2c_entry *i2c) +{ + uint8_t dcb_i2c_ver = dcb_version, headerlen = 0, entry_len = 4; + int i2c_entries = DCB_MAX_NUM_I2C_ENTRIES; + int recordoffset = 0, rdofs = 1, wrofs = 0; + uint8_t port_type = 0; + + if (!i2ctable) + return -EINVAL; + + if (dcb_version >= 0x30) { + if (i2ctable[0] != dcb_version) /* necessary? */ + NV_WARN(dev, + "DCB I2C table version mismatch (%02X vs %02X)\n", + i2ctable[0], dcb_version); + dcb_i2c_ver = i2ctable[0]; + headerlen = i2ctable[1]; + if (i2ctable[2] <= DCB_MAX_NUM_I2C_ENTRIES) + i2c_entries = i2ctable[2]; + else + NV_WARN(dev, + "DCB I2C table has more entries than indexable " + "(%d entries, max %d)\n", i2ctable[2], + DCB_MAX_NUM_I2C_ENTRIES); + entry_len = i2ctable[3]; + /* [4] is i2c_default_indices, read in parse_dcb_table() */ + } + /* + * It's your own fault if you call this function on a DCB 1.1 BIOS -- + * the test below is for DCB 1.2 + */ + if (dcb_version < 0x14) { + recordoffset = 2; + rdofs = 0; + wrofs = 1; + } + + if (index == 0xf) + return 0; + if (index >= i2c_entries) { + NV_ERROR(dev, "DCB I2C index too big (%d >= %d)\n", + index, i2ctable[2]); + return -ENOENT; + } + if (i2ctable[headerlen + entry_len * index + 3] == 0xff) { + NV_ERROR(dev, "DCB I2C entry invalid\n"); + return -EINVAL; + } + + if (dcb_i2c_ver >= 0x30) { + port_type = i2ctable[headerlen + recordoffset + 3 + entry_len * index]; + + /* + * Fixup for chips using same address offset for read and + * write. + */ + if (port_type == 4) /* seen on C51 */ + rdofs = wrofs = 1; + if (port_type >= 5) /* G80+ */ + rdofs = wrofs = 0; + } + + if (dcb_i2c_ver >= 0x40) { + if (port_type != 5 && port_type != 6) + NV_WARN(dev, "DCB I2C table has port type %d\n", port_type); + + i2c->entry = ROM32(i2ctable[headerlen + recordoffset + entry_len * index]); + } + + i2c->port_type = port_type; + i2c->read = i2ctable[headerlen + recordoffset + rdofs + entry_len * index]; + i2c->write = i2ctable[headerlen + recordoffset + wrofs + entry_len * index]; + + return 0; +} + +static struct nouveau_i2c_chan * +init_i2c_device_find(struct drm_device *dev, int i2c_index) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct dcb_table *dcb = &dev_priv->vbios.dcb; + + if (i2c_index == 0xff) { + /* note: dcb_entry_idx_from_crtchead needs pre-script set-up */ + int idx = dcb_entry_idx_from_crtchead(dev), shift = 0; + int default_indices = dcb->i2c_default_indices; + + if (idx != 0x7f && dcb->entry[idx].i2c_upper_default) + shift = 4; + + i2c_index = (default_indices >> shift) & 0xf; + } + if (i2c_index == 0x80) /* g80+ */ + i2c_index = dcb->i2c_default_indices & 0xf; + else + if (i2c_index == 0x81) + i2c_index = (dcb->i2c_default_indices & 0xf0) >> 4; + + if (i2c_index >= DCB_MAX_NUM_I2C_ENTRIES) { + NV_ERROR(dev, "invalid i2c_index 0x%x\n", i2c_index); + return NULL; + } + + /* Make sure i2c table entry has been parsed, it may not + * have been if this is a bus not referenced by a DCB encoder + */ + read_dcb_i2c_entry(dev, dcb->version, dcb->i2c_table, + i2c_index, &dcb->i2c[i2c_index]); + + return nouveau_i2c_find(dev, i2c_index); +} + +static uint32_t +get_tmds_index_reg(struct drm_device *dev, uint8_t mlv) +{ + /* + * For mlv < 0x80, it is an index into a table of TMDS base addresses. + * For mlv == 0x80 use the "or" value of the dcb_entry indexed by + * CR58 for CR57 = 0 to index a table of offsets to the basic + * 0x6808b0 address. + * For mlv == 0x81 use the "or" value of the dcb_entry indexed by + * CR58 for CR57 = 0 to index a table of offsets to the basic + * 0x6808b0 address, and then flip the offset by 8. + */ + + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nvbios *bios = &dev_priv->vbios; + const int pramdac_offset[13] = { + 0, 0, 0x8, 0, 0x2000, 0, 0, 0, 0x2008, 0, 0, 0, 0x2000 }; + const uint32_t pramdac_table[4] = { + 0x6808b0, 0x6808b8, 0x6828b0, 0x6828b8 }; + + if (mlv >= 0x80) { + int dcb_entry, dacoffset; + + /* note: dcb_entry_idx_from_crtchead needs pre-script set-up */ + dcb_entry = dcb_entry_idx_from_crtchead(dev); + if (dcb_entry == 0x7f) + return 0; + dacoffset = pramdac_offset[bios->dcb.entry[dcb_entry].or]; + if (mlv == 0x81) + dacoffset ^= 8; + return 0x6808b0 + dacoffset; + } else { + if (mlv >= ARRAY_SIZE(pramdac_table)) { + NV_ERROR(dev, "Magic Lookup Value too big (%02X)\n", + mlv); + return 0; + } + return pramdac_table[mlv]; + } +} + +static int +init_io_restrict_prog(struct nvbios *bios, uint16_t offset, + struct init_exec *iexec) +{ + /* + * INIT_IO_RESTRICT_PROG opcode: 0x32 ('2') + * + * offset (8 bit): opcode + * offset + 1 (16 bit): CRTC port + * offset + 3 (8 bit): CRTC index + * offset + 4 (8 bit): mask + * offset + 5 (8 bit): shift + * offset + 6 (8 bit): count + * offset + 7 (32 bit): register + * offset + 11 (32 bit): configuration 1 + * ... + * + * Starting at offset + 11 there are "count" 32 bit values. + * To find out which value to use read index "CRTC index" on "CRTC + * port", AND this value with "mask" and then bit shift right "shift" + * bits. Read the appropriate value using this index and write to + * "register" + */ + + uint16_t crtcport = ROM16(bios->data[offset + 1]); + uint8_t crtcindex = bios->data[offset + 3]; + uint8_t mask = bios->data[offset + 4]; + uint8_t shift = bios->data[offset + 5]; + uint8_t count = bios->data[offset + 6]; + uint32_t reg = ROM32(bios->data[offset + 7]); + uint8_t config; + uint32_t configval; + int len = 11 + count * 4; + + if (!iexec->execute) + return len; + + BIOSLOG(bios, "0x%04X: Port: 0x%04X, Index: 0x%02X, Mask: 0x%02X, " + "Shift: 0x%02X, Count: 0x%02X, Reg: 0x%08X\n", + offset, crtcport, crtcindex, mask, shift, count, reg); + + config = (bios_idxprt_rd(bios, crtcport, crtcindex) & mask) >> shift; + if (config > count) { + NV_ERROR(bios->dev, + "0x%04X: Config 0x%02X exceeds maximal bound 0x%02X\n", + offset, config, count); + return len; + } + + configval = ROM32(bios->data[offset + 11 + config * 4]); + + BIOSLOG(bios, "0x%04X: Writing config %02X\n", offset, config); + + bios_wr32(bios, reg, configval); + + return len; +} + +static int +init_repeat(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_REPEAT opcode: 0x33 ('3') + * + * offset (8 bit): opcode + * offset + 1 (8 bit): count + * + * Execute script following this opcode up to INIT_REPEAT_END + * "count" times + */ + + uint8_t count = bios->data[offset + 1]; + uint8_t i; + + /* no iexec->execute check by design */ + + BIOSLOG(bios, "0x%04X: Repeating following segment %d times\n", + offset, count); + + iexec->repeat = true; + + /* + * count - 1, as the script block will execute once when we leave this + * opcode -- this is compatible with bios behaviour as: + * a) the block is always executed at least once, even if count == 0 + * b) the bios interpreter skips to the op following INIT_END_REPEAT, + * while we don't + */ + for (i = 0; i < count - 1; i++) + parse_init_table(bios, offset + 2, iexec); + + iexec->repeat = false; + + return 2; +} + +static int +init_io_restrict_pll(struct nvbios *bios, uint16_t offset, + struct init_exec *iexec) +{ + /* + * INIT_IO_RESTRICT_PLL opcode: 0x34 ('4') + * + * offset (8 bit): opcode + * offset + 1 (16 bit): CRTC port + * offset + 3 (8 bit): CRTC index + * offset + 4 (8 bit): mask + * offset + 5 (8 bit): shift + * offset + 6 (8 bit): IO flag condition index + * offset + 7 (8 bit): count + * offset + 8 (32 bit): register + * offset + 12 (16 bit): frequency 1 + * ... + * + * Starting at offset + 12 there are "count" 16 bit frequencies (10kHz). + * Set PLL register "register" to coefficients for frequency n, + * selected by reading index "CRTC index" of "CRTC port" ANDed with + * "mask" and shifted right by "shift". + * + * If "IO flag condition index" > 0, and condition met, double + * frequency before setting it. + */ + + uint16_t crtcport = ROM16(bios->data[offset + 1]); + uint8_t crtcindex = bios->data[offset + 3]; + uint8_t mask = bios->data[offset + 4]; + uint8_t shift = bios->data[offset + 5]; + int8_t io_flag_condition_idx = bios->data[offset + 6]; + uint8_t count = bios->data[offset + 7]; + uint32_t reg = ROM32(bios->data[offset + 8]); + uint8_t config; + uint16_t freq; + int len = 12 + count * 2; + + if (!iexec->execute) + return len; + + BIOSLOG(bios, "0x%04X: Port: 0x%04X, Index: 0x%02X, Mask: 0x%02X, " + "Shift: 0x%02X, IO Flag Condition: 0x%02X, " + "Count: 0x%02X, Reg: 0x%08X\n", + offset, crtcport, crtcindex, mask, shift, + io_flag_condition_idx, count, reg); + + config = (bios_idxprt_rd(bios, crtcport, crtcindex) & mask) >> shift; + if (config > count) { + NV_ERROR(bios->dev, + "0x%04X: Config 0x%02X exceeds maximal bound 0x%02X\n", + offset, config, count); + return len; + } + + freq = ROM16(bios->data[offset + 12 + config * 2]); + + if (io_flag_condition_idx > 0) { + if (io_flag_condition_met(bios, offset, io_flag_condition_idx)) { + BIOSLOG(bios, "0x%04X: Condition fulfilled -- " + "frequency doubled\n", offset); + freq *= 2; + } else + BIOSLOG(bios, "0x%04X: Condition not fulfilled -- " + "frequency unchanged\n", offset); + } + + BIOSLOG(bios, "0x%04X: Reg: 0x%08X, Config: 0x%02X, Freq: %d0kHz\n", + offset, reg, config, freq); + + setPLL(bios, reg, freq * 10); + + return len; +} + +static int +init_end_repeat(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_END_REPEAT opcode: 0x36 ('6') + * + * offset (8 bit): opcode + * + * Marks the end of the block for INIT_REPEAT to repeat + */ + + /* no iexec->execute check by design */ + + /* + * iexec->repeat flag necessary to go past INIT_END_REPEAT opcode when + * we're not in repeat mode + */ + if (iexec->repeat) + return 0; + + return 1; +} + +static int +init_copy(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_COPY opcode: 0x37 ('7') + * + * offset (8 bit): opcode + * offset + 1 (32 bit): register + * offset + 5 (8 bit): shift + * offset + 6 (8 bit): srcmask + * offset + 7 (16 bit): CRTC port + * offset + 9 (8 bit): CRTC index + * offset + 10 (8 bit): mask + * + * Read index "CRTC index" on "CRTC port", AND with "mask", OR with + * (REGVAL("register") >> "shift" & "srcmask") and write-back to CRTC + * port + */ + + uint32_t reg = ROM32(bios->data[offset + 1]); + uint8_t shift = bios->data[offset + 5]; + uint8_t srcmask = bios->data[offset + 6]; + uint16_t crtcport = ROM16(bios->data[offset + 7]); + uint8_t crtcindex = bios->data[offset + 9]; + uint8_t mask = bios->data[offset + 10]; + uint32_t data; + uint8_t crtcdata; + + if (!iexec->execute) + return 11; + + BIOSLOG(bios, "0x%04X: Reg: 0x%08X, Shift: 0x%02X, SrcMask: 0x%02X, " + "Port: 0x%04X, Index: 0x%02X, Mask: 0x%02X\n", + offset, reg, shift, srcmask, crtcport, crtcindex, mask); + + data = bios_rd32(bios, reg); + + if (shift < 0x80) + data >>= shift; + else + data <<= (0x100 - shift); + + data &= srcmask; + + crtcdata = bios_idxprt_rd(bios, crtcport, crtcindex) & mask; + crtcdata |= (uint8_t)data; + bios_idxprt_wr(bios, crtcport, crtcindex, crtcdata); + + return 11; +} + +static int +init_not(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_NOT opcode: 0x38 ('8') + * + * offset (8 bit): opcode + * + * Invert the current execute / no-execute condition (i.e. "else") + */ + if (iexec->execute) + BIOSLOG(bios, "0x%04X: ------ Skipping following commands ------\n", offset); + else + BIOSLOG(bios, "0x%04X: ------ Executing following commands ------\n", offset); + + iexec->execute = !iexec->execute; + return 1; +} + +static int +init_io_flag_condition(struct nvbios *bios, uint16_t offset, + struct init_exec *iexec) +{ + /* + * INIT_IO_FLAG_CONDITION opcode: 0x39 ('9') + * + * offset (8 bit): opcode + * offset + 1 (8 bit): condition number + * + * Check condition "condition number" in the IO flag condition table. + * If condition not met skip subsequent opcodes until condition is + * inverted (INIT_NOT), or we hit INIT_RESUME + */ + + uint8_t cond = bios->data[offset + 1]; + + if (!iexec->execute) + return 2; + + if (io_flag_condition_met(bios, offset, cond)) + BIOSLOG(bios, "0x%04X: Condition fulfilled -- continuing to execute\n", offset); + else { + BIOSLOG(bios, "0x%04X: Condition not fulfilled -- skipping following commands\n", offset); + iexec->execute = false; + } + + return 2; +} + +static int +init_dp_condition(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_DP_CONDITION opcode: 0x3A ('') + * + * offset (8 bit): opcode + * offset + 1 (8 bit): "sub" opcode + * offset + 2 (8 bit): unknown + * + */ + + struct bit_displayport_encoder_table *dpe = NULL; + struct dcb_entry *dcb = bios->display.output; + struct drm_device *dev = bios->dev; + uint8_t cond = bios->data[offset + 1]; + int dummy; + + BIOSLOG(bios, "0x%04X: subop 0x%02X\n", offset, cond); + + if (!iexec->execute) + return 3; + + dpe = nouveau_bios_dp_table(dev, dcb, &dummy); + if (!dpe) { + NV_ERROR(dev, "0x%04X: INIT_3A: no encoder table!!\n", offset); + return 3; + } + + switch (cond) { + case 0: + { + struct dcb_connector_table_entry *ent = + &bios->dcb.connector.entry[dcb->connector]; + + if (ent->type != DCB_CONNECTOR_eDP) + iexec->execute = false; + } + break; + case 1: + case 2: + if (!(dpe->unknown & cond)) + iexec->execute = false; + break; + case 5: + { + struct nouveau_i2c_chan *auxch; + int ret; + + auxch = nouveau_i2c_find(dev, bios->display.output->i2c_index); + if (!auxch) { + NV_ERROR(dev, "0x%04X: couldn't get auxch\n", offset); + return 3; + } + + ret = nouveau_dp_auxch(auxch, 9, 0xd, &cond, 1); + if (ret) { + NV_ERROR(dev, "0x%04X: auxch rd fail: %d\n", offset, ret); + return 3; + } + + if (!(cond & 1)) + iexec->execute = false; + } + break; + default: + NV_WARN(dev, "0x%04X: unknown INIT_3A op: %d\n", offset, cond); + break; + } + + if (iexec->execute) + BIOSLOG(bios, "0x%04X: continuing to execute\n", offset); + else + BIOSLOG(bios, "0x%04X: skipping following commands\n", offset); + + return 3; +} + +static int +init_op_3b(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_3B opcode: 0x3B ('') + * + * offset (8 bit): opcode + * offset + 1 (8 bit): crtc index + * + */ + + uint8_t or = ffs(bios->display.output->or) - 1; + uint8_t index = bios->data[offset + 1]; + uint8_t data; + + if (!iexec->execute) + return 2; + + data = bios_idxprt_rd(bios, 0x3d4, index); + bios_idxprt_wr(bios, 0x3d4, index, data & ~(1 << or)); + return 2; +} + +static int +init_op_3c(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_3C opcode: 0x3C ('') + * + * offset (8 bit): opcode + * offset + 1 (8 bit): crtc index + * + */ + + uint8_t or = ffs(bios->display.output->or) - 1; + uint8_t index = bios->data[offset + 1]; + uint8_t data; + + if (!iexec->execute) + return 2; + + data = bios_idxprt_rd(bios, 0x3d4, index); + bios_idxprt_wr(bios, 0x3d4, index, data | (1 << or)); + return 2; +} + +static int +init_idx_addr_latched(struct nvbios *bios, uint16_t offset, + struct init_exec *iexec) +{ + /* + * INIT_INDEX_ADDRESS_LATCHED opcode: 0x49 ('I') + * + * offset (8 bit): opcode + * offset + 1 (32 bit): control register + * offset + 5 (32 bit): data register + * offset + 9 (32 bit): mask + * offset + 13 (32 bit): data + * offset + 17 (8 bit): count + * offset + 18 (8 bit): address 1 + * offset + 19 (8 bit): data 1 + * ... + * + * For each of "count" address and data pairs, write "data n" to + * "data register", read the current value of "control register", + * and write it back once ANDed with "mask", ORed with "data", + * and ORed with "address n" + */ + + uint32_t controlreg = ROM32(bios->data[offset + 1]); + uint32_t datareg = ROM32(bios->data[offset + 5]); + uint32_t mask = ROM32(bios->data[offset + 9]); + uint32_t data = ROM32(bios->data[offset + 13]); + uint8_t count = bios->data[offset + 17]; + int len = 18 + count * 2; + uint32_t value; + int i; + + if (!iexec->execute) + return len; + + BIOSLOG(bios, "0x%04X: ControlReg: 0x%08X, DataReg: 0x%08X, " + "Mask: 0x%08X, Data: 0x%08X, Count: 0x%02X\n", + offset, controlreg, datareg, mask, data, count); + + for (i = 0; i < count; i++) { + uint8_t instaddress = bios->data[offset + 18 + i * 2]; + uint8_t instdata = bios->data[offset + 19 + i * 2]; + + BIOSLOG(bios, "0x%04X: Address: 0x%02X, Data: 0x%02X\n", + offset, instaddress, instdata); + + bios_wr32(bios, datareg, instdata); + value = bios_rd32(bios, controlreg) & mask; + value |= data; + value |= instaddress; + bios_wr32(bios, controlreg, value); + } + + return len; +} + +static int +init_io_restrict_pll2(struct nvbios *bios, uint16_t offset, + struct init_exec *iexec) +{ + /* + * INIT_IO_RESTRICT_PLL2 opcode: 0x4A ('J') + * + * offset (8 bit): opcode + * offset + 1 (16 bit): CRTC port + * offset + 3 (8 bit): CRTC index + * offset + 4 (8 bit): mask + * offset + 5 (8 bit): shift + * offset + 6 (8 bit): count + * offset + 7 (32 bit): register + * offset + 11 (32 bit): frequency 1 + * ... + * + * Starting at offset + 11 there are "count" 32 bit frequencies (kHz). + * Set PLL register "register" to coefficients for frequency n, + * selected by reading index "CRTC index" of "CRTC port" ANDed with + * "mask" and shifted right by "shift". + */ + + uint16_t crtcport = ROM16(bios->data[offset + 1]); + uint8_t crtcindex = bios->data[offset + 3]; + uint8_t mask = bios->data[offset + 4]; + uint8_t shift = bios->data[offset + 5]; + uint8_t count = bios->data[offset + 6]; + uint32_t reg = ROM32(bios->data[offset + 7]); + int len = 11 + count * 4; + uint8_t config; + uint32_t freq; + + if (!iexec->execute) + return len; + + BIOSLOG(bios, "0x%04X: Port: 0x%04X, Index: 0x%02X, Mask: 0x%02X, " + "Shift: 0x%02X, Count: 0x%02X, Reg: 0x%08X\n", + offset, crtcport, crtcindex, mask, shift, count, reg); + + if (!reg) + return len; + + config = (bios_idxprt_rd(bios, crtcport, crtcindex) & mask) >> shift; + if (config > count) { + NV_ERROR(bios->dev, + "0x%04X: Config 0x%02X exceeds maximal bound 0x%02X\n", + offset, config, count); + return len; + } + + freq = ROM32(bios->data[offset + 11 + config * 4]); + + BIOSLOG(bios, "0x%04X: Reg: 0x%08X, Config: 0x%02X, Freq: %dkHz\n", + offset, reg, config, freq); + + setPLL(bios, reg, freq); + + return len; +} + +static int +init_pll2(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_PLL2 opcode: 0x4B ('K') + * + * offset (8 bit): opcode + * offset + 1 (32 bit): register + * offset + 5 (32 bit): freq + * + * Set PLL register "register" to coefficients for frequency "freq" + */ + + uint32_t reg = ROM32(bios->data[offset + 1]); + uint32_t freq = ROM32(bios->data[offset + 5]); + + if (!iexec->execute) + return 9; + + BIOSLOG(bios, "0x%04X: Reg: 0x%04X, Freq: %dkHz\n", + offset, reg, freq); + + setPLL(bios, reg, freq); + return 9; +} + +static int +init_i2c_byte(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_I2C_BYTE opcode: 0x4C ('L') + * + * offset (8 bit): opcode + * offset + 1 (8 bit): DCB I2C table entry index + * offset + 2 (8 bit): I2C slave address + * offset + 3 (8 bit): count + * offset + 4 (8 bit): I2C register 1 + * offset + 5 (8 bit): mask 1 + * offset + 6 (8 bit): data 1 + * ... + * + * For each of "count" registers given by "I2C register n" on the device + * addressed by "I2C slave address" on the I2C bus given by + * "DCB I2C table entry index", read the register, AND the result with + * "mask n" and OR it with "data n" before writing it back to the device + */ + + struct drm_device *dev = bios->dev; + uint8_t i2c_index = bios->data[offset + 1]; + uint8_t i2c_address = bios->data[offset + 2] >> 1; + uint8_t count = bios->data[offset + 3]; + struct nouveau_i2c_chan *chan; + int len = 4 + count * 3; + int ret, i; + + if (!iexec->execute) + return len; + + BIOSLOG(bios, "0x%04X: DCBI2CIndex: 0x%02X, I2CAddress: 0x%02X, " + "Count: 0x%02X\n", + offset, i2c_index, i2c_address, count); + + chan = init_i2c_device_find(dev, i2c_index); + if (!chan) { + NV_ERROR(dev, "0x%04X: i2c bus not found\n", offset); + return len; + } + + for (i = 0; i < count; i++) { + uint8_t reg = bios->data[offset + 4 + i * 3]; + uint8_t mask = bios->data[offset + 5 + i * 3]; + uint8_t data = bios->data[offset + 6 + i * 3]; + union i2c_smbus_data val; + + ret = i2c_smbus_xfer(&chan->adapter, i2c_address, 0, + I2C_SMBUS_READ, reg, + I2C_SMBUS_BYTE_DATA, &val); + if (ret < 0) { + NV_ERROR(dev, "0x%04X: i2c rd fail: %d\n", offset, ret); + return len; + } + + BIOSLOG(bios, "0x%04X: I2CReg: 0x%02X, Value: 0x%02X, " + "Mask: 0x%02X, Data: 0x%02X\n", + offset, reg, val.byte, mask, data); + + if (!bios->execute) + continue; + + val.byte &= mask; + val.byte |= data; + ret = i2c_smbus_xfer(&chan->adapter, i2c_address, 0, + I2C_SMBUS_WRITE, reg, + I2C_SMBUS_BYTE_DATA, &val); + if (ret < 0) { + NV_ERROR(dev, "0x%04X: i2c wr fail: %d\n", offset, ret); + return len; + } + } + + return len; +} + +static int +init_zm_i2c_byte(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_ZM_I2C_BYTE opcode: 0x4D ('M') + * + * offset (8 bit): opcode + * offset + 1 (8 bit): DCB I2C table entry index + * offset + 2 (8 bit): I2C slave address + * offset + 3 (8 bit): count + * offset + 4 (8 bit): I2C register 1 + * offset + 5 (8 bit): data 1 + * ... + * + * For each of "count" registers given by "I2C register n" on the device + * addressed by "I2C slave address" on the I2C bus given by + * "DCB I2C table entry index", set the register to "data n" + */ + + struct drm_device *dev = bios->dev; + uint8_t i2c_index = bios->data[offset + 1]; + uint8_t i2c_address = bios->data[offset + 2] >> 1; + uint8_t count = bios->data[offset + 3]; + struct nouveau_i2c_chan *chan; + int len = 4 + count * 2; + int ret, i; + + if (!iexec->execute) + return len; + + BIOSLOG(bios, "0x%04X: DCBI2CIndex: 0x%02X, I2CAddress: 0x%02X, " + "Count: 0x%02X\n", + offset, i2c_index, i2c_address, count); + + chan = init_i2c_device_find(dev, i2c_index); + if (!chan) { + NV_ERROR(dev, "0x%04X: i2c bus not found\n", offset); + return len; + } + + for (i = 0; i < count; i++) { + uint8_t reg = bios->data[offset + 4 + i * 2]; + union i2c_smbus_data val; + + val.byte = bios->data[offset + 5 + i * 2]; + + BIOSLOG(bios, "0x%04X: I2CReg: 0x%02X, Data: 0x%02X\n", + offset, reg, val.byte); + + if (!bios->execute) + continue; + + ret = i2c_smbus_xfer(&chan->adapter, i2c_address, 0, + I2C_SMBUS_WRITE, reg, + I2C_SMBUS_BYTE_DATA, &val); + if (ret < 0) { + NV_ERROR(dev, "0x%04X: i2c wr fail: %d\n", offset, ret); + return len; + } + } + + return len; +} + +static int +init_zm_i2c(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_ZM_I2C opcode: 0x4E ('N') + * + * offset (8 bit): opcode + * offset + 1 (8 bit): DCB I2C table entry index + * offset + 2 (8 bit): I2C slave address + * offset + 3 (8 bit): count + * offset + 4 (8 bit): data 1 + * ... + * + * Send "count" bytes ("data n") to the device addressed by "I2C slave + * address" on the I2C bus given by "DCB I2C table entry index" + */ + + struct drm_device *dev = bios->dev; + uint8_t i2c_index = bios->data[offset + 1]; + uint8_t i2c_address = bios->data[offset + 2] >> 1; + uint8_t count = bios->data[offset + 3]; + int len = 4 + count; + struct nouveau_i2c_chan *chan; + struct i2c_msg msg; + uint8_t data[256]; + int ret, i; + + if (!iexec->execute) + return len; + + BIOSLOG(bios, "0x%04X: DCBI2CIndex: 0x%02X, I2CAddress: 0x%02X, " + "Count: 0x%02X\n", + offset, i2c_index, i2c_address, count); + + chan = init_i2c_device_find(dev, i2c_index); + if (!chan) { + NV_ERROR(dev, "0x%04X: i2c bus not found\n", offset); + return len; + } + + for (i = 0; i < count; i++) { + data[i] = bios->data[offset + 4 + i]; + + BIOSLOG(bios, "0x%04X: Data: 0x%02X\n", offset, data[i]); + } + + if (bios->execute) { + msg.addr = i2c_address; + msg.flags = 0; + msg.len = count; + msg.buf = data; + ret = i2c_transfer(&chan->adapter, &msg, 1); + if (ret != 1) { + NV_ERROR(dev, "0x%04X: i2c wr fail: %d\n", offset, ret); + return len; + } + } + + return len; +} + +static int +init_tmds(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_TMDS opcode: 0x4F ('O') (non-canon name) + * + * offset (8 bit): opcode + * offset + 1 (8 bit): magic lookup value + * offset + 2 (8 bit): TMDS address + * offset + 3 (8 bit): mask + * offset + 4 (8 bit): data + * + * Read the data reg for TMDS address "TMDS address", AND it with mask + * and OR it with data, then write it back + * "magic lookup value" determines which TMDS base address register is + * used -- see get_tmds_index_reg() + */ + + struct drm_device *dev = bios->dev; + uint8_t mlv = bios->data[offset + 1]; + uint32_t tmdsaddr = bios->data[offset + 2]; + uint8_t mask = bios->data[offset + 3]; + uint8_t data = bios->data[offset + 4]; + uint32_t reg, value; + + if (!iexec->execute) + return 5; + + BIOSLOG(bios, "0x%04X: MagicLookupValue: 0x%02X, TMDSAddr: 0x%02X, " + "Mask: 0x%02X, Data: 0x%02X\n", + offset, mlv, tmdsaddr, mask, data); + + reg = get_tmds_index_reg(bios->dev, mlv); + if (!reg) { + NV_ERROR(dev, "0x%04X: no tmds_index_reg\n", offset); + return 5; + } + + bios_wr32(bios, reg, + tmdsaddr | NV_PRAMDAC_FP_TMDS_CONTROL_WRITE_DISABLE); + value = (bios_rd32(bios, reg + 4) & mask) | data; + bios_wr32(bios, reg + 4, value); + bios_wr32(bios, reg, tmdsaddr); + + return 5; +} + +static int +init_zm_tmds_group(struct nvbios *bios, uint16_t offset, + struct init_exec *iexec) +{ + /* + * INIT_ZM_TMDS_GROUP opcode: 0x50 ('P') (non-canon name) + * + * offset (8 bit): opcode + * offset + 1 (8 bit): magic lookup value + * offset + 2 (8 bit): count + * offset + 3 (8 bit): addr 1 + * offset + 4 (8 bit): data 1 + * ... + * + * For each of "count" TMDS address and data pairs write "data n" to + * "addr n". "magic lookup value" determines which TMDS base address + * register is used -- see get_tmds_index_reg() + */ + + struct drm_device *dev = bios->dev; + uint8_t mlv = bios->data[offset + 1]; + uint8_t count = bios->data[offset + 2]; + int len = 3 + count * 2; + uint32_t reg; + int i; + + if (!iexec->execute) + return len; + + BIOSLOG(bios, "0x%04X: MagicLookupValue: 0x%02X, Count: 0x%02X\n", + offset, mlv, count); + + reg = get_tmds_index_reg(bios->dev, mlv); + if (!reg) { + NV_ERROR(dev, "0x%04X: no tmds_index_reg\n", offset); + return len; + } + + for (i = 0; i < count; i++) { + uint8_t tmdsaddr = bios->data[offset + 3 + i * 2]; + uint8_t tmdsdata = bios->data[offset + 4 + i * 2]; + + bios_wr32(bios, reg + 4, tmdsdata); + bios_wr32(bios, reg, tmdsaddr); + } + + return len; +} + +static int +init_cr_idx_adr_latch(struct nvbios *bios, uint16_t offset, + struct init_exec *iexec) +{ + /* + * INIT_CR_INDEX_ADDRESS_LATCHED opcode: 0x51 ('Q') + * + * offset (8 bit): opcode + * offset + 1 (8 bit): CRTC index1 + * offset + 2 (8 bit): CRTC index2 + * offset + 3 (8 bit): baseaddr + * offset + 4 (8 bit): count + * offset + 5 (8 bit): data 1 + * ... + * + * For each of "count" address and data pairs, write "baseaddr + n" to + * "CRTC index1" and "data n" to "CRTC index2" + * Once complete, restore initial value read from "CRTC index1" + */ + uint8_t crtcindex1 = bios->data[offset + 1]; + uint8_t crtcindex2 = bios->data[offset + 2]; + uint8_t baseaddr = bios->data[offset + 3]; + uint8_t count = bios->data[offset + 4]; + int len = 5 + count; + uint8_t oldaddr, data; + int i; + + if (!iexec->execute) + return len; + + BIOSLOG(bios, "0x%04X: Index1: 0x%02X, Index2: 0x%02X, " + "BaseAddr: 0x%02X, Count: 0x%02X\n", + offset, crtcindex1, crtcindex2, baseaddr, count); + + oldaddr = bios_idxprt_rd(bios, NV_CIO_CRX__COLOR, crtcindex1); + + for (i = 0; i < count; i++) { + bios_idxprt_wr(bios, NV_CIO_CRX__COLOR, crtcindex1, + baseaddr + i); + data = bios->data[offset + 5 + i]; + bios_idxprt_wr(bios, NV_CIO_CRX__COLOR, crtcindex2, data); + } + + bios_idxprt_wr(bios, NV_CIO_CRX__COLOR, crtcindex1, oldaddr); + + return len; +} + +static int +init_cr(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_CR opcode: 0x52 ('R') + * + * offset (8 bit): opcode + * offset + 1 (8 bit): CRTC index + * offset + 2 (8 bit): mask + * offset + 3 (8 bit): data + * + * Assign the value of at "CRTC index" ANDed with mask and ORed with + * data back to "CRTC index" + */ + + uint8_t crtcindex = bios->data[offset + 1]; + uint8_t mask = bios->data[offset + 2]; + uint8_t data = bios->data[offset + 3]; + uint8_t value; + + if (!iexec->execute) + return 4; + + BIOSLOG(bios, "0x%04X: Index: 0x%02X, Mask: 0x%02X, Data: 0x%02X\n", + offset, crtcindex, mask, data); + + value = bios_idxprt_rd(bios, NV_CIO_CRX__COLOR, crtcindex) & mask; + value |= data; + bios_idxprt_wr(bios, NV_CIO_CRX__COLOR, crtcindex, value); + + return 4; +} + +static int +init_zm_cr(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_ZM_CR opcode: 0x53 ('S') + * + * offset (8 bit): opcode + * offset + 1 (8 bit): CRTC index + * offset + 2 (8 bit): value + * + * Assign "value" to CRTC register with index "CRTC index". + */ + + uint8_t crtcindex = ROM32(bios->data[offset + 1]); + uint8_t data = bios->data[offset + 2]; + + if (!iexec->execute) + return 3; + + bios_idxprt_wr(bios, NV_CIO_CRX__COLOR, crtcindex, data); + + return 3; +} + +static int +init_zm_cr_group(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_ZM_CR_GROUP opcode: 0x54 ('T') + * + * offset (8 bit): opcode + * offset + 1 (8 bit): count + * offset + 2 (8 bit): CRTC index 1 + * offset + 3 (8 bit): value 1 + * ... + * + * For "count", assign "value n" to CRTC register with index + * "CRTC index n". + */ + + uint8_t count = bios->data[offset + 1]; + int len = 2 + count * 2; + int i; + + if (!iexec->execute) + return len; + + for (i = 0; i < count; i++) + init_zm_cr(bios, offset + 2 + 2 * i - 1, iexec); + + return len; +} + +static int +init_condition_time(struct nvbios *bios, uint16_t offset, + struct init_exec *iexec) +{ + /* + * INIT_CONDITION_TIME opcode: 0x56 ('V') + * + * offset (8 bit): opcode + * offset + 1 (8 bit): condition number + * offset + 2 (8 bit): retries / 50 + * + * Check condition "condition number" in the condition table. + * Bios code then sleeps for 2ms if the condition is not met, and + * repeats up to "retries" times, but on one C51 this has proved + * insufficient. In mmiotraces the driver sleeps for 20ms, so we do + * this, and bail after "retries" times, or 2s, whichever is less. + * If still not met after retries, clear execution flag for this table. + */ + + uint8_t cond = bios->data[offset + 1]; + uint16_t retries = bios->data[offset + 2] * 50; + unsigned cnt; + + if (!iexec->execute) + return 3; + + if (retries > 100) + retries = 100; + + BIOSLOG(bios, "0x%04X: Condition: 0x%02X, Retries: 0x%02X\n", + offset, cond, retries); + + if (!bios->execute) /* avoid 2s delays when "faking" execution */ + retries = 1; + + for (cnt = 0; cnt < retries; cnt++) { + if (bios_condition_met(bios, offset, cond)) { + BIOSLOG(bios, "0x%04X: Condition met, continuing\n", + offset); + break; + } else { + BIOSLOG(bios, "0x%04X: " + "Condition not met, sleeping for 20ms\n", + offset); + mdelay(20); + } + } + + if (!bios_condition_met(bios, offset, cond)) { + NV_WARN(bios->dev, + "0x%04X: Condition still not met after %dms, " + "skipping following opcodes\n", offset, 20 * retries); + iexec->execute = false; + } + + return 3; +} + +static int +init_ltime(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_LTIME opcode: 0x57 ('V') + * + * offset (8 bit): opcode + * offset + 1 (16 bit): time + * + * Sleep for "time" milliseconds. + */ + + unsigned time = ROM16(bios->data[offset + 1]); + + if (!iexec->execute) + return 3; + + BIOSLOG(bios, "0x%04X: Sleeping for 0x%04X milliseconds\n", + offset, time); + + mdelay(time); + + return 3; +} + +static int +init_zm_reg_sequence(struct nvbios *bios, uint16_t offset, + struct init_exec *iexec) +{ + /* + * INIT_ZM_REG_SEQUENCE opcode: 0x58 ('X') + * + * offset (8 bit): opcode + * offset + 1 (32 bit): base register + * offset + 5 (8 bit): count + * offset + 6 (32 bit): value 1 + * ... + * + * Starting at offset + 6 there are "count" 32 bit values. + * For "count" iterations set "base register" + 4 * current_iteration + * to "value current_iteration" + */ + + uint32_t basereg = ROM32(bios->data[offset + 1]); + uint32_t count = bios->data[offset + 5]; + int len = 6 + count * 4; + int i; + + if (!iexec->execute) + return len; + + BIOSLOG(bios, "0x%04X: BaseReg: 0x%08X, Count: 0x%02X\n", + offset, basereg, count); + + for (i = 0; i < count; i++) { + uint32_t reg = basereg + i * 4; + uint32_t data = ROM32(bios->data[offset + 6 + i * 4]); + + bios_wr32(bios, reg, data); + } + + return len; +} + +static int +init_sub_direct(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_SUB_DIRECT opcode: 0x5B ('[') + * + * offset (8 bit): opcode + * offset + 1 (16 bit): subroutine offset (in bios) + * + * Calls a subroutine that will execute commands until INIT_DONE + * is found. + */ + + uint16_t sub_offset = ROM16(bios->data[offset + 1]); + + if (!iexec->execute) + return 3; + + BIOSLOG(bios, "0x%04X: Executing subroutine at 0x%04X\n", + offset, sub_offset); + + parse_init_table(bios, sub_offset, iexec); + + BIOSLOG(bios, "0x%04X: End of 0x%04X subroutine\n", offset, sub_offset); + + return 3; +} + +static int +init_jump(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_JUMP opcode: 0x5C ('\') + * + * offset (8 bit): opcode + * offset + 1 (16 bit): offset (in bios) + * + * Continue execution of init table from 'offset' + */ + + uint16_t jmp_offset = ROM16(bios->data[offset + 1]); + + if (!iexec->execute) + return 3; + + BIOSLOG(bios, "0x%04X: Jump to 0x%04X\n", offset, jmp_offset); + return jmp_offset - offset; +} + +static int +init_i2c_if(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_I2C_IF opcode: 0x5E ('^') + * + * offset (8 bit): opcode + * offset + 1 (8 bit): DCB I2C table entry index + * offset + 2 (8 bit): I2C slave address + * offset + 3 (8 bit): I2C register + * offset + 4 (8 bit): mask + * offset + 5 (8 bit): data + * + * Read the register given by "I2C register" on the device addressed + * by "I2C slave address" on the I2C bus given by "DCB I2C table + * entry index". Compare the result AND "mask" to "data". + * If they're not equal, skip subsequent opcodes until condition is + * inverted (INIT_NOT), or we hit INIT_RESUME + */ + + uint8_t i2c_index = bios->data[offset + 1]; + uint8_t i2c_address = bios->data[offset + 2] >> 1; + uint8_t reg = bios->data[offset + 3]; + uint8_t mask = bios->data[offset + 4]; + uint8_t data = bios->data[offset + 5]; + struct nouveau_i2c_chan *chan; + union i2c_smbus_data val; + int ret; + + /* no execute check by design */ + + BIOSLOG(bios, "0x%04X: DCBI2CIndex: 0x%02X, I2CAddress: 0x%02X\n", + offset, i2c_index, i2c_address); + + chan = init_i2c_device_find(bios->dev, i2c_index); + if (!chan) + return -ENODEV; + + ret = i2c_smbus_xfer(&chan->adapter, i2c_address, 0, + I2C_SMBUS_READ, reg, + I2C_SMBUS_BYTE_DATA, &val); + if (ret < 0) { + BIOSLOG(bios, "0x%04X: I2CReg: 0x%02X, Value: [no device], " + "Mask: 0x%02X, Data: 0x%02X\n", + offset, reg, mask, data); + iexec->execute = 0; + return 6; + } + + BIOSLOG(bios, "0x%04X: I2CReg: 0x%02X, Value: 0x%02X, " + "Mask: 0x%02X, Data: 0x%02X\n", + offset, reg, val.byte, mask, data); + + iexec->execute = ((val.byte & mask) == data); + + return 6; +} + +static int +init_copy_nv_reg(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_COPY_NV_REG opcode: 0x5F ('_') + * + * offset (8 bit): opcode + * offset + 1 (32 bit): src reg + * offset + 5 (8 bit): shift + * offset + 6 (32 bit): src mask + * offset + 10 (32 bit): xor + * offset + 14 (32 bit): dst reg + * offset + 18 (32 bit): dst mask + * + * Shift REGVAL("src reg") right by (signed) "shift", AND result with + * "src mask", then XOR with "xor". Write this OR'd with + * (REGVAL("dst reg") AND'd with "dst mask") to "dst reg" + */ + + uint32_t srcreg = *((uint32_t *)(&bios->data[offset + 1])); + uint8_t shift = bios->data[offset + 5]; + uint32_t srcmask = *((uint32_t *)(&bios->data[offset + 6])); + uint32_t xor = *((uint32_t *)(&bios->data[offset + 10])); + uint32_t dstreg = *((uint32_t *)(&bios->data[offset + 14])); + uint32_t dstmask = *((uint32_t *)(&bios->data[offset + 18])); + uint32_t srcvalue, dstvalue; + + if (!iexec->execute) + return 22; + + BIOSLOG(bios, "0x%04X: SrcReg: 0x%08X, Shift: 0x%02X, SrcMask: 0x%08X, " + "Xor: 0x%08X, DstReg: 0x%08X, DstMask: 0x%08X\n", + offset, srcreg, shift, srcmask, xor, dstreg, dstmask); + + srcvalue = bios_rd32(bios, srcreg); + + if (shift < 0x80) + srcvalue >>= shift; + else + srcvalue <<= (0x100 - shift); + + srcvalue = (srcvalue & srcmask) ^ xor; + + dstvalue = bios_rd32(bios, dstreg) & dstmask; + + bios_wr32(bios, dstreg, dstvalue | srcvalue); + + return 22; +} + +static int +init_zm_index_io(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_ZM_INDEX_IO opcode: 0x62 ('b') + * + * offset (8 bit): opcode + * offset + 1 (16 bit): CRTC port + * offset + 3 (8 bit): CRTC index + * offset + 4 (8 bit): data + * + * Write "data" to index "CRTC index" of "CRTC port" + */ + uint16_t crtcport = ROM16(bios->data[offset + 1]); + uint8_t crtcindex = bios->data[offset + 3]; + uint8_t data = bios->data[offset + 4]; + + if (!iexec->execute) + return 5; + + bios_idxprt_wr(bios, crtcport, crtcindex, data); + + return 5; +} + +static inline void +bios_md32(struct nvbios *bios, uint32_t reg, + uint32_t mask, uint32_t val) +{ + bios_wr32(bios, reg, (bios_rd32(bios, reg) & ~mask) | val); +} + +static uint32_t +peek_fb(struct drm_device *dev, struct io_mapping *fb, + uint32_t off) +{ + uint32_t val = 0; + + if (off < pci_resource_len(dev->pdev, 1)) { + uint8_t __iomem *p = + io_mapping_map_atomic_wc(fb, off & PAGE_MASK); + + val = ioread32(p + (off & ~PAGE_MASK)); + + io_mapping_unmap_atomic(p); + } + + return val; +} + +static void +poke_fb(struct drm_device *dev, struct io_mapping *fb, + uint32_t off, uint32_t val) +{ + if (off < pci_resource_len(dev->pdev, 1)) { + uint8_t __iomem *p = + io_mapping_map_atomic_wc(fb, off & PAGE_MASK); + + iowrite32(val, p + (off & ~PAGE_MASK)); + wmb(); + + io_mapping_unmap_atomic(p); + } +} + +static inline bool +read_back_fb(struct drm_device *dev, struct io_mapping *fb, + uint32_t off, uint32_t val) +{ + poke_fb(dev, fb, off, val); + return val == peek_fb(dev, fb, off); +} + +static int +nv04_init_compute_mem(struct nvbios *bios) +{ + struct drm_device *dev = bios->dev; + uint32_t patt = 0xdeadbeef; + struct io_mapping *fb; + int i; + + /* Map the framebuffer aperture */ + fb = io_mapping_create_wc(pci_resource_start(dev->pdev, 1), + pci_resource_len(dev->pdev, 1)); + if (!fb) + return -ENOMEM; + + /* Sequencer and refresh off */ + NVWriteVgaSeq(dev, 0, 1, NVReadVgaSeq(dev, 0, 1) | 0x20); + bios_md32(bios, NV04_PFB_DEBUG_0, 0, NV04_PFB_DEBUG_0_REFRESH_OFF); + + bios_md32(bios, NV04_PFB_BOOT_0, ~0, + NV04_PFB_BOOT_0_RAM_AMOUNT_16MB | + NV04_PFB_BOOT_0_RAM_WIDTH_128 | + NV04_PFB_BOOT_0_RAM_TYPE_SGRAM_16MBIT); + + for (i = 0; i < 4; i++) + poke_fb(dev, fb, 4 * i, patt); + + poke_fb(dev, fb, 0x400000, patt + 1); + + if (peek_fb(dev, fb, 0) == patt + 1) { + bios_md32(bios, NV04_PFB_BOOT_0, NV04_PFB_BOOT_0_RAM_TYPE, + NV04_PFB_BOOT_0_RAM_TYPE_SDRAM_16MBIT); + bios_md32(bios, NV04_PFB_DEBUG_0, + NV04_PFB_DEBUG_0_REFRESH_OFF, 0); + + for (i = 0; i < 4; i++) + poke_fb(dev, fb, 4 * i, patt); + + if ((peek_fb(dev, fb, 0xc) & 0xffff) != (patt & 0xffff)) + bios_md32(bios, NV04_PFB_BOOT_0, + NV04_PFB_BOOT_0_RAM_WIDTH_128 | + NV04_PFB_BOOT_0_RAM_AMOUNT, + NV04_PFB_BOOT_0_RAM_AMOUNT_8MB); + + } else if ((peek_fb(dev, fb, 0xc) & 0xffff0000) != + (patt & 0xffff0000)) { + bios_md32(bios, NV04_PFB_BOOT_0, + NV04_PFB_BOOT_0_RAM_WIDTH_128 | + NV04_PFB_BOOT_0_RAM_AMOUNT, + NV04_PFB_BOOT_0_RAM_AMOUNT_4MB); + + } else if (peek_fb(dev, fb, 0) != patt) { + if (read_back_fb(dev, fb, 0x800000, patt)) + bios_md32(bios, NV04_PFB_BOOT_0, + NV04_PFB_BOOT_0_RAM_AMOUNT, + NV04_PFB_BOOT_0_RAM_AMOUNT_8MB); + else + bios_md32(bios, NV04_PFB_BOOT_0, + NV04_PFB_BOOT_0_RAM_AMOUNT, + NV04_PFB_BOOT_0_RAM_AMOUNT_4MB); + + bios_md32(bios, NV04_PFB_BOOT_0, NV04_PFB_BOOT_0_RAM_TYPE, + NV04_PFB_BOOT_0_RAM_TYPE_SGRAM_8MBIT); + + } else if (!read_back_fb(dev, fb, 0x800000, patt)) { + bios_md32(bios, NV04_PFB_BOOT_0, NV04_PFB_BOOT_0_RAM_AMOUNT, + NV04_PFB_BOOT_0_RAM_AMOUNT_8MB); + + } + + /* Refresh on, sequencer on */ + bios_md32(bios, NV04_PFB_DEBUG_0, NV04_PFB_DEBUG_0_REFRESH_OFF, 0); + NVWriteVgaSeq(dev, 0, 1, NVReadVgaSeq(dev, 0, 1) & ~0x20); + + io_mapping_free(fb); + return 0; +} + +static const uint8_t * +nv05_memory_config(struct nvbios *bios) +{ + /* Defaults for BIOSes lacking a memory config table */ + static const uint8_t default_config_tab[][2] = { + { 0x24, 0x00 }, + { 0x28, 0x00 }, + { 0x24, 0x01 }, + { 0x1f, 0x00 }, + { 0x0f, 0x00 }, + { 0x17, 0x00 }, + { 0x06, 0x00 }, + { 0x00, 0x00 } + }; + int i = (bios_rd32(bios, NV_PEXTDEV_BOOT_0) & + NV_PEXTDEV_BOOT_0_RAMCFG) >> 2; + + if (bios->legacy.mem_init_tbl_ptr) + return &bios->data[bios->legacy.mem_init_tbl_ptr + 2 * i]; + else + return default_config_tab[i]; +} + +static int +nv05_init_compute_mem(struct nvbios *bios) +{ + struct drm_device *dev = bios->dev; + const uint8_t *ramcfg = nv05_memory_config(bios); + uint32_t patt = 0xdeadbeef; + struct io_mapping *fb; + int i, v; + + /* Map the framebuffer aperture */ + fb = io_mapping_create_wc(pci_resource_start(dev->pdev, 1), + pci_resource_len(dev->pdev, 1)); + if (!fb) + return -ENOMEM; + + /* Sequencer off */ + NVWriteVgaSeq(dev, 0, 1, NVReadVgaSeq(dev, 0, 1) | 0x20); + + if (bios_rd32(bios, NV04_PFB_BOOT_0) & NV04_PFB_BOOT_0_UMA_ENABLE) + goto out; + + bios_md32(bios, NV04_PFB_DEBUG_0, NV04_PFB_DEBUG_0_REFRESH_OFF, 0); + + /* If present load the hardcoded scrambling table */ + if (bios->legacy.mem_init_tbl_ptr) { + uint32_t *scramble_tab = (uint32_t *)&bios->data[ + bios->legacy.mem_init_tbl_ptr + 0x10]; + + for (i = 0; i < 8; i++) + bios_wr32(bios, NV04_PFB_SCRAMBLE(i), + ROM32(scramble_tab[i])); + } + + /* Set memory type/width/length defaults depending on the straps */ + bios_md32(bios, NV04_PFB_BOOT_0, 0x3f, ramcfg[0]); + + if (ramcfg[1] & 0x80) + bios_md32(bios, NV04_PFB_CFG0, 0, NV04_PFB_CFG0_SCRAMBLE); + + bios_md32(bios, NV04_PFB_CFG1, 0x700001, (ramcfg[1] & 1) << 20); + bios_md32(bios, NV04_PFB_CFG1, 0, 1); + + /* Probe memory bus width */ + for (i = 0; i < 4; i++) + poke_fb(dev, fb, 4 * i, patt); + + if (peek_fb(dev, fb, 0xc) != patt) + bios_md32(bios, NV04_PFB_BOOT_0, + NV04_PFB_BOOT_0_RAM_WIDTH_128, 0); + + /* Probe memory length */ + v = bios_rd32(bios, NV04_PFB_BOOT_0) & NV04_PFB_BOOT_0_RAM_AMOUNT; + + if (v == NV04_PFB_BOOT_0_RAM_AMOUNT_32MB && + (!read_back_fb(dev, fb, 0x1000000, ++patt) || + !read_back_fb(dev, fb, 0, ++patt))) + bios_md32(bios, NV04_PFB_BOOT_0, NV04_PFB_BOOT_0_RAM_AMOUNT, + NV04_PFB_BOOT_0_RAM_AMOUNT_16MB); + + if (v == NV04_PFB_BOOT_0_RAM_AMOUNT_16MB && + !read_back_fb(dev, fb, 0x800000, ++patt)) + bios_md32(bios, NV04_PFB_BOOT_0, NV04_PFB_BOOT_0_RAM_AMOUNT, + NV04_PFB_BOOT_0_RAM_AMOUNT_8MB); + + if (!read_back_fb(dev, fb, 0x400000, ++patt)) + bios_md32(bios, NV04_PFB_BOOT_0, NV04_PFB_BOOT_0_RAM_AMOUNT, + NV04_PFB_BOOT_0_RAM_AMOUNT_4MB); + +out: + /* Sequencer on */ + NVWriteVgaSeq(dev, 0, 1, NVReadVgaSeq(dev, 0, 1) & ~0x20); + + io_mapping_free(fb); + return 0; +} + +static int +nv10_init_compute_mem(struct nvbios *bios) +{ + struct drm_device *dev = bios->dev; + struct drm_nouveau_private *dev_priv = bios->dev->dev_private; + const int mem_width[] = { 0x10, 0x00, 0x20 }; + const int mem_width_count = (dev_priv->chipset >= 0x17 ? 3 : 2); + uint32_t patt = 0xdeadbeef; + struct io_mapping *fb; + int i, j, k; + + /* Map the framebuffer aperture */ + fb = io_mapping_create_wc(pci_resource_start(dev->pdev, 1), + pci_resource_len(dev->pdev, 1)); + if (!fb) + return -ENOMEM; + + bios_wr32(bios, NV10_PFB_REFCTRL, NV10_PFB_REFCTRL_VALID_1); + + /* Probe memory bus width */ + for (i = 0; i < mem_width_count; i++) { + bios_md32(bios, NV04_PFB_CFG0, 0x30, mem_width[i]); + + for (j = 0; j < 4; j++) { + for (k = 0; k < 4; k++) + poke_fb(dev, fb, 0x1c, 0); + + poke_fb(dev, fb, 0x1c, patt); + poke_fb(dev, fb, 0x3c, 0); + + if (peek_fb(dev, fb, 0x1c) == patt) + goto mem_width_found; + } + } + +mem_width_found: + patt <<= 1; + + /* Probe amount of installed memory */ + for (i = 0; i < 4; i++) { + int off = bios_rd32(bios, NV04_PFB_FIFO_DATA) - 0x100000; + + poke_fb(dev, fb, off, patt); + poke_fb(dev, fb, 0, 0); + + peek_fb(dev, fb, 0); + peek_fb(dev, fb, 0); + peek_fb(dev, fb, 0); + peek_fb(dev, fb, 0); + + if (peek_fb(dev, fb, off) == patt) + goto amount_found; + } + + /* IC missing - disable the upper half memory space. */ + bios_md32(bios, NV04_PFB_CFG0, 0x1000, 0); + +amount_found: + io_mapping_free(fb); + return 0; +} + +static int +nv20_init_compute_mem(struct nvbios *bios) +{ + struct drm_device *dev = bios->dev; + struct drm_nouveau_private *dev_priv = bios->dev->dev_private; + uint32_t mask = (dev_priv->chipset >= 0x25 ? 0x300 : 0x900); + uint32_t amount, off; + struct io_mapping *fb; + + /* Map the framebuffer aperture */ + fb = io_mapping_create_wc(pci_resource_start(dev->pdev, 1), + pci_resource_len(dev->pdev, 1)); + if (!fb) + return -ENOMEM; + + bios_wr32(bios, NV10_PFB_REFCTRL, NV10_PFB_REFCTRL_VALID_1); + + /* Allow full addressing */ + bios_md32(bios, NV04_PFB_CFG0, 0, mask); + + amount = bios_rd32(bios, NV04_PFB_FIFO_DATA); + for (off = amount; off > 0x2000000; off -= 0x2000000) + poke_fb(dev, fb, off - 4, off); + + amount = bios_rd32(bios, NV04_PFB_FIFO_DATA); + if (amount != peek_fb(dev, fb, amount - 4)) + /* IC missing - disable the upper half memory space. */ + bios_md32(bios, NV04_PFB_CFG0, mask, 0); + + io_mapping_free(fb); + return 0; +} + +static int +init_compute_mem(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_COMPUTE_MEM opcode: 0x63 ('c') + * + * offset (8 bit): opcode + * + * This opcode is meant to set the PFB memory config registers + * appropriately so that we can correctly calculate how much VRAM it + * has (on nv10 and better chipsets the amount of installed VRAM is + * subsequently reported in NV_PFB_CSTATUS (0x10020C)). + * + * The implementation of this opcode in general consists of several + * parts: + * + * 1) Determination of memory type and density. Only necessary for + * really old chipsets, the memory type reported by the strap bits + * (0x101000) is assumed to be accurate on nv05 and newer. + * + * 2) Determination of the memory bus width. Usually done by a cunning + * combination of writes to offsets 0x1c and 0x3c in the fb, and + * seeing whether the written values are read back correctly. + * + * Only necessary on nv0x-nv1x and nv34, on the other cards we can + * trust the straps. + * + * 3) Determination of how many of the card's RAM pads have ICs + * attached, usually done by a cunning combination of writes to an + * offset slightly less than the maximum memory reported by + * NV_PFB_CSTATUS, then seeing if the test pattern can be read back. + * + * This appears to be a NOP on IGPs and NV4x or newer chipsets, both io + * logs of the VBIOS and kmmio traces of the binary driver POSTing the + * card show nothing being done for this opcode. Why is it still listed + * in the table?! + */ + + /* no iexec->execute check by design */ + + struct drm_nouveau_private *dev_priv = bios->dev->dev_private; + int ret; + + if (dev_priv->chipset >= 0x40 || + dev_priv->chipset == 0x1a || + dev_priv->chipset == 0x1f) + ret = 0; + else if (dev_priv->chipset >= 0x20 && + dev_priv->chipset != 0x34) + ret = nv20_init_compute_mem(bios); + else if (dev_priv->chipset >= 0x10) + ret = nv10_init_compute_mem(bios); + else if (dev_priv->chipset >= 0x5) + ret = nv05_init_compute_mem(bios); + else + ret = nv04_init_compute_mem(bios); + + if (ret) + return ret; + + return 1; +} + +static int +init_reset(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_RESET opcode: 0x65 ('e') + * + * offset (8 bit): opcode + * offset + 1 (32 bit): register + * offset + 5 (32 bit): value1 + * offset + 9 (32 bit): value2 + * + * Assign "value1" to "register", then assign "value2" to "register" + */ + + uint32_t reg = ROM32(bios->data[offset + 1]); + uint32_t value1 = ROM32(bios->data[offset + 5]); + uint32_t value2 = ROM32(bios->data[offset + 9]); + uint32_t pci_nv_19, pci_nv_20; + + /* no iexec->execute check by design */ + + pci_nv_19 = bios_rd32(bios, NV_PBUS_PCI_NV_19); + bios_wr32(bios, NV_PBUS_PCI_NV_19, pci_nv_19 & ~0xf00); + + bios_wr32(bios, reg, value1); + + udelay(10); + + bios_wr32(bios, reg, value2); + bios_wr32(bios, NV_PBUS_PCI_NV_19, pci_nv_19); + + pci_nv_20 = bios_rd32(bios, NV_PBUS_PCI_NV_20); + pci_nv_20 &= ~NV_PBUS_PCI_NV_20_ROM_SHADOW_ENABLED; /* 0xfffffffe */ + bios_wr32(bios, NV_PBUS_PCI_NV_20, pci_nv_20); + + return 13; +} + +static int +init_configure_mem(struct nvbios *bios, uint16_t offset, + struct init_exec *iexec) +{ + /* + * INIT_CONFIGURE_MEM opcode: 0x66 ('f') + * + * offset (8 bit): opcode + * + * Equivalent to INIT_DONE on bios version 3 or greater. + * For early bios versions, sets up the memory registers, using values + * taken from the memory init table + */ + + /* no iexec->execute check by design */ + + uint16_t meminitoffs = bios->legacy.mem_init_tbl_ptr + MEM_INIT_SIZE * (bios_idxprt_rd(bios, NV_CIO_CRX__COLOR, NV_CIO_CRE_SCRATCH4__INDEX) >> 4); + uint16_t seqtbloffs = bios->legacy.sdr_seq_tbl_ptr, meminitdata = meminitoffs + 6; + uint32_t reg, data; + + if (bios->major_version > 2) + return 0; + + bios_idxprt_wr(bios, NV_VIO_SRX, NV_VIO_SR_CLOCK_INDEX, bios_idxprt_rd( + bios, NV_VIO_SRX, NV_VIO_SR_CLOCK_INDEX) | 0x20); + + if (bios->data[meminitoffs] & 1) + seqtbloffs = bios->legacy.ddr_seq_tbl_ptr; + + for (reg = ROM32(bios->data[seqtbloffs]); + reg != 0xffffffff; + reg = ROM32(bios->data[seqtbloffs += 4])) { + + switch (reg) { + case NV04_PFB_PRE: + data = NV04_PFB_PRE_CMD_PRECHARGE; + break; + case NV04_PFB_PAD: + data = NV04_PFB_PAD_CKE_NORMAL; + break; + case NV04_PFB_REF: + data = NV04_PFB_REF_CMD_REFRESH; + break; + default: + data = ROM32(bios->data[meminitdata]); + meminitdata += 4; + if (data == 0xffffffff) + continue; + } + + bios_wr32(bios, reg, data); + } + + return 1; +} + +static int +init_configure_clk(struct nvbios *bios, uint16_t offset, + struct init_exec *iexec) +{ + /* + * INIT_CONFIGURE_CLK opcode: 0x67 ('g') + * + * offset (8 bit): opcode + * + * Equivalent to INIT_DONE on bios version 3 or greater. + * For early bios versions, sets up the NVClk and MClk PLLs, using + * values taken from the memory init table + */ + + /* no iexec->execute check by design */ + + uint16_t meminitoffs = bios->legacy.mem_init_tbl_ptr + MEM_INIT_SIZE * (bios_idxprt_rd(bios, NV_CIO_CRX__COLOR, NV_CIO_CRE_SCRATCH4__INDEX) >> 4); + int clock; + + if (bios->major_version > 2) + return 0; + + clock = ROM16(bios->data[meminitoffs + 4]) * 10; + setPLL(bios, NV_PRAMDAC_NVPLL_COEFF, clock); + + clock = ROM16(bios->data[meminitoffs + 2]) * 10; + if (bios->data[meminitoffs] & 1) /* DDR */ + clock *= 2; + setPLL(bios, NV_PRAMDAC_MPLL_COEFF, clock); + + return 1; +} + +static int +init_configure_preinit(struct nvbios *bios, uint16_t offset, + struct init_exec *iexec) +{ + /* + * INIT_CONFIGURE_PREINIT opcode: 0x68 ('h') + * + * offset (8 bit): opcode + * + * Equivalent to INIT_DONE on bios version 3 or greater. + * For early bios versions, does early init, loading ram and crystal + * configuration from straps into CR3C + */ + + /* no iexec->execute check by design */ + + uint32_t straps = bios_rd32(bios, NV_PEXTDEV_BOOT_0); + uint8_t cr3c = ((straps << 2) & 0xf0) | (straps & 0x40) >> 6; + + if (bios->major_version > 2) + return 0; + + bios_idxprt_wr(bios, NV_CIO_CRX__COLOR, + NV_CIO_CRE_SCRATCH4__INDEX, cr3c); + + return 1; +} + +static int +init_io(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_IO opcode: 0x69 ('i') + * + * offset (8 bit): opcode + * offset + 1 (16 bit): CRTC port + * offset + 3 (8 bit): mask + * offset + 4 (8 bit): data + * + * Assign ((IOVAL("crtc port") & "mask") | "data") to "crtc port" + */ + + struct drm_nouveau_private *dev_priv = bios->dev->dev_private; + uint16_t crtcport = ROM16(bios->data[offset + 1]); + uint8_t mask = bios->data[offset + 3]; + uint8_t data = bios->data[offset + 4]; + + if (!iexec->execute) + return 5; + + BIOSLOG(bios, "0x%04X: Port: 0x%04X, Mask: 0x%02X, Data: 0x%02X\n", + offset, crtcport, mask, data); + + /* + * I have no idea what this does, but NVIDIA do this magic sequence + * in the places where this INIT_IO happens.. + */ + if (dev_priv->card_type >= NV_50 && crtcport == 0x3c3 && data == 1) { + int i; + + bios_wr32(bios, 0x614100, (bios_rd32( + bios, 0x614100) & 0x0fffffff) | 0x00800000); + + bios_wr32(bios, 0x00e18c, bios_rd32( + bios, 0x00e18c) | 0x00020000); + + bios_wr32(bios, 0x614900, (bios_rd32( + bios, 0x614900) & 0x0fffffff) | 0x00800000); + + bios_wr32(bios, 0x000200, bios_rd32( + bios, 0x000200) & ~0x40000000); + + mdelay(10); + + bios_wr32(bios, 0x00e18c, bios_rd32( + bios, 0x00e18c) & ~0x00020000); + + bios_wr32(bios, 0x000200, bios_rd32( + bios, 0x000200) | 0x40000000); + + bios_wr32(bios, 0x614100, 0x00800018); + bios_wr32(bios, 0x614900, 0x00800018); + + mdelay(10); + + bios_wr32(bios, 0x614100, 0x10000018); + bios_wr32(bios, 0x614900, 0x10000018); + + for (i = 0; i < 3; i++) + bios_wr32(bios, 0x614280 + (i*0x800), bios_rd32( + bios, 0x614280 + (i*0x800)) & 0xf0f0f0f0); + + for (i = 0; i < 2; i++) + bios_wr32(bios, 0x614300 + (i*0x800), bios_rd32( + bios, 0x614300 + (i*0x800)) & 0xfffff0f0); + + for (i = 0; i < 3; i++) + bios_wr32(bios, 0x614380 + (i*0x800), bios_rd32( + bios, 0x614380 + (i*0x800)) & 0xfffff0f0); + + for (i = 0; i < 2; i++) + bios_wr32(bios, 0x614200 + (i*0x800), bios_rd32( + bios, 0x614200 + (i*0x800)) & 0xfffffff0); + + for (i = 0; i < 2; i++) + bios_wr32(bios, 0x614108 + (i*0x800), bios_rd32( + bios, 0x614108 + (i*0x800)) & 0x0fffffff); + return 5; + } + + bios_port_wr(bios, crtcport, (bios_port_rd(bios, crtcport) & mask) | + data); + return 5; +} + +static int +init_sub(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_SUB opcode: 0x6B ('k') + * + * offset (8 bit): opcode + * offset + 1 (8 bit): script number + * + * Execute script number "script number", as a subroutine + */ + + uint8_t sub = bios->data[offset + 1]; + + if (!iexec->execute) + return 2; + + BIOSLOG(bios, "0x%04X: Calling script %d\n", offset, sub); + + parse_init_table(bios, + ROM16(bios->data[bios->init_script_tbls_ptr + sub * 2]), + iexec); + + BIOSLOG(bios, "0x%04X: End of script %d\n", offset, sub); + + return 2; +} + +static int +init_ram_condition(struct nvbios *bios, uint16_t offset, + struct init_exec *iexec) +{ + /* + * INIT_RAM_CONDITION opcode: 0x6D ('m') + * + * offset (8 bit): opcode + * offset + 1 (8 bit): mask + * offset + 2 (8 bit): cmpval + * + * Test if (NV04_PFB_BOOT_0 & "mask") equals "cmpval". + * If condition not met skip subsequent opcodes until condition is + * inverted (INIT_NOT), or we hit INIT_RESUME + */ + + uint8_t mask = bios->data[offset + 1]; + uint8_t cmpval = bios->data[offset + 2]; + uint8_t data; + + if (!iexec->execute) + return 3; + + data = bios_rd32(bios, NV04_PFB_BOOT_0) & mask; + + BIOSLOG(bios, "0x%04X: Checking if 0x%08X equals 0x%08X\n", + offset, data, cmpval); + + if (data == cmpval) + BIOSLOG(bios, "0x%04X: Condition fulfilled -- continuing to execute\n", offset); + else { + BIOSLOG(bios, "0x%04X: Condition not fulfilled -- skipping following commands\n", offset); + iexec->execute = false; + } + + return 3; +} + +static int +init_nv_reg(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_NV_REG opcode: 0x6E ('n') + * + * offset (8 bit): opcode + * offset + 1 (32 bit): register + * offset + 5 (32 bit): mask + * offset + 9 (32 bit): data + * + * Assign ((REGVAL("register") & "mask") | "data") to "register" + */ + + uint32_t reg = ROM32(bios->data[offset + 1]); + uint32_t mask = ROM32(bios->data[offset + 5]); + uint32_t data = ROM32(bios->data[offset + 9]); + + if (!iexec->execute) + return 13; + + BIOSLOG(bios, "0x%04X: Reg: 0x%08X, Mask: 0x%08X, Data: 0x%08X\n", + offset, reg, mask, data); + + bios_wr32(bios, reg, (bios_rd32(bios, reg) & mask) | data); + + return 13; +} + +static int +init_macro(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_MACRO opcode: 0x6F ('o') + * + * offset (8 bit): opcode + * offset + 1 (8 bit): macro number + * + * Look up macro index "macro number" in the macro index table. + * The macro index table entry has 1 byte for the index in the macro + * table, and 1 byte for the number of times to repeat the macro. + * The macro table entry has 4 bytes for the register address and + * 4 bytes for the value to write to that register + */ + + uint8_t macro_index_tbl_idx = bios->data[offset + 1]; + uint16_t tmp = bios->macro_index_tbl_ptr + (macro_index_tbl_idx * MACRO_INDEX_SIZE); + uint8_t macro_tbl_idx = bios->data[tmp]; + uint8_t count = bios->data[tmp + 1]; + uint32_t reg, data; + int i; + + if (!iexec->execute) + return 2; + + BIOSLOG(bios, "0x%04X: Macro: 0x%02X, MacroTableIndex: 0x%02X, " + "Count: 0x%02X\n", + offset, macro_index_tbl_idx, macro_tbl_idx, count); + + for (i = 0; i < count; i++) { + uint16_t macroentryptr = bios->macro_tbl_ptr + (macro_tbl_idx + i) * MACRO_SIZE; + + reg = ROM32(bios->data[macroentryptr]); + data = ROM32(bios->data[macroentryptr + 4]); + + bios_wr32(bios, reg, data); + } + + return 2; +} + +static int +init_done(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_DONE opcode: 0x71 ('q') + * + * offset (8 bit): opcode + * + * End the current script + */ + + /* mild retval abuse to stop parsing this table */ + return 0; +} + +static int +init_resume(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_RESUME opcode: 0x72 ('r') + * + * offset (8 bit): opcode + * + * End the current execute / no-execute condition + */ + + if (iexec->execute) + return 1; + + iexec->execute = true; + BIOSLOG(bios, "0x%04X: ---- Executing following commands ----\n", offset); + + return 1; +} + +static int +init_time(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_TIME opcode: 0x74 ('t') + * + * offset (8 bit): opcode + * offset + 1 (16 bit): time + * + * Sleep for "time" microseconds. + */ + + unsigned time = ROM16(bios->data[offset + 1]); + + if (!iexec->execute) + return 3; + + BIOSLOG(bios, "0x%04X: Sleeping for 0x%04X microseconds\n", + offset, time); + + if (time < 1000) + udelay(time); + else + mdelay((time + 900) / 1000); + + return 3; +} + +static int +init_condition(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_CONDITION opcode: 0x75 ('u') + * + * offset (8 bit): opcode + * offset + 1 (8 bit): condition number + * + * Check condition "condition number" in the condition table. + * If condition not met skip subsequent opcodes until condition is + * inverted (INIT_NOT), or we hit INIT_RESUME + */ + + uint8_t cond = bios->data[offset + 1]; + + if (!iexec->execute) + return 2; + + BIOSLOG(bios, "0x%04X: Condition: 0x%02X\n", offset, cond); + + if (bios_condition_met(bios, offset, cond)) + BIOSLOG(bios, "0x%04X: Condition fulfilled -- continuing to execute\n", offset); + else { + BIOSLOG(bios, "0x%04X: Condition not fulfilled -- skipping following commands\n", offset); + iexec->execute = false; + } + + return 2; +} + +static int +init_io_condition(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_IO_CONDITION opcode: 0x76 + * + * offset (8 bit): opcode + * offset + 1 (8 bit): condition number + * + * Check condition "condition number" in the io condition table. + * If condition not met skip subsequent opcodes until condition is + * inverted (INIT_NOT), or we hit INIT_RESUME + */ + + uint8_t cond = bios->data[offset + 1]; + + if (!iexec->execute) + return 2; + + BIOSLOG(bios, "0x%04X: IO condition: 0x%02X\n", offset, cond); + + if (io_condition_met(bios, offset, cond)) + BIOSLOG(bios, "0x%04X: Condition fulfilled -- continuing to execute\n", offset); + else { + BIOSLOG(bios, "0x%04X: Condition not fulfilled -- skipping following commands\n", offset); + iexec->execute = false; + } + + return 2; +} + +static int +init_index_io(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_INDEX_IO opcode: 0x78 ('x') + * + * offset (8 bit): opcode + * offset + 1 (16 bit): CRTC port + * offset + 3 (8 bit): CRTC index + * offset + 4 (8 bit): mask + * offset + 5 (8 bit): data + * + * Read value at index "CRTC index" on "CRTC port", AND with "mask", + * OR with "data", write-back + */ + + uint16_t crtcport = ROM16(bios->data[offset + 1]); + uint8_t crtcindex = bios->data[offset + 3]; + uint8_t mask = bios->data[offset + 4]; + uint8_t data = bios->data[offset + 5]; + uint8_t value; + + if (!iexec->execute) + return 6; + + BIOSLOG(bios, "0x%04X: Port: 0x%04X, Index: 0x%02X, Mask: 0x%02X, " + "Data: 0x%02X\n", + offset, crtcport, crtcindex, mask, data); + + value = (bios_idxprt_rd(bios, crtcport, crtcindex) & mask) | data; + bios_idxprt_wr(bios, crtcport, crtcindex, value); + + return 6; +} + +static int +init_pll(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_PLL opcode: 0x79 ('y') + * + * offset (8 bit): opcode + * offset + 1 (32 bit): register + * offset + 5 (16 bit): freq + * + * Set PLL register "register" to coefficients for frequency (10kHz) + * "freq" + */ + + uint32_t reg = ROM32(bios->data[offset + 1]); + uint16_t freq = ROM16(bios->data[offset + 5]); + + if (!iexec->execute) + return 7; + + BIOSLOG(bios, "0x%04X: Reg: 0x%08X, Freq: %d0kHz\n", offset, reg, freq); + + setPLL(bios, reg, freq * 10); + + return 7; +} + +static int +init_zm_reg(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_ZM_REG opcode: 0x7A ('z') + * + * offset (8 bit): opcode + * offset + 1 (32 bit): register + * offset + 5 (32 bit): value + * + * Assign "value" to "register" + */ + + uint32_t reg = ROM32(bios->data[offset + 1]); + uint32_t value = ROM32(bios->data[offset + 5]); + + if (!iexec->execute) + return 9; + + if (reg == 0x000200) + value |= 1; + + bios_wr32(bios, reg, value); + + return 9; +} + +static int +init_ram_restrict_pll(struct nvbios *bios, uint16_t offset, + struct init_exec *iexec) +{ + /* + * INIT_RAM_RESTRICT_PLL opcode: 0x87 ('') + * + * offset (8 bit): opcode + * offset + 1 (8 bit): PLL type + * offset + 2 (32 bit): frequency 0 + * + * Uses the RAMCFG strap of PEXTDEV_BOOT as an index into the table at + * ram_restrict_table_ptr. The value read from there is used to select + * a frequency from the table starting at 'frequency 0' to be + * programmed into the PLL corresponding to 'type'. + * + * The PLL limits table on cards using this opcode has a mapping of + * 'type' to the relevant registers. + */ + + struct drm_device *dev = bios->dev; + uint32_t strap = (bios_rd32(bios, NV_PEXTDEV_BOOT_0) & 0x0000003c) >> 2; + uint8_t index = bios->data[bios->ram_restrict_tbl_ptr + strap]; + uint8_t type = bios->data[offset + 1]; + uint32_t freq = ROM32(bios->data[offset + 2 + (index * 4)]); + uint8_t *pll_limits = &bios->data[bios->pll_limit_tbl_ptr], *entry; + int len = 2 + bios->ram_restrict_group_count * 4; + int i; + + if (!iexec->execute) + return len; + + if (!bios->pll_limit_tbl_ptr || (pll_limits[0] & 0xf0) != 0x30) { + NV_ERROR(dev, "PLL limits table not version 3.x\n"); + return len; /* deliberate, allow default clocks to remain */ + } + + entry = pll_limits + pll_limits[1]; + for (i = 0; i < pll_limits[3]; i++, entry += pll_limits[2]) { + if (entry[0] == type) { + uint32_t reg = ROM32(entry[3]); + + BIOSLOG(bios, "0x%04X: " + "Type %02x Reg 0x%08x Freq %dKHz\n", + offset, type, reg, freq); + + setPLL(bios, reg, freq); + return len; + } + } + + NV_ERROR(dev, "PLL type 0x%02x not found in PLL limits table", type); + return len; +} + +static int +init_8c(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_8C opcode: 0x8C ('') + * + * NOP so far.... + * + */ + + return 1; +} + +static int +init_8d(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_8D opcode: 0x8D ('') + * + * NOP so far.... + * + */ + + return 1; +} + +static int +init_gpio(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_GPIO opcode: 0x8E ('') + * + * offset (8 bit): opcode + * + * Loop over all entries in the DCB GPIO table, and initialise + * each GPIO according to various values listed in each entry + */ + + struct drm_nouveau_private *dev_priv = bios->dev->dev_private; + struct nouveau_gpio_engine *pgpio = &dev_priv->engine.gpio; + const uint32_t nv50_gpio_ctl[2] = { 0xe100, 0xe28c }; + int i; + + if (dev_priv->card_type < NV_50) { + NV_ERROR(bios->dev, "INIT_GPIO on unsupported chipset\n"); + return 1; + } + + if (!iexec->execute) + return 1; + + for (i = 0; i < bios->dcb.gpio.entries; i++) { + struct dcb_gpio_entry *gpio = &bios->dcb.gpio.entry[i]; + uint32_t r, s, v; + + BIOSLOG(bios, "0x%04X: Entry: 0x%08X\n", offset, gpio->entry); + + BIOSLOG(bios, "0x%04X: set gpio 0x%02x, state %d\n", + offset, gpio->tag, gpio->state_default); + if (bios->execute) + pgpio->set(bios->dev, gpio->tag, gpio->state_default); + + /* The NVIDIA binary driver doesn't appear to actually do + * any of this, my VBIOS does however. + */ + /* Not a clue, needs de-magicing */ + r = nv50_gpio_ctl[gpio->line >> 4]; + s = (gpio->line & 0x0f); + v = bios_rd32(bios, r) & ~(0x00010001 << s); + switch ((gpio->entry & 0x06000000) >> 25) { + case 1: + v |= (0x00000001 << s); + break; + case 2: + v |= (0x00010000 << s); + break; + default: + break; + } + bios_wr32(bios, r, v); + } + + return 1; +} + +static int +init_ram_restrict_zm_reg_group(struct nvbios *bios, uint16_t offset, + struct init_exec *iexec) +{ + /* + * INIT_RAM_RESTRICT_ZM_REG_GROUP opcode: 0x8F ('') + * + * offset (8 bit): opcode + * offset + 1 (32 bit): reg + * offset + 5 (8 bit): regincrement + * offset + 6 (8 bit): count + * offset + 7 (32 bit): value 1,1 + * ... + * + * Use the RAMCFG strap of PEXTDEV_BOOT as an index into the table at + * ram_restrict_table_ptr. The value read from here is 'n', and + * "value 1,n" gets written to "reg". This repeats "count" times and on + * each iteration 'm', "reg" increases by "regincrement" and + * "value m,n" is used. The extent of n is limited by a number read + * from the 'M' BIT table, herein called "blocklen" + */ + + uint32_t reg = ROM32(bios->data[offset + 1]); + uint8_t regincrement = bios->data[offset + 5]; + uint8_t count = bios->data[offset + 6]; + uint32_t strap_ramcfg, data; + /* previously set by 'M' BIT table */ + uint16_t blocklen = bios->ram_restrict_group_count * 4; + int len = 7 + count * blocklen; + uint8_t index; + int i; + + /* critical! to know the length of the opcode */; + if (!blocklen) { + NV_ERROR(bios->dev, + "0x%04X: Zero block length - has the M table " + "been parsed?\n", offset); + return -EINVAL; + } + + if (!iexec->execute) + return len; + + strap_ramcfg = (bios_rd32(bios, NV_PEXTDEV_BOOT_0) >> 2) & 0xf; + index = bios->data[bios->ram_restrict_tbl_ptr + strap_ramcfg]; + + BIOSLOG(bios, "0x%04X: Reg: 0x%08X, RegIncrement: 0x%02X, " + "Count: 0x%02X, StrapRamCfg: 0x%02X, Index: 0x%02X\n", + offset, reg, regincrement, count, strap_ramcfg, index); + + for (i = 0; i < count; i++) { + data = ROM32(bios->data[offset + 7 + index * 4 + blocklen * i]); + + bios_wr32(bios, reg, data); + + reg += regincrement; + } + + return len; +} + +static int +init_copy_zm_reg(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_COPY_ZM_REG opcode: 0x90 ('') + * + * offset (8 bit): opcode + * offset + 1 (32 bit): src reg + * offset + 5 (32 bit): dst reg + * + * Put contents of "src reg" into "dst reg" + */ + + uint32_t srcreg = ROM32(bios->data[offset + 1]); + uint32_t dstreg = ROM32(bios->data[offset + 5]); + + if (!iexec->execute) + return 9; + + bios_wr32(bios, dstreg, bios_rd32(bios, srcreg)); + + return 9; +} + +static int +init_zm_reg_group_addr_latched(struct nvbios *bios, uint16_t offset, + struct init_exec *iexec) +{ + /* + * INIT_ZM_REG_GROUP_ADDRESS_LATCHED opcode: 0x91 ('') + * + * offset (8 bit): opcode + * offset + 1 (32 bit): dst reg + * offset + 5 (8 bit): count + * offset + 6 (32 bit): data 1 + * ... + * + * For each of "count" values write "data n" to "dst reg" + */ + + uint32_t reg = ROM32(bios->data[offset + 1]); + uint8_t count = bios->data[offset + 5]; + int len = 6 + count * 4; + int i; + + if (!iexec->execute) + return len; + + for (i = 0; i < count; i++) { + uint32_t data = ROM32(bios->data[offset + 6 + 4 * i]); + bios_wr32(bios, reg, data); + } + + return len; +} + +static int +init_reserved(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_RESERVED opcode: 0x92 ('') + * + * offset (8 bit): opcode + * + * Seemingly does nothing + */ + + return 1; +} + +static int +init_96(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_96 opcode: 0x96 ('') + * + * offset (8 bit): opcode + * offset + 1 (32 bit): sreg + * offset + 5 (8 bit): sshift + * offset + 6 (8 bit): smask + * offset + 7 (8 bit): index + * offset + 8 (32 bit): reg + * offset + 12 (32 bit): mask + * offset + 16 (8 bit): shift + * + */ + + uint16_t xlatptr = bios->init96_tbl_ptr + (bios->data[offset + 7] * 2); + uint32_t reg = ROM32(bios->data[offset + 8]); + uint32_t mask = ROM32(bios->data[offset + 12]); + uint32_t val; + + val = bios_rd32(bios, ROM32(bios->data[offset + 1])); + if (bios->data[offset + 5] < 0x80) + val >>= bios->data[offset + 5]; + else + val <<= (0x100 - bios->data[offset + 5]); + val &= bios->data[offset + 6]; + + val = bios->data[ROM16(bios->data[xlatptr]) + val]; + val <<= bios->data[offset + 16]; + + if (!iexec->execute) + return 17; + + bios_wr32(bios, reg, (bios_rd32(bios, reg) & mask) | val); + return 17; +} + +static int +init_97(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_97 opcode: 0x97 ('') + * + * offset (8 bit): opcode + * offset + 1 (32 bit): register + * offset + 5 (32 bit): mask + * offset + 9 (32 bit): value + * + * Adds "value" to "register" preserving the fields specified + * by "mask" + */ + + uint32_t reg = ROM32(bios->data[offset + 1]); + uint32_t mask = ROM32(bios->data[offset + 5]); + uint32_t add = ROM32(bios->data[offset + 9]); + uint32_t val; + + val = bios_rd32(bios, reg); + val = (val & mask) | ((val + add) & ~mask); + + if (!iexec->execute) + return 13; + + bios_wr32(bios, reg, val); + return 13; +} + +static int +init_auxch(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_AUXCH opcode: 0x98 ('') + * + * offset (8 bit): opcode + * offset + 1 (32 bit): address + * offset + 5 (8 bit): count + * offset + 6 (8 bit): mask 0 + * offset + 7 (8 bit): data 0 + * ... + * + */ + + struct drm_device *dev = bios->dev; + struct nouveau_i2c_chan *auxch; + uint32_t addr = ROM32(bios->data[offset + 1]); + uint8_t count = bios->data[offset + 5]; + int len = 6 + count * 2; + int ret, i; + + if (!bios->display.output) { + NV_ERROR(dev, "INIT_AUXCH: no active output\n"); + return len; + } + + auxch = init_i2c_device_find(dev, bios->display.output->i2c_index); + if (!auxch) { + NV_ERROR(dev, "INIT_AUXCH: couldn't get auxch %d\n", + bios->display.output->i2c_index); + return len; + } + + if (!iexec->execute) + return len; + + offset += 6; + for (i = 0; i < count; i++, offset += 2) { + uint8_t data; + + ret = nouveau_dp_auxch(auxch, 9, addr, &data, 1); + if (ret) { + NV_ERROR(dev, "INIT_AUXCH: rd auxch fail %d\n", ret); + return len; + } + + data &= bios->data[offset + 0]; + data |= bios->data[offset + 1]; + + ret = nouveau_dp_auxch(auxch, 8, addr, &data, 1); + if (ret) { + NV_ERROR(dev, "INIT_AUXCH: wr auxch fail %d\n", ret); + return len; + } + } + + return len; +} + +static int +init_zm_auxch(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_ZM_AUXCH opcode: 0x99 ('') + * + * offset (8 bit): opcode + * offset + 1 (32 bit): address + * offset + 5 (8 bit): count + * offset + 6 (8 bit): data 0 + * ... + * + */ + + struct drm_device *dev = bios->dev; + struct nouveau_i2c_chan *auxch; + uint32_t addr = ROM32(bios->data[offset + 1]); + uint8_t count = bios->data[offset + 5]; + int len = 6 + count; + int ret, i; + + if (!bios->display.output) { + NV_ERROR(dev, "INIT_ZM_AUXCH: no active output\n"); + return len; + } + + auxch = init_i2c_device_find(dev, bios->display.output->i2c_index); + if (!auxch) { + NV_ERROR(dev, "INIT_ZM_AUXCH: couldn't get auxch %d\n", + bios->display.output->i2c_index); + return len; + } + + if (!iexec->execute) + return len; + + offset += 6; + for (i = 0; i < count; i++, offset++) { + ret = nouveau_dp_auxch(auxch, 8, addr, &bios->data[offset], 1); + if (ret) { + NV_ERROR(dev, "INIT_ZM_AUXCH: wr auxch fail %d\n", ret); + return len; + } + } + + return len; +} + +static int +init_i2c_long_if(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * INIT_I2C_LONG_IF opcode: 0x9A ('') + * + * offset (8 bit): opcode + * offset + 1 (8 bit): DCB I2C table entry index + * offset + 2 (8 bit): I2C slave address + * offset + 3 (16 bit): I2C register + * offset + 5 (8 bit): mask + * offset + 6 (8 bit): data + * + * Read the register given by "I2C register" on the device addressed + * by "I2C slave address" on the I2C bus given by "DCB I2C table + * entry index". Compare the result AND "mask" to "data". + * If they're not equal, skip subsequent opcodes until condition is + * inverted (INIT_NOT), or we hit INIT_RESUME + */ + + uint8_t i2c_index = bios->data[offset + 1]; + uint8_t i2c_address = bios->data[offset + 2] >> 1; + uint8_t reglo = bios->data[offset + 3]; + uint8_t reghi = bios->data[offset + 4]; + uint8_t mask = bios->data[offset + 5]; + uint8_t data = bios->data[offset + 6]; + struct nouveau_i2c_chan *chan; + uint8_t buf0[2] = { reghi, reglo }; + uint8_t buf1[1]; + struct i2c_msg msg[2] = { + { i2c_address, 0, 1, buf0 }, + { i2c_address, I2C_M_RD, 1, buf1 }, + }; + int ret; + + /* no execute check by design */ + + BIOSLOG(bios, "0x%04X: DCBI2CIndex: 0x%02X, I2CAddress: 0x%02X\n", + offset, i2c_index, i2c_address); + + chan = init_i2c_device_find(bios->dev, i2c_index); + if (!chan) + return -ENODEV; + + + ret = i2c_transfer(&chan->adapter, msg, 2); + if (ret < 0) { + BIOSLOG(bios, "0x%04X: I2CReg: 0x%02X:0x%02X, Value: [no device], " + "Mask: 0x%02X, Data: 0x%02X\n", + offset, reghi, reglo, mask, data); + iexec->execute = 0; + return 7; + } + + BIOSLOG(bios, "0x%04X: I2CReg: 0x%02X:0x%02X, Value: 0x%02X, " + "Mask: 0x%02X, Data: 0x%02X\n", + offset, reghi, reglo, buf1[0], mask, data); + + iexec->execute = ((buf1[0] & mask) == data); + + return 7; +} + +static struct init_tbl_entry itbl_entry[] = { + /* command name , id , length , offset , mult , command handler */ + /* INIT_PROG (0x31, 15, 10, 4) removed due to no example of use */ + { "INIT_IO_RESTRICT_PROG" , 0x32, init_io_restrict_prog }, + { "INIT_REPEAT" , 0x33, init_repeat }, + { "INIT_IO_RESTRICT_PLL" , 0x34, init_io_restrict_pll }, + { "INIT_END_REPEAT" , 0x36, init_end_repeat }, + { "INIT_COPY" , 0x37, init_copy }, + { "INIT_NOT" , 0x38, init_not }, + { "INIT_IO_FLAG_CONDITION" , 0x39, init_io_flag_condition }, + { "INIT_DP_CONDITION" , 0x3A, init_dp_condition }, + { "INIT_OP_3B" , 0x3B, init_op_3b }, + { "INIT_OP_3C" , 0x3C, init_op_3c }, + { "INIT_INDEX_ADDRESS_LATCHED" , 0x49, init_idx_addr_latched }, + { "INIT_IO_RESTRICT_PLL2" , 0x4A, init_io_restrict_pll2 }, + { "INIT_PLL2" , 0x4B, init_pll2 }, + { "INIT_I2C_BYTE" , 0x4C, init_i2c_byte }, + { "INIT_ZM_I2C_BYTE" , 0x4D, init_zm_i2c_byte }, + { "INIT_ZM_I2C" , 0x4E, init_zm_i2c }, + { "INIT_TMDS" , 0x4F, init_tmds }, + { "INIT_ZM_TMDS_GROUP" , 0x50, init_zm_tmds_group }, + { "INIT_CR_INDEX_ADDRESS_LATCHED" , 0x51, init_cr_idx_adr_latch }, + { "INIT_CR" , 0x52, init_cr }, + { "INIT_ZM_CR" , 0x53, init_zm_cr }, + { "INIT_ZM_CR_GROUP" , 0x54, init_zm_cr_group }, + { "INIT_CONDITION_TIME" , 0x56, init_condition_time }, + { "INIT_LTIME" , 0x57, init_ltime }, + { "INIT_ZM_REG_SEQUENCE" , 0x58, init_zm_reg_sequence }, + /* INIT_INDIRECT_REG (0x5A, 7, 0, 0) removed due to no example of use */ + { "INIT_SUB_DIRECT" , 0x5B, init_sub_direct }, + { "INIT_JUMP" , 0x5C, init_jump }, + { "INIT_I2C_IF" , 0x5E, init_i2c_if }, + { "INIT_COPY_NV_REG" , 0x5F, init_copy_nv_reg }, + { "INIT_ZM_INDEX_IO" , 0x62, init_zm_index_io }, + { "INIT_COMPUTE_MEM" , 0x63, init_compute_mem }, + { "INIT_RESET" , 0x65, init_reset }, + { "INIT_CONFIGURE_MEM" , 0x66, init_configure_mem }, + { "INIT_CONFIGURE_CLK" , 0x67, init_configure_clk }, + { "INIT_CONFIGURE_PREINIT" , 0x68, init_configure_preinit }, + { "INIT_IO" , 0x69, init_io }, + { "INIT_SUB" , 0x6B, init_sub }, + { "INIT_RAM_CONDITION" , 0x6D, init_ram_condition }, + { "INIT_NV_REG" , 0x6E, init_nv_reg }, + { "INIT_MACRO" , 0x6F, init_macro }, + { "INIT_DONE" , 0x71, init_done }, + { "INIT_RESUME" , 0x72, init_resume }, + /* INIT_RAM_CONDITION2 (0x73, 9, 0, 0) removed due to no example of use */ + { "INIT_TIME" , 0x74, init_time }, + { "INIT_CONDITION" , 0x75, init_condition }, + { "INIT_IO_CONDITION" , 0x76, init_io_condition }, + { "INIT_INDEX_IO" , 0x78, init_index_io }, + { "INIT_PLL" , 0x79, init_pll }, + { "INIT_ZM_REG" , 0x7A, init_zm_reg }, + { "INIT_RAM_RESTRICT_PLL" , 0x87, init_ram_restrict_pll }, + { "INIT_8C" , 0x8C, init_8c }, + { "INIT_8D" , 0x8D, init_8d }, + { "INIT_GPIO" , 0x8E, init_gpio }, + { "INIT_RAM_RESTRICT_ZM_REG_GROUP" , 0x8F, init_ram_restrict_zm_reg_group }, + { "INIT_COPY_ZM_REG" , 0x90, init_copy_zm_reg }, + { "INIT_ZM_REG_GROUP_ADDRESS_LATCHED" , 0x91, init_zm_reg_group_addr_latched }, + { "INIT_RESERVED" , 0x92, init_reserved }, + { "INIT_96" , 0x96, init_96 }, + { "INIT_97" , 0x97, init_97 }, + { "INIT_AUXCH" , 0x98, init_auxch }, + { "INIT_ZM_AUXCH" , 0x99, init_zm_auxch }, + { "INIT_I2C_LONG_IF" , 0x9A, init_i2c_long_if }, + { NULL , 0 , NULL } +}; + +#define MAX_TABLE_OPS 1000 + +static int +parse_init_table(struct nvbios *bios, uint16_t offset, struct init_exec *iexec) +{ + /* + * Parses all commands in an init table. + * + * We start out executing all commands found in the init table. Some + * opcodes may change the status of iexec->execute to SKIP, which will + * cause the following opcodes to perform no operation until the value + * is changed back to EXECUTE. + */ + + int count = 0, i, ret; + uint8_t id; + + /* + * Loop until INIT_DONE causes us to break out of the loop + * (or until offset > bios length just in case... ) + * (and no more than MAX_TABLE_OPS iterations, just in case... ) + */ + while ((offset < bios->length) && (count++ < MAX_TABLE_OPS)) { + id = bios->data[offset]; + + /* Find matching id in itbl_entry */ + for (i = 0; itbl_entry[i].name && (itbl_entry[i].id != id); i++) + ; + + if (!itbl_entry[i].name) { + NV_ERROR(bios->dev, + "0x%04X: Init table command not found: " + "0x%02X\n", offset, id); + return -ENOENT; + } + + BIOSLOG(bios, "0x%04X: [ (0x%02X) - %s ]\n", offset, + itbl_entry[i].id, itbl_entry[i].name); + + /* execute eventual command handler */ + ret = (*itbl_entry[i].handler)(bios, offset, iexec); + if (ret < 0) { + NV_ERROR(bios->dev, "0x%04X: Failed parsing init " + "table opcode: %s %d\n", offset, + itbl_entry[i].name, ret); + } + + if (ret <= 0) + break; + + /* + * Add the offset of the current command including all data + * of that command. The offset will then be pointing on the + * next op code. + */ + offset += ret; + } + + if (offset >= bios->length) + NV_WARN(bios->dev, + "Offset 0x%04X greater than known bios image length. " + "Corrupt image?\n", offset); + if (count >= MAX_TABLE_OPS) + NV_WARN(bios->dev, + "More than %d opcodes to a table is unlikely, " + "is the bios image corrupt?\n", MAX_TABLE_OPS); + + return 0; +} + +static void +parse_init_tables(struct nvbios *bios) +{ + /* Loops and calls parse_init_table() for each present table. */ + + int i = 0; + uint16_t table; + struct init_exec iexec = {true, false}; + + if (bios->old_style_init) { + if (bios->init_script_tbls_ptr) + parse_init_table(bios, bios->init_script_tbls_ptr, &iexec); + if (bios->extra_init_script_tbl_ptr) + parse_init_table(bios, bios->extra_init_script_tbl_ptr, &iexec); + + return; + } + + while ((table = ROM16(bios->data[bios->init_script_tbls_ptr + i]))) { + NV_INFO(bios->dev, + "Parsing VBIOS init table %d at offset 0x%04X\n", + i / 2, table); + BIOSLOG(bios, "0x%04X: ------ Executing following commands ------\n", table); + + parse_init_table(bios, table, &iexec); + i += 2; + } +} + +static uint16_t clkcmptable(struct nvbios *bios, uint16_t clktable, int pxclk) +{ + int compare_record_len, i = 0; + uint16_t compareclk, scriptptr = 0; + + if (bios->major_version < 5) /* pre BIT */ + compare_record_len = 3; + else + compare_record_len = 4; + + do { + compareclk = ROM16(bios->data[clktable + compare_record_len * i]); + if (pxclk >= compareclk * 10) { + if (bios->major_version < 5) { + uint8_t tmdssub = bios->data[clktable + 2 + compare_record_len * i]; + scriptptr = ROM16(bios->data[bios->init_script_tbls_ptr + tmdssub * 2]); + } else + scriptptr = ROM16(bios->data[clktable + 2 + compare_record_len * i]); + break; + } + i++; + } while (compareclk); + + return scriptptr; +} + +static void +run_digital_op_script(struct drm_device *dev, uint16_t scriptptr, + struct dcb_entry *dcbent, int head, bool dl) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nvbios *bios = &dev_priv->vbios; + struct init_exec iexec = {true, false}; + + NV_TRACE(dev, "0x%04X: Parsing digital output script table\n", + scriptptr); + bios_idxprt_wr(bios, NV_CIO_CRX__COLOR, NV_CIO_CRE_44, + head ? NV_CIO_CRE_44_HEADB : NV_CIO_CRE_44_HEADA); + /* note: if dcb entries have been merged, index may be misleading */ + NVWriteVgaCrtc5758(dev, head, 0, dcbent->index); + parse_init_table(bios, scriptptr, &iexec); + + nv04_dfp_bind_head(dev, dcbent, head, dl); +} + +static int call_lvds_manufacturer_script(struct drm_device *dev, struct dcb_entry *dcbent, int head, enum LVDS_script script) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nvbios *bios = &dev_priv->vbios; + uint8_t sub = bios->data[bios->fp.xlated_entry + script] + (bios->fp.link_c_increment && dcbent->or & OUTPUT_C ? 1 : 0); + uint16_t scriptofs = ROM16(bios->data[bios->init_script_tbls_ptr + sub * 2]); + + if (!bios->fp.xlated_entry || !sub || !scriptofs) + return -EINVAL; + + run_digital_op_script(dev, scriptofs, dcbent, head, bios->fp.dual_link); + + if (script == LVDS_PANEL_OFF) { + /* off-on delay in ms */ + mdelay(ROM16(bios->data[bios->fp.xlated_entry + 7])); + } +#ifdef __powerpc__ + /* Powerbook specific quirks */ + if (script == LVDS_RESET && + (dev->pci_device == 0x0179 || dev->pci_device == 0x0189 || + dev->pci_device == 0x0329)) + nv_write_tmds(dev, dcbent->or, 0, 0x02, 0x72); +#endif + + return 0; +} + +static int run_lvds_table(struct drm_device *dev, struct dcb_entry *dcbent, int head, enum LVDS_script script, int pxclk) +{ + /* + * The BIT LVDS table's header has the information to setup the + * necessary registers. Following the standard 4 byte header are: + * A bitmask byte and a dual-link transition pxclk value for use in + * selecting the init script when not using straps; 4 script pointers + * for panel power, selected by output and on/off; and 8 table pointers + * for panel init, the needed one determined by output, and bits in the + * conf byte. These tables are similar to the TMDS tables, consisting + * of a list of pxclks and script pointers. + */ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nvbios *bios = &dev_priv->vbios; + unsigned int outputset = (dcbent->or == 4) ? 1 : 0; + uint16_t scriptptr = 0, clktable; + + /* + * For now we assume version 3.0 table - g80 support will need some + * changes + */ + + switch (script) { + case LVDS_INIT: + return -ENOSYS; + case LVDS_BACKLIGHT_ON: + case LVDS_PANEL_ON: + scriptptr = ROM16(bios->data[bios->fp.lvdsmanufacturerpointer + 7 + outputset * 2]); + break; + case LVDS_BACKLIGHT_OFF: + case LVDS_PANEL_OFF: + scriptptr = ROM16(bios->data[bios->fp.lvdsmanufacturerpointer + 11 + outputset * 2]); + break; + case LVDS_RESET: + clktable = bios->fp.lvdsmanufacturerpointer + 15; + if (dcbent->or == 4) + clktable += 8; + + if (dcbent->lvdsconf.use_straps_for_mode) { + if (bios->fp.dual_link) + clktable += 4; + if (bios->fp.if_is_24bit) + clktable += 2; + } else { + /* using EDID */ + int cmpval_24bit = (dcbent->or == 4) ? 4 : 1; + + if (bios->fp.dual_link) { + clktable += 4; + cmpval_24bit <<= 1; + } + + if (bios->fp.strapless_is_24bit & cmpval_24bit) + clktable += 2; + } + + clktable = ROM16(bios->data[clktable]); + if (!clktable) { + NV_ERROR(dev, "Pixel clock comparison table not found\n"); + return -ENOENT; + } + scriptptr = clkcmptable(bios, clktable, pxclk); + } + + if (!scriptptr) { + NV_ERROR(dev, "LVDS output init script not found\n"); + return -ENOENT; + } + run_digital_op_script(dev, scriptptr, dcbent, head, bios->fp.dual_link); + + return 0; +} + +int call_lvds_script(struct drm_device *dev, struct dcb_entry *dcbent, int head, enum LVDS_script script, int pxclk) +{ + /* + * LVDS operations are multiplexed in an effort to present a single API + * which works with two vastly differing underlying structures. + * This acts as the demux + */ + + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nvbios *bios = &dev_priv->vbios; + uint8_t lvds_ver = bios->data[bios->fp.lvdsmanufacturerpointer]; + uint32_t sel_clk_binding, sel_clk; + int ret; + + if (bios->fp.last_script_invoc == (script << 1 | head) || !lvds_ver || + (lvds_ver >= 0x30 && script == LVDS_INIT)) + return 0; + + if (!bios->fp.lvds_init_run) { + bios->fp.lvds_init_run = true; + call_lvds_script(dev, dcbent, head, LVDS_INIT, pxclk); + } + + if (script == LVDS_PANEL_ON && bios->fp.reset_after_pclk_change) + call_lvds_script(dev, dcbent, head, LVDS_RESET, pxclk); + if (script == LVDS_RESET && bios->fp.power_off_for_reset) + call_lvds_script(dev, dcbent, head, LVDS_PANEL_OFF, pxclk); + + NV_TRACE(dev, "Calling LVDS script %d:\n", script); + + /* don't let script change pll->head binding */ + sel_clk_binding = bios_rd32(bios, NV_PRAMDAC_SEL_CLK) & 0x50000; + + if (lvds_ver < 0x30) + ret = call_lvds_manufacturer_script(dev, dcbent, head, script); + else + ret = run_lvds_table(dev, dcbent, head, script, pxclk); + + bios->fp.last_script_invoc = (script << 1 | head); + + sel_clk = NVReadRAMDAC(dev, 0, NV_PRAMDAC_SEL_CLK) & ~0x50000; + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_SEL_CLK, sel_clk | sel_clk_binding); + /* some scripts set a value in NV_PBUS_POWERCTRL_2 and break video overlay */ + nvWriteMC(dev, NV_PBUS_POWERCTRL_2, 0); + + return ret; +} + +struct lvdstableheader { + uint8_t lvds_ver, headerlen, recordlen; +}; + +static int parse_lvds_manufacturer_table_header(struct drm_device *dev, struct nvbios *bios, struct lvdstableheader *lth) +{ + /* + * BMP version (0xa) LVDS table has a simple header of version and + * record length. The BIT LVDS table has the typical BIT table header: + * version byte, header length byte, record length byte, and a byte for + * the maximum number of records that can be held in the table. + */ + + uint8_t lvds_ver, headerlen, recordlen; + + memset(lth, 0, sizeof(struct lvdstableheader)); + + if (bios->fp.lvdsmanufacturerpointer == 0x0) { + NV_ERROR(dev, "Pointer to LVDS manufacturer table invalid\n"); + return -EINVAL; + } + + lvds_ver = bios->data[bios->fp.lvdsmanufacturerpointer]; + + switch (lvds_ver) { + case 0x0a: /* pre NV40 */ + headerlen = 2; + recordlen = bios->data[bios->fp.lvdsmanufacturerpointer + 1]; + break; + case 0x30: /* NV4x */ + headerlen = bios->data[bios->fp.lvdsmanufacturerpointer + 1]; + if (headerlen < 0x1f) { + NV_ERROR(dev, "LVDS table header not understood\n"); + return -EINVAL; + } + recordlen = bios->data[bios->fp.lvdsmanufacturerpointer + 2]; + break; + case 0x40: /* G80/G90 */ + headerlen = bios->data[bios->fp.lvdsmanufacturerpointer + 1]; + if (headerlen < 0x7) { + NV_ERROR(dev, "LVDS table header not understood\n"); + return -EINVAL; + } + recordlen = bios->data[bios->fp.lvdsmanufacturerpointer + 2]; + break; + default: + NV_ERROR(dev, + "LVDS table revision %d.%d not currently supported\n", + lvds_ver >> 4, lvds_ver & 0xf); + return -ENOSYS; + } + + lth->lvds_ver = lvds_ver; + lth->headerlen = headerlen; + lth->recordlen = recordlen; + + return 0; +} + +static int +get_fp_strap(struct drm_device *dev, struct nvbios *bios) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + /* + * The fp strap is normally dictated by the "User Strap" in + * PEXTDEV_BOOT_0[20:16], but on BMP cards when bit 2 of the + * Internal_Flags struct at 0x48 is set, the user strap gets overriden + * by the PCI subsystem ID during POST, but not before the previous user + * strap has been committed to CR58 for CR57=0xf on head A, which may be + * read and used instead + */ + + if (bios->major_version < 5 && bios->data[0x48] & 0x4) + return NVReadVgaCrtc5758(dev, 0, 0xf) & 0xf; + + if (dev_priv->card_type >= NV_50) + return (bios_rd32(bios, NV_PEXTDEV_BOOT_0) >> 24) & 0xf; + else + return (bios_rd32(bios, NV_PEXTDEV_BOOT_0) >> 16) & 0xf; +} + +static int parse_fp_mode_table(struct drm_device *dev, struct nvbios *bios) +{ + uint8_t *fptable; + uint8_t fptable_ver, headerlen = 0, recordlen, fpentries = 0xf, fpindex; + int ret, ofs, fpstrapping; + struct lvdstableheader lth; + + if (bios->fp.fptablepointer == 0x0) { + /* Apple cards don't have the fp table; the laptops use DDC */ + /* The table is also missing on some x86 IGPs */ +#ifndef __powerpc__ + NV_ERROR(dev, "Pointer to flat panel table invalid\n"); +#endif + bios->digital_min_front_porch = 0x4b; + return 0; + } + + fptable = &bios->data[bios->fp.fptablepointer]; + fptable_ver = fptable[0]; + + switch (fptable_ver) { + /* + * BMP version 0x5.0x11 BIOSen have version 1 like tables, but no + * version field, and miss one of the spread spectrum/PWM bytes. + * This could affect early GF2Go parts (not seen any appropriate ROMs + * though). Here we assume that a version of 0x05 matches this case + * (combining with a BMP version check would be better), as the + * common case for the panel type field is 0x0005, and that is in + * fact what we are reading the first byte of. + */ + case 0x05: /* some NV10, 11, 15, 16 */ + recordlen = 42; + ofs = -1; + break; + case 0x10: /* some NV15/16, and NV11+ */ + recordlen = 44; + ofs = 0; + break; + case 0x20: /* NV40+ */ + headerlen = fptable[1]; + recordlen = fptable[2]; + fpentries = fptable[3]; + /* + * fptable[4] is the minimum + * RAMDAC_FP_HCRTC -> RAMDAC_FP_HSYNC_START gap + */ + bios->digital_min_front_porch = fptable[4]; + ofs = -7; + break; + default: + NV_ERROR(dev, + "FP table revision %d.%d not currently supported\n", + fptable_ver >> 4, fptable_ver & 0xf); + return -ENOSYS; + } + + if (!bios->is_mobile) /* !mobile only needs digital_min_front_porch */ + return 0; + + ret = parse_lvds_manufacturer_table_header(dev, bios, <h); + if (ret) + return ret; + + if (lth.lvds_ver == 0x30 || lth.lvds_ver == 0x40) { + bios->fp.fpxlatetableptr = bios->fp.lvdsmanufacturerpointer + + lth.headerlen + 1; + bios->fp.xlatwidth = lth.recordlen; + } + if (bios->fp.fpxlatetableptr == 0x0) { + NV_ERROR(dev, "Pointer to flat panel xlat table invalid\n"); + return -EINVAL; + } + + fpstrapping = get_fp_strap(dev, bios); + + fpindex = bios->data[bios->fp.fpxlatetableptr + + fpstrapping * bios->fp.xlatwidth]; + + if (fpindex > fpentries) { + NV_ERROR(dev, "Bad flat panel table index\n"); + return -ENOENT; + } + + /* nv4x cards need both a strap value and fpindex of 0xf to use DDC */ + if (lth.lvds_ver > 0x10) + bios->fp_no_ddc = fpstrapping != 0xf || fpindex != 0xf; + + /* + * If either the strap or xlated fpindex value are 0xf there is no + * panel using a strap-derived bios mode present. this condition + * includes, but is different from, the DDC panel indicator above + */ + if (fpstrapping == 0xf || fpindex == 0xf) + return 0; + + bios->fp.mode_ptr = bios->fp.fptablepointer + headerlen + + recordlen * fpindex + ofs; + + NV_TRACE(dev, "BIOS FP mode: %dx%d (%dkHz pixel clock)\n", + ROM16(bios->data[bios->fp.mode_ptr + 11]) + 1, + ROM16(bios->data[bios->fp.mode_ptr + 25]) + 1, + ROM16(bios->data[bios->fp.mode_ptr + 7]) * 10); + + return 0; +} + +bool nouveau_bios_fp_mode(struct drm_device *dev, struct drm_display_mode *mode) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nvbios *bios = &dev_priv->vbios; + uint8_t *mode_entry = &bios->data[bios->fp.mode_ptr]; + + if (!mode) /* just checking whether we can produce a mode */ + return bios->fp.mode_ptr; + + memset(mode, 0, sizeof(struct drm_display_mode)); + /* + * For version 1.0 (version in byte 0): + * bytes 1-2 are "panel type", including bits on whether Colour/mono, + * single/dual link, and type (TFT etc.) + * bytes 3-6 are bits per colour in RGBX + */ + mode->clock = ROM16(mode_entry[7]) * 10; + /* bytes 9-10 is HActive */ + mode->hdisplay = ROM16(mode_entry[11]) + 1; + /* + * bytes 13-14 is HValid Start + * bytes 15-16 is HValid End + */ + mode->hsync_start = ROM16(mode_entry[17]) + 1; + mode->hsync_end = ROM16(mode_entry[19]) + 1; + mode->htotal = ROM16(mode_entry[21]) + 1; + /* bytes 23-24, 27-30 similarly, but vertical */ + mode->vdisplay = ROM16(mode_entry[25]) + 1; + mode->vsync_start = ROM16(mode_entry[31]) + 1; + mode->vsync_end = ROM16(mode_entry[33]) + 1; + mode->vtotal = ROM16(mode_entry[35]) + 1; + mode->flags |= (mode_entry[37] & 0x10) ? + DRM_MODE_FLAG_PHSYNC : DRM_MODE_FLAG_NHSYNC; + mode->flags |= (mode_entry[37] & 0x1) ? + DRM_MODE_FLAG_PVSYNC : DRM_MODE_FLAG_NVSYNC; + /* + * bytes 38-39 relate to spread spectrum settings + * bytes 40-43 are something to do with PWM + */ + + mode->status = MODE_OK; + mode->type = DRM_MODE_TYPE_DRIVER | DRM_MODE_TYPE_PREFERRED; + drm_mode_set_name(mode); + return bios->fp.mode_ptr; +} + +int nouveau_bios_parse_lvds_table(struct drm_device *dev, int pxclk, bool *dl, bool *if_is_24bit) +{ + /* + * The LVDS table header is (mostly) described in + * parse_lvds_manufacturer_table_header(): the BIT header additionally + * contains the dual-link transition pxclk (in 10s kHz), at byte 5 - if + * straps are not being used for the panel, this specifies the frequency + * at which modes should be set up in the dual link style. + * + * Following the header, the BMP (ver 0xa) table has several records, + * indexed by a separate xlat table, indexed in turn by the fp strap in + * EXTDEV_BOOT. Each record had a config byte, followed by 6 script + * numbers for use by INIT_SUB which controlled panel init and power, + * and finally a dword of ms to sleep between power off and on + * operations. + * + * In the BIT versions, the table following the header serves as an + * integrated config and xlat table: the records in the table are + * indexed by the FP strap nibble in EXTDEV_BOOT, and each record has + * two bytes - the first as a config byte, the second for indexing the + * fp mode table pointed to by the BIT 'D' table + * + * DDC is not used until after card init, so selecting the correct table + * entry and setting the dual link flag for EDID equipped panels, + * requiring tests against the native-mode pixel clock, cannot be done + * until later, when this function should be called with non-zero pxclk + */ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nvbios *bios = &dev_priv->vbios; + int fpstrapping = get_fp_strap(dev, bios), lvdsmanufacturerindex = 0; + struct lvdstableheader lth; + uint16_t lvdsofs; + int ret, chip_version = bios->chip_version; + + ret = parse_lvds_manufacturer_table_header(dev, bios, <h); + if (ret) + return ret; + + switch (lth.lvds_ver) { + case 0x0a: /* pre NV40 */ + lvdsmanufacturerindex = bios->data[ + bios->fp.fpxlatemanufacturertableptr + + fpstrapping]; + + /* we're done if this isn't the EDID panel case */ + if (!pxclk) + break; + + if (chip_version < 0x25) { + /* nv17 behaviour + * + * It seems the old style lvds script pointer is reused + * to select 18/24 bit colour depth for EDID panels. + */ + lvdsmanufacturerindex = + (bios->legacy.lvds_single_a_script_ptr & 1) ? + 2 : 0; + if (pxclk >= bios->fp.duallink_transition_clk) + lvdsmanufacturerindex++; + } else if (chip_version < 0x30) { + /* nv28 behaviour (off-chip encoder) + * + * nv28 does a complex dance of first using byte 121 of + * the EDID to choose the lvdsmanufacturerindex, then + * later attempting to match the EDID manufacturer and + * product IDs in a table (signature 'pidt' (panel id + * table?)), setting an lvdsmanufacturerindex of 0 and + * an fp strap of the match index (or 0xf if none) + */ + lvdsmanufacturerindex = 0; + } else { + /* nv31, nv34 behaviour */ + lvdsmanufacturerindex = 0; + if (pxclk >= bios->fp.duallink_transition_clk) + lvdsmanufacturerindex = 2; + if (pxclk >= 140000) + lvdsmanufacturerindex = 3; + } + + /* + * nvidia set the high nibble of (cr57=f, cr58) to + * lvdsmanufacturerindex in this case; we don't + */ + break; + case 0x30: /* NV4x */ + case 0x40: /* G80/G90 */ + lvdsmanufacturerindex = fpstrapping; + break; + default: + NV_ERROR(dev, "LVDS table revision not currently supported\n"); + return -ENOSYS; + } + + lvdsofs = bios->fp.xlated_entry = bios->fp.lvdsmanufacturerpointer + lth.headerlen + lth.recordlen * lvdsmanufacturerindex; + switch (lth.lvds_ver) { + case 0x0a: + bios->fp.power_off_for_reset = bios->data[lvdsofs] & 1; + bios->fp.reset_after_pclk_change = bios->data[lvdsofs] & 2; + bios->fp.dual_link = bios->data[lvdsofs] & 4; + bios->fp.link_c_increment = bios->data[lvdsofs] & 8; + *if_is_24bit = bios->data[lvdsofs] & 16; + break; + case 0x30: + case 0x40: + /* + * No sign of the "power off for reset" or "reset for panel + * on" bits, but it's safer to assume we should + */ + bios->fp.power_off_for_reset = true; + bios->fp.reset_after_pclk_change = true; + + /* + * It's ok lvdsofs is wrong for nv4x edid case; dual_link is + * over-written, and if_is_24bit isn't used + */ + bios->fp.dual_link = bios->data[lvdsofs] & 1; + bios->fp.if_is_24bit = bios->data[lvdsofs] & 2; + bios->fp.strapless_is_24bit = bios->data[bios->fp.lvdsmanufacturerpointer + 4]; + bios->fp.duallink_transition_clk = ROM16(bios->data[bios->fp.lvdsmanufacturerpointer + 5]) * 10; + break; + } + + /* Dell Latitude D620 reports a too-high value for the dual-link + * transition freq, causing us to program the panel incorrectly. + * + * It doesn't appear the VBIOS actually uses its transition freq + * (90000kHz), instead it uses the "Number of LVDS channels" field + * out of the panel ID structure (http://www.spwg.org/). + * + * For the moment, a quirk will do :) + */ + if (nv_match_device(dev, 0x01d7, 0x1028, 0x01c2)) + bios->fp.duallink_transition_clk = 80000; + + /* set dual_link flag for EDID case */ + if (pxclk && (chip_version < 0x25 || chip_version > 0x28)) + bios->fp.dual_link = (pxclk >= bios->fp.duallink_transition_clk); + + *dl = bios->fp.dual_link; + + return 0; +} + +static uint8_t * +bios_output_config_match(struct drm_device *dev, struct dcb_entry *dcbent, + uint16_t record, int record_len, int record_nr, + bool match_link) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nvbios *bios = &dev_priv->vbios; + uint32_t entry; + uint16_t table; + int i, v; + + switch (dcbent->type) { + case OUTPUT_TMDS: + case OUTPUT_LVDS: + case OUTPUT_DP: + break; + default: + match_link = false; + break; + } + + for (i = 0; i < record_nr; i++, record += record_len) { + table = ROM16(bios->data[record]); + if (!table) + continue; + entry = ROM32(bios->data[table]); + + if (match_link) { + v = (entry & 0x00c00000) >> 22; + if (!(v & dcbent->sorconf.link)) + continue; + } + + v = (entry & 0x000f0000) >> 16; + if (!(v & dcbent->or)) + continue; + + v = (entry & 0x000000f0) >> 4; + if (v != dcbent->location) + continue; + + v = (entry & 0x0000000f); + if (v != dcbent->type) + continue; + + return &bios->data[table]; + } + + return NULL; +} + +void * +nouveau_bios_dp_table(struct drm_device *dev, struct dcb_entry *dcbent, + int *length) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nvbios *bios = &dev_priv->vbios; + uint8_t *table; + + if (!bios->display.dp_table_ptr) { + NV_ERROR(dev, "No pointer to DisplayPort table\n"); + return NULL; + } + table = &bios->data[bios->display.dp_table_ptr]; + + if (table[0] != 0x20 && table[0] != 0x21) { + NV_ERROR(dev, "DisplayPort table version 0x%02x unknown\n", + table[0]); + return NULL; + } + + *length = table[4]; + return bios_output_config_match(dev, dcbent, + bios->display.dp_table_ptr + table[1], + table[2], table[3], table[0] >= 0x21); +} + +int +nouveau_bios_run_display_table(struct drm_device *dev, struct dcb_entry *dcbent, + uint32_t sub, int pxclk) +{ + /* + * The display script table is located by the BIT 'U' table. + * + * It contains an array of pointers to various tables describing + * a particular output type. The first 32-bits of the output + * tables contains similar information to a DCB entry, and is + * used to decide whether that particular table is suitable for + * the output you want to access. + * + * The "record header length" field here seems to indicate the + * offset of the first configuration entry in the output tables. + * This is 10 on most cards I've seen, but 12 has been witnessed + * on DP cards, and there's another script pointer within the + * header. + * + * offset + 0 ( 8 bits): version + * offset + 1 ( 8 bits): header length + * offset + 2 ( 8 bits): record length + * offset + 3 ( 8 bits): number of records + * offset + 4 ( 8 bits): record header length + * offset + 5 (16 bits): pointer to first output script table + */ + + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nvbios *bios = &dev_priv->vbios; + uint8_t *table = &bios->data[bios->display.script_table_ptr]; + uint8_t *otable = NULL; + uint16_t script; + int i = 0; + + if (!bios->display.script_table_ptr) { + NV_ERROR(dev, "No pointer to output script table\n"); + return 1; + } + + /* + * Nothing useful has been in any of the pre-2.0 tables I've seen, + * so until they are, we really don't need to care. + */ + if (table[0] < 0x20) + return 1; + + if (table[0] != 0x20 && table[0] != 0x21) { + NV_ERROR(dev, "Output script table version 0x%02x unknown\n", + table[0]); + return 1; + } + + /* + * The output script tables describing a particular output type + * look as follows: + * + * offset + 0 (32 bits): output this table matches (hash of DCB) + * offset + 4 ( 8 bits): unknown + * offset + 5 ( 8 bits): number of configurations + * offset + 6 (16 bits): pointer to some script + * offset + 8 (16 bits): pointer to some script + * + * headerlen == 10 + * offset + 10 : configuration 0 + * + * headerlen == 12 + * offset + 10 : pointer to some script + * offset + 12 : configuration 0 + * + * Each config entry is as follows: + * + * offset + 0 (16 bits): unknown, assumed to be a match value + * offset + 2 (16 bits): pointer to script table (clock set?) + * offset + 4 (16 bits): pointer to script table (reset?) + * + * There doesn't appear to be a count value to say how many + * entries exist in each script table, instead, a 0 value in + * the first 16-bit word seems to indicate both the end of the + * list and the default entry. The second 16-bit word in the + * script tables is a pointer to the script to execute. + */ + + NV_DEBUG_KMS(dev, "Searching for output entry for %d %d %d\n", + dcbent->type, dcbent->location, dcbent->or); + otable = bios_output_config_match(dev, dcbent, table[1] + + bios->display.script_table_ptr, + table[2], table[3], table[0] >= 0x21); + if (!otable) { + NV_DEBUG_KMS(dev, "failed to match any output table\n"); + return 1; + } + + if (pxclk < -2 || pxclk > 0) { + /* Try to find matching script table entry */ + for (i = 0; i < otable[5]; i++) { + if (ROM16(otable[table[4] + i*6]) == sub) + break; + } + + if (i == otable[5]) { + NV_ERROR(dev, "Table 0x%04x not found for %d/%d, " + "using first\n", + sub, dcbent->type, dcbent->or); + i = 0; + } + } + + if (pxclk == 0) { + script = ROM16(otable[6]); + if (!script) { + NV_DEBUG_KMS(dev, "output script 0 not found\n"); + return 1; + } + + NV_DEBUG_KMS(dev, "0x%04X: parsing output script 0\n", script); + nouveau_bios_run_init_table(dev, script, dcbent); + } else + if (pxclk == -1) { + script = ROM16(otable[8]); + if (!script) { + NV_DEBUG_KMS(dev, "output script 1 not found\n"); + return 1; + } + + NV_DEBUG_KMS(dev, "0x%04X: parsing output script 1\n", script); + nouveau_bios_run_init_table(dev, script, dcbent); + } else + if (pxclk == -2) { + if (table[4] >= 12) + script = ROM16(otable[10]); + else + script = 0; + if (!script) { + NV_DEBUG_KMS(dev, "output script 2 not found\n"); + return 1; + } + + NV_DEBUG_KMS(dev, "0x%04X: parsing output script 2\n", script); + nouveau_bios_run_init_table(dev, script, dcbent); + } else + if (pxclk > 0) { + script = ROM16(otable[table[4] + i*6 + 2]); + if (script) + script = clkcmptable(bios, script, pxclk); + if (!script) { + NV_DEBUG_KMS(dev, "clock script 0 not found\n"); + return 1; + } + + NV_DEBUG_KMS(dev, "0x%04X: parsing clock script 0\n", script); + nouveau_bios_run_init_table(dev, script, dcbent); + } else + if (pxclk < 0) { + script = ROM16(otable[table[4] + i*6 + 4]); + if (script) + script = clkcmptable(bios, script, -pxclk); + if (!script) { + NV_DEBUG_KMS(dev, "clock script 1 not found\n"); + return 1; + } + + NV_DEBUG_KMS(dev, "0x%04X: parsing clock script 1\n", script); + nouveau_bios_run_init_table(dev, script, dcbent); + } + + return 0; +} + + +int run_tmds_table(struct drm_device *dev, struct dcb_entry *dcbent, int head, int pxclk) +{ + /* + * the pxclk parameter is in kHz + * + * This runs the TMDS regs setting code found on BIT bios cards + * + * For ffs(or) == 1 use the first table, for ffs(or) == 2 and + * ffs(or) == 3, use the second. + */ + + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nvbios *bios = &dev_priv->vbios; + int cv = bios->chip_version; + uint16_t clktable = 0, scriptptr; + uint32_t sel_clk_binding, sel_clk; + + /* pre-nv17 off-chip tmds uses scripts, post nv17 doesn't */ + if (cv >= 0x17 && cv != 0x1a && cv != 0x20 && + dcbent->location != DCB_LOC_ON_CHIP) + return 0; + + switch (ffs(dcbent->or)) { + case 1: + clktable = bios->tmds.output0_script_ptr; + break; + case 2: + case 3: + clktable = bios->tmds.output1_script_ptr; + break; + } + + if (!clktable) { + NV_ERROR(dev, "Pixel clock comparison table not found\n"); + return -EINVAL; + } + + scriptptr = clkcmptable(bios, clktable, pxclk); + + if (!scriptptr) { + NV_ERROR(dev, "TMDS output init script not found\n"); + return -ENOENT; + } + + /* don't let script change pll->head binding */ + sel_clk_binding = bios_rd32(bios, NV_PRAMDAC_SEL_CLK) & 0x50000; + run_digital_op_script(dev, scriptptr, dcbent, head, pxclk >= 165000); + sel_clk = NVReadRAMDAC(dev, 0, NV_PRAMDAC_SEL_CLK) & ~0x50000; + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_SEL_CLK, sel_clk | sel_clk_binding); + + return 0; +} + +struct pll_mapping { + u8 type; + u32 reg; +}; + +static struct pll_mapping nv04_pll_mapping[] = { + { PLL_CORE , NV_PRAMDAC_NVPLL_COEFF }, + { PLL_MEMORY, NV_PRAMDAC_MPLL_COEFF }, + { PLL_VPLL0 , NV_PRAMDAC_VPLL_COEFF }, + { PLL_VPLL1 , NV_RAMDAC_VPLL2 }, + {} +}; + +static struct pll_mapping nv40_pll_mapping[] = { + { PLL_CORE , 0x004000 }, + { PLL_MEMORY, 0x004020 }, + { PLL_VPLL0 , NV_PRAMDAC_VPLL_COEFF }, + { PLL_VPLL1 , NV_RAMDAC_VPLL2 }, + {} +}; + +static struct pll_mapping nv50_pll_mapping[] = { + { PLL_CORE , 0x004028 }, + { PLL_SHADER, 0x004020 }, + { PLL_UNK03 , 0x004000 }, + { PLL_MEMORY, 0x004008 }, + { PLL_UNK40 , 0x00e810 }, + { PLL_UNK41 , 0x00e818 }, + { PLL_UNK42 , 0x00e824 }, + { PLL_VPLL0 , 0x614100 }, + { PLL_VPLL1 , 0x614900 }, + {} +}; + +static struct pll_mapping nv84_pll_mapping[] = { + { PLL_CORE , 0x004028 }, + { PLL_SHADER, 0x004020 }, + { PLL_MEMORY, 0x004008 }, + { PLL_UNK05 , 0x004030 }, + { PLL_UNK41 , 0x00e818 }, + { PLL_VPLL0 , 0x614100 }, + { PLL_VPLL1 , 0x614900 }, + {} +}; + +u32 +get_pll_register(struct drm_device *dev, enum pll_types type) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nvbios *bios = &dev_priv->vbios; + struct pll_mapping *map; + int i; + + if (dev_priv->card_type < NV_40) + map = nv04_pll_mapping; + else + if (dev_priv->card_type < NV_50) + map = nv40_pll_mapping; + else { + u8 *plim = &bios->data[bios->pll_limit_tbl_ptr]; + + if (plim[0] >= 0x30) { + u8 *entry = plim + plim[1]; + for (i = 0; i < plim[3]; i++, entry += plim[2]) { + if (entry[0] == type) + return ROM32(entry[3]); + } + + return 0; + } + + if (dev_priv->chipset == 0x50) + map = nv50_pll_mapping; + else + map = nv84_pll_mapping; + } + + while (map->reg) { + if (map->type == type) + return map->reg; + map++; + } + + return 0; +} + +int get_pll_limits(struct drm_device *dev, uint32_t limit_match, struct pll_lims *pll_lim) +{ + /* + * PLL limits table + * + * Version 0x10: NV30, NV31 + * One byte header (version), one record of 24 bytes + * Version 0x11: NV36 - Not implemented + * Seems to have same record style as 0x10, but 3 records rather than 1 + * Version 0x20: Found on Geforce 6 cards + * Trivial 4 byte BIT header. 31 (0x1f) byte record length + * Version 0x21: Found on Geforce 7, 8 and some Geforce 6 cards + * 5 byte header, fifth byte of unknown purpose. 35 (0x23) byte record + * length in general, some (integrated) have an extra configuration byte + * Version 0x30: Found on Geforce 8, separates the register mapping + * from the limits tables. + */ + + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nvbios *bios = &dev_priv->vbios; + int cv = bios->chip_version, pllindex = 0; + uint8_t pll_lim_ver = 0, headerlen = 0, recordlen = 0, entries = 0; + uint32_t crystal_strap_mask, crystal_straps; + + if (!bios->pll_limit_tbl_ptr) { + if (cv == 0x30 || cv == 0x31 || cv == 0x35 || cv == 0x36 || + cv >= 0x40) { + NV_ERROR(dev, "Pointer to PLL limits table invalid\n"); + return -EINVAL; + } + } else + pll_lim_ver = bios->data[bios->pll_limit_tbl_ptr]; + + crystal_strap_mask = 1 << 6; + /* open coded dev->twoHeads test */ + if (cv > 0x10 && cv != 0x15 && cv != 0x1a && cv != 0x20) + crystal_strap_mask |= 1 << 22; + crystal_straps = nvReadEXTDEV(dev, NV_PEXTDEV_BOOT_0) & + crystal_strap_mask; + + switch (pll_lim_ver) { + /* + * We use version 0 to indicate a pre limit table bios (single stage + * pll) and load the hard coded limits instead. + */ + case 0: + break; + case 0x10: + case 0x11: + /* + * Strictly v0x11 has 3 entries, but the last two don't seem + * to get used. + */ + headerlen = 1; + recordlen = 0x18; + entries = 1; + pllindex = 0; + break; + case 0x20: + case 0x21: + case 0x30: + case 0x40: + headerlen = bios->data[bios->pll_limit_tbl_ptr + 1]; + recordlen = bios->data[bios->pll_limit_tbl_ptr + 2]; + entries = bios->data[bios->pll_limit_tbl_ptr + 3]; + break; + default: + NV_ERROR(dev, "PLL limits table revision 0x%X not currently " + "supported\n", pll_lim_ver); + return -ENOSYS; + } + + /* initialize all members to zero */ + memset(pll_lim, 0, sizeof(struct pll_lims)); + + /* if we were passed a type rather than a register, figure + * out the register and store it + */ + if (limit_match > PLL_MAX) + pll_lim->reg = limit_match; + else { + pll_lim->reg = get_pll_register(dev, limit_match); + if (!pll_lim->reg) + return -ENOENT; + } + + if (pll_lim_ver == 0x10 || pll_lim_ver == 0x11) { + uint8_t *pll_rec = &bios->data[bios->pll_limit_tbl_ptr + headerlen + recordlen * pllindex]; + + pll_lim->vco1.minfreq = ROM32(pll_rec[0]); + pll_lim->vco1.maxfreq = ROM32(pll_rec[4]); + pll_lim->vco2.minfreq = ROM32(pll_rec[8]); + pll_lim->vco2.maxfreq = ROM32(pll_rec[12]); + pll_lim->vco1.min_inputfreq = ROM32(pll_rec[16]); + pll_lim->vco2.min_inputfreq = ROM32(pll_rec[20]); + pll_lim->vco1.max_inputfreq = pll_lim->vco2.max_inputfreq = INT_MAX; + + /* these values taken from nv30/31/36 */ + pll_lim->vco1.min_n = 0x1; + if (cv == 0x36) + pll_lim->vco1.min_n = 0x5; + pll_lim->vco1.max_n = 0xff; + pll_lim->vco1.min_m = 0x1; + pll_lim->vco1.max_m = 0xd; + pll_lim->vco2.min_n = 0x4; + /* + * On nv30, 31, 36 (i.e. all cards with two stage PLLs with this + * table version (apart from nv35)), N2 is compared to + * maxN2 (0x46) and 10 * maxM2 (0x4), so set maxN2 to 0x28 and + * save a comparison + */ + pll_lim->vco2.max_n = 0x28; + if (cv == 0x30 || cv == 0x35) + /* only 5 bits available for N2 on nv30/35 */ + pll_lim->vco2.max_n = 0x1f; + pll_lim->vco2.min_m = 0x1; + pll_lim->vco2.max_m = 0x4; + pll_lim->max_log2p = 0x7; + pll_lim->max_usable_log2p = 0x6; + } else if (pll_lim_ver == 0x20 || pll_lim_ver == 0x21) { + uint16_t plloffs = bios->pll_limit_tbl_ptr + headerlen; + uint8_t *pll_rec; + int i; + + /* + * First entry is default match, if nothing better. warn if + * reg field nonzero + */ + if (ROM32(bios->data[plloffs])) + NV_WARN(dev, "Default PLL limit entry has non-zero " + "register field\n"); + + for (i = 1; i < entries; i++) + if (ROM32(bios->data[plloffs + recordlen * i]) == pll_lim->reg) { + pllindex = i; + break; + } + + if ((dev_priv->card_type >= NV_50) && (pllindex == 0)) { + NV_ERROR(dev, "Register 0x%08x not found in PLL " + "limits table", pll_lim->reg); + return -ENOENT; + } + + pll_rec = &bios->data[plloffs + recordlen * pllindex]; + + BIOSLOG(bios, "Loading PLL limits for reg 0x%08x\n", + pllindex ? pll_lim->reg : 0); + + /* + * Frequencies are stored in tables in MHz, kHz are more + * useful, so we convert. + */ + + /* What output frequencies can each VCO generate? */ + pll_lim->vco1.minfreq = ROM16(pll_rec[4]) * 1000; + pll_lim->vco1.maxfreq = ROM16(pll_rec[6]) * 1000; + pll_lim->vco2.minfreq = ROM16(pll_rec[8]) * 1000; + pll_lim->vco2.maxfreq = ROM16(pll_rec[10]) * 1000; + + /* What input frequencies they accept (past the m-divider)? */ + pll_lim->vco1.min_inputfreq = ROM16(pll_rec[12]) * 1000; + pll_lim->vco2.min_inputfreq = ROM16(pll_rec[14]) * 1000; + pll_lim->vco1.max_inputfreq = ROM16(pll_rec[16]) * 1000; + pll_lim->vco2.max_inputfreq = ROM16(pll_rec[18]) * 1000; + + /* What values are accepted as multiplier and divider? */ + pll_lim->vco1.min_n = pll_rec[20]; + pll_lim->vco1.max_n = pll_rec[21]; + pll_lim->vco1.min_m = pll_rec[22]; + pll_lim->vco1.max_m = pll_rec[23]; + pll_lim->vco2.min_n = pll_rec[24]; + pll_lim->vco2.max_n = pll_rec[25]; + pll_lim->vco2.min_m = pll_rec[26]; + pll_lim->vco2.max_m = pll_rec[27]; + + pll_lim->max_usable_log2p = pll_lim->max_log2p = pll_rec[29]; + if (pll_lim->max_log2p > 0x7) + /* pll decoding in nv_hw.c assumes never > 7 */ + NV_WARN(dev, "Max log2 P value greater than 7 (%d)\n", + pll_lim->max_log2p); + if (cv < 0x60) + pll_lim->max_usable_log2p = 0x6; + pll_lim->log2p_bias = pll_rec[30]; + + if (recordlen > 0x22) + pll_lim->refclk = ROM32(pll_rec[31]); + + if (recordlen > 0x23 && pll_rec[35]) + NV_WARN(dev, + "Bits set in PLL configuration byte (%x)\n", + pll_rec[35]); + + /* C51 special not seen elsewhere */ + if (cv == 0x51 && !pll_lim->refclk) { + uint32_t sel_clk = bios_rd32(bios, NV_PRAMDAC_SEL_CLK); + + if ((pll_lim->reg == NV_PRAMDAC_VPLL_COEFF && sel_clk & 0x20) || + (pll_lim->reg == NV_RAMDAC_VPLL2 && sel_clk & 0x80)) { + if (bios_idxprt_rd(bios, NV_CIO_CRX__COLOR, NV_CIO_CRE_CHIP_ID_INDEX) < 0xa3) + pll_lim->refclk = 200000; + else + pll_lim->refclk = 25000; + } + } + } else if (pll_lim_ver == 0x30) { /* ver 0x30 */ + uint8_t *entry = &bios->data[bios->pll_limit_tbl_ptr + headerlen]; + uint8_t *record = NULL; + int i; + + BIOSLOG(bios, "Loading PLL limits for register 0x%08x\n", + pll_lim->reg); + + for (i = 0; i < entries; i++, entry += recordlen) { + if (ROM32(entry[3]) == pll_lim->reg) { + record = &bios->data[ROM16(entry[1])]; + break; + } + } + + if (!record) { + NV_ERROR(dev, "Register 0x%08x not found in PLL " + "limits table", pll_lim->reg); + return -ENOENT; + } + + pll_lim->vco1.minfreq = ROM16(record[0]) * 1000; + pll_lim->vco1.maxfreq = ROM16(record[2]) * 1000; + pll_lim->vco2.minfreq = ROM16(record[4]) * 1000; + pll_lim->vco2.maxfreq = ROM16(record[6]) * 1000; + pll_lim->vco1.min_inputfreq = ROM16(record[8]) * 1000; + pll_lim->vco2.min_inputfreq = ROM16(record[10]) * 1000; + pll_lim->vco1.max_inputfreq = ROM16(record[12]) * 1000; + pll_lim->vco2.max_inputfreq = ROM16(record[14]) * 1000; + pll_lim->vco1.min_n = record[16]; + pll_lim->vco1.max_n = record[17]; + pll_lim->vco1.min_m = record[18]; + pll_lim->vco1.max_m = record[19]; + pll_lim->vco2.min_n = record[20]; + pll_lim->vco2.max_n = record[21]; + pll_lim->vco2.min_m = record[22]; + pll_lim->vco2.max_m = record[23]; + pll_lim->max_usable_log2p = pll_lim->max_log2p = record[25]; + pll_lim->log2p_bias = record[27]; + pll_lim->refclk = ROM32(record[28]); + } else if (pll_lim_ver) { /* ver 0x40 */ + uint8_t *entry = &bios->data[bios->pll_limit_tbl_ptr + headerlen]; + uint8_t *record = NULL; + int i; + + BIOSLOG(bios, "Loading PLL limits for register 0x%08x\n", + pll_lim->reg); + + for (i = 0; i < entries; i++, entry += recordlen) { + if (ROM32(entry[3]) == pll_lim->reg) { + record = &bios->data[ROM16(entry[1])]; + break; + } + } + + if (!record) { + NV_ERROR(dev, "Register 0x%08x not found in PLL " + "limits table", pll_lim->reg); + return -ENOENT; + } + + pll_lim->vco1.minfreq = ROM16(record[0]) * 1000; + pll_lim->vco1.maxfreq = ROM16(record[2]) * 1000; + pll_lim->vco1.min_inputfreq = ROM16(record[4]) * 1000; + pll_lim->vco1.max_inputfreq = ROM16(record[6]) * 1000; + pll_lim->vco1.min_m = record[8]; + pll_lim->vco1.max_m = record[9]; + pll_lim->vco1.min_n = record[10]; + pll_lim->vco1.max_n = record[11]; + pll_lim->min_p = record[12]; + pll_lim->max_p = record[13]; + pll_lim->refclk = ROM16(entry[9]) * 1000; + } + + /* + * By now any valid limit table ought to have set a max frequency for + * vco1, so if it's zero it's either a pre limit table bios, or one + * with an empty limit table (seen on nv18) + */ + if (!pll_lim->vco1.maxfreq) { + pll_lim->vco1.minfreq = bios->fminvco; + pll_lim->vco1.maxfreq = bios->fmaxvco; + pll_lim->vco1.min_inputfreq = 0; + pll_lim->vco1.max_inputfreq = INT_MAX; + pll_lim->vco1.min_n = 0x1; + pll_lim->vco1.max_n = 0xff; + pll_lim->vco1.min_m = 0x1; + if (crystal_straps == 0) { + /* nv05 does this, nv11 doesn't, nv10 unknown */ + if (cv < 0x11) + pll_lim->vco1.min_m = 0x7; + pll_lim->vco1.max_m = 0xd; + } else { + if (cv < 0x11) + pll_lim->vco1.min_m = 0x8; + pll_lim->vco1.max_m = 0xe; + } + if (cv < 0x17 || cv == 0x1a || cv == 0x20) + pll_lim->max_log2p = 4; + else + pll_lim->max_log2p = 5; + pll_lim->max_usable_log2p = pll_lim->max_log2p; + } + + if (!pll_lim->refclk) + switch (crystal_straps) { + case 0: + pll_lim->refclk = 13500; + break; + case (1 << 6): + pll_lim->refclk = 14318; + break; + case (1 << 22): + pll_lim->refclk = 27000; + break; + case (1 << 22 | 1 << 6): + pll_lim->refclk = 25000; + break; + } + + NV_DEBUG(dev, "pll.vco1.minfreq: %d\n", pll_lim->vco1.minfreq); + NV_DEBUG(dev, "pll.vco1.maxfreq: %d\n", pll_lim->vco1.maxfreq); + NV_DEBUG(dev, "pll.vco1.min_inputfreq: %d\n", pll_lim->vco1.min_inputfreq); + NV_DEBUG(dev, "pll.vco1.max_inputfreq: %d\n", pll_lim->vco1.max_inputfreq); + NV_DEBUG(dev, "pll.vco1.min_n: %d\n", pll_lim->vco1.min_n); + NV_DEBUG(dev, "pll.vco1.max_n: %d\n", pll_lim->vco1.max_n); + NV_DEBUG(dev, "pll.vco1.min_m: %d\n", pll_lim->vco1.min_m); + NV_DEBUG(dev, "pll.vco1.max_m: %d\n", pll_lim->vco1.max_m); + if (pll_lim->vco2.maxfreq) { + NV_DEBUG(dev, "pll.vco2.minfreq: %d\n", pll_lim->vco2.minfreq); + NV_DEBUG(dev, "pll.vco2.maxfreq: %d\n", pll_lim->vco2.maxfreq); + NV_DEBUG(dev, "pll.vco2.min_inputfreq: %d\n", pll_lim->vco2.min_inputfreq); + NV_DEBUG(dev, "pll.vco2.max_inputfreq: %d\n", pll_lim->vco2.max_inputfreq); + NV_DEBUG(dev, "pll.vco2.min_n: %d\n", pll_lim->vco2.min_n); + NV_DEBUG(dev, "pll.vco2.max_n: %d\n", pll_lim->vco2.max_n); + NV_DEBUG(dev, "pll.vco2.min_m: %d\n", pll_lim->vco2.min_m); + NV_DEBUG(dev, "pll.vco2.max_m: %d\n", pll_lim->vco2.max_m); + } + if (!pll_lim->max_p) { + NV_DEBUG(dev, "pll.max_log2p: %d\n", pll_lim->max_log2p); + NV_DEBUG(dev, "pll.log2p_bias: %d\n", pll_lim->log2p_bias); + } else { + NV_DEBUG(dev, "pll.min_p: %d\n", pll_lim->min_p); + NV_DEBUG(dev, "pll.max_p: %d\n", pll_lim->max_p); + } + NV_DEBUG(dev, "pll.refclk: %d\n", pll_lim->refclk); + + return 0; +} + +static void parse_bios_version(struct drm_device *dev, struct nvbios *bios, uint16_t offset) +{ + /* + * offset + 0 (8 bits): Micro version + * offset + 1 (8 bits): Minor version + * offset + 2 (8 bits): Chip version + * offset + 3 (8 bits): Major version + */ + + bios->major_version = bios->data[offset + 3]; + bios->chip_version = bios->data[offset + 2]; + NV_TRACE(dev, "Bios version %02x.%02x.%02x.%02x\n", + bios->data[offset + 3], bios->data[offset + 2], + bios->data[offset + 1], bios->data[offset]); +} + +static void parse_script_table_pointers(struct nvbios *bios, uint16_t offset) +{ + /* + * Parses the init table segment for pointers used in script execution. + * + * offset + 0 (16 bits): init script tables pointer + * offset + 2 (16 bits): macro index table pointer + * offset + 4 (16 bits): macro table pointer + * offset + 6 (16 bits): condition table pointer + * offset + 8 (16 bits): io condition table pointer + * offset + 10 (16 bits): io flag condition table pointer + * offset + 12 (16 bits): init function table pointer + */ + + bios->init_script_tbls_ptr = ROM16(bios->data[offset]); + bios->macro_index_tbl_ptr = ROM16(bios->data[offset + 2]); + bios->macro_tbl_ptr = ROM16(bios->data[offset + 4]); + bios->condition_tbl_ptr = ROM16(bios->data[offset + 6]); + bios->io_condition_tbl_ptr = ROM16(bios->data[offset + 8]); + bios->io_flag_condition_tbl_ptr = ROM16(bios->data[offset + 10]); + bios->init_function_tbl_ptr = ROM16(bios->data[offset + 12]); +} + +static int parse_bit_A_tbl_entry(struct drm_device *dev, struct nvbios *bios, struct bit_entry *bitentry) +{ + /* + * Parses the load detect values for g80 cards. + * + * offset + 0 (16 bits): loadval table pointer + */ + + uint16_t load_table_ptr; + uint8_t version, headerlen, entrylen, num_entries; + + if (bitentry->length != 3) { + NV_ERROR(dev, "Do not understand BIT A table\n"); + return -EINVAL; + } + + load_table_ptr = ROM16(bios->data[bitentry->offset]); + + if (load_table_ptr == 0x0) { + NV_ERROR(dev, "Pointer to BIT loadval table invalid\n"); + return -EINVAL; + } + + version = bios->data[load_table_ptr]; + + if (version != 0x10) { + NV_ERROR(dev, "BIT loadval table version %d.%d not supported\n", + version >> 4, version & 0xF); + return -ENOSYS; + } + + headerlen = bios->data[load_table_ptr + 1]; + entrylen = bios->data[load_table_ptr + 2]; + num_entries = bios->data[load_table_ptr + 3]; + + if (headerlen != 4 || entrylen != 4 || num_entries != 2) { + NV_ERROR(dev, "Do not understand BIT loadval table\n"); + return -EINVAL; + } + + /* First entry is normal dac, 2nd tv-out perhaps? */ + bios->dactestval = ROM32(bios->data[load_table_ptr + headerlen]) & 0x3ff; + + return 0; +} + +static int parse_bit_C_tbl_entry(struct drm_device *dev, struct nvbios *bios, struct bit_entry *bitentry) +{ + /* + * offset + 8 (16 bits): PLL limits table pointer + * + * There's more in here, but that's unknown. + */ + + if (bitentry->length < 10) { + NV_ERROR(dev, "Do not understand BIT C table\n"); + return -EINVAL; + } + + bios->pll_limit_tbl_ptr = ROM16(bios->data[bitentry->offset + 8]); + + return 0; +} + +static int parse_bit_display_tbl_entry(struct drm_device *dev, struct nvbios *bios, struct bit_entry *bitentry) +{ + /* + * Parses the flat panel table segment that the bit entry points to. + * Starting at bitentry->offset: + * + * offset + 0 (16 bits): ??? table pointer - seems to have 18 byte + * records beginning with a freq. + * offset + 2 (16 bits): mode table pointer + */ + + if (bitentry->length != 4) { + NV_ERROR(dev, "Do not understand BIT display table\n"); + return -EINVAL; + } + + bios->fp.fptablepointer = ROM16(bios->data[bitentry->offset + 2]); + + return 0; +} + +static int parse_bit_init_tbl_entry(struct drm_device *dev, struct nvbios *bios, struct bit_entry *bitentry) +{ + /* + * Parses the init table segment that the bit entry points to. + * + * See parse_script_table_pointers for layout + */ + + if (bitentry->length < 14) { + NV_ERROR(dev, "Do not understand init table\n"); + return -EINVAL; + } + + parse_script_table_pointers(bios, bitentry->offset); + + if (bitentry->length >= 16) + bios->some_script_ptr = ROM16(bios->data[bitentry->offset + 14]); + if (bitentry->length >= 18) + bios->init96_tbl_ptr = ROM16(bios->data[bitentry->offset + 16]); + + return 0; +} + +static int parse_bit_i_tbl_entry(struct drm_device *dev, struct nvbios *bios, struct bit_entry *bitentry) +{ + /* + * BIT 'i' (info?) table + * + * offset + 0 (32 bits): BIOS version dword (as in B table) + * offset + 5 (8 bits): BIOS feature byte (same as for BMP?) + * offset + 13 (16 bits): pointer to table containing DAC load + * detection comparison values + * + * There's other things in the table, purpose unknown + */ + + uint16_t daccmpoffset; + uint8_t dacver, dacheaderlen; + + if (bitentry->length < 6) { + NV_ERROR(dev, "BIT i table too short for needed information\n"); + return -EINVAL; + } + + parse_bios_version(dev, bios, bitentry->offset); + + /* + * bit 4 seems to indicate a mobile bios (doesn't suffer from BMP's + * Quadro identity crisis), other bits possibly as for BMP feature byte + */ + bios->feature_byte = bios->data[bitentry->offset + 5]; + bios->is_mobile = bios->feature_byte & FEATURE_MOBILE; + + if (bitentry->length < 15) { + NV_WARN(dev, "BIT i table not long enough for DAC load " + "detection comparison table\n"); + return -EINVAL; + } + + daccmpoffset = ROM16(bios->data[bitentry->offset + 13]); + + /* doesn't exist on g80 */ + if (!daccmpoffset) + return 0; + + /* + * The first value in the table, following the header, is the + * comparison value, the second entry is a comparison value for + * TV load detection. + */ + + dacver = bios->data[daccmpoffset]; + dacheaderlen = bios->data[daccmpoffset + 1]; + + if (dacver != 0x00 && dacver != 0x10) { + NV_WARN(dev, "DAC load detection comparison table version " + "%d.%d not known\n", dacver >> 4, dacver & 0xf); + return -ENOSYS; + } + + bios->dactestval = ROM32(bios->data[daccmpoffset + dacheaderlen]); + bios->tvdactestval = ROM32(bios->data[daccmpoffset + dacheaderlen + 4]); + + return 0; +} + +static int parse_bit_lvds_tbl_entry(struct drm_device *dev, struct nvbios *bios, struct bit_entry *bitentry) +{ + /* + * Parses the LVDS table segment that the bit entry points to. + * Starting at bitentry->offset: + * + * offset + 0 (16 bits): LVDS strap xlate table pointer + */ + + if (bitentry->length != 2) { + NV_ERROR(dev, "Do not understand BIT LVDS table\n"); + return -EINVAL; + } + + /* + * No idea if it's still called the LVDS manufacturer table, but + * the concept's close enough. + */ + bios->fp.lvdsmanufacturerpointer = ROM16(bios->data[bitentry->offset]); + + return 0; +} + +static int +parse_bit_M_tbl_entry(struct drm_device *dev, struct nvbios *bios, + struct bit_entry *bitentry) +{ + /* + * offset + 2 (8 bits): number of options in an + * INIT_RAM_RESTRICT_ZM_REG_GROUP opcode option set + * offset + 3 (16 bits): pointer to strap xlate table for RAM + * restrict option selection + * + * There's a bunch of bits in this table other than the RAM restrict + * stuff that we don't use - their use currently unknown + */ + + /* + * Older bios versions don't have a sufficiently long table for + * what we want + */ + if (bitentry->length < 0x5) + return 0; + + if (bitentry->version < 2) { + bios->ram_restrict_group_count = bios->data[bitentry->offset + 2]; + bios->ram_restrict_tbl_ptr = ROM16(bios->data[bitentry->offset + 3]); + } else { + bios->ram_restrict_group_count = bios->data[bitentry->offset + 0]; + bios->ram_restrict_tbl_ptr = ROM16(bios->data[bitentry->offset + 1]); + } + + return 0; +} + +static int parse_bit_tmds_tbl_entry(struct drm_device *dev, struct nvbios *bios, struct bit_entry *bitentry) +{ + /* + * Parses the pointer to the TMDS table + * + * Starting at bitentry->offset: + * + * offset + 0 (16 bits): TMDS table pointer + * + * The TMDS table is typically found just before the DCB table, with a + * characteristic signature of 0x11,0x13 (1.1 being version, 0x13 being + * length?) + * + * At offset +7 is a pointer to a script, which I don't know how to + * run yet. + * At offset +9 is a pointer to another script, likewise + * Offset +11 has a pointer to a table where the first word is a pxclk + * frequency and the second word a pointer to a script, which should be + * run if the comparison pxclk frequency is less than the pxclk desired. + * This repeats for decreasing comparison frequencies + * Offset +13 has a pointer to a similar table + * The selection of table (and possibly +7/+9 script) is dictated by + * "or" from the DCB. + */ + + uint16_t tmdstableptr, script1, script2; + + if (bitentry->length != 2) { + NV_ERROR(dev, "Do not understand BIT TMDS table\n"); + return -EINVAL; + } + + tmdstableptr = ROM16(bios->data[bitentry->offset]); + if (!tmdstableptr) { + NV_ERROR(dev, "Pointer to TMDS table invalid\n"); + return -EINVAL; + } + + NV_INFO(dev, "TMDS table version %d.%d\n", + bios->data[tmdstableptr] >> 4, bios->data[tmdstableptr] & 0xf); + + /* nv50+ has v2.0, but we don't parse it atm */ + if (bios->data[tmdstableptr] != 0x11) + return -ENOSYS; + + /* + * These two scripts are odd: they don't seem to get run even when + * they are not stubbed. + */ + script1 = ROM16(bios->data[tmdstableptr + 7]); + script2 = ROM16(bios->data[tmdstableptr + 9]); + if (bios->data[script1] != 'q' || bios->data[script2] != 'q') + NV_WARN(dev, "TMDS table script pointers not stubbed\n"); + + bios->tmds.output0_script_ptr = ROM16(bios->data[tmdstableptr + 11]); + bios->tmds.output1_script_ptr = ROM16(bios->data[tmdstableptr + 13]); + + return 0; +} + +static int +parse_bit_U_tbl_entry(struct drm_device *dev, struct nvbios *bios, + struct bit_entry *bitentry) +{ + /* + * Parses the pointer to the G80 output script tables + * + * Starting at bitentry->offset: + * + * offset + 0 (16 bits): output script table pointer + */ + + uint16_t outputscripttableptr; + + if (bitentry->length != 3) { + NV_ERROR(dev, "Do not understand BIT U table\n"); + return -EINVAL; + } + + outputscripttableptr = ROM16(bios->data[bitentry->offset]); + bios->display.script_table_ptr = outputscripttableptr; + return 0; +} + +static int +parse_bit_displayport_tbl_entry(struct drm_device *dev, struct nvbios *bios, + struct bit_entry *bitentry) +{ + bios->display.dp_table_ptr = ROM16(bios->data[bitentry->offset]); + return 0; +} + +struct bit_table { + const char id; + int (* const parse_fn)(struct drm_device *, struct nvbios *, struct bit_entry *); +}; + +#define BIT_TABLE(id, funcid) ((struct bit_table){ id, parse_bit_##funcid##_tbl_entry }) + +int +bit_table(struct drm_device *dev, u8 id, struct bit_entry *bit) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nvbios *bios = &dev_priv->vbios; + u8 entries, *entry; + + entries = bios->data[bios->offset + 10]; + entry = &bios->data[bios->offset + 12]; + while (entries--) { + if (entry[0] == id) { + bit->id = entry[0]; + bit->version = entry[1]; + bit->length = ROM16(entry[2]); + bit->offset = ROM16(entry[4]); + bit->data = ROMPTR(bios, entry[4]); + return 0; + } + + entry += bios->data[bios->offset + 9]; + } + + return -ENOENT; +} + +static int +parse_bit_table(struct nvbios *bios, const uint16_t bitoffset, + struct bit_table *table) +{ + struct drm_device *dev = bios->dev; + struct bit_entry bitentry; + + if (bit_table(dev, table->id, &bitentry) == 0) + return table->parse_fn(dev, bios, &bitentry); + + NV_INFO(dev, "BIT table '%c' not found\n", table->id); + return -ENOSYS; +} + +static int +parse_bit_structure(struct nvbios *bios, const uint16_t bitoffset) +{ + int ret; + + /* + * The only restriction on parsing order currently is having 'i' first + * for use of bios->*_version or bios->feature_byte while parsing; + * functions shouldn't be actually *doing* anything apart from pulling + * data from the image into the bios struct, thus no interdependencies + */ + ret = parse_bit_table(bios, bitoffset, &BIT_TABLE('i', i)); + if (ret) /* info? */ + return ret; + if (bios->major_version >= 0x60) /* g80+ */ + parse_bit_table(bios, bitoffset, &BIT_TABLE('A', A)); + ret = parse_bit_table(bios, bitoffset, &BIT_TABLE('C', C)); + if (ret) + return ret; + parse_bit_table(bios, bitoffset, &BIT_TABLE('D', display)); + ret = parse_bit_table(bios, bitoffset, &BIT_TABLE('I', init)); + if (ret) + return ret; + parse_bit_table(bios, bitoffset, &BIT_TABLE('M', M)); /* memory? */ + parse_bit_table(bios, bitoffset, &BIT_TABLE('L', lvds)); + parse_bit_table(bios, bitoffset, &BIT_TABLE('T', tmds)); + parse_bit_table(bios, bitoffset, &BIT_TABLE('U', U)); + parse_bit_table(bios, bitoffset, &BIT_TABLE('d', displayport)); + + return 0; +} + +static int parse_bmp_structure(struct drm_device *dev, struct nvbios *bios, unsigned int offset) +{ + /* + * Parses the BMP structure for useful things, but does not act on them + * + * offset + 5: BMP major version + * offset + 6: BMP minor version + * offset + 9: BMP feature byte + * offset + 10: BCD encoded BIOS version + * + * offset + 18: init script table pointer (for bios versions < 5.10h) + * offset + 20: extra init script table pointer (for bios + * versions < 5.10h) + * + * offset + 24: memory init table pointer (used on early bios versions) + * offset + 26: SDR memory sequencing setup data table + * offset + 28: DDR memory sequencing setup data table + * + * offset + 54: index of I2C CRTC pair to use for CRT output + * offset + 55: index of I2C CRTC pair to use for TV output + * offset + 56: index of I2C CRTC pair to use for flat panel output + * offset + 58: write CRTC index for I2C pair 0 + * offset + 59: read CRTC index for I2C pair 0 + * offset + 60: write CRTC index for I2C pair 1 + * offset + 61: read CRTC index for I2C pair 1 + * + * offset + 67: maximum internal PLL frequency (single stage PLL) + * offset + 71: minimum internal PLL frequency (single stage PLL) + * + * offset + 75: script table pointers, as described in + * parse_script_table_pointers + * + * offset + 89: TMDS single link output A table pointer + * offset + 91: TMDS single link output B table pointer + * offset + 95: LVDS single link output A table pointer + * offset + 105: flat panel timings table pointer + * offset + 107: flat panel strapping translation table pointer + * offset + 117: LVDS manufacturer panel config table pointer + * offset + 119: LVDS manufacturer strapping translation table pointer + * + * offset + 142: PLL limits table pointer + * + * offset + 156: minimum pixel clock for LVDS dual link + */ + + uint8_t *bmp = &bios->data[offset], bmp_version_major, bmp_version_minor; + uint16_t bmplength; + uint16_t legacy_scripts_offset, legacy_i2c_offset; + + /* load needed defaults in case we can't parse this info */ + bios->dcb.i2c[0].write = NV_CIO_CRE_DDC_WR__INDEX; + bios->dcb.i2c[0].read = NV_CIO_CRE_DDC_STATUS__INDEX; + bios->dcb.i2c[1].write = NV_CIO_CRE_DDC0_WR__INDEX; + bios->dcb.i2c[1].read = NV_CIO_CRE_DDC0_STATUS__INDEX; + bios->digital_min_front_porch = 0x4b; + bios->fmaxvco = 256000; + bios->fminvco = 128000; + bios->fp.duallink_transition_clk = 90000; + + bmp_version_major = bmp[5]; + bmp_version_minor = bmp[6]; + + NV_TRACE(dev, "BMP version %d.%d\n", + bmp_version_major, bmp_version_minor); + + /* + * Make sure that 0x36 is blank and can't be mistaken for a DCB + * pointer on early versions + */ + if (bmp_version_major < 5) + *(uint16_t *)&bios->data[0x36] = 0; + + /* + * Seems that the minor version was 1 for all major versions prior + * to 5. Version 6 could theoretically exist, but I suspect BIT + * happened instead. + */ + if ((bmp_version_major < 5 && bmp_version_minor != 1) || bmp_version_major > 5) { + NV_ERROR(dev, "You have an unsupported BMP version. " + "Please send in your bios\n"); + return -ENOSYS; + } + + if (bmp_version_major == 0) + /* nothing that's currently useful in this version */ + return 0; + else if (bmp_version_major == 1) + bmplength = 44; /* exact for 1.01 */ + else if (bmp_version_major == 2) + bmplength = 48; /* exact for 2.01 */ + else if (bmp_version_major == 3) + bmplength = 54; + /* guessed - mem init tables added in this version */ + else if (bmp_version_major == 4 || bmp_version_minor < 0x1) + /* don't know if 5.0 exists... */ + bmplength = 62; + /* guessed - BMP I2C indices added in version 4*/ + else if (bmp_version_minor < 0x6) + bmplength = 67; /* exact for 5.01 */ + else if (bmp_version_minor < 0x10) + bmplength = 75; /* exact for 5.06 */ + else if (bmp_version_minor == 0x10) + bmplength = 89; /* exact for 5.10h */ + else if (bmp_version_minor < 0x14) + bmplength = 118; /* exact for 5.11h */ + else if (bmp_version_minor < 0x24) + /* + * Not sure of version where pll limits came in; + * certainly exist by 0x24 though. + */ + /* length not exact: this is long enough to get lvds members */ + bmplength = 123; + else if (bmp_version_minor < 0x27) + /* + * Length not exact: this is long enough to get pll limit + * member + */ + bmplength = 144; + else + /* + * Length not exact: this is long enough to get dual link + * transition clock. + */ + bmplength = 158; + + /* checksum */ + if (nv_cksum(bmp, 8)) { + NV_ERROR(dev, "Bad BMP checksum\n"); + return -EINVAL; + } + + /* + * Bit 4 seems to indicate either a mobile bios or a quadro card -- + * mobile behaviour consistent (nv11+), quadro only seen nv18gl-nv36gl + * (not nv10gl), bit 5 that the flat panel tables are present, and + * bit 6 a tv bios. + */ + bios->feature_byte = bmp[9]; + + parse_bios_version(dev, bios, offset + 10); + + if (bmp_version_major < 5 || bmp_version_minor < 0x10) + bios->old_style_init = true; + legacy_scripts_offset = 18; + if (bmp_version_major < 2) + legacy_scripts_offset -= 4; + bios->init_script_tbls_ptr = ROM16(bmp[legacy_scripts_offset]); + bios->extra_init_script_tbl_ptr = ROM16(bmp[legacy_scripts_offset + 2]); + + if (bmp_version_major > 2) { /* appears in BMP 3 */ + bios->legacy.mem_init_tbl_ptr = ROM16(bmp[24]); + bios->legacy.sdr_seq_tbl_ptr = ROM16(bmp[26]); + bios->legacy.ddr_seq_tbl_ptr = ROM16(bmp[28]); + } + + legacy_i2c_offset = 0x48; /* BMP version 2 & 3 */ + if (bmplength > 61) + legacy_i2c_offset = offset + 54; + bios->legacy.i2c_indices.crt = bios->data[legacy_i2c_offset]; + bios->legacy.i2c_indices.tv = bios->data[legacy_i2c_offset + 1]; + bios->legacy.i2c_indices.panel = bios->data[legacy_i2c_offset + 2]; + if (bios->data[legacy_i2c_offset + 4]) + bios->dcb.i2c[0].write = bios->data[legacy_i2c_offset + 4]; + if (bios->data[legacy_i2c_offset + 5]) + bios->dcb.i2c[0].read = bios->data[legacy_i2c_offset + 5]; + if (bios->data[legacy_i2c_offset + 6]) + bios->dcb.i2c[1].write = bios->data[legacy_i2c_offset + 6]; + if (bios->data[legacy_i2c_offset + 7]) + bios->dcb.i2c[1].read = bios->data[legacy_i2c_offset + 7]; + + if (bmplength > 74) { + bios->fmaxvco = ROM32(bmp[67]); + bios->fminvco = ROM32(bmp[71]); + } + if (bmplength > 88) + parse_script_table_pointers(bios, offset + 75); + if (bmplength > 94) { + bios->tmds.output0_script_ptr = ROM16(bmp[89]); + bios->tmds.output1_script_ptr = ROM16(bmp[91]); + /* + * Never observed in use with lvds scripts, but is reused for + * 18/24 bit panel interface default for EDID equipped panels + * (if_is_24bit not set directly to avoid any oscillation). + */ + bios->legacy.lvds_single_a_script_ptr = ROM16(bmp[95]); + } + if (bmplength > 108) { + bios->fp.fptablepointer = ROM16(bmp[105]); + bios->fp.fpxlatetableptr = ROM16(bmp[107]); + bios->fp.xlatwidth = 1; + } + if (bmplength > 120) { + bios->fp.lvdsmanufacturerpointer = ROM16(bmp[117]); + bios->fp.fpxlatemanufacturertableptr = ROM16(bmp[119]); + } + if (bmplength > 143) + bios->pll_limit_tbl_ptr = ROM16(bmp[142]); + + if (bmplength > 157) + bios->fp.duallink_transition_clk = ROM16(bmp[156]) * 10; + + return 0; +} + +static uint16_t findstr(uint8_t *data, int n, const uint8_t *str, int len) +{ + int i, j; + + for (i = 0; i <= (n - len); i++) { + for (j = 0; j < len; j++) + if (data[i + j] != str[j]) + break; + if (j == len) + return i; + } + + return 0; +} + +static struct dcb_gpio_entry * +new_gpio_entry(struct nvbios *bios) +{ + struct drm_device *dev = bios->dev; + struct dcb_gpio_table *gpio = &bios->dcb.gpio; + + if (gpio->entries >= DCB_MAX_NUM_GPIO_ENTRIES) { + NV_ERROR(dev, "exceeded maximum number of gpio entries!!\n"); + return NULL; + } + + return &gpio->entry[gpio->entries++]; +} + +struct dcb_gpio_entry * +nouveau_bios_gpio_entry(struct drm_device *dev, enum dcb_gpio_tag tag) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nvbios *bios = &dev_priv->vbios; + int i; + + for (i = 0; i < bios->dcb.gpio.entries; i++) { + if (bios->dcb.gpio.entry[i].tag != tag) + continue; + + return &bios->dcb.gpio.entry[i]; + } + + return NULL; +} + +static void +parse_dcb_gpio_table(struct nvbios *bios) +{ + struct drm_device *dev = bios->dev; + struct dcb_gpio_entry *e; + u8 headerlen, entries, recordlen; + u8 *dcb, *gpio = NULL, *entry; + int i; + + dcb = ROMPTR(bios, bios->data[0x36]); + if (dcb[0] >= 0x30) { + gpio = ROMPTR(bios, dcb[10]); + if (!gpio) + goto no_table; + + headerlen = gpio[1]; + entries = gpio[2]; + recordlen = gpio[3]; + } else + if (dcb[0] >= 0x22 && dcb[-1] >= 0x13) { + gpio = ROMPTR(bios, dcb[-15]); + if (!gpio) + goto no_table; + + headerlen = 3; + entries = gpio[2]; + recordlen = gpio[1]; + } else + if (dcb[0] >= 0x22) { + /* No GPIO table present, parse the TVDAC GPIO data. */ + uint8_t *tvdac_gpio = &dcb[-5]; + + if (tvdac_gpio[0] & 1) { + e = new_gpio_entry(bios); + e->tag = DCB_GPIO_TVDAC0; + e->line = tvdac_gpio[1] >> 4; + e->invert = tvdac_gpio[0] & 2; + } + + goto no_table; + } else { + NV_DEBUG(dev, "no/unknown gpio table on DCB 0x%02x\n", dcb[0]); + goto no_table; + } + + entry = gpio + headerlen; + for (i = 0; i < entries; i++, entry += recordlen) { + e = new_gpio_entry(bios); + if (!e) + break; + + if (gpio[0] < 0x40) { + e->entry = ROM16(entry[0]); + e->tag = (e->entry & 0x07e0) >> 5; + if (e->tag == 0x3f) { + bios->dcb.gpio.entries--; + continue; + } + + e->line = (e->entry & 0x001f); + e->invert = ((e->entry & 0xf800) >> 11) != 4; + } else { + e->entry = ROM32(entry[0]); + e->tag = (e->entry & 0x0000ff00) >> 8; + if (e->tag == 0xff) { + bios->dcb.gpio.entries--; + continue; + } + + e->line = (e->entry & 0x0000001f) >> 0; + e->state_default = (e->entry & 0x01000000) >> 24; + e->state[0] = (e->entry & 0x18000000) >> 27; + e->state[1] = (e->entry & 0x60000000) >> 29; + } + } + +no_table: + /* Apple iMac G4 NV18 */ + if (nv_match_device(dev, 0x0189, 0x10de, 0x0010)) { + e = new_gpio_entry(bios); + if (e) { + e->tag = DCB_GPIO_TVDAC0; + e->line = 4; + } + } +} + +struct dcb_connector_table_entry * +nouveau_bios_connector_entry(struct drm_device *dev, int index) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nvbios *bios = &dev_priv->vbios; + struct dcb_connector_table_entry *cte; + + if (index >= bios->dcb.connector.entries) + return NULL; + + cte = &bios->dcb.connector.entry[index]; + if (cte->type == 0xff) + return NULL; + + return cte; +} + +static enum dcb_connector_type +divine_connector_type(struct nvbios *bios, int index) +{ + struct dcb_table *dcb = &bios->dcb; + unsigned encoders = 0, type = DCB_CONNECTOR_NONE; + int i; + + for (i = 0; i < dcb->entries; i++) { + if (dcb->entry[i].connector == index) + encoders |= (1 << dcb->entry[i].type); + } + + if (encoders & (1 << OUTPUT_DP)) { + if (encoders & (1 << OUTPUT_TMDS)) + type = DCB_CONNECTOR_DP; + else + type = DCB_CONNECTOR_eDP; + } else + if (encoders & (1 << OUTPUT_TMDS)) { + if (encoders & (1 << OUTPUT_ANALOG)) + type = DCB_CONNECTOR_DVI_I; + else + type = DCB_CONNECTOR_DVI_D; + } else + if (encoders & (1 << OUTPUT_ANALOG)) { + type = DCB_CONNECTOR_VGA; + } else + if (encoders & (1 << OUTPUT_LVDS)) { + type = DCB_CONNECTOR_LVDS; + } else + if (encoders & (1 << OUTPUT_TV)) { + type = DCB_CONNECTOR_TV_0; + } + + return type; +} + +static void +apply_dcb_connector_quirks(struct nvbios *bios, int idx) +{ + struct dcb_connector_table_entry *cte = &bios->dcb.connector.entry[idx]; + struct drm_device *dev = bios->dev; + + /* Gigabyte NX85T */ + if (nv_match_device(dev, 0x0421, 0x1458, 0x344c)) { + if (cte->type == DCB_CONNECTOR_HDMI_1) + cte->type = DCB_CONNECTOR_DVI_I; + } +} + +static const u8 hpd_gpio[16] = { + 0xff, 0x07, 0x08, 0xff, 0xff, 0x51, 0x52, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0x5e, 0x5f, 0x60, +}; + +static void +parse_dcb_connector_table(struct nvbios *bios) +{ + struct drm_device *dev = bios->dev; + struct dcb_connector_table *ct = &bios->dcb.connector; + struct dcb_connector_table_entry *cte; + uint8_t *conntab = &bios->data[bios->dcb.connector_table_ptr]; + uint8_t *entry; + int i; + + if (!bios->dcb.connector_table_ptr) { + NV_DEBUG_KMS(dev, "No DCB connector table present\n"); + return; + } + + NV_INFO(dev, "DCB connector table: VHER 0x%02x %d %d %d\n", + conntab[0], conntab[1], conntab[2], conntab[3]); + if ((conntab[0] != 0x30 && conntab[0] != 0x40) || + (conntab[3] != 2 && conntab[3] != 4)) { + NV_ERROR(dev, " Unknown! Please report.\n"); + return; + } + + ct->entries = conntab[2]; + + entry = conntab + conntab[1]; + cte = &ct->entry[0]; + for (i = 0; i < conntab[2]; i++, entry += conntab[3], cte++) { + cte->index = i; + if (conntab[3] == 2) + cte->entry = ROM16(entry[0]); + else + cte->entry = ROM32(entry[0]); + + cte->type = (cte->entry & 0x000000ff) >> 0; + cte->index2 = (cte->entry & 0x00000f00) >> 8; + + cte->gpio_tag = ffs((cte->entry & 0x07033000) >> 12); + cte->gpio_tag = hpd_gpio[cte->gpio_tag]; + + if (cte->type == 0xff) + continue; + + apply_dcb_connector_quirks(bios, i); + + NV_INFO(dev, " %d: 0x%08x: type 0x%02x idx %d tag 0x%02x\n", + i, cte->entry, cte->type, cte->index, cte->gpio_tag); + + /* check for known types, fallback to guessing the type + * from attached encoders if we hit an unknown. + */ + switch (cte->type) { + case DCB_CONNECTOR_VGA: + case DCB_CONNECTOR_TV_0: + case DCB_CONNECTOR_TV_1: + case DCB_CONNECTOR_TV_3: + case DCB_CONNECTOR_DVI_I: + case DCB_CONNECTOR_DVI_D: + case DCB_CONNECTOR_LVDS: + case DCB_CONNECTOR_LVDS_SPWG: + case DCB_CONNECTOR_DP: + case DCB_CONNECTOR_eDP: + case DCB_CONNECTOR_HDMI_0: + case DCB_CONNECTOR_HDMI_1: + break; + default: + cte->type = divine_connector_type(bios, cte->index); + NV_WARN(dev, "unknown type, using 0x%02x\n", cte->type); + break; + } + + if (nouveau_override_conntype) { + int type = divine_connector_type(bios, cte->index); + if (type != cte->type) + NV_WARN(dev, " -> type 0x%02x\n", cte->type); + } + + } +} + +static struct dcb_entry *new_dcb_entry(struct dcb_table *dcb) +{ + struct dcb_entry *entry = &dcb->entry[dcb->entries]; + + memset(entry, 0, sizeof(struct dcb_entry)); + entry->index = dcb->entries++; + + return entry; +} + +static void fabricate_dcb_output(struct dcb_table *dcb, int type, int i2c, + int heads, int or) +{ + struct dcb_entry *entry = new_dcb_entry(dcb); + + entry->type = type; + entry->i2c_index = i2c; + entry->heads = heads; + if (type != OUTPUT_ANALOG) + entry->location = !DCB_LOC_ON_CHIP; /* ie OFF CHIP */ + entry->or = or; +} + +static bool +parse_dcb20_entry(struct drm_device *dev, struct dcb_table *dcb, + uint32_t conn, uint32_t conf, struct dcb_entry *entry) +{ + entry->type = conn & 0xf; + entry->i2c_index = (conn >> 4) & 0xf; + entry->heads = (conn >> 8) & 0xf; + if (dcb->version >= 0x40) + entry->connector = (conn >> 12) & 0xf; + entry->bus = (conn >> 16) & 0xf; + entry->location = (conn >> 20) & 0x3; + entry->or = (conn >> 24) & 0xf; + + switch (entry->type) { + case OUTPUT_ANALOG: + /* + * Although the rest of a CRT conf dword is usually + * zeros, mac biosen have stuff there so we must mask + */ + entry->crtconf.maxfreq = (dcb->version < 0x30) ? + (conf & 0xffff) * 10 : + (conf & 0xff) * 10000; + break; + case OUTPUT_LVDS: + { + uint32_t mask; + if (conf & 0x1) + entry->lvdsconf.use_straps_for_mode = true; + if (dcb->version < 0x22) { + mask = ~0xd; + /* + * The laptop in bug 14567 lies and claims to not use + * straps when it does, so assume all DCB 2.0 laptops + * use straps, until a broken EDID using one is produced + */ + entry->lvdsconf.use_straps_for_mode = true; + /* + * Both 0x4 and 0x8 show up in v2.0 tables; assume they + * mean the same thing (probably wrong, but might work) + */ + if (conf & 0x4 || conf & 0x8) + entry->lvdsconf.use_power_scripts = true; + } else { + mask = ~0x7; + if (conf & 0x2) + entry->lvdsconf.use_acpi_for_edid = true; + if (conf & 0x4) + entry->lvdsconf.use_power_scripts = true; + entry->lvdsconf.sor.link = (conf & 0x00000030) >> 4; + } + if (conf & mask) { + /* + * Until we even try to use these on G8x, it's + * useless reporting unknown bits. They all are. + */ + if (dcb->version >= 0x40) + break; + + NV_ERROR(dev, "Unknown LVDS configuration bits, " + "please report\n"); + } + break; + } + case OUTPUT_TV: + { + if (dcb->version >= 0x30) + entry->tvconf.has_component_output = conf & (0x8 << 4); + else + entry->tvconf.has_component_output = false; + + break; + } + case OUTPUT_DP: + entry->dpconf.sor.link = (conf & 0x00000030) >> 4; + entry->dpconf.link_bw = (conf & 0x00e00000) >> 21; + switch ((conf & 0x0f000000) >> 24) { + case 0xf: + entry->dpconf.link_nr = 4; + break; + case 0x3: + entry->dpconf.link_nr = 2; + break; + default: + entry->dpconf.link_nr = 1; + break; + } + break; + case OUTPUT_TMDS: + if (dcb->version >= 0x40) + entry->tmdsconf.sor.link = (conf & 0x00000030) >> 4; + else if (dcb->version >= 0x30) + entry->tmdsconf.slave_addr = (conf & 0x00000700) >> 8; + else if (dcb->version >= 0x22) + entry->tmdsconf.slave_addr = (conf & 0x00000070) >> 4; + + break; + case OUTPUT_EOL: + /* weird g80 mobile type that "nv" treats as a terminator */ + dcb->entries--; + return false; + default: + break; + } + + if (dcb->version < 0x40) { + /* Normal entries consist of a single bit, but dual link has + * the next most significant bit set too + */ + entry->duallink_possible = + ((1 << (ffs(entry->or) - 1)) * 3 == entry->or); + } else { + entry->duallink_possible = (entry->sorconf.link == 3); + } + + /* unsure what DCB version introduces this, 3.0? */ + if (conf & 0x100000) + entry->i2c_upper_default = true; + + return true; +} + +static bool +parse_dcb15_entry(struct drm_device *dev, struct dcb_table *dcb, + uint32_t conn, uint32_t conf, struct dcb_entry *entry) +{ + switch (conn & 0x0000000f) { + case 0: + entry->type = OUTPUT_ANALOG; + break; + case 1: + entry->type = OUTPUT_TV; + break; + case 2: + case 4: + if (conn & 0x10) + entry->type = OUTPUT_LVDS; + else + entry->type = OUTPUT_TMDS; + break; + case 3: + entry->type = OUTPUT_LVDS; + break; + default: + NV_ERROR(dev, "Unknown DCB type %d\n", conn & 0x0000000f); + return false; + } + + entry->i2c_index = (conn & 0x0003c000) >> 14; + entry->heads = ((conn & 0x001c0000) >> 18) + 1; + entry->or = entry->heads; /* same as heads, hopefully safe enough */ + entry->location = (conn & 0x01e00000) >> 21; + entry->bus = (conn & 0x0e000000) >> 25; + entry->duallink_possible = false; + + switch (entry->type) { + case OUTPUT_ANALOG: + entry->crtconf.maxfreq = (conf & 0xffff) * 10; + break; + case OUTPUT_TV: + entry->tvconf.has_component_output = false; + break; + case OUTPUT_LVDS: + if ((conn & 0x00003f00) >> 8 != 0x10) + entry->lvdsconf.use_straps_for_mode = true; + entry->lvdsconf.use_power_scripts = true; + break; + default: + break; + } + + return true; +} + +static bool parse_dcb_entry(struct drm_device *dev, struct dcb_table *dcb, + uint32_t conn, uint32_t conf) +{ + struct dcb_entry *entry = new_dcb_entry(dcb); + bool ret; + + if (dcb->version >= 0x20) + ret = parse_dcb20_entry(dev, dcb, conn, conf, entry); + else + ret = parse_dcb15_entry(dev, dcb, conn, conf, entry); + if (!ret) + return ret; + + read_dcb_i2c_entry(dev, dcb->version, dcb->i2c_table, + entry->i2c_index, &dcb->i2c[entry->i2c_index]); + + return true; +} + +static +void merge_like_dcb_entries(struct drm_device *dev, struct dcb_table *dcb) +{ + /* + * DCB v2.0 lists each output combination separately. + * Here we merge compatible entries to have fewer outputs, with + * more options + */ + + int i, newentries = 0; + + for (i = 0; i < dcb->entries; i++) { + struct dcb_entry *ient = &dcb->entry[i]; + int j; + + for (j = i + 1; j < dcb->entries; j++) { + struct dcb_entry *jent = &dcb->entry[j]; + + if (jent->type == 100) /* already merged entry */ + continue; + + /* merge heads field when all other fields the same */ + if (jent->i2c_index == ient->i2c_index && + jent->type == ient->type && + jent->location == ient->location && + jent->or == ient->or) { + NV_TRACE(dev, "Merging DCB entries %d and %d\n", + i, j); + ient->heads |= jent->heads; + jent->type = 100; /* dummy value */ + } + } + } + + /* Compact entries merged into others out of dcb */ + for (i = 0; i < dcb->entries; i++) { + if (dcb->entry[i].type == 100) + continue; + + if (newentries != i) { + dcb->entry[newentries] = dcb->entry[i]; + dcb->entry[newentries].index = newentries; + } + newentries++; + } + + dcb->entries = newentries; +} + +static bool +apply_dcb_encoder_quirks(struct drm_device *dev, int idx, u32 *conn, u32 *conf) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct dcb_table *dcb = &dev_priv->vbios.dcb; + + /* Dell Precision M6300 + * DCB entry 2: 02025312 00000010 + * DCB entry 3: 02026312 00000020 + * + * Identical, except apparently a different connector on a + * different SOR link. Not a clue how we're supposed to know + * which one is in use if it even shares an i2c line... + * + * Ignore the connector on the second SOR link to prevent + * nasty problems until this is sorted (assuming it's not a + * VBIOS bug). + */ + if (nv_match_device(dev, 0x040d, 0x1028, 0x019b)) { + if (*conn == 0x02026312 && *conf == 0x00000020) + return false; + } + + /* GeForce3 Ti 200 + * + * DCB reports an LVDS output that should be TMDS: + * DCB entry 1: f2005014 ffffffff + */ + if (nv_match_device(dev, 0x0201, 0x1462, 0x8851)) { + if (*conn == 0xf2005014 && *conf == 0xffffffff) { + fabricate_dcb_output(dcb, OUTPUT_TMDS, 1, 1, 1); + return false; + } + } + + /* XFX GT-240X-YA + * + * So many things wrong here, replace the entire encoder table.. + */ + if (nv_match_device(dev, 0x0ca3, 0x1682, 0x3003)) { + if (idx == 0) { + *conn = 0x02001300; /* VGA, connector 1 */ + *conf = 0x00000028; + } else + if (idx == 1) { + *conn = 0x01010312; /* DVI, connector 0 */ + *conf = 0x00020030; + } else + if (idx == 2) { + *conn = 0x01010310; /* VGA, connector 0 */ + *conf = 0x00000028; + } else + if (idx == 3) { + *conn = 0x02022362; /* HDMI, connector 2 */ + *conf = 0x00020010; + } else { + *conn = 0x0000000e; /* EOL */ + *conf = 0x00000000; + } + } + + return true; +} + +static void +fabricate_dcb_encoder_table(struct drm_device *dev, struct nvbios *bios) +{ + struct dcb_table *dcb = &bios->dcb; + int all_heads = (nv_two_heads(dev) ? 3 : 1); + +#ifdef __powerpc__ + /* Apple iMac G4 NV17 */ + if (of_machine_is_compatible("PowerMac4,5")) { + fabricate_dcb_output(dcb, OUTPUT_TMDS, 0, all_heads, 1); + fabricate_dcb_output(dcb, OUTPUT_ANALOG, 1, all_heads, 2); + return; + } +#endif + + /* Make up some sane defaults */ + fabricate_dcb_output(dcb, OUTPUT_ANALOG, LEGACY_I2C_CRT, 1, 1); + + if (nv04_tv_identify(dev, bios->legacy.i2c_indices.tv) >= 0) + fabricate_dcb_output(dcb, OUTPUT_TV, LEGACY_I2C_TV, + all_heads, 0); + + else if (bios->tmds.output0_script_ptr || + bios->tmds.output1_script_ptr) + fabricate_dcb_output(dcb, OUTPUT_TMDS, LEGACY_I2C_PANEL, + all_heads, 1); +} + +static int +parse_dcb_table(struct drm_device *dev, struct nvbios *bios) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct dcb_table *dcb = &bios->dcb; + uint16_t dcbptr = 0, i2ctabptr = 0; + uint8_t *dcbtable; + uint8_t headerlen = 0x4, entries = DCB_MAX_NUM_ENTRIES; + bool configblock = true; + int recordlength = 8, confofs = 4; + int i; + + /* get the offset from 0x36 */ + if (dev_priv->card_type > NV_04) { + dcbptr = ROM16(bios->data[0x36]); + if (dcbptr == 0x0000) + NV_WARN(dev, "No output data (DCB) found in BIOS\n"); + } + + /* this situation likely means a really old card, pre DCB */ + if (dcbptr == 0x0) { + fabricate_dcb_encoder_table(dev, bios); + return 0; + } + + dcbtable = &bios->data[dcbptr]; + + /* get DCB version */ + dcb->version = dcbtable[0]; + NV_TRACE(dev, "Found Display Configuration Block version %d.%d\n", + dcb->version >> 4, dcb->version & 0xf); + + if (dcb->version >= 0x20) { /* NV17+ */ + uint32_t sig; + + if (dcb->version >= 0x30) { /* NV40+ */ + headerlen = dcbtable[1]; + entries = dcbtable[2]; + recordlength = dcbtable[3]; + i2ctabptr = ROM16(dcbtable[4]); + sig = ROM32(dcbtable[6]); + dcb->gpio_table_ptr = ROM16(dcbtable[10]); + dcb->connector_table_ptr = ROM16(dcbtable[20]); + } else { + i2ctabptr = ROM16(dcbtable[2]); + sig = ROM32(dcbtable[4]); + headerlen = 8; + } + + if (sig != 0x4edcbdcb) { + NV_ERROR(dev, "Bad Display Configuration Block " + "signature (%08X)\n", sig); + return -EINVAL; + } + } else if (dcb->version >= 0x15) { /* some NV11 and NV20 */ + char sig[8] = { 0 }; + + strncpy(sig, (char *)&dcbtable[-7], 7); + i2ctabptr = ROM16(dcbtable[2]); + recordlength = 10; + confofs = 6; + + if (strcmp(sig, "DEV_REC")) { + NV_ERROR(dev, "Bad Display Configuration Block " + "signature (%s)\n", sig); + return -EINVAL; + } + } else { + /* + * v1.4 (some NV15/16, NV11+) seems the same as v1.5, but always + * has the same single (crt) entry, even when tv-out present, so + * the conclusion is this version cannot really be used. + * v1.2 tables (some NV6/10, and NV15+) normally have the same + * 5 entries, which are not specific to the card and so no use. + * v1.2 does have an I2C table that read_dcb_i2c_table can + * handle, but cards exist (nv11 in #14821) with a bad i2c table + * pointer, so use the indices parsed in parse_bmp_structure. + * v1.1 (NV5+, maybe some NV4) is entirely unhelpful + */ + NV_TRACEWARN(dev, "No useful information in BIOS output table; " + "adding all possible outputs\n"); + fabricate_dcb_encoder_table(dev, bios); + return 0; + } + + if (!i2ctabptr) + NV_WARN(dev, "No pointer to DCB I2C port table\n"); + else { + dcb->i2c_table = &bios->data[i2ctabptr]; + if (dcb->version >= 0x30) + dcb->i2c_default_indices = dcb->i2c_table[4]; + + /* + * Parse the "management" I2C bus, used for hardware + * monitoring and some external TMDS transmitters. + */ + if (dcb->version >= 0x22) { + int idx = (dcb->version >= 0x40 ? + dcb->i2c_default_indices & 0xf : + 2); + + read_dcb_i2c_entry(dev, dcb->version, dcb->i2c_table, + idx, &dcb->i2c[idx]); + } + } + + if (entries > DCB_MAX_NUM_ENTRIES) + entries = DCB_MAX_NUM_ENTRIES; + + for (i = 0; i < entries; i++) { + uint32_t connection, config = 0; + + connection = ROM32(dcbtable[headerlen + recordlength * i]); + if (configblock) + config = ROM32(dcbtable[headerlen + confofs + recordlength * i]); + + /* seen on an NV11 with DCB v1.5 */ + if (connection == 0x00000000) + break; + + /* seen on an NV17 with DCB v2.0 */ + if (connection == 0xffffffff) + break; + + if ((connection & 0x0000000f) == 0x0000000f) + continue; + + if (!apply_dcb_encoder_quirks(dev, i, &connection, &config)) + continue; + + NV_TRACEWARN(dev, "Raw DCB entry %d: %08x %08x\n", + dcb->entries, connection, config); + + if (!parse_dcb_entry(dev, dcb, connection, config)) + break; + } + + /* + * apart for v2.1+ not being known for requiring merging, this + * guarantees dcbent->index is the index of the entry in the rom image + */ + if (dcb->version < 0x21) + merge_like_dcb_entries(dev, dcb); + + if (!dcb->entries) + return -ENXIO; + + parse_dcb_gpio_table(bios); + parse_dcb_connector_table(bios); + return 0; +} + +static void +fixup_legacy_connector(struct nvbios *bios) +{ + struct dcb_table *dcb = &bios->dcb; + int i, i2c, i2c_conn[DCB_MAX_NUM_I2C_ENTRIES] = { }; + + /* + * DCB 3.0 also has the table in most cases, but there are some cards + * where the table is filled with stub entries, and the DCB entriy + * indices are all 0. We don't need the connector indices on pre-G80 + * chips (yet?) so limit the use to DCB 4.0 and above. + */ + if (dcb->version >= 0x40) + return; + + dcb->connector.entries = 0; + + /* + * No known connector info before v3.0, so make it up. the rule here + * is: anything on the same i2c bus is considered to be on the same + * connector. any output without an associated i2c bus is assigned + * its own unique connector index. + */ + for (i = 0; i < dcb->entries; i++) { + /* + * Ignore the I2C index for on-chip TV-out, as there + * are cards with bogus values (nv31m in bug 23212), + * and it's otherwise useless. + */ + if (dcb->entry[i].type == OUTPUT_TV && + dcb->entry[i].location == DCB_LOC_ON_CHIP) + dcb->entry[i].i2c_index = 0xf; + i2c = dcb->entry[i].i2c_index; + + if (i2c_conn[i2c]) { + dcb->entry[i].connector = i2c_conn[i2c] - 1; + continue; + } + + dcb->entry[i].connector = dcb->connector.entries++; + if (i2c != 0xf) + i2c_conn[i2c] = dcb->connector.entries; + } + + /* Fake the connector table as well as just connector indices */ + for (i = 0; i < dcb->connector.entries; i++) { + dcb->connector.entry[i].index = i; + dcb->connector.entry[i].type = divine_connector_type(bios, i); + dcb->connector.entry[i].gpio_tag = 0xff; + } +} + +static void +fixup_legacy_i2c(struct nvbios *bios) +{ + struct dcb_table *dcb = &bios->dcb; + int i; + + for (i = 0; i < dcb->entries; i++) { + if (dcb->entry[i].i2c_index == LEGACY_I2C_CRT) + dcb->entry[i].i2c_index = bios->legacy.i2c_indices.crt; + if (dcb->entry[i].i2c_index == LEGACY_I2C_PANEL) + dcb->entry[i].i2c_index = bios->legacy.i2c_indices.panel; + if (dcb->entry[i].i2c_index == LEGACY_I2C_TV) + dcb->entry[i].i2c_index = bios->legacy.i2c_indices.tv; + } +} + +static int load_nv17_hwsq_ucode_entry(struct drm_device *dev, struct nvbios *bios, uint16_t hwsq_offset, int entry) +{ + /* + * The header following the "HWSQ" signature has the number of entries, + * and the entry size + * + * An entry consists of a dword to write to the sequencer control reg + * (0x00001304), followed by the ucode bytes, written sequentially, + * starting at reg 0x00001400 + */ + + uint8_t bytes_to_write; + uint16_t hwsq_entry_offset; + int i; + + if (bios->data[hwsq_offset] <= entry) { + NV_ERROR(dev, "Too few entries in HW sequencer table for " + "requested entry\n"); + return -ENOENT; + } + + bytes_to_write = bios->data[hwsq_offset + 1]; + + if (bytes_to_write != 36) { + NV_ERROR(dev, "Unknown HW sequencer entry size\n"); + return -EINVAL; + } + + NV_TRACE(dev, "Loading NV17 power sequencing microcode\n"); + + hwsq_entry_offset = hwsq_offset + 2 + entry * bytes_to_write; + + /* set sequencer control */ + bios_wr32(bios, 0x00001304, ROM32(bios->data[hwsq_entry_offset])); + bytes_to_write -= 4; + + /* write ucode */ + for (i = 0; i < bytes_to_write; i += 4) + bios_wr32(bios, 0x00001400 + i, ROM32(bios->data[hwsq_entry_offset + i + 4])); + + /* twiddle NV_PBUS_DEBUG_4 */ + bios_wr32(bios, NV_PBUS_DEBUG_4, bios_rd32(bios, NV_PBUS_DEBUG_4) | 0x18); + + return 0; +} + +static int load_nv17_hw_sequencer_ucode(struct drm_device *dev, + struct nvbios *bios) +{ + /* + * BMP based cards, from NV17, need a microcode loading to correctly + * control the GPIO etc for LVDS panels + * + * BIT based cards seem to do this directly in the init scripts + * + * The microcode entries are found by the "HWSQ" signature. + */ + + const uint8_t hwsq_signature[] = { 'H', 'W', 'S', 'Q' }; + const int sz = sizeof(hwsq_signature); + int hwsq_offset; + + hwsq_offset = findstr(bios->data, bios->length, hwsq_signature, sz); + if (!hwsq_offset) + return 0; + + /* always use entry 0? */ + return load_nv17_hwsq_ucode_entry(dev, bios, hwsq_offset + sz, 0); +} + +uint8_t *nouveau_bios_embedded_edid(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nvbios *bios = &dev_priv->vbios; + const uint8_t edid_sig[] = { + 0x00, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0x00 }; + uint16_t offset = 0; + uint16_t newoffset; + int searchlen = NV_PROM_SIZE; + + if (bios->fp.edid) + return bios->fp.edid; + + while (searchlen) { + newoffset = findstr(&bios->data[offset], searchlen, + edid_sig, 8); + if (!newoffset) + return NULL; + offset += newoffset; + if (!nv_cksum(&bios->data[offset], EDID1_LEN)) + break; + + searchlen -= offset; + offset++; + } + + NV_TRACE(dev, "Found EDID in BIOS\n"); + + return bios->fp.edid = &bios->data[offset]; +} + +void +nouveau_bios_run_init_table(struct drm_device *dev, uint16_t table, + struct dcb_entry *dcbent) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nvbios *bios = &dev_priv->vbios; + struct init_exec iexec = { true, false }; + + spin_lock_bh(&bios->lock); + bios->display.output = dcbent; + parse_init_table(bios, table, &iexec); + bios->display.output = NULL; + spin_unlock_bh(&bios->lock); +} + +static bool NVInitVBIOS(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nvbios *bios = &dev_priv->vbios; + + memset(bios, 0, sizeof(struct nvbios)); + spin_lock_init(&bios->lock); + bios->dev = dev; + + if (!NVShadowVBIOS(dev, bios->data)) + return false; + + bios->length = NV_PROM_SIZE; + return true; +} + +static int nouveau_parse_vbios_struct(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nvbios *bios = &dev_priv->vbios; + const uint8_t bit_signature[] = { 0xff, 0xb8, 'B', 'I', 'T' }; + const uint8_t bmp_signature[] = { 0xff, 0x7f, 'N', 'V', 0x0 }; + int offset; + + offset = findstr(bios->data, bios->length, + bit_signature, sizeof(bit_signature)); + if (offset) { + NV_TRACE(dev, "BIT BIOS found\n"); + bios->type = NVBIOS_BIT; + bios->offset = offset; + return parse_bit_structure(bios, offset + 6); + } + + offset = findstr(bios->data, bios->length, + bmp_signature, sizeof(bmp_signature)); + if (offset) { + NV_TRACE(dev, "BMP BIOS found\n"); + bios->type = NVBIOS_BMP; + bios->offset = offset; + return parse_bmp_structure(dev, bios, offset); + } + + NV_ERROR(dev, "No known BIOS signature found\n"); + return -ENODEV; +} + +int +nouveau_run_vbios_init(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nvbios *bios = &dev_priv->vbios; + int i, ret = 0; + + /* Reset the BIOS head to 0. */ + bios->state.crtchead = 0; + + if (bios->major_version < 5) /* BMP only */ + load_nv17_hw_sequencer_ucode(dev, bios); + + if (bios->execute) { + bios->fp.last_script_invoc = 0; + bios->fp.lvds_init_run = false; + } + + parse_init_tables(bios); + + /* + * Runs some additional script seen on G8x VBIOSen. The VBIOS' + * parser will run this right after the init tables, the binary + * driver appears to run it at some point later. + */ + if (bios->some_script_ptr) { + struct init_exec iexec = {true, false}; + + NV_INFO(dev, "Parsing VBIOS init table at offset 0x%04X\n", + bios->some_script_ptr); + parse_init_table(bios, bios->some_script_ptr, &iexec); + } + + if (dev_priv->card_type >= NV_50) { + for (i = 0; i < bios->dcb.entries; i++) { + nouveau_bios_run_display_table(dev, + &bios->dcb.entry[i], + 0, 0); + } + } + + return ret; +} + +static void +nouveau_bios_i2c_devices_takedown(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nvbios *bios = &dev_priv->vbios; + struct dcb_i2c_entry *entry; + int i; + + entry = &bios->dcb.i2c[0]; + for (i = 0; i < DCB_MAX_NUM_I2C_ENTRIES; i++, entry++) + nouveau_i2c_fini(dev, entry); +} + +static bool +nouveau_bios_posted(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + unsigned htotal; + + if (dev_priv->card_type >= NV_50) { + if (NVReadVgaCrtc(dev, 0, 0x00) == 0 && + NVReadVgaCrtc(dev, 0, 0x1a) == 0) + return false; + return true; + } + + htotal = NVReadVgaCrtc(dev, 0, 0x06); + htotal |= (NVReadVgaCrtc(dev, 0, 0x07) & 0x01) << 8; + htotal |= (NVReadVgaCrtc(dev, 0, 0x07) & 0x20) << 4; + htotal |= (NVReadVgaCrtc(dev, 0, 0x25) & 0x01) << 10; + htotal |= (NVReadVgaCrtc(dev, 0, 0x41) & 0x01) << 11; + + return (htotal != 0); +} + +int +nouveau_bios_init(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nvbios *bios = &dev_priv->vbios; + int ret; + + if (!NVInitVBIOS(dev)) + return -ENODEV; + + ret = nouveau_parse_vbios_struct(dev); + if (ret) + return ret; + + ret = parse_dcb_table(dev, bios); + if (ret) + return ret; + + fixup_legacy_i2c(bios); + fixup_legacy_connector(bios); + + if (!bios->major_version) /* we don't run version 0 bios */ + return 0; + + /* init script execution disabled */ + bios->execute = false; + + /* ... unless card isn't POSTed already */ + if (!nouveau_bios_posted(dev)) { + NV_INFO(dev, "Adaptor not initialised, " + "running VBIOS init tables.\n"); + bios->execute = true; + } + if (nouveau_force_post) + bios->execute = true; + + ret = nouveau_run_vbios_init(dev); + if (ret) + return ret; + + /* feature_byte on BMP is poor, but init always sets CR4B */ + if (bios->major_version < 5) + bios->is_mobile = NVReadVgaCrtc(dev, 0, NV_CIO_CRE_4B) & 0x40; + + /* all BIT systems need p_f_m_t for digital_min_front_porch */ + if (bios->is_mobile || bios->major_version >= 5) + ret = parse_fp_mode_table(dev, bios); + + /* allow subsequent scripts to execute */ + bios->execute = true; + + return 0; +} + +void +nouveau_bios_takedown(struct drm_device *dev) +{ + nouveau_bios_i2c_devices_takedown(dev); +} diff --git a/drivers/gpu/drm/nouveau/nouveau_bios.h b/drivers/gpu/drm/nouveau/nouveau_bios.h new file mode 100644 index 00000000..050c3141 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_bios.h @@ -0,0 +1,336 @@ +/* + * Copyright 2007-2008 Nouveau Project + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + */ + +#ifndef __NOUVEAU_BIOS_H__ +#define __NOUVEAU_BIOS_H__ + +#include "nvreg.h" +#include "nouveau_i2c.h" + +#define DCB_MAX_NUM_ENTRIES 16 +#define DCB_MAX_NUM_I2C_ENTRIES 16 +#define DCB_MAX_NUM_GPIO_ENTRIES 32 +#define DCB_MAX_NUM_CONNECTOR_ENTRIES 16 + +#define DCB_LOC_ON_CHIP 0 + +#define ROM16(x) le16_to_cpu(*(uint16_t *)&(x)) +#define ROM32(x) le32_to_cpu(*(uint32_t *)&(x)) +#define ROMPTR(bios, x) (ROM16(x) ? &(bios)->data[ROM16(x)] : NULL) + +struct bit_entry { + uint8_t id; + uint8_t version; + uint16_t length; + uint16_t offset; + uint8_t *data; +}; + +int bit_table(struct drm_device *, u8 id, struct bit_entry *); + +struct dcb_i2c_entry { + uint32_t entry; + uint8_t port_type; + uint8_t read, write; + struct nouveau_i2c_chan *chan; +}; + +enum dcb_gpio_tag { + DCB_GPIO_TVDAC0 = 0xc, + DCB_GPIO_TVDAC1 = 0x2d, +}; + +struct dcb_gpio_entry { + enum dcb_gpio_tag tag; + int line; + bool invert; + uint32_t entry; + uint8_t state_default; + uint8_t state[2]; +}; + +struct dcb_gpio_table { + int entries; + struct dcb_gpio_entry entry[DCB_MAX_NUM_GPIO_ENTRIES]; +}; + +enum dcb_connector_type { + DCB_CONNECTOR_VGA = 0x00, + DCB_CONNECTOR_TV_0 = 0x10, + DCB_CONNECTOR_TV_1 = 0x11, + DCB_CONNECTOR_TV_3 = 0x13, + DCB_CONNECTOR_DVI_I = 0x30, + DCB_CONNECTOR_DVI_D = 0x31, + DCB_CONNECTOR_LVDS = 0x40, + DCB_CONNECTOR_LVDS_SPWG = 0x41, + DCB_CONNECTOR_DP = 0x46, + DCB_CONNECTOR_eDP = 0x47, + DCB_CONNECTOR_HDMI_0 = 0x60, + DCB_CONNECTOR_HDMI_1 = 0x61, + DCB_CONNECTOR_NONE = 0xff +}; + +struct dcb_connector_table_entry { + uint8_t index; + uint32_t entry; + enum dcb_connector_type type; + uint8_t index2; + uint8_t gpio_tag; + void *drm; +}; + +struct dcb_connector_table { + int entries; + struct dcb_connector_table_entry entry[DCB_MAX_NUM_CONNECTOR_ENTRIES]; +}; + +enum dcb_type { + OUTPUT_ANALOG = 0, + OUTPUT_TV = 1, + OUTPUT_TMDS = 2, + OUTPUT_LVDS = 3, + OUTPUT_DP = 6, + OUTPUT_EOL = 14, /* DCB 4.0+, appears to be end-of-list */ + OUTPUT_ANY = -1 +}; + +struct dcb_entry { + int index; /* may not be raw dcb index if merging has happened */ + enum dcb_type type; + uint8_t i2c_index; + uint8_t heads; + uint8_t connector; + uint8_t bus; + uint8_t location; + uint8_t or; + bool duallink_possible; + union { + struct sor_conf { + int link; + } sorconf; + struct { + int maxfreq; + } crtconf; + struct { + struct sor_conf sor; + bool use_straps_for_mode; + bool use_acpi_for_edid; + bool use_power_scripts; + } lvdsconf; + struct { + bool has_component_output; + } tvconf; + struct { + struct sor_conf sor; + int link_nr; + int link_bw; + } dpconf; + struct { + struct sor_conf sor; + int slave_addr; + } tmdsconf; + }; + bool i2c_upper_default; +}; + +struct dcb_table { + uint8_t version; + + int entries; + struct dcb_entry entry[DCB_MAX_NUM_ENTRIES]; + + uint8_t *i2c_table; + uint8_t i2c_default_indices; + struct dcb_i2c_entry i2c[DCB_MAX_NUM_I2C_ENTRIES]; + + uint16_t gpio_table_ptr; + struct dcb_gpio_table gpio; + uint16_t connector_table_ptr; + struct dcb_connector_table connector; +}; + +enum nouveau_or { + OUTPUT_A = (1 << 0), + OUTPUT_B = (1 << 1), + OUTPUT_C = (1 << 2) +}; + +enum LVDS_script { + /* Order *does* matter here */ + LVDS_INIT = 1, + LVDS_RESET, + LVDS_BACKLIGHT_ON, + LVDS_BACKLIGHT_OFF, + LVDS_PANEL_ON, + LVDS_PANEL_OFF +}; + +/* these match types in pll limits table version 0x40, + * nouveau uses them on all chipsets internally where a + * specific pll needs to be referenced, but the exact + * register isn't known. + */ +enum pll_types { + PLL_CORE = 0x01, + PLL_SHADER = 0x02, + PLL_UNK03 = 0x03, + PLL_MEMORY = 0x04, + PLL_UNK05 = 0x05, + PLL_UNK40 = 0x40, + PLL_UNK41 = 0x41, + PLL_UNK42 = 0x42, + PLL_VPLL0 = 0x80, + PLL_VPLL1 = 0x81, + PLL_MAX = 0xff +}; + +struct pll_lims { + u32 reg; + + struct { + int minfreq; + int maxfreq; + int min_inputfreq; + int max_inputfreq; + + uint8_t min_m; + uint8_t max_m; + uint8_t min_n; + uint8_t max_n; + } vco1, vco2; + + uint8_t max_log2p; + /* + * for most pre nv50 cards setting a log2P of 7 (the common max_log2p + * value) is no different to 6 (at least for vplls) so allowing the MNP + * calc to use 7 causes the generated clock to be out by a factor of 2. + * however, max_log2p cannot be fixed-up during parsing as the + * unmodified max_log2p value is still needed for setting mplls, hence + * an additional max_usable_log2p member + */ + uint8_t max_usable_log2p; + uint8_t log2p_bias; + + uint8_t min_p; + uint8_t max_p; + + int refclk; +}; + +struct nvbios { + struct drm_device *dev; + enum { + NVBIOS_BMP, + NVBIOS_BIT + } type; + uint16_t offset; + + uint8_t chip_version; + + uint32_t dactestval; + uint32_t tvdactestval; + uint8_t digital_min_front_porch; + bool fp_no_ddc; + + spinlock_t lock; + + uint8_t data[NV_PROM_SIZE]; + unsigned int length; + bool execute; + + uint8_t major_version; + uint8_t feature_byte; + bool is_mobile; + + uint32_t fmaxvco, fminvco; + + bool old_style_init; + uint16_t init_script_tbls_ptr; + uint16_t extra_init_script_tbl_ptr; + uint16_t macro_index_tbl_ptr; + uint16_t macro_tbl_ptr; + uint16_t condition_tbl_ptr; + uint16_t io_condition_tbl_ptr; + uint16_t io_flag_condition_tbl_ptr; + uint16_t init_function_tbl_ptr; + + uint16_t pll_limit_tbl_ptr; + uint16_t ram_restrict_tbl_ptr; + uint8_t ram_restrict_group_count; + + uint16_t some_script_ptr; /* BIT I + 14 */ + uint16_t init96_tbl_ptr; /* BIT I + 16 */ + + struct dcb_table dcb; + + struct { + int crtchead; + } state; + + struct { + struct dcb_entry *output; + uint16_t script_table_ptr; + uint16_t dp_table_ptr; + } display; + + struct { + uint16_t fptablepointer; /* also used by tmds */ + uint16_t fpxlatetableptr; + int xlatwidth; + uint16_t lvdsmanufacturerpointer; + uint16_t fpxlatemanufacturertableptr; + uint16_t mode_ptr; + uint16_t xlated_entry; + bool power_off_for_reset; + bool reset_after_pclk_change; + bool dual_link; + bool link_c_increment; + bool if_is_24bit; + int duallink_transition_clk; + uint8_t strapless_is_24bit; + uint8_t *edid; + + /* will need resetting after suspend */ + int last_script_invoc; + bool lvds_init_run; + } fp; + + struct { + uint16_t output0_script_ptr; + uint16_t output1_script_ptr; + } tmds; + + struct { + uint16_t mem_init_tbl_ptr; + uint16_t sdr_seq_tbl_ptr; + uint16_t ddr_seq_tbl_ptr; + + struct { + uint8_t crt, tv, panel; + } i2c_indices; + + uint16_t lvds_single_a_script_ptr; + } legacy; +}; + +#endif diff --git a/drivers/gpu/drm/nouveau/nouveau_bo.c b/drivers/gpu/drm/nouveau/nouveau_bo.c new file mode 100644 index 00000000..5fb98de0 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_bo.c @@ -0,0 +1,1115 @@ +/* + * Copyright 2007 Dave Airlied + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * VA LINUX SYSTEMS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +/* + * Authors: Dave Airlied + * Ben Skeggs + * Jeremy Kolb + */ + +#include "drmP.h" + +#include "nouveau_drm.h" +#include "nouveau_drv.h" +#include "nouveau_dma.h" +#include "nouveau_mm.h" +#include "nouveau_vm.h" + +#include +#include + +static void +nouveau_bo_del_ttm(struct ttm_buffer_object *bo) +{ + struct drm_nouveau_private *dev_priv = nouveau_bdev(bo->bdev); + struct drm_device *dev = dev_priv->dev; + struct nouveau_bo *nvbo = nouveau_bo(bo); + + if (unlikely(nvbo->gem)) + DRM_ERROR("bo %p still attached to GEM object\n", bo); + + nv10_mem_put_tile_region(dev, nvbo->tile, NULL); + if (nvbo->vma.node) { + nouveau_vm_unmap(&nvbo->vma); + nouveau_vm_put(&nvbo->vma); + } + kfree(nvbo); +} + +static void +nouveau_bo_fixup_align(struct nouveau_bo *nvbo, u32 flags, + int *align, int *size, int *page_shift) +{ + struct drm_nouveau_private *dev_priv = nouveau_bdev(nvbo->bo.bdev); + + if (dev_priv->card_type < NV_50) { + if (nvbo->tile_mode) { + if (dev_priv->chipset >= 0x40) { + *align = 65536; + *size = roundup(*size, 64 * nvbo->tile_mode); + + } else if (dev_priv->chipset >= 0x30) { + *align = 32768; + *size = roundup(*size, 64 * nvbo->tile_mode); + + } else if (dev_priv->chipset >= 0x20) { + *align = 16384; + *size = roundup(*size, 64 * nvbo->tile_mode); + + } else if (dev_priv->chipset >= 0x10) { + *align = 16384; + *size = roundup(*size, 32 * nvbo->tile_mode); + } + } + } else { + if (likely(dev_priv->chan_vm)) { + if (!(flags & TTM_PL_FLAG_TT) && *size > 256 * 1024) + *page_shift = dev_priv->chan_vm->lpg_shift; + else + *page_shift = dev_priv->chan_vm->spg_shift; + } else { + *page_shift = 12; + } + + *size = roundup(*size, (1 << *page_shift)); + *align = max((1 << *page_shift), *align); + } + + *size = roundup(*size, PAGE_SIZE); +} + +int +nouveau_bo_new(struct drm_device *dev, struct nouveau_channel *chan, + int size, int align, uint32_t flags, uint32_t tile_mode, + uint32_t tile_flags, struct nouveau_bo **pnvbo) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_bo *nvbo; + int ret = 0, page_shift = 0; + + nvbo = kzalloc(sizeof(struct nouveau_bo), GFP_KERNEL); + if (!nvbo) + return -ENOMEM; + INIT_LIST_HEAD(&nvbo->head); + INIT_LIST_HEAD(&nvbo->entry); + nvbo->tile_mode = tile_mode; + nvbo->tile_flags = tile_flags; + nvbo->bo.bdev = &dev_priv->ttm.bdev; + + nouveau_bo_fixup_align(nvbo, flags, &align, &size, &page_shift); + align >>= PAGE_SHIFT; + + if (dev_priv->chan_vm) { + ret = nouveau_vm_get(dev_priv->chan_vm, size, page_shift, + NV_MEM_ACCESS_RW, &nvbo->vma); + if (ret) { + kfree(nvbo); + return ret; + } + } + + nvbo->bo.mem.num_pages = size >> PAGE_SHIFT; + nouveau_bo_placement_set(nvbo, flags, 0); + + nvbo->channel = chan; + ret = ttm_bo_init(&dev_priv->ttm.bdev, &nvbo->bo, size, + ttm_bo_type_device, &nvbo->placement, align, 0, + false, NULL, size, nouveau_bo_del_ttm); + if (ret) { + /* ttm will call nouveau_bo_del_ttm if it fails.. */ + return ret; + } + nvbo->channel = NULL; + + if (nvbo->vma.node) + nvbo->bo.offset = nvbo->vma.offset; + *pnvbo = nvbo; + return 0; +} + +static void +set_placement_list(uint32_t *pl, unsigned *n, uint32_t type, uint32_t flags) +{ + *n = 0; + + if (type & TTM_PL_FLAG_VRAM) + pl[(*n)++] = TTM_PL_FLAG_VRAM | flags; + if (type & TTM_PL_FLAG_TT) + pl[(*n)++] = TTM_PL_FLAG_TT | flags; + if (type & TTM_PL_FLAG_SYSTEM) + pl[(*n)++] = TTM_PL_FLAG_SYSTEM | flags; +} + +static void +set_placement_range(struct nouveau_bo *nvbo, uint32_t type) +{ + struct drm_nouveau_private *dev_priv = nouveau_bdev(nvbo->bo.bdev); + int vram_pages = dev_priv->vram_size >> PAGE_SHIFT; + + if (dev_priv->card_type == NV_10 && + nvbo->tile_mode && (type & TTM_PL_FLAG_VRAM) && + nvbo->bo.mem.num_pages < vram_pages / 2) { + /* + * Make sure that the color and depth buffers are handled + * by independent memory controller units. Up to a 9x + * speed up when alpha-blending and depth-test are enabled + * at the same time. + */ + if (nvbo->tile_flags & NOUVEAU_GEM_TILE_ZETA) { + nvbo->placement.fpfn = vram_pages / 2; + nvbo->placement.lpfn = ~0; + } else { + nvbo->placement.fpfn = 0; + nvbo->placement.lpfn = vram_pages / 2; + } + } +} + +void +nouveau_bo_placement_set(struct nouveau_bo *nvbo, uint32_t type, uint32_t busy) +{ + struct ttm_placement *pl = &nvbo->placement; + uint32_t flags = TTM_PL_MASK_CACHING | + (nvbo->pin_refcnt ? TTM_PL_FLAG_NO_EVICT : 0); + + pl->placement = nvbo->placements; + set_placement_list(nvbo->placements, &pl->num_placement, + type, flags); + + pl->busy_placement = nvbo->busy_placements; + set_placement_list(nvbo->busy_placements, &pl->num_busy_placement, + type | busy, flags); + + set_placement_range(nvbo, type); +} + +int +nouveau_bo_pin(struct nouveau_bo *nvbo, uint32_t memtype) +{ + struct drm_nouveau_private *dev_priv = nouveau_bdev(nvbo->bo.bdev); + struct ttm_buffer_object *bo = &nvbo->bo; + int ret; + + if (nvbo->pin_refcnt && !(memtype & (1 << bo->mem.mem_type))) { + NV_ERROR(nouveau_bdev(bo->bdev)->dev, + "bo %p pinned elsewhere: 0x%08x vs 0x%08x\n", bo, + 1 << bo->mem.mem_type, memtype); + return -EINVAL; + } + + if (nvbo->pin_refcnt++) + return 0; + + ret = ttm_bo_reserve(bo, false, false, false, 0); + if (ret) + goto out; + + nouveau_bo_placement_set(nvbo, memtype, 0); + + ret = nouveau_bo_validate(nvbo, false, false, false); + if (ret == 0) { + switch (bo->mem.mem_type) { + case TTM_PL_VRAM: + dev_priv->fb_aper_free -= bo->mem.size; + break; + case TTM_PL_TT: + dev_priv->gart_info.aper_free -= bo->mem.size; + break; + default: + break; + } + } + ttm_bo_unreserve(bo); +out: + if (unlikely(ret)) + nvbo->pin_refcnt--; + return ret; +} + +int +nouveau_bo_unpin(struct nouveau_bo *nvbo) +{ + struct drm_nouveau_private *dev_priv = nouveau_bdev(nvbo->bo.bdev); + struct ttm_buffer_object *bo = &nvbo->bo; + int ret; + + if (--nvbo->pin_refcnt) + return 0; + + ret = ttm_bo_reserve(bo, false, false, false, 0); + if (ret) + return ret; + + nouveau_bo_placement_set(nvbo, bo->mem.placement, 0); + + ret = nouveau_bo_validate(nvbo, false, false, false); + if (ret == 0) { + switch (bo->mem.mem_type) { + case TTM_PL_VRAM: + dev_priv->fb_aper_free += bo->mem.size; + break; + case TTM_PL_TT: + dev_priv->gart_info.aper_free += bo->mem.size; + break; + default: + break; + } + } + + ttm_bo_unreserve(bo); + return ret; +} + +int +nouveau_bo_map(struct nouveau_bo *nvbo) +{ + int ret; + + ret = ttm_bo_reserve(&nvbo->bo, false, false, false, 0); + if (ret) + return ret; + + ret = ttm_bo_kmap(&nvbo->bo, 0, nvbo->bo.mem.num_pages, &nvbo->kmap); + ttm_bo_unreserve(&nvbo->bo); + return ret; +} + +void +nouveau_bo_unmap(struct nouveau_bo *nvbo) +{ + if (nvbo) + ttm_bo_kunmap(&nvbo->kmap); +} + +int +nouveau_bo_validate(struct nouveau_bo *nvbo, bool interruptible, + bool no_wait_reserve, bool no_wait_gpu) +{ + int ret; + + ret = ttm_bo_validate(&nvbo->bo, &nvbo->placement, interruptible, + no_wait_reserve, no_wait_gpu); + if (ret) + return ret; + + if (nvbo->vma.node) + nvbo->bo.offset = nvbo->vma.offset; + return 0; +} + +u16 +nouveau_bo_rd16(struct nouveau_bo *nvbo, unsigned index) +{ + bool is_iomem; + u16 *mem = ttm_kmap_obj_virtual(&nvbo->kmap, &is_iomem); + mem = &mem[index]; + if (is_iomem) + return ioread16_native((void __force __iomem *)mem); + else + return *mem; +} + +void +nouveau_bo_wr16(struct nouveau_bo *nvbo, unsigned index, u16 val) +{ + bool is_iomem; + u16 *mem = ttm_kmap_obj_virtual(&nvbo->kmap, &is_iomem); + mem = &mem[index]; + if (is_iomem) + iowrite16_native(val, (void __force __iomem *)mem); + else + *mem = val; +} + +u32 +nouveau_bo_rd32(struct nouveau_bo *nvbo, unsigned index) +{ + bool is_iomem; + u32 *mem = ttm_kmap_obj_virtual(&nvbo->kmap, &is_iomem); + mem = &mem[index]; + if (is_iomem) + return ioread32_native((void __force __iomem *)mem); + else + return *mem; +} + +void +nouveau_bo_wr32(struct nouveau_bo *nvbo, unsigned index, u32 val) +{ + bool is_iomem; + u32 *mem = ttm_kmap_obj_virtual(&nvbo->kmap, &is_iomem); + mem = &mem[index]; + if (is_iomem) + iowrite32_native(val, (void __force __iomem *)mem); + else + *mem = val; +} + +static struct ttm_backend * +nouveau_bo_create_ttm_backend_entry(struct ttm_bo_device *bdev) +{ + struct drm_nouveau_private *dev_priv = nouveau_bdev(bdev); + struct drm_device *dev = dev_priv->dev; + + switch (dev_priv->gart_info.type) { +#if __OS_HAS_AGP + case NOUVEAU_GART_AGP: + return ttm_agp_backend_init(bdev, dev->agp->bridge); +#endif + case NOUVEAU_GART_PDMA: + case NOUVEAU_GART_HW: + return nouveau_sgdma_init_ttm(dev); + default: + NV_ERROR(dev, "Unknown GART type %d\n", + dev_priv->gart_info.type); + break; + } + + return NULL; +} + +static int +nouveau_bo_invalidate_caches(struct ttm_bo_device *bdev, uint32_t flags) +{ + /* We'll do this from user space. */ + return 0; +} + +static int +nouveau_bo_init_mem_type(struct ttm_bo_device *bdev, uint32_t type, + struct ttm_mem_type_manager *man) +{ + struct drm_nouveau_private *dev_priv = nouveau_bdev(bdev); + struct drm_device *dev = dev_priv->dev; + + switch (type) { + case TTM_PL_SYSTEM: + man->flags = TTM_MEMTYPE_FLAG_MAPPABLE; + man->available_caching = TTM_PL_MASK_CACHING; + man->default_caching = TTM_PL_FLAG_CACHED; + break; + case TTM_PL_VRAM: + if (dev_priv->card_type >= NV_50) { + man->func = &nouveau_vram_manager; + man->io_reserve_fastpath = false; + man->use_io_reserve_lru = true; + } else { + man->func = &ttm_bo_manager_func; + } + man->flags = TTM_MEMTYPE_FLAG_FIXED | + TTM_MEMTYPE_FLAG_MAPPABLE; + man->available_caching = TTM_PL_FLAG_UNCACHED | + TTM_PL_FLAG_WC; + man->default_caching = TTM_PL_FLAG_WC; + break; + case TTM_PL_TT: + if (dev_priv->card_type >= NV_50) + man->func = &nouveau_gart_manager; + else + man->func = &ttm_bo_manager_func; + switch (dev_priv->gart_info.type) { + case NOUVEAU_GART_AGP: + man->flags = TTM_MEMTYPE_FLAG_MAPPABLE; + man->available_caching = TTM_PL_FLAG_UNCACHED | + TTM_PL_FLAG_WC; + man->default_caching = TTM_PL_FLAG_WC; + break; + case NOUVEAU_GART_PDMA: + case NOUVEAU_GART_HW: + man->flags = TTM_MEMTYPE_FLAG_MAPPABLE | + TTM_MEMTYPE_FLAG_CMA; + man->available_caching = TTM_PL_MASK_CACHING; + man->default_caching = TTM_PL_FLAG_CACHED; + man->gpu_offset = dev_priv->gart_info.aper_base; + break; + default: + NV_ERROR(dev, "Unknown GART type: %d\n", + dev_priv->gart_info.type); + return -EINVAL; + } + break; + default: + NV_ERROR(dev, "Unsupported memory type %u\n", (unsigned)type); + return -EINVAL; + } + return 0; +} + +static void +nouveau_bo_evict_flags(struct ttm_buffer_object *bo, struct ttm_placement *pl) +{ + struct nouveau_bo *nvbo = nouveau_bo(bo); + + switch (bo->mem.mem_type) { + case TTM_PL_VRAM: + nouveau_bo_placement_set(nvbo, TTM_PL_FLAG_TT, + TTM_PL_FLAG_SYSTEM); + break; + default: + nouveau_bo_placement_set(nvbo, TTM_PL_FLAG_SYSTEM, 0); + break; + } + + *pl = nvbo->placement; +} + + +/* GPU-assisted copy using NV_MEMORY_TO_MEMORY_FORMAT, can access + * TTM_PL_{VRAM,TT} directly. + */ + +static int +nouveau_bo_move_accel_cleanup(struct nouveau_channel *chan, + struct nouveau_bo *nvbo, bool evict, + bool no_wait_reserve, bool no_wait_gpu, + struct ttm_mem_reg *new_mem) +{ + struct nouveau_fence *fence = NULL; + int ret; + + ret = nouveau_fence_new(chan, &fence, true); + if (ret) + return ret; + + ret = ttm_bo_move_accel_cleanup(&nvbo->bo, fence, NULL, evict, + no_wait_reserve, no_wait_gpu, new_mem); + nouveau_fence_unref(&fence); + return ret; +} + +static int +nvc0_bo_move_m2mf(struct nouveau_channel *chan, struct ttm_buffer_object *bo, + struct ttm_mem_reg *old_mem, struct ttm_mem_reg *new_mem) +{ + struct nouveau_mem *old_node = old_mem->mm_node; + struct nouveau_mem *new_node = new_mem->mm_node; + struct nouveau_bo *nvbo = nouveau_bo(bo); + u32 page_count = new_mem->num_pages; + u64 src_offset, dst_offset; + int ret; + + src_offset = old_node->tmp_vma.offset; + if (new_node->tmp_vma.node) + dst_offset = new_node->tmp_vma.offset; + else + dst_offset = nvbo->vma.offset; + + page_count = new_mem->num_pages; + while (page_count) { + int line_count = (page_count > 2047) ? 2047 : page_count; + + ret = RING_SPACE(chan, 12); + if (ret) + return ret; + + BEGIN_NVC0(chan, 2, NvSubM2MF, 0x0238, 2); + OUT_RING (chan, upper_32_bits(dst_offset)); + OUT_RING (chan, lower_32_bits(dst_offset)); + BEGIN_NVC0(chan, 2, NvSubM2MF, 0x030c, 6); + OUT_RING (chan, upper_32_bits(src_offset)); + OUT_RING (chan, lower_32_bits(src_offset)); + OUT_RING (chan, PAGE_SIZE); /* src_pitch */ + OUT_RING (chan, PAGE_SIZE); /* dst_pitch */ + OUT_RING (chan, PAGE_SIZE); /* line_length */ + OUT_RING (chan, line_count); + BEGIN_NVC0(chan, 2, NvSubM2MF, 0x0300, 1); + OUT_RING (chan, 0x00100110); + + page_count -= line_count; + src_offset += (PAGE_SIZE * line_count); + dst_offset += (PAGE_SIZE * line_count); + } + + return 0; +} + +static int +nv50_bo_move_m2mf(struct nouveau_channel *chan, struct ttm_buffer_object *bo, + struct ttm_mem_reg *old_mem, struct ttm_mem_reg *new_mem) +{ + struct nouveau_mem *old_node = old_mem->mm_node; + struct nouveau_mem *new_node = new_mem->mm_node; + struct nouveau_bo *nvbo = nouveau_bo(bo); + u64 length = (new_mem->num_pages << PAGE_SHIFT); + u64 src_offset, dst_offset; + int ret; + + src_offset = old_node->tmp_vma.offset; + if (new_node->tmp_vma.node) + dst_offset = new_node->tmp_vma.offset; + else + dst_offset = nvbo->vma.offset; + + while (length) { + u32 amount, stride, height; + + amount = min(length, (u64)(4 * 1024 * 1024)); + stride = 16 * 4; + height = amount / stride; + + if (new_mem->mem_type == TTM_PL_VRAM && + nouveau_bo_tile_layout(nvbo)) { + ret = RING_SPACE(chan, 8); + if (ret) + return ret; + + BEGIN_RING(chan, NvSubM2MF, 0x0200, 7); + OUT_RING (chan, 0); + OUT_RING (chan, 0); + OUT_RING (chan, stride); + OUT_RING (chan, height); + OUT_RING (chan, 1); + OUT_RING (chan, 0); + OUT_RING (chan, 0); + } else { + ret = RING_SPACE(chan, 2); + if (ret) + return ret; + + BEGIN_RING(chan, NvSubM2MF, 0x0200, 1); + OUT_RING (chan, 1); + } + if (old_mem->mem_type == TTM_PL_VRAM && + nouveau_bo_tile_layout(nvbo)) { + ret = RING_SPACE(chan, 8); + if (ret) + return ret; + + BEGIN_RING(chan, NvSubM2MF, 0x021c, 7); + OUT_RING (chan, 0); + OUT_RING (chan, 0); + OUT_RING (chan, stride); + OUT_RING (chan, height); + OUT_RING (chan, 1); + OUT_RING (chan, 0); + OUT_RING (chan, 0); + } else { + ret = RING_SPACE(chan, 2); + if (ret) + return ret; + + BEGIN_RING(chan, NvSubM2MF, 0x021c, 1); + OUT_RING (chan, 1); + } + + ret = RING_SPACE(chan, 14); + if (ret) + return ret; + + BEGIN_RING(chan, NvSubM2MF, 0x0238, 2); + OUT_RING (chan, upper_32_bits(src_offset)); + OUT_RING (chan, upper_32_bits(dst_offset)); + BEGIN_RING(chan, NvSubM2MF, 0x030c, 8); + OUT_RING (chan, lower_32_bits(src_offset)); + OUT_RING (chan, lower_32_bits(dst_offset)); + OUT_RING (chan, stride); + OUT_RING (chan, stride); + OUT_RING (chan, stride); + OUT_RING (chan, height); + OUT_RING (chan, 0x00000101); + OUT_RING (chan, 0x00000000); + BEGIN_RING(chan, NvSubM2MF, NV_MEMORY_TO_MEMORY_FORMAT_NOP, 1); + OUT_RING (chan, 0); + + length -= amount; + src_offset += amount; + dst_offset += amount; + } + + return 0; +} + +static inline uint32_t +nouveau_bo_mem_ctxdma(struct ttm_buffer_object *bo, + struct nouveau_channel *chan, struct ttm_mem_reg *mem) +{ + if (mem->mem_type == TTM_PL_TT) + return chan->gart_handle; + return chan->vram_handle; +} + +static int +nv04_bo_move_m2mf(struct nouveau_channel *chan, struct ttm_buffer_object *bo, + struct ttm_mem_reg *old_mem, struct ttm_mem_reg *new_mem) +{ + u32 src_offset = old_mem->start << PAGE_SHIFT; + u32 dst_offset = new_mem->start << PAGE_SHIFT; + u32 page_count = new_mem->num_pages; + int ret; + + ret = RING_SPACE(chan, 3); + if (ret) + return ret; + + BEGIN_RING(chan, NvSubM2MF, NV_MEMORY_TO_MEMORY_FORMAT_DMA_SOURCE, 2); + OUT_RING (chan, nouveau_bo_mem_ctxdma(bo, chan, old_mem)); + OUT_RING (chan, nouveau_bo_mem_ctxdma(bo, chan, new_mem)); + + page_count = new_mem->num_pages; + while (page_count) { + int line_count = (page_count > 2047) ? 2047 : page_count; + + ret = RING_SPACE(chan, 11); + if (ret) + return ret; + + BEGIN_RING(chan, NvSubM2MF, + NV_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN, 8); + OUT_RING (chan, src_offset); + OUT_RING (chan, dst_offset); + OUT_RING (chan, PAGE_SIZE); /* src_pitch */ + OUT_RING (chan, PAGE_SIZE); /* dst_pitch */ + OUT_RING (chan, PAGE_SIZE); /* line_length */ + OUT_RING (chan, line_count); + OUT_RING (chan, 0x00000101); + OUT_RING (chan, 0x00000000); + BEGIN_RING(chan, NvSubM2MF, NV_MEMORY_TO_MEMORY_FORMAT_NOP, 1); + OUT_RING (chan, 0); + + page_count -= line_count; + src_offset += (PAGE_SIZE * line_count); + dst_offset += (PAGE_SIZE * line_count); + } + + return 0; +} + +static int +nouveau_bo_move_m2mf(struct ttm_buffer_object *bo, int evict, bool intr, + bool no_wait_reserve, bool no_wait_gpu, + struct ttm_mem_reg *new_mem) +{ + struct drm_nouveau_private *dev_priv = nouveau_bdev(bo->bdev); + struct nouveau_bo *nvbo = nouveau_bo(bo); + struct ttm_mem_reg *old_mem = &bo->mem; + struct nouveau_channel *chan; + int ret; + + chan = nvbo->channel; + if (!chan) { + chan = dev_priv->channel; + mutex_lock_nested(&chan->mutex, NOUVEAU_KCHANNEL_MUTEX); + } + + /* create temporary vma for old memory, this will get cleaned + * up after ttm destroys the ttm_mem_reg + */ + if (dev_priv->card_type >= NV_50) { + struct nouveau_mem *node = old_mem->mm_node; + if (!node->tmp_vma.node) { + u32 page_shift = nvbo->vma.node->type; + if (old_mem->mem_type == TTM_PL_TT) + page_shift = nvbo->vma.vm->spg_shift; + + ret = nouveau_vm_get(chan->vm, + old_mem->num_pages << PAGE_SHIFT, + page_shift, NV_MEM_ACCESS_RO, + &node->tmp_vma); + if (ret) + goto out; + } + + if (old_mem->mem_type == TTM_PL_VRAM) + nouveau_vm_map(&node->tmp_vma, node); + else { + nouveau_vm_map_sg(&node->tmp_vma, 0, + old_mem->num_pages << PAGE_SHIFT, + node, node->pages); + } + } + + if (dev_priv->card_type < NV_50) + ret = nv04_bo_move_m2mf(chan, bo, &bo->mem, new_mem); + else + if (dev_priv->card_type < NV_C0) + ret = nv50_bo_move_m2mf(chan, bo, &bo->mem, new_mem); + else + ret = nvc0_bo_move_m2mf(chan, bo, &bo->mem, new_mem); + if (ret == 0) { + ret = nouveau_bo_move_accel_cleanup(chan, nvbo, evict, + no_wait_reserve, + no_wait_gpu, new_mem); + } + +out: + if (chan == dev_priv->channel) + mutex_unlock(&chan->mutex); + return ret; +} + +static int +nouveau_bo_move_flipd(struct ttm_buffer_object *bo, bool evict, bool intr, + bool no_wait_reserve, bool no_wait_gpu, + struct ttm_mem_reg *new_mem) +{ + struct drm_nouveau_private *dev_priv = nouveau_bdev(bo->bdev); + u32 placement_memtype = TTM_PL_FLAG_TT | TTM_PL_MASK_CACHING; + struct ttm_placement placement; + struct ttm_mem_reg tmp_mem; + int ret; + + placement.fpfn = placement.lpfn = 0; + placement.num_placement = placement.num_busy_placement = 1; + placement.placement = placement.busy_placement = &placement_memtype; + + tmp_mem = *new_mem; + tmp_mem.mm_node = NULL; + ret = ttm_bo_mem_space(bo, &placement, &tmp_mem, intr, no_wait_reserve, no_wait_gpu); + if (ret) + return ret; + + ret = ttm_tt_bind(bo->ttm, &tmp_mem); + if (ret) + goto out; + + if (dev_priv->card_type >= NV_50) { + struct nouveau_bo *nvbo = nouveau_bo(bo); + struct nouveau_mem *node = tmp_mem.mm_node; + struct nouveau_vma *vma = &nvbo->vma; + if (vma->node->type != vma->vm->spg_shift) + vma = &node->tmp_vma; + nouveau_vm_map_sg(vma, 0, tmp_mem.num_pages << PAGE_SHIFT, + node, node->pages); + } + + ret = nouveau_bo_move_m2mf(bo, true, intr, no_wait_reserve, no_wait_gpu, &tmp_mem); + + if (dev_priv->card_type >= NV_50) { + struct nouveau_bo *nvbo = nouveau_bo(bo); + nouveau_vm_unmap(&nvbo->vma); + } + + if (ret) + goto out; + + ret = ttm_bo_move_ttm(bo, true, no_wait_reserve, no_wait_gpu, new_mem); +out: + ttm_bo_mem_put(bo, &tmp_mem); + return ret; +} + +static int +nouveau_bo_move_flips(struct ttm_buffer_object *bo, bool evict, bool intr, + bool no_wait_reserve, bool no_wait_gpu, + struct ttm_mem_reg *new_mem) +{ + u32 placement_memtype = TTM_PL_FLAG_TT | TTM_PL_MASK_CACHING; + struct ttm_placement placement; + struct ttm_mem_reg tmp_mem; + int ret; + + placement.fpfn = placement.lpfn = 0; + placement.num_placement = placement.num_busy_placement = 1; + placement.placement = placement.busy_placement = &placement_memtype; + + tmp_mem = *new_mem; + tmp_mem.mm_node = NULL; + ret = ttm_bo_mem_space(bo, &placement, &tmp_mem, intr, no_wait_reserve, no_wait_gpu); + if (ret) + return ret; + + ret = ttm_bo_move_ttm(bo, true, no_wait_reserve, no_wait_gpu, &tmp_mem); + if (ret) + goto out; + + ret = nouveau_bo_move_m2mf(bo, true, intr, no_wait_reserve, no_wait_gpu, new_mem); + if (ret) + goto out; + +out: + ttm_bo_mem_put(bo, &tmp_mem); + return ret; +} + +static void +nouveau_bo_move_ntfy(struct ttm_buffer_object *bo, struct ttm_mem_reg *new_mem) +{ + struct drm_nouveau_private *dev_priv = nouveau_bdev(bo->bdev); + struct nouveau_mem *node = new_mem->mm_node; + struct nouveau_bo *nvbo = nouveau_bo(bo); + struct nouveau_vma *vma = &nvbo->vma; + struct nouveau_vm *vm = vma->vm; + + if (dev_priv->card_type < NV_50) + return; + + switch (new_mem->mem_type) { + case TTM_PL_VRAM: + nouveau_vm_map(vma, node); + break; + case TTM_PL_TT: + if (vma->node->type != vm->spg_shift) { + nouveau_vm_unmap(vma); + vma = &node->tmp_vma; + } + nouveau_vm_map_sg(vma, 0, new_mem->num_pages << PAGE_SHIFT, + node, node->pages); + break; + default: + nouveau_vm_unmap(&nvbo->vma); + break; + } +} + +static int +nouveau_bo_vm_bind(struct ttm_buffer_object *bo, struct ttm_mem_reg *new_mem, + struct nouveau_tile_reg **new_tile) +{ + struct drm_nouveau_private *dev_priv = nouveau_bdev(bo->bdev); + struct drm_device *dev = dev_priv->dev; + struct nouveau_bo *nvbo = nouveau_bo(bo); + u64 offset = new_mem->start << PAGE_SHIFT; + + *new_tile = NULL; + if (new_mem->mem_type != TTM_PL_VRAM) + return 0; + + if (dev_priv->card_type >= NV_10) { + *new_tile = nv10_mem_set_tiling(dev, offset, new_mem->size, + nvbo->tile_mode, + nvbo->tile_flags); + } + + return 0; +} + +static void +nouveau_bo_vm_cleanup(struct ttm_buffer_object *bo, + struct nouveau_tile_reg *new_tile, + struct nouveau_tile_reg **old_tile) +{ + struct drm_nouveau_private *dev_priv = nouveau_bdev(bo->bdev); + struct drm_device *dev = dev_priv->dev; + + nv10_mem_put_tile_region(dev, *old_tile, bo->sync_obj); + *old_tile = new_tile; +} + +static int +nouveau_bo_move(struct ttm_buffer_object *bo, bool evict, bool intr, + bool no_wait_reserve, bool no_wait_gpu, + struct ttm_mem_reg *new_mem) +{ + struct drm_nouveau_private *dev_priv = nouveau_bdev(bo->bdev); + struct nouveau_bo *nvbo = nouveau_bo(bo); + struct ttm_mem_reg *old_mem = &bo->mem; + struct nouveau_tile_reg *new_tile = NULL; + int ret = 0; + + if (dev_priv->card_type < NV_50) { + ret = nouveau_bo_vm_bind(bo, new_mem, &new_tile); + if (ret) + return ret; + } + + /* Fake bo copy. */ + if (old_mem->mem_type == TTM_PL_SYSTEM && !bo->ttm) { + BUG_ON(bo->mem.mm_node != NULL); + bo->mem = *new_mem; + new_mem->mm_node = NULL; + goto out; + } + + /* Software copy if the card isn't up and running yet. */ + if (!dev_priv->channel) { + ret = ttm_bo_move_memcpy(bo, evict, no_wait_reserve, no_wait_gpu, new_mem); + goto out; + } + + /* Hardware assisted copy. */ + if (new_mem->mem_type == TTM_PL_SYSTEM) + ret = nouveau_bo_move_flipd(bo, evict, intr, no_wait_reserve, no_wait_gpu, new_mem); + else if (old_mem->mem_type == TTM_PL_SYSTEM) + ret = nouveau_bo_move_flips(bo, evict, intr, no_wait_reserve, no_wait_gpu, new_mem); + else + ret = nouveau_bo_move_m2mf(bo, evict, intr, no_wait_reserve, no_wait_gpu, new_mem); + + if (!ret) + goto out; + + /* Fallback to software copy. */ + ret = ttm_bo_move_memcpy(bo, evict, no_wait_reserve, no_wait_gpu, new_mem); + +out: + if (dev_priv->card_type < NV_50) { + if (ret) + nouveau_bo_vm_cleanup(bo, NULL, &new_tile); + else + nouveau_bo_vm_cleanup(bo, new_tile, &nvbo->tile); + } + + return ret; +} + +static int +nouveau_bo_verify_access(struct ttm_buffer_object *bo, struct file *filp) +{ + return 0; +} + +static int +nouveau_ttm_io_mem_reserve(struct ttm_bo_device *bdev, struct ttm_mem_reg *mem) +{ + struct ttm_mem_type_manager *man = &bdev->man[mem->mem_type]; + struct drm_nouveau_private *dev_priv = nouveau_bdev(bdev); + struct drm_device *dev = dev_priv->dev; + int ret; + + mem->bus.addr = NULL; + mem->bus.offset = 0; + mem->bus.size = mem->num_pages << PAGE_SHIFT; + mem->bus.base = 0; + mem->bus.is_iomem = false; + if (!(man->flags & TTM_MEMTYPE_FLAG_MAPPABLE)) + return -EINVAL; + switch (mem->mem_type) { + case TTM_PL_SYSTEM: + /* System memory */ + return 0; + case TTM_PL_TT: +#if __OS_HAS_AGP + if (dev_priv->gart_info.type == NOUVEAU_GART_AGP) { + mem->bus.offset = mem->start << PAGE_SHIFT; + mem->bus.base = dev_priv->gart_info.aper_base; + mem->bus.is_iomem = true; + } +#endif + break; + case TTM_PL_VRAM: + { + struct nouveau_mem *node = mem->mm_node; + u8 page_shift; + + if (!dev_priv->bar1_vm) { + mem->bus.offset = mem->start << PAGE_SHIFT; + mem->bus.base = pci_resource_start(dev->pdev, 1); + mem->bus.is_iomem = true; + break; + } + + if (dev_priv->card_type == NV_C0) + page_shift = node->page_shift; + else + page_shift = 12; + + ret = nouveau_vm_get(dev_priv->bar1_vm, mem->bus.size, + page_shift, NV_MEM_ACCESS_RW, + &node->bar_vma); + if (ret) + return ret; + + nouveau_vm_map(&node->bar_vma, node); + if (ret) { + nouveau_vm_put(&node->bar_vma); + return ret; + } + + mem->bus.offset = node->bar_vma.offset; + if (dev_priv->card_type == NV_50) /*XXX*/ + mem->bus.offset -= 0x0020000000ULL; + mem->bus.base = pci_resource_start(dev->pdev, 1); + mem->bus.is_iomem = true; + } + break; + default: + return -EINVAL; + } + return 0; +} + +static void +nouveau_ttm_io_mem_free(struct ttm_bo_device *bdev, struct ttm_mem_reg *mem) +{ + struct drm_nouveau_private *dev_priv = nouveau_bdev(bdev); + struct nouveau_mem *node = mem->mm_node; + + if (!dev_priv->bar1_vm || mem->mem_type != TTM_PL_VRAM) + return; + + if (!node->bar_vma.node) + return; + + nouveau_vm_unmap(&node->bar_vma); + nouveau_vm_put(&node->bar_vma); +} + +static int +nouveau_ttm_fault_reserve_notify(struct ttm_buffer_object *bo) +{ + struct drm_nouveau_private *dev_priv = nouveau_bdev(bo->bdev); + struct nouveau_bo *nvbo = nouveau_bo(bo); + + /* as long as the bo isn't in vram, and isn't tiled, we've got + * nothing to do here. + */ + if (bo->mem.mem_type != TTM_PL_VRAM) { + if (dev_priv->card_type < NV_50 || + !nouveau_bo_tile_layout(nvbo)) + return 0; + } + + /* make sure bo is in mappable vram */ + if (bo->mem.start + bo->mem.num_pages < dev_priv->fb_mappable_pages) + return 0; + + + nvbo->placement.fpfn = 0; + nvbo->placement.lpfn = dev_priv->fb_mappable_pages; + nouveau_bo_placement_set(nvbo, TTM_PL_FLAG_VRAM, 0); + return nouveau_bo_validate(nvbo, false, true, false); +} + +void +nouveau_bo_fence(struct nouveau_bo *nvbo, struct nouveau_fence *fence) +{ + struct nouveau_fence *old_fence; + + if (likely(fence)) + nouveau_fence_ref(fence); + + spin_lock(&nvbo->bo.bdev->fence_lock); + old_fence = nvbo->bo.sync_obj; + nvbo->bo.sync_obj = fence; + spin_unlock(&nvbo->bo.bdev->fence_lock); + + nouveau_fence_unref(&old_fence); +} + +struct ttm_bo_driver nouveau_bo_driver = { + .create_ttm_backend_entry = nouveau_bo_create_ttm_backend_entry, + .invalidate_caches = nouveau_bo_invalidate_caches, + .init_mem_type = nouveau_bo_init_mem_type, + .evict_flags = nouveau_bo_evict_flags, + .move_notify = nouveau_bo_move_ntfy, + .move = nouveau_bo_move, + .verify_access = nouveau_bo_verify_access, + .sync_obj_signaled = __nouveau_fence_signalled, + .sync_obj_wait = __nouveau_fence_wait, + .sync_obj_flush = __nouveau_fence_flush, + .sync_obj_unref = __nouveau_fence_unref, + .sync_obj_ref = __nouveau_fence_ref, + .fault_reserve_notify = &nouveau_ttm_fault_reserve_notify, + .io_mem_reserve = &nouveau_ttm_io_mem_reserve, + .io_mem_free = &nouveau_ttm_io_mem_free, +}; + diff --git a/drivers/gpu/drm/nouveau/nouveau_calc.c b/drivers/gpu/drm/nouveau/nouveau_calc.c new file mode 100644 index 00000000..dad96cce --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_calc.c @@ -0,0 +1,478 @@ +/* + * Copyright 1993-2003 NVIDIA, Corporation + * Copyright 2007-2009 Stuart Bennett + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#include "drmP.h" +#include "nouveau_drv.h" +#include "nouveau_hw.h" + +/****************************************************************************\ +* * +* The video arbitration routines calculate some "magic" numbers. Fixes * +* the snow seen when accessing the framebuffer without it. * +* It just works (I hope). * +* * +\****************************************************************************/ + +struct nv_fifo_info { + int lwm; + int burst; +}; + +struct nv_sim_state { + int pclk_khz; + int mclk_khz; + int nvclk_khz; + int bpp; + int mem_page_miss; + int mem_latency; + int memory_type; + int memory_width; + int two_heads; +}; + +static void +nv04_calc_arb(struct nv_fifo_info *fifo, struct nv_sim_state *arb) +{ + int pagemiss, cas, width, bpp; + int nvclks, mclks, pclks, crtpagemiss; + int found, mclk_extra, mclk_loop, cbs, m1, p1; + int mclk_freq, pclk_freq, nvclk_freq; + int us_m, us_n, us_p, crtc_drain_rate; + int cpm_us, us_crt, clwm; + + pclk_freq = arb->pclk_khz; + mclk_freq = arb->mclk_khz; + nvclk_freq = arb->nvclk_khz; + pagemiss = arb->mem_page_miss; + cas = arb->mem_latency; + width = arb->memory_width >> 6; + bpp = arb->bpp; + cbs = 128; + + pclks = 2; + nvclks = 10; + mclks = 13 + cas; + mclk_extra = 3; + found = 0; + + while (!found) { + found = 1; + + mclk_loop = mclks + mclk_extra; + us_m = mclk_loop * 1000 * 1000 / mclk_freq; + us_n = nvclks * 1000 * 1000 / nvclk_freq; + us_p = nvclks * 1000 * 1000 / pclk_freq; + + crtc_drain_rate = pclk_freq * bpp / 8; + crtpagemiss = 2; + crtpagemiss += 1; + cpm_us = crtpagemiss * pagemiss * 1000 * 1000 / mclk_freq; + us_crt = cpm_us + us_m + us_n + us_p; + clwm = us_crt * crtc_drain_rate / (1000 * 1000); + clwm++; + + m1 = clwm + cbs - 512; + p1 = m1 * pclk_freq / mclk_freq; + p1 = p1 * bpp / 8; + if ((p1 < m1 && m1 > 0) || clwm > 519) { + found = !mclk_extra; + mclk_extra--; + } + if (clwm < 384) + clwm = 384; + + fifo->lwm = clwm; + fifo->burst = cbs; + } +} + +static void +nv10_calc_arb(struct nv_fifo_info *fifo, struct nv_sim_state *arb) +{ + int fill_rate, drain_rate; + int pclks, nvclks, mclks, xclks; + int pclk_freq, nvclk_freq, mclk_freq; + int fill_lat, extra_lat; + int max_burst_o, max_burst_l; + int fifo_len, min_lwm, max_lwm; + const int burst_lat = 80; /* Maximum allowable latency due + * to the CRTC FIFO burst. (ns) */ + + pclk_freq = arb->pclk_khz; + nvclk_freq = arb->nvclk_khz; + mclk_freq = arb->mclk_khz; + + fill_rate = mclk_freq * arb->memory_width / 8; /* kB/s */ + drain_rate = pclk_freq * arb->bpp / 8; /* kB/s */ + + fifo_len = arb->two_heads ? 1536 : 1024; /* B */ + + /* Fixed FIFO refill latency. */ + + pclks = 4; /* lwm detect. */ + + nvclks = 3 /* lwm -> sync. */ + + 2 /* fbi bus cycles (1 req + 1 busy) */ + + 1 /* 2 edge sync. may be very close to edge so + * just put one. */ + + 1 /* fbi_d_rdv_n */ + + 1 /* Fbi_d_rdata */ + + 1; /* crtfifo load */ + + mclks = 1 /* 2 edge sync. may be very close to edge so + * just put one. */ + + 1 /* arb_hp_req */ + + 5 /* tiling pipeline */ + + 2 /* latency fifo */ + + 2 /* memory request to fbio block */ + + 7; /* data returned from fbio block */ + + /* Need to accumulate 256 bits for read */ + mclks += (arb->memory_type == 0 ? 2 : 1) + * arb->memory_width / 32; + + fill_lat = mclks * 1000 * 1000 / mclk_freq /* minimum mclk latency */ + + nvclks * 1000 * 1000 / nvclk_freq /* nvclk latency */ + + pclks * 1000 * 1000 / pclk_freq; /* pclk latency */ + + /* Conditional FIFO refill latency. */ + + xclks = 2 * arb->mem_page_miss + mclks /* Extra latency due to + * the overlay. */ + + 2 * arb->mem_page_miss /* Extra pagemiss latency. */ + + (arb->bpp == 32 ? 8 : 4); /* Margin of error. */ + + extra_lat = xclks * 1000 * 1000 / mclk_freq; + + if (arb->two_heads) + /* Account for another CRTC. */ + extra_lat += fill_lat + extra_lat + burst_lat; + + /* FIFO burst */ + + /* Max burst not leading to overflows. */ + max_burst_o = (1 + fifo_len - extra_lat * drain_rate / (1000 * 1000)) + * (fill_rate / 1000) / ((fill_rate - drain_rate) / 1000); + fifo->burst = min(max_burst_o, 1024); + + /* Max burst value with an acceptable latency. */ + max_burst_l = burst_lat * fill_rate / (1000 * 1000); + fifo->burst = min(max_burst_l, fifo->burst); + + fifo->burst = rounddown_pow_of_two(fifo->burst); + + /* FIFO low watermark */ + + min_lwm = (fill_lat + extra_lat) * drain_rate / (1000 * 1000) + 1; + max_lwm = fifo_len - fifo->burst + + fill_lat * drain_rate / (1000 * 1000) + + fifo->burst * drain_rate / fill_rate; + + fifo->lwm = min_lwm + 10 * (max_lwm - min_lwm) / 100; /* Empirical. */ +} + +static void +nv04_update_arb(struct drm_device *dev, int VClk, int bpp, + int *burst, int *lwm) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nv_fifo_info fifo_data; + struct nv_sim_state sim_data; + int MClk = nouveau_hw_get_clock(dev, PLL_MEMORY); + int NVClk = nouveau_hw_get_clock(dev, PLL_CORE); + uint32_t cfg1 = nvReadFB(dev, NV04_PFB_CFG1); + + sim_data.pclk_khz = VClk; + sim_data.mclk_khz = MClk; + sim_data.nvclk_khz = NVClk; + sim_data.bpp = bpp; + sim_data.two_heads = nv_two_heads(dev); + if ((dev->pci_device & 0xffff) == 0x01a0 /*CHIPSET_NFORCE*/ || + (dev->pci_device & 0xffff) == 0x01f0 /*CHIPSET_NFORCE2*/) { + uint32_t type; + + pci_read_config_dword(pci_get_bus_and_slot(0, 1), 0x7c, &type); + + sim_data.memory_type = (type >> 12) & 1; + sim_data.memory_width = 64; + sim_data.mem_latency = 3; + sim_data.mem_page_miss = 10; + } else { + sim_data.memory_type = nvReadFB(dev, NV04_PFB_CFG0) & 0x1; + sim_data.memory_width = (nvReadEXTDEV(dev, NV_PEXTDEV_BOOT_0) & 0x10) ? 128 : 64; + sim_data.mem_latency = cfg1 & 0xf; + sim_data.mem_page_miss = ((cfg1 >> 4) & 0xf) + ((cfg1 >> 31) & 0x1); + } + + if (dev_priv->card_type == NV_04) + nv04_calc_arb(&fifo_data, &sim_data); + else + nv10_calc_arb(&fifo_data, &sim_data); + + *burst = ilog2(fifo_data.burst >> 4); + *lwm = fifo_data.lwm >> 3; +} + +static void +nv20_update_arb(int *burst, int *lwm) +{ + unsigned int fifo_size, burst_size, graphics_lwm; + + fifo_size = 2048; + burst_size = 512; + graphics_lwm = fifo_size - burst_size; + + *burst = ilog2(burst_size >> 5); + *lwm = graphics_lwm >> 3; +} + +void +nouveau_calc_arb(struct drm_device *dev, int vclk, int bpp, int *burst, int *lwm) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + if (dev_priv->card_type < NV_20) + nv04_update_arb(dev, vclk, bpp, burst, lwm); + else if ((dev->pci_device & 0xfff0) == 0x0240 /*CHIPSET_C51*/ || + (dev->pci_device & 0xfff0) == 0x03d0 /*CHIPSET_C512*/) { + *burst = 128; + *lwm = 0x0480; + } else + nv20_update_arb(burst, lwm); +} + +static int +getMNP_single(struct drm_device *dev, struct pll_lims *pll_lim, int clk, + struct nouveau_pll_vals *bestpv) +{ + /* Find M, N and P for a single stage PLL + * + * Note that some bioses (NV3x) have lookup tables of precomputed MNP + * values, but we're too lazy to use those atm + * + * "clk" parameter in kHz + * returns calculated clock + */ + struct drm_nouveau_private *dev_priv = dev->dev_private; + int cv = dev_priv->vbios.chip_version; + int minvco = pll_lim->vco1.minfreq, maxvco = pll_lim->vco1.maxfreq; + int minM = pll_lim->vco1.min_m, maxM = pll_lim->vco1.max_m; + int minN = pll_lim->vco1.min_n, maxN = pll_lim->vco1.max_n; + int minU = pll_lim->vco1.min_inputfreq; + int maxU = pll_lim->vco1.max_inputfreq; + int minP = pll_lim->max_p ? pll_lim->min_p : 0; + int maxP = pll_lim->max_p ? pll_lim->max_p : pll_lim->max_usable_log2p; + int crystal = pll_lim->refclk; + int M, N, thisP, P; + int clkP, calcclk; + int delta, bestdelta = INT_MAX; + int bestclk = 0; + + /* this division verified for nv20, nv18, nv28 (Haiku), and nv34 */ + /* possibly correlated with introduction of 27MHz crystal */ + if (dev_priv->card_type < NV_50) { + if (cv < 0x17 || cv == 0x1a || cv == 0x20) { + if (clk > 250000) + maxM = 6; + if (clk > 340000) + maxM = 2; + } else if (cv < 0x40) { + if (clk > 150000) + maxM = 6; + if (clk > 200000) + maxM = 4; + if (clk > 340000) + maxM = 2; + } + } + + P = pll_lim->max_p ? maxP : (1 << maxP); + if ((clk * P) < minvco) { + minvco = clk * maxP; + maxvco = minvco * 2; + } + + if (clk + clk/200 > maxvco) /* +0.5% */ + maxvco = clk + clk/200; + + /* NV34 goes maxlog2P->0, NV20 goes 0->maxlog2P */ + for (thisP = minP; thisP <= maxP; thisP++) { + P = pll_lim->max_p ? thisP : (1 << thisP); + clkP = clk * P; + + if (clkP < minvco) + continue; + if (clkP > maxvco) + return bestclk; + + for (M = minM; M <= maxM; M++) { + if (crystal/M < minU) + return bestclk; + if (crystal/M > maxU) + continue; + + /* add crystal/2 to round better */ + N = (clkP * M + crystal/2) / crystal; + + if (N < minN) + continue; + if (N > maxN) + break; + + /* more rounding additions */ + calcclk = ((N * crystal + P/2) / P + M/2) / M; + delta = abs(calcclk - clk); + /* we do an exhaustive search rather than terminating + * on an optimality condition... + */ + if (delta < bestdelta) { + bestdelta = delta; + bestclk = calcclk; + bestpv->N1 = N; + bestpv->M1 = M; + bestpv->log2P = thisP; + if (delta == 0) /* except this one */ + return bestclk; + } + } + } + + return bestclk; +} + +static int +getMNP_double(struct drm_device *dev, struct pll_lims *pll_lim, int clk, + struct nouveau_pll_vals *bestpv) +{ + /* Find M, N and P for a two stage PLL + * + * Note that some bioses (NV30+) have lookup tables of precomputed MNP + * values, but we're too lazy to use those atm + * + * "clk" parameter in kHz + * returns calculated clock + */ + struct drm_nouveau_private *dev_priv = dev->dev_private; + int chip_version = dev_priv->vbios.chip_version; + int minvco1 = pll_lim->vco1.minfreq, maxvco1 = pll_lim->vco1.maxfreq; + int minvco2 = pll_lim->vco2.minfreq, maxvco2 = pll_lim->vco2.maxfreq; + int minU1 = pll_lim->vco1.min_inputfreq, minU2 = pll_lim->vco2.min_inputfreq; + int maxU1 = pll_lim->vco1.max_inputfreq, maxU2 = pll_lim->vco2.max_inputfreq; + int minM1 = pll_lim->vco1.min_m, maxM1 = pll_lim->vco1.max_m; + int minN1 = pll_lim->vco1.min_n, maxN1 = pll_lim->vco1.max_n; + int minM2 = pll_lim->vco2.min_m, maxM2 = pll_lim->vco2.max_m; + int minN2 = pll_lim->vco2.min_n, maxN2 = pll_lim->vco2.max_n; + int maxlog2P = pll_lim->max_usable_log2p; + int crystal = pll_lim->refclk; + bool fixedgain2 = (minM2 == maxM2 && minN2 == maxN2); + int M1, N1, M2, N2, log2P; + int clkP, calcclk1, calcclk2, calcclkout; + int delta, bestdelta = INT_MAX; + int bestclk = 0; + + int vco2 = (maxvco2 - maxvco2/200) / 2; + for (log2P = 0; clk && log2P < maxlog2P && clk <= (vco2 >> log2P); log2P++) + ; + clkP = clk << log2P; + + if (maxvco2 < clk + clk/200) /* +0.5% */ + maxvco2 = clk + clk/200; + + for (M1 = minM1; M1 <= maxM1; M1++) { + if (crystal/M1 < minU1) + return bestclk; + if (crystal/M1 > maxU1) + continue; + + for (N1 = minN1; N1 <= maxN1; N1++) { + calcclk1 = crystal * N1 / M1; + if (calcclk1 < minvco1) + continue; + if (calcclk1 > maxvco1) + break; + + for (M2 = minM2; M2 <= maxM2; M2++) { + if (calcclk1/M2 < minU2) + break; + if (calcclk1/M2 > maxU2) + continue; + + /* add calcclk1/2 to round better */ + N2 = (clkP * M2 + calcclk1/2) / calcclk1; + if (N2 < minN2) + continue; + if (N2 > maxN2) + break; + + if (!fixedgain2) { + if (chip_version < 0x60) + if (N2/M2 < 4 || N2/M2 > 10) + continue; + + calcclk2 = calcclk1 * N2 / M2; + if (calcclk2 < minvco2) + break; + if (calcclk2 > maxvco2) + continue; + } else + calcclk2 = calcclk1; + + calcclkout = calcclk2 >> log2P; + delta = abs(calcclkout - clk); + /* we do an exhaustive search rather than terminating + * on an optimality condition... + */ + if (delta < bestdelta) { + bestdelta = delta; + bestclk = calcclkout; + bestpv->N1 = N1; + bestpv->M1 = M1; + bestpv->N2 = N2; + bestpv->M2 = M2; + bestpv->log2P = log2P; + if (delta == 0) /* except this one */ + return bestclk; + } + } + } + } + + return bestclk; +} + +int +nouveau_calc_pll_mnp(struct drm_device *dev, struct pll_lims *pll_lim, int clk, + struct nouveau_pll_vals *pv) +{ + int outclk; + + if (!pll_lim->vco2.maxfreq) + outclk = getMNP_single(dev, pll_lim, clk, pv); + else + outclk = getMNP_double(dev, pll_lim, clk, pv); + + if (!outclk) + NV_ERROR(dev, "Could not find a compatible set of PLL values\n"); + + return outclk; +} diff --git a/drivers/gpu/drm/nouveau/nouveau_channel.c b/drivers/gpu/drm/nouveau/nouveau_channel.c new file mode 100644 index 00000000..d31d355f --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_channel.c @@ -0,0 +1,491 @@ +/* + * Copyright 2005-2006 Stephane Marchesin + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + */ + +#include "drmP.h" +#include "drm.h" +#include "nouveau_drv.h" +#include "nouveau_drm.h" +#include "nouveau_dma.h" + +static int +nouveau_channel_pushbuf_ctxdma_init(struct nouveau_channel *chan) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_bo *pb = chan->pushbuf_bo; + struct nouveau_gpuobj *pushbuf = NULL; + int ret = 0; + + if (dev_priv->card_type >= NV_50) { + if (dev_priv->card_type < NV_C0) { + ret = nouveau_gpuobj_dma_new(chan, + NV_CLASS_DMA_IN_MEMORY, 0, + (1ULL << 40), + NV_MEM_ACCESS_RO, + NV_MEM_TARGET_VM, + &pushbuf); + } + chan->pushbuf_base = pb->bo.offset; + } else + if (pb->bo.mem.mem_type == TTM_PL_TT) { + ret = nouveau_gpuobj_dma_new(chan, NV_CLASS_DMA_IN_MEMORY, 0, + dev_priv->gart_info.aper_size, + NV_MEM_ACCESS_RO, + NV_MEM_TARGET_GART, &pushbuf); + chan->pushbuf_base = pb->bo.mem.start << PAGE_SHIFT; + } else + if (dev_priv->card_type != NV_04) { + ret = nouveau_gpuobj_dma_new(chan, NV_CLASS_DMA_IN_MEMORY, 0, + dev_priv->fb_available_size, + NV_MEM_ACCESS_RO, + NV_MEM_TARGET_VRAM, &pushbuf); + chan->pushbuf_base = pb->bo.mem.start << PAGE_SHIFT; + } else { + /* NV04 cmdbuf hack, from original ddx.. not sure of it's + * exact reason for existing :) PCI access to cmdbuf in + * VRAM. + */ + ret = nouveau_gpuobj_dma_new(chan, NV_CLASS_DMA_IN_MEMORY, + pci_resource_start(dev->pdev, 1), + dev_priv->fb_available_size, + NV_MEM_ACCESS_RO, + NV_MEM_TARGET_PCI, &pushbuf); + chan->pushbuf_base = pb->bo.mem.start << PAGE_SHIFT; + } + + nouveau_gpuobj_ref(pushbuf, &chan->pushbuf); + nouveau_gpuobj_ref(NULL, &pushbuf); + return ret; +} + +static struct nouveau_bo * +nouveau_channel_user_pushbuf_alloc(struct drm_device *dev) +{ + struct nouveau_bo *pushbuf = NULL; + int location, ret; + + if (nouveau_vram_pushbuf) + location = TTM_PL_FLAG_VRAM; + else + location = TTM_PL_FLAG_TT; + + ret = nouveau_bo_new(dev, NULL, 65536, 0, location, 0, 0x0000, &pushbuf); + if (ret) { + NV_ERROR(dev, "error allocating DMA push buffer: %d\n", ret); + return NULL; + } + + ret = nouveau_bo_pin(pushbuf, location); + if (ret) { + NV_ERROR(dev, "error pinning DMA push buffer: %d\n", ret); + nouveau_bo_ref(NULL, &pushbuf); + return NULL; + } + + ret = nouveau_bo_map(pushbuf); + if (ret) { + nouveau_bo_unpin(pushbuf); + nouveau_bo_ref(NULL, &pushbuf); + return NULL; + } + + return pushbuf; +} + +/* allocates and initializes a fifo for user space consumption */ +int +nouveau_channel_alloc(struct drm_device *dev, struct nouveau_channel **chan_ret, + struct drm_file *file_priv, + uint32_t vram_handle, uint32_t gart_handle) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; + struct nouveau_channel *chan; + unsigned long flags; + int ret; + + /* allocate and lock channel structure */ + chan = kzalloc(sizeof(*chan), GFP_KERNEL); + if (!chan) + return -ENOMEM; + chan->dev = dev; + chan->file_priv = file_priv; + chan->vram_handle = vram_handle; + chan->gart_handle = gart_handle; + + kref_init(&chan->ref); + atomic_set(&chan->users, 1); + mutex_init(&chan->mutex); + mutex_lock(&chan->mutex); + + /* allocate hw channel id */ + spin_lock_irqsave(&dev_priv->channels.lock, flags); + for (chan->id = 0; chan->id < pfifo->channels; chan->id++) { + if (!dev_priv->channels.ptr[chan->id]) { + nouveau_channel_ref(chan, &dev_priv->channels.ptr[chan->id]); + break; + } + } + spin_unlock_irqrestore(&dev_priv->channels.lock, flags); + + if (chan->id == pfifo->channels) { + mutex_unlock(&chan->mutex); + kfree(chan); + return -ENODEV; + } + + NV_DEBUG(dev, "initialising channel %d\n", chan->id); + INIT_LIST_HEAD(&chan->nvsw.vbl_wait); + INIT_LIST_HEAD(&chan->nvsw.flip); + INIT_LIST_HEAD(&chan->fence.pending); + spin_lock_init(&chan->fence.lock); + + /* Allocate DMA push buffer */ + chan->pushbuf_bo = nouveau_channel_user_pushbuf_alloc(dev); + if (!chan->pushbuf_bo) { + ret = -ENOMEM; + NV_ERROR(dev, "pushbuf %d\n", ret); + nouveau_channel_put(&chan); + return ret; + } + + nouveau_dma_pre_init(chan); + chan->user_put = 0x40; + chan->user_get = 0x44; + + /* Allocate space for per-channel fixed notifier memory */ + ret = nouveau_notifier_init_channel(chan); + if (ret) { + NV_ERROR(dev, "ntfy %d\n", ret); + nouveau_channel_put(&chan); + return ret; + } + + /* Setup channel's default objects */ + ret = nouveau_gpuobj_channel_init(chan, vram_handle, gart_handle); + if (ret) { + NV_ERROR(dev, "gpuobj %d\n", ret); + nouveau_channel_put(&chan); + return ret; + } + + /* Create a dma object for the push buffer */ + ret = nouveau_channel_pushbuf_ctxdma_init(chan); + if (ret) { + NV_ERROR(dev, "pbctxdma %d\n", ret); + nouveau_channel_put(&chan); + return ret; + } + + /* disable the fifo caches */ + pfifo->reassign(dev, false); + + /* Construct initial RAMFC for new channel */ + ret = pfifo->create_context(chan); + if (ret) { + nouveau_channel_put(&chan); + return ret; + } + + pfifo->reassign(dev, true); + + ret = nouveau_dma_init(chan); + if (!ret) + ret = nouveau_fence_channel_init(chan); + if (ret) { + nouveau_channel_put(&chan); + return ret; + } + + nouveau_debugfs_channel_init(chan); + + NV_DEBUG(dev, "channel %d initialised\n", chan->id); + *chan_ret = chan; + return 0; +} + +struct nouveau_channel * +nouveau_channel_get_unlocked(struct nouveau_channel *ref) +{ + struct nouveau_channel *chan = NULL; + + if (likely(ref && atomic_inc_not_zero(&ref->users))) + nouveau_channel_ref(ref, &chan); + + return chan; +} + +struct nouveau_channel * +nouveau_channel_get(struct drm_device *dev, struct drm_file *file_priv, int id) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_channel *chan; + unsigned long flags; + + if (unlikely(id < 0 || id >= NOUVEAU_MAX_CHANNEL_NR)) + return ERR_PTR(-EINVAL); + + spin_lock_irqsave(&dev_priv->channels.lock, flags); + chan = nouveau_channel_get_unlocked(dev_priv->channels.ptr[id]); + spin_unlock_irqrestore(&dev_priv->channels.lock, flags); + + if (unlikely(!chan)) + return ERR_PTR(-EINVAL); + + if (unlikely(file_priv && chan->file_priv != file_priv)) { + nouveau_channel_put_unlocked(&chan); + return ERR_PTR(-EINVAL); + } + + mutex_lock(&chan->mutex); + return chan; +} + +void +nouveau_channel_put_unlocked(struct nouveau_channel **pchan) +{ + struct nouveau_channel *chan = *pchan; + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; + unsigned long flags; + int i; + + /* decrement the refcount, and we're done if there's still refs */ + if (likely(!atomic_dec_and_test(&chan->users))) { + nouveau_channel_ref(NULL, pchan); + return; + } + + /* no one wants the channel anymore */ + NV_DEBUG(dev, "freeing channel %d\n", chan->id); + nouveau_debugfs_channel_fini(chan); + + /* give it chance to idle */ + nouveau_channel_idle(chan); + + /* ensure all outstanding fences are signaled. they should be if the + * above attempts at idling were OK, but if we failed this'll tell TTM + * we're done with the buffers. + */ + nouveau_fence_channel_fini(chan); + + /* boot it off the hardware */ + pfifo->reassign(dev, false); + + /* destroy the engine specific contexts */ + pfifo->destroy_context(chan); + for (i = 0; i < NVOBJ_ENGINE_NR; i++) { + if (chan->engctx[i]) + dev_priv->eng[i]->context_del(chan, i); + } + + pfifo->reassign(dev, true); + + /* aside from its resources, the channel should now be dead, + * remove it from the channel list + */ + spin_lock_irqsave(&dev_priv->channels.lock, flags); + nouveau_channel_ref(NULL, &dev_priv->channels.ptr[chan->id]); + spin_unlock_irqrestore(&dev_priv->channels.lock, flags); + + /* destroy any resources the channel owned */ + nouveau_gpuobj_ref(NULL, &chan->pushbuf); + if (chan->pushbuf_bo) { + nouveau_bo_unmap(chan->pushbuf_bo); + nouveau_bo_unpin(chan->pushbuf_bo); + nouveau_bo_ref(NULL, &chan->pushbuf_bo); + } + nouveau_gpuobj_channel_takedown(chan); + nouveau_notifier_takedown_channel(chan); + + nouveau_channel_ref(NULL, pchan); +} + +void +nouveau_channel_put(struct nouveau_channel **pchan) +{ + mutex_unlock(&(*pchan)->mutex); + nouveau_channel_put_unlocked(pchan); +} + +static void +nouveau_channel_del(struct kref *ref) +{ + struct nouveau_channel *chan = + container_of(ref, struct nouveau_channel, ref); + + kfree(chan); +} + +void +nouveau_channel_ref(struct nouveau_channel *chan, + struct nouveau_channel **pchan) +{ + if (chan) + kref_get(&chan->ref); + + if (*pchan) + kref_put(&(*pchan)->ref, nouveau_channel_del); + + *pchan = chan; +} + +void +nouveau_channel_idle(struct nouveau_channel *chan) +{ + struct drm_device *dev = chan->dev; + struct nouveau_fence *fence = NULL; + int ret; + + nouveau_fence_update(chan); + + if (chan->fence.sequence != chan->fence.sequence_ack) { + ret = nouveau_fence_new(chan, &fence, true); + if (!ret) { + ret = nouveau_fence_wait(fence, false, false); + nouveau_fence_unref(&fence); + } + + if (ret) + NV_ERROR(dev, "Failed to idle channel %d.\n", chan->id); + } +} + +/* cleans up all the fifos from file_priv */ +void +nouveau_channel_cleanup(struct drm_device *dev, struct drm_file *file_priv) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_engine *engine = &dev_priv->engine; + struct nouveau_channel *chan; + int i; + + NV_DEBUG(dev, "clearing FIFO enables from file_priv\n"); + for (i = 0; i < engine->fifo.channels; i++) { + chan = nouveau_channel_get(dev, file_priv, i); + if (IS_ERR(chan)) + continue; + + atomic_dec(&chan->users); + nouveau_channel_put(&chan); + } +} + + +/*********************************** + * ioctls wrapping the functions + ***********************************/ + +static int +nouveau_ioctl_fifo_alloc(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct drm_nouveau_channel_alloc *init = data; + struct nouveau_channel *chan; + int ret; + + if (!dev_priv->eng[NVOBJ_ENGINE_GR]) + return -ENODEV; + + if (init->fb_ctxdma_handle == ~0 || init->tt_ctxdma_handle == ~0) + return -EINVAL; + + ret = nouveau_channel_alloc(dev, &chan, file_priv, + init->fb_ctxdma_handle, + init->tt_ctxdma_handle); + if (ret) + return ret; + init->channel = chan->id; + + if (chan->dma.ib_max) + init->pushbuf_domains = NOUVEAU_GEM_DOMAIN_VRAM | + NOUVEAU_GEM_DOMAIN_GART; + else if (chan->pushbuf_bo->bo.mem.mem_type == TTM_PL_VRAM) + init->pushbuf_domains = NOUVEAU_GEM_DOMAIN_VRAM; + else + init->pushbuf_domains = NOUVEAU_GEM_DOMAIN_GART; + + if (dev_priv->card_type < NV_C0) { + init->subchan[0].handle = NvM2MF; + if (dev_priv->card_type < NV_50) + init->subchan[0].grclass = 0x0039; + else + init->subchan[0].grclass = 0x5039; + init->subchan[1].handle = NvSw; + init->subchan[1].grclass = NV_SW; + init->nr_subchan = 2; + } else { + init->subchan[0].handle = 0x9039; + init->subchan[0].grclass = 0x9039; + init->nr_subchan = 1; + } + + /* Named memory object area */ + ret = drm_gem_handle_create(file_priv, chan->notifier_bo->gem, + &init->notifier_handle); + + if (ret == 0) + atomic_inc(&chan->users); /* userspace reference */ + nouveau_channel_put(&chan); + return ret; +} + +static int +nouveau_ioctl_fifo_free(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_nouveau_channel_free *req = data; + struct nouveau_channel *chan; + + chan = nouveau_channel_get(dev, file_priv, req->channel); + if (IS_ERR(chan)) + return PTR_ERR(chan); + + atomic_dec(&chan->users); + nouveau_channel_put(&chan); + return 0; +} + +/*********************************** + * finally, the ioctl table + ***********************************/ + +struct drm_ioctl_desc nouveau_ioctls[] = { + DRM_IOCTL_DEF_DRV(NOUVEAU_GETPARAM, nouveau_ioctl_getparam, DRM_UNLOCKED|DRM_AUTH), + DRM_IOCTL_DEF_DRV(NOUVEAU_SETPARAM, nouveau_ioctl_setparam, DRM_UNLOCKED|DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF_DRV(NOUVEAU_CHANNEL_ALLOC, nouveau_ioctl_fifo_alloc, DRM_UNLOCKED|DRM_AUTH), + DRM_IOCTL_DEF_DRV(NOUVEAU_CHANNEL_FREE, nouveau_ioctl_fifo_free, DRM_UNLOCKED|DRM_AUTH), + DRM_IOCTL_DEF_DRV(NOUVEAU_GROBJ_ALLOC, nouveau_ioctl_grobj_alloc, DRM_UNLOCKED|DRM_AUTH), + DRM_IOCTL_DEF_DRV(NOUVEAU_NOTIFIEROBJ_ALLOC, nouveau_ioctl_notifier_alloc, DRM_UNLOCKED|DRM_AUTH), + DRM_IOCTL_DEF_DRV(NOUVEAU_GPUOBJ_FREE, nouveau_ioctl_gpuobj_free, DRM_UNLOCKED|DRM_AUTH), + DRM_IOCTL_DEF_DRV(NOUVEAU_GEM_NEW, nouveau_gem_ioctl_new, DRM_UNLOCKED|DRM_AUTH), + DRM_IOCTL_DEF_DRV(NOUVEAU_GEM_PUSHBUF, nouveau_gem_ioctl_pushbuf, DRM_UNLOCKED|DRM_AUTH), + DRM_IOCTL_DEF_DRV(NOUVEAU_GEM_CPU_PREP, nouveau_gem_ioctl_cpu_prep, DRM_UNLOCKED|DRM_AUTH), + DRM_IOCTL_DEF_DRV(NOUVEAU_GEM_CPU_FINI, nouveau_gem_ioctl_cpu_fini, DRM_UNLOCKED|DRM_AUTH), + DRM_IOCTL_DEF_DRV(NOUVEAU_GEM_INFO, nouveau_gem_ioctl_info, DRM_UNLOCKED|DRM_AUTH), +}; + +int nouveau_max_ioctl = DRM_ARRAY_SIZE(nouveau_ioctls); diff --git a/drivers/gpu/drm/nouveau/nouveau_connector.c b/drivers/gpu/drm/nouveau/nouveau_connector.c new file mode 100644 index 00000000..1595d0b6 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_connector.c @@ -0,0 +1,942 @@ +/* + * Copyright (C) 2008 Maarten Maathuis. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include + +#include "drmP.h" +#include "drm_edid.h" +#include "drm_crtc_helper.h" + +#include "nouveau_reg.h" +#include "nouveau_drv.h" +#include "nouveau_encoder.h" +#include "nouveau_crtc.h" +#include "nouveau_connector.h" +#include "nouveau_hw.h" + +static void nouveau_connector_hotplug(void *, int); + +static struct nouveau_encoder * +find_encoder_by_type(struct drm_connector *connector, int type) +{ + struct drm_device *dev = connector->dev; + struct nouveau_encoder *nv_encoder; + struct drm_mode_object *obj; + int i, id; + + for (i = 0; i < DRM_CONNECTOR_MAX_ENCODER; i++) { + id = connector->encoder_ids[i]; + if (!id) + break; + + obj = drm_mode_object_find(dev, id, DRM_MODE_OBJECT_ENCODER); + if (!obj) + continue; + nv_encoder = nouveau_encoder(obj_to_encoder(obj)); + + if (type == OUTPUT_ANY || nv_encoder->dcb->type == type) + return nv_encoder; + } + + return NULL; +} + +struct nouveau_connector * +nouveau_encoder_connector_get(struct nouveau_encoder *encoder) +{ + struct drm_device *dev = to_drm_encoder(encoder)->dev; + struct drm_connector *drm_connector; + + list_for_each_entry(drm_connector, &dev->mode_config.connector_list, head) { + if (drm_connector->encoder == to_drm_encoder(encoder)) + return nouveau_connector(drm_connector); + } + + return NULL; +} + +/*TODO: This could use improvement, and learn to handle the fixed + * BIOS tables etc. It's fine currently, for its only user. + */ +int +nouveau_connector_bpp(struct drm_connector *connector) +{ + struct nouveau_connector *nv_connector = nouveau_connector(connector); + + if (nv_connector->edid && nv_connector->edid->revision >= 4) { + u8 bpc = ((nv_connector->edid->input & 0x70) >> 3) + 4; + if (bpc > 4) + return bpc; + } + + return 18; +} + +static void +nouveau_connector_destroy(struct drm_connector *connector) +{ + struct nouveau_connector *nv_connector = nouveau_connector(connector); + struct drm_nouveau_private *dev_priv; + struct nouveau_gpio_engine *pgpio; + struct drm_device *dev; + + if (!nv_connector) + return; + + dev = nv_connector->base.dev; + dev_priv = dev->dev_private; + NV_DEBUG_KMS(dev, "\n"); + + pgpio = &dev_priv->engine.gpio; + if (pgpio->irq_unregister) { + pgpio->irq_unregister(dev, nv_connector->dcb->gpio_tag, + nouveau_connector_hotplug, connector); + } + + if (connector->connector_type == DRM_MODE_CONNECTOR_LVDS || + connector->connector_type == DRM_MODE_CONNECTOR_eDP) + nouveau_backlight_exit(connector); + + kfree(nv_connector->edid); + drm_sysfs_connector_remove(connector); + drm_connector_cleanup(connector); + kfree(connector); +} + +static struct nouveau_i2c_chan * +nouveau_connector_ddc_detect(struct drm_connector *connector, + struct nouveau_encoder **pnv_encoder) +{ + struct drm_device *dev = connector->dev; + int i; + + for (i = 0; i < DRM_CONNECTOR_MAX_ENCODER; i++) { + struct nouveau_i2c_chan *i2c = NULL; + struct nouveau_encoder *nv_encoder; + struct drm_mode_object *obj; + int id; + + id = connector->encoder_ids[i]; + if (!id) + break; + + obj = drm_mode_object_find(dev, id, DRM_MODE_OBJECT_ENCODER); + if (!obj) + continue; + nv_encoder = nouveau_encoder(obj_to_encoder(obj)); + + if (nv_encoder->dcb->i2c_index < 0xf) + i2c = nouveau_i2c_find(dev, nv_encoder->dcb->i2c_index); + + if (i2c && nouveau_probe_i2c_addr(i2c, 0x50)) { + *pnv_encoder = nv_encoder; + return i2c; + } + } + + return NULL; +} + +static struct nouveau_encoder * +nouveau_connector_of_detect(struct drm_connector *connector) +{ +#ifdef __powerpc__ + struct drm_device *dev = connector->dev; + struct nouveau_connector *nv_connector = nouveau_connector(connector); + struct nouveau_encoder *nv_encoder; + struct device_node *cn, *dn = pci_device_to_OF_node(dev->pdev); + + if (!dn || + !((nv_encoder = find_encoder_by_type(connector, OUTPUT_TMDS)) || + (nv_encoder = find_encoder_by_type(connector, OUTPUT_ANALOG)))) + return NULL; + + for_each_child_of_node(dn, cn) { + const char *name = of_get_property(cn, "name", NULL); + const void *edid = of_get_property(cn, "EDID", NULL); + int idx = name ? name[strlen(name) - 1] - 'A' : 0; + + if (nv_encoder->dcb->i2c_index == idx && edid) { + nv_connector->edid = + kmemdup(edid, EDID_LENGTH, GFP_KERNEL); + of_node_put(cn); + return nv_encoder; + } + } +#endif + return NULL; +} + +static void +nouveau_connector_set_encoder(struct drm_connector *connector, + struct nouveau_encoder *nv_encoder) +{ + struct nouveau_connector *nv_connector = nouveau_connector(connector); + struct drm_nouveau_private *dev_priv = connector->dev->dev_private; + struct drm_device *dev = connector->dev; + + if (nv_connector->detected_encoder == nv_encoder) + return; + nv_connector->detected_encoder = nv_encoder; + + if (nv_encoder->dcb->type == OUTPUT_LVDS || + nv_encoder->dcb->type == OUTPUT_TMDS) { + connector->doublescan_allowed = false; + connector->interlace_allowed = false; + } else { + connector->doublescan_allowed = true; + if (dev_priv->card_type == NV_20 || + (dev_priv->card_type == NV_10 && + (dev->pci_device & 0x0ff0) != 0x0100 && + (dev->pci_device & 0x0ff0) != 0x0150)) + /* HW is broken */ + connector->interlace_allowed = false; + else + connector->interlace_allowed = true; + } + + if (nv_connector->dcb->type == DCB_CONNECTOR_DVI_I) { + drm_connector_property_set_value(connector, + dev->mode_config.dvi_i_subconnector_property, + nv_encoder->dcb->type == OUTPUT_TMDS ? + DRM_MODE_SUBCONNECTOR_DVID : + DRM_MODE_SUBCONNECTOR_DVIA); + } +} + +static enum drm_connector_status +nouveau_connector_detect(struct drm_connector *connector, bool force) +{ + struct drm_device *dev = connector->dev; + struct nouveau_connector *nv_connector = nouveau_connector(connector); + struct nouveau_encoder *nv_encoder = NULL; + struct nouveau_i2c_chan *i2c; + int type; + + /* Cleanup the previous EDID block. */ + if (nv_connector->edid) { + drm_mode_connector_update_edid_property(connector, NULL); + kfree(nv_connector->edid); + nv_connector->edid = NULL; + } + + i2c = nouveau_connector_ddc_detect(connector, &nv_encoder); + if (i2c) { + nv_connector->edid = drm_get_edid(connector, &i2c->adapter); + drm_mode_connector_update_edid_property(connector, + nv_connector->edid); + if (!nv_connector->edid) { + NV_ERROR(dev, "DDC responded, but no EDID for %s\n", + drm_get_connector_name(connector)); + goto detect_analog; + } + + if (nv_encoder->dcb->type == OUTPUT_DP && + !nouveau_dp_detect(to_drm_encoder(nv_encoder))) { + NV_ERROR(dev, "Detected %s, but failed init\n", + drm_get_connector_name(connector)); + return connector_status_disconnected; + } + + /* Override encoder type for DVI-I based on whether EDID + * says the display is digital or analog, both use the + * same i2c channel so the value returned from ddc_detect + * isn't necessarily correct. + */ + if (nv_connector->dcb->type == DCB_CONNECTOR_DVI_I) { + if (nv_connector->edid->input & DRM_EDID_INPUT_DIGITAL) + type = OUTPUT_TMDS; + else + type = OUTPUT_ANALOG; + + nv_encoder = find_encoder_by_type(connector, type); + if (!nv_encoder) { + NV_ERROR(dev, "Detected %d encoder on %s, " + "but no object!\n", type, + drm_get_connector_name(connector)); + return connector_status_disconnected; + } + } + + nouveau_connector_set_encoder(connector, nv_encoder); + return connector_status_connected; + } + + nv_encoder = nouveau_connector_of_detect(connector); + if (nv_encoder) { + nouveau_connector_set_encoder(connector, nv_encoder); + return connector_status_connected; + } + +detect_analog: + nv_encoder = find_encoder_by_type(connector, OUTPUT_ANALOG); + if (!nv_encoder && !nouveau_tv_disable) + nv_encoder = find_encoder_by_type(connector, OUTPUT_TV); + if (nv_encoder && force) { + struct drm_encoder *encoder = to_drm_encoder(nv_encoder); + struct drm_encoder_helper_funcs *helper = + encoder->helper_private; + + if (helper->detect(encoder, connector) == + connector_status_connected) { + nouveau_connector_set_encoder(connector, nv_encoder); + return connector_status_connected; + } + + } + + return connector_status_disconnected; +} + +static enum drm_connector_status +nouveau_connector_detect_lvds(struct drm_connector *connector, bool force) +{ + struct drm_device *dev = connector->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_connector *nv_connector = nouveau_connector(connector); + struct nouveau_encoder *nv_encoder = NULL; + enum drm_connector_status status = connector_status_disconnected; + + /* Cleanup the previous EDID block. */ + if (nv_connector->edid) { + drm_mode_connector_update_edid_property(connector, NULL); + kfree(nv_connector->edid); + nv_connector->edid = NULL; + } + + nv_encoder = find_encoder_by_type(connector, OUTPUT_LVDS); + if (!nv_encoder) + return connector_status_disconnected; + + /* Try retrieving EDID via DDC */ + if (!dev_priv->vbios.fp_no_ddc) { + status = nouveau_connector_detect(connector, force); + if (status == connector_status_connected) + goto out; + } + + /* On some laptops (Sony, i'm looking at you) there appears to + * be no direct way of accessing the panel's EDID. The only + * option available to us appears to be to ask ACPI for help.. + * + * It's important this check's before trying straps, one of the + * said manufacturer's laptops are configured in such a way + * the nouveau decides an entry in the VBIOS FP mode table is + * valid - it's not (rh#613284) + */ + if (nv_encoder->dcb->lvdsconf.use_acpi_for_edid) { + if (!nouveau_acpi_edid(dev, connector)) { + status = connector_status_connected; + goto out; + } + } + + /* If no EDID found above, and the VBIOS indicates a hardcoded + * modeline is avalilable for the panel, set it as the panel's + * native mode and exit. + */ + if (nouveau_bios_fp_mode(dev, NULL) && (dev_priv->vbios.fp_no_ddc || + nv_encoder->dcb->lvdsconf.use_straps_for_mode)) { + status = connector_status_connected; + goto out; + } + + /* Still nothing, some VBIOS images have a hardcoded EDID block + * stored for the panel stored in them. + */ + if (!dev_priv->vbios.fp_no_ddc) { + struct edid *edid = + (struct edid *)nouveau_bios_embedded_edid(dev); + if (edid) { + nv_connector->edid = kmalloc(EDID_LENGTH, GFP_KERNEL); + *(nv_connector->edid) = *edid; + status = connector_status_connected; + } + } + +out: +#if defined(CONFIG_ACPI_BUTTON) || \ + (defined(CONFIG_ACPI_BUTTON_MODULE) && defined(MODULE)) + if (status == connector_status_connected && + !nouveau_ignorelid && !acpi_lid_open()) + status = connector_status_unknown; +#endif + + drm_mode_connector_update_edid_property(connector, nv_connector->edid); + nouveau_connector_set_encoder(connector, nv_encoder); + return status; +} + +static void +nouveau_connector_force(struct drm_connector *connector) +{ + struct nouveau_connector *nv_connector = nouveau_connector(connector); + struct nouveau_encoder *nv_encoder; + int type; + + if (nv_connector->dcb->type == DCB_CONNECTOR_DVI_I) { + if (connector->force == DRM_FORCE_ON_DIGITAL) + type = OUTPUT_TMDS; + else + type = OUTPUT_ANALOG; + } else + type = OUTPUT_ANY; + + nv_encoder = find_encoder_by_type(connector, type); + if (!nv_encoder) { + NV_ERROR(connector->dev, "can't find encoder to force %s on!\n", + drm_get_connector_name(connector)); + connector->status = connector_status_disconnected; + return; + } + + nouveau_connector_set_encoder(connector, nv_encoder); +} + +static int +nouveau_connector_set_property(struct drm_connector *connector, + struct drm_property *property, uint64_t value) +{ + struct nouveau_connector *nv_connector = nouveau_connector(connector); + struct nouveau_encoder *nv_encoder = nv_connector->detected_encoder; + struct drm_encoder *encoder = to_drm_encoder(nv_encoder); + struct drm_device *dev = connector->dev; + int ret; + + /* Scaling mode */ + if (property == dev->mode_config.scaling_mode_property) { + struct nouveau_crtc *nv_crtc = NULL; + bool modeset = false; + + switch (value) { + case DRM_MODE_SCALE_NONE: + case DRM_MODE_SCALE_FULLSCREEN: + case DRM_MODE_SCALE_CENTER: + case DRM_MODE_SCALE_ASPECT: + break; + default: + return -EINVAL; + } + + /* LVDS always needs gpu scaling */ + if (connector->connector_type == DRM_MODE_CONNECTOR_LVDS && + value == DRM_MODE_SCALE_NONE) + return -EINVAL; + + /* Changing between GPU and panel scaling requires a full + * modeset + */ + if ((nv_connector->scaling_mode == DRM_MODE_SCALE_NONE) || + (value == DRM_MODE_SCALE_NONE)) + modeset = true; + nv_connector->scaling_mode = value; + + if (connector->encoder && connector->encoder->crtc) + nv_crtc = nouveau_crtc(connector->encoder->crtc); + if (!nv_crtc) + return 0; + + if (modeset || !nv_crtc->set_scale) { + ret = drm_crtc_helper_set_mode(&nv_crtc->base, + &nv_crtc->base.mode, + nv_crtc->base.x, + nv_crtc->base.y, NULL); + if (!ret) + return -EINVAL; + } else { + ret = nv_crtc->set_scale(nv_crtc, value, true); + if (ret) + return ret; + } + + return 0; + } + + /* Dithering */ + if (property == dev->mode_config.dithering_mode_property) { + struct nouveau_crtc *nv_crtc = NULL; + + if (value == DRM_MODE_DITHERING_ON) + nv_connector->use_dithering = true; + else + nv_connector->use_dithering = false; + + if (connector->encoder && connector->encoder->crtc) + nv_crtc = nouveau_crtc(connector->encoder->crtc); + + if (!nv_crtc || !nv_crtc->set_dither) + return 0; + + return nv_crtc->set_dither(nv_crtc, nv_connector->use_dithering, + true); + } + + if (nv_encoder && nv_encoder->dcb->type == OUTPUT_TV) + return get_slave_funcs(encoder)->set_property( + encoder, connector, property, value); + + return -EINVAL; +} + +static struct drm_display_mode * +nouveau_connector_native_mode(struct drm_connector *connector) +{ + struct drm_connector_helper_funcs *helper = connector->helper_private; + struct nouveau_connector *nv_connector = nouveau_connector(connector); + struct drm_device *dev = connector->dev; + struct drm_display_mode *mode, *largest = NULL; + int high_w = 0, high_h = 0, high_v = 0; + + list_for_each_entry(mode, &nv_connector->base.probed_modes, head) { + mode->vrefresh = drm_mode_vrefresh(mode); + if (helper->mode_valid(connector, mode) != MODE_OK || + (mode->flags & DRM_MODE_FLAG_INTERLACE)) + continue; + + /* Use preferred mode if there is one.. */ + if (mode->type & DRM_MODE_TYPE_PREFERRED) { + NV_DEBUG_KMS(dev, "native mode from preferred\n"); + return drm_mode_duplicate(dev, mode); + } + + /* Otherwise, take the resolution with the largest width, then + * height, then vertical refresh + */ + if (mode->hdisplay < high_w) + continue; + + if (mode->hdisplay == high_w && mode->vdisplay < high_h) + continue; + + if (mode->hdisplay == high_w && mode->vdisplay == high_h && + mode->vrefresh < high_v) + continue; + + high_w = mode->hdisplay; + high_h = mode->vdisplay; + high_v = mode->vrefresh; + largest = mode; + } + + NV_DEBUG_KMS(dev, "native mode from largest: %dx%d@%d\n", + high_w, high_h, high_v); + return largest ? drm_mode_duplicate(dev, largest) : NULL; +} + +struct moderec { + int hdisplay; + int vdisplay; +}; + +static struct moderec scaler_modes[] = { + { 1920, 1200 }, + { 1920, 1080 }, + { 1680, 1050 }, + { 1600, 1200 }, + { 1400, 1050 }, + { 1280, 1024 }, + { 1280, 960 }, + { 1152, 864 }, + { 1024, 768 }, + { 800, 600 }, + { 720, 400 }, + { 640, 480 }, + { 640, 400 }, + { 640, 350 }, + {} +}; + +static int +nouveau_connector_scaler_modes_add(struct drm_connector *connector) +{ + struct nouveau_connector *nv_connector = nouveau_connector(connector); + struct drm_display_mode *native = nv_connector->native_mode, *m; + struct drm_device *dev = connector->dev; + struct moderec *mode = &scaler_modes[0]; + int modes = 0; + + if (!native) + return 0; + + while (mode->hdisplay) { + if (mode->hdisplay <= native->hdisplay && + mode->vdisplay <= native->vdisplay) { + m = drm_cvt_mode(dev, mode->hdisplay, mode->vdisplay, + drm_mode_vrefresh(native), false, + false, false); + if (!m) + continue; + + m->type |= DRM_MODE_TYPE_DRIVER; + + drm_mode_probed_add(connector, m); + modes++; + } + + mode++; + } + + return modes; +} + +static int +nouveau_connector_get_modes(struct drm_connector *connector) +{ + struct drm_device *dev = connector->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_connector *nv_connector = nouveau_connector(connector); + struct nouveau_encoder *nv_encoder = nv_connector->detected_encoder; + struct drm_encoder *encoder = to_drm_encoder(nv_encoder); + int ret = 0; + + /* destroy the native mode, the attached monitor could have changed. + */ + if (nv_connector->native_mode) { + drm_mode_destroy(dev, nv_connector->native_mode); + nv_connector->native_mode = NULL; + } + + if (nv_connector->edid) + ret = drm_add_edid_modes(connector, nv_connector->edid); + else + if (nv_encoder->dcb->type == OUTPUT_LVDS && + (nv_encoder->dcb->lvdsconf.use_straps_for_mode || + dev_priv->vbios.fp_no_ddc) && nouveau_bios_fp_mode(dev, NULL)) { + struct drm_display_mode mode; + + nouveau_bios_fp_mode(dev, &mode); + nv_connector->native_mode = drm_mode_duplicate(dev, &mode); + } + + /* Find the native mode if this is a digital panel, if we didn't + * find any modes through DDC previously add the native mode to + * the list of modes. + */ + if (!nv_connector->native_mode) + nv_connector->native_mode = + nouveau_connector_native_mode(connector); + if (ret == 0 && nv_connector->native_mode) { + struct drm_display_mode *mode; + + mode = drm_mode_duplicate(dev, nv_connector->native_mode); + drm_mode_probed_add(connector, mode); + ret = 1; + } + + if (nv_encoder->dcb->type == OUTPUT_TV) + ret = get_slave_funcs(encoder)->get_modes(encoder, connector); + + if (nv_connector->dcb->type == DCB_CONNECTOR_LVDS || + nv_connector->dcb->type == DCB_CONNECTOR_LVDS_SPWG || + nv_connector->dcb->type == DCB_CONNECTOR_eDP) + ret += nouveau_connector_scaler_modes_add(connector); + + return ret; +} + +static unsigned +get_tmds_link_bandwidth(struct drm_connector *connector) +{ + struct nouveau_connector *nv_connector = nouveau_connector(connector); + struct drm_nouveau_private *dev_priv = connector->dev->dev_private; + struct dcb_entry *dcb = nv_connector->detected_encoder->dcb; + + if (dcb->location != DCB_LOC_ON_CHIP || + dev_priv->chipset >= 0x46) + return 165000; + else if (dev_priv->chipset >= 0x40) + return 155000; + else if (dev_priv->chipset >= 0x18) + return 135000; + else + return 112000; +} + +static int +nouveau_connector_mode_valid(struct drm_connector *connector, + struct drm_display_mode *mode) +{ + struct nouveau_connector *nv_connector = nouveau_connector(connector); + struct nouveau_encoder *nv_encoder = nv_connector->detected_encoder; + struct drm_encoder *encoder = to_drm_encoder(nv_encoder); + unsigned min_clock = 25000, max_clock = min_clock; + unsigned clock = mode->clock; + + switch (nv_encoder->dcb->type) { + case OUTPUT_LVDS: + if (nv_connector->native_mode && + (mode->hdisplay > nv_connector->native_mode->hdisplay || + mode->vdisplay > nv_connector->native_mode->vdisplay)) + return MODE_PANEL; + + min_clock = 0; + max_clock = 400000; + break; + case OUTPUT_TMDS: + max_clock = get_tmds_link_bandwidth(connector); + if (nouveau_duallink && nv_encoder->dcb->duallink_possible) + max_clock *= 2; + break; + case OUTPUT_ANALOG: + max_clock = nv_encoder->dcb->crtconf.maxfreq; + if (!max_clock) + max_clock = 350000; + break; + case OUTPUT_TV: + return get_slave_funcs(encoder)->mode_valid(encoder, mode); + case OUTPUT_DP: + if (nv_encoder->dp.link_bw == DP_LINK_BW_2_7) + max_clock = nv_encoder->dp.link_nr * 270000; + else + max_clock = nv_encoder->dp.link_nr * 162000; + + clock = clock * nouveau_connector_bpp(connector) / 8; + break; + default: + BUG_ON(1); + return MODE_BAD; + } + + if (clock < min_clock) + return MODE_CLOCK_LOW; + + if (clock > max_clock) + return MODE_CLOCK_HIGH; + + return MODE_OK; +} + +static struct drm_encoder * +nouveau_connector_best_encoder(struct drm_connector *connector) +{ + struct nouveau_connector *nv_connector = nouveau_connector(connector); + + if (nv_connector->detected_encoder) + return to_drm_encoder(nv_connector->detected_encoder); + + return NULL; +} + +static const struct drm_connector_helper_funcs +nouveau_connector_helper_funcs = { + .get_modes = nouveau_connector_get_modes, + .mode_valid = nouveau_connector_mode_valid, + .best_encoder = nouveau_connector_best_encoder, +}; + +static const struct drm_connector_funcs +nouveau_connector_funcs = { + .dpms = drm_helper_connector_dpms, + .save = NULL, + .restore = NULL, + .detect = nouveau_connector_detect, + .destroy = nouveau_connector_destroy, + .fill_modes = drm_helper_probe_single_connector_modes, + .set_property = nouveau_connector_set_property, + .force = nouveau_connector_force +}; + +static const struct drm_connector_funcs +nouveau_connector_funcs_lvds = { + .dpms = drm_helper_connector_dpms, + .save = NULL, + .restore = NULL, + .detect = nouveau_connector_detect_lvds, + .destroy = nouveau_connector_destroy, + .fill_modes = drm_helper_probe_single_connector_modes, + .set_property = nouveau_connector_set_property, + .force = nouveau_connector_force +}; + +struct drm_connector * +nouveau_connector_create(struct drm_device *dev, int index) +{ + const struct drm_connector_funcs *funcs = &nouveau_connector_funcs; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpio_engine *pgpio = &dev_priv->engine.gpio; + struct nouveau_connector *nv_connector = NULL; + struct dcb_connector_table_entry *dcb = NULL; + struct drm_connector *connector; + int type, ret = 0; + + NV_DEBUG_KMS(dev, "\n"); + + if (index >= dev_priv->vbios.dcb.connector.entries) + return ERR_PTR(-EINVAL); + + dcb = &dev_priv->vbios.dcb.connector.entry[index]; + if (dcb->drm) + return dcb->drm; + + switch (dcb->type) { + case DCB_CONNECTOR_VGA: + type = DRM_MODE_CONNECTOR_VGA; + break; + case DCB_CONNECTOR_TV_0: + case DCB_CONNECTOR_TV_1: + case DCB_CONNECTOR_TV_3: + type = DRM_MODE_CONNECTOR_TV; + break; + case DCB_CONNECTOR_DVI_I: + type = DRM_MODE_CONNECTOR_DVII; + break; + case DCB_CONNECTOR_DVI_D: + type = DRM_MODE_CONNECTOR_DVID; + break; + case DCB_CONNECTOR_HDMI_0: + case DCB_CONNECTOR_HDMI_1: + type = DRM_MODE_CONNECTOR_HDMIA; + break; + case DCB_CONNECTOR_LVDS: + case DCB_CONNECTOR_LVDS_SPWG: + type = DRM_MODE_CONNECTOR_LVDS; + funcs = &nouveau_connector_funcs_lvds; + break; + case DCB_CONNECTOR_DP: + type = DRM_MODE_CONNECTOR_DisplayPort; + break; + case DCB_CONNECTOR_eDP: + type = DRM_MODE_CONNECTOR_eDP; + break; + default: + NV_ERROR(dev, "unknown connector type: 0x%02x!!\n", dcb->type); + return ERR_PTR(-EINVAL); + } + + nv_connector = kzalloc(sizeof(*nv_connector), GFP_KERNEL); + if (!nv_connector) + return ERR_PTR(-ENOMEM); + nv_connector->dcb = dcb; + connector = &nv_connector->base; + + /* defaults, will get overridden in detect() */ + connector->interlace_allowed = false; + connector->doublescan_allowed = false; + + drm_connector_init(dev, connector, funcs, type); + drm_connector_helper_add(connector, &nouveau_connector_helper_funcs); + + /* Check if we need dithering enabled */ + if (connector->connector_type == DRM_MODE_CONNECTOR_LVDS) { + bool dummy, is_24bit = false; + + ret = nouveau_bios_parse_lvds_table(dev, 0, &dummy, &is_24bit); + if (ret) { + NV_ERROR(dev, "Error parsing LVDS table, disabling " + "LVDS\n"); + goto fail; + } + + nv_connector->use_dithering = !is_24bit; + } + + /* Init DVI-I specific properties */ + if (dcb->type == DCB_CONNECTOR_DVI_I) { + drm_mode_create_dvi_i_properties(dev); + drm_connector_attach_property(connector, dev->mode_config.dvi_i_subconnector_property, 0); + drm_connector_attach_property(connector, dev->mode_config.dvi_i_select_subconnector_property, 0); + } + + switch (dcb->type) { + case DCB_CONNECTOR_VGA: + if (dev_priv->card_type >= NV_50) { + drm_connector_attach_property(connector, + dev->mode_config.scaling_mode_property, + nv_connector->scaling_mode); + } + connector->polled = DRM_CONNECTOR_POLL_CONNECT; + /* fall-through */ + case DCB_CONNECTOR_TV_0: + case DCB_CONNECTOR_TV_1: + case DCB_CONNECTOR_TV_3: + nv_connector->scaling_mode = DRM_MODE_SCALE_NONE; + break; + default: + nv_connector->scaling_mode = DRM_MODE_SCALE_FULLSCREEN; + + drm_connector_attach_property(connector, + dev->mode_config.scaling_mode_property, + nv_connector->scaling_mode); + drm_connector_attach_property(connector, + dev->mode_config.dithering_mode_property, + nv_connector->use_dithering ? + DRM_MODE_DITHERING_ON : DRM_MODE_DITHERING_OFF); + + if (connector->connector_type != DRM_MODE_CONNECTOR_LVDS) { + if (dev_priv->card_type >= NV_50) + connector->polled = DRM_CONNECTOR_POLL_HPD; + else + connector->polled = DRM_CONNECTOR_POLL_CONNECT; + } + break; + } + + if (pgpio->irq_register) { + pgpio->irq_register(dev, nv_connector->dcb->gpio_tag, + nouveau_connector_hotplug, connector); + } + + drm_sysfs_connector_add(connector); + + if (connector->connector_type == DRM_MODE_CONNECTOR_LVDS || + connector->connector_type == DRM_MODE_CONNECTOR_eDP) + nouveau_backlight_init(connector); + + dcb->drm = connector; + return dcb->drm; + +fail: + drm_connector_cleanup(connector); + kfree(connector); + return ERR_PTR(ret); + +} + +static void +nouveau_connector_hotplug(void *data, int plugged) +{ + struct drm_connector *connector = data; + struct drm_device *dev = connector->dev; + + NV_INFO(dev, "%splugged %s\n", plugged ? "" : "un", + drm_get_connector_name(connector)); + + if (connector->encoder && connector->encoder->crtc && + connector->encoder->crtc->enabled) { + struct nouveau_encoder *nv_encoder = nouveau_encoder(connector->encoder); + struct drm_encoder_helper_funcs *helper = + connector->encoder->helper_private; + + if (nv_encoder->dcb->type == OUTPUT_DP) { + if (plugged) + helper->dpms(connector->encoder, DRM_MODE_DPMS_ON); + else + helper->dpms(connector->encoder, DRM_MODE_DPMS_OFF); + } + } + + drm_helper_hpd_irq_event(dev); +} diff --git a/drivers/gpu/drm/nouveau/nouveau_connector.h b/drivers/gpu/drm/nouveau/nouveau_connector.h new file mode 100644 index 00000000..711b1e92 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_connector.h @@ -0,0 +1,58 @@ +/* + * Copyright (C) 2008 Maarten Maathuis. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __NOUVEAU_CONNECTOR_H__ +#define __NOUVEAU_CONNECTOR_H__ + +#include "drm_edid.h" +#include "nouveau_i2c.h" + +struct nouveau_connector { + struct drm_connector base; + + struct dcb_connector_table_entry *dcb; + + int scaling_mode; + bool use_dithering; + + struct nouveau_encoder *detected_encoder; + struct edid *edid; + struct drm_display_mode *native_mode; +}; + +static inline struct nouveau_connector *nouveau_connector( + struct drm_connector *con) +{ + return container_of(con, struct nouveau_connector, base); +} + +struct drm_connector * +nouveau_connector_create(struct drm_device *, int index); + +int +nouveau_connector_bpp(struct drm_connector *); + +#endif /* __NOUVEAU_CONNECTOR_H__ */ diff --git a/drivers/gpu/drm/nouveau/nouveau_crtc.h b/drivers/gpu/drm/nouveau/nouveau_crtc.h new file mode 100644 index 00000000..cb1ce2a0 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_crtc.h @@ -0,0 +1,97 @@ +/* + * Copyright (C) 2008 Maarten Maathuis. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __NOUVEAU_CRTC_H__ +#define __NOUVEAU_CRTC_H__ + +struct nouveau_crtc { + struct drm_crtc base; + + int index; + + struct drm_display_mode *mode; + + uint32_t dpms_saved_fp_control; + uint32_t fp_users; + int saturation; + int sharpness; + int last_dpms; + + int cursor_saved_x, cursor_saved_y; + + struct { + int cpp; + bool blanked; + uint32_t offset; + uint32_t tile_flags; + } fb; + + struct { + struct nouveau_bo *nvbo; + bool visible; + uint32_t offset; + void (*set_offset)(struct nouveau_crtc *, uint32_t offset); + void (*set_pos)(struct nouveau_crtc *, int x, int y); + void (*hide)(struct nouveau_crtc *, bool update); + void (*show)(struct nouveau_crtc *, bool update); + } cursor; + + struct { + struct nouveau_bo *nvbo; + uint16_t r[256]; + uint16_t g[256]; + uint16_t b[256]; + int depth; + } lut; + + int (*set_dither)(struct nouveau_crtc *crtc, bool on, bool update); + int (*set_scale)(struct nouveau_crtc *crtc, int mode, bool update); +}; + +static inline struct nouveau_crtc *nouveau_crtc(struct drm_crtc *crtc) +{ + return container_of(crtc, struct nouveau_crtc, base); +} + +static inline struct drm_crtc *to_drm_crtc(struct nouveau_crtc *crtc) +{ + return &crtc->base; +} + +int nv50_crtc_create(struct drm_device *dev, int index); +int nv50_cursor_init(struct nouveau_crtc *); +void nv50_cursor_fini(struct nouveau_crtc *); +int nv50_crtc_cursor_set(struct drm_crtc *drm_crtc, struct drm_file *file_priv, + uint32_t buffer_handle, uint32_t width, + uint32_t height); +int nv50_crtc_cursor_move(struct drm_crtc *drm_crtc, int x, int y); + +int nv04_cursor_init(struct nouveau_crtc *); + +struct nouveau_connector * +nouveau_crtc_connector_get(struct nouveau_crtc *crtc); + +#endif /* __NOUVEAU_CRTC_H__ */ diff --git a/drivers/gpu/drm/nouveau/nouveau_debugfs.c b/drivers/gpu/drm/nouveau/nouveau_debugfs.c new file mode 100644 index 00000000..8e159236 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_debugfs.c @@ -0,0 +1,197 @@ +/* + * Copyright (C) 2009 Red Hat + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +/* + * Authors: + * Ben Skeggs + */ + +#include + +#include "drmP.h" +#include "nouveau_drv.h" + +#include + +static int +nouveau_debugfs_channel_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct nouveau_channel *chan = node->info_ent->data; + + seq_printf(m, "channel id : %d\n", chan->id); + + seq_printf(m, "cpu fifo state:\n"); + seq_printf(m, " base: 0x%08x\n", chan->pushbuf_base); + seq_printf(m, " max: 0x%08x\n", chan->dma.max << 2); + seq_printf(m, " cur: 0x%08x\n", chan->dma.cur << 2); + seq_printf(m, " put: 0x%08x\n", chan->dma.put << 2); + seq_printf(m, " free: 0x%08x\n", chan->dma.free << 2); + if (chan->dma.ib_max) { + seq_printf(m, " ib max: 0x%08x\n", chan->dma.ib_max); + seq_printf(m, " ib put: 0x%08x\n", chan->dma.ib_put); + seq_printf(m, " ib free: 0x%08x\n", chan->dma.ib_free); + } + + seq_printf(m, "gpu fifo state:\n"); + seq_printf(m, " get: 0x%08x\n", + nvchan_rd32(chan, chan->user_get)); + seq_printf(m, " put: 0x%08x\n", + nvchan_rd32(chan, chan->user_put)); + if (chan->dma.ib_max) { + seq_printf(m, " ib get: 0x%08x\n", + nvchan_rd32(chan, 0x88)); + seq_printf(m, " ib put: 0x%08x\n", + nvchan_rd32(chan, 0x8c)); + } + + seq_printf(m, "last fence : %d\n", chan->fence.sequence); + seq_printf(m, "last signalled: %d\n", chan->fence.sequence_ack); + return 0; +} + +int +nouveau_debugfs_channel_init(struct nouveau_channel *chan) +{ + struct drm_nouveau_private *dev_priv = chan->dev->dev_private; + struct drm_minor *minor = chan->dev->primary; + int ret; + + if (!dev_priv->debugfs.channel_root) { + dev_priv->debugfs.channel_root = + debugfs_create_dir("channel", minor->debugfs_root); + if (!dev_priv->debugfs.channel_root) + return -ENOENT; + } + + snprintf(chan->debugfs.name, 32, "%d", chan->id); + chan->debugfs.info.name = chan->debugfs.name; + chan->debugfs.info.show = nouveau_debugfs_channel_info; + chan->debugfs.info.driver_features = 0; + chan->debugfs.info.data = chan; + + ret = drm_debugfs_create_files(&chan->debugfs.info, 1, + dev_priv->debugfs.channel_root, + chan->dev->primary); + if (ret == 0) + chan->debugfs.active = true; + return ret; +} + +void +nouveau_debugfs_channel_fini(struct nouveau_channel *chan) +{ + struct drm_nouveau_private *dev_priv = chan->dev->dev_private; + + if (!chan->debugfs.active) + return; + + drm_debugfs_remove_files(&chan->debugfs.info, 1, chan->dev->primary); + chan->debugfs.active = false; + + if (chan == dev_priv->channel) { + debugfs_remove(dev_priv->debugfs.channel_root); + dev_priv->debugfs.channel_root = NULL; + } +} + +static int +nouveau_debugfs_chipset_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_minor *minor = node->minor; + struct drm_device *dev = minor->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + uint32_t ppci_0; + + ppci_0 = nv_rd32(dev, dev_priv->chipset >= 0x40 ? 0x88000 : 0x1800); + + seq_printf(m, "PMC_BOOT_0: 0x%08x\n", nv_rd32(dev, NV03_PMC_BOOT_0)); + seq_printf(m, "PCI ID : 0x%04x:0x%04x\n", + ppci_0 & 0xffff, ppci_0 >> 16); + return 0; +} + +static int +nouveau_debugfs_memory_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_minor *minor = node->minor; + struct drm_nouveau_private *dev_priv = minor->dev->dev_private; + + seq_printf(m, "VRAM total: %dKiB\n", (int)(dev_priv->vram_size >> 10)); + return 0; +} + +static int +nouveau_debugfs_vbios_image(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_nouveau_private *dev_priv = node->minor->dev->dev_private; + int i; + + for (i = 0; i < dev_priv->vbios.length; i++) + seq_printf(m, "%c", dev_priv->vbios.data[i]); + return 0; +} + +static int +nouveau_debugfs_evict_vram(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_nouveau_private *dev_priv = node->minor->dev->dev_private; + int ret; + + ret = ttm_bo_evict_mm(&dev_priv->ttm.bdev, TTM_PL_VRAM); + if (ret) + seq_printf(m, "failed: %d", ret); + else + seq_printf(m, "succeeded\n"); + return 0; +} + +static struct drm_info_list nouveau_debugfs_list[] = { + { "evict_vram", nouveau_debugfs_evict_vram, 0, NULL }, + { "chipset", nouveau_debugfs_chipset_info, 0, NULL }, + { "memory", nouveau_debugfs_memory_info, 0, NULL }, + { "vbios.rom", nouveau_debugfs_vbios_image, 0, NULL }, + { "ttm_page_pool", ttm_page_alloc_debugfs, 0, NULL }, +}; +#define NOUVEAU_DEBUGFS_ENTRIES ARRAY_SIZE(nouveau_debugfs_list) + +int +nouveau_debugfs_init(struct drm_minor *minor) +{ + drm_debugfs_create_files(nouveau_debugfs_list, NOUVEAU_DEBUGFS_ENTRIES, + minor->debugfs_root, minor); + return 0; +} + +void +nouveau_debugfs_takedown(struct drm_minor *minor) +{ + drm_debugfs_remove_files(nouveau_debugfs_list, NOUVEAU_DEBUGFS_ENTRIES, + minor); +} diff --git a/drivers/gpu/drm/nouveau/nouveau_display.c b/drivers/gpu/drm/nouveau/nouveau_display.c new file mode 100644 index 00000000..eb514ea2 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_display.c @@ -0,0 +1,368 @@ +/* + * Copyright (C) 2008 Maarten Maathuis. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "drmP.h" +#include "drm_crtc_helper.h" +#include "nouveau_drv.h" +#include "nouveau_fb.h" +#include "nouveau_fbcon.h" +#include "nouveau_hw.h" +#include "nouveau_crtc.h" +#include "nouveau_dma.h" +#include "nv50_display.h" + +static void +nouveau_user_framebuffer_destroy(struct drm_framebuffer *drm_fb) +{ + struct nouveau_framebuffer *fb = nouveau_framebuffer(drm_fb); + + if (fb->nvbo) + drm_gem_object_unreference_unlocked(fb->nvbo->gem); + + drm_framebuffer_cleanup(drm_fb); + kfree(fb); +} + +static int +nouveau_user_framebuffer_create_handle(struct drm_framebuffer *drm_fb, + struct drm_file *file_priv, + unsigned int *handle) +{ + struct nouveau_framebuffer *fb = nouveau_framebuffer(drm_fb); + + return drm_gem_handle_create(file_priv, fb->nvbo->gem, handle); +} + +static const struct drm_framebuffer_funcs nouveau_framebuffer_funcs = { + .destroy = nouveau_user_framebuffer_destroy, + .create_handle = nouveau_user_framebuffer_create_handle, +}; + +int +nouveau_framebuffer_init(struct drm_device *dev, + struct nouveau_framebuffer *nv_fb, + struct drm_mode_fb_cmd *mode_cmd, + struct nouveau_bo *nvbo) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct drm_framebuffer *fb = &nv_fb->base; + int ret; + + ret = drm_framebuffer_init(dev, fb, &nouveau_framebuffer_funcs); + if (ret) { + return ret; + } + + drm_helper_mode_fill_fb_struct(fb, mode_cmd); + nv_fb->nvbo = nvbo; + + if (dev_priv->card_type >= NV_50) { + u32 tile_flags = nouveau_bo_tile_layout(nvbo); + if (tile_flags == 0x7a00 || + tile_flags == 0xfe00) + nv_fb->r_dma = NvEvoFB32; + else + if (tile_flags == 0x7000) + nv_fb->r_dma = NvEvoFB16; + else + nv_fb->r_dma = NvEvoVRAM_LP; + + switch (fb->depth) { + case 8: nv_fb->r_format = NV50_EVO_CRTC_FB_DEPTH_8; break; + case 15: nv_fb->r_format = NV50_EVO_CRTC_FB_DEPTH_15; break; + case 16: nv_fb->r_format = NV50_EVO_CRTC_FB_DEPTH_16; break; + case 24: + case 32: nv_fb->r_format = NV50_EVO_CRTC_FB_DEPTH_24; break; + case 30: nv_fb->r_format = NV50_EVO_CRTC_FB_DEPTH_30; break; + default: + NV_ERROR(dev, "unknown depth %d\n", fb->depth); + return -EINVAL; + } + + if (dev_priv->chipset == 0x50) + nv_fb->r_format |= (tile_flags << 8); + + if (!tile_flags) + nv_fb->r_pitch = 0x00100000 | fb->pitch; + else { + u32 mode = nvbo->tile_mode; + if (dev_priv->card_type >= NV_C0) + mode >>= 4; + nv_fb->r_pitch = ((fb->pitch / 4) << 4) | mode; + } + } + + return 0; +} + +static struct drm_framebuffer * +nouveau_user_framebuffer_create(struct drm_device *dev, + struct drm_file *file_priv, + struct drm_mode_fb_cmd *mode_cmd) +{ + struct nouveau_framebuffer *nouveau_fb; + struct drm_gem_object *gem; + int ret; + + gem = drm_gem_object_lookup(dev, file_priv, mode_cmd->handle); + if (!gem) + return ERR_PTR(-ENOENT); + + nouveau_fb = kzalloc(sizeof(struct nouveau_framebuffer), GFP_KERNEL); + if (!nouveau_fb) + return ERR_PTR(-ENOMEM); + + ret = nouveau_framebuffer_init(dev, nouveau_fb, mode_cmd, nouveau_gem_object(gem)); + if (ret) { + drm_gem_object_unreference(gem); + return ERR_PTR(ret); + } + + return &nouveau_fb->base; +} + +const struct drm_mode_config_funcs nouveau_mode_config_funcs = { + .fb_create = nouveau_user_framebuffer_create, + .output_poll_changed = nouveau_fbcon_output_poll_changed, +}; + +int +nouveau_vblank_enable(struct drm_device *dev, int crtc) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + if (dev_priv->card_type >= NV_50) + nv_mask(dev, NV50_PDISPLAY_INTR_EN_1, 0, + NV50_PDISPLAY_INTR_EN_1_VBLANK_CRTC_(crtc)); + else + NVWriteCRTC(dev, crtc, NV_PCRTC_INTR_EN_0, + NV_PCRTC_INTR_0_VBLANK); + + return 0; +} + +void +nouveau_vblank_disable(struct drm_device *dev, int crtc) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + if (dev_priv->card_type >= NV_50) + nv_mask(dev, NV50_PDISPLAY_INTR_EN_1, + NV50_PDISPLAY_INTR_EN_1_VBLANK_CRTC_(crtc), 0); + else + NVWriteCRTC(dev, crtc, NV_PCRTC_INTR_EN_0, 0); +} + +static int +nouveau_page_flip_reserve(struct nouveau_bo *old_bo, + struct nouveau_bo *new_bo) +{ + int ret; + + ret = nouveau_bo_pin(new_bo, TTM_PL_FLAG_VRAM); + if (ret) + return ret; + + ret = ttm_bo_reserve(&new_bo->bo, false, false, false, 0); + if (ret) + goto fail; + + ret = ttm_bo_reserve(&old_bo->bo, false, false, false, 0); + if (ret) + goto fail_unreserve; + + return 0; + +fail_unreserve: + ttm_bo_unreserve(&new_bo->bo); +fail: + nouveau_bo_unpin(new_bo); + return ret; +} + +static void +nouveau_page_flip_unreserve(struct nouveau_bo *old_bo, + struct nouveau_bo *new_bo, + struct nouveau_fence *fence) +{ + nouveau_bo_fence(new_bo, fence); + ttm_bo_unreserve(&new_bo->bo); + + nouveau_bo_fence(old_bo, fence); + ttm_bo_unreserve(&old_bo->bo); + + nouveau_bo_unpin(old_bo); +} + +static int +nouveau_page_flip_emit(struct nouveau_channel *chan, + struct nouveau_bo *old_bo, + struct nouveau_bo *new_bo, + struct nouveau_page_flip_state *s, + struct nouveau_fence **pfence) +{ + struct drm_nouveau_private *dev_priv = chan->dev->dev_private; + struct drm_device *dev = chan->dev; + unsigned long flags; + int ret; + + /* Queue it to the pending list */ + spin_lock_irqsave(&dev->event_lock, flags); + list_add_tail(&s->head, &chan->nvsw.flip); + spin_unlock_irqrestore(&dev->event_lock, flags); + + /* Synchronize with the old framebuffer */ + ret = nouveau_fence_sync(old_bo->bo.sync_obj, chan); + if (ret) + goto fail; + + /* Emit the pageflip */ + ret = RING_SPACE(chan, 2); + if (ret) + goto fail; + + if (dev_priv->card_type < NV_C0) + BEGIN_RING(chan, NvSubSw, NV_SW_PAGE_FLIP, 1); + else + BEGIN_NVC0(chan, 2, NvSubM2MF, 0x0500, 1); + OUT_RING (chan, 0); + FIRE_RING (chan); + + ret = nouveau_fence_new(chan, pfence, true); + if (ret) + goto fail; + + return 0; +fail: + spin_lock_irqsave(&dev->event_lock, flags); + list_del(&s->head); + spin_unlock_irqrestore(&dev->event_lock, flags); + return ret; +} + +int +nouveau_crtc_page_flip(struct drm_crtc *crtc, struct drm_framebuffer *fb, + struct drm_pending_vblank_event *event) +{ + struct drm_device *dev = crtc->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_bo *old_bo = nouveau_framebuffer(crtc->fb)->nvbo; + struct nouveau_bo *new_bo = nouveau_framebuffer(fb)->nvbo; + struct nouveau_page_flip_state *s; + struct nouveau_channel *chan; + struct nouveau_fence *fence; + int ret; + + if (!dev_priv->channel) + return -ENODEV; + + s = kzalloc(sizeof(*s), GFP_KERNEL); + if (!s) + return -ENOMEM; + + /* Don't let the buffers go away while we flip */ + ret = nouveau_page_flip_reserve(old_bo, new_bo); + if (ret) + goto fail_free; + + /* Initialize a page flip struct */ + *s = (struct nouveau_page_flip_state) + { { }, event, nouveau_crtc(crtc)->index, + fb->bits_per_pixel, fb->pitch, crtc->x, crtc->y, + new_bo->bo.offset }; + + /* Choose the channel the flip will be handled in */ + chan = nouveau_fence_channel(new_bo->bo.sync_obj); + if (!chan) + chan = nouveau_channel_get_unlocked(dev_priv->channel); + mutex_lock(&chan->mutex); + + /* Emit a page flip */ + if (dev_priv->card_type >= NV_50) { + ret = nv50_display_flip_next(crtc, fb, chan); + if (ret) { + nouveau_channel_put(&chan); + goto fail_unreserve; + } + } + + ret = nouveau_page_flip_emit(chan, old_bo, new_bo, s, &fence); + nouveau_channel_put(&chan); + if (ret) + goto fail_unreserve; + + /* Update the crtc struct and cleanup */ + crtc->fb = fb; + + nouveau_page_flip_unreserve(old_bo, new_bo, fence); + nouveau_fence_unref(&fence); + return 0; + +fail_unreserve: + nouveau_page_flip_unreserve(old_bo, new_bo, NULL); +fail_free: + kfree(s); + return ret; +} + +int +nouveau_finish_page_flip(struct nouveau_channel *chan, + struct nouveau_page_flip_state *ps) +{ + struct drm_device *dev = chan->dev; + struct nouveau_page_flip_state *s; + unsigned long flags; + + spin_lock_irqsave(&dev->event_lock, flags); + + if (list_empty(&chan->nvsw.flip)) { + NV_ERROR(dev, "Unexpected pageflip in channel %d.\n", chan->id); + spin_unlock_irqrestore(&dev->event_lock, flags); + return -EINVAL; + } + + s = list_first_entry(&chan->nvsw.flip, + struct nouveau_page_flip_state, head); + if (s->event) { + struct drm_pending_vblank_event *e = s->event; + struct timeval now; + + do_gettimeofday(&now); + e->event.sequence = 0; + e->event.tv_sec = now.tv_sec; + e->event.tv_usec = now.tv_usec; + list_add_tail(&e->base.link, &e->base.file_priv->event_list); + wake_up_interruptible(&e->base.file_priv->event_wait); + } + + list_del(&s->head); + if (ps) + *ps = *s; + kfree(s); + + spin_unlock_irqrestore(&dev->event_lock, flags); + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nouveau_dma.c b/drivers/gpu/drm/nouveau/nouveau_dma.c new file mode 100644 index 00000000..568caedd --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_dma.c @@ -0,0 +1,338 @@ +/* + * Copyright (C) 2007 Ben Skeggs. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "drmP.h" +#include "drm.h" +#include "nouveau_drv.h" +#include "nouveau_dma.h" +#include "nouveau_ramht.h" + +void +nouveau_dma_pre_init(struct nouveau_channel *chan) +{ + struct drm_nouveau_private *dev_priv = chan->dev->dev_private; + struct nouveau_bo *pushbuf = chan->pushbuf_bo; + + if (dev_priv->card_type >= NV_50) { + const int ib_size = pushbuf->bo.mem.size / 2; + + chan->dma.ib_base = (pushbuf->bo.mem.size - ib_size) >> 2; + chan->dma.ib_max = (ib_size / 8) - 1; + chan->dma.ib_put = 0; + chan->dma.ib_free = chan->dma.ib_max - chan->dma.ib_put; + + chan->dma.max = (pushbuf->bo.mem.size - ib_size) >> 2; + } else { + chan->dma.max = (pushbuf->bo.mem.size >> 2) - 2; + } + + chan->dma.put = 0; + chan->dma.cur = chan->dma.put; + chan->dma.free = chan->dma.max - chan->dma.cur; +} + +int +nouveau_dma_init(struct nouveau_channel *chan) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + int ret, i; + + if (dev_priv->card_type >= NV_C0) { + ret = nouveau_gpuobj_gr_new(chan, 0x9039, 0x9039); + if (ret) + return ret; + + ret = RING_SPACE(chan, 2); + if (ret) + return ret; + + BEGIN_NVC0(chan, 2, NvSubM2MF, 0x0000, 1); + OUT_RING (chan, 0x00009039); + FIRE_RING (chan); + return 0; + } + + /* Create NV_MEMORY_TO_MEMORY_FORMAT for buffer moves */ + ret = nouveau_gpuobj_gr_new(chan, NvM2MF, dev_priv->card_type < NV_50 ? + 0x0039 : 0x5039); + if (ret) + return ret; + + /* NV_MEMORY_TO_MEMORY_FORMAT requires a notifier object */ + ret = nouveau_notifier_alloc(chan, NvNotify0, 32, 0xfe0, 0x1000, + &chan->m2mf_ntfy); + if (ret) + return ret; + + /* Insert NOPS for NOUVEAU_DMA_SKIPS */ + ret = RING_SPACE(chan, NOUVEAU_DMA_SKIPS); + if (ret) + return ret; + + for (i = 0; i < NOUVEAU_DMA_SKIPS; i++) + OUT_RING(chan, 0); + + /* Initialise NV_MEMORY_TO_MEMORY_FORMAT */ + ret = RING_SPACE(chan, 6); + if (ret) + return ret; + BEGIN_RING(chan, NvSubM2MF, NV_MEMORY_TO_MEMORY_FORMAT_NAME, 1); + OUT_RING (chan, NvM2MF); + BEGIN_RING(chan, NvSubM2MF, NV_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY, 3); + OUT_RING (chan, NvNotify0); + OUT_RING (chan, chan->vram_handle); + OUT_RING (chan, chan->gart_handle); + + /* Sit back and pray the channel works.. */ + FIRE_RING(chan); + + return 0; +} + +void +OUT_RINGp(struct nouveau_channel *chan, const void *data, unsigned nr_dwords) +{ + bool is_iomem; + u32 *mem = ttm_kmap_obj_virtual(&chan->pushbuf_bo->kmap, &is_iomem); + mem = &mem[chan->dma.cur]; + if (is_iomem) + memcpy_toio((void __force __iomem *)mem, data, nr_dwords * 4); + else + memcpy(mem, data, nr_dwords * 4); + chan->dma.cur += nr_dwords; +} + +/* Fetch and adjust GPU GET pointer + * + * Returns: + * value >= 0, the adjusted GET pointer + * -EINVAL if GET pointer currently outside main push buffer + * -EBUSY if timeout exceeded + */ +static inline int +READ_GET(struct nouveau_channel *chan, uint32_t *prev_get, uint32_t *timeout) +{ + uint32_t val; + + val = nvchan_rd32(chan, chan->user_get); + + /* reset counter as long as GET is still advancing, this is + * to avoid misdetecting a GPU lockup if the GPU happens to + * just be processing an operation that takes a long time + */ + if (val != *prev_get) { + *prev_get = val; + *timeout = 0; + } + + if ((++*timeout & 0xff) == 0) { + DRM_UDELAY(1); + if (*timeout > 100000) + return -EBUSY; + } + + if (val < chan->pushbuf_base || + val > chan->pushbuf_base + (chan->dma.max << 2)) + return -EINVAL; + + return (val - chan->pushbuf_base) >> 2; +} + +void +nv50_dma_push(struct nouveau_channel *chan, struct nouveau_bo *bo, + int delta, int length) +{ + struct nouveau_bo *pb = chan->pushbuf_bo; + uint64_t offset = bo->bo.offset + delta; + int ip = (chan->dma.ib_put * 2) + chan->dma.ib_base; + + BUG_ON(chan->dma.ib_free < 1); + nouveau_bo_wr32(pb, ip++, lower_32_bits(offset)); + nouveau_bo_wr32(pb, ip++, upper_32_bits(offset) | length << 8); + + chan->dma.ib_put = (chan->dma.ib_put + 1) & chan->dma.ib_max; + + DRM_MEMORYBARRIER(); + /* Flush writes. */ + nouveau_bo_rd32(pb, 0); + + nvchan_wr32(chan, 0x8c, chan->dma.ib_put); + chan->dma.ib_free--; +} + +static int +nv50_dma_push_wait(struct nouveau_channel *chan, int count) +{ + uint32_t cnt = 0, prev_get = 0; + + while (chan->dma.ib_free < count) { + uint32_t get = nvchan_rd32(chan, 0x88); + if (get != prev_get) { + prev_get = get; + cnt = 0; + } + + if ((++cnt & 0xff) == 0) { + DRM_UDELAY(1); + if (cnt > 100000) + return -EBUSY; + } + + chan->dma.ib_free = get - chan->dma.ib_put; + if (chan->dma.ib_free <= 0) + chan->dma.ib_free += chan->dma.ib_max; + } + + return 0; +} + +static int +nv50_dma_wait(struct nouveau_channel *chan, int slots, int count) +{ + uint32_t cnt = 0, prev_get = 0; + int ret; + + ret = nv50_dma_push_wait(chan, slots + 1); + if (unlikely(ret)) + return ret; + + while (chan->dma.free < count) { + int get = READ_GET(chan, &prev_get, &cnt); + if (unlikely(get < 0)) { + if (get == -EINVAL) + continue; + + return get; + } + + if (get <= chan->dma.cur) { + chan->dma.free = chan->dma.max - chan->dma.cur; + if (chan->dma.free >= count) + break; + + FIRE_RING(chan); + do { + get = READ_GET(chan, &prev_get, &cnt); + if (unlikely(get < 0)) { + if (get == -EINVAL) + continue; + return get; + } + } while (get == 0); + chan->dma.cur = 0; + chan->dma.put = 0; + } + + chan->dma.free = get - chan->dma.cur - 1; + } + + return 0; +} + +int +nouveau_dma_wait(struct nouveau_channel *chan, int slots, int size) +{ + uint32_t prev_get = 0, cnt = 0; + int get; + + if (chan->dma.ib_max) + return nv50_dma_wait(chan, slots, size); + + while (chan->dma.free < size) { + get = READ_GET(chan, &prev_get, &cnt); + if (unlikely(get == -EBUSY)) + return -EBUSY; + + /* loop until we have a usable GET pointer. the value + * we read from the GPU may be outside the main ring if + * PFIFO is processing a buffer called from the main ring, + * discard these values until something sensible is seen. + * + * the other case we discard GET is while the GPU is fetching + * from the SKIPS area, so the code below doesn't have to deal + * with some fun corner cases. + */ + if (unlikely(get == -EINVAL) || get < NOUVEAU_DMA_SKIPS) + continue; + + if (get <= chan->dma.cur) { + /* engine is fetching behind us, or is completely + * idle (GET == PUT) so we have free space up until + * the end of the push buffer + * + * we can only hit that path once per call due to + * looping back to the beginning of the push buffer, + * we'll hit the fetching-ahead-of-us path from that + * point on. + * + * the *one* exception to that rule is if we read + * GET==PUT, in which case the below conditional will + * always succeed and break us out of the wait loop. + */ + chan->dma.free = chan->dma.max - chan->dma.cur; + if (chan->dma.free >= size) + break; + + /* not enough space left at the end of the push buffer, + * instruct the GPU to jump back to the start right + * after processing the currently pending commands. + */ + OUT_RING(chan, chan->pushbuf_base | 0x20000000); + + /* wait for GET to depart from the skips area. + * prevents writing GET==PUT and causing a race + * condition that causes us to think the GPU is + * idle when it's not. + */ + do { + get = READ_GET(chan, &prev_get, &cnt); + if (unlikely(get == -EBUSY)) + return -EBUSY; + if (unlikely(get == -EINVAL)) + continue; + } while (get <= NOUVEAU_DMA_SKIPS); + WRITE_PUT(NOUVEAU_DMA_SKIPS); + + /* we're now submitting commands at the start of + * the push buffer. + */ + chan->dma.cur = + chan->dma.put = NOUVEAU_DMA_SKIPS; + } + + /* engine fetching ahead of us, we have space up until the + * current GET pointer. the "- 1" is to ensure there's + * space left to emit a jump back to the beginning of the + * push buffer if we require it. we can never get GET == PUT + * here, so this is safe. + */ + chan->dma.free = get - chan->dma.cur - 1; + } + + return 0; +} + diff --git a/drivers/gpu/drm/nouveau/nouveau_dma.h b/drivers/gpu/drm/nouveau/nouveau_dma.h new file mode 100644 index 00000000..23d4edf9 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_dma.h @@ -0,0 +1,175 @@ +/* + * Copyright (C) 2007 Ben Skeggs. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __NOUVEAU_DMA_H__ +#define __NOUVEAU_DMA_H__ + +#ifndef NOUVEAU_DMA_DEBUG +#define NOUVEAU_DMA_DEBUG 0 +#endif + +void nv50_dma_push(struct nouveau_channel *, struct nouveau_bo *, + int delta, int length); + +/* + * There's a hw race condition where you can't jump to your PUT offset, + * to avoid this we jump to offset + SKIPS and fill the difference with + * NOPs. + * + * xf86-video-nv configures the DMA fetch size to 32 bytes, and uses + * a SKIPS value of 8. Lets assume that the race condition is to do + * with writing into the fetch area, we configure a fetch size of 128 + * bytes so we need a larger SKIPS value. + */ +#define NOUVEAU_DMA_SKIPS (128 / 4) + +/* Hardcoded object assignments to subchannels (subchannel id). */ +enum { + NvSubM2MF = 0, + NvSubSw = 1, + NvSub2D = 2, + NvSubCtxSurf2D = 2, + NvSubGdiRect = 3, + NvSubImageBlit = 4 +}; + +/* Object handles. */ +enum { + NvM2MF = 0x80000001, + NvDmaFB = 0x80000002, + NvDmaTT = 0x80000003, + NvNotify0 = 0x80000006, + Nv2D = 0x80000007, + NvCtxSurf2D = 0x80000008, + NvRop = 0x80000009, + NvImagePatt = 0x8000000a, + NvClipRect = 0x8000000b, + NvGdiRect = 0x8000000c, + NvImageBlit = 0x8000000d, + NvSw = 0x8000000e, + NvSema = 0x8000000f, + NvEvoSema0 = 0x80000010, + NvEvoSema1 = 0x80000011, + + /* G80+ display objects */ + NvEvoVRAM = 0x01000000, + NvEvoFB16 = 0x01000001, + NvEvoFB32 = 0x01000002, + NvEvoVRAM_LP = 0x01000003, + NvEvoSync = 0xcafe0000 +}; + +#define NV_MEMORY_TO_MEMORY_FORMAT 0x00000039 +#define NV_MEMORY_TO_MEMORY_FORMAT_NAME 0x00000000 +#define NV_MEMORY_TO_MEMORY_FORMAT_SET_REF 0x00000050 +#define NV_MEMORY_TO_MEMORY_FORMAT_NOP 0x00000100 +#define NV_MEMORY_TO_MEMORY_FORMAT_NOTIFY 0x00000104 +#define NV_MEMORY_TO_MEMORY_FORMAT_NOTIFY_STYLE_WRITE 0x00000000 +#define NV_MEMORY_TO_MEMORY_FORMAT_NOTIFY_STYLE_WRITE_LE_AWAKEN 0x00000001 +#define NV_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY 0x00000180 +#define NV_MEMORY_TO_MEMORY_FORMAT_DMA_SOURCE 0x00000184 +#define NV_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN 0x0000030c + +#define NV50_MEMORY_TO_MEMORY_FORMAT 0x00005039 +#define NV50_MEMORY_TO_MEMORY_FORMAT_UNK200 0x00000200 +#define NV50_MEMORY_TO_MEMORY_FORMAT_UNK21C 0x0000021c +#define NV50_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN_HIGH 0x00000238 +#define NV50_MEMORY_TO_MEMORY_FORMAT_OFFSET_OUT_HIGH 0x0000023c + +static __must_check inline int +RING_SPACE(struct nouveau_channel *chan, int size) +{ + int ret; + + ret = nouveau_dma_wait(chan, 1, size); + if (ret) + return ret; + + chan->dma.free -= size; + return 0; +} + +static inline void +OUT_RING(struct nouveau_channel *chan, int data) +{ + if (NOUVEAU_DMA_DEBUG) { + NV_INFO(chan->dev, "Ch%d/0x%08x: 0x%08x\n", + chan->id, chan->dma.cur << 2, data); + } + + nouveau_bo_wr32(chan->pushbuf_bo, chan->dma.cur++, data); +} + +extern void +OUT_RINGp(struct nouveau_channel *chan, const void *data, unsigned nr_dwords); + +static inline void +BEGIN_NVC0(struct nouveau_channel *chan, int op, int subc, int mthd, int size) +{ + OUT_RING(chan, (op << 28) | (size << 16) | (subc << 13) | (mthd >> 2)); +} + +static inline void +BEGIN_RING(struct nouveau_channel *chan, int subc, int mthd, int size) +{ + OUT_RING(chan, (subc << 13) | (size << 18) | mthd); +} + +#define WRITE_PUT(val) do { \ + DRM_MEMORYBARRIER(); \ + nouveau_bo_rd32(chan->pushbuf_bo, 0); \ + nvchan_wr32(chan, chan->user_put, ((val) << 2) + chan->pushbuf_base); \ +} while (0) + +static inline void +FIRE_RING(struct nouveau_channel *chan) +{ + if (NOUVEAU_DMA_DEBUG) { + NV_INFO(chan->dev, "Ch%d/0x%08x: PUSH!\n", + chan->id, chan->dma.cur << 2); + } + + if (chan->dma.cur == chan->dma.put) + return; + chan->accel_done = true; + + if (chan->dma.ib_max) { + nv50_dma_push(chan, chan->pushbuf_bo, chan->dma.put << 2, + (chan->dma.cur - chan->dma.put) << 2); + } else { + WRITE_PUT(chan->dma.cur); + } + + chan->dma.put = chan->dma.cur; +} + +static inline void +WIND_RING(struct nouveau_channel *chan) +{ + chan->dma.cur = chan->dma.put; +} + +#endif diff --git a/drivers/gpu/drm/nouveau/nouveau_dp.c b/drivers/gpu/drm/nouveau/nouveau_dp.c new file mode 100644 index 00000000..7beb82a0 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_dp.c @@ -0,0 +1,637 @@ +/* + * Copyright 2009 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include "drmP.h" + +#include "nouveau_drv.h" +#include "nouveau_i2c.h" +#include "nouveau_connector.h" +#include "nouveau_encoder.h" + +static int +auxch_rd(struct drm_encoder *encoder, int address, uint8_t *buf, int size) +{ + struct drm_device *dev = encoder->dev; + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + struct nouveau_i2c_chan *auxch; + int ret; + + auxch = nouveau_i2c_find(dev, nv_encoder->dcb->i2c_index); + if (!auxch) + return -ENODEV; + + ret = nouveau_dp_auxch(auxch, 9, address, buf, size); + if (ret) + return ret; + + return 0; +} + +static int +auxch_wr(struct drm_encoder *encoder, int address, uint8_t *buf, int size) +{ + struct drm_device *dev = encoder->dev; + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + struct nouveau_i2c_chan *auxch; + int ret; + + auxch = nouveau_i2c_find(dev, nv_encoder->dcb->i2c_index); + if (!auxch) + return -ENODEV; + + ret = nouveau_dp_auxch(auxch, 8, address, buf, size); + return ret; +} + +static int +nouveau_dp_lane_count_set(struct drm_encoder *encoder, uint8_t cmd) +{ + struct drm_device *dev = encoder->dev; + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + uint32_t tmp; + int or = nv_encoder->or, link = !(nv_encoder->dcb->sorconf.link & 1); + + tmp = nv_rd32(dev, NV50_SOR_DP_CTRL(or, link)); + tmp &= ~(NV50_SOR_DP_CTRL_ENHANCED_FRAME_ENABLED | + NV50_SOR_DP_CTRL_LANE_MASK); + tmp |= ((1 << (cmd & DP_LANE_COUNT_MASK)) - 1) << 16; + if (cmd & DP_LANE_COUNT_ENHANCED_FRAME_EN) + tmp |= NV50_SOR_DP_CTRL_ENHANCED_FRAME_ENABLED; + nv_wr32(dev, NV50_SOR_DP_CTRL(or, link), tmp); + + return auxch_wr(encoder, DP_LANE_COUNT_SET, &cmd, 1); +} + +static int +nouveau_dp_link_bw_set(struct drm_encoder *encoder, uint8_t cmd) +{ + struct drm_device *dev = encoder->dev; + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + uint32_t tmp; + int reg = 0x614300 + (nv_encoder->or * 0x800); + + tmp = nv_rd32(dev, reg); + tmp &= 0xfff3ffff; + if (cmd == DP_LINK_BW_2_7) + tmp |= 0x00040000; + nv_wr32(dev, reg, tmp); + + return auxch_wr(encoder, DP_LINK_BW_SET, &cmd, 1); +} + +static int +nouveau_dp_link_train_set(struct drm_encoder *encoder, int pattern) +{ + struct drm_device *dev = encoder->dev; + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + uint32_t tmp; + uint8_t cmd; + int or = nv_encoder->or, link = !(nv_encoder->dcb->sorconf.link & 1); + int ret; + + tmp = nv_rd32(dev, NV50_SOR_DP_CTRL(or, link)); + tmp &= ~NV50_SOR_DP_CTRL_TRAINING_PATTERN; + tmp |= (pattern << 24); + nv_wr32(dev, NV50_SOR_DP_CTRL(or, link), tmp); + + ret = auxch_rd(encoder, DP_TRAINING_PATTERN_SET, &cmd, 1); + if (ret) + return ret; + cmd &= ~DP_TRAINING_PATTERN_MASK; + cmd |= (pattern & DP_TRAINING_PATTERN_MASK); + return auxch_wr(encoder, DP_TRAINING_PATTERN_SET, &cmd, 1); +} + +static int +nouveau_dp_max_voltage_swing(struct drm_encoder *encoder) +{ + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + struct drm_device *dev = encoder->dev; + struct bit_displayport_encoder_table_entry *dpse; + struct bit_displayport_encoder_table *dpe; + int i, dpe_headerlen, max_vs = 0; + + dpe = nouveau_bios_dp_table(dev, nv_encoder->dcb, &dpe_headerlen); + if (!dpe) + return false; + dpse = (void *)((char *)dpe + dpe_headerlen); + + for (i = 0; i < dpe_headerlen; i++, dpse++) { + if (dpse->vs_level > max_vs) + max_vs = dpse->vs_level; + } + + return max_vs; +} + +static int +nouveau_dp_max_pre_emphasis(struct drm_encoder *encoder, int vs) +{ + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + struct drm_device *dev = encoder->dev; + struct bit_displayport_encoder_table_entry *dpse; + struct bit_displayport_encoder_table *dpe; + int i, dpe_headerlen, max_pre = 0; + + dpe = nouveau_bios_dp_table(dev, nv_encoder->dcb, &dpe_headerlen); + if (!dpe) + return false; + dpse = (void *)((char *)dpe + dpe_headerlen); + + for (i = 0; i < dpe_headerlen; i++, dpse++) { + if (dpse->vs_level != vs) + continue; + + if (dpse->pre_level > max_pre) + max_pre = dpse->pre_level; + } + + return max_pre; +} + +static bool +nouveau_dp_link_train_adjust(struct drm_encoder *encoder, uint8_t *config) +{ + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + struct drm_device *dev = encoder->dev; + struct bit_displayport_encoder_table *dpe; + int ret, i, dpe_headerlen, vs = 0, pre = 0; + uint8_t request[2]; + + dpe = nouveau_bios_dp_table(dev, nv_encoder->dcb, &dpe_headerlen); + if (!dpe) + return false; + + ret = auxch_rd(encoder, DP_ADJUST_REQUEST_LANE0_1, request, 2); + if (ret) + return false; + + NV_DEBUG_KMS(dev, "\t\tadjust 0x%02x 0x%02x\n", request[0], request[1]); + + /* Keep all lanes at the same level.. */ + for (i = 0; i < nv_encoder->dp.link_nr; i++) { + int lane_req = (request[i >> 1] >> ((i & 1) << 2)) & 0xf; + int lane_vs = lane_req & 3; + int lane_pre = (lane_req >> 2) & 3; + + if (lane_vs > vs) + vs = lane_vs; + if (lane_pre > pre) + pre = lane_pre; + } + + if (vs >= nouveau_dp_max_voltage_swing(encoder)) { + vs = nouveau_dp_max_voltage_swing(encoder); + vs |= 4; + } + + if (pre >= nouveau_dp_max_pre_emphasis(encoder, vs & 3)) { + pre = nouveau_dp_max_pre_emphasis(encoder, vs & 3); + pre |= 4; + } + + /* Update the configuration for all lanes.. */ + for (i = 0; i < nv_encoder->dp.link_nr; i++) + config[i] = (pre << 3) | vs; + + return true; +} + +static bool +nouveau_dp_link_train_commit(struct drm_encoder *encoder, uint8_t *config) +{ + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + struct drm_device *dev = encoder->dev; + struct bit_displayport_encoder_table_entry *dpse; + struct bit_displayport_encoder_table *dpe; + int or = nv_encoder->or, link = !(nv_encoder->dcb->sorconf.link & 1); + int dpe_headerlen, ret, i; + + NV_DEBUG_KMS(dev, "\t\tconfig 0x%02x 0x%02x 0x%02x 0x%02x\n", + config[0], config[1], config[2], config[3]); + + dpe = nouveau_bios_dp_table(dev, nv_encoder->dcb, &dpe_headerlen); + if (!dpe) + return false; + dpse = (void *)((char *)dpe + dpe_headerlen); + + for (i = 0; i < dpe->record_nr; i++, dpse++) { + if (dpse->vs_level == (config[0] & 3) && + dpse->pre_level == ((config[0] >> 3) & 3)) + break; + } + BUG_ON(i == dpe->record_nr); + + for (i = 0; i < nv_encoder->dp.link_nr; i++) { + const int shift[4] = { 16, 8, 0, 24 }; + uint32_t mask = 0xff << shift[i]; + uint32_t reg0, reg1, reg2; + + reg0 = nv_rd32(dev, NV50_SOR_DP_UNK118(or, link)) & ~mask; + reg0 |= (dpse->reg0 << shift[i]); + reg1 = nv_rd32(dev, NV50_SOR_DP_UNK120(or, link)) & ~mask; + reg1 |= (dpse->reg1 << shift[i]); + reg2 = nv_rd32(dev, NV50_SOR_DP_UNK130(or, link)) & 0xffff00ff; + reg2 |= (dpse->reg2 << 8); + nv_wr32(dev, NV50_SOR_DP_UNK118(or, link), reg0); + nv_wr32(dev, NV50_SOR_DP_UNK120(or, link), reg1); + nv_wr32(dev, NV50_SOR_DP_UNK130(or, link), reg2); + } + + ret = auxch_wr(encoder, DP_TRAINING_LANE0_SET, config, 4); + if (ret) + return false; + + return true; +} + +bool +nouveau_dp_link_train(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpio_engine *pgpio = &dev_priv->engine.gpio; + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + struct nouveau_connector *nv_connector; + struct bit_displayport_encoder_table *dpe; + int dpe_headerlen; + uint8_t config[4], status[3]; + bool cr_done, cr_max_vs, eq_done, hpd_state; + int ret = 0, i, tries, voltage; + + NV_DEBUG_KMS(dev, "link training!!\n"); + + nv_connector = nouveau_encoder_connector_get(nv_encoder); + if (!nv_connector) + return false; + + dpe = nouveau_bios_dp_table(dev, nv_encoder->dcb, &dpe_headerlen); + if (!dpe) { + NV_ERROR(dev, "SOR-%d: no DP encoder table!\n", nv_encoder->or); + return false; + } + + /* disable hotplug detect, this flips around on some panels during + * link training. + */ + hpd_state = pgpio->irq_enable(dev, nv_connector->dcb->gpio_tag, false); + + if (dpe->script0) { + NV_DEBUG_KMS(dev, "SOR-%d: running DP script 0\n", nv_encoder->or); + nouveau_bios_run_init_table(dev, le16_to_cpu(dpe->script0), + nv_encoder->dcb); + } + +train: + cr_done = eq_done = false; + + /* set link configuration */ + NV_DEBUG_KMS(dev, "\tbegin train: bw %d, lanes %d\n", + nv_encoder->dp.link_bw, nv_encoder->dp.link_nr); + + ret = nouveau_dp_link_bw_set(encoder, nv_encoder->dp.link_bw); + if (ret) + return false; + + config[0] = nv_encoder->dp.link_nr; + if (nv_encoder->dp.dpcd_version >= 0x11 && + nv_encoder->dp.enhanced_frame) + config[0] |= DP_LANE_COUNT_ENHANCED_FRAME_EN; + + ret = nouveau_dp_lane_count_set(encoder, config[0]); + if (ret) + return false; + + /* clock recovery */ + NV_DEBUG_KMS(dev, "\tbegin cr\n"); + ret = nouveau_dp_link_train_set(encoder, DP_TRAINING_PATTERN_1); + if (ret) + goto stop; + + tries = 0; + voltage = -1; + memset(config, 0x00, sizeof(config)); + for (;;) { + if (!nouveau_dp_link_train_commit(encoder, config)) + break; + + udelay(100); + + ret = auxch_rd(encoder, DP_LANE0_1_STATUS, status, 2); + if (ret) + break; + NV_DEBUG_KMS(dev, "\t\tstatus: 0x%02x 0x%02x\n", + status[0], status[1]); + + cr_done = true; + cr_max_vs = false; + for (i = 0; i < nv_encoder->dp.link_nr; i++) { + int lane = (status[i >> 1] >> ((i & 1) * 4)) & 0xf; + + if (!(lane & DP_LANE_CR_DONE)) { + cr_done = false; + if (config[i] & DP_TRAIN_MAX_PRE_EMPHASIS_REACHED) + cr_max_vs = true; + break; + } + } + + if ((config[0] & DP_TRAIN_VOLTAGE_SWING_MASK) != voltage) { + voltage = config[0] & DP_TRAIN_VOLTAGE_SWING_MASK; + tries = 0; + } + + if (cr_done || cr_max_vs || (++tries == 5)) + break; + + if (!nouveau_dp_link_train_adjust(encoder, config)) + break; + } + + if (!cr_done) + goto stop; + + /* channel equalisation */ + NV_DEBUG_KMS(dev, "\tbegin eq\n"); + ret = nouveau_dp_link_train_set(encoder, DP_TRAINING_PATTERN_2); + if (ret) + goto stop; + + for (tries = 0; tries <= 5; tries++) { + udelay(400); + + ret = auxch_rd(encoder, DP_LANE0_1_STATUS, status, 3); + if (ret) + break; + NV_DEBUG_KMS(dev, "\t\tstatus: 0x%02x 0x%02x\n", + status[0], status[1]); + + eq_done = true; + if (!(status[2] & DP_INTERLANE_ALIGN_DONE)) + eq_done = false; + + for (i = 0; eq_done && i < nv_encoder->dp.link_nr; i++) { + int lane = (status[i >> 1] >> ((i & 1) * 4)) & 0xf; + + if (!(lane & DP_LANE_CR_DONE)) { + cr_done = false; + break; + } + + if (!(lane & DP_LANE_CHANNEL_EQ_DONE) || + !(lane & DP_LANE_SYMBOL_LOCKED)) { + eq_done = false; + break; + } + } + + if (eq_done || !cr_done) + break; + + if (!nouveau_dp_link_train_adjust(encoder, config) || + !nouveau_dp_link_train_commit(encoder, config)) + break; + } + +stop: + /* end link training */ + ret = nouveau_dp_link_train_set(encoder, DP_TRAINING_PATTERN_DISABLE); + if (ret) + return false; + + /* retry at a lower setting, if possible */ + if (!ret && !(eq_done && cr_done)) { + NV_DEBUG_KMS(dev, "\twe failed\n"); + if (nv_encoder->dp.link_bw != DP_LINK_BW_1_62) { + NV_DEBUG_KMS(dev, "retry link training at low rate\n"); + nv_encoder->dp.link_bw = DP_LINK_BW_1_62; + goto train; + } + } + + if (dpe->script1) { + NV_DEBUG_KMS(dev, "SOR-%d: running DP script 1\n", nv_encoder->or); + nouveau_bios_run_init_table(dev, le16_to_cpu(dpe->script1), + nv_encoder->dcb); + } + + /* re-enable hotplug detect */ + pgpio->irq_enable(dev, nv_connector->dcb->gpio_tag, hpd_state); + + return eq_done; +} + +bool +nouveau_dp_detect(struct drm_encoder *encoder) +{ + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + struct drm_device *dev = encoder->dev; + uint8_t dpcd[4]; + int ret; + + ret = auxch_rd(encoder, 0x0000, dpcd, 4); + if (ret) + return false; + + NV_DEBUG_KMS(dev, "encoder: link_bw %d, link_nr %d\n" + "display: link_bw %d, link_nr %d version 0x%02x\n", + nv_encoder->dcb->dpconf.link_bw, + nv_encoder->dcb->dpconf.link_nr, + dpcd[1], dpcd[2] & 0x0f, dpcd[0]); + + nv_encoder->dp.dpcd_version = dpcd[0]; + + nv_encoder->dp.link_bw = dpcd[1]; + if (nv_encoder->dp.link_bw != DP_LINK_BW_1_62 && + !nv_encoder->dcb->dpconf.link_bw) + nv_encoder->dp.link_bw = DP_LINK_BW_1_62; + + nv_encoder->dp.link_nr = dpcd[2] & DP_MAX_LANE_COUNT_MASK; + if (nv_encoder->dp.link_nr > nv_encoder->dcb->dpconf.link_nr) + nv_encoder->dp.link_nr = nv_encoder->dcb->dpconf.link_nr; + + nv_encoder->dp.enhanced_frame = (dpcd[2] & DP_ENHANCED_FRAME_CAP); + + return true; +} + +int +nouveau_dp_auxch(struct nouveau_i2c_chan *auxch, int cmd, int addr, + uint8_t *data, int data_nr) +{ + struct drm_device *dev = auxch->dev; + uint32_t tmp, ctrl, stat = 0, data32[4] = {}; + int ret = 0, i, index = auxch->rd; + + NV_DEBUG_KMS(dev, "ch %d cmd %d addr 0x%x len %d\n", index, cmd, addr, data_nr); + + tmp = nv_rd32(dev, NV50_AUXCH_CTRL(auxch->rd)); + nv_wr32(dev, NV50_AUXCH_CTRL(auxch->rd), tmp | 0x00100000); + tmp = nv_rd32(dev, NV50_AUXCH_CTRL(auxch->rd)); + if (!(tmp & 0x01000000)) { + NV_ERROR(dev, "expected bit 24 == 1, got 0x%08x\n", tmp); + ret = -EIO; + goto out; + } + + for (i = 0; i < 3; i++) { + tmp = nv_rd32(dev, NV50_AUXCH_STAT(auxch->rd)); + if (tmp & NV50_AUXCH_STAT_STATE_READY) + break; + udelay(100); + } + + if (i == 3) { + ret = -EBUSY; + goto out; + } + + if (!(cmd & 1)) { + memcpy(data32, data, data_nr); + for (i = 0; i < 4; i++) { + NV_DEBUG_KMS(dev, "wr %d: 0x%08x\n", i, data32[i]); + nv_wr32(dev, NV50_AUXCH_DATA_OUT(index, i), data32[i]); + } + } + + nv_wr32(dev, NV50_AUXCH_ADDR(index), addr); + ctrl = nv_rd32(dev, NV50_AUXCH_CTRL(index)); + ctrl &= ~(NV50_AUXCH_CTRL_CMD | NV50_AUXCH_CTRL_LEN); + ctrl |= (cmd << NV50_AUXCH_CTRL_CMD_SHIFT); + ctrl |= ((data_nr - 1) << NV50_AUXCH_CTRL_LEN_SHIFT); + + for (i = 0; i < 16; i++) { + nv_wr32(dev, NV50_AUXCH_CTRL(index), ctrl | 0x80000000); + nv_wr32(dev, NV50_AUXCH_CTRL(index), ctrl); + nv_wr32(dev, NV50_AUXCH_CTRL(index), ctrl | 0x00010000); + if (!nv_wait(dev, NV50_AUXCH_CTRL(index), + 0x00010000, 0x00000000)) { + NV_ERROR(dev, "expected bit 16 == 0, got 0x%08x\n", + nv_rd32(dev, NV50_AUXCH_CTRL(index))); + ret = -EBUSY; + goto out; + } + + udelay(400); + + stat = nv_rd32(dev, NV50_AUXCH_STAT(index)); + if ((stat & NV50_AUXCH_STAT_REPLY_AUX) != + NV50_AUXCH_STAT_REPLY_AUX_DEFER) + break; + } + + if (i == 16) { + NV_ERROR(dev, "auxch DEFER too many times, bailing\n"); + ret = -EREMOTEIO; + goto out; + } + + if (cmd & 1) { + if ((stat & NV50_AUXCH_STAT_COUNT) != data_nr) { + ret = -EREMOTEIO; + goto out; + } + + for (i = 0; i < 4; i++) { + data32[i] = nv_rd32(dev, NV50_AUXCH_DATA_IN(index, i)); + NV_DEBUG_KMS(dev, "rd %d: 0x%08x\n", i, data32[i]); + } + memcpy(data, data32, data_nr); + } + +out: + tmp = nv_rd32(dev, NV50_AUXCH_CTRL(auxch->rd)); + nv_wr32(dev, NV50_AUXCH_CTRL(auxch->rd), tmp & ~0x00100000); + tmp = nv_rd32(dev, NV50_AUXCH_CTRL(auxch->rd)); + if (tmp & 0x01000000) { + NV_ERROR(dev, "expected bit 24 == 0, got 0x%08x\n", tmp); + ret = -EIO; + } + + udelay(400); + + return ret ? ret : (stat & NV50_AUXCH_STAT_REPLY); +} + +static int +nouveau_dp_i2c_xfer(struct i2c_adapter *adap, struct i2c_msg *msgs, int num) +{ + struct nouveau_i2c_chan *auxch = (struct nouveau_i2c_chan *)adap; + struct drm_device *dev = auxch->dev; + struct i2c_msg *msg = msgs; + int ret, mcnt = num; + + while (mcnt--) { + u8 remaining = msg->len; + u8 *ptr = msg->buf; + + while (remaining) { + u8 cnt = (remaining > 16) ? 16 : remaining; + u8 cmd; + + if (msg->flags & I2C_M_RD) + cmd = AUX_I2C_READ; + else + cmd = AUX_I2C_WRITE; + + if (mcnt || remaining > 16) + cmd |= AUX_I2C_MOT; + + ret = nouveau_dp_auxch(auxch, cmd, msg->addr, ptr, cnt); + if (ret < 0) + return ret; + + switch (ret & NV50_AUXCH_STAT_REPLY_I2C) { + case NV50_AUXCH_STAT_REPLY_I2C_ACK: + break; + case NV50_AUXCH_STAT_REPLY_I2C_NACK: + return -EREMOTEIO; + case NV50_AUXCH_STAT_REPLY_I2C_DEFER: + udelay(100); + continue; + default: + NV_ERROR(dev, "bad auxch reply: 0x%08x\n", ret); + return -EREMOTEIO; + } + + ptr += cnt; + remaining -= cnt; + } + + msg++; + } + + return num; +} + +static u32 +nouveau_dp_i2c_func(struct i2c_adapter *adap) +{ + return I2C_FUNC_I2C | I2C_FUNC_SMBUS_EMUL; +} + +const struct i2c_algorithm nouveau_dp_i2c_algo = { + .master_xfer = nouveau_dp_i2c_xfer, + .functionality = nouveau_dp_i2c_func +}; diff --git a/drivers/gpu/drm/nouveau/nouveau_drv.c b/drivers/gpu/drm/nouveau/nouveau_drv.c new file mode 100644 index 00000000..02c6f37d --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_drv.c @@ -0,0 +1,479 @@ +/* + * Copyright 2005 Stephane Marchesin. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * VA LINUX SYSTEMS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#include + +#include "drmP.h" +#include "drm.h" +#include "drm_crtc_helper.h" +#include "nouveau_drv.h" +#include "nouveau_hw.h" +#include "nouveau_fb.h" +#include "nouveau_fbcon.h" +#include "nouveau_pm.h" +#include "nv50_display.h" + +#include "drm_pciids.h" + +MODULE_PARM_DESC(agpmode, "AGP mode (0 to disable AGP)"); +int nouveau_agpmode = -1; +module_param_named(agpmode, nouveau_agpmode, int, 0400); + +MODULE_PARM_DESC(modeset, "Enable kernel modesetting"); +static int nouveau_modeset = -1; /* kms */ +module_param_named(modeset, nouveau_modeset, int, 0400); + +MODULE_PARM_DESC(vbios, "Override default VBIOS location"); +char *nouveau_vbios; +module_param_named(vbios, nouveau_vbios, charp, 0400); + +MODULE_PARM_DESC(vram_pushbuf, "Force DMA push buffers to be in VRAM"); +int nouveau_vram_pushbuf; +module_param_named(vram_pushbuf, nouveau_vram_pushbuf, int, 0400); + +MODULE_PARM_DESC(vram_notify, "Force DMA notifiers to be in VRAM"); +int nouveau_vram_notify = 0; +module_param_named(vram_notify, nouveau_vram_notify, int, 0400); + +MODULE_PARM_DESC(duallink, "Allow dual-link TMDS (>=GeForce 8)"); +int nouveau_duallink = 1; +module_param_named(duallink, nouveau_duallink, int, 0400); + +MODULE_PARM_DESC(uscript_lvds, "LVDS output script table ID (>=GeForce 8)"); +int nouveau_uscript_lvds = -1; +module_param_named(uscript_lvds, nouveau_uscript_lvds, int, 0400); + +MODULE_PARM_DESC(uscript_tmds, "TMDS output script table ID (>=GeForce 8)"); +int nouveau_uscript_tmds = -1; +module_param_named(uscript_tmds, nouveau_uscript_tmds, int, 0400); + +MODULE_PARM_DESC(ignorelid, "Ignore ACPI lid status"); +int nouveau_ignorelid = 0; +module_param_named(ignorelid, nouveau_ignorelid, int, 0400); + +MODULE_PARM_DESC(noaccel, "Disable all acceleration"); +int nouveau_noaccel = 0; +module_param_named(noaccel, nouveau_noaccel, int, 0400); + +MODULE_PARM_DESC(nofbaccel, "Disable fbcon acceleration"); +int nouveau_nofbaccel = 0; +module_param_named(nofbaccel, nouveau_nofbaccel, int, 0400); + +MODULE_PARM_DESC(force_post, "Force POST"); +int nouveau_force_post = 0; +module_param_named(force_post, nouveau_force_post, int, 0400); + +MODULE_PARM_DESC(override_conntype, "Ignore DCB connector type"); +int nouveau_override_conntype = 0; +module_param_named(override_conntype, nouveau_override_conntype, int, 0400); + +MODULE_PARM_DESC(tv_disable, "Disable TV-out detection\n"); +int nouveau_tv_disable = 0; +module_param_named(tv_disable, nouveau_tv_disable, int, 0400); + +MODULE_PARM_DESC(tv_norm, "Default TV norm.\n" + "\t\tSupported: PAL, PAL-M, PAL-N, PAL-Nc, NTSC-M, NTSC-J,\n" + "\t\t\thd480i, hd480p, hd576i, hd576p, hd720p, hd1080i.\n" + "\t\tDefault: PAL\n" + "\t\t*NOTE* Ignored for cards with external TV encoders."); +char *nouveau_tv_norm; +module_param_named(tv_norm, nouveau_tv_norm, charp, 0400); + +MODULE_PARM_DESC(reg_debug, "Register access debug bitmask:\n" + "\t\t0x1 mc, 0x2 video, 0x4 fb, 0x8 extdev,\n" + "\t\t0x10 crtc, 0x20 ramdac, 0x40 vgacrtc, 0x80 rmvio,\n" + "\t\t0x100 vgaattr, 0x200 EVO (G80+). "); +int nouveau_reg_debug; +module_param_named(reg_debug, nouveau_reg_debug, int, 0600); + +MODULE_PARM_DESC(perflvl, "Performance level (default: boot)\n"); +char *nouveau_perflvl; +module_param_named(perflvl, nouveau_perflvl, charp, 0400); + +MODULE_PARM_DESC(perflvl_wr, "Allow perflvl changes (warning: dangerous!)\n"); +int nouveau_perflvl_wr; +module_param_named(perflvl_wr, nouveau_perflvl_wr, int, 0400); + +MODULE_PARM_DESC(msi, "Enable MSI (default: off)\n"); +int nouveau_msi; +module_param_named(msi, nouveau_msi, int, 0400); + +int nouveau_fbpercrtc; +#if 0 +module_param_named(fbpercrtc, nouveau_fbpercrtc, int, 0400); +#endif + +static struct pci_device_id pciidlist[] = { + { + PCI_DEVICE(PCI_VENDOR_ID_NVIDIA, PCI_ANY_ID), + .class = PCI_BASE_CLASS_DISPLAY << 16, + .class_mask = 0xff << 16, + }, + { + PCI_DEVICE(PCI_VENDOR_ID_NVIDIA_SGS, PCI_ANY_ID), + .class = PCI_BASE_CLASS_DISPLAY << 16, + .class_mask = 0xff << 16, + }, + {} +}; + +MODULE_DEVICE_TABLE(pci, pciidlist); + +static struct drm_driver driver; + +static int __devinit +nouveau_pci_probe(struct pci_dev *pdev, const struct pci_device_id *ent) +{ + return drm_get_pci_dev(pdev, ent, &driver); +} + +static void +nouveau_pci_remove(struct pci_dev *pdev) +{ + struct drm_device *dev = pci_get_drvdata(pdev); + + drm_put_dev(dev); +} + +int +nouveau_pci_suspend(struct pci_dev *pdev, pm_message_t pm_state) +{ + struct drm_device *dev = pci_get_drvdata(pdev); + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_instmem_engine *pinstmem = &dev_priv->engine.instmem; + struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; + struct nouveau_channel *chan; + struct drm_crtc *crtc; + int ret, i, e; + + if (pm_state.event == PM_EVENT_PRETHAW) + return 0; + + if (dev->switch_power_state == DRM_SWITCH_POWER_OFF) + return 0; + + NV_INFO(dev, "Disabling fbcon acceleration...\n"); + nouveau_fbcon_save_disable_accel(dev); + + NV_INFO(dev, "Unpinning framebuffer(s)...\n"); + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { + struct nouveau_framebuffer *nouveau_fb; + + nouveau_fb = nouveau_framebuffer(crtc->fb); + if (!nouveau_fb || !nouveau_fb->nvbo) + continue; + + nouveau_bo_unpin(nouveau_fb->nvbo); + } + + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + + nouveau_bo_unmap(nv_crtc->cursor.nvbo); + nouveau_bo_unpin(nv_crtc->cursor.nvbo); + } + + NV_INFO(dev, "Evicting buffers...\n"); + ttm_bo_evict_mm(&dev_priv->ttm.bdev, TTM_PL_VRAM); + + NV_INFO(dev, "Idling channels...\n"); + for (i = 0; i < pfifo->channels; i++) { + chan = dev_priv->channels.ptr[i]; + + if (chan && chan->pushbuf_bo) + nouveau_channel_idle(chan); + } + + pfifo->reassign(dev, false); + pfifo->disable(dev); + pfifo->unload_context(dev); + + for (e = NVOBJ_ENGINE_NR - 1; e >= 0; e--) { + if (dev_priv->eng[e]) { + ret = dev_priv->eng[e]->fini(dev, e); + if (ret) + goto out_abort; + } + } + + ret = pinstmem->suspend(dev); + if (ret) { + NV_ERROR(dev, "... failed: %d\n", ret); + goto out_abort; + } + + NV_INFO(dev, "Suspending GPU objects...\n"); + ret = nouveau_gpuobj_suspend(dev); + if (ret) { + NV_ERROR(dev, "... failed: %d\n", ret); + pinstmem->resume(dev); + goto out_abort; + } + + NV_INFO(dev, "And we're gone!\n"); + pci_save_state(pdev); + if (pm_state.event == PM_EVENT_SUSPEND) { + pci_disable_device(pdev); + pci_set_power_state(pdev, PCI_D3hot); + } + + console_lock(); + nouveau_fbcon_set_suspend(dev, 1); + console_unlock(); + nouveau_fbcon_restore_accel(dev); + return 0; + +out_abort: + NV_INFO(dev, "Re-enabling acceleration..\n"); + for (e = e + 1; e < NVOBJ_ENGINE_NR; e++) { + if (dev_priv->eng[e]) + dev_priv->eng[e]->init(dev, e); + } + pfifo->enable(dev); + pfifo->reassign(dev, true); + return ret; +} + +int +nouveau_pci_resume(struct pci_dev *pdev) +{ + struct drm_device *dev = pci_get_drvdata(pdev); + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_engine *engine = &dev_priv->engine; + struct drm_crtc *crtc; + int ret, i; + + if (dev->switch_power_state == DRM_SWITCH_POWER_OFF) + return 0; + + nouveau_fbcon_save_disable_accel(dev); + + NV_INFO(dev, "We're back, enabling device...\n"); + pci_set_power_state(pdev, PCI_D0); + pci_restore_state(pdev); + if (pci_enable_device(pdev)) + return -1; + pci_set_master(dev->pdev); + + /* Make sure the AGP controller is in a consistent state */ + if (dev_priv->gart_info.type == NOUVEAU_GART_AGP) + nouveau_mem_reset_agp(dev); + + /* Make the CRTCs accessible */ + engine->display.early_init(dev); + + NV_INFO(dev, "POSTing device...\n"); + ret = nouveau_run_vbios_init(dev); + if (ret) + return ret; + + nouveau_pm_resume(dev); + + if (dev_priv->gart_info.type == NOUVEAU_GART_AGP) { + ret = nouveau_mem_init_agp(dev); + if (ret) { + NV_ERROR(dev, "error reinitialising AGP: %d\n", ret); + return ret; + } + } + + NV_INFO(dev, "Restoring GPU objects...\n"); + nouveau_gpuobj_resume(dev); + + NV_INFO(dev, "Reinitialising engines...\n"); + engine->instmem.resume(dev); + engine->mc.init(dev); + engine->timer.init(dev); + engine->fb.init(dev); + for (i = 0; i < NVOBJ_ENGINE_NR; i++) { + if (dev_priv->eng[i]) + dev_priv->eng[i]->init(dev, i); + } + engine->fifo.init(dev); + + nouveau_irq_postinstall(dev); + + /* Re-write SKIPS, they'll have been lost over the suspend */ + if (nouveau_vram_pushbuf) { + struct nouveau_channel *chan; + int j; + + for (i = 0; i < dev_priv->engine.fifo.channels; i++) { + chan = dev_priv->channels.ptr[i]; + if (!chan || !chan->pushbuf_bo) + continue; + + for (j = 0; j < NOUVEAU_DMA_SKIPS; j++) + nouveau_bo_wr32(chan->pushbuf_bo, i, 0); + } + } + + NV_INFO(dev, "Restoring mode...\n"); + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { + struct nouveau_framebuffer *nouveau_fb; + + nouveau_fb = nouveau_framebuffer(crtc->fb); + if (!nouveau_fb || !nouveau_fb->nvbo) + continue; + + nouveau_bo_pin(nouveau_fb->nvbo, TTM_PL_FLAG_VRAM); + } + + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + + ret = nouveau_bo_pin(nv_crtc->cursor.nvbo, TTM_PL_FLAG_VRAM); + if (!ret) + ret = nouveau_bo_map(nv_crtc->cursor.nvbo); + if (ret) + NV_ERROR(dev, "Could not pin/map cursor.\n"); + } + + engine->display.init(dev); + + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + u32 offset = nv_crtc->cursor.nvbo->bo.mem.start << PAGE_SHIFT; + + nv_crtc->cursor.set_offset(nv_crtc, offset); + nv_crtc->cursor.set_pos(nv_crtc, nv_crtc->cursor_saved_x, + nv_crtc->cursor_saved_y); + } + + /* Force CLUT to get re-loaded during modeset */ + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + + nv_crtc->lut.depth = 0; + } + + console_lock(); + nouveau_fbcon_set_suspend(dev, 0); + console_unlock(); + + nouveau_fbcon_zfill_all(dev); + + drm_helper_resume_force_mode(dev); + + nouveau_fbcon_restore_accel(dev); + return 0; +} + +static struct drm_driver driver = { + .driver_features = + DRIVER_USE_AGP | DRIVER_PCI_DMA | DRIVER_SG | + DRIVER_HAVE_IRQ | DRIVER_IRQ_SHARED | DRIVER_GEM | + DRIVER_MODESET, + .load = nouveau_load, + .firstopen = nouveau_firstopen, + .lastclose = nouveau_lastclose, + .unload = nouveau_unload, + .preclose = nouveau_preclose, +#if defined(CONFIG_DRM_NOUVEAU_DEBUG) + .debugfs_init = nouveau_debugfs_init, + .debugfs_cleanup = nouveau_debugfs_takedown, +#endif + .irq_preinstall = nouveau_irq_preinstall, + .irq_postinstall = nouveau_irq_postinstall, + .irq_uninstall = nouveau_irq_uninstall, + .irq_handler = nouveau_irq_handler, + .get_vblank_counter = drm_vblank_count, + .enable_vblank = nouveau_vblank_enable, + .disable_vblank = nouveau_vblank_disable, + .reclaim_buffers = drm_core_reclaim_buffers, + .ioctls = nouveau_ioctls, + .fops = { + .owner = THIS_MODULE, + .open = drm_open, + .release = drm_release, + .unlocked_ioctl = drm_ioctl, + .mmap = nouveau_ttm_mmap, + .poll = drm_poll, + .fasync = drm_fasync, + .read = drm_read, +#if defined(CONFIG_COMPAT) + .compat_ioctl = nouveau_compat_ioctl, +#endif + .llseek = noop_llseek, + }, + + .gem_init_object = nouveau_gem_object_new, + .gem_free_object = nouveau_gem_object_del, + + .name = DRIVER_NAME, + .desc = DRIVER_DESC, +#ifdef GIT_REVISION + .date = GIT_REVISION, +#else + .date = DRIVER_DATE, +#endif + .major = DRIVER_MAJOR, + .minor = DRIVER_MINOR, + .patchlevel = DRIVER_PATCHLEVEL, +}; + +static struct pci_driver nouveau_pci_driver = { + .name = DRIVER_NAME, + .id_table = pciidlist, + .probe = nouveau_pci_probe, + .remove = nouveau_pci_remove, + .suspend = nouveau_pci_suspend, + .resume = nouveau_pci_resume +}; + +static int __init nouveau_init(void) +{ + driver.num_ioctls = nouveau_max_ioctl; + + if (nouveau_modeset == -1) { +#ifdef CONFIG_VGA_CONSOLE + if (vgacon_text_force()) + nouveau_modeset = 0; + else +#endif + nouveau_modeset = 1; + } + + if (!nouveau_modeset) + return 0; + + nouveau_register_dsm_handler(); + return drm_pci_init(&driver, &nouveau_pci_driver); +} + +static void __exit nouveau_exit(void) +{ + if (!nouveau_modeset) + return; + + drm_pci_exit(&driver, &nouveau_pci_driver); + nouveau_unregister_dsm_handler(); +} + +module_init(nouveau_init); +module_exit(nouveau_exit); + +MODULE_AUTHOR(DRIVER_AUTHOR); +MODULE_DESCRIPTION(DRIVER_DESC); +MODULE_LICENSE("GPL and additional rights"); diff --git a/drivers/gpu/drm/nouveau/nouveau_drv.h b/drivers/gpu/drm/nouveau/nouveau_drv.h new file mode 100644 index 00000000..9c563319 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_drv.h @@ -0,0 +1,1583 @@ +/* + * Copyright 2005 Stephane Marchesin. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * VA LINUX SYSTEMS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#ifndef __NOUVEAU_DRV_H__ +#define __NOUVEAU_DRV_H__ + +#define DRIVER_AUTHOR "Stephane Marchesin" +#define DRIVER_EMAIL "dri-devel@lists.sourceforge.net" + +#define DRIVER_NAME "nouveau" +#define DRIVER_DESC "nVidia Riva/TNT/GeForce" +#define DRIVER_DATE "20090420" + +#define DRIVER_MAJOR 0 +#define DRIVER_MINOR 0 +#define DRIVER_PATCHLEVEL 16 + +#define NOUVEAU_FAMILY 0x0000FFFF +#define NOUVEAU_FLAGS 0xFFFF0000 + +#include "ttm/ttm_bo_api.h" +#include "ttm/ttm_bo_driver.h" +#include "ttm/ttm_placement.h" +#include "ttm/ttm_memory.h" +#include "ttm/ttm_module.h" + +struct nouveau_fpriv { + struct ttm_object_file *tfile; +}; + +#define DRM_FILE_PAGE_OFFSET (0x100000000ULL >> PAGE_SHIFT) + +#include "nouveau_drm.h" +#include "nouveau_reg.h" +#include "nouveau_bios.h" +#include "nouveau_util.h" + +struct nouveau_grctx; +struct nouveau_mem; +#include "nouveau_vm.h" + +#define MAX_NUM_DCB_ENTRIES 16 + +#define NOUVEAU_MAX_CHANNEL_NR 128 +#define NOUVEAU_MAX_TILE_NR 15 + +struct nouveau_mem { + struct drm_device *dev; + + struct nouveau_vma bar_vma; + struct nouveau_vma tmp_vma; + u8 page_shift; + + struct drm_mm_node *tag; + struct list_head regions; + dma_addr_t *pages; + u32 memtype; + u64 offset; + u64 size; +}; + +struct nouveau_tile_reg { + bool used; + uint32_t addr; + uint32_t limit; + uint32_t pitch; + uint32_t zcomp; + struct drm_mm_node *tag_mem; + struct nouveau_fence *fence; +}; + +struct nouveau_bo { + struct ttm_buffer_object bo; + struct ttm_placement placement; + u32 valid_domains; + u32 placements[3]; + u32 busy_placements[3]; + struct ttm_bo_kmap_obj kmap; + struct list_head head; + + /* protected by ttm_bo_reserve() */ + struct drm_file *reserved_by; + struct list_head entry; + int pbbo_index; + bool validate_mapped; + + struct nouveau_channel *channel; + + struct nouveau_vma vma; + + uint32_t tile_mode; + uint32_t tile_flags; + struct nouveau_tile_reg *tile; + + struct drm_gem_object *gem; + int pin_refcnt; +}; + +#define nouveau_bo_tile_layout(nvbo) \ + ((nvbo)->tile_flags & NOUVEAU_GEM_TILE_LAYOUT_MASK) + +static inline struct nouveau_bo * +nouveau_bo(struct ttm_buffer_object *bo) +{ + return container_of(bo, struct nouveau_bo, bo); +} + +static inline struct nouveau_bo * +nouveau_gem_object(struct drm_gem_object *gem) +{ + return gem ? gem->driver_private : NULL; +} + +/* TODO: submit equivalent to TTM generic API upstream? */ +static inline void __iomem * +nvbo_kmap_obj_iovirtual(struct nouveau_bo *nvbo) +{ + bool is_iomem; + void __iomem *ioptr = (void __force __iomem *)ttm_kmap_obj_virtual( + &nvbo->kmap, &is_iomem); + WARN_ON_ONCE(ioptr && !is_iomem); + return ioptr; +} + +enum nouveau_flags { + NV_NFORCE = 0x10000000, + NV_NFORCE2 = 0x20000000 +}; + +#define NVOBJ_ENGINE_SW 0 +#define NVOBJ_ENGINE_GR 1 +#define NVOBJ_ENGINE_CRYPT 2 +#define NVOBJ_ENGINE_COPY0 3 +#define NVOBJ_ENGINE_COPY1 4 +#define NVOBJ_ENGINE_MPEG 5 +#define NVOBJ_ENGINE_DISPLAY 15 +#define NVOBJ_ENGINE_NR 16 + +#define NVOBJ_FLAG_DONT_MAP (1 << 0) +#define NVOBJ_FLAG_ZERO_ALLOC (1 << 1) +#define NVOBJ_FLAG_ZERO_FREE (1 << 2) +#define NVOBJ_FLAG_VM (1 << 3) +#define NVOBJ_FLAG_VM_USER (1 << 4) + +#define NVOBJ_CINST_GLOBAL 0xdeadbeef + +struct nouveau_gpuobj { + struct drm_device *dev; + struct kref refcount; + struct list_head list; + + void *node; + u32 *suspend; + + uint32_t flags; + + u32 size; + u32 pinst; + u32 cinst; + u64 vinst; + + uint32_t engine; + uint32_t class; + + void (*dtor)(struct drm_device *, struct nouveau_gpuobj *); + void *priv; +}; + +struct nouveau_page_flip_state { + struct list_head head; + struct drm_pending_vblank_event *event; + int crtc, bpp, pitch, x, y; + uint64_t offset; +}; + +enum nouveau_channel_mutex_class { + NOUVEAU_UCHANNEL_MUTEX, + NOUVEAU_KCHANNEL_MUTEX +}; + +struct nouveau_channel { + struct drm_device *dev; + int id; + + /* references to the channel data structure */ + struct kref ref; + /* users of the hardware channel resources, the hardware + * context will be kicked off when it reaches zero. */ + atomic_t users; + struct mutex mutex; + + /* owner of this fifo */ + struct drm_file *file_priv; + /* mapping of the fifo itself */ + struct drm_local_map *map; + + /* mapping of the regs controlling the fifo */ + void __iomem *user; + uint32_t user_get; + uint32_t user_put; + + /* Fencing */ + struct { + /* lock protects the pending list only */ + spinlock_t lock; + struct list_head pending; + uint32_t sequence; + uint32_t sequence_ack; + atomic_t last_sequence_irq; + } fence; + + /* DMA push buffer */ + struct nouveau_gpuobj *pushbuf; + struct nouveau_bo *pushbuf_bo; + uint32_t pushbuf_base; + + /* Notifier memory */ + struct nouveau_bo *notifier_bo; + struct drm_mm notifier_heap; + + /* PFIFO context */ + struct nouveau_gpuobj *ramfc; + struct nouveau_gpuobj *cache; + void *fifo_priv; + + /* Execution engine contexts */ + void *engctx[NVOBJ_ENGINE_NR]; + + /* NV50 VM */ + struct nouveau_vm *vm; + struct nouveau_gpuobj *vm_pd; + + /* Objects */ + struct nouveau_gpuobj *ramin; /* Private instmem */ + struct drm_mm ramin_heap; /* Private PRAMIN heap */ + struct nouveau_ramht *ramht; /* Hash table */ + + /* GPU object info for stuff used in-kernel (mm_enabled) */ + uint32_t m2mf_ntfy; + uint32_t vram_handle; + uint32_t gart_handle; + bool accel_done; + + /* Push buffer state (only for drm's channel on !mm_enabled) */ + struct { + int max; + int free; + int cur; + int put; + /* access via pushbuf_bo */ + + int ib_base; + int ib_max; + int ib_free; + int ib_put; + } dma; + + uint32_t sw_subchannel[8]; + + struct { + struct nouveau_gpuobj *vblsem; + uint32_t vblsem_head; + uint32_t vblsem_offset; + uint32_t vblsem_rval; + struct list_head vbl_wait; + struct list_head flip; + } nvsw; + + struct { + bool active; + char name[32]; + struct drm_info_list info; + } debugfs; +}; + +struct nouveau_exec_engine { + void (*destroy)(struct drm_device *, int engine); + int (*init)(struct drm_device *, int engine); + int (*fini)(struct drm_device *, int engine); + int (*context_new)(struct nouveau_channel *, int engine); + void (*context_del)(struct nouveau_channel *, int engine); + int (*object_new)(struct nouveau_channel *, int engine, + u32 handle, u16 class); + void (*set_tile_region)(struct drm_device *dev, int i); + void (*tlb_flush)(struct drm_device *, int engine); +}; + +struct nouveau_instmem_engine { + void *priv; + + int (*init)(struct drm_device *dev); + void (*takedown)(struct drm_device *dev); + int (*suspend)(struct drm_device *dev); + void (*resume)(struct drm_device *dev); + + int (*get)(struct nouveau_gpuobj *, u32 size, u32 align); + void (*put)(struct nouveau_gpuobj *); + int (*map)(struct nouveau_gpuobj *); + void (*unmap)(struct nouveau_gpuobj *); + + void (*flush)(struct drm_device *); +}; + +struct nouveau_mc_engine { + int (*init)(struct drm_device *dev); + void (*takedown)(struct drm_device *dev); +}; + +struct nouveau_timer_engine { + int (*init)(struct drm_device *dev); + void (*takedown)(struct drm_device *dev); + uint64_t (*read)(struct drm_device *dev); +}; + +struct nouveau_fb_engine { + int num_tiles; + struct drm_mm tag_heap; + void *priv; + + int (*init)(struct drm_device *dev); + void (*takedown)(struct drm_device *dev); + + void (*init_tile_region)(struct drm_device *dev, int i, + uint32_t addr, uint32_t size, + uint32_t pitch, uint32_t flags); + void (*set_tile_region)(struct drm_device *dev, int i); + void (*free_tile_region)(struct drm_device *dev, int i); +}; + +struct nouveau_fifo_engine { + void *priv; + int channels; + + struct nouveau_gpuobj *playlist[2]; + int cur_playlist; + + int (*init)(struct drm_device *); + void (*takedown)(struct drm_device *); + + void (*disable)(struct drm_device *); + void (*enable)(struct drm_device *); + bool (*reassign)(struct drm_device *, bool enable); + bool (*cache_pull)(struct drm_device *dev, bool enable); + + int (*channel_id)(struct drm_device *); + + int (*create_context)(struct nouveau_channel *); + void (*destroy_context)(struct nouveau_channel *); + int (*load_context)(struct nouveau_channel *); + int (*unload_context)(struct drm_device *); + void (*tlb_flush)(struct drm_device *dev); +}; + +struct nouveau_display_engine { + void *priv; + int (*early_init)(struct drm_device *); + void (*late_takedown)(struct drm_device *); + int (*create)(struct drm_device *); + int (*init)(struct drm_device *); + void (*destroy)(struct drm_device *); +}; + +struct nouveau_gpio_engine { + void *priv; + + int (*init)(struct drm_device *); + void (*takedown)(struct drm_device *); + + int (*get)(struct drm_device *, enum dcb_gpio_tag); + int (*set)(struct drm_device *, enum dcb_gpio_tag, int state); + + int (*irq_register)(struct drm_device *, enum dcb_gpio_tag, + void (*)(void *, int), void *); + void (*irq_unregister)(struct drm_device *, enum dcb_gpio_tag, + void (*)(void *, int), void *); + bool (*irq_enable)(struct drm_device *, enum dcb_gpio_tag, bool on); +}; + +struct nouveau_pm_voltage_level { + u8 voltage; + u8 vid; +}; + +struct nouveau_pm_voltage { + bool supported; + u8 vid_mask; + + struct nouveau_pm_voltage_level *level; + int nr_level; +}; + +struct nouveau_pm_memtiming { + int id; + u32 reg_100220; + u32 reg_100224; + u32 reg_100228; + u32 reg_10022c; + u32 reg_100230; + u32 reg_100234; + u32 reg_100238; + u32 reg_10023c; + u32 reg_100240; +}; + +#define NOUVEAU_PM_MAX_LEVEL 8 +struct nouveau_pm_level { + struct device_attribute dev_attr; + char name[32]; + int id; + + u32 core; + u32 memory; + u32 shader; + u32 unk05; + u32 unk0a; + + u8 voltage; + u8 fanspeed; + + u16 memscript; + struct nouveau_pm_memtiming *timing; +}; + +struct nouveau_pm_temp_sensor_constants { + u16 offset_constant; + s16 offset_mult; + u16 offset_div; + u16 slope_mult; + u16 slope_div; +}; + +struct nouveau_pm_threshold_temp { + s16 critical; + s16 down_clock; + s16 fan_boost; +}; + +struct nouveau_pm_memtimings { + bool supported; + struct nouveau_pm_memtiming *timing; + int nr_timing; +}; + +struct nouveau_pm_engine { + struct nouveau_pm_voltage voltage; + struct nouveau_pm_level perflvl[NOUVEAU_PM_MAX_LEVEL]; + int nr_perflvl; + struct nouveau_pm_memtimings memtimings; + struct nouveau_pm_temp_sensor_constants sensor_constants; + struct nouveau_pm_threshold_temp threshold_temp; + + struct nouveau_pm_level boot; + struct nouveau_pm_level *cur; + + struct device *hwmon; + struct notifier_block acpi_nb; + + int (*clock_get)(struct drm_device *, u32 id); + void *(*clock_pre)(struct drm_device *, struct nouveau_pm_level *, + u32 id, int khz); + void (*clock_set)(struct drm_device *, void *); + int (*voltage_get)(struct drm_device *); + int (*voltage_set)(struct drm_device *, int voltage); + int (*fanspeed_get)(struct drm_device *); + int (*fanspeed_set)(struct drm_device *, int fanspeed); + int (*temp_get)(struct drm_device *); +}; + +struct nouveau_vram_engine { + int (*init)(struct drm_device *); + int (*get)(struct drm_device *, u64, u32 align, u32 size_nc, + u32 type, struct nouveau_mem **); + void (*put)(struct drm_device *, struct nouveau_mem **); + + bool (*flags_valid)(struct drm_device *, u32 tile_flags); +}; + +struct nouveau_engine { + struct nouveau_instmem_engine instmem; + struct nouveau_mc_engine mc; + struct nouveau_timer_engine timer; + struct nouveau_fb_engine fb; + struct nouveau_fifo_engine fifo; + struct nouveau_display_engine display; + struct nouveau_gpio_engine gpio; + struct nouveau_pm_engine pm; + struct nouveau_vram_engine vram; +}; + +struct nouveau_pll_vals { + union { + struct { +#ifdef __BIG_ENDIAN + uint8_t N1, M1, N2, M2; +#else + uint8_t M1, N1, M2, N2; +#endif + }; + struct { + uint16_t NM1, NM2; + } __attribute__((packed)); + }; + int log2P; + + int refclk; +}; + +enum nv04_fp_display_regs { + FP_DISPLAY_END, + FP_TOTAL, + FP_CRTC, + FP_SYNC_START, + FP_SYNC_END, + FP_VALID_START, + FP_VALID_END +}; + +struct nv04_crtc_reg { + unsigned char MiscOutReg; + uint8_t CRTC[0xa0]; + uint8_t CR58[0x10]; + uint8_t Sequencer[5]; + uint8_t Graphics[9]; + uint8_t Attribute[21]; + unsigned char DAC[768]; + + /* PCRTC regs */ + uint32_t fb_start; + uint32_t crtc_cfg; + uint32_t cursor_cfg; + uint32_t gpio_ext; + uint32_t crtc_830; + uint32_t crtc_834; + uint32_t crtc_850; + uint32_t crtc_eng_ctrl; + + /* PRAMDAC regs */ + uint32_t nv10_cursync; + struct nouveau_pll_vals pllvals; + uint32_t ramdac_gen_ctrl; + uint32_t ramdac_630; + uint32_t ramdac_634; + uint32_t tv_setup; + uint32_t tv_vtotal; + uint32_t tv_vskew; + uint32_t tv_vsync_delay; + uint32_t tv_htotal; + uint32_t tv_hskew; + uint32_t tv_hsync_delay; + uint32_t tv_hsync_delay2; + uint32_t fp_horiz_regs[7]; + uint32_t fp_vert_regs[7]; + uint32_t dither; + uint32_t fp_control; + uint32_t dither_regs[6]; + uint32_t fp_debug_0; + uint32_t fp_debug_1; + uint32_t fp_debug_2; + uint32_t fp_margin_color; + uint32_t ramdac_8c0; + uint32_t ramdac_a20; + uint32_t ramdac_a24; + uint32_t ramdac_a34; + uint32_t ctv_regs[38]; +}; + +struct nv04_output_reg { + uint32_t output; + int head; +}; + +struct nv04_mode_state { + struct nv04_crtc_reg crtc_reg[2]; + uint32_t pllsel; + uint32_t sel_clk; +}; + +enum nouveau_card_type { + NV_04 = 0x00, + NV_10 = 0x10, + NV_20 = 0x20, + NV_30 = 0x30, + NV_40 = 0x40, + NV_50 = 0x50, + NV_C0 = 0xc0, +}; + +struct drm_nouveau_private { + struct drm_device *dev; + + /* the card type, takes NV_* as values */ + enum nouveau_card_type card_type; + /* exact chipset, derived from NV_PMC_BOOT_0 */ + int chipset; + int stepping; + int flags; + + void __iomem *mmio; + + spinlock_t ramin_lock; + void __iomem *ramin; + u32 ramin_size; + u32 ramin_base; + bool ramin_available; + struct drm_mm ramin_heap; + struct nouveau_exec_engine *eng[NVOBJ_ENGINE_NR]; + struct list_head gpuobj_list; + struct list_head classes; + + struct nouveau_bo *vga_ram; + + /* interrupt handling */ + void (*irq_handler[32])(struct drm_device *); + bool msi_enabled; + + struct list_head vbl_waiting; + + struct { + struct drm_global_reference mem_global_ref; + struct ttm_bo_global_ref bo_global_ref; + struct ttm_bo_device bdev; + atomic_t validate_sequence; + } ttm; + + struct { + spinlock_t lock; + struct drm_mm heap; + struct nouveau_bo *bo; + } fence; + + struct { + spinlock_t lock; + struct nouveau_channel *ptr[NOUVEAU_MAX_CHANNEL_NR]; + } channels; + + struct nouveau_engine engine; + struct nouveau_channel *channel; + + /* For PFIFO and PGRAPH. */ + spinlock_t context_switch_lock; + + /* VM/PRAMIN flush, legacy PRAMIN aperture */ + spinlock_t vm_lock; + + /* RAMIN configuration, RAMFC, RAMHT and RAMRO offsets */ + struct nouveau_ramht *ramht; + struct nouveau_gpuobj *ramfc; + struct nouveau_gpuobj *ramro; + + uint32_t ramin_rsvd_vram; + + struct { + enum { + NOUVEAU_GART_NONE = 0, + NOUVEAU_GART_AGP, /* AGP */ + NOUVEAU_GART_PDMA, /* paged dma object */ + NOUVEAU_GART_HW /* on-chip gart/vm */ + } type; + uint64_t aper_base; + uint64_t aper_size; + uint64_t aper_free; + + struct ttm_backend_func *func; + + struct { + struct page *page; + dma_addr_t addr; + } dummy; + + struct nouveau_gpuobj *sg_ctxdma; + } gart_info; + + /* nv10-nv40 tiling regions */ + struct { + struct nouveau_tile_reg reg[NOUVEAU_MAX_TILE_NR]; + spinlock_t lock; + } tile; + + /* VRAM/fb configuration */ + uint64_t vram_size; + uint64_t vram_sys_base; + u32 vram_rblock_size; + + uint64_t fb_phys; + uint64_t fb_available_size; + uint64_t fb_mappable_pages; + uint64_t fb_aper_free; + int fb_mtrr; + + /* BAR control (NV50-) */ + struct nouveau_vm *bar1_vm; + struct nouveau_vm *bar3_vm; + + /* G8x/G9x virtual address space */ + struct nouveau_vm *chan_vm; + + struct nvbios vbios; + + struct nv04_mode_state mode_reg; + struct nv04_mode_state saved_reg; + uint32_t saved_vga_font[4][16384]; + uint32_t crtc_owner; + uint32_t dac_users[4]; + + struct backlight_device *backlight; + + struct { + struct dentry *channel_root; + } debugfs; + + struct nouveau_fbdev *nfbdev; + struct apertures_struct *apertures; +}; + +static inline struct drm_nouveau_private * +nouveau_private(struct drm_device *dev) +{ + return dev->dev_private; +} + +static inline struct drm_nouveau_private * +nouveau_bdev(struct ttm_bo_device *bd) +{ + return container_of(bd, struct drm_nouveau_private, ttm.bdev); +} + +static inline int +nouveau_bo_ref(struct nouveau_bo *ref, struct nouveau_bo **pnvbo) +{ + struct nouveau_bo *prev; + + if (!pnvbo) + return -EINVAL; + prev = *pnvbo; + + *pnvbo = ref ? nouveau_bo(ttm_bo_reference(&ref->bo)) : NULL; + if (prev) { + struct ttm_buffer_object *bo = &prev->bo; + + ttm_bo_unref(&bo); + } + + return 0; +} + +/* nouveau_drv.c */ +extern int nouveau_agpmode; +extern int nouveau_duallink; +extern int nouveau_uscript_lvds; +extern int nouveau_uscript_tmds; +extern int nouveau_vram_pushbuf; +extern int nouveau_vram_notify; +extern int nouveau_fbpercrtc; +extern int nouveau_tv_disable; +extern char *nouveau_tv_norm; +extern int nouveau_reg_debug; +extern char *nouveau_vbios; +extern int nouveau_ignorelid; +extern int nouveau_nofbaccel; +extern int nouveau_noaccel; +extern int nouveau_force_post; +extern int nouveau_override_conntype; +extern char *nouveau_perflvl; +extern int nouveau_perflvl_wr; +extern int nouveau_msi; + +extern int nouveau_pci_suspend(struct pci_dev *pdev, pm_message_t pm_state); +extern int nouveau_pci_resume(struct pci_dev *pdev); + +/* nouveau_state.c */ +extern void nouveau_preclose(struct drm_device *dev, struct drm_file *); +extern int nouveau_load(struct drm_device *, unsigned long flags); +extern int nouveau_firstopen(struct drm_device *); +extern void nouveau_lastclose(struct drm_device *); +extern int nouveau_unload(struct drm_device *); +extern int nouveau_ioctl_getparam(struct drm_device *, void *data, + struct drm_file *); +extern int nouveau_ioctl_setparam(struct drm_device *, void *data, + struct drm_file *); +extern bool nouveau_wait_eq(struct drm_device *, uint64_t timeout, + uint32_t reg, uint32_t mask, uint32_t val); +extern bool nouveau_wait_ne(struct drm_device *, uint64_t timeout, + uint32_t reg, uint32_t mask, uint32_t val); +extern bool nouveau_wait_for_idle(struct drm_device *); +extern int nouveau_card_init(struct drm_device *); + +/* nouveau_mem.c */ +extern int nouveau_mem_vram_init(struct drm_device *); +extern void nouveau_mem_vram_fini(struct drm_device *); +extern int nouveau_mem_gart_init(struct drm_device *); +extern void nouveau_mem_gart_fini(struct drm_device *); +extern int nouveau_mem_init_agp(struct drm_device *); +extern int nouveau_mem_reset_agp(struct drm_device *); +extern void nouveau_mem_close(struct drm_device *); +extern int nouveau_mem_detect(struct drm_device *); +extern bool nouveau_mem_flags_valid(struct drm_device *, u32 tile_flags); +extern struct nouveau_tile_reg *nv10_mem_set_tiling( + struct drm_device *dev, uint32_t addr, uint32_t size, + uint32_t pitch, uint32_t flags); +extern void nv10_mem_put_tile_region(struct drm_device *dev, + struct nouveau_tile_reg *tile, + struct nouveau_fence *fence); +extern const struct ttm_mem_type_manager_func nouveau_vram_manager; +extern const struct ttm_mem_type_manager_func nouveau_gart_manager; + +/* nouveau_notifier.c */ +extern int nouveau_notifier_init_channel(struct nouveau_channel *); +extern void nouveau_notifier_takedown_channel(struct nouveau_channel *); +extern int nouveau_notifier_alloc(struct nouveau_channel *, uint32_t handle, + int cout, uint32_t start, uint32_t end, + uint32_t *offset); +extern int nouveau_notifier_offset(struct nouveau_gpuobj *, uint32_t *); +extern int nouveau_ioctl_notifier_alloc(struct drm_device *, void *data, + struct drm_file *); +extern int nouveau_ioctl_notifier_free(struct drm_device *, void *data, + struct drm_file *); + +/* nouveau_channel.c */ +extern struct drm_ioctl_desc nouveau_ioctls[]; +extern int nouveau_max_ioctl; +extern void nouveau_channel_cleanup(struct drm_device *, struct drm_file *); +extern int nouveau_channel_alloc(struct drm_device *dev, + struct nouveau_channel **chan, + struct drm_file *file_priv, + uint32_t fb_ctxdma, uint32_t tt_ctxdma); +extern struct nouveau_channel * +nouveau_channel_get_unlocked(struct nouveau_channel *); +extern struct nouveau_channel * +nouveau_channel_get(struct drm_device *, struct drm_file *, int id); +extern void nouveau_channel_put_unlocked(struct nouveau_channel **); +extern void nouveau_channel_put(struct nouveau_channel **); +extern void nouveau_channel_ref(struct nouveau_channel *chan, + struct nouveau_channel **pchan); +extern void nouveau_channel_idle(struct nouveau_channel *chan); + +/* nouveau_object.c */ +#define NVOBJ_ENGINE_ADD(d, e, p) do { \ + struct drm_nouveau_private *dev_priv = (d)->dev_private; \ + dev_priv->eng[NVOBJ_ENGINE_##e] = (p); \ +} while (0) + +#define NVOBJ_ENGINE_DEL(d, e) do { \ + struct drm_nouveau_private *dev_priv = (d)->dev_private; \ + dev_priv->eng[NVOBJ_ENGINE_##e] = NULL; \ +} while (0) + +#define NVOBJ_CLASS(d, c, e) do { \ + int ret = nouveau_gpuobj_class_new((d), (c), NVOBJ_ENGINE_##e); \ + if (ret) \ + return ret; \ +} while (0) + +#define NVOBJ_MTHD(d, c, m, e) do { \ + int ret = nouveau_gpuobj_mthd_new((d), (c), (m), (e)); \ + if (ret) \ + return ret; \ +} while (0) + +extern int nouveau_gpuobj_early_init(struct drm_device *); +extern int nouveau_gpuobj_init(struct drm_device *); +extern void nouveau_gpuobj_takedown(struct drm_device *); +extern int nouveau_gpuobj_suspend(struct drm_device *dev); +extern void nouveau_gpuobj_resume(struct drm_device *dev); +extern int nouveau_gpuobj_class_new(struct drm_device *, u32 class, u32 eng); +extern int nouveau_gpuobj_mthd_new(struct drm_device *, u32 class, u32 mthd, + int (*exec)(struct nouveau_channel *, + u32 class, u32 mthd, u32 data)); +extern int nouveau_gpuobj_mthd_call(struct nouveau_channel *, u32, u32, u32); +extern int nouveau_gpuobj_mthd_call2(struct drm_device *, int, u32, u32, u32); +extern int nouveau_gpuobj_channel_init(struct nouveau_channel *, + uint32_t vram_h, uint32_t tt_h); +extern void nouveau_gpuobj_channel_takedown(struct nouveau_channel *); +extern int nouveau_gpuobj_new(struct drm_device *, struct nouveau_channel *, + uint32_t size, int align, uint32_t flags, + struct nouveau_gpuobj **); +extern void nouveau_gpuobj_ref(struct nouveau_gpuobj *, + struct nouveau_gpuobj **); +extern int nouveau_gpuobj_new_fake(struct drm_device *, u32 pinst, u64 vinst, + u32 size, u32 flags, + struct nouveau_gpuobj **); +extern int nouveau_gpuobj_dma_new(struct nouveau_channel *, int class, + uint64_t offset, uint64_t size, int access, + int target, struct nouveau_gpuobj **); +extern int nouveau_gpuobj_gr_new(struct nouveau_channel *, u32 handle, int class); +extern int nv50_gpuobj_dma_new(struct nouveau_channel *, int class, u64 base, + u64 size, int target, int access, u32 type, + u32 comp, struct nouveau_gpuobj **pobj); +extern void nv50_gpuobj_dma_init(struct nouveau_gpuobj *, u32 offset, + int class, u64 base, u64 size, int target, + int access, u32 type, u32 comp); +extern int nouveau_ioctl_grobj_alloc(struct drm_device *, void *data, + struct drm_file *); +extern int nouveau_ioctl_gpuobj_free(struct drm_device *, void *data, + struct drm_file *); + +/* nouveau_irq.c */ +extern int nouveau_irq_init(struct drm_device *); +extern void nouveau_irq_fini(struct drm_device *); +extern irqreturn_t nouveau_irq_handler(DRM_IRQ_ARGS); +extern void nouveau_irq_register(struct drm_device *, int status_bit, + void (*)(struct drm_device *)); +extern void nouveau_irq_unregister(struct drm_device *, int status_bit); +extern void nouveau_irq_preinstall(struct drm_device *); +extern int nouveau_irq_postinstall(struct drm_device *); +extern void nouveau_irq_uninstall(struct drm_device *); + +/* nouveau_sgdma.c */ +extern int nouveau_sgdma_init(struct drm_device *); +extern void nouveau_sgdma_takedown(struct drm_device *); +extern uint32_t nouveau_sgdma_get_physical(struct drm_device *, + uint32_t offset); +extern struct ttm_backend *nouveau_sgdma_init_ttm(struct drm_device *); + +/* nouveau_debugfs.c */ +#if defined(CONFIG_DRM_NOUVEAU_DEBUG) +extern int nouveau_debugfs_init(struct drm_minor *); +extern void nouveau_debugfs_takedown(struct drm_minor *); +extern int nouveau_debugfs_channel_init(struct nouveau_channel *); +extern void nouveau_debugfs_channel_fini(struct nouveau_channel *); +#else +static inline int +nouveau_debugfs_init(struct drm_minor *minor) +{ + return 0; +} + +static inline void nouveau_debugfs_takedown(struct drm_minor *minor) +{ +} + +static inline int +nouveau_debugfs_channel_init(struct nouveau_channel *chan) +{ + return 0; +} + +static inline void +nouveau_debugfs_channel_fini(struct nouveau_channel *chan) +{ +} +#endif + +/* nouveau_dma.c */ +extern void nouveau_dma_pre_init(struct nouveau_channel *); +extern int nouveau_dma_init(struct nouveau_channel *); +extern int nouveau_dma_wait(struct nouveau_channel *, int slots, int size); + +/* nouveau_acpi.c */ +#define ROM_BIOS_PAGE 4096 +#if defined(CONFIG_ACPI) +void nouveau_register_dsm_handler(void); +void nouveau_unregister_dsm_handler(void); +int nouveau_acpi_get_bios_chunk(uint8_t *bios, int offset, int len); +bool nouveau_acpi_rom_supported(struct pci_dev *pdev); +int nouveau_acpi_edid(struct drm_device *, struct drm_connector *); +#else +static inline void nouveau_register_dsm_handler(void) {} +static inline void nouveau_unregister_dsm_handler(void) {} +static inline bool nouveau_acpi_rom_supported(struct pci_dev *pdev) { return false; } +static inline int nouveau_acpi_get_bios_chunk(uint8_t *bios, int offset, int len) { return -EINVAL; } +static inline int nouveau_acpi_edid(struct drm_device *dev, struct drm_connector *connector) { return -EINVAL; } +#endif + +/* nouveau_backlight.c */ +#ifdef CONFIG_DRM_NOUVEAU_BACKLIGHT +extern int nouveau_backlight_init(struct drm_connector *); +extern void nouveau_backlight_exit(struct drm_connector *); +#else +static inline int nouveau_backlight_init(struct drm_connector *dev) +{ + return 0; +} + +static inline void nouveau_backlight_exit(struct drm_connector *dev) { } +#endif + +/* nouveau_bios.c */ +extern int nouveau_bios_init(struct drm_device *); +extern void nouveau_bios_takedown(struct drm_device *dev); +extern int nouveau_run_vbios_init(struct drm_device *); +extern void nouveau_bios_run_init_table(struct drm_device *, uint16_t table, + struct dcb_entry *); +extern struct dcb_gpio_entry *nouveau_bios_gpio_entry(struct drm_device *, + enum dcb_gpio_tag); +extern struct dcb_connector_table_entry * +nouveau_bios_connector_entry(struct drm_device *, int index); +extern u32 get_pll_register(struct drm_device *, enum pll_types); +extern int get_pll_limits(struct drm_device *, uint32_t limit_match, + struct pll_lims *); +extern int nouveau_bios_run_display_table(struct drm_device *, + struct dcb_entry *, + uint32_t script, int pxclk); +extern void *nouveau_bios_dp_table(struct drm_device *, struct dcb_entry *, + int *length); +extern bool nouveau_bios_fp_mode(struct drm_device *, struct drm_display_mode *); +extern uint8_t *nouveau_bios_embedded_edid(struct drm_device *); +extern int nouveau_bios_parse_lvds_table(struct drm_device *, int pxclk, + bool *dl, bool *if_is_24bit); +extern int run_tmds_table(struct drm_device *, struct dcb_entry *, + int head, int pxclk); +extern int call_lvds_script(struct drm_device *, struct dcb_entry *, int head, + enum LVDS_script, int pxclk); + +/* nouveau_ttm.c */ +int nouveau_ttm_global_init(struct drm_nouveau_private *); +void nouveau_ttm_global_release(struct drm_nouveau_private *); +int nouveau_ttm_mmap(struct file *, struct vm_area_struct *); + +/* nouveau_dp.c */ +int nouveau_dp_auxch(struct nouveau_i2c_chan *auxch, int cmd, int addr, + uint8_t *data, int data_nr); +bool nouveau_dp_detect(struct drm_encoder *); +bool nouveau_dp_link_train(struct drm_encoder *); + +/* nv04_fb.c */ +extern int nv04_fb_init(struct drm_device *); +extern void nv04_fb_takedown(struct drm_device *); + +/* nv10_fb.c */ +extern int nv10_fb_init(struct drm_device *); +extern void nv10_fb_takedown(struct drm_device *); +extern void nv10_fb_init_tile_region(struct drm_device *dev, int i, + uint32_t addr, uint32_t size, + uint32_t pitch, uint32_t flags); +extern void nv10_fb_set_tile_region(struct drm_device *dev, int i); +extern void nv10_fb_free_tile_region(struct drm_device *dev, int i); + +/* nv30_fb.c */ +extern int nv30_fb_init(struct drm_device *); +extern void nv30_fb_takedown(struct drm_device *); +extern void nv30_fb_init_tile_region(struct drm_device *dev, int i, + uint32_t addr, uint32_t size, + uint32_t pitch, uint32_t flags); +extern void nv30_fb_free_tile_region(struct drm_device *dev, int i); + +/* nv40_fb.c */ +extern int nv40_fb_init(struct drm_device *); +extern void nv40_fb_takedown(struct drm_device *); +extern void nv40_fb_set_tile_region(struct drm_device *dev, int i); + +/* nv50_fb.c */ +extern int nv50_fb_init(struct drm_device *); +extern void nv50_fb_takedown(struct drm_device *); +extern void nv50_fb_vm_trap(struct drm_device *, int display); + +/* nvc0_fb.c */ +extern int nvc0_fb_init(struct drm_device *); +extern void nvc0_fb_takedown(struct drm_device *); + +/* nv04_fifo.c */ +extern int nv04_fifo_init(struct drm_device *); +extern void nv04_fifo_fini(struct drm_device *); +extern void nv04_fifo_disable(struct drm_device *); +extern void nv04_fifo_enable(struct drm_device *); +extern bool nv04_fifo_reassign(struct drm_device *, bool); +extern bool nv04_fifo_cache_pull(struct drm_device *, bool); +extern int nv04_fifo_channel_id(struct drm_device *); +extern int nv04_fifo_create_context(struct nouveau_channel *); +extern void nv04_fifo_destroy_context(struct nouveau_channel *); +extern int nv04_fifo_load_context(struct nouveau_channel *); +extern int nv04_fifo_unload_context(struct drm_device *); +extern void nv04_fifo_isr(struct drm_device *); + +/* nv10_fifo.c */ +extern int nv10_fifo_init(struct drm_device *); +extern int nv10_fifo_channel_id(struct drm_device *); +extern int nv10_fifo_create_context(struct nouveau_channel *); +extern int nv10_fifo_load_context(struct nouveau_channel *); +extern int nv10_fifo_unload_context(struct drm_device *); + +/* nv40_fifo.c */ +extern int nv40_fifo_init(struct drm_device *); +extern int nv40_fifo_create_context(struct nouveau_channel *); +extern int nv40_fifo_load_context(struct nouveau_channel *); +extern int nv40_fifo_unload_context(struct drm_device *); + +/* nv50_fifo.c */ +extern int nv50_fifo_init(struct drm_device *); +extern void nv50_fifo_takedown(struct drm_device *); +extern int nv50_fifo_channel_id(struct drm_device *); +extern int nv50_fifo_create_context(struct nouveau_channel *); +extern void nv50_fifo_destroy_context(struct nouveau_channel *); +extern int nv50_fifo_load_context(struct nouveau_channel *); +extern int nv50_fifo_unload_context(struct drm_device *); +extern void nv50_fifo_tlb_flush(struct drm_device *dev); + +/* nvc0_fifo.c */ +extern int nvc0_fifo_init(struct drm_device *); +extern void nvc0_fifo_takedown(struct drm_device *); +extern void nvc0_fifo_disable(struct drm_device *); +extern void nvc0_fifo_enable(struct drm_device *); +extern bool nvc0_fifo_reassign(struct drm_device *, bool); +extern bool nvc0_fifo_cache_pull(struct drm_device *, bool); +extern int nvc0_fifo_channel_id(struct drm_device *); +extern int nvc0_fifo_create_context(struct nouveau_channel *); +extern void nvc0_fifo_destroy_context(struct nouveau_channel *); +extern int nvc0_fifo_load_context(struct nouveau_channel *); +extern int nvc0_fifo_unload_context(struct drm_device *); + +/* nv04_graph.c */ +extern int nv04_graph_create(struct drm_device *); +extern void nv04_graph_fifo_access(struct drm_device *, bool); +extern int nv04_graph_object_new(struct nouveau_channel *, int, u32, u16); +extern int nv04_graph_mthd_page_flip(struct nouveau_channel *chan, + u32 class, u32 mthd, u32 data); +extern struct nouveau_bitfield nv04_graph_nsource[]; + +/* nv10_graph.c */ +extern int nv10_graph_create(struct drm_device *); +extern struct nouveau_channel *nv10_graph_channel(struct drm_device *); +extern struct nouveau_bitfield nv10_graph_intr[]; +extern struct nouveau_bitfield nv10_graph_nstatus[]; + +/* nv20_graph.c */ +extern int nv20_graph_create(struct drm_device *); + +/* nv40_graph.c */ +extern int nv40_graph_create(struct drm_device *); +extern void nv40_grctx_init(struct nouveau_grctx *); + +/* nv50_graph.c */ +extern int nv50_graph_create(struct drm_device *); +extern int nv50_grctx_init(struct nouveau_grctx *); +extern struct nouveau_enum nv50_data_error_names[]; +extern int nv50_graph_isr_chid(struct drm_device *dev, u64 inst); + +/* nvc0_graph.c */ +extern int nvc0_graph_create(struct drm_device *); +extern int nvc0_graph_isr_chid(struct drm_device *dev, u64 inst); + +/* nv84_crypt.c */ +extern int nv84_crypt_create(struct drm_device *); + +/* nva3_copy.c */ +extern int nva3_copy_create(struct drm_device *dev); + +/* nvc0_copy.c */ +extern int nvc0_copy_create(struct drm_device *dev, int engine); + +/* nv40_mpeg.c */ +extern int nv40_mpeg_create(struct drm_device *dev); + +/* nv50_mpeg.c */ +extern int nv50_mpeg_create(struct drm_device *dev); + +/* nv04_instmem.c */ +extern int nv04_instmem_init(struct drm_device *); +extern void nv04_instmem_takedown(struct drm_device *); +extern int nv04_instmem_suspend(struct drm_device *); +extern void nv04_instmem_resume(struct drm_device *); +extern int nv04_instmem_get(struct nouveau_gpuobj *, u32 size, u32 align); +extern void nv04_instmem_put(struct nouveau_gpuobj *); +extern int nv04_instmem_map(struct nouveau_gpuobj *); +extern void nv04_instmem_unmap(struct nouveau_gpuobj *); +extern void nv04_instmem_flush(struct drm_device *); + +/* nv50_instmem.c */ +extern int nv50_instmem_init(struct drm_device *); +extern void nv50_instmem_takedown(struct drm_device *); +extern int nv50_instmem_suspend(struct drm_device *); +extern void nv50_instmem_resume(struct drm_device *); +extern int nv50_instmem_get(struct nouveau_gpuobj *, u32 size, u32 align); +extern void nv50_instmem_put(struct nouveau_gpuobj *); +extern int nv50_instmem_map(struct nouveau_gpuobj *); +extern void nv50_instmem_unmap(struct nouveau_gpuobj *); +extern void nv50_instmem_flush(struct drm_device *); +extern void nv84_instmem_flush(struct drm_device *); + +/* nvc0_instmem.c */ +extern int nvc0_instmem_init(struct drm_device *); +extern void nvc0_instmem_takedown(struct drm_device *); +extern int nvc0_instmem_suspend(struct drm_device *); +extern void nvc0_instmem_resume(struct drm_device *); + +/* nv04_mc.c */ +extern int nv04_mc_init(struct drm_device *); +extern void nv04_mc_takedown(struct drm_device *); + +/* nv40_mc.c */ +extern int nv40_mc_init(struct drm_device *); +extern void nv40_mc_takedown(struct drm_device *); + +/* nv50_mc.c */ +extern int nv50_mc_init(struct drm_device *); +extern void nv50_mc_takedown(struct drm_device *); + +/* nv04_timer.c */ +extern int nv04_timer_init(struct drm_device *); +extern uint64_t nv04_timer_read(struct drm_device *); +extern void nv04_timer_takedown(struct drm_device *); + +extern long nouveau_compat_ioctl(struct file *file, unsigned int cmd, + unsigned long arg); + +/* nv04_dac.c */ +extern int nv04_dac_create(struct drm_connector *, struct dcb_entry *); +extern uint32_t nv17_dac_sample_load(struct drm_encoder *encoder); +extern int nv04_dac_output_offset(struct drm_encoder *encoder); +extern void nv04_dac_update_dacclk(struct drm_encoder *encoder, bool enable); +extern bool nv04_dac_in_use(struct drm_encoder *encoder); + +/* nv04_dfp.c */ +extern int nv04_dfp_create(struct drm_connector *, struct dcb_entry *); +extern int nv04_dfp_get_bound_head(struct drm_device *dev, struct dcb_entry *dcbent); +extern void nv04_dfp_bind_head(struct drm_device *dev, struct dcb_entry *dcbent, + int head, bool dl); +extern void nv04_dfp_disable(struct drm_device *dev, int head); +extern void nv04_dfp_update_fp_control(struct drm_encoder *encoder, int mode); + +/* nv04_tv.c */ +extern int nv04_tv_identify(struct drm_device *dev, int i2c_index); +extern int nv04_tv_create(struct drm_connector *, struct dcb_entry *); + +/* nv17_tv.c */ +extern int nv17_tv_create(struct drm_connector *, struct dcb_entry *); + +/* nv04_display.c */ +extern int nv04_display_early_init(struct drm_device *); +extern void nv04_display_late_takedown(struct drm_device *); +extern int nv04_display_create(struct drm_device *); +extern int nv04_display_init(struct drm_device *); +extern void nv04_display_destroy(struct drm_device *); + +/* nv04_crtc.c */ +extern int nv04_crtc_create(struct drm_device *, int index); + +/* nouveau_bo.c */ +extern struct ttm_bo_driver nouveau_bo_driver; +extern int nouveau_bo_new(struct drm_device *, struct nouveau_channel *, + int size, int align, uint32_t flags, + uint32_t tile_mode, uint32_t tile_flags, + struct nouveau_bo **); +extern int nouveau_bo_pin(struct nouveau_bo *, uint32_t flags); +extern int nouveau_bo_unpin(struct nouveau_bo *); +extern int nouveau_bo_map(struct nouveau_bo *); +extern void nouveau_bo_unmap(struct nouveau_bo *); +extern void nouveau_bo_placement_set(struct nouveau_bo *, uint32_t type, + uint32_t busy); +extern u16 nouveau_bo_rd16(struct nouveau_bo *nvbo, unsigned index); +extern void nouveau_bo_wr16(struct nouveau_bo *nvbo, unsigned index, u16 val); +extern u32 nouveau_bo_rd32(struct nouveau_bo *nvbo, unsigned index); +extern void nouveau_bo_wr32(struct nouveau_bo *nvbo, unsigned index, u32 val); +extern void nouveau_bo_fence(struct nouveau_bo *, struct nouveau_fence *); +extern int nouveau_bo_validate(struct nouveau_bo *, bool interruptible, + bool no_wait_reserve, bool no_wait_gpu); + +/* nouveau_fence.c */ +struct nouveau_fence; +extern int nouveau_fence_init(struct drm_device *); +extern void nouveau_fence_fini(struct drm_device *); +extern int nouveau_fence_channel_init(struct nouveau_channel *); +extern void nouveau_fence_channel_fini(struct nouveau_channel *); +extern void nouveau_fence_update(struct nouveau_channel *); +extern int nouveau_fence_new(struct nouveau_channel *, struct nouveau_fence **, + bool emit); +extern int nouveau_fence_emit(struct nouveau_fence *); +extern void nouveau_fence_work(struct nouveau_fence *fence, + void (*work)(void *priv, bool signalled), + void *priv); +struct nouveau_channel *nouveau_fence_channel(struct nouveau_fence *); + +extern bool __nouveau_fence_signalled(void *obj, void *arg); +extern int __nouveau_fence_wait(void *obj, void *arg, bool lazy, bool intr); +extern int __nouveau_fence_flush(void *obj, void *arg); +extern void __nouveau_fence_unref(void **obj); +extern void *__nouveau_fence_ref(void *obj); + +static inline bool nouveau_fence_signalled(struct nouveau_fence *obj) +{ + return __nouveau_fence_signalled(obj, NULL); +} +static inline int +nouveau_fence_wait(struct nouveau_fence *obj, bool lazy, bool intr) +{ + return __nouveau_fence_wait(obj, NULL, lazy, intr); +} +extern int nouveau_fence_sync(struct nouveau_fence *, struct nouveau_channel *); +static inline int nouveau_fence_flush(struct nouveau_fence *obj) +{ + return __nouveau_fence_flush(obj, NULL); +} +static inline void nouveau_fence_unref(struct nouveau_fence **obj) +{ + __nouveau_fence_unref((void **)obj); +} +static inline struct nouveau_fence *nouveau_fence_ref(struct nouveau_fence *obj) +{ + return __nouveau_fence_ref(obj); +} + +/* nouveau_gem.c */ +extern int nouveau_gem_new(struct drm_device *, struct nouveau_channel *, + int size, int align, uint32_t domain, + uint32_t tile_mode, uint32_t tile_flags, + struct nouveau_bo **); +extern int nouveau_gem_object_new(struct drm_gem_object *); +extern void nouveau_gem_object_del(struct drm_gem_object *); +extern int nouveau_gem_ioctl_new(struct drm_device *, void *, + struct drm_file *); +extern int nouveau_gem_ioctl_pushbuf(struct drm_device *, void *, + struct drm_file *); +extern int nouveau_gem_ioctl_cpu_prep(struct drm_device *, void *, + struct drm_file *); +extern int nouveau_gem_ioctl_cpu_fini(struct drm_device *, void *, + struct drm_file *); +extern int nouveau_gem_ioctl_info(struct drm_device *, void *, + struct drm_file *); + +/* nouveau_display.c */ +int nouveau_vblank_enable(struct drm_device *dev, int crtc); +void nouveau_vblank_disable(struct drm_device *dev, int crtc); +int nouveau_crtc_page_flip(struct drm_crtc *crtc, struct drm_framebuffer *fb, + struct drm_pending_vblank_event *event); +int nouveau_finish_page_flip(struct nouveau_channel *, + struct nouveau_page_flip_state *); + +/* nv10_gpio.c */ +int nv10_gpio_get(struct drm_device *dev, enum dcb_gpio_tag tag); +int nv10_gpio_set(struct drm_device *dev, enum dcb_gpio_tag tag, int state); + +/* nv50_gpio.c */ +int nv50_gpio_init(struct drm_device *dev); +void nv50_gpio_fini(struct drm_device *dev); +int nv50_gpio_get(struct drm_device *dev, enum dcb_gpio_tag tag); +int nv50_gpio_set(struct drm_device *dev, enum dcb_gpio_tag tag, int state); +int nv50_gpio_irq_register(struct drm_device *, enum dcb_gpio_tag, + void (*)(void *, int), void *); +void nv50_gpio_irq_unregister(struct drm_device *, enum dcb_gpio_tag, + void (*)(void *, int), void *); +bool nv50_gpio_irq_enable(struct drm_device *, enum dcb_gpio_tag, bool on); + +/* nv50_calc. */ +int nv50_calc_pll(struct drm_device *, struct pll_lims *, int clk, + int *N1, int *M1, int *N2, int *M2, int *P); +int nva3_calc_pll(struct drm_device *, struct pll_lims *, + int clk, int *N, int *fN, int *M, int *P); + +#ifndef ioread32_native +#ifdef __BIG_ENDIAN +#define ioread16_native ioread16be +#define iowrite16_native iowrite16be +#define ioread32_native ioread32be +#define iowrite32_native iowrite32be +#else /* def __BIG_ENDIAN */ +#define ioread16_native ioread16 +#define iowrite16_native iowrite16 +#define ioread32_native ioread32 +#define iowrite32_native iowrite32 +#endif /* def __BIG_ENDIAN else */ +#endif /* !ioread32_native */ + +/* channel control reg access */ +static inline u32 nvchan_rd32(struct nouveau_channel *chan, unsigned reg) +{ + return ioread32_native(chan->user + reg); +} + +static inline void nvchan_wr32(struct nouveau_channel *chan, + unsigned reg, u32 val) +{ + iowrite32_native(val, chan->user + reg); +} + +/* register access */ +static inline u32 nv_rd32(struct drm_device *dev, unsigned reg) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + return ioread32_native(dev_priv->mmio + reg); +} + +static inline void nv_wr32(struct drm_device *dev, unsigned reg, u32 val) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + iowrite32_native(val, dev_priv->mmio + reg); +} + +static inline u32 nv_mask(struct drm_device *dev, u32 reg, u32 mask, u32 val) +{ + u32 tmp = nv_rd32(dev, reg); + nv_wr32(dev, reg, (tmp & ~mask) | val); + return tmp; +} + +static inline u8 nv_rd08(struct drm_device *dev, unsigned reg) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + return ioread8(dev_priv->mmio + reg); +} + +static inline void nv_wr08(struct drm_device *dev, unsigned reg, u8 val) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + iowrite8(val, dev_priv->mmio + reg); +} + +#define nv_wait(dev, reg, mask, val) \ + nouveau_wait_eq(dev, 2000000000ULL, (reg), (mask), (val)) +#define nv_wait_ne(dev, reg, mask, val) \ + nouveau_wait_ne(dev, 2000000000ULL, (reg), (mask), (val)) + +/* PRAMIN access */ +static inline u32 nv_ri32(struct drm_device *dev, unsigned offset) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + return ioread32_native(dev_priv->ramin + offset); +} + +static inline void nv_wi32(struct drm_device *dev, unsigned offset, u32 val) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + iowrite32_native(val, dev_priv->ramin + offset); +} + +/* object access */ +extern u32 nv_ro32(struct nouveau_gpuobj *, u32 offset); +extern void nv_wo32(struct nouveau_gpuobj *, u32 offset, u32 val); + +/* + * Logging + * Argument d is (struct drm_device *). + */ +#define NV_PRINTK(level, d, fmt, arg...) \ + printk(level "[" DRM_NAME "] " DRIVER_NAME " %s: " fmt, \ + pci_name(d->pdev), ##arg) +#ifndef NV_DEBUG_NOTRACE +#define NV_DEBUG(d, fmt, arg...) do { \ + if (drm_debug & DRM_UT_DRIVER) { \ + NV_PRINTK(KERN_DEBUG, d, "%s:%d - " fmt, __func__, \ + __LINE__, ##arg); \ + } \ +} while (0) +#define NV_DEBUG_KMS(d, fmt, arg...) do { \ + if (drm_debug & DRM_UT_KMS) { \ + NV_PRINTK(KERN_DEBUG, d, "%s:%d - " fmt, __func__, \ + __LINE__, ##arg); \ + } \ +} while (0) +#else +#define NV_DEBUG(d, fmt, arg...) do { \ + if (drm_debug & DRM_UT_DRIVER) \ + NV_PRINTK(KERN_DEBUG, d, fmt, ##arg); \ +} while (0) +#define NV_DEBUG_KMS(d, fmt, arg...) do { \ + if (drm_debug & DRM_UT_KMS) \ + NV_PRINTK(KERN_DEBUG, d, fmt, ##arg); \ +} while (0) +#endif +#define NV_ERROR(d, fmt, arg...) NV_PRINTK(KERN_ERR, d, fmt, ##arg) +#define NV_INFO(d, fmt, arg...) NV_PRINTK(KERN_INFO, d, fmt, ##arg) +#define NV_TRACEWARN(d, fmt, arg...) NV_PRINTK(KERN_NOTICE, d, fmt, ##arg) +#define NV_TRACE(d, fmt, arg...) NV_PRINTK(KERN_INFO, d, fmt, ##arg) +#define NV_WARN(d, fmt, arg...) NV_PRINTK(KERN_WARNING, d, fmt, ##arg) + +/* nouveau_reg_debug bitmask */ +enum { + NOUVEAU_REG_DEBUG_MC = 0x1, + NOUVEAU_REG_DEBUG_VIDEO = 0x2, + NOUVEAU_REG_DEBUG_FB = 0x4, + NOUVEAU_REG_DEBUG_EXTDEV = 0x8, + NOUVEAU_REG_DEBUG_CRTC = 0x10, + NOUVEAU_REG_DEBUG_RAMDAC = 0x20, + NOUVEAU_REG_DEBUG_VGACRTC = 0x40, + NOUVEAU_REG_DEBUG_RMVIO = 0x80, + NOUVEAU_REG_DEBUG_VGAATTR = 0x100, + NOUVEAU_REG_DEBUG_EVO = 0x200, +}; + +#define NV_REG_DEBUG(type, dev, fmt, arg...) do { \ + if (nouveau_reg_debug & NOUVEAU_REG_DEBUG_##type) \ + NV_PRINTK(KERN_DEBUG, dev, "%s: " fmt, __func__, ##arg); \ +} while (0) + +static inline bool +nv_two_heads(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + const int impl = dev->pci_device & 0x0ff0; + + if (dev_priv->card_type >= NV_10 && impl != 0x0100 && + impl != 0x0150 && impl != 0x01a0 && impl != 0x0200) + return true; + + return false; +} + +static inline bool +nv_gf4_disp_arch(struct drm_device *dev) +{ + return nv_two_heads(dev) && (dev->pci_device & 0x0ff0) != 0x0110; +} + +static inline bool +nv_two_reg_pll(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + const int impl = dev->pci_device & 0x0ff0; + + if (impl == 0x0310 || impl == 0x0340 || dev_priv->card_type >= NV_40) + return true; + return false; +} + +static inline bool +nv_match_device(struct drm_device *dev, unsigned device, + unsigned sub_vendor, unsigned sub_device) +{ + return dev->pdev->device == device && + dev->pdev->subsystem_vendor == sub_vendor && + dev->pdev->subsystem_device == sub_device; +} + +static inline void * +nv_engine(struct drm_device *dev, int engine) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + return (void *)dev_priv->eng[engine]; +} + +/* returns 1 if device is one of the nv4x using the 0x4497 object class, + * helpful to determine a number of other hardware features + */ +static inline int +nv44_graph_class(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + if ((dev_priv->chipset & 0xf0) == 0x60) + return 1; + + return !(0x0baf & (1 << (dev_priv->chipset & 0x0f))); +} + +/* memory type/access flags, do not match hardware values */ +#define NV_MEM_ACCESS_RO 1 +#define NV_MEM_ACCESS_WO 2 +#define NV_MEM_ACCESS_RW (NV_MEM_ACCESS_RO | NV_MEM_ACCESS_WO) +#define NV_MEM_ACCESS_SYS 4 +#define NV_MEM_ACCESS_VM 8 + +#define NV_MEM_TARGET_VRAM 0 +#define NV_MEM_TARGET_PCI 1 +#define NV_MEM_TARGET_PCI_NOSNOOP 2 +#define NV_MEM_TARGET_VM 3 +#define NV_MEM_TARGET_GART 4 + +#define NV_MEM_TYPE_VM 0x7f +#define NV_MEM_COMP_VM 0x03 + +/* NV_SW object class */ +#define NV_SW 0x0000506e +#define NV_SW_DMA_SEMAPHORE 0x00000060 +#define NV_SW_SEMAPHORE_OFFSET 0x00000064 +#define NV_SW_SEMAPHORE_ACQUIRE 0x00000068 +#define NV_SW_SEMAPHORE_RELEASE 0x0000006c +#define NV_SW_YIELD 0x00000080 +#define NV_SW_DMA_VBLSEM 0x0000018c +#define NV_SW_VBLSEM_OFFSET 0x00000400 +#define NV_SW_VBLSEM_RELEASE_VALUE 0x00000404 +#define NV_SW_VBLSEM_RELEASE 0x00000408 +#define NV_SW_PAGE_FLIP 0x00000500 + +#endif /* __NOUVEAU_DRV_H__ */ diff --git a/drivers/gpu/drm/nouveau/nouveau_encoder.h b/drivers/gpu/drm/nouveau/nouveau_encoder.h new file mode 100644 index 00000000..ae69b61d --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_encoder.h @@ -0,0 +1,103 @@ +/* + * Copyright (C) 2008 Maarten Maathuis. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __NOUVEAU_ENCODER_H__ +#define __NOUVEAU_ENCODER_H__ + +#include "drm_encoder_slave.h" +#include "nouveau_drv.h" + +#define NV_DPMS_CLEARED 0x80 + +struct nouveau_encoder { + struct drm_encoder_slave base; + + struct dcb_entry *dcb; + int or; + + /* different to drm_encoder.crtc, this reflects what's + * actually programmed on the hw, not the proposed crtc */ + struct drm_crtc *crtc; + + struct drm_display_mode mode; + int last_dpms; + + struct nv04_output_reg restore; + + union { + struct { + int mc_unknown; + uint32_t unk0; + uint32_t unk1; + int dpcd_version; + int link_nr; + int link_bw; + bool enhanced_frame; + } dp; + }; +}; + +static inline struct nouveau_encoder *nouveau_encoder(struct drm_encoder *enc) +{ + struct drm_encoder_slave *slave = to_encoder_slave(enc); + + return container_of(slave, struct nouveau_encoder, base); +} + +static inline struct drm_encoder *to_drm_encoder(struct nouveau_encoder *enc) +{ + return &enc->base.base; +} + +static inline struct drm_encoder_slave_funcs * +get_slave_funcs(struct drm_encoder *enc) +{ + return to_encoder_slave(enc)->slave_funcs; +} + +struct nouveau_connector * +nouveau_encoder_connector_get(struct nouveau_encoder *encoder); +int nv50_sor_create(struct drm_connector *, struct dcb_entry *); +int nv50_dac_create(struct drm_connector *, struct dcb_entry *); + +struct bit_displayport_encoder_table { + uint32_t match; + uint8_t record_nr; + uint8_t unknown; + uint16_t script0; + uint16_t script1; + uint16_t unknown_table; +} __attribute__ ((packed)); + +struct bit_displayport_encoder_table_entry { + uint8_t vs_level; + uint8_t pre_level; + uint8_t reg0; + uint8_t reg1; + uint8_t reg2; +} __attribute__ ((packed)); + +#endif /* __NOUVEAU_ENCODER_H__ */ diff --git a/drivers/gpu/drm/nouveau/nouveau_fb.h b/drivers/gpu/drm/nouveau/nouveau_fb.h new file mode 100644 index 00000000..a3a88ad0 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_fb.h @@ -0,0 +1,48 @@ +/* + * Copyright (C) 2008 Maarten Maathuis. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __NOUVEAU_FB_H__ +#define __NOUVEAU_FB_H__ + +struct nouveau_framebuffer { + struct drm_framebuffer base; + struct nouveau_bo *nvbo; + u32 r_dma; + u32 r_format; + u32 r_pitch; +}; + +static inline struct nouveau_framebuffer * +nouveau_framebuffer(struct drm_framebuffer *fb) +{ + return container_of(fb, struct nouveau_framebuffer, base); +} + +extern const struct drm_mode_config_funcs nouveau_mode_config_funcs; + +int nouveau_framebuffer_init(struct drm_device *dev, struct nouveau_framebuffer *nouveau_fb, + struct drm_mode_fb_cmd *mode_cmd, struct nouveau_bo *nvbo); +#endif /* __NOUVEAU_FB_H__ */ diff --git a/drivers/gpu/drm/nouveau/nouveau_fbcon.c b/drivers/gpu/drm/nouveau/nouveau_fbcon.c new file mode 100644 index 00000000..39aee6d4 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_fbcon.c @@ -0,0 +1,537 @@ +/* + * Copyright © 2007 David Airlie + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * David Airlie + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include "drmP.h" +#include "drm.h" +#include "drm_crtc.h" +#include "drm_crtc_helper.h" +#include "drm_fb_helper.h" +#include "nouveau_drv.h" +#include "nouveau_drm.h" +#include "nouveau_crtc.h" +#include "nouveau_fb.h" +#include "nouveau_fbcon.h" +#include "nouveau_dma.h" + +static void +nouveau_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect) +{ + struct nouveau_fbdev *nfbdev = info->par; + struct drm_device *dev = nfbdev->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + int ret; + + if (info->state != FBINFO_STATE_RUNNING) + return; + + ret = -ENODEV; + if (!in_interrupt() && !(info->flags & FBINFO_HWACCEL_DISABLED) && + mutex_trylock(&dev_priv->channel->mutex)) { + if (dev_priv->card_type < NV_50) + ret = nv04_fbcon_fillrect(info, rect); + else + if (dev_priv->card_type < NV_C0) + ret = nv50_fbcon_fillrect(info, rect); + else + ret = nvc0_fbcon_fillrect(info, rect); + mutex_unlock(&dev_priv->channel->mutex); + } + + if (ret == 0) + return; + + if (ret != -ENODEV) + nouveau_fbcon_gpu_lockup(info); + cfb_fillrect(info, rect); +} + +static void +nouveau_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *image) +{ + struct nouveau_fbdev *nfbdev = info->par; + struct drm_device *dev = nfbdev->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + int ret; + + if (info->state != FBINFO_STATE_RUNNING) + return; + + ret = -ENODEV; + if (!in_interrupt() && !(info->flags & FBINFO_HWACCEL_DISABLED) && + mutex_trylock(&dev_priv->channel->mutex)) { + if (dev_priv->card_type < NV_50) + ret = nv04_fbcon_copyarea(info, image); + else + if (dev_priv->card_type < NV_C0) + ret = nv50_fbcon_copyarea(info, image); + else + ret = nvc0_fbcon_copyarea(info, image); + mutex_unlock(&dev_priv->channel->mutex); + } + + if (ret == 0) + return; + + if (ret != -ENODEV) + nouveau_fbcon_gpu_lockup(info); + cfb_copyarea(info, image); +} + +static void +nouveau_fbcon_imageblit(struct fb_info *info, const struct fb_image *image) +{ + struct nouveau_fbdev *nfbdev = info->par; + struct drm_device *dev = nfbdev->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + int ret; + + if (info->state != FBINFO_STATE_RUNNING) + return; + + ret = -ENODEV; + if (!in_interrupt() && !(info->flags & FBINFO_HWACCEL_DISABLED) && + mutex_trylock(&dev_priv->channel->mutex)) { + if (dev_priv->card_type < NV_50) + ret = nv04_fbcon_imageblit(info, image); + else + if (dev_priv->card_type < NV_C0) + ret = nv50_fbcon_imageblit(info, image); + else + ret = nvc0_fbcon_imageblit(info, image); + mutex_unlock(&dev_priv->channel->mutex); + } + + if (ret == 0) + return; + + if (ret != -ENODEV) + nouveau_fbcon_gpu_lockup(info); + cfb_imageblit(info, image); +} + +static int +nouveau_fbcon_sync(struct fb_info *info) +{ + struct nouveau_fbdev *nfbdev = info->par; + struct drm_device *dev = nfbdev->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_channel *chan = dev_priv->channel; + int ret, i; + + if (!chan || !chan->accel_done || in_interrupt() || + info->state != FBINFO_STATE_RUNNING || + info->flags & FBINFO_HWACCEL_DISABLED) + return 0; + + if (!mutex_trylock(&chan->mutex)) + return 0; + + ret = RING_SPACE(chan, 4); + if (ret) { + mutex_unlock(&chan->mutex); + nouveau_fbcon_gpu_lockup(info); + return 0; + } + + if (dev_priv->card_type >= NV_C0) { + BEGIN_NVC0(chan, 2, NvSub2D, 0x010c, 1); + OUT_RING (chan, 0); + BEGIN_NVC0(chan, 2, NvSub2D, 0x0100, 1); + OUT_RING (chan, 0); + } else { + BEGIN_RING(chan, 0, 0x0104, 1); + OUT_RING (chan, 0); + BEGIN_RING(chan, 0, 0x0100, 1); + OUT_RING (chan, 0); + } + + nouveau_bo_wr32(chan->notifier_bo, chan->m2mf_ntfy/4 + 3, 0xffffffff); + FIRE_RING(chan); + mutex_unlock(&chan->mutex); + + ret = -EBUSY; + for (i = 0; i < 100000; i++) { + if (!nouveau_bo_rd32(chan->notifier_bo, chan->m2mf_ntfy/4 + 3)) { + ret = 0; + break; + } + DRM_UDELAY(1); + } + + if (ret) { + nouveau_fbcon_gpu_lockup(info); + return 0; + } + + chan->accel_done = false; + return 0; +} + +static struct fb_ops nouveau_fbcon_ops = { + .owner = THIS_MODULE, + .fb_check_var = drm_fb_helper_check_var, + .fb_set_par = drm_fb_helper_set_par, + .fb_fillrect = nouveau_fbcon_fillrect, + .fb_copyarea = nouveau_fbcon_copyarea, + .fb_imageblit = nouveau_fbcon_imageblit, + .fb_sync = nouveau_fbcon_sync, + .fb_pan_display = drm_fb_helper_pan_display, + .fb_blank = drm_fb_helper_blank, + .fb_setcmap = drm_fb_helper_setcmap, + .fb_debug_enter = drm_fb_helper_debug_enter, + .fb_debug_leave = drm_fb_helper_debug_leave, +}; + +static struct fb_ops nouveau_fbcon_sw_ops = { + .owner = THIS_MODULE, + .fb_check_var = drm_fb_helper_check_var, + .fb_set_par = drm_fb_helper_set_par, + .fb_fillrect = cfb_fillrect, + .fb_copyarea = cfb_copyarea, + .fb_imageblit = cfb_imageblit, + .fb_pan_display = drm_fb_helper_pan_display, + .fb_blank = drm_fb_helper_blank, + .fb_setcmap = drm_fb_helper_setcmap, + .fb_debug_enter = drm_fb_helper_debug_enter, + .fb_debug_leave = drm_fb_helper_debug_leave, +}; + +static void nouveau_fbcon_gamma_set(struct drm_crtc *crtc, u16 red, u16 green, + u16 blue, int regno) +{ + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + + nv_crtc->lut.r[regno] = red; + nv_crtc->lut.g[regno] = green; + nv_crtc->lut.b[regno] = blue; +} + +static void nouveau_fbcon_gamma_get(struct drm_crtc *crtc, u16 *red, u16 *green, + u16 *blue, int regno) +{ + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + + *red = nv_crtc->lut.r[regno]; + *green = nv_crtc->lut.g[regno]; + *blue = nv_crtc->lut.b[regno]; +} + +static void +nouveau_fbcon_zfill(struct drm_device *dev, struct nouveau_fbdev *nfbdev) +{ + struct fb_info *info = nfbdev->helper.fbdev; + struct fb_fillrect rect; + + /* Clear the entire fbcon. The drm will program every connector + * with it's preferred mode. If the sizes differ, one display will + * quite likely have garbage around the console. + */ + rect.dx = rect.dy = 0; + rect.width = info->var.xres_virtual; + rect.height = info->var.yres_virtual; + rect.color = 0; + rect.rop = ROP_COPY; + info->fbops->fb_fillrect(info, &rect); +} + +static int +nouveau_fbcon_create(struct nouveau_fbdev *nfbdev, + struct drm_fb_helper_surface_size *sizes) +{ + struct drm_device *dev = nfbdev->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct fb_info *info; + struct drm_framebuffer *fb; + struct nouveau_framebuffer *nouveau_fb; + struct nouveau_bo *nvbo; + struct drm_mode_fb_cmd mode_cmd; + struct pci_dev *pdev = dev->pdev; + struct device *device = &pdev->dev; + int size, ret; + + mode_cmd.width = sizes->surface_width; + mode_cmd.height = sizes->surface_height; + + mode_cmd.bpp = sizes->surface_bpp; + mode_cmd.pitch = mode_cmd.width * (mode_cmd.bpp >> 3); + mode_cmd.pitch = roundup(mode_cmd.pitch, 256); + mode_cmd.depth = sizes->surface_depth; + + size = mode_cmd.pitch * mode_cmd.height; + size = roundup(size, PAGE_SIZE); + + ret = nouveau_gem_new(dev, dev_priv->channel, size, 0, + NOUVEAU_GEM_DOMAIN_VRAM, 0, 0x0000, &nvbo); + if (ret) { + NV_ERROR(dev, "failed to allocate framebuffer\n"); + goto out; + } + + ret = nouveau_bo_pin(nvbo, TTM_PL_FLAG_VRAM); + if (ret) { + NV_ERROR(dev, "failed to pin fb: %d\n", ret); + nouveau_bo_ref(NULL, &nvbo); + goto out; + } + + ret = nouveau_bo_map(nvbo); + if (ret) { + NV_ERROR(dev, "failed to map fb: %d\n", ret); + nouveau_bo_unpin(nvbo); + nouveau_bo_ref(NULL, &nvbo); + goto out; + } + + mutex_lock(&dev->struct_mutex); + + info = framebuffer_alloc(0, device); + if (!info) { + ret = -ENOMEM; + goto out_unref; + } + + ret = fb_alloc_cmap(&info->cmap, 256, 0); + if (ret) { + ret = -ENOMEM; + goto out_unref; + } + + info->par = nfbdev; + + nouveau_framebuffer_init(dev, &nfbdev->nouveau_fb, &mode_cmd, nvbo); + + nouveau_fb = &nfbdev->nouveau_fb; + fb = &nouveau_fb->base; + + /* setup helper */ + nfbdev->helper.fb = fb; + nfbdev->helper.fbdev = info; + + strcpy(info->fix.id, "nouveaufb"); + if (nouveau_nofbaccel) + info->flags = FBINFO_DEFAULT | FBINFO_HWACCEL_DISABLED; + else + info->flags = FBINFO_DEFAULT | FBINFO_HWACCEL_COPYAREA | + FBINFO_HWACCEL_FILLRECT | + FBINFO_HWACCEL_IMAGEBLIT; + info->flags |= FBINFO_CAN_FORCE_OUTPUT; + info->fbops = &nouveau_fbcon_sw_ops; + info->fix.smem_start = nvbo->bo.mem.bus.base + + nvbo->bo.mem.bus.offset; + info->fix.smem_len = size; + + info->screen_base = nvbo_kmap_obj_iovirtual(nouveau_fb->nvbo); + info->screen_size = size; + + drm_fb_helper_fill_fix(info, fb->pitch, fb->depth); + drm_fb_helper_fill_var(info, &nfbdev->helper, sizes->fb_width, sizes->fb_height); + + /* Set aperture base/size for vesafb takeover */ + info->apertures = dev_priv->apertures; + if (!info->apertures) { + ret = -ENOMEM; + goto out_unref; + } + + info->pixmap.size = 64*1024; + info->pixmap.buf_align = 8; + info->pixmap.access_align = 32; + info->pixmap.flags = FB_PIXMAP_SYSTEM; + info->pixmap.scan_align = 1; + + mutex_unlock(&dev->struct_mutex); + + if (dev_priv->channel && !nouveau_nofbaccel) { + ret = -ENODEV; + if (dev_priv->card_type < NV_50) + ret = nv04_fbcon_accel_init(info); + else + if (dev_priv->card_type < NV_C0) + ret = nv50_fbcon_accel_init(info); + else + ret = nvc0_fbcon_accel_init(info); + + if (ret == 0) + info->fbops = &nouveau_fbcon_ops; + } + + nouveau_fbcon_zfill(dev, nfbdev); + + /* To allow resizeing without swapping buffers */ + NV_INFO(dev, "allocated %dx%d fb: 0x%lx, bo %p\n", + nouveau_fb->base.width, + nouveau_fb->base.height, + nvbo->bo.offset, nvbo); + + vga_switcheroo_client_fb_set(dev->pdev, info); + return 0; + +out_unref: + mutex_unlock(&dev->struct_mutex); +out: + return ret; +} + +static int +nouveau_fbcon_find_or_create_single(struct drm_fb_helper *helper, + struct drm_fb_helper_surface_size *sizes) +{ + struct nouveau_fbdev *nfbdev = (struct nouveau_fbdev *)helper; + int new_fb = 0; + int ret; + + if (!helper->fb) { + ret = nouveau_fbcon_create(nfbdev, sizes); + if (ret) + return ret; + new_fb = 1; + } + return new_fb; +} + +void +nouveau_fbcon_output_poll_changed(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + drm_fb_helper_hotplug_event(&dev_priv->nfbdev->helper); +} + +static int +nouveau_fbcon_destroy(struct drm_device *dev, struct nouveau_fbdev *nfbdev) +{ + struct nouveau_framebuffer *nouveau_fb = &nfbdev->nouveau_fb; + struct fb_info *info; + + if (nfbdev->helper.fbdev) { + info = nfbdev->helper.fbdev; + unregister_framebuffer(info); + if (info->cmap.len) + fb_dealloc_cmap(&info->cmap); + framebuffer_release(info); + } + + if (nouveau_fb->nvbo) { + nouveau_bo_unmap(nouveau_fb->nvbo); + drm_gem_object_unreference_unlocked(nouveau_fb->nvbo->gem); + nouveau_fb->nvbo = NULL; + } + drm_fb_helper_fini(&nfbdev->helper); + drm_framebuffer_cleanup(&nouveau_fb->base); + return 0; +} + +void nouveau_fbcon_gpu_lockup(struct fb_info *info) +{ + struct nouveau_fbdev *nfbdev = info->par; + struct drm_device *dev = nfbdev->dev; + + NV_ERROR(dev, "GPU lockup - switching to software fbcon\n"); + info->flags |= FBINFO_HWACCEL_DISABLED; +} + +static struct drm_fb_helper_funcs nouveau_fbcon_helper_funcs = { + .gamma_set = nouveau_fbcon_gamma_set, + .gamma_get = nouveau_fbcon_gamma_get, + .fb_probe = nouveau_fbcon_find_or_create_single, +}; + + +int nouveau_fbcon_init(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fbdev *nfbdev; + int ret; + + nfbdev = kzalloc(sizeof(struct nouveau_fbdev), GFP_KERNEL); + if (!nfbdev) + return -ENOMEM; + + nfbdev->dev = dev; + dev_priv->nfbdev = nfbdev; + nfbdev->helper.funcs = &nouveau_fbcon_helper_funcs; + + ret = drm_fb_helper_init(dev, &nfbdev->helper, + nv_two_heads(dev) ? 2 : 1, 4); + if (ret) { + kfree(nfbdev); + return ret; + } + + drm_fb_helper_single_add_all_connectors(&nfbdev->helper); + drm_fb_helper_initial_config(&nfbdev->helper, 32); + return 0; +} + +void nouveau_fbcon_fini(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + if (!dev_priv->nfbdev) + return; + + nouveau_fbcon_destroy(dev, dev_priv->nfbdev); + kfree(dev_priv->nfbdev); + dev_priv->nfbdev = NULL; +} + +void nouveau_fbcon_save_disable_accel(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + dev_priv->nfbdev->saved_flags = dev_priv->nfbdev->helper.fbdev->flags; + dev_priv->nfbdev->helper.fbdev->flags |= FBINFO_HWACCEL_DISABLED; +} + +void nouveau_fbcon_restore_accel(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + dev_priv->nfbdev->helper.fbdev->flags = dev_priv->nfbdev->saved_flags; +} + +void nouveau_fbcon_set_suspend(struct drm_device *dev, int state) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + fb_set_suspend(dev_priv->nfbdev->helper.fbdev, state); +} + +void nouveau_fbcon_zfill_all(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + nouveau_fbcon_zfill(dev, dev_priv->nfbdev); +} diff --git a/drivers/gpu/drm/nouveau/nouveau_fbcon.h b/drivers/gpu/drm/nouveau/nouveau_fbcon.h new file mode 100644 index 00000000..b73c29f8 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_fbcon.h @@ -0,0 +1,69 @@ +/* + * Copyright (C) 2008 Maarten Maathuis. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __NOUVEAU_FBCON_H__ +#define __NOUVEAU_FBCON_H__ + +#include "drm_fb_helper.h" + +#include "nouveau_fb.h" +struct nouveau_fbdev { + struct drm_fb_helper helper; + struct nouveau_framebuffer nouveau_fb; + struct list_head fbdev_list; + struct drm_device *dev; + unsigned int saved_flags; +}; + +void nouveau_fbcon_restore(void); + +int nv04_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *region); +int nv04_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect); +int nv04_fbcon_imageblit(struct fb_info *info, const struct fb_image *image); +int nv04_fbcon_accel_init(struct fb_info *info); + +int nv50_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect); +int nv50_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *region); +int nv50_fbcon_imageblit(struct fb_info *info, const struct fb_image *image); +int nv50_fbcon_accel_init(struct fb_info *info); + +int nvc0_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect); +int nvc0_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *region); +int nvc0_fbcon_imageblit(struct fb_info *info, const struct fb_image *image); +int nvc0_fbcon_accel_init(struct fb_info *info); + +void nouveau_fbcon_gpu_lockup(struct fb_info *info); + +int nouveau_fbcon_init(struct drm_device *dev); +void nouveau_fbcon_fini(struct drm_device *dev); +void nouveau_fbcon_set_suspend(struct drm_device *dev, int state); +void nouveau_fbcon_zfill_all(struct drm_device *dev); +void nouveau_fbcon_save_disable_accel(struct drm_device *dev); +void nouveau_fbcon_restore_accel(struct drm_device *dev); + +void nouveau_fbcon_output_poll_changed(struct drm_device *dev); +#endif /* __NV50_FBCON_H__ */ + diff --git a/drivers/gpu/drm/nouveau/nouveau_fence.c b/drivers/gpu/drm/nouveau/nouveau_fence.c new file mode 100644 index 00000000..56f06b0c --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_fence.c @@ -0,0 +1,617 @@ +/* + * Copyright (C) 2007 Ben Skeggs. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "drmP.h" +#include "drm.h" + +#include +#include + +#include "nouveau_drv.h" +#include "nouveau_ramht.h" +#include "nouveau_dma.h" + +#define USE_REFCNT(dev) (nouveau_private(dev)->chipset >= 0x10) +#define USE_SEMA(dev) (nouveau_private(dev)->chipset >= 0x17) + +struct nouveau_fence { + struct nouveau_channel *channel; + struct kref refcount; + struct list_head entry; + + uint32_t sequence; + bool signalled; + + void (*work)(void *priv, bool signalled); + void *priv; +}; + +struct nouveau_semaphore { + struct kref ref; + struct drm_device *dev; + struct drm_mm_node *mem; +}; + +static inline struct nouveau_fence * +nouveau_fence(void *sync_obj) +{ + return (struct nouveau_fence *)sync_obj; +} + +static void +nouveau_fence_del(struct kref *ref) +{ + struct nouveau_fence *fence = + container_of(ref, struct nouveau_fence, refcount); + + nouveau_channel_ref(NULL, &fence->channel); + kfree(fence); +} + +void +nouveau_fence_update(struct nouveau_channel *chan) +{ + struct drm_device *dev = chan->dev; + struct nouveau_fence *tmp, *fence; + uint32_t sequence; + + spin_lock(&chan->fence.lock); + + /* Fetch the last sequence if the channel is still up and running */ + if (likely(!list_empty(&chan->fence.pending))) { + if (USE_REFCNT(dev)) + sequence = nvchan_rd32(chan, 0x48); + else + sequence = atomic_read(&chan->fence.last_sequence_irq); + + if (chan->fence.sequence_ack == sequence) + goto out; + chan->fence.sequence_ack = sequence; + } + + list_for_each_entry_safe(fence, tmp, &chan->fence.pending, entry) { + sequence = fence->sequence; + fence->signalled = true; + list_del(&fence->entry); + + if (unlikely(fence->work)) + fence->work(fence->priv, true); + + kref_put(&fence->refcount, nouveau_fence_del); + + if (sequence == chan->fence.sequence_ack) + break; + } +out: + spin_unlock(&chan->fence.lock); +} + +int +nouveau_fence_new(struct nouveau_channel *chan, struct nouveau_fence **pfence, + bool emit) +{ + struct nouveau_fence *fence; + int ret = 0; + + fence = kzalloc(sizeof(*fence), GFP_KERNEL); + if (!fence) + return -ENOMEM; + kref_init(&fence->refcount); + nouveau_channel_ref(chan, &fence->channel); + + if (emit) + ret = nouveau_fence_emit(fence); + + if (ret) + nouveau_fence_unref(&fence); + *pfence = fence; + return ret; +} + +struct nouveau_channel * +nouveau_fence_channel(struct nouveau_fence *fence) +{ + return fence ? nouveau_channel_get_unlocked(fence->channel) : NULL; +} + +int +nouveau_fence_emit(struct nouveau_fence *fence) +{ + struct nouveau_channel *chan = fence->channel; + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + int ret; + + ret = RING_SPACE(chan, 2); + if (ret) + return ret; + + if (unlikely(chan->fence.sequence == chan->fence.sequence_ack - 1)) { + nouveau_fence_update(chan); + + BUG_ON(chan->fence.sequence == + chan->fence.sequence_ack - 1); + } + + fence->sequence = ++chan->fence.sequence; + + kref_get(&fence->refcount); + spin_lock(&chan->fence.lock); + list_add_tail(&fence->entry, &chan->fence.pending); + spin_unlock(&chan->fence.lock); + + if (USE_REFCNT(dev)) { + if (dev_priv->card_type < NV_C0) + BEGIN_RING(chan, NvSubSw, 0x0050, 1); + else + BEGIN_NVC0(chan, 2, NvSubM2MF, 0x0050, 1); + } else { + BEGIN_RING(chan, NvSubSw, 0x0150, 1); + } + OUT_RING (chan, fence->sequence); + FIRE_RING(chan); + + return 0; +} + +void +nouveau_fence_work(struct nouveau_fence *fence, + void (*work)(void *priv, bool signalled), + void *priv) +{ + BUG_ON(fence->work); + + spin_lock(&fence->channel->fence.lock); + + if (fence->signalled) { + work(priv, true); + } else { + fence->work = work; + fence->priv = priv; + } + + spin_unlock(&fence->channel->fence.lock); +} + +void +__nouveau_fence_unref(void **sync_obj) +{ + struct nouveau_fence *fence = nouveau_fence(*sync_obj); + + if (fence) + kref_put(&fence->refcount, nouveau_fence_del); + *sync_obj = NULL; +} + +void * +__nouveau_fence_ref(void *sync_obj) +{ + struct nouveau_fence *fence = nouveau_fence(sync_obj); + + kref_get(&fence->refcount); + return sync_obj; +} + +bool +__nouveau_fence_signalled(void *sync_obj, void *sync_arg) +{ + struct nouveau_fence *fence = nouveau_fence(sync_obj); + struct nouveau_channel *chan = fence->channel; + + if (fence->signalled) + return true; + + nouveau_fence_update(chan); + return fence->signalled; +} + +int +__nouveau_fence_wait(void *sync_obj, void *sync_arg, bool lazy, bool intr) +{ + unsigned long timeout = jiffies + (3 * DRM_HZ); + unsigned long sleep_time = NSEC_PER_MSEC / 1000; + ktime_t t; + int ret = 0; + + while (1) { + if (__nouveau_fence_signalled(sync_obj, sync_arg)) + break; + + if (time_after_eq(jiffies, timeout)) { + ret = -EBUSY; + break; + } + + __set_current_state(intr ? TASK_INTERRUPTIBLE + : TASK_UNINTERRUPTIBLE); + if (lazy) { + t = ktime_set(0, sleep_time); + schedule_hrtimeout(&t, HRTIMER_MODE_REL); + sleep_time *= 2; + if (sleep_time > NSEC_PER_MSEC) + sleep_time = NSEC_PER_MSEC; + } + + if (intr && signal_pending(current)) { + ret = -ERESTARTSYS; + break; + } + } + + __set_current_state(TASK_RUNNING); + + return ret; +} + +static struct nouveau_semaphore * +semaphore_alloc(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_semaphore *sema; + int size = (dev_priv->chipset < 0x84) ? 4 : 16; + int ret, i; + + if (!USE_SEMA(dev)) + return NULL; + + sema = kmalloc(sizeof(*sema), GFP_KERNEL); + if (!sema) + goto fail; + + ret = drm_mm_pre_get(&dev_priv->fence.heap); + if (ret) + goto fail; + + spin_lock(&dev_priv->fence.lock); + sema->mem = drm_mm_search_free(&dev_priv->fence.heap, size, 0, 0); + if (sema->mem) + sema->mem = drm_mm_get_block_atomic(sema->mem, size, 0); + spin_unlock(&dev_priv->fence.lock); + + if (!sema->mem) + goto fail; + + kref_init(&sema->ref); + sema->dev = dev; + for (i = sema->mem->start; i < sema->mem->start + size; i += 4) + nouveau_bo_wr32(dev_priv->fence.bo, i / 4, 0); + + return sema; +fail: + kfree(sema); + return NULL; +} + +static void +semaphore_free(struct kref *ref) +{ + struct nouveau_semaphore *sema = + container_of(ref, struct nouveau_semaphore, ref); + struct drm_nouveau_private *dev_priv = sema->dev->dev_private; + + spin_lock(&dev_priv->fence.lock); + drm_mm_put_block(sema->mem); + spin_unlock(&dev_priv->fence.lock); + + kfree(sema); +} + +static void +semaphore_work(void *priv, bool signalled) +{ + struct nouveau_semaphore *sema = priv; + struct drm_nouveau_private *dev_priv = sema->dev->dev_private; + + if (unlikely(!signalled)) + nouveau_bo_wr32(dev_priv->fence.bo, sema->mem->start / 4, 1); + + kref_put(&sema->ref, semaphore_free); +} + +static int +semaphore_acquire(struct nouveau_channel *chan, struct nouveau_semaphore *sema) +{ + struct drm_nouveau_private *dev_priv = chan->dev->dev_private; + struct nouveau_fence *fence = NULL; + int ret; + + if (dev_priv->chipset < 0x84) { + ret = RING_SPACE(chan, 4); + if (ret) + return ret; + + BEGIN_RING(chan, NvSubSw, NV_SW_DMA_SEMAPHORE, 3); + OUT_RING (chan, NvSema); + OUT_RING (chan, sema->mem->start); + OUT_RING (chan, 1); + } else + if (dev_priv->chipset < 0xc0) { + struct nouveau_vma *vma = &dev_priv->fence.bo->vma; + u64 offset = vma->offset + sema->mem->start; + + ret = RING_SPACE(chan, 7); + if (ret) + return ret; + + BEGIN_RING(chan, NvSubSw, NV_SW_DMA_SEMAPHORE, 1); + OUT_RING (chan, chan->vram_handle); + BEGIN_RING(chan, NvSubSw, 0x0010, 4); + OUT_RING (chan, upper_32_bits(offset)); + OUT_RING (chan, lower_32_bits(offset)); + OUT_RING (chan, 1); + OUT_RING (chan, 1); /* ACQUIRE_EQ */ + } else { + struct nouveau_vma *vma = &dev_priv->fence.bo->vma; + u64 offset = vma->offset + sema->mem->start; + + ret = RING_SPACE(chan, 5); + if (ret) + return ret; + + BEGIN_NVC0(chan, 2, NvSubM2MF, 0x0010, 4); + OUT_RING (chan, upper_32_bits(offset)); + OUT_RING (chan, lower_32_bits(offset)); + OUT_RING (chan, 1); + OUT_RING (chan, 0x1001); /* ACQUIRE_EQ */ + } + + /* Delay semaphore destruction until its work is done */ + ret = nouveau_fence_new(chan, &fence, true); + if (ret) + return ret; + + kref_get(&sema->ref); + nouveau_fence_work(fence, semaphore_work, sema); + nouveau_fence_unref(&fence); + return 0; +} + +static int +semaphore_release(struct nouveau_channel *chan, struct nouveau_semaphore *sema) +{ + struct drm_nouveau_private *dev_priv = chan->dev->dev_private; + struct nouveau_fence *fence = NULL; + int ret; + + if (dev_priv->chipset < 0x84) { + ret = RING_SPACE(chan, 5); + if (ret) + return ret; + + BEGIN_RING(chan, NvSubSw, NV_SW_DMA_SEMAPHORE, 2); + OUT_RING (chan, NvSema); + OUT_RING (chan, sema->mem->start); + BEGIN_RING(chan, NvSubSw, NV_SW_SEMAPHORE_RELEASE, 1); + OUT_RING (chan, 1); + } else + if (dev_priv->chipset < 0xc0) { + struct nouveau_vma *vma = &dev_priv->fence.bo->vma; + u64 offset = vma->offset + sema->mem->start; + + ret = RING_SPACE(chan, 7); + if (ret) + return ret; + + BEGIN_RING(chan, NvSubSw, NV_SW_DMA_SEMAPHORE, 1); + OUT_RING (chan, chan->vram_handle); + BEGIN_RING(chan, NvSubSw, 0x0010, 4); + OUT_RING (chan, upper_32_bits(offset)); + OUT_RING (chan, lower_32_bits(offset)); + OUT_RING (chan, 1); + OUT_RING (chan, 2); /* RELEASE */ + } else { + struct nouveau_vma *vma = &dev_priv->fence.bo->vma; + u64 offset = vma->offset + sema->mem->start; + + ret = RING_SPACE(chan, 5); + if (ret) + return ret; + + BEGIN_NVC0(chan, 2, NvSubM2MF, 0x0010, 4); + OUT_RING (chan, upper_32_bits(offset)); + OUT_RING (chan, lower_32_bits(offset)); + OUT_RING (chan, 1); + OUT_RING (chan, 0x1002); /* RELEASE */ + } + + /* Delay semaphore destruction until its work is done */ + ret = nouveau_fence_new(chan, &fence, true); + if (ret) + return ret; + + kref_get(&sema->ref); + nouveau_fence_work(fence, semaphore_work, sema); + nouveau_fence_unref(&fence); + return 0; +} + +int +nouveau_fence_sync(struct nouveau_fence *fence, + struct nouveau_channel *wchan) +{ + struct nouveau_channel *chan = nouveau_fence_channel(fence); + struct drm_device *dev = wchan->dev; + struct nouveau_semaphore *sema; + int ret = 0; + + if (likely(!chan || chan == wchan || + nouveau_fence_signalled(fence))) + goto out; + + sema = semaphore_alloc(dev); + if (!sema) { + /* Early card or broken userspace, fall back to + * software sync. */ + ret = nouveau_fence_wait(fence, true, false); + goto out; + } + + /* try to take chan's mutex, if we can't take it right away + * we have to fallback to software sync to prevent locking + * order issues + */ + if (!mutex_trylock(&chan->mutex)) { + ret = nouveau_fence_wait(fence, true, false); + goto out_unref; + } + + /* Make wchan wait until it gets signalled */ + ret = semaphore_acquire(wchan, sema); + if (ret) + goto out_unlock; + + /* Signal the semaphore from chan */ + ret = semaphore_release(chan, sema); + +out_unlock: + mutex_unlock(&chan->mutex); +out_unref: + kref_put(&sema->ref, semaphore_free); +out: + if (chan) + nouveau_channel_put_unlocked(&chan); + return ret; +} + +int +__nouveau_fence_flush(void *sync_obj, void *sync_arg) +{ + return 0; +} + +int +nouveau_fence_channel_init(struct nouveau_channel *chan) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *obj = NULL; + int ret; + + if (dev_priv->card_type < NV_C0) { + /* Create an NV_SW object for various sync purposes */ + ret = nouveau_gpuobj_gr_new(chan, NvSw, NV_SW); + if (ret) + return ret; + + ret = RING_SPACE(chan, 2); + if (ret) + return ret; + + BEGIN_RING(chan, NvSubSw, 0, 1); + OUT_RING (chan, NvSw); + FIRE_RING (chan); + } + + /* Setup area of memory shared between all channels for x-chan sync */ + if (USE_SEMA(dev) && dev_priv->chipset < 0x84) { + struct ttm_mem_reg *mem = &dev_priv->fence.bo->bo.mem; + + ret = nouveau_gpuobj_dma_new(chan, NV_CLASS_DMA_IN_MEMORY, + mem->start << PAGE_SHIFT, + mem->size, NV_MEM_ACCESS_RW, + NV_MEM_TARGET_VRAM, &obj); + if (ret) + return ret; + + ret = nouveau_ramht_insert(chan, NvSema, obj); + nouveau_gpuobj_ref(NULL, &obj); + if (ret) + return ret; + } + + atomic_set(&chan->fence.last_sequence_irq, 0); + return 0; +} + +void +nouveau_fence_channel_fini(struct nouveau_channel *chan) +{ + struct nouveau_fence *tmp, *fence; + + spin_lock(&chan->fence.lock); + + list_for_each_entry_safe(fence, tmp, &chan->fence.pending, entry) { + fence->signalled = true; + list_del(&fence->entry); + + if (unlikely(fence->work)) + fence->work(fence->priv, false); + + kref_put(&fence->refcount, nouveau_fence_del); + } + + spin_unlock(&chan->fence.lock); +} + +int +nouveau_fence_init(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + int size = (dev_priv->chipset < 0x84) ? 4096 : 16384; + int ret; + + /* Create a shared VRAM heap for cross-channel sync. */ + if (USE_SEMA(dev)) { + ret = nouveau_bo_new(dev, NULL, size, 0, TTM_PL_FLAG_VRAM, + 0, 0, &dev_priv->fence.bo); + if (ret) + return ret; + + ret = nouveau_bo_pin(dev_priv->fence.bo, TTM_PL_FLAG_VRAM); + if (ret) + goto fail; + + ret = nouveau_bo_map(dev_priv->fence.bo); + if (ret) + goto fail; + + ret = drm_mm_init(&dev_priv->fence.heap, 0, + dev_priv->fence.bo->bo.mem.size); + if (ret) + goto fail; + + spin_lock_init(&dev_priv->fence.lock); + } + + return 0; +fail: + nouveau_bo_unmap(dev_priv->fence.bo); + nouveau_bo_ref(NULL, &dev_priv->fence.bo); + return ret; +} + +void +nouveau_fence_fini(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + if (USE_SEMA(dev)) { + drm_mm_takedown(&dev_priv->fence.heap); + nouveau_bo_unmap(dev_priv->fence.bo); + nouveau_bo_unpin(dev_priv->fence.bo); + nouveau_bo_ref(NULL, &dev_priv->fence.bo); + } +} diff --git a/drivers/gpu/drm/nouveau/nouveau_gem.c b/drivers/gpu/drm/nouveau/nouveau_gem.c new file mode 100644 index 00000000..cee78b26 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_gem.c @@ -0,0 +1,808 @@ +/* + * Copyright (C) 2008 Ben Skeggs. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "drmP.h" +#include "drm.h" + +#include "nouveau_drv.h" +#include "nouveau_drm.h" +#include "nouveau_dma.h" + +#define nouveau_gem_pushbuf_sync(chan) 0 + +int +nouveau_gem_object_new(struct drm_gem_object *gem) +{ + return 0; +} + +void +nouveau_gem_object_del(struct drm_gem_object *gem) +{ + struct nouveau_bo *nvbo = gem->driver_private; + struct ttm_buffer_object *bo = &nvbo->bo; + + if (!nvbo) + return; + nvbo->gem = NULL; + + if (unlikely(nvbo->pin_refcnt)) { + nvbo->pin_refcnt = 1; + nouveau_bo_unpin(nvbo); + } + + ttm_bo_unref(&bo); + + drm_gem_object_release(gem); + kfree(gem); +} + +int +nouveau_gem_new(struct drm_device *dev, struct nouveau_channel *chan, + int size, int align, uint32_t domain, uint32_t tile_mode, + uint32_t tile_flags, struct nouveau_bo **pnvbo) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_bo *nvbo; + u32 flags = 0; + int ret; + + if (domain & NOUVEAU_GEM_DOMAIN_VRAM) + flags |= TTM_PL_FLAG_VRAM; + if (domain & NOUVEAU_GEM_DOMAIN_GART) + flags |= TTM_PL_FLAG_TT; + if (!flags || domain & NOUVEAU_GEM_DOMAIN_CPU) + flags |= TTM_PL_FLAG_SYSTEM; + + ret = nouveau_bo_new(dev, chan, size, align, flags, tile_mode, + tile_flags, pnvbo); + if (ret) + return ret; + nvbo = *pnvbo; + + /* we restrict allowed domains on nv50+ to only the types + * that were requested at creation time. not possibly on + * earlier chips without busting the ABI. + */ + nvbo->valid_domains = NOUVEAU_GEM_DOMAIN_VRAM | + NOUVEAU_GEM_DOMAIN_GART; + if (dev_priv->card_type >= NV_50) + nvbo->valid_domains &= domain; + + nvbo->gem = drm_gem_object_alloc(dev, nvbo->bo.mem.size); + if (!nvbo->gem) { + nouveau_bo_ref(NULL, pnvbo); + return -ENOMEM; + } + + nvbo->bo.persistent_swap_storage = nvbo->gem->filp; + nvbo->gem->driver_private = nvbo; + return 0; +} + +static int +nouveau_gem_info(struct drm_gem_object *gem, struct drm_nouveau_gem_info *rep) +{ + struct nouveau_bo *nvbo = nouveau_gem_object(gem); + + if (nvbo->bo.mem.mem_type == TTM_PL_TT) + rep->domain = NOUVEAU_GEM_DOMAIN_GART; + else + rep->domain = NOUVEAU_GEM_DOMAIN_VRAM; + + rep->size = nvbo->bo.mem.num_pages << PAGE_SHIFT; + rep->offset = nvbo->bo.offset; + rep->map_handle = nvbo->bo.addr_space_offset; + rep->tile_mode = nvbo->tile_mode; + rep->tile_flags = nvbo->tile_flags; + return 0; +} + +int +nouveau_gem_ioctl_new(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct drm_nouveau_gem_new *req = data; + struct nouveau_bo *nvbo = NULL; + struct nouveau_channel *chan = NULL; + int ret = 0; + + if (unlikely(dev_priv->ttm.bdev.dev_mapping == NULL)) + dev_priv->ttm.bdev.dev_mapping = dev_priv->dev->dev_mapping; + + if (!dev_priv->engine.vram.flags_valid(dev, req->info.tile_flags)) { + NV_ERROR(dev, "bad page flags: 0x%08x\n", req->info.tile_flags); + return -EINVAL; + } + + if (req->channel_hint) { + chan = nouveau_channel_get(dev, file_priv, req->channel_hint); + if (IS_ERR(chan)) + return PTR_ERR(chan); + } + + ret = nouveau_gem_new(dev, chan, req->info.size, req->align, + req->info.domain, req->info.tile_mode, + req->info.tile_flags, &nvbo); + if (chan) + nouveau_channel_put(&chan); + if (ret) + return ret; + + ret = nouveau_gem_info(nvbo->gem, &req->info); + if (ret) + goto out; + + ret = drm_gem_handle_create(file_priv, nvbo->gem, &req->info.handle); + /* drop reference from allocate - handle holds it now */ + drm_gem_object_unreference_unlocked(nvbo->gem); +out: + return ret; +} + +static int +nouveau_gem_set_domain(struct drm_gem_object *gem, uint32_t read_domains, + uint32_t write_domains, uint32_t valid_domains) +{ + struct nouveau_bo *nvbo = gem->driver_private; + struct ttm_buffer_object *bo = &nvbo->bo; + uint32_t domains = valid_domains & nvbo->valid_domains & + (write_domains ? write_domains : read_domains); + uint32_t pref_flags = 0, valid_flags = 0; + + if (!domains) + return -EINVAL; + + if (valid_domains & NOUVEAU_GEM_DOMAIN_VRAM) + valid_flags |= TTM_PL_FLAG_VRAM; + + if (valid_domains & NOUVEAU_GEM_DOMAIN_GART) + valid_flags |= TTM_PL_FLAG_TT; + + if ((domains & NOUVEAU_GEM_DOMAIN_VRAM) && + bo->mem.mem_type == TTM_PL_VRAM) + pref_flags |= TTM_PL_FLAG_VRAM; + + else if ((domains & NOUVEAU_GEM_DOMAIN_GART) && + bo->mem.mem_type == TTM_PL_TT) + pref_flags |= TTM_PL_FLAG_TT; + + else if (domains & NOUVEAU_GEM_DOMAIN_VRAM) + pref_flags |= TTM_PL_FLAG_VRAM; + + else + pref_flags |= TTM_PL_FLAG_TT; + + nouveau_bo_placement_set(nvbo, pref_flags, valid_flags); + + return 0; +} + +struct validate_op { + struct list_head vram_list; + struct list_head gart_list; + struct list_head both_list; +}; + +static void +validate_fini_list(struct list_head *list, struct nouveau_fence *fence) +{ + struct list_head *entry, *tmp; + struct nouveau_bo *nvbo; + + list_for_each_safe(entry, tmp, list) { + nvbo = list_entry(entry, struct nouveau_bo, entry); + + nouveau_bo_fence(nvbo, fence); + + if (unlikely(nvbo->validate_mapped)) { + ttm_bo_kunmap(&nvbo->kmap); + nvbo->validate_mapped = false; + } + + list_del(&nvbo->entry); + nvbo->reserved_by = NULL; + ttm_bo_unreserve(&nvbo->bo); + drm_gem_object_unreference_unlocked(nvbo->gem); + } +} + +static void +validate_fini(struct validate_op *op, struct nouveau_fence* fence) +{ + validate_fini_list(&op->vram_list, fence); + validate_fini_list(&op->gart_list, fence); + validate_fini_list(&op->both_list, fence); +} + +static int +validate_init(struct nouveau_channel *chan, struct drm_file *file_priv, + struct drm_nouveau_gem_pushbuf_bo *pbbo, + int nr_buffers, struct validate_op *op) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + uint32_t sequence; + int trycnt = 0; + int ret, i; + + sequence = atomic_add_return(1, &dev_priv->ttm.validate_sequence); +retry: + if (++trycnt > 100000) { + NV_ERROR(dev, "%s failed and gave up.\n", __func__); + return -EINVAL; + } + + for (i = 0; i < nr_buffers; i++) { + struct drm_nouveau_gem_pushbuf_bo *b = &pbbo[i]; + struct drm_gem_object *gem; + struct nouveau_bo *nvbo; + + gem = drm_gem_object_lookup(dev, file_priv, b->handle); + if (!gem) { + NV_ERROR(dev, "Unknown handle 0x%08x\n", b->handle); + validate_fini(op, NULL); + return -ENOENT; + } + nvbo = gem->driver_private; + + if (nvbo->reserved_by && nvbo->reserved_by == file_priv) { + NV_ERROR(dev, "multiple instances of buffer %d on " + "validation list\n", b->handle); + validate_fini(op, NULL); + return -EINVAL; + } + + ret = ttm_bo_reserve(&nvbo->bo, true, false, true, sequence); + if (ret) { + validate_fini(op, NULL); + if (unlikely(ret == -EAGAIN)) + ret = ttm_bo_wait_unreserved(&nvbo->bo, true); + drm_gem_object_unreference_unlocked(gem); + if (unlikely(ret)) { + if (ret != -ERESTARTSYS) + NV_ERROR(dev, "fail reserve\n"); + return ret; + } + goto retry; + } + + b->user_priv = (uint64_t)(unsigned long)nvbo; + nvbo->reserved_by = file_priv; + nvbo->pbbo_index = i; + if ((b->valid_domains & NOUVEAU_GEM_DOMAIN_VRAM) && + (b->valid_domains & NOUVEAU_GEM_DOMAIN_GART)) + list_add_tail(&nvbo->entry, &op->both_list); + else + if (b->valid_domains & NOUVEAU_GEM_DOMAIN_VRAM) + list_add_tail(&nvbo->entry, &op->vram_list); + else + if (b->valid_domains & NOUVEAU_GEM_DOMAIN_GART) + list_add_tail(&nvbo->entry, &op->gart_list); + else { + NV_ERROR(dev, "invalid valid domains: 0x%08x\n", + b->valid_domains); + list_add_tail(&nvbo->entry, &op->both_list); + validate_fini(op, NULL); + return -EINVAL; + } + } + + return 0; +} + +static int +validate_sync(struct nouveau_channel *chan, struct nouveau_bo *nvbo) +{ + struct nouveau_fence *fence = NULL; + int ret = 0; + + spin_lock(&nvbo->bo.bdev->fence_lock); + if (nvbo->bo.sync_obj) + fence = nouveau_fence_ref(nvbo->bo.sync_obj); + spin_unlock(&nvbo->bo.bdev->fence_lock); + + if (fence) { + ret = nouveau_fence_sync(fence, chan); + nouveau_fence_unref(&fence); + } + + return ret; +} + +static int +validate_list(struct nouveau_channel *chan, struct list_head *list, + struct drm_nouveau_gem_pushbuf_bo *pbbo, uint64_t user_pbbo_ptr) +{ + struct drm_nouveau_gem_pushbuf_bo __user *upbbo = + (void __force __user *)(uintptr_t)user_pbbo_ptr; + struct drm_device *dev = chan->dev; + struct nouveau_bo *nvbo; + int ret, relocs = 0; + + list_for_each_entry(nvbo, list, entry) { + struct drm_nouveau_gem_pushbuf_bo *b = &pbbo[nvbo->pbbo_index]; + + ret = validate_sync(chan, nvbo); + if (unlikely(ret)) { + NV_ERROR(dev, "fail pre-validate sync\n"); + return ret; + } + + ret = nouveau_gem_set_domain(nvbo->gem, b->read_domains, + b->write_domains, + b->valid_domains); + if (unlikely(ret)) { + NV_ERROR(dev, "fail set_domain\n"); + return ret; + } + + nvbo->channel = (b->read_domains & (1 << 31)) ? NULL : chan; + ret = nouveau_bo_validate(nvbo, true, false, false); + nvbo->channel = NULL; + if (unlikely(ret)) { + if (ret != -ERESTARTSYS) + NV_ERROR(dev, "fail ttm_validate\n"); + return ret; + } + + ret = validate_sync(chan, nvbo); + if (unlikely(ret)) { + NV_ERROR(dev, "fail post-validate sync\n"); + return ret; + } + + if (nvbo->bo.offset == b->presumed.offset && + ((nvbo->bo.mem.mem_type == TTM_PL_VRAM && + b->presumed.domain & NOUVEAU_GEM_DOMAIN_VRAM) || + (nvbo->bo.mem.mem_type == TTM_PL_TT && + b->presumed.domain & NOUVEAU_GEM_DOMAIN_GART))) + continue; + + if (nvbo->bo.mem.mem_type == TTM_PL_TT) + b->presumed.domain = NOUVEAU_GEM_DOMAIN_GART; + else + b->presumed.domain = NOUVEAU_GEM_DOMAIN_VRAM; + b->presumed.offset = nvbo->bo.offset; + b->presumed.valid = 0; + relocs++; + + if (DRM_COPY_TO_USER(&upbbo[nvbo->pbbo_index].presumed, + &b->presumed, sizeof(b->presumed))) + return -EFAULT; + } + + return relocs; +} + +static int +nouveau_gem_pushbuf_validate(struct nouveau_channel *chan, + struct drm_file *file_priv, + struct drm_nouveau_gem_pushbuf_bo *pbbo, + uint64_t user_buffers, int nr_buffers, + struct validate_op *op, int *apply_relocs) +{ + struct drm_device *dev = chan->dev; + int ret, relocs = 0; + + INIT_LIST_HEAD(&op->vram_list); + INIT_LIST_HEAD(&op->gart_list); + INIT_LIST_HEAD(&op->both_list); + + if (nr_buffers == 0) + return 0; + + ret = validate_init(chan, file_priv, pbbo, nr_buffers, op); + if (unlikely(ret)) { + if (ret != -ERESTARTSYS) + NV_ERROR(dev, "validate_init\n"); + return ret; + } + + ret = validate_list(chan, &op->vram_list, pbbo, user_buffers); + if (unlikely(ret < 0)) { + if (ret != -ERESTARTSYS) + NV_ERROR(dev, "validate vram_list\n"); + validate_fini(op, NULL); + return ret; + } + relocs += ret; + + ret = validate_list(chan, &op->gart_list, pbbo, user_buffers); + if (unlikely(ret < 0)) { + if (ret != -ERESTARTSYS) + NV_ERROR(dev, "validate gart_list\n"); + validate_fini(op, NULL); + return ret; + } + relocs += ret; + + ret = validate_list(chan, &op->both_list, pbbo, user_buffers); + if (unlikely(ret < 0)) { + if (ret != -ERESTARTSYS) + NV_ERROR(dev, "validate both_list\n"); + validate_fini(op, NULL); + return ret; + } + relocs += ret; + + *apply_relocs = relocs; + return 0; +} + +static inline void * +u_memcpya(uint64_t user, unsigned nmemb, unsigned size) +{ + void *mem; + void __user *userptr = (void __force __user *)(uintptr_t)user; + + mem = kmalloc(nmemb * size, GFP_KERNEL); + if (!mem) + return ERR_PTR(-ENOMEM); + + if (DRM_COPY_FROM_USER(mem, userptr, nmemb * size)) { + kfree(mem); + return ERR_PTR(-EFAULT); + } + + return mem; +} + +static int +nouveau_gem_pushbuf_reloc_apply(struct drm_device *dev, + struct drm_nouveau_gem_pushbuf *req, + struct drm_nouveau_gem_pushbuf_bo *bo) +{ + struct drm_nouveau_gem_pushbuf_reloc *reloc = NULL; + int ret = 0; + unsigned i; + + reloc = u_memcpya(req->relocs, req->nr_relocs, sizeof(*reloc)); + if (IS_ERR(reloc)) + return PTR_ERR(reloc); + + for (i = 0; i < req->nr_relocs; i++) { + struct drm_nouveau_gem_pushbuf_reloc *r = &reloc[i]; + struct drm_nouveau_gem_pushbuf_bo *b; + struct nouveau_bo *nvbo; + uint32_t data; + + if (unlikely(r->bo_index > req->nr_buffers)) { + NV_ERROR(dev, "reloc bo index invalid\n"); + ret = -EINVAL; + break; + } + + b = &bo[r->bo_index]; + if (b->presumed.valid) + continue; + + if (unlikely(r->reloc_bo_index > req->nr_buffers)) { + NV_ERROR(dev, "reloc container bo index invalid\n"); + ret = -EINVAL; + break; + } + nvbo = (void *)(unsigned long)bo[r->reloc_bo_index].user_priv; + + if (unlikely(r->reloc_bo_offset + 4 > + nvbo->bo.mem.num_pages << PAGE_SHIFT)) { + NV_ERROR(dev, "reloc outside of bo\n"); + ret = -EINVAL; + break; + } + + if (!nvbo->kmap.virtual) { + ret = ttm_bo_kmap(&nvbo->bo, 0, nvbo->bo.mem.num_pages, + &nvbo->kmap); + if (ret) { + NV_ERROR(dev, "failed kmap for reloc\n"); + break; + } + nvbo->validate_mapped = true; + } + + if (r->flags & NOUVEAU_GEM_RELOC_LOW) + data = b->presumed.offset + r->data; + else + if (r->flags & NOUVEAU_GEM_RELOC_HIGH) + data = (b->presumed.offset + r->data) >> 32; + else + data = r->data; + + if (r->flags & NOUVEAU_GEM_RELOC_OR) { + if (b->presumed.domain == NOUVEAU_GEM_DOMAIN_GART) + data |= r->tor; + else + data |= r->vor; + } + + spin_lock(&nvbo->bo.bdev->fence_lock); + ret = ttm_bo_wait(&nvbo->bo, false, false, false); + spin_unlock(&nvbo->bo.bdev->fence_lock); + if (ret) { + NV_ERROR(dev, "reloc wait_idle failed: %d\n", ret); + break; + } + + nouveau_bo_wr32(nvbo, r->reloc_bo_offset >> 2, data); + } + + kfree(reloc); + return ret; +} + +int +nouveau_gem_ioctl_pushbuf(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct drm_nouveau_gem_pushbuf *req = data; + struct drm_nouveau_gem_pushbuf_push *push; + struct drm_nouveau_gem_pushbuf_bo *bo; + struct nouveau_channel *chan; + struct validate_op op; + struct nouveau_fence *fence = NULL; + int i, j, ret = 0, do_reloc = 0; + + chan = nouveau_channel_get(dev, file_priv, req->channel); + if (IS_ERR(chan)) + return PTR_ERR(chan); + + req->vram_available = dev_priv->fb_aper_free; + req->gart_available = dev_priv->gart_info.aper_free; + if (unlikely(req->nr_push == 0)) + goto out_next; + + if (unlikely(req->nr_push > NOUVEAU_GEM_MAX_PUSH)) { + NV_ERROR(dev, "pushbuf push count exceeds limit: %d max %d\n", + req->nr_push, NOUVEAU_GEM_MAX_PUSH); + nouveau_channel_put(&chan); + return -EINVAL; + } + + if (unlikely(req->nr_buffers > NOUVEAU_GEM_MAX_BUFFERS)) { + NV_ERROR(dev, "pushbuf bo count exceeds limit: %d max %d\n", + req->nr_buffers, NOUVEAU_GEM_MAX_BUFFERS); + nouveau_channel_put(&chan); + return -EINVAL; + } + + if (unlikely(req->nr_relocs > NOUVEAU_GEM_MAX_RELOCS)) { + NV_ERROR(dev, "pushbuf reloc count exceeds limit: %d max %d\n", + req->nr_relocs, NOUVEAU_GEM_MAX_RELOCS); + nouveau_channel_put(&chan); + return -EINVAL; + } + + push = u_memcpya(req->push, req->nr_push, sizeof(*push)); + if (IS_ERR(push)) { + nouveau_channel_put(&chan); + return PTR_ERR(push); + } + + bo = u_memcpya(req->buffers, req->nr_buffers, sizeof(*bo)); + if (IS_ERR(bo)) { + kfree(push); + nouveau_channel_put(&chan); + return PTR_ERR(bo); + } + + /* Mark push buffers as being used on PFIFO, the validation code + * will then make sure that if the pushbuf bo moves, that they + * happen on the kernel channel, which will in turn cause a sync + * to happen before we try and submit the push buffer. + */ + for (i = 0; i < req->nr_push; i++) { + if (push[i].bo_index >= req->nr_buffers) { + NV_ERROR(dev, "push %d buffer not in list\n", i); + ret = -EINVAL; + goto out_prevalid; + } + + bo[push[i].bo_index].read_domains |= (1 << 31); + } + + /* Validate buffer list */ + ret = nouveau_gem_pushbuf_validate(chan, file_priv, bo, req->buffers, + req->nr_buffers, &op, &do_reloc); + if (ret) { + if (ret != -ERESTARTSYS) + NV_ERROR(dev, "validate: %d\n", ret); + goto out_prevalid; + } + + /* Apply any relocations that are required */ + if (do_reloc) { + ret = nouveau_gem_pushbuf_reloc_apply(dev, req, bo); + if (ret) { + NV_ERROR(dev, "reloc apply: %d\n", ret); + goto out; + } + } + + if (chan->dma.ib_max) { + ret = nouveau_dma_wait(chan, req->nr_push + 1, 6); + if (ret) { + NV_INFO(dev, "nv50cal_space: %d\n", ret); + goto out; + } + + for (i = 0; i < req->nr_push; i++) { + struct nouveau_bo *nvbo = (void *)(unsigned long) + bo[push[i].bo_index].user_priv; + + nv50_dma_push(chan, nvbo, push[i].offset, + push[i].length); + } + } else + if (dev_priv->chipset >= 0x25) { + ret = RING_SPACE(chan, req->nr_push * 2); + if (ret) { + NV_ERROR(dev, "cal_space: %d\n", ret); + goto out; + } + + for (i = 0; i < req->nr_push; i++) { + struct nouveau_bo *nvbo = (void *)(unsigned long) + bo[push[i].bo_index].user_priv; + struct drm_mm_node *mem = nvbo->bo.mem.mm_node; + + OUT_RING(chan, ((mem->start << PAGE_SHIFT) + + push[i].offset) | 2); + OUT_RING(chan, 0); + } + } else { + ret = RING_SPACE(chan, req->nr_push * (2 + NOUVEAU_DMA_SKIPS)); + if (ret) { + NV_ERROR(dev, "jmp_space: %d\n", ret); + goto out; + } + + for (i = 0; i < req->nr_push; i++) { + struct nouveau_bo *nvbo = (void *)(unsigned long) + bo[push[i].bo_index].user_priv; + struct drm_mm_node *mem = nvbo->bo.mem.mm_node; + uint32_t cmd; + + cmd = chan->pushbuf_base + ((chan->dma.cur + 2) << 2); + cmd |= 0x20000000; + if (unlikely(cmd != req->suffix0)) { + if (!nvbo->kmap.virtual) { + ret = ttm_bo_kmap(&nvbo->bo, 0, + nvbo->bo.mem. + num_pages, + &nvbo->kmap); + if (ret) { + WIND_RING(chan); + goto out; + } + nvbo->validate_mapped = true; + } + + nouveau_bo_wr32(nvbo, (push[i].offset + + push[i].length - 8) / 4, cmd); + } + + OUT_RING(chan, ((mem->start << PAGE_SHIFT) + + push[i].offset) | 0x20000000); + OUT_RING(chan, 0); + for (j = 0; j < NOUVEAU_DMA_SKIPS; j++) + OUT_RING(chan, 0); + } + } + + ret = nouveau_fence_new(chan, &fence, true); + if (ret) { + NV_ERROR(dev, "error fencing pushbuf: %d\n", ret); + WIND_RING(chan); + goto out; + } + +out: + validate_fini(&op, fence); + nouveau_fence_unref(&fence); + +out_prevalid: + kfree(bo); + kfree(push); + +out_next: + if (chan->dma.ib_max) { + req->suffix0 = 0x00000000; + req->suffix1 = 0x00000000; + } else + if (dev_priv->chipset >= 0x25) { + req->suffix0 = 0x00020000; + req->suffix1 = 0x00000000; + } else { + req->suffix0 = 0x20000000 | + (chan->pushbuf_base + ((chan->dma.cur + 2) << 2)); + req->suffix1 = 0x00000000; + } + + nouveau_channel_put(&chan); + return ret; +} + +static inline uint32_t +domain_to_ttm(struct nouveau_bo *nvbo, uint32_t domain) +{ + uint32_t flags = 0; + + if (domain & NOUVEAU_GEM_DOMAIN_VRAM) + flags |= TTM_PL_FLAG_VRAM; + if (domain & NOUVEAU_GEM_DOMAIN_GART) + flags |= TTM_PL_FLAG_TT; + + return flags; +} + +int +nouveau_gem_ioctl_cpu_prep(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_nouveau_gem_cpu_prep *req = data; + struct drm_gem_object *gem; + struct nouveau_bo *nvbo; + bool no_wait = !!(req->flags & NOUVEAU_GEM_CPU_PREP_NOWAIT); + int ret = -EINVAL; + + gem = drm_gem_object_lookup(dev, file_priv, req->handle); + if (!gem) + return -ENOENT; + nvbo = nouveau_gem_object(gem); + + spin_lock(&nvbo->bo.bdev->fence_lock); + ret = ttm_bo_wait(&nvbo->bo, true, true, no_wait); + spin_unlock(&nvbo->bo.bdev->fence_lock); + drm_gem_object_unreference_unlocked(gem); + return ret; +} + +int +nouveau_gem_ioctl_cpu_fini(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + return 0; +} + +int +nouveau_gem_ioctl_info(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_nouveau_gem_info *req = data; + struct drm_gem_object *gem; + int ret; + + gem = drm_gem_object_lookup(dev, file_priv, req->handle); + if (!gem) + return -ENOENT; + + ret = nouveau_gem_info(gem, req); + drm_gem_object_unreference_unlocked(gem); + return ret; +} + diff --git a/drivers/gpu/drm/nouveau/nouveau_grctx.h b/drivers/gpu/drm/nouveau/nouveau_grctx.h new file mode 100644 index 00000000..86c2e374 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_grctx.h @@ -0,0 +1,133 @@ +#ifndef __NOUVEAU_GRCTX_H__ +#define __NOUVEAU_GRCTX_H__ + +struct nouveau_grctx { + struct drm_device *dev; + + enum { + NOUVEAU_GRCTX_PROG, + NOUVEAU_GRCTX_VALS + } mode; + void *data; + + uint32_t ctxprog_max; + uint32_t ctxprog_len; + uint32_t ctxprog_reg; + int ctxprog_label[32]; + uint32_t ctxvals_pos; + uint32_t ctxvals_base; +}; + +#ifdef CP_CTX +static inline void +cp_out(struct nouveau_grctx *ctx, uint32_t inst) +{ + uint32_t *ctxprog = ctx->data; + + if (ctx->mode != NOUVEAU_GRCTX_PROG) + return; + + BUG_ON(ctx->ctxprog_len == ctx->ctxprog_max); + ctxprog[ctx->ctxprog_len++] = inst; +} + +static inline void +cp_lsr(struct nouveau_grctx *ctx, uint32_t val) +{ + cp_out(ctx, CP_LOAD_SR | val); +} + +static inline void +cp_ctx(struct nouveau_grctx *ctx, uint32_t reg, uint32_t length) +{ + ctx->ctxprog_reg = (reg - 0x00400000) >> 2; + + ctx->ctxvals_base = ctx->ctxvals_pos; + ctx->ctxvals_pos = ctx->ctxvals_base + length; + + if (length > (CP_CTX_COUNT >> CP_CTX_COUNT_SHIFT)) { + cp_lsr(ctx, length); + length = 0; + } + + cp_out(ctx, CP_CTX | (length << CP_CTX_COUNT_SHIFT) | ctx->ctxprog_reg); +} + +static inline void +cp_name(struct nouveau_grctx *ctx, int name) +{ + uint32_t *ctxprog = ctx->data; + int i; + + if (ctx->mode != NOUVEAU_GRCTX_PROG) + return; + + ctx->ctxprog_label[name] = ctx->ctxprog_len; + for (i = 0; i < ctx->ctxprog_len; i++) { + if ((ctxprog[i] & 0xfff00000) != 0xff400000) + continue; + if ((ctxprog[i] & CP_BRA_IP) != ((name) << CP_BRA_IP_SHIFT)) + continue; + ctxprog[i] = (ctxprog[i] & 0x00ff00ff) | + (ctx->ctxprog_len << CP_BRA_IP_SHIFT); + } +} + +static inline void +_cp_bra(struct nouveau_grctx *ctx, u32 mod, int flag, int state, int name) +{ + int ip = 0; + + if (mod != 2) { + ip = ctx->ctxprog_label[name] << CP_BRA_IP_SHIFT; + if (ip == 0) + ip = 0xff000000 | (name << CP_BRA_IP_SHIFT); + } + + cp_out(ctx, CP_BRA | (mod << 18) | ip | flag | + (state ? 0 : CP_BRA_IF_CLEAR)); +} +#define cp_bra(c, f, s, n) _cp_bra((c), 0, CP_FLAG_##f, CP_FLAG_##f##_##s, n) +#ifdef CP_BRA_MOD +#define cp_cal(c, f, s, n) _cp_bra((c), 1, CP_FLAG_##f, CP_FLAG_##f##_##s, n) +#define cp_ret(c, f, s) _cp_bra((c), 2, CP_FLAG_##f, CP_FLAG_##f##_##s, 0) +#endif + +static inline void +_cp_wait(struct nouveau_grctx *ctx, int flag, int state) +{ + cp_out(ctx, CP_WAIT | flag | (state ? CP_WAIT_SET : 0)); +} +#define cp_wait(c, f, s) _cp_wait((c), CP_FLAG_##f, CP_FLAG_##f##_##s) + +static inline void +_cp_set(struct nouveau_grctx *ctx, int flag, int state) +{ + cp_out(ctx, CP_SET | flag | (state ? CP_SET_1 : 0)); +} +#define cp_set(c, f, s) _cp_set((c), CP_FLAG_##f, CP_FLAG_##f##_##s) + +static inline void +cp_pos(struct nouveau_grctx *ctx, int offset) +{ + ctx->ctxvals_pos = offset; + ctx->ctxvals_base = ctx->ctxvals_pos; + + cp_lsr(ctx, ctx->ctxvals_pos); + cp_out(ctx, CP_SET_CONTEXT_POINTER); +} + +static inline void +gr_def(struct nouveau_grctx *ctx, uint32_t reg, uint32_t val) +{ + if (ctx->mode != NOUVEAU_GRCTX_VALS) + return; + + reg = (reg - 0x00400000) / 4; + reg = (reg - ctx->ctxprog_reg) + ctx->ctxvals_base; + + nv_wo32(ctx->data, reg * 4, val); +} +#endif + +#endif diff --git a/drivers/gpu/drm/nouveau/nouveau_hw.c b/drivers/gpu/drm/nouveau/nouveau_hw.c new file mode 100644 index 00000000..ba896e54 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_hw.c @@ -0,0 +1,1087 @@ +/* + * Copyright 2006 Dave Airlie + * Copyright 2007 Maarten Maathuis + * Copyright 2007-2009 Stuart Bennett + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#include "drmP.h" +#include "nouveau_drv.h" +#include "nouveau_hw.h" + +#define CHIPSET_NFORCE 0x01a0 +#define CHIPSET_NFORCE2 0x01f0 + +/* + * misc hw access wrappers/control functions + */ + +void +NVWriteVgaSeq(struct drm_device *dev, int head, uint8_t index, uint8_t value) +{ + NVWritePRMVIO(dev, head, NV_PRMVIO_SRX, index); + NVWritePRMVIO(dev, head, NV_PRMVIO_SR, value); +} + +uint8_t +NVReadVgaSeq(struct drm_device *dev, int head, uint8_t index) +{ + NVWritePRMVIO(dev, head, NV_PRMVIO_SRX, index); + return NVReadPRMVIO(dev, head, NV_PRMVIO_SR); +} + +void +NVWriteVgaGr(struct drm_device *dev, int head, uint8_t index, uint8_t value) +{ + NVWritePRMVIO(dev, head, NV_PRMVIO_GRX, index); + NVWritePRMVIO(dev, head, NV_PRMVIO_GX, value); +} + +uint8_t +NVReadVgaGr(struct drm_device *dev, int head, uint8_t index) +{ + NVWritePRMVIO(dev, head, NV_PRMVIO_GRX, index); + return NVReadPRMVIO(dev, head, NV_PRMVIO_GX); +} + +/* CR44 takes values 0 (head A), 3 (head B) and 4 (heads tied) + * it affects only the 8 bit vga io regs, which we access using mmio at + * 0xc{0,2}3c*, 0x60{1,3}3*, and 0x68{1,3}3d* + * in general, the set value of cr44 does not matter: reg access works as + * expected and values can be set for the appropriate head by using a 0x2000 + * offset as required + * however: + * a) pre nv40, the head B range of PRMVIO regs at 0xc23c* was not exposed and + * cr44 must be set to 0 or 3 for accessing values on the correct head + * through the common 0xc03c* addresses + * b) in tied mode (4) head B is programmed to the values set on head A, and + * access using the head B addresses can have strange results, ergo we leave + * tied mode in init once we know to what cr44 should be restored on exit + * + * the owner parameter is slightly abused: + * 0 and 1 are treated as head values and so the set value is (owner * 3) + * other values are treated as literal values to set + */ +void +NVSetOwner(struct drm_device *dev, int owner) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + if (owner == 1) + owner *= 3; + + if (dev_priv->chipset == 0x11) { + /* This might seem stupid, but the blob does it and + * omitting it often locks the system up. + */ + NVReadVgaCrtc(dev, 0, NV_CIO_SR_LOCK_INDEX); + NVReadVgaCrtc(dev, 1, NV_CIO_SR_LOCK_INDEX); + } + + /* CR44 is always changed on CRTC0 */ + NVWriteVgaCrtc(dev, 0, NV_CIO_CRE_44, owner); + + if (dev_priv->chipset == 0x11) { /* set me harder */ + NVWriteVgaCrtc(dev, 0, NV_CIO_CRE_2E, owner); + NVWriteVgaCrtc(dev, 0, NV_CIO_CRE_2E, owner); + } +} + +void +NVBlankScreen(struct drm_device *dev, int head, bool blank) +{ + unsigned char seq1; + + if (nv_two_heads(dev)) + NVSetOwner(dev, head); + + seq1 = NVReadVgaSeq(dev, head, NV_VIO_SR_CLOCK_INDEX); + + NVVgaSeqReset(dev, head, true); + if (blank) + NVWriteVgaSeq(dev, head, NV_VIO_SR_CLOCK_INDEX, seq1 | 0x20); + else + NVWriteVgaSeq(dev, head, NV_VIO_SR_CLOCK_INDEX, seq1 & ~0x20); + NVVgaSeqReset(dev, head, false); +} + +/* + * PLL setting + */ + +static int +powerctrl_1_shift(int chip_version, int reg) +{ + int shift = -4; + + if (chip_version < 0x17 || chip_version == 0x1a || chip_version == 0x20) + return shift; + + switch (reg) { + case NV_RAMDAC_VPLL2: + shift += 4; + case NV_PRAMDAC_VPLL_COEFF: + shift += 4; + case NV_PRAMDAC_MPLL_COEFF: + shift += 4; + case NV_PRAMDAC_NVPLL_COEFF: + shift += 4; + } + + /* + * the shift for vpll regs is only used for nv3x chips with a single + * stage pll + */ + if (shift > 4 && (chip_version < 0x32 || chip_version == 0x35 || + chip_version == 0x36 || chip_version >= 0x40)) + shift = -4; + + return shift; +} + +static void +setPLL_single(struct drm_device *dev, uint32_t reg, struct nouveau_pll_vals *pv) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + int chip_version = dev_priv->vbios.chip_version; + uint32_t oldpll = NVReadRAMDAC(dev, 0, reg); + int oldN = (oldpll >> 8) & 0xff, oldM = oldpll & 0xff; + uint32_t pll = (oldpll & 0xfff80000) | pv->log2P << 16 | pv->NM1; + uint32_t saved_powerctrl_1 = 0; + int shift_powerctrl_1 = powerctrl_1_shift(chip_version, reg); + + if (oldpll == pll) + return; /* already set */ + + if (shift_powerctrl_1 >= 0) { + saved_powerctrl_1 = nvReadMC(dev, NV_PBUS_POWERCTRL_1); + nvWriteMC(dev, NV_PBUS_POWERCTRL_1, + (saved_powerctrl_1 & ~(0xf << shift_powerctrl_1)) | + 1 << shift_powerctrl_1); + } + + if (oldM && pv->M1 && (oldN / oldM < pv->N1 / pv->M1)) + /* upclock -- write new post divider first */ + NVWriteRAMDAC(dev, 0, reg, pv->log2P << 16 | (oldpll & 0xffff)); + else + /* downclock -- write new NM first */ + NVWriteRAMDAC(dev, 0, reg, (oldpll & 0xffff0000) | pv->NM1); + + if (chip_version < 0x17 && chip_version != 0x11) + /* wait a bit on older chips */ + msleep(64); + NVReadRAMDAC(dev, 0, reg); + + /* then write the other half as well */ + NVWriteRAMDAC(dev, 0, reg, pll); + + if (shift_powerctrl_1 >= 0) + nvWriteMC(dev, NV_PBUS_POWERCTRL_1, saved_powerctrl_1); +} + +static uint32_t +new_ramdac580(uint32_t reg1, bool ss, uint32_t ramdac580) +{ + bool head_a = (reg1 == NV_PRAMDAC_VPLL_COEFF); + + if (ss) /* single stage pll mode */ + ramdac580 |= head_a ? NV_RAMDAC_580_VPLL1_ACTIVE : + NV_RAMDAC_580_VPLL2_ACTIVE; + else + ramdac580 &= head_a ? ~NV_RAMDAC_580_VPLL1_ACTIVE : + ~NV_RAMDAC_580_VPLL2_ACTIVE; + + return ramdac580; +} + +static void +setPLL_double_highregs(struct drm_device *dev, uint32_t reg1, + struct nouveau_pll_vals *pv) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + int chip_version = dev_priv->vbios.chip_version; + bool nv3035 = chip_version == 0x30 || chip_version == 0x35; + uint32_t reg2 = reg1 + ((reg1 == NV_RAMDAC_VPLL2) ? 0x5c : 0x70); + uint32_t oldpll1 = NVReadRAMDAC(dev, 0, reg1); + uint32_t oldpll2 = !nv3035 ? NVReadRAMDAC(dev, 0, reg2) : 0; + uint32_t pll1 = (oldpll1 & 0xfff80000) | pv->log2P << 16 | pv->NM1; + uint32_t pll2 = (oldpll2 & 0x7fff0000) | 1 << 31 | pv->NM2; + uint32_t oldramdac580 = 0, ramdac580 = 0; + bool single_stage = !pv->NM2 || pv->N2 == pv->M2; /* nv41+ only */ + uint32_t saved_powerctrl_1 = 0, savedc040 = 0; + int shift_powerctrl_1 = powerctrl_1_shift(chip_version, reg1); + + /* model specific additions to generic pll1 and pll2 set up above */ + if (nv3035) { + pll1 = (pll1 & 0xfcc7ffff) | (pv->N2 & 0x18) << 21 | + (pv->N2 & 0x7) << 19 | 8 << 4 | (pv->M2 & 7) << 4; + pll2 = 0; + } + if (chip_version > 0x40 && reg1 >= NV_PRAMDAC_VPLL_COEFF) { /* !nv40 */ + oldramdac580 = NVReadRAMDAC(dev, 0, NV_PRAMDAC_580); + ramdac580 = new_ramdac580(reg1, single_stage, oldramdac580); + if (oldramdac580 != ramdac580) + oldpll1 = ~0; /* force mismatch */ + if (single_stage) + /* magic value used by nvidia in single stage mode */ + pll2 |= 0x011f; + } + if (chip_version > 0x70) + /* magic bits set by the blob (but not the bios) on g71-73 */ + pll1 = (pll1 & 0x7fffffff) | (single_stage ? 0x4 : 0xc) << 28; + + if (oldpll1 == pll1 && oldpll2 == pll2) + return; /* already set */ + + if (shift_powerctrl_1 >= 0) { + saved_powerctrl_1 = nvReadMC(dev, NV_PBUS_POWERCTRL_1); + nvWriteMC(dev, NV_PBUS_POWERCTRL_1, + (saved_powerctrl_1 & ~(0xf << shift_powerctrl_1)) | + 1 << shift_powerctrl_1); + } + + if (chip_version >= 0x40) { + int shift_c040 = 14; + + switch (reg1) { + case NV_PRAMDAC_MPLL_COEFF: + shift_c040 += 2; + case NV_PRAMDAC_NVPLL_COEFF: + shift_c040 += 2; + case NV_RAMDAC_VPLL2: + shift_c040 += 2; + case NV_PRAMDAC_VPLL_COEFF: + shift_c040 += 2; + } + + savedc040 = nvReadMC(dev, 0xc040); + if (shift_c040 != 14) + nvWriteMC(dev, 0xc040, savedc040 & ~(3 << shift_c040)); + } + + if (oldramdac580 != ramdac580) + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_580, ramdac580); + + if (!nv3035) + NVWriteRAMDAC(dev, 0, reg2, pll2); + NVWriteRAMDAC(dev, 0, reg1, pll1); + + if (shift_powerctrl_1 >= 0) + nvWriteMC(dev, NV_PBUS_POWERCTRL_1, saved_powerctrl_1); + if (chip_version >= 0x40) + nvWriteMC(dev, 0xc040, savedc040); +} + +static void +setPLL_double_lowregs(struct drm_device *dev, uint32_t NMNMreg, + struct nouveau_pll_vals *pv) +{ + /* When setting PLLs, there is a merry game of disabling and enabling + * various bits of hardware during the process. This function is a + * synthesis of six nv4x traces, nearly each card doing a subtly + * different thing. With luck all the necessary bits for each card are + * combined herein. Without luck it deviates from each card's formula + * so as to not work on any :) + */ + + uint32_t Preg = NMNMreg - 4; + bool mpll = Preg == 0x4020; + uint32_t oldPval = nvReadMC(dev, Preg); + uint32_t NMNM = pv->NM2 << 16 | pv->NM1; + uint32_t Pval = (oldPval & (mpll ? ~(0x77 << 16) : ~(7 << 16))) | + 0xc << 28 | pv->log2P << 16; + uint32_t saved4600 = 0; + /* some cards have different maskc040s */ + uint32_t maskc040 = ~(3 << 14), savedc040; + bool single_stage = !pv->NM2 || pv->N2 == pv->M2; + + if (nvReadMC(dev, NMNMreg) == NMNM && (oldPval & 0xc0070000) == Pval) + return; + + if (Preg == 0x4000) + maskc040 = ~0x333; + if (Preg == 0x4058) + maskc040 = ~(0xc << 24); + + if (mpll) { + struct pll_lims pll_lim; + uint8_t Pval2; + + if (get_pll_limits(dev, Preg, &pll_lim)) + return; + + Pval2 = pv->log2P + pll_lim.log2p_bias; + if (Pval2 > pll_lim.max_log2p) + Pval2 = pll_lim.max_log2p; + Pval |= 1 << 28 | Pval2 << 20; + + saved4600 = nvReadMC(dev, 0x4600); + nvWriteMC(dev, 0x4600, saved4600 | 8 << 28); + } + if (single_stage) + Pval |= mpll ? 1 << 12 : 1 << 8; + + nvWriteMC(dev, Preg, oldPval | 1 << 28); + nvWriteMC(dev, Preg, Pval & ~(4 << 28)); + if (mpll) { + Pval |= 8 << 20; + nvWriteMC(dev, 0x4020, Pval & ~(0xc << 28)); + nvWriteMC(dev, 0x4038, Pval & ~(0xc << 28)); + } + + savedc040 = nvReadMC(dev, 0xc040); + nvWriteMC(dev, 0xc040, savedc040 & maskc040); + + nvWriteMC(dev, NMNMreg, NMNM); + if (NMNMreg == 0x4024) + nvWriteMC(dev, 0x403c, NMNM); + + nvWriteMC(dev, Preg, Pval); + if (mpll) { + Pval &= ~(8 << 20); + nvWriteMC(dev, 0x4020, Pval); + nvWriteMC(dev, 0x4038, Pval); + nvWriteMC(dev, 0x4600, saved4600); + } + + nvWriteMC(dev, 0xc040, savedc040); + + if (mpll) { + nvWriteMC(dev, 0x4020, Pval & ~(1 << 28)); + nvWriteMC(dev, 0x4038, Pval & ~(1 << 28)); + } +} + +void +nouveau_hw_setpll(struct drm_device *dev, uint32_t reg1, + struct nouveau_pll_vals *pv) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + int cv = dev_priv->vbios.chip_version; + + if (cv == 0x30 || cv == 0x31 || cv == 0x35 || cv == 0x36 || + cv >= 0x40) { + if (reg1 > 0x405c) + setPLL_double_highregs(dev, reg1, pv); + else + setPLL_double_lowregs(dev, reg1, pv); + } else + setPLL_single(dev, reg1, pv); +} + +/* + * PLL getting + */ + +static void +nouveau_hw_decode_pll(struct drm_device *dev, uint32_t reg1, uint32_t pll1, + uint32_t pll2, struct nouveau_pll_vals *pllvals) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + /* to force parsing as single stage (i.e. nv40 vplls) pass pll2 as 0 */ + + /* log2P is & 0x7 as never more than 7, and nv30/35 only uses 3 bits */ + pllvals->log2P = (pll1 >> 16) & 0x7; + pllvals->N2 = pllvals->M2 = 1; + + if (reg1 <= 0x405c) { + pllvals->NM1 = pll2 & 0xffff; + /* single stage NVPLL and VPLLs use 1 << 8, MPLL uses 1 << 12 */ + if (!(pll1 & 0x1100)) + pllvals->NM2 = pll2 >> 16; + } else { + pllvals->NM1 = pll1 & 0xffff; + if (nv_two_reg_pll(dev) && pll2 & NV31_RAMDAC_ENABLE_VCO2) + pllvals->NM2 = pll2 & 0xffff; + else if (dev_priv->chipset == 0x30 || dev_priv->chipset == 0x35) { + pllvals->M1 &= 0xf; /* only 4 bits */ + if (pll1 & NV30_RAMDAC_ENABLE_VCO2) { + pllvals->M2 = (pll1 >> 4) & 0x7; + pllvals->N2 = ((pll1 >> 21) & 0x18) | + ((pll1 >> 19) & 0x7); + } + } + } +} + +int +nouveau_hw_get_pllvals(struct drm_device *dev, enum pll_types plltype, + struct nouveau_pll_vals *pllvals) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + uint32_t reg1 = get_pll_register(dev, plltype), pll1, pll2 = 0; + struct pll_lims pll_lim; + int ret; + + if (reg1 == 0) + return -ENOENT; + + pll1 = nvReadMC(dev, reg1); + + if (reg1 <= 0x405c) + pll2 = nvReadMC(dev, reg1 + 4); + else if (nv_two_reg_pll(dev)) { + uint32_t reg2 = reg1 + (reg1 == NV_RAMDAC_VPLL2 ? 0x5c : 0x70); + + pll2 = nvReadMC(dev, reg2); + } + + if (dev_priv->card_type == 0x40 && reg1 >= NV_PRAMDAC_VPLL_COEFF) { + uint32_t ramdac580 = NVReadRAMDAC(dev, 0, NV_PRAMDAC_580); + + /* check whether vpll has been forced into single stage mode */ + if (reg1 == NV_PRAMDAC_VPLL_COEFF) { + if (ramdac580 & NV_RAMDAC_580_VPLL1_ACTIVE) + pll2 = 0; + } else + if (ramdac580 & NV_RAMDAC_580_VPLL2_ACTIVE) + pll2 = 0; + } + + nouveau_hw_decode_pll(dev, reg1, pll1, pll2, pllvals); + + ret = get_pll_limits(dev, plltype, &pll_lim); + if (ret) + return ret; + + pllvals->refclk = pll_lim.refclk; + + return 0; +} + +int +nouveau_hw_pllvals_to_clk(struct nouveau_pll_vals *pv) +{ + /* Avoid divide by zero if called at an inappropriate time */ + if (!pv->M1 || !pv->M2) + return 0; + + return pv->N1 * pv->N2 * pv->refclk / (pv->M1 * pv->M2) >> pv->log2P; +} + +int +nouveau_hw_get_clock(struct drm_device *dev, enum pll_types plltype) +{ + struct nouveau_pll_vals pllvals; + int ret; + + if (plltype == PLL_MEMORY && + (dev->pci_device & 0x0ff0) == CHIPSET_NFORCE) { + uint32_t mpllP; + + pci_read_config_dword(pci_get_bus_and_slot(0, 3), 0x6c, &mpllP); + if (!mpllP) + mpllP = 4; + + return 400000 / mpllP; + } else + if (plltype == PLL_MEMORY && + (dev->pci_device & 0xff0) == CHIPSET_NFORCE2) { + uint32_t clock; + + pci_read_config_dword(pci_get_bus_and_slot(0, 5), 0x4c, &clock); + return clock; + } + + ret = nouveau_hw_get_pllvals(dev, plltype, &pllvals); + if (ret) + return ret; + + return nouveau_hw_pllvals_to_clk(&pllvals); +} + +static void +nouveau_hw_fix_bad_vpll(struct drm_device *dev, int head) +{ + /* the vpll on an unused head can come up with a random value, way + * beyond the pll limits. for some reason this causes the chip to + * lock up when reading the dac palette regs, so set a valid pll here + * when such a condition detected. only seen on nv11 to date + */ + + struct pll_lims pll_lim; + struct nouveau_pll_vals pv; + enum pll_types pll = head ? PLL_VPLL1 : PLL_VPLL0; + + if (get_pll_limits(dev, pll, &pll_lim)) + return; + nouveau_hw_get_pllvals(dev, pll, &pv); + + if (pv.M1 >= pll_lim.vco1.min_m && pv.M1 <= pll_lim.vco1.max_m && + pv.N1 >= pll_lim.vco1.min_n && pv.N1 <= pll_lim.vco1.max_n && + pv.log2P <= pll_lim.max_log2p) + return; + + NV_WARN(dev, "VPLL %d outwith limits, attempting to fix\n", head + 1); + + /* set lowest clock within static limits */ + pv.M1 = pll_lim.vco1.max_m; + pv.N1 = pll_lim.vco1.min_n; + pv.log2P = pll_lim.max_usable_log2p; + nouveau_hw_setpll(dev, pll_lim.reg, &pv); +} + +/* + * vga font save/restore + */ + +static void nouveau_vga_font_io(struct drm_device *dev, + void __iomem *iovram, + bool save, unsigned plane) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + unsigned i; + + NVWriteVgaSeq(dev, 0, NV_VIO_SR_PLANE_MASK_INDEX, 1 << plane); + NVWriteVgaGr(dev, 0, NV_VIO_GX_READ_MAP_INDEX, plane); + for (i = 0; i < 16384; i++) { + if (save) { + dev_priv->saved_vga_font[plane][i] = + ioread32_native(iovram + i * 4); + } else { + iowrite32_native(dev_priv->saved_vga_font[plane][i], + iovram + i * 4); + } + } +} + +void +nouveau_hw_save_vga_fonts(struct drm_device *dev, bool save) +{ + uint8_t misc, gr4, gr5, gr6, seq2, seq4; + bool graphicsmode; + unsigned plane; + void __iomem *iovram; + + if (nv_two_heads(dev)) + NVSetOwner(dev, 0); + + NVSetEnablePalette(dev, 0, true); + graphicsmode = NVReadVgaAttr(dev, 0, NV_CIO_AR_MODE_INDEX) & 1; + NVSetEnablePalette(dev, 0, false); + + if (graphicsmode) /* graphics mode => framebuffer => no need to save */ + return; + + NV_INFO(dev, "%sing VGA fonts\n", save ? "Sav" : "Restor"); + + /* map first 64KiB of VRAM, holds VGA fonts etc */ + iovram = ioremap(pci_resource_start(dev->pdev, 1), 65536); + if (!iovram) { + NV_ERROR(dev, "Failed to map VRAM, " + "cannot save/restore VGA fonts.\n"); + return; + } + + if (nv_two_heads(dev)) + NVBlankScreen(dev, 1, true); + NVBlankScreen(dev, 0, true); + + /* save control regs */ + misc = NVReadPRMVIO(dev, 0, NV_PRMVIO_MISC__READ); + seq2 = NVReadVgaSeq(dev, 0, NV_VIO_SR_PLANE_MASK_INDEX); + seq4 = NVReadVgaSeq(dev, 0, NV_VIO_SR_MEM_MODE_INDEX); + gr4 = NVReadVgaGr(dev, 0, NV_VIO_GX_READ_MAP_INDEX); + gr5 = NVReadVgaGr(dev, 0, NV_VIO_GX_MODE_INDEX); + gr6 = NVReadVgaGr(dev, 0, NV_VIO_GX_MISC_INDEX); + + NVWritePRMVIO(dev, 0, NV_PRMVIO_MISC__WRITE, 0x67); + NVWriteVgaSeq(dev, 0, NV_VIO_SR_MEM_MODE_INDEX, 0x6); + NVWriteVgaGr(dev, 0, NV_VIO_GX_MODE_INDEX, 0x0); + NVWriteVgaGr(dev, 0, NV_VIO_GX_MISC_INDEX, 0x5); + + /* store font in planes 0..3 */ + for (plane = 0; plane < 4; plane++) + nouveau_vga_font_io(dev, iovram, save, plane); + + /* restore control regs */ + NVWritePRMVIO(dev, 0, NV_PRMVIO_MISC__WRITE, misc); + NVWriteVgaGr(dev, 0, NV_VIO_GX_READ_MAP_INDEX, gr4); + NVWriteVgaGr(dev, 0, NV_VIO_GX_MODE_INDEX, gr5); + NVWriteVgaGr(dev, 0, NV_VIO_GX_MISC_INDEX, gr6); + NVWriteVgaSeq(dev, 0, NV_VIO_SR_PLANE_MASK_INDEX, seq2); + NVWriteVgaSeq(dev, 0, NV_VIO_SR_MEM_MODE_INDEX, seq4); + + if (nv_two_heads(dev)) + NVBlankScreen(dev, 1, false); + NVBlankScreen(dev, 0, false); + + iounmap(iovram); +} + +/* + * mode state save/load + */ + +static void +rd_cio_state(struct drm_device *dev, int head, + struct nv04_crtc_reg *crtcstate, int index) +{ + crtcstate->CRTC[index] = NVReadVgaCrtc(dev, head, index); +} + +static void +wr_cio_state(struct drm_device *dev, int head, + struct nv04_crtc_reg *crtcstate, int index) +{ + NVWriteVgaCrtc(dev, head, index, crtcstate->CRTC[index]); +} + +static void +nv_save_state_ramdac(struct drm_device *dev, int head, + struct nv04_mode_state *state) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nv04_crtc_reg *regp = &state->crtc_reg[head]; + int i; + + if (dev_priv->card_type >= NV_10) + regp->nv10_cursync = NVReadRAMDAC(dev, head, NV_RAMDAC_NV10_CURSYNC); + + nouveau_hw_get_pllvals(dev, head ? PLL_VPLL1 : PLL_VPLL0, ®p->pllvals); + state->pllsel = NVReadRAMDAC(dev, 0, NV_PRAMDAC_PLL_COEFF_SELECT); + if (nv_two_heads(dev)) + state->sel_clk = NVReadRAMDAC(dev, 0, NV_PRAMDAC_SEL_CLK); + if (dev_priv->chipset == 0x11) + regp->dither = NVReadRAMDAC(dev, head, NV_RAMDAC_DITHER_NV11); + + regp->ramdac_gen_ctrl = NVReadRAMDAC(dev, head, NV_PRAMDAC_GENERAL_CONTROL); + + if (nv_gf4_disp_arch(dev)) + regp->ramdac_630 = NVReadRAMDAC(dev, head, NV_PRAMDAC_630); + if (dev_priv->chipset >= 0x30) + regp->ramdac_634 = NVReadRAMDAC(dev, head, NV_PRAMDAC_634); + + regp->tv_setup = NVReadRAMDAC(dev, head, NV_PRAMDAC_TV_SETUP); + regp->tv_vtotal = NVReadRAMDAC(dev, head, NV_PRAMDAC_TV_VTOTAL); + regp->tv_vskew = NVReadRAMDAC(dev, head, NV_PRAMDAC_TV_VSKEW); + regp->tv_vsync_delay = NVReadRAMDAC(dev, head, NV_PRAMDAC_TV_VSYNC_DELAY); + regp->tv_htotal = NVReadRAMDAC(dev, head, NV_PRAMDAC_TV_HTOTAL); + regp->tv_hskew = NVReadRAMDAC(dev, head, NV_PRAMDAC_TV_HSKEW); + regp->tv_hsync_delay = NVReadRAMDAC(dev, head, NV_PRAMDAC_TV_HSYNC_DELAY); + regp->tv_hsync_delay2 = NVReadRAMDAC(dev, head, NV_PRAMDAC_TV_HSYNC_DELAY2); + + for (i = 0; i < 7; i++) { + uint32_t ramdac_reg = NV_PRAMDAC_FP_VDISPLAY_END + (i * 4); + regp->fp_vert_regs[i] = NVReadRAMDAC(dev, head, ramdac_reg); + regp->fp_horiz_regs[i] = NVReadRAMDAC(dev, head, ramdac_reg + 0x20); + } + + if (nv_gf4_disp_arch(dev)) { + regp->dither = NVReadRAMDAC(dev, head, NV_RAMDAC_FP_DITHER); + for (i = 0; i < 3; i++) { + regp->dither_regs[i] = NVReadRAMDAC(dev, head, NV_PRAMDAC_850 + i * 4); + regp->dither_regs[i + 3] = NVReadRAMDAC(dev, head, NV_PRAMDAC_85C + i * 4); + } + } + + regp->fp_control = NVReadRAMDAC(dev, head, NV_PRAMDAC_FP_TG_CONTROL); + regp->fp_debug_0 = NVReadRAMDAC(dev, head, NV_PRAMDAC_FP_DEBUG_0); + if (!nv_gf4_disp_arch(dev) && head == 0) { + /* early chips don't allow access to PRAMDAC_TMDS_* without + * the head A FPCLK on (nv11 even locks up) */ + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_FP_DEBUG_0, regp->fp_debug_0 & + ~NV_PRAMDAC_FP_DEBUG_0_PWRDOWN_FPCLK); + } + regp->fp_debug_1 = NVReadRAMDAC(dev, head, NV_PRAMDAC_FP_DEBUG_1); + regp->fp_debug_2 = NVReadRAMDAC(dev, head, NV_PRAMDAC_FP_DEBUG_2); + + regp->fp_margin_color = NVReadRAMDAC(dev, head, NV_PRAMDAC_FP_MARGIN_COLOR); + + if (nv_gf4_disp_arch(dev)) + regp->ramdac_8c0 = NVReadRAMDAC(dev, head, NV_PRAMDAC_8C0); + + if (dev_priv->card_type == NV_40) { + regp->ramdac_a20 = NVReadRAMDAC(dev, head, NV_PRAMDAC_A20); + regp->ramdac_a24 = NVReadRAMDAC(dev, head, NV_PRAMDAC_A24); + regp->ramdac_a34 = NVReadRAMDAC(dev, head, NV_PRAMDAC_A34); + + for (i = 0; i < 38; i++) + regp->ctv_regs[i] = NVReadRAMDAC(dev, head, + NV_PRAMDAC_CTV + 4*i); + } +} + +static void +nv_load_state_ramdac(struct drm_device *dev, int head, + struct nv04_mode_state *state) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nv04_crtc_reg *regp = &state->crtc_reg[head]; + uint32_t pllreg = head ? NV_RAMDAC_VPLL2 : NV_PRAMDAC_VPLL_COEFF; + int i; + + if (dev_priv->card_type >= NV_10) + NVWriteRAMDAC(dev, head, NV_RAMDAC_NV10_CURSYNC, regp->nv10_cursync); + + nouveau_hw_setpll(dev, pllreg, ®p->pllvals); + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_PLL_COEFF_SELECT, state->pllsel); + if (nv_two_heads(dev)) + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_SEL_CLK, state->sel_clk); + if (dev_priv->chipset == 0x11) + NVWriteRAMDAC(dev, head, NV_RAMDAC_DITHER_NV11, regp->dither); + + NVWriteRAMDAC(dev, head, NV_PRAMDAC_GENERAL_CONTROL, regp->ramdac_gen_ctrl); + + if (nv_gf4_disp_arch(dev)) + NVWriteRAMDAC(dev, head, NV_PRAMDAC_630, regp->ramdac_630); + if (dev_priv->chipset >= 0x30) + NVWriteRAMDAC(dev, head, NV_PRAMDAC_634, regp->ramdac_634); + + NVWriteRAMDAC(dev, head, NV_PRAMDAC_TV_SETUP, regp->tv_setup); + NVWriteRAMDAC(dev, head, NV_PRAMDAC_TV_VTOTAL, regp->tv_vtotal); + NVWriteRAMDAC(dev, head, NV_PRAMDAC_TV_VSKEW, regp->tv_vskew); + NVWriteRAMDAC(dev, head, NV_PRAMDAC_TV_VSYNC_DELAY, regp->tv_vsync_delay); + NVWriteRAMDAC(dev, head, NV_PRAMDAC_TV_HTOTAL, regp->tv_htotal); + NVWriteRAMDAC(dev, head, NV_PRAMDAC_TV_HSKEW, regp->tv_hskew); + NVWriteRAMDAC(dev, head, NV_PRAMDAC_TV_HSYNC_DELAY, regp->tv_hsync_delay); + NVWriteRAMDAC(dev, head, NV_PRAMDAC_TV_HSYNC_DELAY2, regp->tv_hsync_delay2); + + for (i = 0; i < 7; i++) { + uint32_t ramdac_reg = NV_PRAMDAC_FP_VDISPLAY_END + (i * 4); + + NVWriteRAMDAC(dev, head, ramdac_reg, regp->fp_vert_regs[i]); + NVWriteRAMDAC(dev, head, ramdac_reg + 0x20, regp->fp_horiz_regs[i]); + } + + if (nv_gf4_disp_arch(dev)) { + NVWriteRAMDAC(dev, head, NV_RAMDAC_FP_DITHER, regp->dither); + for (i = 0; i < 3; i++) { + NVWriteRAMDAC(dev, head, NV_PRAMDAC_850 + i * 4, regp->dither_regs[i]); + NVWriteRAMDAC(dev, head, NV_PRAMDAC_85C + i * 4, regp->dither_regs[i + 3]); + } + } + + NVWriteRAMDAC(dev, head, NV_PRAMDAC_FP_TG_CONTROL, regp->fp_control); + NVWriteRAMDAC(dev, head, NV_PRAMDAC_FP_DEBUG_0, regp->fp_debug_0); + NVWriteRAMDAC(dev, head, NV_PRAMDAC_FP_DEBUG_1, regp->fp_debug_1); + NVWriteRAMDAC(dev, head, NV_PRAMDAC_FP_DEBUG_2, regp->fp_debug_2); + + NVWriteRAMDAC(dev, head, NV_PRAMDAC_FP_MARGIN_COLOR, regp->fp_margin_color); + + if (nv_gf4_disp_arch(dev)) + NVWriteRAMDAC(dev, head, NV_PRAMDAC_8C0, regp->ramdac_8c0); + + if (dev_priv->card_type == NV_40) { + NVWriteRAMDAC(dev, head, NV_PRAMDAC_A20, regp->ramdac_a20); + NVWriteRAMDAC(dev, head, NV_PRAMDAC_A24, regp->ramdac_a24); + NVWriteRAMDAC(dev, head, NV_PRAMDAC_A34, regp->ramdac_a34); + + for (i = 0; i < 38; i++) + NVWriteRAMDAC(dev, head, + NV_PRAMDAC_CTV + 4*i, regp->ctv_regs[i]); + } +} + +static void +nv_save_state_vga(struct drm_device *dev, int head, + struct nv04_mode_state *state) +{ + struct nv04_crtc_reg *regp = &state->crtc_reg[head]; + int i; + + regp->MiscOutReg = NVReadPRMVIO(dev, head, NV_PRMVIO_MISC__READ); + + for (i = 0; i < 25; i++) + rd_cio_state(dev, head, regp, i); + + NVSetEnablePalette(dev, head, true); + for (i = 0; i < 21; i++) + regp->Attribute[i] = NVReadVgaAttr(dev, head, i); + NVSetEnablePalette(dev, head, false); + + for (i = 0; i < 9; i++) + regp->Graphics[i] = NVReadVgaGr(dev, head, i); + + for (i = 0; i < 5; i++) + regp->Sequencer[i] = NVReadVgaSeq(dev, head, i); +} + +static void +nv_load_state_vga(struct drm_device *dev, int head, + struct nv04_mode_state *state) +{ + struct nv04_crtc_reg *regp = &state->crtc_reg[head]; + int i; + + NVWritePRMVIO(dev, head, NV_PRMVIO_MISC__WRITE, regp->MiscOutReg); + + for (i = 0; i < 5; i++) + NVWriteVgaSeq(dev, head, i, regp->Sequencer[i]); + + nv_lock_vga_crtc_base(dev, head, false); + for (i = 0; i < 25; i++) + wr_cio_state(dev, head, regp, i); + nv_lock_vga_crtc_base(dev, head, true); + + for (i = 0; i < 9; i++) + NVWriteVgaGr(dev, head, i, regp->Graphics[i]); + + NVSetEnablePalette(dev, head, true); + for (i = 0; i < 21; i++) + NVWriteVgaAttr(dev, head, i, regp->Attribute[i]); + NVSetEnablePalette(dev, head, false); +} + +static void +nv_save_state_ext(struct drm_device *dev, int head, + struct nv04_mode_state *state) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nv04_crtc_reg *regp = &state->crtc_reg[head]; + int i; + + rd_cio_state(dev, head, regp, NV_CIO_CRE_LCD__INDEX); + rd_cio_state(dev, head, regp, NV_CIO_CRE_RPC0_INDEX); + rd_cio_state(dev, head, regp, NV_CIO_CRE_RPC1_INDEX); + rd_cio_state(dev, head, regp, NV_CIO_CRE_LSR_INDEX); + rd_cio_state(dev, head, regp, NV_CIO_CRE_PIXEL_INDEX); + rd_cio_state(dev, head, regp, NV_CIO_CRE_HEB__INDEX); + rd_cio_state(dev, head, regp, NV_CIO_CRE_ENH_INDEX); + + rd_cio_state(dev, head, regp, NV_CIO_CRE_FF_INDEX); + rd_cio_state(dev, head, regp, NV_CIO_CRE_FFLWM__INDEX); + rd_cio_state(dev, head, regp, NV_CIO_CRE_21); + + if (dev_priv->card_type >= NV_20) + rd_cio_state(dev, head, regp, NV_CIO_CRE_47); + + if (dev_priv->card_type >= NV_30) + rd_cio_state(dev, head, regp, 0x9f); + + rd_cio_state(dev, head, regp, NV_CIO_CRE_49); + rd_cio_state(dev, head, regp, NV_CIO_CRE_HCUR_ADDR0_INDEX); + rd_cio_state(dev, head, regp, NV_CIO_CRE_HCUR_ADDR1_INDEX); + rd_cio_state(dev, head, regp, NV_CIO_CRE_HCUR_ADDR2_INDEX); + rd_cio_state(dev, head, regp, NV_CIO_CRE_ILACE__INDEX); + + if (dev_priv->card_type >= NV_10) { + regp->crtc_830 = NVReadCRTC(dev, head, NV_PCRTC_830); + regp->crtc_834 = NVReadCRTC(dev, head, NV_PCRTC_834); + + if (dev_priv->card_type >= NV_30) + regp->gpio_ext = NVReadCRTC(dev, head, NV_PCRTC_GPIO_EXT); + + if (dev_priv->card_type == NV_40) + regp->crtc_850 = NVReadCRTC(dev, head, NV_PCRTC_850); + + if (nv_two_heads(dev)) + regp->crtc_eng_ctrl = NVReadCRTC(dev, head, NV_PCRTC_ENGINE_CTRL); + regp->cursor_cfg = NVReadCRTC(dev, head, NV_PCRTC_CURSOR_CONFIG); + } + + regp->crtc_cfg = NVReadCRTC(dev, head, NV_PCRTC_CONFIG); + + rd_cio_state(dev, head, regp, NV_CIO_CRE_SCRATCH3__INDEX); + rd_cio_state(dev, head, regp, NV_CIO_CRE_SCRATCH4__INDEX); + if (dev_priv->card_type >= NV_10) { + rd_cio_state(dev, head, regp, NV_CIO_CRE_EBR_INDEX); + rd_cio_state(dev, head, regp, NV_CIO_CRE_CSB); + rd_cio_state(dev, head, regp, NV_CIO_CRE_4B); + rd_cio_state(dev, head, regp, NV_CIO_CRE_TVOUT_LATENCY); + } + /* NV11 and NV20 don't have this, they stop at 0x52. */ + if (nv_gf4_disp_arch(dev)) { + rd_cio_state(dev, head, regp, NV_CIO_CRE_42); + rd_cio_state(dev, head, regp, NV_CIO_CRE_53); + rd_cio_state(dev, head, regp, NV_CIO_CRE_54); + + for (i = 0; i < 0x10; i++) + regp->CR58[i] = NVReadVgaCrtc5758(dev, head, i); + rd_cio_state(dev, head, regp, NV_CIO_CRE_59); + rd_cio_state(dev, head, regp, NV_CIO_CRE_5B); + + rd_cio_state(dev, head, regp, NV_CIO_CRE_85); + rd_cio_state(dev, head, regp, NV_CIO_CRE_86); + } + + regp->fb_start = NVReadCRTC(dev, head, NV_PCRTC_START); +} + +static void +nv_load_state_ext(struct drm_device *dev, int head, + struct nv04_mode_state *state) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nv04_crtc_reg *regp = &state->crtc_reg[head]; + uint32_t reg900; + int i; + + if (dev_priv->card_type >= NV_10) { + if (nv_two_heads(dev)) + /* setting ENGINE_CTRL (EC) *must* come before + * CIO_CRE_LCD, as writing CRE_LCD sets bits 16 & 17 in + * EC that should not be overwritten by writing stale EC + */ + NVWriteCRTC(dev, head, NV_PCRTC_ENGINE_CTRL, regp->crtc_eng_ctrl); + + nvWriteVIDEO(dev, NV_PVIDEO_STOP, 1); + nvWriteVIDEO(dev, NV_PVIDEO_INTR_EN, 0); + nvWriteVIDEO(dev, NV_PVIDEO_OFFSET_BUFF(0), 0); + nvWriteVIDEO(dev, NV_PVIDEO_OFFSET_BUFF(1), 0); + nvWriteVIDEO(dev, NV_PVIDEO_LIMIT(0), dev_priv->fb_available_size - 1); + nvWriteVIDEO(dev, NV_PVIDEO_LIMIT(1), dev_priv->fb_available_size - 1); + nvWriteVIDEO(dev, NV_PVIDEO_UVPLANE_LIMIT(0), dev_priv->fb_available_size - 1); + nvWriteVIDEO(dev, NV_PVIDEO_UVPLANE_LIMIT(1), dev_priv->fb_available_size - 1); + nvWriteMC(dev, NV_PBUS_POWERCTRL_2, 0); + + NVWriteCRTC(dev, head, NV_PCRTC_CURSOR_CONFIG, regp->cursor_cfg); + NVWriteCRTC(dev, head, NV_PCRTC_830, regp->crtc_830); + NVWriteCRTC(dev, head, NV_PCRTC_834, regp->crtc_834); + + if (dev_priv->card_type >= NV_30) + NVWriteCRTC(dev, head, NV_PCRTC_GPIO_EXT, regp->gpio_ext); + + if (dev_priv->card_type == NV_40) { + NVWriteCRTC(dev, head, NV_PCRTC_850, regp->crtc_850); + + reg900 = NVReadRAMDAC(dev, head, NV_PRAMDAC_900); + if (regp->crtc_cfg == NV10_PCRTC_CONFIG_START_ADDRESS_HSYNC) + NVWriteRAMDAC(dev, head, NV_PRAMDAC_900, reg900 | 0x10000); + else + NVWriteRAMDAC(dev, head, NV_PRAMDAC_900, reg900 & ~0x10000); + } + } + + NVWriteCRTC(dev, head, NV_PCRTC_CONFIG, regp->crtc_cfg); + + wr_cio_state(dev, head, regp, NV_CIO_CRE_RPC0_INDEX); + wr_cio_state(dev, head, regp, NV_CIO_CRE_RPC1_INDEX); + wr_cio_state(dev, head, regp, NV_CIO_CRE_LSR_INDEX); + wr_cio_state(dev, head, regp, NV_CIO_CRE_PIXEL_INDEX); + wr_cio_state(dev, head, regp, NV_CIO_CRE_LCD__INDEX); + wr_cio_state(dev, head, regp, NV_CIO_CRE_HEB__INDEX); + wr_cio_state(dev, head, regp, NV_CIO_CRE_ENH_INDEX); + wr_cio_state(dev, head, regp, NV_CIO_CRE_FF_INDEX); + wr_cio_state(dev, head, regp, NV_CIO_CRE_FFLWM__INDEX); + + if (dev_priv->card_type >= NV_20) + wr_cio_state(dev, head, regp, NV_CIO_CRE_47); + + if (dev_priv->card_type >= NV_30) + wr_cio_state(dev, head, regp, 0x9f); + + wr_cio_state(dev, head, regp, NV_CIO_CRE_49); + wr_cio_state(dev, head, regp, NV_CIO_CRE_HCUR_ADDR0_INDEX); + wr_cio_state(dev, head, regp, NV_CIO_CRE_HCUR_ADDR1_INDEX); + wr_cio_state(dev, head, regp, NV_CIO_CRE_HCUR_ADDR2_INDEX); + if (dev_priv->card_type == NV_40) + nv_fix_nv40_hw_cursor(dev, head); + wr_cio_state(dev, head, regp, NV_CIO_CRE_ILACE__INDEX); + + wr_cio_state(dev, head, regp, NV_CIO_CRE_SCRATCH3__INDEX); + wr_cio_state(dev, head, regp, NV_CIO_CRE_SCRATCH4__INDEX); + if (dev_priv->card_type >= NV_10) { + wr_cio_state(dev, head, regp, NV_CIO_CRE_EBR_INDEX); + wr_cio_state(dev, head, regp, NV_CIO_CRE_CSB); + wr_cio_state(dev, head, regp, NV_CIO_CRE_4B); + wr_cio_state(dev, head, regp, NV_CIO_CRE_TVOUT_LATENCY); + } + /* NV11 and NV20 stop at 0x52. */ + if (nv_gf4_disp_arch(dev)) { + if (dev_priv->card_type == NV_10) { + /* Not waiting for vertical retrace before modifying + CRE_53/CRE_54 causes lockups. */ + nouveau_wait_eq(dev, 650000000, NV_PRMCIO_INP0__COLOR, 0x8, 0x8); + nouveau_wait_eq(dev, 650000000, NV_PRMCIO_INP0__COLOR, 0x8, 0x0); + } + + wr_cio_state(dev, head, regp, NV_CIO_CRE_42); + wr_cio_state(dev, head, regp, NV_CIO_CRE_53); + wr_cio_state(dev, head, regp, NV_CIO_CRE_54); + + for (i = 0; i < 0x10; i++) + NVWriteVgaCrtc5758(dev, head, i, regp->CR58[i]); + wr_cio_state(dev, head, regp, NV_CIO_CRE_59); + wr_cio_state(dev, head, regp, NV_CIO_CRE_5B); + + wr_cio_state(dev, head, regp, NV_CIO_CRE_85); + wr_cio_state(dev, head, regp, NV_CIO_CRE_86); + } + + NVWriteCRTC(dev, head, NV_PCRTC_START, regp->fb_start); + + /* Enable vblank interrupts. */ + NVWriteCRTC(dev, head, NV_PCRTC_INTR_EN_0, + (dev->vblank_enabled[head] ? 1 : 0)); + NVWriteCRTC(dev, head, NV_PCRTC_INTR_0, NV_PCRTC_INTR_0_VBLANK); +} + +static void +nv_save_state_palette(struct drm_device *dev, int head, + struct nv04_mode_state *state) +{ + int head_offset = head * NV_PRMDIO_SIZE, i; + + nv_wr08(dev, NV_PRMDIO_PIXEL_MASK + head_offset, + NV_PRMDIO_PIXEL_MASK_MASK); + nv_wr08(dev, NV_PRMDIO_READ_MODE_ADDRESS + head_offset, 0x0); + + for (i = 0; i < 768; i++) { + state->crtc_reg[head].DAC[i] = nv_rd08(dev, + NV_PRMDIO_PALETTE_DATA + head_offset); + } + + NVSetEnablePalette(dev, head, false); +} + +void +nouveau_hw_load_state_palette(struct drm_device *dev, int head, + struct nv04_mode_state *state) +{ + int head_offset = head * NV_PRMDIO_SIZE, i; + + nv_wr08(dev, NV_PRMDIO_PIXEL_MASK + head_offset, + NV_PRMDIO_PIXEL_MASK_MASK); + nv_wr08(dev, NV_PRMDIO_WRITE_MODE_ADDRESS + head_offset, 0x0); + + for (i = 0; i < 768; i++) { + nv_wr08(dev, NV_PRMDIO_PALETTE_DATA + head_offset, + state->crtc_reg[head].DAC[i]); + } + + NVSetEnablePalette(dev, head, false); +} + +void nouveau_hw_save_state(struct drm_device *dev, int head, + struct nv04_mode_state *state) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + if (dev_priv->chipset == 0x11) + /* NB: no attempt is made to restore the bad pll later on */ + nouveau_hw_fix_bad_vpll(dev, head); + nv_save_state_ramdac(dev, head, state); + nv_save_state_vga(dev, head, state); + nv_save_state_palette(dev, head, state); + nv_save_state_ext(dev, head, state); +} + +void nouveau_hw_load_state(struct drm_device *dev, int head, + struct nv04_mode_state *state) +{ + NVVgaProtect(dev, head, true); + nv_load_state_ramdac(dev, head, state); + nv_load_state_ext(dev, head, state); + nouveau_hw_load_state_palette(dev, head, state); + nv_load_state_vga(dev, head, state); + NVVgaProtect(dev, head, false); +} diff --git a/drivers/gpu/drm/nouveau/nouveau_hw.h b/drivers/gpu/drm/nouveau/nouveau_hw.h new file mode 100644 index 00000000..2989090b --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_hw.h @@ -0,0 +1,474 @@ +/* + * Copyright 2008 Stuart Bennett + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +#ifndef __NOUVEAU_HW_H__ +#define __NOUVEAU_HW_H__ + +#include "drmP.h" +#include "nouveau_drv.h" + +#define MASK(field) ( \ + (0xffffffff >> (31 - ((1 ? field) - (0 ? field)))) << (0 ? field)) + +#define XLATE(src, srclowbit, outfield) ( \ + (((src) >> (srclowbit)) << (0 ? outfield)) & MASK(outfield)) + +void NVWriteVgaSeq(struct drm_device *, int head, uint8_t index, uint8_t value); +uint8_t NVReadVgaSeq(struct drm_device *, int head, uint8_t index); +void NVWriteVgaGr(struct drm_device *, int head, uint8_t index, uint8_t value); +uint8_t NVReadVgaGr(struct drm_device *, int head, uint8_t index); +void NVSetOwner(struct drm_device *, int owner); +void NVBlankScreen(struct drm_device *, int head, bool blank); +void nouveau_hw_setpll(struct drm_device *, uint32_t reg1, + struct nouveau_pll_vals *pv); +int nouveau_hw_get_pllvals(struct drm_device *, enum pll_types plltype, + struct nouveau_pll_vals *pllvals); +int nouveau_hw_pllvals_to_clk(struct nouveau_pll_vals *pllvals); +int nouveau_hw_get_clock(struct drm_device *, enum pll_types plltype); +void nouveau_hw_save_vga_fonts(struct drm_device *, bool save); +void nouveau_hw_save_state(struct drm_device *, int head, + struct nv04_mode_state *state); +void nouveau_hw_load_state(struct drm_device *, int head, + struct nv04_mode_state *state); +void nouveau_hw_load_state_palette(struct drm_device *, int head, + struct nv04_mode_state *state); + +/* nouveau_calc.c */ +extern void nouveau_calc_arb(struct drm_device *, int vclk, int bpp, + int *burst, int *lwm); +extern int nouveau_calc_pll_mnp(struct drm_device *, struct pll_lims *pll_lim, + int clk, struct nouveau_pll_vals *pv); + +static inline uint32_t +nvReadMC(struct drm_device *dev, uint32_t reg) +{ + uint32_t val = nv_rd32(dev, reg); + NV_REG_DEBUG(MC, dev, "reg %08x val %08x\n", reg, val); + return val; +} + +static inline void +nvWriteMC(struct drm_device *dev, uint32_t reg, uint32_t val) +{ + NV_REG_DEBUG(MC, dev, "reg %08x val %08x\n", reg, val); + nv_wr32(dev, reg, val); +} + +static inline uint32_t +nvReadVIDEO(struct drm_device *dev, uint32_t reg) +{ + uint32_t val = nv_rd32(dev, reg); + NV_REG_DEBUG(VIDEO, dev, "reg %08x val %08x\n", reg, val); + return val; +} + +static inline void +nvWriteVIDEO(struct drm_device *dev, uint32_t reg, uint32_t val) +{ + NV_REG_DEBUG(VIDEO, dev, "reg %08x val %08x\n", reg, val); + nv_wr32(dev, reg, val); +} + +static inline uint32_t +nvReadFB(struct drm_device *dev, uint32_t reg) +{ + uint32_t val = nv_rd32(dev, reg); + NV_REG_DEBUG(FB, dev, "reg %08x val %08x\n", reg, val); + return val; +} + +static inline void +nvWriteFB(struct drm_device *dev, uint32_t reg, uint32_t val) +{ + NV_REG_DEBUG(FB, dev, "reg %08x val %08x\n", reg, val); + nv_wr32(dev, reg, val); +} + +static inline uint32_t +nvReadEXTDEV(struct drm_device *dev, uint32_t reg) +{ + uint32_t val = nv_rd32(dev, reg); + NV_REG_DEBUG(EXTDEV, dev, "reg %08x val %08x\n", reg, val); + return val; +} + +static inline void +nvWriteEXTDEV(struct drm_device *dev, uint32_t reg, uint32_t val) +{ + NV_REG_DEBUG(EXTDEV, dev, "reg %08x val %08x\n", reg, val); + nv_wr32(dev, reg, val); +} + +static inline uint32_t NVReadCRTC(struct drm_device *dev, + int head, uint32_t reg) +{ + uint32_t val; + if (head) + reg += NV_PCRTC0_SIZE; + val = nv_rd32(dev, reg); + NV_REG_DEBUG(CRTC, dev, "head %d reg %08x val %08x\n", head, reg, val); + return val; +} + +static inline void NVWriteCRTC(struct drm_device *dev, + int head, uint32_t reg, uint32_t val) +{ + if (head) + reg += NV_PCRTC0_SIZE; + NV_REG_DEBUG(CRTC, dev, "head %d reg %08x val %08x\n", head, reg, val); + nv_wr32(dev, reg, val); +} + +static inline uint32_t NVReadRAMDAC(struct drm_device *dev, + int head, uint32_t reg) +{ + uint32_t val; + if (head) + reg += NV_PRAMDAC0_SIZE; + val = nv_rd32(dev, reg); + NV_REG_DEBUG(RAMDAC, dev, "head %d reg %08x val %08x\n", + head, reg, val); + return val; +} + +static inline void NVWriteRAMDAC(struct drm_device *dev, + int head, uint32_t reg, uint32_t val) +{ + if (head) + reg += NV_PRAMDAC0_SIZE; + NV_REG_DEBUG(RAMDAC, dev, "head %d reg %08x val %08x\n", + head, reg, val); + nv_wr32(dev, reg, val); +} + +static inline uint8_t nv_read_tmds(struct drm_device *dev, + int or, int dl, uint8_t address) +{ + int ramdac = (or & OUTPUT_C) >> 2; + + NVWriteRAMDAC(dev, ramdac, NV_PRAMDAC_FP_TMDS_CONTROL + dl * 8, + NV_PRAMDAC_FP_TMDS_CONTROL_WRITE_DISABLE | address); + return NVReadRAMDAC(dev, ramdac, NV_PRAMDAC_FP_TMDS_DATA + dl * 8); +} + +static inline void nv_write_tmds(struct drm_device *dev, + int or, int dl, uint8_t address, + uint8_t data) +{ + int ramdac = (or & OUTPUT_C) >> 2; + + NVWriteRAMDAC(dev, ramdac, NV_PRAMDAC_FP_TMDS_DATA + dl * 8, data); + NVWriteRAMDAC(dev, ramdac, NV_PRAMDAC_FP_TMDS_CONTROL + dl * 8, address); +} + +static inline void NVWriteVgaCrtc(struct drm_device *dev, + int head, uint8_t index, uint8_t value) +{ + NV_REG_DEBUG(VGACRTC, dev, "head %d index 0x%02x data 0x%02x\n", + head, index, value); + nv_wr08(dev, NV_PRMCIO_CRX__COLOR + head * NV_PRMCIO_SIZE, index); + nv_wr08(dev, NV_PRMCIO_CR__COLOR + head * NV_PRMCIO_SIZE, value); +} + +static inline uint8_t NVReadVgaCrtc(struct drm_device *dev, + int head, uint8_t index) +{ + uint8_t val; + nv_wr08(dev, NV_PRMCIO_CRX__COLOR + head * NV_PRMCIO_SIZE, index); + val = nv_rd08(dev, NV_PRMCIO_CR__COLOR + head * NV_PRMCIO_SIZE); + NV_REG_DEBUG(VGACRTC, dev, "head %d index 0x%02x data 0x%02x\n", + head, index, val); + return val; +} + +/* CR57 and CR58 are a fun pair of regs. CR57 provides an index (0-0xf) for CR58 + * I suspect they in fact do nothing, but are merely a way to carry useful + * per-head variables around + * + * Known uses: + * CR57 CR58 + * 0x00 index to the appropriate dcb entry (or 7f for inactive) + * 0x02 dcb entry's "or" value (or 00 for inactive) + * 0x03 bit0 set for dual link (LVDS, possibly elsewhere too) + * 0x08 or 0x09 pxclk in MHz + * 0x0f laptop panel info - low nibble for PEXTDEV_BOOT_0 strap + * high nibble for xlat strap value + */ + +static inline void +NVWriteVgaCrtc5758(struct drm_device *dev, int head, uint8_t index, uint8_t value) +{ + NVWriteVgaCrtc(dev, head, NV_CIO_CRE_57, index); + NVWriteVgaCrtc(dev, head, NV_CIO_CRE_58, value); +} + +static inline uint8_t NVReadVgaCrtc5758(struct drm_device *dev, int head, uint8_t index) +{ + NVWriteVgaCrtc(dev, head, NV_CIO_CRE_57, index); + return NVReadVgaCrtc(dev, head, NV_CIO_CRE_58); +} + +static inline uint8_t NVReadPRMVIO(struct drm_device *dev, + int head, uint32_t reg) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + uint8_t val; + + /* Only NV4x have two pvio ranges; other twoHeads cards MUST call + * NVSetOwner for the relevant head to be programmed */ + if (head && dev_priv->card_type == NV_40) + reg += NV_PRMVIO_SIZE; + + val = nv_rd08(dev, reg); + NV_REG_DEBUG(RMVIO, dev, "head %d reg %08x val %02x\n", head, reg, val); + return val; +} + +static inline void NVWritePRMVIO(struct drm_device *dev, + int head, uint32_t reg, uint8_t value) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + /* Only NV4x have two pvio ranges; other twoHeads cards MUST call + * NVSetOwner for the relevant head to be programmed */ + if (head && dev_priv->card_type == NV_40) + reg += NV_PRMVIO_SIZE; + + NV_REG_DEBUG(RMVIO, dev, "head %d reg %08x val %02x\n", + head, reg, value); + nv_wr08(dev, reg, value); +} + +static inline void NVSetEnablePalette(struct drm_device *dev, int head, bool enable) +{ + nv_rd08(dev, NV_PRMCIO_INP0__COLOR + head * NV_PRMCIO_SIZE); + nv_wr08(dev, NV_PRMCIO_ARX + head * NV_PRMCIO_SIZE, enable ? 0 : 0x20); +} + +static inline bool NVGetEnablePalette(struct drm_device *dev, int head) +{ + nv_rd08(dev, NV_PRMCIO_INP0__COLOR + head * NV_PRMCIO_SIZE); + return !(nv_rd08(dev, NV_PRMCIO_ARX + head * NV_PRMCIO_SIZE) & 0x20); +} + +static inline void NVWriteVgaAttr(struct drm_device *dev, + int head, uint8_t index, uint8_t value) +{ + if (NVGetEnablePalette(dev, head)) + index &= ~0x20; + else + index |= 0x20; + + nv_rd08(dev, NV_PRMCIO_INP0__COLOR + head * NV_PRMCIO_SIZE); + NV_REG_DEBUG(VGAATTR, dev, "head %d index 0x%02x data 0x%02x\n", + head, index, value); + nv_wr08(dev, NV_PRMCIO_ARX + head * NV_PRMCIO_SIZE, index); + nv_wr08(dev, NV_PRMCIO_AR__WRITE + head * NV_PRMCIO_SIZE, value); +} + +static inline uint8_t NVReadVgaAttr(struct drm_device *dev, + int head, uint8_t index) +{ + uint8_t val; + if (NVGetEnablePalette(dev, head)) + index &= ~0x20; + else + index |= 0x20; + + nv_rd08(dev, NV_PRMCIO_INP0__COLOR + head * NV_PRMCIO_SIZE); + nv_wr08(dev, NV_PRMCIO_ARX + head * NV_PRMCIO_SIZE, index); + val = nv_rd08(dev, NV_PRMCIO_AR__READ + head * NV_PRMCIO_SIZE); + NV_REG_DEBUG(VGAATTR, dev, "head %d index 0x%02x data 0x%02x\n", + head, index, val); + return val; +} + +static inline void NVVgaSeqReset(struct drm_device *dev, int head, bool start) +{ + NVWriteVgaSeq(dev, head, NV_VIO_SR_RESET_INDEX, start ? 0x1 : 0x3); +} + +static inline void NVVgaProtect(struct drm_device *dev, int head, bool protect) +{ + uint8_t seq1 = NVReadVgaSeq(dev, head, NV_VIO_SR_CLOCK_INDEX); + + if (protect) { + NVVgaSeqReset(dev, head, true); + NVWriteVgaSeq(dev, head, NV_VIO_SR_CLOCK_INDEX, seq1 | 0x20); + } else { + /* Reenable sequencer, then turn on screen */ + NVWriteVgaSeq(dev, head, NV_VIO_SR_CLOCK_INDEX, seq1 & ~0x20); /* reenable display */ + NVVgaSeqReset(dev, head, false); + } + NVSetEnablePalette(dev, head, protect); +} + +static inline bool +nv_heads_tied(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + if (dev_priv->chipset == 0x11) + return !!(nvReadMC(dev, NV_PBUS_DEBUG_1) & (1 << 28)); + + return NVReadVgaCrtc(dev, 0, NV_CIO_CRE_44) & 0x4; +} + +/* makes cr0-7 on the specified head read-only */ +static inline bool +nv_lock_vga_crtc_base(struct drm_device *dev, int head, bool lock) +{ + uint8_t cr11 = NVReadVgaCrtc(dev, head, NV_CIO_CR_VRE_INDEX); + bool waslocked = cr11 & 0x80; + + if (lock) + cr11 |= 0x80; + else + cr11 &= ~0x80; + NVWriteVgaCrtc(dev, head, NV_CIO_CR_VRE_INDEX, cr11); + + return waslocked; +} + +static inline void +nv_lock_vga_crtc_shadow(struct drm_device *dev, int head, int lock) +{ + /* shadow lock: connects 0x60?3d? regs to "real" 0x3d? regs + * bit7: unlocks HDT, HBS, HBE, HRS, HRE, HEB + * bit6: seems to have some effect on CR09 (double scan, VBS_9) + * bit5: unlocks HDE + * bit4: unlocks VDE + * bit3: unlocks VDT, OVL, VRS, ?VRE?, VBS, VBE, LSR, EBR + * bit2: same as bit 1 of 0x60?804 + * bit0: same as bit 0 of 0x60?804 + */ + + uint8_t cr21 = lock; + + if (lock < 0) + /* 0xfa is generic "unlock all" mask */ + cr21 = NVReadVgaCrtc(dev, head, NV_CIO_CRE_21) | 0xfa; + + NVWriteVgaCrtc(dev, head, NV_CIO_CRE_21, cr21); +} + +/* renders the extended crtc regs (cr19+) on all crtcs impervious: + * immutable and unreadable + */ +static inline bool +NVLockVgaCrtcs(struct drm_device *dev, bool lock) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + bool waslocked = !NVReadVgaCrtc(dev, 0, NV_CIO_SR_LOCK_INDEX); + + NVWriteVgaCrtc(dev, 0, NV_CIO_SR_LOCK_INDEX, + lock ? NV_CIO_SR_LOCK_VALUE : NV_CIO_SR_UNLOCK_RW_VALUE); + /* NV11 has independently lockable extended crtcs, except when tied */ + if (dev_priv->chipset == 0x11 && !nv_heads_tied(dev)) + NVWriteVgaCrtc(dev, 1, NV_CIO_SR_LOCK_INDEX, + lock ? NV_CIO_SR_LOCK_VALUE : + NV_CIO_SR_UNLOCK_RW_VALUE); + + return waslocked; +} + +/* nv04 cursor max dimensions of 32x32 (A1R5G5B5) */ +#define NV04_CURSOR_SIZE 32 +/* limit nv10 cursors to 64x64 (ARGB8) (we could go to 64x255) */ +#define NV10_CURSOR_SIZE 64 + +static inline int nv_cursor_width(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + return dev_priv->card_type >= NV_10 ? NV10_CURSOR_SIZE : NV04_CURSOR_SIZE; +} + +static inline void +nv_fix_nv40_hw_cursor(struct drm_device *dev, int head) +{ + /* on some nv40 (such as the "true" (in the NV_PFB_BOOT_0 sense) nv40, + * the gf6800gt) a hardware bug requires a write to PRAMDAC_CURSOR_POS + * for changes to the CRTC CURCTL regs to take effect, whether changing + * the pixmap location, or just showing/hiding the cursor + */ + uint32_t curpos = NVReadRAMDAC(dev, head, NV_PRAMDAC_CU_START_POS); + NVWriteRAMDAC(dev, head, NV_PRAMDAC_CU_START_POS, curpos); +} + +static inline void +nv_set_crtc_base(struct drm_device *dev, int head, uint32_t offset) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + NVWriteCRTC(dev, head, NV_PCRTC_START, offset); + + if (dev_priv->card_type == NV_04) { + /* + * Hilarious, the 24th bit doesn't want to stick to + * PCRTC_START... + */ + int cre_heb = NVReadVgaCrtc(dev, head, NV_CIO_CRE_HEB__INDEX); + + NVWriteVgaCrtc(dev, head, NV_CIO_CRE_HEB__INDEX, + (cre_heb & ~0x40) | ((offset >> 18) & 0x40)); + } +} + +static inline void +nv_show_cursor(struct drm_device *dev, int head, bool show) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + uint8_t *curctl1 = + &dev_priv->mode_reg.crtc_reg[head].CRTC[NV_CIO_CRE_HCUR_ADDR1_INDEX]; + + if (show) + *curctl1 |= MASK(NV_CIO_CRE_HCUR_ADDR1_ENABLE); + else + *curctl1 &= ~MASK(NV_CIO_CRE_HCUR_ADDR1_ENABLE); + NVWriteVgaCrtc(dev, head, NV_CIO_CRE_HCUR_ADDR1_INDEX, *curctl1); + + if (dev_priv->card_type == NV_40) + nv_fix_nv40_hw_cursor(dev, head); +} + +static inline uint32_t +nv_pitch_align(struct drm_device *dev, uint32_t width, int bpp) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + int mask; + + if (bpp == 15) + bpp = 16; + if (bpp == 24) + bpp = 8; + + /* Alignment requirements taken from the Haiku driver */ + if (dev_priv->card_type == NV_04) + mask = 128 / bpp - 1; + else + mask = 512 / bpp - 1; + + return (width + mask) & ~mask; +} + +#endif /* __NOUVEAU_HW_H__ */ diff --git a/drivers/gpu/drm/nouveau/nouveau_i2c.c b/drivers/gpu/drm/nouveau/nouveau_i2c.c new file mode 100644 index 00000000..cb389d01 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_i2c.c @@ -0,0 +1,323 @@ +/* + * Copyright 2009 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include "drmP.h" +#include "nouveau_drv.h" +#include "nouveau_i2c.h" +#include "nouveau_hw.h" + +static void +nv04_i2c_setscl(void *data, int state) +{ + struct nouveau_i2c_chan *i2c = data; + struct drm_device *dev = i2c->dev; + uint8_t val; + + val = (NVReadVgaCrtc(dev, 0, i2c->wr) & 0xd0) | (state ? 0x20 : 0); + NVWriteVgaCrtc(dev, 0, i2c->wr, val | 0x01); +} + +static void +nv04_i2c_setsda(void *data, int state) +{ + struct nouveau_i2c_chan *i2c = data; + struct drm_device *dev = i2c->dev; + uint8_t val; + + val = (NVReadVgaCrtc(dev, 0, i2c->wr) & 0xe0) | (state ? 0x10 : 0); + NVWriteVgaCrtc(dev, 0, i2c->wr, val | 0x01); +} + +static int +nv04_i2c_getscl(void *data) +{ + struct nouveau_i2c_chan *i2c = data; + struct drm_device *dev = i2c->dev; + + return !!(NVReadVgaCrtc(dev, 0, i2c->rd) & 4); +} + +static int +nv04_i2c_getsda(void *data) +{ + struct nouveau_i2c_chan *i2c = data; + struct drm_device *dev = i2c->dev; + + return !!(NVReadVgaCrtc(dev, 0, i2c->rd) & 8); +} + +static void +nv4e_i2c_setscl(void *data, int state) +{ + struct nouveau_i2c_chan *i2c = data; + struct drm_device *dev = i2c->dev; + uint8_t val; + + val = (nv_rd32(dev, i2c->wr) & 0xd0) | (state ? 0x20 : 0); + nv_wr32(dev, i2c->wr, val | 0x01); +} + +static void +nv4e_i2c_setsda(void *data, int state) +{ + struct nouveau_i2c_chan *i2c = data; + struct drm_device *dev = i2c->dev; + uint8_t val; + + val = (nv_rd32(dev, i2c->wr) & 0xe0) | (state ? 0x10 : 0); + nv_wr32(dev, i2c->wr, val | 0x01); +} + +static int +nv4e_i2c_getscl(void *data) +{ + struct nouveau_i2c_chan *i2c = data; + struct drm_device *dev = i2c->dev; + + return !!((nv_rd32(dev, i2c->rd) >> 16) & 4); +} + +static int +nv4e_i2c_getsda(void *data) +{ + struct nouveau_i2c_chan *i2c = data; + struct drm_device *dev = i2c->dev; + + return !!((nv_rd32(dev, i2c->rd) >> 16) & 8); +} + +static int +nv50_i2c_getscl(void *data) +{ + struct nouveau_i2c_chan *i2c = data; + struct drm_device *dev = i2c->dev; + + return !!(nv_rd32(dev, i2c->rd) & 1); +} + + +static int +nv50_i2c_getsda(void *data) +{ + struct nouveau_i2c_chan *i2c = data; + struct drm_device *dev = i2c->dev; + + return !!(nv_rd32(dev, i2c->rd) & 2); +} + +static void +nv50_i2c_setscl(void *data, int state) +{ + struct nouveau_i2c_chan *i2c = data; + struct drm_device *dev = i2c->dev; + + nv_wr32(dev, i2c->wr, 4 | (i2c->data ? 2 : 0) | (state ? 1 : 0)); +} + +static void +nv50_i2c_setsda(void *data, int state) +{ + struct nouveau_i2c_chan *i2c = data; + struct drm_device *dev = i2c->dev; + + nv_wr32(dev, i2c->wr, + (nv_rd32(dev, i2c->rd) & 1) | 4 | (state ? 2 : 0)); + i2c->data = state; +} + +static const uint32_t nv50_i2c_port[] = { + 0x00e138, 0x00e150, 0x00e168, 0x00e180, + 0x00e254, 0x00e274, 0x00e764, 0x00e780, + 0x00e79c, 0x00e7b8 +}; +#define NV50_I2C_PORTS ARRAY_SIZE(nv50_i2c_port) + +int +nouveau_i2c_init(struct drm_device *dev, struct dcb_i2c_entry *entry, int index) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_i2c_chan *i2c; + int ret; + + if (entry->chan) + return -EEXIST; + + if (dev_priv->card_type >= NV_50 && entry->read >= NV50_I2C_PORTS) { + NV_ERROR(dev, "unknown i2c port %d\n", entry->read); + return -EINVAL; + } + + i2c = kzalloc(sizeof(*i2c), GFP_KERNEL); + if (i2c == NULL) + return -ENOMEM; + + switch (entry->port_type) { + case 0: + i2c->bit.setsda = nv04_i2c_setsda; + i2c->bit.setscl = nv04_i2c_setscl; + i2c->bit.getsda = nv04_i2c_getsda; + i2c->bit.getscl = nv04_i2c_getscl; + i2c->rd = entry->read; + i2c->wr = entry->write; + break; + case 4: + i2c->bit.setsda = nv4e_i2c_setsda; + i2c->bit.setscl = nv4e_i2c_setscl; + i2c->bit.getsda = nv4e_i2c_getsda; + i2c->bit.getscl = nv4e_i2c_getscl; + i2c->rd = 0x600800 + entry->read; + i2c->wr = 0x600800 + entry->write; + break; + case 5: + i2c->bit.setsda = nv50_i2c_setsda; + i2c->bit.setscl = nv50_i2c_setscl; + i2c->bit.getsda = nv50_i2c_getsda; + i2c->bit.getscl = nv50_i2c_getscl; + i2c->rd = nv50_i2c_port[entry->read]; + i2c->wr = i2c->rd; + break; + case 6: + i2c->rd = entry->read; + i2c->wr = entry->write; + break; + default: + NV_ERROR(dev, "DCB I2C port type %d unknown\n", + entry->port_type); + kfree(i2c); + return -EINVAL; + } + + snprintf(i2c->adapter.name, sizeof(i2c->adapter.name), + "nouveau-%s-%d", pci_name(dev->pdev), index); + i2c->adapter.owner = THIS_MODULE; + i2c->adapter.dev.parent = &dev->pdev->dev; + i2c->dev = dev; + i2c_set_adapdata(&i2c->adapter, i2c); + + if (entry->port_type < 6) { + i2c->adapter.algo_data = &i2c->bit; + i2c->bit.udelay = 40; + i2c->bit.timeout = usecs_to_jiffies(5000); + i2c->bit.data = i2c; + ret = i2c_bit_add_bus(&i2c->adapter); + } else { + i2c->adapter.algo = &nouveau_dp_i2c_algo; + ret = i2c_add_adapter(&i2c->adapter); + } + + if (ret) { + NV_ERROR(dev, "Failed to register i2c %d\n", index); + kfree(i2c); + return ret; + } + + entry->chan = i2c; + return 0; +} + +void +nouveau_i2c_fini(struct drm_device *dev, struct dcb_i2c_entry *entry) +{ + if (!entry->chan) + return; + + i2c_del_adapter(&entry->chan->adapter); + kfree(entry->chan); + entry->chan = NULL; +} + +struct nouveau_i2c_chan * +nouveau_i2c_find(struct drm_device *dev, int index) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct dcb_i2c_entry *i2c = &dev_priv->vbios.dcb.i2c[index]; + + if (index >= DCB_MAX_NUM_I2C_ENTRIES) + return NULL; + + if (dev_priv->card_type >= NV_50 && (i2c->entry & 0x00000100)) { + uint32_t reg = 0xe500, val; + + if (i2c->port_type == 6) { + reg += i2c->read * 0x50; + val = 0x2002; + } else { + reg += ((i2c->entry & 0x1e00) >> 9) * 0x50; + val = 0xe001; + } + + nv_wr32(dev, reg, (nv_rd32(dev, reg) & ~0xf003) | val); + } + + if (!i2c->chan && nouveau_i2c_init(dev, i2c, index)) + return NULL; + return i2c->chan; +} + +bool +nouveau_probe_i2c_addr(struct nouveau_i2c_chan *i2c, int addr) +{ + uint8_t buf[] = { 0 }; + struct i2c_msg msgs[] = { + { + .addr = addr, + .flags = 0, + .len = 1, + .buf = buf, + }, + { + .addr = addr, + .flags = I2C_M_RD, + .len = 1, + .buf = buf, + } + }; + + return i2c_transfer(&i2c->adapter, msgs, 2) == 2; +} + +int +nouveau_i2c_identify(struct drm_device *dev, const char *what, + struct i2c_board_info *info, + bool (*match)(struct nouveau_i2c_chan *, + struct i2c_board_info *), + int index) +{ + struct nouveau_i2c_chan *i2c = nouveau_i2c_find(dev, index); + int i; + + NV_DEBUG(dev, "Probing %ss on I2C bus: %d\n", what, index); + + for (i = 0; info[i].addr; i++) { + if (nouveau_probe_i2c_addr(i2c, info[i].addr) && + (!match || match(i2c, &info[i]))) { + NV_INFO(dev, "Detected %s: %s\n", what, info[i].type); + return i; + } + } + + NV_DEBUG(dev, "No devices found.\n"); + + return -ENODEV; +} diff --git a/drivers/gpu/drm/nouveau/nouveau_i2c.h b/drivers/gpu/drm/nouveau/nouveau_i2c.h new file mode 100644 index 00000000..422b62fd --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_i2c.h @@ -0,0 +1,53 @@ +/* + * Copyright 2009 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#ifndef __NOUVEAU_I2C_H__ +#define __NOUVEAU_I2C_H__ + +#include +#include +#include "drm_dp_helper.h" + +struct dcb_i2c_entry; + +struct nouveau_i2c_chan { + struct i2c_adapter adapter; + struct drm_device *dev; + struct i2c_algo_bit_data bit; + unsigned rd; + unsigned wr; + unsigned data; +}; + +int nouveau_i2c_init(struct drm_device *, struct dcb_i2c_entry *, int index); +void nouveau_i2c_fini(struct drm_device *, struct dcb_i2c_entry *); +struct nouveau_i2c_chan *nouveau_i2c_find(struct drm_device *, int index); +bool nouveau_probe_i2c_addr(struct nouveau_i2c_chan *i2c, int addr); +int nouveau_i2c_identify(struct drm_device *dev, const char *what, + struct i2c_board_info *info, + bool (*match)(struct nouveau_i2c_chan *, + struct i2c_board_info *), + int index); + +extern const struct i2c_algorithm nouveau_dp_i2c_algo; + +#endif /* __NOUVEAU_I2C_H__ */ diff --git a/drivers/gpu/drm/nouveau/nouveau_ioc32.c b/drivers/gpu/drm/nouveau/nouveau_ioc32.c new file mode 100644 index 00000000..475ba810 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_ioc32.c @@ -0,0 +1,70 @@ +/** + * \file mga_ioc32.c + * + * 32-bit ioctl compatibility routines for the MGA DRM. + * + * \author Dave Airlie with code from patches by Egbert Eich + * + * + * Copyright (C) Paul Mackerras 2005 + * Copyright (C) Egbert Eich 2003,2004 + * Copyright (C) Dave Airlie 2005 + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHOR BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS + * IN THE SOFTWARE. + */ + +#include + +#include "drmP.h" +#include "drm.h" + +#include "nouveau_drv.h" + +/** + * Called whenever a 32-bit process running under a 64-bit kernel + * performs an ioctl on /dev/dri/card. + * + * \param filp file pointer. + * \param cmd command. + * \param arg user argument. + * \return zero on success or negative number on failure. + */ +long nouveau_compat_ioctl(struct file *filp, unsigned int cmd, + unsigned long arg) +{ + unsigned int nr = DRM_IOCTL_NR(cmd); + drm_ioctl_compat_t *fn = NULL; + int ret; + + if (nr < DRM_COMMAND_BASE) + return drm_compat_ioctl(filp, cmd, arg); + +#if 0 + if (nr < DRM_COMMAND_BASE + DRM_ARRAY_SIZE(mga_compat_ioctls)) + fn = nouveau_compat_ioctls[nr - DRM_COMMAND_BASE]; +#endif + if (fn != NULL) + ret = (*fn)(filp, cmd, arg); + else + ret = drm_ioctl(filp, cmd, arg); + + return ret; +} diff --git a/drivers/gpu/drm/nouveau/nouveau_irq.c b/drivers/gpu/drm/nouveau/nouveau_irq.c new file mode 100644 index 00000000..2ba7265b --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_irq.c @@ -0,0 +1,151 @@ +/* + * Copyright (C) 2006 Ben Skeggs. + * + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +/* + * Authors: + * Ben Skeggs + */ + +#include "drmP.h" +#include "drm.h" +#include "nouveau_drm.h" +#include "nouveau_drv.h" +#include "nouveau_reg.h" +#include "nouveau_ramht.h" +#include "nouveau_util.h" + +void +nouveau_irq_preinstall(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + /* Master disable */ + nv_wr32(dev, NV03_PMC_INTR_EN_0, 0); + + INIT_LIST_HEAD(&dev_priv->vbl_waiting); +} + +int +nouveau_irq_postinstall(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + /* Master enable */ + nv_wr32(dev, NV03_PMC_INTR_EN_0, NV_PMC_INTR_EN_0_MASTER_ENABLE); + if (dev_priv->msi_enabled) + nv_wr08(dev, 0x00088068, 0xff); + + return 0; +} + +void +nouveau_irq_uninstall(struct drm_device *dev) +{ + /* Master disable */ + nv_wr32(dev, NV03_PMC_INTR_EN_0, 0); +} + +irqreturn_t +nouveau_irq_handler(DRM_IRQ_ARGS) +{ + struct drm_device *dev = (struct drm_device *)arg; + struct drm_nouveau_private *dev_priv = dev->dev_private; + unsigned long flags; + u32 stat; + int i; + + stat = nv_rd32(dev, NV03_PMC_INTR_0); + if (!stat) + return IRQ_NONE; + + spin_lock_irqsave(&dev_priv->context_switch_lock, flags); + for (i = 0; i < 32 && stat; i++) { + if (!(stat & (1 << i)) || !dev_priv->irq_handler[i]) + continue; + + dev_priv->irq_handler[i](dev); + stat &= ~(1 << i); + } + + if (dev_priv->msi_enabled) + nv_wr08(dev, 0x00088068, 0xff); + spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); + + if (stat && nouveau_ratelimit()) + NV_ERROR(dev, "PMC - unhandled INTR 0x%08x\n", stat); + return IRQ_HANDLED; +} + +int +nouveau_irq_init(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + int ret; + + if (nouveau_msi != 0 && dev_priv->card_type >= NV_50) { + ret = pci_enable_msi(dev->pdev); + if (ret == 0) { + NV_INFO(dev, "enabled MSI\n"); + dev_priv->msi_enabled = true; + } + } + + return drm_irq_install(dev); +} + +void +nouveau_irq_fini(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + drm_irq_uninstall(dev); + if (dev_priv->msi_enabled) + pci_disable_msi(dev->pdev); +} + +void +nouveau_irq_register(struct drm_device *dev, int status_bit, + void (*handler)(struct drm_device *)) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + unsigned long flags; + + spin_lock_irqsave(&dev_priv->context_switch_lock, flags); + dev_priv->irq_handler[status_bit] = handler; + spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); +} + +void +nouveau_irq_unregister(struct drm_device *dev, int status_bit) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + unsigned long flags; + + spin_lock_irqsave(&dev_priv->context_switch_lock, flags); + dev_priv->irq_handler[status_bit] = NULL; + spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); +} diff --git a/drivers/gpu/drm/nouveau/nouveau_mem.c b/drivers/gpu/drm/nouveau/nouveau_mem.c new file mode 100644 index 00000000..5ee14d21 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_mem.c @@ -0,0 +1,930 @@ +/* + * Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + * Copyright 2005 Stephane Marchesin + * + * The Weather Channel (TM) funded Tungsten Graphics to develop the + * initial release of the Radeon 8500 driver under the XFree86 license. + * This notice must be preserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Keith Whitwell + */ + + +#include "drmP.h" +#include "drm.h" +#include "drm_sarea.h" + +#include "nouveau_drv.h" +#include "nouveau_pm.h" +#include "nouveau_mm.h" +#include "nouveau_vm.h" + +/* + * NV10-NV40 tiling helpers + */ + +static void +nv10_mem_update_tile_region(struct drm_device *dev, + struct nouveau_tile_reg *tile, uint32_t addr, + uint32_t size, uint32_t pitch, uint32_t flags) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; + struct nouveau_fb_engine *pfb = &dev_priv->engine.fb; + int i = tile - dev_priv->tile.reg, j; + unsigned long save; + + nouveau_fence_unref(&tile->fence); + + if (tile->pitch) + pfb->free_tile_region(dev, i); + + if (pitch) + pfb->init_tile_region(dev, i, addr, size, pitch, flags); + + spin_lock_irqsave(&dev_priv->context_switch_lock, save); + pfifo->reassign(dev, false); + pfifo->cache_pull(dev, false); + + nouveau_wait_for_idle(dev); + + pfb->set_tile_region(dev, i); + for (j = 0; j < NVOBJ_ENGINE_NR; j++) { + if (dev_priv->eng[j] && dev_priv->eng[j]->set_tile_region) + dev_priv->eng[j]->set_tile_region(dev, i); + } + + pfifo->cache_pull(dev, true); + pfifo->reassign(dev, true); + spin_unlock_irqrestore(&dev_priv->context_switch_lock, save); +} + +static struct nouveau_tile_reg * +nv10_mem_get_tile_region(struct drm_device *dev, int i) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_tile_reg *tile = &dev_priv->tile.reg[i]; + + spin_lock(&dev_priv->tile.lock); + + if (!tile->used && + (!tile->fence || nouveau_fence_signalled(tile->fence))) + tile->used = true; + else + tile = NULL; + + spin_unlock(&dev_priv->tile.lock); + return tile; +} + +void +nv10_mem_put_tile_region(struct drm_device *dev, struct nouveau_tile_reg *tile, + struct nouveau_fence *fence) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + if (tile) { + spin_lock(&dev_priv->tile.lock); + if (fence) { + /* Mark it as pending. */ + tile->fence = fence; + nouveau_fence_ref(fence); + } + + tile->used = false; + spin_unlock(&dev_priv->tile.lock); + } +} + +struct nouveau_tile_reg * +nv10_mem_set_tiling(struct drm_device *dev, uint32_t addr, uint32_t size, + uint32_t pitch, uint32_t flags) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fb_engine *pfb = &dev_priv->engine.fb; + struct nouveau_tile_reg *tile, *found = NULL; + int i; + + for (i = 0; i < pfb->num_tiles; i++) { + tile = nv10_mem_get_tile_region(dev, i); + + if (pitch && !found) { + found = tile; + continue; + + } else if (tile && tile->pitch) { + /* Kill an unused tile region. */ + nv10_mem_update_tile_region(dev, tile, 0, 0, 0, 0); + } + + nv10_mem_put_tile_region(dev, tile, NULL); + } + + if (found) + nv10_mem_update_tile_region(dev, found, addr, size, + pitch, flags); + return found; +} + +/* + * Cleanup everything + */ +void +nouveau_mem_vram_fini(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + ttm_bo_device_release(&dev_priv->ttm.bdev); + + nouveau_ttm_global_release(dev_priv); + + if (dev_priv->fb_mtrr >= 0) { + drm_mtrr_del(dev_priv->fb_mtrr, + pci_resource_start(dev->pdev, 1), + pci_resource_len(dev->pdev, 1), DRM_MTRR_WC); + dev_priv->fb_mtrr = -1; + } +} + +void +nouveau_mem_gart_fini(struct drm_device *dev) +{ + nouveau_sgdma_takedown(dev); + + if (drm_core_has_AGP(dev) && dev->agp) { + struct drm_agp_mem *entry, *tempe; + + /* Remove AGP resources, but leave dev->agp + intact until drv_cleanup is called. */ + list_for_each_entry_safe(entry, tempe, &dev->agp->memory, head) { + if (entry->bound) + drm_unbind_agp(entry->memory); + drm_free_agp(entry->memory, entry->pages); + kfree(entry); + } + INIT_LIST_HEAD(&dev->agp->memory); + + if (dev->agp->acquired) + drm_agp_release(dev); + + dev->agp->acquired = 0; + dev->agp->enabled = 0; + } +} + +static uint32_t +nouveau_mem_detect_nv04(struct drm_device *dev) +{ + uint32_t boot0 = nv_rd32(dev, NV04_PFB_BOOT_0); + + if (boot0 & 0x00000100) + return (((boot0 >> 12) & 0xf) * 2 + 2) * 1024 * 1024; + + switch (boot0 & NV04_PFB_BOOT_0_RAM_AMOUNT) { + case NV04_PFB_BOOT_0_RAM_AMOUNT_32MB: + return 32 * 1024 * 1024; + case NV04_PFB_BOOT_0_RAM_AMOUNT_16MB: + return 16 * 1024 * 1024; + case NV04_PFB_BOOT_0_RAM_AMOUNT_8MB: + return 8 * 1024 * 1024; + case NV04_PFB_BOOT_0_RAM_AMOUNT_4MB: + return 4 * 1024 * 1024; + } + + return 0; +} + +static uint32_t +nouveau_mem_detect_nforce(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct pci_dev *bridge; + uint32_t mem; + + bridge = pci_get_bus_and_slot(0, PCI_DEVFN(0, 1)); + if (!bridge) { + NV_ERROR(dev, "no bridge device\n"); + return 0; + } + + if (dev_priv->flags & NV_NFORCE) { + pci_read_config_dword(bridge, 0x7C, &mem); + return (uint64_t)(((mem >> 6) & 31) + 1)*1024*1024; + } else + if (dev_priv->flags & NV_NFORCE2) { + pci_read_config_dword(bridge, 0x84, &mem); + return (uint64_t)(((mem >> 4) & 127) + 1)*1024*1024; + } + + NV_ERROR(dev, "impossible!\n"); + return 0; +} + +int +nouveau_mem_detect(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + if (dev_priv->card_type == NV_04) { + dev_priv->vram_size = nouveau_mem_detect_nv04(dev); + } else + if (dev_priv->flags & (NV_NFORCE | NV_NFORCE2)) { + dev_priv->vram_size = nouveau_mem_detect_nforce(dev); + } else + if (dev_priv->card_type < NV_50) { + dev_priv->vram_size = nv_rd32(dev, NV04_PFB_FIFO_DATA); + dev_priv->vram_size &= NV10_PFB_FIFO_DATA_RAM_AMOUNT_MB_MASK; + } + + if (dev_priv->vram_size) + return 0; + return -ENOMEM; +} + +bool +nouveau_mem_flags_valid(struct drm_device *dev, u32 tile_flags) +{ + if (!(tile_flags & NOUVEAU_GEM_TILE_LAYOUT_MASK)) + return true; + + return false; +} + +#if __OS_HAS_AGP +static unsigned long +get_agp_mode(struct drm_device *dev, unsigned long mode) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + /* + * FW seems to be broken on nv18, it makes the card lock up + * randomly. + */ + if (dev_priv->chipset == 0x18) + mode &= ~PCI_AGP_COMMAND_FW; + + /* + * AGP mode set in the command line. + */ + if (nouveau_agpmode > 0) { + bool agpv3 = mode & 0x8; + int rate = agpv3 ? nouveau_agpmode / 4 : nouveau_agpmode; + + mode = (mode & ~0x7) | (rate & 0x7); + } + + return mode; +} +#endif + +int +nouveau_mem_reset_agp(struct drm_device *dev) +{ +#if __OS_HAS_AGP + uint32_t saved_pci_nv_1, pmc_enable; + int ret; + + /* First of all, disable fast writes, otherwise if it's + * already enabled in the AGP bridge and we disable the card's + * AGP controller we might be locking ourselves out of it. */ + if ((nv_rd32(dev, NV04_PBUS_PCI_NV_19) | + dev->agp->mode) & PCI_AGP_COMMAND_FW) { + struct drm_agp_info info; + struct drm_agp_mode mode; + + ret = drm_agp_info(dev, &info); + if (ret) + return ret; + + mode.mode = get_agp_mode(dev, info.mode) & ~PCI_AGP_COMMAND_FW; + ret = drm_agp_enable(dev, mode); + if (ret) + return ret; + } + + saved_pci_nv_1 = nv_rd32(dev, NV04_PBUS_PCI_NV_1); + + /* clear busmaster bit */ + nv_wr32(dev, NV04_PBUS_PCI_NV_1, saved_pci_nv_1 & ~0x4); + /* disable AGP */ + nv_wr32(dev, NV04_PBUS_PCI_NV_19, 0); + + /* power cycle pgraph, if enabled */ + pmc_enable = nv_rd32(dev, NV03_PMC_ENABLE); + if (pmc_enable & NV_PMC_ENABLE_PGRAPH) { + nv_wr32(dev, NV03_PMC_ENABLE, + pmc_enable & ~NV_PMC_ENABLE_PGRAPH); + nv_wr32(dev, NV03_PMC_ENABLE, nv_rd32(dev, NV03_PMC_ENABLE) | + NV_PMC_ENABLE_PGRAPH); + } + + /* and restore (gives effect of resetting AGP) */ + nv_wr32(dev, NV04_PBUS_PCI_NV_1, saved_pci_nv_1); +#endif + + return 0; +} + +int +nouveau_mem_init_agp(struct drm_device *dev) +{ +#if __OS_HAS_AGP + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct drm_agp_info info; + struct drm_agp_mode mode; + int ret; + + if (!dev->agp->acquired) { + ret = drm_agp_acquire(dev); + if (ret) { + NV_ERROR(dev, "Unable to acquire AGP: %d\n", ret); + return ret; + } + } + + nouveau_mem_reset_agp(dev); + + ret = drm_agp_info(dev, &info); + if (ret) { + NV_ERROR(dev, "Unable to get AGP info: %d\n", ret); + return ret; + } + + /* see agp.h for the AGPSTAT_* modes available */ + mode.mode = get_agp_mode(dev, info.mode); + ret = drm_agp_enable(dev, mode); + if (ret) { + NV_ERROR(dev, "Unable to enable AGP: %d\n", ret); + return ret; + } + + dev_priv->gart_info.type = NOUVEAU_GART_AGP; + dev_priv->gart_info.aper_base = info.aperture_base; + dev_priv->gart_info.aper_size = info.aperture_size; +#endif + return 0; +} + +int +nouveau_mem_vram_init(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct ttm_bo_device *bdev = &dev_priv->ttm.bdev; + int ret, dma_bits; + + dma_bits = 32; + if (dev_priv->card_type >= NV_50) { + if (pci_dma_supported(dev->pdev, DMA_BIT_MASK(40))) + dma_bits = 40; + } else + if (0 && drm_pci_device_is_pcie(dev) && + dev_priv->chipset > 0x40 && + dev_priv->chipset != 0x45) { + if (pci_dma_supported(dev->pdev, DMA_BIT_MASK(39))) + dma_bits = 39; + } + + ret = pci_set_dma_mask(dev->pdev, DMA_BIT_MASK(dma_bits)); + if (ret) + return ret; + + dev_priv->fb_phys = pci_resource_start(dev->pdev, 1); + + ret = nouveau_ttm_global_init(dev_priv); + if (ret) + return ret; + + ret = ttm_bo_device_init(&dev_priv->ttm.bdev, + dev_priv->ttm.bo_global_ref.ref.object, + &nouveau_bo_driver, DRM_FILE_PAGE_OFFSET, + dma_bits <= 32 ? true : false); + if (ret) { + NV_ERROR(dev, "Error initialising bo driver: %d\n", ret); + return ret; + } + + /* reserve space at end of VRAM for PRAMIN */ + if (dev_priv->card_type >= NV_50) { + dev_priv->ramin_rsvd_vram = 1 * 1024 * 1024; + } else + if (dev_priv->card_type >= NV_40) { + u32 vs = hweight8((nv_rd32(dev, 0x001540) & 0x0000ff00) >> 8); + u32 rsvd; + + /* estimate grctx size, the magics come from nv40_grctx.c */ + if (dev_priv->chipset == 0x40) rsvd = 0x6aa0 * vs; + else if (dev_priv->chipset < 0x43) rsvd = 0x4f00 * vs; + else if (nv44_graph_class(dev)) rsvd = 0x4980 * vs; + else rsvd = 0x4a40 * vs; + rsvd += 16 * 1024; + rsvd *= dev_priv->engine.fifo.channels; + + /* pciegart table */ + if (drm_pci_device_is_pcie(dev)) + rsvd += 512 * 1024; + + /* object storage */ + rsvd += 512 * 1024; + + dev_priv->ramin_rsvd_vram = round_up(rsvd, 4096); + } else { + dev_priv->ramin_rsvd_vram = 512 * 1024; + } + + ret = dev_priv->engine.vram.init(dev); + if (ret) + return ret; + + NV_INFO(dev, "Detected %dMiB VRAM\n", (int)(dev_priv->vram_size >> 20)); + if (dev_priv->vram_sys_base) { + NV_INFO(dev, "Stolen system memory at: 0x%010llx\n", + dev_priv->vram_sys_base); + } + + dev_priv->fb_available_size = dev_priv->vram_size; + dev_priv->fb_mappable_pages = dev_priv->fb_available_size; + if (dev_priv->fb_mappable_pages > pci_resource_len(dev->pdev, 1)) + dev_priv->fb_mappable_pages = pci_resource_len(dev->pdev, 1); + dev_priv->fb_mappable_pages >>= PAGE_SHIFT; + + dev_priv->fb_available_size -= dev_priv->ramin_rsvd_vram; + dev_priv->fb_aper_free = dev_priv->fb_available_size; + + /* mappable vram */ + ret = ttm_bo_init_mm(bdev, TTM_PL_VRAM, + dev_priv->fb_available_size >> PAGE_SHIFT); + if (ret) { + NV_ERROR(dev, "Failed VRAM mm init: %d\n", ret); + return ret; + } + + if (dev_priv->card_type < NV_50) { + ret = nouveau_bo_new(dev, NULL, 256*1024, 0, TTM_PL_FLAG_VRAM, + 0, 0, &dev_priv->vga_ram); + if (ret == 0) + ret = nouveau_bo_pin(dev_priv->vga_ram, + TTM_PL_FLAG_VRAM); + + if (ret) { + NV_WARN(dev, "failed to reserve VGA memory\n"); + nouveau_bo_ref(NULL, &dev_priv->vga_ram); + } + } + + dev_priv->fb_mtrr = drm_mtrr_add(pci_resource_start(dev->pdev, 1), + pci_resource_len(dev->pdev, 1), + DRM_MTRR_WC); + return 0; +} + +int +nouveau_mem_gart_init(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct ttm_bo_device *bdev = &dev_priv->ttm.bdev; + int ret; + + dev_priv->gart_info.type = NOUVEAU_GART_NONE; + +#if !defined(__powerpc__) && !defined(__ia64__) + if (drm_pci_device_is_agp(dev) && dev->agp && nouveau_agpmode) { + ret = nouveau_mem_init_agp(dev); + if (ret) + NV_ERROR(dev, "Error initialising AGP: %d\n", ret); + } +#endif + + if (dev_priv->gart_info.type == NOUVEAU_GART_NONE) { + ret = nouveau_sgdma_init(dev); + if (ret) { + NV_ERROR(dev, "Error initialising PCI(E): %d\n", ret); + return ret; + } + } + + NV_INFO(dev, "%d MiB GART (aperture)\n", + (int)(dev_priv->gart_info.aper_size >> 20)); + dev_priv->gart_info.aper_free = dev_priv->gart_info.aper_size; + + ret = ttm_bo_init_mm(bdev, TTM_PL_TT, + dev_priv->gart_info.aper_size >> PAGE_SHIFT); + if (ret) { + NV_ERROR(dev, "Failed TT mm init: %d\n", ret); + return ret; + } + + return 0; +} + +void +nouveau_mem_timing_init(struct drm_device *dev) +{ + /* cards < NVC0 only */ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_pm_engine *pm = &dev_priv->engine.pm; + struct nouveau_pm_memtimings *memtimings = &pm->memtimings; + struct nvbios *bios = &dev_priv->vbios; + struct bit_entry P; + u8 tUNK_0, tUNK_1, tUNK_2; + u8 tRP; /* Byte 3 */ + u8 tRAS; /* Byte 5 */ + u8 tRFC; /* Byte 7 */ + u8 tRC; /* Byte 9 */ + u8 tUNK_10, tUNK_11, tUNK_12, tUNK_13, tUNK_14; + u8 tUNK_18, tUNK_19, tUNK_20, tUNK_21; + u8 magic_number = 0; /* Yeah... sorry*/ + u8 *mem = NULL, *entry; + int i, recordlen, entries; + + if (bios->type == NVBIOS_BIT) { + if (bit_table(dev, 'P', &P)) + return; + + if (P.version == 1) + mem = ROMPTR(bios, P.data[4]); + else + if (P.version == 2) + mem = ROMPTR(bios, P.data[8]); + else { + NV_WARN(dev, "unknown mem for BIT P %d\n", P.version); + } + } else { + NV_DEBUG(dev, "BMP version too old for memory\n"); + return; + } + + if (!mem) { + NV_DEBUG(dev, "memory timing table pointer invalid\n"); + return; + } + + if (mem[0] != 0x10) { + NV_WARN(dev, "memory timing table 0x%02x unknown\n", mem[0]); + return; + } + + /* validate record length */ + entries = mem[2]; + recordlen = mem[3]; + if (recordlen < 15) { + NV_ERROR(dev, "mem timing table length unknown: %d\n", mem[3]); + return; + } + + /* parse vbios entries into common format */ + memtimings->timing = + kcalloc(entries, sizeof(*memtimings->timing), GFP_KERNEL); + if (!memtimings->timing) + return; + + /* Get "some number" from the timing reg for NV_40 and NV_50 + * Used in calculations later */ + if (dev_priv->card_type >= NV_40 && dev_priv->chipset < 0x98) { + magic_number = (nv_rd32(dev, 0x100228) & 0x0f000000) >> 24; + } + + entry = mem + mem[1]; + for (i = 0; i < entries; i++, entry += recordlen) { + struct nouveau_pm_memtiming *timing = &pm->memtimings.timing[i]; + if (entry[0] == 0) + continue; + + tUNK_18 = 1; + tUNK_19 = 1; + tUNK_20 = 0; + tUNK_21 = 0; + switch (min(recordlen, 22)) { + case 22: + tUNK_21 = entry[21]; + case 21: + tUNK_20 = entry[20]; + case 20: + tUNK_19 = entry[19]; + case 19: + tUNK_18 = entry[18]; + default: + tUNK_0 = entry[0]; + tUNK_1 = entry[1]; + tUNK_2 = entry[2]; + tRP = entry[3]; + tRAS = entry[5]; + tRFC = entry[7]; + tRC = entry[9]; + tUNK_10 = entry[10]; + tUNK_11 = entry[11]; + tUNK_12 = entry[12]; + tUNK_13 = entry[13]; + tUNK_14 = entry[14]; + break; + } + + timing->reg_100220 = (tRC << 24 | tRFC << 16 | tRAS << 8 | tRP); + + /* XXX: I don't trust the -1's and +1's... they must come + * from somewhere! */ + timing->reg_100224 = (tUNK_0 + tUNK_19 + 1 + magic_number) << 24 | + max(tUNK_18, (u8) 1) << 16 | + (tUNK_1 + tUNK_19 + 1 + magic_number) << 8; + if (dev_priv->chipset == 0xa8) { + timing->reg_100224 |= (tUNK_2 - 1); + } else { + timing->reg_100224 |= (tUNK_2 + 2 - magic_number); + } + + timing->reg_100228 = (tUNK_12 << 16 | tUNK_11 << 8 | tUNK_10); + if (dev_priv->chipset >= 0xa3 && dev_priv->chipset < 0xaa) + timing->reg_100228 |= (tUNK_19 - 1) << 24; + else + timing->reg_100228 |= magic_number << 24; + + if (dev_priv->card_type == NV_40) { + /* NV40: don't know what the rest of the regs are.. + * And don't need to know either */ + timing->reg_100228 |= 0x20200000; + } else if (dev_priv->card_type >= NV_50) { + if (dev_priv->chipset < 0x98 || + (dev_priv->chipset == 0x98 && + dev_priv->stepping <= 0xa1)) { + timing->reg_10022c = (0x14 + tUNK_2) << 24 | + 0x16 << 16 | + (tUNK_2 - 1) << 8 | + (tUNK_2 - 1); + } else { + /* XXX: reg_10022c for recentish cards */ + timing->reg_10022c = tUNK_2 - 1; + } + + timing->reg_100230 = (tUNK_20 << 24 | tUNK_21 << 16 | + tUNK_13 << 8 | tUNK_13); + + timing->reg_100234 = (tRAS << 24 | tRC); + timing->reg_100234 += max(tUNK_10, tUNK_11) << 16; + + if (dev_priv->chipset < 0x98 || + (dev_priv->chipset == 0x98 && + dev_priv->stepping <= 0xa1)) { + timing->reg_100234 |= (tUNK_2 + 2) << 8; + } else { + /* XXX: +6? */ + timing->reg_100234 |= (tUNK_19 + 6) << 8; + } + + /* XXX; reg_100238 + * reg_100238: 0x00?????? */ + timing->reg_10023c = 0x202; + if (dev_priv->chipset < 0x98 || + (dev_priv->chipset == 0x98 && + dev_priv->stepping <= 0xa1)) { + timing->reg_10023c |= 0x4000000 | (tUNK_2 - 1) << 16; + } else { + /* XXX: reg_10023c + * currently unknown + * 10023c seen as 06xxxxxx, 0bxxxxxx or 0fxxxxxx */ + } + + /* XXX: reg_100240? */ + } + timing->id = i; + + NV_DEBUG(dev, "Entry %d: 220: %08x %08x %08x %08x\n", i, + timing->reg_100220, timing->reg_100224, + timing->reg_100228, timing->reg_10022c); + NV_DEBUG(dev, " 230: %08x %08x %08x %08x\n", + timing->reg_100230, timing->reg_100234, + timing->reg_100238, timing->reg_10023c); + NV_DEBUG(dev, " 240: %08x\n", timing->reg_100240); + } + + memtimings->nr_timing = entries; + memtimings->supported = (dev_priv->chipset <= 0x98); +} + +void +nouveau_mem_timing_fini(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_pm_memtimings *mem = &dev_priv->engine.pm.memtimings; + + kfree(mem->timing); +} + +static int +nouveau_vram_manager_init(struct ttm_mem_type_manager *man, unsigned long p_size) +{ + struct drm_nouveau_private *dev_priv = nouveau_bdev(man->bdev); + struct nouveau_mm *mm; + u64 size, block, rsvd; + int ret; + + rsvd = (256 * 1024); /* vga memory */ + size = (p_size << PAGE_SHIFT) - rsvd; + block = dev_priv->vram_rblock_size; + + ret = nouveau_mm_init(&mm, rsvd >> 12, size >> 12, block >> 12); + if (ret) + return ret; + + man->priv = mm; + return 0; +} + +static int +nouveau_vram_manager_fini(struct ttm_mem_type_manager *man) +{ + struct nouveau_mm *mm = man->priv; + int ret; + + ret = nouveau_mm_fini(&mm); + if (ret) + return ret; + + man->priv = NULL; + return 0; +} + +static void +nouveau_vram_manager_del(struct ttm_mem_type_manager *man, + struct ttm_mem_reg *mem) +{ + struct drm_nouveau_private *dev_priv = nouveau_bdev(man->bdev); + struct nouveau_vram_engine *vram = &dev_priv->engine.vram; + struct nouveau_mem *node = mem->mm_node; + struct drm_device *dev = dev_priv->dev; + + if (node->tmp_vma.node) { + nouveau_vm_unmap(&node->tmp_vma); + nouveau_vm_put(&node->tmp_vma); + } + + vram->put(dev, (struct nouveau_mem **)&mem->mm_node); +} + +static int +nouveau_vram_manager_new(struct ttm_mem_type_manager *man, + struct ttm_buffer_object *bo, + struct ttm_placement *placement, + struct ttm_mem_reg *mem) +{ + struct drm_nouveau_private *dev_priv = nouveau_bdev(man->bdev); + struct nouveau_vram_engine *vram = &dev_priv->engine.vram; + struct drm_device *dev = dev_priv->dev; + struct nouveau_bo *nvbo = nouveau_bo(bo); + struct nouveau_mem *node; + u32 size_nc = 0; + int ret; + + if (nvbo->tile_flags & NOUVEAU_GEM_TILE_NONCONTIG) + size_nc = 1 << nvbo->vma.node->type; + + ret = vram->get(dev, mem->num_pages << PAGE_SHIFT, + mem->page_alignment << PAGE_SHIFT, size_nc, + (nvbo->tile_flags >> 8) & 0x3ff, &node); + if (ret) { + mem->mm_node = NULL; + return (ret == -ENOSPC) ? 0 : ret; + } + + node->page_shift = 12; + if (nvbo->vma.node) + node->page_shift = nvbo->vma.node->type; + + mem->mm_node = node; + mem->start = node->offset >> PAGE_SHIFT; + return 0; +} + +void +nouveau_vram_manager_debug(struct ttm_mem_type_manager *man, const char *prefix) +{ + struct nouveau_mm *mm = man->priv; + struct nouveau_mm_node *r; + u32 total = 0, free = 0; + + mutex_lock(&mm->mutex); + list_for_each_entry(r, &mm->nodes, nl_entry) { + printk(KERN_DEBUG "%s %d: 0x%010llx 0x%010llx\n", + prefix, r->type, ((u64)r->offset << 12), + (((u64)r->offset + r->length) << 12)); + + total += r->length; + if (!r->type) + free += r->length; + } + mutex_unlock(&mm->mutex); + + printk(KERN_DEBUG "%s total: 0x%010llx free: 0x%010llx\n", + prefix, (u64)total << 12, (u64)free << 12); + printk(KERN_DEBUG "%s block: 0x%08x\n", + prefix, mm->block_size << 12); +} + +const struct ttm_mem_type_manager_func nouveau_vram_manager = { + nouveau_vram_manager_init, + nouveau_vram_manager_fini, + nouveau_vram_manager_new, + nouveau_vram_manager_del, + nouveau_vram_manager_debug +}; + +static int +nouveau_gart_manager_init(struct ttm_mem_type_manager *man, unsigned long psize) +{ + return 0; +} + +static int +nouveau_gart_manager_fini(struct ttm_mem_type_manager *man) +{ + return 0; +} + +static void +nouveau_gart_manager_del(struct ttm_mem_type_manager *man, + struct ttm_mem_reg *mem) +{ + struct nouveau_mem *node = mem->mm_node; + + if (node->tmp_vma.node) { + nouveau_vm_unmap(&node->tmp_vma); + nouveau_vm_put(&node->tmp_vma); + } + + mem->mm_node = NULL; + kfree(node); +} + +static int +nouveau_gart_manager_new(struct ttm_mem_type_manager *man, + struct ttm_buffer_object *bo, + struct ttm_placement *placement, + struct ttm_mem_reg *mem) +{ + struct drm_nouveau_private *dev_priv = nouveau_bdev(bo->bdev); + struct nouveau_bo *nvbo = nouveau_bo(bo); + struct nouveau_vma *vma = &nvbo->vma; + struct nouveau_vm *vm = vma->vm; + struct nouveau_mem *node; + int ret; + + if (unlikely((mem->num_pages << PAGE_SHIFT) >= + dev_priv->gart_info.aper_size)) + return -ENOMEM; + + node = kzalloc(sizeof(*node), GFP_KERNEL); + if (!node) + return -ENOMEM; + + /* This node must be for evicting large-paged VRAM + * to system memory. Due to a nv50 limitation of + * not being able to mix large/small pages within + * the same PDE, we need to create a temporary + * small-paged VMA for the eviction. + */ + if (vma->node->type != vm->spg_shift) { + ret = nouveau_vm_get(vm, (u64)vma->node->length << 12, + vm->spg_shift, NV_MEM_ACCESS_RW, + &node->tmp_vma); + if (ret) { + kfree(node); + return ret; + } + } + + node->page_shift = nvbo->vma.node->type; + mem->mm_node = node; + mem->start = 0; + return 0; +} + +void +nouveau_gart_manager_debug(struct ttm_mem_type_manager *man, const char *prefix) +{ +} + +const struct ttm_mem_type_manager_func nouveau_gart_manager = { + nouveau_gart_manager_init, + nouveau_gart_manager_fini, + nouveau_gart_manager_new, + nouveau_gart_manager_del, + nouveau_gart_manager_debug +}; diff --git a/drivers/gpu/drm/nouveau/nouveau_mm.c b/drivers/gpu/drm/nouveau/nouveau_mm.c new file mode 100644 index 00000000..7609756b --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_mm.c @@ -0,0 +1,171 @@ +/* + * Copyright 2010 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include "drmP.h" +#include "nouveau_drv.h" +#include "nouveau_mm.h" + +static inline void +region_put(struct nouveau_mm *rmm, struct nouveau_mm_node *a) +{ + list_del(&a->nl_entry); + list_del(&a->fl_entry); + kfree(a); +} + +static struct nouveau_mm_node * +region_split(struct nouveau_mm *rmm, struct nouveau_mm_node *a, u32 size) +{ + struct nouveau_mm_node *b; + + if (a->length == size) + return a; + + b = kmalloc(sizeof(*b), GFP_KERNEL); + if (unlikely(b == NULL)) + return NULL; + + b->offset = a->offset; + b->length = size; + b->type = a->type; + a->offset += size; + a->length -= size; + list_add_tail(&b->nl_entry, &a->nl_entry); + if (b->type == 0) + list_add_tail(&b->fl_entry, &a->fl_entry); + return b; +} + +#define node(root, dir) ((root)->nl_entry.dir == &rmm->nodes) ? NULL : \ + list_entry((root)->nl_entry.dir, struct nouveau_mm_node, nl_entry) + +void +nouveau_mm_put(struct nouveau_mm *rmm, struct nouveau_mm_node *this) +{ + struct nouveau_mm_node *prev = node(this, prev); + struct nouveau_mm_node *next = node(this, next); + + list_add(&this->fl_entry, &rmm->free); + this->type = 0; + + if (prev && prev->type == 0) { + prev->length += this->length; + region_put(rmm, this); + this = prev; + } + + if (next && next->type == 0) { + next->offset = this->offset; + next->length += this->length; + region_put(rmm, this); + } +} + +int +nouveau_mm_get(struct nouveau_mm *rmm, int type, u32 size, u32 size_nc, + u32 align, struct nouveau_mm_node **pnode) +{ + struct nouveau_mm_node *prev, *this, *next; + u32 min = size_nc ? size_nc : size; + u32 align_mask = align - 1; + u32 splitoff; + u32 s, e; + + list_for_each_entry(this, &rmm->free, fl_entry) { + e = this->offset + this->length; + s = this->offset; + + prev = node(this, prev); + if (prev && prev->type != type) + s = roundup(s, rmm->block_size); + + next = node(this, next); + if (next && next->type != type) + e = rounddown(e, rmm->block_size); + + s = (s + align_mask) & ~align_mask; + e &= ~align_mask; + if (s > e || e - s < min) + continue; + + splitoff = s - this->offset; + if (splitoff && !region_split(rmm, this, splitoff)) + return -ENOMEM; + + this = region_split(rmm, this, min(size, e - s)); + if (!this) + return -ENOMEM; + + this->type = type; + list_del(&this->fl_entry); + *pnode = this; + return 0; + } + + return -ENOSPC; +} + +int +nouveau_mm_init(struct nouveau_mm **prmm, u32 offset, u32 length, u32 block) +{ + struct nouveau_mm *rmm; + struct nouveau_mm_node *heap; + + heap = kzalloc(sizeof(*heap), GFP_KERNEL); + if (!heap) + return -ENOMEM; + heap->offset = roundup(offset, block); + heap->length = rounddown(offset + length, block) - heap->offset; + + rmm = kzalloc(sizeof(*rmm), GFP_KERNEL); + if (!rmm) { + kfree(heap); + return -ENOMEM; + } + rmm->block_size = block; + mutex_init(&rmm->mutex); + INIT_LIST_HEAD(&rmm->nodes); + INIT_LIST_HEAD(&rmm->free); + list_add(&heap->nl_entry, &rmm->nodes); + list_add(&heap->fl_entry, &rmm->free); + + *prmm = rmm; + return 0; +} + +int +nouveau_mm_fini(struct nouveau_mm **prmm) +{ + struct nouveau_mm *rmm = *prmm; + struct nouveau_mm_node *heap = + list_first_entry(&rmm->nodes, struct nouveau_mm_node, nl_entry); + + if (!list_is_singular(&rmm->nodes)) + return -EBUSY; + + kfree(heap); + kfree(rmm); + *prmm = NULL; + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nouveau_mm.h b/drivers/gpu/drm/nouveau/nouveau_mm.h new file mode 100644 index 00000000..1f7483aa --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_mm.h @@ -0,0 +1,65 @@ +/* + * Copyright 2010 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#ifndef __NOUVEAU_REGION_H__ +#define __NOUVEAU_REGION_H__ + +struct nouveau_mm_node { + struct list_head nl_entry; + struct list_head fl_entry; + struct list_head rl_entry; + + u8 type; + u32 offset; + u32 length; +}; + +struct nouveau_mm { + struct list_head nodes; + struct list_head free; + + struct mutex mutex; + + u32 block_size; +}; + +int nouveau_mm_init(struct nouveau_mm **, u32 offset, u32 length, u32 block); +int nouveau_mm_fini(struct nouveau_mm **); +int nouveau_mm_pre(struct nouveau_mm *); +int nouveau_mm_get(struct nouveau_mm *, int type, u32 size, u32 size_nc, + u32 align, struct nouveau_mm_node **); +void nouveau_mm_put(struct nouveau_mm *, struct nouveau_mm_node *); + +int nv50_vram_init(struct drm_device *); +int nv50_vram_new(struct drm_device *, u64 size, u32 align, u32 size_nc, + u32 memtype, struct nouveau_mem **); +void nv50_vram_del(struct drm_device *, struct nouveau_mem **); +bool nv50_vram_flags_valid(struct drm_device *, u32 tile_flags); + +int nvc0_vram_init(struct drm_device *); +int nvc0_vram_new(struct drm_device *, u64 size, u32 align, u32 ncmin, + u32 memtype, struct nouveau_mem **); +bool nvc0_vram_flags_valid(struct drm_device *, u32 tile_flags); + +#endif diff --git a/drivers/gpu/drm/nouveau/nouveau_notifier.c b/drivers/gpu/drm/nouveau/nouveau_notifier.c new file mode 100644 index 00000000..5b39718a --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_notifier.c @@ -0,0 +1,194 @@ +/* + * Copyright (C) 2007 Ben Skeggs. + * + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "drmP.h" +#include "drm.h" +#include "nouveau_drv.h" +#include "nouveau_ramht.h" + +int +nouveau_notifier_init_channel(struct nouveau_channel *chan) +{ + struct drm_device *dev = chan->dev; + struct nouveau_bo *ntfy = NULL; + uint32_t flags, ttmpl; + int ret; + + if (nouveau_vram_notify) { + flags = NOUVEAU_GEM_DOMAIN_VRAM; + ttmpl = TTM_PL_FLAG_VRAM; + } else { + flags = NOUVEAU_GEM_DOMAIN_GART; + ttmpl = TTM_PL_FLAG_TT; + } + + ret = nouveau_gem_new(dev, NULL, PAGE_SIZE, 0, flags, 0, 0, &ntfy); + if (ret) + return ret; + + ret = nouveau_bo_pin(ntfy, ttmpl); + if (ret) + goto out_err; + + ret = nouveau_bo_map(ntfy); + if (ret) + goto out_err; + + ret = drm_mm_init(&chan->notifier_heap, 0, ntfy->bo.mem.size); + if (ret) + goto out_err; + + chan->notifier_bo = ntfy; +out_err: + if (ret) + drm_gem_object_unreference_unlocked(ntfy->gem); + + return ret; +} + +void +nouveau_notifier_takedown_channel(struct nouveau_channel *chan) +{ + struct drm_device *dev = chan->dev; + + if (!chan->notifier_bo) + return; + + nouveau_bo_unmap(chan->notifier_bo); + mutex_lock(&dev->struct_mutex); + nouveau_bo_unpin(chan->notifier_bo); + mutex_unlock(&dev->struct_mutex); + drm_gem_object_unreference_unlocked(chan->notifier_bo->gem); + drm_mm_takedown(&chan->notifier_heap); +} + +static void +nouveau_notifier_gpuobj_dtor(struct drm_device *dev, + struct nouveau_gpuobj *gpuobj) +{ + NV_DEBUG(dev, "\n"); + + if (gpuobj->priv) + drm_mm_put_block(gpuobj->priv); +} + +int +nouveau_notifier_alloc(struct nouveau_channel *chan, uint32_t handle, + int size, uint32_t start, uint32_t end, + uint32_t *b_offset) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *nobj = NULL; + struct drm_mm_node *mem; + uint32_t offset; + int target, ret; + + mem = drm_mm_search_free_in_range(&chan->notifier_heap, size, 0, + start, end, 0); + if (mem) + mem = drm_mm_get_block_range(mem, size, 0, start, end); + if (!mem) { + NV_ERROR(dev, "Channel %d notifier block full\n", chan->id); + return -ENOMEM; + } + + if (dev_priv->card_type < NV_50) { + if (chan->notifier_bo->bo.mem.mem_type == TTM_PL_VRAM) + target = NV_MEM_TARGET_VRAM; + else + target = NV_MEM_TARGET_GART; + offset = chan->notifier_bo->bo.mem.start << PAGE_SHIFT; + } else { + target = NV_MEM_TARGET_VM; + offset = chan->notifier_bo->vma.offset; + } + offset += mem->start; + + ret = nouveau_gpuobj_dma_new(chan, NV_CLASS_DMA_IN_MEMORY, offset, + mem->size, NV_MEM_ACCESS_RW, target, + &nobj); + if (ret) { + drm_mm_put_block(mem); + NV_ERROR(dev, "Error creating notifier ctxdma: %d\n", ret); + return ret; + } + nobj->dtor = nouveau_notifier_gpuobj_dtor; + nobj->priv = mem; + + ret = nouveau_ramht_insert(chan, handle, nobj); + nouveau_gpuobj_ref(NULL, &nobj); + if (ret) { + drm_mm_put_block(mem); + NV_ERROR(dev, "Error adding notifier to ramht: %d\n", ret); + return ret; + } + + *b_offset = mem->start; + return 0; +} + +int +nouveau_notifier_offset(struct nouveau_gpuobj *nobj, uint32_t *poffset) +{ + if (!nobj || nobj->dtor != nouveau_notifier_gpuobj_dtor) + return -EINVAL; + + if (poffset) { + struct drm_mm_node *mem = nobj->priv; + + if (*poffset >= mem->size) + return false; + + *poffset += mem->start; + } + + return 0; +} + +int +nouveau_ioctl_notifier_alloc(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct drm_nouveau_notifierobj_alloc *na = data; + struct nouveau_channel *chan; + int ret; + + /* completely unnecessary for these chipsets... */ + if (unlikely(dev_priv->card_type >= NV_C0)) + return -EINVAL; + + chan = nouveau_channel_get(dev, file_priv, na->channel); + if (IS_ERR(chan)) + return PTR_ERR(chan); + + ret = nouveau_notifier_alloc(chan, na->handle, na->size, 0, 0x1000, + &na->offset); + nouveau_channel_put(&chan); + return ret; +} diff --git a/drivers/gpu/drm/nouveau/nouveau_object.c b/drivers/gpu/drm/nouveau/nouveau_object.c new file mode 100644 index 00000000..8f97016f --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_object.c @@ -0,0 +1,998 @@ +/* + * Copyright (C) 2006 Ben Skeggs. + * + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +/* + * Authors: + * Ben Skeggs + */ + +#include "drmP.h" +#include "drm.h" +#include "nouveau_drv.h" +#include "nouveau_drm.h" +#include "nouveau_ramht.h" +#include "nouveau_vm.h" +#include "nv50_display.h" + +struct nouveau_gpuobj_method { + struct list_head head; + u32 mthd; + int (*exec)(struct nouveau_channel *, u32 class, u32 mthd, u32 data); +}; + +struct nouveau_gpuobj_class { + struct list_head head; + struct list_head methods; + u32 id; + u32 engine; +}; + +int +nouveau_gpuobj_class_new(struct drm_device *dev, u32 class, u32 engine) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj_class *oc; + + oc = kzalloc(sizeof(*oc), GFP_KERNEL); + if (!oc) + return -ENOMEM; + + INIT_LIST_HEAD(&oc->methods); + oc->id = class; + oc->engine = engine; + list_add(&oc->head, &dev_priv->classes); + return 0; +} + +int +nouveau_gpuobj_mthd_new(struct drm_device *dev, u32 class, u32 mthd, + int (*exec)(struct nouveau_channel *, u32, u32, u32)) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj_method *om; + struct nouveau_gpuobj_class *oc; + + list_for_each_entry(oc, &dev_priv->classes, head) { + if (oc->id == class) + goto found; + } + + return -EINVAL; + +found: + om = kzalloc(sizeof(*om), GFP_KERNEL); + if (!om) + return -ENOMEM; + + om->mthd = mthd; + om->exec = exec; + list_add(&om->head, &oc->methods); + return 0; +} + +int +nouveau_gpuobj_mthd_call(struct nouveau_channel *chan, + u32 class, u32 mthd, u32 data) +{ + struct drm_nouveau_private *dev_priv = chan->dev->dev_private; + struct nouveau_gpuobj_method *om; + struct nouveau_gpuobj_class *oc; + + list_for_each_entry(oc, &dev_priv->classes, head) { + if (oc->id != class) + continue; + + list_for_each_entry(om, &oc->methods, head) { + if (om->mthd == mthd) + return om->exec(chan, class, mthd, data); + } + } + + return -ENOENT; +} + +int +nouveau_gpuobj_mthd_call2(struct drm_device *dev, int chid, + u32 class, u32 mthd, u32 data) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_channel *chan = NULL; + unsigned long flags; + int ret = -EINVAL; + + spin_lock_irqsave(&dev_priv->channels.lock, flags); + if (chid > 0 && chid < dev_priv->engine.fifo.channels) + chan = dev_priv->channels.ptr[chid]; + if (chan) + ret = nouveau_gpuobj_mthd_call(chan, class, mthd, data); + spin_unlock_irqrestore(&dev_priv->channels.lock, flags); + return ret; +} + +/* NVidia uses context objects to drive drawing operations. + + Context objects can be selected into 8 subchannels in the FIFO, + and then used via DMA command buffers. + + A context object is referenced by a user defined handle (CARD32). The HW + looks up graphics objects in a hash table in the instance RAM. + + An entry in the hash table consists of 2 CARD32. The first CARD32 contains + the handle, the second one a bitfield, that contains the address of the + object in instance RAM. + + The format of the second CARD32 seems to be: + + NV4 to NV30: + + 15: 0 instance_addr >> 4 + 17:16 engine (here uses 1 = graphics) + 28:24 channel id (here uses 0) + 31 valid (use 1) + + NV40: + + 15: 0 instance_addr >> 4 (maybe 19-0) + 21:20 engine (here uses 1 = graphics) + I'm unsure about the other bits, but using 0 seems to work. + + The key into the hash table depends on the object handle and channel id and + is given as: +*/ + +int +nouveau_gpuobj_new(struct drm_device *dev, struct nouveau_channel *chan, + uint32_t size, int align, uint32_t flags, + struct nouveau_gpuobj **gpuobj_ret) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_instmem_engine *instmem = &dev_priv->engine.instmem; + struct nouveau_gpuobj *gpuobj; + struct drm_mm_node *ramin = NULL; + int ret, i; + + NV_DEBUG(dev, "ch%d size=%u align=%d flags=0x%08x\n", + chan ? chan->id : -1, size, align, flags); + + gpuobj = kzalloc(sizeof(*gpuobj), GFP_KERNEL); + if (!gpuobj) + return -ENOMEM; + NV_DEBUG(dev, "gpuobj %p\n", gpuobj); + gpuobj->dev = dev; + gpuobj->flags = flags; + kref_init(&gpuobj->refcount); + gpuobj->size = size; + + spin_lock(&dev_priv->ramin_lock); + list_add_tail(&gpuobj->list, &dev_priv->gpuobj_list); + spin_unlock(&dev_priv->ramin_lock); + + if (chan) { + ramin = drm_mm_search_free(&chan->ramin_heap, size, align, 0); + if (ramin) + ramin = drm_mm_get_block(ramin, size, align); + if (!ramin) { + nouveau_gpuobj_ref(NULL, &gpuobj); + return -ENOMEM; + } + + gpuobj->pinst = chan->ramin->pinst; + if (gpuobj->pinst != ~0) + gpuobj->pinst += ramin->start; + + gpuobj->cinst = ramin->start; + gpuobj->vinst = ramin->start + chan->ramin->vinst; + gpuobj->node = ramin; + } else { + ret = instmem->get(gpuobj, size, align); + if (ret) { + nouveau_gpuobj_ref(NULL, &gpuobj); + return ret; + } + + ret = -ENOSYS; + if (!(flags & NVOBJ_FLAG_DONT_MAP)) + ret = instmem->map(gpuobj); + if (ret) + gpuobj->pinst = ~0; + + gpuobj->cinst = NVOBJ_CINST_GLOBAL; + } + + if (gpuobj->flags & NVOBJ_FLAG_ZERO_ALLOC) { + for (i = 0; i < gpuobj->size; i += 4) + nv_wo32(gpuobj, i, 0); + instmem->flush(dev); + } + + + *gpuobj_ret = gpuobj; + return 0; +} + +int +nouveau_gpuobj_init(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + NV_DEBUG(dev, "\n"); + + INIT_LIST_HEAD(&dev_priv->gpuobj_list); + INIT_LIST_HEAD(&dev_priv->classes); + spin_lock_init(&dev_priv->ramin_lock); + dev_priv->ramin_base = ~0; + + return 0; +} + +void +nouveau_gpuobj_takedown(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj_method *om, *tm; + struct nouveau_gpuobj_class *oc, *tc; + + NV_DEBUG(dev, "\n"); + + list_for_each_entry_safe(oc, tc, &dev_priv->classes, head) { + list_for_each_entry_safe(om, tm, &oc->methods, head) { + list_del(&om->head); + kfree(om); + } + list_del(&oc->head); + kfree(oc); + } + + BUG_ON(!list_empty(&dev_priv->gpuobj_list)); +} + + +static void +nouveau_gpuobj_del(struct kref *ref) +{ + struct nouveau_gpuobj *gpuobj = + container_of(ref, struct nouveau_gpuobj, refcount); + struct drm_device *dev = gpuobj->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_instmem_engine *instmem = &dev_priv->engine.instmem; + int i; + + NV_DEBUG(dev, "gpuobj %p\n", gpuobj); + + if (gpuobj->node && (gpuobj->flags & NVOBJ_FLAG_ZERO_FREE)) { + for (i = 0; i < gpuobj->size; i += 4) + nv_wo32(gpuobj, i, 0); + instmem->flush(dev); + } + + if (gpuobj->dtor) + gpuobj->dtor(dev, gpuobj); + + if (gpuobj->cinst == NVOBJ_CINST_GLOBAL) { + if (gpuobj->node) { + instmem->unmap(gpuobj); + instmem->put(gpuobj); + } + } else { + if (gpuobj->node) { + spin_lock(&dev_priv->ramin_lock); + drm_mm_put_block(gpuobj->node); + spin_unlock(&dev_priv->ramin_lock); + } + } + + spin_lock(&dev_priv->ramin_lock); + list_del(&gpuobj->list); + spin_unlock(&dev_priv->ramin_lock); + + kfree(gpuobj); +} + +void +nouveau_gpuobj_ref(struct nouveau_gpuobj *ref, struct nouveau_gpuobj **ptr) +{ + if (ref) + kref_get(&ref->refcount); + + if (*ptr) + kref_put(&(*ptr)->refcount, nouveau_gpuobj_del); + + *ptr = ref; +} + +int +nouveau_gpuobj_new_fake(struct drm_device *dev, u32 pinst, u64 vinst, + u32 size, u32 flags, struct nouveau_gpuobj **pgpuobj) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *gpuobj = NULL; + int i; + + NV_DEBUG(dev, + "pinst=0x%08x vinst=0x%010llx size=0x%08x flags=0x%08x\n", + pinst, vinst, size, flags); + + gpuobj = kzalloc(sizeof(*gpuobj), GFP_KERNEL); + if (!gpuobj) + return -ENOMEM; + NV_DEBUG(dev, "gpuobj %p\n", gpuobj); + gpuobj->dev = dev; + gpuobj->flags = flags; + kref_init(&gpuobj->refcount); + gpuobj->size = size; + gpuobj->pinst = pinst; + gpuobj->cinst = NVOBJ_CINST_GLOBAL; + gpuobj->vinst = vinst; + + if (gpuobj->flags & NVOBJ_FLAG_ZERO_ALLOC) { + for (i = 0; i < gpuobj->size; i += 4) + nv_wo32(gpuobj, i, 0); + dev_priv->engine.instmem.flush(dev); + } + + spin_lock(&dev_priv->ramin_lock); + list_add_tail(&gpuobj->list, &dev_priv->gpuobj_list); + spin_unlock(&dev_priv->ramin_lock); + *pgpuobj = gpuobj; + return 0; +} + +/* + DMA objects are used to reference a piece of memory in the + framebuffer, PCI or AGP address space. Each object is 16 bytes big + and looks as follows: + + entry[0] + 11:0 class (seems like I can always use 0 here) + 12 page table present? + 13 page entry linear? + 15:14 access: 0 rw, 1 ro, 2 wo + 17:16 target: 0 NV memory, 1 NV memory tiled, 2 PCI, 3 AGP + 31:20 dma adjust (bits 0-11 of the address) + entry[1] + dma limit (size of transfer) + entry[X] + 1 0 readonly, 1 readwrite + 31:12 dma frame address of the page (bits 12-31 of the address) + entry[N] + page table terminator, same value as the first pte, as does nvidia + rivatv uses 0xffffffff + + Non linear page tables need a list of frame addresses afterwards, + the rivatv project has some info on this. + + The method below creates a DMA object in instance RAM and returns a handle + to it that can be used to set up context objects. +*/ + +void +nv50_gpuobj_dma_init(struct nouveau_gpuobj *obj, u32 offset, int class, + u64 base, u64 size, int target, int access, + u32 type, u32 comp) +{ + struct drm_nouveau_private *dev_priv = obj->dev->dev_private; + struct nouveau_instmem_engine *pinstmem = &dev_priv->engine.instmem; + u32 flags0; + + flags0 = (comp << 29) | (type << 22) | class; + flags0 |= 0x00100000; + + switch (access) { + case NV_MEM_ACCESS_RO: flags0 |= 0x00040000; break; + case NV_MEM_ACCESS_RW: + case NV_MEM_ACCESS_WO: flags0 |= 0x00080000; break; + default: + break; + } + + switch (target) { + case NV_MEM_TARGET_VRAM: + flags0 |= 0x00010000; + break; + case NV_MEM_TARGET_PCI: + flags0 |= 0x00020000; + break; + case NV_MEM_TARGET_PCI_NOSNOOP: + flags0 |= 0x00030000; + break; + case NV_MEM_TARGET_GART: + base += dev_priv->gart_info.aper_base; + default: + flags0 &= ~0x00100000; + break; + } + + /* convert to base + limit */ + size = (base + size) - 1; + + nv_wo32(obj, offset + 0x00, flags0); + nv_wo32(obj, offset + 0x04, lower_32_bits(size)); + nv_wo32(obj, offset + 0x08, lower_32_bits(base)); + nv_wo32(obj, offset + 0x0c, upper_32_bits(size) << 24 | + upper_32_bits(base)); + nv_wo32(obj, offset + 0x10, 0x00000000); + nv_wo32(obj, offset + 0x14, 0x00000000); + + pinstmem->flush(obj->dev); +} + +int +nv50_gpuobj_dma_new(struct nouveau_channel *chan, int class, u64 base, u64 size, + int target, int access, u32 type, u32 comp, + struct nouveau_gpuobj **pobj) +{ + struct drm_device *dev = chan->dev; + int ret; + + ret = nouveau_gpuobj_new(dev, chan, 24, 16, NVOBJ_FLAG_ZERO_FREE, pobj); + if (ret) + return ret; + + nv50_gpuobj_dma_init(*pobj, 0, class, base, size, target, + access, type, comp); + return 0; +} + +int +nouveau_gpuobj_dma_new(struct nouveau_channel *chan, int class, u64 base, + u64 size, int access, int target, + struct nouveau_gpuobj **pobj) +{ + struct drm_nouveau_private *dev_priv = chan->dev->dev_private; + struct drm_device *dev = chan->dev; + struct nouveau_gpuobj *obj; + u32 flags0, flags2; + int ret; + + if (dev_priv->card_type >= NV_50) { + u32 comp = (target == NV_MEM_TARGET_VM) ? NV_MEM_COMP_VM : 0; + u32 type = (target == NV_MEM_TARGET_VM) ? NV_MEM_TYPE_VM : 0; + + return nv50_gpuobj_dma_new(chan, class, base, size, + target, access, type, comp, pobj); + } + + if (target == NV_MEM_TARGET_GART) { + struct nouveau_gpuobj *gart = dev_priv->gart_info.sg_ctxdma; + + if (dev_priv->gart_info.type == NOUVEAU_GART_PDMA) { + if (base == 0) { + nouveau_gpuobj_ref(gart, pobj); + return 0; + } + + base = nouveau_sgdma_get_physical(dev, base); + target = NV_MEM_TARGET_PCI; + } else { + base += dev_priv->gart_info.aper_base; + if (dev_priv->gart_info.type == NOUVEAU_GART_AGP) + target = NV_MEM_TARGET_PCI_NOSNOOP; + else + target = NV_MEM_TARGET_PCI; + } + } + + flags0 = class; + flags0 |= 0x00003000; /* PT present, PT linear */ + flags2 = 0; + + switch (target) { + case NV_MEM_TARGET_PCI: + flags0 |= 0x00020000; + break; + case NV_MEM_TARGET_PCI_NOSNOOP: + flags0 |= 0x00030000; + break; + default: + break; + } + + switch (access) { + case NV_MEM_ACCESS_RO: + flags0 |= 0x00004000; + break; + case NV_MEM_ACCESS_WO: + flags0 |= 0x00008000; + default: + flags2 |= 0x00000002; + break; + } + + flags0 |= (base & 0x00000fff) << 20; + flags2 |= (base & 0xfffff000); + + ret = nouveau_gpuobj_new(dev, chan, 16, 16, NVOBJ_FLAG_ZERO_FREE, &obj); + if (ret) + return ret; + + nv_wo32(obj, 0x00, flags0); + nv_wo32(obj, 0x04, size - 1); + nv_wo32(obj, 0x08, flags2); + nv_wo32(obj, 0x0c, flags2); + + obj->engine = NVOBJ_ENGINE_SW; + obj->class = class; + *pobj = obj; + return 0; +} + +/* Context objects in the instance RAM have the following structure. + * On NV40 they are 32 byte long, on NV30 and smaller 16 bytes. + + NV4 - NV30: + + entry[0] + 11:0 class + 12 chroma key enable + 13 user clip enable + 14 swizzle enable + 17:15 patch config: + scrcopy_and, rop_and, blend_and, scrcopy, srccopy_pre, blend_pre + 18 synchronize enable + 19 endian: 1 big, 0 little + 21:20 dither mode + 23 single step enable + 24 patch status: 0 invalid, 1 valid + 25 context_surface 0: 1 valid + 26 context surface 1: 1 valid + 27 context pattern: 1 valid + 28 context rop: 1 valid + 29,30 context beta, beta4 + entry[1] + 7:0 mono format + 15:8 color format + 31:16 notify instance address + entry[2] + 15:0 dma 0 instance address + 31:16 dma 1 instance address + entry[3] + dma method traps + + NV40: + No idea what the exact format is. Here's what can be deducted: + + entry[0]: + 11:0 class (maybe uses more bits here?) + 17 user clip enable + 21:19 patch config + 25 patch status valid ? + entry[1]: + 15:0 DMA notifier (maybe 20:0) + entry[2]: + 15:0 DMA 0 instance (maybe 20:0) + 24 big endian + entry[3]: + 15:0 DMA 1 instance (maybe 20:0) + entry[4]: + entry[5]: + set to 0? +*/ +static int +nouveau_gpuobj_sw_new(struct nouveau_channel *chan, u32 handle, u16 class) +{ + struct drm_nouveau_private *dev_priv = chan->dev->dev_private; + struct nouveau_gpuobj *gpuobj; + int ret; + + gpuobj = kzalloc(sizeof(*gpuobj), GFP_KERNEL); + if (!gpuobj) + return -ENOMEM; + gpuobj->dev = chan->dev; + gpuobj->engine = NVOBJ_ENGINE_SW; + gpuobj->class = class; + kref_init(&gpuobj->refcount); + gpuobj->cinst = 0x40; + + spin_lock(&dev_priv->ramin_lock); + list_add_tail(&gpuobj->list, &dev_priv->gpuobj_list); + spin_unlock(&dev_priv->ramin_lock); + + ret = nouveau_ramht_insert(chan, handle, gpuobj); + nouveau_gpuobj_ref(NULL, &gpuobj); + return ret; +} + +int +nouveau_gpuobj_gr_new(struct nouveau_channel *chan, u32 handle, int class) +{ + struct drm_nouveau_private *dev_priv = chan->dev->dev_private; + struct drm_device *dev = chan->dev; + struct nouveau_gpuobj_class *oc; + int ret; + + NV_DEBUG(dev, "ch%d class=0x%04x\n", chan->id, class); + + list_for_each_entry(oc, &dev_priv->classes, head) { + struct nouveau_exec_engine *eng = dev_priv->eng[oc->engine]; + + if (oc->id != class) + continue; + + if (oc->engine == NVOBJ_ENGINE_SW) + return nouveau_gpuobj_sw_new(chan, handle, class); + + if (!chan->engctx[oc->engine]) { + ret = eng->context_new(chan, oc->engine); + if (ret) + return ret; + } + + return eng->object_new(chan, oc->engine, handle, class); + } + + NV_ERROR(dev, "illegal object class: 0x%x\n", class); + return -EINVAL; +} + +static int +nouveau_gpuobj_channel_init_pramin(struct nouveau_channel *chan) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + uint32_t size; + uint32_t base; + int ret; + + NV_DEBUG(dev, "ch%d\n", chan->id); + + /* Base amount for object storage (4KiB enough?) */ + size = 0x2000; + base = 0; + + if (dev_priv->card_type == NV_50) { + /* Various fixed table thingos */ + size += 0x1400; /* mostly unknown stuff */ + size += 0x4000; /* vm pd */ + base = 0x6000; + /* RAMHT, not sure about setting size yet, 32KiB to be safe */ + size += 0x8000; + /* RAMFC */ + size += 0x1000; + } + + ret = nouveau_gpuobj_new(dev, NULL, size, 0x1000, 0, &chan->ramin); + if (ret) { + NV_ERROR(dev, "Error allocating channel PRAMIN: %d\n", ret); + return ret; + } + + ret = drm_mm_init(&chan->ramin_heap, base, size); + if (ret) { + NV_ERROR(dev, "Error creating PRAMIN heap: %d\n", ret); + nouveau_gpuobj_ref(NULL, &chan->ramin); + return ret; + } + + return 0; +} + +int +nouveau_gpuobj_channel_init(struct nouveau_channel *chan, + uint32_t vram_h, uint32_t tt_h) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *vram = NULL, *tt = NULL; + int ret, i; + + NV_DEBUG(dev, "ch%d vram=0x%08x tt=0x%08x\n", chan->id, vram_h, tt_h); + + if (dev_priv->card_type == NV_C0) { + struct nouveau_vm *vm = dev_priv->chan_vm; + struct nouveau_vm_pgd *vpgd; + + ret = nouveau_gpuobj_new(dev, NULL, 4096, 0x1000, 0, + &chan->ramin); + if (ret) + return ret; + + nouveau_vm_ref(vm, &chan->vm, NULL); + + vpgd = list_first_entry(&vm->pgd_list, struct nouveau_vm_pgd, head); + nv_wo32(chan->ramin, 0x0200, lower_32_bits(vpgd->obj->vinst)); + nv_wo32(chan->ramin, 0x0204, upper_32_bits(vpgd->obj->vinst)); + nv_wo32(chan->ramin, 0x0208, 0xffffffff); + nv_wo32(chan->ramin, 0x020c, 0x000000ff); + return 0; + } + + /* Allocate a chunk of memory for per-channel object storage */ + ret = nouveau_gpuobj_channel_init_pramin(chan); + if (ret) { + NV_ERROR(dev, "init pramin\n"); + return ret; + } + + /* NV50 VM + * - Allocate per-channel page-directory + * - Link with shared channel VM + */ + if (dev_priv->chan_vm) { + u32 pgd_offs = (dev_priv->chipset == 0x50) ? 0x1400 : 0x0200; + u64 vm_vinst = chan->ramin->vinst + pgd_offs; + u32 vm_pinst = chan->ramin->pinst; + + if (vm_pinst != ~0) + vm_pinst += pgd_offs; + + ret = nouveau_gpuobj_new_fake(dev, vm_pinst, vm_vinst, 0x4000, + 0, &chan->vm_pd); + if (ret) + return ret; + + nouveau_vm_ref(dev_priv->chan_vm, &chan->vm, chan->vm_pd); + } + + /* RAMHT */ + if (dev_priv->card_type < NV_50) { + nouveau_ramht_ref(dev_priv->ramht, &chan->ramht, NULL); + } else { + struct nouveau_gpuobj *ramht = NULL; + + ret = nouveau_gpuobj_new(dev, chan, 0x8000, 16, + NVOBJ_FLAG_ZERO_ALLOC, &ramht); + if (ret) + return ret; + + ret = nouveau_ramht_new(dev, ramht, &chan->ramht); + nouveau_gpuobj_ref(NULL, &ramht); + if (ret) + return ret; + + /* dma objects for display sync channel semaphore blocks */ + for (i = 0; i < 2; i++) { + struct nouveau_gpuobj *sem = NULL; + struct nv50_display_crtc *dispc = + &nv50_display(dev)->crtc[i]; + u64 offset = dispc->sem.bo->bo.mem.start << PAGE_SHIFT; + + ret = nouveau_gpuobj_dma_new(chan, 0x3d, offset, 0xfff, + NV_MEM_ACCESS_RW, + NV_MEM_TARGET_VRAM, &sem); + if (ret) + return ret; + + ret = nouveau_ramht_insert(chan, NvEvoSema0 + i, sem); + nouveau_gpuobj_ref(NULL, &sem); + if (ret) + return ret; + } + } + + /* VRAM ctxdma */ + if (dev_priv->card_type >= NV_50) { + ret = nouveau_gpuobj_dma_new(chan, NV_CLASS_DMA_IN_MEMORY, + 0, (1ULL << 40), NV_MEM_ACCESS_RW, + NV_MEM_TARGET_VM, &vram); + if (ret) { + NV_ERROR(dev, "Error creating VRAM ctxdma: %d\n", ret); + return ret; + } + } else { + ret = nouveau_gpuobj_dma_new(chan, NV_CLASS_DMA_IN_MEMORY, + 0, dev_priv->fb_available_size, + NV_MEM_ACCESS_RW, + NV_MEM_TARGET_VRAM, &vram); + if (ret) { + NV_ERROR(dev, "Error creating VRAM ctxdma: %d\n", ret); + return ret; + } + } + + ret = nouveau_ramht_insert(chan, vram_h, vram); + nouveau_gpuobj_ref(NULL, &vram); + if (ret) { + NV_ERROR(dev, "Error adding VRAM ctxdma to RAMHT: %d\n", ret); + return ret; + } + + /* TT memory ctxdma */ + if (dev_priv->card_type >= NV_50) { + ret = nouveau_gpuobj_dma_new(chan, NV_CLASS_DMA_IN_MEMORY, + 0, (1ULL << 40), NV_MEM_ACCESS_RW, + NV_MEM_TARGET_VM, &tt); + } else { + ret = nouveau_gpuobj_dma_new(chan, NV_CLASS_DMA_IN_MEMORY, + 0, dev_priv->gart_info.aper_size, + NV_MEM_ACCESS_RW, + NV_MEM_TARGET_GART, &tt); + } + + if (ret) { + NV_ERROR(dev, "Error creating TT ctxdma: %d\n", ret); + return ret; + } + + ret = nouveau_ramht_insert(chan, tt_h, tt); + nouveau_gpuobj_ref(NULL, &tt); + if (ret) { + NV_ERROR(dev, "Error adding TT ctxdma to RAMHT: %d\n", ret); + return ret; + } + + return 0; +} + +void +nouveau_gpuobj_channel_takedown(struct nouveau_channel *chan) +{ + struct drm_device *dev = chan->dev; + + NV_DEBUG(dev, "ch%d\n", chan->id); + + nouveau_ramht_ref(NULL, &chan->ramht, chan); + + nouveau_vm_ref(NULL, &chan->vm, chan->vm_pd); + nouveau_gpuobj_ref(NULL, &chan->vm_pd); + + if (drm_mm_initialized(&chan->ramin_heap)) + drm_mm_takedown(&chan->ramin_heap); + nouveau_gpuobj_ref(NULL, &chan->ramin); +} + +int +nouveau_gpuobj_suspend(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *gpuobj; + int i; + + list_for_each_entry(gpuobj, &dev_priv->gpuobj_list, list) { + if (gpuobj->cinst != NVOBJ_CINST_GLOBAL) + continue; + + gpuobj->suspend = vmalloc(gpuobj->size); + if (!gpuobj->suspend) { + nouveau_gpuobj_resume(dev); + return -ENOMEM; + } + + for (i = 0; i < gpuobj->size; i += 4) + gpuobj->suspend[i/4] = nv_ro32(gpuobj, i); + } + + return 0; +} + +void +nouveau_gpuobj_resume(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *gpuobj; + int i; + + list_for_each_entry(gpuobj, &dev_priv->gpuobj_list, list) { + if (!gpuobj->suspend) + continue; + + for (i = 0; i < gpuobj->size; i += 4) + nv_wo32(gpuobj, i, gpuobj->suspend[i/4]); + + vfree(gpuobj->suspend); + gpuobj->suspend = NULL; + } + + dev_priv->engine.instmem.flush(dev); +} + +int nouveau_ioctl_grobj_alloc(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_nouveau_grobj_alloc *init = data; + struct nouveau_channel *chan; + int ret; + + if (init->handle == ~0) + return -EINVAL; + + chan = nouveau_channel_get(dev, file_priv, init->channel); + if (IS_ERR(chan)) + return PTR_ERR(chan); + + if (nouveau_ramht_find(chan, init->handle)) { + ret = -EEXIST; + goto out; + } + + ret = nouveau_gpuobj_gr_new(chan, init->handle, init->class); + if (ret) { + NV_ERROR(dev, "Error creating object: %d (%d/0x%08x)\n", + ret, init->channel, init->handle); + } + +out: + nouveau_channel_put(&chan); + return ret; +} + +int nouveau_ioctl_gpuobj_free(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_nouveau_gpuobj_free *objfree = data; + struct nouveau_channel *chan; + int ret; + + chan = nouveau_channel_get(dev, file_priv, objfree->channel); + if (IS_ERR(chan)) + return PTR_ERR(chan); + + /* Synchronize with the user channel */ + nouveau_channel_idle(chan); + + ret = nouveau_ramht_remove(chan, objfree->handle); + nouveau_channel_put(&chan); + return ret; +} + +u32 +nv_ro32(struct nouveau_gpuobj *gpuobj, u32 offset) +{ + struct drm_nouveau_private *dev_priv = gpuobj->dev->dev_private; + struct drm_device *dev = gpuobj->dev; + unsigned long flags; + + if (gpuobj->pinst == ~0 || !dev_priv->ramin_available) { + u64 ptr = gpuobj->vinst + offset; + u32 base = ptr >> 16; + u32 val; + + spin_lock_irqsave(&dev_priv->vm_lock, flags); + if (dev_priv->ramin_base != base) { + dev_priv->ramin_base = base; + nv_wr32(dev, 0x001700, dev_priv->ramin_base); + } + val = nv_rd32(dev, 0x700000 + (ptr & 0xffff)); + spin_unlock_irqrestore(&dev_priv->vm_lock, flags); + return val; + } + + return nv_ri32(dev, gpuobj->pinst + offset); +} + +void +nv_wo32(struct nouveau_gpuobj *gpuobj, u32 offset, u32 val) +{ + struct drm_nouveau_private *dev_priv = gpuobj->dev->dev_private; + struct drm_device *dev = gpuobj->dev; + unsigned long flags; + + if (gpuobj->pinst == ~0 || !dev_priv->ramin_available) { + u64 ptr = gpuobj->vinst + offset; + u32 base = ptr >> 16; + + spin_lock_irqsave(&dev_priv->vm_lock, flags); + if (dev_priv->ramin_base != base) { + dev_priv->ramin_base = base; + nv_wr32(dev, 0x001700, dev_priv->ramin_base); + } + nv_wr32(dev, 0x700000 + (ptr & 0xffff), val); + spin_unlock_irqrestore(&dev_priv->vm_lock, flags); + return; + } + + nv_wi32(dev, gpuobj->pinst + offset, val); +} diff --git a/drivers/gpu/drm/nouveau/nouveau_perf.c b/drivers/gpu/drm/nouveau/nouveau_perf.c new file mode 100644 index 00000000..ef9dec0e --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_perf.c @@ -0,0 +1,296 @@ +/* + * Copyright 2010 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include "drmP.h" + +#include "nouveau_drv.h" +#include "nouveau_pm.h" + +static void +legacy_perf_init(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nvbios *bios = &dev_priv->vbios; + struct nouveau_pm_engine *pm = &dev_priv->engine.pm; + char *perf, *entry, *bmp = &bios->data[bios->offset]; + int headerlen, use_straps; + + if (bmp[5] < 0x5 || bmp[6] < 0x14) { + NV_DEBUG(dev, "BMP version too old for perf\n"); + return; + } + + perf = ROMPTR(bios, bmp[0x73]); + if (!perf) { + NV_DEBUG(dev, "No memclock table pointer found.\n"); + return; + } + + switch (perf[0]) { + case 0x12: + case 0x14: + case 0x18: + use_straps = 0; + headerlen = 1; + break; + case 0x01: + use_straps = perf[1] & 1; + headerlen = (use_straps ? 8 : 2); + break; + default: + NV_WARN(dev, "Unknown memclock table version %x.\n", perf[0]); + return; + } + + entry = perf + headerlen; + if (use_straps) + entry += (nv_rd32(dev, NV_PEXTDEV_BOOT_0) & 0x3c) >> 1; + + sprintf(pm->perflvl[0].name, "performance_level_0"); + pm->perflvl[0].memory = ROM16(entry[0]) * 20; + pm->nr_perflvl = 1; +} + +static struct nouveau_pm_memtiming * +nouveau_perf_timing(struct drm_device *dev, struct bit_entry *P, + u16 memclk, u8 *entry, u8 recordlen, u8 entries) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_pm_engine *pm = &dev_priv->engine.pm; + struct nvbios *bios = &dev_priv->vbios; + u8 ramcfg; + int i; + + /* perf v2 has a separate "timing map" table, we have to match + * the target memory clock to a specific entry, *then* use + * ramcfg to select the correct subentry + */ + if (P->version == 2) { + u8 *tmap = ROMPTR(bios, P->data[4]); + if (!tmap) { + NV_DEBUG(dev, "no timing map pointer\n"); + return NULL; + } + + if (tmap[0] != 0x10) { + NV_WARN(dev, "timing map 0x%02x unknown\n", tmap[0]); + return NULL; + } + + entry = tmap + tmap[1]; + recordlen = tmap[2] + (tmap[4] * tmap[3]); + for (i = 0; i < tmap[5]; i++, entry += recordlen) { + if (memclk >= ROM16(entry[0]) && + memclk <= ROM16(entry[2])) + break; + } + + if (i == tmap[5]) { + NV_WARN(dev, "no match in timing map table\n"); + return NULL; + } + + entry += tmap[2]; + recordlen = tmap[3]; + entries = tmap[4]; + } + + ramcfg = (nv_rd32(dev, NV_PEXTDEV_BOOT_0) & 0x0000003c) >> 2; + if (bios->ram_restrict_tbl_ptr) + ramcfg = bios->data[bios->ram_restrict_tbl_ptr + ramcfg]; + + if (ramcfg >= entries) { + NV_WARN(dev, "ramcfg strap out of bounds!\n"); + return NULL; + } + + entry += ramcfg * recordlen; + if (entry[1] >= pm->memtimings.nr_timing) { + NV_WARN(dev, "timingset %d does not exist\n", entry[1]); + return NULL; + } + + return &pm->memtimings.timing[entry[1]]; +} + +void +nouveau_perf_init(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_pm_engine *pm = &dev_priv->engine.pm; + struct nvbios *bios = &dev_priv->vbios; + struct bit_entry P; + u8 version, headerlen, recordlen, entries; + u8 *perf, *entry; + int vid, i; + + if (bios->type == NVBIOS_BIT) { + if (bit_table(dev, 'P', &P)) + return; + + if (P.version != 1 && P.version != 2) { + NV_WARN(dev, "unknown perf for BIT P %d\n", P.version); + return; + } + + perf = ROMPTR(bios, P.data[0]); + version = perf[0]; + headerlen = perf[1]; + if (version < 0x40) { + recordlen = perf[3] + (perf[4] * perf[5]); + entries = perf[2]; + } else { + recordlen = perf[2] + (perf[3] * perf[4]); + entries = perf[5]; + } + } else { + if (bios->data[bios->offset + 6] < 0x25) { + legacy_perf_init(dev); + return; + } + + perf = ROMPTR(bios, bios->data[bios->offset + 0x94]); + if (!perf) { + NV_DEBUG(dev, "perf table pointer invalid\n"); + return; + } + + version = perf[1]; + headerlen = perf[0]; + recordlen = perf[3]; + entries = perf[2]; + } + + if (entries > NOUVEAU_PM_MAX_LEVEL) { + NV_DEBUG(dev, "perf table has too many entries - buggy vbios?\n"); + entries = NOUVEAU_PM_MAX_LEVEL; + } + + entry = perf + headerlen; + for (i = 0; i < entries; i++) { + struct nouveau_pm_level *perflvl = &pm->perflvl[pm->nr_perflvl]; + + perflvl->timing = NULL; + + if (entry[0] == 0xff) { + entry += recordlen; + continue; + } + + switch (version) { + case 0x12: + case 0x13: + case 0x15: + perflvl->fanspeed = entry[55]; + perflvl->voltage = (recordlen > 56) ? entry[56] : 0; + perflvl->core = ROM32(entry[1]) * 10; + perflvl->memory = ROM32(entry[5]) * 20; + break; + case 0x21: + case 0x23: + case 0x24: + perflvl->fanspeed = entry[4]; + perflvl->voltage = entry[5]; + perflvl->core = ROM16(entry[6]) * 1000; + + if (dev_priv->chipset == 0x49 || + dev_priv->chipset == 0x4b) + perflvl->memory = ROM16(entry[11]) * 1000; + else + perflvl->memory = ROM16(entry[11]) * 2000; + + break; + case 0x25: + perflvl->fanspeed = entry[4]; + perflvl->voltage = entry[5]; + perflvl->core = ROM16(entry[6]) * 1000; + perflvl->shader = ROM16(entry[10]) * 1000; + perflvl->memory = ROM16(entry[12]) * 1000; + break; + case 0x30: + perflvl->memscript = ROM16(entry[2]); + case 0x35: + perflvl->fanspeed = entry[6]; + perflvl->voltage = entry[7]; + perflvl->core = ROM16(entry[8]) * 1000; + perflvl->shader = ROM16(entry[10]) * 1000; + perflvl->memory = ROM16(entry[12]) * 1000; + /*XXX: confirm on 0x35 */ + perflvl->unk05 = ROM16(entry[16]) * 1000; + break; + case 0x40: +#define subent(n) entry[perf[2] + ((n) * perf[3])] + perflvl->fanspeed = 0; /*XXX*/ + perflvl->voltage = entry[2]; + if (dev_priv->card_type == NV_50) { + perflvl->core = ROM16(subent(0)) & 0xfff; + perflvl->shader = ROM16(subent(1)) & 0xfff; + perflvl->memory = ROM16(subent(2)) & 0xfff; + } else { + perflvl->shader = ROM16(subent(3)) & 0xfff; + perflvl->core = perflvl->shader / 2; + perflvl->unk0a = ROM16(subent(4)) & 0xfff; + perflvl->memory = ROM16(subent(5)) & 0xfff; + } + + perflvl->core *= 1000; + perflvl->shader *= 1000; + perflvl->memory *= 1000; + perflvl->unk0a *= 1000; + break; + } + + /* make sure vid is valid */ + if (pm->voltage.supported && perflvl->voltage) { + vid = nouveau_volt_vid_lookup(dev, perflvl->voltage); + if (vid < 0) { + NV_DEBUG(dev, "drop perflvl %d, bad vid\n", i); + entry += recordlen; + continue; + } + } + + /* get the corresponding memory timings */ + if (version > 0x15) { + /* last 3 args are for < 0x40, ignored for >= 0x40 */ + perflvl->timing = + nouveau_perf_timing(dev, &P, + perflvl->memory / 1000, + entry + perf[3], + perf[5], perf[4]); + } + + snprintf(perflvl->name, sizeof(perflvl->name), + "performance_level_%d", i); + perflvl->id = i; + pm->nr_perflvl++; + + entry += recordlen; + } +} + +void +nouveau_perf_fini(struct drm_device *dev) +{ +} diff --git a/drivers/gpu/drm/nouveau/nouveau_pm.c b/drivers/gpu/drm/nouveau/nouveau_pm.c new file mode 100644 index 00000000..da8d994d --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_pm.c @@ -0,0 +1,557 @@ +/* + * Copyright 2010 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include "drmP.h" + +#include "nouveau_drv.h" +#include "nouveau_pm.h" + +#ifdef CONFIG_ACPI +#include +#endif +#include +#include +#include + +static int +nouveau_pm_clock_set(struct drm_device *dev, struct nouveau_pm_level *perflvl, + u8 id, u32 khz) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_pm_engine *pm = &dev_priv->engine.pm; + void *pre_state; + + if (khz == 0) + return 0; + + pre_state = pm->clock_pre(dev, perflvl, id, khz); + if (IS_ERR(pre_state)) + return PTR_ERR(pre_state); + + if (pre_state) + pm->clock_set(dev, pre_state); + return 0; +} + +static int +nouveau_pm_perflvl_set(struct drm_device *dev, struct nouveau_pm_level *perflvl) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_pm_engine *pm = &dev_priv->engine.pm; + int ret; + + if (perflvl == pm->cur) + return 0; + + if (pm->voltage.supported && pm->voltage_set && perflvl->voltage) { + ret = pm->voltage_set(dev, perflvl->voltage); + if (ret) { + NV_ERROR(dev, "voltage_set %d failed: %d\n", + perflvl->voltage, ret); + } + } + + nouveau_pm_clock_set(dev, perflvl, PLL_CORE, perflvl->core); + nouveau_pm_clock_set(dev, perflvl, PLL_SHADER, perflvl->shader); + nouveau_pm_clock_set(dev, perflvl, PLL_MEMORY, perflvl->memory); + nouveau_pm_clock_set(dev, perflvl, PLL_UNK05, perflvl->unk05); + + pm->cur = perflvl; + return 0; +} + +static int +nouveau_pm_profile_set(struct drm_device *dev, const char *profile) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_pm_engine *pm = &dev_priv->engine.pm; + struct nouveau_pm_level *perflvl = NULL; + + /* safety precaution, for now */ + if (nouveau_perflvl_wr != 7777) + return -EPERM; + + if (!pm->clock_set) + return -EINVAL; + + if (!strncmp(profile, "boot", 4)) + perflvl = &pm->boot; + else { + int pl = simple_strtol(profile, NULL, 10); + int i; + + for (i = 0; i < pm->nr_perflvl; i++) { + if (pm->perflvl[i].id == pl) { + perflvl = &pm->perflvl[i]; + break; + } + } + + if (!perflvl) + return -EINVAL; + } + + NV_INFO(dev, "setting performance level: %s\n", profile); + return nouveau_pm_perflvl_set(dev, perflvl); +} + +static int +nouveau_pm_perflvl_get(struct drm_device *dev, struct nouveau_pm_level *perflvl) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_pm_engine *pm = &dev_priv->engine.pm; + int ret; + + if (!pm->clock_get) + return -EINVAL; + + memset(perflvl, 0, sizeof(*perflvl)); + + ret = pm->clock_get(dev, PLL_CORE); + if (ret > 0) + perflvl->core = ret; + + ret = pm->clock_get(dev, PLL_MEMORY); + if (ret > 0) + perflvl->memory = ret; + + ret = pm->clock_get(dev, PLL_SHADER); + if (ret > 0) + perflvl->shader = ret; + + ret = pm->clock_get(dev, PLL_UNK05); + if (ret > 0) + perflvl->unk05 = ret; + + if (pm->voltage.supported && pm->voltage_get) { + ret = pm->voltage_get(dev); + if (ret > 0) + perflvl->voltage = ret; + } + + return 0; +} + +static void +nouveau_pm_perflvl_info(struct nouveau_pm_level *perflvl, char *ptr, int len) +{ + char c[16], s[16], v[16], f[16], t[16]; + + c[0] = '\0'; + if (perflvl->core) + snprintf(c, sizeof(c), " core %dMHz", perflvl->core / 1000); + + s[0] = '\0'; + if (perflvl->shader) + snprintf(s, sizeof(s), " shader %dMHz", perflvl->shader / 1000); + + v[0] = '\0'; + if (perflvl->voltage) + snprintf(v, sizeof(v), " voltage %dmV", perflvl->voltage * 10); + + f[0] = '\0'; + if (perflvl->fanspeed) + snprintf(f, sizeof(f), " fanspeed %d%%", perflvl->fanspeed); + + t[0] = '\0'; + if (perflvl->timing) + snprintf(t, sizeof(t), " timing %d", perflvl->timing->id); + + snprintf(ptr, len, "memory %dMHz%s%s%s%s%s\n", perflvl->memory / 1000, + c, s, v, f, t); +} + +static ssize_t +nouveau_pm_get_perflvl_info(struct device *d, + struct device_attribute *a, char *buf) +{ + struct nouveau_pm_level *perflvl = (struct nouveau_pm_level *)a; + char *ptr = buf; + int len = PAGE_SIZE; + + snprintf(ptr, len, "%d: ", perflvl->id); + ptr += strlen(buf); + len -= strlen(buf); + + nouveau_pm_perflvl_info(perflvl, ptr, len); + return strlen(buf); +} + +static ssize_t +nouveau_pm_get_perflvl(struct device *d, struct device_attribute *a, char *buf) +{ + struct drm_device *dev = pci_get_drvdata(to_pci_dev(d)); + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_pm_engine *pm = &dev_priv->engine.pm; + struct nouveau_pm_level cur; + int len = PAGE_SIZE, ret; + char *ptr = buf; + + if (!pm->cur) + snprintf(ptr, len, "setting: boot\n"); + else if (pm->cur == &pm->boot) + snprintf(ptr, len, "setting: boot\nc: "); + else + snprintf(ptr, len, "setting: static %d\nc: ", pm->cur->id); + ptr += strlen(buf); + len -= strlen(buf); + + ret = nouveau_pm_perflvl_get(dev, &cur); + if (ret == 0) + nouveau_pm_perflvl_info(&cur, ptr, len); + return strlen(buf); +} + +static ssize_t +nouveau_pm_set_perflvl(struct device *d, struct device_attribute *a, + const char *buf, size_t count) +{ + struct drm_device *dev = pci_get_drvdata(to_pci_dev(d)); + int ret; + + ret = nouveau_pm_profile_set(dev, buf); + if (ret) + return ret; + return strlen(buf); +} + +static DEVICE_ATTR(performance_level, S_IRUGO | S_IWUSR, + nouveau_pm_get_perflvl, nouveau_pm_set_perflvl); + +static int +nouveau_sysfs_init(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_pm_engine *pm = &dev_priv->engine.pm; + struct device *d = &dev->pdev->dev; + int ret, i; + + ret = device_create_file(d, &dev_attr_performance_level); + if (ret) + return ret; + + for (i = 0; i < pm->nr_perflvl; i++) { + struct nouveau_pm_level *perflvl = &pm->perflvl[i]; + + perflvl->dev_attr.attr.name = perflvl->name; + perflvl->dev_attr.attr.mode = S_IRUGO; + perflvl->dev_attr.show = nouveau_pm_get_perflvl_info; + perflvl->dev_attr.store = NULL; + sysfs_attr_init(&perflvl->dev_attr.attr); + + ret = device_create_file(d, &perflvl->dev_attr); + if (ret) { + NV_ERROR(dev, "failed pervlvl %d sysfs: %d\n", + perflvl->id, i); + perflvl->dev_attr.attr.name = NULL; + nouveau_pm_fini(dev); + return ret; + } + } + + return 0; +} + +static void +nouveau_sysfs_fini(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_pm_engine *pm = &dev_priv->engine.pm; + struct device *d = &dev->pdev->dev; + int i; + + device_remove_file(d, &dev_attr_performance_level); + for (i = 0; i < pm->nr_perflvl; i++) { + struct nouveau_pm_level *pl = &pm->perflvl[i]; + + if (!pl->dev_attr.attr.name) + break; + + device_remove_file(d, &pl->dev_attr); + } +} + +#ifdef CONFIG_HWMON +static ssize_t +nouveau_hwmon_show_temp(struct device *d, struct device_attribute *a, char *buf) +{ + struct drm_device *dev = dev_get_drvdata(d); + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_pm_engine *pm = &dev_priv->engine.pm; + + return snprintf(buf, PAGE_SIZE, "%d\n", pm->temp_get(dev)*1000); +} +static SENSOR_DEVICE_ATTR(temp1_input, S_IRUGO, nouveau_hwmon_show_temp, + NULL, 0); + +static ssize_t +nouveau_hwmon_max_temp(struct device *d, struct device_attribute *a, char *buf) +{ + struct drm_device *dev = dev_get_drvdata(d); + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_pm_engine *pm = &dev_priv->engine.pm; + struct nouveau_pm_threshold_temp *temp = &pm->threshold_temp; + + return snprintf(buf, PAGE_SIZE, "%d\n", temp->down_clock*1000); +} +static ssize_t +nouveau_hwmon_set_max_temp(struct device *d, struct device_attribute *a, + const char *buf, size_t count) +{ + struct drm_device *dev = dev_get_drvdata(d); + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_pm_engine *pm = &dev_priv->engine.pm; + struct nouveau_pm_threshold_temp *temp = &pm->threshold_temp; + long value; + + if (strict_strtol(buf, 10, &value) == -EINVAL) + return count; + + temp->down_clock = value/1000; + + nouveau_temp_safety_checks(dev); + + return count; +} +static SENSOR_DEVICE_ATTR(temp1_max, S_IRUGO | S_IWUSR, nouveau_hwmon_max_temp, + nouveau_hwmon_set_max_temp, + 0); + +static ssize_t +nouveau_hwmon_critical_temp(struct device *d, struct device_attribute *a, + char *buf) +{ + struct drm_device *dev = dev_get_drvdata(d); + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_pm_engine *pm = &dev_priv->engine.pm; + struct nouveau_pm_threshold_temp *temp = &pm->threshold_temp; + + return snprintf(buf, PAGE_SIZE, "%d\n", temp->critical*1000); +} +static ssize_t +nouveau_hwmon_set_critical_temp(struct device *d, struct device_attribute *a, + const char *buf, + size_t count) +{ + struct drm_device *dev = dev_get_drvdata(d); + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_pm_engine *pm = &dev_priv->engine.pm; + struct nouveau_pm_threshold_temp *temp = &pm->threshold_temp; + long value; + + if (strict_strtol(buf, 10, &value) == -EINVAL) + return count; + + temp->critical = value/1000; + + nouveau_temp_safety_checks(dev); + + return count; +} +static SENSOR_DEVICE_ATTR(temp1_crit, S_IRUGO | S_IWUSR, + nouveau_hwmon_critical_temp, + nouveau_hwmon_set_critical_temp, + 0); + +static ssize_t nouveau_hwmon_show_name(struct device *dev, + struct device_attribute *attr, + char *buf) +{ + return sprintf(buf, "nouveau\n"); +} +static SENSOR_DEVICE_ATTR(name, S_IRUGO, nouveau_hwmon_show_name, NULL, 0); + +static ssize_t nouveau_hwmon_show_update_rate(struct device *dev, + struct device_attribute *attr, + char *buf) +{ + return sprintf(buf, "1000\n"); +} +static SENSOR_DEVICE_ATTR(update_rate, S_IRUGO, + nouveau_hwmon_show_update_rate, + NULL, 0); + +static struct attribute *hwmon_attributes[] = { + &sensor_dev_attr_temp1_input.dev_attr.attr, + &sensor_dev_attr_temp1_max.dev_attr.attr, + &sensor_dev_attr_temp1_crit.dev_attr.attr, + &sensor_dev_attr_name.dev_attr.attr, + &sensor_dev_attr_update_rate.dev_attr.attr, + NULL +}; + +static const struct attribute_group hwmon_attrgroup = { + .attrs = hwmon_attributes, +}; +#endif + +static int +nouveau_hwmon_init(struct drm_device *dev) +{ +#ifdef CONFIG_HWMON + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_pm_engine *pm = &dev_priv->engine.pm; + struct device *hwmon_dev; + int ret; + + if (!pm->temp_get) + return -ENODEV; + + hwmon_dev = hwmon_device_register(&dev->pdev->dev); + if (IS_ERR(hwmon_dev)) { + ret = PTR_ERR(hwmon_dev); + NV_ERROR(dev, + "Unable to register hwmon device: %d\n", ret); + return ret; + } + dev_set_drvdata(hwmon_dev, dev); + ret = sysfs_create_group(&dev->pdev->dev.kobj, &hwmon_attrgroup); + if (ret) { + NV_ERROR(dev, + "Unable to create hwmon sysfs file: %d\n", ret); + hwmon_device_unregister(hwmon_dev); + return ret; + } + + pm->hwmon = hwmon_dev; +#endif + return 0; +} + +static void +nouveau_hwmon_fini(struct drm_device *dev) +{ +#ifdef CONFIG_HWMON + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_pm_engine *pm = &dev_priv->engine.pm; + + if (pm->hwmon) { + sysfs_remove_group(&dev->pdev->dev.kobj, &hwmon_attrgroup); + hwmon_device_unregister(pm->hwmon); + } +#endif +} + +#if defined(CONFIG_ACPI) && defined(CONFIG_POWER_SUPPLY) +static int +nouveau_pm_acpi_event(struct notifier_block *nb, unsigned long val, void *data) +{ + struct drm_nouveau_private *dev_priv = + container_of(nb, struct drm_nouveau_private, engine.pm.acpi_nb); + struct drm_device *dev = dev_priv->dev; + struct acpi_bus_event *entry = (struct acpi_bus_event *)data; + + if (strcmp(entry->device_class, "ac_adapter") == 0) { + bool ac = power_supply_is_system_supplied(); + + NV_DEBUG(dev, "power supply changed: %s\n", ac ? "AC" : "DC"); + } + + return NOTIFY_OK; +} +#endif + +int +nouveau_pm_init(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_pm_engine *pm = &dev_priv->engine.pm; + char info[256]; + int ret, i; + + nouveau_mem_timing_init(dev); + nouveau_volt_init(dev); + nouveau_perf_init(dev); + nouveau_temp_init(dev); + + NV_INFO(dev, "%d available performance level(s)\n", pm->nr_perflvl); + for (i = 0; i < pm->nr_perflvl; i++) { + nouveau_pm_perflvl_info(&pm->perflvl[i], info, sizeof(info)); + NV_INFO(dev, "%d: %s", pm->perflvl[i].id, info); + } + + /* determine current ("boot") performance level */ + ret = nouveau_pm_perflvl_get(dev, &pm->boot); + if (ret == 0) { + strncpy(pm->boot.name, "boot", 4); + pm->cur = &pm->boot; + + nouveau_pm_perflvl_info(&pm->boot, info, sizeof(info)); + NV_INFO(dev, "c: %s", info); + } + + /* switch performance levels now if requested */ + if (nouveau_perflvl != NULL) { + ret = nouveau_pm_profile_set(dev, nouveau_perflvl); + if (ret) { + NV_ERROR(dev, "error setting perflvl \"%s\": %d\n", + nouveau_perflvl, ret); + } + } + + nouveau_sysfs_init(dev); + nouveau_hwmon_init(dev); +#if defined(CONFIG_ACPI) && defined(CONFIG_POWER_SUPPLY) + pm->acpi_nb.notifier_call = nouveau_pm_acpi_event; + register_acpi_notifier(&pm->acpi_nb); +#endif + + return 0; +} + +void +nouveau_pm_fini(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_pm_engine *pm = &dev_priv->engine.pm; + + if (pm->cur != &pm->boot) + nouveau_pm_perflvl_set(dev, &pm->boot); + + nouveau_temp_fini(dev); + nouveau_perf_fini(dev); + nouveau_volt_fini(dev); + nouveau_mem_timing_fini(dev); + +#if defined(CONFIG_ACPI) && defined(CONFIG_POWER_SUPPLY) + unregister_acpi_notifier(&pm->acpi_nb); +#endif + nouveau_hwmon_fini(dev); + nouveau_sysfs_fini(dev); +} + +void +nouveau_pm_resume(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_pm_engine *pm = &dev_priv->engine.pm; + struct nouveau_pm_level *perflvl; + + if (!pm->cur || pm->cur == &pm->boot) + return; + + perflvl = pm->cur; + pm->cur = &pm->boot; + nouveau_pm_perflvl_set(dev, perflvl); +} diff --git a/drivers/gpu/drm/nouveau/nouveau_pm.h b/drivers/gpu/drm/nouveau/nouveau_pm.h new file mode 100644 index 00000000..4a9838dd --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_pm.h @@ -0,0 +1,74 @@ +/* + * Copyright 2010 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#ifndef __NOUVEAU_PM_H__ +#define __NOUVEAU_PM_H__ + +/* nouveau_pm.c */ +int nouveau_pm_init(struct drm_device *dev); +void nouveau_pm_fini(struct drm_device *dev); +void nouveau_pm_resume(struct drm_device *dev); + +/* nouveau_volt.c */ +void nouveau_volt_init(struct drm_device *); +void nouveau_volt_fini(struct drm_device *); +int nouveau_volt_vid_lookup(struct drm_device *, int voltage); +int nouveau_volt_lvl_lookup(struct drm_device *, int vid); +int nouveau_voltage_gpio_get(struct drm_device *); +int nouveau_voltage_gpio_set(struct drm_device *, int voltage); + +/* nouveau_perf.c */ +void nouveau_perf_init(struct drm_device *); +void nouveau_perf_fini(struct drm_device *); + +/* nouveau_mem.c */ +void nouveau_mem_timing_init(struct drm_device *); +void nouveau_mem_timing_fini(struct drm_device *); + +/* nv04_pm.c */ +int nv04_pm_clock_get(struct drm_device *, u32 id); +void *nv04_pm_clock_pre(struct drm_device *, struct nouveau_pm_level *, + u32 id, int khz); +void nv04_pm_clock_set(struct drm_device *, void *); + +/* nv50_pm.c */ +int nv50_pm_clock_get(struct drm_device *, u32 id); +void *nv50_pm_clock_pre(struct drm_device *, struct nouveau_pm_level *, + u32 id, int khz); +void nv50_pm_clock_set(struct drm_device *, void *); + +/* nva3_pm.c */ +int nva3_pm_clock_get(struct drm_device *, u32 id); +void *nva3_pm_clock_pre(struct drm_device *, struct nouveau_pm_level *, + u32 id, int khz); +void nva3_pm_clock_set(struct drm_device *, void *); + +/* nouveau_temp.c */ +void nouveau_temp_init(struct drm_device *dev); +void nouveau_temp_fini(struct drm_device *dev); +void nouveau_temp_safety_checks(struct drm_device *dev); +int nv40_temp_get(struct drm_device *dev); +int nv84_temp_get(struct drm_device *dev); + +#endif diff --git a/drivers/gpu/drm/nouveau/nouveau_ramht.c b/drivers/gpu/drm/nouveau/nouveau_ramht.c new file mode 100644 index 00000000..a24a81f5 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_ramht.c @@ -0,0 +1,309 @@ +/* + * Copyright 2010 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include "drmP.h" + +#include "nouveau_drv.h" +#include "nouveau_ramht.h" + +static u32 +nouveau_ramht_hash_handle(struct nouveau_channel *chan, u32 handle) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_ramht *ramht = chan->ramht; + u32 hash = 0; + int i; + + NV_DEBUG(dev, "ch%d handle=0x%08x\n", chan->id, handle); + + for (i = 32; i > 0; i -= ramht->bits) { + hash ^= (handle & ((1 << ramht->bits) - 1)); + handle >>= ramht->bits; + } + + if (dev_priv->card_type < NV_50) + hash ^= chan->id << (ramht->bits - 4); + hash <<= 3; + + NV_DEBUG(dev, "hash=0x%08x\n", hash); + return hash; +} + +static int +nouveau_ramht_entry_valid(struct drm_device *dev, struct nouveau_gpuobj *ramht, + u32 offset) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + u32 ctx = nv_ro32(ramht, offset + 4); + + if (dev_priv->card_type < NV_40) + return ((ctx & NV_RAMHT_CONTEXT_VALID) != 0); + return (ctx != 0); +} + +static int +nouveau_ramht_entry_same_channel(struct nouveau_channel *chan, + struct nouveau_gpuobj *ramht, u32 offset) +{ + struct drm_nouveau_private *dev_priv = chan->dev->dev_private; + u32 ctx = nv_ro32(ramht, offset + 4); + + if (dev_priv->card_type >= NV_50) + return true; + else if (dev_priv->card_type >= NV_40) + return chan->id == + ((ctx >> NV40_RAMHT_CONTEXT_CHANNEL_SHIFT) & 0x1f); + else + return chan->id == + ((ctx >> NV_RAMHT_CONTEXT_CHANNEL_SHIFT) & 0x1f); +} + +int +nouveau_ramht_insert(struct nouveau_channel *chan, u32 handle, + struct nouveau_gpuobj *gpuobj) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_instmem_engine *instmem = &dev_priv->engine.instmem; + struct nouveau_ramht_entry *entry; + struct nouveau_gpuobj *ramht = chan->ramht->gpuobj; + unsigned long flags; + u32 ctx, co, ho; + + if (nouveau_ramht_find(chan, handle)) + return -EEXIST; + + entry = kmalloc(sizeof(*entry), GFP_KERNEL); + if (!entry) + return -ENOMEM; + entry->channel = chan; + entry->gpuobj = NULL; + entry->handle = handle; + nouveau_gpuobj_ref(gpuobj, &entry->gpuobj); + + if (dev_priv->card_type < NV_40) { + ctx = NV_RAMHT_CONTEXT_VALID | (gpuobj->pinst >> 4) | + (chan->id << NV_RAMHT_CONTEXT_CHANNEL_SHIFT) | + (gpuobj->engine << NV_RAMHT_CONTEXT_ENGINE_SHIFT); + } else + if (dev_priv->card_type < NV_50) { + ctx = (gpuobj->pinst >> 4) | + (chan->id << NV40_RAMHT_CONTEXT_CHANNEL_SHIFT) | + (gpuobj->engine << NV40_RAMHT_CONTEXT_ENGINE_SHIFT); + } else { + if (gpuobj->engine == NVOBJ_ENGINE_DISPLAY) { + ctx = (gpuobj->cinst << 10) | + (chan->id << 28) | + chan->id; /* HASH_TAG */ + } else { + ctx = (gpuobj->cinst >> 4) | + ((gpuobj->engine << + NV40_RAMHT_CONTEXT_ENGINE_SHIFT)); + } + } + + spin_lock_irqsave(&chan->ramht->lock, flags); + list_add(&entry->head, &chan->ramht->entries); + + co = ho = nouveau_ramht_hash_handle(chan, handle); + do { + if (!nouveau_ramht_entry_valid(dev, ramht, co)) { + NV_DEBUG(dev, + "insert ch%d 0x%08x: h=0x%08x, c=0x%08x\n", + chan->id, co, handle, ctx); + nv_wo32(ramht, co + 0, handle); + nv_wo32(ramht, co + 4, ctx); + + spin_unlock_irqrestore(&chan->ramht->lock, flags); + instmem->flush(dev); + return 0; + } + NV_DEBUG(dev, "collision ch%d 0x%08x: h=0x%08x\n", + chan->id, co, nv_ro32(ramht, co)); + + co += 8; + if (co >= ramht->size) + co = 0; + } while (co != ho); + + NV_ERROR(dev, "RAMHT space exhausted. ch=%d\n", chan->id); + list_del(&entry->head); + spin_unlock_irqrestore(&chan->ramht->lock, flags); + kfree(entry); + return -ENOMEM; +} + +static struct nouveau_ramht_entry * +nouveau_ramht_remove_entry(struct nouveau_channel *chan, u32 handle) +{ + struct nouveau_ramht *ramht = chan ? chan->ramht : NULL; + struct nouveau_ramht_entry *entry; + unsigned long flags; + + if (!ramht) + return NULL; + + spin_lock_irqsave(&ramht->lock, flags); + list_for_each_entry(entry, &ramht->entries, head) { + if (entry->channel == chan && + (!handle || entry->handle == handle)) { + list_del(&entry->head); + spin_unlock_irqrestore(&ramht->lock, flags); + + return entry; + } + } + spin_unlock_irqrestore(&ramht->lock, flags); + + return NULL; +} + +static void +nouveau_ramht_remove_hash(struct nouveau_channel *chan, u32 handle) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_instmem_engine *instmem = &dev_priv->engine.instmem; + struct nouveau_gpuobj *ramht = chan->ramht->gpuobj; + unsigned long flags; + u32 co, ho; + + spin_lock_irqsave(&chan->ramht->lock, flags); + co = ho = nouveau_ramht_hash_handle(chan, handle); + do { + if (nouveau_ramht_entry_valid(dev, ramht, co) && + nouveau_ramht_entry_same_channel(chan, ramht, co) && + (handle == nv_ro32(ramht, co))) { + NV_DEBUG(dev, + "remove ch%d 0x%08x: h=0x%08x, c=0x%08x\n", + chan->id, co, handle, nv_ro32(ramht, co + 4)); + nv_wo32(ramht, co + 0, 0x00000000); + nv_wo32(ramht, co + 4, 0x00000000); + instmem->flush(dev); + goto out; + } + + co += 8; + if (co >= ramht->size) + co = 0; + } while (co != ho); + + NV_ERROR(dev, "RAMHT entry not found. ch=%d, handle=0x%08x\n", + chan->id, handle); +out: + spin_unlock_irqrestore(&chan->ramht->lock, flags); +} + +int +nouveau_ramht_remove(struct nouveau_channel *chan, u32 handle) +{ + struct nouveau_ramht_entry *entry; + + entry = nouveau_ramht_remove_entry(chan, handle); + if (!entry) + return -ENOENT; + + nouveau_ramht_remove_hash(chan, entry->handle); + nouveau_gpuobj_ref(NULL, &entry->gpuobj); + kfree(entry); + return 0; +} + +struct nouveau_gpuobj * +nouveau_ramht_find(struct nouveau_channel *chan, u32 handle) +{ + struct nouveau_ramht *ramht = chan->ramht; + struct nouveau_ramht_entry *entry; + struct nouveau_gpuobj *gpuobj = NULL; + unsigned long flags; + + if (unlikely(!chan->ramht)) + return NULL; + + spin_lock_irqsave(&ramht->lock, flags); + list_for_each_entry(entry, &chan->ramht->entries, head) { + if (entry->channel == chan && entry->handle == handle) { + gpuobj = entry->gpuobj; + break; + } + } + spin_unlock_irqrestore(&ramht->lock, flags); + + return gpuobj; +} + +int +nouveau_ramht_new(struct drm_device *dev, struct nouveau_gpuobj *gpuobj, + struct nouveau_ramht **pramht) +{ + struct nouveau_ramht *ramht; + + ramht = kzalloc(sizeof(*ramht), GFP_KERNEL); + if (!ramht) + return -ENOMEM; + + ramht->dev = dev; + kref_init(&ramht->refcount); + ramht->bits = drm_order(gpuobj->size / 8); + INIT_LIST_HEAD(&ramht->entries); + spin_lock_init(&ramht->lock); + nouveau_gpuobj_ref(gpuobj, &ramht->gpuobj); + + *pramht = ramht; + return 0; +} + +static void +nouveau_ramht_del(struct kref *ref) +{ + struct nouveau_ramht *ramht = + container_of(ref, struct nouveau_ramht, refcount); + + nouveau_gpuobj_ref(NULL, &ramht->gpuobj); + kfree(ramht); +} + +void +nouveau_ramht_ref(struct nouveau_ramht *ref, struct nouveau_ramht **ptr, + struct nouveau_channel *chan) +{ + struct nouveau_ramht_entry *entry; + struct nouveau_ramht *ramht; + + if (ref) + kref_get(&ref->refcount); + + ramht = *ptr; + if (ramht) { + while ((entry = nouveau_ramht_remove_entry(chan, 0))) { + nouveau_ramht_remove_hash(chan, entry->handle); + nouveau_gpuobj_ref(NULL, &entry->gpuobj); + kfree(entry); + } + + kref_put(&ramht->refcount, nouveau_ramht_del); + } + *ptr = ref; +} diff --git a/drivers/gpu/drm/nouveau/nouveau_ramht.h b/drivers/gpu/drm/nouveau/nouveau_ramht.h new file mode 100644 index 00000000..c82de98f --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_ramht.h @@ -0,0 +1,55 @@ +/* + * Copyright 2010 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#ifndef __NOUVEAU_RAMHT_H__ +#define __NOUVEAU_RAMHT_H__ + +struct nouveau_ramht_entry { + struct list_head head; + struct nouveau_channel *channel; + struct nouveau_gpuobj *gpuobj; + u32 handle; +}; + +struct nouveau_ramht { + struct drm_device *dev; + struct kref refcount; + spinlock_t lock; + struct nouveau_gpuobj *gpuobj; + struct list_head entries; + int bits; +}; + +extern int nouveau_ramht_new(struct drm_device *, struct nouveau_gpuobj *, + struct nouveau_ramht **); +extern void nouveau_ramht_ref(struct nouveau_ramht *, struct nouveau_ramht **, + struct nouveau_channel *unref_channel); + +extern int nouveau_ramht_insert(struct nouveau_channel *, u32 handle, + struct nouveau_gpuobj *); +extern int nouveau_ramht_remove(struct nouveau_channel *, u32 handle); +extern struct nouveau_gpuobj * +nouveau_ramht_find(struct nouveau_channel *chan, u32 handle); + +#endif diff --git a/drivers/gpu/drm/nouveau/nouveau_reg.h b/drivers/gpu/drm/nouveau/nouveau_reg.h new file mode 100644 index 00000000..f18cdfc3 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_reg.h @@ -0,0 +1,855 @@ + +#define NV04_PFB_BOOT_0 0x00100000 +# define NV04_PFB_BOOT_0_RAM_AMOUNT 0x00000003 +# define NV04_PFB_BOOT_0_RAM_AMOUNT_32MB 0x00000000 +# define NV04_PFB_BOOT_0_RAM_AMOUNT_4MB 0x00000001 +# define NV04_PFB_BOOT_0_RAM_AMOUNT_8MB 0x00000002 +# define NV04_PFB_BOOT_0_RAM_AMOUNT_16MB 0x00000003 +# define NV04_PFB_BOOT_0_RAM_WIDTH_128 0x00000004 +# define NV04_PFB_BOOT_0_RAM_TYPE 0x00000028 +# define NV04_PFB_BOOT_0_RAM_TYPE_SGRAM_8MBIT 0x00000000 +# define NV04_PFB_BOOT_0_RAM_TYPE_SGRAM_16MBIT 0x00000008 +# define NV04_PFB_BOOT_0_RAM_TYPE_SGRAM_16MBIT_4BANK 0x00000010 +# define NV04_PFB_BOOT_0_RAM_TYPE_SDRAM_16MBIT 0x00000018 +# define NV04_PFB_BOOT_0_RAM_TYPE_SDRAM_64MBIT 0x00000020 +# define NV04_PFB_BOOT_0_RAM_TYPE_SDRAM_64MBITX16 0x00000028 +# define NV04_PFB_BOOT_0_UMA_ENABLE 0x00000100 +# define NV04_PFB_BOOT_0_UMA_SIZE 0x0000f000 +#define NV04_PFB_DEBUG_0 0x00100080 +# define NV04_PFB_DEBUG_0_PAGE_MODE 0x00000001 +# define NV04_PFB_DEBUG_0_REFRESH_OFF 0x00000010 +# define NV04_PFB_DEBUG_0_REFRESH_COUNTX64 0x00003f00 +# define NV04_PFB_DEBUG_0_REFRESH_SLOW_CLK 0x00004000 +# define NV04_PFB_DEBUG_0_SAFE_MODE 0x00008000 +# define NV04_PFB_DEBUG_0_ALOM_ENABLE 0x00010000 +# define NV04_PFB_DEBUG_0_CASOE 0x00100000 +# define NV04_PFB_DEBUG_0_CKE_INVERT 0x10000000 +# define NV04_PFB_DEBUG_0_REFINC 0x20000000 +# define NV04_PFB_DEBUG_0_SAVE_POWER_OFF 0x40000000 +#define NV04_PFB_CFG0 0x00100200 +# define NV04_PFB_CFG0_SCRAMBLE 0x20000000 +#define NV04_PFB_CFG1 0x00100204 +#define NV04_PFB_FIFO_DATA 0x0010020c +# define NV10_PFB_FIFO_DATA_RAM_AMOUNT_MB_MASK 0xfff00000 +# define NV10_PFB_FIFO_DATA_RAM_AMOUNT_MB_SHIFT 20 +#define NV10_PFB_REFCTRL 0x00100210 +# define NV10_PFB_REFCTRL_VALID_1 (1 << 31) +#define NV04_PFB_PAD 0x0010021c +# define NV04_PFB_PAD_CKE_NORMAL (1 << 0) +#define NV10_PFB_TILE(i) (0x00100240 + (i*16)) +#define NV10_PFB_TILE__SIZE 8 +#define NV10_PFB_TLIMIT(i) (0x00100244 + (i*16)) +#define NV10_PFB_TSIZE(i) (0x00100248 + (i*16)) +#define NV10_PFB_TSTATUS(i) (0x0010024c + (i*16)) +#define NV04_PFB_REF 0x001002d0 +# define NV04_PFB_REF_CMD_REFRESH (1 << 0) +#define NV04_PFB_PRE 0x001002d4 +# define NV04_PFB_PRE_CMD_PRECHARGE (1 << 0) +#define NV20_PFB_ZCOMP(i) (0x00100300 + 4*(i)) +# define NV20_PFB_ZCOMP_MODE_32 (4 << 24) +# define NV20_PFB_ZCOMP_EN (1 << 31) +# define NV25_PFB_ZCOMP_MODE_16 (1 << 20) +# define NV25_PFB_ZCOMP_MODE_32 (2 << 20) +#define NV10_PFB_CLOSE_PAGE2 0x0010033c +#define NV04_PFB_SCRAMBLE(i) (0x00100400 + 4 * (i)) +#define NV40_PFB_TILE(i) (0x00100600 + (i*16)) +#define NV40_PFB_TILE__SIZE_0 12 +#define NV40_PFB_TILE__SIZE_1 15 +#define NV40_PFB_TLIMIT(i) (0x00100604 + (i*16)) +#define NV40_PFB_TSIZE(i) (0x00100608 + (i*16)) +#define NV40_PFB_TSTATUS(i) (0x0010060c + (i*16)) +#define NV40_PFB_UNK_800 0x00100800 + +#define NV_PEXTDEV_BOOT_0 0x00101000 +#define NV_PEXTDEV_BOOT_0_RAMCFG 0x0000003c +# define NV_PEXTDEV_BOOT_0_STRAP_FP_IFACE_12BIT (8 << 12) +#define NV_PEXTDEV_BOOT_3 0x0010100c + +#define NV_RAMIN 0x00700000 + +#define NV_RAMHT_HANDLE_OFFSET 0 +#define NV_RAMHT_CONTEXT_OFFSET 4 +# define NV_RAMHT_CONTEXT_VALID (1<<31) +# define NV_RAMHT_CONTEXT_CHANNEL_SHIFT 24 +# define NV_RAMHT_CONTEXT_ENGINE_SHIFT 16 +# define NV_RAMHT_CONTEXT_ENGINE_SOFTWARE 0 +# define NV_RAMHT_CONTEXT_ENGINE_GRAPHICS 1 +# define NV_RAMHT_CONTEXT_INSTANCE_SHIFT 0 +# define NV40_RAMHT_CONTEXT_CHANNEL_SHIFT 23 +# define NV40_RAMHT_CONTEXT_ENGINE_SHIFT 20 +# define NV40_RAMHT_CONTEXT_INSTANCE_SHIFT 0 + +/* Some object classes we care about in the drm */ +#define NV_CLASS_DMA_FROM_MEMORY 0x00000002 +#define NV_CLASS_DMA_TO_MEMORY 0x00000003 +#define NV_CLASS_NULL 0x00000030 +#define NV_CLASS_DMA_IN_MEMORY 0x0000003D + +#define NV03_USER(i) (0x00800000+(i*NV03_USER_SIZE)) +#define NV03_USER__SIZE 16 +#define NV10_USER__SIZE 32 +#define NV03_USER_SIZE 0x00010000 +#define NV03_USER_DMA_PUT(i) (0x00800040+(i*NV03_USER_SIZE)) +#define NV03_USER_DMA_PUT__SIZE 16 +#define NV10_USER_DMA_PUT__SIZE 32 +#define NV03_USER_DMA_GET(i) (0x00800044+(i*NV03_USER_SIZE)) +#define NV03_USER_DMA_GET__SIZE 16 +#define NV10_USER_DMA_GET__SIZE 32 +#define NV03_USER_REF_CNT(i) (0x00800048+(i*NV03_USER_SIZE)) +#define NV03_USER_REF_CNT__SIZE 16 +#define NV10_USER_REF_CNT__SIZE 32 + +#define NV40_USER(i) (0x00c00000+(i*NV40_USER_SIZE)) +#define NV40_USER_SIZE 0x00001000 +#define NV40_USER_DMA_PUT(i) (0x00c00040+(i*NV40_USER_SIZE)) +#define NV40_USER_DMA_PUT__SIZE 32 +#define NV40_USER_DMA_GET(i) (0x00c00044+(i*NV40_USER_SIZE)) +#define NV40_USER_DMA_GET__SIZE 32 +#define NV40_USER_REF_CNT(i) (0x00c00048+(i*NV40_USER_SIZE)) +#define NV40_USER_REF_CNT__SIZE 32 + +#define NV50_USER(i) (0x00c00000+(i*NV50_USER_SIZE)) +#define NV50_USER_SIZE 0x00002000 +#define NV50_USER_DMA_PUT(i) (0x00c00040+(i*NV50_USER_SIZE)) +#define NV50_USER_DMA_PUT__SIZE 128 +#define NV50_USER_DMA_GET(i) (0x00c00044+(i*NV50_USER_SIZE)) +#define NV50_USER_DMA_GET__SIZE 128 +#define NV50_USER_REF_CNT(i) (0x00c00048+(i*NV50_USER_SIZE)) +#define NV50_USER_REF_CNT__SIZE 128 + +#define NV03_FIFO_SIZE 0x8000UL + +#define NV03_PMC_BOOT_0 0x00000000 +#define NV03_PMC_BOOT_1 0x00000004 +#define NV03_PMC_INTR_0 0x00000100 +# define NV_PMC_INTR_0_PFIFO_PENDING (1<<8) +# define NV_PMC_INTR_0_PGRAPH_PENDING (1<<12) +# define NV_PMC_INTR_0_NV50_I2C_PENDING (1<<21) +# define NV_PMC_INTR_0_CRTC0_PENDING (1<<24) +# define NV_PMC_INTR_0_CRTC1_PENDING (1<<25) +# define NV_PMC_INTR_0_NV50_DISPLAY_PENDING (1<<26) +# define NV_PMC_INTR_0_CRTCn_PENDING (3<<24) +#define NV03_PMC_INTR_EN_0 0x00000140 +# define NV_PMC_INTR_EN_0_MASTER_ENABLE (1<<0) +#define NV03_PMC_ENABLE 0x00000200 +# define NV_PMC_ENABLE_PFIFO (1<<8) +# define NV_PMC_ENABLE_PGRAPH (1<<12) +/* Disabling the below bit breaks newer (G7X only?) mobile chipsets, + * the card will hang early on in the X init process. + */ +# define NV_PMC_ENABLE_UNK13 (1<<13) +#define NV40_PMC_GRAPH_UNITS 0x00001540 +#define NV40_PMC_BACKLIGHT 0x000015f0 +# define NV40_PMC_BACKLIGHT_MASK 0x001f0000 +#define NV40_PMC_1700 0x00001700 +#define NV40_PMC_1704 0x00001704 +#define NV40_PMC_1708 0x00001708 +#define NV40_PMC_170C 0x0000170C + +/* probably PMC ? */ +#define NV50_PUNK_BAR0_PRAMIN 0x00001700 +#define NV50_PUNK_BAR_CFG_BASE 0x00001704 +#define NV50_PUNK_BAR_CFG_BASE_VALID (1<<30) +#define NV50_PUNK_BAR1_CTXDMA 0x00001708 +#define NV50_PUNK_BAR1_CTXDMA_VALID (1<<31) +#define NV50_PUNK_BAR3_CTXDMA 0x0000170C +#define NV50_PUNK_BAR3_CTXDMA_VALID (1<<31) +#define NV50_PUNK_UNK1710 0x00001710 + +#define NV04_PBUS_PCI_NV_1 0x00001804 +#define NV04_PBUS_PCI_NV_19 0x0000184C +#define NV04_PBUS_PCI_NV_20 0x00001850 +# define NV04_PBUS_PCI_NV_20_ROM_SHADOW_DISABLED (0 << 0) +# define NV04_PBUS_PCI_NV_20_ROM_SHADOW_ENABLED (1 << 0) + +#define NV04_PTIMER_INTR_0 0x00009100 +#define NV04_PTIMER_INTR_EN_0 0x00009140 +#define NV04_PTIMER_NUMERATOR 0x00009200 +#define NV04_PTIMER_DENOMINATOR 0x00009210 +#define NV04_PTIMER_TIME_0 0x00009400 +#define NV04_PTIMER_TIME_1 0x00009410 +#define NV04_PTIMER_ALARM_0 0x00009420 + +#define NV04_PGRAPH_DEBUG_0 0x00400080 +#define NV04_PGRAPH_DEBUG_1 0x00400084 +#define NV04_PGRAPH_DEBUG_2 0x00400088 +#define NV04_PGRAPH_DEBUG_3 0x0040008c +#define NV10_PGRAPH_DEBUG_4 0x00400090 +#define NV03_PGRAPH_INTR 0x00400100 +#define NV03_PGRAPH_NSTATUS 0x00400104 +# define NV04_PGRAPH_NSTATUS_STATE_IN_USE (1<<11) +# define NV04_PGRAPH_NSTATUS_INVALID_STATE (1<<12) +# define NV04_PGRAPH_NSTATUS_BAD_ARGUMENT (1<<13) +# define NV04_PGRAPH_NSTATUS_PROTECTION_FAULT (1<<14) +# define NV10_PGRAPH_NSTATUS_STATE_IN_USE (1<<23) +# define NV10_PGRAPH_NSTATUS_INVALID_STATE (1<<24) +# define NV10_PGRAPH_NSTATUS_BAD_ARGUMENT (1<<25) +# define NV10_PGRAPH_NSTATUS_PROTECTION_FAULT (1<<26) +#define NV03_PGRAPH_NSOURCE 0x00400108 +# define NV03_PGRAPH_NSOURCE_NOTIFICATION (1<<0) +# define NV03_PGRAPH_NSOURCE_DATA_ERROR (1<<1) +# define NV03_PGRAPH_NSOURCE_PROTECTION_ERROR (1<<2) +# define NV03_PGRAPH_NSOURCE_RANGE_EXCEPTION (1<<3) +# define NV03_PGRAPH_NSOURCE_LIMIT_COLOR (1<<4) +# define NV03_PGRAPH_NSOURCE_LIMIT_ZETA (1<<5) +# define NV03_PGRAPH_NSOURCE_ILLEGAL_MTHD (1<<6) +# define NV03_PGRAPH_NSOURCE_DMA_R_PROTECTION (1<<7) +# define NV03_PGRAPH_NSOURCE_DMA_W_PROTECTION (1<<8) +# define NV03_PGRAPH_NSOURCE_FORMAT_EXCEPTION (1<<9) +# define NV03_PGRAPH_NSOURCE_PATCH_EXCEPTION (1<<10) +# define NV03_PGRAPH_NSOURCE_STATE_INVALID (1<<11) +# define NV03_PGRAPH_NSOURCE_DOUBLE_NOTIFY (1<<12) +# define NV03_PGRAPH_NSOURCE_NOTIFY_IN_USE (1<<13) +# define NV03_PGRAPH_NSOURCE_METHOD_CNT (1<<14) +# define NV03_PGRAPH_NSOURCE_BFR_NOTIFICATION (1<<15) +# define NV03_PGRAPH_NSOURCE_DMA_VTX_PROTECTION (1<<16) +# define NV03_PGRAPH_NSOURCE_DMA_WIDTH_A (1<<17) +# define NV03_PGRAPH_NSOURCE_DMA_WIDTH_B (1<<18) +#define NV03_PGRAPH_INTR_EN 0x00400140 +#define NV40_PGRAPH_INTR_EN 0x0040013C +# define NV_PGRAPH_INTR_NOTIFY (1<<0) +# define NV_PGRAPH_INTR_MISSING_HW (1<<4) +# define NV_PGRAPH_INTR_CONTEXT_SWITCH (1<<12) +# define NV_PGRAPH_INTR_BUFFER_NOTIFY (1<<16) +# define NV_PGRAPH_INTR_ERROR (1<<20) +#define NV10_PGRAPH_CTX_CONTROL 0x00400144 +#define NV10_PGRAPH_CTX_USER 0x00400148 +#define NV10_PGRAPH_CTX_SWITCH(i) (0x0040014C + 0x4*(i)) +#define NV04_PGRAPH_CTX_SWITCH1 0x00400160 +#define NV10_PGRAPH_CTX_CACHE(i, j) (0x00400160 \ + + 0x4*(i) + 0x20*(j)) +#define NV04_PGRAPH_CTX_SWITCH2 0x00400164 +#define NV04_PGRAPH_CTX_SWITCH3 0x00400168 +#define NV04_PGRAPH_CTX_SWITCH4 0x0040016C +#define NV04_PGRAPH_CTX_CONTROL 0x00400170 +#define NV04_PGRAPH_CTX_USER 0x00400174 +#define NV04_PGRAPH_CTX_CACHE1 0x00400180 +#define NV03_PGRAPH_CTX_CONTROL 0x00400190 +#define NV03_PGRAPH_CTX_USER 0x00400194 +#define NV04_PGRAPH_CTX_CACHE2 0x004001A0 +#define NV04_PGRAPH_CTX_CACHE3 0x004001C0 +#define NV04_PGRAPH_CTX_CACHE4 0x004001E0 +#define NV40_PGRAPH_CTXCTL_0304 0x00400304 +#define NV40_PGRAPH_CTXCTL_0304_XFER_CTX 0x00000001 +#define NV40_PGRAPH_CTXCTL_UCODE_STAT 0x00400308 +#define NV40_PGRAPH_CTXCTL_UCODE_STAT_IP_MASK 0xff000000 +#define NV40_PGRAPH_CTXCTL_UCODE_STAT_IP_SHIFT 24 +#define NV40_PGRAPH_CTXCTL_UCODE_STAT_OP_MASK 0x00ffffff +#define NV40_PGRAPH_CTXCTL_0310 0x00400310 +#define NV40_PGRAPH_CTXCTL_0310_XFER_SAVE 0x00000020 +#define NV40_PGRAPH_CTXCTL_0310_XFER_LOAD 0x00000040 +#define NV40_PGRAPH_CTXCTL_030C 0x0040030c +#define NV40_PGRAPH_CTXCTL_UCODE_INDEX 0x00400324 +#define NV40_PGRAPH_CTXCTL_UCODE_DATA 0x00400328 +#define NV40_PGRAPH_CTXCTL_CUR 0x0040032c +#define NV40_PGRAPH_CTXCTL_CUR_LOADED 0x01000000 +#define NV40_PGRAPH_CTXCTL_CUR_INSTANCE 0x000FFFFF +#define NV40_PGRAPH_CTXCTL_NEXT 0x00400330 +#define NV40_PGRAPH_CTXCTL_NEXT_INSTANCE 0x000fffff +#define NV50_PGRAPH_CTXCTL_CUR 0x0040032c +#define NV50_PGRAPH_CTXCTL_CUR_LOADED 0x80000000 +#define NV50_PGRAPH_CTXCTL_CUR_INSTANCE 0x00ffffff +#define NV50_PGRAPH_CTXCTL_NEXT 0x00400330 +#define NV50_PGRAPH_CTXCTL_NEXT_INSTANCE 0x00ffffff +#define NV03_PGRAPH_ABS_X_RAM 0x00400400 +#define NV03_PGRAPH_ABS_Y_RAM 0x00400480 +#define NV03_PGRAPH_X_MISC 0x00400500 +#define NV03_PGRAPH_Y_MISC 0x00400504 +#define NV04_PGRAPH_VALID1 0x00400508 +#define NV04_PGRAPH_SOURCE_COLOR 0x0040050C +#define NV04_PGRAPH_MISC24_0 0x00400510 +#define NV03_PGRAPH_XY_LOGIC_MISC0 0x00400514 +#define NV03_PGRAPH_XY_LOGIC_MISC1 0x00400518 +#define NV03_PGRAPH_XY_LOGIC_MISC2 0x0040051C +#define NV03_PGRAPH_XY_LOGIC_MISC3 0x00400520 +#define NV03_PGRAPH_CLIPX_0 0x00400524 +#define NV03_PGRAPH_CLIPX_1 0x00400528 +#define NV03_PGRAPH_CLIPY_0 0x0040052C +#define NV03_PGRAPH_CLIPY_1 0x00400530 +#define NV03_PGRAPH_ABS_ICLIP_XMAX 0x00400534 +#define NV03_PGRAPH_ABS_ICLIP_YMAX 0x00400538 +#define NV03_PGRAPH_ABS_UCLIP_XMIN 0x0040053C +#define NV03_PGRAPH_ABS_UCLIP_YMIN 0x00400540 +#define NV03_PGRAPH_ABS_UCLIP_XMAX 0x00400544 +#define NV03_PGRAPH_ABS_UCLIP_YMAX 0x00400548 +#define NV03_PGRAPH_ABS_UCLIPA_XMIN 0x00400560 +#define NV03_PGRAPH_ABS_UCLIPA_YMIN 0x00400564 +#define NV03_PGRAPH_ABS_UCLIPA_XMAX 0x00400568 +#define NV03_PGRAPH_ABS_UCLIPA_YMAX 0x0040056C +#define NV04_PGRAPH_MISC24_1 0x00400570 +#define NV04_PGRAPH_MISC24_2 0x00400574 +#define NV04_PGRAPH_VALID2 0x00400578 +#define NV04_PGRAPH_PASSTHRU_0 0x0040057C +#define NV04_PGRAPH_PASSTHRU_1 0x00400580 +#define NV04_PGRAPH_PASSTHRU_2 0x00400584 +#define NV10_PGRAPH_DIMX_TEXTURE 0x00400588 +#define NV10_PGRAPH_WDIMX_TEXTURE 0x0040058C +#define NV04_PGRAPH_COMBINE_0_ALPHA 0x00400590 +#define NV04_PGRAPH_COMBINE_0_COLOR 0x00400594 +#define NV04_PGRAPH_COMBINE_1_ALPHA 0x00400598 +#define NV04_PGRAPH_COMBINE_1_COLOR 0x0040059C +#define NV04_PGRAPH_FORMAT_0 0x004005A8 +#define NV04_PGRAPH_FORMAT_1 0x004005AC +#define NV04_PGRAPH_FILTER_0 0x004005B0 +#define NV04_PGRAPH_FILTER_1 0x004005B4 +#define NV03_PGRAPH_MONO_COLOR0 0x00400600 +#define NV04_PGRAPH_ROP3 0x00400604 +#define NV04_PGRAPH_BETA_AND 0x00400608 +#define NV04_PGRAPH_BETA_PREMULT 0x0040060C +#define NV04_PGRAPH_LIMIT_VIOL_PIX 0x00400610 +#define NV04_PGRAPH_FORMATS 0x00400618 +#define NV10_PGRAPH_DEBUG_2 0x00400620 +#define NV04_PGRAPH_BOFFSET0 0x00400640 +#define NV04_PGRAPH_BOFFSET1 0x00400644 +#define NV04_PGRAPH_BOFFSET2 0x00400648 +#define NV04_PGRAPH_BOFFSET3 0x0040064C +#define NV04_PGRAPH_BOFFSET4 0x00400650 +#define NV04_PGRAPH_BOFFSET5 0x00400654 +#define NV04_PGRAPH_BBASE0 0x00400658 +#define NV04_PGRAPH_BBASE1 0x0040065C +#define NV04_PGRAPH_BBASE2 0x00400660 +#define NV04_PGRAPH_BBASE3 0x00400664 +#define NV04_PGRAPH_BBASE4 0x00400668 +#define NV04_PGRAPH_BBASE5 0x0040066C +#define NV04_PGRAPH_BPITCH0 0x00400670 +#define NV04_PGRAPH_BPITCH1 0x00400674 +#define NV04_PGRAPH_BPITCH2 0x00400678 +#define NV04_PGRAPH_BPITCH3 0x0040067C +#define NV04_PGRAPH_BPITCH4 0x00400680 +#define NV04_PGRAPH_BLIMIT0 0x00400684 +#define NV04_PGRAPH_BLIMIT1 0x00400688 +#define NV04_PGRAPH_BLIMIT2 0x0040068C +#define NV04_PGRAPH_BLIMIT3 0x00400690 +#define NV04_PGRAPH_BLIMIT4 0x00400694 +#define NV04_PGRAPH_BLIMIT5 0x00400698 +#define NV04_PGRAPH_BSWIZZLE2 0x0040069C +#define NV04_PGRAPH_BSWIZZLE5 0x004006A0 +#define NV03_PGRAPH_STATUS 0x004006B0 +#define NV04_PGRAPH_STATUS 0x00400700 +# define NV40_PGRAPH_STATUS_SYNC_STALL 0x00004000 +#define NV04_PGRAPH_TRAPPED_ADDR 0x00400704 +#define NV04_PGRAPH_TRAPPED_DATA 0x00400708 +#define NV04_PGRAPH_SURFACE 0x0040070C +#define NV10_PGRAPH_TRAPPED_DATA_HIGH 0x0040070C +#define NV04_PGRAPH_STATE 0x00400710 +#define NV10_PGRAPH_SURFACE 0x00400710 +#define NV04_PGRAPH_NOTIFY 0x00400714 +#define NV10_PGRAPH_STATE 0x00400714 +#define NV10_PGRAPH_NOTIFY 0x00400718 + +#define NV04_PGRAPH_FIFO 0x00400720 + +#define NV04_PGRAPH_BPIXEL 0x00400724 +#define NV10_PGRAPH_RDI_INDEX 0x00400750 +#define NV04_PGRAPH_FFINTFC_ST2 0x00400754 +#define NV10_PGRAPH_RDI_DATA 0x00400754 +#define NV04_PGRAPH_DMA_PITCH 0x00400760 +#define NV10_PGRAPH_FFINTFC_FIFO_PTR 0x00400760 +#define NV04_PGRAPH_DVD_COLORFMT 0x00400764 +#define NV10_PGRAPH_FFINTFC_ST2 0x00400764 +#define NV04_PGRAPH_SCALED_FORMAT 0x00400768 +#define NV10_PGRAPH_FFINTFC_ST2_DL 0x00400768 +#define NV10_PGRAPH_FFINTFC_ST2_DH 0x0040076c +#define NV10_PGRAPH_DMA_PITCH 0x00400770 +#define NV10_PGRAPH_DVD_COLORFMT 0x00400774 +#define NV10_PGRAPH_SCALED_FORMAT 0x00400778 +#define NV20_PGRAPH_CHANNEL_CTX_TABLE 0x00400780 +#define NV20_PGRAPH_CHANNEL_CTX_POINTER 0x00400784 +#define NV20_PGRAPH_CHANNEL_CTX_XFER 0x00400788 +#define NV20_PGRAPH_CHANNEL_CTX_XFER_LOAD 0x00000001 +#define NV20_PGRAPH_CHANNEL_CTX_XFER_SAVE 0x00000002 +#define NV04_PGRAPH_PATT_COLOR0 0x00400800 +#define NV04_PGRAPH_PATT_COLOR1 0x00400804 +#define NV04_PGRAPH_PATTERN 0x00400808 +#define NV04_PGRAPH_PATTERN_SHAPE 0x00400810 +#define NV04_PGRAPH_CHROMA 0x00400814 +#define NV04_PGRAPH_CONTROL0 0x00400818 +#define NV04_PGRAPH_CONTROL1 0x0040081C +#define NV04_PGRAPH_CONTROL2 0x00400820 +#define NV04_PGRAPH_BLEND 0x00400824 +#define NV04_PGRAPH_STORED_FMT 0x00400830 +#define NV04_PGRAPH_PATT_COLORRAM 0x00400900 +#define NV20_PGRAPH_TILE(i) (0x00400900 + (i*16)) +#define NV20_PGRAPH_TLIMIT(i) (0x00400904 + (i*16)) +#define NV20_PGRAPH_TSIZE(i) (0x00400908 + (i*16)) +#define NV20_PGRAPH_TSTATUS(i) (0x0040090C + (i*16)) +#define NV20_PGRAPH_ZCOMP(i) (0x00400980 + 4*(i)) +#define NV10_PGRAPH_TILE(i) (0x00400B00 + (i*16)) +#define NV10_PGRAPH_TLIMIT(i) (0x00400B04 + (i*16)) +#define NV10_PGRAPH_TSIZE(i) (0x00400B08 + (i*16)) +#define NV10_PGRAPH_TSTATUS(i) (0x00400B0C + (i*16)) +#define NV04_PGRAPH_U_RAM 0x00400D00 +#define NV47_PGRAPH_TILE(i) (0x00400D00 + (i*16)) +#define NV47_PGRAPH_TLIMIT(i) (0x00400D04 + (i*16)) +#define NV47_PGRAPH_TSIZE(i) (0x00400D08 + (i*16)) +#define NV47_PGRAPH_TSTATUS(i) (0x00400D0C + (i*16)) +#define NV04_PGRAPH_V_RAM 0x00400D40 +#define NV04_PGRAPH_W_RAM 0x00400D80 +#define NV10_PGRAPH_COMBINER0_IN_ALPHA 0x00400E40 +#define NV10_PGRAPH_COMBINER1_IN_ALPHA 0x00400E44 +#define NV10_PGRAPH_COMBINER0_IN_RGB 0x00400E48 +#define NV10_PGRAPH_COMBINER1_IN_RGB 0x00400E4C +#define NV10_PGRAPH_COMBINER_COLOR0 0x00400E50 +#define NV10_PGRAPH_COMBINER_COLOR1 0x00400E54 +#define NV10_PGRAPH_COMBINER0_OUT_ALPHA 0x00400E58 +#define NV10_PGRAPH_COMBINER1_OUT_ALPHA 0x00400E5C +#define NV10_PGRAPH_COMBINER0_OUT_RGB 0x00400E60 +#define NV10_PGRAPH_COMBINER1_OUT_RGB 0x00400E64 +#define NV10_PGRAPH_COMBINER_FINAL0 0x00400E68 +#define NV10_PGRAPH_COMBINER_FINAL1 0x00400E6C +#define NV10_PGRAPH_WINDOWCLIP_HORIZONTAL 0x00400F00 +#define NV10_PGRAPH_WINDOWCLIP_VERTICAL 0x00400F20 +#define NV10_PGRAPH_XFMODE0 0x00400F40 +#define NV10_PGRAPH_XFMODE1 0x00400F44 +#define NV10_PGRAPH_GLOBALSTATE0 0x00400F48 +#define NV10_PGRAPH_GLOBALSTATE1 0x00400F4C +#define NV10_PGRAPH_PIPE_ADDRESS 0x00400F50 +#define NV10_PGRAPH_PIPE_DATA 0x00400F54 +#define NV04_PGRAPH_DMA_START_0 0x00401000 +#define NV04_PGRAPH_DMA_START_1 0x00401004 +#define NV04_PGRAPH_DMA_LENGTH 0x00401008 +#define NV04_PGRAPH_DMA_MISC 0x0040100C +#define NV04_PGRAPH_DMA_DATA_0 0x00401020 +#define NV04_PGRAPH_DMA_DATA_1 0x00401024 +#define NV04_PGRAPH_DMA_RM 0x00401030 +#define NV04_PGRAPH_DMA_A_XLATE_INST 0x00401040 +#define NV04_PGRAPH_DMA_A_CONTROL 0x00401044 +#define NV04_PGRAPH_DMA_A_LIMIT 0x00401048 +#define NV04_PGRAPH_DMA_A_TLB_PTE 0x0040104C +#define NV04_PGRAPH_DMA_A_TLB_TAG 0x00401050 +#define NV04_PGRAPH_DMA_A_ADJ_OFFSET 0x00401054 +#define NV04_PGRAPH_DMA_A_OFFSET 0x00401058 +#define NV04_PGRAPH_DMA_A_SIZE 0x0040105C +#define NV04_PGRAPH_DMA_A_Y_SIZE 0x00401060 +#define NV04_PGRAPH_DMA_B_XLATE_INST 0x00401080 +#define NV04_PGRAPH_DMA_B_CONTROL 0x00401084 +#define NV04_PGRAPH_DMA_B_LIMIT 0x00401088 +#define NV04_PGRAPH_DMA_B_TLB_PTE 0x0040108C +#define NV04_PGRAPH_DMA_B_TLB_TAG 0x00401090 +#define NV04_PGRAPH_DMA_B_ADJ_OFFSET 0x00401094 +#define NV04_PGRAPH_DMA_B_OFFSET 0x00401098 +#define NV04_PGRAPH_DMA_B_SIZE 0x0040109C +#define NV04_PGRAPH_DMA_B_Y_SIZE 0x004010A0 +#define NV40_PGRAPH_TILE1(i) (0x00406900 + (i*16)) +#define NV40_PGRAPH_TLIMIT1(i) (0x00406904 + (i*16)) +#define NV40_PGRAPH_TSIZE1(i) (0x00406908 + (i*16)) +#define NV40_PGRAPH_TSTATUS1(i) (0x0040690C + (i*16)) + + +/* It's a guess that this works on NV03. Confirmed on NV04, though */ +#define NV04_PFIFO_DELAY_0 0x00002040 +#define NV04_PFIFO_DMA_TIMESLICE 0x00002044 +#define NV04_PFIFO_NEXT_CHANNEL 0x00002050 +#define NV03_PFIFO_INTR_0 0x00002100 +#define NV03_PFIFO_INTR_EN_0 0x00002140 +# define NV_PFIFO_INTR_CACHE_ERROR (1<<0) +# define NV_PFIFO_INTR_RUNOUT (1<<4) +# define NV_PFIFO_INTR_RUNOUT_OVERFLOW (1<<8) +# define NV_PFIFO_INTR_DMA_PUSHER (1<<12) +# define NV_PFIFO_INTR_DMA_PT (1<<16) +# define NV_PFIFO_INTR_SEMAPHORE (1<<20) +# define NV_PFIFO_INTR_ACQUIRE_TIMEOUT (1<<24) +#define NV03_PFIFO_RAMHT 0x00002210 +#define NV03_PFIFO_RAMFC 0x00002214 +#define NV03_PFIFO_RAMRO 0x00002218 +#define NV40_PFIFO_RAMFC 0x00002220 +#define NV03_PFIFO_CACHES 0x00002500 +#define NV04_PFIFO_MODE 0x00002504 +#define NV04_PFIFO_DMA 0x00002508 +#define NV04_PFIFO_SIZE 0x0000250c +#define NV50_PFIFO_CTX_TABLE(c) (0x2600+(c)*4) +#define NV50_PFIFO_CTX_TABLE__SIZE 128 +#define NV50_PFIFO_CTX_TABLE_CHANNEL_ENABLED (1<<31) +#define NV50_PFIFO_CTX_TABLE_UNK30_BAD (1<<30) +#define NV50_PFIFO_CTX_TABLE_INSTANCE_MASK_G80 0x0FFFFFFF +#define NV50_PFIFO_CTX_TABLE_INSTANCE_MASK_G84 0x00FFFFFF +#define NV03_PFIFO_CACHE0_PUSH0 0x00003000 +#define NV03_PFIFO_CACHE0_PULL0 0x00003040 +#define NV04_PFIFO_CACHE0_PULL0 0x00003050 +#define NV04_PFIFO_CACHE0_PULL1 0x00003054 +#define NV03_PFIFO_CACHE1_PUSH0 0x00003200 +#define NV03_PFIFO_CACHE1_PUSH1 0x00003204 +#define NV03_PFIFO_CACHE1_PUSH1_DMA (1<<8) +#define NV40_PFIFO_CACHE1_PUSH1_DMA (1<<16) +#define NV03_PFIFO_CACHE1_PUSH1_CHID_MASK 0x0000000f +#define NV10_PFIFO_CACHE1_PUSH1_CHID_MASK 0x0000001f +#define NV50_PFIFO_CACHE1_PUSH1_CHID_MASK 0x0000007f +#define NV03_PFIFO_CACHE1_PUT 0x00003210 +#define NV04_PFIFO_CACHE1_DMA_PUSH 0x00003220 +#define NV04_PFIFO_CACHE1_DMA_FETCH 0x00003224 +# define NV_PFIFO_CACHE1_DMA_FETCH_TRIG_8_BYTES 0x00000000 +# define NV_PFIFO_CACHE1_DMA_FETCH_TRIG_16_BYTES 0x00000008 +# define NV_PFIFO_CACHE1_DMA_FETCH_TRIG_24_BYTES 0x00000010 +# define NV_PFIFO_CACHE1_DMA_FETCH_TRIG_32_BYTES 0x00000018 +# define NV_PFIFO_CACHE1_DMA_FETCH_TRIG_40_BYTES 0x00000020 +# define NV_PFIFO_CACHE1_DMA_FETCH_TRIG_48_BYTES 0x00000028 +# define NV_PFIFO_CACHE1_DMA_FETCH_TRIG_56_BYTES 0x00000030 +# define NV_PFIFO_CACHE1_DMA_FETCH_TRIG_64_BYTES 0x00000038 +# define NV_PFIFO_CACHE1_DMA_FETCH_TRIG_72_BYTES 0x00000040 +# define NV_PFIFO_CACHE1_DMA_FETCH_TRIG_80_BYTES 0x00000048 +# define NV_PFIFO_CACHE1_DMA_FETCH_TRIG_88_BYTES 0x00000050 +# define NV_PFIFO_CACHE1_DMA_FETCH_TRIG_96_BYTES 0x00000058 +# define NV_PFIFO_CACHE1_DMA_FETCH_TRIG_104_BYTES 0x00000060 +# define NV_PFIFO_CACHE1_DMA_FETCH_TRIG_112_BYTES 0x00000068 +# define NV_PFIFO_CACHE1_DMA_FETCH_TRIG_120_BYTES 0x00000070 +# define NV_PFIFO_CACHE1_DMA_FETCH_TRIG_128_BYTES 0x00000078 +# define NV_PFIFO_CACHE1_DMA_FETCH_TRIG_136_BYTES 0x00000080 +# define NV_PFIFO_CACHE1_DMA_FETCH_TRIG_144_BYTES 0x00000088 +# define NV_PFIFO_CACHE1_DMA_FETCH_TRIG_152_BYTES 0x00000090 +# define NV_PFIFO_CACHE1_DMA_FETCH_TRIG_160_BYTES 0x00000098 +# define NV_PFIFO_CACHE1_DMA_FETCH_TRIG_168_BYTES 0x000000A0 +# define NV_PFIFO_CACHE1_DMA_FETCH_TRIG_176_BYTES 0x000000A8 +# define NV_PFIFO_CACHE1_DMA_FETCH_TRIG_184_BYTES 0x000000B0 +# define NV_PFIFO_CACHE1_DMA_FETCH_TRIG_192_BYTES 0x000000B8 +# define NV_PFIFO_CACHE1_DMA_FETCH_TRIG_200_BYTES 0x000000C0 +# define NV_PFIFO_CACHE1_DMA_FETCH_TRIG_208_BYTES 0x000000C8 +# define NV_PFIFO_CACHE1_DMA_FETCH_TRIG_216_BYTES 0x000000D0 +# define NV_PFIFO_CACHE1_DMA_FETCH_TRIG_224_BYTES 0x000000D8 +# define NV_PFIFO_CACHE1_DMA_FETCH_TRIG_232_BYTES 0x000000E0 +# define NV_PFIFO_CACHE1_DMA_FETCH_TRIG_240_BYTES 0x000000E8 +# define NV_PFIFO_CACHE1_DMA_FETCH_TRIG_248_BYTES 0x000000F0 +# define NV_PFIFO_CACHE1_DMA_FETCH_TRIG_256_BYTES 0x000000F8 +# define NV_PFIFO_CACHE1_DMA_FETCH_SIZE 0x0000E000 +# define NV_PFIFO_CACHE1_DMA_FETCH_SIZE_32_BYTES 0x00000000 +# define NV_PFIFO_CACHE1_DMA_FETCH_SIZE_64_BYTES 0x00002000 +# define NV_PFIFO_CACHE1_DMA_FETCH_SIZE_96_BYTES 0x00004000 +# define NV_PFIFO_CACHE1_DMA_FETCH_SIZE_128_BYTES 0x00006000 +# define NV_PFIFO_CACHE1_DMA_FETCH_SIZE_160_BYTES 0x00008000 +# define NV_PFIFO_CACHE1_DMA_FETCH_SIZE_192_BYTES 0x0000A000 +# define NV_PFIFO_CACHE1_DMA_FETCH_SIZE_224_BYTES 0x0000C000 +# define NV_PFIFO_CACHE1_DMA_FETCH_SIZE_256_BYTES 0x0000E000 +# define NV_PFIFO_CACHE1_DMA_FETCH_MAX_REQS 0x001F0000 +# define NV_PFIFO_CACHE1_DMA_FETCH_MAX_REQS_0 0x00000000 +# define NV_PFIFO_CACHE1_DMA_FETCH_MAX_REQS_1 0x00010000 +# define NV_PFIFO_CACHE1_DMA_FETCH_MAX_REQS_2 0x00020000 +# define NV_PFIFO_CACHE1_DMA_FETCH_MAX_REQS_3 0x00030000 +# define NV_PFIFO_CACHE1_DMA_FETCH_MAX_REQS_4 0x00040000 +# define NV_PFIFO_CACHE1_DMA_FETCH_MAX_REQS_5 0x00050000 +# define NV_PFIFO_CACHE1_DMA_FETCH_MAX_REQS_6 0x00060000 +# define NV_PFIFO_CACHE1_DMA_FETCH_MAX_REQS_7 0x00070000 +# define NV_PFIFO_CACHE1_DMA_FETCH_MAX_REQS_8 0x00080000 +# define NV_PFIFO_CACHE1_DMA_FETCH_MAX_REQS_9 0x00090000 +# define NV_PFIFO_CACHE1_DMA_FETCH_MAX_REQS_10 0x000A0000 +# define NV_PFIFO_CACHE1_DMA_FETCH_MAX_REQS_11 0x000B0000 +# define NV_PFIFO_CACHE1_DMA_FETCH_MAX_REQS_12 0x000C0000 +# define NV_PFIFO_CACHE1_DMA_FETCH_MAX_REQS_13 0x000D0000 +# define NV_PFIFO_CACHE1_DMA_FETCH_MAX_REQS_14 0x000E0000 +# define NV_PFIFO_CACHE1_DMA_FETCH_MAX_REQS_15 0x000F0000 +# define NV_PFIFO_CACHE1_ENDIAN 0x80000000 +# define NV_PFIFO_CACHE1_LITTLE_ENDIAN 0x7FFFFFFF +# define NV_PFIFO_CACHE1_BIG_ENDIAN 0x80000000 +#define NV04_PFIFO_CACHE1_DMA_STATE 0x00003228 +#define NV04_PFIFO_CACHE1_DMA_INSTANCE 0x0000322c +#define NV04_PFIFO_CACHE1_DMA_CTL 0x00003230 +#define NV04_PFIFO_CACHE1_DMA_PUT 0x00003240 +#define NV04_PFIFO_CACHE1_DMA_GET 0x00003244 +#define NV10_PFIFO_CACHE1_REF_CNT 0x00003248 +#define NV10_PFIFO_CACHE1_DMA_SUBROUTINE 0x0000324C +#define NV03_PFIFO_CACHE1_PULL0 0x00003240 +#define NV04_PFIFO_CACHE1_PULL0 0x00003250 +# define NV04_PFIFO_CACHE1_PULL0_HASH_FAILED 0x00000010 +# define NV04_PFIFO_CACHE1_PULL0_HASH_BUSY 0x00001000 +#define NV03_PFIFO_CACHE1_PULL1 0x00003250 +#define NV04_PFIFO_CACHE1_PULL1 0x00003254 +#define NV04_PFIFO_CACHE1_HASH 0x00003258 +#define NV10_PFIFO_CACHE1_ACQUIRE_TIMEOUT 0x00003260 +#define NV10_PFIFO_CACHE1_ACQUIRE_TIMESTAMP 0x00003264 +#define NV10_PFIFO_CACHE1_ACQUIRE_VALUE 0x00003268 +#define NV10_PFIFO_CACHE1_SEMAPHORE 0x0000326C +#define NV03_PFIFO_CACHE1_GET 0x00003270 +#define NV04_PFIFO_CACHE1_ENGINE 0x00003280 +#define NV04_PFIFO_CACHE1_DMA_DCOUNT 0x000032A0 +#define NV40_PFIFO_GRCTX_INSTANCE 0x000032E0 +#define NV40_PFIFO_UNK32E4 0x000032E4 +#define NV04_PFIFO_CACHE1_METHOD(i) (0x00003800+(i*8)) +#define NV04_PFIFO_CACHE1_DATA(i) (0x00003804+(i*8)) +#define NV40_PFIFO_CACHE1_METHOD(i) (0x00090000+(i*8)) +#define NV40_PFIFO_CACHE1_DATA(i) (0x00090004+(i*8)) + +#define NV_CRTC0_INTSTAT 0x00600100 +#define NV_CRTC0_INTEN 0x00600140 +#define NV_CRTC1_INTSTAT 0x00602100 +#define NV_CRTC1_INTEN 0x00602140 +# define NV_CRTC_INTR_VBLANK (1<<0) + +#define NV04_PRAMIN 0x00700000 + +/* Fifo commands. These are not regs, neither masks */ +#define NV03_FIFO_CMD_JUMP 0x20000000 +#define NV03_FIFO_CMD_JUMP_OFFSET_MASK 0x1ffffffc +#define NV03_FIFO_CMD_REWIND (NV03_FIFO_CMD_JUMP | (0 & NV03_FIFO_CMD_JUMP_OFFSET_MASK)) + +/* This is a partial import from rules-ng, a few things may be duplicated. + * Eventually we should completely import everything from rules-ng. + * For the moment check rules-ng for docs. + */ + +#define NV50_PMC 0x00000000 +#define NV50_PMC__LEN 0x1 +#define NV50_PMC__ESIZE 0x2000 +# define NV50_PMC_BOOT_0 0x00000000 +# define NV50_PMC_BOOT_0_REVISION 0x000000ff +# define NV50_PMC_BOOT_0_REVISION__SHIFT 0 +# define NV50_PMC_BOOT_0_ARCH 0x0ff00000 +# define NV50_PMC_BOOT_0_ARCH__SHIFT 20 +# define NV50_PMC_INTR_0 0x00000100 +# define NV50_PMC_INTR_0_PFIFO (1<<8) +# define NV50_PMC_INTR_0_PGRAPH (1<<12) +# define NV50_PMC_INTR_0_PTIMER (1<<20) +# define NV50_PMC_INTR_0_HOTPLUG (1<<21) +# define NV50_PMC_INTR_0_DISPLAY (1<<26) +# define NV50_PMC_INTR_EN_0 0x00000140 +# define NV50_PMC_INTR_EN_0_MASTER (1<<0) +# define NV50_PMC_INTR_EN_0_MASTER_DISABLED (0<<0) +# define NV50_PMC_INTR_EN_0_MASTER_ENABLED (1<<0) +# define NV50_PMC_ENABLE 0x00000200 +# define NV50_PMC_ENABLE_PFIFO (1<<8) +# define NV50_PMC_ENABLE_PGRAPH (1<<12) + +#define NV50_PCONNECTOR 0x0000e000 +#define NV50_PCONNECTOR__LEN 0x1 +#define NV50_PCONNECTOR__ESIZE 0x1000 +# define NV50_PCONNECTOR_HOTPLUG_INTR 0x0000e050 +# define NV50_PCONNECTOR_HOTPLUG_INTR_PLUG_I2C0 (1<<0) +# define NV50_PCONNECTOR_HOTPLUG_INTR_PLUG_I2C1 (1<<1) +# define NV50_PCONNECTOR_HOTPLUG_INTR_PLUG_I2C2 (1<<2) +# define NV50_PCONNECTOR_HOTPLUG_INTR_PLUG_I2C3 (1<<3) +# define NV50_PCONNECTOR_HOTPLUG_INTR_UNPLUG_I2C0 (1<<16) +# define NV50_PCONNECTOR_HOTPLUG_INTR_UNPLUG_I2C1 (1<<17) +# define NV50_PCONNECTOR_HOTPLUG_INTR_UNPLUG_I2C2 (1<<18) +# define NV50_PCONNECTOR_HOTPLUG_INTR_UNPLUG_I2C3 (1<<19) +# define NV50_PCONNECTOR_HOTPLUG_CTRL 0x0000e054 +# define NV50_PCONNECTOR_HOTPLUG_CTRL_PLUG_I2C0 (1<<0) +# define NV50_PCONNECTOR_HOTPLUG_CTRL_PLUG_I2C1 (1<<1) +# define NV50_PCONNECTOR_HOTPLUG_CTRL_PLUG_I2C2 (1<<2) +# define NV50_PCONNECTOR_HOTPLUG_CTRL_PLUG_I2C3 (1<<3) +# define NV50_PCONNECTOR_HOTPLUG_CTRL_UNPLUG_I2C0 (1<<16) +# define NV50_PCONNECTOR_HOTPLUG_CTRL_UNPLUG_I2C1 (1<<17) +# define NV50_PCONNECTOR_HOTPLUG_CTRL_UNPLUG_I2C2 (1<<18) +# define NV50_PCONNECTOR_HOTPLUG_CTRL_UNPLUG_I2C3 (1<<19) +# define NV50_PCONNECTOR_HOTPLUG_STATE 0x0000e104 +# define NV50_PCONNECTOR_HOTPLUG_STATE_PIN_CONNECTED_I2C0 (1<<2) +# define NV50_PCONNECTOR_HOTPLUG_STATE_PIN_CONNECTED_I2C1 (1<<6) +# define NV50_PCONNECTOR_HOTPLUG_STATE_PIN_CONNECTED_I2C2 (1<<10) +# define NV50_PCONNECTOR_HOTPLUG_STATE_PIN_CONNECTED_I2C3 (1<<14) +# define NV50_PCONNECTOR_I2C_PORT_0 0x0000e138 +# define NV50_PCONNECTOR_I2C_PORT_1 0x0000e150 +# define NV50_PCONNECTOR_I2C_PORT_2 0x0000e168 +# define NV50_PCONNECTOR_I2C_PORT_3 0x0000e180 +# define NV50_PCONNECTOR_I2C_PORT_4 0x0000e240 +# define NV50_PCONNECTOR_I2C_PORT_5 0x0000e258 + +#define NV50_AUXCH_DATA_OUT(i, n) ((n) * 4 + (i) * 0x50 + 0x0000e4c0) +#define NV50_AUXCH_DATA_OUT__SIZE 4 +#define NV50_AUXCH_DATA_IN(i, n) ((n) * 4 + (i) * 0x50 + 0x0000e4d0) +#define NV50_AUXCH_DATA_IN__SIZE 4 +#define NV50_AUXCH_ADDR(i) ((i) * 0x50 + 0x0000e4e0) +#define NV50_AUXCH_CTRL(i) ((i) * 0x50 + 0x0000e4e4) +#define NV50_AUXCH_CTRL_LINKSTAT 0x01000000 +#define NV50_AUXCH_CTRL_LINKSTAT_NOT_READY 0x00000000 +#define NV50_AUXCH_CTRL_LINKSTAT_READY 0x01000000 +#define NV50_AUXCH_CTRL_LINKEN 0x00100000 +#define NV50_AUXCH_CTRL_LINKEN_DISABLED 0x00000000 +#define NV50_AUXCH_CTRL_LINKEN_ENABLED 0x00100000 +#define NV50_AUXCH_CTRL_EXEC 0x00010000 +#define NV50_AUXCH_CTRL_EXEC_COMPLETE 0x00000000 +#define NV50_AUXCH_CTRL_EXEC_IN_PROCESS 0x00010000 +#define NV50_AUXCH_CTRL_CMD 0x0000f000 +#define NV50_AUXCH_CTRL_CMD_SHIFT 12 +#define NV50_AUXCH_CTRL_LEN 0x0000000f +#define NV50_AUXCH_CTRL_LEN_SHIFT 0 +#define NV50_AUXCH_STAT(i) ((i) * 0x50 + 0x0000e4e8) +#define NV50_AUXCH_STAT_STATE 0x10000000 +#define NV50_AUXCH_STAT_STATE_NOT_READY 0x00000000 +#define NV50_AUXCH_STAT_STATE_READY 0x10000000 +#define NV50_AUXCH_STAT_REPLY 0x000f0000 +#define NV50_AUXCH_STAT_REPLY_AUX 0x00030000 +#define NV50_AUXCH_STAT_REPLY_AUX_ACK 0x00000000 +#define NV50_AUXCH_STAT_REPLY_AUX_NACK 0x00010000 +#define NV50_AUXCH_STAT_REPLY_AUX_DEFER 0x00020000 +#define NV50_AUXCH_STAT_REPLY_I2C 0x000c0000 +#define NV50_AUXCH_STAT_REPLY_I2C_ACK 0x00000000 +#define NV50_AUXCH_STAT_REPLY_I2C_NACK 0x00040000 +#define NV50_AUXCH_STAT_REPLY_I2C_DEFER 0x00080000 +#define NV50_AUXCH_STAT_COUNT 0x0000001f + +#define NV50_PBUS 0x00088000 +#define NV50_PBUS__LEN 0x1 +#define NV50_PBUS__ESIZE 0x1000 +# define NV50_PBUS_PCI_ID 0x00088000 +# define NV50_PBUS_PCI_ID_VENDOR_ID 0x0000ffff +# define NV50_PBUS_PCI_ID_VENDOR_ID__SHIFT 0 +# define NV50_PBUS_PCI_ID_DEVICE_ID 0xffff0000 +# define NV50_PBUS_PCI_ID_DEVICE_ID__SHIFT 16 + +#define NV50_PFB 0x00100000 +#define NV50_PFB__LEN 0x1 +#define NV50_PFB__ESIZE 0x1000 + +#define NV50_PEXTDEV 0x00101000 +#define NV50_PEXTDEV__LEN 0x1 +#define NV50_PEXTDEV__ESIZE 0x1000 + +#define NV50_PROM 0x00300000 +#define NV50_PROM__LEN 0x1 +#define NV50_PROM__ESIZE 0x10000 + +#define NV50_PGRAPH 0x00400000 +#define NV50_PGRAPH__LEN 0x1 +#define NV50_PGRAPH__ESIZE 0x10000 + +#define NV50_PDISPLAY 0x00610000 +#define NV50_PDISPLAY_OBJECTS 0x00610010 +#define NV50_PDISPLAY_INTR_0 0x00610020 +#define NV50_PDISPLAY_INTR_1 0x00610024 +#define NV50_PDISPLAY_INTR_1_VBLANK_CRTC 0x0000000c +#define NV50_PDISPLAY_INTR_1_VBLANK_CRTC_SHIFT 2 +#define NV50_PDISPLAY_INTR_1_VBLANK_CRTC_(n) (1 << ((n) + 2)) +#define NV50_PDISPLAY_INTR_1_VBLANK_CRTC_0 0x00000004 +#define NV50_PDISPLAY_INTR_1_VBLANK_CRTC_1 0x00000008 +#define NV50_PDISPLAY_INTR_1_CLK_UNK10 0x00000010 +#define NV50_PDISPLAY_INTR_1_CLK_UNK20 0x00000020 +#define NV50_PDISPLAY_INTR_1_CLK_UNK40 0x00000040 +#define NV50_PDISPLAY_INTR_EN_0 0x00610028 +#define NV50_PDISPLAY_INTR_EN_1 0x0061002c +#define NV50_PDISPLAY_INTR_EN_1_VBLANK_CRTC 0x0000000c +#define NV50_PDISPLAY_INTR_EN_1_VBLANK_CRTC_(n) (1 << ((n) + 2)) +#define NV50_PDISPLAY_INTR_EN_1_VBLANK_CRTC_0 0x00000004 +#define NV50_PDISPLAY_INTR_EN_1_VBLANK_CRTC_1 0x00000008 +#define NV50_PDISPLAY_INTR_EN_1_CLK_UNK10 0x00000010 +#define NV50_PDISPLAY_INTR_EN_1_CLK_UNK20 0x00000020 +#define NV50_PDISPLAY_INTR_EN_1_CLK_UNK40 0x00000040 +#define NV50_PDISPLAY_UNK30_CTRL 0x00610030 +#define NV50_PDISPLAY_UNK30_CTRL_UPDATE_VCLK0 0x00000200 +#define NV50_PDISPLAY_UNK30_CTRL_UPDATE_VCLK1 0x00000400 +#define NV50_PDISPLAY_UNK30_CTRL_PENDING 0x80000000 +#define NV50_PDISPLAY_TRAPPED_ADDR(i) ((i) * 0x08 + 0x00610080) +#define NV50_PDISPLAY_TRAPPED_DATA(i) ((i) * 0x08 + 0x00610084) +#define NV50_PDISPLAY_EVO_CTRL(i) ((i) * 0x10 + 0x00610200) +#define NV50_PDISPLAY_EVO_CTRL_DMA 0x00000010 +#define NV50_PDISPLAY_EVO_CTRL_DMA_DISABLED 0x00000000 +#define NV50_PDISPLAY_EVO_CTRL_DMA_ENABLED 0x00000010 +#define NV50_PDISPLAY_EVO_DMA_CB(i) ((i) * 0x10 + 0x00610204) +#define NV50_PDISPLAY_EVO_DMA_CB_LOCATION 0x00000002 +#define NV50_PDISPLAY_EVO_DMA_CB_LOCATION_VRAM 0x00000000 +#define NV50_PDISPLAY_EVO_DMA_CB_LOCATION_SYSTEM 0x00000002 +#define NV50_PDISPLAY_EVO_DMA_CB_VALID 0x00000001 +#define NV50_PDISPLAY_EVO_UNK2(i) ((i) * 0x10 + 0x00610208) +#define NV50_PDISPLAY_EVO_HASH_TAG(i) ((i) * 0x10 + 0x0061020c) + +#define NV50_PDISPLAY_CURSOR 0x00610270 +#define NV50_PDISPLAY_CURSOR_CURSOR_CTRL2(i) ((i) * 0x10 + 0x00610270) +#define NV50_PDISPLAY_CURSOR_CURSOR_CTRL2_ON 0x00000001 +#define NV50_PDISPLAY_CURSOR_CURSOR_CTRL2_STATUS 0x00030000 +#define NV50_PDISPLAY_CURSOR_CURSOR_CTRL2_STATUS_ACTIVE 0x00010000 + +#define NV50_PDISPLAY_PIO_CTRL 0x00610300 +#define NV50_PDISPLAY_PIO_CTRL_PENDING 0x80000000 +#define NV50_PDISPLAY_PIO_CTRL_MTHD 0x00001ffc +#define NV50_PDISPLAY_PIO_CTRL_ENABLED 0x00000001 +#define NV50_PDISPLAY_PIO_DATA 0x00610304 + +#define NV50_PDISPLAY_CRTC_P(i, r) ((i) * 0x540 + NV50_PDISPLAY_CRTC_##r) +#define NV50_PDISPLAY_CRTC_C(i, r) (4 + (i) * 0x540 + NV50_PDISPLAY_CRTC_##r) +#define NV50_PDISPLAY_CRTC_UNK_0A18 /* mthd 0x0900 */ 0x00610a18 +#define NV50_PDISPLAY_CRTC_CLUT_MODE 0x00610a24 +#define NV50_PDISPLAY_CRTC_INTERLACE 0x00610a48 +#define NV50_PDISPLAY_CRTC_SCALE_CTRL 0x00610a50 +#define NV50_PDISPLAY_CRTC_CURSOR_CTRL 0x00610a58 +#define NV50_PDISPLAY_CRTC_UNK0A78 /* mthd 0x0904 */ 0x00610a78 +#define NV50_PDISPLAY_CRTC_UNK0AB8 0x00610ab8 +#define NV50_PDISPLAY_CRTC_DEPTH 0x00610ac8 +#define NV50_PDISPLAY_CRTC_CLOCK 0x00610ad0 +#define NV50_PDISPLAY_CRTC_COLOR_CTRL 0x00610ae0 +#define NV50_PDISPLAY_CRTC_SYNC_START_TO_BLANK_END 0x00610ae8 +#define NV50_PDISPLAY_CRTC_MODE_UNK1 0x00610af0 +#define NV50_PDISPLAY_CRTC_DISPLAY_TOTAL 0x00610af8 +#define NV50_PDISPLAY_CRTC_SYNC_DURATION 0x00610b00 +#define NV50_PDISPLAY_CRTC_MODE_UNK2 0x00610b08 +#define NV50_PDISPLAY_CRTC_UNK_0B10 /* mthd 0x0828 */ 0x00610b10 +#define NV50_PDISPLAY_CRTC_FB_SIZE 0x00610b18 +#define NV50_PDISPLAY_CRTC_FB_PITCH 0x00610b20 +#define NV50_PDISPLAY_CRTC_FB_PITCH_LINEAR 0x00100000 +#define NV50_PDISPLAY_CRTC_FB_POS 0x00610b28 +#define NV50_PDISPLAY_CRTC_SCALE_CENTER_OFFSET 0x00610b38 +#define NV50_PDISPLAY_CRTC_REAL_RES 0x00610b40 +#define NV50_PDISPLAY_CRTC_SCALE_RES1 0x00610b48 +#define NV50_PDISPLAY_CRTC_SCALE_RES2 0x00610b50 + +#define NV50_PDISPLAY_DAC_MODE_CTRL_P(i) (0x00610b58 + (i) * 0x8) +#define NV50_PDISPLAY_DAC_MODE_CTRL_C(i) (0x00610b5c + (i) * 0x8) +#define NV50_PDISPLAY_SOR_MODE_CTRL_P(i) (0x00610b70 + (i) * 0x8) +#define NV50_PDISPLAY_SOR_MODE_CTRL_C(i) (0x00610b74 + (i) * 0x8) +#define NV50_PDISPLAY_EXT_MODE_CTRL_P(i) (0x00610b80 + (i) * 0x8) +#define NV50_PDISPLAY_EXT_MODE_CTRL_C(i) (0x00610b84 + (i) * 0x8) +#define NV50_PDISPLAY_DAC_MODE_CTRL2_P(i) (0x00610bdc + (i) * 0x8) +#define NV50_PDISPLAY_DAC_MODE_CTRL2_C(i) (0x00610be0 + (i) * 0x8) +#define NV90_PDISPLAY_SOR_MODE_CTRL_P(i) (0x00610794 + (i) * 0x8) +#define NV90_PDISPLAY_SOR_MODE_CTRL_C(i) (0x00610798 + (i) * 0x8) + +#define NV50_PDISPLAY_CRTC_CLK 0x00614000 +#define NV50_PDISPLAY_CRTC_CLK_CTRL1(i) ((i) * 0x800 + 0x614100) +#define NV50_PDISPLAY_CRTC_CLK_CTRL1_CONNECTED 0x00000600 +#define NV50_PDISPLAY_CRTC_CLK_VPLL_A(i) ((i) * 0x800 + 0x614104) +#define NV50_PDISPLAY_CRTC_CLK_VPLL_B(i) ((i) * 0x800 + 0x614108) +#define NV50_PDISPLAY_CRTC_CLK_CTRL2(i) ((i) * 0x800 + 0x614200) + +#define NV50_PDISPLAY_DAC_CLK 0x00614000 +#define NV50_PDISPLAY_DAC_CLK_CTRL2(i) ((i) * 0x800 + 0x614280) + +#define NV50_PDISPLAY_SOR_CLK 0x00614000 +#define NV50_PDISPLAY_SOR_CLK_CTRL2(i) ((i) * 0x800 + 0x614300) + +#define NV50_PDISPLAY_VGACRTC(r) ((r) + 0x619400) + +#define NV50_PDISPLAY_DAC 0x0061a000 +#define NV50_PDISPLAY_DAC_DPMS_CTRL(i) (0x0061a004 + (i) * 0x800) +#define NV50_PDISPLAY_DAC_DPMS_CTRL_HSYNC_OFF 0x00000001 +#define NV50_PDISPLAY_DAC_DPMS_CTRL_VSYNC_OFF 0x00000004 +#define NV50_PDISPLAY_DAC_DPMS_CTRL_BLANKED 0x00000010 +#define NV50_PDISPLAY_DAC_DPMS_CTRL_OFF 0x00000040 +#define NV50_PDISPLAY_DAC_DPMS_CTRL_PENDING 0x80000000 +#define NV50_PDISPLAY_DAC_LOAD_CTRL(i) (0x0061a00c + (i) * 0x800) +#define NV50_PDISPLAY_DAC_LOAD_CTRL_ACTIVE 0x00100000 +#define NV50_PDISPLAY_DAC_LOAD_CTRL_PRESENT 0x38000000 +#define NV50_PDISPLAY_DAC_LOAD_CTRL_DONE 0x80000000 +#define NV50_PDISPLAY_DAC_CLK_CTRL1(i) (0x0061a010 + (i) * 0x800) +#define NV50_PDISPLAY_DAC_CLK_CTRL1_CONNECTED 0x00000600 + +#define NV50_PDISPLAY_SOR 0x0061c000 +#define NV50_PDISPLAY_SOR_DPMS_CTRL(i) (0x0061c004 + (i) * 0x800) +#define NV50_PDISPLAY_SOR_DPMS_CTRL_PENDING 0x80000000 +#define NV50_PDISPLAY_SOR_DPMS_CTRL_ON 0x00000001 +#define NV50_PDISPLAY_SOR_CLK_CTRL1(i) (0x0061c008 + (i) * 0x800) +#define NV50_PDISPLAY_SOR_CLK_CTRL1_CONNECTED 0x00000600 +#define NV50_PDISPLAY_SOR_DPMS_STATE(i) (0x0061c030 + (i) * 0x800) +#define NV50_PDISPLAY_SOR_DPMS_STATE_ACTIVE 0x00030000 +#define NV50_PDISPLAY_SOR_DPMS_STATE_BLANKED 0x00080000 +#define NV50_PDISPLAY_SOR_DPMS_STATE_WAIT 0x10000000 +#define NV50_PDISPLAY_SOR_BACKLIGHT 0x0061c084 +#define NV50_PDISPLAY_SOR_BACKLIGHT_ENABLE 0x80000000 +#define NV50_PDISPLAY_SOR_BACKLIGHT_LEVEL 0x00000fff +#define NV50_SOR_DP_CTRL(i, l) (0x0061c10c + (i) * 0x800 + (l) * 0x80) +#define NV50_SOR_DP_CTRL_ENABLED 0x00000001 +#define NV50_SOR_DP_CTRL_ENHANCED_FRAME_ENABLED 0x00004000 +#define NV50_SOR_DP_CTRL_LANE_MASK 0x001f0000 +#define NV50_SOR_DP_CTRL_LANE_0_ENABLED 0x00010000 +#define NV50_SOR_DP_CTRL_LANE_1_ENABLED 0x00020000 +#define NV50_SOR_DP_CTRL_LANE_2_ENABLED 0x00040000 +#define NV50_SOR_DP_CTRL_LANE_3_ENABLED 0x00080000 +#define NV50_SOR_DP_CTRL_TRAINING_PATTERN 0x0f000000 +#define NV50_SOR_DP_CTRL_TRAINING_PATTERN_DISABLED 0x00000000 +#define NV50_SOR_DP_CTRL_TRAINING_PATTERN_1 0x01000000 +#define NV50_SOR_DP_CTRL_TRAINING_PATTERN_2 0x02000000 +#define NV50_SOR_DP_UNK118(i, l) (0x0061c118 + (i) * 0x800 + (l) * 0x80) +#define NV50_SOR_DP_UNK120(i, l) (0x0061c120 + (i) * 0x800 + (l) * 0x80) +#define NV50_SOR_DP_UNK128(i, l) (0x0061c128 + (i) * 0x800 + (l) * 0x80) +#define NV50_SOR_DP_UNK130(i, l) (0x0061c130 + (i) * 0x800 + (l) * 0x80) + +#define NV50_PDISPLAY_USER(i) ((i) * 0x1000 + 0x00640000) +#define NV50_PDISPLAY_USER_PUT(i) ((i) * 0x1000 + 0x00640000) +#define NV50_PDISPLAY_USER_GET(i) ((i) * 0x1000 + 0x00640004) + +#define NV50_PDISPLAY_CURSOR_USER 0x00647000 +#define NV50_PDISPLAY_CURSOR_USER_POS_CTRL(i) ((i) * 0x1000 + 0x00647080) +#define NV50_PDISPLAY_CURSOR_USER_POS(i) ((i) * 0x1000 + 0x00647084) diff --git a/drivers/gpu/drm/nouveau/nouveau_sgdma.c b/drivers/gpu/drm/nouveau/nouveau_sgdma.c new file mode 100644 index 00000000..ca6028f2 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_sgdma.c @@ -0,0 +1,538 @@ +#include "drmP.h" +#include "nouveau_drv.h" +#include +#include + +#define NV_CTXDMA_PAGE_SHIFT 12 +#define NV_CTXDMA_PAGE_SIZE (1 << NV_CTXDMA_PAGE_SHIFT) +#define NV_CTXDMA_PAGE_MASK (NV_CTXDMA_PAGE_SIZE - 1) + +struct nouveau_sgdma_be { + struct ttm_backend backend; + struct drm_device *dev; + + dma_addr_t *pages; + bool *ttm_alloced; + unsigned nr_pages; + + u64 offset; + bool bound; +}; + +static int +nouveau_sgdma_populate(struct ttm_backend *be, unsigned long num_pages, + struct page **pages, struct page *dummy_read_page, + dma_addr_t *dma_addrs) +{ + struct nouveau_sgdma_be *nvbe = (struct nouveau_sgdma_be *)be; + struct drm_device *dev = nvbe->dev; + + NV_DEBUG(nvbe->dev, "num_pages = %ld\n", num_pages); + + if (nvbe->pages) + return -EINVAL; + + nvbe->pages = kmalloc(sizeof(dma_addr_t) * num_pages, GFP_KERNEL); + if (!nvbe->pages) + return -ENOMEM; + + nvbe->ttm_alloced = kmalloc(sizeof(bool) * num_pages, GFP_KERNEL); + if (!nvbe->ttm_alloced) { + kfree(nvbe->pages); + nvbe->pages = NULL; + return -ENOMEM; + } + + nvbe->nr_pages = 0; + while (num_pages--) { + /* this code path isn't called and is incorrect anyways */ + if (0) { /*dma_addrs[nvbe->nr_pages] != DMA_ERROR_CODE)*/ + nvbe->pages[nvbe->nr_pages] = + dma_addrs[nvbe->nr_pages]; + nvbe->ttm_alloced[nvbe->nr_pages] = true; + } else { + nvbe->pages[nvbe->nr_pages] = + pci_map_page(dev->pdev, pages[nvbe->nr_pages], 0, + PAGE_SIZE, PCI_DMA_BIDIRECTIONAL); + if (pci_dma_mapping_error(dev->pdev, + nvbe->pages[nvbe->nr_pages])) { + be->func->clear(be); + return -EFAULT; + } + nvbe->ttm_alloced[nvbe->nr_pages] = false; + } + + nvbe->nr_pages++; + } + + return 0; +} + +static void +nouveau_sgdma_clear(struct ttm_backend *be) +{ + struct nouveau_sgdma_be *nvbe = (struct nouveau_sgdma_be *)be; + struct drm_device *dev; + + if (nvbe && nvbe->pages) { + dev = nvbe->dev; + NV_DEBUG(dev, "\n"); + + if (nvbe->bound) + be->func->unbind(be); + + while (nvbe->nr_pages--) { + if (!nvbe->ttm_alloced[nvbe->nr_pages]) + pci_unmap_page(dev->pdev, nvbe->pages[nvbe->nr_pages], + PAGE_SIZE, PCI_DMA_BIDIRECTIONAL); + } + kfree(nvbe->pages); + kfree(nvbe->ttm_alloced); + nvbe->pages = NULL; + nvbe->ttm_alloced = NULL; + nvbe->nr_pages = 0; + } +} + +static void +nouveau_sgdma_destroy(struct ttm_backend *be) +{ + struct nouveau_sgdma_be *nvbe = (struct nouveau_sgdma_be *)be; + + if (be) { + NV_DEBUG(nvbe->dev, "\n"); + + if (nvbe) { + if (nvbe->pages) + be->func->clear(be); + kfree(nvbe); + } + } +} + +static int +nv04_sgdma_bind(struct ttm_backend *be, struct ttm_mem_reg *mem) +{ + struct nouveau_sgdma_be *nvbe = (struct nouveau_sgdma_be *)be; + struct drm_device *dev = nvbe->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *gpuobj = dev_priv->gart_info.sg_ctxdma; + unsigned i, j, pte; + + NV_DEBUG(dev, "pg=0x%lx\n", mem->start); + + nvbe->offset = mem->start << PAGE_SHIFT; + pte = (nvbe->offset >> NV_CTXDMA_PAGE_SHIFT) + 2; + for (i = 0; i < nvbe->nr_pages; i++) { + dma_addr_t dma_offset = nvbe->pages[i]; + uint32_t offset_l = lower_32_bits(dma_offset); + + for (j = 0; j < PAGE_SIZE / NV_CTXDMA_PAGE_SIZE; j++, pte++) { + nv_wo32(gpuobj, (pte * 4) + 0, offset_l | 3); + dma_offset += NV_CTXDMA_PAGE_SIZE; + } + } + + nvbe->bound = true; + return 0; +} + +static int +nv04_sgdma_unbind(struct ttm_backend *be) +{ + struct nouveau_sgdma_be *nvbe = (struct nouveau_sgdma_be *)be; + struct drm_device *dev = nvbe->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *gpuobj = dev_priv->gart_info.sg_ctxdma; + unsigned i, j, pte; + + NV_DEBUG(dev, "\n"); + + if (!nvbe->bound) + return 0; + + pte = (nvbe->offset >> NV_CTXDMA_PAGE_SHIFT) + 2; + for (i = 0; i < nvbe->nr_pages; i++) { + for (j = 0; j < PAGE_SIZE / NV_CTXDMA_PAGE_SIZE; j++, pte++) + nv_wo32(gpuobj, (pte * 4) + 0, 0x00000000); + } + + nvbe->bound = false; + return 0; +} + +static struct ttm_backend_func nv04_sgdma_backend = { + .populate = nouveau_sgdma_populate, + .clear = nouveau_sgdma_clear, + .bind = nv04_sgdma_bind, + .unbind = nv04_sgdma_unbind, + .destroy = nouveau_sgdma_destroy +}; + +static void +nv41_sgdma_flush(struct nouveau_sgdma_be *nvbe) +{ + struct drm_device *dev = nvbe->dev; + + nv_wr32(dev, 0x100810, 0x00000022); + if (!nv_wait(dev, 0x100810, 0x00000100, 0x00000100)) + NV_ERROR(dev, "vm flush timeout: 0x%08x\n", + nv_rd32(dev, 0x100810)); + nv_wr32(dev, 0x100810, 0x00000000); +} + +static int +nv41_sgdma_bind(struct ttm_backend *be, struct ttm_mem_reg *mem) +{ + struct nouveau_sgdma_be *nvbe = (struct nouveau_sgdma_be *)be; + struct drm_nouveau_private *dev_priv = nvbe->dev->dev_private; + struct nouveau_gpuobj *pgt = dev_priv->gart_info.sg_ctxdma; + dma_addr_t *list = nvbe->pages; + u32 pte = mem->start << 2; + u32 cnt = nvbe->nr_pages; + + nvbe->offset = mem->start << PAGE_SHIFT; + + while (cnt--) { + nv_wo32(pgt, pte, (*list++ >> 7) | 1); + pte += 4; + } + + nv41_sgdma_flush(nvbe); + nvbe->bound = true; + return 0; +} + +static int +nv41_sgdma_unbind(struct ttm_backend *be) +{ + struct nouveau_sgdma_be *nvbe = (struct nouveau_sgdma_be *)be; + struct drm_nouveau_private *dev_priv = nvbe->dev->dev_private; + struct nouveau_gpuobj *pgt = dev_priv->gart_info.sg_ctxdma; + u32 pte = (nvbe->offset >> 12) << 2; + u32 cnt = nvbe->nr_pages; + + while (cnt--) { + nv_wo32(pgt, pte, 0x00000000); + pte += 4; + } + + nv41_sgdma_flush(nvbe); + nvbe->bound = false; + return 0; +} + +static struct ttm_backend_func nv41_sgdma_backend = { + .populate = nouveau_sgdma_populate, + .clear = nouveau_sgdma_clear, + .bind = nv41_sgdma_bind, + .unbind = nv41_sgdma_unbind, + .destroy = nouveau_sgdma_destroy +}; + +static void +nv44_sgdma_flush(struct nouveau_sgdma_be *nvbe) +{ + struct drm_device *dev = nvbe->dev; + + nv_wr32(dev, 0x100814, (nvbe->nr_pages - 1) << 12); + nv_wr32(dev, 0x100808, nvbe->offset | 0x20); + if (!nv_wait(dev, 0x100808, 0x00000001, 0x00000001)) + NV_ERROR(dev, "gart flush timeout: 0x%08x\n", + nv_rd32(dev, 0x100808)); + nv_wr32(dev, 0x100808, 0x00000000); +} + +static void +nv44_sgdma_fill(struct nouveau_gpuobj *pgt, dma_addr_t *list, u32 base, u32 cnt) +{ + struct drm_nouveau_private *dev_priv = pgt->dev->dev_private; + dma_addr_t dummy = dev_priv->gart_info.dummy.addr; + u32 pte, tmp[4]; + + pte = base >> 2; + base &= ~0x0000000f; + + tmp[0] = nv_ro32(pgt, base + 0x0); + tmp[1] = nv_ro32(pgt, base + 0x4); + tmp[2] = nv_ro32(pgt, base + 0x8); + tmp[3] = nv_ro32(pgt, base + 0xc); + while (cnt--) { + u32 addr = list ? (*list++ >> 12) : (dummy >> 12); + switch (pte++ & 0x3) { + case 0: + tmp[0] &= ~0x07ffffff; + tmp[0] |= addr; + break; + case 1: + tmp[0] &= ~0xf8000000; + tmp[0] |= addr << 27; + tmp[1] &= ~0x003fffff; + tmp[1] |= addr >> 5; + break; + case 2: + tmp[1] &= ~0xffc00000; + tmp[1] |= addr << 22; + tmp[2] &= ~0x0001ffff; + tmp[2] |= addr >> 10; + break; + case 3: + tmp[2] &= ~0xfffe0000; + tmp[2] |= addr << 17; + tmp[3] &= ~0x00000fff; + tmp[3] |= addr >> 15; + break; + } + } + + tmp[3] |= 0x40000000; + + nv_wo32(pgt, base + 0x0, tmp[0]); + nv_wo32(pgt, base + 0x4, tmp[1]); + nv_wo32(pgt, base + 0x8, tmp[2]); + nv_wo32(pgt, base + 0xc, tmp[3]); +} + +static int +nv44_sgdma_bind(struct ttm_backend *be, struct ttm_mem_reg *mem) +{ + struct nouveau_sgdma_be *nvbe = (struct nouveau_sgdma_be *)be; + struct drm_nouveau_private *dev_priv = nvbe->dev->dev_private; + struct nouveau_gpuobj *pgt = dev_priv->gart_info.sg_ctxdma; + dma_addr_t *list = nvbe->pages; + u32 pte = mem->start << 2, tmp[4]; + u32 cnt = nvbe->nr_pages; + int i; + + nvbe->offset = mem->start << PAGE_SHIFT; + + if (pte & 0x0000000c) { + u32 max = 4 - ((pte >> 2) & 0x3); + u32 part = (cnt > max) ? max : cnt; + nv44_sgdma_fill(pgt, list, pte, part); + pte += (part << 2); + list += part; + cnt -= part; + } + + while (cnt >= 4) { + for (i = 0; i < 4; i++) + tmp[i] = *list++ >> 12; + nv_wo32(pgt, pte + 0x0, tmp[0] >> 0 | tmp[1] << 27); + nv_wo32(pgt, pte + 0x4, tmp[1] >> 5 | tmp[2] << 22); + nv_wo32(pgt, pte + 0x8, tmp[2] >> 10 | tmp[3] << 17); + nv_wo32(pgt, pte + 0xc, tmp[3] >> 15 | 0x40000000); + pte += 0x10; + cnt -= 4; + } + + if (cnt) + nv44_sgdma_fill(pgt, list, pte, cnt); + + nv44_sgdma_flush(nvbe); + nvbe->bound = true; + return 0; +} + +static int +nv44_sgdma_unbind(struct ttm_backend *be) +{ + struct nouveau_sgdma_be *nvbe = (struct nouveau_sgdma_be *)be; + struct drm_nouveau_private *dev_priv = nvbe->dev->dev_private; + struct nouveau_gpuobj *pgt = dev_priv->gart_info.sg_ctxdma; + u32 pte = (nvbe->offset >> 12) << 2; + u32 cnt = nvbe->nr_pages; + + if (pte & 0x0000000c) { + u32 max = 4 - ((pte >> 2) & 0x3); + u32 part = (cnt > max) ? max : cnt; + nv44_sgdma_fill(pgt, NULL, pte, part); + pte += (part << 2); + cnt -= part; + } + + while (cnt >= 4) { + nv_wo32(pgt, pte + 0x0, 0x00000000); + nv_wo32(pgt, pte + 0x4, 0x00000000); + nv_wo32(pgt, pte + 0x8, 0x00000000); + nv_wo32(pgt, pte + 0xc, 0x00000000); + pte += 0x10; + cnt -= 4; + } + + if (cnt) + nv44_sgdma_fill(pgt, NULL, pte, cnt); + + nv44_sgdma_flush(nvbe); + nvbe->bound = false; + return 0; +} + +static struct ttm_backend_func nv44_sgdma_backend = { + .populate = nouveau_sgdma_populate, + .clear = nouveau_sgdma_clear, + .bind = nv44_sgdma_bind, + .unbind = nv44_sgdma_unbind, + .destroy = nouveau_sgdma_destroy +}; + +static int +nv50_sgdma_bind(struct ttm_backend *be, struct ttm_mem_reg *mem) +{ + struct nouveau_sgdma_be *nvbe = (struct nouveau_sgdma_be *)be; + struct nouveau_mem *node = mem->mm_node; + /* noop: bound in move_notify() */ + node->pages = nvbe->pages; + nvbe->pages = (dma_addr_t *)node; + nvbe->bound = true; + return 0; +} + +static int +nv50_sgdma_unbind(struct ttm_backend *be) +{ + struct nouveau_sgdma_be *nvbe = (struct nouveau_sgdma_be *)be; + struct nouveau_mem *node = (struct nouveau_mem *)nvbe->pages; + /* noop: unbound in move_notify() */ + nvbe->pages = node->pages; + node->pages = NULL; + nvbe->bound = false; + return 0; +} + +static struct ttm_backend_func nv50_sgdma_backend = { + .populate = nouveau_sgdma_populate, + .clear = nouveau_sgdma_clear, + .bind = nv50_sgdma_bind, + .unbind = nv50_sgdma_unbind, + .destroy = nouveau_sgdma_destroy +}; + +struct ttm_backend * +nouveau_sgdma_init_ttm(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_sgdma_be *nvbe; + + nvbe = kzalloc(sizeof(*nvbe), GFP_KERNEL); + if (!nvbe) + return NULL; + + nvbe->dev = dev; + + nvbe->backend.func = dev_priv->gart_info.func; + return &nvbe->backend; +} + +int +nouveau_sgdma_init(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *gpuobj = NULL; + u32 aper_size, align; + int ret; + + if (dev_priv->card_type >= NV_40 && drm_pci_device_is_pcie(dev)) + aper_size = 512 * 1024 * 1024; + else + aper_size = 64 * 1024 * 1024; + + /* Dear NVIDIA, NV44+ would like proper present bits in PTEs for + * christmas. The cards before it have them, the cards after + * it have them, why is NV44 so unloved? + */ + dev_priv->gart_info.dummy.page = alloc_page(GFP_DMA32 | GFP_KERNEL); + if (!dev_priv->gart_info.dummy.page) + return -ENOMEM; + + dev_priv->gart_info.dummy.addr = + pci_map_page(dev->pdev, dev_priv->gart_info.dummy.page, + 0, PAGE_SIZE, PCI_DMA_BIDIRECTIONAL); + if (pci_dma_mapping_error(dev->pdev, dev_priv->gart_info.dummy.addr)) { + NV_ERROR(dev, "error mapping dummy page\n"); + __free_page(dev_priv->gart_info.dummy.page); + dev_priv->gart_info.dummy.page = NULL; + return -ENOMEM; + } + + if (dev_priv->card_type >= NV_50) { + dev_priv->gart_info.aper_base = 0; + dev_priv->gart_info.aper_size = aper_size; + dev_priv->gart_info.type = NOUVEAU_GART_HW; + dev_priv->gart_info.func = &nv50_sgdma_backend; + } else + if (0 && drm_pci_device_is_pcie(dev) && + dev_priv->chipset > 0x40 && dev_priv->chipset != 0x45) { + if (nv44_graph_class(dev)) { + dev_priv->gart_info.func = &nv44_sgdma_backend; + align = 512 * 1024; + } else { + dev_priv->gart_info.func = &nv41_sgdma_backend; + align = 16; + } + + ret = nouveau_gpuobj_new(dev, NULL, aper_size / 1024, align, + NVOBJ_FLAG_ZERO_ALLOC | + NVOBJ_FLAG_ZERO_FREE, &gpuobj); + if (ret) { + NV_ERROR(dev, "Error creating sgdma object: %d\n", ret); + return ret; + } + + dev_priv->gart_info.sg_ctxdma = gpuobj; + dev_priv->gart_info.aper_base = 0; + dev_priv->gart_info.aper_size = aper_size; + dev_priv->gart_info.type = NOUVEAU_GART_HW; + } else { + ret = nouveau_gpuobj_new(dev, NULL, (aper_size / 1024) + 8, 16, + NVOBJ_FLAG_ZERO_ALLOC | + NVOBJ_FLAG_ZERO_FREE, &gpuobj); + if (ret) { + NV_ERROR(dev, "Error creating sgdma object: %d\n", ret); + return ret; + } + + nv_wo32(gpuobj, 0, NV_CLASS_DMA_IN_MEMORY | + (1 << 12) /* PT present */ | + (0 << 13) /* PT *not* linear */ | + (0 << 14) /* RW */ | + (2 << 16) /* PCI */); + nv_wo32(gpuobj, 4, aper_size - 1); + + dev_priv->gart_info.sg_ctxdma = gpuobj; + dev_priv->gart_info.aper_base = 0; + dev_priv->gart_info.aper_size = aper_size; + dev_priv->gart_info.type = NOUVEAU_GART_PDMA; + dev_priv->gart_info.func = &nv04_sgdma_backend; + } + + return 0; +} + +void +nouveau_sgdma_takedown(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + nouveau_gpuobj_ref(NULL, &dev_priv->gart_info.sg_ctxdma); + + if (dev_priv->gart_info.dummy.page) { + pci_unmap_page(dev->pdev, dev_priv->gart_info.dummy.addr, + PAGE_SIZE, PCI_DMA_BIDIRECTIONAL); + __free_page(dev_priv->gart_info.dummy.page); + dev_priv->gart_info.dummy.page = NULL; + } +} + +uint32_t +nouveau_sgdma_get_physical(struct drm_device *dev, uint32_t offset) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *gpuobj = dev_priv->gart_info.sg_ctxdma; + int pte = (offset >> NV_CTXDMA_PAGE_SHIFT) + 2; + + BUG_ON(dev_priv->card_type >= NV_50); + + return (nv_ro32(gpuobj, 4 * pte) & ~NV_CTXDMA_PAGE_MASK) | + (offset & NV_CTXDMA_PAGE_MASK); +} diff --git a/drivers/gpu/drm/nouveau/nouveau_state.c b/drivers/gpu/drm/nouveau/nouveau_state.c new file mode 100644 index 00000000..731acea8 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_state.c @@ -0,0 +1,1140 @@ +/* + * Copyright 2005 Stephane Marchesin + * Copyright 2008 Stuart Bennett + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + */ + +#include +#include +#include "drmP.h" +#include "drm.h" +#include "drm_sarea.h" +#include "drm_crtc_helper.h" +#include +#include + +#include "nouveau_drv.h" +#include "nouveau_drm.h" +#include "nouveau_fbcon.h" +#include "nouveau_ramht.h" +#include "nouveau_pm.h" +#include "nv50_display.h" + +static void nouveau_stub_takedown(struct drm_device *dev) {} +static int nouveau_stub_init(struct drm_device *dev) { return 0; } + +static int nouveau_init_engine_ptrs(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_engine *engine = &dev_priv->engine; + + switch (dev_priv->chipset & 0xf0) { + case 0x00: + engine->instmem.init = nv04_instmem_init; + engine->instmem.takedown = nv04_instmem_takedown; + engine->instmem.suspend = nv04_instmem_suspend; + engine->instmem.resume = nv04_instmem_resume; + engine->instmem.get = nv04_instmem_get; + engine->instmem.put = nv04_instmem_put; + engine->instmem.map = nv04_instmem_map; + engine->instmem.unmap = nv04_instmem_unmap; + engine->instmem.flush = nv04_instmem_flush; + engine->mc.init = nv04_mc_init; + engine->mc.takedown = nv04_mc_takedown; + engine->timer.init = nv04_timer_init; + engine->timer.read = nv04_timer_read; + engine->timer.takedown = nv04_timer_takedown; + engine->fb.init = nv04_fb_init; + engine->fb.takedown = nv04_fb_takedown; + engine->fifo.channels = 16; + engine->fifo.init = nv04_fifo_init; + engine->fifo.takedown = nv04_fifo_fini; + engine->fifo.disable = nv04_fifo_disable; + engine->fifo.enable = nv04_fifo_enable; + engine->fifo.reassign = nv04_fifo_reassign; + engine->fifo.cache_pull = nv04_fifo_cache_pull; + engine->fifo.channel_id = nv04_fifo_channel_id; + engine->fifo.create_context = nv04_fifo_create_context; + engine->fifo.destroy_context = nv04_fifo_destroy_context; + engine->fifo.load_context = nv04_fifo_load_context; + engine->fifo.unload_context = nv04_fifo_unload_context; + engine->display.early_init = nv04_display_early_init; + engine->display.late_takedown = nv04_display_late_takedown; + engine->display.create = nv04_display_create; + engine->display.init = nv04_display_init; + engine->display.destroy = nv04_display_destroy; + engine->gpio.init = nouveau_stub_init; + engine->gpio.takedown = nouveau_stub_takedown; + engine->gpio.get = NULL; + engine->gpio.set = NULL; + engine->gpio.irq_enable = NULL; + engine->pm.clock_get = nv04_pm_clock_get; + engine->pm.clock_pre = nv04_pm_clock_pre; + engine->pm.clock_set = nv04_pm_clock_set; + engine->vram.init = nouveau_mem_detect; + engine->vram.flags_valid = nouveau_mem_flags_valid; + break; + case 0x10: + engine->instmem.init = nv04_instmem_init; + engine->instmem.takedown = nv04_instmem_takedown; + engine->instmem.suspend = nv04_instmem_suspend; + engine->instmem.resume = nv04_instmem_resume; + engine->instmem.get = nv04_instmem_get; + engine->instmem.put = nv04_instmem_put; + engine->instmem.map = nv04_instmem_map; + engine->instmem.unmap = nv04_instmem_unmap; + engine->instmem.flush = nv04_instmem_flush; + engine->mc.init = nv04_mc_init; + engine->mc.takedown = nv04_mc_takedown; + engine->timer.init = nv04_timer_init; + engine->timer.read = nv04_timer_read; + engine->timer.takedown = nv04_timer_takedown; + engine->fb.init = nv10_fb_init; + engine->fb.takedown = nv10_fb_takedown; + engine->fb.init_tile_region = nv10_fb_init_tile_region; + engine->fb.set_tile_region = nv10_fb_set_tile_region; + engine->fb.free_tile_region = nv10_fb_free_tile_region; + engine->fifo.channels = 32; + engine->fifo.init = nv10_fifo_init; + engine->fifo.takedown = nv04_fifo_fini; + engine->fifo.disable = nv04_fifo_disable; + engine->fifo.enable = nv04_fifo_enable; + engine->fifo.reassign = nv04_fifo_reassign; + engine->fifo.cache_pull = nv04_fifo_cache_pull; + engine->fifo.channel_id = nv10_fifo_channel_id; + engine->fifo.create_context = nv10_fifo_create_context; + engine->fifo.destroy_context = nv04_fifo_destroy_context; + engine->fifo.load_context = nv10_fifo_load_context; + engine->fifo.unload_context = nv10_fifo_unload_context; + engine->display.early_init = nv04_display_early_init; + engine->display.late_takedown = nv04_display_late_takedown; + engine->display.create = nv04_display_create; + engine->display.init = nv04_display_init; + engine->display.destroy = nv04_display_destroy; + engine->gpio.init = nouveau_stub_init; + engine->gpio.takedown = nouveau_stub_takedown; + engine->gpio.get = nv10_gpio_get; + engine->gpio.set = nv10_gpio_set; + engine->gpio.irq_enable = NULL; + engine->pm.clock_get = nv04_pm_clock_get; + engine->pm.clock_pre = nv04_pm_clock_pre; + engine->pm.clock_set = nv04_pm_clock_set; + engine->vram.init = nouveau_mem_detect; + engine->vram.flags_valid = nouveau_mem_flags_valid; + break; + case 0x20: + engine->instmem.init = nv04_instmem_init; + engine->instmem.takedown = nv04_instmem_takedown; + engine->instmem.suspend = nv04_instmem_suspend; + engine->instmem.resume = nv04_instmem_resume; + engine->instmem.get = nv04_instmem_get; + engine->instmem.put = nv04_instmem_put; + engine->instmem.map = nv04_instmem_map; + engine->instmem.unmap = nv04_instmem_unmap; + engine->instmem.flush = nv04_instmem_flush; + engine->mc.init = nv04_mc_init; + engine->mc.takedown = nv04_mc_takedown; + engine->timer.init = nv04_timer_init; + engine->timer.read = nv04_timer_read; + engine->timer.takedown = nv04_timer_takedown; + engine->fb.init = nv10_fb_init; + engine->fb.takedown = nv10_fb_takedown; + engine->fb.init_tile_region = nv10_fb_init_tile_region; + engine->fb.set_tile_region = nv10_fb_set_tile_region; + engine->fb.free_tile_region = nv10_fb_free_tile_region; + engine->fifo.channels = 32; + engine->fifo.init = nv10_fifo_init; + engine->fifo.takedown = nv04_fifo_fini; + engine->fifo.disable = nv04_fifo_disable; + engine->fifo.enable = nv04_fifo_enable; + engine->fifo.reassign = nv04_fifo_reassign; + engine->fifo.cache_pull = nv04_fifo_cache_pull; + engine->fifo.channel_id = nv10_fifo_channel_id; + engine->fifo.create_context = nv10_fifo_create_context; + engine->fifo.destroy_context = nv04_fifo_destroy_context; + engine->fifo.load_context = nv10_fifo_load_context; + engine->fifo.unload_context = nv10_fifo_unload_context; + engine->display.early_init = nv04_display_early_init; + engine->display.late_takedown = nv04_display_late_takedown; + engine->display.create = nv04_display_create; + engine->display.init = nv04_display_init; + engine->display.destroy = nv04_display_destroy; + engine->gpio.init = nouveau_stub_init; + engine->gpio.takedown = nouveau_stub_takedown; + engine->gpio.get = nv10_gpio_get; + engine->gpio.set = nv10_gpio_set; + engine->gpio.irq_enable = NULL; + engine->pm.clock_get = nv04_pm_clock_get; + engine->pm.clock_pre = nv04_pm_clock_pre; + engine->pm.clock_set = nv04_pm_clock_set; + engine->vram.init = nouveau_mem_detect; + engine->vram.flags_valid = nouveau_mem_flags_valid; + break; + case 0x30: + engine->instmem.init = nv04_instmem_init; + engine->instmem.takedown = nv04_instmem_takedown; + engine->instmem.suspend = nv04_instmem_suspend; + engine->instmem.resume = nv04_instmem_resume; + engine->instmem.get = nv04_instmem_get; + engine->instmem.put = nv04_instmem_put; + engine->instmem.map = nv04_instmem_map; + engine->instmem.unmap = nv04_instmem_unmap; + engine->instmem.flush = nv04_instmem_flush; + engine->mc.init = nv04_mc_init; + engine->mc.takedown = nv04_mc_takedown; + engine->timer.init = nv04_timer_init; + engine->timer.read = nv04_timer_read; + engine->timer.takedown = nv04_timer_takedown; + engine->fb.init = nv30_fb_init; + engine->fb.takedown = nv30_fb_takedown; + engine->fb.init_tile_region = nv30_fb_init_tile_region; + engine->fb.set_tile_region = nv10_fb_set_tile_region; + engine->fb.free_tile_region = nv30_fb_free_tile_region; + engine->fifo.channels = 32; + engine->fifo.init = nv10_fifo_init; + engine->fifo.takedown = nv04_fifo_fini; + engine->fifo.disable = nv04_fifo_disable; + engine->fifo.enable = nv04_fifo_enable; + engine->fifo.reassign = nv04_fifo_reassign; + engine->fifo.cache_pull = nv04_fifo_cache_pull; + engine->fifo.channel_id = nv10_fifo_channel_id; + engine->fifo.create_context = nv10_fifo_create_context; + engine->fifo.destroy_context = nv04_fifo_destroy_context; + engine->fifo.load_context = nv10_fifo_load_context; + engine->fifo.unload_context = nv10_fifo_unload_context; + engine->display.early_init = nv04_display_early_init; + engine->display.late_takedown = nv04_display_late_takedown; + engine->display.create = nv04_display_create; + engine->display.init = nv04_display_init; + engine->display.destroy = nv04_display_destroy; + engine->gpio.init = nouveau_stub_init; + engine->gpio.takedown = nouveau_stub_takedown; + engine->gpio.get = nv10_gpio_get; + engine->gpio.set = nv10_gpio_set; + engine->gpio.irq_enable = NULL; + engine->pm.clock_get = nv04_pm_clock_get; + engine->pm.clock_pre = nv04_pm_clock_pre; + engine->pm.clock_set = nv04_pm_clock_set; + engine->pm.voltage_get = nouveau_voltage_gpio_get; + engine->pm.voltage_set = nouveau_voltage_gpio_set; + engine->vram.init = nouveau_mem_detect; + engine->vram.flags_valid = nouveau_mem_flags_valid; + break; + case 0x40: + case 0x60: + engine->instmem.init = nv04_instmem_init; + engine->instmem.takedown = nv04_instmem_takedown; + engine->instmem.suspend = nv04_instmem_suspend; + engine->instmem.resume = nv04_instmem_resume; + engine->instmem.get = nv04_instmem_get; + engine->instmem.put = nv04_instmem_put; + engine->instmem.map = nv04_instmem_map; + engine->instmem.unmap = nv04_instmem_unmap; + engine->instmem.flush = nv04_instmem_flush; + engine->mc.init = nv40_mc_init; + engine->mc.takedown = nv40_mc_takedown; + engine->timer.init = nv04_timer_init; + engine->timer.read = nv04_timer_read; + engine->timer.takedown = nv04_timer_takedown; + engine->fb.init = nv40_fb_init; + engine->fb.takedown = nv40_fb_takedown; + engine->fb.init_tile_region = nv30_fb_init_tile_region; + engine->fb.set_tile_region = nv40_fb_set_tile_region; + engine->fb.free_tile_region = nv30_fb_free_tile_region; + engine->fifo.channels = 32; + engine->fifo.init = nv40_fifo_init; + engine->fifo.takedown = nv04_fifo_fini; + engine->fifo.disable = nv04_fifo_disable; + engine->fifo.enable = nv04_fifo_enable; + engine->fifo.reassign = nv04_fifo_reassign; + engine->fifo.cache_pull = nv04_fifo_cache_pull; + engine->fifo.channel_id = nv10_fifo_channel_id; + engine->fifo.create_context = nv40_fifo_create_context; + engine->fifo.destroy_context = nv04_fifo_destroy_context; + engine->fifo.load_context = nv40_fifo_load_context; + engine->fifo.unload_context = nv40_fifo_unload_context; + engine->display.early_init = nv04_display_early_init; + engine->display.late_takedown = nv04_display_late_takedown; + engine->display.create = nv04_display_create; + engine->display.init = nv04_display_init; + engine->display.destroy = nv04_display_destroy; + engine->gpio.init = nouveau_stub_init; + engine->gpio.takedown = nouveau_stub_takedown; + engine->gpio.get = nv10_gpio_get; + engine->gpio.set = nv10_gpio_set; + engine->gpio.irq_enable = NULL; + engine->pm.clock_get = nv04_pm_clock_get; + engine->pm.clock_pre = nv04_pm_clock_pre; + engine->pm.clock_set = nv04_pm_clock_set; + engine->pm.voltage_get = nouveau_voltage_gpio_get; + engine->pm.voltage_set = nouveau_voltage_gpio_set; + engine->pm.temp_get = nv40_temp_get; + engine->vram.init = nouveau_mem_detect; + engine->vram.flags_valid = nouveau_mem_flags_valid; + break; + case 0x50: + case 0x80: /* gotta love NVIDIA's consistency.. */ + case 0x90: + case 0xA0: + engine->instmem.init = nv50_instmem_init; + engine->instmem.takedown = nv50_instmem_takedown; + engine->instmem.suspend = nv50_instmem_suspend; + engine->instmem.resume = nv50_instmem_resume; + engine->instmem.get = nv50_instmem_get; + engine->instmem.put = nv50_instmem_put; + engine->instmem.map = nv50_instmem_map; + engine->instmem.unmap = nv50_instmem_unmap; + if (dev_priv->chipset == 0x50) + engine->instmem.flush = nv50_instmem_flush; + else + engine->instmem.flush = nv84_instmem_flush; + engine->mc.init = nv50_mc_init; + engine->mc.takedown = nv50_mc_takedown; + engine->timer.init = nv04_timer_init; + engine->timer.read = nv04_timer_read; + engine->timer.takedown = nv04_timer_takedown; + engine->fb.init = nv50_fb_init; + engine->fb.takedown = nv50_fb_takedown; + engine->fifo.channels = 128; + engine->fifo.init = nv50_fifo_init; + engine->fifo.takedown = nv50_fifo_takedown; + engine->fifo.disable = nv04_fifo_disable; + engine->fifo.enable = nv04_fifo_enable; + engine->fifo.reassign = nv04_fifo_reassign; + engine->fifo.channel_id = nv50_fifo_channel_id; + engine->fifo.create_context = nv50_fifo_create_context; + engine->fifo.destroy_context = nv50_fifo_destroy_context; + engine->fifo.load_context = nv50_fifo_load_context; + engine->fifo.unload_context = nv50_fifo_unload_context; + engine->fifo.tlb_flush = nv50_fifo_tlb_flush; + engine->display.early_init = nv50_display_early_init; + engine->display.late_takedown = nv50_display_late_takedown; + engine->display.create = nv50_display_create; + engine->display.init = nv50_display_init; + engine->display.destroy = nv50_display_destroy; + engine->gpio.init = nv50_gpio_init; + engine->gpio.takedown = nv50_gpio_fini; + engine->gpio.get = nv50_gpio_get; + engine->gpio.set = nv50_gpio_set; + engine->gpio.irq_register = nv50_gpio_irq_register; + engine->gpio.irq_unregister = nv50_gpio_irq_unregister; + engine->gpio.irq_enable = nv50_gpio_irq_enable; + switch (dev_priv->chipset) { + case 0x84: + case 0x86: + case 0x92: + case 0x94: + case 0x96: + case 0x98: + case 0xa0: + case 0xaa: + case 0xac: + case 0x50: + engine->pm.clock_get = nv50_pm_clock_get; + engine->pm.clock_pre = nv50_pm_clock_pre; + engine->pm.clock_set = nv50_pm_clock_set; + break; + default: + engine->pm.clock_get = nva3_pm_clock_get; + engine->pm.clock_pre = nva3_pm_clock_pre; + engine->pm.clock_set = nva3_pm_clock_set; + break; + } + engine->pm.voltage_get = nouveau_voltage_gpio_get; + engine->pm.voltage_set = nouveau_voltage_gpio_set; + if (dev_priv->chipset >= 0x84) + engine->pm.temp_get = nv84_temp_get; + else + engine->pm.temp_get = nv40_temp_get; + engine->vram.init = nv50_vram_init; + engine->vram.get = nv50_vram_new; + engine->vram.put = nv50_vram_del; + engine->vram.flags_valid = nv50_vram_flags_valid; + break; + case 0xC0: + engine->instmem.init = nvc0_instmem_init; + engine->instmem.takedown = nvc0_instmem_takedown; + engine->instmem.suspend = nvc0_instmem_suspend; + engine->instmem.resume = nvc0_instmem_resume; + engine->instmem.get = nv50_instmem_get; + engine->instmem.put = nv50_instmem_put; + engine->instmem.map = nv50_instmem_map; + engine->instmem.unmap = nv50_instmem_unmap; + engine->instmem.flush = nv84_instmem_flush; + engine->mc.init = nv50_mc_init; + engine->mc.takedown = nv50_mc_takedown; + engine->timer.init = nv04_timer_init; + engine->timer.read = nv04_timer_read; + engine->timer.takedown = nv04_timer_takedown; + engine->fb.init = nvc0_fb_init; + engine->fb.takedown = nvc0_fb_takedown; + engine->fifo.channels = 128; + engine->fifo.init = nvc0_fifo_init; + engine->fifo.takedown = nvc0_fifo_takedown; + engine->fifo.disable = nvc0_fifo_disable; + engine->fifo.enable = nvc0_fifo_enable; + engine->fifo.reassign = nvc0_fifo_reassign; + engine->fifo.channel_id = nvc0_fifo_channel_id; + engine->fifo.create_context = nvc0_fifo_create_context; + engine->fifo.destroy_context = nvc0_fifo_destroy_context; + engine->fifo.load_context = nvc0_fifo_load_context; + engine->fifo.unload_context = nvc0_fifo_unload_context; + engine->display.early_init = nv50_display_early_init; + engine->display.late_takedown = nv50_display_late_takedown; + engine->display.create = nv50_display_create; + engine->display.init = nv50_display_init; + engine->display.destroy = nv50_display_destroy; + engine->gpio.init = nv50_gpio_init; + engine->gpio.takedown = nouveau_stub_takedown; + engine->gpio.get = nv50_gpio_get; + engine->gpio.set = nv50_gpio_set; + engine->gpio.irq_register = nv50_gpio_irq_register; + engine->gpio.irq_unregister = nv50_gpio_irq_unregister; + engine->gpio.irq_enable = nv50_gpio_irq_enable; + engine->vram.init = nvc0_vram_init; + engine->vram.get = nvc0_vram_new; + engine->vram.put = nv50_vram_del; + engine->vram.flags_valid = nvc0_vram_flags_valid; + break; + default: + NV_ERROR(dev, "NV%02x unsupported\n", dev_priv->chipset); + return 1; + } + + return 0; +} + +static unsigned int +nouveau_vga_set_decode(void *priv, bool state) +{ + struct drm_device *dev = priv; + struct drm_nouveau_private *dev_priv = dev->dev_private; + + if (dev_priv->chipset >= 0x40) + nv_wr32(dev, 0x88054, state); + else + nv_wr32(dev, 0x1854, state); + + if (state) + return VGA_RSRC_LEGACY_IO | VGA_RSRC_LEGACY_MEM | + VGA_RSRC_NORMAL_IO | VGA_RSRC_NORMAL_MEM; + else + return VGA_RSRC_NORMAL_IO | VGA_RSRC_NORMAL_MEM; +} + +static int +nouveau_card_init_channel(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + int ret; + + ret = nouveau_channel_alloc(dev, &dev_priv->channel, + (struct drm_file *)-2, NvDmaFB, NvDmaTT); + if (ret) + return ret; + + mutex_unlock(&dev_priv->channel->mutex); + return 0; +} + +static void nouveau_switcheroo_set_state(struct pci_dev *pdev, + enum vga_switcheroo_state state) +{ + struct drm_device *dev = pci_get_drvdata(pdev); + pm_message_t pmm = { .event = PM_EVENT_SUSPEND }; + if (state == VGA_SWITCHEROO_ON) { + printk(KERN_ERR "VGA switcheroo: switched nouveau on\n"); + dev->switch_power_state = DRM_SWITCH_POWER_CHANGING; + nouveau_pci_resume(pdev); + drm_kms_helper_poll_enable(dev); + dev->switch_power_state = DRM_SWITCH_POWER_ON; + } else { + printk(KERN_ERR "VGA switcheroo: switched nouveau off\n"); + dev->switch_power_state = DRM_SWITCH_POWER_CHANGING; + drm_kms_helper_poll_disable(dev); + nouveau_pci_suspend(pdev, pmm); + dev->switch_power_state = DRM_SWITCH_POWER_OFF; + } +} + +static void nouveau_switcheroo_reprobe(struct pci_dev *pdev) +{ + struct drm_device *dev = pci_get_drvdata(pdev); + nouveau_fbcon_output_poll_changed(dev); +} + +static bool nouveau_switcheroo_can_switch(struct pci_dev *pdev) +{ + struct drm_device *dev = pci_get_drvdata(pdev); + bool can_switch; + + spin_lock(&dev->count_lock); + can_switch = (dev->open_count == 0); + spin_unlock(&dev->count_lock); + return can_switch; +} + +int +nouveau_card_init(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_engine *engine; + int ret, e = 0; + + vga_client_register(dev->pdev, dev, NULL, nouveau_vga_set_decode); + vga_switcheroo_register_client(dev->pdev, nouveau_switcheroo_set_state, + nouveau_switcheroo_reprobe, + nouveau_switcheroo_can_switch); + + /* Initialise internal driver API hooks */ + ret = nouveau_init_engine_ptrs(dev); + if (ret) + goto out; + engine = &dev_priv->engine; + spin_lock_init(&dev_priv->channels.lock); + spin_lock_init(&dev_priv->tile.lock); + spin_lock_init(&dev_priv->context_switch_lock); + spin_lock_init(&dev_priv->vm_lock); + + /* Make the CRTCs and I2C buses accessible */ + ret = engine->display.early_init(dev); + if (ret) + goto out; + + /* Parse BIOS tables / Run init tables if card not POSTed */ + ret = nouveau_bios_init(dev); + if (ret) + goto out_display_early; + + nouveau_pm_init(dev); + + ret = nouveau_mem_vram_init(dev); + if (ret) + goto out_bios; + + ret = nouveau_gpuobj_init(dev); + if (ret) + goto out_vram; + + ret = engine->instmem.init(dev); + if (ret) + goto out_gpuobj; + + ret = nouveau_mem_gart_init(dev); + if (ret) + goto out_instmem; + + /* PMC */ + ret = engine->mc.init(dev); + if (ret) + goto out_gart; + + /* PGPIO */ + ret = engine->gpio.init(dev); + if (ret) + goto out_mc; + + /* PTIMER */ + ret = engine->timer.init(dev); + if (ret) + goto out_gpio; + + /* PFB */ + ret = engine->fb.init(dev); + if (ret) + goto out_timer; + + if (!nouveau_noaccel) { + switch (dev_priv->card_type) { + case NV_04: + nv04_graph_create(dev); + break; + case NV_10: + nv10_graph_create(dev); + break; + case NV_20: + case NV_30: + nv20_graph_create(dev); + break; + case NV_40: + nv40_graph_create(dev); + break; + case NV_50: + nv50_graph_create(dev); + break; + case NV_C0: + nvc0_graph_create(dev); + break; + default: + break; + } + + switch (dev_priv->chipset) { + case 0x84: + case 0x86: + case 0x92: + case 0x94: + case 0x96: + case 0xa0: + nv84_crypt_create(dev); + break; + } + + switch (dev_priv->card_type) { + case NV_50: + switch (dev_priv->chipset) { + case 0xa3: + case 0xa5: + case 0xa8: + case 0xaf: + nva3_copy_create(dev); + break; + } + break; + case NV_C0: + nvc0_copy_create(dev, 0); + nvc0_copy_create(dev, 1); + break; + default: + break; + } + + if (dev_priv->card_type == NV_40) + nv40_mpeg_create(dev); + else + if (dev_priv->card_type == NV_50 && + (dev_priv->chipset < 0x98 || dev_priv->chipset == 0xa0)) + nv50_mpeg_create(dev); + + for (e = 0; e < NVOBJ_ENGINE_NR; e++) { + if (dev_priv->eng[e]) { + ret = dev_priv->eng[e]->init(dev, e); + if (ret) + goto out_engine; + } + } + + /* PFIFO */ + ret = engine->fifo.init(dev); + if (ret) + goto out_engine; + } + + ret = engine->display.create(dev); + if (ret) + goto out_fifo; + + ret = drm_vblank_init(dev, nv_two_heads(dev) ? 2 : 1); + if (ret) + goto out_vblank; + + ret = nouveau_irq_init(dev); + if (ret) + goto out_vblank; + + /* what about PVIDEO/PCRTC/PRAMDAC etc? */ + + if (dev_priv->eng[NVOBJ_ENGINE_GR]) { + ret = nouveau_fence_init(dev); + if (ret) + goto out_irq; + + ret = nouveau_card_init_channel(dev); + if (ret) + goto out_fence; + } + + nouveau_fbcon_init(dev); + drm_kms_helper_poll_init(dev); + return 0; + +out_fence: + nouveau_fence_fini(dev); +out_irq: + nouveau_irq_fini(dev); +out_vblank: + drm_vblank_cleanup(dev); + engine->display.destroy(dev); +out_fifo: + if (!nouveau_noaccel) + engine->fifo.takedown(dev); +out_engine: + if (!nouveau_noaccel) { + for (e = e - 1; e >= 0; e--) { + if (!dev_priv->eng[e]) + continue; + dev_priv->eng[e]->fini(dev, e); + dev_priv->eng[e]->destroy(dev,e ); + } + } + + engine->fb.takedown(dev); +out_timer: + engine->timer.takedown(dev); +out_gpio: + engine->gpio.takedown(dev); +out_mc: + engine->mc.takedown(dev); +out_gart: + nouveau_mem_gart_fini(dev); +out_instmem: + engine->instmem.takedown(dev); +out_gpuobj: + nouveau_gpuobj_takedown(dev); +out_vram: + nouveau_mem_vram_fini(dev); +out_bios: + nouveau_pm_fini(dev); + nouveau_bios_takedown(dev); +out_display_early: + engine->display.late_takedown(dev); +out: + vga_client_register(dev->pdev, NULL, NULL, NULL); + return ret; +} + +static void nouveau_card_takedown(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_engine *engine = &dev_priv->engine; + int e; + + if (dev_priv->channel) { + nouveau_fence_fini(dev); + nouveau_channel_put_unlocked(&dev_priv->channel); + } + + if (!nouveau_noaccel) { + engine->fifo.takedown(dev); + for (e = NVOBJ_ENGINE_NR - 1; e >= 0; e--) { + if (dev_priv->eng[e]) { + dev_priv->eng[e]->fini(dev, e); + dev_priv->eng[e]->destroy(dev,e ); + } + } + } + engine->fb.takedown(dev); + engine->timer.takedown(dev); + engine->gpio.takedown(dev); + engine->mc.takedown(dev); + engine->display.late_takedown(dev); + + if (dev_priv->vga_ram) { + nouveau_bo_unpin(dev_priv->vga_ram); + nouveau_bo_ref(NULL, &dev_priv->vga_ram); + } + + mutex_lock(&dev->struct_mutex); + ttm_bo_clean_mm(&dev_priv->ttm.bdev, TTM_PL_VRAM); + ttm_bo_clean_mm(&dev_priv->ttm.bdev, TTM_PL_TT); + mutex_unlock(&dev->struct_mutex); + nouveau_mem_gart_fini(dev); + + engine->instmem.takedown(dev); + nouveau_gpuobj_takedown(dev); + nouveau_mem_vram_fini(dev); + + nouveau_irq_fini(dev); + drm_vblank_cleanup(dev); + + nouveau_pm_fini(dev); + nouveau_bios_takedown(dev); + + vga_client_register(dev->pdev, NULL, NULL, NULL); +} + +/* here a client dies, release the stuff that was allocated for its + * file_priv */ +void nouveau_preclose(struct drm_device *dev, struct drm_file *file_priv) +{ + nouveau_channel_cleanup(dev, file_priv); +} + +/* first module load, setup the mmio/fb mapping */ +/* KMS: we need mmio at load time, not when the first drm client opens. */ +int nouveau_firstopen(struct drm_device *dev) +{ + return 0; +} + +/* if we have an OF card, copy vbios to RAMIN */ +static void nouveau_OF_copy_vbios_to_ramin(struct drm_device *dev) +{ +#if defined(__powerpc__) + int size, i; + const uint32_t *bios; + struct device_node *dn = pci_device_to_OF_node(dev->pdev); + if (!dn) { + NV_INFO(dev, "Unable to get the OF node\n"); + return; + } + + bios = of_get_property(dn, "NVDA,BMP", &size); + if (bios) { + for (i = 0; i < size; i += 4) + nv_wi32(dev, i, bios[i/4]); + NV_INFO(dev, "OF bios successfully copied (%d bytes)\n", size); + } else { + NV_INFO(dev, "Unable to get the OF bios\n"); + } +#endif +} + +static struct apertures_struct *nouveau_get_apertures(struct drm_device *dev) +{ + struct pci_dev *pdev = dev->pdev; + struct apertures_struct *aper = alloc_apertures(3); + if (!aper) + return NULL; + + aper->ranges[0].base = pci_resource_start(pdev, 1); + aper->ranges[0].size = pci_resource_len(pdev, 1); + aper->count = 1; + + if (pci_resource_len(pdev, 2)) { + aper->ranges[aper->count].base = pci_resource_start(pdev, 2); + aper->ranges[aper->count].size = pci_resource_len(pdev, 2); + aper->count++; + } + + if (pci_resource_len(pdev, 3)) { + aper->ranges[aper->count].base = pci_resource_start(pdev, 3); + aper->ranges[aper->count].size = pci_resource_len(pdev, 3); + aper->count++; + } + + return aper; +} + +static int nouveau_remove_conflicting_drivers(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + bool primary = false; + dev_priv->apertures = nouveau_get_apertures(dev); + if (!dev_priv->apertures) + return -ENOMEM; + +#ifdef CONFIG_X86 + primary = dev->pdev->resource[PCI_ROM_RESOURCE].flags & IORESOURCE_ROM_SHADOW; +#endif + + remove_conflicting_framebuffers(dev_priv->apertures, "nouveaufb", primary); + return 0; +} + +int nouveau_load(struct drm_device *dev, unsigned long flags) +{ + struct drm_nouveau_private *dev_priv; + uint32_t reg0; + resource_size_t mmio_start_offs; + int ret; + + dev_priv = kzalloc(sizeof(*dev_priv), GFP_KERNEL); + if (!dev_priv) { + ret = -ENOMEM; + goto err_out; + } + dev->dev_private = dev_priv; + dev_priv->dev = dev; + + dev_priv->flags = flags & NOUVEAU_FLAGS; + + NV_DEBUG(dev, "vendor: 0x%X device: 0x%X class: 0x%X\n", + dev->pci_vendor, dev->pci_device, dev->pdev->class); + + /* resource 0 is mmio regs */ + /* resource 1 is linear FB */ + /* resource 2 is RAMIN (mmio regs + 0x1000000) */ + /* resource 6 is bios */ + + /* map the mmio regs */ + mmio_start_offs = pci_resource_start(dev->pdev, 0); + dev_priv->mmio = ioremap(mmio_start_offs, 0x00800000); + if (!dev_priv->mmio) { + NV_ERROR(dev, "Unable to initialize the mmio mapping. " + "Please report your setup to " DRIVER_EMAIL "\n"); + ret = -EINVAL; + goto err_priv; + } + NV_DEBUG(dev, "regs mapped ok at 0x%llx\n", + (unsigned long long)mmio_start_offs); + +#ifdef __BIG_ENDIAN + /* Put the card in BE mode if it's not */ + if (nv_rd32(dev, NV03_PMC_BOOT_1) != 0x01000001) + nv_wr32(dev, NV03_PMC_BOOT_1, 0x01000001); + + DRM_MEMORYBARRIER(); +#endif + + /* Time to determine the card architecture */ + reg0 = nv_rd32(dev, NV03_PMC_BOOT_0); + dev_priv->stepping = 0; /* XXX: add stepping for pre-NV10? */ + + /* We're dealing with >=NV10 */ + if ((reg0 & 0x0f000000) > 0) { + /* Bit 27-20 contain the architecture in hex */ + dev_priv->chipset = (reg0 & 0xff00000) >> 20; + dev_priv->stepping = (reg0 & 0xff); + /* NV04 or NV05 */ + } else if ((reg0 & 0xff00fff0) == 0x20004000) { + if (reg0 & 0x00f00000) + dev_priv->chipset = 0x05; + else + dev_priv->chipset = 0x04; + } else + dev_priv->chipset = 0xff; + + switch (dev_priv->chipset & 0xf0) { + case 0x00: + case 0x10: + case 0x20: + case 0x30: + dev_priv->card_type = dev_priv->chipset & 0xf0; + break; + case 0x40: + case 0x60: + dev_priv->card_type = NV_40; + break; + case 0x50: + case 0x80: + case 0x90: + case 0xa0: + dev_priv->card_type = NV_50; + break; + case 0xc0: + dev_priv->card_type = NV_C0; + break; + default: + NV_INFO(dev, "Unsupported chipset 0x%08x\n", reg0); + ret = -EINVAL; + goto err_mmio; + } + + NV_INFO(dev, "Detected an NV%2x generation card (0x%08x)\n", + dev_priv->card_type, reg0); + + ret = nouveau_remove_conflicting_drivers(dev); + if (ret) + goto err_mmio; + + /* Map PRAMIN BAR, or on older cards, the aperture within BAR0 */ + if (dev_priv->card_type >= NV_40) { + int ramin_bar = 2; + if (pci_resource_len(dev->pdev, ramin_bar) == 0) + ramin_bar = 3; + + dev_priv->ramin_size = pci_resource_len(dev->pdev, ramin_bar); + dev_priv->ramin = + ioremap(pci_resource_start(dev->pdev, ramin_bar), + dev_priv->ramin_size); + if (!dev_priv->ramin) { + NV_ERROR(dev, "Failed to PRAMIN BAR"); + ret = -ENOMEM; + goto err_mmio; + } + } else { + dev_priv->ramin_size = 1 * 1024 * 1024; + dev_priv->ramin = ioremap(mmio_start_offs + NV_RAMIN, + dev_priv->ramin_size); + if (!dev_priv->ramin) { + NV_ERROR(dev, "Failed to map BAR0 PRAMIN.\n"); + ret = -ENOMEM; + goto err_mmio; + } + } + + nouveau_OF_copy_vbios_to_ramin(dev); + + /* Special flags */ + if (dev->pci_device == 0x01a0) + dev_priv->flags |= NV_NFORCE; + else if (dev->pci_device == 0x01f0) + dev_priv->flags |= NV_NFORCE2; + + /* For kernel modesetting, init card now and bring up fbcon */ + ret = nouveau_card_init(dev); + if (ret) + goto err_ramin; + + return 0; + +err_ramin: + iounmap(dev_priv->ramin); +err_mmio: + iounmap(dev_priv->mmio); +err_priv: + kfree(dev_priv); + dev->dev_private = NULL; +err_out: + return ret; +} + +void nouveau_lastclose(struct drm_device *dev) +{ + vga_switcheroo_process_delayed_switch(); +} + +int nouveau_unload(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_engine *engine = &dev_priv->engine; + + drm_kms_helper_poll_fini(dev); + nouveau_fbcon_fini(dev); + engine->display.destroy(dev); + nouveau_card_takedown(dev); + + iounmap(dev_priv->mmio); + iounmap(dev_priv->ramin); + + kfree(dev_priv); + dev->dev_private = NULL; + return 0; +} + +int nouveau_ioctl_getparam(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct drm_nouveau_getparam *getparam = data; + + switch (getparam->param) { + case NOUVEAU_GETPARAM_CHIPSET_ID: + getparam->value = dev_priv->chipset; + break; + case NOUVEAU_GETPARAM_PCI_VENDOR: + getparam->value = dev->pci_vendor; + break; + case NOUVEAU_GETPARAM_PCI_DEVICE: + getparam->value = dev->pci_device; + break; + case NOUVEAU_GETPARAM_BUS_TYPE: + if (drm_pci_device_is_agp(dev)) + getparam->value = NV_AGP; + else if (drm_pci_device_is_pcie(dev)) + getparam->value = NV_PCIE; + else + getparam->value = NV_PCI; + break; + case NOUVEAU_GETPARAM_FB_SIZE: + getparam->value = dev_priv->fb_available_size; + break; + case NOUVEAU_GETPARAM_AGP_SIZE: + getparam->value = dev_priv->gart_info.aper_size; + break; + case NOUVEAU_GETPARAM_VM_VRAM_BASE: + getparam->value = 0; /* deprecated */ + break; + case NOUVEAU_GETPARAM_PTIMER_TIME: + getparam->value = dev_priv->engine.timer.read(dev); + break; + case NOUVEAU_GETPARAM_HAS_BO_USAGE: + getparam->value = 1; + break; + case NOUVEAU_GETPARAM_HAS_PAGEFLIP: + getparam->value = 1; + break; + case NOUVEAU_GETPARAM_GRAPH_UNITS: + /* NV40 and NV50 versions are quite different, but register + * address is the same. User is supposed to know the card + * family anyway... */ + if (dev_priv->chipset >= 0x40) { + getparam->value = nv_rd32(dev, NV40_PMC_GRAPH_UNITS); + break; + } + /* FALLTHRU */ + default: + NV_DEBUG(dev, "unknown parameter %lld\n", getparam->param); + return -EINVAL; + } + + return 0; +} + +int +nouveau_ioctl_setparam(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_nouveau_setparam *setparam = data; + + switch (setparam->param) { + default: + NV_DEBUG(dev, "unknown parameter %lld\n", setparam->param); + return -EINVAL; + } + + return 0; +} + +/* Wait until (value(reg) & mask) == val, up until timeout has hit */ +bool +nouveau_wait_eq(struct drm_device *dev, uint64_t timeout, + uint32_t reg, uint32_t mask, uint32_t val) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_timer_engine *ptimer = &dev_priv->engine.timer; + uint64_t start = ptimer->read(dev); + + do { + if ((nv_rd32(dev, reg) & mask) == val) + return true; + } while (ptimer->read(dev) - start < timeout); + + return false; +} + +/* Wait until (value(reg) & mask) != val, up until timeout has hit */ +bool +nouveau_wait_ne(struct drm_device *dev, uint64_t timeout, + uint32_t reg, uint32_t mask, uint32_t val) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_timer_engine *ptimer = &dev_priv->engine.timer; + uint64_t start = ptimer->read(dev); + + do { + if ((nv_rd32(dev, reg) & mask) != val) + return true; + } while (ptimer->read(dev) - start < timeout); + + return false; +} + +/* Waits for PGRAPH to go completely idle */ +bool nouveau_wait_for_idle(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + uint32_t mask = ~0; + + if (dev_priv->card_type == NV_40) + mask &= ~NV40_PGRAPH_STATUS_SYNC_STALL; + + if (!nv_wait(dev, NV04_PGRAPH_STATUS, mask, 0)) { + NV_ERROR(dev, "PGRAPH idle timed out with status 0x%08x\n", + nv_rd32(dev, NV04_PGRAPH_STATUS)); + return false; + } + + return true; +} + diff --git a/drivers/gpu/drm/nouveau/nouveau_temp.c b/drivers/gpu/drm/nouveau/nouveau_temp.c new file mode 100644 index 00000000..649b0413 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_temp.c @@ -0,0 +1,307 @@ +/* + * Copyright 2010 PathScale inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Martin Peres + */ + +#include "drmP.h" + +#include "nouveau_drv.h" +#include "nouveau_pm.h" + +static void +nouveau_temp_vbios_parse(struct drm_device *dev, u8 *temp) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_pm_engine *pm = &dev_priv->engine.pm; + struct nouveau_pm_temp_sensor_constants *sensor = &pm->sensor_constants; + struct nouveau_pm_threshold_temp *temps = &pm->threshold_temp; + int i, headerlen, recordlen, entries; + + if (!temp) { + NV_DEBUG(dev, "temperature table pointer invalid\n"); + return; + } + + /* Set the default sensor's contants */ + sensor->offset_constant = 0; + sensor->offset_mult = 1; + sensor->offset_div = 1; + sensor->slope_mult = 1; + sensor->slope_div = 1; + + /* Set the default temperature thresholds */ + temps->critical = 110; + temps->down_clock = 100; + temps->fan_boost = 90; + + /* Set the known default values to setup the temperature sensor */ + if (dev_priv->card_type >= NV_40) { + switch (dev_priv->chipset) { + case 0x43: + sensor->offset_mult = 32060; + sensor->offset_div = 1000; + sensor->slope_mult = 792; + sensor->slope_div = 1000; + break; + + case 0x44: + case 0x47: + case 0x4a: + sensor->offset_mult = 27839; + sensor->offset_div = 1000; + sensor->slope_mult = 780; + sensor->slope_div = 1000; + break; + + case 0x46: + sensor->offset_mult = -24775; + sensor->offset_div = 100; + sensor->slope_mult = 467; + sensor->slope_div = 10000; + break; + + case 0x49: + sensor->offset_mult = -25051; + sensor->offset_div = 100; + sensor->slope_mult = 458; + sensor->slope_div = 10000; + break; + + case 0x4b: + sensor->offset_mult = -24088; + sensor->offset_div = 100; + sensor->slope_mult = 442; + sensor->slope_div = 10000; + break; + + case 0x50: + sensor->offset_mult = -22749; + sensor->offset_div = 100; + sensor->slope_mult = 431; + sensor->slope_div = 10000; + break; + } + } + + headerlen = temp[1]; + recordlen = temp[2]; + entries = temp[3]; + temp = temp + headerlen; + + /* Read the entries from the table */ + for (i = 0; i < entries; i++) { + u16 value = ROM16(temp[1]); + + switch (temp[0]) { + case 0x01: + if ((value & 0x8f) == 0) + sensor->offset_constant = (value >> 9) & 0x7f; + break; + + case 0x04: + if ((value & 0xf00f) == 0xa000) /* core */ + temps->critical = (value&0x0ff0) >> 4; + break; + + case 0x07: + if ((value & 0xf00f) == 0xa000) /* core */ + temps->down_clock = (value&0x0ff0) >> 4; + break; + + case 0x08: + if ((value & 0xf00f) == 0xa000) /* core */ + temps->fan_boost = (value&0x0ff0) >> 4; + break; + + case 0x10: + sensor->offset_mult = value; + break; + + case 0x11: + sensor->offset_div = value; + break; + + case 0x12: + sensor->slope_mult = value; + break; + + case 0x13: + sensor->slope_div = value; + break; + } + temp += recordlen; + } + + nouveau_temp_safety_checks(dev); +} + +static int +nv40_sensor_setup(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_pm_engine *pm = &dev_priv->engine.pm; + struct nouveau_pm_temp_sensor_constants *sensor = &pm->sensor_constants; + u32 offset = sensor->offset_mult / sensor->offset_div; + u32 sensor_calibration; + + /* set up the sensors */ + sensor_calibration = 120 - offset - sensor->offset_constant; + sensor_calibration = sensor_calibration * sensor->slope_div / + sensor->slope_mult; + + if (dev_priv->chipset >= 0x46) + sensor_calibration |= 0x80000000; + else + sensor_calibration |= 0x10000000; + + nv_wr32(dev, 0x0015b0, sensor_calibration); + + /* Wait for the sensor to update */ + msleep(5); + + /* read */ + return nv_rd32(dev, 0x0015b4) & 0x1fff; +} + +int +nv40_temp_get(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_pm_engine *pm = &dev_priv->engine.pm; + struct nouveau_pm_temp_sensor_constants *sensor = &pm->sensor_constants; + int offset = sensor->offset_mult / sensor->offset_div; + int core_temp; + + if (dev_priv->card_type >= NV_50) { + core_temp = nv_rd32(dev, 0x20008); + } else { + core_temp = nv_rd32(dev, 0x0015b4) & 0x1fff; + /* Setup the sensor if the temperature is 0 */ + if (core_temp == 0) + core_temp = nv40_sensor_setup(dev); + } + + core_temp = core_temp * sensor->slope_mult / sensor->slope_div; + core_temp = core_temp + offset + sensor->offset_constant; + + return core_temp; +} + +int +nv84_temp_get(struct drm_device *dev) +{ + return nv_rd32(dev, 0x20400); +} + +void +nouveau_temp_safety_checks(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_pm_engine *pm = &dev_priv->engine.pm; + struct nouveau_pm_threshold_temp *temps = &pm->threshold_temp; + + if (temps->critical > 120) + temps->critical = 120; + else if (temps->critical < 80) + temps->critical = 80; + + if (temps->down_clock > 110) + temps->down_clock = 110; + else if (temps->down_clock < 60) + temps->down_clock = 60; + + if (temps->fan_boost > 100) + temps->fan_boost = 100; + else if (temps->fan_boost < 40) + temps->fan_boost = 40; +} + +static bool +probe_monitoring_device(struct nouveau_i2c_chan *i2c, + struct i2c_board_info *info) +{ + struct i2c_client *client; + + request_module("%s%s", I2C_MODULE_PREFIX, info->type); + + client = i2c_new_device(&i2c->adapter, info); + if (!client) + return false; + + if (!client->driver || client->driver->detect(client, info)) { + i2c_unregister_device(client); + return false; + } + + return true; +} + +static void +nouveau_temp_probe_i2c(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct dcb_table *dcb = &dev_priv->vbios.dcb; + struct i2c_board_info info[] = { + { I2C_BOARD_INFO("w83l785ts", 0x2d) }, + { I2C_BOARD_INFO("w83781d", 0x2d) }, + { I2C_BOARD_INFO("adt7473", 0x2e) }, + { I2C_BOARD_INFO("f75375", 0x2e) }, + { I2C_BOARD_INFO("lm99", 0x4c) }, + { } + }; + int idx = (dcb->version >= 0x40 ? + dcb->i2c_default_indices & 0xf : 2); + + nouveau_i2c_identify(dev, "monitoring device", info, + probe_monitoring_device, idx); +} + +void +nouveau_temp_init(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nvbios *bios = &dev_priv->vbios; + struct bit_entry P; + u8 *temp = NULL; + + if (bios->type == NVBIOS_BIT) { + if (bit_table(dev, 'P', &P)) + return; + + if (P.version == 1) + temp = ROMPTR(bios, P.data[12]); + else if (P.version == 2) + temp = ROMPTR(bios, P.data[16]); + else + NV_WARN(dev, "unknown temp for BIT P %d\n", P.version); + + nouveau_temp_vbios_parse(dev, temp); + } + + nouveau_temp_probe_i2c(dev); +} + +void +nouveau_temp_fini(struct drm_device *dev) +{ + +} diff --git a/drivers/gpu/drm/nouveau/nouveau_ttm.c b/drivers/gpu/drm/nouveau/nouveau_ttm.c new file mode 100644 index 00000000..bd35f930 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_ttm.c @@ -0,0 +1,103 @@ +/* + * Copyright (c) 2007-2008 Tungsten Graphics, Inc., Cedar Park, TX., USA, + * All Rights Reserved. + * Copyright (c) 2009 VMware, Inc., Palo Alto, CA., USA, + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sub license, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. + */ + +#include "drmP.h" + +#include "nouveau_drv.h" + +int +nouveau_ttm_mmap(struct file *filp, struct vm_area_struct *vma) +{ + struct drm_file *file_priv = filp->private_data; + struct drm_nouveau_private *dev_priv = + file_priv->minor->dev->dev_private; + + if (unlikely(vma->vm_pgoff < DRM_FILE_PAGE_OFFSET)) + return drm_mmap(filp, vma); + + return ttm_bo_mmap(filp, vma, &dev_priv->ttm.bdev); +} + +static int +nouveau_ttm_mem_global_init(struct drm_global_reference *ref) +{ + return ttm_mem_global_init(ref->object); +} + +static void +nouveau_ttm_mem_global_release(struct drm_global_reference *ref) +{ + ttm_mem_global_release(ref->object); +} + +int +nouveau_ttm_global_init(struct drm_nouveau_private *dev_priv) +{ + struct drm_global_reference *global_ref; + int ret; + + global_ref = &dev_priv->ttm.mem_global_ref; + global_ref->global_type = DRM_GLOBAL_TTM_MEM; + global_ref->size = sizeof(struct ttm_mem_global); + global_ref->init = &nouveau_ttm_mem_global_init; + global_ref->release = &nouveau_ttm_mem_global_release; + + ret = drm_global_item_ref(global_ref); + if (unlikely(ret != 0)) { + DRM_ERROR("Failed setting up TTM memory accounting\n"); + dev_priv->ttm.mem_global_ref.release = NULL; + return ret; + } + + dev_priv->ttm.bo_global_ref.mem_glob = global_ref->object; + global_ref = &dev_priv->ttm.bo_global_ref.ref; + global_ref->global_type = DRM_GLOBAL_TTM_BO; + global_ref->size = sizeof(struct ttm_bo_global); + global_ref->init = &ttm_bo_global_init; + global_ref->release = &ttm_bo_global_release; + + ret = drm_global_item_ref(global_ref); + if (unlikely(ret != 0)) { + DRM_ERROR("Failed setting up TTM BO subsystem\n"); + drm_global_item_unref(&dev_priv->ttm.mem_global_ref); + dev_priv->ttm.mem_global_ref.release = NULL; + return ret; + } + + return 0; +} + +void +nouveau_ttm_global_release(struct drm_nouveau_private *dev_priv) +{ + if (dev_priv->ttm.mem_global_ref.release == NULL) + return; + + drm_global_item_unref(&dev_priv->ttm.bo_global_ref.ref); + drm_global_item_unref(&dev_priv->ttm.mem_global_ref); + dev_priv->ttm.mem_global_ref.release = NULL; +} + diff --git a/drivers/gpu/drm/nouveau/nouveau_util.c b/drivers/gpu/drm/nouveau/nouveau_util.c new file mode 100644 index 00000000..e51b5150 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_util.c @@ -0,0 +1,78 @@ +/* + * Copyright (C) 2010 Nouveau Project + * + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include + +#include "nouveau_util.h" + +static DEFINE_RATELIMIT_STATE(nouveau_ratelimit_state, 3 * HZ, 20); + +void +nouveau_bitfield_print(const struct nouveau_bitfield *bf, u32 value) +{ + while (bf->name) { + if (value & bf->mask) { + printk(" %s", bf->name); + value &= ~bf->mask; + } + + bf++; + } + + if (value) + printk(" (unknown bits 0x%08x)", value); +} + +const struct nouveau_enum * +nouveau_enum_find(const struct nouveau_enum *en, u32 value) +{ + while (en->name) { + if (en->value == value) + return en; + en++; + } + + return NULL; +} + +void +nouveau_enum_print(const struct nouveau_enum *en, u32 value) +{ + en = nouveau_enum_find(en, value); + if (en) { + printk("%s", en->name); + return; + } + + printk("(unknown enum 0x%08x)", value); +} + +int +nouveau_ratelimit(void) +{ + return __ratelimit(&nouveau_ratelimit_state); +} diff --git a/drivers/gpu/drm/nouveau/nouveau_util.h b/drivers/gpu/drm/nouveau/nouveau_util.h new file mode 100644 index 00000000..b97719fb --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_util.h @@ -0,0 +1,49 @@ +/* + * Copyright (C) 2010 Nouveau Project + * + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __NOUVEAU_UTIL_H__ +#define __NOUVEAU_UTIL_H__ + +struct nouveau_bitfield { + u32 mask; + const char *name; +}; + +struct nouveau_enum { + u32 value; + const char *name; + void *data; +}; + +void nouveau_bitfield_print(const struct nouveau_bitfield *, u32 value); +void nouveau_enum_print(const struct nouveau_enum *, u32 value); +const struct nouveau_enum * +nouveau_enum_find(const struct nouveau_enum *, u32 value); + +int nouveau_ratelimit(void); + +#endif diff --git a/drivers/gpu/drm/nouveau/nouveau_vm.c b/drivers/gpu/drm/nouveau/nouveau_vm.c new file mode 100644 index 00000000..519a6b4b --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_vm.c @@ -0,0 +1,432 @@ +/* + * Copyright 2010 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include "drmP.h" +#include "nouveau_drv.h" +#include "nouveau_mm.h" +#include "nouveau_vm.h" + +void +nouveau_vm_map_at(struct nouveau_vma *vma, u64 delta, struct nouveau_mem *node) +{ + struct nouveau_vm *vm = vma->vm; + struct nouveau_mm_node *r; + int big = vma->node->type != vm->spg_shift; + u32 offset = vma->node->offset + (delta >> 12); + u32 bits = vma->node->type - 12; + u32 pde = (offset >> vm->pgt_bits) - vm->fpde; + u32 pte = (offset & ((1 << vm->pgt_bits) - 1)) >> bits; + u32 max = 1 << (vm->pgt_bits - bits); + u32 end, len; + + delta = 0; + list_for_each_entry(r, &node->regions, rl_entry) { + u64 phys = (u64)r->offset << 12; + u32 num = r->length >> bits; + + while (num) { + struct nouveau_gpuobj *pgt = vm->pgt[pde].obj[big]; + + end = (pte + num); + if (unlikely(end >= max)) + end = max; + len = end - pte; + + vm->map(vma, pgt, node, pte, len, phys, delta); + + num -= len; + pte += len; + if (unlikely(end >= max)) { + phys += len << (bits + 12); + pde++; + pte = 0; + } + + delta += (u64)len << vma->node->type; + } + } + + vm->flush(vm); +} + +void +nouveau_vm_map(struct nouveau_vma *vma, struct nouveau_mem *node) +{ + nouveau_vm_map_at(vma, 0, node); +} + +void +nouveau_vm_map_sg(struct nouveau_vma *vma, u64 delta, u64 length, + struct nouveau_mem *mem, dma_addr_t *list) +{ + struct nouveau_vm *vm = vma->vm; + int big = vma->node->type != vm->spg_shift; + u32 offset = vma->node->offset + (delta >> 12); + u32 bits = vma->node->type - 12; + u32 num = length >> vma->node->type; + u32 pde = (offset >> vm->pgt_bits) - vm->fpde; + u32 pte = (offset & ((1 << vm->pgt_bits) - 1)) >> bits; + u32 max = 1 << (vm->pgt_bits - bits); + u32 end, len; + + while (num) { + struct nouveau_gpuobj *pgt = vm->pgt[pde].obj[big]; + + end = (pte + num); + if (unlikely(end >= max)) + end = max; + len = end - pte; + + vm->map_sg(vma, pgt, mem, pte, len, list); + + num -= len; + pte += len; + list += len; + if (unlikely(end >= max)) { + pde++; + pte = 0; + } + } + + vm->flush(vm); +} + +void +nouveau_vm_unmap_at(struct nouveau_vma *vma, u64 delta, u64 length) +{ + struct nouveau_vm *vm = vma->vm; + int big = vma->node->type != vm->spg_shift; + u32 offset = vma->node->offset + (delta >> 12); + u32 bits = vma->node->type - 12; + u32 num = length >> vma->node->type; + u32 pde = (offset >> vm->pgt_bits) - vm->fpde; + u32 pte = (offset & ((1 << vm->pgt_bits) - 1)) >> bits; + u32 max = 1 << (vm->pgt_bits - bits); + u32 end, len; + + while (num) { + struct nouveau_gpuobj *pgt = vm->pgt[pde].obj[big]; + + end = (pte + num); + if (unlikely(end >= max)) + end = max; + len = end - pte; + + vm->unmap(pgt, pte, len); + + num -= len; + pte += len; + if (unlikely(end >= max)) { + pde++; + pte = 0; + } + } + + vm->flush(vm); +} + +void +nouveau_vm_unmap(struct nouveau_vma *vma) +{ + nouveau_vm_unmap_at(vma, 0, (u64)vma->node->length << 12); +} + +static void +nouveau_vm_unmap_pgt(struct nouveau_vm *vm, int big, u32 fpde, u32 lpde) +{ + struct nouveau_vm_pgd *vpgd; + struct nouveau_vm_pgt *vpgt; + struct nouveau_gpuobj *pgt; + u32 pde; + + for (pde = fpde; pde <= lpde; pde++) { + vpgt = &vm->pgt[pde - vm->fpde]; + if (--vpgt->refcount[big]) + continue; + + pgt = vpgt->obj[big]; + vpgt->obj[big] = NULL; + + list_for_each_entry(vpgd, &vm->pgd_list, head) { + vm->map_pgt(vpgd->obj, pde, vpgt->obj); + } + + mutex_unlock(&vm->mm->mutex); + nouveau_gpuobj_ref(NULL, &pgt); + mutex_lock(&vm->mm->mutex); + } +} + +static int +nouveau_vm_map_pgt(struct nouveau_vm *vm, u32 pde, u32 type) +{ + struct nouveau_vm_pgt *vpgt = &vm->pgt[pde - vm->fpde]; + struct nouveau_vm_pgd *vpgd; + struct nouveau_gpuobj *pgt; + int big = (type != vm->spg_shift); + u32 pgt_size; + int ret; + + pgt_size = (1 << (vm->pgt_bits + 12)) >> type; + pgt_size *= 8; + + mutex_unlock(&vm->mm->mutex); + ret = nouveau_gpuobj_new(vm->dev, NULL, pgt_size, 0x1000, + NVOBJ_FLAG_ZERO_ALLOC, &pgt); + mutex_lock(&vm->mm->mutex); + if (unlikely(ret)) + return ret; + + /* someone beat us to filling the PDE while we didn't have the lock */ + if (unlikely(vpgt->refcount[big]++)) { + mutex_unlock(&vm->mm->mutex); + nouveau_gpuobj_ref(NULL, &pgt); + mutex_lock(&vm->mm->mutex); + return 0; + } + + vpgt->obj[big] = pgt; + list_for_each_entry(vpgd, &vm->pgd_list, head) { + vm->map_pgt(vpgd->obj, pde, vpgt->obj); + } + + return 0; +} + +int +nouveau_vm_get(struct nouveau_vm *vm, u64 size, u32 page_shift, + u32 access, struct nouveau_vma *vma) +{ + u32 align = (1 << page_shift) >> 12; + u32 msize = size >> 12; + u32 fpde, lpde, pde; + int ret; + + mutex_lock(&vm->mm->mutex); + ret = nouveau_mm_get(vm->mm, page_shift, msize, 0, align, &vma->node); + if (unlikely(ret != 0)) { + mutex_unlock(&vm->mm->mutex); + return ret; + } + + fpde = (vma->node->offset >> vm->pgt_bits); + lpde = (vma->node->offset + vma->node->length - 1) >> vm->pgt_bits; + for (pde = fpde; pde <= lpde; pde++) { + struct nouveau_vm_pgt *vpgt = &vm->pgt[pde - vm->fpde]; + int big = (vma->node->type != vm->spg_shift); + + if (likely(vpgt->refcount[big])) { + vpgt->refcount[big]++; + continue; + } + + ret = nouveau_vm_map_pgt(vm, pde, vma->node->type); + if (ret) { + if (pde != fpde) + nouveau_vm_unmap_pgt(vm, big, fpde, pde - 1); + nouveau_mm_put(vm->mm, vma->node); + mutex_unlock(&vm->mm->mutex); + vma->node = NULL; + return ret; + } + } + mutex_unlock(&vm->mm->mutex); + + vma->vm = vm; + vma->offset = (u64)vma->node->offset << 12; + vma->access = access; + return 0; +} + +void +nouveau_vm_put(struct nouveau_vma *vma) +{ + struct nouveau_vm *vm = vma->vm; + u32 fpde, lpde; + + if (unlikely(vma->node == NULL)) + return; + fpde = (vma->node->offset >> vm->pgt_bits); + lpde = (vma->node->offset + vma->node->length - 1) >> vm->pgt_bits; + + mutex_lock(&vm->mm->mutex); + nouveau_vm_unmap_pgt(vm, vma->node->type != vm->spg_shift, fpde, lpde); + nouveau_mm_put(vm->mm, vma->node); + vma->node = NULL; + mutex_unlock(&vm->mm->mutex); +} + +int +nouveau_vm_new(struct drm_device *dev, u64 offset, u64 length, u64 mm_offset, + struct nouveau_vm **pvm) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_vm *vm; + u64 mm_length = (offset + length) - mm_offset; + u32 block, pgt_bits; + int ret; + + vm = kzalloc(sizeof(*vm), GFP_KERNEL); + if (!vm) + return -ENOMEM; + + if (dev_priv->card_type == NV_50) { + vm->map_pgt = nv50_vm_map_pgt; + vm->map = nv50_vm_map; + vm->map_sg = nv50_vm_map_sg; + vm->unmap = nv50_vm_unmap; + vm->flush = nv50_vm_flush; + vm->spg_shift = 12; + vm->lpg_shift = 16; + + pgt_bits = 29; + block = (1 << pgt_bits); + if (length < block) + block = length; + + } else + if (dev_priv->card_type == NV_C0) { + vm->map_pgt = nvc0_vm_map_pgt; + vm->map = nvc0_vm_map; + vm->map_sg = nvc0_vm_map_sg; + vm->unmap = nvc0_vm_unmap; + vm->flush = nvc0_vm_flush; + vm->spg_shift = 12; + vm->lpg_shift = 17; + pgt_bits = 27; + block = 4096; + } else { + kfree(vm); + return -ENOSYS; + } + + vm->fpde = offset >> pgt_bits; + vm->lpde = (offset + length - 1) >> pgt_bits; + vm->pgt = kcalloc(vm->lpde - vm->fpde + 1, sizeof(*vm->pgt), GFP_KERNEL); + if (!vm->pgt) { + kfree(vm); + return -ENOMEM; + } + + INIT_LIST_HEAD(&vm->pgd_list); + vm->dev = dev; + vm->refcount = 1; + vm->pgt_bits = pgt_bits - 12; + + ret = nouveau_mm_init(&vm->mm, mm_offset >> 12, mm_length >> 12, + block >> 12); + if (ret) { + kfree(vm); + return ret; + } + + *pvm = vm; + return 0; +} + +static int +nouveau_vm_link(struct nouveau_vm *vm, struct nouveau_gpuobj *pgd) +{ + struct nouveau_vm_pgd *vpgd; + int i; + + if (!pgd) + return 0; + + vpgd = kzalloc(sizeof(*vpgd), GFP_KERNEL); + if (!vpgd) + return -ENOMEM; + + nouveau_gpuobj_ref(pgd, &vpgd->obj); + + mutex_lock(&vm->mm->mutex); + for (i = vm->fpde; i <= vm->lpde; i++) + vm->map_pgt(pgd, i, vm->pgt[i - vm->fpde].obj); + list_add(&vpgd->head, &vm->pgd_list); + mutex_unlock(&vm->mm->mutex); + return 0; +} + +static void +nouveau_vm_unlink(struct nouveau_vm *vm, struct nouveau_gpuobj *pgd) +{ + struct nouveau_vm_pgd *vpgd, *tmp; + + if (!pgd) + return; + + mutex_lock(&vm->mm->mutex); + list_for_each_entry_safe(vpgd, tmp, &vm->pgd_list, head) { + if (vpgd->obj != pgd) + continue; + + list_del(&vpgd->head); + nouveau_gpuobj_ref(NULL, &vpgd->obj); + kfree(vpgd); + } + mutex_unlock(&vm->mm->mutex); +} + +static void +nouveau_vm_del(struct nouveau_vm *vm) +{ + struct nouveau_vm_pgd *vpgd, *tmp; + + list_for_each_entry_safe(vpgd, tmp, &vm->pgd_list, head) { + nouveau_vm_unlink(vm, vpgd->obj); + } + WARN_ON(nouveau_mm_fini(&vm->mm) != 0); + + kfree(vm->pgt); + kfree(vm); +} + +int +nouveau_vm_ref(struct nouveau_vm *ref, struct nouveau_vm **ptr, + struct nouveau_gpuobj *pgd) +{ + struct nouveau_vm *vm; + int ret; + + vm = ref; + if (vm) { + ret = nouveau_vm_link(vm, pgd); + if (ret) + return ret; + + vm->refcount++; + } + + vm = *ptr; + *ptr = ref; + + if (vm) { + nouveau_vm_unlink(vm, pgd); + + if (--vm->refcount == 0) + nouveau_vm_del(vm); + } + + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nouveau_vm.h b/drivers/gpu/drm/nouveau/nouveau_vm.h new file mode 100644 index 00000000..c48a9fc2 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_vm.h @@ -0,0 +1,113 @@ +/* + * Copyright 2010 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#ifndef __NOUVEAU_VM_H__ +#define __NOUVEAU_VM_H__ + +#include "drmP.h" + +#include "nouveau_drv.h" +#include "nouveau_mm.h" + +struct nouveau_vm_pgt { + struct nouveau_gpuobj *obj[2]; + u32 refcount[2]; +}; + +struct nouveau_vm_pgd { + struct list_head head; + struct nouveau_gpuobj *obj; +}; + +struct nouveau_vma { + struct nouveau_vm *vm; + struct nouveau_mm_node *node; + u64 offset; + u32 access; +}; + +struct nouveau_vm { + struct drm_device *dev; + struct nouveau_mm *mm; + int refcount; + + struct list_head pgd_list; + atomic_t engref[16]; + + struct nouveau_vm_pgt *pgt; + u32 fpde; + u32 lpde; + + u32 pgt_bits; + u8 spg_shift; + u8 lpg_shift; + + void (*map_pgt)(struct nouveau_gpuobj *pgd, u32 pde, + struct nouveau_gpuobj *pgt[2]); + void (*map)(struct nouveau_vma *, struct nouveau_gpuobj *, + struct nouveau_mem *, u32 pte, u32 cnt, + u64 phys, u64 delta); + void (*map_sg)(struct nouveau_vma *, struct nouveau_gpuobj *, + struct nouveau_mem *, u32 pte, u32 cnt, dma_addr_t *); + void (*unmap)(struct nouveau_gpuobj *pgt, u32 pte, u32 cnt); + void (*flush)(struct nouveau_vm *); +}; + +/* nouveau_vm.c */ +int nouveau_vm_new(struct drm_device *, u64 offset, u64 length, u64 mm_offset, + struct nouveau_vm **); +int nouveau_vm_ref(struct nouveau_vm *, struct nouveau_vm **, + struct nouveau_gpuobj *pgd); +int nouveau_vm_get(struct nouveau_vm *, u64 size, u32 page_shift, + u32 access, struct nouveau_vma *); +void nouveau_vm_put(struct nouveau_vma *); +void nouveau_vm_map(struct nouveau_vma *, struct nouveau_mem *); +void nouveau_vm_map_at(struct nouveau_vma *, u64 offset, struct nouveau_mem *); +void nouveau_vm_unmap(struct nouveau_vma *); +void nouveau_vm_unmap_at(struct nouveau_vma *, u64 offset, u64 length); +void nouveau_vm_map_sg(struct nouveau_vma *, u64 offset, u64 length, + struct nouveau_mem *, dma_addr_t *); + +/* nv50_vm.c */ +void nv50_vm_map_pgt(struct nouveau_gpuobj *pgd, u32 pde, + struct nouveau_gpuobj *pgt[2]); +void nv50_vm_map(struct nouveau_vma *, struct nouveau_gpuobj *, + struct nouveau_mem *, u32 pte, u32 cnt, u64 phys, u64 delta); +void nv50_vm_map_sg(struct nouveau_vma *, struct nouveau_gpuobj *, + struct nouveau_mem *, u32 pte, u32 cnt, dma_addr_t *); +void nv50_vm_unmap(struct nouveau_gpuobj *, u32 pte, u32 cnt); +void nv50_vm_flush(struct nouveau_vm *); +void nv50_vm_flush_engine(struct drm_device *, int engine); + +/* nvc0_vm.c */ +void nvc0_vm_map_pgt(struct nouveau_gpuobj *pgd, u32 pde, + struct nouveau_gpuobj *pgt[2]); +void nvc0_vm_map(struct nouveau_vma *, struct nouveau_gpuobj *, + struct nouveau_mem *, u32 pte, u32 cnt, u64 phys, u64 delta); +void nvc0_vm_map_sg(struct nouveau_vma *, struct nouveau_gpuobj *, + struct nouveau_mem *, u32 pte, u32 cnt, dma_addr_t *); +void nvc0_vm_unmap(struct nouveau_gpuobj *, u32 pte, u32 cnt); +void nvc0_vm_flush(struct nouveau_vm *); + +#endif diff --git a/drivers/gpu/drm/nouveau/nouveau_volt.c b/drivers/gpu/drm/nouveau/nouveau_volt.c new file mode 100644 index 00000000..75e87274 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nouveau_volt.c @@ -0,0 +1,220 @@ +/* + * Copyright 2010 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include "drmP.h" + +#include "nouveau_drv.h" +#include "nouveau_pm.h" + +static const enum dcb_gpio_tag vidtag[] = { 0x04, 0x05, 0x06, 0x1a }; +static int nr_vidtag = sizeof(vidtag) / sizeof(vidtag[0]); + +int +nouveau_voltage_gpio_get(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpio_engine *gpio = &dev_priv->engine.gpio; + struct nouveau_pm_voltage *volt = &dev_priv->engine.pm.voltage; + u8 vid = 0; + int i; + + for (i = 0; i < nr_vidtag; i++) { + if (!(volt->vid_mask & (1 << i))) + continue; + + vid |= gpio->get(dev, vidtag[i]) << i; + } + + return nouveau_volt_lvl_lookup(dev, vid); +} + +int +nouveau_voltage_gpio_set(struct drm_device *dev, int voltage) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpio_engine *gpio = &dev_priv->engine.gpio; + struct nouveau_pm_voltage *volt = &dev_priv->engine.pm.voltage; + int vid, i; + + vid = nouveau_volt_vid_lookup(dev, voltage); + if (vid < 0) + return vid; + + for (i = 0; i < nr_vidtag; i++) { + if (!(volt->vid_mask & (1 << i))) + continue; + + gpio->set(dev, vidtag[i], !!(vid & (1 << i))); + } + + return 0; +} + +int +nouveau_volt_vid_lookup(struct drm_device *dev, int voltage) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_pm_voltage *volt = &dev_priv->engine.pm.voltage; + int i; + + for (i = 0; i < volt->nr_level; i++) { + if (volt->level[i].voltage == voltage) + return volt->level[i].vid; + } + + return -ENOENT; +} + +int +nouveau_volt_lvl_lookup(struct drm_device *dev, int vid) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_pm_voltage *volt = &dev_priv->engine.pm.voltage; + int i; + + for (i = 0; i < volt->nr_level; i++) { + if (volt->level[i].vid == vid) + return volt->level[i].voltage; + } + + return -ENOENT; +} + +void +nouveau_volt_init(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_pm_engine *pm = &dev_priv->engine.pm; + struct nouveau_pm_voltage *voltage = &pm->voltage; + struct nvbios *bios = &dev_priv->vbios; + struct bit_entry P; + u8 *volt = NULL, *entry; + int i, headerlen, recordlen, entries, vidmask, vidshift; + + if (bios->type == NVBIOS_BIT) { + if (bit_table(dev, 'P', &P)) + return; + + if (P.version == 1) + volt = ROMPTR(bios, P.data[16]); + else + if (P.version == 2) + volt = ROMPTR(bios, P.data[12]); + else { + NV_WARN(dev, "unknown volt for BIT P %d\n", P.version); + } + } else { + if (bios->data[bios->offset + 6] < 0x27) { + NV_DEBUG(dev, "BMP version too old for voltage\n"); + return; + } + + volt = ROMPTR(bios, bios->data[bios->offset + 0x98]); + } + + if (!volt) { + NV_DEBUG(dev, "voltage table pointer invalid\n"); + return; + } + + switch (volt[0]) { + case 0x10: + case 0x11: + case 0x12: + headerlen = 5; + recordlen = volt[1]; + entries = volt[2]; + vidshift = 0; + vidmask = volt[4]; + break; + case 0x20: + headerlen = volt[1]; + recordlen = volt[3]; + entries = volt[2]; + vidshift = 0; /* could be vidshift like 0x30? */ + vidmask = volt[5]; + break; + case 0x30: + headerlen = volt[1]; + recordlen = volt[2]; + entries = volt[3]; + vidmask = volt[4]; + /* no longer certain what volt[5] is, if it's related to + * the vid shift then it's definitely not a function of + * how many bits are set. + * + * after looking at a number of nva3+ vbios images, they + * all seem likely to have a static shift of 2.. lets + * go with that for now until proven otherwise. + */ + vidshift = 2; + break; + default: + NV_WARN(dev, "voltage table 0x%02x unknown\n", volt[0]); + return; + } + + /* validate vid mask */ + voltage->vid_mask = vidmask; + if (!voltage->vid_mask) + return; + + i = 0; + while (vidmask) { + if (i > nr_vidtag) { + NV_DEBUG(dev, "vid bit %d unknown\n", i); + return; + } + + if (!nouveau_bios_gpio_entry(dev, vidtag[i])) { + NV_DEBUG(dev, "vid bit %d has no gpio tag\n", i); + return; + } + + vidmask >>= 1; + i++; + } + + /* parse vbios entries into common format */ + voltage->level = kcalloc(entries, sizeof(*voltage->level), GFP_KERNEL); + if (!voltage->level) + return; + + entry = volt + headerlen; + for (i = 0; i < entries; i++, entry += recordlen) { + voltage->level[i].voltage = entry[0]; + voltage->level[i].vid = entry[1] >> vidshift; + } + voltage->nr_level = entries; + voltage->supported = true; +} + +void +nouveau_volt_fini(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_pm_voltage *volt = &dev_priv->engine.pm.voltage; + + kfree(volt->level); +} diff --git a/drivers/gpu/drm/nouveau/nv04_crtc.c b/drivers/gpu/drm/nouveau/nv04_crtc.c new file mode 100644 index 00000000..f1a3ae49 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv04_crtc.c @@ -0,0 +1,1052 @@ +/* + * Copyright 1993-2003 NVIDIA, Corporation + * Copyright 2006 Dave Airlie + * Copyright 2007 Maarten Maathuis + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + */ + +#include "drmP.h" +#include "drm_crtc_helper.h" + +#include "nouveau_drv.h" +#include "nouveau_encoder.h" +#include "nouveau_connector.h" +#include "nouveau_crtc.h" +#include "nouveau_fb.h" +#include "nouveau_hw.h" +#include "nvreg.h" +#include "nouveau_fbcon.h" + +static int +nv04_crtc_mode_set_base(struct drm_crtc *crtc, int x, int y, + struct drm_framebuffer *old_fb); + +static void +crtc_wr_cio_state(struct drm_crtc *crtc, struct nv04_crtc_reg *crtcstate, int index) +{ + NVWriteVgaCrtc(crtc->dev, nouveau_crtc(crtc)->index, index, + crtcstate->CRTC[index]); +} + +static void nv_crtc_set_digital_vibrance(struct drm_crtc *crtc, int level) +{ + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + struct drm_nouveau_private *dev_priv = crtc->dev->dev_private; + struct nv04_crtc_reg *regp = &dev_priv->mode_reg.crtc_reg[nv_crtc->index]; + + regp->CRTC[NV_CIO_CRE_CSB] = nv_crtc->saturation = level; + if (nv_crtc->saturation && nv_gf4_disp_arch(crtc->dev)) { + regp->CRTC[NV_CIO_CRE_CSB] = 0x80; + regp->CRTC[NV_CIO_CRE_5B] = nv_crtc->saturation << 2; + crtc_wr_cio_state(crtc, regp, NV_CIO_CRE_5B); + } + crtc_wr_cio_state(crtc, regp, NV_CIO_CRE_CSB); +} + +static void nv_crtc_set_image_sharpening(struct drm_crtc *crtc, int level) +{ + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + struct drm_nouveau_private *dev_priv = crtc->dev->dev_private; + struct nv04_crtc_reg *regp = &dev_priv->mode_reg.crtc_reg[nv_crtc->index]; + + nv_crtc->sharpness = level; + if (level < 0) /* blur is in hw range 0x3f -> 0x20 */ + level += 0x40; + regp->ramdac_634 = level; + NVWriteRAMDAC(crtc->dev, nv_crtc->index, NV_PRAMDAC_634, regp->ramdac_634); +} + +#define PLLSEL_VPLL1_MASK \ + (NV_PRAMDAC_PLL_COEFF_SELECT_SOURCE_PROG_VPLL \ + | NV_PRAMDAC_PLL_COEFF_SELECT_VCLK_RATIO_DB2) +#define PLLSEL_VPLL2_MASK \ + (NV_PRAMDAC_PLL_COEFF_SELECT_PLL_SOURCE_VPLL2 \ + | NV_PRAMDAC_PLL_COEFF_SELECT_VCLK2_RATIO_DB2) +#define PLLSEL_TV_MASK \ + (NV_PRAMDAC_PLL_COEFF_SELECT_TV_VSCLK1 \ + | NV_PRAMDAC_PLL_COEFF_SELECT_TV_PCLK1 \ + | NV_PRAMDAC_PLL_COEFF_SELECT_TV_VSCLK2 \ + | NV_PRAMDAC_PLL_COEFF_SELECT_TV_PCLK2) + +/* NV4x 0x40.. pll notes: + * gpu pll: 0x4000 + 0x4004 + * ?gpu? pll: 0x4008 + 0x400c + * vpll1: 0x4010 + 0x4014 + * vpll2: 0x4018 + 0x401c + * mpll: 0x4020 + 0x4024 + * mpll: 0x4038 + 0x403c + * + * the first register of each pair has some unknown details: + * bits 0-7: redirected values from elsewhere? (similar to PLL_SETUP_CONTROL?) + * bits 20-23: (mpll) something to do with post divider? + * bits 28-31: related to single stage mode? (bit 8/12) + */ + +static void nv_crtc_calc_state_ext(struct drm_crtc *crtc, struct drm_display_mode * mode, int dot_clock) +{ + struct drm_device *dev = crtc->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + struct nv04_mode_state *state = &dev_priv->mode_reg; + struct nv04_crtc_reg *regp = &state->crtc_reg[nv_crtc->index]; + struct nouveau_pll_vals *pv = ®p->pllvals; + struct pll_lims pll_lim; + + if (get_pll_limits(dev, nv_crtc->index ? PLL_VPLL1 : PLL_VPLL0, &pll_lim)) + return; + + /* NM2 == 0 is used to determine single stage mode on two stage plls */ + pv->NM2 = 0; + + /* for newer nv4x the blob uses only the first stage of the vpll below a + * certain clock. for a certain nv4b this is 150MHz. since the max + * output frequency of the first stage for this card is 300MHz, it is + * assumed the threshold is given by vco1 maxfreq/2 + */ + /* for early nv4x, specifically nv40 and *some* nv43 (devids 0 and 6, + * not 8, others unknown), the blob always uses both plls. no problem + * has yet been observed in allowing the use a single stage pll on all + * nv43 however. the behaviour of single stage use is untested on nv40 + */ + if (dev_priv->chipset > 0x40 && dot_clock <= (pll_lim.vco1.maxfreq / 2)) + memset(&pll_lim.vco2, 0, sizeof(pll_lim.vco2)); + + if (!nouveau_calc_pll_mnp(dev, &pll_lim, dot_clock, pv)) + return; + + state->pllsel &= PLLSEL_VPLL1_MASK | PLLSEL_VPLL2_MASK | PLLSEL_TV_MASK; + + /* The blob uses this always, so let's do the same */ + if (dev_priv->card_type == NV_40) + state->pllsel |= NV_PRAMDAC_PLL_COEFF_SELECT_USE_VPLL2_TRUE; + /* again nv40 and some nv43 act more like nv3x as described above */ + if (dev_priv->chipset < 0x41) + state->pllsel |= NV_PRAMDAC_PLL_COEFF_SELECT_SOURCE_PROG_MPLL | + NV_PRAMDAC_PLL_COEFF_SELECT_SOURCE_PROG_NVPLL; + state->pllsel |= nv_crtc->index ? PLLSEL_VPLL2_MASK : PLLSEL_VPLL1_MASK; + + if (pv->NM2) + NV_DEBUG_KMS(dev, "vpll: n1 %d n2 %d m1 %d m2 %d log2p %d\n", + pv->N1, pv->N2, pv->M1, pv->M2, pv->log2P); + else + NV_DEBUG_KMS(dev, "vpll: n %d m %d log2p %d\n", + pv->N1, pv->M1, pv->log2P); + + nv_crtc->cursor.set_offset(nv_crtc, nv_crtc->cursor.offset); +} + +static void +nv_crtc_dpms(struct drm_crtc *crtc, int mode) +{ + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + struct drm_device *dev = crtc->dev; + unsigned char seq1 = 0, crtc17 = 0; + unsigned char crtc1A; + + NV_DEBUG_KMS(dev, "Setting dpms mode %d on CRTC %d\n", mode, + nv_crtc->index); + + if (nv_crtc->last_dpms == mode) /* Don't do unnecessary mode changes. */ + return; + + nv_crtc->last_dpms = mode; + + if (nv_two_heads(dev)) + NVSetOwner(dev, nv_crtc->index); + + /* nv4ref indicates these two RPC1 bits inhibit h/v sync */ + crtc1A = NVReadVgaCrtc(dev, nv_crtc->index, + NV_CIO_CRE_RPC1_INDEX) & ~0xC0; + switch (mode) { + case DRM_MODE_DPMS_STANDBY: + /* Screen: Off; HSync: Off, VSync: On -- Not Supported */ + seq1 = 0x20; + crtc17 = 0x80; + crtc1A |= 0x80; + break; + case DRM_MODE_DPMS_SUSPEND: + /* Screen: Off; HSync: On, VSync: Off -- Not Supported */ + seq1 = 0x20; + crtc17 = 0x80; + crtc1A |= 0x40; + break; + case DRM_MODE_DPMS_OFF: + /* Screen: Off; HSync: Off, VSync: Off */ + seq1 = 0x20; + crtc17 = 0x00; + crtc1A |= 0xC0; + break; + case DRM_MODE_DPMS_ON: + default: + /* Screen: On; HSync: On, VSync: On */ + seq1 = 0x00; + crtc17 = 0x80; + break; + } + + NVVgaSeqReset(dev, nv_crtc->index, true); + /* Each head has it's own sequencer, so we can turn it off when we want */ + seq1 |= (NVReadVgaSeq(dev, nv_crtc->index, NV_VIO_SR_CLOCK_INDEX) & ~0x20); + NVWriteVgaSeq(dev, nv_crtc->index, NV_VIO_SR_CLOCK_INDEX, seq1); + crtc17 |= (NVReadVgaCrtc(dev, nv_crtc->index, NV_CIO_CR_MODE_INDEX) & ~0x80); + mdelay(10); + NVWriteVgaCrtc(dev, nv_crtc->index, NV_CIO_CR_MODE_INDEX, crtc17); + NVVgaSeqReset(dev, nv_crtc->index, false); + + NVWriteVgaCrtc(dev, nv_crtc->index, NV_CIO_CRE_RPC1_INDEX, crtc1A); +} + +static bool +nv_crtc_mode_fixup(struct drm_crtc *crtc, struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + return true; +} + +static void +nv_crtc_mode_set_vga(struct drm_crtc *crtc, struct drm_display_mode *mode) +{ + struct drm_device *dev = crtc->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + struct nv04_crtc_reg *regp = &dev_priv->mode_reg.crtc_reg[nv_crtc->index]; + struct drm_framebuffer *fb = crtc->fb; + + /* Calculate our timings */ + int horizDisplay = (mode->crtc_hdisplay >> 3) - 1; + int horizStart = (mode->crtc_hsync_start >> 3) + 1; + int horizEnd = (mode->crtc_hsync_end >> 3) + 1; + int horizTotal = (mode->crtc_htotal >> 3) - 5; + int horizBlankStart = (mode->crtc_hdisplay >> 3) - 1; + int horizBlankEnd = (mode->crtc_htotal >> 3) - 1; + int vertDisplay = mode->crtc_vdisplay - 1; + int vertStart = mode->crtc_vsync_start - 1; + int vertEnd = mode->crtc_vsync_end - 1; + int vertTotal = mode->crtc_vtotal - 2; + int vertBlankStart = mode->crtc_vdisplay - 1; + int vertBlankEnd = mode->crtc_vtotal - 1; + + struct drm_encoder *encoder; + bool fp_output = false; + + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + + if (encoder->crtc == crtc && + (nv_encoder->dcb->type == OUTPUT_LVDS || + nv_encoder->dcb->type == OUTPUT_TMDS)) + fp_output = true; + } + + if (fp_output) { + vertStart = vertTotal - 3; + vertEnd = vertTotal - 2; + vertBlankStart = vertStart; + horizStart = horizTotal - 5; + horizEnd = horizTotal - 2; + horizBlankEnd = horizTotal + 4; +#if 0 + if (dev->overlayAdaptor && dev_priv->card_type >= NV_10) + /* This reportedly works around some video overlay bandwidth problems */ + horizTotal += 2; +#endif + } + + if (mode->flags & DRM_MODE_FLAG_INTERLACE) + vertTotal |= 1; + +#if 0 + ErrorF("horizDisplay: 0x%X \n", horizDisplay); + ErrorF("horizStart: 0x%X \n", horizStart); + ErrorF("horizEnd: 0x%X \n", horizEnd); + ErrorF("horizTotal: 0x%X \n", horizTotal); + ErrorF("horizBlankStart: 0x%X \n", horizBlankStart); + ErrorF("horizBlankEnd: 0x%X \n", horizBlankEnd); + ErrorF("vertDisplay: 0x%X \n", vertDisplay); + ErrorF("vertStart: 0x%X \n", vertStart); + ErrorF("vertEnd: 0x%X \n", vertEnd); + ErrorF("vertTotal: 0x%X \n", vertTotal); + ErrorF("vertBlankStart: 0x%X \n", vertBlankStart); + ErrorF("vertBlankEnd: 0x%X \n", vertBlankEnd); +#endif + + /* + * compute correct Hsync & Vsync polarity + */ + if ((mode->flags & (DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_NHSYNC)) + && (mode->flags & (DRM_MODE_FLAG_PVSYNC | DRM_MODE_FLAG_NVSYNC))) { + + regp->MiscOutReg = 0x23; + if (mode->flags & DRM_MODE_FLAG_NHSYNC) + regp->MiscOutReg |= 0x40; + if (mode->flags & DRM_MODE_FLAG_NVSYNC) + regp->MiscOutReg |= 0x80; + } else { + int vdisplay = mode->vdisplay; + if (mode->flags & DRM_MODE_FLAG_DBLSCAN) + vdisplay *= 2; + if (mode->vscan > 1) + vdisplay *= mode->vscan; + if (vdisplay < 400) + regp->MiscOutReg = 0xA3; /* +hsync -vsync */ + else if (vdisplay < 480) + regp->MiscOutReg = 0x63; /* -hsync +vsync */ + else if (vdisplay < 768) + regp->MiscOutReg = 0xE3; /* -hsync -vsync */ + else + regp->MiscOutReg = 0x23; /* +hsync +vsync */ + } + + regp->MiscOutReg |= (mode->clock_index & 0x03) << 2; + + /* + * Time Sequencer + */ + regp->Sequencer[NV_VIO_SR_RESET_INDEX] = 0x00; + /* 0x20 disables the sequencer */ + if (mode->flags & DRM_MODE_FLAG_CLKDIV2) + regp->Sequencer[NV_VIO_SR_CLOCK_INDEX] = 0x29; + else + regp->Sequencer[NV_VIO_SR_CLOCK_INDEX] = 0x21; + regp->Sequencer[NV_VIO_SR_PLANE_MASK_INDEX] = 0x0F; + regp->Sequencer[NV_VIO_SR_CHAR_MAP_INDEX] = 0x00; + regp->Sequencer[NV_VIO_SR_MEM_MODE_INDEX] = 0x0E; + + /* + * CRTC + */ + regp->CRTC[NV_CIO_CR_HDT_INDEX] = horizTotal; + regp->CRTC[NV_CIO_CR_HDE_INDEX] = horizDisplay; + regp->CRTC[NV_CIO_CR_HBS_INDEX] = horizBlankStart; + regp->CRTC[NV_CIO_CR_HBE_INDEX] = (1 << 7) | + XLATE(horizBlankEnd, 0, NV_CIO_CR_HBE_4_0); + regp->CRTC[NV_CIO_CR_HRS_INDEX] = horizStart; + regp->CRTC[NV_CIO_CR_HRE_INDEX] = XLATE(horizBlankEnd, 5, NV_CIO_CR_HRE_HBE_5) | + XLATE(horizEnd, 0, NV_CIO_CR_HRE_4_0); + regp->CRTC[NV_CIO_CR_VDT_INDEX] = vertTotal; + regp->CRTC[NV_CIO_CR_OVL_INDEX] = XLATE(vertStart, 9, NV_CIO_CR_OVL_VRS_9) | + XLATE(vertDisplay, 9, NV_CIO_CR_OVL_VDE_9) | + XLATE(vertTotal, 9, NV_CIO_CR_OVL_VDT_9) | + (1 << 4) | + XLATE(vertBlankStart, 8, NV_CIO_CR_OVL_VBS_8) | + XLATE(vertStart, 8, NV_CIO_CR_OVL_VRS_8) | + XLATE(vertDisplay, 8, NV_CIO_CR_OVL_VDE_8) | + XLATE(vertTotal, 8, NV_CIO_CR_OVL_VDT_8); + regp->CRTC[NV_CIO_CR_RSAL_INDEX] = 0x00; + regp->CRTC[NV_CIO_CR_CELL_HT_INDEX] = ((mode->flags & DRM_MODE_FLAG_DBLSCAN) ? MASK(NV_CIO_CR_CELL_HT_SCANDBL) : 0) | + 1 << 6 | + XLATE(vertBlankStart, 9, NV_CIO_CR_CELL_HT_VBS_9); + regp->CRTC[NV_CIO_CR_CURS_ST_INDEX] = 0x00; + regp->CRTC[NV_CIO_CR_CURS_END_INDEX] = 0x00; + regp->CRTC[NV_CIO_CR_SA_HI_INDEX] = 0x00; + regp->CRTC[NV_CIO_CR_SA_LO_INDEX] = 0x00; + regp->CRTC[NV_CIO_CR_TCOFF_HI_INDEX] = 0x00; + regp->CRTC[NV_CIO_CR_TCOFF_LO_INDEX] = 0x00; + regp->CRTC[NV_CIO_CR_VRS_INDEX] = vertStart; + regp->CRTC[NV_CIO_CR_VRE_INDEX] = 1 << 5 | XLATE(vertEnd, 0, NV_CIO_CR_VRE_3_0); + regp->CRTC[NV_CIO_CR_VDE_INDEX] = vertDisplay; + /* framebuffer can be larger than crtc scanout area. */ + regp->CRTC[NV_CIO_CR_OFFSET_INDEX] = fb->pitch / 8; + regp->CRTC[NV_CIO_CR_ULINE_INDEX] = 0x00; + regp->CRTC[NV_CIO_CR_VBS_INDEX] = vertBlankStart; + regp->CRTC[NV_CIO_CR_VBE_INDEX] = vertBlankEnd; + regp->CRTC[NV_CIO_CR_MODE_INDEX] = 0x43; + regp->CRTC[NV_CIO_CR_LCOMP_INDEX] = 0xff; + + /* + * Some extended CRTC registers (they are not saved with the rest of the vga regs). + */ + + /* framebuffer can be larger than crtc scanout area. */ + regp->CRTC[NV_CIO_CRE_RPC0_INDEX] = + XLATE(fb->pitch / 8, 8, NV_CIO_CRE_RPC0_OFFSET_10_8); + regp->CRTC[NV_CIO_CRE_42] = + XLATE(fb->pitch / 8, 11, NV_CIO_CRE_42_OFFSET_11); + regp->CRTC[NV_CIO_CRE_RPC1_INDEX] = mode->crtc_hdisplay < 1280 ? + MASK(NV_CIO_CRE_RPC1_LARGE) : 0x00; + regp->CRTC[NV_CIO_CRE_LSR_INDEX] = XLATE(horizBlankEnd, 6, NV_CIO_CRE_LSR_HBE_6) | + XLATE(vertBlankStart, 10, NV_CIO_CRE_LSR_VBS_10) | + XLATE(vertStart, 10, NV_CIO_CRE_LSR_VRS_10) | + XLATE(vertDisplay, 10, NV_CIO_CRE_LSR_VDE_10) | + XLATE(vertTotal, 10, NV_CIO_CRE_LSR_VDT_10); + regp->CRTC[NV_CIO_CRE_HEB__INDEX] = XLATE(horizStart, 8, NV_CIO_CRE_HEB_HRS_8) | + XLATE(horizBlankStart, 8, NV_CIO_CRE_HEB_HBS_8) | + XLATE(horizDisplay, 8, NV_CIO_CRE_HEB_HDE_8) | + XLATE(horizTotal, 8, NV_CIO_CRE_HEB_HDT_8); + regp->CRTC[NV_CIO_CRE_EBR_INDEX] = XLATE(vertBlankStart, 11, NV_CIO_CRE_EBR_VBS_11) | + XLATE(vertStart, 11, NV_CIO_CRE_EBR_VRS_11) | + XLATE(vertDisplay, 11, NV_CIO_CRE_EBR_VDE_11) | + XLATE(vertTotal, 11, NV_CIO_CRE_EBR_VDT_11); + + if (mode->flags & DRM_MODE_FLAG_INTERLACE) { + horizTotal = (horizTotal >> 1) & ~1; + regp->CRTC[NV_CIO_CRE_ILACE__INDEX] = horizTotal; + regp->CRTC[NV_CIO_CRE_HEB__INDEX] |= XLATE(horizTotal, 8, NV_CIO_CRE_HEB_ILC_8); + } else + regp->CRTC[NV_CIO_CRE_ILACE__INDEX] = 0xff; /* interlace off */ + + /* + * Graphics Display Controller + */ + regp->Graphics[NV_VIO_GX_SR_INDEX] = 0x00; + regp->Graphics[NV_VIO_GX_SREN_INDEX] = 0x00; + regp->Graphics[NV_VIO_GX_CCOMP_INDEX] = 0x00; + regp->Graphics[NV_VIO_GX_ROP_INDEX] = 0x00; + regp->Graphics[NV_VIO_GX_READ_MAP_INDEX] = 0x00; + regp->Graphics[NV_VIO_GX_MODE_INDEX] = 0x40; /* 256 color mode */ + regp->Graphics[NV_VIO_GX_MISC_INDEX] = 0x05; /* map 64k mem + graphic mode */ + regp->Graphics[NV_VIO_GX_DONT_CARE_INDEX] = 0x0F; + regp->Graphics[NV_VIO_GX_BIT_MASK_INDEX] = 0xFF; + + regp->Attribute[0] = 0x00; /* standard colormap translation */ + regp->Attribute[1] = 0x01; + regp->Attribute[2] = 0x02; + regp->Attribute[3] = 0x03; + regp->Attribute[4] = 0x04; + regp->Attribute[5] = 0x05; + regp->Attribute[6] = 0x06; + regp->Attribute[7] = 0x07; + regp->Attribute[8] = 0x08; + regp->Attribute[9] = 0x09; + regp->Attribute[10] = 0x0A; + regp->Attribute[11] = 0x0B; + regp->Attribute[12] = 0x0C; + regp->Attribute[13] = 0x0D; + regp->Attribute[14] = 0x0E; + regp->Attribute[15] = 0x0F; + regp->Attribute[NV_CIO_AR_MODE_INDEX] = 0x01; /* Enable graphic mode */ + /* Non-vga */ + regp->Attribute[NV_CIO_AR_OSCAN_INDEX] = 0x00; + regp->Attribute[NV_CIO_AR_PLANE_INDEX] = 0x0F; /* enable all color planes */ + regp->Attribute[NV_CIO_AR_HPP_INDEX] = 0x00; + regp->Attribute[NV_CIO_AR_CSEL_INDEX] = 0x00; +} + +/** + * Sets up registers for the given mode/adjusted_mode pair. + * + * The clocks, CRTCs and outputs attached to this CRTC must be off. + * + * This shouldn't enable any clocks, CRTCs, or outputs, but they should + * be easily turned on/off after this. + */ +static void +nv_crtc_mode_set_regs(struct drm_crtc *crtc, struct drm_display_mode * mode) +{ + struct drm_device *dev = crtc->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + struct nv04_crtc_reg *regp = &dev_priv->mode_reg.crtc_reg[nv_crtc->index]; + struct nv04_crtc_reg *savep = &dev_priv->saved_reg.crtc_reg[nv_crtc->index]; + struct drm_encoder *encoder; + bool lvds_output = false, tmds_output = false, tv_output = false, + off_chip_digital = false; + + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + bool digital = false; + + if (encoder->crtc != crtc) + continue; + + if (nv_encoder->dcb->type == OUTPUT_LVDS) + digital = lvds_output = true; + if (nv_encoder->dcb->type == OUTPUT_TV) + tv_output = true; + if (nv_encoder->dcb->type == OUTPUT_TMDS) + digital = tmds_output = true; + if (nv_encoder->dcb->location != DCB_LOC_ON_CHIP && digital) + off_chip_digital = true; + } + + /* Registers not directly related to the (s)vga mode */ + + /* What is the meaning of this register? */ + /* A few popular values are 0x18, 0x1c, 0x38, 0x3c */ + regp->CRTC[NV_CIO_CRE_ENH_INDEX] = savep->CRTC[NV_CIO_CRE_ENH_INDEX] & ~(1<<5); + + regp->crtc_eng_ctrl = 0; + /* Except for rare conditions I2C is enabled on the primary crtc */ + if (nv_crtc->index == 0) + regp->crtc_eng_ctrl |= NV_CRTC_FSEL_I2C; +#if 0 + /* Set overlay to desired crtc. */ + if (dev->overlayAdaptor) { + NVPortPrivPtr pPriv = GET_OVERLAY_PRIVATE(dev); + if (pPriv->overlayCRTC == nv_crtc->index) + regp->crtc_eng_ctrl |= NV_CRTC_FSEL_OVERLAY; + } +#endif + + /* ADDRESS_SPACE_PNVM is the same as setting HCUR_ASI */ + regp->cursor_cfg = NV_PCRTC_CURSOR_CONFIG_CUR_LINES_64 | + NV_PCRTC_CURSOR_CONFIG_CUR_PIXELS_64 | + NV_PCRTC_CURSOR_CONFIG_ADDRESS_SPACE_PNVM; + if (dev_priv->chipset >= 0x11) + regp->cursor_cfg |= NV_PCRTC_CURSOR_CONFIG_CUR_BPP_32; + if (mode->flags & DRM_MODE_FLAG_DBLSCAN) + regp->cursor_cfg |= NV_PCRTC_CURSOR_CONFIG_DOUBLE_SCAN_ENABLE; + + /* Unblock some timings */ + regp->CRTC[NV_CIO_CRE_53] = 0; + regp->CRTC[NV_CIO_CRE_54] = 0; + + /* 0x00 is disabled, 0x11 is lvds, 0x22 crt and 0x88 tmds */ + if (lvds_output) + regp->CRTC[NV_CIO_CRE_SCRATCH3__INDEX] = 0x11; + else if (tmds_output) + regp->CRTC[NV_CIO_CRE_SCRATCH3__INDEX] = 0x88; + else + regp->CRTC[NV_CIO_CRE_SCRATCH3__INDEX] = 0x22; + + /* These values seem to vary */ + /* This register seems to be used by the bios to make certain decisions on some G70 cards? */ + regp->CRTC[NV_CIO_CRE_SCRATCH4__INDEX] = savep->CRTC[NV_CIO_CRE_SCRATCH4__INDEX]; + + nv_crtc_set_digital_vibrance(crtc, nv_crtc->saturation); + + /* probably a scratch reg, but kept for cargo-cult purposes: + * bit0: crtc0?, head A + * bit6: lvds, head A + * bit7: (only in X), head A + */ + if (nv_crtc->index == 0) + regp->CRTC[NV_CIO_CRE_4B] = savep->CRTC[NV_CIO_CRE_4B] | 0x80; + + /* The blob seems to take the current value from crtc 0, add 4 to that + * and reuse the old value for crtc 1 */ + regp->CRTC[NV_CIO_CRE_TVOUT_LATENCY] = dev_priv->saved_reg.crtc_reg[0].CRTC[NV_CIO_CRE_TVOUT_LATENCY]; + if (!nv_crtc->index) + regp->CRTC[NV_CIO_CRE_TVOUT_LATENCY] += 4; + + /* the blob sometimes sets |= 0x10 (which is the same as setting |= + * 1 << 30 on 0x60.830), for no apparent reason */ + regp->CRTC[NV_CIO_CRE_59] = off_chip_digital; + + if (dev_priv->card_type >= NV_30) + regp->CRTC[0x9f] = off_chip_digital ? 0x11 : 0x1; + + regp->crtc_830 = mode->crtc_vdisplay - 3; + regp->crtc_834 = mode->crtc_vdisplay - 1; + + if (dev_priv->card_type == NV_40) + /* This is what the blob does */ + regp->crtc_850 = NVReadCRTC(dev, 0, NV_PCRTC_850); + + if (dev_priv->card_type >= NV_30) + regp->gpio_ext = NVReadCRTC(dev, 0, NV_PCRTC_GPIO_EXT); + + if (dev_priv->card_type >= NV_10) + regp->crtc_cfg = NV10_PCRTC_CONFIG_START_ADDRESS_HSYNC; + else + regp->crtc_cfg = NV04_PCRTC_CONFIG_START_ADDRESS_HSYNC; + + /* Some misc regs */ + if (dev_priv->card_type == NV_40) { + regp->CRTC[NV_CIO_CRE_85] = 0xFF; + regp->CRTC[NV_CIO_CRE_86] = 0x1; + } + + regp->CRTC[NV_CIO_CRE_PIXEL_INDEX] = (crtc->fb->depth + 1) / 8; + /* Enable slaved mode (called MODE_TV in nv4ref.h) */ + if (lvds_output || tmds_output || tv_output) + regp->CRTC[NV_CIO_CRE_PIXEL_INDEX] |= (1 << 7); + + /* Generic PRAMDAC regs */ + + if (dev_priv->card_type >= NV_10) + /* Only bit that bios and blob set. */ + regp->nv10_cursync = (1 << 25); + + regp->ramdac_gen_ctrl = NV_PRAMDAC_GENERAL_CONTROL_BPC_8BITS | + NV_PRAMDAC_GENERAL_CONTROL_VGA_STATE_SEL | + NV_PRAMDAC_GENERAL_CONTROL_PIXMIX_ON; + if (crtc->fb->depth == 16) + regp->ramdac_gen_ctrl |= NV_PRAMDAC_GENERAL_CONTROL_ALT_MODE_SEL; + if (dev_priv->chipset >= 0x11) + regp->ramdac_gen_ctrl |= NV_PRAMDAC_GENERAL_CONTROL_PIPE_LONG; + + regp->ramdac_630 = 0; /* turn off green mode (tv test pattern?) */ + regp->tv_setup = 0; + + nv_crtc_set_image_sharpening(crtc, nv_crtc->sharpness); + + /* Some values the blob sets */ + regp->ramdac_8c0 = 0x100; + regp->ramdac_a20 = 0x0; + regp->ramdac_a24 = 0xfffff; + regp->ramdac_a34 = 0x1; +} + +/** + * Sets up registers for the given mode/adjusted_mode pair. + * + * The clocks, CRTCs and outputs attached to this CRTC must be off. + * + * This shouldn't enable any clocks, CRTCs, or outputs, but they should + * be easily turned on/off after this. + */ +static int +nv_crtc_mode_set(struct drm_crtc *crtc, struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode, + int x, int y, struct drm_framebuffer *old_fb) +{ + struct drm_device *dev = crtc->dev; + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + struct drm_nouveau_private *dev_priv = dev->dev_private; + + NV_DEBUG_KMS(dev, "CTRC mode on CRTC %d:\n", nv_crtc->index); + drm_mode_debug_printmodeline(adjusted_mode); + + /* unlock must come after turning off FP_TG_CONTROL in output_prepare */ + nv_lock_vga_crtc_shadow(dev, nv_crtc->index, -1); + + nv_crtc_mode_set_vga(crtc, adjusted_mode); + /* calculated in nv04_dfp_prepare, nv40 needs it written before calculating PLLs */ + if (dev_priv->card_type == NV_40) + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_SEL_CLK, dev_priv->mode_reg.sel_clk); + nv_crtc_mode_set_regs(crtc, adjusted_mode); + nv_crtc_calc_state_ext(crtc, mode, adjusted_mode->clock); + return 0; +} + +static void nv_crtc_save(struct drm_crtc *crtc) +{ + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + struct drm_nouveau_private *dev_priv = crtc->dev->dev_private; + struct nv04_mode_state *state = &dev_priv->mode_reg; + struct nv04_crtc_reg *crtc_state = &state->crtc_reg[nv_crtc->index]; + struct nv04_mode_state *saved = &dev_priv->saved_reg; + struct nv04_crtc_reg *crtc_saved = &saved->crtc_reg[nv_crtc->index]; + + if (nv_two_heads(crtc->dev)) + NVSetOwner(crtc->dev, nv_crtc->index); + + nouveau_hw_save_state(crtc->dev, nv_crtc->index, saved); + + /* init some state to saved value */ + state->sel_clk = saved->sel_clk & ~(0x5 << 16); + crtc_state->CRTC[NV_CIO_CRE_LCD__INDEX] = crtc_saved->CRTC[NV_CIO_CRE_LCD__INDEX]; + state->pllsel = saved->pllsel & ~(PLLSEL_VPLL1_MASK | PLLSEL_VPLL2_MASK | PLLSEL_TV_MASK); + crtc_state->gpio_ext = crtc_saved->gpio_ext; +} + +static void nv_crtc_restore(struct drm_crtc *crtc) +{ + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + struct drm_nouveau_private *dev_priv = crtc->dev->dev_private; + int head = nv_crtc->index; + uint8_t saved_cr21 = dev_priv->saved_reg.crtc_reg[head].CRTC[NV_CIO_CRE_21]; + + if (nv_two_heads(crtc->dev)) + NVSetOwner(crtc->dev, head); + + nouveau_hw_load_state(crtc->dev, head, &dev_priv->saved_reg); + nv_lock_vga_crtc_shadow(crtc->dev, head, saved_cr21); + + nv_crtc->last_dpms = NV_DPMS_CLEARED; +} + +static void nv_crtc_prepare(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + struct drm_crtc_helper_funcs *funcs = crtc->helper_private; + + if (nv_two_heads(dev)) + NVSetOwner(dev, nv_crtc->index); + + drm_vblank_pre_modeset(dev, nv_crtc->index); + funcs->dpms(crtc, DRM_MODE_DPMS_OFF); + + NVBlankScreen(dev, nv_crtc->index, true); + + /* Some more preparation. */ + NVWriteCRTC(dev, nv_crtc->index, NV_PCRTC_CONFIG, NV_PCRTC_CONFIG_START_ADDRESS_NON_VGA); + if (dev_priv->card_type == NV_40) { + uint32_t reg900 = NVReadRAMDAC(dev, nv_crtc->index, NV_PRAMDAC_900); + NVWriteRAMDAC(dev, nv_crtc->index, NV_PRAMDAC_900, reg900 & ~0x10000); + } +} + +static void nv_crtc_commit(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + struct drm_crtc_helper_funcs *funcs = crtc->helper_private; + struct drm_nouveau_private *dev_priv = crtc->dev->dev_private; + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + + nouveau_hw_load_state(dev, nv_crtc->index, &dev_priv->mode_reg); + nv04_crtc_mode_set_base(crtc, crtc->x, crtc->y, NULL); + +#ifdef __BIG_ENDIAN + /* turn on LFB swapping */ + { + uint8_t tmp = NVReadVgaCrtc(dev, nv_crtc->index, NV_CIO_CRE_RCR); + tmp |= MASK(NV_CIO_CRE_RCR_ENDIAN_BIG); + NVWriteVgaCrtc(dev, nv_crtc->index, NV_CIO_CRE_RCR, tmp); + } +#endif + + funcs->dpms(crtc, DRM_MODE_DPMS_ON); + drm_vblank_post_modeset(dev, nv_crtc->index); +} + +static void nv_crtc_destroy(struct drm_crtc *crtc) +{ + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + + NV_DEBUG_KMS(crtc->dev, "\n"); + + if (!nv_crtc) + return; + + drm_crtc_cleanup(crtc); + + nouveau_bo_unmap(nv_crtc->cursor.nvbo); + nouveau_bo_ref(NULL, &nv_crtc->cursor.nvbo); + kfree(nv_crtc); +} + +static void +nv_crtc_gamma_load(struct drm_crtc *crtc) +{ + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + struct drm_device *dev = nv_crtc->base.dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct rgb { uint8_t r, g, b; } __attribute__((packed)) *rgbs; + int i; + + rgbs = (struct rgb *)dev_priv->mode_reg.crtc_reg[nv_crtc->index].DAC; + for (i = 0; i < 256; i++) { + rgbs[i].r = nv_crtc->lut.r[i] >> 8; + rgbs[i].g = nv_crtc->lut.g[i] >> 8; + rgbs[i].b = nv_crtc->lut.b[i] >> 8; + } + + nouveau_hw_load_state_palette(dev, nv_crtc->index, &dev_priv->mode_reg); +} + +static void +nv_crtc_gamma_set(struct drm_crtc *crtc, u16 *r, u16 *g, u16 *b, uint32_t start, + uint32_t size) +{ + int end = (start + size > 256) ? 256 : start + size, i; + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + + for (i = start; i < end; i++) { + nv_crtc->lut.r[i] = r[i]; + nv_crtc->lut.g[i] = g[i]; + nv_crtc->lut.b[i] = b[i]; + } + + /* We need to know the depth before we upload, but it's possible to + * get called before a framebuffer is bound. If this is the case, + * mark the lut values as dirty by setting depth==0, and it'll be + * uploaded on the first mode_set_base() + */ + if (!nv_crtc->base.fb) { + nv_crtc->lut.depth = 0; + return; + } + + nv_crtc_gamma_load(crtc); +} + +static int +nv04_crtc_do_mode_set_base(struct drm_crtc *crtc, + struct drm_framebuffer *passed_fb, + int x, int y, bool atomic) +{ + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + struct drm_device *dev = crtc->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nv04_crtc_reg *regp = &dev_priv->mode_reg.crtc_reg[nv_crtc->index]; + struct drm_framebuffer *drm_fb = nv_crtc->base.fb; + struct nouveau_framebuffer *fb = nouveau_framebuffer(drm_fb); + int arb_burst, arb_lwm; + int ret; + + /* If atomic, we want to switch to the fb we were passed, so + * now we update pointers to do that. (We don't pin; just + * assume we're already pinned and update the base address.) + */ + if (atomic) { + drm_fb = passed_fb; + fb = nouveau_framebuffer(passed_fb); + } else { + /* If not atomic, we can go ahead and pin, and unpin the + * old fb we were passed. + */ + ret = nouveau_bo_pin(fb->nvbo, TTM_PL_FLAG_VRAM); + if (ret) + return ret; + + if (passed_fb) { + struct nouveau_framebuffer *ofb = nouveau_framebuffer(passed_fb); + nouveau_bo_unpin(ofb->nvbo); + } + } + + nv_crtc->fb.offset = fb->nvbo->bo.offset; + + if (nv_crtc->lut.depth != drm_fb->depth) { + nv_crtc->lut.depth = drm_fb->depth; + nv_crtc_gamma_load(crtc); + } + + /* Update the framebuffer format. */ + regp->CRTC[NV_CIO_CRE_PIXEL_INDEX] &= ~3; + regp->CRTC[NV_CIO_CRE_PIXEL_INDEX] |= (crtc->fb->depth + 1) / 8; + regp->ramdac_gen_ctrl &= ~NV_PRAMDAC_GENERAL_CONTROL_ALT_MODE_SEL; + if (crtc->fb->depth == 16) + regp->ramdac_gen_ctrl |= NV_PRAMDAC_GENERAL_CONTROL_ALT_MODE_SEL; + crtc_wr_cio_state(crtc, regp, NV_CIO_CRE_PIXEL_INDEX); + NVWriteRAMDAC(dev, nv_crtc->index, NV_PRAMDAC_GENERAL_CONTROL, + regp->ramdac_gen_ctrl); + + regp->CRTC[NV_CIO_CR_OFFSET_INDEX] = drm_fb->pitch >> 3; + regp->CRTC[NV_CIO_CRE_RPC0_INDEX] = + XLATE(drm_fb->pitch >> 3, 8, NV_CIO_CRE_RPC0_OFFSET_10_8); + regp->CRTC[NV_CIO_CRE_42] = + XLATE(drm_fb->pitch / 8, 11, NV_CIO_CRE_42_OFFSET_11); + crtc_wr_cio_state(crtc, regp, NV_CIO_CRE_RPC0_INDEX); + crtc_wr_cio_state(crtc, regp, NV_CIO_CR_OFFSET_INDEX); + crtc_wr_cio_state(crtc, regp, NV_CIO_CRE_42); + + /* Update the framebuffer location. */ + regp->fb_start = nv_crtc->fb.offset & ~3; + regp->fb_start += (y * drm_fb->pitch) + (x * drm_fb->bits_per_pixel / 8); + nv_set_crtc_base(dev, nv_crtc->index, regp->fb_start); + + /* Update the arbitration parameters. */ + nouveau_calc_arb(dev, crtc->mode.clock, drm_fb->bits_per_pixel, + &arb_burst, &arb_lwm); + + regp->CRTC[NV_CIO_CRE_FF_INDEX] = arb_burst; + regp->CRTC[NV_CIO_CRE_FFLWM__INDEX] = arb_lwm & 0xff; + crtc_wr_cio_state(crtc, regp, NV_CIO_CRE_FF_INDEX); + crtc_wr_cio_state(crtc, regp, NV_CIO_CRE_FFLWM__INDEX); + + if (dev_priv->card_type >= NV_20) { + regp->CRTC[NV_CIO_CRE_47] = arb_lwm >> 8; + crtc_wr_cio_state(crtc, regp, NV_CIO_CRE_47); + } + + return 0; +} + +static int +nv04_crtc_mode_set_base(struct drm_crtc *crtc, int x, int y, + struct drm_framebuffer *old_fb) +{ + return nv04_crtc_do_mode_set_base(crtc, old_fb, x, y, false); +} + +static int +nv04_crtc_mode_set_base_atomic(struct drm_crtc *crtc, + struct drm_framebuffer *fb, + int x, int y, enum mode_set_atomic state) +{ + struct drm_nouveau_private *dev_priv = crtc->dev->dev_private; + struct drm_device *dev = dev_priv->dev; + + if (state == ENTER_ATOMIC_MODE_SET) + nouveau_fbcon_save_disable_accel(dev); + else + nouveau_fbcon_restore_accel(dev); + + return nv04_crtc_do_mode_set_base(crtc, fb, x, y, true); +} + +static void nv04_cursor_upload(struct drm_device *dev, struct nouveau_bo *src, + struct nouveau_bo *dst) +{ + int width = nv_cursor_width(dev); + uint32_t pixel; + int i, j; + + for (i = 0; i < width; i++) { + for (j = 0; j < width; j++) { + pixel = nouveau_bo_rd32(src, i*64 + j); + + nouveau_bo_wr16(dst, i*width + j, (pixel & 0x80000000) >> 16 + | (pixel & 0xf80000) >> 9 + | (pixel & 0xf800) >> 6 + | (pixel & 0xf8) >> 3); + } + } +} + +static void nv11_cursor_upload(struct drm_device *dev, struct nouveau_bo *src, + struct nouveau_bo *dst) +{ + uint32_t pixel; + int alpha, i; + + /* nv11+ supports premultiplied (PM), or non-premultiplied (NPM) alpha + * cursors (though NPM in combination with fp dithering may not work on + * nv11, from "nv" driver history) + * NPM mode needs NV_PCRTC_CURSOR_CONFIG_ALPHA_BLEND set and is what the + * blob uses, however we get given PM cursors so we use PM mode + */ + for (i = 0; i < 64 * 64; i++) { + pixel = nouveau_bo_rd32(src, i); + + /* hw gets unhappy if alpha <= rgb values. for a PM image "less + * than" shouldn't happen; fix "equal to" case by adding one to + * alpha channel (slightly inaccurate, but so is attempting to + * get back to NPM images, due to limits of integer precision) + */ + alpha = pixel >> 24; + if (alpha > 0 && alpha < 255) + pixel = (pixel & 0x00ffffff) | ((alpha + 1) << 24); + +#ifdef __BIG_ENDIAN + { + struct drm_nouveau_private *dev_priv = dev->dev_private; + + if (dev_priv->chipset == 0x11) { + pixel = ((pixel & 0x000000ff) << 24) | + ((pixel & 0x0000ff00) << 8) | + ((pixel & 0x00ff0000) >> 8) | + ((pixel & 0xff000000) >> 24); + } + } +#endif + + nouveau_bo_wr32(dst, i, pixel); + } +} + +static int +nv04_crtc_cursor_set(struct drm_crtc *crtc, struct drm_file *file_priv, + uint32_t buffer_handle, uint32_t width, uint32_t height) +{ + struct drm_nouveau_private *dev_priv = crtc->dev->dev_private; + struct drm_device *dev = dev_priv->dev; + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + struct nouveau_bo *cursor = NULL; + struct drm_gem_object *gem; + int ret = 0; + + if (!buffer_handle) { + nv_crtc->cursor.hide(nv_crtc, true); + return 0; + } + + if (width != 64 || height != 64) + return -EINVAL; + + gem = drm_gem_object_lookup(dev, file_priv, buffer_handle); + if (!gem) + return -ENOENT; + cursor = nouveau_gem_object(gem); + + ret = nouveau_bo_map(cursor); + if (ret) + goto out; + + if (dev_priv->chipset >= 0x11) + nv11_cursor_upload(dev, cursor, nv_crtc->cursor.nvbo); + else + nv04_cursor_upload(dev, cursor, nv_crtc->cursor.nvbo); + + nouveau_bo_unmap(cursor); + nv_crtc->cursor.offset = nv_crtc->cursor.nvbo->bo.offset; + nv_crtc->cursor.set_offset(nv_crtc, nv_crtc->cursor.offset); + nv_crtc->cursor.show(nv_crtc, true); +out: + drm_gem_object_unreference_unlocked(gem); + return ret; +} + +static int +nv04_crtc_cursor_move(struct drm_crtc *crtc, int x, int y) +{ + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + + nv_crtc->cursor.set_pos(nv_crtc, x, y); + return 0; +} + +static const struct drm_crtc_funcs nv04_crtc_funcs = { + .save = nv_crtc_save, + .restore = nv_crtc_restore, + .cursor_set = nv04_crtc_cursor_set, + .cursor_move = nv04_crtc_cursor_move, + .gamma_set = nv_crtc_gamma_set, + .set_config = drm_crtc_helper_set_config, + .page_flip = nouveau_crtc_page_flip, + .destroy = nv_crtc_destroy, +}; + +static const struct drm_crtc_helper_funcs nv04_crtc_helper_funcs = { + .dpms = nv_crtc_dpms, + .prepare = nv_crtc_prepare, + .commit = nv_crtc_commit, + .mode_fixup = nv_crtc_mode_fixup, + .mode_set = nv_crtc_mode_set, + .mode_set_base = nv04_crtc_mode_set_base, + .mode_set_base_atomic = nv04_crtc_mode_set_base_atomic, + .load_lut = nv_crtc_gamma_load, +}; + +int +nv04_crtc_create(struct drm_device *dev, int crtc_num) +{ + struct nouveau_crtc *nv_crtc; + int ret, i; + + nv_crtc = kzalloc(sizeof(*nv_crtc), GFP_KERNEL); + if (!nv_crtc) + return -ENOMEM; + + for (i = 0; i < 256; i++) { + nv_crtc->lut.r[i] = i << 8; + nv_crtc->lut.g[i] = i << 8; + nv_crtc->lut.b[i] = i << 8; + } + nv_crtc->lut.depth = 0; + + nv_crtc->index = crtc_num; + nv_crtc->last_dpms = NV_DPMS_CLEARED; + + drm_crtc_init(dev, &nv_crtc->base, &nv04_crtc_funcs); + drm_crtc_helper_add(&nv_crtc->base, &nv04_crtc_helper_funcs); + drm_mode_crtc_set_gamma_size(&nv_crtc->base, 256); + + ret = nouveau_bo_new(dev, NULL, 64*64*4, 0x100, TTM_PL_FLAG_VRAM, + 0, 0x0000, &nv_crtc->cursor.nvbo); + if (!ret) { + ret = nouveau_bo_pin(nv_crtc->cursor.nvbo, TTM_PL_FLAG_VRAM); + if (!ret) + ret = nouveau_bo_map(nv_crtc->cursor.nvbo); + if (ret) + nouveau_bo_ref(NULL, &nv_crtc->cursor.nvbo); + } + + nv04_cursor_init(nv_crtc); + + return 0; +} + diff --git a/drivers/gpu/drm/nouveau/nv04_cursor.c b/drivers/gpu/drm/nouveau/nv04_cursor.c new file mode 100644 index 00000000..aaf3de3b --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv04_cursor.c @@ -0,0 +1,71 @@ +#include "drmP.h" +#include "drm_mode.h" +#include "nouveau_reg.h" +#include "nouveau_drv.h" +#include "nouveau_crtc.h" +#include "nouveau_hw.h" + +static void +nv04_cursor_show(struct nouveau_crtc *nv_crtc, bool update) +{ + nv_show_cursor(nv_crtc->base.dev, nv_crtc->index, true); +} + +static void +nv04_cursor_hide(struct nouveau_crtc *nv_crtc, bool update) +{ + nv_show_cursor(nv_crtc->base.dev, nv_crtc->index, false); +} + +static void +nv04_cursor_set_pos(struct nouveau_crtc *nv_crtc, int x, int y) +{ + nv_crtc->cursor_saved_x = x; nv_crtc->cursor_saved_y = y; + NVWriteRAMDAC(nv_crtc->base.dev, nv_crtc->index, + NV_PRAMDAC_CU_START_POS, + XLATE(y, 0, NV_PRAMDAC_CU_START_POS_Y) | + XLATE(x, 0, NV_PRAMDAC_CU_START_POS_X)); +} + +static void +crtc_wr_cio_state(struct drm_crtc *crtc, struct nv04_crtc_reg *crtcstate, int index) +{ + NVWriteVgaCrtc(crtc->dev, nouveau_crtc(crtc)->index, index, + crtcstate->CRTC[index]); +} + +static void +nv04_cursor_set_offset(struct nouveau_crtc *nv_crtc, uint32_t offset) +{ + struct drm_device *dev = nv_crtc->base.dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nv04_crtc_reg *regp = &dev_priv->mode_reg.crtc_reg[nv_crtc->index]; + struct drm_crtc *crtc = &nv_crtc->base; + + regp->CRTC[NV_CIO_CRE_HCUR_ADDR0_INDEX] = + MASK(NV_CIO_CRE_HCUR_ASI) | + XLATE(offset, 17, NV_CIO_CRE_HCUR_ADDR0_ADR); + regp->CRTC[NV_CIO_CRE_HCUR_ADDR1_INDEX] = + XLATE(offset, 11, NV_CIO_CRE_HCUR_ADDR1_ADR); + if (crtc->mode.flags & DRM_MODE_FLAG_DBLSCAN) + regp->CRTC[NV_CIO_CRE_HCUR_ADDR1_INDEX] |= + MASK(NV_CIO_CRE_HCUR_ADDR1_CUR_DBL); + regp->CRTC[NV_CIO_CRE_HCUR_ADDR2_INDEX] = offset >> 24; + + crtc_wr_cio_state(crtc, regp, NV_CIO_CRE_HCUR_ADDR0_INDEX); + crtc_wr_cio_state(crtc, regp, NV_CIO_CRE_HCUR_ADDR1_INDEX); + crtc_wr_cio_state(crtc, regp, NV_CIO_CRE_HCUR_ADDR2_INDEX); + if (dev_priv->card_type == NV_40) + nv_fix_nv40_hw_cursor(dev, nv_crtc->index); +} + +int +nv04_cursor_init(struct nouveau_crtc *crtc) +{ + crtc->cursor.set_offset = nv04_cursor_set_offset; + crtc->cursor.set_pos = nv04_cursor_set_pos; + crtc->cursor.hide = nv04_cursor_hide; + crtc->cursor.show = nv04_cursor_show; + return 0; +} + diff --git a/drivers/gpu/drm/nouveau/nv04_dac.c b/drivers/gpu/drm/nouveau/nv04_dac.c new file mode 100644 index 00000000..e000455e --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv04_dac.c @@ -0,0 +1,544 @@ +/* + * Copyright 2003 NVIDIA, Corporation + * Copyright 2006 Dave Airlie + * Copyright 2007 Maarten Maathuis + * Copyright 2007-2009 Stuart Bennett + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + */ + +#include "drmP.h" +#include "drm_crtc_helper.h" + +#include "nouveau_drv.h" +#include "nouveau_encoder.h" +#include "nouveau_connector.h" +#include "nouveau_crtc.h" +#include "nouveau_hw.h" +#include "nvreg.h" + +int nv04_dac_output_offset(struct drm_encoder *encoder) +{ + struct dcb_entry *dcb = nouveau_encoder(encoder)->dcb; + int offset = 0; + + if (dcb->or & (8 | OUTPUT_C)) + offset += 0x68; + if (dcb->or & (8 | OUTPUT_B)) + offset += 0x2000; + + return offset; +} + +/* + * arbitrary limit to number of sense oscillations tolerated in one sample + * period (observed to be at least 13 in "nvidia") + */ +#define MAX_HBLANK_OSC 20 + +/* + * arbitrary limit to number of conflicting sample pairs to tolerate at a + * voltage step (observed to be at least 5 in "nvidia") + */ +#define MAX_SAMPLE_PAIRS 10 + +static int sample_load_twice(struct drm_device *dev, bool sense[2]) +{ + int i; + + for (i = 0; i < 2; i++) { + bool sense_a, sense_b, sense_b_prime; + int j = 0; + + /* + * wait for bit 0 clear -- out of hblank -- (say reg value 0x4), + * then wait for transition 0x4->0x5->0x4: enter hblank, leave + * hblank again + * use a 10ms timeout (guards against crtc being inactive, in + * which case blank state would never change) + */ + if (!nouveau_wait_eq(dev, 10000000, NV_PRMCIO_INP0__COLOR, + 0x00000001, 0x00000000)) + return -EBUSY; + if (!nouveau_wait_eq(dev, 10000000, NV_PRMCIO_INP0__COLOR, + 0x00000001, 0x00000001)) + return -EBUSY; + if (!nouveau_wait_eq(dev, 10000000, NV_PRMCIO_INP0__COLOR, + 0x00000001, 0x00000000)) + return -EBUSY; + + udelay(100); + /* when level triggers, sense is _LO_ */ + sense_a = nv_rd08(dev, NV_PRMCIO_INP0) & 0x10; + + /* take another reading until it agrees with sense_a... */ + do { + udelay(100); + sense_b = nv_rd08(dev, NV_PRMCIO_INP0) & 0x10; + if (sense_a != sense_b) { + sense_b_prime = + nv_rd08(dev, NV_PRMCIO_INP0) & 0x10; + if (sense_b == sense_b_prime) { + /* ... unless two consecutive subsequent + * samples agree; sense_a is replaced */ + sense_a = sense_b; + /* force mis-match so we loop */ + sense_b = !sense_a; + } + } + } while ((sense_a != sense_b) && ++j < MAX_HBLANK_OSC); + + if (j == MAX_HBLANK_OSC) + /* with so much oscillation, default to sense:LO */ + sense[i] = false; + else + sense[i] = sense_a; + } + + return 0; +} + +static enum drm_connector_status nv04_dac_detect(struct drm_encoder *encoder, + struct drm_connector *connector) +{ + struct drm_device *dev = encoder->dev; + uint8_t saved_seq1, saved_pi, saved_rpc1, saved_cr_mode; + uint8_t saved_palette0[3], saved_palette_mask; + uint32_t saved_rtest_ctrl, saved_rgen_ctrl; + int i; + uint8_t blue; + bool sense = true; + + /* + * for this detection to work, there needs to be a mode set up on the + * CRTC. this is presumed to be the case + */ + + if (nv_two_heads(dev)) + /* only implemented for head A for now */ + NVSetOwner(dev, 0); + + saved_cr_mode = NVReadVgaCrtc(dev, 0, NV_CIO_CR_MODE_INDEX); + NVWriteVgaCrtc(dev, 0, NV_CIO_CR_MODE_INDEX, saved_cr_mode | 0x80); + + saved_seq1 = NVReadVgaSeq(dev, 0, NV_VIO_SR_CLOCK_INDEX); + NVWriteVgaSeq(dev, 0, NV_VIO_SR_CLOCK_INDEX, saved_seq1 & ~0x20); + + saved_rtest_ctrl = NVReadRAMDAC(dev, 0, NV_PRAMDAC_TEST_CONTROL); + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_TEST_CONTROL, + saved_rtest_ctrl & ~NV_PRAMDAC_TEST_CONTROL_PWRDWN_DAC_OFF); + + msleep(10); + + saved_pi = NVReadVgaCrtc(dev, 0, NV_CIO_CRE_PIXEL_INDEX); + NVWriteVgaCrtc(dev, 0, NV_CIO_CRE_PIXEL_INDEX, + saved_pi & ~(0x80 | MASK(NV_CIO_CRE_PIXEL_FORMAT))); + saved_rpc1 = NVReadVgaCrtc(dev, 0, NV_CIO_CRE_RPC1_INDEX); + NVWriteVgaCrtc(dev, 0, NV_CIO_CRE_RPC1_INDEX, saved_rpc1 & ~0xc0); + + nv_wr08(dev, NV_PRMDIO_READ_MODE_ADDRESS, 0x0); + for (i = 0; i < 3; i++) + saved_palette0[i] = nv_rd08(dev, NV_PRMDIO_PALETTE_DATA); + saved_palette_mask = nv_rd08(dev, NV_PRMDIO_PIXEL_MASK); + nv_wr08(dev, NV_PRMDIO_PIXEL_MASK, 0); + + saved_rgen_ctrl = NVReadRAMDAC(dev, 0, NV_PRAMDAC_GENERAL_CONTROL); + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_GENERAL_CONTROL, + (saved_rgen_ctrl & ~(NV_PRAMDAC_GENERAL_CONTROL_BPC_8BITS | + NV_PRAMDAC_GENERAL_CONTROL_TERMINATION_75OHM)) | + NV_PRAMDAC_GENERAL_CONTROL_PIXMIX_ON); + + blue = 8; /* start of test range */ + + do { + bool sense_pair[2]; + + nv_wr08(dev, NV_PRMDIO_WRITE_MODE_ADDRESS, 0); + nv_wr08(dev, NV_PRMDIO_PALETTE_DATA, 0); + nv_wr08(dev, NV_PRMDIO_PALETTE_DATA, 0); + /* testing blue won't find monochrome monitors. I don't care */ + nv_wr08(dev, NV_PRMDIO_PALETTE_DATA, blue); + + i = 0; + /* take sample pairs until both samples in the pair agree */ + do { + if (sample_load_twice(dev, sense_pair)) + goto out; + } while ((sense_pair[0] != sense_pair[1]) && + ++i < MAX_SAMPLE_PAIRS); + + if (i == MAX_SAMPLE_PAIRS) + /* too much oscillation defaults to LO */ + sense = false; + else + sense = sense_pair[0]; + + /* + * if sense goes LO before blue ramps to 0x18, monitor is not connected. + * ergo, if blue gets to 0x18, monitor must be connected + */ + } while (++blue < 0x18 && sense); + +out: + nv_wr08(dev, NV_PRMDIO_PIXEL_MASK, saved_palette_mask); + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_GENERAL_CONTROL, saved_rgen_ctrl); + nv_wr08(dev, NV_PRMDIO_WRITE_MODE_ADDRESS, 0); + for (i = 0; i < 3; i++) + nv_wr08(dev, NV_PRMDIO_PALETTE_DATA, saved_palette0[i]); + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_TEST_CONTROL, saved_rtest_ctrl); + NVWriteVgaCrtc(dev, 0, NV_CIO_CRE_PIXEL_INDEX, saved_pi); + NVWriteVgaCrtc(dev, 0, NV_CIO_CRE_RPC1_INDEX, saved_rpc1); + NVWriteVgaSeq(dev, 0, NV_VIO_SR_CLOCK_INDEX, saved_seq1); + NVWriteVgaCrtc(dev, 0, NV_CIO_CR_MODE_INDEX, saved_cr_mode); + + if (blue == 0x18) { + NV_INFO(dev, "Load detected on head A\n"); + return connector_status_connected; + } + + return connector_status_disconnected; +} + +uint32_t nv17_dac_sample_load(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpio_engine *gpio = &dev_priv->engine.gpio; + struct dcb_entry *dcb = nouveau_encoder(encoder)->dcb; + uint32_t sample, testval, regoffset = nv04_dac_output_offset(encoder); + uint32_t saved_powerctrl_2 = 0, saved_powerctrl_4 = 0, saved_routput, + saved_rtest_ctrl, saved_gpio0, saved_gpio1, temp, routput; + int head; + +#define RGB_TEST_DATA(r, g, b) (r << 0 | g << 10 | b << 20) + if (dcb->type == OUTPUT_TV) { + testval = RGB_TEST_DATA(0xa0, 0xa0, 0xa0); + + if (dev_priv->vbios.tvdactestval) + testval = dev_priv->vbios.tvdactestval; + } else { + testval = RGB_TEST_DATA(0x140, 0x140, 0x140); /* 0x94050140 */ + + if (dev_priv->vbios.dactestval) + testval = dev_priv->vbios.dactestval; + } + + saved_rtest_ctrl = NVReadRAMDAC(dev, 0, NV_PRAMDAC_TEST_CONTROL + regoffset); + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_TEST_CONTROL + regoffset, + saved_rtest_ctrl & ~NV_PRAMDAC_TEST_CONTROL_PWRDWN_DAC_OFF); + + saved_powerctrl_2 = nvReadMC(dev, NV_PBUS_POWERCTRL_2); + + nvWriteMC(dev, NV_PBUS_POWERCTRL_2, saved_powerctrl_2 & 0xd7ffffff); + if (regoffset == 0x68) { + saved_powerctrl_4 = nvReadMC(dev, NV_PBUS_POWERCTRL_4); + nvWriteMC(dev, NV_PBUS_POWERCTRL_4, saved_powerctrl_4 & 0xffffffcf); + } + + saved_gpio1 = gpio->get(dev, DCB_GPIO_TVDAC1); + saved_gpio0 = gpio->get(dev, DCB_GPIO_TVDAC0); + + gpio->set(dev, DCB_GPIO_TVDAC1, dcb->type == OUTPUT_TV); + gpio->set(dev, DCB_GPIO_TVDAC0, dcb->type == OUTPUT_TV); + + msleep(4); + + saved_routput = NVReadRAMDAC(dev, 0, NV_PRAMDAC_DACCLK + regoffset); + head = (saved_routput & 0x100) >> 8; + + /* if there's a spare crtc, using it will minimise flicker */ + if (!(NVReadVgaCrtc(dev, head, NV_CIO_CRE_RPC1_INDEX) & 0xC0)) + head ^= 1; + + /* nv driver and nv31 use 0xfffffeee, nv34 and 6600 use 0xfffffece */ + routput = (saved_routput & 0xfffffece) | head << 8; + + if (dev_priv->card_type >= NV_40) { + if (dcb->type == OUTPUT_TV) + routput |= 0x1a << 16; + else + routput &= ~(0x1a << 16); + } + + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_DACCLK + regoffset, routput); + msleep(1); + + temp = NVReadRAMDAC(dev, 0, NV_PRAMDAC_DACCLK + regoffset); + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_DACCLK + regoffset, temp | 1); + + NVWriteRAMDAC(dev, head, NV_PRAMDAC_TESTPOINT_DATA, + NV_PRAMDAC_TESTPOINT_DATA_NOTBLANK | testval); + temp = NVReadRAMDAC(dev, head, NV_PRAMDAC_TEST_CONTROL); + NVWriteRAMDAC(dev, head, NV_PRAMDAC_TEST_CONTROL, + temp | NV_PRAMDAC_TEST_CONTROL_TP_INS_EN_ASSERTED); + msleep(5); + + sample = NVReadRAMDAC(dev, 0, NV_PRAMDAC_TEST_CONTROL + regoffset); + /* do it again just in case it's a residual current */ + sample &= NVReadRAMDAC(dev, 0, NV_PRAMDAC_TEST_CONTROL + regoffset); + + temp = NVReadRAMDAC(dev, head, NV_PRAMDAC_TEST_CONTROL); + NVWriteRAMDAC(dev, head, NV_PRAMDAC_TEST_CONTROL, + temp & ~NV_PRAMDAC_TEST_CONTROL_TP_INS_EN_ASSERTED); + NVWriteRAMDAC(dev, head, NV_PRAMDAC_TESTPOINT_DATA, 0); + + /* bios does something more complex for restoring, but I think this is good enough */ + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_DACCLK + regoffset, saved_routput); + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_TEST_CONTROL + regoffset, saved_rtest_ctrl); + if (regoffset == 0x68) + nvWriteMC(dev, NV_PBUS_POWERCTRL_4, saved_powerctrl_4); + nvWriteMC(dev, NV_PBUS_POWERCTRL_2, saved_powerctrl_2); + + gpio->set(dev, DCB_GPIO_TVDAC1, saved_gpio1); + gpio->set(dev, DCB_GPIO_TVDAC0, saved_gpio0); + + return sample; +} + +static enum drm_connector_status +nv17_dac_detect(struct drm_encoder *encoder, struct drm_connector *connector) +{ + struct drm_device *dev = encoder->dev; + struct dcb_entry *dcb = nouveau_encoder(encoder)->dcb; + + if (nv04_dac_in_use(encoder)) + return connector_status_disconnected; + + if (nv17_dac_sample_load(encoder) & + NV_PRAMDAC_TEST_CONTROL_SENSEB_ALLHI) { + NV_INFO(dev, "Load detected on output %c\n", + '@' + ffs(dcb->or)); + return connector_status_connected; + } else { + return connector_status_disconnected; + } +} + +static bool nv04_dac_mode_fixup(struct drm_encoder *encoder, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + if (nv04_dac_in_use(encoder)) + return false; + + return true; +} + +static void nv04_dac_prepare(struct drm_encoder *encoder) +{ + struct drm_encoder_helper_funcs *helper = encoder->helper_private; + struct drm_device *dev = encoder->dev; + int head = nouveau_crtc(encoder->crtc)->index; + + helper->dpms(encoder, DRM_MODE_DPMS_OFF); + + nv04_dfp_disable(dev, head); +} + +static void nv04_dac_mode_set(struct drm_encoder *encoder, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct drm_device *dev = encoder->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + int head = nouveau_crtc(encoder->crtc)->index; + + if (nv_gf4_disp_arch(dev)) { + struct drm_encoder *rebind; + uint32_t dac_offset = nv04_dac_output_offset(encoder); + uint32_t otherdac; + + /* bit 16-19 are bits that are set on some G70 cards, + * but don't seem to have much effect */ + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_DACCLK + dac_offset, + head << 8 | NV_PRAMDAC_DACCLK_SEL_DACCLK); + /* force any other vga encoders to bind to the other crtc */ + list_for_each_entry(rebind, &dev->mode_config.encoder_list, head) { + if (rebind == encoder + || nouveau_encoder(rebind)->dcb->type != OUTPUT_ANALOG) + continue; + + dac_offset = nv04_dac_output_offset(rebind); + otherdac = NVReadRAMDAC(dev, 0, NV_PRAMDAC_DACCLK + dac_offset); + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_DACCLK + dac_offset, + (otherdac & ~0x0100) | (head ^ 1) << 8); + } + } + + /* This could use refinement for flatpanels, but it should work this way */ + if (dev_priv->chipset < 0x44) + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_TEST_CONTROL + nv04_dac_output_offset(encoder), 0xf0000000); + else + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_TEST_CONTROL + nv04_dac_output_offset(encoder), 0x00100000); +} + +static void nv04_dac_commit(struct drm_encoder *encoder) +{ + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + struct drm_device *dev = encoder->dev; + struct nouveau_crtc *nv_crtc = nouveau_crtc(encoder->crtc); + struct drm_encoder_helper_funcs *helper = encoder->helper_private; + + helper->dpms(encoder, DRM_MODE_DPMS_ON); + + NV_INFO(dev, "Output %s is running on CRTC %d using output %c\n", + drm_get_connector_name(&nouveau_encoder_connector_get(nv_encoder)->base), + nv_crtc->index, '@' + ffs(nv_encoder->dcb->or)); +} + +void nv04_dac_update_dacclk(struct drm_encoder *encoder, bool enable) +{ + struct drm_device *dev = encoder->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct dcb_entry *dcb = nouveau_encoder(encoder)->dcb; + + if (nv_gf4_disp_arch(dev)) { + uint32_t *dac_users = &dev_priv->dac_users[ffs(dcb->or) - 1]; + int dacclk_off = NV_PRAMDAC_DACCLK + nv04_dac_output_offset(encoder); + uint32_t dacclk = NVReadRAMDAC(dev, 0, dacclk_off); + + if (enable) { + *dac_users |= 1 << dcb->index; + NVWriteRAMDAC(dev, 0, dacclk_off, dacclk | NV_PRAMDAC_DACCLK_SEL_DACCLK); + + } else { + *dac_users &= ~(1 << dcb->index); + if (!*dac_users) + NVWriteRAMDAC(dev, 0, dacclk_off, + dacclk & ~NV_PRAMDAC_DACCLK_SEL_DACCLK); + } + } +} + +/* Check if the DAC corresponding to 'encoder' is being used by + * someone else. */ +bool nv04_dac_in_use(struct drm_encoder *encoder) +{ + struct drm_nouveau_private *dev_priv = encoder->dev->dev_private; + struct dcb_entry *dcb = nouveau_encoder(encoder)->dcb; + + return nv_gf4_disp_arch(encoder->dev) && + (dev_priv->dac_users[ffs(dcb->or) - 1] & ~(1 << dcb->index)); +} + +static void nv04_dac_dpms(struct drm_encoder *encoder, int mode) +{ + struct drm_device *dev = encoder->dev; + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + + if (nv_encoder->last_dpms == mode) + return; + nv_encoder->last_dpms = mode; + + NV_INFO(dev, "Setting dpms mode %d on vga encoder (output %d)\n", + mode, nv_encoder->dcb->index); + + nv04_dac_update_dacclk(encoder, mode == DRM_MODE_DPMS_ON); +} + +static void nv04_dac_save(struct drm_encoder *encoder) +{ + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + struct drm_device *dev = encoder->dev; + + if (nv_gf4_disp_arch(dev)) + nv_encoder->restore.output = NVReadRAMDAC(dev, 0, NV_PRAMDAC_DACCLK + + nv04_dac_output_offset(encoder)); +} + +static void nv04_dac_restore(struct drm_encoder *encoder) +{ + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + struct drm_device *dev = encoder->dev; + + if (nv_gf4_disp_arch(dev)) + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_DACCLK + nv04_dac_output_offset(encoder), + nv_encoder->restore.output); + + nv_encoder->last_dpms = NV_DPMS_CLEARED; +} + +static void nv04_dac_destroy(struct drm_encoder *encoder) +{ + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + + NV_DEBUG_KMS(encoder->dev, "\n"); + + drm_encoder_cleanup(encoder); + kfree(nv_encoder); +} + +static const struct drm_encoder_helper_funcs nv04_dac_helper_funcs = { + .dpms = nv04_dac_dpms, + .save = nv04_dac_save, + .restore = nv04_dac_restore, + .mode_fixup = nv04_dac_mode_fixup, + .prepare = nv04_dac_prepare, + .commit = nv04_dac_commit, + .mode_set = nv04_dac_mode_set, + .detect = nv04_dac_detect +}; + +static const struct drm_encoder_helper_funcs nv17_dac_helper_funcs = { + .dpms = nv04_dac_dpms, + .save = nv04_dac_save, + .restore = nv04_dac_restore, + .mode_fixup = nv04_dac_mode_fixup, + .prepare = nv04_dac_prepare, + .commit = nv04_dac_commit, + .mode_set = nv04_dac_mode_set, + .detect = nv17_dac_detect +}; + +static const struct drm_encoder_funcs nv04_dac_funcs = { + .destroy = nv04_dac_destroy, +}; + +int +nv04_dac_create(struct drm_connector *connector, struct dcb_entry *entry) +{ + const struct drm_encoder_helper_funcs *helper; + struct nouveau_encoder *nv_encoder = NULL; + struct drm_device *dev = connector->dev; + struct drm_encoder *encoder; + + nv_encoder = kzalloc(sizeof(*nv_encoder), GFP_KERNEL); + if (!nv_encoder) + return -ENOMEM; + + encoder = to_drm_encoder(nv_encoder); + + nv_encoder->dcb = entry; + nv_encoder->or = ffs(entry->or) - 1; + + if (nv_gf4_disp_arch(dev)) + helper = &nv17_dac_helper_funcs; + else + helper = &nv04_dac_helper_funcs; + + drm_encoder_init(dev, encoder, &nv04_dac_funcs, DRM_MODE_ENCODER_DAC); + drm_encoder_helper_add(encoder, helper); + + encoder->possible_crtcs = entry->heads; + encoder->possible_clones = 0; + + drm_mode_connector_attach_encoder(connector, encoder); + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nv04_dfp.c b/drivers/gpu/drm/nouveau/nv04_dfp.c new file mode 100644 index 00000000..12098bf8 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv04_dfp.c @@ -0,0 +1,712 @@ +/* + * Copyright 2003 NVIDIA, Corporation + * Copyright 2006 Dave Airlie + * Copyright 2007 Maarten Maathuis + * Copyright 2007-2009 Stuart Bennett + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + */ + +#include "drmP.h" +#include "drm_crtc_helper.h" + +#include "nouveau_drv.h" +#include "nouveau_encoder.h" +#include "nouveau_connector.h" +#include "nouveau_crtc.h" +#include "nouveau_hw.h" +#include "nvreg.h" + +#include "i2c/sil164.h" + +#define FP_TG_CONTROL_ON (NV_PRAMDAC_FP_TG_CONTROL_DISPEN_POS | \ + NV_PRAMDAC_FP_TG_CONTROL_HSYNC_POS | \ + NV_PRAMDAC_FP_TG_CONTROL_VSYNC_POS) +#define FP_TG_CONTROL_OFF (NV_PRAMDAC_FP_TG_CONTROL_DISPEN_DISABLE | \ + NV_PRAMDAC_FP_TG_CONTROL_HSYNC_DISABLE | \ + NV_PRAMDAC_FP_TG_CONTROL_VSYNC_DISABLE) + +static inline bool is_fpc_off(uint32_t fpc) +{ + return ((fpc & (FP_TG_CONTROL_ON | FP_TG_CONTROL_OFF)) == + FP_TG_CONTROL_OFF); +} + +int nv04_dfp_get_bound_head(struct drm_device *dev, struct dcb_entry *dcbent) +{ + /* special case of nv_read_tmds to find crtc associated with an output. + * this does not give a correct answer for off-chip dvi, but there's no + * use for such an answer anyway + */ + int ramdac = (dcbent->or & OUTPUT_C) >> 2; + + NVWriteRAMDAC(dev, ramdac, NV_PRAMDAC_FP_TMDS_CONTROL, + NV_PRAMDAC_FP_TMDS_CONTROL_WRITE_DISABLE | 0x4); + return ((NVReadRAMDAC(dev, ramdac, NV_PRAMDAC_FP_TMDS_DATA) & 0x8) >> 3) ^ ramdac; +} + +void nv04_dfp_bind_head(struct drm_device *dev, struct dcb_entry *dcbent, + int head, bool dl) +{ + /* The BIOS scripts don't do this for us, sadly + * Luckily we do know the values ;-) + * + * head < 0 indicates we wish to force a setting with the overrideval + * (for VT restore etc.) + */ + + int ramdac = (dcbent->or & OUTPUT_C) >> 2; + uint8_t tmds04 = 0x80; + + if (head != ramdac) + tmds04 = 0x88; + + if (dcbent->type == OUTPUT_LVDS) + tmds04 |= 0x01; + + nv_write_tmds(dev, dcbent->or, 0, 0x04, tmds04); + + if (dl) /* dual link */ + nv_write_tmds(dev, dcbent->or, 1, 0x04, tmds04 ^ 0x08); +} + +void nv04_dfp_disable(struct drm_device *dev, int head) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nv04_crtc_reg *crtcstate = dev_priv->mode_reg.crtc_reg; + + if (NVReadRAMDAC(dev, head, NV_PRAMDAC_FP_TG_CONTROL) & + FP_TG_CONTROL_ON) { + /* digital remnants must be cleaned before new crtc + * values programmed. delay is time for the vga stuff + * to realise it's in control again + */ + NVWriteRAMDAC(dev, head, NV_PRAMDAC_FP_TG_CONTROL, + FP_TG_CONTROL_OFF); + msleep(50); + } + /* don't inadvertently turn it on when state written later */ + crtcstate[head].fp_control = FP_TG_CONTROL_OFF; + crtcstate[head].CRTC[NV_CIO_CRE_LCD__INDEX] &= + ~NV_CIO_CRE_LCD_ROUTE_MASK; +} + +void nv04_dfp_update_fp_control(struct drm_encoder *encoder, int mode) +{ + struct drm_device *dev = encoder->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct drm_crtc *crtc; + struct nouveau_crtc *nv_crtc; + uint32_t *fpc; + + if (mode == DRM_MODE_DPMS_ON) { + nv_crtc = nouveau_crtc(encoder->crtc); + fpc = &dev_priv->mode_reg.crtc_reg[nv_crtc->index].fp_control; + + if (is_fpc_off(*fpc)) { + /* using saved value is ok, as (is_digital && dpms_on && + * fp_control==OFF) is (at present) *only* true when + * fpc's most recent change was by below "off" code + */ + *fpc = nv_crtc->dpms_saved_fp_control; + } + + nv_crtc->fp_users |= 1 << nouveau_encoder(encoder)->dcb->index; + NVWriteRAMDAC(dev, nv_crtc->index, NV_PRAMDAC_FP_TG_CONTROL, *fpc); + } else { + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { + nv_crtc = nouveau_crtc(crtc); + fpc = &dev_priv->mode_reg.crtc_reg[nv_crtc->index].fp_control; + + nv_crtc->fp_users &= ~(1 << nouveau_encoder(encoder)->dcb->index); + if (!is_fpc_off(*fpc) && !nv_crtc->fp_users) { + nv_crtc->dpms_saved_fp_control = *fpc; + /* cut the FP output */ + *fpc &= ~FP_TG_CONTROL_ON; + *fpc |= FP_TG_CONTROL_OFF; + NVWriteRAMDAC(dev, nv_crtc->index, + NV_PRAMDAC_FP_TG_CONTROL, *fpc); + } + } + } +} + +static struct drm_encoder *get_tmds_slave(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct dcb_entry *dcb = nouveau_encoder(encoder)->dcb; + struct drm_encoder *slave; + + if (dcb->type != OUTPUT_TMDS || dcb->location == DCB_LOC_ON_CHIP) + return NULL; + + /* Some BIOSes (e.g. the one in a Quadro FX1000) report several + * TMDS transmitters at the same I2C address, in the same I2C + * bus. This can still work because in that case one of them is + * always hard-wired to a reasonable configuration using straps, + * and the other one needs to be programmed. + * + * I don't think there's a way to know which is which, even the + * blob programs the one exposed via I2C for *both* heads, so + * let's do the same. + */ + list_for_each_entry(slave, &dev->mode_config.encoder_list, head) { + struct dcb_entry *slave_dcb = nouveau_encoder(slave)->dcb; + + if (slave_dcb->type == OUTPUT_TMDS && get_slave_funcs(slave) && + slave_dcb->tmdsconf.slave_addr == dcb->tmdsconf.slave_addr) + return slave; + } + + return NULL; +} + +static bool nv04_dfp_mode_fixup(struct drm_encoder *encoder, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + struct nouveau_connector *nv_connector = nouveau_encoder_connector_get(nv_encoder); + + if (!nv_connector->native_mode || + nv_connector->scaling_mode == DRM_MODE_SCALE_NONE || + mode->hdisplay > nv_connector->native_mode->hdisplay || + mode->vdisplay > nv_connector->native_mode->vdisplay) { + nv_encoder->mode = *adjusted_mode; + + } else { + nv_encoder->mode = *nv_connector->native_mode; + adjusted_mode->clock = nv_connector->native_mode->clock; + } + + return true; +} + +static void nv04_dfp_prepare_sel_clk(struct drm_device *dev, + struct nouveau_encoder *nv_encoder, int head) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nv04_mode_state *state = &dev_priv->mode_reg; + uint32_t bits1618 = nv_encoder->dcb->or & OUTPUT_A ? 0x10000 : 0x40000; + + if (nv_encoder->dcb->location != DCB_LOC_ON_CHIP) + return; + + /* SEL_CLK is only used on the primary ramdac + * It toggles spread spectrum PLL output and sets the bindings of PLLs + * to heads on digital outputs + */ + if (head) + state->sel_clk |= bits1618; + else + state->sel_clk &= ~bits1618; + + /* nv30: + * bit 0 NVClk spread spectrum on/off + * bit 2 MemClk spread spectrum on/off + * bit 4 PixClk1 spread spectrum on/off toggle + * bit 6 PixClk2 spread spectrum on/off toggle + * + * nv40 (observations from bios behaviour and mmio traces): + * bits 4&6 as for nv30 + * bits 5&7 head dependent as for bits 4&6, but do not appear with 4&6; + * maybe a different spread mode + * bits 8&10 seen on dual-link dvi outputs, purpose unknown (set by POST scripts) + * The logic behind turning spread spectrum on/off in the first place, + * and which bit-pair to use, is unclear on nv40 (for earlier cards, the fp table + * entry has the necessary info) + */ + if (nv_encoder->dcb->type == OUTPUT_LVDS && dev_priv->saved_reg.sel_clk & 0xf0) { + int shift = (dev_priv->saved_reg.sel_clk & 0x50) ? 0 : 1; + + state->sel_clk &= ~0xf0; + state->sel_clk |= (head ? 0x40 : 0x10) << shift; + } +} + +static void nv04_dfp_prepare(struct drm_encoder *encoder) +{ + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + struct drm_encoder_helper_funcs *helper = encoder->helper_private; + struct drm_device *dev = encoder->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + int head = nouveau_crtc(encoder->crtc)->index; + struct nv04_crtc_reg *crtcstate = dev_priv->mode_reg.crtc_reg; + uint8_t *cr_lcd = &crtcstate[head].CRTC[NV_CIO_CRE_LCD__INDEX]; + uint8_t *cr_lcd_oth = &crtcstate[head ^ 1].CRTC[NV_CIO_CRE_LCD__INDEX]; + + helper->dpms(encoder, DRM_MODE_DPMS_OFF); + + nv04_dfp_prepare_sel_clk(dev, nv_encoder, head); + + *cr_lcd = (*cr_lcd & ~NV_CIO_CRE_LCD_ROUTE_MASK) | 0x3; + + if (nv_two_heads(dev)) { + if (nv_encoder->dcb->location == DCB_LOC_ON_CHIP) + *cr_lcd |= head ? 0x0 : 0x8; + else { + *cr_lcd |= (nv_encoder->dcb->or << 4) & 0x30; + if (nv_encoder->dcb->type == OUTPUT_LVDS) + *cr_lcd |= 0x30; + if ((*cr_lcd & 0x30) == (*cr_lcd_oth & 0x30)) { + /* avoid being connected to both crtcs */ + *cr_lcd_oth &= ~0x30; + NVWriteVgaCrtc(dev, head ^ 1, + NV_CIO_CRE_LCD__INDEX, + *cr_lcd_oth); + } + } + } +} + + +static void nv04_dfp_mode_set(struct drm_encoder *encoder, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct drm_device *dev = encoder->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_crtc *nv_crtc = nouveau_crtc(encoder->crtc); + struct nv04_crtc_reg *regp = &dev_priv->mode_reg.crtc_reg[nv_crtc->index]; + struct nv04_crtc_reg *savep = &dev_priv->saved_reg.crtc_reg[nv_crtc->index]; + struct nouveau_connector *nv_connector = nouveau_crtc_connector_get(nv_crtc); + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + struct drm_display_mode *output_mode = &nv_encoder->mode; + uint32_t mode_ratio, panel_ratio; + + NV_DEBUG_KMS(dev, "Output mode on CRTC %d:\n", nv_crtc->index); + drm_mode_debug_printmodeline(output_mode); + + /* Initialize the FP registers in this CRTC. */ + regp->fp_horiz_regs[FP_DISPLAY_END] = output_mode->hdisplay - 1; + regp->fp_horiz_regs[FP_TOTAL] = output_mode->htotal - 1; + if (!nv_gf4_disp_arch(dev) || + (output_mode->hsync_start - output_mode->hdisplay) >= + dev_priv->vbios.digital_min_front_porch) + regp->fp_horiz_regs[FP_CRTC] = output_mode->hdisplay; + else + regp->fp_horiz_regs[FP_CRTC] = output_mode->hsync_start - dev_priv->vbios.digital_min_front_porch - 1; + regp->fp_horiz_regs[FP_SYNC_START] = output_mode->hsync_start - 1; + regp->fp_horiz_regs[FP_SYNC_END] = output_mode->hsync_end - 1; + regp->fp_horiz_regs[FP_VALID_START] = output_mode->hskew; + regp->fp_horiz_regs[FP_VALID_END] = output_mode->hdisplay - 1; + + regp->fp_vert_regs[FP_DISPLAY_END] = output_mode->vdisplay - 1; + regp->fp_vert_regs[FP_TOTAL] = output_mode->vtotal - 1; + regp->fp_vert_regs[FP_CRTC] = output_mode->vtotal - 5 - 1; + regp->fp_vert_regs[FP_SYNC_START] = output_mode->vsync_start - 1; + regp->fp_vert_regs[FP_SYNC_END] = output_mode->vsync_end - 1; + regp->fp_vert_regs[FP_VALID_START] = 0; + regp->fp_vert_regs[FP_VALID_END] = output_mode->vdisplay - 1; + + /* bit26: a bit seen on some g7x, no as yet discernable purpose */ + regp->fp_control = NV_PRAMDAC_FP_TG_CONTROL_DISPEN_POS | + (savep->fp_control & (1 << 26 | NV_PRAMDAC_FP_TG_CONTROL_READ_PROG)); + /* Deal with vsync/hsync polarity */ + /* LVDS screens do set this, but modes with +ve syncs are very rare */ + if (output_mode->flags & DRM_MODE_FLAG_PVSYNC) + regp->fp_control |= NV_PRAMDAC_FP_TG_CONTROL_VSYNC_POS; + if (output_mode->flags & DRM_MODE_FLAG_PHSYNC) + regp->fp_control |= NV_PRAMDAC_FP_TG_CONTROL_HSYNC_POS; + /* panel scaling first, as native would get set otherwise */ + if (nv_connector->scaling_mode == DRM_MODE_SCALE_NONE || + nv_connector->scaling_mode == DRM_MODE_SCALE_CENTER) /* panel handles it */ + regp->fp_control |= NV_PRAMDAC_FP_TG_CONTROL_MODE_CENTER; + else if (adjusted_mode->hdisplay == output_mode->hdisplay && + adjusted_mode->vdisplay == output_mode->vdisplay) /* native mode */ + regp->fp_control |= NV_PRAMDAC_FP_TG_CONTROL_MODE_NATIVE; + else /* gpu needs to scale */ + regp->fp_control |= NV_PRAMDAC_FP_TG_CONTROL_MODE_SCALE; + if (nvReadEXTDEV(dev, NV_PEXTDEV_BOOT_0) & NV_PEXTDEV_BOOT_0_STRAP_FP_IFACE_12BIT) + regp->fp_control |= NV_PRAMDAC_FP_TG_CONTROL_WIDTH_12; + if (nv_encoder->dcb->location != DCB_LOC_ON_CHIP && + output_mode->clock > 165000) + regp->fp_control |= (2 << 24); + if (nv_encoder->dcb->type == OUTPUT_LVDS) { + bool duallink, dummy; + + nouveau_bios_parse_lvds_table(dev, output_mode->clock, + &duallink, &dummy); + if (duallink) + regp->fp_control |= (8 << 28); + } else + if (output_mode->clock > 165000) + regp->fp_control |= (8 << 28); + + regp->fp_debug_0 = NV_PRAMDAC_FP_DEBUG_0_YWEIGHT_ROUND | + NV_PRAMDAC_FP_DEBUG_0_XWEIGHT_ROUND | + NV_PRAMDAC_FP_DEBUG_0_YINTERP_BILINEAR | + NV_PRAMDAC_FP_DEBUG_0_XINTERP_BILINEAR | + NV_RAMDAC_FP_DEBUG_0_TMDS_ENABLED | + NV_PRAMDAC_FP_DEBUG_0_YSCALE_ENABLE | + NV_PRAMDAC_FP_DEBUG_0_XSCALE_ENABLE; + + /* We want automatic scaling */ + regp->fp_debug_1 = 0; + /* This can override HTOTAL and VTOTAL */ + regp->fp_debug_2 = 0; + + /* Use 20.12 fixed point format to avoid floats */ + mode_ratio = (1 << 12) * adjusted_mode->hdisplay / adjusted_mode->vdisplay; + panel_ratio = (1 << 12) * output_mode->hdisplay / output_mode->vdisplay; + /* if ratios are equal, SCALE_ASPECT will automatically (and correctly) + * get treated the same as SCALE_FULLSCREEN */ + if (nv_connector->scaling_mode == DRM_MODE_SCALE_ASPECT && + mode_ratio != panel_ratio) { + uint32_t diff, scale; + bool divide_by_2 = nv_gf4_disp_arch(dev); + + if (mode_ratio < panel_ratio) { + /* vertical needs to expand to glass size (automatic) + * horizontal needs to be scaled at vertical scale factor + * to maintain aspect */ + + scale = (1 << 12) * adjusted_mode->vdisplay / output_mode->vdisplay; + regp->fp_debug_1 = NV_PRAMDAC_FP_DEBUG_1_XSCALE_TESTMODE_ENABLE | + XLATE(scale, divide_by_2, NV_PRAMDAC_FP_DEBUG_1_XSCALE_VALUE); + + /* restrict area of screen used, horizontally */ + diff = output_mode->hdisplay - + output_mode->vdisplay * mode_ratio / (1 << 12); + regp->fp_horiz_regs[FP_VALID_START] += diff / 2; + regp->fp_horiz_regs[FP_VALID_END] -= diff / 2; + } + + if (mode_ratio > panel_ratio) { + /* horizontal needs to expand to glass size (automatic) + * vertical needs to be scaled at horizontal scale factor + * to maintain aspect */ + + scale = (1 << 12) * adjusted_mode->hdisplay / output_mode->hdisplay; + regp->fp_debug_1 = NV_PRAMDAC_FP_DEBUG_1_YSCALE_TESTMODE_ENABLE | + XLATE(scale, divide_by_2, NV_PRAMDAC_FP_DEBUG_1_YSCALE_VALUE); + + /* restrict area of screen used, vertically */ + diff = output_mode->vdisplay - + (1 << 12) * output_mode->hdisplay / mode_ratio; + regp->fp_vert_regs[FP_VALID_START] += diff / 2; + regp->fp_vert_regs[FP_VALID_END] -= diff / 2; + } + } + + /* Output property. */ + if (nv_connector->use_dithering) { + if (dev_priv->chipset == 0x11) + regp->dither = savep->dither | 0x00010000; + else { + int i; + regp->dither = savep->dither | 0x00000001; + for (i = 0; i < 3; i++) { + regp->dither_regs[i] = 0xe4e4e4e4; + regp->dither_regs[i + 3] = 0x44444444; + } + } + } else { + if (dev_priv->chipset != 0x11) { + /* reset them */ + int i; + for (i = 0; i < 3; i++) { + regp->dither_regs[i] = savep->dither_regs[i]; + regp->dither_regs[i + 3] = savep->dither_regs[i + 3]; + } + } + regp->dither = savep->dither; + } + + regp->fp_margin_color = 0; +} + +static void nv04_dfp_commit(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct drm_encoder_helper_funcs *helper = encoder->helper_private; + struct nouveau_crtc *nv_crtc = nouveau_crtc(encoder->crtc); + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + struct dcb_entry *dcbe = nv_encoder->dcb; + int head = nouveau_crtc(encoder->crtc)->index; + struct drm_encoder *slave_encoder; + + if (dcbe->type == OUTPUT_TMDS) + run_tmds_table(dev, dcbe, head, nv_encoder->mode.clock); + else if (dcbe->type == OUTPUT_LVDS) + call_lvds_script(dev, dcbe, head, LVDS_RESET, nv_encoder->mode.clock); + + /* update fp_control state for any changes made by scripts, + * so correct value is written at DPMS on */ + dev_priv->mode_reg.crtc_reg[head].fp_control = + NVReadRAMDAC(dev, head, NV_PRAMDAC_FP_TG_CONTROL); + + /* This could use refinement for flatpanels, but it should work this way */ + if (dev_priv->chipset < 0x44) + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_TEST_CONTROL + nv04_dac_output_offset(encoder), 0xf0000000); + else + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_TEST_CONTROL + nv04_dac_output_offset(encoder), 0x00100000); + + /* Init external transmitters */ + slave_encoder = get_tmds_slave(encoder); + if (slave_encoder) + get_slave_funcs(slave_encoder)->mode_set( + slave_encoder, &nv_encoder->mode, &nv_encoder->mode); + + helper->dpms(encoder, DRM_MODE_DPMS_ON); + + NV_INFO(dev, "Output %s is running on CRTC %d using output %c\n", + drm_get_connector_name(&nouveau_encoder_connector_get(nv_encoder)->base), + nv_crtc->index, '@' + ffs(nv_encoder->dcb->or)); +} + +static void nv04_dfp_update_backlight(struct drm_encoder *encoder, int mode) +{ +#ifdef __powerpc__ + struct drm_device *dev = encoder->dev; + + /* BIOS scripts usually take care of the backlight, thanks + * Apple for your consistency. + */ + if (dev->pci_device == 0x0179 || dev->pci_device == 0x0189 || + dev->pci_device == 0x0329) { + if (mode == DRM_MODE_DPMS_ON) { + nv_mask(dev, NV_PBUS_DEBUG_DUALHEAD_CTL, 0, 1 << 31); + nv_mask(dev, NV_PCRTC_GPIO_EXT, 3, 1); + } else { + nv_mask(dev, NV_PBUS_DEBUG_DUALHEAD_CTL, 1 << 31, 0); + nv_mask(dev, NV_PCRTC_GPIO_EXT, 3, 0); + } + } +#endif +} + +static inline bool is_powersaving_dpms(int mode) +{ + return (mode != DRM_MODE_DPMS_ON); +} + +static void nv04_lvds_dpms(struct drm_encoder *encoder, int mode) +{ + struct drm_device *dev = encoder->dev; + struct drm_crtc *crtc = encoder->crtc; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + bool was_powersaving = is_powersaving_dpms(nv_encoder->last_dpms); + + if (nv_encoder->last_dpms == mode) + return; + nv_encoder->last_dpms = mode; + + NV_INFO(dev, "Setting dpms mode %d on lvds encoder (output %d)\n", + mode, nv_encoder->dcb->index); + + if (was_powersaving && is_powersaving_dpms(mode)) + return; + + if (nv_encoder->dcb->lvdsconf.use_power_scripts) { + /* when removing an output, crtc may not be set, but PANEL_OFF + * must still be run + */ + int head = crtc ? nouveau_crtc(crtc)->index : + nv04_dfp_get_bound_head(dev, nv_encoder->dcb); + + if (mode == DRM_MODE_DPMS_ON) { + call_lvds_script(dev, nv_encoder->dcb, head, + LVDS_PANEL_ON, nv_encoder->mode.clock); + } else + /* pxclk of 0 is fine for PANEL_OFF, and for a + * disconnected LVDS encoder there is no native_mode + */ + call_lvds_script(dev, nv_encoder->dcb, head, + LVDS_PANEL_OFF, 0); + } + + nv04_dfp_update_backlight(encoder, mode); + nv04_dfp_update_fp_control(encoder, mode); + + if (mode == DRM_MODE_DPMS_ON) + nv04_dfp_prepare_sel_clk(dev, nv_encoder, nouveau_crtc(crtc)->index); + else { + dev_priv->mode_reg.sel_clk = NVReadRAMDAC(dev, 0, NV_PRAMDAC_SEL_CLK); + dev_priv->mode_reg.sel_clk &= ~0xf0; + } + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_SEL_CLK, dev_priv->mode_reg.sel_clk); +} + +static void nv04_tmds_dpms(struct drm_encoder *encoder, int mode) +{ + struct drm_device *dev = encoder->dev; + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + + if (nv_encoder->last_dpms == mode) + return; + nv_encoder->last_dpms = mode; + + NV_INFO(dev, "Setting dpms mode %d on tmds encoder (output %d)\n", + mode, nv_encoder->dcb->index); + + nv04_dfp_update_backlight(encoder, mode); + nv04_dfp_update_fp_control(encoder, mode); +} + +static void nv04_dfp_save(struct drm_encoder *encoder) +{ + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + struct drm_device *dev = encoder->dev; + + if (nv_two_heads(dev)) + nv_encoder->restore.head = + nv04_dfp_get_bound_head(dev, nv_encoder->dcb); +} + +static void nv04_dfp_restore(struct drm_encoder *encoder) +{ + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + struct drm_device *dev = encoder->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + int head = nv_encoder->restore.head; + + if (nv_encoder->dcb->type == OUTPUT_LVDS) { + struct nouveau_connector *connector = + nouveau_encoder_connector_get(nv_encoder); + + if (connector && connector->native_mode) + call_lvds_script(dev, nv_encoder->dcb, head, + LVDS_PANEL_ON, + connector->native_mode->clock); + + } else if (nv_encoder->dcb->type == OUTPUT_TMDS) { + int clock = nouveau_hw_pllvals_to_clk + (&dev_priv->saved_reg.crtc_reg[head].pllvals); + + run_tmds_table(dev, nv_encoder->dcb, head, clock); + } + + nv_encoder->last_dpms = NV_DPMS_CLEARED; +} + +static void nv04_dfp_destroy(struct drm_encoder *encoder) +{ + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + + NV_DEBUG_KMS(encoder->dev, "\n"); + + if (get_slave_funcs(encoder)) + get_slave_funcs(encoder)->destroy(encoder); + + drm_encoder_cleanup(encoder); + kfree(nv_encoder); +} + +static void nv04_tmds_slave_init(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct dcb_entry *dcb = nouveau_encoder(encoder)->dcb; + struct nouveau_i2c_chan *i2c = nouveau_i2c_find(dev, 2); + struct i2c_board_info info[] = { + { + .type = "sil164", + .addr = (dcb->tmdsconf.slave_addr == 0x7 ? 0x3a : 0x38), + .platform_data = &(struct sil164_encoder_params) { + SIL164_INPUT_EDGE_RISING + } + }, + { } + }; + int type; + + if (!nv_gf4_disp_arch(dev) || !i2c || + get_tmds_slave(encoder)) + return; + + type = nouveau_i2c_identify(dev, "TMDS transmitter", info, NULL, 2); + if (type < 0) + return; + + drm_i2c_encoder_init(dev, to_encoder_slave(encoder), + &i2c->adapter, &info[type]); +} + +static const struct drm_encoder_helper_funcs nv04_lvds_helper_funcs = { + .dpms = nv04_lvds_dpms, + .save = nv04_dfp_save, + .restore = nv04_dfp_restore, + .mode_fixup = nv04_dfp_mode_fixup, + .prepare = nv04_dfp_prepare, + .commit = nv04_dfp_commit, + .mode_set = nv04_dfp_mode_set, + .detect = NULL, +}; + +static const struct drm_encoder_helper_funcs nv04_tmds_helper_funcs = { + .dpms = nv04_tmds_dpms, + .save = nv04_dfp_save, + .restore = nv04_dfp_restore, + .mode_fixup = nv04_dfp_mode_fixup, + .prepare = nv04_dfp_prepare, + .commit = nv04_dfp_commit, + .mode_set = nv04_dfp_mode_set, + .detect = NULL, +}; + +static const struct drm_encoder_funcs nv04_dfp_funcs = { + .destroy = nv04_dfp_destroy, +}; + +int +nv04_dfp_create(struct drm_connector *connector, struct dcb_entry *entry) +{ + const struct drm_encoder_helper_funcs *helper; + struct nouveau_encoder *nv_encoder = NULL; + struct drm_encoder *encoder; + int type; + + switch (entry->type) { + case OUTPUT_TMDS: + type = DRM_MODE_ENCODER_TMDS; + helper = &nv04_tmds_helper_funcs; + break; + case OUTPUT_LVDS: + type = DRM_MODE_ENCODER_LVDS; + helper = &nv04_lvds_helper_funcs; + break; + default: + return -EINVAL; + } + + nv_encoder = kzalloc(sizeof(*nv_encoder), GFP_KERNEL); + if (!nv_encoder) + return -ENOMEM; + + encoder = to_drm_encoder(nv_encoder); + + nv_encoder->dcb = entry; + nv_encoder->or = ffs(entry->or) - 1; + + drm_encoder_init(connector->dev, encoder, &nv04_dfp_funcs, type); + drm_encoder_helper_add(encoder, helper); + + encoder->possible_crtcs = entry->heads; + encoder->possible_clones = 0; + + if (entry->type == OUTPUT_TMDS && + entry->location != DCB_LOC_ON_CHIP) + nv04_tmds_slave_init(encoder); + + drm_mode_connector_attach_encoder(connector, encoder); + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nv04_display.c b/drivers/gpu/drm/nouveau/nv04_display.c new file mode 100644 index 00000000..1715e146 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv04_display.c @@ -0,0 +1,281 @@ +/* + * Copyright 2009 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Author: Ben Skeggs + */ + +#include "drmP.h" +#include "drm.h" +#include "drm_crtc_helper.h" + +#include "nouveau_drv.h" +#include "nouveau_fb.h" +#include "nouveau_hw.h" +#include "nouveau_encoder.h" +#include "nouveau_connector.h" + +static void nv04_vblank_crtc0_isr(struct drm_device *); +static void nv04_vblank_crtc1_isr(struct drm_device *); + +static void +nv04_display_store_initial_head_owner(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + if (dev_priv->chipset != 0x11) { + dev_priv->crtc_owner = NVReadVgaCrtc(dev, 0, NV_CIO_CRE_44); + return; + } + + /* reading CR44 is broken on nv11, so we attempt to infer it */ + if (nvReadMC(dev, NV_PBUS_DEBUG_1) & (1 << 28)) /* heads tied, restore both */ + dev_priv->crtc_owner = 0x4; + else { + uint8_t slaved_on_A, slaved_on_B; + bool tvA = false; + bool tvB = false; + + slaved_on_B = NVReadVgaCrtc(dev, 1, NV_CIO_CRE_PIXEL_INDEX) & + 0x80; + if (slaved_on_B) + tvB = !(NVReadVgaCrtc(dev, 1, NV_CIO_CRE_LCD__INDEX) & + MASK(NV_CIO_CRE_LCD_LCD_SELECT)); + + slaved_on_A = NVReadVgaCrtc(dev, 0, NV_CIO_CRE_PIXEL_INDEX) & + 0x80; + if (slaved_on_A) + tvA = !(NVReadVgaCrtc(dev, 0, NV_CIO_CRE_LCD__INDEX) & + MASK(NV_CIO_CRE_LCD_LCD_SELECT)); + + if (slaved_on_A && !tvA) + dev_priv->crtc_owner = 0x0; + else if (slaved_on_B && !tvB) + dev_priv->crtc_owner = 0x3; + else if (slaved_on_A) + dev_priv->crtc_owner = 0x0; + else if (slaved_on_B) + dev_priv->crtc_owner = 0x3; + else + dev_priv->crtc_owner = 0x0; + } +} + +int +nv04_display_early_init(struct drm_device *dev) +{ + /* Make the I2C buses accessible. */ + if (!nv_gf4_disp_arch(dev)) { + uint32_t pmc_enable = nv_rd32(dev, NV03_PMC_ENABLE); + + if (!(pmc_enable & 1)) + nv_wr32(dev, NV03_PMC_ENABLE, pmc_enable | 1); + } + + /* Unlock the VGA CRTCs. */ + NVLockVgaCrtcs(dev, false); + + /* Make sure the CRTCs aren't in slaved mode. */ + if (nv_two_heads(dev)) { + nv04_display_store_initial_head_owner(dev); + NVSetOwner(dev, 0); + } + + return 0; +} + +void +nv04_display_late_takedown(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + if (nv_two_heads(dev)) + NVSetOwner(dev, dev_priv->crtc_owner); + + NVLockVgaCrtcs(dev, true); +} + +int +nv04_display_create(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct dcb_table *dcb = &dev_priv->vbios.dcb; + struct drm_connector *connector, *ct; + struct drm_encoder *encoder; + struct drm_crtc *crtc; + int i, ret; + + NV_DEBUG_KMS(dev, "\n"); + + nouveau_hw_save_vga_fonts(dev, 1); + + drm_mode_config_init(dev); + drm_mode_create_scaling_mode_property(dev); + drm_mode_create_dithering_property(dev); + + dev->mode_config.funcs = (void *)&nouveau_mode_config_funcs; + + dev->mode_config.min_width = 0; + dev->mode_config.min_height = 0; + switch (dev_priv->card_type) { + case NV_04: + dev->mode_config.max_width = 2048; + dev->mode_config.max_height = 2048; + break; + default: + dev->mode_config.max_width = 4096; + dev->mode_config.max_height = 4096; + break; + } + + dev->mode_config.fb_base = dev_priv->fb_phys; + + nv04_crtc_create(dev, 0); + if (nv_two_heads(dev)) + nv04_crtc_create(dev, 1); + + for (i = 0; i < dcb->entries; i++) { + struct dcb_entry *dcbent = &dcb->entry[i]; + + connector = nouveau_connector_create(dev, dcbent->connector); + if (IS_ERR(connector)) + continue; + + switch (dcbent->type) { + case OUTPUT_ANALOG: + ret = nv04_dac_create(connector, dcbent); + break; + case OUTPUT_LVDS: + case OUTPUT_TMDS: + ret = nv04_dfp_create(connector, dcbent); + break; + case OUTPUT_TV: + if (dcbent->location == DCB_LOC_ON_CHIP) + ret = nv17_tv_create(connector, dcbent); + else + ret = nv04_tv_create(connector, dcbent); + break; + default: + NV_WARN(dev, "DCB type %d not known\n", dcbent->type); + continue; + } + + if (ret) + continue; + } + + list_for_each_entry_safe(connector, ct, + &dev->mode_config.connector_list, head) { + if (!connector->encoder_ids[0]) { + NV_WARN(dev, "%s has no encoders, removing\n", + drm_get_connector_name(connector)); + connector->funcs->destroy(connector); + } + } + + /* Save previous state */ + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) + crtc->funcs->save(crtc); + + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + struct drm_encoder_helper_funcs *func = encoder->helper_private; + + func->save(encoder); + } + + nouveau_irq_register(dev, 24, nv04_vblank_crtc0_isr); + nouveau_irq_register(dev, 25, nv04_vblank_crtc1_isr); + return 0; +} + +void +nv04_display_destroy(struct drm_device *dev) +{ + struct drm_encoder *encoder; + struct drm_crtc *crtc; + + NV_DEBUG_KMS(dev, "\n"); + + nouveau_irq_unregister(dev, 24); + nouveau_irq_unregister(dev, 25); + + /* Turn every CRTC off. */ + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { + struct drm_mode_set modeset = { + .crtc = crtc, + }; + + crtc->funcs->set_config(&modeset); + } + + /* Restore state */ + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + struct drm_encoder_helper_funcs *func = encoder->helper_private; + + func->restore(encoder); + } + + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) + crtc->funcs->restore(crtc); + + drm_mode_config_cleanup(dev); + + nouveau_hw_save_vga_fonts(dev, 0); +} + +int +nv04_display_init(struct drm_device *dev) +{ + struct drm_encoder *encoder; + struct drm_crtc *crtc; + + /* meh.. modeset apparently doesn't setup all the regs and depends + * on pre-existing state, for now load the state of the card *before* + * nouveau was loaded, and then do a modeset. + * + * best thing to do probably is to make save/restore routines not + * save/restore "pre-load" state, but more general so we can save + * on suspend too. + */ + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + struct drm_encoder_helper_funcs *func = encoder->helper_private; + + func->restore(encoder); + } + + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) + crtc->funcs->restore(crtc); + + return 0; +} + +static void +nv04_vblank_crtc0_isr(struct drm_device *dev) +{ + nv_wr32(dev, NV_CRTC0_INTSTAT, NV_CRTC_INTR_VBLANK); + drm_handle_vblank(dev, 0); +} + +static void +nv04_vblank_crtc1_isr(struct drm_device *dev) +{ + nv_wr32(dev, NV_CRTC1_INTSTAT, NV_CRTC_INTR_VBLANK); + drm_handle_vblank(dev, 1); +} diff --git a/drivers/gpu/drm/nouveau/nv04_fb.c b/drivers/gpu/drm/nouveau/nv04_fb.c new file mode 100644 index 00000000..638cf601 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv04_fb.c @@ -0,0 +1,21 @@ +#include "drmP.h" +#include "drm.h" +#include "nouveau_drv.h" +#include "nouveau_drm.h" + +int +nv04_fb_init(struct drm_device *dev) +{ + /* This is what the DDX did for NV_ARCH_04, but a mmio-trace shows + * nvidia reading PFB_CFG_0, then writing back its original value. + * (which was 0x701114 in this case) + */ + + nv_wr32(dev, NV04_PFB_CFG0, 0x1114); + return 0; +} + +void +nv04_fb_takedown(struct drm_device *dev) +{ +} diff --git a/drivers/gpu/drm/nouveau/nv04_fbcon.c b/drivers/gpu/drm/nouveau/nv04_fbcon.c new file mode 100644 index 00000000..7a118937 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv04_fbcon.c @@ -0,0 +1,275 @@ +/* + * Copyright 2009 Ben Skeggs + * Copyright 2008 Stuart Bennett + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + */ + +#include "drmP.h" +#include "nouveau_drv.h" +#include "nouveau_dma.h" +#include "nouveau_ramht.h" +#include "nouveau_fbcon.h" + +int +nv04_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *region) +{ + struct nouveau_fbdev *nfbdev = info->par; + struct drm_device *dev = nfbdev->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_channel *chan = dev_priv->channel; + int ret; + + ret = RING_SPACE(chan, 4); + if (ret) + return ret; + + BEGIN_RING(chan, NvSubImageBlit, 0x0300, 3); + OUT_RING(chan, (region->sy << 16) | region->sx); + OUT_RING(chan, (region->dy << 16) | region->dx); + OUT_RING(chan, (region->height << 16) | region->width); + FIRE_RING(chan); + return 0; +} + +int +nv04_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect) +{ + struct nouveau_fbdev *nfbdev = info->par; + struct drm_device *dev = nfbdev->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_channel *chan = dev_priv->channel; + int ret; + + ret = RING_SPACE(chan, 7); + if (ret) + return ret; + + BEGIN_RING(chan, NvSubGdiRect, 0x02fc, 1); + OUT_RING(chan, (rect->rop != ROP_COPY) ? 1 : 3); + BEGIN_RING(chan, NvSubGdiRect, 0x03fc, 1); + if (info->fix.visual == FB_VISUAL_TRUECOLOR || + info->fix.visual == FB_VISUAL_DIRECTCOLOR) + OUT_RING(chan, ((uint32_t *)info->pseudo_palette)[rect->color]); + else + OUT_RING(chan, rect->color); + BEGIN_RING(chan, NvSubGdiRect, 0x0400, 2); + OUT_RING(chan, (rect->dx << 16) | rect->dy); + OUT_RING(chan, (rect->width << 16) | rect->height); + FIRE_RING(chan); + return 0; +} + +int +nv04_fbcon_imageblit(struct fb_info *info, const struct fb_image *image) +{ + struct nouveau_fbdev *nfbdev = info->par; + struct drm_device *dev = nfbdev->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_channel *chan = dev_priv->channel; + uint32_t fg; + uint32_t bg; + uint32_t dsize; + uint32_t width; + uint32_t *data = (uint32_t *)image->data; + int ret; + + if (image->depth != 1) + return -ENODEV; + + ret = RING_SPACE(chan, 8); + if (ret) + return ret; + + width = ALIGN(image->width, 8); + dsize = ALIGN(width * image->height, 32) >> 5; + + if (info->fix.visual == FB_VISUAL_TRUECOLOR || + info->fix.visual == FB_VISUAL_DIRECTCOLOR) { + fg = ((uint32_t *) info->pseudo_palette)[image->fg_color]; + bg = ((uint32_t *) info->pseudo_palette)[image->bg_color]; + } else { + fg = image->fg_color; + bg = image->bg_color; + } + + BEGIN_RING(chan, NvSubGdiRect, 0x0be4, 7); + OUT_RING(chan, (image->dy << 16) | (image->dx & 0xffff)); + OUT_RING(chan, ((image->dy + image->height) << 16) | + ((image->dx + image->width) & 0xffff)); + OUT_RING(chan, bg); + OUT_RING(chan, fg); + OUT_RING(chan, (image->height << 16) | width); + OUT_RING(chan, (image->height << 16) | image->width); + OUT_RING(chan, (image->dy << 16) | (image->dx & 0xffff)); + + while (dsize) { + int iter_len = dsize > 128 ? 128 : dsize; + + ret = RING_SPACE(chan, iter_len + 1); + if (ret) + return ret; + + BEGIN_RING(chan, NvSubGdiRect, 0x0c00, iter_len); + OUT_RINGp(chan, data, iter_len); + data += iter_len; + dsize -= iter_len; + } + + FIRE_RING(chan); + return 0; +} + +int +nv04_fbcon_accel_init(struct fb_info *info) +{ + struct nouveau_fbdev *nfbdev = info->par; + struct drm_device *dev = nfbdev->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_channel *chan = dev_priv->channel; + const int sub = NvSubCtxSurf2D; + int surface_fmt, pattern_fmt, rect_fmt; + int ret; + + switch (info->var.bits_per_pixel) { + case 8: + surface_fmt = 1; + pattern_fmt = 3; + rect_fmt = 3; + break; + case 16: + surface_fmt = 4; + pattern_fmt = 1; + rect_fmt = 1; + break; + case 32: + switch (info->var.transp.length) { + case 0: /* depth 24 */ + case 8: /* depth 32 */ + break; + default: + return -EINVAL; + } + + surface_fmt = 6; + pattern_fmt = 3; + rect_fmt = 3; + break; + default: + return -EINVAL; + } + + ret = nouveau_gpuobj_gr_new(chan, NvCtxSurf2D, + dev_priv->card_type >= NV_10 ? + 0x0062 : 0x0042); + if (ret) + return ret; + + ret = nouveau_gpuobj_gr_new(chan, NvClipRect, 0x0019); + if (ret) + return ret; + + ret = nouveau_gpuobj_gr_new(chan, NvRop, 0x0043); + if (ret) + return ret; + + ret = nouveau_gpuobj_gr_new(chan, NvImagePatt, 0x0044); + if (ret) + return ret; + + ret = nouveau_gpuobj_gr_new(chan, NvGdiRect, 0x004a); + if (ret) + return ret; + + ret = nouveau_gpuobj_gr_new(chan, NvImageBlit, + dev_priv->chipset >= 0x11 ? + 0x009f : 0x005f); + if (ret) + return ret; + + if (RING_SPACE(chan, 49)) { + nouveau_fbcon_gpu_lockup(info); + return 0; + } + + BEGIN_RING(chan, sub, 0x0000, 1); + OUT_RING(chan, NvCtxSurf2D); + BEGIN_RING(chan, sub, 0x0184, 2); + OUT_RING(chan, NvDmaFB); + OUT_RING(chan, NvDmaFB); + BEGIN_RING(chan, sub, 0x0300, 4); + OUT_RING(chan, surface_fmt); + OUT_RING(chan, info->fix.line_length | (info->fix.line_length << 16)); + OUT_RING(chan, info->fix.smem_start - dev->mode_config.fb_base); + OUT_RING(chan, info->fix.smem_start - dev->mode_config.fb_base); + + BEGIN_RING(chan, sub, 0x0000, 1); + OUT_RING(chan, NvRop); + BEGIN_RING(chan, sub, 0x0300, 1); + OUT_RING(chan, 0x55); + + BEGIN_RING(chan, sub, 0x0000, 1); + OUT_RING(chan, NvImagePatt); + BEGIN_RING(chan, sub, 0x0300, 8); + OUT_RING(chan, pattern_fmt); +#ifdef __BIG_ENDIAN + OUT_RING(chan, 2); +#else + OUT_RING(chan, 1); +#endif + OUT_RING(chan, 0); + OUT_RING(chan, 1); + OUT_RING(chan, ~0); + OUT_RING(chan, ~0); + OUT_RING(chan, ~0); + OUT_RING(chan, ~0); + + BEGIN_RING(chan, sub, 0x0000, 1); + OUT_RING(chan, NvClipRect); + BEGIN_RING(chan, sub, 0x0300, 2); + OUT_RING(chan, 0); + OUT_RING(chan, (info->var.yres_virtual << 16) | info->var.xres_virtual); + + BEGIN_RING(chan, NvSubImageBlit, 0x0000, 1); + OUT_RING(chan, NvImageBlit); + BEGIN_RING(chan, NvSubImageBlit, 0x019c, 1); + OUT_RING(chan, NvCtxSurf2D); + BEGIN_RING(chan, NvSubImageBlit, 0x02fc, 1); + OUT_RING(chan, 3); + + BEGIN_RING(chan, NvSubGdiRect, 0x0000, 1); + OUT_RING(chan, NvGdiRect); + BEGIN_RING(chan, NvSubGdiRect, 0x0198, 1); + OUT_RING(chan, NvCtxSurf2D); + BEGIN_RING(chan, NvSubGdiRect, 0x0188, 2); + OUT_RING(chan, NvImagePatt); + OUT_RING(chan, NvRop); + BEGIN_RING(chan, NvSubGdiRect, 0x0304, 1); + OUT_RING(chan, 1); + BEGIN_RING(chan, NvSubGdiRect, 0x0300, 1); + OUT_RING(chan, rect_fmt); + BEGIN_RING(chan, NvSubGdiRect, 0x02fc, 1); + OUT_RING(chan, 3); + + FIRE_RING(chan); + + return 0; +} + diff --git a/drivers/gpu/drm/nouveau/nv04_fifo.c b/drivers/gpu/drm/nouveau/nv04_fifo.c new file mode 100644 index 00000000..db465a3e --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv04_fifo.c @@ -0,0 +1,543 @@ +/* + * Copyright (C) 2007 Ben Skeggs. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "drmP.h" +#include "drm.h" +#include "nouveau_drv.h" +#include "nouveau_ramht.h" +#include "nouveau_util.h" + +#define NV04_RAMFC(c) (dev_priv->ramfc->pinst + ((c) * NV04_RAMFC__SIZE)) +#define NV04_RAMFC__SIZE 32 +#define NV04_RAMFC_DMA_PUT 0x00 +#define NV04_RAMFC_DMA_GET 0x04 +#define NV04_RAMFC_DMA_INSTANCE 0x08 +#define NV04_RAMFC_DMA_STATE 0x0C +#define NV04_RAMFC_DMA_FETCH 0x10 +#define NV04_RAMFC_ENGINE 0x14 +#define NV04_RAMFC_PULL1_ENGINE 0x18 + +#define RAMFC_WR(offset, val) nv_wo32(chan->ramfc, NV04_RAMFC_##offset, (val)) +#define RAMFC_RD(offset) nv_ro32(chan->ramfc, NV04_RAMFC_##offset) + +void +nv04_fifo_disable(struct drm_device *dev) +{ + uint32_t tmp; + + tmp = nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_PUSH); + nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_PUSH, tmp & ~1); + nv_wr32(dev, NV03_PFIFO_CACHE1_PUSH0, 0); + tmp = nv_rd32(dev, NV03_PFIFO_CACHE1_PULL1); + nv_wr32(dev, NV04_PFIFO_CACHE1_PULL0, tmp & ~1); +} + +void +nv04_fifo_enable(struct drm_device *dev) +{ + nv_wr32(dev, NV03_PFIFO_CACHE1_PUSH0, 1); + nv_wr32(dev, NV04_PFIFO_CACHE1_PULL0, 1); +} + +bool +nv04_fifo_reassign(struct drm_device *dev, bool enable) +{ + uint32_t reassign = nv_rd32(dev, NV03_PFIFO_CACHES); + + nv_wr32(dev, NV03_PFIFO_CACHES, enable ? 1 : 0); + return (reassign == 1); +} + +bool +nv04_fifo_cache_pull(struct drm_device *dev, bool enable) +{ + int pull = nv_mask(dev, NV04_PFIFO_CACHE1_PULL0, 1, enable); + + if (!enable) { + /* In some cases the PFIFO puller may be left in an + * inconsistent state if you try to stop it when it's + * busy translating handles. Sometimes you get a + * PFIFO_CACHE_ERROR, sometimes it just fails silently + * sending incorrect instance offsets to PGRAPH after + * it's started up again. To avoid the latter we + * invalidate the most recently calculated instance. + */ + if (!nv_wait(dev, NV04_PFIFO_CACHE1_PULL0, + NV04_PFIFO_CACHE1_PULL0_HASH_BUSY, 0)) + NV_ERROR(dev, "Timeout idling the PFIFO puller.\n"); + + if (nv_rd32(dev, NV04_PFIFO_CACHE1_PULL0) & + NV04_PFIFO_CACHE1_PULL0_HASH_FAILED) + nv_wr32(dev, NV03_PFIFO_INTR_0, + NV_PFIFO_INTR_CACHE_ERROR); + + nv_wr32(dev, NV04_PFIFO_CACHE1_HASH, 0); + } + + return pull & 1; +} + +int +nv04_fifo_channel_id(struct drm_device *dev) +{ + return nv_rd32(dev, NV03_PFIFO_CACHE1_PUSH1) & + NV03_PFIFO_CACHE1_PUSH1_CHID_MASK; +} + +#ifdef __BIG_ENDIAN +#define DMA_FETCH_ENDIANNESS NV_PFIFO_CACHE1_BIG_ENDIAN +#else +#define DMA_FETCH_ENDIANNESS 0 +#endif + +int +nv04_fifo_create_context(struct nouveau_channel *chan) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + unsigned long flags; + int ret; + + ret = nouveau_gpuobj_new_fake(dev, NV04_RAMFC(chan->id), ~0, + NV04_RAMFC__SIZE, + NVOBJ_FLAG_ZERO_ALLOC | + NVOBJ_FLAG_ZERO_FREE, + &chan->ramfc); + if (ret) + return ret; + + chan->user = ioremap(pci_resource_start(dev->pdev, 0) + + NV03_USER(chan->id), PAGE_SIZE); + if (!chan->user) + return -ENOMEM; + + spin_lock_irqsave(&dev_priv->context_switch_lock, flags); + + /* Setup initial state */ + RAMFC_WR(DMA_PUT, chan->pushbuf_base); + RAMFC_WR(DMA_GET, chan->pushbuf_base); + RAMFC_WR(DMA_INSTANCE, chan->pushbuf->pinst >> 4); + RAMFC_WR(DMA_FETCH, (NV_PFIFO_CACHE1_DMA_FETCH_TRIG_128_BYTES | + NV_PFIFO_CACHE1_DMA_FETCH_SIZE_128_BYTES | + NV_PFIFO_CACHE1_DMA_FETCH_MAX_REQS_8 | + DMA_FETCH_ENDIANNESS)); + + /* enable the fifo dma operation */ + nv_wr32(dev, NV04_PFIFO_MODE, + nv_rd32(dev, NV04_PFIFO_MODE) | (1 << chan->id)); + + spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); + return 0; +} + +void +nv04_fifo_destroy_context(struct nouveau_channel *chan) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; + unsigned long flags; + + spin_lock_irqsave(&dev_priv->context_switch_lock, flags); + pfifo->reassign(dev, false); + + /* Unload the context if it's the currently active one */ + if (pfifo->channel_id(dev) == chan->id) { + pfifo->disable(dev); + pfifo->unload_context(dev); + pfifo->enable(dev); + } + + /* Keep it from being rescheduled */ + nv_mask(dev, NV04_PFIFO_MODE, 1 << chan->id, 0); + + pfifo->reassign(dev, true); + spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); + + /* Free the channel resources */ + if (chan->user) { + iounmap(chan->user); + chan->user = NULL; + } + nouveau_gpuobj_ref(NULL, &chan->ramfc); +} + +static void +nv04_fifo_do_load_context(struct drm_device *dev, int chid) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + uint32_t fc = NV04_RAMFC(chid), tmp; + + nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_PUT, nv_ri32(dev, fc + 0)); + nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_GET, nv_ri32(dev, fc + 4)); + tmp = nv_ri32(dev, fc + 8); + nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_INSTANCE, tmp & 0xFFFF); + nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_DCOUNT, tmp >> 16); + nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_STATE, nv_ri32(dev, fc + 12)); + nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_FETCH, nv_ri32(dev, fc + 16)); + nv_wr32(dev, NV04_PFIFO_CACHE1_ENGINE, nv_ri32(dev, fc + 20)); + nv_wr32(dev, NV04_PFIFO_CACHE1_PULL1, nv_ri32(dev, fc + 24)); + + nv_wr32(dev, NV03_PFIFO_CACHE1_GET, 0); + nv_wr32(dev, NV03_PFIFO_CACHE1_PUT, 0); +} + +int +nv04_fifo_load_context(struct nouveau_channel *chan) +{ + uint32_t tmp; + + nv_wr32(chan->dev, NV03_PFIFO_CACHE1_PUSH1, + NV03_PFIFO_CACHE1_PUSH1_DMA | chan->id); + nv04_fifo_do_load_context(chan->dev, chan->id); + nv_wr32(chan->dev, NV04_PFIFO_CACHE1_DMA_PUSH, 1); + + /* Reset NV04_PFIFO_CACHE1_DMA_CTL_AT_INFO to INVALID */ + tmp = nv_rd32(chan->dev, NV04_PFIFO_CACHE1_DMA_CTL) & ~(1 << 31); + nv_wr32(chan->dev, NV04_PFIFO_CACHE1_DMA_CTL, tmp); + + return 0; +} + +int +nv04_fifo_unload_context(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; + struct nouveau_channel *chan = NULL; + uint32_t tmp; + int chid; + + chid = pfifo->channel_id(dev); + if (chid < 0 || chid >= dev_priv->engine.fifo.channels) + return 0; + + chan = dev_priv->channels.ptr[chid]; + if (!chan) { + NV_ERROR(dev, "Inactive channel on PFIFO: %d\n", chid); + return -EINVAL; + } + + RAMFC_WR(DMA_PUT, nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_PUT)); + RAMFC_WR(DMA_GET, nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_GET)); + tmp = nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_DCOUNT) << 16; + tmp |= nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_INSTANCE); + RAMFC_WR(DMA_INSTANCE, tmp); + RAMFC_WR(DMA_STATE, nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_STATE)); + RAMFC_WR(DMA_FETCH, nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_FETCH)); + RAMFC_WR(ENGINE, nv_rd32(dev, NV04_PFIFO_CACHE1_ENGINE)); + RAMFC_WR(PULL1_ENGINE, nv_rd32(dev, NV04_PFIFO_CACHE1_PULL1)); + + nv04_fifo_do_load_context(dev, pfifo->channels - 1); + nv_wr32(dev, NV03_PFIFO_CACHE1_PUSH1, pfifo->channels - 1); + return 0; +} + +static void +nv04_fifo_init_reset(struct drm_device *dev) +{ + nv_wr32(dev, NV03_PMC_ENABLE, + nv_rd32(dev, NV03_PMC_ENABLE) & ~NV_PMC_ENABLE_PFIFO); + nv_wr32(dev, NV03_PMC_ENABLE, + nv_rd32(dev, NV03_PMC_ENABLE) | NV_PMC_ENABLE_PFIFO); + + nv_wr32(dev, 0x003224, 0x000f0078); + nv_wr32(dev, 0x002044, 0x0101ffff); + nv_wr32(dev, 0x002040, 0x000000ff); + nv_wr32(dev, 0x002500, 0x00000000); + nv_wr32(dev, 0x003000, 0x00000000); + nv_wr32(dev, 0x003050, 0x00000000); + nv_wr32(dev, 0x003200, 0x00000000); + nv_wr32(dev, 0x003250, 0x00000000); + nv_wr32(dev, 0x003220, 0x00000000); + + nv_wr32(dev, 0x003250, 0x00000000); + nv_wr32(dev, 0x003270, 0x00000000); + nv_wr32(dev, 0x003210, 0x00000000); +} + +static void +nv04_fifo_init_ramxx(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + nv_wr32(dev, NV03_PFIFO_RAMHT, (0x03 << 24) /* search 128 */ | + ((dev_priv->ramht->bits - 9) << 16) | + (dev_priv->ramht->gpuobj->pinst >> 8)); + nv_wr32(dev, NV03_PFIFO_RAMRO, dev_priv->ramro->pinst >> 8); + nv_wr32(dev, NV03_PFIFO_RAMFC, dev_priv->ramfc->pinst >> 8); +} + +static void +nv04_fifo_init_intr(struct drm_device *dev) +{ + nouveau_irq_register(dev, 8, nv04_fifo_isr); + nv_wr32(dev, 0x002100, 0xffffffff); + nv_wr32(dev, 0x002140, 0xffffffff); +} + +int +nv04_fifo_init(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; + int i; + + nv04_fifo_init_reset(dev); + nv04_fifo_init_ramxx(dev); + + nv04_fifo_do_load_context(dev, pfifo->channels - 1); + nv_wr32(dev, NV03_PFIFO_CACHE1_PUSH1, pfifo->channels - 1); + + nv04_fifo_init_intr(dev); + pfifo->enable(dev); + pfifo->reassign(dev, true); + + for (i = 0; i < dev_priv->engine.fifo.channels; i++) { + if (dev_priv->channels.ptr[i]) { + uint32_t mode = nv_rd32(dev, NV04_PFIFO_MODE); + nv_wr32(dev, NV04_PFIFO_MODE, mode | (1 << i)); + } + } + + return 0; +} + +void +nv04_fifo_fini(struct drm_device *dev) +{ + nv_wr32(dev, 0x2140, 0x00000000); + nouveau_irq_unregister(dev, 8); +} + +static bool +nouveau_fifo_swmthd(struct drm_device *dev, u32 chid, u32 addr, u32 data) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_channel *chan = NULL; + struct nouveau_gpuobj *obj; + unsigned long flags; + const int subc = (addr >> 13) & 0x7; + const int mthd = addr & 0x1ffc; + bool handled = false; + u32 engine; + + spin_lock_irqsave(&dev_priv->channels.lock, flags); + if (likely(chid >= 0 && chid < dev_priv->engine.fifo.channels)) + chan = dev_priv->channels.ptr[chid]; + if (unlikely(!chan)) + goto out; + + switch (mthd) { + case 0x0000: /* bind object to subchannel */ + obj = nouveau_ramht_find(chan, data); + if (unlikely(!obj || obj->engine != NVOBJ_ENGINE_SW)) + break; + + chan->sw_subchannel[subc] = obj->class; + engine = 0x0000000f << (subc * 4); + + nv_mask(dev, NV04_PFIFO_CACHE1_ENGINE, engine, 0x00000000); + handled = true; + break; + default: + engine = nv_rd32(dev, NV04_PFIFO_CACHE1_ENGINE); + if (unlikely(((engine >> (subc * 4)) & 0xf) != 0)) + break; + + if (!nouveau_gpuobj_mthd_call(chan, chan->sw_subchannel[subc], + mthd, data)) + handled = true; + break; + } + +out: + spin_unlock_irqrestore(&dev_priv->channels.lock, flags); + return handled; +} + +static const char *nv_dma_state_err(u32 state) +{ + static const char * const desc[] = { + "NONE", "CALL_SUBR_ACTIVE", "INVALID_MTHD", "RET_SUBR_INACTIVE", + "INVALID_CMD", "IB_EMPTY"/* NV50+ */, "MEM_FAULT", "UNK" + }; + return desc[(state >> 29) & 0x7]; +} + +void +nv04_fifo_isr(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_engine *engine = &dev_priv->engine; + uint32_t status, reassign; + int cnt = 0; + + reassign = nv_rd32(dev, NV03_PFIFO_CACHES) & 1; + while ((status = nv_rd32(dev, NV03_PFIFO_INTR_0)) && (cnt++ < 100)) { + uint32_t chid, get; + + nv_wr32(dev, NV03_PFIFO_CACHES, 0); + + chid = engine->fifo.channel_id(dev); + get = nv_rd32(dev, NV03_PFIFO_CACHE1_GET); + + if (status & NV_PFIFO_INTR_CACHE_ERROR) { + uint32_t mthd, data; + int ptr; + + /* NV_PFIFO_CACHE1_GET actually goes to 0xffc before + * wrapping on my G80 chips, but CACHE1 isn't big + * enough for this much data.. Tests show that it + * wraps around to the start at GET=0x800.. No clue + * as to why.. + */ + ptr = (get & 0x7ff) >> 2; + + if (dev_priv->card_type < NV_40) { + mthd = nv_rd32(dev, + NV04_PFIFO_CACHE1_METHOD(ptr)); + data = nv_rd32(dev, + NV04_PFIFO_CACHE1_DATA(ptr)); + } else { + mthd = nv_rd32(dev, + NV40_PFIFO_CACHE1_METHOD(ptr)); + data = nv_rd32(dev, + NV40_PFIFO_CACHE1_DATA(ptr)); + } + + if (!nouveau_fifo_swmthd(dev, chid, mthd, data)) { + NV_INFO(dev, "PFIFO_CACHE_ERROR - Ch %d/%d " + "Mthd 0x%04x Data 0x%08x\n", + chid, (mthd >> 13) & 7, mthd & 0x1ffc, + data); + } + + nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_PUSH, 0); + nv_wr32(dev, NV03_PFIFO_INTR_0, + NV_PFIFO_INTR_CACHE_ERROR); + + nv_wr32(dev, NV03_PFIFO_CACHE1_PUSH0, + nv_rd32(dev, NV03_PFIFO_CACHE1_PUSH0) & ~1); + nv_wr32(dev, NV03_PFIFO_CACHE1_GET, get + 4); + nv_wr32(dev, NV03_PFIFO_CACHE1_PUSH0, + nv_rd32(dev, NV03_PFIFO_CACHE1_PUSH0) | 1); + nv_wr32(dev, NV04_PFIFO_CACHE1_HASH, 0); + + nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_PUSH, + nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_PUSH) | 1); + nv_wr32(dev, NV04_PFIFO_CACHE1_PULL0, 1); + + status &= ~NV_PFIFO_INTR_CACHE_ERROR; + } + + if (status & NV_PFIFO_INTR_DMA_PUSHER) { + u32 dma_get = nv_rd32(dev, 0x003244); + u32 dma_put = nv_rd32(dev, 0x003240); + u32 push = nv_rd32(dev, 0x003220); + u32 state = nv_rd32(dev, 0x003228); + + if (dev_priv->card_type == NV_50) { + u32 ho_get = nv_rd32(dev, 0x003328); + u32 ho_put = nv_rd32(dev, 0x003320); + u32 ib_get = nv_rd32(dev, 0x003334); + u32 ib_put = nv_rd32(dev, 0x003330); + + if (nouveau_ratelimit()) + NV_INFO(dev, "PFIFO_DMA_PUSHER - Ch %d Get 0x%02x%08x " + "Put 0x%02x%08x IbGet 0x%08x IbPut 0x%08x " + "State 0x%08x (err: %s) Push 0x%08x\n", + chid, ho_get, dma_get, ho_put, + dma_put, ib_get, ib_put, state, + nv_dma_state_err(state), + push); + + /* METHOD_COUNT, in DMA_STATE on earlier chipsets */ + nv_wr32(dev, 0x003364, 0x00000000); + if (dma_get != dma_put || ho_get != ho_put) { + nv_wr32(dev, 0x003244, dma_put); + nv_wr32(dev, 0x003328, ho_put); + } else + if (ib_get != ib_put) { + nv_wr32(dev, 0x003334, ib_put); + } + } else { + NV_INFO(dev, "PFIFO_DMA_PUSHER - Ch %d Get 0x%08x " + "Put 0x%08x State 0x%08x (err: %s) Push 0x%08x\n", + chid, dma_get, dma_put, state, + nv_dma_state_err(state), push); + + if (dma_get != dma_put) + nv_wr32(dev, 0x003244, dma_put); + } + + nv_wr32(dev, 0x003228, 0x00000000); + nv_wr32(dev, 0x003220, 0x00000001); + nv_wr32(dev, 0x002100, NV_PFIFO_INTR_DMA_PUSHER); + status &= ~NV_PFIFO_INTR_DMA_PUSHER; + } + + if (status & NV_PFIFO_INTR_SEMAPHORE) { + uint32_t sem; + + status &= ~NV_PFIFO_INTR_SEMAPHORE; + nv_wr32(dev, NV03_PFIFO_INTR_0, + NV_PFIFO_INTR_SEMAPHORE); + + sem = nv_rd32(dev, NV10_PFIFO_CACHE1_SEMAPHORE); + nv_wr32(dev, NV10_PFIFO_CACHE1_SEMAPHORE, sem | 0x1); + + nv_wr32(dev, NV03_PFIFO_CACHE1_GET, get + 4); + nv_wr32(dev, NV04_PFIFO_CACHE1_PULL0, 1); + } + + if (dev_priv->card_type == NV_50) { + if (status & 0x00000010) { + nv50_fb_vm_trap(dev, nouveau_ratelimit()); + status &= ~0x00000010; + nv_wr32(dev, 0x002100, 0x00000010); + } + } + + if (status) { + if (nouveau_ratelimit()) + NV_INFO(dev, "PFIFO_INTR 0x%08x - Ch %d\n", + status, chid); + nv_wr32(dev, NV03_PFIFO_INTR_0, status); + status = 0; + } + + nv_wr32(dev, NV03_PFIFO_CACHES, reassign); + } + + if (status) { + NV_INFO(dev, "PFIFO still angry after %d spins, halt\n", cnt); + nv_wr32(dev, 0x2140, 0); + nv_wr32(dev, 0x140, 0); + } + + nv_wr32(dev, NV03_PMC_INTR_0, NV_PMC_INTR_0_PFIFO_PENDING); +} diff --git a/drivers/gpu/drm/nouveau/nv04_graph.c b/drivers/gpu/drm/nouveau/nv04_graph.c new file mode 100644 index 00000000..3626ee7d --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv04_graph.c @@ -0,0 +1,1359 @@ +/* + * Copyright 2007 Stephane Marchesin + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + */ + +#include "drmP.h" +#include "drm.h" +#include "nouveau_drm.h" +#include "nouveau_drv.h" +#include "nouveau_hw.h" +#include "nouveau_util.h" +#include "nouveau_ramht.h" + +struct nv04_graph_engine { + struct nouveau_exec_engine base; +}; + +static uint32_t nv04_graph_ctx_regs[] = { + 0x0040053c, + 0x00400544, + 0x00400540, + 0x00400548, + NV04_PGRAPH_CTX_SWITCH1, + NV04_PGRAPH_CTX_SWITCH2, + NV04_PGRAPH_CTX_SWITCH3, + NV04_PGRAPH_CTX_SWITCH4, + NV04_PGRAPH_CTX_CACHE1, + NV04_PGRAPH_CTX_CACHE2, + NV04_PGRAPH_CTX_CACHE3, + NV04_PGRAPH_CTX_CACHE4, + 0x00400184, + 0x004001a4, + 0x004001c4, + 0x004001e4, + 0x00400188, + 0x004001a8, + 0x004001c8, + 0x004001e8, + 0x0040018c, + 0x004001ac, + 0x004001cc, + 0x004001ec, + 0x00400190, + 0x004001b0, + 0x004001d0, + 0x004001f0, + 0x00400194, + 0x004001b4, + 0x004001d4, + 0x004001f4, + 0x00400198, + 0x004001b8, + 0x004001d8, + 0x004001f8, + 0x0040019c, + 0x004001bc, + 0x004001dc, + 0x004001fc, + 0x00400174, + NV04_PGRAPH_DMA_START_0, + NV04_PGRAPH_DMA_START_1, + NV04_PGRAPH_DMA_LENGTH, + NV04_PGRAPH_DMA_MISC, + NV04_PGRAPH_DMA_PITCH, + NV04_PGRAPH_BOFFSET0, + NV04_PGRAPH_BBASE0, + NV04_PGRAPH_BLIMIT0, + NV04_PGRAPH_BOFFSET1, + NV04_PGRAPH_BBASE1, + NV04_PGRAPH_BLIMIT1, + NV04_PGRAPH_BOFFSET2, + NV04_PGRAPH_BBASE2, + NV04_PGRAPH_BLIMIT2, + NV04_PGRAPH_BOFFSET3, + NV04_PGRAPH_BBASE3, + NV04_PGRAPH_BLIMIT3, + NV04_PGRAPH_BOFFSET4, + NV04_PGRAPH_BBASE4, + NV04_PGRAPH_BLIMIT4, + NV04_PGRAPH_BOFFSET5, + NV04_PGRAPH_BBASE5, + NV04_PGRAPH_BLIMIT5, + NV04_PGRAPH_BPITCH0, + NV04_PGRAPH_BPITCH1, + NV04_PGRAPH_BPITCH2, + NV04_PGRAPH_BPITCH3, + NV04_PGRAPH_BPITCH4, + NV04_PGRAPH_SURFACE, + NV04_PGRAPH_STATE, + NV04_PGRAPH_BSWIZZLE2, + NV04_PGRAPH_BSWIZZLE5, + NV04_PGRAPH_BPIXEL, + NV04_PGRAPH_NOTIFY, + NV04_PGRAPH_PATT_COLOR0, + NV04_PGRAPH_PATT_COLOR1, + NV04_PGRAPH_PATT_COLORRAM+0x00, + NV04_PGRAPH_PATT_COLORRAM+0x04, + NV04_PGRAPH_PATT_COLORRAM+0x08, + NV04_PGRAPH_PATT_COLORRAM+0x0c, + NV04_PGRAPH_PATT_COLORRAM+0x10, + NV04_PGRAPH_PATT_COLORRAM+0x14, + NV04_PGRAPH_PATT_COLORRAM+0x18, + NV04_PGRAPH_PATT_COLORRAM+0x1c, + NV04_PGRAPH_PATT_COLORRAM+0x20, + NV04_PGRAPH_PATT_COLORRAM+0x24, + NV04_PGRAPH_PATT_COLORRAM+0x28, + NV04_PGRAPH_PATT_COLORRAM+0x2c, + NV04_PGRAPH_PATT_COLORRAM+0x30, + NV04_PGRAPH_PATT_COLORRAM+0x34, + NV04_PGRAPH_PATT_COLORRAM+0x38, + NV04_PGRAPH_PATT_COLORRAM+0x3c, + NV04_PGRAPH_PATT_COLORRAM+0x40, + NV04_PGRAPH_PATT_COLORRAM+0x44, + NV04_PGRAPH_PATT_COLORRAM+0x48, + NV04_PGRAPH_PATT_COLORRAM+0x4c, + NV04_PGRAPH_PATT_COLORRAM+0x50, + NV04_PGRAPH_PATT_COLORRAM+0x54, + NV04_PGRAPH_PATT_COLORRAM+0x58, + NV04_PGRAPH_PATT_COLORRAM+0x5c, + NV04_PGRAPH_PATT_COLORRAM+0x60, + NV04_PGRAPH_PATT_COLORRAM+0x64, + NV04_PGRAPH_PATT_COLORRAM+0x68, + NV04_PGRAPH_PATT_COLORRAM+0x6c, + NV04_PGRAPH_PATT_COLORRAM+0x70, + NV04_PGRAPH_PATT_COLORRAM+0x74, + NV04_PGRAPH_PATT_COLORRAM+0x78, + NV04_PGRAPH_PATT_COLORRAM+0x7c, + NV04_PGRAPH_PATT_COLORRAM+0x80, + NV04_PGRAPH_PATT_COLORRAM+0x84, + NV04_PGRAPH_PATT_COLORRAM+0x88, + NV04_PGRAPH_PATT_COLORRAM+0x8c, + NV04_PGRAPH_PATT_COLORRAM+0x90, + NV04_PGRAPH_PATT_COLORRAM+0x94, + NV04_PGRAPH_PATT_COLORRAM+0x98, + NV04_PGRAPH_PATT_COLORRAM+0x9c, + NV04_PGRAPH_PATT_COLORRAM+0xa0, + NV04_PGRAPH_PATT_COLORRAM+0xa4, + NV04_PGRAPH_PATT_COLORRAM+0xa8, + NV04_PGRAPH_PATT_COLORRAM+0xac, + NV04_PGRAPH_PATT_COLORRAM+0xb0, + NV04_PGRAPH_PATT_COLORRAM+0xb4, + NV04_PGRAPH_PATT_COLORRAM+0xb8, + NV04_PGRAPH_PATT_COLORRAM+0xbc, + NV04_PGRAPH_PATT_COLORRAM+0xc0, + NV04_PGRAPH_PATT_COLORRAM+0xc4, + NV04_PGRAPH_PATT_COLORRAM+0xc8, + NV04_PGRAPH_PATT_COLORRAM+0xcc, + NV04_PGRAPH_PATT_COLORRAM+0xd0, + NV04_PGRAPH_PATT_COLORRAM+0xd4, + NV04_PGRAPH_PATT_COLORRAM+0xd8, + NV04_PGRAPH_PATT_COLORRAM+0xdc, + NV04_PGRAPH_PATT_COLORRAM+0xe0, + NV04_PGRAPH_PATT_COLORRAM+0xe4, + NV04_PGRAPH_PATT_COLORRAM+0xe8, + NV04_PGRAPH_PATT_COLORRAM+0xec, + NV04_PGRAPH_PATT_COLORRAM+0xf0, + NV04_PGRAPH_PATT_COLORRAM+0xf4, + NV04_PGRAPH_PATT_COLORRAM+0xf8, + NV04_PGRAPH_PATT_COLORRAM+0xfc, + NV04_PGRAPH_PATTERN, + 0x0040080c, + NV04_PGRAPH_PATTERN_SHAPE, + 0x00400600, + NV04_PGRAPH_ROP3, + NV04_PGRAPH_CHROMA, + NV04_PGRAPH_BETA_AND, + NV04_PGRAPH_BETA_PREMULT, + NV04_PGRAPH_CONTROL0, + NV04_PGRAPH_CONTROL1, + NV04_PGRAPH_CONTROL2, + NV04_PGRAPH_BLEND, + NV04_PGRAPH_STORED_FMT, + NV04_PGRAPH_SOURCE_COLOR, + 0x00400560, + 0x00400568, + 0x00400564, + 0x0040056c, + 0x00400400, + 0x00400480, + 0x00400404, + 0x00400484, + 0x00400408, + 0x00400488, + 0x0040040c, + 0x0040048c, + 0x00400410, + 0x00400490, + 0x00400414, + 0x00400494, + 0x00400418, + 0x00400498, + 0x0040041c, + 0x0040049c, + 0x00400420, + 0x004004a0, + 0x00400424, + 0x004004a4, + 0x00400428, + 0x004004a8, + 0x0040042c, + 0x004004ac, + 0x00400430, + 0x004004b0, + 0x00400434, + 0x004004b4, + 0x00400438, + 0x004004b8, + 0x0040043c, + 0x004004bc, + 0x00400440, + 0x004004c0, + 0x00400444, + 0x004004c4, + 0x00400448, + 0x004004c8, + 0x0040044c, + 0x004004cc, + 0x00400450, + 0x004004d0, + 0x00400454, + 0x004004d4, + 0x00400458, + 0x004004d8, + 0x0040045c, + 0x004004dc, + 0x00400460, + 0x004004e0, + 0x00400464, + 0x004004e4, + 0x00400468, + 0x004004e8, + 0x0040046c, + 0x004004ec, + 0x00400470, + 0x004004f0, + 0x00400474, + 0x004004f4, + 0x00400478, + 0x004004f8, + 0x0040047c, + 0x004004fc, + 0x00400534, + 0x00400538, + 0x00400514, + 0x00400518, + 0x0040051c, + 0x00400520, + 0x00400524, + 0x00400528, + 0x0040052c, + 0x00400530, + 0x00400d00, + 0x00400d40, + 0x00400d80, + 0x00400d04, + 0x00400d44, + 0x00400d84, + 0x00400d08, + 0x00400d48, + 0x00400d88, + 0x00400d0c, + 0x00400d4c, + 0x00400d8c, + 0x00400d10, + 0x00400d50, + 0x00400d90, + 0x00400d14, + 0x00400d54, + 0x00400d94, + 0x00400d18, + 0x00400d58, + 0x00400d98, + 0x00400d1c, + 0x00400d5c, + 0x00400d9c, + 0x00400d20, + 0x00400d60, + 0x00400da0, + 0x00400d24, + 0x00400d64, + 0x00400da4, + 0x00400d28, + 0x00400d68, + 0x00400da8, + 0x00400d2c, + 0x00400d6c, + 0x00400dac, + 0x00400d30, + 0x00400d70, + 0x00400db0, + 0x00400d34, + 0x00400d74, + 0x00400db4, + 0x00400d38, + 0x00400d78, + 0x00400db8, + 0x00400d3c, + 0x00400d7c, + 0x00400dbc, + 0x00400590, + 0x00400594, + 0x00400598, + 0x0040059c, + 0x004005a8, + 0x004005ac, + 0x004005b0, + 0x004005b4, + 0x004005c0, + 0x004005c4, + 0x004005c8, + 0x004005cc, + 0x004005d0, + 0x004005d4, + 0x004005d8, + 0x004005dc, + 0x004005e0, + NV04_PGRAPH_PASSTHRU_0, + NV04_PGRAPH_PASSTHRU_1, + NV04_PGRAPH_PASSTHRU_2, + NV04_PGRAPH_DVD_COLORFMT, + NV04_PGRAPH_SCALED_FORMAT, + NV04_PGRAPH_MISC24_0, + NV04_PGRAPH_MISC24_1, + NV04_PGRAPH_MISC24_2, + 0x00400500, + 0x00400504, + NV04_PGRAPH_VALID1, + NV04_PGRAPH_VALID2, + NV04_PGRAPH_DEBUG_3 +}; + +struct graph_state { + uint32_t nv04[ARRAY_SIZE(nv04_graph_ctx_regs)]; +}; + +static struct nouveau_channel * +nv04_graph_channel(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + int chid = dev_priv->engine.fifo.channels; + + if (nv_rd32(dev, NV04_PGRAPH_CTX_CONTROL) & 0x00010000) + chid = nv_rd32(dev, NV04_PGRAPH_CTX_USER) >> 24; + + if (chid >= dev_priv->engine.fifo.channels) + return NULL; + + return dev_priv->channels.ptr[chid]; +} + +static uint32_t *ctx_reg(struct graph_state *ctx, uint32_t reg) +{ + int i; + + for (i = 0; i < ARRAY_SIZE(nv04_graph_ctx_regs); i++) { + if (nv04_graph_ctx_regs[i] == reg) + return &ctx->nv04[i]; + } + + return NULL; +} + +static int +nv04_graph_load_context(struct nouveau_channel *chan) +{ + struct graph_state *pgraph_ctx = chan->engctx[NVOBJ_ENGINE_GR]; + struct drm_device *dev = chan->dev; + uint32_t tmp; + int i; + + for (i = 0; i < ARRAY_SIZE(nv04_graph_ctx_regs); i++) + nv_wr32(dev, nv04_graph_ctx_regs[i], pgraph_ctx->nv04[i]); + + nv_wr32(dev, NV04_PGRAPH_CTX_CONTROL, 0x10010100); + + tmp = nv_rd32(dev, NV04_PGRAPH_CTX_USER) & 0x00ffffff; + nv_wr32(dev, NV04_PGRAPH_CTX_USER, tmp | chan->id << 24); + + tmp = nv_rd32(dev, NV04_PGRAPH_FFINTFC_ST2); + nv_wr32(dev, NV04_PGRAPH_FFINTFC_ST2, tmp & 0x000fffff); + + return 0; +} + +static int +nv04_graph_unload_context(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_channel *chan = NULL; + struct graph_state *ctx; + uint32_t tmp; + int i; + + chan = nv04_graph_channel(dev); + if (!chan) + return 0; + ctx = chan->engctx[NVOBJ_ENGINE_GR]; + + for (i = 0; i < ARRAY_SIZE(nv04_graph_ctx_regs); i++) + ctx->nv04[i] = nv_rd32(dev, nv04_graph_ctx_regs[i]); + + nv_wr32(dev, NV04_PGRAPH_CTX_CONTROL, 0x10000000); + tmp = nv_rd32(dev, NV04_PGRAPH_CTX_USER) & 0x00ffffff; + tmp |= (dev_priv->engine.fifo.channels - 1) << 24; + nv_wr32(dev, NV04_PGRAPH_CTX_USER, tmp); + return 0; +} + +static int +nv04_graph_context_new(struct nouveau_channel *chan, int engine) +{ + struct graph_state *pgraph_ctx; + NV_DEBUG(chan->dev, "nv04_graph_context_create %d\n", chan->id); + + pgraph_ctx = kzalloc(sizeof(*pgraph_ctx), GFP_KERNEL); + if (pgraph_ctx == NULL) + return -ENOMEM; + + *ctx_reg(pgraph_ctx, NV04_PGRAPH_DEBUG_3) = 0xfad4ff31; + + chan->engctx[engine] = pgraph_ctx; + return 0; +} + +static void +nv04_graph_context_del(struct nouveau_channel *chan, int engine) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct graph_state *pgraph_ctx = chan->engctx[engine]; + unsigned long flags; + + spin_lock_irqsave(&dev_priv->context_switch_lock, flags); + nv04_graph_fifo_access(dev, false); + + /* Unload the context if it's the currently active one */ + if (nv04_graph_channel(dev) == chan) + nv04_graph_unload_context(dev); + + nv04_graph_fifo_access(dev, true); + spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); + + /* Free the context resources */ + kfree(pgraph_ctx); + chan->engctx[engine] = NULL; +} + +int +nv04_graph_object_new(struct nouveau_channel *chan, int engine, + u32 handle, u16 class) +{ + struct drm_device *dev = chan->dev; + struct nouveau_gpuobj *obj = NULL; + int ret; + + ret = nouveau_gpuobj_new(dev, chan, 16, 16, NVOBJ_FLAG_ZERO_FREE, &obj); + if (ret) + return ret; + obj->engine = 1; + obj->class = class; + +#ifdef __BIG_ENDIAN + nv_wo32(obj, 0x00, 0x00080000 | class); +#else + nv_wo32(obj, 0x00, class); +#endif + nv_wo32(obj, 0x04, 0x00000000); + nv_wo32(obj, 0x08, 0x00000000); + nv_wo32(obj, 0x0c, 0x00000000); + + ret = nouveau_ramht_insert(chan, handle, obj); + nouveau_gpuobj_ref(NULL, &obj); + return ret; +} + +static int +nv04_graph_init(struct drm_device *dev, int engine) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + uint32_t tmp; + + nv_wr32(dev, NV03_PMC_ENABLE, nv_rd32(dev, NV03_PMC_ENABLE) & + ~NV_PMC_ENABLE_PGRAPH); + nv_wr32(dev, NV03_PMC_ENABLE, nv_rd32(dev, NV03_PMC_ENABLE) | + NV_PMC_ENABLE_PGRAPH); + + /* Enable PGRAPH interrupts */ + nv_wr32(dev, NV03_PGRAPH_INTR, 0xFFFFFFFF); + nv_wr32(dev, NV03_PGRAPH_INTR_EN, 0xFFFFFFFF); + + nv_wr32(dev, NV04_PGRAPH_VALID1, 0); + nv_wr32(dev, NV04_PGRAPH_VALID2, 0); + /*nv_wr32(dev, NV04_PGRAPH_DEBUG_0, 0x000001FF); + nv_wr32(dev, NV04_PGRAPH_DEBUG_0, 0x001FFFFF);*/ + nv_wr32(dev, NV04_PGRAPH_DEBUG_0, 0x1231c000); + /*1231C000 blob, 001 haiku*/ + /*V_WRITE(NV04_PGRAPH_DEBUG_1, 0xf2d91100);*/ + nv_wr32(dev, NV04_PGRAPH_DEBUG_1, 0x72111100); + /*0x72111100 blob , 01 haiku*/ + /*nv_wr32(dev, NV04_PGRAPH_DEBUG_2, 0x11d5f870);*/ + nv_wr32(dev, NV04_PGRAPH_DEBUG_2, 0x11d5f071); + /*haiku same*/ + + /*nv_wr32(dev, NV04_PGRAPH_DEBUG_3, 0xfad4ff31);*/ + nv_wr32(dev, NV04_PGRAPH_DEBUG_3, 0xf0d4ff31); + /*haiku and blob 10d4*/ + + nv_wr32(dev, NV04_PGRAPH_STATE , 0xFFFFFFFF); + nv_wr32(dev, NV04_PGRAPH_CTX_CONTROL , 0x10000100); + tmp = nv_rd32(dev, NV04_PGRAPH_CTX_USER) & 0x00ffffff; + tmp |= (dev_priv->engine.fifo.channels - 1) << 24; + nv_wr32(dev, NV04_PGRAPH_CTX_USER, tmp); + + /* These don't belong here, they're part of a per-channel context */ + nv_wr32(dev, NV04_PGRAPH_PATTERN_SHAPE, 0x00000000); + nv_wr32(dev, NV04_PGRAPH_BETA_AND , 0xFFFFFFFF); + + return 0; +} + +static int +nv04_graph_fini(struct drm_device *dev, int engine) +{ + nv04_graph_unload_context(dev); + nv_wr32(dev, NV03_PGRAPH_INTR_EN, 0x00000000); + return 0; +} + +void +nv04_graph_fifo_access(struct drm_device *dev, bool enabled) +{ + if (enabled) + nv_wr32(dev, NV04_PGRAPH_FIFO, + nv_rd32(dev, NV04_PGRAPH_FIFO) | 1); + else + nv_wr32(dev, NV04_PGRAPH_FIFO, + nv_rd32(dev, NV04_PGRAPH_FIFO) & ~1); +} + +static int +nv04_graph_mthd_set_ref(struct nouveau_channel *chan, + u32 class, u32 mthd, u32 data) +{ + atomic_set(&chan->fence.last_sequence_irq, data); + return 0; +} + +int +nv04_graph_mthd_page_flip(struct nouveau_channel *chan, + u32 class, u32 mthd, u32 data) +{ + struct drm_device *dev = chan->dev; + struct nouveau_page_flip_state s; + + if (!nouveau_finish_page_flip(chan, &s)) + nv_set_crtc_base(dev, s.crtc, + s.offset + s.y * s.pitch + s.x * s.bpp / 8); + + return 0; +} + +/* + * Software methods, why they are needed, and how they all work: + * + * NV04 and NV05 keep most of the state in PGRAPH context itself, but some + * 2d engine settings are kept inside the grobjs themselves. The grobjs are + * 3 words long on both. grobj format on NV04 is: + * + * word 0: + * - bits 0-7: class + * - bit 12: color key active + * - bit 13: clip rect active + * - bit 14: if set, destination surface is swizzled and taken from buffer 5 + * [set by NV04_SWIZZLED_SURFACE], otherwise it's linear and taken + * from buffer 0 [set by NV04_CONTEXT_SURFACES_2D or + * NV03_CONTEXT_SURFACE_DST]. + * - bits 15-17: 2d operation [aka patch config] + * - bit 24: patch valid [enables rendering using this object] + * - bit 25: surf3d valid [for tex_tri and multitex_tri only] + * word 1: + * - bits 0-1: mono format + * - bits 8-13: color format + * - bits 16-31: DMA_NOTIFY instance + * word 2: + * - bits 0-15: DMA_A instance + * - bits 16-31: DMA_B instance + * + * On NV05 it's: + * + * word 0: + * - bits 0-7: class + * - bit 12: color key active + * - bit 13: clip rect active + * - bit 14: if set, destination surface is swizzled and taken from buffer 5 + * [set by NV04_SWIZZLED_SURFACE], otherwise it's linear and taken + * from buffer 0 [set by NV04_CONTEXT_SURFACES_2D or + * NV03_CONTEXT_SURFACE_DST]. + * - bits 15-17: 2d operation [aka patch config] + * - bits 20-22: dither mode + * - bit 24: patch valid [enables rendering using this object] + * - bit 25: surface_dst/surface_color/surf2d/surf3d valid + * - bit 26: surface_src/surface_zeta valid + * - bit 27: pattern valid + * - bit 28: rop valid + * - bit 29: beta1 valid + * - bit 30: beta4 valid + * word 1: + * - bits 0-1: mono format + * - bits 8-13: color format + * - bits 16-31: DMA_NOTIFY instance + * word 2: + * - bits 0-15: DMA_A instance + * - bits 16-31: DMA_B instance + * + * NV05 will set/unset the relevant valid bits when you poke the relevant + * object-binding methods with object of the proper type, or with the NULL + * type. It'll only allow rendering using the grobj if all needed objects + * are bound. The needed set of objects depends on selected operation: for + * example rop object is needed by ROP_AND, but not by SRCCOPY_AND. + * + * NV04 doesn't have these methods implemented at all, and doesn't have the + * relevant bits in grobj. Instead, it'll allow rendering whenever bit 24 + * is set. So we have to emulate them in software, internally keeping the + * same bits as NV05 does. Since grobjs are aligned to 16 bytes on nv04, + * but the last word isn't actually used for anything, we abuse it for this + * purpose. + * + * Actually, NV05 can optionally check bit 24 too, but we disable this since + * there's no use for it. + * + * For unknown reasons, NV04 implements surf3d binding in hardware as an + * exception. Also for unknown reasons, NV04 doesn't implement the clipping + * methods on the surf3d object, so we have to emulate them too. + */ + +static void +nv04_graph_set_ctx1(struct nouveau_channel *chan, u32 mask, u32 value) +{ + struct drm_device *dev = chan->dev; + u32 instance = (nv_rd32(dev, NV04_PGRAPH_CTX_SWITCH4) & 0xffff) << 4; + int subc = (nv_rd32(dev, NV04_PGRAPH_TRAPPED_ADDR) >> 13) & 0x7; + u32 tmp; + + tmp = nv_ri32(dev, instance); + tmp &= ~mask; + tmp |= value; + + nv_wi32(dev, instance, tmp); + nv_wr32(dev, NV04_PGRAPH_CTX_SWITCH1, tmp); + nv_wr32(dev, NV04_PGRAPH_CTX_CACHE1 + (subc<<2), tmp); +} + +static void +nv04_graph_set_ctx_val(struct nouveau_channel *chan, u32 mask, u32 value) +{ + struct drm_device *dev = chan->dev; + u32 instance = (nv_rd32(dev, NV04_PGRAPH_CTX_SWITCH4) & 0xffff) << 4; + u32 tmp, ctx1; + int class, op, valid = 1; + + ctx1 = nv_ri32(dev, instance); + class = ctx1 & 0xff; + op = (ctx1 >> 15) & 7; + tmp = nv_ri32(dev, instance + 0xc); + tmp &= ~mask; + tmp |= value; + nv_wi32(dev, instance + 0xc, tmp); + + /* check for valid surf2d/surf_dst/surf_color */ + if (!(tmp & 0x02000000)) + valid = 0; + /* check for valid surf_src/surf_zeta */ + if ((class == 0x1f || class == 0x48) && !(tmp & 0x04000000)) + valid = 0; + + switch (op) { + /* SRCCOPY_AND, SRCCOPY: no extra objects required */ + case 0: + case 3: + break; + /* ROP_AND: requires pattern and rop */ + case 1: + if (!(tmp & 0x18000000)) + valid = 0; + break; + /* BLEND_AND: requires beta1 */ + case 2: + if (!(tmp & 0x20000000)) + valid = 0; + break; + /* SRCCOPY_PREMULT, BLEND_PREMULT: beta4 required */ + case 4: + case 5: + if (!(tmp & 0x40000000)) + valid = 0; + break; + } + + nv04_graph_set_ctx1(chan, 0x01000000, valid << 24); +} + +static int +nv04_graph_mthd_set_operation(struct nouveau_channel *chan, + u32 class, u32 mthd, u32 data) +{ + if (data > 5) + return 1; + /* Old versions of the objects only accept first three operations. */ + if (data > 2 && class < 0x40) + return 1; + nv04_graph_set_ctx1(chan, 0x00038000, data << 15); + /* changing operation changes set of objects needed for validation */ + nv04_graph_set_ctx_val(chan, 0, 0); + return 0; +} + +static int +nv04_graph_mthd_surf3d_clip_h(struct nouveau_channel *chan, + u32 class, u32 mthd, u32 data) +{ + uint32_t min = data & 0xffff, max; + uint32_t w = data >> 16; + if (min & 0x8000) + /* too large */ + return 1; + if (w & 0x8000) + /* yes, it accepts negative for some reason. */ + w |= 0xffff0000; + max = min + w; + max &= 0x3ffff; + nv_wr32(chan->dev, 0x40053c, min); + nv_wr32(chan->dev, 0x400544, max); + return 0; +} + +static int +nv04_graph_mthd_surf3d_clip_v(struct nouveau_channel *chan, + u32 class, u32 mthd, u32 data) +{ + uint32_t min = data & 0xffff, max; + uint32_t w = data >> 16; + if (min & 0x8000) + /* too large */ + return 1; + if (w & 0x8000) + /* yes, it accepts negative for some reason. */ + w |= 0xffff0000; + max = min + w; + max &= 0x3ffff; + nv_wr32(chan->dev, 0x400540, min); + nv_wr32(chan->dev, 0x400548, max); + return 0; +} + +static int +nv04_graph_mthd_bind_surf2d(struct nouveau_channel *chan, + u32 class, u32 mthd, u32 data) +{ + switch (nv_ri32(chan->dev, data << 4) & 0xff) { + case 0x30: + nv04_graph_set_ctx1(chan, 0x00004000, 0); + nv04_graph_set_ctx_val(chan, 0x02000000, 0); + return 0; + case 0x42: + nv04_graph_set_ctx1(chan, 0x00004000, 0); + nv04_graph_set_ctx_val(chan, 0x02000000, 0x02000000); + return 0; + } + return 1; +} + +static int +nv04_graph_mthd_bind_surf2d_swzsurf(struct nouveau_channel *chan, + u32 class, u32 mthd, u32 data) +{ + switch (nv_ri32(chan->dev, data << 4) & 0xff) { + case 0x30: + nv04_graph_set_ctx1(chan, 0x00004000, 0); + nv04_graph_set_ctx_val(chan, 0x02000000, 0); + return 0; + case 0x42: + nv04_graph_set_ctx1(chan, 0x00004000, 0); + nv04_graph_set_ctx_val(chan, 0x02000000, 0x02000000); + return 0; + case 0x52: + nv04_graph_set_ctx1(chan, 0x00004000, 0x00004000); + nv04_graph_set_ctx_val(chan, 0x02000000, 0x02000000); + return 0; + } + return 1; +} + +static int +nv04_graph_mthd_bind_nv01_patt(struct nouveau_channel *chan, + u32 class, u32 mthd, u32 data) +{ + switch (nv_ri32(chan->dev, data << 4) & 0xff) { + case 0x30: + nv04_graph_set_ctx_val(chan, 0x08000000, 0); + return 0; + case 0x18: + nv04_graph_set_ctx_val(chan, 0x08000000, 0x08000000); + return 0; + } + return 1; +} + +static int +nv04_graph_mthd_bind_nv04_patt(struct nouveau_channel *chan, + u32 class, u32 mthd, u32 data) +{ + switch (nv_ri32(chan->dev, data << 4) & 0xff) { + case 0x30: + nv04_graph_set_ctx_val(chan, 0x08000000, 0); + return 0; + case 0x44: + nv04_graph_set_ctx_val(chan, 0x08000000, 0x08000000); + return 0; + } + return 1; +} + +static int +nv04_graph_mthd_bind_rop(struct nouveau_channel *chan, + u32 class, u32 mthd, u32 data) +{ + switch (nv_ri32(chan->dev, data << 4) & 0xff) { + case 0x30: + nv04_graph_set_ctx_val(chan, 0x10000000, 0); + return 0; + case 0x43: + nv04_graph_set_ctx_val(chan, 0x10000000, 0x10000000); + return 0; + } + return 1; +} + +static int +nv04_graph_mthd_bind_beta1(struct nouveau_channel *chan, + u32 class, u32 mthd, u32 data) +{ + switch (nv_ri32(chan->dev, data << 4) & 0xff) { + case 0x30: + nv04_graph_set_ctx_val(chan, 0x20000000, 0); + return 0; + case 0x12: + nv04_graph_set_ctx_val(chan, 0x20000000, 0x20000000); + return 0; + } + return 1; +} + +static int +nv04_graph_mthd_bind_beta4(struct nouveau_channel *chan, + u32 class, u32 mthd, u32 data) +{ + switch (nv_ri32(chan->dev, data << 4) & 0xff) { + case 0x30: + nv04_graph_set_ctx_val(chan, 0x40000000, 0); + return 0; + case 0x72: + nv04_graph_set_ctx_val(chan, 0x40000000, 0x40000000); + return 0; + } + return 1; +} + +static int +nv04_graph_mthd_bind_surf_dst(struct nouveau_channel *chan, + u32 class, u32 mthd, u32 data) +{ + switch (nv_ri32(chan->dev, data << 4) & 0xff) { + case 0x30: + nv04_graph_set_ctx_val(chan, 0x02000000, 0); + return 0; + case 0x58: + nv04_graph_set_ctx_val(chan, 0x02000000, 0x02000000); + return 0; + } + return 1; +} + +static int +nv04_graph_mthd_bind_surf_src(struct nouveau_channel *chan, + u32 class, u32 mthd, u32 data) +{ + switch (nv_ri32(chan->dev, data << 4) & 0xff) { + case 0x30: + nv04_graph_set_ctx_val(chan, 0x04000000, 0); + return 0; + case 0x59: + nv04_graph_set_ctx_val(chan, 0x04000000, 0x04000000); + return 0; + } + return 1; +} + +static int +nv04_graph_mthd_bind_surf_color(struct nouveau_channel *chan, + u32 class, u32 mthd, u32 data) +{ + switch (nv_ri32(chan->dev, data << 4) & 0xff) { + case 0x30: + nv04_graph_set_ctx_val(chan, 0x02000000, 0); + return 0; + case 0x5a: + nv04_graph_set_ctx_val(chan, 0x02000000, 0x02000000); + return 0; + } + return 1; +} + +static int +nv04_graph_mthd_bind_surf_zeta(struct nouveau_channel *chan, + u32 class, u32 mthd, u32 data) +{ + switch (nv_ri32(chan->dev, data << 4) & 0xff) { + case 0x30: + nv04_graph_set_ctx_val(chan, 0x04000000, 0); + return 0; + case 0x5b: + nv04_graph_set_ctx_val(chan, 0x04000000, 0x04000000); + return 0; + } + return 1; +} + +static int +nv04_graph_mthd_bind_clip(struct nouveau_channel *chan, + u32 class, u32 mthd, u32 data) +{ + switch (nv_ri32(chan->dev, data << 4) & 0xff) { + case 0x30: + nv04_graph_set_ctx1(chan, 0x2000, 0); + return 0; + case 0x19: + nv04_graph_set_ctx1(chan, 0x2000, 0x2000); + return 0; + } + return 1; +} + +static int +nv04_graph_mthd_bind_chroma(struct nouveau_channel *chan, + u32 class, u32 mthd, u32 data) +{ + switch (nv_ri32(chan->dev, data << 4) & 0xff) { + case 0x30: + nv04_graph_set_ctx1(chan, 0x1000, 0); + return 0; + /* Yes, for some reason even the old versions of objects + * accept 0x57 and not 0x17. Consistency be damned. + */ + case 0x57: + nv04_graph_set_ctx1(chan, 0x1000, 0x1000); + return 0; + } + return 1; +} + +static struct nouveau_bitfield nv04_graph_intr[] = { + { NV_PGRAPH_INTR_NOTIFY, "NOTIFY" }, + {} +}; + +static struct nouveau_bitfield nv04_graph_nstatus[] = { + { NV04_PGRAPH_NSTATUS_STATE_IN_USE, "STATE_IN_USE" }, + { NV04_PGRAPH_NSTATUS_INVALID_STATE, "INVALID_STATE" }, + { NV04_PGRAPH_NSTATUS_BAD_ARGUMENT, "BAD_ARGUMENT" }, + { NV04_PGRAPH_NSTATUS_PROTECTION_FAULT, "PROTECTION_FAULT" }, + {} +}; + +struct nouveau_bitfield nv04_graph_nsource[] = { + { NV03_PGRAPH_NSOURCE_NOTIFICATION, "NOTIFICATION" }, + { NV03_PGRAPH_NSOURCE_DATA_ERROR, "DATA_ERROR" }, + { NV03_PGRAPH_NSOURCE_PROTECTION_ERROR, "PROTECTION_ERROR" }, + { NV03_PGRAPH_NSOURCE_RANGE_EXCEPTION, "RANGE_EXCEPTION" }, + { NV03_PGRAPH_NSOURCE_LIMIT_COLOR, "LIMIT_COLOR" }, + { NV03_PGRAPH_NSOURCE_LIMIT_ZETA, "LIMIT_ZETA" }, + { NV03_PGRAPH_NSOURCE_ILLEGAL_MTHD, "ILLEGAL_MTHD" }, + { NV03_PGRAPH_NSOURCE_DMA_R_PROTECTION, "DMA_R_PROTECTION" }, + { NV03_PGRAPH_NSOURCE_DMA_W_PROTECTION, "DMA_W_PROTECTION" }, + { NV03_PGRAPH_NSOURCE_FORMAT_EXCEPTION, "FORMAT_EXCEPTION" }, + { NV03_PGRAPH_NSOURCE_PATCH_EXCEPTION, "PATCH_EXCEPTION" }, + { NV03_PGRAPH_NSOURCE_STATE_INVALID, "STATE_INVALID" }, + { NV03_PGRAPH_NSOURCE_DOUBLE_NOTIFY, "DOUBLE_NOTIFY" }, + { NV03_PGRAPH_NSOURCE_NOTIFY_IN_USE, "NOTIFY_IN_USE" }, + { NV03_PGRAPH_NSOURCE_METHOD_CNT, "METHOD_CNT" }, + { NV03_PGRAPH_NSOURCE_BFR_NOTIFICATION, "BFR_NOTIFICATION" }, + { NV03_PGRAPH_NSOURCE_DMA_VTX_PROTECTION, "DMA_VTX_PROTECTION" }, + { NV03_PGRAPH_NSOURCE_DMA_WIDTH_A, "DMA_WIDTH_A" }, + { NV03_PGRAPH_NSOURCE_DMA_WIDTH_B, "DMA_WIDTH_B" }, + {} +}; + +static void +nv04_graph_context_switch(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_channel *chan = NULL; + int chid; + + nouveau_wait_for_idle(dev); + + /* If previous context is valid, we need to save it */ + nv04_graph_unload_context(dev); + + /* Load context for next channel */ + chid = dev_priv->engine.fifo.channel_id(dev); + chan = dev_priv->channels.ptr[chid]; + if (chan) + nv04_graph_load_context(chan); +} + +static void +nv04_graph_isr(struct drm_device *dev) +{ + u32 stat; + + while ((stat = nv_rd32(dev, NV03_PGRAPH_INTR))) { + u32 nsource = nv_rd32(dev, NV03_PGRAPH_NSOURCE); + u32 nstatus = nv_rd32(dev, NV03_PGRAPH_NSTATUS); + u32 addr = nv_rd32(dev, NV04_PGRAPH_TRAPPED_ADDR); + u32 chid = (addr & 0x0f000000) >> 24; + u32 subc = (addr & 0x0000e000) >> 13; + u32 mthd = (addr & 0x00001ffc); + u32 data = nv_rd32(dev, NV04_PGRAPH_TRAPPED_DATA); + u32 class = nv_rd32(dev, 0x400180 + subc * 4) & 0xff; + u32 show = stat; + + if (stat & NV_PGRAPH_INTR_NOTIFY) { + if (nsource & NV03_PGRAPH_NSOURCE_ILLEGAL_MTHD) { + if (!nouveau_gpuobj_mthd_call2(dev, chid, class, mthd, data)) + show &= ~NV_PGRAPH_INTR_NOTIFY; + } + } + + if (stat & NV_PGRAPH_INTR_CONTEXT_SWITCH) { + nv_wr32(dev, NV03_PGRAPH_INTR, NV_PGRAPH_INTR_CONTEXT_SWITCH); + stat &= ~NV_PGRAPH_INTR_CONTEXT_SWITCH; + show &= ~NV_PGRAPH_INTR_CONTEXT_SWITCH; + nv04_graph_context_switch(dev); + } + + nv_wr32(dev, NV03_PGRAPH_INTR, stat); + nv_wr32(dev, NV04_PGRAPH_FIFO, 0x00000001); + + if (show && nouveau_ratelimit()) { + NV_INFO(dev, "PGRAPH -"); + nouveau_bitfield_print(nv04_graph_intr, show); + printk(" nsource:"); + nouveau_bitfield_print(nv04_graph_nsource, nsource); + printk(" nstatus:"); + nouveau_bitfield_print(nv04_graph_nstatus, nstatus); + printk("\n"); + NV_INFO(dev, "PGRAPH - ch %d/%d class 0x%04x " + "mthd 0x%04x data 0x%08x\n", + chid, subc, class, mthd, data); + } + } +} + +static void +nv04_graph_destroy(struct drm_device *dev, int engine) +{ + struct nv04_graph_engine *pgraph = nv_engine(dev, engine); + + nouveau_irq_unregister(dev, 12); + + NVOBJ_ENGINE_DEL(dev, GR); + kfree(pgraph); +} + +int +nv04_graph_create(struct drm_device *dev) +{ + struct nv04_graph_engine *pgraph; + + pgraph = kzalloc(sizeof(*pgraph), GFP_KERNEL); + if (!pgraph) + return -ENOMEM; + + pgraph->base.destroy = nv04_graph_destroy; + pgraph->base.init = nv04_graph_init; + pgraph->base.fini = nv04_graph_fini; + pgraph->base.context_new = nv04_graph_context_new; + pgraph->base.context_del = nv04_graph_context_del; + pgraph->base.object_new = nv04_graph_object_new; + + NVOBJ_ENGINE_ADD(dev, GR, &pgraph->base); + nouveau_irq_register(dev, 12, nv04_graph_isr); + + /* dvd subpicture */ + NVOBJ_CLASS(dev, 0x0038, GR); + + /* m2mf */ + NVOBJ_CLASS(dev, 0x0039, GR); + + /* nv03 gdirect */ + NVOBJ_CLASS(dev, 0x004b, GR); + NVOBJ_MTHD (dev, 0x004b, 0x0184, nv04_graph_mthd_bind_nv01_patt); + NVOBJ_MTHD (dev, 0x004b, 0x0188, nv04_graph_mthd_bind_rop); + NVOBJ_MTHD (dev, 0x004b, 0x018c, nv04_graph_mthd_bind_beta1); + NVOBJ_MTHD (dev, 0x004b, 0x0190, nv04_graph_mthd_bind_surf_dst); + NVOBJ_MTHD (dev, 0x004b, 0x02fc, nv04_graph_mthd_set_operation); + + /* nv04 gdirect */ + NVOBJ_CLASS(dev, 0x004a, GR); + NVOBJ_MTHD (dev, 0x004a, 0x0188, nv04_graph_mthd_bind_nv04_patt); + NVOBJ_MTHD (dev, 0x004a, 0x018c, nv04_graph_mthd_bind_rop); + NVOBJ_MTHD (dev, 0x004a, 0x0190, nv04_graph_mthd_bind_beta1); + NVOBJ_MTHD (dev, 0x004a, 0x0194, nv04_graph_mthd_bind_beta4); + NVOBJ_MTHD (dev, 0x004a, 0x0198, nv04_graph_mthd_bind_surf2d); + NVOBJ_MTHD (dev, 0x004a, 0x02fc, nv04_graph_mthd_set_operation); + + /* nv01 imageblit */ + NVOBJ_CLASS(dev, 0x001f, GR); + NVOBJ_MTHD (dev, 0x001f, 0x0184, nv04_graph_mthd_bind_chroma); + NVOBJ_MTHD (dev, 0x001f, 0x0188, nv04_graph_mthd_bind_clip); + NVOBJ_MTHD (dev, 0x001f, 0x018c, nv04_graph_mthd_bind_nv01_patt); + NVOBJ_MTHD (dev, 0x001f, 0x0190, nv04_graph_mthd_bind_rop); + NVOBJ_MTHD (dev, 0x001f, 0x0194, nv04_graph_mthd_bind_beta1); + NVOBJ_MTHD (dev, 0x001f, 0x0198, nv04_graph_mthd_bind_surf_dst); + NVOBJ_MTHD (dev, 0x001f, 0x019c, nv04_graph_mthd_bind_surf_src); + NVOBJ_MTHD (dev, 0x001f, 0x02fc, nv04_graph_mthd_set_operation); + + /* nv04 imageblit */ + NVOBJ_CLASS(dev, 0x005f, GR); + NVOBJ_MTHD (dev, 0x005f, 0x0184, nv04_graph_mthd_bind_chroma); + NVOBJ_MTHD (dev, 0x005f, 0x0188, nv04_graph_mthd_bind_clip); + NVOBJ_MTHD (dev, 0x005f, 0x018c, nv04_graph_mthd_bind_nv04_patt); + NVOBJ_MTHD (dev, 0x005f, 0x0190, nv04_graph_mthd_bind_rop); + NVOBJ_MTHD (dev, 0x005f, 0x0194, nv04_graph_mthd_bind_beta1); + NVOBJ_MTHD (dev, 0x005f, 0x0198, nv04_graph_mthd_bind_beta4); + NVOBJ_MTHD (dev, 0x005f, 0x019c, nv04_graph_mthd_bind_surf2d); + NVOBJ_MTHD (dev, 0x005f, 0x02fc, nv04_graph_mthd_set_operation); + + /* nv04 iifc */ + NVOBJ_CLASS(dev, 0x0060, GR); + NVOBJ_MTHD (dev, 0x0060, 0x0188, nv04_graph_mthd_bind_chroma); + NVOBJ_MTHD (dev, 0x0060, 0x018c, nv04_graph_mthd_bind_clip); + NVOBJ_MTHD (dev, 0x0060, 0x0190, nv04_graph_mthd_bind_nv04_patt); + NVOBJ_MTHD (dev, 0x0060, 0x0194, nv04_graph_mthd_bind_rop); + NVOBJ_MTHD (dev, 0x0060, 0x0198, nv04_graph_mthd_bind_beta1); + NVOBJ_MTHD (dev, 0x0060, 0x019c, nv04_graph_mthd_bind_beta4); + NVOBJ_MTHD (dev, 0x0060, 0x01a0, nv04_graph_mthd_bind_surf2d_swzsurf); + NVOBJ_MTHD (dev, 0x0060, 0x03e4, nv04_graph_mthd_set_operation); + + /* nv05 iifc */ + NVOBJ_CLASS(dev, 0x0064, GR); + + /* nv01 ifc */ + NVOBJ_CLASS(dev, 0x0021, GR); + NVOBJ_MTHD (dev, 0x0021, 0x0184, nv04_graph_mthd_bind_chroma); + NVOBJ_MTHD (dev, 0x0021, 0x0188, nv04_graph_mthd_bind_clip); + NVOBJ_MTHD (dev, 0x0021, 0x018c, nv04_graph_mthd_bind_nv01_patt); + NVOBJ_MTHD (dev, 0x0021, 0x0190, nv04_graph_mthd_bind_rop); + NVOBJ_MTHD (dev, 0x0021, 0x0194, nv04_graph_mthd_bind_beta1); + NVOBJ_MTHD (dev, 0x0021, 0x0198, nv04_graph_mthd_bind_surf_dst); + NVOBJ_MTHD (dev, 0x0021, 0x02fc, nv04_graph_mthd_set_operation); + + /* nv04 ifc */ + NVOBJ_CLASS(dev, 0x0061, GR); + NVOBJ_MTHD (dev, 0x0061, 0x0184, nv04_graph_mthd_bind_chroma); + NVOBJ_MTHD (dev, 0x0061, 0x0188, nv04_graph_mthd_bind_clip); + NVOBJ_MTHD (dev, 0x0061, 0x018c, nv04_graph_mthd_bind_nv04_patt); + NVOBJ_MTHD (dev, 0x0061, 0x0190, nv04_graph_mthd_bind_rop); + NVOBJ_MTHD (dev, 0x0061, 0x0194, nv04_graph_mthd_bind_beta1); + NVOBJ_MTHD (dev, 0x0061, 0x0198, nv04_graph_mthd_bind_beta4); + NVOBJ_MTHD (dev, 0x0061, 0x019c, nv04_graph_mthd_bind_surf2d); + NVOBJ_MTHD (dev, 0x0061, 0x02fc, nv04_graph_mthd_set_operation); + + /* nv05 ifc */ + NVOBJ_CLASS(dev, 0x0065, GR); + + /* nv03 sifc */ + NVOBJ_CLASS(dev, 0x0036, GR); + NVOBJ_MTHD (dev, 0x0036, 0x0184, nv04_graph_mthd_bind_chroma); + NVOBJ_MTHD (dev, 0x0036, 0x0188, nv04_graph_mthd_bind_nv01_patt); + NVOBJ_MTHD (dev, 0x0036, 0x018c, nv04_graph_mthd_bind_rop); + NVOBJ_MTHD (dev, 0x0036, 0x0190, nv04_graph_mthd_bind_beta1); + NVOBJ_MTHD (dev, 0x0036, 0x0194, nv04_graph_mthd_bind_surf_dst); + NVOBJ_MTHD (dev, 0x0036, 0x02fc, nv04_graph_mthd_set_operation); + + /* nv04 sifc */ + NVOBJ_CLASS(dev, 0x0076, GR); + NVOBJ_MTHD (dev, 0x0076, 0x0184, nv04_graph_mthd_bind_chroma); + NVOBJ_MTHD (dev, 0x0076, 0x0188, nv04_graph_mthd_bind_nv04_patt); + NVOBJ_MTHD (dev, 0x0076, 0x018c, nv04_graph_mthd_bind_rop); + NVOBJ_MTHD (dev, 0x0076, 0x0190, nv04_graph_mthd_bind_beta1); + NVOBJ_MTHD (dev, 0x0076, 0x0194, nv04_graph_mthd_bind_beta4); + NVOBJ_MTHD (dev, 0x0076, 0x0198, nv04_graph_mthd_bind_surf2d); + NVOBJ_MTHD (dev, 0x0076, 0x02fc, nv04_graph_mthd_set_operation); + + /* nv05 sifc */ + NVOBJ_CLASS(dev, 0x0066, GR); + + /* nv03 sifm */ + NVOBJ_CLASS(dev, 0x0037, GR); + NVOBJ_MTHD (dev, 0x0037, 0x0188, nv04_graph_mthd_bind_nv01_patt); + NVOBJ_MTHD (dev, 0x0037, 0x018c, nv04_graph_mthd_bind_rop); + NVOBJ_MTHD (dev, 0x0037, 0x0190, nv04_graph_mthd_bind_beta1); + NVOBJ_MTHD (dev, 0x0037, 0x0194, nv04_graph_mthd_bind_surf_dst); + NVOBJ_MTHD (dev, 0x0037, 0x0304, nv04_graph_mthd_set_operation); + + /* nv04 sifm */ + NVOBJ_CLASS(dev, 0x0077, GR); + NVOBJ_MTHD (dev, 0x0077, 0x0188, nv04_graph_mthd_bind_nv04_patt); + NVOBJ_MTHD (dev, 0x0077, 0x018c, nv04_graph_mthd_bind_rop); + NVOBJ_MTHD (dev, 0x0077, 0x0190, nv04_graph_mthd_bind_beta1); + NVOBJ_MTHD (dev, 0x0077, 0x0194, nv04_graph_mthd_bind_beta4); + NVOBJ_MTHD (dev, 0x0077, 0x0198, nv04_graph_mthd_bind_surf2d_swzsurf); + NVOBJ_MTHD (dev, 0x0077, 0x0304, nv04_graph_mthd_set_operation); + + /* null */ + NVOBJ_CLASS(dev, 0x0030, GR); + + /* surf2d */ + NVOBJ_CLASS(dev, 0x0042, GR); + + /* rop */ + NVOBJ_CLASS(dev, 0x0043, GR); + + /* beta1 */ + NVOBJ_CLASS(dev, 0x0012, GR); + + /* beta4 */ + NVOBJ_CLASS(dev, 0x0072, GR); + + /* cliprect */ + NVOBJ_CLASS(dev, 0x0019, GR); + + /* nv01 pattern */ + NVOBJ_CLASS(dev, 0x0018, GR); + + /* nv04 pattern */ + NVOBJ_CLASS(dev, 0x0044, GR); + + /* swzsurf */ + NVOBJ_CLASS(dev, 0x0052, GR); + + /* surf3d */ + NVOBJ_CLASS(dev, 0x0053, GR); + NVOBJ_MTHD (dev, 0x0053, 0x02f8, nv04_graph_mthd_surf3d_clip_h); + NVOBJ_MTHD (dev, 0x0053, 0x02fc, nv04_graph_mthd_surf3d_clip_v); + + /* nv03 tex_tri */ + NVOBJ_CLASS(dev, 0x0048, GR); + NVOBJ_MTHD (dev, 0x0048, 0x0188, nv04_graph_mthd_bind_clip); + NVOBJ_MTHD (dev, 0x0048, 0x018c, nv04_graph_mthd_bind_surf_color); + NVOBJ_MTHD (dev, 0x0048, 0x0190, nv04_graph_mthd_bind_surf_zeta); + + /* tex_tri */ + NVOBJ_CLASS(dev, 0x0054, GR); + + /* multitex_tri */ + NVOBJ_CLASS(dev, 0x0055, GR); + + /* nv01 chroma */ + NVOBJ_CLASS(dev, 0x0017, GR); + + /* nv04 chroma */ + NVOBJ_CLASS(dev, 0x0057, GR); + + /* surf_dst */ + NVOBJ_CLASS(dev, 0x0058, GR); + + /* surf_src */ + NVOBJ_CLASS(dev, 0x0059, GR); + + /* surf_color */ + NVOBJ_CLASS(dev, 0x005a, GR); + + /* surf_zeta */ + NVOBJ_CLASS(dev, 0x005b, GR); + + /* nv01 line */ + NVOBJ_CLASS(dev, 0x001c, GR); + NVOBJ_MTHD (dev, 0x001c, 0x0184, nv04_graph_mthd_bind_clip); + NVOBJ_MTHD (dev, 0x001c, 0x0188, nv04_graph_mthd_bind_nv01_patt); + NVOBJ_MTHD (dev, 0x001c, 0x018c, nv04_graph_mthd_bind_rop); + NVOBJ_MTHD (dev, 0x001c, 0x0190, nv04_graph_mthd_bind_beta1); + NVOBJ_MTHD (dev, 0x001c, 0x0194, nv04_graph_mthd_bind_surf_dst); + NVOBJ_MTHD (dev, 0x001c, 0x02fc, nv04_graph_mthd_set_operation); + + /* nv04 line */ + NVOBJ_CLASS(dev, 0x005c, GR); + NVOBJ_MTHD (dev, 0x005c, 0x0184, nv04_graph_mthd_bind_clip); + NVOBJ_MTHD (dev, 0x005c, 0x0188, nv04_graph_mthd_bind_nv04_patt); + NVOBJ_MTHD (dev, 0x005c, 0x018c, nv04_graph_mthd_bind_rop); + NVOBJ_MTHD (dev, 0x005c, 0x0190, nv04_graph_mthd_bind_beta1); + NVOBJ_MTHD (dev, 0x005c, 0x0194, nv04_graph_mthd_bind_beta4); + NVOBJ_MTHD (dev, 0x005c, 0x0198, nv04_graph_mthd_bind_surf2d); + NVOBJ_MTHD (dev, 0x005c, 0x02fc, nv04_graph_mthd_set_operation); + + /* nv01 tri */ + NVOBJ_CLASS(dev, 0x001d, GR); + NVOBJ_MTHD (dev, 0x001d, 0x0184, nv04_graph_mthd_bind_clip); + NVOBJ_MTHD (dev, 0x001d, 0x0188, nv04_graph_mthd_bind_nv01_patt); + NVOBJ_MTHD (dev, 0x001d, 0x018c, nv04_graph_mthd_bind_rop); + NVOBJ_MTHD (dev, 0x001d, 0x0190, nv04_graph_mthd_bind_beta1); + NVOBJ_MTHD (dev, 0x001d, 0x0194, nv04_graph_mthd_bind_surf_dst); + NVOBJ_MTHD (dev, 0x001d, 0x02fc, nv04_graph_mthd_set_operation); + + /* nv04 tri */ + NVOBJ_CLASS(dev, 0x005d, GR); + NVOBJ_MTHD (dev, 0x005d, 0x0184, nv04_graph_mthd_bind_clip); + NVOBJ_MTHD (dev, 0x005d, 0x0188, nv04_graph_mthd_bind_nv04_patt); + NVOBJ_MTHD (dev, 0x005d, 0x018c, nv04_graph_mthd_bind_rop); + NVOBJ_MTHD (dev, 0x005d, 0x0190, nv04_graph_mthd_bind_beta1); + NVOBJ_MTHD (dev, 0x005d, 0x0194, nv04_graph_mthd_bind_beta4); + NVOBJ_MTHD (dev, 0x005d, 0x0198, nv04_graph_mthd_bind_surf2d); + NVOBJ_MTHD (dev, 0x005d, 0x02fc, nv04_graph_mthd_set_operation); + + /* nv01 rect */ + NVOBJ_CLASS(dev, 0x001e, GR); + NVOBJ_MTHD (dev, 0x001e, 0x0184, nv04_graph_mthd_bind_clip); + NVOBJ_MTHD (dev, 0x001e, 0x0188, nv04_graph_mthd_bind_nv01_patt); + NVOBJ_MTHD (dev, 0x001e, 0x018c, nv04_graph_mthd_bind_rop); + NVOBJ_MTHD (dev, 0x001e, 0x0190, nv04_graph_mthd_bind_beta1); + NVOBJ_MTHD (dev, 0x001e, 0x0194, nv04_graph_mthd_bind_surf_dst); + NVOBJ_MTHD (dev, 0x001e, 0x02fc, nv04_graph_mthd_set_operation); + + /* nv04 rect */ + NVOBJ_CLASS(dev, 0x005e, GR); + NVOBJ_MTHD (dev, 0x005e, 0x0184, nv04_graph_mthd_bind_clip); + NVOBJ_MTHD (dev, 0x005e, 0x0188, nv04_graph_mthd_bind_nv04_patt); + NVOBJ_MTHD (dev, 0x005e, 0x018c, nv04_graph_mthd_bind_rop); + NVOBJ_MTHD (dev, 0x005e, 0x0190, nv04_graph_mthd_bind_beta1); + NVOBJ_MTHD (dev, 0x005e, 0x0194, nv04_graph_mthd_bind_beta4); + NVOBJ_MTHD (dev, 0x005e, 0x0198, nv04_graph_mthd_bind_surf2d); + NVOBJ_MTHD (dev, 0x005e, 0x02fc, nv04_graph_mthd_set_operation); + + /* nvsw */ + NVOBJ_CLASS(dev, 0x506e, SW); + NVOBJ_MTHD (dev, 0x506e, 0x0150, nv04_graph_mthd_set_ref); + NVOBJ_MTHD (dev, 0x506e, 0x0500, nv04_graph_mthd_page_flip); + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nv04_instmem.c b/drivers/gpu/drm/nouveau/nv04_instmem.c new file mode 100644 index 00000000..b8611b95 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv04_instmem.c @@ -0,0 +1,166 @@ +#include "drmP.h" +#include "drm.h" +#include "nouveau_drv.h" +#include "nouveau_ramht.h" + +/* returns the size of fifo context */ +static int +nouveau_fifo_ctx_size(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + if (dev_priv->chipset >= 0x40) + return 128; + else + if (dev_priv->chipset >= 0x17) + return 64; + + return 32; +} + +int nv04_instmem_init(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *ramht = NULL; + u32 offset, length; + int ret; + + /* RAMIN always available */ + dev_priv->ramin_available = true; + + /* Setup shared RAMHT */ + ret = nouveau_gpuobj_new_fake(dev, 0x10000, ~0, 4096, + NVOBJ_FLAG_ZERO_ALLOC, &ramht); + if (ret) + return ret; + + ret = nouveau_ramht_new(dev, ramht, &dev_priv->ramht); + nouveau_gpuobj_ref(NULL, &ramht); + if (ret) + return ret; + + /* And RAMRO */ + ret = nouveau_gpuobj_new_fake(dev, 0x11200, ~0, 512, + NVOBJ_FLAG_ZERO_ALLOC, &dev_priv->ramro); + if (ret) + return ret; + + /* And RAMFC */ + length = dev_priv->engine.fifo.channels * nouveau_fifo_ctx_size(dev); + switch (dev_priv->card_type) { + case NV_40: + offset = 0x20000; + break; + default: + offset = 0x11400; + break; + } + + ret = nouveau_gpuobj_new_fake(dev, offset, ~0, length, + NVOBJ_FLAG_ZERO_ALLOC, &dev_priv->ramfc); + if (ret) + return ret; + + /* Only allow space after RAMFC to be used for object allocation */ + offset += length; + + /* It appears RAMRO (or something?) is controlled by 0x2220/0x2230 + * on certain NV4x chipsets as well as RAMFC. When 0x2230 == 0 + * ("new style" control) the upper 16-bits of 0x2220 points at this + * other mysterious table that's clobbering important things. + * + * We're now pointing this at RAMIN+0x30000 to avoid RAMFC getting + * smashed to pieces on us, so reserve 0x30000-0x40000 too.. + */ + if (dev_priv->card_type >= NV_40) { + if (offset < 0x40000) + offset = 0x40000; + } + + ret = drm_mm_init(&dev_priv->ramin_heap, offset, + dev_priv->ramin_rsvd_vram - offset); + if (ret) { + NV_ERROR(dev, "Failed to init RAMIN heap: %d\n", ret); + return ret; + } + + return 0; +} + +void +nv04_instmem_takedown(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + nouveau_ramht_ref(NULL, &dev_priv->ramht, NULL); + nouveau_gpuobj_ref(NULL, &dev_priv->ramro); + nouveau_gpuobj_ref(NULL, &dev_priv->ramfc); + + if (drm_mm_initialized(&dev_priv->ramin_heap)) + drm_mm_takedown(&dev_priv->ramin_heap); +} + +int +nv04_instmem_suspend(struct drm_device *dev) +{ + return 0; +} + +void +nv04_instmem_resume(struct drm_device *dev) +{ +} + +int +nv04_instmem_get(struct nouveau_gpuobj *gpuobj, u32 size, u32 align) +{ + struct drm_nouveau_private *dev_priv = gpuobj->dev->dev_private; + struct drm_mm_node *ramin = NULL; + + do { + if (drm_mm_pre_get(&dev_priv->ramin_heap)) + return -ENOMEM; + + spin_lock(&dev_priv->ramin_lock); + ramin = drm_mm_search_free(&dev_priv->ramin_heap, size, align, 0); + if (ramin == NULL) { + spin_unlock(&dev_priv->ramin_lock); + return -ENOMEM; + } + + ramin = drm_mm_get_block_atomic(ramin, size, align); + spin_unlock(&dev_priv->ramin_lock); + } while (ramin == NULL); + + gpuobj->node = ramin; + gpuobj->vinst = ramin->start; + return 0; +} + +void +nv04_instmem_put(struct nouveau_gpuobj *gpuobj) +{ + struct drm_nouveau_private *dev_priv = gpuobj->dev->dev_private; + + spin_lock(&dev_priv->ramin_lock); + drm_mm_put_block(gpuobj->node); + gpuobj->node = NULL; + spin_unlock(&dev_priv->ramin_lock); +} + +int +nv04_instmem_map(struct nouveau_gpuobj *gpuobj) +{ + gpuobj->pinst = gpuobj->vinst; + return 0; +} + +void +nv04_instmem_unmap(struct nouveau_gpuobj *gpuobj) +{ +} + +void +nv04_instmem_flush(struct drm_device *dev) +{ +} diff --git a/drivers/gpu/drm/nouveau/nv04_mc.c b/drivers/gpu/drm/nouveau/nv04_mc.c new file mode 100644 index 00000000..2af43a1c --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv04_mc.c @@ -0,0 +1,24 @@ +#include "drmP.h" +#include "drm.h" +#include "nouveau_drv.h" +#include "nouveau_drm.h" + +int +nv04_mc_init(struct drm_device *dev) +{ + /* Power up everything, resetting each individual unit will + * be done later if needed. + */ + + nv_wr32(dev, NV03_PMC_ENABLE, 0xFFFFFFFF); + + /* Disable PROM access. */ + nv_wr32(dev, NV_PBUS_PCI_NV_20, NV_PBUS_PCI_NV_20_ROM_SHADOW_ENABLED); + + return 0; +} + +void +nv04_mc_takedown(struct drm_device *dev) +{ +} diff --git a/drivers/gpu/drm/nouveau/nv04_pm.c b/drivers/gpu/drm/nouveau/nv04_pm.c new file mode 100644 index 00000000..eb1c70dd --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv04_pm.c @@ -0,0 +1,90 @@ +/* + * Copyright 2010 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include "drmP.h" +#include "nouveau_drv.h" +#include "nouveau_hw.h" +#include "nouveau_pm.h" + +struct nv04_pm_state { + struct pll_lims pll; + struct nouveau_pll_vals calc; +}; + +int +nv04_pm_clock_get(struct drm_device *dev, u32 id) +{ + return nouveau_hw_get_clock(dev, id); +} + +void * +nv04_pm_clock_pre(struct drm_device *dev, struct nouveau_pm_level *perflvl, + u32 id, int khz) +{ + struct nv04_pm_state *state; + int ret; + + state = kzalloc(sizeof(*state), GFP_KERNEL); + if (!state) + return ERR_PTR(-ENOMEM); + + ret = get_pll_limits(dev, id, &state->pll); + if (ret) { + kfree(state); + return (ret == -ENOENT) ? NULL : ERR_PTR(ret); + } + + ret = nouveau_calc_pll_mnp(dev, &state->pll, khz, &state->calc); + if (!ret) { + kfree(state); + return ERR_PTR(-EINVAL); + } + + return state; +} + +void +nv04_pm_clock_set(struct drm_device *dev, void *pre_state) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nv04_pm_state *state = pre_state; + u32 reg = state->pll.reg; + + /* thank the insane nouveau_hw_setpll() interface for this */ + if (dev_priv->card_type >= NV_40) + reg += 4; + + nouveau_hw_setpll(dev, reg, &state->calc); + + if (dev_priv->card_type < NV_30 && reg == NV_PRAMDAC_MPLL_COEFF) { + if (dev_priv->card_type == NV_20) + nv_mask(dev, 0x1002c4, 0, 1 << 20); + + /* Reset the DLLs */ + nv_mask(dev, 0x1002c0, 0, 1 << 8); + } + + kfree(state); +} + diff --git a/drivers/gpu/drm/nouveau/nv04_timer.c b/drivers/gpu/drm/nouveau/nv04_timer.c new file mode 100644 index 00000000..1d09ddd5 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv04_timer.c @@ -0,0 +1,51 @@ +#include "drmP.h" +#include "drm.h" +#include "nouveau_drv.h" +#include "nouveau_drm.h" + +int +nv04_timer_init(struct drm_device *dev) +{ + nv_wr32(dev, NV04_PTIMER_INTR_EN_0, 0x00000000); + nv_wr32(dev, NV04_PTIMER_INTR_0, 0xFFFFFFFF); + + /* Just use the pre-existing values when possible for now; these regs + * are not written in nv (driver writer missed a /4 on the address), and + * writing 8 and 3 to the correct regs breaks the timings on the LVDS + * hardware sequencing microcode. + * A correct solution (involving calculations with the GPU PLL) can + * be done when kernel modesetting lands + */ + if (!nv_rd32(dev, NV04_PTIMER_NUMERATOR) || + !nv_rd32(dev, NV04_PTIMER_DENOMINATOR)) { + nv_wr32(dev, NV04_PTIMER_NUMERATOR, 0x00000008); + nv_wr32(dev, NV04_PTIMER_DENOMINATOR, 0x00000003); + } + + return 0; +} + +uint64_t +nv04_timer_read(struct drm_device *dev) +{ + uint32_t low; + /* From kmmio dumps on nv28 this looks like how the blob does this. + * It reads the high dword twice, before and after. + * The only explanation seems to be that the 64-bit timer counter + * advances between high and low dword reads and may corrupt the + * result. Not confirmed. + */ + uint32_t high2 = nv_rd32(dev, NV04_PTIMER_TIME_1); + uint32_t high1; + do { + high1 = high2; + low = nv_rd32(dev, NV04_PTIMER_TIME_0); + high2 = nv_rd32(dev, NV04_PTIMER_TIME_1); + } while (high1 != high2); + return (((uint64_t)high2) << 32) | (uint64_t)low; +} + +void +nv04_timer_takedown(struct drm_device *dev) +{ +} diff --git a/drivers/gpu/drm/nouveau/nv04_tv.c b/drivers/gpu/drm/nouveau/nv04_tv.c new file mode 100644 index 00000000..3eb605dd --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv04_tv.c @@ -0,0 +1,254 @@ +/* + * Copyright (C) 2009 Francisco Jerez. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "drmP.h" +#include "nouveau_drv.h" +#include "nouveau_encoder.h" +#include "nouveau_connector.h" +#include "nouveau_crtc.h" +#include "nouveau_hw.h" +#include "drm_crtc_helper.h" + +#include "i2c/ch7006.h" + +static struct i2c_board_info nv04_tv_encoder_info[] = { + { + I2C_BOARD_INFO("ch7006", 0x75), + .platform_data = &(struct ch7006_encoder_params) { + CH7006_FORMAT_RGB24m12I, CH7006_CLOCK_MASTER, + 0, 0, 0, + CH7006_SYNC_SLAVE, CH7006_SYNC_SEPARATED, + CH7006_POUT_3_3V, CH7006_ACTIVE_HSYNC + } + }, + { } +}; + +int nv04_tv_identify(struct drm_device *dev, int i2c_index) +{ + return nouveau_i2c_identify(dev, "TV encoder", nv04_tv_encoder_info, + NULL, i2c_index); +} + + +#define PLLSEL_TV_CRTC1_MASK \ + (NV_PRAMDAC_PLL_COEFF_SELECT_TV_VSCLK1 \ + | NV_PRAMDAC_PLL_COEFF_SELECT_TV_PCLK1) +#define PLLSEL_TV_CRTC2_MASK \ + (NV_PRAMDAC_PLL_COEFF_SELECT_TV_VSCLK2 \ + | NV_PRAMDAC_PLL_COEFF_SELECT_TV_PCLK2) + +static void nv04_tv_dpms(struct drm_encoder *encoder, int mode) +{ + struct drm_device *dev = encoder->dev; + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nv04_mode_state *state = &dev_priv->mode_reg; + uint8_t crtc1A; + + NV_INFO(dev, "Setting dpms mode %d on TV encoder (output %d)\n", + mode, nv_encoder->dcb->index); + + state->pllsel &= ~(PLLSEL_TV_CRTC1_MASK | PLLSEL_TV_CRTC2_MASK); + + if (mode == DRM_MODE_DPMS_ON) { + int head = nouveau_crtc(encoder->crtc)->index; + crtc1A = NVReadVgaCrtc(dev, head, NV_CIO_CRE_RPC1_INDEX); + + state->pllsel |= head ? PLLSEL_TV_CRTC2_MASK : + PLLSEL_TV_CRTC1_MASK; + + /* Inhibit hsync */ + crtc1A |= 0x80; + + NVWriteVgaCrtc(dev, head, NV_CIO_CRE_RPC1_INDEX, crtc1A); + } + + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_PLL_COEFF_SELECT, state->pllsel); + + get_slave_funcs(encoder)->dpms(encoder, mode); +} + +static void nv04_tv_bind(struct drm_device *dev, int head, bool bind) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nv04_crtc_reg *state = &dev_priv->mode_reg.crtc_reg[head]; + + state->tv_setup = 0; + + if (bind) + state->CRTC[NV_CIO_CRE_49] |= 0x10; + else + state->CRTC[NV_CIO_CRE_49] &= ~0x10; + + NVWriteVgaCrtc(dev, head, NV_CIO_CRE_LCD__INDEX, + state->CRTC[NV_CIO_CRE_LCD__INDEX]); + NVWriteVgaCrtc(dev, head, NV_CIO_CRE_49, + state->CRTC[NV_CIO_CRE_49]); + NVWriteRAMDAC(dev, head, NV_PRAMDAC_TV_SETUP, + state->tv_setup); +} + +static void nv04_tv_prepare(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + int head = nouveau_crtc(encoder->crtc)->index; + struct drm_encoder_helper_funcs *helper = encoder->helper_private; + + helper->dpms(encoder, DRM_MODE_DPMS_OFF); + + nv04_dfp_disable(dev, head); + + if (nv_two_heads(dev)) + nv04_tv_bind(dev, head ^ 1, false); + + nv04_tv_bind(dev, head, true); +} + +static void nv04_tv_mode_set(struct drm_encoder *encoder, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct drm_device *dev = encoder->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_crtc *nv_crtc = nouveau_crtc(encoder->crtc); + struct nv04_crtc_reg *regp = &dev_priv->mode_reg.crtc_reg[nv_crtc->index]; + + regp->tv_htotal = adjusted_mode->htotal; + regp->tv_vtotal = adjusted_mode->vtotal; + + /* These delay the TV signals with respect to the VGA port, + * they might be useful if we ever allow a CRTC to drive + * multiple outputs. + */ + regp->tv_hskew = 1; + regp->tv_hsync_delay = 1; + regp->tv_hsync_delay2 = 64; + regp->tv_vskew = 1; + regp->tv_vsync_delay = 1; + + get_slave_funcs(encoder)->mode_set(encoder, mode, adjusted_mode); +} + +static void nv04_tv_commit(struct drm_encoder *encoder) +{ + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + struct drm_device *dev = encoder->dev; + struct nouveau_crtc *nv_crtc = nouveau_crtc(encoder->crtc); + struct drm_encoder_helper_funcs *helper = encoder->helper_private; + + helper->dpms(encoder, DRM_MODE_DPMS_ON); + + NV_INFO(dev, "Output %s is running on CRTC %d using output %c\n", + drm_get_connector_name(&nouveau_encoder_connector_get(nv_encoder)->base), nv_crtc->index, + '@' + ffs(nv_encoder->dcb->or)); +} + +static void nv04_tv_destroy(struct drm_encoder *encoder) +{ + get_slave_funcs(encoder)->destroy(encoder); + drm_encoder_cleanup(encoder); + + kfree(encoder->helper_private); + kfree(nouveau_encoder(encoder)); +} + +static const struct drm_encoder_funcs nv04_tv_funcs = { + .destroy = nv04_tv_destroy, +}; + +int +nv04_tv_create(struct drm_connector *connector, struct dcb_entry *entry) +{ + struct nouveau_encoder *nv_encoder; + struct drm_encoder *encoder; + struct drm_device *dev = connector->dev; + struct drm_encoder_helper_funcs *hfuncs; + struct drm_encoder_slave_funcs *sfuncs; + struct nouveau_i2c_chan *i2c = + nouveau_i2c_find(dev, entry->i2c_index); + int type, ret; + + /* Ensure that we can talk to this encoder */ + type = nv04_tv_identify(dev, entry->i2c_index); + if (type < 0) + return type; + + /* Allocate the necessary memory */ + nv_encoder = kzalloc(sizeof(*nv_encoder), GFP_KERNEL); + if (!nv_encoder) + return -ENOMEM; + + hfuncs = kzalloc(sizeof(*hfuncs), GFP_KERNEL); + if (!hfuncs) { + ret = -ENOMEM; + goto fail_free; + } + + /* Initialize the common members */ + encoder = to_drm_encoder(nv_encoder); + + drm_encoder_init(dev, encoder, &nv04_tv_funcs, DRM_MODE_ENCODER_TVDAC); + drm_encoder_helper_add(encoder, hfuncs); + + encoder->possible_crtcs = entry->heads; + encoder->possible_clones = 0; + nv_encoder->dcb = entry; + nv_encoder->or = ffs(entry->or) - 1; + + /* Run the slave-specific initialization */ + ret = drm_i2c_encoder_init(dev, to_encoder_slave(encoder), + &i2c->adapter, &nv04_tv_encoder_info[type]); + if (ret < 0) + goto fail_cleanup; + + /* Fill the function pointers */ + sfuncs = get_slave_funcs(encoder); + + *hfuncs = (struct drm_encoder_helper_funcs) { + .dpms = nv04_tv_dpms, + .save = sfuncs->save, + .restore = sfuncs->restore, + .mode_fixup = sfuncs->mode_fixup, + .prepare = nv04_tv_prepare, + .commit = nv04_tv_commit, + .mode_set = nv04_tv_mode_set, + .detect = sfuncs->detect, + }; + + /* Attach it to the specified connector. */ + sfuncs->create_resources(encoder, connector); + drm_mode_connector_attach_encoder(connector, encoder); + + return 0; + +fail_cleanup: + drm_encoder_cleanup(encoder); + kfree(hfuncs); +fail_free: + kfree(nv_encoder); + return ret; +} diff --git a/drivers/gpu/drm/nouveau/nv10_fb.c b/drivers/gpu/drm/nouveau/nv10_fb.c new file mode 100644 index 00000000..f78181a5 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv10_fb.c @@ -0,0 +1,144 @@ +#include "drmP.h" +#include "drm.h" +#include "nouveau_drv.h" +#include "nouveau_drm.h" + +static struct drm_mm_node * +nv20_fb_alloc_tag(struct drm_device *dev, uint32_t size) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fb_engine *pfb = &dev_priv->engine.fb; + struct drm_mm_node *mem; + int ret; + + ret = drm_mm_pre_get(&pfb->tag_heap); + if (ret) + return NULL; + + spin_lock(&dev_priv->tile.lock); + mem = drm_mm_search_free(&pfb->tag_heap, size, 0, 0); + if (mem) + mem = drm_mm_get_block_atomic(mem, size, 0); + spin_unlock(&dev_priv->tile.lock); + + return mem; +} + +static void +nv20_fb_free_tag(struct drm_device *dev, struct drm_mm_node *mem) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + spin_lock(&dev_priv->tile.lock); + drm_mm_put_block(mem); + spin_unlock(&dev_priv->tile.lock); +} + +void +nv10_fb_init_tile_region(struct drm_device *dev, int i, uint32_t addr, + uint32_t size, uint32_t pitch, uint32_t flags) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_tile_reg *tile = &dev_priv->tile.reg[i]; + int bpp = (flags & NOUVEAU_GEM_TILE_32BPP ? 32 : 16); + + tile->addr = addr; + tile->limit = max(1u, addr + size) - 1; + tile->pitch = pitch; + + if (dev_priv->card_type == NV_20) { + if (flags & NOUVEAU_GEM_TILE_ZETA) { + /* + * Allocate some of the on-die tag memory, + * used to store Z compression meta-data (most + * likely just a bitmap determining if a given + * tile is compressed or not). + */ + tile->tag_mem = nv20_fb_alloc_tag(dev, size / 256); + + if (tile->tag_mem) { + /* Enable Z compression */ + if (dev_priv->chipset >= 0x25) + tile->zcomp = tile->tag_mem->start | + (bpp == 16 ? + NV25_PFB_ZCOMP_MODE_16 : + NV25_PFB_ZCOMP_MODE_32); + else + tile->zcomp = tile->tag_mem->start | + NV20_PFB_ZCOMP_EN | + (bpp == 16 ? 0 : + NV20_PFB_ZCOMP_MODE_32); + } + + tile->addr |= 3; + } else { + tile->addr |= 1; + } + + } else { + tile->addr |= 1 << 31; + } +} + +void +nv10_fb_free_tile_region(struct drm_device *dev, int i) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_tile_reg *tile = &dev_priv->tile.reg[i]; + + if (tile->tag_mem) { + nv20_fb_free_tag(dev, tile->tag_mem); + tile->tag_mem = NULL; + } + + tile->addr = tile->limit = tile->pitch = tile->zcomp = 0; +} + +void +nv10_fb_set_tile_region(struct drm_device *dev, int i) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_tile_reg *tile = &dev_priv->tile.reg[i]; + + nv_wr32(dev, NV10_PFB_TLIMIT(i), tile->limit); + nv_wr32(dev, NV10_PFB_TSIZE(i), tile->pitch); + nv_wr32(dev, NV10_PFB_TILE(i), tile->addr); + + if (dev_priv->card_type == NV_20) + nv_wr32(dev, NV20_PFB_ZCOMP(i), tile->zcomp); +} + +int +nv10_fb_init(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fb_engine *pfb = &dev_priv->engine.fb; + int i; + + pfb->num_tiles = NV10_PFB_TILE__SIZE; + + if (dev_priv->card_type == NV_20) + drm_mm_init(&pfb->tag_heap, 0, + (dev_priv->chipset >= 0x25 ? + 64 * 1024 : 32 * 1024)); + + /* Turn all the tiling regions off. */ + for (i = 0; i < pfb->num_tiles; i++) + pfb->set_tile_region(dev, i); + + return 0; +} + +void +nv10_fb_takedown(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fb_engine *pfb = &dev_priv->engine.fb; + int i; + + for (i = 0; i < pfb->num_tiles; i++) + pfb->free_tile_region(dev, i); + + if (dev_priv->card_type == NV_20) + drm_mm_takedown(&pfb->tag_heap); +} diff --git a/drivers/gpu/drm/nouveau/nv10_fifo.c b/drivers/gpu/drm/nouveau/nv10_fifo.c new file mode 100644 index 00000000..d2ecbff4 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv10_fifo.c @@ -0,0 +1,246 @@ +/* + * Copyright (C) 2007 Ben Skeggs. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "drmP.h" +#include "drm.h" +#include "nouveau_drv.h" +#include "nouveau_ramht.h" + +#define NV10_RAMFC(c) (dev_priv->ramfc->pinst + ((c) * NV10_RAMFC__SIZE)) +#define NV10_RAMFC__SIZE ((dev_priv->chipset) >= 0x17 ? 64 : 32) + +int +nv10_fifo_channel_id(struct drm_device *dev) +{ + return nv_rd32(dev, NV03_PFIFO_CACHE1_PUSH1) & + NV10_PFIFO_CACHE1_PUSH1_CHID_MASK; +} + +int +nv10_fifo_create_context(struct nouveau_channel *chan) +{ + struct drm_nouveau_private *dev_priv = chan->dev->dev_private; + struct drm_device *dev = chan->dev; + uint32_t fc = NV10_RAMFC(chan->id); + int ret; + + ret = nouveau_gpuobj_new_fake(dev, NV10_RAMFC(chan->id), ~0, + NV10_RAMFC__SIZE, NVOBJ_FLAG_ZERO_ALLOC | + NVOBJ_FLAG_ZERO_FREE, &chan->ramfc); + if (ret) + return ret; + + chan->user = ioremap(pci_resource_start(dev->pdev, 0) + + NV03_USER(chan->id), PAGE_SIZE); + if (!chan->user) + return -ENOMEM; + + /* Fill entries that are seen filled in dumps of nvidia driver just + * after channel's is put into DMA mode + */ + nv_wi32(dev, fc + 0, chan->pushbuf_base); + nv_wi32(dev, fc + 4, chan->pushbuf_base); + nv_wi32(dev, fc + 12, chan->pushbuf->pinst >> 4); + nv_wi32(dev, fc + 20, NV_PFIFO_CACHE1_DMA_FETCH_TRIG_128_BYTES | + NV_PFIFO_CACHE1_DMA_FETCH_SIZE_128_BYTES | + NV_PFIFO_CACHE1_DMA_FETCH_MAX_REQS_8 | +#ifdef __BIG_ENDIAN + NV_PFIFO_CACHE1_BIG_ENDIAN | +#endif + 0); + + /* enable the fifo dma operation */ + nv_wr32(dev, NV04_PFIFO_MODE, + nv_rd32(dev, NV04_PFIFO_MODE) | (1 << chan->id)); + return 0; +} + +static void +nv10_fifo_do_load_context(struct drm_device *dev, int chid) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + uint32_t fc = NV10_RAMFC(chid), tmp; + + nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_PUT, nv_ri32(dev, fc + 0)); + nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_GET, nv_ri32(dev, fc + 4)); + nv_wr32(dev, NV10_PFIFO_CACHE1_REF_CNT, nv_ri32(dev, fc + 8)); + + tmp = nv_ri32(dev, fc + 12); + nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_INSTANCE, tmp & 0xFFFF); + nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_DCOUNT, tmp >> 16); + + nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_STATE, nv_ri32(dev, fc + 16)); + nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_FETCH, nv_ri32(dev, fc + 20)); + nv_wr32(dev, NV04_PFIFO_CACHE1_ENGINE, nv_ri32(dev, fc + 24)); + nv_wr32(dev, NV04_PFIFO_CACHE1_PULL1, nv_ri32(dev, fc + 28)); + + if (dev_priv->chipset < 0x17) + goto out; + + nv_wr32(dev, NV10_PFIFO_CACHE1_ACQUIRE_VALUE, nv_ri32(dev, fc + 32)); + tmp = nv_ri32(dev, fc + 36); + nv_wr32(dev, NV10_PFIFO_CACHE1_ACQUIRE_TIMESTAMP, tmp); + nv_wr32(dev, NV10_PFIFO_CACHE1_ACQUIRE_TIMEOUT, nv_ri32(dev, fc + 40)); + nv_wr32(dev, NV10_PFIFO_CACHE1_SEMAPHORE, nv_ri32(dev, fc + 44)); + nv_wr32(dev, NV10_PFIFO_CACHE1_DMA_SUBROUTINE, nv_ri32(dev, fc + 48)); + +out: + nv_wr32(dev, NV03_PFIFO_CACHE1_GET, 0); + nv_wr32(dev, NV03_PFIFO_CACHE1_PUT, 0); +} + +int +nv10_fifo_load_context(struct nouveau_channel *chan) +{ + struct drm_device *dev = chan->dev; + uint32_t tmp; + + nv10_fifo_do_load_context(dev, chan->id); + + nv_wr32(dev, NV03_PFIFO_CACHE1_PUSH1, + NV03_PFIFO_CACHE1_PUSH1_DMA | chan->id); + nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_PUSH, 1); + + /* Reset NV04_PFIFO_CACHE1_DMA_CTL_AT_INFO to INVALID */ + tmp = nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_CTL) & ~(1 << 31); + nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_CTL, tmp); + + return 0; +} + +int +nv10_fifo_unload_context(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; + uint32_t fc, tmp; + int chid; + + chid = pfifo->channel_id(dev); + if (chid < 0 || chid >= dev_priv->engine.fifo.channels) + return 0; + fc = NV10_RAMFC(chid); + + nv_wi32(dev, fc + 0, nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_PUT)); + nv_wi32(dev, fc + 4, nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_GET)); + nv_wi32(dev, fc + 8, nv_rd32(dev, NV10_PFIFO_CACHE1_REF_CNT)); + tmp = nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_INSTANCE) & 0xFFFF; + tmp |= (nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_DCOUNT) << 16); + nv_wi32(dev, fc + 12, tmp); + nv_wi32(dev, fc + 16, nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_STATE)); + nv_wi32(dev, fc + 20, nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_FETCH)); + nv_wi32(dev, fc + 24, nv_rd32(dev, NV04_PFIFO_CACHE1_ENGINE)); + nv_wi32(dev, fc + 28, nv_rd32(dev, NV04_PFIFO_CACHE1_PULL1)); + + if (dev_priv->chipset < 0x17) + goto out; + + nv_wi32(dev, fc + 32, nv_rd32(dev, NV10_PFIFO_CACHE1_ACQUIRE_VALUE)); + tmp = nv_rd32(dev, NV10_PFIFO_CACHE1_ACQUIRE_TIMESTAMP); + nv_wi32(dev, fc + 36, tmp); + nv_wi32(dev, fc + 40, nv_rd32(dev, NV10_PFIFO_CACHE1_ACQUIRE_TIMEOUT)); + nv_wi32(dev, fc + 44, nv_rd32(dev, NV10_PFIFO_CACHE1_SEMAPHORE)); + nv_wi32(dev, fc + 48, nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_GET)); + +out: + nv10_fifo_do_load_context(dev, pfifo->channels - 1); + nv_wr32(dev, NV03_PFIFO_CACHE1_PUSH1, pfifo->channels - 1); + return 0; +} + +static void +nv10_fifo_init_reset(struct drm_device *dev) +{ + nv_wr32(dev, NV03_PMC_ENABLE, + nv_rd32(dev, NV03_PMC_ENABLE) & ~NV_PMC_ENABLE_PFIFO); + nv_wr32(dev, NV03_PMC_ENABLE, + nv_rd32(dev, NV03_PMC_ENABLE) | NV_PMC_ENABLE_PFIFO); + + nv_wr32(dev, 0x003224, 0x000f0078); + nv_wr32(dev, 0x002044, 0x0101ffff); + nv_wr32(dev, 0x002040, 0x000000ff); + nv_wr32(dev, 0x002500, 0x00000000); + nv_wr32(dev, 0x003000, 0x00000000); + nv_wr32(dev, 0x003050, 0x00000000); + + nv_wr32(dev, 0x003258, 0x00000000); + nv_wr32(dev, 0x003210, 0x00000000); + nv_wr32(dev, 0x003270, 0x00000000); +} + +static void +nv10_fifo_init_ramxx(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + nv_wr32(dev, NV03_PFIFO_RAMHT, (0x03 << 24) /* search 128 */ | + ((dev_priv->ramht->bits - 9) << 16) | + (dev_priv->ramht->gpuobj->pinst >> 8)); + nv_wr32(dev, NV03_PFIFO_RAMRO, dev_priv->ramro->pinst >> 8); + + if (dev_priv->chipset < 0x17) { + nv_wr32(dev, NV03_PFIFO_RAMFC, dev_priv->ramfc->pinst >> 8); + } else { + nv_wr32(dev, NV03_PFIFO_RAMFC, (dev_priv->ramfc->pinst >> 8) | + (1 << 16) /* 64 Bytes entry*/); + /* XXX nvidia blob set bit 18, 21,23 for nv20 & nv30 */ + } +} + +static void +nv10_fifo_init_intr(struct drm_device *dev) +{ + nouveau_irq_register(dev, 8, nv04_fifo_isr); + nv_wr32(dev, 0x002100, 0xffffffff); + nv_wr32(dev, 0x002140, 0xffffffff); +} + +int +nv10_fifo_init(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; + int i; + + nv10_fifo_init_reset(dev); + nv10_fifo_init_ramxx(dev); + + nv10_fifo_do_load_context(dev, pfifo->channels - 1); + nv_wr32(dev, NV03_PFIFO_CACHE1_PUSH1, pfifo->channels - 1); + + nv10_fifo_init_intr(dev); + pfifo->enable(dev); + pfifo->reassign(dev, true); + + for (i = 0; i < dev_priv->engine.fifo.channels; i++) { + if (dev_priv->channels.ptr[i]) { + uint32_t mode = nv_rd32(dev, NV04_PFIFO_MODE); + nv_wr32(dev, NV04_PFIFO_MODE, mode | (1 << i)); + } + } + + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nv10_gpio.c b/drivers/gpu/drm/nouveau/nv10_gpio.c new file mode 100644 index 00000000..007fc29e --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv10_gpio.c @@ -0,0 +1,92 @@ +/* + * Copyright (C) 2009 Francisco Jerez. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "drmP.h" +#include "nouveau_drv.h" +#include "nouveau_hw.h" + +static bool +get_gpio_location(struct dcb_gpio_entry *ent, uint32_t *reg, uint32_t *shift, + uint32_t *mask) +{ + if (ent->line < 2) { + *reg = NV_PCRTC_GPIO; + *shift = ent->line * 16; + *mask = 0x11; + + } else if (ent->line < 10) { + *reg = NV_PCRTC_GPIO_EXT; + *shift = (ent->line - 2) * 4; + *mask = 0x3; + + } else if (ent->line < 14) { + *reg = NV_PCRTC_850; + *shift = (ent->line - 10) * 4; + *mask = 0x3; + + } else { + return false; + } + + return true; +} + +int +nv10_gpio_get(struct drm_device *dev, enum dcb_gpio_tag tag) +{ + struct dcb_gpio_entry *ent = nouveau_bios_gpio_entry(dev, tag); + uint32_t reg, shift, mask, value; + + if (!ent) + return -ENODEV; + + if (!get_gpio_location(ent, ®, &shift, &mask)) + return -ENODEV; + + value = NVReadCRTC(dev, 0, reg) >> shift; + + return (ent->invert ? 1 : 0) ^ (value & 1); +} + +int +nv10_gpio_set(struct drm_device *dev, enum dcb_gpio_tag tag, int state) +{ + struct dcb_gpio_entry *ent = nouveau_bios_gpio_entry(dev, tag); + uint32_t reg, shift, mask, value; + + if (!ent) + return -ENODEV; + + if (!get_gpio_location(ent, ®, &shift, &mask)) + return -ENODEV; + + value = ((ent->invert ? 1 : 0) ^ (state ? 1 : 0)) << shift; + mask = ~(mask << shift); + + NVWriteCRTC(dev, 0, reg, value | (NVReadCRTC(dev, 0, reg) & mask)); + + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nv10_graph.c b/drivers/gpu/drm/nouveau/nv10_graph.c new file mode 100644 index 00000000..0930c6cb --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv10_graph.c @@ -0,0 +1,1189 @@ +/* + * Copyright 2007 Matthieu CASTET + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + */ + +#include "drmP.h" +#include "drm.h" +#include "nouveau_drm.h" +#include "nouveau_drv.h" +#include "nouveau_util.h" + +struct nv10_graph_engine { + struct nouveau_exec_engine base; +}; + +struct pipe_state { + uint32_t pipe_0x0000[0x040/4]; + uint32_t pipe_0x0040[0x010/4]; + uint32_t pipe_0x0200[0x0c0/4]; + uint32_t pipe_0x4400[0x080/4]; + uint32_t pipe_0x6400[0x3b0/4]; + uint32_t pipe_0x6800[0x2f0/4]; + uint32_t pipe_0x6c00[0x030/4]; + uint32_t pipe_0x7000[0x130/4]; + uint32_t pipe_0x7400[0x0c0/4]; + uint32_t pipe_0x7800[0x0c0/4]; +}; + +static int nv10_graph_ctx_regs[] = { + NV10_PGRAPH_CTX_SWITCH(0), + NV10_PGRAPH_CTX_SWITCH(1), + NV10_PGRAPH_CTX_SWITCH(2), + NV10_PGRAPH_CTX_SWITCH(3), + NV10_PGRAPH_CTX_SWITCH(4), + NV10_PGRAPH_CTX_CACHE(0, 0), + NV10_PGRAPH_CTX_CACHE(0, 1), + NV10_PGRAPH_CTX_CACHE(0, 2), + NV10_PGRAPH_CTX_CACHE(0, 3), + NV10_PGRAPH_CTX_CACHE(0, 4), + NV10_PGRAPH_CTX_CACHE(1, 0), + NV10_PGRAPH_CTX_CACHE(1, 1), + NV10_PGRAPH_CTX_CACHE(1, 2), + NV10_PGRAPH_CTX_CACHE(1, 3), + NV10_PGRAPH_CTX_CACHE(1, 4), + NV10_PGRAPH_CTX_CACHE(2, 0), + NV10_PGRAPH_CTX_CACHE(2, 1), + NV10_PGRAPH_CTX_CACHE(2, 2), + NV10_PGRAPH_CTX_CACHE(2, 3), + NV10_PGRAPH_CTX_CACHE(2, 4), + NV10_PGRAPH_CTX_CACHE(3, 0), + NV10_PGRAPH_CTX_CACHE(3, 1), + NV10_PGRAPH_CTX_CACHE(3, 2), + NV10_PGRAPH_CTX_CACHE(3, 3), + NV10_PGRAPH_CTX_CACHE(3, 4), + NV10_PGRAPH_CTX_CACHE(4, 0), + NV10_PGRAPH_CTX_CACHE(4, 1), + NV10_PGRAPH_CTX_CACHE(4, 2), + NV10_PGRAPH_CTX_CACHE(4, 3), + NV10_PGRAPH_CTX_CACHE(4, 4), + NV10_PGRAPH_CTX_CACHE(5, 0), + NV10_PGRAPH_CTX_CACHE(5, 1), + NV10_PGRAPH_CTX_CACHE(5, 2), + NV10_PGRAPH_CTX_CACHE(5, 3), + NV10_PGRAPH_CTX_CACHE(5, 4), + NV10_PGRAPH_CTX_CACHE(6, 0), + NV10_PGRAPH_CTX_CACHE(6, 1), + NV10_PGRAPH_CTX_CACHE(6, 2), + NV10_PGRAPH_CTX_CACHE(6, 3), + NV10_PGRAPH_CTX_CACHE(6, 4), + NV10_PGRAPH_CTX_CACHE(7, 0), + NV10_PGRAPH_CTX_CACHE(7, 1), + NV10_PGRAPH_CTX_CACHE(7, 2), + NV10_PGRAPH_CTX_CACHE(7, 3), + NV10_PGRAPH_CTX_CACHE(7, 4), + NV10_PGRAPH_CTX_USER, + NV04_PGRAPH_DMA_START_0, + NV04_PGRAPH_DMA_START_1, + NV04_PGRAPH_DMA_LENGTH, + NV04_PGRAPH_DMA_MISC, + NV10_PGRAPH_DMA_PITCH, + NV04_PGRAPH_BOFFSET0, + NV04_PGRAPH_BBASE0, + NV04_PGRAPH_BLIMIT0, + NV04_PGRAPH_BOFFSET1, + NV04_PGRAPH_BBASE1, + NV04_PGRAPH_BLIMIT1, + NV04_PGRAPH_BOFFSET2, + NV04_PGRAPH_BBASE2, + NV04_PGRAPH_BLIMIT2, + NV04_PGRAPH_BOFFSET3, + NV04_PGRAPH_BBASE3, + NV04_PGRAPH_BLIMIT3, + NV04_PGRAPH_BOFFSET4, + NV04_PGRAPH_BBASE4, + NV04_PGRAPH_BLIMIT4, + NV04_PGRAPH_BOFFSET5, + NV04_PGRAPH_BBASE5, + NV04_PGRAPH_BLIMIT5, + NV04_PGRAPH_BPITCH0, + NV04_PGRAPH_BPITCH1, + NV04_PGRAPH_BPITCH2, + NV04_PGRAPH_BPITCH3, + NV04_PGRAPH_BPITCH4, + NV10_PGRAPH_SURFACE, + NV10_PGRAPH_STATE, + NV04_PGRAPH_BSWIZZLE2, + NV04_PGRAPH_BSWIZZLE5, + NV04_PGRAPH_BPIXEL, + NV10_PGRAPH_NOTIFY, + NV04_PGRAPH_PATT_COLOR0, + NV04_PGRAPH_PATT_COLOR1, + NV04_PGRAPH_PATT_COLORRAM, /* 64 values from 0x400900 to 0x4009fc */ + 0x00400904, + 0x00400908, + 0x0040090c, + 0x00400910, + 0x00400914, + 0x00400918, + 0x0040091c, + 0x00400920, + 0x00400924, + 0x00400928, + 0x0040092c, + 0x00400930, + 0x00400934, + 0x00400938, + 0x0040093c, + 0x00400940, + 0x00400944, + 0x00400948, + 0x0040094c, + 0x00400950, + 0x00400954, + 0x00400958, + 0x0040095c, + 0x00400960, + 0x00400964, + 0x00400968, + 0x0040096c, + 0x00400970, + 0x00400974, + 0x00400978, + 0x0040097c, + 0x00400980, + 0x00400984, + 0x00400988, + 0x0040098c, + 0x00400990, + 0x00400994, + 0x00400998, + 0x0040099c, + 0x004009a0, + 0x004009a4, + 0x004009a8, + 0x004009ac, + 0x004009b0, + 0x004009b4, + 0x004009b8, + 0x004009bc, + 0x004009c0, + 0x004009c4, + 0x004009c8, + 0x004009cc, + 0x004009d0, + 0x004009d4, + 0x004009d8, + 0x004009dc, + 0x004009e0, + 0x004009e4, + 0x004009e8, + 0x004009ec, + 0x004009f0, + 0x004009f4, + 0x004009f8, + 0x004009fc, + NV04_PGRAPH_PATTERN, /* 2 values from 0x400808 to 0x40080c */ + 0x0040080c, + NV04_PGRAPH_PATTERN_SHAPE, + NV03_PGRAPH_MONO_COLOR0, + NV04_PGRAPH_ROP3, + NV04_PGRAPH_CHROMA, + NV04_PGRAPH_BETA_AND, + NV04_PGRAPH_BETA_PREMULT, + 0x00400e70, + 0x00400e74, + 0x00400e78, + 0x00400e7c, + 0x00400e80, + 0x00400e84, + 0x00400e88, + 0x00400e8c, + 0x00400ea0, + 0x00400ea4, + 0x00400ea8, + 0x00400e90, + 0x00400e94, + 0x00400e98, + 0x00400e9c, + NV10_PGRAPH_WINDOWCLIP_HORIZONTAL, /* 8 values from 0x400f00-0x400f1c */ + NV10_PGRAPH_WINDOWCLIP_VERTICAL, /* 8 values from 0x400f20-0x400f3c */ + 0x00400f04, + 0x00400f24, + 0x00400f08, + 0x00400f28, + 0x00400f0c, + 0x00400f2c, + 0x00400f10, + 0x00400f30, + 0x00400f14, + 0x00400f34, + 0x00400f18, + 0x00400f38, + 0x00400f1c, + 0x00400f3c, + NV10_PGRAPH_XFMODE0, + NV10_PGRAPH_XFMODE1, + NV10_PGRAPH_GLOBALSTATE0, + NV10_PGRAPH_GLOBALSTATE1, + NV04_PGRAPH_STORED_FMT, + NV04_PGRAPH_SOURCE_COLOR, + NV03_PGRAPH_ABS_X_RAM, /* 32 values from 0x400400 to 0x40047c */ + NV03_PGRAPH_ABS_Y_RAM, /* 32 values from 0x400480 to 0x4004fc */ + 0x00400404, + 0x00400484, + 0x00400408, + 0x00400488, + 0x0040040c, + 0x0040048c, + 0x00400410, + 0x00400490, + 0x00400414, + 0x00400494, + 0x00400418, + 0x00400498, + 0x0040041c, + 0x0040049c, + 0x00400420, + 0x004004a0, + 0x00400424, + 0x004004a4, + 0x00400428, + 0x004004a8, + 0x0040042c, + 0x004004ac, + 0x00400430, + 0x004004b0, + 0x00400434, + 0x004004b4, + 0x00400438, + 0x004004b8, + 0x0040043c, + 0x004004bc, + 0x00400440, + 0x004004c0, + 0x00400444, + 0x004004c4, + 0x00400448, + 0x004004c8, + 0x0040044c, + 0x004004cc, + 0x00400450, + 0x004004d0, + 0x00400454, + 0x004004d4, + 0x00400458, + 0x004004d8, + 0x0040045c, + 0x004004dc, + 0x00400460, + 0x004004e0, + 0x00400464, + 0x004004e4, + 0x00400468, + 0x004004e8, + 0x0040046c, + 0x004004ec, + 0x00400470, + 0x004004f0, + 0x00400474, + 0x004004f4, + 0x00400478, + 0x004004f8, + 0x0040047c, + 0x004004fc, + NV03_PGRAPH_ABS_UCLIP_XMIN, + NV03_PGRAPH_ABS_UCLIP_XMAX, + NV03_PGRAPH_ABS_UCLIP_YMIN, + NV03_PGRAPH_ABS_UCLIP_YMAX, + 0x00400550, + 0x00400558, + 0x00400554, + 0x0040055c, + NV03_PGRAPH_ABS_UCLIPA_XMIN, + NV03_PGRAPH_ABS_UCLIPA_XMAX, + NV03_PGRAPH_ABS_UCLIPA_YMIN, + NV03_PGRAPH_ABS_UCLIPA_YMAX, + NV03_PGRAPH_ABS_ICLIP_XMAX, + NV03_PGRAPH_ABS_ICLIP_YMAX, + NV03_PGRAPH_XY_LOGIC_MISC0, + NV03_PGRAPH_XY_LOGIC_MISC1, + NV03_PGRAPH_XY_LOGIC_MISC2, + NV03_PGRAPH_XY_LOGIC_MISC3, + NV03_PGRAPH_CLIPX_0, + NV03_PGRAPH_CLIPX_1, + NV03_PGRAPH_CLIPY_0, + NV03_PGRAPH_CLIPY_1, + NV10_PGRAPH_COMBINER0_IN_ALPHA, + NV10_PGRAPH_COMBINER1_IN_ALPHA, + NV10_PGRAPH_COMBINER0_IN_RGB, + NV10_PGRAPH_COMBINER1_IN_RGB, + NV10_PGRAPH_COMBINER_COLOR0, + NV10_PGRAPH_COMBINER_COLOR1, + NV10_PGRAPH_COMBINER0_OUT_ALPHA, + NV10_PGRAPH_COMBINER1_OUT_ALPHA, + NV10_PGRAPH_COMBINER0_OUT_RGB, + NV10_PGRAPH_COMBINER1_OUT_RGB, + NV10_PGRAPH_COMBINER_FINAL0, + NV10_PGRAPH_COMBINER_FINAL1, + 0x00400e00, + 0x00400e04, + 0x00400e08, + 0x00400e0c, + 0x00400e10, + 0x00400e14, + 0x00400e18, + 0x00400e1c, + 0x00400e20, + 0x00400e24, + 0x00400e28, + 0x00400e2c, + 0x00400e30, + 0x00400e34, + 0x00400e38, + 0x00400e3c, + NV04_PGRAPH_PASSTHRU_0, + NV04_PGRAPH_PASSTHRU_1, + NV04_PGRAPH_PASSTHRU_2, + NV10_PGRAPH_DIMX_TEXTURE, + NV10_PGRAPH_WDIMX_TEXTURE, + NV10_PGRAPH_DVD_COLORFMT, + NV10_PGRAPH_SCALED_FORMAT, + NV04_PGRAPH_MISC24_0, + NV04_PGRAPH_MISC24_1, + NV04_PGRAPH_MISC24_2, + NV03_PGRAPH_X_MISC, + NV03_PGRAPH_Y_MISC, + NV04_PGRAPH_VALID1, + NV04_PGRAPH_VALID2, +}; + +static int nv17_graph_ctx_regs[] = { + NV10_PGRAPH_DEBUG_4, + 0x004006b0, + 0x00400eac, + 0x00400eb0, + 0x00400eb4, + 0x00400eb8, + 0x00400ebc, + 0x00400ec0, + 0x00400ec4, + 0x00400ec8, + 0x00400ecc, + 0x00400ed0, + 0x00400ed4, + 0x00400ed8, + 0x00400edc, + 0x00400ee0, + 0x00400a00, + 0x00400a04, +}; + +struct graph_state { + int nv10[ARRAY_SIZE(nv10_graph_ctx_regs)]; + int nv17[ARRAY_SIZE(nv17_graph_ctx_regs)]; + struct pipe_state pipe_state; + uint32_t lma_window[4]; +}; + +#define PIPE_SAVE(dev, state, addr) \ + do { \ + int __i; \ + nv_wr32(dev, NV10_PGRAPH_PIPE_ADDRESS, addr); \ + for (__i = 0; __i < ARRAY_SIZE(state); __i++) \ + state[__i] = nv_rd32(dev, NV10_PGRAPH_PIPE_DATA); \ + } while (0) + +#define PIPE_RESTORE(dev, state, addr) \ + do { \ + int __i; \ + nv_wr32(dev, NV10_PGRAPH_PIPE_ADDRESS, addr); \ + for (__i = 0; __i < ARRAY_SIZE(state); __i++) \ + nv_wr32(dev, NV10_PGRAPH_PIPE_DATA, state[__i]); \ + } while (0) + +static void nv10_graph_save_pipe(struct nouveau_channel *chan) +{ + struct graph_state *pgraph_ctx = chan->engctx[NVOBJ_ENGINE_GR]; + struct pipe_state *pipe = &pgraph_ctx->pipe_state; + struct drm_device *dev = chan->dev; + + PIPE_SAVE(dev, pipe->pipe_0x4400, 0x4400); + PIPE_SAVE(dev, pipe->pipe_0x0200, 0x0200); + PIPE_SAVE(dev, pipe->pipe_0x6400, 0x6400); + PIPE_SAVE(dev, pipe->pipe_0x6800, 0x6800); + PIPE_SAVE(dev, pipe->pipe_0x6c00, 0x6c00); + PIPE_SAVE(dev, pipe->pipe_0x7000, 0x7000); + PIPE_SAVE(dev, pipe->pipe_0x7400, 0x7400); + PIPE_SAVE(dev, pipe->pipe_0x7800, 0x7800); + PIPE_SAVE(dev, pipe->pipe_0x0040, 0x0040); + PIPE_SAVE(dev, pipe->pipe_0x0000, 0x0000); +} + +static void nv10_graph_load_pipe(struct nouveau_channel *chan) +{ + struct graph_state *pgraph_ctx = chan->engctx[NVOBJ_ENGINE_GR]; + struct pipe_state *pipe = &pgraph_ctx->pipe_state; + struct drm_device *dev = chan->dev; + uint32_t xfmode0, xfmode1; + int i; + + nouveau_wait_for_idle(dev); + /* XXX check haiku comments */ + xfmode0 = nv_rd32(dev, NV10_PGRAPH_XFMODE0); + xfmode1 = nv_rd32(dev, NV10_PGRAPH_XFMODE1); + nv_wr32(dev, NV10_PGRAPH_XFMODE0, 0x10000000); + nv_wr32(dev, NV10_PGRAPH_XFMODE1, 0x00000000); + nv_wr32(dev, NV10_PGRAPH_PIPE_ADDRESS, 0x000064c0); + for (i = 0; i < 4; i++) + nv_wr32(dev, NV10_PGRAPH_PIPE_DATA, 0x3f800000); + for (i = 0; i < 4; i++) + nv_wr32(dev, NV10_PGRAPH_PIPE_DATA, 0x00000000); + + nv_wr32(dev, NV10_PGRAPH_PIPE_ADDRESS, 0x00006ab0); + for (i = 0; i < 3; i++) + nv_wr32(dev, NV10_PGRAPH_PIPE_DATA, 0x3f800000); + + nv_wr32(dev, NV10_PGRAPH_PIPE_ADDRESS, 0x00006a80); + for (i = 0; i < 3; i++) + nv_wr32(dev, NV10_PGRAPH_PIPE_DATA, 0x00000000); + + nv_wr32(dev, NV10_PGRAPH_PIPE_ADDRESS, 0x00000040); + nv_wr32(dev, NV10_PGRAPH_PIPE_DATA, 0x00000008); + + + PIPE_RESTORE(dev, pipe->pipe_0x0200, 0x0200); + nouveau_wait_for_idle(dev); + + /* restore XFMODE */ + nv_wr32(dev, NV10_PGRAPH_XFMODE0, xfmode0); + nv_wr32(dev, NV10_PGRAPH_XFMODE1, xfmode1); + PIPE_RESTORE(dev, pipe->pipe_0x6400, 0x6400); + PIPE_RESTORE(dev, pipe->pipe_0x6800, 0x6800); + PIPE_RESTORE(dev, pipe->pipe_0x6c00, 0x6c00); + PIPE_RESTORE(dev, pipe->pipe_0x7000, 0x7000); + PIPE_RESTORE(dev, pipe->pipe_0x7400, 0x7400); + PIPE_RESTORE(dev, pipe->pipe_0x7800, 0x7800); + PIPE_RESTORE(dev, pipe->pipe_0x4400, 0x4400); + PIPE_RESTORE(dev, pipe->pipe_0x0000, 0x0000); + PIPE_RESTORE(dev, pipe->pipe_0x0040, 0x0040); + nouveau_wait_for_idle(dev); +} + +static void nv10_graph_create_pipe(struct nouveau_channel *chan) +{ + struct graph_state *pgraph_ctx = chan->engctx[NVOBJ_ENGINE_GR]; + struct pipe_state *fifo_pipe_state = &pgraph_ctx->pipe_state; + struct drm_device *dev = chan->dev; + uint32_t *fifo_pipe_state_addr; + int i; +#define PIPE_INIT(addr) \ + do { \ + fifo_pipe_state_addr = fifo_pipe_state->pipe_##addr; \ + } while (0) +#define PIPE_INIT_END(addr) \ + do { \ + uint32_t *__end_addr = fifo_pipe_state->pipe_##addr + \ + ARRAY_SIZE(fifo_pipe_state->pipe_##addr); \ + if (fifo_pipe_state_addr != __end_addr) \ + NV_ERROR(dev, "incomplete pipe init for 0x%x : %p/%p\n", \ + addr, fifo_pipe_state_addr, __end_addr); \ + } while (0) +#define NV_WRITE_PIPE_INIT(value) *(fifo_pipe_state_addr++) = value + + PIPE_INIT(0x0200); + for (i = 0; i < 48; i++) + NV_WRITE_PIPE_INIT(0x00000000); + PIPE_INIT_END(0x0200); + + PIPE_INIT(0x6400); + for (i = 0; i < 211; i++) + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x3f800000); + NV_WRITE_PIPE_INIT(0x40000000); + NV_WRITE_PIPE_INIT(0x40000000); + NV_WRITE_PIPE_INIT(0x40000000); + NV_WRITE_PIPE_INIT(0x40000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x3f800000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x3f000000); + NV_WRITE_PIPE_INIT(0x3f000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x3f800000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x3f800000); + NV_WRITE_PIPE_INIT(0x3f800000); + NV_WRITE_PIPE_INIT(0x3f800000); + NV_WRITE_PIPE_INIT(0x3f800000); + PIPE_INIT_END(0x6400); + + PIPE_INIT(0x6800); + for (i = 0; i < 162; i++) + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x3f800000); + for (i = 0; i < 25; i++) + NV_WRITE_PIPE_INIT(0x00000000); + PIPE_INIT_END(0x6800); + + PIPE_INIT(0x6c00); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0xbf800000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + PIPE_INIT_END(0x6c00); + + PIPE_INIT(0x7000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x7149f2ca); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x7149f2ca); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x7149f2ca); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x7149f2ca); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x7149f2ca); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x7149f2ca); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x7149f2ca); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x00000000); + NV_WRITE_PIPE_INIT(0x7149f2ca); + for (i = 0; i < 35; i++) + NV_WRITE_PIPE_INIT(0x00000000); + PIPE_INIT_END(0x7000); + + PIPE_INIT(0x7400); + for (i = 0; i < 48; i++) + NV_WRITE_PIPE_INIT(0x00000000); + PIPE_INIT_END(0x7400); + + PIPE_INIT(0x7800); + for (i = 0; i < 48; i++) + NV_WRITE_PIPE_INIT(0x00000000); + PIPE_INIT_END(0x7800); + + PIPE_INIT(0x4400); + for (i = 0; i < 32; i++) + NV_WRITE_PIPE_INIT(0x00000000); + PIPE_INIT_END(0x4400); + + PIPE_INIT(0x0000); + for (i = 0; i < 16; i++) + NV_WRITE_PIPE_INIT(0x00000000); + PIPE_INIT_END(0x0000); + + PIPE_INIT(0x0040); + for (i = 0; i < 4; i++) + NV_WRITE_PIPE_INIT(0x00000000); + PIPE_INIT_END(0x0040); + +#undef PIPE_INIT +#undef PIPE_INIT_END +#undef NV_WRITE_PIPE_INIT +} + +static int nv10_graph_ctx_regs_find_offset(struct drm_device *dev, int reg) +{ + int i; + for (i = 0; i < ARRAY_SIZE(nv10_graph_ctx_regs); i++) { + if (nv10_graph_ctx_regs[i] == reg) + return i; + } + NV_ERROR(dev, "unknow offset nv10_ctx_regs %d\n", reg); + return -1; +} + +static int nv17_graph_ctx_regs_find_offset(struct drm_device *dev, int reg) +{ + int i; + for (i = 0; i < ARRAY_SIZE(nv17_graph_ctx_regs); i++) { + if (nv17_graph_ctx_regs[i] == reg) + return i; + } + NV_ERROR(dev, "unknow offset nv17_ctx_regs %d\n", reg); + return -1; +} + +static void nv10_graph_load_dma_vtxbuf(struct nouveau_channel *chan, + uint32_t inst) +{ + struct drm_device *dev = chan->dev; + uint32_t st2, st2_dl, st2_dh, fifo_ptr, fifo[0x60/4]; + uint32_t ctx_user, ctx_switch[5]; + int i, subchan = -1; + + /* NV10TCL_DMA_VTXBUF (method 0x18c) modifies hidden state + * that cannot be restored via MMIO. Do it through the FIFO + * instead. + */ + + /* Look for a celsius object */ + for (i = 0; i < 8; i++) { + int class = nv_rd32(dev, NV10_PGRAPH_CTX_CACHE(i, 0)) & 0xfff; + + if (class == 0x56 || class == 0x96 || class == 0x99) { + subchan = i; + break; + } + } + + if (subchan < 0 || !inst) + return; + + /* Save the current ctx object */ + ctx_user = nv_rd32(dev, NV10_PGRAPH_CTX_USER); + for (i = 0; i < 5; i++) + ctx_switch[i] = nv_rd32(dev, NV10_PGRAPH_CTX_SWITCH(i)); + + /* Save the FIFO state */ + st2 = nv_rd32(dev, NV10_PGRAPH_FFINTFC_ST2); + st2_dl = nv_rd32(dev, NV10_PGRAPH_FFINTFC_ST2_DL); + st2_dh = nv_rd32(dev, NV10_PGRAPH_FFINTFC_ST2_DH); + fifo_ptr = nv_rd32(dev, NV10_PGRAPH_FFINTFC_FIFO_PTR); + + for (i = 0; i < ARRAY_SIZE(fifo); i++) + fifo[i] = nv_rd32(dev, 0x4007a0 + 4 * i); + + /* Switch to the celsius subchannel */ + for (i = 0; i < 5; i++) + nv_wr32(dev, NV10_PGRAPH_CTX_SWITCH(i), + nv_rd32(dev, NV10_PGRAPH_CTX_CACHE(subchan, i))); + nv_mask(dev, NV10_PGRAPH_CTX_USER, 0xe000, subchan << 13); + + /* Inject NV10TCL_DMA_VTXBUF */ + nv_wr32(dev, NV10_PGRAPH_FFINTFC_FIFO_PTR, 0); + nv_wr32(dev, NV10_PGRAPH_FFINTFC_ST2, + 0x2c000000 | chan->id << 20 | subchan << 16 | 0x18c); + nv_wr32(dev, NV10_PGRAPH_FFINTFC_ST2_DL, inst); + nv_mask(dev, NV10_PGRAPH_CTX_CONTROL, 0, 0x10000); + nv04_graph_fifo_access(dev, true); + nv04_graph_fifo_access(dev, false); + + /* Restore the FIFO state */ + for (i = 0; i < ARRAY_SIZE(fifo); i++) + nv_wr32(dev, 0x4007a0 + 4 * i, fifo[i]); + + nv_wr32(dev, NV10_PGRAPH_FFINTFC_FIFO_PTR, fifo_ptr); + nv_wr32(dev, NV10_PGRAPH_FFINTFC_ST2, st2); + nv_wr32(dev, NV10_PGRAPH_FFINTFC_ST2_DL, st2_dl); + nv_wr32(dev, NV10_PGRAPH_FFINTFC_ST2_DH, st2_dh); + + /* Restore the current ctx object */ + for (i = 0; i < 5; i++) + nv_wr32(dev, NV10_PGRAPH_CTX_SWITCH(i), ctx_switch[i]); + nv_wr32(dev, NV10_PGRAPH_CTX_USER, ctx_user); +} + +static int +nv10_graph_load_context(struct nouveau_channel *chan) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct graph_state *pgraph_ctx = chan->engctx[NVOBJ_ENGINE_GR]; + uint32_t tmp; + int i; + + for (i = 0; i < ARRAY_SIZE(nv10_graph_ctx_regs); i++) + nv_wr32(dev, nv10_graph_ctx_regs[i], pgraph_ctx->nv10[i]); + if (dev_priv->chipset >= 0x17) { + for (i = 0; i < ARRAY_SIZE(nv17_graph_ctx_regs); i++) + nv_wr32(dev, nv17_graph_ctx_regs[i], + pgraph_ctx->nv17[i]); + } + + nv10_graph_load_pipe(chan); + nv10_graph_load_dma_vtxbuf(chan, (nv_rd32(dev, NV10_PGRAPH_GLOBALSTATE1) + & 0xffff)); + + nv_wr32(dev, NV10_PGRAPH_CTX_CONTROL, 0x10010100); + tmp = nv_rd32(dev, NV10_PGRAPH_CTX_USER); + nv_wr32(dev, NV10_PGRAPH_CTX_USER, (tmp & 0xffffff) | chan->id << 24); + tmp = nv_rd32(dev, NV10_PGRAPH_FFINTFC_ST2); + nv_wr32(dev, NV10_PGRAPH_FFINTFC_ST2, tmp & 0xcfffffff); + return 0; +} + +static int +nv10_graph_unload_context(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; + struct nouveau_channel *chan; + struct graph_state *ctx; + uint32_t tmp; + int i; + + chan = nv10_graph_channel(dev); + if (!chan) + return 0; + ctx = chan->engctx[NVOBJ_ENGINE_GR]; + + for (i = 0; i < ARRAY_SIZE(nv10_graph_ctx_regs); i++) + ctx->nv10[i] = nv_rd32(dev, nv10_graph_ctx_regs[i]); + + if (dev_priv->chipset >= 0x17) { + for (i = 0; i < ARRAY_SIZE(nv17_graph_ctx_regs); i++) + ctx->nv17[i] = nv_rd32(dev, nv17_graph_ctx_regs[i]); + } + + nv10_graph_save_pipe(chan); + + nv_wr32(dev, NV10_PGRAPH_CTX_CONTROL, 0x10000000); + tmp = nv_rd32(dev, NV10_PGRAPH_CTX_USER) & 0x00ffffff; + tmp |= (pfifo->channels - 1) << 24; + nv_wr32(dev, NV10_PGRAPH_CTX_USER, tmp); + return 0; +} + +static void +nv10_graph_context_switch(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_channel *chan = NULL; + int chid; + + nouveau_wait_for_idle(dev); + + /* If previous context is valid, we need to save it */ + nv10_graph_unload_context(dev); + + /* Load context for next channel */ + chid = (nv_rd32(dev, NV04_PGRAPH_TRAPPED_ADDR) >> 20) & 0x1f; + chan = dev_priv->channels.ptr[chid]; + if (chan && chan->engctx[NVOBJ_ENGINE_GR]) + nv10_graph_load_context(chan); +} + +#define NV_WRITE_CTX(reg, val) do { \ + int offset = nv10_graph_ctx_regs_find_offset(dev, reg); \ + if (offset > 0) \ + pgraph_ctx->nv10[offset] = val; \ + } while (0) + +#define NV17_WRITE_CTX(reg, val) do { \ + int offset = nv17_graph_ctx_regs_find_offset(dev, reg); \ + if (offset > 0) \ + pgraph_ctx->nv17[offset] = val; \ + } while (0) + +struct nouveau_channel * +nv10_graph_channel(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + int chid = dev_priv->engine.fifo.channels; + + if (nv_rd32(dev, NV10_PGRAPH_CTX_CONTROL) & 0x00010000) + chid = nv_rd32(dev, NV10_PGRAPH_CTX_USER) >> 24; + + if (chid >= dev_priv->engine.fifo.channels) + return NULL; + + return dev_priv->channels.ptr[chid]; +} + +static int +nv10_graph_context_new(struct nouveau_channel *chan, int engine) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct graph_state *pgraph_ctx; + + NV_DEBUG(dev, "nv10_graph_context_create %d\n", chan->id); + + pgraph_ctx = kzalloc(sizeof(*pgraph_ctx), GFP_KERNEL); + if (pgraph_ctx == NULL) + return -ENOMEM; + chan->engctx[engine] = pgraph_ctx; + + NV_WRITE_CTX(0x00400e88, 0x08000000); + NV_WRITE_CTX(0x00400e9c, 0x4b7fffff); + NV_WRITE_CTX(NV03_PGRAPH_XY_LOGIC_MISC0, 0x0001ffff); + NV_WRITE_CTX(0x00400e10, 0x00001000); + NV_WRITE_CTX(0x00400e14, 0x00001000); + NV_WRITE_CTX(0x00400e30, 0x00080008); + NV_WRITE_CTX(0x00400e34, 0x00080008); + if (dev_priv->chipset >= 0x17) { + /* is it really needed ??? */ + NV17_WRITE_CTX(NV10_PGRAPH_DEBUG_4, + nv_rd32(dev, NV10_PGRAPH_DEBUG_4)); + NV17_WRITE_CTX(0x004006b0, nv_rd32(dev, 0x004006b0)); + NV17_WRITE_CTX(0x00400eac, 0x0fff0000); + NV17_WRITE_CTX(0x00400eb0, 0x0fff0000); + NV17_WRITE_CTX(0x00400ec0, 0x00000080); + NV17_WRITE_CTX(0x00400ed0, 0x00000080); + } + NV_WRITE_CTX(NV10_PGRAPH_CTX_USER, chan->id << 24); + + nv10_graph_create_pipe(chan); + return 0; +} + +static void +nv10_graph_context_del(struct nouveau_channel *chan, int engine) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct graph_state *pgraph_ctx = chan->engctx[engine]; + unsigned long flags; + + spin_lock_irqsave(&dev_priv->context_switch_lock, flags); + nv04_graph_fifo_access(dev, false); + + /* Unload the context if it's the currently active one */ + if (nv10_graph_channel(dev) == chan) + nv10_graph_unload_context(dev); + + nv04_graph_fifo_access(dev, true); + spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); + + /* Free the context resources */ + chan->engctx[engine] = NULL; + kfree(pgraph_ctx); +} + +static void +nv10_graph_set_tile_region(struct drm_device *dev, int i) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_tile_reg *tile = &dev_priv->tile.reg[i]; + + nv_wr32(dev, NV10_PGRAPH_TLIMIT(i), tile->limit); + nv_wr32(dev, NV10_PGRAPH_TSIZE(i), tile->pitch); + nv_wr32(dev, NV10_PGRAPH_TILE(i), tile->addr); +} + +static int +nv10_graph_init(struct drm_device *dev, int engine) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + u32 tmp; + int i; + + nv_wr32(dev, NV03_PMC_ENABLE, nv_rd32(dev, NV03_PMC_ENABLE) & + ~NV_PMC_ENABLE_PGRAPH); + nv_wr32(dev, NV03_PMC_ENABLE, nv_rd32(dev, NV03_PMC_ENABLE) | + NV_PMC_ENABLE_PGRAPH); + + nv_wr32(dev, NV03_PGRAPH_INTR , 0xFFFFFFFF); + nv_wr32(dev, NV03_PGRAPH_INTR_EN, 0xFFFFFFFF); + + nv_wr32(dev, NV04_PGRAPH_DEBUG_0, 0xFFFFFFFF); + nv_wr32(dev, NV04_PGRAPH_DEBUG_0, 0x00000000); + nv_wr32(dev, NV04_PGRAPH_DEBUG_1, 0x00118700); + /* nv_wr32(dev, NV04_PGRAPH_DEBUG_2, 0x24E00810); */ /* 0x25f92ad9 */ + nv_wr32(dev, NV04_PGRAPH_DEBUG_2, 0x25f92ad9); + nv_wr32(dev, NV04_PGRAPH_DEBUG_3, 0x55DE0830 | + (1<<29) | + (1<<31)); + if (dev_priv->chipset >= 0x17) { + nv_wr32(dev, NV10_PGRAPH_DEBUG_4, 0x1f000000); + nv_wr32(dev, 0x400a10, 0x3ff3fb6); + nv_wr32(dev, 0x400838, 0x2f8684); + nv_wr32(dev, 0x40083c, 0x115f3f); + nv_wr32(dev, 0x004006b0, 0x40000020); + } else + nv_wr32(dev, NV10_PGRAPH_DEBUG_4, 0x00000000); + + /* Turn all the tiling regions off. */ + for (i = 0; i < NV10_PFB_TILE__SIZE; i++) + nv10_graph_set_tile_region(dev, i); + + nv_wr32(dev, NV10_PGRAPH_CTX_SWITCH(0), 0x00000000); + nv_wr32(dev, NV10_PGRAPH_CTX_SWITCH(1), 0x00000000); + nv_wr32(dev, NV10_PGRAPH_CTX_SWITCH(2), 0x00000000); + nv_wr32(dev, NV10_PGRAPH_CTX_SWITCH(3), 0x00000000); + nv_wr32(dev, NV10_PGRAPH_CTX_SWITCH(4), 0x00000000); + nv_wr32(dev, NV10_PGRAPH_STATE, 0xFFFFFFFF); + + tmp = nv_rd32(dev, NV10_PGRAPH_CTX_USER) & 0x00ffffff; + tmp |= (dev_priv->engine.fifo.channels - 1) << 24; + nv_wr32(dev, NV10_PGRAPH_CTX_USER, tmp); + nv_wr32(dev, NV10_PGRAPH_CTX_CONTROL, 0x10000100); + nv_wr32(dev, NV10_PGRAPH_FFINTFC_ST2, 0x08000000); + + return 0; +} + +static int +nv10_graph_fini(struct drm_device *dev, int engine) +{ + nv10_graph_unload_context(dev); + nv_wr32(dev, NV03_PGRAPH_INTR_EN, 0x00000000); + return 0; +} + +static int +nv17_graph_mthd_lma_window(struct nouveau_channel *chan, + u32 class, u32 mthd, u32 data) +{ + struct graph_state *ctx = chan->engctx[NVOBJ_ENGINE_GR]; + struct drm_device *dev = chan->dev; + struct pipe_state *pipe = &ctx->pipe_state; + uint32_t pipe_0x0040[1], pipe_0x64c0[8], pipe_0x6a80[3], pipe_0x6ab0[3]; + uint32_t xfmode0, xfmode1; + int i; + + ctx->lma_window[(mthd - 0x1638) / 4] = data; + + if (mthd != 0x1644) + return 0; + + nouveau_wait_for_idle(dev); + + PIPE_SAVE(dev, pipe_0x0040, 0x0040); + PIPE_SAVE(dev, pipe->pipe_0x0200, 0x0200); + + PIPE_RESTORE(dev, ctx->lma_window, 0x6790); + + nouveau_wait_for_idle(dev); + + xfmode0 = nv_rd32(dev, NV10_PGRAPH_XFMODE0); + xfmode1 = nv_rd32(dev, NV10_PGRAPH_XFMODE1); + + PIPE_SAVE(dev, pipe->pipe_0x4400, 0x4400); + PIPE_SAVE(dev, pipe_0x64c0, 0x64c0); + PIPE_SAVE(dev, pipe_0x6ab0, 0x6ab0); + PIPE_SAVE(dev, pipe_0x6a80, 0x6a80); + + nouveau_wait_for_idle(dev); + + nv_wr32(dev, NV10_PGRAPH_XFMODE0, 0x10000000); + nv_wr32(dev, NV10_PGRAPH_XFMODE1, 0x00000000); + nv_wr32(dev, NV10_PGRAPH_PIPE_ADDRESS, 0x000064c0); + for (i = 0; i < 4; i++) + nv_wr32(dev, NV10_PGRAPH_PIPE_DATA, 0x3f800000); + for (i = 0; i < 4; i++) + nv_wr32(dev, NV10_PGRAPH_PIPE_DATA, 0x00000000); + + nv_wr32(dev, NV10_PGRAPH_PIPE_ADDRESS, 0x00006ab0); + for (i = 0; i < 3; i++) + nv_wr32(dev, NV10_PGRAPH_PIPE_DATA, 0x3f800000); + + nv_wr32(dev, NV10_PGRAPH_PIPE_ADDRESS, 0x00006a80); + for (i = 0; i < 3; i++) + nv_wr32(dev, NV10_PGRAPH_PIPE_DATA, 0x00000000); + + nv_wr32(dev, NV10_PGRAPH_PIPE_ADDRESS, 0x00000040); + nv_wr32(dev, NV10_PGRAPH_PIPE_DATA, 0x00000008); + + PIPE_RESTORE(dev, pipe->pipe_0x0200, 0x0200); + + nouveau_wait_for_idle(dev); + + PIPE_RESTORE(dev, pipe_0x0040, 0x0040); + + nv_wr32(dev, NV10_PGRAPH_XFMODE0, xfmode0); + nv_wr32(dev, NV10_PGRAPH_XFMODE1, xfmode1); + + PIPE_RESTORE(dev, pipe_0x64c0, 0x64c0); + PIPE_RESTORE(dev, pipe_0x6ab0, 0x6ab0); + PIPE_RESTORE(dev, pipe_0x6a80, 0x6a80); + PIPE_RESTORE(dev, pipe->pipe_0x4400, 0x4400); + + nv_wr32(dev, NV10_PGRAPH_PIPE_ADDRESS, 0x000000c0); + nv_wr32(dev, NV10_PGRAPH_PIPE_DATA, 0x00000000); + + nouveau_wait_for_idle(dev); + + return 0; +} + +static int +nv17_graph_mthd_lma_enable(struct nouveau_channel *chan, + u32 class, u32 mthd, u32 data) +{ + struct drm_device *dev = chan->dev; + + nouveau_wait_for_idle(dev); + + nv_wr32(dev, NV10_PGRAPH_DEBUG_4, + nv_rd32(dev, NV10_PGRAPH_DEBUG_4) | 0x1 << 8); + nv_wr32(dev, 0x004006b0, + nv_rd32(dev, 0x004006b0) | 0x8 << 24); + + return 0; +} + +struct nouveau_bitfield nv10_graph_intr[] = { + { NV_PGRAPH_INTR_NOTIFY, "NOTIFY" }, + { NV_PGRAPH_INTR_ERROR, "ERROR" }, + {} +}; + +struct nouveau_bitfield nv10_graph_nstatus[] = { + { NV10_PGRAPH_NSTATUS_STATE_IN_USE, "STATE_IN_USE" }, + { NV10_PGRAPH_NSTATUS_INVALID_STATE, "INVALID_STATE" }, + { NV10_PGRAPH_NSTATUS_BAD_ARGUMENT, "BAD_ARGUMENT" }, + { NV10_PGRAPH_NSTATUS_PROTECTION_FAULT, "PROTECTION_FAULT" }, + {} +}; + +static void +nv10_graph_isr(struct drm_device *dev) +{ + u32 stat; + + while ((stat = nv_rd32(dev, NV03_PGRAPH_INTR))) { + u32 nsource = nv_rd32(dev, NV03_PGRAPH_NSOURCE); + u32 nstatus = nv_rd32(dev, NV03_PGRAPH_NSTATUS); + u32 addr = nv_rd32(dev, NV04_PGRAPH_TRAPPED_ADDR); + u32 chid = (addr & 0x01f00000) >> 20; + u32 subc = (addr & 0x00070000) >> 16; + u32 mthd = (addr & 0x00001ffc); + u32 data = nv_rd32(dev, NV04_PGRAPH_TRAPPED_DATA); + u32 class = nv_rd32(dev, 0x400160 + subc * 4) & 0xfff; + u32 show = stat; + + if (stat & NV_PGRAPH_INTR_ERROR) { + if (nsource & NV03_PGRAPH_NSOURCE_ILLEGAL_MTHD) { + if (!nouveau_gpuobj_mthd_call2(dev, chid, class, mthd, data)) + show &= ~NV_PGRAPH_INTR_ERROR; + } + } + + if (stat & NV_PGRAPH_INTR_CONTEXT_SWITCH) { + nv_wr32(dev, NV03_PGRAPH_INTR, NV_PGRAPH_INTR_CONTEXT_SWITCH); + stat &= ~NV_PGRAPH_INTR_CONTEXT_SWITCH; + show &= ~NV_PGRAPH_INTR_CONTEXT_SWITCH; + nv10_graph_context_switch(dev); + } + + nv_wr32(dev, NV03_PGRAPH_INTR, stat); + nv_wr32(dev, NV04_PGRAPH_FIFO, 0x00000001); + + if (show && nouveau_ratelimit()) { + NV_INFO(dev, "PGRAPH -"); + nouveau_bitfield_print(nv10_graph_intr, show); + printk(" nsource:"); + nouveau_bitfield_print(nv04_graph_nsource, nsource); + printk(" nstatus:"); + nouveau_bitfield_print(nv10_graph_nstatus, nstatus); + printk("\n"); + NV_INFO(dev, "PGRAPH - ch %d/%d class 0x%04x " + "mthd 0x%04x data 0x%08x\n", + chid, subc, class, mthd, data); + } + } +} + +static void +nv10_graph_destroy(struct drm_device *dev, int engine) +{ + struct nv10_graph_engine *pgraph = nv_engine(dev, engine); + + nouveau_irq_unregister(dev, 12); + kfree(pgraph); +} + +int +nv10_graph_create(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nv10_graph_engine *pgraph; + + pgraph = kzalloc(sizeof(*pgraph), GFP_KERNEL); + if (!pgraph) + return -ENOMEM; + + pgraph->base.destroy = nv10_graph_destroy; + pgraph->base.init = nv10_graph_init; + pgraph->base.fini = nv10_graph_fini; + pgraph->base.context_new = nv10_graph_context_new; + pgraph->base.context_del = nv10_graph_context_del; + pgraph->base.object_new = nv04_graph_object_new; + pgraph->base.set_tile_region = nv10_graph_set_tile_region; + + NVOBJ_ENGINE_ADD(dev, GR, &pgraph->base); + nouveau_irq_register(dev, 12, nv10_graph_isr); + + /* nvsw */ + NVOBJ_CLASS(dev, 0x506e, SW); + NVOBJ_MTHD (dev, 0x506e, 0x0500, nv04_graph_mthd_page_flip); + + NVOBJ_CLASS(dev, 0x0030, GR); /* null */ + NVOBJ_CLASS(dev, 0x0039, GR); /* m2mf */ + NVOBJ_CLASS(dev, 0x004a, GR); /* gdirect */ + NVOBJ_CLASS(dev, 0x005f, GR); /* imageblit */ + NVOBJ_CLASS(dev, 0x009f, GR); /* imageblit (nv12) */ + NVOBJ_CLASS(dev, 0x008a, GR); /* ifc */ + NVOBJ_CLASS(dev, 0x0089, GR); /* sifm */ + NVOBJ_CLASS(dev, 0x0062, GR); /* surf2d */ + NVOBJ_CLASS(dev, 0x0043, GR); /* rop */ + NVOBJ_CLASS(dev, 0x0012, GR); /* beta1 */ + NVOBJ_CLASS(dev, 0x0072, GR); /* beta4 */ + NVOBJ_CLASS(dev, 0x0019, GR); /* cliprect */ + NVOBJ_CLASS(dev, 0x0044, GR); /* pattern */ + NVOBJ_CLASS(dev, 0x0052, GR); /* swzsurf */ + NVOBJ_CLASS(dev, 0x0093, GR); /* surf3d */ + NVOBJ_CLASS(dev, 0x0094, GR); /* tex_tri */ + NVOBJ_CLASS(dev, 0x0095, GR); /* multitex_tri */ + + /* celcius */ + if (dev_priv->chipset <= 0x10) { + NVOBJ_CLASS(dev, 0x0056, GR); + } else + if (dev_priv->chipset < 0x17 || dev_priv->chipset == 0x1a) { + NVOBJ_CLASS(dev, 0x0096, GR); + } else { + NVOBJ_CLASS(dev, 0x0099, GR); + NVOBJ_MTHD (dev, 0x0099, 0x1638, nv17_graph_mthd_lma_window); + NVOBJ_MTHD (dev, 0x0099, 0x163c, nv17_graph_mthd_lma_window); + NVOBJ_MTHD (dev, 0x0099, 0x1640, nv17_graph_mthd_lma_window); + NVOBJ_MTHD (dev, 0x0099, 0x1644, nv17_graph_mthd_lma_window); + NVOBJ_MTHD (dev, 0x0099, 0x1658, nv17_graph_mthd_lma_enable); + } + + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nv17_tv.c b/drivers/gpu/drm/nouveau/nv17_tv.c new file mode 100644 index 00000000..3900cebb --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv17_tv.c @@ -0,0 +1,827 @@ +/* + * Copyright (C) 2009 Francisco Jerez. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "drmP.h" +#include "drm_crtc_helper.h" +#include "nouveau_drv.h" +#include "nouveau_encoder.h" +#include "nouveau_connector.h" +#include "nouveau_crtc.h" +#include "nouveau_hw.h" +#include "nv17_tv.h" + +static uint32_t nv42_tv_sample_load(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpio_engine *gpio = &dev_priv->engine.gpio; + uint32_t testval, regoffset = nv04_dac_output_offset(encoder); + uint32_t gpio0, gpio1, fp_htotal, fp_hsync_start, fp_hsync_end, + fp_control, test_ctrl, dacclk, ctv_14, ctv_1c, ctv_6c; + uint32_t sample = 0; + int head; + +#define RGB_TEST_DATA(r, g, b) (r << 0 | g << 10 | b << 20) + testval = RGB_TEST_DATA(0x82, 0xeb, 0x82); + if (dev_priv->vbios.tvdactestval) + testval = dev_priv->vbios.tvdactestval; + + dacclk = NVReadRAMDAC(dev, 0, NV_PRAMDAC_DACCLK + regoffset); + head = (dacclk & 0x100) >> 8; + + /* Save the previous state. */ + gpio1 = gpio->get(dev, DCB_GPIO_TVDAC1); + gpio0 = gpio->get(dev, DCB_GPIO_TVDAC0); + fp_htotal = NVReadRAMDAC(dev, head, NV_PRAMDAC_FP_HTOTAL); + fp_hsync_start = NVReadRAMDAC(dev, head, NV_PRAMDAC_FP_HSYNC_START); + fp_hsync_end = NVReadRAMDAC(dev, head, NV_PRAMDAC_FP_HSYNC_END); + fp_control = NVReadRAMDAC(dev, head, NV_PRAMDAC_FP_TG_CONTROL); + test_ctrl = NVReadRAMDAC(dev, 0, NV_PRAMDAC_TEST_CONTROL + regoffset); + ctv_1c = NVReadRAMDAC(dev, head, 0x680c1c); + ctv_14 = NVReadRAMDAC(dev, head, 0x680c14); + ctv_6c = NVReadRAMDAC(dev, head, 0x680c6c); + + /* Prepare the DAC for load detection. */ + gpio->set(dev, DCB_GPIO_TVDAC1, true); + gpio->set(dev, DCB_GPIO_TVDAC0, true); + + NVWriteRAMDAC(dev, head, NV_PRAMDAC_FP_HTOTAL, 1343); + NVWriteRAMDAC(dev, head, NV_PRAMDAC_FP_HSYNC_START, 1047); + NVWriteRAMDAC(dev, head, NV_PRAMDAC_FP_HSYNC_END, 1183); + NVWriteRAMDAC(dev, head, NV_PRAMDAC_FP_TG_CONTROL, + NV_PRAMDAC_FP_TG_CONTROL_DISPEN_POS | + NV_PRAMDAC_FP_TG_CONTROL_WIDTH_12 | + NV_PRAMDAC_FP_TG_CONTROL_READ_PROG | + NV_PRAMDAC_FP_TG_CONTROL_HSYNC_POS | + NV_PRAMDAC_FP_TG_CONTROL_VSYNC_POS); + + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_TEST_CONTROL + regoffset, 0); + + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_DACCLK + regoffset, + (dacclk & ~0xff) | 0x22); + msleep(1); + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_DACCLK + regoffset, + (dacclk & ~0xff) | 0x21); + + NVWriteRAMDAC(dev, head, 0x680c1c, 1 << 20); + NVWriteRAMDAC(dev, head, 0x680c14, 4 << 16); + + /* Sample pin 0x4 (usually S-video luma). */ + NVWriteRAMDAC(dev, head, 0x680c6c, testval >> 10 & 0x3ff); + msleep(20); + sample |= NVReadRAMDAC(dev, 0, NV_PRAMDAC_TEST_CONTROL + regoffset) + & 0x4 << 28; + + /* Sample the remaining pins. */ + NVWriteRAMDAC(dev, head, 0x680c6c, testval & 0x3ff); + msleep(20); + sample |= NVReadRAMDAC(dev, 0, NV_PRAMDAC_TEST_CONTROL + regoffset) + & 0xa << 28; + + /* Restore the previous state. */ + NVWriteRAMDAC(dev, head, 0x680c1c, ctv_1c); + NVWriteRAMDAC(dev, head, 0x680c14, ctv_14); + NVWriteRAMDAC(dev, head, 0x680c6c, ctv_6c); + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_DACCLK + regoffset, dacclk); + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_TEST_CONTROL + regoffset, test_ctrl); + NVWriteRAMDAC(dev, head, NV_PRAMDAC_FP_TG_CONTROL, fp_control); + NVWriteRAMDAC(dev, head, NV_PRAMDAC_FP_HSYNC_END, fp_hsync_end); + NVWriteRAMDAC(dev, head, NV_PRAMDAC_FP_HSYNC_START, fp_hsync_start); + NVWriteRAMDAC(dev, head, NV_PRAMDAC_FP_HTOTAL, fp_htotal); + gpio->set(dev, DCB_GPIO_TVDAC1, gpio1); + gpio->set(dev, DCB_GPIO_TVDAC0, gpio0); + + return sample; +} + +static bool +get_tv_detect_quirks(struct drm_device *dev, uint32_t *pin_mask) +{ + /* Zotac FX5200 */ + if (nv_match_device(dev, 0x0322, 0x19da, 0x1035) || + nv_match_device(dev, 0x0322, 0x19da, 0x2035)) { + *pin_mask = 0xc; + return false; + } + + /* MSI nForce2 IGP */ + if (nv_match_device(dev, 0x01f0, 0x1462, 0x5710)) { + *pin_mask = 0xc; + return false; + } + + return true; +} + +static enum drm_connector_status +nv17_tv_detect(struct drm_encoder *encoder, struct drm_connector *connector) +{ + struct drm_device *dev = encoder->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct drm_mode_config *conf = &dev->mode_config; + struct nv17_tv_encoder *tv_enc = to_tv_enc(encoder); + struct dcb_entry *dcb = tv_enc->base.dcb; + bool reliable = get_tv_detect_quirks(dev, &tv_enc->pin_mask); + + if (nv04_dac_in_use(encoder)) + return connector_status_disconnected; + + if (reliable) { + if (dev_priv->chipset == 0x42 || + dev_priv->chipset == 0x43) + tv_enc->pin_mask = + nv42_tv_sample_load(encoder) >> 28 & 0xe; + else + tv_enc->pin_mask = + nv17_dac_sample_load(encoder) >> 28 & 0xe; + } + + switch (tv_enc->pin_mask) { + case 0x2: + case 0x4: + tv_enc->subconnector = DRM_MODE_SUBCONNECTOR_Composite; + break; + case 0xc: + tv_enc->subconnector = DRM_MODE_SUBCONNECTOR_SVIDEO; + break; + case 0xe: + if (dcb->tvconf.has_component_output) + tv_enc->subconnector = DRM_MODE_SUBCONNECTOR_Component; + else + tv_enc->subconnector = DRM_MODE_SUBCONNECTOR_SCART; + break; + default: + tv_enc->subconnector = DRM_MODE_SUBCONNECTOR_Unknown; + break; + } + + drm_connector_property_set_value(connector, + conf->tv_subconnector_property, + tv_enc->subconnector); + + if (!reliable) { + return connector_status_unknown; + } else if (tv_enc->subconnector) { + NV_INFO(dev, "Load detected on output %c\n", + '@' + ffs(dcb->or)); + return connector_status_connected; + } else { + return connector_status_disconnected; + } +} + +static int nv17_tv_get_ld_modes(struct drm_encoder *encoder, + struct drm_connector *connector) +{ + struct nv17_tv_norm_params *tv_norm = get_tv_norm(encoder); + const struct drm_display_mode *tv_mode; + int n = 0; + + for (tv_mode = nv17_tv_modes; tv_mode->hdisplay; tv_mode++) { + struct drm_display_mode *mode; + + mode = drm_mode_duplicate(encoder->dev, tv_mode); + + mode->clock = tv_norm->tv_enc_mode.vrefresh * + mode->htotal / 1000 * + mode->vtotal / 1000; + + if (mode->flags & DRM_MODE_FLAG_DBLSCAN) + mode->clock *= 2; + + if (mode->hdisplay == tv_norm->tv_enc_mode.hdisplay && + mode->vdisplay == tv_norm->tv_enc_mode.vdisplay) + mode->type |= DRM_MODE_TYPE_PREFERRED; + + drm_mode_probed_add(connector, mode); + n++; + } + + return n; +} + +static int nv17_tv_get_hd_modes(struct drm_encoder *encoder, + struct drm_connector *connector) +{ + struct nv17_tv_norm_params *tv_norm = get_tv_norm(encoder); + struct drm_display_mode *output_mode = &tv_norm->ctv_enc_mode.mode; + struct drm_display_mode *mode; + const struct { + int hdisplay; + int vdisplay; + } modes[] = { + { 640, 400 }, + { 640, 480 }, + { 720, 480 }, + { 720, 576 }, + { 800, 600 }, + { 1024, 768 }, + { 1280, 720 }, + { 1280, 1024 }, + { 1920, 1080 } + }; + int i, n = 0; + + for (i = 0; i < ARRAY_SIZE(modes); i++) { + if (modes[i].hdisplay > output_mode->hdisplay || + modes[i].vdisplay > output_mode->vdisplay) + continue; + + if (modes[i].hdisplay == output_mode->hdisplay && + modes[i].vdisplay == output_mode->vdisplay) { + mode = drm_mode_duplicate(encoder->dev, output_mode); + mode->type |= DRM_MODE_TYPE_PREFERRED; + + } else { + mode = drm_cvt_mode(encoder->dev, modes[i].hdisplay, + modes[i].vdisplay, 60, false, + (output_mode->flags & + DRM_MODE_FLAG_INTERLACE), false); + } + + /* CVT modes are sometimes unsuitable... */ + if (output_mode->hdisplay <= 720 + || output_mode->hdisplay >= 1920) { + mode->htotal = output_mode->htotal; + mode->hsync_start = (mode->hdisplay + (mode->htotal + - mode->hdisplay) * 9 / 10) & ~7; + mode->hsync_end = mode->hsync_start + 8; + } + + if (output_mode->vdisplay >= 1024) { + mode->vtotal = output_mode->vtotal; + mode->vsync_start = output_mode->vsync_start; + mode->vsync_end = output_mode->vsync_end; + } + + mode->type |= DRM_MODE_TYPE_DRIVER; + drm_mode_probed_add(connector, mode); + n++; + } + + return n; +} + +static int nv17_tv_get_modes(struct drm_encoder *encoder, + struct drm_connector *connector) +{ + struct nv17_tv_norm_params *tv_norm = get_tv_norm(encoder); + + if (tv_norm->kind == CTV_ENC_MODE) + return nv17_tv_get_hd_modes(encoder, connector); + else + return nv17_tv_get_ld_modes(encoder, connector); +} + +static int nv17_tv_mode_valid(struct drm_encoder *encoder, + struct drm_display_mode *mode) +{ + struct nv17_tv_norm_params *tv_norm = get_tv_norm(encoder); + + if (tv_norm->kind == CTV_ENC_MODE) { + struct drm_display_mode *output_mode = + &tv_norm->ctv_enc_mode.mode; + + if (mode->clock > 400000) + return MODE_CLOCK_HIGH; + + if (mode->hdisplay > output_mode->hdisplay || + mode->vdisplay > output_mode->vdisplay) + return MODE_BAD; + + if ((mode->flags & DRM_MODE_FLAG_INTERLACE) != + (output_mode->flags & DRM_MODE_FLAG_INTERLACE)) + return MODE_NO_INTERLACE; + + if (mode->flags & DRM_MODE_FLAG_DBLSCAN) + return MODE_NO_DBLESCAN; + + } else { + const int vsync_tolerance = 600; + + if (mode->clock > 70000) + return MODE_CLOCK_HIGH; + + if (abs(drm_mode_vrefresh(mode) * 1000 - + tv_norm->tv_enc_mode.vrefresh) > vsync_tolerance) + return MODE_VSYNC; + + /* The encoder takes care of the actual interlacing */ + if (mode->flags & DRM_MODE_FLAG_INTERLACE) + return MODE_NO_INTERLACE; + } + + return MODE_OK; +} + +static bool nv17_tv_mode_fixup(struct drm_encoder *encoder, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct nv17_tv_norm_params *tv_norm = get_tv_norm(encoder); + + if (nv04_dac_in_use(encoder)) + return false; + + if (tv_norm->kind == CTV_ENC_MODE) + adjusted_mode->clock = tv_norm->ctv_enc_mode.mode.clock; + else + adjusted_mode->clock = 90000; + + return true; +} + +static void nv17_tv_dpms(struct drm_encoder *encoder, int mode) +{ + struct drm_device *dev = encoder->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpio_engine *gpio = &dev_priv->engine.gpio; + struct nv17_tv_state *regs = &to_tv_enc(encoder)->state; + struct nv17_tv_norm_params *tv_norm = get_tv_norm(encoder); + + if (nouveau_encoder(encoder)->last_dpms == mode) + return; + nouveau_encoder(encoder)->last_dpms = mode; + + NV_INFO(dev, "Setting dpms mode %d on TV encoder (output %d)\n", + mode, nouveau_encoder(encoder)->dcb->index); + + regs->ptv_200 &= ~1; + + if (tv_norm->kind == CTV_ENC_MODE) { + nv04_dfp_update_fp_control(encoder, mode); + + } else { + nv04_dfp_update_fp_control(encoder, DRM_MODE_DPMS_OFF); + + if (mode == DRM_MODE_DPMS_ON) + regs->ptv_200 |= 1; + } + + nv_load_ptv(dev, regs, 200); + + gpio->set(dev, DCB_GPIO_TVDAC1, mode == DRM_MODE_DPMS_ON); + gpio->set(dev, DCB_GPIO_TVDAC0, mode == DRM_MODE_DPMS_ON); + + nv04_dac_update_dacclk(encoder, mode == DRM_MODE_DPMS_ON); +} + +static void nv17_tv_prepare(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct drm_encoder_helper_funcs *helper = encoder->helper_private; + struct nv17_tv_norm_params *tv_norm = get_tv_norm(encoder); + int head = nouveau_crtc(encoder->crtc)->index; + uint8_t *cr_lcd = &dev_priv->mode_reg.crtc_reg[head].CRTC[ + NV_CIO_CRE_LCD__INDEX]; + uint32_t dacclk_off = NV_PRAMDAC_DACCLK + + nv04_dac_output_offset(encoder); + uint32_t dacclk; + + helper->dpms(encoder, DRM_MODE_DPMS_OFF); + + nv04_dfp_disable(dev, head); + + /* Unbind any FP encoders from this head if we need the FP + * stuff enabled. */ + if (tv_norm->kind == CTV_ENC_MODE) { + struct drm_encoder *enc; + + list_for_each_entry(enc, &dev->mode_config.encoder_list, head) { + struct dcb_entry *dcb = nouveau_encoder(enc)->dcb; + + if ((dcb->type == OUTPUT_TMDS || + dcb->type == OUTPUT_LVDS) && + !enc->crtc && + nv04_dfp_get_bound_head(dev, dcb) == head) { + nv04_dfp_bind_head(dev, dcb, head ^ 1, + dev_priv->vbios.fp.dual_link); + } + } + + } + + if (tv_norm->kind == CTV_ENC_MODE) + *cr_lcd |= 0x1 | (head ? 0x0 : 0x8); + + /* Set the DACCLK register */ + dacclk = (NVReadRAMDAC(dev, 0, dacclk_off) & ~0x30) | 0x1; + + if (dev_priv->card_type == NV_40) + dacclk |= 0x1a << 16; + + if (tv_norm->kind == CTV_ENC_MODE) { + dacclk |= 0x20; + + if (head) + dacclk |= 0x100; + else + dacclk &= ~0x100; + + } else { + dacclk |= 0x10; + + } + + NVWriteRAMDAC(dev, 0, dacclk_off, dacclk); +} + +static void nv17_tv_mode_set(struct drm_encoder *encoder, + struct drm_display_mode *drm_mode, + struct drm_display_mode *adjusted_mode) +{ + struct drm_device *dev = encoder->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + int head = nouveau_crtc(encoder->crtc)->index; + struct nv04_crtc_reg *regs = &dev_priv->mode_reg.crtc_reg[head]; + struct nv17_tv_state *tv_regs = &to_tv_enc(encoder)->state; + struct nv17_tv_norm_params *tv_norm = get_tv_norm(encoder); + int i; + + regs->CRTC[NV_CIO_CRE_53] = 0x40; /* FP_HTIMING */ + regs->CRTC[NV_CIO_CRE_54] = 0; /* FP_VTIMING */ + regs->ramdac_630 = 0x2; /* turn off green mode (tv test pattern?) */ + regs->tv_setup = 1; + regs->ramdac_8c0 = 0x0; + + if (tv_norm->kind == TV_ENC_MODE) { + tv_regs->ptv_200 = 0x13111100; + if (head) + tv_regs->ptv_200 |= 0x10; + + tv_regs->ptv_20c = 0x808010; + tv_regs->ptv_304 = 0x2d00000; + tv_regs->ptv_600 = 0x0; + tv_regs->ptv_60c = 0x0; + tv_regs->ptv_610 = 0x1e00000; + + if (tv_norm->tv_enc_mode.vdisplay == 576) { + tv_regs->ptv_508 = 0x1200000; + tv_regs->ptv_614 = 0x33; + + } else if (tv_norm->tv_enc_mode.vdisplay == 480) { + tv_regs->ptv_508 = 0xf00000; + tv_regs->ptv_614 = 0x13; + } + + if (dev_priv->card_type >= NV_30) { + tv_regs->ptv_500 = 0xe8e0; + tv_regs->ptv_504 = 0x1710; + tv_regs->ptv_604 = 0x0; + tv_regs->ptv_608 = 0x0; + } else { + if (tv_norm->tv_enc_mode.vdisplay == 576) { + tv_regs->ptv_604 = 0x20; + tv_regs->ptv_608 = 0x10; + tv_regs->ptv_500 = 0x19710; + tv_regs->ptv_504 = 0x68f0; + + } else if (tv_norm->tv_enc_mode.vdisplay == 480) { + tv_regs->ptv_604 = 0x10; + tv_regs->ptv_608 = 0x20; + tv_regs->ptv_500 = 0x4b90; + tv_regs->ptv_504 = 0x1b480; + } + } + + for (i = 0; i < 0x40; i++) + tv_regs->tv_enc[i] = tv_norm->tv_enc_mode.tv_enc[i]; + + } else { + struct drm_display_mode *output_mode = + &tv_norm->ctv_enc_mode.mode; + + /* The registers in PRAMDAC+0xc00 control some timings and CSC + * parameters for the CTV encoder (It's only used for "HD" TV + * modes, I don't think I have enough working to guess what + * they exactly mean...), it's probably connected at the + * output of the FP encoder, but it also needs the analog + * encoder in its OR enabled and routed to the head it's + * using. It's enabled with the DACCLK register, bits [5:4]. + */ + for (i = 0; i < 38; i++) + regs->ctv_regs[i] = tv_norm->ctv_enc_mode.ctv_regs[i]; + + regs->fp_horiz_regs[FP_DISPLAY_END] = output_mode->hdisplay - 1; + regs->fp_horiz_regs[FP_TOTAL] = output_mode->htotal - 1; + regs->fp_horiz_regs[FP_SYNC_START] = + output_mode->hsync_start - 1; + regs->fp_horiz_regs[FP_SYNC_END] = output_mode->hsync_end - 1; + regs->fp_horiz_regs[FP_CRTC] = output_mode->hdisplay + + max((output_mode->hdisplay-600)/40 - 1, 1); + + regs->fp_vert_regs[FP_DISPLAY_END] = output_mode->vdisplay - 1; + regs->fp_vert_regs[FP_TOTAL] = output_mode->vtotal - 1; + regs->fp_vert_regs[FP_SYNC_START] = + output_mode->vsync_start - 1; + regs->fp_vert_regs[FP_SYNC_END] = output_mode->vsync_end - 1; + regs->fp_vert_regs[FP_CRTC] = output_mode->vdisplay - 1; + + regs->fp_control = NV_PRAMDAC_FP_TG_CONTROL_DISPEN_POS | + NV_PRAMDAC_FP_TG_CONTROL_READ_PROG | + NV_PRAMDAC_FP_TG_CONTROL_WIDTH_12; + + if (output_mode->flags & DRM_MODE_FLAG_PVSYNC) + regs->fp_control |= NV_PRAMDAC_FP_TG_CONTROL_VSYNC_POS; + if (output_mode->flags & DRM_MODE_FLAG_PHSYNC) + regs->fp_control |= NV_PRAMDAC_FP_TG_CONTROL_HSYNC_POS; + + regs->fp_debug_0 = NV_PRAMDAC_FP_DEBUG_0_YWEIGHT_ROUND | + NV_PRAMDAC_FP_DEBUG_0_XWEIGHT_ROUND | + NV_PRAMDAC_FP_DEBUG_0_YINTERP_BILINEAR | + NV_PRAMDAC_FP_DEBUG_0_XINTERP_BILINEAR | + NV_RAMDAC_FP_DEBUG_0_TMDS_ENABLED | + NV_PRAMDAC_FP_DEBUG_0_YSCALE_ENABLE | + NV_PRAMDAC_FP_DEBUG_0_XSCALE_ENABLE; + + regs->fp_debug_2 = 0; + + regs->fp_margin_color = 0x801080; + + } +} + +static void nv17_tv_commit(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_crtc *nv_crtc = nouveau_crtc(encoder->crtc); + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + struct drm_encoder_helper_funcs *helper = encoder->helper_private; + + if (get_tv_norm(encoder)->kind == TV_ENC_MODE) { + nv17_tv_update_rescaler(encoder); + nv17_tv_update_properties(encoder); + } else { + nv17_ctv_update_rescaler(encoder); + } + + nv17_tv_state_load(dev, &to_tv_enc(encoder)->state); + + /* This could use refinement for flatpanels, but it should work */ + if (dev_priv->chipset < 0x44) + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_TEST_CONTROL + + nv04_dac_output_offset(encoder), + 0xf0000000); + else + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_TEST_CONTROL + + nv04_dac_output_offset(encoder), + 0x00100000); + + helper->dpms(encoder, DRM_MODE_DPMS_ON); + + NV_INFO(dev, "Output %s is running on CRTC %d using output %c\n", + drm_get_connector_name( + &nouveau_encoder_connector_get(nv_encoder)->base), + nv_crtc->index, '@' + ffs(nv_encoder->dcb->or)); +} + +static void nv17_tv_save(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct nv17_tv_encoder *tv_enc = to_tv_enc(encoder); + + nouveau_encoder(encoder)->restore.output = + NVReadRAMDAC(dev, 0, + NV_PRAMDAC_DACCLK + + nv04_dac_output_offset(encoder)); + + nv17_tv_state_save(dev, &tv_enc->saved_state); + + tv_enc->state.ptv_200 = tv_enc->saved_state.ptv_200; +} + +static void nv17_tv_restore(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + + NVWriteRAMDAC(dev, 0, NV_PRAMDAC_DACCLK + + nv04_dac_output_offset(encoder), + nouveau_encoder(encoder)->restore.output); + + nv17_tv_state_load(dev, &to_tv_enc(encoder)->saved_state); + + nouveau_encoder(encoder)->last_dpms = NV_DPMS_CLEARED; +} + +static int nv17_tv_create_resources(struct drm_encoder *encoder, + struct drm_connector *connector) +{ + struct drm_device *dev = encoder->dev; + struct drm_mode_config *conf = &dev->mode_config; + struct nv17_tv_encoder *tv_enc = to_tv_enc(encoder); + struct dcb_entry *dcb = nouveau_encoder(encoder)->dcb; + int num_tv_norms = dcb->tvconf.has_component_output ? NUM_TV_NORMS : + NUM_LD_TV_NORMS; + int i; + + if (nouveau_tv_norm) { + for (i = 0; i < num_tv_norms; i++) { + if (!strcmp(nv17_tv_norm_names[i], nouveau_tv_norm)) { + tv_enc->tv_norm = i; + break; + } + } + + if (i == num_tv_norms) + NV_WARN(dev, "Invalid TV norm setting \"%s\"\n", + nouveau_tv_norm); + } + + drm_mode_create_tv_properties(dev, num_tv_norms, nv17_tv_norm_names); + + drm_connector_attach_property(connector, + conf->tv_select_subconnector_property, + tv_enc->select_subconnector); + drm_connector_attach_property(connector, + conf->tv_subconnector_property, + tv_enc->subconnector); + drm_connector_attach_property(connector, + conf->tv_mode_property, + tv_enc->tv_norm); + drm_connector_attach_property(connector, + conf->tv_flicker_reduction_property, + tv_enc->flicker); + drm_connector_attach_property(connector, + conf->tv_saturation_property, + tv_enc->saturation); + drm_connector_attach_property(connector, + conf->tv_hue_property, + tv_enc->hue); + drm_connector_attach_property(connector, + conf->tv_overscan_property, + tv_enc->overscan); + + return 0; +} + +static int nv17_tv_set_property(struct drm_encoder *encoder, + struct drm_connector *connector, + struct drm_property *property, + uint64_t val) +{ + struct drm_mode_config *conf = &encoder->dev->mode_config; + struct drm_crtc *crtc = encoder->crtc; + struct nv17_tv_encoder *tv_enc = to_tv_enc(encoder); + struct nv17_tv_norm_params *tv_norm = get_tv_norm(encoder); + bool modes_changed = false; + + if (property == conf->tv_overscan_property) { + tv_enc->overscan = val; + if (encoder->crtc) { + if (tv_norm->kind == CTV_ENC_MODE) + nv17_ctv_update_rescaler(encoder); + else + nv17_tv_update_rescaler(encoder); + } + + } else if (property == conf->tv_saturation_property) { + if (tv_norm->kind != TV_ENC_MODE) + return -EINVAL; + + tv_enc->saturation = val; + nv17_tv_update_properties(encoder); + + } else if (property == conf->tv_hue_property) { + if (tv_norm->kind != TV_ENC_MODE) + return -EINVAL; + + tv_enc->hue = val; + nv17_tv_update_properties(encoder); + + } else if (property == conf->tv_flicker_reduction_property) { + if (tv_norm->kind != TV_ENC_MODE) + return -EINVAL; + + tv_enc->flicker = val; + if (encoder->crtc) + nv17_tv_update_rescaler(encoder); + + } else if (property == conf->tv_mode_property) { + if (connector->dpms != DRM_MODE_DPMS_OFF) + return -EINVAL; + + tv_enc->tv_norm = val; + + modes_changed = true; + + } else if (property == conf->tv_select_subconnector_property) { + if (tv_norm->kind != TV_ENC_MODE) + return -EINVAL; + + tv_enc->select_subconnector = val; + nv17_tv_update_properties(encoder); + + } else { + return -EINVAL; + } + + if (modes_changed) { + drm_helper_probe_single_connector_modes(connector, 0, 0); + + /* Disable the crtc to ensure a full modeset is + * performed whenever it's turned on again. */ + if (crtc) { + struct drm_mode_set modeset = { + .crtc = crtc, + }; + + crtc->funcs->set_config(&modeset); + } + } + + return 0; +} + +static void nv17_tv_destroy(struct drm_encoder *encoder) +{ + struct nv17_tv_encoder *tv_enc = to_tv_enc(encoder); + + NV_DEBUG_KMS(encoder->dev, "\n"); + + drm_encoder_cleanup(encoder); + kfree(tv_enc); +} + +static struct drm_encoder_helper_funcs nv17_tv_helper_funcs = { + .dpms = nv17_tv_dpms, + .save = nv17_tv_save, + .restore = nv17_tv_restore, + .mode_fixup = nv17_tv_mode_fixup, + .prepare = nv17_tv_prepare, + .commit = nv17_tv_commit, + .mode_set = nv17_tv_mode_set, + .detect = nv17_tv_detect, +}; + +static struct drm_encoder_slave_funcs nv17_tv_slave_funcs = { + .get_modes = nv17_tv_get_modes, + .mode_valid = nv17_tv_mode_valid, + .create_resources = nv17_tv_create_resources, + .set_property = nv17_tv_set_property, +}; + +static struct drm_encoder_funcs nv17_tv_funcs = { + .destroy = nv17_tv_destroy, +}; + +int +nv17_tv_create(struct drm_connector *connector, struct dcb_entry *entry) +{ + struct drm_device *dev = connector->dev; + struct drm_encoder *encoder; + struct nv17_tv_encoder *tv_enc = NULL; + + tv_enc = kzalloc(sizeof(*tv_enc), GFP_KERNEL); + if (!tv_enc) + return -ENOMEM; + + tv_enc->overscan = 50; + tv_enc->flicker = 50; + tv_enc->saturation = 50; + tv_enc->hue = 0; + tv_enc->tv_norm = TV_NORM_PAL; + tv_enc->subconnector = DRM_MODE_SUBCONNECTOR_Unknown; + tv_enc->select_subconnector = DRM_MODE_SUBCONNECTOR_Automatic; + tv_enc->pin_mask = 0; + + encoder = to_drm_encoder(&tv_enc->base); + + tv_enc->base.dcb = entry; + tv_enc->base.or = ffs(entry->or) - 1; + + drm_encoder_init(dev, encoder, &nv17_tv_funcs, DRM_MODE_ENCODER_TVDAC); + drm_encoder_helper_add(encoder, &nv17_tv_helper_funcs); + to_encoder_slave(encoder)->slave_funcs = &nv17_tv_slave_funcs; + + encoder->possible_crtcs = entry->heads; + encoder->possible_clones = 0; + + nv17_tv_create_resources(encoder, connector); + drm_mode_connector_attach_encoder(connector, encoder); + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nv17_tv.h b/drivers/gpu/drm/nouveau/nv17_tv.h new file mode 100644 index 00000000..622e7222 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv17_tv.h @@ -0,0 +1,161 @@ +/* + * Copyright (C) 2009 Francisco Jerez. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __NV17_TV_H__ +#define __NV17_TV_H__ + +struct nv17_tv_state { + uint8_t tv_enc[0x40]; + + uint32_t hfilter[4][7]; + uint32_t hfilter2[4][7]; + uint32_t vfilter[4][7]; + + uint32_t ptv_200; + uint32_t ptv_204; + uint32_t ptv_208; + uint32_t ptv_20c; + uint32_t ptv_304; + uint32_t ptv_500; + uint32_t ptv_504; + uint32_t ptv_508; + uint32_t ptv_600; + uint32_t ptv_604; + uint32_t ptv_608; + uint32_t ptv_60c; + uint32_t ptv_610; + uint32_t ptv_614; +}; + +enum nv17_tv_norm{ + TV_NORM_PAL, + TV_NORM_PAL_M, + TV_NORM_PAL_N, + TV_NORM_PAL_NC, + TV_NORM_NTSC_M, + TV_NORM_NTSC_J, + NUM_LD_TV_NORMS, + TV_NORM_HD480I = NUM_LD_TV_NORMS, + TV_NORM_HD480P, + TV_NORM_HD576I, + TV_NORM_HD576P, + TV_NORM_HD720P, + TV_NORM_HD1080I, + NUM_TV_NORMS +}; + +struct nv17_tv_encoder { + struct nouveau_encoder base; + + struct nv17_tv_state state; + struct nv17_tv_state saved_state; + + int overscan; + int flicker; + int saturation; + int hue; + enum nv17_tv_norm tv_norm; + int subconnector; + int select_subconnector; + uint32_t pin_mask; +}; +#define to_tv_enc(x) container_of(nouveau_encoder(x), \ + struct nv17_tv_encoder, base) + +extern char *nv17_tv_norm_names[NUM_TV_NORMS]; + +extern struct nv17_tv_norm_params { + enum { + TV_ENC_MODE, + CTV_ENC_MODE, + } kind; + + union { + struct { + int hdisplay; + int vdisplay; + int vrefresh; /* mHz */ + + uint8_t tv_enc[0x40]; + } tv_enc_mode; + + struct { + struct drm_display_mode mode; + + uint32_t ctv_regs[38]; + } ctv_enc_mode; + }; + +} nv17_tv_norms[NUM_TV_NORMS]; +#define get_tv_norm(enc) (&nv17_tv_norms[to_tv_enc(enc)->tv_norm]) + +extern const struct drm_display_mode nv17_tv_modes[]; + +static inline int interpolate(int y0, int y1, int y2, int x) +{ + return y1 + (x < 50 ? y1 - y0 : y2 - y1) * (x - 50) / 50; +} + +void nv17_tv_state_save(struct drm_device *dev, struct nv17_tv_state *state); +void nv17_tv_state_load(struct drm_device *dev, struct nv17_tv_state *state); +void nv17_tv_update_properties(struct drm_encoder *encoder); +void nv17_tv_update_rescaler(struct drm_encoder *encoder); +void nv17_ctv_update_rescaler(struct drm_encoder *encoder); + +/* TV hardware access functions */ + +static inline void nv_write_ptv(struct drm_device *dev, uint32_t reg, + uint32_t val) +{ + nv_wr32(dev, reg, val); +} + +static inline uint32_t nv_read_ptv(struct drm_device *dev, uint32_t reg) +{ + return nv_rd32(dev, reg); +} + +static inline void nv_write_tv_enc(struct drm_device *dev, uint8_t reg, + uint8_t val) +{ + nv_write_ptv(dev, NV_PTV_TV_INDEX, reg); + nv_write_ptv(dev, NV_PTV_TV_DATA, val); +} + +static inline uint8_t nv_read_tv_enc(struct drm_device *dev, uint8_t reg) +{ + nv_write_ptv(dev, NV_PTV_TV_INDEX, reg); + return nv_read_ptv(dev, NV_PTV_TV_DATA); +} + +#define nv_load_ptv(dev, state, reg) \ + nv_write_ptv(dev, NV_PTV_OFFSET + 0x##reg, state->ptv_##reg) +#define nv_save_ptv(dev, state, reg) \ + state->ptv_##reg = nv_read_ptv(dev, NV_PTV_OFFSET + 0x##reg) +#define nv_load_tv_enc(dev, state, reg) \ + nv_write_tv_enc(dev, 0x##reg, state->tv_enc[0x##reg]) + +#endif diff --git a/drivers/gpu/drm/nouveau/nv17_tv_modes.c b/drivers/gpu/drm/nouveau/nv17_tv_modes.c new file mode 100644 index 00000000..4d1d29f6 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv17_tv_modes.c @@ -0,0 +1,593 @@ +/* + * Copyright (C) 2009 Francisco Jerez. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "drmP.h" +#include "drm_crtc_helper.h" +#include "nouveau_drv.h" +#include "nouveau_encoder.h" +#include "nouveau_crtc.h" +#include "nouveau_hw.h" +#include "nv17_tv.h" + +char *nv17_tv_norm_names[NUM_TV_NORMS] = { + [TV_NORM_PAL] = "PAL", + [TV_NORM_PAL_M] = "PAL-M", + [TV_NORM_PAL_N] = "PAL-N", + [TV_NORM_PAL_NC] = "PAL-Nc", + [TV_NORM_NTSC_M] = "NTSC-M", + [TV_NORM_NTSC_J] = "NTSC-J", + [TV_NORM_HD480I] = "hd480i", + [TV_NORM_HD480P] = "hd480p", + [TV_NORM_HD576I] = "hd576i", + [TV_NORM_HD576P] = "hd576p", + [TV_NORM_HD720P] = "hd720p", + [TV_NORM_HD1080I] = "hd1080i" +}; + +/* TV standard specific parameters */ + +struct nv17_tv_norm_params nv17_tv_norms[NUM_TV_NORMS] = { + [TV_NORM_PAL] = { TV_ENC_MODE, { + .tv_enc_mode = { 720, 576, 50000, { + 0x2a, 0x9, 0x8a, 0xcb, 0x0, 0x0, 0xb, 0x18, + 0x7e, 0x40, 0x8a, 0x35, 0x27, 0x0, 0x34, 0x3, + 0x3e, 0x3, 0x17, 0x21, 0x1b, 0x1b, 0x24, 0x9c, + 0x1, 0x0, 0xf, 0xf, 0x60, 0x5, 0xd3, 0x3, + 0xd3, 0x4, 0xd4, 0x1, 0x2, 0x0, 0xa, 0x5, + 0x0, 0x1a, 0xff, 0x3, 0x18, 0xf, 0x78, 0x0, + 0x0, 0xb4, 0x0, 0x15, 0x49, 0x10, 0x0, 0x9b, + 0xbd, 0x15, 0x5, 0x15, 0x3e, 0x3, 0x0, 0x0 + } } } }, + + [TV_NORM_PAL_M] = { TV_ENC_MODE, { + .tv_enc_mode = { 720, 480, 59940, { + 0x21, 0xe6, 0xef, 0xe3, 0x0, 0x0, 0xb, 0x18, + 0x7e, 0x44, 0x76, 0x32, 0x25, 0x0, 0x3c, 0x0, + 0x3c, 0x0, 0x17, 0x21, 0x1b, 0x1b, 0x24, 0x83, + 0x1, 0x0, 0xf, 0xf, 0x60, 0x5, 0xd3, 0x1, + 0xc5, 0x4, 0xc5, 0x1, 0x2, 0x0, 0xa, 0x5, + 0x0, 0x18, 0xff, 0x3, 0x20, 0xf, 0x78, 0x0, + 0x0, 0xb4, 0x0, 0x15, 0x40, 0x10, 0x0, 0x9c, + 0xc8, 0x15, 0x5, 0x15, 0x3c, 0x0, 0x0, 0x0 + } } } }, + + [TV_NORM_PAL_N] = { TV_ENC_MODE, { + .tv_enc_mode = { 720, 576, 50000, { + 0x2a, 0x9, 0x8a, 0xcb, 0x0, 0x0, 0xb, 0x18, + 0x7e, 0x40, 0x8a, 0x32, 0x25, 0x0, 0x3c, 0x0, + 0x3c, 0x0, 0x17, 0x21, 0x1b, 0x1b, 0x24, 0x9c, + 0x1, 0x0, 0xf, 0xf, 0x60, 0x5, 0xd3, 0x1, + 0xc5, 0x4, 0xc5, 0x1, 0x2, 0x0, 0xa, 0x5, + 0x0, 0x1a, 0xff, 0x3, 0x18, 0xf, 0x78, 0x0, + 0x0, 0xb4, 0x0, 0x15, 0x49, 0x10, 0x0, 0x9b, + 0xbd, 0x15, 0x5, 0x15, 0x3c, 0x0, 0x0, 0x0 + } } } }, + + [TV_NORM_PAL_NC] = { TV_ENC_MODE, { + .tv_enc_mode = { 720, 576, 50000, { + 0x21, 0xf6, 0x94, 0x46, 0x0, 0x0, 0xb, 0x18, + 0x7e, 0x44, 0x8a, 0x35, 0x27, 0x0, 0x34, 0x3, + 0x3e, 0x3, 0x17, 0x21, 0x1b, 0x1b, 0x24, 0x9c, + 0x1, 0x0, 0xf, 0xf, 0x60, 0x5, 0xd3, 0x3, + 0xd3, 0x4, 0xd4, 0x1, 0x2, 0x0, 0xa, 0x5, + 0x0, 0x1a, 0xff, 0x3, 0x18, 0xf, 0x78, 0x0, + 0x0, 0xb4, 0x0, 0x15, 0x49, 0x10, 0x0, 0x9b, + 0xbd, 0x15, 0x5, 0x15, 0x3e, 0x3, 0x0, 0x0 + } } } }, + + [TV_NORM_NTSC_M] = { TV_ENC_MODE, { + .tv_enc_mode = { 720, 480, 59940, { + 0x21, 0xf0, 0x7c, 0x1f, 0x0, 0x0, 0xb, 0x18, + 0x7e, 0x44, 0x76, 0x48, 0x0, 0x0, 0x3c, 0x0, + 0x3c, 0x0, 0x17, 0x21, 0x1b, 0x1b, 0x24, 0x83, + 0x1, 0x0, 0xf, 0xf, 0x60, 0x5, 0xd3, 0x1, + 0xc5, 0x4, 0xc5, 0x1, 0x2, 0x0, 0xa, 0x5, + 0x0, 0x16, 0xff, 0x3, 0x20, 0xf, 0x78, 0x0, + 0x0, 0xb4, 0x0, 0x15, 0x4, 0x10, 0x0, 0x9c, + 0xc8, 0x15, 0x5, 0x15, 0x3c, 0x0, 0x0, 0x0 + } } } }, + + [TV_NORM_NTSC_J] = { TV_ENC_MODE, { + .tv_enc_mode = { 720, 480, 59940, { + 0x21, 0xf0, 0x7c, 0x1f, 0x0, 0x0, 0xb, 0x18, + 0x7e, 0x44, 0x76, 0x48, 0x0, 0x0, 0x32, 0x0, + 0x3c, 0x0, 0x17, 0x21, 0x1b, 0x1b, 0x24, 0x83, + 0x1, 0x0, 0xf, 0xf, 0x60, 0x5, 0xd3, 0x1, + 0xcf, 0x4, 0xcf, 0x1, 0x2, 0x0, 0xa, 0x5, + 0x0, 0x16, 0xff, 0x3, 0x20, 0xf, 0x78, 0x0, + 0x0, 0xb4, 0x0, 0x15, 0x4, 0x10, 0x0, 0xa4, + 0xc8, 0x15, 0x5, 0x15, 0x3c, 0x0, 0x0, 0x0 + } } } }, + + [TV_NORM_HD480I] = { TV_ENC_MODE, { + .tv_enc_mode = { 720, 480, 59940, { + 0x21, 0xf0, 0x7c, 0x1f, 0x0, 0x0, 0xb, 0x18, + 0x7e, 0x44, 0x76, 0x48, 0x0, 0x0, 0x32, 0x0, + 0x3c, 0x0, 0x17, 0x21, 0x1b, 0x1b, 0x24, 0x83, + 0x1, 0x0, 0xf, 0xf, 0x60, 0x5, 0xd3, 0x1, + 0xcf, 0x4, 0xcf, 0x1, 0x2, 0x0, 0xa, 0x5, + 0x0, 0x16, 0xff, 0x3, 0x20, 0xf, 0x78, 0x0, + 0x0, 0xb4, 0x0, 0x15, 0x4, 0x10, 0x0, 0xa4, + 0xc8, 0x15, 0x5, 0x15, 0x3c, 0x0, 0x0, 0x0 + } } } }, + + [TV_NORM_HD576I] = { TV_ENC_MODE, { + .tv_enc_mode = { 720, 576, 50000, { + 0x2a, 0x9, 0x8a, 0xcb, 0x0, 0x0, 0xb, 0x18, + 0x7e, 0x40, 0x8a, 0x35, 0x27, 0x0, 0x34, 0x3, + 0x3e, 0x3, 0x17, 0x21, 0x1b, 0x1b, 0x24, 0x9c, + 0x1, 0x0, 0xf, 0xf, 0x60, 0x5, 0xd3, 0x3, + 0xd3, 0x4, 0xd4, 0x1, 0x2, 0x0, 0xa, 0x5, + 0x0, 0x1a, 0xff, 0x3, 0x18, 0xf, 0x78, 0x0, + 0x0, 0xb4, 0x0, 0x15, 0x49, 0x10, 0x0, 0x9b, + 0xbd, 0x15, 0x5, 0x15, 0x3e, 0x3, 0x0, 0x0 + } } } }, + + + [TV_NORM_HD480P] = { CTV_ENC_MODE, { + .ctv_enc_mode = { + .mode = { DRM_MODE("720x480", DRM_MODE_TYPE_DRIVER, 27000, + 720, 735, 743, 858, 0, 480, 490, 494, 525, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + .ctv_regs = { 0x3540000, 0x0, 0x0, 0x314, + 0x354003a, 0x40000, 0x6f0344, 0x18100000, + 0x10160004, 0x10060005, 0x1006000c, 0x10060020, + 0x10060021, 0x140e0022, 0x10060202, 0x1802020a, + 0x1810020b, 0x10000fff, 0x10000fff, 0x10000fff, + 0x10000fff, 0x10000fff, 0x10000fff, 0x70, + 0x3ff0000, 0x57, 0x2e001e, 0x258012c, + 0xa0aa04ec, 0x30, 0x80960019, 0x12c0300, + 0x2019, 0x600, 0x32060019, 0x0, 0x0, 0x400 + } } } }, + + [TV_NORM_HD576P] = { CTV_ENC_MODE, { + .ctv_enc_mode = { + .mode = { DRM_MODE("720x576", DRM_MODE_TYPE_DRIVER, 27000, + 720, 730, 738, 864, 0, 576, 581, 585, 625, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + .ctv_regs = { 0x3540000, 0x0, 0x0, 0x314, + 0x354003a, 0x40000, 0x6f0344, 0x18100000, + 0x10060001, 0x10060009, 0x10060026, 0x10060027, + 0x140e0028, 0x10060268, 0x1810026d, 0x10000fff, + 0x10000fff, 0x10000fff, 0x10000fff, 0x10000fff, + 0x10000fff, 0x10000fff, 0x10000fff, 0x69, + 0x3ff0000, 0x57, 0x2e001e, 0x258012c, + 0xa0aa04ec, 0x30, 0x80960019, 0x12c0300, + 0x2019, 0x600, 0x32060019, 0x0, 0x0, 0x400 + } } } }, + + [TV_NORM_HD720P] = { CTV_ENC_MODE, { + .ctv_enc_mode = { + .mode = { DRM_MODE("1280x720", DRM_MODE_TYPE_DRIVER, 74250, + 1280, 1349, 1357, 1650, 0, 720, 725, 730, 750, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + .ctv_regs = { 0x1260394, 0x0, 0x0, 0x622, + 0x66b0021, 0x6004a, 0x1210626, 0x8170000, + 0x70004, 0x70016, 0x70017, 0x40f0018, + 0x702e8, 0x81702ed, 0xfff, 0xfff, + 0xfff, 0xfff, 0xfff, 0xfff, + 0xfff, 0xfff, 0xfff, 0x0, + 0x2e40001, 0x58, 0x2e001e, 0x258012c, + 0xa0aa04ec, 0x30, 0x810c0039, 0x12c0300, + 0xc0002039, 0x600, 0x32060039, 0x0, 0x0, 0x0 + } } } }, + + [TV_NORM_HD1080I] = { CTV_ENC_MODE, { + .ctv_enc_mode = { + .mode = { DRM_MODE("1920x1080", DRM_MODE_TYPE_DRIVER, 74250, + 1920, 1961, 2049, 2200, 0, 1080, 1084, 1088, 1125, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC + | DRM_MODE_FLAG_INTERLACE) }, + .ctv_regs = { 0xac0420, 0x44c0478, 0x4a4, 0x4fc0868, + 0x8940028, 0x60054, 0xe80870, 0xbf70000, + 0xbc70004, 0x70005, 0x70012, 0x70013, + 0x40f0014, 0x70230, 0xbf70232, 0xbf70233, + 0x1c70237, 0x70238, 0x70244, 0x70245, + 0x40f0246, 0x70462, 0x1f70464, 0x0, + 0x2e40001, 0x58, 0x2e001e, 0x258012c, + 0xa0aa04ec, 0x30, 0x815f004c, 0x12c0300, + 0xc000204c, 0x600, 0x3206004c, 0x0, 0x0, 0x0 + } } } } +}; + +/* + * The following is some guesswork on how the TV encoder flicker + * filter/rescaler works: + * + * It seems to use some sort of resampling filter, it is controlled + * through the registers at NV_PTV_HFILTER and NV_PTV_VFILTER, they + * control the horizontal and vertical stage respectively, there is + * also NV_PTV_HFILTER2 the blob fills identically to NV_PTV_HFILTER, + * but they seem to do nothing. A rough guess might be that they could + * be used to independently control the filtering of each interlaced + * field, but I don't know how they are enabled. The whole filtering + * process seems to be disabled with bits 26:27 of PTV_200, but we + * aren't doing that. + * + * The layout of both register sets is the same: + * + * A: [BASE+0x18]...[BASE+0x0] [BASE+0x58]..[BASE+0x40] + * B: [BASE+0x34]...[BASE+0x1c] [BASE+0x74]..[BASE+0x5c] + * + * Each coefficient is stored in bits [31],[15:9] in two's complement + * format. They seem to be some kind of weights used in a low-pass + * filter. Both A and B coefficients are applied to the 14 nearest + * samples on each side (Listed from nearest to furthermost. They + * roughly cover 2 framebuffer pixels on each side). They are + * probably multiplied with some more hardwired weights before being + * used: B-coefficients are applied the same on both sides, + * A-coefficients are inverted before being applied to the opposite + * side. + * + * After all the hassle, I got the following formula by empirical + * means... + */ + +#define calc_overscan(o) interpolate(0x100, 0xe1, 0xc1, o) + +#define id1 (1LL << 8) +#define id2 (1LL << 16) +#define id3 (1LL << 24) +#define id4 (1LL << 32) +#define id5 (1LL << 48) + +static struct filter_params{ + int64_t k1; + int64_t ki; + int64_t ki2; + int64_t ki3; + int64_t kr; + int64_t kir; + int64_t ki2r; + int64_t ki3r; + int64_t kf; + int64_t kif; + int64_t ki2f; + int64_t ki3f; + int64_t krf; + int64_t kirf; + int64_t ki2rf; + int64_t ki3rf; +} fparams[2][4] = { + /* Horizontal filter parameters */ + { + {64.311690 * id5, -39.516924 * id5, 6.586143 * id5, 0.000002 * id5, + 0.051285 * id4, 26.168746 * id4, -4.361449 * id4, -0.000001 * id4, + 9.308169 * id3, 78.180965 * id3, -13.030158 * id3, -0.000001 * id3, + -8.801540 * id1, -46.572890 * id1, 7.762145 * id1, -0.000000 * id1}, + {-44.565569 * id5, -68.081246 * id5, 39.812074 * id5, -4.009316 * id5, + 29.832207 * id4, 50.047322 * id4, -25.380017 * id4, 2.546422 * id4, + 104.605622 * id3, 141.908641 * id3, -74.322319 * id3, 7.484316 * id3, + -37.081621 * id1, -90.397510 * id1, 42.784229 * id1, -4.289952 * id1}, + {-56.793244 * id5, 31.153584 * id5, -5.192247 * id5, -0.000003 * id5, + 33.541131 * id4, -34.149302 * id4, 5.691537 * id4, 0.000002 * id4, + 87.196610 * id3, -88.995169 * id3, 14.832456 * id3, 0.000012 * id3, + 17.288138 * id1, 71.864786 * id1, -11.977408 * id1, -0.000009 * id1}, + {51.787796 * id5, 21.211771 * id5, -18.993730 * id5, 1.853310 * id5, + -41.470726 * id4, -17.775823 * id4, 13.057821 * id4, -1.15823 * id4, + -154.235673 * id3, -44.878641 * id3, 40.656077 * id3, -3.695595 * id3, + 112.201065 * id1, 39.992155 * id1, -25.155714 * id1, 2.113984 * id1}, + }, + + /* Vertical filter parameters */ + { + {67.601979 * id5, 0.428319 * id5, -0.071318 * id5, -0.000012 * id5, + -3.402339 * id4, 0.000209 * id4, -0.000092 * id4, 0.000010 * id4, + -9.180996 * id3, 6.111270 * id3, -1.024457 * id3, 0.001043 * id3, + 6.060315 * id1, -0.017425 * id1, 0.007830 * id1, -0.000869 * id1}, + {6.755647 * id5, 5.841348 * id5, 1.469734 * id5, -0.149656 * id5, + 8.293120 * id4, -1.192888 * id4, -0.947652 * id4, 0.094507 * id4, + 37.526655 * id3, 10.257875 * id3, -10.823275 * id3, 1.081497 * id3, + -2.361928 * id1, -2.059432 * id1, 1.840671 * id1, -0.168100 * id1}, + {-14.780391 * id5, -16.042148 * id5, 2.673692 * id5, -0.000000 * id5, + 39.541978 * id4, 5.680053 * id4, -0.946676 * id4, 0.000000 * id4, + 152.994486 * id3, 12.625439 * id3, -2.119579 * id3, 0.002708 * id3, + -38.125089 * id1, -0.855880 * id1, 0.155359 * id1, -0.002245 * id1}, + {-27.476193 * id5, -1.454976 * id5, 1.286557 * id5, 0.025346 * id5, + 20.687300 * id4, 3.014003 * id4, -0.557786 * id4, -0.01311 * id4, + 60.008737 * id3, -0.738273 * id3, 5.408217 * id3, -0.796798 * id3, + -17.296835 * id1, 4.438577 * id1, -2.809420 * id1, 0.385491 * id1}, + } +}; + +static void tv_setup_filter(struct drm_encoder *encoder) +{ + struct nv17_tv_encoder *tv_enc = to_tv_enc(encoder); + struct nv17_tv_norm_params *tv_norm = get_tv_norm(encoder); + struct drm_display_mode *mode = &encoder->crtc->mode; + uint32_t (*filters[])[4][7] = {&tv_enc->state.hfilter, + &tv_enc->state.vfilter}; + int i, j, k; + int32_t overscan = calc_overscan(tv_enc->overscan); + int64_t flicker = (tv_enc->flicker - 50) * (id3 / 100); + uint64_t rs[] = {mode->hdisplay * id3, + mode->vdisplay * id3}; + + do_div(rs[0], overscan * tv_norm->tv_enc_mode.hdisplay); + do_div(rs[1], overscan * tv_norm->tv_enc_mode.vdisplay); + + for (k = 0; k < 2; k++) { + rs[k] = max((int64_t)rs[k], id2); + + for (j = 0; j < 4; j++) { + struct filter_params *p = &fparams[k][j]; + + for (i = 0; i < 7; i++) { + int64_t c = (p->k1 + p->ki*i + p->ki2*i*i + + p->ki3*i*i*i) + + (p->kr + p->kir*i + p->ki2r*i*i + + p->ki3r*i*i*i) * rs[k] + + (p->kf + p->kif*i + p->ki2f*i*i + + p->ki3f*i*i*i) * flicker + + (p->krf + p->kirf*i + p->ki2rf*i*i + + p->ki3rf*i*i*i) * flicker * rs[k]; + + (*filters[k])[j][i] = (c + id5/2) >> 39 + & (0x1 << 31 | 0x7f << 9); + } + } + } +} + +/* Hardware state saving/restoring */ + +static void tv_save_filter(struct drm_device *dev, uint32_t base, + uint32_t regs[4][7]) +{ + int i, j; + uint32_t offsets[] = { base, base + 0x1c, base + 0x40, base + 0x5c }; + + for (i = 0; i < 4; i++) { + for (j = 0; j < 7; j++) + regs[i][j] = nv_read_ptv(dev, offsets[i]+4*j); + } +} + +static void tv_load_filter(struct drm_device *dev, uint32_t base, + uint32_t regs[4][7]) +{ + int i, j; + uint32_t offsets[] = { base, base + 0x1c, base + 0x40, base + 0x5c }; + + for (i = 0; i < 4; i++) { + for (j = 0; j < 7; j++) + nv_write_ptv(dev, offsets[i]+4*j, regs[i][j]); + } +} + +void nv17_tv_state_save(struct drm_device *dev, struct nv17_tv_state *state) +{ + int i; + + for (i = 0; i < 0x40; i++) + state->tv_enc[i] = nv_read_tv_enc(dev, i); + + tv_save_filter(dev, NV_PTV_HFILTER, state->hfilter); + tv_save_filter(dev, NV_PTV_HFILTER2, state->hfilter2); + tv_save_filter(dev, NV_PTV_VFILTER, state->vfilter); + + nv_save_ptv(dev, state, 200); + nv_save_ptv(dev, state, 204); + nv_save_ptv(dev, state, 208); + nv_save_ptv(dev, state, 20c); + nv_save_ptv(dev, state, 304); + nv_save_ptv(dev, state, 500); + nv_save_ptv(dev, state, 504); + nv_save_ptv(dev, state, 508); + nv_save_ptv(dev, state, 600); + nv_save_ptv(dev, state, 604); + nv_save_ptv(dev, state, 608); + nv_save_ptv(dev, state, 60c); + nv_save_ptv(dev, state, 610); + nv_save_ptv(dev, state, 614); +} + +void nv17_tv_state_load(struct drm_device *dev, struct nv17_tv_state *state) +{ + int i; + + for (i = 0; i < 0x40; i++) + nv_write_tv_enc(dev, i, state->tv_enc[i]); + + tv_load_filter(dev, NV_PTV_HFILTER, state->hfilter); + tv_load_filter(dev, NV_PTV_HFILTER2, state->hfilter2); + tv_load_filter(dev, NV_PTV_VFILTER, state->vfilter); + + nv_load_ptv(dev, state, 200); + nv_load_ptv(dev, state, 204); + nv_load_ptv(dev, state, 208); + nv_load_ptv(dev, state, 20c); + nv_load_ptv(dev, state, 304); + nv_load_ptv(dev, state, 500); + nv_load_ptv(dev, state, 504); + nv_load_ptv(dev, state, 508); + nv_load_ptv(dev, state, 600); + nv_load_ptv(dev, state, 604); + nv_load_ptv(dev, state, 608); + nv_load_ptv(dev, state, 60c); + nv_load_ptv(dev, state, 610); + nv_load_ptv(dev, state, 614); + + /* This is required for some settings to kick in. */ + nv_write_tv_enc(dev, 0x3e, 1); + nv_write_tv_enc(dev, 0x3e, 0); +} + +/* Timings similar to the ones the blob sets */ + +const struct drm_display_mode nv17_tv_modes[] = { + { DRM_MODE("320x200", DRM_MODE_TYPE_DRIVER, 0, + 320, 344, 392, 560, 0, 200, 200, 202, 220, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_NVSYNC + | DRM_MODE_FLAG_DBLSCAN | DRM_MODE_FLAG_CLKDIV2) }, + { DRM_MODE("320x240", DRM_MODE_TYPE_DRIVER, 0, + 320, 344, 392, 560, 0, 240, 240, 246, 263, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_NVSYNC + | DRM_MODE_FLAG_DBLSCAN | DRM_MODE_FLAG_CLKDIV2) }, + { DRM_MODE("400x300", DRM_MODE_TYPE_DRIVER, 0, + 400, 432, 496, 640, 0, 300, 300, 303, 314, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC + | DRM_MODE_FLAG_DBLSCAN | DRM_MODE_FLAG_CLKDIV2) }, + { DRM_MODE("640x480", DRM_MODE_TYPE_DRIVER, 0, + 640, 672, 768, 880, 0, 480, 480, 492, 525, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_NVSYNC) }, + { DRM_MODE("720x480", DRM_MODE_TYPE_DRIVER, 0, + 720, 752, 872, 960, 0, 480, 480, 493, 525, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_NVSYNC) }, + { DRM_MODE("720x576", DRM_MODE_TYPE_DRIVER, 0, + 720, 776, 856, 960, 0, 576, 576, 588, 597, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_NVSYNC) }, + { DRM_MODE("800x600", DRM_MODE_TYPE_DRIVER, 0, + 800, 840, 920, 1040, 0, 600, 600, 604, 618, 0, + DRM_MODE_FLAG_PHSYNC | DRM_MODE_FLAG_PVSYNC) }, + { DRM_MODE("1024x768", DRM_MODE_TYPE_DRIVER, 0, + 1024, 1064, 1200, 1344, 0, 768, 768, 777, 806, 0, + DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_NVSYNC) }, + {} +}; + +void nv17_tv_update_properties(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct nv17_tv_encoder *tv_enc = to_tv_enc(encoder); + struct nv17_tv_state *regs = &tv_enc->state; + struct nv17_tv_norm_params *tv_norm = get_tv_norm(encoder); + int subconnector = tv_enc->select_subconnector ? + tv_enc->select_subconnector : + tv_enc->subconnector; + + switch (subconnector) { + case DRM_MODE_SUBCONNECTOR_Composite: + { + regs->ptv_204 = 0x2; + + /* The composite connector may be found on either pin. */ + if (tv_enc->pin_mask & 0x4) + regs->ptv_204 |= 0x010000; + else if (tv_enc->pin_mask & 0x2) + regs->ptv_204 |= 0x100000; + else + regs->ptv_204 |= 0x110000; + + regs->tv_enc[0x7] = 0x10; + break; + } + case DRM_MODE_SUBCONNECTOR_SVIDEO: + regs->ptv_204 = 0x11012; + regs->tv_enc[0x7] = 0x18; + break; + + case DRM_MODE_SUBCONNECTOR_Component: + regs->ptv_204 = 0x111333; + regs->tv_enc[0x7] = 0x14; + break; + + case DRM_MODE_SUBCONNECTOR_SCART: + regs->ptv_204 = 0x111012; + regs->tv_enc[0x7] = 0x18; + break; + } + + regs->tv_enc[0x20] = interpolate(0, tv_norm->tv_enc_mode.tv_enc[0x20], + 255, tv_enc->saturation); + regs->tv_enc[0x22] = interpolate(0, tv_norm->tv_enc_mode.tv_enc[0x22], + 255, tv_enc->saturation); + regs->tv_enc[0x25] = tv_enc->hue * 255 / 100; + + nv_load_ptv(dev, regs, 204); + nv_load_tv_enc(dev, regs, 7); + nv_load_tv_enc(dev, regs, 20); + nv_load_tv_enc(dev, regs, 22); + nv_load_tv_enc(dev, regs, 25); +} + +void nv17_tv_update_rescaler(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct nv17_tv_encoder *tv_enc = to_tv_enc(encoder); + struct nv17_tv_state *regs = &tv_enc->state; + + regs->ptv_208 = 0x40 | (calc_overscan(tv_enc->overscan) << 8); + + tv_setup_filter(encoder); + + nv_load_ptv(dev, regs, 208); + tv_load_filter(dev, NV_PTV_HFILTER, regs->hfilter); + tv_load_filter(dev, NV_PTV_HFILTER2, regs->hfilter2); + tv_load_filter(dev, NV_PTV_VFILTER, regs->vfilter); +} + +void nv17_ctv_update_rescaler(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nv17_tv_encoder *tv_enc = to_tv_enc(encoder); + int head = nouveau_crtc(encoder->crtc)->index; + struct nv04_crtc_reg *regs = &dev_priv->mode_reg.crtc_reg[head]; + struct drm_display_mode *crtc_mode = &encoder->crtc->mode; + struct drm_display_mode *output_mode = + &get_tv_norm(encoder)->ctv_enc_mode.mode; + int overscan, hmargin, vmargin, hratio, vratio; + + /* The rescaler doesn't do the right thing for interlaced modes. */ + if (output_mode->flags & DRM_MODE_FLAG_INTERLACE) + overscan = 100; + else + overscan = tv_enc->overscan; + + hmargin = (output_mode->hdisplay - crtc_mode->hdisplay) / 2; + vmargin = (output_mode->vdisplay - crtc_mode->vdisplay) / 2; + + hmargin = interpolate(0, min(hmargin, output_mode->hdisplay/20), + hmargin, overscan); + vmargin = interpolate(0, min(vmargin, output_mode->vdisplay/20), + vmargin, overscan); + + hratio = crtc_mode->hdisplay * 0x800 / + (output_mode->hdisplay - 2*hmargin); + vratio = crtc_mode->vdisplay * 0x800 / + (output_mode->vdisplay - 2*vmargin) & ~3; + + regs->fp_horiz_regs[FP_VALID_START] = hmargin; + regs->fp_horiz_regs[FP_VALID_END] = output_mode->hdisplay - hmargin - 1; + regs->fp_vert_regs[FP_VALID_START] = vmargin; + regs->fp_vert_regs[FP_VALID_END] = output_mode->vdisplay - vmargin - 1; + + regs->fp_debug_1 = NV_PRAMDAC_FP_DEBUG_1_YSCALE_TESTMODE_ENABLE | + XLATE(vratio, 0, NV_PRAMDAC_FP_DEBUG_1_YSCALE_VALUE) | + NV_PRAMDAC_FP_DEBUG_1_XSCALE_TESTMODE_ENABLE | + XLATE(hratio, 0, NV_PRAMDAC_FP_DEBUG_1_XSCALE_VALUE); + + NVWriteRAMDAC(dev, head, NV_PRAMDAC_FP_HVALID_START, + regs->fp_horiz_regs[FP_VALID_START]); + NVWriteRAMDAC(dev, head, NV_PRAMDAC_FP_HVALID_END, + regs->fp_horiz_regs[FP_VALID_END]); + NVWriteRAMDAC(dev, head, NV_PRAMDAC_FP_VVALID_START, + regs->fp_vert_regs[FP_VALID_START]); + NVWriteRAMDAC(dev, head, NV_PRAMDAC_FP_VVALID_END, + regs->fp_vert_regs[FP_VALID_END]); + NVWriteRAMDAC(dev, head, NV_PRAMDAC_FP_DEBUG_1, regs->fp_debug_1); +} diff --git a/drivers/gpu/drm/nouveau/nv20_graph.c b/drivers/gpu/drm/nouveau/nv20_graph.c new file mode 100644 index 00000000..affc7d7d --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv20_graph.c @@ -0,0 +1,835 @@ +#include "drmP.h" +#include "drm.h" +#include "nouveau_drv.h" +#include "nouveau_drm.h" + +/* + * NV20 + * ----- + * There are 3 families : + * NV20 is 0x10de:0x020* + * NV25/28 is 0x10de:0x025* / 0x10de:0x028* + * NV2A is 0x10de:0x02A0 + * + * NV30 + * ----- + * There are 3 families : + * NV30/31 is 0x10de:0x030* / 0x10de:0x031* + * NV34 is 0x10de:0x032* + * NV35/36 is 0x10de:0x033* / 0x10de:0x034* + * + * Not seen in the wild, no dumps (probably NV35) : + * NV37 is 0x10de:0x00fc, 0x10de:0x00fd + * NV38 is 0x10de:0x0333, 0x10de:0x00fe + * + */ + +struct nv20_graph_engine { + struct nouveau_exec_engine base; + struct nouveau_gpuobj *ctxtab; + void (*grctx_init)(struct nouveau_gpuobj *); + u32 grctx_size; + u32 grctx_user; +}; + +#define NV20_GRCTX_SIZE (3580*4) +#define NV25_GRCTX_SIZE (3529*4) +#define NV2A_GRCTX_SIZE (3500*4) + +#define NV30_31_GRCTX_SIZE (24392) +#define NV34_GRCTX_SIZE (18140) +#define NV35_36_GRCTX_SIZE (22396) + +int +nv20_graph_unload_context(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; + struct nouveau_channel *chan; + struct nouveau_gpuobj *grctx; + u32 tmp; + + chan = nv10_graph_channel(dev); + if (!chan) + return 0; + grctx = chan->engctx[NVOBJ_ENGINE_GR]; + + nv_wr32(dev, NV20_PGRAPH_CHANNEL_CTX_POINTER, grctx->pinst >> 4); + nv_wr32(dev, NV20_PGRAPH_CHANNEL_CTX_XFER, + NV20_PGRAPH_CHANNEL_CTX_XFER_SAVE); + + nouveau_wait_for_idle(dev); + + nv_wr32(dev, NV10_PGRAPH_CTX_CONTROL, 0x10000000); + tmp = nv_rd32(dev, NV10_PGRAPH_CTX_USER) & 0x00ffffff; + tmp |= (pfifo->channels - 1) << 24; + nv_wr32(dev, NV10_PGRAPH_CTX_USER, tmp); + return 0; +} + +static void +nv20_graph_rdi(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + int i, writecount = 32; + uint32_t rdi_index = 0x2c80000; + + if (dev_priv->chipset == 0x20) { + rdi_index = 0x3d0000; + writecount = 15; + } + + nv_wr32(dev, NV10_PGRAPH_RDI_INDEX, rdi_index); + for (i = 0; i < writecount; i++) + nv_wr32(dev, NV10_PGRAPH_RDI_DATA, 0); + + nouveau_wait_for_idle(dev); +} + +static void +nv20_graph_context_init(struct nouveau_gpuobj *ctx) +{ + int i; + + nv_wo32(ctx, 0x033c, 0xffff0000); + nv_wo32(ctx, 0x03a0, 0x0fff0000); + nv_wo32(ctx, 0x03a4, 0x0fff0000); + nv_wo32(ctx, 0x047c, 0x00000101); + nv_wo32(ctx, 0x0490, 0x00000111); + nv_wo32(ctx, 0x04a8, 0x44400000); + for (i = 0x04d4; i <= 0x04e0; i += 4) + nv_wo32(ctx, i, 0x00030303); + for (i = 0x04f4; i <= 0x0500; i += 4) + nv_wo32(ctx, i, 0x00080000); + for (i = 0x050c; i <= 0x0518; i += 4) + nv_wo32(ctx, i, 0x01012000); + for (i = 0x051c; i <= 0x0528; i += 4) + nv_wo32(ctx, i, 0x000105b8); + for (i = 0x052c; i <= 0x0538; i += 4) + nv_wo32(ctx, i, 0x00080008); + for (i = 0x055c; i <= 0x0598; i += 4) + nv_wo32(ctx, i, 0x07ff0000); + nv_wo32(ctx, 0x05a4, 0x4b7fffff); + nv_wo32(ctx, 0x05fc, 0x00000001); + nv_wo32(ctx, 0x0604, 0x00004000); + nv_wo32(ctx, 0x0610, 0x00000001); + nv_wo32(ctx, 0x0618, 0x00040000); + nv_wo32(ctx, 0x061c, 0x00010000); + for (i = 0x1c1c; i <= 0x248c; i += 16) { + nv_wo32(ctx, (i + 0), 0x10700ff9); + nv_wo32(ctx, (i + 4), 0x0436086c); + nv_wo32(ctx, (i + 8), 0x000c001b); + } + nv_wo32(ctx, 0x281c, 0x3f800000); + nv_wo32(ctx, 0x2830, 0x3f800000); + nv_wo32(ctx, 0x285c, 0x40000000); + nv_wo32(ctx, 0x2860, 0x3f800000); + nv_wo32(ctx, 0x2864, 0x3f000000); + nv_wo32(ctx, 0x286c, 0x40000000); + nv_wo32(ctx, 0x2870, 0x3f800000); + nv_wo32(ctx, 0x2878, 0xbf800000); + nv_wo32(ctx, 0x2880, 0xbf800000); + nv_wo32(ctx, 0x34a4, 0x000fe000); + nv_wo32(ctx, 0x3530, 0x000003f8); + nv_wo32(ctx, 0x3540, 0x002fe000); + for (i = 0x355c; i <= 0x3578; i += 4) + nv_wo32(ctx, i, 0x001c527c); +} + +static void +nv25_graph_context_init(struct nouveau_gpuobj *ctx) +{ + int i; + + nv_wo32(ctx, 0x035c, 0xffff0000); + nv_wo32(ctx, 0x03c0, 0x0fff0000); + nv_wo32(ctx, 0x03c4, 0x0fff0000); + nv_wo32(ctx, 0x049c, 0x00000101); + nv_wo32(ctx, 0x04b0, 0x00000111); + nv_wo32(ctx, 0x04c8, 0x00000080); + nv_wo32(ctx, 0x04cc, 0xffff0000); + nv_wo32(ctx, 0x04d0, 0x00000001); + nv_wo32(ctx, 0x04e4, 0x44400000); + nv_wo32(ctx, 0x04fc, 0x4b800000); + for (i = 0x0510; i <= 0x051c; i += 4) + nv_wo32(ctx, i, 0x00030303); + for (i = 0x0530; i <= 0x053c; i += 4) + nv_wo32(ctx, i, 0x00080000); + for (i = 0x0548; i <= 0x0554; i += 4) + nv_wo32(ctx, i, 0x01012000); + for (i = 0x0558; i <= 0x0564; i += 4) + nv_wo32(ctx, i, 0x000105b8); + for (i = 0x0568; i <= 0x0574; i += 4) + nv_wo32(ctx, i, 0x00080008); + for (i = 0x0598; i <= 0x05d4; i += 4) + nv_wo32(ctx, i, 0x07ff0000); + nv_wo32(ctx, 0x05e0, 0x4b7fffff); + nv_wo32(ctx, 0x0620, 0x00000080); + nv_wo32(ctx, 0x0624, 0x30201000); + nv_wo32(ctx, 0x0628, 0x70605040); + nv_wo32(ctx, 0x062c, 0xb0a09080); + nv_wo32(ctx, 0x0630, 0xf0e0d0c0); + nv_wo32(ctx, 0x0664, 0x00000001); + nv_wo32(ctx, 0x066c, 0x00004000); + nv_wo32(ctx, 0x0678, 0x00000001); + nv_wo32(ctx, 0x0680, 0x00040000); + nv_wo32(ctx, 0x0684, 0x00010000); + for (i = 0x1b04; i <= 0x2374; i += 16) { + nv_wo32(ctx, (i + 0), 0x10700ff9); + nv_wo32(ctx, (i + 4), 0x0436086c); + nv_wo32(ctx, (i + 8), 0x000c001b); + } + nv_wo32(ctx, 0x2704, 0x3f800000); + nv_wo32(ctx, 0x2718, 0x3f800000); + nv_wo32(ctx, 0x2744, 0x40000000); + nv_wo32(ctx, 0x2748, 0x3f800000); + nv_wo32(ctx, 0x274c, 0x3f000000); + nv_wo32(ctx, 0x2754, 0x40000000); + nv_wo32(ctx, 0x2758, 0x3f800000); + nv_wo32(ctx, 0x2760, 0xbf800000); + nv_wo32(ctx, 0x2768, 0xbf800000); + nv_wo32(ctx, 0x308c, 0x000fe000); + nv_wo32(ctx, 0x3108, 0x000003f8); + nv_wo32(ctx, 0x3468, 0x002fe000); + for (i = 0x3484; i <= 0x34a0; i += 4) + nv_wo32(ctx, i, 0x001c527c); +} + +static void +nv2a_graph_context_init(struct nouveau_gpuobj *ctx) +{ + int i; + + nv_wo32(ctx, 0x033c, 0xffff0000); + nv_wo32(ctx, 0x03a0, 0x0fff0000); + nv_wo32(ctx, 0x03a4, 0x0fff0000); + nv_wo32(ctx, 0x047c, 0x00000101); + nv_wo32(ctx, 0x0490, 0x00000111); + nv_wo32(ctx, 0x04a8, 0x44400000); + for (i = 0x04d4; i <= 0x04e0; i += 4) + nv_wo32(ctx, i, 0x00030303); + for (i = 0x04f4; i <= 0x0500; i += 4) + nv_wo32(ctx, i, 0x00080000); + for (i = 0x050c; i <= 0x0518; i += 4) + nv_wo32(ctx, i, 0x01012000); + for (i = 0x051c; i <= 0x0528; i += 4) + nv_wo32(ctx, i, 0x000105b8); + for (i = 0x052c; i <= 0x0538; i += 4) + nv_wo32(ctx, i, 0x00080008); + for (i = 0x055c; i <= 0x0598; i += 4) + nv_wo32(ctx, i, 0x07ff0000); + nv_wo32(ctx, 0x05a4, 0x4b7fffff); + nv_wo32(ctx, 0x05fc, 0x00000001); + nv_wo32(ctx, 0x0604, 0x00004000); + nv_wo32(ctx, 0x0610, 0x00000001); + nv_wo32(ctx, 0x0618, 0x00040000); + nv_wo32(ctx, 0x061c, 0x00010000); + for (i = 0x1a9c; i <= 0x22fc; i += 16) { /*XXX: check!! */ + nv_wo32(ctx, (i + 0), 0x10700ff9); + nv_wo32(ctx, (i + 4), 0x0436086c); + nv_wo32(ctx, (i + 8), 0x000c001b); + } + nv_wo32(ctx, 0x269c, 0x3f800000); + nv_wo32(ctx, 0x26b0, 0x3f800000); + nv_wo32(ctx, 0x26dc, 0x40000000); + nv_wo32(ctx, 0x26e0, 0x3f800000); + nv_wo32(ctx, 0x26e4, 0x3f000000); + nv_wo32(ctx, 0x26ec, 0x40000000); + nv_wo32(ctx, 0x26f0, 0x3f800000); + nv_wo32(ctx, 0x26f8, 0xbf800000); + nv_wo32(ctx, 0x2700, 0xbf800000); + nv_wo32(ctx, 0x3024, 0x000fe000); + nv_wo32(ctx, 0x30a0, 0x000003f8); + nv_wo32(ctx, 0x33fc, 0x002fe000); + for (i = 0x341c; i <= 0x3438; i += 4) + nv_wo32(ctx, i, 0x001c527c); +} + +static void +nv30_31_graph_context_init(struct nouveau_gpuobj *ctx) +{ + int i; + + nv_wo32(ctx, 0x0410, 0x00000101); + nv_wo32(ctx, 0x0424, 0x00000111); + nv_wo32(ctx, 0x0428, 0x00000060); + nv_wo32(ctx, 0x0444, 0x00000080); + nv_wo32(ctx, 0x0448, 0xffff0000); + nv_wo32(ctx, 0x044c, 0x00000001); + nv_wo32(ctx, 0x0460, 0x44400000); + nv_wo32(ctx, 0x048c, 0xffff0000); + for (i = 0x04e0; i < 0x04e8; i += 4) + nv_wo32(ctx, i, 0x0fff0000); + nv_wo32(ctx, 0x04ec, 0x00011100); + for (i = 0x0508; i < 0x0548; i += 4) + nv_wo32(ctx, i, 0x07ff0000); + nv_wo32(ctx, 0x0550, 0x4b7fffff); + nv_wo32(ctx, 0x058c, 0x00000080); + nv_wo32(ctx, 0x0590, 0x30201000); + nv_wo32(ctx, 0x0594, 0x70605040); + nv_wo32(ctx, 0x0598, 0xb8a89888); + nv_wo32(ctx, 0x059c, 0xf8e8d8c8); + nv_wo32(ctx, 0x05b0, 0xb0000000); + for (i = 0x0600; i < 0x0640; i += 4) + nv_wo32(ctx, i, 0x00010588); + for (i = 0x0640; i < 0x0680; i += 4) + nv_wo32(ctx, i, 0x00030303); + for (i = 0x06c0; i < 0x0700; i += 4) + nv_wo32(ctx, i, 0x0008aae4); + for (i = 0x0700; i < 0x0740; i += 4) + nv_wo32(ctx, i, 0x01012000); + for (i = 0x0740; i < 0x0780; i += 4) + nv_wo32(ctx, i, 0x00080008); + nv_wo32(ctx, 0x085c, 0x00040000); + nv_wo32(ctx, 0x0860, 0x00010000); + for (i = 0x0864; i < 0x0874; i += 4) + nv_wo32(ctx, i, 0x00040004); + for (i = 0x1f18; i <= 0x3088 ; i += 16) { + nv_wo32(ctx, i + 0, 0x10700ff9); + nv_wo32(ctx, i + 1, 0x0436086c); + nv_wo32(ctx, i + 2, 0x000c001b); + } + for (i = 0x30b8; i < 0x30c8; i += 4) + nv_wo32(ctx, i, 0x0000ffff); + nv_wo32(ctx, 0x344c, 0x3f800000); + nv_wo32(ctx, 0x3808, 0x3f800000); + nv_wo32(ctx, 0x381c, 0x3f800000); + nv_wo32(ctx, 0x3848, 0x40000000); + nv_wo32(ctx, 0x384c, 0x3f800000); + nv_wo32(ctx, 0x3850, 0x3f000000); + nv_wo32(ctx, 0x3858, 0x40000000); + nv_wo32(ctx, 0x385c, 0x3f800000); + nv_wo32(ctx, 0x3864, 0xbf800000); + nv_wo32(ctx, 0x386c, 0xbf800000); +} + +static void +nv34_graph_context_init(struct nouveau_gpuobj *ctx) +{ + int i; + + nv_wo32(ctx, 0x040c, 0x01000101); + nv_wo32(ctx, 0x0420, 0x00000111); + nv_wo32(ctx, 0x0424, 0x00000060); + nv_wo32(ctx, 0x0440, 0x00000080); + nv_wo32(ctx, 0x0444, 0xffff0000); + nv_wo32(ctx, 0x0448, 0x00000001); + nv_wo32(ctx, 0x045c, 0x44400000); + nv_wo32(ctx, 0x0480, 0xffff0000); + for (i = 0x04d4; i < 0x04dc; i += 4) + nv_wo32(ctx, i, 0x0fff0000); + nv_wo32(ctx, 0x04e0, 0x00011100); + for (i = 0x04fc; i < 0x053c; i += 4) + nv_wo32(ctx, i, 0x07ff0000); + nv_wo32(ctx, 0x0544, 0x4b7fffff); + nv_wo32(ctx, 0x057c, 0x00000080); + nv_wo32(ctx, 0x0580, 0x30201000); + nv_wo32(ctx, 0x0584, 0x70605040); + nv_wo32(ctx, 0x0588, 0xb8a89888); + nv_wo32(ctx, 0x058c, 0xf8e8d8c8); + nv_wo32(ctx, 0x05a0, 0xb0000000); + for (i = 0x05f0; i < 0x0630; i += 4) + nv_wo32(ctx, i, 0x00010588); + for (i = 0x0630; i < 0x0670; i += 4) + nv_wo32(ctx, i, 0x00030303); + for (i = 0x06b0; i < 0x06f0; i += 4) + nv_wo32(ctx, i, 0x0008aae4); + for (i = 0x06f0; i < 0x0730; i += 4) + nv_wo32(ctx, i, 0x01012000); + for (i = 0x0730; i < 0x0770; i += 4) + nv_wo32(ctx, i, 0x00080008); + nv_wo32(ctx, 0x0850, 0x00040000); + nv_wo32(ctx, 0x0854, 0x00010000); + for (i = 0x0858; i < 0x0868; i += 4) + nv_wo32(ctx, i, 0x00040004); + for (i = 0x15ac; i <= 0x271c ; i += 16) { + nv_wo32(ctx, i + 0, 0x10700ff9); + nv_wo32(ctx, i + 1, 0x0436086c); + nv_wo32(ctx, i + 2, 0x000c001b); + } + for (i = 0x274c; i < 0x275c; i += 4) + nv_wo32(ctx, i, 0x0000ffff); + nv_wo32(ctx, 0x2ae0, 0x3f800000); + nv_wo32(ctx, 0x2e9c, 0x3f800000); + nv_wo32(ctx, 0x2eb0, 0x3f800000); + nv_wo32(ctx, 0x2edc, 0x40000000); + nv_wo32(ctx, 0x2ee0, 0x3f800000); + nv_wo32(ctx, 0x2ee4, 0x3f000000); + nv_wo32(ctx, 0x2eec, 0x40000000); + nv_wo32(ctx, 0x2ef0, 0x3f800000); + nv_wo32(ctx, 0x2ef8, 0xbf800000); + nv_wo32(ctx, 0x2f00, 0xbf800000); +} + +static void +nv35_36_graph_context_init(struct nouveau_gpuobj *ctx) +{ + int i; + + nv_wo32(ctx, 0x040c, 0x00000101); + nv_wo32(ctx, 0x0420, 0x00000111); + nv_wo32(ctx, 0x0424, 0x00000060); + nv_wo32(ctx, 0x0440, 0x00000080); + nv_wo32(ctx, 0x0444, 0xffff0000); + nv_wo32(ctx, 0x0448, 0x00000001); + nv_wo32(ctx, 0x045c, 0x44400000); + nv_wo32(ctx, 0x0488, 0xffff0000); + for (i = 0x04dc; i < 0x04e4; i += 4) + nv_wo32(ctx, i, 0x0fff0000); + nv_wo32(ctx, 0x04e8, 0x00011100); + for (i = 0x0504; i < 0x0544; i += 4) + nv_wo32(ctx, i, 0x07ff0000); + nv_wo32(ctx, 0x054c, 0x4b7fffff); + nv_wo32(ctx, 0x0588, 0x00000080); + nv_wo32(ctx, 0x058c, 0x30201000); + nv_wo32(ctx, 0x0590, 0x70605040); + nv_wo32(ctx, 0x0594, 0xb8a89888); + nv_wo32(ctx, 0x0598, 0xf8e8d8c8); + nv_wo32(ctx, 0x05ac, 0xb0000000); + for (i = 0x0604; i < 0x0644; i += 4) + nv_wo32(ctx, i, 0x00010588); + for (i = 0x0644; i < 0x0684; i += 4) + nv_wo32(ctx, i, 0x00030303); + for (i = 0x06c4; i < 0x0704; i += 4) + nv_wo32(ctx, i, 0x0008aae4); + for (i = 0x0704; i < 0x0744; i += 4) + nv_wo32(ctx, i, 0x01012000); + for (i = 0x0744; i < 0x0784; i += 4) + nv_wo32(ctx, i, 0x00080008); + nv_wo32(ctx, 0x0860, 0x00040000); + nv_wo32(ctx, 0x0864, 0x00010000); + for (i = 0x0868; i < 0x0878; i += 4) + nv_wo32(ctx, i, 0x00040004); + for (i = 0x1f1c; i <= 0x308c ; i += 16) { + nv_wo32(ctx, i + 0, 0x10700ff9); + nv_wo32(ctx, i + 4, 0x0436086c); + nv_wo32(ctx, i + 8, 0x000c001b); + } + for (i = 0x30bc; i < 0x30cc; i += 4) + nv_wo32(ctx, i, 0x0000ffff); + nv_wo32(ctx, 0x3450, 0x3f800000); + nv_wo32(ctx, 0x380c, 0x3f800000); + nv_wo32(ctx, 0x3820, 0x3f800000); + nv_wo32(ctx, 0x384c, 0x40000000); + nv_wo32(ctx, 0x3850, 0x3f800000); + nv_wo32(ctx, 0x3854, 0x3f000000); + nv_wo32(ctx, 0x385c, 0x40000000); + nv_wo32(ctx, 0x3860, 0x3f800000); + nv_wo32(ctx, 0x3868, 0xbf800000); + nv_wo32(ctx, 0x3870, 0xbf800000); +} + +int +nv20_graph_context_new(struct nouveau_channel *chan, int engine) +{ + struct nv20_graph_engine *pgraph = nv_engine(chan->dev, engine); + struct nouveau_gpuobj *grctx = NULL; + struct drm_device *dev = chan->dev; + int ret; + + ret = nouveau_gpuobj_new(dev, NULL, pgraph->grctx_size, 16, + NVOBJ_FLAG_ZERO_ALLOC, &grctx); + if (ret) + return ret; + + /* Initialise default context values */ + pgraph->grctx_init(grctx); + + /* nv20: nv_wo32(dev, chan->ramin_grctx->gpuobj, 10, chan->id<<24); */ + /* CTX_USER */ + nv_wo32(grctx, pgraph->grctx_user, (chan->id << 24) | 0x1); + + nv_wo32(pgraph->ctxtab, chan->id * 4, grctx->pinst >> 4); + chan->engctx[engine] = grctx; + return 0; +} + +void +nv20_graph_context_del(struct nouveau_channel *chan, int engine) +{ + struct nv20_graph_engine *pgraph = nv_engine(chan->dev, engine); + struct nouveau_gpuobj *grctx = chan->engctx[engine]; + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + unsigned long flags; + + spin_lock_irqsave(&dev_priv->context_switch_lock, flags); + nv04_graph_fifo_access(dev, false); + + /* Unload the context if it's the currently active one */ + if (nv10_graph_channel(dev) == chan) + nv20_graph_unload_context(dev); + + nv04_graph_fifo_access(dev, true); + spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); + + /* Free the context resources */ + nv_wo32(pgraph->ctxtab, chan->id * 4, 0); + + nouveau_gpuobj_ref(NULL, &grctx); + chan->engctx[engine] = NULL; +} + +static void +nv20_graph_set_tile_region(struct drm_device *dev, int i) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_tile_reg *tile = &dev_priv->tile.reg[i]; + + nv_wr32(dev, NV20_PGRAPH_TLIMIT(i), tile->limit); + nv_wr32(dev, NV20_PGRAPH_TSIZE(i), tile->pitch); + nv_wr32(dev, NV20_PGRAPH_TILE(i), tile->addr); + + nv_wr32(dev, NV10_PGRAPH_RDI_INDEX, 0x00EA0030 + 4 * i); + nv_wr32(dev, NV10_PGRAPH_RDI_DATA, tile->limit); + nv_wr32(dev, NV10_PGRAPH_RDI_INDEX, 0x00EA0050 + 4 * i); + nv_wr32(dev, NV10_PGRAPH_RDI_DATA, tile->pitch); + nv_wr32(dev, NV10_PGRAPH_RDI_INDEX, 0x00EA0010 + 4 * i); + nv_wr32(dev, NV10_PGRAPH_RDI_DATA, tile->addr); + + if (dev_priv->card_type == NV_20) { + nv_wr32(dev, NV20_PGRAPH_ZCOMP(i), tile->zcomp); + nv_wr32(dev, NV10_PGRAPH_RDI_INDEX, 0x00ea0090 + 4 * i); + nv_wr32(dev, NV10_PGRAPH_RDI_DATA, tile->zcomp); + } +} + +int +nv20_graph_init(struct drm_device *dev, int engine) +{ + struct nv20_graph_engine *pgraph = nv_engine(dev, engine); + struct drm_nouveau_private *dev_priv = dev->dev_private; + uint32_t tmp, vramsz; + int i; + + nv_wr32(dev, NV03_PMC_ENABLE, + nv_rd32(dev, NV03_PMC_ENABLE) & ~NV_PMC_ENABLE_PGRAPH); + nv_wr32(dev, NV03_PMC_ENABLE, + nv_rd32(dev, NV03_PMC_ENABLE) | NV_PMC_ENABLE_PGRAPH); + + nv_wr32(dev, NV20_PGRAPH_CHANNEL_CTX_TABLE, pgraph->ctxtab->pinst >> 4); + + nv20_graph_rdi(dev); + + nv_wr32(dev, NV03_PGRAPH_INTR , 0xFFFFFFFF); + nv_wr32(dev, NV03_PGRAPH_INTR_EN, 0xFFFFFFFF); + + nv_wr32(dev, NV04_PGRAPH_DEBUG_0, 0xFFFFFFFF); + nv_wr32(dev, NV04_PGRAPH_DEBUG_0, 0x00000000); + nv_wr32(dev, NV04_PGRAPH_DEBUG_1, 0x00118700); + nv_wr32(dev, NV04_PGRAPH_DEBUG_3, 0xF3CE0475); /* 0x4 = auto ctx switch */ + nv_wr32(dev, NV10_PGRAPH_DEBUG_4, 0x00000000); + nv_wr32(dev, 0x40009C , 0x00000040); + + if (dev_priv->chipset >= 0x25) { + nv_wr32(dev, 0x400890, 0x00a8cfff); + nv_wr32(dev, 0x400610, 0x304B1FB6); + nv_wr32(dev, 0x400B80, 0x1cbd3883); + nv_wr32(dev, 0x400B84, 0x44000000); + nv_wr32(dev, 0x400098, 0x40000080); + nv_wr32(dev, 0x400B88, 0x000000ff); + + } else { + nv_wr32(dev, 0x400880, 0x0008c7df); + nv_wr32(dev, 0x400094, 0x00000005); + nv_wr32(dev, 0x400B80, 0x45eae20e); + nv_wr32(dev, 0x400B84, 0x24000000); + nv_wr32(dev, 0x400098, 0x00000040); + nv_wr32(dev, NV10_PGRAPH_RDI_INDEX, 0x00E00038); + nv_wr32(dev, NV10_PGRAPH_RDI_DATA , 0x00000030); + nv_wr32(dev, NV10_PGRAPH_RDI_INDEX, 0x00E10038); + nv_wr32(dev, NV10_PGRAPH_RDI_DATA , 0x00000030); + } + + /* Turn all the tiling regions off. */ + for (i = 0; i < NV10_PFB_TILE__SIZE; i++) + nv20_graph_set_tile_region(dev, i); + + nv_wr32(dev, 0x4009a0, nv_rd32(dev, 0x100324)); + nv_wr32(dev, NV10_PGRAPH_RDI_INDEX, 0x00EA000C); + nv_wr32(dev, NV10_PGRAPH_RDI_DATA, nv_rd32(dev, 0x100324)); + + nv_wr32(dev, NV10_PGRAPH_CTX_CONTROL, 0x10000100); + nv_wr32(dev, NV10_PGRAPH_STATE , 0xFFFFFFFF); + + tmp = nv_rd32(dev, NV10_PGRAPH_SURFACE) & 0x0007ff00; + nv_wr32(dev, NV10_PGRAPH_SURFACE, tmp); + tmp = nv_rd32(dev, NV10_PGRAPH_SURFACE) | 0x00020100; + nv_wr32(dev, NV10_PGRAPH_SURFACE, tmp); + + /* begin RAM config */ + vramsz = pci_resource_len(dev->pdev, 0) - 1; + nv_wr32(dev, 0x4009A4, nv_rd32(dev, NV04_PFB_CFG0)); + nv_wr32(dev, 0x4009A8, nv_rd32(dev, NV04_PFB_CFG1)); + nv_wr32(dev, NV10_PGRAPH_RDI_INDEX, 0x00EA0000); + nv_wr32(dev, NV10_PGRAPH_RDI_DATA , nv_rd32(dev, NV04_PFB_CFG0)); + nv_wr32(dev, NV10_PGRAPH_RDI_INDEX, 0x00EA0004); + nv_wr32(dev, NV10_PGRAPH_RDI_DATA , nv_rd32(dev, NV04_PFB_CFG1)); + nv_wr32(dev, 0x400820, 0); + nv_wr32(dev, 0x400824, 0); + nv_wr32(dev, 0x400864, vramsz - 1); + nv_wr32(dev, 0x400868, vramsz - 1); + + /* interesting.. the below overwrites some of the tile setup above.. */ + nv_wr32(dev, 0x400B20, 0x00000000); + nv_wr32(dev, 0x400B04, 0xFFFFFFFF); + + nv_wr32(dev, NV03_PGRAPH_ABS_UCLIP_XMIN, 0); + nv_wr32(dev, NV03_PGRAPH_ABS_UCLIP_YMIN, 0); + nv_wr32(dev, NV03_PGRAPH_ABS_UCLIP_XMAX, 0x7fff); + nv_wr32(dev, NV03_PGRAPH_ABS_UCLIP_YMAX, 0x7fff); + + return 0; +} + +int +nv30_graph_init(struct drm_device *dev, int engine) +{ + struct nv20_graph_engine *pgraph = nv_engine(dev, engine); + struct drm_nouveau_private *dev_priv = dev->dev_private; + int i; + + nv_wr32(dev, NV03_PMC_ENABLE, + nv_rd32(dev, NV03_PMC_ENABLE) & ~NV_PMC_ENABLE_PGRAPH); + nv_wr32(dev, NV03_PMC_ENABLE, + nv_rd32(dev, NV03_PMC_ENABLE) | NV_PMC_ENABLE_PGRAPH); + + nv_wr32(dev, NV20_PGRAPH_CHANNEL_CTX_TABLE, pgraph->ctxtab->pinst >> 4); + + nv_wr32(dev, NV03_PGRAPH_INTR , 0xFFFFFFFF); + nv_wr32(dev, NV03_PGRAPH_INTR_EN, 0xFFFFFFFF); + + nv_wr32(dev, NV04_PGRAPH_DEBUG_0, 0xFFFFFFFF); + nv_wr32(dev, NV04_PGRAPH_DEBUG_0, 0x00000000); + nv_wr32(dev, NV04_PGRAPH_DEBUG_1, 0x401287c0); + nv_wr32(dev, 0x400890, 0x01b463ff); + nv_wr32(dev, NV04_PGRAPH_DEBUG_3, 0xf2de0475); + nv_wr32(dev, NV10_PGRAPH_DEBUG_4, 0x00008000); + nv_wr32(dev, NV04_PGRAPH_LIMIT_VIOL_PIX, 0xf04bdff6); + nv_wr32(dev, 0x400B80, 0x1003d888); + nv_wr32(dev, 0x400B84, 0x0c000000); + nv_wr32(dev, 0x400098, 0x00000000); + nv_wr32(dev, 0x40009C, 0x0005ad00); + nv_wr32(dev, 0x400B88, 0x62ff00ff); /* suspiciously like PGRAPH_DEBUG_2 */ + nv_wr32(dev, 0x4000a0, 0x00000000); + nv_wr32(dev, 0x4000a4, 0x00000008); + nv_wr32(dev, 0x4008a8, 0xb784a400); + nv_wr32(dev, 0x400ba0, 0x002f8685); + nv_wr32(dev, 0x400ba4, 0x00231f3f); + nv_wr32(dev, 0x4008a4, 0x40000020); + + if (dev_priv->chipset == 0x34) { + nv_wr32(dev, NV10_PGRAPH_RDI_INDEX, 0x00EA0004); + nv_wr32(dev, NV10_PGRAPH_RDI_DATA , 0x00200201); + nv_wr32(dev, NV10_PGRAPH_RDI_INDEX, 0x00EA0008); + nv_wr32(dev, NV10_PGRAPH_RDI_DATA , 0x00000008); + nv_wr32(dev, NV10_PGRAPH_RDI_INDEX, 0x00EA0000); + nv_wr32(dev, NV10_PGRAPH_RDI_DATA , 0x00000032); + nv_wr32(dev, NV10_PGRAPH_RDI_INDEX, 0x00E00004); + nv_wr32(dev, NV10_PGRAPH_RDI_DATA , 0x00000002); + } + + nv_wr32(dev, 0x4000c0, 0x00000016); + + /* Turn all the tiling regions off. */ + for (i = 0; i < NV10_PFB_TILE__SIZE; i++) + nv20_graph_set_tile_region(dev, i); + + nv_wr32(dev, NV10_PGRAPH_CTX_CONTROL, 0x10000100); + nv_wr32(dev, NV10_PGRAPH_STATE , 0xFFFFFFFF); + nv_wr32(dev, 0x0040075c , 0x00000001); + + /* begin RAM config */ + /* vramsz = pci_resource_len(dev->pdev, 0) - 1; */ + nv_wr32(dev, 0x4009A4, nv_rd32(dev, NV04_PFB_CFG0)); + nv_wr32(dev, 0x4009A8, nv_rd32(dev, NV04_PFB_CFG1)); + if (dev_priv->chipset != 0x34) { + nv_wr32(dev, 0x400750, 0x00EA0000); + nv_wr32(dev, 0x400754, nv_rd32(dev, NV04_PFB_CFG0)); + nv_wr32(dev, 0x400750, 0x00EA0004); + nv_wr32(dev, 0x400754, nv_rd32(dev, NV04_PFB_CFG1)); + } + + return 0; +} + +int +nv20_graph_fini(struct drm_device *dev, int engine) +{ + nv20_graph_unload_context(dev); + nv_wr32(dev, NV03_PGRAPH_INTR_EN, 0x00000000); + return 0; +} + +static void +nv20_graph_isr(struct drm_device *dev) +{ + u32 stat; + + while ((stat = nv_rd32(dev, NV03_PGRAPH_INTR))) { + u32 nsource = nv_rd32(dev, NV03_PGRAPH_NSOURCE); + u32 nstatus = nv_rd32(dev, NV03_PGRAPH_NSTATUS); + u32 addr = nv_rd32(dev, NV04_PGRAPH_TRAPPED_ADDR); + u32 chid = (addr & 0x01f00000) >> 20; + u32 subc = (addr & 0x00070000) >> 16; + u32 mthd = (addr & 0x00001ffc); + u32 data = nv_rd32(dev, NV04_PGRAPH_TRAPPED_DATA); + u32 class = nv_rd32(dev, 0x400160 + subc * 4) & 0xfff; + u32 show = stat; + + if (stat & NV_PGRAPH_INTR_ERROR) { + if (nsource & NV03_PGRAPH_NSOURCE_ILLEGAL_MTHD) { + if (!nouveau_gpuobj_mthd_call2(dev, chid, class, mthd, data)) + show &= ~NV_PGRAPH_INTR_ERROR; + } + } + + nv_wr32(dev, NV03_PGRAPH_INTR, stat); + nv_wr32(dev, NV04_PGRAPH_FIFO, 0x00000001); + + if (show && nouveau_ratelimit()) { + NV_INFO(dev, "PGRAPH -"); + nouveau_bitfield_print(nv10_graph_intr, show); + printk(" nsource:"); + nouveau_bitfield_print(nv04_graph_nsource, nsource); + printk(" nstatus:"); + nouveau_bitfield_print(nv10_graph_nstatus, nstatus); + printk("\n"); + NV_INFO(dev, "PGRAPH - ch %d/%d class 0x%04x " + "mthd 0x%04x data 0x%08x\n", + chid, subc, class, mthd, data); + } + } +} + +static void +nv20_graph_destroy(struct drm_device *dev, int engine) +{ + struct nv20_graph_engine *pgraph = nv_engine(dev, engine); + + nouveau_irq_unregister(dev, 12); + nouveau_gpuobj_ref(NULL, &pgraph->ctxtab); + + NVOBJ_ENGINE_DEL(dev, GR); + kfree(pgraph); +} + +int +nv20_graph_create(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nv20_graph_engine *pgraph; + int ret; + + pgraph = kzalloc(sizeof(*pgraph), GFP_KERNEL); + if (!pgraph) + return -ENOMEM; + + pgraph->base.destroy = nv20_graph_destroy; + pgraph->base.fini = nv20_graph_fini; + pgraph->base.context_new = nv20_graph_context_new; + pgraph->base.context_del = nv20_graph_context_del; + pgraph->base.object_new = nv04_graph_object_new; + pgraph->base.set_tile_region = nv20_graph_set_tile_region; + + pgraph->grctx_user = 0x0028; + if (dev_priv->card_type == NV_20) { + pgraph->base.init = nv20_graph_init; + switch (dev_priv->chipset) { + case 0x20: + pgraph->grctx_init = nv20_graph_context_init; + pgraph->grctx_size = NV20_GRCTX_SIZE; + pgraph->grctx_user = 0x0000; + break; + case 0x25: + case 0x28: + pgraph->grctx_init = nv25_graph_context_init; + pgraph->grctx_size = NV25_GRCTX_SIZE; + break; + case 0x2a: + pgraph->grctx_init = nv2a_graph_context_init; + pgraph->grctx_size = NV2A_GRCTX_SIZE; + pgraph->grctx_user = 0x0000; + break; + default: + NV_ERROR(dev, "PGRAPH: unknown chipset\n"); + return 0; + } + } else { + pgraph->base.init = nv30_graph_init; + switch (dev_priv->chipset) { + case 0x30: + case 0x31: + pgraph->grctx_init = nv30_31_graph_context_init; + pgraph->grctx_size = NV30_31_GRCTX_SIZE; + break; + case 0x34: + pgraph->grctx_init = nv34_graph_context_init; + pgraph->grctx_size = NV34_GRCTX_SIZE; + break; + case 0x35: + case 0x36: + pgraph->grctx_init = nv35_36_graph_context_init; + pgraph->grctx_size = NV35_36_GRCTX_SIZE; + break; + default: + NV_ERROR(dev, "PGRAPH: unknown chipset\n"); + return 0; + } + } + + /* Create Context Pointer Table */ + ret = nouveau_gpuobj_new(dev, NULL, 32 * 4, 16, NVOBJ_FLAG_ZERO_ALLOC, + &pgraph->ctxtab); + if (ret) { + kfree(pgraph); + return ret; + } + + NVOBJ_ENGINE_ADD(dev, GR, &pgraph->base); + nouveau_irq_register(dev, 12, nv20_graph_isr); + + /* nvsw */ + NVOBJ_CLASS(dev, 0x506e, SW); + NVOBJ_MTHD (dev, 0x506e, 0x0500, nv04_graph_mthd_page_flip); + + NVOBJ_CLASS(dev, 0x0030, GR); /* null */ + NVOBJ_CLASS(dev, 0x0039, GR); /* m2mf */ + NVOBJ_CLASS(dev, 0x004a, GR); /* gdirect */ + NVOBJ_CLASS(dev, 0x009f, GR); /* imageblit (nv12) */ + NVOBJ_CLASS(dev, 0x008a, GR); /* ifc */ + NVOBJ_CLASS(dev, 0x0089, GR); /* sifm */ + NVOBJ_CLASS(dev, 0x0062, GR); /* surf2d */ + NVOBJ_CLASS(dev, 0x0043, GR); /* rop */ + NVOBJ_CLASS(dev, 0x0012, GR); /* beta1 */ + NVOBJ_CLASS(dev, 0x0072, GR); /* beta4 */ + NVOBJ_CLASS(dev, 0x0019, GR); /* cliprect */ + NVOBJ_CLASS(dev, 0x0044, GR); /* pattern */ + if (dev_priv->card_type == NV_20) { + NVOBJ_CLASS(dev, 0x009e, GR); /* swzsurf */ + NVOBJ_CLASS(dev, 0x0096, GR); /* celcius */ + + /* kelvin */ + if (dev_priv->chipset < 0x25) + NVOBJ_CLASS(dev, 0x0097, GR); + else + NVOBJ_CLASS(dev, 0x0597, GR); + } else { + NVOBJ_CLASS(dev, 0x038a, GR); /* ifc (nv30) */ + NVOBJ_CLASS(dev, 0x0389, GR); /* sifm (nv30) */ + NVOBJ_CLASS(dev, 0x0362, GR); /* surf2d (nv30) */ + NVOBJ_CLASS(dev, 0x039e, GR); /* swzsurf */ + + /* rankine */ + if (0x00000003 & (1 << (dev_priv->chipset & 0x0f))) + NVOBJ_CLASS(dev, 0x0397, GR); + else + if (0x00000010 & (1 << (dev_priv->chipset & 0x0f))) + NVOBJ_CLASS(dev, 0x0697, GR); + else + if (0x000001e0 & (1 << (dev_priv->chipset & 0x0f))) + NVOBJ_CLASS(dev, 0x0497, GR); + } + + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nv30_fb.c b/drivers/gpu/drm/nouveau/nv30_fb.c new file mode 100644 index 00000000..e0135f0e --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv30_fb.c @@ -0,0 +1,116 @@ +/* + * Copyright (C) 2010 Francisco Jerez. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "drmP.h" +#include "drm.h" +#include "nouveau_drv.h" +#include "nouveau_drm.h" + +void +nv30_fb_init_tile_region(struct drm_device *dev, int i, uint32_t addr, + uint32_t size, uint32_t pitch, uint32_t flags) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_tile_reg *tile = &dev_priv->tile.reg[i]; + + tile->addr = addr | 1; + tile->limit = max(1u, addr + size) - 1; + tile->pitch = pitch; +} + +void +nv30_fb_free_tile_region(struct drm_device *dev, int i) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_tile_reg *tile = &dev_priv->tile.reg[i]; + + tile->addr = tile->limit = tile->pitch = 0; +} + +static int +calc_bias(struct drm_device *dev, int k, int i, int j) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + int b = (dev_priv->chipset > 0x30 ? + nv_rd32(dev, 0x122c + 0x10 * k + 0x4 * j) >> (4 * (i ^ 1)) : + 0) & 0xf; + + return 2 * (b & 0x8 ? b - 0x10 : b); +} + +static int +calc_ref(struct drm_device *dev, int l, int k, int i) +{ + int j, x = 0; + + for (j = 0; j < 4; j++) { + int m = (l >> (8 * i) & 0xff) + calc_bias(dev, k, i, j); + + x |= (0x80 | clamp(m, 0, 0x1f)) << (8 * j); + } + + return x; +} + +int +nv30_fb_init(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fb_engine *pfb = &dev_priv->engine.fb; + int i, j; + + pfb->num_tiles = NV10_PFB_TILE__SIZE; + + /* Turn all the tiling regions off. */ + for (i = 0; i < pfb->num_tiles; i++) + pfb->set_tile_region(dev, i); + + /* Init the memory timing regs at 0x10037c/0x1003ac */ + if (dev_priv->chipset == 0x30 || + dev_priv->chipset == 0x31 || + dev_priv->chipset == 0x35) { + /* Related to ROP count */ + int n = (dev_priv->chipset == 0x31 ? 2 : 4); + int l = nv_rd32(dev, 0x1003d0); + + for (i = 0; i < n; i++) { + for (j = 0; j < 3; j++) + nv_wr32(dev, 0x10037c + 0xc * i + 0x4 * j, + calc_ref(dev, l, 0, j)); + + for (j = 0; j < 2; j++) + nv_wr32(dev, 0x1003ac + 0x8 * i + 0x4 * j, + calc_ref(dev, l, 1, j)); + } + } + + return 0; +} + +void +nv30_fb_takedown(struct drm_device *dev) +{ +} diff --git a/drivers/gpu/drm/nouveau/nv40_fb.c b/drivers/gpu/drm/nouveau/nv40_fb.c new file mode 100644 index 00000000..f0ac2a76 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv40_fb.c @@ -0,0 +1,118 @@ +#include "drmP.h" +#include "drm.h" +#include "nouveau_drv.h" +#include "nouveau_drm.h" + +void +nv40_fb_set_tile_region(struct drm_device *dev, int i) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_tile_reg *tile = &dev_priv->tile.reg[i]; + + switch (dev_priv->chipset) { + case 0x40: + nv_wr32(dev, NV10_PFB_TLIMIT(i), tile->limit); + nv_wr32(dev, NV10_PFB_TSIZE(i), tile->pitch); + nv_wr32(dev, NV10_PFB_TILE(i), tile->addr); + break; + + default: + nv_wr32(dev, NV40_PFB_TLIMIT(i), tile->limit); + nv_wr32(dev, NV40_PFB_TSIZE(i), tile->pitch); + nv_wr32(dev, NV40_PFB_TILE(i), tile->addr); + break; + } +} + +static void +nv40_fb_init_gart(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *gart = dev_priv->gart_info.sg_ctxdma; + + if (dev_priv->gart_info.type != NOUVEAU_GART_HW) { + nv_wr32(dev, 0x100800, 0x00000001); + return; + } + + nv_wr32(dev, 0x100800, gart->pinst | 0x00000002); + nv_mask(dev, 0x10008c, 0x00000100, 0x00000100); + nv_wr32(dev, 0x100820, 0x00000000); +} + +static void +nv44_fb_init_gart(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *gart = dev_priv->gart_info.sg_ctxdma; + u32 vinst; + + if (dev_priv->gart_info.type != NOUVEAU_GART_HW) { + nv_wr32(dev, 0x100850, 0x80000000); + nv_wr32(dev, 0x100800, 0x00000001); + return; + } + + /* calculate vram address of this PRAMIN block, object + * must be allocated on 512KiB alignment, and not exceed + * a total size of 512KiB for this to work correctly + */ + vinst = nv_rd32(dev, 0x10020c); + vinst -= ((gart->pinst >> 19) + 1) << 19; + + nv_wr32(dev, 0x100850, 0x80000000); + nv_wr32(dev, 0x100818, dev_priv->gart_info.dummy.addr); + + nv_wr32(dev, 0x100804, dev_priv->gart_info.aper_size); + nv_wr32(dev, 0x100850, 0x00008000); + nv_mask(dev, 0x10008c, 0x00000200, 0x00000200); + nv_wr32(dev, 0x100820, 0x00000000); + nv_wr32(dev, 0x10082c, 0x00000001); + nv_wr32(dev, 0x100800, vinst | 0x00000010); +} + +int +nv40_fb_init(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fb_engine *pfb = &dev_priv->engine.fb; + uint32_t tmp; + int i; + + if (dev_priv->chipset != 0x40 && dev_priv->chipset != 0x45) { + if (nv44_graph_class(dev)) + nv44_fb_init_gart(dev); + else + nv40_fb_init_gart(dev); + } + + switch (dev_priv->chipset) { + case 0x40: + case 0x45: + tmp = nv_rd32(dev, NV10_PFB_CLOSE_PAGE2); + nv_wr32(dev, NV10_PFB_CLOSE_PAGE2, tmp & ~(1 << 15)); + pfb->num_tiles = NV10_PFB_TILE__SIZE; + break; + case 0x46: /* G72 */ + case 0x47: /* G70 */ + case 0x49: /* G71 */ + case 0x4b: /* G73 */ + case 0x4c: /* C51 (G7X version) */ + pfb->num_tiles = NV40_PFB_TILE__SIZE_1; + break; + default: + pfb->num_tiles = NV40_PFB_TILE__SIZE_0; + break; + } + + /* Turn all the tiling regions off. */ + for (i = 0; i < pfb->num_tiles; i++) + pfb->set_tile_region(dev, i); + + return 0; +} + +void +nv40_fb_takedown(struct drm_device *dev) +{ +} diff --git a/drivers/gpu/drm/nouveau/nv40_fifo.c b/drivers/gpu/drm/nouveau/nv40_fifo.c new file mode 100644 index 00000000..68cb2d99 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv40_fifo.c @@ -0,0 +1,307 @@ +/* + * Copyright (C) 2007 Ben Skeggs. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "drmP.h" +#include "nouveau_drv.h" +#include "nouveau_drm.h" +#include "nouveau_ramht.h" + +#define NV40_RAMFC(c) (dev_priv->ramfc->pinst + ((c) * NV40_RAMFC__SIZE)) +#define NV40_RAMFC__SIZE 128 + +int +nv40_fifo_create_context(struct nouveau_channel *chan) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + uint32_t fc = NV40_RAMFC(chan->id); + unsigned long flags; + int ret; + + ret = nouveau_gpuobj_new_fake(dev, NV40_RAMFC(chan->id), ~0, + NV40_RAMFC__SIZE, NVOBJ_FLAG_ZERO_ALLOC | + NVOBJ_FLAG_ZERO_FREE, &chan->ramfc); + if (ret) + return ret; + + chan->user = ioremap(pci_resource_start(dev->pdev, 0) + + NV40_USER(chan->id), PAGE_SIZE); + if (!chan->user) + return -ENOMEM; + + spin_lock_irqsave(&dev_priv->context_switch_lock, flags); + + nv_wi32(dev, fc + 0, chan->pushbuf_base); + nv_wi32(dev, fc + 4, chan->pushbuf_base); + nv_wi32(dev, fc + 12, chan->pushbuf->pinst >> 4); + nv_wi32(dev, fc + 24, NV_PFIFO_CACHE1_DMA_FETCH_TRIG_128_BYTES | + NV_PFIFO_CACHE1_DMA_FETCH_SIZE_128_BYTES | + NV_PFIFO_CACHE1_DMA_FETCH_MAX_REQS_8 | +#ifdef __BIG_ENDIAN + NV_PFIFO_CACHE1_BIG_ENDIAN | +#endif + 0x30000000 /* no idea.. */); + nv_wi32(dev, fc + 60, 0x0001FFFF); + + /* enable the fifo dma operation */ + nv_wr32(dev, NV04_PFIFO_MODE, + nv_rd32(dev, NV04_PFIFO_MODE) | (1 << chan->id)); + + spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); + return 0; +} + +static void +nv40_fifo_do_load_context(struct drm_device *dev, int chid) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + uint32_t fc = NV40_RAMFC(chid), tmp, tmp2; + + nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_PUT, nv_ri32(dev, fc + 0)); + nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_GET, nv_ri32(dev, fc + 4)); + nv_wr32(dev, NV10_PFIFO_CACHE1_REF_CNT, nv_ri32(dev, fc + 8)); + nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_INSTANCE, nv_ri32(dev, fc + 12)); + nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_DCOUNT, nv_ri32(dev, fc + 16)); + nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_STATE, nv_ri32(dev, fc + 20)); + + /* No idea what 0x2058 is.. */ + tmp = nv_ri32(dev, fc + 24); + tmp2 = nv_rd32(dev, 0x2058) & 0xFFF; + tmp2 |= (tmp & 0x30000000); + nv_wr32(dev, 0x2058, tmp2); + tmp &= ~0x30000000; + nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_FETCH, tmp); + + nv_wr32(dev, NV04_PFIFO_CACHE1_ENGINE, nv_ri32(dev, fc + 28)); + nv_wr32(dev, NV04_PFIFO_CACHE1_PULL1, nv_ri32(dev, fc + 32)); + nv_wr32(dev, NV10_PFIFO_CACHE1_ACQUIRE_VALUE, nv_ri32(dev, fc + 36)); + tmp = nv_ri32(dev, fc + 40); + nv_wr32(dev, NV10_PFIFO_CACHE1_ACQUIRE_TIMESTAMP, tmp); + nv_wr32(dev, NV10_PFIFO_CACHE1_ACQUIRE_TIMEOUT, nv_ri32(dev, fc + 44)); + nv_wr32(dev, NV10_PFIFO_CACHE1_SEMAPHORE, nv_ri32(dev, fc + 48)); + nv_wr32(dev, NV10_PFIFO_CACHE1_DMA_SUBROUTINE, nv_ri32(dev, fc + 52)); + nv_wr32(dev, NV40_PFIFO_GRCTX_INSTANCE, nv_ri32(dev, fc + 56)); + + /* Don't clobber the TIMEOUT_ENABLED flag when restoring from RAMFC */ + tmp = nv_rd32(dev, NV04_PFIFO_DMA_TIMESLICE) & ~0x1FFFF; + tmp |= nv_ri32(dev, fc + 60) & 0x1FFFF; + nv_wr32(dev, NV04_PFIFO_DMA_TIMESLICE, tmp); + + nv_wr32(dev, 0x32e4, nv_ri32(dev, fc + 64)); + /* NVIDIA does this next line twice... */ + nv_wr32(dev, 0x32e8, nv_ri32(dev, fc + 68)); + nv_wr32(dev, 0x2088, nv_ri32(dev, fc + 76)); + nv_wr32(dev, 0x3300, nv_ri32(dev, fc + 80)); + nv_wr32(dev, 0x330c, nv_ri32(dev, fc + 84)); + + nv_wr32(dev, NV03_PFIFO_CACHE1_GET, 0); + nv_wr32(dev, NV03_PFIFO_CACHE1_PUT, 0); +} + +int +nv40_fifo_load_context(struct nouveau_channel *chan) +{ + struct drm_device *dev = chan->dev; + uint32_t tmp; + + nv40_fifo_do_load_context(dev, chan->id); + + /* Set channel active, and in DMA mode */ + nv_wr32(dev, NV03_PFIFO_CACHE1_PUSH1, + NV40_PFIFO_CACHE1_PUSH1_DMA | chan->id); + nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_PUSH, 1); + + /* Reset DMA_CTL_AT_INFO to INVALID */ + tmp = nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_CTL) & ~(1 << 31); + nv_wr32(dev, NV04_PFIFO_CACHE1_DMA_CTL, tmp); + + return 0; +} + +int +nv40_fifo_unload_context(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; + uint32_t fc, tmp; + int chid; + + chid = pfifo->channel_id(dev); + if (chid < 0 || chid >= dev_priv->engine.fifo.channels) + return 0; + fc = NV40_RAMFC(chid); + + nv_wi32(dev, fc + 0, nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_PUT)); + nv_wi32(dev, fc + 4, nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_GET)); + nv_wi32(dev, fc + 8, nv_rd32(dev, NV10_PFIFO_CACHE1_REF_CNT)); + nv_wi32(dev, fc + 12, nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_INSTANCE)); + nv_wi32(dev, fc + 16, nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_DCOUNT)); + nv_wi32(dev, fc + 20, nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_STATE)); + tmp = nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_FETCH); + tmp |= nv_rd32(dev, 0x2058) & 0x30000000; + nv_wi32(dev, fc + 24, tmp); + nv_wi32(dev, fc + 28, nv_rd32(dev, NV04_PFIFO_CACHE1_ENGINE)); + nv_wi32(dev, fc + 32, nv_rd32(dev, NV04_PFIFO_CACHE1_PULL1)); + nv_wi32(dev, fc + 36, nv_rd32(dev, NV10_PFIFO_CACHE1_ACQUIRE_VALUE)); + tmp = nv_rd32(dev, NV10_PFIFO_CACHE1_ACQUIRE_TIMESTAMP); + nv_wi32(dev, fc + 40, tmp); + nv_wi32(dev, fc + 44, nv_rd32(dev, NV10_PFIFO_CACHE1_ACQUIRE_TIMEOUT)); + nv_wi32(dev, fc + 48, nv_rd32(dev, NV10_PFIFO_CACHE1_SEMAPHORE)); + /* NVIDIA read 0x3228 first, then write DMA_GET here.. maybe something + * more involved depending on the value of 0x3228? + */ + nv_wi32(dev, fc + 52, nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_GET)); + nv_wi32(dev, fc + 56, nv_rd32(dev, NV40_PFIFO_GRCTX_INSTANCE)); + nv_wi32(dev, fc + 60, nv_rd32(dev, NV04_PFIFO_DMA_TIMESLICE) & 0x1ffff); + /* No idea what the below is for exactly, ripped from a mmio-trace */ + nv_wi32(dev, fc + 64, nv_rd32(dev, NV40_PFIFO_UNK32E4)); + /* NVIDIA do this next line twice.. bug? */ + nv_wi32(dev, fc + 68, nv_rd32(dev, 0x32e8)); + nv_wi32(dev, fc + 76, nv_rd32(dev, 0x2088)); + nv_wi32(dev, fc + 80, nv_rd32(dev, 0x3300)); +#if 0 /* no real idea which is PUT/GET in UNK_48.. */ + tmp = nv_rd32(dev, NV04_PFIFO_CACHE1_GET); + tmp |= (nv_rd32(dev, NV04_PFIFO_CACHE1_PUT) << 16); + nv_wi32(dev, fc + 72, tmp); +#endif + nv_wi32(dev, fc + 84, nv_rd32(dev, 0x330c)); + + nv40_fifo_do_load_context(dev, pfifo->channels - 1); + nv_wr32(dev, NV03_PFIFO_CACHE1_PUSH1, + NV40_PFIFO_CACHE1_PUSH1_DMA | (pfifo->channels - 1)); + return 0; +} + +static void +nv40_fifo_init_reset(struct drm_device *dev) +{ + int i; + + nv_wr32(dev, NV03_PMC_ENABLE, + nv_rd32(dev, NV03_PMC_ENABLE) & ~NV_PMC_ENABLE_PFIFO); + nv_wr32(dev, NV03_PMC_ENABLE, + nv_rd32(dev, NV03_PMC_ENABLE) | NV_PMC_ENABLE_PFIFO); + + nv_wr32(dev, 0x003224, 0x000f0078); + nv_wr32(dev, 0x003210, 0x00000000); + nv_wr32(dev, 0x003270, 0x00000000); + nv_wr32(dev, 0x003240, 0x00000000); + nv_wr32(dev, 0x003244, 0x00000000); + nv_wr32(dev, 0x003258, 0x00000000); + nv_wr32(dev, 0x002504, 0x00000000); + for (i = 0; i < 16; i++) + nv_wr32(dev, 0x002510 + (i * 4), 0x00000000); + nv_wr32(dev, 0x00250c, 0x0000ffff); + nv_wr32(dev, 0x002048, 0x00000000); + nv_wr32(dev, 0x003228, 0x00000000); + nv_wr32(dev, 0x0032e8, 0x00000000); + nv_wr32(dev, 0x002410, 0x00000000); + nv_wr32(dev, 0x002420, 0x00000000); + nv_wr32(dev, 0x002058, 0x00000001); + nv_wr32(dev, 0x00221c, 0x00000000); + /* something with 0x2084, read/modify/write, no change */ + nv_wr32(dev, 0x002040, 0x000000ff); + nv_wr32(dev, 0x002500, 0x00000000); + nv_wr32(dev, 0x003200, 0x00000000); + + nv_wr32(dev, NV04_PFIFO_DMA_TIMESLICE, 0x2101ffff); +} + +static void +nv40_fifo_init_ramxx(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + nv_wr32(dev, NV03_PFIFO_RAMHT, (0x03 << 24) /* search 128 */ | + ((dev_priv->ramht->bits - 9) << 16) | + (dev_priv->ramht->gpuobj->pinst >> 8)); + nv_wr32(dev, NV03_PFIFO_RAMRO, dev_priv->ramro->pinst >> 8); + + switch (dev_priv->chipset) { + case 0x47: + case 0x49: + case 0x4b: + nv_wr32(dev, 0x2230, 1); + break; + default: + break; + } + + switch (dev_priv->chipset) { + case 0x40: + case 0x41: + case 0x42: + case 0x43: + case 0x45: + case 0x47: + case 0x48: + case 0x49: + case 0x4b: + nv_wr32(dev, NV40_PFIFO_RAMFC, 0x30002); + break; + default: + nv_wr32(dev, 0x2230, 0); + nv_wr32(dev, NV40_PFIFO_RAMFC, + ((dev_priv->vram_size - 512 * 1024 + + dev_priv->ramfc->pinst) >> 16) | (3 << 16)); + break; + } +} + +static void +nv40_fifo_init_intr(struct drm_device *dev) +{ + nouveau_irq_register(dev, 8, nv04_fifo_isr); + nv_wr32(dev, 0x002100, 0xffffffff); + nv_wr32(dev, 0x002140, 0xffffffff); +} + +int +nv40_fifo_init(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; + int i; + + nv40_fifo_init_reset(dev); + nv40_fifo_init_ramxx(dev); + + nv40_fifo_do_load_context(dev, pfifo->channels - 1); + nv_wr32(dev, NV03_PFIFO_CACHE1_PUSH1, pfifo->channels - 1); + + nv40_fifo_init_intr(dev); + pfifo->enable(dev); + pfifo->reassign(dev, true); + + for (i = 0; i < dev_priv->engine.fifo.channels; i++) { + if (dev_priv->channels.ptr[i]) { + uint32_t mode = nv_rd32(dev, NV04_PFIFO_MODE); + nv_wr32(dev, NV04_PFIFO_MODE, mode | (1 << i)); + } + } + + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nv40_graph.c b/drivers/gpu/drm/nouveau/nv40_graph.c new file mode 100644 index 00000000..5beb01b8 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv40_graph.c @@ -0,0 +1,562 @@ +/* + * Copyright (C) 2007 Ben Skeggs. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "drmP.h" +#include "drm.h" +#include "nouveau_drv.h" +#include "nouveau_grctx.h" +#include "nouveau_ramht.h" + +struct nv40_graph_engine { + struct nouveau_exec_engine base; + u32 grctx_size; +}; + +static struct nouveau_channel * +nv40_graph_channel(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *grctx; + uint32_t inst; + int i; + + inst = nv_rd32(dev, NV40_PGRAPH_CTXCTL_CUR); + if (!(inst & NV40_PGRAPH_CTXCTL_CUR_LOADED)) + return NULL; + inst = (inst & NV40_PGRAPH_CTXCTL_CUR_INSTANCE) << 4; + + for (i = 0; i < dev_priv->engine.fifo.channels; i++) { + if (!dev_priv->channels.ptr[i]) + continue; + + grctx = dev_priv->channels.ptr[i]->engctx[NVOBJ_ENGINE_GR]; + if (grctx && grctx->pinst == inst) + return dev_priv->channels.ptr[i]; + } + + return NULL; +} + +static int +nv40_graph_transfer_context(struct drm_device *dev, uint32_t inst, int save) +{ + uint32_t old_cp, tv = 1000, tmp; + int i; + + old_cp = nv_rd32(dev, NV20_PGRAPH_CHANNEL_CTX_POINTER); + nv_wr32(dev, NV20_PGRAPH_CHANNEL_CTX_POINTER, inst); + + tmp = nv_rd32(dev, NV40_PGRAPH_CTXCTL_0310); + tmp |= save ? NV40_PGRAPH_CTXCTL_0310_XFER_SAVE : + NV40_PGRAPH_CTXCTL_0310_XFER_LOAD; + nv_wr32(dev, NV40_PGRAPH_CTXCTL_0310, tmp); + + tmp = nv_rd32(dev, NV40_PGRAPH_CTXCTL_0304); + tmp |= NV40_PGRAPH_CTXCTL_0304_XFER_CTX; + nv_wr32(dev, NV40_PGRAPH_CTXCTL_0304, tmp); + + nouveau_wait_for_idle(dev); + + for (i = 0; i < tv; i++) { + if (nv_rd32(dev, NV40_PGRAPH_CTXCTL_030C) == 0) + break; + } + + nv_wr32(dev, NV20_PGRAPH_CHANNEL_CTX_POINTER, old_cp); + + if (i == tv) { + uint32_t ucstat = nv_rd32(dev, NV40_PGRAPH_CTXCTL_UCODE_STAT); + NV_ERROR(dev, "Failed: Instance=0x%08x Save=%d\n", inst, save); + NV_ERROR(dev, "IP: 0x%02x, Opcode: 0x%08x\n", + ucstat >> NV40_PGRAPH_CTXCTL_UCODE_STAT_IP_SHIFT, + ucstat & NV40_PGRAPH_CTXCTL_UCODE_STAT_OP_MASK); + NV_ERROR(dev, "0x40030C = 0x%08x\n", + nv_rd32(dev, NV40_PGRAPH_CTXCTL_030C)); + return -EBUSY; + } + + return 0; +} + +static int +nv40_graph_unload_context(struct drm_device *dev) +{ + uint32_t inst; + int ret; + + inst = nv_rd32(dev, NV40_PGRAPH_CTXCTL_CUR); + if (!(inst & NV40_PGRAPH_CTXCTL_CUR_LOADED)) + return 0; + inst &= NV40_PGRAPH_CTXCTL_CUR_INSTANCE; + + ret = nv40_graph_transfer_context(dev, inst, 1); + + nv_wr32(dev, NV40_PGRAPH_CTXCTL_CUR, inst); + return ret; +} + +static int +nv40_graph_context_new(struct nouveau_channel *chan, int engine) +{ + struct nv40_graph_engine *pgraph = nv_engine(chan->dev, engine); + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *grctx = NULL; + struct nouveau_grctx ctx = {}; + unsigned long flags; + int ret; + + ret = nouveau_gpuobj_new(dev, NULL, pgraph->grctx_size, 16, + NVOBJ_FLAG_ZERO_ALLOC, &grctx); + if (ret) + return ret; + + /* Initialise default context values */ + ctx.dev = chan->dev; + ctx.mode = NOUVEAU_GRCTX_VALS; + ctx.data = grctx; + nv40_grctx_init(&ctx); + + nv_wo32(grctx, 0, grctx->vinst); + + /* init grctx pointer in ramfc, and on PFIFO if channel is + * already active there + */ + spin_lock_irqsave(&dev_priv->context_switch_lock, flags); + nv_wo32(chan->ramfc, 0x38, grctx->vinst >> 4); + nv_mask(dev, 0x002500, 0x00000001, 0x00000000); + if ((nv_rd32(dev, 0x003204) & 0x0000001f) == chan->id) + nv_wr32(dev, 0x0032e0, grctx->vinst >> 4); + nv_mask(dev, 0x002500, 0x00000001, 0x00000001); + spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); + + chan->engctx[engine] = grctx; + return 0; +} + +static void +nv40_graph_context_del(struct nouveau_channel *chan, int engine) +{ + struct nouveau_gpuobj *grctx = chan->engctx[engine]; + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + unsigned long flags; + + spin_lock_irqsave(&dev_priv->context_switch_lock, flags); + nv04_graph_fifo_access(dev, false); + + /* Unload the context if it's the currently active one */ + if (nv40_graph_channel(dev) == chan) + nv40_graph_unload_context(dev); + + nv04_graph_fifo_access(dev, true); + spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); + + /* Free the context resources */ + nouveau_gpuobj_ref(NULL, &grctx); + chan->engctx[engine] = NULL; +} + +int +nv40_graph_object_new(struct nouveau_channel *chan, int engine, + u32 handle, u16 class) +{ + struct drm_device *dev = chan->dev; + struct nouveau_gpuobj *obj = NULL; + int ret; + + ret = nouveau_gpuobj_new(dev, chan, 20, 16, NVOBJ_FLAG_ZERO_FREE, &obj); + if (ret) + return ret; + obj->engine = 1; + obj->class = class; + + nv_wo32(obj, 0x00, class); + nv_wo32(obj, 0x04, 0x00000000); +#ifndef __BIG_ENDIAN + nv_wo32(obj, 0x08, 0x00000000); +#else + nv_wo32(obj, 0x08, 0x01000000); +#endif + nv_wo32(obj, 0x0c, 0x00000000); + nv_wo32(obj, 0x10, 0x00000000); + + ret = nouveau_ramht_insert(chan, handle, obj); + nouveau_gpuobj_ref(NULL, &obj); + return ret; +} + +static void +nv40_graph_set_tile_region(struct drm_device *dev, int i) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_tile_reg *tile = &dev_priv->tile.reg[i]; + + switch (dev_priv->chipset) { + case 0x40: + case 0x41: /* guess */ + case 0x42: + case 0x43: + case 0x45: /* guess */ + case 0x4e: + nv_wr32(dev, NV20_PGRAPH_TSIZE(i), tile->pitch); + nv_wr32(dev, NV20_PGRAPH_TLIMIT(i), tile->limit); + nv_wr32(dev, NV20_PGRAPH_TILE(i), tile->addr); + nv_wr32(dev, NV40_PGRAPH_TSIZE1(i), tile->pitch); + nv_wr32(dev, NV40_PGRAPH_TLIMIT1(i), tile->limit); + nv_wr32(dev, NV40_PGRAPH_TILE1(i), tile->addr); + break; + case 0x44: + case 0x4a: + nv_wr32(dev, NV20_PGRAPH_TSIZE(i), tile->pitch); + nv_wr32(dev, NV20_PGRAPH_TLIMIT(i), tile->limit); + nv_wr32(dev, NV20_PGRAPH_TILE(i), tile->addr); + break; + case 0x46: + case 0x47: + case 0x49: + case 0x4b: + case 0x4c: + case 0x67: + default: + nv_wr32(dev, NV47_PGRAPH_TSIZE(i), tile->pitch); + nv_wr32(dev, NV47_PGRAPH_TLIMIT(i), tile->limit); + nv_wr32(dev, NV47_PGRAPH_TILE(i), tile->addr); + nv_wr32(dev, NV40_PGRAPH_TSIZE1(i), tile->pitch); + nv_wr32(dev, NV40_PGRAPH_TLIMIT1(i), tile->limit); + nv_wr32(dev, NV40_PGRAPH_TILE1(i), tile->addr); + break; + } +} + +/* + * G70 0x47 + * G71 0x49 + * NV45 0x48 + * G72[M] 0x46 + * G73 0x4b + * C51_G7X 0x4c + * C51 0x4e + */ +int +nv40_graph_init(struct drm_device *dev, int engine) +{ + struct nv40_graph_engine *pgraph = nv_engine(dev, engine); + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fb_engine *pfb = &dev_priv->engine.fb; + struct nouveau_grctx ctx = {}; + uint32_t vramsz, *cp; + int i, j; + + nv_wr32(dev, NV03_PMC_ENABLE, nv_rd32(dev, NV03_PMC_ENABLE) & + ~NV_PMC_ENABLE_PGRAPH); + nv_wr32(dev, NV03_PMC_ENABLE, nv_rd32(dev, NV03_PMC_ENABLE) | + NV_PMC_ENABLE_PGRAPH); + + cp = kmalloc(sizeof(*cp) * 256, GFP_KERNEL); + if (!cp) + return -ENOMEM; + + ctx.dev = dev; + ctx.mode = NOUVEAU_GRCTX_PROG; + ctx.data = cp; + ctx.ctxprog_max = 256; + nv40_grctx_init(&ctx); + pgraph->grctx_size = ctx.ctxvals_pos * 4; + + nv_wr32(dev, NV40_PGRAPH_CTXCTL_UCODE_INDEX, 0); + for (i = 0; i < ctx.ctxprog_len; i++) + nv_wr32(dev, NV40_PGRAPH_CTXCTL_UCODE_DATA, cp[i]); + + kfree(cp); + + /* No context present currently */ + nv_wr32(dev, NV40_PGRAPH_CTXCTL_CUR, 0x00000000); + + nv_wr32(dev, NV03_PGRAPH_INTR , 0xFFFFFFFF); + nv_wr32(dev, NV40_PGRAPH_INTR_EN, 0xFFFFFFFF); + + nv_wr32(dev, NV04_PGRAPH_DEBUG_0, 0xFFFFFFFF); + nv_wr32(dev, NV04_PGRAPH_DEBUG_0, 0x00000000); + nv_wr32(dev, NV04_PGRAPH_DEBUG_1, 0x401287c0); + nv_wr32(dev, NV04_PGRAPH_DEBUG_3, 0xe0de8055); + nv_wr32(dev, NV10_PGRAPH_DEBUG_4, 0x00008000); + nv_wr32(dev, NV04_PGRAPH_LIMIT_VIOL_PIX, 0x00be3c5f); + + nv_wr32(dev, NV10_PGRAPH_CTX_CONTROL, 0x10010100); + nv_wr32(dev, NV10_PGRAPH_STATE , 0xFFFFFFFF); + + j = nv_rd32(dev, 0x1540) & 0xff; + if (j) { + for (i = 0; !(j & 1); j >>= 1, i++) + ; + nv_wr32(dev, 0x405000, i); + } + + if (dev_priv->chipset == 0x40) { + nv_wr32(dev, 0x4009b0, 0x83280fff); + nv_wr32(dev, 0x4009b4, 0x000000a0); + } else { + nv_wr32(dev, 0x400820, 0x83280eff); + nv_wr32(dev, 0x400824, 0x000000a0); + } + + switch (dev_priv->chipset) { + case 0x40: + case 0x45: + nv_wr32(dev, 0x4009b8, 0x0078e366); + nv_wr32(dev, 0x4009bc, 0x0000014c); + break; + case 0x41: + case 0x42: /* pciid also 0x00Cx */ + /* case 0x0120: XXX (pciid) */ + nv_wr32(dev, 0x400828, 0x007596ff); + nv_wr32(dev, 0x40082c, 0x00000108); + break; + case 0x43: + nv_wr32(dev, 0x400828, 0x0072cb77); + nv_wr32(dev, 0x40082c, 0x00000108); + break; + case 0x44: + case 0x46: /* G72 */ + case 0x4a: + case 0x4c: /* G7x-based C51 */ + case 0x4e: + nv_wr32(dev, 0x400860, 0); + nv_wr32(dev, 0x400864, 0); + break; + case 0x47: /* G70 */ + case 0x49: /* G71 */ + case 0x4b: /* G73 */ + nv_wr32(dev, 0x400828, 0x07830610); + nv_wr32(dev, 0x40082c, 0x0000016A); + break; + default: + break; + } + + nv_wr32(dev, 0x400b38, 0x2ffff800); + nv_wr32(dev, 0x400b3c, 0x00006000); + + /* Tiling related stuff. */ + switch (dev_priv->chipset) { + case 0x44: + case 0x4a: + nv_wr32(dev, 0x400bc4, 0x1003d888); + nv_wr32(dev, 0x400bbc, 0xb7a7b500); + break; + case 0x46: + nv_wr32(dev, 0x400bc4, 0x0000e024); + nv_wr32(dev, 0x400bbc, 0xb7a7b520); + break; + case 0x4c: + case 0x4e: + case 0x67: + nv_wr32(dev, 0x400bc4, 0x1003d888); + nv_wr32(dev, 0x400bbc, 0xb7a7b540); + break; + default: + break; + } + + /* Turn all the tiling regions off. */ + for (i = 0; i < pfb->num_tiles; i++) + nv40_graph_set_tile_region(dev, i); + + /* begin RAM config */ + vramsz = pci_resource_len(dev->pdev, 0) - 1; + switch (dev_priv->chipset) { + case 0x40: + nv_wr32(dev, 0x4009A4, nv_rd32(dev, NV04_PFB_CFG0)); + nv_wr32(dev, 0x4009A8, nv_rd32(dev, NV04_PFB_CFG1)); + nv_wr32(dev, 0x4069A4, nv_rd32(dev, NV04_PFB_CFG0)); + nv_wr32(dev, 0x4069A8, nv_rd32(dev, NV04_PFB_CFG1)); + nv_wr32(dev, 0x400820, 0); + nv_wr32(dev, 0x400824, 0); + nv_wr32(dev, 0x400864, vramsz); + nv_wr32(dev, 0x400868, vramsz); + break; + default: + switch (dev_priv->chipset) { + case 0x41: + case 0x42: + case 0x43: + case 0x45: + case 0x4e: + case 0x44: + case 0x4a: + nv_wr32(dev, 0x4009F0, nv_rd32(dev, NV04_PFB_CFG0)); + nv_wr32(dev, 0x4009F4, nv_rd32(dev, NV04_PFB_CFG1)); + break; + default: + nv_wr32(dev, 0x400DF0, nv_rd32(dev, NV04_PFB_CFG0)); + nv_wr32(dev, 0x400DF4, nv_rd32(dev, NV04_PFB_CFG1)); + break; + } + nv_wr32(dev, 0x4069F0, nv_rd32(dev, NV04_PFB_CFG0)); + nv_wr32(dev, 0x4069F4, nv_rd32(dev, NV04_PFB_CFG1)); + nv_wr32(dev, 0x400840, 0); + nv_wr32(dev, 0x400844, 0); + nv_wr32(dev, 0x4008A0, vramsz); + nv_wr32(dev, 0x4008A4, vramsz); + break; + } + + return 0; +} + +static int +nv40_graph_fini(struct drm_device *dev, int engine) +{ + nv40_graph_unload_context(dev); + return 0; +} + +static int +nv40_graph_isr_chid(struct drm_device *dev, u32 inst) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *grctx; + unsigned long flags; + int i; + + spin_lock_irqsave(&dev_priv->channels.lock, flags); + for (i = 0; i < dev_priv->engine.fifo.channels; i++) { + if (!dev_priv->channels.ptr[i]) + continue; + grctx = dev_priv->channels.ptr[i]->engctx[NVOBJ_ENGINE_GR]; + + if (grctx && grctx->pinst == inst) + break; + } + spin_unlock_irqrestore(&dev_priv->channels.lock, flags); + return i; +} + +static void +nv40_graph_isr(struct drm_device *dev) +{ + u32 stat; + + while ((stat = nv_rd32(dev, NV03_PGRAPH_INTR))) { + u32 nsource = nv_rd32(dev, NV03_PGRAPH_NSOURCE); + u32 nstatus = nv_rd32(dev, NV03_PGRAPH_NSTATUS); + u32 inst = (nv_rd32(dev, 0x40032c) & 0x000fffff) << 4; + u32 chid = nv40_graph_isr_chid(dev, inst); + u32 addr = nv_rd32(dev, NV04_PGRAPH_TRAPPED_ADDR); + u32 subc = (addr & 0x00070000) >> 16; + u32 mthd = (addr & 0x00001ffc); + u32 data = nv_rd32(dev, NV04_PGRAPH_TRAPPED_DATA); + u32 class = nv_rd32(dev, 0x400160 + subc * 4) & 0xffff; + u32 show = stat; + + if (stat & NV_PGRAPH_INTR_ERROR) { + if (nsource & NV03_PGRAPH_NSOURCE_ILLEGAL_MTHD) { + if (!nouveau_gpuobj_mthd_call2(dev, chid, class, mthd, data)) + show &= ~NV_PGRAPH_INTR_ERROR; + } else + if (nsource & NV03_PGRAPH_NSOURCE_DMA_VTX_PROTECTION) { + nv_mask(dev, 0x402000, 0, 0); + } + } + + nv_wr32(dev, NV03_PGRAPH_INTR, stat); + nv_wr32(dev, NV04_PGRAPH_FIFO, 0x00000001); + + if (show && nouveau_ratelimit()) { + NV_INFO(dev, "PGRAPH -"); + nouveau_bitfield_print(nv10_graph_intr, show); + printk(" nsource:"); + nouveau_bitfield_print(nv04_graph_nsource, nsource); + printk(" nstatus:"); + nouveau_bitfield_print(nv10_graph_nstatus, nstatus); + printk("\n"); + NV_INFO(dev, "PGRAPH - ch %d (0x%08x) subc %d " + "class 0x%04x mthd 0x%04x data 0x%08x\n", + chid, inst, subc, class, mthd, data); + } + } +} + +static void +nv40_graph_destroy(struct drm_device *dev, int engine) +{ + struct nv40_graph_engine *pgraph = nv_engine(dev, engine); + + nouveau_irq_unregister(dev, 12); + + NVOBJ_ENGINE_DEL(dev, GR); + kfree(pgraph); +} + +int +nv40_graph_create(struct drm_device *dev) +{ + struct nv40_graph_engine *pgraph; + + pgraph = kzalloc(sizeof(*pgraph), GFP_KERNEL); + if (!pgraph) + return -ENOMEM; + + pgraph->base.destroy = nv40_graph_destroy; + pgraph->base.init = nv40_graph_init; + pgraph->base.fini = nv40_graph_fini; + pgraph->base.context_new = nv40_graph_context_new; + pgraph->base.context_del = nv40_graph_context_del; + pgraph->base.object_new = nv40_graph_object_new; + pgraph->base.set_tile_region = nv40_graph_set_tile_region; + + NVOBJ_ENGINE_ADD(dev, GR, &pgraph->base); + nouveau_irq_register(dev, 12, nv40_graph_isr); + + NVOBJ_CLASS(dev, 0x506e, SW); /* nvsw */ + NVOBJ_CLASS(dev, 0x0030, GR); /* null */ + NVOBJ_CLASS(dev, 0x0039, GR); /* m2mf */ + NVOBJ_CLASS(dev, 0x004a, GR); /* gdirect */ + NVOBJ_CLASS(dev, 0x009f, GR); /* imageblit (nv12) */ + NVOBJ_CLASS(dev, 0x008a, GR); /* ifc */ + NVOBJ_CLASS(dev, 0x0089, GR); /* sifm */ + NVOBJ_CLASS(dev, 0x3089, GR); /* sifm (nv40) */ + NVOBJ_CLASS(dev, 0x0062, GR); /* surf2d */ + NVOBJ_CLASS(dev, 0x3062, GR); /* surf2d (nv40) */ + NVOBJ_CLASS(dev, 0x0043, GR); /* rop */ + NVOBJ_CLASS(dev, 0x0012, GR); /* beta1 */ + NVOBJ_CLASS(dev, 0x0072, GR); /* beta4 */ + NVOBJ_CLASS(dev, 0x0019, GR); /* cliprect */ + NVOBJ_CLASS(dev, 0x0044, GR); /* pattern */ + NVOBJ_CLASS(dev, 0x309e, GR); /* swzsurf */ + + /* curie */ + if (nv44_graph_class(dev)) + NVOBJ_CLASS(dev, 0x4497, GR); + else + NVOBJ_CLASS(dev, 0x4097, GR); + + /* nvsw */ + NVOBJ_CLASS(dev, 0x506e, SW); + NVOBJ_MTHD (dev, 0x506e, 0x0500, nv04_graph_mthd_page_flip); + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nv40_grctx.c b/drivers/gpu/drm/nouveau/nv40_grctx.c new file mode 100644 index 00000000..f70447d1 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv40_grctx.c @@ -0,0 +1,662 @@ +/* + * Copyright 2009 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +/* NVIDIA context programs handle a number of other conditions which are + * not implemented in our versions. It's not clear why NVIDIA context + * programs have this code, nor whether it's strictly necessary for + * correct operation. We'll implement additional handling if/when we + * discover it's necessary. + * + * - On context save, NVIDIA set 0x400314 bit 0 to 1 if the "3D state" + * flag is set, this gets saved into the context. + * - On context save, the context program for all cards load nsource + * into a flag register and check for ILLEGAL_MTHD. If it's set, + * opcode 0x60000d is called before resuming normal operation. + * - Some context programs check more conditions than the above. NV44 + * checks: ((nsource & 0x0857) || (0x400718 & 0x0100) || (intr & 0x0001)) + * and calls 0x60000d before resuming normal operation. + * - At the very beginning of NVIDIA's context programs, flag 9 is checked + * and if true 0x800001 is called with count=0, pos=0, the flag is cleared + * and then the ctxprog is aborted. It looks like a complicated NOP, + * its purpose is unknown. + * - In the section of code that loads the per-vs state, NVIDIA check + * flag 10. If it's set, they only transfer the small 0x300 byte block + * of state + the state for a single vs as opposed to the state for + * all vs units. It doesn't seem likely that it'll occur in normal + * operation, especially seeing as it appears NVIDIA may have screwed + * up the ctxprogs for some cards and have an invalid instruction + * rather than a cp_lsr(ctx, dwords_for_1_vs_unit) instruction. + * - There's a number of places where context offset 0 (where we place + * the PRAMIN offset of the context) is loaded into either 0x408000, + * 0x408004 or 0x408008. Not sure what's up there either. + * - The ctxprogs for some cards save 0x400a00 again during the cleanup + * path for auto-loadctx. + */ + +#define CP_FLAG_CLEAR 0 +#define CP_FLAG_SET 1 +#define CP_FLAG_SWAP_DIRECTION ((0 * 32) + 0) +#define CP_FLAG_SWAP_DIRECTION_LOAD 0 +#define CP_FLAG_SWAP_DIRECTION_SAVE 1 +#define CP_FLAG_USER_SAVE ((0 * 32) + 5) +#define CP_FLAG_USER_SAVE_NOT_PENDING 0 +#define CP_FLAG_USER_SAVE_PENDING 1 +#define CP_FLAG_USER_LOAD ((0 * 32) + 6) +#define CP_FLAG_USER_LOAD_NOT_PENDING 0 +#define CP_FLAG_USER_LOAD_PENDING 1 +#define CP_FLAG_STATUS ((3 * 32) + 0) +#define CP_FLAG_STATUS_IDLE 0 +#define CP_FLAG_STATUS_BUSY 1 +#define CP_FLAG_AUTO_SAVE ((3 * 32) + 4) +#define CP_FLAG_AUTO_SAVE_NOT_PENDING 0 +#define CP_FLAG_AUTO_SAVE_PENDING 1 +#define CP_FLAG_AUTO_LOAD ((3 * 32) + 5) +#define CP_FLAG_AUTO_LOAD_NOT_PENDING 0 +#define CP_FLAG_AUTO_LOAD_PENDING 1 +#define CP_FLAG_UNK54 ((3 * 32) + 6) +#define CP_FLAG_UNK54_CLEAR 0 +#define CP_FLAG_UNK54_SET 1 +#define CP_FLAG_ALWAYS ((3 * 32) + 8) +#define CP_FLAG_ALWAYS_FALSE 0 +#define CP_FLAG_ALWAYS_TRUE 1 +#define CP_FLAG_UNK57 ((3 * 32) + 9) +#define CP_FLAG_UNK57_CLEAR 0 +#define CP_FLAG_UNK57_SET 1 + +#define CP_CTX 0x00100000 +#define CP_CTX_COUNT 0x000fc000 +#define CP_CTX_COUNT_SHIFT 14 +#define CP_CTX_REG 0x00003fff +#define CP_LOAD_SR 0x00200000 +#define CP_LOAD_SR_VALUE 0x000fffff +#define CP_BRA 0x00400000 +#define CP_BRA_IP 0x0000ff00 +#define CP_BRA_IP_SHIFT 8 +#define CP_BRA_IF_CLEAR 0x00000080 +#define CP_BRA_FLAG 0x0000007f +#define CP_WAIT 0x00500000 +#define CP_WAIT_SET 0x00000080 +#define CP_WAIT_FLAG 0x0000007f +#define CP_SET 0x00700000 +#define CP_SET_1 0x00000080 +#define CP_SET_FLAG 0x0000007f +#define CP_NEXT_TO_SWAP 0x00600007 +#define CP_NEXT_TO_CURRENT 0x00600009 +#define CP_SET_CONTEXT_POINTER 0x0060000a +#define CP_END 0x0060000e +#define CP_LOAD_MAGIC_UNK01 0x00800001 /* unknown */ +#define CP_LOAD_MAGIC_NV44TCL 0x00800029 /* per-vs state (0x4497) */ +#define CP_LOAD_MAGIC_NV40TCL 0x00800041 /* per-vs state (0x4097) */ + +#include "drmP.h" +#include "nouveau_drv.h" +#include "nouveau_grctx.h" + +/* TODO: + * - get vs count from 0x1540 + */ + +static int +nv40_graph_vs_count(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + switch (dev_priv->chipset) { + case 0x47: + case 0x49: + case 0x4b: + return 8; + case 0x40: + return 6; + case 0x41: + case 0x42: + return 5; + case 0x43: + case 0x44: + case 0x46: + case 0x4a: + return 3; + case 0x4c: + case 0x4e: + case 0x67: + default: + return 1; + } +} + + +enum cp_label { + cp_check_load = 1, + cp_setup_auto_load, + cp_setup_load, + cp_setup_save, + cp_swap_state, + cp_swap_state3d_3_is_save, + cp_prepare_exit, + cp_exit, +}; + +static void +nv40_graph_construct_general(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + int i; + + cp_ctx(ctx, 0x4000a4, 1); + gr_def(ctx, 0x4000a4, 0x00000008); + cp_ctx(ctx, 0x400144, 58); + gr_def(ctx, 0x400144, 0x00000001); + cp_ctx(ctx, 0x400314, 1); + gr_def(ctx, 0x400314, 0x00000000); + cp_ctx(ctx, 0x400400, 10); + cp_ctx(ctx, 0x400480, 10); + cp_ctx(ctx, 0x400500, 19); + gr_def(ctx, 0x400514, 0x00040000); + gr_def(ctx, 0x400524, 0x55555555); + gr_def(ctx, 0x400528, 0x55555555); + gr_def(ctx, 0x40052c, 0x55555555); + gr_def(ctx, 0x400530, 0x55555555); + cp_ctx(ctx, 0x400560, 6); + gr_def(ctx, 0x400568, 0x0000ffff); + gr_def(ctx, 0x40056c, 0x0000ffff); + cp_ctx(ctx, 0x40057c, 5); + cp_ctx(ctx, 0x400710, 3); + gr_def(ctx, 0x400710, 0x20010001); + gr_def(ctx, 0x400714, 0x0f73ef00); + cp_ctx(ctx, 0x400724, 1); + gr_def(ctx, 0x400724, 0x02008821); + cp_ctx(ctx, 0x400770, 3); + if (dev_priv->chipset == 0x40) { + cp_ctx(ctx, 0x400814, 4); + cp_ctx(ctx, 0x400828, 5); + cp_ctx(ctx, 0x400840, 5); + gr_def(ctx, 0x400850, 0x00000040); + cp_ctx(ctx, 0x400858, 4); + gr_def(ctx, 0x400858, 0x00000040); + gr_def(ctx, 0x40085c, 0x00000040); + gr_def(ctx, 0x400864, 0x80000000); + cp_ctx(ctx, 0x40086c, 9); + gr_def(ctx, 0x40086c, 0x80000000); + gr_def(ctx, 0x400870, 0x80000000); + gr_def(ctx, 0x400874, 0x80000000); + gr_def(ctx, 0x400878, 0x80000000); + gr_def(ctx, 0x400888, 0x00000040); + gr_def(ctx, 0x40088c, 0x80000000); + cp_ctx(ctx, 0x4009c0, 8); + gr_def(ctx, 0x4009cc, 0x80000000); + gr_def(ctx, 0x4009dc, 0x80000000); + } else { + cp_ctx(ctx, 0x400840, 20); + if (nv44_graph_class(ctx->dev)) { + for (i = 0; i < 8; i++) + gr_def(ctx, 0x400860 + (i * 4), 0x00000001); + } + gr_def(ctx, 0x400880, 0x00000040); + gr_def(ctx, 0x400884, 0x00000040); + gr_def(ctx, 0x400888, 0x00000040); + cp_ctx(ctx, 0x400894, 11); + gr_def(ctx, 0x400894, 0x00000040); + if (!nv44_graph_class(ctx->dev)) { + for (i = 0; i < 8; i++) + gr_def(ctx, 0x4008a0 + (i * 4), 0x80000000); + } + cp_ctx(ctx, 0x4008e0, 2); + cp_ctx(ctx, 0x4008f8, 2); + if (dev_priv->chipset == 0x4c || + (dev_priv->chipset & 0xf0) == 0x60) + cp_ctx(ctx, 0x4009f8, 1); + } + cp_ctx(ctx, 0x400a00, 73); + gr_def(ctx, 0x400b0c, 0x0b0b0b0c); + cp_ctx(ctx, 0x401000, 4); + cp_ctx(ctx, 0x405004, 1); + switch (dev_priv->chipset) { + case 0x47: + case 0x49: + case 0x4b: + cp_ctx(ctx, 0x403448, 1); + gr_def(ctx, 0x403448, 0x00001010); + break; + default: + cp_ctx(ctx, 0x403440, 1); + switch (dev_priv->chipset) { + case 0x40: + gr_def(ctx, 0x403440, 0x00000010); + break; + case 0x44: + case 0x46: + case 0x4a: + gr_def(ctx, 0x403440, 0x00003010); + break; + case 0x41: + case 0x42: + case 0x43: + case 0x4c: + case 0x4e: + case 0x67: + default: + gr_def(ctx, 0x403440, 0x00001010); + break; + } + break; + } +} + +static void +nv40_graph_construct_state3d(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + int i; + + if (dev_priv->chipset == 0x40) { + cp_ctx(ctx, 0x401880, 51); + gr_def(ctx, 0x401940, 0x00000100); + } else + if (dev_priv->chipset == 0x46 || dev_priv->chipset == 0x47 || + dev_priv->chipset == 0x49 || dev_priv->chipset == 0x4b) { + cp_ctx(ctx, 0x401880, 32); + for (i = 0; i < 16; i++) + gr_def(ctx, 0x401880 + (i * 4), 0x00000111); + if (dev_priv->chipset == 0x46) + cp_ctx(ctx, 0x401900, 16); + cp_ctx(ctx, 0x401940, 3); + } + cp_ctx(ctx, 0x40194c, 18); + gr_def(ctx, 0x401954, 0x00000111); + gr_def(ctx, 0x401958, 0x00080060); + gr_def(ctx, 0x401974, 0x00000080); + gr_def(ctx, 0x401978, 0xffff0000); + gr_def(ctx, 0x40197c, 0x00000001); + gr_def(ctx, 0x401990, 0x46400000); + if (dev_priv->chipset == 0x40) { + cp_ctx(ctx, 0x4019a0, 2); + cp_ctx(ctx, 0x4019ac, 5); + } else { + cp_ctx(ctx, 0x4019a0, 1); + cp_ctx(ctx, 0x4019b4, 3); + } + gr_def(ctx, 0x4019bc, 0xffff0000); + switch (dev_priv->chipset) { + case 0x46: + case 0x47: + case 0x49: + case 0x4b: + cp_ctx(ctx, 0x4019c0, 18); + for (i = 0; i < 16; i++) + gr_def(ctx, 0x4019c0 + (i * 4), 0x88888888); + break; + } + cp_ctx(ctx, 0x401a08, 8); + gr_def(ctx, 0x401a10, 0x0fff0000); + gr_def(ctx, 0x401a14, 0x0fff0000); + gr_def(ctx, 0x401a1c, 0x00011100); + cp_ctx(ctx, 0x401a2c, 4); + cp_ctx(ctx, 0x401a44, 26); + for (i = 0; i < 16; i++) + gr_def(ctx, 0x401a44 + (i * 4), 0x07ff0000); + gr_def(ctx, 0x401a8c, 0x4b7fffff); + if (dev_priv->chipset == 0x40) { + cp_ctx(ctx, 0x401ab8, 3); + } else { + cp_ctx(ctx, 0x401ab8, 1); + cp_ctx(ctx, 0x401ac0, 1); + } + cp_ctx(ctx, 0x401ad0, 8); + gr_def(ctx, 0x401ad0, 0x30201000); + gr_def(ctx, 0x401ad4, 0x70605040); + gr_def(ctx, 0x401ad8, 0xb8a89888); + gr_def(ctx, 0x401adc, 0xf8e8d8c8); + cp_ctx(ctx, 0x401b10, dev_priv->chipset == 0x40 ? 2 : 1); + gr_def(ctx, 0x401b10, 0x40100000); + cp_ctx(ctx, 0x401b18, dev_priv->chipset == 0x40 ? 6 : 5); + gr_def(ctx, 0x401b28, dev_priv->chipset == 0x40 ? + 0x00000004 : 0x00000000); + cp_ctx(ctx, 0x401b30, 25); + gr_def(ctx, 0x401b34, 0x0000ffff); + gr_def(ctx, 0x401b68, 0x435185d6); + gr_def(ctx, 0x401b6c, 0x2155b699); + gr_def(ctx, 0x401b70, 0xfedcba98); + gr_def(ctx, 0x401b74, 0x00000098); + gr_def(ctx, 0x401b84, 0xffffffff); + gr_def(ctx, 0x401b88, 0x00ff7000); + gr_def(ctx, 0x401b8c, 0x0000ffff); + if (dev_priv->chipset != 0x44 && dev_priv->chipset != 0x4a && + dev_priv->chipset != 0x4e) + cp_ctx(ctx, 0x401b94, 1); + cp_ctx(ctx, 0x401b98, 8); + gr_def(ctx, 0x401b9c, 0x00ff0000); + cp_ctx(ctx, 0x401bc0, 9); + gr_def(ctx, 0x401be0, 0x00ffff00); + cp_ctx(ctx, 0x401c00, 192); + for (i = 0; i < 16; i++) { /* fragment texture units */ + gr_def(ctx, 0x401c40 + (i * 4), 0x00018488); + gr_def(ctx, 0x401c80 + (i * 4), 0x00028202); + gr_def(ctx, 0x401d00 + (i * 4), 0x0000aae4); + gr_def(ctx, 0x401d40 + (i * 4), 0x01012000); + gr_def(ctx, 0x401d80 + (i * 4), 0x00080008); + gr_def(ctx, 0x401e00 + (i * 4), 0x00100008); + } + for (i = 0; i < 4; i++) { /* vertex texture units */ + gr_def(ctx, 0x401e90 + (i * 4), 0x0001bc80); + gr_def(ctx, 0x401ea0 + (i * 4), 0x00000202); + gr_def(ctx, 0x401ec0 + (i * 4), 0x00000008); + gr_def(ctx, 0x401ee0 + (i * 4), 0x00080008); + } + cp_ctx(ctx, 0x400f5c, 3); + gr_def(ctx, 0x400f5c, 0x00000002); + cp_ctx(ctx, 0x400f84, 1); +} + +static void +nv40_graph_construct_state3d_2(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + int i; + + cp_ctx(ctx, 0x402000, 1); + cp_ctx(ctx, 0x402404, dev_priv->chipset == 0x40 ? 1 : 2); + switch (dev_priv->chipset) { + case 0x40: + gr_def(ctx, 0x402404, 0x00000001); + break; + case 0x4c: + case 0x4e: + case 0x67: + gr_def(ctx, 0x402404, 0x00000020); + break; + case 0x46: + case 0x49: + case 0x4b: + gr_def(ctx, 0x402404, 0x00000421); + break; + default: + gr_def(ctx, 0x402404, 0x00000021); + } + if (dev_priv->chipset != 0x40) + gr_def(ctx, 0x402408, 0x030c30c3); + switch (dev_priv->chipset) { + case 0x44: + case 0x46: + case 0x4a: + case 0x4c: + case 0x4e: + case 0x67: + cp_ctx(ctx, 0x402440, 1); + gr_def(ctx, 0x402440, 0x00011001); + break; + default: + break; + } + cp_ctx(ctx, 0x402480, dev_priv->chipset == 0x40 ? 8 : 9); + gr_def(ctx, 0x402488, 0x3e020200); + gr_def(ctx, 0x40248c, 0x00ffffff); + switch (dev_priv->chipset) { + case 0x40: + gr_def(ctx, 0x402490, 0x60103f00); + break; + case 0x47: + gr_def(ctx, 0x402490, 0x40103f00); + break; + case 0x41: + case 0x42: + case 0x49: + case 0x4b: + gr_def(ctx, 0x402490, 0x20103f00); + break; + default: + gr_def(ctx, 0x402490, 0x0c103f00); + break; + } + gr_def(ctx, 0x40249c, dev_priv->chipset <= 0x43 ? + 0x00020000 : 0x00040000); + cp_ctx(ctx, 0x402500, 31); + gr_def(ctx, 0x402530, 0x00008100); + if (dev_priv->chipset == 0x40) + cp_ctx(ctx, 0x40257c, 6); + cp_ctx(ctx, 0x402594, 16); + cp_ctx(ctx, 0x402800, 17); + gr_def(ctx, 0x402800, 0x00000001); + switch (dev_priv->chipset) { + case 0x47: + case 0x49: + case 0x4b: + cp_ctx(ctx, 0x402864, 1); + gr_def(ctx, 0x402864, 0x00001001); + cp_ctx(ctx, 0x402870, 3); + gr_def(ctx, 0x402878, 0x00000003); + if (dev_priv->chipset != 0x47) { /* belong at end!! */ + cp_ctx(ctx, 0x402900, 1); + cp_ctx(ctx, 0x402940, 1); + cp_ctx(ctx, 0x402980, 1); + cp_ctx(ctx, 0x4029c0, 1); + cp_ctx(ctx, 0x402a00, 1); + cp_ctx(ctx, 0x402a40, 1); + cp_ctx(ctx, 0x402a80, 1); + cp_ctx(ctx, 0x402ac0, 1); + } + break; + case 0x40: + cp_ctx(ctx, 0x402844, 1); + gr_def(ctx, 0x402844, 0x00000001); + cp_ctx(ctx, 0x402850, 1); + break; + default: + cp_ctx(ctx, 0x402844, 1); + gr_def(ctx, 0x402844, 0x00001001); + cp_ctx(ctx, 0x402850, 2); + gr_def(ctx, 0x402854, 0x00000003); + break; + } + + cp_ctx(ctx, 0x402c00, 4); + gr_def(ctx, 0x402c00, dev_priv->chipset == 0x40 ? + 0x80800001 : 0x00888001); + switch (dev_priv->chipset) { + case 0x47: + case 0x49: + case 0x4b: + cp_ctx(ctx, 0x402c20, 40); + for (i = 0; i < 32; i++) + gr_def(ctx, 0x402c40 + (i * 4), 0xffffffff); + cp_ctx(ctx, 0x4030b8, 13); + gr_def(ctx, 0x4030dc, 0x00000005); + gr_def(ctx, 0x4030e8, 0x0000ffff); + break; + default: + cp_ctx(ctx, 0x402c10, 4); + if (dev_priv->chipset == 0x40) + cp_ctx(ctx, 0x402c20, 36); + else + if (dev_priv->chipset <= 0x42) + cp_ctx(ctx, 0x402c20, 24); + else + if (dev_priv->chipset <= 0x4a) + cp_ctx(ctx, 0x402c20, 16); + else + cp_ctx(ctx, 0x402c20, 8); + cp_ctx(ctx, 0x402cb0, dev_priv->chipset == 0x40 ? 12 : 13); + gr_def(ctx, 0x402cd4, 0x00000005); + if (dev_priv->chipset != 0x40) + gr_def(ctx, 0x402ce0, 0x0000ffff); + break; + } + + cp_ctx(ctx, 0x403400, dev_priv->chipset == 0x40 ? 4 : 3); + cp_ctx(ctx, 0x403410, dev_priv->chipset == 0x40 ? 4 : 3); + cp_ctx(ctx, 0x403420, nv40_graph_vs_count(ctx->dev)); + for (i = 0; i < nv40_graph_vs_count(ctx->dev); i++) + gr_def(ctx, 0x403420 + (i * 4), 0x00005555); + + if (dev_priv->chipset != 0x40) { + cp_ctx(ctx, 0x403600, 1); + gr_def(ctx, 0x403600, 0x00000001); + } + cp_ctx(ctx, 0x403800, 1); + + cp_ctx(ctx, 0x403c18, 1); + gr_def(ctx, 0x403c18, 0x00000001); + switch (dev_priv->chipset) { + case 0x46: + case 0x47: + case 0x49: + case 0x4b: + cp_ctx(ctx, 0x405018, 1); + gr_def(ctx, 0x405018, 0x08e00001); + cp_ctx(ctx, 0x405c24, 1); + gr_def(ctx, 0x405c24, 0x000e3000); + break; + } + if (dev_priv->chipset != 0x4e) + cp_ctx(ctx, 0x405800, 11); + cp_ctx(ctx, 0x407000, 1); +} + +static void +nv40_graph_construct_state3d_3(struct nouveau_grctx *ctx) +{ + int len = nv44_graph_class(ctx->dev) ? 0x0084 : 0x0684; + + cp_out (ctx, 0x300000); + cp_lsr (ctx, len - 4); + cp_bra (ctx, SWAP_DIRECTION, SAVE, cp_swap_state3d_3_is_save); + cp_lsr (ctx, len); + cp_name(ctx, cp_swap_state3d_3_is_save); + cp_out (ctx, 0x800001); + + ctx->ctxvals_pos += len; +} + +static void +nv40_graph_construct_shader(struct nouveau_grctx *ctx) +{ + struct drm_device *dev = ctx->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *obj = ctx->data; + int vs, vs_nr, vs_len, vs_nr_b0, vs_nr_b1, b0_offset, b1_offset; + int offset, i; + + vs_nr = nv40_graph_vs_count(ctx->dev); + vs_nr_b0 = 363; + vs_nr_b1 = dev_priv->chipset == 0x40 ? 128 : 64; + if (dev_priv->chipset == 0x40) { + b0_offset = 0x2200/4; /* 33a0 */ + b1_offset = 0x55a0/4; /* 1500 */ + vs_len = 0x6aa0/4; + } else + if (dev_priv->chipset == 0x41 || dev_priv->chipset == 0x42) { + b0_offset = 0x2200/4; /* 2200 */ + b1_offset = 0x4400/4; /* 0b00 */ + vs_len = 0x4f00/4; + } else { + b0_offset = 0x1d40/4; /* 2200 */ + b1_offset = 0x3f40/4; /* 0b00 : 0a40 */ + vs_len = nv44_graph_class(dev) ? 0x4980/4 : 0x4a40/4; + } + + cp_lsr(ctx, vs_len * vs_nr + 0x300/4); + cp_out(ctx, nv44_graph_class(dev) ? 0x800029 : 0x800041); + + offset = ctx->ctxvals_pos; + ctx->ctxvals_pos += (0x0300/4 + (vs_nr * vs_len)); + + if (ctx->mode != NOUVEAU_GRCTX_VALS) + return; + + offset += 0x0280/4; + for (i = 0; i < 16; i++, offset += 2) + nv_wo32(obj, offset * 4, 0x3f800000); + + for (vs = 0; vs < vs_nr; vs++, offset += vs_len) { + for (i = 0; i < vs_nr_b0 * 6; i += 6) + nv_wo32(obj, (offset + b0_offset + i) * 4, 0x00000001); + for (i = 0; i < vs_nr_b1 * 4; i += 4) + nv_wo32(obj, (offset + b1_offset + i) * 4, 0x3f800000); + } +} + +void +nv40_grctx_init(struct nouveau_grctx *ctx) +{ + /* decide whether we're loading/unloading the context */ + cp_bra (ctx, AUTO_SAVE, PENDING, cp_setup_save); + cp_bra (ctx, USER_SAVE, PENDING, cp_setup_save); + + cp_name(ctx, cp_check_load); + cp_bra (ctx, AUTO_LOAD, PENDING, cp_setup_auto_load); + cp_bra (ctx, USER_LOAD, PENDING, cp_setup_load); + cp_bra (ctx, ALWAYS, TRUE, cp_exit); + + /* setup for context load */ + cp_name(ctx, cp_setup_auto_load); + cp_wait(ctx, STATUS, IDLE); + cp_out (ctx, CP_NEXT_TO_SWAP); + cp_name(ctx, cp_setup_load); + cp_wait(ctx, STATUS, IDLE); + cp_set (ctx, SWAP_DIRECTION, LOAD); + cp_out (ctx, 0x00910880); /* ?? */ + cp_out (ctx, 0x00901ffe); /* ?? */ + cp_out (ctx, 0x01940000); /* ?? */ + cp_lsr (ctx, 0x20); + cp_out (ctx, 0x0060000b); /* ?? */ + cp_wait(ctx, UNK57, CLEAR); + cp_out (ctx, 0x0060000c); /* ?? */ + cp_bra (ctx, ALWAYS, TRUE, cp_swap_state); + + /* setup for context save */ + cp_name(ctx, cp_setup_save); + cp_set (ctx, SWAP_DIRECTION, SAVE); + + /* general PGRAPH state */ + cp_name(ctx, cp_swap_state); + cp_pos (ctx, 0x00020/4); + nv40_graph_construct_general(ctx); + cp_wait(ctx, STATUS, IDLE); + + /* 3D state, block 1 */ + cp_bra (ctx, UNK54, CLEAR, cp_prepare_exit); + nv40_graph_construct_state3d(ctx); + cp_wait(ctx, STATUS, IDLE); + + /* 3D state, block 2 */ + nv40_graph_construct_state3d_2(ctx); + + /* Some other block of "random" state */ + nv40_graph_construct_state3d_3(ctx); + + /* Per-vertex shader state */ + cp_pos (ctx, ctx->ctxvals_pos); + nv40_graph_construct_shader(ctx); + + /* pre-exit state updates */ + cp_name(ctx, cp_prepare_exit); + cp_bra (ctx, SWAP_DIRECTION, SAVE, cp_check_load); + cp_bra (ctx, USER_SAVE, PENDING, cp_exit); + cp_out (ctx, CP_NEXT_TO_CURRENT); + + cp_name(ctx, cp_exit); + cp_set (ctx, USER_SAVE, NOT_PENDING); + cp_set (ctx, USER_LOAD, NOT_PENDING); + cp_out (ctx, CP_END); +} + diff --git a/drivers/gpu/drm/nouveau/nv40_mc.c b/drivers/gpu/drm/nouveau/nv40_mc.c new file mode 100644 index 00000000..03c0d4c3 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv40_mc.c @@ -0,0 +1,28 @@ +#include "drmP.h" +#include "drm.h" +#include "nouveau_drv.h" +#include "nouveau_drm.h" + +int +nv40_mc_init(struct drm_device *dev) +{ + /* Power up everything, resetting each individual unit will + * be done later if needed. + */ + nv_wr32(dev, NV03_PMC_ENABLE, 0xFFFFFFFF); + + if (nv44_graph_class(dev)) { + u32 tmp = nv_rd32(dev, NV04_PFB_FIFO_DATA); + nv_wr32(dev, NV40_PMC_1700, tmp); + nv_wr32(dev, NV40_PMC_1704, 0); + nv_wr32(dev, NV40_PMC_1708, 0); + nv_wr32(dev, NV40_PMC_170C, tmp); + } + + return 0; +} + +void +nv40_mc_takedown(struct drm_device *dev) +{ +} diff --git a/drivers/gpu/drm/nouveau/nv40_mpeg.c b/drivers/gpu/drm/nouveau/nv40_mpeg.c new file mode 100644 index 00000000..6d2af292 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv40_mpeg.c @@ -0,0 +1,311 @@ +/* + * Copyright 2011 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include "drmP.h" +#include "nouveau_drv.h" +#include "nouveau_ramht.h" + +struct nv40_mpeg_engine { + struct nouveau_exec_engine base; +}; + +static int +nv40_mpeg_context_new(struct nouveau_channel *chan, int engine) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *ctx = NULL; + unsigned long flags; + int ret; + + NV_DEBUG(dev, "ch%d\n", chan->id); + + ret = nouveau_gpuobj_new(dev, NULL, 264 * 4, 16, NVOBJ_FLAG_ZERO_ALLOC | + NVOBJ_FLAG_ZERO_FREE, &ctx); + if (ret) + return ret; + + nv_wo32(ctx, 0x78, 0x02001ec1); + + spin_lock_irqsave(&dev_priv->context_switch_lock, flags); + nv_mask(dev, 0x002500, 0x00000001, 0x00000000); + if ((nv_rd32(dev, 0x003204) & 0x1f) == chan->id) + nv_wr32(dev, 0x00330c, ctx->pinst >> 4); + nv_wo32(chan->ramfc, 0x54, ctx->pinst >> 4); + nv_mask(dev, 0x002500, 0x00000001, 0x00000001); + spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); + + chan->engctx[engine] = ctx; + return 0; +} + +static void +nv40_mpeg_context_del(struct nouveau_channel *chan, int engine) +{ + struct drm_nouveau_private *dev_priv = chan->dev->dev_private; + struct nouveau_gpuobj *ctx = chan->engctx[engine]; + struct drm_device *dev = chan->dev; + unsigned long flags; + u32 inst = 0x80000000 | (ctx->pinst >> 4); + + spin_lock_irqsave(&dev_priv->context_switch_lock, flags); + nv_mask(dev, 0x00b32c, 0x00000001, 0x00000000); + if (nv_rd32(dev, 0x00b318) == inst) + nv_mask(dev, 0x00b318, 0x80000000, 0x00000000); + nv_mask(dev, 0x00b32c, 0x00000001, 0x00000001); + spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); + + nouveau_gpuobj_ref(NULL, &ctx); + chan->engctx[engine] = NULL; +} + +static int +nv40_mpeg_object_new(struct nouveau_channel *chan, int engine, + u32 handle, u16 class) +{ + struct drm_device *dev = chan->dev; + struct nouveau_gpuobj *obj = NULL; + int ret; + + ret = nouveau_gpuobj_new(dev, chan, 20, 16, NVOBJ_FLAG_ZERO_ALLOC | + NVOBJ_FLAG_ZERO_FREE, &obj); + if (ret) + return ret; + obj->engine = 2; + obj->class = class; + + nv_wo32(obj, 0x00, class); + + ret = nouveau_ramht_insert(chan, handle, obj); + nouveau_gpuobj_ref(NULL, &obj); + return ret; +} + +static int +nv40_mpeg_init(struct drm_device *dev, int engine) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nv40_mpeg_engine *pmpeg = nv_engine(dev, engine); + int i; + + /* VPE init */ + nv_mask(dev, 0x000200, 0x00000002, 0x00000000); + nv_mask(dev, 0x000200, 0x00000002, 0x00000002); + nv_wr32(dev, 0x00b0e0, 0x00000020); /* nvidia: rd 0x01, wr 0x20 */ + nv_wr32(dev, 0x00b0e8, 0x00000020); /* nvidia: rd 0x01, wr 0x20 */ + + for (i = 0; i < dev_priv->engine.fb.num_tiles; i++) + pmpeg->base.set_tile_region(dev, i); + + /* PMPEG init */ + nv_wr32(dev, 0x00b32c, 0x00000000); + nv_wr32(dev, 0x00b314, 0x00000100); + nv_wr32(dev, 0x00b220, 0x00000044); + nv_wr32(dev, 0x00b300, 0x02001ec1); + nv_mask(dev, 0x00b32c, 0x00000001, 0x00000001); + + nv_wr32(dev, 0x00b100, 0xffffffff); + nv_wr32(dev, 0x00b140, 0xffffffff); + + if (!nv_wait(dev, 0x00b200, 0x00000001, 0x00000000)) { + NV_ERROR(dev, "PMPEG init: 0x%08x\n", nv_rd32(dev, 0x00b200)); + return -EBUSY; + } + + return 0; +} + +static int +nv40_mpeg_fini(struct drm_device *dev, int engine) +{ + /*XXX: context save? */ + nv_mask(dev, 0x00b32c, 0x00000001, 0x00000000); + nv_wr32(dev, 0x00b140, 0x00000000); + return 0; +} + +static int +nv40_mpeg_mthd_dma(struct nouveau_channel *chan, u32 class, u32 mthd, u32 data) +{ + struct drm_device *dev = chan->dev; + u32 inst = data << 4; + u32 dma0 = nv_ri32(dev, inst + 0); + u32 dma1 = nv_ri32(dev, inst + 4); + u32 dma2 = nv_ri32(dev, inst + 8); + u32 base = (dma2 & 0xfffff000) | (dma0 >> 20); + u32 size = dma1 + 1; + + /* only allow linear DMA objects */ + if (!(dma0 & 0x00002000)) + return -EINVAL; + + if (mthd == 0x0190) { + /* DMA_CMD */ + nv_mask(dev, 0x00b300, 0x00030000, (dma0 & 0x00030000)); + nv_wr32(dev, 0x00b334, base); + nv_wr32(dev, 0x00b324, size); + } else + if (mthd == 0x01a0) { + /* DMA_DATA */ + nv_mask(dev, 0x00b300, 0x000c0000, (dma0 & 0x00030000) << 2); + nv_wr32(dev, 0x00b360, base); + nv_wr32(dev, 0x00b364, size); + } else { + /* DMA_IMAGE, VRAM only */ + if (dma0 & 0x000c0000) + return -EINVAL; + + nv_wr32(dev, 0x00b370, base); + nv_wr32(dev, 0x00b374, size); + } + + return 0; +} + +static int +nv40_mpeg_isr_chid(struct drm_device *dev, u32 inst) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *ctx; + unsigned long flags; + int i; + + spin_lock_irqsave(&dev_priv->channels.lock, flags); + for (i = 0; i < dev_priv->engine.fifo.channels; i++) { + if (!dev_priv->channels.ptr[i]) + continue; + + ctx = dev_priv->channels.ptr[i]->engctx[NVOBJ_ENGINE_MPEG]; + if (ctx && ctx->pinst == inst) + break; + } + spin_unlock_irqrestore(&dev_priv->channels.lock, flags); + return i; +} + +static void +nv40_vpe_set_tile_region(struct drm_device *dev, int i) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_tile_reg *tile = &dev_priv->tile.reg[i]; + + nv_wr32(dev, 0x00b008 + (i * 0x10), tile->pitch); + nv_wr32(dev, 0x00b004 + (i * 0x10), tile->limit); + nv_wr32(dev, 0x00b000 + (i * 0x10), tile->addr); +} + +static void +nv40_mpeg_isr(struct drm_device *dev) +{ + u32 inst = (nv_rd32(dev, 0x00b318) & 0x000fffff) << 4; + u32 chid = nv40_mpeg_isr_chid(dev, inst); + u32 stat = nv_rd32(dev, 0x00b100); + u32 type = nv_rd32(dev, 0x00b230); + u32 mthd = nv_rd32(dev, 0x00b234); + u32 data = nv_rd32(dev, 0x00b238); + u32 show = stat; + + if (stat & 0x01000000) { + /* happens on initial binding of the object */ + if (type == 0x00000020 && mthd == 0x0000) { + nv_mask(dev, 0x00b308, 0x00000000, 0x00000000); + show &= ~0x01000000; + } + + if (type == 0x00000010) { + if (!nouveau_gpuobj_mthd_call2(dev, chid, 0x3174, mthd, data)) + show &= ~0x01000000; + } + } + + nv_wr32(dev, 0x00b100, stat); + nv_wr32(dev, 0x00b230, 0x00000001); + + if (show && nouveau_ratelimit()) { + NV_INFO(dev, "PMPEG: Ch %d [0x%08x] 0x%08x 0x%08x 0x%08x 0x%08x\n", + chid, inst, stat, type, mthd, data); + } +} + +static void +nv40_vpe_isr(struct drm_device *dev) +{ + if (nv_rd32(dev, 0x00b100)) + nv40_mpeg_isr(dev); + + if (nv_rd32(dev, 0x00b800)) { + u32 stat = nv_rd32(dev, 0x00b800); + NV_INFO(dev, "PMSRCH: 0x%08x\n", stat); + nv_wr32(dev, 0xb800, stat); + } +} + +static void +nv40_mpeg_destroy(struct drm_device *dev, int engine) +{ + struct nv40_mpeg_engine *pmpeg = nv_engine(dev, engine); + + nouveau_irq_unregister(dev, 0); + + NVOBJ_ENGINE_DEL(dev, MPEG); + kfree(pmpeg); +} + +int +nv40_mpeg_create(struct drm_device *dev) +{ + struct nv40_mpeg_engine *pmpeg; + + pmpeg = kzalloc(sizeof(*pmpeg), GFP_KERNEL); + if (!pmpeg) + return -ENOMEM; + + pmpeg->base.destroy = nv40_mpeg_destroy; + pmpeg->base.init = nv40_mpeg_init; + pmpeg->base.fini = nv40_mpeg_fini; + pmpeg->base.context_new = nv40_mpeg_context_new; + pmpeg->base.context_del = nv40_mpeg_context_del; + pmpeg->base.object_new = nv40_mpeg_object_new; + + /* ISR vector, PMC_ENABLE bit, and TILE regs are shared between + * all VPE engines, for this driver's purposes the PMPEG engine + * will be treated as the "master" and handle the global VPE + * bits too + */ + pmpeg->base.set_tile_region = nv40_vpe_set_tile_region; + nouveau_irq_register(dev, 0, nv40_vpe_isr); + + NVOBJ_ENGINE_ADD(dev, MPEG, &pmpeg->base); + NVOBJ_CLASS(dev, 0x3174, MPEG); + NVOBJ_MTHD (dev, 0x3174, 0x0190, nv40_mpeg_mthd_dma); + NVOBJ_MTHD (dev, 0x3174, 0x01a0, nv40_mpeg_mthd_dma); + NVOBJ_MTHD (dev, 0x3174, 0x01b0, nv40_mpeg_mthd_dma); + +#if 0 + NVOBJ_ENGINE_ADD(dev, ME, &pme->base); + NVOBJ_CLASS(dev, 0x4075, ME); +#endif + return 0; + +} diff --git a/drivers/gpu/drm/nouveau/nv50_calc.c b/drivers/gpu/drm/nouveau/nv50_calc.c new file mode 100644 index 00000000..8cf63a8b --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv50_calc.c @@ -0,0 +1,97 @@ +/* + * Copyright 2010 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include "drmP.h" +#include "nouveau_drv.h" +#include "nouveau_hw.h" + +int +nv50_calc_pll(struct drm_device *dev, struct pll_lims *pll, int clk, + int *N1, int *M1, int *N2, int *M2, int *P) +{ + struct nouveau_pll_vals pll_vals; + int ret; + + ret = nouveau_calc_pll_mnp(dev, pll, clk, &pll_vals); + if (ret <= 0) + return ret; + + *N1 = pll_vals.N1; + *M1 = pll_vals.M1; + *N2 = pll_vals.N2; + *M2 = pll_vals.M2; + *P = pll_vals.log2P; + return ret; +} + +int +nva3_calc_pll(struct drm_device *dev, struct pll_lims *pll, int clk, + int *pN, int *pfN, int *pM, int *P) +{ + u32 best_err = ~0, err; + int M, lM, hM, N, fN; + + *P = pll->vco1.maxfreq / clk; + if (*P > pll->max_p) + *P = pll->max_p; + if (*P < pll->min_p) + *P = pll->min_p; + + lM = (pll->refclk + pll->vco1.max_inputfreq) / pll->vco1.max_inputfreq; + lM = max(lM, (int)pll->vco1.min_m); + hM = (pll->refclk + pll->vco1.min_inputfreq) / pll->vco1.min_inputfreq; + hM = min(hM, (int)pll->vco1.max_m); + + for (M = lM; M <= hM; M++) { + u32 tmp = clk * *P * M; + N = tmp / pll->refclk; + fN = tmp % pll->refclk; + if (!pfN && fN >= pll->refclk / 2) + N++; + + if (N < pll->vco1.min_n) + continue; + if (N > pll->vco1.max_n) + break; + + err = abs(clk - (pll->refclk * N / M / *P)); + if (err < best_err) { + best_err = err; + *pN = N; + *pM = M; + } + + if (pfN) { + *pfN = (((fN << 13) / pll->refclk) - 4096) & 0xffff; + return clk; + } + } + + if (unlikely(best_err == ~0)) { + NV_ERROR(dev, "unable to find matching pll values\n"); + return -EINVAL; + } + + return pll->refclk * *pN / *pM / *P; +} diff --git a/drivers/gpu/drm/nouveau/nv50_crtc.c b/drivers/gpu/drm/nouveau/nv50_crtc.c new file mode 100644 index 00000000..ebabacf3 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv50_crtc.c @@ -0,0 +1,788 @@ +/* + * Copyright (C) 2008 Maarten Maathuis. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "drmP.h" +#include "drm_mode.h" +#include "drm_crtc_helper.h" + +#define NOUVEAU_DMA_DEBUG (nouveau_reg_debug & NOUVEAU_REG_DEBUG_EVO) +#include "nouveau_reg.h" +#include "nouveau_drv.h" +#include "nouveau_hw.h" +#include "nouveau_encoder.h" +#include "nouveau_crtc.h" +#include "nouveau_fb.h" +#include "nouveau_connector.h" +#include "nv50_display.h" + +static void +nv50_crtc_lut_load(struct drm_crtc *crtc) +{ + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + void __iomem *lut = nvbo_kmap_obj_iovirtual(nv_crtc->lut.nvbo); + int i; + + NV_DEBUG_KMS(crtc->dev, "\n"); + + for (i = 0; i < 256; i++) { + writew(nv_crtc->lut.r[i] >> 2, lut + 8*i + 0); + writew(nv_crtc->lut.g[i] >> 2, lut + 8*i + 2); + writew(nv_crtc->lut.b[i] >> 2, lut + 8*i + 4); + } + + if (nv_crtc->lut.depth == 30) { + writew(nv_crtc->lut.r[i - 1] >> 2, lut + 8*i + 0); + writew(nv_crtc->lut.g[i - 1] >> 2, lut + 8*i + 2); + writew(nv_crtc->lut.b[i - 1] >> 2, lut + 8*i + 4); + } +} + +int +nv50_crtc_blank(struct nouveau_crtc *nv_crtc, bool blanked) +{ + struct drm_device *dev = nv_crtc->base.dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_channel *evo = nv50_display(dev)->master; + int index = nv_crtc->index, ret; + + NV_DEBUG_KMS(dev, "index %d\n", nv_crtc->index); + NV_DEBUG_KMS(dev, "%s\n", blanked ? "blanked" : "unblanked"); + + if (blanked) { + nv_crtc->cursor.hide(nv_crtc, false); + + ret = RING_SPACE(evo, dev_priv->chipset != 0x50 ? 7 : 5); + if (ret) { + NV_ERROR(dev, "no space while blanking crtc\n"); + return ret; + } + BEGIN_RING(evo, 0, NV50_EVO_CRTC(index, CLUT_MODE), 2); + OUT_RING(evo, NV50_EVO_CRTC_CLUT_MODE_BLANK); + OUT_RING(evo, 0); + if (dev_priv->chipset != 0x50) { + BEGIN_RING(evo, 0, NV84_EVO_CRTC(index, CLUT_DMA), 1); + OUT_RING(evo, NV84_EVO_CRTC_CLUT_DMA_HANDLE_NONE); + } + + BEGIN_RING(evo, 0, NV50_EVO_CRTC(index, FB_DMA), 1); + OUT_RING(evo, NV50_EVO_CRTC_FB_DMA_HANDLE_NONE); + } else { + if (nv_crtc->cursor.visible) + nv_crtc->cursor.show(nv_crtc, false); + else + nv_crtc->cursor.hide(nv_crtc, false); + + ret = RING_SPACE(evo, dev_priv->chipset != 0x50 ? 10 : 8); + if (ret) { + NV_ERROR(dev, "no space while unblanking crtc\n"); + return ret; + } + BEGIN_RING(evo, 0, NV50_EVO_CRTC(index, CLUT_MODE), 2); + OUT_RING(evo, nv_crtc->lut.depth == 8 ? + NV50_EVO_CRTC_CLUT_MODE_OFF : + NV50_EVO_CRTC_CLUT_MODE_ON); + OUT_RING(evo, (nv_crtc->lut.nvbo->bo.mem.start << PAGE_SHIFT) >> 8); + if (dev_priv->chipset != 0x50) { + BEGIN_RING(evo, 0, NV84_EVO_CRTC(index, CLUT_DMA), 1); + OUT_RING(evo, NvEvoVRAM); + } + + BEGIN_RING(evo, 0, NV50_EVO_CRTC(index, FB_OFFSET), 2); + OUT_RING(evo, nv_crtc->fb.offset >> 8); + OUT_RING(evo, 0); + BEGIN_RING(evo, 0, NV50_EVO_CRTC(index, FB_DMA), 1); + if (dev_priv->chipset != 0x50) + if (nv_crtc->fb.tile_flags == 0x7a00 || + nv_crtc->fb.tile_flags == 0xfe00) + OUT_RING(evo, NvEvoFB32); + else + if (nv_crtc->fb.tile_flags == 0x7000) + OUT_RING(evo, NvEvoFB16); + else + OUT_RING(evo, NvEvoVRAM_LP); + else + OUT_RING(evo, NvEvoVRAM_LP); + } + + nv_crtc->fb.blanked = blanked; + return 0; +} + +static int +nv50_crtc_set_dither(struct nouveau_crtc *nv_crtc, bool on, bool update) +{ + struct drm_device *dev = nv_crtc->base.dev; + struct nouveau_channel *evo = nv50_display(dev)->master; + int ret; + + NV_DEBUG_KMS(dev, "\n"); + + ret = RING_SPACE(evo, 2 + (update ? 2 : 0)); + if (ret) { + NV_ERROR(dev, "no space while setting dither\n"); + return ret; + } + + BEGIN_RING(evo, 0, NV50_EVO_CRTC(nv_crtc->index, DITHER_CTRL), 1); + if (on) + OUT_RING(evo, NV50_EVO_CRTC_DITHER_CTRL_ON); + else + OUT_RING(evo, NV50_EVO_CRTC_DITHER_CTRL_OFF); + + if (update) { + BEGIN_RING(evo, 0, NV50_EVO_UPDATE, 1); + OUT_RING(evo, 0); + FIRE_RING(evo); + } + + return 0; +} + +struct nouveau_connector * +nouveau_crtc_connector_get(struct nouveau_crtc *nv_crtc) +{ + struct drm_device *dev = nv_crtc->base.dev; + struct drm_connector *connector; + struct drm_crtc *crtc = to_drm_crtc(nv_crtc); + + /* The safest approach is to find an encoder with the right crtc, that + * is also linked to a connector. */ + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + if (connector->encoder) + if (connector->encoder->crtc == crtc) + return nouveau_connector(connector); + } + + return NULL; +} + +static int +nv50_crtc_set_scale(struct nouveau_crtc *nv_crtc, int scaling_mode, bool update) +{ + struct nouveau_connector *nv_connector = + nouveau_crtc_connector_get(nv_crtc); + struct drm_device *dev = nv_crtc->base.dev; + struct nouveau_channel *evo = nv50_display(dev)->master; + struct drm_display_mode *native_mode = NULL; + struct drm_display_mode *mode = &nv_crtc->base.mode; + uint32_t outX, outY, horiz, vert; + int ret; + + NV_DEBUG_KMS(dev, "\n"); + + switch (scaling_mode) { + case DRM_MODE_SCALE_NONE: + break; + default: + if (!nv_connector || !nv_connector->native_mode) { + NV_ERROR(dev, "No native mode, forcing panel scaling\n"); + scaling_mode = DRM_MODE_SCALE_NONE; + } else { + native_mode = nv_connector->native_mode; + } + break; + } + + switch (scaling_mode) { + case DRM_MODE_SCALE_ASPECT: + horiz = (native_mode->hdisplay << 19) / mode->hdisplay; + vert = (native_mode->vdisplay << 19) / mode->vdisplay; + + if (vert > horiz) { + outX = (mode->hdisplay * horiz) >> 19; + outY = (mode->vdisplay * horiz) >> 19; + } else { + outX = (mode->hdisplay * vert) >> 19; + outY = (mode->vdisplay * vert) >> 19; + } + break; + case DRM_MODE_SCALE_FULLSCREEN: + outX = native_mode->hdisplay; + outY = native_mode->vdisplay; + break; + case DRM_MODE_SCALE_CENTER: + case DRM_MODE_SCALE_NONE: + default: + outX = mode->hdisplay; + outY = mode->vdisplay; + break; + } + + ret = RING_SPACE(evo, update ? 7 : 5); + if (ret) + return ret; + + /* Got a better name for SCALER_ACTIVE? */ + /* One day i've got to really figure out why this is needed. */ + BEGIN_RING(evo, 0, NV50_EVO_CRTC(nv_crtc->index, SCALE_CTRL), 1); + if ((mode->flags & DRM_MODE_FLAG_DBLSCAN) || + (mode->flags & DRM_MODE_FLAG_INTERLACE) || + mode->hdisplay != outX || mode->vdisplay != outY) { + OUT_RING(evo, NV50_EVO_CRTC_SCALE_CTRL_ACTIVE); + } else { + OUT_RING(evo, NV50_EVO_CRTC_SCALE_CTRL_INACTIVE); + } + + BEGIN_RING(evo, 0, NV50_EVO_CRTC(nv_crtc->index, SCALE_RES1), 2); + OUT_RING(evo, outY << 16 | outX); + OUT_RING(evo, outY << 16 | outX); + + if (update) { + BEGIN_RING(evo, 0, NV50_EVO_UPDATE, 1); + OUT_RING(evo, 0); + FIRE_RING(evo); + } + + return 0; +} + +int +nv50_crtc_set_clock(struct drm_device *dev, int head, int pclk) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct pll_lims pll; + uint32_t reg1, reg2; + int ret, N1, M1, N2, M2, P; + + ret = get_pll_limits(dev, PLL_VPLL0 + head, &pll); + if (ret) + return ret; + + if (pll.vco2.maxfreq) { + ret = nv50_calc_pll(dev, &pll, pclk, &N1, &M1, &N2, &M2, &P); + if (ret <= 0) + return 0; + + NV_DEBUG(dev, "pclk %d out %d NM1 %d %d NM2 %d %d P %d\n", + pclk, ret, N1, M1, N2, M2, P); + + reg1 = nv_rd32(dev, pll.reg + 4) & 0xff00ff00; + reg2 = nv_rd32(dev, pll.reg + 8) & 0x8000ff00; + nv_wr32(dev, pll.reg + 0, 0x10000611); + nv_wr32(dev, pll.reg + 4, reg1 | (M1 << 16) | N1); + nv_wr32(dev, pll.reg + 8, reg2 | (P << 28) | (M2 << 16) | N2); + } else + if (dev_priv->chipset < NV_C0) { + ret = nva3_calc_pll(dev, &pll, pclk, &N1, &N2, &M1, &P); + if (ret <= 0) + return 0; + + NV_DEBUG(dev, "pclk %d out %d N %d fN 0x%04x M %d P %d\n", + pclk, ret, N1, N2, M1, P); + + reg1 = nv_rd32(dev, pll.reg + 4) & 0xffc00000; + nv_wr32(dev, pll.reg + 0, 0x50000610); + nv_wr32(dev, pll.reg + 4, reg1 | (P << 16) | (M1 << 8) | N1); + nv_wr32(dev, pll.reg + 8, N2); + } else { + ret = nva3_calc_pll(dev, &pll, pclk, &N1, &N2, &M1, &P); + if (ret <= 0) + return 0; + + NV_DEBUG(dev, "pclk %d out %d N %d fN 0x%04x M %d P %d\n", + pclk, ret, N1, N2, M1, P); + + nv_mask(dev, pll.reg + 0x0c, 0x00000000, 0x00000100); + nv_wr32(dev, pll.reg + 0x04, (P << 16) | (N1 << 8) | M1); + nv_wr32(dev, pll.reg + 0x10, N2 << 16); + } + + return 0; +} + +static void +nv50_crtc_destroy(struct drm_crtc *crtc) +{ + struct drm_device *dev; + struct nouveau_crtc *nv_crtc; + + if (!crtc) + return; + + dev = crtc->dev; + nv_crtc = nouveau_crtc(crtc); + + NV_DEBUG_KMS(dev, "\n"); + + drm_crtc_cleanup(&nv_crtc->base); + + nv50_cursor_fini(nv_crtc); + + nouveau_bo_unmap(nv_crtc->lut.nvbo); + nouveau_bo_ref(NULL, &nv_crtc->lut.nvbo); + nouveau_bo_unmap(nv_crtc->cursor.nvbo); + nouveau_bo_ref(NULL, &nv_crtc->cursor.nvbo); + kfree(nv_crtc->mode); + kfree(nv_crtc); +} + +int +nv50_crtc_cursor_set(struct drm_crtc *crtc, struct drm_file *file_priv, + uint32_t buffer_handle, uint32_t width, uint32_t height) +{ + struct drm_device *dev = crtc->dev; + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + struct nouveau_bo *cursor = NULL; + struct drm_gem_object *gem; + int ret = 0, i; + + if (!buffer_handle) { + nv_crtc->cursor.hide(nv_crtc, true); + return 0; + } + + if (width != 64 || height != 64) + return -EINVAL; + + gem = drm_gem_object_lookup(dev, file_priv, buffer_handle); + if (!gem) + return -ENOENT; + cursor = nouveau_gem_object(gem); + + ret = nouveau_bo_map(cursor); + if (ret) + goto out; + + /* The simple will do for now. */ + for (i = 0; i < 64 * 64; i++) + nouveau_bo_wr32(nv_crtc->cursor.nvbo, i, nouveau_bo_rd32(cursor, i)); + + nouveau_bo_unmap(cursor); + + nv_crtc->cursor.set_offset(nv_crtc, nv_crtc->cursor.nvbo->bo.mem.start << PAGE_SHIFT); + nv_crtc->cursor.show(nv_crtc, true); + +out: + drm_gem_object_unreference_unlocked(gem); + return ret; +} + +int +nv50_crtc_cursor_move(struct drm_crtc *crtc, int x, int y) +{ + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + + nv_crtc->cursor.set_pos(nv_crtc, x, y); + return 0; +} + +static void +nv50_crtc_gamma_set(struct drm_crtc *crtc, u16 *r, u16 *g, u16 *b, + uint32_t start, uint32_t size) +{ + int end = (start + size > 256) ? 256 : start + size, i; + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + + for (i = start; i < end; i++) { + nv_crtc->lut.r[i] = r[i]; + nv_crtc->lut.g[i] = g[i]; + nv_crtc->lut.b[i] = b[i]; + } + + /* We need to know the depth before we upload, but it's possible to + * get called before a framebuffer is bound. If this is the case, + * mark the lut values as dirty by setting depth==0, and it'll be + * uploaded on the first mode_set_base() + */ + if (!nv_crtc->base.fb) { + nv_crtc->lut.depth = 0; + return; + } + + nv50_crtc_lut_load(crtc); +} + +static void +nv50_crtc_save(struct drm_crtc *crtc) +{ + NV_ERROR(crtc->dev, "!!\n"); +} + +static void +nv50_crtc_restore(struct drm_crtc *crtc) +{ + NV_ERROR(crtc->dev, "!!\n"); +} + +static const struct drm_crtc_funcs nv50_crtc_funcs = { + .save = nv50_crtc_save, + .restore = nv50_crtc_restore, + .cursor_set = nv50_crtc_cursor_set, + .cursor_move = nv50_crtc_cursor_move, + .gamma_set = nv50_crtc_gamma_set, + .set_config = drm_crtc_helper_set_config, + .page_flip = nouveau_crtc_page_flip, + .destroy = nv50_crtc_destroy, +}; + +static void +nv50_crtc_dpms(struct drm_crtc *crtc, int mode) +{ +} + +static int +nv50_crtc_wait_complete(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_timer_engine *ptimer = &dev_priv->engine.timer; + struct nv50_display *disp = nv50_display(dev); + struct nouveau_channel *evo = disp->master; + u64 start; + int ret; + + ret = RING_SPACE(evo, 6); + if (ret) + return ret; + BEGIN_RING(evo, 0, 0x0084, 1); + OUT_RING (evo, 0x80000000); + BEGIN_RING(evo, 0, 0x0080, 1); + OUT_RING (evo, 0); + BEGIN_RING(evo, 0, 0x0084, 1); + OUT_RING (evo, 0x00000000); + + nv_wo32(disp->ntfy, 0x000, 0x00000000); + FIRE_RING (evo); + + start = ptimer->read(dev); + do { + if (nv_ro32(disp->ntfy, 0x000)) + return 0; + } while (ptimer->read(dev) - start < 2000000000ULL); + + return -EBUSY; +} + +static void +nv50_crtc_prepare(struct drm_crtc *crtc) +{ + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + struct drm_device *dev = crtc->dev; + + NV_DEBUG_KMS(dev, "index %d\n", nv_crtc->index); + + nv50_display_flip_stop(crtc); + drm_vblank_pre_modeset(dev, nv_crtc->index); + nv50_crtc_blank(nv_crtc, true); +} + +static void +nv50_crtc_commit(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + + NV_DEBUG_KMS(dev, "index %d\n", nv_crtc->index); + + nv50_crtc_blank(nv_crtc, false); + drm_vblank_post_modeset(dev, nv_crtc->index); + nv50_crtc_wait_complete(crtc); + nv50_display_flip_next(crtc, crtc->fb, NULL); +} + +static bool +nv50_crtc_mode_fixup(struct drm_crtc *crtc, struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + return true; +} + +static int +nv50_crtc_do_mode_set_base(struct drm_crtc *crtc, + struct drm_framebuffer *passed_fb, + int x, int y, bool atomic) +{ + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + struct drm_device *dev = nv_crtc->base.dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_channel *evo = nv50_display(dev)->master; + struct drm_framebuffer *drm_fb = nv_crtc->base.fb; + struct nouveau_framebuffer *fb = nouveau_framebuffer(drm_fb); + int ret; + + NV_DEBUG_KMS(dev, "index %d\n", nv_crtc->index); + + /* If atomic, we want to switch to the fb we were passed, so + * now we update pointers to do that. (We don't pin; just + * assume we're already pinned and update the base address.) + */ + if (atomic) { + drm_fb = passed_fb; + fb = nouveau_framebuffer(passed_fb); + } else { + /* If not atomic, we can go ahead and pin, and unpin the + * old fb we were passed. + */ + ret = nouveau_bo_pin(fb->nvbo, TTM_PL_FLAG_VRAM); + if (ret) + return ret; + + if (passed_fb) { + struct nouveau_framebuffer *ofb = nouveau_framebuffer(passed_fb); + nouveau_bo_unpin(ofb->nvbo); + } + } + + nv_crtc->fb.offset = fb->nvbo->bo.mem.start << PAGE_SHIFT; + nv_crtc->fb.tile_flags = nouveau_bo_tile_layout(fb->nvbo); + nv_crtc->fb.cpp = drm_fb->bits_per_pixel / 8; + if (!nv_crtc->fb.blanked && dev_priv->chipset != 0x50) { + ret = RING_SPACE(evo, 2); + if (ret) + return ret; + + BEGIN_RING(evo, 0, NV50_EVO_CRTC(nv_crtc->index, FB_DMA), 1); + OUT_RING (evo, fb->r_dma); + } + + ret = RING_SPACE(evo, 12); + if (ret) + return ret; + + BEGIN_RING(evo, 0, NV50_EVO_CRTC(nv_crtc->index, FB_OFFSET), 5); + OUT_RING (evo, nv_crtc->fb.offset >> 8); + OUT_RING (evo, 0); + OUT_RING (evo, (drm_fb->height << 16) | drm_fb->width); + OUT_RING (evo, fb->r_pitch); + OUT_RING (evo, fb->r_format); + + BEGIN_RING(evo, 0, NV50_EVO_CRTC(nv_crtc->index, CLUT_MODE), 1); + OUT_RING (evo, fb->base.depth == 8 ? + NV50_EVO_CRTC_CLUT_MODE_OFF : NV50_EVO_CRTC_CLUT_MODE_ON); + + BEGIN_RING(evo, 0, NV50_EVO_CRTC(nv_crtc->index, COLOR_CTRL), 1); + OUT_RING (evo, NV50_EVO_CRTC_COLOR_CTRL_COLOR); + BEGIN_RING(evo, 0, NV50_EVO_CRTC(nv_crtc->index, FB_POS), 1); + OUT_RING (evo, (y << 16) | x); + + if (nv_crtc->lut.depth != fb->base.depth) { + nv_crtc->lut.depth = fb->base.depth; + nv50_crtc_lut_load(crtc); + } + + return 0; +} + +static int +nv50_crtc_mode_set(struct drm_crtc *crtc, struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode, int x, int y, + struct drm_framebuffer *old_fb) +{ + struct drm_device *dev = crtc->dev; + struct nouveau_channel *evo = nv50_display(dev)->master; + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + struct nouveau_connector *nv_connector = NULL; + uint32_t hsync_dur, vsync_dur, hsync_start_to_end, vsync_start_to_end; + uint32_t hunk1, vunk1, vunk2a, vunk2b; + int ret; + + /* Find the connector attached to this CRTC */ + nv_connector = nouveau_crtc_connector_get(nv_crtc); + + *nv_crtc->mode = *adjusted_mode; + + NV_DEBUG_KMS(dev, "index %d\n", nv_crtc->index); + + hsync_dur = adjusted_mode->hsync_end - adjusted_mode->hsync_start; + vsync_dur = adjusted_mode->vsync_end - adjusted_mode->vsync_start; + hsync_start_to_end = adjusted_mode->htotal - adjusted_mode->hsync_start; + vsync_start_to_end = adjusted_mode->vtotal - adjusted_mode->vsync_start; + /* I can't give this a proper name, anyone else can? */ + hunk1 = adjusted_mode->htotal - + adjusted_mode->hsync_start + adjusted_mode->hdisplay; + vunk1 = adjusted_mode->vtotal - + adjusted_mode->vsync_start + adjusted_mode->vdisplay; + /* Another strange value, this time only for interlaced adjusted_modes. */ + vunk2a = 2 * adjusted_mode->vtotal - + adjusted_mode->vsync_start + adjusted_mode->vdisplay; + vunk2b = adjusted_mode->vtotal - + adjusted_mode->vsync_start + adjusted_mode->vtotal; + + if (adjusted_mode->flags & DRM_MODE_FLAG_INTERLACE) { + vsync_dur /= 2; + vsync_start_to_end /= 2; + vunk1 /= 2; + vunk2a /= 2; + vunk2b /= 2; + /* magic */ + if (adjusted_mode->flags & DRM_MODE_FLAG_DBLSCAN) { + vsync_start_to_end -= 1; + vunk1 -= 1; + vunk2a -= 1; + vunk2b -= 1; + } + } + + ret = RING_SPACE(evo, 17); + if (ret) + return ret; + + BEGIN_RING(evo, 0, NV50_EVO_CRTC(nv_crtc->index, CLOCK), 2); + OUT_RING(evo, adjusted_mode->clock | 0x800000); + OUT_RING(evo, (adjusted_mode->flags & DRM_MODE_FLAG_INTERLACE) ? 2 : 0); + + BEGIN_RING(evo, 0, NV50_EVO_CRTC(nv_crtc->index, DISPLAY_START), 5); + OUT_RING(evo, 0); + OUT_RING(evo, (adjusted_mode->vtotal << 16) | adjusted_mode->htotal); + OUT_RING(evo, (vsync_dur - 1) << 16 | (hsync_dur - 1)); + OUT_RING(evo, (vsync_start_to_end - 1) << 16 | + (hsync_start_to_end - 1)); + OUT_RING(evo, (vunk1 - 1) << 16 | (hunk1 - 1)); + + if (adjusted_mode->flags & DRM_MODE_FLAG_INTERLACE) { + BEGIN_RING(evo, 0, NV50_EVO_CRTC(nv_crtc->index, UNK0824), 1); + OUT_RING(evo, (vunk2b - 1) << 16 | (vunk2a - 1)); + } else { + OUT_RING(evo, 0); + OUT_RING(evo, 0); + } + + BEGIN_RING(evo, 0, NV50_EVO_CRTC(nv_crtc->index, UNK082C), 1); + OUT_RING(evo, 0); + + /* This is the actual resolution of the mode. */ + BEGIN_RING(evo, 0, NV50_EVO_CRTC(nv_crtc->index, REAL_RES), 1); + OUT_RING(evo, (mode->vdisplay << 16) | mode->hdisplay); + BEGIN_RING(evo, 0, NV50_EVO_CRTC(nv_crtc->index, SCALE_CENTER_OFFSET), 1); + OUT_RING(evo, NV50_EVO_CRTC_SCALE_CENTER_OFFSET_VAL(0, 0)); + + nv_crtc->set_dither(nv_crtc, nv_connector->use_dithering, false); + nv_crtc->set_scale(nv_crtc, nv_connector->scaling_mode, false); + + return nv50_crtc_do_mode_set_base(crtc, old_fb, x, y, false); +} + +static int +nv50_crtc_mode_set_base(struct drm_crtc *crtc, int x, int y, + struct drm_framebuffer *old_fb) +{ + int ret; + + nv50_display_flip_stop(crtc); + ret = nv50_crtc_do_mode_set_base(crtc, old_fb, x, y, false); + if (ret) + return ret; + + ret = nv50_crtc_wait_complete(crtc); + if (ret) + return ret; + + return nv50_display_flip_next(crtc, crtc->fb, NULL); +} + +static int +nv50_crtc_mode_set_base_atomic(struct drm_crtc *crtc, + struct drm_framebuffer *fb, + int x, int y, enum mode_set_atomic state) +{ + int ret; + + nv50_display_flip_stop(crtc); + ret = nv50_crtc_do_mode_set_base(crtc, fb, x, y, true); + if (ret) + return ret; + + return nv50_crtc_wait_complete(crtc); +} + +static const struct drm_crtc_helper_funcs nv50_crtc_helper_funcs = { + .dpms = nv50_crtc_dpms, + .prepare = nv50_crtc_prepare, + .commit = nv50_crtc_commit, + .mode_fixup = nv50_crtc_mode_fixup, + .mode_set = nv50_crtc_mode_set, + .mode_set_base = nv50_crtc_mode_set_base, + .mode_set_base_atomic = nv50_crtc_mode_set_base_atomic, + .load_lut = nv50_crtc_lut_load, +}; + +int +nv50_crtc_create(struct drm_device *dev, int index) +{ + struct nouveau_crtc *nv_crtc = NULL; + int ret, i; + + NV_DEBUG_KMS(dev, "\n"); + + nv_crtc = kzalloc(sizeof(*nv_crtc), GFP_KERNEL); + if (!nv_crtc) + return -ENOMEM; + + nv_crtc->mode = kzalloc(sizeof(*nv_crtc->mode), GFP_KERNEL); + if (!nv_crtc->mode) { + kfree(nv_crtc); + return -ENOMEM; + } + + /* Default CLUT parameters, will be activated on the hw upon + * first mode set. + */ + for (i = 0; i < 256; i++) { + nv_crtc->lut.r[i] = i << 8; + nv_crtc->lut.g[i] = i << 8; + nv_crtc->lut.b[i] = i << 8; + } + nv_crtc->lut.depth = 0; + + ret = nouveau_bo_new(dev, NULL, 4096, 0x100, TTM_PL_FLAG_VRAM, + 0, 0x0000, &nv_crtc->lut.nvbo); + if (!ret) { + ret = nouveau_bo_pin(nv_crtc->lut.nvbo, TTM_PL_FLAG_VRAM); + if (!ret) + ret = nouveau_bo_map(nv_crtc->lut.nvbo); + if (ret) + nouveau_bo_ref(NULL, &nv_crtc->lut.nvbo); + } + + if (ret) { + kfree(nv_crtc->mode); + kfree(nv_crtc); + return ret; + } + + nv_crtc->index = index; + + /* set function pointers */ + nv_crtc->set_dither = nv50_crtc_set_dither; + nv_crtc->set_scale = nv50_crtc_set_scale; + + drm_crtc_init(dev, &nv_crtc->base, &nv50_crtc_funcs); + drm_crtc_helper_add(&nv_crtc->base, &nv50_crtc_helper_funcs); + drm_mode_crtc_set_gamma_size(&nv_crtc->base, 256); + + ret = nouveau_bo_new(dev, NULL, 64*64*4, 0x100, TTM_PL_FLAG_VRAM, + 0, 0x0000, &nv_crtc->cursor.nvbo); + if (!ret) { + ret = nouveau_bo_pin(nv_crtc->cursor.nvbo, TTM_PL_FLAG_VRAM); + if (!ret) + ret = nouveau_bo_map(nv_crtc->cursor.nvbo); + if (ret) + nouveau_bo_ref(NULL, &nv_crtc->cursor.nvbo); + } + + nv50_cursor_init(nv_crtc); + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nv50_cursor.c b/drivers/gpu/drm/nouveau/nv50_cursor.c new file mode 100644 index 00000000..9752c35b --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv50_cursor.c @@ -0,0 +1,157 @@ +/* + * Copyright (C) 2008 Maarten Maathuis. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "drmP.h" +#include "drm_mode.h" + +#define NOUVEAU_DMA_DEBUG (nouveau_reg_debug & NOUVEAU_REG_DEBUG_EVO) +#include "nouveau_reg.h" +#include "nouveau_drv.h" +#include "nouveau_crtc.h" +#include "nv50_display.h" + +static void +nv50_cursor_show(struct nouveau_crtc *nv_crtc, bool update) +{ + struct drm_device *dev = nv_crtc->base.dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_channel *evo = nv50_display(dev)->master; + int ret; + + NV_DEBUG_KMS(dev, "\n"); + + if (update && nv_crtc->cursor.visible) + return; + + ret = RING_SPACE(evo, (dev_priv->chipset != 0x50 ? 5 : 3) + update * 2); + if (ret) { + NV_ERROR(dev, "no space while unhiding cursor\n"); + return; + } + + if (dev_priv->chipset != 0x50) { + BEGIN_RING(evo, 0, NV84_EVO_CRTC(nv_crtc->index, CURSOR_DMA), 1); + OUT_RING(evo, NvEvoVRAM); + } + BEGIN_RING(evo, 0, NV50_EVO_CRTC(nv_crtc->index, CURSOR_CTRL), 2); + OUT_RING(evo, NV50_EVO_CRTC_CURSOR_CTRL_SHOW); + OUT_RING(evo, nv_crtc->cursor.offset >> 8); + + if (update) { + BEGIN_RING(evo, 0, NV50_EVO_UPDATE, 1); + OUT_RING(evo, 0); + FIRE_RING(evo); + nv_crtc->cursor.visible = true; + } +} + +static void +nv50_cursor_hide(struct nouveau_crtc *nv_crtc, bool update) +{ + struct drm_device *dev = nv_crtc->base.dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_channel *evo = nv50_display(dev)->master; + int ret; + + NV_DEBUG_KMS(dev, "\n"); + + if (update && !nv_crtc->cursor.visible) + return; + + ret = RING_SPACE(evo, (dev_priv->chipset != 0x50 ? 5 : 3) + update * 2); + if (ret) { + NV_ERROR(dev, "no space while hiding cursor\n"); + return; + } + BEGIN_RING(evo, 0, NV50_EVO_CRTC(nv_crtc->index, CURSOR_CTRL), 2); + OUT_RING(evo, NV50_EVO_CRTC_CURSOR_CTRL_HIDE); + OUT_RING(evo, 0); + if (dev_priv->chipset != 0x50) { + BEGIN_RING(evo, 0, NV84_EVO_CRTC(nv_crtc->index, CURSOR_DMA), 1); + OUT_RING(evo, NV84_EVO_CRTC_CURSOR_DMA_HANDLE_NONE); + } + + if (update) { + BEGIN_RING(evo, 0, NV50_EVO_UPDATE, 1); + OUT_RING(evo, 0); + FIRE_RING(evo); + nv_crtc->cursor.visible = false; + } +} + +static void +nv50_cursor_set_pos(struct nouveau_crtc *nv_crtc, int x, int y) +{ + struct drm_device *dev = nv_crtc->base.dev; + + nv_crtc->cursor_saved_x = x; nv_crtc->cursor_saved_y = y; + nv_wr32(dev, NV50_PDISPLAY_CURSOR_USER_POS(nv_crtc->index), + ((y & 0xFFFF) << 16) | (x & 0xFFFF)); + /* Needed to make the cursor move. */ + nv_wr32(dev, NV50_PDISPLAY_CURSOR_USER_POS_CTRL(nv_crtc->index), 0); +} + +static void +nv50_cursor_set_offset(struct nouveau_crtc *nv_crtc, uint32_t offset) +{ + NV_DEBUG_KMS(nv_crtc->base.dev, "\n"); + if (offset == nv_crtc->cursor.offset) + return; + + nv_crtc->cursor.offset = offset; + if (nv_crtc->cursor.visible) { + nv_crtc->cursor.visible = false; + nv_crtc->cursor.show(nv_crtc, true); + } +} + +int +nv50_cursor_init(struct nouveau_crtc *nv_crtc) +{ + nv_crtc->cursor.set_offset = nv50_cursor_set_offset; + nv_crtc->cursor.set_pos = nv50_cursor_set_pos; + nv_crtc->cursor.hide = nv50_cursor_hide; + nv_crtc->cursor.show = nv50_cursor_show; + return 0; +} + +void +nv50_cursor_fini(struct nouveau_crtc *nv_crtc) +{ + struct drm_device *dev = nv_crtc->base.dev; + int idx = nv_crtc->index; + + NV_DEBUG_KMS(dev, "\n"); + + nv_wr32(dev, NV50_PDISPLAY_CURSOR_CURSOR_CTRL2(idx), 0); + if (!nv_wait(dev, NV50_PDISPLAY_CURSOR_CURSOR_CTRL2(idx), + NV50_PDISPLAY_CURSOR_CURSOR_CTRL2_STATUS, 0)) { + NV_ERROR(dev, "timeout: CURSOR_CTRL2_STATUS == 0\n"); + NV_ERROR(dev, "CURSOR_CTRL2 = 0x%08x\n", + nv_rd32(dev, NV50_PDISPLAY_CURSOR_CURSOR_CTRL2(idx))); + } +} + diff --git a/drivers/gpu/drm/nouveau/nv50_dac.c b/drivers/gpu/drm/nouveau/nv50_dac.c new file mode 100644 index 00000000..808f3ec8 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv50_dac.c @@ -0,0 +1,319 @@ +/* + * Copyright (C) 2008 Maarten Maathuis. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "drmP.h" +#include "drm_crtc_helper.h" + +#define NOUVEAU_DMA_DEBUG (nouveau_reg_debug & NOUVEAU_REG_DEBUG_EVO) +#include "nouveau_reg.h" +#include "nouveau_drv.h" +#include "nouveau_dma.h" +#include "nouveau_encoder.h" +#include "nouveau_connector.h" +#include "nouveau_crtc.h" +#include "nv50_display.h" + +static void +nv50_dac_disconnect(struct drm_encoder *encoder) +{ + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + struct drm_device *dev = encoder->dev; + struct nouveau_channel *evo = nv50_display(dev)->master; + int ret; + + if (!nv_encoder->crtc) + return; + nv50_crtc_blank(nouveau_crtc(nv_encoder->crtc), true); + + NV_DEBUG_KMS(dev, "Disconnecting DAC %d\n", nv_encoder->or); + + ret = RING_SPACE(evo, 4); + if (ret) { + NV_ERROR(dev, "no space while disconnecting DAC\n"); + return; + } + BEGIN_RING(evo, 0, NV50_EVO_DAC(nv_encoder->or, MODE_CTRL), 1); + OUT_RING (evo, 0); + BEGIN_RING(evo, 0, NV50_EVO_UPDATE, 1); + OUT_RING (evo, 0); + + nv_encoder->crtc = NULL; +} + +static enum drm_connector_status +nv50_dac_detect(struct drm_encoder *encoder, struct drm_connector *connector) +{ + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + struct drm_device *dev = encoder->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + enum drm_connector_status status = connector_status_disconnected; + uint32_t dpms_state, load_pattern, load_state; + int or = nv_encoder->or; + + nv_wr32(dev, NV50_PDISPLAY_DAC_CLK_CTRL1(or), 0x00000001); + dpms_state = nv_rd32(dev, NV50_PDISPLAY_DAC_DPMS_CTRL(or)); + + nv_wr32(dev, NV50_PDISPLAY_DAC_DPMS_CTRL(or), + 0x00150000 | NV50_PDISPLAY_DAC_DPMS_CTRL_PENDING); + if (!nv_wait(dev, NV50_PDISPLAY_DAC_DPMS_CTRL(or), + NV50_PDISPLAY_DAC_DPMS_CTRL_PENDING, 0)) { + NV_ERROR(dev, "timeout: DAC_DPMS_CTRL_PENDING(%d) == 0\n", or); + NV_ERROR(dev, "DAC_DPMS_CTRL(%d) = 0x%08x\n", or, + nv_rd32(dev, NV50_PDISPLAY_DAC_DPMS_CTRL(or))); + return status; + } + + /* Use bios provided value if possible. */ + if (dev_priv->vbios.dactestval) { + load_pattern = dev_priv->vbios.dactestval; + NV_DEBUG_KMS(dev, "Using bios provided load_pattern of %d\n", + load_pattern); + } else { + load_pattern = 340; + NV_DEBUG_KMS(dev, "Using default load_pattern of %d\n", + load_pattern); + } + + nv_wr32(dev, NV50_PDISPLAY_DAC_LOAD_CTRL(or), + NV50_PDISPLAY_DAC_LOAD_CTRL_ACTIVE | load_pattern); + mdelay(45); /* give it some time to process */ + load_state = nv_rd32(dev, NV50_PDISPLAY_DAC_LOAD_CTRL(or)); + + nv_wr32(dev, NV50_PDISPLAY_DAC_LOAD_CTRL(or), 0); + nv_wr32(dev, NV50_PDISPLAY_DAC_DPMS_CTRL(or), dpms_state | + NV50_PDISPLAY_DAC_DPMS_CTRL_PENDING); + + if ((load_state & NV50_PDISPLAY_DAC_LOAD_CTRL_PRESENT) == + NV50_PDISPLAY_DAC_LOAD_CTRL_PRESENT) + status = connector_status_connected; + + if (status == connector_status_connected) + NV_DEBUG_KMS(dev, "Load was detected on output with or %d\n", or); + else + NV_DEBUG_KMS(dev, "Load was not detected on output with or %d\n", or); + + return status; +} + +static void +nv50_dac_dpms(struct drm_encoder *encoder, int mode) +{ + struct drm_device *dev = encoder->dev; + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + uint32_t val; + int or = nv_encoder->or; + + NV_DEBUG_KMS(dev, "or %d mode %d\n", or, mode); + + /* wait for it to be done */ + if (!nv_wait(dev, NV50_PDISPLAY_DAC_DPMS_CTRL(or), + NV50_PDISPLAY_DAC_DPMS_CTRL_PENDING, 0)) { + NV_ERROR(dev, "timeout: DAC_DPMS_CTRL_PENDING(%d) == 0\n", or); + NV_ERROR(dev, "DAC_DPMS_CTRL(%d) = 0x%08x\n", or, + nv_rd32(dev, NV50_PDISPLAY_DAC_DPMS_CTRL(or))); + return; + } + + val = nv_rd32(dev, NV50_PDISPLAY_DAC_DPMS_CTRL(or)) & ~0x7F; + + if (mode != DRM_MODE_DPMS_ON) + val |= NV50_PDISPLAY_DAC_DPMS_CTRL_BLANKED; + + switch (mode) { + case DRM_MODE_DPMS_STANDBY: + val |= NV50_PDISPLAY_DAC_DPMS_CTRL_HSYNC_OFF; + break; + case DRM_MODE_DPMS_SUSPEND: + val |= NV50_PDISPLAY_DAC_DPMS_CTRL_VSYNC_OFF; + break; + case DRM_MODE_DPMS_OFF: + val |= NV50_PDISPLAY_DAC_DPMS_CTRL_OFF; + val |= NV50_PDISPLAY_DAC_DPMS_CTRL_HSYNC_OFF; + val |= NV50_PDISPLAY_DAC_DPMS_CTRL_VSYNC_OFF; + break; + default: + break; + } + + nv_wr32(dev, NV50_PDISPLAY_DAC_DPMS_CTRL(or), val | + NV50_PDISPLAY_DAC_DPMS_CTRL_PENDING); +} + +static void +nv50_dac_save(struct drm_encoder *encoder) +{ + NV_ERROR(encoder->dev, "!!\n"); +} + +static void +nv50_dac_restore(struct drm_encoder *encoder) +{ + NV_ERROR(encoder->dev, "!!\n"); +} + +static bool +nv50_dac_mode_fixup(struct drm_encoder *encoder, struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + struct nouveau_connector *connector; + + NV_DEBUG_KMS(encoder->dev, "or %d\n", nv_encoder->or); + + connector = nouveau_encoder_connector_get(nv_encoder); + if (!connector) { + NV_ERROR(encoder->dev, "Encoder has no connector\n"); + return false; + } + + if (connector->scaling_mode != DRM_MODE_SCALE_NONE && + connector->native_mode) { + int id = adjusted_mode->base.id; + *adjusted_mode = *connector->native_mode; + adjusted_mode->base.id = id; + } + + return true; +} + +static void +nv50_dac_prepare(struct drm_encoder *encoder) +{ +} + +static void +nv50_dac_commit(struct drm_encoder *encoder) +{ +} + +static void +nv50_dac_mode_set(struct drm_encoder *encoder, struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + struct drm_device *dev = encoder->dev; + struct nouveau_channel *evo = nv50_display(dev)->master; + struct nouveau_crtc *crtc = nouveau_crtc(encoder->crtc); + uint32_t mode_ctl = 0, mode_ctl2 = 0; + int ret; + + NV_DEBUG_KMS(dev, "or %d type %d crtc %d\n", + nv_encoder->or, nv_encoder->dcb->type, crtc->index); + + nv50_dac_dpms(encoder, DRM_MODE_DPMS_ON); + + if (crtc->index == 1) + mode_ctl |= NV50_EVO_DAC_MODE_CTRL_CRTC1; + else + mode_ctl |= NV50_EVO_DAC_MODE_CTRL_CRTC0; + + /* Lacking a working tv-out, this is not a 100% sure. */ + if (nv_encoder->dcb->type == OUTPUT_ANALOG) + mode_ctl |= 0x40; + else + if (nv_encoder->dcb->type == OUTPUT_TV) + mode_ctl |= 0x100; + + if (adjusted_mode->flags & DRM_MODE_FLAG_NHSYNC) + mode_ctl2 |= NV50_EVO_DAC_MODE_CTRL2_NHSYNC; + + if (adjusted_mode->flags & DRM_MODE_FLAG_NVSYNC) + mode_ctl2 |= NV50_EVO_DAC_MODE_CTRL2_NVSYNC; + + ret = RING_SPACE(evo, 3); + if (ret) { + NV_ERROR(dev, "no space while connecting DAC\n"); + return; + } + BEGIN_RING(evo, 0, NV50_EVO_DAC(nv_encoder->or, MODE_CTRL), 2); + OUT_RING(evo, mode_ctl); + OUT_RING(evo, mode_ctl2); + + nv_encoder->crtc = encoder->crtc; +} + +static struct drm_crtc * +nv50_dac_crtc_get(struct drm_encoder *encoder) +{ + return nouveau_encoder(encoder)->crtc; +} + +static const struct drm_encoder_helper_funcs nv50_dac_helper_funcs = { + .dpms = nv50_dac_dpms, + .save = nv50_dac_save, + .restore = nv50_dac_restore, + .mode_fixup = nv50_dac_mode_fixup, + .prepare = nv50_dac_prepare, + .commit = nv50_dac_commit, + .mode_set = nv50_dac_mode_set, + .get_crtc = nv50_dac_crtc_get, + .detect = nv50_dac_detect, + .disable = nv50_dac_disconnect +}; + +static void +nv50_dac_destroy(struct drm_encoder *encoder) +{ + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + + if (!encoder) + return; + + NV_DEBUG_KMS(encoder->dev, "\n"); + + drm_encoder_cleanup(encoder); + kfree(nv_encoder); +} + +static const struct drm_encoder_funcs nv50_dac_encoder_funcs = { + .destroy = nv50_dac_destroy, +}; + +int +nv50_dac_create(struct drm_connector *connector, struct dcb_entry *entry) +{ + struct nouveau_encoder *nv_encoder; + struct drm_encoder *encoder; + + nv_encoder = kzalloc(sizeof(*nv_encoder), GFP_KERNEL); + if (!nv_encoder) + return -ENOMEM; + encoder = to_drm_encoder(nv_encoder); + + nv_encoder->dcb = entry; + nv_encoder->or = ffs(entry->or) - 1; + + drm_encoder_init(connector->dev, encoder, &nv50_dac_encoder_funcs, + DRM_MODE_ENCODER_DAC); + drm_encoder_helper_add(encoder, &nv50_dac_helper_funcs); + + encoder->possible_crtcs = entry->heads; + encoder->possible_clones = 0; + + drm_mode_connector_attach_encoder(connector, encoder); + return 0; +} + diff --git a/drivers/gpu/drm/nouveau/nv50_display.c b/drivers/gpu/drm/nouveau/nv50_display.c new file mode 100644 index 00000000..08da478b --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv50_display.c @@ -0,0 +1,1011 @@ +/* + * Copyright (C) 2008 Maarten Maathuis. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#define NOUVEAU_DMA_DEBUG (nouveau_reg_debug & NOUVEAU_REG_DEBUG_EVO) +#include "nv50_display.h" +#include "nouveau_crtc.h" +#include "nouveau_encoder.h" +#include "nouveau_connector.h" +#include "nouveau_fb.h" +#include "nouveau_fbcon.h" +#include "nouveau_ramht.h" +#include "drm_crtc_helper.h" + +static void nv50_display_isr(struct drm_device *); +static void nv50_display_bh(unsigned long); + +static inline int +nv50_sor_nr(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + if (dev_priv->chipset < 0x90 || + dev_priv->chipset == 0x92 || + dev_priv->chipset == 0xa0) + return 2; + + return 4; +} + +int +nv50_display_early_init(struct drm_device *dev) +{ + return 0; +} + +void +nv50_display_late_takedown(struct drm_device *dev) +{ +} + +int +nv50_display_init(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpio_engine *pgpio = &dev_priv->engine.gpio; + struct drm_connector *connector; + struct nouveau_channel *evo; + int ret, i; + u32 val; + + NV_DEBUG_KMS(dev, "\n"); + + nv_wr32(dev, 0x00610184, nv_rd32(dev, 0x00614004)); + + /* + * I think the 0x006101XX range is some kind of main control area + * that enables things. + */ + /* CRTC? */ + for (i = 0; i < 2; i++) { + val = nv_rd32(dev, 0x00616100 + (i * 0x800)); + nv_wr32(dev, 0x00610190 + (i * 0x10), val); + val = nv_rd32(dev, 0x00616104 + (i * 0x800)); + nv_wr32(dev, 0x00610194 + (i * 0x10), val); + val = nv_rd32(dev, 0x00616108 + (i * 0x800)); + nv_wr32(dev, 0x00610198 + (i * 0x10), val); + val = nv_rd32(dev, 0x0061610c + (i * 0x800)); + nv_wr32(dev, 0x0061019c + (i * 0x10), val); + } + + /* DAC */ + for (i = 0; i < 3; i++) { + val = nv_rd32(dev, 0x0061a000 + (i * 0x800)); + nv_wr32(dev, 0x006101d0 + (i * 0x04), val); + } + + /* SOR */ + for (i = 0; i < nv50_sor_nr(dev); i++) { + val = nv_rd32(dev, 0x0061c000 + (i * 0x800)); + nv_wr32(dev, 0x006101e0 + (i * 0x04), val); + } + + /* EXT */ + for (i = 0; i < 3; i++) { + val = nv_rd32(dev, 0x0061e000 + (i * 0x800)); + nv_wr32(dev, 0x006101f0 + (i * 0x04), val); + } + + for (i = 0; i < 3; i++) { + nv_wr32(dev, NV50_PDISPLAY_DAC_DPMS_CTRL(i), 0x00550000 | + NV50_PDISPLAY_DAC_DPMS_CTRL_PENDING); + nv_wr32(dev, NV50_PDISPLAY_DAC_CLK_CTRL1(i), 0x00000001); + } + + /* The precise purpose is unknown, i suspect it has something to do + * with text mode. + */ + if (nv_rd32(dev, NV50_PDISPLAY_INTR_1) & 0x100) { + nv_wr32(dev, NV50_PDISPLAY_INTR_1, 0x100); + nv_wr32(dev, 0x006194e8, nv_rd32(dev, 0x006194e8) & ~1); + if (!nv_wait(dev, 0x006194e8, 2, 0)) { + NV_ERROR(dev, "timeout: (0x6194e8 & 2) != 0\n"); + NV_ERROR(dev, "0x6194e8 = 0x%08x\n", + nv_rd32(dev, 0x6194e8)); + return -EBUSY; + } + } + + for (i = 0; i < 2; i++) { + nv_wr32(dev, NV50_PDISPLAY_CURSOR_CURSOR_CTRL2(i), 0x2000); + if (!nv_wait(dev, NV50_PDISPLAY_CURSOR_CURSOR_CTRL2(i), + NV50_PDISPLAY_CURSOR_CURSOR_CTRL2_STATUS, 0)) { + NV_ERROR(dev, "timeout: CURSOR_CTRL2_STATUS == 0\n"); + NV_ERROR(dev, "CURSOR_CTRL2 = 0x%08x\n", + nv_rd32(dev, NV50_PDISPLAY_CURSOR_CURSOR_CTRL2(i))); + return -EBUSY; + } + + nv_wr32(dev, NV50_PDISPLAY_CURSOR_CURSOR_CTRL2(i), + NV50_PDISPLAY_CURSOR_CURSOR_CTRL2_ON); + if (!nv_wait(dev, NV50_PDISPLAY_CURSOR_CURSOR_CTRL2(i), + NV50_PDISPLAY_CURSOR_CURSOR_CTRL2_STATUS, + NV50_PDISPLAY_CURSOR_CURSOR_CTRL2_STATUS_ACTIVE)) { + NV_ERROR(dev, "timeout: " + "CURSOR_CTRL2_STATUS_ACTIVE(%d)\n", i); + NV_ERROR(dev, "CURSOR_CTRL2(%d) = 0x%08x\n", i, + nv_rd32(dev, NV50_PDISPLAY_CURSOR_CURSOR_CTRL2(i))); + return -EBUSY; + } + } + + nv_wr32(dev, NV50_PDISPLAY_PIO_CTRL, 0x00000000); + nv_mask(dev, NV50_PDISPLAY_INTR_0, 0x00000000, 0x00000000); + nv_wr32(dev, NV50_PDISPLAY_INTR_EN_0, 0x00000000); + nv_mask(dev, NV50_PDISPLAY_INTR_1, 0x00000000, 0x00000000); + nv_wr32(dev, NV50_PDISPLAY_INTR_EN_1, + NV50_PDISPLAY_INTR_EN_1_CLK_UNK10 | + NV50_PDISPLAY_INTR_EN_1_CLK_UNK20 | + NV50_PDISPLAY_INTR_EN_1_CLK_UNK40); + + /* enable hotplug interrupts */ + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + struct nouveau_connector *conn = nouveau_connector(connector); + + if (conn->dcb->gpio_tag == 0xff) + continue; + + pgpio->irq_enable(dev, conn->dcb->gpio_tag, true); + } + + ret = nv50_evo_init(dev); + if (ret) + return ret; + evo = nv50_display(dev)->master; + + nv_wr32(dev, NV50_PDISPLAY_OBJECTS, (evo->ramin->vinst >> 8) | 9); + + ret = RING_SPACE(evo, 15); + if (ret) + return ret; + BEGIN_RING(evo, 0, NV50_EVO_UNK84, 2); + OUT_RING(evo, NV50_EVO_UNK84_NOTIFY_DISABLED); + OUT_RING(evo, NvEvoSync); + BEGIN_RING(evo, 0, NV50_EVO_CRTC(0, FB_DMA), 1); + OUT_RING(evo, NV50_EVO_CRTC_FB_DMA_HANDLE_NONE); + BEGIN_RING(evo, 0, NV50_EVO_CRTC(0, UNK0800), 1); + OUT_RING(evo, 0); + BEGIN_RING(evo, 0, NV50_EVO_CRTC(0, DISPLAY_START), 1); + OUT_RING(evo, 0); + BEGIN_RING(evo, 0, NV50_EVO_CRTC(0, UNK082C), 1); + OUT_RING(evo, 0); + /* required to make display sync channels not hate life */ + BEGIN_RING(evo, 0, NV50_EVO_CRTC(0, UNK900), 1); + OUT_RING (evo, 0x00000311); + BEGIN_RING(evo, 0, NV50_EVO_CRTC(1, UNK900), 1); + OUT_RING (evo, 0x00000311); + FIRE_RING(evo); + if (!nv_wait(dev, 0x640004, 0xffffffff, evo->dma.put << 2)) + NV_ERROR(dev, "evo pushbuf stalled\n"); + + + return 0; +} + +static int nv50_display_disable(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nv50_display *disp = nv50_display(dev); + struct nouveau_channel *evo = disp->master; + struct drm_crtc *drm_crtc; + int ret, i; + + NV_DEBUG_KMS(dev, "\n"); + + list_for_each_entry(drm_crtc, &dev->mode_config.crtc_list, head) { + struct nouveau_crtc *crtc = nouveau_crtc(drm_crtc); + + nv50_crtc_blank(crtc, true); + } + + ret = RING_SPACE(evo, 2); + if (ret == 0) { + BEGIN_RING(evo, 0, NV50_EVO_UPDATE, 1); + OUT_RING(evo, 0); + } + FIRE_RING(evo); + + /* Almost like ack'ing a vblank interrupt, maybe in the spirit of + * cleaning up? + */ + list_for_each_entry(drm_crtc, &dev->mode_config.crtc_list, head) { + struct nouveau_crtc *crtc = nouveau_crtc(drm_crtc); + uint32_t mask = NV50_PDISPLAY_INTR_1_VBLANK_CRTC_(crtc->index); + + if (!crtc->base.enabled) + continue; + + nv_wr32(dev, NV50_PDISPLAY_INTR_1, mask); + if (!nv_wait(dev, NV50_PDISPLAY_INTR_1, mask, mask)) { + NV_ERROR(dev, "timeout: (0x610024 & 0x%08x) == " + "0x%08x\n", mask, mask); + NV_ERROR(dev, "0x610024 = 0x%08x\n", + nv_rd32(dev, NV50_PDISPLAY_INTR_1)); + } + } + + nv50_evo_fini(dev); + + for (i = 0; i < 3; i++) { + if (!nv_wait(dev, NV50_PDISPLAY_SOR_DPMS_STATE(i), + NV50_PDISPLAY_SOR_DPMS_STATE_WAIT, 0)) { + NV_ERROR(dev, "timeout: SOR_DPMS_STATE_WAIT(%d) == 0\n", i); + NV_ERROR(dev, "SOR_DPMS_STATE(%d) = 0x%08x\n", i, + nv_rd32(dev, NV50_PDISPLAY_SOR_DPMS_STATE(i))); + } + } + + /* disable interrupts. */ + nv_wr32(dev, NV50_PDISPLAY_INTR_EN_1, 0x00000000); + + /* disable hotplug interrupts */ + nv_wr32(dev, 0xe054, 0xffffffff); + nv_wr32(dev, 0xe050, 0x00000000); + if (dev_priv->chipset >= 0x90) { + nv_wr32(dev, 0xe074, 0xffffffff); + nv_wr32(dev, 0xe070, 0x00000000); + } + return 0; +} + +int nv50_display_create(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct dcb_table *dcb = &dev_priv->vbios.dcb; + struct drm_connector *connector, *ct; + struct nv50_display *priv; + int ret, i; + + NV_DEBUG_KMS(dev, "\n"); + + priv = kzalloc(sizeof(*priv), GFP_KERNEL); + if (!priv) + return -ENOMEM; + dev_priv->engine.display.priv = priv; + + /* init basic kernel modesetting */ + drm_mode_config_init(dev); + + /* Initialise some optional connector properties. */ + drm_mode_create_scaling_mode_property(dev); + drm_mode_create_dithering_property(dev); + + dev->mode_config.min_width = 0; + dev->mode_config.min_height = 0; + + dev->mode_config.funcs = (void *)&nouveau_mode_config_funcs; + + dev->mode_config.max_width = 8192; + dev->mode_config.max_height = 8192; + + dev->mode_config.fb_base = dev_priv->fb_phys; + + /* Create CRTC objects */ + for (i = 0; i < 2; i++) + nv50_crtc_create(dev, i); + + /* We setup the encoders from the BIOS table */ + for (i = 0 ; i < dcb->entries; i++) { + struct dcb_entry *entry = &dcb->entry[i]; + + if (entry->location != DCB_LOC_ON_CHIP) { + NV_WARN(dev, "Off-chip encoder %d/%d unsupported\n", + entry->type, ffs(entry->or) - 1); + continue; + } + + connector = nouveau_connector_create(dev, entry->connector); + if (IS_ERR(connector)) + continue; + + switch (entry->type) { + case OUTPUT_TMDS: + case OUTPUT_LVDS: + case OUTPUT_DP: + nv50_sor_create(connector, entry); + break; + case OUTPUT_ANALOG: + nv50_dac_create(connector, entry); + break; + default: + NV_WARN(dev, "DCB encoder %d unknown\n", entry->type); + continue; + } + } + + list_for_each_entry_safe(connector, ct, + &dev->mode_config.connector_list, head) { + if (!connector->encoder_ids[0]) { + NV_WARN(dev, "%s has no encoders, removing\n", + drm_get_connector_name(connector)); + connector->funcs->destroy(connector); + } + } + + tasklet_init(&priv->tasklet, nv50_display_bh, (unsigned long)dev); + nouveau_irq_register(dev, 26, nv50_display_isr); + + ret = nv50_display_init(dev); + if (ret) { + nv50_display_destroy(dev); + return ret; + } + + return 0; +} + +void +nv50_display_destroy(struct drm_device *dev) +{ + struct nv50_display *disp = nv50_display(dev); + + NV_DEBUG_KMS(dev, "\n"); + + drm_mode_config_cleanup(dev); + + nv50_display_disable(dev); + nouveau_irq_unregister(dev, 26); + kfree(disp); +} + +void +nv50_display_flip_stop(struct drm_crtc *crtc) +{ + struct nv50_display *disp = nv50_display(crtc->dev); + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + struct nv50_display_crtc *dispc = &disp->crtc[nv_crtc->index]; + struct nouveau_channel *evo = dispc->sync; + int ret; + + ret = RING_SPACE(evo, 8); + if (ret) { + WARN_ON(1); + return; + } + + BEGIN_RING(evo, 0, 0x0084, 1); + OUT_RING (evo, 0x00000000); + BEGIN_RING(evo, 0, 0x0094, 1); + OUT_RING (evo, 0x00000000); + BEGIN_RING(evo, 0, 0x00c0, 1); + OUT_RING (evo, 0x00000000); + BEGIN_RING(evo, 0, 0x0080, 1); + OUT_RING (evo, 0x00000000); + FIRE_RING (evo); +} + +int +nv50_display_flip_next(struct drm_crtc *crtc, struct drm_framebuffer *fb, + struct nouveau_channel *chan) +{ + struct drm_nouveau_private *dev_priv = crtc->dev->dev_private; + struct nouveau_framebuffer *nv_fb = nouveau_framebuffer(fb); + struct nv50_display *disp = nv50_display(crtc->dev); + struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); + struct nv50_display_crtc *dispc = &disp->crtc[nv_crtc->index]; + struct nouveau_channel *evo = dispc->sync; + int ret; + + ret = RING_SPACE(evo, chan ? 25 : 27); + if (unlikely(ret)) + return ret; + + /* synchronise with the rendering channel, if necessary */ + if (likely(chan)) { + u64 offset = dispc->sem.bo->vma.offset + dispc->sem.offset; + + ret = RING_SPACE(chan, 10); + if (ret) { + WIND_RING(evo); + return ret; + } + + if (dev_priv->chipset < 0xc0) { + BEGIN_RING(chan, NvSubSw, 0x0060, 2); + OUT_RING (chan, NvEvoSema0 + nv_crtc->index); + OUT_RING (chan, dispc->sem.offset); + BEGIN_RING(chan, NvSubSw, 0x006c, 1); + OUT_RING (chan, 0xf00d0000 | dispc->sem.value); + BEGIN_RING(chan, NvSubSw, 0x0064, 2); + OUT_RING (chan, dispc->sem.offset ^ 0x10); + OUT_RING (chan, 0x74b1e000); + BEGIN_RING(chan, NvSubSw, 0x0060, 1); + if (dev_priv->chipset < 0x84) + OUT_RING (chan, NvSema); + else + OUT_RING (chan, chan->vram_handle); + } else { + BEGIN_NVC0(chan, 2, NvSubM2MF, 0x0010, 4); + OUT_RING (chan, upper_32_bits(offset)); + OUT_RING (chan, lower_32_bits(offset)); + OUT_RING (chan, 0xf00d0000 | dispc->sem.value); + OUT_RING (chan, 0x1002); + BEGIN_NVC0(chan, 2, NvSubM2MF, 0x0010, 4); + OUT_RING (chan, upper_32_bits(offset)); + OUT_RING (chan, lower_32_bits(offset ^ 0x10)); + OUT_RING (chan, 0x74b1e000); + OUT_RING (chan, 0x1001); + } + FIRE_RING (chan); + } else { + nouveau_bo_wr32(dispc->sem.bo, dispc->sem.offset / 4, + 0xf00d0000 | dispc->sem.value); + } + + /* queue the flip on the crtc's "display sync" channel */ + BEGIN_RING(evo, 0, 0x0100, 1); + OUT_RING (evo, 0xfffe0000); + if (chan) { + BEGIN_RING(evo, 0, 0x0084, 1); + OUT_RING (evo, 0x00000100); + } else { + BEGIN_RING(evo, 0, 0x0084, 1); + OUT_RING (evo, 0x00000010); + /* allows gamma somehow, PDISP will bitch at you if + * you don't wait for vblank before changing this.. + */ + BEGIN_RING(evo, 0, 0x00e0, 1); + OUT_RING (evo, 0x40000000); + } + BEGIN_RING(evo, 0, 0x0088, 4); + OUT_RING (evo, dispc->sem.offset); + OUT_RING (evo, 0xf00d0000 | dispc->sem.value); + OUT_RING (evo, 0x74b1e000); + OUT_RING (evo, NvEvoSync); + BEGIN_RING(evo, 0, 0x00a0, 2); + OUT_RING (evo, 0x00000000); + OUT_RING (evo, 0x00000000); + BEGIN_RING(evo, 0, 0x00c0, 1); + OUT_RING (evo, nv_fb->r_dma); + BEGIN_RING(evo, 0, 0x0110, 2); + OUT_RING (evo, 0x00000000); + OUT_RING (evo, 0x00000000); + BEGIN_RING(evo, 0, 0x0800, 5); + OUT_RING (evo, (nv_fb->nvbo->bo.mem.start << PAGE_SHIFT) >> 8); + OUT_RING (evo, 0); + OUT_RING (evo, (fb->height << 16) | fb->width); + OUT_RING (evo, nv_fb->r_pitch); + OUT_RING (evo, nv_fb->r_format); + BEGIN_RING(evo, 0, 0x0080, 1); + OUT_RING (evo, 0x00000000); + FIRE_RING (evo); + + dispc->sem.offset ^= 0x10; + dispc->sem.value++; + return 0; +} + +static u16 +nv50_display_script_select(struct drm_device *dev, struct dcb_entry *dcb, + u32 mc, int pxclk) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_connector *nv_connector = NULL; + struct drm_encoder *encoder; + struct nvbios *bios = &dev_priv->vbios; + u32 script = 0, or; + + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + + if (nv_encoder->dcb != dcb) + continue; + + nv_connector = nouveau_encoder_connector_get(nv_encoder); + break; + } + + or = ffs(dcb->or) - 1; + switch (dcb->type) { + case OUTPUT_LVDS: + script = (mc >> 8) & 0xf; + if (bios->fp_no_ddc) { + if (bios->fp.dual_link) + script |= 0x0100; + if (bios->fp.if_is_24bit) + script |= 0x0200; + } else { + /* determine number of lvds links */ + if (nv_connector && nv_connector->edid && + nv_connector->dcb->type == DCB_CONNECTOR_LVDS_SPWG) { + /* http://www.spwg.org */ + if (((u8 *)nv_connector->edid)[121] == 2) + script |= 0x0100; + } else + if (pxclk >= bios->fp.duallink_transition_clk) { + script |= 0x0100; + } + + /* determine panel depth */ + if (script & 0x0100) { + if (bios->fp.strapless_is_24bit & 2) + script |= 0x0200; + } else { + if (bios->fp.strapless_is_24bit & 1) + script |= 0x0200; + } + + if (nv_connector && nv_connector->edid && + (nv_connector->edid->revision >= 4) && + (nv_connector->edid->input & 0x70) >= 0x20) + script |= 0x0200; + } + + if (nouveau_uscript_lvds >= 0) { + NV_INFO(dev, "override script 0x%04x with 0x%04x " + "for output LVDS-%d\n", script, + nouveau_uscript_lvds, or); + script = nouveau_uscript_lvds; + } + break; + case OUTPUT_TMDS: + script = (mc >> 8) & 0xf; + if (pxclk >= 165000) + script |= 0x0100; + + if (nouveau_uscript_tmds >= 0) { + NV_INFO(dev, "override script 0x%04x with 0x%04x " + "for output TMDS-%d\n", script, + nouveau_uscript_tmds, or); + script = nouveau_uscript_tmds; + } + break; + case OUTPUT_DP: + script = (mc >> 8) & 0xf; + break; + case OUTPUT_ANALOG: + script = 0xff; + break; + default: + NV_ERROR(dev, "modeset on unsupported output type!\n"); + break; + } + + return script; +} + +static void +nv50_display_vblank_crtc_handler(struct drm_device *dev, int crtc) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_channel *chan, *tmp; + + list_for_each_entry_safe(chan, tmp, &dev_priv->vbl_waiting, + nvsw.vbl_wait) { + if (chan->nvsw.vblsem_head != crtc) + continue; + + nouveau_bo_wr32(chan->notifier_bo, chan->nvsw.vblsem_offset, + chan->nvsw.vblsem_rval); + list_del(&chan->nvsw.vbl_wait); + drm_vblank_put(dev, crtc); + } + + drm_handle_vblank(dev, crtc); +} + +static void +nv50_display_vblank_handler(struct drm_device *dev, uint32_t intr) +{ + if (intr & NV50_PDISPLAY_INTR_1_VBLANK_CRTC_0) + nv50_display_vblank_crtc_handler(dev, 0); + + if (intr & NV50_PDISPLAY_INTR_1_VBLANK_CRTC_1) + nv50_display_vblank_crtc_handler(dev, 1); + + nv_wr32(dev, NV50_PDISPLAY_INTR_1, NV50_PDISPLAY_INTR_1_VBLANK_CRTC); +} + +static void +nv50_display_unk10_handler(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nv50_display *disp = nv50_display(dev); + u32 unk30 = nv_rd32(dev, 0x610030), mc; + int i, crtc, or, type = OUTPUT_ANY; + + NV_DEBUG_KMS(dev, "0x610030: 0x%08x\n", unk30); + disp->irq.dcb = NULL; + + nv_wr32(dev, 0x619494, nv_rd32(dev, 0x619494) & ~8); + + /* Determine which CRTC we're dealing with, only 1 ever will be + * signalled at the same time with the current nouveau code. + */ + crtc = ffs((unk30 & 0x00000060) >> 5) - 1; + if (crtc < 0) + goto ack; + + /* Nothing needs to be done for the encoder */ + crtc = ffs((unk30 & 0x00000180) >> 7) - 1; + if (crtc < 0) + goto ack; + + /* Find which encoder was connected to the CRTC */ + for (i = 0; type == OUTPUT_ANY && i < 3; i++) { + mc = nv_rd32(dev, NV50_PDISPLAY_DAC_MODE_CTRL_C(i)); + NV_DEBUG_KMS(dev, "DAC-%d mc: 0x%08x\n", i, mc); + if (!(mc & (1 << crtc))) + continue; + + switch ((mc & 0x00000f00) >> 8) { + case 0: type = OUTPUT_ANALOG; break; + case 1: type = OUTPUT_TV; break; + default: + NV_ERROR(dev, "invalid mc, DAC-%d: 0x%08x\n", i, mc); + goto ack; + } + + or = i; + } + + for (i = 0; type == OUTPUT_ANY && i < nv50_sor_nr(dev); i++) { + if (dev_priv->chipset < 0x90 || + dev_priv->chipset == 0x92 || + dev_priv->chipset == 0xa0) + mc = nv_rd32(dev, NV50_PDISPLAY_SOR_MODE_CTRL_C(i)); + else + mc = nv_rd32(dev, NV90_PDISPLAY_SOR_MODE_CTRL_C(i)); + + NV_DEBUG_KMS(dev, "SOR-%d mc: 0x%08x\n", i, mc); + if (!(mc & (1 << crtc))) + continue; + + switch ((mc & 0x00000f00) >> 8) { + case 0: type = OUTPUT_LVDS; break; + case 1: type = OUTPUT_TMDS; break; + case 2: type = OUTPUT_TMDS; break; + case 5: type = OUTPUT_TMDS; break; + case 8: type = OUTPUT_DP; break; + case 9: type = OUTPUT_DP; break; + default: + NV_ERROR(dev, "invalid mc, SOR-%d: 0x%08x\n", i, mc); + goto ack; + } + + or = i; + } + + /* There was no encoder to disable */ + if (type == OUTPUT_ANY) + goto ack; + + /* Disable the encoder */ + for (i = 0; i < dev_priv->vbios.dcb.entries; i++) { + struct dcb_entry *dcb = &dev_priv->vbios.dcb.entry[i]; + + if (dcb->type == type && (dcb->or & (1 << or))) { + nouveau_bios_run_display_table(dev, dcb, 0, -1); + disp->irq.dcb = dcb; + goto ack; + } + } + + NV_ERROR(dev, "no dcb for %d %d 0x%08x\n", or, type, mc); +ack: + nv_wr32(dev, NV50_PDISPLAY_INTR_1, NV50_PDISPLAY_INTR_1_CLK_UNK10); + nv_wr32(dev, 0x610030, 0x80000000); +} + +static void +nv50_display_unk20_dp_hack(struct drm_device *dev, struct dcb_entry *dcb) +{ + int or = ffs(dcb->or) - 1, link = !(dcb->dpconf.sor.link & 1); + struct drm_encoder *encoder; + uint32_t tmp, unk0 = 0, unk1 = 0; + + if (dcb->type != OUTPUT_DP) + return; + + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + + if (nv_encoder->dcb == dcb) { + unk0 = nv_encoder->dp.unk0; + unk1 = nv_encoder->dp.unk1; + break; + } + } + + if (unk0 || unk1) { + tmp = nv_rd32(dev, NV50_SOR_DP_CTRL(or, link)); + tmp &= 0xfffffe03; + nv_wr32(dev, NV50_SOR_DP_CTRL(or, link), tmp | unk0); + + tmp = nv_rd32(dev, NV50_SOR_DP_UNK128(or, link)); + tmp &= 0xfef080c0; + nv_wr32(dev, NV50_SOR_DP_UNK128(or, link), tmp | unk1); + } +} + +static void +nv50_display_unk20_handler(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nv50_display *disp = nv50_display(dev); + u32 unk30 = nv_rd32(dev, 0x610030), tmp, pclk, script, mc = 0; + struct dcb_entry *dcb; + int i, crtc, or, type = OUTPUT_ANY; + + NV_DEBUG_KMS(dev, "0x610030: 0x%08x\n", unk30); + dcb = disp->irq.dcb; + if (dcb) { + nouveau_bios_run_display_table(dev, dcb, 0, -2); + disp->irq.dcb = NULL; + } + + /* CRTC clock change requested? */ + crtc = ffs((unk30 & 0x00000600) >> 9) - 1; + if (crtc >= 0) { + pclk = nv_rd32(dev, NV50_PDISPLAY_CRTC_P(crtc, CLOCK)); + pclk &= 0x003fffff; + + nv50_crtc_set_clock(dev, crtc, pclk); + + tmp = nv_rd32(dev, NV50_PDISPLAY_CRTC_CLK_CTRL2(crtc)); + tmp &= ~0x000000f; + nv_wr32(dev, NV50_PDISPLAY_CRTC_CLK_CTRL2(crtc), tmp); + } + + /* Nothing needs to be done for the encoder */ + crtc = ffs((unk30 & 0x00000180) >> 7) - 1; + if (crtc < 0) + goto ack; + pclk = nv_rd32(dev, NV50_PDISPLAY_CRTC_P(crtc, CLOCK)) & 0x003fffff; + + /* Find which encoder is connected to the CRTC */ + for (i = 0; type == OUTPUT_ANY && i < 3; i++) { + mc = nv_rd32(dev, NV50_PDISPLAY_DAC_MODE_CTRL_P(i)); + NV_DEBUG_KMS(dev, "DAC-%d mc: 0x%08x\n", i, mc); + if (!(mc & (1 << crtc))) + continue; + + switch ((mc & 0x00000f00) >> 8) { + case 0: type = OUTPUT_ANALOG; break; + case 1: type = OUTPUT_TV; break; + default: + NV_ERROR(dev, "invalid mc, DAC-%d: 0x%08x\n", i, mc); + goto ack; + } + + or = i; + } + + for (i = 0; type == OUTPUT_ANY && i < nv50_sor_nr(dev); i++) { + if (dev_priv->chipset < 0x90 || + dev_priv->chipset == 0x92 || + dev_priv->chipset == 0xa0) + mc = nv_rd32(dev, NV50_PDISPLAY_SOR_MODE_CTRL_P(i)); + else + mc = nv_rd32(dev, NV90_PDISPLAY_SOR_MODE_CTRL_P(i)); + + NV_DEBUG_KMS(dev, "SOR-%d mc: 0x%08x\n", i, mc); + if (!(mc & (1 << crtc))) + continue; + + switch ((mc & 0x00000f00) >> 8) { + case 0: type = OUTPUT_LVDS; break; + case 1: type = OUTPUT_TMDS; break; + case 2: type = OUTPUT_TMDS; break; + case 5: type = OUTPUT_TMDS; break; + case 8: type = OUTPUT_DP; break; + case 9: type = OUTPUT_DP; break; + default: + NV_ERROR(dev, "invalid mc, SOR-%d: 0x%08x\n", i, mc); + goto ack; + } + + or = i; + } + + if (type == OUTPUT_ANY) + goto ack; + + /* Enable the encoder */ + for (i = 0; i < dev_priv->vbios.dcb.entries; i++) { + dcb = &dev_priv->vbios.dcb.entry[i]; + if (dcb->type == type && (dcb->or & (1 << or))) + break; + } + + if (i == dev_priv->vbios.dcb.entries) { + NV_ERROR(dev, "no dcb for %d %d 0x%08x\n", or, type, mc); + goto ack; + } + + script = nv50_display_script_select(dev, dcb, mc, pclk); + nouveau_bios_run_display_table(dev, dcb, script, pclk); + + nv50_display_unk20_dp_hack(dev, dcb); + + if (dcb->type != OUTPUT_ANALOG) { + tmp = nv_rd32(dev, NV50_PDISPLAY_SOR_CLK_CTRL2(or)); + tmp &= ~0x00000f0f; + if (script & 0x0100) + tmp |= 0x00000101; + nv_wr32(dev, NV50_PDISPLAY_SOR_CLK_CTRL2(or), tmp); + } else { + nv_wr32(dev, NV50_PDISPLAY_DAC_CLK_CTRL2(or), 0); + } + + disp->irq.dcb = dcb; + disp->irq.pclk = pclk; + disp->irq.script = script; + +ack: + nv_wr32(dev, NV50_PDISPLAY_INTR_1, NV50_PDISPLAY_INTR_1_CLK_UNK20); + nv_wr32(dev, 0x610030, 0x80000000); +} + +/* If programming a TMDS output on a SOR that can also be configured for + * DisplayPort, make sure NV50_SOR_DP_CTRL_ENABLE is forced off. + * + * It looks like the VBIOS TMDS scripts make an attempt at this, however, + * the VBIOS scripts on at least one board I have only switch it off on + * link 0, causing a blank display if the output has previously been + * programmed for DisplayPort. + */ +static void +nv50_display_unk40_dp_set_tmds(struct drm_device *dev, struct dcb_entry *dcb) +{ + int or = ffs(dcb->or) - 1, link = !(dcb->dpconf.sor.link & 1); + struct drm_encoder *encoder; + u32 tmp; + + if (dcb->type != OUTPUT_TMDS) + return; + + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + + if (nv_encoder->dcb->type == OUTPUT_DP && + nv_encoder->dcb->or & (1 << or)) { + tmp = nv_rd32(dev, NV50_SOR_DP_CTRL(or, link)); + tmp &= ~NV50_SOR_DP_CTRL_ENABLED; + nv_wr32(dev, NV50_SOR_DP_CTRL(or, link), tmp); + break; + } + } +} + +static void +nv50_display_unk40_handler(struct drm_device *dev) +{ + struct nv50_display *disp = nv50_display(dev); + struct dcb_entry *dcb = disp->irq.dcb; + u16 script = disp->irq.script; + u32 unk30 = nv_rd32(dev, 0x610030), pclk = disp->irq.pclk; + + NV_DEBUG_KMS(dev, "0x610030: 0x%08x\n", unk30); + disp->irq.dcb = NULL; + if (!dcb) + goto ack; + + nouveau_bios_run_display_table(dev, dcb, script, -pclk); + nv50_display_unk40_dp_set_tmds(dev, dcb); + +ack: + nv_wr32(dev, NV50_PDISPLAY_INTR_1, NV50_PDISPLAY_INTR_1_CLK_UNK40); + nv_wr32(dev, 0x610030, 0x80000000); + nv_wr32(dev, 0x619494, nv_rd32(dev, 0x619494) | 8); +} + +static void +nv50_display_bh(unsigned long data) +{ + struct drm_device *dev = (struct drm_device *)data; + + for (;;) { + uint32_t intr0 = nv_rd32(dev, NV50_PDISPLAY_INTR_0); + uint32_t intr1 = nv_rd32(dev, NV50_PDISPLAY_INTR_1); + + NV_DEBUG_KMS(dev, "PDISPLAY_INTR_BH 0x%08x 0x%08x\n", intr0, intr1); + + if (intr1 & NV50_PDISPLAY_INTR_1_CLK_UNK10) + nv50_display_unk10_handler(dev); + else + if (intr1 & NV50_PDISPLAY_INTR_1_CLK_UNK20) + nv50_display_unk20_handler(dev); + else + if (intr1 & NV50_PDISPLAY_INTR_1_CLK_UNK40) + nv50_display_unk40_handler(dev); + else + break; + } + + nv_wr32(dev, NV03_PMC_INTR_EN_0, 1); +} + +static void +nv50_display_error_handler(struct drm_device *dev) +{ + u32 channels = (nv_rd32(dev, NV50_PDISPLAY_INTR_0) & 0x001f0000) >> 16; + u32 addr, data; + int chid; + + for (chid = 0; chid < 5; chid++) { + if (!(channels & (1 << chid))) + continue; + + nv_wr32(dev, NV50_PDISPLAY_INTR_0, 0x00010000 << chid); + addr = nv_rd32(dev, NV50_PDISPLAY_TRAPPED_ADDR(chid)); + data = nv_rd32(dev, NV50_PDISPLAY_TRAPPED_DATA(chid)); + NV_ERROR(dev, "EvoCh %d Mthd 0x%04x Data 0x%08x " + "(0x%04x 0x%02x)\n", chid, + addr & 0xffc, data, addr >> 16, (addr >> 12) & 0xf); + + nv_wr32(dev, NV50_PDISPLAY_TRAPPED_ADDR(chid), 0x90000000); + } +} + +static void +nv50_display_isr(struct drm_device *dev) +{ + struct nv50_display *disp = nv50_display(dev); + uint32_t delayed = 0; + + while (nv_rd32(dev, NV50_PMC_INTR_0) & NV50_PMC_INTR_0_DISPLAY) { + uint32_t intr0 = nv_rd32(dev, NV50_PDISPLAY_INTR_0); + uint32_t intr1 = nv_rd32(dev, NV50_PDISPLAY_INTR_1); + uint32_t clock; + + NV_DEBUG_KMS(dev, "PDISPLAY_INTR 0x%08x 0x%08x\n", intr0, intr1); + + if (!intr0 && !(intr1 & ~delayed)) + break; + + if (intr0 & 0x001f0000) { + nv50_display_error_handler(dev); + intr0 &= ~0x001f0000; + } + + if (intr1 & NV50_PDISPLAY_INTR_1_VBLANK_CRTC) { + nv50_display_vblank_handler(dev, intr1); + intr1 &= ~NV50_PDISPLAY_INTR_1_VBLANK_CRTC; + } + + clock = (intr1 & (NV50_PDISPLAY_INTR_1_CLK_UNK10 | + NV50_PDISPLAY_INTR_1_CLK_UNK20 | + NV50_PDISPLAY_INTR_1_CLK_UNK40)); + if (clock) { + nv_wr32(dev, NV03_PMC_INTR_EN_0, 0); + tasklet_schedule(&disp->tasklet); + delayed |= clock; + intr1 &= ~clock; + } + + if (intr0) { + NV_ERROR(dev, "unknown PDISPLAY_INTR_0: 0x%08x\n", intr0); + nv_wr32(dev, NV50_PDISPLAY_INTR_0, intr0); + } + + if (intr1) { + NV_ERROR(dev, + "unknown PDISPLAY_INTR_1: 0x%08x\n", intr1); + nv_wr32(dev, NV50_PDISPLAY_INTR_1, intr1); + } + } +} diff --git a/drivers/gpu/drm/nouveau/nv50_display.h b/drivers/gpu/drm/nouveau/nv50_display.h new file mode 100644 index 00000000..c2da503a --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv50_display.h @@ -0,0 +1,87 @@ +/* + * Copyright (C) 2008 Maarten Maathuis. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __NV50_DISPLAY_H__ +#define __NV50_DISPLAY_H__ + +#include "drmP.h" +#include "drm.h" +#include "nouveau_drv.h" +#include "nouveau_dma.h" +#include "nouveau_reg.h" +#include "nouveau_crtc.h" +#include "nv50_evo.h" + +struct nv50_display_crtc { + struct nouveau_channel *sync; + struct { + struct nouveau_bo *bo; + u32 offset; + u16 value; + } sem; +}; + +struct nv50_display { + struct nouveau_channel *master; + struct nouveau_gpuobj *ntfy; + + struct nv50_display_crtc crtc[2]; + + struct tasklet_struct tasklet; + struct { + struct dcb_entry *dcb; + u16 script; + u32 pclk; + } irq; +}; + +static inline struct nv50_display * +nv50_display(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + return dev_priv->engine.display.priv; +} + +int nv50_display_early_init(struct drm_device *dev); +void nv50_display_late_takedown(struct drm_device *dev); +int nv50_display_create(struct drm_device *dev); +int nv50_display_init(struct drm_device *dev); +void nv50_display_destroy(struct drm_device *dev); +int nv50_crtc_blank(struct nouveau_crtc *, bool blank); +int nv50_crtc_set_clock(struct drm_device *, int head, int pclk); + +int nv50_display_flip_next(struct drm_crtc *, struct drm_framebuffer *, + struct nouveau_channel *chan); +void nv50_display_flip_stop(struct drm_crtc *); + +int nv50_evo_init(struct drm_device *dev); +void nv50_evo_fini(struct drm_device *dev); +void nv50_evo_dmaobj_init(struct nouveau_gpuobj *, u32 memtype, u64 base, + u64 size); +int nv50_evo_dmaobj_new(struct nouveau_channel *, u32 handle, u32 memtype, + u64 base, u64 size, struct nouveau_gpuobj **); + +#endif /* __NV50_DISPLAY_H__ */ diff --git a/drivers/gpu/drm/nouveau/nv50_evo.c b/drivers/gpu/drm/nouveau/nv50_evo.c new file mode 100644 index 00000000..c8e83c1a --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv50_evo.c @@ -0,0 +1,425 @@ +/* + * Copyright 2010 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include "drmP.h" + +#include "nouveau_drv.h" +#include "nouveau_dma.h" +#include "nouveau_ramht.h" +#include "nv50_display.h" + +static void +nv50_evo_channel_del(struct nouveau_channel **pevo) +{ + struct nouveau_channel *evo = *pevo; + + if (!evo) + return; + *pevo = NULL; + + nouveau_gpuobj_channel_takedown(evo); + nouveau_bo_unmap(evo->pushbuf_bo); + nouveau_bo_ref(NULL, &evo->pushbuf_bo); + + if (evo->user) + iounmap(evo->user); + + kfree(evo); +} + +void +nv50_evo_dmaobj_init(struct nouveau_gpuobj *obj, u32 memtype, u64 base, u64 size) +{ + struct drm_nouveau_private *dev_priv = obj->dev->dev_private; + u32 flags5; + + if (dev_priv->chipset < 0xc0) { + /* not supported on 0x50, specified in format mthd */ + if (dev_priv->chipset == 0x50) + memtype = 0; + flags5 = 0x00010000; + } else { + if (memtype & 0x80000000) + flags5 = 0x00000000; /* large pages */ + else + flags5 = 0x00020000; + } + + nv50_gpuobj_dma_init(obj, 0, 0x3d, base, size, NV_MEM_TARGET_VRAM, + NV_MEM_ACCESS_RW, (memtype >> 8) & 0xff, 0); + nv_wo32(obj, 0x14, flags5); + dev_priv->engine.instmem.flush(obj->dev); +} + +int +nv50_evo_dmaobj_new(struct nouveau_channel *evo, u32 handle, u32 memtype, + u64 base, u64 size, struct nouveau_gpuobj **pobj) +{ + struct nv50_display *disp = nv50_display(evo->dev); + struct nouveau_gpuobj *obj = NULL; + int ret; + + ret = nouveau_gpuobj_new(evo->dev, disp->master, 6*4, 32, 0, &obj); + if (ret) + return ret; + obj->engine = NVOBJ_ENGINE_DISPLAY; + + nv50_evo_dmaobj_init(obj, memtype, base, size); + + ret = nouveau_ramht_insert(evo, handle, obj); + if (ret) + goto out; + + if (pobj) + nouveau_gpuobj_ref(obj, pobj); +out: + nouveau_gpuobj_ref(NULL, &obj); + return ret; +} + +static int +nv50_evo_channel_new(struct drm_device *dev, int chid, + struct nouveau_channel **pevo) +{ + struct nv50_display *disp = nv50_display(dev); + struct nouveau_channel *evo; + int ret; + + evo = kzalloc(sizeof(struct nouveau_channel), GFP_KERNEL); + if (!evo) + return -ENOMEM; + *pevo = evo; + + evo->id = chid; + evo->dev = dev; + evo->user_get = 4; + evo->user_put = 0; + + ret = nouveau_bo_new(dev, NULL, 4096, 0, TTM_PL_FLAG_VRAM, 0, 0, + &evo->pushbuf_bo); + if (ret == 0) + ret = nouveau_bo_pin(evo->pushbuf_bo, TTM_PL_FLAG_VRAM); + if (ret) { + NV_ERROR(dev, "Error creating EVO DMA push buffer: %d\n", ret); + nv50_evo_channel_del(pevo); + return ret; + } + + ret = nouveau_bo_map(evo->pushbuf_bo); + if (ret) { + NV_ERROR(dev, "Error mapping EVO DMA push buffer: %d\n", ret); + nv50_evo_channel_del(pevo); + return ret; + } + + evo->user = ioremap(pci_resource_start(dev->pdev, 0) + + NV50_PDISPLAY_USER(evo->id), PAGE_SIZE); + if (!evo->user) { + NV_ERROR(dev, "Error mapping EVO control regs.\n"); + nv50_evo_channel_del(pevo); + return -ENOMEM; + } + + /* bind primary evo channel's ramht to the channel */ + if (disp->master && evo != disp->master) + nouveau_ramht_ref(disp->master->ramht, &evo->ramht, NULL); + + return 0; +} + +static int +nv50_evo_channel_init(struct nouveau_channel *evo) +{ + struct drm_device *dev = evo->dev; + int id = evo->id, ret, i; + u64 pushbuf = evo->pushbuf_bo->bo.mem.start << PAGE_SHIFT; + u32 tmp; + + tmp = nv_rd32(dev, NV50_PDISPLAY_EVO_CTRL(id)); + if ((tmp & 0x009f0000) == 0x00020000) + nv_wr32(dev, NV50_PDISPLAY_EVO_CTRL(id), tmp | 0x00800000); + + tmp = nv_rd32(dev, NV50_PDISPLAY_EVO_CTRL(id)); + if ((tmp & 0x003f0000) == 0x00030000) + nv_wr32(dev, NV50_PDISPLAY_EVO_CTRL(id), tmp | 0x00600000); + + /* initialise fifo */ + nv_wr32(dev, NV50_PDISPLAY_EVO_DMA_CB(id), pushbuf >> 8 | + NV50_PDISPLAY_EVO_DMA_CB_LOCATION_VRAM | + NV50_PDISPLAY_EVO_DMA_CB_VALID); + nv_wr32(dev, NV50_PDISPLAY_EVO_UNK2(id), 0x00010000); + nv_wr32(dev, NV50_PDISPLAY_EVO_HASH_TAG(id), id); + nv_mask(dev, NV50_PDISPLAY_EVO_CTRL(id), NV50_PDISPLAY_EVO_CTRL_DMA, + NV50_PDISPLAY_EVO_CTRL_DMA_ENABLED); + + nv_wr32(dev, NV50_PDISPLAY_USER_PUT(id), 0x00000000); + nv_wr32(dev, NV50_PDISPLAY_EVO_CTRL(id), 0x01000003 | + NV50_PDISPLAY_EVO_CTRL_DMA_ENABLED); + if (!nv_wait(dev, NV50_PDISPLAY_EVO_CTRL(id), 0x80000000, 0x00000000)) { + NV_ERROR(dev, "EvoCh %d init timeout: 0x%08x\n", id, + nv_rd32(dev, NV50_PDISPLAY_EVO_CTRL(id))); + return -EBUSY; + } + + /* enable error reporting on the channel */ + nv_mask(dev, 0x610028, 0x00000000, 0x00010001 << id); + + evo->dma.max = (4096/4) - 2; + evo->dma.max &= ~7; + evo->dma.put = 0; + evo->dma.cur = evo->dma.put; + evo->dma.free = evo->dma.max - evo->dma.cur; + + ret = RING_SPACE(evo, NOUVEAU_DMA_SKIPS); + if (ret) + return ret; + + for (i = 0; i < NOUVEAU_DMA_SKIPS; i++) + OUT_RING(evo, 0); + + return 0; +} + +static void +nv50_evo_channel_fini(struct nouveau_channel *evo) +{ + struct drm_device *dev = evo->dev; + int id = evo->id; + + nv_mask(dev, 0x610028, 0x00010001 << id, 0x00000000); + nv_mask(dev, NV50_PDISPLAY_EVO_CTRL(id), 0x00001010, 0x00001000); + nv_wr32(dev, NV50_PDISPLAY_INTR_0, (1 << id)); + nv_mask(dev, NV50_PDISPLAY_EVO_CTRL(id), 0x00000003, 0x00000000); + if (!nv_wait(dev, NV50_PDISPLAY_EVO_CTRL(id), 0x001e0000, 0x00000000)) { + NV_ERROR(dev, "EvoCh %d takedown timeout: 0x%08x\n", id, + nv_rd32(dev, NV50_PDISPLAY_EVO_CTRL(id))); + } +} + +static void +nv50_evo_destroy(struct drm_device *dev) +{ + struct nv50_display *disp = nv50_display(dev); + int i; + + for (i = 0; i < 2; i++) { + if (disp->crtc[i].sem.bo) { + nouveau_bo_unmap(disp->crtc[i].sem.bo); + nouveau_bo_ref(NULL, &disp->crtc[i].sem.bo); + } + nv50_evo_channel_del(&disp->crtc[i].sync); + } + nouveau_gpuobj_ref(NULL, &disp->ntfy); + nv50_evo_channel_del(&disp->master); +} + +static int +nv50_evo_create(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nv50_display *disp = nv50_display(dev); + struct nouveau_gpuobj *ramht = NULL; + struct nouveau_channel *evo; + int ret, i, j; + + /* create primary evo channel, the one we use for modesetting + * purporses + */ + ret = nv50_evo_channel_new(dev, 0, &disp->master); + if (ret) + return ret; + evo = disp->master; + + /* setup object management on it, any other evo channel will + * use this also as there's no per-channel support on the + * hardware + */ + ret = nouveau_gpuobj_new(dev, NULL, 32768, 65536, + NVOBJ_FLAG_ZERO_ALLOC, &evo->ramin); + if (ret) { + NV_ERROR(dev, "Error allocating EVO channel memory: %d\n", ret); + goto err; + } + + ret = drm_mm_init(&evo->ramin_heap, 0, 32768); + if (ret) { + NV_ERROR(dev, "Error initialising EVO PRAMIN heap: %d\n", ret); + goto err; + } + + ret = nouveau_gpuobj_new(dev, evo, 4096, 16, 0, &ramht); + if (ret) { + NV_ERROR(dev, "Unable to allocate EVO RAMHT: %d\n", ret); + goto err; + } + + ret = nouveau_ramht_new(dev, ramht, &evo->ramht); + nouveau_gpuobj_ref(NULL, &ramht); + if (ret) + goto err; + + /* not sure exactly what this is.. + * + * the first dword of the structure is used by nvidia to wait on + * full completion of an EVO "update" command. + * + * method 0x8c on the master evo channel will fill a lot more of + * this structure with some undefined info + */ + ret = nouveau_gpuobj_new(dev, disp->master, 0x1000, 0, + NVOBJ_FLAG_ZERO_ALLOC, &disp->ntfy); + if (ret) + goto err; + + ret = nv50_evo_dmaobj_new(disp->master, NvEvoSync, 0x0000, + disp->ntfy->vinst, disp->ntfy->size, NULL); + if (ret) + goto err; + + /* create some default objects for the scanout memtypes we support */ + ret = nv50_evo_dmaobj_new(disp->master, NvEvoVRAM, 0x0000, + 0, dev_priv->vram_size, NULL); + if (ret) + goto err; + + ret = nv50_evo_dmaobj_new(disp->master, NvEvoVRAM_LP, 0x80000000, + 0, dev_priv->vram_size, NULL); + if (ret) + goto err; + + ret = nv50_evo_dmaobj_new(disp->master, NvEvoFB32, 0x80000000 | + (dev_priv->chipset < 0xc0 ? 0x7a00 : 0xfe00), + 0, dev_priv->vram_size, NULL); + if (ret) + goto err; + + ret = nv50_evo_dmaobj_new(disp->master, NvEvoFB16, 0x80000000 | + (dev_priv->chipset < 0xc0 ? 0x7000 : 0xfe00), + 0, dev_priv->vram_size, NULL); + if (ret) + goto err; + + /* create "display sync" channels and other structures we need + * to implement page flipping + */ + for (i = 0; i < 2; i++) { + struct nv50_display_crtc *dispc = &disp->crtc[i]; + u64 offset; + + ret = nv50_evo_channel_new(dev, 1 + i, &dispc->sync); + if (ret) + goto err; + + ret = nouveau_bo_new(dev, NULL, 4096, 0x1000, TTM_PL_FLAG_VRAM, + 0, 0x0000, &dispc->sem.bo); + if (!ret) { + offset = dispc->sem.bo->bo.mem.start << PAGE_SHIFT; + + ret = nouveau_bo_pin(dispc->sem.bo, TTM_PL_FLAG_VRAM); + if (!ret) + ret = nouveau_bo_map(dispc->sem.bo); + if (ret) + nouveau_bo_ref(NULL, &dispc->sem.bo); + } + + if (ret) + goto err; + + ret = nv50_evo_dmaobj_new(dispc->sync, NvEvoSync, 0x0000, + offset, 4096, NULL); + if (ret) + goto err; + + ret = nv50_evo_dmaobj_new(dispc->sync, NvEvoVRAM_LP, 0x80000000, + 0, dev_priv->vram_size, NULL); + if (ret) + goto err; + + ret = nv50_evo_dmaobj_new(dispc->sync, NvEvoFB32, 0x80000000 | + (dev_priv->chipset < 0xc0 ? + 0x7a00 : 0xfe00), + 0, dev_priv->vram_size, NULL); + if (ret) + goto err; + + ret = nv50_evo_dmaobj_new(dispc->sync, NvEvoFB16, 0x80000000 | + (dev_priv->chipset < 0xc0 ? + 0x7000 : 0xfe00), + 0, dev_priv->vram_size, NULL); + if (ret) + goto err; + + for (j = 0; j < 4096; j += 4) + nouveau_bo_wr32(dispc->sem.bo, j / 4, 0x74b1e000); + dispc->sem.offset = 0; + } + + return 0; + +err: + nv50_evo_destroy(dev); + return ret; +} + +int +nv50_evo_init(struct drm_device *dev) +{ + struct nv50_display *disp = nv50_display(dev); + int ret, i; + + if (!disp->master) { + ret = nv50_evo_create(dev); + if (ret) + return ret; + } + + ret = nv50_evo_channel_init(disp->master); + if (ret) + return ret; + + for (i = 0; i < 2; i++) { + ret = nv50_evo_channel_init(disp->crtc[i].sync); + if (ret) + return ret; + } + + return 0; +} + +void +nv50_evo_fini(struct drm_device *dev) +{ + struct nv50_display *disp = nv50_display(dev); + int i; + + for (i = 0; i < 2; i++) { + if (disp->crtc[i].sync) + nv50_evo_channel_fini(disp->crtc[i].sync); + } + + if (disp->master) + nv50_evo_channel_fini(disp->master); + + nv50_evo_destroy(dev); +} diff --git a/drivers/gpu/drm/nouveau/nv50_evo.h b/drivers/gpu/drm/nouveau/nv50_evo.h new file mode 100644 index 00000000..3860ca62 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv50_evo.h @@ -0,0 +1,119 @@ +/* + * Copyright (C) 2008 Maarten Maathuis. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __NV50_EVO_H__ +#define __NV50_EVO_H__ + +#define NV50_EVO_UPDATE 0x00000080 +#define NV50_EVO_UNK84 0x00000084 +#define NV50_EVO_UNK84_NOTIFY 0x40000000 +#define NV50_EVO_UNK84_NOTIFY_DISABLED 0x00000000 +#define NV50_EVO_UNK84_NOTIFY_ENABLED 0x40000000 +#define NV50_EVO_DMA_NOTIFY 0x00000088 +#define NV50_EVO_DMA_NOTIFY_HANDLE 0xffffffff +#define NV50_EVO_DMA_NOTIFY_HANDLE_NONE 0x00000000 +#define NV50_EVO_UNK8C 0x0000008C + +#define NV50_EVO_DAC(n, r) ((n) * 0x80 + NV50_EVO_DAC_##r) +#define NV50_EVO_DAC_MODE_CTRL 0x00000400 +#define NV50_EVO_DAC_MODE_CTRL_CRTC0 0x00000001 +#define NV50_EVO_DAC_MODE_CTRL_CRTC1 0x00000002 +#define NV50_EVO_DAC_MODE_CTRL2 0x00000404 +#define NV50_EVO_DAC_MODE_CTRL2_NHSYNC 0x00000001 +#define NV50_EVO_DAC_MODE_CTRL2_NVSYNC 0x00000002 + +#define NV50_EVO_SOR(n, r) ((n) * 0x40 + NV50_EVO_SOR_##r) +#define NV50_EVO_SOR_MODE_CTRL 0x00000600 +#define NV50_EVO_SOR_MODE_CTRL_CRTC0 0x00000001 +#define NV50_EVO_SOR_MODE_CTRL_CRTC1 0x00000002 +#define NV50_EVO_SOR_MODE_CTRL_TMDS 0x00000100 +#define NV50_EVO_SOR_MODE_CTRL_TMDS_DUAL_LINK 0x00000400 +#define NV50_EVO_SOR_MODE_CTRL_NHSYNC 0x00001000 +#define NV50_EVO_SOR_MODE_CTRL_NVSYNC 0x00002000 + +#define NV50_EVO_CRTC(n, r) ((n) * 0x400 + NV50_EVO_CRTC_##r) +#define NV84_EVO_CRTC(n, r) ((n) * 0x400 + NV84_EVO_CRTC_##r) +#define NV50_EVO_CRTC_UNK0800 0x00000800 +#define NV50_EVO_CRTC_CLOCK 0x00000804 +#define NV50_EVO_CRTC_INTERLACE 0x00000808 +#define NV50_EVO_CRTC_DISPLAY_START 0x00000810 +#define NV50_EVO_CRTC_DISPLAY_TOTAL 0x00000814 +#define NV50_EVO_CRTC_SYNC_DURATION 0x00000818 +#define NV50_EVO_CRTC_SYNC_START_TO_BLANK_END 0x0000081c +#define NV50_EVO_CRTC_UNK0820 0x00000820 +#define NV50_EVO_CRTC_UNK0824 0x00000824 +#define NV50_EVO_CRTC_UNK082C 0x0000082c +#define NV50_EVO_CRTC_CLUT_MODE 0x00000840 +/* You can't have a palette in 8 bit mode (=OFF) */ +#define NV50_EVO_CRTC_CLUT_MODE_BLANK 0x00000000 +#define NV50_EVO_CRTC_CLUT_MODE_OFF 0x80000000 +#define NV50_EVO_CRTC_CLUT_MODE_ON 0xC0000000 +#define NV50_EVO_CRTC_CLUT_OFFSET 0x00000844 +#define NV84_EVO_CRTC_CLUT_DMA 0x0000085C +#define NV84_EVO_CRTC_CLUT_DMA_HANDLE 0xffffffff +#define NV84_EVO_CRTC_CLUT_DMA_HANDLE_NONE 0x00000000 +#define NV50_EVO_CRTC_FB_OFFSET 0x00000860 +#define NV50_EVO_CRTC_FB_SIZE 0x00000868 +#define NV50_EVO_CRTC_FB_CONFIG 0x0000086c +#define NV50_EVO_CRTC_FB_CONFIG_MODE 0x00100000 +#define NV50_EVO_CRTC_FB_CONFIG_MODE_TILE 0x00000000 +#define NV50_EVO_CRTC_FB_CONFIG_MODE_PITCH 0x00100000 +#define NV50_EVO_CRTC_FB_DEPTH 0x00000870 +#define NV50_EVO_CRTC_FB_DEPTH_8 0x00001e00 +#define NV50_EVO_CRTC_FB_DEPTH_15 0x0000e900 +#define NV50_EVO_CRTC_FB_DEPTH_16 0x0000e800 +#define NV50_EVO_CRTC_FB_DEPTH_24 0x0000cf00 +#define NV50_EVO_CRTC_FB_DEPTH_30 0x0000d100 +#define NV50_EVO_CRTC_FB_DMA 0x00000874 +#define NV50_EVO_CRTC_FB_DMA_HANDLE 0xffffffff +#define NV50_EVO_CRTC_FB_DMA_HANDLE_NONE 0x00000000 +#define NV50_EVO_CRTC_CURSOR_CTRL 0x00000880 +#define NV50_EVO_CRTC_CURSOR_CTRL_HIDE 0x05000000 +#define NV50_EVO_CRTC_CURSOR_CTRL_SHOW 0x85000000 +#define NV50_EVO_CRTC_CURSOR_OFFSET 0x00000884 +#define NV84_EVO_CRTC_CURSOR_DMA 0x0000089c +#define NV84_EVO_CRTC_CURSOR_DMA_HANDLE 0xffffffff +#define NV84_EVO_CRTC_CURSOR_DMA_HANDLE_NONE 0x00000000 +#define NV50_EVO_CRTC_DITHER_CTRL 0x000008a0 +#define NV50_EVO_CRTC_DITHER_CTRL_OFF 0x00000000 +#define NV50_EVO_CRTC_DITHER_CTRL_ON 0x00000011 +#define NV50_EVO_CRTC_SCALE_CTRL 0x000008a4 +#define NV50_EVO_CRTC_SCALE_CTRL_INACTIVE 0x00000000 +#define NV50_EVO_CRTC_SCALE_CTRL_ACTIVE 0x00000009 +#define NV50_EVO_CRTC_COLOR_CTRL 0x000008a8 +#define NV50_EVO_CRTC_COLOR_CTRL_COLOR 0x00040000 +#define NV50_EVO_CRTC_FB_POS 0x000008c0 +#define NV50_EVO_CRTC_REAL_RES 0x000008c8 +#define NV50_EVO_CRTC_SCALE_CENTER_OFFSET 0x000008d4 +#define NV50_EVO_CRTC_SCALE_CENTER_OFFSET_VAL(x, y) \ + ((((unsigned)y << 16) & 0xFFFF0000) | (((unsigned)x) & 0x0000FFFF)) +/* Both of these are needed, otherwise nothing happens. */ +#define NV50_EVO_CRTC_SCALE_RES1 0x000008d8 +#define NV50_EVO_CRTC_SCALE_RES2 0x000008dc +#define NV50_EVO_CRTC_UNK900 0x00000900 +#define NV50_EVO_CRTC_UNK904 0x00000904 + +#endif diff --git a/drivers/gpu/drm/nouveau/nv50_fb.c b/drivers/gpu/drm/nouveau/nv50_fb.c new file mode 100644 index 00000000..bdd2afe2 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv50_fb.c @@ -0,0 +1,294 @@ +#include "drmP.h" +#include "drm.h" +#include "nouveau_drv.h" +#include "nouveau_drm.h" + +struct nv50_fb_priv { + struct page *r100c08_page; + dma_addr_t r100c08; +}; + +static void +nv50_fb_destroy(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fb_engine *pfb = &dev_priv->engine.fb; + struct nv50_fb_priv *priv = pfb->priv; + + if (drm_mm_initialized(&pfb->tag_heap)) + drm_mm_takedown(&pfb->tag_heap); + + if (priv->r100c08_page) { + pci_unmap_page(dev->pdev, priv->r100c08, PAGE_SIZE, + PCI_DMA_BIDIRECTIONAL); + __free_page(priv->r100c08_page); + } + + kfree(priv); + pfb->priv = NULL; +} + +static int +nv50_fb_create(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fb_engine *pfb = &dev_priv->engine.fb; + struct nv50_fb_priv *priv; + u32 tagmem; + int ret; + + priv = kzalloc(sizeof(*priv), GFP_KERNEL); + if (!priv) + return -ENOMEM; + pfb->priv = priv; + + priv->r100c08_page = alloc_page(GFP_KERNEL | __GFP_ZERO); + if (!priv->r100c08_page) { + nv50_fb_destroy(dev); + return -ENOMEM; + } + + priv->r100c08 = pci_map_page(dev->pdev, priv->r100c08_page, 0, + PAGE_SIZE, PCI_DMA_BIDIRECTIONAL); + if (pci_dma_mapping_error(dev->pdev, priv->r100c08)) { + nv50_fb_destroy(dev); + return -EFAULT; + } + + tagmem = nv_rd32(dev, 0x100320); + NV_DEBUG(dev, "%d tags available\n", tagmem); + ret = drm_mm_init(&pfb->tag_heap, 0, tagmem); + if (ret) { + nv50_fb_destroy(dev); + return ret; + } + + return 0; +} + +int +nv50_fb_init(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nv50_fb_priv *priv; + int ret; + + if (!dev_priv->engine.fb.priv) { + ret = nv50_fb_create(dev); + if (ret) + return ret; + } + priv = dev_priv->engine.fb.priv; + + /* Not a clue what this is exactly. Without pointing it at a + * scratch page, VRAM->GART blits with M2MF (as in DDX DFS) + * cause IOMMU "read from address 0" errors (rh#561267) + */ + nv_wr32(dev, 0x100c08, priv->r100c08 >> 8); + + /* This is needed to get meaningful information from 100c90 + * on traps. No idea what these values mean exactly. */ + switch (dev_priv->chipset) { + case 0x50: + nv_wr32(dev, 0x100c90, 0x000707ff); + break; + case 0xa3: + case 0xa5: + case 0xa8: + nv_wr32(dev, 0x100c90, 0x000d0fff); + break; + case 0xaf: + nv_wr32(dev, 0x100c90, 0x089d1fff); + break; + default: + nv_wr32(dev, 0x100c90, 0x001d07ff); + break; + } + + return 0; +} + +void +nv50_fb_takedown(struct drm_device *dev) +{ + nv50_fb_destroy(dev); +} + +static struct nouveau_enum vm_dispatch_subclients[] = { + { 0x00000000, "GRCTX", NULL }, + { 0x00000001, "NOTIFY", NULL }, + { 0x00000002, "QUERY", NULL }, + { 0x00000003, "COND", NULL }, + { 0x00000004, "M2M_IN", NULL }, + { 0x00000005, "M2M_OUT", NULL }, + { 0x00000006, "M2M_NOTIFY", NULL }, + {} +}; + +static struct nouveau_enum vm_ccache_subclients[] = { + { 0x00000000, "CB", NULL }, + { 0x00000001, "TIC", NULL }, + { 0x00000002, "TSC", NULL }, + {} +}; + +static struct nouveau_enum vm_prop_subclients[] = { + { 0x00000000, "RT0", NULL }, + { 0x00000001, "RT1", NULL }, + { 0x00000002, "RT2", NULL }, + { 0x00000003, "RT3", NULL }, + { 0x00000004, "RT4", NULL }, + { 0x00000005, "RT5", NULL }, + { 0x00000006, "RT6", NULL }, + { 0x00000007, "RT7", NULL }, + { 0x00000008, "ZETA", NULL }, + { 0x00000009, "LOCAL", NULL }, + { 0x0000000a, "GLOBAL", NULL }, + { 0x0000000b, "STACK", NULL }, + { 0x0000000c, "DST2D", NULL }, + {} +}; + +static struct nouveau_enum vm_pfifo_subclients[] = { + { 0x00000000, "PUSHBUF", NULL }, + { 0x00000001, "SEMAPHORE", NULL }, + {} +}; + +static struct nouveau_enum vm_bar_subclients[] = { + { 0x00000000, "FB", NULL }, + { 0x00000001, "IN", NULL }, + {} +}; + +static struct nouveau_enum vm_client[] = { + { 0x00000000, "STRMOUT", NULL }, + { 0x00000003, "DISPATCH", vm_dispatch_subclients }, + { 0x00000004, "PFIFO_WRITE", NULL }, + { 0x00000005, "CCACHE", vm_ccache_subclients }, + { 0x00000006, "PPPP", NULL }, + { 0x00000007, "CLIPID", NULL }, + { 0x00000008, "PFIFO_READ", NULL }, + { 0x00000009, "VFETCH", NULL }, + { 0x0000000a, "TEXTURE", NULL }, + { 0x0000000b, "PROP", vm_prop_subclients }, + { 0x0000000c, "PVP", NULL }, + { 0x0000000d, "PBSP", NULL }, + { 0x0000000e, "PCRYPT", NULL }, + { 0x0000000f, "PCOUNTER", NULL }, + { 0x00000011, "PDAEMON", NULL }, + {} +}; + +static struct nouveau_enum vm_engine[] = { + { 0x00000000, "PGRAPH", NULL }, + { 0x00000001, "PVP", NULL }, + { 0x00000004, "PEEPHOLE", NULL }, + { 0x00000005, "PFIFO", vm_pfifo_subclients }, + { 0x00000006, "BAR", vm_bar_subclients }, + { 0x00000008, "PPPP", NULL }, + { 0x00000009, "PBSP", NULL }, + { 0x0000000a, "PCRYPT", NULL }, + { 0x0000000b, "PCOUNTER", NULL }, + { 0x0000000c, "SEMAPHORE_BG", NULL }, + { 0x0000000d, "PCOPY", NULL }, + { 0x0000000e, "PDAEMON", NULL }, + {} +}; + +static struct nouveau_enum vm_fault[] = { + { 0x00000000, "PT_NOT_PRESENT", NULL }, + { 0x00000001, "PT_TOO_SHORT", NULL }, + { 0x00000002, "PAGE_NOT_PRESENT", NULL }, + { 0x00000003, "PAGE_SYSTEM_ONLY", NULL }, + { 0x00000004, "PAGE_READ_ONLY", NULL }, + { 0x00000006, "NULL_DMAOBJ", NULL }, + { 0x00000007, "WRONG_MEMTYPE", NULL }, + { 0x0000000b, "VRAM_LIMIT", NULL }, + { 0x0000000f, "DMAOBJ_LIMIT", NULL }, + {} +}; + +void +nv50_fb_vm_trap(struct drm_device *dev, int display) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + const struct nouveau_enum *en, *cl; + unsigned long flags; + u32 trap[6], idx, chinst; + u8 st0, st1, st2, st3; + int i, ch; + + idx = nv_rd32(dev, 0x100c90); + if (!(idx & 0x80000000)) + return; + idx &= 0x00ffffff; + + for (i = 0; i < 6; i++) { + nv_wr32(dev, 0x100c90, idx | i << 24); + trap[i] = nv_rd32(dev, 0x100c94); + } + nv_wr32(dev, 0x100c90, idx | 0x80000000); + + if (!display) + return; + + /* lookup channel id */ + chinst = (trap[2] << 16) | trap[1]; + spin_lock_irqsave(&dev_priv->channels.lock, flags); + for (ch = 0; ch < dev_priv->engine.fifo.channels; ch++) { + struct nouveau_channel *chan = dev_priv->channels.ptr[ch]; + + if (!chan || !chan->ramin) + continue; + + if (chinst == chan->ramin->vinst >> 12) + break; + } + spin_unlock_irqrestore(&dev_priv->channels.lock, flags); + + /* decode status bits into something more useful */ + if (dev_priv->chipset < 0xa3 || + dev_priv->chipset == 0xaa || dev_priv->chipset == 0xac) { + st0 = (trap[0] & 0x0000000f) >> 0; + st1 = (trap[0] & 0x000000f0) >> 4; + st2 = (trap[0] & 0x00000f00) >> 8; + st3 = (trap[0] & 0x0000f000) >> 12; + } else { + st0 = (trap[0] & 0x000000ff) >> 0; + st1 = (trap[0] & 0x0000ff00) >> 8; + st2 = (trap[0] & 0x00ff0000) >> 16; + st3 = (trap[0] & 0xff000000) >> 24; + } + + NV_INFO(dev, "VM: trapped %s at 0x%02x%04x%04x on ch %d [0x%08x] ", + (trap[5] & 0x00000100) ? "read" : "write", + trap[5] & 0xff, trap[4] & 0xffff, trap[3] & 0xffff, ch, chinst); + + en = nouveau_enum_find(vm_engine, st0); + if (en) + printk("%s/", en->name); + else + printk("%02x/", st0); + + cl = nouveau_enum_find(vm_client, st2); + if (cl) + printk("%s/", cl->name); + else + printk("%02x/", st2); + + if (cl && cl->data) cl = nouveau_enum_find(cl->data, st3); + else if (en && en->data) cl = nouveau_enum_find(en->data, st3); + else cl = NULL; + if (cl) + printk("%s", cl->name); + else + printk("%02x", st3); + + printk(" reason: "); + en = nouveau_enum_find(vm_fault, st1); + if (en) + printk("%s\n", en->name); + else + printk("0x%08x\n", st1); +} diff --git a/drivers/gpu/drm/nouveau/nv50_fbcon.c b/drivers/gpu/drm/nouveau/nv50_fbcon.c new file mode 100644 index 00000000..791ded1c --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv50_fbcon.c @@ -0,0 +1,264 @@ +/* + * Copyright 2010 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include "drmP.h" +#include "nouveau_drv.h" +#include "nouveau_dma.h" +#include "nouveau_ramht.h" +#include "nouveau_fbcon.h" +#include "nouveau_mm.h" + +int +nv50_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect) +{ + struct nouveau_fbdev *nfbdev = info->par; + struct drm_device *dev = nfbdev->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_channel *chan = dev_priv->channel; + int ret; + + ret = RING_SPACE(chan, rect->rop == ROP_COPY ? 7 : 11); + if (ret) + return ret; + + if (rect->rop != ROP_COPY) { + BEGIN_RING(chan, NvSub2D, 0x02ac, 1); + OUT_RING(chan, 1); + } + BEGIN_RING(chan, NvSub2D, 0x0588, 1); + if (info->fix.visual == FB_VISUAL_TRUECOLOR || + info->fix.visual == FB_VISUAL_DIRECTCOLOR) + OUT_RING(chan, ((uint32_t *)info->pseudo_palette)[rect->color]); + else + OUT_RING(chan, rect->color); + BEGIN_RING(chan, NvSub2D, 0x0600, 4); + OUT_RING(chan, rect->dx); + OUT_RING(chan, rect->dy); + OUT_RING(chan, rect->dx + rect->width); + OUT_RING(chan, rect->dy + rect->height); + if (rect->rop != ROP_COPY) { + BEGIN_RING(chan, NvSub2D, 0x02ac, 1); + OUT_RING(chan, 3); + } + FIRE_RING(chan); + return 0; +} + +int +nv50_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *region) +{ + struct nouveau_fbdev *nfbdev = info->par; + struct drm_device *dev = nfbdev->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_channel *chan = dev_priv->channel; + int ret; + + ret = RING_SPACE(chan, 12); + if (ret) + return ret; + + BEGIN_RING(chan, NvSub2D, 0x0110, 1); + OUT_RING(chan, 0); + BEGIN_RING(chan, NvSub2D, 0x08b0, 4); + OUT_RING(chan, region->dx); + OUT_RING(chan, region->dy); + OUT_RING(chan, region->width); + OUT_RING(chan, region->height); + BEGIN_RING(chan, NvSub2D, 0x08d0, 4); + OUT_RING(chan, 0); + OUT_RING(chan, region->sx); + OUT_RING(chan, 0); + OUT_RING(chan, region->sy); + FIRE_RING(chan); + return 0; +} + +int +nv50_fbcon_imageblit(struct fb_info *info, const struct fb_image *image) +{ + struct nouveau_fbdev *nfbdev = info->par; + struct drm_device *dev = nfbdev->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_channel *chan = dev_priv->channel; + uint32_t width, dwords, *data = (uint32_t *)image->data; + uint32_t mask = ~(~0 >> (32 - info->var.bits_per_pixel)); + uint32_t *palette = info->pseudo_palette; + int ret; + + if (image->depth != 1) + return -ENODEV; + + ret = RING_SPACE(chan, 11); + if (ret) + return ret; + + width = ALIGN(image->width, 32); + dwords = (width * image->height) >> 5; + + BEGIN_RING(chan, NvSub2D, 0x0814, 2); + if (info->fix.visual == FB_VISUAL_TRUECOLOR || + info->fix.visual == FB_VISUAL_DIRECTCOLOR) { + OUT_RING(chan, palette[image->bg_color] | mask); + OUT_RING(chan, palette[image->fg_color] | mask); + } else { + OUT_RING(chan, image->bg_color); + OUT_RING(chan, image->fg_color); + } + BEGIN_RING(chan, NvSub2D, 0x0838, 2); + OUT_RING(chan, image->width); + OUT_RING(chan, image->height); + BEGIN_RING(chan, NvSub2D, 0x0850, 4); + OUT_RING(chan, 0); + OUT_RING(chan, image->dx); + OUT_RING(chan, 0); + OUT_RING(chan, image->dy); + + while (dwords) { + int push = dwords > 2047 ? 2047 : dwords; + + ret = RING_SPACE(chan, push + 1); + if (ret) + return ret; + + dwords -= push; + + BEGIN_RING(chan, NvSub2D, 0x40000860, push); + OUT_RINGp(chan, data, push); + data += push; + } + + FIRE_RING(chan); + return 0; +} + +int +nv50_fbcon_accel_init(struct fb_info *info) +{ + struct nouveau_fbdev *nfbdev = info->par; + struct drm_device *dev = nfbdev->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_channel *chan = dev_priv->channel; + struct nouveau_bo *nvbo = nfbdev->nouveau_fb.nvbo; + int ret, format; + + switch (info->var.bits_per_pixel) { + case 8: + format = 0xf3; + break; + case 15: + format = 0xf8; + break; + case 16: + format = 0xe8; + break; + case 32: + switch (info->var.transp.length) { + case 0: /* depth 24 */ + case 8: /* depth 32, just use 24.. */ + format = 0xe6; + break; + case 2: /* depth 30 */ + format = 0xd1; + break; + default: + return -EINVAL; + } + break; + default: + return -EINVAL; + } + + ret = nouveau_gpuobj_gr_new(dev_priv->channel, Nv2D, 0x502d); + if (ret) + return ret; + + ret = RING_SPACE(chan, 59); + if (ret) { + nouveau_fbcon_gpu_lockup(info); + return ret; + } + + BEGIN_RING(chan, NvSub2D, 0x0000, 1); + OUT_RING(chan, Nv2D); + BEGIN_RING(chan, NvSub2D, 0x0180, 4); + OUT_RING(chan, NvNotify0); + OUT_RING(chan, chan->vram_handle); + OUT_RING(chan, chan->vram_handle); + OUT_RING(chan, chan->vram_handle); + BEGIN_RING(chan, NvSub2D, 0x0290, 1); + OUT_RING(chan, 0); + BEGIN_RING(chan, NvSub2D, 0x0888, 1); + OUT_RING(chan, 1); + BEGIN_RING(chan, NvSub2D, 0x02ac, 1); + OUT_RING(chan, 3); + BEGIN_RING(chan, NvSub2D, 0x02a0, 1); + OUT_RING(chan, 0x55); + BEGIN_RING(chan, NvSub2D, 0x08c0, 4); + OUT_RING(chan, 0); + OUT_RING(chan, 1); + OUT_RING(chan, 0); + OUT_RING(chan, 1); + BEGIN_RING(chan, NvSub2D, 0x0580, 2); + OUT_RING(chan, 4); + OUT_RING(chan, format); + BEGIN_RING(chan, NvSub2D, 0x02e8, 2); + OUT_RING(chan, 2); + OUT_RING(chan, 1); + BEGIN_RING(chan, NvSub2D, 0x0804, 1); + OUT_RING(chan, format); + BEGIN_RING(chan, NvSub2D, 0x0800, 1); + OUT_RING(chan, 1); + BEGIN_RING(chan, NvSub2D, 0x0808, 3); + OUT_RING(chan, 0); + OUT_RING(chan, 0); + OUT_RING(chan, 1); + BEGIN_RING(chan, NvSub2D, 0x081c, 1); + OUT_RING(chan, 1); + BEGIN_RING(chan, NvSub2D, 0x0840, 4); + OUT_RING(chan, 0); + OUT_RING(chan, 1); + OUT_RING(chan, 0); + OUT_RING(chan, 1); + BEGIN_RING(chan, NvSub2D, 0x0200, 2); + OUT_RING(chan, format); + OUT_RING(chan, 1); + BEGIN_RING(chan, NvSub2D, 0x0214, 5); + OUT_RING(chan, info->fix.line_length); + OUT_RING(chan, info->var.xres_virtual); + OUT_RING(chan, info->var.yres_virtual); + OUT_RING(chan, upper_32_bits(nvbo->vma.offset)); + OUT_RING(chan, lower_32_bits(nvbo->vma.offset)); + BEGIN_RING(chan, NvSub2D, 0x0230, 2); + OUT_RING(chan, format); + OUT_RING(chan, 1); + BEGIN_RING(chan, NvSub2D, 0x0244, 5); + OUT_RING(chan, info->fix.line_length); + OUT_RING(chan, info->var.xres_virtual); + OUT_RING(chan, info->var.yres_virtual); + OUT_RING(chan, upper_32_bits(nvbo->vma.offset)); + OUT_RING(chan, lower_32_bits(nvbo->vma.offset)); + + return 0; +} + diff --git a/drivers/gpu/drm/nouveau/nv50_fifo.c b/drivers/gpu/drm/nouveau/nv50_fifo.c new file mode 100644 index 00000000..c34a074f --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv50_fifo.c @@ -0,0 +1,504 @@ +/* + * Copyright (C) 2007 Ben Skeggs. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "drmP.h" +#include "drm.h" +#include "nouveau_drv.h" +#include "nouveau_ramht.h" +#include "nouveau_vm.h" + +static void +nv50_fifo_playlist_update(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; + struct nouveau_gpuobj *cur; + int i, nr; + + NV_DEBUG(dev, "\n"); + + cur = pfifo->playlist[pfifo->cur_playlist]; + pfifo->cur_playlist = !pfifo->cur_playlist; + + /* We never schedule channel 0 or 127 */ + for (i = 1, nr = 0; i < 127; i++) { + if (dev_priv->channels.ptr[i] && + dev_priv->channels.ptr[i]->ramfc) { + nv_wo32(cur, (nr * 4), i); + nr++; + } + } + dev_priv->engine.instmem.flush(dev); + + nv_wr32(dev, 0x32f4, cur->vinst >> 12); + nv_wr32(dev, 0x32ec, nr); + nv_wr32(dev, 0x2500, 0x101); +} + +static void +nv50_fifo_channel_enable(struct drm_device *dev, int channel) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_channel *chan = dev_priv->channels.ptr[channel]; + uint32_t inst; + + NV_DEBUG(dev, "ch%d\n", channel); + + if (dev_priv->chipset == 0x50) + inst = chan->ramfc->vinst >> 12; + else + inst = chan->ramfc->vinst >> 8; + + nv_wr32(dev, NV50_PFIFO_CTX_TABLE(channel), inst | + NV50_PFIFO_CTX_TABLE_CHANNEL_ENABLED); +} + +static void +nv50_fifo_channel_disable(struct drm_device *dev, int channel) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + uint32_t inst; + + NV_DEBUG(dev, "ch%d\n", channel); + + if (dev_priv->chipset == 0x50) + inst = NV50_PFIFO_CTX_TABLE_INSTANCE_MASK_G80; + else + inst = NV50_PFIFO_CTX_TABLE_INSTANCE_MASK_G84; + nv_wr32(dev, NV50_PFIFO_CTX_TABLE(channel), inst); +} + +static void +nv50_fifo_init_reset(struct drm_device *dev) +{ + uint32_t pmc_e = NV_PMC_ENABLE_PFIFO; + + NV_DEBUG(dev, "\n"); + + nv_wr32(dev, NV03_PMC_ENABLE, nv_rd32(dev, NV03_PMC_ENABLE) & ~pmc_e); + nv_wr32(dev, NV03_PMC_ENABLE, nv_rd32(dev, NV03_PMC_ENABLE) | pmc_e); +} + +static void +nv50_fifo_init_intr(struct drm_device *dev) +{ + NV_DEBUG(dev, "\n"); + + nouveau_irq_register(dev, 8, nv04_fifo_isr); + nv_wr32(dev, NV03_PFIFO_INTR_0, 0xFFFFFFFF); + nv_wr32(dev, NV03_PFIFO_INTR_EN_0, 0xFFFFFFFF); +} + +static void +nv50_fifo_init_context_table(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + int i; + + NV_DEBUG(dev, "\n"); + + for (i = 0; i < NV50_PFIFO_CTX_TABLE__SIZE; i++) { + if (dev_priv->channels.ptr[i]) + nv50_fifo_channel_enable(dev, i); + else + nv50_fifo_channel_disable(dev, i); + } + + nv50_fifo_playlist_update(dev); +} + +static void +nv50_fifo_init_regs__nv(struct drm_device *dev) +{ + NV_DEBUG(dev, "\n"); + + nv_wr32(dev, 0x250c, 0x6f3cfc34); +} + +static void +nv50_fifo_init_regs(struct drm_device *dev) +{ + NV_DEBUG(dev, "\n"); + + nv_wr32(dev, 0x2500, 0); + nv_wr32(dev, 0x3250, 0); + nv_wr32(dev, 0x3220, 0); + nv_wr32(dev, 0x3204, 0); + nv_wr32(dev, 0x3210, 0); + nv_wr32(dev, 0x3270, 0); + nv_wr32(dev, 0x2044, 0x01003fff); + + /* Enable dummy channels setup by nv50_instmem.c */ + nv50_fifo_channel_enable(dev, 0); + nv50_fifo_channel_enable(dev, 127); +} + +int +nv50_fifo_init(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; + int ret; + + NV_DEBUG(dev, "\n"); + + if (pfifo->playlist[0]) { + pfifo->cur_playlist = !pfifo->cur_playlist; + goto just_reset; + } + + ret = nouveau_gpuobj_new(dev, NULL, 128*4, 0x1000, + NVOBJ_FLAG_ZERO_ALLOC, + &pfifo->playlist[0]); + if (ret) { + NV_ERROR(dev, "error creating playlist 0: %d\n", ret); + return ret; + } + + ret = nouveau_gpuobj_new(dev, NULL, 128*4, 0x1000, + NVOBJ_FLAG_ZERO_ALLOC, + &pfifo->playlist[1]); + if (ret) { + nouveau_gpuobj_ref(NULL, &pfifo->playlist[0]); + NV_ERROR(dev, "error creating playlist 1: %d\n", ret); + return ret; + } + +just_reset: + nv50_fifo_init_reset(dev); + nv50_fifo_init_intr(dev); + nv50_fifo_init_context_table(dev); + nv50_fifo_init_regs__nv(dev); + nv50_fifo_init_regs(dev); + dev_priv->engine.fifo.enable(dev); + dev_priv->engine.fifo.reassign(dev, true); + + return 0; +} + +void +nv50_fifo_takedown(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; + + NV_DEBUG(dev, "\n"); + + if (!pfifo->playlist[0]) + return; + + nv_wr32(dev, 0x2140, 0x00000000); + nouveau_irq_unregister(dev, 8); + + nouveau_gpuobj_ref(NULL, &pfifo->playlist[0]); + nouveau_gpuobj_ref(NULL, &pfifo->playlist[1]); +} + +int +nv50_fifo_channel_id(struct drm_device *dev) +{ + return nv_rd32(dev, NV03_PFIFO_CACHE1_PUSH1) & + NV50_PFIFO_CACHE1_PUSH1_CHID_MASK; +} + +int +nv50_fifo_create_context(struct nouveau_channel *chan) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *ramfc = NULL; + unsigned long flags; + int ret; + + NV_DEBUG(dev, "ch%d\n", chan->id); + + if (dev_priv->chipset == 0x50) { + ret = nouveau_gpuobj_new_fake(dev, chan->ramin->pinst, + chan->ramin->vinst, 0x100, + NVOBJ_FLAG_ZERO_ALLOC | + NVOBJ_FLAG_ZERO_FREE, + &chan->ramfc); + if (ret) + return ret; + + ret = nouveau_gpuobj_new_fake(dev, chan->ramin->pinst + 0x0400, + chan->ramin->vinst + 0x0400, + 4096, 0, &chan->cache); + if (ret) + return ret; + } else { + ret = nouveau_gpuobj_new(dev, chan, 0x100, 256, + NVOBJ_FLAG_ZERO_ALLOC | + NVOBJ_FLAG_ZERO_FREE, &chan->ramfc); + if (ret) + return ret; + + ret = nouveau_gpuobj_new(dev, chan, 4096, 1024, + 0, &chan->cache); + if (ret) + return ret; + } + ramfc = chan->ramfc; + + chan->user = ioremap(pci_resource_start(dev->pdev, 0) + + NV50_USER(chan->id), PAGE_SIZE); + if (!chan->user) + return -ENOMEM; + + spin_lock_irqsave(&dev_priv->context_switch_lock, flags); + + nv_wo32(ramfc, 0x48, chan->pushbuf->cinst >> 4); + nv_wo32(ramfc, 0x80, ((chan->ramht->bits - 9) << 27) | + (4 << 24) /* SEARCH_FULL */ | + (chan->ramht->gpuobj->cinst >> 4)); + nv_wo32(ramfc, 0x44, 0x01003fff); + nv_wo32(ramfc, 0x60, 0x7fffffff); + nv_wo32(ramfc, 0x40, 0x00000000); + nv_wo32(ramfc, 0x7c, 0x30000001); + nv_wo32(ramfc, 0x78, 0x00000000); + nv_wo32(ramfc, 0x3c, 0x403f6078); + nv_wo32(ramfc, 0x50, chan->pushbuf_base + chan->dma.ib_base * 4); + nv_wo32(ramfc, 0x54, drm_order(chan->dma.ib_max + 1) << 16); + + if (dev_priv->chipset != 0x50) { + nv_wo32(chan->ramin, 0, chan->id); + nv_wo32(chan->ramin, 4, chan->ramfc->vinst >> 8); + + nv_wo32(ramfc, 0x88, chan->cache->vinst >> 10); + nv_wo32(ramfc, 0x98, chan->ramin->vinst >> 12); + } + + dev_priv->engine.instmem.flush(dev); + + nv50_fifo_channel_enable(dev, chan->id); + nv50_fifo_playlist_update(dev); + spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); + return 0; +} + +void +nv50_fifo_destroy_context(struct nouveau_channel *chan) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; + struct nouveau_gpuobj *ramfc = NULL; + unsigned long flags; + + NV_DEBUG(dev, "ch%d\n", chan->id); + + spin_lock_irqsave(&dev_priv->context_switch_lock, flags); + pfifo->reassign(dev, false); + + /* Unload the context if it's the currently active one */ + if (pfifo->channel_id(dev) == chan->id) { + pfifo->disable(dev); + pfifo->unload_context(dev); + pfifo->enable(dev); + } + + /* This will ensure the channel is seen as disabled. */ + nouveau_gpuobj_ref(chan->ramfc, &ramfc); + nouveau_gpuobj_ref(NULL, &chan->ramfc); + nv50_fifo_channel_disable(dev, chan->id); + + /* Dummy channel, also used on ch 127 */ + if (chan->id == 0) + nv50_fifo_channel_disable(dev, 127); + nv50_fifo_playlist_update(dev); + + pfifo->reassign(dev, true); + spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); + + /* Free the channel resources */ + if (chan->user) { + iounmap(chan->user); + chan->user = NULL; + } + nouveau_gpuobj_ref(NULL, &ramfc); + nouveau_gpuobj_ref(NULL, &chan->cache); +} + +int +nv50_fifo_load_context(struct nouveau_channel *chan) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *ramfc = chan->ramfc; + struct nouveau_gpuobj *cache = chan->cache; + int ptr, cnt; + + NV_DEBUG(dev, "ch%d\n", chan->id); + + nv_wr32(dev, 0x3330, nv_ro32(ramfc, 0x00)); + nv_wr32(dev, 0x3334, nv_ro32(ramfc, 0x04)); + nv_wr32(dev, 0x3240, nv_ro32(ramfc, 0x08)); + nv_wr32(dev, 0x3320, nv_ro32(ramfc, 0x0c)); + nv_wr32(dev, 0x3244, nv_ro32(ramfc, 0x10)); + nv_wr32(dev, 0x3328, nv_ro32(ramfc, 0x14)); + nv_wr32(dev, 0x3368, nv_ro32(ramfc, 0x18)); + nv_wr32(dev, 0x336c, nv_ro32(ramfc, 0x1c)); + nv_wr32(dev, 0x3370, nv_ro32(ramfc, 0x20)); + nv_wr32(dev, 0x3374, nv_ro32(ramfc, 0x24)); + nv_wr32(dev, 0x3378, nv_ro32(ramfc, 0x28)); + nv_wr32(dev, 0x337c, nv_ro32(ramfc, 0x2c)); + nv_wr32(dev, 0x3228, nv_ro32(ramfc, 0x30)); + nv_wr32(dev, 0x3364, nv_ro32(ramfc, 0x34)); + nv_wr32(dev, 0x32a0, nv_ro32(ramfc, 0x38)); + nv_wr32(dev, 0x3224, nv_ro32(ramfc, 0x3c)); + nv_wr32(dev, 0x324c, nv_ro32(ramfc, 0x40)); + nv_wr32(dev, 0x2044, nv_ro32(ramfc, 0x44)); + nv_wr32(dev, 0x322c, nv_ro32(ramfc, 0x48)); + nv_wr32(dev, 0x3234, nv_ro32(ramfc, 0x4c)); + nv_wr32(dev, 0x3340, nv_ro32(ramfc, 0x50)); + nv_wr32(dev, 0x3344, nv_ro32(ramfc, 0x54)); + nv_wr32(dev, 0x3280, nv_ro32(ramfc, 0x58)); + nv_wr32(dev, 0x3254, nv_ro32(ramfc, 0x5c)); + nv_wr32(dev, 0x3260, nv_ro32(ramfc, 0x60)); + nv_wr32(dev, 0x3264, nv_ro32(ramfc, 0x64)); + nv_wr32(dev, 0x3268, nv_ro32(ramfc, 0x68)); + nv_wr32(dev, 0x326c, nv_ro32(ramfc, 0x6c)); + nv_wr32(dev, 0x32e4, nv_ro32(ramfc, 0x70)); + nv_wr32(dev, 0x3248, nv_ro32(ramfc, 0x74)); + nv_wr32(dev, 0x2088, nv_ro32(ramfc, 0x78)); + nv_wr32(dev, 0x2058, nv_ro32(ramfc, 0x7c)); + nv_wr32(dev, 0x2210, nv_ro32(ramfc, 0x80)); + + cnt = nv_ro32(ramfc, 0x84); + for (ptr = 0; ptr < cnt; ptr++) { + nv_wr32(dev, NV40_PFIFO_CACHE1_METHOD(ptr), + nv_ro32(cache, (ptr * 8) + 0)); + nv_wr32(dev, NV40_PFIFO_CACHE1_DATA(ptr), + nv_ro32(cache, (ptr * 8) + 4)); + } + nv_wr32(dev, NV03_PFIFO_CACHE1_PUT, cnt << 2); + nv_wr32(dev, NV03_PFIFO_CACHE1_GET, 0); + + /* guessing that all the 0x34xx regs aren't on NV50 */ + if (dev_priv->chipset != 0x50) { + nv_wr32(dev, 0x340c, nv_ro32(ramfc, 0x88)); + nv_wr32(dev, 0x3400, nv_ro32(ramfc, 0x8c)); + nv_wr32(dev, 0x3404, nv_ro32(ramfc, 0x90)); + nv_wr32(dev, 0x3408, nv_ro32(ramfc, 0x94)); + nv_wr32(dev, 0x3410, nv_ro32(ramfc, 0x98)); + } + + nv_wr32(dev, NV03_PFIFO_CACHE1_PUSH1, chan->id | (1<<16)); + return 0; +} + +int +nv50_fifo_unload_context(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; + struct nouveau_gpuobj *ramfc, *cache; + struct nouveau_channel *chan = NULL; + int chid, get, put, ptr; + + NV_DEBUG(dev, "\n"); + + chid = pfifo->channel_id(dev); + if (chid < 1 || chid >= dev_priv->engine.fifo.channels - 1) + return 0; + + chan = dev_priv->channels.ptr[chid]; + if (!chan) { + NV_ERROR(dev, "Inactive channel on PFIFO: %d\n", chid); + return -EINVAL; + } + NV_DEBUG(dev, "ch%d\n", chan->id); + ramfc = chan->ramfc; + cache = chan->cache; + + nv_wo32(ramfc, 0x00, nv_rd32(dev, 0x3330)); + nv_wo32(ramfc, 0x04, nv_rd32(dev, 0x3334)); + nv_wo32(ramfc, 0x08, nv_rd32(dev, 0x3240)); + nv_wo32(ramfc, 0x0c, nv_rd32(dev, 0x3320)); + nv_wo32(ramfc, 0x10, nv_rd32(dev, 0x3244)); + nv_wo32(ramfc, 0x14, nv_rd32(dev, 0x3328)); + nv_wo32(ramfc, 0x18, nv_rd32(dev, 0x3368)); + nv_wo32(ramfc, 0x1c, nv_rd32(dev, 0x336c)); + nv_wo32(ramfc, 0x20, nv_rd32(dev, 0x3370)); + nv_wo32(ramfc, 0x24, nv_rd32(dev, 0x3374)); + nv_wo32(ramfc, 0x28, nv_rd32(dev, 0x3378)); + nv_wo32(ramfc, 0x2c, nv_rd32(dev, 0x337c)); + nv_wo32(ramfc, 0x30, nv_rd32(dev, 0x3228)); + nv_wo32(ramfc, 0x34, nv_rd32(dev, 0x3364)); + nv_wo32(ramfc, 0x38, nv_rd32(dev, 0x32a0)); + nv_wo32(ramfc, 0x3c, nv_rd32(dev, 0x3224)); + nv_wo32(ramfc, 0x40, nv_rd32(dev, 0x324c)); + nv_wo32(ramfc, 0x44, nv_rd32(dev, 0x2044)); + nv_wo32(ramfc, 0x48, nv_rd32(dev, 0x322c)); + nv_wo32(ramfc, 0x4c, nv_rd32(dev, 0x3234)); + nv_wo32(ramfc, 0x50, nv_rd32(dev, 0x3340)); + nv_wo32(ramfc, 0x54, nv_rd32(dev, 0x3344)); + nv_wo32(ramfc, 0x58, nv_rd32(dev, 0x3280)); + nv_wo32(ramfc, 0x5c, nv_rd32(dev, 0x3254)); + nv_wo32(ramfc, 0x60, nv_rd32(dev, 0x3260)); + nv_wo32(ramfc, 0x64, nv_rd32(dev, 0x3264)); + nv_wo32(ramfc, 0x68, nv_rd32(dev, 0x3268)); + nv_wo32(ramfc, 0x6c, nv_rd32(dev, 0x326c)); + nv_wo32(ramfc, 0x70, nv_rd32(dev, 0x32e4)); + nv_wo32(ramfc, 0x74, nv_rd32(dev, 0x3248)); + nv_wo32(ramfc, 0x78, nv_rd32(dev, 0x2088)); + nv_wo32(ramfc, 0x7c, nv_rd32(dev, 0x2058)); + nv_wo32(ramfc, 0x80, nv_rd32(dev, 0x2210)); + + put = (nv_rd32(dev, NV03_PFIFO_CACHE1_PUT) & 0x7ff) >> 2; + get = (nv_rd32(dev, NV03_PFIFO_CACHE1_GET) & 0x7ff) >> 2; + ptr = 0; + while (put != get) { + nv_wo32(cache, ptr + 0, + nv_rd32(dev, NV40_PFIFO_CACHE1_METHOD(get))); + nv_wo32(cache, ptr + 4, + nv_rd32(dev, NV40_PFIFO_CACHE1_DATA(get))); + get = (get + 1) & 0x1ff; + ptr += 8; + } + + /* guessing that all the 0x34xx regs aren't on NV50 */ + if (dev_priv->chipset != 0x50) { + nv_wo32(ramfc, 0x84, ptr >> 3); + nv_wo32(ramfc, 0x88, nv_rd32(dev, 0x340c)); + nv_wo32(ramfc, 0x8c, nv_rd32(dev, 0x3400)); + nv_wo32(ramfc, 0x90, nv_rd32(dev, 0x3404)); + nv_wo32(ramfc, 0x94, nv_rd32(dev, 0x3408)); + nv_wo32(ramfc, 0x98, nv_rd32(dev, 0x3410)); + } + + dev_priv->engine.instmem.flush(dev); + + /*XXX: probably reload ch127 (NULL) state back too */ + nv_wr32(dev, NV03_PFIFO_CACHE1_PUSH1, 127); + return 0; +} + +void +nv50_fifo_tlb_flush(struct drm_device *dev) +{ + nv50_vm_flush_engine(dev, 5); +} diff --git a/drivers/gpu/drm/nouveau/nv50_gpio.c b/drivers/gpu/drm/nouveau/nv50_gpio.c new file mode 100644 index 00000000..d4f4206d --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv50_gpio.c @@ -0,0 +1,302 @@ +/* + * Copyright 2010 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include "drmP.h" +#include "nouveau_drv.h" +#include "nouveau_hw.h" + +#include "nv50_display.h" + +static void nv50_gpio_isr(struct drm_device *dev); +static void nv50_gpio_isr_bh(struct work_struct *work); + +struct nv50_gpio_priv { + struct list_head handlers; + spinlock_t lock; +}; + +struct nv50_gpio_handler { + struct drm_device *dev; + struct list_head head; + struct work_struct work; + bool inhibit; + + struct dcb_gpio_entry *gpio; + + void (*handler)(void *data, int state); + void *data; +}; + +static int +nv50_gpio_location(struct dcb_gpio_entry *gpio, uint32_t *reg, uint32_t *shift) +{ + const uint32_t nv50_gpio_reg[4] = { 0xe104, 0xe108, 0xe280, 0xe284 }; + + if (gpio->line >= 32) + return -EINVAL; + + *reg = nv50_gpio_reg[gpio->line >> 3]; + *shift = (gpio->line & 7) << 2; + return 0; +} + +int +nv50_gpio_get(struct drm_device *dev, enum dcb_gpio_tag tag) +{ + struct dcb_gpio_entry *gpio; + uint32_t r, s, v; + + gpio = nouveau_bios_gpio_entry(dev, tag); + if (!gpio) + return -ENOENT; + + if (nv50_gpio_location(gpio, &r, &s)) + return -EINVAL; + + v = nv_rd32(dev, r) >> (s + 2); + return ((v & 1) == (gpio->state[1] & 1)); +} + +int +nv50_gpio_set(struct drm_device *dev, enum dcb_gpio_tag tag, int state) +{ + struct dcb_gpio_entry *gpio; + uint32_t r, s, v; + + gpio = nouveau_bios_gpio_entry(dev, tag); + if (!gpio) + return -ENOENT; + + if (nv50_gpio_location(gpio, &r, &s)) + return -EINVAL; + + v = nv_rd32(dev, r) & ~(0x3 << s); + v |= (gpio->state[state] ^ 2) << s; + nv_wr32(dev, r, v); + return 0; +} + +int +nv50_gpio_irq_register(struct drm_device *dev, enum dcb_gpio_tag tag, + void (*handler)(void *, int), void *data) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpio_engine *pgpio = &dev_priv->engine.gpio; + struct nv50_gpio_priv *priv = pgpio->priv; + struct nv50_gpio_handler *gpioh; + struct dcb_gpio_entry *gpio; + unsigned long flags; + + gpio = nouveau_bios_gpio_entry(dev, tag); + if (!gpio) + return -ENOENT; + + gpioh = kzalloc(sizeof(*gpioh), GFP_KERNEL); + if (!gpioh) + return -ENOMEM; + + INIT_WORK(&gpioh->work, nv50_gpio_isr_bh); + gpioh->dev = dev; + gpioh->gpio = gpio; + gpioh->handler = handler; + gpioh->data = data; + + spin_lock_irqsave(&priv->lock, flags); + list_add(&gpioh->head, &priv->handlers); + spin_unlock_irqrestore(&priv->lock, flags); + return 0; +} + +void +nv50_gpio_irq_unregister(struct drm_device *dev, enum dcb_gpio_tag tag, + void (*handler)(void *, int), void *data) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpio_engine *pgpio = &dev_priv->engine.gpio; + struct nv50_gpio_priv *priv = pgpio->priv; + struct nv50_gpio_handler *gpioh, *tmp; + struct dcb_gpio_entry *gpio; + LIST_HEAD(tofree); + unsigned long flags; + + gpio = nouveau_bios_gpio_entry(dev, tag); + if (!gpio) + return; + + spin_lock_irqsave(&priv->lock, flags); + list_for_each_entry_safe(gpioh, tmp, &priv->handlers, head) { + if (gpioh->gpio != gpio || + gpioh->handler != handler || + gpioh->data != data) + continue; + list_move(&gpioh->head, &tofree); + } + spin_unlock_irqrestore(&priv->lock, flags); + + list_for_each_entry_safe(gpioh, tmp, &tofree, head) { + flush_work_sync(&gpioh->work); + kfree(gpioh); + } +} + +bool +nv50_gpio_irq_enable(struct drm_device *dev, enum dcb_gpio_tag tag, bool on) +{ + struct dcb_gpio_entry *gpio; + u32 reg, mask; + + gpio = nouveau_bios_gpio_entry(dev, tag); + if (!gpio) + return false; + + reg = gpio->line < 16 ? 0xe050 : 0xe070; + mask = 0x00010001 << (gpio->line & 0xf); + + nv_wr32(dev, reg + 4, mask); + reg = nv_mask(dev, reg + 0, mask, on ? mask : 0); + return (reg & mask) == mask; +} + +static int +nv50_gpio_create(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpio_engine *pgpio = &dev_priv->engine.gpio; + struct nv50_gpio_priv *priv; + + priv = kzalloc(sizeof(*priv), GFP_KERNEL); + if (!priv) + return -ENOMEM; + + INIT_LIST_HEAD(&priv->handlers); + spin_lock_init(&priv->lock); + pgpio->priv = priv; + return 0; +} + +static void +nv50_gpio_destroy(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpio_engine *pgpio = &dev_priv->engine.gpio; + + kfree(pgpio->priv); + pgpio->priv = NULL; +} + +int +nv50_gpio_init(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpio_engine *pgpio = &dev_priv->engine.gpio; + int ret; + + if (!pgpio->priv) { + ret = nv50_gpio_create(dev); + if (ret) + return ret; + } + + /* disable, and ack any pending gpio interrupts */ + nv_wr32(dev, 0xe050, 0x00000000); + nv_wr32(dev, 0xe054, 0xffffffff); + if (dev_priv->chipset >= 0x90) { + nv_wr32(dev, 0xe070, 0x00000000); + nv_wr32(dev, 0xe074, 0xffffffff); + } + + nouveau_irq_register(dev, 21, nv50_gpio_isr); + return 0; +} + +void +nv50_gpio_fini(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + nv_wr32(dev, 0xe050, 0x00000000); + if (dev_priv->chipset >= 0x90) + nv_wr32(dev, 0xe070, 0x00000000); + nouveau_irq_unregister(dev, 21); + + nv50_gpio_destroy(dev); +} + +static void +nv50_gpio_isr_bh(struct work_struct *work) +{ + struct nv50_gpio_handler *gpioh = + container_of(work, struct nv50_gpio_handler, work); + struct drm_nouveau_private *dev_priv = gpioh->dev->dev_private; + struct nouveau_gpio_engine *pgpio = &dev_priv->engine.gpio; + struct nv50_gpio_priv *priv = pgpio->priv; + unsigned long flags; + int state; + + state = pgpio->get(gpioh->dev, gpioh->gpio->tag); + if (state < 0) + return; + + gpioh->handler(gpioh->data, state); + + spin_lock_irqsave(&priv->lock, flags); + gpioh->inhibit = false; + spin_unlock_irqrestore(&priv->lock, flags); +} + +static void +nv50_gpio_isr(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpio_engine *pgpio = &dev_priv->engine.gpio; + struct nv50_gpio_priv *priv = pgpio->priv; + struct nv50_gpio_handler *gpioh; + u32 intr0, intr1 = 0; + u32 hi, lo, ch; + + intr0 = nv_rd32(dev, 0xe054) & nv_rd32(dev, 0xe050); + if (dev_priv->chipset >= 0x90) + intr1 = nv_rd32(dev, 0xe074) & nv_rd32(dev, 0xe070); + + hi = (intr0 & 0x0000ffff) | (intr1 << 16); + lo = (intr0 >> 16) | (intr1 & 0xffff0000); + ch = hi | lo; + + nv_wr32(dev, 0xe054, intr0); + if (dev_priv->chipset >= 0x90) + nv_wr32(dev, 0xe074, intr1); + + spin_lock(&priv->lock); + list_for_each_entry(gpioh, &priv->handlers, head) { + if (!(ch & (1 << gpioh->gpio->line))) + continue; + + if (gpioh->inhibit) + continue; + gpioh->inhibit = true; + + schedule_work(&gpioh->work); + } + spin_unlock(&priv->lock); +} diff --git a/drivers/gpu/drm/nouveau/nv50_graph.c b/drivers/gpu/drm/nouveau/nv50_graph.c new file mode 100644 index 00000000..e25cbb46 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv50_graph.c @@ -0,0 +1,1121 @@ +/* + * Copyright (C) 2007 Ben Skeggs. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "drmP.h" +#include "drm.h" +#include "nouveau_drv.h" +#include "nouveau_ramht.h" +#include "nouveau_grctx.h" +#include "nouveau_dma.h" +#include "nouveau_vm.h" +#include "nouveau_ramht.h" +#include "nv50_evo.h" + +struct nv50_graph_engine { + struct nouveau_exec_engine base; + u32 ctxprog[512]; + u32 ctxprog_size; + u32 grctx_size; +}; + +static void +nv50_graph_fifo_access(struct drm_device *dev, bool enabled) +{ + const uint32_t mask = 0x00010001; + + if (enabled) + nv_wr32(dev, 0x400500, nv_rd32(dev, 0x400500) | mask); + else + nv_wr32(dev, 0x400500, nv_rd32(dev, 0x400500) & ~mask); +} + +static struct nouveau_channel * +nv50_graph_channel(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + uint32_t inst; + int i; + + /* Be sure we're not in the middle of a context switch or bad things + * will happen, such as unloading the wrong pgraph context. + */ + if (!nv_wait(dev, 0x400300, 0x00000001, 0x00000000)) + NV_ERROR(dev, "Ctxprog is still running\n"); + + inst = nv_rd32(dev, NV50_PGRAPH_CTXCTL_CUR); + if (!(inst & NV50_PGRAPH_CTXCTL_CUR_LOADED)) + return NULL; + inst = (inst & NV50_PGRAPH_CTXCTL_CUR_INSTANCE) << 12; + + for (i = 0; i < dev_priv->engine.fifo.channels; i++) { + struct nouveau_channel *chan = dev_priv->channels.ptr[i]; + + if (chan && chan->ramin && chan->ramin->vinst == inst) + return chan; + } + + return NULL; +} + +static int +nv50_graph_do_load_context(struct drm_device *dev, uint32_t inst) +{ + uint32_t fifo = nv_rd32(dev, 0x400500); + + nv_wr32(dev, 0x400500, fifo & ~1); + nv_wr32(dev, 0x400784, inst); + nv_wr32(dev, 0x400824, nv_rd32(dev, 0x400824) | 0x40); + nv_wr32(dev, 0x400320, nv_rd32(dev, 0x400320) | 0x11); + nv_wr32(dev, 0x400040, 0xffffffff); + (void)nv_rd32(dev, 0x400040); + nv_wr32(dev, 0x400040, 0x00000000); + nv_wr32(dev, 0x400304, nv_rd32(dev, 0x400304) | 1); + + if (nouveau_wait_for_idle(dev)) + nv_wr32(dev, 0x40032c, inst | (1<<31)); + nv_wr32(dev, 0x400500, fifo); + + return 0; +} + +static int +nv50_graph_unload_context(struct drm_device *dev) +{ + uint32_t inst; + + inst = nv_rd32(dev, NV50_PGRAPH_CTXCTL_CUR); + if (!(inst & NV50_PGRAPH_CTXCTL_CUR_LOADED)) + return 0; + inst &= NV50_PGRAPH_CTXCTL_CUR_INSTANCE; + + nouveau_wait_for_idle(dev); + nv_wr32(dev, 0x400784, inst); + nv_wr32(dev, 0x400824, nv_rd32(dev, 0x400824) | 0x20); + nv_wr32(dev, 0x400304, nv_rd32(dev, 0x400304) | 0x01); + nouveau_wait_for_idle(dev); + + nv_wr32(dev, NV50_PGRAPH_CTXCTL_CUR, inst); + return 0; +} + +static void +nv50_graph_init_reset(struct drm_device *dev) +{ + uint32_t pmc_e = NV_PMC_ENABLE_PGRAPH | (1 << 21); + + NV_DEBUG(dev, "\n"); + + nv_wr32(dev, NV03_PMC_ENABLE, nv_rd32(dev, NV03_PMC_ENABLE) & ~pmc_e); + nv_wr32(dev, NV03_PMC_ENABLE, nv_rd32(dev, NV03_PMC_ENABLE) | pmc_e); +} + +static void +nv50_graph_init_intr(struct drm_device *dev) +{ + NV_DEBUG(dev, "\n"); + + nv_wr32(dev, NV03_PGRAPH_INTR, 0xffffffff); + nv_wr32(dev, 0x400138, 0xffffffff); + nv_wr32(dev, NV40_PGRAPH_INTR_EN, 0xffffffff); +} + +static void +nv50_graph_init_regs__nv(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + uint32_t units = nv_rd32(dev, 0x1540); + int i; + + NV_DEBUG(dev, "\n"); + + nv_wr32(dev, 0x400804, 0xc0000000); + nv_wr32(dev, 0x406800, 0xc0000000); + nv_wr32(dev, 0x400c04, 0xc0000000); + nv_wr32(dev, 0x401800, 0xc0000000); + nv_wr32(dev, 0x405018, 0xc0000000); + nv_wr32(dev, 0x402000, 0xc0000000); + + for (i = 0; i < 16; i++) { + if (units & 1 << i) { + if (dev_priv->chipset < 0xa0) { + nv_wr32(dev, 0x408900 + (i << 12), 0xc0000000); + nv_wr32(dev, 0x408e08 + (i << 12), 0xc0000000); + nv_wr32(dev, 0x408314 + (i << 12), 0xc0000000); + } else { + nv_wr32(dev, 0x408600 + (i << 11), 0xc0000000); + nv_wr32(dev, 0x408708 + (i << 11), 0xc0000000); + nv_wr32(dev, 0x40831c + (i << 11), 0xc0000000); + } + } + } + + nv_wr32(dev, 0x400108, 0xffffffff); + + nv_wr32(dev, 0x400824, 0x00004000); + nv_wr32(dev, 0x400500, 0x00010001); +} + +static void +nv50_graph_init_zcull(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + int i; + + NV_DEBUG(dev, "\n"); + + switch (dev_priv->chipset & 0xf0) { + case 0x50: + case 0x80: + case 0x90: + nv_wr32(dev, 0x402ca8, 0x00000800); + break; + case 0xa0: + default: + nv_wr32(dev, 0x402cc0, 0x00000000); + if (dev_priv->chipset == 0xa0 || + dev_priv->chipset == 0xaa || + dev_priv->chipset == 0xac) { + nv_wr32(dev, 0x402ca8, 0x00000802); + } else { + nv_wr32(dev, 0x402cc0, 0x00000000); + nv_wr32(dev, 0x402ca8, 0x00000002); + } + + break; + } + + /* zero out zcull regions */ + for (i = 0; i < 8; i++) { + nv_wr32(dev, 0x402c20 + (i * 8), 0x00000000); + nv_wr32(dev, 0x402c24 + (i * 8), 0x00000000); + nv_wr32(dev, 0x402c28 + (i * 8), 0x00000000); + nv_wr32(dev, 0x402c2c + (i * 8), 0x00000000); + } +} + +static int +nv50_graph_init_ctxctl(struct drm_device *dev) +{ + struct nv50_graph_engine *pgraph = nv_engine(dev, NVOBJ_ENGINE_GR); + int i; + + NV_DEBUG(dev, "\n"); + + nv_wr32(dev, NV40_PGRAPH_CTXCTL_UCODE_INDEX, 0); + for (i = 0; i < pgraph->ctxprog_size; i++) + nv_wr32(dev, NV40_PGRAPH_CTXCTL_UCODE_DATA, pgraph->ctxprog[i]); + + nv_wr32(dev, 0x40008c, 0x00000004); /* HW_CTX_SWITCH_ENABLED */ + nv_wr32(dev, 0x400320, 4); + nv_wr32(dev, NV40_PGRAPH_CTXCTL_CUR, 0); + nv_wr32(dev, NV20_PGRAPH_CHANNEL_CTX_POINTER, 0); + return 0; +} + +static int +nv50_graph_init(struct drm_device *dev, int engine) +{ + int ret; + + NV_DEBUG(dev, "\n"); + + nv50_graph_init_reset(dev); + nv50_graph_init_regs__nv(dev); + nv50_graph_init_zcull(dev); + + ret = nv50_graph_init_ctxctl(dev); + if (ret) + return ret; + + nv50_graph_init_intr(dev); + return 0; +} + +static int +nv50_graph_fini(struct drm_device *dev, int engine) +{ + NV_DEBUG(dev, "\n"); + nv50_graph_unload_context(dev); + nv_wr32(dev, 0x40013c, 0x00000000); + return 0; +} + +static int +nv50_graph_context_new(struct nouveau_channel *chan, int engine) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *ramin = chan->ramin; + struct nouveau_gpuobj *grctx = NULL; + struct nv50_graph_engine *pgraph = nv_engine(dev, engine); + struct nouveau_grctx ctx = {}; + int hdr, ret; + + NV_DEBUG(dev, "ch%d\n", chan->id); + + ret = nouveau_gpuobj_new(dev, NULL, pgraph->grctx_size, 0, + NVOBJ_FLAG_ZERO_ALLOC | + NVOBJ_FLAG_ZERO_FREE, &grctx); + if (ret) + return ret; + + hdr = (dev_priv->chipset == 0x50) ? 0x200 : 0x20; + nv_wo32(ramin, hdr + 0x00, 0x00190002); + nv_wo32(ramin, hdr + 0x04, grctx->vinst + grctx->size - 1); + nv_wo32(ramin, hdr + 0x08, grctx->vinst); + nv_wo32(ramin, hdr + 0x0c, 0); + nv_wo32(ramin, hdr + 0x10, 0); + nv_wo32(ramin, hdr + 0x14, 0x00010000); + + ctx.dev = chan->dev; + ctx.mode = NOUVEAU_GRCTX_VALS; + ctx.data = grctx; + nv50_grctx_init(&ctx); + + nv_wo32(grctx, 0x00000, chan->ramin->vinst >> 12); + + dev_priv->engine.instmem.flush(dev); + + atomic_inc(&chan->vm->engref[NVOBJ_ENGINE_GR]); + chan->engctx[NVOBJ_ENGINE_GR] = grctx; + return 0; +} + +static void +nv50_graph_context_del(struct nouveau_channel *chan, int engine) +{ + struct nouveau_gpuobj *grctx = chan->engctx[engine]; + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; + int i, hdr = (dev_priv->chipset == 0x50) ? 0x200 : 0x20; + unsigned long flags; + + NV_DEBUG(dev, "ch%d\n", chan->id); + + if (!chan->ramin) + return; + + spin_lock_irqsave(&dev_priv->context_switch_lock, flags); + pfifo->reassign(dev, false); + nv50_graph_fifo_access(dev, false); + + if (nv50_graph_channel(dev) == chan) + nv50_graph_unload_context(dev); + + for (i = hdr; i < hdr + 24; i += 4) + nv_wo32(chan->ramin, i, 0); + dev_priv->engine.instmem.flush(dev); + + nv50_graph_fifo_access(dev, true); + pfifo->reassign(dev, true); + spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); + + nouveau_gpuobj_ref(NULL, &grctx); + + atomic_dec(&chan->vm->engref[engine]); + chan->engctx[engine] = NULL; +} + +static int +nv50_graph_object_new(struct nouveau_channel *chan, int engine, + u32 handle, u16 class) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *obj = NULL; + int ret; + + ret = nouveau_gpuobj_new(dev, chan, 16, 16, NVOBJ_FLAG_ZERO_FREE, &obj); + if (ret) + return ret; + obj->engine = 1; + obj->class = class; + + nv_wo32(obj, 0x00, class); + nv_wo32(obj, 0x04, 0x00000000); + nv_wo32(obj, 0x08, 0x00000000); + nv_wo32(obj, 0x0c, 0x00000000); + dev_priv->engine.instmem.flush(dev); + + ret = nouveau_ramht_insert(chan, handle, obj); + nouveau_gpuobj_ref(NULL, &obj); + return ret; +} + +static void +nv50_graph_context_switch(struct drm_device *dev) +{ + uint32_t inst; + + nv50_graph_unload_context(dev); + + inst = nv_rd32(dev, NV50_PGRAPH_CTXCTL_NEXT); + inst &= NV50_PGRAPH_CTXCTL_NEXT_INSTANCE; + nv50_graph_do_load_context(dev, inst); + + nv_wr32(dev, NV40_PGRAPH_INTR_EN, nv_rd32(dev, + NV40_PGRAPH_INTR_EN) | NV_PGRAPH_INTR_CONTEXT_SWITCH); +} + +static int +nv50_graph_nvsw_dma_vblsem(struct nouveau_channel *chan, + u32 class, u32 mthd, u32 data) +{ + struct nouveau_gpuobj *gpuobj; + + gpuobj = nouveau_ramht_find(chan, data); + if (!gpuobj) + return -ENOENT; + + if (nouveau_notifier_offset(gpuobj, NULL)) + return -EINVAL; + + chan->nvsw.vblsem = gpuobj; + chan->nvsw.vblsem_offset = ~0; + return 0; +} + +static int +nv50_graph_nvsw_vblsem_offset(struct nouveau_channel *chan, + u32 class, u32 mthd, u32 data) +{ + if (nouveau_notifier_offset(chan->nvsw.vblsem, &data)) + return -ERANGE; + + chan->nvsw.vblsem_offset = data >> 2; + return 0; +} + +static int +nv50_graph_nvsw_vblsem_release_val(struct nouveau_channel *chan, + u32 class, u32 mthd, u32 data) +{ + chan->nvsw.vblsem_rval = data; + return 0; +} + +static int +nv50_graph_nvsw_vblsem_release(struct nouveau_channel *chan, + u32 class, u32 mthd, u32 data) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + + if (!chan->nvsw.vblsem || chan->nvsw.vblsem_offset == ~0 || data > 1) + return -EINVAL; + + drm_vblank_get(dev, data); + + chan->nvsw.vblsem_head = data; + list_add(&chan->nvsw.vbl_wait, &dev_priv->vbl_waiting); + + return 0; +} + +static int +nv50_graph_nvsw_mthd_page_flip(struct nouveau_channel *chan, + u32 class, u32 mthd, u32 data) +{ + nouveau_finish_page_flip(chan, NULL); + return 0; +} + + +static void +nv50_graph_tlb_flush(struct drm_device *dev, int engine) +{ + nv50_vm_flush_engine(dev, 0); +} + +static void +nv84_graph_tlb_flush(struct drm_device *dev, int engine) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_timer_engine *ptimer = &dev_priv->engine.timer; + bool idle, timeout = false; + unsigned long flags; + u64 start; + u32 tmp; + + spin_lock_irqsave(&dev_priv->context_switch_lock, flags); + nv_mask(dev, 0x400500, 0x00000001, 0x00000000); + + start = ptimer->read(dev); + do { + idle = true; + + for (tmp = nv_rd32(dev, 0x400380); tmp && idle; tmp >>= 3) { + if ((tmp & 7) == 1) + idle = false; + } + + for (tmp = nv_rd32(dev, 0x400384); tmp && idle; tmp >>= 3) { + if ((tmp & 7) == 1) + idle = false; + } + + for (tmp = nv_rd32(dev, 0x400388); tmp && idle; tmp >>= 3) { + if ((tmp & 7) == 1) + idle = false; + } + } while (!idle && !(timeout = ptimer->read(dev) - start > 2000000000)); + + if (timeout) { + NV_ERROR(dev, "PGRAPH TLB flush idle timeout fail: " + "0x%08x 0x%08x 0x%08x 0x%08x\n", + nv_rd32(dev, 0x400700), nv_rd32(dev, 0x400380), + nv_rd32(dev, 0x400384), nv_rd32(dev, 0x400388)); + } + + nv50_vm_flush_engine(dev, 0); + + nv_mask(dev, 0x400500, 0x00000001, 0x00000001); + spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); +} + +static struct nouveau_enum nv50_mp_exec_error_names[] = { + { 3, "STACK_UNDERFLOW", NULL }, + { 4, "QUADON_ACTIVE", NULL }, + { 8, "TIMEOUT", NULL }, + { 0x10, "INVALID_OPCODE", NULL }, + { 0x40, "BREAKPOINT", NULL }, + {} +}; + +static struct nouveau_bitfield nv50_graph_trap_m2mf[] = { + { 0x00000001, "NOTIFY" }, + { 0x00000002, "IN" }, + { 0x00000004, "OUT" }, + {} +}; + +static struct nouveau_bitfield nv50_graph_trap_vfetch[] = { + { 0x00000001, "FAULT" }, + {} +}; + +static struct nouveau_bitfield nv50_graph_trap_strmout[] = { + { 0x00000001, "FAULT" }, + {} +}; + +static struct nouveau_bitfield nv50_graph_trap_ccache[] = { + { 0x00000001, "FAULT" }, + {} +}; + +/* There must be a *lot* of these. Will take some time to gather them up. */ +struct nouveau_enum nv50_data_error_names[] = { + { 0x00000003, "INVALID_QUERY_OR_TEXTURE", NULL }, + { 0x00000004, "INVALID_VALUE", NULL }, + { 0x00000005, "INVALID_ENUM", NULL }, + { 0x00000008, "INVALID_OBJECT", NULL }, + { 0x00000009, "READ_ONLY_OBJECT", NULL }, + { 0x0000000a, "SUPERVISOR_OBJECT", NULL }, + { 0x0000000b, "INVALID_ADDRESS_ALIGNMENT", NULL }, + { 0x0000000c, "INVALID_BITFIELD", NULL }, + { 0x0000000d, "BEGIN_END_ACTIVE", NULL }, + { 0x0000000e, "SEMANTIC_COLOR_BACK_OVER_LIMIT", NULL }, + { 0x0000000f, "VIEWPORT_ID_NEEDS_GP", NULL }, + { 0x00000010, "RT_DOUBLE_BIND", NULL }, + { 0x00000011, "RT_TYPES_MISMATCH", NULL }, + { 0x00000012, "RT_LINEAR_WITH_ZETA", NULL }, + { 0x00000015, "FP_TOO_FEW_REGS", NULL }, + { 0x00000016, "ZETA_FORMAT_CSAA_MISMATCH", NULL }, + { 0x00000017, "RT_LINEAR_WITH_MSAA", NULL }, + { 0x00000018, "FP_INTERPOLANT_START_OVER_LIMIT", NULL }, + { 0x00000019, "SEMANTIC_LAYER_OVER_LIMIT", NULL }, + { 0x0000001a, "RT_INVALID_ALIGNMENT", NULL }, + { 0x0000001b, "SAMPLER_OVER_LIMIT", NULL }, + { 0x0000001c, "TEXTURE_OVER_LIMIT", NULL }, + { 0x0000001e, "GP_TOO_MANY_OUTPUTS", NULL }, + { 0x0000001f, "RT_BPP128_WITH_MS8", NULL }, + { 0x00000021, "Z_OUT_OF_BOUNDS", NULL }, + { 0x00000023, "XY_OUT_OF_BOUNDS", NULL }, + { 0x00000027, "CP_MORE_PARAMS_THAN_SHARED", NULL }, + { 0x00000028, "CP_NO_REG_SPACE_STRIPED", NULL }, + { 0x00000029, "CP_NO_REG_SPACE_PACKED", NULL }, + { 0x0000002a, "CP_NOT_ENOUGH_WARPS", NULL }, + { 0x0000002b, "CP_BLOCK_SIZE_MISMATCH", NULL }, + { 0x0000002c, "CP_NOT_ENOUGH_LOCAL_WARPS", NULL }, + { 0x0000002d, "CP_NOT_ENOUGH_STACK_WARPS", NULL }, + { 0x0000002e, "CP_NO_BLOCKDIM_LATCH", NULL }, + { 0x00000031, "ENG2D_FORMAT_MISMATCH", NULL }, + { 0x0000003f, "PRIMITIVE_ID_NEEDS_GP", NULL }, + { 0x00000044, "SEMANTIC_VIEWPORT_OVER_LIMIT", NULL }, + { 0x00000045, "SEMANTIC_COLOR_FRONT_OVER_LIMIT", NULL }, + { 0x00000046, "LAYER_ID_NEEDS_GP", NULL }, + { 0x00000047, "SEMANTIC_CLIP_OVER_LIMIT", NULL }, + { 0x00000048, "SEMANTIC_PTSZ_OVER_LIMIT", NULL }, + {} +}; + +static struct nouveau_bitfield nv50_graph_intr[] = { + { 0x00000001, "NOTIFY" }, + { 0x00000002, "COMPUTE_QUERY" }, + { 0x00000010, "ILLEGAL_MTHD" }, + { 0x00000020, "ILLEGAL_CLASS" }, + { 0x00000040, "DOUBLE_NOTIFY" }, + { 0x00001000, "CONTEXT_SWITCH" }, + { 0x00010000, "BUFFER_NOTIFY" }, + { 0x00100000, "DATA_ERROR" }, + { 0x00200000, "TRAP" }, + { 0x01000000, "SINGLE_STEP" }, + {} +}; + +static void +nv50_pgraph_mp_trap(struct drm_device *dev, int tpid, int display) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + uint32_t units = nv_rd32(dev, 0x1540); + uint32_t addr, mp10, status, pc, oplow, ophigh; + int i; + int mps = 0; + for (i = 0; i < 4; i++) { + if (!(units & 1 << (i+24))) + continue; + if (dev_priv->chipset < 0xa0) + addr = 0x408200 + (tpid << 12) + (i << 7); + else + addr = 0x408100 + (tpid << 11) + (i << 7); + mp10 = nv_rd32(dev, addr + 0x10); + status = nv_rd32(dev, addr + 0x14); + if (!status) + continue; + if (display) { + nv_rd32(dev, addr + 0x20); + pc = nv_rd32(dev, addr + 0x24); + oplow = nv_rd32(dev, addr + 0x70); + ophigh = nv_rd32(dev, addr + 0x74); + NV_INFO(dev, "PGRAPH_TRAP_MP_EXEC - " + "TP %d MP %d: ", tpid, i); + nouveau_enum_print(nv50_mp_exec_error_names, status); + printk(" at %06x warp %d, opcode %08x %08x\n", + pc&0xffffff, pc >> 24, + oplow, ophigh); + } + nv_wr32(dev, addr + 0x10, mp10); + nv_wr32(dev, addr + 0x14, 0); + mps++; + } + if (!mps && display) + NV_INFO(dev, "PGRAPH_TRAP_MP_EXEC - TP %d: " + "No MPs claiming errors?\n", tpid); +} + +static void +nv50_pgraph_tp_trap(struct drm_device *dev, int type, uint32_t ustatus_old, + uint32_t ustatus_new, int display, const char *name) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + int tps = 0; + uint32_t units = nv_rd32(dev, 0x1540); + int i, r; + uint32_t ustatus_addr, ustatus; + for (i = 0; i < 16; i++) { + if (!(units & (1 << i))) + continue; + if (dev_priv->chipset < 0xa0) + ustatus_addr = ustatus_old + (i << 12); + else + ustatus_addr = ustatus_new + (i << 11); + ustatus = nv_rd32(dev, ustatus_addr) & 0x7fffffff; + if (!ustatus) + continue; + tps++; + switch (type) { + case 6: /* texture error... unknown for now */ + if (display) { + NV_ERROR(dev, "magic set %d:\n", i); + for (r = ustatus_addr + 4; r <= ustatus_addr + 0x10; r += 4) + NV_ERROR(dev, "\t0x%08x: 0x%08x\n", r, + nv_rd32(dev, r)); + } + break; + case 7: /* MP error */ + if (ustatus & 0x00010000) { + nv50_pgraph_mp_trap(dev, i, display); + ustatus &= ~0x00010000; + } + break; + case 8: /* TPDMA error */ + { + uint32_t e0c = nv_rd32(dev, ustatus_addr + 4); + uint32_t e10 = nv_rd32(dev, ustatus_addr + 8); + uint32_t e14 = nv_rd32(dev, ustatus_addr + 0xc); + uint32_t e18 = nv_rd32(dev, ustatus_addr + 0x10); + uint32_t e1c = nv_rd32(dev, ustatus_addr + 0x14); + uint32_t e20 = nv_rd32(dev, ustatus_addr + 0x18); + uint32_t e24 = nv_rd32(dev, ustatus_addr + 0x1c); + /* 2d engine destination */ + if (ustatus & 0x00000010) { + if (display) { + NV_INFO(dev, "PGRAPH_TRAP_TPDMA_2D - TP %d - Unknown fault at address %02x%08x\n", + i, e14, e10); + NV_INFO(dev, "PGRAPH_TRAP_TPDMA_2D - TP %d - e0c: %08x, e18: %08x, e1c: %08x, e20: %08x, e24: %08x\n", + i, e0c, e18, e1c, e20, e24); + } + ustatus &= ~0x00000010; + } + /* Render target */ + if (ustatus & 0x00000040) { + if (display) { + NV_INFO(dev, "PGRAPH_TRAP_TPDMA_RT - TP %d - Unknown fault at address %02x%08x\n", + i, e14, e10); + NV_INFO(dev, "PGRAPH_TRAP_TPDMA_RT - TP %d - e0c: %08x, e18: %08x, e1c: %08x, e20: %08x, e24: %08x\n", + i, e0c, e18, e1c, e20, e24); + } + ustatus &= ~0x00000040; + } + /* CUDA memory: l[], g[] or stack. */ + if (ustatus & 0x00000080) { + if (display) { + if (e18 & 0x80000000) { + /* g[] read fault? */ + NV_INFO(dev, "PGRAPH_TRAP_TPDMA - TP %d - Global read fault at address %02x%08x\n", + i, e14, e10 | ((e18 >> 24) & 0x1f)); + e18 &= ~0x1f000000; + } else if (e18 & 0xc) { + /* g[] write fault? */ + NV_INFO(dev, "PGRAPH_TRAP_TPDMA - TP %d - Global write fault at address %02x%08x\n", + i, e14, e10 | ((e18 >> 7) & 0x1f)); + e18 &= ~0x00000f80; + } else { + NV_INFO(dev, "PGRAPH_TRAP_TPDMA - TP %d - Unknown CUDA fault at address %02x%08x\n", + i, e14, e10); + } + NV_INFO(dev, "PGRAPH_TRAP_TPDMA - TP %d - e0c: %08x, e18: %08x, e1c: %08x, e20: %08x, e24: %08x\n", + i, e0c, e18, e1c, e20, e24); + } + ustatus &= ~0x00000080; + } + } + break; + } + if (ustatus) { + if (display) + NV_INFO(dev, "%s - TP%d: Unhandled ustatus 0x%08x\n", name, i, ustatus); + } + nv_wr32(dev, ustatus_addr, 0xc0000000); + } + + if (!tps && display) + NV_INFO(dev, "%s - No TPs claiming errors?\n", name); +} + +static int +nv50_pgraph_trap_handler(struct drm_device *dev, u32 display, u64 inst, u32 chid) +{ + u32 status = nv_rd32(dev, 0x400108); + u32 ustatus; + + if (!status && display) { + NV_INFO(dev, "PGRAPH - TRAP: no units reporting traps?\n"); + return 1; + } + + /* DISPATCH: Relays commands to other units and handles NOTIFY, + * COND, QUERY. If you get a trap from it, the command is still stuck + * in DISPATCH and you need to do something about it. */ + if (status & 0x001) { + ustatus = nv_rd32(dev, 0x400804) & 0x7fffffff; + if (!ustatus && display) { + NV_INFO(dev, "PGRAPH_TRAP_DISPATCH - no ustatus?\n"); + } + + nv_wr32(dev, 0x400500, 0x00000000); + + /* Known to be triggered by screwed up NOTIFY and COND... */ + if (ustatus & 0x00000001) { + u32 addr = nv_rd32(dev, 0x400808); + u32 subc = (addr & 0x00070000) >> 16; + u32 mthd = (addr & 0x00001ffc); + u32 datal = nv_rd32(dev, 0x40080c); + u32 datah = nv_rd32(dev, 0x400810); + u32 class = nv_rd32(dev, 0x400814); + u32 r848 = nv_rd32(dev, 0x400848); + + NV_INFO(dev, "PGRAPH - TRAP DISPATCH_FAULT\n"); + if (display && (addr & 0x80000000)) { + NV_INFO(dev, "PGRAPH - ch %d (0x%010llx) " + "subc %d class 0x%04x mthd 0x%04x " + "data 0x%08x%08x " + "400808 0x%08x 400848 0x%08x\n", + chid, inst, subc, class, mthd, datah, + datal, addr, r848); + } else + if (display) { + NV_INFO(dev, "PGRAPH - no stuck command?\n"); + } + + nv_wr32(dev, 0x400808, 0); + nv_wr32(dev, 0x4008e8, nv_rd32(dev, 0x4008e8) & 3); + nv_wr32(dev, 0x400848, 0); + ustatus &= ~0x00000001; + } + + if (ustatus & 0x00000002) { + u32 addr = nv_rd32(dev, 0x40084c); + u32 subc = (addr & 0x00070000) >> 16; + u32 mthd = (addr & 0x00001ffc); + u32 data = nv_rd32(dev, 0x40085c); + u32 class = nv_rd32(dev, 0x400814); + + NV_INFO(dev, "PGRAPH - TRAP DISPATCH_QUERY\n"); + if (display && (addr & 0x80000000)) { + NV_INFO(dev, "PGRAPH - ch %d (0x%010llx) " + "subc %d class 0x%04x mthd 0x%04x " + "data 0x%08x 40084c 0x%08x\n", + chid, inst, subc, class, mthd, + data, addr); + } else + if (display) { + NV_INFO(dev, "PGRAPH - no stuck command?\n"); + } + + nv_wr32(dev, 0x40084c, 0); + ustatus &= ~0x00000002; + } + + if (ustatus && display) { + NV_INFO(dev, "PGRAPH - TRAP_DISPATCH (unknown " + "0x%08x)\n", ustatus); + } + + nv_wr32(dev, 0x400804, 0xc0000000); + nv_wr32(dev, 0x400108, 0x001); + status &= ~0x001; + if (!status) + return 0; + } + + /* M2MF: Memory to memory copy engine. */ + if (status & 0x002) { + u32 ustatus = nv_rd32(dev, 0x406800) & 0x7fffffff; + if (display) { + NV_INFO(dev, "PGRAPH - TRAP_M2MF"); + nouveau_bitfield_print(nv50_graph_trap_m2mf, ustatus); + printk("\n"); + NV_INFO(dev, "PGRAPH - TRAP_M2MF %08x %08x %08x %08x\n", + nv_rd32(dev, 0x406804), nv_rd32(dev, 0x406808), + nv_rd32(dev, 0x40680c), nv_rd32(dev, 0x406810)); + + } + + /* No sane way found yet -- just reset the bugger. */ + nv_wr32(dev, 0x400040, 2); + nv_wr32(dev, 0x400040, 0); + nv_wr32(dev, 0x406800, 0xc0000000); + nv_wr32(dev, 0x400108, 0x002); + status &= ~0x002; + } + + /* VFETCH: Fetches data from vertex buffers. */ + if (status & 0x004) { + u32 ustatus = nv_rd32(dev, 0x400c04) & 0x7fffffff; + if (display) { + NV_INFO(dev, "PGRAPH - TRAP_VFETCH"); + nouveau_bitfield_print(nv50_graph_trap_vfetch, ustatus); + printk("\n"); + NV_INFO(dev, "PGRAPH - TRAP_VFETCH %08x %08x %08x %08x\n", + nv_rd32(dev, 0x400c00), nv_rd32(dev, 0x400c08), + nv_rd32(dev, 0x400c0c), nv_rd32(dev, 0x400c10)); + } + + nv_wr32(dev, 0x400c04, 0xc0000000); + nv_wr32(dev, 0x400108, 0x004); + status &= ~0x004; + } + + /* STRMOUT: DirectX streamout / OpenGL transform feedback. */ + if (status & 0x008) { + ustatus = nv_rd32(dev, 0x401800) & 0x7fffffff; + if (display) { + NV_INFO(dev, "PGRAPH - TRAP_STRMOUT"); + nouveau_bitfield_print(nv50_graph_trap_strmout, ustatus); + printk("\n"); + NV_INFO(dev, "PGRAPH - TRAP_STRMOUT %08x %08x %08x %08x\n", + nv_rd32(dev, 0x401804), nv_rd32(dev, 0x401808), + nv_rd32(dev, 0x40180c), nv_rd32(dev, 0x401810)); + + } + + /* No sane way found yet -- just reset the bugger. */ + nv_wr32(dev, 0x400040, 0x80); + nv_wr32(dev, 0x400040, 0); + nv_wr32(dev, 0x401800, 0xc0000000); + nv_wr32(dev, 0x400108, 0x008); + status &= ~0x008; + } + + /* CCACHE: Handles code and c[] caches and fills them. */ + if (status & 0x010) { + ustatus = nv_rd32(dev, 0x405018) & 0x7fffffff; + if (display) { + NV_INFO(dev, "PGRAPH - TRAP_CCACHE"); + nouveau_bitfield_print(nv50_graph_trap_ccache, ustatus); + printk("\n"); + NV_INFO(dev, "PGRAPH - TRAP_CCACHE %08x %08x %08x %08x" + " %08x %08x %08x\n", + nv_rd32(dev, 0x405000), nv_rd32(dev, 0x405004), + nv_rd32(dev, 0x405008), nv_rd32(dev, 0x40500c), + nv_rd32(dev, 0x405010), nv_rd32(dev, 0x405014), + nv_rd32(dev, 0x40501c)); + + } + + nv_wr32(dev, 0x405018, 0xc0000000); + nv_wr32(dev, 0x400108, 0x010); + status &= ~0x010; + } + + /* Unknown, not seen yet... 0x402000 is the only trap status reg + * remaining, so try to handle it anyway. Perhaps related to that + * unknown DMA slot on tesla? */ + if (status & 0x20) { + ustatus = nv_rd32(dev, 0x402000) & 0x7fffffff; + if (display) + NV_INFO(dev, "PGRAPH - TRAP_UNKC04 0x%08x\n", ustatus); + nv_wr32(dev, 0x402000, 0xc0000000); + /* no status modifiction on purpose */ + } + + /* TEXTURE: CUDA texturing units */ + if (status & 0x040) { + nv50_pgraph_tp_trap(dev, 6, 0x408900, 0x408600, display, + "PGRAPH - TRAP_TEXTURE"); + nv_wr32(dev, 0x400108, 0x040); + status &= ~0x040; + } + + /* MP: CUDA execution engines. */ + if (status & 0x080) { + nv50_pgraph_tp_trap(dev, 7, 0x408314, 0x40831c, display, + "PGRAPH - TRAP_MP"); + nv_wr32(dev, 0x400108, 0x080); + status &= ~0x080; + } + + /* TPDMA: Handles TP-initiated uncached memory accesses: + * l[], g[], stack, 2d surfaces, render targets. */ + if (status & 0x100) { + nv50_pgraph_tp_trap(dev, 8, 0x408e08, 0x408708, display, + "PGRAPH - TRAP_TPDMA"); + nv_wr32(dev, 0x400108, 0x100); + status &= ~0x100; + } + + if (status) { + if (display) + NV_INFO(dev, "PGRAPH - TRAP: unknown 0x%08x\n", status); + nv_wr32(dev, 0x400108, status); + } + + return 1; +} + +int +nv50_graph_isr_chid(struct drm_device *dev, u64 inst) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_channel *chan; + unsigned long flags; + int i; + + spin_lock_irqsave(&dev_priv->channels.lock, flags); + for (i = 0; i < dev_priv->engine.fifo.channels; i++) { + chan = dev_priv->channels.ptr[i]; + if (!chan || !chan->ramin) + continue; + + if (inst == chan->ramin->vinst) + break; + } + spin_unlock_irqrestore(&dev_priv->channels.lock, flags); + return i; +} + +static void +nv50_graph_isr(struct drm_device *dev) +{ + u32 stat; + + while ((stat = nv_rd32(dev, 0x400100))) { + u64 inst = (u64)(nv_rd32(dev, 0x40032c) & 0x0fffffff) << 12; + u32 chid = nv50_graph_isr_chid(dev, inst); + u32 addr = nv_rd32(dev, NV04_PGRAPH_TRAPPED_ADDR); + u32 subc = (addr & 0x00070000) >> 16; + u32 mthd = (addr & 0x00001ffc); + u32 data = nv_rd32(dev, NV04_PGRAPH_TRAPPED_DATA); + u32 class = nv_rd32(dev, 0x400814); + u32 show = stat; + + if (stat & 0x00000010) { + if (!nouveau_gpuobj_mthd_call2(dev, chid, class, + mthd, data)) + show &= ~0x00000010; + } + + if (stat & 0x00001000) { + nv_wr32(dev, 0x400500, 0x00000000); + nv_wr32(dev, 0x400100, 0x00001000); + nv_mask(dev, 0x40013c, 0x00001000, 0x00000000); + nv50_graph_context_switch(dev); + stat &= ~0x00001000; + show &= ~0x00001000; + } + + show = (show && nouveau_ratelimit()) ? show : 0; + + if (show & 0x00100000) { + u32 ecode = nv_rd32(dev, 0x400110); + NV_INFO(dev, "PGRAPH - DATA_ERROR "); + nouveau_enum_print(nv50_data_error_names, ecode); + printk("\n"); + } + + if (stat & 0x00200000) { + if (!nv50_pgraph_trap_handler(dev, show, inst, chid)) + show &= ~0x00200000; + } + + nv_wr32(dev, 0x400100, stat); + nv_wr32(dev, 0x400500, 0x00010001); + + if (show) { + NV_INFO(dev, "PGRAPH -"); + nouveau_bitfield_print(nv50_graph_intr, show); + printk("\n"); + NV_INFO(dev, "PGRAPH - ch %d (0x%010llx) subc %d " + "class 0x%04x mthd 0x%04x data 0x%08x\n", + chid, inst, subc, class, mthd, data); + nv50_fb_vm_trap(dev, 1); + } + } + + if (nv_rd32(dev, 0x400824) & (1 << 31)) + nv_wr32(dev, 0x400824, nv_rd32(dev, 0x400824) & ~(1 << 31)); +} + +static void +nv50_graph_destroy(struct drm_device *dev, int engine) +{ + struct nv50_graph_engine *pgraph = nv_engine(dev, engine); + + NVOBJ_ENGINE_DEL(dev, GR); + + nouveau_irq_unregister(dev, 12); + kfree(pgraph); +} + +int +nv50_graph_create(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nv50_graph_engine *pgraph; + struct nouveau_grctx ctx = {}; + int ret; + + pgraph = kzalloc(sizeof(*pgraph),GFP_KERNEL); + if (!pgraph) + return -ENOMEM; + + ctx.dev = dev; + ctx.mode = NOUVEAU_GRCTX_PROG; + ctx.data = pgraph->ctxprog; + ctx.ctxprog_max = ARRAY_SIZE(pgraph->ctxprog); + + ret = nv50_grctx_init(&ctx); + if (ret) { + NV_ERROR(dev, "PGRAPH: ctxprog build failed\n"); + kfree(pgraph); + return 0; + } + + pgraph->grctx_size = ctx.ctxvals_pos * 4; + pgraph->ctxprog_size = ctx.ctxprog_len; + + pgraph->base.destroy = nv50_graph_destroy; + pgraph->base.init = nv50_graph_init; + pgraph->base.fini = nv50_graph_fini; + pgraph->base.context_new = nv50_graph_context_new; + pgraph->base.context_del = nv50_graph_context_del; + pgraph->base.object_new = nv50_graph_object_new; + if (dev_priv->chipset == 0x50 || dev_priv->chipset == 0xac) + pgraph->base.tlb_flush = nv50_graph_tlb_flush; + else + pgraph->base.tlb_flush = nv84_graph_tlb_flush; + + nouveau_irq_register(dev, 12, nv50_graph_isr); + + /* NVSW really doesn't live here... */ + NVOBJ_CLASS(dev, 0x506e, SW); /* nvsw */ + NVOBJ_MTHD (dev, 0x506e, 0x018c, nv50_graph_nvsw_dma_vblsem); + NVOBJ_MTHD (dev, 0x506e, 0x0400, nv50_graph_nvsw_vblsem_offset); + NVOBJ_MTHD (dev, 0x506e, 0x0404, nv50_graph_nvsw_vblsem_release_val); + NVOBJ_MTHD (dev, 0x506e, 0x0408, nv50_graph_nvsw_vblsem_release); + NVOBJ_MTHD (dev, 0x506e, 0x0500, nv50_graph_nvsw_mthd_page_flip); + + NVOBJ_ENGINE_ADD(dev, GR, &pgraph->base); + NVOBJ_CLASS(dev, 0x0030, GR); /* null */ + NVOBJ_CLASS(dev, 0x5039, GR); /* m2mf */ + NVOBJ_CLASS(dev, 0x502d, GR); /* 2d */ + + /* tesla */ + if (dev_priv->chipset == 0x50) + NVOBJ_CLASS(dev, 0x5097, GR); /* tesla (nv50) */ + else + if (dev_priv->chipset < 0xa0) + NVOBJ_CLASS(dev, 0x8297, GR); /* tesla (nv8x/nv9x) */ + else { + switch (dev_priv->chipset) { + case 0xa0: + case 0xaa: + case 0xac: + NVOBJ_CLASS(dev, 0x8397, GR); + break; + case 0xa3: + case 0xa5: + case 0xa8: + NVOBJ_CLASS(dev, 0x8597, GR); + break; + case 0xaf: + NVOBJ_CLASS(dev, 0x8697, GR); + break; + } + } + + /* compute */ + NVOBJ_CLASS(dev, 0x50c0, GR); + if (dev_priv->chipset > 0xa0 && + dev_priv->chipset != 0xaa && + dev_priv->chipset != 0xac) + NVOBJ_CLASS(dev, 0x85c0, GR); + + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nv50_grctx.c b/drivers/gpu/drm/nouveau/nv50_grctx.c new file mode 100644 index 00000000..de9abff1 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv50_grctx.c @@ -0,0 +1,3322 @@ +/* + * Copyright 2009 Marcin KoÅ›cielnicki + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#define CP_FLAG_CLEAR 0 +#define CP_FLAG_SET 1 +#define CP_FLAG_SWAP_DIRECTION ((0 * 32) + 0) +#define CP_FLAG_SWAP_DIRECTION_LOAD 0 +#define CP_FLAG_SWAP_DIRECTION_SAVE 1 +#define CP_FLAG_UNK01 ((0 * 32) + 1) +#define CP_FLAG_UNK01_CLEAR 0 +#define CP_FLAG_UNK01_SET 1 +#define CP_FLAG_UNK03 ((0 * 32) + 3) +#define CP_FLAG_UNK03_CLEAR 0 +#define CP_FLAG_UNK03_SET 1 +#define CP_FLAG_USER_SAVE ((0 * 32) + 5) +#define CP_FLAG_USER_SAVE_NOT_PENDING 0 +#define CP_FLAG_USER_SAVE_PENDING 1 +#define CP_FLAG_USER_LOAD ((0 * 32) + 6) +#define CP_FLAG_USER_LOAD_NOT_PENDING 0 +#define CP_FLAG_USER_LOAD_PENDING 1 +#define CP_FLAG_UNK0B ((0 * 32) + 0xb) +#define CP_FLAG_UNK0B_CLEAR 0 +#define CP_FLAG_UNK0B_SET 1 +#define CP_FLAG_UNK1D ((0 * 32) + 0x1d) +#define CP_FLAG_UNK1D_CLEAR 0 +#define CP_FLAG_UNK1D_SET 1 +#define CP_FLAG_UNK20 ((1 * 32) + 0) +#define CP_FLAG_UNK20_CLEAR 0 +#define CP_FLAG_UNK20_SET 1 +#define CP_FLAG_STATUS ((2 * 32) + 0) +#define CP_FLAG_STATUS_BUSY 0 +#define CP_FLAG_STATUS_IDLE 1 +#define CP_FLAG_AUTO_SAVE ((2 * 32) + 4) +#define CP_FLAG_AUTO_SAVE_NOT_PENDING 0 +#define CP_FLAG_AUTO_SAVE_PENDING 1 +#define CP_FLAG_AUTO_LOAD ((2 * 32) + 5) +#define CP_FLAG_AUTO_LOAD_NOT_PENDING 0 +#define CP_FLAG_AUTO_LOAD_PENDING 1 +#define CP_FLAG_NEWCTX ((2 * 32) + 10) +#define CP_FLAG_NEWCTX_BUSY 0 +#define CP_FLAG_NEWCTX_DONE 1 +#define CP_FLAG_XFER ((2 * 32) + 11) +#define CP_FLAG_XFER_IDLE 0 +#define CP_FLAG_XFER_BUSY 1 +#define CP_FLAG_ALWAYS ((2 * 32) + 13) +#define CP_FLAG_ALWAYS_FALSE 0 +#define CP_FLAG_ALWAYS_TRUE 1 +#define CP_FLAG_INTR ((2 * 32) + 15) +#define CP_FLAG_INTR_NOT_PENDING 0 +#define CP_FLAG_INTR_PENDING 1 + +#define CP_CTX 0x00100000 +#define CP_CTX_COUNT 0x000f0000 +#define CP_CTX_COUNT_SHIFT 16 +#define CP_CTX_REG 0x00003fff +#define CP_LOAD_SR 0x00200000 +#define CP_LOAD_SR_VALUE 0x000fffff +#define CP_BRA 0x00400000 +#define CP_BRA_IP 0x0001ff00 +#define CP_BRA_IP_SHIFT 8 +#define CP_BRA_IF_CLEAR 0x00000080 +#define CP_BRA_FLAG 0x0000007f +#define CP_WAIT 0x00500000 +#define CP_WAIT_SET 0x00000080 +#define CP_WAIT_FLAG 0x0000007f +#define CP_SET 0x00700000 +#define CP_SET_1 0x00000080 +#define CP_SET_FLAG 0x0000007f +#define CP_NEWCTX 0x00600004 +#define CP_NEXT_TO_SWAP 0x00600005 +#define CP_SET_CONTEXT_POINTER 0x00600006 +#define CP_SET_XFER_POINTER 0x00600007 +#define CP_ENABLE 0x00600009 +#define CP_END 0x0060000c +#define CP_NEXT_TO_CURRENT 0x0060000d +#define CP_DISABLE1 0x0090ffff +#define CP_DISABLE2 0x0091ffff +#define CP_XFER_1 0x008000ff +#define CP_XFER_2 0x008800ff +#define CP_SEEK_1 0x00c000ff +#define CP_SEEK_2 0x00c800ff + +#include "drmP.h" +#include "nouveau_drv.h" +#include "nouveau_grctx.h" + +#define IS_NVA3F(x) (((x) > 0xa0 && (x) < 0xaa) || (x) == 0xaf) +#define IS_NVAAF(x) ((x) >= 0xaa && (x) <= 0xac) + +/* + * This code deals with PGRAPH contexts on NV50 family cards. Like NV40, it's + * the GPU itself that does context-switching, but it needs a special + * microcode to do it. And it's the driver's task to supply this microcode, + * further known as ctxprog, as well as the initial context values, known + * as ctxvals. + * + * Without ctxprog, you cannot switch contexts. Not even in software, since + * the majority of context [xfer strands] isn't accessible directly. You're + * stuck with a single channel, and you also suffer all the problems resulting + * from missing ctxvals, since you cannot load them. + * + * Without ctxvals, you're stuck with PGRAPH's default context. It's enough to + * run 2d operations, but trying to utilise 3d or CUDA will just lock you up, + * since you don't have... some sort of needed setup. + * + * Nouveau will just disable acceleration if not given ctxprog + ctxvals, since + * it's too much hassle to handle no-ctxprog as a special case. + */ + +/* + * How ctxprogs work. + * + * The ctxprog is written in its own kind of microcode, with very small and + * crappy set of available commands. You upload it to a small [512 insns] + * area of memory on PGRAPH, and it'll be run when PFIFO wants PGRAPH to + * switch channel. or when the driver explicitely requests it. Stuff visible + * to ctxprog consists of: PGRAPH MMIO registers, PGRAPH context strands, + * the per-channel context save area in VRAM [known as ctxvals or grctx], + * 4 flags registers, a scratch register, two grctx pointers, plus many + * random poorly-understood details. + * + * When ctxprog runs, it's supposed to check what operations are asked of it, + * save old context if requested, optionally reset PGRAPH and switch to the + * new channel, and load the new context. Context consists of three major + * parts: subset of MMIO registers and two "xfer areas". + */ + +/* TODO: + * - document unimplemented bits compared to nvidia + * - NVAx: make a TP subroutine, use it. + * - use 0x4008fc instead of 0x1540? + */ + +enum cp_label { + cp_check_load = 1, + cp_setup_auto_load, + cp_setup_load, + cp_setup_save, + cp_swap_state, + cp_prepare_exit, + cp_exit, +}; + +static void nv50_graph_construct_mmio(struct nouveau_grctx *ctx); +static void nv50_graph_construct_xfer1(struct nouveau_grctx *ctx); +static void nv50_graph_construct_xfer2(struct nouveau_grctx *ctx); + +/* Main function: construct the ctxprog skeleton, call the other functions. */ + +int +nv50_grctx_init(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + + switch (dev_priv->chipset) { + case 0x50: + case 0x84: + case 0x86: + case 0x92: + case 0x94: + case 0x96: + case 0x98: + case 0xa0: + case 0xa3: + case 0xa5: + case 0xa8: + case 0xaa: + case 0xac: + case 0xaf: + break; + default: + NV_ERROR(ctx->dev, "I don't know how to make a ctxprog for " + "your NV%x card.\n", dev_priv->chipset); + NV_ERROR(ctx->dev, "Disabling acceleration. Please contact " + "the devs.\n"); + return -ENOSYS; + } + /* decide whether we're loading/unloading the context */ + cp_bra (ctx, AUTO_SAVE, PENDING, cp_setup_save); + cp_bra (ctx, USER_SAVE, PENDING, cp_setup_save); + + cp_name(ctx, cp_check_load); + cp_bra (ctx, AUTO_LOAD, PENDING, cp_setup_auto_load); + cp_bra (ctx, USER_LOAD, PENDING, cp_setup_load); + cp_bra (ctx, ALWAYS, TRUE, cp_exit); + + /* setup for context load */ + cp_name(ctx, cp_setup_auto_load); + cp_out (ctx, CP_DISABLE1); + cp_out (ctx, CP_DISABLE2); + cp_out (ctx, CP_ENABLE); + cp_out (ctx, CP_NEXT_TO_SWAP); + cp_set (ctx, UNK01, SET); + cp_name(ctx, cp_setup_load); + cp_out (ctx, CP_NEWCTX); + cp_wait(ctx, NEWCTX, BUSY); + cp_set (ctx, UNK1D, CLEAR); + cp_set (ctx, SWAP_DIRECTION, LOAD); + cp_bra (ctx, UNK0B, SET, cp_prepare_exit); + cp_bra (ctx, ALWAYS, TRUE, cp_swap_state); + + /* setup for context save */ + cp_name(ctx, cp_setup_save); + cp_set (ctx, UNK1D, SET); + cp_wait(ctx, STATUS, BUSY); + cp_wait(ctx, INTR, PENDING); + cp_bra (ctx, STATUS, BUSY, cp_setup_save); + cp_set (ctx, UNK01, SET); + cp_set (ctx, SWAP_DIRECTION, SAVE); + + /* general PGRAPH state */ + cp_name(ctx, cp_swap_state); + cp_set (ctx, UNK03, SET); + cp_pos (ctx, 0x00004/4); + cp_ctx (ctx, 0x400828, 1); /* needed. otherwise, flickering happens. */ + cp_pos (ctx, 0x00100/4); + nv50_graph_construct_mmio(ctx); + nv50_graph_construct_xfer1(ctx); + nv50_graph_construct_xfer2(ctx); + + cp_bra (ctx, SWAP_DIRECTION, SAVE, cp_check_load); + + cp_set (ctx, UNK20, SET); + cp_set (ctx, SWAP_DIRECTION, SAVE); /* no idea why this is needed, but fixes at least one lockup. */ + cp_lsr (ctx, ctx->ctxvals_base); + cp_out (ctx, CP_SET_XFER_POINTER); + cp_lsr (ctx, 4); + cp_out (ctx, CP_SEEK_1); + cp_out (ctx, CP_XFER_1); + cp_wait(ctx, XFER, BUSY); + + /* pre-exit state updates */ + cp_name(ctx, cp_prepare_exit); + cp_set (ctx, UNK01, CLEAR); + cp_set (ctx, UNK03, CLEAR); + cp_set (ctx, UNK1D, CLEAR); + + cp_bra (ctx, USER_SAVE, PENDING, cp_exit); + cp_out (ctx, CP_NEXT_TO_CURRENT); + + cp_name(ctx, cp_exit); + cp_set (ctx, USER_SAVE, NOT_PENDING); + cp_set (ctx, USER_LOAD, NOT_PENDING); + cp_out (ctx, CP_END); + ctx->ctxvals_pos += 0x400; /* padding... no idea why you need it */ + + return 0; +} + +/* + * Constructs MMIO part of ctxprog and ctxvals. Just a matter of knowing which + * registers to save/restore and the default values for them. + */ + +static void +nv50_graph_construct_mmio_ddata(struct nouveau_grctx *ctx); + +static void +nv50_graph_construct_mmio(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + int i, j; + int offset, base; + uint32_t units = nv_rd32 (ctx->dev, 0x1540); + + /* 0800: DISPATCH */ + cp_ctx(ctx, 0x400808, 7); + gr_def(ctx, 0x400814, 0x00000030); + cp_ctx(ctx, 0x400834, 0x32); + if (dev_priv->chipset == 0x50) { + gr_def(ctx, 0x400834, 0xff400040); + gr_def(ctx, 0x400838, 0xfff00080); + gr_def(ctx, 0x40083c, 0xfff70090); + gr_def(ctx, 0x400840, 0xffe806a8); + } + gr_def(ctx, 0x400844, 0x00000002); + if (IS_NVA3F(dev_priv->chipset)) + gr_def(ctx, 0x400894, 0x00001000); + gr_def(ctx, 0x4008e8, 0x00000003); + gr_def(ctx, 0x4008ec, 0x00001000); + if (dev_priv->chipset == 0x50) + cp_ctx(ctx, 0x400908, 0xb); + else if (dev_priv->chipset < 0xa0) + cp_ctx(ctx, 0x400908, 0xc); + else + cp_ctx(ctx, 0x400908, 0xe); + + if (dev_priv->chipset >= 0xa0) + cp_ctx(ctx, 0x400b00, 0x1); + if (IS_NVA3F(dev_priv->chipset)) { + cp_ctx(ctx, 0x400b10, 0x1); + gr_def(ctx, 0x400b10, 0x0001629d); + cp_ctx(ctx, 0x400b20, 0x1); + gr_def(ctx, 0x400b20, 0x0001629d); + } + + nv50_graph_construct_mmio_ddata(ctx); + + /* 0C00: VFETCH */ + cp_ctx(ctx, 0x400c08, 0x2); + gr_def(ctx, 0x400c08, 0x0000fe0c); + + /* 1000 */ + if (dev_priv->chipset < 0xa0) { + cp_ctx(ctx, 0x401008, 0x4); + gr_def(ctx, 0x401014, 0x00001000); + } else if (!IS_NVA3F(dev_priv->chipset)) { + cp_ctx(ctx, 0x401008, 0x5); + gr_def(ctx, 0x401018, 0x00001000); + } else { + cp_ctx(ctx, 0x401008, 0x5); + gr_def(ctx, 0x401018, 0x00004000); + } + + /* 1400 */ + cp_ctx(ctx, 0x401400, 0x8); + cp_ctx(ctx, 0x401424, 0x3); + if (dev_priv->chipset == 0x50) + gr_def(ctx, 0x40142c, 0x0001fd87); + else + gr_def(ctx, 0x40142c, 0x00000187); + cp_ctx(ctx, 0x401540, 0x5); + gr_def(ctx, 0x401550, 0x00001018); + + /* 1800: STREAMOUT */ + cp_ctx(ctx, 0x401814, 0x1); + gr_def(ctx, 0x401814, 0x000000ff); + if (dev_priv->chipset == 0x50) { + cp_ctx(ctx, 0x40181c, 0xe); + gr_def(ctx, 0x401850, 0x00000004); + } else if (dev_priv->chipset < 0xa0) { + cp_ctx(ctx, 0x40181c, 0xf); + gr_def(ctx, 0x401854, 0x00000004); + } else { + cp_ctx(ctx, 0x40181c, 0x13); + gr_def(ctx, 0x401864, 0x00000004); + } + + /* 1C00 */ + cp_ctx(ctx, 0x401c00, 0x1); + switch (dev_priv->chipset) { + case 0x50: + gr_def(ctx, 0x401c00, 0x0001005f); + break; + case 0x84: + case 0x86: + case 0x94: + gr_def(ctx, 0x401c00, 0x044d00df); + break; + case 0x92: + case 0x96: + case 0x98: + case 0xa0: + case 0xaa: + case 0xac: + gr_def(ctx, 0x401c00, 0x042500df); + break; + case 0xa3: + case 0xa5: + case 0xa8: + case 0xaf: + gr_def(ctx, 0x401c00, 0x142500df); + break; + } + + /* 2000 */ + + /* 2400 */ + cp_ctx(ctx, 0x402400, 0x1); + if (dev_priv->chipset == 0x50) + cp_ctx(ctx, 0x402408, 0x1); + else + cp_ctx(ctx, 0x402408, 0x2); + gr_def(ctx, 0x402408, 0x00000600); + + /* 2800: CSCHED */ + cp_ctx(ctx, 0x402800, 0x1); + if (dev_priv->chipset == 0x50) + gr_def(ctx, 0x402800, 0x00000006); + + /* 2C00: ZCULL */ + cp_ctx(ctx, 0x402c08, 0x6); + if (dev_priv->chipset != 0x50) + gr_def(ctx, 0x402c14, 0x01000000); + gr_def(ctx, 0x402c18, 0x000000ff); + if (dev_priv->chipset == 0x50) + cp_ctx(ctx, 0x402ca0, 0x1); + else + cp_ctx(ctx, 0x402ca0, 0x2); + if (dev_priv->chipset < 0xa0) + gr_def(ctx, 0x402ca0, 0x00000400); + else if (!IS_NVA3F(dev_priv->chipset)) + gr_def(ctx, 0x402ca0, 0x00000800); + else + gr_def(ctx, 0x402ca0, 0x00000400); + cp_ctx(ctx, 0x402cac, 0x4); + + /* 3000: ENG2D */ + cp_ctx(ctx, 0x403004, 0x1); + gr_def(ctx, 0x403004, 0x00000001); + + /* 3400 */ + if (dev_priv->chipset >= 0xa0) { + cp_ctx(ctx, 0x403404, 0x1); + gr_def(ctx, 0x403404, 0x00000001); + } + + /* 5000: CCACHE */ + cp_ctx(ctx, 0x405000, 0x1); + switch (dev_priv->chipset) { + case 0x50: + gr_def(ctx, 0x405000, 0x00300080); + break; + case 0x84: + case 0xa0: + case 0xa3: + case 0xa5: + case 0xa8: + case 0xaa: + case 0xac: + case 0xaf: + gr_def(ctx, 0x405000, 0x000e0080); + break; + case 0x86: + case 0x92: + case 0x94: + case 0x96: + case 0x98: + gr_def(ctx, 0x405000, 0x00000080); + break; + } + cp_ctx(ctx, 0x405014, 0x1); + gr_def(ctx, 0x405014, 0x00000004); + cp_ctx(ctx, 0x40501c, 0x1); + cp_ctx(ctx, 0x405024, 0x1); + cp_ctx(ctx, 0x40502c, 0x1); + + /* 6000? */ + if (dev_priv->chipset == 0x50) + cp_ctx(ctx, 0x4063e0, 0x1); + + /* 6800: M2MF */ + if (dev_priv->chipset < 0x90) { + cp_ctx(ctx, 0x406814, 0x2b); + gr_def(ctx, 0x406818, 0x00000f80); + gr_def(ctx, 0x406860, 0x007f0080); + gr_def(ctx, 0x40689c, 0x007f0080); + } else { + cp_ctx(ctx, 0x406814, 0x4); + if (dev_priv->chipset == 0x98) + gr_def(ctx, 0x406818, 0x00000f80); + else + gr_def(ctx, 0x406818, 0x00001f80); + if (IS_NVA3F(dev_priv->chipset)) + gr_def(ctx, 0x40681c, 0x00000030); + cp_ctx(ctx, 0x406830, 0x3); + } + + /* 7000: per-ROP group state */ + for (i = 0; i < 8; i++) { + if (units & (1<<(i+16))) { + cp_ctx(ctx, 0x407000 + (i<<8), 3); + if (dev_priv->chipset == 0x50) + gr_def(ctx, 0x407000 + (i<<8), 0x1b74f820); + else if (dev_priv->chipset != 0xa5) + gr_def(ctx, 0x407000 + (i<<8), 0x3b74f821); + else + gr_def(ctx, 0x407000 + (i<<8), 0x7b74f821); + gr_def(ctx, 0x407004 + (i<<8), 0x89058001); + + if (dev_priv->chipset == 0x50) { + cp_ctx(ctx, 0x407010 + (i<<8), 1); + } else if (dev_priv->chipset < 0xa0) { + cp_ctx(ctx, 0x407010 + (i<<8), 2); + gr_def(ctx, 0x407010 + (i<<8), 0x00001000); + gr_def(ctx, 0x407014 + (i<<8), 0x0000001f); + } else { + cp_ctx(ctx, 0x407010 + (i<<8), 3); + gr_def(ctx, 0x407010 + (i<<8), 0x00001000); + if (dev_priv->chipset != 0xa5) + gr_def(ctx, 0x407014 + (i<<8), 0x000000ff); + else + gr_def(ctx, 0x407014 + (i<<8), 0x000001ff); + } + + cp_ctx(ctx, 0x407080 + (i<<8), 4); + if (dev_priv->chipset != 0xa5) + gr_def(ctx, 0x407080 + (i<<8), 0x027c10fa); + else + gr_def(ctx, 0x407080 + (i<<8), 0x827c10fa); + if (dev_priv->chipset == 0x50) + gr_def(ctx, 0x407084 + (i<<8), 0x000000c0); + else + gr_def(ctx, 0x407084 + (i<<8), 0x400000c0); + gr_def(ctx, 0x407088 + (i<<8), 0xb7892080); + + if (dev_priv->chipset < 0xa0) + cp_ctx(ctx, 0x407094 + (i<<8), 1); + else if (!IS_NVA3F(dev_priv->chipset)) + cp_ctx(ctx, 0x407094 + (i<<8), 3); + else { + cp_ctx(ctx, 0x407094 + (i<<8), 4); + gr_def(ctx, 0x4070a0 + (i<<8), 1); + } + } + } + + cp_ctx(ctx, 0x407c00, 0x3); + if (dev_priv->chipset < 0x90) + gr_def(ctx, 0x407c00, 0x00010040); + else if (dev_priv->chipset < 0xa0) + gr_def(ctx, 0x407c00, 0x00390040); + else + gr_def(ctx, 0x407c00, 0x003d0040); + gr_def(ctx, 0x407c08, 0x00000022); + if (dev_priv->chipset >= 0xa0) { + cp_ctx(ctx, 0x407c10, 0x3); + cp_ctx(ctx, 0x407c20, 0x1); + cp_ctx(ctx, 0x407c2c, 0x1); + } + + if (dev_priv->chipset < 0xa0) { + cp_ctx(ctx, 0x407d00, 0x9); + } else { + cp_ctx(ctx, 0x407d00, 0x15); + } + if (dev_priv->chipset == 0x98) + gr_def(ctx, 0x407d08, 0x00380040); + else { + if (dev_priv->chipset < 0x90) + gr_def(ctx, 0x407d08, 0x00010040); + else if (dev_priv->chipset < 0xa0) + gr_def(ctx, 0x407d08, 0x00390040); + else + gr_def(ctx, 0x407d08, 0x003d0040); + gr_def(ctx, 0x407d0c, 0x00000022); + } + + /* 8000+: per-TP state */ + for (i = 0; i < 10; i++) { + if (units & (1<chipset < 0xa0) + base = 0x408000 + (i<<12); + else + base = 0x408000 + (i<<11); + if (dev_priv->chipset < 0xa0) + offset = base + 0xc00; + else + offset = base + 0x80; + cp_ctx(ctx, offset + 0x00, 1); + gr_def(ctx, offset + 0x00, 0x0000ff0a); + cp_ctx(ctx, offset + 0x08, 1); + + /* per-MP state */ + for (j = 0; j < (dev_priv->chipset < 0xa0 ? 2 : 4); j++) { + if (!(units & (1 << (j+24)))) continue; + if (dev_priv->chipset < 0xa0) + offset = base + 0x200 + (j<<7); + else + offset = base + 0x100 + (j<<7); + cp_ctx(ctx, offset, 0x20); + gr_def(ctx, offset + 0x00, 0x01800000); + gr_def(ctx, offset + 0x04, 0x00160000); + gr_def(ctx, offset + 0x08, 0x01800000); + gr_def(ctx, offset + 0x18, 0x0003ffff); + switch (dev_priv->chipset) { + case 0x50: + gr_def(ctx, offset + 0x1c, 0x00080000); + break; + case 0x84: + gr_def(ctx, offset + 0x1c, 0x00880000); + break; + case 0x86: + gr_def(ctx, offset + 0x1c, 0x008c0000); + break; + case 0x92: + case 0x96: + case 0x98: + gr_def(ctx, offset + 0x1c, 0x118c0000); + break; + case 0x94: + gr_def(ctx, offset + 0x1c, 0x10880000); + break; + case 0xa0: + case 0xa5: + gr_def(ctx, offset + 0x1c, 0x310c0000); + break; + case 0xa3: + case 0xa8: + case 0xaa: + case 0xac: + case 0xaf: + gr_def(ctx, offset + 0x1c, 0x300c0000); + break; + } + gr_def(ctx, offset + 0x40, 0x00010401); + if (dev_priv->chipset == 0x50) + gr_def(ctx, offset + 0x48, 0x00000040); + else + gr_def(ctx, offset + 0x48, 0x00000078); + gr_def(ctx, offset + 0x50, 0x000000bf); + gr_def(ctx, offset + 0x58, 0x00001210); + if (dev_priv->chipset == 0x50) + gr_def(ctx, offset + 0x5c, 0x00000080); + else + gr_def(ctx, offset + 0x5c, 0x08000080); + if (dev_priv->chipset >= 0xa0) + gr_def(ctx, offset + 0x68, 0x0000003e); + } + + if (dev_priv->chipset < 0xa0) + cp_ctx(ctx, base + 0x300, 0x4); + else + cp_ctx(ctx, base + 0x300, 0x5); + if (dev_priv->chipset == 0x50) + gr_def(ctx, base + 0x304, 0x00007070); + else if (dev_priv->chipset < 0xa0) + gr_def(ctx, base + 0x304, 0x00027070); + else if (!IS_NVA3F(dev_priv->chipset)) + gr_def(ctx, base + 0x304, 0x01127070); + else + gr_def(ctx, base + 0x304, 0x05127070); + + if (dev_priv->chipset < 0xa0) + cp_ctx(ctx, base + 0x318, 1); + else + cp_ctx(ctx, base + 0x320, 1); + if (dev_priv->chipset == 0x50) + gr_def(ctx, base + 0x318, 0x0003ffff); + else if (dev_priv->chipset < 0xa0) + gr_def(ctx, base + 0x318, 0x03ffffff); + else + gr_def(ctx, base + 0x320, 0x07ffffff); + + if (dev_priv->chipset < 0xa0) + cp_ctx(ctx, base + 0x324, 5); + else + cp_ctx(ctx, base + 0x328, 4); + + if (dev_priv->chipset < 0xa0) { + cp_ctx(ctx, base + 0x340, 9); + offset = base + 0x340; + } else if (!IS_NVA3F(dev_priv->chipset)) { + cp_ctx(ctx, base + 0x33c, 0xb); + offset = base + 0x344; + } else { + cp_ctx(ctx, base + 0x33c, 0xd); + offset = base + 0x344; + } + gr_def(ctx, offset + 0x0, 0x00120407); + gr_def(ctx, offset + 0x4, 0x05091507); + if (dev_priv->chipset == 0x84) + gr_def(ctx, offset + 0x8, 0x05100202); + else + gr_def(ctx, offset + 0x8, 0x05010202); + gr_def(ctx, offset + 0xc, 0x00030201); + if (dev_priv->chipset == 0xa3) + cp_ctx(ctx, base + 0x36c, 1); + + cp_ctx(ctx, base + 0x400, 2); + gr_def(ctx, base + 0x404, 0x00000040); + cp_ctx(ctx, base + 0x40c, 2); + gr_def(ctx, base + 0x40c, 0x0d0c0b0a); + gr_def(ctx, base + 0x410, 0x00141210); + + if (dev_priv->chipset < 0xa0) + offset = base + 0x800; + else + offset = base + 0x500; + cp_ctx(ctx, offset, 6); + gr_def(ctx, offset + 0x0, 0x000001f0); + gr_def(ctx, offset + 0x4, 0x00000001); + gr_def(ctx, offset + 0x8, 0x00000003); + if (dev_priv->chipset == 0x50 || IS_NVAAF(dev_priv->chipset)) + gr_def(ctx, offset + 0xc, 0x00008000); + gr_def(ctx, offset + 0x14, 0x00039e00); + cp_ctx(ctx, offset + 0x1c, 2); + if (dev_priv->chipset == 0x50) + gr_def(ctx, offset + 0x1c, 0x00000040); + else + gr_def(ctx, offset + 0x1c, 0x00000100); + gr_def(ctx, offset + 0x20, 0x00003800); + + if (dev_priv->chipset >= 0xa0) { + cp_ctx(ctx, base + 0x54c, 2); + if (!IS_NVA3F(dev_priv->chipset)) + gr_def(ctx, base + 0x54c, 0x003fe006); + else + gr_def(ctx, base + 0x54c, 0x003fe007); + gr_def(ctx, base + 0x550, 0x003fe000); + } + + if (dev_priv->chipset < 0xa0) + offset = base + 0xa00; + else + offset = base + 0x680; + cp_ctx(ctx, offset, 1); + gr_def(ctx, offset, 0x00404040); + + if (dev_priv->chipset < 0xa0) + offset = base + 0xe00; + else + offset = base + 0x700; + cp_ctx(ctx, offset, 2); + if (dev_priv->chipset < 0xa0) + gr_def(ctx, offset, 0x0077f005); + else if (dev_priv->chipset == 0xa5) + gr_def(ctx, offset, 0x6cf7f007); + else if (dev_priv->chipset == 0xa8) + gr_def(ctx, offset, 0x6cfff007); + else if (dev_priv->chipset == 0xac) + gr_def(ctx, offset, 0x0cfff007); + else + gr_def(ctx, offset, 0x0cf7f007); + if (dev_priv->chipset == 0x50) + gr_def(ctx, offset + 0x4, 0x00007fff); + else if (dev_priv->chipset < 0xa0) + gr_def(ctx, offset + 0x4, 0x003f7fff); + else + gr_def(ctx, offset + 0x4, 0x02bf7fff); + cp_ctx(ctx, offset + 0x2c, 1); + if (dev_priv->chipset == 0x50) { + cp_ctx(ctx, offset + 0x50, 9); + gr_def(ctx, offset + 0x54, 0x000003ff); + gr_def(ctx, offset + 0x58, 0x00000003); + gr_def(ctx, offset + 0x5c, 0x00000003); + gr_def(ctx, offset + 0x60, 0x000001ff); + gr_def(ctx, offset + 0x64, 0x0000001f); + gr_def(ctx, offset + 0x68, 0x0000000f); + gr_def(ctx, offset + 0x6c, 0x0000000f); + } else if (dev_priv->chipset < 0xa0) { + cp_ctx(ctx, offset + 0x50, 1); + cp_ctx(ctx, offset + 0x70, 1); + } else { + cp_ctx(ctx, offset + 0x50, 1); + cp_ctx(ctx, offset + 0x60, 5); + } + } + } +} + +static void +dd_emit(struct nouveau_grctx *ctx, int num, uint32_t val) { + int i; + if (val && ctx->mode == NOUVEAU_GRCTX_VALS) + for (i = 0; i < num; i++) + nv_wo32(ctx->data, 4 * (ctx->ctxvals_pos + i), val); + ctx->ctxvals_pos += num; +} + +static void +nv50_graph_construct_mmio_ddata(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + int base, num; + base = ctx->ctxvals_pos; + + /* tesla state */ + dd_emit(ctx, 1, 0); /* 00000001 UNK0F90 */ + dd_emit(ctx, 1, 0); /* 00000001 UNK135C */ + + /* SRC_TIC state */ + dd_emit(ctx, 1, 0); /* 00000007 SRC_TILE_MODE_Z */ + dd_emit(ctx, 1, 2); /* 00000007 SRC_TILE_MODE_Y */ + dd_emit(ctx, 1, 1); /* 00000001 SRC_LINEAR #1 */ + dd_emit(ctx, 1, 0); /* 000000ff SRC_ADDRESS_HIGH */ + dd_emit(ctx, 1, 0); /* 00000001 SRC_SRGB */ + if (dev_priv->chipset >= 0x94) + dd_emit(ctx, 1, 0); /* 00000003 eng2d UNK0258 */ + dd_emit(ctx, 1, 1); /* 00000fff SRC_DEPTH */ + dd_emit(ctx, 1, 0x100); /* 0000ffff SRC_HEIGHT */ + + /* turing state */ + dd_emit(ctx, 1, 0); /* 0000000f TEXTURES_LOG2 */ + dd_emit(ctx, 1, 0); /* 0000000f SAMPLERS_LOG2 */ + dd_emit(ctx, 1, 0); /* 000000ff CB_DEF_ADDRESS_HIGH */ + dd_emit(ctx, 1, 0); /* ffffffff CB_DEF_ADDRESS_LOW */ + dd_emit(ctx, 1, 0); /* ffffffff SHARED_SIZE */ + dd_emit(ctx, 1, 2); /* ffffffff REG_MODE */ + dd_emit(ctx, 1, 1); /* 0000ffff BLOCK_ALLOC_THREADS */ + dd_emit(ctx, 1, 1); /* 00000001 LANES32 */ + dd_emit(ctx, 1, 0); /* 000000ff UNK370 */ + dd_emit(ctx, 1, 0); /* 000000ff USER_PARAM_UNK */ + dd_emit(ctx, 1, 0); /* 000000ff USER_PARAM_COUNT */ + dd_emit(ctx, 1, 1); /* 000000ff UNK384 bits 8-15 */ + dd_emit(ctx, 1, 0x3fffff); /* 003fffff TIC_LIMIT */ + dd_emit(ctx, 1, 0x1fff); /* 000fffff TSC_LIMIT */ + dd_emit(ctx, 1, 0); /* 0000ffff CB_ADDR_INDEX */ + dd_emit(ctx, 1, 1); /* 000007ff BLOCKDIM_X */ + dd_emit(ctx, 1, 1); /* 000007ff BLOCKDIM_XMY */ + dd_emit(ctx, 1, 0); /* 00000001 BLOCKDIM_XMY_OVERFLOW */ + dd_emit(ctx, 1, 1); /* 0003ffff BLOCKDIM_XMYMZ */ + dd_emit(ctx, 1, 1); /* 000007ff BLOCKDIM_Y */ + dd_emit(ctx, 1, 1); /* 0000007f BLOCKDIM_Z */ + dd_emit(ctx, 1, 4); /* 000000ff CP_REG_ALLOC_TEMP */ + dd_emit(ctx, 1, 1); /* 00000001 BLOCKDIM_DIRTY */ + if (IS_NVA3F(dev_priv->chipset)) + dd_emit(ctx, 1, 0); /* 00000003 UNK03E8 */ + dd_emit(ctx, 1, 1); /* 0000007f BLOCK_ALLOC_HALFWARPS */ + dd_emit(ctx, 1, 1); /* 00000007 LOCAL_WARPS_NO_CLAMP */ + dd_emit(ctx, 1, 7); /* 00000007 LOCAL_WARPS_LOG_ALLOC */ + dd_emit(ctx, 1, 1); /* 00000007 STACK_WARPS_NO_CLAMP */ + dd_emit(ctx, 1, 7); /* 00000007 STACK_WARPS_LOG_ALLOC */ + dd_emit(ctx, 1, 1); /* 00001fff BLOCK_ALLOC_REGSLOTS_PACKED */ + dd_emit(ctx, 1, 1); /* 00001fff BLOCK_ALLOC_REGSLOTS_STRIDED */ + dd_emit(ctx, 1, 1); /* 000007ff BLOCK_ALLOC_THREADS */ + + /* compat 2d state */ + if (dev_priv->chipset == 0x50) { + dd_emit(ctx, 4, 0); /* 0000ffff clip X, Y, W, H */ + + dd_emit(ctx, 1, 1); /* ffffffff chroma COLOR_FORMAT */ + + dd_emit(ctx, 1, 1); /* ffffffff pattern COLOR_FORMAT */ + dd_emit(ctx, 1, 0); /* ffffffff pattern SHAPE */ + dd_emit(ctx, 1, 1); /* ffffffff pattern PATTERN_SELECT */ + + dd_emit(ctx, 1, 0xa); /* ffffffff surf2d SRC_FORMAT */ + dd_emit(ctx, 1, 0); /* ffffffff surf2d DMA_SRC */ + dd_emit(ctx, 1, 0); /* 000000ff surf2d SRC_ADDRESS_HIGH */ + dd_emit(ctx, 1, 0); /* ffffffff surf2d SRC_ADDRESS_LOW */ + dd_emit(ctx, 1, 0x40); /* 0000ffff surf2d SRC_PITCH */ + dd_emit(ctx, 1, 0); /* 0000000f surf2d SRC_TILE_MODE_Z */ + dd_emit(ctx, 1, 2); /* 0000000f surf2d SRC_TILE_MODE_Y */ + dd_emit(ctx, 1, 0x100); /* ffffffff surf2d SRC_HEIGHT */ + dd_emit(ctx, 1, 1); /* 00000001 surf2d SRC_LINEAR */ + dd_emit(ctx, 1, 0x100); /* ffffffff surf2d SRC_WIDTH */ + + dd_emit(ctx, 1, 0); /* 0000ffff gdirect CLIP_B_X */ + dd_emit(ctx, 1, 0); /* 0000ffff gdirect CLIP_B_Y */ + dd_emit(ctx, 1, 0); /* 0000ffff gdirect CLIP_C_X */ + dd_emit(ctx, 1, 0); /* 0000ffff gdirect CLIP_C_Y */ + dd_emit(ctx, 1, 0); /* 0000ffff gdirect CLIP_D_X */ + dd_emit(ctx, 1, 0); /* 0000ffff gdirect CLIP_D_Y */ + dd_emit(ctx, 1, 1); /* ffffffff gdirect COLOR_FORMAT */ + dd_emit(ctx, 1, 0); /* ffffffff gdirect OPERATION */ + dd_emit(ctx, 1, 0); /* 0000ffff gdirect POINT_X */ + dd_emit(ctx, 1, 0); /* 0000ffff gdirect POINT_Y */ + + dd_emit(ctx, 1, 0); /* 0000ffff blit SRC_Y */ + dd_emit(ctx, 1, 0); /* ffffffff blit OPERATION */ + + dd_emit(ctx, 1, 0); /* ffffffff ifc OPERATION */ + + dd_emit(ctx, 1, 0); /* ffffffff iifc INDEX_FORMAT */ + dd_emit(ctx, 1, 0); /* ffffffff iifc LUT_OFFSET */ + dd_emit(ctx, 1, 4); /* ffffffff iifc COLOR_FORMAT */ + dd_emit(ctx, 1, 0); /* ffffffff iifc OPERATION */ + } + + /* m2mf state */ + dd_emit(ctx, 1, 0); /* ffffffff m2mf LINE_COUNT */ + dd_emit(ctx, 1, 0); /* ffffffff m2mf LINE_LENGTH_IN */ + dd_emit(ctx, 2, 0); /* ffffffff m2mf OFFSET_IN, OFFSET_OUT */ + dd_emit(ctx, 1, 1); /* ffffffff m2mf TILING_DEPTH_OUT */ + dd_emit(ctx, 1, 0x100); /* ffffffff m2mf TILING_HEIGHT_OUT */ + dd_emit(ctx, 1, 0); /* ffffffff m2mf TILING_POSITION_OUT_Z */ + dd_emit(ctx, 1, 1); /* 00000001 m2mf LINEAR_OUT */ + dd_emit(ctx, 2, 0); /* 0000ffff m2mf TILING_POSITION_OUT_X, Y */ + dd_emit(ctx, 1, 0x100); /* ffffffff m2mf TILING_PITCH_OUT */ + dd_emit(ctx, 1, 1); /* ffffffff m2mf TILING_DEPTH_IN */ + dd_emit(ctx, 1, 0x100); /* ffffffff m2mf TILING_HEIGHT_IN */ + dd_emit(ctx, 1, 0); /* ffffffff m2mf TILING_POSITION_IN_Z */ + dd_emit(ctx, 1, 1); /* 00000001 m2mf LINEAR_IN */ + dd_emit(ctx, 2, 0); /* 0000ffff m2mf TILING_POSITION_IN_X, Y */ + dd_emit(ctx, 1, 0x100); /* ffffffff m2mf TILING_PITCH_IN */ + + /* more compat 2d state */ + if (dev_priv->chipset == 0x50) { + dd_emit(ctx, 1, 1); /* ffffffff line COLOR_FORMAT */ + dd_emit(ctx, 1, 0); /* ffffffff line OPERATION */ + + dd_emit(ctx, 1, 1); /* ffffffff triangle COLOR_FORMAT */ + dd_emit(ctx, 1, 0); /* ffffffff triangle OPERATION */ + + dd_emit(ctx, 1, 0); /* 0000000f sifm TILE_MODE_Z */ + dd_emit(ctx, 1, 2); /* 0000000f sifm TILE_MODE_Y */ + dd_emit(ctx, 1, 0); /* 000000ff sifm FORMAT_FILTER */ + dd_emit(ctx, 1, 1); /* 000000ff sifm FORMAT_ORIGIN */ + dd_emit(ctx, 1, 0); /* 0000ffff sifm SRC_PITCH */ + dd_emit(ctx, 1, 1); /* 00000001 sifm SRC_LINEAR */ + dd_emit(ctx, 1, 0); /* 000000ff sifm SRC_OFFSET_HIGH */ + dd_emit(ctx, 1, 0); /* ffffffff sifm SRC_OFFSET */ + dd_emit(ctx, 1, 0); /* 0000ffff sifm SRC_HEIGHT */ + dd_emit(ctx, 1, 0); /* 0000ffff sifm SRC_WIDTH */ + dd_emit(ctx, 1, 3); /* ffffffff sifm COLOR_FORMAT */ + dd_emit(ctx, 1, 0); /* ffffffff sifm OPERATION */ + + dd_emit(ctx, 1, 0); /* ffffffff sifc OPERATION */ + } + + /* tesla state */ + dd_emit(ctx, 1, 0); /* 0000000f GP_TEXTURES_LOG2 */ + dd_emit(ctx, 1, 0); /* 0000000f GP_SAMPLERS_LOG2 */ + dd_emit(ctx, 1, 0); /* 000000ff */ + dd_emit(ctx, 1, 0); /* ffffffff */ + dd_emit(ctx, 1, 4); /* 000000ff UNK12B0_0 */ + dd_emit(ctx, 1, 0x70); /* 000000ff UNK12B0_1 */ + dd_emit(ctx, 1, 0x80); /* 000000ff UNK12B0_3 */ + dd_emit(ctx, 1, 0); /* 000000ff UNK12B0_2 */ + dd_emit(ctx, 1, 0); /* 0000000f FP_TEXTURES_LOG2 */ + dd_emit(ctx, 1, 0); /* 0000000f FP_SAMPLERS_LOG2 */ + if (IS_NVA3F(dev_priv->chipset)) { + dd_emit(ctx, 1, 0); /* ffffffff */ + dd_emit(ctx, 1, 0); /* 0000007f MULTISAMPLE_SAMPLES_LOG2 */ + } else { + dd_emit(ctx, 1, 0); /* 0000000f MULTISAMPLE_SAMPLES_LOG2 */ + } + dd_emit(ctx, 1, 0xc); /* 000000ff SEMANTIC_COLOR.BFC0_ID */ + if (dev_priv->chipset != 0x50) + dd_emit(ctx, 1, 0); /* 00000001 SEMANTIC_COLOR.CLMP_EN */ + dd_emit(ctx, 1, 8); /* 000000ff SEMANTIC_COLOR.COLR_NR */ + dd_emit(ctx, 1, 0x14); /* 000000ff SEMANTIC_COLOR.FFC0_ID */ + if (dev_priv->chipset == 0x50) { + dd_emit(ctx, 1, 0); /* 000000ff SEMANTIC_LAYER */ + dd_emit(ctx, 1, 0); /* 00000001 */ + } else { + dd_emit(ctx, 1, 0); /* 00000001 SEMANTIC_PTSZ.ENABLE */ + dd_emit(ctx, 1, 0x29); /* 000000ff SEMANTIC_PTSZ.PTSZ_ID */ + dd_emit(ctx, 1, 0x27); /* 000000ff SEMANTIC_PRIM */ + dd_emit(ctx, 1, 0x26); /* 000000ff SEMANTIC_LAYER */ + dd_emit(ctx, 1, 8); /* 0000000f SMENATIC_CLIP.CLIP_HIGH */ + dd_emit(ctx, 1, 4); /* 000000ff SEMANTIC_CLIP.CLIP_LO */ + dd_emit(ctx, 1, 0x27); /* 000000ff UNK0FD4 */ + dd_emit(ctx, 1, 0); /* 00000001 UNK1900 */ + } + dd_emit(ctx, 1, 0); /* 00000007 RT_CONTROL_MAP0 */ + dd_emit(ctx, 1, 1); /* 00000007 RT_CONTROL_MAP1 */ + dd_emit(ctx, 1, 2); /* 00000007 RT_CONTROL_MAP2 */ + dd_emit(ctx, 1, 3); /* 00000007 RT_CONTROL_MAP3 */ + dd_emit(ctx, 1, 4); /* 00000007 RT_CONTROL_MAP4 */ + dd_emit(ctx, 1, 5); /* 00000007 RT_CONTROL_MAP5 */ + dd_emit(ctx, 1, 6); /* 00000007 RT_CONTROL_MAP6 */ + dd_emit(ctx, 1, 7); /* 00000007 RT_CONTROL_MAP7 */ + dd_emit(ctx, 1, 1); /* 0000000f RT_CONTROL_COUNT */ + dd_emit(ctx, 8, 0); /* 00000001 RT_HORIZ_UNK */ + dd_emit(ctx, 8, 0); /* ffffffff RT_ADDRESS_LOW */ + dd_emit(ctx, 1, 0xcf); /* 000000ff RT_FORMAT */ + dd_emit(ctx, 7, 0); /* 000000ff RT_FORMAT */ + if (dev_priv->chipset != 0x50) + dd_emit(ctx, 3, 0); /* 1, 1, 1 */ + else + dd_emit(ctx, 2, 0); /* 1, 1 */ + dd_emit(ctx, 1, 0); /* ffffffff GP_ENABLE */ + dd_emit(ctx, 1, 0x80); /* 0000ffff GP_VERTEX_OUTPUT_COUNT*/ + dd_emit(ctx, 1, 4); /* 000000ff GP_REG_ALLOC_RESULT */ + dd_emit(ctx, 1, 4); /* 000000ff GP_RESULT_MAP_SIZE */ + if (IS_NVA3F(dev_priv->chipset)) { + dd_emit(ctx, 1, 3); /* 00000003 */ + dd_emit(ctx, 1, 0); /* 00000001 UNK1418. Alone. */ + } + if (dev_priv->chipset != 0x50) + dd_emit(ctx, 1, 3); /* 00000003 UNK15AC */ + dd_emit(ctx, 1, 1); /* ffffffff RASTERIZE_ENABLE */ + dd_emit(ctx, 1, 0); /* 00000001 FP_CONTROL.EXPORTS_Z */ + if (dev_priv->chipset != 0x50) + dd_emit(ctx, 1, 0); /* 00000001 FP_CONTROL.MULTIPLE_RESULTS */ + dd_emit(ctx, 1, 0x12); /* 000000ff FP_INTERPOLANT_CTRL.COUNT */ + dd_emit(ctx, 1, 0x10); /* 000000ff FP_INTERPOLANT_CTRL.COUNT_NONFLAT */ + dd_emit(ctx, 1, 0xc); /* 000000ff FP_INTERPOLANT_CTRL.OFFSET */ + dd_emit(ctx, 1, 1); /* 00000001 FP_INTERPOLANT_CTRL.UMASK.W */ + dd_emit(ctx, 1, 0); /* 00000001 FP_INTERPOLANT_CTRL.UMASK.X */ + dd_emit(ctx, 1, 0); /* 00000001 FP_INTERPOLANT_CTRL.UMASK.Y */ + dd_emit(ctx, 1, 0); /* 00000001 FP_INTERPOLANT_CTRL.UMASK.Z */ + dd_emit(ctx, 1, 4); /* 000000ff FP_RESULT_COUNT */ + dd_emit(ctx, 1, 2); /* ffffffff REG_MODE */ + dd_emit(ctx, 1, 4); /* 000000ff FP_REG_ALLOC_TEMP */ + if (dev_priv->chipset >= 0xa0) + dd_emit(ctx, 1, 0); /* ffffffff */ + dd_emit(ctx, 1, 0); /* 00000001 GP_BUILTIN_RESULT_EN.LAYER_IDX */ + dd_emit(ctx, 1, 0); /* ffffffff STRMOUT_ENABLE */ + dd_emit(ctx, 1, 0x3fffff); /* 003fffff TIC_LIMIT */ + dd_emit(ctx, 1, 0x1fff); /* 000fffff TSC_LIMIT */ + dd_emit(ctx, 1, 0); /* 00000001 VERTEX_TWO_SIDE_ENABLE*/ + if (dev_priv->chipset != 0x50) + dd_emit(ctx, 8, 0); /* 00000001 */ + if (dev_priv->chipset >= 0xa0) { + dd_emit(ctx, 1, 1); /* 00000007 VTX_ATTR_DEFINE.COMP */ + dd_emit(ctx, 1, 1); /* 00000007 VTX_ATTR_DEFINE.SIZE */ + dd_emit(ctx, 1, 2); /* 00000007 VTX_ATTR_DEFINE.TYPE */ + dd_emit(ctx, 1, 0); /* 000000ff VTX_ATTR_DEFINE.ATTR */ + } + dd_emit(ctx, 1, 4); /* 0000007f VP_RESULT_MAP_SIZE */ + dd_emit(ctx, 1, 0x14); /* 0000001f ZETA_FORMAT */ + dd_emit(ctx, 1, 1); /* 00000001 ZETA_ENABLE */ + dd_emit(ctx, 1, 0); /* 0000000f VP_TEXTURES_LOG2 */ + dd_emit(ctx, 1, 0); /* 0000000f VP_SAMPLERS_LOG2 */ + if (IS_NVA3F(dev_priv->chipset)) + dd_emit(ctx, 1, 0); /* 00000001 */ + dd_emit(ctx, 1, 2); /* 00000003 POLYGON_MODE_BACK */ + if (dev_priv->chipset >= 0xa0) + dd_emit(ctx, 1, 0); /* 00000003 VTX_ATTR_DEFINE.SIZE - 1 */ + dd_emit(ctx, 1, 0); /* 0000ffff CB_ADDR_INDEX */ + if (dev_priv->chipset >= 0xa0) + dd_emit(ctx, 1, 0); /* 00000003 */ + dd_emit(ctx, 1, 0); /* 00000001 CULL_FACE_ENABLE */ + dd_emit(ctx, 1, 1); /* 00000003 CULL_FACE */ + dd_emit(ctx, 1, 0); /* 00000001 FRONT_FACE */ + dd_emit(ctx, 1, 2); /* 00000003 POLYGON_MODE_FRONT */ + dd_emit(ctx, 1, 0x1000); /* 00007fff UNK141C */ + if (dev_priv->chipset != 0x50) { + dd_emit(ctx, 1, 0xe00); /* 7fff */ + dd_emit(ctx, 1, 0x1000); /* 7fff */ + dd_emit(ctx, 1, 0x1e00); /* 7fff */ + } + dd_emit(ctx, 1, 0); /* 00000001 BEGIN_END_ACTIVE */ + dd_emit(ctx, 1, 1); /* 00000001 POLYGON_MODE_??? */ + dd_emit(ctx, 1, 1); /* 000000ff GP_REG_ALLOC_TEMP / 4 rounded up */ + dd_emit(ctx, 1, 1); /* 000000ff FP_REG_ALLOC_TEMP... without /4? */ + dd_emit(ctx, 1, 1); /* 000000ff VP_REG_ALLOC_TEMP / 4 rounded up */ + dd_emit(ctx, 1, 1); /* 00000001 */ + dd_emit(ctx, 1, 0); /* 00000001 */ + dd_emit(ctx, 1, 0); /* 00000001 VTX_ATTR_MASK_UNK0 nonempty */ + dd_emit(ctx, 1, 0); /* 00000001 VTX_ATTR_MASK_UNK1 nonempty */ + dd_emit(ctx, 1, 0x200); /* 0003ffff GP_VERTEX_OUTPUT_COUNT*GP_REG_ALLOC_RESULT */ + if (IS_NVA3F(dev_priv->chipset)) + dd_emit(ctx, 1, 0x200); + dd_emit(ctx, 1, 0); /* 00000001 */ + if (dev_priv->chipset < 0xa0) { + dd_emit(ctx, 1, 1); /* 00000001 */ + dd_emit(ctx, 1, 0x70); /* 000000ff */ + dd_emit(ctx, 1, 0x80); /* 000000ff */ + dd_emit(ctx, 1, 0); /* 000000ff */ + dd_emit(ctx, 1, 0); /* 00000001 */ + dd_emit(ctx, 1, 1); /* 00000001 */ + dd_emit(ctx, 1, 0x70); /* 000000ff */ + dd_emit(ctx, 1, 0x80); /* 000000ff */ + dd_emit(ctx, 1, 0); /* 000000ff */ + } else { + dd_emit(ctx, 1, 1); /* 00000001 */ + dd_emit(ctx, 1, 0xf0); /* 000000ff */ + dd_emit(ctx, 1, 0xff); /* 000000ff */ + dd_emit(ctx, 1, 0); /* 000000ff */ + dd_emit(ctx, 1, 0); /* 00000001 */ + dd_emit(ctx, 1, 1); /* 00000001 */ + dd_emit(ctx, 1, 0xf0); /* 000000ff */ + dd_emit(ctx, 1, 0xff); /* 000000ff */ + dd_emit(ctx, 1, 0); /* 000000ff */ + dd_emit(ctx, 1, 9); /* 0000003f UNK114C.COMP,SIZE */ + } + + /* eng2d state */ + dd_emit(ctx, 1, 0); /* 00000001 eng2d COLOR_KEY_ENABLE */ + dd_emit(ctx, 1, 0); /* 00000007 eng2d COLOR_KEY_FORMAT */ + dd_emit(ctx, 1, 1); /* ffffffff eng2d DST_DEPTH */ + dd_emit(ctx, 1, 0xcf); /* 000000ff eng2d DST_FORMAT */ + dd_emit(ctx, 1, 0); /* ffffffff eng2d DST_LAYER */ + dd_emit(ctx, 1, 1); /* 00000001 eng2d DST_LINEAR */ + dd_emit(ctx, 1, 0); /* 00000007 eng2d PATTERN_COLOR_FORMAT */ + dd_emit(ctx, 1, 0); /* 00000007 eng2d OPERATION */ + dd_emit(ctx, 1, 0); /* 00000003 eng2d PATTERN_SELECT */ + dd_emit(ctx, 1, 0xcf); /* 000000ff eng2d SIFC_FORMAT */ + dd_emit(ctx, 1, 0); /* 00000001 eng2d SIFC_BITMAP_ENABLE */ + dd_emit(ctx, 1, 2); /* 00000003 eng2d SIFC_BITMAP_UNK808 */ + dd_emit(ctx, 1, 0); /* ffffffff eng2d BLIT_DU_DX_FRACT */ + dd_emit(ctx, 1, 1); /* ffffffff eng2d BLIT_DU_DX_INT */ + dd_emit(ctx, 1, 0); /* ffffffff eng2d BLIT_DV_DY_FRACT */ + dd_emit(ctx, 1, 1); /* ffffffff eng2d BLIT_DV_DY_INT */ + dd_emit(ctx, 1, 0); /* 00000001 eng2d BLIT_CONTROL_FILTER */ + dd_emit(ctx, 1, 0xcf); /* 000000ff eng2d DRAW_COLOR_FORMAT */ + dd_emit(ctx, 1, 0xcf); /* 000000ff eng2d SRC_FORMAT */ + dd_emit(ctx, 1, 1); /* 00000001 eng2d SRC_LINEAR #2 */ + + num = ctx->ctxvals_pos - base; + ctx->ctxvals_pos = base; + if (IS_NVA3F(dev_priv->chipset)) + cp_ctx(ctx, 0x404800, num); + else + cp_ctx(ctx, 0x405400, num); +} + +/* + * xfer areas. These are a pain. + * + * There are 2 xfer areas: the first one is big and contains all sorts of + * stuff, the second is small and contains some per-TP context. + * + * Each area is split into 8 "strands". The areas, when saved to grctx, + * are made of 8-word blocks. Each block contains a single word from + * each strand. The strands are independent of each other, their + * addresses are unrelated to each other, and data in them is closely + * packed together. The strand layout varies a bit between cards: here + * and there, a single word is thrown out in the middle and the whole + * strand is offset by a bit from corresponding one on another chipset. + * For this reason, addresses of stuff in strands are almost useless. + * Knowing sequence of stuff and size of gaps between them is much more + * useful, and that's how we build the strands in our generator. + * + * NVA0 takes this mess to a whole new level by cutting the old strands + * into a few dozen pieces [known as genes], rearranging them randomly, + * and putting them back together to make new strands. Hopefully these + * genes correspond more or less directly to the same PGRAPH subunits + * as in 400040 register. + * + * The most common value in default context is 0, and when the genes + * are separated by 0's, gene bounduaries are quite speculative... + * some of them can be clearly deduced, others can be guessed, and yet + * others won't be resolved without figuring out the real meaning of + * given ctxval. For the same reason, ending point of each strand + * is unknown. Except for strand 0, which is the longest strand and + * its end corresponds to end of the whole xfer. + * + * An unsolved mystery is the seek instruction: it takes an argument + * in bits 8-18, and that argument is clearly the place in strands to + * seek to... but the offsets don't seem to correspond to offsets as + * seen in grctx. Perhaps there's another, real, not randomly-changing + * addressing in strands, and the xfer insn just happens to skip over + * the unused bits? NV10-NV30 PIPE comes to mind... + * + * As far as I know, there's no way to access the xfer areas directly + * without the help of ctxprog. + */ + +static void +xf_emit(struct nouveau_grctx *ctx, int num, uint32_t val) { + int i; + if (val && ctx->mode == NOUVEAU_GRCTX_VALS) + for (i = 0; i < num; i++) + nv_wo32(ctx->data, 4 * (ctx->ctxvals_pos + (i << 3)), val); + ctx->ctxvals_pos += num << 3; +} + +/* Gene declarations... */ + +static void nv50_graph_construct_gene_dispatch(struct nouveau_grctx *ctx); +static void nv50_graph_construct_gene_m2mf(struct nouveau_grctx *ctx); +static void nv50_graph_construct_gene_ccache(struct nouveau_grctx *ctx); +static void nv50_graph_construct_gene_unk10xx(struct nouveau_grctx *ctx); +static void nv50_graph_construct_gene_unk14xx(struct nouveau_grctx *ctx); +static void nv50_graph_construct_gene_zcull(struct nouveau_grctx *ctx); +static void nv50_graph_construct_gene_clipid(struct nouveau_grctx *ctx); +static void nv50_graph_construct_gene_unk24xx(struct nouveau_grctx *ctx); +static void nv50_graph_construct_gene_vfetch(struct nouveau_grctx *ctx); +static void nv50_graph_construct_gene_eng2d(struct nouveau_grctx *ctx); +static void nv50_graph_construct_gene_csched(struct nouveau_grctx *ctx); +static void nv50_graph_construct_gene_unk1cxx(struct nouveau_grctx *ctx); +static void nv50_graph_construct_gene_strmout(struct nouveau_grctx *ctx); +static void nv50_graph_construct_gene_unk34xx(struct nouveau_grctx *ctx); +static void nv50_graph_construct_gene_ropm1(struct nouveau_grctx *ctx); +static void nv50_graph_construct_gene_ropm2(struct nouveau_grctx *ctx); +static void nv50_graph_construct_gene_ropc(struct nouveau_grctx *ctx); +static void nv50_graph_construct_xfer_tp(struct nouveau_grctx *ctx); + +static void +nv50_graph_construct_xfer1(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + int i; + int offset; + int size = 0; + uint32_t units = nv_rd32 (ctx->dev, 0x1540); + + offset = (ctx->ctxvals_pos+0x3f)&~0x3f; + ctx->ctxvals_base = offset; + + if (dev_priv->chipset < 0xa0) { + /* Strand 0 */ + ctx->ctxvals_pos = offset; + nv50_graph_construct_gene_dispatch(ctx); + nv50_graph_construct_gene_m2mf(ctx); + nv50_graph_construct_gene_unk24xx(ctx); + nv50_graph_construct_gene_clipid(ctx); + nv50_graph_construct_gene_zcull(ctx); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + + /* Strand 1 */ + ctx->ctxvals_pos = offset + 0x1; + nv50_graph_construct_gene_vfetch(ctx); + nv50_graph_construct_gene_eng2d(ctx); + nv50_graph_construct_gene_csched(ctx); + nv50_graph_construct_gene_ropm1(ctx); + nv50_graph_construct_gene_ropm2(ctx); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + + /* Strand 2 */ + ctx->ctxvals_pos = offset + 0x2; + nv50_graph_construct_gene_ccache(ctx); + nv50_graph_construct_gene_unk1cxx(ctx); + nv50_graph_construct_gene_strmout(ctx); + nv50_graph_construct_gene_unk14xx(ctx); + nv50_graph_construct_gene_unk10xx(ctx); + nv50_graph_construct_gene_unk34xx(ctx); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + + /* Strand 3: per-ROP group state */ + ctx->ctxvals_pos = offset + 3; + for (i = 0; i < 6; i++) + if (units & (1 << (i + 16))) + nv50_graph_construct_gene_ropc(ctx); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + + /* Strands 4-7: per-TP state */ + for (i = 0; i < 4; i++) { + ctx->ctxvals_pos = offset + 4 + i; + if (units & (1 << (2 * i))) + nv50_graph_construct_xfer_tp(ctx); + if (units & (1 << (2 * i + 1))) + nv50_graph_construct_xfer_tp(ctx); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + } + } else { + /* Strand 0 */ + ctx->ctxvals_pos = offset; + nv50_graph_construct_gene_dispatch(ctx); + nv50_graph_construct_gene_m2mf(ctx); + nv50_graph_construct_gene_unk34xx(ctx); + nv50_graph_construct_gene_csched(ctx); + nv50_graph_construct_gene_unk1cxx(ctx); + nv50_graph_construct_gene_strmout(ctx); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + + /* Strand 1 */ + ctx->ctxvals_pos = offset + 1; + nv50_graph_construct_gene_unk10xx(ctx); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + + /* Strand 2 */ + ctx->ctxvals_pos = offset + 2; + if (dev_priv->chipset == 0xa0) + nv50_graph_construct_gene_unk14xx(ctx); + nv50_graph_construct_gene_unk24xx(ctx); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + + /* Strand 3 */ + ctx->ctxvals_pos = offset + 3; + nv50_graph_construct_gene_vfetch(ctx); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + + /* Strand 4 */ + ctx->ctxvals_pos = offset + 4; + nv50_graph_construct_gene_ccache(ctx); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + + /* Strand 5 */ + ctx->ctxvals_pos = offset + 5; + nv50_graph_construct_gene_ropm2(ctx); + nv50_graph_construct_gene_ropm1(ctx); + /* per-ROP context */ + for (i = 0; i < 8; i++) + if (units & (1<<(i+16))) + nv50_graph_construct_gene_ropc(ctx); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + + /* Strand 6 */ + ctx->ctxvals_pos = offset + 6; + nv50_graph_construct_gene_zcull(ctx); + nv50_graph_construct_gene_clipid(ctx); + nv50_graph_construct_gene_eng2d(ctx); + if (units & (1 << 0)) + nv50_graph_construct_xfer_tp(ctx); + if (units & (1 << 1)) + nv50_graph_construct_xfer_tp(ctx); + if (units & (1 << 2)) + nv50_graph_construct_xfer_tp(ctx); + if (units & (1 << 3)) + nv50_graph_construct_xfer_tp(ctx); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + + /* Strand 7 */ + ctx->ctxvals_pos = offset + 7; + if (dev_priv->chipset == 0xa0) { + if (units & (1 << 4)) + nv50_graph_construct_xfer_tp(ctx); + if (units & (1 << 5)) + nv50_graph_construct_xfer_tp(ctx); + if (units & (1 << 6)) + nv50_graph_construct_xfer_tp(ctx); + if (units & (1 << 7)) + nv50_graph_construct_xfer_tp(ctx); + if (units & (1 << 8)) + nv50_graph_construct_xfer_tp(ctx); + if (units & (1 << 9)) + nv50_graph_construct_xfer_tp(ctx); + } else { + nv50_graph_construct_gene_unk14xx(ctx); + } + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + } + + ctx->ctxvals_pos = offset + size * 8; + ctx->ctxvals_pos = (ctx->ctxvals_pos+0x3f)&~0x3f; + cp_lsr (ctx, offset); + cp_out (ctx, CP_SET_XFER_POINTER); + cp_lsr (ctx, size); + cp_out (ctx, CP_SEEK_1); + cp_out (ctx, CP_XFER_1); + cp_wait(ctx, XFER, BUSY); +} + +/* + * non-trivial demagiced parts of ctx init go here + */ + +static void +nv50_graph_construct_gene_dispatch(struct nouveau_grctx *ctx) +{ + /* start of strand 0 */ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + /* SEEK */ + if (dev_priv->chipset == 0x50) + xf_emit(ctx, 5, 0); + else if (!IS_NVA3F(dev_priv->chipset)) + xf_emit(ctx, 6, 0); + else + xf_emit(ctx, 4, 0); + /* SEEK */ + /* the PGRAPH's internal FIFO */ + if (dev_priv->chipset == 0x50) + xf_emit(ctx, 8*3, 0); + else + xf_emit(ctx, 0x100*3, 0); + /* and another bonus slot?!? */ + xf_emit(ctx, 3, 0); + /* and YET ANOTHER bonus slot? */ + if (IS_NVA3F(dev_priv->chipset)) + xf_emit(ctx, 3, 0); + /* SEEK */ + /* CTX_SWITCH: caches of gr objects bound to subchannels. 8 values, last used index */ + xf_emit(ctx, 9, 0); + /* SEEK */ + xf_emit(ctx, 9, 0); + /* SEEK */ + xf_emit(ctx, 9, 0); + /* SEEK */ + xf_emit(ctx, 9, 0); + /* SEEK */ + if (dev_priv->chipset < 0x90) + xf_emit(ctx, 4, 0); + /* SEEK */ + xf_emit(ctx, 2, 0); + /* SEEK */ + xf_emit(ctx, 6*2, 0); + xf_emit(ctx, 2, 0); + /* SEEK */ + xf_emit(ctx, 2, 0); + /* SEEK */ + xf_emit(ctx, 6*2, 0); + xf_emit(ctx, 2, 0); + /* SEEK */ + if (dev_priv->chipset == 0x50) + xf_emit(ctx, 0x1c, 0); + else if (dev_priv->chipset < 0xa0) + xf_emit(ctx, 0x1e, 0); + else + xf_emit(ctx, 0x22, 0); + /* SEEK */ + xf_emit(ctx, 0x15, 0); +} + +static void +nv50_graph_construct_gene_m2mf(struct nouveau_grctx *ctx) +{ + /* Strand 0, right after dispatch */ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + int smallm2mf = 0; + if (dev_priv->chipset < 0x92 || dev_priv->chipset == 0x98) + smallm2mf = 1; + /* SEEK */ + xf_emit (ctx, 1, 0); /* DMA_NOTIFY instance >> 4 */ + xf_emit (ctx, 1, 0); /* DMA_BUFFER_IN instance >> 4 */ + xf_emit (ctx, 1, 0); /* DMA_BUFFER_OUT instance >> 4 */ + xf_emit (ctx, 1, 0); /* OFFSET_IN */ + xf_emit (ctx, 1, 0); /* OFFSET_OUT */ + xf_emit (ctx, 1, 0); /* PITCH_IN */ + xf_emit (ctx, 1, 0); /* PITCH_OUT */ + xf_emit (ctx, 1, 0); /* LINE_LENGTH */ + xf_emit (ctx, 1, 0); /* LINE_COUNT */ + xf_emit (ctx, 1, 0x21); /* FORMAT: bits 0-4 INPUT_INC, bits 5-9 OUTPUT_INC */ + xf_emit (ctx, 1, 1); /* LINEAR_IN */ + xf_emit (ctx, 1, 0x2); /* TILING_MODE_IN: bits 0-2 y tiling, bits 3-5 z tiling */ + xf_emit (ctx, 1, 0x100); /* TILING_PITCH_IN */ + xf_emit (ctx, 1, 0x100); /* TILING_HEIGHT_IN */ + xf_emit (ctx, 1, 1); /* TILING_DEPTH_IN */ + xf_emit (ctx, 1, 0); /* TILING_POSITION_IN_Z */ + xf_emit (ctx, 1, 0); /* TILING_POSITION_IN */ + xf_emit (ctx, 1, 1); /* LINEAR_OUT */ + xf_emit (ctx, 1, 0x2); /* TILING_MODE_OUT: bits 0-2 y tiling, bits 3-5 z tiling */ + xf_emit (ctx, 1, 0x100); /* TILING_PITCH_OUT */ + xf_emit (ctx, 1, 0x100); /* TILING_HEIGHT_OUT */ + xf_emit (ctx, 1, 1); /* TILING_DEPTH_OUT */ + xf_emit (ctx, 1, 0); /* TILING_POSITION_OUT_Z */ + xf_emit (ctx, 1, 0); /* TILING_POSITION_OUT */ + xf_emit (ctx, 1, 0); /* OFFSET_IN_HIGH */ + xf_emit (ctx, 1, 0); /* OFFSET_OUT_HIGH */ + /* SEEK */ + if (smallm2mf) + xf_emit(ctx, 0x40, 0); /* 20 * ffffffff, 3ffff */ + else + xf_emit(ctx, 0x100, 0); /* 80 * ffffffff, 3ffff */ + xf_emit(ctx, 4, 0); /* 1f/7f, 0, 1f/7f, 0 [1f for smallm2mf, 7f otherwise] */ + /* SEEK */ + if (smallm2mf) + xf_emit(ctx, 0x400, 0); /* ffffffff */ + else + xf_emit(ctx, 0x800, 0); /* ffffffff */ + xf_emit(ctx, 4, 0); /* ff/1ff, 0, 0, 0 [ff for smallm2mf, 1ff otherwise] */ + /* SEEK */ + xf_emit(ctx, 0x40, 0); /* 20 * bits ffffffff, 3ffff */ + xf_emit(ctx, 0x6, 0); /* 1f, 0, 1f, 0, 1f, 0 */ +} + +static void +nv50_graph_construct_gene_ccache(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + xf_emit(ctx, 2, 0); /* RO */ + xf_emit(ctx, 0x800, 0); /* ffffffff */ + switch (dev_priv->chipset) { + case 0x50: + case 0x92: + case 0xa0: + xf_emit(ctx, 0x2b, 0); + break; + case 0x84: + xf_emit(ctx, 0x29, 0); + break; + case 0x94: + case 0x96: + case 0xa3: + xf_emit(ctx, 0x27, 0); + break; + case 0x86: + case 0x98: + case 0xa5: + case 0xa8: + case 0xaa: + case 0xac: + case 0xaf: + xf_emit(ctx, 0x25, 0); + break; + } + /* CB bindings, 0x80 of them. first word is address >> 8, second is + * size >> 4 | valid << 24 */ + xf_emit(ctx, 0x100, 0); /* ffffffff CB_DEF */ + xf_emit(ctx, 1, 0); /* 0000007f CB_ADDR_BUFFER */ + xf_emit(ctx, 1, 0); /* 0 */ + xf_emit(ctx, 0x30, 0); /* ff SET_PROGRAM_CB */ + xf_emit(ctx, 1, 0); /* 3f last SET_PROGRAM_CB */ + xf_emit(ctx, 4, 0); /* RO */ + xf_emit(ctx, 0x100, 0); /* ffffffff */ + xf_emit(ctx, 8, 0); /* 1f, 0, 0, ... */ + xf_emit(ctx, 8, 0); /* ffffffff */ + xf_emit(ctx, 4, 0); /* ffffffff */ + xf_emit(ctx, 1, 0); /* 3 */ + xf_emit(ctx, 1, 0); /* ffffffff */ + xf_emit(ctx, 1, 0); /* 0000ffff DMA_CODE_CB */ + xf_emit(ctx, 1, 0); /* 0000ffff DMA_TIC */ + xf_emit(ctx, 1, 0); /* 0000ffff DMA_TSC */ + xf_emit(ctx, 1, 0); /* 00000001 LINKED_TSC */ + xf_emit(ctx, 1, 0); /* 000000ff TIC_ADDRESS_HIGH */ + xf_emit(ctx, 1, 0); /* ffffffff TIC_ADDRESS_LOW */ + xf_emit(ctx, 1, 0x3fffff); /* 003fffff TIC_LIMIT */ + xf_emit(ctx, 1, 0); /* 000000ff TSC_ADDRESS_HIGH */ + xf_emit(ctx, 1, 0); /* ffffffff TSC_ADDRESS_LOW */ + xf_emit(ctx, 1, 0x1fff); /* 000fffff TSC_LIMIT */ + xf_emit(ctx, 1, 0); /* 000000ff VP_ADDRESS_HIGH */ + xf_emit(ctx, 1, 0); /* ffffffff VP_ADDRESS_LOW */ + xf_emit(ctx, 1, 0); /* 00ffffff VP_START_ID */ + xf_emit(ctx, 1, 0); /* 000000ff CB_DEF_ADDRESS_HIGH */ + xf_emit(ctx, 1, 0); /* ffffffff CB_DEF_ADDRESS_LOW */ + xf_emit(ctx, 1, 0); /* 00000001 GP_ENABLE */ + xf_emit(ctx, 1, 0); /* 000000ff GP_ADDRESS_HIGH */ + xf_emit(ctx, 1, 0); /* ffffffff GP_ADDRESS_LOW */ + xf_emit(ctx, 1, 0); /* 00ffffff GP_START_ID */ + xf_emit(ctx, 1, 0); /* 000000ff FP_ADDRESS_HIGH */ + xf_emit(ctx, 1, 0); /* ffffffff FP_ADDRESS_LOW */ + xf_emit(ctx, 1, 0); /* 00ffffff FP_START_ID */ +} + +static void +nv50_graph_construct_gene_unk10xx(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + int i; + /* end of area 2 on pre-NVA0, area 1 on NVAx */ + xf_emit(ctx, 1, 4); /* 000000ff GP_RESULT_MAP_SIZE */ + xf_emit(ctx, 1, 4); /* 0000007f VP_RESULT_MAP_SIZE */ + xf_emit(ctx, 1, 0); /* 00000001 GP_ENABLE */ + xf_emit(ctx, 1, 0x80); /* 0000ffff GP_VERTEX_OUTPUT_COUNT */ + xf_emit(ctx, 1, 4); /* 000000ff GP_REG_ALLOC_RESULT */ + xf_emit(ctx, 1, 0x80c14); /* 01ffffff SEMANTIC_COLOR */ + xf_emit(ctx, 1, 0); /* 00000001 VERTEX_TWO_SIDE_ENABLE */ + if (dev_priv->chipset == 0x50) + xf_emit(ctx, 1, 0x3ff); + else + xf_emit(ctx, 1, 0x7ff); /* 000007ff */ + xf_emit(ctx, 1, 0); /* 111/113 */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A30 */ + for (i = 0; i < 8; i++) { + switch (dev_priv->chipset) { + case 0x50: + case 0x86: + case 0x98: + case 0xaa: + case 0xac: + xf_emit(ctx, 0xa0, 0); /* ffffffff */ + break; + case 0x84: + case 0x92: + case 0x94: + case 0x96: + xf_emit(ctx, 0x120, 0); + break; + case 0xa5: + case 0xa8: + xf_emit(ctx, 0x100, 0); /* ffffffff */ + break; + case 0xa0: + case 0xa3: + case 0xaf: + xf_emit(ctx, 0x400, 0); /* ffffffff */ + break; + } + xf_emit(ctx, 4, 0); /* 3f, 0, 0, 0 */ + xf_emit(ctx, 4, 0); /* ffffffff */ + } + xf_emit(ctx, 1, 4); /* 000000ff GP_RESULT_MAP_SIZE */ + xf_emit(ctx, 1, 4); /* 0000007f VP_RESULT_MAP_SIZE */ + xf_emit(ctx, 1, 0); /* 00000001 GP_ENABLE */ + xf_emit(ctx, 1, 0x80); /* 0000ffff GP_VERTEX_OUTPUT_COUNT */ + xf_emit(ctx, 1, 4); /* 000000ff GP_REG_ALLOC_TEMP */ + xf_emit(ctx, 1, 1); /* 00000001 RASTERIZE_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK1900 */ + xf_emit(ctx, 1, 0x27); /* 000000ff UNK0FD4 */ + xf_emit(ctx, 1, 0); /* 0001ffff GP_BUILTIN_RESULT_EN */ + xf_emit(ctx, 1, 0x26); /* 000000ff SEMANTIC_LAYER */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A30 */ +} + +static void +nv50_graph_construct_gene_unk34xx(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + /* end of area 2 on pre-NVA0, area 1 on NVAx */ + xf_emit(ctx, 1, 0); /* 00000001 VIEWPORT_CLIP_RECTS_EN */ + xf_emit(ctx, 1, 0); /* 00000003 VIEWPORT_CLIP_MODE */ + xf_emit(ctx, 0x10, 0x04000000); /* 07ffffff VIEWPORT_CLIP_HORIZ*8, VIEWPORT_CLIP_VERT*8 */ + xf_emit(ctx, 1, 0); /* 00000001 POLYGON_STIPPLE_ENABLE */ + xf_emit(ctx, 0x20, 0); /* ffffffff POLYGON_STIPPLE */ + xf_emit(ctx, 2, 0); /* 00007fff WINDOW_OFFSET_XY */ + xf_emit(ctx, 1, 0); /* ffff0ff3 */ + xf_emit(ctx, 1, 0x04e3bfdf); /* ffffffff UNK0D64 */ + xf_emit(ctx, 1, 0x04e3bfdf); /* ffffffff UNK0DF4 */ + xf_emit(ctx, 1, 0); /* 00000003 WINDOW_ORIGIN */ + xf_emit(ctx, 1, 0); /* 00000007 */ + xf_emit(ctx, 1, 0x1fe21); /* 0001ffff tesla UNK0FAC */ + if (dev_priv->chipset >= 0xa0) + xf_emit(ctx, 1, 0x0fac6881); + if (IS_NVA3F(dev_priv->chipset)) { + xf_emit(ctx, 1, 1); + xf_emit(ctx, 3, 0); + } +} + +static void +nv50_graph_construct_gene_unk14xx(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + /* middle of area 2 on pre-NVA0, beginning of area 2 on NVA0, area 7 on >NVA0 */ + if (dev_priv->chipset != 0x50) { + xf_emit(ctx, 5, 0); /* ffffffff */ + xf_emit(ctx, 1, 0x80c14); /* 01ffffff SEMANTIC_COLOR */ + xf_emit(ctx, 1, 0); /* 00000001 */ + xf_emit(ctx, 1, 0); /* 000003ff */ + xf_emit(ctx, 1, 0x804); /* 00000fff SEMANTIC_CLIP */ + xf_emit(ctx, 1, 0); /* 00000001 */ + xf_emit(ctx, 2, 4); /* 7f, ff */ + xf_emit(ctx, 1, 0x8100c12); /* 1fffffff FP_INTERPOLANT_CTRL */ + } + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A30 */ + xf_emit(ctx, 1, 4); /* 0000007f VP_RESULT_MAP_SIZE */ + xf_emit(ctx, 1, 4); /* 000000ff GP_RESULT_MAP_SIZE */ + xf_emit(ctx, 1, 0); /* 00000001 GP_ENABLE */ + xf_emit(ctx, 1, 0x10); /* 7f/ff VIEW_VOLUME_CLIP_CTRL */ + xf_emit(ctx, 1, 0); /* 000000ff VP_CLIP_DISTANCE_ENABLE */ + if (dev_priv->chipset != 0x50) + xf_emit(ctx, 1, 0); /* 3ff */ + xf_emit(ctx, 1, 0); /* 000000ff tesla UNK1940 */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK0D7C */ + xf_emit(ctx, 1, 0x804); /* 00000fff SEMANTIC_CLIP */ + xf_emit(ctx, 1, 1); /* 00000001 VIEWPORT_TRANSFORM_EN */ + xf_emit(ctx, 1, 0x1a); /* 0000001f POLYGON_MODE */ + if (dev_priv->chipset != 0x50) + xf_emit(ctx, 1, 0x7f); /* 000000ff tesla UNK0FFC */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A30 */ + xf_emit(ctx, 1, 1); /* 00000001 SHADE_MODEL */ + xf_emit(ctx, 1, 0x80c14); /* 01ffffff SEMANTIC_COLOR */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK1900 */ + xf_emit(ctx, 1, 0x8100c12); /* 1fffffff FP_INTERPOLANT_CTRL */ + xf_emit(ctx, 1, 4); /* 0000007f VP_RESULT_MAP_SIZE */ + xf_emit(ctx, 1, 4); /* 000000ff GP_RESULT_MAP_SIZE */ + xf_emit(ctx, 1, 0); /* 00000001 GP_ENABLE */ + xf_emit(ctx, 1, 0x10); /* 7f/ff VIEW_VOLUME_CLIP_CTRL */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK0D7C */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK0F8C */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A30 */ + xf_emit(ctx, 1, 1); /* 00000001 VIEWPORT_TRANSFORM_EN */ + xf_emit(ctx, 1, 0x8100c12); /* 1fffffff FP_INTERPOLANT_CTRL */ + xf_emit(ctx, 4, 0); /* ffffffff NOPERSPECTIVE_BITMAP */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK1900 */ + xf_emit(ctx, 1, 0); /* 0000000f */ + if (dev_priv->chipset == 0x50) + xf_emit(ctx, 1, 0x3ff); /* 000003ff tesla UNK0D68 */ + else + xf_emit(ctx, 1, 0x7ff); /* 000007ff tesla UNK0D68 */ + xf_emit(ctx, 1, 0x80c14); /* 01ffffff SEMANTIC_COLOR */ + xf_emit(ctx, 1, 0); /* 00000001 VERTEX_TWO_SIDE_ENABLE */ + xf_emit(ctx, 0x30, 0); /* ffffffff VIEWPORT_SCALE: X0, Y0, Z0, X1, Y1, ... */ + xf_emit(ctx, 3, 0); /* f, 0, 0 */ + xf_emit(ctx, 3, 0); /* ffffffff last VIEWPORT_SCALE? */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A30 */ + xf_emit(ctx, 1, 1); /* 00000001 VIEWPORT_TRANSFORM_EN */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK1900 */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK1924 */ + xf_emit(ctx, 1, 0x10); /* 000000ff VIEW_VOLUME_CLIP_CTRL */ + xf_emit(ctx, 1, 0); /* 00000001 */ + xf_emit(ctx, 0x30, 0); /* ffffffff VIEWPORT_TRANSLATE */ + xf_emit(ctx, 3, 0); /* f, 0, 0 */ + xf_emit(ctx, 3, 0); /* ffffffff */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A30 */ + xf_emit(ctx, 2, 0x88); /* 000001ff tesla UNK19D8 */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK1924 */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A30 */ + xf_emit(ctx, 1, 4); /* 0000000f CULL_MODE */ + xf_emit(ctx, 2, 0); /* 07ffffff SCREEN_SCISSOR */ + xf_emit(ctx, 2, 0); /* 00007fff WINDOW_OFFSET_XY */ + xf_emit(ctx, 1, 0); /* 00000003 WINDOW_ORIGIN */ + xf_emit(ctx, 0x10, 0); /* 00000001 SCISSOR_ENABLE */ + xf_emit(ctx, 1, 0); /* 0001ffff GP_BUILTIN_RESULT_EN */ + xf_emit(ctx, 1, 0x26); /* 000000ff SEMANTIC_LAYER */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK1900 */ + xf_emit(ctx, 1, 0); /* 0000000f */ + xf_emit(ctx, 1, 0x3f800000); /* ffffffff LINE_WIDTH */ + xf_emit(ctx, 1, 0); /* 00000001 LINE_STIPPLE_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000001 LINE_SMOOTH_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000007 MULTISAMPLE_SAMPLES_LOG2 */ + if (IS_NVA3F(dev_priv->chipset)) + xf_emit(ctx, 1, 0); /* 00000001 */ + xf_emit(ctx, 1, 0x1a); /* 0000001f POLYGON_MODE */ + xf_emit(ctx, 1, 0x10); /* 000000ff VIEW_VOLUME_CLIP_CTRL */ + if (dev_priv->chipset != 0x50) { + xf_emit(ctx, 1, 0); /* ffffffff */ + xf_emit(ctx, 1, 0); /* 00000001 */ + xf_emit(ctx, 1, 0); /* 000003ff */ + } + xf_emit(ctx, 0x20, 0); /* 10xbits ffffffff, 3fffff. SCISSOR_* */ + xf_emit(ctx, 1, 0); /* f */ + xf_emit(ctx, 1, 0); /* 0? */ + xf_emit(ctx, 1, 0); /* ffffffff */ + xf_emit(ctx, 1, 0); /* 003fffff */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A30 */ + xf_emit(ctx, 1, 0x52); /* 000001ff SEMANTIC_PTSZ */ + xf_emit(ctx, 1, 0); /* 0001ffff GP_BUILTIN_RESULT_EN */ + xf_emit(ctx, 1, 0x26); /* 000000ff SEMANTIC_LAYER */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK1900 */ + xf_emit(ctx, 1, 4); /* 0000007f VP_RESULT_MAP_SIZE */ + xf_emit(ctx, 1, 4); /* 000000ff GP_RESULT_MAP_SIZE */ + xf_emit(ctx, 1, 0); /* 00000001 GP_ENABLE */ + xf_emit(ctx, 1, 0x1a); /* 0000001f POLYGON_MODE */ + xf_emit(ctx, 1, 0); /* 00000001 LINE_SMOOTH_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000001 LINE_STIPPLE_ENABLE */ + xf_emit(ctx, 1, 0x00ffff00); /* 00ffffff LINE_STIPPLE_PATTERN */ + xf_emit(ctx, 1, 0); /* 0000000f */ +} + +static void +nv50_graph_construct_gene_zcull(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + /* end of strand 0 on pre-NVA0, beginning of strand 6 on NVAx */ + /* SEEK */ + xf_emit(ctx, 1, 0x3f); /* 0000003f UNK1590 */ + xf_emit(ctx, 1, 0); /* 00000001 ALPHA_TEST_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000007 MULTISAMPLE_SAMPLES_LOG2 */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK1534 */ + xf_emit(ctx, 1, 0); /* 00000007 STENCIL_BACK_FUNC_FUNC */ + xf_emit(ctx, 1, 0); /* 000000ff STENCIL_BACK_FUNC_MASK */ + xf_emit(ctx, 1, 0); /* 000000ff STENCIL_BACK_FUNC_REF */ + xf_emit(ctx, 1, 0); /* 000000ff STENCIL_BACK_MASK */ + xf_emit(ctx, 3, 0); /* 00000007 STENCIL_BACK_OP_FAIL, ZFAIL, ZPASS */ + xf_emit(ctx, 1, 2); /* 00000003 tesla UNK143C */ + xf_emit(ctx, 2, 0x04000000); /* 07ffffff tesla UNK0D6C */ + xf_emit(ctx, 1, 0); /* ffff0ff3 */ + xf_emit(ctx, 1, 0); /* 00000001 CLIPID_ENABLE */ + xf_emit(ctx, 2, 0); /* ffffffff DEPTH_BOUNDS */ + xf_emit(ctx, 1, 0); /* 00000001 */ + xf_emit(ctx, 1, 0); /* 00000007 DEPTH_TEST_FUNC */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_TEST_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_WRITE_ENABLE */ + xf_emit(ctx, 1, 4); /* 0000000f CULL_MODE */ + xf_emit(ctx, 1, 0); /* 0000ffff */ + xf_emit(ctx, 1, 0); /* 00000001 UNK0FB0 */ + xf_emit(ctx, 1, 0); /* 00000001 POLYGON_STIPPLE_ENABLE */ + xf_emit(ctx, 1, 4); /* 00000007 FP_CONTROL */ + xf_emit(ctx, 1, 0); /* ffffffff */ + xf_emit(ctx, 1, 0); /* 0001ffff GP_BUILTIN_RESULT_EN */ + xf_emit(ctx, 1, 0); /* 000000ff CLEAR_STENCIL */ + xf_emit(ctx, 1, 0); /* 00000007 STENCIL_FRONT_FUNC_FUNC */ + xf_emit(ctx, 1, 0); /* 000000ff STENCIL_FRONT_FUNC_MASK */ + xf_emit(ctx, 1, 0); /* 000000ff STENCIL_FRONT_FUNC_REF */ + xf_emit(ctx, 1, 0); /* 000000ff STENCIL_FRONT_MASK */ + xf_emit(ctx, 3, 0); /* 00000007 STENCIL_FRONT_OP_FAIL, ZFAIL, ZPASS */ + xf_emit(ctx, 1, 0); /* 00000001 STENCIL_FRONT_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000001 STENCIL_BACK_ENABLE */ + xf_emit(ctx, 1, 0); /* ffffffff CLEAR_DEPTH */ + xf_emit(ctx, 1, 0); /* 00000007 */ + if (dev_priv->chipset != 0x50) + xf_emit(ctx, 1, 0); /* 00000003 tesla UNK1108 */ + xf_emit(ctx, 1, 0); /* 00000001 SAMPLECNT_ENABLE */ + xf_emit(ctx, 1, 0); /* 0000000f ZETA_FORMAT */ + xf_emit(ctx, 1, 1); /* 00000001 ZETA_ENABLE */ + xf_emit(ctx, 1, 0x1001); /* 00001fff ZETA_ARRAY_MODE */ + /* SEEK */ + xf_emit(ctx, 4, 0xffff); /* 0000ffff MSAA_MASK */ + xf_emit(ctx, 0x10, 0); /* 00000001 SCISSOR_ENABLE */ + xf_emit(ctx, 0x10, 0); /* ffffffff DEPTH_RANGE_NEAR */ + xf_emit(ctx, 0x10, 0x3f800000); /* ffffffff DEPTH_RANGE_FAR */ + xf_emit(ctx, 1, 0x10); /* 7f/ff/3ff VIEW_VOLUME_CLIP_CTRL */ + xf_emit(ctx, 1, 0); /* 00000001 VIEWPORT_CLIP_RECTS_EN */ + xf_emit(ctx, 1, 3); /* 00000003 FP_CTRL_UNK196C */ + xf_emit(ctx, 1, 0); /* 00000003 tesla UNK1968 */ + if (dev_priv->chipset != 0x50) + xf_emit(ctx, 1, 0); /* 0fffffff tesla UNK1104 */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK151C */ +} + +static void +nv50_graph_construct_gene_clipid(struct nouveau_grctx *ctx) +{ + /* middle of strand 0 on pre-NVA0 [after 24xx], middle of area 6 on NVAx */ + /* SEEK */ + xf_emit(ctx, 1, 0); /* 00000007 UNK0FB4 */ + /* SEEK */ + xf_emit(ctx, 4, 0); /* 07ffffff CLIPID_REGION_HORIZ */ + xf_emit(ctx, 4, 0); /* 07ffffff CLIPID_REGION_VERT */ + xf_emit(ctx, 2, 0); /* 07ffffff SCREEN_SCISSOR */ + xf_emit(ctx, 2, 0x04000000); /* 07ffffff UNK1508 */ + xf_emit(ctx, 1, 0); /* 00000001 CLIPID_ENABLE */ + xf_emit(ctx, 1, 0x80); /* 00003fff CLIPID_WIDTH */ + xf_emit(ctx, 1, 0); /* 000000ff CLIPID_ID */ + xf_emit(ctx, 1, 0); /* 000000ff CLIPID_ADDRESS_HIGH */ + xf_emit(ctx, 1, 0); /* ffffffff CLIPID_ADDRESS_LOW */ + xf_emit(ctx, 1, 0x80); /* 00003fff CLIPID_HEIGHT */ + xf_emit(ctx, 1, 0); /* 0000ffff DMA_CLIPID */ +} + +static void +nv50_graph_construct_gene_unk24xx(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + int i; + /* middle of strand 0 on pre-NVA0 [after m2mf], end of strand 2 on NVAx */ + /* SEEK */ + xf_emit(ctx, 0x33, 0); + /* SEEK */ + xf_emit(ctx, 2, 0); + /* SEEK */ + xf_emit(ctx, 1, 0); /* 00000001 GP_ENABLE */ + xf_emit(ctx, 1, 4); /* 0000007f VP_RESULT_MAP_SIZE */ + xf_emit(ctx, 1, 4); /* 000000ff GP_RESULT_MAP_SIZE */ + /* SEEK */ + if (IS_NVA3F(dev_priv->chipset)) { + xf_emit(ctx, 4, 0); /* RO */ + xf_emit(ctx, 0xe10, 0); /* 190 * 9: 8*ffffffff, 7ff */ + xf_emit(ctx, 1, 0); /* 1ff */ + xf_emit(ctx, 8, 0); /* 0? */ + xf_emit(ctx, 9, 0); /* ffffffff, 7ff */ + + xf_emit(ctx, 4, 0); /* RO */ + xf_emit(ctx, 0xe10, 0); /* 190 * 9: 8*ffffffff, 7ff */ + xf_emit(ctx, 1, 0); /* 1ff */ + xf_emit(ctx, 8, 0); /* 0? */ + xf_emit(ctx, 9, 0); /* ffffffff, 7ff */ + } else { + xf_emit(ctx, 0xc, 0); /* RO */ + /* SEEK */ + xf_emit(ctx, 0xe10, 0); /* 190 * 9: 8*ffffffff, 7ff */ + xf_emit(ctx, 1, 0); /* 1ff */ + xf_emit(ctx, 8, 0); /* 0? */ + + /* SEEK */ + xf_emit(ctx, 0xc, 0); /* RO */ + /* SEEK */ + xf_emit(ctx, 0xe10, 0); /* 190 * 9: 8*ffffffff, 7ff */ + xf_emit(ctx, 1, 0); /* 1ff */ + xf_emit(ctx, 8, 0); /* 0? */ + } + /* SEEK */ + xf_emit(ctx, 1, 0); /* 00000001 GP_ENABLE */ + xf_emit(ctx, 1, 4); /* 000000ff GP_RESULT_MAP_SIZE */ + xf_emit(ctx, 1, 4); /* 0000007f VP_RESULT_MAP_SIZE */ + xf_emit(ctx, 1, 0x8100c12); /* 1fffffff FP_INTERPOLANT_CTRL */ + if (dev_priv->chipset != 0x50) + xf_emit(ctx, 1, 3); /* 00000003 tesla UNK1100 */ + /* SEEK */ + xf_emit(ctx, 1, 0); /* 00000001 GP_ENABLE */ + xf_emit(ctx, 1, 0x8100c12); /* 1fffffff FP_INTERPOLANT_CTRL */ + xf_emit(ctx, 1, 0); /* 0000000f VP_GP_BUILTIN_ATTR_EN */ + xf_emit(ctx, 1, 0x80c14); /* 01ffffff SEMANTIC_COLOR */ + xf_emit(ctx, 1, 1); /* 00000001 */ + /* SEEK */ + if (dev_priv->chipset >= 0xa0) + xf_emit(ctx, 2, 4); /* 000000ff */ + xf_emit(ctx, 1, 0x80c14); /* 01ffffff SEMANTIC_COLOR */ + xf_emit(ctx, 1, 0); /* 00000001 VERTEX_TWO_SIDE_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000001 POINT_SPRITE_ENABLE */ + xf_emit(ctx, 1, 0x8100c12); /* 1fffffff FP_INTERPOLANT_CTRL */ + xf_emit(ctx, 1, 0x27); /* 000000ff SEMANTIC_PRIM_ID */ + xf_emit(ctx, 1, 0); /* 00000001 GP_ENABLE */ + xf_emit(ctx, 1, 0); /* 0000000f */ + xf_emit(ctx, 1, 1); /* 00000001 */ + for (i = 0; i < 10; i++) { + /* SEEK */ + xf_emit(ctx, 0x40, 0); /* ffffffff */ + xf_emit(ctx, 0x10, 0); /* 3, 0, 0.... */ + xf_emit(ctx, 0x10, 0); /* ffffffff */ + } + /* SEEK */ + xf_emit(ctx, 1, 0); /* 00000001 POINT_SPRITE_CTRL */ + xf_emit(ctx, 1, 1); /* 00000001 */ + xf_emit(ctx, 1, 0); /* ffffffff */ + xf_emit(ctx, 4, 0); /* ffffffff NOPERSPECTIVE_BITMAP */ + xf_emit(ctx, 0x10, 0); /* 00ffffff POINT_COORD_REPLACE_MAP */ + xf_emit(ctx, 1, 0); /* 00000003 WINDOW_ORIGIN */ + xf_emit(ctx, 1, 0x8100c12); /* 1fffffff FP_INTERPOLANT_CTRL */ + if (dev_priv->chipset != 0x50) + xf_emit(ctx, 1, 0); /* 000003ff */ +} + +static void +nv50_graph_construct_gene_vfetch(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + int acnt = 0x10, rep, i; + /* beginning of strand 1 on pre-NVA0, strand 3 on NVAx */ + if (IS_NVA3F(dev_priv->chipset)) + acnt = 0x20; + /* SEEK */ + if (dev_priv->chipset >= 0xa0) { + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK13A4 */ + xf_emit(ctx, 1, 1); /* 00000fff tesla UNK1318 */ + } + xf_emit(ctx, 1, 0); /* ffffffff VERTEX_BUFFER_FIRST */ + xf_emit(ctx, 1, 0); /* 00000001 PRIMITIVE_RESTART_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000001 UNK0DE8 */ + xf_emit(ctx, 1, 0); /* ffffffff PRIMITIVE_RESTART_INDEX */ + xf_emit(ctx, 1, 0xf); /* ffffffff VP_ATTR_EN */ + xf_emit(ctx, (acnt/8)-1, 0); /* ffffffff VP_ATTR_EN */ + xf_emit(ctx, acnt/8, 0); /* ffffffff VTX_ATR_MASK_UNK0DD0 */ + xf_emit(ctx, 1, 0); /* 0000000f VP_GP_BUILTIN_ATTR_EN */ + xf_emit(ctx, 1, 0x20); /* 0000ffff tesla UNK129C */ + xf_emit(ctx, 1, 0); /* 000000ff turing UNK370??? */ + xf_emit(ctx, 1, 0); /* 0000ffff turing USER_PARAM_COUNT */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A30 */ + /* SEEK */ + if (IS_NVA3F(dev_priv->chipset)) + xf_emit(ctx, 0xb, 0); /* RO */ + else if (dev_priv->chipset >= 0xa0) + xf_emit(ctx, 0x9, 0); /* RO */ + else + xf_emit(ctx, 0x8, 0); /* RO */ + /* SEEK */ + xf_emit(ctx, 1, 0); /* 00000001 EDGE_FLAG */ + xf_emit(ctx, 1, 0); /* 00000001 PROVOKING_VERTEX_LAST */ + xf_emit(ctx, 1, 0); /* 00000001 GP_ENABLE */ + xf_emit(ctx, 1, 0x1a); /* 0000001f POLYGON_MODE */ + /* SEEK */ + xf_emit(ctx, 0xc, 0); /* RO */ + /* SEEK */ + xf_emit(ctx, 1, 0); /* 7f/ff */ + xf_emit(ctx, 1, 4); /* 7f/ff VP_REG_ALLOC_RESULT */ + xf_emit(ctx, 1, 4); /* 7f/ff VP_RESULT_MAP_SIZE */ + xf_emit(ctx, 1, 0); /* 0000000f VP_GP_BUILTIN_ATTR_EN */ + xf_emit(ctx, 1, 4); /* 000001ff UNK1A28 */ + xf_emit(ctx, 1, 8); /* 000001ff UNK0DF0 */ + xf_emit(ctx, 1, 0); /* 00000001 GP_ENABLE */ + if (dev_priv->chipset == 0x50) + xf_emit(ctx, 1, 0x3ff); /* 3ff tesla UNK0D68 */ + else + xf_emit(ctx, 1, 0x7ff); /* 7ff tesla UNK0D68 */ + if (dev_priv->chipset == 0xa8) + xf_emit(ctx, 1, 0x1e00); /* 7fff */ + /* SEEK */ + xf_emit(ctx, 0xc, 0); /* RO or close */ + /* SEEK */ + xf_emit(ctx, 1, 0xf); /* ffffffff VP_ATTR_EN */ + xf_emit(ctx, (acnt/8)-1, 0); /* ffffffff VP_ATTR_EN */ + xf_emit(ctx, 1, 0); /* 0000000f VP_GP_BUILTIN_ATTR_EN */ + if (dev_priv->chipset > 0x50 && dev_priv->chipset < 0xa0) + xf_emit(ctx, 2, 0); /* ffffffff */ + else + xf_emit(ctx, 1, 0); /* ffffffff */ + xf_emit(ctx, 1, 0); /* 00000003 tesla UNK0FD8 */ + /* SEEK */ + if (IS_NVA3F(dev_priv->chipset)) { + xf_emit(ctx, 0x10, 0); /* 0? */ + xf_emit(ctx, 2, 0); /* weird... */ + xf_emit(ctx, 2, 0); /* RO */ + } else { + xf_emit(ctx, 8, 0); /* 0? */ + xf_emit(ctx, 1, 0); /* weird... */ + xf_emit(ctx, 2, 0); /* RO */ + } + /* SEEK */ + xf_emit(ctx, 1, 0); /* ffffffff VB_ELEMENT_BASE */ + xf_emit(ctx, 1, 0); /* ffffffff UNK1438 */ + xf_emit(ctx, acnt, 0); /* 1 tesla UNK1000 */ + if (dev_priv->chipset >= 0xa0) + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1118? */ + /* SEEK */ + xf_emit(ctx, acnt, 0); /* ffffffff VERTEX_ARRAY_UNK90C */ + xf_emit(ctx, 1, 0); /* f/1f */ + /* SEEK */ + xf_emit(ctx, acnt, 0); /* ffffffff VERTEX_ARRAY_UNK90C */ + xf_emit(ctx, 1, 0); /* f/1f */ + /* SEEK */ + xf_emit(ctx, acnt, 0); /* RO */ + xf_emit(ctx, 2, 0); /* RO */ + /* SEEK */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK111C? */ + xf_emit(ctx, 1, 0); /* RO */ + /* SEEK */ + xf_emit(ctx, 1, 0); /* 000000ff UNK15F4_ADDRESS_HIGH */ + xf_emit(ctx, 1, 0); /* ffffffff UNK15F4_ADDRESS_LOW */ + xf_emit(ctx, 1, 0); /* 000000ff UNK0F84_ADDRESS_HIGH */ + xf_emit(ctx, 1, 0); /* ffffffff UNK0F84_ADDRESS_LOW */ + /* SEEK */ + xf_emit(ctx, acnt, 0); /* 00003fff VERTEX_ARRAY_ATTRIB_OFFSET */ + xf_emit(ctx, 3, 0); /* f/1f */ + /* SEEK */ + xf_emit(ctx, acnt, 0); /* 00000fff VERTEX_ARRAY_STRIDE */ + xf_emit(ctx, 3, 0); /* f/1f */ + /* SEEK */ + xf_emit(ctx, acnt, 0); /* ffffffff VERTEX_ARRAY_LOW */ + xf_emit(ctx, 3, 0); /* f/1f */ + /* SEEK */ + xf_emit(ctx, acnt, 0); /* 000000ff VERTEX_ARRAY_HIGH */ + xf_emit(ctx, 3, 0); /* f/1f */ + /* SEEK */ + xf_emit(ctx, acnt, 0); /* ffffffff VERTEX_LIMIT_LOW */ + xf_emit(ctx, 3, 0); /* f/1f */ + /* SEEK */ + xf_emit(ctx, acnt, 0); /* 000000ff VERTEX_LIMIT_HIGH */ + xf_emit(ctx, 3, 0); /* f/1f */ + /* SEEK */ + if (IS_NVA3F(dev_priv->chipset)) { + xf_emit(ctx, acnt, 0); /* f */ + xf_emit(ctx, 3, 0); /* f/1f */ + } + /* SEEK */ + if (IS_NVA3F(dev_priv->chipset)) + xf_emit(ctx, 2, 0); /* RO */ + else + xf_emit(ctx, 5, 0); /* RO */ + /* SEEK */ + xf_emit(ctx, 1, 0); /* ffff DMA_VTXBUF */ + /* SEEK */ + if (dev_priv->chipset < 0xa0) { + xf_emit(ctx, 0x41, 0); /* RO */ + /* SEEK */ + xf_emit(ctx, 0x11, 0); /* RO */ + } else if (!IS_NVA3F(dev_priv->chipset)) + xf_emit(ctx, 0x50, 0); /* RO */ + else + xf_emit(ctx, 0x58, 0); /* RO */ + /* SEEK */ + xf_emit(ctx, 1, 0xf); /* ffffffff VP_ATTR_EN */ + xf_emit(ctx, (acnt/8)-1, 0); /* ffffffff VP_ATTR_EN */ + xf_emit(ctx, 1, 1); /* 1 UNK0DEC */ + /* SEEK */ + xf_emit(ctx, acnt*4, 0); /* ffffffff VTX_ATTR */ + xf_emit(ctx, 4, 0); /* f/1f, 0, 0, 0 */ + /* SEEK */ + if (IS_NVA3F(dev_priv->chipset)) + xf_emit(ctx, 0x1d, 0); /* RO */ + else + xf_emit(ctx, 0x16, 0); /* RO */ + /* SEEK */ + xf_emit(ctx, 1, 0xf); /* ffffffff VP_ATTR_EN */ + xf_emit(ctx, (acnt/8)-1, 0); /* ffffffff VP_ATTR_EN */ + /* SEEK */ + if (dev_priv->chipset < 0xa0) + xf_emit(ctx, 8, 0); /* RO */ + else if (IS_NVA3F(dev_priv->chipset)) + xf_emit(ctx, 0xc, 0); /* RO */ + else + xf_emit(ctx, 7, 0); /* RO */ + /* SEEK */ + xf_emit(ctx, 0xa, 0); /* RO */ + if (dev_priv->chipset == 0xa0) + rep = 0xc; + else + rep = 4; + for (i = 0; i < rep; i++) { + /* SEEK */ + if (IS_NVA3F(dev_priv->chipset)) + xf_emit(ctx, 0x20, 0); /* ffffffff */ + xf_emit(ctx, 0x200, 0); /* ffffffff */ + xf_emit(ctx, 4, 0); /* 7f/ff, 0, 0, 0 */ + xf_emit(ctx, 4, 0); /* ffffffff */ + } + /* SEEK */ + xf_emit(ctx, 1, 0); /* 113/111 */ + xf_emit(ctx, 1, 0xf); /* ffffffff VP_ATTR_EN */ + xf_emit(ctx, (acnt/8)-1, 0); /* ffffffff VP_ATTR_EN */ + xf_emit(ctx, acnt/8, 0); /* ffffffff VTX_ATTR_MASK_UNK0DD0 */ + xf_emit(ctx, 1, 0); /* 0000000f VP_GP_BUILTIN_ATTR_EN */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A30 */ + /* SEEK */ + if (IS_NVA3F(dev_priv->chipset)) + xf_emit(ctx, 7, 0); /* weird... */ + else + xf_emit(ctx, 5, 0); /* weird... */ +} + +static void +nv50_graph_construct_gene_eng2d(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + /* middle of strand 1 on pre-NVA0 [after vfetch], middle of strand 6 on NVAx */ + /* SEEK */ + xf_emit(ctx, 2, 0); /* 0001ffff CLIP_X, CLIP_Y */ + xf_emit(ctx, 2, 0); /* 0000ffff CLIP_W, CLIP_H */ + xf_emit(ctx, 1, 0); /* 00000001 CLIP_ENABLE */ + if (dev_priv->chipset < 0xa0) { + /* this is useless on everything but the original NV50, + * guess they forgot to nuke it. Or just didn't bother. */ + xf_emit(ctx, 2, 0); /* 0000ffff IFC_CLIP_X, Y */ + xf_emit(ctx, 2, 1); /* 0000ffff IFC_CLIP_W, H */ + xf_emit(ctx, 1, 0); /* 00000001 IFC_CLIP_ENABLE */ + } + xf_emit(ctx, 1, 1); /* 00000001 DST_LINEAR */ + xf_emit(ctx, 1, 0x100); /* 0001ffff DST_WIDTH */ + xf_emit(ctx, 1, 0x100); /* 0001ffff DST_HEIGHT */ + xf_emit(ctx, 1, 0x11); /* 3f[NV50]/7f[NV84+] DST_FORMAT */ + xf_emit(ctx, 1, 0); /* 0001ffff DRAW_POINT_X */ + xf_emit(ctx, 1, 8); /* 0000000f DRAW_UNK58C */ + xf_emit(ctx, 1, 0); /* 000fffff SIFC_DST_X_FRACT */ + xf_emit(ctx, 1, 0); /* 0001ffff SIFC_DST_X_INT */ + xf_emit(ctx, 1, 0); /* 000fffff SIFC_DST_Y_FRACT */ + xf_emit(ctx, 1, 0); /* 0001ffff SIFC_DST_Y_INT */ + xf_emit(ctx, 1, 0); /* 000fffff SIFC_DX_DU_FRACT */ + xf_emit(ctx, 1, 1); /* 0001ffff SIFC_DX_DU_INT */ + xf_emit(ctx, 1, 0); /* 000fffff SIFC_DY_DV_FRACT */ + xf_emit(ctx, 1, 1); /* 0001ffff SIFC_DY_DV_INT */ + xf_emit(ctx, 1, 1); /* 0000ffff SIFC_WIDTH */ + xf_emit(ctx, 1, 1); /* 0000ffff SIFC_HEIGHT */ + xf_emit(ctx, 1, 0xcf); /* 000000ff SIFC_FORMAT */ + xf_emit(ctx, 1, 2); /* 00000003 SIFC_BITMAP_UNK808 */ + xf_emit(ctx, 1, 0); /* 00000003 SIFC_BITMAP_LINE_PACK_MODE */ + xf_emit(ctx, 1, 0); /* 00000001 SIFC_BITMAP_LSB_FIRST */ + xf_emit(ctx, 1, 0); /* 00000001 SIFC_BITMAP_ENABLE */ + xf_emit(ctx, 1, 0); /* 0000ffff BLIT_DST_X */ + xf_emit(ctx, 1, 0); /* 0000ffff BLIT_DST_Y */ + xf_emit(ctx, 1, 0); /* 000fffff BLIT_DU_DX_FRACT */ + xf_emit(ctx, 1, 1); /* 0001ffff BLIT_DU_DX_INT */ + xf_emit(ctx, 1, 0); /* 000fffff BLIT_DV_DY_FRACT */ + xf_emit(ctx, 1, 1); /* 0001ffff BLIT_DV_DY_INT */ + xf_emit(ctx, 1, 1); /* 0000ffff BLIT_DST_W */ + xf_emit(ctx, 1, 1); /* 0000ffff BLIT_DST_H */ + xf_emit(ctx, 1, 0); /* 000fffff BLIT_SRC_X_FRACT */ + xf_emit(ctx, 1, 0); /* 0001ffff BLIT_SRC_X_INT */ + xf_emit(ctx, 1, 0); /* 000fffff BLIT_SRC_Y_FRACT */ + xf_emit(ctx, 1, 0); /* 00000001 UNK888 */ + xf_emit(ctx, 1, 4); /* 0000003f UNK884 */ + xf_emit(ctx, 1, 0); /* 00000007 UNK880 */ + xf_emit(ctx, 1, 1); /* 0000001f tesla UNK0FB8 */ + xf_emit(ctx, 1, 0x15); /* 000000ff tesla UNK128C */ + xf_emit(ctx, 2, 0); /* 00000007, ffff0ff3 */ + xf_emit(ctx, 1, 0); /* 00000001 UNK260 */ + xf_emit(ctx, 1, 0x4444480); /* 1fffffff UNK870 */ + /* SEEK */ + xf_emit(ctx, 0x10, 0); + /* SEEK */ + xf_emit(ctx, 0x27, 0); +} + +static void +nv50_graph_construct_gene_csched(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + /* middle of strand 1 on pre-NVA0 [after eng2d], middle of strand 0 on NVAx */ + /* SEEK */ + xf_emit(ctx, 2, 0); /* 00007fff WINDOW_OFFSET_XY... what is it doing here??? */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK1924 */ + xf_emit(ctx, 1, 0); /* 00000003 WINDOW_ORIGIN */ + xf_emit(ctx, 1, 0x8100c12); /* 1fffffff FP_INTERPOLANT_CTRL */ + xf_emit(ctx, 1, 0); /* 000003ff */ + /* SEEK */ + xf_emit(ctx, 1, 0); /* ffffffff turing UNK364 */ + xf_emit(ctx, 1, 0); /* 0000000f turing UNK36C */ + xf_emit(ctx, 1, 0); /* 0000ffff USER_PARAM_COUNT */ + xf_emit(ctx, 1, 0x100); /* 00ffffff turing UNK384 */ + xf_emit(ctx, 1, 0); /* 0000000f turing UNK2A0 */ + xf_emit(ctx, 1, 0); /* 0000ffff GRIDID */ + xf_emit(ctx, 1, 0x10001); /* ffffffff GRIDDIM_XY */ + xf_emit(ctx, 1, 0); /* ffffffff */ + xf_emit(ctx, 1, 0x10001); /* ffffffff BLOCKDIM_XY */ + xf_emit(ctx, 1, 1); /* 0000ffff BLOCKDIM_Z */ + xf_emit(ctx, 1, 0x10001); /* 00ffffff BLOCK_ALLOC */ + xf_emit(ctx, 1, 1); /* 00000001 LANES32 */ + xf_emit(ctx, 1, 4); /* 000000ff FP_REG_ALLOC_TEMP */ + xf_emit(ctx, 1, 2); /* 00000003 REG_MODE */ + /* SEEK */ + xf_emit(ctx, 0x40, 0); /* ffffffff USER_PARAM */ + switch (dev_priv->chipset) { + case 0x50: + case 0x92: + xf_emit(ctx, 8, 0); /* 7, 0, 0, 0, ... */ + xf_emit(ctx, 0x80, 0); /* fff */ + xf_emit(ctx, 2, 0); /* ff, fff */ + xf_emit(ctx, 0x10*2, 0); /* ffffffff, 1f */ + break; + case 0x84: + xf_emit(ctx, 8, 0); /* 7, 0, 0, 0, ... */ + xf_emit(ctx, 0x60, 0); /* fff */ + xf_emit(ctx, 2, 0); /* ff, fff */ + xf_emit(ctx, 0xc*2, 0); /* ffffffff, 1f */ + break; + case 0x94: + case 0x96: + xf_emit(ctx, 8, 0); /* 7, 0, 0, 0, ... */ + xf_emit(ctx, 0x40, 0); /* fff */ + xf_emit(ctx, 2, 0); /* ff, fff */ + xf_emit(ctx, 8*2, 0); /* ffffffff, 1f */ + break; + case 0x86: + case 0x98: + xf_emit(ctx, 4, 0); /* f, 0, 0, 0 */ + xf_emit(ctx, 0x10, 0); /* fff */ + xf_emit(ctx, 2, 0); /* ff, fff */ + xf_emit(ctx, 2*2, 0); /* ffffffff, 1f */ + break; + case 0xa0: + xf_emit(ctx, 8, 0); /* 7, 0, 0, 0, ... */ + xf_emit(ctx, 0xf0, 0); /* fff */ + xf_emit(ctx, 2, 0); /* ff, fff */ + xf_emit(ctx, 0x1e*2, 0); /* ffffffff, 1f */ + break; + case 0xa3: + xf_emit(ctx, 8, 0); /* 7, 0, 0, 0, ... */ + xf_emit(ctx, 0x60, 0); /* fff */ + xf_emit(ctx, 2, 0); /* ff, fff */ + xf_emit(ctx, 0xc*2, 0); /* ffffffff, 1f */ + break; + case 0xa5: + case 0xaf: + xf_emit(ctx, 8, 0); /* 7, 0, 0, 0, ... */ + xf_emit(ctx, 0x30, 0); /* fff */ + xf_emit(ctx, 2, 0); /* ff, fff */ + xf_emit(ctx, 6*2, 0); /* ffffffff, 1f */ + break; + case 0xaa: + xf_emit(ctx, 0x12, 0); + break; + case 0xa8: + case 0xac: + xf_emit(ctx, 4, 0); /* f, 0, 0, 0 */ + xf_emit(ctx, 0x10, 0); /* fff */ + xf_emit(ctx, 2, 0); /* ff, fff */ + xf_emit(ctx, 2*2, 0); /* ffffffff, 1f */ + break; + } + xf_emit(ctx, 1, 0); /* 0000000f */ + xf_emit(ctx, 1, 0); /* 00000000 */ + xf_emit(ctx, 1, 0); /* ffffffff */ + xf_emit(ctx, 1, 0); /* 0000001f */ + xf_emit(ctx, 4, 0); /* ffffffff */ + xf_emit(ctx, 1, 0); /* 00000003 turing UNK35C */ + xf_emit(ctx, 1, 0); /* ffffffff */ + xf_emit(ctx, 4, 0); /* ffffffff */ + xf_emit(ctx, 1, 0); /* 00000003 turing UNK35C */ + xf_emit(ctx, 1, 0); /* ffffffff */ + xf_emit(ctx, 1, 0); /* 000000ff */ +} + +static void +nv50_graph_construct_gene_unk1cxx(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + xf_emit(ctx, 2, 0); /* 00007fff WINDOW_OFFSET_XY */ + xf_emit(ctx, 1, 0x3f800000); /* ffffffff LINE_WIDTH */ + xf_emit(ctx, 1, 0); /* 00000001 LINE_SMOOTH_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK1658 */ + xf_emit(ctx, 1, 0); /* 00000001 POLYGON_SMOOTH_ENABLE */ + xf_emit(ctx, 3, 0); /* 00000001 POLYGON_OFFSET_*_ENABLE */ + xf_emit(ctx, 1, 4); /* 0000000f CULL_MODE */ + xf_emit(ctx, 1, 0x1a); /* 0000001f POLYGON_MODE */ + xf_emit(ctx, 1, 0); /* 0000000f ZETA_FORMAT */ + xf_emit(ctx, 1, 0); /* 00000001 POINT_SPRITE_ENABLE */ + xf_emit(ctx, 1, 1); /* 00000001 tesla UNK165C */ + xf_emit(ctx, 0x10, 0); /* 00000001 SCISSOR_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK1534 */ + xf_emit(ctx, 1, 0); /* 00000001 LINE_STIPPLE_ENABLE */ + xf_emit(ctx, 1, 0x00ffff00); /* 00ffffff LINE_STIPPLE_PATTERN */ + xf_emit(ctx, 1, 0); /* ffffffff POLYGON_OFFSET_UNITS */ + xf_emit(ctx, 1, 0); /* ffffffff POLYGON_OFFSET_FACTOR */ + xf_emit(ctx, 1, 0); /* 00000003 tesla UNK1668 */ + xf_emit(ctx, 2, 0); /* 07ffffff SCREEN_SCISSOR */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK1900 */ + xf_emit(ctx, 1, 0xf); /* 0000000f COLOR_MASK */ + xf_emit(ctx, 7, 0); /* 0000000f COLOR_MASK */ + xf_emit(ctx, 1, 0x0fac6881); /* 0fffffff RT_CONTROL */ + xf_emit(ctx, 1, 0x11); /* 0000007f RT_FORMAT */ + xf_emit(ctx, 7, 0); /* 0000007f RT_FORMAT */ + xf_emit(ctx, 8, 0); /* 00000001 RT_HORIZ_LINEAR */ + xf_emit(ctx, 1, 4); /* 00000007 FP_CONTROL */ + xf_emit(ctx, 1, 0); /* 00000001 ALPHA_TEST_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000007 ALPHA_TEST_FUNC */ + if (IS_NVA3F(dev_priv->chipset)) + xf_emit(ctx, 1, 3); /* 00000003 UNK16B4 */ + else if (dev_priv->chipset >= 0xa0) + xf_emit(ctx, 1, 1); /* 00000001 UNK16B4 */ + xf_emit(ctx, 1, 0); /* 00000003 MULTISAMPLE_CTRL */ + xf_emit(ctx, 1, 0); /* 00000003 tesla UNK0F90 */ + xf_emit(ctx, 1, 2); /* 00000003 tesla UNK143C */ + xf_emit(ctx, 2, 0x04000000); /* 07ffffff tesla UNK0D6C */ + xf_emit(ctx, 1, 0); /* 000000ff STENCIL_FRONT_MASK */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_WRITE_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000001 SAMPLECNT_ENABLE */ + xf_emit(ctx, 1, 5); /* 0000000f UNK1408 */ + xf_emit(ctx, 1, 0x52); /* 000001ff SEMANTIC_PTSZ */ + xf_emit(ctx, 1, 0); /* ffffffff POINT_SIZE */ + xf_emit(ctx, 1, 0); /* 00000001 */ + xf_emit(ctx, 1, 0); /* 00000007 tesla UNK0FB4 */ + if (dev_priv->chipset != 0x50) { + xf_emit(ctx, 1, 0); /* 3ff */ + xf_emit(ctx, 1, 1); /* 00000001 tesla UNK1110 */ + } + if (IS_NVA3F(dev_priv->chipset)) + xf_emit(ctx, 1, 0); /* 00000003 tesla UNK1928 */ + xf_emit(ctx, 0x10, 0); /* ffffffff DEPTH_RANGE_NEAR */ + xf_emit(ctx, 0x10, 0x3f800000); /* ffffffff DEPTH_RANGE_FAR */ + xf_emit(ctx, 1, 0x10); /* 000000ff VIEW_VOLUME_CLIP_CTRL */ + xf_emit(ctx, 0x20, 0); /* 07ffffff VIEWPORT_HORIZ, then VIEWPORT_VERT. (W&0x3fff)<<13 | (X&0x1fff). */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK187C */ + xf_emit(ctx, 1, 0); /* 00000003 WINDOW_ORIGIN */ + xf_emit(ctx, 1, 0); /* 00000001 STENCIL_FRONT_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_TEST_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000001 STENCIL_BACK_ENABLE */ + xf_emit(ctx, 1, 0); /* 000000ff STENCIL_BACK_MASK */ + xf_emit(ctx, 1, 0x8100c12); /* 1fffffff FP_INTERPOLANT_CTRL */ + xf_emit(ctx, 1, 5); /* 0000000f tesla UNK1220 */ + xf_emit(ctx, 1, 0); /* 00000007 MULTISAMPLE_SAMPLES_LOG2 */ + xf_emit(ctx, 1, 0); /* 000000ff tesla UNK1A20 */ + xf_emit(ctx, 1, 1); /* 00000001 ZETA_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000001 VERTEX_TWO_SIDE_ENABLE */ + xf_emit(ctx, 4, 0xffff); /* 0000ffff MSAA_MASK */ + if (dev_priv->chipset != 0x50) + xf_emit(ctx, 1, 3); /* 00000003 tesla UNK1100 */ + if (dev_priv->chipset < 0xa0) + xf_emit(ctx, 0x1c, 0); /* RO */ + else if (IS_NVA3F(dev_priv->chipset)) + xf_emit(ctx, 0x9, 0); + xf_emit(ctx, 1, 0); /* 00000001 UNK1534 */ + xf_emit(ctx, 1, 0); /* 00000001 LINE_SMOOTH_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000001 LINE_STIPPLE_ENABLE */ + xf_emit(ctx, 1, 0x00ffff00); /* 00ffffff LINE_STIPPLE_PATTERN */ + xf_emit(ctx, 1, 0x1a); /* 0000001f POLYGON_MODE */ + xf_emit(ctx, 1, 0); /* 00000003 WINDOW_ORIGIN */ + if (dev_priv->chipset != 0x50) { + xf_emit(ctx, 1, 3); /* 00000003 tesla UNK1100 */ + xf_emit(ctx, 1, 0); /* 3ff */ + } + /* XXX: the following block could belong either to unk1cxx, or + * to STRMOUT. Rather hard to tell. */ + if (dev_priv->chipset < 0xa0) + xf_emit(ctx, 0x25, 0); + else + xf_emit(ctx, 0x3b, 0); +} + +static void +nv50_graph_construct_gene_strmout(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + xf_emit(ctx, 1, 0x102); /* 0000ffff STRMOUT_BUFFER_CTRL */ + xf_emit(ctx, 1, 0); /* ffffffff STRMOUT_PRIMITIVE_COUNT */ + xf_emit(ctx, 4, 4); /* 000000ff STRMOUT_NUM_ATTRIBS */ + if (dev_priv->chipset >= 0xa0) { + xf_emit(ctx, 4, 0); /* ffffffff UNK1A8C */ + xf_emit(ctx, 4, 0); /* ffffffff UNK1780 */ + } + xf_emit(ctx, 1, 4); /* 000000ff GP_RESULT_MAP_SIZE */ + xf_emit(ctx, 1, 4); /* 0000007f VP_RESULT_MAP_SIZE */ + xf_emit(ctx, 1, 0); /* 00000001 GP_ENABLE */ + if (dev_priv->chipset == 0x50) + xf_emit(ctx, 1, 0x3ff); /* 000003ff tesla UNK0D68 */ + else + xf_emit(ctx, 1, 0x7ff); /* 000007ff tesla UNK0D68 */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A30 */ + /* SEEK */ + xf_emit(ctx, 1, 0x102); /* 0000ffff STRMOUT_BUFFER_CTRL */ + xf_emit(ctx, 1, 0); /* ffffffff STRMOUT_PRIMITIVE_COUNT */ + xf_emit(ctx, 4, 0); /* 000000ff STRMOUT_ADDRESS_HIGH */ + xf_emit(ctx, 4, 0); /* ffffffff STRMOUT_ADDRESS_LOW */ + xf_emit(ctx, 4, 4); /* 000000ff STRMOUT_NUM_ATTRIBS */ + if (dev_priv->chipset >= 0xa0) { + xf_emit(ctx, 4, 0); /* ffffffff UNK1A8C */ + xf_emit(ctx, 4, 0); /* ffffffff UNK1780 */ + } + xf_emit(ctx, 1, 0); /* 0000ffff DMA_STRMOUT */ + xf_emit(ctx, 1, 0); /* 0000ffff DMA_QUERY */ + xf_emit(ctx, 1, 0); /* 000000ff QUERY_ADDRESS_HIGH */ + xf_emit(ctx, 2, 0); /* ffffffff QUERY_ADDRESS_LOW QUERY_COUNTER */ + xf_emit(ctx, 2, 0); /* ffffffff */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A30 */ + /* SEEK */ + xf_emit(ctx, 0x20, 0); /* ffffffff STRMOUT_MAP */ + xf_emit(ctx, 1, 0); /* 0000000f */ + xf_emit(ctx, 1, 0); /* 00000000? */ + xf_emit(ctx, 2, 0); /* ffffffff */ +} + +static void +nv50_graph_construct_gene_ropm1(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + xf_emit(ctx, 1, 0x4e3bfdf); /* ffffffff UNK0D64 */ + xf_emit(ctx, 1, 0x4e3bfdf); /* ffffffff UNK0DF4 */ + xf_emit(ctx, 1, 0); /* 00000007 */ + xf_emit(ctx, 1, 0); /* 000003ff */ + if (IS_NVA3F(dev_priv->chipset)) + xf_emit(ctx, 1, 0x11); /* 000000ff tesla UNK1968 */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A3C */ +} + +static void +nv50_graph_construct_gene_ropm2(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + /* SEEK */ + xf_emit(ctx, 1, 0); /* 0000ffff DMA_QUERY */ + xf_emit(ctx, 1, 0x0fac6881); /* 0fffffff RT_CONTROL */ + xf_emit(ctx, 2, 0); /* ffffffff */ + xf_emit(ctx, 1, 0); /* 000000ff QUERY_ADDRESS_HIGH */ + xf_emit(ctx, 2, 0); /* ffffffff QUERY_ADDRESS_LOW, COUNTER */ + xf_emit(ctx, 1, 0); /* 00000001 SAMPLECNT_ENABLE */ + xf_emit(ctx, 1, 0); /* 7 */ + /* SEEK */ + xf_emit(ctx, 1, 0); /* 0000ffff DMA_QUERY */ + xf_emit(ctx, 1, 0); /* 000000ff QUERY_ADDRESS_HIGH */ + xf_emit(ctx, 2, 0); /* ffffffff QUERY_ADDRESS_LOW, COUNTER */ + xf_emit(ctx, 1, 0x4e3bfdf); /* ffffffff UNK0D64 */ + xf_emit(ctx, 1, 0x4e3bfdf); /* ffffffff UNK0DF4 */ + xf_emit(ctx, 1, 0); /* 00000001 eng2d UNK260 */ + xf_emit(ctx, 1, 0); /* ff/3ff */ + xf_emit(ctx, 1, 0); /* 00000007 */ + if (IS_NVA3F(dev_priv->chipset)) + xf_emit(ctx, 1, 0x11); /* 000000ff tesla UNK1968 */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A3C */ +} + +static void +nv50_graph_construct_gene_ropc(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + int magic2; + if (dev_priv->chipset == 0x50) { + magic2 = 0x00003e60; + } else if (!IS_NVA3F(dev_priv->chipset)) { + magic2 = 0x001ffe67; + } else { + magic2 = 0x00087e67; + } + xf_emit(ctx, 1, 0); /* f/7 MUTISAMPLE_SAMPLES_LOG2 */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK1534 */ + xf_emit(ctx, 1, 0); /* 00000007 STENCIL_BACK_FUNC_FUNC */ + xf_emit(ctx, 1, 0); /* 000000ff STENCIL_BACK_FUNC_MASK */ + xf_emit(ctx, 1, 0); /* 000000ff STENCIL_BACK_MASK */ + xf_emit(ctx, 3, 0); /* 00000007 STENCIL_BACK_OP_FAIL, ZFAIL, ZPASS */ + xf_emit(ctx, 1, 2); /* 00000003 tesla UNK143C */ + xf_emit(ctx, 1, 0); /* ffff0ff3 */ + xf_emit(ctx, 1, magic2); /* 001fffff tesla UNK0F78 */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_BOUNDS_EN */ + xf_emit(ctx, 1, 0); /* 00000007 DEPTH_TEST_FUNC */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_TEST_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_WRITE_ENABLE */ + if (IS_NVA3F(dev_priv->chipset)) + xf_emit(ctx, 1, 1); /* 0000001f tesla UNK169C */ + xf_emit(ctx, 1, 0); /* 00000007 STENCIL_FRONT_FUNC_FUNC */ + xf_emit(ctx, 1, 0); /* 000000ff STENCIL_FRONT_FUNC_MASK */ + xf_emit(ctx, 1, 0); /* 000000ff STENCIL_FRONT_MASK */ + xf_emit(ctx, 3, 0); /* 00000007 STENCIL_FRONT_OP_FAIL, ZFAIL, ZPASS */ + xf_emit(ctx, 1, 0); /* 00000001 STENCIL_FRONT_ENABLE */ + if (dev_priv->chipset >= 0xa0 && !IS_NVAAF(dev_priv->chipset)) + xf_emit(ctx, 1, 0x15); /* 000000ff */ + xf_emit(ctx, 1, 0); /* 00000001 STENCIL_BACK_ENABLE */ + xf_emit(ctx, 1, 1); /* 00000001 tesla UNK15B4 */ + xf_emit(ctx, 1, 0x10); /* 3ff/ff VIEW_VOLUME_CLIP_CTRL */ + xf_emit(ctx, 1, 0); /* ffffffff CLEAR_DEPTH */ + xf_emit(ctx, 1, 0); /* 0000000f ZETA_FORMAT */ + xf_emit(ctx, 1, 1); /* 00000001 ZETA_ENABLE */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A3C */ + if (dev_priv->chipset == 0x86 || dev_priv->chipset == 0x92 || dev_priv->chipset == 0x98 || dev_priv->chipset >= 0xa0) { + xf_emit(ctx, 3, 0); /* ff, ffffffff, ffffffff */ + xf_emit(ctx, 1, 4); /* 7 */ + xf_emit(ctx, 1, 0x400); /* fffffff */ + xf_emit(ctx, 1, 0x300); /* ffff */ + xf_emit(ctx, 1, 0x1001); /* 1fff */ + if (dev_priv->chipset != 0xa0) { + if (IS_NVA3F(dev_priv->chipset)) + xf_emit(ctx, 1, 0); /* 0000000f UNK15C8 */ + else + xf_emit(ctx, 1, 0x15); /* ff */ + } + } + xf_emit(ctx, 1, 0); /* 00000007 MULTISAMPLE_SAMPLES_LOG2 */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK1534 */ + xf_emit(ctx, 1, 0); /* 00000007 STENCIL_BACK_FUNC_FUNC */ + xf_emit(ctx, 1, 0); /* 000000ff STENCIL_BACK_FUNC_MASK */ + xf_emit(ctx, 1, 0); /* ffff0ff3 */ + xf_emit(ctx, 1, 2); /* 00000003 tesla UNK143C */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_BOUNDS_EN */ + xf_emit(ctx, 1, 0); /* 00000007 DEPTH_TEST_FUNC */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_TEST_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_WRITE_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000007 STENCIL_FRONT_FUNC_FUNC */ + xf_emit(ctx, 1, 0); /* 000000ff STENCIL_FRONT_FUNC_MASK */ + xf_emit(ctx, 1, 0); /* 00000001 STENCIL_FRONT_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000001 STENCIL_BACK_ENABLE */ + xf_emit(ctx, 1, 1); /* 00000001 tesla UNK15B4 */ + xf_emit(ctx, 1, 0x10); /* 7f/ff VIEW_VOLUME_CLIP_CTRL */ + xf_emit(ctx, 1, 0); /* 0000000f ZETA_FORMAT */ + xf_emit(ctx, 1, 1); /* 00000001 ZETA_ENABLE */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A3C */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK1534 */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK1900 */ + xf_emit(ctx, 1, 0); /* 00000007 STENCIL_BACK_FUNC_FUNC */ + xf_emit(ctx, 1, 0); /* 000000ff STENCIL_BACK_FUNC_MASK */ + xf_emit(ctx, 1, 0); /* 000000ff STENCIL_BACK_FUNC_REF */ + xf_emit(ctx, 2, 0); /* ffffffff DEPTH_BOUNDS */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_BOUNDS_EN */ + xf_emit(ctx, 1, 0); /* 00000007 DEPTH_TEST_FUNC */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_TEST_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_WRITE_ENABLE */ + xf_emit(ctx, 1, 0); /* 0000000f */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK0FB0 */ + xf_emit(ctx, 1, 0); /* 00000007 STENCIL_FRONT_FUNC_FUNC */ + xf_emit(ctx, 1, 0); /* 000000ff STENCIL_FRONT_FUNC_MASK */ + xf_emit(ctx, 1, 0); /* 000000ff STENCIL_FRONT_FUNC_REF */ + xf_emit(ctx, 1, 0); /* 00000001 STENCIL_FRONT_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000001 STENCIL_BACK_ENABLE */ + xf_emit(ctx, 1, 0x10); /* 7f/ff VIEW_VOLUME_CLIP_CTRL */ + xf_emit(ctx, 0x10, 0); /* ffffffff DEPTH_RANGE_NEAR */ + xf_emit(ctx, 0x10, 0x3f800000); /* ffffffff DEPTH_RANGE_FAR */ + xf_emit(ctx, 1, 0); /* 0000000f ZETA_FORMAT */ + xf_emit(ctx, 1, 0); /* 00000007 MULTISAMPLE_SAMPLES_LOG2 */ + xf_emit(ctx, 1, 0); /* 00000007 STENCIL_BACK_FUNC_FUNC */ + xf_emit(ctx, 1, 0); /* 000000ff STENCIL_BACK_FUNC_MASK */ + xf_emit(ctx, 1, 0); /* 000000ff STENCIL_BACK_FUNC_REF */ + xf_emit(ctx, 1, 0); /* 000000ff STENCIL_BACK_MASK */ + xf_emit(ctx, 3, 0); /* 00000007 STENCIL_BACK_OP_FAIL, ZFAIL, ZPASS */ + xf_emit(ctx, 2, 0); /* ffffffff DEPTH_BOUNDS */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_BOUNDS_EN */ + xf_emit(ctx, 1, 0); /* 00000007 DEPTH_TEST_FUNC */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_TEST_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_WRITE_ENABLE */ + xf_emit(ctx, 1, 0); /* 000000ff CLEAR_STENCIL */ + xf_emit(ctx, 1, 0); /* 00000007 STENCIL_FRONT_FUNC_FUNC */ + xf_emit(ctx, 1, 0); /* 000000ff STENCIL_FRONT_FUNC_MASK */ + xf_emit(ctx, 1, 0); /* 000000ff STENCIL_FRONT_FUNC_REF */ + xf_emit(ctx, 1, 0); /* 000000ff STENCIL_FRONT_MASK */ + xf_emit(ctx, 3, 0); /* 00000007 STENCIL_FRONT_OP_FAIL, ZFAIL, ZPASS */ + xf_emit(ctx, 1, 0); /* 00000001 STENCIL_FRONT_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000001 STENCIL_BACK_ENABLE */ + xf_emit(ctx, 1, 0x10); /* 7f/ff VIEW_VOLUME_CLIP_CTRL */ + xf_emit(ctx, 1, 0); /* 0000000f ZETA_FORMAT */ + xf_emit(ctx, 1, 0x3f); /* 0000003f UNK1590 */ + xf_emit(ctx, 1, 0); /* 00000007 MULTISAMPLE_SAMPLES_LOG2 */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK1534 */ + xf_emit(ctx, 2, 0); /* ffff0ff3, ffff */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK0FB0 */ + xf_emit(ctx, 1, 0); /* 0001ffff GP_BUILTIN_RESULT_EN */ + xf_emit(ctx, 1, 1); /* 00000001 tesla UNK15B4 */ + xf_emit(ctx, 1, 0); /* 0000000f ZETA_FORMAT */ + xf_emit(ctx, 1, 1); /* 00000001 ZETA_ENABLE */ + xf_emit(ctx, 1, 0); /* ffffffff CLEAR_DEPTH */ + xf_emit(ctx, 1, 1); /* 00000001 tesla UNK19CC */ + if (dev_priv->chipset >= 0xa0) { + xf_emit(ctx, 2, 0); + xf_emit(ctx, 1, 0x1001); + xf_emit(ctx, 0xb, 0); + } else { + xf_emit(ctx, 1, 0); /* 00000007 */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK1534 */ + xf_emit(ctx, 1, 0); /* 00000007 MULTISAMPLE_SAMPLES_LOG2 */ + xf_emit(ctx, 8, 0); /* 00000001 BLEND_ENABLE */ + xf_emit(ctx, 1, 0); /* ffff0ff3 */ + } + xf_emit(ctx, 1, 0x11); /* 3f/7f RT_FORMAT */ + xf_emit(ctx, 7, 0); /* 3f/7f RT_FORMAT */ + xf_emit(ctx, 1, 0xf); /* 0000000f COLOR_MASK */ + xf_emit(ctx, 7, 0); /* 0000000f COLOR_MASK */ + xf_emit(ctx, 1, 0x11); /* 3f/7f */ + xf_emit(ctx, 1, 0); /* 00000001 LOGIC_OP_ENABLE */ + if (dev_priv->chipset != 0x50) { + xf_emit(ctx, 1, 0); /* 0000000f LOGIC_OP */ + xf_emit(ctx, 1, 0); /* 000000ff */ + } + xf_emit(ctx, 1, 0); /* 00000007 OPERATION */ + xf_emit(ctx, 1, 0); /* ff/3ff */ + xf_emit(ctx, 1, 0); /* 00000003 UNK0F90 */ + xf_emit(ctx, 2, 1); /* 00000007 BLEND_EQUATION_RGB, ALPHA */ + xf_emit(ctx, 1, 1); /* 00000001 UNK133C */ + xf_emit(ctx, 1, 2); /* 0000001f BLEND_FUNC_SRC_RGB */ + xf_emit(ctx, 1, 1); /* 0000001f BLEND_FUNC_DST_RGB */ + xf_emit(ctx, 1, 2); /* 0000001f BLEND_FUNC_SRC_ALPHA */ + xf_emit(ctx, 1, 1); /* 0000001f BLEND_FUNC_DST_ALPHA */ + xf_emit(ctx, 1, 0); /* 00000001 */ + xf_emit(ctx, 1, magic2); /* 001fffff tesla UNK0F78 */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A3C */ + xf_emit(ctx, 1, 0x0fac6881); /* 0fffffff RT_CONTROL */ + if (IS_NVA3F(dev_priv->chipset)) { + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK12E4 */ + xf_emit(ctx, 8, 1); /* 00000007 IBLEND_EQUATION_RGB */ + xf_emit(ctx, 8, 1); /* 00000007 IBLEND_EQUATION_ALPHA */ + xf_emit(ctx, 8, 1); /* 00000001 IBLEND_UNK00 */ + xf_emit(ctx, 8, 2); /* 0000001f IBLEND_FUNC_SRC_RGB */ + xf_emit(ctx, 8, 1); /* 0000001f IBLEND_FUNC_DST_RGB */ + xf_emit(ctx, 8, 2); /* 0000001f IBLEND_FUNC_SRC_ALPHA */ + xf_emit(ctx, 8, 1); /* 0000001f IBLEND_FUNC_DST_ALPHA */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK1140 */ + xf_emit(ctx, 2, 0); /* 00000001 */ + xf_emit(ctx, 1, 1); /* 0000001f tesla UNK169C */ + xf_emit(ctx, 1, 0); /* 0000000f */ + xf_emit(ctx, 1, 0); /* 00000003 */ + xf_emit(ctx, 1, 0); /* ffffffff */ + xf_emit(ctx, 2, 0); /* 00000001 */ + xf_emit(ctx, 1, 1); /* 0000001f tesla UNK169C */ + xf_emit(ctx, 1, 0); /* 00000001 */ + xf_emit(ctx, 1, 0); /* 000003ff */ + } else if (dev_priv->chipset >= 0xa0) { + xf_emit(ctx, 2, 0); /* 00000001 */ + xf_emit(ctx, 1, 0); /* 00000007 */ + xf_emit(ctx, 1, 0); /* 00000003 */ + xf_emit(ctx, 1, 0); /* ffffffff */ + xf_emit(ctx, 2, 0); /* 00000001 */ + } else { + xf_emit(ctx, 1, 0); /* 00000007 MULTISAMPLE_SAMPLES_LOG2 */ + xf_emit(ctx, 1, 0); /* 00000003 tesla UNK1430 */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A3C */ + } + xf_emit(ctx, 4, 0); /* ffffffff CLEAR_COLOR */ + xf_emit(ctx, 4, 0); /* ffffffff BLEND_COLOR A R G B */ + xf_emit(ctx, 1, 0); /* 00000fff eng2d UNK2B0 */ + if (dev_priv->chipset >= 0xa0) + xf_emit(ctx, 2, 0); /* 00000001 */ + xf_emit(ctx, 1, 0); /* 000003ff */ + xf_emit(ctx, 8, 0); /* 00000001 BLEND_ENABLE */ + xf_emit(ctx, 1, 1); /* 00000001 UNK133C */ + xf_emit(ctx, 1, 2); /* 0000001f BLEND_FUNC_SRC_RGB */ + xf_emit(ctx, 1, 1); /* 0000001f BLEND_FUNC_DST_RGB */ + xf_emit(ctx, 1, 1); /* 00000007 BLEND_EQUATION_RGB */ + xf_emit(ctx, 1, 2); /* 0000001f BLEND_FUNC_SRC_ALPHA */ + xf_emit(ctx, 1, 1); /* 0000001f BLEND_FUNC_DST_ALPHA */ + xf_emit(ctx, 1, 1); /* 00000007 BLEND_EQUATION_ALPHA */ + xf_emit(ctx, 1, 0); /* 00000001 UNK19C0 */ + xf_emit(ctx, 1, 0); /* 00000001 LOGIC_OP_ENABLE */ + xf_emit(ctx, 1, 0); /* 0000000f LOGIC_OP */ + if (dev_priv->chipset >= 0xa0) + xf_emit(ctx, 1, 0); /* 00000001 UNK12E4? NVA3+ only? */ + if (IS_NVA3F(dev_priv->chipset)) { + xf_emit(ctx, 8, 1); /* 00000001 IBLEND_UNK00 */ + xf_emit(ctx, 8, 1); /* 00000007 IBLEND_EQUATION_RGB */ + xf_emit(ctx, 8, 2); /* 0000001f IBLEND_FUNC_SRC_RGB */ + xf_emit(ctx, 8, 1); /* 0000001f IBLEND_FUNC_DST_RGB */ + xf_emit(ctx, 8, 1); /* 00000007 IBLEND_EQUATION_ALPHA */ + xf_emit(ctx, 8, 2); /* 0000001f IBLEND_FUNC_SRC_ALPHA */ + xf_emit(ctx, 8, 1); /* 0000001f IBLEND_FUNC_DST_ALPHA */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK15C4 */ + xf_emit(ctx, 1, 0); /* 00000001 */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK1140 */ + } + xf_emit(ctx, 1, 0x11); /* 3f/7f DST_FORMAT */ + xf_emit(ctx, 1, 1); /* 00000001 DST_LINEAR */ + xf_emit(ctx, 1, 0); /* 00000007 PATTERN_COLOR_FORMAT */ + xf_emit(ctx, 2, 0); /* ffffffff PATTERN_MONO_COLOR */ + xf_emit(ctx, 1, 0); /* 00000001 PATTERN_MONO_FORMAT */ + xf_emit(ctx, 2, 0); /* ffffffff PATTERN_MONO_BITMAP */ + xf_emit(ctx, 1, 0); /* 00000003 PATTERN_SELECT */ + xf_emit(ctx, 1, 0); /* 000000ff ROP */ + xf_emit(ctx, 1, 0); /* ffffffff BETA1 */ + xf_emit(ctx, 1, 0); /* ffffffff BETA4 */ + xf_emit(ctx, 1, 0); /* 00000007 OPERATION */ + xf_emit(ctx, 0x50, 0); /* 10x ffffff, ffffff, ffffff, ffffff, 3 PATTERN */ +} + +static void +nv50_graph_construct_xfer_unk84xx(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + int magic3; + switch (dev_priv->chipset) { + case 0x50: + magic3 = 0x1000; + break; + case 0x86: + case 0x98: + case 0xa8: + case 0xaa: + case 0xac: + case 0xaf: + magic3 = 0x1e00; + break; + default: + magic3 = 0; + } + xf_emit(ctx, 1, 0); /* 00000001 GP_ENABLE */ + xf_emit(ctx, 1, 4); /* 7f/ff[NVA0+] VP_REG_ALLOC_RESULT */ + xf_emit(ctx, 1, 0); /* 00000001 GP_ENABLE */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A30 */ + xf_emit(ctx, 1, 0); /* 111/113[NVA0+] */ + if (IS_NVA3F(dev_priv->chipset)) + xf_emit(ctx, 0x1f, 0); /* ffffffff */ + else if (dev_priv->chipset >= 0xa0) + xf_emit(ctx, 0x0f, 0); /* ffffffff */ + else + xf_emit(ctx, 0x10, 0); /* fffffff VP_RESULT_MAP_1 up */ + xf_emit(ctx, 2, 0); /* f/1f[NVA3], fffffff/ffffffff[NVA0+] */ + xf_emit(ctx, 1, 4); /* 7f/ff VP_REG_ALLOC_RESULT */ + xf_emit(ctx, 1, 4); /* 7f/ff VP_RESULT_MAP_SIZE */ + if (dev_priv->chipset >= 0xa0) + xf_emit(ctx, 1, 0x03020100); /* ffffffff */ + else + xf_emit(ctx, 1, 0x00608080); /* fffffff VP_RESULT_MAP_0 */ + xf_emit(ctx, 1, 0); /* 00000001 GP_ENABLE */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A30 */ + xf_emit(ctx, 2, 0); /* 111/113, 7f/ff */ + xf_emit(ctx, 1, 4); /* 7f/ff VP_RESULT_MAP_SIZE */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A30 */ + xf_emit(ctx, 1, 0); /* 00000001 GP_ENABLE */ + xf_emit(ctx, 1, 4); /* 000000ff GP_REG_ALLOC_RESULT */ + xf_emit(ctx, 1, 4); /* 000000ff GP_RESULT_MAP_SIZE */ + xf_emit(ctx, 1, 0x80); /* 0000ffff GP_VERTEX_OUTPUT_COUNT */ + if (magic3) + xf_emit(ctx, 1, magic3); /* 00007fff tesla UNK141C */ + xf_emit(ctx, 1, 4); /* 7f/ff VP_RESULT_MAP_SIZE */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A30 */ + xf_emit(ctx, 1, 0); /* 111/113 */ + xf_emit(ctx, 0x1f, 0); /* ffffffff GP_RESULT_MAP_1 up */ + xf_emit(ctx, 1, 0); /* 0000001f */ + xf_emit(ctx, 1, 0); /* ffffffff */ + xf_emit(ctx, 1, 0); /* 00000001 GP_ENABLE */ + xf_emit(ctx, 1, 4); /* 000000ff GP_REG_ALLOC_RESULT */ + xf_emit(ctx, 1, 0x80); /* 0000ffff GP_VERTEX_OUTPUT_COUNT */ + xf_emit(ctx, 1, 4); /* 000000ff GP_RESULT_MAP_SIZE */ + xf_emit(ctx, 1, 0x03020100); /* ffffffff GP_RESULT_MAP_0 */ + xf_emit(ctx, 1, 3); /* 00000003 GP_OUTPUT_PRIMITIVE_TYPE */ + if (magic3) + xf_emit(ctx, 1, magic3); /* 7fff tesla UNK141C */ + xf_emit(ctx, 1, 4); /* 7f/ff VP_RESULT_MAP_SIZE */ + xf_emit(ctx, 1, 0); /* 00000001 PROVOKING_VERTEX_LAST */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A30 */ + xf_emit(ctx, 1, 0); /* 111/113 */ + xf_emit(ctx, 1, 0); /* 00000001 GP_ENABLE */ + xf_emit(ctx, 1, 4); /* 000000ff GP_RESULT_MAP_SIZE */ + xf_emit(ctx, 1, 3); /* 00000003 GP_OUTPUT_PRIMITIVE_TYPE */ + xf_emit(ctx, 1, 0); /* 00000001 PROVOKING_VERTEX_LAST */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A30 */ + xf_emit(ctx, 1, 0); /* 00000003 tesla UNK13A0 */ + xf_emit(ctx, 1, 4); /* 7f/ff VP_REG_ALLOC_RESULT */ + xf_emit(ctx, 1, 0); /* 00000001 GP_ENABLE */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A30 */ + xf_emit(ctx, 1, 0); /* 111/113 */ + if (dev_priv->chipset == 0x94 || dev_priv->chipset == 0x96) + xf_emit(ctx, 0x1020, 0); /* 4 x (0x400 x 0xffffffff, ff, 0, 0, 0, 4 x ffffffff) */ + else if (dev_priv->chipset < 0xa0) + xf_emit(ctx, 0xa20, 0); /* 4 x (0x280 x 0xffffffff, ff, 0, 0, 0, 4 x ffffffff) */ + else if (!IS_NVA3F(dev_priv->chipset)) + xf_emit(ctx, 0x210, 0); /* ffffffff */ + else + xf_emit(ctx, 0x410, 0); /* ffffffff */ + xf_emit(ctx, 1, 0); /* 00000001 GP_ENABLE */ + xf_emit(ctx, 1, 4); /* 000000ff GP_RESULT_MAP_SIZE */ + xf_emit(ctx, 1, 3); /* 00000003 GP_OUTPUT_PRIMITIVE_TYPE */ + xf_emit(ctx, 1, 0); /* 00000001 PROVOKING_VERTEX_LAST */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A30 */ +} + +static void +nv50_graph_construct_xfer_tprop(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + int magic1, magic2; + if (dev_priv->chipset == 0x50) { + magic1 = 0x3ff; + magic2 = 0x00003e60; + } else if (!IS_NVA3F(dev_priv->chipset)) { + magic1 = 0x7ff; + magic2 = 0x001ffe67; + } else { + magic1 = 0x7ff; + magic2 = 0x00087e67; + } + xf_emit(ctx, 1, 0); /* 00000007 ALPHA_TEST_FUNC */ + xf_emit(ctx, 1, 0); /* ffffffff ALPHA_TEST_REF */ + xf_emit(ctx, 1, 0); /* 00000001 ALPHA_TEST_ENABLE */ + if (IS_NVA3F(dev_priv->chipset)) + xf_emit(ctx, 1, 1); /* 0000000f UNK16A0 */ + xf_emit(ctx, 1, 0); /* 7/f MULTISAMPLE_SAMPLES_LOG2 */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK1534 */ + xf_emit(ctx, 1, 0); /* 000000ff STENCIL_BACK_MASK */ + xf_emit(ctx, 3, 0); /* 00000007 STENCIL_BACK_OP_FAIL, ZFAIL, ZPASS */ + xf_emit(ctx, 4, 0); /* ffffffff BLEND_COLOR */ + xf_emit(ctx, 1, 0); /* 00000001 UNK19C0 */ + xf_emit(ctx, 1, 0); /* 00000001 UNK0FDC */ + xf_emit(ctx, 1, 0xf); /* 0000000f COLOR_MASK */ + xf_emit(ctx, 7, 0); /* 0000000f COLOR_MASK */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_TEST_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_WRITE_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000001 LOGIC_OP_ENABLE */ + xf_emit(ctx, 1, 0); /* ff[NV50]/3ff[NV84+] */ + xf_emit(ctx, 1, 4); /* 00000007 FP_CONTROL */ + xf_emit(ctx, 4, 0xffff); /* 0000ffff MSAA_MASK */ + xf_emit(ctx, 1, 0); /* 000000ff STENCIL_FRONT_MASK */ + xf_emit(ctx, 3, 0); /* 00000007 STENCIL_FRONT_OP_FAIL, ZFAIL, ZPASS */ + xf_emit(ctx, 1, 0); /* 00000001 STENCIL_FRONT_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000001 STENCIL_BACK_ENABLE */ + xf_emit(ctx, 2, 0); /* 00007fff WINDOW_OFFSET_XY */ + xf_emit(ctx, 1, 1); /* 00000001 tesla UNK19CC */ + xf_emit(ctx, 1, 0); /* 7 */ + xf_emit(ctx, 1, 0); /* 00000001 SAMPLECNT_ENABLE */ + xf_emit(ctx, 1, 0); /* 0000000f ZETA_FORMAT */ + xf_emit(ctx, 1, 1); /* 00000001 ZETA_ENABLE */ + xf_emit(ctx, 1, 0); /* ffffffff COLOR_KEY */ + xf_emit(ctx, 1, 0); /* 00000001 COLOR_KEY_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000007 COLOR_KEY_FORMAT */ + xf_emit(ctx, 2, 0); /* ffffffff SIFC_BITMAP_COLOR */ + xf_emit(ctx, 1, 1); /* 00000001 SIFC_BITMAP_WRITE_BIT0_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000007 ALPHA_TEST_FUNC */ + xf_emit(ctx, 1, 0); /* 00000001 ALPHA_TEST_ENABLE */ + if (IS_NVA3F(dev_priv->chipset)) { + xf_emit(ctx, 1, 3); /* 00000003 tesla UNK16B4 */ + xf_emit(ctx, 1, 0); /* 00000003 */ + xf_emit(ctx, 1, 0); /* 00000003 tesla UNK1298 */ + } else if (dev_priv->chipset >= 0xa0) { + xf_emit(ctx, 1, 1); /* 00000001 tesla UNK16B4 */ + xf_emit(ctx, 1, 0); /* 00000003 */ + } else { + xf_emit(ctx, 1, 0); /* 00000003 MULTISAMPLE_CTRL */ + } + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK1534 */ + xf_emit(ctx, 8, 0); /* 00000001 BLEND_ENABLE */ + xf_emit(ctx, 1, 1); /* 0000001f BLEND_FUNC_DST_ALPHA */ + xf_emit(ctx, 1, 1); /* 00000007 BLEND_EQUATION_ALPHA */ + xf_emit(ctx, 1, 2); /* 0000001f BLEND_FUNC_SRC_ALPHA */ + xf_emit(ctx, 1, 1); /* 0000001f BLEND_FUNC_DST_RGB */ + xf_emit(ctx, 1, 1); /* 00000007 BLEND_EQUATION_RGB */ + xf_emit(ctx, 1, 2); /* 0000001f BLEND_FUNC_SRC_RGB */ + if (IS_NVA3F(dev_priv->chipset)) { + xf_emit(ctx, 1, 0); /* 00000001 UNK12E4 */ + xf_emit(ctx, 8, 1); /* 00000007 IBLEND_EQUATION_RGB */ + xf_emit(ctx, 8, 1); /* 00000007 IBLEND_EQUATION_ALPHA */ + xf_emit(ctx, 8, 1); /* 00000001 IBLEND_UNK00 */ + xf_emit(ctx, 8, 2); /* 0000001f IBLEND_SRC_RGB */ + xf_emit(ctx, 8, 1); /* 0000001f IBLEND_DST_RGB */ + xf_emit(ctx, 8, 2); /* 0000001f IBLEND_SRC_ALPHA */ + xf_emit(ctx, 8, 1); /* 0000001f IBLEND_DST_ALPHA */ + xf_emit(ctx, 1, 0); /* 00000001 UNK1140 */ + } + xf_emit(ctx, 1, 1); /* 00000001 UNK133C */ + xf_emit(ctx, 1, 0); /* ffff0ff3 */ + xf_emit(ctx, 1, 0x11); /* 3f/7f RT_FORMAT */ + xf_emit(ctx, 7, 0); /* 3f/7f RT_FORMAT */ + xf_emit(ctx, 1, 0x0fac6881); /* 0fffffff RT_CONTROL */ + xf_emit(ctx, 1, 0); /* 00000001 LOGIC_OP_ENABLE */ + xf_emit(ctx, 1, 0); /* ff/3ff */ + xf_emit(ctx, 1, 4); /* 00000007 FP_CONTROL */ + xf_emit(ctx, 1, 0); /* 00000003 UNK0F90 */ + xf_emit(ctx, 1, 0); /* 00000001 FRAMEBUFFER_SRGB */ + xf_emit(ctx, 1, 0); /* 7 */ + xf_emit(ctx, 1, 0x11); /* 3f/7f DST_FORMAT */ + xf_emit(ctx, 1, 1); /* 00000001 DST_LINEAR */ + xf_emit(ctx, 1, 0); /* 00000007 OPERATION */ + xf_emit(ctx, 1, 0xcf); /* 000000ff SIFC_FORMAT */ + xf_emit(ctx, 1, 0xcf); /* 000000ff DRAW_COLOR_FORMAT */ + xf_emit(ctx, 1, 0xcf); /* 000000ff SRC_FORMAT */ + if (IS_NVA3F(dev_priv->chipset)) + xf_emit(ctx, 1, 1); /* 0000001f tesla UNK169C */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A3C */ + xf_emit(ctx, 1, 0); /* 7/f[NVA3] MULTISAMPLE_SAMPLES_LOG2 */ + xf_emit(ctx, 8, 0); /* 00000001 BLEND_ENABLE */ + xf_emit(ctx, 1, 1); /* 0000001f BLEND_FUNC_DST_ALPHA */ + xf_emit(ctx, 1, 1); /* 00000007 BLEND_EQUATION_ALPHA */ + xf_emit(ctx, 1, 2); /* 0000001f BLEND_FUNC_SRC_ALPHA */ + xf_emit(ctx, 1, 1); /* 0000001f BLEND_FUNC_DST_RGB */ + xf_emit(ctx, 1, 1); /* 00000007 BLEND_EQUATION_RGB */ + xf_emit(ctx, 1, 2); /* 0000001f BLEND_FUNC_SRC_RGB */ + xf_emit(ctx, 1, 1); /* 00000001 UNK133C */ + xf_emit(ctx, 1, 0); /* ffff0ff3 */ + xf_emit(ctx, 8, 1); /* 00000001 UNK19E0 */ + xf_emit(ctx, 1, 0x11); /* 3f/7f RT_FORMAT */ + xf_emit(ctx, 7, 0); /* 3f/7f RT_FORMAT */ + xf_emit(ctx, 1, 0x0fac6881); /* 0fffffff RT_CONTROL */ + xf_emit(ctx, 1, 0xf); /* 0000000f COLOR_MASK */ + xf_emit(ctx, 7, 0); /* 0000000f COLOR_MASK */ + xf_emit(ctx, 1, magic2); /* 001fffff tesla UNK0F78 */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_BOUNDS_EN */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_TEST_ENABLE */ + xf_emit(ctx, 1, 0x11); /* 3f/7f DST_FORMAT */ + xf_emit(ctx, 1, 1); /* 00000001 DST_LINEAR */ + if (IS_NVA3F(dev_priv->chipset)) + xf_emit(ctx, 1, 1); /* 0000001f tesla UNK169C */ + if (dev_priv->chipset == 0x50) + xf_emit(ctx, 1, 0); /* ff */ + else + xf_emit(ctx, 3, 0); /* 1, 7, 3ff */ + xf_emit(ctx, 1, 4); /* 00000007 FP_CONTROL */ + xf_emit(ctx, 1, 0); /* 00000003 UNK0F90 */ + xf_emit(ctx, 1, 0); /* 00000001 STENCIL_FRONT_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000007 */ + xf_emit(ctx, 1, 0); /* 00000001 SAMPLECNT_ENABLE */ + xf_emit(ctx, 1, 0); /* 0000000f ZETA_FORMAT */ + xf_emit(ctx, 1, 1); /* 00000001 ZETA_ENABLE */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A3C */ + xf_emit(ctx, 1, 0); /* 7/f MULTISAMPLE_SAMPLES_LOG2 */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK1534 */ + xf_emit(ctx, 1, 0); /* ffff0ff3 */ + xf_emit(ctx, 1, 0x11); /* 3f/7f RT_FORMAT */ + xf_emit(ctx, 7, 0); /* 3f/7f RT_FORMAT */ + xf_emit(ctx, 1, 0x0fac6881); /* 0fffffff RT_CONTROL */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_BOUNDS_EN */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_TEST_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_WRITE_ENABLE */ + xf_emit(ctx, 1, 0x11); /* 3f/7f DST_FORMAT */ + xf_emit(ctx, 1, 1); /* 00000001 DST_LINEAR */ + xf_emit(ctx, 1, 0); /* 000fffff BLIT_DU_DX_FRACT */ + xf_emit(ctx, 1, 1); /* 0001ffff BLIT_DU_DX_INT */ + xf_emit(ctx, 1, 0); /* 000fffff BLIT_DV_DY_FRACT */ + xf_emit(ctx, 1, 1); /* 0001ffff BLIT_DV_DY_INT */ + xf_emit(ctx, 1, 0); /* ff/3ff */ + xf_emit(ctx, 1, magic1); /* 3ff/7ff tesla UNK0D68 */ + xf_emit(ctx, 1, 0); /* 00000001 STENCIL_FRONT_ENABLE */ + xf_emit(ctx, 1, 1); /* 00000001 tesla UNK15B4 */ + xf_emit(ctx, 1, 0); /* 0000000f ZETA_FORMAT */ + xf_emit(ctx, 1, 1); /* 00000001 ZETA_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000007 */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A3C */ + if (IS_NVA3F(dev_priv->chipset)) + xf_emit(ctx, 1, 1); /* 0000001f tesla UNK169C */ + xf_emit(ctx, 8, 0); /* 0000ffff DMA_COLOR */ + xf_emit(ctx, 1, 0); /* 0000ffff DMA_GLOBAL */ + xf_emit(ctx, 1, 0); /* 0000ffff DMA_LOCAL */ + xf_emit(ctx, 1, 0); /* 0000ffff DMA_STACK */ + xf_emit(ctx, 1, 0); /* ff/3ff */ + xf_emit(ctx, 1, 0); /* 0000ffff DMA_DST */ + xf_emit(ctx, 1, 0); /* 7 */ + xf_emit(ctx, 1, 0); /* 7/f MULTISAMPLE_SAMPLES_LOG2 */ + xf_emit(ctx, 1, 0); /* ffff0ff3 */ + xf_emit(ctx, 8, 0); /* 000000ff RT_ADDRESS_HIGH */ + xf_emit(ctx, 8, 0); /* ffffffff RT_LAYER_STRIDE */ + xf_emit(ctx, 8, 0); /* ffffffff RT_ADDRESS_LOW */ + xf_emit(ctx, 8, 8); /* 0000007f RT_TILE_MODE */ + xf_emit(ctx, 1, 0x11); /* 3f/7f RT_FORMAT */ + xf_emit(ctx, 7, 0); /* 3f/7f RT_FORMAT */ + xf_emit(ctx, 1, 0x0fac6881); /* 0fffffff RT_CONTROL */ + xf_emit(ctx, 8, 0x400); /* 0fffffff RT_HORIZ */ + xf_emit(ctx, 8, 0x300); /* 0000ffff RT_VERT */ + xf_emit(ctx, 1, 1); /* 00001fff RT_ARRAY_MODE */ + xf_emit(ctx, 1, 0xf); /* 0000000f COLOR_MASK */ + xf_emit(ctx, 7, 0); /* 0000000f COLOR_MASK */ + xf_emit(ctx, 1, 0x20); /* 00000fff DST_TILE_MODE */ + xf_emit(ctx, 1, 0x11); /* 3f/7f DST_FORMAT */ + xf_emit(ctx, 1, 0x100); /* 0001ffff DST_HEIGHT */ + xf_emit(ctx, 1, 0); /* 000007ff DST_LAYER */ + xf_emit(ctx, 1, 1); /* 00000001 DST_LINEAR */ + xf_emit(ctx, 1, 0); /* ffffffff DST_ADDRESS_LOW */ + xf_emit(ctx, 1, 0); /* 000000ff DST_ADDRESS_HIGH */ + xf_emit(ctx, 1, 0x40); /* 0007ffff DST_PITCH */ + xf_emit(ctx, 1, 0x100); /* 0001ffff DST_WIDTH */ + xf_emit(ctx, 1, 0); /* 0000ffff */ + xf_emit(ctx, 1, 3); /* 00000003 tesla UNK15AC */ + xf_emit(ctx, 1, 0); /* ff/3ff */ + xf_emit(ctx, 1, 0); /* 0001ffff GP_BUILTIN_RESULT_EN */ + xf_emit(ctx, 1, 0); /* 00000003 UNK0F90 */ + xf_emit(ctx, 1, 0); /* 00000007 */ + if (IS_NVA3F(dev_priv->chipset)) + xf_emit(ctx, 1, 1); /* 0000001f tesla UNK169C */ + xf_emit(ctx, 1, magic2); /* 001fffff tesla UNK0F78 */ + xf_emit(ctx, 1, 0); /* 7/f MULTISAMPLE_SAMPLES_LOG2 */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK1534 */ + xf_emit(ctx, 1, 0); /* ffff0ff3 */ + xf_emit(ctx, 1, 2); /* 00000003 tesla UNK143C */ + xf_emit(ctx, 1, 0x0fac6881); /* 0fffffff RT_CONTROL */ + xf_emit(ctx, 1, 0); /* 0000ffff DMA_ZETA */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_BOUNDS_EN */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_TEST_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_WRITE_ENABLE */ + xf_emit(ctx, 2, 0); /* ffff, ff/3ff */ + xf_emit(ctx, 1, 0); /* 0001ffff GP_BUILTIN_RESULT_EN */ + xf_emit(ctx, 1, 0); /* 00000001 STENCIL_FRONT_ENABLE */ + xf_emit(ctx, 1, 0); /* 000000ff STENCIL_FRONT_MASK */ + xf_emit(ctx, 1, 1); /* 00000001 tesla UNK15B4 */ + xf_emit(ctx, 1, 0); /* 00000007 */ + xf_emit(ctx, 1, 0); /* ffffffff ZETA_LAYER_STRIDE */ + xf_emit(ctx, 1, 0); /* 000000ff ZETA_ADDRESS_HIGH */ + xf_emit(ctx, 1, 0); /* ffffffff ZETA_ADDRESS_LOW */ + xf_emit(ctx, 1, 4); /* 00000007 ZETA_TILE_MODE */ + xf_emit(ctx, 1, 0); /* 0000000f ZETA_FORMAT */ + xf_emit(ctx, 1, 1); /* 00000001 ZETA_ENABLE */ + xf_emit(ctx, 1, 0x400); /* 0fffffff ZETA_HORIZ */ + xf_emit(ctx, 1, 0x300); /* 0000ffff ZETA_VERT */ + xf_emit(ctx, 1, 0x1001); /* 00001fff ZETA_ARRAY_MODE */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A3C */ + xf_emit(ctx, 1, 0); /* 7/f MULTISAMPLE_SAMPLES_LOG2 */ + if (IS_NVA3F(dev_priv->chipset)) + xf_emit(ctx, 1, 0); /* 00000001 */ + xf_emit(ctx, 1, 0); /* ffff0ff3 */ + xf_emit(ctx, 1, 0x11); /* 3f/7f RT_FORMAT */ + xf_emit(ctx, 7, 0); /* 3f/7f RT_FORMAT */ + xf_emit(ctx, 1, 0x0fac6881); /* 0fffffff RT_CONTROL */ + xf_emit(ctx, 1, 0xf); /* 0000000f COLOR_MASK */ + xf_emit(ctx, 7, 0); /* 0000000f COLOR_MASK */ + xf_emit(ctx, 1, 0); /* ff/3ff */ + xf_emit(ctx, 8, 0); /* 00000001 BLEND_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000003 UNK0F90 */ + xf_emit(ctx, 1, 0); /* 00000001 FRAMEBUFFER_SRGB */ + xf_emit(ctx, 1, 0); /* 7 */ + xf_emit(ctx, 1, 0); /* 00000001 LOGIC_OP_ENABLE */ + if (IS_NVA3F(dev_priv->chipset)) { + xf_emit(ctx, 1, 0); /* 00000001 UNK1140 */ + xf_emit(ctx, 1, 1); /* 0000001f tesla UNK169C */ + } + xf_emit(ctx, 1, 0); /* 7/f MULTISAMPLE_SAMPLES_LOG2 */ + xf_emit(ctx, 1, 0); /* 00000001 UNK1534 */ + xf_emit(ctx, 1, 0); /* ffff0ff3 */ + if (dev_priv->chipset >= 0xa0) + xf_emit(ctx, 1, 0x0fac6881); /* fffffff */ + xf_emit(ctx, 1, magic2); /* 001fffff tesla UNK0F78 */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_BOUNDS_EN */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_TEST_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_WRITE_ENABLE */ + xf_emit(ctx, 1, 0x11); /* 3f/7f DST_FORMAT */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK0FB0 */ + xf_emit(ctx, 1, 0); /* ff/3ff */ + xf_emit(ctx, 1, 4); /* 00000007 FP_CONTROL */ + xf_emit(ctx, 1, 0); /* 00000001 STENCIL_FRONT_ENABLE */ + xf_emit(ctx, 1, 1); /* 00000001 tesla UNK15B4 */ + xf_emit(ctx, 1, 1); /* 00000001 tesla UNK19CC */ + xf_emit(ctx, 1, 0); /* 00000007 */ + xf_emit(ctx, 1, 0); /* 00000001 SAMPLECNT_ENABLE */ + xf_emit(ctx, 1, 0); /* 0000000f ZETA_FORMAT */ + xf_emit(ctx, 1, 1); /* 00000001 ZETA_ENABLE */ + if (IS_NVA3F(dev_priv->chipset)) { + xf_emit(ctx, 1, 1); /* 0000001f tesla UNK169C */ + xf_emit(ctx, 1, 0); /* 0000000f tesla UNK15C8 */ + } + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A3C */ + if (dev_priv->chipset >= 0xa0) { + xf_emit(ctx, 3, 0); /* 7/f, 1, ffff0ff3 */ + xf_emit(ctx, 1, 0xfac6881); /* fffffff */ + xf_emit(ctx, 4, 0); /* 1, 1, 1, 3ff */ + xf_emit(ctx, 1, 4); /* 7 */ + xf_emit(ctx, 1, 0); /* 1 */ + xf_emit(ctx, 2, 1); /* 1 */ + xf_emit(ctx, 2, 0); /* 7, f */ + xf_emit(ctx, 1, 1); /* 1 */ + xf_emit(ctx, 1, 0); /* 7/f */ + if (IS_NVA3F(dev_priv->chipset)) + xf_emit(ctx, 0x9, 0); /* 1 */ + else + xf_emit(ctx, 0x8, 0); /* 1 */ + xf_emit(ctx, 1, 0); /* ffff0ff3 */ + xf_emit(ctx, 8, 1); /* 1 */ + xf_emit(ctx, 1, 0x11); /* 7f */ + xf_emit(ctx, 7, 0); /* 7f */ + xf_emit(ctx, 1, 0xfac6881); /* fffffff */ + xf_emit(ctx, 1, 0xf); /* f */ + xf_emit(ctx, 7, 0); /* f */ + xf_emit(ctx, 1, 0x11); /* 7f */ + xf_emit(ctx, 1, 1); /* 1 */ + xf_emit(ctx, 5, 0); /* 1, 7, 3ff, 3, 7 */ + if (IS_NVA3F(dev_priv->chipset)) { + xf_emit(ctx, 1, 0); /* 00000001 UNK1140 */ + xf_emit(ctx, 1, 1); /* 0000001f tesla UNK169C */ + } + } +} + +static void +nv50_graph_construct_xfer_tex(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + xf_emit(ctx, 2, 0); /* 1 LINKED_TSC. yes, 2. */ + if (dev_priv->chipset != 0x50) + xf_emit(ctx, 1, 0); /* 3 */ + xf_emit(ctx, 1, 1); /* 1ffff BLIT_DU_DX_INT */ + xf_emit(ctx, 1, 0); /* fffff BLIT_DU_DX_FRACT */ + xf_emit(ctx, 1, 1); /* 1ffff BLIT_DV_DY_INT */ + xf_emit(ctx, 1, 0); /* fffff BLIT_DV_DY_FRACT */ + if (dev_priv->chipset == 0x50) + xf_emit(ctx, 1, 0); /* 3 BLIT_CONTROL */ + else + xf_emit(ctx, 2, 0); /* 3ff, 1 */ + xf_emit(ctx, 1, 0x2a712488); /* ffffffff SRC_TIC_0 */ + xf_emit(ctx, 1, 0); /* ffffffff SRC_TIC_1 */ + xf_emit(ctx, 1, 0x4085c000); /* ffffffff SRC_TIC_2 */ + xf_emit(ctx, 1, 0x40); /* ffffffff SRC_TIC_3 */ + xf_emit(ctx, 1, 0x100); /* ffffffff SRC_TIC_4 */ + xf_emit(ctx, 1, 0x10100); /* ffffffff SRC_TIC_5 */ + xf_emit(ctx, 1, 0x02800000); /* ffffffff SRC_TIC_6 */ + xf_emit(ctx, 1, 0); /* ffffffff SRC_TIC_7 */ + if (dev_priv->chipset == 0x50) { + xf_emit(ctx, 1, 0); /* 00000001 turing UNK358 */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A34? */ + xf_emit(ctx, 1, 0); /* 00000003 turing UNK37C tesla UNK1690 */ + xf_emit(ctx, 1, 0); /* 00000003 BLIT_CONTROL */ + xf_emit(ctx, 1, 0); /* 00000001 turing UNK32C tesla UNK0F94 */ + } else if (!IS_NVAAF(dev_priv->chipset)) { + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A34? */ + xf_emit(ctx, 1, 0); /* 00000003 */ + xf_emit(ctx, 1, 0); /* 000003ff */ + xf_emit(ctx, 1, 0); /* 00000003 */ + xf_emit(ctx, 1, 0); /* 000003ff */ + xf_emit(ctx, 1, 0); /* 00000003 tesla UNK1664 / turing UNK03E8 */ + xf_emit(ctx, 1, 0); /* 00000003 */ + xf_emit(ctx, 1, 0); /* 000003ff */ + } else { + xf_emit(ctx, 0x6, 0); + } + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A34 */ + xf_emit(ctx, 1, 0); /* 0000ffff DMA_TEXTURE */ + xf_emit(ctx, 1, 0); /* 0000ffff DMA_SRC */ +} + +static void +nv50_graph_construct_xfer_unk8cxx(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + xf_emit(ctx, 1, 0); /* 00000001 UNK1534 */ + xf_emit(ctx, 1, 0); /* 7/f MULTISAMPLE_SAMPLES_LOG2 */ + xf_emit(ctx, 2, 0); /* 7, ffff0ff3 */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_TEST_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_WRITE */ + xf_emit(ctx, 1, 0x04e3bfdf); /* ffffffff UNK0D64 */ + xf_emit(ctx, 1, 0x04e3bfdf); /* ffffffff UNK0DF4 */ + xf_emit(ctx, 1, 1); /* 00000001 UNK15B4 */ + xf_emit(ctx, 1, 0); /* 00000001 LINE_STIPPLE_ENABLE */ + xf_emit(ctx, 1, 0x00ffff00); /* 00ffffff LINE_STIPPLE_PATTERN */ + xf_emit(ctx, 1, 1); /* 00000001 tesla UNK0F98 */ + if (IS_NVA3F(dev_priv->chipset)) + xf_emit(ctx, 1, 1); /* 0000001f tesla UNK169C */ + xf_emit(ctx, 1, 0); /* 00000003 tesla UNK1668 */ + xf_emit(ctx, 1, 0); /* 00000001 LINE_STIPPLE_ENABLE */ + xf_emit(ctx, 1, 0x00ffff00); /* 00ffffff LINE_STIPPLE_PATTERN */ + xf_emit(ctx, 1, 0); /* 00000001 POLYGON_SMOOTH_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000001 UNK1534 */ + xf_emit(ctx, 1, 0); /* 7/f MULTISAMPLE_SAMPLES_LOG2 */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK1658 */ + xf_emit(ctx, 1, 0); /* 00000001 LINE_SMOOTH_ENABLE */ + xf_emit(ctx, 1, 0); /* ffff0ff3 */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_TEST_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000001 DEPTH_WRITE */ + xf_emit(ctx, 1, 1); /* 00000001 UNK15B4 */ + xf_emit(ctx, 1, 0); /* 00000001 POINT_SPRITE_ENABLE */ + xf_emit(ctx, 1, 1); /* 00000001 tesla UNK165C */ + xf_emit(ctx, 1, 0x30201000); /* ffffffff tesla UNK1670 */ + xf_emit(ctx, 1, 0x70605040); /* ffffffff tesla UNK1670 */ + xf_emit(ctx, 1, 0xb8a89888); /* ffffffff tesla UNK1670 */ + xf_emit(ctx, 1, 0xf8e8d8c8); /* ffffffff tesla UNK1670 */ + xf_emit(ctx, 1, 0); /* 00000001 VERTEX_TWO_SIDE_ENABLE */ + xf_emit(ctx, 1, 0x1a); /* 0000001f POLYGON_MODE */ +} + +static void +nv50_graph_construct_xfer_tp(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + if (dev_priv->chipset < 0xa0) { + nv50_graph_construct_xfer_unk84xx(ctx); + nv50_graph_construct_xfer_tprop(ctx); + nv50_graph_construct_xfer_tex(ctx); + nv50_graph_construct_xfer_unk8cxx(ctx); + } else { + nv50_graph_construct_xfer_tex(ctx); + nv50_graph_construct_xfer_tprop(ctx); + nv50_graph_construct_xfer_unk8cxx(ctx); + nv50_graph_construct_xfer_unk84xx(ctx); + } +} + +static void +nv50_graph_construct_xfer_mpc(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + int i, mpcnt = 2; + switch (dev_priv->chipset) { + case 0x98: + case 0xaa: + mpcnt = 1; + break; + case 0x50: + case 0x84: + case 0x86: + case 0x92: + case 0x94: + case 0x96: + case 0xa8: + case 0xac: + mpcnt = 2; + break; + case 0xa0: + case 0xa3: + case 0xa5: + case 0xaf: + mpcnt = 3; + break; + } + for (i = 0; i < mpcnt; i++) { + xf_emit(ctx, 1, 0); /* ff */ + xf_emit(ctx, 1, 0x80); /* ffffffff tesla UNK1404 */ + xf_emit(ctx, 1, 0x80007004); /* ffffffff tesla UNK12B0 */ + xf_emit(ctx, 1, 0x04000400); /* ffffffff */ + if (dev_priv->chipset >= 0xa0) + xf_emit(ctx, 1, 0xc0); /* 00007fff tesla UNK152C */ + xf_emit(ctx, 1, 0x1000); /* 0000ffff tesla UNK0D60 */ + xf_emit(ctx, 1, 0); /* ff/3ff */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A30 */ + if (dev_priv->chipset == 0x86 || dev_priv->chipset == 0x98 || dev_priv->chipset == 0xa8 || IS_NVAAF(dev_priv->chipset)) { + xf_emit(ctx, 1, 0xe00); /* 7fff */ + xf_emit(ctx, 1, 0x1e00); /* 7fff */ + } + xf_emit(ctx, 1, 1); /* 000000ff VP_REG_ALLOC_TEMP */ + xf_emit(ctx, 1, 0); /* 00000001 LINKED_TSC */ + xf_emit(ctx, 1, 0); /* 00000001 GP_ENABLE */ + if (dev_priv->chipset == 0x50) + xf_emit(ctx, 2, 0x1000); /* 7fff tesla UNK141C */ + xf_emit(ctx, 1, 1); /* 000000ff GP_REG_ALLOC_TEMP */ + xf_emit(ctx, 1, 0); /* 00000001 GP_ENABLE */ + xf_emit(ctx, 1, 4); /* 000000ff FP_REG_ALLOC_TEMP */ + xf_emit(ctx, 1, 2); /* 00000003 REG_MODE */ + if (IS_NVAAF(dev_priv->chipset)) + xf_emit(ctx, 0xb, 0); /* RO */ + else if (dev_priv->chipset >= 0xa0) + xf_emit(ctx, 0xc, 0); /* RO */ + else + xf_emit(ctx, 0xa, 0); /* RO */ + } + xf_emit(ctx, 1, 0x08100c12); /* 1fffffff FP_INTERPOLANT_CTRL */ + xf_emit(ctx, 1, 0); /* ff/3ff */ + if (dev_priv->chipset >= 0xa0) { + xf_emit(ctx, 1, 0x1fe21); /* 0003ffff tesla UNK0FAC */ + } + xf_emit(ctx, 3, 0); /* 7fff, 0, 0 */ + xf_emit(ctx, 1, 0); /* 00000001 tesla UNK1534 */ + xf_emit(ctx, 1, 0); /* 7/f MULTISAMPLE_SAMPLES_LOG2 */ + xf_emit(ctx, 4, 0xffff); /* 0000ffff MSAA_MASK */ + xf_emit(ctx, 1, 1); /* 00000001 LANES32 */ + xf_emit(ctx, 1, 0x10001); /* 00ffffff BLOCK_ALLOC */ + xf_emit(ctx, 1, 0x10001); /* ffffffff BLOCKDIM_XY */ + xf_emit(ctx, 1, 1); /* 0000ffff BLOCKDIM_Z */ + xf_emit(ctx, 1, 0); /* ffffffff SHARED_SIZE */ + xf_emit(ctx, 1, 0x1fe21); /* 1ffff/3ffff[NVA0+] tesla UNk0FAC */ + xf_emit(ctx, 1, 0); /* ffffffff tesla UNK1A34 */ + if (IS_NVA3F(dev_priv->chipset)) + xf_emit(ctx, 1, 1); /* 0000001f tesla UNK169C */ + xf_emit(ctx, 1, 0); /* ff/3ff */ + xf_emit(ctx, 1, 0); /* 1 LINKED_TSC */ + xf_emit(ctx, 1, 0); /* ff FP_ADDRESS_HIGH */ + xf_emit(ctx, 1, 0); /* ffffffff FP_ADDRESS_LOW */ + xf_emit(ctx, 1, 0x08100c12); /* 1fffffff FP_INTERPOLANT_CTRL */ + xf_emit(ctx, 1, 4); /* 00000007 FP_CONTROL */ + xf_emit(ctx, 1, 0); /* 000000ff FRAG_COLOR_CLAMP_EN */ + xf_emit(ctx, 1, 2); /* 00000003 REG_MODE */ + xf_emit(ctx, 1, 0x11); /* 0000007f RT_FORMAT */ + xf_emit(ctx, 7, 0); /* 0000007f RT_FORMAT */ + xf_emit(ctx, 1, 0); /* 00000007 */ + xf_emit(ctx, 1, 0xfac6881); /* 0fffffff RT_CONTROL */ + xf_emit(ctx, 1, 0); /* 00000003 MULTISAMPLE_CTRL */ + if (IS_NVA3F(dev_priv->chipset)) + xf_emit(ctx, 1, 3); /* 00000003 tesla UNK16B4 */ + xf_emit(ctx, 1, 0); /* 00000001 ALPHA_TEST_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000007 ALPHA_TEST_FUNC */ + xf_emit(ctx, 1, 0); /* 00000001 FRAMEBUFFER_SRGB */ + xf_emit(ctx, 1, 4); /* ffffffff tesla UNK1400 */ + xf_emit(ctx, 8, 0); /* 00000001 BLEND_ENABLE */ + xf_emit(ctx, 1, 0); /* 00000001 LOGIC_OP_ENABLE */ + xf_emit(ctx, 1, 2); /* 0000001f BLEND_FUNC_SRC_RGB */ + xf_emit(ctx, 1, 1); /* 0000001f BLEND_FUNC_DST_RGB */ + xf_emit(ctx, 1, 1); /* 00000007 BLEND_EQUATION_RGB */ + xf_emit(ctx, 1, 2); /* 0000001f BLEND_FUNC_SRC_ALPHA */ + xf_emit(ctx, 1, 1); /* 0000001f BLEND_FUNC_DST_ALPHA */ + xf_emit(ctx, 1, 1); /* 00000007 BLEND_EQUATION_ALPHA */ + xf_emit(ctx, 1, 1); /* 00000001 UNK133C */ + if (IS_NVA3F(dev_priv->chipset)) { + xf_emit(ctx, 1, 0); /* 00000001 UNK12E4 */ + xf_emit(ctx, 8, 2); /* 0000001f IBLEND_FUNC_SRC_RGB */ + xf_emit(ctx, 8, 1); /* 0000001f IBLEND_FUNC_DST_RGB */ + xf_emit(ctx, 8, 1); /* 00000007 IBLEND_EQUATION_RGB */ + xf_emit(ctx, 8, 2); /* 0000001f IBLEND_FUNC_SRC_ALPHA */ + xf_emit(ctx, 8, 1); /* 0000001f IBLEND_FUNC_DST_ALPHA */ + xf_emit(ctx, 8, 1); /* 00000007 IBLEND_EQUATION_ALPHA */ + xf_emit(ctx, 8, 1); /* 00000001 IBLEND_UNK00 */ + xf_emit(ctx, 1, 0); /* 00000003 tesla UNK1928 */ + xf_emit(ctx, 1, 0); /* 00000001 UNK1140 */ + } + xf_emit(ctx, 1, 0); /* 00000003 tesla UNK0F90 */ + xf_emit(ctx, 1, 4); /* 000000ff FP_RESULT_COUNT */ + /* XXX: demagic this part some day */ + if (dev_priv->chipset == 0x50) + xf_emit(ctx, 0x3a0, 0); + else if (dev_priv->chipset < 0x94) + xf_emit(ctx, 0x3a2, 0); + else if (dev_priv->chipset == 0x98 || dev_priv->chipset == 0xaa) + xf_emit(ctx, 0x39f, 0); + else + xf_emit(ctx, 0x3a3, 0); + xf_emit(ctx, 1, 0x11); /* 3f/7f DST_FORMAT */ + xf_emit(ctx, 1, 0); /* 7 OPERATION */ + xf_emit(ctx, 1, 1); /* 1 DST_LINEAR */ + xf_emit(ctx, 0x2d, 0); +} + +static void +nv50_graph_construct_xfer2(struct nouveau_grctx *ctx) +{ + struct drm_nouveau_private *dev_priv = ctx->dev->dev_private; + int i; + uint32_t offset; + uint32_t units = nv_rd32 (ctx->dev, 0x1540); + int size = 0; + + offset = (ctx->ctxvals_pos+0x3f)&~0x3f; + + if (dev_priv->chipset < 0xa0) { + for (i = 0; i < 8; i++) { + ctx->ctxvals_pos = offset + i; + /* that little bugger belongs to csched. No idea + * what it's doing here. */ + if (i == 0) + xf_emit(ctx, 1, 0x08100c12); /* FP_INTERPOLANT_CTRL */ + if (units & (1 << i)) + nv50_graph_construct_xfer_mpc(ctx); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + } + } else { + /* Strand 0: TPs 0, 1 */ + ctx->ctxvals_pos = offset; + /* that little bugger belongs to csched. No idea + * what it's doing here. */ + xf_emit(ctx, 1, 0x08100c12); /* FP_INTERPOLANT_CTRL */ + if (units & (1 << 0)) + nv50_graph_construct_xfer_mpc(ctx); + if (units & (1 << 1)) + nv50_graph_construct_xfer_mpc(ctx); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + + /* Strand 1: TPs 2, 3 */ + ctx->ctxvals_pos = offset + 1; + if (units & (1 << 2)) + nv50_graph_construct_xfer_mpc(ctx); + if (units & (1 << 3)) + nv50_graph_construct_xfer_mpc(ctx); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + + /* Strand 2: TPs 4, 5, 6 */ + ctx->ctxvals_pos = offset + 2; + if (units & (1 << 4)) + nv50_graph_construct_xfer_mpc(ctx); + if (units & (1 << 5)) + nv50_graph_construct_xfer_mpc(ctx); + if (units & (1 << 6)) + nv50_graph_construct_xfer_mpc(ctx); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + + /* Strand 3: TPs 7, 8, 9 */ + ctx->ctxvals_pos = offset + 3; + if (units & (1 << 7)) + nv50_graph_construct_xfer_mpc(ctx); + if (units & (1 << 8)) + nv50_graph_construct_xfer_mpc(ctx); + if (units & (1 << 9)) + nv50_graph_construct_xfer_mpc(ctx); + if ((ctx->ctxvals_pos-offset)/8 > size) + size = (ctx->ctxvals_pos-offset)/8; + } + ctx->ctxvals_pos = offset + size * 8; + ctx->ctxvals_pos = (ctx->ctxvals_pos+0x3f)&~0x3f; + cp_lsr (ctx, offset); + cp_out (ctx, CP_SET_XFER_POINTER); + cp_lsr (ctx, size); + cp_out (ctx, CP_SEEK_2); + cp_out (ctx, CP_XFER_2); + cp_wait(ctx, XFER, BUSY); +} diff --git a/drivers/gpu/drm/nouveau/nv50_instmem.c b/drivers/gpu/drm/nouveau/nv50_instmem.c new file mode 100644 index 00000000..4f95a1e5 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv50_instmem.c @@ -0,0 +1,428 @@ +/* + * Copyright (C) 2007 Ben Skeggs. + * + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "drmP.h" +#include "drm.h" + +#include "nouveau_drv.h" +#include "nouveau_vm.h" + +#define BAR1_VM_BASE 0x0020000000ULL +#define BAR1_VM_SIZE pci_resource_len(dev->pdev, 1) +#define BAR3_VM_BASE 0x0000000000ULL +#define BAR3_VM_SIZE pci_resource_len(dev->pdev, 3) + +struct nv50_instmem_priv { + uint32_t save1700[5]; /* 0x1700->0x1710 */ + + struct nouveau_gpuobj *bar1_dmaobj; + struct nouveau_gpuobj *bar3_dmaobj; +}; + +static void +nv50_channel_del(struct nouveau_channel **pchan) +{ + struct nouveau_channel *chan; + + chan = *pchan; + *pchan = NULL; + if (!chan) + return; + + nouveau_gpuobj_ref(NULL, &chan->ramfc); + nouveau_vm_ref(NULL, &chan->vm, chan->vm_pd); + nouveau_gpuobj_ref(NULL, &chan->vm_pd); + if (drm_mm_initialized(&chan->ramin_heap)) + drm_mm_takedown(&chan->ramin_heap); + nouveau_gpuobj_ref(NULL, &chan->ramin); + kfree(chan); +} + +static int +nv50_channel_new(struct drm_device *dev, u32 size, struct nouveau_vm *vm, + struct nouveau_channel **pchan) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + u32 pgd = (dev_priv->chipset == 0x50) ? 0x1400 : 0x0200; + u32 fc = (dev_priv->chipset == 0x50) ? 0x0000 : 0x4200; + struct nouveau_channel *chan; + int ret, i; + + chan = kzalloc(sizeof(*chan), GFP_KERNEL); + if (!chan) + return -ENOMEM; + chan->dev = dev; + + ret = nouveau_gpuobj_new(dev, NULL, size, 0x1000, 0, &chan->ramin); + if (ret) { + nv50_channel_del(&chan); + return ret; + } + + ret = drm_mm_init(&chan->ramin_heap, 0x6000, chan->ramin->size); + if (ret) { + nv50_channel_del(&chan); + return ret; + } + + ret = nouveau_gpuobj_new_fake(dev, chan->ramin->pinst == ~0 ? ~0 : + chan->ramin->pinst + pgd, + chan->ramin->vinst + pgd, + 0x4000, NVOBJ_FLAG_ZERO_ALLOC, + &chan->vm_pd); + if (ret) { + nv50_channel_del(&chan); + return ret; + } + + for (i = 0; i < 0x4000; i += 8) { + nv_wo32(chan->vm_pd, i + 0, 0x00000000); + nv_wo32(chan->vm_pd, i + 4, 0xdeadcafe); + } + + ret = nouveau_vm_ref(vm, &chan->vm, chan->vm_pd); + if (ret) { + nv50_channel_del(&chan); + return ret; + } + + ret = nouveau_gpuobj_new_fake(dev, chan->ramin->pinst == ~0 ? ~0 : + chan->ramin->pinst + fc, + chan->ramin->vinst + fc, 0x100, + NVOBJ_FLAG_ZERO_ALLOC, &chan->ramfc); + if (ret) { + nv50_channel_del(&chan); + return ret; + } + + *pchan = chan; + return 0; +} + +int +nv50_instmem_init(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nv50_instmem_priv *priv; + struct nouveau_channel *chan; + struct nouveau_vm *vm; + int ret, i; + u32 tmp; + + priv = kzalloc(sizeof(*priv), GFP_KERNEL); + if (!priv) + return -ENOMEM; + dev_priv->engine.instmem.priv = priv; + + /* Save state, will restore at takedown. */ + for (i = 0x1700; i <= 0x1710; i += 4) + priv->save1700[(i-0x1700)/4] = nv_rd32(dev, i); + + /* Global PRAMIN heap */ + ret = drm_mm_init(&dev_priv->ramin_heap, 0, dev_priv->ramin_size); + if (ret) { + NV_ERROR(dev, "Failed to init RAMIN heap\n"); + goto error; + } + + /* BAR3 */ + ret = nouveau_vm_new(dev, BAR3_VM_BASE, BAR3_VM_SIZE, BAR3_VM_BASE, + &dev_priv->bar3_vm); + if (ret) + goto error; + + ret = nouveau_gpuobj_new(dev, NULL, (BAR3_VM_SIZE >> 12) * 8, + 0x1000, NVOBJ_FLAG_DONT_MAP | + NVOBJ_FLAG_ZERO_ALLOC, + &dev_priv->bar3_vm->pgt[0].obj[0]); + if (ret) + goto error; + dev_priv->bar3_vm->pgt[0].refcount[0] = 1; + + nv50_instmem_map(dev_priv->bar3_vm->pgt[0].obj[0]); + + ret = nv50_channel_new(dev, 128 * 1024, dev_priv->bar3_vm, &chan); + if (ret) + goto error; + dev_priv->channels.ptr[0] = dev_priv->channels.ptr[127] = chan; + + ret = nv50_gpuobj_dma_new(chan, 0x0000, BAR3_VM_BASE, BAR3_VM_SIZE, + NV_MEM_TARGET_VM, NV_MEM_ACCESS_VM, + NV_MEM_TYPE_VM, NV_MEM_COMP_VM, + &priv->bar3_dmaobj); + if (ret) + goto error; + + nv_wr32(dev, 0x001704, 0x00000000 | (chan->ramin->vinst >> 12)); + nv_wr32(dev, 0x001704, 0x40000000 | (chan->ramin->vinst >> 12)); + nv_wr32(dev, 0x00170c, 0x80000000 | (priv->bar3_dmaobj->cinst >> 4)); + + dev_priv->engine.instmem.flush(dev); + dev_priv->ramin_available = true; + + tmp = nv_ro32(chan->ramin, 0); + nv_wo32(chan->ramin, 0, ~tmp); + if (nv_ro32(chan->ramin, 0) != ~tmp) { + NV_ERROR(dev, "PRAMIN readback failed\n"); + ret = -EIO; + goto error; + } + nv_wo32(chan->ramin, 0, tmp); + + /* BAR1 */ + ret = nouveau_vm_new(dev, BAR1_VM_BASE, BAR1_VM_SIZE, BAR1_VM_BASE, &vm); + if (ret) + goto error; + + ret = nouveau_vm_ref(vm, &dev_priv->bar1_vm, chan->vm_pd); + if (ret) + goto error; + nouveau_vm_ref(NULL, &vm, NULL); + + ret = nv50_gpuobj_dma_new(chan, 0x0000, BAR1_VM_BASE, BAR1_VM_SIZE, + NV_MEM_TARGET_VM, NV_MEM_ACCESS_VM, + NV_MEM_TYPE_VM, NV_MEM_COMP_VM, + &priv->bar1_dmaobj); + if (ret) + goto error; + + nv_wr32(dev, 0x001708, 0x80000000 | (priv->bar1_dmaobj->cinst >> 4)); + for (i = 0; i < 8; i++) + nv_wr32(dev, 0x1900 + (i*4), 0); + + /* Create shared channel VM, space is reserved at the beginning + * to catch "NULL pointer" references + */ + ret = nouveau_vm_new(dev, 0, (1ULL << 40), 0x0020000000ULL, + &dev_priv->chan_vm); + if (ret) + return ret; + + return 0; + +error: + nv50_instmem_takedown(dev); + return ret; +} + +void +nv50_instmem_takedown(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nv50_instmem_priv *priv = dev_priv->engine.instmem.priv; + struct nouveau_channel *chan = dev_priv->channels.ptr[0]; + int i; + + NV_DEBUG(dev, "\n"); + + if (!priv) + return; + + dev_priv->ramin_available = false; + + nouveau_vm_ref(NULL, &dev_priv->chan_vm, NULL); + + for (i = 0x1700; i <= 0x1710; i += 4) + nv_wr32(dev, i, priv->save1700[(i - 0x1700) / 4]); + + nouveau_gpuobj_ref(NULL, &priv->bar3_dmaobj); + nouveau_gpuobj_ref(NULL, &priv->bar1_dmaobj); + + nouveau_vm_ref(NULL, &dev_priv->bar1_vm, chan->vm_pd); + dev_priv->channels.ptr[127] = 0; + nv50_channel_del(&dev_priv->channels.ptr[0]); + + nouveau_gpuobj_ref(NULL, &dev_priv->bar3_vm->pgt[0].obj[0]); + nouveau_vm_ref(NULL, &dev_priv->bar3_vm, NULL); + + if (drm_mm_initialized(&dev_priv->ramin_heap)) + drm_mm_takedown(&dev_priv->ramin_heap); + + dev_priv->engine.instmem.priv = NULL; + kfree(priv); +} + +int +nv50_instmem_suspend(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + dev_priv->ramin_available = false; + return 0; +} + +void +nv50_instmem_resume(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nv50_instmem_priv *priv = dev_priv->engine.instmem.priv; + struct nouveau_channel *chan = dev_priv->channels.ptr[0]; + int i; + + /* Poke the relevant regs, and pray it works :) */ + nv_wr32(dev, NV50_PUNK_BAR_CFG_BASE, (chan->ramin->vinst >> 12)); + nv_wr32(dev, NV50_PUNK_UNK1710, 0); + nv_wr32(dev, NV50_PUNK_BAR_CFG_BASE, (chan->ramin->vinst >> 12) | + NV50_PUNK_BAR_CFG_BASE_VALID); + nv_wr32(dev, NV50_PUNK_BAR1_CTXDMA, (priv->bar1_dmaobj->cinst >> 4) | + NV50_PUNK_BAR1_CTXDMA_VALID); + nv_wr32(dev, NV50_PUNK_BAR3_CTXDMA, (priv->bar3_dmaobj->cinst >> 4) | + NV50_PUNK_BAR3_CTXDMA_VALID); + + for (i = 0; i < 8; i++) + nv_wr32(dev, 0x1900 + (i*4), 0); + + dev_priv->ramin_available = true; +} + +struct nv50_gpuobj_node { + struct nouveau_mem *vram; + struct nouveau_vma chan_vma; + u32 align; +}; + + +int +nv50_instmem_get(struct nouveau_gpuobj *gpuobj, u32 size, u32 align) +{ + struct drm_device *dev = gpuobj->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_vram_engine *vram = &dev_priv->engine.vram; + struct nv50_gpuobj_node *node = NULL; + int ret; + + node = kzalloc(sizeof(*node), GFP_KERNEL); + if (!node) + return -ENOMEM; + node->align = align; + + size = (size + 4095) & ~4095; + align = max(align, (u32)4096); + + ret = vram->get(dev, size, align, 0, 0, &node->vram); + if (ret) { + kfree(node); + return ret; + } + + gpuobj->vinst = node->vram->offset; + + if (gpuobj->flags & NVOBJ_FLAG_VM) { + u32 flags = NV_MEM_ACCESS_RW; + if (!(gpuobj->flags & NVOBJ_FLAG_VM_USER)) + flags |= NV_MEM_ACCESS_SYS; + + ret = nouveau_vm_get(dev_priv->chan_vm, size, 12, flags, + &node->chan_vma); + if (ret) { + vram->put(dev, &node->vram); + kfree(node); + return ret; + } + + nouveau_vm_map(&node->chan_vma, node->vram); + gpuobj->vinst = node->chan_vma.offset; + } + + gpuobj->size = size; + gpuobj->node = node; + return 0; +} + +void +nv50_instmem_put(struct nouveau_gpuobj *gpuobj) +{ + struct drm_device *dev = gpuobj->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_vram_engine *vram = &dev_priv->engine.vram; + struct nv50_gpuobj_node *node; + + node = gpuobj->node; + gpuobj->node = NULL; + + if (node->chan_vma.node) { + nouveau_vm_unmap(&node->chan_vma); + nouveau_vm_put(&node->chan_vma); + } + vram->put(dev, &node->vram); + kfree(node); +} + +int +nv50_instmem_map(struct nouveau_gpuobj *gpuobj) +{ + struct drm_nouveau_private *dev_priv = gpuobj->dev->dev_private; + struct nv50_gpuobj_node *node = gpuobj->node; + int ret; + + ret = nouveau_vm_get(dev_priv->bar3_vm, gpuobj->size, 12, + NV_MEM_ACCESS_RW, &node->vram->bar_vma); + if (ret) + return ret; + + nouveau_vm_map(&node->vram->bar_vma, node->vram); + gpuobj->pinst = node->vram->bar_vma.offset; + return 0; +} + +void +nv50_instmem_unmap(struct nouveau_gpuobj *gpuobj) +{ + struct nv50_gpuobj_node *node = gpuobj->node; + + if (node->vram->bar_vma.node) { + nouveau_vm_unmap(&node->vram->bar_vma); + nouveau_vm_put(&node->vram->bar_vma); + } +} + +void +nv50_instmem_flush(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + unsigned long flags; + + spin_lock_irqsave(&dev_priv->vm_lock, flags); + nv_wr32(dev, 0x00330c, 0x00000001); + if (!nv_wait(dev, 0x00330c, 0x00000002, 0x00000000)) + NV_ERROR(dev, "PRAMIN flush timeout\n"); + spin_unlock_irqrestore(&dev_priv->vm_lock, flags); +} + +void +nv84_instmem_flush(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + unsigned long flags; + + spin_lock_irqsave(&dev_priv->vm_lock, flags); + nv_wr32(dev, 0x070000, 0x00000001); + if (!nv_wait(dev, 0x070000, 0x00000002, 0x00000000)) + NV_ERROR(dev, "PRAMIN flush timeout\n"); + spin_unlock_irqrestore(&dev_priv->vm_lock, flags); +} + diff --git a/drivers/gpu/drm/nouveau/nv50_mc.c b/drivers/gpu/drm/nouveau/nv50_mc.c new file mode 100644 index 00000000..e0a9c3fa --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv50_mc.c @@ -0,0 +1,40 @@ +/* + * Copyright (C) 2007 Ben Skeggs. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "drmP.h" +#include "drm.h" +#include "nouveau_drv.h" + +int +nv50_mc_init(struct drm_device *dev) +{ + nv_wr32(dev, NV03_PMC_ENABLE, 0xFFFFFFFF); + return 0; +} + +void nv50_mc_takedown(struct drm_device *dev) +{ +} diff --git a/drivers/gpu/drm/nouveau/nv50_mpeg.c b/drivers/gpu/drm/nouveau/nv50_mpeg.c new file mode 100644 index 00000000..1dc5913f --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv50_mpeg.c @@ -0,0 +1,256 @@ +/* + * Copyright 2011 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include "drmP.h" +#include "nouveau_drv.h" +#include "nouveau_ramht.h" + +struct nv50_mpeg_engine { + struct nouveau_exec_engine base; +}; + +static inline u32 +CTX_PTR(struct drm_device *dev, u32 offset) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + if (dev_priv->chipset == 0x50) + offset += 0x0260; + else + offset += 0x0060; + + return offset; +} + +static int +nv50_mpeg_context_new(struct nouveau_channel *chan, int engine) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *ramin = chan->ramin; + struct nouveau_gpuobj *ctx = NULL; + int ret; + + NV_DEBUG(dev, "ch%d\n", chan->id); + + ret = nouveau_gpuobj_new(dev, chan, 128 * 4, 0, NVOBJ_FLAG_ZERO_ALLOC | + NVOBJ_FLAG_ZERO_FREE, &ctx); + if (ret) + return ret; + + nv_wo32(ramin, CTX_PTR(dev, 0x00), 0x80190002); + nv_wo32(ramin, CTX_PTR(dev, 0x04), ctx->vinst + ctx->size - 1); + nv_wo32(ramin, CTX_PTR(dev, 0x08), ctx->vinst); + nv_wo32(ramin, CTX_PTR(dev, 0x0c), 0); + nv_wo32(ramin, CTX_PTR(dev, 0x10), 0); + nv_wo32(ramin, CTX_PTR(dev, 0x14), 0x00010000); + + nv_wo32(ctx, 0x70, 0x00801ec1); + nv_wo32(ctx, 0x7c, 0x0000037c); + dev_priv->engine.instmem.flush(dev); + + chan->engctx[engine] = ctx; + return 0; +} + +static void +nv50_mpeg_context_del(struct nouveau_channel *chan, int engine) +{ + struct drm_nouveau_private *dev_priv = chan->dev->dev_private; + struct nouveau_gpuobj *ctx = chan->engctx[engine]; + struct drm_device *dev = chan->dev; + unsigned long flags; + u32 inst, i; + + if (!chan->ramin) + return; + + inst = chan->ramin->vinst >> 12; + inst |= 0x80000000; + + spin_lock_irqsave(&dev_priv->context_switch_lock, flags); + nv_mask(dev, 0x00b32c, 0x00000001, 0x00000000); + if (nv_rd32(dev, 0x00b318) == inst) + nv_mask(dev, 0x00b318, 0x80000000, 0x00000000); + nv_mask(dev, 0x00b32c, 0x00000001, 0x00000001); + spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); + + for (i = 0x00; i <= 0x14; i += 4) + nv_wo32(chan->ramin, CTX_PTR(dev, i), 0x00000000); + nouveau_gpuobj_ref(NULL, &ctx); + chan->engctx[engine] = NULL; +} + +static int +nv50_mpeg_object_new(struct nouveau_channel *chan, int engine, + u32 handle, u16 class) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *obj = NULL; + int ret; + + ret = nouveau_gpuobj_new(dev, chan, 16, 16, NVOBJ_FLAG_ZERO_FREE, &obj); + if (ret) + return ret; + obj->engine = 2; + obj->class = class; + + nv_wo32(obj, 0x00, class); + nv_wo32(obj, 0x04, 0x00000000); + nv_wo32(obj, 0x08, 0x00000000); + nv_wo32(obj, 0x0c, 0x00000000); + dev_priv->engine.instmem.flush(dev); + + ret = nouveau_ramht_insert(chan, handle, obj); + nouveau_gpuobj_ref(NULL, &obj); + return ret; +} + +static void +nv50_mpeg_tlb_flush(struct drm_device *dev, int engine) +{ + nv50_vm_flush_engine(dev, 0x08); +} + +static int +nv50_mpeg_init(struct drm_device *dev, int engine) +{ + nv_wr32(dev, 0x00b32c, 0x00000000); + nv_wr32(dev, 0x00b314, 0x00000100); + nv_wr32(dev, 0x00b0e0, 0x0000001a); + + nv_wr32(dev, 0x00b220, 0x00000044); + nv_wr32(dev, 0x00b300, 0x00801ec1); + nv_wr32(dev, 0x00b390, 0x00000000); + nv_wr32(dev, 0x00b394, 0x00000000); + nv_wr32(dev, 0x00b398, 0x00000000); + nv_mask(dev, 0x00b32c, 0x00000001, 0x00000001); + + nv_wr32(dev, 0x00b100, 0xffffffff); + nv_wr32(dev, 0x00b140, 0xffffffff); + + if (!nv_wait(dev, 0x00b200, 0x00000001, 0x00000000)) { + NV_ERROR(dev, "PMPEG init: 0x%08x\n", nv_rd32(dev, 0x00b200)); + return -EBUSY; + } + + return 0; +} + +static int +nv50_mpeg_fini(struct drm_device *dev, int engine) +{ + /*XXX: context save for s/r */ + nv_mask(dev, 0x00b32c, 0x00000001, 0x00000000); + nv_wr32(dev, 0x00b140, 0x00000000); + return 0; +} + +static void +nv50_mpeg_isr(struct drm_device *dev) +{ + u32 stat = nv_rd32(dev, 0x00b100); + u32 type = nv_rd32(dev, 0x00b230); + u32 mthd = nv_rd32(dev, 0x00b234); + u32 data = nv_rd32(dev, 0x00b238); + u32 show = stat; + + if (stat & 0x01000000) { + /* happens on initial binding of the object */ + if (type == 0x00000020 && mthd == 0x0000) { + nv_wr32(dev, 0x00b308, 0x00000100); + show &= ~0x01000000; + } + } + + if (show && nouveau_ratelimit()) { + NV_INFO(dev, "PMPEG - 0x%08x 0x%08x 0x%08x 0x%08x\n", + stat, type, mthd, data); + } + + nv_wr32(dev, 0x00b100, stat); + nv_wr32(dev, 0x00b230, 0x00000001); + nv50_fb_vm_trap(dev, 1); +} + +static void +nv50_vpe_isr(struct drm_device *dev) +{ + if (nv_rd32(dev, 0x00b100)) + nv50_mpeg_isr(dev); + + if (nv_rd32(dev, 0x00b800)) { + u32 stat = nv_rd32(dev, 0x00b800); + NV_INFO(dev, "PMSRCH: 0x%08x\n", stat); + nv_wr32(dev, 0xb800, stat); + } +} + +static void +nv50_mpeg_destroy(struct drm_device *dev, int engine) +{ + struct nv50_mpeg_engine *pmpeg = nv_engine(dev, engine); + + nouveau_irq_unregister(dev, 0); + + NVOBJ_ENGINE_DEL(dev, MPEG); + kfree(pmpeg); +} + +int +nv50_mpeg_create(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nv50_mpeg_engine *pmpeg; + + pmpeg = kzalloc(sizeof(*pmpeg), GFP_KERNEL); + if (!pmpeg) + return -ENOMEM; + + pmpeg->base.destroy = nv50_mpeg_destroy; + pmpeg->base.init = nv50_mpeg_init; + pmpeg->base.fini = nv50_mpeg_fini; + pmpeg->base.context_new = nv50_mpeg_context_new; + pmpeg->base.context_del = nv50_mpeg_context_del; + pmpeg->base.object_new = nv50_mpeg_object_new; + pmpeg->base.tlb_flush = nv50_mpeg_tlb_flush; + + if (dev_priv->chipset == 0x50) { + nouveau_irq_register(dev, 0, nv50_vpe_isr); + NVOBJ_ENGINE_ADD(dev, MPEG, &pmpeg->base); + NVOBJ_CLASS(dev, 0x3174, MPEG); +#if 0 + NVOBJ_ENGINE_ADD(dev, ME, &pme->base); + NVOBJ_CLASS(dev, 0x4075, ME); +#endif + } else { + nouveau_irq_register(dev, 0, nv50_mpeg_isr); + NVOBJ_ENGINE_ADD(dev, MPEG, &pmpeg->base); + NVOBJ_CLASS(dev, 0x8274, MPEG); + } + + return 0; + +} diff --git a/drivers/gpu/drm/nouveau/nv50_pm.c b/drivers/gpu/drm/nouveau/nv50_pm.c new file mode 100644 index 00000000..8a281001 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv50_pm.c @@ -0,0 +1,146 @@ +/* + * Copyright 2010 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include "drmP.h" +#include "nouveau_drv.h" +#include "nouveau_bios.h" +#include "nouveau_pm.h" + +struct nv50_pm_state { + struct nouveau_pm_level *perflvl; + struct pll_lims pll; + enum pll_types type; + int N, M, P; +}; + +int +nv50_pm_clock_get(struct drm_device *dev, u32 id) +{ + struct pll_lims pll; + int P, N, M, ret; + u32 reg0, reg1; + + ret = get_pll_limits(dev, id, &pll); + if (ret) + return ret; + + reg0 = nv_rd32(dev, pll.reg + 0); + reg1 = nv_rd32(dev, pll.reg + 4); + + if ((reg0 & 0x80000000) == 0) { + if (id == PLL_SHADER) { + NV_DEBUG(dev, "Shader PLL is disabled. " + "Shader clock is twice the core\n"); + ret = nv50_pm_clock_get(dev, PLL_CORE); + if (ret > 0) + return ret << 1; + } else if (id == PLL_MEMORY) { + NV_DEBUG(dev, "Memory PLL is disabled. " + "Memory clock is equal to the ref_clk\n"); + return pll.refclk; + } + } + + P = (reg0 & 0x00070000) >> 16; + N = (reg1 & 0x0000ff00) >> 8; + M = (reg1 & 0x000000ff); + + return ((pll.refclk * N / M) >> P); +} + +void * +nv50_pm_clock_pre(struct drm_device *dev, struct nouveau_pm_level *perflvl, + u32 id, int khz) +{ + struct nv50_pm_state *state; + int dummy, ret; + + state = kzalloc(sizeof(*state), GFP_KERNEL); + if (!state) + return ERR_PTR(-ENOMEM); + state->type = id; + state->perflvl = perflvl; + + ret = get_pll_limits(dev, id, &state->pll); + if (ret < 0) { + kfree(state); + return (ret == -ENOENT) ? NULL : ERR_PTR(ret); + } + + ret = nv50_calc_pll(dev, &state->pll, khz, &state->N, &state->M, + &dummy, &dummy, &state->P); + if (ret < 0) { + kfree(state); + return ERR_PTR(ret); + } + + return state; +} + +void +nv50_pm_clock_set(struct drm_device *dev, void *pre_state) +{ + struct nv50_pm_state *state = pre_state; + struct nouveau_pm_level *perflvl = state->perflvl; + u32 reg = state->pll.reg, tmp; + struct bit_entry BIT_M; + u16 script; + int N = state->N; + int M = state->M; + int P = state->P; + + if (state->type == PLL_MEMORY && perflvl->memscript && + bit_table(dev, 'M', &BIT_M) == 0 && + BIT_M.version == 1 && BIT_M.length >= 0x0b) { + script = ROM16(BIT_M.data[0x05]); + if (script) + nouveau_bios_run_init_table(dev, script, NULL); + script = ROM16(BIT_M.data[0x07]); + if (script) + nouveau_bios_run_init_table(dev, script, NULL); + script = ROM16(BIT_M.data[0x09]); + if (script) + nouveau_bios_run_init_table(dev, script, NULL); + + nouveau_bios_run_init_table(dev, perflvl->memscript, NULL); + } + + if (state->type == PLL_MEMORY) { + nv_wr32(dev, 0x100210, 0); + nv_wr32(dev, 0x1002dc, 1); + } + + tmp = nv_rd32(dev, reg + 0) & 0xfff8ffff; + tmp |= 0x80000000 | (P << 16); + nv_wr32(dev, reg + 0, tmp); + nv_wr32(dev, reg + 4, (N << 8) | M); + + if (state->type == PLL_MEMORY) { + nv_wr32(dev, 0x1002dc, 0); + nv_wr32(dev, 0x100210, 0x80000000); + } + + kfree(state); +} + diff --git a/drivers/gpu/drm/nouveau/nv50_sor.c b/drivers/gpu/drm/nouveau/nv50_sor.c new file mode 100644 index 00000000..c25c5938 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv50_sor.c @@ -0,0 +1,341 @@ +/* + * Copyright (C) 2008 Maarten Maathuis. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE + * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION + * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "drmP.h" +#include "drm_crtc_helper.h" + +#define NOUVEAU_DMA_DEBUG (nouveau_reg_debug & NOUVEAU_REG_DEBUG_EVO) +#include "nouveau_reg.h" +#include "nouveau_drv.h" +#include "nouveau_dma.h" +#include "nouveau_encoder.h" +#include "nouveau_connector.h" +#include "nouveau_crtc.h" +#include "nv50_display.h" + +static void +nv50_sor_disconnect(struct drm_encoder *encoder) +{ + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + struct drm_device *dev = encoder->dev; + struct nouveau_channel *evo = nv50_display(dev)->master; + int ret; + + if (!nv_encoder->crtc) + return; + nv50_crtc_blank(nouveau_crtc(nv_encoder->crtc), true); + + NV_DEBUG_KMS(dev, "Disconnecting SOR %d\n", nv_encoder->or); + + ret = RING_SPACE(evo, 4); + if (ret) { + NV_ERROR(dev, "no space while disconnecting SOR\n"); + return; + } + BEGIN_RING(evo, 0, NV50_EVO_SOR(nv_encoder->or, MODE_CTRL), 1); + OUT_RING (evo, 0); + BEGIN_RING(evo, 0, NV50_EVO_UPDATE, 1); + OUT_RING (evo, 0); + + nv_encoder->crtc = NULL; + nv_encoder->last_dpms = DRM_MODE_DPMS_OFF; +} + +static void +nv50_sor_dpms(struct drm_encoder *encoder, int mode) +{ + struct drm_device *dev = encoder->dev; + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + struct drm_encoder *enc; + uint32_t val; + int or = nv_encoder->or; + + NV_DEBUG_KMS(dev, "or %d type %d mode %d\n", or, nv_encoder->dcb->type, mode); + + nv_encoder->last_dpms = mode; + list_for_each_entry(enc, &dev->mode_config.encoder_list, head) { + struct nouveau_encoder *nvenc = nouveau_encoder(enc); + + if (nvenc == nv_encoder || + (nvenc->dcb->type != OUTPUT_TMDS && + nvenc->dcb->type != OUTPUT_LVDS && + nvenc->dcb->type != OUTPUT_DP) || + nvenc->dcb->or != nv_encoder->dcb->or) + continue; + + if (nvenc->last_dpms == DRM_MODE_DPMS_ON) + return; + } + + /* wait for it to be done */ + if (!nv_wait(dev, NV50_PDISPLAY_SOR_DPMS_CTRL(or), + NV50_PDISPLAY_SOR_DPMS_CTRL_PENDING, 0)) { + NV_ERROR(dev, "timeout: SOR_DPMS_CTRL_PENDING(%d) == 0\n", or); + NV_ERROR(dev, "SOR_DPMS_CTRL(%d) = 0x%08x\n", or, + nv_rd32(dev, NV50_PDISPLAY_SOR_DPMS_CTRL(or))); + } + + val = nv_rd32(dev, NV50_PDISPLAY_SOR_DPMS_CTRL(or)); + + if (mode == DRM_MODE_DPMS_ON) + val |= NV50_PDISPLAY_SOR_DPMS_CTRL_ON; + else + val &= ~NV50_PDISPLAY_SOR_DPMS_CTRL_ON; + + nv_wr32(dev, NV50_PDISPLAY_SOR_DPMS_CTRL(or), val | + NV50_PDISPLAY_SOR_DPMS_CTRL_PENDING); + if (!nv_wait(dev, NV50_PDISPLAY_SOR_DPMS_STATE(or), + NV50_PDISPLAY_SOR_DPMS_STATE_WAIT, 0)) { + NV_ERROR(dev, "timeout: SOR_DPMS_STATE_WAIT(%d) == 0\n", or); + NV_ERROR(dev, "SOR_DPMS_STATE(%d) = 0x%08x\n", or, + nv_rd32(dev, NV50_PDISPLAY_SOR_DPMS_STATE(or))); + } + + if (nv_encoder->dcb->type == OUTPUT_DP) { + struct nouveau_i2c_chan *auxch; + + auxch = nouveau_i2c_find(dev, nv_encoder->dcb->i2c_index); + if (!auxch) + return; + + if (mode == DRM_MODE_DPMS_ON) { + u8 status = DP_SET_POWER_D0; + nouveau_dp_auxch(auxch, 8, DP_SET_POWER, &status, 1); + nouveau_dp_link_train(encoder); + } else { + u8 status = DP_SET_POWER_D3; + nouveau_dp_auxch(auxch, 8, DP_SET_POWER, &status, 1); + } + } +} + +static void +nv50_sor_save(struct drm_encoder *encoder) +{ + NV_ERROR(encoder->dev, "!!\n"); +} + +static void +nv50_sor_restore(struct drm_encoder *encoder) +{ + NV_ERROR(encoder->dev, "!!\n"); +} + +static bool +nv50_sor_mode_fixup(struct drm_encoder *encoder, struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + struct nouveau_connector *connector; + + NV_DEBUG_KMS(encoder->dev, "or %d\n", nv_encoder->or); + + connector = nouveau_encoder_connector_get(nv_encoder); + if (!connector) { + NV_ERROR(encoder->dev, "Encoder has no connector\n"); + return false; + } + + if (connector->scaling_mode != DRM_MODE_SCALE_NONE && + connector->native_mode) { + int id = adjusted_mode->base.id; + *adjusted_mode = *connector->native_mode; + adjusted_mode->base.id = id; + } + + return true; +} + +static void +nv50_sor_prepare(struct drm_encoder *encoder) +{ +} + +static void +nv50_sor_commit(struct drm_encoder *encoder) +{ +} + +static void +nv50_sor_mode_set(struct drm_encoder *encoder, struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct nouveau_channel *evo = nv50_display(encoder->dev)->master; + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + struct drm_device *dev = encoder->dev; + struct nouveau_crtc *crtc = nouveau_crtc(encoder->crtc); + uint32_t mode_ctl = 0; + int ret; + + NV_DEBUG_KMS(dev, "or %d type %d -> crtc %d\n", + nv_encoder->or, nv_encoder->dcb->type, crtc->index); + + nv50_sor_dpms(encoder, DRM_MODE_DPMS_ON); + + switch (nv_encoder->dcb->type) { + case OUTPUT_TMDS: + if (nv_encoder->dcb->sorconf.link & 1) { + if (adjusted_mode->clock < 165000) + mode_ctl = 0x0100; + else + mode_ctl = 0x0500; + } else + mode_ctl = 0x0200; + break; + case OUTPUT_DP: + mode_ctl |= (nv_encoder->dp.mc_unknown << 16); + if (nv_encoder->dcb->sorconf.link & 1) + mode_ctl |= 0x00000800; + else + mode_ctl |= 0x00000900; + break; + default: + break; + } + + if (crtc->index == 1) + mode_ctl |= NV50_EVO_SOR_MODE_CTRL_CRTC1; + else + mode_ctl |= NV50_EVO_SOR_MODE_CTRL_CRTC0; + + if (adjusted_mode->flags & DRM_MODE_FLAG_NHSYNC) + mode_ctl |= NV50_EVO_SOR_MODE_CTRL_NHSYNC; + + if (adjusted_mode->flags & DRM_MODE_FLAG_NVSYNC) + mode_ctl |= NV50_EVO_SOR_MODE_CTRL_NVSYNC; + + ret = RING_SPACE(evo, 2); + if (ret) { + NV_ERROR(dev, "no space while connecting SOR\n"); + return; + } + BEGIN_RING(evo, 0, NV50_EVO_SOR(nv_encoder->or, MODE_CTRL), 1); + OUT_RING(evo, mode_ctl); + + nv_encoder->crtc = encoder->crtc; +} + +static struct drm_crtc * +nv50_sor_crtc_get(struct drm_encoder *encoder) +{ + return nouveau_encoder(encoder)->crtc; +} + +static const struct drm_encoder_helper_funcs nv50_sor_helper_funcs = { + .dpms = nv50_sor_dpms, + .save = nv50_sor_save, + .restore = nv50_sor_restore, + .mode_fixup = nv50_sor_mode_fixup, + .prepare = nv50_sor_prepare, + .commit = nv50_sor_commit, + .mode_set = nv50_sor_mode_set, + .get_crtc = nv50_sor_crtc_get, + .detect = NULL, + .disable = nv50_sor_disconnect +}; + +static void +nv50_sor_destroy(struct drm_encoder *encoder) +{ + struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); + + if (!encoder) + return; + + NV_DEBUG_KMS(encoder->dev, "\n"); + + drm_encoder_cleanup(encoder); + + kfree(nv_encoder); +} + +static const struct drm_encoder_funcs nv50_sor_encoder_funcs = { + .destroy = nv50_sor_destroy, +}; + +int +nv50_sor_create(struct drm_connector *connector, struct dcb_entry *entry) +{ + struct nouveau_encoder *nv_encoder = NULL; + struct drm_device *dev = connector->dev; + struct drm_encoder *encoder; + int type; + + NV_DEBUG_KMS(dev, "\n"); + + switch (entry->type) { + case OUTPUT_TMDS: + case OUTPUT_DP: + type = DRM_MODE_ENCODER_TMDS; + break; + case OUTPUT_LVDS: + type = DRM_MODE_ENCODER_LVDS; + break; + default: + return -EINVAL; + } + + nv_encoder = kzalloc(sizeof(*nv_encoder), GFP_KERNEL); + if (!nv_encoder) + return -ENOMEM; + encoder = to_drm_encoder(nv_encoder); + + nv_encoder->dcb = entry; + nv_encoder->or = ffs(entry->or) - 1; + nv_encoder->last_dpms = DRM_MODE_DPMS_OFF; + + drm_encoder_init(dev, encoder, &nv50_sor_encoder_funcs, type); + drm_encoder_helper_add(encoder, &nv50_sor_helper_funcs); + + encoder->possible_crtcs = entry->heads; + encoder->possible_clones = 0; + + if (nv_encoder->dcb->type == OUTPUT_DP) { + int or = nv_encoder->or, link = !(entry->dpconf.sor.link & 1); + uint32_t tmp; + + tmp = nv_rd32(dev, 0x61c700 + (or * 0x800)); + + switch ((tmp & 0x00000f00) >> 8) { + case 8: + case 9: + nv_encoder->dp.mc_unknown = (tmp & 0x000f0000) >> 16; + tmp = nv_rd32(dev, NV50_SOR_DP_CTRL(or, link)); + nv_encoder->dp.unk0 = tmp & 0x000001fc; + tmp = nv_rd32(dev, NV50_SOR_DP_UNK128(or, link)); + nv_encoder->dp.unk1 = tmp & 0x010f7f3f; + break; + default: + break; + } + + if (!nv_encoder->dp.mc_unknown) + nv_encoder->dp.mc_unknown = 5; + } + + drm_mode_connector_attach_encoder(connector, encoder); + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nv50_vm.c b/drivers/gpu/drm/nouveau/nv50_vm.c new file mode 100644 index 00000000..1a0dd491 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv50_vm.c @@ -0,0 +1,182 @@ +/* + * Copyright 2010 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include "drmP.h" + +#include "nouveau_drv.h" +#include "nouveau_vm.h" + +void +nv50_vm_map_pgt(struct nouveau_gpuobj *pgd, u32 pde, + struct nouveau_gpuobj *pgt[2]) +{ + u64 phys = 0xdeadcafe00000000ULL; + u32 coverage = 0; + + if (pgt[0]) { + phys = 0x00000003 | pgt[0]->vinst; /* present, 4KiB pages */ + coverage = (pgt[0]->size >> 3) << 12; + } else + if (pgt[1]) { + phys = 0x00000001 | pgt[1]->vinst; /* present */ + coverage = (pgt[1]->size >> 3) << 16; + } + + if (phys & 1) { + if (coverage <= 32 * 1024 * 1024) + phys |= 0x60; + else if (coverage <= 64 * 1024 * 1024) + phys |= 0x40; + else if (coverage < 128 * 1024 * 1024) + phys |= 0x20; + } + + nv_wo32(pgd, (pde * 8) + 0, lower_32_bits(phys)); + nv_wo32(pgd, (pde * 8) + 4, upper_32_bits(phys)); +} + +static inline u64 +nv50_vm_addr(struct nouveau_vma *vma, u64 phys, u32 memtype, u32 target) +{ + struct drm_nouveau_private *dev_priv = vma->vm->dev->dev_private; + + phys |= 1; /* present */ + phys |= (u64)memtype << 40; + + /* IGPs don't have real VRAM, re-target to stolen system memory */ + if (target == 0 && dev_priv->vram_sys_base) { + phys += dev_priv->vram_sys_base; + target = 3; + } + + phys |= target << 4; + + if (vma->access & NV_MEM_ACCESS_SYS) + phys |= (1 << 6); + + if (!(vma->access & NV_MEM_ACCESS_WO)) + phys |= (1 << 3); + + return phys; +} + +void +nv50_vm_map(struct nouveau_vma *vma, struct nouveau_gpuobj *pgt, + struct nouveau_mem *mem, u32 pte, u32 cnt, u64 phys, u64 delta) +{ + u32 comp = (mem->memtype & 0x180) >> 7; + u32 block; + int i; + + phys = nv50_vm_addr(vma, phys, mem->memtype, 0); + pte <<= 3; + cnt <<= 3; + + while (cnt) { + u32 offset_h = upper_32_bits(phys); + u32 offset_l = lower_32_bits(phys); + + for (i = 7; i >= 0; i--) { + block = 1 << (i + 3); + if (cnt >= block && !(pte & (block - 1))) + break; + } + offset_l |= (i << 7); + + phys += block << (vma->node->type - 3); + cnt -= block; + if (comp) { + u32 tag = mem->tag->start + ((delta >> 16) * comp); + offset_h |= (tag << 17); + delta += block << (vma->node->type - 3); + } + + while (block) { + nv_wo32(pgt, pte + 0, offset_l); + nv_wo32(pgt, pte + 4, offset_h); + pte += 8; + block -= 8; + } + } +} + +void +nv50_vm_map_sg(struct nouveau_vma *vma, struct nouveau_gpuobj *pgt, + struct nouveau_mem *mem, u32 pte, u32 cnt, dma_addr_t *list) +{ + pte <<= 3; + while (cnt--) { + u64 phys = nv50_vm_addr(vma, (u64)*list++, mem->memtype, 2); + nv_wo32(pgt, pte + 0, lower_32_bits(phys)); + nv_wo32(pgt, pte + 4, upper_32_bits(phys)); + pte += 8; + } +} + +void +nv50_vm_unmap(struct nouveau_gpuobj *pgt, u32 pte, u32 cnt) +{ + pte <<= 3; + while (cnt--) { + nv_wo32(pgt, pte + 0, 0x00000000); + nv_wo32(pgt, pte + 4, 0x00000000); + pte += 8; + } +} + +void +nv50_vm_flush(struct nouveau_vm *vm) +{ + struct drm_nouveau_private *dev_priv = vm->dev->dev_private; + struct nouveau_instmem_engine *pinstmem = &dev_priv->engine.instmem; + struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; + int i; + + pinstmem->flush(vm->dev); + + /* BAR */ + if (vm != dev_priv->chan_vm) { + nv50_vm_flush_engine(vm->dev, 6); + return; + } + + pfifo->tlb_flush(vm->dev); + for (i = 0; i < NVOBJ_ENGINE_NR; i++) { + if (atomic_read(&vm->engref[i])) + dev_priv->eng[i]->tlb_flush(vm->dev, i); + } +} + +void +nv50_vm_flush_engine(struct drm_device *dev, int engine) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + unsigned long flags; + + spin_lock_irqsave(&dev_priv->vm_lock, flags); + nv_wr32(dev, 0x100c80, (engine << 16) | 1); + if (!nv_wait(dev, 0x100c80, 0x00000001, 0x00000000)) + NV_ERROR(dev, "vm flush timeout: engine %d\n", engine); + spin_unlock_irqrestore(&dev_priv->vm_lock, flags); +} diff --git a/drivers/gpu/drm/nouveau/nv50_vram.c b/drivers/gpu/drm/nouveau/nv50_vram.c new file mode 100644 index 00000000..ffbc3d8c --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv50_vram.c @@ -0,0 +1,211 @@ +/* + * Copyright 2010 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include "drmP.h" +#include "nouveau_drv.h" +#include "nouveau_mm.h" + +static int types[0x80] = { + 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 1, 1, 1, 1, 0, 0, 0, 0, 2, 2, 2, 2, 0, 0, 0, 0, + 1, 1, 1, 1, 1, 1, 1, 0, 2, 2, 2, 2, 2, 2, 2, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2, 0, 0, + 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 2, 2, 2, 2, + 1, 0, 2, 0, 1, 0, 2, 0, 1, 1, 2, 2, 1, 1, 0, 0 +}; + +bool +nv50_vram_flags_valid(struct drm_device *dev, u32 tile_flags) +{ + int type = (tile_flags & NOUVEAU_GEM_TILE_LAYOUT_MASK) >> 8; + + if (likely(type < ARRAY_SIZE(types) && types[type])) + return true; + return false; +} + +void +nv50_vram_del(struct drm_device *dev, struct nouveau_mem **pmem) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct ttm_bo_device *bdev = &dev_priv->ttm.bdev; + struct ttm_mem_type_manager *man = &bdev->man[TTM_PL_VRAM]; + struct nouveau_mm *mm = man->priv; + struct nouveau_mm_node *this; + struct nouveau_mem *mem; + + mem = *pmem; + *pmem = NULL; + if (unlikely(mem == NULL)) + return; + + mutex_lock(&mm->mutex); + while (!list_empty(&mem->regions)) { + this = list_first_entry(&mem->regions, struct nouveau_mm_node, rl_entry); + + list_del(&this->rl_entry); + nouveau_mm_put(mm, this); + } + + if (mem->tag) { + drm_mm_put_block(mem->tag); + mem->tag = NULL; + } + mutex_unlock(&mm->mutex); + + kfree(mem); +} + +int +nv50_vram_new(struct drm_device *dev, u64 size, u32 align, u32 size_nc, + u32 memtype, struct nouveau_mem **pmem) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct ttm_bo_device *bdev = &dev_priv->ttm.bdev; + struct ttm_mem_type_manager *man = &bdev->man[TTM_PL_VRAM]; + struct nouveau_mm *mm = man->priv; + struct nouveau_mm_node *r; + struct nouveau_mem *mem; + int comp = (memtype & 0x300) >> 8; + int type = (memtype & 0x07f); + int ret; + + if (!types[type]) + return -EINVAL; + size >>= 12; + align >>= 12; + size_nc >>= 12; + + mem = kzalloc(sizeof(*mem), GFP_KERNEL); + if (!mem) + return -ENOMEM; + + mutex_lock(&mm->mutex); + if (comp) { + if (align == 16) { + struct nouveau_fb_engine *pfb = &dev_priv->engine.fb; + int n = (size >> 4) * comp; + + mem->tag = drm_mm_search_free(&pfb->tag_heap, n, 0, 0); + if (mem->tag) + mem->tag = drm_mm_get_block(mem->tag, n, 0); + } + + if (unlikely(!mem->tag)) + comp = 0; + } + + INIT_LIST_HEAD(&mem->regions); + mem->dev = dev_priv->dev; + mem->memtype = (comp << 7) | type; + mem->size = size; + + do { + ret = nouveau_mm_get(mm, types[type], size, size_nc, align, &r); + if (ret) { + mutex_unlock(&mm->mutex); + nv50_vram_del(dev, &mem); + return ret; + } + + list_add_tail(&r->rl_entry, &mem->regions); + size -= r->length; + } while (size); + mutex_unlock(&mm->mutex); + + r = list_first_entry(&mem->regions, struct nouveau_mm_node, rl_entry); + mem->offset = (u64)r->offset << 12; + *pmem = mem; + return 0; +} + +static u32 +nv50_vram_rblock(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + int i, parts, colbits, rowbitsa, rowbitsb, banks; + u64 rowsize, predicted; + u32 r0, r4, rt, ru, rblock_size; + + r0 = nv_rd32(dev, 0x100200); + r4 = nv_rd32(dev, 0x100204); + rt = nv_rd32(dev, 0x100250); + ru = nv_rd32(dev, 0x001540); + NV_DEBUG(dev, "memcfg 0x%08x 0x%08x 0x%08x 0x%08x\n", r0, r4, rt, ru); + + for (i = 0, parts = 0; i < 8; i++) { + if (ru & (0x00010000 << i)) + parts++; + } + + colbits = (r4 & 0x0000f000) >> 12; + rowbitsa = ((r4 & 0x000f0000) >> 16) + 8; + rowbitsb = ((r4 & 0x00f00000) >> 20) + 8; + banks = ((r4 & 0x01000000) ? 8 : 4); + + rowsize = parts * banks * (1 << colbits) * 8; + predicted = rowsize << rowbitsa; + if (r0 & 0x00000004) + predicted += rowsize << rowbitsb; + + if (predicted != dev_priv->vram_size) { + NV_WARN(dev, "memory controller reports %dMiB VRAM\n", + (u32)(dev_priv->vram_size >> 20)); + NV_WARN(dev, "we calculated %dMiB VRAM\n", + (u32)(predicted >> 20)); + } + + rblock_size = rowsize; + if (rt & 1) + rblock_size *= 3; + + NV_DEBUG(dev, "rblock %d bytes\n", rblock_size); + return rblock_size; +} + +int +nv50_vram_init(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + dev_priv->vram_size = nv_rd32(dev, 0x10020c); + dev_priv->vram_size |= (dev_priv->vram_size & 0xff) << 32; + dev_priv->vram_size &= 0xffffffff00ULL; + + switch (dev_priv->chipset) { + case 0xaa: + case 0xac: + case 0xaf: + dev_priv->vram_sys_base = (u64)nv_rd32(dev, 0x100e10) << 12; + dev_priv->vram_rblock_size = 4096; + break; + default: + dev_priv->vram_rblock_size = nv50_vram_rblock(dev); + break; + } + + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nv84_crypt.c b/drivers/gpu/drm/nouveau/nv84_crypt.c new file mode 100644 index 00000000..75b809a5 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv84_crypt.c @@ -0,0 +1,193 @@ +/* + * Copyright 2010 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include "drmP.h" +#include "nouveau_drv.h" +#include "nouveau_util.h" +#include "nouveau_vm.h" +#include "nouveau_ramht.h" + +struct nv84_crypt_engine { + struct nouveau_exec_engine base; +}; + +static int +nv84_crypt_context_new(struct nouveau_channel *chan, int engine) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *ramin = chan->ramin; + struct nouveau_gpuobj *ctx; + int ret; + + NV_DEBUG(dev, "ch%d\n", chan->id); + + ret = nouveau_gpuobj_new(dev, chan, 256, 0, NVOBJ_FLAG_ZERO_ALLOC | + NVOBJ_FLAG_ZERO_FREE, &ctx); + if (ret) + return ret; + + nv_wo32(ramin, 0xa0, 0x00190000); + nv_wo32(ramin, 0xa4, ctx->vinst + ctx->size - 1); + nv_wo32(ramin, 0xa8, ctx->vinst); + nv_wo32(ramin, 0xac, 0); + nv_wo32(ramin, 0xb0, 0); + nv_wo32(ramin, 0xb4, 0); + dev_priv->engine.instmem.flush(dev); + + atomic_inc(&chan->vm->engref[engine]); + chan->engctx[engine] = ctx; + return 0; +} + +static void +nv84_crypt_context_del(struct nouveau_channel *chan, int engine) +{ + struct nouveau_gpuobj *ctx = chan->engctx[engine]; + struct drm_device *dev = chan->dev; + u32 inst; + + inst = (chan->ramin->vinst >> 12); + inst |= 0x80000000; + + /* mark context as invalid if still on the hardware, not + * doing this causes issues the next time PCRYPT is used, + * unsurprisingly :) + */ + nv_wr32(dev, 0x10200c, 0x00000000); + if (nv_rd32(dev, 0x102188) == inst) + nv_mask(dev, 0x102188, 0x80000000, 0x00000000); + if (nv_rd32(dev, 0x10218c) == inst) + nv_mask(dev, 0x10218c, 0x80000000, 0x00000000); + nv_wr32(dev, 0x10200c, 0x00000010); + + nouveau_gpuobj_ref(NULL, &ctx); + + atomic_dec(&chan->vm->engref[engine]); + chan->engctx[engine] = NULL; +} + +static int +nv84_crypt_object_new(struct nouveau_channel *chan, int engine, + u32 handle, u16 class) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *obj = NULL; + int ret; + + ret = nouveau_gpuobj_new(dev, chan, 16, 16, NVOBJ_FLAG_ZERO_FREE, &obj); + if (ret) + return ret; + obj->engine = 5; + obj->class = class; + + nv_wo32(obj, 0x00, class); + dev_priv->engine.instmem.flush(dev); + + ret = nouveau_ramht_insert(chan, handle, obj); + nouveau_gpuobj_ref(NULL, &obj); + return ret; +} + +static void +nv84_crypt_tlb_flush(struct drm_device *dev, int engine) +{ + nv50_vm_flush_engine(dev, 0x0a); +} + +static void +nv84_crypt_isr(struct drm_device *dev) +{ + u32 stat = nv_rd32(dev, 0x102130); + u32 mthd = nv_rd32(dev, 0x102190); + u32 data = nv_rd32(dev, 0x102194); + u32 inst = nv_rd32(dev, 0x102188) & 0x7fffffff; + int show = nouveau_ratelimit(); + + if (show) { + NV_INFO(dev, "PCRYPT_INTR: 0x%08x 0x%08x 0x%08x 0x%08x\n", + stat, mthd, data, inst); + } + + nv_wr32(dev, 0x102130, stat); + nv_wr32(dev, 0x10200c, 0x10); + + nv50_fb_vm_trap(dev, show); +} + +static int +nv84_crypt_fini(struct drm_device *dev, int engine) +{ + nv_wr32(dev, 0x102140, 0x00000000); + return 0; +} + +static int +nv84_crypt_init(struct drm_device *dev, int engine) +{ + nv_mask(dev, 0x000200, 0x00004000, 0x00000000); + nv_mask(dev, 0x000200, 0x00004000, 0x00004000); + + nv_wr32(dev, 0x102130, 0xffffffff); + nv_wr32(dev, 0x102140, 0xffffffbf); + + nv_wr32(dev, 0x10200c, 0x00000010); + return 0; +} + +static void +nv84_crypt_destroy(struct drm_device *dev, int engine) +{ + struct nv84_crypt_engine *pcrypt = nv_engine(dev, engine); + + NVOBJ_ENGINE_DEL(dev, CRYPT); + + nouveau_irq_unregister(dev, 14); + kfree(pcrypt); +} + +int +nv84_crypt_create(struct drm_device *dev) +{ + struct nv84_crypt_engine *pcrypt; + + pcrypt = kzalloc(sizeof(*pcrypt), GFP_KERNEL); + if (!pcrypt) + return -ENOMEM; + + pcrypt->base.destroy = nv84_crypt_destroy; + pcrypt->base.init = nv84_crypt_init; + pcrypt->base.fini = nv84_crypt_fini; + pcrypt->base.context_new = nv84_crypt_context_new; + pcrypt->base.context_del = nv84_crypt_context_del; + pcrypt->base.object_new = nv84_crypt_object_new; + pcrypt->base.tlb_flush = nv84_crypt_tlb_flush; + + nouveau_irq_register(dev, 14, nv84_crypt_isr); + + NVOBJ_ENGINE_ADD(dev, CRYPT, &pcrypt->base); + NVOBJ_CLASS (dev, 0x74c1, CRYPT); + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nva3_copy.c b/drivers/gpu/drm/nouveau/nva3_copy.c new file mode 100644 index 00000000..b86820a6 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nva3_copy.c @@ -0,0 +1,226 @@ +/* + * Copyright 2011 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include +#include "drmP.h" +#include "nouveau_drv.h" +#include "nouveau_util.h" +#include "nouveau_vm.h" +#include "nouveau_ramht.h" +#include "nva3_copy.fuc.h" + +struct nva3_copy_engine { + struct nouveau_exec_engine base; +}; + +static int +nva3_copy_context_new(struct nouveau_channel *chan, int engine) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *ramin = chan->ramin; + struct nouveau_gpuobj *ctx = NULL; + int ret; + + NV_DEBUG(dev, "ch%d\n", chan->id); + + ret = nouveau_gpuobj_new(dev, chan, 256, 0, NVOBJ_FLAG_ZERO_ALLOC | + NVOBJ_FLAG_ZERO_FREE, &ctx); + if (ret) + return ret; + + nv_wo32(ramin, 0xc0, 0x00190000); + nv_wo32(ramin, 0xc4, ctx->vinst + ctx->size - 1); + nv_wo32(ramin, 0xc8, ctx->vinst); + nv_wo32(ramin, 0xcc, 0x00000000); + nv_wo32(ramin, 0xd0, 0x00000000); + nv_wo32(ramin, 0xd4, 0x00000000); + dev_priv->engine.instmem.flush(dev); + + atomic_inc(&chan->vm->engref[engine]); + chan->engctx[engine] = ctx; + return 0; +} + +static int +nva3_copy_object_new(struct nouveau_channel *chan, int engine, + u32 handle, u16 class) +{ + struct nouveau_gpuobj *ctx = chan->engctx[engine]; + + /* fuc engine doesn't need an object, our ramht code does.. */ + ctx->engine = 3; + ctx->class = class; + return nouveau_ramht_insert(chan, handle, ctx); +} + +static void +nva3_copy_context_del(struct nouveau_channel *chan, int engine) +{ + struct nouveau_gpuobj *ctx = chan->engctx[engine]; + struct drm_device *dev = chan->dev; + u32 inst; + + inst = (chan->ramin->vinst >> 12); + inst |= 0x40000000; + + /* disable fifo access */ + nv_wr32(dev, 0x104048, 0x00000000); + /* mark channel as unloaded if it's currently active */ + if (nv_rd32(dev, 0x104050) == inst) + nv_mask(dev, 0x104050, 0x40000000, 0x00000000); + /* mark next channel as invalid if it's about to be loaded */ + if (nv_rd32(dev, 0x104054) == inst) + nv_mask(dev, 0x104054, 0x40000000, 0x00000000); + /* restore fifo access */ + nv_wr32(dev, 0x104048, 0x00000003); + + for (inst = 0xc0; inst <= 0xd4; inst += 4) + nv_wo32(chan->ramin, inst, 0x00000000); + + nouveau_gpuobj_ref(NULL, &ctx); + + atomic_dec(&chan->vm->engref[engine]); + chan->engctx[engine] = ctx; +} + +static void +nva3_copy_tlb_flush(struct drm_device *dev, int engine) +{ + nv50_vm_flush_engine(dev, 0x0d); +} + +static int +nva3_copy_init(struct drm_device *dev, int engine) +{ + int i; + + nv_mask(dev, 0x000200, 0x00002000, 0x00000000); + nv_mask(dev, 0x000200, 0x00002000, 0x00002000); + nv_wr32(dev, 0x104014, 0xffffffff); /* disable all interrupts */ + + /* upload ucode */ + nv_wr32(dev, 0x1041c0, 0x01000000); + for (i = 0; i < sizeof(nva3_pcopy_data) / 4; i++) + nv_wr32(dev, 0x1041c4, nva3_pcopy_data[i]); + + nv_wr32(dev, 0x104180, 0x01000000); + for (i = 0; i < sizeof(nva3_pcopy_code) / 4; i++) { + if ((i & 0x3f) == 0) + nv_wr32(dev, 0x104188, i >> 6); + nv_wr32(dev, 0x104184, nva3_pcopy_code[i]); + } + + /* start it running */ + nv_wr32(dev, 0x10410c, 0x00000000); + nv_wr32(dev, 0x104104, 0x00000000); /* ENTRY */ + nv_wr32(dev, 0x104100, 0x00000002); /* TRIGGER */ + return 0; +} + +static int +nva3_copy_fini(struct drm_device *dev, int engine) +{ + nv_mask(dev, 0x104048, 0x00000003, 0x00000000); + + /* trigger fuc context unload */ + nv_wait(dev, 0x104008, 0x0000000c, 0x00000000); + nv_mask(dev, 0x104054, 0x40000000, 0x00000000); + nv_wr32(dev, 0x104000, 0x00000008); + nv_wait(dev, 0x104008, 0x00000008, 0x00000000); + + nv_wr32(dev, 0x104014, 0xffffffff); + return 0; +} + +static struct nouveau_enum nva3_copy_isr_error_name[] = { + { 0x0001, "ILLEGAL_MTHD" }, + { 0x0002, "INVALID_ENUM" }, + { 0x0003, "INVALID_BITFIELD" }, + {} +}; + +static void +nva3_copy_isr(struct drm_device *dev) +{ + u32 dispatch = nv_rd32(dev, 0x10401c); + u32 stat = nv_rd32(dev, 0x104008) & dispatch & ~(dispatch >> 16); + u32 inst = nv_rd32(dev, 0x104050) & 0x3fffffff; + u32 ssta = nv_rd32(dev, 0x104040) & 0x0000ffff; + u32 addr = nv_rd32(dev, 0x104040) >> 16; + u32 mthd = (addr & 0x07ff) << 2; + u32 subc = (addr & 0x3800) >> 11; + u32 data = nv_rd32(dev, 0x104044); + int chid = nv50_graph_isr_chid(dev, inst); + + if (stat & 0x00000040) { + NV_INFO(dev, "PCOPY: DISPATCH_ERROR ["); + nouveau_enum_print(nva3_copy_isr_error_name, ssta); + printk("] ch %d [0x%08x] subc %d mthd 0x%04x data 0x%08x\n", + chid, inst, subc, mthd, data); + nv_wr32(dev, 0x104004, 0x00000040); + stat &= ~0x00000040; + } + + if (stat) { + NV_INFO(dev, "PCOPY: unhandled intr 0x%08x\n", stat); + nv_wr32(dev, 0x104004, stat); + } + nv50_fb_vm_trap(dev, 1); +} + +static void +nva3_copy_destroy(struct drm_device *dev, int engine) +{ + struct nva3_copy_engine *pcopy = nv_engine(dev, engine); + + nouveau_irq_unregister(dev, 22); + + NVOBJ_ENGINE_DEL(dev, COPY0); + kfree(pcopy); +} + +int +nva3_copy_create(struct drm_device *dev) +{ + struct nva3_copy_engine *pcopy; + + pcopy = kzalloc(sizeof(*pcopy), GFP_KERNEL); + if (!pcopy) + return -ENOMEM; + + pcopy->base.destroy = nva3_copy_destroy; + pcopy->base.init = nva3_copy_init; + pcopy->base.fini = nva3_copy_fini; + pcopy->base.context_new = nva3_copy_context_new; + pcopy->base.context_del = nva3_copy_context_del; + pcopy->base.object_new = nva3_copy_object_new; + pcopy->base.tlb_flush = nva3_copy_tlb_flush; + + nouveau_irq_register(dev, 22, nva3_copy_isr); + + NVOBJ_ENGINE_ADD(dev, COPY0, &pcopy->base); + NVOBJ_CLASS(dev, 0x85b5, COPY0); + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nva3_copy.fuc b/drivers/gpu/drm/nouveau/nva3_copy.fuc new file mode 100644 index 00000000..eaf35f83 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nva3_copy.fuc @@ -0,0 +1,870 @@ +/* fuc microcode for copy engine on nva3- chipsets + * + * Copyright 2011 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +/* To build for nva3:nvc0 + * m4 -DNVA3 nva3_copy.fuc | envyas -a -w -m fuc -V nva3 -o nva3_copy.fuc.h + * + * To build for nvc0- + * m4 -DNVC0 nva3_copy.fuc | envyas -a -w -m fuc -V nva3 -o nvc0_copy.fuc.h + */ + +ifdef(`NVA3', +.section nva3_pcopy_data, +.section nvc0_pcopy_data +) + +ctx_object: .b32 0 +ifdef(`NVA3', +ctx_dma: +ctx_dma_query: .b32 0 +ctx_dma_src: .b32 0 +ctx_dma_dst: .b32 0 +,) +.equ ctx_dma_count 3 +ctx_query_address_high: .b32 0 +ctx_query_address_low: .b32 0 +ctx_query_counter: .b32 0 +ctx_src_address_high: .b32 0 +ctx_src_address_low: .b32 0 +ctx_src_pitch: .b32 0 +ctx_src_tile_mode: .b32 0 +ctx_src_xsize: .b32 0 +ctx_src_ysize: .b32 0 +ctx_src_zsize: .b32 0 +ctx_src_zoff: .b32 0 +ctx_src_xoff: .b32 0 +ctx_src_yoff: .b32 0 +ctx_src_cpp: .b32 0 +ctx_dst_address_high: .b32 0 +ctx_dst_address_low: .b32 0 +ctx_dst_pitch: .b32 0 +ctx_dst_tile_mode: .b32 0 +ctx_dst_xsize: .b32 0 +ctx_dst_ysize: .b32 0 +ctx_dst_zsize: .b32 0 +ctx_dst_zoff: .b32 0 +ctx_dst_xoff: .b32 0 +ctx_dst_yoff: .b32 0 +ctx_dst_cpp: .b32 0 +ctx_format: .b32 0 +ctx_swz_const0: .b32 0 +ctx_swz_const1: .b32 0 +ctx_xcnt: .b32 0 +ctx_ycnt: .b32 0 +.align 256 + +dispatch_table: +// mthd 0x0000, NAME +.b16 0x000 1 +.b32 ctx_object ~0xffffffff +// mthd 0x0100, NOP +.b16 0x040 1 +.b32 0x00010000 + cmd_nop ~0xffffffff +// mthd 0x0140, PM_TRIGGER +.b16 0x050 1 +.b32 0x00010000 + cmd_pm_trigger ~0xffffffff +ifdef(`NVA3', ` +// mthd 0x0180-0x018c, DMA_ +.b16 0x060 ctx_dma_count +dispatch_dma: +.b32 0x00010000 + cmd_dma ~0xffffffff +.b32 0x00010000 + cmd_dma ~0xffffffff +.b32 0x00010000 + cmd_dma ~0xffffffff +',) +// mthd 0x0200-0x0218, SRC_TILE +.b16 0x80 7 +.b32 ctx_src_tile_mode ~0x00000fff +.b32 ctx_src_xsize ~0x0007ffff +.b32 ctx_src_ysize ~0x00001fff +.b32 ctx_src_zsize ~0x000007ff +.b32 ctx_src_zoff ~0x00000fff +.b32 ctx_src_xoff ~0x0007ffff +.b32 ctx_src_yoff ~0x00001fff +// mthd 0x0220-0x0238, DST_TILE +.b16 0x88 7 +.b32 ctx_dst_tile_mode ~0x00000fff +.b32 ctx_dst_xsize ~0x0007ffff +.b32 ctx_dst_ysize ~0x00001fff +.b32 ctx_dst_zsize ~0x000007ff +.b32 ctx_dst_zoff ~0x00000fff +.b32 ctx_dst_xoff ~0x0007ffff +.b32 ctx_dst_yoff ~0x00001fff +// mthd 0x0300-0x0304, EXEC, WRCACHE_FLUSH +.b16 0xc0 2 +.b32 0x00010000 + cmd_exec ~0xffffffff +.b32 0x00010000 + cmd_wrcache_flush ~0xffffffff +// mthd 0x030c-0x0340, various stuff +.b16 0xc3 14 +.b32 ctx_src_address_high ~0x000000ff +.b32 ctx_src_address_low ~0xfffffff0 +.b32 ctx_dst_address_high ~0x000000ff +.b32 ctx_dst_address_low ~0xfffffff0 +.b32 ctx_src_pitch ~0x0007ffff +.b32 ctx_dst_pitch ~0x0007ffff +.b32 ctx_xcnt ~0x0000ffff +.b32 ctx_ycnt ~0x00001fff +.b32 ctx_format ~0x0333ffff +.b32 ctx_swz_const0 ~0xffffffff +.b32 ctx_swz_const1 ~0xffffffff +.b32 ctx_query_address_high ~0x000000ff +.b32 ctx_query_address_low ~0xffffffff +.b32 ctx_query_counter ~0xffffffff +.b16 0x800 0 + +ifdef(`NVA3', +.section nva3_pcopy_code, +.section nvc0_pcopy_code +) + +main: + clear b32 $r0 + mov $sp $r0 + + // setup i0 handler and route fifo and ctxswitch to it + mov $r1 ih + mov $iv0 $r1 + mov $r1 0x400 + movw $r2 0xfff3 + sethi $r2 0 + iowr I[$r2 + 0x300] $r2 + + // enable interrupts + or $r2 0xc + iowr I[$r1] $r2 + bset $flags ie0 + + // enable fifo access and context switching + mov $r1 0x1200 + mov $r2 3 + iowr I[$r1] $r2 + + // sleep forever, waking for interrupts + bset $flags $p0 + spin: + sleep $p0 + bra spin + +// i0 handler +ih: + iord $r1 I[$r0 + 0x200] + + and $r2 $r1 0x00000008 + bra e ih_no_chsw + call chsw + ih_no_chsw: + and $r2 $r1 0x00000004 + bra e ih_no_cmd + call dispatch + + ih_no_cmd: + and $r1 $r1 0x0000000c + iowr I[$r0 + 0x100] $r1 + iret + +// $p1 direction (0 = unload, 1 = load) +// $r3 channel +swctx: + mov $r4 0x7700 + mov $xtargets $r4 +ifdef(`NVA3', ` + // target 7 hardcoded to ctx dma object + mov $xdbase $r0 +', ` // NVC0 + // read SCRATCH3 to decide if we are PCOPY0 or PCOPY1 + mov $r4 0x2100 + iord $r4 I[$r4 + 0] + and $r4 1 + shl b32 $r4 4 + add b32 $r4 0x30 + + // channel is in vram + mov $r15 0x61c + shl b32 $r15 6 + mov $r5 0x114 + iowrs I[$r15] $r5 + + // read 16-byte PCOPYn info, containing context pointer, from channel + shl b32 $r5 $r3 4 + add b32 $r5 2 + mov $xdbase $r5 + mov $r5 $sp + // get a chunk of stack space, aligned to 256 byte boundary + sub b32 $r5 0x100 + mov $r6 0xff + not b32 $r6 + and $r5 $r6 + sethi $r5 0x00020000 + xdld $r4 $r5 + xdwait + sethi $r5 0 + + // set context pointer, from within channel VM + mov $r14 0 + iowrs I[$r15] $r14 + ld b32 $r4 D[$r5 + 0] + shr b32 $r4 8 + ld b32 $r6 D[$r5 + 4] + shl b32 $r6 24 + or $r4 $r6 + mov $xdbase $r4 +') + // 256-byte context, at start of data segment + mov b32 $r4 $r0 + sethi $r4 0x60000 + + // swap! + bra $p1 swctx_load + xdst $r0 $r4 + bra swctx_done + swctx_load: + xdld $r0 $r4 + swctx_done: + xdwait + ret + +chsw: + // read current channel + mov $r2 0x1400 + iord $r3 I[$r2] + + // if it's active, unload it and return + xbit $r15 $r3 0x1e + bra e chsw_no_unload + bclr $flags $p1 + call swctx + bclr $r3 0x1e + iowr I[$r2] $r3 + mov $r4 1 + iowr I[$r2 + 0x200] $r4 + ret + + // read next channel + chsw_no_unload: + iord $r3 I[$r2 + 0x100] + + // is there a channel waiting to be loaded? + xbit $r13 $r3 0x1e + bra e chsw_finish_load + bset $flags $p1 + call swctx +ifdef(`NVA3', + // load dma objects back into TARGET regs + mov $r5 ctx_dma + mov $r6 ctx_dma_count + chsw_load_ctx_dma: + ld b32 $r7 D[$r5 + $r6 * 4] + add b32 $r8 $r6 0x180 + shl b32 $r8 8 + iowr I[$r8] $r7 + sub b32 $r6 1 + bra nc chsw_load_ctx_dma +,) + + chsw_finish_load: + mov $r3 2 + iowr I[$r2 + 0x200] $r3 + ret + +dispatch: + // read incoming fifo command + mov $r3 0x1900 + iord $r2 I[$r3 + 0x100] + iord $r3 I[$r3 + 0x000] + and $r4 $r2 0x7ff + // $r2 will be used to store exception data + shl b32 $r2 0x10 + + // lookup method in the dispatch table, ILLEGAL_MTHD if not found + mov $r5 dispatch_table + clear b32 $r6 + clear b32 $r7 + dispatch_loop: + ld b16 $r6 D[$r5 + 0] + ld b16 $r7 D[$r5 + 2] + add b32 $r5 4 + cmpu b32 $r4 $r6 + bra c dispatch_illegal_mthd + add b32 $r7 $r6 + cmpu b32 $r4 $r7 + bra c dispatch_valid_mthd + sub b32 $r7 $r6 + shl b32 $r7 3 + add b32 $r5 $r7 + bra dispatch_loop + + // ensure no bits set in reserved fields, INVALID_BITFIELD + dispatch_valid_mthd: + sub b32 $r4 $r6 + shl b32 $r4 3 + add b32 $r4 $r5 + ld b32 $r5 D[$r4 + 4] + and $r5 $r3 + cmpu b32 $r5 0 + bra ne dispatch_invalid_bitfield + + // depending on dispatch flags: execute method, or save data as state + ld b16 $r5 D[$r4 + 0] + ld b16 $r6 D[$r4 + 2] + cmpu b32 $r6 0 + bra ne dispatch_cmd + st b32 D[$r5] $r3 + bra dispatch_done + dispatch_cmd: + bclr $flags $p1 + call $r5 + bra $p1 dispatch_error + bra dispatch_done + + dispatch_invalid_bitfield: + or $r2 2 + dispatch_illegal_mthd: + or $r2 1 + + // store exception data in SCRATCH0/SCRATCH1, signal hostirq + dispatch_error: + mov $r4 0x1000 + iowr I[$r4 + 0x000] $r2 + iowr I[$r4 + 0x100] $r3 + mov $r2 0x40 + iowr I[$r0] $r2 + hostirq_wait: + iord $r2 I[$r0 + 0x200] + and $r2 0x40 + cmpu b32 $r2 0 + bra ne hostirq_wait + + dispatch_done: + mov $r2 0x1d00 + mov $r3 1 + iowr I[$r2] $r3 + ret + +// No-operation +// +// Inputs: +// $r1: irqh state +// $r2: hostirq state +// $r3: data +// $r4: dispatch table entry +// Outputs: +// $r1: irqh state +// $p1: set on error +// $r2: hostirq state +// $r3: data +cmd_nop: + ret + +// PM_TRIGGER +// +// Inputs: +// $r1: irqh state +// $r2: hostirq state +// $r3: data +// $r4: dispatch table entry +// Outputs: +// $r1: irqh state +// $p1: set on error +// $r2: hostirq state +// $r3: data +cmd_pm_trigger: + mov $r2 0x2200 + clear b32 $r3 + sethi $r3 0x20000 + iowr I[$r2] $r3 + ret + +ifdef(`NVA3', +// SET_DMA_* method handler +// +// Inputs: +// $r1: irqh state +// $r2: hostirq state +// $r3: data +// $r4: dispatch table entry +// Outputs: +// $r1: irqh state +// $p1: set on error +// $r2: hostirq state +// $r3: data +cmd_dma: + sub b32 $r4 dispatch_dma + shr b32 $r4 1 + bset $r3 0x1e + st b32 D[$r4 + ctx_dma] $r3 + add b32 $r4 0x600 + shl b32 $r4 6 + iowr I[$r4] $r3 + ret +,) + +// Calculates the hw swizzle mask and adjusts the surface's xcnt to match +// +cmd_exec_set_format: + // zero out a chunk of the stack to store the swizzle into + add $sp -0x10 + st b32 D[$sp + 0x00] $r0 + st b32 D[$sp + 0x04] $r0 + st b32 D[$sp + 0x08] $r0 + st b32 D[$sp + 0x0c] $r0 + + // extract cpp, src_ncomp and dst_ncomp from FORMAT + ld b32 $r4 D[$r0 + ctx_format] + extr $r5 $r4 16:17 + add b32 $r5 1 + extr $r6 $r4 20:21 + add b32 $r6 1 + extr $r7 $r4 24:25 + add b32 $r7 1 + + // convert FORMAT swizzle mask to hw swizzle mask + bclr $flags $p2 + clear b32 $r8 + clear b32 $r9 + ncomp_loop: + and $r10 $r4 0xf + shr b32 $r4 4 + clear b32 $r11 + bpc_loop: + cmpu b8 $r10 4 + bra nc cmp_c0 + mulu $r12 $r10 $r5 + add b32 $r12 $r11 + bset $flags $p2 + bra bpc_next + cmp_c0: + bra ne cmp_c1 + mov $r12 0x10 + add b32 $r12 $r11 + bra bpc_next + cmp_c1: + cmpu b8 $r10 6 + bra nc cmp_zero + mov $r12 0x14 + add b32 $r12 $r11 + bra bpc_next + cmp_zero: + mov $r12 0x80 + bpc_next: + st b8 D[$sp + $r8] $r12 + add b32 $r8 1 + add b32 $r11 1 + cmpu b32 $r11 $r5 + bra c bpc_loop + add b32 $r9 1 + cmpu b32 $r9 $r7 + bra c ncomp_loop + + // SRC_XCNT = (xcnt * src_cpp), or 0 if no src ref in swz (hw will hang) + mulu $r6 $r5 + st b32 D[$r0 + ctx_src_cpp] $r6 + ld b32 $r8 D[$r0 + ctx_xcnt] + mulu $r6 $r8 + bra $p2 dst_xcnt + clear b32 $r6 + + dst_xcnt: + mulu $r7 $r5 + st b32 D[$r0 + ctx_dst_cpp] $r7 + mulu $r7 $r8 + + mov $r5 0x810 + shl b32 $r5 6 + iowr I[$r5 + 0x000] $r6 + iowr I[$r5 + 0x100] $r7 + add b32 $r5 0x800 + ld b32 $r6 D[$r0 + ctx_dst_cpp] + sub b32 $r6 1 + shl b32 $r6 8 + ld b32 $r7 D[$r0 + ctx_src_cpp] + sub b32 $r7 1 + or $r6 $r7 + iowr I[$r5 + 0x000] $r6 + add b32 $r5 0x100 + ld b32 $r6 D[$sp + 0x00] + iowr I[$r5 + 0x000] $r6 + ld b32 $r6 D[$sp + 0x04] + iowr I[$r5 + 0x100] $r6 + ld b32 $r6 D[$sp + 0x08] + iowr I[$r5 + 0x200] $r6 + ld b32 $r6 D[$sp + 0x0c] + iowr I[$r5 + 0x300] $r6 + add b32 $r5 0x400 + ld b32 $r6 D[$r0 + ctx_swz_const0] + iowr I[$r5 + 0x000] $r6 + ld b32 $r6 D[$r0 + ctx_swz_const1] + iowr I[$r5 + 0x100] $r6 + add $sp 0x10 + ret + +// Setup to handle a tiled surface +// +// Calculates a number of parameters the hardware requires in order +// to correctly handle tiling. +// +// Offset calculation is performed as follows (Tp/Th/Td from TILE_MODE): +// nTx = round_up(w * cpp, 1 << Tp) >> Tp +// nTy = round_up(h, 1 << Th) >> Th +// Txo = (x * cpp) & ((1 << Tp) - 1) +// Tx = (x * cpp) >> Tp +// Tyo = y & ((1 << Th) - 1) +// Ty = y >> Th +// Tzo = z & ((1 << Td) - 1) +// Tz = z >> Td +// +// off = (Tzo << Tp << Th) + (Tyo << Tp) + Txo +// off += ((Tz * nTy * nTx)) + (Ty * nTx) + Tx) << Td << Th << Tp; +// +// Inputs: +// $r4: hw command (0x104800) +// $r5: ctx offset adjustment for src/dst selection +// $p2: set if dst surface +// +cmd_exec_set_surface_tiled: + // translate TILE_MODE into Tp, Th, Td shift values + ld b32 $r7 D[$r5 + ctx_src_tile_mode] + extr $r9 $r7 8:11 + extr $r8 $r7 4:7 +ifdef(`NVA3', + add b32 $r8 2 +, + add b32 $r8 3 +) + extr $r7 $r7 0:3 + cmp b32 $r7 0xe + bra ne xtile64 + mov $r7 4 + bra xtileok + xtile64: + xbit $r7 $flags $p2 + add b32 $r7 17 + bset $r4 $r7 + mov $r7 6 + xtileok: + + // Op = (x * cpp) & ((1 << Tp) - 1) + // Tx = (x * cpp) >> Tp + ld b32 $r10 D[$r5 + ctx_src_xoff] + ld b32 $r11 D[$r5 + ctx_src_cpp] + mulu $r10 $r11 + mov $r11 1 + shl b32 $r11 $r7 + sub b32 $r11 1 + and $r12 $r10 $r11 + shr b32 $r10 $r7 + + // Tyo = y & ((1 << Th) - 1) + // Ty = y >> Th + ld b32 $r13 D[$r5 + ctx_src_yoff] + mov $r14 1 + shl b32 $r14 $r8 + sub b32 $r14 1 + and $r11 $r13 $r14 + shr b32 $r13 $r8 + + // YTILE = ((1 << Th) << 12) | ((1 << Th) - Tyo) + add b32 $r14 1 + shl b32 $r15 $r14 12 + sub b32 $r14 $r11 + or $r15 $r14 + xbit $r6 $flags $p2 + add b32 $r6 0x208 + shl b32 $r6 8 + iowr I[$r6 + 0x000] $r15 + + // Op += Tyo << Tp + shl b32 $r11 $r7 + add b32 $r12 $r11 + + // nTx = ((w * cpp) + ((1 << Tp) - 1) >> Tp) + ld b32 $r15 D[$r5 + ctx_src_xsize] + ld b32 $r11 D[$r5 + ctx_src_cpp] + mulu $r15 $r11 + mov $r11 1 + shl b32 $r11 $r7 + sub b32 $r11 1 + add b32 $r15 $r11 + shr b32 $r15 $r7 + push $r15 + + // nTy = (h + ((1 << Th) - 1)) >> Th + ld b32 $r15 D[$r5 + ctx_src_ysize] + mov $r11 1 + shl b32 $r11 $r8 + sub b32 $r11 1 + add b32 $r15 $r11 + shr b32 $r15 $r8 + push $r15 + + // Tys = Tp + Th + // CFG_YZ_TILE_SIZE = ((1 << Th) >> 2) << Td + add b32 $r7 $r8 + sub b32 $r8 2 + mov $r11 1 + shl b32 $r11 $r8 + shl b32 $r11 $r9 + + // Tzo = z & ((1 << Td) - 1) + // Tz = z >> Td + // Op += Tzo << Tys + // Ts = Tys + Td + ld b32 $r8 D[$r5 + ctx_src_zoff] + mov $r14 1 + shl b32 $r14 $r9 + sub b32 $r14 1 + and $r15 $r8 $r14 + shl b32 $r15 $r7 + add b32 $r12 $r15 + add b32 $r7 $r9 + shr b32 $r8 $r9 + + // Ot = ((Tz * nTy * nTx) + (Ty * nTx) + Tx) << Ts + pop $r15 + pop $r9 + mulu $r13 $r9 + add b32 $r10 $r13 + mulu $r8 $r9 + mulu $r8 $r15 + add b32 $r10 $r8 + shl b32 $r10 $r7 + + // PITCH = (nTx - 1) << Ts + sub b32 $r9 1 + shl b32 $r9 $r7 + iowr I[$r6 + 0x200] $r9 + + // SRC_ADDRESS_LOW = (Ot + Op) & 0xffffffff + // CFG_ADDRESS_HIGH |= ((Ot + Op) >> 32) << 16 + ld b32 $r7 D[$r5 + ctx_src_address_low] + ld b32 $r8 D[$r5 + ctx_src_address_high] + add b32 $r10 $r12 + add b32 $r7 $r10 + adc b32 $r8 0 + shl b32 $r8 16 + or $r8 $r11 + sub b32 $r6 0x600 + iowr I[$r6 + 0x000] $r7 + add b32 $r6 0x400 + iowr I[$r6 + 0x000] $r8 + ret + +// Setup to handle a linear surface +// +// Nothing to see here.. Sets ADDRESS and PITCH, pretty non-exciting +// +cmd_exec_set_surface_linear: + xbit $r6 $flags $p2 + add b32 $r6 0x202 + shl b32 $r6 8 + ld b32 $r7 D[$r5 + ctx_src_address_low] + iowr I[$r6 + 0x000] $r7 + add b32 $r6 0x400 + ld b32 $r7 D[$r5 + ctx_src_address_high] + shl b32 $r7 16 + iowr I[$r6 + 0x000] $r7 + add b32 $r6 0x400 + ld b32 $r7 D[$r5 + ctx_src_pitch] + iowr I[$r6 + 0x000] $r7 + ret + +// wait for regs to be available for use +cmd_exec_wait: + push $r0 + push $r1 + mov $r0 0x800 + shl b32 $r0 6 + loop: + iord $r1 I[$r0] + and $r1 1 + bra ne loop + pop $r1 + pop $r0 + ret + +cmd_exec_query: + // if QUERY_SHORT not set, write out { -, 0, TIME_LO, TIME_HI } + xbit $r4 $r3 13 + bra ne query_counter + call cmd_exec_wait + mov $r4 0x80c + shl b32 $r4 6 + ld b32 $r5 D[$r0 + ctx_query_address_low] + add b32 $r5 4 + iowr I[$r4 + 0x000] $r5 + iowr I[$r4 + 0x100] $r0 + mov $r5 0xc + iowr I[$r4 + 0x200] $r5 + add b32 $r4 0x400 + ld b32 $r5 D[$r0 + ctx_query_address_high] + shl b32 $r5 16 + iowr I[$r4 + 0x000] $r5 + add b32 $r4 0x500 + mov $r5 0x00000b00 + sethi $r5 0x00010000 + iowr I[$r4 + 0x000] $r5 + mov $r5 0x00004040 + shl b32 $r5 1 + sethi $r5 0x80800000 + iowr I[$r4 + 0x100] $r5 + mov $r5 0x00001110 + sethi $r5 0x13120000 + iowr I[$r4 + 0x200] $r5 + mov $r5 0x00001514 + sethi $r5 0x17160000 + iowr I[$r4 + 0x300] $r5 + mov $r5 0x00002601 + sethi $r5 0x00010000 + mov $r4 0x800 + shl b32 $r4 6 + iowr I[$r4 + 0x000] $r5 + + // write COUNTER + query_counter: + call cmd_exec_wait + mov $r4 0x80c + shl b32 $r4 6 + ld b32 $r5 D[$r0 + ctx_query_address_low] + iowr I[$r4 + 0x000] $r5 + iowr I[$r4 + 0x100] $r0 + mov $r5 0x4 + iowr I[$r4 + 0x200] $r5 + add b32 $r4 0x400 + ld b32 $r5 D[$r0 + ctx_query_address_high] + shl b32 $r5 16 + iowr I[$r4 + 0x000] $r5 + add b32 $r4 0x500 + mov $r5 0x00000300 + iowr I[$r4 + 0x000] $r5 + mov $r5 0x00001110 + sethi $r5 0x13120000 + iowr I[$r4 + 0x100] $r5 + ld b32 $r5 D[$r0 + ctx_query_counter] + add b32 $r4 0x500 + iowr I[$r4 + 0x000] $r5 + mov $r5 0x00002601 + sethi $r5 0x00010000 + mov $r4 0x800 + shl b32 $r4 6 + iowr I[$r4 + 0x000] $r5 + ret + +// Execute a copy operation +// +// Inputs: +// $r1: irqh state +// $r2: hostirq state +// $r3: data +// 000002000 QUERY_SHORT +// 000001000 QUERY +// 000000100 DST_LINEAR +// 000000010 SRC_LINEAR +// 000000001 FORMAT +// $r4: dispatch table entry +// Outputs: +// $r1: irqh state +// $p1: set on error +// $r2: hostirq state +// $r3: data +cmd_exec: + call cmd_exec_wait + + // if format requested, call function to calculate it, otherwise + // fill in cpp/xcnt for both surfaces as if (cpp == 1) + xbit $r15 $r3 0 + bra e cmd_exec_no_format + call cmd_exec_set_format + mov $r4 0x200 + bra cmd_exec_init_src_surface + cmd_exec_no_format: + mov $r6 0x810 + shl b32 $r6 6 + mov $r7 1 + st b32 D[$r0 + ctx_src_cpp] $r7 + st b32 D[$r0 + ctx_dst_cpp] $r7 + ld b32 $r7 D[$r0 + ctx_xcnt] + iowr I[$r6 + 0x000] $r7 + iowr I[$r6 + 0x100] $r7 + clear b32 $r4 + + cmd_exec_init_src_surface: + bclr $flags $p2 + clear b32 $r5 + xbit $r15 $r3 4 + bra e src_tiled + call cmd_exec_set_surface_linear + bra cmd_exec_init_dst_surface + src_tiled: + call cmd_exec_set_surface_tiled + bset $r4 7 + + cmd_exec_init_dst_surface: + bset $flags $p2 + mov $r5 ctx_dst_address_high - ctx_src_address_high + xbit $r15 $r3 8 + bra e dst_tiled + call cmd_exec_set_surface_linear + bra cmd_exec_kick + dst_tiled: + call cmd_exec_set_surface_tiled + bset $r4 8 + + cmd_exec_kick: + mov $r5 0x800 + shl b32 $r5 6 + ld b32 $r6 D[$r0 + ctx_ycnt] + iowr I[$r5 + 0x100] $r6 + mov $r6 0x0041 + // SRC_TARGET = 1, DST_TARGET = 2 + sethi $r6 0x44000000 + or $r4 $r6 + iowr I[$r5] $r4 + + // if requested, queue up a QUERY write after the copy has completed + xbit $r15 $r3 12 + bra e cmd_exec_done + call cmd_exec_query + + cmd_exec_done: + ret + +// Flush write cache +// +// Inputs: +// $r1: irqh state +// $r2: hostirq state +// $r3: data +// $r4: dispatch table entry +// Outputs: +// $r1: irqh state +// $p1: set on error +// $r2: hostirq state +// $r3: data +cmd_wrcache_flush: + mov $r2 0x2200 + clear b32 $r3 + sethi $r3 0x10000 + iowr I[$r2] $r3 + ret + +.align 0x100 diff --git a/drivers/gpu/drm/nouveau/nva3_copy.fuc.h b/drivers/gpu/drm/nouveau/nva3_copy.fuc.h new file mode 100644 index 00000000..2731de22 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nva3_copy.fuc.h @@ -0,0 +1,534 @@ +uint32_t nva3_pcopy_data[] = { + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00010000, + 0x00000000, + 0x00000000, + 0x00010040, + 0x00010160, + 0x00000000, + 0x00010050, + 0x00010162, + 0x00000000, + 0x00030060, + 0x00010170, + 0x00000000, + 0x00010170, + 0x00000000, + 0x00010170, + 0x00000000, + 0x00070080, + 0x00000028, + 0xfffff000, + 0x0000002c, + 0xfff80000, + 0x00000030, + 0xffffe000, + 0x00000034, + 0xfffff800, + 0x00000038, + 0xfffff000, + 0x0000003c, + 0xfff80000, + 0x00000040, + 0xffffe000, + 0x00070088, + 0x00000054, + 0xfffff000, + 0x00000058, + 0xfff80000, + 0x0000005c, + 0xffffe000, + 0x00000060, + 0xfffff800, + 0x00000064, + 0xfffff000, + 0x00000068, + 0xfff80000, + 0x0000006c, + 0xffffe000, + 0x000200c0, + 0x00010492, + 0x00000000, + 0x0001051b, + 0x00000000, + 0x000e00c3, + 0x0000001c, + 0xffffff00, + 0x00000020, + 0x0000000f, + 0x00000048, + 0xffffff00, + 0x0000004c, + 0x0000000f, + 0x00000024, + 0xfff80000, + 0x00000050, + 0xfff80000, + 0x00000080, + 0xffff0000, + 0x00000084, + 0xffffe000, + 0x00000074, + 0xfccc0000, + 0x00000078, + 0x00000000, + 0x0000007c, + 0x00000000, + 0x00000010, + 0xffffff00, + 0x00000014, + 0x00000000, + 0x00000018, + 0x00000000, + 0x00000800, +}; + +uint32_t nva3_pcopy_code[] = { + 0x04fe04bd, + 0x3517f000, + 0xf10010fe, + 0xf1040017, + 0xf0fff327, + 0x22d00023, + 0x0c25f0c0, + 0xf40012d0, + 0x17f11031, + 0x27f01200, + 0x0012d003, + 0xf40031f4, + 0x0ef40028, + 0x8001cffd, + 0xf40812c4, + 0x21f4060b, + 0x0412c472, + 0xf4060bf4, + 0x11c4c321, + 0x4001d00c, + 0x47f101f8, + 0x4bfe7700, + 0x0007fe00, + 0xf00204b9, + 0x01f40643, + 0x0604fa09, + 0xfa060ef4, + 0x03f80504, + 0x27f100f8, + 0x23cf1400, + 0x1e3fc800, + 0xf4170bf4, + 0x21f40132, + 0x1e3af052, + 0xf00023d0, + 0x24d00147, + 0xcf00f880, + 0x3dc84023, + 0x220bf41e, + 0xf40131f4, + 0x57f05221, + 0x0367f004, + 0xa07856bc, + 0xb6018068, + 0x87d00884, + 0x0162b600, + 0xf0f018f4, + 0x23d00237, + 0xf100f880, + 0xcf190037, + 0x33cf4032, + 0xff24e400, + 0x1024b607, + 0x010057f1, + 0x74bd64bd, + 0x58005658, + 0x50b60157, + 0x0446b804, + 0xbb4d08f4, + 0x47b80076, + 0x0f08f404, + 0xb60276bb, + 0x57bb0374, + 0xdf0ef400, + 0xb60246bb, + 0x45bb0344, + 0x01459800, + 0xb00453fd, + 0x1bf40054, + 0x00455820, + 0xb0014658, + 0x1bf40064, + 0x00538009, + 0xf4300ef4, + 0x55f90132, + 0xf40c01f4, + 0x25f0250e, + 0x0125f002, + 0x100047f1, + 0xd00042d0, + 0x27f04043, + 0x0002d040, + 0xf08002cf, + 0x24b04024, + 0xf71bf400, + 0x1d0027f1, + 0xd00137f0, + 0x00f80023, + 0x27f100f8, + 0x34bd2200, + 0xd00233f0, + 0x00f80023, + 0x012842b7, + 0xf00145b6, + 0x43801e39, + 0x0040b701, + 0x0644b606, + 0xf80043d0, + 0xf030f400, + 0xb00001b0, + 0x01b00101, + 0x0301b002, + 0xc71d0498, + 0x50b63045, + 0x3446c701, + 0xc70160b6, + 0x70b63847, + 0x0232f401, + 0x94bd84bd, + 0xb60f4ac4, + 0xb4bd0445, + 0xf404a430, + 0xa5ff0f18, + 0x00cbbbc0, + 0xf40231f4, + 0x1bf4220e, + 0x10c7f00c, + 0xf400cbbb, + 0xa430160e, + 0x0c18f406, + 0xbb14c7f0, + 0x0ef400cb, + 0x80c7f107, + 0x01c83800, + 0xb60180b6, + 0xb5b801b0, + 0xc308f404, + 0xb80190b6, + 0x08f40497, + 0x0065fdb2, + 0x98110680, + 0x68fd2008, + 0x0502f400, + 0x75fd64bd, + 0x1c078000, + 0xf10078fd, + 0xb6081057, + 0x56d00654, + 0x4057d000, + 0x080050b7, + 0xb61c0698, + 0x64b60162, + 0x11079808, + 0xfd0172b6, + 0x56d00567, + 0x0050b700, + 0x0060b401, + 0xb40056d0, + 0x56d00160, + 0x0260b440, + 0xb48056d0, + 0x56d00360, + 0x0050b7c0, + 0x1e069804, + 0x980056d0, + 0x56d01f06, + 0x1030f440, + 0x579800f8, + 0x6879c70a, + 0xb66478c7, + 0x77c70280, + 0x0e76b060, + 0xf0091bf4, + 0x0ef40477, + 0x027cf00f, + 0xfd1170b6, + 0x77f00947, + 0x0f5a9806, + 0xfd115b98, + 0xb7f000ab, + 0x04b7bb01, + 0xff01b2b6, + 0xa7bbc4ab, + 0x105d9805, + 0xbb01e7f0, + 0xe2b604e8, + 0xb4deff01, + 0xb605d8bb, + 0xef9401e0, + 0x02ebbb0c, + 0xf005fefd, + 0x60b7026c, + 0x64b60208, + 0x006fd008, + 0xbb04b7bb, + 0x5f9800cb, + 0x115b980b, + 0xf000fbfd, + 0xb7bb01b7, + 0x01b2b604, + 0xbb00fbbb, + 0xf0f905f7, + 0xf00c5f98, + 0xb8bb01b7, + 0x01b2b604, + 0xbb00fbbb, + 0xf0f905f8, + 0xb60078bb, + 0xb7f00282, + 0x04b8bb01, + 0x9804b9bb, + 0xe7f00e58, + 0x04e9bb01, + 0xff01e2b6, + 0xf7bbf48e, + 0x00cfbb04, + 0xbb0079bb, + 0xf0fc0589, + 0xd9fd90fc, + 0x00adbb00, + 0xfd0089fd, + 0xa8bb008f, + 0x04a7bb00, + 0xbb0192b6, + 0x69d00497, + 0x08579880, + 0xbb075898, + 0x7abb00ac, + 0x0081b600, + 0xfd1084b6, + 0x62b7058b, + 0x67d00600, + 0x0060b700, + 0x0068d004, + 0x6cf000f8, + 0x0260b702, + 0x0864b602, + 0xd0085798, + 0x60b70067, + 0x57980400, + 0x1074b607, + 0xb70067d0, + 0x98040060, + 0x67d00957, + 0xf900f800, + 0xf110f900, + 0xb6080007, + 0x01cf0604, + 0x0114f000, + 0xfcfa1bf4, + 0xf800fc10, + 0x0d34c800, + 0xf5701bf4, + 0xf103ab21, + 0xb6080c47, + 0x05980644, + 0x0450b605, + 0xd00045d0, + 0x57f04040, + 0x8045d00c, + 0x040040b7, + 0xb6040598, + 0x45d01054, + 0x0040b700, + 0x0057f105, + 0x0153f00b, + 0xf10045d0, + 0xb6404057, + 0x53f10154, + 0x45d08080, + 0x1057f140, + 0x1253f111, + 0x8045d013, + 0x151457f1, + 0x171653f1, + 0xf1c045d0, + 0xf0260157, + 0x47f10153, + 0x44b60800, + 0x0045d006, + 0x03ab21f5, + 0x080c47f1, + 0x980644b6, + 0x45d00505, + 0x4040d000, + 0xd00457f0, + 0x40b78045, + 0x05980400, + 0x1054b604, + 0xb70045d0, + 0xf1050040, + 0xd0030057, + 0x57f10045, + 0x53f11110, + 0x45d01312, + 0x06059840, + 0x050040b7, + 0xf10045d0, + 0xf0260157, + 0x47f10153, + 0x44b60800, + 0x0045d006, + 0x21f500f8, + 0x3fc803ab, + 0x0e0bf400, + 0x018921f5, + 0x020047f1, + 0xf11e0ef4, + 0xb6081067, + 0x77f00664, + 0x11078001, + 0x981c0780, + 0x67d02007, + 0x4067d000, + 0x32f444bd, + 0xc854bd02, + 0x0bf4043f, + 0x8221f50a, + 0x0a0ef403, + 0x027621f5, + 0xf40749f0, + 0x57f00231, + 0x083fc82c, + 0xf50a0bf4, + 0xf4038221, + 0x21f50a0e, + 0x49f00276, + 0x0057f108, + 0x0654b608, + 0xd0210698, + 0x67f04056, + 0x0063f141, + 0x0546fd44, + 0xc80054d0, + 0x0bf40c3f, + 0xc521f507, + 0xf100f803, + 0xbd220027, + 0x0133f034, + 0xf80023d0, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, +}; diff --git a/drivers/gpu/drm/nouveau/nva3_pm.c b/drivers/gpu/drm/nouveau/nva3_pm.c new file mode 100644 index 00000000..e4b2b9e9 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nva3_pm.c @@ -0,0 +1,204 @@ +/* + * Copyright 2010 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include "drmP.h" +#include "nouveau_drv.h" +#include "nouveau_bios.h" +#include "nouveau_pm.h" + +/* This is actually a lot more complex than it appears here, but hopefully + * this should be able to deal with what the VBIOS leaves for us.. + * + * If not, well, I'll jump off that bridge when I come to it. + */ + +struct nva3_pm_state { + enum pll_types type; + u32 src0; + u32 src1; + u32 ctrl; + u32 coef; + u32 old_pnm; + u32 new_pnm; + u32 new_div; +}; + +static int +nva3_pm_pll_offset(u32 id) +{ + static const u32 pll_map[] = { + 0x00, PLL_CORE, + 0x01, PLL_SHADER, + 0x02, PLL_MEMORY, + 0x00, 0x00 + }; + const u32 *map = pll_map; + + while (map[1]) { + if (id == map[1]) + return map[0]; + map += 2; + } + + return -ENOENT; +} + +int +nva3_pm_clock_get(struct drm_device *dev, u32 id) +{ + u32 src0, src1, ctrl, coef; + struct pll_lims pll; + int ret, off; + int P, N, M; + + ret = get_pll_limits(dev, id, &pll); + if (ret) + return ret; + + off = nva3_pm_pll_offset(id); + if (off < 0) + return off; + + src0 = nv_rd32(dev, 0x4120 + (off * 4)); + src1 = nv_rd32(dev, 0x4160 + (off * 4)); + ctrl = nv_rd32(dev, pll.reg + 0); + coef = nv_rd32(dev, pll.reg + 4); + NV_DEBUG(dev, "PLL %02x: 0x%08x 0x%08x 0x%08x 0x%08x\n", + id, src0, src1, ctrl, coef); + + if (ctrl & 0x00000008) { + u32 div = ((src1 & 0x003c0000) >> 18) + 1; + return (pll.refclk * 2) / div; + } + + P = (coef & 0x003f0000) >> 16; + N = (coef & 0x0000ff00) >> 8; + M = (coef & 0x000000ff); + return pll.refclk * N / M / P; +} + +void * +nva3_pm_clock_pre(struct drm_device *dev, struct nouveau_pm_level *perflvl, + u32 id, int khz) +{ + struct nva3_pm_state *pll; + struct pll_lims limits; + int N, M, P, diff; + int ret, off; + + ret = get_pll_limits(dev, id, &limits); + if (ret < 0) + return (ret == -ENOENT) ? NULL : ERR_PTR(ret); + + off = nva3_pm_pll_offset(id); + if (id < 0) + return ERR_PTR(-EINVAL); + + + pll = kzalloc(sizeof(*pll), GFP_KERNEL); + if (!pll) + return ERR_PTR(-ENOMEM); + pll->type = id; + pll->src0 = 0x004120 + (off * 4); + pll->src1 = 0x004160 + (off * 4); + pll->ctrl = limits.reg + 0; + pll->coef = limits.reg + 4; + + /* If target clock is within [-2, 3) MHz of a divisor, we'll + * use that instead of calculating MNP values + */ + pll->new_div = min((limits.refclk * 2) / (khz - 2999), 16); + if (pll->new_div) { + diff = khz - ((limits.refclk * 2) / pll->new_div); + if (diff < -2000 || diff >= 3000) + pll->new_div = 0; + } + + if (!pll->new_div) { + ret = nva3_calc_pll(dev, &limits, khz, &N, NULL, &M, &P); + if (ret < 0) + return ERR_PTR(ret); + + pll->new_pnm = (P << 16) | (N << 8) | M; + pll->new_div = 2 - 1; + } else { + pll->new_pnm = 0; + pll->new_div--; + } + + if ((nv_rd32(dev, pll->src1) & 0x00000101) != 0x00000101) + pll->old_pnm = nv_rd32(dev, pll->coef); + return pll; +} + +void +nva3_pm_clock_set(struct drm_device *dev, void *pre_state) +{ + struct nva3_pm_state *pll = pre_state; + u32 ctrl = 0; + + /* For the memory clock, NVIDIA will build a "script" describing + * the reclocking process and ask PDAEMON to execute it. + */ + if (pll->type == PLL_MEMORY) { + nv_wr32(dev, 0x100210, 0); + nv_wr32(dev, 0x1002dc, 1); + nv_wr32(dev, 0x004018, 0x00001000); + ctrl = 0x18000100; + } + + if (pll->old_pnm || !pll->new_pnm) { + nv_mask(dev, pll->src1, 0x003c0101, 0x00000101 | + (pll->new_div << 18)); + nv_wr32(dev, pll->ctrl, 0x0001001d | ctrl); + nv_mask(dev, pll->ctrl, 0x00000001, 0x00000000); + } + + if (pll->new_pnm) { + nv_mask(dev, pll->src0, 0x00000101, 0x00000101); + nv_wr32(dev, pll->coef, pll->new_pnm); + nv_wr32(dev, pll->ctrl, 0x0001001d | ctrl); + nv_mask(dev, pll->ctrl, 0x00000010, 0x00000000); + nv_mask(dev, pll->ctrl, 0x00020010, 0x00020010); + nv_wr32(dev, pll->ctrl, 0x00010015 | ctrl); + nv_mask(dev, pll->src1, 0x00000100, 0x00000000); + nv_mask(dev, pll->src1, 0x00000001, 0x00000000); + if (pll->type == PLL_MEMORY) + nv_wr32(dev, 0x4018, 0x10005000); + } else { + nv_mask(dev, pll->ctrl, 0x00000001, 0x00000000); + nv_mask(dev, pll->src0, 0x00000100, 0x00000000); + nv_mask(dev, pll->src0, 0x00000001, 0x00000000); + if (pll->type == PLL_MEMORY) + nv_wr32(dev, 0x4018, 0x1000d000); + } + + if (pll->type == PLL_MEMORY) { + nv_wr32(dev, 0x1002dc, 0); + nv_wr32(dev, 0x100210, 0x80000000); + } + + kfree(pll); +} + diff --git a/drivers/gpu/drm/nouveau/nvc0_copy.c b/drivers/gpu/drm/nouveau/nvc0_copy.c new file mode 100644 index 00000000..208fa7ab --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvc0_copy.c @@ -0,0 +1,243 @@ +/* + * Copyright 2011 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include +#include "drmP.h" +#include "nouveau_drv.h" +#include "nouveau_util.h" +#include "nouveau_vm.h" +#include "nouveau_ramht.h" +#include "nvc0_copy.fuc.h" + +struct nvc0_copy_engine { + struct nouveau_exec_engine base; + u32 irq; + u32 pmc; + u32 fuc; + u32 ctx; +}; + +static int +nvc0_copy_context_new(struct nouveau_channel *chan, int engine) +{ + struct nvc0_copy_engine *pcopy = nv_engine(chan->dev, engine); + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *ramin = chan->ramin; + struct nouveau_gpuobj *ctx = NULL; + int ret; + + ret = nouveau_gpuobj_new(dev, NULL, 256, 256, + NVOBJ_FLAG_VM | NVOBJ_FLAG_VM_USER | + NVOBJ_FLAG_ZERO_ALLOC, &ctx); + if (ret) + return ret; + + nv_wo32(ramin, pcopy->ctx + 0, lower_32_bits(ctx->vinst)); + nv_wo32(ramin, pcopy->ctx + 4, upper_32_bits(ctx->vinst)); + dev_priv->engine.instmem.flush(dev); + + chan->engctx[engine] = ctx; + return 0; +} + +static int +nvc0_copy_object_new(struct nouveau_channel *chan, int engine, + u32 handle, u16 class) +{ + return 0; +} + +static void +nvc0_copy_context_del(struct nouveau_channel *chan, int engine) +{ + struct nvc0_copy_engine *pcopy = nv_engine(chan->dev, engine); + struct nouveau_gpuobj *ctx = chan->engctx[engine]; + struct drm_device *dev = chan->dev; + u32 inst; + + inst = (chan->ramin->vinst >> 12); + inst |= 0x40000000; + + /* disable fifo access */ + nv_wr32(dev, pcopy->fuc + 0x048, 0x00000000); + /* mark channel as unloaded if it's currently active */ + if (nv_rd32(dev, pcopy->fuc + 0x050) == inst) + nv_mask(dev, pcopy->fuc + 0x050, 0x40000000, 0x00000000); + /* mark next channel as invalid if it's about to be loaded */ + if (nv_rd32(dev, pcopy->fuc + 0x054) == inst) + nv_mask(dev, pcopy->fuc + 0x054, 0x40000000, 0x00000000); + /* restore fifo access */ + nv_wr32(dev, pcopy->fuc + 0x048, 0x00000003); + + nv_wo32(chan->ramin, pcopy->ctx + 0, 0x00000000); + nv_wo32(chan->ramin, pcopy->ctx + 4, 0x00000000); + nouveau_gpuobj_ref(NULL, &ctx); + + chan->engctx[engine] = ctx; +} + +static int +nvc0_copy_init(struct drm_device *dev, int engine) +{ + struct nvc0_copy_engine *pcopy = nv_engine(dev, engine); + int i; + + nv_mask(dev, 0x000200, pcopy->pmc, 0x00000000); + nv_mask(dev, 0x000200, pcopy->pmc, pcopy->pmc); + nv_wr32(dev, pcopy->fuc + 0x014, 0xffffffff); + + nv_wr32(dev, pcopy->fuc + 0x1c0, 0x01000000); + for (i = 0; i < sizeof(nvc0_pcopy_data) / 4; i++) + nv_wr32(dev, pcopy->fuc + 0x1c4, nvc0_pcopy_data[i]); + + nv_wr32(dev, pcopy->fuc + 0x180, 0x01000000); + for (i = 0; i < sizeof(nvc0_pcopy_code) / 4; i++) { + if ((i & 0x3f) == 0) + nv_wr32(dev, pcopy->fuc + 0x188, i >> 6); + nv_wr32(dev, pcopy->fuc + 0x184, nvc0_pcopy_code[i]); + } + + nv_wr32(dev, pcopy->fuc + 0x084, engine - NVOBJ_ENGINE_COPY0); + nv_wr32(dev, pcopy->fuc + 0x10c, 0x00000000); + nv_wr32(dev, pcopy->fuc + 0x104, 0x00000000); /* ENTRY */ + nv_wr32(dev, pcopy->fuc + 0x100, 0x00000002); /* TRIGGER */ + return 0; +} + +static int +nvc0_copy_fini(struct drm_device *dev, int engine) +{ + struct nvc0_copy_engine *pcopy = nv_engine(dev, engine); + + nv_mask(dev, pcopy->fuc + 0x048, 0x00000003, 0x00000000); + + /* trigger fuc context unload */ + nv_wait(dev, pcopy->fuc + 0x008, 0x0000000c, 0x00000000); + nv_mask(dev, pcopy->fuc + 0x054, 0x40000000, 0x00000000); + nv_wr32(dev, pcopy->fuc + 0x000, 0x00000008); + nv_wait(dev, pcopy->fuc + 0x008, 0x00000008, 0x00000000); + + nv_wr32(dev, pcopy->fuc + 0x014, 0xffffffff); + return 0; +} + +static struct nouveau_enum nvc0_copy_isr_error_name[] = { + { 0x0001, "ILLEGAL_MTHD" }, + { 0x0002, "INVALID_ENUM" }, + { 0x0003, "INVALID_BITFIELD" }, + {} +}; + +static void +nvc0_copy_isr(struct drm_device *dev, int engine) +{ + struct nvc0_copy_engine *pcopy = nv_engine(dev, engine); + u32 disp = nv_rd32(dev, pcopy->fuc + 0x01c); + u32 stat = nv_rd32(dev, pcopy->fuc + 0x008) & disp & ~(disp >> 16); + u64 inst = (u64)(nv_rd32(dev, pcopy->fuc + 0x050) & 0x0fffffff) << 12; + u32 chid = nvc0_graph_isr_chid(dev, inst); + u32 ssta = nv_rd32(dev, pcopy->fuc + 0x040) & 0x0000ffff; + u32 addr = nv_rd32(dev, pcopy->fuc + 0x040) >> 16; + u32 mthd = (addr & 0x07ff) << 2; + u32 subc = (addr & 0x3800) >> 11; + u32 data = nv_rd32(dev, pcopy->fuc + 0x044); + + if (stat & 0x00000040) { + NV_INFO(dev, "PCOPY: DISPATCH_ERROR ["); + nouveau_enum_print(nvc0_copy_isr_error_name, ssta); + printk("] ch %d [0x%010llx] subc %d mthd 0x%04x data 0x%08x\n", + chid, inst, subc, mthd, data); + nv_wr32(dev, pcopy->fuc + 0x004, 0x00000040); + stat &= ~0x00000040; + } + + if (stat) { + NV_INFO(dev, "PCOPY: unhandled intr 0x%08x\n", stat); + nv_wr32(dev, pcopy->fuc + 0x004, stat); + } +} + +static void +nvc0_copy_isr_0(struct drm_device *dev) +{ + nvc0_copy_isr(dev, NVOBJ_ENGINE_COPY0); +} + +static void +nvc0_copy_isr_1(struct drm_device *dev) +{ + nvc0_copy_isr(dev, NVOBJ_ENGINE_COPY1); +} + +static void +nvc0_copy_destroy(struct drm_device *dev, int engine) +{ + struct nvc0_copy_engine *pcopy = nv_engine(dev, engine); + + nouveau_irq_unregister(dev, pcopy->irq); + + if (engine == NVOBJ_ENGINE_COPY0) + NVOBJ_ENGINE_DEL(dev, COPY0); + else + NVOBJ_ENGINE_DEL(dev, COPY1); + kfree(pcopy); +} + +int +nvc0_copy_create(struct drm_device *dev, int engine) +{ + struct nvc0_copy_engine *pcopy; + + pcopy = kzalloc(sizeof(*pcopy), GFP_KERNEL); + if (!pcopy) + return -ENOMEM; + + pcopy->base.destroy = nvc0_copy_destroy; + pcopy->base.init = nvc0_copy_init; + pcopy->base.fini = nvc0_copy_fini; + pcopy->base.context_new = nvc0_copy_context_new; + pcopy->base.context_del = nvc0_copy_context_del; + pcopy->base.object_new = nvc0_copy_object_new; + + if (engine == 0) { + pcopy->irq = 5; + pcopy->pmc = 0x00000040; + pcopy->fuc = 0x104000; + pcopy->ctx = 0x0230; + nouveau_irq_register(dev, pcopy->irq, nvc0_copy_isr_0); + NVOBJ_ENGINE_ADD(dev, COPY0, &pcopy->base); + NVOBJ_CLASS(dev, 0x90b5, COPY0); + } else { + pcopy->irq = 6; + pcopy->pmc = 0x00000080; + pcopy->fuc = 0x105000; + pcopy->ctx = 0x0240; + nouveau_irq_register(dev, pcopy->irq, nvc0_copy_isr_1); + NVOBJ_ENGINE_ADD(dev, COPY1, &pcopy->base); + NVOBJ_CLASS(dev, 0x90b8, COPY1); + } + + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nvc0_copy.fuc.h b/drivers/gpu/drm/nouveau/nvc0_copy.fuc.h new file mode 100644 index 00000000..41990388 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvc0_copy.fuc.h @@ -0,0 +1,527 @@ +uint32_t nvc0_pcopy_data[] = { + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00010000, + 0x00000000, + 0x00000000, + 0x00010040, + 0x0001019f, + 0x00000000, + 0x00010050, + 0x000101a1, + 0x00000000, + 0x00070080, + 0x0000001c, + 0xfffff000, + 0x00000020, + 0xfff80000, + 0x00000024, + 0xffffe000, + 0x00000028, + 0xfffff800, + 0x0000002c, + 0xfffff000, + 0x00000030, + 0xfff80000, + 0x00000034, + 0xffffe000, + 0x00070088, + 0x00000048, + 0xfffff000, + 0x0000004c, + 0xfff80000, + 0x00000050, + 0xffffe000, + 0x00000054, + 0xfffff800, + 0x00000058, + 0xfffff000, + 0x0000005c, + 0xfff80000, + 0x00000060, + 0xffffe000, + 0x000200c0, + 0x000104b8, + 0x00000000, + 0x00010541, + 0x00000000, + 0x000e00c3, + 0x00000010, + 0xffffff00, + 0x00000014, + 0x0000000f, + 0x0000003c, + 0xffffff00, + 0x00000040, + 0x0000000f, + 0x00000018, + 0xfff80000, + 0x00000044, + 0xfff80000, + 0x00000074, + 0xffff0000, + 0x00000078, + 0xffffe000, + 0x00000068, + 0xfccc0000, + 0x0000006c, + 0x00000000, + 0x00000070, + 0x00000000, + 0x00000004, + 0xffffff00, + 0x00000008, + 0x00000000, + 0x0000000c, + 0x00000000, + 0x00000800, +}; + +uint32_t nvc0_pcopy_code[] = { + 0x04fe04bd, + 0x3517f000, + 0xf10010fe, + 0xf1040017, + 0xf0fff327, + 0x22d00023, + 0x0c25f0c0, + 0xf40012d0, + 0x17f11031, + 0x27f01200, + 0x0012d003, + 0xf40031f4, + 0x0ef40028, + 0x8001cffd, + 0xf40812c4, + 0x21f4060b, + 0x0412c4ca, + 0xf5070bf4, + 0xc4010221, + 0x01d00c11, + 0xf101f840, + 0xfe770047, + 0x47f1004b, + 0x44cf2100, + 0x0144f000, + 0xb60444b6, + 0xf7f13040, + 0xf4b6061c, + 0x1457f106, + 0x00f5d101, + 0xb6043594, + 0x57fe0250, + 0x0145fe00, + 0x010052b7, + 0x00ff67f1, + 0x56fd60bd, + 0x0253f004, + 0xf80545fa, + 0x0053f003, + 0xd100e7f0, + 0x549800fe, + 0x0845b600, + 0xb6015698, + 0x46fd1864, + 0x0047fe05, + 0xf00204b9, + 0x01f40643, + 0x0604fa09, + 0xfa060ef4, + 0x03f80504, + 0x27f100f8, + 0x23cf1400, + 0x1e3fc800, + 0xf4170bf4, + 0x21f40132, + 0x1e3af053, + 0xf00023d0, + 0x24d00147, + 0xcf00f880, + 0x3dc84023, + 0x090bf41e, + 0xf40131f4, + 0x37f05321, + 0x8023d002, + 0x37f100f8, + 0x32cf1900, + 0x0033cf40, + 0x07ff24e4, + 0xf11024b6, + 0xbd010057, + 0x5874bd64, + 0x57580056, + 0x0450b601, + 0xf40446b8, + 0x76bb4d08, + 0x0447b800, + 0xbb0f08f4, + 0x74b60276, + 0x0057bb03, + 0xbbdf0ef4, + 0x44b60246, + 0x0045bb03, + 0xfd014598, + 0x54b00453, + 0x201bf400, + 0x58004558, + 0x64b00146, + 0x091bf400, + 0xf4005380, + 0x32f4300e, + 0xf455f901, + 0x0ef40c01, + 0x0225f025, + 0xf10125f0, + 0xd0100047, + 0x43d00042, + 0x4027f040, + 0xcf0002d0, + 0x24f08002, + 0x0024b040, + 0xf1f71bf4, + 0xf01d0027, + 0x23d00137, + 0xf800f800, + 0x0027f100, + 0xf034bd22, + 0x23d00233, + 0xf400f800, + 0x01b0f030, + 0x0101b000, + 0xb00201b0, + 0x04980301, + 0x3045c71a, + 0xc70150b6, + 0x60b63446, + 0x3847c701, + 0xf40170b6, + 0x84bd0232, + 0x4ac494bd, + 0x0445b60f, + 0xa430b4bd, + 0x0f18f404, + 0xbbc0a5ff, + 0x31f400cb, + 0x220ef402, + 0xf00c1bf4, + 0xcbbb10c7, + 0x160ef400, + 0xf406a430, + 0xc7f00c18, + 0x00cbbb14, + 0xf1070ef4, + 0x380080c7, + 0x80b601c8, + 0x01b0b601, + 0xf404b5b8, + 0x90b6c308, + 0x0497b801, + 0xfdb208f4, + 0x06800065, + 0x1d08980e, + 0xf40068fd, + 0x64bd0502, + 0x800075fd, + 0x78fd1907, + 0x1057f100, + 0x0654b608, + 0xd00056d0, + 0x50b74057, + 0x06980800, + 0x0162b619, + 0x980864b6, + 0x72b60e07, + 0x0567fd01, + 0xb70056d0, + 0xb4010050, + 0x56d00060, + 0x0160b400, + 0xb44056d0, + 0x56d00260, + 0x0360b480, + 0xb7c056d0, + 0x98040050, + 0x56d01b06, + 0x1c069800, + 0xf44056d0, + 0x00f81030, + 0xc7075798, + 0x78c76879, + 0x0380b664, + 0xb06077c7, + 0x1bf40e76, + 0x0477f009, + 0xf00f0ef4, + 0x70b6027c, + 0x0947fd11, + 0x980677f0, + 0x5b980c5a, + 0x00abfd0e, + 0xbb01b7f0, + 0xb2b604b7, + 0xc4abff01, + 0x9805a7bb, + 0xe7f00d5d, + 0x04e8bb01, + 0xff01e2b6, + 0xd8bbb4de, + 0x01e0b605, + 0xbb0cef94, + 0xfefd02eb, + 0x026cf005, + 0x020860b7, + 0xd00864b6, + 0xb7bb006f, + 0x00cbbb04, + 0x98085f98, + 0xfbfd0e5b, + 0x01b7f000, + 0xb604b7bb, + 0xfbbb01b2, + 0x05f7bb00, + 0x5f98f0f9, + 0x01b7f009, + 0xb604b8bb, + 0xfbbb01b2, + 0x05f8bb00, + 0x78bbf0f9, + 0x0282b600, + 0xbb01b7f0, + 0xb9bb04b8, + 0x0b589804, + 0xbb01e7f0, + 0xe2b604e9, + 0xf48eff01, + 0xbb04f7bb, + 0x79bb00cf, + 0x0589bb00, + 0x90fcf0fc, + 0xbb00d9fd, + 0x89fd00ad, + 0x008ffd00, + 0xbb00a8bb, + 0x92b604a7, + 0x0497bb01, + 0x988069d0, + 0x58980557, + 0x00acbb04, + 0xb6007abb, + 0x84b60081, + 0x058bfd10, + 0x060062b7, + 0xb70067d0, + 0xd0040060, + 0x00f80068, + 0xb7026cf0, + 0xb6020260, + 0x57980864, + 0x0067d005, + 0x040060b7, + 0xb6045798, + 0x67d01074, + 0x0060b700, + 0x06579804, + 0xf80067d0, + 0xf900f900, + 0x0007f110, + 0x0604b608, + 0xf00001cf, + 0x1bf40114, + 0xfc10fcfa, + 0xc800f800, + 0x1bf40d34, + 0xd121f570, + 0x0c47f103, + 0x0644b608, + 0xb6020598, + 0x45d00450, + 0x4040d000, + 0xd00c57f0, + 0x40b78045, + 0x05980400, + 0x1054b601, + 0xb70045d0, + 0xf1050040, + 0xf00b0057, + 0x45d00153, + 0x4057f100, + 0x0154b640, + 0x808053f1, + 0xf14045d0, + 0xf1111057, + 0xd0131253, + 0x57f18045, + 0x53f11514, + 0x45d01716, + 0x0157f1c0, + 0x0153f026, + 0x080047f1, + 0xd00644b6, + 0x21f50045, + 0x47f103d1, + 0x44b6080c, + 0x02059806, + 0xd00045d0, + 0x57f04040, + 0x8045d004, + 0x040040b7, + 0xb6010598, + 0x45d01054, + 0x0040b700, + 0x0057f105, + 0x0045d003, + 0x111057f1, + 0x131253f1, + 0x984045d0, + 0x40b70305, + 0x45d00500, + 0x0157f100, + 0x0153f026, + 0x080047f1, + 0xd00644b6, + 0x00f80045, + 0x03d121f5, + 0xf4003fc8, + 0x21f50e0b, + 0x47f101af, + 0x0ef40200, + 0x1067f11e, + 0x0664b608, + 0x800177f0, + 0x07800e07, + 0x1d079819, + 0xd00067d0, + 0x44bd4067, + 0xbd0232f4, + 0x043fc854, + 0xf50a0bf4, + 0xf403a821, + 0x21f50a0e, + 0x49f0029c, + 0x0231f407, + 0xc82c57f0, + 0x0bf4083f, + 0xa821f50a, + 0x0a0ef403, + 0x029c21f5, + 0xf10849f0, + 0xb6080057, + 0x06980654, + 0x4056d01e, + 0xf14167f0, + 0xfd440063, + 0x54d00546, + 0x0c3fc800, + 0xf5070bf4, + 0xf803eb21, + 0x0027f100, + 0xf034bd22, + 0x23d00133, + 0x0000f800, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, +}; diff --git a/drivers/gpu/drm/nouveau/nvc0_fb.c b/drivers/gpu/drm/nouveau/nvc0_fb.c new file mode 100644 index 00000000..26a99602 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvc0_fb.c @@ -0,0 +1,38 @@ +/* + * Copyright 2010 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include "drmP.h" + +#include "nouveau_drv.h" + +int +nvc0_fb_init(struct drm_device *dev) +{ + return 0; +} + +void +nvc0_fb_takedown(struct drm_device *dev) +{ +} diff --git a/drivers/gpu/drm/nouveau/nvc0_fbcon.c b/drivers/gpu/drm/nouveau/nvc0_fbcon.c new file mode 100644 index 00000000..fa5d4c23 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvc0_fbcon.c @@ -0,0 +1,269 @@ +/* + * Copyright 2010 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include "drmP.h" +#include "nouveau_drv.h" +#include "nouveau_dma.h" +#include "nouveau_ramht.h" +#include "nouveau_fbcon.h" +#include "nouveau_mm.h" + +int +nvc0_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect) +{ + struct nouveau_fbdev *nfbdev = info->par; + struct drm_device *dev = nfbdev->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_channel *chan = dev_priv->channel; + int ret; + + ret = RING_SPACE(chan, rect->rop == ROP_COPY ? 7 : 11); + if (ret) + return ret; + + if (rect->rop != ROP_COPY) { + BEGIN_NVC0(chan, 2, NvSub2D, 0x02ac, 1); + OUT_RING (chan, 1); + } + BEGIN_NVC0(chan, 2, NvSub2D, 0x0588, 1); + if (info->fix.visual == FB_VISUAL_TRUECOLOR || + info->fix.visual == FB_VISUAL_DIRECTCOLOR) + OUT_RING (chan, ((uint32_t *)info->pseudo_palette)[rect->color]); + else + OUT_RING (chan, rect->color); + BEGIN_NVC0(chan, 2, NvSub2D, 0x0600, 4); + OUT_RING (chan, rect->dx); + OUT_RING (chan, rect->dy); + OUT_RING (chan, rect->dx + rect->width); + OUT_RING (chan, rect->dy + rect->height); + if (rect->rop != ROP_COPY) { + BEGIN_NVC0(chan, 2, NvSub2D, 0x02ac, 1); + OUT_RING (chan, 3); + } + FIRE_RING(chan); + return 0; +} + +int +nvc0_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *region) +{ + struct nouveau_fbdev *nfbdev = info->par; + struct drm_device *dev = nfbdev->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_channel *chan = dev_priv->channel; + int ret; + + ret = RING_SPACE(chan, 12); + if (ret) + return ret; + + BEGIN_NVC0(chan, 2, NvSub2D, 0x0110, 1); + OUT_RING (chan, 0); + BEGIN_NVC0(chan, 2, NvSub2D, 0x08b0, 4); + OUT_RING (chan, region->dx); + OUT_RING (chan, region->dy); + OUT_RING (chan, region->width); + OUT_RING (chan, region->height); + BEGIN_NVC0(chan, 2, NvSub2D, 0x08d0, 4); + OUT_RING (chan, 0); + OUT_RING (chan, region->sx); + OUT_RING (chan, 0); + OUT_RING (chan, region->sy); + FIRE_RING(chan); + return 0; +} + +int +nvc0_fbcon_imageblit(struct fb_info *info, const struct fb_image *image) +{ + struct nouveau_fbdev *nfbdev = info->par; + struct drm_device *dev = nfbdev->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_channel *chan = dev_priv->channel; + uint32_t width, dwords, *data = (uint32_t *)image->data; + uint32_t mask = ~(~0 >> (32 - info->var.bits_per_pixel)); + uint32_t *palette = info->pseudo_palette; + int ret; + + if (image->depth != 1) + return -ENODEV; + + ret = RING_SPACE(chan, 11); + if (ret) + return ret; + + width = ALIGN(image->width, 32); + dwords = (width * image->height) >> 5; + + BEGIN_NVC0(chan, 2, NvSub2D, 0x0814, 2); + if (info->fix.visual == FB_VISUAL_TRUECOLOR || + info->fix.visual == FB_VISUAL_DIRECTCOLOR) { + OUT_RING (chan, palette[image->bg_color] | mask); + OUT_RING (chan, palette[image->fg_color] | mask); + } else { + OUT_RING (chan, image->bg_color); + OUT_RING (chan, image->fg_color); + } + BEGIN_NVC0(chan, 2, NvSub2D, 0x0838, 2); + OUT_RING (chan, image->width); + OUT_RING (chan, image->height); + BEGIN_NVC0(chan, 2, NvSub2D, 0x0850, 4); + OUT_RING (chan, 0); + OUT_RING (chan, image->dx); + OUT_RING (chan, 0); + OUT_RING (chan, image->dy); + + while (dwords) { + int push = dwords > 2047 ? 2047 : dwords; + + ret = RING_SPACE(chan, push + 1); + if (ret) + return ret; + + dwords -= push; + + BEGIN_NVC0(chan, 6, NvSub2D, 0x0860, push); + OUT_RINGp(chan, data, push); + data += push; + } + + FIRE_RING(chan); + return 0; +} + +int +nvc0_fbcon_accel_init(struct fb_info *info) +{ + struct nouveau_fbdev *nfbdev = info->par; + struct drm_device *dev = nfbdev->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_channel *chan = dev_priv->channel; + struct nouveau_bo *nvbo = nfbdev->nouveau_fb.nvbo; + int ret, format; + + ret = nouveau_gpuobj_gr_new(chan, 0x902d, 0x902d); + if (ret) + return ret; + + switch (info->var.bits_per_pixel) { + case 8: + format = 0xf3; + break; + case 15: + format = 0xf8; + break; + case 16: + format = 0xe8; + break; + case 32: + switch (info->var.transp.length) { + case 0: /* depth 24 */ + case 8: /* depth 32, just use 24.. */ + format = 0xe6; + break; + case 2: /* depth 30 */ + format = 0xd1; + break; + default: + return -EINVAL; + } + break; + default: + return -EINVAL; + } + + ret = RING_SPACE(chan, 60); + if (ret) { + WARN_ON(1); + nouveau_fbcon_gpu_lockup(info); + return ret; + } + + BEGIN_NVC0(chan, 2, NvSub2D, 0x0000, 1); + OUT_RING (chan, 0x0000902d); + BEGIN_NVC0(chan, 2, NvSub2D, 0x0104, 2); + OUT_RING (chan, upper_32_bits(chan->notifier_bo->bo.offset)); + OUT_RING (chan, lower_32_bits(chan->notifier_bo->bo.offset)); + BEGIN_NVC0(chan, 2, NvSub2D, 0x0290, 1); + OUT_RING (chan, 0); + BEGIN_NVC0(chan, 2, NvSub2D, 0x0888, 1); + OUT_RING (chan, 1); + BEGIN_NVC0(chan, 2, NvSub2D, 0x02ac, 1); + OUT_RING (chan, 3); + BEGIN_NVC0(chan, 2, NvSub2D, 0x02a0, 1); + OUT_RING (chan, 0x55); + BEGIN_NVC0(chan, 2, NvSub2D, 0x08c0, 4); + OUT_RING (chan, 0); + OUT_RING (chan, 1); + OUT_RING (chan, 0); + OUT_RING (chan, 1); + BEGIN_NVC0(chan, 2, NvSub2D, 0x0580, 2); + OUT_RING (chan, 4); + OUT_RING (chan, format); + BEGIN_NVC0(chan, 2, NvSub2D, 0x02e8, 2); + OUT_RING (chan, 2); + OUT_RING (chan, 1); + + BEGIN_NVC0(chan, 2, NvSub2D, 0x0804, 1); + OUT_RING (chan, format); + BEGIN_NVC0(chan, 2, NvSub2D, 0x0800, 1); + OUT_RING (chan, 1); + BEGIN_NVC0(chan, 2, NvSub2D, 0x0808, 3); + OUT_RING (chan, 0); + OUT_RING (chan, 0); + OUT_RING (chan, 1); + BEGIN_NVC0(chan, 2, NvSub2D, 0x081c, 1); + OUT_RING (chan, 1); + BEGIN_NVC0(chan, 2, NvSub2D, 0x0840, 4); + OUT_RING (chan, 0); + OUT_RING (chan, 1); + OUT_RING (chan, 0); + OUT_RING (chan, 1); + BEGIN_NVC0(chan, 2, NvSub2D, 0x0200, 10); + OUT_RING (chan, format); + OUT_RING (chan, 1); + OUT_RING (chan, 0); + OUT_RING (chan, 1); + OUT_RING (chan, 0); + OUT_RING (chan, info->fix.line_length); + OUT_RING (chan, info->var.xres_virtual); + OUT_RING (chan, info->var.yres_virtual); + OUT_RING (chan, upper_32_bits(nvbo->vma.offset)); + OUT_RING (chan, lower_32_bits(nvbo->vma.offset)); + BEGIN_NVC0(chan, 2, NvSub2D, 0x0230, 10); + OUT_RING (chan, format); + OUT_RING (chan, 1); + OUT_RING (chan, 0); + OUT_RING (chan, 1); + OUT_RING (chan, 0); + OUT_RING (chan, info->fix.line_length); + OUT_RING (chan, info->var.xres_virtual); + OUT_RING (chan, info->var.yres_virtual); + OUT_RING (chan, upper_32_bits(nvbo->vma.offset)); + OUT_RING (chan, lower_32_bits(nvbo->vma.offset)); + FIRE_RING (chan); + + return 0; +} + diff --git a/drivers/gpu/drm/nouveau/nvc0_fifo.c b/drivers/gpu/drm/nouveau/nvc0_fifo.c new file mode 100644 index 00000000..fb4f5943 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvc0_fifo.c @@ -0,0 +1,508 @@ +/* + * Copyright 2010 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include "drmP.h" + +#include "nouveau_drv.h" +#include "nouveau_mm.h" + +static void nvc0_fifo_isr(struct drm_device *); + +struct nvc0_fifo_priv { + struct nouveau_gpuobj *playlist[2]; + int cur_playlist; + struct nouveau_vma user_vma; + int spoon_nr; +}; + +struct nvc0_fifo_chan { + struct nouveau_gpuobj *user; + struct nouveau_gpuobj *ramfc; +}; + +static void +nvc0_fifo_playlist_update(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_instmem_engine *pinstmem = &dev_priv->engine.instmem; + struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; + struct nvc0_fifo_priv *priv = pfifo->priv; + struct nouveau_gpuobj *cur; + int i, p; + + cur = priv->playlist[priv->cur_playlist]; + priv->cur_playlist = !priv->cur_playlist; + + for (i = 0, p = 0; i < 128; i++) { + if (!(nv_rd32(dev, 0x3004 + (i * 8)) & 1)) + continue; + nv_wo32(cur, p + 0, i); + nv_wo32(cur, p + 4, 0x00000004); + p += 8; + } + pinstmem->flush(dev); + + nv_wr32(dev, 0x002270, cur->vinst >> 12); + nv_wr32(dev, 0x002274, 0x01f00000 | (p >> 3)); + if (!nv_wait(dev, 0x00227c, 0x00100000, 0x00000000)) + NV_ERROR(dev, "PFIFO - playlist update failed\n"); +} + +void +nvc0_fifo_disable(struct drm_device *dev) +{ +} + +void +nvc0_fifo_enable(struct drm_device *dev) +{ +} + +bool +nvc0_fifo_reassign(struct drm_device *dev, bool enable) +{ + return false; +} + +bool +nvc0_fifo_cache_pull(struct drm_device *dev, bool enable) +{ + return false; +} + +int +nvc0_fifo_channel_id(struct drm_device *dev) +{ + return 127; +} + +int +nvc0_fifo_create_context(struct nouveau_channel *chan) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_instmem_engine *pinstmem = &dev_priv->engine.instmem; + struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; + struct nvc0_fifo_priv *priv = pfifo->priv; + struct nvc0_fifo_chan *fifoch; + u64 ib_virt = chan->pushbuf_base + chan->dma.ib_base * 4; + int ret; + + chan->fifo_priv = kzalloc(sizeof(*fifoch), GFP_KERNEL); + if (!chan->fifo_priv) + return -ENOMEM; + fifoch = chan->fifo_priv; + + /* allocate vram for control regs, map into polling area */ + ret = nouveau_gpuobj_new(dev, NULL, 0x1000, 0x1000, + NVOBJ_FLAG_ZERO_ALLOC, &fifoch->user); + if (ret) + goto error; + + nouveau_vm_map_at(&priv->user_vma, chan->id * 0x1000, + *(struct nouveau_mem **)fifoch->user->node); + + chan->user = ioremap_wc(pci_resource_start(dev->pdev, 1) + + priv->user_vma.offset + (chan->id * 0x1000), + PAGE_SIZE); + if (!chan->user) { + ret = -ENOMEM; + goto error; + } + + /* ramfc */ + ret = nouveau_gpuobj_new_fake(dev, chan->ramin->pinst, + chan->ramin->vinst, 0x100, + NVOBJ_FLAG_ZERO_ALLOC, &fifoch->ramfc); + if (ret) + goto error; + + nv_wo32(fifoch->ramfc, 0x08, lower_32_bits(fifoch->user->vinst)); + nv_wo32(fifoch->ramfc, 0x0c, upper_32_bits(fifoch->user->vinst)); + nv_wo32(fifoch->ramfc, 0x10, 0x0000face); + nv_wo32(fifoch->ramfc, 0x30, 0xfffff902); + nv_wo32(fifoch->ramfc, 0x48, lower_32_bits(ib_virt)); + nv_wo32(fifoch->ramfc, 0x4c, drm_order(chan->dma.ib_max + 1) << 16 | + upper_32_bits(ib_virt)); + nv_wo32(fifoch->ramfc, 0x54, 0x00000002); + nv_wo32(fifoch->ramfc, 0x84, 0x20400000); + nv_wo32(fifoch->ramfc, 0x94, 0x30000001); + nv_wo32(fifoch->ramfc, 0x9c, 0x00000100); + nv_wo32(fifoch->ramfc, 0xa4, 0x1f1f1f1f); + nv_wo32(fifoch->ramfc, 0xa8, 0x1f1f1f1f); + nv_wo32(fifoch->ramfc, 0xac, 0x0000001f); + nv_wo32(fifoch->ramfc, 0xb8, 0xf8000000); + nv_wo32(fifoch->ramfc, 0xf8, 0x10003080); /* 0x002310 */ + nv_wo32(fifoch->ramfc, 0xfc, 0x10000010); /* 0x002350 */ + pinstmem->flush(dev); + + nv_wr32(dev, 0x003000 + (chan->id * 8), 0xc0000000 | + (chan->ramin->vinst >> 12)); + nv_wr32(dev, 0x003004 + (chan->id * 8), 0x001f0001); + nvc0_fifo_playlist_update(dev); + return 0; + +error: + pfifo->destroy_context(chan); + return ret; +} + +void +nvc0_fifo_destroy_context(struct nouveau_channel *chan) +{ + struct drm_device *dev = chan->dev; + struct nvc0_fifo_chan *fifoch; + + nv_mask(dev, 0x003004 + (chan->id * 8), 0x00000001, 0x00000000); + nv_wr32(dev, 0x002634, chan->id); + if (!nv_wait(dev, 0x0002634, 0xffffffff, chan->id)) + NV_WARN(dev, "0x2634 != chid: 0x%08x\n", nv_rd32(dev, 0x2634)); + + nvc0_fifo_playlist_update(dev); + + nv_wr32(dev, 0x003000 + (chan->id * 8), 0x00000000); + + if (chan->user) { + iounmap(chan->user); + chan->user = NULL; + } + + fifoch = chan->fifo_priv; + chan->fifo_priv = NULL; + if (!fifoch) + return; + + nouveau_gpuobj_ref(NULL, &fifoch->ramfc); + nouveau_gpuobj_ref(NULL, &fifoch->user); + kfree(fifoch); +} + +int +nvc0_fifo_load_context(struct nouveau_channel *chan) +{ + return 0; +} + +int +nvc0_fifo_unload_context(struct drm_device *dev) +{ + int i; + + for (i = 0; i < 128; i++) { + if (!(nv_rd32(dev, 0x003004 + (i * 4)) & 1)) + continue; + + nv_mask(dev, 0x003004 + (i * 4), 0x00000001, 0x00000000); + nv_wr32(dev, 0x002634, i); + if (!nv_wait(dev, 0x002634, 0xffffffff, i)) { + NV_INFO(dev, "PFIFO: kick ch %d failed: 0x%08x\n", + i, nv_rd32(dev, 0x002634)); + return -EBUSY; + } + } + + return 0; +} + +static void +nvc0_fifo_destroy(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; + struct nvc0_fifo_priv *priv; + + priv = pfifo->priv; + if (!priv) + return; + + nouveau_vm_put(&priv->user_vma); + nouveau_gpuobj_ref(NULL, &priv->playlist[1]); + nouveau_gpuobj_ref(NULL, &priv->playlist[0]); + kfree(priv); +} + +void +nvc0_fifo_takedown(struct drm_device *dev) +{ + nv_wr32(dev, 0x002140, 0x00000000); + nvc0_fifo_destroy(dev); +} + +static int +nvc0_fifo_create(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; + struct nvc0_fifo_priv *priv; + int ret; + + priv = kzalloc(sizeof(*priv), GFP_KERNEL); + if (!priv) + return -ENOMEM; + pfifo->priv = priv; + + ret = nouveau_gpuobj_new(dev, NULL, 0x1000, 0x1000, 0, + &priv->playlist[0]); + if (ret) + goto error; + + ret = nouveau_gpuobj_new(dev, NULL, 0x1000, 0x1000, 0, + &priv->playlist[1]); + if (ret) + goto error; + + ret = nouveau_vm_get(dev_priv->bar1_vm, pfifo->channels * 0x1000, + 12, NV_MEM_ACCESS_RW, &priv->user_vma); + if (ret) + goto error; + + nouveau_irq_register(dev, 8, nvc0_fifo_isr); + NVOBJ_CLASS(dev, 0x506e, SW); /* nvsw */ + return 0; + +error: + nvc0_fifo_destroy(dev); + return ret; +} + +int +nvc0_fifo_init(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; + struct nouveau_channel *chan; + struct nvc0_fifo_priv *priv; + int ret, i; + + if (!pfifo->priv) { + ret = nvc0_fifo_create(dev); + if (ret) + return ret; + } + priv = pfifo->priv; + + /* reset PFIFO, enable all available PSUBFIFO areas */ + nv_mask(dev, 0x000200, 0x00000100, 0x00000000); + nv_mask(dev, 0x000200, 0x00000100, 0x00000100); + nv_wr32(dev, 0x000204, 0xffffffff); + nv_wr32(dev, 0x002204, 0xffffffff); + + priv->spoon_nr = hweight32(nv_rd32(dev, 0x002204)); + NV_DEBUG(dev, "PFIFO: %d subfifo(s)\n", priv->spoon_nr); + + /* assign engines to subfifos */ + if (priv->spoon_nr >= 3) { + nv_wr32(dev, 0x002208, ~(1 << 0)); /* PGRAPH */ + nv_wr32(dev, 0x00220c, ~(1 << 1)); /* PVP */ + nv_wr32(dev, 0x002210, ~(1 << 1)); /* PPP */ + nv_wr32(dev, 0x002214, ~(1 << 1)); /* PBSP */ + nv_wr32(dev, 0x002218, ~(1 << 2)); /* PCE0 */ + nv_wr32(dev, 0x00221c, ~(1 << 1)); /* PCE1 */ + } + + /* PSUBFIFO[n] */ + for (i = 0; i < 3; i++) { + nv_mask(dev, 0x04013c + (i * 0x2000), 0x10000100, 0x00000000); + nv_wr32(dev, 0x040108 + (i * 0x2000), 0xffffffff); /* INTR */ + nv_wr32(dev, 0x04010c + (i * 0x2000), 0xfffffeff); /* INTR_EN */ + } + + nv_mask(dev, 0x002200, 0x00000001, 0x00000001); + nv_wr32(dev, 0x002254, 0x10000000 | priv->user_vma.offset >> 12); + + nv_wr32(dev, 0x002a00, 0xffffffff); /* clears PFIFO.INTR bit 30 */ + nv_wr32(dev, 0x002100, 0xffffffff); + nv_wr32(dev, 0x002140, 0xbfffffff); + + /* restore PFIFO context table */ + for (i = 0; i < 128; i++) { + chan = dev_priv->channels.ptr[i]; + if (!chan || !chan->fifo_priv) + continue; + + nv_wr32(dev, 0x003000 + (i * 8), 0xc0000000 | + (chan->ramin->vinst >> 12)); + nv_wr32(dev, 0x003004 + (i * 8), 0x001f0001); + } + nvc0_fifo_playlist_update(dev); + + return 0; +} + +struct nouveau_enum nvc0_fifo_fault_unit[] = { + { 0x00, "PGRAPH" }, + { 0x03, "PEEPHOLE" }, + { 0x04, "BAR1" }, + { 0x05, "BAR3" }, + { 0x07, "PFIFO" }, + { 0x10, "PBSP" }, + { 0x11, "PPPP" }, + { 0x13, "PCOUNTER" }, + { 0x14, "PVP" }, + { 0x15, "PCOPY0" }, + { 0x16, "PCOPY1" }, + { 0x17, "PDAEMON" }, + {} +}; + +struct nouveau_enum nvc0_fifo_fault_reason[] = { + { 0x00, "PT_NOT_PRESENT" }, + { 0x01, "PT_TOO_SHORT" }, + { 0x02, "PAGE_NOT_PRESENT" }, + { 0x03, "VM_LIMIT_EXCEEDED" }, + { 0x04, "NO_CHANNEL" }, + { 0x05, "PAGE_SYSTEM_ONLY" }, + { 0x06, "PAGE_READ_ONLY" }, + { 0x0a, "COMPRESSED_SYSRAM" }, + { 0x0c, "INVALID_STORAGE_TYPE" }, + {} +}; + +struct nouveau_enum nvc0_fifo_fault_hubclient[] = { + { 0x01, "PCOPY0" }, + { 0x02, "PCOPY1" }, + { 0x04, "DISPATCH" }, + { 0x05, "CTXCTL" }, + { 0x06, "PFIFO" }, + { 0x07, "BAR_READ" }, + { 0x08, "BAR_WRITE" }, + { 0x0b, "PVP" }, + { 0x0c, "PPPP" }, + { 0x0d, "PBSP" }, + { 0x11, "PCOUNTER" }, + { 0x12, "PDAEMON" }, + { 0x14, "CCACHE" }, + { 0x15, "CCACHE_POST" }, + {} +}; + +struct nouveau_enum nvc0_fifo_fault_gpcclient[] = { + { 0x01, "TEX" }, + { 0x0c, "ESETUP" }, + { 0x0e, "CTXCTL" }, + { 0x0f, "PROP" }, + {} +}; + +struct nouveau_bitfield nvc0_fifo_subfifo_intr[] = { +/* { 0x00008000, "" } seen with null ib push */ + { 0x00200000, "ILLEGAL_MTHD" }, + { 0x00800000, "EMPTY_SUBC" }, + {} +}; + +static void +nvc0_fifo_isr_vm_fault(struct drm_device *dev, int unit) +{ + u32 inst = nv_rd32(dev, 0x2800 + (unit * 0x10)); + u32 valo = nv_rd32(dev, 0x2804 + (unit * 0x10)); + u32 vahi = nv_rd32(dev, 0x2808 + (unit * 0x10)); + u32 stat = nv_rd32(dev, 0x280c + (unit * 0x10)); + u32 client = (stat & 0x00001f00) >> 8; + + NV_INFO(dev, "PFIFO: %s fault at 0x%010llx [", + (stat & 0x00000080) ? "write" : "read", (u64)vahi << 32 | valo); + nouveau_enum_print(nvc0_fifo_fault_reason, stat & 0x0000000f); + printk("] from "); + nouveau_enum_print(nvc0_fifo_fault_unit, unit); + if (stat & 0x00000040) { + printk("/"); + nouveau_enum_print(nvc0_fifo_fault_hubclient, client); + } else { + printk("/GPC%d/", (stat & 0x1f000000) >> 24); + nouveau_enum_print(nvc0_fifo_fault_gpcclient, client); + } + printk(" on channel 0x%010llx\n", (u64)inst << 12); +} + +static void +nvc0_fifo_isr_subfifo_intr(struct drm_device *dev, int unit) +{ + u32 stat = nv_rd32(dev, 0x040108 + (unit * 0x2000)); + u32 addr = nv_rd32(dev, 0x0400c0 + (unit * 0x2000)); + u32 data = nv_rd32(dev, 0x0400c4 + (unit * 0x2000)); + u32 chid = nv_rd32(dev, 0x040120 + (unit * 0x2000)) & 0x7f; + u32 subc = (addr & 0x00070000); + u32 mthd = (addr & 0x00003ffc); + + NV_INFO(dev, "PSUBFIFO %d:", unit); + nouveau_bitfield_print(nvc0_fifo_subfifo_intr, stat); + NV_INFO(dev, "PSUBFIFO %d: ch %d subc %d mthd 0x%04x data 0x%08x\n", + unit, chid, subc, mthd, data); + + nv_wr32(dev, 0x0400c0 + (unit * 0x2000), 0x80600008); + nv_wr32(dev, 0x040108 + (unit * 0x2000), stat); +} + +static void +nvc0_fifo_isr(struct drm_device *dev) +{ + u32 stat = nv_rd32(dev, 0x002100); + + if (stat & 0x00000100) { + NV_INFO(dev, "PFIFO: unknown status 0x00000100\n"); + nv_wr32(dev, 0x002100, 0x00000100); + stat &= ~0x00000100; + } + + if (stat & 0x10000000) { + u32 units = nv_rd32(dev, 0x00259c); + u32 u = units; + + while (u) { + int i = ffs(u) - 1; + nvc0_fifo_isr_vm_fault(dev, i); + u &= ~(1 << i); + } + + nv_wr32(dev, 0x00259c, units); + stat &= ~0x10000000; + } + + if (stat & 0x20000000) { + u32 units = nv_rd32(dev, 0x0025a0); + u32 u = units; + + while (u) { + int i = ffs(u) - 1; + nvc0_fifo_isr_subfifo_intr(dev, i); + u &= ~(1 << i); + } + + nv_wr32(dev, 0x0025a0, units); + stat &= ~0x20000000; + } + + if (stat & 0x40000000) { + NV_INFO(dev, "PFIFO: unknown status 0x40000000\n"); + nv_mask(dev, 0x002a00, 0x00000000, 0x00000000); + stat &= ~0x40000000; + } + + if (stat) { + NV_INFO(dev, "PFIFO: unhandled status 0x%08x\n", stat); + nv_wr32(dev, 0x002100, stat); + nv_wr32(dev, 0x002140, 0); + } +} diff --git a/drivers/gpu/drm/nouveau/nvc0_graph.c b/drivers/gpu/drm/nouveau/nvc0_graph.c new file mode 100644 index 00000000..ca6db204 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvc0_graph.c @@ -0,0 +1,796 @@ +/* + * Copyright 2010 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include + +#include "drmP.h" + +#include "nouveau_drv.h" +#include "nouveau_mm.h" +#include "nvc0_graph.h" + +static int +nvc0_graph_load_context(struct nouveau_channel *chan) +{ + struct drm_device *dev = chan->dev; + + nv_wr32(dev, 0x409840, 0x00000030); + nv_wr32(dev, 0x409500, 0x80000000 | chan->ramin->vinst >> 12); + nv_wr32(dev, 0x409504, 0x00000003); + if (!nv_wait(dev, 0x409800, 0x00000010, 0x00000010)) + NV_ERROR(dev, "PGRAPH: load_ctx timeout\n"); + + return 0; +} + +static int +nvc0_graph_unload_context_to(struct drm_device *dev, u64 chan) +{ + nv_wr32(dev, 0x409840, 0x00000003); + nv_wr32(dev, 0x409500, 0x80000000 | chan >> 12); + nv_wr32(dev, 0x409504, 0x00000009); + if (!nv_wait(dev, 0x409800, 0x00000001, 0x00000000)) { + NV_ERROR(dev, "PGRAPH: unload_ctx timeout\n"); + return -EBUSY; + } + + return 0; +} + +static int +nvc0_graph_construct_context(struct nouveau_channel *chan) +{ + struct drm_nouveau_private *dev_priv = chan->dev->dev_private; + struct nvc0_graph_priv *priv = nv_engine(chan->dev, NVOBJ_ENGINE_GR); + struct nvc0_graph_chan *grch = chan->engctx[NVOBJ_ENGINE_GR]; + struct drm_device *dev = chan->dev; + int ret, i; + u32 *ctx; + + ctx = kmalloc(priv->grctx_size, GFP_KERNEL); + if (!ctx) + return -ENOMEM; + + nvc0_graph_load_context(chan); + + nv_wo32(grch->grctx, 0x1c, 1); + nv_wo32(grch->grctx, 0x20, 0); + nv_wo32(grch->grctx, 0x28, 0); + nv_wo32(grch->grctx, 0x2c, 0); + dev_priv->engine.instmem.flush(dev); + + ret = nvc0_grctx_generate(chan); + if (ret) { + kfree(ctx); + return ret; + } + + ret = nvc0_graph_unload_context_to(dev, chan->ramin->vinst); + if (ret) { + kfree(ctx); + return ret; + } + + for (i = 0; i < priv->grctx_size; i += 4) + ctx[i / 4] = nv_ro32(grch->grctx, i); + + priv->grctx_vals = ctx; + return 0; +} + +static int +nvc0_graph_create_context_mmio_list(struct nouveau_channel *chan) +{ + struct nvc0_graph_priv *priv = nv_engine(chan->dev, NVOBJ_ENGINE_GR); + struct nvc0_graph_chan *grch = chan->engctx[NVOBJ_ENGINE_GR]; + struct drm_device *dev = chan->dev; + int i = 0, gpc, tp, ret; + u32 magic; + + ret = nouveau_gpuobj_new(dev, NULL, 0x2000, 256, NVOBJ_FLAG_VM, + &grch->unk408004); + if (ret) + return ret; + + ret = nouveau_gpuobj_new(dev, NULL, 0x8000, 256, NVOBJ_FLAG_VM, + &grch->unk40800c); + if (ret) + return ret; + + ret = nouveau_gpuobj_new(dev, NULL, 384 * 1024, 4096, + NVOBJ_FLAG_VM | NVOBJ_FLAG_VM_USER, + &grch->unk418810); + if (ret) + return ret; + + ret = nouveau_gpuobj_new(dev, NULL, 0x1000, 0, NVOBJ_FLAG_VM, + &grch->mmio); + if (ret) + return ret; + + + nv_wo32(grch->mmio, i++ * 4, 0x00408004); + nv_wo32(grch->mmio, i++ * 4, grch->unk408004->vinst >> 8); + nv_wo32(grch->mmio, i++ * 4, 0x00408008); + nv_wo32(grch->mmio, i++ * 4, 0x80000018); + + nv_wo32(grch->mmio, i++ * 4, 0x0040800c); + nv_wo32(grch->mmio, i++ * 4, grch->unk40800c->vinst >> 8); + nv_wo32(grch->mmio, i++ * 4, 0x00408010); + nv_wo32(grch->mmio, i++ * 4, 0x80000000); + + nv_wo32(grch->mmio, i++ * 4, 0x00418810); + nv_wo32(grch->mmio, i++ * 4, 0x80000000 | grch->unk418810->vinst >> 12); + nv_wo32(grch->mmio, i++ * 4, 0x00419848); + nv_wo32(grch->mmio, i++ * 4, 0x10000000 | grch->unk418810->vinst >> 12); + + nv_wo32(grch->mmio, i++ * 4, 0x00419004); + nv_wo32(grch->mmio, i++ * 4, grch->unk40800c->vinst >> 8); + nv_wo32(grch->mmio, i++ * 4, 0x00419008); + nv_wo32(grch->mmio, i++ * 4, 0x00000000); + + nv_wo32(grch->mmio, i++ * 4, 0x00418808); + nv_wo32(grch->mmio, i++ * 4, grch->unk408004->vinst >> 8); + nv_wo32(grch->mmio, i++ * 4, 0x0041880c); + nv_wo32(grch->mmio, i++ * 4, 0x80000018); + + magic = 0x02180000; + nv_wo32(grch->mmio, i++ * 4, 0x00405830); + nv_wo32(grch->mmio, i++ * 4, magic); + for (gpc = 0; gpc < priv->gpc_nr; gpc++) { + for (tp = 0; tp < priv->tp_nr[gpc]; tp++, magic += 0x02fc) { + u32 reg = 0x504520 + (gpc * 0x8000) + (tp * 0x0800); + nv_wo32(grch->mmio, i++ * 4, reg); + nv_wo32(grch->mmio, i++ * 4, magic); + } + } + + grch->mmio_nr = i / 2; + return 0; +} + +static int +nvc0_graph_context_new(struct nouveau_channel *chan, int engine) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_instmem_engine *pinstmem = &dev_priv->engine.instmem; + struct nvc0_graph_priv *priv = nv_engine(dev, engine); + struct nvc0_graph_chan *grch; + struct nouveau_gpuobj *grctx; + int ret, i; + + grch = kzalloc(sizeof(*grch), GFP_KERNEL); + if (!grch) + return -ENOMEM; + chan->engctx[NVOBJ_ENGINE_GR] = grch; + + ret = nouveau_gpuobj_new(dev, NULL, priv->grctx_size, 256, + NVOBJ_FLAG_VM | NVOBJ_FLAG_ZERO_ALLOC, + &grch->grctx); + if (ret) + goto error; + grctx = grch->grctx; + + ret = nvc0_graph_create_context_mmio_list(chan); + if (ret) + goto error; + + nv_wo32(chan->ramin, 0x0210, lower_32_bits(grctx->vinst) | 4); + nv_wo32(chan->ramin, 0x0214, upper_32_bits(grctx->vinst)); + pinstmem->flush(dev); + + if (!priv->grctx_vals) { + ret = nvc0_graph_construct_context(chan); + if (ret) + goto error; + } + + for (i = 0; i < priv->grctx_size; i += 4) + nv_wo32(grctx, i, priv->grctx_vals[i / 4]); + + nv_wo32(grctx, 0xf4, 0); + nv_wo32(grctx, 0xf8, 0); + nv_wo32(grctx, 0x10, grch->mmio_nr); + nv_wo32(grctx, 0x14, lower_32_bits(grch->mmio->vinst)); + nv_wo32(grctx, 0x18, upper_32_bits(grch->mmio->vinst)); + nv_wo32(grctx, 0x1c, 1); + nv_wo32(grctx, 0x20, 0); + nv_wo32(grctx, 0x28, 0); + nv_wo32(grctx, 0x2c, 0); + pinstmem->flush(dev); + return 0; + +error: + priv->base.context_del(chan, engine); + return ret; +} + +static void +nvc0_graph_context_del(struct nouveau_channel *chan, int engine) +{ + struct nvc0_graph_chan *grch = chan->engctx[engine]; + + nouveau_gpuobj_ref(NULL, &grch->mmio); + nouveau_gpuobj_ref(NULL, &grch->unk418810); + nouveau_gpuobj_ref(NULL, &grch->unk40800c); + nouveau_gpuobj_ref(NULL, &grch->unk408004); + nouveau_gpuobj_ref(NULL, &grch->grctx); + chan->engctx[engine] = NULL; +} + +static int +nvc0_graph_object_new(struct nouveau_channel *chan, int engine, + u32 handle, u16 class) +{ + return 0; +} + +static int +nvc0_graph_fini(struct drm_device *dev, int engine) +{ + return 0; +} + +static int +nvc0_graph_mthd_page_flip(struct nouveau_channel *chan, + u32 class, u32 mthd, u32 data) +{ + nouveau_finish_page_flip(chan, NULL); + return 0; +} + +static void +nvc0_graph_init_obj418880(struct drm_device *dev) +{ + struct nvc0_graph_priv *priv = nv_engine(dev, NVOBJ_ENGINE_GR); + int i; + + nv_wr32(dev, GPC_BCAST(0x0880), 0x00000000); + nv_wr32(dev, GPC_BCAST(0x08a4), 0x00000000); + for (i = 0; i < 4; i++) + nv_wr32(dev, GPC_BCAST(0x0888) + (i * 4), 0x00000000); + nv_wr32(dev, GPC_BCAST(0x08b4), priv->unk4188b4->vinst >> 8); + nv_wr32(dev, GPC_BCAST(0x08b8), priv->unk4188b8->vinst >> 8); +} + +static void +nvc0_graph_init_regs(struct drm_device *dev) +{ + nv_wr32(dev, 0x400080, 0x003083c2); + nv_wr32(dev, 0x400088, 0x00006fe7); + nv_wr32(dev, 0x40008c, 0x00000000); + nv_wr32(dev, 0x400090, 0x00000030); + nv_wr32(dev, 0x40013c, 0x013901f7); + nv_wr32(dev, 0x400140, 0x00000100); + nv_wr32(dev, 0x400144, 0x00000000); + nv_wr32(dev, 0x400148, 0x00000110); + nv_wr32(dev, 0x400138, 0x00000000); + nv_wr32(dev, 0x400130, 0x00000000); + nv_wr32(dev, 0x400134, 0x00000000); + nv_wr32(dev, 0x400124, 0x00000002); +} + +static void +nvc0_graph_init_gpc_0(struct drm_device *dev) +{ + struct nvc0_graph_priv *priv = nv_engine(dev, NVOBJ_ENGINE_GR); + u32 data[TP_MAX / 8]; + u8 tpnr[GPC_MAX]; + int i, gpc, tpc; + + /* + * TP ROP UNKVAL(magic_not_rop_nr) + * 450: 4/0/0/0 2 3 + * 460: 3/4/0/0 4 1 + * 465: 3/4/4/0 4 7 + * 470: 3/3/4/4 5 5 + * 480: 3/4/4/4 6 6 + * + * magicgpc918 + * 450: 00200000 00000000001000000000000000000000 + * 460: 00124925 00000000000100100100100100100101 + * 465: 000ba2e9 00000000000010111010001011101001 + * 470: 00092493 00000000000010010010010010010011 + * 480: 00088889 00000000000010001000100010001001 + */ + + memset(data, 0x00, sizeof(data)); + memcpy(tpnr, priv->tp_nr, sizeof(priv->tp_nr)); + for (i = 0, gpc = -1; i < priv->tp_total; i++) { + do { + gpc = (gpc + 1) % priv->gpc_nr; + } while (!tpnr[gpc]); + tpc = priv->tp_nr[gpc] - tpnr[gpc]--; + + data[i / 8] |= tpc << ((i % 8) * 4); + } + + nv_wr32(dev, GPC_BCAST(0x0980), data[0]); + nv_wr32(dev, GPC_BCAST(0x0984), data[1]); + nv_wr32(dev, GPC_BCAST(0x0988), data[2]); + nv_wr32(dev, GPC_BCAST(0x098c), data[3]); + + for (gpc = 0; gpc < priv->gpc_nr; gpc++) { + nv_wr32(dev, GPC_UNIT(gpc, 0x0914), priv->magic_not_rop_nr << 8 | + priv->tp_nr[gpc]); + nv_wr32(dev, GPC_UNIT(gpc, 0x0910), 0x00040000 | priv->tp_total); + nv_wr32(dev, GPC_UNIT(gpc, 0x0918), priv->magicgpc918); + } + + nv_wr32(dev, GPC_BCAST(0x1bd4), priv->magicgpc918); + nv_wr32(dev, GPC_BCAST(0x08ac), priv->rop_nr); +} + +static void +nvc0_graph_init_units(struct drm_device *dev) +{ + nv_wr32(dev, 0x409c24, 0x000f0000); + nv_wr32(dev, 0x404000, 0xc0000000); /* DISPATCH */ + nv_wr32(dev, 0x404600, 0xc0000000); /* M2MF */ + nv_wr32(dev, 0x408030, 0xc0000000); + nv_wr32(dev, 0x40601c, 0xc0000000); + nv_wr32(dev, 0x404490, 0xc0000000); /* MACRO */ + nv_wr32(dev, 0x406018, 0xc0000000); + nv_wr32(dev, 0x405840, 0xc0000000); + nv_wr32(dev, 0x405844, 0x00ffffff); + nv_mask(dev, 0x419cc0, 0x00000008, 0x00000008); + nv_mask(dev, 0x419eb4, 0x00001000, 0x00001000); +} + +static void +nvc0_graph_init_gpc_1(struct drm_device *dev) +{ + struct nvc0_graph_priv *priv = nv_engine(dev, NVOBJ_ENGINE_GR); + int gpc, tp; + + for (gpc = 0; gpc < priv->gpc_nr; gpc++) { + nv_wr32(dev, GPC_UNIT(gpc, 0x0420), 0xc0000000); + nv_wr32(dev, GPC_UNIT(gpc, 0x0900), 0xc0000000); + nv_wr32(dev, GPC_UNIT(gpc, 0x1028), 0xc0000000); + nv_wr32(dev, GPC_UNIT(gpc, 0x0824), 0xc0000000); + for (tp = 0; tp < priv->tp_nr[gpc]; tp++) { + nv_wr32(dev, TP_UNIT(gpc, tp, 0x508), 0xffffffff); + nv_wr32(dev, TP_UNIT(gpc, tp, 0x50c), 0xffffffff); + nv_wr32(dev, TP_UNIT(gpc, tp, 0x224), 0xc0000000); + nv_wr32(dev, TP_UNIT(gpc, tp, 0x48c), 0xc0000000); + nv_wr32(dev, TP_UNIT(gpc, tp, 0x084), 0xc0000000); + nv_wr32(dev, TP_UNIT(gpc, tp, 0x644), 0x001ffffe); + nv_wr32(dev, TP_UNIT(gpc, tp, 0x64c), 0x0000000f); + } + nv_wr32(dev, GPC_UNIT(gpc, 0x2c90), 0xffffffff); + nv_wr32(dev, GPC_UNIT(gpc, 0x2c94), 0xffffffff); + } +} + +static void +nvc0_graph_init_rop(struct drm_device *dev) +{ + struct nvc0_graph_priv *priv = nv_engine(dev, NVOBJ_ENGINE_GR); + int rop; + + for (rop = 0; rop < priv->rop_nr; rop++) { + nv_wr32(dev, ROP_UNIT(rop, 0x144), 0xc0000000); + nv_wr32(dev, ROP_UNIT(rop, 0x070), 0xc0000000); + nv_wr32(dev, ROP_UNIT(rop, 0x204), 0xffffffff); + nv_wr32(dev, ROP_UNIT(rop, 0x208), 0xffffffff); + } +} + +static void +nvc0_graph_init_fuc(struct drm_device *dev, u32 fuc_base, + struct nvc0_graph_fuc *code, struct nvc0_graph_fuc *data) +{ + int i; + + nv_wr32(dev, fuc_base + 0x01c0, 0x01000000); + for (i = 0; i < data->size / 4; i++) + nv_wr32(dev, fuc_base + 0x01c4, data->data[i]); + + nv_wr32(dev, fuc_base + 0x0180, 0x01000000); + for (i = 0; i < code->size / 4; i++) { + if ((i & 0x3f) == 0) + nv_wr32(dev, fuc_base + 0x0188, i >> 6); + nv_wr32(dev, fuc_base + 0x0184, code->data[i]); + } +} + +static int +nvc0_graph_init_ctxctl(struct drm_device *dev) +{ + struct nvc0_graph_priv *priv = nv_engine(dev, NVOBJ_ENGINE_GR); + u32 r000260; + + /* load fuc microcode */ + r000260 = nv_mask(dev, 0x000260, 0x00000001, 0x00000000); + nvc0_graph_init_fuc(dev, 0x409000, &priv->fuc409c, &priv->fuc409d); + nvc0_graph_init_fuc(dev, 0x41a000, &priv->fuc41ac, &priv->fuc41ad); + nv_wr32(dev, 0x000260, r000260); + + /* start both of them running */ + nv_wr32(dev, 0x409840, 0xffffffff); + nv_wr32(dev, 0x41a10c, 0x00000000); + nv_wr32(dev, 0x40910c, 0x00000000); + nv_wr32(dev, 0x41a100, 0x00000002); + nv_wr32(dev, 0x409100, 0x00000002); + if (!nv_wait(dev, 0x409800, 0x00000001, 0x00000001)) + NV_INFO(dev, "0x409800 wait failed\n"); + + nv_wr32(dev, 0x409840, 0xffffffff); + nv_wr32(dev, 0x409500, 0x7fffffff); + nv_wr32(dev, 0x409504, 0x00000021); + + nv_wr32(dev, 0x409840, 0xffffffff); + nv_wr32(dev, 0x409500, 0x00000000); + nv_wr32(dev, 0x409504, 0x00000010); + if (!nv_wait_ne(dev, 0x409800, 0xffffffff, 0x00000000)) { + NV_ERROR(dev, "fuc09 req 0x10 timeout\n"); + return -EBUSY; + } + priv->grctx_size = nv_rd32(dev, 0x409800); + + nv_wr32(dev, 0x409840, 0xffffffff); + nv_wr32(dev, 0x409500, 0x00000000); + nv_wr32(dev, 0x409504, 0x00000016); + if (!nv_wait_ne(dev, 0x409800, 0xffffffff, 0x00000000)) { + NV_ERROR(dev, "fuc09 req 0x16 timeout\n"); + return -EBUSY; + } + + nv_wr32(dev, 0x409840, 0xffffffff); + nv_wr32(dev, 0x409500, 0x00000000); + nv_wr32(dev, 0x409504, 0x00000025); + if (!nv_wait_ne(dev, 0x409800, 0xffffffff, 0x00000000)) { + NV_ERROR(dev, "fuc09 req 0x25 timeout\n"); + return -EBUSY; + } + + return 0; +} + +static int +nvc0_graph_init(struct drm_device *dev, int engine) +{ + int ret; + + nv_mask(dev, 0x000200, 0x18001000, 0x00000000); + nv_mask(dev, 0x000200, 0x18001000, 0x18001000); + + nvc0_graph_init_obj418880(dev); + nvc0_graph_init_regs(dev); + /*nvc0_graph_init_unitplemented_magics(dev);*/ + nvc0_graph_init_gpc_0(dev); + /*nvc0_graph_init_unitplemented_c242(dev);*/ + + nv_wr32(dev, 0x400500, 0x00010001); + nv_wr32(dev, 0x400100, 0xffffffff); + nv_wr32(dev, 0x40013c, 0xffffffff); + + nvc0_graph_init_units(dev); + nvc0_graph_init_gpc_1(dev); + nvc0_graph_init_rop(dev); + + nv_wr32(dev, 0x400108, 0xffffffff); + nv_wr32(dev, 0x400138, 0xffffffff); + nv_wr32(dev, 0x400118, 0xffffffff); + nv_wr32(dev, 0x400130, 0xffffffff); + nv_wr32(dev, 0x40011c, 0xffffffff); + nv_wr32(dev, 0x400134, 0xffffffff); + nv_wr32(dev, 0x400054, 0x34ce3464); + + ret = nvc0_graph_init_ctxctl(dev); + if (ret) + return ret; + + return 0; +} + +int +nvc0_graph_isr_chid(struct drm_device *dev, u64 inst) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_channel *chan; + unsigned long flags; + int i; + + spin_lock_irqsave(&dev_priv->channels.lock, flags); + for (i = 0; i < dev_priv->engine.fifo.channels; i++) { + chan = dev_priv->channels.ptr[i]; + if (!chan || !chan->ramin) + continue; + + if (inst == chan->ramin->vinst) + break; + } + spin_unlock_irqrestore(&dev_priv->channels.lock, flags); + return i; +} + +static void +nvc0_graph_isr(struct drm_device *dev) +{ + u64 inst = (u64)(nv_rd32(dev, 0x409b00) & 0x0fffffff) << 12; + u32 chid = nvc0_graph_isr_chid(dev, inst); + u32 stat = nv_rd32(dev, 0x400100); + u32 addr = nv_rd32(dev, 0x400704); + u32 mthd = (addr & 0x00003ffc); + u32 subc = (addr & 0x00070000) >> 16; + u32 data = nv_rd32(dev, 0x400708); + u32 code = nv_rd32(dev, 0x400110); + u32 class = nv_rd32(dev, 0x404200 + (subc * 4)); + + if (stat & 0x00000010) { + if (nouveau_gpuobj_mthd_call2(dev, chid, class, mthd, data)) { + NV_INFO(dev, "PGRAPH: ILLEGAL_MTHD ch %d [0x%010llx] " + "subc %d class 0x%04x mthd 0x%04x " + "data 0x%08x\n", + chid, inst, subc, class, mthd, data); + } + nv_wr32(dev, 0x400100, 0x00000010); + stat &= ~0x00000010; + } + + if (stat & 0x00000020) { + NV_INFO(dev, "PGRAPH: ILLEGAL_CLASS ch %d [0x%010llx] subc %d " + "class 0x%04x mthd 0x%04x data 0x%08x\n", + chid, inst, subc, class, mthd, data); + nv_wr32(dev, 0x400100, 0x00000020); + stat &= ~0x00000020; + } + + if (stat & 0x00100000) { + NV_INFO(dev, "PGRAPH: DATA_ERROR ["); + nouveau_enum_print(nv50_data_error_names, code); + printk("] ch %d [0x%010llx] subc %d class 0x%04x " + "mthd 0x%04x data 0x%08x\n", + chid, inst, subc, class, mthd, data); + nv_wr32(dev, 0x400100, 0x00100000); + stat &= ~0x00100000; + } + + if (stat & 0x00200000) { + u32 trap = nv_rd32(dev, 0x400108); + NV_INFO(dev, "PGRAPH: TRAP ch %d status 0x%08x\n", chid, trap); + nv_wr32(dev, 0x400108, trap); + nv_wr32(dev, 0x400100, 0x00200000); + stat &= ~0x00200000; + } + + if (stat & 0x00080000) { + u32 ustat = nv_rd32(dev, 0x409c18); + + NV_INFO(dev, "PGRAPH: CTXCTRL ustat 0x%08x\n", ustat); + + nv_wr32(dev, 0x409c20, ustat); + nv_wr32(dev, 0x400100, 0x00080000); + stat &= ~0x00080000; + } + + if (stat) { + NV_INFO(dev, "PGRAPH: unknown stat 0x%08x\n", stat); + nv_wr32(dev, 0x400100, stat); + } + + nv_wr32(dev, 0x400500, 0x00010001); +} + +static void +nvc0_runk140_isr(struct drm_device *dev) +{ + u32 units = nv_rd32(dev, 0x00017c) & 0x1f; + + while (units) { + u32 unit = ffs(units) - 1; + u32 reg = 0x140000 + unit * 0x2000; + u32 st0 = nv_mask(dev, reg + 0x1020, 0, 0); + u32 st1 = nv_mask(dev, reg + 0x1420, 0, 0); + + NV_INFO(dev, "PRUNK140: %d 0x%08x 0x%08x\n", unit, st0, st1); + units &= ~(1 << unit); + } +} + +static int +nvc0_graph_create_fw(struct drm_device *dev, const char *fwname, + struct nvc0_graph_fuc *fuc) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + const struct firmware *fw; + char f[32]; + int ret; + + snprintf(f, sizeof(f), "nouveau/nv%02x_%s", dev_priv->chipset, fwname); + ret = request_firmware(&fw, f, &dev->pdev->dev); + if (ret) { + snprintf(f, sizeof(f), "nouveau/%s", fwname); + ret = request_firmware(&fw, f, &dev->pdev->dev); + if (ret) { + NV_ERROR(dev, "failed to load %s\n", fwname); + return ret; + } + } + + fuc->size = fw->size; + fuc->data = kmemdup(fw->data, fuc->size, GFP_KERNEL); + release_firmware(fw); + return (fuc->data != NULL) ? 0 : -ENOMEM; +} + +static void +nvc0_graph_destroy_fw(struct nvc0_graph_fuc *fuc) +{ + if (fuc->data) { + kfree(fuc->data); + fuc->data = NULL; + } +} + +static void +nvc0_graph_destroy(struct drm_device *dev, int engine) +{ + struct nvc0_graph_priv *priv = nv_engine(dev, engine); + + nvc0_graph_destroy_fw(&priv->fuc409c); + nvc0_graph_destroy_fw(&priv->fuc409d); + nvc0_graph_destroy_fw(&priv->fuc41ac); + nvc0_graph_destroy_fw(&priv->fuc41ad); + + nouveau_irq_unregister(dev, 12); + nouveau_irq_unregister(dev, 25); + + nouveau_gpuobj_ref(NULL, &priv->unk4188b8); + nouveau_gpuobj_ref(NULL, &priv->unk4188b4); + + if (priv->grctx_vals) + kfree(priv->grctx_vals); + + NVOBJ_ENGINE_DEL(dev, GR); + kfree(priv); +} + +int +nvc0_graph_create(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nvc0_graph_priv *priv; + int ret, gpc, i; + + switch (dev_priv->chipset) { + case 0xc0: + case 0xc3: + case 0xc4: + break; + default: + NV_ERROR(dev, "PGRAPH: unsupported chipset, please report!\n"); + return 0; + } + + priv = kzalloc(sizeof(*priv), GFP_KERNEL); + if (!priv) + return -ENOMEM; + + priv->base.destroy = nvc0_graph_destroy; + priv->base.init = nvc0_graph_init; + priv->base.fini = nvc0_graph_fini; + priv->base.context_new = nvc0_graph_context_new; + priv->base.context_del = nvc0_graph_context_del; + priv->base.object_new = nvc0_graph_object_new; + + NVOBJ_ENGINE_ADD(dev, GR, &priv->base); + nouveau_irq_register(dev, 12, nvc0_graph_isr); + nouveau_irq_register(dev, 25, nvc0_runk140_isr); + + if (nvc0_graph_create_fw(dev, "fuc409c", &priv->fuc409c) || + nvc0_graph_create_fw(dev, "fuc409d", &priv->fuc409d) || + nvc0_graph_create_fw(dev, "fuc41ac", &priv->fuc41ac) || + nvc0_graph_create_fw(dev, "fuc41ad", &priv->fuc41ad)) { + ret = 0; + goto error; + } + + + ret = nouveau_gpuobj_new(dev, NULL, 0x1000, 256, 0, &priv->unk4188b4); + if (ret) + goto error; + + ret = nouveau_gpuobj_new(dev, NULL, 0x1000, 256, 0, &priv->unk4188b8); + if (ret) + goto error; + + for (i = 0; i < 0x1000; i += 4) { + nv_wo32(priv->unk4188b4, i, 0x00000010); + nv_wo32(priv->unk4188b8, i, 0x00000010); + } + + priv->gpc_nr = nv_rd32(dev, 0x409604) & 0x0000001f; + priv->rop_nr = (nv_rd32(dev, 0x409604) & 0x001f0000) >> 16; + for (gpc = 0; gpc < priv->gpc_nr; gpc++) { + priv->tp_nr[gpc] = nv_rd32(dev, GPC_UNIT(gpc, 0x2608)); + priv->tp_total += priv->tp_nr[gpc]; + } + + /*XXX: these need figuring out... */ + switch (dev_priv->chipset) { + case 0xc0: + if (priv->tp_total == 11) { /* 465, 3/4/4/0, 4 */ + priv->magic_not_rop_nr = 0x07; + /* filled values up to tp_total, the rest 0 */ + priv->magicgpc918 = 0x000ba2e9; + } else + if (priv->tp_total == 14) { /* 470, 3/3/4/4, 5 */ + priv->magic_not_rop_nr = 0x05; + priv->magicgpc918 = 0x00092493; + } else + if (priv->tp_total == 15) { /* 480, 3/4/4/4, 6 */ + priv->magic_not_rop_nr = 0x06; + priv->magicgpc918 = 0x00088889; + } + break; + case 0xc3: /* 450, 4/0/0/0, 2 */ + priv->magic_not_rop_nr = 0x03; + priv->magicgpc918 = 0x00200000; + break; + case 0xc4: /* 460, 3/4/0/0, 4 */ + priv->magic_not_rop_nr = 0x01; + priv->magicgpc918 = 0x00124925; + break; + } + + if (!priv->magic_not_rop_nr) { + NV_ERROR(dev, "PGRAPH: unknown config: %d/%d/%d/%d, %d\n", + priv->tp_nr[0], priv->tp_nr[1], priv->tp_nr[2], + priv->tp_nr[3], priv->rop_nr); + /* use 0xc3's values... */ + priv->magic_not_rop_nr = 0x03; + priv->magicgpc918 = 0x00200000; + } + + NVOBJ_CLASS(dev, 0x902d, GR); /* 2D */ + NVOBJ_CLASS(dev, 0x9039, GR); /* M2MF */ + NVOBJ_MTHD (dev, 0x9039, 0x0500, nvc0_graph_mthd_page_flip); + NVOBJ_CLASS(dev, 0x9097, GR); /* 3D */ + NVOBJ_CLASS(dev, 0x90c0, GR); /* COMPUTE */ + return 0; + +error: + nvc0_graph_destroy(dev, NVOBJ_ENGINE_GR); + return ret; +} + +MODULE_FIRMWARE("nouveau/nvc0_fuc409c"); +MODULE_FIRMWARE("nouveau/nvc0_fuc409d"); +MODULE_FIRMWARE("nouveau/nvc0_fuc41ac"); +MODULE_FIRMWARE("nouveau/nvc0_fuc41ad"); +MODULE_FIRMWARE("nouveau/nvc3_fuc409c"); +MODULE_FIRMWARE("nouveau/nvc3_fuc409d"); +MODULE_FIRMWARE("nouveau/nvc3_fuc41ac"); +MODULE_FIRMWARE("nouveau/nvc3_fuc41ad"); +MODULE_FIRMWARE("nouveau/nvc4_fuc409c"); +MODULE_FIRMWARE("nouveau/nvc4_fuc409d"); +MODULE_FIRMWARE("nouveau/nvc4_fuc41ac"); +MODULE_FIRMWARE("nouveau/nvc4_fuc41ad"); +MODULE_FIRMWARE("nouveau/fuc409c"); +MODULE_FIRMWARE("nouveau/fuc409d"); +MODULE_FIRMWARE("nouveau/fuc41ac"); +MODULE_FIRMWARE("nouveau/fuc41ad"); diff --git a/drivers/gpu/drm/nouveau/nvc0_graph.h b/drivers/gpu/drm/nouveau/nvc0_graph.h new file mode 100644 index 00000000..f5d184e0 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvc0_graph.h @@ -0,0 +1,75 @@ +/* + * Copyright 2010 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#ifndef __NVC0_GRAPH_H__ +#define __NVC0_GRAPH_H__ + +#define GPC_MAX 4 +#define TP_MAX 32 + +#define ROP_BCAST(r) (0x408800 + (r)) +#define ROP_UNIT(u, r) (0x410000 + (u) * 0x400 + (r)) +#define GPC_BCAST(r) (0x418000 + (r)) +#define GPC_UNIT(t, r) (0x500000 + (t) * 0x8000 + (r)) +#define TP_UNIT(t, m, r) (0x504000 + (t) * 0x8000 + (m) * 0x800 + (r)) + +struct nvc0_graph_fuc { + u32 *data; + u32 size; +}; + +struct nvc0_graph_priv { + struct nouveau_exec_engine base; + + struct nvc0_graph_fuc fuc409c; + struct nvc0_graph_fuc fuc409d; + struct nvc0_graph_fuc fuc41ac; + struct nvc0_graph_fuc fuc41ad; + + u8 gpc_nr; + u8 rop_nr; + u8 tp_nr[GPC_MAX]; + u8 tp_total; + + u32 grctx_size; + u32 *grctx_vals; + struct nouveau_gpuobj *unk4188b4; + struct nouveau_gpuobj *unk4188b8; + + u8 magic_not_rop_nr; + u32 magicgpc918; +}; + +struct nvc0_graph_chan { + struct nouveau_gpuobj *grctx; + struct nouveau_gpuobj *unk408004; /* 0x418810 too */ + struct nouveau_gpuobj *unk40800c; /* 0x419004 too */ + struct nouveau_gpuobj *unk418810; /* 0x419848 too */ + struct nouveau_gpuobj *mmio; + int mmio_nr; +}; + +int nvc0_grctx_generate(struct nouveau_channel *); + +#endif diff --git a/drivers/gpu/drm/nouveau/nvc0_grctx.c b/drivers/gpu/drm/nouveau/nvc0_grctx.c new file mode 100644 index 00000000..6df06611 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvc0_grctx.c @@ -0,0 +1,2874 @@ +/* + * Copyright 2010 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include "drmP.h" +#include "nouveau_drv.h" +#include "nouveau_mm.h" +#include "nvc0_graph.h" + +static void +nv_icmd(struct drm_device *dev, u32 icmd, u32 data) +{ + nv_wr32(dev, 0x400204, data); + nv_wr32(dev, 0x400200, icmd); + while (nv_rd32(dev, 0x400700) & 2) {} +} + +static void +nv_mthd(struct drm_device *dev, u32 class, u32 mthd, u32 data) +{ + nv_wr32(dev, 0x40448c, data); + nv_wr32(dev, 0x404488, 0x80000000 | (mthd << 14) | class); +} + +static void +nvc0_grctx_generate_9097(struct drm_device *dev) +{ + nv_mthd(dev, 0x9097, 0x0800, 0x00000000); + nv_mthd(dev, 0x9097, 0x0840, 0x00000000); + nv_mthd(dev, 0x9097, 0x0880, 0x00000000); + nv_mthd(dev, 0x9097, 0x08c0, 0x00000000); + nv_mthd(dev, 0x9097, 0x0900, 0x00000000); + nv_mthd(dev, 0x9097, 0x0940, 0x00000000); + nv_mthd(dev, 0x9097, 0x0980, 0x00000000); + nv_mthd(dev, 0x9097, 0x09c0, 0x00000000); + nv_mthd(dev, 0x9097, 0x0804, 0x00000000); + nv_mthd(dev, 0x9097, 0x0844, 0x00000000); + nv_mthd(dev, 0x9097, 0x0884, 0x00000000); + nv_mthd(dev, 0x9097, 0x08c4, 0x00000000); + nv_mthd(dev, 0x9097, 0x0904, 0x00000000); + nv_mthd(dev, 0x9097, 0x0944, 0x00000000); + nv_mthd(dev, 0x9097, 0x0984, 0x00000000); + nv_mthd(dev, 0x9097, 0x09c4, 0x00000000); + nv_mthd(dev, 0x9097, 0x0808, 0x00000400); + nv_mthd(dev, 0x9097, 0x0848, 0x00000400); + nv_mthd(dev, 0x9097, 0x0888, 0x00000400); + nv_mthd(dev, 0x9097, 0x08c8, 0x00000400); + nv_mthd(dev, 0x9097, 0x0908, 0x00000400); + nv_mthd(dev, 0x9097, 0x0948, 0x00000400); + nv_mthd(dev, 0x9097, 0x0988, 0x00000400); + nv_mthd(dev, 0x9097, 0x09c8, 0x00000400); + nv_mthd(dev, 0x9097, 0x080c, 0x00000300); + nv_mthd(dev, 0x9097, 0x084c, 0x00000300); + nv_mthd(dev, 0x9097, 0x088c, 0x00000300); + nv_mthd(dev, 0x9097, 0x08cc, 0x00000300); + nv_mthd(dev, 0x9097, 0x090c, 0x00000300); + nv_mthd(dev, 0x9097, 0x094c, 0x00000300); + nv_mthd(dev, 0x9097, 0x098c, 0x00000300); + nv_mthd(dev, 0x9097, 0x09cc, 0x00000300); + nv_mthd(dev, 0x9097, 0x0810, 0x000000cf); + nv_mthd(dev, 0x9097, 0x0850, 0x00000000); + nv_mthd(dev, 0x9097, 0x0890, 0x00000000); + nv_mthd(dev, 0x9097, 0x08d0, 0x00000000); + nv_mthd(dev, 0x9097, 0x0910, 0x00000000); + nv_mthd(dev, 0x9097, 0x0950, 0x00000000); + nv_mthd(dev, 0x9097, 0x0990, 0x00000000); + nv_mthd(dev, 0x9097, 0x09d0, 0x00000000); + nv_mthd(dev, 0x9097, 0x0814, 0x00000040); + nv_mthd(dev, 0x9097, 0x0854, 0x00000040); + nv_mthd(dev, 0x9097, 0x0894, 0x00000040); + nv_mthd(dev, 0x9097, 0x08d4, 0x00000040); + nv_mthd(dev, 0x9097, 0x0914, 0x00000040); + nv_mthd(dev, 0x9097, 0x0954, 0x00000040); + nv_mthd(dev, 0x9097, 0x0994, 0x00000040); + nv_mthd(dev, 0x9097, 0x09d4, 0x00000040); + nv_mthd(dev, 0x9097, 0x0818, 0x00000001); + nv_mthd(dev, 0x9097, 0x0858, 0x00000001); + nv_mthd(dev, 0x9097, 0x0898, 0x00000001); + nv_mthd(dev, 0x9097, 0x08d8, 0x00000001); + nv_mthd(dev, 0x9097, 0x0918, 0x00000001); + nv_mthd(dev, 0x9097, 0x0958, 0x00000001); + nv_mthd(dev, 0x9097, 0x0998, 0x00000001); + nv_mthd(dev, 0x9097, 0x09d8, 0x00000001); + nv_mthd(dev, 0x9097, 0x081c, 0x00000000); + nv_mthd(dev, 0x9097, 0x085c, 0x00000000); + nv_mthd(dev, 0x9097, 0x089c, 0x00000000); + nv_mthd(dev, 0x9097, 0x08dc, 0x00000000); + nv_mthd(dev, 0x9097, 0x091c, 0x00000000); + nv_mthd(dev, 0x9097, 0x095c, 0x00000000); + nv_mthd(dev, 0x9097, 0x099c, 0x00000000); + nv_mthd(dev, 0x9097, 0x09dc, 0x00000000); + nv_mthd(dev, 0x9097, 0x0820, 0x00000000); + nv_mthd(dev, 0x9097, 0x0860, 0x00000000); + nv_mthd(dev, 0x9097, 0x08a0, 0x00000000); + nv_mthd(dev, 0x9097, 0x08e0, 0x00000000); + nv_mthd(dev, 0x9097, 0x0920, 0x00000000); + nv_mthd(dev, 0x9097, 0x0960, 0x00000000); + nv_mthd(dev, 0x9097, 0x09a0, 0x00000000); + nv_mthd(dev, 0x9097, 0x09e0, 0x00000000); + nv_mthd(dev, 0x9097, 0x2700, 0x00000000); + nv_mthd(dev, 0x9097, 0x2720, 0x00000000); + nv_mthd(dev, 0x9097, 0x2740, 0x00000000); + nv_mthd(dev, 0x9097, 0x2760, 0x00000000); + nv_mthd(dev, 0x9097, 0x2780, 0x00000000); + nv_mthd(dev, 0x9097, 0x27a0, 0x00000000); + nv_mthd(dev, 0x9097, 0x27c0, 0x00000000); + nv_mthd(dev, 0x9097, 0x27e0, 0x00000000); + nv_mthd(dev, 0x9097, 0x2704, 0x00000000); + nv_mthd(dev, 0x9097, 0x2724, 0x00000000); + nv_mthd(dev, 0x9097, 0x2744, 0x00000000); + nv_mthd(dev, 0x9097, 0x2764, 0x00000000); + nv_mthd(dev, 0x9097, 0x2784, 0x00000000); + nv_mthd(dev, 0x9097, 0x27a4, 0x00000000); + nv_mthd(dev, 0x9097, 0x27c4, 0x00000000); + nv_mthd(dev, 0x9097, 0x27e4, 0x00000000); + nv_mthd(dev, 0x9097, 0x2708, 0x00000000); + nv_mthd(dev, 0x9097, 0x2728, 0x00000000); + nv_mthd(dev, 0x9097, 0x2748, 0x00000000); + nv_mthd(dev, 0x9097, 0x2768, 0x00000000); + nv_mthd(dev, 0x9097, 0x2788, 0x00000000); + nv_mthd(dev, 0x9097, 0x27a8, 0x00000000); + nv_mthd(dev, 0x9097, 0x27c8, 0x00000000); + nv_mthd(dev, 0x9097, 0x27e8, 0x00000000); + nv_mthd(dev, 0x9097, 0x270c, 0x00000000); + nv_mthd(dev, 0x9097, 0x272c, 0x00000000); + nv_mthd(dev, 0x9097, 0x274c, 0x00000000); + nv_mthd(dev, 0x9097, 0x276c, 0x00000000); + nv_mthd(dev, 0x9097, 0x278c, 0x00000000); + nv_mthd(dev, 0x9097, 0x27ac, 0x00000000); + nv_mthd(dev, 0x9097, 0x27cc, 0x00000000); + nv_mthd(dev, 0x9097, 0x27ec, 0x00000000); + nv_mthd(dev, 0x9097, 0x2710, 0x00014000); + nv_mthd(dev, 0x9097, 0x2730, 0x00014000); + nv_mthd(dev, 0x9097, 0x2750, 0x00014000); + nv_mthd(dev, 0x9097, 0x2770, 0x00014000); + nv_mthd(dev, 0x9097, 0x2790, 0x00014000); + nv_mthd(dev, 0x9097, 0x27b0, 0x00014000); + nv_mthd(dev, 0x9097, 0x27d0, 0x00014000); + nv_mthd(dev, 0x9097, 0x27f0, 0x00014000); + nv_mthd(dev, 0x9097, 0x2714, 0x00000040); + nv_mthd(dev, 0x9097, 0x2734, 0x00000040); + nv_mthd(dev, 0x9097, 0x2754, 0x00000040); + nv_mthd(dev, 0x9097, 0x2774, 0x00000040); + nv_mthd(dev, 0x9097, 0x2794, 0x00000040); + nv_mthd(dev, 0x9097, 0x27b4, 0x00000040); + nv_mthd(dev, 0x9097, 0x27d4, 0x00000040); + nv_mthd(dev, 0x9097, 0x27f4, 0x00000040); + nv_mthd(dev, 0x9097, 0x1c00, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c10, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c20, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c30, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c40, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c50, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c60, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c70, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c80, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c90, 0x00000000); + nv_mthd(dev, 0x9097, 0x1ca0, 0x00000000); + nv_mthd(dev, 0x9097, 0x1cb0, 0x00000000); + nv_mthd(dev, 0x9097, 0x1cc0, 0x00000000); + nv_mthd(dev, 0x9097, 0x1cd0, 0x00000000); + nv_mthd(dev, 0x9097, 0x1ce0, 0x00000000); + nv_mthd(dev, 0x9097, 0x1cf0, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c04, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c14, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c24, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c34, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c44, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c54, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c64, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c74, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c84, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c94, 0x00000000); + nv_mthd(dev, 0x9097, 0x1ca4, 0x00000000); + nv_mthd(dev, 0x9097, 0x1cb4, 0x00000000); + nv_mthd(dev, 0x9097, 0x1cc4, 0x00000000); + nv_mthd(dev, 0x9097, 0x1cd4, 0x00000000); + nv_mthd(dev, 0x9097, 0x1ce4, 0x00000000); + nv_mthd(dev, 0x9097, 0x1cf4, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c08, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c18, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c28, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c38, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c48, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c58, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c68, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c78, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c88, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c98, 0x00000000); + nv_mthd(dev, 0x9097, 0x1ca8, 0x00000000); + nv_mthd(dev, 0x9097, 0x1cb8, 0x00000000); + nv_mthd(dev, 0x9097, 0x1cc8, 0x00000000); + nv_mthd(dev, 0x9097, 0x1cd8, 0x00000000); + nv_mthd(dev, 0x9097, 0x1ce8, 0x00000000); + nv_mthd(dev, 0x9097, 0x1cf8, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c0c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c1c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c2c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c3c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c4c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c5c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c6c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c7c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c8c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1c9c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1cac, 0x00000000); + nv_mthd(dev, 0x9097, 0x1cbc, 0x00000000); + nv_mthd(dev, 0x9097, 0x1ccc, 0x00000000); + nv_mthd(dev, 0x9097, 0x1cdc, 0x00000000); + nv_mthd(dev, 0x9097, 0x1cec, 0x00000000); + nv_mthd(dev, 0x9097, 0x1cfc, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d00, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d10, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d20, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d30, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d40, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d50, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d60, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d70, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d80, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d90, 0x00000000); + nv_mthd(dev, 0x9097, 0x1da0, 0x00000000); + nv_mthd(dev, 0x9097, 0x1db0, 0x00000000); + nv_mthd(dev, 0x9097, 0x1dc0, 0x00000000); + nv_mthd(dev, 0x9097, 0x1dd0, 0x00000000); + nv_mthd(dev, 0x9097, 0x1de0, 0x00000000); + nv_mthd(dev, 0x9097, 0x1df0, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d04, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d14, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d24, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d34, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d44, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d54, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d64, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d74, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d84, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d94, 0x00000000); + nv_mthd(dev, 0x9097, 0x1da4, 0x00000000); + nv_mthd(dev, 0x9097, 0x1db4, 0x00000000); + nv_mthd(dev, 0x9097, 0x1dc4, 0x00000000); + nv_mthd(dev, 0x9097, 0x1dd4, 0x00000000); + nv_mthd(dev, 0x9097, 0x1de4, 0x00000000); + nv_mthd(dev, 0x9097, 0x1df4, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d08, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d18, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d28, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d38, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d48, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d58, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d68, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d78, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d88, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d98, 0x00000000); + nv_mthd(dev, 0x9097, 0x1da8, 0x00000000); + nv_mthd(dev, 0x9097, 0x1db8, 0x00000000); + nv_mthd(dev, 0x9097, 0x1dc8, 0x00000000); + nv_mthd(dev, 0x9097, 0x1dd8, 0x00000000); + nv_mthd(dev, 0x9097, 0x1de8, 0x00000000); + nv_mthd(dev, 0x9097, 0x1df8, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d0c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d1c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d2c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d3c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d4c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d5c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d6c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d7c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d8c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1d9c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1dac, 0x00000000); + nv_mthd(dev, 0x9097, 0x1dbc, 0x00000000); + nv_mthd(dev, 0x9097, 0x1dcc, 0x00000000); + nv_mthd(dev, 0x9097, 0x1ddc, 0x00000000); + nv_mthd(dev, 0x9097, 0x1dec, 0x00000000); + nv_mthd(dev, 0x9097, 0x1dfc, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f00, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f08, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f10, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f18, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f20, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f28, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f30, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f38, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f40, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f48, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f50, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f58, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f60, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f68, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f70, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f78, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f04, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f0c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f14, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f1c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f24, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f2c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f34, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f3c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f44, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f4c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f54, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f5c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f64, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f6c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f74, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f7c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f80, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f88, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f90, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f98, 0x00000000); + nv_mthd(dev, 0x9097, 0x1fa0, 0x00000000); + nv_mthd(dev, 0x9097, 0x1fa8, 0x00000000); + nv_mthd(dev, 0x9097, 0x1fb0, 0x00000000); + nv_mthd(dev, 0x9097, 0x1fb8, 0x00000000); + nv_mthd(dev, 0x9097, 0x1fc0, 0x00000000); + nv_mthd(dev, 0x9097, 0x1fc8, 0x00000000); + nv_mthd(dev, 0x9097, 0x1fd0, 0x00000000); + nv_mthd(dev, 0x9097, 0x1fd8, 0x00000000); + nv_mthd(dev, 0x9097, 0x1fe0, 0x00000000); + nv_mthd(dev, 0x9097, 0x1fe8, 0x00000000); + nv_mthd(dev, 0x9097, 0x1ff0, 0x00000000); + nv_mthd(dev, 0x9097, 0x1ff8, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f84, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f8c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f94, 0x00000000); + nv_mthd(dev, 0x9097, 0x1f9c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1fa4, 0x00000000); + nv_mthd(dev, 0x9097, 0x1fac, 0x00000000); + nv_mthd(dev, 0x9097, 0x1fb4, 0x00000000); + nv_mthd(dev, 0x9097, 0x1fbc, 0x00000000); + nv_mthd(dev, 0x9097, 0x1fc4, 0x00000000); + nv_mthd(dev, 0x9097, 0x1fcc, 0x00000000); + nv_mthd(dev, 0x9097, 0x1fd4, 0x00000000); + nv_mthd(dev, 0x9097, 0x1fdc, 0x00000000); + nv_mthd(dev, 0x9097, 0x1fe4, 0x00000000); + nv_mthd(dev, 0x9097, 0x1fec, 0x00000000); + nv_mthd(dev, 0x9097, 0x1ff4, 0x00000000); + nv_mthd(dev, 0x9097, 0x1ffc, 0x00000000); + nv_mthd(dev, 0x9097, 0x2200, 0x00000022); + nv_mthd(dev, 0x9097, 0x2210, 0x00000022); + nv_mthd(dev, 0x9097, 0x2220, 0x00000022); + nv_mthd(dev, 0x9097, 0x2230, 0x00000022); + nv_mthd(dev, 0x9097, 0x2240, 0x00000022); + nv_mthd(dev, 0x9097, 0x2000, 0x00000000); + nv_mthd(dev, 0x9097, 0x2040, 0x00000011); + nv_mthd(dev, 0x9097, 0x2080, 0x00000020); + nv_mthd(dev, 0x9097, 0x20c0, 0x00000030); + nv_mthd(dev, 0x9097, 0x2100, 0x00000040); + nv_mthd(dev, 0x9097, 0x2140, 0x00000051); + nv_mthd(dev, 0x9097, 0x200c, 0x00000001); + nv_mthd(dev, 0x9097, 0x204c, 0x00000001); + nv_mthd(dev, 0x9097, 0x208c, 0x00000001); + nv_mthd(dev, 0x9097, 0x20cc, 0x00000001); + nv_mthd(dev, 0x9097, 0x210c, 0x00000001); + nv_mthd(dev, 0x9097, 0x214c, 0x00000001); + nv_mthd(dev, 0x9097, 0x2010, 0x00000000); + nv_mthd(dev, 0x9097, 0x2050, 0x00000000); + nv_mthd(dev, 0x9097, 0x2090, 0x00000001); + nv_mthd(dev, 0x9097, 0x20d0, 0x00000002); + nv_mthd(dev, 0x9097, 0x2110, 0x00000003); + nv_mthd(dev, 0x9097, 0x2150, 0x00000004); + nv_mthd(dev, 0x9097, 0x0380, 0x00000000); + nv_mthd(dev, 0x9097, 0x03a0, 0x00000000); + nv_mthd(dev, 0x9097, 0x03c0, 0x00000000); + nv_mthd(dev, 0x9097, 0x03e0, 0x00000000); + nv_mthd(dev, 0x9097, 0x0384, 0x00000000); + nv_mthd(dev, 0x9097, 0x03a4, 0x00000000); + nv_mthd(dev, 0x9097, 0x03c4, 0x00000000); + nv_mthd(dev, 0x9097, 0x03e4, 0x00000000); + nv_mthd(dev, 0x9097, 0x0388, 0x00000000); + nv_mthd(dev, 0x9097, 0x03a8, 0x00000000); + nv_mthd(dev, 0x9097, 0x03c8, 0x00000000); + nv_mthd(dev, 0x9097, 0x03e8, 0x00000000); + nv_mthd(dev, 0x9097, 0x038c, 0x00000000); + nv_mthd(dev, 0x9097, 0x03ac, 0x00000000); + nv_mthd(dev, 0x9097, 0x03cc, 0x00000000); + nv_mthd(dev, 0x9097, 0x03ec, 0x00000000); + nv_mthd(dev, 0x9097, 0x0700, 0x00000000); + nv_mthd(dev, 0x9097, 0x0710, 0x00000000); + nv_mthd(dev, 0x9097, 0x0720, 0x00000000); + nv_mthd(dev, 0x9097, 0x0730, 0x00000000); + nv_mthd(dev, 0x9097, 0x0704, 0x00000000); + nv_mthd(dev, 0x9097, 0x0714, 0x00000000); + nv_mthd(dev, 0x9097, 0x0724, 0x00000000); + nv_mthd(dev, 0x9097, 0x0734, 0x00000000); + nv_mthd(dev, 0x9097, 0x0708, 0x00000000); + nv_mthd(dev, 0x9097, 0x0718, 0x00000000); + nv_mthd(dev, 0x9097, 0x0728, 0x00000000); + nv_mthd(dev, 0x9097, 0x0738, 0x00000000); + nv_mthd(dev, 0x9097, 0x2800, 0x00000000); + nv_mthd(dev, 0x9097, 0x2804, 0x00000000); + nv_mthd(dev, 0x9097, 0x2808, 0x00000000); + nv_mthd(dev, 0x9097, 0x280c, 0x00000000); + nv_mthd(dev, 0x9097, 0x2810, 0x00000000); + nv_mthd(dev, 0x9097, 0x2814, 0x00000000); + nv_mthd(dev, 0x9097, 0x2818, 0x00000000); + nv_mthd(dev, 0x9097, 0x281c, 0x00000000); + nv_mthd(dev, 0x9097, 0x2820, 0x00000000); + nv_mthd(dev, 0x9097, 0x2824, 0x00000000); + nv_mthd(dev, 0x9097, 0x2828, 0x00000000); + nv_mthd(dev, 0x9097, 0x282c, 0x00000000); + nv_mthd(dev, 0x9097, 0x2830, 0x00000000); + nv_mthd(dev, 0x9097, 0x2834, 0x00000000); + nv_mthd(dev, 0x9097, 0x2838, 0x00000000); + nv_mthd(dev, 0x9097, 0x283c, 0x00000000); + nv_mthd(dev, 0x9097, 0x2840, 0x00000000); + nv_mthd(dev, 0x9097, 0x2844, 0x00000000); + nv_mthd(dev, 0x9097, 0x2848, 0x00000000); + nv_mthd(dev, 0x9097, 0x284c, 0x00000000); + nv_mthd(dev, 0x9097, 0x2850, 0x00000000); + nv_mthd(dev, 0x9097, 0x2854, 0x00000000); + nv_mthd(dev, 0x9097, 0x2858, 0x00000000); + nv_mthd(dev, 0x9097, 0x285c, 0x00000000); + nv_mthd(dev, 0x9097, 0x2860, 0x00000000); + nv_mthd(dev, 0x9097, 0x2864, 0x00000000); + nv_mthd(dev, 0x9097, 0x2868, 0x00000000); + nv_mthd(dev, 0x9097, 0x286c, 0x00000000); + nv_mthd(dev, 0x9097, 0x2870, 0x00000000); + nv_mthd(dev, 0x9097, 0x2874, 0x00000000); + nv_mthd(dev, 0x9097, 0x2878, 0x00000000); + nv_mthd(dev, 0x9097, 0x287c, 0x00000000); + nv_mthd(dev, 0x9097, 0x2880, 0x00000000); + nv_mthd(dev, 0x9097, 0x2884, 0x00000000); + nv_mthd(dev, 0x9097, 0x2888, 0x00000000); + nv_mthd(dev, 0x9097, 0x288c, 0x00000000); + nv_mthd(dev, 0x9097, 0x2890, 0x00000000); + nv_mthd(dev, 0x9097, 0x2894, 0x00000000); + nv_mthd(dev, 0x9097, 0x2898, 0x00000000); + nv_mthd(dev, 0x9097, 0x289c, 0x00000000); + nv_mthd(dev, 0x9097, 0x28a0, 0x00000000); + nv_mthd(dev, 0x9097, 0x28a4, 0x00000000); + nv_mthd(dev, 0x9097, 0x28a8, 0x00000000); + nv_mthd(dev, 0x9097, 0x28ac, 0x00000000); + nv_mthd(dev, 0x9097, 0x28b0, 0x00000000); + nv_mthd(dev, 0x9097, 0x28b4, 0x00000000); + nv_mthd(dev, 0x9097, 0x28b8, 0x00000000); + nv_mthd(dev, 0x9097, 0x28bc, 0x00000000); + nv_mthd(dev, 0x9097, 0x28c0, 0x00000000); + nv_mthd(dev, 0x9097, 0x28c4, 0x00000000); + nv_mthd(dev, 0x9097, 0x28c8, 0x00000000); + nv_mthd(dev, 0x9097, 0x28cc, 0x00000000); + nv_mthd(dev, 0x9097, 0x28d0, 0x00000000); + nv_mthd(dev, 0x9097, 0x28d4, 0x00000000); + nv_mthd(dev, 0x9097, 0x28d8, 0x00000000); + nv_mthd(dev, 0x9097, 0x28dc, 0x00000000); + nv_mthd(dev, 0x9097, 0x28e0, 0x00000000); + nv_mthd(dev, 0x9097, 0x28e4, 0x00000000); + nv_mthd(dev, 0x9097, 0x28e8, 0x00000000); + nv_mthd(dev, 0x9097, 0x28ec, 0x00000000); + nv_mthd(dev, 0x9097, 0x28f0, 0x00000000); + nv_mthd(dev, 0x9097, 0x28f4, 0x00000000); + nv_mthd(dev, 0x9097, 0x28f8, 0x00000000); + nv_mthd(dev, 0x9097, 0x28fc, 0x00000000); + nv_mthd(dev, 0x9097, 0x2900, 0x00000000); + nv_mthd(dev, 0x9097, 0x2904, 0x00000000); + nv_mthd(dev, 0x9097, 0x2908, 0x00000000); + nv_mthd(dev, 0x9097, 0x290c, 0x00000000); + nv_mthd(dev, 0x9097, 0x2910, 0x00000000); + nv_mthd(dev, 0x9097, 0x2914, 0x00000000); + nv_mthd(dev, 0x9097, 0x2918, 0x00000000); + nv_mthd(dev, 0x9097, 0x291c, 0x00000000); + nv_mthd(dev, 0x9097, 0x2920, 0x00000000); + nv_mthd(dev, 0x9097, 0x2924, 0x00000000); + nv_mthd(dev, 0x9097, 0x2928, 0x00000000); + nv_mthd(dev, 0x9097, 0x292c, 0x00000000); + nv_mthd(dev, 0x9097, 0x2930, 0x00000000); + nv_mthd(dev, 0x9097, 0x2934, 0x00000000); + nv_mthd(dev, 0x9097, 0x2938, 0x00000000); + nv_mthd(dev, 0x9097, 0x293c, 0x00000000); + nv_mthd(dev, 0x9097, 0x2940, 0x00000000); + nv_mthd(dev, 0x9097, 0x2944, 0x00000000); + nv_mthd(dev, 0x9097, 0x2948, 0x00000000); + nv_mthd(dev, 0x9097, 0x294c, 0x00000000); + nv_mthd(dev, 0x9097, 0x2950, 0x00000000); + nv_mthd(dev, 0x9097, 0x2954, 0x00000000); + nv_mthd(dev, 0x9097, 0x2958, 0x00000000); + nv_mthd(dev, 0x9097, 0x295c, 0x00000000); + nv_mthd(dev, 0x9097, 0x2960, 0x00000000); + nv_mthd(dev, 0x9097, 0x2964, 0x00000000); + nv_mthd(dev, 0x9097, 0x2968, 0x00000000); + nv_mthd(dev, 0x9097, 0x296c, 0x00000000); + nv_mthd(dev, 0x9097, 0x2970, 0x00000000); + nv_mthd(dev, 0x9097, 0x2974, 0x00000000); + nv_mthd(dev, 0x9097, 0x2978, 0x00000000); + nv_mthd(dev, 0x9097, 0x297c, 0x00000000); + nv_mthd(dev, 0x9097, 0x2980, 0x00000000); + nv_mthd(dev, 0x9097, 0x2984, 0x00000000); + nv_mthd(dev, 0x9097, 0x2988, 0x00000000); + nv_mthd(dev, 0x9097, 0x298c, 0x00000000); + nv_mthd(dev, 0x9097, 0x2990, 0x00000000); + nv_mthd(dev, 0x9097, 0x2994, 0x00000000); + nv_mthd(dev, 0x9097, 0x2998, 0x00000000); + nv_mthd(dev, 0x9097, 0x299c, 0x00000000); + nv_mthd(dev, 0x9097, 0x29a0, 0x00000000); + nv_mthd(dev, 0x9097, 0x29a4, 0x00000000); + nv_mthd(dev, 0x9097, 0x29a8, 0x00000000); + nv_mthd(dev, 0x9097, 0x29ac, 0x00000000); + nv_mthd(dev, 0x9097, 0x29b0, 0x00000000); + nv_mthd(dev, 0x9097, 0x29b4, 0x00000000); + nv_mthd(dev, 0x9097, 0x29b8, 0x00000000); + nv_mthd(dev, 0x9097, 0x29bc, 0x00000000); + nv_mthd(dev, 0x9097, 0x29c0, 0x00000000); + nv_mthd(dev, 0x9097, 0x29c4, 0x00000000); + nv_mthd(dev, 0x9097, 0x29c8, 0x00000000); + nv_mthd(dev, 0x9097, 0x29cc, 0x00000000); + nv_mthd(dev, 0x9097, 0x29d0, 0x00000000); + nv_mthd(dev, 0x9097, 0x29d4, 0x00000000); + nv_mthd(dev, 0x9097, 0x29d8, 0x00000000); + nv_mthd(dev, 0x9097, 0x29dc, 0x00000000); + nv_mthd(dev, 0x9097, 0x29e0, 0x00000000); + nv_mthd(dev, 0x9097, 0x29e4, 0x00000000); + nv_mthd(dev, 0x9097, 0x29e8, 0x00000000); + nv_mthd(dev, 0x9097, 0x29ec, 0x00000000); + nv_mthd(dev, 0x9097, 0x29f0, 0x00000000); + nv_mthd(dev, 0x9097, 0x29f4, 0x00000000); + nv_mthd(dev, 0x9097, 0x29f8, 0x00000000); + nv_mthd(dev, 0x9097, 0x29fc, 0x00000000); + nv_mthd(dev, 0x9097, 0x0a00, 0x00000000); + nv_mthd(dev, 0x9097, 0x0a20, 0x00000000); + nv_mthd(dev, 0x9097, 0x0a40, 0x00000000); + nv_mthd(dev, 0x9097, 0x0a60, 0x00000000); + nv_mthd(dev, 0x9097, 0x0a80, 0x00000000); + nv_mthd(dev, 0x9097, 0x0aa0, 0x00000000); + nv_mthd(dev, 0x9097, 0x0ac0, 0x00000000); + nv_mthd(dev, 0x9097, 0x0ae0, 0x00000000); + nv_mthd(dev, 0x9097, 0x0b00, 0x00000000); + nv_mthd(dev, 0x9097, 0x0b20, 0x00000000); + nv_mthd(dev, 0x9097, 0x0b40, 0x00000000); + nv_mthd(dev, 0x9097, 0x0b60, 0x00000000); + nv_mthd(dev, 0x9097, 0x0b80, 0x00000000); + nv_mthd(dev, 0x9097, 0x0ba0, 0x00000000); + nv_mthd(dev, 0x9097, 0x0bc0, 0x00000000); + nv_mthd(dev, 0x9097, 0x0be0, 0x00000000); + nv_mthd(dev, 0x9097, 0x0a04, 0x00000000); + nv_mthd(dev, 0x9097, 0x0a24, 0x00000000); + nv_mthd(dev, 0x9097, 0x0a44, 0x00000000); + nv_mthd(dev, 0x9097, 0x0a64, 0x00000000); + nv_mthd(dev, 0x9097, 0x0a84, 0x00000000); + nv_mthd(dev, 0x9097, 0x0aa4, 0x00000000); + nv_mthd(dev, 0x9097, 0x0ac4, 0x00000000); + nv_mthd(dev, 0x9097, 0x0ae4, 0x00000000); + nv_mthd(dev, 0x9097, 0x0b04, 0x00000000); + nv_mthd(dev, 0x9097, 0x0b24, 0x00000000); + nv_mthd(dev, 0x9097, 0x0b44, 0x00000000); + nv_mthd(dev, 0x9097, 0x0b64, 0x00000000); + nv_mthd(dev, 0x9097, 0x0b84, 0x00000000); + nv_mthd(dev, 0x9097, 0x0ba4, 0x00000000); + nv_mthd(dev, 0x9097, 0x0bc4, 0x00000000); + nv_mthd(dev, 0x9097, 0x0be4, 0x00000000); + nv_mthd(dev, 0x9097, 0x0a08, 0x00000000); + nv_mthd(dev, 0x9097, 0x0a28, 0x00000000); + nv_mthd(dev, 0x9097, 0x0a48, 0x00000000); + nv_mthd(dev, 0x9097, 0x0a68, 0x00000000); + nv_mthd(dev, 0x9097, 0x0a88, 0x00000000); + nv_mthd(dev, 0x9097, 0x0aa8, 0x00000000); + nv_mthd(dev, 0x9097, 0x0ac8, 0x00000000); + nv_mthd(dev, 0x9097, 0x0ae8, 0x00000000); + nv_mthd(dev, 0x9097, 0x0b08, 0x00000000); + nv_mthd(dev, 0x9097, 0x0b28, 0x00000000); + nv_mthd(dev, 0x9097, 0x0b48, 0x00000000); + nv_mthd(dev, 0x9097, 0x0b68, 0x00000000); + nv_mthd(dev, 0x9097, 0x0b88, 0x00000000); + nv_mthd(dev, 0x9097, 0x0ba8, 0x00000000); + nv_mthd(dev, 0x9097, 0x0bc8, 0x00000000); + nv_mthd(dev, 0x9097, 0x0be8, 0x00000000); + nv_mthd(dev, 0x9097, 0x0a0c, 0x00000000); + nv_mthd(dev, 0x9097, 0x0a2c, 0x00000000); + nv_mthd(dev, 0x9097, 0x0a4c, 0x00000000); + nv_mthd(dev, 0x9097, 0x0a6c, 0x00000000); + nv_mthd(dev, 0x9097, 0x0a8c, 0x00000000); + nv_mthd(dev, 0x9097, 0x0aac, 0x00000000); + nv_mthd(dev, 0x9097, 0x0acc, 0x00000000); + nv_mthd(dev, 0x9097, 0x0aec, 0x00000000); + nv_mthd(dev, 0x9097, 0x0b0c, 0x00000000); + nv_mthd(dev, 0x9097, 0x0b2c, 0x00000000); + nv_mthd(dev, 0x9097, 0x0b4c, 0x00000000); + nv_mthd(dev, 0x9097, 0x0b6c, 0x00000000); + nv_mthd(dev, 0x9097, 0x0b8c, 0x00000000); + nv_mthd(dev, 0x9097, 0x0bac, 0x00000000); + nv_mthd(dev, 0x9097, 0x0bcc, 0x00000000); + nv_mthd(dev, 0x9097, 0x0bec, 0x00000000); + nv_mthd(dev, 0x9097, 0x0a10, 0x00000000); + nv_mthd(dev, 0x9097, 0x0a30, 0x00000000); + nv_mthd(dev, 0x9097, 0x0a50, 0x00000000); + nv_mthd(dev, 0x9097, 0x0a70, 0x00000000); + nv_mthd(dev, 0x9097, 0x0a90, 0x00000000); + nv_mthd(dev, 0x9097, 0x0ab0, 0x00000000); + nv_mthd(dev, 0x9097, 0x0ad0, 0x00000000); + nv_mthd(dev, 0x9097, 0x0af0, 0x00000000); + nv_mthd(dev, 0x9097, 0x0b10, 0x00000000); + nv_mthd(dev, 0x9097, 0x0b30, 0x00000000); + nv_mthd(dev, 0x9097, 0x0b50, 0x00000000); + nv_mthd(dev, 0x9097, 0x0b70, 0x00000000); + nv_mthd(dev, 0x9097, 0x0b90, 0x00000000); + nv_mthd(dev, 0x9097, 0x0bb0, 0x00000000); + nv_mthd(dev, 0x9097, 0x0bd0, 0x00000000); + nv_mthd(dev, 0x9097, 0x0bf0, 0x00000000); + nv_mthd(dev, 0x9097, 0x0a14, 0x00000000); + nv_mthd(dev, 0x9097, 0x0a34, 0x00000000); + nv_mthd(dev, 0x9097, 0x0a54, 0x00000000); + nv_mthd(dev, 0x9097, 0x0a74, 0x00000000); + nv_mthd(dev, 0x9097, 0x0a94, 0x00000000); + nv_mthd(dev, 0x9097, 0x0ab4, 0x00000000); + nv_mthd(dev, 0x9097, 0x0ad4, 0x00000000); + nv_mthd(dev, 0x9097, 0x0af4, 0x00000000); + nv_mthd(dev, 0x9097, 0x0b14, 0x00000000); + nv_mthd(dev, 0x9097, 0x0b34, 0x00000000); + nv_mthd(dev, 0x9097, 0x0b54, 0x00000000); + nv_mthd(dev, 0x9097, 0x0b74, 0x00000000); + nv_mthd(dev, 0x9097, 0x0b94, 0x00000000); + nv_mthd(dev, 0x9097, 0x0bb4, 0x00000000); + nv_mthd(dev, 0x9097, 0x0bd4, 0x00000000); + nv_mthd(dev, 0x9097, 0x0bf4, 0x00000000); + nv_mthd(dev, 0x9097, 0x0c00, 0x00000000); + nv_mthd(dev, 0x9097, 0x0c10, 0x00000000); + nv_mthd(dev, 0x9097, 0x0c20, 0x00000000); + nv_mthd(dev, 0x9097, 0x0c30, 0x00000000); + nv_mthd(dev, 0x9097, 0x0c40, 0x00000000); + nv_mthd(dev, 0x9097, 0x0c50, 0x00000000); + nv_mthd(dev, 0x9097, 0x0c60, 0x00000000); + nv_mthd(dev, 0x9097, 0x0c70, 0x00000000); + nv_mthd(dev, 0x9097, 0x0c80, 0x00000000); + nv_mthd(dev, 0x9097, 0x0c90, 0x00000000); + nv_mthd(dev, 0x9097, 0x0ca0, 0x00000000); + nv_mthd(dev, 0x9097, 0x0cb0, 0x00000000); + nv_mthd(dev, 0x9097, 0x0cc0, 0x00000000); + nv_mthd(dev, 0x9097, 0x0cd0, 0x00000000); + nv_mthd(dev, 0x9097, 0x0ce0, 0x00000000); + nv_mthd(dev, 0x9097, 0x0cf0, 0x00000000); + nv_mthd(dev, 0x9097, 0x0c04, 0x00000000); + nv_mthd(dev, 0x9097, 0x0c14, 0x00000000); + nv_mthd(dev, 0x9097, 0x0c24, 0x00000000); + nv_mthd(dev, 0x9097, 0x0c34, 0x00000000); + nv_mthd(dev, 0x9097, 0x0c44, 0x00000000); + nv_mthd(dev, 0x9097, 0x0c54, 0x00000000); + nv_mthd(dev, 0x9097, 0x0c64, 0x00000000); + nv_mthd(dev, 0x9097, 0x0c74, 0x00000000); + nv_mthd(dev, 0x9097, 0x0c84, 0x00000000); + nv_mthd(dev, 0x9097, 0x0c94, 0x00000000); + nv_mthd(dev, 0x9097, 0x0ca4, 0x00000000); + nv_mthd(dev, 0x9097, 0x0cb4, 0x00000000); + nv_mthd(dev, 0x9097, 0x0cc4, 0x00000000); + nv_mthd(dev, 0x9097, 0x0cd4, 0x00000000); + nv_mthd(dev, 0x9097, 0x0ce4, 0x00000000); + nv_mthd(dev, 0x9097, 0x0cf4, 0x00000000); + nv_mthd(dev, 0x9097, 0x0c08, 0x00000000); + nv_mthd(dev, 0x9097, 0x0c18, 0x00000000); + nv_mthd(dev, 0x9097, 0x0c28, 0x00000000); + nv_mthd(dev, 0x9097, 0x0c38, 0x00000000); + nv_mthd(dev, 0x9097, 0x0c48, 0x00000000); + nv_mthd(dev, 0x9097, 0x0c58, 0x00000000); + nv_mthd(dev, 0x9097, 0x0c68, 0x00000000); + nv_mthd(dev, 0x9097, 0x0c78, 0x00000000); + nv_mthd(dev, 0x9097, 0x0c88, 0x00000000); + nv_mthd(dev, 0x9097, 0x0c98, 0x00000000); + nv_mthd(dev, 0x9097, 0x0ca8, 0x00000000); + nv_mthd(dev, 0x9097, 0x0cb8, 0x00000000); + nv_mthd(dev, 0x9097, 0x0cc8, 0x00000000); + nv_mthd(dev, 0x9097, 0x0cd8, 0x00000000); + nv_mthd(dev, 0x9097, 0x0ce8, 0x00000000); + nv_mthd(dev, 0x9097, 0x0cf8, 0x00000000); + nv_mthd(dev, 0x9097, 0x0c0c, 0x3f800000); + nv_mthd(dev, 0x9097, 0x0c1c, 0x3f800000); + nv_mthd(dev, 0x9097, 0x0c2c, 0x3f800000); + nv_mthd(dev, 0x9097, 0x0c3c, 0x3f800000); + nv_mthd(dev, 0x9097, 0x0c4c, 0x3f800000); + nv_mthd(dev, 0x9097, 0x0c5c, 0x3f800000); + nv_mthd(dev, 0x9097, 0x0c6c, 0x3f800000); + nv_mthd(dev, 0x9097, 0x0c7c, 0x3f800000); + nv_mthd(dev, 0x9097, 0x0c8c, 0x3f800000); + nv_mthd(dev, 0x9097, 0x0c9c, 0x3f800000); + nv_mthd(dev, 0x9097, 0x0cac, 0x3f800000); + nv_mthd(dev, 0x9097, 0x0cbc, 0x3f800000); + nv_mthd(dev, 0x9097, 0x0ccc, 0x3f800000); + nv_mthd(dev, 0x9097, 0x0cdc, 0x3f800000); + nv_mthd(dev, 0x9097, 0x0cec, 0x3f800000); + nv_mthd(dev, 0x9097, 0x0cfc, 0x3f800000); + nv_mthd(dev, 0x9097, 0x0d00, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0d08, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0d10, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0d18, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0d20, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0d28, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0d30, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0d38, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0d04, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0d0c, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0d14, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0d1c, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0d24, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0d2c, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0d34, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0d3c, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0e00, 0x00000000); + nv_mthd(dev, 0x9097, 0x0e10, 0x00000000); + nv_mthd(dev, 0x9097, 0x0e20, 0x00000000); + nv_mthd(dev, 0x9097, 0x0e30, 0x00000000); + nv_mthd(dev, 0x9097, 0x0e40, 0x00000000); + nv_mthd(dev, 0x9097, 0x0e50, 0x00000000); + nv_mthd(dev, 0x9097, 0x0e60, 0x00000000); + nv_mthd(dev, 0x9097, 0x0e70, 0x00000000); + nv_mthd(dev, 0x9097, 0x0e80, 0x00000000); + nv_mthd(dev, 0x9097, 0x0e90, 0x00000000); + nv_mthd(dev, 0x9097, 0x0ea0, 0x00000000); + nv_mthd(dev, 0x9097, 0x0eb0, 0x00000000); + nv_mthd(dev, 0x9097, 0x0ec0, 0x00000000); + nv_mthd(dev, 0x9097, 0x0ed0, 0x00000000); + nv_mthd(dev, 0x9097, 0x0ee0, 0x00000000); + nv_mthd(dev, 0x9097, 0x0ef0, 0x00000000); + nv_mthd(dev, 0x9097, 0x0e04, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0e14, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0e24, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0e34, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0e44, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0e54, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0e64, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0e74, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0e84, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0e94, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0ea4, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0eb4, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0ec4, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0ed4, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0ee4, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0ef4, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0e08, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0e18, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0e28, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0e38, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0e48, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0e58, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0e68, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0e78, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0e88, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0e98, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0ea8, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0eb8, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0ec8, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0ed8, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0ee8, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0ef8, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0d40, 0x00000000); + nv_mthd(dev, 0x9097, 0x0d48, 0x00000000); + nv_mthd(dev, 0x9097, 0x0d50, 0x00000000); + nv_mthd(dev, 0x9097, 0x0d58, 0x00000000); + nv_mthd(dev, 0x9097, 0x0d44, 0x00000000); + nv_mthd(dev, 0x9097, 0x0d4c, 0x00000000); + nv_mthd(dev, 0x9097, 0x0d54, 0x00000000); + nv_mthd(dev, 0x9097, 0x0d5c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1e00, 0x00000001); + nv_mthd(dev, 0x9097, 0x1e20, 0x00000001); + nv_mthd(dev, 0x9097, 0x1e40, 0x00000001); + nv_mthd(dev, 0x9097, 0x1e60, 0x00000001); + nv_mthd(dev, 0x9097, 0x1e80, 0x00000001); + nv_mthd(dev, 0x9097, 0x1ea0, 0x00000001); + nv_mthd(dev, 0x9097, 0x1ec0, 0x00000001); + nv_mthd(dev, 0x9097, 0x1ee0, 0x00000001); + nv_mthd(dev, 0x9097, 0x1e04, 0x00000001); + nv_mthd(dev, 0x9097, 0x1e24, 0x00000001); + nv_mthd(dev, 0x9097, 0x1e44, 0x00000001); + nv_mthd(dev, 0x9097, 0x1e64, 0x00000001); + nv_mthd(dev, 0x9097, 0x1e84, 0x00000001); + nv_mthd(dev, 0x9097, 0x1ea4, 0x00000001); + nv_mthd(dev, 0x9097, 0x1ec4, 0x00000001); + nv_mthd(dev, 0x9097, 0x1ee4, 0x00000001); + nv_mthd(dev, 0x9097, 0x1e08, 0x00000002); + nv_mthd(dev, 0x9097, 0x1e28, 0x00000002); + nv_mthd(dev, 0x9097, 0x1e48, 0x00000002); + nv_mthd(dev, 0x9097, 0x1e68, 0x00000002); + nv_mthd(dev, 0x9097, 0x1e88, 0x00000002); + nv_mthd(dev, 0x9097, 0x1ea8, 0x00000002); + nv_mthd(dev, 0x9097, 0x1ec8, 0x00000002); + nv_mthd(dev, 0x9097, 0x1ee8, 0x00000002); + nv_mthd(dev, 0x9097, 0x1e0c, 0x00000001); + nv_mthd(dev, 0x9097, 0x1e2c, 0x00000001); + nv_mthd(dev, 0x9097, 0x1e4c, 0x00000001); + nv_mthd(dev, 0x9097, 0x1e6c, 0x00000001); + nv_mthd(dev, 0x9097, 0x1e8c, 0x00000001); + nv_mthd(dev, 0x9097, 0x1eac, 0x00000001); + nv_mthd(dev, 0x9097, 0x1ecc, 0x00000001); + nv_mthd(dev, 0x9097, 0x1eec, 0x00000001); + nv_mthd(dev, 0x9097, 0x1e10, 0x00000001); + nv_mthd(dev, 0x9097, 0x1e30, 0x00000001); + nv_mthd(dev, 0x9097, 0x1e50, 0x00000001); + nv_mthd(dev, 0x9097, 0x1e70, 0x00000001); + nv_mthd(dev, 0x9097, 0x1e90, 0x00000001); + nv_mthd(dev, 0x9097, 0x1eb0, 0x00000001); + nv_mthd(dev, 0x9097, 0x1ed0, 0x00000001); + nv_mthd(dev, 0x9097, 0x1ef0, 0x00000001); + nv_mthd(dev, 0x9097, 0x1e14, 0x00000002); + nv_mthd(dev, 0x9097, 0x1e34, 0x00000002); + nv_mthd(dev, 0x9097, 0x1e54, 0x00000002); + nv_mthd(dev, 0x9097, 0x1e74, 0x00000002); + nv_mthd(dev, 0x9097, 0x1e94, 0x00000002); + nv_mthd(dev, 0x9097, 0x1eb4, 0x00000002); + nv_mthd(dev, 0x9097, 0x1ed4, 0x00000002); + nv_mthd(dev, 0x9097, 0x1ef4, 0x00000002); + nv_mthd(dev, 0x9097, 0x1e18, 0x00000001); + nv_mthd(dev, 0x9097, 0x1e38, 0x00000001); + nv_mthd(dev, 0x9097, 0x1e58, 0x00000001); + nv_mthd(dev, 0x9097, 0x1e78, 0x00000001); + nv_mthd(dev, 0x9097, 0x1e98, 0x00000001); + nv_mthd(dev, 0x9097, 0x1eb8, 0x00000001); + nv_mthd(dev, 0x9097, 0x1ed8, 0x00000001); + nv_mthd(dev, 0x9097, 0x1ef8, 0x00000001); + nv_mthd(dev, 0x9097, 0x3400, 0x00000000); + nv_mthd(dev, 0x9097, 0x3404, 0x00000000); + nv_mthd(dev, 0x9097, 0x3408, 0x00000000); + nv_mthd(dev, 0x9097, 0x340c, 0x00000000); + nv_mthd(dev, 0x9097, 0x3410, 0x00000000); + nv_mthd(dev, 0x9097, 0x3414, 0x00000000); + nv_mthd(dev, 0x9097, 0x3418, 0x00000000); + nv_mthd(dev, 0x9097, 0x341c, 0x00000000); + nv_mthd(dev, 0x9097, 0x3420, 0x00000000); + nv_mthd(dev, 0x9097, 0x3424, 0x00000000); + nv_mthd(dev, 0x9097, 0x3428, 0x00000000); + nv_mthd(dev, 0x9097, 0x342c, 0x00000000); + nv_mthd(dev, 0x9097, 0x3430, 0x00000000); + nv_mthd(dev, 0x9097, 0x3434, 0x00000000); + nv_mthd(dev, 0x9097, 0x3438, 0x00000000); + nv_mthd(dev, 0x9097, 0x343c, 0x00000000); + nv_mthd(dev, 0x9097, 0x3440, 0x00000000); + nv_mthd(dev, 0x9097, 0x3444, 0x00000000); + nv_mthd(dev, 0x9097, 0x3448, 0x00000000); + nv_mthd(dev, 0x9097, 0x344c, 0x00000000); + nv_mthd(dev, 0x9097, 0x3450, 0x00000000); + nv_mthd(dev, 0x9097, 0x3454, 0x00000000); + nv_mthd(dev, 0x9097, 0x3458, 0x00000000); + nv_mthd(dev, 0x9097, 0x345c, 0x00000000); + nv_mthd(dev, 0x9097, 0x3460, 0x00000000); + nv_mthd(dev, 0x9097, 0x3464, 0x00000000); + nv_mthd(dev, 0x9097, 0x3468, 0x00000000); + nv_mthd(dev, 0x9097, 0x346c, 0x00000000); + nv_mthd(dev, 0x9097, 0x3470, 0x00000000); + nv_mthd(dev, 0x9097, 0x3474, 0x00000000); + nv_mthd(dev, 0x9097, 0x3478, 0x00000000); + nv_mthd(dev, 0x9097, 0x347c, 0x00000000); + nv_mthd(dev, 0x9097, 0x3480, 0x00000000); + nv_mthd(dev, 0x9097, 0x3484, 0x00000000); + nv_mthd(dev, 0x9097, 0x3488, 0x00000000); + nv_mthd(dev, 0x9097, 0x348c, 0x00000000); + nv_mthd(dev, 0x9097, 0x3490, 0x00000000); + nv_mthd(dev, 0x9097, 0x3494, 0x00000000); + nv_mthd(dev, 0x9097, 0x3498, 0x00000000); + nv_mthd(dev, 0x9097, 0x349c, 0x00000000); + nv_mthd(dev, 0x9097, 0x34a0, 0x00000000); + nv_mthd(dev, 0x9097, 0x34a4, 0x00000000); + nv_mthd(dev, 0x9097, 0x34a8, 0x00000000); + nv_mthd(dev, 0x9097, 0x34ac, 0x00000000); + nv_mthd(dev, 0x9097, 0x34b0, 0x00000000); + nv_mthd(dev, 0x9097, 0x34b4, 0x00000000); + nv_mthd(dev, 0x9097, 0x34b8, 0x00000000); + nv_mthd(dev, 0x9097, 0x34bc, 0x00000000); + nv_mthd(dev, 0x9097, 0x34c0, 0x00000000); + nv_mthd(dev, 0x9097, 0x34c4, 0x00000000); + nv_mthd(dev, 0x9097, 0x34c8, 0x00000000); + nv_mthd(dev, 0x9097, 0x34cc, 0x00000000); + nv_mthd(dev, 0x9097, 0x34d0, 0x00000000); + nv_mthd(dev, 0x9097, 0x34d4, 0x00000000); + nv_mthd(dev, 0x9097, 0x34d8, 0x00000000); + nv_mthd(dev, 0x9097, 0x34dc, 0x00000000); + nv_mthd(dev, 0x9097, 0x34e0, 0x00000000); + nv_mthd(dev, 0x9097, 0x34e4, 0x00000000); + nv_mthd(dev, 0x9097, 0x34e8, 0x00000000); + nv_mthd(dev, 0x9097, 0x34ec, 0x00000000); + nv_mthd(dev, 0x9097, 0x34f0, 0x00000000); + nv_mthd(dev, 0x9097, 0x34f4, 0x00000000); + nv_mthd(dev, 0x9097, 0x34f8, 0x00000000); + nv_mthd(dev, 0x9097, 0x34fc, 0x00000000); + nv_mthd(dev, 0x9097, 0x3500, 0x00000000); + nv_mthd(dev, 0x9097, 0x3504, 0x00000000); + nv_mthd(dev, 0x9097, 0x3508, 0x00000000); + nv_mthd(dev, 0x9097, 0x350c, 0x00000000); + nv_mthd(dev, 0x9097, 0x3510, 0x00000000); + nv_mthd(dev, 0x9097, 0x3514, 0x00000000); + nv_mthd(dev, 0x9097, 0x3518, 0x00000000); + nv_mthd(dev, 0x9097, 0x351c, 0x00000000); + nv_mthd(dev, 0x9097, 0x3520, 0x00000000); + nv_mthd(dev, 0x9097, 0x3524, 0x00000000); + nv_mthd(dev, 0x9097, 0x3528, 0x00000000); + nv_mthd(dev, 0x9097, 0x352c, 0x00000000); + nv_mthd(dev, 0x9097, 0x3530, 0x00000000); + nv_mthd(dev, 0x9097, 0x3534, 0x00000000); + nv_mthd(dev, 0x9097, 0x3538, 0x00000000); + nv_mthd(dev, 0x9097, 0x353c, 0x00000000); + nv_mthd(dev, 0x9097, 0x3540, 0x00000000); + nv_mthd(dev, 0x9097, 0x3544, 0x00000000); + nv_mthd(dev, 0x9097, 0x3548, 0x00000000); + nv_mthd(dev, 0x9097, 0x354c, 0x00000000); + nv_mthd(dev, 0x9097, 0x3550, 0x00000000); + nv_mthd(dev, 0x9097, 0x3554, 0x00000000); + nv_mthd(dev, 0x9097, 0x3558, 0x00000000); + nv_mthd(dev, 0x9097, 0x355c, 0x00000000); + nv_mthd(dev, 0x9097, 0x3560, 0x00000000); + nv_mthd(dev, 0x9097, 0x3564, 0x00000000); + nv_mthd(dev, 0x9097, 0x3568, 0x00000000); + nv_mthd(dev, 0x9097, 0x356c, 0x00000000); + nv_mthd(dev, 0x9097, 0x3570, 0x00000000); + nv_mthd(dev, 0x9097, 0x3574, 0x00000000); + nv_mthd(dev, 0x9097, 0x3578, 0x00000000); + nv_mthd(dev, 0x9097, 0x357c, 0x00000000); + nv_mthd(dev, 0x9097, 0x3580, 0x00000000); + nv_mthd(dev, 0x9097, 0x3584, 0x00000000); + nv_mthd(dev, 0x9097, 0x3588, 0x00000000); + nv_mthd(dev, 0x9097, 0x358c, 0x00000000); + nv_mthd(dev, 0x9097, 0x3590, 0x00000000); + nv_mthd(dev, 0x9097, 0x3594, 0x00000000); + nv_mthd(dev, 0x9097, 0x3598, 0x00000000); + nv_mthd(dev, 0x9097, 0x359c, 0x00000000); + nv_mthd(dev, 0x9097, 0x35a0, 0x00000000); + nv_mthd(dev, 0x9097, 0x35a4, 0x00000000); + nv_mthd(dev, 0x9097, 0x35a8, 0x00000000); + nv_mthd(dev, 0x9097, 0x35ac, 0x00000000); + nv_mthd(dev, 0x9097, 0x35b0, 0x00000000); + nv_mthd(dev, 0x9097, 0x35b4, 0x00000000); + nv_mthd(dev, 0x9097, 0x35b8, 0x00000000); + nv_mthd(dev, 0x9097, 0x35bc, 0x00000000); + nv_mthd(dev, 0x9097, 0x35c0, 0x00000000); + nv_mthd(dev, 0x9097, 0x35c4, 0x00000000); + nv_mthd(dev, 0x9097, 0x35c8, 0x00000000); + nv_mthd(dev, 0x9097, 0x35cc, 0x00000000); + nv_mthd(dev, 0x9097, 0x35d0, 0x00000000); + nv_mthd(dev, 0x9097, 0x35d4, 0x00000000); + nv_mthd(dev, 0x9097, 0x35d8, 0x00000000); + nv_mthd(dev, 0x9097, 0x35dc, 0x00000000); + nv_mthd(dev, 0x9097, 0x35e0, 0x00000000); + nv_mthd(dev, 0x9097, 0x35e4, 0x00000000); + nv_mthd(dev, 0x9097, 0x35e8, 0x00000000); + nv_mthd(dev, 0x9097, 0x35ec, 0x00000000); + nv_mthd(dev, 0x9097, 0x35f0, 0x00000000); + nv_mthd(dev, 0x9097, 0x35f4, 0x00000000); + nv_mthd(dev, 0x9097, 0x35f8, 0x00000000); + nv_mthd(dev, 0x9097, 0x35fc, 0x00000000); + nv_mthd(dev, 0x9097, 0x030c, 0x00000001); + nv_mthd(dev, 0x9097, 0x1944, 0x00000000); + nv_mthd(dev, 0x9097, 0x1514, 0x00000000); + nv_mthd(dev, 0x9097, 0x0d68, 0x0000ffff); + nv_mthd(dev, 0x9097, 0x121c, 0x0fac6881); + nv_mthd(dev, 0x9097, 0x0fac, 0x00000001); + nv_mthd(dev, 0x9097, 0x1538, 0x00000001); + nv_mthd(dev, 0x9097, 0x0fe0, 0x00000000); + nv_mthd(dev, 0x9097, 0x0fe4, 0x00000000); + nv_mthd(dev, 0x9097, 0x0fe8, 0x00000014); + nv_mthd(dev, 0x9097, 0x0fec, 0x00000040); + nv_mthd(dev, 0x9097, 0x0ff0, 0x00000000); + nv_mthd(dev, 0x9097, 0x179c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1228, 0x00000400); + nv_mthd(dev, 0x9097, 0x122c, 0x00000300); + nv_mthd(dev, 0x9097, 0x1230, 0x00010001); + nv_mthd(dev, 0x9097, 0x07f8, 0x00000000); + nv_mthd(dev, 0x9097, 0x15b4, 0x00000001); + nv_mthd(dev, 0x9097, 0x15cc, 0x00000000); + nv_mthd(dev, 0x9097, 0x1534, 0x00000000); + nv_mthd(dev, 0x9097, 0x0fb0, 0x00000000); + nv_mthd(dev, 0x9097, 0x15d0, 0x00000000); + nv_mthd(dev, 0x9097, 0x153c, 0x00000000); + nv_mthd(dev, 0x9097, 0x16b4, 0x00000003); + nv_mthd(dev, 0x9097, 0x0fbc, 0x0000ffff); + nv_mthd(dev, 0x9097, 0x0fc0, 0x0000ffff); + nv_mthd(dev, 0x9097, 0x0fc4, 0x0000ffff); + nv_mthd(dev, 0x9097, 0x0fc8, 0x0000ffff); + nv_mthd(dev, 0x9097, 0x0df8, 0x00000000); + nv_mthd(dev, 0x9097, 0x0dfc, 0x00000000); + nv_mthd(dev, 0x9097, 0x1948, 0x00000000); + nv_mthd(dev, 0x9097, 0x1970, 0x00000001); + nv_mthd(dev, 0x9097, 0x161c, 0x000009f0); + nv_mthd(dev, 0x9097, 0x0dcc, 0x00000010); + nv_mthd(dev, 0x9097, 0x163c, 0x00000000); + nv_mthd(dev, 0x9097, 0x15e4, 0x00000000); + nv_mthd(dev, 0x9097, 0x1160, 0x25e00040); + nv_mthd(dev, 0x9097, 0x1164, 0x25e00040); + nv_mthd(dev, 0x9097, 0x1168, 0x25e00040); + nv_mthd(dev, 0x9097, 0x116c, 0x25e00040); + nv_mthd(dev, 0x9097, 0x1170, 0x25e00040); + nv_mthd(dev, 0x9097, 0x1174, 0x25e00040); + nv_mthd(dev, 0x9097, 0x1178, 0x25e00040); + nv_mthd(dev, 0x9097, 0x117c, 0x25e00040); + nv_mthd(dev, 0x9097, 0x1180, 0x25e00040); + nv_mthd(dev, 0x9097, 0x1184, 0x25e00040); + nv_mthd(dev, 0x9097, 0x1188, 0x25e00040); + nv_mthd(dev, 0x9097, 0x118c, 0x25e00040); + nv_mthd(dev, 0x9097, 0x1190, 0x25e00040); + nv_mthd(dev, 0x9097, 0x1194, 0x25e00040); + nv_mthd(dev, 0x9097, 0x1198, 0x25e00040); + nv_mthd(dev, 0x9097, 0x119c, 0x25e00040); + nv_mthd(dev, 0x9097, 0x11a0, 0x25e00040); + nv_mthd(dev, 0x9097, 0x11a4, 0x25e00040); + nv_mthd(dev, 0x9097, 0x11a8, 0x25e00040); + nv_mthd(dev, 0x9097, 0x11ac, 0x25e00040); + nv_mthd(dev, 0x9097, 0x11b0, 0x25e00040); + nv_mthd(dev, 0x9097, 0x11b4, 0x25e00040); + nv_mthd(dev, 0x9097, 0x11b8, 0x25e00040); + nv_mthd(dev, 0x9097, 0x11bc, 0x25e00040); + nv_mthd(dev, 0x9097, 0x11c0, 0x25e00040); + nv_mthd(dev, 0x9097, 0x11c4, 0x25e00040); + nv_mthd(dev, 0x9097, 0x11c8, 0x25e00040); + nv_mthd(dev, 0x9097, 0x11cc, 0x25e00040); + nv_mthd(dev, 0x9097, 0x11d0, 0x25e00040); + nv_mthd(dev, 0x9097, 0x11d4, 0x25e00040); + nv_mthd(dev, 0x9097, 0x11d8, 0x25e00040); + nv_mthd(dev, 0x9097, 0x11dc, 0x25e00040); + nv_mthd(dev, 0x9097, 0x1880, 0x00000000); + nv_mthd(dev, 0x9097, 0x1884, 0x00000000); + nv_mthd(dev, 0x9097, 0x1888, 0x00000000); + nv_mthd(dev, 0x9097, 0x188c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1890, 0x00000000); + nv_mthd(dev, 0x9097, 0x1894, 0x00000000); + nv_mthd(dev, 0x9097, 0x1898, 0x00000000); + nv_mthd(dev, 0x9097, 0x189c, 0x00000000); + nv_mthd(dev, 0x9097, 0x18a0, 0x00000000); + nv_mthd(dev, 0x9097, 0x18a4, 0x00000000); + nv_mthd(dev, 0x9097, 0x18a8, 0x00000000); + nv_mthd(dev, 0x9097, 0x18ac, 0x00000000); + nv_mthd(dev, 0x9097, 0x18b0, 0x00000000); + nv_mthd(dev, 0x9097, 0x18b4, 0x00000000); + nv_mthd(dev, 0x9097, 0x18b8, 0x00000000); + nv_mthd(dev, 0x9097, 0x18bc, 0x00000000); + nv_mthd(dev, 0x9097, 0x18c0, 0x00000000); + nv_mthd(dev, 0x9097, 0x18c4, 0x00000000); + nv_mthd(dev, 0x9097, 0x18c8, 0x00000000); + nv_mthd(dev, 0x9097, 0x18cc, 0x00000000); + nv_mthd(dev, 0x9097, 0x18d0, 0x00000000); + nv_mthd(dev, 0x9097, 0x18d4, 0x00000000); + nv_mthd(dev, 0x9097, 0x18d8, 0x00000000); + nv_mthd(dev, 0x9097, 0x18dc, 0x00000000); + nv_mthd(dev, 0x9097, 0x18e0, 0x00000000); + nv_mthd(dev, 0x9097, 0x18e4, 0x00000000); + nv_mthd(dev, 0x9097, 0x18e8, 0x00000000); + nv_mthd(dev, 0x9097, 0x18ec, 0x00000000); + nv_mthd(dev, 0x9097, 0x18f0, 0x00000000); + nv_mthd(dev, 0x9097, 0x18f4, 0x00000000); + nv_mthd(dev, 0x9097, 0x18f8, 0x00000000); + nv_mthd(dev, 0x9097, 0x18fc, 0x00000000); + nv_mthd(dev, 0x9097, 0x0f84, 0x00000000); + nv_mthd(dev, 0x9097, 0x0f88, 0x00000000); + nv_mthd(dev, 0x9097, 0x17c8, 0x00000000); + nv_mthd(dev, 0x9097, 0x17cc, 0x00000000); + nv_mthd(dev, 0x9097, 0x17d0, 0x000000ff); + nv_mthd(dev, 0x9097, 0x17d4, 0xffffffff); + nv_mthd(dev, 0x9097, 0x17d8, 0x00000002); + nv_mthd(dev, 0x9097, 0x17dc, 0x00000000); + nv_mthd(dev, 0x9097, 0x15f4, 0x00000000); + nv_mthd(dev, 0x9097, 0x15f8, 0x00000000); + nv_mthd(dev, 0x9097, 0x1434, 0x00000000); + nv_mthd(dev, 0x9097, 0x1438, 0x00000000); + nv_mthd(dev, 0x9097, 0x0d74, 0x00000000); + nv_mthd(dev, 0x9097, 0x0dec, 0x00000001); + nv_mthd(dev, 0x9097, 0x13a4, 0x00000000); + nv_mthd(dev, 0x9097, 0x1318, 0x00000001); + nv_mthd(dev, 0x9097, 0x1644, 0x00000000); + nv_mthd(dev, 0x9097, 0x0748, 0x00000000); + nv_mthd(dev, 0x9097, 0x0de8, 0x00000000); + nv_mthd(dev, 0x9097, 0x1648, 0x00000000); + nv_mthd(dev, 0x9097, 0x12a4, 0x00000000); + nv_mthd(dev, 0x9097, 0x1120, 0x00000000); + nv_mthd(dev, 0x9097, 0x1124, 0x00000000); + nv_mthd(dev, 0x9097, 0x1128, 0x00000000); + nv_mthd(dev, 0x9097, 0x112c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1118, 0x00000000); + nv_mthd(dev, 0x9097, 0x164c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1658, 0x00000000); + nv_mthd(dev, 0x9097, 0x1910, 0x00000290); + nv_mthd(dev, 0x9097, 0x1518, 0x00000000); + nv_mthd(dev, 0x9097, 0x165c, 0x00000001); + nv_mthd(dev, 0x9097, 0x1520, 0x00000000); + nv_mthd(dev, 0x9097, 0x1604, 0x00000000); + nv_mthd(dev, 0x9097, 0x1570, 0x00000000); + nv_mthd(dev, 0x9097, 0x13b0, 0x3f800000); + nv_mthd(dev, 0x9097, 0x13b4, 0x3f800000); + nv_mthd(dev, 0x9097, 0x020c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1670, 0x30201000); + nv_mthd(dev, 0x9097, 0x1674, 0x70605040); + nv_mthd(dev, 0x9097, 0x1678, 0xb8a89888); + nv_mthd(dev, 0x9097, 0x167c, 0xf8e8d8c8); + nv_mthd(dev, 0x9097, 0x166c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1680, 0x00ffff00); + nv_mthd(dev, 0x9097, 0x12d0, 0x00000003); + nv_mthd(dev, 0x9097, 0x12d4, 0x00000002); + nv_mthd(dev, 0x9097, 0x1684, 0x00000000); + nv_mthd(dev, 0x9097, 0x1688, 0x00000000); + nv_mthd(dev, 0x9097, 0x0dac, 0x00001b02); + nv_mthd(dev, 0x9097, 0x0db0, 0x00001b02); + nv_mthd(dev, 0x9097, 0x0db4, 0x00000000); + nv_mthd(dev, 0x9097, 0x168c, 0x00000000); + nv_mthd(dev, 0x9097, 0x15bc, 0x00000000); + nv_mthd(dev, 0x9097, 0x156c, 0x00000000); + nv_mthd(dev, 0x9097, 0x187c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1110, 0x00000001); + nv_mthd(dev, 0x9097, 0x0dc0, 0x00000000); + nv_mthd(dev, 0x9097, 0x0dc4, 0x00000000); + nv_mthd(dev, 0x9097, 0x0dc8, 0x00000000); + nv_mthd(dev, 0x9097, 0x1234, 0x00000000); + nv_mthd(dev, 0x9097, 0x1690, 0x00000000); + nv_mthd(dev, 0x9097, 0x12ac, 0x00000001); + nv_mthd(dev, 0x9097, 0x02c4, 0x00000000); + nv_mthd(dev, 0x9097, 0x0790, 0x00000000); + nv_mthd(dev, 0x9097, 0x0794, 0x00000000); + nv_mthd(dev, 0x9097, 0x0798, 0x00000000); + nv_mthd(dev, 0x9097, 0x079c, 0x00000000); + nv_mthd(dev, 0x9097, 0x07a0, 0x00000000); + nv_mthd(dev, 0x9097, 0x077c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1000, 0x00000010); + nv_mthd(dev, 0x9097, 0x10fc, 0x00000000); + nv_mthd(dev, 0x9097, 0x1290, 0x00000000); + nv_mthd(dev, 0x9097, 0x0218, 0x00000010); + nv_mthd(dev, 0x9097, 0x12d8, 0x00000000); + nv_mthd(dev, 0x9097, 0x12dc, 0x00000010); + nv_mthd(dev, 0x9097, 0x0d94, 0x00000001); + nv_mthd(dev, 0x9097, 0x155c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1560, 0x00000000); + nv_mthd(dev, 0x9097, 0x1564, 0x00001fff); + nv_mthd(dev, 0x9097, 0x1574, 0x00000000); + nv_mthd(dev, 0x9097, 0x1578, 0x00000000); + nv_mthd(dev, 0x9097, 0x157c, 0x003fffff); + nv_mthd(dev, 0x9097, 0x1354, 0x00000000); + nv_mthd(dev, 0x9097, 0x1664, 0x00000000); + nv_mthd(dev, 0x9097, 0x1610, 0x00000012); + nv_mthd(dev, 0x9097, 0x1608, 0x00000000); + nv_mthd(dev, 0x9097, 0x160c, 0x00000000); + nv_mthd(dev, 0x9097, 0x162c, 0x00000003); + nv_mthd(dev, 0x9097, 0x0210, 0x00000000); + nv_mthd(dev, 0x9097, 0x0320, 0x00000000); + nv_mthd(dev, 0x9097, 0x0324, 0x3f800000); + nv_mthd(dev, 0x9097, 0x0328, 0x3f800000); + nv_mthd(dev, 0x9097, 0x032c, 0x3f800000); + nv_mthd(dev, 0x9097, 0x0330, 0x3f800000); + nv_mthd(dev, 0x9097, 0x0334, 0x3f800000); + nv_mthd(dev, 0x9097, 0x0338, 0x3f800000); + nv_mthd(dev, 0x9097, 0x0750, 0x00000000); + nv_mthd(dev, 0x9097, 0x0760, 0x39291909); + nv_mthd(dev, 0x9097, 0x0764, 0x79695949); + nv_mthd(dev, 0x9097, 0x0768, 0xb9a99989); + nv_mthd(dev, 0x9097, 0x076c, 0xf9e9d9c9); + nv_mthd(dev, 0x9097, 0x0770, 0x30201000); + nv_mthd(dev, 0x9097, 0x0774, 0x70605040); + nv_mthd(dev, 0x9097, 0x0778, 0x00009080); + nv_mthd(dev, 0x9097, 0x0780, 0x39291909); + nv_mthd(dev, 0x9097, 0x0784, 0x79695949); + nv_mthd(dev, 0x9097, 0x0788, 0xb9a99989); + nv_mthd(dev, 0x9097, 0x078c, 0xf9e9d9c9); + nv_mthd(dev, 0x9097, 0x07d0, 0x30201000); + nv_mthd(dev, 0x9097, 0x07d4, 0x70605040); + nv_mthd(dev, 0x9097, 0x07d8, 0x00009080); + nv_mthd(dev, 0x9097, 0x037c, 0x00000001); + nv_mthd(dev, 0x9097, 0x0740, 0x00000000); + nv_mthd(dev, 0x9097, 0x0744, 0x00000000); + nv_mthd(dev, 0x9097, 0x2600, 0x00000000); + nv_mthd(dev, 0x9097, 0x1918, 0x00000000); + nv_mthd(dev, 0x9097, 0x191c, 0x00000900); + nv_mthd(dev, 0x9097, 0x1920, 0x00000405); + nv_mthd(dev, 0x9097, 0x1308, 0x00000001); + nv_mthd(dev, 0x9097, 0x1924, 0x00000000); + nv_mthd(dev, 0x9097, 0x13ac, 0x00000000); + nv_mthd(dev, 0x9097, 0x192c, 0x00000001); + nv_mthd(dev, 0x9097, 0x193c, 0x00002c1c); + nv_mthd(dev, 0x9097, 0x0d7c, 0x00000000); + nv_mthd(dev, 0x9097, 0x0f8c, 0x00000000); + nv_mthd(dev, 0x9097, 0x02c0, 0x00000001); + nv_mthd(dev, 0x9097, 0x1510, 0x00000000); + nv_mthd(dev, 0x9097, 0x1940, 0x00000000); + nv_mthd(dev, 0x9097, 0x0ff4, 0x00000000); + nv_mthd(dev, 0x9097, 0x0ff8, 0x00000000); + nv_mthd(dev, 0x9097, 0x194c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1950, 0x00000000); + nv_mthd(dev, 0x9097, 0x1968, 0x00000000); + nv_mthd(dev, 0x9097, 0x1590, 0x0000003f); + nv_mthd(dev, 0x9097, 0x07e8, 0x00000000); + nv_mthd(dev, 0x9097, 0x07ec, 0x00000000); + nv_mthd(dev, 0x9097, 0x07f0, 0x00000000); + nv_mthd(dev, 0x9097, 0x07f4, 0x00000000); + nv_mthd(dev, 0x9097, 0x196c, 0x00000011); + nv_mthd(dev, 0x9097, 0x197c, 0x00000000); + nv_mthd(dev, 0x9097, 0x0fcc, 0x00000000); + nv_mthd(dev, 0x9097, 0x0fd0, 0x00000000); + nv_mthd(dev, 0x9097, 0x02d8, 0x00000040); + nv_mthd(dev, 0x9097, 0x1980, 0x00000080); + nv_mthd(dev, 0x9097, 0x1504, 0x00000080); + nv_mthd(dev, 0x9097, 0x1984, 0x00000000); + nv_mthd(dev, 0x9097, 0x0300, 0x00000001); + nv_mthd(dev, 0x9097, 0x13a8, 0x00000000); + nv_mthd(dev, 0x9097, 0x12ec, 0x00000000); + nv_mthd(dev, 0x9097, 0x1310, 0x00000000); + nv_mthd(dev, 0x9097, 0x1314, 0x00000001); + nv_mthd(dev, 0x9097, 0x1380, 0x00000000); + nv_mthd(dev, 0x9097, 0x1384, 0x00000001); + nv_mthd(dev, 0x9097, 0x1388, 0x00000001); + nv_mthd(dev, 0x9097, 0x138c, 0x00000001); + nv_mthd(dev, 0x9097, 0x1390, 0x00000001); + nv_mthd(dev, 0x9097, 0x1394, 0x00000000); + nv_mthd(dev, 0x9097, 0x139c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1398, 0x00000000); + nv_mthd(dev, 0x9097, 0x1594, 0x00000000); + nv_mthd(dev, 0x9097, 0x1598, 0x00000001); + nv_mthd(dev, 0x9097, 0x159c, 0x00000001); + nv_mthd(dev, 0x9097, 0x15a0, 0x00000001); + nv_mthd(dev, 0x9097, 0x15a4, 0x00000001); + nv_mthd(dev, 0x9097, 0x0f54, 0x00000000); + nv_mthd(dev, 0x9097, 0x0f58, 0x00000000); + nv_mthd(dev, 0x9097, 0x0f5c, 0x00000000); + nv_mthd(dev, 0x9097, 0x19bc, 0x00000000); + nv_mthd(dev, 0x9097, 0x0f9c, 0x00000000); + nv_mthd(dev, 0x9097, 0x0fa0, 0x00000000); + nv_mthd(dev, 0x9097, 0x12cc, 0x00000000); + nv_mthd(dev, 0x9097, 0x12e8, 0x00000000); + nv_mthd(dev, 0x9097, 0x130c, 0x00000001); + nv_mthd(dev, 0x9097, 0x1360, 0x00000000); + nv_mthd(dev, 0x9097, 0x1364, 0x00000000); + nv_mthd(dev, 0x9097, 0x1368, 0x00000000); + nv_mthd(dev, 0x9097, 0x136c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1370, 0x00000000); + nv_mthd(dev, 0x9097, 0x1374, 0x00000000); + nv_mthd(dev, 0x9097, 0x1378, 0x00000000); + nv_mthd(dev, 0x9097, 0x137c, 0x00000000); + nv_mthd(dev, 0x9097, 0x133c, 0x00000001); + nv_mthd(dev, 0x9097, 0x1340, 0x00000001); + nv_mthd(dev, 0x9097, 0x1344, 0x00000002); + nv_mthd(dev, 0x9097, 0x1348, 0x00000001); + nv_mthd(dev, 0x9097, 0x134c, 0x00000001); + nv_mthd(dev, 0x9097, 0x1350, 0x00000002); + nv_mthd(dev, 0x9097, 0x1358, 0x00000001); + nv_mthd(dev, 0x9097, 0x12e4, 0x00000000); + nv_mthd(dev, 0x9097, 0x131c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1320, 0x00000000); + nv_mthd(dev, 0x9097, 0x1324, 0x00000000); + nv_mthd(dev, 0x9097, 0x1328, 0x00000000); + nv_mthd(dev, 0x9097, 0x19c0, 0x00000000); + nv_mthd(dev, 0x9097, 0x1140, 0x00000000); + nv_mthd(dev, 0x9097, 0x19c4, 0x00000000); + nv_mthd(dev, 0x9097, 0x19c8, 0x00001500); + nv_mthd(dev, 0x9097, 0x135c, 0x00000000); + nv_mthd(dev, 0x9097, 0x0f90, 0x00000000); + nv_mthd(dev, 0x9097, 0x19e0, 0x00000001); + nv_mthd(dev, 0x9097, 0x19e4, 0x00000001); + nv_mthd(dev, 0x9097, 0x19e8, 0x00000001); + nv_mthd(dev, 0x9097, 0x19ec, 0x00000001); + nv_mthd(dev, 0x9097, 0x19f0, 0x00000001); + nv_mthd(dev, 0x9097, 0x19f4, 0x00000001); + nv_mthd(dev, 0x9097, 0x19f8, 0x00000001); + nv_mthd(dev, 0x9097, 0x19fc, 0x00000001); + nv_mthd(dev, 0x9097, 0x19cc, 0x00000001); + nv_mthd(dev, 0x9097, 0x15b8, 0x00000000); + nv_mthd(dev, 0x9097, 0x1a00, 0x00001111); + nv_mthd(dev, 0x9097, 0x1a04, 0x00000000); + nv_mthd(dev, 0x9097, 0x1a08, 0x00000000); + nv_mthd(dev, 0x9097, 0x1a0c, 0x00000000); + nv_mthd(dev, 0x9097, 0x1a10, 0x00000000); + nv_mthd(dev, 0x9097, 0x1a14, 0x00000000); + nv_mthd(dev, 0x9097, 0x1a18, 0x00000000); + nv_mthd(dev, 0x9097, 0x1a1c, 0x00000000); + nv_mthd(dev, 0x9097, 0x0d6c, 0xffff0000); + nv_mthd(dev, 0x9097, 0x0d70, 0xffff0000); + nv_mthd(dev, 0x9097, 0x10f8, 0x00001010); + nv_mthd(dev, 0x9097, 0x0d80, 0x00000000); + nv_mthd(dev, 0x9097, 0x0d84, 0x00000000); + nv_mthd(dev, 0x9097, 0x0d88, 0x00000000); + nv_mthd(dev, 0x9097, 0x0d8c, 0x00000000); + nv_mthd(dev, 0x9097, 0x0d90, 0x00000000); + nv_mthd(dev, 0x9097, 0x0da0, 0x00000000); + nv_mthd(dev, 0x9097, 0x1508, 0x80000000); + nv_mthd(dev, 0x9097, 0x150c, 0x40000000); + nv_mthd(dev, 0x9097, 0x1668, 0x00000000); + nv_mthd(dev, 0x9097, 0x0318, 0x00000008); + nv_mthd(dev, 0x9097, 0x031c, 0x00000008); + nv_mthd(dev, 0x9097, 0x0d9c, 0x00000001); + nv_mthd(dev, 0x9097, 0x07dc, 0x00000000); + nv_mthd(dev, 0x9097, 0x074c, 0x00000055); + nv_mthd(dev, 0x9097, 0x1420, 0x00000003); + nv_mthd(dev, 0x9097, 0x17bc, 0x00000000); + nv_mthd(dev, 0x9097, 0x17c0, 0x00000000); + nv_mthd(dev, 0x9097, 0x17c4, 0x00000001); + nv_mthd(dev, 0x9097, 0x1008, 0x00000008); + nv_mthd(dev, 0x9097, 0x100c, 0x00000040); + nv_mthd(dev, 0x9097, 0x1010, 0x0000012c); + nv_mthd(dev, 0x9097, 0x0d60, 0x00000040); + nv_mthd(dev, 0x9097, 0x075c, 0x00000003); + nv_mthd(dev, 0x9097, 0x1018, 0x00000020); + nv_mthd(dev, 0x9097, 0x101c, 0x00000001); + nv_mthd(dev, 0x9097, 0x1020, 0x00000020); + nv_mthd(dev, 0x9097, 0x1024, 0x00000001); + nv_mthd(dev, 0x9097, 0x1444, 0x00000000); + nv_mthd(dev, 0x9097, 0x1448, 0x00000000); + nv_mthd(dev, 0x9097, 0x144c, 0x00000000); + nv_mthd(dev, 0x9097, 0x0360, 0x20164010); + nv_mthd(dev, 0x9097, 0x0364, 0x00000020); + nv_mthd(dev, 0x9097, 0x0368, 0x00000000); + nv_mthd(dev, 0x9097, 0x0de4, 0x00000000); + nv_mthd(dev, 0x9097, 0x0204, 0x00000006); + nv_mthd(dev, 0x9097, 0x0208, 0x00000000); + nv_mthd(dev, 0x9097, 0x02cc, 0x003fffff); + nv_mthd(dev, 0x9097, 0x02d0, 0x00000c48); + nv_mthd(dev, 0x9097, 0x1220, 0x00000005); + nv_mthd(dev, 0x9097, 0x0fdc, 0x00000000); + nv_mthd(dev, 0x9097, 0x0f98, 0x00300008); + nv_mthd(dev, 0x9097, 0x1284, 0x04000080); + nv_mthd(dev, 0x9097, 0x1450, 0x00300008); + nv_mthd(dev, 0x9097, 0x1454, 0x04000080); + nv_mthd(dev, 0x9097, 0x0214, 0x00000000); + /* in trace, right after 0x90c0, not here */ + nv_mthd(dev, 0x9097, 0x3410, 0x80002006); +} + +static void +nvc0_grctx_generate_902d(struct drm_device *dev) +{ + nv_mthd(dev, 0x902d, 0x0200, 0x000000cf); + nv_mthd(dev, 0x902d, 0x0204, 0x00000001); + nv_mthd(dev, 0x902d, 0x0208, 0x00000020); + nv_mthd(dev, 0x902d, 0x020c, 0x00000001); + nv_mthd(dev, 0x902d, 0x0210, 0x00000000); + nv_mthd(dev, 0x902d, 0x0214, 0x00000080); + nv_mthd(dev, 0x902d, 0x0218, 0x00000100); + nv_mthd(dev, 0x902d, 0x021c, 0x00000100); + nv_mthd(dev, 0x902d, 0x0220, 0x00000000); + nv_mthd(dev, 0x902d, 0x0224, 0x00000000); + nv_mthd(dev, 0x902d, 0x0230, 0x000000cf); + nv_mthd(dev, 0x902d, 0x0234, 0x00000001); + nv_mthd(dev, 0x902d, 0x0238, 0x00000020); + nv_mthd(dev, 0x902d, 0x023c, 0x00000001); + nv_mthd(dev, 0x902d, 0x0244, 0x00000080); + nv_mthd(dev, 0x902d, 0x0248, 0x00000100); + nv_mthd(dev, 0x902d, 0x024c, 0x00000100); +} + +static void +nvc0_grctx_generate_9039(struct drm_device *dev) +{ + nv_mthd(dev, 0x9039, 0x030c, 0x00000000); + nv_mthd(dev, 0x9039, 0x0310, 0x00000000); + nv_mthd(dev, 0x9039, 0x0314, 0x00000000); + nv_mthd(dev, 0x9039, 0x0320, 0x00000000); + nv_mthd(dev, 0x9039, 0x0238, 0x00000000); + nv_mthd(dev, 0x9039, 0x023c, 0x00000000); + nv_mthd(dev, 0x9039, 0x0318, 0x00000000); + nv_mthd(dev, 0x9039, 0x031c, 0x00000000); +} + +static void +nvc0_grctx_generate_90c0(struct drm_device *dev) +{ + nv_mthd(dev, 0x90c0, 0x270c, 0x00000000); + nv_mthd(dev, 0x90c0, 0x272c, 0x00000000); + nv_mthd(dev, 0x90c0, 0x274c, 0x00000000); + nv_mthd(dev, 0x90c0, 0x276c, 0x00000000); + nv_mthd(dev, 0x90c0, 0x278c, 0x00000000); + nv_mthd(dev, 0x90c0, 0x27ac, 0x00000000); + nv_mthd(dev, 0x90c0, 0x27cc, 0x00000000); + nv_mthd(dev, 0x90c0, 0x27ec, 0x00000000); + nv_mthd(dev, 0x90c0, 0x030c, 0x00000001); + nv_mthd(dev, 0x90c0, 0x1944, 0x00000000); + nv_mthd(dev, 0x90c0, 0x0758, 0x00000100); + nv_mthd(dev, 0x90c0, 0x02c4, 0x00000000); + nv_mthd(dev, 0x90c0, 0x0790, 0x00000000); + nv_mthd(dev, 0x90c0, 0x0794, 0x00000000); + nv_mthd(dev, 0x90c0, 0x0798, 0x00000000); + nv_mthd(dev, 0x90c0, 0x079c, 0x00000000); + nv_mthd(dev, 0x90c0, 0x07a0, 0x00000000); + nv_mthd(dev, 0x90c0, 0x077c, 0x00000000); + nv_mthd(dev, 0x90c0, 0x0204, 0x00000000); + nv_mthd(dev, 0x90c0, 0x0208, 0x00000000); + nv_mthd(dev, 0x90c0, 0x020c, 0x00000000); + nv_mthd(dev, 0x90c0, 0x0214, 0x00000000); + nv_mthd(dev, 0x90c0, 0x024c, 0x00000000); + nv_mthd(dev, 0x90c0, 0x0d94, 0x00000001); + nv_mthd(dev, 0x90c0, 0x1608, 0x00000000); + nv_mthd(dev, 0x90c0, 0x160c, 0x00000000); + nv_mthd(dev, 0x90c0, 0x1664, 0x00000000); +} + +static void +nvc0_grctx_generate_dispatch(struct drm_device *dev) +{ + int i; + + nv_wr32(dev, 0x404004, 0x00000000); + nv_wr32(dev, 0x404008, 0x00000000); + nv_wr32(dev, 0x40400c, 0x00000000); + nv_wr32(dev, 0x404010, 0x00000000); + nv_wr32(dev, 0x404014, 0x00000000); + nv_wr32(dev, 0x404018, 0x00000000); + nv_wr32(dev, 0x40401c, 0x00000000); + nv_wr32(dev, 0x404020, 0x00000000); + nv_wr32(dev, 0x404024, 0x00000000); + nv_wr32(dev, 0x404028, 0x00000000); + nv_wr32(dev, 0x40402c, 0x00000000); + nv_wr32(dev, 0x404044, 0x00000000); + nv_wr32(dev, 0x404094, 0x00000000); + nv_wr32(dev, 0x404098, 0x00000000); + nv_wr32(dev, 0x40409c, 0x00000000); + nv_wr32(dev, 0x4040a0, 0x00000000); + nv_wr32(dev, 0x4040a4, 0x00000000); + nv_wr32(dev, 0x4040a8, 0x00000000); + nv_wr32(dev, 0x4040ac, 0x00000000); + nv_wr32(dev, 0x4040b0, 0x00000000); + nv_wr32(dev, 0x4040b4, 0x00000000); + nv_wr32(dev, 0x4040b8, 0x00000000); + nv_wr32(dev, 0x4040bc, 0x00000000); + nv_wr32(dev, 0x4040c0, 0x00000000); + nv_wr32(dev, 0x4040c4, 0x00000000); + nv_wr32(dev, 0x4040c8, 0xf0000087); + nv_wr32(dev, 0x4040d4, 0x00000000); + nv_wr32(dev, 0x4040d8, 0x00000000); + nv_wr32(dev, 0x4040dc, 0x00000000); + nv_wr32(dev, 0x4040e0, 0x00000000); + nv_wr32(dev, 0x4040e4, 0x00000000); + nv_wr32(dev, 0x4040e8, 0x00001000); + nv_wr32(dev, 0x4040f8, 0x00000000); + nv_wr32(dev, 0x404130, 0x00000000); + nv_wr32(dev, 0x404134, 0x00000000); + nv_wr32(dev, 0x404138, 0x20000040); + nv_wr32(dev, 0x404150, 0x0000002e); + nv_wr32(dev, 0x404154, 0x00000400); + nv_wr32(dev, 0x404158, 0x00000200); + nv_wr32(dev, 0x404164, 0x00000055); + nv_wr32(dev, 0x404168, 0x00000000); + nv_wr32(dev, 0x404174, 0x00000000); + nv_wr32(dev, 0x404178, 0x00000000); + nv_wr32(dev, 0x40417c, 0x00000000); + for (i = 0; i < 8; i++) + nv_wr32(dev, 0x404200 + (i * 4), 0x00000000); /* subc */ +} + +static void +nvc0_grctx_generate_macro(struct drm_device *dev) +{ + nv_wr32(dev, 0x404404, 0x00000000); + nv_wr32(dev, 0x404408, 0x00000000); + nv_wr32(dev, 0x40440c, 0x00000000); + nv_wr32(dev, 0x404410, 0x00000000); + nv_wr32(dev, 0x404414, 0x00000000); + nv_wr32(dev, 0x404418, 0x00000000); + nv_wr32(dev, 0x40441c, 0x00000000); + nv_wr32(dev, 0x404420, 0x00000000); + nv_wr32(dev, 0x404424, 0x00000000); + nv_wr32(dev, 0x404428, 0x00000000); + nv_wr32(dev, 0x40442c, 0x00000000); + nv_wr32(dev, 0x404430, 0x00000000); + nv_wr32(dev, 0x404434, 0x00000000); + nv_wr32(dev, 0x404438, 0x00000000); + nv_wr32(dev, 0x404460, 0x00000000); + nv_wr32(dev, 0x404464, 0x00000000); + nv_wr32(dev, 0x404468, 0x00ffffff); + nv_wr32(dev, 0x40446c, 0x00000000); + nv_wr32(dev, 0x404480, 0x00000001); + nv_wr32(dev, 0x404498, 0x00000001); +} + +static void +nvc0_grctx_generate_m2mf(struct drm_device *dev) +{ + nv_wr32(dev, 0x404604, 0x00000015); + nv_wr32(dev, 0x404608, 0x00000000); + nv_wr32(dev, 0x40460c, 0x00002e00); + nv_wr32(dev, 0x404610, 0x00000100); + nv_wr32(dev, 0x404618, 0x00000000); + nv_wr32(dev, 0x40461c, 0x00000000); + nv_wr32(dev, 0x404620, 0x00000000); + nv_wr32(dev, 0x404624, 0x00000000); + nv_wr32(dev, 0x404628, 0x00000000); + nv_wr32(dev, 0x40462c, 0x00000000); + nv_wr32(dev, 0x404630, 0x00000000); + nv_wr32(dev, 0x404634, 0x00000000); + nv_wr32(dev, 0x404638, 0x00000004); + nv_wr32(dev, 0x40463c, 0x00000000); + nv_wr32(dev, 0x404640, 0x00000000); + nv_wr32(dev, 0x404644, 0x00000000); + nv_wr32(dev, 0x404648, 0x00000000); + nv_wr32(dev, 0x40464c, 0x00000000); + nv_wr32(dev, 0x404650, 0x00000000); + nv_wr32(dev, 0x404654, 0x00000000); + nv_wr32(dev, 0x404658, 0x00000000); + nv_wr32(dev, 0x40465c, 0x007f0100); + nv_wr32(dev, 0x404660, 0x00000000); + nv_wr32(dev, 0x404664, 0x00000000); + nv_wr32(dev, 0x404668, 0x00000000); + nv_wr32(dev, 0x40466c, 0x00000000); + nv_wr32(dev, 0x404670, 0x00000000); + nv_wr32(dev, 0x404674, 0x00000000); + nv_wr32(dev, 0x404678, 0x00000000); + nv_wr32(dev, 0x40467c, 0x00000002); + nv_wr32(dev, 0x404680, 0x00000000); + nv_wr32(dev, 0x404684, 0x00000000); + nv_wr32(dev, 0x404688, 0x00000000); + nv_wr32(dev, 0x40468c, 0x00000000); + nv_wr32(dev, 0x404690, 0x00000000); + nv_wr32(dev, 0x404694, 0x00000000); + nv_wr32(dev, 0x404698, 0x00000000); + nv_wr32(dev, 0x40469c, 0x00000000); + nv_wr32(dev, 0x4046a0, 0x007f0080); + nv_wr32(dev, 0x4046a4, 0x00000000); + nv_wr32(dev, 0x4046a8, 0x00000000); + nv_wr32(dev, 0x4046ac, 0x00000000); + nv_wr32(dev, 0x4046b0, 0x00000000); + nv_wr32(dev, 0x4046b4, 0x00000000); + nv_wr32(dev, 0x4046b8, 0x00000000); + nv_wr32(dev, 0x4046bc, 0x00000000); + nv_wr32(dev, 0x4046c0, 0x00000000); + nv_wr32(dev, 0x4046c4, 0x00000000); + nv_wr32(dev, 0x4046c8, 0x00000000); + nv_wr32(dev, 0x4046cc, 0x00000000); + nv_wr32(dev, 0x4046d0, 0x00000000); + nv_wr32(dev, 0x4046d4, 0x00000000); + nv_wr32(dev, 0x4046d8, 0x00000000); + nv_wr32(dev, 0x4046dc, 0x00000000); + nv_wr32(dev, 0x4046e0, 0x00000000); + nv_wr32(dev, 0x4046e4, 0x00000000); + nv_wr32(dev, 0x4046e8, 0x00000000); + nv_wr32(dev, 0x4046f0, 0x00000000); + nv_wr32(dev, 0x4046f4, 0x00000000); +} + +static void +nvc0_grctx_generate_unk47xx(struct drm_device *dev) +{ + nv_wr32(dev, 0x404700, 0x00000000); + nv_wr32(dev, 0x404704, 0x00000000); + nv_wr32(dev, 0x404708, 0x00000000); + nv_wr32(dev, 0x40470c, 0x00000000); + nv_wr32(dev, 0x404710, 0x00000000); + nv_wr32(dev, 0x404714, 0x00000000); + nv_wr32(dev, 0x404718, 0x00000000); + nv_wr32(dev, 0x40471c, 0x00000000); + nv_wr32(dev, 0x404720, 0x00000000); + nv_wr32(dev, 0x404724, 0x00000000); + nv_wr32(dev, 0x404728, 0x00000000); + nv_wr32(dev, 0x40472c, 0x00000000); + nv_wr32(dev, 0x404730, 0x00000000); + nv_wr32(dev, 0x404734, 0x00000100); + nv_wr32(dev, 0x404738, 0x00000000); + nv_wr32(dev, 0x40473c, 0x00000000); + nv_wr32(dev, 0x404740, 0x00000000); + nv_wr32(dev, 0x404744, 0x00000000); + nv_wr32(dev, 0x404748, 0x00000000); + nv_wr32(dev, 0x40474c, 0x00000000); + nv_wr32(dev, 0x404750, 0x00000000); + nv_wr32(dev, 0x404754, 0x00000000); +} + +static void +nvc0_grctx_generate_shaders(struct drm_device *dev) +{ + nv_wr32(dev, 0x405800, 0x078000bf); + nv_wr32(dev, 0x405830, 0x02180000); + nv_wr32(dev, 0x405834, 0x00000000); + nv_wr32(dev, 0x405838, 0x00000000); + nv_wr32(dev, 0x405854, 0x00000000); + nv_wr32(dev, 0x405870, 0x00000001); + nv_wr32(dev, 0x405874, 0x00000001); + nv_wr32(dev, 0x405878, 0x00000001); + nv_wr32(dev, 0x40587c, 0x00000001); + nv_wr32(dev, 0x405a00, 0x00000000); + nv_wr32(dev, 0x405a04, 0x00000000); + nv_wr32(dev, 0x405a18, 0x00000000); +} + +static void +nvc0_grctx_generate_unk60xx(struct drm_device *dev) +{ + nv_wr32(dev, 0x406020, 0x000103c1); + nv_wr32(dev, 0x406028, 0x00000001); + nv_wr32(dev, 0x40602c, 0x00000001); + nv_wr32(dev, 0x406030, 0x00000001); + nv_wr32(dev, 0x406034, 0x00000001); +} + +static void +nvc0_grctx_generate_unk64xx(struct drm_device *dev) +{ + nv_wr32(dev, 0x4064a8, 0x00000000); + nv_wr32(dev, 0x4064ac, 0x00003fff); + nv_wr32(dev, 0x4064b4, 0x00000000); + nv_wr32(dev, 0x4064b8, 0x00000000); +} + +static void +nvc0_grctx_generate_tpbus(struct drm_device *dev) +{ + nv_wr32(dev, 0x407804, 0x00000023); + nv_wr32(dev, 0x40780c, 0x0a418820); + nv_wr32(dev, 0x407810, 0x062080e6); + nv_wr32(dev, 0x407814, 0x020398a4); + nv_wr32(dev, 0x407818, 0x0e629062); + nv_wr32(dev, 0x40781c, 0x0a418820); + nv_wr32(dev, 0x407820, 0x000000e6); + nv_wr32(dev, 0x4078bc, 0x00000103); +} + +static void +nvc0_grctx_generate_ccache(struct drm_device *dev) +{ + nv_wr32(dev, 0x408000, 0x00000000); + nv_wr32(dev, 0x408004, 0x00000000); + nv_wr32(dev, 0x408008, 0x00000018); + nv_wr32(dev, 0x40800c, 0x00000000); + nv_wr32(dev, 0x408010, 0x00000000); + nv_wr32(dev, 0x408014, 0x00000069); + nv_wr32(dev, 0x408018, 0xe100e100); + nv_wr32(dev, 0x408064, 0x00000000); +} + +static void +nvc0_grctx_generate_rop(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + /* ROPC_BROADCAST */ + nv_wr32(dev, 0x408800, 0x02802a3c); + nv_wr32(dev, 0x408804, 0x00000040); + nv_wr32(dev, 0x408808, 0x0003e00d); + switch (dev_priv->chipset) { + case 0xc0: + nv_wr32(dev, 0x408900, 0x0080b801); + break; + case 0xc3: + case 0xc4: + nv_wr32(dev, 0x408900, 0x3080b801); + break; + } + nv_wr32(dev, 0x408904, 0x02000001); + nv_wr32(dev, 0x408908, 0x00c80929); + nv_wr32(dev, 0x40890c, 0x00000000); + nv_wr32(dev, 0x408980, 0x0000011d); +} + +static void +nvc0_grctx_generate_gpc(struct drm_device *dev) +{ + int i; + + /* GPC_BROADCAST */ + nv_wr32(dev, 0x418380, 0x00000016); + nv_wr32(dev, 0x418400, 0x38004e00); + nv_wr32(dev, 0x418404, 0x71e0ffff); + nv_wr32(dev, 0x418408, 0x00000000); + nv_wr32(dev, 0x41840c, 0x00001008); + nv_wr32(dev, 0x418410, 0x0fff0fff); + nv_wr32(dev, 0x418414, 0x00200fff); + nv_wr32(dev, 0x418450, 0x00000000); + nv_wr32(dev, 0x418454, 0x00000000); + nv_wr32(dev, 0x418458, 0x00000000); + nv_wr32(dev, 0x41845c, 0x00000000); + nv_wr32(dev, 0x418460, 0x00000000); + nv_wr32(dev, 0x418464, 0x00000000); + nv_wr32(dev, 0x418468, 0x00000001); + nv_wr32(dev, 0x41846c, 0x00000000); + nv_wr32(dev, 0x418470, 0x00000000); + nv_wr32(dev, 0x418600, 0x0000001f); + nv_wr32(dev, 0x418684, 0x0000000f); + nv_wr32(dev, 0x418700, 0x00000002); + nv_wr32(dev, 0x418704, 0x00000080); + nv_wr32(dev, 0x418708, 0x00000000); + nv_wr32(dev, 0x41870c, 0x07c80000); + nv_wr32(dev, 0x418710, 0x00000000); + nv_wr32(dev, 0x418800, 0x0006860a); + nv_wr32(dev, 0x418808, 0x00000000); + nv_wr32(dev, 0x41880c, 0x00000000); + nv_wr32(dev, 0x418810, 0x00000000); + nv_wr32(dev, 0x418828, 0x00008442); + nv_wr32(dev, 0x418830, 0x00000001); + nv_wr32(dev, 0x4188d8, 0x00000008); + nv_wr32(dev, 0x4188e0, 0x01000000); + nv_wr32(dev, 0x4188e8, 0x00000000); + nv_wr32(dev, 0x4188ec, 0x00000000); + nv_wr32(dev, 0x4188f0, 0x00000000); + nv_wr32(dev, 0x4188f4, 0x00000000); + nv_wr32(dev, 0x4188f8, 0x00000000); + nv_wr32(dev, 0x4188fc, 0x00100000); + nv_wr32(dev, 0x41891c, 0x00ff00ff); + nv_wr32(dev, 0x418924, 0x00000000); + nv_wr32(dev, 0x418928, 0x00ffff00); + nv_wr32(dev, 0x41892c, 0x0000ff00); + for (i = 0; i < 8; i++) { + nv_wr32(dev, 0x418a00 + (i * 0x20), 0x00000000); + nv_wr32(dev, 0x418a04 + (i * 0x20), 0x00000000); + nv_wr32(dev, 0x418a08 + (i * 0x20), 0x00000000); + nv_wr32(dev, 0x418a0c + (i * 0x20), 0x00010000); + nv_wr32(dev, 0x418a10 + (i * 0x20), 0x00000000); + nv_wr32(dev, 0x418a14 + (i * 0x20), 0x00000000); + nv_wr32(dev, 0x418a18 + (i * 0x20), 0x00000000); + } + nv_wr32(dev, 0x418b00, 0x00000000); + nv_wr32(dev, 0x418b08, 0x0a418820); + nv_wr32(dev, 0x418b0c, 0x062080e6); + nv_wr32(dev, 0x418b10, 0x020398a4); + nv_wr32(dev, 0x418b14, 0x0e629062); + nv_wr32(dev, 0x418b18, 0x0a418820); + nv_wr32(dev, 0x418b1c, 0x000000e6); + nv_wr32(dev, 0x418bb8, 0x00000103); + nv_wr32(dev, 0x418c08, 0x00000001); + nv_wr32(dev, 0x418c10, 0x00000000); + nv_wr32(dev, 0x418c14, 0x00000000); + nv_wr32(dev, 0x418c18, 0x00000000); + nv_wr32(dev, 0x418c1c, 0x00000000); + nv_wr32(dev, 0x418c20, 0x00000000); + nv_wr32(dev, 0x418c24, 0x00000000); + nv_wr32(dev, 0x418c28, 0x00000000); + nv_wr32(dev, 0x418c2c, 0x00000000); + nv_wr32(dev, 0x418c80, 0x20200004); + nv_wr32(dev, 0x418c8c, 0x00000001); + nv_wr32(dev, 0x419000, 0x00000780); + nv_wr32(dev, 0x419004, 0x00000000); + nv_wr32(dev, 0x419008, 0x00000000); + nv_wr32(dev, 0x419014, 0x00000004); +} + +static void +nvc0_grctx_generate_tp(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + /* GPC_BROADCAST.TP_BROADCAST */ + nv_wr32(dev, 0x419848, 0x00000000); + nv_wr32(dev, 0x419864, 0x0000012a); + nv_wr32(dev, 0x419888, 0x00000000); + nv_wr32(dev, 0x419a00, 0x000001f0); + nv_wr32(dev, 0x419a04, 0x00000001); + nv_wr32(dev, 0x419a08, 0x00000023); + nv_wr32(dev, 0x419a0c, 0x00020000); + nv_wr32(dev, 0x419a10, 0x00000000); + nv_wr32(dev, 0x419a14, 0x00000200); + nv_wr32(dev, 0x419a1c, 0x00000000); + nv_wr32(dev, 0x419a20, 0x00000800); + if (dev_priv->chipset != 0xc0) + nv_wr32(dev, 0x00419ac4, 0x0007f440); /* 0xc3 */ + nv_wr32(dev, 0x419b00, 0x0a418820); + nv_wr32(dev, 0x419b04, 0x062080e6); + nv_wr32(dev, 0x419b08, 0x020398a4); + nv_wr32(dev, 0x419b0c, 0x0e629062); + nv_wr32(dev, 0x419b10, 0x0a418820); + nv_wr32(dev, 0x419b14, 0x000000e6); + nv_wr32(dev, 0x419bd0, 0x00900103); + nv_wr32(dev, 0x419be0, 0x00000001); + nv_wr32(dev, 0x419be4, 0x00000000); + nv_wr32(dev, 0x419c00, 0x00000002); + nv_wr32(dev, 0x419c04, 0x00000006); + nv_wr32(dev, 0x419c08, 0x00000002); + nv_wr32(dev, 0x419c20, 0x00000000); + nv_wr32(dev, 0x419cbc, 0x28137606); + nv_wr32(dev, 0x419ce8, 0x00000000); + nv_wr32(dev, 0x419cf4, 0x00000183); + nv_wr32(dev, 0x419d20, 0x02180000); + nv_wr32(dev, 0x419d24, 0x00001fff); + nv_wr32(dev, 0x419e04, 0x00000000); + nv_wr32(dev, 0x419e08, 0x00000000); + nv_wr32(dev, 0x419e0c, 0x00000000); + nv_wr32(dev, 0x419e10, 0x00000002); + nv_wr32(dev, 0x419e44, 0x001beff2); + nv_wr32(dev, 0x419e48, 0x00000000); + nv_wr32(dev, 0x419e4c, 0x0000000f); + nv_wr32(dev, 0x419e50, 0x00000000); + nv_wr32(dev, 0x419e54, 0x00000000); + nv_wr32(dev, 0x419e58, 0x00000000); + nv_wr32(dev, 0x419e5c, 0x00000000); + nv_wr32(dev, 0x419e60, 0x00000000); + nv_wr32(dev, 0x419e64, 0x00000000); + nv_wr32(dev, 0x419e68, 0x00000000); + nv_wr32(dev, 0x419e6c, 0x00000000); + nv_wr32(dev, 0x419e70, 0x00000000); + nv_wr32(dev, 0x419e74, 0x00000000); + nv_wr32(dev, 0x419e78, 0x00000000); + nv_wr32(dev, 0x419e7c, 0x00000000); + nv_wr32(dev, 0x419e80, 0x00000000); + nv_wr32(dev, 0x419e84, 0x00000000); + nv_wr32(dev, 0x419e88, 0x00000000); + nv_wr32(dev, 0x419e8c, 0x00000000); + nv_wr32(dev, 0x419e90, 0x00000000); + nv_wr32(dev, 0x419e98, 0x00000000); + if (dev_priv->chipset != 0xc0) + nv_wr32(dev, 0x419ee0, 0x00011110); + nv_wr32(dev, 0x419f50, 0x00000000); + nv_wr32(dev, 0x419f54, 0x00000000); + if (dev_priv->chipset != 0xc0) + nv_wr32(dev, 0x419f58, 0x00000000); +} + +int +nvc0_grctx_generate(struct nouveau_channel *chan) +{ + struct drm_nouveau_private *dev_priv = chan->dev->dev_private; + struct nvc0_graph_priv *priv = nv_engine(chan->dev, NVOBJ_ENGINE_GR); + struct nvc0_graph_chan *grch = chan->engctx[NVOBJ_ENGINE_GR]; + struct drm_device *dev = chan->dev; + int i, gpc, tp, id; + u32 r000260, tmp; + + r000260 = nv_rd32(dev, 0x000260); + nv_wr32(dev, 0x000260, r000260 & ~1); + nv_wr32(dev, 0x400208, 0x00000000); + + nvc0_grctx_generate_dispatch(dev); + nvc0_grctx_generate_macro(dev); + nvc0_grctx_generate_m2mf(dev); + nvc0_grctx_generate_unk47xx(dev); + nvc0_grctx_generate_shaders(dev); + nvc0_grctx_generate_unk60xx(dev); + nvc0_grctx_generate_unk64xx(dev); + nvc0_grctx_generate_tpbus(dev); + nvc0_grctx_generate_ccache(dev); + nvc0_grctx_generate_rop(dev); + nvc0_grctx_generate_gpc(dev); + nvc0_grctx_generate_tp(dev); + + nv_wr32(dev, 0x404154, 0x00000000); + + /* fuc "mmio list" writes */ + for (i = 0; i < grch->mmio_nr * 8; i += 8) { + u32 reg = nv_ro32(grch->mmio, i + 0); + nv_wr32(dev, reg, nv_ro32(grch->mmio, i + 4)); + } + + for (tp = 0, id = 0; tp < 4; tp++) { + for (gpc = 0; gpc < priv->gpc_nr; gpc++) { + if (tp < priv->tp_nr[gpc]) { + nv_wr32(dev, TP_UNIT(gpc, tp, 0x698), id); + nv_wr32(dev, TP_UNIT(gpc, tp, 0x4e8), id); + nv_wr32(dev, GPC_UNIT(gpc, 0x0c10 + tp * 4), id); + nv_wr32(dev, TP_UNIT(gpc, tp, 0x088), id); + id++; + } + + nv_wr32(dev, GPC_UNIT(gpc, 0x0c08), priv->tp_nr[gpc]); + nv_wr32(dev, GPC_UNIT(gpc, 0x0c8c), priv->tp_nr[gpc]); + } + } + + tmp = 0; + for (i = 0; i < priv->gpc_nr; i++) + tmp |= priv->tp_nr[i] << (i * 4); + nv_wr32(dev, 0x406028, tmp); + nv_wr32(dev, 0x405870, tmp); + + nv_wr32(dev, 0x40602c, 0x00000000); + nv_wr32(dev, 0x405874, 0x00000000); + nv_wr32(dev, 0x406030, 0x00000000); + nv_wr32(dev, 0x405878, 0x00000000); + nv_wr32(dev, 0x406034, 0x00000000); + nv_wr32(dev, 0x40587c, 0x00000000); + + if (1) { + const u8 chipset_tp_max[] = { 16, 0, 0, 4, 8 }; + u8 max = chipset_tp_max[dev_priv->chipset & 0x0f]; + u8 tpnr[GPC_MAX]; + u8 data[32]; + + memcpy(tpnr, priv->tp_nr, sizeof(priv->tp_nr)); + memset(data, 0x1f, sizeof(data)); + + gpc = -1; + for (tp = 0; tp < priv->tp_total; tp++) { + do { + gpc = (gpc + 1) % priv->gpc_nr; + } while (!tpnr[gpc]); + tpnr[gpc]--; + data[tp] = gpc; + } + + for (i = 0; i < max / 4; i++) + nv_wr32(dev, 0x4060a8 + (i * 4), ((u32 *)data)[i]); + } + + if (1) { + u32 data[6] = {}, data2[2] = {}; + u8 tpnr[GPC_MAX]; + u8 shift, ntpcv; + + /* calculate first set of magics */ + memcpy(tpnr, priv->tp_nr, sizeof(priv->tp_nr)); + + for (tp = 0; tp < priv->tp_total; tp++) { + do { + gpc = (gpc + 1) % priv->gpc_nr; + } while (!tpnr[gpc]); + tpnr[gpc]--; + + data[tp / 6] |= gpc << ((tp % 6) * 5); + } + + for (; tp < 32; tp++) + data[tp / 6] |= 7 << ((tp % 6) * 5); + + /* and the second... */ + shift = 0; + ntpcv = priv->tp_total; + while (!(ntpcv & (1 << 4))) { + ntpcv <<= 1; + shift++; + } + + data2[0] = (ntpcv << 16); + data2[0] |= (shift << 21); + data2[0] |= (((1 << (0 + 5)) % ntpcv) << 24); + for (i = 1; i < 7; i++) + data2[1] |= ((1 << (i + 5)) % ntpcv) << ((i - 1) * 5); + + /* GPC_BROADCAST */ + nv_wr32(dev, 0x418bb8, (priv->tp_total << 8) | + priv->magic_not_rop_nr); + for (i = 0; i < 6; i++) + nv_wr32(dev, 0x418b08 + (i * 4), data[i]); + + /* GPC_BROADCAST.TP_BROADCAST */ + nv_wr32(dev, 0x419bd0, (priv->tp_total << 8) | + priv->magic_not_rop_nr | + data2[0]); + nv_wr32(dev, 0x419be4, data2[1]); + for (i = 0; i < 6; i++) + nv_wr32(dev, 0x419b00 + (i * 4), data[i]); + + /* UNK78xx */ + nv_wr32(dev, 0x4078bc, (priv->tp_total << 8) | + priv->magic_not_rop_nr); + for (i = 0; i < 6; i++) + nv_wr32(dev, 0x40780c + (i * 4), data[i]); + } + + if (1) { + u32 tp_mask = 0, tp_set = 0; + u8 tpnr[GPC_MAX]; + + memcpy(tpnr, priv->tp_nr, sizeof(priv->tp_nr)); + for (gpc = 0; gpc < priv->gpc_nr; gpc++) + tp_mask |= ((1 << priv->tp_nr[gpc]) - 1) << (gpc * 8); + + gpc = -1; + for (i = 0, gpc = -1; i < 32; i++) { + int ltp = i * (priv->tp_total - 1) / 32; + + do { + gpc = (gpc + 1) % priv->gpc_nr; + } while (!tpnr[gpc]); + tp = priv->tp_nr[gpc] - tpnr[gpc]--; + + tp_set |= 1 << ((gpc * 8) + tp); + + do { + nv_wr32(dev, 0x406800 + (i * 0x20), tp_set); + tp_set ^= tp_mask; + nv_wr32(dev, 0x406c00 + (i * 0x20), tp_set); + tp_set ^= tp_mask; + } while (ltp == (++i * (priv->tp_total - 1) / 32)); + i--; + } + } + + nv_wr32(dev, 0x400208, 0x80000000); + + nv_icmd(dev, 0x00001000, 0x00000004); + nv_icmd(dev, 0x000000a9, 0x0000ffff); + nv_icmd(dev, 0x00000038, 0x0fac6881); + nv_icmd(dev, 0x0000003d, 0x00000001); + nv_icmd(dev, 0x000000e8, 0x00000400); + nv_icmd(dev, 0x000000e9, 0x00000400); + nv_icmd(dev, 0x000000ea, 0x00000400); + nv_icmd(dev, 0x000000eb, 0x00000400); + nv_icmd(dev, 0x000000ec, 0x00000400); + nv_icmd(dev, 0x000000ed, 0x00000400); + nv_icmd(dev, 0x000000ee, 0x00000400); + nv_icmd(dev, 0x000000ef, 0x00000400); + nv_icmd(dev, 0x00000078, 0x00000300); + nv_icmd(dev, 0x00000079, 0x00000300); + nv_icmd(dev, 0x0000007a, 0x00000300); + nv_icmd(dev, 0x0000007b, 0x00000300); + nv_icmd(dev, 0x0000007c, 0x00000300); + nv_icmd(dev, 0x0000007d, 0x00000300); + nv_icmd(dev, 0x0000007e, 0x00000300); + nv_icmd(dev, 0x0000007f, 0x00000300); + nv_icmd(dev, 0x00000050, 0x00000011); + nv_icmd(dev, 0x00000058, 0x00000008); + nv_icmd(dev, 0x00000059, 0x00000008); + nv_icmd(dev, 0x0000005a, 0x00000008); + nv_icmd(dev, 0x0000005b, 0x00000008); + nv_icmd(dev, 0x0000005c, 0x00000008); + nv_icmd(dev, 0x0000005d, 0x00000008); + nv_icmd(dev, 0x0000005e, 0x00000008); + nv_icmd(dev, 0x0000005f, 0x00000008); + nv_icmd(dev, 0x00000208, 0x00000001); + nv_icmd(dev, 0x00000209, 0x00000001); + nv_icmd(dev, 0x0000020a, 0x00000001); + nv_icmd(dev, 0x0000020b, 0x00000001); + nv_icmd(dev, 0x0000020c, 0x00000001); + nv_icmd(dev, 0x0000020d, 0x00000001); + nv_icmd(dev, 0x0000020e, 0x00000001); + nv_icmd(dev, 0x0000020f, 0x00000001); + nv_icmd(dev, 0x00000081, 0x00000001); + nv_icmd(dev, 0x00000085, 0x00000004); + nv_icmd(dev, 0x00000088, 0x00000400); + nv_icmd(dev, 0x00000090, 0x00000300); + nv_icmd(dev, 0x00000098, 0x00001001); + nv_icmd(dev, 0x000000e3, 0x00000001); + nv_icmd(dev, 0x000000da, 0x00000001); + nv_icmd(dev, 0x000000f8, 0x00000003); + nv_icmd(dev, 0x000000fa, 0x00000001); + nv_icmd(dev, 0x0000009f, 0x0000ffff); + nv_icmd(dev, 0x000000a0, 0x0000ffff); + nv_icmd(dev, 0x000000a1, 0x0000ffff); + nv_icmd(dev, 0x000000a2, 0x0000ffff); + nv_icmd(dev, 0x000000b1, 0x00000001); + nv_icmd(dev, 0x000000b2, 0x00000000); + nv_icmd(dev, 0x000000b3, 0x00000000); + nv_icmd(dev, 0x000000b4, 0x00000000); + nv_icmd(dev, 0x000000b5, 0x00000000); + nv_icmd(dev, 0x000000b6, 0x00000000); + nv_icmd(dev, 0x000000b7, 0x00000000); + nv_icmd(dev, 0x000000b8, 0x00000000); + nv_icmd(dev, 0x000000b9, 0x00000000); + nv_icmd(dev, 0x000000ba, 0x00000000); + nv_icmd(dev, 0x000000bb, 0x00000000); + nv_icmd(dev, 0x000000bc, 0x00000000); + nv_icmd(dev, 0x000000bd, 0x00000000); + nv_icmd(dev, 0x000000be, 0x00000000); + nv_icmd(dev, 0x000000bf, 0x00000000); + nv_icmd(dev, 0x000000c0, 0x00000000); + nv_icmd(dev, 0x000000c1, 0x00000000); + nv_icmd(dev, 0x000000c2, 0x00000000); + nv_icmd(dev, 0x000000c3, 0x00000000); + nv_icmd(dev, 0x000000c4, 0x00000000); + nv_icmd(dev, 0x000000c5, 0x00000000); + nv_icmd(dev, 0x000000c6, 0x00000000); + nv_icmd(dev, 0x000000c7, 0x00000000); + nv_icmd(dev, 0x000000c8, 0x00000000); + nv_icmd(dev, 0x000000c9, 0x00000000); + nv_icmd(dev, 0x000000ca, 0x00000000); + nv_icmd(dev, 0x000000cb, 0x00000000); + nv_icmd(dev, 0x000000cc, 0x00000000); + nv_icmd(dev, 0x000000cd, 0x00000000); + nv_icmd(dev, 0x000000ce, 0x00000000); + nv_icmd(dev, 0x000000cf, 0x00000000); + nv_icmd(dev, 0x000000d0, 0x00000000); + nv_icmd(dev, 0x000000d1, 0x00000000); + nv_icmd(dev, 0x000000d2, 0x00000000); + nv_icmd(dev, 0x000000d3, 0x00000000); + nv_icmd(dev, 0x000000d4, 0x00000000); + nv_icmd(dev, 0x000000d5, 0x00000000); + nv_icmd(dev, 0x000000d6, 0x00000000); + nv_icmd(dev, 0x000000d7, 0x00000000); + nv_icmd(dev, 0x000000d8, 0x00000000); + nv_icmd(dev, 0x000000d9, 0x00000000); + nv_icmd(dev, 0x00000210, 0x00000040); + nv_icmd(dev, 0x00000211, 0x00000040); + nv_icmd(dev, 0x00000212, 0x00000040); + nv_icmd(dev, 0x00000213, 0x00000040); + nv_icmd(dev, 0x00000214, 0x00000040); + nv_icmd(dev, 0x00000215, 0x00000040); + nv_icmd(dev, 0x00000216, 0x00000040); + nv_icmd(dev, 0x00000217, 0x00000040); + nv_icmd(dev, 0x00000218, 0x0000c080); + nv_icmd(dev, 0x00000219, 0x0000c080); + nv_icmd(dev, 0x0000021a, 0x0000c080); + nv_icmd(dev, 0x0000021b, 0x0000c080); + nv_icmd(dev, 0x0000021c, 0x0000c080); + nv_icmd(dev, 0x0000021d, 0x0000c080); + nv_icmd(dev, 0x0000021e, 0x0000c080); + nv_icmd(dev, 0x0000021f, 0x0000c080); + nv_icmd(dev, 0x000000ad, 0x0000013e); + nv_icmd(dev, 0x000000e1, 0x00000010); + nv_icmd(dev, 0x00000290, 0x00000000); + nv_icmd(dev, 0x00000291, 0x00000000); + nv_icmd(dev, 0x00000292, 0x00000000); + nv_icmd(dev, 0x00000293, 0x00000000); + nv_icmd(dev, 0x00000294, 0x00000000); + nv_icmd(dev, 0x00000295, 0x00000000); + nv_icmd(dev, 0x00000296, 0x00000000); + nv_icmd(dev, 0x00000297, 0x00000000); + nv_icmd(dev, 0x00000298, 0x00000000); + nv_icmd(dev, 0x00000299, 0x00000000); + nv_icmd(dev, 0x0000029a, 0x00000000); + nv_icmd(dev, 0x0000029b, 0x00000000); + nv_icmd(dev, 0x0000029c, 0x00000000); + nv_icmd(dev, 0x0000029d, 0x00000000); + nv_icmd(dev, 0x0000029e, 0x00000000); + nv_icmd(dev, 0x0000029f, 0x00000000); + nv_icmd(dev, 0x000003b0, 0x00000000); + nv_icmd(dev, 0x000003b1, 0x00000000); + nv_icmd(dev, 0x000003b2, 0x00000000); + nv_icmd(dev, 0x000003b3, 0x00000000); + nv_icmd(dev, 0x000003b4, 0x00000000); + nv_icmd(dev, 0x000003b5, 0x00000000); + nv_icmd(dev, 0x000003b6, 0x00000000); + nv_icmd(dev, 0x000003b7, 0x00000000); + nv_icmd(dev, 0x000003b8, 0x00000000); + nv_icmd(dev, 0x000003b9, 0x00000000); + nv_icmd(dev, 0x000003ba, 0x00000000); + nv_icmd(dev, 0x000003bb, 0x00000000); + nv_icmd(dev, 0x000003bc, 0x00000000); + nv_icmd(dev, 0x000003bd, 0x00000000); + nv_icmd(dev, 0x000003be, 0x00000000); + nv_icmd(dev, 0x000003bf, 0x00000000); + nv_icmd(dev, 0x000002a0, 0x00000000); + nv_icmd(dev, 0x000002a1, 0x00000000); + nv_icmd(dev, 0x000002a2, 0x00000000); + nv_icmd(dev, 0x000002a3, 0x00000000); + nv_icmd(dev, 0x000002a4, 0x00000000); + nv_icmd(dev, 0x000002a5, 0x00000000); + nv_icmd(dev, 0x000002a6, 0x00000000); + nv_icmd(dev, 0x000002a7, 0x00000000); + nv_icmd(dev, 0x000002a8, 0x00000000); + nv_icmd(dev, 0x000002a9, 0x00000000); + nv_icmd(dev, 0x000002aa, 0x00000000); + nv_icmd(dev, 0x000002ab, 0x00000000); + nv_icmd(dev, 0x000002ac, 0x00000000); + nv_icmd(dev, 0x000002ad, 0x00000000); + nv_icmd(dev, 0x000002ae, 0x00000000); + nv_icmd(dev, 0x000002af, 0x00000000); + nv_icmd(dev, 0x00000420, 0x00000000); + nv_icmd(dev, 0x00000421, 0x00000000); + nv_icmd(dev, 0x00000422, 0x00000000); + nv_icmd(dev, 0x00000423, 0x00000000); + nv_icmd(dev, 0x00000424, 0x00000000); + nv_icmd(dev, 0x00000425, 0x00000000); + nv_icmd(dev, 0x00000426, 0x00000000); + nv_icmd(dev, 0x00000427, 0x00000000); + nv_icmd(dev, 0x00000428, 0x00000000); + nv_icmd(dev, 0x00000429, 0x00000000); + nv_icmd(dev, 0x0000042a, 0x00000000); + nv_icmd(dev, 0x0000042b, 0x00000000); + nv_icmd(dev, 0x0000042c, 0x00000000); + nv_icmd(dev, 0x0000042d, 0x00000000); + nv_icmd(dev, 0x0000042e, 0x00000000); + nv_icmd(dev, 0x0000042f, 0x00000000); + nv_icmd(dev, 0x000002b0, 0x00000000); + nv_icmd(dev, 0x000002b1, 0x00000000); + nv_icmd(dev, 0x000002b2, 0x00000000); + nv_icmd(dev, 0x000002b3, 0x00000000); + nv_icmd(dev, 0x000002b4, 0x00000000); + nv_icmd(dev, 0x000002b5, 0x00000000); + nv_icmd(dev, 0x000002b6, 0x00000000); + nv_icmd(dev, 0x000002b7, 0x00000000); + nv_icmd(dev, 0x000002b8, 0x00000000); + nv_icmd(dev, 0x000002b9, 0x00000000); + nv_icmd(dev, 0x000002ba, 0x00000000); + nv_icmd(dev, 0x000002bb, 0x00000000); + nv_icmd(dev, 0x000002bc, 0x00000000); + nv_icmd(dev, 0x000002bd, 0x00000000); + nv_icmd(dev, 0x000002be, 0x00000000); + nv_icmd(dev, 0x000002bf, 0x00000000); + nv_icmd(dev, 0x00000430, 0x00000000); + nv_icmd(dev, 0x00000431, 0x00000000); + nv_icmd(dev, 0x00000432, 0x00000000); + nv_icmd(dev, 0x00000433, 0x00000000); + nv_icmd(dev, 0x00000434, 0x00000000); + nv_icmd(dev, 0x00000435, 0x00000000); + nv_icmd(dev, 0x00000436, 0x00000000); + nv_icmd(dev, 0x00000437, 0x00000000); + nv_icmd(dev, 0x00000438, 0x00000000); + nv_icmd(dev, 0x00000439, 0x00000000); + nv_icmd(dev, 0x0000043a, 0x00000000); + nv_icmd(dev, 0x0000043b, 0x00000000); + nv_icmd(dev, 0x0000043c, 0x00000000); + nv_icmd(dev, 0x0000043d, 0x00000000); + nv_icmd(dev, 0x0000043e, 0x00000000); + nv_icmd(dev, 0x0000043f, 0x00000000); + nv_icmd(dev, 0x000002c0, 0x00000000); + nv_icmd(dev, 0x000002c1, 0x00000000); + nv_icmd(dev, 0x000002c2, 0x00000000); + nv_icmd(dev, 0x000002c3, 0x00000000); + nv_icmd(dev, 0x000002c4, 0x00000000); + nv_icmd(dev, 0x000002c5, 0x00000000); + nv_icmd(dev, 0x000002c6, 0x00000000); + nv_icmd(dev, 0x000002c7, 0x00000000); + nv_icmd(dev, 0x000002c8, 0x00000000); + nv_icmd(dev, 0x000002c9, 0x00000000); + nv_icmd(dev, 0x000002ca, 0x00000000); + nv_icmd(dev, 0x000002cb, 0x00000000); + nv_icmd(dev, 0x000002cc, 0x00000000); + nv_icmd(dev, 0x000002cd, 0x00000000); + nv_icmd(dev, 0x000002ce, 0x00000000); + nv_icmd(dev, 0x000002cf, 0x00000000); + nv_icmd(dev, 0x000004d0, 0x00000000); + nv_icmd(dev, 0x000004d1, 0x00000000); + nv_icmd(dev, 0x000004d2, 0x00000000); + nv_icmd(dev, 0x000004d3, 0x00000000); + nv_icmd(dev, 0x000004d4, 0x00000000); + nv_icmd(dev, 0x000004d5, 0x00000000); + nv_icmd(dev, 0x000004d6, 0x00000000); + nv_icmd(dev, 0x000004d7, 0x00000000); + nv_icmd(dev, 0x000004d8, 0x00000000); + nv_icmd(dev, 0x000004d9, 0x00000000); + nv_icmd(dev, 0x000004da, 0x00000000); + nv_icmd(dev, 0x000004db, 0x00000000); + nv_icmd(dev, 0x000004dc, 0x00000000); + nv_icmd(dev, 0x000004dd, 0x00000000); + nv_icmd(dev, 0x000004de, 0x00000000); + nv_icmd(dev, 0x000004df, 0x00000000); + nv_icmd(dev, 0x00000720, 0x00000000); + nv_icmd(dev, 0x00000721, 0x00000000); + nv_icmd(dev, 0x00000722, 0x00000000); + nv_icmd(dev, 0x00000723, 0x00000000); + nv_icmd(dev, 0x00000724, 0x00000000); + nv_icmd(dev, 0x00000725, 0x00000000); + nv_icmd(dev, 0x00000726, 0x00000000); + nv_icmd(dev, 0x00000727, 0x00000000); + nv_icmd(dev, 0x00000728, 0x00000000); + nv_icmd(dev, 0x00000729, 0x00000000); + nv_icmd(dev, 0x0000072a, 0x00000000); + nv_icmd(dev, 0x0000072b, 0x00000000); + nv_icmd(dev, 0x0000072c, 0x00000000); + nv_icmd(dev, 0x0000072d, 0x00000000); + nv_icmd(dev, 0x0000072e, 0x00000000); + nv_icmd(dev, 0x0000072f, 0x00000000); + nv_icmd(dev, 0x000008c0, 0x00000000); + nv_icmd(dev, 0x000008c1, 0x00000000); + nv_icmd(dev, 0x000008c2, 0x00000000); + nv_icmd(dev, 0x000008c3, 0x00000000); + nv_icmd(dev, 0x000008c4, 0x00000000); + nv_icmd(dev, 0x000008c5, 0x00000000); + nv_icmd(dev, 0x000008c6, 0x00000000); + nv_icmd(dev, 0x000008c7, 0x00000000); + nv_icmd(dev, 0x000008c8, 0x00000000); + nv_icmd(dev, 0x000008c9, 0x00000000); + nv_icmd(dev, 0x000008ca, 0x00000000); + nv_icmd(dev, 0x000008cb, 0x00000000); + nv_icmd(dev, 0x000008cc, 0x00000000); + nv_icmd(dev, 0x000008cd, 0x00000000); + nv_icmd(dev, 0x000008ce, 0x00000000); + nv_icmd(dev, 0x000008cf, 0x00000000); + nv_icmd(dev, 0x00000890, 0x00000000); + nv_icmd(dev, 0x00000891, 0x00000000); + nv_icmd(dev, 0x00000892, 0x00000000); + nv_icmd(dev, 0x00000893, 0x00000000); + nv_icmd(dev, 0x00000894, 0x00000000); + nv_icmd(dev, 0x00000895, 0x00000000); + nv_icmd(dev, 0x00000896, 0x00000000); + nv_icmd(dev, 0x00000897, 0x00000000); + nv_icmd(dev, 0x00000898, 0x00000000); + nv_icmd(dev, 0x00000899, 0x00000000); + nv_icmd(dev, 0x0000089a, 0x00000000); + nv_icmd(dev, 0x0000089b, 0x00000000); + nv_icmd(dev, 0x0000089c, 0x00000000); + nv_icmd(dev, 0x0000089d, 0x00000000); + nv_icmd(dev, 0x0000089e, 0x00000000); + nv_icmd(dev, 0x0000089f, 0x00000000); + nv_icmd(dev, 0x000008e0, 0x00000000); + nv_icmd(dev, 0x000008e1, 0x00000000); + nv_icmd(dev, 0x000008e2, 0x00000000); + nv_icmd(dev, 0x000008e3, 0x00000000); + nv_icmd(dev, 0x000008e4, 0x00000000); + nv_icmd(dev, 0x000008e5, 0x00000000); + nv_icmd(dev, 0x000008e6, 0x00000000); + nv_icmd(dev, 0x000008e7, 0x00000000); + nv_icmd(dev, 0x000008e8, 0x00000000); + nv_icmd(dev, 0x000008e9, 0x00000000); + nv_icmd(dev, 0x000008ea, 0x00000000); + nv_icmd(dev, 0x000008eb, 0x00000000); + nv_icmd(dev, 0x000008ec, 0x00000000); + nv_icmd(dev, 0x000008ed, 0x00000000); + nv_icmd(dev, 0x000008ee, 0x00000000); + nv_icmd(dev, 0x000008ef, 0x00000000); + nv_icmd(dev, 0x000008a0, 0x00000000); + nv_icmd(dev, 0x000008a1, 0x00000000); + nv_icmd(dev, 0x000008a2, 0x00000000); + nv_icmd(dev, 0x000008a3, 0x00000000); + nv_icmd(dev, 0x000008a4, 0x00000000); + nv_icmd(dev, 0x000008a5, 0x00000000); + nv_icmd(dev, 0x000008a6, 0x00000000); + nv_icmd(dev, 0x000008a7, 0x00000000); + nv_icmd(dev, 0x000008a8, 0x00000000); + nv_icmd(dev, 0x000008a9, 0x00000000); + nv_icmd(dev, 0x000008aa, 0x00000000); + nv_icmd(dev, 0x000008ab, 0x00000000); + nv_icmd(dev, 0x000008ac, 0x00000000); + nv_icmd(dev, 0x000008ad, 0x00000000); + nv_icmd(dev, 0x000008ae, 0x00000000); + nv_icmd(dev, 0x000008af, 0x00000000); + nv_icmd(dev, 0x000008f0, 0x00000000); + nv_icmd(dev, 0x000008f1, 0x00000000); + nv_icmd(dev, 0x000008f2, 0x00000000); + nv_icmd(dev, 0x000008f3, 0x00000000); + nv_icmd(dev, 0x000008f4, 0x00000000); + nv_icmd(dev, 0x000008f5, 0x00000000); + nv_icmd(dev, 0x000008f6, 0x00000000); + nv_icmd(dev, 0x000008f7, 0x00000000); + nv_icmd(dev, 0x000008f8, 0x00000000); + nv_icmd(dev, 0x000008f9, 0x00000000); + nv_icmd(dev, 0x000008fa, 0x00000000); + nv_icmd(dev, 0x000008fb, 0x00000000); + nv_icmd(dev, 0x000008fc, 0x00000000); + nv_icmd(dev, 0x000008fd, 0x00000000); + nv_icmd(dev, 0x000008fe, 0x00000000); + nv_icmd(dev, 0x000008ff, 0x00000000); + nv_icmd(dev, 0x0000094c, 0x000000ff); + nv_icmd(dev, 0x0000094d, 0xffffffff); + nv_icmd(dev, 0x0000094e, 0x00000002); + nv_icmd(dev, 0x000002ec, 0x00000001); + nv_icmd(dev, 0x00000303, 0x00000001); + nv_icmd(dev, 0x000002e6, 0x00000001); + nv_icmd(dev, 0x00000466, 0x00000052); + nv_icmd(dev, 0x00000301, 0x3f800000); + nv_icmd(dev, 0x00000304, 0x30201000); + nv_icmd(dev, 0x00000305, 0x70605040); + nv_icmd(dev, 0x00000306, 0xb8a89888); + nv_icmd(dev, 0x00000307, 0xf8e8d8c8); + nv_icmd(dev, 0x0000030a, 0x00ffff00); + nv_icmd(dev, 0x0000030b, 0x0000001a); + nv_icmd(dev, 0x0000030c, 0x00000001); + nv_icmd(dev, 0x00000318, 0x00000001); + nv_icmd(dev, 0x00000340, 0x00000000); + nv_icmd(dev, 0x00000375, 0x00000001); + nv_icmd(dev, 0x00000351, 0x00000100); + nv_icmd(dev, 0x0000037d, 0x00000006); + nv_icmd(dev, 0x000003a0, 0x00000002); + nv_icmd(dev, 0x000003aa, 0x00000001); + nv_icmd(dev, 0x000003a9, 0x00000001); + nv_icmd(dev, 0x00000380, 0x00000001); + nv_icmd(dev, 0x00000360, 0x00000040); + nv_icmd(dev, 0x00000366, 0x00000000); + nv_icmd(dev, 0x00000367, 0x00000000); + nv_icmd(dev, 0x00000368, 0x00001fff); + nv_icmd(dev, 0x00000370, 0x00000000); + nv_icmd(dev, 0x00000371, 0x00000000); + nv_icmd(dev, 0x00000372, 0x003fffff); + nv_icmd(dev, 0x0000037a, 0x00000012); + nv_icmd(dev, 0x000005e0, 0x00000022); + nv_icmd(dev, 0x000005e1, 0x00000022); + nv_icmd(dev, 0x000005e2, 0x00000022); + nv_icmd(dev, 0x000005e3, 0x00000022); + nv_icmd(dev, 0x000005e4, 0x00000022); + nv_icmd(dev, 0x00000619, 0x00000003); + nv_icmd(dev, 0x00000811, 0x00000003); + nv_icmd(dev, 0x00000812, 0x00000004); + nv_icmd(dev, 0x00000813, 0x00000006); + nv_icmd(dev, 0x00000814, 0x00000008); + nv_icmd(dev, 0x00000815, 0x0000000b); + nv_icmd(dev, 0x00000800, 0x00000001); + nv_icmd(dev, 0x00000801, 0x00000001); + nv_icmd(dev, 0x00000802, 0x00000001); + nv_icmd(dev, 0x00000803, 0x00000001); + nv_icmd(dev, 0x00000804, 0x00000001); + nv_icmd(dev, 0x00000805, 0x00000001); + nv_icmd(dev, 0x00000632, 0x00000001); + nv_icmd(dev, 0x00000633, 0x00000002); + nv_icmd(dev, 0x00000634, 0x00000003); + nv_icmd(dev, 0x00000635, 0x00000004); + nv_icmd(dev, 0x00000654, 0x3f800000); + nv_icmd(dev, 0x00000657, 0x3f800000); + nv_icmd(dev, 0x00000655, 0x3f800000); + nv_icmd(dev, 0x00000656, 0x3f800000); + nv_icmd(dev, 0x000006cd, 0x3f800000); + nv_icmd(dev, 0x000007f5, 0x3f800000); + nv_icmd(dev, 0x000007dc, 0x39291909); + nv_icmd(dev, 0x000007dd, 0x79695949); + nv_icmd(dev, 0x000007de, 0xb9a99989); + nv_icmd(dev, 0x000007df, 0xf9e9d9c9); + nv_icmd(dev, 0x000007e8, 0x00003210); + nv_icmd(dev, 0x000007e9, 0x00007654); + nv_icmd(dev, 0x000007ea, 0x00000098); + nv_icmd(dev, 0x000007ec, 0x39291909); + nv_icmd(dev, 0x000007ed, 0x79695949); + nv_icmd(dev, 0x000007ee, 0xb9a99989); + nv_icmd(dev, 0x000007ef, 0xf9e9d9c9); + nv_icmd(dev, 0x000007f0, 0x00003210); + nv_icmd(dev, 0x000007f1, 0x00007654); + nv_icmd(dev, 0x000007f2, 0x00000098); + nv_icmd(dev, 0x000005a5, 0x00000001); + nv_icmd(dev, 0x00000980, 0x00000000); + nv_icmd(dev, 0x00000981, 0x00000000); + nv_icmd(dev, 0x00000982, 0x00000000); + nv_icmd(dev, 0x00000983, 0x00000000); + nv_icmd(dev, 0x00000984, 0x00000000); + nv_icmd(dev, 0x00000985, 0x00000000); + nv_icmd(dev, 0x00000986, 0x00000000); + nv_icmd(dev, 0x00000987, 0x00000000); + nv_icmd(dev, 0x00000988, 0x00000000); + nv_icmd(dev, 0x00000989, 0x00000000); + nv_icmd(dev, 0x0000098a, 0x00000000); + nv_icmd(dev, 0x0000098b, 0x00000000); + nv_icmd(dev, 0x0000098c, 0x00000000); + nv_icmd(dev, 0x0000098d, 0x00000000); + nv_icmd(dev, 0x0000098e, 0x00000000); + nv_icmd(dev, 0x0000098f, 0x00000000); + nv_icmd(dev, 0x00000990, 0x00000000); + nv_icmd(dev, 0x00000991, 0x00000000); + nv_icmd(dev, 0x00000992, 0x00000000); + nv_icmd(dev, 0x00000993, 0x00000000); + nv_icmd(dev, 0x00000994, 0x00000000); + nv_icmd(dev, 0x00000995, 0x00000000); + nv_icmd(dev, 0x00000996, 0x00000000); + nv_icmd(dev, 0x00000997, 0x00000000); + nv_icmd(dev, 0x00000998, 0x00000000); + nv_icmd(dev, 0x00000999, 0x00000000); + nv_icmd(dev, 0x0000099a, 0x00000000); + nv_icmd(dev, 0x0000099b, 0x00000000); + nv_icmd(dev, 0x0000099c, 0x00000000); + nv_icmd(dev, 0x0000099d, 0x00000000); + nv_icmd(dev, 0x0000099e, 0x00000000); + nv_icmd(dev, 0x0000099f, 0x00000000); + nv_icmd(dev, 0x000009a0, 0x00000000); + nv_icmd(dev, 0x000009a1, 0x00000000); + nv_icmd(dev, 0x000009a2, 0x00000000); + nv_icmd(dev, 0x000009a3, 0x00000000); + nv_icmd(dev, 0x000009a4, 0x00000000); + nv_icmd(dev, 0x000009a5, 0x00000000); + nv_icmd(dev, 0x000009a6, 0x00000000); + nv_icmd(dev, 0x000009a7, 0x00000000); + nv_icmd(dev, 0x000009a8, 0x00000000); + nv_icmd(dev, 0x000009a9, 0x00000000); + nv_icmd(dev, 0x000009aa, 0x00000000); + nv_icmd(dev, 0x000009ab, 0x00000000); + nv_icmd(dev, 0x000009ac, 0x00000000); + nv_icmd(dev, 0x000009ad, 0x00000000); + nv_icmd(dev, 0x000009ae, 0x00000000); + nv_icmd(dev, 0x000009af, 0x00000000); + nv_icmd(dev, 0x000009b0, 0x00000000); + nv_icmd(dev, 0x000009b1, 0x00000000); + nv_icmd(dev, 0x000009b2, 0x00000000); + nv_icmd(dev, 0x000009b3, 0x00000000); + nv_icmd(dev, 0x000009b4, 0x00000000); + nv_icmd(dev, 0x000009b5, 0x00000000); + nv_icmd(dev, 0x000009b6, 0x00000000); + nv_icmd(dev, 0x000009b7, 0x00000000); + nv_icmd(dev, 0x000009b8, 0x00000000); + nv_icmd(dev, 0x000009b9, 0x00000000); + nv_icmd(dev, 0x000009ba, 0x00000000); + nv_icmd(dev, 0x000009bb, 0x00000000); + nv_icmd(dev, 0x000009bc, 0x00000000); + nv_icmd(dev, 0x000009bd, 0x00000000); + nv_icmd(dev, 0x000009be, 0x00000000); + nv_icmd(dev, 0x000009bf, 0x00000000); + nv_icmd(dev, 0x000009c0, 0x00000000); + nv_icmd(dev, 0x000009c1, 0x00000000); + nv_icmd(dev, 0x000009c2, 0x00000000); + nv_icmd(dev, 0x000009c3, 0x00000000); + nv_icmd(dev, 0x000009c4, 0x00000000); + nv_icmd(dev, 0x000009c5, 0x00000000); + nv_icmd(dev, 0x000009c6, 0x00000000); + nv_icmd(dev, 0x000009c7, 0x00000000); + nv_icmd(dev, 0x000009c8, 0x00000000); + nv_icmd(dev, 0x000009c9, 0x00000000); + nv_icmd(dev, 0x000009ca, 0x00000000); + nv_icmd(dev, 0x000009cb, 0x00000000); + nv_icmd(dev, 0x000009cc, 0x00000000); + nv_icmd(dev, 0x000009cd, 0x00000000); + nv_icmd(dev, 0x000009ce, 0x00000000); + nv_icmd(dev, 0x000009cf, 0x00000000); + nv_icmd(dev, 0x000009d0, 0x00000000); + nv_icmd(dev, 0x000009d1, 0x00000000); + nv_icmd(dev, 0x000009d2, 0x00000000); + nv_icmd(dev, 0x000009d3, 0x00000000); + nv_icmd(dev, 0x000009d4, 0x00000000); + nv_icmd(dev, 0x000009d5, 0x00000000); + nv_icmd(dev, 0x000009d6, 0x00000000); + nv_icmd(dev, 0x000009d7, 0x00000000); + nv_icmd(dev, 0x000009d8, 0x00000000); + nv_icmd(dev, 0x000009d9, 0x00000000); + nv_icmd(dev, 0x000009da, 0x00000000); + nv_icmd(dev, 0x000009db, 0x00000000); + nv_icmd(dev, 0x000009dc, 0x00000000); + nv_icmd(dev, 0x000009dd, 0x00000000); + nv_icmd(dev, 0x000009de, 0x00000000); + nv_icmd(dev, 0x000009df, 0x00000000); + nv_icmd(dev, 0x000009e0, 0x00000000); + nv_icmd(dev, 0x000009e1, 0x00000000); + nv_icmd(dev, 0x000009e2, 0x00000000); + nv_icmd(dev, 0x000009e3, 0x00000000); + nv_icmd(dev, 0x000009e4, 0x00000000); + nv_icmd(dev, 0x000009e5, 0x00000000); + nv_icmd(dev, 0x000009e6, 0x00000000); + nv_icmd(dev, 0x000009e7, 0x00000000); + nv_icmd(dev, 0x000009e8, 0x00000000); + nv_icmd(dev, 0x000009e9, 0x00000000); + nv_icmd(dev, 0x000009ea, 0x00000000); + nv_icmd(dev, 0x000009eb, 0x00000000); + nv_icmd(dev, 0x000009ec, 0x00000000); + nv_icmd(dev, 0x000009ed, 0x00000000); + nv_icmd(dev, 0x000009ee, 0x00000000); + nv_icmd(dev, 0x000009ef, 0x00000000); + nv_icmd(dev, 0x000009f0, 0x00000000); + nv_icmd(dev, 0x000009f1, 0x00000000); + nv_icmd(dev, 0x000009f2, 0x00000000); + nv_icmd(dev, 0x000009f3, 0x00000000); + nv_icmd(dev, 0x000009f4, 0x00000000); + nv_icmd(dev, 0x000009f5, 0x00000000); + nv_icmd(dev, 0x000009f6, 0x00000000); + nv_icmd(dev, 0x000009f7, 0x00000000); + nv_icmd(dev, 0x000009f8, 0x00000000); + nv_icmd(dev, 0x000009f9, 0x00000000); + nv_icmd(dev, 0x000009fa, 0x00000000); + nv_icmd(dev, 0x000009fb, 0x00000000); + nv_icmd(dev, 0x000009fc, 0x00000000); + nv_icmd(dev, 0x000009fd, 0x00000000); + nv_icmd(dev, 0x000009fe, 0x00000000); + nv_icmd(dev, 0x000009ff, 0x00000000); + nv_icmd(dev, 0x00000468, 0x00000004); + nv_icmd(dev, 0x0000046c, 0x00000001); + nv_icmd(dev, 0x00000470, 0x00000000); + nv_icmd(dev, 0x00000471, 0x00000000); + nv_icmd(dev, 0x00000472, 0x00000000); + nv_icmd(dev, 0x00000473, 0x00000000); + nv_icmd(dev, 0x00000474, 0x00000000); + nv_icmd(dev, 0x00000475, 0x00000000); + nv_icmd(dev, 0x00000476, 0x00000000); + nv_icmd(dev, 0x00000477, 0x00000000); + nv_icmd(dev, 0x00000478, 0x00000000); + nv_icmd(dev, 0x00000479, 0x00000000); + nv_icmd(dev, 0x0000047a, 0x00000000); + nv_icmd(dev, 0x0000047b, 0x00000000); + nv_icmd(dev, 0x0000047c, 0x00000000); + nv_icmd(dev, 0x0000047d, 0x00000000); + nv_icmd(dev, 0x0000047e, 0x00000000); + nv_icmd(dev, 0x0000047f, 0x00000000); + nv_icmd(dev, 0x00000480, 0x00000000); + nv_icmd(dev, 0x00000481, 0x00000000); + nv_icmd(dev, 0x00000482, 0x00000000); + nv_icmd(dev, 0x00000483, 0x00000000); + nv_icmd(dev, 0x00000484, 0x00000000); + nv_icmd(dev, 0x00000485, 0x00000000); + nv_icmd(dev, 0x00000486, 0x00000000); + nv_icmd(dev, 0x00000487, 0x00000000); + nv_icmd(dev, 0x00000488, 0x00000000); + nv_icmd(dev, 0x00000489, 0x00000000); + nv_icmd(dev, 0x0000048a, 0x00000000); + nv_icmd(dev, 0x0000048b, 0x00000000); + nv_icmd(dev, 0x0000048c, 0x00000000); + nv_icmd(dev, 0x0000048d, 0x00000000); + nv_icmd(dev, 0x0000048e, 0x00000000); + nv_icmd(dev, 0x0000048f, 0x00000000); + nv_icmd(dev, 0x00000490, 0x00000000); + nv_icmd(dev, 0x00000491, 0x00000000); + nv_icmd(dev, 0x00000492, 0x00000000); + nv_icmd(dev, 0x00000493, 0x00000000); + nv_icmd(dev, 0x00000494, 0x00000000); + nv_icmd(dev, 0x00000495, 0x00000000); + nv_icmd(dev, 0x00000496, 0x00000000); + nv_icmd(dev, 0x00000497, 0x00000000); + nv_icmd(dev, 0x00000498, 0x00000000); + nv_icmd(dev, 0x00000499, 0x00000000); + nv_icmd(dev, 0x0000049a, 0x00000000); + nv_icmd(dev, 0x0000049b, 0x00000000); + nv_icmd(dev, 0x0000049c, 0x00000000); + nv_icmd(dev, 0x0000049d, 0x00000000); + nv_icmd(dev, 0x0000049e, 0x00000000); + nv_icmd(dev, 0x0000049f, 0x00000000); + nv_icmd(dev, 0x000004a0, 0x00000000); + nv_icmd(dev, 0x000004a1, 0x00000000); + nv_icmd(dev, 0x000004a2, 0x00000000); + nv_icmd(dev, 0x000004a3, 0x00000000); + nv_icmd(dev, 0x000004a4, 0x00000000); + nv_icmd(dev, 0x000004a5, 0x00000000); + nv_icmd(dev, 0x000004a6, 0x00000000); + nv_icmd(dev, 0x000004a7, 0x00000000); + nv_icmd(dev, 0x000004a8, 0x00000000); + nv_icmd(dev, 0x000004a9, 0x00000000); + nv_icmd(dev, 0x000004aa, 0x00000000); + nv_icmd(dev, 0x000004ab, 0x00000000); + nv_icmd(dev, 0x000004ac, 0x00000000); + nv_icmd(dev, 0x000004ad, 0x00000000); + nv_icmd(dev, 0x000004ae, 0x00000000); + nv_icmd(dev, 0x000004af, 0x00000000); + nv_icmd(dev, 0x000004b0, 0x00000000); + nv_icmd(dev, 0x000004b1, 0x00000000); + nv_icmd(dev, 0x000004b2, 0x00000000); + nv_icmd(dev, 0x000004b3, 0x00000000); + nv_icmd(dev, 0x000004b4, 0x00000000); + nv_icmd(dev, 0x000004b5, 0x00000000); + nv_icmd(dev, 0x000004b6, 0x00000000); + nv_icmd(dev, 0x000004b7, 0x00000000); + nv_icmd(dev, 0x000004b8, 0x00000000); + nv_icmd(dev, 0x000004b9, 0x00000000); + nv_icmd(dev, 0x000004ba, 0x00000000); + nv_icmd(dev, 0x000004bb, 0x00000000); + nv_icmd(dev, 0x000004bc, 0x00000000); + nv_icmd(dev, 0x000004bd, 0x00000000); + nv_icmd(dev, 0x000004be, 0x00000000); + nv_icmd(dev, 0x000004bf, 0x00000000); + nv_icmd(dev, 0x000004c0, 0x00000000); + nv_icmd(dev, 0x000004c1, 0x00000000); + nv_icmd(dev, 0x000004c2, 0x00000000); + nv_icmd(dev, 0x000004c3, 0x00000000); + nv_icmd(dev, 0x000004c4, 0x00000000); + nv_icmd(dev, 0x000004c5, 0x00000000); + nv_icmd(dev, 0x000004c6, 0x00000000); + nv_icmd(dev, 0x000004c7, 0x00000000); + nv_icmd(dev, 0x000004c8, 0x00000000); + nv_icmd(dev, 0x000004c9, 0x00000000); + nv_icmd(dev, 0x000004ca, 0x00000000); + nv_icmd(dev, 0x000004cb, 0x00000000); + nv_icmd(dev, 0x000004cc, 0x00000000); + nv_icmd(dev, 0x000004cd, 0x00000000); + nv_icmd(dev, 0x000004ce, 0x00000000); + nv_icmd(dev, 0x000004cf, 0x00000000); + nv_icmd(dev, 0x00000510, 0x3f800000); + nv_icmd(dev, 0x00000511, 0x3f800000); + nv_icmd(dev, 0x00000512, 0x3f800000); + nv_icmd(dev, 0x00000513, 0x3f800000); + nv_icmd(dev, 0x00000514, 0x3f800000); + nv_icmd(dev, 0x00000515, 0x3f800000); + nv_icmd(dev, 0x00000516, 0x3f800000); + nv_icmd(dev, 0x00000517, 0x3f800000); + nv_icmd(dev, 0x00000518, 0x3f800000); + nv_icmd(dev, 0x00000519, 0x3f800000); + nv_icmd(dev, 0x0000051a, 0x3f800000); + nv_icmd(dev, 0x0000051b, 0x3f800000); + nv_icmd(dev, 0x0000051c, 0x3f800000); + nv_icmd(dev, 0x0000051d, 0x3f800000); + nv_icmd(dev, 0x0000051e, 0x3f800000); + nv_icmd(dev, 0x0000051f, 0x3f800000); + nv_icmd(dev, 0x00000520, 0x000002b6); + nv_icmd(dev, 0x00000529, 0x00000001); + nv_icmd(dev, 0x00000530, 0xffff0000); + nv_icmd(dev, 0x00000531, 0xffff0000); + nv_icmd(dev, 0x00000532, 0xffff0000); + nv_icmd(dev, 0x00000533, 0xffff0000); + nv_icmd(dev, 0x00000534, 0xffff0000); + nv_icmd(dev, 0x00000535, 0xffff0000); + nv_icmd(dev, 0x00000536, 0xffff0000); + nv_icmd(dev, 0x00000537, 0xffff0000); + nv_icmd(dev, 0x00000538, 0xffff0000); + nv_icmd(dev, 0x00000539, 0xffff0000); + nv_icmd(dev, 0x0000053a, 0xffff0000); + nv_icmd(dev, 0x0000053b, 0xffff0000); + nv_icmd(dev, 0x0000053c, 0xffff0000); + nv_icmd(dev, 0x0000053d, 0xffff0000); + nv_icmd(dev, 0x0000053e, 0xffff0000); + nv_icmd(dev, 0x0000053f, 0xffff0000); + nv_icmd(dev, 0x00000585, 0x0000003f); + nv_icmd(dev, 0x00000576, 0x00000003); + nv_icmd(dev, 0x00000586, 0x00000040); + nv_icmd(dev, 0x00000582, 0x00000080); + nv_icmd(dev, 0x00000583, 0x00000080); + nv_icmd(dev, 0x000005c2, 0x00000001); + nv_icmd(dev, 0x00000638, 0x00000001); + nv_icmd(dev, 0x00000639, 0x00000001); + nv_icmd(dev, 0x0000063a, 0x00000002); + nv_icmd(dev, 0x0000063b, 0x00000001); + nv_icmd(dev, 0x0000063c, 0x00000001); + nv_icmd(dev, 0x0000063d, 0x00000002); + nv_icmd(dev, 0x0000063e, 0x00000001); + nv_icmd(dev, 0x000008b8, 0x00000001); + nv_icmd(dev, 0x000008b9, 0x00000001); + nv_icmd(dev, 0x000008ba, 0x00000001); + nv_icmd(dev, 0x000008bb, 0x00000001); + nv_icmd(dev, 0x000008bc, 0x00000001); + nv_icmd(dev, 0x000008bd, 0x00000001); + nv_icmd(dev, 0x000008be, 0x00000001); + nv_icmd(dev, 0x000008bf, 0x00000001); + nv_icmd(dev, 0x00000900, 0x00000001); + nv_icmd(dev, 0x00000901, 0x00000001); + nv_icmd(dev, 0x00000902, 0x00000001); + nv_icmd(dev, 0x00000903, 0x00000001); + nv_icmd(dev, 0x00000904, 0x00000001); + nv_icmd(dev, 0x00000905, 0x00000001); + nv_icmd(dev, 0x00000906, 0x00000001); + nv_icmd(dev, 0x00000907, 0x00000001); + nv_icmd(dev, 0x00000908, 0x00000002); + nv_icmd(dev, 0x00000909, 0x00000002); + nv_icmd(dev, 0x0000090a, 0x00000002); + nv_icmd(dev, 0x0000090b, 0x00000002); + nv_icmd(dev, 0x0000090c, 0x00000002); + nv_icmd(dev, 0x0000090d, 0x00000002); + nv_icmd(dev, 0x0000090e, 0x00000002); + nv_icmd(dev, 0x0000090f, 0x00000002); + nv_icmd(dev, 0x00000910, 0x00000001); + nv_icmd(dev, 0x00000911, 0x00000001); + nv_icmd(dev, 0x00000912, 0x00000001); + nv_icmd(dev, 0x00000913, 0x00000001); + nv_icmd(dev, 0x00000914, 0x00000001); + nv_icmd(dev, 0x00000915, 0x00000001); + nv_icmd(dev, 0x00000916, 0x00000001); + nv_icmd(dev, 0x00000917, 0x00000001); + nv_icmd(dev, 0x00000918, 0x00000001); + nv_icmd(dev, 0x00000919, 0x00000001); + nv_icmd(dev, 0x0000091a, 0x00000001); + nv_icmd(dev, 0x0000091b, 0x00000001); + nv_icmd(dev, 0x0000091c, 0x00000001); + nv_icmd(dev, 0x0000091d, 0x00000001); + nv_icmd(dev, 0x0000091e, 0x00000001); + nv_icmd(dev, 0x0000091f, 0x00000001); + nv_icmd(dev, 0x00000920, 0x00000002); + nv_icmd(dev, 0x00000921, 0x00000002); + nv_icmd(dev, 0x00000922, 0x00000002); + nv_icmd(dev, 0x00000923, 0x00000002); + nv_icmd(dev, 0x00000924, 0x00000002); + nv_icmd(dev, 0x00000925, 0x00000002); + nv_icmd(dev, 0x00000926, 0x00000002); + nv_icmd(dev, 0x00000927, 0x00000002); + nv_icmd(dev, 0x00000928, 0x00000001); + nv_icmd(dev, 0x00000929, 0x00000001); + nv_icmd(dev, 0x0000092a, 0x00000001); + nv_icmd(dev, 0x0000092b, 0x00000001); + nv_icmd(dev, 0x0000092c, 0x00000001); + nv_icmd(dev, 0x0000092d, 0x00000001); + nv_icmd(dev, 0x0000092e, 0x00000001); + nv_icmd(dev, 0x0000092f, 0x00000001); + nv_icmd(dev, 0x00000648, 0x00000001); + nv_icmd(dev, 0x00000649, 0x00000001); + nv_icmd(dev, 0x0000064a, 0x00000001); + nv_icmd(dev, 0x0000064b, 0x00000001); + nv_icmd(dev, 0x0000064c, 0x00000001); + nv_icmd(dev, 0x0000064d, 0x00000001); + nv_icmd(dev, 0x0000064e, 0x00000001); + nv_icmd(dev, 0x0000064f, 0x00000001); + nv_icmd(dev, 0x00000650, 0x00000001); + nv_icmd(dev, 0x00000658, 0x0000000f); + nv_icmd(dev, 0x000007ff, 0x0000000a); + nv_icmd(dev, 0x0000066a, 0x40000000); + nv_icmd(dev, 0x0000066b, 0x10000000); + nv_icmd(dev, 0x0000066c, 0xffff0000); + nv_icmd(dev, 0x0000066d, 0xffff0000); + nv_icmd(dev, 0x000007af, 0x00000008); + nv_icmd(dev, 0x000007b0, 0x00000008); + nv_icmd(dev, 0x000007f6, 0x00000001); + nv_icmd(dev, 0x000006b2, 0x00000055); + nv_icmd(dev, 0x000007ad, 0x00000003); + nv_icmd(dev, 0x00000937, 0x00000001); + nv_icmd(dev, 0x00000971, 0x00000008); + nv_icmd(dev, 0x00000972, 0x00000040); + nv_icmd(dev, 0x00000973, 0x0000012c); + nv_icmd(dev, 0x0000097c, 0x00000040); + nv_icmd(dev, 0x00000979, 0x00000003); + nv_icmd(dev, 0x00000975, 0x00000020); + nv_icmd(dev, 0x00000976, 0x00000001); + nv_icmd(dev, 0x00000977, 0x00000020); + nv_icmd(dev, 0x00000978, 0x00000001); + nv_icmd(dev, 0x00000957, 0x00000003); + nv_icmd(dev, 0x0000095e, 0x20164010); + nv_icmd(dev, 0x0000095f, 0x00000020); + nv_icmd(dev, 0x00000683, 0x00000006); + nv_icmd(dev, 0x00000685, 0x003fffff); + nv_icmd(dev, 0x00000687, 0x00000c48); + nv_icmd(dev, 0x000006a0, 0x00000005); + nv_icmd(dev, 0x00000840, 0x00300008); + nv_icmd(dev, 0x00000841, 0x04000080); + nv_icmd(dev, 0x00000842, 0x00300008); + nv_icmd(dev, 0x00000843, 0x04000080); + nv_icmd(dev, 0x00000818, 0x00000000); + nv_icmd(dev, 0x00000819, 0x00000000); + nv_icmd(dev, 0x0000081a, 0x00000000); + nv_icmd(dev, 0x0000081b, 0x00000000); + nv_icmd(dev, 0x0000081c, 0x00000000); + nv_icmd(dev, 0x0000081d, 0x00000000); + nv_icmd(dev, 0x0000081e, 0x00000000); + nv_icmd(dev, 0x0000081f, 0x00000000); + nv_icmd(dev, 0x00000848, 0x00000000); + nv_icmd(dev, 0x00000849, 0x00000000); + nv_icmd(dev, 0x0000084a, 0x00000000); + nv_icmd(dev, 0x0000084b, 0x00000000); + nv_icmd(dev, 0x0000084c, 0x00000000); + nv_icmd(dev, 0x0000084d, 0x00000000); + nv_icmd(dev, 0x0000084e, 0x00000000); + nv_icmd(dev, 0x0000084f, 0x00000000); + nv_icmd(dev, 0x00000850, 0x00000000); + nv_icmd(dev, 0x00000851, 0x00000000); + nv_icmd(dev, 0x00000852, 0x00000000); + nv_icmd(dev, 0x00000853, 0x00000000); + nv_icmd(dev, 0x00000854, 0x00000000); + nv_icmd(dev, 0x00000855, 0x00000000); + nv_icmd(dev, 0x00000856, 0x00000000); + nv_icmd(dev, 0x00000857, 0x00000000); + nv_icmd(dev, 0x00000738, 0x00000000); + nv_icmd(dev, 0x000006aa, 0x00000001); + nv_icmd(dev, 0x000006ab, 0x00000002); + nv_icmd(dev, 0x000006ac, 0x00000080); + nv_icmd(dev, 0x000006ad, 0x00000100); + nv_icmd(dev, 0x000006ae, 0x00000100); + nv_icmd(dev, 0x000006b1, 0x00000011); + nv_icmd(dev, 0x000006bb, 0x000000cf); + nv_icmd(dev, 0x000006ce, 0x2a712488); + nv_icmd(dev, 0x00000739, 0x4085c000); + nv_icmd(dev, 0x0000073a, 0x00000080); + nv_icmd(dev, 0x00000786, 0x80000100); + nv_icmd(dev, 0x0000073c, 0x00010100); + nv_icmd(dev, 0x0000073d, 0x02800000); + nv_icmd(dev, 0x00000787, 0x000000cf); + nv_icmd(dev, 0x0000078c, 0x00000008); + nv_icmd(dev, 0x00000792, 0x00000001); + nv_icmd(dev, 0x00000794, 0x00000001); + nv_icmd(dev, 0x00000795, 0x00000001); + nv_icmd(dev, 0x00000796, 0x00000001); + nv_icmd(dev, 0x00000797, 0x000000cf); + nv_icmd(dev, 0x00000836, 0x00000001); + nv_icmd(dev, 0x0000079a, 0x00000002); + nv_icmd(dev, 0x00000833, 0x04444480); + nv_icmd(dev, 0x000007a1, 0x00000001); + nv_icmd(dev, 0x000007a3, 0x00000001); + nv_icmd(dev, 0x000007a4, 0x00000001); + nv_icmd(dev, 0x000007a5, 0x00000001); + nv_icmd(dev, 0x00000831, 0x00000004); + nv_icmd(dev, 0x0000080c, 0x00000002); + nv_icmd(dev, 0x0000080d, 0x00000100); + nv_icmd(dev, 0x0000080e, 0x00000100); + nv_icmd(dev, 0x0000080f, 0x00000001); + nv_icmd(dev, 0x00000823, 0x00000002); + nv_icmd(dev, 0x00000824, 0x00000100); + nv_icmd(dev, 0x00000825, 0x00000100); + nv_icmd(dev, 0x00000826, 0x00000001); + nv_icmd(dev, 0x0000095d, 0x00000001); + nv_icmd(dev, 0x0000082b, 0x00000004); + nv_icmd(dev, 0x00000942, 0x00010001); + nv_icmd(dev, 0x00000943, 0x00000001); + nv_icmd(dev, 0x00000944, 0x00000022); + nv_icmd(dev, 0x000007c5, 0x00010001); + nv_icmd(dev, 0x00000834, 0x00000001); + nv_icmd(dev, 0x000007c7, 0x00000001); + nv_icmd(dev, 0x0000c1b0, 0x0000000f); + nv_icmd(dev, 0x0000c1b1, 0x0000000f); + nv_icmd(dev, 0x0000c1b2, 0x0000000f); + nv_icmd(dev, 0x0000c1b3, 0x0000000f); + nv_icmd(dev, 0x0000c1b4, 0x0000000f); + nv_icmd(dev, 0x0000c1b5, 0x0000000f); + nv_icmd(dev, 0x0000c1b6, 0x0000000f); + nv_icmd(dev, 0x0000c1b7, 0x0000000f); + nv_icmd(dev, 0x0000c1b8, 0x0fac6881); + nv_icmd(dev, 0x0000c1b9, 0x00fac688); + nv_icmd(dev, 0x0001e100, 0x00000001); + nv_icmd(dev, 0x00001000, 0x00000002); + nv_icmd(dev, 0x000006aa, 0x00000001); + nv_icmd(dev, 0x000006ad, 0x00000100); + nv_icmd(dev, 0x000006ae, 0x00000100); + nv_icmd(dev, 0x000006b1, 0x00000011); + nv_icmd(dev, 0x0000078c, 0x00000008); + nv_icmd(dev, 0x00000792, 0x00000001); + nv_icmd(dev, 0x00000794, 0x00000001); + nv_icmd(dev, 0x00000795, 0x00000001); + nv_icmd(dev, 0x00000796, 0x00000001); + nv_icmd(dev, 0x00000797, 0x000000cf); + nv_icmd(dev, 0x0000079a, 0x00000002); + nv_icmd(dev, 0x00000833, 0x04444480); + nv_icmd(dev, 0x000007a1, 0x00000001); + nv_icmd(dev, 0x000007a3, 0x00000001); + nv_icmd(dev, 0x000007a4, 0x00000001); + nv_icmd(dev, 0x000007a5, 0x00000001); + nv_icmd(dev, 0x00000831, 0x00000004); + nv_icmd(dev, 0x0001e100, 0x00000001); + nv_icmd(dev, 0x00001000, 0x00000014); + nv_icmd(dev, 0x00000351, 0x00000100); + nv_icmd(dev, 0x00000957, 0x00000003); + nv_icmd(dev, 0x0000095d, 0x00000001); + nv_icmd(dev, 0x0000082b, 0x00000004); + nv_icmd(dev, 0x00000942, 0x00010001); + nv_icmd(dev, 0x00000943, 0x00000001); + nv_icmd(dev, 0x000007c5, 0x00010001); + nv_icmd(dev, 0x00000834, 0x00000001); + nv_icmd(dev, 0x000007c7, 0x00000001); + nv_icmd(dev, 0x0001e100, 0x00000001); + nv_icmd(dev, 0x00001000, 0x00000001); + nv_icmd(dev, 0x0000080c, 0x00000002); + nv_icmd(dev, 0x0000080d, 0x00000100); + nv_icmd(dev, 0x0000080e, 0x00000100); + nv_icmd(dev, 0x0000080f, 0x00000001); + nv_icmd(dev, 0x00000823, 0x00000002); + nv_icmd(dev, 0x00000824, 0x00000100); + nv_icmd(dev, 0x00000825, 0x00000100); + nv_icmd(dev, 0x00000826, 0x00000001); + nv_icmd(dev, 0x0001e100, 0x00000001); + nv_wr32(dev, 0x400208, 0x00000000); + nv_wr32(dev, 0x404154, 0x00000400); + + nvc0_grctx_generate_9097(dev); + nvc0_grctx_generate_902d(dev); + nvc0_grctx_generate_9039(dev); + nvc0_grctx_generate_90c0(dev); + + nv_wr32(dev, 0x000260, r000260); + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nvc0_instmem.c b/drivers/gpu/drm/nouveau/nvc0_instmem.c new file mode 100644 index 00000000..82357d2d --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvc0_instmem.c @@ -0,0 +1,231 @@ +/* + * Copyright 2010 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include "drmP.h" + +#include "nouveau_drv.h" +#include "nouveau_vm.h" + +struct nvc0_instmem_priv { + struct nouveau_gpuobj *bar1_pgd; + struct nouveau_channel *bar1; + struct nouveau_gpuobj *bar3_pgd; + struct nouveau_channel *bar3; + struct nouveau_gpuobj *chan_pgd; +}; + +int +nvc0_instmem_suspend(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + dev_priv->ramin_available = false; + return 0; +} + +void +nvc0_instmem_resume(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nvc0_instmem_priv *priv = dev_priv->engine.instmem.priv; + + nv_mask(dev, 0x100c80, 0x00000001, 0x00000000); + nv_wr32(dev, 0x001704, 0x80000000 | priv->bar1->ramin->vinst >> 12); + nv_wr32(dev, 0x001714, 0xc0000000 | priv->bar3->ramin->vinst >> 12); + dev_priv->ramin_available = true; +} + +static void +nvc0_channel_del(struct nouveau_channel **pchan) +{ + struct nouveau_channel *chan; + + chan = *pchan; + *pchan = NULL; + if (!chan) + return; + + nouveau_vm_ref(NULL, &chan->vm, NULL); + if (drm_mm_initialized(&chan->ramin_heap)) + drm_mm_takedown(&chan->ramin_heap); + nouveau_gpuobj_ref(NULL, &chan->ramin); + kfree(chan); +} + +static int +nvc0_channel_new(struct drm_device *dev, u32 size, struct nouveau_vm *vm, + struct nouveau_channel **pchan, + struct nouveau_gpuobj *pgd, u64 vm_size) +{ + struct nouveau_channel *chan; + int ret; + + chan = kzalloc(sizeof(*chan), GFP_KERNEL); + if (!chan) + return -ENOMEM; + chan->dev = dev; + + ret = nouveau_gpuobj_new(dev, NULL, size, 0x1000, 0, &chan->ramin); + if (ret) { + nvc0_channel_del(&chan); + return ret; + } + + ret = drm_mm_init(&chan->ramin_heap, 0x1000, size - 0x1000); + if (ret) { + nvc0_channel_del(&chan); + return ret; + } + + ret = nouveau_vm_ref(vm, &chan->vm, NULL); + if (ret) { + nvc0_channel_del(&chan); + return ret; + } + + nv_wo32(chan->ramin, 0x0200, lower_32_bits(pgd->vinst)); + nv_wo32(chan->ramin, 0x0204, upper_32_bits(pgd->vinst)); + nv_wo32(chan->ramin, 0x0208, lower_32_bits(vm_size - 1)); + nv_wo32(chan->ramin, 0x020c, upper_32_bits(vm_size - 1)); + + *pchan = chan; + return 0; +} + +int +nvc0_instmem_init(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_instmem_engine *pinstmem = &dev_priv->engine.instmem; + struct pci_dev *pdev = dev->pdev; + struct nvc0_instmem_priv *priv; + struct nouveau_vm *vm = NULL; + int ret; + + priv = kzalloc(sizeof(*priv), GFP_KERNEL); + if (!priv) + return -ENOMEM; + pinstmem->priv = priv; + + /* BAR3 VM */ + ret = nouveau_vm_new(dev, 0, pci_resource_len(pdev, 3), 0, + &dev_priv->bar3_vm); + if (ret) + goto error; + + ret = nouveau_gpuobj_new(dev, NULL, + (pci_resource_len(pdev, 3) >> 12) * 8, 0, + NVOBJ_FLAG_DONT_MAP | + NVOBJ_FLAG_ZERO_ALLOC, + &dev_priv->bar3_vm->pgt[0].obj[0]); + if (ret) + goto error; + dev_priv->bar3_vm->pgt[0].refcount[0] = 1; + + nv50_instmem_map(dev_priv->bar3_vm->pgt[0].obj[0]); + + ret = nouveau_gpuobj_new(dev, NULL, 0x8000, 4096, + NVOBJ_FLAG_ZERO_ALLOC, &priv->bar3_pgd); + if (ret) + goto error; + + ret = nouveau_vm_ref(dev_priv->bar3_vm, &vm, priv->bar3_pgd); + if (ret) + goto error; + nouveau_vm_ref(NULL, &vm, NULL); + + ret = nvc0_channel_new(dev, 8192, dev_priv->bar3_vm, &priv->bar3, + priv->bar3_pgd, pci_resource_len(dev->pdev, 3)); + if (ret) + goto error; + + /* BAR1 VM */ + ret = nouveau_vm_new(dev, 0, pci_resource_len(pdev, 1), 0, &vm); + if (ret) + goto error; + + ret = nouveau_gpuobj_new(dev, NULL, 0x8000, 4096, + NVOBJ_FLAG_ZERO_ALLOC, &priv->bar1_pgd); + if (ret) + goto error; + + ret = nouveau_vm_ref(vm, &dev_priv->bar1_vm, priv->bar1_pgd); + if (ret) + goto error; + nouveau_vm_ref(NULL, &vm, NULL); + + ret = nvc0_channel_new(dev, 8192, dev_priv->bar1_vm, &priv->bar1, + priv->bar1_pgd, pci_resource_len(dev->pdev, 1)); + if (ret) + goto error; + + /* channel vm */ + ret = nouveau_vm_new(dev, 0, (1ULL << 40), 0x0008000000ULL, &vm); + if (ret) + goto error; + + ret = nouveau_gpuobj_new(dev, NULL, 0x8000, 4096, 0, &priv->chan_pgd); + if (ret) + goto error; + + nouveau_vm_ref(vm, &dev_priv->chan_vm, priv->chan_pgd); + nouveau_vm_ref(NULL, &vm, NULL); + + nvc0_instmem_resume(dev); + return 0; +error: + nvc0_instmem_takedown(dev); + return ret; +} + +void +nvc0_instmem_takedown(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nvc0_instmem_priv *priv = dev_priv->engine.instmem.priv; + struct nouveau_vm *vm = NULL; + + nvc0_instmem_suspend(dev); + + nv_wr32(dev, 0x1704, 0x00000000); + nv_wr32(dev, 0x1714, 0x00000000); + + nouveau_vm_ref(NULL, &dev_priv->chan_vm, priv->chan_pgd); + nouveau_gpuobj_ref(NULL, &priv->chan_pgd); + + nvc0_channel_del(&priv->bar1); + nouveau_vm_ref(NULL, &dev_priv->bar1_vm, priv->bar1_pgd); + nouveau_gpuobj_ref(NULL, &priv->bar1_pgd); + + nvc0_channel_del(&priv->bar3); + nouveau_vm_ref(dev_priv->bar3_vm, &vm, NULL); + nouveau_vm_ref(NULL, &vm, priv->bar3_pgd); + nouveau_gpuobj_ref(NULL, &priv->bar3_pgd); + nouveau_gpuobj_ref(NULL, &dev_priv->bar3_vm->pgt[0].obj[0]); + nouveau_vm_ref(NULL, &dev_priv->bar3_vm, NULL); + + dev_priv->engine.instmem.priv = NULL; + kfree(priv); +} + diff --git a/drivers/gpu/drm/nouveau/nvc0_vm.c b/drivers/gpu/drm/nouveau/nvc0_vm.c new file mode 100644 index 00000000..a179e6c5 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvc0_vm.c @@ -0,0 +1,130 @@ +/* + * Copyright 2010 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include "drmP.h" + +#include "nouveau_drv.h" +#include "nouveau_vm.h" + +void +nvc0_vm_map_pgt(struct nouveau_gpuobj *pgd, u32 index, + struct nouveau_gpuobj *pgt[2]) +{ + u32 pde[2] = { 0, 0 }; + + if (pgt[0]) + pde[1] = 0x00000001 | (pgt[0]->vinst >> 8); + if (pgt[1]) + pde[0] = 0x00000001 | (pgt[1]->vinst >> 8); + + nv_wo32(pgd, (index * 8) + 0, pde[0]); + nv_wo32(pgd, (index * 8) + 4, pde[1]); +} + +static inline u64 +nvc0_vm_addr(struct nouveau_vma *vma, u64 phys, u32 memtype, u32 target) +{ + phys >>= 8; + + phys |= 0x00000001; /* present */ + if (vma->access & NV_MEM_ACCESS_SYS) + phys |= 0x00000002; + + phys |= ((u64)target << 32); + phys |= ((u64)memtype << 36); + + return phys; +} + +void +nvc0_vm_map(struct nouveau_vma *vma, struct nouveau_gpuobj *pgt, + struct nouveau_mem *mem, u32 pte, u32 cnt, u64 phys, u64 delta) +{ + u32 next = 1 << (vma->node->type - 8); + + phys = nvc0_vm_addr(vma, phys, mem->memtype, 0); + pte <<= 3; + while (cnt--) { + nv_wo32(pgt, pte + 0, lower_32_bits(phys)); + nv_wo32(pgt, pte + 4, upper_32_bits(phys)); + phys += next; + pte += 8; + } +} + +void +nvc0_vm_map_sg(struct nouveau_vma *vma, struct nouveau_gpuobj *pgt, + struct nouveau_mem *mem, u32 pte, u32 cnt, dma_addr_t *list) +{ + pte <<= 3; + while (cnt--) { + u64 phys = nvc0_vm_addr(vma, *list++, mem->memtype, 5); + nv_wo32(pgt, pte + 0, lower_32_bits(phys)); + nv_wo32(pgt, pte + 4, upper_32_bits(phys)); + pte += 8; + } +} + +void +nvc0_vm_unmap(struct nouveau_gpuobj *pgt, u32 pte, u32 cnt) +{ + pte <<= 3; + while (cnt--) { + nv_wo32(pgt, pte + 0, 0x00000000); + nv_wo32(pgt, pte + 4, 0x00000000); + pte += 8; + } +} + +void +nvc0_vm_flush(struct nouveau_vm *vm) +{ + struct drm_nouveau_private *dev_priv = vm->dev->dev_private; + struct nouveau_instmem_engine *pinstmem = &dev_priv->engine.instmem; + struct drm_device *dev = vm->dev; + struct nouveau_vm_pgd *vpgd; + unsigned long flags; + u32 engine = (dev_priv->chan_vm == vm) ? 1 : 5; + + pinstmem->flush(vm->dev); + + spin_lock_irqsave(&dev_priv->vm_lock, flags); + list_for_each_entry(vpgd, &vm->pgd_list, head) { + /* looks like maybe a "free flush slots" counter, the + * faster you write to 0x100cbc to more it decreases + */ + if (!nv_wait_ne(dev, 0x100c80, 0x00ff0000, 0x00000000)) { + NV_ERROR(dev, "vm timeout 0: 0x%08x %d\n", + nv_rd32(dev, 0x100c80), engine); + } + nv_wr32(dev, 0x100cb8, vpgd->obj->vinst >> 8); + nv_wr32(dev, 0x100cbc, 0x80000000 | engine); + /* wait for flush to be queued? */ + if (!nv_wait(dev, 0x100c80, 0x00008000, 0x00008000)) { + NV_ERROR(dev, "vm timeout 1: 0x%08x %d\n", + nv_rd32(dev, 0x100c80), engine); + } + } + spin_unlock_irqrestore(&dev_priv->vm_lock, flags); +} diff --git a/drivers/gpu/drm/nouveau/nvc0_vram.c b/drivers/gpu/drm/nouveau/nvc0_vram.c new file mode 100644 index 00000000..67c6ec6f --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvc0_vram.c @@ -0,0 +1,113 @@ +/* + * Copyright 2010 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include "drmP.h" +#include "nouveau_drv.h" +#include "nouveau_mm.h" + +/* 0 = unsupported + * 1 = non-compressed + * 3 = compressed + */ +static const u8 types[256] = { + 1, 1, 3, 3, 3, 3, 0, 3, 3, 3, 3, 0, 0, 0, 0, 0, + 0, 1, 0, 0, 0, 0, 0, 3, 3, 3, 3, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 3, 3, 3, 3, 3, 3, + 3, 3, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 3, 3, 3, 3, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 3, 3, 3, 3, 0, 1, 1, 1, 1, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 3, 3, 3, 3, 1, 1, 1, 1, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 3, + 3, 3, 3, 3, 0, 0, 0, 0, 0, 0, 0, 0, 3, 3, 3, 3, + 3, 3, 0, 0, 0, 0, 0, 0, 3, 0, 0, 3, 0, 3, 0, 3, + 3, 0, 3, 3, 3, 3, 3, 0, 0, 3, 0, 3, 0, 3, 3, 0, + 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 0, 1, 1, 0 +}; + +bool +nvc0_vram_flags_valid(struct drm_device *dev, u32 tile_flags) +{ + u8 memtype = (tile_flags & NOUVEAU_GEM_TILE_LAYOUT_MASK) >> 8; + return likely((types[memtype] == 1)); +} + +int +nvc0_vram_new(struct drm_device *dev, u64 size, u32 align, u32 ncmin, + u32 type, struct nouveau_mem **pmem) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct ttm_bo_device *bdev = &dev_priv->ttm.bdev; + struct ttm_mem_type_manager *man = &bdev->man[TTM_PL_VRAM]; + struct nouveau_mm *mm = man->priv; + struct nouveau_mm_node *r; + struct nouveau_mem *mem; + int ret; + + size >>= 12; + align >>= 12; + ncmin >>= 12; + + mem = kzalloc(sizeof(*mem), GFP_KERNEL); + if (!mem) + return -ENOMEM; + + INIT_LIST_HEAD(&mem->regions); + mem->dev = dev_priv->dev; + mem->memtype = (type & 0xff); + mem->size = size; + + mutex_lock(&mm->mutex); + do { + ret = nouveau_mm_get(mm, 1, size, ncmin, align, &r); + if (ret) { + mutex_unlock(&mm->mutex); + nv50_vram_del(dev, &mem); + return ret; + } + + list_add_tail(&r->rl_entry, &mem->regions); + size -= r->length; + } while (size); + mutex_unlock(&mm->mutex); + + r = list_first_entry(&mem->regions, struct nouveau_mm_node, rl_entry); + mem->offset = (u64)r->offset << 12; + *pmem = mem; + return 0; +} + +int +nvc0_vram_init(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + dev_priv->vram_size = nv_rd32(dev, 0x10f20c) << 20; + dev_priv->vram_size *= nv_rd32(dev, 0x121c74); + dev_priv->vram_rblock_size = 4096; + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nvreg.h b/drivers/gpu/drm/nouveau/nvreg.h new file mode 100644 index 00000000..bbfb1a68 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvreg.h @@ -0,0 +1,517 @@ +/* $XConsortium: nvreg.h /main/2 1996/10/28 05:13:41 kaleb $ */ +/* + * Copyright 1996-1997 David J. McKay + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * DAVID J. MCKAY BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF + * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +/* $XFree86: xc/programs/Xserver/hw/xfree86/drivers/nv/nvreg.h,v 1.6 2002/01/25 21:56:06 tsi Exp $ */ + +#ifndef __NVREG_H_ +#define __NVREG_H_ + +#define NV_PMC_OFFSET 0x00000000 +#define NV_PMC_SIZE 0x00001000 + +#define NV_PBUS_OFFSET 0x00001000 +#define NV_PBUS_SIZE 0x00001000 + +#define NV_PFIFO_OFFSET 0x00002000 +#define NV_PFIFO_SIZE 0x00002000 + +#define NV_HDIAG_OFFSET 0x00005000 +#define NV_HDIAG_SIZE 0x00001000 + +#define NV_PRAM_OFFSET 0x00006000 +#define NV_PRAM_SIZE 0x00001000 + +#define NV_PVIDEO_OFFSET 0x00008000 +#define NV_PVIDEO_SIZE 0x00001000 + +#define NV_PTIMER_OFFSET 0x00009000 +#define NV_PTIMER_SIZE 0x00001000 + +#define NV_PPM_OFFSET 0x0000A000 +#define NV_PPM_SIZE 0x00001000 + +#define NV_PTV_OFFSET 0x0000D000 +#define NV_PTV_SIZE 0x00001000 + +#define NV_PRMVGA_OFFSET 0x000A0000 +#define NV_PRMVGA_SIZE 0x00020000 + +#define NV_PRMVIO0_OFFSET 0x000C0000 +#define NV_PRMVIO_SIZE 0x00002000 +#define NV_PRMVIO1_OFFSET 0x000C2000 + +#define NV_PFB_OFFSET 0x00100000 +#define NV_PFB_SIZE 0x00001000 + +#define NV_PEXTDEV_OFFSET 0x00101000 +#define NV_PEXTDEV_SIZE 0x00001000 + +#define NV_PME_OFFSET 0x00200000 +#define NV_PME_SIZE 0x00001000 + +#define NV_PROM_OFFSET 0x00300000 +#define NV_PROM_SIZE 0x00010000 + +#define NV_PGRAPH_OFFSET 0x00400000 +#define NV_PGRAPH_SIZE 0x00010000 + +#define NV_PCRTC0_OFFSET 0x00600000 +#define NV_PCRTC0_SIZE 0x00002000 /* empirical */ + +#define NV_PRMCIO0_OFFSET 0x00601000 +#define NV_PRMCIO_SIZE 0x00002000 +#define NV_PRMCIO1_OFFSET 0x00603000 + +#define NV50_DISPLAY_OFFSET 0x00610000 +#define NV50_DISPLAY_SIZE 0x0000FFFF + +#define NV_PRAMDAC0_OFFSET 0x00680000 +#define NV_PRAMDAC0_SIZE 0x00002000 + +#define NV_PRMDIO0_OFFSET 0x00681000 +#define NV_PRMDIO_SIZE 0x00002000 +#define NV_PRMDIO1_OFFSET 0x00683000 + +#define NV_PRAMIN_OFFSET 0x00700000 +#define NV_PRAMIN_SIZE 0x00100000 + +#define NV_FIFO_OFFSET 0x00800000 +#define NV_FIFO_SIZE 0x00800000 + +#define NV_PMC_BOOT_0 0x00000000 +#define NV_PMC_ENABLE 0x00000200 + +#define NV_VIO_VSE2 0x000003c3 +#define NV_VIO_SRX 0x000003c4 + +#define NV_CIO_CRX__COLOR 0x000003d4 +#define NV_CIO_CR__COLOR 0x000003d5 + +#define NV_PBUS_DEBUG_1 0x00001084 +#define NV_PBUS_DEBUG_4 0x00001098 +#define NV_PBUS_DEBUG_DUALHEAD_CTL 0x000010f0 +#define NV_PBUS_POWERCTRL_1 0x00001584 +#define NV_PBUS_POWERCTRL_2 0x00001588 +#define NV_PBUS_POWERCTRL_4 0x00001590 +#define NV_PBUS_PCI_NV_19 0x0000184C +#define NV_PBUS_PCI_NV_20 0x00001850 +# define NV_PBUS_PCI_NV_20_ROM_SHADOW_DISABLED (0 << 0) +# define NV_PBUS_PCI_NV_20_ROM_SHADOW_ENABLED (1 << 0) + +#define NV_PFIFO_RAMHT 0x00002210 + +#define NV_PTV_TV_INDEX 0x0000d220 +#define NV_PTV_TV_DATA 0x0000d224 +#define NV_PTV_HFILTER 0x0000d310 +#define NV_PTV_HFILTER2 0x0000d390 +#define NV_PTV_VFILTER 0x0000d510 + +#define NV_PRMVIO_MISC__WRITE 0x000c03c2 +#define NV_PRMVIO_SRX 0x000c03c4 +#define NV_PRMVIO_SR 0x000c03c5 +# define NV_VIO_SR_RESET_INDEX 0x00 +# define NV_VIO_SR_CLOCK_INDEX 0x01 +# define NV_VIO_SR_PLANE_MASK_INDEX 0x02 +# define NV_VIO_SR_CHAR_MAP_INDEX 0x03 +# define NV_VIO_SR_MEM_MODE_INDEX 0x04 +#define NV_PRMVIO_MISC__READ 0x000c03cc +#define NV_PRMVIO_GRX 0x000c03ce +#define NV_PRMVIO_GX 0x000c03cf +# define NV_VIO_GX_SR_INDEX 0x00 +# define NV_VIO_GX_SREN_INDEX 0x01 +# define NV_VIO_GX_CCOMP_INDEX 0x02 +# define NV_VIO_GX_ROP_INDEX 0x03 +# define NV_VIO_GX_READ_MAP_INDEX 0x04 +# define NV_VIO_GX_MODE_INDEX 0x05 +# define NV_VIO_GX_MISC_INDEX 0x06 +# define NV_VIO_GX_DONT_CARE_INDEX 0x07 +# define NV_VIO_GX_BIT_MASK_INDEX 0x08 + +#define NV_PCRTC_INTR_0 0x00600100 +# define NV_PCRTC_INTR_0_VBLANK (1 << 0) +#define NV_PCRTC_INTR_EN_0 0x00600140 +#define NV_PCRTC_START 0x00600800 +#define NV_PCRTC_CONFIG 0x00600804 +# define NV_PCRTC_CONFIG_START_ADDRESS_NON_VGA (1 << 0) +# define NV04_PCRTC_CONFIG_START_ADDRESS_HSYNC (4 << 0) +# define NV10_PCRTC_CONFIG_START_ADDRESS_HSYNC (2 << 0) +#define NV_PCRTC_CURSOR_CONFIG 0x00600810 +# define NV_PCRTC_CURSOR_CONFIG_ENABLE_ENABLE (1 << 0) +# define NV_PCRTC_CURSOR_CONFIG_DOUBLE_SCAN_ENABLE (1 << 4) +# define NV_PCRTC_CURSOR_CONFIG_ADDRESS_SPACE_PNVM (1 << 8) +# define NV_PCRTC_CURSOR_CONFIG_CUR_BPP_32 (1 << 12) +# define NV_PCRTC_CURSOR_CONFIG_CUR_PIXELS_64 (1 << 16) +# define NV_PCRTC_CURSOR_CONFIG_CUR_LINES_32 (2 << 24) +# define NV_PCRTC_CURSOR_CONFIG_CUR_LINES_64 (4 << 24) +# define NV_PCRTC_CURSOR_CONFIG_CUR_BLEND_ALPHA (1 << 28) + +/* note: PCRTC_GPIO is not available on nv10, and in fact aliases 0x600810 */ +#define NV_PCRTC_GPIO 0x00600818 +#define NV_PCRTC_GPIO_EXT 0x0060081c +#define NV_PCRTC_830 0x00600830 +#define NV_PCRTC_834 0x00600834 +#define NV_PCRTC_850 0x00600850 +#define NV_PCRTC_ENGINE_CTRL 0x00600860 +# define NV_CRTC_FSEL_I2C (1 << 4) +# define NV_CRTC_FSEL_OVERLAY (1 << 12) + +#define NV_PRMCIO_ARX 0x006013c0 +#define NV_PRMCIO_AR__WRITE 0x006013c0 +#define NV_PRMCIO_AR__READ 0x006013c1 +# define NV_CIO_AR_MODE_INDEX 0x10 +# define NV_CIO_AR_OSCAN_INDEX 0x11 +# define NV_CIO_AR_PLANE_INDEX 0x12 +# define NV_CIO_AR_HPP_INDEX 0x13 +# define NV_CIO_AR_CSEL_INDEX 0x14 +#define NV_PRMCIO_INP0 0x006013c2 +#define NV_PRMCIO_CRX__COLOR 0x006013d4 +#define NV_PRMCIO_CR__COLOR 0x006013d5 + /* Standard VGA CRTC registers */ +# define NV_CIO_CR_HDT_INDEX 0x00 /* horizontal display total */ +# define NV_CIO_CR_HDE_INDEX 0x01 /* horizontal display end */ +# define NV_CIO_CR_HBS_INDEX 0x02 /* horizontal blanking start */ +# define NV_CIO_CR_HBE_INDEX 0x03 /* horizontal blanking end */ +# define NV_CIO_CR_HBE_4_0 4:0 +# define NV_CIO_CR_HRS_INDEX 0x04 /* horizontal retrace start */ +# define NV_CIO_CR_HRE_INDEX 0x05 /* horizontal retrace end */ +# define NV_CIO_CR_HRE_4_0 4:0 +# define NV_CIO_CR_HRE_HBE_5 7:7 +# define NV_CIO_CR_VDT_INDEX 0x06 /* vertical display total */ +# define NV_CIO_CR_OVL_INDEX 0x07 /* overflow bits */ +# define NV_CIO_CR_OVL_VDT_8 0:0 +# define NV_CIO_CR_OVL_VDE_8 1:1 +# define NV_CIO_CR_OVL_VRS_8 2:2 +# define NV_CIO_CR_OVL_VBS_8 3:3 +# define NV_CIO_CR_OVL_VDT_9 5:5 +# define NV_CIO_CR_OVL_VDE_9 6:6 +# define NV_CIO_CR_OVL_VRS_9 7:7 +# define NV_CIO_CR_RSAL_INDEX 0x08 /* normally "preset row scan" */ +# define NV_CIO_CR_CELL_HT_INDEX 0x09 /* cell height?! normally "max scan line" */ +# define NV_CIO_CR_CELL_HT_VBS_9 5:5 +# define NV_CIO_CR_CELL_HT_SCANDBL 7:7 +# define NV_CIO_CR_CURS_ST_INDEX 0x0a /* cursor start */ +# define NV_CIO_CR_CURS_END_INDEX 0x0b /* cursor end */ +# define NV_CIO_CR_SA_HI_INDEX 0x0c /* screen start address high */ +# define NV_CIO_CR_SA_LO_INDEX 0x0d /* screen start address low */ +# define NV_CIO_CR_TCOFF_HI_INDEX 0x0e /* cursor offset high */ +# define NV_CIO_CR_TCOFF_LO_INDEX 0x0f /* cursor offset low */ +# define NV_CIO_CR_VRS_INDEX 0x10 /* vertical retrace start */ +# define NV_CIO_CR_VRE_INDEX 0x11 /* vertical retrace end */ +# define NV_CIO_CR_VRE_3_0 3:0 +# define NV_CIO_CR_VDE_INDEX 0x12 /* vertical display end */ +# define NV_CIO_CR_OFFSET_INDEX 0x13 /* sets screen pitch */ +# define NV_CIO_CR_ULINE_INDEX 0x14 /* underline location */ +# define NV_CIO_CR_VBS_INDEX 0x15 /* vertical blank start */ +# define NV_CIO_CR_VBE_INDEX 0x16 /* vertical blank end */ +# define NV_CIO_CR_MODE_INDEX 0x17 /* crtc mode control */ +# define NV_CIO_CR_LCOMP_INDEX 0x18 /* line compare */ + /* Extended VGA CRTC registers */ +# define NV_CIO_CRE_RPC0_INDEX 0x19 /* repaint control 0 */ +# define NV_CIO_CRE_RPC0_OFFSET_10_8 7:5 +# define NV_CIO_CRE_RPC1_INDEX 0x1a /* repaint control 1 */ +# define NV_CIO_CRE_RPC1_LARGE 2:2 +# define NV_CIO_CRE_FF_INDEX 0x1b /* fifo control */ +# define NV_CIO_CRE_ENH_INDEX 0x1c /* enhanced? */ +# define NV_CIO_SR_LOCK_INDEX 0x1f /* crtc lock */ +# define NV_CIO_SR_UNLOCK_RW_VALUE 0x57 +# define NV_CIO_SR_LOCK_VALUE 0x99 +# define NV_CIO_CRE_FFLWM__INDEX 0x20 /* fifo low water mark */ +# define NV_CIO_CRE_21 0x21 /* vga shadow crtc lock */ +# define NV_CIO_CRE_LSR_INDEX 0x25 /* ? */ +# define NV_CIO_CRE_LSR_VDT_10 0:0 +# define NV_CIO_CRE_LSR_VDE_10 1:1 +# define NV_CIO_CRE_LSR_VRS_10 2:2 +# define NV_CIO_CRE_LSR_VBS_10 3:3 +# define NV_CIO_CRE_LSR_HBE_6 4:4 +# define NV_CIO_CR_ARX_INDEX 0x26 /* attribute index -- ro copy of 0x60.3c0 */ +# define NV_CIO_CRE_CHIP_ID_INDEX 0x27 /* chip revision */ +# define NV_CIO_CRE_PIXEL_INDEX 0x28 +# define NV_CIO_CRE_PIXEL_FORMAT 1:0 +# define NV_CIO_CRE_HEB__INDEX 0x2d /* horizontal extra bits? */ +# define NV_CIO_CRE_HEB_HDT_8 0:0 +# define NV_CIO_CRE_HEB_HDE_8 1:1 +# define NV_CIO_CRE_HEB_HBS_8 2:2 +# define NV_CIO_CRE_HEB_HRS_8 3:3 +# define NV_CIO_CRE_HEB_ILC_8 4:4 +# define NV_CIO_CRE_2E 0x2e /* some scratch or dummy reg to force writes to sink in */ +# define NV_CIO_CRE_HCUR_ADDR2_INDEX 0x2f /* cursor */ +# define NV_CIO_CRE_HCUR_ADDR0_INDEX 0x30 /* pixmap */ +# define NV_CIO_CRE_HCUR_ADDR0_ADR 6:0 +# define NV_CIO_CRE_HCUR_ASI 7:7 +# define NV_CIO_CRE_HCUR_ADDR1_INDEX 0x31 /* address */ +# define NV_CIO_CRE_HCUR_ADDR1_ENABLE 0:0 +# define NV_CIO_CRE_HCUR_ADDR1_CUR_DBL 1:1 +# define NV_CIO_CRE_HCUR_ADDR1_ADR 7:2 +# define NV_CIO_CRE_LCD__INDEX 0x33 +# define NV_CIO_CRE_LCD_LCD_SELECT 0:0 +# define NV_CIO_CRE_LCD_ROUTE_MASK 0x3b +# define NV_CIO_CRE_DDC0_STATUS__INDEX 0x36 +# define NV_CIO_CRE_DDC0_WR__INDEX 0x37 +# define NV_CIO_CRE_ILACE__INDEX 0x39 /* interlace */ +# define NV_CIO_CRE_SCRATCH3__INDEX 0x3b +# define NV_CIO_CRE_SCRATCH4__INDEX 0x3c +# define NV_CIO_CRE_DDC_STATUS__INDEX 0x3e +# define NV_CIO_CRE_DDC_WR__INDEX 0x3f +# define NV_CIO_CRE_EBR_INDEX 0x41 /* extra bits ? (vertical) */ +# define NV_CIO_CRE_EBR_VDT_11 0:0 +# define NV_CIO_CRE_EBR_VDE_11 2:2 +# define NV_CIO_CRE_EBR_VRS_11 4:4 +# define NV_CIO_CRE_EBR_VBS_11 6:6 +# define NV_CIO_CRE_42 0x42 +# define NV_CIO_CRE_42_OFFSET_11 6:6 +# define NV_CIO_CRE_43 0x43 +# define NV_CIO_CRE_44 0x44 /* head control */ +# define NV_CIO_CRE_CSB 0x45 /* colour saturation boost */ +# define NV_CIO_CRE_RCR 0x46 +# define NV_CIO_CRE_RCR_ENDIAN_BIG 7:7 +# define NV_CIO_CRE_47 0x47 /* extended fifo lwm, used on nv30+ */ +# define NV_CIO_CRE_49 0x49 +# define NV_CIO_CRE_4B 0x4b /* given patterns in 0x[2-3][a-c] regs, probably scratch 6 */ +# define NV_CIO_CRE_TVOUT_LATENCY 0x52 +# define NV_CIO_CRE_53 0x53 /* `fp_htiming' according to Haiku */ +# define NV_CIO_CRE_54 0x54 /* `fp_vtiming' according to Haiku */ +# define NV_CIO_CRE_57 0x57 /* index reg for cr58 */ +# define NV_CIO_CRE_58 0x58 /* data reg for cr57 */ +# define NV_CIO_CRE_59 0x59 /* related to on/off-chip-ness of digital outputs */ +# define NV_CIO_CRE_5B 0x5B /* newer colour saturation reg */ +# define NV_CIO_CRE_85 0x85 +# define NV_CIO_CRE_86 0x86 +#define NV_PRMCIO_INP0__COLOR 0x006013da + +#define NV_PRAMDAC_CU_START_POS 0x00680300 +# define NV_PRAMDAC_CU_START_POS_X 15:0 +# define NV_PRAMDAC_CU_START_POS_Y 31:16 +#define NV_RAMDAC_NV10_CURSYNC 0x00680404 + +#define NV_PRAMDAC_NVPLL_COEFF 0x00680500 +#define NV_PRAMDAC_MPLL_COEFF 0x00680504 +#define NV_PRAMDAC_VPLL_COEFF 0x00680508 +# define NV30_RAMDAC_ENABLE_VCO2 (8 << 4) + +#define NV_PRAMDAC_PLL_COEFF_SELECT 0x0068050c +# define NV_PRAMDAC_PLL_COEFF_SELECT_USE_VPLL2_TRUE (4 << 0) +# define NV_PRAMDAC_PLL_COEFF_SELECT_SOURCE_PROG_MPLL (1 << 8) +# define NV_PRAMDAC_PLL_COEFF_SELECT_SOURCE_PROG_VPLL (2 << 8) +# define NV_PRAMDAC_PLL_COEFF_SELECT_SOURCE_PROG_NVPLL (4 << 8) +# define NV_PRAMDAC_PLL_COEFF_SELECT_PLL_SOURCE_VPLL2 (8 << 8) +# define NV_PRAMDAC_PLL_COEFF_SELECT_TV_VSCLK1 (1 << 16) +# define NV_PRAMDAC_PLL_COEFF_SELECT_TV_PCLK1 (2 << 16) +# define NV_PRAMDAC_PLL_COEFF_SELECT_TV_VSCLK2 (4 << 16) +# define NV_PRAMDAC_PLL_COEFF_SELECT_TV_PCLK2 (8 << 16) +# define NV_PRAMDAC_PLL_COEFF_SELECT_TV_CLK_SOURCE_VIP (1 << 20) +# define NV_PRAMDAC_PLL_COEFF_SELECT_VCLK_RATIO_DB2 (1 << 28) +# define NV_PRAMDAC_PLL_COEFF_SELECT_VCLK2_RATIO_DB2 (2 << 28) + +#define NV_PRAMDAC_PLL_SETUP_CONTROL 0x00680510 +#define NV_RAMDAC_VPLL2 0x00680520 +#define NV_PRAMDAC_SEL_CLK 0x00680524 +#define NV_RAMDAC_DITHER_NV11 0x00680528 +#define NV_PRAMDAC_DACCLK 0x0068052c +# define NV_PRAMDAC_DACCLK_SEL_DACCLK (1 << 0) + +#define NV_RAMDAC_NVPLL_B 0x00680570 +#define NV_RAMDAC_MPLL_B 0x00680574 +#define NV_RAMDAC_VPLL_B 0x00680578 +#define NV_RAMDAC_VPLL2_B 0x0068057c +# define NV31_RAMDAC_ENABLE_VCO2 (8 << 28) +#define NV_PRAMDAC_580 0x00680580 +# define NV_RAMDAC_580_VPLL1_ACTIVE (1 << 8) +# define NV_RAMDAC_580_VPLL2_ACTIVE (1 << 28) + +#define NV_PRAMDAC_GENERAL_CONTROL 0x00680600 +# define NV_PRAMDAC_GENERAL_CONTROL_PIXMIX_ON (3 << 4) +# define NV_PRAMDAC_GENERAL_CONTROL_VGA_STATE_SEL (1 << 8) +# define NV_PRAMDAC_GENERAL_CONTROL_ALT_MODE_SEL (1 << 12) +# define NV_PRAMDAC_GENERAL_CONTROL_TERMINATION_75OHM (2 << 16) +# define NV_PRAMDAC_GENERAL_CONTROL_BPC_8BITS (1 << 20) +# define NV_PRAMDAC_GENERAL_CONTROL_PIPE_LONG (2 << 28) +#define NV_PRAMDAC_TEST_CONTROL 0x00680608 +# define NV_PRAMDAC_TEST_CONTROL_TP_INS_EN_ASSERTED (1 << 12) +# define NV_PRAMDAC_TEST_CONTROL_PWRDWN_DAC_OFF (1 << 16) +# define NV_PRAMDAC_TEST_CONTROL_SENSEB_ALLHI (1 << 28) +#define NV_PRAMDAC_TESTPOINT_DATA 0x00680610 +# define NV_PRAMDAC_TESTPOINT_DATA_NOTBLANK (8 << 28) +#define NV_PRAMDAC_630 0x00680630 +#define NV_PRAMDAC_634 0x00680634 + +#define NV_PRAMDAC_TV_SETUP 0x00680700 +#define NV_PRAMDAC_TV_VTOTAL 0x00680720 +#define NV_PRAMDAC_TV_VSKEW 0x00680724 +#define NV_PRAMDAC_TV_VSYNC_DELAY 0x00680728 +#define NV_PRAMDAC_TV_HTOTAL 0x0068072c +#define NV_PRAMDAC_TV_HSKEW 0x00680730 +#define NV_PRAMDAC_TV_HSYNC_DELAY 0x00680734 +#define NV_PRAMDAC_TV_HSYNC_DELAY2 0x00680738 + +#define NV_PRAMDAC_TV_SETUP 0x00680700 + +#define NV_PRAMDAC_FP_VDISPLAY_END 0x00680800 +#define NV_PRAMDAC_FP_VTOTAL 0x00680804 +#define NV_PRAMDAC_FP_VCRTC 0x00680808 +#define NV_PRAMDAC_FP_VSYNC_START 0x0068080c +#define NV_PRAMDAC_FP_VSYNC_END 0x00680810 +#define NV_PRAMDAC_FP_VVALID_START 0x00680814 +#define NV_PRAMDAC_FP_VVALID_END 0x00680818 +#define NV_PRAMDAC_FP_HDISPLAY_END 0x00680820 +#define NV_PRAMDAC_FP_HTOTAL 0x00680824 +#define NV_PRAMDAC_FP_HCRTC 0x00680828 +#define NV_PRAMDAC_FP_HSYNC_START 0x0068082c +#define NV_PRAMDAC_FP_HSYNC_END 0x00680830 +#define NV_PRAMDAC_FP_HVALID_START 0x00680834 +#define NV_PRAMDAC_FP_HVALID_END 0x00680838 + +#define NV_RAMDAC_FP_DITHER 0x0068083c +#define NV_PRAMDAC_FP_TG_CONTROL 0x00680848 +# define NV_PRAMDAC_FP_TG_CONTROL_VSYNC_POS (1 << 0) +# define NV_PRAMDAC_FP_TG_CONTROL_VSYNC_DISABLE (2 << 0) +# define NV_PRAMDAC_FP_TG_CONTROL_HSYNC_POS (1 << 4) +# define NV_PRAMDAC_FP_TG_CONTROL_HSYNC_DISABLE (2 << 4) +# define NV_PRAMDAC_FP_TG_CONTROL_MODE_SCALE (0 << 8) +# define NV_PRAMDAC_FP_TG_CONTROL_MODE_CENTER (1 << 8) +# define NV_PRAMDAC_FP_TG_CONTROL_MODE_NATIVE (2 << 8) +# define NV_PRAMDAC_FP_TG_CONTROL_READ_PROG (1 << 20) +# define NV_PRAMDAC_FP_TG_CONTROL_WIDTH_12 (1 << 24) +# define NV_PRAMDAC_FP_TG_CONTROL_DISPEN_POS (1 << 28) +# define NV_PRAMDAC_FP_TG_CONTROL_DISPEN_DISABLE (2 << 28) +#define NV_PRAMDAC_FP_MARGIN_COLOR 0x0068084c +#define NV_PRAMDAC_850 0x00680850 +#define NV_PRAMDAC_85C 0x0068085c +#define NV_PRAMDAC_FP_DEBUG_0 0x00680880 +# define NV_PRAMDAC_FP_DEBUG_0_XSCALE_ENABLE (1 << 0) +# define NV_PRAMDAC_FP_DEBUG_0_YSCALE_ENABLE (1 << 4) +/* This doesn't seem to be essential for tmds, but still often set */ +# define NV_RAMDAC_FP_DEBUG_0_TMDS_ENABLED (8 << 4) +# define NV_PRAMDAC_FP_DEBUG_0_XINTERP_BILINEAR (1 << 8) +# define NV_PRAMDAC_FP_DEBUG_0_YINTERP_BILINEAR (1 << 12) +# define NV_PRAMDAC_FP_DEBUG_0_XWEIGHT_ROUND (1 << 20) +# define NV_PRAMDAC_FP_DEBUG_0_YWEIGHT_ROUND (1 << 24) +# define NV_PRAMDAC_FP_DEBUG_0_PWRDOWN_FPCLK (1 << 28) +#define NV_PRAMDAC_FP_DEBUG_1 0x00680884 +# define NV_PRAMDAC_FP_DEBUG_1_XSCALE_VALUE 11:0 +# define NV_PRAMDAC_FP_DEBUG_1_XSCALE_TESTMODE_ENABLE (1 << 12) +# define NV_PRAMDAC_FP_DEBUG_1_YSCALE_VALUE 27:16 +# define NV_PRAMDAC_FP_DEBUG_1_YSCALE_TESTMODE_ENABLE (1 << 28) +#define NV_PRAMDAC_FP_DEBUG_2 0x00680888 +#define NV_PRAMDAC_FP_DEBUG_3 0x0068088C + +/* see NV_PRAMDAC_INDIR_TMDS in rules.xml */ +#define NV_PRAMDAC_FP_TMDS_CONTROL 0x006808b0 +# define NV_PRAMDAC_FP_TMDS_CONTROL_WRITE_DISABLE (1 << 16) +#define NV_PRAMDAC_FP_TMDS_DATA 0x006808b4 + +#define NV_PRAMDAC_8C0 0x006808c0 + +/* Some kind of switch */ +#define NV_PRAMDAC_900 0x00680900 +#define NV_PRAMDAC_A20 0x00680A20 +#define NV_PRAMDAC_A24 0x00680A24 +#define NV_PRAMDAC_A34 0x00680A34 + +#define NV_PRAMDAC_CTV 0x00680c00 + +/* names fabricated from NV_USER_DAC info */ +#define NV_PRMDIO_PIXEL_MASK 0x006813c6 +# define NV_PRMDIO_PIXEL_MASK_MASK 0xff +#define NV_PRMDIO_READ_MODE_ADDRESS 0x006813c7 +#define NV_PRMDIO_WRITE_MODE_ADDRESS 0x006813c8 +#define NV_PRMDIO_PALETTE_DATA 0x006813c9 + +#define NV_PGRAPH_DEBUG_0 0x00400080 +#define NV_PGRAPH_DEBUG_1 0x00400084 +#define NV_PGRAPH_DEBUG_2_NV04 0x00400088 +#define NV_PGRAPH_DEBUG_2 0x00400620 +#define NV_PGRAPH_DEBUG_3 0x0040008c +#define NV_PGRAPH_DEBUG_4 0x00400090 +#define NV_PGRAPH_INTR 0x00400100 +#define NV_PGRAPH_INTR_EN 0x00400140 +#define NV_PGRAPH_CTX_CONTROL 0x00400144 +#define NV_PGRAPH_CTX_CONTROL_NV04 0x00400170 +#define NV_PGRAPH_ABS_UCLIP_XMIN 0x0040053C +#define NV_PGRAPH_ABS_UCLIP_YMIN 0x00400540 +#define NV_PGRAPH_ABS_UCLIP_XMAX 0x00400544 +#define NV_PGRAPH_ABS_UCLIP_YMAX 0x00400548 +#define NV_PGRAPH_BETA_AND 0x00400608 +#define NV_PGRAPH_LIMIT_VIOL_PIX 0x00400610 +#define NV_PGRAPH_BOFFSET0 0x00400640 +#define NV_PGRAPH_BOFFSET1 0x00400644 +#define NV_PGRAPH_BOFFSET2 0x00400648 +#define NV_PGRAPH_BLIMIT0 0x00400684 +#define NV_PGRAPH_BLIMIT1 0x00400688 +#define NV_PGRAPH_BLIMIT2 0x0040068c +#define NV_PGRAPH_STATUS 0x00400700 +#define NV_PGRAPH_SURFACE 0x00400710 +#define NV_PGRAPH_STATE 0x00400714 +#define NV_PGRAPH_FIFO 0x00400720 +#define NV_PGRAPH_PATTERN_SHAPE 0x00400810 +#define NV_PGRAPH_TILE 0x00400b00 + +#define NV_PVIDEO_INTR_EN 0x00008140 +#define NV_PVIDEO_BUFFER 0x00008700 +#define NV_PVIDEO_STOP 0x00008704 +#define NV_PVIDEO_UVPLANE_BASE(buff) (0x00008800+(buff)*4) +#define NV_PVIDEO_UVPLANE_LIMIT(buff) (0x00008808+(buff)*4) +#define NV_PVIDEO_UVPLANE_OFFSET_BUFF(buff) (0x00008820+(buff)*4) +#define NV_PVIDEO_BASE(buff) (0x00008900+(buff)*4) +#define NV_PVIDEO_LIMIT(buff) (0x00008908+(buff)*4) +#define NV_PVIDEO_LUMINANCE(buff) (0x00008910+(buff)*4) +#define NV_PVIDEO_CHROMINANCE(buff) (0x00008918+(buff)*4) +#define NV_PVIDEO_OFFSET_BUFF(buff) (0x00008920+(buff)*4) +#define NV_PVIDEO_SIZE_IN(buff) (0x00008928+(buff)*4) +#define NV_PVIDEO_POINT_IN(buff) (0x00008930+(buff)*4) +#define NV_PVIDEO_DS_DX(buff) (0x00008938+(buff)*4) +#define NV_PVIDEO_DT_DY(buff) (0x00008940+(buff)*4) +#define NV_PVIDEO_POINT_OUT(buff) (0x00008948+(buff)*4) +#define NV_PVIDEO_SIZE_OUT(buff) (0x00008950+(buff)*4) +#define NV_PVIDEO_FORMAT(buff) (0x00008958+(buff)*4) +# define NV_PVIDEO_FORMAT_PLANAR (1 << 0) +# define NV_PVIDEO_FORMAT_COLOR_LE_CR8YB8CB8YA8 (1 << 16) +# define NV_PVIDEO_FORMAT_DISPLAY_COLOR_KEY (1 << 20) +# define NV_PVIDEO_FORMAT_MATRIX_ITURBT709 (1 << 24) +#define NV_PVIDEO_COLOR_KEY 0x00008B00 + +/* NV04 overlay defines from VIDIX & Haiku */ +#define NV_PVIDEO_INTR_EN_0 0x00680140 +#define NV_PVIDEO_STEP_SIZE 0x00680200 +#define NV_PVIDEO_CONTROL_Y 0x00680204 +#define NV_PVIDEO_CONTROL_X 0x00680208 +#define NV_PVIDEO_BUFF0_START_ADDRESS 0x0068020c +#define NV_PVIDEO_BUFF0_PITCH_LENGTH 0x00680214 +#define NV_PVIDEO_BUFF0_OFFSET 0x0068021c +#define NV_PVIDEO_BUFF1_START_ADDRESS 0x00680210 +#define NV_PVIDEO_BUFF1_PITCH_LENGTH 0x00680218 +#define NV_PVIDEO_BUFF1_OFFSET 0x00680220 +#define NV_PVIDEO_OE_STATE 0x00680224 +#define NV_PVIDEO_SU_STATE 0x00680228 +#define NV_PVIDEO_RM_STATE 0x0068022c +#define NV_PVIDEO_WINDOW_START 0x00680230 +#define NV_PVIDEO_WINDOW_SIZE 0x00680234 +#define NV_PVIDEO_FIFO_THRES_SIZE 0x00680238 +#define NV_PVIDEO_FIFO_BURST_LENGTH 0x0068023c +#define NV_PVIDEO_KEY 0x00680240 +#define NV_PVIDEO_OVERLAY 0x00680244 +#define NV_PVIDEO_RED_CSC_OFFSET 0x00680280 +#define NV_PVIDEO_GREEN_CSC_OFFSET 0x00680284 +#define NV_PVIDEO_BLUE_CSC_OFFSET 0x00680288 +#define NV_PVIDEO_CSC_ADJUST 0x0068028c + +#endif diff --git a/drivers/gpu/drm/r128/Makefile b/drivers/gpu/drm/r128/Makefile new file mode 100644 index 00000000..1cc72ae3 --- /dev/null +++ b/drivers/gpu/drm/r128/Makefile @@ -0,0 +1,10 @@ +# +# Makefile for the drm device driver. This driver provides support for the +# Direct Rendering Infrastructure (DRI) in XFree86 4.1.0 and higher. + +ccflags-y := -Iinclude/drm +r128-y := r128_drv.o r128_cce.o r128_state.o r128_irq.o + +r128-$(CONFIG_COMPAT) += r128_ioc32.o + +obj-$(CONFIG_DRM_R128) += r128.o diff --git a/drivers/gpu/drm/r128/r128_cce.c b/drivers/gpu/drm/r128/r128_cce.c new file mode 100644 index 00000000..570e1907 --- /dev/null +++ b/drivers/gpu/drm/r128/r128_cce.c @@ -0,0 +1,937 @@ +/* r128_cce.c -- ATI Rage 128 driver -*- linux-c -*- + * Created: Wed Apr 5 19:24:19 2000 by kevin@precisioninsight.com + */ +/* + * Copyright 2000 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Gareth Hughes + */ + +#include +#include +#include + +#include "drmP.h" +#include "drm.h" +#include "r128_drm.h" +#include "r128_drv.h" + +#define R128_FIFO_DEBUG 0 + +#define FIRMWARE_NAME "r128/r128_cce.bin" + +MODULE_FIRMWARE(FIRMWARE_NAME); + +static int R128_READ_PLL(struct drm_device *dev, int addr) +{ + drm_r128_private_t *dev_priv = dev->dev_private; + + R128_WRITE8(R128_CLOCK_CNTL_INDEX, addr & 0x1f); + return R128_READ(R128_CLOCK_CNTL_DATA); +} + +#if R128_FIFO_DEBUG +static void r128_status(drm_r128_private_t *dev_priv) +{ + printk("GUI_STAT = 0x%08x\n", + (unsigned int)R128_READ(R128_GUI_STAT)); + printk("PM4_STAT = 0x%08x\n", + (unsigned int)R128_READ(R128_PM4_STAT)); + printk("PM4_BUFFER_DL_WPTR = 0x%08x\n", + (unsigned int)R128_READ(R128_PM4_BUFFER_DL_WPTR)); + printk("PM4_BUFFER_DL_RPTR = 0x%08x\n", + (unsigned int)R128_READ(R128_PM4_BUFFER_DL_RPTR)); + printk("PM4_MICRO_CNTL = 0x%08x\n", + (unsigned int)R128_READ(R128_PM4_MICRO_CNTL)); + printk("PM4_BUFFER_CNTL = 0x%08x\n", + (unsigned int)R128_READ(R128_PM4_BUFFER_CNTL)); +} +#endif + +/* ================================================================ + * Engine, FIFO control + */ + +static int r128_do_pixcache_flush(drm_r128_private_t *dev_priv) +{ + u32 tmp; + int i; + + tmp = R128_READ(R128_PC_NGUI_CTLSTAT) | R128_PC_FLUSH_ALL; + R128_WRITE(R128_PC_NGUI_CTLSTAT, tmp); + + for (i = 0; i < dev_priv->usec_timeout; i++) { + if (!(R128_READ(R128_PC_NGUI_CTLSTAT) & R128_PC_BUSY)) + return 0; + DRM_UDELAY(1); + } + +#if R128_FIFO_DEBUG + DRM_ERROR("failed!\n"); +#endif + return -EBUSY; +} + +static int r128_do_wait_for_fifo(drm_r128_private_t *dev_priv, int entries) +{ + int i; + + for (i = 0; i < dev_priv->usec_timeout; i++) { + int slots = R128_READ(R128_GUI_STAT) & R128_GUI_FIFOCNT_MASK; + if (slots >= entries) + return 0; + DRM_UDELAY(1); + } + +#if R128_FIFO_DEBUG + DRM_ERROR("failed!\n"); +#endif + return -EBUSY; +} + +static int r128_do_wait_for_idle(drm_r128_private_t *dev_priv) +{ + int i, ret; + + ret = r128_do_wait_for_fifo(dev_priv, 64); + if (ret) + return ret; + + for (i = 0; i < dev_priv->usec_timeout; i++) { + if (!(R128_READ(R128_GUI_STAT) & R128_GUI_ACTIVE)) { + r128_do_pixcache_flush(dev_priv); + return 0; + } + DRM_UDELAY(1); + } + +#if R128_FIFO_DEBUG + DRM_ERROR("failed!\n"); +#endif + return -EBUSY; +} + +/* ================================================================ + * CCE control, initialization + */ + +/* Load the microcode for the CCE */ +static int r128_cce_load_microcode(drm_r128_private_t *dev_priv) +{ + struct platform_device *pdev; + const struct firmware *fw; + const __be32 *fw_data; + int rc, i; + + DRM_DEBUG("\n"); + + pdev = platform_device_register_simple("r128_cce", 0, NULL, 0); + if (IS_ERR(pdev)) { + printk(KERN_ERR "r128_cce: Failed to register firmware\n"); + return PTR_ERR(pdev); + } + rc = request_firmware(&fw, FIRMWARE_NAME, &pdev->dev); + platform_device_unregister(pdev); + if (rc) { + printk(KERN_ERR "r128_cce: Failed to load firmware \"%s\"\n", + FIRMWARE_NAME); + return rc; + } + + if (fw->size != 256 * 8) { + printk(KERN_ERR + "r128_cce: Bogus length %zu in firmware \"%s\"\n", + fw->size, FIRMWARE_NAME); + rc = -EINVAL; + goto out_release; + } + + r128_do_wait_for_idle(dev_priv); + + fw_data = (const __be32 *)fw->data; + R128_WRITE(R128_PM4_MICROCODE_ADDR, 0); + for (i = 0; i < 256; i++) { + R128_WRITE(R128_PM4_MICROCODE_DATAH, + be32_to_cpup(&fw_data[i * 2])); + R128_WRITE(R128_PM4_MICROCODE_DATAL, + be32_to_cpup(&fw_data[i * 2 + 1])); + } + +out_release: + release_firmware(fw); + return rc; +} + +/* Flush any pending commands to the CCE. This should only be used just + * prior to a wait for idle, as it informs the engine that the command + * stream is ending. + */ +static void r128_do_cce_flush(drm_r128_private_t *dev_priv) +{ + u32 tmp; + + tmp = R128_READ(R128_PM4_BUFFER_DL_WPTR) | R128_PM4_BUFFER_DL_DONE; + R128_WRITE(R128_PM4_BUFFER_DL_WPTR, tmp); +} + +/* Wait for the CCE to go idle. + */ +int r128_do_cce_idle(drm_r128_private_t *dev_priv) +{ + int i; + + for (i = 0; i < dev_priv->usec_timeout; i++) { + if (GET_RING_HEAD(dev_priv) == dev_priv->ring.tail) { + int pm4stat = R128_READ(R128_PM4_STAT); + if (((pm4stat & R128_PM4_FIFOCNT_MASK) >= + dev_priv->cce_fifo_size) && + !(pm4stat & (R128_PM4_BUSY | + R128_PM4_GUI_ACTIVE))) { + return r128_do_pixcache_flush(dev_priv); + } + } + DRM_UDELAY(1); + } + +#if R128_FIFO_DEBUG + DRM_ERROR("failed!\n"); + r128_status(dev_priv); +#endif + return -EBUSY; +} + +/* Start the Concurrent Command Engine. + */ +static void r128_do_cce_start(drm_r128_private_t *dev_priv) +{ + r128_do_wait_for_idle(dev_priv); + + R128_WRITE(R128_PM4_BUFFER_CNTL, + dev_priv->cce_mode | dev_priv->ring.size_l2qw + | R128_PM4_BUFFER_CNTL_NOUPDATE); + R128_READ(R128_PM4_BUFFER_ADDR); /* as per the sample code */ + R128_WRITE(R128_PM4_MICRO_CNTL, R128_PM4_MICRO_FREERUN); + + dev_priv->cce_running = 1; +} + +/* Reset the Concurrent Command Engine. This will not flush any pending + * commands, so you must wait for the CCE command stream to complete + * before calling this routine. + */ +static void r128_do_cce_reset(drm_r128_private_t *dev_priv) +{ + R128_WRITE(R128_PM4_BUFFER_DL_WPTR, 0); + R128_WRITE(R128_PM4_BUFFER_DL_RPTR, 0); + dev_priv->ring.tail = 0; +} + +/* Stop the Concurrent Command Engine. This will not flush any pending + * commands, so you must flush the command stream and wait for the CCE + * to go idle before calling this routine. + */ +static void r128_do_cce_stop(drm_r128_private_t *dev_priv) +{ + R128_WRITE(R128_PM4_MICRO_CNTL, 0); + R128_WRITE(R128_PM4_BUFFER_CNTL, + R128_PM4_NONPM4 | R128_PM4_BUFFER_CNTL_NOUPDATE); + + dev_priv->cce_running = 0; +} + +/* Reset the engine. This will stop the CCE if it is running. + */ +static int r128_do_engine_reset(struct drm_device *dev) +{ + drm_r128_private_t *dev_priv = dev->dev_private; + u32 clock_cntl_index, mclk_cntl, gen_reset_cntl; + + r128_do_pixcache_flush(dev_priv); + + clock_cntl_index = R128_READ(R128_CLOCK_CNTL_INDEX); + mclk_cntl = R128_READ_PLL(dev, R128_MCLK_CNTL); + + R128_WRITE_PLL(R128_MCLK_CNTL, + mclk_cntl | R128_FORCE_GCP | R128_FORCE_PIPE3D_CP); + + gen_reset_cntl = R128_READ(R128_GEN_RESET_CNTL); + + /* Taken from the sample code - do not change */ + R128_WRITE(R128_GEN_RESET_CNTL, gen_reset_cntl | R128_SOFT_RESET_GUI); + R128_READ(R128_GEN_RESET_CNTL); + R128_WRITE(R128_GEN_RESET_CNTL, gen_reset_cntl & ~R128_SOFT_RESET_GUI); + R128_READ(R128_GEN_RESET_CNTL); + + R128_WRITE_PLL(R128_MCLK_CNTL, mclk_cntl); + R128_WRITE(R128_CLOCK_CNTL_INDEX, clock_cntl_index); + R128_WRITE(R128_GEN_RESET_CNTL, gen_reset_cntl); + + /* Reset the CCE ring */ + r128_do_cce_reset(dev_priv); + + /* The CCE is no longer running after an engine reset */ + dev_priv->cce_running = 0; + + /* Reset any pending vertex, indirect buffers */ + r128_freelist_reset(dev); + + return 0; +} + +static void r128_cce_init_ring_buffer(struct drm_device *dev, + drm_r128_private_t *dev_priv) +{ + u32 ring_start; + u32 tmp; + + DRM_DEBUG("\n"); + + /* The manual (p. 2) says this address is in "VM space". This + * means it's an offset from the start of AGP space. + */ +#if __OS_HAS_AGP + if (!dev_priv->is_pci) + ring_start = dev_priv->cce_ring->offset - dev->agp->base; + else +#endif + ring_start = dev_priv->cce_ring->offset - + (unsigned long)dev->sg->virtual; + + R128_WRITE(R128_PM4_BUFFER_OFFSET, ring_start | R128_AGP_OFFSET); + + R128_WRITE(R128_PM4_BUFFER_DL_WPTR, 0); + R128_WRITE(R128_PM4_BUFFER_DL_RPTR, 0); + + /* Set watermark control */ + R128_WRITE(R128_PM4_BUFFER_WM_CNTL, + ((R128_WATERMARK_L / 4) << R128_WMA_SHIFT) + | ((R128_WATERMARK_M / 4) << R128_WMB_SHIFT) + | ((R128_WATERMARK_N / 4) << R128_WMC_SHIFT) + | ((R128_WATERMARK_K / 64) << R128_WB_WM_SHIFT)); + + /* Force read. Why? Because it's in the examples... */ + R128_READ(R128_PM4_BUFFER_ADDR); + + /* Turn on bus mastering */ + tmp = R128_READ(R128_BUS_CNTL) & ~R128_BUS_MASTER_DIS; + R128_WRITE(R128_BUS_CNTL, tmp); +} + +static int r128_do_init_cce(struct drm_device *dev, drm_r128_init_t *init) +{ + drm_r128_private_t *dev_priv; + int rc; + + DRM_DEBUG("\n"); + + if (dev->dev_private) { + DRM_DEBUG("called when already initialized\n"); + return -EINVAL; + } + + dev_priv = kzalloc(sizeof(drm_r128_private_t), GFP_KERNEL); + if (dev_priv == NULL) + return -ENOMEM; + + dev_priv->is_pci = init->is_pci; + + if (dev_priv->is_pci && !dev->sg) { + DRM_ERROR("PCI GART memory not allocated!\n"); + dev->dev_private = (void *)dev_priv; + r128_do_cleanup_cce(dev); + return -EINVAL; + } + + dev_priv->usec_timeout = init->usec_timeout; + if (dev_priv->usec_timeout < 1 || + dev_priv->usec_timeout > R128_MAX_USEC_TIMEOUT) { + DRM_DEBUG("TIMEOUT problem!\n"); + dev->dev_private = (void *)dev_priv; + r128_do_cleanup_cce(dev); + return -EINVAL; + } + + dev_priv->cce_mode = init->cce_mode; + + /* GH: Simple idle check. + */ + atomic_set(&dev_priv->idle_count, 0); + + /* We don't support anything other than bus-mastering ring mode, + * but the ring can be in either AGP or PCI space for the ring + * read pointer. + */ + if ((init->cce_mode != R128_PM4_192BM) && + (init->cce_mode != R128_PM4_128BM_64INDBM) && + (init->cce_mode != R128_PM4_64BM_128INDBM) && + (init->cce_mode != R128_PM4_64BM_64VCBM_64INDBM)) { + DRM_DEBUG("Bad cce_mode!\n"); + dev->dev_private = (void *)dev_priv; + r128_do_cleanup_cce(dev); + return -EINVAL; + } + + switch (init->cce_mode) { + case R128_PM4_NONPM4: + dev_priv->cce_fifo_size = 0; + break; + case R128_PM4_192PIO: + case R128_PM4_192BM: + dev_priv->cce_fifo_size = 192; + break; + case R128_PM4_128PIO_64INDBM: + case R128_PM4_128BM_64INDBM: + dev_priv->cce_fifo_size = 128; + break; + case R128_PM4_64PIO_128INDBM: + case R128_PM4_64BM_128INDBM: + case R128_PM4_64PIO_64VCBM_64INDBM: + case R128_PM4_64BM_64VCBM_64INDBM: + case R128_PM4_64PIO_64VCPIO_64INDPIO: + dev_priv->cce_fifo_size = 64; + break; + } + + switch (init->fb_bpp) { + case 16: + dev_priv->color_fmt = R128_DATATYPE_RGB565; + break; + case 32: + default: + dev_priv->color_fmt = R128_DATATYPE_ARGB8888; + break; + } + dev_priv->front_offset = init->front_offset; + dev_priv->front_pitch = init->front_pitch; + dev_priv->back_offset = init->back_offset; + dev_priv->back_pitch = init->back_pitch; + + switch (init->depth_bpp) { + case 16: + dev_priv->depth_fmt = R128_DATATYPE_RGB565; + break; + case 24: + case 32: + default: + dev_priv->depth_fmt = R128_DATATYPE_ARGB8888; + break; + } + dev_priv->depth_offset = init->depth_offset; + dev_priv->depth_pitch = init->depth_pitch; + dev_priv->span_offset = init->span_offset; + + dev_priv->front_pitch_offset_c = (((dev_priv->front_pitch / 8) << 21) | + (dev_priv->front_offset >> 5)); + dev_priv->back_pitch_offset_c = (((dev_priv->back_pitch / 8) << 21) | + (dev_priv->back_offset >> 5)); + dev_priv->depth_pitch_offset_c = (((dev_priv->depth_pitch / 8) << 21) | + (dev_priv->depth_offset >> 5) | + R128_DST_TILE); + dev_priv->span_pitch_offset_c = (((dev_priv->depth_pitch / 8) << 21) | + (dev_priv->span_offset >> 5)); + + dev_priv->sarea = drm_getsarea(dev); + if (!dev_priv->sarea) { + DRM_ERROR("could not find sarea!\n"); + dev->dev_private = (void *)dev_priv; + r128_do_cleanup_cce(dev); + return -EINVAL; + } + + dev_priv->mmio = drm_core_findmap(dev, init->mmio_offset); + if (!dev_priv->mmio) { + DRM_ERROR("could not find mmio region!\n"); + dev->dev_private = (void *)dev_priv; + r128_do_cleanup_cce(dev); + return -EINVAL; + } + dev_priv->cce_ring = drm_core_findmap(dev, init->ring_offset); + if (!dev_priv->cce_ring) { + DRM_ERROR("could not find cce ring region!\n"); + dev->dev_private = (void *)dev_priv; + r128_do_cleanup_cce(dev); + return -EINVAL; + } + dev_priv->ring_rptr = drm_core_findmap(dev, init->ring_rptr_offset); + if (!dev_priv->ring_rptr) { + DRM_ERROR("could not find ring read pointer!\n"); + dev->dev_private = (void *)dev_priv; + r128_do_cleanup_cce(dev); + return -EINVAL; + } + dev->agp_buffer_token = init->buffers_offset; + dev->agp_buffer_map = drm_core_findmap(dev, init->buffers_offset); + if (!dev->agp_buffer_map) { + DRM_ERROR("could not find dma buffer region!\n"); + dev->dev_private = (void *)dev_priv; + r128_do_cleanup_cce(dev); + return -EINVAL; + } + + if (!dev_priv->is_pci) { + dev_priv->agp_textures = + drm_core_findmap(dev, init->agp_textures_offset); + if (!dev_priv->agp_textures) { + DRM_ERROR("could not find agp texture region!\n"); + dev->dev_private = (void *)dev_priv; + r128_do_cleanup_cce(dev); + return -EINVAL; + } + } + + dev_priv->sarea_priv = + (drm_r128_sarea_t *) ((u8 *) dev_priv->sarea->handle + + init->sarea_priv_offset); + +#if __OS_HAS_AGP + if (!dev_priv->is_pci) { + drm_core_ioremap_wc(dev_priv->cce_ring, dev); + drm_core_ioremap_wc(dev_priv->ring_rptr, dev); + drm_core_ioremap_wc(dev->agp_buffer_map, dev); + if (!dev_priv->cce_ring->handle || + !dev_priv->ring_rptr->handle || + !dev->agp_buffer_map->handle) { + DRM_ERROR("Could not ioremap agp regions!\n"); + dev->dev_private = (void *)dev_priv; + r128_do_cleanup_cce(dev); + return -ENOMEM; + } + } else +#endif + { + dev_priv->cce_ring->handle = + (void *)(unsigned long)dev_priv->cce_ring->offset; + dev_priv->ring_rptr->handle = + (void *)(unsigned long)dev_priv->ring_rptr->offset; + dev->agp_buffer_map->handle = + (void *)(unsigned long)dev->agp_buffer_map->offset; + } + +#if __OS_HAS_AGP + if (!dev_priv->is_pci) + dev_priv->cce_buffers_offset = dev->agp->base; + else +#endif + dev_priv->cce_buffers_offset = (unsigned long)dev->sg->virtual; + + dev_priv->ring.start = (u32 *) dev_priv->cce_ring->handle; + dev_priv->ring.end = ((u32 *) dev_priv->cce_ring->handle + + init->ring_size / sizeof(u32)); + dev_priv->ring.size = init->ring_size; + dev_priv->ring.size_l2qw = drm_order(init->ring_size / 8); + + dev_priv->ring.tail_mask = (dev_priv->ring.size / sizeof(u32)) - 1; + + dev_priv->ring.high_mark = 128; + + dev_priv->sarea_priv->last_frame = 0; + R128_WRITE(R128_LAST_FRAME_REG, dev_priv->sarea_priv->last_frame); + + dev_priv->sarea_priv->last_dispatch = 0; + R128_WRITE(R128_LAST_DISPATCH_REG, dev_priv->sarea_priv->last_dispatch); + +#if __OS_HAS_AGP + if (dev_priv->is_pci) { +#endif + dev_priv->gart_info.table_mask = DMA_BIT_MASK(32); + dev_priv->gart_info.gart_table_location = DRM_ATI_GART_MAIN; + dev_priv->gart_info.table_size = R128_PCIGART_TABLE_SIZE; + dev_priv->gart_info.addr = NULL; + dev_priv->gart_info.bus_addr = 0; + dev_priv->gart_info.gart_reg_if = DRM_ATI_GART_PCI; + if (!drm_ati_pcigart_init(dev, &dev_priv->gart_info)) { + DRM_ERROR("failed to init PCI GART!\n"); + dev->dev_private = (void *)dev_priv; + r128_do_cleanup_cce(dev); + return -ENOMEM; + } + R128_WRITE(R128_PCI_GART_PAGE, dev_priv->gart_info.bus_addr); +#if __OS_HAS_AGP + } +#endif + + r128_cce_init_ring_buffer(dev, dev_priv); + rc = r128_cce_load_microcode(dev_priv); + + dev->dev_private = (void *)dev_priv; + + r128_do_engine_reset(dev); + + if (rc) { + DRM_ERROR("Failed to load firmware!\n"); + r128_do_cleanup_cce(dev); + } + + return rc; +} + +int r128_do_cleanup_cce(struct drm_device *dev) +{ + + /* Make sure interrupts are disabled here because the uninstall ioctl + * may not have been called from userspace and after dev_private + * is freed, it's too late. + */ + if (dev->irq_enabled) + drm_irq_uninstall(dev); + + if (dev->dev_private) { + drm_r128_private_t *dev_priv = dev->dev_private; + +#if __OS_HAS_AGP + if (!dev_priv->is_pci) { + if (dev_priv->cce_ring != NULL) + drm_core_ioremapfree(dev_priv->cce_ring, dev); + if (dev_priv->ring_rptr != NULL) + drm_core_ioremapfree(dev_priv->ring_rptr, dev); + if (dev->agp_buffer_map != NULL) { + drm_core_ioremapfree(dev->agp_buffer_map, dev); + dev->agp_buffer_map = NULL; + } + } else +#endif + { + if (dev_priv->gart_info.bus_addr) + if (!drm_ati_pcigart_cleanup(dev, + &dev_priv->gart_info)) + DRM_ERROR + ("failed to cleanup PCI GART!\n"); + } + + kfree(dev->dev_private); + dev->dev_private = NULL; + } + + return 0; +} + +int r128_cce_init(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_r128_init_t *init = data; + + DRM_DEBUG("\n"); + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + switch (init->func) { + case R128_INIT_CCE: + return r128_do_init_cce(dev, init); + case R128_CLEANUP_CCE: + return r128_do_cleanup_cce(dev); + } + + return -EINVAL; +} + +int r128_cce_start(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_r128_private_t *dev_priv = dev->dev_private; + DRM_DEBUG("\n"); + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + DEV_INIT_TEST_WITH_RETURN(dev_priv); + + if (dev_priv->cce_running || dev_priv->cce_mode == R128_PM4_NONPM4) { + DRM_DEBUG("while CCE running\n"); + return 0; + } + + r128_do_cce_start(dev_priv); + + return 0; +} + +/* Stop the CCE. The engine must have been idled before calling this + * routine. + */ +int r128_cce_stop(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_r128_private_t *dev_priv = dev->dev_private; + drm_r128_cce_stop_t *stop = data; + int ret; + DRM_DEBUG("\n"); + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + DEV_INIT_TEST_WITH_RETURN(dev_priv); + + /* Flush any pending CCE commands. This ensures any outstanding + * commands are exectuted by the engine before we turn it off. + */ + if (stop->flush) + r128_do_cce_flush(dev_priv); + + /* If we fail to make the engine go idle, we return an error + * code so that the DRM ioctl wrapper can try again. + */ + if (stop->idle) { + ret = r128_do_cce_idle(dev_priv); + if (ret) + return ret; + } + + /* Finally, we can turn off the CCE. If the engine isn't idle, + * we will get some dropped triangles as they won't be fully + * rendered before the CCE is shut down. + */ + r128_do_cce_stop(dev_priv); + + /* Reset the engine */ + r128_do_engine_reset(dev); + + return 0; +} + +/* Just reset the CCE ring. Called as part of an X Server engine reset. + */ +int r128_cce_reset(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_r128_private_t *dev_priv = dev->dev_private; + DRM_DEBUG("\n"); + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + DEV_INIT_TEST_WITH_RETURN(dev_priv); + + r128_do_cce_reset(dev_priv); + + /* The CCE is no longer running after an engine reset */ + dev_priv->cce_running = 0; + + return 0; +} + +int r128_cce_idle(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_r128_private_t *dev_priv = dev->dev_private; + DRM_DEBUG("\n"); + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + DEV_INIT_TEST_WITH_RETURN(dev_priv); + + if (dev_priv->cce_running) + r128_do_cce_flush(dev_priv); + + return r128_do_cce_idle(dev_priv); +} + +int r128_engine_reset(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + DRM_DEBUG("\n"); + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + DEV_INIT_TEST_WITH_RETURN(dev->dev_private); + + return r128_do_engine_reset(dev); +} + +int r128_fullscreen(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + return -EINVAL; +} + +/* ================================================================ + * Freelist management + */ +#define R128_BUFFER_USED 0xffffffff +#define R128_BUFFER_FREE 0 + +#if 0 +static int r128_freelist_init(struct drm_device *dev) +{ + struct drm_device_dma *dma = dev->dma; + drm_r128_private_t *dev_priv = dev->dev_private; + struct drm_buf *buf; + drm_r128_buf_priv_t *buf_priv; + drm_r128_freelist_t *entry; + int i; + + dev_priv->head = kzalloc(sizeof(drm_r128_freelist_t), GFP_KERNEL); + if (dev_priv->head == NULL) + return -ENOMEM; + + dev_priv->head->age = R128_BUFFER_USED; + + for (i = 0; i < dma->buf_count; i++) { + buf = dma->buflist[i]; + buf_priv = buf->dev_private; + + entry = kmalloc(sizeof(drm_r128_freelist_t), GFP_KERNEL); + if (!entry) + return -ENOMEM; + + entry->age = R128_BUFFER_FREE; + entry->buf = buf; + entry->prev = dev_priv->head; + entry->next = dev_priv->head->next; + if (!entry->next) + dev_priv->tail = entry; + + buf_priv->discard = 0; + buf_priv->dispatched = 0; + buf_priv->list_entry = entry; + + dev_priv->head->next = entry; + + if (dev_priv->head->next) + dev_priv->head->next->prev = entry; + } + + return 0; + +} +#endif + +static struct drm_buf *r128_freelist_get(struct drm_device * dev) +{ + struct drm_device_dma *dma = dev->dma; + drm_r128_private_t *dev_priv = dev->dev_private; + drm_r128_buf_priv_t *buf_priv; + struct drm_buf *buf; + int i, t; + + /* FIXME: Optimize -- use freelist code */ + + for (i = 0; i < dma->buf_count; i++) { + buf = dma->buflist[i]; + buf_priv = buf->dev_private; + if (!buf->file_priv) + return buf; + } + + for (t = 0; t < dev_priv->usec_timeout; t++) { + u32 done_age = R128_READ(R128_LAST_DISPATCH_REG); + + for (i = 0; i < dma->buf_count; i++) { + buf = dma->buflist[i]; + buf_priv = buf->dev_private; + if (buf->pending && buf_priv->age <= done_age) { + /* The buffer has been processed, so it + * can now be used. + */ + buf->pending = 0; + return buf; + } + } + DRM_UDELAY(1); + } + + DRM_DEBUG("returning NULL!\n"); + return NULL; +} + +void r128_freelist_reset(struct drm_device *dev) +{ + struct drm_device_dma *dma = dev->dma; + int i; + + for (i = 0; i < dma->buf_count; i++) { + struct drm_buf *buf = dma->buflist[i]; + drm_r128_buf_priv_t *buf_priv = buf->dev_private; + buf_priv->age = 0; + } +} + +/* ================================================================ + * CCE command submission + */ + +int r128_wait_ring(drm_r128_private_t *dev_priv, int n) +{ + drm_r128_ring_buffer_t *ring = &dev_priv->ring; + int i; + + for (i = 0; i < dev_priv->usec_timeout; i++) { + r128_update_ring_snapshot(dev_priv); + if (ring->space >= n) + return 0; + DRM_UDELAY(1); + } + + /* FIXME: This is being ignored... */ + DRM_ERROR("failed!\n"); + return -EBUSY; +} + +static int r128_cce_get_buffers(struct drm_device *dev, + struct drm_file *file_priv, + struct drm_dma *d) +{ + int i; + struct drm_buf *buf; + + for (i = d->granted_count; i < d->request_count; i++) { + buf = r128_freelist_get(dev); + if (!buf) + return -EAGAIN; + + buf->file_priv = file_priv; + + if (DRM_COPY_TO_USER(&d->request_indices[i], &buf->idx, + sizeof(buf->idx))) + return -EFAULT; + if (DRM_COPY_TO_USER(&d->request_sizes[i], &buf->total, + sizeof(buf->total))) + return -EFAULT; + + d->granted_count++; + } + return 0; +} + +int r128_cce_buffers(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + struct drm_device_dma *dma = dev->dma; + int ret = 0; + struct drm_dma *d = data; + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + /* Please don't send us buffers. + */ + if (d->send_count != 0) { + DRM_ERROR("Process %d trying to send %d buffers via drmDMA\n", + DRM_CURRENTPID, d->send_count); + return -EINVAL; + } + + /* We'll send you buffers. + */ + if (d->request_count < 0 || d->request_count > dma->buf_count) { + DRM_ERROR("Process %d trying to get %d buffers (of %d max)\n", + DRM_CURRENTPID, d->request_count, dma->buf_count); + return -EINVAL; + } + + d->granted_count = 0; + + if (d->request_count) + ret = r128_cce_get_buffers(dev, file_priv, d); + + return ret; +} diff --git a/drivers/gpu/drm/r128/r128_drv.c b/drivers/gpu/drm/r128/r128_drv.c new file mode 100644 index 00000000..b9e8efd2 --- /dev/null +++ b/drivers/gpu/drm/r128/r128_drv.c @@ -0,0 +1,111 @@ +/* r128_drv.c -- ATI Rage 128 driver -*- linux-c -*- + * Created: Mon Dec 13 09:47:27 1999 by faith@precisioninsight.com + * + * Copyright 1999 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * VA LINUX SYSTEMS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: + * Rickard E. (Rik) Faith + * Gareth Hughes + */ + +#include "drmP.h" +#include "drm.h" +#include "r128_drm.h" +#include "r128_drv.h" + +#include "drm_pciids.h" + +static struct pci_device_id pciidlist[] = { + r128_PCI_IDS +}; + +static struct drm_driver driver = { + .driver_features = + DRIVER_USE_AGP | DRIVER_USE_MTRR | DRIVER_PCI_DMA | DRIVER_SG | + DRIVER_HAVE_DMA | DRIVER_HAVE_IRQ | DRIVER_IRQ_SHARED, + .dev_priv_size = sizeof(drm_r128_buf_priv_t), + .load = r128_driver_load, + .preclose = r128_driver_preclose, + .lastclose = r128_driver_lastclose, + .get_vblank_counter = r128_get_vblank_counter, + .enable_vblank = r128_enable_vblank, + .disable_vblank = r128_disable_vblank, + .irq_preinstall = r128_driver_irq_preinstall, + .irq_postinstall = r128_driver_irq_postinstall, + .irq_uninstall = r128_driver_irq_uninstall, + .irq_handler = r128_driver_irq_handler, + .reclaim_buffers = drm_core_reclaim_buffers, + .ioctls = r128_ioctls, + .dma_ioctl = r128_cce_buffers, + .fops = { + .owner = THIS_MODULE, + .open = drm_open, + .release = drm_release, + .unlocked_ioctl = drm_ioctl, + .mmap = drm_mmap, + .poll = drm_poll, + .fasync = drm_fasync, +#ifdef CONFIG_COMPAT + .compat_ioctl = r128_compat_ioctl, +#endif + .llseek = noop_llseek, + }, + + + .name = DRIVER_NAME, + .desc = DRIVER_DESC, + .date = DRIVER_DATE, + .major = DRIVER_MAJOR, + .minor = DRIVER_MINOR, + .patchlevel = DRIVER_PATCHLEVEL, +}; + +int r128_driver_load(struct drm_device *dev, unsigned long flags) +{ + return drm_vblank_init(dev, 1); +} + +static struct pci_driver r128_pci_driver = { + .name = DRIVER_NAME, + .id_table = pciidlist, +}; + +static int __init r128_init(void) +{ + driver.num_ioctls = r128_max_ioctl; + + return drm_pci_init(&driver, &r128_pci_driver); +} + +static void __exit r128_exit(void) +{ + drm_pci_exit(&driver, &r128_pci_driver); +} + +module_init(r128_init); +module_exit(r128_exit); + +MODULE_AUTHOR(DRIVER_AUTHOR); +MODULE_DESCRIPTION(DRIVER_DESC); +MODULE_LICENSE("GPL and additional rights"); diff --git a/drivers/gpu/drm/r128/r128_drv.h b/drivers/gpu/drm/r128/r128_drv.h new file mode 100644 index 00000000..930c71b2 --- /dev/null +++ b/drivers/gpu/drm/r128/r128_drv.h @@ -0,0 +1,530 @@ +/* r128_drv.h -- Private header for r128 driver -*- linux-c -*- + * Created: Mon Dec 13 09:51:11 1999 by faith@precisioninsight.com + */ +/* + * Copyright 1999 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California. + * All rights reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Rickard E. (Rik) Faith + * Kevin E. Martin + * Gareth Hughes + * Michel D�zer + */ + +#ifndef __R128_DRV_H__ +#define __R128_DRV_H__ + +/* General customization: + */ +#define DRIVER_AUTHOR "Gareth Hughes, VA Linux Systems Inc." + +#define DRIVER_NAME "r128" +#define DRIVER_DESC "ATI Rage 128" +#define DRIVER_DATE "20030725" + +/* Interface history: + * + * ?? - ?? + * 2.4 - Add support for ycbcr textures (no new ioctls) + * 2.5 - Add FLIP ioctl, disable FULLSCREEN. + */ +#define DRIVER_MAJOR 2 +#define DRIVER_MINOR 5 +#define DRIVER_PATCHLEVEL 0 + +#define GET_RING_HEAD(dev_priv) R128_READ(R128_PM4_BUFFER_DL_RPTR) + +typedef struct drm_r128_freelist { + unsigned int age; + struct drm_buf *buf; + struct drm_r128_freelist *next; + struct drm_r128_freelist *prev; +} drm_r128_freelist_t; + +typedef struct drm_r128_ring_buffer { + u32 *start; + u32 *end; + int size; + int size_l2qw; + + u32 tail; + u32 tail_mask; + int space; + + int high_mark; +} drm_r128_ring_buffer_t; + +typedef struct drm_r128_private { + drm_r128_ring_buffer_t ring; + drm_r128_sarea_t *sarea_priv; + + int cce_mode; + int cce_fifo_size; + int cce_running; + + drm_r128_freelist_t *head; + drm_r128_freelist_t *tail; + + int usec_timeout; + int is_pci; + unsigned long cce_buffers_offset; + + atomic_t idle_count; + + int page_flipping; + int current_page; + u32 crtc_offset; + u32 crtc_offset_cntl; + + atomic_t vbl_received; + + u32 color_fmt; + unsigned int front_offset; + unsigned int front_pitch; + unsigned int back_offset; + unsigned int back_pitch; + + u32 depth_fmt; + unsigned int depth_offset; + unsigned int depth_pitch; + unsigned int span_offset; + + u32 front_pitch_offset_c; + u32 back_pitch_offset_c; + u32 depth_pitch_offset_c; + u32 span_pitch_offset_c; + + drm_local_map_t *sarea; + drm_local_map_t *mmio; + drm_local_map_t *cce_ring; + drm_local_map_t *ring_rptr; + drm_local_map_t *agp_textures; + struct drm_ati_pcigart_info gart_info; +} drm_r128_private_t; + +typedef struct drm_r128_buf_priv { + u32 age; + int prim; + int discard; + int dispatched; + drm_r128_freelist_t *list_entry; +} drm_r128_buf_priv_t; + +extern struct drm_ioctl_desc r128_ioctls[]; +extern int r128_max_ioctl; + + /* r128_cce.c */ +extern int r128_cce_init(struct drm_device *dev, void *data, struct drm_file *file_priv); +extern int r128_cce_start(struct drm_device *dev, void *data, struct drm_file *file_priv); +extern int r128_cce_stop(struct drm_device *dev, void *data, struct drm_file *file_priv); +extern int r128_cce_reset(struct drm_device *dev, void *data, struct drm_file *file_priv); +extern int r128_cce_idle(struct drm_device *dev, void *data, struct drm_file *file_priv); +extern int r128_engine_reset(struct drm_device *dev, void *data, struct drm_file *file_priv); +extern int r128_fullscreen(struct drm_device *dev, void *data, struct drm_file *file_priv); +extern int r128_cce_buffers(struct drm_device *dev, void *data, struct drm_file *file_priv); + +extern void r128_freelist_reset(struct drm_device *dev); + +extern int r128_wait_ring(drm_r128_private_t *dev_priv, int n); + +extern int r128_do_cce_idle(drm_r128_private_t *dev_priv); +extern int r128_do_cleanup_cce(struct drm_device *dev); + +extern int r128_enable_vblank(struct drm_device *dev, int crtc); +extern void r128_disable_vblank(struct drm_device *dev, int crtc); +extern u32 r128_get_vblank_counter(struct drm_device *dev, int crtc); +extern irqreturn_t r128_driver_irq_handler(DRM_IRQ_ARGS); +extern void r128_driver_irq_preinstall(struct drm_device *dev); +extern int r128_driver_irq_postinstall(struct drm_device *dev); +extern void r128_driver_irq_uninstall(struct drm_device *dev); +extern void r128_driver_lastclose(struct drm_device *dev); +extern int r128_driver_load(struct drm_device *dev, unsigned long flags); +extern void r128_driver_preclose(struct drm_device *dev, + struct drm_file *file_priv); + +extern long r128_compat_ioctl(struct file *filp, unsigned int cmd, + unsigned long arg); + +/* Register definitions, register access macros and drmAddMap constants + * for Rage 128 kernel driver. + */ + +#define R128_AUX_SC_CNTL 0x1660 +# define R128_AUX1_SC_EN (1 << 0) +# define R128_AUX1_SC_MODE_OR (0 << 1) +# define R128_AUX1_SC_MODE_NAND (1 << 1) +# define R128_AUX2_SC_EN (1 << 2) +# define R128_AUX2_SC_MODE_OR (0 << 3) +# define R128_AUX2_SC_MODE_NAND (1 << 3) +# define R128_AUX3_SC_EN (1 << 4) +# define R128_AUX3_SC_MODE_OR (0 << 5) +# define R128_AUX3_SC_MODE_NAND (1 << 5) +#define R128_AUX1_SC_LEFT 0x1664 +#define R128_AUX1_SC_RIGHT 0x1668 +#define R128_AUX1_SC_TOP 0x166c +#define R128_AUX1_SC_BOTTOM 0x1670 +#define R128_AUX2_SC_LEFT 0x1674 +#define R128_AUX2_SC_RIGHT 0x1678 +#define R128_AUX2_SC_TOP 0x167c +#define R128_AUX2_SC_BOTTOM 0x1680 +#define R128_AUX3_SC_LEFT 0x1684 +#define R128_AUX3_SC_RIGHT 0x1688 +#define R128_AUX3_SC_TOP 0x168c +#define R128_AUX3_SC_BOTTOM 0x1690 + +#define R128_BRUSH_DATA0 0x1480 +#define R128_BUS_CNTL 0x0030 +# define R128_BUS_MASTER_DIS (1 << 6) + +#define R128_CLOCK_CNTL_INDEX 0x0008 +#define R128_CLOCK_CNTL_DATA 0x000c +# define R128_PLL_WR_EN (1 << 7) +#define R128_CONSTANT_COLOR_C 0x1d34 +#define R128_CRTC_OFFSET 0x0224 +#define R128_CRTC_OFFSET_CNTL 0x0228 +# define R128_CRTC_OFFSET_FLIP_CNTL (1 << 16) + +#define R128_DP_GUI_MASTER_CNTL 0x146c +# define R128_GMC_SRC_PITCH_OFFSET_CNTL (1 << 0) +# define R128_GMC_DST_PITCH_OFFSET_CNTL (1 << 1) +# define R128_GMC_BRUSH_SOLID_COLOR (13 << 4) +# define R128_GMC_BRUSH_NONE (15 << 4) +# define R128_GMC_DST_16BPP (4 << 8) +# define R128_GMC_DST_24BPP (5 << 8) +# define R128_GMC_DST_32BPP (6 << 8) +# define R128_GMC_DST_DATATYPE_SHIFT 8 +# define R128_GMC_SRC_DATATYPE_COLOR (3 << 12) +# define R128_DP_SRC_SOURCE_MEMORY (2 << 24) +# define R128_DP_SRC_SOURCE_HOST_DATA (3 << 24) +# define R128_GMC_CLR_CMP_CNTL_DIS (1 << 28) +# define R128_GMC_AUX_CLIP_DIS (1 << 29) +# define R128_GMC_WR_MSK_DIS (1 << 30) +# define R128_ROP3_S 0x00cc0000 +# define R128_ROP3_P 0x00f00000 +#define R128_DP_WRITE_MASK 0x16cc +#define R128_DST_PITCH_OFFSET_C 0x1c80 +# define R128_DST_TILE (1 << 31) + +#define R128_GEN_INT_CNTL 0x0040 +# define R128_CRTC_VBLANK_INT_EN (1 << 0) +#define R128_GEN_INT_STATUS 0x0044 +# define R128_CRTC_VBLANK_INT (1 << 0) +# define R128_CRTC_VBLANK_INT_AK (1 << 0) +#define R128_GEN_RESET_CNTL 0x00f0 +# define R128_SOFT_RESET_GUI (1 << 0) + +#define R128_GUI_SCRATCH_REG0 0x15e0 +#define R128_GUI_SCRATCH_REG1 0x15e4 +#define R128_GUI_SCRATCH_REG2 0x15e8 +#define R128_GUI_SCRATCH_REG3 0x15ec +#define R128_GUI_SCRATCH_REG4 0x15f0 +#define R128_GUI_SCRATCH_REG5 0x15f4 + +#define R128_GUI_STAT 0x1740 +# define R128_GUI_FIFOCNT_MASK 0x0fff +# define R128_GUI_ACTIVE (1 << 31) + +#define R128_MCLK_CNTL 0x000f +# define R128_FORCE_GCP (1 << 16) +# define R128_FORCE_PIPE3D_CP (1 << 17) +# define R128_FORCE_RCP (1 << 18) + +#define R128_PC_GUI_CTLSTAT 0x1748 +#define R128_PC_NGUI_CTLSTAT 0x0184 +# define R128_PC_FLUSH_GUI (3 << 0) +# define R128_PC_RI_GUI (1 << 2) +# define R128_PC_FLUSH_ALL 0x00ff +# define R128_PC_BUSY (1 << 31) + +#define R128_PCI_GART_PAGE 0x017c +#define R128_PRIM_TEX_CNTL_C 0x1cb0 + +#define R128_SCALE_3D_CNTL 0x1a00 +#define R128_SEC_TEX_CNTL_C 0x1d00 +#define R128_SEC_TEXTURE_BORDER_COLOR_C 0x1d3c +#define R128_SETUP_CNTL 0x1bc4 +#define R128_STEN_REF_MASK_C 0x1d40 + +#define R128_TEX_CNTL_C 0x1c9c +# define R128_TEX_CACHE_FLUSH (1 << 23) + +#define R128_WAIT_UNTIL 0x1720 +# define R128_EVENT_CRTC_OFFSET (1 << 0) +#define R128_WINDOW_XY_OFFSET 0x1bcc + +/* CCE registers + */ +#define R128_PM4_BUFFER_OFFSET 0x0700 +#define R128_PM4_BUFFER_CNTL 0x0704 +# define R128_PM4_MASK (15 << 28) +# define R128_PM4_NONPM4 (0 << 28) +# define R128_PM4_192PIO (1 << 28) +# define R128_PM4_192BM (2 << 28) +# define R128_PM4_128PIO_64INDBM (3 << 28) +# define R128_PM4_128BM_64INDBM (4 << 28) +# define R128_PM4_64PIO_128INDBM (5 << 28) +# define R128_PM4_64BM_128INDBM (6 << 28) +# define R128_PM4_64PIO_64VCBM_64INDBM (7 << 28) +# define R128_PM4_64BM_64VCBM_64INDBM (8 << 28) +# define R128_PM4_64PIO_64VCPIO_64INDPIO (15 << 28) +# define R128_PM4_BUFFER_CNTL_NOUPDATE (1 << 27) + +#define R128_PM4_BUFFER_WM_CNTL 0x0708 +# define R128_WMA_SHIFT 0 +# define R128_WMB_SHIFT 8 +# define R128_WMC_SHIFT 16 +# define R128_WB_WM_SHIFT 24 + +#define R128_PM4_BUFFER_DL_RPTR_ADDR 0x070c +#define R128_PM4_BUFFER_DL_RPTR 0x0710 +#define R128_PM4_BUFFER_DL_WPTR 0x0714 +# define R128_PM4_BUFFER_DL_DONE (1 << 31) + +#define R128_PM4_VC_FPU_SETUP 0x071c + +#define R128_PM4_IW_INDOFF 0x0738 +#define R128_PM4_IW_INDSIZE 0x073c + +#define R128_PM4_STAT 0x07b8 +# define R128_PM4_FIFOCNT_MASK 0x0fff +# define R128_PM4_BUSY (1 << 16) +# define R128_PM4_GUI_ACTIVE (1 << 31) + +#define R128_PM4_MICROCODE_ADDR 0x07d4 +#define R128_PM4_MICROCODE_RADDR 0x07d8 +#define R128_PM4_MICROCODE_DATAH 0x07dc +#define R128_PM4_MICROCODE_DATAL 0x07e0 + +#define R128_PM4_BUFFER_ADDR 0x07f0 +#define R128_PM4_MICRO_CNTL 0x07fc +# define R128_PM4_MICRO_FREERUN (1 << 30) + +#define R128_PM4_FIFO_DATA_EVEN 0x1000 +#define R128_PM4_FIFO_DATA_ODD 0x1004 + +/* CCE command packets + */ +#define R128_CCE_PACKET0 0x00000000 +#define R128_CCE_PACKET1 0x40000000 +#define R128_CCE_PACKET2 0x80000000 +#define R128_CCE_PACKET3 0xC0000000 +# define R128_CNTL_HOSTDATA_BLT 0x00009400 +# define R128_CNTL_PAINT_MULTI 0x00009A00 +# define R128_CNTL_BITBLT_MULTI 0x00009B00 +# define R128_3D_RNDR_GEN_INDX_PRIM 0x00002300 + +#define R128_CCE_PACKET_MASK 0xC0000000 +#define R128_CCE_PACKET_COUNT_MASK 0x3fff0000 +#define R128_CCE_PACKET0_REG_MASK 0x000007ff +#define R128_CCE_PACKET1_REG0_MASK 0x000007ff +#define R128_CCE_PACKET1_REG1_MASK 0x003ff800 + +#define R128_CCE_VC_CNTL_PRIM_TYPE_NONE 0x00000000 +#define R128_CCE_VC_CNTL_PRIM_TYPE_POINT 0x00000001 +#define R128_CCE_VC_CNTL_PRIM_TYPE_LINE 0x00000002 +#define R128_CCE_VC_CNTL_PRIM_TYPE_POLY_LINE 0x00000003 +#define R128_CCE_VC_CNTL_PRIM_TYPE_TRI_LIST 0x00000004 +#define R128_CCE_VC_CNTL_PRIM_TYPE_TRI_FAN 0x00000005 +#define R128_CCE_VC_CNTL_PRIM_TYPE_TRI_STRIP 0x00000006 +#define R128_CCE_VC_CNTL_PRIM_TYPE_TRI_TYPE2 0x00000007 +#define R128_CCE_VC_CNTL_PRIM_WALK_IND 0x00000010 +#define R128_CCE_VC_CNTL_PRIM_WALK_LIST 0x00000020 +#define R128_CCE_VC_CNTL_PRIM_WALK_RING 0x00000030 +#define R128_CCE_VC_CNTL_NUM_SHIFT 16 + +#define R128_DATATYPE_VQ 0 +#define R128_DATATYPE_CI4 1 +#define R128_DATATYPE_CI8 2 +#define R128_DATATYPE_ARGB1555 3 +#define R128_DATATYPE_RGB565 4 +#define R128_DATATYPE_RGB888 5 +#define R128_DATATYPE_ARGB8888 6 +#define R128_DATATYPE_RGB332 7 +#define R128_DATATYPE_Y8 8 +#define R128_DATATYPE_RGB8 9 +#define R128_DATATYPE_CI16 10 +#define R128_DATATYPE_YVYU422 11 +#define R128_DATATYPE_VYUY422 12 +#define R128_DATATYPE_AYUV444 14 +#define R128_DATATYPE_ARGB4444 15 + +/* Constants */ +#define R128_AGP_OFFSET 0x02000000 + +#define R128_WATERMARK_L 16 +#define R128_WATERMARK_M 8 +#define R128_WATERMARK_N 8 +#define R128_WATERMARK_K 128 + +#define R128_MAX_USEC_TIMEOUT 100000 /* 100 ms */ + +#define R128_LAST_FRAME_REG R128_GUI_SCRATCH_REG0 +#define R128_LAST_DISPATCH_REG R128_GUI_SCRATCH_REG1 +#define R128_MAX_VB_AGE 0x7fffffff +#define R128_MAX_VB_VERTS (0xffff) + +#define R128_RING_HIGH_MARK 128 + +#define R128_PERFORMANCE_BOXES 0 + +#define R128_PCIGART_TABLE_SIZE 32768 + +#define R128_READ(reg) DRM_READ32(dev_priv->mmio, (reg)) +#define R128_WRITE(reg, val) DRM_WRITE32(dev_priv->mmio, (reg), (val)) +#define R128_READ8(reg) DRM_READ8(dev_priv->mmio, (reg)) +#define R128_WRITE8(reg, val) DRM_WRITE8(dev_priv->mmio, (reg), (val)) + +#define R128_WRITE_PLL(addr, val) \ +do { \ + R128_WRITE8(R128_CLOCK_CNTL_INDEX, \ + ((addr) & 0x1f) | R128_PLL_WR_EN); \ + R128_WRITE(R128_CLOCK_CNTL_DATA, (val)); \ +} while (0) + +#define CCE_PACKET0(reg, n) (R128_CCE_PACKET0 | \ + ((n) << 16) | ((reg) >> 2)) +#define CCE_PACKET1(reg0, reg1) (R128_CCE_PACKET1 | \ + (((reg1) >> 2) << 11) | ((reg0) >> 2)) +#define CCE_PACKET2() (R128_CCE_PACKET2) +#define CCE_PACKET3(pkt, n) (R128_CCE_PACKET3 | \ + (pkt) | ((n) << 16)) + +static __inline__ void r128_update_ring_snapshot(drm_r128_private_t *dev_priv) +{ + drm_r128_ring_buffer_t *ring = &dev_priv->ring; + ring->space = (GET_RING_HEAD(dev_priv) - ring->tail) * sizeof(u32); + if (ring->space <= 0) + ring->space += ring->size; +} + +/* ================================================================ + * Misc helper macros + */ + +#define DEV_INIT_TEST_WITH_RETURN(_dev_priv) \ +do { \ + if (!_dev_priv) { \ + DRM_ERROR("called with no initialization\n"); \ + return -EINVAL; \ + } \ +} while (0) + +#define RING_SPACE_TEST_WITH_RETURN(dev_priv) \ +do { \ + drm_r128_ring_buffer_t *ring = &dev_priv->ring; int i; \ + if (ring->space < ring->high_mark) { \ + for (i = 0 ; i < dev_priv->usec_timeout ; i++) { \ + r128_update_ring_snapshot(dev_priv); \ + if (ring->space >= ring->high_mark) \ + goto __ring_space_done; \ + DRM_UDELAY(1); \ + } \ + DRM_ERROR("ring space check failed!\n"); \ + return -EBUSY; \ + } \ + __ring_space_done: \ + ; \ +} while (0) + +#define VB_AGE_TEST_WITH_RETURN(dev_priv) \ +do { \ + drm_r128_sarea_t *sarea_priv = dev_priv->sarea_priv; \ + if (sarea_priv->last_dispatch >= R128_MAX_VB_AGE) { \ + int __ret = r128_do_cce_idle(dev_priv); \ + if (__ret) \ + return __ret; \ + sarea_priv->last_dispatch = 0; \ + r128_freelist_reset(dev); \ + } \ +} while (0) + +#define R128_WAIT_UNTIL_PAGE_FLIPPED() do { \ + OUT_RING(CCE_PACKET0(R128_WAIT_UNTIL, 0)); \ + OUT_RING(R128_EVENT_CRTC_OFFSET); \ +} while (0) + +/* ================================================================ + * Ring control + */ + +#define R128_VERBOSE 0 + +#define RING_LOCALS \ + int write, _nr; unsigned int tail_mask; volatile u32 *ring; + +#define BEGIN_RING(n) do { \ + if (R128_VERBOSE) \ + DRM_INFO("BEGIN_RING(%d)\n", (n)); \ + if (dev_priv->ring.space <= (n) * sizeof(u32)) { \ + COMMIT_RING(); \ + r128_wait_ring(dev_priv, (n) * sizeof(u32)); \ + } \ + _nr = n; dev_priv->ring.space -= (n) * sizeof(u32); \ + ring = dev_priv->ring.start; \ + write = dev_priv->ring.tail; \ + tail_mask = dev_priv->ring.tail_mask; \ +} while (0) + +/* You can set this to zero if you want. If the card locks up, you'll + * need to keep this set. It works around a bug in early revs of the + * Rage 128 chipset, where the CCE would read 32 dwords past the end of + * the ring buffer before wrapping around. + */ +#define R128_BROKEN_CCE 1 + +#define ADVANCE_RING() do { \ + if (R128_VERBOSE) \ + DRM_INFO("ADVANCE_RING() wr=0x%06x tail=0x%06x\n", \ + write, dev_priv->ring.tail); \ + if (R128_BROKEN_CCE && write < 32) \ + memcpy(dev_priv->ring.end, \ + dev_priv->ring.start, \ + write * sizeof(u32)); \ + if (((dev_priv->ring.tail + _nr) & tail_mask) != write) \ + DRM_ERROR( \ + "ADVANCE_RING(): mismatch: nr: %x write: %x line: %d\n", \ + ((dev_priv->ring.tail + _nr) & tail_mask), \ + write, __LINE__); \ + else \ + dev_priv->ring.tail = write; \ +} while (0) + +#define COMMIT_RING() do { \ + if (R128_VERBOSE) \ + DRM_INFO("COMMIT_RING() tail=0x%06x\n", \ + dev_priv->ring.tail); \ + DRM_MEMORYBARRIER(); \ + R128_WRITE(R128_PM4_BUFFER_DL_WPTR, dev_priv->ring.tail); \ + R128_READ(R128_PM4_BUFFER_DL_WPTR); \ +} while (0) + +#define OUT_RING(x) do { \ + if (R128_VERBOSE) \ + DRM_INFO(" OUT_RING( 0x%08x ) at 0x%x\n", \ + (unsigned int)(x), write); \ + ring[write++] = cpu_to_le32(x); \ + write &= tail_mask; \ +} while (0) + +#endif /* __R128_DRV_H__ */ diff --git a/drivers/gpu/drm/r128/r128_ioc32.c b/drivers/gpu/drm/r128/r128_ioc32.c new file mode 100644 index 00000000..51c99fc4 --- /dev/null +++ b/drivers/gpu/drm/r128/r128_ioc32.c @@ -0,0 +1,215 @@ +/** + * \file r128_ioc32.c + * + * 32-bit ioctl compatibility routines for the R128 DRM. + * + * \author Dave Airlie with code from patches by Egbert Eich + * + * Copyright (C) Paul Mackerras 2005 + * Copyright (C) Egbert Eich 2003,2004 + * Copyright (C) Dave Airlie 2005 + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHOR BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS + * IN THE SOFTWARE. + */ +#include + +#include "drmP.h" +#include "drm.h" +#include "r128_drm.h" + +typedef struct drm_r128_init32 { + int func; + unsigned int sarea_priv_offset; + int is_pci; + int cce_mode; + int cce_secure; + int ring_size; + int usec_timeout; + + unsigned int fb_bpp; + unsigned int front_offset, front_pitch; + unsigned int back_offset, back_pitch; + unsigned int depth_bpp; + unsigned int depth_offset, depth_pitch; + unsigned int span_offset; + + unsigned int fb_offset; + unsigned int mmio_offset; + unsigned int ring_offset; + unsigned int ring_rptr_offset; + unsigned int buffers_offset; + unsigned int agp_textures_offset; +} drm_r128_init32_t; + +static int compat_r128_init(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_r128_init32_t init32; + drm_r128_init_t __user *init; + + if (copy_from_user(&init32, (void __user *)arg, sizeof(init32))) + return -EFAULT; + + init = compat_alloc_user_space(sizeof(*init)); + if (!access_ok(VERIFY_WRITE, init, sizeof(*init)) + || __put_user(init32.func, &init->func) + || __put_user(init32.sarea_priv_offset, &init->sarea_priv_offset) + || __put_user(init32.is_pci, &init->is_pci) + || __put_user(init32.cce_mode, &init->cce_mode) + || __put_user(init32.cce_secure, &init->cce_secure) + || __put_user(init32.ring_size, &init->ring_size) + || __put_user(init32.usec_timeout, &init->usec_timeout) + || __put_user(init32.fb_bpp, &init->fb_bpp) + || __put_user(init32.front_offset, &init->front_offset) + || __put_user(init32.front_pitch, &init->front_pitch) + || __put_user(init32.back_offset, &init->back_offset) + || __put_user(init32.back_pitch, &init->back_pitch) + || __put_user(init32.depth_bpp, &init->depth_bpp) + || __put_user(init32.depth_offset, &init->depth_offset) + || __put_user(init32.depth_pitch, &init->depth_pitch) + || __put_user(init32.span_offset, &init->span_offset) + || __put_user(init32.fb_offset, &init->fb_offset) + || __put_user(init32.mmio_offset, &init->mmio_offset) + || __put_user(init32.ring_offset, &init->ring_offset) + || __put_user(init32.ring_rptr_offset, &init->ring_rptr_offset) + || __put_user(init32.buffers_offset, &init->buffers_offset) + || __put_user(init32.agp_textures_offset, + &init->agp_textures_offset)) + return -EFAULT; + + return drm_ioctl(file, DRM_IOCTL_R128_INIT, (unsigned long)init); +} + +typedef struct drm_r128_depth32 { + int func; + int n; + u32 x; + u32 y; + u32 buffer; + u32 mask; +} drm_r128_depth32_t; + +static int compat_r128_depth(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_r128_depth32_t depth32; + drm_r128_depth_t __user *depth; + + if (copy_from_user(&depth32, (void __user *)arg, sizeof(depth32))) + return -EFAULT; + + depth = compat_alloc_user_space(sizeof(*depth)); + if (!access_ok(VERIFY_WRITE, depth, sizeof(*depth)) + || __put_user(depth32.func, &depth->func) + || __put_user(depth32.n, &depth->n) + || __put_user((int __user *)(unsigned long)depth32.x, &depth->x) + || __put_user((int __user *)(unsigned long)depth32.y, &depth->y) + || __put_user((unsigned int __user *)(unsigned long)depth32.buffer, + &depth->buffer) + || __put_user((unsigned char __user *)(unsigned long)depth32.mask, + &depth->mask)) + return -EFAULT; + + return drm_ioctl(file, DRM_IOCTL_R128_DEPTH, (unsigned long)depth); + +} + +typedef struct drm_r128_stipple32 { + u32 mask; +} drm_r128_stipple32_t; + +static int compat_r128_stipple(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_r128_stipple32_t stipple32; + drm_r128_stipple_t __user *stipple; + + if (copy_from_user(&stipple32, (void __user *)arg, sizeof(stipple32))) + return -EFAULT; + + stipple = compat_alloc_user_space(sizeof(*stipple)); + if (!access_ok(VERIFY_WRITE, stipple, sizeof(*stipple)) + || __put_user((unsigned int __user *)(unsigned long)stipple32.mask, + &stipple->mask)) + return -EFAULT; + + return drm_ioctl(file, DRM_IOCTL_R128_STIPPLE, (unsigned long)stipple); +} + +typedef struct drm_r128_getparam32 { + int param; + u32 value; +} drm_r128_getparam32_t; + +static int compat_r128_getparam(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_r128_getparam32_t getparam32; + drm_r128_getparam_t __user *getparam; + + if (copy_from_user(&getparam32, (void __user *)arg, sizeof(getparam32))) + return -EFAULT; + + getparam = compat_alloc_user_space(sizeof(*getparam)); + if (!access_ok(VERIFY_WRITE, getparam, sizeof(*getparam)) + || __put_user(getparam32.param, &getparam->param) + || __put_user((void __user *)(unsigned long)getparam32.value, + &getparam->value)) + return -EFAULT; + + return drm_ioctl(file, DRM_IOCTL_R128_GETPARAM, (unsigned long)getparam); +} + +drm_ioctl_compat_t *r128_compat_ioctls[] = { + [DRM_R128_INIT] = compat_r128_init, + [DRM_R128_DEPTH] = compat_r128_depth, + [DRM_R128_STIPPLE] = compat_r128_stipple, + [DRM_R128_GETPARAM] = compat_r128_getparam, +}; + +/** + * Called whenever a 32-bit process running under a 64-bit kernel + * performs an ioctl on /dev/dri/card. + * + * \param filp file pointer. + * \param cmd command. + * \param arg user argument. + * \return zero on success or negative number on failure. + */ +long r128_compat_ioctl(struct file *filp, unsigned int cmd, unsigned long arg) +{ + unsigned int nr = DRM_IOCTL_NR(cmd); + drm_ioctl_compat_t *fn = NULL; + int ret; + + if (nr < DRM_COMMAND_BASE) + return drm_compat_ioctl(filp, cmd, arg); + + if (nr < DRM_COMMAND_BASE + DRM_ARRAY_SIZE(r128_compat_ioctls)) + fn = r128_compat_ioctls[nr - DRM_COMMAND_BASE]; + + if (fn != NULL) + ret = (*fn) (filp, cmd, arg); + else + ret = drm_ioctl(filp, cmd, arg); + + return ret; +} diff --git a/drivers/gpu/drm/r128/r128_irq.c b/drivers/gpu/drm/r128/r128_irq.c new file mode 100644 index 00000000..429d5a02 --- /dev/null +++ b/drivers/gpu/drm/r128/r128_irq.c @@ -0,0 +1,116 @@ +/* r128_irq.c -- IRQ handling for radeon -*- linux-c -*- */ +/* + * Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + * + * The Weather Channel (TM) funded Tungsten Graphics to develop the + * initial release of the Radeon 8500 driver under the XFree86 license. + * This notice must be preserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Keith Whitwell + * Eric Anholt + */ + +#include "drmP.h" +#include "drm.h" +#include "r128_drm.h" +#include "r128_drv.h" + +u32 r128_get_vblank_counter(struct drm_device *dev, int crtc) +{ + const drm_r128_private_t *dev_priv = dev->dev_private; + + if (crtc != 0) + return 0; + + return atomic_read(&dev_priv->vbl_received); +} + +irqreturn_t r128_driver_irq_handler(DRM_IRQ_ARGS) +{ + struct drm_device *dev = (struct drm_device *) arg; + drm_r128_private_t *dev_priv = (drm_r128_private_t *) dev->dev_private; + int status; + + status = R128_READ(R128_GEN_INT_STATUS); + + /* VBLANK interrupt */ + if (status & R128_CRTC_VBLANK_INT) { + R128_WRITE(R128_GEN_INT_STATUS, R128_CRTC_VBLANK_INT_AK); + atomic_inc(&dev_priv->vbl_received); + drm_handle_vblank(dev, 0); + return IRQ_HANDLED; + } + return IRQ_NONE; +} + +int r128_enable_vblank(struct drm_device *dev, int crtc) +{ + drm_r128_private_t *dev_priv = dev->dev_private; + + if (crtc != 0) { + DRM_ERROR("%s: bad crtc %d\n", __func__, crtc); + return -EINVAL; + } + + R128_WRITE(R128_GEN_INT_CNTL, R128_CRTC_VBLANK_INT_EN); + return 0; +} + +void r128_disable_vblank(struct drm_device *dev, int crtc) +{ + if (crtc != 0) + DRM_ERROR("%s: bad crtc %d\n", __func__, crtc); + + /* + * FIXME: implement proper interrupt disable by using the vblank + * counter register (if available) + * + * R128_WRITE(R128_GEN_INT_CNTL, + * R128_READ(R128_GEN_INT_CNTL) & ~R128_CRTC_VBLANK_INT_EN); + */ +} + +void r128_driver_irq_preinstall(struct drm_device *dev) +{ + drm_r128_private_t *dev_priv = (drm_r128_private_t *) dev->dev_private; + + /* Disable *all* interrupts */ + R128_WRITE(R128_GEN_INT_CNTL, 0); + /* Clear vblank bit if it's already high */ + R128_WRITE(R128_GEN_INT_STATUS, R128_CRTC_VBLANK_INT_AK); +} + +int r128_driver_irq_postinstall(struct drm_device *dev) +{ + return 0; +} + +void r128_driver_irq_uninstall(struct drm_device *dev) +{ + drm_r128_private_t *dev_priv = (drm_r128_private_t *) dev->dev_private; + if (!dev_priv) + return; + + /* Disable *all* interrupts */ + R128_WRITE(R128_GEN_INT_CNTL, 0); +} diff --git a/drivers/gpu/drm/r128/r128_state.c b/drivers/gpu/drm/r128/r128_state.c new file mode 100644 index 00000000..a9e33ce6 --- /dev/null +++ b/drivers/gpu/drm/r128/r128_state.c @@ -0,0 +1,1667 @@ +/* r128_state.c -- State support for r128 -*- linux-c -*- + * Created: Thu Jan 27 02:53:43 2000 by gareth@valinux.com + */ +/* + * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Gareth Hughes + */ + +#include "drmP.h" +#include "drm.h" +#include "r128_drm.h" +#include "r128_drv.h" + +/* ================================================================ + * CCE hardware state programming functions + */ + +static void r128_emit_clip_rects(drm_r128_private_t *dev_priv, + struct drm_clip_rect *boxes, int count) +{ + u32 aux_sc_cntl = 0x00000000; + RING_LOCALS; + DRM_DEBUG("\n"); + + BEGIN_RING((count < 3 ? count : 3) * 5 + 2); + + if (count >= 1) { + OUT_RING(CCE_PACKET0(R128_AUX1_SC_LEFT, 3)); + OUT_RING(boxes[0].x1); + OUT_RING(boxes[0].x2 - 1); + OUT_RING(boxes[0].y1); + OUT_RING(boxes[0].y2 - 1); + + aux_sc_cntl |= (R128_AUX1_SC_EN | R128_AUX1_SC_MODE_OR); + } + if (count >= 2) { + OUT_RING(CCE_PACKET0(R128_AUX2_SC_LEFT, 3)); + OUT_RING(boxes[1].x1); + OUT_RING(boxes[1].x2 - 1); + OUT_RING(boxes[1].y1); + OUT_RING(boxes[1].y2 - 1); + + aux_sc_cntl |= (R128_AUX2_SC_EN | R128_AUX2_SC_MODE_OR); + } + if (count >= 3) { + OUT_RING(CCE_PACKET0(R128_AUX3_SC_LEFT, 3)); + OUT_RING(boxes[2].x1); + OUT_RING(boxes[2].x2 - 1); + OUT_RING(boxes[2].y1); + OUT_RING(boxes[2].y2 - 1); + + aux_sc_cntl |= (R128_AUX3_SC_EN | R128_AUX3_SC_MODE_OR); + } + + OUT_RING(CCE_PACKET0(R128_AUX_SC_CNTL, 0)); + OUT_RING(aux_sc_cntl); + + ADVANCE_RING(); +} + +static __inline__ void r128_emit_core(drm_r128_private_t *dev_priv) +{ + drm_r128_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_r128_context_regs_t *ctx = &sarea_priv->context_state; + RING_LOCALS; + DRM_DEBUG("\n"); + + BEGIN_RING(2); + + OUT_RING(CCE_PACKET0(R128_SCALE_3D_CNTL, 0)); + OUT_RING(ctx->scale_3d_cntl); + + ADVANCE_RING(); +} + +static __inline__ void r128_emit_context(drm_r128_private_t *dev_priv) +{ + drm_r128_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_r128_context_regs_t *ctx = &sarea_priv->context_state; + RING_LOCALS; + DRM_DEBUG("\n"); + + BEGIN_RING(13); + + OUT_RING(CCE_PACKET0(R128_DST_PITCH_OFFSET_C, 11)); + OUT_RING(ctx->dst_pitch_offset_c); + OUT_RING(ctx->dp_gui_master_cntl_c); + OUT_RING(ctx->sc_top_left_c); + OUT_RING(ctx->sc_bottom_right_c); + OUT_RING(ctx->z_offset_c); + OUT_RING(ctx->z_pitch_c); + OUT_RING(ctx->z_sten_cntl_c); + OUT_RING(ctx->tex_cntl_c); + OUT_RING(ctx->misc_3d_state_cntl_reg); + OUT_RING(ctx->texture_clr_cmp_clr_c); + OUT_RING(ctx->texture_clr_cmp_msk_c); + OUT_RING(ctx->fog_color_c); + + ADVANCE_RING(); +} + +static __inline__ void r128_emit_setup(drm_r128_private_t *dev_priv) +{ + drm_r128_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_r128_context_regs_t *ctx = &sarea_priv->context_state; + RING_LOCALS; + DRM_DEBUG("\n"); + + BEGIN_RING(3); + + OUT_RING(CCE_PACKET1(R128_SETUP_CNTL, R128_PM4_VC_FPU_SETUP)); + OUT_RING(ctx->setup_cntl); + OUT_RING(ctx->pm4_vc_fpu_setup); + + ADVANCE_RING(); +} + +static __inline__ void r128_emit_masks(drm_r128_private_t *dev_priv) +{ + drm_r128_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_r128_context_regs_t *ctx = &sarea_priv->context_state; + RING_LOCALS; + DRM_DEBUG("\n"); + + BEGIN_RING(5); + + OUT_RING(CCE_PACKET0(R128_DP_WRITE_MASK, 0)); + OUT_RING(ctx->dp_write_mask); + + OUT_RING(CCE_PACKET0(R128_STEN_REF_MASK_C, 1)); + OUT_RING(ctx->sten_ref_mask_c); + OUT_RING(ctx->plane_3d_mask_c); + + ADVANCE_RING(); +} + +static __inline__ void r128_emit_window(drm_r128_private_t *dev_priv) +{ + drm_r128_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_r128_context_regs_t *ctx = &sarea_priv->context_state; + RING_LOCALS; + DRM_DEBUG("\n"); + + BEGIN_RING(2); + + OUT_RING(CCE_PACKET0(R128_WINDOW_XY_OFFSET, 0)); + OUT_RING(ctx->window_xy_offset); + + ADVANCE_RING(); +} + +static __inline__ void r128_emit_tex0(drm_r128_private_t *dev_priv) +{ + drm_r128_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_r128_context_regs_t *ctx = &sarea_priv->context_state; + drm_r128_texture_regs_t *tex = &sarea_priv->tex_state[0]; + int i; + RING_LOCALS; + DRM_DEBUG("\n"); + + BEGIN_RING(7 + R128_MAX_TEXTURE_LEVELS); + + OUT_RING(CCE_PACKET0(R128_PRIM_TEX_CNTL_C, + 2 + R128_MAX_TEXTURE_LEVELS)); + OUT_RING(tex->tex_cntl); + OUT_RING(tex->tex_combine_cntl); + OUT_RING(ctx->tex_size_pitch_c); + for (i = 0; i < R128_MAX_TEXTURE_LEVELS; i++) + OUT_RING(tex->tex_offset[i]); + + OUT_RING(CCE_PACKET0(R128_CONSTANT_COLOR_C, 1)); + OUT_RING(ctx->constant_color_c); + OUT_RING(tex->tex_border_color); + + ADVANCE_RING(); +} + +static __inline__ void r128_emit_tex1(drm_r128_private_t *dev_priv) +{ + drm_r128_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_r128_texture_regs_t *tex = &sarea_priv->tex_state[1]; + int i; + RING_LOCALS; + DRM_DEBUG("\n"); + + BEGIN_RING(5 + R128_MAX_TEXTURE_LEVELS); + + OUT_RING(CCE_PACKET0(R128_SEC_TEX_CNTL_C, 1 + R128_MAX_TEXTURE_LEVELS)); + OUT_RING(tex->tex_cntl); + OUT_RING(tex->tex_combine_cntl); + for (i = 0; i < R128_MAX_TEXTURE_LEVELS; i++) + OUT_RING(tex->tex_offset[i]); + + OUT_RING(CCE_PACKET0(R128_SEC_TEXTURE_BORDER_COLOR_C, 0)); + OUT_RING(tex->tex_border_color); + + ADVANCE_RING(); +} + +static void r128_emit_state(drm_r128_private_t *dev_priv) +{ + drm_r128_sarea_t *sarea_priv = dev_priv->sarea_priv; + unsigned int dirty = sarea_priv->dirty; + + DRM_DEBUG("dirty=0x%08x\n", dirty); + + if (dirty & R128_UPLOAD_CORE) { + r128_emit_core(dev_priv); + sarea_priv->dirty &= ~R128_UPLOAD_CORE; + } + + if (dirty & R128_UPLOAD_CONTEXT) { + r128_emit_context(dev_priv); + sarea_priv->dirty &= ~R128_UPLOAD_CONTEXT; + } + + if (dirty & R128_UPLOAD_SETUP) { + r128_emit_setup(dev_priv); + sarea_priv->dirty &= ~R128_UPLOAD_SETUP; + } + + if (dirty & R128_UPLOAD_MASKS) { + r128_emit_masks(dev_priv); + sarea_priv->dirty &= ~R128_UPLOAD_MASKS; + } + + if (dirty & R128_UPLOAD_WINDOW) { + r128_emit_window(dev_priv); + sarea_priv->dirty &= ~R128_UPLOAD_WINDOW; + } + + if (dirty & R128_UPLOAD_TEX0) { + r128_emit_tex0(dev_priv); + sarea_priv->dirty &= ~R128_UPLOAD_TEX0; + } + + if (dirty & R128_UPLOAD_TEX1) { + r128_emit_tex1(dev_priv); + sarea_priv->dirty &= ~R128_UPLOAD_TEX1; + } + + /* Turn off the texture cache flushing */ + sarea_priv->context_state.tex_cntl_c &= ~R128_TEX_CACHE_FLUSH; + + sarea_priv->dirty &= ~R128_REQUIRE_QUIESCENCE; +} + +#if R128_PERFORMANCE_BOXES +/* ================================================================ + * Performance monitoring functions + */ + +static void r128_clear_box(drm_r128_private_t *dev_priv, + int x, int y, int w, int h, int r, int g, int b) +{ + u32 pitch, offset; + u32 fb_bpp, color; + RING_LOCALS; + + switch (dev_priv->fb_bpp) { + case 16: + fb_bpp = R128_GMC_DST_16BPP; + color = (((r & 0xf8) << 8) | + ((g & 0xfc) << 3) | ((b & 0xf8) >> 3)); + break; + case 24: + fb_bpp = R128_GMC_DST_24BPP; + color = ((r << 16) | (g << 8) | b); + break; + case 32: + fb_bpp = R128_GMC_DST_32BPP; + color = (((0xff) << 24) | (r << 16) | (g << 8) | b); + break; + default: + return; + } + + offset = dev_priv->back_offset; + pitch = dev_priv->back_pitch >> 3; + + BEGIN_RING(6); + + OUT_RING(CCE_PACKET3(R128_CNTL_PAINT_MULTI, 4)); + OUT_RING(R128_GMC_DST_PITCH_OFFSET_CNTL | + R128_GMC_BRUSH_SOLID_COLOR | + fb_bpp | + R128_GMC_SRC_DATATYPE_COLOR | + R128_ROP3_P | + R128_GMC_CLR_CMP_CNTL_DIS | R128_GMC_AUX_CLIP_DIS); + + OUT_RING((pitch << 21) | (offset >> 5)); + OUT_RING(color); + + OUT_RING((x << 16) | y); + OUT_RING((w << 16) | h); + + ADVANCE_RING(); +} + +static void r128_cce_performance_boxes(drm_r128_private_t *dev_priv) +{ + if (atomic_read(&dev_priv->idle_count) == 0) + r128_clear_box(dev_priv, 64, 4, 8, 8, 0, 255, 0); + else + atomic_set(&dev_priv->idle_count, 0); +} + +#endif + +/* ================================================================ + * CCE command dispatch functions + */ + +static void r128_print_dirty(const char *msg, unsigned int flags) +{ + DRM_INFO("%s: (0x%x) %s%s%s%s%s%s%s%s%s\n", + msg, + flags, + (flags & R128_UPLOAD_CORE) ? "core, " : "", + (flags & R128_UPLOAD_CONTEXT) ? "context, " : "", + (flags & R128_UPLOAD_SETUP) ? "setup, " : "", + (flags & R128_UPLOAD_TEX0) ? "tex0, " : "", + (flags & R128_UPLOAD_TEX1) ? "tex1, " : "", + (flags & R128_UPLOAD_MASKS) ? "masks, " : "", + (flags & R128_UPLOAD_WINDOW) ? "window, " : "", + (flags & R128_UPLOAD_CLIPRECTS) ? "cliprects, " : "", + (flags & R128_REQUIRE_QUIESCENCE) ? "quiescence, " : ""); +} + +static void r128_cce_dispatch_clear(struct drm_device *dev, + drm_r128_clear_t *clear) +{ + drm_r128_private_t *dev_priv = dev->dev_private; + drm_r128_sarea_t *sarea_priv = dev_priv->sarea_priv; + int nbox = sarea_priv->nbox; + struct drm_clip_rect *pbox = sarea_priv->boxes; + unsigned int flags = clear->flags; + int i; + RING_LOCALS; + DRM_DEBUG("\n"); + + if (dev_priv->page_flipping && dev_priv->current_page == 1) { + unsigned int tmp = flags; + + flags &= ~(R128_FRONT | R128_BACK); + if (tmp & R128_FRONT) + flags |= R128_BACK; + if (tmp & R128_BACK) + flags |= R128_FRONT; + } + + for (i = 0; i < nbox; i++) { + int x = pbox[i].x1; + int y = pbox[i].y1; + int w = pbox[i].x2 - x; + int h = pbox[i].y2 - y; + + DRM_DEBUG("dispatch clear %d,%d-%d,%d flags 0x%x\n", + pbox[i].x1, pbox[i].y1, pbox[i].x2, + pbox[i].y2, flags); + + if (flags & (R128_FRONT | R128_BACK)) { + BEGIN_RING(2); + + OUT_RING(CCE_PACKET0(R128_DP_WRITE_MASK, 0)); + OUT_RING(clear->color_mask); + + ADVANCE_RING(); + } + + if (flags & R128_FRONT) { + BEGIN_RING(6); + + OUT_RING(CCE_PACKET3(R128_CNTL_PAINT_MULTI, 4)); + OUT_RING(R128_GMC_DST_PITCH_OFFSET_CNTL | + R128_GMC_BRUSH_SOLID_COLOR | + (dev_priv->color_fmt << 8) | + R128_GMC_SRC_DATATYPE_COLOR | + R128_ROP3_P | + R128_GMC_CLR_CMP_CNTL_DIS | + R128_GMC_AUX_CLIP_DIS); + + OUT_RING(dev_priv->front_pitch_offset_c); + OUT_RING(clear->clear_color); + + OUT_RING((x << 16) | y); + OUT_RING((w << 16) | h); + + ADVANCE_RING(); + } + + if (flags & R128_BACK) { + BEGIN_RING(6); + + OUT_RING(CCE_PACKET3(R128_CNTL_PAINT_MULTI, 4)); + OUT_RING(R128_GMC_DST_PITCH_OFFSET_CNTL | + R128_GMC_BRUSH_SOLID_COLOR | + (dev_priv->color_fmt << 8) | + R128_GMC_SRC_DATATYPE_COLOR | + R128_ROP3_P | + R128_GMC_CLR_CMP_CNTL_DIS | + R128_GMC_AUX_CLIP_DIS); + + OUT_RING(dev_priv->back_pitch_offset_c); + OUT_RING(clear->clear_color); + + OUT_RING((x << 16) | y); + OUT_RING((w << 16) | h); + + ADVANCE_RING(); + } + + if (flags & R128_DEPTH) { + BEGIN_RING(6); + + OUT_RING(CCE_PACKET3(R128_CNTL_PAINT_MULTI, 4)); + OUT_RING(R128_GMC_DST_PITCH_OFFSET_CNTL | + R128_GMC_BRUSH_SOLID_COLOR | + (dev_priv->depth_fmt << 8) | + R128_GMC_SRC_DATATYPE_COLOR | + R128_ROP3_P | + R128_GMC_CLR_CMP_CNTL_DIS | + R128_GMC_AUX_CLIP_DIS | R128_GMC_WR_MSK_DIS); + + OUT_RING(dev_priv->depth_pitch_offset_c); + OUT_RING(clear->clear_depth); + + OUT_RING((x << 16) | y); + OUT_RING((w << 16) | h); + + ADVANCE_RING(); + } + } +} + +static void r128_cce_dispatch_swap(struct drm_device *dev) +{ + drm_r128_private_t *dev_priv = dev->dev_private; + drm_r128_sarea_t *sarea_priv = dev_priv->sarea_priv; + int nbox = sarea_priv->nbox; + struct drm_clip_rect *pbox = sarea_priv->boxes; + int i; + RING_LOCALS; + DRM_DEBUG("\n"); + +#if R128_PERFORMANCE_BOXES + /* Do some trivial performance monitoring... + */ + r128_cce_performance_boxes(dev_priv); +#endif + + for (i = 0; i < nbox; i++) { + int x = pbox[i].x1; + int y = pbox[i].y1; + int w = pbox[i].x2 - x; + int h = pbox[i].y2 - y; + + BEGIN_RING(7); + + OUT_RING(CCE_PACKET3(R128_CNTL_BITBLT_MULTI, 5)); + OUT_RING(R128_GMC_SRC_PITCH_OFFSET_CNTL | + R128_GMC_DST_PITCH_OFFSET_CNTL | + R128_GMC_BRUSH_NONE | + (dev_priv->color_fmt << 8) | + R128_GMC_SRC_DATATYPE_COLOR | + R128_ROP3_S | + R128_DP_SRC_SOURCE_MEMORY | + R128_GMC_CLR_CMP_CNTL_DIS | + R128_GMC_AUX_CLIP_DIS | R128_GMC_WR_MSK_DIS); + + /* Make this work even if front & back are flipped: + */ + if (dev_priv->current_page == 0) { + OUT_RING(dev_priv->back_pitch_offset_c); + OUT_RING(dev_priv->front_pitch_offset_c); + } else { + OUT_RING(dev_priv->front_pitch_offset_c); + OUT_RING(dev_priv->back_pitch_offset_c); + } + + OUT_RING((x << 16) | y); + OUT_RING((x << 16) | y); + OUT_RING((w << 16) | h); + + ADVANCE_RING(); + } + + /* Increment the frame counter. The client-side 3D driver must + * throttle the framerate by waiting for this value before + * performing the swapbuffer ioctl. + */ + dev_priv->sarea_priv->last_frame++; + + BEGIN_RING(2); + + OUT_RING(CCE_PACKET0(R128_LAST_FRAME_REG, 0)); + OUT_RING(dev_priv->sarea_priv->last_frame); + + ADVANCE_RING(); +} + +static void r128_cce_dispatch_flip(struct drm_device *dev) +{ + drm_r128_private_t *dev_priv = dev->dev_private; + RING_LOCALS; + DRM_DEBUG("page=%d pfCurrentPage=%d\n", + dev_priv->current_page, dev_priv->sarea_priv->pfCurrentPage); + +#if R128_PERFORMANCE_BOXES + /* Do some trivial performance monitoring... + */ + r128_cce_performance_boxes(dev_priv); +#endif + + BEGIN_RING(4); + + R128_WAIT_UNTIL_PAGE_FLIPPED(); + OUT_RING(CCE_PACKET0(R128_CRTC_OFFSET, 0)); + + if (dev_priv->current_page == 0) + OUT_RING(dev_priv->back_offset); + else + OUT_RING(dev_priv->front_offset); + + ADVANCE_RING(); + + /* Increment the frame counter. The client-side 3D driver must + * throttle the framerate by waiting for this value before + * performing the swapbuffer ioctl. + */ + dev_priv->sarea_priv->last_frame++; + dev_priv->sarea_priv->pfCurrentPage = dev_priv->current_page = + 1 - dev_priv->current_page; + + BEGIN_RING(2); + + OUT_RING(CCE_PACKET0(R128_LAST_FRAME_REG, 0)); + OUT_RING(dev_priv->sarea_priv->last_frame); + + ADVANCE_RING(); +} + +static void r128_cce_dispatch_vertex(struct drm_device *dev, struct drm_buf *buf) +{ + drm_r128_private_t *dev_priv = dev->dev_private; + drm_r128_buf_priv_t *buf_priv = buf->dev_private; + drm_r128_sarea_t *sarea_priv = dev_priv->sarea_priv; + int format = sarea_priv->vc_format; + int offset = buf->bus_address; + int size = buf->used; + int prim = buf_priv->prim; + int i = 0; + RING_LOCALS; + DRM_DEBUG("buf=%d nbox=%d\n", buf->idx, sarea_priv->nbox); + + if (0) + r128_print_dirty("dispatch_vertex", sarea_priv->dirty); + + if (buf->used) { + buf_priv->dispatched = 1; + + if (sarea_priv->dirty & ~R128_UPLOAD_CLIPRECTS) + r128_emit_state(dev_priv); + + do { + /* Emit the next set of up to three cliprects */ + if (i < sarea_priv->nbox) { + r128_emit_clip_rects(dev_priv, + &sarea_priv->boxes[i], + sarea_priv->nbox - i); + } + + /* Emit the vertex buffer rendering commands */ + BEGIN_RING(5); + + OUT_RING(CCE_PACKET3(R128_3D_RNDR_GEN_INDX_PRIM, 3)); + OUT_RING(offset); + OUT_RING(size); + OUT_RING(format); + OUT_RING(prim | R128_CCE_VC_CNTL_PRIM_WALK_LIST | + (size << R128_CCE_VC_CNTL_NUM_SHIFT)); + + ADVANCE_RING(); + + i += 3; + } while (i < sarea_priv->nbox); + } + + if (buf_priv->discard) { + buf_priv->age = dev_priv->sarea_priv->last_dispatch; + + /* Emit the vertex buffer age */ + BEGIN_RING(2); + + OUT_RING(CCE_PACKET0(R128_LAST_DISPATCH_REG, 0)); + OUT_RING(buf_priv->age); + + ADVANCE_RING(); + + buf->pending = 1; + buf->used = 0; + /* FIXME: Check dispatched field */ + buf_priv->dispatched = 0; + } + + dev_priv->sarea_priv->last_dispatch++; + + sarea_priv->dirty &= ~R128_UPLOAD_CLIPRECTS; + sarea_priv->nbox = 0; +} + +static void r128_cce_dispatch_indirect(struct drm_device *dev, + struct drm_buf *buf, int start, int end) +{ + drm_r128_private_t *dev_priv = dev->dev_private; + drm_r128_buf_priv_t *buf_priv = buf->dev_private; + RING_LOCALS; + DRM_DEBUG("indirect: buf=%d s=0x%x e=0x%x\n", buf->idx, start, end); + + if (start != end) { + int offset = buf->bus_address + start; + int dwords = (end - start + 3) / sizeof(u32); + + /* Indirect buffer data must be an even number of + * dwords, so if we've been given an odd number we must + * pad the data with a Type-2 CCE packet. + */ + if (dwords & 1) { + u32 *data = (u32 *) + ((char *)dev->agp_buffer_map->handle + + buf->offset + start); + data[dwords++] = cpu_to_le32(R128_CCE_PACKET2); + } + + buf_priv->dispatched = 1; + + /* Fire off the indirect buffer */ + BEGIN_RING(3); + + OUT_RING(CCE_PACKET0(R128_PM4_IW_INDOFF, 1)); + OUT_RING(offset); + OUT_RING(dwords); + + ADVANCE_RING(); + } + + if (buf_priv->discard) { + buf_priv->age = dev_priv->sarea_priv->last_dispatch; + + /* Emit the indirect buffer age */ + BEGIN_RING(2); + + OUT_RING(CCE_PACKET0(R128_LAST_DISPATCH_REG, 0)); + OUT_RING(buf_priv->age); + + ADVANCE_RING(); + + buf->pending = 1; + buf->used = 0; + /* FIXME: Check dispatched field */ + buf_priv->dispatched = 0; + } + + dev_priv->sarea_priv->last_dispatch++; +} + +static void r128_cce_dispatch_indices(struct drm_device *dev, + struct drm_buf *buf, + int start, int end, int count) +{ + drm_r128_private_t *dev_priv = dev->dev_private; + drm_r128_buf_priv_t *buf_priv = buf->dev_private; + drm_r128_sarea_t *sarea_priv = dev_priv->sarea_priv; + int format = sarea_priv->vc_format; + int offset = dev->agp_buffer_map->offset - dev_priv->cce_buffers_offset; + int prim = buf_priv->prim; + u32 *data; + int dwords; + int i = 0; + RING_LOCALS; + DRM_DEBUG("indices: s=%d e=%d c=%d\n", start, end, count); + + if (0) + r128_print_dirty("dispatch_indices", sarea_priv->dirty); + + if (start != end) { + buf_priv->dispatched = 1; + + if (sarea_priv->dirty & ~R128_UPLOAD_CLIPRECTS) + r128_emit_state(dev_priv); + + dwords = (end - start + 3) / sizeof(u32); + + data = (u32 *) ((char *)dev->agp_buffer_map->handle + + buf->offset + start); + + data[0] = cpu_to_le32(CCE_PACKET3(R128_3D_RNDR_GEN_INDX_PRIM, + dwords - 2)); + + data[1] = cpu_to_le32(offset); + data[2] = cpu_to_le32(R128_MAX_VB_VERTS); + data[3] = cpu_to_le32(format); + data[4] = cpu_to_le32((prim | R128_CCE_VC_CNTL_PRIM_WALK_IND | + (count << 16))); + + if (count & 0x1) { +#ifdef __LITTLE_ENDIAN + data[dwords - 1] &= 0x0000ffff; +#else + data[dwords - 1] &= 0xffff0000; +#endif + } + + do { + /* Emit the next set of up to three cliprects */ + if (i < sarea_priv->nbox) { + r128_emit_clip_rects(dev_priv, + &sarea_priv->boxes[i], + sarea_priv->nbox - i); + } + + r128_cce_dispatch_indirect(dev, buf, start, end); + + i += 3; + } while (i < sarea_priv->nbox); + } + + if (buf_priv->discard) { + buf_priv->age = dev_priv->sarea_priv->last_dispatch; + + /* Emit the vertex buffer age */ + BEGIN_RING(2); + + OUT_RING(CCE_PACKET0(R128_LAST_DISPATCH_REG, 0)); + OUT_RING(buf_priv->age); + + ADVANCE_RING(); + + buf->pending = 1; + /* FIXME: Check dispatched field */ + buf_priv->dispatched = 0; + } + + dev_priv->sarea_priv->last_dispatch++; + + sarea_priv->dirty &= ~R128_UPLOAD_CLIPRECTS; + sarea_priv->nbox = 0; +} + +static int r128_cce_dispatch_blit(struct drm_device *dev, + struct drm_file *file_priv, + drm_r128_blit_t *blit) +{ + drm_r128_private_t *dev_priv = dev->dev_private; + struct drm_device_dma *dma = dev->dma; + struct drm_buf *buf; + drm_r128_buf_priv_t *buf_priv; + u32 *data; + int dword_shift, dwords; + RING_LOCALS; + DRM_DEBUG("\n"); + + /* The compiler won't optimize away a division by a variable, + * even if the only legal values are powers of two. Thus, we'll + * use a shift instead. + */ + switch (blit->format) { + case R128_DATATYPE_ARGB8888: + dword_shift = 0; + break; + case R128_DATATYPE_ARGB1555: + case R128_DATATYPE_RGB565: + case R128_DATATYPE_ARGB4444: + case R128_DATATYPE_YVYU422: + case R128_DATATYPE_VYUY422: + dword_shift = 1; + break; + case R128_DATATYPE_CI8: + case R128_DATATYPE_RGB8: + dword_shift = 2; + break; + default: + DRM_ERROR("invalid blit format %d\n", blit->format); + return -EINVAL; + } + + /* Flush the pixel cache, and mark the contents as Read Invalid. + * This ensures no pixel data gets mixed up with the texture + * data from the host data blit, otherwise part of the texture + * image may be corrupted. + */ + BEGIN_RING(2); + + OUT_RING(CCE_PACKET0(R128_PC_GUI_CTLSTAT, 0)); + OUT_RING(R128_PC_RI_GUI | R128_PC_FLUSH_GUI); + + ADVANCE_RING(); + + /* Dispatch the indirect buffer. + */ + buf = dma->buflist[blit->idx]; + buf_priv = buf->dev_private; + + if (buf->file_priv != file_priv) { + DRM_ERROR("process %d using buffer owned by %p\n", + DRM_CURRENTPID, buf->file_priv); + return -EINVAL; + } + if (buf->pending) { + DRM_ERROR("sending pending buffer %d\n", blit->idx); + return -EINVAL; + } + + buf_priv->discard = 1; + + dwords = (blit->width * blit->height) >> dword_shift; + + data = (u32 *) ((char *)dev->agp_buffer_map->handle + buf->offset); + + data[0] = cpu_to_le32(CCE_PACKET3(R128_CNTL_HOSTDATA_BLT, dwords + 6)); + data[1] = cpu_to_le32((R128_GMC_DST_PITCH_OFFSET_CNTL | + R128_GMC_BRUSH_NONE | + (blit->format << 8) | + R128_GMC_SRC_DATATYPE_COLOR | + R128_ROP3_S | + R128_DP_SRC_SOURCE_HOST_DATA | + R128_GMC_CLR_CMP_CNTL_DIS | + R128_GMC_AUX_CLIP_DIS | R128_GMC_WR_MSK_DIS)); + + data[2] = cpu_to_le32((blit->pitch << 21) | (blit->offset >> 5)); + data[3] = cpu_to_le32(0xffffffff); + data[4] = cpu_to_le32(0xffffffff); + data[5] = cpu_to_le32((blit->y << 16) | blit->x); + data[6] = cpu_to_le32((blit->height << 16) | blit->width); + data[7] = cpu_to_le32(dwords); + + buf->used = (dwords + 8) * sizeof(u32); + + r128_cce_dispatch_indirect(dev, buf, 0, buf->used); + + /* Flush the pixel cache after the blit completes. This ensures + * the texture data is written out to memory before rendering + * continues. + */ + BEGIN_RING(2); + + OUT_RING(CCE_PACKET0(R128_PC_GUI_CTLSTAT, 0)); + OUT_RING(R128_PC_FLUSH_GUI); + + ADVANCE_RING(); + + return 0; +} + +/* ================================================================ + * Tiled depth buffer management + * + * FIXME: These should all set the destination write mask for when we + * have hardware stencil support. + */ + +static int r128_cce_dispatch_write_span(struct drm_device *dev, + drm_r128_depth_t *depth) +{ + drm_r128_private_t *dev_priv = dev->dev_private; + int count, x, y; + u32 *buffer; + u8 *mask; + int i, buffer_size, mask_size; + RING_LOCALS; + DRM_DEBUG("\n"); + + count = depth->n; + if (count > 4096 || count <= 0) + return -EMSGSIZE; + + if (DRM_COPY_FROM_USER(&x, depth->x, sizeof(x))) + return -EFAULT; + if (DRM_COPY_FROM_USER(&y, depth->y, sizeof(y))) + return -EFAULT; + + buffer_size = depth->n * sizeof(u32); + buffer = kmalloc(buffer_size, GFP_KERNEL); + if (buffer == NULL) + return -ENOMEM; + if (DRM_COPY_FROM_USER(buffer, depth->buffer, buffer_size)) { + kfree(buffer); + return -EFAULT; + } + + mask_size = depth->n * sizeof(u8); + if (depth->mask) { + mask = kmalloc(mask_size, GFP_KERNEL); + if (mask == NULL) { + kfree(buffer); + return -ENOMEM; + } + if (DRM_COPY_FROM_USER(mask, depth->mask, mask_size)) { + kfree(buffer); + kfree(mask); + return -EFAULT; + } + + for (i = 0; i < count; i++, x++) { + if (mask[i]) { + BEGIN_RING(6); + + OUT_RING(CCE_PACKET3(R128_CNTL_PAINT_MULTI, 4)); + OUT_RING(R128_GMC_DST_PITCH_OFFSET_CNTL | + R128_GMC_BRUSH_SOLID_COLOR | + (dev_priv->depth_fmt << 8) | + R128_GMC_SRC_DATATYPE_COLOR | + R128_ROP3_P | + R128_GMC_CLR_CMP_CNTL_DIS | + R128_GMC_WR_MSK_DIS); + + OUT_RING(dev_priv->depth_pitch_offset_c); + OUT_RING(buffer[i]); + + OUT_RING((x << 16) | y); + OUT_RING((1 << 16) | 1); + + ADVANCE_RING(); + } + } + + kfree(mask); + } else { + for (i = 0; i < count; i++, x++) { + BEGIN_RING(6); + + OUT_RING(CCE_PACKET3(R128_CNTL_PAINT_MULTI, 4)); + OUT_RING(R128_GMC_DST_PITCH_OFFSET_CNTL | + R128_GMC_BRUSH_SOLID_COLOR | + (dev_priv->depth_fmt << 8) | + R128_GMC_SRC_DATATYPE_COLOR | + R128_ROP3_P | + R128_GMC_CLR_CMP_CNTL_DIS | + R128_GMC_WR_MSK_DIS); + + OUT_RING(dev_priv->depth_pitch_offset_c); + OUT_RING(buffer[i]); + + OUT_RING((x << 16) | y); + OUT_RING((1 << 16) | 1); + + ADVANCE_RING(); + } + } + + kfree(buffer); + + return 0; +} + +static int r128_cce_dispatch_write_pixels(struct drm_device *dev, + drm_r128_depth_t *depth) +{ + drm_r128_private_t *dev_priv = dev->dev_private; + int count, *x, *y; + u32 *buffer; + u8 *mask; + int i, xbuf_size, ybuf_size, buffer_size, mask_size; + RING_LOCALS; + DRM_DEBUG("\n"); + + count = depth->n; + if (count > 4096 || count <= 0) + return -EMSGSIZE; + + xbuf_size = count * sizeof(*x); + ybuf_size = count * sizeof(*y); + x = kmalloc(xbuf_size, GFP_KERNEL); + if (x == NULL) + return -ENOMEM; + y = kmalloc(ybuf_size, GFP_KERNEL); + if (y == NULL) { + kfree(x); + return -ENOMEM; + } + if (DRM_COPY_FROM_USER(x, depth->x, xbuf_size)) { + kfree(x); + kfree(y); + return -EFAULT; + } + if (DRM_COPY_FROM_USER(y, depth->y, xbuf_size)) { + kfree(x); + kfree(y); + return -EFAULT; + } + + buffer_size = depth->n * sizeof(u32); + buffer = kmalloc(buffer_size, GFP_KERNEL); + if (buffer == NULL) { + kfree(x); + kfree(y); + return -ENOMEM; + } + if (DRM_COPY_FROM_USER(buffer, depth->buffer, buffer_size)) { + kfree(x); + kfree(y); + kfree(buffer); + return -EFAULT; + } + + if (depth->mask) { + mask_size = depth->n * sizeof(u8); + mask = kmalloc(mask_size, GFP_KERNEL); + if (mask == NULL) { + kfree(x); + kfree(y); + kfree(buffer); + return -ENOMEM; + } + if (DRM_COPY_FROM_USER(mask, depth->mask, mask_size)) { + kfree(x); + kfree(y); + kfree(buffer); + kfree(mask); + return -EFAULT; + } + + for (i = 0; i < count; i++) { + if (mask[i]) { + BEGIN_RING(6); + + OUT_RING(CCE_PACKET3(R128_CNTL_PAINT_MULTI, 4)); + OUT_RING(R128_GMC_DST_PITCH_OFFSET_CNTL | + R128_GMC_BRUSH_SOLID_COLOR | + (dev_priv->depth_fmt << 8) | + R128_GMC_SRC_DATATYPE_COLOR | + R128_ROP3_P | + R128_GMC_CLR_CMP_CNTL_DIS | + R128_GMC_WR_MSK_DIS); + + OUT_RING(dev_priv->depth_pitch_offset_c); + OUT_RING(buffer[i]); + + OUT_RING((x[i] << 16) | y[i]); + OUT_RING((1 << 16) | 1); + + ADVANCE_RING(); + } + } + + kfree(mask); + } else { + for (i = 0; i < count; i++) { + BEGIN_RING(6); + + OUT_RING(CCE_PACKET3(R128_CNTL_PAINT_MULTI, 4)); + OUT_RING(R128_GMC_DST_PITCH_OFFSET_CNTL | + R128_GMC_BRUSH_SOLID_COLOR | + (dev_priv->depth_fmt << 8) | + R128_GMC_SRC_DATATYPE_COLOR | + R128_ROP3_P | + R128_GMC_CLR_CMP_CNTL_DIS | + R128_GMC_WR_MSK_DIS); + + OUT_RING(dev_priv->depth_pitch_offset_c); + OUT_RING(buffer[i]); + + OUT_RING((x[i] << 16) | y[i]); + OUT_RING((1 << 16) | 1); + + ADVANCE_RING(); + } + } + + kfree(x); + kfree(y); + kfree(buffer); + + return 0; +} + +static int r128_cce_dispatch_read_span(struct drm_device *dev, + drm_r128_depth_t *depth) +{ + drm_r128_private_t *dev_priv = dev->dev_private; + int count, x, y; + RING_LOCALS; + DRM_DEBUG("\n"); + + count = depth->n; + if (count > 4096 || count <= 0) + return -EMSGSIZE; + + if (DRM_COPY_FROM_USER(&x, depth->x, sizeof(x))) + return -EFAULT; + if (DRM_COPY_FROM_USER(&y, depth->y, sizeof(y))) + return -EFAULT; + + BEGIN_RING(7); + + OUT_RING(CCE_PACKET3(R128_CNTL_BITBLT_MULTI, 5)); + OUT_RING(R128_GMC_SRC_PITCH_OFFSET_CNTL | + R128_GMC_DST_PITCH_OFFSET_CNTL | + R128_GMC_BRUSH_NONE | + (dev_priv->depth_fmt << 8) | + R128_GMC_SRC_DATATYPE_COLOR | + R128_ROP3_S | + R128_DP_SRC_SOURCE_MEMORY | + R128_GMC_CLR_CMP_CNTL_DIS | R128_GMC_WR_MSK_DIS); + + OUT_RING(dev_priv->depth_pitch_offset_c); + OUT_RING(dev_priv->span_pitch_offset_c); + + OUT_RING((x << 16) | y); + OUT_RING((0 << 16) | 0); + OUT_RING((count << 16) | 1); + + ADVANCE_RING(); + + return 0; +} + +static int r128_cce_dispatch_read_pixels(struct drm_device *dev, + drm_r128_depth_t *depth) +{ + drm_r128_private_t *dev_priv = dev->dev_private; + int count, *x, *y; + int i, xbuf_size, ybuf_size; + RING_LOCALS; + DRM_DEBUG("\n"); + + count = depth->n; + if (count > 4096 || count <= 0) + return -EMSGSIZE; + + if (count > dev_priv->depth_pitch) + count = dev_priv->depth_pitch; + + xbuf_size = count * sizeof(*x); + ybuf_size = count * sizeof(*y); + x = kmalloc(xbuf_size, GFP_KERNEL); + if (x == NULL) + return -ENOMEM; + y = kmalloc(ybuf_size, GFP_KERNEL); + if (y == NULL) { + kfree(x); + return -ENOMEM; + } + if (DRM_COPY_FROM_USER(x, depth->x, xbuf_size)) { + kfree(x); + kfree(y); + return -EFAULT; + } + if (DRM_COPY_FROM_USER(y, depth->y, ybuf_size)) { + kfree(x); + kfree(y); + return -EFAULT; + } + + for (i = 0; i < count; i++) { + BEGIN_RING(7); + + OUT_RING(CCE_PACKET3(R128_CNTL_BITBLT_MULTI, 5)); + OUT_RING(R128_GMC_SRC_PITCH_OFFSET_CNTL | + R128_GMC_DST_PITCH_OFFSET_CNTL | + R128_GMC_BRUSH_NONE | + (dev_priv->depth_fmt << 8) | + R128_GMC_SRC_DATATYPE_COLOR | + R128_ROP3_S | + R128_DP_SRC_SOURCE_MEMORY | + R128_GMC_CLR_CMP_CNTL_DIS | R128_GMC_WR_MSK_DIS); + + OUT_RING(dev_priv->depth_pitch_offset_c); + OUT_RING(dev_priv->span_pitch_offset_c); + + OUT_RING((x[i] << 16) | y[i]); + OUT_RING((i << 16) | 0); + OUT_RING((1 << 16) | 1); + + ADVANCE_RING(); + } + + kfree(x); + kfree(y); + + return 0; +} + +/* ================================================================ + * Polygon stipple + */ + +static void r128_cce_dispatch_stipple(struct drm_device *dev, u32 *stipple) +{ + drm_r128_private_t *dev_priv = dev->dev_private; + int i; + RING_LOCALS; + DRM_DEBUG("\n"); + + BEGIN_RING(33); + + OUT_RING(CCE_PACKET0(R128_BRUSH_DATA0, 31)); + for (i = 0; i < 32; i++) + OUT_RING(stipple[i]); + + ADVANCE_RING(); +} + +/* ================================================================ + * IOCTL functions + */ + +static int r128_cce_clear(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_r128_private_t *dev_priv = dev->dev_private; + drm_r128_sarea_t *sarea_priv; + drm_r128_clear_t *clear = data; + DRM_DEBUG("\n"); + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + DEV_INIT_TEST_WITH_RETURN(dev_priv); + + RING_SPACE_TEST_WITH_RETURN(dev_priv); + + sarea_priv = dev_priv->sarea_priv; + + if (sarea_priv->nbox > R128_NR_SAREA_CLIPRECTS) + sarea_priv->nbox = R128_NR_SAREA_CLIPRECTS; + + r128_cce_dispatch_clear(dev, clear); + COMMIT_RING(); + + /* Make sure we restore the 3D state next time. + */ + dev_priv->sarea_priv->dirty |= R128_UPLOAD_CONTEXT | R128_UPLOAD_MASKS; + + return 0; +} + +static int r128_do_init_pageflip(struct drm_device *dev) +{ + drm_r128_private_t *dev_priv = dev->dev_private; + DRM_DEBUG("\n"); + + dev_priv->crtc_offset = R128_READ(R128_CRTC_OFFSET); + dev_priv->crtc_offset_cntl = R128_READ(R128_CRTC_OFFSET_CNTL); + + R128_WRITE(R128_CRTC_OFFSET, dev_priv->front_offset); + R128_WRITE(R128_CRTC_OFFSET_CNTL, + dev_priv->crtc_offset_cntl | R128_CRTC_OFFSET_FLIP_CNTL); + + dev_priv->page_flipping = 1; + dev_priv->current_page = 0; + dev_priv->sarea_priv->pfCurrentPage = dev_priv->current_page; + + return 0; +} + +static int r128_do_cleanup_pageflip(struct drm_device *dev) +{ + drm_r128_private_t *dev_priv = dev->dev_private; + DRM_DEBUG("\n"); + + R128_WRITE(R128_CRTC_OFFSET, dev_priv->crtc_offset); + R128_WRITE(R128_CRTC_OFFSET_CNTL, dev_priv->crtc_offset_cntl); + + if (dev_priv->current_page != 0) { + r128_cce_dispatch_flip(dev); + COMMIT_RING(); + } + + dev_priv->page_flipping = 0; + return 0; +} + +/* Swapping and flipping are different operations, need different ioctls. + * They can & should be intermixed to support multiple 3d windows. + */ + +static int r128_cce_flip(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_r128_private_t *dev_priv = dev->dev_private; + DRM_DEBUG("\n"); + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + DEV_INIT_TEST_WITH_RETURN(dev_priv); + + RING_SPACE_TEST_WITH_RETURN(dev_priv); + + if (!dev_priv->page_flipping) + r128_do_init_pageflip(dev); + + r128_cce_dispatch_flip(dev); + + COMMIT_RING(); + return 0; +} + +static int r128_cce_swap(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_r128_private_t *dev_priv = dev->dev_private; + drm_r128_sarea_t *sarea_priv = dev_priv->sarea_priv; + DRM_DEBUG("\n"); + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + DEV_INIT_TEST_WITH_RETURN(dev_priv); + + RING_SPACE_TEST_WITH_RETURN(dev_priv); + + if (sarea_priv->nbox > R128_NR_SAREA_CLIPRECTS) + sarea_priv->nbox = R128_NR_SAREA_CLIPRECTS; + + r128_cce_dispatch_swap(dev); + dev_priv->sarea_priv->dirty |= (R128_UPLOAD_CONTEXT | + R128_UPLOAD_MASKS); + + COMMIT_RING(); + return 0; +} + +static int r128_cce_vertex(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_r128_private_t *dev_priv = dev->dev_private; + struct drm_device_dma *dma = dev->dma; + struct drm_buf *buf; + drm_r128_buf_priv_t *buf_priv; + drm_r128_vertex_t *vertex = data; + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + DEV_INIT_TEST_WITH_RETURN(dev_priv); + + DRM_DEBUG("pid=%d index=%d count=%d discard=%d\n", + DRM_CURRENTPID, vertex->idx, vertex->count, vertex->discard); + + if (vertex->idx < 0 || vertex->idx >= dma->buf_count) { + DRM_ERROR("buffer index %d (of %d max)\n", + vertex->idx, dma->buf_count - 1); + return -EINVAL; + } + if (vertex->prim < 0 || + vertex->prim > R128_CCE_VC_CNTL_PRIM_TYPE_TRI_TYPE2) { + DRM_ERROR("buffer prim %d\n", vertex->prim); + return -EINVAL; + } + + RING_SPACE_TEST_WITH_RETURN(dev_priv); + VB_AGE_TEST_WITH_RETURN(dev_priv); + + buf = dma->buflist[vertex->idx]; + buf_priv = buf->dev_private; + + if (buf->file_priv != file_priv) { + DRM_ERROR("process %d using buffer owned by %p\n", + DRM_CURRENTPID, buf->file_priv); + return -EINVAL; + } + if (buf->pending) { + DRM_ERROR("sending pending buffer %d\n", vertex->idx); + return -EINVAL; + } + + buf->used = vertex->count; + buf_priv->prim = vertex->prim; + buf_priv->discard = vertex->discard; + + r128_cce_dispatch_vertex(dev, buf); + + COMMIT_RING(); + return 0; +} + +static int r128_cce_indices(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_r128_private_t *dev_priv = dev->dev_private; + struct drm_device_dma *dma = dev->dma; + struct drm_buf *buf; + drm_r128_buf_priv_t *buf_priv; + drm_r128_indices_t *elts = data; + int count; + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + DEV_INIT_TEST_WITH_RETURN(dev_priv); + + DRM_DEBUG("pid=%d buf=%d s=%d e=%d d=%d\n", DRM_CURRENTPID, + elts->idx, elts->start, elts->end, elts->discard); + + if (elts->idx < 0 || elts->idx >= dma->buf_count) { + DRM_ERROR("buffer index %d (of %d max)\n", + elts->idx, dma->buf_count - 1); + return -EINVAL; + } + if (elts->prim < 0 || + elts->prim > R128_CCE_VC_CNTL_PRIM_TYPE_TRI_TYPE2) { + DRM_ERROR("buffer prim %d\n", elts->prim); + return -EINVAL; + } + + RING_SPACE_TEST_WITH_RETURN(dev_priv); + VB_AGE_TEST_WITH_RETURN(dev_priv); + + buf = dma->buflist[elts->idx]; + buf_priv = buf->dev_private; + + if (buf->file_priv != file_priv) { + DRM_ERROR("process %d using buffer owned by %p\n", + DRM_CURRENTPID, buf->file_priv); + return -EINVAL; + } + if (buf->pending) { + DRM_ERROR("sending pending buffer %d\n", elts->idx); + return -EINVAL; + } + + count = (elts->end - elts->start) / sizeof(u16); + elts->start -= R128_INDEX_PRIM_OFFSET; + + if (elts->start & 0x7) { + DRM_ERROR("misaligned buffer 0x%x\n", elts->start); + return -EINVAL; + } + if (elts->start < buf->used) { + DRM_ERROR("no header 0x%x - 0x%x\n", elts->start, buf->used); + return -EINVAL; + } + + buf->used = elts->end; + buf_priv->prim = elts->prim; + buf_priv->discard = elts->discard; + + r128_cce_dispatch_indices(dev, buf, elts->start, elts->end, count); + + COMMIT_RING(); + return 0; +} + +static int r128_cce_blit(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + struct drm_device_dma *dma = dev->dma; + drm_r128_private_t *dev_priv = dev->dev_private; + drm_r128_blit_t *blit = data; + int ret; + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + DEV_INIT_TEST_WITH_RETURN(dev_priv); + + DRM_DEBUG("pid=%d index=%d\n", DRM_CURRENTPID, blit->idx); + + if (blit->idx < 0 || blit->idx >= dma->buf_count) { + DRM_ERROR("buffer index %d (of %d max)\n", + blit->idx, dma->buf_count - 1); + return -EINVAL; + } + + RING_SPACE_TEST_WITH_RETURN(dev_priv); + VB_AGE_TEST_WITH_RETURN(dev_priv); + + ret = r128_cce_dispatch_blit(dev, file_priv, blit); + + COMMIT_RING(); + return ret; +} + +static int r128_cce_depth(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_r128_private_t *dev_priv = dev->dev_private; + drm_r128_depth_t *depth = data; + int ret; + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + DEV_INIT_TEST_WITH_RETURN(dev_priv); + + RING_SPACE_TEST_WITH_RETURN(dev_priv); + + ret = -EINVAL; + switch (depth->func) { + case R128_WRITE_SPAN: + ret = r128_cce_dispatch_write_span(dev, depth); + break; + case R128_WRITE_PIXELS: + ret = r128_cce_dispatch_write_pixels(dev, depth); + break; + case R128_READ_SPAN: + ret = r128_cce_dispatch_read_span(dev, depth); + break; + case R128_READ_PIXELS: + ret = r128_cce_dispatch_read_pixels(dev, depth); + break; + } + + COMMIT_RING(); + return ret; +} + +static int r128_cce_stipple(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_r128_private_t *dev_priv = dev->dev_private; + drm_r128_stipple_t *stipple = data; + u32 mask[32]; + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + DEV_INIT_TEST_WITH_RETURN(dev_priv); + + if (DRM_COPY_FROM_USER(&mask, stipple->mask, 32 * sizeof(u32))) + return -EFAULT; + + RING_SPACE_TEST_WITH_RETURN(dev_priv); + + r128_cce_dispatch_stipple(dev, mask); + + COMMIT_RING(); + return 0; +} + +static int r128_cce_indirect(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_r128_private_t *dev_priv = dev->dev_private; + struct drm_device_dma *dma = dev->dma; + struct drm_buf *buf; + drm_r128_buf_priv_t *buf_priv; + drm_r128_indirect_t *indirect = data; +#if 0 + RING_LOCALS; +#endif + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + DEV_INIT_TEST_WITH_RETURN(dev_priv); + + DRM_DEBUG("idx=%d s=%d e=%d d=%d\n", + indirect->idx, indirect->start, indirect->end, + indirect->discard); + + if (indirect->idx < 0 || indirect->idx >= dma->buf_count) { + DRM_ERROR("buffer index %d (of %d max)\n", + indirect->idx, dma->buf_count - 1); + return -EINVAL; + } + + buf = dma->buflist[indirect->idx]; + buf_priv = buf->dev_private; + + if (buf->file_priv != file_priv) { + DRM_ERROR("process %d using buffer owned by %p\n", + DRM_CURRENTPID, buf->file_priv); + return -EINVAL; + } + if (buf->pending) { + DRM_ERROR("sending pending buffer %d\n", indirect->idx); + return -EINVAL; + } + + if (indirect->start < buf->used) { + DRM_ERROR("reusing indirect: start=0x%x actual=0x%x\n", + indirect->start, buf->used); + return -EINVAL; + } + + RING_SPACE_TEST_WITH_RETURN(dev_priv); + VB_AGE_TEST_WITH_RETURN(dev_priv); + + buf->used = indirect->end; + buf_priv->discard = indirect->discard; + +#if 0 + /* Wait for the 3D stream to idle before the indirect buffer + * containing 2D acceleration commands is processed. + */ + BEGIN_RING(2); + RADEON_WAIT_UNTIL_3D_IDLE(); + ADVANCE_RING(); +#endif + + /* Dispatch the indirect buffer full of commands from the + * X server. This is insecure and is thus only available to + * privileged clients. + */ + r128_cce_dispatch_indirect(dev, buf, indirect->start, indirect->end); + + COMMIT_RING(); + return 0; +} + +static int r128_getparam(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_r128_private_t *dev_priv = dev->dev_private; + drm_r128_getparam_t *param = data; + int value; + + DEV_INIT_TEST_WITH_RETURN(dev_priv); + + DRM_DEBUG("pid=%d\n", DRM_CURRENTPID); + + switch (param->param) { + case R128_PARAM_IRQ_NR: + value = drm_dev_to_irq(dev); + break; + default: + return -EINVAL; + } + + if (DRM_COPY_TO_USER(param->value, &value, sizeof(int))) { + DRM_ERROR("copy_to_user\n"); + return -EFAULT; + } + + return 0; +} + +void r128_driver_preclose(struct drm_device *dev, struct drm_file *file_priv) +{ + if (dev->dev_private) { + drm_r128_private_t *dev_priv = dev->dev_private; + if (dev_priv->page_flipping) + r128_do_cleanup_pageflip(dev); + } +} +void r128_driver_lastclose(struct drm_device *dev) +{ + r128_do_cleanup_cce(dev); +} + +struct drm_ioctl_desc r128_ioctls[] = { + DRM_IOCTL_DEF_DRV(R128_INIT, r128_cce_init, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF_DRV(R128_CCE_START, r128_cce_start, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF_DRV(R128_CCE_STOP, r128_cce_stop, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF_DRV(R128_CCE_RESET, r128_cce_reset, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF_DRV(R128_CCE_IDLE, r128_cce_idle, DRM_AUTH), + DRM_IOCTL_DEF_DRV(R128_RESET, r128_engine_reset, DRM_AUTH), + DRM_IOCTL_DEF_DRV(R128_FULLSCREEN, r128_fullscreen, DRM_AUTH), + DRM_IOCTL_DEF_DRV(R128_SWAP, r128_cce_swap, DRM_AUTH), + DRM_IOCTL_DEF_DRV(R128_FLIP, r128_cce_flip, DRM_AUTH), + DRM_IOCTL_DEF_DRV(R128_CLEAR, r128_cce_clear, DRM_AUTH), + DRM_IOCTL_DEF_DRV(R128_VERTEX, r128_cce_vertex, DRM_AUTH), + DRM_IOCTL_DEF_DRV(R128_INDICES, r128_cce_indices, DRM_AUTH), + DRM_IOCTL_DEF_DRV(R128_BLIT, r128_cce_blit, DRM_AUTH), + DRM_IOCTL_DEF_DRV(R128_DEPTH, r128_cce_depth, DRM_AUTH), + DRM_IOCTL_DEF_DRV(R128_STIPPLE, r128_cce_stipple, DRM_AUTH), + DRM_IOCTL_DEF_DRV(R128_INDIRECT, r128_cce_indirect, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF_DRV(R128_GETPARAM, r128_getparam, DRM_AUTH), +}; + +int r128_max_ioctl = DRM_ARRAY_SIZE(r128_ioctls); diff --git a/drivers/gpu/drm/radeon/.gitignore b/drivers/gpu/drm/radeon/.gitignore new file mode 100644 index 00000000..403eb3a5 --- /dev/null +++ b/drivers/gpu/drm/radeon/.gitignore @@ -0,0 +1,3 @@ +mkregtable +*_reg_safe.h + diff --git a/drivers/gpu/drm/radeon/Kconfig b/drivers/gpu/drm/radeon/Kconfig new file mode 100644 index 00000000..ea92bbe3 --- /dev/null +++ b/drivers/gpu/drm/radeon/Kconfig @@ -0,0 +1,31 @@ +config DRM_RADEON_KMS + bool "Enable modesetting on radeon by default - NEW DRIVER" + depends on DRM_RADEON + select BACKLIGHT_CLASS_DEVICE + help + Choose this option if you want kernel modesetting enabled by default. + + This is a completely new driver. It's only part of the existing drm + for compatibility reasons. It requires an entirely different graphics + stack above it and works very differently from the old drm stack. + i.e. don't enable this unless you know what you are doing it may + cause issues or bugs compared to the previous userspace driver stack. + + When kernel modesetting is enabled the IOCTL of radeon/drm + driver are considered as invalid and an error message is printed + in the log and they return failure. + + KMS enabled userspace will use new API to talk with the radeon/drm + driver. The new API provide functions to create/destroy/share/mmap + buffer object which are then managed by the kernel memory manager + (here TTM). In order to submit command to the GPU the userspace + provide a buffer holding the command stream, along this buffer + userspace have to provide a list of buffer object used by the + command stream. The kernel radeon driver will then place buffer + in GPU accessible memory and will update command stream to reflect + the position of the different buffers. + + The kernel will also perform security check on command stream + provided by the user, we want to catch and forbid any illegal use + of the GPU such as DMA into random system memory or into memory + not owned by the process supplying the command stream. diff --git a/drivers/gpu/drm/radeon/Makefile b/drivers/gpu/drm/radeon/Makefile new file mode 100644 index 00000000..3896ef81 --- /dev/null +++ b/drivers/gpu/drm/radeon/Makefile @@ -0,0 +1,80 @@ +# +# Makefile for the drm device driver. This driver provides support for the +# Direct Rendering Infrastructure (DRI) in XFree86 4.1.0 and higher. + +ccflags-y := -Iinclude/drm + +hostprogs-y := mkregtable + +quiet_cmd_mkregtable = MKREGTABLE $@ + cmd_mkregtable = $(obj)/mkregtable $< > $@ + +$(obj)/rn50_reg_safe.h: $(src)/reg_srcs/rn50 $(obj)/mkregtable + $(call if_changed,mkregtable) + +$(obj)/r100_reg_safe.h: $(src)/reg_srcs/r100 $(obj)/mkregtable + $(call if_changed,mkregtable) + +$(obj)/r200_reg_safe.h: $(src)/reg_srcs/r200 $(obj)/mkregtable + $(call if_changed,mkregtable) + +$(obj)/rv515_reg_safe.h: $(src)/reg_srcs/rv515 $(obj)/mkregtable + $(call if_changed,mkregtable) + +$(obj)/r300_reg_safe.h: $(src)/reg_srcs/r300 $(obj)/mkregtable + $(call if_changed,mkregtable) + +$(obj)/r420_reg_safe.h: $(src)/reg_srcs/r420 $(obj)/mkregtable + $(call if_changed,mkregtable) + +$(obj)/rs600_reg_safe.h: $(src)/reg_srcs/rs600 $(obj)/mkregtable + $(call if_changed,mkregtable) + +$(obj)/r600_reg_safe.h: $(src)/reg_srcs/r600 $(obj)/mkregtable + $(call if_changed,mkregtable) + +$(obj)/evergreen_reg_safe.h: $(src)/reg_srcs/evergreen $(obj)/mkregtable + $(call if_changed,mkregtable) + +$(obj)/cayman_reg_safe.h: $(src)/reg_srcs/cayman $(obj)/mkregtable + $(call if_changed,mkregtable) + +$(obj)/r100.o: $(obj)/r100_reg_safe.h $(obj)/rn50_reg_safe.h + +$(obj)/r200.o: $(obj)/r200_reg_safe.h + +$(obj)/rv515.o: $(obj)/rv515_reg_safe.h + +$(obj)/r300.o: $(obj)/r300_reg_safe.h + +$(obj)/r420.o: $(obj)/r420_reg_safe.h + +$(obj)/rs600.o: $(obj)/rs600_reg_safe.h + +$(obj)/r600_cs.o: $(obj)/r600_reg_safe.h + +$(obj)/evergreen_cs.o: $(obj)/evergreen_reg_safe.h $(obj)/cayman_reg_safe.h + +radeon-y := radeon_drv.o radeon_cp.o radeon_state.o radeon_mem.o \ + radeon_irq.o r300_cmdbuf.o r600_cp.o +# add KMS driver +radeon-y += radeon_device.o radeon_asic.o radeon_kms.o \ + radeon_atombios.o radeon_agp.o atombios_crtc.o radeon_combios.o \ + atom.o radeon_fence.o radeon_ttm.o radeon_object.o radeon_gart.o \ + radeon_legacy_crtc.o radeon_legacy_encoders.o radeon_connectors.o \ + radeon_encoders.o radeon_display.o radeon_cursor.o radeon_i2c.o \ + radeon_clocks.o radeon_fb.o radeon_gem.o radeon_ring.o radeon_irq_kms.o \ + radeon_cs.o radeon_bios.o radeon_benchmark.o r100.o r300.o r420.o \ + rs400.o rs600.o rs690.o rv515.o r520.o r600.o rv770.o radeon_test.o \ + r200.o radeon_legacy_tv.o r600_cs.o r600_blit.o r600_blit_shaders.o \ + r600_blit_kms.o radeon_pm.o atombios_dp.o r600_audio.o r600_hdmi.o \ + evergreen.o evergreen_cs.o evergreen_blit_shaders.o evergreen_blit_kms.o \ + radeon_trace_points.o ni.o cayman_blit_shaders.o + +radeon-$(CONFIG_COMPAT) += radeon_ioc32.o +radeon-$(CONFIG_VGA_SWITCHEROO) += radeon_atpx_handler.o +radeon-$(CONFIG_ACPI) += radeon_acpi.o + +obj-$(CONFIG_DRM_RADEON)+= radeon.o + +CFLAGS_radeon_trace_points.o := -I$(src) \ No newline at end of file diff --git a/drivers/gpu/drm/radeon/ObjectID.h b/drivers/gpu/drm/radeon/ObjectID.h new file mode 100644 index 00000000..c61c3fe9 --- /dev/null +++ b/drivers/gpu/drm/radeon/ObjectID.h @@ -0,0 +1,691 @@ +/* +* Copyright 2006-2007 Advanced Micro Devices, Inc. +* +* Permission is hereby granted, free of charge, to any person obtaining a +* copy of this software and associated documentation files (the "Software"), +* to deal in the Software without restriction, including without limitation +* the rights to use, copy, modify, merge, publish, distribute, sublicense, +* and/or sell copies of the Software, and to permit persons to whom the +* Software is furnished to do so, subject to the following conditions: +* +* The above copyright notice and this permission notice shall be included in +* all copies or substantial portions of the Software. +* +* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL +* THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR +* OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, +* ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR +* OTHER DEALINGS IN THE SOFTWARE. +*/ +/* based on stg/asic_reg/drivers/inc/asic_reg/ObjectID.h ver 23 */ + +#ifndef _OBJECTID_H +#define _OBJECTID_H + +#if defined(_X86_) +#pragma pack(1) +#endif + +/****************************************************/ +/* Graphics Object Type Definition */ +/****************************************************/ +#define GRAPH_OBJECT_TYPE_NONE 0x0 +#define GRAPH_OBJECT_TYPE_GPU 0x1 +#define GRAPH_OBJECT_TYPE_ENCODER 0x2 +#define GRAPH_OBJECT_TYPE_CONNECTOR 0x3 +#define GRAPH_OBJECT_TYPE_ROUTER 0x4 +/* deleted */ +#define GRAPH_OBJECT_TYPE_DISPLAY_PATH 0x6 +#define GRAPH_OBJECT_TYPE_GENERIC 0x7 + +/****************************************************/ +/* Encoder Object ID Definition */ +/****************************************************/ +#define ENCODER_OBJECT_ID_NONE 0x00 + +/* Radeon Class Display Hardware */ +#define ENCODER_OBJECT_ID_INTERNAL_LVDS 0x01 +#define ENCODER_OBJECT_ID_INTERNAL_TMDS1 0x02 +#define ENCODER_OBJECT_ID_INTERNAL_TMDS2 0x03 +#define ENCODER_OBJECT_ID_INTERNAL_DAC1 0x04 +#define ENCODER_OBJECT_ID_INTERNAL_DAC2 0x05 /* TV/CV DAC */ +#define ENCODER_OBJECT_ID_INTERNAL_SDVOA 0x06 +#define ENCODER_OBJECT_ID_INTERNAL_SDVOB 0x07 + +/* External Third Party Encoders */ +#define ENCODER_OBJECT_ID_SI170B 0x08 +#define ENCODER_OBJECT_ID_CH7303 0x09 +#define ENCODER_OBJECT_ID_CH7301 0x0A +#define ENCODER_OBJECT_ID_INTERNAL_DVO1 0x0B /* This belongs to Radeon Class Display Hardware */ +#define ENCODER_OBJECT_ID_EXTERNAL_SDVOA 0x0C +#define ENCODER_OBJECT_ID_EXTERNAL_SDVOB 0x0D +#define ENCODER_OBJECT_ID_TITFP513 0x0E +#define ENCODER_OBJECT_ID_INTERNAL_LVTM1 0x0F /* not used for Radeon */ +#define ENCODER_OBJECT_ID_VT1623 0x10 +#define ENCODER_OBJECT_ID_HDMI_SI1930 0x11 +#define ENCODER_OBJECT_ID_HDMI_INTERNAL 0x12 +#define ENCODER_OBJECT_ID_ALMOND 0x22 +#define ENCODER_OBJECT_ID_TRAVIS 0x23 +#define ENCODER_OBJECT_ID_NUTMEG 0x22 +/* Kaleidoscope (KLDSCP) Class Display Hardware (internal) */ +#define ENCODER_OBJECT_ID_INTERNAL_KLDSCP_TMDS1 0x13 +#define ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1 0x14 +#define ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1 0x15 +#define ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2 0x16 /* Shared with CV/TV and CRT */ +#define ENCODER_OBJECT_ID_SI178 0X17 /* External TMDS (dual link, no HDCP.) */ +#define ENCODER_OBJECT_ID_MVPU_FPGA 0x18 /* MVPU FPGA chip */ +#define ENCODER_OBJECT_ID_INTERNAL_DDI 0x19 +#define ENCODER_OBJECT_ID_VT1625 0x1A +#define ENCODER_OBJECT_ID_HDMI_SI1932 0x1B +#define ENCODER_OBJECT_ID_DP_AN9801 0x1C +#define ENCODER_OBJECT_ID_DP_DP501 0x1D +#define ENCODER_OBJECT_ID_INTERNAL_UNIPHY 0x1E +#define ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA 0x1F +#define ENCODER_OBJECT_ID_INTERNAL_UNIPHY1 0x20 +#define ENCODER_OBJECT_ID_INTERNAL_UNIPHY2 0x21 + +#define ENCODER_OBJECT_ID_GENERAL_EXTERNAL_DVO 0xFF + +/****************************************************/ +/* Connector Object ID Definition */ +/****************************************************/ +#define CONNECTOR_OBJECT_ID_NONE 0x00 +#define CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I 0x01 +#define CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_I 0x02 +#define CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D 0x03 +#define CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_D 0x04 +#define CONNECTOR_OBJECT_ID_VGA 0x05 +#define CONNECTOR_OBJECT_ID_COMPOSITE 0x06 +#define CONNECTOR_OBJECT_ID_SVIDEO 0x07 +#define CONNECTOR_OBJECT_ID_YPbPr 0x08 +#define CONNECTOR_OBJECT_ID_D_CONNECTOR 0x09 +#define CONNECTOR_OBJECT_ID_9PIN_DIN 0x0A /* Supports both CV & TV */ +#define CONNECTOR_OBJECT_ID_SCART 0x0B +#define CONNECTOR_OBJECT_ID_HDMI_TYPE_A 0x0C +#define CONNECTOR_OBJECT_ID_HDMI_TYPE_B 0x0D +#define CONNECTOR_OBJECT_ID_LVDS 0x0E +#define CONNECTOR_OBJECT_ID_7PIN_DIN 0x0F +#define CONNECTOR_OBJECT_ID_PCIE_CONNECTOR 0x10 +#define CONNECTOR_OBJECT_ID_CROSSFIRE 0x11 +#define CONNECTOR_OBJECT_ID_HARDCODE_DVI 0x12 +#define CONNECTOR_OBJECT_ID_DISPLAYPORT 0x13 +#define CONNECTOR_OBJECT_ID_eDP 0x14 +#define CONNECTOR_OBJECT_ID_MXM 0x15 +#define CONNECTOR_OBJECT_ID_LVDS_eDP 0x16 + +/* deleted */ + +/****************************************************/ +/* Router Object ID Definition */ +/****************************************************/ +#define ROUTER_OBJECT_ID_NONE 0x00 +#define ROUTER_OBJECT_ID_I2C_EXTENDER_CNTL 0x01 + +/****************************************************/ +/* Generic Object ID Definition */ +/****************************************************/ +#define GENERIC_OBJECT_ID_NONE 0x00 +#define GENERIC_OBJECT_ID_GLSYNC 0x01 +#define GENERIC_OBJECT_ID_PX2_NON_DRIVABLE 0x02 +#define GENERIC_OBJECT_ID_MXM_OPM 0x03 +#define GENERIC_OBJECT_ID_STEREO_PIN 0x04 //This object could show up from Misc Object table, it follows ATOM_OBJECT format, and contains one ATOM_OBJECT_GPIO_CNTL_RECORD for the stereo pin + +/****************************************************/ +/* Graphics Object ENUM ID Definition */ +/****************************************************/ +#define GRAPH_OBJECT_ENUM_ID1 0x01 +#define GRAPH_OBJECT_ENUM_ID2 0x02 +#define GRAPH_OBJECT_ENUM_ID3 0x03 +#define GRAPH_OBJECT_ENUM_ID4 0x04 +#define GRAPH_OBJECT_ENUM_ID5 0x05 +#define GRAPH_OBJECT_ENUM_ID6 0x06 +#define GRAPH_OBJECT_ENUM_ID7 0x07 + +/****************************************************/ +/* Graphics Object ID Bit definition */ +/****************************************************/ +#define OBJECT_ID_MASK 0x00FF +#define ENUM_ID_MASK 0x0700 +#define RESERVED1_ID_MASK 0x0800 +#define OBJECT_TYPE_MASK 0x7000 +#define RESERVED2_ID_MASK 0x8000 + +#define OBJECT_ID_SHIFT 0x00 +#define ENUM_ID_SHIFT 0x08 +#define OBJECT_TYPE_SHIFT 0x0C + + +/****************************************************/ +/* Graphics Object family definition */ +/****************************************************/ +#define CONSTRUCTOBJECTFAMILYID(GRAPHICS_OBJECT_TYPE, GRAPHICS_OBJECT_ID) (GRAPHICS_OBJECT_TYPE << OBJECT_TYPE_SHIFT | \ + GRAPHICS_OBJECT_ID << OBJECT_ID_SHIFT) +/****************************************************/ +/* GPU Object ID definition - Shared with BIOS */ +/****************************************************/ +#define GPU_ENUM_ID1 ( GRAPH_OBJECT_TYPE_GPU << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT) + +/****************************************************/ +/* Encoder Object ID definition - Shared with BIOS */ +/****************************************************/ +/* +#define ENCODER_INTERNAL_LVDS_ENUM_ID1 0x2101 +#define ENCODER_INTERNAL_TMDS1_ENUM_ID1 0x2102 +#define ENCODER_INTERNAL_TMDS2_ENUM_ID1 0x2103 +#define ENCODER_INTERNAL_DAC1_ENUM_ID1 0x2104 +#define ENCODER_INTERNAL_DAC2_ENUM_ID1 0x2105 +#define ENCODER_INTERNAL_SDVOA_ENUM_ID1 0x2106 +#define ENCODER_INTERNAL_SDVOB_ENUM_ID1 0x2107 +#define ENCODER_SIL170B_ENUM_ID1 0x2108 +#define ENCODER_CH7303_ENUM_ID1 0x2109 +#define ENCODER_CH7301_ENUM_ID1 0x210A +#define ENCODER_INTERNAL_DVO1_ENUM_ID1 0x210B +#define ENCODER_EXTERNAL_SDVOA_ENUM_ID1 0x210C +#define ENCODER_EXTERNAL_SDVOB_ENUM_ID1 0x210D +#define ENCODER_TITFP513_ENUM_ID1 0x210E +#define ENCODER_INTERNAL_LVTM1_ENUM_ID1 0x210F +#define ENCODER_VT1623_ENUM_ID1 0x2110 +#define ENCODER_HDMI_SI1930_ENUM_ID1 0x2111 +#define ENCODER_HDMI_INTERNAL_ENUM_ID1 0x2112 +#define ENCODER_INTERNAL_KLDSCP_TMDS1_ENUM_ID1 0x2113 +#define ENCODER_INTERNAL_KLDSCP_DVO1_ENUM_ID1 0x2114 +#define ENCODER_INTERNAL_KLDSCP_DAC1_ENUM_ID1 0x2115 +#define ENCODER_INTERNAL_KLDSCP_DAC2_ENUM_ID1 0x2116 +#define ENCODER_SI178_ENUM_ID1 0x2117 +#define ENCODER_MVPU_FPGA_ENUM_ID1 0x2118 +#define ENCODER_INTERNAL_DDI_ENUM_ID1 0x2119 +#define ENCODER_VT1625_ENUM_ID1 0x211A +#define ENCODER_HDMI_SI1932_ENUM_ID1 0x211B +#define ENCODER_ENCODER_DP_AN9801_ENUM_ID1 0x211C +#define ENCODER_DP_DP501_ENUM_ID1 0x211D +#define ENCODER_INTERNAL_UNIPHY_ENUM_ID1 0x211E +*/ +#define ENCODER_INTERNAL_LVDS_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_INTERNAL_LVDS << OBJECT_ID_SHIFT) + +#define ENCODER_INTERNAL_TMDS1_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_INTERNAL_TMDS1 << OBJECT_ID_SHIFT) + +#define ENCODER_INTERNAL_TMDS2_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_INTERNAL_TMDS2 << OBJECT_ID_SHIFT) + +#define ENCODER_INTERNAL_DAC1_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_INTERNAL_DAC1 << OBJECT_ID_SHIFT) + +#define ENCODER_INTERNAL_DAC2_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_INTERNAL_DAC2 << OBJECT_ID_SHIFT) + +#define ENCODER_INTERNAL_SDVOA_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_INTERNAL_SDVOA << OBJECT_ID_SHIFT) + +#define ENCODER_INTERNAL_SDVOA_ENUM_ID2 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_INTERNAL_SDVOA << OBJECT_ID_SHIFT) + +#define ENCODER_INTERNAL_SDVOB_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_INTERNAL_SDVOB << OBJECT_ID_SHIFT) + +#define ENCODER_SIL170B_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_SI170B << OBJECT_ID_SHIFT) + +#define ENCODER_CH7303_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_CH7303 << OBJECT_ID_SHIFT) + +#define ENCODER_CH7301_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_CH7301 << OBJECT_ID_SHIFT) + +#define ENCODER_INTERNAL_DVO1_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_INTERNAL_DVO1 << OBJECT_ID_SHIFT) + +#define ENCODER_EXTERNAL_SDVOA_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_EXTERNAL_SDVOA << OBJECT_ID_SHIFT) + +#define ENCODER_EXTERNAL_SDVOA_ENUM_ID2 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_EXTERNAL_SDVOA << OBJECT_ID_SHIFT) + + +#define ENCODER_EXTERNAL_SDVOB_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_EXTERNAL_SDVOB << OBJECT_ID_SHIFT) + + +#define ENCODER_TITFP513_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_TITFP513 << OBJECT_ID_SHIFT) + +#define ENCODER_INTERNAL_LVTM1_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_INTERNAL_LVTM1 << OBJECT_ID_SHIFT) + +#define ENCODER_VT1623_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_VT1623 << OBJECT_ID_SHIFT) + +#define ENCODER_HDMI_SI1930_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_HDMI_SI1930 << OBJECT_ID_SHIFT) + +#define ENCODER_HDMI_INTERNAL_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_HDMI_INTERNAL << OBJECT_ID_SHIFT) + +#define ENCODER_INTERNAL_KLDSCP_TMDS1_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_INTERNAL_KLDSCP_TMDS1 << OBJECT_ID_SHIFT) + + +#define ENCODER_INTERNAL_KLDSCP_TMDS1_ENUM_ID2 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_INTERNAL_KLDSCP_TMDS1 << OBJECT_ID_SHIFT) + + +#define ENCODER_INTERNAL_KLDSCP_DVO1_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1 << OBJECT_ID_SHIFT) + +#define ENCODER_INTERNAL_KLDSCP_DAC1_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1 << OBJECT_ID_SHIFT) + +#define ENCODER_INTERNAL_KLDSCP_DAC2_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2 << OBJECT_ID_SHIFT) // Shared with CV/TV and CRT + +#define ENCODER_SI178_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_SI178 << OBJECT_ID_SHIFT) + +#define ENCODER_MVPU_FPGA_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_MVPU_FPGA << OBJECT_ID_SHIFT) + +#define ENCODER_INTERNAL_DDI_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_INTERNAL_DDI << OBJECT_ID_SHIFT) + +#define ENCODER_VT1625_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_VT1625 << OBJECT_ID_SHIFT) + +#define ENCODER_HDMI_SI1932_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_HDMI_SI1932 << OBJECT_ID_SHIFT) + +#define ENCODER_DP_DP501_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_DP_DP501 << OBJECT_ID_SHIFT) + +#define ENCODER_DP_AN9801_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_DP_AN9801 << OBJECT_ID_SHIFT) + +#define ENCODER_INTERNAL_UNIPHY_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_INTERNAL_UNIPHY << OBJECT_ID_SHIFT) + +#define ENCODER_INTERNAL_UNIPHY_ENUM_ID2 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_INTERNAL_UNIPHY << OBJECT_ID_SHIFT) + +#define ENCODER_INTERNAL_KLDSCP_LVTMA_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA << OBJECT_ID_SHIFT) + +#define ENCODER_INTERNAL_UNIPHY1_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_INTERNAL_UNIPHY1 << OBJECT_ID_SHIFT) + +#define ENCODER_INTERNAL_UNIPHY1_ENUM_ID2 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_INTERNAL_UNIPHY1 << OBJECT_ID_SHIFT) + +#define ENCODER_INTERNAL_UNIPHY2_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_INTERNAL_UNIPHY2 << OBJECT_ID_SHIFT) + +#define ENCODER_INTERNAL_UNIPHY2_ENUM_ID2 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_INTERNAL_UNIPHY2 << OBJECT_ID_SHIFT) + +#define ENCODER_GENERAL_EXTERNAL_DVO_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_GENERAL_EXTERNAL_DVO << OBJECT_ID_SHIFT) + +#define ENCODER_ALMOND_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_ALMOND << OBJECT_ID_SHIFT) + +#define ENCODER_ALMOND_ENUM_ID2 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_ALMOND << OBJECT_ID_SHIFT) + +#define ENCODER_TRAVIS_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_TRAVIS << OBJECT_ID_SHIFT) + +#define ENCODER_TRAVIS_ENUM_ID2 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_TRAVIS << OBJECT_ID_SHIFT) + +#define ENCODER_NUTMEG_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_NUTMEG << OBJECT_ID_SHIFT) + +/****************************************************/ +/* Connector Object ID definition - Shared with BIOS */ +/****************************************************/ +/* +#define CONNECTOR_SINGLE_LINK_DVI_I_ENUM_ID1 0x3101 +#define CONNECTOR_DUAL_LINK_DVI_I_ENUM_ID1 0x3102 +#define CONNECTOR_SINGLE_LINK_DVI_D_ENUM_ID1 0x3103 +#define CONNECTOR_DUAL_LINK_DVI_D_ENUM_ID1 0x3104 +#define CONNECTOR_VGA_ENUM_ID1 0x3105 +#define CONNECTOR_COMPOSITE_ENUM_ID1 0x3106 +#define CONNECTOR_SVIDEO_ENUM_ID1 0x3107 +#define CONNECTOR_YPbPr_ENUM_ID1 0x3108 +#define CONNECTOR_D_CONNECTORE_ENUM_ID1 0x3109 +#define CONNECTOR_9PIN_DIN_ENUM_ID1 0x310A +#define CONNECTOR_SCART_ENUM_ID1 0x310B +#define CONNECTOR_HDMI_TYPE_A_ENUM_ID1 0x310C +#define CONNECTOR_HDMI_TYPE_B_ENUM_ID1 0x310D +#define CONNECTOR_LVDS_ENUM_ID1 0x310E +#define CONNECTOR_7PIN_DIN_ENUM_ID1 0x310F +#define CONNECTOR_PCIE_CONNECTOR_ENUM_ID1 0x3110 +*/ +#define CONNECTOR_LVDS_ENUM_ID1 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_LVDS << OBJECT_ID_SHIFT) + +#define CONNECTOR_LVDS_ENUM_ID2 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_LVDS << OBJECT_ID_SHIFT) + +#define CONNECTOR_eDP_ENUM_ID1 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_eDP << OBJECT_ID_SHIFT) + +#define CONNECTOR_eDP_ENUM_ID2 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_eDP << OBJECT_ID_SHIFT) + +#define CONNECTOR_SINGLE_LINK_DVI_I_ENUM_ID1 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I << OBJECT_ID_SHIFT) + +#define CONNECTOR_SINGLE_LINK_DVI_I_ENUM_ID2 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I << OBJECT_ID_SHIFT) + +#define CONNECTOR_DUAL_LINK_DVI_I_ENUM_ID1 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_I << OBJECT_ID_SHIFT) + +#define CONNECTOR_DUAL_LINK_DVI_I_ENUM_ID2 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_I << OBJECT_ID_SHIFT) + +#define CONNECTOR_SINGLE_LINK_DVI_D_ENUM_ID1 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D << OBJECT_ID_SHIFT) + +#define CONNECTOR_SINGLE_LINK_DVI_D_ENUM_ID2 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D << OBJECT_ID_SHIFT) + +#define CONNECTOR_SINGLE_LINK_DVI_D_ENUM_ID3 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID3 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D << OBJECT_ID_SHIFT) + +#define CONNECTOR_SINGLE_LINK_DVI_D_ENUM_ID4 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID4 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D << OBJECT_ID_SHIFT) + +#define CONNECTOR_DUAL_LINK_DVI_D_ENUM_ID1 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_D << OBJECT_ID_SHIFT) + +#define CONNECTOR_DUAL_LINK_DVI_D_ENUM_ID2 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_D << OBJECT_ID_SHIFT) + +#define CONNECTOR_DUAL_LINK_DVI_D_ENUM_ID3 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID3 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_D << OBJECT_ID_SHIFT) + +#define CONNECTOR_VGA_ENUM_ID1 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_VGA << OBJECT_ID_SHIFT) + +#define CONNECTOR_VGA_ENUM_ID2 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_VGA << OBJECT_ID_SHIFT) + +#define CONNECTOR_COMPOSITE_ENUM_ID1 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_COMPOSITE << OBJECT_ID_SHIFT) + +#define CONNECTOR_COMPOSITE_ENUM_ID2 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_COMPOSITE << OBJECT_ID_SHIFT) + +#define CONNECTOR_SVIDEO_ENUM_ID1 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_SVIDEO << OBJECT_ID_SHIFT) + +#define CONNECTOR_SVIDEO_ENUM_ID2 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_SVIDEO << OBJECT_ID_SHIFT) + +#define CONNECTOR_YPbPr_ENUM_ID1 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_YPbPr << OBJECT_ID_SHIFT) + +#define CONNECTOR_YPbPr_ENUM_ID2 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_YPbPr << OBJECT_ID_SHIFT) + +#define CONNECTOR_D_CONNECTOR_ENUM_ID1 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_D_CONNECTOR << OBJECT_ID_SHIFT) + +#define CONNECTOR_D_CONNECTOR_ENUM_ID2 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_D_CONNECTOR << OBJECT_ID_SHIFT) + +#define CONNECTOR_9PIN_DIN_ENUM_ID1 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_9PIN_DIN << OBJECT_ID_SHIFT) + +#define CONNECTOR_9PIN_DIN_ENUM_ID2 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_9PIN_DIN << OBJECT_ID_SHIFT) + +#define CONNECTOR_SCART_ENUM_ID1 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_SCART << OBJECT_ID_SHIFT) + +#define CONNECTOR_SCART_ENUM_ID2 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_SCART << OBJECT_ID_SHIFT) + +#define CONNECTOR_HDMI_TYPE_A_ENUM_ID1 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_HDMI_TYPE_A << OBJECT_ID_SHIFT) + +#define CONNECTOR_HDMI_TYPE_A_ENUM_ID2 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_HDMI_TYPE_A << OBJECT_ID_SHIFT) + +#define CONNECTOR_HDMI_TYPE_A_ENUM_ID3 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID3 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_HDMI_TYPE_A << OBJECT_ID_SHIFT) + +#define CONNECTOR_HDMI_TYPE_B_ENUM_ID1 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_HDMI_TYPE_B << OBJECT_ID_SHIFT) + +#define CONNECTOR_HDMI_TYPE_B_ENUM_ID2 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_HDMI_TYPE_B << OBJECT_ID_SHIFT) + +#define CONNECTOR_7PIN_DIN_ENUM_ID1 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_7PIN_DIN << OBJECT_ID_SHIFT) + +#define CONNECTOR_7PIN_DIN_ENUM_ID2 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_7PIN_DIN << OBJECT_ID_SHIFT) + +#define CONNECTOR_PCIE_CONNECTOR_ENUM_ID1 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_PCIE_CONNECTOR << OBJECT_ID_SHIFT) + +#define CONNECTOR_PCIE_CONNECTOR_ENUM_ID2 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_PCIE_CONNECTOR << OBJECT_ID_SHIFT) + +#define CONNECTOR_CROSSFIRE_ENUM_ID1 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_CROSSFIRE << OBJECT_ID_SHIFT) + +#define CONNECTOR_CROSSFIRE_ENUM_ID2 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_CROSSFIRE << OBJECT_ID_SHIFT) + + +#define CONNECTOR_HARDCODE_DVI_ENUM_ID1 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_HARDCODE_DVI << OBJECT_ID_SHIFT) + +#define CONNECTOR_HARDCODE_DVI_ENUM_ID2 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_HARDCODE_DVI << OBJECT_ID_SHIFT) + +#define CONNECTOR_DISPLAYPORT_ENUM_ID1 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_DISPLAYPORT << OBJECT_ID_SHIFT) + +#define CONNECTOR_DISPLAYPORT_ENUM_ID2 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_DISPLAYPORT << OBJECT_ID_SHIFT) + +#define CONNECTOR_DISPLAYPORT_ENUM_ID3 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID3 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_DISPLAYPORT << OBJECT_ID_SHIFT) + +#define CONNECTOR_DISPLAYPORT_ENUM_ID4 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID4 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_DISPLAYPORT << OBJECT_ID_SHIFT) + +#define CONNECTOR_DISPLAYPORT_ENUM_ID5 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID5 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_DISPLAYPORT << OBJECT_ID_SHIFT) + +#define CONNECTOR_DISPLAYPORT_ENUM_ID6 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID6 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_DISPLAYPORT << OBJECT_ID_SHIFT) + +#define CONNECTOR_MXM_ENUM_ID1 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_MXM << OBJECT_ID_SHIFT) //Mapping to MXM_DP_A + +#define CONNECTOR_MXM_ENUM_ID2 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_MXM << OBJECT_ID_SHIFT) //Mapping to MXM_DP_B + +#define CONNECTOR_MXM_ENUM_ID3 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID3 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_MXM << OBJECT_ID_SHIFT) //Mapping to MXM_DP_C + +#define CONNECTOR_MXM_ENUM_ID4 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID4 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_MXM << OBJECT_ID_SHIFT) //Mapping to MXM_DP_D + +#define CONNECTOR_MXM_ENUM_ID5 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID5 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_MXM << OBJECT_ID_SHIFT) //Mapping to MXM_LVDS_TXxx + +#define CONNECTOR_MXM_ENUM_ID6 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID6 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_MXM << OBJECT_ID_SHIFT) //Mapping to MXM_LVDS_UXxx + +#define CONNECTOR_MXM_ENUM_ID7 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID7 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_MXM << OBJECT_ID_SHIFT) //Mapping to MXM_DAC + +#define CONNECTOR_LVDS_eDP_ENUM_ID1 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_LVDS_eDP << OBJECT_ID_SHIFT) + +#define CONNECTOR_LVDS_eDP_ENUM_ID2 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\ + CONNECTOR_OBJECT_ID_LVDS_eDP << OBJECT_ID_SHIFT) + +/****************************************************/ +/* Router Object ID definition - Shared with BIOS */ +/****************************************************/ +#define ROUTER_I2C_EXTENDER_CNTL_ENUM_ID1 ( GRAPH_OBJECT_TYPE_ROUTER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + ROUTER_OBJECT_ID_I2C_EXTENDER_CNTL << OBJECT_ID_SHIFT) + +/* deleted */ + +/****************************************************/ +/* Generic Object ID definition - Shared with BIOS */ +/****************************************************/ +#define GENERICOBJECT_GLSYNC_ENUM_ID1 (GRAPH_OBJECT_TYPE_GENERIC << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + GENERIC_OBJECT_ID_GLSYNC << OBJECT_ID_SHIFT) + +#define GENERICOBJECT_PX2_NON_DRIVABLE_ID1 (GRAPH_OBJECT_TYPE_GENERIC << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + GENERIC_OBJECT_ID_PX2_NON_DRIVABLE<< OBJECT_ID_SHIFT) + +#define GENERICOBJECT_PX2_NON_DRIVABLE_ID2 (GRAPH_OBJECT_TYPE_GENERIC << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\ + GENERIC_OBJECT_ID_PX2_NON_DRIVABLE<< OBJECT_ID_SHIFT) + +#define GENERICOBJECT_MXM_OPM_ENUM_ID1 (GRAPH_OBJECT_TYPE_GENERIC << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + GENERIC_OBJECT_ID_MXM_OPM << OBJECT_ID_SHIFT) + +#define GENERICOBJECT_STEREO_PIN_ENUM_ID1 (GRAPH_OBJECT_TYPE_GENERIC << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ + GENERIC_OBJECT_ID_STEREO_PIN << OBJECT_ID_SHIFT) + +/****************************************************/ +/* Object Cap definition - Shared with BIOS */ +/****************************************************/ +#define GRAPHICS_OBJECT_CAP_I2C 0x00000001L +#define GRAPHICS_OBJECT_CAP_TABLE_ID 0x00000002L + + +#define GRAPHICS_OBJECT_I2CCOMMAND_TABLE_ID 0x01 +#define GRAPHICS_OBJECT_HOTPLUGDETECTIONINTERUPT_TABLE_ID 0x02 +#define GRAPHICS_OBJECT_ENCODER_OUTPUT_PROTECTION_TABLE_ID 0x03 + +#if defined(_X86_) +#pragma pack() +#endif + +#endif /*GRAPHICTYPE */ + + + + diff --git a/drivers/gpu/drm/radeon/atom-bits.h b/drivers/gpu/drm/radeon/atom-bits.h new file mode 100644 index 00000000..e8fae5c7 --- /dev/null +++ b/drivers/gpu/drm/radeon/atom-bits.h @@ -0,0 +1,48 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Author: Stanislaw Skowronek + */ + +#ifndef ATOM_BITS_H +#define ATOM_BITS_H + +static inline uint8_t get_u8(void *bios, int ptr) +{ + return ((unsigned char *)bios)[ptr]; +} +#define U8(ptr) get_u8(ctx->ctx->bios, (ptr)) +#define CU8(ptr) get_u8(ctx->bios, (ptr)) +static inline uint16_t get_u16(void *bios, int ptr) +{ + return get_u8(bios ,ptr)|(((uint16_t)get_u8(bios, ptr+1))<<8); +} +#define U16(ptr) get_u16(ctx->ctx->bios, (ptr)) +#define CU16(ptr) get_u16(ctx->bios, (ptr)) +static inline uint32_t get_u32(void *bios, int ptr) +{ + return get_u16(bios, ptr)|(((uint32_t)get_u16(bios, ptr+2))<<16); +} +#define U32(ptr) get_u32(ctx->ctx->bios, (ptr)) +#define CU32(ptr) get_u32(ctx->bios, (ptr)) +#define CSTR(ptr) (((char *)(ctx->bios))+(ptr)) + +#endif diff --git a/drivers/gpu/drm/radeon/atom-names.h b/drivers/gpu/drm/radeon/atom-names.h new file mode 100644 index 00000000..6f907a5f --- /dev/null +++ b/drivers/gpu/drm/radeon/atom-names.h @@ -0,0 +1,100 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Author: Stanislaw Skowronek + */ + +#ifndef ATOM_NAMES_H +#define ATOM_NAMES_H + +#include "atom.h" + +#ifdef ATOM_DEBUG + +#define ATOM_OP_NAMES_CNT 123 +static char *atom_op_names[ATOM_OP_NAMES_CNT] = { +"RESERVED", "MOVE_REG", "MOVE_PS", "MOVE_WS", "MOVE_FB", "MOVE_PLL", +"MOVE_MC", "AND_REG", "AND_PS", "AND_WS", "AND_FB", "AND_PLL", "AND_MC", +"OR_REG", "OR_PS", "OR_WS", "OR_FB", "OR_PLL", "OR_MC", "SHIFT_LEFT_REG", +"SHIFT_LEFT_PS", "SHIFT_LEFT_WS", "SHIFT_LEFT_FB", "SHIFT_LEFT_PLL", +"SHIFT_LEFT_MC", "SHIFT_RIGHT_REG", "SHIFT_RIGHT_PS", "SHIFT_RIGHT_WS", +"SHIFT_RIGHT_FB", "SHIFT_RIGHT_PLL", "SHIFT_RIGHT_MC", "MUL_REG", +"MUL_PS", "MUL_WS", "MUL_FB", "MUL_PLL", "MUL_MC", "DIV_REG", "DIV_PS", +"DIV_WS", "DIV_FB", "DIV_PLL", "DIV_MC", "ADD_REG", "ADD_PS", "ADD_WS", +"ADD_FB", "ADD_PLL", "ADD_MC", "SUB_REG", "SUB_PS", "SUB_WS", "SUB_FB", +"SUB_PLL", "SUB_MC", "SET_ATI_PORT", "SET_PCI_PORT", "SET_SYS_IO_PORT", +"SET_REG_BLOCK", "SET_FB_BASE", "COMPARE_REG", "COMPARE_PS", +"COMPARE_WS", "COMPARE_FB", "COMPARE_PLL", "COMPARE_MC", "SWITCH", +"JUMP", "JUMP_EQUAL", "JUMP_BELOW", "JUMP_ABOVE", "JUMP_BELOW_OR_EQUAL", +"JUMP_ABOVE_OR_EQUAL", "JUMP_NOT_EQUAL", "TEST_REG", "TEST_PS", "TEST_WS", +"TEST_FB", "TEST_PLL", "TEST_MC", "DELAY_MILLISEC", "DELAY_MICROSEC", +"CALL_TABLE", "REPEAT", "CLEAR_REG", "CLEAR_PS", "CLEAR_WS", "CLEAR_FB", +"CLEAR_PLL", "CLEAR_MC", "NOP", "EOT", "MASK_REG", "MASK_PS", "MASK_WS", +"MASK_FB", "MASK_PLL", "MASK_MC", "POST_CARD", "BEEP", "SAVE_REG", +"RESTORE_REG", "SET_DATA_BLOCK", "XOR_REG", "XOR_PS", "XOR_WS", "XOR_FB", +"XOR_PLL", "XOR_MC", "SHL_REG", "SHL_PS", "SHL_WS", "SHL_FB", "SHL_PLL", +"SHL_MC", "SHR_REG", "SHR_PS", "SHR_WS", "SHR_FB", "SHR_PLL", "SHR_MC", +"DEBUG", "CTB_DS", +}; + +#define ATOM_TABLE_NAMES_CNT 74 +static char *atom_table_names[ATOM_TABLE_NAMES_CNT] = { +"ASIC_Init", "GetDisplaySurfaceSize", "ASIC_RegistersInit", +"VRAM_BlockVenderDetection", "SetClocksRatio", "MemoryControllerInit", +"GPIO_PinInit", "MemoryParamAdjust", "DVOEncoderControl", +"GPIOPinControl", "SetEngineClock", "SetMemoryClock", "SetPixelClock", +"DynamicClockGating", "ResetMemoryDLL", "ResetMemoryDevice", +"MemoryPLLInit", "EnableMemorySelfRefresh", "AdjustMemoryController", +"EnableASIC_StaticPwrMgt", "ASIC_StaticPwrMgtStatusChange", +"DAC_LoadDetection", "TMDS2EncoderControl", "LCD1OutputControl", +"DAC1EncoderControl", "DAC2EncoderControl", "DVOOutputControl", +"CV1OutputControl", "SetCRTC_DPM_State", "TVEncoderControl", +"TMDS1EncoderControl", "LVDSEncoderControl", "TV1OutputControl", +"EnableScaler", "BlankCRTC", "EnableCRTC", "GetPixelClock", +"EnableVGA_Render", "EnableVGA_Access", "SetCRTC_Timing", +"SetCRTC_OverScan", "SetCRTC_Replication", "SelectCRTC_Source", +"EnableGraphSurfaces", "UpdateCRTC_DoubleBufferRegisters", +"LUT_AutoFill", "EnableHW_IconCursor", "GetMemoryClock", +"GetEngineClock", "SetCRTC_UsingDTDTiming", "TVBootUpStdPinDetection", +"DFP2OutputControl", "VRAM_BlockDetectionByStrap", "MemoryCleanUp", +"ReadEDIDFromHWAssistedI2C", "WriteOneByteToHWAssistedI2C", +"ReadHWAssistedI2CStatus", "SpeedFanControl", "PowerConnectorDetection", +"MC_Synchronization", "ComputeMemoryEnginePLL", "MemoryRefreshConversion", +"VRAM_GetCurrentInfoBlock", "DynamicMemorySettings", "MemoryTraining", +"EnableLVDS_SS", "DFP1OutputControl", "SetVoltage", "CRT1OutputControl", +"CRT2OutputControl", "SetupHWAssistedI2CStatus", "ClockSource", +"MemoryDeviceInit", "EnableYUV", +}; + +#define ATOM_IO_NAMES_CNT 5 +static char *atom_io_names[ATOM_IO_NAMES_CNT] = { +"MM", "PLL", "MC", "PCIE", "PCIE PORT", +}; + +#else + +#define ATOM_OP_NAMES_CNT 0 +#define ATOM_TABLE_NAMES_CNT 0 +#define ATOM_IO_NAMES_CNT 0 + +#endif + +#endif diff --git a/drivers/gpu/drm/radeon/atom-types.h b/drivers/gpu/drm/radeon/atom-types.h new file mode 100644 index 00000000..1125b866 --- /dev/null +++ b/drivers/gpu/drm/radeon/atom-types.h @@ -0,0 +1,42 @@ +/* + * Copyright 2008 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Author: Dave Airlie + */ + +#ifndef ATOM_TYPES_H +#define ATOM_TYPES_H + +/* sync atom types to kernel types */ + +typedef uint16_t USHORT; +typedef uint32_t ULONG; +typedef uint8_t UCHAR; + + +#ifndef ATOM_BIG_ENDIAN +#if defined(__BIG_ENDIAN) +#define ATOM_BIG_ENDIAN 1 +#else +#define ATOM_BIG_ENDIAN 0 +#endif +#endif +#endif diff --git a/drivers/gpu/drm/radeon/atom.c b/drivers/gpu/drm/radeon/atom.c new file mode 100644 index 00000000..88661eaa --- /dev/null +++ b/drivers/gpu/drm/radeon/atom.c @@ -0,0 +1,1401 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Author: Stanislaw Skowronek + */ + +#include +#include +#include +#include + +#define ATOM_DEBUG + +#include "atom.h" +#include "atom-names.h" +#include "atom-bits.h" +#include "radeon.h" + +#define ATOM_COND_ABOVE 0 +#define ATOM_COND_ABOVEOREQUAL 1 +#define ATOM_COND_ALWAYS 2 +#define ATOM_COND_BELOW 3 +#define ATOM_COND_BELOWOREQUAL 4 +#define ATOM_COND_EQUAL 5 +#define ATOM_COND_NOTEQUAL 6 + +#define ATOM_PORT_ATI 0 +#define ATOM_PORT_PCI 1 +#define ATOM_PORT_SYSIO 2 + +#define ATOM_UNIT_MICROSEC 0 +#define ATOM_UNIT_MILLISEC 1 + +#define PLL_INDEX 2 +#define PLL_DATA 3 + +typedef struct { + struct atom_context *ctx; + uint32_t *ps, *ws; + int ps_shift; + uint16_t start; + unsigned last_jump; + unsigned long last_jump_jiffies; + bool abort; +} atom_exec_context; + +int atom_debug = 0; +static int atom_execute_table_locked(struct atom_context *ctx, int index, uint32_t * params); +int atom_execute_table(struct atom_context *ctx, int index, uint32_t * params); + +static uint32_t atom_arg_mask[8] = + { 0xFFFFFFFF, 0xFFFF, 0xFFFF00, 0xFFFF0000, 0xFF, 0xFF00, 0xFF0000, +0xFF000000 }; +static int atom_arg_shift[8] = { 0, 0, 8, 16, 0, 8, 16, 24 }; + +static int atom_dst_to_src[8][4] = { + /* translate destination alignment field to the source alignment encoding */ + {0, 0, 0, 0}, + {1, 2, 3, 0}, + {1, 2, 3, 0}, + {1, 2, 3, 0}, + {4, 5, 6, 7}, + {4, 5, 6, 7}, + {4, 5, 6, 7}, + {4, 5, 6, 7}, +}; +static int atom_def_dst[8] = { 0, 0, 1, 2, 0, 1, 2, 3 }; + +static int debug_depth = 0; +#ifdef ATOM_DEBUG +static void debug_print_spaces(int n) +{ + while (n--) + printk(" "); +} + +#define DEBUG(...) do if (atom_debug) { printk(KERN_DEBUG __VA_ARGS__); } while (0) +#define SDEBUG(...) do if (atom_debug) { printk(KERN_DEBUG); debug_print_spaces(debug_depth); printk(__VA_ARGS__); } while (0) +#else +#define DEBUG(...) do { } while (0) +#define SDEBUG(...) do { } while (0) +#endif + +static uint32_t atom_iio_execute(struct atom_context *ctx, int base, + uint32_t index, uint32_t data) +{ + struct radeon_device *rdev = ctx->card->dev->dev_private; + uint32_t temp = 0xCDCDCDCD; + + while (1) + switch (CU8(base)) { + case ATOM_IIO_NOP: + base++; + break; + case ATOM_IIO_READ: + temp = ctx->card->ioreg_read(ctx->card, CU16(base + 1)); + base += 3; + break; + case ATOM_IIO_WRITE: + if (rdev->family == CHIP_RV515) + (void)ctx->card->ioreg_read(ctx->card, CU16(base + 1)); + ctx->card->ioreg_write(ctx->card, CU16(base + 1), temp); + base += 3; + break; + case ATOM_IIO_CLEAR: + temp &= + ~((0xFFFFFFFF >> (32 - CU8(base + 1))) << + CU8(base + 2)); + base += 3; + break; + case ATOM_IIO_SET: + temp |= + (0xFFFFFFFF >> (32 - CU8(base + 1))) << CU8(base + + 2); + base += 3; + break; + case ATOM_IIO_MOVE_INDEX: + temp &= + ~((0xFFFFFFFF >> (32 - CU8(base + 1))) << + CU8(base + 3)); + temp |= + ((index >> CU8(base + 2)) & + (0xFFFFFFFF >> (32 - CU8(base + 1)))) << CU8(base + + 3); + base += 4; + break; + case ATOM_IIO_MOVE_DATA: + temp &= + ~((0xFFFFFFFF >> (32 - CU8(base + 1))) << + CU8(base + 3)); + temp |= + ((data >> CU8(base + 2)) & + (0xFFFFFFFF >> (32 - CU8(base + 1)))) << CU8(base + + 3); + base += 4; + break; + case ATOM_IIO_MOVE_ATTR: + temp &= + ~((0xFFFFFFFF >> (32 - CU8(base + 1))) << + CU8(base + 3)); + temp |= + ((ctx-> + io_attr >> CU8(base + 2)) & (0xFFFFFFFF >> (32 - + CU8 + (base + + + 1)))) + << CU8(base + 3); + base += 4; + break; + case ATOM_IIO_END: + return temp; + default: + printk(KERN_INFO "Unknown IIO opcode.\n"); + return 0; + } +} + +static uint32_t atom_get_src_int(atom_exec_context *ctx, uint8_t attr, + int *ptr, uint32_t *saved, int print) +{ + uint32_t idx, val = 0xCDCDCDCD, align, arg; + struct atom_context *gctx = ctx->ctx; + arg = attr & 7; + align = (attr >> 3) & 7; + switch (arg) { + case ATOM_ARG_REG: + idx = U16(*ptr); + (*ptr) += 2; + if (print) + DEBUG("REG[0x%04X]", idx); + idx += gctx->reg_block; + switch (gctx->io_mode) { + case ATOM_IO_MM: + val = gctx->card->reg_read(gctx->card, idx); + break; + case ATOM_IO_PCI: + printk(KERN_INFO + "PCI registers are not implemented.\n"); + return 0; + case ATOM_IO_SYSIO: + printk(KERN_INFO + "SYSIO registers are not implemented.\n"); + return 0; + default: + if (!(gctx->io_mode & 0x80)) { + printk(KERN_INFO "Bad IO mode.\n"); + return 0; + } + if (!gctx->iio[gctx->io_mode & 0x7F]) { + printk(KERN_INFO + "Undefined indirect IO read method %d.\n", + gctx->io_mode & 0x7F); + return 0; + } + val = + atom_iio_execute(gctx, + gctx->iio[gctx->io_mode & 0x7F], + idx, 0); + } + break; + case ATOM_ARG_PS: + idx = U8(*ptr); + (*ptr)++; + /* get_unaligned_le32 avoids unaligned accesses from atombios + * tables, noticed on a DEC Alpha. */ + val = get_unaligned_le32((u32 *)&ctx->ps[idx]); + if (print) + DEBUG("PS[0x%02X,0x%04X]", idx, val); + break; + case ATOM_ARG_WS: + idx = U8(*ptr); + (*ptr)++; + if (print) + DEBUG("WS[0x%02X]", idx); + switch (idx) { + case ATOM_WS_QUOTIENT: + val = gctx->divmul[0]; + break; + case ATOM_WS_REMAINDER: + val = gctx->divmul[1]; + break; + case ATOM_WS_DATAPTR: + val = gctx->data_block; + break; + case ATOM_WS_SHIFT: + val = gctx->shift; + break; + case ATOM_WS_OR_MASK: + val = 1 << gctx->shift; + break; + case ATOM_WS_AND_MASK: + val = ~(1 << gctx->shift); + break; + case ATOM_WS_FB_WINDOW: + val = gctx->fb_base; + break; + case ATOM_WS_ATTRIBUTES: + val = gctx->io_attr; + break; + case ATOM_WS_REGPTR: + val = gctx->reg_block; + break; + default: + val = ctx->ws[idx]; + } + break; + case ATOM_ARG_ID: + idx = U16(*ptr); + (*ptr) += 2; + if (print) { + if (gctx->data_block) + DEBUG("ID[0x%04X+%04X]", idx, gctx->data_block); + else + DEBUG("ID[0x%04X]", idx); + } + val = U32(idx + gctx->data_block); + break; + case ATOM_ARG_FB: + idx = U8(*ptr); + (*ptr)++; + if ((gctx->fb_base + (idx * 4)) > gctx->scratch_size_bytes) { + DRM_ERROR("ATOM: fb read beyond scratch region: %d vs. %d\n", + gctx->fb_base + (idx * 4), gctx->scratch_size_bytes); + val = 0; + } else + val = gctx->scratch[(gctx->fb_base / 4) + idx]; + if (print) + DEBUG("FB[0x%02X]", idx); + break; + case ATOM_ARG_IMM: + switch (align) { + case ATOM_SRC_DWORD: + val = U32(*ptr); + (*ptr) += 4; + if (print) + DEBUG("IMM 0x%08X\n", val); + return val; + case ATOM_SRC_WORD0: + case ATOM_SRC_WORD8: + case ATOM_SRC_WORD16: + val = U16(*ptr); + (*ptr) += 2; + if (print) + DEBUG("IMM 0x%04X\n", val); + return val; + case ATOM_SRC_BYTE0: + case ATOM_SRC_BYTE8: + case ATOM_SRC_BYTE16: + case ATOM_SRC_BYTE24: + val = U8(*ptr); + (*ptr)++; + if (print) + DEBUG("IMM 0x%02X\n", val); + return val; + } + return 0; + case ATOM_ARG_PLL: + idx = U8(*ptr); + (*ptr)++; + if (print) + DEBUG("PLL[0x%02X]", idx); + val = gctx->card->pll_read(gctx->card, idx); + break; + case ATOM_ARG_MC: + idx = U8(*ptr); + (*ptr)++; + if (print) + DEBUG("MC[0x%02X]", idx); + val = gctx->card->mc_read(gctx->card, idx); + break; + } + if (saved) + *saved = val; + val &= atom_arg_mask[align]; + val >>= atom_arg_shift[align]; + if (print) + switch (align) { + case ATOM_SRC_DWORD: + DEBUG(".[31:0] -> 0x%08X\n", val); + break; + case ATOM_SRC_WORD0: + DEBUG(".[15:0] -> 0x%04X\n", val); + break; + case ATOM_SRC_WORD8: + DEBUG(".[23:8] -> 0x%04X\n", val); + break; + case ATOM_SRC_WORD16: + DEBUG(".[31:16] -> 0x%04X\n", val); + break; + case ATOM_SRC_BYTE0: + DEBUG(".[7:0] -> 0x%02X\n", val); + break; + case ATOM_SRC_BYTE8: + DEBUG(".[15:8] -> 0x%02X\n", val); + break; + case ATOM_SRC_BYTE16: + DEBUG(".[23:16] -> 0x%02X\n", val); + break; + case ATOM_SRC_BYTE24: + DEBUG(".[31:24] -> 0x%02X\n", val); + break; + } + return val; +} + +static void atom_skip_src_int(atom_exec_context *ctx, uint8_t attr, int *ptr) +{ + uint32_t align = (attr >> 3) & 7, arg = attr & 7; + switch (arg) { + case ATOM_ARG_REG: + case ATOM_ARG_ID: + (*ptr) += 2; + break; + case ATOM_ARG_PLL: + case ATOM_ARG_MC: + case ATOM_ARG_PS: + case ATOM_ARG_WS: + case ATOM_ARG_FB: + (*ptr)++; + break; + case ATOM_ARG_IMM: + switch (align) { + case ATOM_SRC_DWORD: + (*ptr) += 4; + return; + case ATOM_SRC_WORD0: + case ATOM_SRC_WORD8: + case ATOM_SRC_WORD16: + (*ptr) += 2; + return; + case ATOM_SRC_BYTE0: + case ATOM_SRC_BYTE8: + case ATOM_SRC_BYTE16: + case ATOM_SRC_BYTE24: + (*ptr)++; + return; + } + return; + } +} + +static uint32_t atom_get_src(atom_exec_context *ctx, uint8_t attr, int *ptr) +{ + return atom_get_src_int(ctx, attr, ptr, NULL, 1); +} + +static uint32_t atom_get_src_direct(atom_exec_context *ctx, uint8_t align, int *ptr) +{ + uint32_t val = 0xCDCDCDCD; + + switch (align) { + case ATOM_SRC_DWORD: + val = U32(*ptr); + (*ptr) += 4; + break; + case ATOM_SRC_WORD0: + case ATOM_SRC_WORD8: + case ATOM_SRC_WORD16: + val = U16(*ptr); + (*ptr) += 2; + break; + case ATOM_SRC_BYTE0: + case ATOM_SRC_BYTE8: + case ATOM_SRC_BYTE16: + case ATOM_SRC_BYTE24: + val = U8(*ptr); + (*ptr)++; + break; + } + return val; +} + +static uint32_t atom_get_dst(atom_exec_context *ctx, int arg, uint8_t attr, + int *ptr, uint32_t *saved, int print) +{ + return atom_get_src_int(ctx, + arg | atom_dst_to_src[(attr >> 3) & + 7][(attr >> 6) & 3] << 3, + ptr, saved, print); +} + +static void atom_skip_dst(atom_exec_context *ctx, int arg, uint8_t attr, int *ptr) +{ + atom_skip_src_int(ctx, + arg | atom_dst_to_src[(attr >> 3) & 7][(attr >> 6) & + 3] << 3, ptr); +} + +static void atom_put_dst(atom_exec_context *ctx, int arg, uint8_t attr, + int *ptr, uint32_t val, uint32_t saved) +{ + uint32_t align = + atom_dst_to_src[(attr >> 3) & 7][(attr >> 6) & 3], old_val = + val, idx; + struct atom_context *gctx = ctx->ctx; + old_val &= atom_arg_mask[align] >> atom_arg_shift[align]; + val <<= atom_arg_shift[align]; + val &= atom_arg_mask[align]; + saved &= ~atom_arg_mask[align]; + val |= saved; + switch (arg) { + case ATOM_ARG_REG: + idx = U16(*ptr); + (*ptr) += 2; + DEBUG("REG[0x%04X]", idx); + idx += gctx->reg_block; + switch (gctx->io_mode) { + case ATOM_IO_MM: + if (idx == 0) + gctx->card->reg_write(gctx->card, idx, + val << 2); + else + gctx->card->reg_write(gctx->card, idx, val); + break; + case ATOM_IO_PCI: + printk(KERN_INFO + "PCI registers are not implemented.\n"); + return; + case ATOM_IO_SYSIO: + printk(KERN_INFO + "SYSIO registers are not implemented.\n"); + return; + default: + if (!(gctx->io_mode & 0x80)) { + printk(KERN_INFO "Bad IO mode.\n"); + return; + } + if (!gctx->iio[gctx->io_mode & 0xFF]) { + printk(KERN_INFO + "Undefined indirect IO write method %d.\n", + gctx->io_mode & 0x7F); + return; + } + atom_iio_execute(gctx, gctx->iio[gctx->io_mode & 0xFF], + idx, val); + } + break; + case ATOM_ARG_PS: + idx = U8(*ptr); + (*ptr)++; + DEBUG("PS[0x%02X]", idx); + ctx->ps[idx] = cpu_to_le32(val); + break; + case ATOM_ARG_WS: + idx = U8(*ptr); + (*ptr)++; + DEBUG("WS[0x%02X]", idx); + switch (idx) { + case ATOM_WS_QUOTIENT: + gctx->divmul[0] = val; + break; + case ATOM_WS_REMAINDER: + gctx->divmul[1] = val; + break; + case ATOM_WS_DATAPTR: + gctx->data_block = val; + break; + case ATOM_WS_SHIFT: + gctx->shift = val; + break; + case ATOM_WS_OR_MASK: + case ATOM_WS_AND_MASK: + break; + case ATOM_WS_FB_WINDOW: + gctx->fb_base = val; + break; + case ATOM_WS_ATTRIBUTES: + gctx->io_attr = val; + break; + case ATOM_WS_REGPTR: + gctx->reg_block = val; + break; + default: + ctx->ws[idx] = val; + } + break; + case ATOM_ARG_FB: + idx = U8(*ptr); + (*ptr)++; + if ((gctx->fb_base + (idx * 4)) > gctx->scratch_size_bytes) { + DRM_ERROR("ATOM: fb write beyond scratch region: %d vs. %d\n", + gctx->fb_base + (idx * 4), gctx->scratch_size_bytes); + } else + gctx->scratch[(gctx->fb_base / 4) + idx] = val; + DEBUG("FB[0x%02X]", idx); + break; + case ATOM_ARG_PLL: + idx = U8(*ptr); + (*ptr)++; + DEBUG("PLL[0x%02X]", idx); + gctx->card->pll_write(gctx->card, idx, val); + break; + case ATOM_ARG_MC: + idx = U8(*ptr); + (*ptr)++; + DEBUG("MC[0x%02X]", idx); + gctx->card->mc_write(gctx->card, idx, val); + return; + } + switch (align) { + case ATOM_SRC_DWORD: + DEBUG(".[31:0] <- 0x%08X\n", old_val); + break; + case ATOM_SRC_WORD0: + DEBUG(".[15:0] <- 0x%04X\n", old_val); + break; + case ATOM_SRC_WORD8: + DEBUG(".[23:8] <- 0x%04X\n", old_val); + break; + case ATOM_SRC_WORD16: + DEBUG(".[31:16] <- 0x%04X\n", old_val); + break; + case ATOM_SRC_BYTE0: + DEBUG(".[7:0] <- 0x%02X\n", old_val); + break; + case ATOM_SRC_BYTE8: + DEBUG(".[15:8] <- 0x%02X\n", old_val); + break; + case ATOM_SRC_BYTE16: + DEBUG(".[23:16] <- 0x%02X\n", old_val); + break; + case ATOM_SRC_BYTE24: + DEBUG(".[31:24] <- 0x%02X\n", old_val); + break; + } +} + +static void atom_op_add(atom_exec_context *ctx, int *ptr, int arg) +{ + uint8_t attr = U8((*ptr)++); + uint32_t dst, src, saved; + int dptr = *ptr; + SDEBUG(" dst: "); + dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1); + SDEBUG(" src: "); + src = atom_get_src(ctx, attr, ptr); + dst += src; + SDEBUG(" dst: "); + atom_put_dst(ctx, arg, attr, &dptr, dst, saved); +} + +static void atom_op_and(atom_exec_context *ctx, int *ptr, int arg) +{ + uint8_t attr = U8((*ptr)++); + uint32_t dst, src, saved; + int dptr = *ptr; + SDEBUG(" dst: "); + dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1); + SDEBUG(" src: "); + src = atom_get_src(ctx, attr, ptr); + dst &= src; + SDEBUG(" dst: "); + atom_put_dst(ctx, arg, attr, &dptr, dst, saved); +} + +static void atom_op_beep(atom_exec_context *ctx, int *ptr, int arg) +{ + printk("ATOM BIOS beeped!\n"); +} + +static void atom_op_calltable(atom_exec_context *ctx, int *ptr, int arg) +{ + int idx = U8((*ptr)++); + int r = 0; + + if (idx < ATOM_TABLE_NAMES_CNT) + SDEBUG(" table: %d (%s)\n", idx, atom_table_names[idx]); + else + SDEBUG(" table: %d\n", idx); + if (U16(ctx->ctx->cmd_table + 4 + 2 * idx)) + r = atom_execute_table_locked(ctx->ctx, idx, ctx->ps + ctx->ps_shift); + if (r) { + ctx->abort = true; + } +} + +static void atom_op_clear(atom_exec_context *ctx, int *ptr, int arg) +{ + uint8_t attr = U8((*ptr)++); + uint32_t saved; + int dptr = *ptr; + attr &= 0x38; + attr |= atom_def_dst[attr >> 3] << 6; + atom_get_dst(ctx, arg, attr, ptr, &saved, 0); + SDEBUG(" dst: "); + atom_put_dst(ctx, arg, attr, &dptr, 0, saved); +} + +static void atom_op_compare(atom_exec_context *ctx, int *ptr, int arg) +{ + uint8_t attr = U8((*ptr)++); + uint32_t dst, src; + SDEBUG(" src1: "); + dst = atom_get_dst(ctx, arg, attr, ptr, NULL, 1); + SDEBUG(" src2: "); + src = atom_get_src(ctx, attr, ptr); + ctx->ctx->cs_equal = (dst == src); + ctx->ctx->cs_above = (dst > src); + SDEBUG(" result: %s %s\n", ctx->ctx->cs_equal ? "EQ" : "NE", + ctx->ctx->cs_above ? "GT" : "LE"); +} + +static void atom_op_delay(atom_exec_context *ctx, int *ptr, int arg) +{ + unsigned count = U8((*ptr)++); + SDEBUG(" count: %d\n", count); + if (arg == ATOM_UNIT_MICROSEC) + udelay(count); + else + msleep(count); +} + +static void atom_op_div(atom_exec_context *ctx, int *ptr, int arg) +{ + uint8_t attr = U8((*ptr)++); + uint32_t dst, src; + SDEBUG(" src1: "); + dst = atom_get_dst(ctx, arg, attr, ptr, NULL, 1); + SDEBUG(" src2: "); + src = atom_get_src(ctx, attr, ptr); + if (src != 0) { + ctx->ctx->divmul[0] = dst / src; + ctx->ctx->divmul[1] = dst % src; + } else { + ctx->ctx->divmul[0] = 0; + ctx->ctx->divmul[1] = 0; + } +} + +static void atom_op_eot(atom_exec_context *ctx, int *ptr, int arg) +{ + /* functionally, a nop */ +} + +static void atom_op_jump(atom_exec_context *ctx, int *ptr, int arg) +{ + int execute = 0, target = U16(*ptr); + unsigned long cjiffies; + + (*ptr) += 2; + switch (arg) { + case ATOM_COND_ABOVE: + execute = ctx->ctx->cs_above; + break; + case ATOM_COND_ABOVEOREQUAL: + execute = ctx->ctx->cs_above || ctx->ctx->cs_equal; + break; + case ATOM_COND_ALWAYS: + execute = 1; + break; + case ATOM_COND_BELOW: + execute = !(ctx->ctx->cs_above || ctx->ctx->cs_equal); + break; + case ATOM_COND_BELOWOREQUAL: + execute = !ctx->ctx->cs_above; + break; + case ATOM_COND_EQUAL: + execute = ctx->ctx->cs_equal; + break; + case ATOM_COND_NOTEQUAL: + execute = !ctx->ctx->cs_equal; + break; + } + if (arg != ATOM_COND_ALWAYS) + SDEBUG(" taken: %s\n", execute ? "yes" : "no"); + SDEBUG(" target: 0x%04X\n", target); + if (execute) { + if (ctx->last_jump == (ctx->start + target)) { + cjiffies = jiffies; + if (time_after(cjiffies, ctx->last_jump_jiffies)) { + cjiffies -= ctx->last_jump_jiffies; + if ((jiffies_to_msecs(cjiffies) > 5000)) { + DRM_ERROR("atombios stuck in loop for more than 5secs aborting\n"); + ctx->abort = true; + } + } else { + /* jiffies wrap around we will just wait a little longer */ + ctx->last_jump_jiffies = jiffies; + } + } else { + ctx->last_jump = ctx->start + target; + ctx->last_jump_jiffies = jiffies; + } + *ptr = ctx->start + target; + } +} + +static void atom_op_mask(atom_exec_context *ctx, int *ptr, int arg) +{ + uint8_t attr = U8((*ptr)++); + uint32_t dst, mask, src, saved; + int dptr = *ptr; + SDEBUG(" dst: "); + dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1); + mask = atom_get_src_direct(ctx, ((attr >> 3) & 7), ptr); + SDEBUG(" mask: 0x%08x", mask); + SDEBUG(" src: "); + src = atom_get_src(ctx, attr, ptr); + dst &= mask; + dst |= src; + SDEBUG(" dst: "); + atom_put_dst(ctx, arg, attr, &dptr, dst, saved); +} + +static void atom_op_move(atom_exec_context *ctx, int *ptr, int arg) +{ + uint8_t attr = U8((*ptr)++); + uint32_t src, saved; + int dptr = *ptr; + if (((attr >> 3) & 7) != ATOM_SRC_DWORD) + atom_get_dst(ctx, arg, attr, ptr, &saved, 0); + else { + atom_skip_dst(ctx, arg, attr, ptr); + saved = 0xCDCDCDCD; + } + SDEBUG(" src: "); + src = atom_get_src(ctx, attr, ptr); + SDEBUG(" dst: "); + atom_put_dst(ctx, arg, attr, &dptr, src, saved); +} + +static void atom_op_mul(atom_exec_context *ctx, int *ptr, int arg) +{ + uint8_t attr = U8((*ptr)++); + uint32_t dst, src; + SDEBUG(" src1: "); + dst = atom_get_dst(ctx, arg, attr, ptr, NULL, 1); + SDEBUG(" src2: "); + src = atom_get_src(ctx, attr, ptr); + ctx->ctx->divmul[0] = dst * src; +} + +static void atom_op_nop(atom_exec_context *ctx, int *ptr, int arg) +{ + /* nothing */ +} + +static void atom_op_or(atom_exec_context *ctx, int *ptr, int arg) +{ + uint8_t attr = U8((*ptr)++); + uint32_t dst, src, saved; + int dptr = *ptr; + SDEBUG(" dst: "); + dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1); + SDEBUG(" src: "); + src = atom_get_src(ctx, attr, ptr); + dst |= src; + SDEBUG(" dst: "); + atom_put_dst(ctx, arg, attr, &dptr, dst, saved); +} + +static void atom_op_postcard(atom_exec_context *ctx, int *ptr, int arg) +{ + uint8_t val = U8((*ptr)++); + SDEBUG("POST card output: 0x%02X\n", val); +} + +static void atom_op_repeat(atom_exec_context *ctx, int *ptr, int arg) +{ + printk(KERN_INFO "unimplemented!\n"); +} + +static void atom_op_restorereg(atom_exec_context *ctx, int *ptr, int arg) +{ + printk(KERN_INFO "unimplemented!\n"); +} + +static void atom_op_savereg(atom_exec_context *ctx, int *ptr, int arg) +{ + printk(KERN_INFO "unimplemented!\n"); +} + +static void atom_op_setdatablock(atom_exec_context *ctx, int *ptr, int arg) +{ + int idx = U8(*ptr); + (*ptr)++; + SDEBUG(" block: %d\n", idx); + if (!idx) + ctx->ctx->data_block = 0; + else if (idx == 255) + ctx->ctx->data_block = ctx->start; + else + ctx->ctx->data_block = U16(ctx->ctx->data_table + 4 + 2 * idx); + SDEBUG(" base: 0x%04X\n", ctx->ctx->data_block); +} + +static void atom_op_setfbbase(atom_exec_context *ctx, int *ptr, int arg) +{ + uint8_t attr = U8((*ptr)++); + SDEBUG(" fb_base: "); + ctx->ctx->fb_base = atom_get_src(ctx, attr, ptr); +} + +static void atom_op_setport(atom_exec_context *ctx, int *ptr, int arg) +{ + int port; + switch (arg) { + case ATOM_PORT_ATI: + port = U16(*ptr); + if (port < ATOM_IO_NAMES_CNT) + SDEBUG(" port: %d (%s)\n", port, atom_io_names[port]); + else + SDEBUG(" port: %d\n", port); + if (!port) + ctx->ctx->io_mode = ATOM_IO_MM; + else + ctx->ctx->io_mode = ATOM_IO_IIO | port; + (*ptr) += 2; + break; + case ATOM_PORT_PCI: + ctx->ctx->io_mode = ATOM_IO_PCI; + (*ptr)++; + break; + case ATOM_PORT_SYSIO: + ctx->ctx->io_mode = ATOM_IO_SYSIO; + (*ptr)++; + break; + } +} + +static void atom_op_setregblock(atom_exec_context *ctx, int *ptr, int arg) +{ + ctx->ctx->reg_block = U16(*ptr); + (*ptr) += 2; + SDEBUG(" base: 0x%04X\n", ctx->ctx->reg_block); +} + +static void atom_op_shift_left(atom_exec_context *ctx, int *ptr, int arg) +{ + uint8_t attr = U8((*ptr)++), shift; + uint32_t saved, dst; + int dptr = *ptr; + attr &= 0x38; + attr |= atom_def_dst[attr >> 3] << 6; + SDEBUG(" dst: "); + dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1); + shift = atom_get_src_direct(ctx, ATOM_SRC_BYTE0, ptr); + SDEBUG(" shift: %d\n", shift); + dst <<= shift; + SDEBUG(" dst: "); + atom_put_dst(ctx, arg, attr, &dptr, dst, saved); +} + +static void atom_op_shift_right(atom_exec_context *ctx, int *ptr, int arg) +{ + uint8_t attr = U8((*ptr)++), shift; + uint32_t saved, dst; + int dptr = *ptr; + attr &= 0x38; + attr |= atom_def_dst[attr >> 3] << 6; + SDEBUG(" dst: "); + dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1); + shift = atom_get_src_direct(ctx, ATOM_SRC_BYTE0, ptr); + SDEBUG(" shift: %d\n", shift); + dst >>= shift; + SDEBUG(" dst: "); + atom_put_dst(ctx, arg, attr, &dptr, dst, saved); +} + +static void atom_op_shl(atom_exec_context *ctx, int *ptr, int arg) +{ + uint8_t attr = U8((*ptr)++), shift; + uint32_t saved, dst; + int dptr = *ptr; + uint32_t dst_align = atom_dst_to_src[(attr >> 3) & 7][(attr >> 6) & 3]; + SDEBUG(" dst: "); + dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1); + /* op needs to full dst value */ + dst = saved; + shift = atom_get_src(ctx, attr, ptr); + SDEBUG(" shift: %d\n", shift); + dst <<= shift; + dst &= atom_arg_mask[dst_align]; + dst >>= atom_arg_shift[dst_align]; + SDEBUG(" dst: "); + atom_put_dst(ctx, arg, attr, &dptr, dst, saved); +} + +static void atom_op_shr(atom_exec_context *ctx, int *ptr, int arg) +{ + uint8_t attr = U8((*ptr)++), shift; + uint32_t saved, dst; + int dptr = *ptr; + uint32_t dst_align = atom_dst_to_src[(attr >> 3) & 7][(attr >> 6) & 3]; + SDEBUG(" dst: "); + dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1); + /* op needs to full dst value */ + dst = saved; + shift = atom_get_src(ctx, attr, ptr); + SDEBUG(" shift: %d\n", shift); + dst >>= shift; + dst &= atom_arg_mask[dst_align]; + dst >>= atom_arg_shift[dst_align]; + SDEBUG(" dst: "); + atom_put_dst(ctx, arg, attr, &dptr, dst, saved); +} + +static void atom_op_sub(atom_exec_context *ctx, int *ptr, int arg) +{ + uint8_t attr = U8((*ptr)++); + uint32_t dst, src, saved; + int dptr = *ptr; + SDEBUG(" dst: "); + dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1); + SDEBUG(" src: "); + src = atom_get_src(ctx, attr, ptr); + dst -= src; + SDEBUG(" dst: "); + atom_put_dst(ctx, arg, attr, &dptr, dst, saved); +} + +static void atom_op_switch(atom_exec_context *ctx, int *ptr, int arg) +{ + uint8_t attr = U8((*ptr)++); + uint32_t src, val, target; + SDEBUG(" switch: "); + src = atom_get_src(ctx, attr, ptr); + while (U16(*ptr) != ATOM_CASE_END) + if (U8(*ptr) == ATOM_CASE_MAGIC) { + (*ptr)++; + SDEBUG(" case: "); + val = + atom_get_src(ctx, (attr & 0x38) | ATOM_ARG_IMM, + ptr); + target = U16(*ptr); + if (val == src) { + SDEBUG(" target: %04X\n", target); + *ptr = ctx->start + target; + return; + } + (*ptr) += 2; + } else { + printk(KERN_INFO "Bad case.\n"); + return; + } + (*ptr) += 2; +} + +static void atom_op_test(atom_exec_context *ctx, int *ptr, int arg) +{ + uint8_t attr = U8((*ptr)++); + uint32_t dst, src; + SDEBUG(" src1: "); + dst = atom_get_dst(ctx, arg, attr, ptr, NULL, 1); + SDEBUG(" src2: "); + src = atom_get_src(ctx, attr, ptr); + ctx->ctx->cs_equal = ((dst & src) == 0); + SDEBUG(" result: %s\n", ctx->ctx->cs_equal ? "EQ" : "NE"); +} + +static void atom_op_xor(atom_exec_context *ctx, int *ptr, int arg) +{ + uint8_t attr = U8((*ptr)++); + uint32_t dst, src, saved; + int dptr = *ptr; + SDEBUG(" dst: "); + dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1); + SDEBUG(" src: "); + src = atom_get_src(ctx, attr, ptr); + dst ^= src; + SDEBUG(" dst: "); + atom_put_dst(ctx, arg, attr, &dptr, dst, saved); +} + +static void atom_op_debug(atom_exec_context *ctx, int *ptr, int arg) +{ + printk(KERN_INFO "unimplemented!\n"); +} + +static struct { + void (*func) (atom_exec_context *, int *, int); + int arg; +} opcode_table[ATOM_OP_CNT] = { + { + NULL, 0}, { + atom_op_move, ATOM_ARG_REG}, { + atom_op_move, ATOM_ARG_PS}, { + atom_op_move, ATOM_ARG_WS}, { + atom_op_move, ATOM_ARG_FB}, { + atom_op_move, ATOM_ARG_PLL}, { + atom_op_move, ATOM_ARG_MC}, { + atom_op_and, ATOM_ARG_REG}, { + atom_op_and, ATOM_ARG_PS}, { + atom_op_and, ATOM_ARG_WS}, { + atom_op_and, ATOM_ARG_FB}, { + atom_op_and, ATOM_ARG_PLL}, { + atom_op_and, ATOM_ARG_MC}, { + atom_op_or, ATOM_ARG_REG}, { + atom_op_or, ATOM_ARG_PS}, { + atom_op_or, ATOM_ARG_WS}, { + atom_op_or, ATOM_ARG_FB}, { + atom_op_or, ATOM_ARG_PLL}, { + atom_op_or, ATOM_ARG_MC}, { + atom_op_shift_left, ATOM_ARG_REG}, { + atom_op_shift_left, ATOM_ARG_PS}, { + atom_op_shift_left, ATOM_ARG_WS}, { + atom_op_shift_left, ATOM_ARG_FB}, { + atom_op_shift_left, ATOM_ARG_PLL}, { + atom_op_shift_left, ATOM_ARG_MC}, { + atom_op_shift_right, ATOM_ARG_REG}, { + atom_op_shift_right, ATOM_ARG_PS}, { + atom_op_shift_right, ATOM_ARG_WS}, { + atom_op_shift_right, ATOM_ARG_FB}, { + atom_op_shift_right, ATOM_ARG_PLL}, { + atom_op_shift_right, ATOM_ARG_MC}, { + atom_op_mul, ATOM_ARG_REG}, { + atom_op_mul, ATOM_ARG_PS}, { + atom_op_mul, ATOM_ARG_WS}, { + atom_op_mul, ATOM_ARG_FB}, { + atom_op_mul, ATOM_ARG_PLL}, { + atom_op_mul, ATOM_ARG_MC}, { + atom_op_div, ATOM_ARG_REG}, { + atom_op_div, ATOM_ARG_PS}, { + atom_op_div, ATOM_ARG_WS}, { + atom_op_div, ATOM_ARG_FB}, { + atom_op_div, ATOM_ARG_PLL}, { + atom_op_div, ATOM_ARG_MC}, { + atom_op_add, ATOM_ARG_REG}, { + atom_op_add, ATOM_ARG_PS}, { + atom_op_add, ATOM_ARG_WS}, { + atom_op_add, ATOM_ARG_FB}, { + atom_op_add, ATOM_ARG_PLL}, { + atom_op_add, ATOM_ARG_MC}, { + atom_op_sub, ATOM_ARG_REG}, { + atom_op_sub, ATOM_ARG_PS}, { + atom_op_sub, ATOM_ARG_WS}, { + atom_op_sub, ATOM_ARG_FB}, { + atom_op_sub, ATOM_ARG_PLL}, { + atom_op_sub, ATOM_ARG_MC}, { + atom_op_setport, ATOM_PORT_ATI}, { + atom_op_setport, ATOM_PORT_PCI}, { + atom_op_setport, ATOM_PORT_SYSIO}, { + atom_op_setregblock, 0}, { + atom_op_setfbbase, 0}, { + atom_op_compare, ATOM_ARG_REG}, { + atom_op_compare, ATOM_ARG_PS}, { + atom_op_compare, ATOM_ARG_WS}, { + atom_op_compare, ATOM_ARG_FB}, { + atom_op_compare, ATOM_ARG_PLL}, { + atom_op_compare, ATOM_ARG_MC}, { + atom_op_switch, 0}, { + atom_op_jump, ATOM_COND_ALWAYS}, { + atom_op_jump, ATOM_COND_EQUAL}, { + atom_op_jump, ATOM_COND_BELOW}, { + atom_op_jump, ATOM_COND_ABOVE}, { + atom_op_jump, ATOM_COND_BELOWOREQUAL}, { + atom_op_jump, ATOM_COND_ABOVEOREQUAL}, { + atom_op_jump, ATOM_COND_NOTEQUAL}, { + atom_op_test, ATOM_ARG_REG}, { + atom_op_test, ATOM_ARG_PS}, { + atom_op_test, ATOM_ARG_WS}, { + atom_op_test, ATOM_ARG_FB}, { + atom_op_test, ATOM_ARG_PLL}, { + atom_op_test, ATOM_ARG_MC}, { + atom_op_delay, ATOM_UNIT_MILLISEC}, { + atom_op_delay, ATOM_UNIT_MICROSEC}, { + atom_op_calltable, 0}, { + atom_op_repeat, 0}, { + atom_op_clear, ATOM_ARG_REG}, { + atom_op_clear, ATOM_ARG_PS}, { + atom_op_clear, ATOM_ARG_WS}, { + atom_op_clear, ATOM_ARG_FB}, { + atom_op_clear, ATOM_ARG_PLL}, { + atom_op_clear, ATOM_ARG_MC}, { + atom_op_nop, 0}, { + atom_op_eot, 0}, { + atom_op_mask, ATOM_ARG_REG}, { + atom_op_mask, ATOM_ARG_PS}, { + atom_op_mask, ATOM_ARG_WS}, { + atom_op_mask, ATOM_ARG_FB}, { + atom_op_mask, ATOM_ARG_PLL}, { + atom_op_mask, ATOM_ARG_MC}, { + atom_op_postcard, 0}, { + atom_op_beep, 0}, { + atom_op_savereg, 0}, { + atom_op_restorereg, 0}, { + atom_op_setdatablock, 0}, { + atom_op_xor, ATOM_ARG_REG}, { + atom_op_xor, ATOM_ARG_PS}, { + atom_op_xor, ATOM_ARG_WS}, { + atom_op_xor, ATOM_ARG_FB}, { + atom_op_xor, ATOM_ARG_PLL}, { + atom_op_xor, ATOM_ARG_MC}, { + atom_op_shl, ATOM_ARG_REG}, { + atom_op_shl, ATOM_ARG_PS}, { + atom_op_shl, ATOM_ARG_WS}, { + atom_op_shl, ATOM_ARG_FB}, { + atom_op_shl, ATOM_ARG_PLL}, { + atom_op_shl, ATOM_ARG_MC}, { + atom_op_shr, ATOM_ARG_REG}, { + atom_op_shr, ATOM_ARG_PS}, { + atom_op_shr, ATOM_ARG_WS}, { + atom_op_shr, ATOM_ARG_FB}, { + atom_op_shr, ATOM_ARG_PLL}, { + atom_op_shr, ATOM_ARG_MC}, { +atom_op_debug, 0},}; + +static int atom_execute_table_locked(struct atom_context *ctx, int index, uint32_t * params) +{ + int base = CU16(ctx->cmd_table + 4 + 2 * index); + int len, ws, ps, ptr; + unsigned char op; + atom_exec_context ectx; + int ret = 0; + + if (!base) + return -EINVAL; + + len = CU16(base + ATOM_CT_SIZE_PTR); + ws = CU8(base + ATOM_CT_WS_PTR); + ps = CU8(base + ATOM_CT_PS_PTR) & ATOM_CT_PS_MASK; + ptr = base + ATOM_CT_CODE_PTR; + + SDEBUG(">> execute %04X (len %d, WS %d, PS %d)\n", base, len, ws, ps); + + ectx.ctx = ctx; + ectx.ps_shift = ps / 4; + ectx.start = base; + ectx.ps = params; + ectx.abort = false; + ectx.last_jump = 0; + if (ws) + ectx.ws = kzalloc(4 * ws, GFP_KERNEL); + else + ectx.ws = NULL; + + debug_depth++; + while (1) { + op = CU8(ptr++); + if (op < ATOM_OP_NAMES_CNT) + SDEBUG("%s @ 0x%04X\n", atom_op_names[op], ptr - 1); + else + SDEBUG("[%d] @ 0x%04X\n", op, ptr - 1); + if (ectx.abort) { + DRM_ERROR("atombios stuck executing %04X (len %d, WS %d, PS %d) @ 0x%04X\n", + base, len, ws, ps, ptr - 1); + ret = -EINVAL; + goto free; + } + + if (op < ATOM_OP_CNT && op > 0) + opcode_table[op].func(&ectx, &ptr, + opcode_table[op].arg); + else + break; + + if (op == ATOM_OP_EOT) + break; + } + debug_depth--; + SDEBUG("<<\n"); + +free: + if (ws) + kfree(ectx.ws); + return ret; +} + +int atom_execute_table(struct atom_context *ctx, int index, uint32_t * params) +{ + int r; + + mutex_lock(&ctx->mutex); + /* reset reg block */ + ctx->reg_block = 0; + /* reset fb window */ + ctx->fb_base = 0; + /* reset io mode */ + ctx->io_mode = ATOM_IO_MM; + r = atom_execute_table_locked(ctx, index, params); + mutex_unlock(&ctx->mutex); + return r; +} + +static int atom_iio_len[] = { 1, 2, 3, 3, 3, 3, 4, 4, 4, 3 }; + +static void atom_index_iio(struct atom_context *ctx, int base) +{ + ctx->iio = kzalloc(2 * 256, GFP_KERNEL); + while (CU8(base) == ATOM_IIO_START) { + ctx->iio[CU8(base + 1)] = base + 2; + base += 2; + while (CU8(base) != ATOM_IIO_END) + base += atom_iio_len[CU8(base)]; + base += 3; + } +} + +struct atom_context *atom_parse(struct card_info *card, void *bios) +{ + int base; + struct atom_context *ctx = + kzalloc(sizeof(struct atom_context), GFP_KERNEL); + char *str; + char name[512]; + int i; + + ctx->card = card; + ctx->bios = bios; + + if (CU16(0) != ATOM_BIOS_MAGIC) { + printk(KERN_INFO "Invalid BIOS magic.\n"); + kfree(ctx); + return NULL; + } + if (strncmp + (CSTR(ATOM_ATI_MAGIC_PTR), ATOM_ATI_MAGIC, + strlen(ATOM_ATI_MAGIC))) { + printk(KERN_INFO "Invalid ATI magic.\n"); + kfree(ctx); + return NULL; + } + + base = CU16(ATOM_ROM_TABLE_PTR); + if (strncmp + (CSTR(base + ATOM_ROM_MAGIC_PTR), ATOM_ROM_MAGIC, + strlen(ATOM_ROM_MAGIC))) { + printk(KERN_INFO "Invalid ATOM magic.\n"); + kfree(ctx); + return NULL; + } + + ctx->cmd_table = CU16(base + ATOM_ROM_CMD_PTR); + ctx->data_table = CU16(base + ATOM_ROM_DATA_PTR); + atom_index_iio(ctx, CU16(ctx->data_table + ATOM_DATA_IIO_PTR) + 4); + + str = CSTR(CU16(base + ATOM_ROM_MSG_PTR)); + while (*str && ((*str == '\n') || (*str == '\r'))) + str++; + /* name string isn't always 0 terminated */ + for (i = 0; i < 511; i++) { + name[i] = str[i]; + if (name[i] < '.' || name[i] > 'z') { + name[i] = 0; + break; + } + } + printk(KERN_INFO "ATOM BIOS: %s\n", name); + + return ctx; +} + +int atom_asic_init(struct atom_context *ctx) +{ + struct radeon_device *rdev = ctx->card->dev->dev_private; + int hwi = CU16(ctx->data_table + ATOM_DATA_FWI_PTR); + uint32_t ps[16]; + int ret; + + memset(ps, 0, 64); + + ps[0] = cpu_to_le32(CU32(hwi + ATOM_FWI_DEFSCLK_PTR)); + ps[1] = cpu_to_le32(CU32(hwi + ATOM_FWI_DEFMCLK_PTR)); + if (!ps[0] || !ps[1]) + return 1; + + if (!CU16(ctx->cmd_table + 4 + 2 * ATOM_CMD_INIT)) + return 1; + ret = atom_execute_table(ctx, ATOM_CMD_INIT, ps); + if (ret) + return ret; + + memset(ps, 0, 64); + + if (rdev->family < CHIP_R600) { + if (CU16(ctx->cmd_table + 4 + 2 * ATOM_CMD_SPDFANCNTL)) + atom_execute_table(ctx, ATOM_CMD_SPDFANCNTL, ps); + } + return ret; +} + +void atom_destroy(struct atom_context *ctx) +{ + if (ctx->iio) + kfree(ctx->iio); + kfree(ctx); +} + +bool atom_parse_data_header(struct atom_context *ctx, int index, + uint16_t * size, uint8_t * frev, uint8_t * crev, + uint16_t * data_start) +{ + int offset = index * 2 + 4; + int idx = CU16(ctx->data_table + offset); + u16 *mdt = (u16 *)(ctx->bios + ctx->data_table + 4); + + if (!mdt[index]) + return false; + + if (size) + *size = CU16(idx); + if (frev) + *frev = CU8(idx + 2); + if (crev) + *crev = CU8(idx + 3); + *data_start = idx; + return true; +} + +bool atom_parse_cmd_header(struct atom_context *ctx, int index, uint8_t * frev, + uint8_t * crev) +{ + int offset = index * 2 + 4; + int idx = CU16(ctx->cmd_table + offset); + u16 *mct = (u16 *)(ctx->bios + ctx->cmd_table + 4); + + if (!mct[index]) + return false; + + if (frev) + *frev = CU8(idx + 2); + if (crev) + *crev = CU8(idx + 3); + return true; +} + +int atom_allocate_fb_scratch(struct atom_context *ctx) +{ + int index = GetIndexIntoMasterTable(DATA, VRAM_UsageByFirmware); + uint16_t data_offset; + int usage_bytes = 0; + struct _ATOM_VRAM_USAGE_BY_FIRMWARE *firmware_usage; + + if (atom_parse_data_header(ctx, index, NULL, NULL, NULL, &data_offset)) { + firmware_usage = (struct _ATOM_VRAM_USAGE_BY_FIRMWARE *)(ctx->bios + data_offset); + + DRM_DEBUG("atom firmware requested %08x %dkb\n", + firmware_usage->asFirmwareVramReserveInfo[0].ulStartAddrUsedByFirmware, + firmware_usage->asFirmwareVramReserveInfo[0].usFirmwareUseInKb); + + usage_bytes = firmware_usage->asFirmwareVramReserveInfo[0].usFirmwareUseInKb * 1024; + } + ctx->scratch_size_bytes = 0; + if (usage_bytes == 0) + usage_bytes = 20 * 1024; + /* allocate some scratch memory */ + ctx->scratch = kzalloc(usage_bytes, GFP_KERNEL); + if (!ctx->scratch) + return -ENOMEM; + ctx->scratch_size_bytes = usage_bytes; + return 0; +} diff --git a/drivers/gpu/drm/radeon/atom.h b/drivers/gpu/drm/radeon/atom.h new file mode 100644 index 00000000..25fea631 --- /dev/null +++ b/drivers/gpu/drm/radeon/atom.h @@ -0,0 +1,159 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Author: Stanislaw Skowronek + */ + +#ifndef ATOM_H +#define ATOM_H + +#include +#include "drmP.h" + +#define ATOM_BIOS_MAGIC 0xAA55 +#define ATOM_ATI_MAGIC_PTR 0x30 +#define ATOM_ATI_MAGIC " 761295520" +#define ATOM_ROM_TABLE_PTR 0x48 + +#define ATOM_ROM_MAGIC "ATOM" +#define ATOM_ROM_MAGIC_PTR 4 + +#define ATOM_ROM_MSG_PTR 0x10 +#define ATOM_ROM_CMD_PTR 0x1E +#define ATOM_ROM_DATA_PTR 0x20 + +#define ATOM_CMD_INIT 0 +#define ATOM_CMD_SETSCLK 0x0A +#define ATOM_CMD_SETMCLK 0x0B +#define ATOM_CMD_SETPCLK 0x0C +#define ATOM_CMD_SPDFANCNTL 0x39 + +#define ATOM_DATA_FWI_PTR 0xC +#define ATOM_DATA_IIO_PTR 0x32 + +#define ATOM_FWI_DEFSCLK_PTR 8 +#define ATOM_FWI_DEFMCLK_PTR 0xC +#define ATOM_FWI_MAXSCLK_PTR 0x24 +#define ATOM_FWI_MAXMCLK_PTR 0x28 + +#define ATOM_CT_SIZE_PTR 0 +#define ATOM_CT_WS_PTR 4 +#define ATOM_CT_PS_PTR 5 +#define ATOM_CT_PS_MASK 0x7F +#define ATOM_CT_CODE_PTR 6 + +#define ATOM_OP_CNT 123 +#define ATOM_OP_EOT 91 + +#define ATOM_CASE_MAGIC 0x63 +#define ATOM_CASE_END 0x5A5A + +#define ATOM_ARG_REG 0 +#define ATOM_ARG_PS 1 +#define ATOM_ARG_WS 2 +#define ATOM_ARG_FB 3 +#define ATOM_ARG_ID 4 +#define ATOM_ARG_IMM 5 +#define ATOM_ARG_PLL 6 +#define ATOM_ARG_MC 7 + +#define ATOM_SRC_DWORD 0 +#define ATOM_SRC_WORD0 1 +#define ATOM_SRC_WORD8 2 +#define ATOM_SRC_WORD16 3 +#define ATOM_SRC_BYTE0 4 +#define ATOM_SRC_BYTE8 5 +#define ATOM_SRC_BYTE16 6 +#define ATOM_SRC_BYTE24 7 + +#define ATOM_WS_QUOTIENT 0x40 +#define ATOM_WS_REMAINDER 0x41 +#define ATOM_WS_DATAPTR 0x42 +#define ATOM_WS_SHIFT 0x43 +#define ATOM_WS_OR_MASK 0x44 +#define ATOM_WS_AND_MASK 0x45 +#define ATOM_WS_FB_WINDOW 0x46 +#define ATOM_WS_ATTRIBUTES 0x47 +#define ATOM_WS_REGPTR 0x48 + +#define ATOM_IIO_NOP 0 +#define ATOM_IIO_START 1 +#define ATOM_IIO_READ 2 +#define ATOM_IIO_WRITE 3 +#define ATOM_IIO_CLEAR 4 +#define ATOM_IIO_SET 5 +#define ATOM_IIO_MOVE_INDEX 6 +#define ATOM_IIO_MOVE_ATTR 7 +#define ATOM_IIO_MOVE_DATA 8 +#define ATOM_IIO_END 9 + +#define ATOM_IO_MM 0 +#define ATOM_IO_PCI 1 +#define ATOM_IO_SYSIO 2 +#define ATOM_IO_IIO 0x80 + +struct card_info { + struct drm_device *dev; + void (* reg_write)(struct card_info *, uint32_t, uint32_t); /* filled by driver */ + uint32_t (* reg_read)(struct card_info *, uint32_t); /* filled by driver */ + void (* ioreg_write)(struct card_info *, uint32_t, uint32_t); /* filled by driver */ + uint32_t (* ioreg_read)(struct card_info *, uint32_t); /* filled by driver */ + void (* mc_write)(struct card_info *, uint32_t, uint32_t); /* filled by driver */ + uint32_t (* mc_read)(struct card_info *, uint32_t); /* filled by driver */ + void (* pll_write)(struct card_info *, uint32_t, uint32_t); /* filled by driver */ + uint32_t (* pll_read)(struct card_info *, uint32_t); /* filled by driver */ +}; + +struct atom_context { + struct card_info *card; + struct mutex mutex; + void *bios; + uint32_t cmd_table, data_table; + uint16_t *iio; + + uint16_t data_block; + uint32_t fb_base; + uint32_t divmul[2]; + uint16_t io_attr; + uint16_t reg_block; + uint8_t shift; + int cs_equal, cs_above; + int io_mode; + uint32_t *scratch; + int scratch_size_bytes; +}; + +extern int atom_debug; + +struct atom_context *atom_parse(struct card_info *, void *); +int atom_execute_table(struct atom_context *, int, uint32_t *); +int atom_asic_init(struct atom_context *); +void atom_destroy(struct atom_context *); +bool atom_parse_data_header(struct atom_context *ctx, int index, uint16_t *size, + uint8_t *frev, uint8_t *crev, uint16_t *data_start); +bool atom_parse_cmd_header(struct atom_context *ctx, int index, + uint8_t *frev, uint8_t *crev); +int atom_allocate_fb_scratch(struct atom_context *ctx); +#include "atom-types.h" +#include "atombios.h" +#include "ObjectID.h" + +#endif diff --git a/drivers/gpu/drm/radeon/atombios.h b/drivers/gpu/drm/radeon/atombios.h new file mode 100644 index 00000000..1b50ad89 --- /dev/null +++ b/drivers/gpu/drm/radeon/atombios.h @@ -0,0 +1,7023 @@ +/* + * Copyright 2006-2007 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + + +/****************************************************************************/ +/*Portion I: Definitions shared between VBIOS and Driver */ +/****************************************************************************/ + + +#ifndef _ATOMBIOS_H +#define _ATOMBIOS_H + +#define ATOM_VERSION_MAJOR 0x00020000 +#define ATOM_VERSION_MINOR 0x00000002 + +#define ATOM_HEADER_VERSION (ATOM_VERSION_MAJOR | ATOM_VERSION_MINOR) + +/* Endianness should be specified before inclusion, + * default to little endian + */ +#ifndef ATOM_BIG_ENDIAN +#error Endian not specified +#endif + +#ifdef _H2INC + #ifndef ULONG + typedef unsigned long ULONG; + #endif + + #ifndef UCHAR + typedef unsigned char UCHAR; + #endif + + #ifndef USHORT + typedef unsigned short USHORT; + #endif +#endif + +#define ATOM_DAC_A 0 +#define ATOM_DAC_B 1 +#define ATOM_EXT_DAC 2 + +#define ATOM_CRTC1 0 +#define ATOM_CRTC2 1 +#define ATOM_CRTC3 2 +#define ATOM_CRTC4 3 +#define ATOM_CRTC5 4 +#define ATOM_CRTC6 5 +#define ATOM_CRTC_INVALID 0xFF + +#define ATOM_DIGA 0 +#define ATOM_DIGB 1 + +#define ATOM_PPLL1 0 +#define ATOM_PPLL2 1 +#define ATOM_DCPLL 2 +#define ATOM_PPLL0 2 +#define ATOM_EXT_PLL1 8 +#define ATOM_EXT_PLL2 9 +#define ATOM_EXT_CLOCK 10 +#define ATOM_PPLL_INVALID 0xFF + +#define ENCODER_REFCLK_SRC_P1PLL 0 +#define ENCODER_REFCLK_SRC_P2PLL 1 +#define ENCODER_REFCLK_SRC_DCPLL 2 +#define ENCODER_REFCLK_SRC_EXTCLK 3 +#define ENCODER_REFCLK_SRC_INVALID 0xFF + +#define ATOM_SCALER1 0 +#define ATOM_SCALER2 1 + +#define ATOM_SCALER_DISABLE 0 +#define ATOM_SCALER_CENTER 1 +#define ATOM_SCALER_EXPANSION 2 +#define ATOM_SCALER_MULTI_EX 3 + +#define ATOM_DISABLE 0 +#define ATOM_ENABLE 1 +#define ATOM_LCD_BLOFF (ATOM_DISABLE+2) +#define ATOM_LCD_BLON (ATOM_ENABLE+2) +#define ATOM_LCD_BL_BRIGHTNESS_CONTROL (ATOM_ENABLE+3) +#define ATOM_LCD_SELFTEST_START (ATOM_DISABLE+5) +#define ATOM_LCD_SELFTEST_STOP (ATOM_ENABLE+5) +#define ATOM_ENCODER_INIT (ATOM_DISABLE+7) +#define ATOM_GET_STATUS (ATOM_DISABLE+8) + +#define ATOM_BLANKING 1 +#define ATOM_BLANKING_OFF 0 + +#define ATOM_CURSOR1 0 +#define ATOM_CURSOR2 1 + +#define ATOM_ICON1 0 +#define ATOM_ICON2 1 + +#define ATOM_CRT1 0 +#define ATOM_CRT2 1 + +#define ATOM_TV_NTSC 1 +#define ATOM_TV_NTSCJ 2 +#define ATOM_TV_PAL 3 +#define ATOM_TV_PALM 4 +#define ATOM_TV_PALCN 5 +#define ATOM_TV_PALN 6 +#define ATOM_TV_PAL60 7 +#define ATOM_TV_SECAM 8 +#define ATOM_TV_CV 16 + +#define ATOM_DAC1_PS2 1 +#define ATOM_DAC1_CV 2 +#define ATOM_DAC1_NTSC 3 +#define ATOM_DAC1_PAL 4 + +#define ATOM_DAC2_PS2 ATOM_DAC1_PS2 +#define ATOM_DAC2_CV ATOM_DAC1_CV +#define ATOM_DAC2_NTSC ATOM_DAC1_NTSC +#define ATOM_DAC2_PAL ATOM_DAC1_PAL + +#define ATOM_PM_ON 0 +#define ATOM_PM_STANDBY 1 +#define ATOM_PM_SUSPEND 2 +#define ATOM_PM_OFF 3 + +/* Bit0:{=0:single, =1:dual}, + Bit1 {=0:666RGB, =1:888RGB}, + Bit2:3:{Grey level} + Bit4:{=0:LDI format for RGB888, =1 FPDI format for RGB888}*/ + +#define ATOM_PANEL_MISC_DUAL 0x00000001 +#define ATOM_PANEL_MISC_888RGB 0x00000002 +#define ATOM_PANEL_MISC_GREY_LEVEL 0x0000000C +#define ATOM_PANEL_MISC_FPDI 0x00000010 +#define ATOM_PANEL_MISC_GREY_LEVEL_SHIFT 2 +#define ATOM_PANEL_MISC_SPATIAL 0x00000020 +#define ATOM_PANEL_MISC_TEMPORAL 0x00000040 +#define ATOM_PANEL_MISC_API_ENABLED 0x00000080 + + +#define MEMTYPE_DDR1 "DDR1" +#define MEMTYPE_DDR2 "DDR2" +#define MEMTYPE_DDR3 "DDR3" +#define MEMTYPE_DDR4 "DDR4" + +#define ASIC_BUS_TYPE_PCI "PCI" +#define ASIC_BUS_TYPE_AGP "AGP" +#define ASIC_BUS_TYPE_PCIE "PCI_EXPRESS" + +/* Maximum size of that FireGL flag string */ + +#define ATOM_FIREGL_FLAG_STRING "FGL" //Flag used to enable FireGL Support +#define ATOM_MAX_SIZE_OF_FIREGL_FLAG_STRING 3 //sizeof( ATOM_FIREGL_FLAG_STRING ) + +#define ATOM_FAKE_DESKTOP_STRING "DSK" //Flag used to enable mobile ASIC on Desktop +#define ATOM_MAX_SIZE_OF_FAKE_DESKTOP_STRING ATOM_MAX_SIZE_OF_FIREGL_FLAG_STRING + +#define ATOM_M54T_FLAG_STRING "M54T" //Flag used to enable M54T Support +#define ATOM_MAX_SIZE_OF_M54T_FLAG_STRING 4 //sizeof( ATOM_M54T_FLAG_STRING ) + +#define HW_ASSISTED_I2C_STATUS_FAILURE 2 +#define HW_ASSISTED_I2C_STATUS_SUCCESS 1 + +#pragma pack(1) /* BIOS data must use byte aligment */ + +/* Define offset to location of ROM header. */ + +#define OFFSET_TO_POINTER_TO_ATOM_ROM_HEADER 0x00000048L +#define OFFSET_TO_ATOM_ROM_IMAGE_SIZE 0x00000002L + +#define OFFSET_TO_ATOMBIOS_ASIC_BUS_MEM_TYPE 0x94 +#define MAXSIZE_OF_ATOMBIOS_ASIC_BUS_MEM_TYPE 20 /* including the terminator 0x0! */ +#define OFFSET_TO_GET_ATOMBIOS_STRINGS_NUMBER 0x002f +#define OFFSET_TO_GET_ATOMBIOS_STRINGS_START 0x006e + +/* Common header for all ROM Data tables. + Every table pointed _ATOM_MASTER_DATA_TABLE has this common header. + And the pointer actually points to this header. */ + +typedef struct _ATOM_COMMON_TABLE_HEADER +{ + USHORT usStructureSize; + UCHAR ucTableFormatRevision; /*Change it when the Parser is not backward compatible */ + UCHAR ucTableContentRevision; /*Change it only when the table needs to change but the firmware */ + /*Image can't be updated, while Driver needs to carry the new table! */ +}ATOM_COMMON_TABLE_HEADER; + +/****************************************************************************/ +// Structure stores the ROM header. +/****************************************************************************/ +typedef struct _ATOM_ROM_HEADER +{ + ATOM_COMMON_TABLE_HEADER sHeader; + UCHAR uaFirmWareSignature[4]; /*Signature to distinguish between Atombios and non-atombios, + atombios should init it as "ATOM", don't change the position */ + USHORT usBiosRuntimeSegmentAddress; + USHORT usProtectedModeInfoOffset; + USHORT usConfigFilenameOffset; + USHORT usCRC_BlockOffset; + USHORT usBIOS_BootupMessageOffset; + USHORT usInt10Offset; + USHORT usPciBusDevInitCode; + USHORT usIoBaseAddress; + USHORT usSubsystemVendorID; + USHORT usSubsystemID; + USHORT usPCI_InfoOffset; + USHORT usMasterCommandTableOffset; /*Offset for SW to get all command table offsets, Don't change the position */ + USHORT usMasterDataTableOffset; /*Offset for SW to get all data table offsets, Don't change the position */ + UCHAR ucExtendedFunctionCode; + UCHAR ucReserved; +}ATOM_ROM_HEADER; + +/*==============================Command Table Portion==================================== */ + +#ifdef UEFI_BUILD + #define UTEMP USHORT + #define USHORT void* +#endif + +/****************************************************************************/ +// Structures used in Command.mtb +/****************************************************************************/ +typedef struct _ATOM_MASTER_LIST_OF_COMMAND_TABLES{ + USHORT ASIC_Init; //Function Table, used by various SW components,latest version 1.1 + USHORT GetDisplaySurfaceSize; //Atomic Table, Used by Bios when enabling HW ICON + USHORT ASIC_RegistersInit; //Atomic Table, indirectly used by various SW components,called from ASIC_Init + USHORT VRAM_BlockVenderDetection; //Atomic Table, used only by Bios + USHORT DIGxEncoderControl; //Only used by Bios + USHORT MemoryControllerInit; //Atomic Table, indirectly used by various SW components,called from ASIC_Init + USHORT EnableCRTCMemReq; //Function Table,directly used by various SW components,latest version 2.1 + USHORT MemoryParamAdjust; //Atomic Table, indirectly used by various SW components,called from SetMemoryClock if needed + USHORT DVOEncoderControl; //Function Table,directly used by various SW components,latest version 1.2 + USHORT GPIOPinControl; //Atomic Table, only used by Bios + USHORT SetEngineClock; //Function Table,directly used by various SW components,latest version 1.1 + USHORT SetMemoryClock; //Function Table,directly used by various SW components,latest version 1.1 + USHORT SetPixelClock; //Function Table,directly used by various SW components,latest version 1.2 + USHORT DynamicClockGating; //Atomic Table, indirectly used by various SW components,called from ASIC_Init + USHORT ResetMemoryDLL; //Atomic Table, indirectly used by various SW components,called from SetMemoryClock + USHORT ResetMemoryDevice; //Atomic Table, indirectly used by various SW components,called from SetMemoryClock + USHORT MemoryPLLInit; + USHORT AdjustDisplayPll; //only used by Bios + USHORT AdjustMemoryController; //Atomic Table, indirectly used by various SW components,called from SetMemoryClock + USHORT EnableASIC_StaticPwrMgt; //Atomic Table, only used by Bios + USHORT ASIC_StaticPwrMgtStatusChange; //Obsolete , only used by Bios + USHORT DAC_LoadDetection; //Atomic Table, directly used by various SW components,latest version 1.2 + USHORT LVTMAEncoderControl; //Atomic Table,directly used by various SW components,latest version 1.3 + USHORT LCD1OutputControl; //Atomic Table, directly used by various SW components,latest version 1.1 + USHORT DAC1EncoderControl; //Atomic Table, directly used by various SW components,latest version 1.1 + USHORT DAC2EncoderControl; //Atomic Table, directly used by various SW components,latest version 1.1 + USHORT DVOOutputControl; //Atomic Table, directly used by various SW components,latest version 1.1 + USHORT CV1OutputControl; //Atomic Table, Atomic Table, Obsolete from Ry6xx, use DAC2 Output instead + USHORT GetConditionalGoldenSetting; //only used by Bios + USHORT TVEncoderControl; //Function Table,directly used by various SW components,latest version 1.1 + USHORT TMDSAEncoderControl; //Atomic Table, directly used by various SW components,latest version 1.3 + USHORT LVDSEncoderControl; //Atomic Table, directly used by various SW components,latest version 1.3 + USHORT TV1OutputControl; //Atomic Table, Obsolete from Ry6xx, use DAC2 Output instead + USHORT EnableScaler; //Atomic Table, used only by Bios + USHORT BlankCRTC; //Atomic Table, directly used by various SW components,latest version 1.1 + USHORT EnableCRTC; //Atomic Table, directly used by various SW components,latest version 1.1 + USHORT GetPixelClock; //Atomic Table, directly used by various SW components,latest version 1.1 + USHORT EnableVGA_Render; //Function Table,directly used by various SW components,latest version 1.1 + USHORT GetSCLKOverMCLKRatio; //Atomic Table, only used by Bios + USHORT SetCRTC_Timing; //Atomic Table, directly used by various SW components,latest version 1.1 + USHORT SetCRTC_OverScan; //Atomic Table, used by various SW components,latest version 1.1 + USHORT SetCRTC_Replication; //Atomic Table, used only by Bios + USHORT SelectCRTC_Source; //Atomic Table, directly used by various SW components,latest version 1.1 + USHORT EnableGraphSurfaces; //Atomic Table, used only by Bios + USHORT UpdateCRTC_DoubleBufferRegisters; + USHORT LUT_AutoFill; //Atomic Table, only used by Bios + USHORT EnableHW_IconCursor; //Atomic Table, only used by Bios + USHORT GetMemoryClock; //Atomic Table, directly used by various SW components,latest version 1.1 + USHORT GetEngineClock; //Atomic Table, directly used by various SW components,latest version 1.1 + USHORT SetCRTC_UsingDTDTiming; //Atomic Table, directly used by various SW components,latest version 1.1 + USHORT ExternalEncoderControl; //Atomic Table, directly used by various SW components,latest version 2.1 + USHORT LVTMAOutputControl; //Atomic Table, directly used by various SW components,latest version 1.1 + USHORT VRAM_BlockDetectionByStrap; //Atomic Table, used only by Bios + USHORT MemoryCleanUp; //Atomic Table, only used by Bios + USHORT ProcessI2cChannelTransaction; //Function Table,only used by Bios + USHORT WriteOneByteToHWAssistedI2C; //Function Table,indirectly used by various SW components + USHORT ReadHWAssistedI2CStatus; //Atomic Table, indirectly used by various SW components + USHORT SpeedFanControl; //Function Table,indirectly used by various SW components,called from ASIC_Init + USHORT PowerConnectorDetection; //Atomic Table, directly used by various SW components,latest version 1.1 + USHORT MC_Synchronization; //Atomic Table, indirectly used by various SW components,called from SetMemoryClock + USHORT ComputeMemoryEnginePLL; //Atomic Table, indirectly used by various SW components,called from SetMemory/EngineClock + USHORT MemoryRefreshConversion; //Atomic Table, indirectly used by various SW components,called from SetMemory or SetEngineClock + USHORT VRAM_GetCurrentInfoBlock; //Atomic Table, used only by Bios + USHORT DynamicMemorySettings; //Atomic Table, indirectly used by various SW components,called from SetMemoryClock + USHORT MemoryTraining; //Atomic Table, used only by Bios + USHORT EnableSpreadSpectrumOnPPLL; //Atomic Table, directly used by various SW components,latest version 1.2 + USHORT TMDSAOutputControl; //Atomic Table, directly used by various SW components,latest version 1.1 + USHORT SetVoltage; //Function Table,directly and/or indirectly used by various SW components,latest version 1.1 + USHORT DAC1OutputControl; //Atomic Table, directly used by various SW components,latest version 1.1 + USHORT DAC2OutputControl; //Atomic Table, directly used by various SW components,latest version 1.1 + USHORT SetupHWAssistedI2CStatus; //Function Table,only used by Bios, obsolete soon.Switch to use "ReadEDIDFromHWAssistedI2C" + USHORT ClockSource; //Atomic Table, indirectly used by various SW components,called from ASIC_Init + USHORT MemoryDeviceInit; //Atomic Table, indirectly used by various SW components,called from SetMemoryClock + USHORT EnableYUV; //Atomic Table, indirectly used by various SW components,called from EnableVGARender + USHORT DIG1EncoderControl; //Atomic Table,directly used by various SW components,latest version 1.1 + USHORT DIG2EncoderControl; //Atomic Table,directly used by various SW components,latest version 1.1 + USHORT DIG1TransmitterControl; //Atomic Table,directly used by various SW components,latest version 1.1 + USHORT DIG2TransmitterControl; //Atomic Table,directly used by various SW components,latest version 1.1 + USHORT ProcessAuxChannelTransaction; //Function Table,only used by Bios + USHORT DPEncoderService; //Function Table,only used by Bios +}ATOM_MASTER_LIST_OF_COMMAND_TABLES; + +// For backward compatible +#define ReadEDIDFromHWAssistedI2C ProcessI2cChannelTransaction +#define UNIPHYTransmitterControl DIG1TransmitterControl +#define LVTMATransmitterControl DIG2TransmitterControl +#define SetCRTC_DPM_State GetConditionalGoldenSetting +#define SetUniphyInstance ASIC_StaticPwrMgtStatusChange +#define HPDInterruptService ReadHWAssistedI2CStatus +#define EnableVGA_Access GetSCLKOverMCLKRatio +#define GetDispObjectInfo EnableYUV + +typedef struct _ATOM_MASTER_COMMAND_TABLE +{ + ATOM_COMMON_TABLE_HEADER sHeader; + ATOM_MASTER_LIST_OF_COMMAND_TABLES ListOfCommandTables; +}ATOM_MASTER_COMMAND_TABLE; + +/****************************************************************************/ +// Structures used in every command table +/****************************************************************************/ +typedef struct _ATOM_TABLE_ATTRIBUTE +{ +#if ATOM_BIG_ENDIAN + USHORT UpdatedByUtility:1; //[15]=Table updated by utility flag + USHORT PS_SizeInBytes:7; //[14:8]=Size of parameter space in Bytes (multiple of a dword), + USHORT WS_SizeInBytes:8; //[7:0]=Size of workspace in Bytes (in multiple of a dword), +#else + USHORT WS_SizeInBytes:8; //[7:0]=Size of workspace in Bytes (in multiple of a dword), + USHORT PS_SizeInBytes:7; //[14:8]=Size of parameter space in Bytes (multiple of a dword), + USHORT UpdatedByUtility:1; //[15]=Table updated by utility flag +#endif +}ATOM_TABLE_ATTRIBUTE; + +typedef union _ATOM_TABLE_ATTRIBUTE_ACCESS +{ + ATOM_TABLE_ATTRIBUTE sbfAccess; + USHORT susAccess; +}ATOM_TABLE_ATTRIBUTE_ACCESS; + +/****************************************************************************/ +// Common header for all command tables. +// Every table pointed by _ATOM_MASTER_COMMAND_TABLE has this common header. +// And the pointer actually points to this header. +/****************************************************************************/ +typedef struct _ATOM_COMMON_ROM_COMMAND_TABLE_HEADER +{ + ATOM_COMMON_TABLE_HEADER CommonHeader; + ATOM_TABLE_ATTRIBUTE TableAttribute; +}ATOM_COMMON_ROM_COMMAND_TABLE_HEADER; + +/****************************************************************************/ +// Structures used by ComputeMemoryEnginePLLTable +/****************************************************************************/ +#define COMPUTE_MEMORY_PLL_PARAM 1 +#define COMPUTE_ENGINE_PLL_PARAM 2 +#define ADJUST_MC_SETTING_PARAM 3 + +/****************************************************************************/ +// Structures used by AdjustMemoryControllerTable +/****************************************************************************/ +typedef struct _ATOM_ADJUST_MEMORY_CLOCK_FREQ +{ +#if ATOM_BIG_ENDIAN + ULONG ulPointerReturnFlag:1; // BYTE_3[7]=1 - Return the pointer to the right Data Block; BYTE_3[7]=0 - Program the right Data Block + ULONG ulMemoryModuleNumber:7; // BYTE_3[6:0] + ULONG ulClockFreq:24; +#else + ULONG ulClockFreq:24; + ULONG ulMemoryModuleNumber:7; // BYTE_3[6:0] + ULONG ulPointerReturnFlag:1; // BYTE_3[7]=1 - Return the pointer to the right Data Block; BYTE_3[7]=0 - Program the right Data Block +#endif +}ATOM_ADJUST_MEMORY_CLOCK_FREQ; +#define POINTER_RETURN_FLAG 0x80 + +typedef struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS +{ + ULONG ulClock; //When returen, it's the re-calculated clock based on given Fb_div Post_Div and ref_div + UCHAR ucAction; //0:reserved //1:Memory //2:Engine + UCHAR ucReserved; //may expand to return larger Fbdiv later + UCHAR ucFbDiv; //return value + UCHAR ucPostDiv; //return value +}COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS; + +typedef struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V2 +{ + ULONG ulClock; //When return, [23:0] return real clock + UCHAR ucAction; //0:reserved;COMPUTE_MEMORY_PLL_PARAM:Memory;COMPUTE_ENGINE_PLL_PARAM:Engine. it return ref_div to be written to register + USHORT usFbDiv; //return Feedback value to be written to register + UCHAR ucPostDiv; //return post div to be written to register +}COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V2; +#define COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_PS_ALLOCATION COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS + + +#define SET_CLOCK_FREQ_MASK 0x00FFFFFF //Clock change tables only take bit [23:0] as the requested clock value +#define USE_NON_BUS_CLOCK_MASK 0x01000000 //Applicable to both memory and engine clock change, when set, it uses another clock as the temporary clock (engine uses memory and vice versa) +#define USE_MEMORY_SELF_REFRESH_MASK 0x02000000 //Only applicable to memory clock change, when set, using memory self refresh during clock transition +#define SKIP_INTERNAL_MEMORY_PARAMETER_CHANGE 0x04000000 //Only applicable to memory clock change, when set, the table will skip predefined internal memory parameter change +#define FIRST_TIME_CHANGE_CLOCK 0x08000000 //Applicable to both memory and engine clock change,when set, it means this is 1st time to change clock after ASIC bootup +#define SKIP_SW_PROGRAM_PLL 0x10000000 //Applicable to both memory and engine clock change, when set, it means the table will not program SPLL/MPLL +#define USE_SS_ENABLED_PIXEL_CLOCK USE_NON_BUS_CLOCK_MASK + +#define b3USE_NON_BUS_CLOCK_MASK 0x01 //Applicable to both memory and engine clock change, when set, it uses another clock as the temporary clock (engine uses memory and vice versa) +#define b3USE_MEMORY_SELF_REFRESH 0x02 //Only applicable to memory clock change, when set, using memory self refresh during clock transition +#define b3SKIP_INTERNAL_MEMORY_PARAMETER_CHANGE 0x04 //Only applicable to memory clock change, when set, the table will skip predefined internal memory parameter change +#define b3FIRST_TIME_CHANGE_CLOCK 0x08 //Applicable to both memory and engine clock change,when set, it means this is 1st time to change clock after ASIC bootup +#define b3SKIP_SW_PROGRAM_PLL 0x10 //Applicable to both memory and engine clock change, when set, it means the table will not program SPLL/MPLL + +typedef struct _ATOM_COMPUTE_CLOCK_FREQ +{ +#if ATOM_BIG_ENDIAN + ULONG ulComputeClockFlag:8; // =1: COMPUTE_MEMORY_PLL_PARAM, =2: COMPUTE_ENGINE_PLL_PARAM + ULONG ulClockFreq:24; // in unit of 10kHz +#else + ULONG ulClockFreq:24; // in unit of 10kHz + ULONG ulComputeClockFlag:8; // =1: COMPUTE_MEMORY_PLL_PARAM, =2: COMPUTE_ENGINE_PLL_PARAM +#endif +}ATOM_COMPUTE_CLOCK_FREQ; + +typedef struct _ATOM_S_MPLL_FB_DIVIDER +{ + USHORT usFbDivFrac; + USHORT usFbDiv; +}ATOM_S_MPLL_FB_DIVIDER; + +typedef struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V3 +{ + union + { + ATOM_COMPUTE_CLOCK_FREQ ulClock; //Input Parameter + ATOM_S_MPLL_FB_DIVIDER ulFbDiv; //Output Parameter + }; + UCHAR ucRefDiv; //Output Parameter + UCHAR ucPostDiv; //Output Parameter + UCHAR ucCntlFlag; //Output Parameter + UCHAR ucReserved; +}COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V3; + +// ucCntlFlag +#define ATOM_PLL_CNTL_FLAG_PLL_POST_DIV_EN 1 +#define ATOM_PLL_CNTL_FLAG_MPLL_VCO_MODE 2 +#define ATOM_PLL_CNTL_FLAG_FRACTION_DISABLE 4 +#define ATOM_PLL_CNTL_FLAG_SPLL_ISPARE_9 8 + + +// V4 are only used for APU which PLL outside GPU +typedef struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V4 +{ +#if ATOM_BIG_ENDIAN + ULONG ucPostDiv; //return parameter: post divider which is used to program to register directly + ULONG ulClock:24; //Input= target clock, output = actual clock +#else + ULONG ulClock:24; //Input= target clock, output = actual clock + ULONG ucPostDiv; //return parameter: post divider which is used to program to register directly +#endif +}COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V4; + +typedef struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V5 +{ + union + { + ATOM_COMPUTE_CLOCK_FREQ ulClock; //Input Parameter + ATOM_S_MPLL_FB_DIVIDER ulFbDiv; //Output Parameter + }; + UCHAR ucRefDiv; //Output Parameter + UCHAR ucPostDiv; //Output Parameter + union + { + UCHAR ucCntlFlag; //Output Flags + UCHAR ucInputFlag; //Input Flags. ucInputFlag[0] - Strobe(1)/Performance(0) mode + }; + UCHAR ucReserved; +}COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V5; + +// ucInputFlag +#define ATOM_PLL_INPUT_FLAG_PLL_STROBE_MODE_EN 1 // 1-StrobeMode, 0-PerformanceMode + +typedef struct _DYNAMICE_MEMORY_SETTINGS_PARAMETER +{ + ATOM_COMPUTE_CLOCK_FREQ ulClock; + ULONG ulReserved[2]; +}DYNAMICE_MEMORY_SETTINGS_PARAMETER; + +typedef struct _DYNAMICE_ENGINE_SETTINGS_PARAMETER +{ + ATOM_COMPUTE_CLOCK_FREQ ulClock; + ULONG ulMemoryClock; + ULONG ulReserved; +}DYNAMICE_ENGINE_SETTINGS_PARAMETER; + +/****************************************************************************/ +// Structures used by SetEngineClockTable +/****************************************************************************/ +typedef struct _SET_ENGINE_CLOCK_PARAMETERS +{ + ULONG ulTargetEngineClock; //In 10Khz unit +}SET_ENGINE_CLOCK_PARAMETERS; + +typedef struct _SET_ENGINE_CLOCK_PS_ALLOCATION +{ + ULONG ulTargetEngineClock; //In 10Khz unit + COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_PS_ALLOCATION sReserved; +}SET_ENGINE_CLOCK_PS_ALLOCATION; + +/****************************************************************************/ +// Structures used by SetMemoryClockTable +/****************************************************************************/ +typedef struct _SET_MEMORY_CLOCK_PARAMETERS +{ + ULONG ulTargetMemoryClock; //In 10Khz unit +}SET_MEMORY_CLOCK_PARAMETERS; + +typedef struct _SET_MEMORY_CLOCK_PS_ALLOCATION +{ + ULONG ulTargetMemoryClock; //In 10Khz unit + COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_PS_ALLOCATION sReserved; +}SET_MEMORY_CLOCK_PS_ALLOCATION; + +/****************************************************************************/ +// Structures used by ASIC_Init.ctb +/****************************************************************************/ +typedef struct _ASIC_INIT_PARAMETERS +{ + ULONG ulDefaultEngineClock; //In 10Khz unit + ULONG ulDefaultMemoryClock; //In 10Khz unit +}ASIC_INIT_PARAMETERS; + +typedef struct _ASIC_INIT_PS_ALLOCATION +{ + ASIC_INIT_PARAMETERS sASICInitClocks; + SET_ENGINE_CLOCK_PS_ALLOCATION sReserved; //Caller doesn't need to init this structure +}ASIC_INIT_PS_ALLOCATION; + +/****************************************************************************/ +// Structure used by DynamicClockGatingTable.ctb +/****************************************************************************/ +typedef struct _DYNAMIC_CLOCK_GATING_PARAMETERS +{ + UCHAR ucEnable; // ATOM_ENABLE or ATOM_DISABLE + UCHAR ucPadding[3]; +}DYNAMIC_CLOCK_GATING_PARAMETERS; +#define DYNAMIC_CLOCK_GATING_PS_ALLOCATION DYNAMIC_CLOCK_GATING_PARAMETERS + +/****************************************************************************/ +// Structure used by EnableASIC_StaticPwrMgtTable.ctb +/****************************************************************************/ +typedef struct _ENABLE_ASIC_STATIC_PWR_MGT_PARAMETERS +{ + UCHAR ucEnable; // ATOM_ENABLE or ATOM_DISABLE + UCHAR ucPadding[3]; +}ENABLE_ASIC_STATIC_PWR_MGT_PARAMETERS; +#define ENABLE_ASIC_STATIC_PWR_MGT_PS_ALLOCATION ENABLE_ASIC_STATIC_PWR_MGT_PARAMETERS + +/****************************************************************************/ +// Structures used by DAC_LoadDetectionTable.ctb +/****************************************************************************/ +typedef struct _DAC_LOAD_DETECTION_PARAMETERS +{ + USHORT usDeviceID; //{ATOM_DEVICE_CRTx_SUPPORT,ATOM_DEVICE_TVx_SUPPORT,ATOM_DEVICE_CVx_SUPPORT} + UCHAR ucDacType; //{ATOM_DAC_A,ATOM_DAC_B, ATOM_EXT_DAC} + UCHAR ucMisc; //Valid only when table revision =1.3 and above +}DAC_LOAD_DETECTION_PARAMETERS; + +// DAC_LOAD_DETECTION_PARAMETERS.ucMisc +#define DAC_LOAD_MISC_YPrPb 0x01 + +typedef struct _DAC_LOAD_DETECTION_PS_ALLOCATION +{ + DAC_LOAD_DETECTION_PARAMETERS sDacload; + ULONG Reserved[2];// Don't set this one, allocation for EXT DAC +}DAC_LOAD_DETECTION_PS_ALLOCATION; + +/****************************************************************************/ +// Structures used by DAC1EncoderControlTable.ctb and DAC2EncoderControlTable.ctb +/****************************************************************************/ +typedef struct _DAC_ENCODER_CONTROL_PARAMETERS +{ + USHORT usPixelClock; // in 10KHz; for bios convenient + UCHAR ucDacStandard; // See definition of ATOM_DACx_xxx, For DEC3.0, bit 7 used as internal flag to indicate DAC2 (==1) or DAC1 (==0) + UCHAR ucAction; // 0: turn off encoder + // 1: setup and turn on encoder + // 7: ATOM_ENCODER_INIT Initialize DAC +}DAC_ENCODER_CONTROL_PARAMETERS; + +#define DAC_ENCODER_CONTROL_PS_ALLOCATION DAC_ENCODER_CONTROL_PARAMETERS + +/****************************************************************************/ +// Structures used by DIG1EncoderControlTable +// DIG2EncoderControlTable +// ExternalEncoderControlTable +/****************************************************************************/ +typedef struct _DIG_ENCODER_CONTROL_PARAMETERS +{ + USHORT usPixelClock; // in 10KHz; for bios convenient + UCHAR ucConfig; + // [2] Link Select: + // =0: PHY linkA if bfLane<3 + // =1: PHY linkB if bfLanes<3 + // =0: PHY linkA+B if bfLanes=3 + // [3] Transmitter Sel + // =0: UNIPHY or PCIEPHY + // =1: LVTMA + UCHAR ucAction; // =0: turn off encoder + // =1: turn on encoder + UCHAR ucEncoderMode; + // =0: DP encoder + // =1: LVDS encoder + // =2: DVI encoder + // =3: HDMI encoder + // =4: SDVO encoder + UCHAR ucLaneNum; // how many lanes to enable + UCHAR ucReserved[2]; +}DIG_ENCODER_CONTROL_PARAMETERS; +#define DIG_ENCODER_CONTROL_PS_ALLOCATION DIG_ENCODER_CONTROL_PARAMETERS +#define EXTERNAL_ENCODER_CONTROL_PARAMETER DIG_ENCODER_CONTROL_PARAMETERS + +//ucConfig +#define ATOM_ENCODER_CONFIG_DPLINKRATE_MASK 0x01 +#define ATOM_ENCODER_CONFIG_DPLINKRATE_1_62GHZ 0x00 +#define ATOM_ENCODER_CONFIG_DPLINKRATE_2_70GHZ 0x01 +#define ATOM_ENCODER_CONFIG_DPLINKRATE_5_40GHZ 0x02 +#define ATOM_ENCODER_CONFIG_LINK_SEL_MASK 0x04 +#define ATOM_ENCODER_CONFIG_LINKA 0x00 +#define ATOM_ENCODER_CONFIG_LINKB 0x04 +#define ATOM_ENCODER_CONFIG_LINKA_B ATOM_TRANSMITTER_CONFIG_LINKA +#define ATOM_ENCODER_CONFIG_LINKB_A ATOM_ENCODER_CONFIG_LINKB +#define ATOM_ENCODER_CONFIG_TRANSMITTER_SEL_MASK 0x08 +#define ATOM_ENCODER_CONFIG_UNIPHY 0x00 +#define ATOM_ENCODER_CONFIG_LVTMA 0x08 +#define ATOM_ENCODER_CONFIG_TRANSMITTER1 0x00 +#define ATOM_ENCODER_CONFIG_TRANSMITTER2 0x08 +#define ATOM_ENCODER_CONFIG_DIGB 0x80 // VBIOS Internal use, outside SW should set this bit=0 +// ucAction +// ATOM_ENABLE: Enable Encoder +// ATOM_DISABLE: Disable Encoder + +//ucEncoderMode +#define ATOM_ENCODER_MODE_DP 0 +#define ATOM_ENCODER_MODE_LVDS 1 +#define ATOM_ENCODER_MODE_DVI 2 +#define ATOM_ENCODER_MODE_HDMI 3 +#define ATOM_ENCODER_MODE_SDVO 4 +#define ATOM_ENCODER_MODE_DP_AUDIO 5 +#define ATOM_ENCODER_MODE_TV 13 +#define ATOM_ENCODER_MODE_CV 14 +#define ATOM_ENCODER_MODE_CRT 15 +#define ATOM_ENCODER_MODE_DVO 16 +#define ATOM_ENCODER_MODE_DP_SST ATOM_ENCODER_MODE_DP // For DP1.2 +#define ATOM_ENCODER_MODE_DP_MST 5 // For DP1.2 + +typedef struct _ATOM_DIG_ENCODER_CONFIG_V2 +{ +#if ATOM_BIG_ENDIAN + UCHAR ucReserved1:2; + UCHAR ucTransmitterSel:2; // =0: UniphyAB, =1: UniphyCD =2: UniphyEF + UCHAR ucLinkSel:1; // =0: linkA/C/E =1: linkB/D/F + UCHAR ucReserved:1; + UCHAR ucDPLinkRate:1; // =0: 1.62Ghz, =1: 2.7Ghz +#else + UCHAR ucDPLinkRate:1; // =0: 1.62Ghz, =1: 2.7Ghz + UCHAR ucReserved:1; + UCHAR ucLinkSel:1; // =0: linkA/C/E =1: linkB/D/F + UCHAR ucTransmitterSel:2; // =0: UniphyAB, =1: UniphyCD =2: UniphyEF + UCHAR ucReserved1:2; +#endif +}ATOM_DIG_ENCODER_CONFIG_V2; + + +typedef struct _DIG_ENCODER_CONTROL_PARAMETERS_V2 +{ + USHORT usPixelClock; // in 10KHz; for bios convenient + ATOM_DIG_ENCODER_CONFIG_V2 acConfig; + UCHAR ucAction; + UCHAR ucEncoderMode; + // =0: DP encoder + // =1: LVDS encoder + // =2: DVI encoder + // =3: HDMI encoder + // =4: SDVO encoder + UCHAR ucLaneNum; // how many lanes to enable + UCHAR ucStatus; // = DP_LINK_TRAINING_COMPLETE or DP_LINK_TRAINING_INCOMPLETE, only used by VBIOS with command ATOM_ENCODER_CMD_QUERY_DP_LINK_TRAINING_STATUS + UCHAR ucReserved; +}DIG_ENCODER_CONTROL_PARAMETERS_V2; + +//ucConfig +#define ATOM_ENCODER_CONFIG_V2_DPLINKRATE_MASK 0x01 +#define ATOM_ENCODER_CONFIG_V2_DPLINKRATE_1_62GHZ 0x00 +#define ATOM_ENCODER_CONFIG_V2_DPLINKRATE_2_70GHZ 0x01 +#define ATOM_ENCODER_CONFIG_V2_LINK_SEL_MASK 0x04 +#define ATOM_ENCODER_CONFIG_V2_LINKA 0x00 +#define ATOM_ENCODER_CONFIG_V2_LINKB 0x04 +#define ATOM_ENCODER_CONFIG_V2_TRANSMITTER_SEL_MASK 0x18 +#define ATOM_ENCODER_CONFIG_V2_TRANSMITTER1 0x00 +#define ATOM_ENCODER_CONFIG_V2_TRANSMITTER2 0x08 +#define ATOM_ENCODER_CONFIG_V2_TRANSMITTER3 0x10 + +// ucAction: +// ATOM_DISABLE +// ATOM_ENABLE +#define ATOM_ENCODER_CMD_DP_LINK_TRAINING_START 0x08 +#define ATOM_ENCODER_CMD_DP_LINK_TRAINING_PATTERN1 0x09 +#define ATOM_ENCODER_CMD_DP_LINK_TRAINING_PATTERN2 0x0a +#define ATOM_ENCODER_CMD_DP_LINK_TRAINING_PATTERN3 0x13 +#define ATOM_ENCODER_CMD_DP_LINK_TRAINING_COMPLETE 0x0b +#define ATOM_ENCODER_CMD_DP_VIDEO_OFF 0x0c +#define ATOM_ENCODER_CMD_DP_VIDEO_ON 0x0d +#define ATOM_ENCODER_CMD_QUERY_DP_LINK_TRAINING_STATUS 0x0e +#define ATOM_ENCODER_CMD_SETUP 0x0f +#define ATOM_ENCODER_CMD_SETUP_PANEL_MODE 0x10 + +// ucStatus +#define ATOM_ENCODER_STATUS_LINK_TRAINING_COMPLETE 0x10 +#define ATOM_ENCODER_STATUS_LINK_TRAINING_INCOMPLETE 0x00 + +//ucTableFormatRevision=1 +//ucTableContentRevision=3 +// Following function ENABLE sub-function will be used by driver when TMDS/HDMI/LVDS is used, disable function will be used by driver +typedef struct _ATOM_DIG_ENCODER_CONFIG_V3 +{ +#if ATOM_BIG_ENDIAN + UCHAR ucReserved1:1; + UCHAR ucDigSel:3; // =0/1/2/3/4/5: DIG0/1/2/3/4/5 (In register spec also referred as DIGA/B/C/D/E/F) + UCHAR ucReserved:3; + UCHAR ucDPLinkRate:1; // =0: 1.62Ghz, =1: 2.7Ghz +#else + UCHAR ucDPLinkRate:1; // =0: 1.62Ghz, =1: 2.7Ghz + UCHAR ucReserved:3; + UCHAR ucDigSel:3; // =0/1/2/3/4/5: DIG0/1/2/3/4/5 (In register spec also referred as DIGA/B/C/D/E/F) + UCHAR ucReserved1:1; +#endif +}ATOM_DIG_ENCODER_CONFIG_V3; + +#define ATOM_ENCODER_CONFIG_V3_DPLINKRATE_MASK 0x03 +#define ATOM_ENCODER_CONFIG_V3_DPLINKRATE_1_62GHZ 0x00 +#define ATOM_ENCODER_CONFIG_V3_DPLINKRATE_2_70GHZ 0x01 +#define ATOM_ENCODER_CONFIG_V3_ENCODER_SEL 0x70 +#define ATOM_ENCODER_CONFIG_V3_DIG0_ENCODER 0x00 +#define ATOM_ENCODER_CONFIG_V3_DIG1_ENCODER 0x10 +#define ATOM_ENCODER_CONFIG_V3_DIG2_ENCODER 0x20 +#define ATOM_ENCODER_CONFIG_V3_DIG3_ENCODER 0x30 +#define ATOM_ENCODER_CONFIG_V3_DIG4_ENCODER 0x40 +#define ATOM_ENCODER_CONFIG_V3_DIG5_ENCODER 0x50 + +typedef struct _DIG_ENCODER_CONTROL_PARAMETERS_V3 +{ + USHORT usPixelClock; // in 10KHz; for bios convenient + ATOM_DIG_ENCODER_CONFIG_V3 acConfig; + UCHAR ucAction; + union { + UCHAR ucEncoderMode; + // =0: DP encoder + // =1: LVDS encoder + // =2: DVI encoder + // =3: HDMI encoder + // =4: SDVO encoder + // =5: DP audio + UCHAR ucPanelMode; // only valid when ucAction == ATOM_ENCODER_CMD_SETUP_PANEL_MODE + // =0: external DP + // =1: internal DP2 + // =0x11: internal DP1 for NutMeg/Travis DP translator + }; + UCHAR ucLaneNum; // how many lanes to enable + UCHAR ucBitPerColor; // only valid for DP mode when ucAction = ATOM_ENCODER_CMD_SETUP + UCHAR ucReserved; +}DIG_ENCODER_CONTROL_PARAMETERS_V3; + +//ucTableFormatRevision=1 +//ucTableContentRevision=4 +// start from NI +// Following function ENABLE sub-function will be used by driver when TMDS/HDMI/LVDS is used, disable function will be used by driver +typedef struct _ATOM_DIG_ENCODER_CONFIG_V4 +{ +#if ATOM_BIG_ENDIAN + UCHAR ucReserved1:1; + UCHAR ucDigSel:3; // =0/1/2/3/4/5: DIG0/1/2/3/4/5 (In register spec also referred as DIGA/B/C/D/E/F) + UCHAR ucReserved:2; + UCHAR ucDPLinkRate:2; // =0: 1.62Ghz, =1: 2.7Ghz, 2=5.4Ghz <= Changed comparing to previous version +#else + UCHAR ucDPLinkRate:2; // =0: 1.62Ghz, =1: 2.7Ghz, 2=5.4Ghz <= Changed comparing to previous version + UCHAR ucReserved:2; + UCHAR ucDigSel:3; // =0/1/2/3/4/5: DIG0/1/2/3/4/5 (In register spec also referred as DIGA/B/C/D/E/F) + UCHAR ucReserved1:1; +#endif +}ATOM_DIG_ENCODER_CONFIG_V4; + +#define ATOM_ENCODER_CONFIG_V4_DPLINKRATE_MASK 0x03 +#define ATOM_ENCODER_CONFIG_V4_DPLINKRATE_1_62GHZ 0x00 +#define ATOM_ENCODER_CONFIG_V4_DPLINKRATE_2_70GHZ 0x01 +#define ATOM_ENCODER_CONFIG_V4_DPLINKRATE_5_40GHZ 0x02 +#define ATOM_ENCODER_CONFIG_V4_ENCODER_SEL 0x70 +#define ATOM_ENCODER_CONFIG_V4_DIG0_ENCODER 0x00 +#define ATOM_ENCODER_CONFIG_V4_DIG1_ENCODER 0x10 +#define ATOM_ENCODER_CONFIG_V4_DIG2_ENCODER 0x20 +#define ATOM_ENCODER_CONFIG_V4_DIG3_ENCODER 0x30 +#define ATOM_ENCODER_CONFIG_V4_DIG4_ENCODER 0x40 +#define ATOM_ENCODER_CONFIG_V4_DIG5_ENCODER 0x50 + +typedef struct _DIG_ENCODER_CONTROL_PARAMETERS_V4 +{ + USHORT usPixelClock; // in 10KHz; for bios convenient + union{ + ATOM_DIG_ENCODER_CONFIG_V4 acConfig; + UCHAR ucConfig; + }; + UCHAR ucAction; + union { + UCHAR ucEncoderMode; + // =0: DP encoder + // =1: LVDS encoder + // =2: DVI encoder + // =3: HDMI encoder + // =4: SDVO encoder + // =5: DP audio + UCHAR ucPanelMode; // only valid when ucAction == ATOM_ENCODER_CMD_SETUP_PANEL_MODE + // =0: external DP + // =1: internal DP2 + // =0x11: internal DP1 for NutMeg/Travis DP translator + }; + UCHAR ucLaneNum; // how many lanes to enable + UCHAR ucBitPerColor; // only valid for DP mode when ucAction = ATOM_ENCODER_CMD_SETUP + UCHAR ucHPD_ID; // HPD ID (1-6). =0 means to skip HDP programming. New comparing to previous version +}DIG_ENCODER_CONTROL_PARAMETERS_V4; + +// define ucBitPerColor: +#define PANEL_BPC_UNDEFINE 0x00 +#define PANEL_6BIT_PER_COLOR 0x01 +#define PANEL_8BIT_PER_COLOR 0x02 +#define PANEL_10BIT_PER_COLOR 0x03 +#define PANEL_12BIT_PER_COLOR 0x04 +#define PANEL_16BIT_PER_COLOR 0x05 + +//define ucPanelMode +#define DP_PANEL_MODE_EXTERNAL_DP_MODE 0x00 +#define DP_PANEL_MODE_INTERNAL_DP2_MODE 0x01 +#define DP_PANEL_MODE_INTERNAL_DP1_MODE 0x11 + +/****************************************************************************/ +// Structures used by UNIPHYTransmitterControlTable +// LVTMATransmitterControlTable +// DVOOutputControlTable +/****************************************************************************/ +typedef struct _ATOM_DP_VS_MODE +{ + UCHAR ucLaneSel; + UCHAR ucLaneSet; +}ATOM_DP_VS_MODE; + +typedef struct _DIG_TRANSMITTER_CONTROL_PARAMETERS +{ + union + { + USHORT usPixelClock; // in 10KHz; for bios convenient + USHORT usInitInfo; // when init uniphy,lower 8bit is used for connector type defined in objectid.h + ATOM_DP_VS_MODE asMode; // DP Voltage swing mode + }; + UCHAR ucConfig; + // [0]=0: 4 lane Link, + // =1: 8 lane Link ( Dual Links TMDS ) + // [1]=0: InCoherent mode + // =1: Coherent Mode + // [2] Link Select: + // =0: PHY linkA if bfLane<3 + // =1: PHY linkB if bfLanes<3 + // =0: PHY linkA+B if bfLanes=3 + // [5:4]PCIE lane Sel + // =0: lane 0~3 or 0~7 + // =1: lane 4~7 + // =2: lane 8~11 or 8~15 + // =3: lane 12~15 + UCHAR ucAction; // =0: turn off encoder + // =1: turn on encoder + UCHAR ucReserved[4]; +}DIG_TRANSMITTER_CONTROL_PARAMETERS; + +#define DIG_TRANSMITTER_CONTROL_PS_ALLOCATION DIG_TRANSMITTER_CONTROL_PARAMETERS + +//ucInitInfo +#define ATOM_TRAMITTER_INITINFO_CONNECTOR_MASK 0x00ff + +//ucConfig +#define ATOM_TRANSMITTER_CONFIG_8LANE_LINK 0x01 +#define ATOM_TRANSMITTER_CONFIG_COHERENT 0x02 +#define ATOM_TRANSMITTER_CONFIG_LINK_SEL_MASK 0x04 +#define ATOM_TRANSMITTER_CONFIG_LINKA 0x00 +#define ATOM_TRANSMITTER_CONFIG_LINKB 0x04 +#define ATOM_TRANSMITTER_CONFIG_LINKA_B 0x00 +#define ATOM_TRANSMITTER_CONFIG_LINKB_A 0x04 + +#define ATOM_TRANSMITTER_CONFIG_ENCODER_SEL_MASK 0x08 // only used when ATOM_TRANSMITTER_ACTION_ENABLE +#define ATOM_TRANSMITTER_CONFIG_DIG1_ENCODER 0x00 // only used when ATOM_TRANSMITTER_ACTION_ENABLE +#define ATOM_TRANSMITTER_CONFIG_DIG2_ENCODER 0x08 // only used when ATOM_TRANSMITTER_ACTION_ENABLE + +#define ATOM_TRANSMITTER_CONFIG_CLKSRC_MASK 0x30 +#define ATOM_TRANSMITTER_CONFIG_CLKSRC_PPLL 0x00 +#define ATOM_TRANSMITTER_CONFIG_CLKSRC_PCIE 0x20 +#define ATOM_TRANSMITTER_CONFIG_CLKSRC_XTALIN 0x30 +#define ATOM_TRANSMITTER_CONFIG_LANE_SEL_MASK 0xc0 +#define ATOM_TRANSMITTER_CONFIG_LANE_0_3 0x00 +#define ATOM_TRANSMITTER_CONFIG_LANE_0_7 0x00 +#define ATOM_TRANSMITTER_CONFIG_LANE_4_7 0x40 +#define ATOM_TRANSMITTER_CONFIG_LANE_8_11 0x80 +#define ATOM_TRANSMITTER_CONFIG_LANE_8_15 0x80 +#define ATOM_TRANSMITTER_CONFIG_LANE_12_15 0xc0 + +//ucAction +#define ATOM_TRANSMITTER_ACTION_DISABLE 0 +#define ATOM_TRANSMITTER_ACTION_ENABLE 1 +#define ATOM_TRANSMITTER_ACTION_LCD_BLOFF 2 +#define ATOM_TRANSMITTER_ACTION_LCD_BLON 3 +#define ATOM_TRANSMITTER_ACTION_BL_BRIGHTNESS_CONTROL 4 +#define ATOM_TRANSMITTER_ACTION_LCD_SELFTEST_START 5 +#define ATOM_TRANSMITTER_ACTION_LCD_SELFTEST_STOP 6 +#define ATOM_TRANSMITTER_ACTION_INIT 7 +#define ATOM_TRANSMITTER_ACTION_DISABLE_OUTPUT 8 +#define ATOM_TRANSMITTER_ACTION_ENABLE_OUTPUT 9 +#define ATOM_TRANSMITTER_ACTION_SETUP 10 +#define ATOM_TRANSMITTER_ACTION_SETUP_VSEMPH 11 +#define ATOM_TRANSMITTER_ACTION_POWER_ON 12 +#define ATOM_TRANSMITTER_ACTION_POWER_OFF 13 + +// Following are used for DigTransmitterControlTable ver1.2 +typedef struct _ATOM_DIG_TRANSMITTER_CONFIG_V2 +{ +#if ATOM_BIG_ENDIAN + UCHAR ucTransmitterSel:2; //bit7:6: =0 Dig Transmitter 1 ( Uniphy AB ) + // =1 Dig Transmitter 2 ( Uniphy CD ) + // =2 Dig Transmitter 3 ( Uniphy EF ) + UCHAR ucReserved:1; + UCHAR fDPConnector:1; //bit4=0: DP connector =1: None DP connector + UCHAR ucEncoderSel:1; //bit3=0: Data/Clk path source from DIGA( DIG inst0 ). =1: Data/clk path source from DIGB ( DIG inst1 ) + UCHAR ucLinkSel:1; //bit2=0: Uniphy LINKA or C or E when fDualLinkConnector=0. when fDualLinkConnector=1, it means master link of dual link is A or C or E + // =1: Uniphy LINKB or D or F when fDualLinkConnector=0. when fDualLinkConnector=1, it means master link of dual link is B or D or F + + UCHAR fCoherentMode:1; //bit1=1: Coherent Mode ( for DVI/HDMI mode ) + UCHAR fDualLinkConnector:1; //bit0=1: Dual Link DVI connector +#else + UCHAR fDualLinkConnector:1; //bit0=1: Dual Link DVI connector + UCHAR fCoherentMode:1; //bit1=1: Coherent Mode ( for DVI/HDMI mode ) + UCHAR ucLinkSel:1; //bit2=0: Uniphy LINKA or C or E when fDualLinkConnector=0. when fDualLinkConnector=1, it means master link of dual link is A or C or E + // =1: Uniphy LINKB or D or F when fDualLinkConnector=0. when fDualLinkConnector=1, it means master link of dual link is B or D or F + UCHAR ucEncoderSel:1; //bit3=0: Data/Clk path source from DIGA( DIG inst0 ). =1: Data/clk path source from DIGB ( DIG inst1 ) + UCHAR fDPConnector:1; //bit4=0: DP connector =1: None DP connector + UCHAR ucReserved:1; + UCHAR ucTransmitterSel:2; //bit7:6: =0 Dig Transmitter 1 ( Uniphy AB ) + // =1 Dig Transmitter 2 ( Uniphy CD ) + // =2 Dig Transmitter 3 ( Uniphy EF ) +#endif +}ATOM_DIG_TRANSMITTER_CONFIG_V2; + +//ucConfig +//Bit0 +#define ATOM_TRANSMITTER_CONFIG_V2_DUAL_LINK_CONNECTOR 0x01 + +//Bit1 +#define ATOM_TRANSMITTER_CONFIG_V2_COHERENT 0x02 + +//Bit2 +#define ATOM_TRANSMITTER_CONFIG_V2_LINK_SEL_MASK 0x04 +#define ATOM_TRANSMITTER_CONFIG_V2_LINKA 0x00 +#define ATOM_TRANSMITTER_CONFIG_V2_LINKB 0x04 + +// Bit3 +#define ATOM_TRANSMITTER_CONFIG_V2_ENCODER_SEL_MASK 0x08 +#define ATOM_TRANSMITTER_CONFIG_V2_DIG1_ENCODER 0x00 // only used when ucAction == ATOM_TRANSMITTER_ACTION_ENABLE or ATOM_TRANSMITTER_ACTION_SETUP +#define ATOM_TRANSMITTER_CONFIG_V2_DIG2_ENCODER 0x08 // only used when ucAction == ATOM_TRANSMITTER_ACTION_ENABLE or ATOM_TRANSMITTER_ACTION_SETUP + +// Bit4 +#define ATOM_TRASMITTER_CONFIG_V2_DP_CONNECTOR 0x10 + +// Bit7:6 +#define ATOM_TRANSMITTER_CONFIG_V2_TRANSMITTER_SEL_MASK 0xC0 +#define ATOM_TRANSMITTER_CONFIG_V2_TRANSMITTER1 0x00 //AB +#define ATOM_TRANSMITTER_CONFIG_V2_TRANSMITTER2 0x40 //CD +#define ATOM_TRANSMITTER_CONFIG_V2_TRANSMITTER3 0x80 //EF + +typedef struct _DIG_TRANSMITTER_CONTROL_PARAMETERS_V2 +{ + union + { + USHORT usPixelClock; // in 10KHz; for bios convenient + USHORT usInitInfo; // when init uniphy,lower 8bit is used for connector type defined in objectid.h + ATOM_DP_VS_MODE asMode; // DP Voltage swing mode + }; + ATOM_DIG_TRANSMITTER_CONFIG_V2 acConfig; + UCHAR ucAction; // define as ATOM_TRANSMITER_ACTION_XXX + UCHAR ucReserved[4]; +}DIG_TRANSMITTER_CONTROL_PARAMETERS_V2; + +typedef struct _ATOM_DIG_TRANSMITTER_CONFIG_V3 +{ +#if ATOM_BIG_ENDIAN + UCHAR ucTransmitterSel:2; //bit7:6: =0 Dig Transmitter 1 ( Uniphy AB ) + // =1 Dig Transmitter 2 ( Uniphy CD ) + // =2 Dig Transmitter 3 ( Uniphy EF ) + UCHAR ucRefClkSource:2; //bit5:4: PPLL1 =0, PPLL2=1, EXT_CLK=2 + UCHAR ucEncoderSel:1; //bit3=0: Data/Clk path source from DIGA/C/E. =1: Data/clk path source from DIGB/D/F + UCHAR ucLinkSel:1; //bit2=0: Uniphy LINKA or C or E when fDualLinkConnector=0. when fDualLinkConnector=1, it means master link of dual link is A or C or E + // =1: Uniphy LINKB or D or F when fDualLinkConnector=0. when fDualLinkConnector=1, it means master link of dual link is B or D or F + UCHAR fCoherentMode:1; //bit1=1: Coherent Mode ( for DVI/HDMI mode ) + UCHAR fDualLinkConnector:1; //bit0=1: Dual Link DVI connector +#else + UCHAR fDualLinkConnector:1; //bit0=1: Dual Link DVI connector + UCHAR fCoherentMode:1; //bit1=1: Coherent Mode ( for DVI/HDMI mode ) + UCHAR ucLinkSel:1; //bit2=0: Uniphy LINKA or C or E when fDualLinkConnector=0. when fDualLinkConnector=1, it means master link of dual link is A or C or E + // =1: Uniphy LINKB or D or F when fDualLinkConnector=0. when fDualLinkConnector=1, it means master link of dual link is B or D or F + UCHAR ucEncoderSel:1; //bit3=0: Data/Clk path source from DIGA/C/E. =1: Data/clk path source from DIGB/D/F + UCHAR ucRefClkSource:2; //bit5:4: PPLL1 =0, PPLL2=1, EXT_CLK=2 + UCHAR ucTransmitterSel:2; //bit7:6: =0 Dig Transmitter 1 ( Uniphy AB ) + // =1 Dig Transmitter 2 ( Uniphy CD ) + // =2 Dig Transmitter 3 ( Uniphy EF ) +#endif +}ATOM_DIG_TRANSMITTER_CONFIG_V3; + + +typedef struct _DIG_TRANSMITTER_CONTROL_PARAMETERS_V3 +{ + union + { + USHORT usPixelClock; // in 10KHz; for bios convenient + USHORT usInitInfo; // when init uniphy,lower 8bit is used for connector type defined in objectid.h + ATOM_DP_VS_MODE asMode; // DP Voltage swing mode + }; + ATOM_DIG_TRANSMITTER_CONFIG_V3 acConfig; + UCHAR ucAction; // define as ATOM_TRANSMITER_ACTION_XXX + UCHAR ucLaneNum; + UCHAR ucReserved[3]; +}DIG_TRANSMITTER_CONTROL_PARAMETERS_V3; + +//ucConfig +//Bit0 +#define ATOM_TRANSMITTER_CONFIG_V3_DUAL_LINK_CONNECTOR 0x01 + +//Bit1 +#define ATOM_TRANSMITTER_CONFIG_V3_COHERENT 0x02 + +//Bit2 +#define ATOM_TRANSMITTER_CONFIG_V3_LINK_SEL_MASK 0x04 +#define ATOM_TRANSMITTER_CONFIG_V3_LINKA 0x00 +#define ATOM_TRANSMITTER_CONFIG_V3_LINKB 0x04 + +// Bit3 +#define ATOM_TRANSMITTER_CONFIG_V3_ENCODER_SEL_MASK 0x08 +#define ATOM_TRANSMITTER_CONFIG_V3_DIG1_ENCODER 0x00 +#define ATOM_TRANSMITTER_CONFIG_V3_DIG2_ENCODER 0x08 + +// Bit5:4 +#define ATOM_TRASMITTER_CONFIG_V3_REFCLK_SEL_MASK 0x30 +#define ATOM_TRASMITTER_CONFIG_V3_P1PLL 0x00 +#define ATOM_TRASMITTER_CONFIG_V3_P2PLL 0x10 +#define ATOM_TRASMITTER_CONFIG_V3_REFCLK_SRC_EXT 0x20 + +// Bit7:6 +#define ATOM_TRANSMITTER_CONFIG_V3_TRANSMITTER_SEL_MASK 0xC0 +#define ATOM_TRANSMITTER_CONFIG_V3_TRANSMITTER1 0x00 //AB +#define ATOM_TRANSMITTER_CONFIG_V3_TRANSMITTER2 0x40 //CD +#define ATOM_TRANSMITTER_CONFIG_V3_TRANSMITTER3 0x80 //EF + + +/****************************************************************************/ +// Structures used by UNIPHYTransmitterControlTable V1.4 +// ASIC Families: NI +// ucTableFormatRevision=1 +// ucTableContentRevision=4 +/****************************************************************************/ +typedef struct _ATOM_DP_VS_MODE_V4 +{ + UCHAR ucLaneSel; + union + { + UCHAR ucLaneSet; + struct { +#if ATOM_BIG_ENDIAN + UCHAR ucPOST_CURSOR2:2; //Bit[7:6] Post Cursor2 Level <= New in V4 + UCHAR ucPRE_EMPHASIS:3; //Bit[5:3] Pre-emphasis Level + UCHAR ucVOLTAGE_SWING:3; //Bit[2:0] Voltage Swing Level +#else + UCHAR ucVOLTAGE_SWING:3; //Bit[2:0] Voltage Swing Level + UCHAR ucPRE_EMPHASIS:3; //Bit[5:3] Pre-emphasis Level + UCHAR ucPOST_CURSOR2:2; //Bit[7:6] Post Cursor2 Level <= New in V4 +#endif + }; + }; +}ATOM_DP_VS_MODE_V4; + +typedef struct _ATOM_DIG_TRANSMITTER_CONFIG_V4 +{ +#if ATOM_BIG_ENDIAN + UCHAR ucTransmitterSel:2; //bit7:6: =0 Dig Transmitter 1 ( Uniphy AB ) + // =1 Dig Transmitter 2 ( Uniphy CD ) + // =2 Dig Transmitter 3 ( Uniphy EF ) + UCHAR ucRefClkSource:2; //bit5:4: PPLL1 =0, PPLL2=1, DCPLL=2, EXT_CLK=3 <= New + UCHAR ucEncoderSel:1; //bit3=0: Data/Clk path source from DIGA/C/E. =1: Data/clk path source from DIGB/D/F + UCHAR ucLinkSel:1; //bit2=0: Uniphy LINKA or C or E when fDualLinkConnector=0. when fDualLinkConnector=1, it means master link of dual link is A or C or E + // =1: Uniphy LINKB or D or F when fDualLinkConnector=0. when fDualLinkConnector=1, it means master link of dual link is B or D or F + UCHAR fCoherentMode:1; //bit1=1: Coherent Mode ( for DVI/HDMI mode ) + UCHAR fDualLinkConnector:1; //bit0=1: Dual Link DVI connector +#else + UCHAR fDualLinkConnector:1; //bit0=1: Dual Link DVI connector + UCHAR fCoherentMode:1; //bit1=1: Coherent Mode ( for DVI/HDMI mode ) + UCHAR ucLinkSel:1; //bit2=0: Uniphy LINKA or C or E when fDualLinkConnector=0. when fDualLinkConnector=1, it means master link of dual link is A or C or E + // =1: Uniphy LINKB or D or F when fDualLinkConnector=0. when fDualLinkConnector=1, it means master link of dual link is B or D or F + UCHAR ucEncoderSel:1; //bit3=0: Data/Clk path source from DIGA/C/E. =1: Data/clk path source from DIGB/D/F + UCHAR ucRefClkSource:2; //bit5:4: PPLL1 =0, PPLL2=1, DCPLL=2, EXT_CLK=3 <= New + UCHAR ucTransmitterSel:2; //bit7:6: =0 Dig Transmitter 1 ( Uniphy AB ) + // =1 Dig Transmitter 2 ( Uniphy CD ) + // =2 Dig Transmitter 3 ( Uniphy EF ) +#endif +}ATOM_DIG_TRANSMITTER_CONFIG_V4; + +typedef struct _DIG_TRANSMITTER_CONTROL_PARAMETERS_V4 +{ + union + { + USHORT usPixelClock; // in 10KHz; for bios convenient + USHORT usInitInfo; // when init uniphy,lower 8bit is used for connector type defined in objectid.h + ATOM_DP_VS_MODE_V4 asMode; // DP Voltage swing mode Redefined comparing to previous version + }; + union + { + ATOM_DIG_TRANSMITTER_CONFIG_V4 acConfig; + UCHAR ucConfig; + }; + UCHAR ucAction; // define as ATOM_TRANSMITER_ACTION_XXX + UCHAR ucLaneNum; + UCHAR ucReserved[3]; +}DIG_TRANSMITTER_CONTROL_PARAMETERS_V4; + +//ucConfig +//Bit0 +#define ATOM_TRANSMITTER_CONFIG_V4_DUAL_LINK_CONNECTOR 0x01 +//Bit1 +#define ATOM_TRANSMITTER_CONFIG_V4_COHERENT 0x02 +//Bit2 +#define ATOM_TRANSMITTER_CONFIG_V4_LINK_SEL_MASK 0x04 +#define ATOM_TRANSMITTER_CONFIG_V4_LINKA 0x00 +#define ATOM_TRANSMITTER_CONFIG_V4_LINKB 0x04 +// Bit3 +#define ATOM_TRANSMITTER_CONFIG_V4_ENCODER_SEL_MASK 0x08 +#define ATOM_TRANSMITTER_CONFIG_V4_DIG1_ENCODER 0x00 +#define ATOM_TRANSMITTER_CONFIG_V4_DIG2_ENCODER 0x08 +// Bit5:4 +#define ATOM_TRANSMITTER_CONFIG_V4_REFCLK_SEL_MASK 0x30 +#define ATOM_TRANSMITTER_CONFIG_V4_P1PLL 0x00 +#define ATOM_TRANSMITTER_CONFIG_V4_P2PLL 0x10 +#define ATOM_TRANSMITTER_CONFIG_V4_DCPLL 0x20 // New in _V4 +#define ATOM_TRANSMITTER_CONFIG_V4_REFCLK_SRC_EXT 0x30 // Changed comparing to V3 +// Bit7:6 +#define ATOM_TRANSMITTER_CONFIG_V4_TRANSMITTER_SEL_MASK 0xC0 +#define ATOM_TRANSMITTER_CONFIG_V4_TRANSMITTER1 0x00 //AB +#define ATOM_TRANSMITTER_CONFIG_V4_TRANSMITTER2 0x40 //CD +#define ATOM_TRANSMITTER_CONFIG_V4_TRANSMITTER3 0x80 //EF + + +/****************************************************************************/ +// Structures used by ExternalEncoderControlTable V1.3 +// ASIC Families: Evergreen, Llano, NI +// ucTableFormatRevision=1 +// ucTableContentRevision=3 +/****************************************************************************/ + +typedef struct _EXTERNAL_ENCODER_CONTROL_PARAMETERS_V3 +{ + union{ + USHORT usPixelClock; // pixel clock in 10Khz, valid when ucAction=SETUP/ENABLE_OUTPUT + USHORT usConnectorId; // connector id, valid when ucAction = INIT + }; + UCHAR ucConfig; // indicate which encoder, and DP link rate when ucAction = SETUP/ENABLE_OUTPUT + UCHAR ucAction; // + UCHAR ucEncoderMode; // encoder mode, only used when ucAction = SETUP/ENABLE_OUTPUT + UCHAR ucLaneNum; // lane number, only used when ucAction = SETUP/ENABLE_OUTPUT + UCHAR ucBitPerColor; // output bit per color, only valid when ucAction = SETUP/ENABLE_OUTPUT and ucEncodeMode= DP + UCHAR ucReserved; +}EXTERNAL_ENCODER_CONTROL_PARAMETERS_V3; + +// ucAction +#define EXTERNAL_ENCODER_ACTION_V3_DISABLE_OUTPUT 0x00 +#define EXTERNAL_ENCODER_ACTION_V3_ENABLE_OUTPUT 0x01 +#define EXTERNAL_ENCODER_ACTION_V3_ENCODER_INIT 0x07 +#define EXTERNAL_ENCODER_ACTION_V3_ENCODER_SETUP 0x0f +#define EXTERNAL_ENCODER_ACTION_V3_ENCODER_BLANKING_OFF 0x10 +#define EXTERNAL_ENCODER_ACTION_V3_ENCODER_BLANKING 0x11 +#define EXTERNAL_ENCODER_ACTION_V3_DACLOAD_DETECTION 0x12 +#define EXTERNAL_ENCODER_ACTION_V3_DDC_SETUP 0x14 + +// ucConfig +#define EXTERNAL_ENCODER_CONFIG_V3_DPLINKRATE_MASK 0x03 +#define EXTERNAL_ENCODER_CONFIG_V3_DPLINKRATE_1_62GHZ 0x00 +#define EXTERNAL_ENCODER_CONFIG_V3_DPLINKRATE_2_70GHZ 0x01 +#define EXTERNAL_ENCODER_CONFIG_V3_DPLINKRATE_5_40GHZ 0x02 +#define EXTERNAL_ENCODER_CONFIG_V3_ENCODER_SEL_MASK 0x70 +#define EXTERNAL_ENCODER_CONFIG_V3_ENCODER1 0x00 +#define EXTERNAL_ENCODER_CONFIG_V3_ENCODER2 0x10 +#define EXTERNAL_ENCODER_CONFIG_V3_ENCODER3 0x20 + +typedef struct _EXTERNAL_ENCODER_CONTROL_PS_ALLOCATION_V3 +{ + EXTERNAL_ENCODER_CONTROL_PARAMETERS_V3 sExtEncoder; + ULONG ulReserved[2]; +}EXTERNAL_ENCODER_CONTROL_PS_ALLOCATION_V3; + + +/****************************************************************************/ +// Structures used by DAC1OuputControlTable +// DAC2OuputControlTable +// LVTMAOutputControlTable (Before DEC30) +// TMDSAOutputControlTable (Before DEC30) +/****************************************************************************/ +typedef struct _DISPLAY_DEVICE_OUTPUT_CONTROL_PARAMETERS +{ + UCHAR ucAction; // Possible input:ATOM_ENABLE||ATOMDISABLE + // When the display is LCD, in addition to above: + // ATOM_LCD_BLOFF|| ATOM_LCD_BLON ||ATOM_LCD_BL_BRIGHTNESS_CONTROL||ATOM_LCD_SELFTEST_START|| + // ATOM_LCD_SELFTEST_STOP + + UCHAR aucPadding[3]; // padding to DWORD aligned +}DISPLAY_DEVICE_OUTPUT_CONTROL_PARAMETERS; + +#define DISPLAY_DEVICE_OUTPUT_CONTROL_PS_ALLOCATION DISPLAY_DEVICE_OUTPUT_CONTROL_PARAMETERS + + +#define CRT1_OUTPUT_CONTROL_PARAMETERS DISPLAY_DEVICE_OUTPUT_CONTROL_PARAMETERS +#define CRT1_OUTPUT_CONTROL_PS_ALLOCATION DISPLAY_DEVICE_OUTPUT_CONTROL_PS_ALLOCATION + +#define CRT2_OUTPUT_CONTROL_PARAMETERS DISPLAY_DEVICE_OUTPUT_CONTROL_PARAMETERS +#define CRT2_OUTPUT_CONTROL_PS_ALLOCATION DISPLAY_DEVICE_OUTPUT_CONTROL_PS_ALLOCATION + +#define CV1_OUTPUT_CONTROL_PARAMETERS DISPLAY_DEVICE_OUTPUT_CONTROL_PARAMETERS +#define CV1_OUTPUT_CONTROL_PS_ALLOCATION DISPLAY_DEVICE_OUTPUT_CONTROL_PS_ALLOCATION + +#define TV1_OUTPUT_CONTROL_PARAMETERS DISPLAY_DEVICE_OUTPUT_CONTROL_PARAMETERS +#define TV1_OUTPUT_CONTROL_PS_ALLOCATION DISPLAY_DEVICE_OUTPUT_CONTROL_PS_ALLOCATION + +#define DFP1_OUTPUT_CONTROL_PARAMETERS DISPLAY_DEVICE_OUTPUT_CONTROL_PARAMETERS +#define DFP1_OUTPUT_CONTROL_PS_ALLOCATION DISPLAY_DEVICE_OUTPUT_CONTROL_PS_ALLOCATION + +#define DFP2_OUTPUT_CONTROL_PARAMETERS DISPLAY_DEVICE_OUTPUT_CONTROL_PARAMETERS +#define DFP2_OUTPUT_CONTROL_PS_ALLOCATION DISPLAY_DEVICE_OUTPUT_CONTROL_PS_ALLOCATION + +#define LCD1_OUTPUT_CONTROL_PARAMETERS DISPLAY_DEVICE_OUTPUT_CONTROL_PARAMETERS +#define LCD1_OUTPUT_CONTROL_PS_ALLOCATION DISPLAY_DEVICE_OUTPUT_CONTROL_PS_ALLOCATION + +#define DVO_OUTPUT_CONTROL_PARAMETERS DISPLAY_DEVICE_OUTPUT_CONTROL_PARAMETERS +#define DVO_OUTPUT_CONTROL_PS_ALLOCATION DIG_TRANSMITTER_CONTROL_PS_ALLOCATION +#define DVO_OUTPUT_CONTROL_PARAMETERS_V3 DIG_TRANSMITTER_CONTROL_PARAMETERS + +/****************************************************************************/ +// Structures used by BlankCRTCTable +/****************************************************************************/ +typedef struct _BLANK_CRTC_PARAMETERS +{ + UCHAR ucCRTC; // ATOM_CRTC1 or ATOM_CRTC2 + UCHAR ucBlanking; // ATOM_BLANKING or ATOM_BLANKINGOFF + USHORT usBlackColorRCr; + USHORT usBlackColorGY; + USHORT usBlackColorBCb; +}BLANK_CRTC_PARAMETERS; +#define BLANK_CRTC_PS_ALLOCATION BLANK_CRTC_PARAMETERS + +/****************************************************************************/ +// Structures used by EnableCRTCTable +// EnableCRTCMemReqTable +// UpdateCRTC_DoubleBufferRegistersTable +/****************************************************************************/ +typedef struct _ENABLE_CRTC_PARAMETERS +{ + UCHAR ucCRTC; // ATOM_CRTC1 or ATOM_CRTC2 + UCHAR ucEnable; // ATOM_ENABLE or ATOM_DISABLE + UCHAR ucPadding[2]; +}ENABLE_CRTC_PARAMETERS; +#define ENABLE_CRTC_PS_ALLOCATION ENABLE_CRTC_PARAMETERS + +/****************************************************************************/ +// Structures used by SetCRTC_OverScanTable +/****************************************************************************/ +typedef struct _SET_CRTC_OVERSCAN_PARAMETERS +{ + USHORT usOverscanRight; // right + USHORT usOverscanLeft; // left + USHORT usOverscanBottom; // bottom + USHORT usOverscanTop; // top + UCHAR ucCRTC; // ATOM_CRTC1 or ATOM_CRTC2 + UCHAR ucPadding[3]; +}SET_CRTC_OVERSCAN_PARAMETERS; +#define SET_CRTC_OVERSCAN_PS_ALLOCATION SET_CRTC_OVERSCAN_PARAMETERS + +/****************************************************************************/ +// Structures used by SetCRTC_ReplicationTable +/****************************************************************************/ +typedef struct _SET_CRTC_REPLICATION_PARAMETERS +{ + UCHAR ucH_Replication; // horizontal replication + UCHAR ucV_Replication; // vertical replication + UCHAR usCRTC; // ATOM_CRTC1 or ATOM_CRTC2 + UCHAR ucPadding; +}SET_CRTC_REPLICATION_PARAMETERS; +#define SET_CRTC_REPLICATION_PS_ALLOCATION SET_CRTC_REPLICATION_PARAMETERS + +/****************************************************************************/ +// Structures used by SelectCRTC_SourceTable +/****************************************************************************/ +typedef struct _SELECT_CRTC_SOURCE_PARAMETERS +{ + UCHAR ucCRTC; // ATOM_CRTC1 or ATOM_CRTC2 + UCHAR ucDevice; // ATOM_DEVICE_CRT1|ATOM_DEVICE_CRT2|.... + UCHAR ucPadding[2]; +}SELECT_CRTC_SOURCE_PARAMETERS; +#define SELECT_CRTC_SOURCE_PS_ALLOCATION SELECT_CRTC_SOURCE_PARAMETERS + +typedef struct _SELECT_CRTC_SOURCE_PARAMETERS_V2 +{ + UCHAR ucCRTC; // ATOM_CRTC1 or ATOM_CRTC2 + UCHAR ucEncoderID; // DAC1/DAC2/TVOUT/DIG1/DIG2/DVO + UCHAR ucEncodeMode; // Encoding mode, only valid when using DIG1/DIG2/DVO + UCHAR ucPadding; +}SELECT_CRTC_SOURCE_PARAMETERS_V2; + +//ucEncoderID +//#define ASIC_INT_DAC1_ENCODER_ID 0x00 +//#define ASIC_INT_TV_ENCODER_ID 0x02 +//#define ASIC_INT_DIG1_ENCODER_ID 0x03 +//#define ASIC_INT_DAC2_ENCODER_ID 0x04 +//#define ASIC_EXT_TV_ENCODER_ID 0x06 +//#define ASIC_INT_DVO_ENCODER_ID 0x07 +//#define ASIC_INT_DIG2_ENCODER_ID 0x09 +//#define ASIC_EXT_DIG_ENCODER_ID 0x05 + +//ucEncodeMode +//#define ATOM_ENCODER_MODE_DP 0 +//#define ATOM_ENCODER_MODE_LVDS 1 +//#define ATOM_ENCODER_MODE_DVI 2 +//#define ATOM_ENCODER_MODE_HDMI 3 +//#define ATOM_ENCODER_MODE_SDVO 4 +//#define ATOM_ENCODER_MODE_TV 13 +//#define ATOM_ENCODER_MODE_CV 14 +//#define ATOM_ENCODER_MODE_CRT 15 + +/****************************************************************************/ +// Structures used by SetPixelClockTable +// GetPixelClockTable +/****************************************************************************/ +//Major revision=1., Minor revision=1 +typedef struct _PIXEL_CLOCK_PARAMETERS +{ + USHORT usPixelClock; // in 10kHz unit; for bios convenient = (RefClk*FB_Div)/(Ref_Div*Post_Div) + // 0 means disable PPLL + USHORT usRefDiv; // Reference divider + USHORT usFbDiv; // feedback divider + UCHAR ucPostDiv; // post divider + UCHAR ucFracFbDiv; // fractional feedback divider + UCHAR ucPpll; // ATOM_PPLL1 or ATOM_PPL2 + UCHAR ucRefDivSrc; // ATOM_PJITTER or ATO_NONPJITTER + UCHAR ucCRTC; // Which CRTC uses this Ppll + UCHAR ucPadding; +}PIXEL_CLOCK_PARAMETERS; + +//Major revision=1., Minor revision=2, add ucMiscIfno +//ucMiscInfo: +#define MISC_FORCE_REPROG_PIXEL_CLOCK 0x1 +#define MISC_DEVICE_INDEX_MASK 0xF0 +#define MISC_DEVICE_INDEX_SHIFT 4 + +typedef struct _PIXEL_CLOCK_PARAMETERS_V2 +{ + USHORT usPixelClock; // in 10kHz unit; for bios convenient = (RefClk*FB_Div)/(Ref_Div*Post_Div) + // 0 means disable PPLL + USHORT usRefDiv; // Reference divider + USHORT usFbDiv; // feedback divider + UCHAR ucPostDiv; // post divider + UCHAR ucFracFbDiv; // fractional feedback divider + UCHAR ucPpll; // ATOM_PPLL1 or ATOM_PPL2 + UCHAR ucRefDivSrc; // ATOM_PJITTER or ATO_NONPJITTER + UCHAR ucCRTC; // Which CRTC uses this Ppll + UCHAR ucMiscInfo; // Different bits for different purpose, bit [7:4] as device index, bit[0]=Force prog +}PIXEL_CLOCK_PARAMETERS_V2; + +//Major revision=1., Minor revision=3, structure/definition change +//ucEncoderMode: +//ATOM_ENCODER_MODE_DP +//ATOM_ENOCDER_MODE_LVDS +//ATOM_ENOCDER_MODE_DVI +//ATOM_ENOCDER_MODE_HDMI +//ATOM_ENOCDER_MODE_SDVO +//ATOM_ENCODER_MODE_TV 13 +//ATOM_ENCODER_MODE_CV 14 +//ATOM_ENCODER_MODE_CRT 15 + +//ucDVOConfig +//#define DVO_ENCODER_CONFIG_RATE_SEL 0x01 +//#define DVO_ENCODER_CONFIG_DDR_SPEED 0x00 +//#define DVO_ENCODER_CONFIG_SDR_SPEED 0x01 +//#define DVO_ENCODER_CONFIG_OUTPUT_SEL 0x0c +//#define DVO_ENCODER_CONFIG_LOW12BIT 0x00 +//#define DVO_ENCODER_CONFIG_UPPER12BIT 0x04 +//#define DVO_ENCODER_CONFIG_24BIT 0x08 + +//ucMiscInfo: also changed, see below +#define PIXEL_CLOCK_MISC_FORCE_PROG_PPLL 0x01 +#define PIXEL_CLOCK_MISC_VGA_MODE 0x02 +#define PIXEL_CLOCK_MISC_CRTC_SEL_MASK 0x04 +#define PIXEL_CLOCK_MISC_CRTC_SEL_CRTC1 0x00 +#define PIXEL_CLOCK_MISC_CRTC_SEL_CRTC2 0x04 +#define PIXEL_CLOCK_MISC_USE_ENGINE_FOR_DISPCLK 0x08 +#define PIXEL_CLOCK_MISC_REF_DIV_SRC 0x10 +// V1.4 for RoadRunner +#define PIXEL_CLOCK_V4_MISC_SS_ENABLE 0x10 +#define PIXEL_CLOCK_V4_MISC_COHERENT_MODE 0x20 + + +typedef struct _PIXEL_CLOCK_PARAMETERS_V3 +{ + USHORT usPixelClock; // in 10kHz unit; for bios convenient = (RefClk*FB_Div)/(Ref_Div*Post_Div) + // 0 means disable PPLL. For VGA PPLL,make sure this value is not 0. + USHORT usRefDiv; // Reference divider + USHORT usFbDiv; // feedback divider + UCHAR ucPostDiv; // post divider + UCHAR ucFracFbDiv; // fractional feedback divider + UCHAR ucPpll; // ATOM_PPLL1 or ATOM_PPL2 + UCHAR ucTransmitterId; // graphic encoder id defined in objectId.h + union + { + UCHAR ucEncoderMode; // encoder type defined as ATOM_ENCODER_MODE_DP/DVI/HDMI/ + UCHAR ucDVOConfig; // when use DVO, need to know SDR/DDR, 12bit or 24bit + }; + UCHAR ucMiscInfo; // bit[0]=Force program, bit[1]= set pclk for VGA, b[2]= CRTC sel + // bit[3]=0:use PPLL for dispclk source, =1: use engine clock for dispclock source + // bit[4]=0:use XTALIN as the source of reference divider,=1 use the pre-defined clock as the source of reference divider +}PIXEL_CLOCK_PARAMETERS_V3; + +#define PIXEL_CLOCK_PARAMETERS_LAST PIXEL_CLOCK_PARAMETERS_V2 +#define GET_PIXEL_CLOCK_PS_ALLOCATION PIXEL_CLOCK_PARAMETERS_LAST + +typedef struct _PIXEL_CLOCK_PARAMETERS_V5 +{ + UCHAR ucCRTC; // ATOM_CRTC1~6, indicate the CRTC controller to + // drive the pixel clock. not used for DCPLL case. + union{ + UCHAR ucReserved; + UCHAR ucFracFbDiv; // [gphan] temporary to prevent build problem. remove it after driver code is changed. + }; + USHORT usPixelClock; // target the pixel clock to drive the CRTC timing + // 0 means disable PPLL/DCPLL. + USHORT usFbDiv; // feedback divider integer part. + UCHAR ucPostDiv; // post divider. + UCHAR ucRefDiv; // Reference divider + UCHAR ucPpll; // ATOM_PPLL1/ATOM_PPLL2/ATOM_DCPLL + UCHAR ucTransmitterID; // ASIC encoder id defined in objectId.h, + // indicate which graphic encoder will be used. + UCHAR ucEncoderMode; // Encoder mode: + UCHAR ucMiscInfo; // bit[0]= Force program PPLL + // bit[1]= when VGA timing is used. + // bit[3:2]= HDMI panel bit depth: =0: 24bpp =1:30bpp, =2:32bpp + // bit[4]= RefClock source for PPLL. + // =0: XTLAIN( default mode ) + // =1: other external clock source, which is pre-defined + // by VBIOS depend on the feature required. + // bit[7:5]: reserved. + ULONG ulFbDivDecFrac; // 20 bit feedback divider decimal fraction part, range from 1~999999 ( 0.000001 to 0.999999 ) + +}PIXEL_CLOCK_PARAMETERS_V5; + +#define PIXEL_CLOCK_V5_MISC_FORCE_PROG_PPLL 0x01 +#define PIXEL_CLOCK_V5_MISC_VGA_MODE 0x02 +#define PIXEL_CLOCK_V5_MISC_HDMI_BPP_MASK 0x0c +#define PIXEL_CLOCK_V5_MISC_HDMI_24BPP 0x00 +#define PIXEL_CLOCK_V5_MISC_HDMI_30BPP 0x04 +#define PIXEL_CLOCK_V5_MISC_HDMI_32BPP 0x08 +#define PIXEL_CLOCK_V5_MISC_REF_DIV_SRC 0x10 + +typedef struct _CRTC_PIXEL_CLOCK_FREQ +{ +#if ATOM_BIG_ENDIAN + ULONG ucCRTC:8; // ATOM_CRTC1~6, indicate the CRTC controller to + // drive the pixel clock. not used for DCPLL case. + ULONG ulPixelClock:24; // target the pixel clock to drive the CRTC timing. + // 0 means disable PPLL/DCPLL. Expanded to 24 bits comparing to previous version. +#else + ULONG ulPixelClock:24; // target the pixel clock to drive the CRTC timing. + // 0 means disable PPLL/DCPLL. Expanded to 24 bits comparing to previous version. + ULONG ucCRTC:8; // ATOM_CRTC1~6, indicate the CRTC controller to + // drive the pixel clock. not used for DCPLL case. +#endif +}CRTC_PIXEL_CLOCK_FREQ; + +typedef struct _PIXEL_CLOCK_PARAMETERS_V6 +{ + union{ + CRTC_PIXEL_CLOCK_FREQ ulCrtcPclkFreq; // pixel clock and CRTC id frequency + ULONG ulDispEngClkFreq; // dispclk frequency + }; + USHORT usFbDiv; // feedback divider integer part. + UCHAR ucPostDiv; // post divider. + UCHAR ucRefDiv; // Reference divider + UCHAR ucPpll; // ATOM_PPLL1/ATOM_PPLL2/ATOM_DCPLL + UCHAR ucTransmitterID; // ASIC encoder id defined in objectId.h, + // indicate which graphic encoder will be used. + UCHAR ucEncoderMode; // Encoder mode: + UCHAR ucMiscInfo; // bit[0]= Force program PPLL + // bit[1]= when VGA timing is used. + // bit[3:2]= HDMI panel bit depth: =0: 24bpp =1:30bpp, =2:32bpp + // bit[4]= RefClock source for PPLL. + // =0: XTLAIN( default mode ) + // =1: other external clock source, which is pre-defined + // by VBIOS depend on the feature required. + // bit[7:5]: reserved. + ULONG ulFbDivDecFrac; // 20 bit feedback divider decimal fraction part, range from 1~999999 ( 0.000001 to 0.999999 ) + +}PIXEL_CLOCK_PARAMETERS_V6; + +#define PIXEL_CLOCK_V6_MISC_FORCE_PROG_PPLL 0x01 +#define PIXEL_CLOCK_V6_MISC_VGA_MODE 0x02 +#define PIXEL_CLOCK_V6_MISC_HDMI_BPP_MASK 0x0c +#define PIXEL_CLOCK_V6_MISC_HDMI_24BPP 0x00 +#define PIXEL_CLOCK_V6_MISC_HDMI_36BPP 0x04 +#define PIXEL_CLOCK_V6_MISC_HDMI_30BPP 0x08 +#define PIXEL_CLOCK_V6_MISC_HDMI_48BPP 0x0c +#define PIXEL_CLOCK_V6_MISC_REF_DIV_SRC 0x10 + +typedef struct _GET_DISP_PLL_STATUS_INPUT_PARAMETERS_V2 +{ + PIXEL_CLOCK_PARAMETERS_V3 sDispClkInput; +}GET_DISP_PLL_STATUS_INPUT_PARAMETERS_V2; + +typedef struct _GET_DISP_PLL_STATUS_OUTPUT_PARAMETERS_V2 +{ + UCHAR ucStatus; + UCHAR ucRefDivSrc; // =1: reference clock source from XTALIN, =0: source from PCIE ref clock + UCHAR ucReserved[2]; +}GET_DISP_PLL_STATUS_OUTPUT_PARAMETERS_V2; + +typedef struct _GET_DISP_PLL_STATUS_INPUT_PARAMETERS_V3 +{ + PIXEL_CLOCK_PARAMETERS_V5 sDispClkInput; +}GET_DISP_PLL_STATUS_INPUT_PARAMETERS_V3; + +/****************************************************************************/ +// Structures used by AdjustDisplayPllTable +/****************************************************************************/ +typedef struct _ADJUST_DISPLAY_PLL_PARAMETERS +{ + USHORT usPixelClock; + UCHAR ucTransmitterID; + UCHAR ucEncodeMode; + union + { + UCHAR ucDVOConfig; //if DVO, need passing link rate and output 12bitlow or 24bit + UCHAR ucConfig; //if none DVO, not defined yet + }; + UCHAR ucReserved[3]; +}ADJUST_DISPLAY_PLL_PARAMETERS; + +#define ADJUST_DISPLAY_CONFIG_SS_ENABLE 0x10 +#define ADJUST_DISPLAY_PLL_PS_ALLOCATION ADJUST_DISPLAY_PLL_PARAMETERS + +typedef struct _ADJUST_DISPLAY_PLL_INPUT_PARAMETERS_V3 +{ + USHORT usPixelClock; // target pixel clock + UCHAR ucTransmitterID; // GPU transmitter id defined in objectid.h + UCHAR ucEncodeMode; // encoder mode: CRT, LVDS, DP, TMDS or HDMI + UCHAR ucDispPllConfig; // display pll configure parameter defined as following DISPPLL_CONFIG_XXXX + UCHAR ucExtTransmitterID; // external encoder id. + UCHAR ucReserved[2]; +}ADJUST_DISPLAY_PLL_INPUT_PARAMETERS_V3; + +// usDispPllConfig v1.2 for RoadRunner +#define DISPPLL_CONFIG_DVO_RATE_SEL 0x0001 // need only when ucTransmitterID = DVO +#define DISPPLL_CONFIG_DVO_DDR_SPEED 0x0000 // need only when ucTransmitterID = DVO +#define DISPPLL_CONFIG_DVO_SDR_SPEED 0x0001 // need only when ucTransmitterID = DVO +#define DISPPLL_CONFIG_DVO_OUTPUT_SEL 0x000c // need only when ucTransmitterID = DVO +#define DISPPLL_CONFIG_DVO_LOW12BIT 0x0000 // need only when ucTransmitterID = DVO +#define DISPPLL_CONFIG_DVO_UPPER12BIT 0x0004 // need only when ucTransmitterID = DVO +#define DISPPLL_CONFIG_DVO_24BIT 0x0008 // need only when ucTransmitterID = DVO +#define DISPPLL_CONFIG_SS_ENABLE 0x0010 // Only used when ucEncoderMode = DP or LVDS +#define DISPPLL_CONFIG_COHERENT_MODE 0x0020 // Only used when ucEncoderMode = TMDS or HDMI +#define DISPPLL_CONFIG_DUAL_LINK 0x0040 // Only used when ucEncoderMode = TMDS or LVDS + + +typedef struct _ADJUST_DISPLAY_PLL_OUTPUT_PARAMETERS_V3 +{ + ULONG ulDispPllFreq; // return display PPLL freq which is used to generate the pixclock, and related idclk, symclk etc + UCHAR ucRefDiv; // if it is none-zero, it is used to be calculated the other ppll parameter fb_divider and post_div ( if it is not given ) + UCHAR ucPostDiv; // if it is none-zero, it is used to be calculated the other ppll parameter fb_divider + UCHAR ucReserved[2]; +}ADJUST_DISPLAY_PLL_OUTPUT_PARAMETERS_V3; + +typedef struct _ADJUST_DISPLAY_PLL_PS_ALLOCATION_V3 +{ + union + { + ADJUST_DISPLAY_PLL_INPUT_PARAMETERS_V3 sInput; + ADJUST_DISPLAY_PLL_OUTPUT_PARAMETERS_V3 sOutput; + }; +} ADJUST_DISPLAY_PLL_PS_ALLOCATION_V3; + +/****************************************************************************/ +// Structures used by EnableYUVTable +/****************************************************************************/ +typedef struct _ENABLE_YUV_PARAMETERS +{ + UCHAR ucEnable; // ATOM_ENABLE:Enable YUV or ATOM_DISABLE:Disable YUV (RGB) + UCHAR ucCRTC; // Which CRTC needs this YUV or RGB format + UCHAR ucPadding[2]; +}ENABLE_YUV_PARAMETERS; +#define ENABLE_YUV_PS_ALLOCATION ENABLE_YUV_PARAMETERS + +/****************************************************************************/ +// Structures used by GetMemoryClockTable +/****************************************************************************/ +typedef struct _GET_MEMORY_CLOCK_PARAMETERS +{ + ULONG ulReturnMemoryClock; // current memory speed in 10KHz unit +} GET_MEMORY_CLOCK_PARAMETERS; +#define GET_MEMORY_CLOCK_PS_ALLOCATION GET_MEMORY_CLOCK_PARAMETERS + +/****************************************************************************/ +// Structures used by GetEngineClockTable +/****************************************************************************/ +typedef struct _GET_ENGINE_CLOCK_PARAMETERS +{ + ULONG ulReturnEngineClock; // current engine speed in 10KHz unit +} GET_ENGINE_CLOCK_PARAMETERS; +#define GET_ENGINE_CLOCK_PS_ALLOCATION GET_ENGINE_CLOCK_PARAMETERS + +/****************************************************************************/ +// Following Structures and constant may be obsolete +/****************************************************************************/ +//Maxium 8 bytes,the data read in will be placed in the parameter space. +//Read operaion successeful when the paramter space is non-zero, otherwise read operation failed +typedef struct _READ_EDID_FROM_HW_I2C_DATA_PARAMETERS +{ + USHORT usPrescale; //Ratio between Engine clock and I2C clock + USHORT usVRAMAddress; //Address in Frame Buffer where to pace raw EDID + USHORT usStatus; //When use output: lower byte EDID checksum, high byte hardware status + //WHen use input: lower byte as 'byte to read':currently limited to 128byte or 1byte + UCHAR ucSlaveAddr; //Read from which slave + UCHAR ucLineNumber; //Read from which HW assisted line +}READ_EDID_FROM_HW_I2C_DATA_PARAMETERS; +#define READ_EDID_FROM_HW_I2C_DATA_PS_ALLOCATION READ_EDID_FROM_HW_I2C_DATA_PARAMETERS + + +#define ATOM_WRITE_I2C_FORMAT_PSOFFSET_PSDATABYTE 0 +#define ATOM_WRITE_I2C_FORMAT_PSOFFSET_PSTWODATABYTES 1 +#define ATOM_WRITE_I2C_FORMAT_PSCOUNTER_PSOFFSET_IDDATABLOCK 2 +#define ATOM_WRITE_I2C_FORMAT_PSCOUNTER_IDOFFSET_PLUS_IDDATABLOCK 3 +#define ATOM_WRITE_I2C_FORMAT_IDCOUNTER_IDOFFSET_IDDATABLOCK 4 + +typedef struct _WRITE_ONE_BYTE_HW_I2C_DATA_PARAMETERS +{ + USHORT usPrescale; //Ratio between Engine clock and I2C clock + USHORT usByteOffset; //Write to which byte + //Upper portion of usByteOffset is Format of data + //1bytePS+offsetPS + //2bytesPS+offsetPS + //blockID+offsetPS + //blockID+offsetID + //blockID+counterID+offsetID + UCHAR ucData; //PS data1 + UCHAR ucStatus; //Status byte 1=success, 2=failure, Also is used as PS data2 + UCHAR ucSlaveAddr; //Write to which slave + UCHAR ucLineNumber; //Write from which HW assisted line +}WRITE_ONE_BYTE_HW_I2C_DATA_PARAMETERS; + +#define WRITE_ONE_BYTE_HW_I2C_DATA_PS_ALLOCATION WRITE_ONE_BYTE_HW_I2C_DATA_PARAMETERS + +typedef struct _SET_UP_HW_I2C_DATA_PARAMETERS +{ + USHORT usPrescale; //Ratio between Engine clock and I2C clock + UCHAR ucSlaveAddr; //Write to which slave + UCHAR ucLineNumber; //Write from which HW assisted line +}SET_UP_HW_I2C_DATA_PARAMETERS; + + +/**************************************************************************/ +#define SPEED_FAN_CONTROL_PS_ALLOCATION WRITE_ONE_BYTE_HW_I2C_DATA_PARAMETERS + + +/****************************************************************************/ +// Structures used by PowerConnectorDetectionTable +/****************************************************************************/ +typedef struct _POWER_CONNECTOR_DETECTION_PARAMETERS +{ + UCHAR ucPowerConnectorStatus; //Used for return value 0: detected, 1:not detected + UCHAR ucPwrBehaviorId; + USHORT usPwrBudget; //how much power currently boot to in unit of watt +}POWER_CONNECTOR_DETECTION_PARAMETERS; + +typedef struct POWER_CONNECTOR_DETECTION_PS_ALLOCATION +{ + UCHAR ucPowerConnectorStatus; //Used for return value 0: detected, 1:not detected + UCHAR ucReserved; + USHORT usPwrBudget; //how much power currently boot to in unit of watt + WRITE_ONE_BYTE_HW_I2C_DATA_PS_ALLOCATION sReserved; +}POWER_CONNECTOR_DETECTION_PS_ALLOCATION; + +/****************************LVDS SS Command Table Definitions**********************/ + +/****************************************************************************/ +// Structures used by EnableSpreadSpectrumOnPPLLTable +/****************************************************************************/ +typedef struct _ENABLE_LVDS_SS_PARAMETERS +{ + USHORT usSpreadSpectrumPercentage; + UCHAR ucSpreadSpectrumType; //Bit1=0 Down Spread,=1 Center Spread. Bit1=1 Ext. =0 Int. Others:TBD + UCHAR ucSpreadSpectrumStepSize_Delay; //bits3:2 SS_STEP_SIZE; bit 6:4 SS_DELAY + UCHAR ucEnable; //ATOM_ENABLE or ATOM_DISABLE + UCHAR ucPadding[3]; +}ENABLE_LVDS_SS_PARAMETERS; + +//ucTableFormatRevision=1,ucTableContentRevision=2 +typedef struct _ENABLE_LVDS_SS_PARAMETERS_V2 +{ + USHORT usSpreadSpectrumPercentage; + UCHAR ucSpreadSpectrumType; //Bit1=0 Down Spread,=1 Center Spread. Bit1=1 Ext. =0 Int. Others:TBD + UCHAR ucSpreadSpectrumStep; // + UCHAR ucEnable; //ATOM_ENABLE or ATOM_DISABLE + UCHAR ucSpreadSpectrumDelay; + UCHAR ucSpreadSpectrumRange; + UCHAR ucPadding; +}ENABLE_LVDS_SS_PARAMETERS_V2; + +//This new structure is based on ENABLE_LVDS_SS_PARAMETERS but expands to SS on PPLL, so other devices can use SS. +typedef struct _ENABLE_SPREAD_SPECTRUM_ON_PPLL +{ + USHORT usSpreadSpectrumPercentage; + UCHAR ucSpreadSpectrumType; // Bit1=0 Down Spread,=1 Center Spread. Bit1=1 Ext. =0 Int. Others:TBD + UCHAR ucSpreadSpectrumStep; // + UCHAR ucEnable; // ATOM_ENABLE or ATOM_DISABLE + UCHAR ucSpreadSpectrumDelay; + UCHAR ucSpreadSpectrumRange; + UCHAR ucPpll; // ATOM_PPLL1/ATOM_PPLL2 +}ENABLE_SPREAD_SPECTRUM_ON_PPLL; + +typedef struct _ENABLE_SPREAD_SPECTRUM_ON_PPLL_V2 +{ + USHORT usSpreadSpectrumPercentage; + UCHAR ucSpreadSpectrumType; // Bit[0]: 0-Down Spread,1-Center Spread. + // Bit[1]: 1-Ext. 0-Int. + // Bit[3:2]: =0 P1PLL =1 P2PLL =2 DCPLL + // Bits[7:4] reserved + UCHAR ucEnable; // ATOM_ENABLE or ATOM_DISABLE + USHORT usSpreadSpectrumAmount; // Includes SS_AMOUNT_FBDIV[7:0] and SS_AMOUNT_NFRAC_SLIP[11:8] + USHORT usSpreadSpectrumStep; // SS_STEP_SIZE_DSFRAC +}ENABLE_SPREAD_SPECTRUM_ON_PPLL_V2; + +#define ATOM_PPLL_SS_TYPE_V2_DOWN_SPREAD 0x00 +#define ATOM_PPLL_SS_TYPE_V2_CENTRE_SPREAD 0x01 +#define ATOM_PPLL_SS_TYPE_V2_EXT_SPREAD 0x02 +#define ATOM_PPLL_SS_TYPE_V2_PPLL_SEL_MASK 0x0c +#define ATOM_PPLL_SS_TYPE_V2_P1PLL 0x00 +#define ATOM_PPLL_SS_TYPE_V2_P2PLL 0x04 +#define ATOM_PPLL_SS_TYPE_V2_DCPLL 0x08 +#define ATOM_PPLL_SS_AMOUNT_V2_FBDIV_MASK 0x00FF +#define ATOM_PPLL_SS_AMOUNT_V2_FBDIV_SHIFT 0 +#define ATOM_PPLL_SS_AMOUNT_V2_NFRAC_MASK 0x0F00 +#define ATOM_PPLL_SS_AMOUNT_V2_NFRAC_SHIFT 8 + +// Used by DCE5.0 + typedef struct _ENABLE_SPREAD_SPECTRUM_ON_PPLL_V3 +{ + USHORT usSpreadSpectrumAmountFrac; // SS_AMOUNT_DSFRAC New in DCE5.0 + UCHAR ucSpreadSpectrumType; // Bit[0]: 0-Down Spread,1-Center Spread. + // Bit[1]: 1-Ext. 0-Int. + // Bit[3:2]: =0 P1PLL =1 P2PLL =2 DCPLL + // Bits[7:4] reserved + UCHAR ucEnable; // ATOM_ENABLE or ATOM_DISABLE + USHORT usSpreadSpectrumAmount; // Includes SS_AMOUNT_FBDIV[7:0] and SS_AMOUNT_NFRAC_SLIP[11:8] + USHORT usSpreadSpectrumStep; // SS_STEP_SIZE_DSFRAC +}ENABLE_SPREAD_SPECTRUM_ON_PPLL_V3; + +#define ATOM_PPLL_SS_TYPE_V3_DOWN_SPREAD 0x00 +#define ATOM_PPLL_SS_TYPE_V3_CENTRE_SPREAD 0x01 +#define ATOM_PPLL_SS_TYPE_V3_EXT_SPREAD 0x02 +#define ATOM_PPLL_SS_TYPE_V3_PPLL_SEL_MASK 0x0c +#define ATOM_PPLL_SS_TYPE_V3_P1PLL 0x00 +#define ATOM_PPLL_SS_TYPE_V3_P2PLL 0x04 +#define ATOM_PPLL_SS_TYPE_V3_DCPLL 0x08 +#define ATOM_PPLL_SS_AMOUNT_V3_FBDIV_MASK 0x00FF +#define ATOM_PPLL_SS_AMOUNT_V3_FBDIV_SHIFT 0 +#define ATOM_PPLL_SS_AMOUNT_V3_NFRAC_MASK 0x0F00 +#define ATOM_PPLL_SS_AMOUNT_V3_NFRAC_SHIFT 8 + +#define ENABLE_SPREAD_SPECTRUM_ON_PPLL_PS_ALLOCATION ENABLE_SPREAD_SPECTRUM_ON_PPLL + +/**************************************************************************/ + +typedef struct _SET_PIXEL_CLOCK_PS_ALLOCATION +{ + PIXEL_CLOCK_PARAMETERS sPCLKInput; + ENABLE_SPREAD_SPECTRUM_ON_PPLL sReserved;//Caller doesn't need to init this portion +}SET_PIXEL_CLOCK_PS_ALLOCATION; + +#define ENABLE_VGA_RENDER_PS_ALLOCATION SET_PIXEL_CLOCK_PS_ALLOCATION + +/****************************************************************************/ +// Structures used by ### +/****************************************************************************/ +typedef struct _MEMORY_TRAINING_PARAMETERS +{ + ULONG ulTargetMemoryClock; //In 10Khz unit +}MEMORY_TRAINING_PARAMETERS; +#define MEMORY_TRAINING_PS_ALLOCATION MEMORY_TRAINING_PARAMETERS + + +/****************************LVDS and other encoder command table definitions **********************/ + + +/****************************************************************************/ +// Structures used by LVDSEncoderControlTable (Before DCE30) +// LVTMAEncoderControlTable (Before DCE30) +// TMDSAEncoderControlTable (Before DCE30) +/****************************************************************************/ +typedef struct _LVDS_ENCODER_CONTROL_PARAMETERS +{ + USHORT usPixelClock; // in 10KHz; for bios convenient + UCHAR ucMisc; // bit0=0: Enable single link + // =1: Enable dual link + // Bit1=0: 666RGB + // =1: 888RGB + UCHAR ucAction; // 0: turn off encoder + // 1: setup and turn on encoder +}LVDS_ENCODER_CONTROL_PARAMETERS; + +#define LVDS_ENCODER_CONTROL_PS_ALLOCATION LVDS_ENCODER_CONTROL_PARAMETERS + +#define TMDS1_ENCODER_CONTROL_PARAMETERS LVDS_ENCODER_CONTROL_PARAMETERS +#define TMDS1_ENCODER_CONTROL_PS_ALLOCATION TMDS1_ENCODER_CONTROL_PARAMETERS + +#define TMDS2_ENCODER_CONTROL_PARAMETERS TMDS1_ENCODER_CONTROL_PARAMETERS +#define TMDS2_ENCODER_CONTROL_PS_ALLOCATION TMDS2_ENCODER_CONTROL_PARAMETERS + + +//ucTableFormatRevision=1,ucTableContentRevision=2 +typedef struct _LVDS_ENCODER_CONTROL_PARAMETERS_V2 +{ + USHORT usPixelClock; // in 10KHz; for bios convenient + UCHAR ucMisc; // see PANEL_ENCODER_MISC_xx defintions below + UCHAR ucAction; // 0: turn off encoder + // 1: setup and turn on encoder + UCHAR ucTruncate; // bit0=0: Disable truncate + // =1: Enable truncate + // bit4=0: 666RGB + // =1: 888RGB + UCHAR ucSpatial; // bit0=0: Disable spatial dithering + // =1: Enable spatial dithering + // bit4=0: 666RGB + // =1: 888RGB + UCHAR ucTemporal; // bit0=0: Disable temporal dithering + // =1: Enable temporal dithering + // bit4=0: 666RGB + // =1: 888RGB + // bit5=0: Gray level 2 + // =1: Gray level 4 + UCHAR ucFRC; // bit4=0: 25FRC_SEL pattern E + // =1: 25FRC_SEL pattern F + // bit6:5=0: 50FRC_SEL pattern A + // =1: 50FRC_SEL pattern B + // =2: 50FRC_SEL pattern C + // =3: 50FRC_SEL pattern D + // bit7=0: 75FRC_SEL pattern E + // =1: 75FRC_SEL pattern F +}LVDS_ENCODER_CONTROL_PARAMETERS_V2; + +#define LVDS_ENCODER_CONTROL_PS_ALLOCATION_V2 LVDS_ENCODER_CONTROL_PARAMETERS_V2 + +#define TMDS1_ENCODER_CONTROL_PARAMETERS_V2 LVDS_ENCODER_CONTROL_PARAMETERS_V2 +#define TMDS1_ENCODER_CONTROL_PS_ALLOCATION_V2 TMDS1_ENCODER_CONTROL_PARAMETERS_V2 + +#define TMDS2_ENCODER_CONTROL_PARAMETERS_V2 TMDS1_ENCODER_CONTROL_PARAMETERS_V2 +#define TMDS2_ENCODER_CONTROL_PS_ALLOCATION_V2 TMDS2_ENCODER_CONTROL_PARAMETERS_V2 + +#define LVDS_ENCODER_CONTROL_PARAMETERS_V3 LVDS_ENCODER_CONTROL_PARAMETERS_V2 +#define LVDS_ENCODER_CONTROL_PS_ALLOCATION_V3 LVDS_ENCODER_CONTROL_PARAMETERS_V3 + +#define TMDS1_ENCODER_CONTROL_PARAMETERS_V3 LVDS_ENCODER_CONTROL_PARAMETERS_V3 +#define TMDS1_ENCODER_CONTROL_PS_ALLOCATION_V3 TMDS1_ENCODER_CONTROL_PARAMETERS_V3 + +#define TMDS2_ENCODER_CONTROL_PARAMETERS_V3 LVDS_ENCODER_CONTROL_PARAMETERS_V3 +#define TMDS2_ENCODER_CONTROL_PS_ALLOCATION_V3 TMDS2_ENCODER_CONTROL_PARAMETERS_V3 + +/****************************************************************************/ +// Structures used by ### +/****************************************************************************/ +typedef struct _ENABLE_EXTERNAL_TMDS_ENCODER_PARAMETERS +{ + UCHAR ucEnable; // Enable or Disable External TMDS encoder + UCHAR ucMisc; // Bit0=0:Enable Single link;=1:Enable Dual link;Bit1 {=0:666RGB, =1:888RGB} + UCHAR ucPadding[2]; +}ENABLE_EXTERNAL_TMDS_ENCODER_PARAMETERS; + +typedef struct _ENABLE_EXTERNAL_TMDS_ENCODER_PS_ALLOCATION +{ + ENABLE_EXTERNAL_TMDS_ENCODER_PARAMETERS sXTmdsEncoder; + WRITE_ONE_BYTE_HW_I2C_DATA_PS_ALLOCATION sReserved; //Caller doesn't need to init this portion +}ENABLE_EXTERNAL_TMDS_ENCODER_PS_ALLOCATION; + +#define ENABLE_EXTERNAL_TMDS_ENCODER_PARAMETERS_V2 LVDS_ENCODER_CONTROL_PARAMETERS_V2 + +typedef struct _ENABLE_EXTERNAL_TMDS_ENCODER_PS_ALLOCATION_V2 +{ + ENABLE_EXTERNAL_TMDS_ENCODER_PARAMETERS_V2 sXTmdsEncoder; + WRITE_ONE_BYTE_HW_I2C_DATA_PS_ALLOCATION sReserved; //Caller doesn't need to init this portion +}ENABLE_EXTERNAL_TMDS_ENCODER_PS_ALLOCATION_V2; + +typedef struct _EXTERNAL_ENCODER_CONTROL_PS_ALLOCATION +{ + DIG_ENCODER_CONTROL_PARAMETERS sDigEncoder; + WRITE_ONE_BYTE_HW_I2C_DATA_PS_ALLOCATION sReserved; +}EXTERNAL_ENCODER_CONTROL_PS_ALLOCATION; + +/****************************************************************************/ +// Structures used by DVOEncoderControlTable +/****************************************************************************/ +//ucTableFormatRevision=1,ucTableContentRevision=3 + +//ucDVOConfig: +#define DVO_ENCODER_CONFIG_RATE_SEL 0x01 +#define DVO_ENCODER_CONFIG_DDR_SPEED 0x00 +#define DVO_ENCODER_CONFIG_SDR_SPEED 0x01 +#define DVO_ENCODER_CONFIG_OUTPUT_SEL 0x0c +#define DVO_ENCODER_CONFIG_LOW12BIT 0x00 +#define DVO_ENCODER_CONFIG_UPPER12BIT 0x04 +#define DVO_ENCODER_CONFIG_24BIT 0x08 + +typedef struct _DVO_ENCODER_CONTROL_PARAMETERS_V3 +{ + USHORT usPixelClock; + UCHAR ucDVOConfig; + UCHAR ucAction; //ATOM_ENABLE/ATOM_DISABLE/ATOM_HPD_INIT + UCHAR ucReseved[4]; +}DVO_ENCODER_CONTROL_PARAMETERS_V3; +#define DVO_ENCODER_CONTROL_PS_ALLOCATION_V3 DVO_ENCODER_CONTROL_PARAMETERS_V3 + +//ucTableFormatRevision=1 +//ucTableContentRevision=3 structure is not changed but usMisc add bit 1 as another input for +// bit1=0: non-coherent mode +// =1: coherent mode + +//========================================================================================== +//Only change is here next time when changing encoder parameter definitions again! +#define LVDS_ENCODER_CONTROL_PARAMETERS_LAST LVDS_ENCODER_CONTROL_PARAMETERS_V3 +#define LVDS_ENCODER_CONTROL_PS_ALLOCATION_LAST LVDS_ENCODER_CONTROL_PARAMETERS_LAST + +#define TMDS1_ENCODER_CONTROL_PARAMETERS_LAST LVDS_ENCODER_CONTROL_PARAMETERS_V3 +#define TMDS1_ENCODER_CONTROL_PS_ALLOCATION_LAST TMDS1_ENCODER_CONTROL_PARAMETERS_LAST + +#define TMDS2_ENCODER_CONTROL_PARAMETERS_LAST LVDS_ENCODER_CONTROL_PARAMETERS_V3 +#define TMDS2_ENCODER_CONTROL_PS_ALLOCATION_LAST TMDS2_ENCODER_CONTROL_PARAMETERS_LAST + +#define DVO_ENCODER_CONTROL_PARAMETERS_LAST DVO_ENCODER_CONTROL_PARAMETERS +#define DVO_ENCODER_CONTROL_PS_ALLOCATION_LAST DVO_ENCODER_CONTROL_PS_ALLOCATION + +//========================================================================================== +#define PANEL_ENCODER_MISC_DUAL 0x01 +#define PANEL_ENCODER_MISC_COHERENT 0x02 +#define PANEL_ENCODER_MISC_TMDS_LINKB 0x04 +#define PANEL_ENCODER_MISC_HDMI_TYPE 0x08 + +#define PANEL_ENCODER_ACTION_DISABLE ATOM_DISABLE +#define PANEL_ENCODER_ACTION_ENABLE ATOM_ENABLE +#define PANEL_ENCODER_ACTION_COHERENTSEQ (ATOM_ENABLE+1) + +#define PANEL_ENCODER_TRUNCATE_EN 0x01 +#define PANEL_ENCODER_TRUNCATE_DEPTH 0x10 +#define PANEL_ENCODER_SPATIAL_DITHER_EN 0x01 +#define PANEL_ENCODER_SPATIAL_DITHER_DEPTH 0x10 +#define PANEL_ENCODER_TEMPORAL_DITHER_EN 0x01 +#define PANEL_ENCODER_TEMPORAL_DITHER_DEPTH 0x10 +#define PANEL_ENCODER_TEMPORAL_LEVEL_4 0x20 +#define PANEL_ENCODER_25FRC_MASK 0x10 +#define PANEL_ENCODER_25FRC_E 0x00 +#define PANEL_ENCODER_25FRC_F 0x10 +#define PANEL_ENCODER_50FRC_MASK 0x60 +#define PANEL_ENCODER_50FRC_A 0x00 +#define PANEL_ENCODER_50FRC_B 0x20 +#define PANEL_ENCODER_50FRC_C 0x40 +#define PANEL_ENCODER_50FRC_D 0x60 +#define PANEL_ENCODER_75FRC_MASK 0x80 +#define PANEL_ENCODER_75FRC_E 0x00 +#define PANEL_ENCODER_75FRC_F 0x80 + +/****************************************************************************/ +// Structures used by SetVoltageTable +/****************************************************************************/ +#define SET_VOLTAGE_TYPE_ASIC_VDDC 1 +#define SET_VOLTAGE_TYPE_ASIC_MVDDC 2 +#define SET_VOLTAGE_TYPE_ASIC_MVDDQ 3 +#define SET_VOLTAGE_TYPE_ASIC_VDDCI 4 +#define SET_VOLTAGE_INIT_MODE 5 +#define SET_VOLTAGE_GET_MAX_VOLTAGE 6 //Gets the Max. voltage for the soldered Asic + +#define SET_ASIC_VOLTAGE_MODE_ALL_SOURCE 0x1 +#define SET_ASIC_VOLTAGE_MODE_SOURCE_A 0x2 +#define SET_ASIC_VOLTAGE_MODE_SOURCE_B 0x4 + +#define SET_ASIC_VOLTAGE_MODE_SET_VOLTAGE 0x0 +#define SET_ASIC_VOLTAGE_MODE_GET_GPIOVAL 0x1 +#define SET_ASIC_VOLTAGE_MODE_GET_GPIOMASK 0x2 + +typedef struct _SET_VOLTAGE_PARAMETERS +{ + UCHAR ucVoltageType; // To tell which voltage to set up, VDDC/MVDDC/MVDDQ + UCHAR ucVoltageMode; // To set all, to set source A or source B or ... + UCHAR ucVoltageIndex; // An index to tell which voltage level + UCHAR ucReserved; +}SET_VOLTAGE_PARAMETERS; + +typedef struct _SET_VOLTAGE_PARAMETERS_V2 +{ + UCHAR ucVoltageType; // To tell which voltage to set up, VDDC/MVDDC/MVDDQ + UCHAR ucVoltageMode; // Not used, maybe use for state machine for differen power mode + USHORT usVoltageLevel; // real voltage level +}SET_VOLTAGE_PARAMETERS_V2; + +typedef struct _SET_VOLTAGE_PS_ALLOCATION +{ + SET_VOLTAGE_PARAMETERS sASICSetVoltage; + WRITE_ONE_BYTE_HW_I2C_DATA_PS_ALLOCATION sReserved; +}SET_VOLTAGE_PS_ALLOCATION; + +/****************************************************************************/ +// Structures used by TVEncoderControlTable +/****************************************************************************/ +typedef struct _TV_ENCODER_CONTROL_PARAMETERS +{ + USHORT usPixelClock; // in 10KHz; for bios convenient + UCHAR ucTvStandard; // See definition "ATOM_TV_NTSC ..." + UCHAR ucAction; // 0: turn off encoder + // 1: setup and turn on encoder +}TV_ENCODER_CONTROL_PARAMETERS; + +typedef struct _TV_ENCODER_CONTROL_PS_ALLOCATION +{ + TV_ENCODER_CONTROL_PARAMETERS sTVEncoder; + WRITE_ONE_BYTE_HW_I2C_DATA_PS_ALLOCATION sReserved; // Don't set this one +}TV_ENCODER_CONTROL_PS_ALLOCATION; + +//==============================Data Table Portion==================================== + +/****************************************************************************/ +// Structure used in Data.mtb +/****************************************************************************/ +typedef struct _ATOM_MASTER_LIST_OF_DATA_TABLES +{ + USHORT UtilityPipeLine; // Offest for the utility to get parser info,Don't change this position! + USHORT MultimediaCapabilityInfo; // Only used by MM Lib,latest version 1.1, not configuable from Bios, need to include the table to build Bios + USHORT MultimediaConfigInfo; // Only used by MM Lib,latest version 2.1, not configuable from Bios, need to include the table to build Bios + USHORT StandardVESA_Timing; // Only used by Bios + USHORT FirmwareInfo; // Shared by various SW components,latest version 1.4 + USHORT DAC_Info; // Will be obsolete from R600 + USHORT LCD_Info; // Shared by various SW components,latest version 1.3, was called LVDS_Info + USHORT TMDS_Info; // Will be obsolete from R600 + USHORT AnalogTV_Info; // Shared by various SW components,latest version 1.1 + USHORT SupportedDevicesInfo; // Will be obsolete from R600 + USHORT GPIO_I2C_Info; // Shared by various SW components,latest version 1.2 will be used from R600 + USHORT VRAM_UsageByFirmware; // Shared by various SW components,latest version 1.3 will be used from R600 + USHORT GPIO_Pin_LUT; // Shared by various SW components,latest version 1.1 + USHORT VESA_ToInternalModeLUT; // Only used by Bios + USHORT ComponentVideoInfo; // Shared by various SW components,latest version 2.1 will be used from R600 + USHORT PowerPlayInfo; // Shared by various SW components,latest version 2.1,new design from R600 + USHORT CompassionateData; // Will be obsolete from R600 + USHORT SaveRestoreInfo; // Only used by Bios + USHORT PPLL_SS_Info; // Shared by various SW components,latest version 1.2, used to call SS_Info, change to new name because of int ASIC SS info + USHORT OemInfo; // Defined and used by external SW, should be obsolete soon + USHORT XTMDS_Info; // Will be obsolete from R600 + USHORT MclkSS_Info; // Shared by various SW components,latest version 1.1, only enabled when ext SS chip is used + USHORT Object_Header; // Shared by various SW components,latest version 1.1 + USHORT IndirectIOAccess; // Only used by Bios,this table position can't change at all!! + USHORT MC_InitParameter; // Only used by command table + USHORT ASIC_VDDC_Info; // Will be obsolete from R600 + USHORT ASIC_InternalSS_Info; // New tabel name from R600, used to be called "ASIC_MVDDC_Info" + USHORT TV_VideoMode; // Only used by command table + USHORT VRAM_Info; // Only used by command table, latest version 1.3 + USHORT MemoryTrainingInfo; // Used for VBIOS and Diag utility for memory training purpose since R600. the new table rev start from 2.1 + USHORT IntegratedSystemInfo; // Shared by various SW components + USHORT ASIC_ProfilingInfo; // New table name from R600, used to be called "ASIC_VDDCI_Info" for pre-R600 + USHORT VoltageObjectInfo; // Shared by various SW components, latest version 1.1 + USHORT PowerSourceInfo; // Shared by various SW components, latest versoin 1.1 +}ATOM_MASTER_LIST_OF_DATA_TABLES; + +// For backward compatible +#define LVDS_Info LCD_Info + +typedef struct _ATOM_MASTER_DATA_TABLE +{ + ATOM_COMMON_TABLE_HEADER sHeader; + ATOM_MASTER_LIST_OF_DATA_TABLES ListOfDataTables; +}ATOM_MASTER_DATA_TABLE; + + +/****************************************************************************/ +// Structure used in MultimediaCapabilityInfoTable +/****************************************************************************/ +typedef struct _ATOM_MULTIMEDIA_CAPABILITY_INFO +{ + ATOM_COMMON_TABLE_HEADER sHeader; + ULONG ulSignature; // HW info table signature string "$ATI" + UCHAR ucI2C_Type; // I2C type (normal GP_IO, ImpactTV GP_IO, Dedicated I2C pin, etc) + UCHAR ucTV_OutInfo; // Type of TV out supported (3:0) and video out crystal frequency (6:4) and TV data port (7) + UCHAR ucVideoPortInfo; // Provides the video port capabilities + UCHAR ucHostPortInfo; // Provides host port configuration information +}ATOM_MULTIMEDIA_CAPABILITY_INFO; + +/****************************************************************************/ +// Structure used in MultimediaConfigInfoTable +/****************************************************************************/ +typedef struct _ATOM_MULTIMEDIA_CONFIG_INFO +{ + ATOM_COMMON_TABLE_HEADER sHeader; + ULONG ulSignature; // MM info table signature sting "$MMT" + UCHAR ucTunerInfo; // Type of tuner installed on the adapter (4:0) and video input for tuner (7:5) + UCHAR ucAudioChipInfo; // List the audio chip type (3:0) product type (4) and OEM revision (7:5) + UCHAR ucProductID; // Defines as OEM ID or ATI board ID dependent on product type setting + UCHAR ucMiscInfo1; // Tuner voltage (1:0) HW teletext support (3:2) FM audio decoder (5:4) reserved (6) audio scrambling (7) + UCHAR ucMiscInfo2; // I2S input config (0) I2S output config (1) I2S Audio Chip (4:2) SPDIF Output Config (5) reserved (7:6) + UCHAR ucMiscInfo3; // Video Decoder Type (3:0) Video In Standard/Crystal (7:4) + UCHAR ucMiscInfo4; // Video Decoder Host Config (2:0) reserved (7:3) + UCHAR ucVideoInput0Info;// Video Input 0 Type (1:0) F/B setting (2) physical connector ID (5:3) reserved (7:6) + UCHAR ucVideoInput1Info;// Video Input 1 Type (1:0) F/B setting (2) physical connector ID (5:3) reserved (7:6) + UCHAR ucVideoInput2Info;// Video Input 2 Type (1:0) F/B setting (2) physical connector ID (5:3) reserved (7:6) + UCHAR ucVideoInput3Info;// Video Input 3 Type (1:0) F/B setting (2) physical connector ID (5:3) reserved (7:6) + UCHAR ucVideoInput4Info;// Video Input 4 Type (1:0) F/B setting (2) physical connector ID (5:3) reserved (7:6) +}ATOM_MULTIMEDIA_CONFIG_INFO; + + +/****************************************************************************/ +// Structures used in FirmwareInfoTable +/****************************************************************************/ + +// usBIOSCapability Definition: +// Bit 0 = 0: Bios image is not Posted, =1:Bios image is Posted; +// Bit 1 = 0: Dual CRTC is not supported, =1: Dual CRTC is supported; +// Bit 2 = 0: Extended Desktop is not supported, =1: Extended Desktop is supported; +// Others: Reserved +#define ATOM_BIOS_INFO_ATOM_FIRMWARE_POSTED 0x0001 +#define ATOM_BIOS_INFO_DUAL_CRTC_SUPPORT 0x0002 +#define ATOM_BIOS_INFO_EXTENDED_DESKTOP_SUPPORT 0x0004 +#define ATOM_BIOS_INFO_MEMORY_CLOCK_SS_SUPPORT 0x0008 // (valid from v1.1 ~v1.4):=1: memclk SS enable, =0 memclk SS disable. +#define ATOM_BIOS_INFO_ENGINE_CLOCK_SS_SUPPORT 0x0010 // (valid from v1.1 ~v1.4):=1: engclk SS enable, =0 engclk SS disable. +#define ATOM_BIOS_INFO_BL_CONTROLLED_BY_GPU 0x0020 +#define ATOM_BIOS_INFO_WMI_SUPPORT 0x0040 +#define ATOM_BIOS_INFO_PPMODE_ASSIGNGED_BY_SYSTEM 0x0080 +#define ATOM_BIOS_INFO_HYPERMEMORY_SUPPORT 0x0100 +#define ATOM_BIOS_INFO_HYPERMEMORY_SIZE_MASK 0x1E00 +#define ATOM_BIOS_INFO_VPOST_WITHOUT_FIRST_MODE_SET 0x2000 +#define ATOM_BIOS_INFO_BIOS_SCRATCH6_SCL2_REDEFINE 0x4000 +#define ATOM_BIOS_INFO_MEMORY_CLOCK_EXT_SS_SUPPORT 0x0008 // (valid from v2.1 ): =1: memclk ss enable with external ss chip +#define ATOM_BIOS_INFO_ENGINE_CLOCK_EXT_SS_SUPPORT 0x0010 // (valid from v2.1 ): =1: engclk ss enable with external ss chip + +#ifndef _H2INC + +//Please don't add or expand this bitfield structure below, this one will retire soon.! +typedef struct _ATOM_FIRMWARE_CAPABILITY +{ +#if ATOM_BIG_ENDIAN + USHORT Reserved:3; + USHORT HyperMemory_Size:4; + USHORT HyperMemory_Support:1; + USHORT PPMode_Assigned:1; + USHORT WMI_SUPPORT:1; + USHORT GPUControlsBL:1; + USHORT EngineClockSS_Support:1; + USHORT MemoryClockSS_Support:1; + USHORT ExtendedDesktopSupport:1; + USHORT DualCRTC_Support:1; + USHORT FirmwarePosted:1; +#else + USHORT FirmwarePosted:1; + USHORT DualCRTC_Support:1; + USHORT ExtendedDesktopSupport:1; + USHORT MemoryClockSS_Support:1; + USHORT EngineClockSS_Support:1; + USHORT GPUControlsBL:1; + USHORT WMI_SUPPORT:1; + USHORT PPMode_Assigned:1; + USHORT HyperMemory_Support:1; + USHORT HyperMemory_Size:4; + USHORT Reserved:3; +#endif +}ATOM_FIRMWARE_CAPABILITY; + +typedef union _ATOM_FIRMWARE_CAPABILITY_ACCESS +{ + ATOM_FIRMWARE_CAPABILITY sbfAccess; + USHORT susAccess; +}ATOM_FIRMWARE_CAPABILITY_ACCESS; + +#else + +typedef union _ATOM_FIRMWARE_CAPABILITY_ACCESS +{ + USHORT susAccess; +}ATOM_FIRMWARE_CAPABILITY_ACCESS; + +#endif + +typedef struct _ATOM_FIRMWARE_INFO +{ + ATOM_COMMON_TABLE_HEADER sHeader; + ULONG ulFirmwareRevision; + ULONG ulDefaultEngineClock; //In 10Khz unit + ULONG ulDefaultMemoryClock; //In 10Khz unit + ULONG ulDriverTargetEngineClock; //In 10Khz unit + ULONG ulDriverTargetMemoryClock; //In 10Khz unit + ULONG ulMaxEngineClockPLL_Output; //In 10Khz unit + ULONG ulMaxMemoryClockPLL_Output; //In 10Khz unit + ULONG ulMaxPixelClockPLL_Output; //In 10Khz unit + ULONG ulASICMaxEngineClock; //In 10Khz unit + ULONG ulASICMaxMemoryClock; //In 10Khz unit + UCHAR ucASICMaxTemperature; + UCHAR ucPadding[3]; //Don't use them + ULONG aulReservedForBIOS[3]; //Don't use them + USHORT usMinEngineClockPLL_Input; //In 10Khz unit + USHORT usMaxEngineClockPLL_Input; //In 10Khz unit + USHORT usMinEngineClockPLL_Output; //In 10Khz unit + USHORT usMinMemoryClockPLL_Input; //In 10Khz unit + USHORT usMaxMemoryClockPLL_Input; //In 10Khz unit + USHORT usMinMemoryClockPLL_Output; //In 10Khz unit + USHORT usMaxPixelClock; //In 10Khz unit, Max. Pclk + USHORT usMinPixelClockPLL_Input; //In 10Khz unit + USHORT usMaxPixelClockPLL_Input; //In 10Khz unit + USHORT usMinPixelClockPLL_Output; //In 10Khz unit, the definitions above can't change!!! + ATOM_FIRMWARE_CAPABILITY_ACCESS usFirmwareCapability; + USHORT usReferenceClock; //In 10Khz unit + USHORT usPM_RTS_Location; //RTS PM4 starting location in ROM in 1Kb unit + UCHAR ucPM_RTS_StreamSize; //RTS PM4 packets in Kb unit + UCHAR ucDesign_ID; //Indicate what is the board design + UCHAR ucMemoryModule_ID; //Indicate what is the board design +}ATOM_FIRMWARE_INFO; + +typedef struct _ATOM_FIRMWARE_INFO_V1_2 +{ + ATOM_COMMON_TABLE_HEADER sHeader; + ULONG ulFirmwareRevision; + ULONG ulDefaultEngineClock; //In 10Khz unit + ULONG ulDefaultMemoryClock; //In 10Khz unit + ULONG ulDriverTargetEngineClock; //In 10Khz unit + ULONG ulDriverTargetMemoryClock; //In 10Khz unit + ULONG ulMaxEngineClockPLL_Output; //In 10Khz unit + ULONG ulMaxMemoryClockPLL_Output; //In 10Khz unit + ULONG ulMaxPixelClockPLL_Output; //In 10Khz unit + ULONG ulASICMaxEngineClock; //In 10Khz unit + ULONG ulASICMaxMemoryClock; //In 10Khz unit + UCHAR ucASICMaxTemperature; + UCHAR ucMinAllowedBL_Level; + UCHAR ucPadding[2]; //Don't use them + ULONG aulReservedForBIOS[2]; //Don't use them + ULONG ulMinPixelClockPLL_Output; //In 10Khz unit + USHORT usMinEngineClockPLL_Input; //In 10Khz unit + USHORT usMaxEngineClockPLL_Input; //In 10Khz unit + USHORT usMinEngineClockPLL_Output; //In 10Khz unit + USHORT usMinMemoryClockPLL_Input; //In 10Khz unit + USHORT usMaxMemoryClockPLL_Input; //In 10Khz unit + USHORT usMinMemoryClockPLL_Output; //In 10Khz unit + USHORT usMaxPixelClock; //In 10Khz unit, Max. Pclk + USHORT usMinPixelClockPLL_Input; //In 10Khz unit + USHORT usMaxPixelClockPLL_Input; //In 10Khz unit + USHORT usMinPixelClockPLL_Output; //In 10Khz unit - lower 16bit of ulMinPixelClockPLL_Output + ATOM_FIRMWARE_CAPABILITY_ACCESS usFirmwareCapability; + USHORT usReferenceClock; //In 10Khz unit + USHORT usPM_RTS_Location; //RTS PM4 starting location in ROM in 1Kb unit + UCHAR ucPM_RTS_StreamSize; //RTS PM4 packets in Kb unit + UCHAR ucDesign_ID; //Indicate what is the board design + UCHAR ucMemoryModule_ID; //Indicate what is the board design +}ATOM_FIRMWARE_INFO_V1_2; + +typedef struct _ATOM_FIRMWARE_INFO_V1_3 +{ + ATOM_COMMON_TABLE_HEADER sHeader; + ULONG ulFirmwareRevision; + ULONG ulDefaultEngineClock; //In 10Khz unit + ULONG ulDefaultMemoryClock; //In 10Khz unit + ULONG ulDriverTargetEngineClock; //In 10Khz unit + ULONG ulDriverTargetMemoryClock; //In 10Khz unit + ULONG ulMaxEngineClockPLL_Output; //In 10Khz unit + ULONG ulMaxMemoryClockPLL_Output; //In 10Khz unit + ULONG ulMaxPixelClockPLL_Output; //In 10Khz unit + ULONG ulASICMaxEngineClock; //In 10Khz unit + ULONG ulASICMaxMemoryClock; //In 10Khz unit + UCHAR ucASICMaxTemperature; + UCHAR ucMinAllowedBL_Level; + UCHAR ucPadding[2]; //Don't use them + ULONG aulReservedForBIOS; //Don't use them + ULONG ul3DAccelerationEngineClock;//In 10Khz unit + ULONG ulMinPixelClockPLL_Output; //In 10Khz unit + USHORT usMinEngineClockPLL_Input; //In 10Khz unit + USHORT usMaxEngineClockPLL_Input; //In 10Khz unit + USHORT usMinEngineClockPLL_Output; //In 10Khz unit + USHORT usMinMemoryClockPLL_Input; //In 10Khz unit + USHORT usMaxMemoryClockPLL_Input; //In 10Khz unit + USHORT usMinMemoryClockPLL_Output; //In 10Khz unit + USHORT usMaxPixelClock; //In 10Khz unit, Max. Pclk + USHORT usMinPixelClockPLL_Input; //In 10Khz unit + USHORT usMaxPixelClockPLL_Input; //In 10Khz unit + USHORT usMinPixelClockPLL_Output; //In 10Khz unit - lower 16bit of ulMinPixelClockPLL_Output + ATOM_FIRMWARE_CAPABILITY_ACCESS usFirmwareCapability; + USHORT usReferenceClock; //In 10Khz unit + USHORT usPM_RTS_Location; //RTS PM4 starting location in ROM in 1Kb unit + UCHAR ucPM_RTS_StreamSize; //RTS PM4 packets in Kb unit + UCHAR ucDesign_ID; //Indicate what is the board design + UCHAR ucMemoryModule_ID; //Indicate what is the board design +}ATOM_FIRMWARE_INFO_V1_3; + +typedef struct _ATOM_FIRMWARE_INFO_V1_4 +{ + ATOM_COMMON_TABLE_HEADER sHeader; + ULONG ulFirmwareRevision; + ULONG ulDefaultEngineClock; //In 10Khz unit + ULONG ulDefaultMemoryClock; //In 10Khz unit + ULONG ulDriverTargetEngineClock; //In 10Khz unit + ULONG ulDriverTargetMemoryClock; //In 10Khz unit + ULONG ulMaxEngineClockPLL_Output; //In 10Khz unit + ULONG ulMaxMemoryClockPLL_Output; //In 10Khz unit + ULONG ulMaxPixelClockPLL_Output; //In 10Khz unit + ULONG ulASICMaxEngineClock; //In 10Khz unit + ULONG ulASICMaxMemoryClock; //In 10Khz unit + UCHAR ucASICMaxTemperature; + UCHAR ucMinAllowedBL_Level; + USHORT usBootUpVDDCVoltage; //In MV unit + USHORT usLcdMinPixelClockPLL_Output; // In MHz unit + USHORT usLcdMaxPixelClockPLL_Output; // In MHz unit + ULONG ul3DAccelerationEngineClock;//In 10Khz unit + ULONG ulMinPixelClockPLL_Output; //In 10Khz unit + USHORT usMinEngineClockPLL_Input; //In 10Khz unit + USHORT usMaxEngineClockPLL_Input; //In 10Khz unit + USHORT usMinEngineClockPLL_Output; //In 10Khz unit + USHORT usMinMemoryClockPLL_Input; //In 10Khz unit + USHORT usMaxMemoryClockPLL_Input; //In 10Khz unit + USHORT usMinMemoryClockPLL_Output; //In 10Khz unit + USHORT usMaxPixelClock; //In 10Khz unit, Max. Pclk + USHORT usMinPixelClockPLL_Input; //In 10Khz unit + USHORT usMaxPixelClockPLL_Input; //In 10Khz unit + USHORT usMinPixelClockPLL_Output; //In 10Khz unit - lower 16bit of ulMinPixelClockPLL_Output + ATOM_FIRMWARE_CAPABILITY_ACCESS usFirmwareCapability; + USHORT usReferenceClock; //In 10Khz unit + USHORT usPM_RTS_Location; //RTS PM4 starting location in ROM in 1Kb unit + UCHAR ucPM_RTS_StreamSize; //RTS PM4 packets in Kb unit + UCHAR ucDesign_ID; //Indicate what is the board design + UCHAR ucMemoryModule_ID; //Indicate what is the board design +}ATOM_FIRMWARE_INFO_V1_4; + +//the structure below to be used from Cypress +typedef struct _ATOM_FIRMWARE_INFO_V2_1 +{ + ATOM_COMMON_TABLE_HEADER sHeader; + ULONG ulFirmwareRevision; + ULONG ulDefaultEngineClock; //In 10Khz unit + ULONG ulDefaultMemoryClock; //In 10Khz unit + ULONG ulReserved1; + ULONG ulReserved2; + ULONG ulMaxEngineClockPLL_Output; //In 10Khz unit + ULONG ulMaxMemoryClockPLL_Output; //In 10Khz unit + ULONG ulMaxPixelClockPLL_Output; //In 10Khz unit + ULONG ulBinaryAlteredInfo; //Was ulASICMaxEngineClock + ULONG ulDefaultDispEngineClkFreq; //In 10Khz unit + UCHAR ucReserved1; //Was ucASICMaxTemperature; + UCHAR ucMinAllowedBL_Level; + USHORT usBootUpVDDCVoltage; //In MV unit + USHORT usLcdMinPixelClockPLL_Output; // In MHz unit + USHORT usLcdMaxPixelClockPLL_Output; // In MHz unit + ULONG ulReserved4; //Was ulAsicMaximumVoltage + ULONG ulMinPixelClockPLL_Output; //In 10Khz unit + USHORT usMinEngineClockPLL_Input; //In 10Khz unit + USHORT usMaxEngineClockPLL_Input; //In 10Khz unit + USHORT usMinEngineClockPLL_Output; //In 10Khz unit + USHORT usMinMemoryClockPLL_Input; //In 10Khz unit + USHORT usMaxMemoryClockPLL_Input; //In 10Khz unit + USHORT usMinMemoryClockPLL_Output; //In 10Khz unit + USHORT usMaxPixelClock; //In 10Khz unit, Max. Pclk + USHORT usMinPixelClockPLL_Input; //In 10Khz unit + USHORT usMaxPixelClockPLL_Input; //In 10Khz unit + USHORT usMinPixelClockPLL_Output; //In 10Khz unit - lower 16bit of ulMinPixelClockPLL_Output + ATOM_FIRMWARE_CAPABILITY_ACCESS usFirmwareCapability; + USHORT usCoreReferenceClock; //In 10Khz unit + USHORT usMemoryReferenceClock; //In 10Khz unit + USHORT usUniphyDPModeExtClkFreq; //In 10Khz unit, if it is 0, In DP Mode Uniphy Input clock from internal PPLL, otherwise Input clock from external Spread clock + UCHAR ucMemoryModule_ID; //Indicate what is the board design + UCHAR ucReserved4[3]; +}ATOM_FIRMWARE_INFO_V2_1; + +//the structure below to be used from NI +//ucTableFormatRevision=2 +//ucTableContentRevision=2 +typedef struct _ATOM_FIRMWARE_INFO_V2_2 +{ + ATOM_COMMON_TABLE_HEADER sHeader; + ULONG ulFirmwareRevision; + ULONG ulDefaultEngineClock; //In 10Khz unit + ULONG ulDefaultMemoryClock; //In 10Khz unit + ULONG ulReserved[2]; + ULONG ulReserved1; //Was ulMaxEngineClockPLL_Output; //In 10Khz unit* + ULONG ulReserved2; //Was ulMaxMemoryClockPLL_Output; //In 10Khz unit* + ULONG ulMaxPixelClockPLL_Output; //In 10Khz unit + ULONG ulBinaryAlteredInfo; //Was ulASICMaxEngineClock ? + ULONG ulDefaultDispEngineClkFreq; //In 10Khz unit. This is the frequency before DCDTO, corresponding to usBootUpVDDCVoltage. + UCHAR ucReserved3; //Was ucASICMaxTemperature; + UCHAR ucMinAllowedBL_Level; + USHORT usBootUpVDDCVoltage; //In MV unit + USHORT usLcdMinPixelClockPLL_Output; // In MHz unit + USHORT usLcdMaxPixelClockPLL_Output; // In MHz unit + ULONG ulReserved4; //Was ulAsicMaximumVoltage + ULONG ulMinPixelClockPLL_Output; //In 10Khz unit + ULONG ulReserved5; //Was usMinEngineClockPLL_Input and usMaxEngineClockPLL_Input + ULONG ulReserved6; //Was usMinEngineClockPLL_Output and usMinMemoryClockPLL_Input + ULONG ulReserved7; //Was usMaxMemoryClockPLL_Input and usMinMemoryClockPLL_Output + USHORT usReserved11; //Was usMaxPixelClock; //In 10Khz unit, Max. Pclk used only for DAC + USHORT usMinPixelClockPLL_Input; //In 10Khz unit + USHORT usMaxPixelClockPLL_Input; //In 10Khz unit + USHORT usBootUpVDDCIVoltage; //In unit of mv; Was usMinPixelClockPLL_Output; + ATOM_FIRMWARE_CAPABILITY_ACCESS usFirmwareCapability; + USHORT usCoreReferenceClock; //In 10Khz unit + USHORT usMemoryReferenceClock; //In 10Khz unit + USHORT usUniphyDPModeExtClkFreq; //In 10Khz unit, if it is 0, In DP Mode Uniphy Input clock from internal PPLL, otherwise Input clock from external Spread clock + UCHAR ucMemoryModule_ID; //Indicate what is the board design + UCHAR ucReserved9[3]; + USHORT usBootUpMVDDCVoltage; //In unit of mv; Was usMinPixelClockPLL_Output; + USHORT usReserved12; + ULONG ulReserved10[3]; // New added comparing to previous version +}ATOM_FIRMWARE_INFO_V2_2; + +#define ATOM_FIRMWARE_INFO_LAST ATOM_FIRMWARE_INFO_V2_2 + +/****************************************************************************/ +// Structures used in IntegratedSystemInfoTable +/****************************************************************************/ +#define IGP_CAP_FLAG_DYNAMIC_CLOCK_EN 0x2 +#define IGP_CAP_FLAG_AC_CARD 0x4 +#define IGP_CAP_FLAG_SDVO_CARD 0x8 +#define IGP_CAP_FLAG_POSTDIV_BY_2_MODE 0x10 + +typedef struct _ATOM_INTEGRATED_SYSTEM_INFO +{ + ATOM_COMMON_TABLE_HEADER sHeader; + ULONG ulBootUpEngineClock; //in 10kHz unit + ULONG ulBootUpMemoryClock; //in 10kHz unit + ULONG ulMaxSystemMemoryClock; //in 10kHz unit + ULONG ulMinSystemMemoryClock; //in 10kHz unit + UCHAR ucNumberOfCyclesInPeriodHi; + UCHAR ucLCDTimingSel; //=0:not valid.!=0 sel this timing descriptor from LCD EDID. + USHORT usReserved1; + USHORT usInterNBVoltageLow; //An intermidiate PMW value to set the voltage + USHORT usInterNBVoltageHigh; //Another intermidiate PMW value to set the voltage + ULONG ulReserved[2]; + + USHORT usFSBClock; //In MHz unit + USHORT usCapabilityFlag; //Bit0=1 indicates the fake HDMI support,Bit1=0/1 for Dynamic clocking dis/enable + //Bit[3:2]== 0:No PCIE card, 1:AC card, 2:SDVO card + //Bit[4]==1: P/2 mode, ==0: P/1 mode + USHORT usPCIENBCfgReg7; //bit[7:0]=MUX_Sel, bit[9:8]=MUX_SEL_LEVEL2, bit[10]=Lane_Reversal + USHORT usK8MemoryClock; //in MHz unit + USHORT usK8SyncStartDelay; //in 0.01 us unit + USHORT usK8DataReturnTime; //in 0.01 us unit + UCHAR ucMaxNBVoltage; + UCHAR ucMinNBVoltage; + UCHAR ucMemoryType; //[7:4]=1:DDR1;=2:DDR2;=3:DDR3.[3:0] is reserved + UCHAR ucNumberOfCyclesInPeriod; //CG.FVTHROT_PWM_CTRL_REG0.NumberOfCyclesInPeriod + UCHAR ucStartingPWM_HighTime; //CG.FVTHROT_PWM_CTRL_REG0.StartingPWM_HighTime + UCHAR ucHTLinkWidth; //16 bit vs. 8 bit + UCHAR ucMaxNBVoltageHigh; + UCHAR ucMinNBVoltageHigh; +}ATOM_INTEGRATED_SYSTEM_INFO; + +/* Explanation on entries in ATOM_INTEGRATED_SYSTEM_INFO +ulBootUpMemoryClock: For Intel IGP,it's the UMA system memory clock + For AMD IGP,it's 0 if no SidePort memory installed or it's the boot-up SidePort memory clock +ulMaxSystemMemoryClock: For Intel IGP,it's the Max freq from memory SPD if memory runs in ASYNC mode or otherwise (SYNC mode) it's 0 + For AMD IGP,for now this can be 0 +ulMinSystemMemoryClock: For Intel IGP,it's 133MHz if memory runs in ASYNC mode or otherwise (SYNC mode) it's 0 + For AMD IGP,for now this can be 0 + +usFSBClock: For Intel IGP,it's FSB Freq + For AMD IGP,it's HT Link Speed + +usK8MemoryClock: For AMD IGP only. For RevF CPU, set it to 200 +usK8SyncStartDelay: For AMD IGP only. Memory access latency in K8, required for watermark calculation +usK8DataReturnTime: For AMD IGP only. Memory access latency in K8, required for watermark calculation + +VC:Voltage Control +ucMaxNBVoltage: Voltage regulator dependent PWM value. Low 8 bits of the value for the max voltage.Set this one to 0xFF if VC without PWM. Set this to 0x0 if no VC at all. +ucMinNBVoltage: Voltage regulator dependent PWM value. Low 8 bits of the value for the min voltage.Set this one to 0x00 if VC without PWM or no VC at all. + +ucNumberOfCyclesInPeriod: Indicate how many cycles when PWM duty is 100%. low 8 bits of the value. +ucNumberOfCyclesInPeriodHi: Indicate how many cycles when PWM duty is 100%. high 8 bits of the value.If the PWM has an inverter,set bit [7]==1,otherwise set it 0 + +ucMaxNBVoltageHigh: Voltage regulator dependent PWM value. High 8 bits of the value for the max voltage.Set this one to 0xFF if VC without PWM. Set this to 0x0 if no VC at all. +ucMinNBVoltageHigh: Voltage regulator dependent PWM value. High 8 bits of the value for the min voltage.Set this one to 0x00 if VC without PWM or no VC at all. + + +usInterNBVoltageLow: Voltage regulator dependent PWM value. The value makes the the voltage >=Min NB voltage but <=InterNBVoltageHigh. Set this to 0x0000 if VC without PWM or no VC at all. +usInterNBVoltageHigh: Voltage regulator dependent PWM value. The value makes the the voltage >=InterNBVoltageLow but <=Max NB voltage.Set this to 0x0000 if VC without PWM or no VC at all. +*/ + + +/* +The following IGP table is introduced from RS780, which is supposed to be put by SBIOS in FB before IGP VBIOS starts VPOST; +Then VBIOS will copy the whole structure to its image so all GPU SW components can access this data structure to get whatever they need. +The enough reservation should allow us to never change table revisions. Whenever needed, a GPU SW component can use reserved portion for new data entries. + +SW components can access the IGP system infor structure in the same way as before +*/ + + +typedef struct _ATOM_INTEGRATED_SYSTEM_INFO_V2 +{ + ATOM_COMMON_TABLE_HEADER sHeader; + ULONG ulBootUpEngineClock; //in 10kHz unit + ULONG ulReserved1[2]; //must be 0x0 for the reserved + ULONG ulBootUpUMAClock; //in 10kHz unit + ULONG ulBootUpSidePortClock; //in 10kHz unit + ULONG ulMinSidePortClock; //in 10kHz unit + ULONG ulReserved2[6]; //must be 0x0 for the reserved + ULONG ulSystemConfig; //see explanation below + ULONG ulBootUpReqDisplayVector; + ULONG ulOtherDisplayMisc; + ULONG ulDDISlot1Config; + ULONG ulDDISlot2Config; + UCHAR ucMemoryType; //[3:0]=1:DDR1;=2:DDR2;=3:DDR3.[7:4] is reserved + UCHAR ucUMAChannelNumber; + UCHAR ucDockingPinBit; + UCHAR ucDockingPinPolarity; + ULONG ulDockingPinCFGInfo; + ULONG ulCPUCapInfo; + USHORT usNumberOfCyclesInPeriod; + USHORT usMaxNBVoltage; + USHORT usMinNBVoltage; + USHORT usBootUpNBVoltage; + ULONG ulHTLinkFreq; //in 10Khz + USHORT usMinHTLinkWidth; + USHORT usMaxHTLinkWidth; + USHORT usUMASyncStartDelay; + USHORT usUMADataReturnTime; + USHORT usLinkStatusZeroTime; + USHORT usDACEfuse; //for storing badgap value (for RS880 only) + ULONG ulHighVoltageHTLinkFreq; // in 10Khz + ULONG ulLowVoltageHTLinkFreq; // in 10Khz + USHORT usMaxUpStreamHTLinkWidth; + USHORT usMaxDownStreamHTLinkWidth; + USHORT usMinUpStreamHTLinkWidth; + USHORT usMinDownStreamHTLinkWidth; + USHORT usFirmwareVersion; //0 means FW is not supported. Otherwise it's the FW version loaded by SBIOS and driver should enable FW. + USHORT usFullT0Time; // Input to calculate minimum HT link change time required by NB P-State. Unit is 0.01us. + ULONG ulReserved3[96]; //must be 0x0 +}ATOM_INTEGRATED_SYSTEM_INFO_V2; + +/* +ulBootUpEngineClock: Boot-up Engine Clock in 10Khz; +ulBootUpUMAClock: Boot-up UMA Clock in 10Khz; it must be 0x0 when UMA is not present +ulBootUpSidePortClock: Boot-up SidePort Clock in 10Khz; it must be 0x0 when SidePort Memory is not present,this could be equal to or less than maximum supported Sideport memory clock + +ulSystemConfig: +Bit[0]=1: PowerExpress mode =0 Non-PowerExpress mode; +Bit[1]=1: system boots up at AMD overdrived state or user customized mode. In this case, driver will just stick to this boot-up mode. No other PowerPlay state + =0: system boots up at driver control state. Power state depends on PowerPlay table. +Bit[2]=1: PWM method is used on NB voltage control. =0: GPIO method is used. +Bit[3]=1: Only one power state(Performance) will be supported. + =0: Multiple power states supported from PowerPlay table. +Bit[4]=1: CLMC is supported and enabled on current system. + =0: CLMC is not supported or enabled on current system. SBIOS need to support HT link/freq change through ATIF interface. +Bit[5]=1: Enable CDLW for all driver control power states. Max HT width is from SBIOS, while Min HT width is determined by display requirement. + =0: CDLW is disabled. If CLMC is enabled case, Min HT width will be set equal to Max HT width. If CLMC disabled case, Max HT width will be applied. +Bit[6]=1: High Voltage requested for all power states. In this case, voltage will be forced at 1.1v and powerplay table voltage drop/throttling request will be ignored. + =0: Voltage settings is determined by powerplay table. +Bit[7]=1: Enable CLMC as hybrid Mode. CDLD and CILR will be disabled in this case and we're using legacy C1E. This is workaround for CPU(Griffin) performance issue. + =0: Enable CLMC as regular mode, CDLD and CILR will be enabled. +Bit[8]=1: CDLF is supported and enabled on current system. + =0: CDLF is not supported or enabled on current system. +Bit[9]=1: DLL Shut Down feature is enabled on current system. + =0: DLL Shut Down feature is not enabled or supported on current system. + +ulBootUpReqDisplayVector: This dword is a bit vector indicates what display devices are requested during boot-up. Refer to ATOM_DEVICE_xxx_SUPPORT for the bit vector definitions. + +ulOtherDisplayMisc: [15:8]- Bootup LCD Expansion selection; 0-center, 1-full panel size expansion; + [7:0] - BootupTV standard selection; This is a bit vector to indicate what TV standards are supported by the system. Refer to ucTVSupportedStd definition; + +ulDDISlot1Config: Describes the PCIE lane configuration on this DDI PCIE slot (ADD2 card) or connector (Mobile design). + [3:0] - Bit vector to indicate PCIE lane config of the DDI slot/connector on chassis (bit 0=1 lane 3:0; bit 1=1 lane 7:4; bit 2=1 lane 11:8; bit 3=1 lane 15:12) + [7:4] - Bit vector to indicate PCIE lane config of the same DDI slot/connector on docking station (bit 4=1 lane 3:0; bit 5=1 lane 7:4; bit 6=1 lane 11:8; bit 7=1 lane 15:12) + When a DDI connector is not "paired" (meaming two connections mutualexclusive on chassis or docking, only one of them can be connected at one time. + in both chassis and docking, SBIOS has to duplicate the same PCIE lane info from chassis to docking or vice versa. For example: + one DDI connector is only populated in docking with PCIE lane 8-11, but there is no paired connection on chassis, SBIOS has to copy bit 6 to bit 2. + + [15:8] - Lane configuration attribute; + [23:16]- Connector type, possible value: + CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D + CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_D + CONNECTOR_OBJECT_ID_HDMI_TYPE_A + CONNECTOR_OBJECT_ID_DISPLAYPORT + CONNECTOR_OBJECT_ID_eDP + [31:24]- Reserved + +ulDDISlot2Config: Same as Slot1. +ucMemoryType: SidePort memory type, set it to 0x0 when Sideport memory is not installed. Driver needs this info to change sideport memory clock. Not for display in CCC. +For IGP, Hypermemory is the only memory type showed in CCC. + +ucUMAChannelNumber: how many channels for the UMA; + +ulDockingPinCFGInfo: [15:0]-Bus/Device/Function # to CFG to read this Docking Pin; [31:16]-reg offset in CFG to read this pin +ucDockingPinBit: which bit in this register to read the pin status; +ucDockingPinPolarity:Polarity of the pin when docked; + +ulCPUCapInfo: [7:0]=1:Griffin;[7:0]=2:Greyhound;[7:0]=3:K8, [7:0]=4:Pharaoh, other bits reserved for now and must be 0x0 + +usNumberOfCyclesInPeriod:Indicate how many cycles when PWM duty is 100%. + +usMaxNBVoltage:Max. voltage control value in either PWM or GPIO mode. +usMinNBVoltage:Min. voltage control value in either PWM or GPIO mode. + GPIO mode: both usMaxNBVoltage & usMinNBVoltage have a valid value ulSystemConfig.SYSTEM_CONFIG_USE_PWM_ON_VOLTAGE=0 + PWM mode: both usMaxNBVoltage & usMinNBVoltage have a valid value ulSystemConfig.SYSTEM_CONFIG_USE_PWM_ON_VOLTAGE=1 + GPU SW don't control mode: usMaxNBVoltage & usMinNBVoltage=0 and no care about ulSystemConfig.SYSTEM_CONFIG_USE_PWM_ON_VOLTAGE + +usBootUpNBVoltage:Boot-up voltage regulator dependent PWM value. + +ulHTLinkFreq: Bootup HT link Frequency in 10Khz. +usMinHTLinkWidth: Bootup minimum HT link width. If CDLW disabled, this is equal to usMaxHTLinkWidth. + If CDLW enabled, both upstream and downstream width should be the same during bootup. +usMaxHTLinkWidth: Bootup maximum HT link width. If CDLW disabled, this is equal to usMinHTLinkWidth. + If CDLW enabled, both upstream and downstream width should be the same during bootup. + +usUMASyncStartDelay: Memory access latency, required for watermark calculation +usUMADataReturnTime: Memory access latency, required for watermark calculation +usLinkStatusZeroTime:Memory access latency required for watermark calculation, set this to 0x0 for K8 CPU, set a proper value in 0.01 the unit of us +for Griffin or Greyhound. SBIOS needs to convert to actual time by: + if T0Ttime [5:4]=00b, then usLinkStatusZeroTime=T0Ttime [3:0]*0.1us (0.0 to 1.5us) + if T0Ttime [5:4]=01b, then usLinkStatusZeroTime=T0Ttime [3:0]*0.5us (0.0 to 7.5us) + if T0Ttime [5:4]=10b, then usLinkStatusZeroTime=T0Ttime [3:0]*2.0us (0.0 to 30us) + if T0Ttime [5:4]=11b, and T0Ttime [3:0]=0x0 to 0xa, then usLinkStatusZeroTime=T0Ttime [3:0]*20us (0.0 to 200us) + +ulHighVoltageHTLinkFreq: HT link frequency for power state with low voltage. If boot up runs in HT1, this must be 0. + This must be less than or equal to ulHTLinkFreq(bootup frequency). +ulLowVoltageHTLinkFreq: HT link frequency for power state with low voltage or voltage scaling 1.0v~1.1v. If boot up runs in HT1, this must be 0. + This must be less than or equal to ulHighVoltageHTLinkFreq. + +usMaxUpStreamHTLinkWidth: Asymmetric link width support in the future, to replace usMaxHTLinkWidth. Not used for now. +usMaxDownStreamHTLinkWidth: same as above. +usMinUpStreamHTLinkWidth: Asymmetric link width support in the future, to replace usMinHTLinkWidth. Not used for now. +usMinDownStreamHTLinkWidth: same as above. +*/ + +// ATOM_INTEGRATED_SYSTEM_INFO::ulCPUCapInfo - CPU type definition +#define INTEGRATED_SYSTEM_INFO__UNKNOWN_CPU 0 +#define INTEGRATED_SYSTEM_INFO__AMD_CPU__GRIFFIN 1 +#define INTEGRATED_SYSTEM_INFO__AMD_CPU__GREYHOUND 2 +#define INTEGRATED_SYSTEM_INFO__AMD_CPU__K8 3 +#define INTEGRATED_SYSTEM_INFO__AMD_CPU__PHARAOH 4 + +#define INTEGRATED_SYSTEM_INFO__AMD_CPU__MAX_CODE INTEGRATED_SYSTEM_INFO__AMD_CPU__PHARAOH // this deff reflects max defined CPU code + +#define SYSTEM_CONFIG_POWEREXPRESS_ENABLE 0x00000001 +#define SYSTEM_CONFIG_RUN_AT_OVERDRIVE_ENGINE 0x00000002 +#define SYSTEM_CONFIG_USE_PWM_ON_VOLTAGE 0x00000004 +#define SYSTEM_CONFIG_PERFORMANCE_POWERSTATE_ONLY 0x00000008 +#define SYSTEM_CONFIG_CLMC_ENABLED 0x00000010 +#define SYSTEM_CONFIG_CDLW_ENABLED 0x00000020 +#define SYSTEM_CONFIG_HIGH_VOLTAGE_REQUESTED 0x00000040 +#define SYSTEM_CONFIG_CLMC_HYBRID_MODE_ENABLED 0x00000080 +#define SYSTEM_CONFIG_CDLF_ENABLED 0x00000100 +#define SYSTEM_CONFIG_DLL_SHUTDOWN_ENABLED 0x00000200 + +#define IGP_DDI_SLOT_LANE_CONFIG_MASK 0x000000FF + +#define b0IGP_DDI_SLOT_LANE_MAP_MASK 0x0F +#define b0IGP_DDI_SLOT_DOCKING_LANE_MAP_MASK 0xF0 +#define b0IGP_DDI_SLOT_CONFIG_LANE_0_3 0x01 +#define b0IGP_DDI_SLOT_CONFIG_LANE_4_7 0x02 +#define b0IGP_DDI_SLOT_CONFIG_LANE_8_11 0x04 +#define b0IGP_DDI_SLOT_CONFIG_LANE_12_15 0x08 + +#define IGP_DDI_SLOT_ATTRIBUTE_MASK 0x0000FF00 +#define IGP_DDI_SLOT_CONFIG_REVERSED 0x00000100 +#define b1IGP_DDI_SLOT_CONFIG_REVERSED 0x01 + +#define IGP_DDI_SLOT_CONNECTOR_TYPE_MASK 0x00FF0000 + +// IntegratedSystemInfoTable new Rev is V5 after V2, because of the real rev of V2 is v1.4. This rev is used for RR +typedef struct _ATOM_INTEGRATED_SYSTEM_INFO_V5 +{ + ATOM_COMMON_TABLE_HEADER sHeader; + ULONG ulBootUpEngineClock; //in 10kHz unit + ULONG ulDentistVCOFreq; //Dentist VCO clock in 10kHz unit, the source of GPU SCLK, LCLK, UCLK and VCLK. + ULONG ulLClockFreq; //GPU Lclk freq in 10kHz unit, have relationship with NCLK in NorthBridge + ULONG ulBootUpUMAClock; //in 10kHz unit + ULONG ulReserved1[8]; //must be 0x0 for the reserved + ULONG ulBootUpReqDisplayVector; + ULONG ulOtherDisplayMisc; + ULONG ulReserved2[4]; //must be 0x0 for the reserved + ULONG ulSystemConfig; //TBD + ULONG ulCPUCapInfo; //TBD + USHORT usMaxNBVoltage; //high NB voltage, calculated using current VDDNB (D24F2xDC) and VDDNB offset fuse; + USHORT usMinNBVoltage; //low NB voltage, calculated using current VDDNB (D24F2xDC) and VDDNB offset fuse; + USHORT usBootUpNBVoltage; //boot up NB voltage + UCHAR ucHtcTmpLmt; //bit [22:16] of D24F3x64 Hardware Thermal Control (HTC) Register, may not be needed, TBD + UCHAR ucTjOffset; //bit [28:22] of D24F3xE4 Thermtrip Status Register,may not be needed, TBD + ULONG ulReserved3[4]; //must be 0x0 for the reserved + ULONG ulDDISlot1Config; //see above ulDDISlot1Config definition + ULONG ulDDISlot2Config; + ULONG ulDDISlot3Config; + ULONG ulDDISlot4Config; + ULONG ulReserved4[4]; //must be 0x0 for the reserved + UCHAR ucMemoryType; //[3:0]=1:DDR1;=2:DDR2;=3:DDR3.[7:4] is reserved + UCHAR ucUMAChannelNumber; + USHORT usReserved; + ULONG ulReserved5[4]; //must be 0x0 for the reserved + ULONG ulCSR_M3_ARB_CNTL_DEFAULT[10];//arrays with values for CSR M3 arbiter for default + ULONG ulCSR_M3_ARB_CNTL_UVD[10]; //arrays with values for CSR M3 arbiter for UVD playback + ULONG ulCSR_M3_ARB_CNTL_FS3D[10];//arrays with values for CSR M3 arbiter for Full Screen 3D applications + ULONG ulReserved6[61]; //must be 0x0 +}ATOM_INTEGRATED_SYSTEM_INFO_V5; + +#define ATOM_CRT_INT_ENCODER1_INDEX 0x00000000 +#define ATOM_LCD_INT_ENCODER1_INDEX 0x00000001 +#define ATOM_TV_INT_ENCODER1_INDEX 0x00000002 +#define ATOM_DFP_INT_ENCODER1_INDEX 0x00000003 +#define ATOM_CRT_INT_ENCODER2_INDEX 0x00000004 +#define ATOM_LCD_EXT_ENCODER1_INDEX 0x00000005 +#define ATOM_TV_EXT_ENCODER1_INDEX 0x00000006 +#define ATOM_DFP_EXT_ENCODER1_INDEX 0x00000007 +#define ATOM_CV_INT_ENCODER1_INDEX 0x00000008 +#define ATOM_DFP_INT_ENCODER2_INDEX 0x00000009 +#define ATOM_CRT_EXT_ENCODER1_INDEX 0x0000000A +#define ATOM_CV_EXT_ENCODER1_INDEX 0x0000000B +#define ATOM_DFP_INT_ENCODER3_INDEX 0x0000000C +#define ATOM_DFP_INT_ENCODER4_INDEX 0x0000000D + +// define ASIC internal encoder id ( bit vector ), used for CRTC_SourceSelTable +#define ASIC_INT_DAC1_ENCODER_ID 0x00 +#define ASIC_INT_TV_ENCODER_ID 0x02 +#define ASIC_INT_DIG1_ENCODER_ID 0x03 +#define ASIC_INT_DAC2_ENCODER_ID 0x04 +#define ASIC_EXT_TV_ENCODER_ID 0x06 +#define ASIC_INT_DVO_ENCODER_ID 0x07 +#define ASIC_INT_DIG2_ENCODER_ID 0x09 +#define ASIC_EXT_DIG_ENCODER_ID 0x05 +#define ASIC_EXT_DIG2_ENCODER_ID 0x08 +#define ASIC_INT_DIG3_ENCODER_ID 0x0a +#define ASIC_INT_DIG4_ENCODER_ID 0x0b +#define ASIC_INT_DIG5_ENCODER_ID 0x0c +#define ASIC_INT_DIG6_ENCODER_ID 0x0d + +//define Encoder attribute +#define ATOM_ANALOG_ENCODER 0 +#define ATOM_DIGITAL_ENCODER 1 +#define ATOM_DP_ENCODER 2 + +#define ATOM_ENCODER_ENUM_MASK 0x70 +#define ATOM_ENCODER_ENUM_ID1 0x00 +#define ATOM_ENCODER_ENUM_ID2 0x10 +#define ATOM_ENCODER_ENUM_ID3 0x20 +#define ATOM_ENCODER_ENUM_ID4 0x30 +#define ATOM_ENCODER_ENUM_ID5 0x40 +#define ATOM_ENCODER_ENUM_ID6 0x50 + +#define ATOM_DEVICE_CRT1_INDEX 0x00000000 +#define ATOM_DEVICE_LCD1_INDEX 0x00000001 +#define ATOM_DEVICE_TV1_INDEX 0x00000002 +#define ATOM_DEVICE_DFP1_INDEX 0x00000003 +#define ATOM_DEVICE_CRT2_INDEX 0x00000004 +#define ATOM_DEVICE_LCD2_INDEX 0x00000005 +#define ATOM_DEVICE_DFP6_INDEX 0x00000006 +#define ATOM_DEVICE_DFP2_INDEX 0x00000007 +#define ATOM_DEVICE_CV_INDEX 0x00000008 +#define ATOM_DEVICE_DFP3_INDEX 0x00000009 +#define ATOM_DEVICE_DFP4_INDEX 0x0000000A +#define ATOM_DEVICE_DFP5_INDEX 0x0000000B + +#define ATOM_DEVICE_RESERVEDC_INDEX 0x0000000C +#define ATOM_DEVICE_RESERVEDD_INDEX 0x0000000D +#define ATOM_DEVICE_RESERVEDE_INDEX 0x0000000E +#define ATOM_DEVICE_RESERVEDF_INDEX 0x0000000F +#define ATOM_MAX_SUPPORTED_DEVICE_INFO (ATOM_DEVICE_DFP3_INDEX+1) +#define ATOM_MAX_SUPPORTED_DEVICE_INFO_2 ATOM_MAX_SUPPORTED_DEVICE_INFO +#define ATOM_MAX_SUPPORTED_DEVICE_INFO_3 (ATOM_DEVICE_DFP5_INDEX + 1 ) + +#define ATOM_MAX_SUPPORTED_DEVICE (ATOM_DEVICE_RESERVEDF_INDEX+1) + +#define ATOM_DEVICE_CRT1_SUPPORT (0x1L << ATOM_DEVICE_CRT1_INDEX ) +#define ATOM_DEVICE_LCD1_SUPPORT (0x1L << ATOM_DEVICE_LCD1_INDEX ) +#define ATOM_DEVICE_TV1_SUPPORT (0x1L << ATOM_DEVICE_TV1_INDEX ) +#define ATOM_DEVICE_DFP1_SUPPORT (0x1L << ATOM_DEVICE_DFP1_INDEX ) +#define ATOM_DEVICE_CRT2_SUPPORT (0x1L << ATOM_DEVICE_CRT2_INDEX ) +#define ATOM_DEVICE_LCD2_SUPPORT (0x1L << ATOM_DEVICE_LCD2_INDEX ) +#define ATOM_DEVICE_DFP6_SUPPORT (0x1L << ATOM_DEVICE_DFP6_INDEX ) +#define ATOM_DEVICE_DFP2_SUPPORT (0x1L << ATOM_DEVICE_DFP2_INDEX ) +#define ATOM_DEVICE_CV_SUPPORT (0x1L << ATOM_DEVICE_CV_INDEX ) +#define ATOM_DEVICE_DFP3_SUPPORT (0x1L << ATOM_DEVICE_DFP3_INDEX ) +#define ATOM_DEVICE_DFP4_SUPPORT (0x1L << ATOM_DEVICE_DFP4_INDEX ) +#define ATOM_DEVICE_DFP5_SUPPORT (0x1L << ATOM_DEVICE_DFP5_INDEX ) + +#define ATOM_DEVICE_CRT_SUPPORT (ATOM_DEVICE_CRT1_SUPPORT | ATOM_DEVICE_CRT2_SUPPORT) +#define ATOM_DEVICE_DFP_SUPPORT (ATOM_DEVICE_DFP1_SUPPORT | ATOM_DEVICE_DFP2_SUPPORT | ATOM_DEVICE_DFP3_SUPPORT | ATOM_DEVICE_DFP4_SUPPORT | ATOM_DEVICE_DFP5_SUPPORT | ATOM_DEVICE_DFP6_SUPPORT) +#define ATOM_DEVICE_TV_SUPPORT (ATOM_DEVICE_TV1_SUPPORT) +#define ATOM_DEVICE_LCD_SUPPORT (ATOM_DEVICE_LCD1_SUPPORT | ATOM_DEVICE_LCD2_SUPPORT) + +#define ATOM_DEVICE_CONNECTOR_TYPE_MASK 0x000000F0 +#define ATOM_DEVICE_CONNECTOR_TYPE_SHIFT 0x00000004 +#define ATOM_DEVICE_CONNECTOR_VGA 0x00000001 +#define ATOM_DEVICE_CONNECTOR_DVI_I 0x00000002 +#define ATOM_DEVICE_CONNECTOR_DVI_D 0x00000003 +#define ATOM_DEVICE_CONNECTOR_DVI_A 0x00000004 +#define ATOM_DEVICE_CONNECTOR_SVIDEO 0x00000005 +#define ATOM_DEVICE_CONNECTOR_COMPOSITE 0x00000006 +#define ATOM_DEVICE_CONNECTOR_LVDS 0x00000007 +#define ATOM_DEVICE_CONNECTOR_DIGI_LINK 0x00000008 +#define ATOM_DEVICE_CONNECTOR_SCART 0x00000009 +#define ATOM_DEVICE_CONNECTOR_HDMI_TYPE_A 0x0000000A +#define ATOM_DEVICE_CONNECTOR_HDMI_TYPE_B 0x0000000B +#define ATOM_DEVICE_CONNECTOR_CASE_1 0x0000000E +#define ATOM_DEVICE_CONNECTOR_DISPLAYPORT 0x0000000F + + +#define ATOM_DEVICE_DAC_INFO_MASK 0x0000000F +#define ATOM_DEVICE_DAC_INFO_SHIFT 0x00000000 +#define ATOM_DEVICE_DAC_INFO_NODAC 0x00000000 +#define ATOM_DEVICE_DAC_INFO_DACA 0x00000001 +#define ATOM_DEVICE_DAC_INFO_DACB 0x00000002 +#define ATOM_DEVICE_DAC_INFO_EXDAC 0x00000003 + +#define ATOM_DEVICE_I2C_ID_NOI2C 0x00000000 + +#define ATOM_DEVICE_I2C_LINEMUX_MASK 0x0000000F +#define ATOM_DEVICE_I2C_LINEMUX_SHIFT 0x00000000 + +#define ATOM_DEVICE_I2C_ID_MASK 0x00000070 +#define ATOM_DEVICE_I2C_ID_SHIFT 0x00000004 +#define ATOM_DEVICE_I2C_ID_IS_FOR_NON_MM_USE 0x00000001 +#define ATOM_DEVICE_I2C_ID_IS_FOR_MM_USE 0x00000002 +#define ATOM_DEVICE_I2C_ID_IS_FOR_SDVO_USE 0x00000003 //For IGP RS600 +#define ATOM_DEVICE_I2C_ID_IS_FOR_DAC_SCL 0x00000004 //For IGP RS690 + +#define ATOM_DEVICE_I2C_HARDWARE_CAP_MASK 0x00000080 +#define ATOM_DEVICE_I2C_HARDWARE_CAP_SHIFT 0x00000007 +#define ATOM_DEVICE_USES_SOFTWARE_ASSISTED_I2C 0x00000000 +#define ATOM_DEVICE_USES_HARDWARE_ASSISTED_I2C 0x00000001 + +// usDeviceSupport: +// Bits0 = 0 - no CRT1 support= 1- CRT1 is supported +// Bit 1 = 0 - no LCD1 support= 1- LCD1 is supported +// Bit 2 = 0 - no TV1 support= 1- TV1 is supported +// Bit 3 = 0 - no DFP1 support= 1- DFP1 is supported +// Bit 4 = 0 - no CRT2 support= 1- CRT2 is supported +// Bit 5 = 0 - no LCD2 support= 1- LCD2 is supported +// Bit 6 = 0 - no DFP6 support= 1- DFP6 is supported +// Bit 7 = 0 - no DFP2 support= 1- DFP2 is supported +// Bit 8 = 0 - no CV support= 1- CV is supported +// Bit 9 = 0 - no DFP3 support= 1- DFP3 is supported +// Bit 10 = 0 - no DFP4 support= 1- DFP4 is supported +// Bit 11 = 0 - no DFP5 support= 1- DFP5 is supported +// +// + +/****************************************************************************/ +/* Structure used in MclkSS_InfoTable */ +/****************************************************************************/ +// ucI2C_ConfigID +// [7:0] - I2C LINE Associate ID +// = 0 - no I2C +// [7] - HW_Cap = 1, [6:0]=HW assisted I2C ID(HW line selection) +// = 0, [6:0]=SW assisted I2C ID +// [6-4] - HW_ENGINE_ID = 1, HW engine for NON multimedia use +// = 2, HW engine for Multimedia use +// = 3-7 Reserved for future I2C engines +// [3-0] - I2C_LINE_MUX = A Mux number when it's HW assisted I2C or GPIO ID when it's SW I2C + +typedef struct _ATOM_I2C_ID_CONFIG +{ +#if ATOM_BIG_ENDIAN + UCHAR bfHW_Capable:1; + UCHAR bfHW_EngineID:3; + UCHAR bfI2C_LineMux:4; +#else + UCHAR bfI2C_LineMux:4; + UCHAR bfHW_EngineID:3; + UCHAR bfHW_Capable:1; +#endif +}ATOM_I2C_ID_CONFIG; + +typedef union _ATOM_I2C_ID_CONFIG_ACCESS +{ + ATOM_I2C_ID_CONFIG sbfAccess; + UCHAR ucAccess; +}ATOM_I2C_ID_CONFIG_ACCESS; + + +/****************************************************************************/ +// Structure used in GPIO_I2C_InfoTable +/****************************************************************************/ +typedef struct _ATOM_GPIO_I2C_ASSIGMENT +{ + USHORT usClkMaskRegisterIndex; + USHORT usClkEnRegisterIndex; + USHORT usClkY_RegisterIndex; + USHORT usClkA_RegisterIndex; + USHORT usDataMaskRegisterIndex; + USHORT usDataEnRegisterIndex; + USHORT usDataY_RegisterIndex; + USHORT usDataA_RegisterIndex; + ATOM_I2C_ID_CONFIG_ACCESS sucI2cId; + UCHAR ucClkMaskShift; + UCHAR ucClkEnShift; + UCHAR ucClkY_Shift; + UCHAR ucClkA_Shift; + UCHAR ucDataMaskShift; + UCHAR ucDataEnShift; + UCHAR ucDataY_Shift; + UCHAR ucDataA_Shift; + UCHAR ucReserved1; + UCHAR ucReserved2; +}ATOM_GPIO_I2C_ASSIGMENT; + +typedef struct _ATOM_GPIO_I2C_INFO +{ + ATOM_COMMON_TABLE_HEADER sHeader; + ATOM_GPIO_I2C_ASSIGMENT asGPIO_Info[ATOM_MAX_SUPPORTED_DEVICE]; +}ATOM_GPIO_I2C_INFO; + +/****************************************************************************/ +// Common Structure used in other structures +/****************************************************************************/ + +#ifndef _H2INC + +//Please don't add or expand this bitfield structure below, this one will retire soon.! +typedef struct _ATOM_MODE_MISC_INFO +{ +#if ATOM_BIG_ENDIAN + USHORT Reserved:6; + USHORT RGB888:1; + USHORT DoubleClock:1; + USHORT Interlace:1; + USHORT CompositeSync:1; + USHORT V_ReplicationBy2:1; + USHORT H_ReplicationBy2:1; + USHORT VerticalCutOff:1; + USHORT VSyncPolarity:1; //0=Active High, 1=Active Low + USHORT HSyncPolarity:1; //0=Active High, 1=Active Low + USHORT HorizontalCutOff:1; +#else + USHORT HorizontalCutOff:1; + USHORT HSyncPolarity:1; //0=Active High, 1=Active Low + USHORT VSyncPolarity:1; //0=Active High, 1=Active Low + USHORT VerticalCutOff:1; + USHORT H_ReplicationBy2:1; + USHORT V_ReplicationBy2:1; + USHORT CompositeSync:1; + USHORT Interlace:1; + USHORT DoubleClock:1; + USHORT RGB888:1; + USHORT Reserved:6; +#endif +}ATOM_MODE_MISC_INFO; + +typedef union _ATOM_MODE_MISC_INFO_ACCESS +{ + ATOM_MODE_MISC_INFO sbfAccess; + USHORT usAccess; +}ATOM_MODE_MISC_INFO_ACCESS; + +#else + +typedef union _ATOM_MODE_MISC_INFO_ACCESS +{ + USHORT usAccess; +}ATOM_MODE_MISC_INFO_ACCESS; + +#endif + +// usModeMiscInfo- +#define ATOM_H_CUTOFF 0x01 +#define ATOM_HSYNC_POLARITY 0x02 //0=Active High, 1=Active Low +#define ATOM_VSYNC_POLARITY 0x04 //0=Active High, 1=Active Low +#define ATOM_V_CUTOFF 0x08 +#define ATOM_H_REPLICATIONBY2 0x10 +#define ATOM_V_REPLICATIONBY2 0x20 +#define ATOM_COMPOSITESYNC 0x40 +#define ATOM_INTERLACE 0x80 +#define ATOM_DOUBLE_CLOCK_MODE 0x100 +#define ATOM_RGB888_MODE 0x200 + +//usRefreshRate- +#define ATOM_REFRESH_43 43 +#define ATOM_REFRESH_47 47 +#define ATOM_REFRESH_56 56 +#define ATOM_REFRESH_60 60 +#define ATOM_REFRESH_65 65 +#define ATOM_REFRESH_70 70 +#define ATOM_REFRESH_72 72 +#define ATOM_REFRESH_75 75 +#define ATOM_REFRESH_85 85 + +// ATOM_MODE_TIMING data are exactly the same as VESA timing data. +// Translation from EDID to ATOM_MODE_TIMING, use the following formula. +// +// VESA_HTOTAL = VESA_ACTIVE + 2* VESA_BORDER + VESA_BLANK +// = EDID_HA + EDID_HBL +// VESA_HDISP = VESA_ACTIVE = EDID_HA +// VESA_HSYNC_START = VESA_ACTIVE + VESA_BORDER + VESA_FRONT_PORCH +// = EDID_HA + EDID_HSO +// VESA_HSYNC_WIDTH = VESA_HSYNC_TIME = EDID_HSPW +// VESA_BORDER = EDID_BORDER + +/****************************************************************************/ +// Structure used in SetCRTC_UsingDTDTimingTable +/****************************************************************************/ +typedef struct _SET_CRTC_USING_DTD_TIMING_PARAMETERS +{ + USHORT usH_Size; + USHORT usH_Blanking_Time; + USHORT usV_Size; + USHORT usV_Blanking_Time; + USHORT usH_SyncOffset; + USHORT usH_SyncWidth; + USHORT usV_SyncOffset; + USHORT usV_SyncWidth; + ATOM_MODE_MISC_INFO_ACCESS susModeMiscInfo; + UCHAR ucH_Border; // From DFP EDID + UCHAR ucV_Border; + UCHAR ucCRTC; // ATOM_CRTC1 or ATOM_CRTC2 + UCHAR ucPadding[3]; +}SET_CRTC_USING_DTD_TIMING_PARAMETERS; + +/****************************************************************************/ +// Structure used in SetCRTC_TimingTable +/****************************************************************************/ +typedef struct _SET_CRTC_TIMING_PARAMETERS +{ + USHORT usH_Total; // horizontal total + USHORT usH_Disp; // horizontal display + USHORT usH_SyncStart; // horozontal Sync start + USHORT usH_SyncWidth; // horizontal Sync width + USHORT usV_Total; // vertical total + USHORT usV_Disp; // vertical display + USHORT usV_SyncStart; // vertical Sync start + USHORT usV_SyncWidth; // vertical Sync width + ATOM_MODE_MISC_INFO_ACCESS susModeMiscInfo; + UCHAR ucCRTC; // ATOM_CRTC1 or ATOM_CRTC2 + UCHAR ucOverscanRight; // right + UCHAR ucOverscanLeft; // left + UCHAR ucOverscanBottom; // bottom + UCHAR ucOverscanTop; // top + UCHAR ucReserved; +}SET_CRTC_TIMING_PARAMETERS; +#define SET_CRTC_TIMING_PARAMETERS_PS_ALLOCATION SET_CRTC_TIMING_PARAMETERS + +/****************************************************************************/ +// Structure used in StandardVESA_TimingTable +// AnalogTV_InfoTable +// ComponentVideoInfoTable +/****************************************************************************/ +typedef struct _ATOM_MODE_TIMING +{ + USHORT usCRTC_H_Total; + USHORT usCRTC_H_Disp; + USHORT usCRTC_H_SyncStart; + USHORT usCRTC_H_SyncWidth; + USHORT usCRTC_V_Total; + USHORT usCRTC_V_Disp; + USHORT usCRTC_V_SyncStart; + USHORT usCRTC_V_SyncWidth; + USHORT usPixelClock; //in 10Khz unit + ATOM_MODE_MISC_INFO_ACCESS susModeMiscInfo; + USHORT usCRTC_OverscanRight; + USHORT usCRTC_OverscanLeft; + USHORT usCRTC_OverscanBottom; + USHORT usCRTC_OverscanTop; + USHORT usReserve; + UCHAR ucInternalModeNumber; + UCHAR ucRefreshRate; +}ATOM_MODE_TIMING; + +typedef struct _ATOM_DTD_FORMAT +{ + USHORT usPixClk; + USHORT usHActive; + USHORT usHBlanking_Time; + USHORT usVActive; + USHORT usVBlanking_Time; + USHORT usHSyncOffset; + USHORT usHSyncWidth; + USHORT usVSyncOffset; + USHORT usVSyncWidth; + USHORT usImageHSize; + USHORT usImageVSize; + UCHAR ucHBorder; + UCHAR ucVBorder; + ATOM_MODE_MISC_INFO_ACCESS susModeMiscInfo; + UCHAR ucInternalModeNumber; + UCHAR ucRefreshRate; +}ATOM_DTD_FORMAT; + +/****************************************************************************/ +// Structure used in LVDS_InfoTable +// * Need a document to describe this table +/****************************************************************************/ +#define SUPPORTED_LCD_REFRESHRATE_30Hz 0x0004 +#define SUPPORTED_LCD_REFRESHRATE_40Hz 0x0008 +#define SUPPORTED_LCD_REFRESHRATE_50Hz 0x0010 +#define SUPPORTED_LCD_REFRESHRATE_60Hz 0x0020 + +//ucTableFormatRevision=1 +//ucTableContentRevision=1 +typedef struct _ATOM_LVDS_INFO +{ + ATOM_COMMON_TABLE_HEADER sHeader; + ATOM_DTD_FORMAT sLCDTiming; + USHORT usModePatchTableOffset; + USHORT usSupportedRefreshRate; //Refer to panel info table in ATOMBIOS extension Spec. + USHORT usOffDelayInMs; + UCHAR ucPowerSequenceDigOntoDEin10Ms; + UCHAR ucPowerSequenceDEtoBLOnin10Ms; + UCHAR ucLVDS_Misc; // Bit0:{=0:single, =1:dual},Bit1 {=0:666RGB, =1:888RGB},Bit2:3:{Grey level} + // Bit4:{=0:LDI format for RGB888, =1 FPDI format for RGB888} + // Bit5:{=0:Spatial Dithering disabled;1 Spatial Dithering enabled} + // Bit6:{=0:Temporal Dithering disabled;1 Temporal Dithering enabled} + UCHAR ucPanelDefaultRefreshRate; + UCHAR ucPanelIdentification; + UCHAR ucSS_Id; +}ATOM_LVDS_INFO; + +//ucTableFormatRevision=1 +//ucTableContentRevision=2 +typedef struct _ATOM_LVDS_INFO_V12 +{ + ATOM_COMMON_TABLE_HEADER sHeader; + ATOM_DTD_FORMAT sLCDTiming; + USHORT usExtInfoTableOffset; + USHORT usSupportedRefreshRate; //Refer to panel info table in ATOMBIOS extension Spec. + USHORT usOffDelayInMs; + UCHAR ucPowerSequenceDigOntoDEin10Ms; + UCHAR ucPowerSequenceDEtoBLOnin10Ms; + UCHAR ucLVDS_Misc; // Bit0:{=0:single, =1:dual},Bit1 {=0:666RGB, =1:888RGB},Bit2:3:{Grey level} + // Bit4:{=0:LDI format for RGB888, =1 FPDI format for RGB888} + // Bit5:{=0:Spatial Dithering disabled;1 Spatial Dithering enabled} + // Bit6:{=0:Temporal Dithering disabled;1 Temporal Dithering enabled} + UCHAR ucPanelDefaultRefreshRate; + UCHAR ucPanelIdentification; + UCHAR ucSS_Id; + USHORT usLCDVenderID; + USHORT usLCDProductID; + UCHAR ucLCDPanel_SpecialHandlingCap; + UCHAR ucPanelInfoSize; // start from ATOM_DTD_FORMAT to end of panel info, include ExtInfoTable + UCHAR ucReserved[2]; +}ATOM_LVDS_INFO_V12; + +//Definitions for ucLCDPanel_SpecialHandlingCap: + +//Once DAL sees this CAP is set, it will read EDID from LCD on its own instead of using sLCDTiming in ATOM_LVDS_INFO_V12. +//Other entries in ATOM_LVDS_INFO_V12 are still valid/useful to DAL +#define LCDPANEL_CAP_READ_EDID 0x1 + +//If a design supports DRR (dynamic refresh rate) on internal panels (LVDS or EDP), this cap is set in ucLCDPanel_SpecialHandlingCap together +//with multiple supported refresh rates@usSupportedRefreshRate. This cap should not be set when only slow refresh rate is supported (static +//refresh rate switch by SW. This is only valid from ATOM_LVDS_INFO_V12 +#define LCDPANEL_CAP_DRR_SUPPORTED 0x2 + +//Use this cap bit for a quick reference whether an embadded panel (LCD1 ) is LVDS or eDP. +#define LCDPANEL_CAP_eDP 0x4 + + +//Color Bit Depth definition in EDID V1.4 @BYTE 14h +//Bit 6 5 4 + // 0 0 0 - Color bit depth is undefined + // 0 0 1 - 6 Bits per Primary Color + // 0 1 0 - 8 Bits per Primary Color + // 0 1 1 - 10 Bits per Primary Color + // 1 0 0 - 12 Bits per Primary Color + // 1 0 1 - 14 Bits per Primary Color + // 1 1 0 - 16 Bits per Primary Color + // 1 1 1 - Reserved + +#define PANEL_COLOR_BIT_DEPTH_MASK 0x70 + +// Bit7:{=0:Random Dithering disabled;1 Random Dithering enabled} +#define PANEL_RANDOM_DITHER 0x80 +#define PANEL_RANDOM_DITHER_MASK 0x80 + +#define ATOM_LVDS_INFO_LAST ATOM_LVDS_INFO_V12 // no need to change this + +/****************************************************************************/ +// Structures used by LCD_InfoTable V1.3 Note: previous version was called ATOM_LVDS_INFO_V12 +// ASIC Families: NI +// ucTableFormatRevision=1 +// ucTableContentRevision=3 +/****************************************************************************/ +typedef struct _ATOM_LCD_INFO_V13 +{ + ATOM_COMMON_TABLE_HEADER sHeader; + ATOM_DTD_FORMAT sLCDTiming; + USHORT usExtInfoTableOffset; + USHORT usSupportedRefreshRate; //Refer to panel info table in ATOMBIOS extension Spec. + ULONG ulReserved0; + UCHAR ucLCD_Misc; // Reorganized in V13 + // Bit0: {=0:single, =1:dual}, + // Bit1: {=0:LDI format for RGB888, =1 FPDI format for RGB888} // was {=0:666RGB, =1:888RGB}, + // Bit3:2: {Grey level} + // Bit6:4 Color Bit Depth definition (see below definition in EDID V1.4 @BYTE 14h) + // Bit7 Reserved. was for ATOM_PANEL_MISC_API_ENABLED, still need it? + UCHAR ucPanelDefaultRefreshRate; + UCHAR ucPanelIdentification; + UCHAR ucSS_Id; + USHORT usLCDVenderID; + USHORT usLCDProductID; + UCHAR ucLCDPanel_SpecialHandlingCap; // Reorganized in V13 + // Bit0: Once DAL sees this CAP is set, it will read EDID from LCD on its own + // Bit1: See LCDPANEL_CAP_DRR_SUPPORTED + // Bit2: a quick reference whether an embadded panel (LCD1 ) is LVDS (0) or eDP (1) + // Bit7-3: Reserved + UCHAR ucPanelInfoSize; // start from ATOM_DTD_FORMAT to end of panel info, include ExtInfoTable + USHORT usBacklightPWM; // Backlight PWM in Hz. New in _V13 + + UCHAR ucPowerSequenceDIGONtoDE_in4Ms; + UCHAR ucPowerSequenceDEtoVARY_BL_in4Ms; + UCHAR ucPowerSequenceDEtoDIGON_in4Ms; + UCHAR ucPowerSequenceVARY_BLtoDE_in4Ms; + + UCHAR ucOffDelay_in4Ms; + UCHAR ucPowerSequenceVARY_BLtoBLON_in4Ms; + UCHAR ucPowerSequenceBLONtoVARY_BL_in4Ms; + UCHAR ucReserved1; + + ULONG ulReserved[4]; +}ATOM_LCD_INFO_V13; + +#define ATOM_LCD_INFO_LAST ATOM_LCD_INFO_V13 + +//Definitions for ucLCD_Misc +#define ATOM_PANEL_MISC_V13_DUAL 0x00000001 +#define ATOM_PANEL_MISC_V13_FPDI 0x00000002 +#define ATOM_PANEL_MISC_V13_GREY_LEVEL 0x0000000C +#define ATOM_PANEL_MISC_V13_GREY_LEVEL_SHIFT 2 +#define ATOM_PANEL_MISC_V13_COLOR_BIT_DEPTH_MASK 0x70 +#define ATOM_PANEL_MISC_V13_6BIT_PER_COLOR 0x10 +#define ATOM_PANEL_MISC_V13_8BIT_PER_COLOR 0x20 + +//Color Bit Depth definition in EDID V1.4 @BYTE 14h +//Bit 6 5 4 + // 0 0 0 - Color bit depth is undefined + // 0 0 1 - 6 Bits per Primary Color + // 0 1 0 - 8 Bits per Primary Color + // 0 1 1 - 10 Bits per Primary Color + // 1 0 0 - 12 Bits per Primary Color + // 1 0 1 - 14 Bits per Primary Color + // 1 1 0 - 16 Bits per Primary Color + // 1 1 1 - Reserved + +//Definitions for ucLCDPanel_SpecialHandlingCap: + +//Once DAL sees this CAP is set, it will read EDID from LCD on its own instead of using sLCDTiming in ATOM_LVDS_INFO_V12. +//Other entries in ATOM_LVDS_INFO_V12 are still valid/useful to DAL +#define LCDPANEL_CAP_V13_READ_EDID 0x1 // = LCDPANEL_CAP_READ_EDID no change comparing to previous version + +//If a design supports DRR (dynamic refresh rate) on internal panels (LVDS or EDP), this cap is set in ucLCDPanel_SpecialHandlingCap together +//with multiple supported refresh rates@usSupportedRefreshRate. This cap should not be set when only slow refresh rate is supported (static +//refresh rate switch by SW. This is only valid from ATOM_LVDS_INFO_V12 +#define LCDPANEL_CAP_V13_DRR_SUPPORTED 0x2 // = LCDPANEL_CAP_DRR_SUPPORTED no change comparing to previous version + +//Use this cap bit for a quick reference whether an embadded panel (LCD1 ) is LVDS or eDP. +#define LCDPANEL_CAP_V13_eDP 0x4 // = LCDPANEL_CAP_eDP no change comparing to previous version + +typedef struct _ATOM_PATCH_RECORD_MODE +{ + UCHAR ucRecordType; + USHORT usHDisp; + USHORT usVDisp; +}ATOM_PATCH_RECORD_MODE; + +typedef struct _ATOM_LCD_RTS_RECORD +{ + UCHAR ucRecordType; + UCHAR ucRTSValue; +}ATOM_LCD_RTS_RECORD; + +//!! If the record below exits, it shoud always be the first record for easy use in command table!!! +// The record below is only used when LVDS_Info is present. From ATOM_LVDS_INFO_V12, use ucLCDPanel_SpecialHandlingCap instead. +typedef struct _ATOM_LCD_MODE_CONTROL_CAP +{ + UCHAR ucRecordType; + USHORT usLCDCap; +}ATOM_LCD_MODE_CONTROL_CAP; + +#define LCD_MODE_CAP_BL_OFF 1 +#define LCD_MODE_CAP_CRTC_OFF 2 +#define LCD_MODE_CAP_PANEL_OFF 4 + +typedef struct _ATOM_FAKE_EDID_PATCH_RECORD +{ + UCHAR ucRecordType; + UCHAR ucFakeEDIDLength; + UCHAR ucFakeEDIDString[1]; // This actually has ucFakeEdidLength elements. +} ATOM_FAKE_EDID_PATCH_RECORD; + +typedef struct _ATOM_PANEL_RESOLUTION_PATCH_RECORD +{ + UCHAR ucRecordType; + USHORT usHSize; + USHORT usVSize; +}ATOM_PANEL_RESOLUTION_PATCH_RECORD; + +#define LCD_MODE_PATCH_RECORD_MODE_TYPE 1 +#define LCD_RTS_RECORD_TYPE 2 +#define LCD_CAP_RECORD_TYPE 3 +#define LCD_FAKE_EDID_PATCH_RECORD_TYPE 4 +#define LCD_PANEL_RESOLUTION_RECORD_TYPE 5 +#define ATOM_RECORD_END_TYPE 0xFF + +/****************************Spread Spectrum Info Table Definitions **********************/ + +//ucTableFormatRevision=1 +//ucTableContentRevision=2 +typedef struct _ATOM_SPREAD_SPECTRUM_ASSIGNMENT +{ + USHORT usSpreadSpectrumPercentage; + UCHAR ucSpreadSpectrumType; //Bit1=0 Down Spread,=1 Center Spread. Bit1=1 Ext. =0 Int. Bit2=1: PCIE REFCLK SS =0 iternal PPLL SS Others:TBD + UCHAR ucSS_Step; + UCHAR ucSS_Delay; + UCHAR ucSS_Id; + UCHAR ucRecommendedRef_Div; + UCHAR ucSS_Range; //it was reserved for V11 +}ATOM_SPREAD_SPECTRUM_ASSIGNMENT; + +#define ATOM_MAX_SS_ENTRY 16 +#define ATOM_DP_SS_ID1 0x0f1 // SS ID for internal DP stream at 2.7Ghz. if ATOM_DP_SS_ID2 does not exist in SS_InfoTable, it is used for internal DP stream at 1.62Ghz as well. +#define ATOM_DP_SS_ID2 0x0f2 // SS ID for internal DP stream at 1.62Ghz, if it exists in SS_InfoTable. +#define ATOM_LVLINK_2700MHz_SS_ID 0x0f3 // SS ID for LV link translator chip at 2.7Ghz +#define ATOM_LVLINK_1620MHz_SS_ID 0x0f4 // SS ID for LV link translator chip at 1.62Ghz + + +#define ATOM_SS_DOWN_SPREAD_MODE_MASK 0x00000000 +#define ATOM_SS_DOWN_SPREAD_MODE 0x00000000 +#define ATOM_SS_CENTRE_SPREAD_MODE_MASK 0x00000001 +#define ATOM_SS_CENTRE_SPREAD_MODE 0x00000001 +#define ATOM_INTERNAL_SS_MASK 0x00000000 +#define ATOM_EXTERNAL_SS_MASK 0x00000002 +#define EXEC_SS_STEP_SIZE_SHIFT 2 +#define EXEC_SS_DELAY_SHIFT 4 +#define ACTIVEDATA_TO_BLON_DELAY_SHIFT 4 + +typedef struct _ATOM_SPREAD_SPECTRUM_INFO +{ + ATOM_COMMON_TABLE_HEADER sHeader; + ATOM_SPREAD_SPECTRUM_ASSIGNMENT asSS_Info[ATOM_MAX_SS_ENTRY]; +}ATOM_SPREAD_SPECTRUM_INFO; + +/****************************************************************************/ +// Structure used in AnalogTV_InfoTable (Top level) +/****************************************************************************/ +//ucTVBootUpDefaultStd definition: + +//ATOM_TV_NTSC 1 +//ATOM_TV_NTSCJ 2 +//ATOM_TV_PAL 3 +//ATOM_TV_PALM 4 +//ATOM_TV_PALCN 5 +//ATOM_TV_PALN 6 +//ATOM_TV_PAL60 7 +//ATOM_TV_SECAM 8 + +//ucTVSupportedStd definition: +#define NTSC_SUPPORT 0x1 +#define NTSCJ_SUPPORT 0x2 + +#define PAL_SUPPORT 0x4 +#define PALM_SUPPORT 0x8 +#define PALCN_SUPPORT 0x10 +#define PALN_SUPPORT 0x20 +#define PAL60_SUPPORT 0x40 +#define SECAM_SUPPORT 0x80 + +#define MAX_SUPPORTED_TV_TIMING 2 + +typedef struct _ATOM_ANALOG_TV_INFO +{ + ATOM_COMMON_TABLE_HEADER sHeader; + UCHAR ucTV_SupportedStandard; + UCHAR ucTV_BootUpDefaultStandard; + UCHAR ucExt_TV_ASIC_ID; + UCHAR ucExt_TV_ASIC_SlaveAddr; + /*ATOM_DTD_FORMAT aModeTimings[MAX_SUPPORTED_TV_TIMING];*/ + ATOM_MODE_TIMING aModeTimings[MAX_SUPPORTED_TV_TIMING]; +}ATOM_ANALOG_TV_INFO; + +#define MAX_SUPPORTED_TV_TIMING_V1_2 3 + +typedef struct _ATOM_ANALOG_TV_INFO_V1_2 +{ + ATOM_COMMON_TABLE_HEADER sHeader; + UCHAR ucTV_SupportedStandard; + UCHAR ucTV_BootUpDefaultStandard; + UCHAR ucExt_TV_ASIC_ID; + UCHAR ucExt_TV_ASIC_SlaveAddr; + ATOM_DTD_FORMAT aModeTimings[MAX_SUPPORTED_TV_TIMING_V1_2]; +}ATOM_ANALOG_TV_INFO_V1_2; + +typedef struct _ATOM_DPCD_INFO +{ + UCHAR ucRevisionNumber; //10h : Revision 1.0; 11h : Revision 1.1 + UCHAR ucMaxLinkRate; //06h : 1.62Gbps per lane; 0Ah = 2.7Gbps per lane + UCHAR ucMaxLane; //Bits 4:0 = MAX_LANE_COUNT (1/2/4). Bit 7 = ENHANCED_FRAME_CAP + UCHAR ucMaxDownSpread; //Bit0 = 0: No Down spread; Bit0 = 1: 0.5% (Subject to change according to DP spec) +}ATOM_DPCD_INFO; + +#define ATOM_DPCD_MAX_LANE_MASK 0x1F + +/**************************************************************************/ +// VRAM usage and their defintions + +// One chunk of VRAM used by Bios are for HWICON surfaces,EDID data. +// Current Mode timing and Dail Timing and/or STD timing data EACH device. They can be broken down as below. +// All the addresses below are the offsets from the frame buffer start.They all MUST be Dword aligned! +// To driver: The physical address of this memory portion=mmFB_START(4K aligned)+ATOMBIOS_VRAM_USAGE_START_ADDR+ATOM_x_ADDR +// To Bios: ATOMBIOS_VRAM_USAGE_START_ADDR+ATOM_x_ADDR->MM_INDEX + +#ifndef VESA_MEMORY_IN_64K_BLOCK +#define VESA_MEMORY_IN_64K_BLOCK 0x100 //256*64K=16Mb (Max. VESA memory is 16Mb!) +#endif + +#define ATOM_EDID_RAW_DATASIZE 256 //In Bytes +#define ATOM_HWICON_SURFACE_SIZE 4096 //In Bytes +#define ATOM_HWICON_INFOTABLE_SIZE 32 +#define MAX_DTD_MODE_IN_VRAM 6 +#define ATOM_DTD_MODE_SUPPORT_TBL_SIZE (MAX_DTD_MODE_IN_VRAM*28) //28= (SIZEOF ATOM_DTD_FORMAT) +#define ATOM_STD_MODE_SUPPORT_TBL_SIZE 32*8 //32 is a predefined number,8= (SIZEOF ATOM_STD_FORMAT) +//20 bytes for Encoder Type and DPCD in STD EDID area +#define DFP_ENCODER_TYPE_OFFSET (ATOM_EDID_RAW_DATASIZE + ATOM_DTD_MODE_SUPPORT_TBL_SIZE + ATOM_STD_MODE_SUPPORT_TBL_SIZE - 20) +#define ATOM_DP_DPCD_OFFSET (DFP_ENCODER_TYPE_OFFSET + 4 ) + +#define ATOM_HWICON1_SURFACE_ADDR 0 +#define ATOM_HWICON2_SURFACE_ADDR (ATOM_HWICON1_SURFACE_ADDR + ATOM_HWICON_SURFACE_SIZE) +#define ATOM_HWICON_INFOTABLE_ADDR (ATOM_HWICON2_SURFACE_ADDR + ATOM_HWICON_SURFACE_SIZE) +#define ATOM_CRT1_EDID_ADDR (ATOM_HWICON_INFOTABLE_ADDR + ATOM_HWICON_INFOTABLE_SIZE) +#define ATOM_CRT1_DTD_MODE_TBL_ADDR (ATOM_CRT1_EDID_ADDR + ATOM_EDID_RAW_DATASIZE) +#define ATOM_CRT1_STD_MODE_TBL_ADDR (ATOM_CRT1_DTD_MODE_TBL_ADDR + ATOM_DTD_MODE_SUPPORT_TBL_SIZE) + +#define ATOM_LCD1_EDID_ADDR (ATOM_CRT1_STD_MODE_TBL_ADDR + ATOM_STD_MODE_SUPPORT_TBL_SIZE) +#define ATOM_LCD1_DTD_MODE_TBL_ADDR (ATOM_LCD1_EDID_ADDR + ATOM_EDID_RAW_DATASIZE) +#define ATOM_LCD1_STD_MODE_TBL_ADDR (ATOM_LCD1_DTD_MODE_TBL_ADDR + ATOM_DTD_MODE_SUPPORT_TBL_SIZE) + +#define ATOM_TV1_DTD_MODE_TBL_ADDR (ATOM_LCD1_STD_MODE_TBL_ADDR + ATOM_STD_MODE_SUPPORT_TBL_SIZE) + +#define ATOM_DFP1_EDID_ADDR (ATOM_TV1_DTD_MODE_TBL_ADDR + ATOM_DTD_MODE_SUPPORT_TBL_SIZE) +#define ATOM_DFP1_DTD_MODE_TBL_ADDR (ATOM_DFP1_EDID_ADDR + ATOM_EDID_RAW_DATASIZE) +#define ATOM_DFP1_STD_MODE_TBL_ADDR (ATOM_DFP1_DTD_MODE_TBL_ADDR + ATOM_DTD_MODE_SUPPORT_TBL_SIZE) + +#define ATOM_CRT2_EDID_ADDR (ATOM_DFP1_STD_MODE_TBL_ADDR + ATOM_STD_MODE_SUPPORT_TBL_SIZE) +#define ATOM_CRT2_DTD_MODE_TBL_ADDR (ATOM_CRT2_EDID_ADDR + ATOM_EDID_RAW_DATASIZE) +#define ATOM_CRT2_STD_MODE_TBL_ADDR (ATOM_CRT2_DTD_MODE_TBL_ADDR + ATOM_DTD_MODE_SUPPORT_TBL_SIZE) + +#define ATOM_LCD2_EDID_ADDR (ATOM_CRT2_STD_MODE_TBL_ADDR + ATOM_STD_MODE_SUPPORT_TBL_SIZE) +#define ATOM_LCD2_DTD_MODE_TBL_ADDR (ATOM_LCD2_EDID_ADDR + ATOM_EDID_RAW_DATASIZE) +#define ATOM_LCD2_STD_MODE_TBL_ADDR (ATOM_LCD2_DTD_MODE_TBL_ADDR + ATOM_DTD_MODE_SUPPORT_TBL_SIZE) + +#define ATOM_DFP6_EDID_ADDR (ATOM_LCD2_STD_MODE_TBL_ADDR + ATOM_STD_MODE_SUPPORT_TBL_SIZE) +#define ATOM_DFP6_DTD_MODE_TBL_ADDR (ATOM_DFP6_EDID_ADDR + ATOM_EDID_RAW_DATASIZE) +#define ATOM_DFP6_STD_MODE_TBL_ADDR (ATOM_DFP6_DTD_MODE_TBL_ADDR + ATOM_DTD_MODE_SUPPORT_TBL_SIZE) + +#define ATOM_DFP2_EDID_ADDR (ATOM_DFP6_STD_MODE_TBL_ADDR + ATOM_STD_MODE_SUPPORT_TBL_SIZE) +#define ATOM_DFP2_DTD_MODE_TBL_ADDR (ATOM_DFP2_EDID_ADDR + ATOM_EDID_RAW_DATASIZE) +#define ATOM_DFP2_STD_MODE_TBL_ADDR (ATOM_DFP2_DTD_MODE_TBL_ADDR + ATOM_DTD_MODE_SUPPORT_TBL_SIZE) + +#define ATOM_CV_EDID_ADDR (ATOM_DFP2_STD_MODE_TBL_ADDR + ATOM_STD_MODE_SUPPORT_TBL_SIZE) +#define ATOM_CV_DTD_MODE_TBL_ADDR (ATOM_CV_EDID_ADDR + ATOM_EDID_RAW_DATASIZE) +#define ATOM_CV_STD_MODE_TBL_ADDR (ATOM_CV_DTD_MODE_TBL_ADDR + ATOM_DTD_MODE_SUPPORT_TBL_SIZE) + +#define ATOM_DFP3_EDID_ADDR (ATOM_CV_STD_MODE_TBL_ADDR + ATOM_STD_MODE_SUPPORT_TBL_SIZE) +#define ATOM_DFP3_DTD_MODE_TBL_ADDR (ATOM_DFP3_EDID_ADDR + ATOM_EDID_RAW_DATASIZE) +#define ATOM_DFP3_STD_MODE_TBL_ADDR (ATOM_DFP3_DTD_MODE_TBL_ADDR + ATOM_DTD_MODE_SUPPORT_TBL_SIZE) + +#define ATOM_DFP4_EDID_ADDR (ATOM_DFP3_STD_MODE_TBL_ADDR + ATOM_STD_MODE_SUPPORT_TBL_SIZE) +#define ATOM_DFP4_DTD_MODE_TBL_ADDR (ATOM_DFP4_EDID_ADDR + ATOM_EDID_RAW_DATASIZE) +#define ATOM_DFP4_STD_MODE_TBL_ADDR (ATOM_DFP4_DTD_MODE_TBL_ADDR + ATOM_DTD_MODE_SUPPORT_TBL_SIZE) + +#define ATOM_DFP5_EDID_ADDR (ATOM_DFP4_STD_MODE_TBL_ADDR + ATOM_STD_MODE_SUPPORT_TBL_SIZE) +#define ATOM_DFP5_DTD_MODE_TBL_ADDR (ATOM_DFP5_EDID_ADDR + ATOM_EDID_RAW_DATASIZE) +#define ATOM_DFP5_STD_MODE_TBL_ADDR (ATOM_DFP5_DTD_MODE_TBL_ADDR + ATOM_DTD_MODE_SUPPORT_TBL_SIZE) + +#define ATOM_DP_TRAINING_TBL_ADDR (ATOM_DFP5_STD_MODE_TBL_ADDR + ATOM_STD_MODE_SUPPORT_TBL_SIZE) + +#define ATOM_STACK_STORAGE_START (ATOM_DP_TRAINING_TBL_ADDR + 1024) +#define ATOM_STACK_STORAGE_END ATOM_STACK_STORAGE_START + 512 + +//The size below is in Kb! +#define ATOM_VRAM_RESERVE_SIZE ((((ATOM_STACK_STORAGE_END - ATOM_HWICON1_SURFACE_ADDR)>>10)+4)&0xFFFC) + +#define ATOM_VRAM_RESERVE_V2_SIZE 32 + +#define ATOM_VRAM_OPERATION_FLAGS_MASK 0xC0000000L +#define ATOM_VRAM_OPERATION_FLAGS_SHIFT 30 +#define ATOM_VRAM_BLOCK_NEEDS_NO_RESERVATION 0x1 +#define ATOM_VRAM_BLOCK_NEEDS_RESERVATION 0x0 + +/***********************************************************************************/ +// Structure used in VRAM_UsageByFirmwareTable +// Note1: This table is filled by SetBiosReservationStartInFB in CoreCommSubs.asm +// at running time. +// note2: From RV770, the memory is more than 32bit addressable, so we will change +// ucTableFormatRevision=1,ucTableContentRevision=4, the strcuture remains +// exactly same as 1.1 and 1.2 (1.3 is never in use), but ulStartAddrUsedByFirmware +// (in offset to start of memory address) is KB aligned instead of byte aligend. +/***********************************************************************************/ +// Note3: +/* If we change usReserved to "usFBUsedbyDrvInKB", then to VBIOS this usFBUsedbyDrvInKB is a predefined, unchanged constant across VGA or non VGA adapter, +for CAIL, The size of FB access area is known, only thing missing is the Offset of FB Access area, so we can have: + +If (ulStartAddrUsedByFirmware!=0) +FBAccessAreaOffset= ulStartAddrUsedByFirmware - usFBUsedbyDrvInKB; +Reserved area has been claimed by VBIOS including this FB access area; CAIL doesn't need to reserve any extra area for this purpose +else //Non VGA case + if (FB_Size<=2Gb) + FBAccessAreaOffset= FB_Size - usFBUsedbyDrvInKB; + else + FBAccessAreaOffset= Aper_Size - usFBUsedbyDrvInKB + +CAIL needs to claim an reserved area defined by FBAccessAreaOffset and usFBUsedbyDrvInKB in non VGA case.*/ + +#define ATOM_MAX_FIRMWARE_VRAM_USAGE_INFO 1 + +typedef struct _ATOM_FIRMWARE_VRAM_RESERVE_INFO +{ + ULONG ulStartAddrUsedByFirmware; + USHORT usFirmwareUseInKb; + USHORT usReserved; +}ATOM_FIRMWARE_VRAM_RESERVE_INFO; + +typedef struct _ATOM_VRAM_USAGE_BY_FIRMWARE +{ + ATOM_COMMON_TABLE_HEADER sHeader; + ATOM_FIRMWARE_VRAM_RESERVE_INFO asFirmwareVramReserveInfo[ATOM_MAX_FIRMWARE_VRAM_USAGE_INFO]; +}ATOM_VRAM_USAGE_BY_FIRMWARE; + +// change verion to 1.5, when allow driver to allocate the vram area for command table access. +typedef struct _ATOM_FIRMWARE_VRAM_RESERVE_INFO_V1_5 +{ + ULONG ulStartAddrUsedByFirmware; + USHORT usFirmwareUseInKb; + USHORT usFBUsedByDrvInKb; +}ATOM_FIRMWARE_VRAM_RESERVE_INFO_V1_5; + +typedef struct _ATOM_VRAM_USAGE_BY_FIRMWARE_V1_5 +{ + ATOM_COMMON_TABLE_HEADER sHeader; + ATOM_FIRMWARE_VRAM_RESERVE_INFO_V1_5 asFirmwareVramReserveInfo[ATOM_MAX_FIRMWARE_VRAM_USAGE_INFO]; +}ATOM_VRAM_USAGE_BY_FIRMWARE_V1_5; + +/****************************************************************************/ +// Structure used in GPIO_Pin_LUTTable +/****************************************************************************/ +typedef struct _ATOM_GPIO_PIN_ASSIGNMENT +{ + USHORT usGpioPin_AIndex; + UCHAR ucGpioPinBitShift; + UCHAR ucGPIO_ID; +}ATOM_GPIO_PIN_ASSIGNMENT; + +typedef struct _ATOM_GPIO_PIN_LUT +{ + ATOM_COMMON_TABLE_HEADER sHeader; + ATOM_GPIO_PIN_ASSIGNMENT asGPIO_Pin[1]; +}ATOM_GPIO_PIN_LUT; + +/****************************************************************************/ +// Structure used in ComponentVideoInfoTable +/****************************************************************************/ +#define GPIO_PIN_ACTIVE_HIGH 0x1 + +#define MAX_SUPPORTED_CV_STANDARDS 5 + +// definitions for ATOM_D_INFO.ucSettings +#define ATOM_GPIO_SETTINGS_BITSHIFT_MASK 0x1F // [4:0] +#define ATOM_GPIO_SETTINGS_RESERVED_MASK 0x60 // [6:5] = must be zeroed out +#define ATOM_GPIO_SETTINGS_ACTIVE_MASK 0x80 // [7] + +typedef struct _ATOM_GPIO_INFO +{ + USHORT usAOffset; + UCHAR ucSettings; + UCHAR ucReserved; +}ATOM_GPIO_INFO; + +// definitions for ATOM_COMPONENT_VIDEO_INFO.ucMiscInfo (bit vector) +#define ATOM_CV_RESTRICT_FORMAT_SELECTION 0x2 + +// definitions for ATOM_COMPONENT_VIDEO_INFO.uc480i/uc480p/uc720p/uc1080i +#define ATOM_GPIO_DEFAULT_MODE_EN 0x80 //[7]; +#define ATOM_GPIO_SETTING_PERMODE_MASK 0x7F //[6:0] + +// definitions for ATOM_COMPONENT_VIDEO_INFO.ucLetterBoxMode +//Line 3 out put 5V. +#define ATOM_CV_LINE3_ASPECTRATIO_16_9_GPIO_A 0x01 //represent gpio 3 state for 16:9 +#define ATOM_CV_LINE3_ASPECTRATIO_16_9_GPIO_B 0x02 //represent gpio 4 state for 16:9 +#define ATOM_CV_LINE3_ASPECTRATIO_16_9_GPIO_SHIFT 0x0 + +//Line 3 out put 2.2V +#define ATOM_CV_LINE3_ASPECTRATIO_4_3_LETBOX_GPIO_A 0x04 //represent gpio 3 state for 4:3 Letter box +#define ATOM_CV_LINE3_ASPECTRATIO_4_3_LETBOX_GPIO_B 0x08 //represent gpio 4 state for 4:3 Letter box +#define ATOM_CV_LINE3_ASPECTRATIO_4_3_LETBOX_GPIO_SHIFT 0x2 + +//Line 3 out put 0V +#define ATOM_CV_LINE3_ASPECTRATIO_4_3_GPIO_A 0x10 //represent gpio 3 state for 4:3 +#define ATOM_CV_LINE3_ASPECTRATIO_4_3_GPIO_B 0x20 //represent gpio 4 state for 4:3 +#define ATOM_CV_LINE3_ASPECTRATIO_4_3_GPIO_SHIFT 0x4 + +#define ATOM_CV_LINE3_ASPECTRATIO_MASK 0x3F // bit [5:0] + +#define ATOM_CV_LINE3_ASPECTRATIO_EXIST 0x80 //bit 7 + +//GPIO bit index in gpio setting per mode value, also represend the block no. in gpio blocks. +#define ATOM_GPIO_INDEX_LINE3_ASPECRATIO_GPIO_A 3 //bit 3 in uc480i/uc480p/uc720p/uc1080i, which represend the default gpio bit setting for the mode. +#define ATOM_GPIO_INDEX_LINE3_ASPECRATIO_GPIO_B 4 //bit 4 in uc480i/uc480p/uc720p/uc1080i, which represend the default gpio bit setting for the mode. + + +typedef struct _ATOM_COMPONENT_VIDEO_INFO +{ + ATOM_COMMON_TABLE_HEADER sHeader; + USHORT usMask_PinRegisterIndex; + USHORT usEN_PinRegisterIndex; + USHORT usY_PinRegisterIndex; + USHORT usA_PinRegisterIndex; + UCHAR ucBitShift; + UCHAR ucPinActiveState; //ucPinActiveState: Bit0=1 active high, =0 active low + ATOM_DTD_FORMAT sReserved; // must be zeroed out + UCHAR ucMiscInfo; + UCHAR uc480i; + UCHAR uc480p; + UCHAR uc720p; + UCHAR uc1080i; + UCHAR ucLetterBoxMode; + UCHAR ucReserved[3]; + UCHAR ucNumOfWbGpioBlocks; //For Component video D-Connector support. If zere, NTSC type connector + ATOM_GPIO_INFO aWbGpioStateBlock[MAX_SUPPORTED_CV_STANDARDS]; + ATOM_DTD_FORMAT aModeTimings[MAX_SUPPORTED_CV_STANDARDS]; +}ATOM_COMPONENT_VIDEO_INFO; + +//ucTableFormatRevision=2 +//ucTableContentRevision=1 +typedef struct _ATOM_COMPONENT_VIDEO_INFO_V21 +{ + ATOM_COMMON_TABLE_HEADER sHeader; + UCHAR ucMiscInfo; + UCHAR uc480i; + UCHAR uc480p; + UCHAR uc720p; + UCHAR uc1080i; + UCHAR ucReserved; + UCHAR ucLetterBoxMode; + UCHAR ucNumOfWbGpioBlocks; //For Component video D-Connector support. If zere, NTSC type connector + ATOM_GPIO_INFO aWbGpioStateBlock[MAX_SUPPORTED_CV_STANDARDS]; + ATOM_DTD_FORMAT aModeTimings[MAX_SUPPORTED_CV_STANDARDS]; +}ATOM_COMPONENT_VIDEO_INFO_V21; + +#define ATOM_COMPONENT_VIDEO_INFO_LAST ATOM_COMPONENT_VIDEO_INFO_V21 + +/****************************************************************************/ +// Structure used in object_InfoTable +/****************************************************************************/ +typedef struct _ATOM_OBJECT_HEADER +{ + ATOM_COMMON_TABLE_HEADER sHeader; + USHORT usDeviceSupport; + USHORT usConnectorObjectTableOffset; + USHORT usRouterObjectTableOffset; + USHORT usEncoderObjectTableOffset; + USHORT usProtectionObjectTableOffset; //only available when Protection block is independent. + USHORT usDisplayPathTableOffset; +}ATOM_OBJECT_HEADER; + +typedef struct _ATOM_OBJECT_HEADER_V3 +{ + ATOM_COMMON_TABLE_HEADER sHeader; + USHORT usDeviceSupport; + USHORT usConnectorObjectTableOffset; + USHORT usRouterObjectTableOffset; + USHORT usEncoderObjectTableOffset; + USHORT usProtectionObjectTableOffset; //only available when Protection block is independent. + USHORT usDisplayPathTableOffset; + USHORT usMiscObjectTableOffset; +}ATOM_OBJECT_HEADER_V3; + +typedef struct _ATOM_DISPLAY_OBJECT_PATH +{ + USHORT usDeviceTag; //supported device + USHORT usSize; //the size of ATOM_DISPLAY_OBJECT_PATH + USHORT usConnObjectId; //Connector Object ID + USHORT usGPUObjectId; //GPU ID + USHORT usGraphicObjIds[1]; //1st Encoder Obj source from GPU to last Graphic Obj destinate to connector. +}ATOM_DISPLAY_OBJECT_PATH; + +typedef struct _ATOM_DISPLAY_EXTERNAL_OBJECT_PATH +{ + USHORT usDeviceTag; //supported device + USHORT usSize; //the size of ATOM_DISPLAY_OBJECT_PATH + USHORT usConnObjectId; //Connector Object ID + USHORT usGPUObjectId; //GPU ID + USHORT usGraphicObjIds[2]; //usGraphicObjIds[0]= GPU internal encoder, usGraphicObjIds[1]= external encoder +}ATOM_DISPLAY_EXTERNAL_OBJECT_PATH; + +typedef struct _ATOM_DISPLAY_OBJECT_PATH_TABLE +{ + UCHAR ucNumOfDispPath; + UCHAR ucVersion; + UCHAR ucPadding[2]; + ATOM_DISPLAY_OBJECT_PATH asDispPath[1]; +}ATOM_DISPLAY_OBJECT_PATH_TABLE; + + +typedef struct _ATOM_OBJECT //each object has this structure +{ + USHORT usObjectID; + USHORT usSrcDstTableOffset; + USHORT usRecordOffset; //this pointing to a bunch of records defined below + USHORT usReserved; +}ATOM_OBJECT; + +typedef struct _ATOM_OBJECT_TABLE //Above 4 object table offset pointing to a bunch of objects all have this structure +{ + UCHAR ucNumberOfObjects; + UCHAR ucPadding[3]; + ATOM_OBJECT asObjects[1]; +}ATOM_OBJECT_TABLE; + +typedef struct _ATOM_SRC_DST_TABLE_FOR_ONE_OBJECT //usSrcDstTableOffset pointing to this structure +{ + UCHAR ucNumberOfSrc; + USHORT usSrcObjectID[1]; + UCHAR ucNumberOfDst; + USHORT usDstObjectID[1]; +}ATOM_SRC_DST_TABLE_FOR_ONE_OBJECT; + + +//Two definitions below are for OPM on MXM module designs + +#define EXT_HPDPIN_LUTINDEX_0 0 +#define EXT_HPDPIN_LUTINDEX_1 1 +#define EXT_HPDPIN_LUTINDEX_2 2 +#define EXT_HPDPIN_LUTINDEX_3 3 +#define EXT_HPDPIN_LUTINDEX_4 4 +#define EXT_HPDPIN_LUTINDEX_5 5 +#define EXT_HPDPIN_LUTINDEX_6 6 +#define EXT_HPDPIN_LUTINDEX_7 7 +#define MAX_NUMBER_OF_EXT_HPDPIN_LUT_ENTRIES (EXT_HPDPIN_LUTINDEX_7+1) + +#define EXT_AUXDDC_LUTINDEX_0 0 +#define EXT_AUXDDC_LUTINDEX_1 1 +#define EXT_AUXDDC_LUTINDEX_2 2 +#define EXT_AUXDDC_LUTINDEX_3 3 +#define EXT_AUXDDC_LUTINDEX_4 4 +#define EXT_AUXDDC_LUTINDEX_5 5 +#define EXT_AUXDDC_LUTINDEX_6 6 +#define EXT_AUXDDC_LUTINDEX_7 7 +#define MAX_NUMBER_OF_EXT_AUXDDC_LUT_ENTRIES (EXT_AUXDDC_LUTINDEX_7+1) + +//ucChannelMapping are defined as following +//for DP connector, eDP, DP to VGA/LVDS +//Bit[1:0]: Define which pin connect to DP connector DP_Lane0, =0: source from GPU pin TX0, =1: from GPU pin TX1, =2: from GPU pin TX2, =3 from GPU pin TX3 +//Bit[3:2]: Define which pin connect to DP connector DP_Lane1, =0: source from GPU pin TX0, =1: from GPU pin TX1, =2: from GPU pin TX2, =3 from GPU pin TX3 +//Bit[5:4]: Define which pin connect to DP connector DP_Lane2, =0: source from GPU pin TX0, =1: from GPU pin TX1, =2: from GPU pin TX2, =3 from GPU pin TX3 +//Bit[7:6]: Define which pin connect to DP connector DP_Lane3, =0: source from GPU pin TX0, =1: from GPU pin TX1, =2: from GPU pin TX2, =3 from GPU pin TX3 +typedef struct _ATOM_DP_CONN_CHANNEL_MAPPING +{ +#if ATOM_BIG_ENDIAN + UCHAR ucDP_Lane3_Source:2; + UCHAR ucDP_Lane2_Source:2; + UCHAR ucDP_Lane1_Source:2; + UCHAR ucDP_Lane0_Source:2; +#else + UCHAR ucDP_Lane0_Source:2; + UCHAR ucDP_Lane1_Source:2; + UCHAR ucDP_Lane2_Source:2; + UCHAR ucDP_Lane3_Source:2; +#endif +}ATOM_DP_CONN_CHANNEL_MAPPING; + +//for DVI/HDMI, in dual link case, both links have to have same mapping. +//Bit[1:0]: Define which pin connect to DVI connector data Lane2, =0: source from GPU pin TX0, =1: from GPU pin TX1, =2: from GPU pin TX2, =3 from GPU pin TX3 +//Bit[3:2]: Define which pin connect to DVI connector data Lane1, =0: source from GPU pin TX0, =1: from GPU pin TX1, =2: from GPU pin TX2, =3 from GPU pin TX3 +//Bit[5:4]: Define which pin connect to DVI connector data Lane0, =0: source from GPU pin TX0, =1: from GPU pin TX1, =2: from GPU pin TX2, =3 from GPU pin TX3 +//Bit[7:6]: Define which pin connect to DVI connector clock lane, =0: source from GPU pin TX0, =1: from GPU pin TX1, =2: from GPU pin TX2, =3 from GPU pin TX3 +typedef struct _ATOM_DVI_CONN_CHANNEL_MAPPING +{ +#if ATOM_BIG_ENDIAN + UCHAR ucDVI_CLK_Source:2; + UCHAR ucDVI_DATA0_Source:2; + UCHAR ucDVI_DATA1_Source:2; + UCHAR ucDVI_DATA2_Source:2; +#else + UCHAR ucDVI_DATA2_Source:2; + UCHAR ucDVI_DATA1_Source:2; + UCHAR ucDVI_DATA0_Source:2; + UCHAR ucDVI_CLK_Source:2; +#endif +}ATOM_DVI_CONN_CHANNEL_MAPPING; + +typedef struct _EXT_DISPLAY_PATH +{ + USHORT usDeviceTag; //A bit vector to show what devices are supported + USHORT usDeviceACPIEnum; //16bit device ACPI id. + USHORT usDeviceConnector; //A physical connector for displays to plug in, using object connector definitions + UCHAR ucExtAUXDDCLutIndex; //An index into external AUX/DDC channel LUT + UCHAR ucExtHPDPINLutIndex; //An index into external HPD pin LUT + USHORT usExtEncoderObjId; //external encoder object id + union{ + UCHAR ucChannelMapping; // if ucChannelMapping=0, using default one to one mapping + ATOM_DP_CONN_CHANNEL_MAPPING asDPMapping; + ATOM_DVI_CONN_CHANNEL_MAPPING asDVIMapping; + }; + UCHAR ucReserved; + USHORT usReserved[2]; +}EXT_DISPLAY_PATH; + +#define NUMBER_OF_UCHAR_FOR_GUID 16 +#define MAX_NUMBER_OF_EXT_DISPLAY_PATH 7 + +typedef struct _ATOM_EXTERNAL_DISPLAY_CONNECTION_INFO +{ + ATOM_COMMON_TABLE_HEADER sHeader; + UCHAR ucGuid [NUMBER_OF_UCHAR_FOR_GUID]; // a GUID is a 16 byte long string + EXT_DISPLAY_PATH sPath[MAX_NUMBER_OF_EXT_DISPLAY_PATH]; // total of fixed 7 entries. + UCHAR ucChecksum; // a simple Checksum of the sum of whole structure equal to 0x0. + UCHAR uc3DStereoPinId; // use for eDP panel + UCHAR Reserved [6]; // for potential expansion +}ATOM_EXTERNAL_DISPLAY_CONNECTION_INFO; + +//Related definitions, all records are different but they have a commond header +typedef struct _ATOM_COMMON_RECORD_HEADER +{ + UCHAR ucRecordType; //An emun to indicate the record type + UCHAR ucRecordSize; //The size of the whole record in byte +}ATOM_COMMON_RECORD_HEADER; + + +#define ATOM_I2C_RECORD_TYPE 1 +#define ATOM_HPD_INT_RECORD_TYPE 2 +#define ATOM_OUTPUT_PROTECTION_RECORD_TYPE 3 +#define ATOM_CONNECTOR_DEVICE_TAG_RECORD_TYPE 4 +#define ATOM_CONNECTOR_DVI_EXT_INPUT_RECORD_TYPE 5 //Obsolete, switch to use GPIO_CNTL_RECORD_TYPE +#define ATOM_ENCODER_FPGA_CONTROL_RECORD_TYPE 6 //Obsolete, switch to use GPIO_CNTL_RECORD_TYPE +#define ATOM_CONNECTOR_CVTV_SHARE_DIN_RECORD_TYPE 7 +#define ATOM_JTAG_RECORD_TYPE 8 //Obsolete, switch to use GPIO_CNTL_RECORD_TYPE +#define ATOM_OBJECT_GPIO_CNTL_RECORD_TYPE 9 +#define ATOM_ENCODER_DVO_CF_RECORD_TYPE 10 +#define ATOM_CONNECTOR_CF_RECORD_TYPE 11 +#define ATOM_CONNECTOR_HARDCODE_DTD_RECORD_TYPE 12 +#define ATOM_CONNECTOR_PCIE_SUBCONNECTOR_RECORD_TYPE 13 +#define ATOM_ROUTER_DDC_PATH_SELECT_RECORD_TYPE 14 +#define ATOM_ROUTER_DATA_CLOCK_PATH_SELECT_RECORD_TYPE 15 +#define ATOM_CONNECTOR_HPDPIN_LUT_RECORD_TYPE 16 //This is for the case when connectors are not known to object table +#define ATOM_CONNECTOR_AUXDDC_LUT_RECORD_TYPE 17 //This is for the case when connectors are not known to object table +#define ATOM_OBJECT_LINK_RECORD_TYPE 18 //Once this record is present under one object, it indicats the oobject is linked to another obj described by the record +#define ATOM_CONNECTOR_REMOTE_CAP_RECORD_TYPE 19 +#define ATOM_ENCODER_CAP_RECORD_TYPE 20 + + +//Must be updated when new record type is added,equal to that record definition! +#define ATOM_MAX_OBJECT_RECORD_NUMBER ATOM_ENCODER_CAP_RECORD_TYPE + +typedef struct _ATOM_I2C_RECORD +{ + ATOM_COMMON_RECORD_HEADER sheader; + ATOM_I2C_ID_CONFIG sucI2cId; + UCHAR ucI2CAddr; //The slave address, it's 0 when the record is attached to connector for DDC +}ATOM_I2C_RECORD; + +typedef struct _ATOM_HPD_INT_RECORD +{ + ATOM_COMMON_RECORD_HEADER sheader; + UCHAR ucHPDIntGPIOID; //Corresponding block in GPIO_PIN_INFO table gives the pin info + UCHAR ucPlugged_PinState; +}ATOM_HPD_INT_RECORD; + + +typedef struct _ATOM_OUTPUT_PROTECTION_RECORD +{ + ATOM_COMMON_RECORD_HEADER sheader; + UCHAR ucProtectionFlag; + UCHAR ucReserved; +}ATOM_OUTPUT_PROTECTION_RECORD; + +typedef struct _ATOM_CONNECTOR_DEVICE_TAG +{ + ULONG ulACPIDeviceEnum; //Reserved for now + USHORT usDeviceID; //This Id is same as "ATOM_DEVICE_XXX_SUPPORT" + USHORT usPadding; +}ATOM_CONNECTOR_DEVICE_TAG; + +typedef struct _ATOM_CONNECTOR_DEVICE_TAG_RECORD +{ + ATOM_COMMON_RECORD_HEADER sheader; + UCHAR ucNumberOfDevice; + UCHAR ucReserved; + ATOM_CONNECTOR_DEVICE_TAG asDeviceTag[1]; //This Id is same as "ATOM_DEVICE_XXX_SUPPORT", 1 is only for allocation +}ATOM_CONNECTOR_DEVICE_TAG_RECORD; + + +typedef struct _ATOM_CONNECTOR_DVI_EXT_INPUT_RECORD +{ + ATOM_COMMON_RECORD_HEADER sheader; + UCHAR ucConfigGPIOID; + UCHAR ucConfigGPIOState; //Set to 1 when it's active high to enable external flow in + UCHAR ucFlowinGPIPID; + UCHAR ucExtInGPIPID; +}ATOM_CONNECTOR_DVI_EXT_INPUT_RECORD; + +typedef struct _ATOM_ENCODER_FPGA_CONTROL_RECORD +{ + ATOM_COMMON_RECORD_HEADER sheader; + UCHAR ucCTL1GPIO_ID; + UCHAR ucCTL1GPIOState; //Set to 1 when it's active high + UCHAR ucCTL2GPIO_ID; + UCHAR ucCTL2GPIOState; //Set to 1 when it's active high + UCHAR ucCTL3GPIO_ID; + UCHAR ucCTL3GPIOState; //Set to 1 when it's active high + UCHAR ucCTLFPGA_IN_ID; + UCHAR ucPadding[3]; +}ATOM_ENCODER_FPGA_CONTROL_RECORD; + +typedef struct _ATOM_CONNECTOR_CVTV_SHARE_DIN_RECORD +{ + ATOM_COMMON_RECORD_HEADER sheader; + UCHAR ucGPIOID; //Corresponding block in GPIO_PIN_INFO table gives the pin info + UCHAR ucTVActiveState; //Indicating when the pin==0 or 1 when TV is connected +}ATOM_CONNECTOR_CVTV_SHARE_DIN_RECORD; + +typedef struct _ATOM_JTAG_RECORD +{ + ATOM_COMMON_RECORD_HEADER sheader; + UCHAR ucTMSGPIO_ID; + UCHAR ucTMSGPIOState; //Set to 1 when it's active high + UCHAR ucTCKGPIO_ID; + UCHAR ucTCKGPIOState; //Set to 1 when it's active high + UCHAR ucTDOGPIO_ID; + UCHAR ucTDOGPIOState; //Set to 1 when it's active high + UCHAR ucTDIGPIO_ID; + UCHAR ucTDIGPIOState; //Set to 1 when it's active high + UCHAR ucPadding[2]; +}ATOM_JTAG_RECORD; + + +//The following generic object gpio pin control record type will replace JTAG_RECORD/FPGA_CONTROL_RECORD/DVI_EXT_INPUT_RECORD above gradually +typedef struct _ATOM_GPIO_PIN_CONTROL_PAIR +{ + UCHAR ucGPIOID; // GPIO_ID, find the corresponding ID in GPIO_LUT table + UCHAR ucGPIO_PinState; // Pin state showing how to set-up the pin +}ATOM_GPIO_PIN_CONTROL_PAIR; + +typedef struct _ATOM_OBJECT_GPIO_CNTL_RECORD +{ + ATOM_COMMON_RECORD_HEADER sheader; + UCHAR ucFlags; // Future expnadibility + UCHAR ucNumberOfPins; // Number of GPIO pins used to control the object + ATOM_GPIO_PIN_CONTROL_PAIR asGpio[1]; // the real gpio pin pair determined by number of pins ucNumberOfPins +}ATOM_OBJECT_GPIO_CNTL_RECORD; + +//Definitions for GPIO pin state +#define GPIO_PIN_TYPE_INPUT 0x00 +#define GPIO_PIN_TYPE_OUTPUT 0x10 +#define GPIO_PIN_TYPE_HW_CONTROL 0x20 + +//For GPIO_PIN_TYPE_OUTPUT the following is defined +#define GPIO_PIN_OUTPUT_STATE_MASK 0x01 +#define GPIO_PIN_OUTPUT_STATE_SHIFT 0 +#define GPIO_PIN_STATE_ACTIVE_LOW 0x0 +#define GPIO_PIN_STATE_ACTIVE_HIGH 0x1 + +// Indexes to GPIO array in GLSync record +#define ATOM_GPIO_INDEX_GLSYNC_REFCLK 0 +#define ATOM_GPIO_INDEX_GLSYNC_HSYNC 1 +#define ATOM_GPIO_INDEX_GLSYNC_VSYNC 2 +#define ATOM_GPIO_INDEX_GLSYNC_SWAP_REQ 3 +#define ATOM_GPIO_INDEX_GLSYNC_SWAP_GNT 4 +#define ATOM_GPIO_INDEX_GLSYNC_INTERRUPT 5 +#define ATOM_GPIO_INDEX_GLSYNC_V_RESET 6 +#define ATOM_GPIO_INDEX_GLSYNC_MAX 7 + +typedef struct _ATOM_ENCODER_DVO_CF_RECORD +{ + ATOM_COMMON_RECORD_HEADER sheader; + ULONG ulStrengthControl; // DVOA strength control for CF + UCHAR ucPadding[2]; +}ATOM_ENCODER_DVO_CF_RECORD; + +// Bit maps for ATOM_ENCODER_CAP_RECORD.ucEncoderCap +#define ATOM_ENCODER_CAP_RECORD_HBR2 0x01 // DP1.2 HBR2 is supported by this path + +typedef struct _ATOM_ENCODER_CAP_RECORD +{ + ATOM_COMMON_RECORD_HEADER sheader; + union { + USHORT usEncoderCap; + struct { +#if ATOM_BIG_ENDIAN + USHORT usReserved:15; // Bit1-15 may be defined for other capability in future + USHORT usHBR2Cap:1; // Bit0 is for DP1.2 HBR2 capability. +#else + USHORT usHBR2Cap:1; // Bit0 is for DP1.2 HBR2 capability. + USHORT usReserved:15; // Bit1-15 may be defined for other capability in future +#endif + }; + }; +}ATOM_ENCODER_CAP_RECORD; + +// value for ATOM_CONNECTOR_CF_RECORD.ucConnectedDvoBundle +#define ATOM_CONNECTOR_CF_RECORD_CONNECTED_UPPER12BITBUNDLEA 1 +#define ATOM_CONNECTOR_CF_RECORD_CONNECTED_LOWER12BITBUNDLEB 2 + +typedef struct _ATOM_CONNECTOR_CF_RECORD +{ + ATOM_COMMON_RECORD_HEADER sheader; + USHORT usMaxPixClk; + UCHAR ucFlowCntlGpioId; + UCHAR ucSwapCntlGpioId; + UCHAR ucConnectedDvoBundle; + UCHAR ucPadding; +}ATOM_CONNECTOR_CF_RECORD; + +typedef struct _ATOM_CONNECTOR_HARDCODE_DTD_RECORD +{ + ATOM_COMMON_RECORD_HEADER sheader; + ATOM_DTD_FORMAT asTiming; +}ATOM_CONNECTOR_HARDCODE_DTD_RECORD; + +typedef struct _ATOM_CONNECTOR_PCIE_SUBCONNECTOR_RECORD +{ + ATOM_COMMON_RECORD_HEADER sheader; //ATOM_CONNECTOR_PCIE_SUBCONNECTOR_RECORD_TYPE + UCHAR ucSubConnectorType; //CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D|X_ID_DUAL_LINK_DVI_D|HDMI_TYPE_A + UCHAR ucReserved; +}ATOM_CONNECTOR_PCIE_SUBCONNECTOR_RECORD; + + +typedef struct _ATOM_ROUTER_DDC_PATH_SELECT_RECORD +{ + ATOM_COMMON_RECORD_HEADER sheader; + UCHAR ucMuxType; //decide the number of ucMuxState, =0, no pin state, =1: single state with complement, >1: multiple state + UCHAR ucMuxControlPin; + UCHAR ucMuxState[2]; //for alligment purpose +}ATOM_ROUTER_DDC_PATH_SELECT_RECORD; + +typedef struct _ATOM_ROUTER_DATA_CLOCK_PATH_SELECT_RECORD +{ + ATOM_COMMON_RECORD_HEADER sheader; + UCHAR ucMuxType; + UCHAR ucMuxControlPin; + UCHAR ucMuxState[2]; //for alligment purpose +}ATOM_ROUTER_DATA_CLOCK_PATH_SELECT_RECORD; + +// define ucMuxType +#define ATOM_ROUTER_MUX_PIN_STATE_MASK 0x0f +#define ATOM_ROUTER_MUX_PIN_SINGLE_STATE_COMPLEMENT 0x01 + +typedef struct _ATOM_CONNECTOR_HPDPIN_LUT_RECORD //record for ATOM_CONNECTOR_HPDPIN_LUT_RECORD_TYPE +{ + ATOM_COMMON_RECORD_HEADER sheader; + UCHAR ucHPDPINMap[MAX_NUMBER_OF_EXT_HPDPIN_LUT_ENTRIES]; //An fixed size array which maps external pins to internal GPIO_PIN_INFO table +}ATOM_CONNECTOR_HPDPIN_LUT_RECORD; + +typedef struct _ATOM_CONNECTOR_AUXDDC_LUT_RECORD //record for ATOM_CONNECTOR_AUXDDC_LUT_RECORD_TYPE +{ + ATOM_COMMON_RECORD_HEADER sheader; + ATOM_I2C_ID_CONFIG ucAUXDDCMap[MAX_NUMBER_OF_EXT_AUXDDC_LUT_ENTRIES]; //An fixed size array which maps external pins to internal DDC ID +}ATOM_CONNECTOR_AUXDDC_LUT_RECORD; + +typedef struct _ATOM_OBJECT_LINK_RECORD +{ + ATOM_COMMON_RECORD_HEADER sheader; + USHORT usObjectID; //could be connector, encorder or other object in object.h +}ATOM_OBJECT_LINK_RECORD; + +typedef struct _ATOM_CONNECTOR_REMOTE_CAP_RECORD +{ + ATOM_COMMON_RECORD_HEADER sheader; + USHORT usReserved; +}ATOM_CONNECTOR_REMOTE_CAP_RECORD; + +/****************************************************************************/ +// ASIC voltage data table +/****************************************************************************/ +typedef struct _ATOM_VOLTAGE_INFO_HEADER +{ + USHORT usVDDCBaseLevel; //In number of 50mv unit + USHORT usReserved; //For possible extension table offset + UCHAR ucNumOfVoltageEntries; + UCHAR ucBytesPerVoltageEntry; + UCHAR ucVoltageStep; //Indicating in how many mv increament is one step, 0.5mv unit + UCHAR ucDefaultVoltageEntry; + UCHAR ucVoltageControlI2cLine; + UCHAR ucVoltageControlAddress; + UCHAR ucVoltageControlOffset; +}ATOM_VOLTAGE_INFO_HEADER; + +typedef struct _ATOM_VOLTAGE_INFO +{ + ATOM_COMMON_TABLE_HEADER sHeader; + ATOM_VOLTAGE_INFO_HEADER viHeader; + UCHAR ucVoltageEntries[64]; //64 is for allocation, the actual number of entry is present at ucNumOfVoltageEntries*ucBytesPerVoltageEntry +}ATOM_VOLTAGE_INFO; + + +typedef struct _ATOM_VOLTAGE_FORMULA +{ + USHORT usVoltageBaseLevel; // In number of 1mv unit + USHORT usVoltageStep; // Indicating in how many mv increament is one step, 1mv unit + UCHAR ucNumOfVoltageEntries; // Number of Voltage Entry, which indicate max Voltage + UCHAR ucFlag; // bit0=0 :step is 1mv =1 0.5mv + UCHAR ucBaseVID; // if there is no lookup table, VID= BaseVID + ( Vol - BaseLevle ) /VoltageStep + UCHAR ucReserved; + UCHAR ucVIDAdjustEntries[32]; // 32 is for allocation, the actual number of entry is present at ucNumOfVoltageEntries +}ATOM_VOLTAGE_FORMULA; + +typedef struct _VOLTAGE_LUT_ENTRY +{ + USHORT usVoltageCode; // The Voltage ID, either GPIO or I2C code + USHORT usVoltageValue; // The corresponding Voltage Value, in mV +}VOLTAGE_LUT_ENTRY; + +typedef struct _ATOM_VOLTAGE_FORMULA_V2 +{ + UCHAR ucNumOfVoltageEntries; // Number of Voltage Entry, which indicate max Voltage + UCHAR ucReserved[3]; + VOLTAGE_LUT_ENTRY asVIDAdjustEntries[32];// 32 is for allocation, the actual number of entries is in ucNumOfVoltageEntries +}ATOM_VOLTAGE_FORMULA_V2; + +typedef struct _ATOM_VOLTAGE_CONTROL +{ + UCHAR ucVoltageControlId; //Indicate it is controlled by I2C or GPIO or HW state machine + UCHAR ucVoltageControlI2cLine; + UCHAR ucVoltageControlAddress; + UCHAR ucVoltageControlOffset; + USHORT usGpioPin_AIndex; //GPIO_PAD register index + UCHAR ucGpioPinBitShift[9]; //at most 8 pin support 255 VIDs, termintate with 0xff + UCHAR ucReserved; +}ATOM_VOLTAGE_CONTROL; + +// Define ucVoltageControlId +#define VOLTAGE_CONTROLLED_BY_HW 0x00 +#define VOLTAGE_CONTROLLED_BY_I2C_MASK 0x7F +#define VOLTAGE_CONTROLLED_BY_GPIO 0x80 +#define VOLTAGE_CONTROL_ID_LM64 0x01 //I2C control, used for R5xx Core Voltage +#define VOLTAGE_CONTROL_ID_DAC 0x02 //I2C control, used for R5xx/R6xx MVDDC,MVDDQ or VDDCI +#define VOLTAGE_CONTROL_ID_VT116xM 0x03 //I2C control, used for R6xx Core Voltage +#define VOLTAGE_CONTROL_ID_DS4402 0x04 +#define VOLTAGE_CONTROL_ID_UP6266 0x05 +#define VOLTAGE_CONTROL_ID_SCORPIO 0x06 +#define VOLTAGE_CONTROL_ID_VT1556M 0x07 +#define VOLTAGE_CONTROL_ID_CHL822x 0x08 +#define VOLTAGE_CONTROL_ID_VT1586M 0x09 + +typedef struct _ATOM_VOLTAGE_OBJECT +{ + UCHAR ucVoltageType; //Indicate Voltage Source: VDDC, MVDDC, MVDDQ or MVDDCI + UCHAR ucSize; //Size of Object + ATOM_VOLTAGE_CONTROL asControl; //describ how to control + ATOM_VOLTAGE_FORMULA asFormula; //Indicate How to convert real Voltage to VID +}ATOM_VOLTAGE_OBJECT; + +typedef struct _ATOM_VOLTAGE_OBJECT_V2 +{ + UCHAR ucVoltageType; //Indicate Voltage Source: VDDC, MVDDC, MVDDQ or MVDDCI + UCHAR ucSize; //Size of Object + ATOM_VOLTAGE_CONTROL asControl; //describ how to control + ATOM_VOLTAGE_FORMULA_V2 asFormula; //Indicate How to convert real Voltage to VID +}ATOM_VOLTAGE_OBJECT_V2; + +typedef struct _ATOM_VOLTAGE_OBJECT_INFO +{ + ATOM_COMMON_TABLE_HEADER sHeader; + ATOM_VOLTAGE_OBJECT asVoltageObj[3]; //Info for Voltage control +}ATOM_VOLTAGE_OBJECT_INFO; + +typedef struct _ATOM_VOLTAGE_OBJECT_INFO_V2 +{ + ATOM_COMMON_TABLE_HEADER sHeader; + ATOM_VOLTAGE_OBJECT_V2 asVoltageObj[3]; //Info for Voltage control +}ATOM_VOLTAGE_OBJECT_INFO_V2; + +typedef struct _ATOM_LEAKID_VOLTAGE +{ + UCHAR ucLeakageId; + UCHAR ucReserved; + USHORT usVoltage; +}ATOM_LEAKID_VOLTAGE; + +typedef struct _ATOM_ASIC_PROFILE_VOLTAGE +{ + UCHAR ucProfileId; + UCHAR ucReserved; + USHORT usSize; + USHORT usEfuseSpareStartAddr; + USHORT usFuseIndex[8]; //from LSB to MSB, Max 8bit,end of 0xffff if less than 8 efuse id, + ATOM_LEAKID_VOLTAGE asLeakVol[2]; //Leakid and relatd voltage +}ATOM_ASIC_PROFILE_VOLTAGE; + +//ucProfileId +#define ATOM_ASIC_PROFILE_ID_EFUSE_VOLTAGE 1 +#define ATOM_ASIC_PROFILE_ID_EFUSE_PERFORMANCE_VOLTAGE 1 +#define ATOM_ASIC_PROFILE_ID_EFUSE_THERMAL_VOLTAGE 2 + +typedef struct _ATOM_ASIC_PROFILING_INFO +{ + ATOM_COMMON_TABLE_HEADER asHeader; + ATOM_ASIC_PROFILE_VOLTAGE asVoltage; +}ATOM_ASIC_PROFILING_INFO; + +typedef struct _ATOM_POWER_SOURCE_OBJECT +{ + UCHAR ucPwrSrcId; // Power source + UCHAR ucPwrSensorType; // GPIO, I2C or none + UCHAR ucPwrSensId; // if GPIO detect, it is GPIO id, if I2C detect, it is I2C id + UCHAR ucPwrSensSlaveAddr; // Slave address if I2C detect + UCHAR ucPwrSensRegIndex; // I2C register Index if I2C detect + UCHAR ucPwrSensRegBitMask; // detect which bit is used if I2C detect + UCHAR ucPwrSensActiveState; // high active or low active + UCHAR ucReserve[3]; // reserve + USHORT usSensPwr; // in unit of watt +}ATOM_POWER_SOURCE_OBJECT; + +typedef struct _ATOM_POWER_SOURCE_INFO +{ + ATOM_COMMON_TABLE_HEADER asHeader; + UCHAR asPwrbehave[16]; + ATOM_POWER_SOURCE_OBJECT asPwrObj[1]; +}ATOM_POWER_SOURCE_INFO; + + +//Define ucPwrSrcId +#define POWERSOURCE_PCIE_ID1 0x00 +#define POWERSOURCE_6PIN_CONNECTOR_ID1 0x01 +#define POWERSOURCE_8PIN_CONNECTOR_ID1 0x02 +#define POWERSOURCE_6PIN_CONNECTOR_ID2 0x04 +#define POWERSOURCE_8PIN_CONNECTOR_ID2 0x08 + +//define ucPwrSensorId +#define POWER_SENSOR_ALWAYS 0x00 +#define POWER_SENSOR_GPIO 0x01 +#define POWER_SENSOR_I2C 0x02 + +typedef struct _ATOM_CLK_VOLT_CAPABILITY +{ + ULONG ulVoltageIndex; // The Voltage Index indicated by FUSE, same voltage index shared with SCLK DPM fuse table + ULONG ulMaximumSupportedCLK; // Maximum clock supported with specified voltage index, unit in 10kHz +}ATOM_CLK_VOLT_CAPABILITY; + +typedef struct _ATOM_AVAILABLE_SCLK_LIST +{ + ULONG ulSupportedSCLK; // Maximum clock supported with specified voltage index, unit in 10kHz + USHORT usVoltageIndex; // The Voltage Index indicated by FUSE for specified SCLK + USHORT usVoltageID; // The Voltage ID indicated by FUSE for specified SCLK +}ATOM_AVAILABLE_SCLK_LIST; + +// ATOM_INTEGRATED_SYSTEM_INFO_V6 ulSystemConfig cap definition +#define ATOM_IGP_INFO_V6_SYSTEM_CONFIG__PCIE_POWER_GATING_ENABLE 1 // refer to ulSystemConfig bit[0] + +// this IntegrateSystemInfoTable is used for Liano/Ontario APU +typedef struct _ATOM_INTEGRATED_SYSTEM_INFO_V6 +{ + ATOM_COMMON_TABLE_HEADER sHeader; + ULONG ulBootUpEngineClock; + ULONG ulDentistVCOFreq; + ULONG ulBootUpUMAClock; + ATOM_CLK_VOLT_CAPABILITY sDISPCLK_Voltage[4]; + ULONG ulBootUpReqDisplayVector; + ULONG ulOtherDisplayMisc; + ULONG ulGPUCapInfo; + ULONG ulSB_MMIO_Base_Addr; + USHORT usRequestedPWMFreqInHz; + UCHAR ucHtcTmpLmt; + UCHAR ucHtcHystLmt; + ULONG ulMinEngineClock; + ULONG ulSystemConfig; + ULONG ulCPUCapInfo; + USHORT usNBP0Voltage; + USHORT usNBP1Voltage; + USHORT usBootUpNBVoltage; + USHORT usExtDispConnInfoOffset; + USHORT usPanelRefreshRateRange; + UCHAR ucMemoryType; + UCHAR ucUMAChannelNumber; + ULONG ulCSR_M3_ARB_CNTL_DEFAULT[10]; + ULONG ulCSR_M3_ARB_CNTL_UVD[10]; + ULONG ulCSR_M3_ARB_CNTL_FS3D[10]; + ATOM_AVAILABLE_SCLK_LIST sAvail_SCLK[5]; + ULONG ulGMCRestoreResetTime; + ULONG ulMinimumNClk; + ULONG ulIdleNClk; + ULONG ulDDR_DLL_PowerUpTime; + ULONG ulDDR_PLL_PowerUpTime; + USHORT usPCIEClkSSPercentage; + USHORT usPCIEClkSSType; + USHORT usLvdsSSPercentage; + USHORT usLvdsSSpreadRateIn10Hz; + USHORT usHDMISSPercentage; + USHORT usHDMISSpreadRateIn10Hz; + USHORT usDVISSPercentage; + USHORT usDVISSpreadRateIn10Hz; + ULONG ulReserved3[21]; + ATOM_EXTERNAL_DISPLAY_CONNECTION_INFO sExtDispConnInfo; +}ATOM_INTEGRATED_SYSTEM_INFO_V6; + +// ulGPUCapInfo +#define INTEGRATED_SYSTEM_INFO_V6_GPUCAPINFO__TMDSHDMI_COHERENT_SINGLEPLL_MODE 0x01 +#define INTEGRATED_SYSTEM_INFO_V6_GPUCAPINFO__DISABLE_AUX_HW_MODE_DETECTION 0x08 + +// ulOtherDisplayMisc +#define INTEGRATED_SYSTEM_INFO__GET_EDID_CALLBACK_FUNC_SUPPORT 0x01 + + +/********************************************************************************************************************** + ATOM_INTEGRATED_SYSTEM_INFO_V6 Description +ulBootUpEngineClock: VBIOS bootup Engine clock frequency, in 10kHz unit. if it is equal 0, then VBIOS use pre-defined bootup engine clock +ulDentistVCOFreq: Dentist VCO clock in 10kHz unit. +ulBootUpUMAClock: System memory boot up clock frequency in 10Khz unit. +sDISPCLK_Voltage: Report Display clock voltage requirement. + +ulBootUpReqDisplayVector: VBIOS boot up display IDs, following are supported devices in Liano/Ontaio projects: + ATOM_DEVICE_CRT1_SUPPORT 0x0001 + ATOM_DEVICE_CRT2_SUPPORT 0x0010 + ATOM_DEVICE_DFP1_SUPPORT 0x0008 + ATOM_DEVICE_DFP6_SUPPORT 0x0040 + ATOM_DEVICE_DFP2_SUPPORT 0x0080 + ATOM_DEVICE_DFP3_SUPPORT 0x0200 + ATOM_DEVICE_DFP4_SUPPORT 0x0400 + ATOM_DEVICE_DFP5_SUPPORT 0x0800 + ATOM_DEVICE_LCD1_SUPPORT 0x0002 +ulOtherDisplayMisc: Other display related flags, not defined yet. +ulGPUCapInfo: bit[0]=0: TMDS/HDMI Coherent Mode use cascade PLL mode. + =1: TMDS/HDMI Coherent Mode use signel PLL mode. + bit[3]=0: Enable HW AUX mode detection logic + =1: Disable HW AUX mode dettion logic +ulSB_MMIO_Base_Addr: Physical Base address to SB MMIO space. Driver needs to initialize it for SMU usage. + +usRequestedPWMFreqInHz: When it's set to 0x0 by SBIOS: the LCD BackLight is not controlled by GPU(SW). + Any attempt to change BL using VBIOS function or enable VariBri from PP table is not effective since ATOM_BIOS_INFO_BL_CONTROLLED_BY_GPU==0; + + When it's set to a non-zero frequency, the BackLight is controlled by GPU (SW) in one of two ways below: + 1. SW uses the GPU BL PWM output to control the BL, in chis case, this non-zero frequency determines what freq GPU should use; + VBIOS will set up proper PWM frequency and ATOM_BIOS_INFO_BL_CONTROLLED_BY_GPU==1,as the result, + Changing BL using VBIOS function is functional in both driver and non-driver present environment; + and enabling VariBri under the driver environment from PP table is optional. + + 2. SW uses other means to control BL (like DPCD),this non-zero frequency serves as a flag only indicating + that BL control from GPU is expected. + VBIOS will NOT set up PWM frequency but make ATOM_BIOS_INFO_BL_CONTROLLED_BY_GPU==1 + Changing BL using VBIOS function could be functional in both driver and non-driver present environment,but + it's per platform + and enabling VariBri under the driver environment from PP table is optional. + +ucHtcTmpLmt: Refer to D18F3x64 bit[22:16], HtcTmpLmt. + Threshold on value to enter HTC_active state. +ucHtcHystLmt: Refer to D18F3x64 bit[27:24], HtcHystLmt. + To calculate threshold off value to exit HTC_active state, which is Threshold on vlaue minus ucHtcHystLmt. +ulMinEngineClock: Minimum SCLK allowed in 10kHz unit. This is calculated based on WRCK Fuse settings. +ulSystemConfig: Bit[0]=0: PCIE Power Gating Disabled + =1: PCIE Power Gating Enabled + Bit[1]=0: DDR-DLL shut-down feature disabled. + 1: DDR-DLL shut-down feature enabled. + Bit[2]=0: DDR-PLL Power down feature disabled. + 1: DDR-PLL Power down feature enabled. +ulCPUCapInfo: TBD +usNBP0Voltage: VID for voltage on NB P0 State +usNBP1Voltage: VID for voltage on NB P1 State +usBootUpNBVoltage: Voltage Index of GNB voltage configured by SBIOS, which is suffcient to support VBIOS DISPCLK requirement. +usExtDispConnInfoOffset: Offset to sExtDispConnInfo inside the structure +usPanelRefreshRateRange: Bit vector for LCD supported refresh rate range. If DRR is requestd by the platform, at least two bits need to be set + to indicate a range. + SUPPORTED_LCD_REFRESHRATE_30Hz 0x0004 + SUPPORTED_LCD_REFRESHRATE_40Hz 0x0008 + SUPPORTED_LCD_REFRESHRATE_50Hz 0x0010 + SUPPORTED_LCD_REFRESHRATE_60Hz 0x0020 +ucMemoryType: [3:0]=1:DDR1;=2:DDR2;=3:DDR3.[7:4] is reserved. +ucUMAChannelNumber: System memory channel numbers. +ulCSR_M3_ARB_CNTL_DEFAULT[10]: Arrays with values for CSR M3 arbiter for default +ulCSR_M3_ARB_CNTL_UVD[10]: Arrays with values for CSR M3 arbiter for UVD playback. +ulCSR_M3_ARB_CNTL_FS3D[10]: Arrays with values for CSR M3 arbiter for Full Screen 3D applications. +sAvail_SCLK[5]: Arrays to provide available list of SLCK and corresponding voltage, order from low to high +ulGMCRestoreResetTime: GMC power restore and GMC reset time to calculate data reconnection latency. Unit in ns. +ulMinimumNClk: Minimum NCLK speed among all NB-Pstates to calcualte data reconnection latency. Unit in 10kHz. +ulIdleNClk: NCLK speed while memory runs in self-refresh state. Unit in 10kHz. +ulDDR_DLL_PowerUpTime: DDR PHY DLL power up time. Unit in ns. +ulDDR_PLL_PowerUpTime: DDR PHY PLL power up time. Unit in ns. +usPCIEClkSSPercentage: PCIE Clock Spread Spectrum Percentage in unit 0.01%; 100 mean 1%. +usPCIEClkSSType: PCIE Clock Spread Spectrum Type. 0 for Down spread(default); 1 for Center spread. +usLvdsSSPercentage: LVDS panel ( not include eDP ) Spread Spectrum Percentage in unit of 0.01%, =0, use VBIOS default setting. +usLvdsSSpreadRateIn10Hz: LVDS panel ( not include eDP ) Spread Spectrum frequency in unit of 10Hz, =0, use VBIOS default setting. +usHDMISSPercentage: HDMI Spread Spectrum Percentage in unit 0.01%; 100 mean 1%, =0, use VBIOS default setting. +usHDMISSpreadRateIn10Hz: HDMI Spread Spectrum frequency in unit of 10Hz, =0, use VBIOS default setting. +usDVISSPercentage: DVI Spread Spectrum Percentage in unit 0.01%; 100 mean 1%, =0, use VBIOS default setting. +usDVISSpreadRateIn10Hz: DVI Spread Spectrum frequency in unit of 10Hz, =0, use VBIOS default setting. +**********************************************************************************************************************/ + +/**************************************************************************/ +// This portion is only used when ext thermal chip or engine/memory clock SS chip is populated on a design +//Memory SS Info Table +//Define Memory Clock SS chip ID +#define ICS91719 1 +#define ICS91720 2 + +//Define one structure to inform SW a "block of data" writing to external SS chip via I2C protocol +typedef struct _ATOM_I2C_DATA_RECORD +{ + UCHAR ucNunberOfBytes; //Indicates how many bytes SW needs to write to the external ASIC for one block, besides to "Start" and "Stop" + UCHAR ucI2CData[1]; //I2C data in bytes, should be less than 16 bytes usually +}ATOM_I2C_DATA_RECORD; + + +//Define one structure to inform SW how many blocks of data writing to external SS chip via I2C protocol, in addition to other information +typedef struct _ATOM_I2C_DEVICE_SETUP_INFO +{ + ATOM_I2C_ID_CONFIG_ACCESS sucI2cId; //I2C line and HW/SW assisted cap. + UCHAR ucSSChipID; //SS chip being used + UCHAR ucSSChipSlaveAddr; //Slave Address to set up this SS chip + UCHAR ucNumOfI2CDataRecords; //number of data block + ATOM_I2C_DATA_RECORD asI2CData[1]; +}ATOM_I2C_DEVICE_SETUP_INFO; + +//========================================================================================== +typedef struct _ATOM_ASIC_MVDD_INFO +{ + ATOM_COMMON_TABLE_HEADER sHeader; + ATOM_I2C_DEVICE_SETUP_INFO asI2CSetup[1]; +}ATOM_ASIC_MVDD_INFO; + +//========================================================================================== +#define ATOM_MCLK_SS_INFO ATOM_ASIC_MVDD_INFO + +//========================================================================================== +/**************************************************************************/ + +typedef struct _ATOM_ASIC_SS_ASSIGNMENT +{ + ULONG ulTargetClockRange; //Clock Out frequence (VCO ), in unit of 10Khz + USHORT usSpreadSpectrumPercentage; //in unit of 0.01% + USHORT usSpreadRateInKhz; //in unit of kHz, modulation freq + UCHAR ucClockIndication; //Indicate which clock source needs SS + UCHAR ucSpreadSpectrumMode; //Bit1=0 Down Spread,=1 Center Spread. + UCHAR ucReserved[2]; +}ATOM_ASIC_SS_ASSIGNMENT; + +//Define ucClockIndication, SW uses the IDs below to search if the SS is required/enabled on a clock branch/signal type. +//SS is not required or enabled if a match is not found. +#define ASIC_INTERNAL_MEMORY_SS 1 +#define ASIC_INTERNAL_ENGINE_SS 2 +#define ASIC_INTERNAL_UVD_SS 3 +#define ASIC_INTERNAL_SS_ON_TMDS 4 +#define ASIC_INTERNAL_SS_ON_HDMI 5 +#define ASIC_INTERNAL_SS_ON_LVDS 6 +#define ASIC_INTERNAL_SS_ON_DP 7 +#define ASIC_INTERNAL_SS_ON_DCPLL 8 +#define ASIC_EXTERNAL_SS_ON_DP_CLOCK 9 + +typedef struct _ATOM_ASIC_SS_ASSIGNMENT_V2 +{ + ULONG ulTargetClockRange; //For mem/engine/uvd, Clock Out frequence (VCO ), in unit of 10Khz + //For TMDS/HDMI/LVDS, it is pixel clock , for DP, it is link clock ( 27000 or 16200 ) + USHORT usSpreadSpectrumPercentage; //in unit of 0.01% + USHORT usSpreadRateIn10Hz; //in unit of 10Hz, modulation freq + UCHAR ucClockIndication; //Indicate which clock source needs SS + UCHAR ucSpreadSpectrumMode; //Bit0=0 Down Spread,=1 Center Spread, bit1=0: internal SS bit1=1: external SS + UCHAR ucReserved[2]; +}ATOM_ASIC_SS_ASSIGNMENT_V2; + +//ucSpreadSpectrumMode +//#define ATOM_SS_DOWN_SPREAD_MODE_MASK 0x00000000 +//#define ATOM_SS_DOWN_SPREAD_MODE 0x00000000 +//#define ATOM_SS_CENTRE_SPREAD_MODE_MASK 0x00000001 +//#define ATOM_SS_CENTRE_SPREAD_MODE 0x00000001 +//#define ATOM_INTERNAL_SS_MASK 0x00000000 +//#define ATOM_EXTERNAL_SS_MASK 0x00000002 + +typedef struct _ATOM_ASIC_INTERNAL_SS_INFO +{ + ATOM_COMMON_TABLE_HEADER sHeader; + ATOM_ASIC_SS_ASSIGNMENT asSpreadSpectrum[4]; +}ATOM_ASIC_INTERNAL_SS_INFO; + +typedef struct _ATOM_ASIC_INTERNAL_SS_INFO_V2 +{ + ATOM_COMMON_TABLE_HEADER sHeader; + ATOM_ASIC_SS_ASSIGNMENT_V2 asSpreadSpectrum[1]; //this is point only. +}ATOM_ASIC_INTERNAL_SS_INFO_V2; + +typedef struct _ATOM_ASIC_SS_ASSIGNMENT_V3 +{ + ULONG ulTargetClockRange; //For mem/engine/uvd, Clock Out frequence (VCO ), in unit of 10Khz + //For TMDS/HDMI/LVDS, it is pixel clock , for DP, it is link clock ( 27000 or 16200 ) + USHORT usSpreadSpectrumPercentage; //in unit of 0.01% + USHORT usSpreadRateIn10Hz; //in unit of 10Hz, modulation freq + UCHAR ucClockIndication; //Indicate which clock source needs SS + UCHAR ucSpreadSpectrumMode; //Bit0=0 Down Spread,=1 Center Spread, bit1=0: internal SS bit1=1: external SS + UCHAR ucReserved[2]; +}ATOM_ASIC_SS_ASSIGNMENT_V3; + +typedef struct _ATOM_ASIC_INTERNAL_SS_INFO_V3 +{ + ATOM_COMMON_TABLE_HEADER sHeader; + ATOM_ASIC_SS_ASSIGNMENT_V3 asSpreadSpectrum[1]; //this is pointer only. +}ATOM_ASIC_INTERNAL_SS_INFO_V3; + + +//==============================Scratch Pad Definition Portion=============================== +#define ATOM_DEVICE_CONNECT_INFO_DEF 0 +#define ATOM_ROM_LOCATION_DEF 1 +#define ATOM_TV_STANDARD_DEF 2 +#define ATOM_ACTIVE_INFO_DEF 3 +#define ATOM_LCD_INFO_DEF 4 +#define ATOM_DOS_REQ_INFO_DEF 5 +#define ATOM_ACC_CHANGE_INFO_DEF 6 +#define ATOM_DOS_MODE_INFO_DEF 7 +#define ATOM_I2C_CHANNEL_STATUS_DEF 8 +#define ATOM_I2C_CHANNEL_STATUS1_DEF 9 + + +// BIOS_0_SCRATCH Definition +#define ATOM_S0_CRT1_MONO 0x00000001L +#define ATOM_S0_CRT1_COLOR 0x00000002L +#define ATOM_S0_CRT1_MASK (ATOM_S0_CRT1_MONO+ATOM_S0_CRT1_COLOR) + +#define ATOM_S0_TV1_COMPOSITE_A 0x00000004L +#define ATOM_S0_TV1_SVIDEO_A 0x00000008L +#define ATOM_S0_TV1_MASK_A (ATOM_S0_TV1_COMPOSITE_A+ATOM_S0_TV1_SVIDEO_A) + +#define ATOM_S0_CV_A 0x00000010L +#define ATOM_S0_CV_DIN_A 0x00000020L +#define ATOM_S0_CV_MASK_A (ATOM_S0_CV_A+ATOM_S0_CV_DIN_A) + + +#define ATOM_S0_CRT2_MONO 0x00000100L +#define ATOM_S0_CRT2_COLOR 0x00000200L +#define ATOM_S0_CRT2_MASK (ATOM_S0_CRT2_MONO+ATOM_S0_CRT2_COLOR) + +#define ATOM_S0_TV1_COMPOSITE 0x00000400L +#define ATOM_S0_TV1_SVIDEO 0x00000800L +#define ATOM_S0_TV1_SCART 0x00004000L +#define ATOM_S0_TV1_MASK (ATOM_S0_TV1_COMPOSITE+ATOM_S0_TV1_SVIDEO+ATOM_S0_TV1_SCART) + +#define ATOM_S0_CV 0x00001000L +#define ATOM_S0_CV_DIN 0x00002000L +#define ATOM_S0_CV_MASK (ATOM_S0_CV+ATOM_S0_CV_DIN) + +#define ATOM_S0_DFP1 0x00010000L +#define ATOM_S0_DFP2 0x00020000L +#define ATOM_S0_LCD1 0x00040000L +#define ATOM_S0_LCD2 0x00080000L +#define ATOM_S0_DFP6 0x00100000L +#define ATOM_S0_DFP3 0x00200000L +#define ATOM_S0_DFP4 0x00400000L +#define ATOM_S0_DFP5 0x00800000L + +#define ATOM_S0_DFP_MASK ATOM_S0_DFP1 | ATOM_S0_DFP2 | ATOM_S0_DFP3 | ATOM_S0_DFP4 | ATOM_S0_DFP5 | ATOM_S0_DFP6 + +#define ATOM_S0_FAD_REGISTER_BUG 0x02000000L // If set, indicates we are running a PCIE asic with + // the FAD/HDP reg access bug. Bit is read by DAL, this is obsolete from RV5xx + +#define ATOM_S0_THERMAL_STATE_MASK 0x1C000000L +#define ATOM_S0_THERMAL_STATE_SHIFT 26 + +#define ATOM_S0_SYSTEM_POWER_STATE_MASK 0xE0000000L +#define ATOM_S0_SYSTEM_POWER_STATE_SHIFT 29 + +#define ATOM_S0_SYSTEM_POWER_STATE_VALUE_AC 1 +#define ATOM_S0_SYSTEM_POWER_STATE_VALUE_DC 2 +#define ATOM_S0_SYSTEM_POWER_STATE_VALUE_LITEAC 3 +#define ATOM_S0_SYSTEM_POWER_STATE_VALUE_LIT2AC 4 + +//Byte aligned definition for BIOS usage +#define ATOM_S0_CRT1_MONOb0 0x01 +#define ATOM_S0_CRT1_COLORb0 0x02 +#define ATOM_S0_CRT1_MASKb0 (ATOM_S0_CRT1_MONOb0+ATOM_S0_CRT1_COLORb0) + +#define ATOM_S0_TV1_COMPOSITEb0 0x04 +#define ATOM_S0_TV1_SVIDEOb0 0x08 +#define ATOM_S0_TV1_MASKb0 (ATOM_S0_TV1_COMPOSITEb0+ATOM_S0_TV1_SVIDEOb0) + +#define ATOM_S0_CVb0 0x10 +#define ATOM_S0_CV_DINb0 0x20 +#define ATOM_S0_CV_MASKb0 (ATOM_S0_CVb0+ATOM_S0_CV_DINb0) + +#define ATOM_S0_CRT2_MONOb1 0x01 +#define ATOM_S0_CRT2_COLORb1 0x02 +#define ATOM_S0_CRT2_MASKb1 (ATOM_S0_CRT2_MONOb1+ATOM_S0_CRT2_COLORb1) + +#define ATOM_S0_TV1_COMPOSITEb1 0x04 +#define ATOM_S0_TV1_SVIDEOb1 0x08 +#define ATOM_S0_TV1_SCARTb1 0x40 +#define ATOM_S0_TV1_MASKb1 (ATOM_S0_TV1_COMPOSITEb1+ATOM_S0_TV1_SVIDEOb1+ATOM_S0_TV1_SCARTb1) + +#define ATOM_S0_CVb1 0x10 +#define ATOM_S0_CV_DINb1 0x20 +#define ATOM_S0_CV_MASKb1 (ATOM_S0_CVb1+ATOM_S0_CV_DINb1) + +#define ATOM_S0_DFP1b2 0x01 +#define ATOM_S0_DFP2b2 0x02 +#define ATOM_S0_LCD1b2 0x04 +#define ATOM_S0_LCD2b2 0x08 +#define ATOM_S0_DFP6b2 0x10 +#define ATOM_S0_DFP3b2 0x20 +#define ATOM_S0_DFP4b2 0x40 +#define ATOM_S0_DFP5b2 0x80 + + +#define ATOM_S0_THERMAL_STATE_MASKb3 0x1C +#define ATOM_S0_THERMAL_STATE_SHIFTb3 2 + +#define ATOM_S0_SYSTEM_POWER_STATE_MASKb3 0xE0 +#define ATOM_S0_LCD1_SHIFT 18 + +// BIOS_1_SCRATCH Definition +#define ATOM_S1_ROM_LOCATION_MASK 0x0000FFFFL +#define ATOM_S1_PCI_BUS_DEV_MASK 0xFFFF0000L + +// BIOS_2_SCRATCH Definition +#define ATOM_S2_TV1_STANDARD_MASK 0x0000000FL +#define ATOM_S2_CURRENT_BL_LEVEL_MASK 0x0000FF00L +#define ATOM_S2_CURRENT_BL_LEVEL_SHIFT 8 + +#define ATOM_S2_FORCEDLOWPWRMODE_STATE_MASK 0x0C000000L +#define ATOM_S2_FORCEDLOWPWRMODE_STATE_MASK_SHIFT 26 +#define ATOM_S2_FORCEDLOWPWRMODE_STATE_CHANGE 0x10000000L + +#define ATOM_S2_DEVICE_DPMS_STATE 0x00010000L +#define ATOM_S2_VRI_BRIGHT_ENABLE 0x20000000L + +#define ATOM_S2_DISPLAY_ROTATION_0_DEGREE 0x0 +#define ATOM_S2_DISPLAY_ROTATION_90_DEGREE 0x1 +#define ATOM_S2_DISPLAY_ROTATION_180_DEGREE 0x2 +#define ATOM_S2_DISPLAY_ROTATION_270_DEGREE 0x3 +#define ATOM_S2_DISPLAY_ROTATION_DEGREE_SHIFT 30 +#define ATOM_S2_DISPLAY_ROTATION_ANGLE_MASK 0xC0000000L + + +//Byte aligned definition for BIOS usage +#define ATOM_S2_TV1_STANDARD_MASKb0 0x0F +#define ATOM_S2_CURRENT_BL_LEVEL_MASKb1 0xFF +#define ATOM_S2_DEVICE_DPMS_STATEb2 0x01 + +#define ATOM_S2_DEVICE_DPMS_MASKw1 0x3FF +#define ATOM_S2_FORCEDLOWPWRMODE_STATE_MASKb3 0x0C +#define ATOM_S2_FORCEDLOWPWRMODE_STATE_CHANGEb3 0x10 +#define ATOM_S2_VRI_BRIGHT_ENABLEb3 0x20 +#define ATOM_S2_ROTATION_STATE_MASKb3 0xC0 + + +// BIOS_3_SCRATCH Definition +#define ATOM_S3_CRT1_ACTIVE 0x00000001L +#define ATOM_S3_LCD1_ACTIVE 0x00000002L +#define ATOM_S3_TV1_ACTIVE 0x00000004L +#define ATOM_S3_DFP1_ACTIVE 0x00000008L +#define ATOM_S3_CRT2_ACTIVE 0x00000010L +#define ATOM_S3_LCD2_ACTIVE 0x00000020L +#define ATOM_S3_DFP6_ACTIVE 0x00000040L +#define ATOM_S3_DFP2_ACTIVE 0x00000080L +#define ATOM_S3_CV_ACTIVE 0x00000100L +#define ATOM_S3_DFP3_ACTIVE 0x00000200L +#define ATOM_S3_DFP4_ACTIVE 0x00000400L +#define ATOM_S3_DFP5_ACTIVE 0x00000800L + +#define ATOM_S3_DEVICE_ACTIVE_MASK 0x00000FFFL + +#define ATOM_S3_LCD_FULLEXPANSION_ACTIVE 0x00001000L +#define ATOM_S3_LCD_EXPANSION_ASPEC_RATIO_ACTIVE 0x00002000L + +#define ATOM_S3_CRT1_CRTC_ACTIVE 0x00010000L +#define ATOM_S3_LCD1_CRTC_ACTIVE 0x00020000L +#define ATOM_S3_TV1_CRTC_ACTIVE 0x00040000L +#define ATOM_S3_DFP1_CRTC_ACTIVE 0x00080000L +#define ATOM_S3_CRT2_CRTC_ACTIVE 0x00100000L +#define ATOM_S3_LCD2_CRTC_ACTIVE 0x00200000L +#define ATOM_S3_DFP6_CRTC_ACTIVE 0x00400000L +#define ATOM_S3_DFP2_CRTC_ACTIVE 0x00800000L +#define ATOM_S3_CV_CRTC_ACTIVE 0x01000000L +#define ATOM_S3_DFP3_CRTC_ACTIVE 0x02000000L +#define ATOM_S3_DFP4_CRTC_ACTIVE 0x04000000L +#define ATOM_S3_DFP5_CRTC_ACTIVE 0x08000000L + +#define ATOM_S3_DEVICE_CRTC_ACTIVE_MASK 0x0FFF0000L +#define ATOM_S3_ASIC_GUI_ENGINE_HUNG 0x20000000L +//Below two definitions are not supported in pplib, but in the old powerplay in DAL +#define ATOM_S3_ALLOW_FAST_PWR_SWITCH 0x40000000L +#define ATOM_S3_RQST_GPU_USE_MIN_PWR 0x80000000L + +//Byte aligned definition for BIOS usage +#define ATOM_S3_CRT1_ACTIVEb0 0x01 +#define ATOM_S3_LCD1_ACTIVEb0 0x02 +#define ATOM_S3_TV1_ACTIVEb0 0x04 +#define ATOM_S3_DFP1_ACTIVEb0 0x08 +#define ATOM_S3_CRT2_ACTIVEb0 0x10 +#define ATOM_S3_LCD2_ACTIVEb0 0x20 +#define ATOM_S3_DFP6_ACTIVEb0 0x40 +#define ATOM_S3_DFP2_ACTIVEb0 0x80 +#define ATOM_S3_CV_ACTIVEb1 0x01 +#define ATOM_S3_DFP3_ACTIVEb1 0x02 +#define ATOM_S3_DFP4_ACTIVEb1 0x04 +#define ATOM_S3_DFP5_ACTIVEb1 0x08 + +#define ATOM_S3_ACTIVE_CRTC1w0 0xFFF + +#define ATOM_S3_CRT1_CRTC_ACTIVEb2 0x01 +#define ATOM_S3_LCD1_CRTC_ACTIVEb2 0x02 +#define ATOM_S3_TV1_CRTC_ACTIVEb2 0x04 +#define ATOM_S3_DFP1_CRTC_ACTIVEb2 0x08 +#define ATOM_S3_CRT2_CRTC_ACTIVEb2 0x10 +#define ATOM_S3_LCD2_CRTC_ACTIVEb2 0x20 +#define ATOM_S3_DFP6_CRTC_ACTIVEb2 0x40 +#define ATOM_S3_DFP2_CRTC_ACTIVEb2 0x80 +#define ATOM_S3_CV_CRTC_ACTIVEb3 0x01 +#define ATOM_S3_DFP3_CRTC_ACTIVEb3 0x02 +#define ATOM_S3_DFP4_CRTC_ACTIVEb3 0x04 +#define ATOM_S3_DFP5_CRTC_ACTIVEb3 0x08 + +#define ATOM_S3_ACTIVE_CRTC2w1 0xFFF + +// BIOS_4_SCRATCH Definition +#define ATOM_S4_LCD1_PANEL_ID_MASK 0x000000FFL +#define ATOM_S4_LCD1_REFRESH_MASK 0x0000FF00L +#define ATOM_S4_LCD1_REFRESH_SHIFT 8 + +//Byte aligned definition for BIOS usage +#define ATOM_S4_LCD1_PANEL_ID_MASKb0 0x0FF +#define ATOM_S4_LCD1_REFRESH_MASKb1 ATOM_S4_LCD1_PANEL_ID_MASKb0 +#define ATOM_S4_VRAM_INFO_MASKb2 ATOM_S4_LCD1_PANEL_ID_MASKb0 + +// BIOS_5_SCRATCH Definition, BIOS_5_SCRATCH is used by Firmware only !!!! +#define ATOM_S5_DOS_REQ_CRT1b0 0x01 +#define ATOM_S5_DOS_REQ_LCD1b0 0x02 +#define ATOM_S5_DOS_REQ_TV1b0 0x04 +#define ATOM_S5_DOS_REQ_DFP1b0 0x08 +#define ATOM_S5_DOS_REQ_CRT2b0 0x10 +#define ATOM_S5_DOS_REQ_LCD2b0 0x20 +#define ATOM_S5_DOS_REQ_DFP6b0 0x40 +#define ATOM_S5_DOS_REQ_DFP2b0 0x80 +#define ATOM_S5_DOS_REQ_CVb1 0x01 +#define ATOM_S5_DOS_REQ_DFP3b1 0x02 +#define ATOM_S5_DOS_REQ_DFP4b1 0x04 +#define ATOM_S5_DOS_REQ_DFP5b1 0x08 + +#define ATOM_S5_DOS_REQ_DEVICEw0 0x0FFF + +#define ATOM_S5_DOS_REQ_CRT1 0x0001 +#define ATOM_S5_DOS_REQ_LCD1 0x0002 +#define ATOM_S5_DOS_REQ_TV1 0x0004 +#define ATOM_S5_DOS_REQ_DFP1 0x0008 +#define ATOM_S5_DOS_REQ_CRT2 0x0010 +#define ATOM_S5_DOS_REQ_LCD2 0x0020 +#define ATOM_S5_DOS_REQ_DFP6 0x0040 +#define ATOM_S5_DOS_REQ_DFP2 0x0080 +#define ATOM_S5_DOS_REQ_CV 0x0100 +#define ATOM_S5_DOS_REQ_DFP3 0x0200 +#define ATOM_S5_DOS_REQ_DFP4 0x0400 +#define ATOM_S5_DOS_REQ_DFP5 0x0800 + +#define ATOM_S5_DOS_FORCE_CRT1b2 ATOM_S5_DOS_REQ_CRT1b0 +#define ATOM_S5_DOS_FORCE_TV1b2 ATOM_S5_DOS_REQ_TV1b0 +#define ATOM_S5_DOS_FORCE_CRT2b2 ATOM_S5_DOS_REQ_CRT2b0 +#define ATOM_S5_DOS_FORCE_CVb3 ATOM_S5_DOS_REQ_CVb1 +#define ATOM_S5_DOS_FORCE_DEVICEw1 (ATOM_S5_DOS_FORCE_CRT1b2+ATOM_S5_DOS_FORCE_TV1b2+ATOM_S5_DOS_FORCE_CRT2b2+\ + (ATOM_S5_DOS_FORCE_CVb3<<8)) + +// BIOS_6_SCRATCH Definition +#define ATOM_S6_DEVICE_CHANGE 0x00000001L +#define ATOM_S6_SCALER_CHANGE 0x00000002L +#define ATOM_S6_LID_CHANGE 0x00000004L +#define ATOM_S6_DOCKING_CHANGE 0x00000008L +#define ATOM_S6_ACC_MODE 0x00000010L +#define ATOM_S6_EXT_DESKTOP_MODE 0x00000020L +#define ATOM_S6_LID_STATE 0x00000040L +#define ATOM_S6_DOCK_STATE 0x00000080L +#define ATOM_S6_CRITICAL_STATE 0x00000100L +#define ATOM_S6_HW_I2C_BUSY_STATE 0x00000200L +#define ATOM_S6_THERMAL_STATE_CHANGE 0x00000400L +#define ATOM_S6_INTERRUPT_SET_BY_BIOS 0x00000800L +#define ATOM_S6_REQ_LCD_EXPANSION_FULL 0x00001000L //Normal expansion Request bit for LCD +#define ATOM_S6_REQ_LCD_EXPANSION_ASPEC_RATIO 0x00002000L //Aspect ratio expansion Request bit for LCD + +#define ATOM_S6_DISPLAY_STATE_CHANGE 0x00004000L //This bit is recycled when ATOM_BIOS_INFO_BIOS_SCRATCH6_SCL2_REDEFINE is set,previously it's SCL2_H_expansion +#define ATOM_S6_I2C_STATE_CHANGE 0x00008000L //This bit is recycled,when ATOM_BIOS_INFO_BIOS_SCRATCH6_SCL2_REDEFINE is set,previously it's SCL2_V_expansion + +#define ATOM_S6_ACC_REQ_CRT1 0x00010000L +#define ATOM_S6_ACC_REQ_LCD1 0x00020000L +#define ATOM_S6_ACC_REQ_TV1 0x00040000L +#define ATOM_S6_ACC_REQ_DFP1 0x00080000L +#define ATOM_S6_ACC_REQ_CRT2 0x00100000L +#define ATOM_S6_ACC_REQ_LCD2 0x00200000L +#define ATOM_S6_ACC_REQ_DFP6 0x00400000L +#define ATOM_S6_ACC_REQ_DFP2 0x00800000L +#define ATOM_S6_ACC_REQ_CV 0x01000000L +#define ATOM_S6_ACC_REQ_DFP3 0x02000000L +#define ATOM_S6_ACC_REQ_DFP4 0x04000000L +#define ATOM_S6_ACC_REQ_DFP5 0x08000000L + +#define ATOM_S6_ACC_REQ_MASK 0x0FFF0000L +#define ATOM_S6_SYSTEM_POWER_MODE_CHANGE 0x10000000L +#define ATOM_S6_ACC_BLOCK_DISPLAY_SWITCH 0x20000000L +#define ATOM_S6_VRI_BRIGHTNESS_CHANGE 0x40000000L +#define ATOM_S6_CONFIG_DISPLAY_CHANGE_MASK 0x80000000L + +//Byte aligned definition for BIOS usage +#define ATOM_S6_DEVICE_CHANGEb0 0x01 +#define ATOM_S6_SCALER_CHANGEb0 0x02 +#define ATOM_S6_LID_CHANGEb0 0x04 +#define ATOM_S6_DOCKING_CHANGEb0 0x08 +#define ATOM_S6_ACC_MODEb0 0x10 +#define ATOM_S6_EXT_DESKTOP_MODEb0 0x20 +#define ATOM_S6_LID_STATEb0 0x40 +#define ATOM_S6_DOCK_STATEb0 0x80 +#define ATOM_S6_CRITICAL_STATEb1 0x01 +#define ATOM_S6_HW_I2C_BUSY_STATEb1 0x02 +#define ATOM_S6_THERMAL_STATE_CHANGEb1 0x04 +#define ATOM_S6_INTERRUPT_SET_BY_BIOSb1 0x08 +#define ATOM_S6_REQ_LCD_EXPANSION_FULLb1 0x10 +#define ATOM_S6_REQ_LCD_EXPANSION_ASPEC_RATIOb1 0x20 + +#define ATOM_S6_ACC_REQ_CRT1b2 0x01 +#define ATOM_S6_ACC_REQ_LCD1b2 0x02 +#define ATOM_S6_ACC_REQ_TV1b2 0x04 +#define ATOM_S6_ACC_REQ_DFP1b2 0x08 +#define ATOM_S6_ACC_REQ_CRT2b2 0x10 +#define ATOM_S6_ACC_REQ_LCD2b2 0x20 +#define ATOM_S6_ACC_REQ_DFP6b2 0x40 +#define ATOM_S6_ACC_REQ_DFP2b2 0x80 +#define ATOM_S6_ACC_REQ_CVb3 0x01 +#define ATOM_S6_ACC_REQ_DFP3b3 0x02 +#define ATOM_S6_ACC_REQ_DFP4b3 0x04 +#define ATOM_S6_ACC_REQ_DFP5b3 0x08 + +#define ATOM_S6_ACC_REQ_DEVICEw1 ATOM_S5_DOS_REQ_DEVICEw0 +#define ATOM_S6_SYSTEM_POWER_MODE_CHANGEb3 0x10 +#define ATOM_S6_ACC_BLOCK_DISPLAY_SWITCHb3 0x20 +#define ATOM_S6_VRI_BRIGHTNESS_CHANGEb3 0x40 +#define ATOM_S6_CONFIG_DISPLAY_CHANGEb3 0x80 + +#define ATOM_S6_DEVICE_CHANGE_SHIFT 0 +#define ATOM_S6_SCALER_CHANGE_SHIFT 1 +#define ATOM_S6_LID_CHANGE_SHIFT 2 +#define ATOM_S6_DOCKING_CHANGE_SHIFT 3 +#define ATOM_S6_ACC_MODE_SHIFT 4 +#define ATOM_S6_EXT_DESKTOP_MODE_SHIFT 5 +#define ATOM_S6_LID_STATE_SHIFT 6 +#define ATOM_S6_DOCK_STATE_SHIFT 7 +#define ATOM_S6_CRITICAL_STATE_SHIFT 8 +#define ATOM_S6_HW_I2C_BUSY_STATE_SHIFT 9 +#define ATOM_S6_THERMAL_STATE_CHANGE_SHIFT 10 +#define ATOM_S6_INTERRUPT_SET_BY_BIOS_SHIFT 11 +#define ATOM_S6_REQ_SCALER_SHIFT 12 +#define ATOM_S6_REQ_SCALER_ARATIO_SHIFT 13 +#define ATOM_S6_DISPLAY_STATE_CHANGE_SHIFT 14 +#define ATOM_S6_I2C_STATE_CHANGE_SHIFT 15 +#define ATOM_S6_SYSTEM_POWER_MODE_CHANGE_SHIFT 28 +#define ATOM_S6_ACC_BLOCK_DISPLAY_SWITCH_SHIFT 29 +#define ATOM_S6_VRI_BRIGHTNESS_CHANGE_SHIFT 30 +#define ATOM_S6_CONFIG_DISPLAY_CHANGE_SHIFT 31 + +// BIOS_7_SCRATCH Definition, BIOS_7_SCRATCH is used by Firmware only !!!! +#define ATOM_S7_DOS_MODE_TYPEb0 0x03 +#define ATOM_S7_DOS_MODE_VGAb0 0x00 +#define ATOM_S7_DOS_MODE_VESAb0 0x01 +#define ATOM_S7_DOS_MODE_EXTb0 0x02 +#define ATOM_S7_DOS_MODE_PIXEL_DEPTHb0 0x0C +#define ATOM_S7_DOS_MODE_PIXEL_FORMATb0 0xF0 +#define ATOM_S7_DOS_8BIT_DAC_ENb1 0x01 +#define ATOM_S7_DOS_MODE_NUMBERw1 0x0FFFF + +#define ATOM_S7_DOS_8BIT_DAC_EN_SHIFT 8 + +// BIOS_8_SCRATCH Definition +#define ATOM_S8_I2C_CHANNEL_BUSY_MASK 0x00000FFFF +#define ATOM_S8_I2C_HW_ENGINE_BUSY_MASK 0x0FFFF0000 + +#define ATOM_S8_I2C_CHANNEL_BUSY_SHIFT 0 +#define ATOM_S8_I2C_ENGINE_BUSY_SHIFT 16 + +// BIOS_9_SCRATCH Definition +#ifndef ATOM_S9_I2C_CHANNEL_COMPLETED_MASK +#define ATOM_S9_I2C_CHANNEL_COMPLETED_MASK 0x0000FFFF +#endif +#ifndef ATOM_S9_I2C_CHANNEL_ABORTED_MASK +#define ATOM_S9_I2C_CHANNEL_ABORTED_MASK 0xFFFF0000 +#endif +#ifndef ATOM_S9_I2C_CHANNEL_COMPLETED_SHIFT +#define ATOM_S9_I2C_CHANNEL_COMPLETED_SHIFT 0 +#endif +#ifndef ATOM_S9_I2C_CHANNEL_ABORTED_SHIFT +#define ATOM_S9_I2C_CHANNEL_ABORTED_SHIFT 16 +#endif + + +#define ATOM_FLAG_SET 0x20 +#define ATOM_FLAG_CLEAR 0 +#define CLEAR_ATOM_S6_ACC_MODE ((ATOM_ACC_CHANGE_INFO_DEF << 8 )|ATOM_S6_ACC_MODE_SHIFT | ATOM_FLAG_CLEAR) +#define SET_ATOM_S6_DEVICE_CHANGE ((ATOM_ACC_CHANGE_INFO_DEF << 8 )|ATOM_S6_DEVICE_CHANGE_SHIFT | ATOM_FLAG_SET) +#define SET_ATOM_S6_VRI_BRIGHTNESS_CHANGE ((ATOM_ACC_CHANGE_INFO_DEF << 8 )|ATOM_S6_VRI_BRIGHTNESS_CHANGE_SHIFT | ATOM_FLAG_SET) +#define SET_ATOM_S6_SCALER_CHANGE ((ATOM_ACC_CHANGE_INFO_DEF << 8 )|ATOM_S6_SCALER_CHANGE_SHIFT | ATOM_FLAG_SET) +#define SET_ATOM_S6_LID_CHANGE ((ATOM_ACC_CHANGE_INFO_DEF << 8 )|ATOM_S6_LID_CHANGE_SHIFT | ATOM_FLAG_SET) + +#define SET_ATOM_S6_LID_STATE ((ATOM_ACC_CHANGE_INFO_DEF << 8 )|ATOM_S6_LID_STATE_SHIFT | ATOM_FLAG_SET) +#define CLEAR_ATOM_S6_LID_STATE ((ATOM_ACC_CHANGE_INFO_DEF << 8 )|ATOM_S6_LID_STATE_SHIFT | ATOM_FLAG_CLEAR) + +#define SET_ATOM_S6_DOCK_CHANGE ((ATOM_ACC_CHANGE_INFO_DEF << 8 )|ATOM_S6_DOCKING_CHANGE_SHIFT | ATOM_FLAG_SET) +#define SET_ATOM_S6_DOCK_STATE ((ATOM_ACC_CHANGE_INFO_DEF << 8 )|ATOM_S6_DOCK_STATE_SHIFT | ATOM_FLAG_SET) +#define CLEAR_ATOM_S6_DOCK_STATE ((ATOM_ACC_CHANGE_INFO_DEF << 8 )|ATOM_S6_DOCK_STATE_SHIFT | ATOM_FLAG_CLEAR) + +#define SET_ATOM_S6_THERMAL_STATE_CHANGE ((ATOM_ACC_CHANGE_INFO_DEF << 8 )|ATOM_S6_THERMAL_STATE_CHANGE_SHIFT | ATOM_FLAG_SET) +#define SET_ATOM_S6_SYSTEM_POWER_MODE_CHANGE ((ATOM_ACC_CHANGE_INFO_DEF << 8 )|ATOM_S6_SYSTEM_POWER_MODE_CHANGE_SHIFT | ATOM_FLAG_SET) +#define SET_ATOM_S6_INTERRUPT_SET_BY_BIOS ((ATOM_ACC_CHANGE_INFO_DEF << 8 )|ATOM_S6_INTERRUPT_SET_BY_BIOS_SHIFT | ATOM_FLAG_SET) + +#define SET_ATOM_S6_CRITICAL_STATE ((ATOM_ACC_CHANGE_INFO_DEF << 8 )|ATOM_S6_CRITICAL_STATE_SHIFT | ATOM_FLAG_SET) +#define CLEAR_ATOM_S6_CRITICAL_STATE ((ATOM_ACC_CHANGE_INFO_DEF << 8 )|ATOM_S6_CRITICAL_STATE_SHIFT | ATOM_FLAG_CLEAR) + +#define SET_ATOM_S6_REQ_SCALER ((ATOM_ACC_CHANGE_INFO_DEF << 8 )|ATOM_S6_REQ_SCALER_SHIFT | ATOM_FLAG_SET) +#define CLEAR_ATOM_S6_REQ_SCALER ((ATOM_ACC_CHANGE_INFO_DEF << 8 )|ATOM_S6_REQ_SCALER_SHIFT | ATOM_FLAG_CLEAR ) + +#define SET_ATOM_S6_REQ_SCALER_ARATIO ((ATOM_ACC_CHANGE_INFO_DEF << 8 )|ATOM_S6_REQ_SCALER_ARATIO_SHIFT | ATOM_FLAG_SET ) +#define CLEAR_ATOM_S6_REQ_SCALER_ARATIO ((ATOM_ACC_CHANGE_INFO_DEF << 8 )|ATOM_S6_REQ_SCALER_ARATIO_SHIFT | ATOM_FLAG_CLEAR ) + +#define SET_ATOM_S6_I2C_STATE_CHANGE ((ATOM_ACC_CHANGE_INFO_DEF << 8 )|ATOM_S6_I2C_STATE_CHANGE_SHIFT | ATOM_FLAG_SET ) + +#define SET_ATOM_S6_DISPLAY_STATE_CHANGE ((ATOM_ACC_CHANGE_INFO_DEF << 8 )|ATOM_S6_DISPLAY_STATE_CHANGE_SHIFT | ATOM_FLAG_SET ) + +#define SET_ATOM_S6_DEVICE_RECONFIG ((ATOM_ACC_CHANGE_INFO_DEF << 8 )|ATOM_S6_CONFIG_DISPLAY_CHANGE_SHIFT | ATOM_FLAG_SET) +#define CLEAR_ATOM_S0_LCD1 ((ATOM_DEVICE_CONNECT_INFO_DEF << 8 )| ATOM_S0_LCD1_SHIFT | ATOM_FLAG_CLEAR ) +#define SET_ATOM_S7_DOS_8BIT_DAC_EN ((ATOM_DOS_MODE_INFO_DEF << 8 )|ATOM_S7_DOS_8BIT_DAC_EN_SHIFT | ATOM_FLAG_SET ) +#define CLEAR_ATOM_S7_DOS_8BIT_DAC_EN ((ATOM_DOS_MODE_INFO_DEF << 8 )|ATOM_S7_DOS_8BIT_DAC_EN_SHIFT | ATOM_FLAG_CLEAR ) + +/****************************************************************************/ +//Portion II: Definitinos only used in Driver +/****************************************************************************/ + +// Macros used by driver +#ifdef __cplusplus +#define GetIndexIntoMasterTable(MasterOrData, FieldName) ((reinterpret_cast(&(static_cast(0))->FieldName)-static_cast(0))/sizeof(USHORT)) + +#define GET_COMMAND_TABLE_COMMANDSET_REVISION(TABLE_HEADER_OFFSET) (((static_cast(TABLE_HEADER_OFFSET))->ucTableFormatRevision )&0x3F) +#define GET_COMMAND_TABLE_PARAMETER_REVISION(TABLE_HEADER_OFFSET) (((static_cast(TABLE_HEADER_OFFSET))->ucTableContentRevision)&0x3F) +#else // not __cplusplus +#define GetIndexIntoMasterTable(MasterOrData, FieldName) (((char*)(&((ATOM_MASTER_LIST_OF_##MasterOrData##_TABLES*)0)->FieldName)-(char*)0)/sizeof(USHORT)) + +#define GET_COMMAND_TABLE_COMMANDSET_REVISION(TABLE_HEADER_OFFSET) ((((ATOM_COMMON_TABLE_HEADER*)TABLE_HEADER_OFFSET)->ucTableFormatRevision)&0x3F) +#define GET_COMMAND_TABLE_PARAMETER_REVISION(TABLE_HEADER_OFFSET) ((((ATOM_COMMON_TABLE_HEADER*)TABLE_HEADER_OFFSET)->ucTableContentRevision)&0x3F) +#endif // __cplusplus + +#define GET_DATA_TABLE_MAJOR_REVISION GET_COMMAND_TABLE_COMMANDSET_REVISION +#define GET_DATA_TABLE_MINOR_REVISION GET_COMMAND_TABLE_PARAMETER_REVISION + +/****************************************************************************/ +//Portion III: Definitinos only used in VBIOS +/****************************************************************************/ +#define ATOM_DAC_SRC 0x80 +#define ATOM_SRC_DAC1 0 +#define ATOM_SRC_DAC2 0x80 + +typedef struct _MEMORY_PLLINIT_PARAMETERS +{ + ULONG ulTargetMemoryClock; //In 10Khz unit + UCHAR ucAction; //not define yet + UCHAR ucFbDiv_Hi; //Fbdiv Hi byte + UCHAR ucFbDiv; //FB value + UCHAR ucPostDiv; //Post div +}MEMORY_PLLINIT_PARAMETERS; + +#define MEMORY_PLLINIT_PS_ALLOCATION MEMORY_PLLINIT_PARAMETERS + + +#define GPIO_PIN_WRITE 0x01 +#define GPIO_PIN_READ 0x00 + +typedef struct _GPIO_PIN_CONTROL_PARAMETERS +{ + UCHAR ucGPIO_ID; //return value, read from GPIO pins + UCHAR ucGPIOBitShift; //define which bit in uGPIOBitVal need to be update + UCHAR ucGPIOBitVal; //Set/Reset corresponding bit defined in ucGPIOBitMask + UCHAR ucAction; //=GPIO_PIN_WRITE: Read; =GPIO_PIN_READ: Write +}GPIO_PIN_CONTROL_PARAMETERS; + +typedef struct _ENABLE_SCALER_PARAMETERS +{ + UCHAR ucScaler; // ATOM_SCALER1, ATOM_SCALER2 + UCHAR ucEnable; // ATOM_SCALER_DISABLE or ATOM_SCALER_CENTER or ATOM_SCALER_EXPANSION + UCHAR ucTVStandard; // + UCHAR ucPadding[1]; +}ENABLE_SCALER_PARAMETERS; +#define ENABLE_SCALER_PS_ALLOCATION ENABLE_SCALER_PARAMETERS + +//ucEnable: +#define SCALER_BYPASS_AUTO_CENTER_NO_REPLICATION 0 +#define SCALER_BYPASS_AUTO_CENTER_AUTO_REPLICATION 1 +#define SCALER_ENABLE_2TAP_ALPHA_MODE 2 +#define SCALER_ENABLE_MULTITAP_MODE 3 + +typedef struct _ENABLE_HARDWARE_ICON_CURSOR_PARAMETERS +{ + ULONG usHWIconHorzVertPosn; // Hardware Icon Vertical position + UCHAR ucHWIconVertOffset; // Hardware Icon Vertical offset + UCHAR ucHWIconHorzOffset; // Hardware Icon Horizontal offset + UCHAR ucSelection; // ATOM_CURSOR1 or ATOM_ICON1 or ATOM_CURSOR2 or ATOM_ICON2 + UCHAR ucEnable; // ATOM_ENABLE or ATOM_DISABLE +}ENABLE_HARDWARE_ICON_CURSOR_PARAMETERS; + +typedef struct _ENABLE_HARDWARE_ICON_CURSOR_PS_ALLOCATION +{ + ENABLE_HARDWARE_ICON_CURSOR_PARAMETERS sEnableIcon; + ENABLE_CRTC_PARAMETERS sReserved; +}ENABLE_HARDWARE_ICON_CURSOR_PS_ALLOCATION; + +typedef struct _ENABLE_GRAPH_SURFACE_PARAMETERS +{ + USHORT usHight; // Image Hight + USHORT usWidth; // Image Width + UCHAR ucSurface; // Surface 1 or 2 + UCHAR ucPadding[3]; +}ENABLE_GRAPH_SURFACE_PARAMETERS; + +typedef struct _ENABLE_GRAPH_SURFACE_PARAMETERS_V1_2 +{ + USHORT usHight; // Image Hight + USHORT usWidth; // Image Width + UCHAR ucSurface; // Surface 1 or 2 + UCHAR ucEnable; // ATOM_ENABLE or ATOM_DISABLE + UCHAR ucPadding[2]; +}ENABLE_GRAPH_SURFACE_PARAMETERS_V1_2; + +typedef struct _ENABLE_GRAPH_SURFACE_PARAMETERS_V1_3 +{ + USHORT usHight; // Image Hight + USHORT usWidth; // Image Width + UCHAR ucSurface; // Surface 1 or 2 + UCHAR ucEnable; // ATOM_ENABLE or ATOM_DISABLE + USHORT usDeviceId; // Active Device Id for this surface. If no device, set to 0. +}ENABLE_GRAPH_SURFACE_PARAMETERS_V1_3; + +typedef struct _ENABLE_GRAPH_SURFACE_PS_ALLOCATION +{ + ENABLE_GRAPH_SURFACE_PARAMETERS sSetSurface; + ENABLE_YUV_PS_ALLOCATION sReserved; // Don't set this one +}ENABLE_GRAPH_SURFACE_PS_ALLOCATION; + +typedef struct _MEMORY_CLEAN_UP_PARAMETERS +{ + USHORT usMemoryStart; //in 8Kb boundary, offset from memory base address + USHORT usMemorySize; //8Kb blocks aligned +}MEMORY_CLEAN_UP_PARAMETERS; +#define MEMORY_CLEAN_UP_PS_ALLOCATION MEMORY_CLEAN_UP_PARAMETERS + +typedef struct _GET_DISPLAY_SURFACE_SIZE_PARAMETERS +{ + USHORT usX_Size; //When use as input parameter, usX_Size indicates which CRTC + USHORT usY_Size; +}GET_DISPLAY_SURFACE_SIZE_PARAMETERS; + +typedef struct _INDIRECT_IO_ACCESS +{ + ATOM_COMMON_TABLE_HEADER sHeader; + UCHAR IOAccessSequence[256]; +} INDIRECT_IO_ACCESS; + +#define INDIRECT_READ 0x00 +#define INDIRECT_WRITE 0x80 + +#define INDIRECT_IO_MM 0 +#define INDIRECT_IO_PLL 1 +#define INDIRECT_IO_MC 2 +#define INDIRECT_IO_PCIE 3 +#define INDIRECT_IO_PCIEP 4 +#define INDIRECT_IO_NBMISC 5 + +#define INDIRECT_IO_PLL_READ INDIRECT_IO_PLL | INDIRECT_READ +#define INDIRECT_IO_PLL_WRITE INDIRECT_IO_PLL | INDIRECT_WRITE +#define INDIRECT_IO_MC_READ INDIRECT_IO_MC | INDIRECT_READ +#define INDIRECT_IO_MC_WRITE INDIRECT_IO_MC | INDIRECT_WRITE +#define INDIRECT_IO_PCIE_READ INDIRECT_IO_PCIE | INDIRECT_READ +#define INDIRECT_IO_PCIE_WRITE INDIRECT_IO_PCIE | INDIRECT_WRITE +#define INDIRECT_IO_PCIEP_READ INDIRECT_IO_PCIEP | INDIRECT_READ +#define INDIRECT_IO_PCIEP_WRITE INDIRECT_IO_PCIEP | INDIRECT_WRITE +#define INDIRECT_IO_NBMISC_READ INDIRECT_IO_NBMISC | INDIRECT_READ +#define INDIRECT_IO_NBMISC_WRITE INDIRECT_IO_NBMISC | INDIRECT_WRITE + +typedef struct _ATOM_OEM_INFO +{ + ATOM_COMMON_TABLE_HEADER sHeader; + ATOM_I2C_ID_CONFIG_ACCESS sucI2cId; +}ATOM_OEM_INFO; + +typedef struct _ATOM_TV_MODE +{ + UCHAR ucVMode_Num; //Video mode number + UCHAR ucTV_Mode_Num; //Internal TV mode number +}ATOM_TV_MODE; + +typedef struct _ATOM_BIOS_INT_TVSTD_MODE +{ + ATOM_COMMON_TABLE_HEADER sHeader; + USHORT usTV_Mode_LUT_Offset; // Pointer to standard to internal number conversion table + USHORT usTV_FIFO_Offset; // Pointer to FIFO entry table + USHORT usNTSC_Tbl_Offset; // Pointer to SDTV_Mode_NTSC table + USHORT usPAL_Tbl_Offset; // Pointer to SDTV_Mode_PAL table + USHORT usCV_Tbl_Offset; // Pointer to SDTV_Mode_PAL table +}ATOM_BIOS_INT_TVSTD_MODE; + + +typedef struct _ATOM_TV_MODE_SCALER_PTR +{ + USHORT ucFilter0_Offset; //Pointer to filter format 0 coefficients + USHORT usFilter1_Offset; //Pointer to filter format 0 coefficients + UCHAR ucTV_Mode_Num; +}ATOM_TV_MODE_SCALER_PTR; + +typedef struct _ATOM_STANDARD_VESA_TIMING +{ + ATOM_COMMON_TABLE_HEADER sHeader; + ATOM_DTD_FORMAT aModeTimings[16]; // 16 is not the real array number, just for initial allocation +}ATOM_STANDARD_VESA_TIMING; + + +typedef struct _ATOM_STD_FORMAT +{ + USHORT usSTD_HDisp; + USHORT usSTD_VDisp; + USHORT usSTD_RefreshRate; + USHORT usReserved; +}ATOM_STD_FORMAT; + +typedef struct _ATOM_VESA_TO_EXTENDED_MODE +{ + USHORT usVESA_ModeNumber; + USHORT usExtendedModeNumber; +}ATOM_VESA_TO_EXTENDED_MODE; + +typedef struct _ATOM_VESA_TO_INTENAL_MODE_LUT +{ + ATOM_COMMON_TABLE_HEADER sHeader; + ATOM_VESA_TO_EXTENDED_MODE asVESA_ToExtendedModeInfo[76]; +}ATOM_VESA_TO_INTENAL_MODE_LUT; + +/*************** ATOM Memory Related Data Structure ***********************/ +typedef struct _ATOM_MEMORY_VENDOR_BLOCK{ + UCHAR ucMemoryType; + UCHAR ucMemoryVendor; + UCHAR ucAdjMCId; + UCHAR ucDynClkId; + ULONG ulDllResetClkRange; +}ATOM_MEMORY_VENDOR_BLOCK; + + +typedef struct _ATOM_MEMORY_SETTING_ID_CONFIG{ +#if ATOM_BIG_ENDIAN + ULONG ucMemBlkId:8; + ULONG ulMemClockRange:24; +#else + ULONG ulMemClockRange:24; + ULONG ucMemBlkId:8; +#endif +}ATOM_MEMORY_SETTING_ID_CONFIG; + +typedef union _ATOM_MEMORY_SETTING_ID_CONFIG_ACCESS +{ + ATOM_MEMORY_SETTING_ID_CONFIG slAccess; + ULONG ulAccess; +}ATOM_MEMORY_SETTING_ID_CONFIG_ACCESS; + + +typedef struct _ATOM_MEMORY_SETTING_DATA_BLOCK{ + ATOM_MEMORY_SETTING_ID_CONFIG_ACCESS ulMemoryID; + ULONG aulMemData[1]; +}ATOM_MEMORY_SETTING_DATA_BLOCK; + + +typedef struct _ATOM_INIT_REG_INDEX_FORMAT{ + USHORT usRegIndex; // MC register index + UCHAR ucPreRegDataLength; // offset in ATOM_INIT_REG_DATA_BLOCK.saRegDataBuf +}ATOM_INIT_REG_INDEX_FORMAT; + + +typedef struct _ATOM_INIT_REG_BLOCK{ + USHORT usRegIndexTblSize; //size of asRegIndexBuf + USHORT usRegDataBlkSize; //size of ATOM_MEMORY_SETTING_DATA_BLOCK + ATOM_INIT_REG_INDEX_FORMAT asRegIndexBuf[1]; + ATOM_MEMORY_SETTING_DATA_BLOCK asRegDataBuf[1]; +}ATOM_INIT_REG_BLOCK; + +#define END_OF_REG_INDEX_BLOCK 0x0ffff +#define END_OF_REG_DATA_BLOCK 0x00000000 +#define ATOM_INIT_REG_MASK_FLAG 0x80 +#define CLOCK_RANGE_HIGHEST 0x00ffffff + +#define VALUE_DWORD SIZEOF ULONG +#define VALUE_SAME_AS_ABOVE 0 +#define VALUE_MASK_DWORD 0x84 + +#define INDEX_ACCESS_RANGE_BEGIN (VALUE_DWORD + 1) +#define INDEX_ACCESS_RANGE_END (INDEX_ACCESS_RANGE_BEGIN + 1) +#define VALUE_INDEX_ACCESS_SINGLE (INDEX_ACCESS_RANGE_END + 1) +//#define ACCESS_MCIODEBUGIND 0x40 //defined in BIOS code +#define ACCESS_PLACEHOLDER 0x80 + +typedef struct _ATOM_MC_INIT_PARAM_TABLE +{ + ATOM_COMMON_TABLE_HEADER sHeader; + USHORT usAdjustARB_SEQDataOffset; + USHORT usMCInitMemTypeTblOffset; + USHORT usMCInitCommonTblOffset; + USHORT usMCInitPowerDownTblOffset; + ULONG ulARB_SEQDataBuf[32]; + ATOM_INIT_REG_BLOCK asMCInitMemType; + ATOM_INIT_REG_BLOCK asMCInitCommon; +}ATOM_MC_INIT_PARAM_TABLE; + + +#define _4Mx16 0x2 +#define _4Mx32 0x3 +#define _8Mx16 0x12 +#define _8Mx32 0x13 +#define _16Mx16 0x22 +#define _16Mx32 0x23 +#define _32Mx16 0x32 +#define _32Mx32 0x33 +#define _64Mx8 0x41 +#define _64Mx16 0x42 +#define _64Mx32 0x43 +#define _128Mx8 0x51 +#define _128Mx16 0x52 +#define _256Mx8 0x61 + +#define SAMSUNG 0x1 +#define INFINEON 0x2 +#define ELPIDA 0x3 +#define ETRON 0x4 +#define NANYA 0x5 +#define HYNIX 0x6 +#define MOSEL 0x7 +#define WINBOND 0x8 +#define ESMT 0x9 +#define MICRON 0xF + +#define QIMONDA INFINEON +#define PROMOS MOSEL +#define KRETON INFINEON +#define ELIXIR NANYA + +/////////////Support for GDDR5 MC uCode to reside in upper 64K of ROM///////////// + +#define UCODE_ROM_START_ADDRESS 0x1b800 +#define UCODE_SIGNATURE 0x4375434d // 'MCuC' - MC uCode + +//uCode block header for reference + +typedef struct _MCuCodeHeader +{ + ULONG ulSignature; + UCHAR ucRevision; + UCHAR ucChecksum; + UCHAR ucReserved1; + UCHAR ucReserved2; + USHORT usParametersLength; + USHORT usUCodeLength; + USHORT usReserved1; + USHORT usReserved2; +} MCuCodeHeader; + +////////////////////////////////////////////////////////////////////////////////// + +#define ATOM_MAX_NUMBER_OF_VRAM_MODULE 16 + +#define ATOM_VRAM_MODULE_MEMORY_VENDOR_ID_MASK 0xF +typedef struct _ATOM_VRAM_MODULE_V1 +{ + ULONG ulReserved; + USHORT usEMRSValue; + USHORT usMRSValue; + USHORT usReserved; + UCHAR ucExtMemoryID; // An external indicator (by hardcode, callback or pin) to tell what is the current memory module + UCHAR ucMemoryType; // [7:4]=0x1:DDR1;=0x2:DDR2;=0x3:DDR3;=0x4:DDR4;[3:0] reserved; + UCHAR ucMemoryVenderID; // Predefined,never change across designs or memory type/vender + UCHAR ucMemoryDeviceCfg; // [7:4]=0x0:4M;=0x1:8M;=0x2:16M;0x3:32M....[3:0]=0x0:x4;=0x1:x8;=0x2:x16;=0x3:x32... + UCHAR ucRow; // Number of Row,in power of 2; + UCHAR ucColumn; // Number of Column,in power of 2; + UCHAR ucBank; // Nunber of Bank; + UCHAR ucRank; // Number of Rank, in power of 2 + UCHAR ucChannelNum; // Number of channel; + UCHAR ucChannelConfig; // [3:0]=Indication of what channel combination;[4:7]=Channel bit width, in number of 2 + UCHAR ucDefaultMVDDQ_ID; // Default MVDDQ setting for this memory block, ID linking to MVDDQ info table to find real set-up data; + UCHAR ucDefaultMVDDC_ID; // Default MVDDC setting for this memory block, ID linking to MVDDC info table to find real set-up data; + UCHAR ucReserved[2]; +}ATOM_VRAM_MODULE_V1; + + +typedef struct _ATOM_VRAM_MODULE_V2 +{ + ULONG ulReserved; + ULONG ulFlags; // To enable/disable functionalities based on memory type + ULONG ulEngineClock; // Override of default engine clock for particular memory type + ULONG ulMemoryClock; // Override of default memory clock for particular memory type + USHORT usEMRS2Value; // EMRS2 Value is used for GDDR2 and GDDR4 memory type + USHORT usEMRS3Value; // EMRS3 Value is used for GDDR2 and GDDR4 memory type + USHORT usEMRSValue; + USHORT usMRSValue; + USHORT usReserved; + UCHAR ucExtMemoryID; // An external indicator (by hardcode, callback or pin) to tell what is the current memory module + UCHAR ucMemoryType; // [7:4]=0x1:DDR1;=0x2:DDR2;=0x3:DDR3;=0x4:DDR4;[3:0] - must not be used for now; + UCHAR ucMemoryVenderID; // Predefined,never change across designs or memory type/vender. If not predefined, vendor detection table gets executed + UCHAR ucMemoryDeviceCfg; // [7:4]=0x0:4M;=0x1:8M;=0x2:16M;0x3:32M....[3:0]=0x0:x4;=0x1:x8;=0x2:x16;=0x3:x32... + UCHAR ucRow; // Number of Row,in power of 2; + UCHAR ucColumn; // Number of Column,in power of 2; + UCHAR ucBank; // Nunber of Bank; + UCHAR ucRank; // Number of Rank, in power of 2 + UCHAR ucChannelNum; // Number of channel; + UCHAR ucChannelConfig; // [3:0]=Indication of what channel combination;[4:7]=Channel bit width, in number of 2 + UCHAR ucDefaultMVDDQ_ID; // Default MVDDQ setting for this memory block, ID linking to MVDDQ info table to find real set-up data; + UCHAR ucDefaultMVDDC_ID; // Default MVDDC setting for this memory block, ID linking to MVDDC info table to find real set-up data; + UCHAR ucRefreshRateFactor; + UCHAR ucReserved[3]; +}ATOM_VRAM_MODULE_V2; + + +typedef struct _ATOM_MEMORY_TIMING_FORMAT +{ + ULONG ulClkRange; // memory clock in 10kHz unit, when target memory clock is below this clock, use this memory timing + union{ + USHORT usMRS; // mode register + USHORT usDDR3_MR0; + }; + union{ + USHORT usEMRS; // extended mode register + USHORT usDDR3_MR1; + }; + UCHAR ucCL; // CAS latency + UCHAR ucWL; // WRITE Latency + UCHAR uctRAS; // tRAS + UCHAR uctRC; // tRC + UCHAR uctRFC; // tRFC + UCHAR uctRCDR; // tRCDR + UCHAR uctRCDW; // tRCDW + UCHAR uctRP; // tRP + UCHAR uctRRD; // tRRD + UCHAR uctWR; // tWR + UCHAR uctWTR; // tWTR + UCHAR uctPDIX; // tPDIX + UCHAR uctFAW; // tFAW + UCHAR uctAOND; // tAOND + union + { + struct { + UCHAR ucflag; // flag to control memory timing calculation. bit0= control EMRS2 Infineon + UCHAR ucReserved; + }; + USHORT usDDR3_MR2; + }; +}ATOM_MEMORY_TIMING_FORMAT; + + +typedef struct _ATOM_MEMORY_TIMING_FORMAT_V1 +{ + ULONG ulClkRange; // memory clock in 10kHz unit, when target memory clock is below this clock, use this memory timing + USHORT usMRS; // mode register + USHORT usEMRS; // extended mode register + UCHAR ucCL; // CAS latency + UCHAR ucWL; // WRITE Latency + UCHAR uctRAS; // tRAS + UCHAR uctRC; // tRC + UCHAR uctRFC; // tRFC + UCHAR uctRCDR; // tRCDR + UCHAR uctRCDW; // tRCDW + UCHAR uctRP; // tRP + UCHAR uctRRD; // tRRD + UCHAR uctWR; // tWR + UCHAR uctWTR; // tWTR + UCHAR uctPDIX; // tPDIX + UCHAR uctFAW; // tFAW + UCHAR uctAOND; // tAOND + UCHAR ucflag; // flag to control memory timing calculation. bit0= control EMRS2 Infineon +////////////////////////////////////GDDR parameters/////////////////////////////////// + UCHAR uctCCDL; // + UCHAR uctCRCRL; // + UCHAR uctCRCWL; // + UCHAR uctCKE; // + UCHAR uctCKRSE; // + UCHAR uctCKRSX; // + UCHAR uctFAW32; // + UCHAR ucMR5lo; // + UCHAR ucMR5hi; // + UCHAR ucTerminator; +}ATOM_MEMORY_TIMING_FORMAT_V1; + +typedef struct _ATOM_MEMORY_TIMING_FORMAT_V2 +{ + ULONG ulClkRange; // memory clock in 10kHz unit, when target memory clock is below this clock, use this memory timing + USHORT usMRS; // mode register + USHORT usEMRS; // extended mode register + UCHAR ucCL; // CAS latency + UCHAR ucWL; // WRITE Latency + UCHAR uctRAS; // tRAS + UCHAR uctRC; // tRC + UCHAR uctRFC; // tRFC + UCHAR uctRCDR; // tRCDR + UCHAR uctRCDW; // tRCDW + UCHAR uctRP; // tRP + UCHAR uctRRD; // tRRD + UCHAR uctWR; // tWR + UCHAR uctWTR; // tWTR + UCHAR uctPDIX; // tPDIX + UCHAR uctFAW; // tFAW + UCHAR uctAOND; // tAOND + UCHAR ucflag; // flag to control memory timing calculation. bit0= control EMRS2 Infineon +////////////////////////////////////GDDR parameters/////////////////////////////////// + UCHAR uctCCDL; // + UCHAR uctCRCRL; // + UCHAR uctCRCWL; // + UCHAR uctCKE; // + UCHAR uctCKRSE; // + UCHAR uctCKRSX; // + UCHAR uctFAW32; // + UCHAR ucMR4lo; // + UCHAR ucMR4hi; // + UCHAR ucMR5lo; // + UCHAR ucMR5hi; // + UCHAR ucTerminator; + UCHAR ucReserved; +}ATOM_MEMORY_TIMING_FORMAT_V2; + +typedef struct _ATOM_MEMORY_FORMAT +{ + ULONG ulDllDisClock; // memory DLL will be disable when target memory clock is below this clock + union{ + USHORT usEMRS2Value; // EMRS2 Value is used for GDDR2 and GDDR4 memory type + USHORT usDDR3_Reserved; // Not used for DDR3 memory + }; + union{ + USHORT usEMRS3Value; // EMRS3 Value is used for GDDR2 and GDDR4 memory type + USHORT usDDR3_MR3; // Used for DDR3 memory + }; + UCHAR ucMemoryType; // [7:4]=0x1:DDR1;=0x2:DDR2;=0x3:DDR3;=0x4:DDR4;[3:0] - must not be used for now; + UCHAR ucMemoryVenderID; // Predefined,never change across designs or memory type/vender. If not predefined, vendor detection table gets executed + UCHAR ucRow; // Number of Row,in power of 2; + UCHAR ucColumn; // Number of Column,in power of 2; + UCHAR ucBank; // Nunber of Bank; + UCHAR ucRank; // Number of Rank, in power of 2 + UCHAR ucBurstSize; // burst size, 0= burst size=4 1= burst size=8 + UCHAR ucDllDisBit; // position of DLL Enable/Disable bit in EMRS ( Extended Mode Register ) + UCHAR ucRefreshRateFactor; // memory refresh rate in unit of ms + UCHAR ucDensity; // _8Mx32, _16Mx32, _16Mx16, _32Mx16 + UCHAR ucPreamble; //[7:4] Write Preamble, [3:0] Read Preamble + UCHAR ucMemAttrib; // Memory Device Addribute, like RDBI/WDBI etc + ATOM_MEMORY_TIMING_FORMAT asMemTiming[5]; //Memory Timing block sort from lower clock to higher clock +}ATOM_MEMORY_FORMAT; + + +typedef struct _ATOM_VRAM_MODULE_V3 +{ + ULONG ulChannelMapCfg; // board dependent paramenter:Channel combination + USHORT usSize; // size of ATOM_VRAM_MODULE_V3 + USHORT usDefaultMVDDQ; // board dependent parameter:Default Memory Core Voltage + USHORT usDefaultMVDDC; // board dependent parameter:Default Memory IO Voltage + UCHAR ucExtMemoryID; // An external indicator (by hardcode, callback or pin) to tell what is the current memory module + UCHAR ucChannelNum; // board dependent parameter:Number of channel; + UCHAR ucChannelSize; // board dependent parameter:32bit or 64bit + UCHAR ucVREFI; // board dependnt parameter: EXT or INT +160mv to -140mv + UCHAR ucNPL_RT; // board dependent parameter:NPL round trip delay, used for calculate memory timing parameters + UCHAR ucFlag; // To enable/disable functionalities based on memory type + ATOM_MEMORY_FORMAT asMemory; // describ all of video memory parameters from memory spec +}ATOM_VRAM_MODULE_V3; + + +//ATOM_VRAM_MODULE_V3.ucNPL_RT +#define NPL_RT_MASK 0x0f +#define BATTERY_ODT_MASK 0xc0 + +#define ATOM_VRAM_MODULE ATOM_VRAM_MODULE_V3 + +typedef struct _ATOM_VRAM_MODULE_V4 +{ + ULONG ulChannelMapCfg; // board dependent parameter: Channel combination + USHORT usModuleSize; // size of ATOM_VRAM_MODULE_V4, make it easy for VBIOS to look for next entry of VRAM_MODULE + USHORT usPrivateReserved; // BIOS internal reserved space to optimize code size, updated by the compiler, shouldn't be modified manually!! + // MC_ARB_RAMCFG (includes NOOFBANK,NOOFRANKS,NOOFROWS,NOOFCOLS) + USHORT usReserved; + UCHAR ucExtMemoryID; // An external indicator (by hardcode, callback or pin) to tell what is the current memory module + UCHAR ucMemoryType; // [7:4]=0x1:DDR1;=0x2:DDR2;=0x3:DDR3;=0x4:DDR4; 0x5:DDR5 [3:0] - Must be 0x0 for now; + UCHAR ucChannelNum; // Number of channels present in this module config + UCHAR ucChannelWidth; // 0 - 32 bits; 1 - 64 bits + UCHAR ucDensity; // _8Mx32, _16Mx32, _16Mx16, _32Mx16 + UCHAR ucFlag; // To enable/disable functionalities based on memory type + UCHAR ucMisc; // bit0: 0 - single rank; 1 - dual rank; bit2: 0 - burstlength 4, 1 - burstlength 8 + UCHAR ucVREFI; // board dependent parameter + UCHAR ucNPL_RT; // board dependent parameter:NPL round trip delay, used for calculate memory timing parameters + UCHAR ucPreamble; // [7:4] Write Preamble, [3:0] Read Preamble + UCHAR ucMemorySize; // BIOS internal reserved space to optimize code size, updated by the compiler, shouldn't be modified manually!! + // Total memory size in unit of 16MB for CONFIG_MEMSIZE - bit[23:0] zeros + UCHAR ucReserved[3]; + +//compare with V3, we flat the struct by merging ATOM_MEMORY_FORMAT (as is) into V4 as the same level + union{ + USHORT usEMRS2Value; // EMRS2 Value is used for GDDR2 and GDDR4 memory type + USHORT usDDR3_Reserved; + }; + union{ + USHORT usEMRS3Value; // EMRS3 Value is used for GDDR2 and GDDR4 memory type + USHORT usDDR3_MR3; // Used for DDR3 memory + }; + UCHAR ucMemoryVenderID; // Predefined, If not predefined, vendor detection table gets executed + UCHAR ucRefreshRateFactor; // [1:0]=RefreshFactor (00=8ms, 01=16ms, 10=32ms,11=64ms) + UCHAR ucReserved2[2]; + ATOM_MEMORY_TIMING_FORMAT asMemTiming[5];//Memory Timing block sort from lower clock to higher clock +}ATOM_VRAM_MODULE_V4; + +#define VRAM_MODULE_V4_MISC_RANK_MASK 0x3 +#define VRAM_MODULE_V4_MISC_DUAL_RANK 0x1 +#define VRAM_MODULE_V4_MISC_BL_MASK 0x4 +#define VRAM_MODULE_V4_MISC_BL8 0x4 +#define VRAM_MODULE_V4_MISC_DUAL_CS 0x10 + +typedef struct _ATOM_VRAM_MODULE_V5 +{ + ULONG ulChannelMapCfg; // board dependent parameter: Channel combination + USHORT usModuleSize; // size of ATOM_VRAM_MODULE_V4, make it easy for VBIOS to look for next entry of VRAM_MODULE + USHORT usPrivateReserved; // BIOS internal reserved space to optimize code size, updated by the compiler, shouldn't be modified manually!! + // MC_ARB_RAMCFG (includes NOOFBANK,NOOFRANKS,NOOFROWS,NOOFCOLS) + USHORT usReserved; + UCHAR ucExtMemoryID; // An external indicator (by hardcode, callback or pin) to tell what is the current memory module + UCHAR ucMemoryType; // [7:4]=0x1:DDR1;=0x2:DDR2;=0x3:DDR3;=0x4:DDR4; 0x5:DDR5 [3:0] - Must be 0x0 for now; + UCHAR ucChannelNum; // Number of channels present in this module config + UCHAR ucChannelWidth; // 0 - 32 bits; 1 - 64 bits + UCHAR ucDensity; // _8Mx32, _16Mx32, _16Mx16, _32Mx16 + UCHAR ucFlag; // To enable/disable functionalities based on memory type + UCHAR ucMisc; // bit0: 0 - single rank; 1 - dual rank; bit2: 0 - burstlength 4, 1 - burstlength 8 + UCHAR ucVREFI; // board dependent parameter + UCHAR ucNPL_RT; // board dependent parameter:NPL round trip delay, used for calculate memory timing parameters + UCHAR ucPreamble; // [7:4] Write Preamble, [3:0] Read Preamble + UCHAR ucMemorySize; // BIOS internal reserved space to optimize code size, updated by the compiler, shouldn't be modified manually!! + // Total memory size in unit of 16MB for CONFIG_MEMSIZE - bit[23:0] zeros + UCHAR ucReserved[3]; + +//compare with V3, we flat the struct by merging ATOM_MEMORY_FORMAT (as is) into V4 as the same level + USHORT usEMRS2Value; // EMRS2 Value is used for GDDR2 and GDDR4 memory type + USHORT usEMRS3Value; // EMRS3 Value is used for GDDR2 and GDDR4 memory type + UCHAR ucMemoryVenderID; // Predefined, If not predefined, vendor detection table gets executed + UCHAR ucRefreshRateFactor; // [1:0]=RefreshFactor (00=8ms, 01=16ms, 10=32ms,11=64ms) + UCHAR ucFIFODepth; // FIFO depth supposes to be detected during vendor detection, but if we dont do vendor detection we have to hardcode FIFO Depth + UCHAR ucCDR_Bandwidth; // [0:3]=Read CDR bandwidth, [4:7] - Write CDR Bandwidth + ATOM_MEMORY_TIMING_FORMAT_V1 asMemTiming[5];//Memory Timing block sort from lower clock to higher clock +}ATOM_VRAM_MODULE_V5; + +typedef struct _ATOM_VRAM_MODULE_V6 +{ + ULONG ulChannelMapCfg; // board dependent parameter: Channel combination + USHORT usModuleSize; // size of ATOM_VRAM_MODULE_V4, make it easy for VBIOS to look for next entry of VRAM_MODULE + USHORT usPrivateReserved; // BIOS internal reserved space to optimize code size, updated by the compiler, shouldn't be modified manually!! + // MC_ARB_RAMCFG (includes NOOFBANK,NOOFRANKS,NOOFROWS,NOOFCOLS) + USHORT usReserved; + UCHAR ucExtMemoryID; // An external indicator (by hardcode, callback or pin) to tell what is the current memory module + UCHAR ucMemoryType; // [7:4]=0x1:DDR1;=0x2:DDR2;=0x3:DDR3;=0x4:DDR4; 0x5:DDR5 [3:0] - Must be 0x0 for now; + UCHAR ucChannelNum; // Number of channels present in this module config + UCHAR ucChannelWidth; // 0 - 32 bits; 1 - 64 bits + UCHAR ucDensity; // _8Mx32, _16Mx32, _16Mx16, _32Mx16 + UCHAR ucFlag; // To enable/disable functionalities based on memory type + UCHAR ucMisc; // bit0: 0 - single rank; 1 - dual rank; bit2: 0 - burstlength 4, 1 - burstlength 8 + UCHAR ucVREFI; // board dependent parameter + UCHAR ucNPL_RT; // board dependent parameter:NPL round trip delay, used for calculate memory timing parameters + UCHAR ucPreamble; // [7:4] Write Preamble, [3:0] Read Preamble + UCHAR ucMemorySize; // BIOS internal reserved space to optimize code size, updated by the compiler, shouldn't be modified manually!! + // Total memory size in unit of 16MB for CONFIG_MEMSIZE - bit[23:0] zeros + UCHAR ucReserved[3]; + +//compare with V3, we flat the struct by merging ATOM_MEMORY_FORMAT (as is) into V4 as the same level + USHORT usEMRS2Value; // EMRS2 Value is used for GDDR2 and GDDR4 memory type + USHORT usEMRS3Value; // EMRS3 Value is used for GDDR2 and GDDR4 memory type + UCHAR ucMemoryVenderID; // Predefined, If not predefined, vendor detection table gets executed + UCHAR ucRefreshRateFactor; // [1:0]=RefreshFactor (00=8ms, 01=16ms, 10=32ms,11=64ms) + UCHAR ucFIFODepth; // FIFO depth supposes to be detected during vendor detection, but if we dont do vendor detection we have to hardcode FIFO Depth + UCHAR ucCDR_Bandwidth; // [0:3]=Read CDR bandwidth, [4:7] - Write CDR Bandwidth + ATOM_MEMORY_TIMING_FORMAT_V2 asMemTiming[5];//Memory Timing block sort from lower clock to higher clock +}ATOM_VRAM_MODULE_V6; + +typedef struct _ATOM_VRAM_MODULE_V7 +{ +// Design Specific Values + ULONG ulChannelMapCfg; // mmMC_SHARED_CHREMAP + USHORT usModuleSize; // Size of ATOM_VRAM_MODULE_V7 + USHORT usPrivateReserved; // MC_ARB_RAMCFG (includes NOOFBANK,NOOFRANKS,NOOFROWS,NOOFCOLS) + USHORT usReserved; + UCHAR ucExtMemoryID; // Current memory module ID + UCHAR ucMemoryType; // MEM_TYPE_DDR2/DDR3/GDDR3/GDDR5 + UCHAR ucChannelNum; // Number of mem. channels supported in this module + UCHAR ucChannelWidth; // CHANNEL_16BIT/CHANNEL_32BIT/CHANNEL_64BIT + UCHAR ucDensity; // _8Mx32, _16Mx32, _16Mx16, _32Mx16 + UCHAR ucReserve; // Former container for Mx_FLAGS like DBI_AC_MODE_ENABLE_ASIC for GDDR4. Not used now. + UCHAR ucMisc; // RANK_OF_THISMEMORY etc. + UCHAR ucVREFI; // Not used. + UCHAR ucNPL_RT; // Round trip delay (MC_SEQ_CAS_TIMING [28:24]:TCL=CL+NPL_RT-2). Always 2. + UCHAR ucPreamble; // [7:4] Write Preamble, [3:0] Read Preamble + UCHAR ucMemorySize; // Total memory size in unit of 16MB for CONFIG_MEMSIZE - bit[23:0] zeros + UCHAR ucReserved[3]; +// Memory Module specific values + USHORT usEMRS2Value; // EMRS2/MR2 Value. + USHORT usEMRS3Value; // EMRS3/MR3 Value. + UCHAR ucMemoryVenderID; // [7:4] Revision, [3:0] Vendor code + UCHAR ucRefreshRateFactor; // [1:0]=RefreshFactor (00=8ms, 01=16ms, 10=32ms,11=64ms) + UCHAR ucFIFODepth; // FIFO depth can be detected during vendor detection, here is hardcoded per memory + UCHAR ucCDR_Bandwidth; // [0:3]=Read CDR bandwidth, [4:7] - Write CDR Bandwidth + char strMemPNString[20]; // part number end with '0'. +}ATOM_VRAM_MODULE_V7; + +typedef struct _ATOM_VRAM_INFO_V2 +{ + ATOM_COMMON_TABLE_HEADER sHeader; + UCHAR ucNumOfVRAMModule; + ATOM_VRAM_MODULE aVramInfo[ATOM_MAX_NUMBER_OF_VRAM_MODULE]; // just for allocation, real number of blocks is in ucNumOfVRAMModule; +}ATOM_VRAM_INFO_V2; + +typedef struct _ATOM_VRAM_INFO_V3 +{ + ATOM_COMMON_TABLE_HEADER sHeader; + USHORT usMemAdjustTblOffset; // offset of ATOM_INIT_REG_BLOCK structure for memory vendor specific MC adjust setting + USHORT usMemClkPatchTblOffset; // offset of ATOM_INIT_REG_BLOCK structure for memory clock specific MC setting + USHORT usRerseved; + UCHAR aVID_PinsShift[9]; // 8 bit strap maximum+terminator + UCHAR ucNumOfVRAMModule; + ATOM_VRAM_MODULE aVramInfo[ATOM_MAX_NUMBER_OF_VRAM_MODULE]; // just for allocation, real number of blocks is in ucNumOfVRAMModule; + ATOM_INIT_REG_BLOCK asMemPatch; // for allocation + // ATOM_INIT_REG_BLOCK aMemAdjust; +}ATOM_VRAM_INFO_V3; + +#define ATOM_VRAM_INFO_LAST ATOM_VRAM_INFO_V3 + +typedef struct _ATOM_VRAM_INFO_V4 +{ + ATOM_COMMON_TABLE_HEADER sHeader; + USHORT usMemAdjustTblOffset; // offset of ATOM_INIT_REG_BLOCK structure for memory vendor specific MC adjust setting + USHORT usMemClkPatchTblOffset; // offset of ATOM_INIT_REG_BLOCK structure for memory clock specific MC setting + USHORT usRerseved; + UCHAR ucMemDQ7_0ByteRemap; // DQ line byte remap, =0: Memory Data line BYTE0, =1: BYTE1, =2: BYTE2, =3: BYTE3 + ULONG ulMemDQ7_0BitRemap; // each DQ line ( 7~0) use 3bits, like: DQ0=Bit[2:0], DQ1:[5:3], ... DQ7:[23:21] + UCHAR ucReservde[4]; + UCHAR ucNumOfVRAMModule; + ATOM_VRAM_MODULE_V4 aVramInfo[ATOM_MAX_NUMBER_OF_VRAM_MODULE]; // just for allocation, real number of blocks is in ucNumOfVRAMModule; + ATOM_INIT_REG_BLOCK asMemPatch; // for allocation + // ATOM_INIT_REG_BLOCK aMemAdjust; +}ATOM_VRAM_INFO_V4; + +typedef struct _ATOM_VRAM_INFO_HEADER_V2_1 +{ + ATOM_COMMON_TABLE_HEADER sHeader; + USHORT usMemAdjustTblOffset; // offset of ATOM_INIT_REG_BLOCK structure for memory vendor specific MC adjust setting + USHORT usMemClkPatchTblOffset; // offset of ATOM_INIT_REG_BLOCK structure for memory clock specific MC setting + USHORT usReserved[4]; + UCHAR ucNumOfVRAMModule; // indicate number of VRAM module + UCHAR ucMemoryClkPatchTblVer; // version of memory AC timing register list + UCHAR ucVramModuleVer; // indicate ATOM_VRAM_MODUE version + UCHAR ucReserved; + ATOM_VRAM_MODULE_V7 aVramInfo[ATOM_MAX_NUMBER_OF_VRAM_MODULE]; // just for allocation, real number of blocks is in ucNumOfVRAMModule; +}ATOM_VRAM_INFO_HEADER_V2_1; + + +typedef struct _ATOM_VRAM_GPIO_DETECTION_INFO +{ + ATOM_COMMON_TABLE_HEADER sHeader; + UCHAR aVID_PinsShift[9]; //8 bit strap maximum+terminator +}ATOM_VRAM_GPIO_DETECTION_INFO; + + +typedef struct _ATOM_MEMORY_TRAINING_INFO +{ + ATOM_COMMON_TABLE_HEADER sHeader; + UCHAR ucTrainingLoop; + UCHAR ucReserved[3]; + ATOM_INIT_REG_BLOCK asMemTrainingSetting; +}ATOM_MEMORY_TRAINING_INFO; + + +typedef struct SW_I2C_CNTL_DATA_PARAMETERS +{ + UCHAR ucControl; + UCHAR ucData; + UCHAR ucSatus; + UCHAR ucTemp; +} SW_I2C_CNTL_DATA_PARAMETERS; + +#define SW_I2C_CNTL_DATA_PS_ALLOCATION SW_I2C_CNTL_DATA_PARAMETERS + +typedef struct _SW_I2C_IO_DATA_PARAMETERS +{ + USHORT GPIO_Info; + UCHAR ucAct; + UCHAR ucData; + } SW_I2C_IO_DATA_PARAMETERS; + +#define SW_I2C_IO_DATA_PS_ALLOCATION SW_I2C_IO_DATA_PARAMETERS + +/****************************SW I2C CNTL DEFINITIONS**********************/ +#define SW_I2C_IO_RESET 0 +#define SW_I2C_IO_GET 1 +#define SW_I2C_IO_DRIVE 2 +#define SW_I2C_IO_SET 3 +#define SW_I2C_IO_START 4 + +#define SW_I2C_IO_CLOCK 0 +#define SW_I2C_IO_DATA 0x80 + +#define SW_I2C_IO_ZERO 0 +#define SW_I2C_IO_ONE 0x100 + +#define SW_I2C_CNTL_READ 0 +#define SW_I2C_CNTL_WRITE 1 +#define SW_I2C_CNTL_START 2 +#define SW_I2C_CNTL_STOP 3 +#define SW_I2C_CNTL_OPEN 4 +#define SW_I2C_CNTL_CLOSE 5 +#define SW_I2C_CNTL_WRITE1BIT 6 + +//==============================VESA definition Portion=============================== +#define VESA_OEM_PRODUCT_REV "01.00" +#define VESA_MODE_ATTRIBUTE_MODE_SUPPORT 0xBB //refer to VBE spec p.32, no TTY support +#define VESA_MODE_WIN_ATTRIBUTE 7 +#define VESA_WIN_SIZE 64 + +typedef struct _PTR_32_BIT_STRUCTURE +{ + USHORT Offset16; + USHORT Segment16; +} PTR_32_BIT_STRUCTURE; + +typedef union _PTR_32_BIT_UNION +{ + PTR_32_BIT_STRUCTURE SegmentOffset; + ULONG Ptr32_Bit; +} PTR_32_BIT_UNION; + +typedef struct _VBE_1_2_INFO_BLOCK_UPDATABLE +{ + UCHAR VbeSignature[4]; + USHORT VbeVersion; + PTR_32_BIT_UNION OemStringPtr; + UCHAR Capabilities[4]; + PTR_32_BIT_UNION VideoModePtr; + USHORT TotalMemory; +} VBE_1_2_INFO_BLOCK_UPDATABLE; + + +typedef struct _VBE_2_0_INFO_BLOCK_UPDATABLE +{ + VBE_1_2_INFO_BLOCK_UPDATABLE CommonBlock; + USHORT OemSoftRev; + PTR_32_BIT_UNION OemVendorNamePtr; + PTR_32_BIT_UNION OemProductNamePtr; + PTR_32_BIT_UNION OemProductRevPtr; +} VBE_2_0_INFO_BLOCK_UPDATABLE; + +typedef union _VBE_VERSION_UNION +{ + VBE_2_0_INFO_BLOCK_UPDATABLE VBE_2_0_InfoBlock; + VBE_1_2_INFO_BLOCK_UPDATABLE VBE_1_2_InfoBlock; +} VBE_VERSION_UNION; + +typedef struct _VBE_INFO_BLOCK +{ + VBE_VERSION_UNION UpdatableVBE_Info; + UCHAR Reserved[222]; + UCHAR OemData[256]; +} VBE_INFO_BLOCK; + +typedef struct _VBE_FP_INFO +{ + USHORT HSize; + USHORT VSize; + USHORT FPType; + UCHAR RedBPP; + UCHAR GreenBPP; + UCHAR BlueBPP; + UCHAR ReservedBPP; + ULONG RsvdOffScrnMemSize; + ULONG RsvdOffScrnMEmPtr; + UCHAR Reserved[14]; +} VBE_FP_INFO; + +typedef struct _VESA_MODE_INFO_BLOCK +{ +// Mandatory information for all VBE revisions + USHORT ModeAttributes; // dw ? ; mode attributes + UCHAR WinAAttributes; // db ? ; window A attributes + UCHAR WinBAttributes; // db ? ; window B attributes + USHORT WinGranularity; // dw ? ; window granularity + USHORT WinSize; // dw ? ; window size + USHORT WinASegment; // dw ? ; window A start segment + USHORT WinBSegment; // dw ? ; window B start segment + ULONG WinFuncPtr; // dd ? ; real mode pointer to window function + USHORT BytesPerScanLine;// dw ? ; bytes per scan line + +//; Mandatory information for VBE 1.2 and above + USHORT XResolution; // dw ? ; horizontal resolution in pixels or characters + USHORT YResolution; // dw ? ; vertical resolution in pixels or characters + UCHAR XCharSize; // db ? ; character cell width in pixels + UCHAR YCharSize; // db ? ; character cell height in pixels + UCHAR NumberOfPlanes; // db ? ; number of memory planes + UCHAR BitsPerPixel; // db ? ; bits per pixel + UCHAR NumberOfBanks; // db ? ; number of banks + UCHAR MemoryModel; // db ? ; memory model type + UCHAR BankSize; // db ? ; bank size in KB + UCHAR NumberOfImagePages;// db ? ; number of images + UCHAR ReservedForPageFunction;//db 1 ; reserved for page function + +//; Direct Color fields(required for direct/6 and YUV/7 memory models) + UCHAR RedMaskSize; // db ? ; size of direct color red mask in bits + UCHAR RedFieldPosition; // db ? ; bit position of lsb of red mask + UCHAR GreenMaskSize; // db ? ; size of direct color green mask in bits + UCHAR GreenFieldPosition; // db ? ; bit position of lsb of green mask + UCHAR BlueMaskSize; // db ? ; size of direct color blue mask in bits + UCHAR BlueFieldPosition; // db ? ; bit position of lsb of blue mask + UCHAR RsvdMaskSize; // db ? ; size of direct color reserved mask in bits + UCHAR RsvdFieldPosition; // db ? ; bit position of lsb of reserved mask + UCHAR DirectColorModeInfo;// db ? ; direct color mode attributes + +//; Mandatory information for VBE 2.0 and above + ULONG PhysBasePtr; // dd ? ; physical address for flat memory frame buffer + ULONG Reserved_1; // dd 0 ; reserved - always set to 0 + USHORT Reserved_2; // dw 0 ; reserved - always set to 0 + +//; Mandatory information for VBE 3.0 and above + USHORT LinBytesPerScanLine; // dw ? ; bytes per scan line for linear modes + UCHAR BnkNumberOfImagePages;// db ? ; number of images for banked modes + UCHAR LinNumberOfImagPages; // db ? ; number of images for linear modes + UCHAR LinRedMaskSize; // db ? ; size of direct color red mask(linear modes) + UCHAR LinRedFieldPosition; // db ? ; bit position of lsb of red mask(linear modes) + UCHAR LinGreenMaskSize; // db ? ; size of direct color green mask(linear modes) + UCHAR LinGreenFieldPosition;// db ? ; bit position of lsb of green mask(linear modes) + UCHAR LinBlueMaskSize; // db ? ; size of direct color blue mask(linear modes) + UCHAR LinBlueFieldPosition; // db ? ; bit position of lsb of blue mask(linear modes) + UCHAR LinRsvdMaskSize; // db ? ; size of direct color reserved mask(linear modes) + UCHAR LinRsvdFieldPosition; // db ? ; bit position of lsb of reserved mask(linear modes) + ULONG MaxPixelClock; // dd ? ; maximum pixel clock(in Hz) for graphics mode + UCHAR Reserved; // db 190 dup (0) +} VESA_MODE_INFO_BLOCK; + +// BIOS function CALLS +#define ATOM_BIOS_EXTENDED_FUNCTION_CODE 0xA0 // ATI Extended Function code +#define ATOM_BIOS_FUNCTION_COP_MODE 0x00 +#define ATOM_BIOS_FUNCTION_SHORT_QUERY1 0x04 +#define ATOM_BIOS_FUNCTION_SHORT_QUERY2 0x05 +#define ATOM_BIOS_FUNCTION_SHORT_QUERY3 0x06 +#define ATOM_BIOS_FUNCTION_GET_DDC 0x0B +#define ATOM_BIOS_FUNCTION_ASIC_DSTATE 0x0E +#define ATOM_BIOS_FUNCTION_DEBUG_PLAY 0x0F +#define ATOM_BIOS_FUNCTION_STV_STD 0x16 +#define ATOM_BIOS_FUNCTION_DEVICE_DET 0x17 +#define ATOM_BIOS_FUNCTION_DEVICE_SWITCH 0x18 + +#define ATOM_BIOS_FUNCTION_PANEL_CONTROL 0x82 +#define ATOM_BIOS_FUNCTION_OLD_DEVICE_DET 0x83 +#define ATOM_BIOS_FUNCTION_OLD_DEVICE_SWITCH 0x84 +#define ATOM_BIOS_FUNCTION_HW_ICON 0x8A +#define ATOM_BIOS_FUNCTION_SET_CMOS 0x8B +#define SUB_FUNCTION_UPDATE_DISPLAY_INFO 0x8000 // Sub function 80 +#define SUB_FUNCTION_UPDATE_EXPANSION_INFO 0x8100 // Sub function 80 + +#define ATOM_BIOS_FUNCTION_DISPLAY_INFO 0x8D +#define ATOM_BIOS_FUNCTION_DEVICE_ON_OFF 0x8E +#define ATOM_BIOS_FUNCTION_VIDEO_STATE 0x8F +#define ATOM_SUB_FUNCTION_GET_CRITICAL_STATE 0x0300 // Sub function 03 +#define ATOM_SUB_FUNCTION_GET_LIDSTATE 0x0700 // Sub function 7 +#define ATOM_SUB_FUNCTION_THERMAL_STATE_NOTICE 0x1400 // Notify caller the current thermal state +#define ATOM_SUB_FUNCTION_CRITICAL_STATE_NOTICE 0x8300 // Notify caller the current critical state +#define ATOM_SUB_FUNCTION_SET_LIDSTATE 0x8500 // Sub function 85 +#define ATOM_SUB_FUNCTION_GET_REQ_DISPLAY_FROM_SBIOS_MODE 0x8900// Sub function 89 +#define ATOM_SUB_FUNCTION_INFORM_ADC_SUPPORT 0x9400 // Notify caller that ADC is supported + + +#define ATOM_BIOS_FUNCTION_VESA_DPMS 0x4F10 // Set DPMS +#define ATOM_SUB_FUNCTION_SET_DPMS 0x0001 // BL: Sub function 01 +#define ATOM_SUB_FUNCTION_GET_DPMS 0x0002 // BL: Sub function 02 +#define ATOM_PARAMETER_VESA_DPMS_ON 0x0000 // BH Parameter for DPMS ON. +#define ATOM_PARAMETER_VESA_DPMS_STANDBY 0x0100 // BH Parameter for DPMS STANDBY +#define ATOM_PARAMETER_VESA_DPMS_SUSPEND 0x0200 // BH Parameter for DPMS SUSPEND +#define ATOM_PARAMETER_VESA_DPMS_OFF 0x0400 // BH Parameter for DPMS OFF +#define ATOM_PARAMETER_VESA_DPMS_REDUCE_ON 0x0800 // BH Parameter for DPMS REDUCE ON (NOT SUPPORTED) + +#define ATOM_BIOS_RETURN_CODE_MASK 0x0000FF00L +#define ATOM_BIOS_REG_HIGH_MASK 0x0000FF00L +#define ATOM_BIOS_REG_LOW_MASK 0x000000FFL + +// structure used for VBIOS only + +//DispOutInfoTable +typedef struct _ASIC_TRANSMITTER_INFO +{ + USHORT usTransmitterObjId; + USHORT usSupportDevice; + UCHAR ucTransmitterCmdTblId; + UCHAR ucConfig; + UCHAR ucEncoderID; //available 1st encoder ( default ) + UCHAR ucOptionEncoderID; //available 2nd encoder ( optional ) + UCHAR uc2ndEncoderID; + UCHAR ucReserved; +}ASIC_TRANSMITTER_INFO; + +#define ASIC_TRANSMITTER_INFO_CONFIG__DVO_SDR_MODE 0x01 +#define ASIC_TRANSMITTER_INFO_CONFIG__COHERENT_MODE 0x02 +#define ASIC_TRANSMITTER_INFO_CONFIG__ENCODEROBJ_ID_MASK 0xc4 +#define ASIC_TRANSMITTER_INFO_CONFIG__ENCODER_A 0x00 +#define ASIC_TRANSMITTER_INFO_CONFIG__ENCODER_B 0x04 +#define ASIC_TRANSMITTER_INFO_CONFIG__ENCODER_C 0x40 +#define ASIC_TRANSMITTER_INFO_CONFIG__ENCODER_D 0x44 +#define ASIC_TRANSMITTER_INFO_CONFIG__ENCODER_E 0x80 +#define ASIC_TRANSMITTER_INFO_CONFIG__ENCODER_F 0x84 + +typedef struct _ASIC_ENCODER_INFO +{ + UCHAR ucEncoderID; + UCHAR ucEncoderConfig; + USHORT usEncoderCmdTblId; +}ASIC_ENCODER_INFO; + +typedef struct _ATOM_DISP_OUT_INFO +{ + ATOM_COMMON_TABLE_HEADER sHeader; + USHORT ptrTransmitterInfo; + USHORT ptrEncoderInfo; + ASIC_TRANSMITTER_INFO asTransmitterInfo[1]; + ASIC_ENCODER_INFO asEncoderInfo[1]; +}ATOM_DISP_OUT_INFO; + +typedef struct _ATOM_DISP_OUT_INFO_V2 +{ + ATOM_COMMON_TABLE_HEADER sHeader; + USHORT ptrTransmitterInfo; + USHORT ptrEncoderInfo; + USHORT ptrMainCallParserFar; // direct address of main parser call in VBIOS binary. + ASIC_TRANSMITTER_INFO asTransmitterInfo[1]; + ASIC_ENCODER_INFO asEncoderInfo[1]; +}ATOM_DISP_OUT_INFO_V2; + +// DispDevicePriorityInfo +typedef struct _ATOM_DISPLAY_DEVICE_PRIORITY_INFO +{ + ATOM_COMMON_TABLE_HEADER sHeader; + USHORT asDevicePriority[16]; +}ATOM_DISPLAY_DEVICE_PRIORITY_INFO; + +//ProcessAuxChannelTransactionTable +typedef struct _PROCESS_AUX_CHANNEL_TRANSACTION_PARAMETERS +{ + USHORT lpAuxRequest; + USHORT lpDataOut; + UCHAR ucChannelID; + union + { + UCHAR ucReplyStatus; + UCHAR ucDelay; + }; + UCHAR ucDataOutLen; + UCHAR ucReserved; +}PROCESS_AUX_CHANNEL_TRANSACTION_PARAMETERS; + +//ProcessAuxChannelTransactionTable +typedef struct _PROCESS_AUX_CHANNEL_TRANSACTION_PARAMETERS_V2 +{ + USHORT lpAuxRequest; + USHORT lpDataOut; + UCHAR ucChannelID; + union + { + UCHAR ucReplyStatus; + UCHAR ucDelay; + }; + UCHAR ucDataOutLen; + UCHAR ucHPD_ID; //=0: HPD1, =1: HPD2, =2: HPD3, =3: HPD4, =4: HPD5, =5: HPD6 +}PROCESS_AUX_CHANNEL_TRANSACTION_PARAMETERS_V2; + +#define PROCESS_AUX_CHANNEL_TRANSACTION_PS_ALLOCATION PROCESS_AUX_CHANNEL_TRANSACTION_PARAMETERS + +//GetSinkType + +typedef struct _DP_ENCODER_SERVICE_PARAMETERS +{ + USHORT ucLinkClock; + union + { + UCHAR ucConfig; // for DP training command + UCHAR ucI2cId; // use for GET_SINK_TYPE command + }; + UCHAR ucAction; + UCHAR ucStatus; + UCHAR ucLaneNum; + UCHAR ucReserved[2]; +}DP_ENCODER_SERVICE_PARAMETERS; + +// ucAction +#define ATOM_DP_ACTION_GET_SINK_TYPE 0x01 +/* obselete */ +#define ATOM_DP_ACTION_TRAINING_START 0x02 +#define ATOM_DP_ACTION_TRAINING_COMPLETE 0x03 +#define ATOM_DP_ACTION_TRAINING_PATTERN_SEL 0x04 +#define ATOM_DP_ACTION_SET_VSWING_PREEMP 0x05 +#define ATOM_DP_ACTION_GET_VSWING_PREEMP 0x06 +#define ATOM_DP_ACTION_BLANKING 0x07 + +// ucConfig +#define ATOM_DP_CONFIG_ENCODER_SEL_MASK 0x03 +#define ATOM_DP_CONFIG_DIG1_ENCODER 0x00 +#define ATOM_DP_CONFIG_DIG2_ENCODER 0x01 +#define ATOM_DP_CONFIG_EXTERNAL_ENCODER 0x02 +#define ATOM_DP_CONFIG_LINK_SEL_MASK 0x04 +#define ATOM_DP_CONFIG_LINK_A 0x00 +#define ATOM_DP_CONFIG_LINK_B 0x04 +/* /obselete */ +#define DP_ENCODER_SERVICE_PS_ALLOCATION WRITE_ONE_BYTE_HW_I2C_DATA_PARAMETERS + + +typedef struct _DP_ENCODER_SERVICE_PARAMETERS_V2 +{ + USHORT usExtEncoderObjId; // External Encoder Object Id, output parameter only, use when ucAction = DP_SERVICE_V2_ACTION_DET_EXT_CONNECTION + UCHAR ucAuxId; + UCHAR ucAction; + UCHAR ucSinkType; // Iput and Output parameters. + UCHAR ucHPDId; // Input parameter, used when ucAction = DP_SERVICE_V2_ACTION_DET_EXT_CONNECTION + UCHAR ucReserved[2]; +}DP_ENCODER_SERVICE_PARAMETERS_V2; + +typedef struct _DP_ENCODER_SERVICE_PS_ALLOCATION_V2 +{ + DP_ENCODER_SERVICE_PARAMETERS_V2 asDPServiceParam; + PROCESS_AUX_CHANNEL_TRANSACTION_PARAMETERS_V2 asAuxParam; +}DP_ENCODER_SERVICE_PS_ALLOCATION_V2; + +// ucAction +#define DP_SERVICE_V2_ACTION_GET_SINK_TYPE 0x01 +#define DP_SERVICE_V2_ACTION_DET_LCD_CONNECTION 0x02 + + +// DP_TRAINING_TABLE +#define DPCD_SET_LINKRATE_LANENUM_PATTERN1_TBL_ADDR ATOM_DP_TRAINING_TBL_ADDR +#define DPCD_SET_SS_CNTL_TBL_ADDR (ATOM_DP_TRAINING_TBL_ADDR + 8 ) +#define DPCD_SET_LANE_VSWING_PREEMP_TBL_ADDR (ATOM_DP_TRAINING_TBL_ADDR + 16 ) +#define DPCD_SET_TRAINING_PATTERN0_TBL_ADDR (ATOM_DP_TRAINING_TBL_ADDR + 24 ) +#define DPCD_SET_TRAINING_PATTERN2_TBL_ADDR (ATOM_DP_TRAINING_TBL_ADDR + 32) +#define DPCD_GET_LINKRATE_LANENUM_SS_TBL_ADDR (ATOM_DP_TRAINING_TBL_ADDR + 40) +#define DPCD_GET_LANE_STATUS_ADJUST_TBL_ADDR (ATOM_DP_TRAINING_TBL_ADDR + 48) +#define DP_I2C_AUX_DDC_WRITE_START_TBL_ADDR (ATOM_DP_TRAINING_TBL_ADDR + 60) +#define DP_I2C_AUX_DDC_WRITE_TBL_ADDR (ATOM_DP_TRAINING_TBL_ADDR + 64) +#define DP_I2C_AUX_DDC_READ_START_TBL_ADDR (ATOM_DP_TRAINING_TBL_ADDR + 72) +#define DP_I2C_AUX_DDC_READ_TBL_ADDR (ATOM_DP_TRAINING_TBL_ADDR + 76) +#define DP_I2C_AUX_DDC_WRITE_END_TBL_ADDR (ATOM_DP_TRAINING_TBL_ADDR + 80) +#define DP_I2C_AUX_DDC_READ_END_TBL_ADDR (ATOM_DP_TRAINING_TBL_ADDR + 84) + +typedef struct _PROCESS_I2C_CHANNEL_TRANSACTION_PARAMETERS +{ + UCHAR ucI2CSpeed; + union + { + UCHAR ucRegIndex; + UCHAR ucStatus; + }; + USHORT lpI2CDataOut; + UCHAR ucFlag; + UCHAR ucTransBytes; + UCHAR ucSlaveAddr; + UCHAR ucLineNumber; +}PROCESS_I2C_CHANNEL_TRANSACTION_PARAMETERS; + +#define PROCESS_I2C_CHANNEL_TRANSACTION_PS_ALLOCATION PROCESS_I2C_CHANNEL_TRANSACTION_PARAMETERS + +//ucFlag +#define HW_I2C_WRITE 1 +#define HW_I2C_READ 0 +#define I2C_2BYTE_ADDR 0x02 + +typedef struct _SET_HWBLOCK_INSTANCE_PARAMETER_V2 +{ + UCHAR ucHWBlkInst; // HW block instance, 0, 1, 2, ... + UCHAR ucReserved[3]; +}SET_HWBLOCK_INSTANCE_PARAMETER_V2; + +#define HWBLKINST_INSTANCE_MASK 0x07 +#define HWBLKINST_HWBLK_MASK 0xF0 +#define HWBLKINST_HWBLK_SHIFT 0x04 + +//ucHWBlock +#define SELECT_DISP_ENGINE 0 +#define SELECT_DISP_PLL 1 +#define SELECT_DCIO_UNIPHY_LINK0 2 +#define SELECT_DCIO_UNIPHY_LINK1 3 +#define SELECT_DCIO_IMPCAL 4 +#define SELECT_DCIO_DIG 6 +#define SELECT_CRTC_PIXEL_RATE 7 +#define SELECT_VGA_BLK 8 + +/****************************************************************************/ +//Portion VI: Definitinos for vbios MC scratch registers that driver used +/****************************************************************************/ + +#define MC_MISC0__MEMORY_TYPE_MASK 0xF0000000 +#define MC_MISC0__MEMORY_TYPE__GDDR1 0x10000000 +#define MC_MISC0__MEMORY_TYPE__DDR2 0x20000000 +#define MC_MISC0__MEMORY_TYPE__GDDR3 0x30000000 +#define MC_MISC0__MEMORY_TYPE__GDDR4 0x40000000 +#define MC_MISC0__MEMORY_TYPE__GDDR5 0x50000000 +#define MC_MISC0__MEMORY_TYPE__DDR3 0xB0000000 + +/****************************************************************************/ +//Portion VI: Definitinos being oboselete +/****************************************************************************/ + +//========================================================================================== +//Remove the definitions below when driver is ready! +typedef struct _ATOM_DAC_INFO +{ + ATOM_COMMON_TABLE_HEADER sHeader; + USHORT usMaxFrequency; // in 10kHz unit + USHORT usReserved; +}ATOM_DAC_INFO; + + +typedef struct _COMPASSIONATE_DATA +{ + ATOM_COMMON_TABLE_HEADER sHeader; + + //============================== DAC1 portion + UCHAR ucDAC1_BG_Adjustment; + UCHAR ucDAC1_DAC_Adjustment; + USHORT usDAC1_FORCE_Data; + //============================== DAC2 portion + UCHAR ucDAC2_CRT2_BG_Adjustment; + UCHAR ucDAC2_CRT2_DAC_Adjustment; + USHORT usDAC2_CRT2_FORCE_Data; + USHORT usDAC2_CRT2_MUX_RegisterIndex; + UCHAR ucDAC2_CRT2_MUX_RegisterInfo; //Bit[4:0]=Bit position,Bit[7]=1:Active High;=0 Active Low + UCHAR ucDAC2_NTSC_BG_Adjustment; + UCHAR ucDAC2_NTSC_DAC_Adjustment; + USHORT usDAC2_TV1_FORCE_Data; + USHORT usDAC2_TV1_MUX_RegisterIndex; + UCHAR ucDAC2_TV1_MUX_RegisterInfo; //Bit[4:0]=Bit position,Bit[7]=1:Active High;=0 Active Low + UCHAR ucDAC2_CV_BG_Adjustment; + UCHAR ucDAC2_CV_DAC_Adjustment; + USHORT usDAC2_CV_FORCE_Data; + USHORT usDAC2_CV_MUX_RegisterIndex; + UCHAR ucDAC2_CV_MUX_RegisterInfo; //Bit[4:0]=Bit position,Bit[7]=1:Active High;=0 Active Low + UCHAR ucDAC2_PAL_BG_Adjustment; + UCHAR ucDAC2_PAL_DAC_Adjustment; + USHORT usDAC2_TV2_FORCE_Data; +}COMPASSIONATE_DATA; + +/****************************Supported Device Info Table Definitions**********************/ +// ucConnectInfo: +// [7:4] - connector type +// = 1 - VGA connector +// = 2 - DVI-I +// = 3 - DVI-D +// = 4 - DVI-A +// = 5 - SVIDEO +// = 6 - COMPOSITE +// = 7 - LVDS +// = 8 - DIGITAL LINK +// = 9 - SCART +// = 0xA - HDMI_type A +// = 0xB - HDMI_type B +// = 0xE - Special case1 (DVI+DIN) +// Others=TBD +// [3:0] - DAC Associated +// = 0 - no DAC +// = 1 - DACA +// = 2 - DACB +// = 3 - External DAC +// Others=TBD +// + +typedef struct _ATOM_CONNECTOR_INFO +{ +#if ATOM_BIG_ENDIAN + UCHAR bfConnectorType:4; + UCHAR bfAssociatedDAC:4; +#else + UCHAR bfAssociatedDAC:4; + UCHAR bfConnectorType:4; +#endif +}ATOM_CONNECTOR_INFO; + +typedef union _ATOM_CONNECTOR_INFO_ACCESS +{ + ATOM_CONNECTOR_INFO sbfAccess; + UCHAR ucAccess; +}ATOM_CONNECTOR_INFO_ACCESS; + +typedef struct _ATOM_CONNECTOR_INFO_I2C +{ + ATOM_CONNECTOR_INFO_ACCESS sucConnectorInfo; + ATOM_I2C_ID_CONFIG_ACCESS sucI2cId; +}ATOM_CONNECTOR_INFO_I2C; + + +typedef struct _ATOM_SUPPORTED_DEVICES_INFO +{ + ATOM_COMMON_TABLE_HEADER sHeader; + USHORT usDeviceSupport; + ATOM_CONNECTOR_INFO_I2C asConnInfo[ATOM_MAX_SUPPORTED_DEVICE_INFO]; +}ATOM_SUPPORTED_DEVICES_INFO; + +#define NO_INT_SRC_MAPPED 0xFF + +typedef struct _ATOM_CONNECTOR_INC_SRC_BITMAP +{ + UCHAR ucIntSrcBitmap; +}ATOM_CONNECTOR_INC_SRC_BITMAP; + +typedef struct _ATOM_SUPPORTED_DEVICES_INFO_2 +{ + ATOM_COMMON_TABLE_HEADER sHeader; + USHORT usDeviceSupport; + ATOM_CONNECTOR_INFO_I2C asConnInfo[ATOM_MAX_SUPPORTED_DEVICE_INFO_2]; + ATOM_CONNECTOR_INC_SRC_BITMAP asIntSrcInfo[ATOM_MAX_SUPPORTED_DEVICE_INFO_2]; +}ATOM_SUPPORTED_DEVICES_INFO_2; + +typedef struct _ATOM_SUPPORTED_DEVICES_INFO_2d1 +{ + ATOM_COMMON_TABLE_HEADER sHeader; + USHORT usDeviceSupport; + ATOM_CONNECTOR_INFO_I2C asConnInfo[ATOM_MAX_SUPPORTED_DEVICE]; + ATOM_CONNECTOR_INC_SRC_BITMAP asIntSrcInfo[ATOM_MAX_SUPPORTED_DEVICE]; +}ATOM_SUPPORTED_DEVICES_INFO_2d1; + +#define ATOM_SUPPORTED_DEVICES_INFO_LAST ATOM_SUPPORTED_DEVICES_INFO_2d1 + + + +typedef struct _ATOM_MISC_CONTROL_INFO +{ + USHORT usFrequency; + UCHAR ucPLL_ChargePump; // PLL charge-pump gain control + UCHAR ucPLL_DutyCycle; // PLL duty cycle control + UCHAR ucPLL_VCO_Gain; // PLL VCO gain control + UCHAR ucPLL_VoltageSwing; // PLL driver voltage swing control +}ATOM_MISC_CONTROL_INFO; + + +#define ATOM_MAX_MISC_INFO 4 + +typedef struct _ATOM_TMDS_INFO +{ + ATOM_COMMON_TABLE_HEADER sHeader; + USHORT usMaxFrequency; // in 10Khz + ATOM_MISC_CONTROL_INFO asMiscInfo[ATOM_MAX_MISC_INFO]; +}ATOM_TMDS_INFO; + + +typedef struct _ATOM_ENCODER_ANALOG_ATTRIBUTE +{ + UCHAR ucTVStandard; //Same as TV standards defined above, + UCHAR ucPadding[1]; +}ATOM_ENCODER_ANALOG_ATTRIBUTE; + +typedef struct _ATOM_ENCODER_DIGITAL_ATTRIBUTE +{ + UCHAR ucAttribute; //Same as other digital encoder attributes defined above + UCHAR ucPadding[1]; +}ATOM_ENCODER_DIGITAL_ATTRIBUTE; + +typedef union _ATOM_ENCODER_ATTRIBUTE +{ + ATOM_ENCODER_ANALOG_ATTRIBUTE sAlgAttrib; + ATOM_ENCODER_DIGITAL_ATTRIBUTE sDigAttrib; +}ATOM_ENCODER_ATTRIBUTE; + + +typedef struct _DVO_ENCODER_CONTROL_PARAMETERS +{ + USHORT usPixelClock; + USHORT usEncoderID; + UCHAR ucDeviceType; //Use ATOM_DEVICE_xxx1_Index to indicate device type only. + UCHAR ucAction; //ATOM_ENABLE/ATOM_DISABLE/ATOM_HPD_INIT + ATOM_ENCODER_ATTRIBUTE usDevAttr; +}DVO_ENCODER_CONTROL_PARAMETERS; + +typedef struct _DVO_ENCODER_CONTROL_PS_ALLOCATION +{ + DVO_ENCODER_CONTROL_PARAMETERS sDVOEncoder; + WRITE_ONE_BYTE_HW_I2C_DATA_PS_ALLOCATION sReserved; //Caller doesn't need to init this portion +}DVO_ENCODER_CONTROL_PS_ALLOCATION; + + +#define ATOM_XTMDS_ASIC_SI164_ID 1 +#define ATOM_XTMDS_ASIC_SI178_ID 2 +#define ATOM_XTMDS_ASIC_TFP513_ID 3 +#define ATOM_XTMDS_SUPPORTED_SINGLELINK 0x00000001 +#define ATOM_XTMDS_SUPPORTED_DUALLINK 0x00000002 +#define ATOM_XTMDS_MVPU_FPGA 0x00000004 + + +typedef struct _ATOM_XTMDS_INFO +{ + ATOM_COMMON_TABLE_HEADER sHeader; + USHORT usSingleLinkMaxFrequency; + ATOM_I2C_ID_CONFIG_ACCESS sucI2cId; //Point the ID on which I2C is used to control external chip + UCHAR ucXtransimitterID; + UCHAR ucSupportedLink; // Bit field, bit0=1, single link supported;bit1=1,dual link supported + UCHAR ucSequnceAlterID; // Even with the same external TMDS asic, it's possible that the program seqence alters + // due to design. This ID is used to alert driver that the sequence is not "standard"! + UCHAR ucMasterAddress; // Address to control Master xTMDS Chip + UCHAR ucSlaveAddress; // Address to control Slave xTMDS Chip +}ATOM_XTMDS_INFO; + +typedef struct _DFP_DPMS_STATUS_CHANGE_PARAMETERS +{ + UCHAR ucEnable; // ATOM_ENABLE=On or ATOM_DISABLE=Off + UCHAR ucDevice; // ATOM_DEVICE_DFP1_INDEX.... + UCHAR ucPadding[2]; +}DFP_DPMS_STATUS_CHANGE_PARAMETERS; + +/****************************Legacy Power Play Table Definitions **********************/ + +//Definitions for ulPowerPlayMiscInfo +#define ATOM_PM_MISCINFO_SPLIT_CLOCK 0x00000000L +#define ATOM_PM_MISCINFO_USING_MCLK_SRC 0x00000001L +#define ATOM_PM_MISCINFO_USING_SCLK_SRC 0x00000002L + +#define ATOM_PM_MISCINFO_VOLTAGE_DROP_SUPPORT 0x00000004L +#define ATOM_PM_MISCINFO_VOLTAGE_DROP_ACTIVE_HIGH 0x00000008L + +#define ATOM_PM_MISCINFO_LOAD_PERFORMANCE_EN 0x00000010L + +#define ATOM_PM_MISCINFO_ENGINE_CLOCK_CONTRL_EN 0x00000020L +#define ATOM_PM_MISCINFO_MEMORY_CLOCK_CONTRL_EN 0x00000040L +#define ATOM_PM_MISCINFO_PROGRAM_VOLTAGE 0x00000080L //When this bit set, ucVoltageDropIndex is not an index for GPIO pin, but a voltage ID that SW needs program + +#define ATOM_PM_MISCINFO_ASIC_REDUCED_SPEED_SCLK_EN 0x00000100L +#define ATOM_PM_MISCINFO_ASIC_DYNAMIC_VOLTAGE_EN 0x00000200L +#define ATOM_PM_MISCINFO_ASIC_SLEEP_MODE_EN 0x00000400L +#define ATOM_PM_MISCINFO_LOAD_BALANCE_EN 0x00000800L +#define ATOM_PM_MISCINFO_DEFAULT_DC_STATE_ENTRY_TRUE 0x00001000L +#define ATOM_PM_MISCINFO_DEFAULT_LOW_DC_STATE_ENTRY_TRUE 0x00002000L +#define ATOM_PM_MISCINFO_LOW_LCD_REFRESH_RATE 0x00004000L + +#define ATOM_PM_MISCINFO_DRIVER_DEFAULT_MODE 0x00008000L +#define ATOM_PM_MISCINFO_OVER_CLOCK_MODE 0x00010000L +#define ATOM_PM_MISCINFO_OVER_DRIVE_MODE 0x00020000L +#define ATOM_PM_MISCINFO_POWER_SAVING_MODE 0x00040000L +#define ATOM_PM_MISCINFO_THERMAL_DIODE_MODE 0x00080000L + +#define ATOM_PM_MISCINFO_FRAME_MODULATION_MASK 0x00300000L //0-FM Disable, 1-2 level FM, 2-4 level FM, 3-Reserved +#define ATOM_PM_MISCINFO_FRAME_MODULATION_SHIFT 20 + +#define ATOM_PM_MISCINFO_DYN_CLK_3D_IDLE 0x00400000L +#define ATOM_PM_MISCINFO_DYNAMIC_CLOCK_DIVIDER_BY_2 0x00800000L +#define ATOM_PM_MISCINFO_DYNAMIC_CLOCK_DIVIDER_BY_4 0x01000000L +#define ATOM_PM_MISCINFO_DYNAMIC_HDP_BLOCK_EN 0x02000000L //When set, Dynamic +#define ATOM_PM_MISCINFO_DYNAMIC_MC_HOST_BLOCK_EN 0x04000000L //When set, Dynamic +#define ATOM_PM_MISCINFO_3D_ACCELERATION_EN 0x08000000L //When set, This mode is for acceleated 3D mode + +#define ATOM_PM_MISCINFO_POWERPLAY_SETTINGS_GROUP_MASK 0x70000000L //1-Optimal Battery Life Group, 2-High Battery, 3-Balanced, 4-High Performance, 5- Optimal Performance (Default state with Default clocks) +#define ATOM_PM_MISCINFO_POWERPLAY_SETTINGS_GROUP_SHIFT 28 +#define ATOM_PM_MISCINFO_ENABLE_BACK_BIAS 0x80000000L + +#define ATOM_PM_MISCINFO2_SYSTEM_AC_LITE_MODE 0x00000001L +#define ATOM_PM_MISCINFO2_MULTI_DISPLAY_SUPPORT 0x00000002L +#define ATOM_PM_MISCINFO2_DYNAMIC_BACK_BIAS_EN 0x00000004L +#define ATOM_PM_MISCINFO2_FS3D_OVERDRIVE_INFO 0x00000008L +#define ATOM_PM_MISCINFO2_FORCEDLOWPWR_MODE 0x00000010L +#define ATOM_PM_MISCINFO2_VDDCI_DYNAMIC_VOLTAGE_EN 0x00000020L +#define ATOM_PM_MISCINFO2_VIDEO_PLAYBACK_CAPABLE 0x00000040L //If this bit is set in multi-pp mode, then driver will pack up one with the minior power consumption. + //If it's not set in any pp mode, driver will use its default logic to pick a pp mode in video playback +#define ATOM_PM_MISCINFO2_NOT_VALID_ON_DC 0x00000080L +#define ATOM_PM_MISCINFO2_STUTTER_MODE_EN 0x00000100L +#define ATOM_PM_MISCINFO2_UVD_SUPPORT_MODE 0x00000200L + +//ucTableFormatRevision=1 +//ucTableContentRevision=1 +typedef struct _ATOM_POWERMODE_INFO +{ + ULONG ulMiscInfo; //The power level should be arranged in ascending order + ULONG ulReserved1; // must set to 0 + ULONG ulReserved2; // must set to 0 + USHORT usEngineClock; + USHORT usMemoryClock; + UCHAR ucVoltageDropIndex; // index to GPIO table + UCHAR ucSelectedPanel_RefreshRate;// panel refresh rate + UCHAR ucMinTemperature; + UCHAR ucMaxTemperature; + UCHAR ucNumPciELanes; // number of PCIE lanes +}ATOM_POWERMODE_INFO; + +//ucTableFormatRevision=2 +//ucTableContentRevision=1 +typedef struct _ATOM_POWERMODE_INFO_V2 +{ + ULONG ulMiscInfo; //The power level should be arranged in ascending order + ULONG ulMiscInfo2; + ULONG ulEngineClock; + ULONG ulMemoryClock; + UCHAR ucVoltageDropIndex; // index to GPIO table + UCHAR ucSelectedPanel_RefreshRate;// panel refresh rate + UCHAR ucMinTemperature; + UCHAR ucMaxTemperature; + UCHAR ucNumPciELanes; // number of PCIE lanes +}ATOM_POWERMODE_INFO_V2; + +//ucTableFormatRevision=2 +//ucTableContentRevision=2 +typedef struct _ATOM_POWERMODE_INFO_V3 +{ + ULONG ulMiscInfo; //The power level should be arranged in ascending order + ULONG ulMiscInfo2; + ULONG ulEngineClock; + ULONG ulMemoryClock; + UCHAR ucVoltageDropIndex; // index to Core (VDDC) votage table + UCHAR ucSelectedPanel_RefreshRate;// panel refresh rate + UCHAR ucMinTemperature; + UCHAR ucMaxTemperature; + UCHAR ucNumPciELanes; // number of PCIE lanes + UCHAR ucVDDCI_VoltageDropIndex; // index to VDDCI votage table +}ATOM_POWERMODE_INFO_V3; + + +#define ATOM_MAX_NUMBEROF_POWER_BLOCK 8 + +#define ATOM_PP_OVERDRIVE_INTBITMAP_AUXWIN 0x01 +#define ATOM_PP_OVERDRIVE_INTBITMAP_OVERDRIVE 0x02 + +#define ATOM_PP_OVERDRIVE_THERMALCONTROLLER_LM63 0x01 +#define ATOM_PP_OVERDRIVE_THERMALCONTROLLER_ADM1032 0x02 +#define ATOM_PP_OVERDRIVE_THERMALCONTROLLER_ADM1030 0x03 +#define ATOM_PP_OVERDRIVE_THERMALCONTROLLER_MUA6649 0x04 +#define ATOM_PP_OVERDRIVE_THERMALCONTROLLER_LM64 0x05 +#define ATOM_PP_OVERDRIVE_THERMALCONTROLLER_F75375 0x06 +#define ATOM_PP_OVERDRIVE_THERMALCONTROLLER_ASC7512 0x07 // Andigilog + + +typedef struct _ATOM_POWERPLAY_INFO +{ + ATOM_COMMON_TABLE_HEADER sHeader; + UCHAR ucOverdriveThermalController; + UCHAR ucOverdriveI2cLine; + UCHAR ucOverdriveIntBitmap; + UCHAR ucOverdriveControllerAddress; + UCHAR ucSizeOfPowerModeEntry; + UCHAR ucNumOfPowerModeEntries; + ATOM_POWERMODE_INFO asPowerPlayInfo[ATOM_MAX_NUMBEROF_POWER_BLOCK]; +}ATOM_POWERPLAY_INFO; + +typedef struct _ATOM_POWERPLAY_INFO_V2 +{ + ATOM_COMMON_TABLE_HEADER sHeader; + UCHAR ucOverdriveThermalController; + UCHAR ucOverdriveI2cLine; + UCHAR ucOverdriveIntBitmap; + UCHAR ucOverdriveControllerAddress; + UCHAR ucSizeOfPowerModeEntry; + UCHAR ucNumOfPowerModeEntries; + ATOM_POWERMODE_INFO_V2 asPowerPlayInfo[ATOM_MAX_NUMBEROF_POWER_BLOCK]; +}ATOM_POWERPLAY_INFO_V2; + +typedef struct _ATOM_POWERPLAY_INFO_V3 +{ + ATOM_COMMON_TABLE_HEADER sHeader; + UCHAR ucOverdriveThermalController; + UCHAR ucOverdriveI2cLine; + UCHAR ucOverdriveIntBitmap; + UCHAR ucOverdriveControllerAddress; + UCHAR ucSizeOfPowerModeEntry; + UCHAR ucNumOfPowerModeEntries; + ATOM_POWERMODE_INFO_V3 asPowerPlayInfo[ATOM_MAX_NUMBEROF_POWER_BLOCK]; +}ATOM_POWERPLAY_INFO_V3; + +/* New PPlib */ +/**************************************************************************/ +typedef struct _ATOM_PPLIB_THERMALCONTROLLER + +{ + UCHAR ucType; // one of ATOM_PP_THERMALCONTROLLER_* + UCHAR ucI2cLine; // as interpreted by DAL I2C + UCHAR ucI2cAddress; + UCHAR ucFanParameters; // Fan Control Parameters. + UCHAR ucFanMinRPM; // Fan Minimum RPM (hundreds) -- for display purposes only. + UCHAR ucFanMaxRPM; // Fan Maximum RPM (hundreds) -- for display purposes only. + UCHAR ucReserved; // ---- + UCHAR ucFlags; // to be defined +} ATOM_PPLIB_THERMALCONTROLLER; + +#define ATOM_PP_FANPARAMETERS_TACHOMETER_PULSES_PER_REVOLUTION_MASK 0x0f +#define ATOM_PP_FANPARAMETERS_NOFAN 0x80 // No fan is connected to this controller. + +#define ATOM_PP_THERMALCONTROLLER_NONE 0 +#define ATOM_PP_THERMALCONTROLLER_LM63 1 // Not used by PPLib +#define ATOM_PP_THERMALCONTROLLER_ADM1032 2 // Not used by PPLib +#define ATOM_PP_THERMALCONTROLLER_ADM1030 3 // Not used by PPLib +#define ATOM_PP_THERMALCONTROLLER_MUA6649 4 // Not used by PPLib +#define ATOM_PP_THERMALCONTROLLER_LM64 5 +#define ATOM_PP_THERMALCONTROLLER_F75375 6 // Not used by PPLib +#define ATOM_PP_THERMALCONTROLLER_RV6xx 7 +#define ATOM_PP_THERMALCONTROLLER_RV770 8 +#define ATOM_PP_THERMALCONTROLLER_ADT7473 9 +#define ATOM_PP_THERMALCONTROLLER_EXTERNAL_GPIO 11 +#define ATOM_PP_THERMALCONTROLLER_EVERGREEN 12 +#define ATOM_PP_THERMALCONTROLLER_EMC2103 13 /* 0x0D */ // Only fan control will be implemented, do NOT show this in PPGen. +#define ATOM_PP_THERMALCONTROLLER_SUMO 14 /* 0x0E */ // Sumo type, used internally +#define ATOM_PP_THERMALCONTROLLER_NISLANDS 15 + +// Thermal controller 'combo type' to use an external controller for Fan control and an internal controller for thermal. +// We probably should reserve the bit 0x80 for this use. +// To keep the number of these types low we should also use the same code for all ASICs (i.e. do not distinguish RV6xx and RV7xx Internal here). +// The driver can pick the correct internal controller based on the ASIC. + +#define ATOM_PP_THERMALCONTROLLER_ADT7473_WITH_INTERNAL 0x89 // ADT7473 Fan Control + Internal Thermal Controller +#define ATOM_PP_THERMALCONTROLLER_EMC2103_WITH_INTERNAL 0x8D // EMC2103 Fan Control + Internal Thermal Controller + +typedef struct _ATOM_PPLIB_STATE +{ + UCHAR ucNonClockStateIndex; + UCHAR ucClockStateIndices[1]; // variable-sized +} ATOM_PPLIB_STATE; + +typedef struct _ATOM_PPLIB_FANTABLE +{ + UCHAR ucFanTableFormat; // Change this if the table format changes or version changes so that the other fields are not the same. + UCHAR ucTHyst; // Temperature hysteresis. Integer. + USHORT usTMin; // The temperature, in 0.01 centigrades, below which we just run at a minimal PWM. + USHORT usTMed; // The middle temperature where we change slopes. + USHORT usTHigh; // The high point above TMed for adjusting the second slope. + USHORT usPWMMin; // The minimum PWM value in percent (0.01% increments). + USHORT usPWMMed; // The PWM value (in percent) at TMed. + USHORT usPWMHigh; // The PWM value at THigh. +} ATOM_PPLIB_FANTABLE; + +typedef struct _ATOM_PPLIB_EXTENDEDHEADER +{ + USHORT usSize; + ULONG ulMaxEngineClock; // For Overdrive. + ULONG ulMaxMemoryClock; // For Overdrive. + // Add extra system parameters here, always adjust size to include all fields. +} ATOM_PPLIB_EXTENDEDHEADER; + +//// ATOM_PPLIB_POWERPLAYTABLE::ulPlatformCaps +#define ATOM_PP_PLATFORM_CAP_BACKBIAS 1 +#define ATOM_PP_PLATFORM_CAP_POWERPLAY 2 +#define ATOM_PP_PLATFORM_CAP_SBIOSPOWERSOURCE 4 +#define ATOM_PP_PLATFORM_CAP_ASPM_L0s 8 +#define ATOM_PP_PLATFORM_CAP_ASPM_L1 16 +#define ATOM_PP_PLATFORM_CAP_HARDWAREDC 32 +#define ATOM_PP_PLATFORM_CAP_GEMINIPRIMARY 64 +#define ATOM_PP_PLATFORM_CAP_STEPVDDC 128 +#define ATOM_PP_PLATFORM_CAP_VOLTAGECONTROL 256 +#define ATOM_PP_PLATFORM_CAP_SIDEPORTCONTROL 512 +#define ATOM_PP_PLATFORM_CAP_TURNOFFPLL_ASPML1 1024 +#define ATOM_PP_PLATFORM_CAP_HTLINKCONTROL 2048 +#define ATOM_PP_PLATFORM_CAP_MVDDCONTROL 4096 +#define ATOM_PP_PLATFORM_CAP_GOTO_BOOT_ON_ALERT 0x2000 // Go to boot state on alerts, e.g. on an AC->DC transition. +#define ATOM_PP_PLATFORM_CAP_DONT_WAIT_FOR_VBLANK_ON_ALERT 0x4000 // Do NOT wait for VBLANK during an alert (e.g. AC->DC transition). +#define ATOM_PP_PLATFORM_CAP_VDDCI_CONTROL 0x8000 // Does the driver control VDDCI independently from VDDC. +#define ATOM_PP_PLATFORM_CAP_REGULATOR_HOT 0x00010000 // Enable the 'regulator hot' feature. +#define ATOM_PP_PLATFORM_CAP_BACO 0x00020000 // Does the driver supports BACO state. + +typedef struct _ATOM_PPLIB_POWERPLAYTABLE +{ + ATOM_COMMON_TABLE_HEADER sHeader; + + UCHAR ucDataRevision; + + UCHAR ucNumStates; + UCHAR ucStateEntrySize; + UCHAR ucClockInfoSize; + UCHAR ucNonClockSize; + + // offset from start of this table to array of ucNumStates ATOM_PPLIB_STATE structures + USHORT usStateArrayOffset; + + // offset from start of this table to array of ASIC-specific structures, + // currently ATOM_PPLIB_CLOCK_INFO. + USHORT usClockInfoArrayOffset; + + // offset from start of this table to array of ATOM_PPLIB_NONCLOCK_INFO + USHORT usNonClockInfoArrayOffset; + + USHORT usBackbiasTime; // in microseconds + USHORT usVoltageTime; // in microseconds + USHORT usTableSize; //the size of this structure, or the extended structure + + ULONG ulPlatformCaps; // See ATOM_PPLIB_CAPS_* + + ATOM_PPLIB_THERMALCONTROLLER sThermalController; + + USHORT usBootClockInfoOffset; + USHORT usBootNonClockInfoOffset; + +} ATOM_PPLIB_POWERPLAYTABLE; + +typedef struct _ATOM_PPLIB_POWERPLAYTABLE2 +{ + ATOM_PPLIB_POWERPLAYTABLE basicTable; + UCHAR ucNumCustomThermalPolicy; + USHORT usCustomThermalPolicyArrayOffset; +}ATOM_PPLIB_POWERPLAYTABLE2, *LPATOM_PPLIB_POWERPLAYTABLE2; + +typedef struct _ATOM_PPLIB_POWERPLAYTABLE3 +{ + ATOM_PPLIB_POWERPLAYTABLE2 basicTable2; + USHORT usFormatID; // To be used ONLY by PPGen. + USHORT usFanTableOffset; + USHORT usExtendendedHeaderOffset; +} ATOM_PPLIB_POWERPLAYTABLE3, *LPATOM_PPLIB_POWERPLAYTABLE3; + +typedef struct _ATOM_PPLIB_POWERPLAYTABLE4 +{ + ATOM_PPLIB_POWERPLAYTABLE3 basicTable3; + ULONG ulGoldenPPID; // PPGen use only + ULONG ulGoldenRevision; // PPGen use only + USHORT usVddcDependencyOnSCLKOffset; + USHORT usVddciDependencyOnMCLKOffset; + USHORT usVddcDependencyOnMCLKOffset; + USHORT usMaxClockVoltageOnDCOffset; + USHORT usReserved[2]; +} ATOM_PPLIB_POWERPLAYTABLE4, *LPATOM_PPLIB_POWERPLAYTABLE4; + +typedef struct _ATOM_PPLIB_POWERPLAYTABLE5 +{ + ATOM_PPLIB_POWERPLAYTABLE4 basicTable4; + ULONG ulTDPLimit; + ULONG ulNearTDPLimit; + ULONG ulSQRampingThreshold; + USHORT usCACLeakageTableOffset; // Points to ATOM_PPLIB_CAC_Leakage_Table + ULONG ulCACLeakage; // TBD, this parameter is still under discussion. Change to ulReserved if not needed. + ULONG ulReserved; +} ATOM_PPLIB_POWERPLAYTABLE5, *LPATOM_PPLIB_POWERPLAYTABLE5; + +//// ATOM_PPLIB_NONCLOCK_INFO::usClassification +#define ATOM_PPLIB_CLASSIFICATION_UI_MASK 0x0007 +#define ATOM_PPLIB_CLASSIFICATION_UI_SHIFT 0 +#define ATOM_PPLIB_CLASSIFICATION_UI_NONE 0 +#define ATOM_PPLIB_CLASSIFICATION_UI_BATTERY 1 +#define ATOM_PPLIB_CLASSIFICATION_UI_BALANCED 3 +#define ATOM_PPLIB_CLASSIFICATION_UI_PERFORMANCE 5 +// 2, 4, 6, 7 are reserved + +#define ATOM_PPLIB_CLASSIFICATION_BOOT 0x0008 +#define ATOM_PPLIB_CLASSIFICATION_THERMAL 0x0010 +#define ATOM_PPLIB_CLASSIFICATION_LIMITEDPOWERSOURCE 0x0020 +#define ATOM_PPLIB_CLASSIFICATION_REST 0x0040 +#define ATOM_PPLIB_CLASSIFICATION_FORCED 0x0080 +#define ATOM_PPLIB_CLASSIFICATION_3DPERFORMANCE 0x0100 +#define ATOM_PPLIB_CLASSIFICATION_OVERDRIVETEMPLATE 0x0200 +#define ATOM_PPLIB_CLASSIFICATION_UVDSTATE 0x0400 +#define ATOM_PPLIB_CLASSIFICATION_3DLOW 0x0800 +#define ATOM_PPLIB_CLASSIFICATION_ACPI 0x1000 +#define ATOM_PPLIB_CLASSIFICATION_HD2STATE 0x2000 +#define ATOM_PPLIB_CLASSIFICATION_HDSTATE 0x4000 +#define ATOM_PPLIB_CLASSIFICATION_SDSTATE 0x8000 + +//// ATOM_PPLIB_NONCLOCK_INFO::usClassification2 +#define ATOM_PPLIB_CLASSIFICATION2_LIMITEDPOWERSOURCE_2 0x0001 +#define ATOM_PPLIB_CLASSIFICATION2_ULV 0x0002 + +//// ATOM_PPLIB_NONCLOCK_INFO::ulCapsAndSettings +#define ATOM_PPLIB_SINGLE_DISPLAY_ONLY 0x00000001 +#define ATOM_PPLIB_SUPPORTS_VIDEO_PLAYBACK 0x00000002 + +// 0 is 2.5Gb/s, 1 is 5Gb/s +#define ATOM_PPLIB_PCIE_LINK_SPEED_MASK 0x00000004 +#define ATOM_PPLIB_PCIE_LINK_SPEED_SHIFT 2 + +// lanes - 1: 1, 2, 4, 8, 12, 16 permitted by PCIE spec +#define ATOM_PPLIB_PCIE_LINK_WIDTH_MASK 0x000000F8 +#define ATOM_PPLIB_PCIE_LINK_WIDTH_SHIFT 3 + +// lookup into reduced refresh-rate table +#define ATOM_PPLIB_LIMITED_REFRESHRATE_VALUE_MASK 0x00000F00 +#define ATOM_PPLIB_LIMITED_REFRESHRATE_VALUE_SHIFT 8 + +#define ATOM_PPLIB_LIMITED_REFRESHRATE_UNLIMITED 0 +#define ATOM_PPLIB_LIMITED_REFRESHRATE_50HZ 1 +// 2-15 TBD as needed. + +#define ATOM_PPLIB_SOFTWARE_DISABLE_LOADBALANCING 0x00001000 +#define ATOM_PPLIB_SOFTWARE_ENABLE_SLEEP_FOR_TIMESTAMPS 0x00002000 +#define ATOM_PPLIB_DISALLOW_ON_DC 0x00004000 +#define ATOM_PPLIB_ENABLE_VARIBRIGHT 0x00008000 + +//memory related flags +#define ATOM_PPLIB_SWSTATE_MEMORY_DLL_OFF 0x000010000 + +//M3 Arb //2bits, current 3 sets of parameters in total +#define ATOM_PPLIB_M3ARB_MASK 0x00060000 +#define ATOM_PPLIB_M3ARB_SHIFT 17 + +#define ATOM_PPLIB_ENABLE_DRR 0x00080000 + +// remaining 16 bits are reserved +typedef struct _ATOM_PPLIB_THERMAL_STATE +{ + UCHAR ucMinTemperature; + UCHAR ucMaxTemperature; + UCHAR ucThermalAction; +}ATOM_PPLIB_THERMAL_STATE, *LPATOM_PPLIB_THERMAL_STATE; + +// Contained in an array starting at the offset +// in ATOM_PPLIB_POWERPLAYTABLE::usNonClockInfoArrayOffset. +// referenced from ATOM_PPLIB_STATE_INFO::ucNonClockStateIndex +#define ATOM_PPLIB_NONCLOCKINFO_VER1 12 +#define ATOM_PPLIB_NONCLOCKINFO_VER2 24 +typedef struct _ATOM_PPLIB_NONCLOCK_INFO +{ + USHORT usClassification; + UCHAR ucMinTemperature; + UCHAR ucMaxTemperature; + ULONG ulCapsAndSettings; + UCHAR ucRequiredPower; + USHORT usClassification2; + ULONG ulVCLK; + ULONG ulDCLK; + UCHAR ucUnused[5]; +} ATOM_PPLIB_NONCLOCK_INFO; + +// Contained in an array starting at the offset +// in ATOM_PPLIB_POWERPLAYTABLE::usClockInfoArrayOffset. +// referenced from ATOM_PPLIB_STATE::ucClockStateIndices +typedef struct _ATOM_PPLIB_R600_CLOCK_INFO +{ + USHORT usEngineClockLow; + UCHAR ucEngineClockHigh; + + USHORT usMemoryClockLow; + UCHAR ucMemoryClockHigh; + + USHORT usVDDC; + USHORT usUnused1; + USHORT usUnused2; + + ULONG ulFlags; // ATOM_PPLIB_R600_FLAGS_* + +} ATOM_PPLIB_R600_CLOCK_INFO; + +// ulFlags in ATOM_PPLIB_R600_CLOCK_INFO +#define ATOM_PPLIB_R600_FLAGS_PCIEGEN2 1 +#define ATOM_PPLIB_R600_FLAGS_UVDSAFE 2 +#define ATOM_PPLIB_R600_FLAGS_BACKBIASENABLE 4 +#define ATOM_PPLIB_R600_FLAGS_MEMORY_ODT_OFF 8 +#define ATOM_PPLIB_R600_FLAGS_MEMORY_DLL_OFF 16 +#define ATOM_PPLIB_R600_FLAGS_LOWPOWER 32 // On the RV770 use 'low power' setting (sequencer S0). + +typedef struct _ATOM_PPLIB_EVERGREEN_CLOCK_INFO +{ + USHORT usEngineClockLow; + UCHAR ucEngineClockHigh; + + USHORT usMemoryClockLow; + UCHAR ucMemoryClockHigh; + + USHORT usVDDC; + USHORT usVDDCI; + USHORT usUnused; + + ULONG ulFlags; // ATOM_PPLIB_R600_FLAGS_* + +} ATOM_PPLIB_EVERGREEN_CLOCK_INFO; + +typedef struct _ATOM_PPLIB_RS780_CLOCK_INFO + +{ + USHORT usLowEngineClockLow; // Low Engine clock in MHz (the same way as on the R600). + UCHAR ucLowEngineClockHigh; + USHORT usHighEngineClockLow; // High Engine clock in MHz. + UCHAR ucHighEngineClockHigh; + USHORT usMemoryClockLow; // For now one of the ATOM_PPLIB_RS780_SPMCLK_XXXX constants. + UCHAR ucMemoryClockHigh; // Currentyl unused. + UCHAR ucPadding; // For proper alignment and size. + USHORT usVDDC; // For the 780, use: None, Low, High, Variable + UCHAR ucMaxHTLinkWidth; // From SBIOS - {2, 4, 8, 16} + UCHAR ucMinHTLinkWidth; // From SBIOS - {2, 4, 8, 16}. Effective only if CDLW enabled. Minimum down stream width could be bigger as display BW requirement. + USHORT usHTLinkFreq; // See definition ATOM_PPLIB_RS780_HTLINKFREQ_xxx or in MHz(>=200). + ULONG ulFlags; +} ATOM_PPLIB_RS780_CLOCK_INFO; + +#define ATOM_PPLIB_RS780_VOLTAGE_NONE 0 +#define ATOM_PPLIB_RS780_VOLTAGE_LOW 1 +#define ATOM_PPLIB_RS780_VOLTAGE_HIGH 2 +#define ATOM_PPLIB_RS780_VOLTAGE_VARIABLE 3 + +#define ATOM_PPLIB_RS780_SPMCLK_NONE 0 // We cannot change the side port memory clock, leave it as it is. +#define ATOM_PPLIB_RS780_SPMCLK_LOW 1 +#define ATOM_PPLIB_RS780_SPMCLK_HIGH 2 + +#define ATOM_PPLIB_RS780_HTLINKFREQ_NONE 0 +#define ATOM_PPLIB_RS780_HTLINKFREQ_LOW 1 +#define ATOM_PPLIB_RS780_HTLINKFREQ_HIGH 2 + +typedef struct _ATOM_PPLIB_SUMO_CLOCK_INFO{ + USHORT usEngineClockLow; //clockfrequency & 0xFFFF. The unit is in 10khz + UCHAR ucEngineClockHigh; //clockfrequency >> 16. + UCHAR vddcIndex; //2-bit vddc index; + UCHAR leakage; //please use 8-bit absolute value, not the 6-bit % value + //please initalize to 0 + UCHAR rsv; + //please initalize to 0 + USHORT rsv1; + //please initialize to 0s + ULONG rsv2[2]; +}ATOM_PPLIB_SUMO_CLOCK_INFO; + + + +typedef struct _ATOM_PPLIB_STATE_V2 +{ + //number of valid dpm levels in this state; Driver uses it to calculate the whole + //size of the state: sizeof(ATOM_PPLIB_STATE_V2) + (ucNumDPMLevels - 1) * sizeof(UCHAR) + UCHAR ucNumDPMLevels; + + //a index to the array of nonClockInfos + UCHAR nonClockInfoIndex; + /** + * Driver will read the first ucNumDPMLevels in this array + */ + UCHAR clockInfoIndex[1]; +} ATOM_PPLIB_STATE_V2; + +typedef struct StateArray{ + //how many states we have + UCHAR ucNumEntries; + + ATOM_PPLIB_STATE_V2 states[1]; +}StateArray; + + +typedef struct ClockInfoArray{ + //how many clock levels we have + UCHAR ucNumEntries; + + //sizeof(ATOM_PPLIB_SUMO_CLOCK_INFO) + UCHAR ucEntrySize; + + //this is for Sumo + ATOM_PPLIB_SUMO_CLOCK_INFO clockInfo[1]; +}ClockInfoArray; + +typedef struct NonClockInfoArray{ + + //how many non-clock levels we have. normally should be same as number of states + UCHAR ucNumEntries; + //sizeof(ATOM_PPLIB_NONCLOCK_INFO) + UCHAR ucEntrySize; + + ATOM_PPLIB_NONCLOCK_INFO nonClockInfo[1]; +}NonClockInfoArray; + +typedef struct _ATOM_PPLIB_Clock_Voltage_Dependency_Record +{ + USHORT usClockLow; + UCHAR ucClockHigh; + USHORT usVoltage; +}ATOM_PPLIB_Clock_Voltage_Dependency_Record; + +typedef struct _ATOM_PPLIB_Clock_Voltage_Dependency_Table +{ + UCHAR ucNumEntries; // Number of entries. + ATOM_PPLIB_Clock_Voltage_Dependency_Record entries[1]; // Dynamically allocate entries. +}ATOM_PPLIB_Clock_Voltage_Dependency_Table; + +typedef struct _ATOM_PPLIB_Clock_Voltage_Limit_Record +{ + USHORT usSclkLow; + UCHAR ucSclkHigh; + USHORT usMclkLow; + UCHAR ucMclkHigh; + USHORT usVddc; + USHORT usVddci; +}ATOM_PPLIB_Clock_Voltage_Limit_Record; + +typedef struct _ATOM_PPLIB_Clock_Voltage_Limit_Table +{ + UCHAR ucNumEntries; // Number of entries. + ATOM_PPLIB_Clock_Voltage_Limit_Record entries[1]; // Dynamically allocate entries. +}ATOM_PPLIB_Clock_Voltage_Limit_Table; + +/**************************************************************************/ + + +// Following definitions are for compatibility issue in different SW components. +#define ATOM_MASTER_DATA_TABLE_REVISION 0x01 +#define Object_Info Object_Header +#define AdjustARB_SEQ MC_InitParameter +#define VRAM_GPIO_DetectionInfo VoltageObjectInfo +#define ASIC_VDDCI_Info ASIC_ProfilingInfo +#define ASIC_MVDDQ_Info MemoryTrainingInfo +#define SS_Info PPLL_SS_Info +#define ASIC_MVDDC_Info ASIC_InternalSS_Info +#define DispDevicePriorityInfo SaveRestoreInfo +#define DispOutInfo TV_VideoMode + + +#define ATOM_ENCODER_OBJECT_TABLE ATOM_OBJECT_TABLE +#define ATOM_CONNECTOR_OBJECT_TABLE ATOM_OBJECT_TABLE + +//New device naming, remove them when both DAL/VBIOS is ready +#define DFP2I_OUTPUT_CONTROL_PARAMETERS CRT1_OUTPUT_CONTROL_PARAMETERS +#define DFP2I_OUTPUT_CONTROL_PS_ALLOCATION DFP2I_OUTPUT_CONTROL_PARAMETERS + +#define DFP1X_OUTPUT_CONTROL_PARAMETERS CRT1_OUTPUT_CONTROL_PARAMETERS +#define DFP1X_OUTPUT_CONTROL_PS_ALLOCATION DFP1X_OUTPUT_CONTROL_PARAMETERS + +#define DFP1I_OUTPUT_CONTROL_PARAMETERS DFP1_OUTPUT_CONTROL_PARAMETERS +#define DFP1I_OUTPUT_CONTROL_PS_ALLOCATION DFP1_OUTPUT_CONTROL_PS_ALLOCATION + +#define ATOM_DEVICE_DFP1I_SUPPORT ATOM_DEVICE_DFP1_SUPPORT +#define ATOM_DEVICE_DFP1X_SUPPORT ATOM_DEVICE_DFP2_SUPPORT + +#define ATOM_DEVICE_DFP1I_INDEX ATOM_DEVICE_DFP1_INDEX +#define ATOM_DEVICE_DFP1X_INDEX ATOM_DEVICE_DFP2_INDEX + +#define ATOM_DEVICE_DFP2I_INDEX 0x00000009 +#define ATOM_DEVICE_DFP2I_SUPPORT (0x1L << ATOM_DEVICE_DFP2I_INDEX) + +#define ATOM_S0_DFP1I ATOM_S0_DFP1 +#define ATOM_S0_DFP1X ATOM_S0_DFP2 + +#define ATOM_S0_DFP2I 0x00200000L +#define ATOM_S0_DFP2Ib2 0x20 + +#define ATOM_S2_DFP1I_DPMS_STATE ATOM_S2_DFP1_DPMS_STATE +#define ATOM_S2_DFP1X_DPMS_STATE ATOM_S2_DFP2_DPMS_STATE + +#define ATOM_S2_DFP2I_DPMS_STATE 0x02000000L +#define ATOM_S2_DFP2I_DPMS_STATEb3 0x02 + +#define ATOM_S3_DFP2I_ACTIVEb1 0x02 + +#define ATOM_S3_DFP1I_ACTIVE ATOM_S3_DFP1_ACTIVE +#define ATOM_S3_DFP1X_ACTIVE ATOM_S3_DFP2_ACTIVE + +#define ATOM_S3_DFP2I_ACTIVE 0x00000200L + +#define ATOM_S3_DFP1I_CRTC_ACTIVE ATOM_S3_DFP1_CRTC_ACTIVE +#define ATOM_S3_DFP1X_CRTC_ACTIVE ATOM_S3_DFP2_CRTC_ACTIVE +#define ATOM_S3_DFP2I_CRTC_ACTIVE 0x02000000L + +#define ATOM_S3_DFP2I_CRTC_ACTIVEb3 0x02 +#define ATOM_S5_DOS_REQ_DFP2Ib1 0x02 + +#define ATOM_S5_DOS_REQ_DFP2I 0x0200 +#define ATOM_S6_ACC_REQ_DFP1I ATOM_S6_ACC_REQ_DFP1 +#define ATOM_S6_ACC_REQ_DFP1X ATOM_S6_ACC_REQ_DFP2 + +#define ATOM_S6_ACC_REQ_DFP2Ib3 0x02 +#define ATOM_S6_ACC_REQ_DFP2I 0x02000000L + +#define TMDS1XEncoderControl DVOEncoderControl +#define DFP1XOutputControl DVOOutputControl + +#define ExternalDFPOutputControl DFP1XOutputControl +#define EnableExternalTMDS_Encoder TMDS1XEncoderControl + +#define DFP1IOutputControl TMDSAOutputControl +#define DFP2IOutputControl LVTMAOutputControl + +#define DAC1_ENCODER_CONTROL_PARAMETERS DAC_ENCODER_CONTROL_PARAMETERS +#define DAC1_ENCODER_CONTROL_PS_ALLOCATION DAC_ENCODER_CONTROL_PS_ALLOCATION + +#define DAC2_ENCODER_CONTROL_PARAMETERS DAC_ENCODER_CONTROL_PARAMETERS +#define DAC2_ENCODER_CONTROL_PS_ALLOCATION DAC_ENCODER_CONTROL_PS_ALLOCATION + +#define ucDac1Standard ucDacStandard +#define ucDac2Standard ucDacStandard + +#define TMDS1EncoderControl TMDSAEncoderControl +#define TMDS2EncoderControl LVTMAEncoderControl + +#define DFP1OutputControl TMDSAOutputControl +#define DFP2OutputControl LVTMAOutputControl +#define CRT1OutputControl DAC1OutputControl +#define CRT2OutputControl DAC2OutputControl + +//These two lines will be removed for sure in a few days, will follow up with Michael V. +#define EnableLVDS_SS EnableSpreadSpectrumOnPPLL +#define ENABLE_LVDS_SS_PARAMETERS_V3 ENABLE_SPREAD_SPECTRUM_ON_PPLL + +//#define ATOM_S2_CRT1_DPMS_STATE 0x00010000L +//#define ATOM_S2_LCD1_DPMS_STATE ATOM_S2_CRT1_DPMS_STATE +//#define ATOM_S2_TV1_DPMS_STATE ATOM_S2_CRT1_DPMS_STATE +//#define ATOM_S2_DFP1_DPMS_STATE ATOM_S2_CRT1_DPMS_STATE +//#define ATOM_S2_CRT2_DPMS_STATE ATOM_S2_CRT1_DPMS_STATE + +#define ATOM_S6_ACC_REQ_TV2 0x00400000L +#define ATOM_DEVICE_TV2_INDEX 0x00000006 +#define ATOM_DEVICE_TV2_SUPPORT (0x1L << ATOM_DEVICE_TV2_INDEX) +#define ATOM_S0_TV2 0x00100000L +#define ATOM_S3_TV2_ACTIVE ATOM_S3_DFP6_ACTIVE +#define ATOM_S3_TV2_CRTC_ACTIVE ATOM_S3_DFP6_CRTC_ACTIVE + +// +#define ATOM_S2_CRT1_DPMS_STATE 0x00010000L +#define ATOM_S2_LCD1_DPMS_STATE 0x00020000L +#define ATOM_S2_TV1_DPMS_STATE 0x00040000L +#define ATOM_S2_DFP1_DPMS_STATE 0x00080000L +#define ATOM_S2_CRT2_DPMS_STATE 0x00100000L +#define ATOM_S2_LCD2_DPMS_STATE 0x00200000L +#define ATOM_S2_TV2_DPMS_STATE 0x00400000L +#define ATOM_S2_DFP2_DPMS_STATE 0x00800000L +#define ATOM_S2_CV_DPMS_STATE 0x01000000L +#define ATOM_S2_DFP3_DPMS_STATE 0x02000000L +#define ATOM_S2_DFP4_DPMS_STATE 0x04000000L +#define ATOM_S2_DFP5_DPMS_STATE 0x08000000L + +#define ATOM_S2_CRT1_DPMS_STATEb2 0x01 +#define ATOM_S2_LCD1_DPMS_STATEb2 0x02 +#define ATOM_S2_TV1_DPMS_STATEb2 0x04 +#define ATOM_S2_DFP1_DPMS_STATEb2 0x08 +#define ATOM_S2_CRT2_DPMS_STATEb2 0x10 +#define ATOM_S2_LCD2_DPMS_STATEb2 0x20 +#define ATOM_S2_TV2_DPMS_STATEb2 0x40 +#define ATOM_S2_DFP2_DPMS_STATEb2 0x80 +#define ATOM_S2_CV_DPMS_STATEb3 0x01 +#define ATOM_S2_DFP3_DPMS_STATEb3 0x02 +#define ATOM_S2_DFP4_DPMS_STATEb3 0x04 +#define ATOM_S2_DFP5_DPMS_STATEb3 0x08 + +#define ATOM_S3_ASIC_GUI_ENGINE_HUNGb3 0x20 +#define ATOM_S3_ALLOW_FAST_PWR_SWITCHb3 0x40 +#define ATOM_S3_RQST_GPU_USE_MIN_PWRb3 0x80 + +/*********************************************************************************/ + +#pragma pack() // BIOS data must use byte aligment + +#endif /* _ATOMBIOS_H */ diff --git a/drivers/gpu/drm/radeon/atombios_crtc.c b/drivers/gpu/drm/radeon/atombios_crtc.c new file mode 100644 index 00000000..071ded11 --- /dev/null +++ b/drivers/gpu/drm/radeon/atombios_crtc.c @@ -0,0 +1,1636 @@ +/* + * Copyright 2007-8 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + */ +#include +#include +#include +#include +#include "radeon.h" +#include "atom.h" +#include "atom-bits.h" + +static void atombios_overscan_setup(struct drm_crtc *crtc, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + SET_CRTC_OVERSCAN_PS_ALLOCATION args; + int index = GetIndexIntoMasterTable(COMMAND, SetCRTC_OverScan); + int a1, a2; + + memset(&args, 0, sizeof(args)); + + args.ucCRTC = radeon_crtc->crtc_id; + + switch (radeon_crtc->rmx_type) { + case RMX_CENTER: + args.usOverscanTop = cpu_to_le16((adjusted_mode->crtc_vdisplay - mode->crtc_vdisplay) / 2); + args.usOverscanBottom = cpu_to_le16((adjusted_mode->crtc_vdisplay - mode->crtc_vdisplay) / 2); + args.usOverscanLeft = cpu_to_le16((adjusted_mode->crtc_hdisplay - mode->crtc_hdisplay) / 2); + args.usOverscanRight = cpu_to_le16((adjusted_mode->crtc_hdisplay - mode->crtc_hdisplay) / 2); + break; + case RMX_ASPECT: + a1 = mode->crtc_vdisplay * adjusted_mode->crtc_hdisplay; + a2 = adjusted_mode->crtc_vdisplay * mode->crtc_hdisplay; + + if (a1 > a2) { + args.usOverscanLeft = cpu_to_le16((adjusted_mode->crtc_hdisplay - (a2 / mode->crtc_vdisplay)) / 2); + args.usOverscanRight = cpu_to_le16((adjusted_mode->crtc_hdisplay - (a2 / mode->crtc_vdisplay)) / 2); + } else if (a2 > a1) { + args.usOverscanTop = cpu_to_le16((adjusted_mode->crtc_vdisplay - (a1 / mode->crtc_hdisplay)) / 2); + args.usOverscanBottom = cpu_to_le16((adjusted_mode->crtc_vdisplay - (a1 / mode->crtc_hdisplay)) / 2); + } + break; + case RMX_FULL: + default: + args.usOverscanRight = cpu_to_le16(radeon_crtc->h_border); + args.usOverscanLeft = cpu_to_le16(radeon_crtc->h_border); + args.usOverscanBottom = cpu_to_le16(radeon_crtc->v_border); + args.usOverscanTop = cpu_to_le16(radeon_crtc->v_border); + break; + } + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); +} + +static void atombios_scaler_setup(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + ENABLE_SCALER_PS_ALLOCATION args; + int index = GetIndexIntoMasterTable(COMMAND, EnableScaler); + + /* fixme - fill in enc_priv for atom dac */ + enum radeon_tv_std tv_std = TV_STD_NTSC; + bool is_tv = false, is_cv = false; + struct drm_encoder *encoder; + + if (!ASIC_IS_AVIVO(rdev) && radeon_crtc->crtc_id) + return; + + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + /* find tv std */ + if (encoder->crtc == crtc) { + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + if (radeon_encoder->active_device & ATOM_DEVICE_TV_SUPPORT) { + struct radeon_encoder_atom_dac *tv_dac = radeon_encoder->enc_priv; + tv_std = tv_dac->tv_std; + is_tv = true; + } + } + } + + memset(&args, 0, sizeof(args)); + + args.ucScaler = radeon_crtc->crtc_id; + + if (is_tv) { + switch (tv_std) { + case TV_STD_NTSC: + default: + args.ucTVStandard = ATOM_TV_NTSC; + break; + case TV_STD_PAL: + args.ucTVStandard = ATOM_TV_PAL; + break; + case TV_STD_PAL_M: + args.ucTVStandard = ATOM_TV_PALM; + break; + case TV_STD_PAL_60: + args.ucTVStandard = ATOM_TV_PAL60; + break; + case TV_STD_NTSC_J: + args.ucTVStandard = ATOM_TV_NTSCJ; + break; + case TV_STD_SCART_PAL: + args.ucTVStandard = ATOM_TV_PAL; /* ??? */ + break; + case TV_STD_SECAM: + args.ucTVStandard = ATOM_TV_SECAM; + break; + case TV_STD_PAL_CN: + args.ucTVStandard = ATOM_TV_PALCN; + break; + } + args.ucEnable = SCALER_ENABLE_MULTITAP_MODE; + } else if (is_cv) { + args.ucTVStandard = ATOM_TV_CV; + args.ucEnable = SCALER_ENABLE_MULTITAP_MODE; + } else { + switch (radeon_crtc->rmx_type) { + case RMX_FULL: + args.ucEnable = ATOM_SCALER_EXPANSION; + break; + case RMX_CENTER: + args.ucEnable = ATOM_SCALER_CENTER; + break; + case RMX_ASPECT: + args.ucEnable = ATOM_SCALER_EXPANSION; + break; + default: + if (ASIC_IS_AVIVO(rdev)) + args.ucEnable = ATOM_SCALER_DISABLE; + else + args.ucEnable = ATOM_SCALER_CENTER; + break; + } + } + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); + if ((is_tv || is_cv) + && rdev->family >= CHIP_RV515 && rdev->family <= CHIP_R580) { + atom_rv515_force_tv_scaler(rdev, radeon_crtc); + } +} + +static void atombios_lock_crtc(struct drm_crtc *crtc, int lock) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + int index = + GetIndexIntoMasterTable(COMMAND, UpdateCRTC_DoubleBufferRegisters); + ENABLE_CRTC_PS_ALLOCATION args; + + memset(&args, 0, sizeof(args)); + + args.ucCRTC = radeon_crtc->crtc_id; + args.ucEnable = lock; + + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); +} + +static void atombios_enable_crtc(struct drm_crtc *crtc, int state) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + int index = GetIndexIntoMasterTable(COMMAND, EnableCRTC); + ENABLE_CRTC_PS_ALLOCATION args; + + memset(&args, 0, sizeof(args)); + + args.ucCRTC = radeon_crtc->crtc_id; + args.ucEnable = state; + + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); +} + +static void atombios_enable_crtc_memreq(struct drm_crtc *crtc, int state) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + int index = GetIndexIntoMasterTable(COMMAND, EnableCRTCMemReq); + ENABLE_CRTC_PS_ALLOCATION args; + + memset(&args, 0, sizeof(args)); + + args.ucCRTC = radeon_crtc->crtc_id; + args.ucEnable = state; + + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); +} + +static void atombios_blank_crtc(struct drm_crtc *crtc, int state) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + int index = GetIndexIntoMasterTable(COMMAND, BlankCRTC); + BLANK_CRTC_PS_ALLOCATION args; + + memset(&args, 0, sizeof(args)); + + args.ucCRTC = radeon_crtc->crtc_id; + args.ucBlanking = state; + + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); +} + +void atombios_crtc_dpms(struct drm_crtc *crtc, int mode) +{ + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + + switch (mode) { + case DRM_MODE_DPMS_ON: + radeon_crtc->enabled = true; + /* adjust pm to dpms changes BEFORE enabling crtcs */ + radeon_pm_compute_clocks(rdev); + atombios_enable_crtc(crtc, ATOM_ENABLE); + if (ASIC_IS_DCE3(rdev)) + atombios_enable_crtc_memreq(crtc, ATOM_ENABLE); + atombios_blank_crtc(crtc, ATOM_DISABLE); + drm_vblank_post_modeset(dev, radeon_crtc->crtc_id); + radeon_crtc_load_lut(crtc); + break; + case DRM_MODE_DPMS_STANDBY: + case DRM_MODE_DPMS_SUSPEND: + case DRM_MODE_DPMS_OFF: + drm_vblank_pre_modeset(dev, radeon_crtc->crtc_id); + if (radeon_crtc->enabled) + atombios_blank_crtc(crtc, ATOM_ENABLE); + if (ASIC_IS_DCE3(rdev)) + atombios_enable_crtc_memreq(crtc, ATOM_DISABLE); + atombios_enable_crtc(crtc, ATOM_DISABLE); + radeon_crtc->enabled = false; + /* adjust pm to dpms changes AFTER disabling crtcs */ + radeon_pm_compute_clocks(rdev); + break; + } +} + +static void +atombios_set_crtc_dtd_timing(struct drm_crtc *crtc, + struct drm_display_mode *mode) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + SET_CRTC_USING_DTD_TIMING_PARAMETERS args; + int index = GetIndexIntoMasterTable(COMMAND, SetCRTC_UsingDTDTiming); + u16 misc = 0; + + memset(&args, 0, sizeof(args)); + args.usH_Size = cpu_to_le16(mode->crtc_hdisplay - (radeon_crtc->h_border * 2)); + args.usH_Blanking_Time = + cpu_to_le16(mode->crtc_hblank_end - mode->crtc_hdisplay + (radeon_crtc->h_border * 2)); + args.usV_Size = cpu_to_le16(mode->crtc_vdisplay - (radeon_crtc->v_border * 2)); + args.usV_Blanking_Time = + cpu_to_le16(mode->crtc_vblank_end - mode->crtc_vdisplay + (radeon_crtc->v_border * 2)); + args.usH_SyncOffset = + cpu_to_le16(mode->crtc_hsync_start - mode->crtc_hdisplay + radeon_crtc->h_border); + args.usH_SyncWidth = + cpu_to_le16(mode->crtc_hsync_end - mode->crtc_hsync_start); + args.usV_SyncOffset = + cpu_to_le16(mode->crtc_vsync_start - mode->crtc_vdisplay + radeon_crtc->v_border); + args.usV_SyncWidth = + cpu_to_le16(mode->crtc_vsync_end - mode->crtc_vsync_start); + args.ucH_Border = radeon_crtc->h_border; + args.ucV_Border = radeon_crtc->v_border; + + if (mode->flags & DRM_MODE_FLAG_NVSYNC) + misc |= ATOM_VSYNC_POLARITY; + if (mode->flags & DRM_MODE_FLAG_NHSYNC) + misc |= ATOM_HSYNC_POLARITY; + if (mode->flags & DRM_MODE_FLAG_CSYNC) + misc |= ATOM_COMPOSITESYNC; + if (mode->flags & DRM_MODE_FLAG_INTERLACE) + misc |= ATOM_INTERLACE; + if (mode->flags & DRM_MODE_FLAG_DBLSCAN) + misc |= ATOM_DOUBLE_CLOCK_MODE; + + args.susModeMiscInfo.usAccess = cpu_to_le16(misc); + args.ucCRTC = radeon_crtc->crtc_id; + + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); +} + +static void atombios_crtc_set_timing(struct drm_crtc *crtc, + struct drm_display_mode *mode) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + SET_CRTC_TIMING_PARAMETERS_PS_ALLOCATION args; + int index = GetIndexIntoMasterTable(COMMAND, SetCRTC_Timing); + u16 misc = 0; + + memset(&args, 0, sizeof(args)); + args.usH_Total = cpu_to_le16(mode->crtc_htotal); + args.usH_Disp = cpu_to_le16(mode->crtc_hdisplay); + args.usH_SyncStart = cpu_to_le16(mode->crtc_hsync_start); + args.usH_SyncWidth = + cpu_to_le16(mode->crtc_hsync_end - mode->crtc_hsync_start); + args.usV_Total = cpu_to_le16(mode->crtc_vtotal); + args.usV_Disp = cpu_to_le16(mode->crtc_vdisplay); + args.usV_SyncStart = cpu_to_le16(mode->crtc_vsync_start); + args.usV_SyncWidth = + cpu_to_le16(mode->crtc_vsync_end - mode->crtc_vsync_start); + + args.ucOverscanRight = radeon_crtc->h_border; + args.ucOverscanLeft = radeon_crtc->h_border; + args.ucOverscanBottom = radeon_crtc->v_border; + args.ucOverscanTop = radeon_crtc->v_border; + + if (mode->flags & DRM_MODE_FLAG_NVSYNC) + misc |= ATOM_VSYNC_POLARITY; + if (mode->flags & DRM_MODE_FLAG_NHSYNC) + misc |= ATOM_HSYNC_POLARITY; + if (mode->flags & DRM_MODE_FLAG_CSYNC) + misc |= ATOM_COMPOSITESYNC; + if (mode->flags & DRM_MODE_FLAG_INTERLACE) + misc |= ATOM_INTERLACE; + if (mode->flags & DRM_MODE_FLAG_DBLSCAN) + misc |= ATOM_DOUBLE_CLOCK_MODE; + + args.susModeMiscInfo.usAccess = cpu_to_le16(misc); + args.ucCRTC = radeon_crtc->crtc_id; + + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); +} + +static void atombios_disable_ss(struct drm_crtc *crtc) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + u32 ss_cntl; + + if (ASIC_IS_DCE4(rdev)) { + switch (radeon_crtc->pll_id) { + case ATOM_PPLL1: + ss_cntl = RREG32(EVERGREEN_P1PLL_SS_CNTL); + ss_cntl &= ~EVERGREEN_PxPLL_SS_EN; + WREG32(EVERGREEN_P1PLL_SS_CNTL, ss_cntl); + break; + case ATOM_PPLL2: + ss_cntl = RREG32(EVERGREEN_P2PLL_SS_CNTL); + ss_cntl &= ~EVERGREEN_PxPLL_SS_EN; + WREG32(EVERGREEN_P2PLL_SS_CNTL, ss_cntl); + break; + case ATOM_DCPLL: + case ATOM_PPLL_INVALID: + return; + } + } else if (ASIC_IS_AVIVO(rdev)) { + switch (radeon_crtc->pll_id) { + case ATOM_PPLL1: + ss_cntl = RREG32(AVIVO_P1PLL_INT_SS_CNTL); + ss_cntl &= ~1; + WREG32(AVIVO_P1PLL_INT_SS_CNTL, ss_cntl); + break; + case ATOM_PPLL2: + ss_cntl = RREG32(AVIVO_P2PLL_INT_SS_CNTL); + ss_cntl &= ~1; + WREG32(AVIVO_P2PLL_INT_SS_CNTL, ss_cntl); + break; + case ATOM_DCPLL: + case ATOM_PPLL_INVALID: + return; + } + } +} + + +union atom_enable_ss { + ENABLE_LVDS_SS_PARAMETERS lvds_ss; + ENABLE_LVDS_SS_PARAMETERS_V2 lvds_ss_2; + ENABLE_SPREAD_SPECTRUM_ON_PPLL_PS_ALLOCATION v1; + ENABLE_SPREAD_SPECTRUM_ON_PPLL_V2 v2; + ENABLE_SPREAD_SPECTRUM_ON_PPLL_V3 v3; +}; + +static void atombios_crtc_program_ss(struct drm_crtc *crtc, + int enable, + int pll_id, + struct radeon_atom_ss *ss) +{ + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + int index = GetIndexIntoMasterTable(COMMAND, EnableSpreadSpectrumOnPPLL); + union atom_enable_ss args; + + memset(&args, 0, sizeof(args)); + + if (ASIC_IS_DCE5(rdev)) { + args.v3.usSpreadSpectrumAmountFrac = cpu_to_le16(0); + args.v3.ucSpreadSpectrumType = ss->type & ATOM_SS_CENTRE_SPREAD_MODE_MASK; + switch (pll_id) { + case ATOM_PPLL1: + args.v3.ucSpreadSpectrumType |= ATOM_PPLL_SS_TYPE_V3_P1PLL; + args.v3.usSpreadSpectrumAmount = cpu_to_le16(ss->amount); + args.v3.usSpreadSpectrumStep = cpu_to_le16(ss->step); + break; + case ATOM_PPLL2: + args.v3.ucSpreadSpectrumType |= ATOM_PPLL_SS_TYPE_V3_P2PLL; + args.v3.usSpreadSpectrumAmount = cpu_to_le16(ss->amount); + args.v3.usSpreadSpectrumStep = cpu_to_le16(ss->step); + break; + case ATOM_DCPLL: + args.v3.ucSpreadSpectrumType |= ATOM_PPLL_SS_TYPE_V3_DCPLL; + args.v3.usSpreadSpectrumAmount = cpu_to_le16(0); + args.v3.usSpreadSpectrumStep = cpu_to_le16(0); + break; + case ATOM_PPLL_INVALID: + return; + } + args.v3.ucEnable = enable; + if ((ss->percentage == 0) || (ss->type & ATOM_EXTERNAL_SS_MASK)) + args.v3.ucEnable = ATOM_DISABLE; + } else if (ASIC_IS_DCE4(rdev)) { + args.v2.usSpreadSpectrumPercentage = cpu_to_le16(ss->percentage); + args.v2.ucSpreadSpectrumType = ss->type & ATOM_SS_CENTRE_SPREAD_MODE_MASK; + switch (pll_id) { + case ATOM_PPLL1: + args.v2.ucSpreadSpectrumType |= ATOM_PPLL_SS_TYPE_V2_P1PLL; + args.v2.usSpreadSpectrumAmount = cpu_to_le16(ss->amount); + args.v2.usSpreadSpectrumStep = cpu_to_le16(ss->step); + break; + case ATOM_PPLL2: + args.v2.ucSpreadSpectrumType |= ATOM_PPLL_SS_TYPE_V2_P2PLL; + args.v2.usSpreadSpectrumAmount = cpu_to_le16(ss->amount); + args.v2.usSpreadSpectrumStep = cpu_to_le16(ss->step); + break; + case ATOM_DCPLL: + args.v2.ucSpreadSpectrumType |= ATOM_PPLL_SS_TYPE_V2_DCPLL; + args.v2.usSpreadSpectrumAmount = cpu_to_le16(0); + args.v2.usSpreadSpectrumStep = cpu_to_le16(0); + break; + case ATOM_PPLL_INVALID: + return; + } + args.v2.ucEnable = enable; + if ((ss->percentage == 0) || (ss->type & ATOM_EXTERNAL_SS_MASK)) + args.v2.ucEnable = ATOM_DISABLE; + } else if (ASIC_IS_DCE3(rdev)) { + args.v1.usSpreadSpectrumPercentage = cpu_to_le16(ss->percentage); + args.v1.ucSpreadSpectrumType = ss->type & ATOM_SS_CENTRE_SPREAD_MODE_MASK; + args.v1.ucSpreadSpectrumStep = ss->step; + args.v1.ucSpreadSpectrumDelay = ss->delay; + args.v1.ucSpreadSpectrumRange = ss->range; + args.v1.ucPpll = pll_id; + args.v1.ucEnable = enable; + } else if (ASIC_IS_AVIVO(rdev)) { + if ((enable == ATOM_DISABLE) || (ss->percentage == 0) || + (ss->type & ATOM_EXTERNAL_SS_MASK)) { + atombios_disable_ss(crtc); + return; + } + args.lvds_ss_2.usSpreadSpectrumPercentage = cpu_to_le16(ss->percentage); + args.lvds_ss_2.ucSpreadSpectrumType = ss->type & ATOM_SS_CENTRE_SPREAD_MODE_MASK; + args.lvds_ss_2.ucSpreadSpectrumStep = ss->step; + args.lvds_ss_2.ucSpreadSpectrumDelay = ss->delay; + args.lvds_ss_2.ucSpreadSpectrumRange = ss->range; + args.lvds_ss_2.ucEnable = enable; + } else { + if ((enable == ATOM_DISABLE) || (ss->percentage == 0) || + (ss->type & ATOM_EXTERNAL_SS_MASK)) { + atombios_disable_ss(crtc); + return; + } + args.lvds_ss.usSpreadSpectrumPercentage = cpu_to_le16(ss->percentage); + args.lvds_ss.ucSpreadSpectrumType = ss->type & ATOM_SS_CENTRE_SPREAD_MODE_MASK; + args.lvds_ss.ucSpreadSpectrumStepSize_Delay = (ss->step & 3) << 2; + args.lvds_ss.ucSpreadSpectrumStepSize_Delay |= (ss->delay & 7) << 4; + args.lvds_ss.ucEnable = enable; + } + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); +} + +union adjust_pixel_clock { + ADJUST_DISPLAY_PLL_PS_ALLOCATION v1; + ADJUST_DISPLAY_PLL_PS_ALLOCATION_V3 v3; +}; + +static u32 atombios_adjust_pll(struct drm_crtc *crtc, + struct drm_display_mode *mode, + struct radeon_pll *pll, + bool ss_enabled, + struct radeon_atom_ss *ss) +{ + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + struct drm_encoder *encoder = NULL; + struct radeon_encoder *radeon_encoder = NULL; + struct drm_connector *connector = NULL; + u32 adjusted_clock = mode->clock; + int encoder_mode = 0; + u32 dp_clock = mode->clock; + int bpc = 8; + + /* reset the pll flags */ + pll->flags = 0; + + if (ASIC_IS_AVIVO(rdev)) { + if ((rdev->family == CHIP_RS600) || + (rdev->family == CHIP_RS690) || + (rdev->family == CHIP_RS740)) + pll->flags |= (/*RADEON_PLL_USE_FRAC_FB_DIV |*/ + RADEON_PLL_PREFER_CLOSEST_LOWER); + + if (ASIC_IS_DCE32(rdev) && mode->clock > 200000) /* range limits??? */ + pll->flags |= RADEON_PLL_PREFER_HIGH_FB_DIV; + else + pll->flags |= RADEON_PLL_PREFER_LOW_REF_DIV; + + if (rdev->family < CHIP_RV770) + pll->flags |= RADEON_PLL_PREFER_MINM_OVER_MAXP; + } else { + pll->flags |= RADEON_PLL_LEGACY; + + if (mode->clock > 200000) /* range limits??? */ + pll->flags |= RADEON_PLL_PREFER_HIGH_FB_DIV; + else + pll->flags |= RADEON_PLL_PREFER_LOW_REF_DIV; + } + + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + if (encoder->crtc == crtc) { + radeon_encoder = to_radeon_encoder(encoder); + connector = radeon_get_connector_for_encoder(encoder); + if (connector) + bpc = connector->display_info.bpc; + encoder_mode = atombios_get_encoder_mode(encoder); + if ((radeon_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT | ATOM_DEVICE_DFP_SUPPORT)) || + radeon_encoder_is_dp_bridge(encoder)) { + if (connector) { + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + struct radeon_connector_atom_dig *dig_connector = + radeon_connector->con_priv; + + dp_clock = dig_connector->dp_clock; + } + } + + /* use recommended ref_div for ss */ + if (radeon_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) { + if (ss_enabled) { + if (ss->refdiv) { + pll->flags |= RADEON_PLL_USE_REF_DIV; + pll->reference_div = ss->refdiv; + if (ASIC_IS_AVIVO(rdev)) + pll->flags |= RADEON_PLL_USE_FRAC_FB_DIV; + } + } + } + + if (ASIC_IS_AVIVO(rdev)) { + /* DVO wants 2x pixel clock if the DVO chip is in 12 bit mode */ + if (radeon_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1) + adjusted_clock = mode->clock * 2; + if (radeon_encoder->active_device & (ATOM_DEVICE_TV_SUPPORT)) + pll->flags |= RADEON_PLL_PREFER_CLOSEST_LOWER; + if (radeon_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) + pll->flags |= RADEON_PLL_IS_LCD; + } else { + if (encoder->encoder_type != DRM_MODE_ENCODER_DAC) + pll->flags |= RADEON_PLL_NO_ODD_POST_DIV; + if (encoder->encoder_type == DRM_MODE_ENCODER_LVDS) + pll->flags |= RADEON_PLL_USE_REF_DIV; + } + break; + } + } + + /* DCE3+ has an AdjustDisplayPll that will adjust the pixel clock + * accordingly based on the encoder/transmitter to work around + * special hw requirements. + */ + if (ASIC_IS_DCE3(rdev)) { + union adjust_pixel_clock args; + u8 frev, crev; + int index; + + index = GetIndexIntoMasterTable(COMMAND, AdjustDisplayPll); + if (!atom_parse_cmd_header(rdev->mode_info.atom_context, index, &frev, + &crev)) + return adjusted_clock; + + memset(&args, 0, sizeof(args)); + + switch (frev) { + case 1: + switch (crev) { + case 1: + case 2: + args.v1.usPixelClock = cpu_to_le16(mode->clock / 10); + args.v1.ucTransmitterID = radeon_encoder->encoder_id; + args.v1.ucEncodeMode = encoder_mode; + if (ss_enabled && ss->percentage) + args.v1.ucConfig |= + ADJUST_DISPLAY_CONFIG_SS_ENABLE; + + atom_execute_table(rdev->mode_info.atom_context, + index, (uint32_t *)&args); + adjusted_clock = le16_to_cpu(args.v1.usPixelClock) * 10; + break; + case 3: + args.v3.sInput.usPixelClock = cpu_to_le16(mode->clock / 10); + args.v3.sInput.ucTransmitterID = radeon_encoder->encoder_id; + args.v3.sInput.ucEncodeMode = encoder_mode; + args.v3.sInput.ucDispPllConfig = 0; + if (ss_enabled && ss->percentage) + args.v3.sInput.ucDispPllConfig |= + DISPPLL_CONFIG_SS_ENABLE; + if (radeon_encoder->devices & (ATOM_DEVICE_DFP_SUPPORT) || + radeon_encoder_is_dp_bridge(encoder)) { + struct radeon_encoder_atom_dig *dig = radeon_encoder->enc_priv; + if (encoder_mode == ATOM_ENCODER_MODE_DP) { + args.v3.sInput.ucDispPllConfig |= + DISPPLL_CONFIG_COHERENT_MODE; + /* 16200 or 27000 */ + args.v3.sInput.usPixelClock = cpu_to_le16(dp_clock / 10); + } else { + if (encoder_mode == ATOM_ENCODER_MODE_HDMI) { + /* deep color support */ + args.v3.sInput.usPixelClock = + cpu_to_le16((mode->clock * bpc / 8) / 10); + } + if (dig->coherent_mode) + args.v3.sInput.ucDispPllConfig |= + DISPPLL_CONFIG_COHERENT_MODE; + if (mode->clock > 165000) + args.v3.sInput.ucDispPllConfig |= + DISPPLL_CONFIG_DUAL_LINK; + } + } else if (radeon_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) { + if (encoder_mode == ATOM_ENCODER_MODE_DP) { + args.v3.sInput.ucDispPllConfig |= + DISPPLL_CONFIG_COHERENT_MODE; + /* 16200 or 27000 */ + args.v3.sInput.usPixelClock = cpu_to_le16(dp_clock / 10); + } else if (encoder_mode != ATOM_ENCODER_MODE_LVDS) { + if (mode->clock > 165000) + args.v3.sInput.ucDispPllConfig |= + DISPPLL_CONFIG_DUAL_LINK; + } + } + if (radeon_encoder_is_dp_bridge(encoder)) { + struct drm_encoder *ext_encoder = radeon_atom_get_external_encoder(encoder); + struct radeon_encoder *ext_radeon_encoder = to_radeon_encoder(ext_encoder); + args.v3.sInput.ucExtTransmitterID = ext_radeon_encoder->encoder_id; + } else + args.v3.sInput.ucExtTransmitterID = 0; + + atom_execute_table(rdev->mode_info.atom_context, + index, (uint32_t *)&args); + adjusted_clock = le32_to_cpu(args.v3.sOutput.ulDispPllFreq) * 10; + if (args.v3.sOutput.ucRefDiv) { + pll->flags |= RADEON_PLL_USE_FRAC_FB_DIV; + pll->flags |= RADEON_PLL_USE_REF_DIV; + pll->reference_div = args.v3.sOutput.ucRefDiv; + } + if (args.v3.sOutput.ucPostDiv) { + pll->flags |= RADEON_PLL_USE_FRAC_FB_DIV; + pll->flags |= RADEON_PLL_USE_POST_DIV; + pll->post_div = args.v3.sOutput.ucPostDiv; + } + break; + default: + DRM_ERROR("Unknown table version %d %d\n", frev, crev); + return adjusted_clock; + } + break; + default: + DRM_ERROR("Unknown table version %d %d\n", frev, crev); + return adjusted_clock; + } + } + return adjusted_clock; +} + +union set_pixel_clock { + SET_PIXEL_CLOCK_PS_ALLOCATION base; + PIXEL_CLOCK_PARAMETERS v1; + PIXEL_CLOCK_PARAMETERS_V2 v2; + PIXEL_CLOCK_PARAMETERS_V3 v3; + PIXEL_CLOCK_PARAMETERS_V5 v5; + PIXEL_CLOCK_PARAMETERS_V6 v6; +}; + +/* on DCE5, make sure the voltage is high enough to support the + * required disp clk. + */ +static void atombios_crtc_set_dcpll(struct drm_crtc *crtc, + u32 dispclk) +{ + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + u8 frev, crev; + int index; + union set_pixel_clock args; + + memset(&args, 0, sizeof(args)); + + index = GetIndexIntoMasterTable(COMMAND, SetPixelClock); + if (!atom_parse_cmd_header(rdev->mode_info.atom_context, index, &frev, + &crev)) + return; + + switch (frev) { + case 1: + switch (crev) { + case 5: + /* if the default dcpll clock is specified, + * SetPixelClock provides the dividers + */ + args.v5.ucCRTC = ATOM_CRTC_INVALID; + args.v5.usPixelClock = cpu_to_le16(dispclk); + args.v5.ucPpll = ATOM_DCPLL; + break; + case 6: + /* if the default dcpll clock is specified, + * SetPixelClock provides the dividers + */ + args.v6.ulDispEngClkFreq = cpu_to_le32(dispclk); + args.v6.ucPpll = ATOM_DCPLL; + break; + default: + DRM_ERROR("Unknown table version %d %d\n", frev, crev); + return; + } + break; + default: + DRM_ERROR("Unknown table version %d %d\n", frev, crev); + return; + } + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); +} + +static void atombios_crtc_program_pll(struct drm_crtc *crtc, + int crtc_id, + int pll_id, + u32 encoder_mode, + u32 encoder_id, + u32 clock, + u32 ref_div, + u32 fb_div, + u32 frac_fb_div, + u32 post_div, + int bpc, + bool ss_enabled, + struct radeon_atom_ss *ss) +{ + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + u8 frev, crev; + int index = GetIndexIntoMasterTable(COMMAND, SetPixelClock); + union set_pixel_clock args; + + memset(&args, 0, sizeof(args)); + + if (!atom_parse_cmd_header(rdev->mode_info.atom_context, index, &frev, + &crev)) + return; + + switch (frev) { + case 1: + switch (crev) { + case 1: + if (clock == ATOM_DISABLE) + return; + args.v1.usPixelClock = cpu_to_le16(clock / 10); + args.v1.usRefDiv = cpu_to_le16(ref_div); + args.v1.usFbDiv = cpu_to_le16(fb_div); + args.v1.ucFracFbDiv = frac_fb_div; + args.v1.ucPostDiv = post_div; + args.v1.ucPpll = pll_id; + args.v1.ucCRTC = crtc_id; + args.v1.ucRefDivSrc = 1; + break; + case 2: + args.v2.usPixelClock = cpu_to_le16(clock / 10); + args.v2.usRefDiv = cpu_to_le16(ref_div); + args.v2.usFbDiv = cpu_to_le16(fb_div); + args.v2.ucFracFbDiv = frac_fb_div; + args.v2.ucPostDiv = post_div; + args.v2.ucPpll = pll_id; + args.v2.ucCRTC = crtc_id; + args.v2.ucRefDivSrc = 1; + break; + case 3: + args.v3.usPixelClock = cpu_to_le16(clock / 10); + args.v3.usRefDiv = cpu_to_le16(ref_div); + args.v3.usFbDiv = cpu_to_le16(fb_div); + args.v3.ucFracFbDiv = frac_fb_div; + args.v3.ucPostDiv = post_div; + args.v3.ucPpll = pll_id; + args.v3.ucMiscInfo = (pll_id << 2); + if (ss_enabled && (ss->type & ATOM_EXTERNAL_SS_MASK)) + args.v3.ucMiscInfo |= PIXEL_CLOCK_MISC_REF_DIV_SRC; + args.v3.ucTransmitterId = encoder_id; + args.v3.ucEncoderMode = encoder_mode; + break; + case 5: + args.v5.ucCRTC = crtc_id; + args.v5.usPixelClock = cpu_to_le16(clock / 10); + args.v5.ucRefDiv = ref_div; + args.v5.usFbDiv = cpu_to_le16(fb_div); + args.v5.ulFbDivDecFrac = cpu_to_le32(frac_fb_div * 100000); + args.v5.ucPostDiv = post_div; + args.v5.ucMiscInfo = 0; /* HDMI depth, etc. */ + if (ss_enabled && (ss->type & ATOM_EXTERNAL_SS_MASK)) + args.v5.ucMiscInfo |= PIXEL_CLOCK_V5_MISC_REF_DIV_SRC; + switch (bpc) { + case 8: + default: + args.v5.ucMiscInfo |= PIXEL_CLOCK_V5_MISC_HDMI_24BPP; + break; + case 10: + args.v5.ucMiscInfo |= PIXEL_CLOCK_V5_MISC_HDMI_30BPP; + break; + } + args.v5.ucTransmitterID = encoder_id; + args.v5.ucEncoderMode = encoder_mode; + args.v5.ucPpll = pll_id; + break; + case 6: + args.v6.ulCrtcPclkFreq.ucCRTC = crtc_id; + args.v6.ulCrtcPclkFreq.ulPixelClock = cpu_to_le32(clock / 10); + args.v6.ucRefDiv = ref_div; + args.v6.usFbDiv = cpu_to_le16(fb_div); + args.v6.ulFbDivDecFrac = cpu_to_le32(frac_fb_div * 100000); + args.v6.ucPostDiv = post_div; + args.v6.ucMiscInfo = 0; /* HDMI depth, etc. */ + if (ss_enabled && (ss->type & ATOM_EXTERNAL_SS_MASK)) + args.v6.ucMiscInfo |= PIXEL_CLOCK_V6_MISC_REF_DIV_SRC; + switch (bpc) { + case 8: + default: + args.v6.ucMiscInfo |= PIXEL_CLOCK_V6_MISC_HDMI_24BPP; + break; + case 10: + args.v6.ucMiscInfo |= PIXEL_CLOCK_V6_MISC_HDMI_30BPP; + break; + case 12: + args.v6.ucMiscInfo |= PIXEL_CLOCK_V6_MISC_HDMI_36BPP; + break; + case 16: + args.v6.ucMiscInfo |= PIXEL_CLOCK_V6_MISC_HDMI_48BPP; + break; + } + args.v6.ucTransmitterID = encoder_id; + args.v6.ucEncoderMode = encoder_mode; + args.v6.ucPpll = pll_id; + break; + default: + DRM_ERROR("Unknown table version %d %d\n", frev, crev); + return; + } + break; + default: + DRM_ERROR("Unknown table version %d %d\n", frev, crev); + return; + } + + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); +} + +static void atombios_crtc_set_pll(struct drm_crtc *crtc, struct drm_display_mode *mode) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + struct drm_encoder *encoder = NULL; + struct radeon_encoder *radeon_encoder = NULL; + u32 pll_clock = mode->clock; + u32 ref_div = 0, fb_div = 0, frac_fb_div = 0, post_div = 0; + struct radeon_pll *pll; + u32 adjusted_clock; + int encoder_mode = 0; + struct radeon_atom_ss ss; + bool ss_enabled = false; + int bpc = 8; + + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + if (encoder->crtc == crtc) { + radeon_encoder = to_radeon_encoder(encoder); + encoder_mode = atombios_get_encoder_mode(encoder); + break; + } + } + + if (!radeon_encoder) + return; + + switch (radeon_crtc->pll_id) { + case ATOM_PPLL1: + pll = &rdev->clock.p1pll; + break; + case ATOM_PPLL2: + pll = &rdev->clock.p2pll; + break; + case ATOM_DCPLL: + case ATOM_PPLL_INVALID: + default: + pll = &rdev->clock.dcpll; + break; + } + + if (radeon_encoder->active_device & + (ATOM_DEVICE_LCD_SUPPORT | ATOM_DEVICE_DFP_SUPPORT)) { + struct radeon_encoder_atom_dig *dig = radeon_encoder->enc_priv; + struct drm_connector *connector = + radeon_get_connector_for_encoder(encoder); + struct radeon_connector *radeon_connector = + to_radeon_connector(connector); + struct radeon_connector_atom_dig *dig_connector = + radeon_connector->con_priv; + int dp_clock; + bpc = connector->display_info.bpc; + + switch (encoder_mode) { + case ATOM_ENCODER_MODE_DP: + /* DP/eDP */ + dp_clock = dig_connector->dp_clock / 10; + if (ASIC_IS_DCE4(rdev)) + ss_enabled = + radeon_atombios_get_asic_ss_info(rdev, &ss, + ASIC_INTERNAL_SS_ON_DP, + dp_clock); + else { + if (dp_clock == 16200) { + ss_enabled = + radeon_atombios_get_ppll_ss_info(rdev, &ss, + ATOM_DP_SS_ID2); + if (!ss_enabled) + ss_enabled = + radeon_atombios_get_ppll_ss_info(rdev, &ss, + ATOM_DP_SS_ID1); + } else + ss_enabled = + radeon_atombios_get_ppll_ss_info(rdev, &ss, + ATOM_DP_SS_ID1); + } + break; + case ATOM_ENCODER_MODE_LVDS: + if (ASIC_IS_DCE4(rdev)) + ss_enabled = radeon_atombios_get_asic_ss_info(rdev, &ss, + dig->lcd_ss_id, + mode->clock / 10); + else + ss_enabled = radeon_atombios_get_ppll_ss_info(rdev, &ss, + dig->lcd_ss_id); + break; + case ATOM_ENCODER_MODE_DVI: + if (ASIC_IS_DCE4(rdev)) + ss_enabled = + radeon_atombios_get_asic_ss_info(rdev, &ss, + ASIC_INTERNAL_SS_ON_TMDS, + mode->clock / 10); + break; + case ATOM_ENCODER_MODE_HDMI: + if (ASIC_IS_DCE4(rdev)) + ss_enabled = + radeon_atombios_get_asic_ss_info(rdev, &ss, + ASIC_INTERNAL_SS_ON_HDMI, + mode->clock / 10); + break; + default: + break; + } + } + + /* adjust pixel clock as needed */ + adjusted_clock = atombios_adjust_pll(crtc, mode, pll, ss_enabled, &ss); + + if (radeon_encoder->active_device & (ATOM_DEVICE_TV_SUPPORT)) + /* TV seems to prefer the legacy algo on some boards */ + radeon_compute_pll_legacy(pll, adjusted_clock, &pll_clock, &fb_div, &frac_fb_div, + &ref_div, &post_div); + else if (ASIC_IS_AVIVO(rdev)) + radeon_compute_pll_avivo(pll, adjusted_clock, &pll_clock, &fb_div, &frac_fb_div, + &ref_div, &post_div); + else + radeon_compute_pll_legacy(pll, adjusted_clock, &pll_clock, &fb_div, &frac_fb_div, + &ref_div, &post_div); + + atombios_crtc_program_ss(crtc, ATOM_DISABLE, radeon_crtc->pll_id, &ss); + + atombios_crtc_program_pll(crtc, radeon_crtc->crtc_id, radeon_crtc->pll_id, + encoder_mode, radeon_encoder->encoder_id, mode->clock, + ref_div, fb_div, frac_fb_div, post_div, bpc, ss_enabled, &ss); + + if (ss_enabled) { + /* calculate ss amount and step size */ + if (ASIC_IS_DCE4(rdev)) { + u32 step_size; + u32 amount = (((fb_div * 10) + frac_fb_div) * ss.percentage) / 10000; + ss.amount = (amount / 10) & ATOM_PPLL_SS_AMOUNT_V2_FBDIV_MASK; + ss.amount |= ((amount - (amount / 10)) << ATOM_PPLL_SS_AMOUNT_V2_NFRAC_SHIFT) & + ATOM_PPLL_SS_AMOUNT_V2_NFRAC_MASK; + if (ss.type & ATOM_PPLL_SS_TYPE_V2_CENTRE_SPREAD) + step_size = (4 * amount * ref_div * (ss.rate * 2048)) / + (125 * 25 * pll->reference_freq / 100); + else + step_size = (2 * amount * ref_div * (ss.rate * 2048)) / + (125 * 25 * pll->reference_freq / 100); + ss.step = step_size; + } + + atombios_crtc_program_ss(crtc, ATOM_ENABLE, radeon_crtc->pll_id, &ss); + } +} + +static int dce4_crtc_do_set_base(struct drm_crtc *crtc, + struct drm_framebuffer *fb, + int x, int y, int atomic) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_framebuffer *radeon_fb; + struct drm_framebuffer *target_fb; + struct drm_gem_object *obj; + struct radeon_bo *rbo; + uint64_t fb_location; + uint32_t fb_format, fb_pitch_pixels, tiling_flags; + u32 fb_swap = EVERGREEN_GRPH_ENDIAN_SWAP(EVERGREEN_GRPH_ENDIAN_NONE); + u32 tmp, viewport_w, viewport_h; + int r; + + /* no fb bound */ + if (!atomic && !crtc->fb) { + DRM_DEBUG_KMS("No FB bound\n"); + return 0; + } + + if (atomic) { + radeon_fb = to_radeon_framebuffer(fb); + target_fb = fb; + } + else { + radeon_fb = to_radeon_framebuffer(crtc->fb); + target_fb = crtc->fb; + } + + /* If atomic, assume fb object is pinned & idle & fenced and + * just update base pointers + */ + obj = radeon_fb->obj; + rbo = gem_to_radeon_bo(obj); + r = radeon_bo_reserve(rbo, false); + if (unlikely(r != 0)) + return r; + + if (atomic) + fb_location = radeon_bo_gpu_offset(rbo); + else { + r = radeon_bo_pin(rbo, RADEON_GEM_DOMAIN_VRAM, &fb_location); + if (unlikely(r != 0)) { + radeon_bo_unreserve(rbo); + return -EINVAL; + } + } + + radeon_bo_get_tiling_flags(rbo, &tiling_flags, NULL); + radeon_bo_unreserve(rbo); + + switch (target_fb->bits_per_pixel) { + case 8: + fb_format = (EVERGREEN_GRPH_DEPTH(EVERGREEN_GRPH_DEPTH_8BPP) | + EVERGREEN_GRPH_FORMAT(EVERGREEN_GRPH_FORMAT_INDEXED)); + break; + case 15: + fb_format = (EVERGREEN_GRPH_DEPTH(EVERGREEN_GRPH_DEPTH_16BPP) | + EVERGREEN_GRPH_FORMAT(EVERGREEN_GRPH_FORMAT_ARGB1555)); + break; + case 16: + fb_format = (EVERGREEN_GRPH_DEPTH(EVERGREEN_GRPH_DEPTH_16BPP) | + EVERGREEN_GRPH_FORMAT(EVERGREEN_GRPH_FORMAT_ARGB565)); +#ifdef __BIG_ENDIAN + fb_swap = EVERGREEN_GRPH_ENDIAN_SWAP(EVERGREEN_GRPH_ENDIAN_8IN16); +#endif + break; + case 24: + case 32: + fb_format = (EVERGREEN_GRPH_DEPTH(EVERGREEN_GRPH_DEPTH_32BPP) | + EVERGREEN_GRPH_FORMAT(EVERGREEN_GRPH_FORMAT_ARGB8888)); +#ifdef __BIG_ENDIAN + fb_swap = EVERGREEN_GRPH_ENDIAN_SWAP(EVERGREEN_GRPH_ENDIAN_8IN32); +#endif + break; + default: + DRM_ERROR("Unsupported screen depth %d\n", + target_fb->bits_per_pixel); + return -EINVAL; + } + + if (tiling_flags & RADEON_TILING_MACRO) + fb_format |= EVERGREEN_GRPH_ARRAY_MODE(EVERGREEN_GRPH_ARRAY_2D_TILED_THIN1); + else if (tiling_flags & RADEON_TILING_MICRO) + fb_format |= EVERGREEN_GRPH_ARRAY_MODE(EVERGREEN_GRPH_ARRAY_1D_TILED_THIN1); + + switch (radeon_crtc->crtc_id) { + case 0: + WREG32(AVIVO_D1VGA_CONTROL, 0); + break; + case 1: + WREG32(AVIVO_D2VGA_CONTROL, 0); + break; + case 2: + WREG32(EVERGREEN_D3VGA_CONTROL, 0); + break; + case 3: + WREG32(EVERGREEN_D4VGA_CONTROL, 0); + break; + case 4: + WREG32(EVERGREEN_D5VGA_CONTROL, 0); + break; + case 5: + WREG32(EVERGREEN_D6VGA_CONTROL, 0); + break; + default: + break; + } + + WREG32(EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS_HIGH + radeon_crtc->crtc_offset, + upper_32_bits(fb_location)); + WREG32(EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS_HIGH + radeon_crtc->crtc_offset, + upper_32_bits(fb_location)); + WREG32(EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS + radeon_crtc->crtc_offset, + (u32)fb_location & EVERGREEN_GRPH_SURFACE_ADDRESS_MASK); + WREG32(EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS + radeon_crtc->crtc_offset, + (u32) fb_location & EVERGREEN_GRPH_SURFACE_ADDRESS_MASK); + WREG32(EVERGREEN_GRPH_CONTROL + radeon_crtc->crtc_offset, fb_format); + WREG32(EVERGREEN_GRPH_SWAP_CONTROL + radeon_crtc->crtc_offset, fb_swap); + + WREG32(EVERGREEN_GRPH_SURFACE_OFFSET_X + radeon_crtc->crtc_offset, 0); + WREG32(EVERGREEN_GRPH_SURFACE_OFFSET_Y + radeon_crtc->crtc_offset, 0); + WREG32(EVERGREEN_GRPH_X_START + radeon_crtc->crtc_offset, 0); + WREG32(EVERGREEN_GRPH_Y_START + radeon_crtc->crtc_offset, 0); + WREG32(EVERGREEN_GRPH_X_END + radeon_crtc->crtc_offset, target_fb->width); + WREG32(EVERGREEN_GRPH_Y_END + radeon_crtc->crtc_offset, target_fb->height); + + fb_pitch_pixels = target_fb->pitch / (target_fb->bits_per_pixel / 8); + WREG32(EVERGREEN_GRPH_PITCH + radeon_crtc->crtc_offset, fb_pitch_pixels); + WREG32(EVERGREEN_GRPH_ENABLE + radeon_crtc->crtc_offset, 1); + + WREG32(EVERGREEN_DESKTOP_HEIGHT + radeon_crtc->crtc_offset, + target_fb->height); + x &= ~3; + y &= ~1; + WREG32(EVERGREEN_VIEWPORT_START + radeon_crtc->crtc_offset, + (x << 16) | y); + viewport_w = crtc->mode.hdisplay; + viewport_h = (crtc->mode.vdisplay + 1) & ~1; + WREG32(EVERGREEN_VIEWPORT_SIZE + radeon_crtc->crtc_offset, + (viewport_w << 16) | viewport_h); + + /* pageflip setup */ + /* make sure flip is at vb rather than hb */ + tmp = RREG32(EVERGREEN_GRPH_FLIP_CONTROL + radeon_crtc->crtc_offset); + tmp &= ~EVERGREEN_GRPH_SURFACE_UPDATE_H_RETRACE_EN; + WREG32(EVERGREEN_GRPH_FLIP_CONTROL + radeon_crtc->crtc_offset, tmp); + + /* set pageflip to happen anywhere in vblank interval */ + WREG32(EVERGREEN_MASTER_UPDATE_MODE + radeon_crtc->crtc_offset, 0); + + if (!atomic && fb && fb != crtc->fb) { + radeon_fb = to_radeon_framebuffer(fb); + rbo = gem_to_radeon_bo(radeon_fb->obj); + r = radeon_bo_reserve(rbo, false); + if (unlikely(r != 0)) + return r; + radeon_bo_unpin(rbo); + radeon_bo_unreserve(rbo); + } + + /* Bytes per pixel may have changed */ + radeon_bandwidth_update(rdev); + + return 0; +} + +static int avivo_crtc_do_set_base(struct drm_crtc *crtc, + struct drm_framebuffer *fb, + int x, int y, int atomic) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_framebuffer *radeon_fb; + struct drm_gem_object *obj; + struct radeon_bo *rbo; + struct drm_framebuffer *target_fb; + uint64_t fb_location; + uint32_t fb_format, fb_pitch_pixels, tiling_flags; + u32 fb_swap = R600_D1GRPH_SWAP_ENDIAN_NONE; + u32 tmp, viewport_w, viewport_h; + int r; + + /* no fb bound */ + if (!atomic && !crtc->fb) { + DRM_DEBUG_KMS("No FB bound\n"); + return 0; + } + + if (atomic) { + radeon_fb = to_radeon_framebuffer(fb); + target_fb = fb; + } + else { + radeon_fb = to_radeon_framebuffer(crtc->fb); + target_fb = crtc->fb; + } + + obj = radeon_fb->obj; + rbo = gem_to_radeon_bo(obj); + r = radeon_bo_reserve(rbo, false); + if (unlikely(r != 0)) + return r; + + /* If atomic, assume fb object is pinned & idle & fenced and + * just update base pointers + */ + if (atomic) + fb_location = radeon_bo_gpu_offset(rbo); + else { + r = radeon_bo_pin(rbo, RADEON_GEM_DOMAIN_VRAM, &fb_location); + if (unlikely(r != 0)) { + radeon_bo_unreserve(rbo); + return -EINVAL; + } + } + radeon_bo_get_tiling_flags(rbo, &tiling_flags, NULL); + radeon_bo_unreserve(rbo); + + switch (target_fb->bits_per_pixel) { + case 8: + fb_format = + AVIVO_D1GRPH_CONTROL_DEPTH_8BPP | + AVIVO_D1GRPH_CONTROL_8BPP_INDEXED; + break; + case 15: + fb_format = + AVIVO_D1GRPH_CONTROL_DEPTH_16BPP | + AVIVO_D1GRPH_CONTROL_16BPP_ARGB1555; + break; + case 16: + fb_format = + AVIVO_D1GRPH_CONTROL_DEPTH_16BPP | + AVIVO_D1GRPH_CONTROL_16BPP_RGB565; +#ifdef __BIG_ENDIAN + fb_swap = R600_D1GRPH_SWAP_ENDIAN_16BIT; +#endif + break; + case 24: + case 32: + fb_format = + AVIVO_D1GRPH_CONTROL_DEPTH_32BPP | + AVIVO_D1GRPH_CONTROL_32BPP_ARGB8888; +#ifdef __BIG_ENDIAN + fb_swap = R600_D1GRPH_SWAP_ENDIAN_32BIT; +#endif + break; + default: + DRM_ERROR("Unsupported screen depth %d\n", + target_fb->bits_per_pixel); + return -EINVAL; + } + + if (rdev->family >= CHIP_R600) { + if (tiling_flags & RADEON_TILING_MACRO) + fb_format |= R600_D1GRPH_ARRAY_MODE_2D_TILED_THIN1; + else if (tiling_flags & RADEON_TILING_MICRO) + fb_format |= R600_D1GRPH_ARRAY_MODE_1D_TILED_THIN1; + } else { + if (tiling_flags & RADEON_TILING_MACRO) + fb_format |= AVIVO_D1GRPH_MACRO_ADDRESS_MODE; + + if (tiling_flags & RADEON_TILING_MICRO) + fb_format |= AVIVO_D1GRPH_TILED; + } + + if (radeon_crtc->crtc_id == 0) + WREG32(AVIVO_D1VGA_CONTROL, 0); + else + WREG32(AVIVO_D2VGA_CONTROL, 0); + + if (rdev->family >= CHIP_RV770) { + if (radeon_crtc->crtc_id) { + WREG32(R700_D2GRPH_PRIMARY_SURFACE_ADDRESS_HIGH, upper_32_bits(fb_location)); + WREG32(R700_D2GRPH_SECONDARY_SURFACE_ADDRESS_HIGH, upper_32_bits(fb_location)); + } else { + WREG32(R700_D1GRPH_PRIMARY_SURFACE_ADDRESS_HIGH, upper_32_bits(fb_location)); + WREG32(R700_D1GRPH_SECONDARY_SURFACE_ADDRESS_HIGH, upper_32_bits(fb_location)); + } + } + WREG32(AVIVO_D1GRPH_PRIMARY_SURFACE_ADDRESS + radeon_crtc->crtc_offset, + (u32) fb_location); + WREG32(AVIVO_D1GRPH_SECONDARY_SURFACE_ADDRESS + + radeon_crtc->crtc_offset, (u32) fb_location); + WREG32(AVIVO_D1GRPH_CONTROL + radeon_crtc->crtc_offset, fb_format); + if (rdev->family >= CHIP_R600) + WREG32(R600_D1GRPH_SWAP_CONTROL + radeon_crtc->crtc_offset, fb_swap); + + WREG32(AVIVO_D1GRPH_SURFACE_OFFSET_X + radeon_crtc->crtc_offset, 0); + WREG32(AVIVO_D1GRPH_SURFACE_OFFSET_Y + radeon_crtc->crtc_offset, 0); + WREG32(AVIVO_D1GRPH_X_START + radeon_crtc->crtc_offset, 0); + WREG32(AVIVO_D1GRPH_Y_START + radeon_crtc->crtc_offset, 0); + WREG32(AVIVO_D1GRPH_X_END + radeon_crtc->crtc_offset, target_fb->width); + WREG32(AVIVO_D1GRPH_Y_END + radeon_crtc->crtc_offset, target_fb->height); + + fb_pitch_pixels = target_fb->pitch / (target_fb->bits_per_pixel / 8); + WREG32(AVIVO_D1GRPH_PITCH + radeon_crtc->crtc_offset, fb_pitch_pixels); + WREG32(AVIVO_D1GRPH_ENABLE + radeon_crtc->crtc_offset, 1); + + WREG32(AVIVO_D1MODE_DESKTOP_HEIGHT + radeon_crtc->crtc_offset, + target_fb->height); + x &= ~3; + y &= ~1; + WREG32(AVIVO_D1MODE_VIEWPORT_START + radeon_crtc->crtc_offset, + (x << 16) | y); + viewport_w = crtc->mode.hdisplay; + viewport_h = (crtc->mode.vdisplay + 1) & ~1; + WREG32(AVIVO_D1MODE_VIEWPORT_SIZE + radeon_crtc->crtc_offset, + (viewport_w << 16) | viewport_h); + + /* pageflip setup */ + /* make sure flip is at vb rather than hb */ + tmp = RREG32(AVIVO_D1GRPH_FLIP_CONTROL + radeon_crtc->crtc_offset); + tmp &= ~AVIVO_D1GRPH_SURFACE_UPDATE_H_RETRACE_EN; + WREG32(AVIVO_D1GRPH_FLIP_CONTROL + radeon_crtc->crtc_offset, tmp); + + /* set pageflip to happen anywhere in vblank interval */ + WREG32(AVIVO_D1MODE_MASTER_UPDATE_MODE + radeon_crtc->crtc_offset, 0); + + if (!atomic && fb && fb != crtc->fb) { + radeon_fb = to_radeon_framebuffer(fb); + rbo = gem_to_radeon_bo(radeon_fb->obj); + r = radeon_bo_reserve(rbo, false); + if (unlikely(r != 0)) + return r; + radeon_bo_unpin(rbo); + radeon_bo_unreserve(rbo); + } + + /* Bytes per pixel may have changed */ + radeon_bandwidth_update(rdev); + + return 0; +} + +int atombios_crtc_set_base(struct drm_crtc *crtc, int x, int y, + struct drm_framebuffer *old_fb) +{ + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + + if (ASIC_IS_DCE4(rdev)) + return dce4_crtc_do_set_base(crtc, old_fb, x, y, 0); + else if (ASIC_IS_AVIVO(rdev)) + return avivo_crtc_do_set_base(crtc, old_fb, x, y, 0); + else + return radeon_crtc_do_set_base(crtc, old_fb, x, y, 0); +} + +int atombios_crtc_set_base_atomic(struct drm_crtc *crtc, + struct drm_framebuffer *fb, + int x, int y, enum mode_set_atomic state) +{ + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + + if (ASIC_IS_DCE4(rdev)) + return dce4_crtc_do_set_base(crtc, fb, x, y, 1); + else if (ASIC_IS_AVIVO(rdev)) + return avivo_crtc_do_set_base(crtc, fb, x, y, 1); + else + return radeon_crtc_do_set_base(crtc, fb, x, y, 1); +} + +/* properly set additional regs when using atombios */ +static void radeon_legacy_atom_fixup(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + u32 disp_merge_cntl; + + switch (radeon_crtc->crtc_id) { + case 0: + disp_merge_cntl = RREG32(RADEON_DISP_MERGE_CNTL); + disp_merge_cntl &= ~RADEON_DISP_RGB_OFFSET_EN; + WREG32(RADEON_DISP_MERGE_CNTL, disp_merge_cntl); + break; + case 1: + disp_merge_cntl = RREG32(RADEON_DISP2_MERGE_CNTL); + disp_merge_cntl &= ~RADEON_DISP2_RGB_OFFSET_EN; + WREG32(RADEON_DISP2_MERGE_CNTL, disp_merge_cntl); + WREG32(RADEON_FP_H2_SYNC_STRT_WID, RREG32(RADEON_CRTC2_H_SYNC_STRT_WID)); + WREG32(RADEON_FP_V2_SYNC_STRT_WID, RREG32(RADEON_CRTC2_V_SYNC_STRT_WID)); + break; + } +} + +static int radeon_atom_pick_pll(struct drm_crtc *crtc) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + struct drm_encoder *test_encoder; + struct drm_crtc *test_crtc; + uint32_t pll_in_use = 0; + + if (ASIC_IS_DCE4(rdev)) { + list_for_each_entry(test_encoder, &dev->mode_config.encoder_list, head) { + if (test_encoder->crtc && (test_encoder->crtc == crtc)) { + /* in DP mode, the DP ref clock can come from PPLL, DCPLL, or ext clock, + * depending on the asic: + * DCE4: PPLL or ext clock + * DCE5: DCPLL or ext clock + * + * Setting ATOM_PPLL_INVALID will cause SetPixelClock to skip + * PPLL/DCPLL programming and only program the DP DTO for the + * crtc virtual pixel clock. + */ + if (atombios_get_encoder_mode(test_encoder) == ATOM_ENCODER_MODE_DP) { + if (ASIC_IS_DCE5(rdev) || rdev->clock.dp_extclk) + return ATOM_PPLL_INVALID; + } + } + } + + /* otherwise, pick one of the plls */ + list_for_each_entry(test_crtc, &dev->mode_config.crtc_list, head) { + struct radeon_crtc *radeon_test_crtc; + + if (crtc == test_crtc) + continue; + + radeon_test_crtc = to_radeon_crtc(test_crtc); + if ((radeon_test_crtc->pll_id >= ATOM_PPLL1) && + (radeon_test_crtc->pll_id <= ATOM_PPLL2)) + pll_in_use |= (1 << radeon_test_crtc->pll_id); + } + if (!(pll_in_use & 1)) + return ATOM_PPLL1; + return ATOM_PPLL2; + } else + return radeon_crtc->crtc_id; + +} + +int atombios_crtc_mode_set(struct drm_crtc *crtc, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode, + int x, int y, struct drm_framebuffer *old_fb) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + struct drm_encoder *encoder; + bool is_tvcv = false; + + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + /* find tv std */ + if (encoder->crtc == crtc) { + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + if (radeon_encoder->active_device & + (ATOM_DEVICE_TV_SUPPORT | ATOM_DEVICE_CV_SUPPORT)) + is_tvcv = true; + } + } + + /* always set DCPLL */ + if (ASIC_IS_DCE4(rdev)) { + struct radeon_atom_ss ss; + bool ss_enabled = radeon_atombios_get_asic_ss_info(rdev, &ss, + ASIC_INTERNAL_SS_ON_DCPLL, + rdev->clock.default_dispclk); + if (ss_enabled) + atombios_crtc_program_ss(crtc, ATOM_DISABLE, ATOM_DCPLL, &ss); + /* XXX: DCE5, make sure voltage, dispclk is high enough */ + atombios_crtc_set_dcpll(crtc, rdev->clock.default_dispclk); + if (ss_enabled) + atombios_crtc_program_ss(crtc, ATOM_ENABLE, ATOM_DCPLL, &ss); + } + atombios_crtc_set_pll(crtc, adjusted_mode); + + if (ASIC_IS_DCE4(rdev)) + atombios_set_crtc_dtd_timing(crtc, adjusted_mode); + else if (ASIC_IS_AVIVO(rdev)) { + if (is_tvcv) + atombios_crtc_set_timing(crtc, adjusted_mode); + else + atombios_set_crtc_dtd_timing(crtc, adjusted_mode); + } else { + atombios_crtc_set_timing(crtc, adjusted_mode); + if (radeon_crtc->crtc_id == 0) + atombios_set_crtc_dtd_timing(crtc, adjusted_mode); + radeon_legacy_atom_fixup(crtc); + } + atombios_crtc_set_base(crtc, x, y, old_fb); + atombios_overscan_setup(crtc, mode, adjusted_mode); + atombios_scaler_setup(crtc); + return 0; +} + +static bool atombios_crtc_mode_fixup(struct drm_crtc *crtc, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + + /* adjust pm to upcoming mode change */ + radeon_pm_compute_clocks(rdev); + + if (!radeon_crtc_scaling_mode_fixup(crtc, mode, adjusted_mode)) + return false; + return true; +} + +static void atombios_crtc_prepare(struct drm_crtc *crtc) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + + /* pick pll */ + radeon_crtc->pll_id = radeon_atom_pick_pll(crtc); + + atombios_lock_crtc(crtc, ATOM_ENABLE); + atombios_crtc_dpms(crtc, DRM_MODE_DPMS_OFF); +} + +static void atombios_crtc_commit(struct drm_crtc *crtc) +{ + atombios_crtc_dpms(crtc, DRM_MODE_DPMS_ON); + atombios_lock_crtc(crtc, ATOM_DISABLE); +} + +static void atombios_crtc_disable(struct drm_crtc *crtc) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct radeon_atom_ss ss; + + atombios_crtc_dpms(crtc, DRM_MODE_DPMS_OFF); + + switch (radeon_crtc->pll_id) { + case ATOM_PPLL1: + case ATOM_PPLL2: + /* disable the ppll */ + atombios_crtc_program_pll(crtc, radeon_crtc->crtc_id, radeon_crtc->pll_id, + 0, 0, ATOM_DISABLE, 0, 0, 0, 0, 0, false, &ss); + break; + default: + break; + } + radeon_crtc->pll_id = -1; +} + +static const struct drm_crtc_helper_funcs atombios_helper_funcs = { + .dpms = atombios_crtc_dpms, + .mode_fixup = atombios_crtc_mode_fixup, + .mode_set = atombios_crtc_mode_set, + .mode_set_base = atombios_crtc_set_base, + .mode_set_base_atomic = atombios_crtc_set_base_atomic, + .prepare = atombios_crtc_prepare, + .commit = atombios_crtc_commit, + .load_lut = radeon_crtc_load_lut, + .disable = atombios_crtc_disable, +}; + +void radeon_atombios_init_crtc(struct drm_device *dev, + struct radeon_crtc *radeon_crtc) +{ + struct radeon_device *rdev = dev->dev_private; + + if (ASIC_IS_DCE4(rdev)) { + switch (radeon_crtc->crtc_id) { + case 0: + default: + radeon_crtc->crtc_offset = EVERGREEN_CRTC0_REGISTER_OFFSET; + break; + case 1: + radeon_crtc->crtc_offset = EVERGREEN_CRTC1_REGISTER_OFFSET; + break; + case 2: + radeon_crtc->crtc_offset = EVERGREEN_CRTC2_REGISTER_OFFSET; + break; + case 3: + radeon_crtc->crtc_offset = EVERGREEN_CRTC3_REGISTER_OFFSET; + break; + case 4: + radeon_crtc->crtc_offset = EVERGREEN_CRTC4_REGISTER_OFFSET; + break; + case 5: + radeon_crtc->crtc_offset = EVERGREEN_CRTC5_REGISTER_OFFSET; + break; + } + } else { + if (radeon_crtc->crtc_id == 1) + radeon_crtc->crtc_offset = + AVIVO_D2CRTC_H_TOTAL - AVIVO_D1CRTC_H_TOTAL; + else + radeon_crtc->crtc_offset = 0; + } + radeon_crtc->pll_id = -1; + drm_crtc_helper_add(&radeon_crtc->base, &atombios_helper_funcs); +} diff --git a/drivers/gpu/drm/radeon/atombios_dp.c b/drivers/gpu/drm/radeon/atombios_dp.c new file mode 100644 index 00000000..3b77ad60 --- /dev/null +++ b/drivers/gpu/drm/radeon/atombios_dp.c @@ -0,0 +1,968 @@ +/* + * Copyright 2007-8 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + */ +#include "drmP.h" +#include "radeon_drm.h" +#include "radeon.h" + +#include "atom.h" +#include "atom-bits.h" +#include "drm_dp_helper.h" + +/* move these to drm_dp_helper.c/h */ +#define DP_LINK_CONFIGURATION_SIZE 9 +#define DP_LINK_STATUS_SIZE 6 +#define DP_DPCD_SIZE 8 + +static char *voltage_names[] = { + "0.4V", "0.6V", "0.8V", "1.2V" +}; +static char *pre_emph_names[] = { + "0dB", "3.5dB", "6dB", "9.5dB" +}; + +/***** radeon AUX functions *****/ +union aux_channel_transaction { + PROCESS_AUX_CHANNEL_TRANSACTION_PS_ALLOCATION v1; + PROCESS_AUX_CHANNEL_TRANSACTION_PARAMETERS_V2 v2; +}; + +static int radeon_process_aux_ch(struct radeon_i2c_chan *chan, + u8 *send, int send_bytes, + u8 *recv, int recv_size, + u8 delay, u8 *ack) +{ + struct drm_device *dev = chan->dev; + struct radeon_device *rdev = dev->dev_private; + union aux_channel_transaction args; + int index = GetIndexIntoMasterTable(COMMAND, ProcessAuxChannelTransaction); + unsigned char *base; + int recv_bytes; + + memset(&args, 0, sizeof(args)); + + base = (unsigned char *)rdev->mode_info.atom_context->scratch; + + memcpy(base, send, send_bytes); + + args.v1.lpAuxRequest = 0; + args.v1.lpDataOut = 16; + args.v1.ucDataOutLen = 0; + args.v1.ucChannelID = chan->rec.i2c_id; + args.v1.ucDelay = delay / 10; + if (ASIC_IS_DCE4(rdev)) + args.v2.ucHPD_ID = chan->rec.hpd; + + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); + + *ack = args.v1.ucReplyStatus; + + /* timeout */ + if (args.v1.ucReplyStatus == 1) { + DRM_DEBUG_KMS("dp_aux_ch timeout\n"); + return -ETIMEDOUT; + } + + /* flags not zero */ + if (args.v1.ucReplyStatus == 2) { + DRM_DEBUG_KMS("dp_aux_ch flags not zero\n"); + return -EBUSY; + } + + /* error */ + if (args.v1.ucReplyStatus == 3) { + DRM_DEBUG_KMS("dp_aux_ch error\n"); + return -EIO; + } + + recv_bytes = args.v1.ucDataOutLen; + if (recv_bytes > recv_size) + recv_bytes = recv_size; + + if (recv && recv_size) + memcpy(recv, base + 16, recv_bytes); + + return recv_bytes; +} + +static int radeon_dp_aux_native_write(struct radeon_connector *radeon_connector, + u16 address, u8 *send, u8 send_bytes, u8 delay) +{ + struct radeon_connector_atom_dig *dig_connector = radeon_connector->con_priv; + int ret; + u8 msg[20]; + int msg_bytes = send_bytes + 4; + u8 ack; + unsigned retry; + + if (send_bytes > 16) + return -1; + + msg[0] = address; + msg[1] = address >> 8; + msg[2] = AUX_NATIVE_WRITE << 4; + msg[3] = (msg_bytes << 4) | (send_bytes - 1); + memcpy(&msg[4], send, send_bytes); + + for (retry = 0; retry < 4; retry++) { + ret = radeon_process_aux_ch(dig_connector->dp_i2c_bus, + msg, msg_bytes, NULL, 0, delay, &ack); + if (ret == -EBUSY) + continue; + else if (ret < 0) + return ret; + if ((ack & AUX_NATIVE_REPLY_MASK) == AUX_NATIVE_REPLY_ACK) + return send_bytes; + else if ((ack & AUX_NATIVE_REPLY_MASK) == AUX_NATIVE_REPLY_DEFER) + udelay(400); + else + return -EIO; + } + + return -EIO; +} + +static int radeon_dp_aux_native_read(struct radeon_connector *radeon_connector, + u16 address, u8 *recv, int recv_bytes, u8 delay) +{ + struct radeon_connector_atom_dig *dig_connector = radeon_connector->con_priv; + u8 msg[4]; + int msg_bytes = 4; + u8 ack; + int ret; + unsigned retry; + + msg[0] = address; + msg[1] = address >> 8; + msg[2] = AUX_NATIVE_READ << 4; + msg[3] = (msg_bytes << 4) | (recv_bytes - 1); + + for (retry = 0; retry < 4; retry++) { + ret = radeon_process_aux_ch(dig_connector->dp_i2c_bus, + msg, msg_bytes, recv, recv_bytes, delay, &ack); + if (ret == -EBUSY) + continue; + else if (ret < 0) + return ret; + if ((ack & AUX_NATIVE_REPLY_MASK) == AUX_NATIVE_REPLY_ACK) + return ret; + else if ((ack & AUX_NATIVE_REPLY_MASK) == AUX_NATIVE_REPLY_DEFER) + udelay(400); + else if (ret == 0) + return -EPROTO; + else + return -EIO; + } + + return -EIO; +} + +static void radeon_write_dpcd_reg(struct radeon_connector *radeon_connector, + u16 reg, u8 val) +{ + radeon_dp_aux_native_write(radeon_connector, reg, &val, 1, 0); +} + +static u8 radeon_read_dpcd_reg(struct radeon_connector *radeon_connector, + u16 reg) +{ + u8 val = 0; + + radeon_dp_aux_native_read(radeon_connector, reg, &val, 1, 0); + + return val; +} + +int radeon_dp_i2c_aux_ch(struct i2c_adapter *adapter, int mode, + u8 write_byte, u8 *read_byte) +{ + struct i2c_algo_dp_aux_data *algo_data = adapter->algo_data; + struct radeon_i2c_chan *auxch = (struct radeon_i2c_chan *)adapter; + u16 address = algo_data->address; + u8 msg[5]; + u8 reply[2]; + unsigned retry; + int msg_bytes; + int reply_bytes = 1; + int ret; + u8 ack; + + /* Set up the command byte */ + if (mode & MODE_I2C_READ) + msg[2] = AUX_I2C_READ << 4; + else + msg[2] = AUX_I2C_WRITE << 4; + + if (!(mode & MODE_I2C_STOP)) + msg[2] |= AUX_I2C_MOT << 4; + + msg[0] = address; + msg[1] = address >> 8; + + switch (mode) { + case MODE_I2C_WRITE: + msg_bytes = 5; + msg[3] = msg_bytes << 4; + msg[4] = write_byte; + break; + case MODE_I2C_READ: + msg_bytes = 4; + msg[3] = msg_bytes << 4; + break; + default: + msg_bytes = 4; + msg[3] = 3 << 4; + break; + } + + for (retry = 0; retry < 4; retry++) { + ret = radeon_process_aux_ch(auxch, + msg, msg_bytes, reply, reply_bytes, 0, &ack); + if (ret == -EBUSY) + continue; + else if (ret < 0) { + DRM_DEBUG_KMS("aux_ch failed %d\n", ret); + return ret; + } + + switch (ack & AUX_NATIVE_REPLY_MASK) { + case AUX_NATIVE_REPLY_ACK: + /* I2C-over-AUX Reply field is only valid + * when paired with AUX ACK. + */ + break; + case AUX_NATIVE_REPLY_NACK: + DRM_DEBUG_KMS("aux_ch native nack\n"); + return -EREMOTEIO; + case AUX_NATIVE_REPLY_DEFER: + DRM_DEBUG_KMS("aux_ch native defer\n"); + udelay(400); + continue; + default: + DRM_ERROR("aux_ch invalid native reply 0x%02x\n", ack); + return -EREMOTEIO; + } + + switch (ack & AUX_I2C_REPLY_MASK) { + case AUX_I2C_REPLY_ACK: + if (mode == MODE_I2C_READ) + *read_byte = reply[0]; + return ret; + case AUX_I2C_REPLY_NACK: + DRM_DEBUG_KMS("aux_i2c nack\n"); + return -EREMOTEIO; + case AUX_I2C_REPLY_DEFER: + DRM_DEBUG_KMS("aux_i2c defer\n"); + udelay(400); + break; + default: + DRM_ERROR("aux_i2c invalid reply 0x%02x\n", ack); + return -EREMOTEIO; + } + } + + DRM_DEBUG_KMS("aux i2c too many retries, giving up\n"); + return -EREMOTEIO; +} + +/***** general DP utility functions *****/ + +static u8 dp_link_status(u8 link_status[DP_LINK_STATUS_SIZE], int r) +{ + return link_status[r - DP_LANE0_1_STATUS]; +} + +static u8 dp_get_lane_status(u8 link_status[DP_LINK_STATUS_SIZE], + int lane) +{ + int i = DP_LANE0_1_STATUS + (lane >> 1); + int s = (lane & 1) * 4; + u8 l = dp_link_status(link_status, i); + return (l >> s) & 0xf; +} + +static bool dp_clock_recovery_ok(u8 link_status[DP_LINK_STATUS_SIZE], + int lane_count) +{ + int lane; + u8 lane_status; + + for (lane = 0; lane < lane_count; lane++) { + lane_status = dp_get_lane_status(link_status, lane); + if ((lane_status & DP_LANE_CR_DONE) == 0) + return false; + } + return true; +} + +static bool dp_channel_eq_ok(u8 link_status[DP_LINK_STATUS_SIZE], + int lane_count) +{ + u8 lane_align; + u8 lane_status; + int lane; + + lane_align = dp_link_status(link_status, + DP_LANE_ALIGN_STATUS_UPDATED); + if ((lane_align & DP_INTERLANE_ALIGN_DONE) == 0) + return false; + for (lane = 0; lane < lane_count; lane++) { + lane_status = dp_get_lane_status(link_status, lane); + if ((lane_status & DP_CHANNEL_EQ_BITS) != DP_CHANNEL_EQ_BITS) + return false; + } + return true; +} + +static u8 dp_get_adjust_request_voltage(u8 link_status[DP_LINK_STATUS_SIZE], + int lane) + +{ + int i = DP_ADJUST_REQUEST_LANE0_1 + (lane >> 1); + int s = ((lane & 1) ? + DP_ADJUST_VOLTAGE_SWING_LANE1_SHIFT : + DP_ADJUST_VOLTAGE_SWING_LANE0_SHIFT); + u8 l = dp_link_status(link_status, i); + + return ((l >> s) & 0x3) << DP_TRAIN_VOLTAGE_SWING_SHIFT; +} + +static u8 dp_get_adjust_request_pre_emphasis(u8 link_status[DP_LINK_STATUS_SIZE], + int lane) +{ + int i = DP_ADJUST_REQUEST_LANE0_1 + (lane >> 1); + int s = ((lane & 1) ? + DP_ADJUST_PRE_EMPHASIS_LANE1_SHIFT : + DP_ADJUST_PRE_EMPHASIS_LANE0_SHIFT); + u8 l = dp_link_status(link_status, i); + + return ((l >> s) & 0x3) << DP_TRAIN_PRE_EMPHASIS_SHIFT; +} + +#define DP_VOLTAGE_MAX DP_TRAIN_VOLTAGE_SWING_1200 +#define DP_PRE_EMPHASIS_MAX DP_TRAIN_PRE_EMPHASIS_9_5 + +static void dp_get_adjust_train(u8 link_status[DP_LINK_STATUS_SIZE], + int lane_count, + u8 train_set[4]) +{ + u8 v = 0; + u8 p = 0; + int lane; + + for (lane = 0; lane < lane_count; lane++) { + u8 this_v = dp_get_adjust_request_voltage(link_status, lane); + u8 this_p = dp_get_adjust_request_pre_emphasis(link_status, lane); + + DRM_DEBUG_KMS("requested signal parameters: lane %d voltage %s pre_emph %s\n", + lane, + voltage_names[this_v >> DP_TRAIN_VOLTAGE_SWING_SHIFT], + pre_emph_names[this_p >> DP_TRAIN_PRE_EMPHASIS_SHIFT]); + + if (this_v > v) + v = this_v; + if (this_p > p) + p = this_p; + } + + if (v >= DP_VOLTAGE_MAX) + v |= DP_TRAIN_MAX_SWING_REACHED; + + if (p >= DP_PRE_EMPHASIS_MAX) + p |= DP_TRAIN_MAX_PRE_EMPHASIS_REACHED; + + DRM_DEBUG_KMS("using signal parameters: voltage %s pre_emph %s\n", + voltage_names[(v & DP_TRAIN_VOLTAGE_SWING_MASK) >> DP_TRAIN_VOLTAGE_SWING_SHIFT], + pre_emph_names[(p & DP_TRAIN_PRE_EMPHASIS_MASK) >> DP_TRAIN_PRE_EMPHASIS_SHIFT]); + + for (lane = 0; lane < 4; lane++) + train_set[lane] = v | p; +} + +/* convert bits per color to bits per pixel */ +/* get bpc from the EDID */ +static int convert_bpc_to_bpp(int bpc) +{ + if (bpc == 0) + return 24; + else + return bpc * 3; +} + +/* get the max pix clock supported by the link rate and lane num */ +static int dp_get_max_dp_pix_clock(int link_rate, + int lane_num, + int bpp) +{ + return (link_rate * lane_num * 8) / bpp; +} + +static int dp_get_max_link_rate(u8 dpcd[DP_DPCD_SIZE]) +{ + switch (dpcd[DP_MAX_LINK_RATE]) { + case DP_LINK_BW_1_62: + default: + return 162000; + case DP_LINK_BW_2_7: + return 270000; + case DP_LINK_BW_5_4: + return 540000; + } +} + +static u8 dp_get_max_lane_number(u8 dpcd[DP_DPCD_SIZE]) +{ + return dpcd[DP_MAX_LANE_COUNT] & DP_MAX_LANE_COUNT_MASK; +} + +static u8 dp_get_dp_link_rate_coded(int link_rate) +{ + switch (link_rate) { + case 162000: + default: + return DP_LINK_BW_1_62; + case 270000: + return DP_LINK_BW_2_7; + case 540000: + return DP_LINK_BW_5_4; + } +} + +/***** radeon specific DP functions *****/ + +/* First get the min lane# when low rate is used according to pixel clock + * (prefer low rate), second check max lane# supported by DP panel, + * if the max lane# < low rate lane# then use max lane# instead. + */ +static int radeon_dp_get_dp_lane_number(struct drm_connector *connector, + u8 dpcd[DP_DPCD_SIZE], + int pix_clock) +{ + int bpp = convert_bpc_to_bpp(connector->display_info.bpc); + int max_link_rate = dp_get_max_link_rate(dpcd); + int max_lane_num = dp_get_max_lane_number(dpcd); + int lane_num; + int max_dp_pix_clock; + + for (lane_num = 1; lane_num < max_lane_num; lane_num <<= 1) { + max_dp_pix_clock = dp_get_max_dp_pix_clock(max_link_rate, lane_num, bpp); + if (pix_clock <= max_dp_pix_clock) + break; + } + + return lane_num; +} + +static int radeon_dp_get_dp_link_clock(struct drm_connector *connector, + u8 dpcd[DP_DPCD_SIZE], + int pix_clock) +{ + int bpp = convert_bpc_to_bpp(connector->display_info.bpc); + int lane_num, max_pix_clock; + + if (radeon_connector_encoder_is_dp_bridge(connector)) + return 270000; + + lane_num = radeon_dp_get_dp_lane_number(connector, dpcd, pix_clock); + max_pix_clock = dp_get_max_dp_pix_clock(162000, lane_num, bpp); + if (pix_clock <= max_pix_clock) + return 162000; + max_pix_clock = dp_get_max_dp_pix_clock(270000, lane_num, bpp); + if (pix_clock <= max_pix_clock) + return 270000; + if (radeon_connector_is_dp12_capable(connector)) { + max_pix_clock = dp_get_max_dp_pix_clock(540000, lane_num, bpp); + if (pix_clock <= max_pix_clock) + return 540000; + } + + return dp_get_max_link_rate(dpcd); +} + +static u8 radeon_dp_encoder_service(struct radeon_device *rdev, + int action, int dp_clock, + u8 ucconfig, u8 lane_num) +{ + DP_ENCODER_SERVICE_PARAMETERS args; + int index = GetIndexIntoMasterTable(COMMAND, DPEncoderService); + + memset(&args, 0, sizeof(args)); + args.ucLinkClock = dp_clock / 10; + args.ucConfig = ucconfig; + args.ucAction = action; + args.ucLaneNum = lane_num; + args.ucStatus = 0; + + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); + return args.ucStatus; +} + +u8 radeon_dp_getsinktype(struct radeon_connector *radeon_connector) +{ + struct radeon_connector_atom_dig *dig_connector = radeon_connector->con_priv; + struct drm_device *dev = radeon_connector->base.dev; + struct radeon_device *rdev = dev->dev_private; + + return radeon_dp_encoder_service(rdev, ATOM_DP_ACTION_GET_SINK_TYPE, 0, + dig_connector->dp_i2c_bus->rec.i2c_id, 0); +} + +bool radeon_dp_getdpcd(struct radeon_connector *radeon_connector) +{ + struct radeon_connector_atom_dig *dig_connector = radeon_connector->con_priv; + u8 msg[25]; + int ret, i; + + ret = radeon_dp_aux_native_read(radeon_connector, DP_DPCD_REV, msg, 8, 0); + if (ret > 0) { + memcpy(dig_connector->dpcd, msg, 8); + DRM_DEBUG_KMS("DPCD: "); + for (i = 0; i < 8; i++) + DRM_DEBUG_KMS("%02x ", msg[i]); + DRM_DEBUG_KMS("\n"); + return true; + } + dig_connector->dpcd[0] = 0; + return false; +} + +static void radeon_dp_set_panel_mode(struct drm_encoder *encoder, + struct drm_connector *connector) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + int panel_mode = DP_PANEL_MODE_EXTERNAL_DP_MODE; + + if (!ASIC_IS_DCE4(rdev)) + return; + + if (radeon_connector_encoder_is_dp_bridge(connector)) + panel_mode = DP_PANEL_MODE_INTERNAL_DP1_MODE; + else if (connector->connector_type == DRM_MODE_CONNECTOR_eDP) { + u8 tmp = radeon_read_dpcd_reg(radeon_connector, DP_EDP_CONFIGURATION_CAP); + if (tmp & 1) + panel_mode = DP_PANEL_MODE_INTERNAL_DP2_MODE; + } + + atombios_dig_encoder_setup(encoder, + ATOM_ENCODER_CMD_SETUP_PANEL_MODE, + panel_mode); + + if ((connector->connector_type == DRM_MODE_CONNECTOR_eDP) && + (panel_mode == DP_PANEL_MODE_INTERNAL_DP2_MODE)) { + radeon_write_dpcd_reg(radeon_connector, DP_EDP_CONFIGURATION_SET, 1); + } +} + +void radeon_dp_set_link_config(struct drm_connector *connector, + struct drm_display_mode *mode) +{ + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + struct radeon_connector_atom_dig *dig_connector; + + if (!radeon_connector->con_priv) + return; + dig_connector = radeon_connector->con_priv; + + if ((dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT) || + (dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_eDP)) { + dig_connector->dp_clock = + radeon_dp_get_dp_link_clock(connector, dig_connector->dpcd, mode->clock); + dig_connector->dp_lane_count = + radeon_dp_get_dp_lane_number(connector, dig_connector->dpcd, mode->clock); + } +} + +int radeon_dp_mode_valid_helper(struct drm_connector *connector, + struct drm_display_mode *mode) +{ + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + struct radeon_connector_atom_dig *dig_connector; + int dp_clock; + + if (!radeon_connector->con_priv) + return MODE_CLOCK_HIGH; + dig_connector = radeon_connector->con_priv; + + dp_clock = + radeon_dp_get_dp_link_clock(connector, dig_connector->dpcd, mode->clock); + + if ((dp_clock == 540000) && + (!radeon_connector_is_dp12_capable(connector))) + return MODE_CLOCK_HIGH; + + return MODE_OK; +} + +static bool radeon_dp_get_link_status(struct radeon_connector *radeon_connector, + u8 link_status[DP_LINK_STATUS_SIZE]) +{ + int ret; + ret = radeon_dp_aux_native_read(radeon_connector, DP_LANE0_1_STATUS, + link_status, DP_LINK_STATUS_SIZE, 100); + if (ret <= 0) { + DRM_ERROR("displayport link status failed\n"); + return false; + } + + DRM_DEBUG_KMS("link status %02x %02x %02x %02x %02x %02x\n", + link_status[0], link_status[1], link_status[2], + link_status[3], link_status[4], link_status[5]); + return true; +} + +bool radeon_dp_needs_link_train(struct radeon_connector *radeon_connector) +{ + u8 link_status[DP_LINK_STATUS_SIZE]; + struct radeon_connector_atom_dig *dig = radeon_connector->con_priv; + + if (!radeon_dp_get_link_status(radeon_connector, link_status)) + return false; + if (dp_channel_eq_ok(link_status, dig->dp_lane_count)) + return false; + return true; +} + +struct radeon_dp_link_train_info { + struct radeon_device *rdev; + struct drm_encoder *encoder; + struct drm_connector *connector; + struct radeon_connector *radeon_connector; + int enc_id; + int dp_clock; + int dp_lane_count; + int rd_interval; + bool tp3_supported; + u8 dpcd[8]; + u8 train_set[4]; + u8 link_status[DP_LINK_STATUS_SIZE]; + u8 tries; + bool use_dpencoder; +}; + +static void radeon_dp_update_vs_emph(struct radeon_dp_link_train_info *dp_info) +{ + /* set the initial vs/emph on the source */ + atombios_dig_transmitter_setup(dp_info->encoder, + ATOM_TRANSMITTER_ACTION_SETUP_VSEMPH, + 0, dp_info->train_set[0]); /* sets all lanes at once */ + + /* set the vs/emph on the sink */ + radeon_dp_aux_native_write(dp_info->radeon_connector, DP_TRAINING_LANE0_SET, + dp_info->train_set, dp_info->dp_lane_count, 0); +} + +static void radeon_dp_set_tp(struct radeon_dp_link_train_info *dp_info, int tp) +{ + int rtp = 0; + + /* set training pattern on the source */ + if (ASIC_IS_DCE4(dp_info->rdev) || !dp_info->use_dpencoder) { + switch (tp) { + case DP_TRAINING_PATTERN_1: + rtp = ATOM_ENCODER_CMD_DP_LINK_TRAINING_PATTERN1; + break; + case DP_TRAINING_PATTERN_2: + rtp = ATOM_ENCODER_CMD_DP_LINK_TRAINING_PATTERN2; + break; + case DP_TRAINING_PATTERN_3: + rtp = ATOM_ENCODER_CMD_DP_LINK_TRAINING_PATTERN3; + break; + } + atombios_dig_encoder_setup(dp_info->encoder, rtp, 0); + } else { + switch (tp) { + case DP_TRAINING_PATTERN_1: + rtp = 0; + break; + case DP_TRAINING_PATTERN_2: + rtp = 1; + break; + } + radeon_dp_encoder_service(dp_info->rdev, ATOM_DP_ACTION_TRAINING_PATTERN_SEL, + dp_info->dp_clock, dp_info->enc_id, rtp); + } + + /* enable training pattern on the sink */ + radeon_write_dpcd_reg(dp_info->radeon_connector, DP_TRAINING_PATTERN_SET, tp); +} + +static int radeon_dp_link_train_init(struct radeon_dp_link_train_info *dp_info) +{ + u8 tmp; + + /* power up the sink */ + if (dp_info->dpcd[0] >= 0x11) + radeon_write_dpcd_reg(dp_info->radeon_connector, + DP_SET_POWER, DP_SET_POWER_D0); + + /* possibly enable downspread on the sink */ + if (dp_info->dpcd[3] & 0x1) + radeon_write_dpcd_reg(dp_info->radeon_connector, + DP_DOWNSPREAD_CTRL, DP_SPREAD_AMP_0_5); + else + radeon_write_dpcd_reg(dp_info->radeon_connector, + DP_DOWNSPREAD_CTRL, 0); + + radeon_dp_set_panel_mode(dp_info->encoder, dp_info->connector); + + /* set the lane count on the sink */ + tmp = dp_info->dp_lane_count; + if (dp_info->dpcd[0] >= 0x11) + tmp |= DP_LANE_COUNT_ENHANCED_FRAME_EN; + radeon_write_dpcd_reg(dp_info->radeon_connector, DP_LANE_COUNT_SET, tmp); + + /* set the link rate on the sink */ + tmp = dp_get_dp_link_rate_coded(dp_info->dp_clock); + radeon_write_dpcd_reg(dp_info->radeon_connector, DP_LINK_BW_SET, tmp); + + /* start training on the source */ + if (ASIC_IS_DCE4(dp_info->rdev) || !dp_info->use_dpencoder) + atombios_dig_encoder_setup(dp_info->encoder, + ATOM_ENCODER_CMD_DP_LINK_TRAINING_START, 0); + else + radeon_dp_encoder_service(dp_info->rdev, ATOM_DP_ACTION_TRAINING_START, + dp_info->dp_clock, dp_info->enc_id, 0); + + /* disable the training pattern on the sink */ + radeon_write_dpcd_reg(dp_info->radeon_connector, + DP_TRAINING_PATTERN_SET, + DP_TRAINING_PATTERN_DISABLE); + + return 0; +} + +static int radeon_dp_link_train_finish(struct radeon_dp_link_train_info *dp_info) +{ + udelay(400); + + /* disable the training pattern on the sink */ + radeon_write_dpcd_reg(dp_info->radeon_connector, + DP_TRAINING_PATTERN_SET, + DP_TRAINING_PATTERN_DISABLE); + + /* disable the training pattern on the source */ + if (ASIC_IS_DCE4(dp_info->rdev) || !dp_info->use_dpencoder) + atombios_dig_encoder_setup(dp_info->encoder, + ATOM_ENCODER_CMD_DP_LINK_TRAINING_COMPLETE, 0); + else + radeon_dp_encoder_service(dp_info->rdev, ATOM_DP_ACTION_TRAINING_COMPLETE, + dp_info->dp_clock, dp_info->enc_id, 0); + + return 0; +} + +static int radeon_dp_link_train_cr(struct radeon_dp_link_train_info *dp_info) +{ + bool clock_recovery; + u8 voltage; + int i; + + radeon_dp_set_tp(dp_info, DP_TRAINING_PATTERN_1); + memset(dp_info->train_set, 0, 4); + radeon_dp_update_vs_emph(dp_info); + + udelay(400); + + /* clock recovery loop */ + clock_recovery = false; + dp_info->tries = 0; + voltage = 0xff; + while (1) { + if (dp_info->rd_interval == 0) + udelay(100); + else + mdelay(dp_info->rd_interval * 4); + + if (!radeon_dp_get_link_status(dp_info->radeon_connector, dp_info->link_status)) + break; + + if (dp_clock_recovery_ok(dp_info->link_status, dp_info->dp_lane_count)) { + clock_recovery = true; + break; + } + + for (i = 0; i < dp_info->dp_lane_count; i++) { + if ((dp_info->train_set[i] & DP_TRAIN_MAX_SWING_REACHED) == 0) + break; + } + if (i == dp_info->dp_lane_count) { + DRM_ERROR("clock recovery reached max voltage\n"); + break; + } + + if ((dp_info->train_set[0] & DP_TRAIN_VOLTAGE_SWING_MASK) == voltage) { + ++dp_info->tries; + if (dp_info->tries == 5) { + DRM_ERROR("clock recovery tried 5 times\n"); + break; + } + } else + dp_info->tries = 0; + + voltage = dp_info->train_set[0] & DP_TRAIN_VOLTAGE_SWING_MASK; + + /* Compute new train_set as requested by sink */ + dp_get_adjust_train(dp_info->link_status, dp_info->dp_lane_count, dp_info->train_set); + + radeon_dp_update_vs_emph(dp_info); + } + if (!clock_recovery) { + DRM_ERROR("clock recovery failed\n"); + return -1; + } else { + DRM_DEBUG_KMS("clock recovery at voltage %d pre-emphasis %d\n", + dp_info->train_set[0] & DP_TRAIN_VOLTAGE_SWING_MASK, + (dp_info->train_set[0] & DP_TRAIN_PRE_EMPHASIS_MASK) >> + DP_TRAIN_PRE_EMPHASIS_SHIFT); + return 0; + } +} + +static int radeon_dp_link_train_ce(struct radeon_dp_link_train_info *dp_info) +{ + bool channel_eq; + + if (dp_info->tp3_supported) + radeon_dp_set_tp(dp_info, DP_TRAINING_PATTERN_3); + else + radeon_dp_set_tp(dp_info, DP_TRAINING_PATTERN_2); + + /* channel equalization loop */ + dp_info->tries = 0; + channel_eq = false; + while (1) { + if (dp_info->rd_interval == 0) + udelay(400); + else + mdelay(dp_info->rd_interval * 4); + + if (!radeon_dp_get_link_status(dp_info->radeon_connector, dp_info->link_status)) + break; + + if (dp_channel_eq_ok(dp_info->link_status, dp_info->dp_lane_count)) { + channel_eq = true; + break; + } + + /* Try 5 times */ + if (dp_info->tries > 5) { + DRM_ERROR("channel eq failed: 5 tries\n"); + break; + } + + /* Compute new train_set as requested by sink */ + dp_get_adjust_train(dp_info->link_status, dp_info->dp_lane_count, dp_info->train_set); + + radeon_dp_update_vs_emph(dp_info); + dp_info->tries++; + } + + if (!channel_eq) { + DRM_ERROR("channel eq failed\n"); + return -1; + } else { + DRM_DEBUG_KMS("channel eq at voltage %d pre-emphasis %d\n", + dp_info->train_set[0] & DP_TRAIN_VOLTAGE_SWING_MASK, + (dp_info->train_set[0] & DP_TRAIN_PRE_EMPHASIS_MASK) + >> DP_TRAIN_PRE_EMPHASIS_SHIFT); + return 0; + } +} + +void radeon_dp_link_train(struct drm_encoder *encoder, + struct drm_connector *connector) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct radeon_encoder_atom_dig *dig; + struct radeon_connector *radeon_connector; + struct radeon_connector_atom_dig *dig_connector; + struct radeon_dp_link_train_info dp_info; + int index; + u8 tmp, frev, crev; + + if (!radeon_encoder->enc_priv) + return; + dig = radeon_encoder->enc_priv; + + radeon_connector = to_radeon_connector(connector); + if (!radeon_connector->con_priv) + return; + dig_connector = radeon_connector->con_priv; + + if ((dig_connector->dp_sink_type != CONNECTOR_OBJECT_ID_DISPLAYPORT) && + (dig_connector->dp_sink_type != CONNECTOR_OBJECT_ID_eDP)) + return; + + /* DPEncoderService newer than 1.1 can't program properly the + * training pattern. When facing such version use the + * DIGXEncoderControl (X== 1 | 2) + */ + dp_info.use_dpencoder = true; + index = GetIndexIntoMasterTable(COMMAND, DPEncoderService); + if (atom_parse_cmd_header(rdev->mode_info.atom_context, index, &frev, &crev)) { + if (crev > 1) { + dp_info.use_dpencoder = false; + } + } + + dp_info.enc_id = 0; + if (dig->dig_encoder) + dp_info.enc_id |= ATOM_DP_CONFIG_DIG2_ENCODER; + else + dp_info.enc_id |= ATOM_DP_CONFIG_DIG1_ENCODER; + if (dig->linkb) + dp_info.enc_id |= ATOM_DP_CONFIG_LINK_B; + else + dp_info.enc_id |= ATOM_DP_CONFIG_LINK_A; + + dp_info.rd_interval = radeon_read_dpcd_reg(radeon_connector, DP_TRAINING_AUX_RD_INTERVAL); + tmp = radeon_read_dpcd_reg(radeon_connector, DP_MAX_LANE_COUNT); + if (ASIC_IS_DCE5(rdev) && (tmp & DP_TPS3_SUPPORTED)) + dp_info.tp3_supported = true; + else + dp_info.tp3_supported = false; + + memcpy(dp_info.dpcd, dig_connector->dpcd, 8); + dp_info.rdev = rdev; + dp_info.encoder = encoder; + dp_info.connector = connector; + dp_info.radeon_connector = radeon_connector; + dp_info.dp_lane_count = dig_connector->dp_lane_count; + dp_info.dp_clock = dig_connector->dp_clock; + + if (radeon_dp_link_train_init(&dp_info)) + goto done; + if (radeon_dp_link_train_cr(&dp_info)) + goto done; + if (radeon_dp_link_train_ce(&dp_info)) + goto done; +done: + if (radeon_dp_link_train_finish(&dp_info)) + return; +} diff --git a/drivers/gpu/drm/radeon/avivod.h b/drivers/gpu/drm/radeon/avivod.h new file mode 100644 index 00000000..3c391e7e --- /dev/null +++ b/drivers/gpu/drm/radeon/avivod.h @@ -0,0 +1,62 @@ +/* + * Copyright 2009 Advanced Micro Devices, Inc. + * Copyright 2009 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#ifndef AVIVOD_H +#define AVIVOD_H + + +#define D1CRTC_CONTROL 0x6080 +#define CRTC_EN (1 << 0) +#define D1CRTC_STATUS 0x609c +#define D1CRTC_UPDATE_LOCK 0x60E8 +#define D1GRPH_PRIMARY_SURFACE_ADDRESS 0x6110 +#define D1GRPH_SECONDARY_SURFACE_ADDRESS 0x6118 + +#define D2CRTC_CONTROL 0x6880 +#define D2CRTC_STATUS 0x689c +#define D2CRTC_UPDATE_LOCK 0x68E8 +#define D2GRPH_PRIMARY_SURFACE_ADDRESS 0x6910 +#define D2GRPH_SECONDARY_SURFACE_ADDRESS 0x6918 + +#define D1VGA_CONTROL 0x0330 +#define DVGA_CONTROL_MODE_ENABLE (1 << 0) +#define DVGA_CONTROL_TIMING_SELECT (1 << 8) +#define DVGA_CONTROL_SYNC_POLARITY_SELECT (1 << 9) +#define DVGA_CONTROL_OVERSCAN_TIMING_SELECT (1 << 10) +#define DVGA_CONTROL_OVERSCAN_COLOR_EN (1 << 16) +#define DVGA_CONTROL_ROTATE (1 << 24) +#define D2VGA_CONTROL 0x0338 + +#define VGA_HDP_CONTROL 0x328 +#define VGA_MEM_PAGE_SELECT_EN (1 << 0) +#define VGA_MEMORY_DISABLE (1 << 4) +#define VGA_RBBM_LOCK_DISABLE (1 << 8) +#define VGA_SOFT_RESET (1 << 16) +#define VGA_MEMORY_BASE_ADDRESS 0x0310 +#define VGA_RENDER_CONTROL 0x0300 +#define VGA_VSTATUS_CNTL_MASK 0x00030000 + +#endif diff --git a/drivers/gpu/drm/radeon/cayman_blit_shaders.c b/drivers/gpu/drm/radeon/cayman_blit_shaders.c new file mode 100644 index 00000000..7b4eeb7b --- /dev/null +++ b/drivers/gpu/drm/radeon/cayman_blit_shaders.c @@ -0,0 +1,373 @@ +/* + * Copyright 2010 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Alex Deucher + */ + +#include +#include + +/* + * evergreen cards need to use the 3D engine to blit data which requires + * quite a bit of hw state setup. Rather than pull the whole 3D driver + * (which normally generates the 3D state) into the DRM, we opt to use + * statically generated state tables. The regsiter state and shaders + * were hand generated to support blitting functionality. See the 3D + * driver or documentation for descriptions of the registers and + * shader instructions. + */ + +const u32 cayman_default_state[] = +{ + 0xc0066900, + 0x00000000, + 0x00000060, /* DB_RENDER_CONTROL */ + 0x00000000, /* DB_COUNT_CONTROL */ + 0x00000000, /* DB_DEPTH_VIEW */ + 0x0000002a, /* DB_RENDER_OVERRIDE */ + 0x00000000, /* DB_RENDER_OVERRIDE2 */ + 0x00000000, /* DB_HTILE_DATA_BASE */ + + 0xc0026900, + 0x0000000a, + 0x00000000, /* DB_STENCIL_CLEAR */ + 0x00000000, /* DB_DEPTH_CLEAR */ + + 0xc0036900, + 0x0000000f, + 0x00000000, /* DB_DEPTH_INFO */ + 0x00000000, /* DB_Z_INFO */ + 0x00000000, /* DB_STENCIL_INFO */ + + 0xc0016900, + 0x00000080, + 0x00000000, /* PA_SC_WINDOW_OFFSET */ + + 0xc00d6900, + 0x00000083, + 0x0000ffff, /* PA_SC_CLIPRECT_RULE */ + 0x00000000, /* PA_SC_CLIPRECT_0_TL */ + 0x20002000, /* PA_SC_CLIPRECT_0_BR */ + 0x00000000, + 0x20002000, + 0x00000000, + 0x20002000, + 0x00000000, + 0x20002000, + 0xaaaaaaaa, /* PA_SC_EDGERULE */ + 0x00000000, /* PA_SU_HARDWARE_SCREEN_OFFSET */ + 0x0000000f, /* CB_TARGET_MASK */ + 0x0000000f, /* CB_SHADER_MASK */ + + 0xc0226900, + 0x00000094, + 0x80000000, /* PA_SC_VPORT_SCISSOR_0_TL */ + 0x20002000, /* PA_SC_VPORT_SCISSOR_0_BR */ + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x00000000, /* PA_SC_VPORT_ZMIN_0 */ + 0x3f800000, /* PA_SC_VPORT_ZMAX_0 */ + + 0xc0016900, + 0x000000d4, + 0x00000000, /* SX_MISC */ + + 0xc0026900, + 0x000000d9, + 0x00000000, /* CP_RINGID */ + 0x00000000, /* CP_VMID */ + + 0xc0096900, + 0x00000100, + 0x00ffffff, /* VGT_MAX_VTX_INDX */ + 0x00000000, /* VGT_MIN_VTX_INDX */ + 0x00000000, /* VGT_INDX_OFFSET */ + 0x00000000, /* VGT_MULTI_PRIM_IB_RESET_INDX */ + 0x00000000, /* SX_ALPHA_TEST_CONTROL */ + 0x00000000, /* CB_BLEND_RED */ + 0x00000000, /* CB_BLEND_GREEN */ + 0x00000000, /* CB_BLEND_BLUE */ + 0x00000000, /* CB_BLEND_ALPHA */ + + 0xc0016900, + 0x00000187, + 0x00000100, /* SPI_VS_OUT_ID_0 */ + + 0xc0026900, + 0x00000191, + 0x00000100, /* SPI_PS_INPUT_CNTL_0 */ + 0x00000101, /* SPI_PS_INPUT_CNTL_1 */ + + 0xc0016900, + 0x000001b1, + 0x00000000, /* SPI_VS_OUT_CONFIG */ + + 0xc0106900, + 0x000001b3, + 0x20000001, /* SPI_PS_IN_CONTROL_0 */ + 0x00000000, /* SPI_PS_IN_CONTROL_1 */ + 0x00000000, /* SPI_INTERP_CONTROL_0 */ + 0x00000000, /* SPI_INPUT_Z */ + 0x00000000, /* SPI_FOG_CNTL */ + 0x00100000, /* SPI_BARYC_CNTL */ + 0x00000000, /* SPI_PS_IN_CONTROL_2 */ + 0x00000000, /* SPI_COMPUTE_INPUT_CNTL */ + 0x00000000, /* SPI_COMPUTE_NUM_THREAD_X */ + 0x00000000, /* SPI_COMPUTE_NUM_THREAD_Y */ + 0x00000000, /* SPI_COMPUTE_NUM_THREAD_Z */ + 0x00000000, /* SPI_GPR_MGMT */ + 0x00000000, /* SPI_LDS_MGMT */ + 0x00000000, /* SPI_STACK_MGMT */ + 0x00000000, /* SPI_WAVE_MGMT_1 */ + 0x00000000, /* SPI_WAVE_MGMT_2 */ + + 0xc0016900, + 0x000001e0, + 0x00000000, /* CB_BLEND0_CONTROL */ + + 0xc00e6900, + 0x00000200, + 0x00000000, /* DB_DEPTH_CONTROL */ + 0x00000000, /* DB_EQAA */ + 0x00cc0010, /* CB_COLOR_CONTROL */ + 0x00000210, /* DB_SHADER_CONTROL */ + 0x00010000, /* PA_CL_CLIP_CNTL */ + 0x00000004, /* PA_SU_SC_MODE_CNTL */ + 0x00000100, /* PA_CL_VTE_CNTL */ + 0x00000000, /* PA_CL_VS_OUT_CNTL */ + 0x00000000, /* PA_CL_NANINF_CNTL */ + 0x00000000, /* PA_SU_LINE_STIPPLE_CNTL */ + 0x00000000, /* PA_SU_LINE_STIPPLE_SCALE */ + 0x00000000, /* PA_SU_PRIM_FILTER_CNTL */ + 0x00000000, /* */ + 0x00000000, /* */ + + 0xc0026900, + 0x00000229, + 0x00000000, /* SQ_PGM_START_FS */ + 0x00000000, + + 0xc0016900, + 0x0000023b, + 0x00000000, /* SQ_LDS_ALLOC_PS */ + + 0xc0066900, + 0x00000240, + 0x00000000, /* SQ_ESGS_RING_ITEMSIZE */ + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + + 0xc0046900, + 0x00000247, + 0x00000000, /* SQ_GS_VERT_ITEMSIZE */ + 0x00000000, + 0x00000000, + 0x00000000, + + 0xc0116900, + 0x00000280, + 0x00000000, /* PA_SU_POINT_SIZE */ + 0x00000000, /* PA_SU_POINT_MINMAX */ + 0x00000008, /* PA_SU_LINE_CNTL */ + 0x00000000, /* PA_SC_LINE_STIPPLE */ + 0x00000000, /* VGT_OUTPUT_PATH_CNTL */ + 0x00000000, /* VGT_HOS_CNTL */ + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, /* VGT_GS_MODE */ + + 0xc0026900, + 0x00000292, + 0x00000000, /* PA_SC_MODE_CNTL_0 */ + 0x00000000, /* PA_SC_MODE_CNTL_1 */ + + 0xc0016900, + 0x000002a1, + 0x00000000, /* VGT_PRIMITIVEID_EN */ + + 0xc0016900, + 0x000002a5, + 0x00000000, /* VGT_MULTI_PRIM_IB_RESET_EN */ + + 0xc0026900, + 0x000002a8, + 0x00000000, /* VGT_INSTANCE_STEP_RATE_0 */ + 0x00000000, + + 0xc0026900, + 0x000002ad, + 0x00000000, /* VGT_REUSE_OFF */ + 0x00000000, + + 0xc0016900, + 0x000002d5, + 0x00000000, /* VGT_SHADER_STAGES_EN */ + + 0xc0016900, + 0x000002dc, + 0x0000aa00, /* DB_ALPHA_TO_MASK */ + + 0xc0066900, + 0x000002de, + 0x00000000, /* PA_SU_POLY_OFFSET_DB_FMT_CNTL */ + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + + 0xc0026900, + 0x000002e5, + 0x00000000, /* VGT_STRMOUT_CONFIG */ + 0x00000000, + + 0xc01b6900, + 0x000002f5, + 0x76543210, /* PA_SC_CENTROID_PRIORITY_0 */ + 0xfedcba98, /* PA_SC_CENTROID_PRIORITY_1 */ + 0x00000000, /* PA_SC_LINE_CNTL */ + 0x00000000, /* PA_SC_AA_CONFIG */ + 0x00000005, /* PA_SU_VTX_CNTL */ + 0x3f800000, /* PA_CL_GB_VERT_CLIP_ADJ */ + 0x3f800000, /* PA_CL_GB_VERT_DISC_ADJ */ + 0x3f800000, /* PA_CL_GB_HORZ_CLIP_ADJ */ + 0x3f800000, /* PA_CL_GB_HORZ_DISC_ADJ */ + 0x00000000, /* PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_0 */ + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0xffffffff, /* PA_SC_AA_MASK_X0Y0_X1Y0 */ + 0xffffffff, + + 0xc0026900, + 0x00000316, + 0x0000000e, /* VGT_VERTEX_REUSE_BLOCK_CNTL */ + 0x00000010, /* */ +}; + +const u32 cayman_vs[] = +{ + 0x00000004, + 0x80400400, + 0x0000a03c, + 0x95000688, + 0x00004000, + 0x15000688, + 0x00000000, + 0x88000000, + 0x04000000, + 0x67961001, +#ifdef __BIG_ENDIAN + 0x00020000, +#else + 0x00000000, +#endif + 0x00000000, + 0x04000000, + 0x67961000, +#ifdef __BIG_ENDIAN + 0x00020008, +#else + 0x00000008, +#endif + 0x00000000, +}; + +const u32 cayman_ps[] = +{ + 0x00000004, + 0xa00c0000, + 0x00000008, + 0x80400000, + 0x00000000, + 0x95000688, + 0x00000000, + 0x88000000, + 0x00380400, + 0x00146b10, + 0x00380000, + 0x20146b10, + 0x00380400, + 0x40146b00, + 0x80380000, + 0x60146b00, + 0x00000010, + 0x000d1000, + 0xb0800000, + 0x00000000, +}; + +const u32 cayman_ps_size = ARRAY_SIZE(cayman_ps); +const u32 cayman_vs_size = ARRAY_SIZE(cayman_vs); +const u32 cayman_default_size = ARRAY_SIZE(cayman_default_state); diff --git a/drivers/gpu/drm/radeon/cayman_blit_shaders.h b/drivers/gpu/drm/radeon/cayman_blit_shaders.h new file mode 100644 index 00000000..f5d0e9a6 --- /dev/null +++ b/drivers/gpu/drm/radeon/cayman_blit_shaders.h @@ -0,0 +1,35 @@ +/* + * Copyright 2010 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef CAYMAN_BLIT_SHADERS_H +#define CAYMAN_BLIT_SHADERS_H + +extern const u32 cayman_ps[]; +extern const u32 cayman_vs[]; +extern const u32 cayman_default_state[]; + +extern const u32 cayman_ps_size, cayman_vs_size; +extern const u32 cayman_default_size; + +#endif diff --git a/drivers/gpu/drm/radeon/evergreen.c b/drivers/gpu/drm/radeon/evergreen.c new file mode 100644 index 00000000..8846bad4 --- /dev/null +++ b/drivers/gpu/drm/radeon/evergreen.c @@ -0,0 +1,3348 @@ +/* + * Copyright 2010 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Alex Deucher + */ +#include +#include +#include +#include "drmP.h" +#include "radeon.h" +#include "radeon_asic.h" +#include "radeon_drm.h" +#include "evergreend.h" +#include "atom.h" +#include "avivod.h" +#include "evergreen_reg.h" +#include "evergreen_blit_shaders.h" + +#define EVERGREEN_PFP_UCODE_SIZE 1120 +#define EVERGREEN_PM4_UCODE_SIZE 1376 + +static void evergreen_gpu_init(struct radeon_device *rdev); +void evergreen_fini(struct radeon_device *rdev); +static void evergreen_pcie_gen2_enable(struct radeon_device *rdev); + +void evergreen_fix_pci_max_read_req_size(struct radeon_device *rdev) +{ + u16 ctl, v; + int cap, err; + + cap = pci_pcie_cap(rdev->pdev); + if (!cap) + return; + + err = pci_read_config_word(rdev->pdev, cap + PCI_EXP_DEVCTL, &ctl); + if (err) + return; + + v = (ctl & PCI_EXP_DEVCTL_READRQ) >> 12; + + /* if bios or OS sets MAX_READ_REQUEST_SIZE to an invalid value, fix it + * to avoid hangs or perfomance issues + */ + if ((v == 0) || (v == 6) || (v == 7)) { + ctl &= ~PCI_EXP_DEVCTL_READRQ; + ctl |= (2 << 12); + pci_write_config_word(rdev->pdev, cap + PCI_EXP_DEVCTL, ctl); + } +} + +void evergreen_pre_page_flip(struct radeon_device *rdev, int crtc) +{ + /* enable the pflip int */ + radeon_irq_kms_pflip_irq_get(rdev, crtc); +} + +void evergreen_post_page_flip(struct radeon_device *rdev, int crtc) +{ + /* disable the pflip int */ + radeon_irq_kms_pflip_irq_put(rdev, crtc); +} + +u32 evergreen_page_flip(struct radeon_device *rdev, int crtc_id, u64 crtc_base) +{ + struct radeon_crtc *radeon_crtc = rdev->mode_info.crtcs[crtc_id]; + u32 tmp = RREG32(EVERGREEN_GRPH_UPDATE + radeon_crtc->crtc_offset); + int i; + + /* Lock the graphics update lock */ + tmp |= EVERGREEN_GRPH_UPDATE_LOCK; + WREG32(EVERGREEN_GRPH_UPDATE + radeon_crtc->crtc_offset, tmp); + + /* update the scanout addresses */ + WREG32(EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS_HIGH + radeon_crtc->crtc_offset, + upper_32_bits(crtc_base)); + WREG32(EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS + radeon_crtc->crtc_offset, + (u32)crtc_base); + + WREG32(EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS_HIGH + radeon_crtc->crtc_offset, + upper_32_bits(crtc_base)); + WREG32(EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS + radeon_crtc->crtc_offset, + (u32)crtc_base); + + /* Wait for update_pending to go high. */ + for (i = 0; i < rdev->usec_timeout; i++) { + if (RREG32(EVERGREEN_GRPH_UPDATE + radeon_crtc->crtc_offset) & EVERGREEN_GRPH_SURFACE_UPDATE_PENDING) + break; + udelay(1); + } + DRM_DEBUG("Update pending now high. Unlocking vupdate_lock.\n"); + + /* Unlock the lock, so double-buffering can take place inside vblank */ + tmp &= ~EVERGREEN_GRPH_UPDATE_LOCK; + WREG32(EVERGREEN_GRPH_UPDATE + radeon_crtc->crtc_offset, tmp); + + /* Return current update_pending status: */ + return RREG32(EVERGREEN_GRPH_UPDATE + radeon_crtc->crtc_offset) & EVERGREEN_GRPH_SURFACE_UPDATE_PENDING; +} + +/* get temperature in millidegrees */ +int evergreen_get_temp(struct radeon_device *rdev) +{ + u32 temp, toffset; + int actual_temp = 0; + + if (rdev->family == CHIP_JUNIPER) { + toffset = (RREG32(CG_THERMAL_CTRL) & TOFFSET_MASK) >> + TOFFSET_SHIFT; + temp = (RREG32(CG_TS0_STATUS) & TS0_ADC_DOUT_MASK) >> + TS0_ADC_DOUT_SHIFT; + + if (toffset & 0x100) + actual_temp = temp / 2 - (0x200 - toffset); + else + actual_temp = temp / 2 + toffset; + + actual_temp = actual_temp * 1000; + + } else { + temp = (RREG32(CG_MULT_THERMAL_STATUS) & ASIC_T_MASK) >> + ASIC_T_SHIFT; + + if (temp & 0x400) + actual_temp = -256; + else if (temp & 0x200) + actual_temp = 255; + else if (temp & 0x100) { + actual_temp = temp & 0x1ff; + actual_temp |= ~0x1ff; + } else + actual_temp = temp & 0xff; + + actual_temp = (actual_temp * 1000) / 2; + } + + return actual_temp; +} + +int sumo_get_temp(struct radeon_device *rdev) +{ + u32 temp = RREG32(CG_THERMAL_STATUS) & 0xff; + int actual_temp = temp - 49; + + return actual_temp * 1000; +} + +void evergreen_pm_misc(struct radeon_device *rdev) +{ + int req_ps_idx = rdev->pm.requested_power_state_index; + int req_cm_idx = rdev->pm.requested_clock_mode_index; + struct radeon_power_state *ps = &rdev->pm.power_state[req_ps_idx]; + struct radeon_voltage *voltage = &ps->clock_info[req_cm_idx].voltage; + + if (voltage->type == VOLTAGE_SW) { + /* 0xff01 is a flag rather then an actual voltage */ + if (voltage->voltage == 0xff01) + return; + if (voltage->voltage && (voltage->voltage != rdev->pm.current_vddc)) { + radeon_atom_set_voltage(rdev, voltage->voltage, SET_VOLTAGE_TYPE_ASIC_VDDC); + rdev->pm.current_vddc = voltage->voltage; + DRM_DEBUG("Setting: vddc: %d\n", voltage->voltage); + } + /* 0xff01 is a flag rather then an actual voltage */ + if (voltage->vddci == 0xff01) + return; + if (voltage->vddci && (voltage->vddci != rdev->pm.current_vddci)) { + radeon_atom_set_voltage(rdev, voltage->vddci, SET_VOLTAGE_TYPE_ASIC_VDDCI); + rdev->pm.current_vddci = voltage->vddci; + DRM_DEBUG("Setting: vddci: %d\n", voltage->vddci); + } + } +} + +void evergreen_pm_prepare(struct radeon_device *rdev) +{ + struct drm_device *ddev = rdev->ddev; + struct drm_crtc *crtc; + struct radeon_crtc *radeon_crtc; + u32 tmp; + + /* disable any active CRTCs */ + list_for_each_entry(crtc, &ddev->mode_config.crtc_list, head) { + radeon_crtc = to_radeon_crtc(crtc); + if (radeon_crtc->enabled) { + tmp = RREG32(EVERGREEN_CRTC_CONTROL + radeon_crtc->crtc_offset); + tmp |= EVERGREEN_CRTC_DISP_READ_REQUEST_DISABLE; + WREG32(EVERGREEN_CRTC_CONTROL + radeon_crtc->crtc_offset, tmp); + } + } +} + +void evergreen_pm_finish(struct radeon_device *rdev) +{ + struct drm_device *ddev = rdev->ddev; + struct drm_crtc *crtc; + struct radeon_crtc *radeon_crtc; + u32 tmp; + + /* enable any active CRTCs */ + list_for_each_entry(crtc, &ddev->mode_config.crtc_list, head) { + radeon_crtc = to_radeon_crtc(crtc); + if (radeon_crtc->enabled) { + tmp = RREG32(EVERGREEN_CRTC_CONTROL + radeon_crtc->crtc_offset); + tmp &= ~EVERGREEN_CRTC_DISP_READ_REQUEST_DISABLE; + WREG32(EVERGREEN_CRTC_CONTROL + radeon_crtc->crtc_offset, tmp); + } + } +} + +bool evergreen_hpd_sense(struct radeon_device *rdev, enum radeon_hpd_id hpd) +{ + bool connected = false; + + switch (hpd) { + case RADEON_HPD_1: + if (RREG32(DC_HPD1_INT_STATUS) & DC_HPDx_SENSE) + connected = true; + break; + case RADEON_HPD_2: + if (RREG32(DC_HPD2_INT_STATUS) & DC_HPDx_SENSE) + connected = true; + break; + case RADEON_HPD_3: + if (RREG32(DC_HPD3_INT_STATUS) & DC_HPDx_SENSE) + connected = true; + break; + case RADEON_HPD_4: + if (RREG32(DC_HPD4_INT_STATUS) & DC_HPDx_SENSE) + connected = true; + break; + case RADEON_HPD_5: + if (RREG32(DC_HPD5_INT_STATUS) & DC_HPDx_SENSE) + connected = true; + break; + case RADEON_HPD_6: + if (RREG32(DC_HPD6_INT_STATUS) & DC_HPDx_SENSE) + connected = true; + break; + default: + break; + } + + return connected; +} + +void evergreen_hpd_set_polarity(struct radeon_device *rdev, + enum radeon_hpd_id hpd) +{ + u32 tmp; + bool connected = evergreen_hpd_sense(rdev, hpd); + + switch (hpd) { + case RADEON_HPD_1: + tmp = RREG32(DC_HPD1_INT_CONTROL); + if (connected) + tmp &= ~DC_HPDx_INT_POLARITY; + else + tmp |= DC_HPDx_INT_POLARITY; + WREG32(DC_HPD1_INT_CONTROL, tmp); + break; + case RADEON_HPD_2: + tmp = RREG32(DC_HPD2_INT_CONTROL); + if (connected) + tmp &= ~DC_HPDx_INT_POLARITY; + else + tmp |= DC_HPDx_INT_POLARITY; + WREG32(DC_HPD2_INT_CONTROL, tmp); + break; + case RADEON_HPD_3: + tmp = RREG32(DC_HPD3_INT_CONTROL); + if (connected) + tmp &= ~DC_HPDx_INT_POLARITY; + else + tmp |= DC_HPDx_INT_POLARITY; + WREG32(DC_HPD3_INT_CONTROL, tmp); + break; + case RADEON_HPD_4: + tmp = RREG32(DC_HPD4_INT_CONTROL); + if (connected) + tmp &= ~DC_HPDx_INT_POLARITY; + else + tmp |= DC_HPDx_INT_POLARITY; + WREG32(DC_HPD4_INT_CONTROL, tmp); + break; + case RADEON_HPD_5: + tmp = RREG32(DC_HPD5_INT_CONTROL); + if (connected) + tmp &= ~DC_HPDx_INT_POLARITY; + else + tmp |= DC_HPDx_INT_POLARITY; + WREG32(DC_HPD5_INT_CONTROL, tmp); + break; + case RADEON_HPD_6: + tmp = RREG32(DC_HPD6_INT_CONTROL); + if (connected) + tmp &= ~DC_HPDx_INT_POLARITY; + else + tmp |= DC_HPDx_INT_POLARITY; + WREG32(DC_HPD6_INT_CONTROL, tmp); + break; + default: + break; + } +} + +void evergreen_hpd_init(struct radeon_device *rdev) +{ + struct drm_device *dev = rdev->ddev; + struct drm_connector *connector; + u32 tmp = DC_HPDx_CONNECTION_TIMER(0x9c4) | + DC_HPDx_RX_INT_TIMER(0xfa) | DC_HPDx_EN; + + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + switch (radeon_connector->hpd.hpd) { + case RADEON_HPD_1: + WREG32(DC_HPD1_CONTROL, tmp); + rdev->irq.hpd[0] = true; + break; + case RADEON_HPD_2: + WREG32(DC_HPD2_CONTROL, tmp); + rdev->irq.hpd[1] = true; + break; + case RADEON_HPD_3: + WREG32(DC_HPD3_CONTROL, tmp); + rdev->irq.hpd[2] = true; + break; + case RADEON_HPD_4: + WREG32(DC_HPD4_CONTROL, tmp); + rdev->irq.hpd[3] = true; + break; + case RADEON_HPD_5: + WREG32(DC_HPD5_CONTROL, tmp); + rdev->irq.hpd[4] = true; + break; + case RADEON_HPD_6: + WREG32(DC_HPD6_CONTROL, tmp); + rdev->irq.hpd[5] = true; + break; + default: + break; + } + radeon_hpd_set_polarity(rdev, radeon_connector->hpd.hpd); + } + if (rdev->irq.installed) + evergreen_irq_set(rdev); +} + +void evergreen_hpd_fini(struct radeon_device *rdev) +{ + struct drm_device *dev = rdev->ddev; + struct drm_connector *connector; + + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + switch (radeon_connector->hpd.hpd) { + case RADEON_HPD_1: + WREG32(DC_HPD1_CONTROL, 0); + rdev->irq.hpd[0] = false; + break; + case RADEON_HPD_2: + WREG32(DC_HPD2_CONTROL, 0); + rdev->irq.hpd[1] = false; + break; + case RADEON_HPD_3: + WREG32(DC_HPD3_CONTROL, 0); + rdev->irq.hpd[2] = false; + break; + case RADEON_HPD_4: + WREG32(DC_HPD4_CONTROL, 0); + rdev->irq.hpd[3] = false; + break; + case RADEON_HPD_5: + WREG32(DC_HPD5_CONTROL, 0); + rdev->irq.hpd[4] = false; + break; + case RADEON_HPD_6: + WREG32(DC_HPD6_CONTROL, 0); + rdev->irq.hpd[5] = false; + break; + default: + break; + } + } +} + +/* watermark setup */ + +static u32 evergreen_line_buffer_adjust(struct radeon_device *rdev, + struct radeon_crtc *radeon_crtc, + struct drm_display_mode *mode, + struct drm_display_mode *other_mode) +{ + u32 tmp; + /* + * Line Buffer Setup + * There are 3 line buffers, each one shared by 2 display controllers. + * DC_LB_MEMORY_SPLIT controls how that line buffer is shared between + * the display controllers. The paritioning is done via one of four + * preset allocations specified in bits 2:0: + * first display controller + * 0 - first half of lb (3840 * 2) + * 1 - first 3/4 of lb (5760 * 2) + * 2 - whole lb (7680 * 2), other crtc must be disabled + * 3 - first 1/4 of lb (1920 * 2) + * second display controller + * 4 - second half of lb (3840 * 2) + * 5 - second 3/4 of lb (5760 * 2) + * 6 - whole lb (7680 * 2), other crtc must be disabled + * 7 - last 1/4 of lb (1920 * 2) + */ + /* this can get tricky if we have two large displays on a paired group + * of crtcs. Ideally for multiple large displays we'd assign them to + * non-linked crtcs for maximum line buffer allocation. + */ + if (radeon_crtc->base.enabled && mode) { + if (other_mode) + tmp = 0; /* 1/2 */ + else + tmp = 2; /* whole */ + } else + tmp = 0; + + /* second controller of the pair uses second half of the lb */ + if (radeon_crtc->crtc_id % 2) + tmp += 4; + WREG32(DC_LB_MEMORY_SPLIT + radeon_crtc->crtc_offset, tmp); + + if (radeon_crtc->base.enabled && mode) { + switch (tmp) { + case 0: + case 4: + default: + if (ASIC_IS_DCE5(rdev)) + return 4096 * 2; + else + return 3840 * 2; + case 1: + case 5: + if (ASIC_IS_DCE5(rdev)) + return 6144 * 2; + else + return 5760 * 2; + case 2: + case 6: + if (ASIC_IS_DCE5(rdev)) + return 8192 * 2; + else + return 7680 * 2; + case 3: + case 7: + if (ASIC_IS_DCE5(rdev)) + return 2048 * 2; + else + return 1920 * 2; + } + } + + /* controller not enabled, so no lb used */ + return 0; +} + +static u32 evergreen_get_number_of_dram_channels(struct radeon_device *rdev) +{ + u32 tmp = RREG32(MC_SHARED_CHMAP); + + switch ((tmp & NOOFCHAN_MASK) >> NOOFCHAN_SHIFT) { + case 0: + default: + return 1; + case 1: + return 2; + case 2: + return 4; + case 3: + return 8; + } +} + +struct evergreen_wm_params { + u32 dram_channels; /* number of dram channels */ + u32 yclk; /* bandwidth per dram data pin in kHz */ + u32 sclk; /* engine clock in kHz */ + u32 disp_clk; /* display clock in kHz */ + u32 src_width; /* viewport width */ + u32 active_time; /* active display time in ns */ + u32 blank_time; /* blank time in ns */ + bool interlaced; /* mode is interlaced */ + fixed20_12 vsc; /* vertical scale ratio */ + u32 num_heads; /* number of active crtcs */ + u32 bytes_per_pixel; /* bytes per pixel display + overlay */ + u32 lb_size; /* line buffer allocated to pipe */ + u32 vtaps; /* vertical scaler taps */ +}; + +static u32 evergreen_dram_bandwidth(struct evergreen_wm_params *wm) +{ + /* Calculate DRAM Bandwidth and the part allocated to display. */ + fixed20_12 dram_efficiency; /* 0.7 */ + fixed20_12 yclk, dram_channels, bandwidth; + fixed20_12 a; + + a.full = dfixed_const(1000); + yclk.full = dfixed_const(wm->yclk); + yclk.full = dfixed_div(yclk, a); + dram_channels.full = dfixed_const(wm->dram_channels * 4); + a.full = dfixed_const(10); + dram_efficiency.full = dfixed_const(7); + dram_efficiency.full = dfixed_div(dram_efficiency, a); + bandwidth.full = dfixed_mul(dram_channels, yclk); + bandwidth.full = dfixed_mul(bandwidth, dram_efficiency); + + return dfixed_trunc(bandwidth); +} + +static u32 evergreen_dram_bandwidth_for_display(struct evergreen_wm_params *wm) +{ + /* Calculate DRAM Bandwidth and the part allocated to display. */ + fixed20_12 disp_dram_allocation; /* 0.3 to 0.7 */ + fixed20_12 yclk, dram_channels, bandwidth; + fixed20_12 a; + + a.full = dfixed_const(1000); + yclk.full = dfixed_const(wm->yclk); + yclk.full = dfixed_div(yclk, a); + dram_channels.full = dfixed_const(wm->dram_channels * 4); + a.full = dfixed_const(10); + disp_dram_allocation.full = dfixed_const(3); /* XXX worse case value 0.3 */ + disp_dram_allocation.full = dfixed_div(disp_dram_allocation, a); + bandwidth.full = dfixed_mul(dram_channels, yclk); + bandwidth.full = dfixed_mul(bandwidth, disp_dram_allocation); + + return dfixed_trunc(bandwidth); +} + +static u32 evergreen_data_return_bandwidth(struct evergreen_wm_params *wm) +{ + /* Calculate the display Data return Bandwidth */ + fixed20_12 return_efficiency; /* 0.8 */ + fixed20_12 sclk, bandwidth; + fixed20_12 a; + + a.full = dfixed_const(1000); + sclk.full = dfixed_const(wm->sclk); + sclk.full = dfixed_div(sclk, a); + a.full = dfixed_const(10); + return_efficiency.full = dfixed_const(8); + return_efficiency.full = dfixed_div(return_efficiency, a); + a.full = dfixed_const(32); + bandwidth.full = dfixed_mul(a, sclk); + bandwidth.full = dfixed_mul(bandwidth, return_efficiency); + + return dfixed_trunc(bandwidth); +} + +static u32 evergreen_dmif_request_bandwidth(struct evergreen_wm_params *wm) +{ + /* Calculate the DMIF Request Bandwidth */ + fixed20_12 disp_clk_request_efficiency; /* 0.8 */ + fixed20_12 disp_clk, bandwidth; + fixed20_12 a; + + a.full = dfixed_const(1000); + disp_clk.full = dfixed_const(wm->disp_clk); + disp_clk.full = dfixed_div(disp_clk, a); + a.full = dfixed_const(10); + disp_clk_request_efficiency.full = dfixed_const(8); + disp_clk_request_efficiency.full = dfixed_div(disp_clk_request_efficiency, a); + a.full = dfixed_const(32); + bandwidth.full = dfixed_mul(a, disp_clk); + bandwidth.full = dfixed_mul(bandwidth, disp_clk_request_efficiency); + + return dfixed_trunc(bandwidth); +} + +static u32 evergreen_available_bandwidth(struct evergreen_wm_params *wm) +{ + /* Calculate the Available bandwidth. Display can use this temporarily but not in average. */ + u32 dram_bandwidth = evergreen_dram_bandwidth(wm); + u32 data_return_bandwidth = evergreen_data_return_bandwidth(wm); + u32 dmif_req_bandwidth = evergreen_dmif_request_bandwidth(wm); + + return min(dram_bandwidth, min(data_return_bandwidth, dmif_req_bandwidth)); +} + +static u32 evergreen_average_bandwidth(struct evergreen_wm_params *wm) +{ + /* Calculate the display mode Average Bandwidth + * DisplayMode should contain the source and destination dimensions, + * timing, etc. + */ + fixed20_12 bpp; + fixed20_12 line_time; + fixed20_12 src_width; + fixed20_12 bandwidth; + fixed20_12 a; + + a.full = dfixed_const(1000); + line_time.full = dfixed_const(wm->active_time + wm->blank_time); + line_time.full = dfixed_div(line_time, a); + bpp.full = dfixed_const(wm->bytes_per_pixel); + src_width.full = dfixed_const(wm->src_width); + bandwidth.full = dfixed_mul(src_width, bpp); + bandwidth.full = dfixed_mul(bandwidth, wm->vsc); + bandwidth.full = dfixed_div(bandwidth, line_time); + + return dfixed_trunc(bandwidth); +} + +static u32 evergreen_latency_watermark(struct evergreen_wm_params *wm) +{ + /* First calcualte the latency in ns */ + u32 mc_latency = 2000; /* 2000 ns. */ + u32 available_bandwidth = evergreen_available_bandwidth(wm); + u32 worst_chunk_return_time = (512 * 8 * 1000) / available_bandwidth; + u32 cursor_line_pair_return_time = (128 * 4 * 1000) / available_bandwidth; + u32 dc_latency = 40000000 / wm->disp_clk; /* dc pipe latency */ + u32 other_heads_data_return_time = ((wm->num_heads + 1) * worst_chunk_return_time) + + (wm->num_heads * cursor_line_pair_return_time); + u32 latency = mc_latency + other_heads_data_return_time + dc_latency; + u32 max_src_lines_per_dst_line, lb_fill_bw, line_fill_time; + fixed20_12 a, b, c; + + if (wm->num_heads == 0) + return 0; + + a.full = dfixed_const(2); + b.full = dfixed_const(1); + if ((wm->vsc.full > a.full) || + ((wm->vsc.full > b.full) && (wm->vtaps >= 3)) || + (wm->vtaps >= 5) || + ((wm->vsc.full >= a.full) && wm->interlaced)) + max_src_lines_per_dst_line = 4; + else + max_src_lines_per_dst_line = 2; + + a.full = dfixed_const(available_bandwidth); + b.full = dfixed_const(wm->num_heads); + a.full = dfixed_div(a, b); + + b.full = dfixed_const(1000); + c.full = dfixed_const(wm->disp_clk); + b.full = dfixed_div(c, b); + c.full = dfixed_const(wm->bytes_per_pixel); + b.full = dfixed_mul(b, c); + + lb_fill_bw = min(dfixed_trunc(a), dfixed_trunc(b)); + + a.full = dfixed_const(max_src_lines_per_dst_line * wm->src_width * wm->bytes_per_pixel); + b.full = dfixed_const(1000); + c.full = dfixed_const(lb_fill_bw); + b.full = dfixed_div(c, b); + a.full = dfixed_div(a, b); + line_fill_time = dfixed_trunc(a); + + if (line_fill_time < wm->active_time) + return latency; + else + return latency + (line_fill_time - wm->active_time); + +} + +static bool evergreen_average_bandwidth_vs_dram_bandwidth_for_display(struct evergreen_wm_params *wm) +{ + if (evergreen_average_bandwidth(wm) <= + (evergreen_dram_bandwidth_for_display(wm) / wm->num_heads)) + return true; + else + return false; +}; + +static bool evergreen_average_bandwidth_vs_available_bandwidth(struct evergreen_wm_params *wm) +{ + if (evergreen_average_bandwidth(wm) <= + (evergreen_available_bandwidth(wm) / wm->num_heads)) + return true; + else + return false; +}; + +static bool evergreen_check_latency_hiding(struct evergreen_wm_params *wm) +{ + u32 lb_partitions = wm->lb_size / wm->src_width; + u32 line_time = wm->active_time + wm->blank_time; + u32 latency_tolerant_lines; + u32 latency_hiding; + fixed20_12 a; + + a.full = dfixed_const(1); + if (wm->vsc.full > a.full) + latency_tolerant_lines = 1; + else { + if (lb_partitions <= (wm->vtaps + 1)) + latency_tolerant_lines = 1; + else + latency_tolerant_lines = 2; + } + + latency_hiding = (latency_tolerant_lines * line_time + wm->blank_time); + + if (evergreen_latency_watermark(wm) <= latency_hiding) + return true; + else + return false; +} + +static void evergreen_program_watermarks(struct radeon_device *rdev, + struct radeon_crtc *radeon_crtc, + u32 lb_size, u32 num_heads) +{ + struct drm_display_mode *mode = &radeon_crtc->base.mode; + struct evergreen_wm_params wm; + u32 pixel_period; + u32 line_time = 0; + u32 latency_watermark_a = 0, latency_watermark_b = 0; + u32 priority_a_mark = 0, priority_b_mark = 0; + u32 priority_a_cnt = PRIORITY_OFF; + u32 priority_b_cnt = PRIORITY_OFF; + u32 pipe_offset = radeon_crtc->crtc_id * 16; + u32 tmp, arb_control3; + fixed20_12 a, b, c; + + if (radeon_crtc->base.enabled && num_heads && mode) { + pixel_period = 1000000 / (u32)mode->clock; + line_time = min((u32)mode->crtc_htotal * pixel_period, (u32)65535); + priority_a_cnt = 0; + priority_b_cnt = 0; + + wm.yclk = rdev->pm.current_mclk * 10; + wm.sclk = rdev->pm.current_sclk * 10; + wm.disp_clk = mode->clock; + wm.src_width = mode->crtc_hdisplay; + wm.active_time = mode->crtc_hdisplay * pixel_period; + wm.blank_time = line_time - wm.active_time; + wm.interlaced = false; + if (mode->flags & DRM_MODE_FLAG_INTERLACE) + wm.interlaced = true; + wm.vsc = radeon_crtc->vsc; + wm.vtaps = 1; + if (radeon_crtc->rmx_type != RMX_OFF) + wm.vtaps = 2; + wm.bytes_per_pixel = 4; /* XXX: get this from fb config */ + wm.lb_size = lb_size; + wm.dram_channels = evergreen_get_number_of_dram_channels(rdev); + wm.num_heads = num_heads; + + /* set for high clocks */ + latency_watermark_a = min(evergreen_latency_watermark(&wm), (u32)65535); + /* set for low clocks */ + /* wm.yclk = low clk; wm.sclk = low clk */ + latency_watermark_b = min(evergreen_latency_watermark(&wm), (u32)65535); + + /* possibly force display priority to high */ + /* should really do this at mode validation time... */ + if (!evergreen_average_bandwidth_vs_dram_bandwidth_for_display(&wm) || + !evergreen_average_bandwidth_vs_available_bandwidth(&wm) || + !evergreen_check_latency_hiding(&wm) || + (rdev->disp_priority == 2)) { + DRM_INFO("force priority to high\n"); + priority_a_cnt |= PRIORITY_ALWAYS_ON; + priority_b_cnt |= PRIORITY_ALWAYS_ON; + } + + a.full = dfixed_const(1000); + b.full = dfixed_const(mode->clock); + b.full = dfixed_div(b, a); + c.full = dfixed_const(latency_watermark_a); + c.full = dfixed_mul(c, b); + c.full = dfixed_mul(c, radeon_crtc->hsc); + c.full = dfixed_div(c, a); + a.full = dfixed_const(16); + c.full = dfixed_div(c, a); + priority_a_mark = dfixed_trunc(c); + priority_a_cnt |= priority_a_mark & PRIORITY_MARK_MASK; + + a.full = dfixed_const(1000); + b.full = dfixed_const(mode->clock); + b.full = dfixed_div(b, a); + c.full = dfixed_const(latency_watermark_b); + c.full = dfixed_mul(c, b); + c.full = dfixed_mul(c, radeon_crtc->hsc); + c.full = dfixed_div(c, a); + a.full = dfixed_const(16); + c.full = dfixed_div(c, a); + priority_b_mark = dfixed_trunc(c); + priority_b_cnt |= priority_b_mark & PRIORITY_MARK_MASK; + } + + /* select wm A */ + arb_control3 = RREG32(PIPE0_ARBITRATION_CONTROL3 + pipe_offset); + tmp = arb_control3; + tmp &= ~LATENCY_WATERMARK_MASK(3); + tmp |= LATENCY_WATERMARK_MASK(1); + WREG32(PIPE0_ARBITRATION_CONTROL3 + pipe_offset, tmp); + WREG32(PIPE0_LATENCY_CONTROL + pipe_offset, + (LATENCY_LOW_WATERMARK(latency_watermark_a) | + LATENCY_HIGH_WATERMARK(line_time))); + /* select wm B */ + tmp = RREG32(PIPE0_ARBITRATION_CONTROL3 + pipe_offset); + tmp &= ~LATENCY_WATERMARK_MASK(3); + tmp |= LATENCY_WATERMARK_MASK(2); + WREG32(PIPE0_ARBITRATION_CONTROL3 + pipe_offset, tmp); + WREG32(PIPE0_LATENCY_CONTROL + pipe_offset, + (LATENCY_LOW_WATERMARK(latency_watermark_b) | + LATENCY_HIGH_WATERMARK(line_time))); + /* restore original selection */ + WREG32(PIPE0_ARBITRATION_CONTROL3 + pipe_offset, arb_control3); + + /* write the priority marks */ + WREG32(PRIORITY_A_CNT + radeon_crtc->crtc_offset, priority_a_cnt); + WREG32(PRIORITY_B_CNT + radeon_crtc->crtc_offset, priority_b_cnt); + +} + +void evergreen_bandwidth_update(struct radeon_device *rdev) +{ + struct drm_display_mode *mode0 = NULL; + struct drm_display_mode *mode1 = NULL; + u32 num_heads = 0, lb_size; + int i; + + radeon_update_display_priority(rdev); + + for (i = 0; i < rdev->num_crtc; i++) { + if (rdev->mode_info.crtcs[i]->base.enabled) + num_heads++; + } + for (i = 0; i < rdev->num_crtc; i += 2) { + mode0 = &rdev->mode_info.crtcs[i]->base.mode; + mode1 = &rdev->mode_info.crtcs[i+1]->base.mode; + lb_size = evergreen_line_buffer_adjust(rdev, rdev->mode_info.crtcs[i], mode0, mode1); + evergreen_program_watermarks(rdev, rdev->mode_info.crtcs[i], lb_size, num_heads); + lb_size = evergreen_line_buffer_adjust(rdev, rdev->mode_info.crtcs[i+1], mode1, mode0); + evergreen_program_watermarks(rdev, rdev->mode_info.crtcs[i+1], lb_size, num_heads); + } +} + +int evergreen_mc_wait_for_idle(struct radeon_device *rdev) +{ + unsigned i; + u32 tmp; + + for (i = 0; i < rdev->usec_timeout; i++) { + /* read MC_STATUS */ + tmp = RREG32(SRBM_STATUS) & 0x1F00; + if (!tmp) + return 0; + udelay(1); + } + return -1; +} + +/* + * GART + */ +void evergreen_pcie_gart_tlb_flush(struct radeon_device *rdev) +{ + unsigned i; + u32 tmp; + + WREG32(HDP_MEM_COHERENCY_FLUSH_CNTL, 0x1); + + WREG32(VM_CONTEXT0_REQUEST_RESPONSE, REQUEST_TYPE(1)); + for (i = 0; i < rdev->usec_timeout; i++) { + /* read MC_STATUS */ + tmp = RREG32(VM_CONTEXT0_REQUEST_RESPONSE); + tmp = (tmp & RESPONSE_TYPE_MASK) >> RESPONSE_TYPE_SHIFT; + if (tmp == 2) { + printk(KERN_WARNING "[drm] r600 flush TLB failed\n"); + return; + } + if (tmp) { + return; + } + udelay(1); + } +} + +int evergreen_pcie_gart_enable(struct radeon_device *rdev) +{ + u32 tmp; + int r; + + if (rdev->gart.table.vram.robj == NULL) { + dev_err(rdev->dev, "No VRAM object for PCIE GART.\n"); + return -EINVAL; + } + r = radeon_gart_table_vram_pin(rdev); + if (r) + return r; + radeon_gart_restore(rdev); + /* Setup L2 cache */ + WREG32(VM_L2_CNTL, ENABLE_L2_CACHE | ENABLE_L2_FRAGMENT_PROCESSING | + ENABLE_L2_PTE_CACHE_LRU_UPDATE_BY_WRITE | + EFFECTIVE_L2_QUEUE_SIZE(7)); + WREG32(VM_L2_CNTL2, 0); + WREG32(VM_L2_CNTL3, BANK_SELECT(0) | CACHE_UPDATE_MODE(2)); + /* Setup TLB control */ + tmp = ENABLE_L1_TLB | ENABLE_L1_FRAGMENT_PROCESSING | + SYSTEM_ACCESS_MODE_NOT_IN_SYS | + SYSTEM_APERTURE_UNMAPPED_ACCESS_PASS_THRU | + EFFECTIVE_L1_TLB_SIZE(5) | EFFECTIVE_L1_QUEUE_SIZE(5); + if (rdev->flags & RADEON_IS_IGP) { + WREG32(FUS_MC_VM_MD_L1_TLB0_CNTL, tmp); + WREG32(FUS_MC_VM_MD_L1_TLB1_CNTL, tmp); + WREG32(FUS_MC_VM_MD_L1_TLB2_CNTL, tmp); + } else { + WREG32(MC_VM_MD_L1_TLB0_CNTL, tmp); + WREG32(MC_VM_MD_L1_TLB1_CNTL, tmp); + WREG32(MC_VM_MD_L1_TLB2_CNTL, tmp); + if ((rdev->family == CHIP_JUNIPER) || + (rdev->family == CHIP_CYPRESS) || + (rdev->family == CHIP_HEMLOCK) || + (rdev->family == CHIP_BARTS)) + WREG32(MC_VM_MD_L1_TLB3_CNTL, tmp); + } + WREG32(MC_VM_MB_L1_TLB0_CNTL, tmp); + WREG32(MC_VM_MB_L1_TLB1_CNTL, tmp); + WREG32(MC_VM_MB_L1_TLB2_CNTL, tmp); + WREG32(MC_VM_MB_L1_TLB3_CNTL, tmp); + WREG32(VM_CONTEXT0_PAGE_TABLE_START_ADDR, rdev->mc.gtt_start >> 12); + WREG32(VM_CONTEXT0_PAGE_TABLE_END_ADDR, rdev->mc.gtt_end >> 12); + WREG32(VM_CONTEXT0_PAGE_TABLE_BASE_ADDR, rdev->gart.table_addr >> 12); + WREG32(VM_CONTEXT0_CNTL, ENABLE_CONTEXT | PAGE_TABLE_DEPTH(0) | + RANGE_PROTECTION_FAULT_ENABLE_DEFAULT); + WREG32(VM_CONTEXT0_PROTECTION_FAULT_DEFAULT_ADDR, + (u32)(rdev->dummy_page.addr >> 12)); + WREG32(VM_CONTEXT1_CNTL, 0); + + evergreen_pcie_gart_tlb_flush(rdev); + rdev->gart.ready = true; + return 0; +} + +void evergreen_pcie_gart_disable(struct radeon_device *rdev) +{ + u32 tmp; + int r; + + /* Disable all tables */ + WREG32(VM_CONTEXT0_CNTL, 0); + WREG32(VM_CONTEXT1_CNTL, 0); + + /* Setup L2 cache */ + WREG32(VM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING | + EFFECTIVE_L2_QUEUE_SIZE(7)); + WREG32(VM_L2_CNTL2, 0); + WREG32(VM_L2_CNTL3, BANK_SELECT(0) | CACHE_UPDATE_MODE(2)); + /* Setup TLB control */ + tmp = EFFECTIVE_L1_TLB_SIZE(5) | EFFECTIVE_L1_QUEUE_SIZE(5); + WREG32(MC_VM_MD_L1_TLB0_CNTL, tmp); + WREG32(MC_VM_MD_L1_TLB1_CNTL, tmp); + WREG32(MC_VM_MD_L1_TLB2_CNTL, tmp); + WREG32(MC_VM_MB_L1_TLB0_CNTL, tmp); + WREG32(MC_VM_MB_L1_TLB1_CNTL, tmp); + WREG32(MC_VM_MB_L1_TLB2_CNTL, tmp); + WREG32(MC_VM_MB_L1_TLB3_CNTL, tmp); + if (rdev->gart.table.vram.robj) { + r = radeon_bo_reserve(rdev->gart.table.vram.robj, false); + if (likely(r == 0)) { + radeon_bo_kunmap(rdev->gart.table.vram.robj); + radeon_bo_unpin(rdev->gart.table.vram.robj); + radeon_bo_unreserve(rdev->gart.table.vram.robj); + } + } +} + +void evergreen_pcie_gart_fini(struct radeon_device *rdev) +{ + evergreen_pcie_gart_disable(rdev); + radeon_gart_table_vram_free(rdev); + radeon_gart_fini(rdev); +} + + +void evergreen_agp_enable(struct radeon_device *rdev) +{ + u32 tmp; + + /* Setup L2 cache */ + WREG32(VM_L2_CNTL, ENABLE_L2_CACHE | ENABLE_L2_FRAGMENT_PROCESSING | + ENABLE_L2_PTE_CACHE_LRU_UPDATE_BY_WRITE | + EFFECTIVE_L2_QUEUE_SIZE(7)); + WREG32(VM_L2_CNTL2, 0); + WREG32(VM_L2_CNTL3, BANK_SELECT(0) | CACHE_UPDATE_MODE(2)); + /* Setup TLB control */ + tmp = ENABLE_L1_TLB | ENABLE_L1_FRAGMENT_PROCESSING | + SYSTEM_ACCESS_MODE_NOT_IN_SYS | + SYSTEM_APERTURE_UNMAPPED_ACCESS_PASS_THRU | + EFFECTIVE_L1_TLB_SIZE(5) | EFFECTIVE_L1_QUEUE_SIZE(5); + WREG32(MC_VM_MD_L1_TLB0_CNTL, tmp); + WREG32(MC_VM_MD_L1_TLB1_CNTL, tmp); + WREG32(MC_VM_MD_L1_TLB2_CNTL, tmp); + WREG32(MC_VM_MB_L1_TLB0_CNTL, tmp); + WREG32(MC_VM_MB_L1_TLB1_CNTL, tmp); + WREG32(MC_VM_MB_L1_TLB2_CNTL, tmp); + WREG32(MC_VM_MB_L1_TLB3_CNTL, tmp); + WREG32(VM_CONTEXT0_CNTL, 0); + WREG32(VM_CONTEXT1_CNTL, 0); +} + +void evergreen_mc_stop(struct radeon_device *rdev, struct evergreen_mc_save *save) +{ + save->vga_control[0] = RREG32(D1VGA_CONTROL); + save->vga_control[1] = RREG32(D2VGA_CONTROL); + save->vga_render_control = RREG32(VGA_RENDER_CONTROL); + save->vga_hdp_control = RREG32(VGA_HDP_CONTROL); + save->crtc_control[0] = RREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC0_REGISTER_OFFSET); + save->crtc_control[1] = RREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC1_REGISTER_OFFSET); + if (rdev->num_crtc >= 4) { + save->vga_control[2] = RREG32(EVERGREEN_D3VGA_CONTROL); + save->vga_control[3] = RREG32(EVERGREEN_D4VGA_CONTROL); + save->crtc_control[2] = RREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC2_REGISTER_OFFSET); + save->crtc_control[3] = RREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC3_REGISTER_OFFSET); + } + if (rdev->num_crtc >= 6) { + save->vga_control[4] = RREG32(EVERGREEN_D5VGA_CONTROL); + save->vga_control[5] = RREG32(EVERGREEN_D6VGA_CONTROL); + save->crtc_control[4] = RREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC4_REGISTER_OFFSET); + save->crtc_control[5] = RREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC5_REGISTER_OFFSET); + } + + /* Stop all video */ + WREG32(VGA_RENDER_CONTROL, 0); + WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC0_REGISTER_OFFSET, 1); + WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC1_REGISTER_OFFSET, 1); + if (rdev->num_crtc >= 4) { + WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC2_REGISTER_OFFSET, 1); + WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC3_REGISTER_OFFSET, 1); + } + if (rdev->num_crtc >= 6) { + WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC4_REGISTER_OFFSET, 1); + WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC5_REGISTER_OFFSET, 1); + } + WREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC0_REGISTER_OFFSET, 0); + WREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC1_REGISTER_OFFSET, 0); + if (rdev->num_crtc >= 4) { + WREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC2_REGISTER_OFFSET, 0); + WREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC3_REGISTER_OFFSET, 0); + } + if (rdev->num_crtc >= 6) { + WREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC4_REGISTER_OFFSET, 0); + WREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC5_REGISTER_OFFSET, 0); + } + WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC0_REGISTER_OFFSET, 0); + WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC1_REGISTER_OFFSET, 0); + if (rdev->num_crtc >= 4) { + WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC2_REGISTER_OFFSET, 0); + WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC3_REGISTER_OFFSET, 0); + } + if (rdev->num_crtc >= 6) { + WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC4_REGISTER_OFFSET, 0); + WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC5_REGISTER_OFFSET, 0); + } + + WREG32(D1VGA_CONTROL, 0); + WREG32(D2VGA_CONTROL, 0); + if (rdev->num_crtc >= 4) { + WREG32(EVERGREEN_D3VGA_CONTROL, 0); + WREG32(EVERGREEN_D4VGA_CONTROL, 0); + } + if (rdev->num_crtc >= 6) { + WREG32(EVERGREEN_D5VGA_CONTROL, 0); + WREG32(EVERGREEN_D6VGA_CONTROL, 0); + } +} + +void evergreen_mc_resume(struct radeon_device *rdev, struct evergreen_mc_save *save) +{ + WREG32(EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS_HIGH + EVERGREEN_CRTC0_REGISTER_OFFSET, + upper_32_bits(rdev->mc.vram_start)); + WREG32(EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS_HIGH + EVERGREEN_CRTC0_REGISTER_OFFSET, + upper_32_bits(rdev->mc.vram_start)); + WREG32(EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS + EVERGREEN_CRTC0_REGISTER_OFFSET, + (u32)rdev->mc.vram_start); + WREG32(EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS + EVERGREEN_CRTC0_REGISTER_OFFSET, + (u32)rdev->mc.vram_start); + + WREG32(EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS_HIGH + EVERGREEN_CRTC1_REGISTER_OFFSET, + upper_32_bits(rdev->mc.vram_start)); + WREG32(EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS_HIGH + EVERGREEN_CRTC1_REGISTER_OFFSET, + upper_32_bits(rdev->mc.vram_start)); + WREG32(EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS + EVERGREEN_CRTC1_REGISTER_OFFSET, + (u32)rdev->mc.vram_start); + WREG32(EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS + EVERGREEN_CRTC1_REGISTER_OFFSET, + (u32)rdev->mc.vram_start); + + if (rdev->num_crtc >= 4) { + WREG32(EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS_HIGH + EVERGREEN_CRTC2_REGISTER_OFFSET, + upper_32_bits(rdev->mc.vram_start)); + WREG32(EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS_HIGH + EVERGREEN_CRTC2_REGISTER_OFFSET, + upper_32_bits(rdev->mc.vram_start)); + WREG32(EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS + EVERGREEN_CRTC2_REGISTER_OFFSET, + (u32)rdev->mc.vram_start); + WREG32(EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS + EVERGREEN_CRTC2_REGISTER_OFFSET, + (u32)rdev->mc.vram_start); + + WREG32(EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS_HIGH + EVERGREEN_CRTC3_REGISTER_OFFSET, + upper_32_bits(rdev->mc.vram_start)); + WREG32(EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS_HIGH + EVERGREEN_CRTC3_REGISTER_OFFSET, + upper_32_bits(rdev->mc.vram_start)); + WREG32(EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS + EVERGREEN_CRTC3_REGISTER_OFFSET, + (u32)rdev->mc.vram_start); + WREG32(EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS + EVERGREEN_CRTC3_REGISTER_OFFSET, + (u32)rdev->mc.vram_start); + } + if (rdev->num_crtc >= 6) { + WREG32(EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS_HIGH + EVERGREEN_CRTC4_REGISTER_OFFSET, + upper_32_bits(rdev->mc.vram_start)); + WREG32(EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS_HIGH + EVERGREEN_CRTC4_REGISTER_OFFSET, + upper_32_bits(rdev->mc.vram_start)); + WREG32(EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS + EVERGREEN_CRTC4_REGISTER_OFFSET, + (u32)rdev->mc.vram_start); + WREG32(EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS + EVERGREEN_CRTC4_REGISTER_OFFSET, + (u32)rdev->mc.vram_start); + + WREG32(EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS_HIGH + EVERGREEN_CRTC5_REGISTER_OFFSET, + upper_32_bits(rdev->mc.vram_start)); + WREG32(EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS_HIGH + EVERGREEN_CRTC5_REGISTER_OFFSET, + upper_32_bits(rdev->mc.vram_start)); + WREG32(EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS + EVERGREEN_CRTC5_REGISTER_OFFSET, + (u32)rdev->mc.vram_start); + WREG32(EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS + EVERGREEN_CRTC5_REGISTER_OFFSET, + (u32)rdev->mc.vram_start); + } + + WREG32(EVERGREEN_VGA_MEMORY_BASE_ADDRESS_HIGH, upper_32_bits(rdev->mc.vram_start)); + WREG32(EVERGREEN_VGA_MEMORY_BASE_ADDRESS, (u32)rdev->mc.vram_start); + /* Unlock host access */ + WREG32(VGA_HDP_CONTROL, save->vga_hdp_control); + mdelay(1); + /* Restore video state */ + WREG32(D1VGA_CONTROL, save->vga_control[0]); + WREG32(D2VGA_CONTROL, save->vga_control[1]); + if (rdev->num_crtc >= 4) { + WREG32(EVERGREEN_D3VGA_CONTROL, save->vga_control[2]); + WREG32(EVERGREEN_D4VGA_CONTROL, save->vga_control[3]); + } + if (rdev->num_crtc >= 6) { + WREG32(EVERGREEN_D5VGA_CONTROL, save->vga_control[4]); + WREG32(EVERGREEN_D6VGA_CONTROL, save->vga_control[5]); + } + WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC0_REGISTER_OFFSET, 1); + WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC1_REGISTER_OFFSET, 1); + if (rdev->num_crtc >= 4) { + WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC2_REGISTER_OFFSET, 1); + WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC3_REGISTER_OFFSET, 1); + } + if (rdev->num_crtc >= 6) { + WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC4_REGISTER_OFFSET, 1); + WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC5_REGISTER_OFFSET, 1); + } + WREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC0_REGISTER_OFFSET, save->crtc_control[0]); + WREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC1_REGISTER_OFFSET, save->crtc_control[1]); + if (rdev->num_crtc >= 4) { + WREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC2_REGISTER_OFFSET, save->crtc_control[2]); + WREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC3_REGISTER_OFFSET, save->crtc_control[3]); + } + if (rdev->num_crtc >= 6) { + WREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC4_REGISTER_OFFSET, save->crtc_control[4]); + WREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC5_REGISTER_OFFSET, save->crtc_control[5]); + } + WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC0_REGISTER_OFFSET, 0); + WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC1_REGISTER_OFFSET, 0); + if (rdev->num_crtc >= 4) { + WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC2_REGISTER_OFFSET, 0); + WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC3_REGISTER_OFFSET, 0); + } + if (rdev->num_crtc >= 6) { + WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC4_REGISTER_OFFSET, 0); + WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC5_REGISTER_OFFSET, 0); + } + WREG32(VGA_RENDER_CONTROL, save->vga_render_control); +} + +void evergreen_mc_program(struct radeon_device *rdev) +{ + struct evergreen_mc_save save; + u32 tmp; + int i, j; + + /* Initialize HDP */ + for (i = 0, j = 0; i < 32; i++, j += 0x18) { + WREG32((0x2c14 + j), 0x00000000); + WREG32((0x2c18 + j), 0x00000000); + WREG32((0x2c1c + j), 0x00000000); + WREG32((0x2c20 + j), 0x00000000); + WREG32((0x2c24 + j), 0x00000000); + } + WREG32(HDP_REG_COHERENCY_FLUSH_CNTL, 0); + + evergreen_mc_stop(rdev, &save); + if (evergreen_mc_wait_for_idle(rdev)) { + dev_warn(rdev->dev, "Wait for MC idle timedout !\n"); + } + /* Lockout access through VGA aperture*/ + WREG32(VGA_HDP_CONTROL, VGA_MEMORY_DISABLE); + /* Update configuration */ + if (rdev->flags & RADEON_IS_AGP) { + if (rdev->mc.vram_start < rdev->mc.gtt_start) { + /* VRAM before AGP */ + WREG32(MC_VM_SYSTEM_APERTURE_LOW_ADDR, + rdev->mc.vram_start >> 12); + WREG32(MC_VM_SYSTEM_APERTURE_HIGH_ADDR, + rdev->mc.gtt_end >> 12); + } else { + /* VRAM after AGP */ + WREG32(MC_VM_SYSTEM_APERTURE_LOW_ADDR, + rdev->mc.gtt_start >> 12); + WREG32(MC_VM_SYSTEM_APERTURE_HIGH_ADDR, + rdev->mc.vram_end >> 12); + } + } else { + WREG32(MC_VM_SYSTEM_APERTURE_LOW_ADDR, + rdev->mc.vram_start >> 12); + WREG32(MC_VM_SYSTEM_APERTURE_HIGH_ADDR, + rdev->mc.vram_end >> 12); + } + WREG32(MC_VM_SYSTEM_APERTURE_DEFAULT_ADDR, 0); + if (rdev->flags & RADEON_IS_IGP) { + tmp = RREG32(MC_FUS_VM_FB_OFFSET) & 0x000FFFFF; + tmp |= ((rdev->mc.vram_end >> 20) & 0xF) << 24; + tmp |= ((rdev->mc.vram_start >> 20) & 0xF) << 20; + WREG32(MC_FUS_VM_FB_OFFSET, tmp); + } + tmp = ((rdev->mc.vram_end >> 24) & 0xFFFF) << 16; + tmp |= ((rdev->mc.vram_start >> 24) & 0xFFFF); + WREG32(MC_VM_FB_LOCATION, tmp); + WREG32(HDP_NONSURFACE_BASE, (rdev->mc.vram_start >> 8)); + WREG32(HDP_NONSURFACE_INFO, (2 << 7) | (1 << 30)); + WREG32(HDP_NONSURFACE_SIZE, 0x3FFFFFFF); + if (rdev->flags & RADEON_IS_AGP) { + WREG32(MC_VM_AGP_TOP, rdev->mc.gtt_end >> 16); + WREG32(MC_VM_AGP_BOT, rdev->mc.gtt_start >> 16); + WREG32(MC_VM_AGP_BASE, rdev->mc.agp_base >> 22); + } else { + WREG32(MC_VM_AGP_BASE, 0); + WREG32(MC_VM_AGP_TOP, 0x0FFFFFFF); + WREG32(MC_VM_AGP_BOT, 0x0FFFFFFF); + } + if (evergreen_mc_wait_for_idle(rdev)) { + dev_warn(rdev->dev, "Wait for MC idle timedout !\n"); + } + evergreen_mc_resume(rdev, &save); + /* we need to own VRAM, so turn off the VGA renderer here + * to stop it overwriting our objects */ + rv515_vga_render_disable(rdev); +} + +/* + * CP. + */ +void evergreen_ring_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib) +{ + /* set to DX10/11 mode */ + radeon_ring_write(rdev, PACKET3(PACKET3_MODE_CONTROL, 0)); + radeon_ring_write(rdev, 1); + /* FIXME: implement */ + radeon_ring_write(rdev, PACKET3(PACKET3_INDIRECT_BUFFER, 2)); + radeon_ring_write(rdev, +#ifdef __BIG_ENDIAN + (2 << 0) | +#endif + (ib->gpu_addr & 0xFFFFFFFC)); + radeon_ring_write(rdev, upper_32_bits(ib->gpu_addr) & 0xFF); + radeon_ring_write(rdev, ib->length_dw); +} + + +static int evergreen_cp_load_microcode(struct radeon_device *rdev) +{ + const __be32 *fw_data; + int i; + + if (!rdev->me_fw || !rdev->pfp_fw) + return -EINVAL; + + r700_cp_stop(rdev); + WREG32(CP_RB_CNTL, +#ifdef __BIG_ENDIAN + BUF_SWAP_32BIT | +#endif + RB_NO_UPDATE | RB_BLKSZ(15) | RB_BUFSZ(3)); + + fw_data = (const __be32 *)rdev->pfp_fw->data; + WREG32(CP_PFP_UCODE_ADDR, 0); + for (i = 0; i < EVERGREEN_PFP_UCODE_SIZE; i++) + WREG32(CP_PFP_UCODE_DATA, be32_to_cpup(fw_data++)); + WREG32(CP_PFP_UCODE_ADDR, 0); + + fw_data = (const __be32 *)rdev->me_fw->data; + WREG32(CP_ME_RAM_WADDR, 0); + for (i = 0; i < EVERGREEN_PM4_UCODE_SIZE; i++) + WREG32(CP_ME_RAM_DATA, be32_to_cpup(fw_data++)); + + WREG32(CP_PFP_UCODE_ADDR, 0); + WREG32(CP_ME_RAM_WADDR, 0); + WREG32(CP_ME_RAM_RADDR, 0); + return 0; +} + +static int evergreen_cp_start(struct radeon_device *rdev) +{ + int r, i; + uint32_t cp_me; + + r = radeon_ring_lock(rdev, 7); + if (r) { + DRM_ERROR("radeon: cp failed to lock ring (%d).\n", r); + return r; + } + radeon_ring_write(rdev, PACKET3(PACKET3_ME_INITIALIZE, 5)); + radeon_ring_write(rdev, 0x1); + radeon_ring_write(rdev, 0x0); + radeon_ring_write(rdev, rdev->config.evergreen.max_hw_contexts - 1); + radeon_ring_write(rdev, PACKET3_ME_INITIALIZE_DEVICE_ID(1)); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, 0); + radeon_ring_unlock_commit(rdev); + + cp_me = 0xff; + WREG32(CP_ME_CNTL, cp_me); + + r = radeon_ring_lock(rdev, evergreen_default_size + 19); + if (r) { + DRM_ERROR("radeon: cp failed to lock ring (%d).\n", r); + return r; + } + + /* setup clear context state */ + radeon_ring_write(rdev, PACKET3(PACKET3_PREAMBLE_CNTL, 0)); + radeon_ring_write(rdev, PACKET3_PREAMBLE_BEGIN_CLEAR_STATE); + + for (i = 0; i < evergreen_default_size; i++) + radeon_ring_write(rdev, evergreen_default_state[i]); + + radeon_ring_write(rdev, PACKET3(PACKET3_PREAMBLE_CNTL, 0)); + radeon_ring_write(rdev, PACKET3_PREAMBLE_END_CLEAR_STATE); + + /* set clear context state */ + radeon_ring_write(rdev, PACKET3(PACKET3_CLEAR_STATE, 0)); + radeon_ring_write(rdev, 0); + + /* SQ_VTX_BASE_VTX_LOC */ + radeon_ring_write(rdev, 0xc0026f00); + radeon_ring_write(rdev, 0x00000000); + radeon_ring_write(rdev, 0x00000000); + radeon_ring_write(rdev, 0x00000000); + + /* Clear consts */ + radeon_ring_write(rdev, 0xc0036f00); + radeon_ring_write(rdev, 0x00000bc4); + radeon_ring_write(rdev, 0xffffffff); + radeon_ring_write(rdev, 0xffffffff); + radeon_ring_write(rdev, 0xffffffff); + + radeon_ring_write(rdev, 0xc0026900); + radeon_ring_write(rdev, 0x00000316); + radeon_ring_write(rdev, 0x0000000e); /* VGT_VERTEX_REUSE_BLOCK_CNTL */ + radeon_ring_write(rdev, 0x00000010); /* */ + + radeon_ring_unlock_commit(rdev); + + return 0; +} + +int evergreen_cp_resume(struct radeon_device *rdev) +{ + u32 tmp; + u32 rb_bufsz; + int r; + + /* Reset cp; if cp is reset, then PA, SH, VGT also need to be reset */ + WREG32(GRBM_SOFT_RESET, (SOFT_RESET_CP | + SOFT_RESET_PA | + SOFT_RESET_SH | + SOFT_RESET_VGT | + SOFT_RESET_SPI | + SOFT_RESET_SX)); + RREG32(GRBM_SOFT_RESET); + mdelay(15); + WREG32(GRBM_SOFT_RESET, 0); + RREG32(GRBM_SOFT_RESET); + + /* Set ring buffer size */ + rb_bufsz = drm_order(rdev->cp.ring_size / 8); + tmp = (drm_order(RADEON_GPU_PAGE_SIZE/8) << 8) | rb_bufsz; +#ifdef __BIG_ENDIAN + tmp |= BUF_SWAP_32BIT; +#endif + WREG32(CP_RB_CNTL, tmp); + WREG32(CP_SEM_WAIT_TIMER, 0x4); + + /* Set the write pointer delay */ + WREG32(CP_RB_WPTR_DELAY, 0); + + /* Initialize the ring buffer's read and write pointers */ + WREG32(CP_RB_CNTL, tmp | RB_RPTR_WR_ENA); + WREG32(CP_RB_RPTR_WR, 0); + rdev->cp.wptr = 0; + WREG32(CP_RB_WPTR, rdev->cp.wptr); + + /* set the wb address wether it's enabled or not */ + WREG32(CP_RB_RPTR_ADDR, +#ifdef __BIG_ENDIAN + RB_RPTR_SWAP(2) | +#endif + ((rdev->wb.gpu_addr + RADEON_WB_CP_RPTR_OFFSET) & 0xFFFFFFFC)); + WREG32(CP_RB_RPTR_ADDR_HI, upper_32_bits(rdev->wb.gpu_addr + RADEON_WB_CP_RPTR_OFFSET) & 0xFF); + WREG32(SCRATCH_ADDR, ((rdev->wb.gpu_addr + RADEON_WB_SCRATCH_OFFSET) >> 8) & 0xFFFFFFFF); + + if (rdev->wb.enabled) + WREG32(SCRATCH_UMSK, 0xff); + else { + tmp |= RB_NO_UPDATE; + WREG32(SCRATCH_UMSK, 0); + } + + mdelay(1); + WREG32(CP_RB_CNTL, tmp); + + WREG32(CP_RB_BASE, rdev->cp.gpu_addr >> 8); + WREG32(CP_DEBUG, (1 << 27) | (1 << 28)); + + rdev->cp.rptr = RREG32(CP_RB_RPTR); + + evergreen_cp_start(rdev); + rdev->cp.ready = true; + r = radeon_ring_test(rdev); + if (r) { + rdev->cp.ready = false; + return r; + } + return 0; +} + +/* + * Core functions + */ +static u32 evergreen_get_tile_pipe_to_backend_map(struct radeon_device *rdev, + u32 num_tile_pipes, + u32 num_backends, + u32 backend_disable_mask) +{ + u32 backend_map = 0; + u32 enabled_backends_mask = 0; + u32 enabled_backends_count = 0; + u32 cur_pipe; + u32 swizzle_pipe[EVERGREEN_MAX_PIPES]; + u32 cur_backend = 0; + u32 i; + bool force_no_swizzle; + + if (num_tile_pipes > EVERGREEN_MAX_PIPES) + num_tile_pipes = EVERGREEN_MAX_PIPES; + if (num_tile_pipes < 1) + num_tile_pipes = 1; + if (num_backends > EVERGREEN_MAX_BACKENDS) + num_backends = EVERGREEN_MAX_BACKENDS; + if (num_backends < 1) + num_backends = 1; + + for (i = 0; i < EVERGREEN_MAX_BACKENDS; ++i) { + if (((backend_disable_mask >> i) & 1) == 0) { + enabled_backends_mask |= (1 << i); + ++enabled_backends_count; + } + if (enabled_backends_count == num_backends) + break; + } + + if (enabled_backends_count == 0) { + enabled_backends_mask = 1; + enabled_backends_count = 1; + } + + if (enabled_backends_count != num_backends) + num_backends = enabled_backends_count; + + memset((uint8_t *)&swizzle_pipe[0], 0, sizeof(u32) * EVERGREEN_MAX_PIPES); + switch (rdev->family) { + case CHIP_CEDAR: + case CHIP_REDWOOD: + case CHIP_PALM: + case CHIP_SUMO: + case CHIP_SUMO2: + case CHIP_TURKS: + case CHIP_CAICOS: + force_no_swizzle = false; + break; + case CHIP_CYPRESS: + case CHIP_HEMLOCK: + case CHIP_JUNIPER: + case CHIP_BARTS: + default: + force_no_swizzle = true; + break; + } + if (force_no_swizzle) { + bool last_backend_enabled = false; + + force_no_swizzle = false; + for (i = 0; i < EVERGREEN_MAX_BACKENDS; ++i) { + if (((enabled_backends_mask >> i) & 1) == 1) { + if (last_backend_enabled) + force_no_swizzle = true; + last_backend_enabled = true; + } else + last_backend_enabled = false; + } + } + + switch (num_tile_pipes) { + case 1: + case 3: + case 5: + case 7: + DRM_ERROR("odd number of pipes!\n"); + break; + case 2: + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 1; + break; + case 4: + if (force_no_swizzle) { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 1; + swizzle_pipe[2] = 2; + swizzle_pipe[3] = 3; + } else { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 2; + swizzle_pipe[2] = 1; + swizzle_pipe[3] = 3; + } + break; + case 6: + if (force_no_swizzle) { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 1; + swizzle_pipe[2] = 2; + swizzle_pipe[3] = 3; + swizzle_pipe[4] = 4; + swizzle_pipe[5] = 5; + } else { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 2; + swizzle_pipe[2] = 4; + swizzle_pipe[3] = 1; + swizzle_pipe[4] = 3; + swizzle_pipe[5] = 5; + } + break; + case 8: + if (force_no_swizzle) { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 1; + swizzle_pipe[2] = 2; + swizzle_pipe[3] = 3; + swizzle_pipe[4] = 4; + swizzle_pipe[5] = 5; + swizzle_pipe[6] = 6; + swizzle_pipe[7] = 7; + } else { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 2; + swizzle_pipe[2] = 4; + swizzle_pipe[3] = 6; + swizzle_pipe[4] = 1; + swizzle_pipe[5] = 3; + swizzle_pipe[6] = 5; + swizzle_pipe[7] = 7; + } + break; + } + + for (cur_pipe = 0; cur_pipe < num_tile_pipes; ++cur_pipe) { + while (((1 << cur_backend) & enabled_backends_mask) == 0) + cur_backend = (cur_backend + 1) % EVERGREEN_MAX_BACKENDS; + + backend_map |= (((cur_backend & 0xf) << (swizzle_pipe[cur_pipe] * 4))); + + cur_backend = (cur_backend + 1) % EVERGREEN_MAX_BACKENDS; + } + + return backend_map; +} + +static void evergreen_gpu_init(struct radeon_device *rdev) +{ + u32 cc_rb_backend_disable = 0; + u32 cc_gc_shader_pipe_config; + u32 gb_addr_config = 0; + u32 mc_shared_chmap, mc_arb_ramcfg; + u32 gb_backend_map; + u32 grbm_gfx_index; + u32 sx_debug_1; + u32 smx_dc_ctl0; + u32 sq_config; + u32 sq_lds_resource_mgmt; + u32 sq_gpr_resource_mgmt_1; + u32 sq_gpr_resource_mgmt_2; + u32 sq_gpr_resource_mgmt_3; + u32 sq_thread_resource_mgmt; + u32 sq_thread_resource_mgmt_2; + u32 sq_stack_resource_mgmt_1; + u32 sq_stack_resource_mgmt_2; + u32 sq_stack_resource_mgmt_3; + u32 vgt_cache_invalidation; + u32 hdp_host_path_cntl, tmp; + int i, j, num_shader_engines, ps_thread_count; + + switch (rdev->family) { + case CHIP_CYPRESS: + case CHIP_HEMLOCK: + rdev->config.evergreen.num_ses = 2; + rdev->config.evergreen.max_pipes = 4; + rdev->config.evergreen.max_tile_pipes = 8; + rdev->config.evergreen.max_simds = 10; + rdev->config.evergreen.max_backends = 4 * rdev->config.evergreen.num_ses; + rdev->config.evergreen.max_gprs = 256; + rdev->config.evergreen.max_threads = 248; + rdev->config.evergreen.max_gs_threads = 32; + rdev->config.evergreen.max_stack_entries = 512; + rdev->config.evergreen.sx_num_of_sets = 4; + rdev->config.evergreen.sx_max_export_size = 256; + rdev->config.evergreen.sx_max_export_pos_size = 64; + rdev->config.evergreen.sx_max_export_smx_size = 192; + rdev->config.evergreen.max_hw_contexts = 8; + rdev->config.evergreen.sq_num_cf_insts = 2; + + rdev->config.evergreen.sc_prim_fifo_size = 0x100; + rdev->config.evergreen.sc_hiz_tile_fifo_size = 0x30; + rdev->config.evergreen.sc_earlyz_tile_fifo_size = 0x130; + break; + case CHIP_JUNIPER: + rdev->config.evergreen.num_ses = 1; + rdev->config.evergreen.max_pipes = 4; + rdev->config.evergreen.max_tile_pipes = 4; + rdev->config.evergreen.max_simds = 10; + rdev->config.evergreen.max_backends = 4 * rdev->config.evergreen.num_ses; + rdev->config.evergreen.max_gprs = 256; + rdev->config.evergreen.max_threads = 248; + rdev->config.evergreen.max_gs_threads = 32; + rdev->config.evergreen.max_stack_entries = 512; + rdev->config.evergreen.sx_num_of_sets = 4; + rdev->config.evergreen.sx_max_export_size = 256; + rdev->config.evergreen.sx_max_export_pos_size = 64; + rdev->config.evergreen.sx_max_export_smx_size = 192; + rdev->config.evergreen.max_hw_contexts = 8; + rdev->config.evergreen.sq_num_cf_insts = 2; + + rdev->config.evergreen.sc_prim_fifo_size = 0x100; + rdev->config.evergreen.sc_hiz_tile_fifo_size = 0x30; + rdev->config.evergreen.sc_earlyz_tile_fifo_size = 0x130; + break; + case CHIP_REDWOOD: + rdev->config.evergreen.num_ses = 1; + rdev->config.evergreen.max_pipes = 4; + rdev->config.evergreen.max_tile_pipes = 4; + rdev->config.evergreen.max_simds = 5; + rdev->config.evergreen.max_backends = 2 * rdev->config.evergreen.num_ses; + rdev->config.evergreen.max_gprs = 256; + rdev->config.evergreen.max_threads = 248; + rdev->config.evergreen.max_gs_threads = 32; + rdev->config.evergreen.max_stack_entries = 256; + rdev->config.evergreen.sx_num_of_sets = 4; + rdev->config.evergreen.sx_max_export_size = 256; + rdev->config.evergreen.sx_max_export_pos_size = 64; + rdev->config.evergreen.sx_max_export_smx_size = 192; + rdev->config.evergreen.max_hw_contexts = 8; + rdev->config.evergreen.sq_num_cf_insts = 2; + + rdev->config.evergreen.sc_prim_fifo_size = 0x100; + rdev->config.evergreen.sc_hiz_tile_fifo_size = 0x30; + rdev->config.evergreen.sc_earlyz_tile_fifo_size = 0x130; + break; + case CHIP_CEDAR: + default: + rdev->config.evergreen.num_ses = 1; + rdev->config.evergreen.max_pipes = 2; + rdev->config.evergreen.max_tile_pipes = 2; + rdev->config.evergreen.max_simds = 2; + rdev->config.evergreen.max_backends = 1 * rdev->config.evergreen.num_ses; + rdev->config.evergreen.max_gprs = 256; + rdev->config.evergreen.max_threads = 192; + rdev->config.evergreen.max_gs_threads = 16; + rdev->config.evergreen.max_stack_entries = 256; + rdev->config.evergreen.sx_num_of_sets = 4; + rdev->config.evergreen.sx_max_export_size = 128; + rdev->config.evergreen.sx_max_export_pos_size = 32; + rdev->config.evergreen.sx_max_export_smx_size = 96; + rdev->config.evergreen.max_hw_contexts = 4; + rdev->config.evergreen.sq_num_cf_insts = 1; + + rdev->config.evergreen.sc_prim_fifo_size = 0x40; + rdev->config.evergreen.sc_hiz_tile_fifo_size = 0x30; + rdev->config.evergreen.sc_earlyz_tile_fifo_size = 0x130; + break; + case CHIP_PALM: + rdev->config.evergreen.num_ses = 1; + rdev->config.evergreen.max_pipes = 2; + rdev->config.evergreen.max_tile_pipes = 2; + rdev->config.evergreen.max_simds = 2; + rdev->config.evergreen.max_backends = 1 * rdev->config.evergreen.num_ses; + rdev->config.evergreen.max_gprs = 256; + rdev->config.evergreen.max_threads = 192; + rdev->config.evergreen.max_gs_threads = 16; + rdev->config.evergreen.max_stack_entries = 256; + rdev->config.evergreen.sx_num_of_sets = 4; + rdev->config.evergreen.sx_max_export_size = 128; + rdev->config.evergreen.sx_max_export_pos_size = 32; + rdev->config.evergreen.sx_max_export_smx_size = 96; + rdev->config.evergreen.max_hw_contexts = 4; + rdev->config.evergreen.sq_num_cf_insts = 1; + + rdev->config.evergreen.sc_prim_fifo_size = 0x40; + rdev->config.evergreen.sc_hiz_tile_fifo_size = 0x30; + rdev->config.evergreen.sc_earlyz_tile_fifo_size = 0x130; + break; + case CHIP_SUMO: + rdev->config.evergreen.num_ses = 1; + rdev->config.evergreen.max_pipes = 4; + rdev->config.evergreen.max_tile_pipes = 2; + if (rdev->pdev->device == 0x9648) + rdev->config.evergreen.max_simds = 3; + else if ((rdev->pdev->device == 0x9647) || + (rdev->pdev->device == 0x964a)) + rdev->config.evergreen.max_simds = 4; + else + rdev->config.evergreen.max_simds = 5; + rdev->config.evergreen.max_backends = 2 * rdev->config.evergreen.num_ses; + rdev->config.evergreen.max_gprs = 256; + rdev->config.evergreen.max_threads = 248; + rdev->config.evergreen.max_gs_threads = 32; + rdev->config.evergreen.max_stack_entries = 256; + rdev->config.evergreen.sx_num_of_sets = 4; + rdev->config.evergreen.sx_max_export_size = 256; + rdev->config.evergreen.sx_max_export_pos_size = 64; + rdev->config.evergreen.sx_max_export_smx_size = 192; + rdev->config.evergreen.max_hw_contexts = 8; + rdev->config.evergreen.sq_num_cf_insts = 2; + + rdev->config.evergreen.sc_prim_fifo_size = 0x40; + rdev->config.evergreen.sc_hiz_tile_fifo_size = 0x30; + rdev->config.evergreen.sc_earlyz_tile_fifo_size = 0x130; + break; + case CHIP_SUMO2: + rdev->config.evergreen.num_ses = 1; + rdev->config.evergreen.max_pipes = 4; + rdev->config.evergreen.max_tile_pipes = 4; + rdev->config.evergreen.max_simds = 2; + rdev->config.evergreen.max_backends = 1 * rdev->config.evergreen.num_ses; + rdev->config.evergreen.max_gprs = 256; + rdev->config.evergreen.max_threads = 248; + rdev->config.evergreen.max_gs_threads = 32; + rdev->config.evergreen.max_stack_entries = 512; + rdev->config.evergreen.sx_num_of_sets = 4; + rdev->config.evergreen.sx_max_export_size = 256; + rdev->config.evergreen.sx_max_export_pos_size = 64; + rdev->config.evergreen.sx_max_export_smx_size = 192; + rdev->config.evergreen.max_hw_contexts = 8; + rdev->config.evergreen.sq_num_cf_insts = 2; + + rdev->config.evergreen.sc_prim_fifo_size = 0x40; + rdev->config.evergreen.sc_hiz_tile_fifo_size = 0x30; + rdev->config.evergreen.sc_earlyz_tile_fifo_size = 0x130; + break; + case CHIP_BARTS: + rdev->config.evergreen.num_ses = 2; + rdev->config.evergreen.max_pipes = 4; + rdev->config.evergreen.max_tile_pipes = 8; + rdev->config.evergreen.max_simds = 7; + rdev->config.evergreen.max_backends = 4 * rdev->config.evergreen.num_ses; + rdev->config.evergreen.max_gprs = 256; + rdev->config.evergreen.max_threads = 248; + rdev->config.evergreen.max_gs_threads = 32; + rdev->config.evergreen.max_stack_entries = 512; + rdev->config.evergreen.sx_num_of_sets = 4; + rdev->config.evergreen.sx_max_export_size = 256; + rdev->config.evergreen.sx_max_export_pos_size = 64; + rdev->config.evergreen.sx_max_export_smx_size = 192; + rdev->config.evergreen.max_hw_contexts = 8; + rdev->config.evergreen.sq_num_cf_insts = 2; + + rdev->config.evergreen.sc_prim_fifo_size = 0x100; + rdev->config.evergreen.sc_hiz_tile_fifo_size = 0x30; + rdev->config.evergreen.sc_earlyz_tile_fifo_size = 0x130; + break; + case CHIP_TURKS: + rdev->config.evergreen.num_ses = 1; + rdev->config.evergreen.max_pipes = 4; + rdev->config.evergreen.max_tile_pipes = 4; + rdev->config.evergreen.max_simds = 6; + rdev->config.evergreen.max_backends = 2 * rdev->config.evergreen.num_ses; + rdev->config.evergreen.max_gprs = 256; + rdev->config.evergreen.max_threads = 248; + rdev->config.evergreen.max_gs_threads = 32; + rdev->config.evergreen.max_stack_entries = 256; + rdev->config.evergreen.sx_num_of_sets = 4; + rdev->config.evergreen.sx_max_export_size = 256; + rdev->config.evergreen.sx_max_export_pos_size = 64; + rdev->config.evergreen.sx_max_export_smx_size = 192; + rdev->config.evergreen.max_hw_contexts = 8; + rdev->config.evergreen.sq_num_cf_insts = 2; + + rdev->config.evergreen.sc_prim_fifo_size = 0x100; + rdev->config.evergreen.sc_hiz_tile_fifo_size = 0x30; + rdev->config.evergreen.sc_earlyz_tile_fifo_size = 0x130; + break; + case CHIP_CAICOS: + rdev->config.evergreen.num_ses = 1; + rdev->config.evergreen.max_pipes = 4; + rdev->config.evergreen.max_tile_pipes = 2; + rdev->config.evergreen.max_simds = 2; + rdev->config.evergreen.max_backends = 1 * rdev->config.evergreen.num_ses; + rdev->config.evergreen.max_gprs = 256; + rdev->config.evergreen.max_threads = 192; + rdev->config.evergreen.max_gs_threads = 16; + rdev->config.evergreen.max_stack_entries = 256; + rdev->config.evergreen.sx_num_of_sets = 4; + rdev->config.evergreen.sx_max_export_size = 128; + rdev->config.evergreen.sx_max_export_pos_size = 32; + rdev->config.evergreen.sx_max_export_smx_size = 96; + rdev->config.evergreen.max_hw_contexts = 4; + rdev->config.evergreen.sq_num_cf_insts = 1; + + rdev->config.evergreen.sc_prim_fifo_size = 0x40; + rdev->config.evergreen.sc_hiz_tile_fifo_size = 0x30; + rdev->config.evergreen.sc_earlyz_tile_fifo_size = 0x130; + break; + } + + /* Initialize HDP */ + for (i = 0, j = 0; i < 32; i++, j += 0x18) { + WREG32((0x2c14 + j), 0x00000000); + WREG32((0x2c18 + j), 0x00000000); + WREG32((0x2c1c + j), 0x00000000); + WREG32((0x2c20 + j), 0x00000000); + WREG32((0x2c24 + j), 0x00000000); + } + + WREG32(GRBM_CNTL, GRBM_READ_TIMEOUT(0xff)); + + evergreen_fix_pci_max_read_req_size(rdev); + + cc_gc_shader_pipe_config = RREG32(CC_GC_SHADER_PIPE_CONFIG) & ~2; + + cc_gc_shader_pipe_config |= + INACTIVE_QD_PIPES((EVERGREEN_MAX_PIPES_MASK << rdev->config.evergreen.max_pipes) + & EVERGREEN_MAX_PIPES_MASK); + cc_gc_shader_pipe_config |= + INACTIVE_SIMDS((EVERGREEN_MAX_SIMDS_MASK << rdev->config.evergreen.max_simds) + & EVERGREEN_MAX_SIMDS_MASK); + + cc_rb_backend_disable = + BACKEND_DISABLE((EVERGREEN_MAX_BACKENDS_MASK << rdev->config.evergreen.max_backends) + & EVERGREEN_MAX_BACKENDS_MASK); + + + mc_shared_chmap = RREG32(MC_SHARED_CHMAP); + if (rdev->flags & RADEON_IS_IGP) + mc_arb_ramcfg = RREG32(FUS_MC_ARB_RAMCFG); + else + mc_arb_ramcfg = RREG32(MC_ARB_RAMCFG); + + switch (rdev->config.evergreen.max_tile_pipes) { + case 1: + default: + gb_addr_config |= NUM_PIPES(0); + break; + case 2: + gb_addr_config |= NUM_PIPES(1); + break; + case 4: + gb_addr_config |= NUM_PIPES(2); + break; + case 8: + gb_addr_config |= NUM_PIPES(3); + break; + } + + gb_addr_config |= PIPE_INTERLEAVE_SIZE((mc_arb_ramcfg & BURSTLENGTH_MASK) >> BURSTLENGTH_SHIFT); + gb_addr_config |= BANK_INTERLEAVE_SIZE(0); + gb_addr_config |= NUM_SHADER_ENGINES(rdev->config.evergreen.num_ses - 1); + gb_addr_config |= SHADER_ENGINE_TILE_SIZE(1); + gb_addr_config |= NUM_GPUS(0); /* Hemlock? */ + gb_addr_config |= MULTI_GPU_TILE_SIZE(2); + + if (((mc_arb_ramcfg & NOOFCOLS_MASK) >> NOOFCOLS_SHIFT) > 2) + gb_addr_config |= ROW_SIZE(2); + else + gb_addr_config |= ROW_SIZE((mc_arb_ramcfg & NOOFCOLS_MASK) >> NOOFCOLS_SHIFT); + + if (rdev->ddev->pdev->device == 0x689e) { + u32 efuse_straps_4; + u32 efuse_straps_3; + u8 efuse_box_bit_131_124; + + WREG32(RCU_IND_INDEX, 0x204); + efuse_straps_4 = RREG32(RCU_IND_DATA); + WREG32(RCU_IND_INDEX, 0x203); + efuse_straps_3 = RREG32(RCU_IND_DATA); + efuse_box_bit_131_124 = (u8)(((efuse_straps_4 & 0xf) << 4) | ((efuse_straps_3 & 0xf0000000) >> 28)); + + switch(efuse_box_bit_131_124) { + case 0x00: + gb_backend_map = 0x76543210; + break; + case 0x55: + gb_backend_map = 0x77553311; + break; + case 0x56: + gb_backend_map = 0x77553300; + break; + case 0x59: + gb_backend_map = 0x77552211; + break; + case 0x66: + gb_backend_map = 0x77443300; + break; + case 0x99: + gb_backend_map = 0x66552211; + break; + case 0x5a: + gb_backend_map = 0x77552200; + break; + case 0xaa: + gb_backend_map = 0x66442200; + break; + case 0x95: + gb_backend_map = 0x66553311; + break; + default: + DRM_ERROR("bad backend map, using default\n"); + gb_backend_map = + evergreen_get_tile_pipe_to_backend_map(rdev, + rdev->config.evergreen.max_tile_pipes, + rdev->config.evergreen.max_backends, + ((EVERGREEN_MAX_BACKENDS_MASK << + rdev->config.evergreen.max_backends) & + EVERGREEN_MAX_BACKENDS_MASK)); + break; + } + } else if (rdev->ddev->pdev->device == 0x68b9) { + u32 efuse_straps_3; + u8 efuse_box_bit_127_124; + + WREG32(RCU_IND_INDEX, 0x203); + efuse_straps_3 = RREG32(RCU_IND_DATA); + efuse_box_bit_127_124 = (u8)((efuse_straps_3 & 0xF0000000) >> 28); + + switch(efuse_box_bit_127_124) { + case 0x0: + gb_backend_map = 0x00003210; + break; + case 0x5: + case 0x6: + case 0x9: + case 0xa: + gb_backend_map = 0x00003311; + break; + default: + DRM_ERROR("bad backend map, using default\n"); + gb_backend_map = + evergreen_get_tile_pipe_to_backend_map(rdev, + rdev->config.evergreen.max_tile_pipes, + rdev->config.evergreen.max_backends, + ((EVERGREEN_MAX_BACKENDS_MASK << + rdev->config.evergreen.max_backends) & + EVERGREEN_MAX_BACKENDS_MASK)); + break; + } + } else { + switch (rdev->family) { + case CHIP_CYPRESS: + case CHIP_HEMLOCK: + case CHIP_BARTS: + gb_backend_map = 0x66442200; + break; + case CHIP_JUNIPER: + gb_backend_map = 0x00002200; + break; + default: + gb_backend_map = + evergreen_get_tile_pipe_to_backend_map(rdev, + rdev->config.evergreen.max_tile_pipes, + rdev->config.evergreen.max_backends, + ((EVERGREEN_MAX_BACKENDS_MASK << + rdev->config.evergreen.max_backends) & + EVERGREEN_MAX_BACKENDS_MASK)); + } + } + + /* setup tiling info dword. gb_addr_config is not adequate since it does + * not have bank info, so create a custom tiling dword. + * bits 3:0 num_pipes + * bits 7:4 num_banks + * bits 11:8 group_size + * bits 15:12 row_size + */ + rdev->config.evergreen.tile_config = 0; + switch (rdev->config.evergreen.max_tile_pipes) { + case 1: + default: + rdev->config.evergreen.tile_config |= (0 << 0); + break; + case 2: + rdev->config.evergreen.tile_config |= (1 << 0); + break; + case 4: + rdev->config.evergreen.tile_config |= (2 << 0); + break; + case 8: + rdev->config.evergreen.tile_config |= (3 << 0); + break; + } + /* num banks is 8 on all fusion asics. 0 = 4, 1 = 8, 2 = 16 */ + if (rdev->flags & RADEON_IS_IGP) + rdev->config.evergreen.tile_config |= 1 << 4; + else + rdev->config.evergreen.tile_config |= + ((mc_arb_ramcfg & NOOFBANK_MASK) >> NOOFBANK_SHIFT) << 4; + rdev->config.evergreen.tile_config |= + ((mc_arb_ramcfg & BURSTLENGTH_MASK) >> BURSTLENGTH_SHIFT) << 8; + rdev->config.evergreen.tile_config |= + ((gb_addr_config & 0x30000000) >> 28) << 12; + + WREG32(GB_BACKEND_MAP, gb_backend_map); + WREG32(GB_ADDR_CONFIG, gb_addr_config); + WREG32(DMIF_ADDR_CONFIG, gb_addr_config); + WREG32(HDP_ADDR_CONFIG, gb_addr_config); + + num_shader_engines = ((RREG32(GB_ADDR_CONFIG) & NUM_SHADER_ENGINES(3)) >> 12) + 1; + grbm_gfx_index = INSTANCE_BROADCAST_WRITES; + + for (i = 0; i < rdev->config.evergreen.num_ses; i++) { + u32 rb = cc_rb_backend_disable | (0xf0 << 16); + u32 sp = cc_gc_shader_pipe_config; + u32 gfx = grbm_gfx_index | SE_INDEX(i); + + if (i == num_shader_engines) { + rb |= BACKEND_DISABLE(EVERGREEN_MAX_BACKENDS_MASK); + sp |= INACTIVE_SIMDS(EVERGREEN_MAX_SIMDS_MASK); + } + + WREG32(GRBM_GFX_INDEX, gfx); + WREG32(RLC_GFX_INDEX, gfx); + + WREG32(CC_RB_BACKEND_DISABLE, rb); + WREG32(CC_SYS_RB_BACKEND_DISABLE, rb); + WREG32(GC_USER_RB_BACKEND_DISABLE, rb); + WREG32(CC_GC_SHADER_PIPE_CONFIG, sp); + } + + grbm_gfx_index = INSTANCE_BROADCAST_WRITES | SE_BROADCAST_WRITES; + WREG32(GRBM_GFX_INDEX, grbm_gfx_index); + WREG32(RLC_GFX_INDEX, grbm_gfx_index); + + WREG32(CGTS_SYS_TCC_DISABLE, 0); + WREG32(CGTS_TCC_DISABLE, 0); + WREG32(CGTS_USER_SYS_TCC_DISABLE, 0); + WREG32(CGTS_USER_TCC_DISABLE, 0); + + /* set HW defaults for 3D engine */ + WREG32(CP_QUEUE_THRESHOLDS, (ROQ_IB1_START(0x16) | + ROQ_IB2_START(0x2b))); + + WREG32(CP_MEQ_THRESHOLDS, STQ_SPLIT(0x30)); + + WREG32(TA_CNTL_AUX, (DISABLE_CUBE_ANISO | + SYNC_GRADIENT | + SYNC_WALKER | + SYNC_ALIGNER)); + + sx_debug_1 = RREG32(SX_DEBUG_1); + sx_debug_1 |= ENABLE_NEW_SMX_ADDRESS; + WREG32(SX_DEBUG_1, sx_debug_1); + + + smx_dc_ctl0 = RREG32(SMX_DC_CTL0); + smx_dc_ctl0 &= ~NUMBER_OF_SETS(0x1ff); + smx_dc_ctl0 |= NUMBER_OF_SETS(rdev->config.evergreen.sx_num_of_sets); + WREG32(SMX_DC_CTL0, smx_dc_ctl0); + + WREG32(SX_EXPORT_BUFFER_SIZES, (COLOR_BUFFER_SIZE((rdev->config.evergreen.sx_max_export_size / 4) - 1) | + POSITION_BUFFER_SIZE((rdev->config.evergreen.sx_max_export_pos_size / 4) - 1) | + SMX_BUFFER_SIZE((rdev->config.evergreen.sx_max_export_smx_size / 4) - 1))); + + WREG32(PA_SC_FIFO_SIZE, (SC_PRIM_FIFO_SIZE(rdev->config.evergreen.sc_prim_fifo_size) | + SC_HIZ_TILE_FIFO_SIZE(rdev->config.evergreen.sc_hiz_tile_fifo_size) | + SC_EARLYZ_TILE_FIFO_SIZE(rdev->config.evergreen.sc_earlyz_tile_fifo_size))); + + WREG32(VGT_NUM_INSTANCES, 1); + WREG32(SPI_CONFIG_CNTL, 0); + WREG32(SPI_CONFIG_CNTL_1, VTX_DONE_DELAY(4)); + WREG32(CP_PERFMON_CNTL, 0); + + WREG32(SQ_MS_FIFO_SIZES, (CACHE_FIFO_SIZE(16 * rdev->config.evergreen.sq_num_cf_insts) | + FETCH_FIFO_HIWATER(0x4) | + DONE_FIFO_HIWATER(0xe0) | + ALU_UPDATE_FIFO_HIWATER(0x8))); + + sq_config = RREG32(SQ_CONFIG); + sq_config &= ~(PS_PRIO(3) | + VS_PRIO(3) | + GS_PRIO(3) | + ES_PRIO(3)); + sq_config |= (VC_ENABLE | + EXPORT_SRC_C | + PS_PRIO(0) | + VS_PRIO(1) | + GS_PRIO(2) | + ES_PRIO(3)); + + switch (rdev->family) { + case CHIP_CEDAR: + case CHIP_PALM: + case CHIP_SUMO: + case CHIP_SUMO2: + case CHIP_CAICOS: + /* no vertex cache */ + sq_config &= ~VC_ENABLE; + break; + default: + break; + } + + sq_lds_resource_mgmt = RREG32(SQ_LDS_RESOURCE_MGMT); + + sq_gpr_resource_mgmt_1 = NUM_PS_GPRS((rdev->config.evergreen.max_gprs - (4 * 2))* 12 / 32); + sq_gpr_resource_mgmt_1 |= NUM_VS_GPRS((rdev->config.evergreen.max_gprs - (4 * 2)) * 6 / 32); + sq_gpr_resource_mgmt_1 |= NUM_CLAUSE_TEMP_GPRS(4); + sq_gpr_resource_mgmt_2 = NUM_GS_GPRS((rdev->config.evergreen.max_gprs - (4 * 2)) * 4 / 32); + sq_gpr_resource_mgmt_2 |= NUM_ES_GPRS((rdev->config.evergreen.max_gprs - (4 * 2)) * 4 / 32); + sq_gpr_resource_mgmt_3 = NUM_HS_GPRS((rdev->config.evergreen.max_gprs - (4 * 2)) * 3 / 32); + sq_gpr_resource_mgmt_3 |= NUM_LS_GPRS((rdev->config.evergreen.max_gprs - (4 * 2)) * 3 / 32); + + switch (rdev->family) { + case CHIP_CEDAR: + case CHIP_PALM: + case CHIP_SUMO: + case CHIP_SUMO2: + ps_thread_count = 96; + break; + default: + ps_thread_count = 128; + break; + } + + sq_thread_resource_mgmt = NUM_PS_THREADS(ps_thread_count); + sq_thread_resource_mgmt |= NUM_VS_THREADS((((rdev->config.evergreen.max_threads - ps_thread_count) / 6) / 8) * 8); + sq_thread_resource_mgmt |= NUM_GS_THREADS((((rdev->config.evergreen.max_threads - ps_thread_count) / 6) / 8) * 8); + sq_thread_resource_mgmt |= NUM_ES_THREADS((((rdev->config.evergreen.max_threads - ps_thread_count) / 6) / 8) * 8); + sq_thread_resource_mgmt_2 = NUM_HS_THREADS((((rdev->config.evergreen.max_threads - ps_thread_count) / 6) / 8) * 8); + sq_thread_resource_mgmt_2 |= NUM_LS_THREADS((((rdev->config.evergreen.max_threads - ps_thread_count) / 6) / 8) * 8); + + sq_stack_resource_mgmt_1 = NUM_PS_STACK_ENTRIES((rdev->config.evergreen.max_stack_entries * 1) / 6); + sq_stack_resource_mgmt_1 |= NUM_VS_STACK_ENTRIES((rdev->config.evergreen.max_stack_entries * 1) / 6); + sq_stack_resource_mgmt_2 = NUM_GS_STACK_ENTRIES((rdev->config.evergreen.max_stack_entries * 1) / 6); + sq_stack_resource_mgmt_2 |= NUM_ES_STACK_ENTRIES((rdev->config.evergreen.max_stack_entries * 1) / 6); + sq_stack_resource_mgmt_3 = NUM_HS_STACK_ENTRIES((rdev->config.evergreen.max_stack_entries * 1) / 6); + sq_stack_resource_mgmt_3 |= NUM_LS_STACK_ENTRIES((rdev->config.evergreen.max_stack_entries * 1) / 6); + + WREG32(SQ_CONFIG, sq_config); + WREG32(SQ_GPR_RESOURCE_MGMT_1, sq_gpr_resource_mgmt_1); + WREG32(SQ_GPR_RESOURCE_MGMT_2, sq_gpr_resource_mgmt_2); + WREG32(SQ_GPR_RESOURCE_MGMT_3, sq_gpr_resource_mgmt_3); + WREG32(SQ_THREAD_RESOURCE_MGMT, sq_thread_resource_mgmt); + WREG32(SQ_THREAD_RESOURCE_MGMT_2, sq_thread_resource_mgmt_2); + WREG32(SQ_STACK_RESOURCE_MGMT_1, sq_stack_resource_mgmt_1); + WREG32(SQ_STACK_RESOURCE_MGMT_2, sq_stack_resource_mgmt_2); + WREG32(SQ_STACK_RESOURCE_MGMT_3, sq_stack_resource_mgmt_3); + WREG32(SQ_DYN_GPR_CNTL_PS_FLUSH_REQ, 0); + WREG32(SQ_LDS_RESOURCE_MGMT, sq_lds_resource_mgmt); + + WREG32(PA_SC_FORCE_EOV_MAX_CNTS, (FORCE_EOV_MAX_CLK_CNT(4095) | + FORCE_EOV_MAX_REZ_CNT(255))); + + switch (rdev->family) { + case CHIP_CEDAR: + case CHIP_PALM: + case CHIP_SUMO: + case CHIP_SUMO2: + case CHIP_CAICOS: + vgt_cache_invalidation = CACHE_INVALIDATION(TC_ONLY); + break; + default: + vgt_cache_invalidation = CACHE_INVALIDATION(VC_AND_TC); + break; + } + vgt_cache_invalidation |= AUTO_INVLD_EN(ES_AND_GS_AUTO); + WREG32(VGT_CACHE_INVALIDATION, vgt_cache_invalidation); + + WREG32(VGT_GS_VERTEX_REUSE, 16); + WREG32(PA_SU_LINE_STIPPLE_VALUE, 0); + WREG32(PA_SC_LINE_STIPPLE_STATE, 0); + + WREG32(VGT_VERTEX_REUSE_BLOCK_CNTL, 14); + WREG32(VGT_OUT_DEALLOC_CNTL, 16); + + WREG32(CB_PERF_CTR0_SEL_0, 0); + WREG32(CB_PERF_CTR0_SEL_1, 0); + WREG32(CB_PERF_CTR1_SEL_0, 0); + WREG32(CB_PERF_CTR1_SEL_1, 0); + WREG32(CB_PERF_CTR2_SEL_0, 0); + WREG32(CB_PERF_CTR2_SEL_1, 0); + WREG32(CB_PERF_CTR3_SEL_0, 0); + WREG32(CB_PERF_CTR3_SEL_1, 0); + + /* clear render buffer base addresses */ + WREG32(CB_COLOR0_BASE, 0); + WREG32(CB_COLOR1_BASE, 0); + WREG32(CB_COLOR2_BASE, 0); + WREG32(CB_COLOR3_BASE, 0); + WREG32(CB_COLOR4_BASE, 0); + WREG32(CB_COLOR5_BASE, 0); + WREG32(CB_COLOR6_BASE, 0); + WREG32(CB_COLOR7_BASE, 0); + WREG32(CB_COLOR8_BASE, 0); + WREG32(CB_COLOR9_BASE, 0); + WREG32(CB_COLOR10_BASE, 0); + WREG32(CB_COLOR11_BASE, 0); + + /* set the shader const cache sizes to 0 */ + for (i = SQ_ALU_CONST_BUFFER_SIZE_PS_0; i < 0x28200; i += 4) + WREG32(i, 0); + for (i = SQ_ALU_CONST_BUFFER_SIZE_HS_0; i < 0x29000; i += 4) + WREG32(i, 0); + + tmp = RREG32(HDP_MISC_CNTL); + tmp |= HDP_FLUSH_INVALIDATE_CACHE; + WREG32(HDP_MISC_CNTL, tmp); + + hdp_host_path_cntl = RREG32(HDP_HOST_PATH_CNTL); + WREG32(HDP_HOST_PATH_CNTL, hdp_host_path_cntl); + + WREG32(PA_CL_ENHANCE, CLIP_VTX_REORDER_ENA | NUM_CLIP_SEQ(3)); + + udelay(50); + +} + +int evergreen_mc_init(struct radeon_device *rdev) +{ + u32 tmp; + int chansize, numchan; + + /* Get VRAM informations */ + rdev->mc.vram_is_ddr = true; + if (rdev->flags & RADEON_IS_IGP) + tmp = RREG32(FUS_MC_ARB_RAMCFG); + else + tmp = RREG32(MC_ARB_RAMCFG); + if (tmp & CHANSIZE_OVERRIDE) { + chansize = 16; + } else if (tmp & CHANSIZE_MASK) { + chansize = 64; + } else { + chansize = 32; + } + tmp = RREG32(MC_SHARED_CHMAP); + switch ((tmp & NOOFCHAN_MASK) >> NOOFCHAN_SHIFT) { + case 0: + default: + numchan = 1; + break; + case 1: + numchan = 2; + break; + case 2: + numchan = 4; + break; + case 3: + numchan = 8; + break; + } + rdev->mc.vram_width = numchan * chansize; + /* Could aper size report 0 ? */ + rdev->mc.aper_base = pci_resource_start(rdev->pdev, 0); + rdev->mc.aper_size = pci_resource_len(rdev->pdev, 0); + /* Setup GPU memory space */ + if (rdev->flags & RADEON_IS_IGP) { + /* size in bytes on fusion */ + rdev->mc.mc_vram_size = RREG32(CONFIG_MEMSIZE); + rdev->mc.real_vram_size = RREG32(CONFIG_MEMSIZE); + } else { + /* size in MB on evergreen */ + rdev->mc.mc_vram_size = RREG32(CONFIG_MEMSIZE) * 1024 * 1024; + rdev->mc.real_vram_size = RREG32(CONFIG_MEMSIZE) * 1024 * 1024; + } + rdev->mc.visible_vram_size = rdev->mc.aper_size; + r700_vram_gtt_location(rdev, &rdev->mc); + radeon_update_bandwidth_info(rdev); + + return 0; +} + +bool evergreen_gpu_is_lockup(struct radeon_device *rdev) +{ + u32 srbm_status; + u32 grbm_status; + u32 grbm_status_se0, grbm_status_se1; + struct r100_gpu_lockup *lockup = &rdev->config.evergreen.lockup; + int r; + + srbm_status = RREG32(SRBM_STATUS); + grbm_status = RREG32(GRBM_STATUS); + grbm_status_se0 = RREG32(GRBM_STATUS_SE0); + grbm_status_se1 = RREG32(GRBM_STATUS_SE1); + if (!(grbm_status & GUI_ACTIVE)) { + r100_gpu_lockup_update(lockup, &rdev->cp); + return false; + } + /* force CP activities */ + r = radeon_ring_lock(rdev, 2); + if (!r) { + /* PACKET2 NOP */ + radeon_ring_write(rdev, 0x80000000); + radeon_ring_write(rdev, 0x80000000); + radeon_ring_unlock_commit(rdev); + } + rdev->cp.rptr = RREG32(CP_RB_RPTR); + return r100_gpu_cp_is_lockup(rdev, lockup, &rdev->cp); +} + +static int evergreen_gpu_soft_reset(struct radeon_device *rdev) +{ + struct evergreen_mc_save save; + u32 grbm_reset = 0; + + if (!(RREG32(GRBM_STATUS) & GUI_ACTIVE)) + return 0; + + dev_info(rdev->dev, "GPU softreset \n"); + dev_info(rdev->dev, " GRBM_STATUS=0x%08X\n", + RREG32(GRBM_STATUS)); + dev_info(rdev->dev, " GRBM_STATUS_SE0=0x%08X\n", + RREG32(GRBM_STATUS_SE0)); + dev_info(rdev->dev, " GRBM_STATUS_SE1=0x%08X\n", + RREG32(GRBM_STATUS_SE1)); + dev_info(rdev->dev, " SRBM_STATUS=0x%08X\n", + RREG32(SRBM_STATUS)); + evergreen_mc_stop(rdev, &save); + if (evergreen_mc_wait_for_idle(rdev)) { + dev_warn(rdev->dev, "Wait for MC idle timedout !\n"); + } + /* Disable CP parsing/prefetching */ + WREG32(CP_ME_CNTL, CP_ME_HALT | CP_PFP_HALT); + + /* reset all the gfx blocks */ + grbm_reset = (SOFT_RESET_CP | + SOFT_RESET_CB | + SOFT_RESET_DB | + SOFT_RESET_PA | + SOFT_RESET_SC | + SOFT_RESET_SPI | + SOFT_RESET_SH | + SOFT_RESET_SX | + SOFT_RESET_TC | + SOFT_RESET_TA | + SOFT_RESET_VC | + SOFT_RESET_VGT); + + dev_info(rdev->dev, " GRBM_SOFT_RESET=0x%08X\n", grbm_reset); + WREG32(GRBM_SOFT_RESET, grbm_reset); + (void)RREG32(GRBM_SOFT_RESET); + udelay(50); + WREG32(GRBM_SOFT_RESET, 0); + (void)RREG32(GRBM_SOFT_RESET); + /* Wait a little for things to settle down */ + udelay(50); + dev_info(rdev->dev, " GRBM_STATUS=0x%08X\n", + RREG32(GRBM_STATUS)); + dev_info(rdev->dev, " GRBM_STATUS_SE0=0x%08X\n", + RREG32(GRBM_STATUS_SE0)); + dev_info(rdev->dev, " GRBM_STATUS_SE1=0x%08X\n", + RREG32(GRBM_STATUS_SE1)); + dev_info(rdev->dev, " SRBM_STATUS=0x%08X\n", + RREG32(SRBM_STATUS)); + evergreen_mc_resume(rdev, &save); + return 0; +} + +int evergreen_asic_reset(struct radeon_device *rdev) +{ + return evergreen_gpu_soft_reset(rdev); +} + +/* Interrupts */ + +u32 evergreen_get_vblank_counter(struct radeon_device *rdev, int crtc) +{ + switch (crtc) { + case 0: + return RREG32(CRTC_STATUS_FRAME_COUNT + EVERGREEN_CRTC0_REGISTER_OFFSET); + case 1: + return RREG32(CRTC_STATUS_FRAME_COUNT + EVERGREEN_CRTC1_REGISTER_OFFSET); + case 2: + return RREG32(CRTC_STATUS_FRAME_COUNT + EVERGREEN_CRTC2_REGISTER_OFFSET); + case 3: + return RREG32(CRTC_STATUS_FRAME_COUNT + EVERGREEN_CRTC3_REGISTER_OFFSET); + case 4: + return RREG32(CRTC_STATUS_FRAME_COUNT + EVERGREEN_CRTC4_REGISTER_OFFSET); + case 5: + return RREG32(CRTC_STATUS_FRAME_COUNT + EVERGREEN_CRTC5_REGISTER_OFFSET); + default: + return 0; + } +} + +void evergreen_disable_interrupt_state(struct radeon_device *rdev) +{ + u32 tmp; + + WREG32(CP_INT_CNTL, CNTX_BUSY_INT_ENABLE | CNTX_EMPTY_INT_ENABLE); + WREG32(GRBM_INT_CNTL, 0); + WREG32(INT_MASK + EVERGREEN_CRTC0_REGISTER_OFFSET, 0); + WREG32(INT_MASK + EVERGREEN_CRTC1_REGISTER_OFFSET, 0); + if (rdev->num_crtc >= 4) { + WREG32(INT_MASK + EVERGREEN_CRTC2_REGISTER_OFFSET, 0); + WREG32(INT_MASK + EVERGREEN_CRTC3_REGISTER_OFFSET, 0); + } + if (rdev->num_crtc >= 6) { + WREG32(INT_MASK + EVERGREEN_CRTC4_REGISTER_OFFSET, 0); + WREG32(INT_MASK + EVERGREEN_CRTC5_REGISTER_OFFSET, 0); + } + + WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC0_REGISTER_OFFSET, 0); + WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC1_REGISTER_OFFSET, 0); + if (rdev->num_crtc >= 4) { + WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC2_REGISTER_OFFSET, 0); + WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC3_REGISTER_OFFSET, 0); + } + if (rdev->num_crtc >= 6) { + WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC4_REGISTER_OFFSET, 0); + WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC5_REGISTER_OFFSET, 0); + } + + WREG32(DACA_AUTODETECT_INT_CONTROL, 0); + WREG32(DACB_AUTODETECT_INT_CONTROL, 0); + + tmp = RREG32(DC_HPD1_INT_CONTROL) & DC_HPDx_INT_POLARITY; + WREG32(DC_HPD1_INT_CONTROL, tmp); + tmp = RREG32(DC_HPD2_INT_CONTROL) & DC_HPDx_INT_POLARITY; + WREG32(DC_HPD2_INT_CONTROL, tmp); + tmp = RREG32(DC_HPD3_INT_CONTROL) & DC_HPDx_INT_POLARITY; + WREG32(DC_HPD3_INT_CONTROL, tmp); + tmp = RREG32(DC_HPD4_INT_CONTROL) & DC_HPDx_INT_POLARITY; + WREG32(DC_HPD4_INT_CONTROL, tmp); + tmp = RREG32(DC_HPD5_INT_CONTROL) & DC_HPDx_INT_POLARITY; + WREG32(DC_HPD5_INT_CONTROL, tmp); + tmp = RREG32(DC_HPD6_INT_CONTROL) & DC_HPDx_INT_POLARITY; + WREG32(DC_HPD6_INT_CONTROL, tmp); + +} + +int evergreen_irq_set(struct radeon_device *rdev) +{ + u32 cp_int_cntl = CNTX_BUSY_INT_ENABLE | CNTX_EMPTY_INT_ENABLE; + u32 crtc1 = 0, crtc2 = 0, crtc3 = 0, crtc4 = 0, crtc5 = 0, crtc6 = 0; + u32 hpd1, hpd2, hpd3, hpd4, hpd5, hpd6; + u32 grbm_int_cntl = 0; + u32 grph1 = 0, grph2 = 0, grph3 = 0, grph4 = 0, grph5 = 0, grph6 = 0; + + if (!rdev->irq.installed) { + WARN(1, "Can't enable IRQ/MSI because no handler is installed\n"); + return -EINVAL; + } + /* don't enable anything if the ih is disabled */ + if (!rdev->ih.enabled) { + r600_disable_interrupts(rdev); + /* force the active interrupt state to all disabled */ + evergreen_disable_interrupt_state(rdev); + return 0; + } + + hpd1 = RREG32(DC_HPD1_INT_CONTROL) & ~DC_HPDx_INT_EN; + hpd2 = RREG32(DC_HPD2_INT_CONTROL) & ~DC_HPDx_INT_EN; + hpd3 = RREG32(DC_HPD3_INT_CONTROL) & ~DC_HPDx_INT_EN; + hpd4 = RREG32(DC_HPD4_INT_CONTROL) & ~DC_HPDx_INT_EN; + hpd5 = RREG32(DC_HPD5_INT_CONTROL) & ~DC_HPDx_INT_EN; + hpd6 = RREG32(DC_HPD6_INT_CONTROL) & ~DC_HPDx_INT_EN; + + if (rdev->irq.sw_int) { + DRM_DEBUG("evergreen_irq_set: sw int\n"); + cp_int_cntl |= RB_INT_ENABLE; + cp_int_cntl |= TIME_STAMP_INT_ENABLE; + } + if (rdev->irq.crtc_vblank_int[0] || + rdev->irq.pflip[0]) { + DRM_DEBUG("evergreen_irq_set: vblank 0\n"); + crtc1 |= VBLANK_INT_MASK; + } + if (rdev->irq.crtc_vblank_int[1] || + rdev->irq.pflip[1]) { + DRM_DEBUG("evergreen_irq_set: vblank 1\n"); + crtc2 |= VBLANK_INT_MASK; + } + if (rdev->irq.crtc_vblank_int[2] || + rdev->irq.pflip[2]) { + DRM_DEBUG("evergreen_irq_set: vblank 2\n"); + crtc3 |= VBLANK_INT_MASK; + } + if (rdev->irq.crtc_vblank_int[3] || + rdev->irq.pflip[3]) { + DRM_DEBUG("evergreen_irq_set: vblank 3\n"); + crtc4 |= VBLANK_INT_MASK; + } + if (rdev->irq.crtc_vblank_int[4] || + rdev->irq.pflip[4]) { + DRM_DEBUG("evergreen_irq_set: vblank 4\n"); + crtc5 |= VBLANK_INT_MASK; + } + if (rdev->irq.crtc_vblank_int[5] || + rdev->irq.pflip[5]) { + DRM_DEBUG("evergreen_irq_set: vblank 5\n"); + crtc6 |= VBLANK_INT_MASK; + } + if (rdev->irq.hpd[0]) { + DRM_DEBUG("evergreen_irq_set: hpd 1\n"); + hpd1 |= DC_HPDx_INT_EN; + } + if (rdev->irq.hpd[1]) { + DRM_DEBUG("evergreen_irq_set: hpd 2\n"); + hpd2 |= DC_HPDx_INT_EN; + } + if (rdev->irq.hpd[2]) { + DRM_DEBUG("evergreen_irq_set: hpd 3\n"); + hpd3 |= DC_HPDx_INT_EN; + } + if (rdev->irq.hpd[3]) { + DRM_DEBUG("evergreen_irq_set: hpd 4\n"); + hpd4 |= DC_HPDx_INT_EN; + } + if (rdev->irq.hpd[4]) { + DRM_DEBUG("evergreen_irq_set: hpd 5\n"); + hpd5 |= DC_HPDx_INT_EN; + } + if (rdev->irq.hpd[5]) { + DRM_DEBUG("evergreen_irq_set: hpd 6\n"); + hpd6 |= DC_HPDx_INT_EN; + } + if (rdev->irq.gui_idle) { + DRM_DEBUG("gui idle\n"); + grbm_int_cntl |= GUI_IDLE_INT_ENABLE; + } + + WREG32(CP_INT_CNTL, cp_int_cntl); + WREG32(GRBM_INT_CNTL, grbm_int_cntl); + + WREG32(INT_MASK + EVERGREEN_CRTC0_REGISTER_OFFSET, crtc1); + WREG32(INT_MASK + EVERGREEN_CRTC1_REGISTER_OFFSET, crtc2); + if (rdev->num_crtc >= 4) { + WREG32(INT_MASK + EVERGREEN_CRTC2_REGISTER_OFFSET, crtc3); + WREG32(INT_MASK + EVERGREEN_CRTC3_REGISTER_OFFSET, crtc4); + } + if (rdev->num_crtc >= 6) { + WREG32(INT_MASK + EVERGREEN_CRTC4_REGISTER_OFFSET, crtc5); + WREG32(INT_MASK + EVERGREEN_CRTC5_REGISTER_OFFSET, crtc6); + } + + WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC0_REGISTER_OFFSET, grph1); + WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC1_REGISTER_OFFSET, grph2); + if (rdev->num_crtc >= 4) { + WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC2_REGISTER_OFFSET, grph3); + WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC3_REGISTER_OFFSET, grph4); + } + if (rdev->num_crtc >= 6) { + WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC4_REGISTER_OFFSET, grph5); + WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC5_REGISTER_OFFSET, grph6); + } + + WREG32(DC_HPD1_INT_CONTROL, hpd1); + WREG32(DC_HPD2_INT_CONTROL, hpd2); + WREG32(DC_HPD3_INT_CONTROL, hpd3); + WREG32(DC_HPD4_INT_CONTROL, hpd4); + WREG32(DC_HPD5_INT_CONTROL, hpd5); + WREG32(DC_HPD6_INT_CONTROL, hpd6); + + return 0; +} + +static inline void evergreen_irq_ack(struct radeon_device *rdev) +{ + u32 tmp; + + rdev->irq.stat_regs.evergreen.disp_int = RREG32(DISP_INTERRUPT_STATUS); + rdev->irq.stat_regs.evergreen.disp_int_cont = RREG32(DISP_INTERRUPT_STATUS_CONTINUE); + rdev->irq.stat_regs.evergreen.disp_int_cont2 = RREG32(DISP_INTERRUPT_STATUS_CONTINUE2); + rdev->irq.stat_regs.evergreen.disp_int_cont3 = RREG32(DISP_INTERRUPT_STATUS_CONTINUE3); + rdev->irq.stat_regs.evergreen.disp_int_cont4 = RREG32(DISP_INTERRUPT_STATUS_CONTINUE4); + rdev->irq.stat_regs.evergreen.disp_int_cont5 = RREG32(DISP_INTERRUPT_STATUS_CONTINUE5); + rdev->irq.stat_regs.evergreen.d1grph_int = RREG32(GRPH_INT_STATUS + EVERGREEN_CRTC0_REGISTER_OFFSET); + rdev->irq.stat_regs.evergreen.d2grph_int = RREG32(GRPH_INT_STATUS + EVERGREEN_CRTC1_REGISTER_OFFSET); + if (rdev->num_crtc >= 4) { + rdev->irq.stat_regs.evergreen.d3grph_int = RREG32(GRPH_INT_STATUS + EVERGREEN_CRTC2_REGISTER_OFFSET); + rdev->irq.stat_regs.evergreen.d4grph_int = RREG32(GRPH_INT_STATUS + EVERGREEN_CRTC3_REGISTER_OFFSET); + } + if (rdev->num_crtc >= 6) { + rdev->irq.stat_regs.evergreen.d5grph_int = RREG32(GRPH_INT_STATUS + EVERGREEN_CRTC4_REGISTER_OFFSET); + rdev->irq.stat_regs.evergreen.d6grph_int = RREG32(GRPH_INT_STATUS + EVERGREEN_CRTC5_REGISTER_OFFSET); + } + + if (rdev->irq.stat_regs.evergreen.d1grph_int & GRPH_PFLIP_INT_OCCURRED) + WREG32(GRPH_INT_STATUS + EVERGREEN_CRTC0_REGISTER_OFFSET, GRPH_PFLIP_INT_CLEAR); + if (rdev->irq.stat_regs.evergreen.d2grph_int & GRPH_PFLIP_INT_OCCURRED) + WREG32(GRPH_INT_STATUS + EVERGREEN_CRTC1_REGISTER_OFFSET, GRPH_PFLIP_INT_CLEAR); + if (rdev->irq.stat_regs.evergreen.disp_int & LB_D1_VBLANK_INTERRUPT) + WREG32(VBLANK_STATUS + EVERGREEN_CRTC0_REGISTER_OFFSET, VBLANK_ACK); + if (rdev->irq.stat_regs.evergreen.disp_int & LB_D1_VLINE_INTERRUPT) + WREG32(VLINE_STATUS + EVERGREEN_CRTC0_REGISTER_OFFSET, VLINE_ACK); + if (rdev->irq.stat_regs.evergreen.disp_int_cont & LB_D2_VBLANK_INTERRUPT) + WREG32(VBLANK_STATUS + EVERGREEN_CRTC1_REGISTER_OFFSET, VBLANK_ACK); + if (rdev->irq.stat_regs.evergreen.disp_int_cont & LB_D2_VLINE_INTERRUPT) + WREG32(VLINE_STATUS + EVERGREEN_CRTC1_REGISTER_OFFSET, VLINE_ACK); + + if (rdev->num_crtc >= 4) { + if (rdev->irq.stat_regs.evergreen.d3grph_int & GRPH_PFLIP_INT_OCCURRED) + WREG32(GRPH_INT_STATUS + EVERGREEN_CRTC2_REGISTER_OFFSET, GRPH_PFLIP_INT_CLEAR); + if (rdev->irq.stat_regs.evergreen.d4grph_int & GRPH_PFLIP_INT_OCCURRED) + WREG32(GRPH_INT_STATUS + EVERGREEN_CRTC3_REGISTER_OFFSET, GRPH_PFLIP_INT_CLEAR); + if (rdev->irq.stat_regs.evergreen.disp_int_cont2 & LB_D3_VBLANK_INTERRUPT) + WREG32(VBLANK_STATUS + EVERGREEN_CRTC2_REGISTER_OFFSET, VBLANK_ACK); + if (rdev->irq.stat_regs.evergreen.disp_int_cont2 & LB_D3_VLINE_INTERRUPT) + WREG32(VLINE_STATUS + EVERGREEN_CRTC2_REGISTER_OFFSET, VLINE_ACK); + if (rdev->irq.stat_regs.evergreen.disp_int_cont3 & LB_D4_VBLANK_INTERRUPT) + WREG32(VBLANK_STATUS + EVERGREEN_CRTC3_REGISTER_OFFSET, VBLANK_ACK); + if (rdev->irq.stat_regs.evergreen.disp_int_cont3 & LB_D4_VLINE_INTERRUPT) + WREG32(VLINE_STATUS + EVERGREEN_CRTC3_REGISTER_OFFSET, VLINE_ACK); + } + + if (rdev->num_crtc >= 6) { + if (rdev->irq.stat_regs.evergreen.d5grph_int & GRPH_PFLIP_INT_OCCURRED) + WREG32(GRPH_INT_STATUS + EVERGREEN_CRTC4_REGISTER_OFFSET, GRPH_PFLIP_INT_CLEAR); + if (rdev->irq.stat_regs.evergreen.d6grph_int & GRPH_PFLIP_INT_OCCURRED) + WREG32(GRPH_INT_STATUS + EVERGREEN_CRTC5_REGISTER_OFFSET, GRPH_PFLIP_INT_CLEAR); + if (rdev->irq.stat_regs.evergreen.disp_int_cont4 & LB_D5_VBLANK_INTERRUPT) + WREG32(VBLANK_STATUS + EVERGREEN_CRTC4_REGISTER_OFFSET, VBLANK_ACK); + if (rdev->irq.stat_regs.evergreen.disp_int_cont4 & LB_D5_VLINE_INTERRUPT) + WREG32(VLINE_STATUS + EVERGREEN_CRTC4_REGISTER_OFFSET, VLINE_ACK); + if (rdev->irq.stat_regs.evergreen.disp_int_cont5 & LB_D6_VBLANK_INTERRUPT) + WREG32(VBLANK_STATUS + EVERGREEN_CRTC5_REGISTER_OFFSET, VBLANK_ACK); + if (rdev->irq.stat_regs.evergreen.disp_int_cont5 & LB_D6_VLINE_INTERRUPT) + WREG32(VLINE_STATUS + EVERGREEN_CRTC5_REGISTER_OFFSET, VLINE_ACK); + } + + if (rdev->irq.stat_regs.evergreen.disp_int & DC_HPD1_INTERRUPT) { + tmp = RREG32(DC_HPD1_INT_CONTROL); + tmp |= DC_HPDx_INT_ACK; + WREG32(DC_HPD1_INT_CONTROL, tmp); + } + if (rdev->irq.stat_regs.evergreen.disp_int_cont & DC_HPD2_INTERRUPT) { + tmp = RREG32(DC_HPD2_INT_CONTROL); + tmp |= DC_HPDx_INT_ACK; + WREG32(DC_HPD2_INT_CONTROL, tmp); + } + if (rdev->irq.stat_regs.evergreen.disp_int_cont2 & DC_HPD3_INTERRUPT) { + tmp = RREG32(DC_HPD3_INT_CONTROL); + tmp |= DC_HPDx_INT_ACK; + WREG32(DC_HPD3_INT_CONTROL, tmp); + } + if (rdev->irq.stat_regs.evergreen.disp_int_cont3 & DC_HPD4_INTERRUPT) { + tmp = RREG32(DC_HPD4_INT_CONTROL); + tmp |= DC_HPDx_INT_ACK; + WREG32(DC_HPD4_INT_CONTROL, tmp); + } + if (rdev->irq.stat_regs.evergreen.disp_int_cont4 & DC_HPD5_INTERRUPT) { + tmp = RREG32(DC_HPD5_INT_CONTROL); + tmp |= DC_HPDx_INT_ACK; + WREG32(DC_HPD5_INT_CONTROL, tmp); + } + if (rdev->irq.stat_regs.evergreen.disp_int_cont5 & DC_HPD6_INTERRUPT) { + tmp = RREG32(DC_HPD5_INT_CONTROL); + tmp |= DC_HPDx_INT_ACK; + WREG32(DC_HPD6_INT_CONTROL, tmp); + } +} + +void evergreen_irq_disable(struct radeon_device *rdev) +{ + r600_disable_interrupts(rdev); + /* Wait and acknowledge irq */ + mdelay(1); + evergreen_irq_ack(rdev); + evergreen_disable_interrupt_state(rdev); +} + +void evergreen_irq_suspend(struct radeon_device *rdev) +{ + evergreen_irq_disable(rdev); + r600_rlc_stop(rdev); +} + +static inline u32 evergreen_get_ih_wptr(struct radeon_device *rdev) +{ + u32 wptr, tmp; + + if (rdev->wb.enabled) + wptr = le32_to_cpu(rdev->wb.wb[R600_WB_IH_WPTR_OFFSET/4]); + else + wptr = RREG32(IH_RB_WPTR); + + if (wptr & RB_OVERFLOW) { + /* When a ring buffer overflow happen start parsing interrupt + * from the last not overwritten vector (wptr + 16). Hopefully + * this should allow us to catchup. + */ + dev_warn(rdev->dev, "IH ring buffer overflow (0x%08X, %d, %d)\n", + wptr, rdev->ih.rptr, (wptr + 16) + rdev->ih.ptr_mask); + rdev->ih.rptr = (wptr + 16) & rdev->ih.ptr_mask; + tmp = RREG32(IH_RB_CNTL); + tmp |= IH_WPTR_OVERFLOW_CLEAR; + WREG32(IH_RB_CNTL, tmp); + } + return (wptr & rdev->ih.ptr_mask); +} + +int evergreen_irq_process(struct radeon_device *rdev) +{ + u32 wptr; + u32 rptr; + u32 src_id, src_data; + u32 ring_index; + unsigned long flags; + bool queue_hotplug = false; + + if (!rdev->ih.enabled || rdev->shutdown) + return IRQ_NONE; + + wptr = evergreen_get_ih_wptr(rdev); + rptr = rdev->ih.rptr; + DRM_DEBUG("r600_irq_process start: rptr %d, wptr %d\n", rptr, wptr); + + spin_lock_irqsave(&rdev->ih.lock, flags); + if (rptr == wptr) { + spin_unlock_irqrestore(&rdev->ih.lock, flags); + return IRQ_NONE; + } +restart_ih: + /* display interrupts */ + evergreen_irq_ack(rdev); + + rdev->ih.wptr = wptr; + while (rptr != wptr) { + /* wptr/rptr are in bytes! */ + ring_index = rptr / 4; + src_id = le32_to_cpu(rdev->ih.ring[ring_index]) & 0xff; + src_data = le32_to_cpu(rdev->ih.ring[ring_index + 1]) & 0xfffffff; + + switch (src_id) { + case 1: /* D1 vblank/vline */ + switch (src_data) { + case 0: /* D1 vblank */ + if (rdev->irq.stat_regs.evergreen.disp_int & LB_D1_VBLANK_INTERRUPT) { + if (rdev->irq.crtc_vblank_int[0]) { + drm_handle_vblank(rdev->ddev, 0); + rdev->pm.vblank_sync = true; + wake_up(&rdev->irq.vblank_queue); + } + if (rdev->irq.pflip[0]) + radeon_crtc_handle_flip(rdev, 0); + rdev->irq.stat_regs.evergreen.disp_int &= ~LB_D1_VBLANK_INTERRUPT; + DRM_DEBUG("IH: D1 vblank\n"); + } + break; + case 1: /* D1 vline */ + if (rdev->irq.stat_regs.evergreen.disp_int & LB_D1_VLINE_INTERRUPT) { + rdev->irq.stat_regs.evergreen.disp_int &= ~LB_D1_VLINE_INTERRUPT; + DRM_DEBUG("IH: D1 vline\n"); + } + break; + default: + DRM_DEBUG("Unhandled interrupt: %d %d\n", src_id, src_data); + break; + } + break; + case 2: /* D2 vblank/vline */ + switch (src_data) { + case 0: /* D2 vblank */ + if (rdev->irq.stat_regs.evergreen.disp_int_cont & LB_D2_VBLANK_INTERRUPT) { + if (rdev->irq.crtc_vblank_int[1]) { + drm_handle_vblank(rdev->ddev, 1); + rdev->pm.vblank_sync = true; + wake_up(&rdev->irq.vblank_queue); + } + if (rdev->irq.pflip[1]) + radeon_crtc_handle_flip(rdev, 1); + rdev->irq.stat_regs.evergreen.disp_int_cont &= ~LB_D2_VBLANK_INTERRUPT; + DRM_DEBUG("IH: D2 vblank\n"); + } + break; + case 1: /* D2 vline */ + if (rdev->irq.stat_regs.evergreen.disp_int_cont & LB_D2_VLINE_INTERRUPT) { + rdev->irq.stat_regs.evergreen.disp_int_cont &= ~LB_D2_VLINE_INTERRUPT; + DRM_DEBUG("IH: D2 vline\n"); + } + break; + default: + DRM_DEBUG("Unhandled interrupt: %d %d\n", src_id, src_data); + break; + } + break; + case 3: /* D3 vblank/vline */ + switch (src_data) { + case 0: /* D3 vblank */ + if (rdev->irq.stat_regs.evergreen.disp_int_cont2 & LB_D3_VBLANK_INTERRUPT) { + if (rdev->irq.crtc_vblank_int[2]) { + drm_handle_vblank(rdev->ddev, 2); + rdev->pm.vblank_sync = true; + wake_up(&rdev->irq.vblank_queue); + } + if (rdev->irq.pflip[2]) + radeon_crtc_handle_flip(rdev, 2); + rdev->irq.stat_regs.evergreen.disp_int_cont2 &= ~LB_D3_VBLANK_INTERRUPT; + DRM_DEBUG("IH: D3 vblank\n"); + } + break; + case 1: /* D3 vline */ + if (rdev->irq.stat_regs.evergreen.disp_int_cont2 & LB_D3_VLINE_INTERRUPT) { + rdev->irq.stat_regs.evergreen.disp_int_cont2 &= ~LB_D3_VLINE_INTERRUPT; + DRM_DEBUG("IH: D3 vline\n"); + } + break; + default: + DRM_DEBUG("Unhandled interrupt: %d %d\n", src_id, src_data); + break; + } + break; + case 4: /* D4 vblank/vline */ + switch (src_data) { + case 0: /* D4 vblank */ + if (rdev->irq.stat_regs.evergreen.disp_int_cont3 & LB_D4_VBLANK_INTERRUPT) { + if (rdev->irq.crtc_vblank_int[3]) { + drm_handle_vblank(rdev->ddev, 3); + rdev->pm.vblank_sync = true; + wake_up(&rdev->irq.vblank_queue); + } + if (rdev->irq.pflip[3]) + radeon_crtc_handle_flip(rdev, 3); + rdev->irq.stat_regs.evergreen.disp_int_cont3 &= ~LB_D4_VBLANK_INTERRUPT; + DRM_DEBUG("IH: D4 vblank\n"); + } + break; + case 1: /* D4 vline */ + if (rdev->irq.stat_regs.evergreen.disp_int_cont3 & LB_D4_VLINE_INTERRUPT) { + rdev->irq.stat_regs.evergreen.disp_int_cont3 &= ~LB_D4_VLINE_INTERRUPT; + DRM_DEBUG("IH: D4 vline\n"); + } + break; + default: + DRM_DEBUG("Unhandled interrupt: %d %d\n", src_id, src_data); + break; + } + break; + case 5: /* D5 vblank/vline */ + switch (src_data) { + case 0: /* D5 vblank */ + if (rdev->irq.stat_regs.evergreen.disp_int_cont4 & LB_D5_VBLANK_INTERRUPT) { + if (rdev->irq.crtc_vblank_int[4]) { + drm_handle_vblank(rdev->ddev, 4); + rdev->pm.vblank_sync = true; + wake_up(&rdev->irq.vblank_queue); + } + if (rdev->irq.pflip[4]) + radeon_crtc_handle_flip(rdev, 4); + rdev->irq.stat_regs.evergreen.disp_int_cont4 &= ~LB_D5_VBLANK_INTERRUPT; + DRM_DEBUG("IH: D5 vblank\n"); + } + break; + case 1: /* D5 vline */ + if (rdev->irq.stat_regs.evergreen.disp_int_cont4 & LB_D5_VLINE_INTERRUPT) { + rdev->irq.stat_regs.evergreen.disp_int_cont4 &= ~LB_D5_VLINE_INTERRUPT; + DRM_DEBUG("IH: D5 vline\n"); + } + break; + default: + DRM_DEBUG("Unhandled interrupt: %d %d\n", src_id, src_data); + break; + } + break; + case 6: /* D6 vblank/vline */ + switch (src_data) { + case 0: /* D6 vblank */ + if (rdev->irq.stat_regs.evergreen.disp_int_cont5 & LB_D6_VBLANK_INTERRUPT) { + if (rdev->irq.crtc_vblank_int[5]) { + drm_handle_vblank(rdev->ddev, 5); + rdev->pm.vblank_sync = true; + wake_up(&rdev->irq.vblank_queue); + } + if (rdev->irq.pflip[5]) + radeon_crtc_handle_flip(rdev, 5); + rdev->irq.stat_regs.evergreen.disp_int_cont5 &= ~LB_D6_VBLANK_INTERRUPT; + DRM_DEBUG("IH: D6 vblank\n"); + } + break; + case 1: /* D6 vline */ + if (rdev->irq.stat_regs.evergreen.disp_int_cont5 & LB_D6_VLINE_INTERRUPT) { + rdev->irq.stat_regs.evergreen.disp_int_cont5 &= ~LB_D6_VLINE_INTERRUPT; + DRM_DEBUG("IH: D6 vline\n"); + } + break; + default: + DRM_DEBUG("Unhandled interrupt: %d %d\n", src_id, src_data); + break; + } + break; + case 42: /* HPD hotplug */ + switch (src_data) { + case 0: + if (rdev->irq.stat_regs.evergreen.disp_int & DC_HPD1_INTERRUPT) { + rdev->irq.stat_regs.evergreen.disp_int &= ~DC_HPD1_INTERRUPT; + queue_hotplug = true; + DRM_DEBUG("IH: HPD1\n"); + } + break; + case 1: + if (rdev->irq.stat_regs.evergreen.disp_int_cont & DC_HPD2_INTERRUPT) { + rdev->irq.stat_regs.evergreen.disp_int_cont &= ~DC_HPD2_INTERRUPT; + queue_hotplug = true; + DRM_DEBUG("IH: HPD2\n"); + } + break; + case 2: + if (rdev->irq.stat_regs.evergreen.disp_int_cont2 & DC_HPD3_INTERRUPT) { + rdev->irq.stat_regs.evergreen.disp_int_cont2 &= ~DC_HPD3_INTERRUPT; + queue_hotplug = true; + DRM_DEBUG("IH: HPD3\n"); + } + break; + case 3: + if (rdev->irq.stat_regs.evergreen.disp_int_cont3 & DC_HPD4_INTERRUPT) { + rdev->irq.stat_regs.evergreen.disp_int_cont3 &= ~DC_HPD4_INTERRUPT; + queue_hotplug = true; + DRM_DEBUG("IH: HPD4\n"); + } + break; + case 4: + if (rdev->irq.stat_regs.evergreen.disp_int_cont4 & DC_HPD5_INTERRUPT) { + rdev->irq.stat_regs.evergreen.disp_int_cont4 &= ~DC_HPD5_INTERRUPT; + queue_hotplug = true; + DRM_DEBUG("IH: HPD5\n"); + } + break; + case 5: + if (rdev->irq.stat_regs.evergreen.disp_int_cont5 & DC_HPD6_INTERRUPT) { + rdev->irq.stat_regs.evergreen.disp_int_cont5 &= ~DC_HPD6_INTERRUPT; + queue_hotplug = true; + DRM_DEBUG("IH: HPD6\n"); + } + break; + default: + DRM_DEBUG("Unhandled interrupt: %d %d\n", src_id, src_data); + break; + } + break; + case 176: /* CP_INT in ring buffer */ + case 177: /* CP_INT in IB1 */ + case 178: /* CP_INT in IB2 */ + DRM_DEBUG("IH: CP int: 0x%08x\n", src_data); + radeon_fence_process(rdev); + break; + case 181: /* CP EOP event */ + DRM_DEBUG("IH: CP EOP\n"); + radeon_fence_process(rdev); + break; + case 233: /* GUI IDLE */ + DRM_DEBUG("IH: GUI idle\n"); + rdev->pm.gui_idle = true; + wake_up(&rdev->irq.idle_queue); + break; + default: + DRM_DEBUG("Unhandled interrupt: %d %d\n", src_id, src_data); + break; + } + + /* wptr/rptr are in bytes! */ + rptr += 16; + rptr &= rdev->ih.ptr_mask; + } + /* make sure wptr hasn't changed while processing */ + wptr = evergreen_get_ih_wptr(rdev); + if (wptr != rdev->ih.wptr) + goto restart_ih; + if (queue_hotplug) + schedule_work(&rdev->hotplug_work); + rdev->ih.rptr = rptr; + WREG32(IH_RB_RPTR, rdev->ih.rptr); + spin_unlock_irqrestore(&rdev->ih.lock, flags); + return IRQ_HANDLED; +} + +static int evergreen_startup(struct radeon_device *rdev) +{ + int r; + + /* enable pcie gen2 link */ + if (!ASIC_IS_DCE5(rdev)) + evergreen_pcie_gen2_enable(rdev); + + if (ASIC_IS_DCE5(rdev)) { + if (!rdev->me_fw || !rdev->pfp_fw || !rdev->rlc_fw || !rdev->mc_fw) { + r = ni_init_microcode(rdev); + if (r) { + DRM_ERROR("Failed to load firmware!\n"); + return r; + } + } + r = ni_mc_load_microcode(rdev); + if (r) { + DRM_ERROR("Failed to load MC firmware!\n"); + return r; + } + } else { + if (!rdev->me_fw || !rdev->pfp_fw || !rdev->rlc_fw) { + r = r600_init_microcode(rdev); + if (r) { + DRM_ERROR("Failed to load firmware!\n"); + return r; + } + } + } + + evergreen_mc_program(rdev); + if (rdev->flags & RADEON_IS_AGP) { + evergreen_agp_enable(rdev); + } else { + r = evergreen_pcie_gart_enable(rdev); + if (r) + return r; + } + evergreen_gpu_init(rdev); + + r = evergreen_blit_init(rdev); + if (r) { + evergreen_blit_fini(rdev); + rdev->asic->copy = NULL; + dev_warn(rdev->dev, "failed blitter (%d) falling back to memcpy\n", r); + } + + /* allocate wb buffer */ + r = radeon_wb_init(rdev); + if (r) + return r; + + /* Enable IRQ */ + r = r600_irq_init(rdev); + if (r) { + DRM_ERROR("radeon: IH init failed (%d).\n", r); + radeon_irq_kms_fini(rdev); + return r; + } + evergreen_irq_set(rdev); + + r = radeon_ring_init(rdev, rdev->cp.ring_size); + if (r) + return r; + r = evergreen_cp_load_microcode(rdev); + if (r) + return r; + r = evergreen_cp_resume(rdev); + if (r) + return r; + + return 0; +} + +int evergreen_resume(struct radeon_device *rdev) +{ + int r; + + /* reset the asic, the gfx blocks are often in a bad state + * after the driver is unloaded or after a resume + */ + if (radeon_asic_reset(rdev)) + dev_warn(rdev->dev, "GPU reset failed !\n"); + /* Do not reset GPU before posting, on rv770 hw unlike on r500 hw, + * posting will perform necessary task to bring back GPU into good + * shape. + */ + /* post card */ + atom_asic_init(rdev->mode_info.atom_context); + + r = evergreen_startup(rdev); + if (r) { + DRM_ERROR("evergreen startup failed on resume\n"); + return r; + } + + r = r600_ib_test(rdev); + if (r) { + DRM_ERROR("radeon: failed testing IB (%d).\n", r); + return r; + } + + return r; + +} + +int evergreen_suspend(struct radeon_device *rdev) +{ + int r; + + /* FIXME: we should wait for ring to be empty */ + r700_cp_stop(rdev); + rdev->cp.ready = false; + evergreen_irq_suspend(rdev); + radeon_wb_disable(rdev); + evergreen_pcie_gart_disable(rdev); + + /* unpin shaders bo */ + r = radeon_bo_reserve(rdev->r600_blit.shader_obj, false); + if (likely(r == 0)) { + radeon_bo_unpin(rdev->r600_blit.shader_obj); + radeon_bo_unreserve(rdev->r600_blit.shader_obj); + } + + return 0; +} + +int evergreen_copy_blit(struct radeon_device *rdev, + uint64_t src_offset, + uint64_t dst_offset, + unsigned num_gpu_pages, + struct radeon_fence *fence) +{ + int r; + + mutex_lock(&rdev->r600_blit.mutex); + rdev->r600_blit.vb_ib = NULL; + r = evergreen_blit_prepare_copy(rdev, num_gpu_pages * RADEON_GPU_PAGE_SIZE); + if (r) { + if (rdev->r600_blit.vb_ib) + radeon_ib_free(rdev, &rdev->r600_blit.vb_ib); + mutex_unlock(&rdev->r600_blit.mutex); + return r; + } + evergreen_kms_blit_copy(rdev, src_offset, dst_offset, num_gpu_pages * RADEON_GPU_PAGE_SIZE); + evergreen_blit_done_copy(rdev, fence); + mutex_unlock(&rdev->r600_blit.mutex); + return 0; +} + +/* Plan is to move initialization in that function and use + * helper function so that radeon_device_init pretty much + * do nothing more than calling asic specific function. This + * should also allow to remove a bunch of callback function + * like vram_info. + */ +int evergreen_init(struct radeon_device *rdev) +{ + int r; + + /* This don't do much */ + r = radeon_gem_init(rdev); + if (r) + return r; + /* Read BIOS */ + if (!radeon_get_bios(rdev)) { + if (ASIC_IS_AVIVO(rdev)) + return -EINVAL; + } + /* Must be an ATOMBIOS */ + if (!rdev->is_atom_bios) { + dev_err(rdev->dev, "Expecting atombios for evergreen GPU\n"); + return -EINVAL; + } + r = radeon_atombios_init(rdev); + if (r) + return r; + /* reset the asic, the gfx blocks are often in a bad state + * after the driver is unloaded or after a resume + */ + if (radeon_asic_reset(rdev)) + dev_warn(rdev->dev, "GPU reset failed !\n"); + /* Post card if necessary */ + if (!radeon_card_posted(rdev)) { + if (!rdev->bios) { + dev_err(rdev->dev, "Card not posted and no BIOS - ignoring\n"); + return -EINVAL; + } + DRM_INFO("GPU not posted. posting now...\n"); + atom_asic_init(rdev->mode_info.atom_context); + } + /* Initialize scratch registers */ + r600_scratch_init(rdev); + /* Initialize surface registers */ + radeon_surface_init(rdev); + /* Initialize clocks */ + radeon_get_clock_info(rdev->ddev); + /* Fence driver */ + r = radeon_fence_driver_init(rdev); + if (r) + return r; + /* initialize AGP */ + if (rdev->flags & RADEON_IS_AGP) { + r = radeon_agp_init(rdev); + if (r) + radeon_agp_disable(rdev); + } + /* initialize memory controller */ + r = evergreen_mc_init(rdev); + if (r) + return r; + /* Memory manager */ + r = radeon_bo_init(rdev); + if (r) + return r; + + r = radeon_irq_kms_init(rdev); + if (r) + return r; + + rdev->cp.ring_obj = NULL; + r600_ring_init(rdev, 1024 * 1024); + + rdev->ih.ring_obj = NULL; + r600_ih_ring_init(rdev, 64 * 1024); + + r = r600_pcie_gart_init(rdev); + if (r) + return r; + + rdev->accel_working = true; + r = evergreen_startup(rdev); + if (r) { + dev_err(rdev->dev, "disabling GPU acceleration\n"); + r700_cp_fini(rdev); + r600_irq_fini(rdev); + radeon_wb_fini(rdev); + radeon_irq_kms_fini(rdev); + evergreen_pcie_gart_fini(rdev); + rdev->accel_working = false; + } + if (rdev->accel_working) { + r = radeon_ib_pool_init(rdev); + if (r) { + DRM_ERROR("radeon: failed initializing IB pool (%d).\n", r); + rdev->accel_working = false; + } + r = r600_ib_test(rdev); + if (r) { + DRM_ERROR("radeon: failed testing IB (%d).\n", r); + rdev->accel_working = false; + } + } + + /* Don't start up if the MC ucode is missing on BTC parts. + * The default clocks and voltages before the MC ucode + * is loaded are not suffient for advanced operations. + */ + if (ASIC_IS_DCE5(rdev)) { + if (!rdev->mc_fw && !(rdev->flags & RADEON_IS_IGP)) { + DRM_ERROR("radeon: MC ucode required for NI+.\n"); + return -EINVAL; + } + } + + return 0; +} + +void evergreen_fini(struct radeon_device *rdev) +{ + evergreen_blit_fini(rdev); + r700_cp_fini(rdev); + r600_irq_fini(rdev); + radeon_wb_fini(rdev); + radeon_ib_pool_fini(rdev); + radeon_irq_kms_fini(rdev); + evergreen_pcie_gart_fini(rdev); + radeon_gem_fini(rdev); + radeon_fence_driver_fini(rdev); + radeon_agp_fini(rdev); + radeon_bo_fini(rdev); + radeon_atombios_fini(rdev); + kfree(rdev->bios); + rdev->bios = NULL; +} + +static void evergreen_pcie_gen2_enable(struct radeon_device *rdev) +{ + u32 link_width_cntl, speed_cntl; + + if (radeon_pcie_gen2 == 0) + return; + + if (rdev->flags & RADEON_IS_IGP) + return; + + if (!(rdev->flags & RADEON_IS_PCIE)) + return; + + /* x2 cards have a special sequence */ + if (ASIC_IS_X2(rdev)) + return; + + speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL); + if ((speed_cntl & LC_OTHER_SIDE_EVER_SENT_GEN2) || + (speed_cntl & LC_OTHER_SIDE_SUPPORTS_GEN2)) { + + link_width_cntl = RREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL); + link_width_cntl &= ~LC_UPCONFIGURE_DIS; + WREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl); + + speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL); + speed_cntl &= ~LC_TARGET_LINK_SPEED_OVERRIDE_EN; + WREG32_PCIE_P(PCIE_LC_SPEED_CNTL, speed_cntl); + + speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL); + speed_cntl |= LC_CLR_FAILED_SPD_CHANGE_CNT; + WREG32_PCIE_P(PCIE_LC_SPEED_CNTL, speed_cntl); + + speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL); + speed_cntl &= ~LC_CLR_FAILED_SPD_CHANGE_CNT; + WREG32_PCIE_P(PCIE_LC_SPEED_CNTL, speed_cntl); + + speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL); + speed_cntl |= LC_GEN2_EN_STRAP; + WREG32_PCIE_P(PCIE_LC_SPEED_CNTL, speed_cntl); + + } else { + link_width_cntl = RREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL); + /* XXX: only disable it if gen1 bridge vendor == 0x111d or 0x1106 */ + if (1) + link_width_cntl |= LC_UPCONFIGURE_DIS; + else + link_width_cntl &= ~LC_UPCONFIGURE_DIS; + WREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl); + } +} diff --git a/drivers/gpu/drm/radeon/evergreen_blit_kms.c b/drivers/gpu/drm/radeon/evergreen_blit_kms.c new file mode 100644 index 00000000..2eb25185 --- /dev/null +++ b/drivers/gpu/drm/radeon/evergreen_blit_kms.c @@ -0,0 +1,988 @@ +/* + * Copyright 2010 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Alex Deucher + */ + +#include "drmP.h" +#include "drm.h" +#include "radeon_drm.h" +#include "radeon.h" + +#include "evergreend.h" +#include "evergreen_blit_shaders.h" +#include "cayman_blit_shaders.h" + +#define DI_PT_RECTLIST 0x11 +#define DI_INDEX_SIZE_16_BIT 0x0 +#define DI_SRC_SEL_AUTO_INDEX 0x2 + +#define FMT_8 0x1 +#define FMT_5_6_5 0x8 +#define FMT_8_8_8_8 0x1a +#define COLOR_8 0x1 +#define COLOR_5_6_5 0x8 +#define COLOR_8_8_8_8 0x1a + +/* emits 17 */ +static void +set_render_target(struct radeon_device *rdev, int format, + int w, int h, u64 gpu_addr) +{ + u32 cb_color_info; + int pitch, slice; + + h = ALIGN(h, 8); + if (h < 8) + h = 8; + + cb_color_info = ((format << 2) | (1 << 24) | (1 << 8)); + pitch = (w / 8) - 1; + slice = ((w * h) / 64) - 1; + + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONTEXT_REG, 15)); + radeon_ring_write(rdev, (CB_COLOR0_BASE - PACKET3_SET_CONTEXT_REG_START) >> 2); + radeon_ring_write(rdev, gpu_addr >> 8); + radeon_ring_write(rdev, pitch); + radeon_ring_write(rdev, slice); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, cb_color_info); + radeon_ring_write(rdev, (1 << 4)); + radeon_ring_write(rdev, (w - 1) | ((h - 1) << 16)); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, 0); +} + +/* emits 5dw */ +static void +cp_set_surface_sync(struct radeon_device *rdev, + u32 sync_type, u32 size, + u64 mc_addr) +{ + u32 cp_coher_size; + + if (size == 0xffffffff) + cp_coher_size = 0xffffffff; + else + cp_coher_size = ((size + 255) >> 8); + + radeon_ring_write(rdev, PACKET3(PACKET3_SURFACE_SYNC, 3)); + radeon_ring_write(rdev, sync_type); + radeon_ring_write(rdev, cp_coher_size); + radeon_ring_write(rdev, mc_addr >> 8); + radeon_ring_write(rdev, 10); /* poll interval */ +} + +/* emits 11dw + 1 surface sync = 16dw */ +static void +set_shaders(struct radeon_device *rdev) +{ + u64 gpu_addr; + + /* VS */ + gpu_addr = rdev->r600_blit.shader_gpu_addr + rdev->r600_blit.vs_offset; + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONTEXT_REG, 3)); + radeon_ring_write(rdev, (SQ_PGM_START_VS - PACKET3_SET_CONTEXT_REG_START) >> 2); + radeon_ring_write(rdev, gpu_addr >> 8); + radeon_ring_write(rdev, 2); + radeon_ring_write(rdev, 0); + + /* PS */ + gpu_addr = rdev->r600_blit.shader_gpu_addr + rdev->r600_blit.ps_offset; + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONTEXT_REG, 4)); + radeon_ring_write(rdev, (SQ_PGM_START_PS - PACKET3_SET_CONTEXT_REG_START) >> 2); + radeon_ring_write(rdev, gpu_addr >> 8); + radeon_ring_write(rdev, 1); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, 2); + + gpu_addr = rdev->r600_blit.shader_gpu_addr + rdev->r600_blit.vs_offset; + cp_set_surface_sync(rdev, PACKET3_SH_ACTION_ENA, 512, gpu_addr); +} + +/* emits 10 + 1 sync (5) = 15 */ +static void +set_vtx_resource(struct radeon_device *rdev, u64 gpu_addr) +{ + u32 sq_vtx_constant_word2, sq_vtx_constant_word3; + + /* high addr, stride */ + sq_vtx_constant_word2 = ((upper_32_bits(gpu_addr) & 0xff) | (16 << 8)); +#ifdef __BIG_ENDIAN + sq_vtx_constant_word2 |= (2 << 30); +#endif + /* xyzw swizzles */ + sq_vtx_constant_word3 = (0 << 3) | (1 << 6) | (2 << 9) | (3 << 12); + + radeon_ring_write(rdev, PACKET3(PACKET3_SET_RESOURCE, 8)); + radeon_ring_write(rdev, 0x580); + radeon_ring_write(rdev, gpu_addr & 0xffffffff); + radeon_ring_write(rdev, 48 - 1); /* size */ + radeon_ring_write(rdev, sq_vtx_constant_word2); + radeon_ring_write(rdev, sq_vtx_constant_word3); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, SQ_TEX_VTX_VALID_BUFFER << 30); + + if ((rdev->family == CHIP_CEDAR) || + (rdev->family == CHIP_PALM) || + (rdev->family == CHIP_SUMO) || + (rdev->family == CHIP_SUMO2) || + (rdev->family == CHIP_CAICOS)) + cp_set_surface_sync(rdev, + PACKET3_TC_ACTION_ENA, 48, gpu_addr); + else + cp_set_surface_sync(rdev, + PACKET3_VC_ACTION_ENA, 48, gpu_addr); + +} + +/* emits 10 */ +static void +set_tex_resource(struct radeon_device *rdev, + int format, int w, int h, int pitch, + u64 gpu_addr) +{ + u32 sq_tex_resource_word0, sq_tex_resource_word1; + u32 sq_tex_resource_word4, sq_tex_resource_word7; + + if (h < 1) + h = 1; + + sq_tex_resource_word0 = (1 << 0); /* 2D */ + sq_tex_resource_word0 |= ((((pitch >> 3) - 1) << 6) | + ((w - 1) << 18)); + sq_tex_resource_word1 = ((h - 1) << 0) | (1 << 28); + /* xyzw swizzles */ + sq_tex_resource_word4 = (0 << 16) | (1 << 19) | (2 << 22) | (3 << 25); + + sq_tex_resource_word7 = format | (SQ_TEX_VTX_VALID_TEXTURE << 30); + + radeon_ring_write(rdev, PACKET3(PACKET3_SET_RESOURCE, 8)); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, sq_tex_resource_word0); + radeon_ring_write(rdev, sq_tex_resource_word1); + radeon_ring_write(rdev, gpu_addr >> 8); + radeon_ring_write(rdev, gpu_addr >> 8); + radeon_ring_write(rdev, sq_tex_resource_word4); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, sq_tex_resource_word7); +} + +/* emits 12 */ +static void +set_scissors(struct radeon_device *rdev, int x1, int y1, + int x2, int y2) +{ + /* workaround some hw bugs */ + if (x2 == 0) + x1 = 1; + if (y2 == 0) + y1 = 1; + if (rdev->family == CHIP_CAYMAN) { + if ((x2 == 1) && (y2 == 1)) + x2 = 2; + } + + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONTEXT_REG, 2)); + radeon_ring_write(rdev, (PA_SC_SCREEN_SCISSOR_TL - PACKET3_SET_CONTEXT_REG_START) >> 2); + radeon_ring_write(rdev, (x1 << 0) | (y1 << 16)); + radeon_ring_write(rdev, (x2 << 0) | (y2 << 16)); + + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONTEXT_REG, 2)); + radeon_ring_write(rdev, (PA_SC_GENERIC_SCISSOR_TL - PACKET3_SET_CONTEXT_REG_START) >> 2); + radeon_ring_write(rdev, (x1 << 0) | (y1 << 16) | (1 << 31)); + radeon_ring_write(rdev, (x2 << 0) | (y2 << 16)); + + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONTEXT_REG, 2)); + radeon_ring_write(rdev, (PA_SC_WINDOW_SCISSOR_TL - PACKET3_SET_CONTEXT_REG_START) >> 2); + radeon_ring_write(rdev, (x1 << 0) | (y1 << 16) | (1 << 31)); + radeon_ring_write(rdev, (x2 << 0) | (y2 << 16)); +} + +/* emits 10 */ +static void +draw_auto(struct radeon_device *rdev) +{ + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONFIG_REG, 1)); + radeon_ring_write(rdev, (VGT_PRIMITIVE_TYPE - PACKET3_SET_CONFIG_REG_START) >> 2); + radeon_ring_write(rdev, DI_PT_RECTLIST); + + radeon_ring_write(rdev, PACKET3(PACKET3_INDEX_TYPE, 0)); + radeon_ring_write(rdev, +#ifdef __BIG_ENDIAN + (2 << 2) | +#endif + DI_INDEX_SIZE_16_BIT); + + radeon_ring_write(rdev, PACKET3(PACKET3_NUM_INSTANCES, 0)); + radeon_ring_write(rdev, 1); + + radeon_ring_write(rdev, PACKET3(PACKET3_DRAW_INDEX_AUTO, 1)); + radeon_ring_write(rdev, 3); + radeon_ring_write(rdev, DI_SRC_SEL_AUTO_INDEX); + +} + +/* emits 39 */ +static void +set_default_state(struct radeon_device *rdev) +{ + u32 sq_config, sq_gpr_resource_mgmt_1, sq_gpr_resource_mgmt_2, sq_gpr_resource_mgmt_3; + u32 sq_thread_resource_mgmt, sq_thread_resource_mgmt_2; + u32 sq_stack_resource_mgmt_1, sq_stack_resource_mgmt_2, sq_stack_resource_mgmt_3; + int num_ps_gprs, num_vs_gprs, num_temp_gprs; + int num_gs_gprs, num_es_gprs, num_hs_gprs, num_ls_gprs; + int num_ps_threads, num_vs_threads, num_gs_threads, num_es_threads; + int num_hs_threads, num_ls_threads; + int num_ps_stack_entries, num_vs_stack_entries, num_gs_stack_entries, num_es_stack_entries; + int num_hs_stack_entries, num_ls_stack_entries; + u64 gpu_addr; + int dwords; + + /* set clear context state */ + radeon_ring_write(rdev, PACKET3(PACKET3_CLEAR_STATE, 0)); + radeon_ring_write(rdev, 0); + + if (rdev->family < CHIP_CAYMAN) { + switch (rdev->family) { + case CHIP_CEDAR: + default: + num_ps_gprs = 93; + num_vs_gprs = 46; + num_temp_gprs = 4; + num_gs_gprs = 31; + num_es_gprs = 31; + num_hs_gprs = 23; + num_ls_gprs = 23; + num_ps_threads = 96; + num_vs_threads = 16; + num_gs_threads = 16; + num_es_threads = 16; + num_hs_threads = 16; + num_ls_threads = 16; + num_ps_stack_entries = 42; + num_vs_stack_entries = 42; + num_gs_stack_entries = 42; + num_es_stack_entries = 42; + num_hs_stack_entries = 42; + num_ls_stack_entries = 42; + break; + case CHIP_REDWOOD: + num_ps_gprs = 93; + num_vs_gprs = 46; + num_temp_gprs = 4; + num_gs_gprs = 31; + num_es_gprs = 31; + num_hs_gprs = 23; + num_ls_gprs = 23; + num_ps_threads = 128; + num_vs_threads = 20; + num_gs_threads = 20; + num_es_threads = 20; + num_hs_threads = 20; + num_ls_threads = 20; + num_ps_stack_entries = 42; + num_vs_stack_entries = 42; + num_gs_stack_entries = 42; + num_es_stack_entries = 42; + num_hs_stack_entries = 42; + num_ls_stack_entries = 42; + break; + case CHIP_JUNIPER: + num_ps_gprs = 93; + num_vs_gprs = 46; + num_temp_gprs = 4; + num_gs_gprs = 31; + num_es_gprs = 31; + num_hs_gprs = 23; + num_ls_gprs = 23; + num_ps_threads = 128; + num_vs_threads = 20; + num_gs_threads = 20; + num_es_threads = 20; + num_hs_threads = 20; + num_ls_threads = 20; + num_ps_stack_entries = 85; + num_vs_stack_entries = 85; + num_gs_stack_entries = 85; + num_es_stack_entries = 85; + num_hs_stack_entries = 85; + num_ls_stack_entries = 85; + break; + case CHIP_CYPRESS: + case CHIP_HEMLOCK: + num_ps_gprs = 93; + num_vs_gprs = 46; + num_temp_gprs = 4; + num_gs_gprs = 31; + num_es_gprs = 31; + num_hs_gprs = 23; + num_ls_gprs = 23; + num_ps_threads = 128; + num_vs_threads = 20; + num_gs_threads = 20; + num_es_threads = 20; + num_hs_threads = 20; + num_ls_threads = 20; + num_ps_stack_entries = 85; + num_vs_stack_entries = 85; + num_gs_stack_entries = 85; + num_es_stack_entries = 85; + num_hs_stack_entries = 85; + num_ls_stack_entries = 85; + break; + case CHIP_PALM: + num_ps_gprs = 93; + num_vs_gprs = 46; + num_temp_gprs = 4; + num_gs_gprs = 31; + num_es_gprs = 31; + num_hs_gprs = 23; + num_ls_gprs = 23; + num_ps_threads = 96; + num_vs_threads = 16; + num_gs_threads = 16; + num_es_threads = 16; + num_hs_threads = 16; + num_ls_threads = 16; + num_ps_stack_entries = 42; + num_vs_stack_entries = 42; + num_gs_stack_entries = 42; + num_es_stack_entries = 42; + num_hs_stack_entries = 42; + num_ls_stack_entries = 42; + break; + case CHIP_SUMO: + num_ps_gprs = 93; + num_vs_gprs = 46; + num_temp_gprs = 4; + num_gs_gprs = 31; + num_es_gprs = 31; + num_hs_gprs = 23; + num_ls_gprs = 23; + num_ps_threads = 96; + num_vs_threads = 25; + num_gs_threads = 25; + num_es_threads = 25; + num_hs_threads = 25; + num_ls_threads = 25; + num_ps_stack_entries = 42; + num_vs_stack_entries = 42; + num_gs_stack_entries = 42; + num_es_stack_entries = 42; + num_hs_stack_entries = 42; + num_ls_stack_entries = 42; + break; + case CHIP_SUMO2: + num_ps_gprs = 93; + num_vs_gprs = 46; + num_temp_gprs = 4; + num_gs_gprs = 31; + num_es_gprs = 31; + num_hs_gprs = 23; + num_ls_gprs = 23; + num_ps_threads = 96; + num_vs_threads = 25; + num_gs_threads = 25; + num_es_threads = 25; + num_hs_threads = 25; + num_ls_threads = 25; + num_ps_stack_entries = 85; + num_vs_stack_entries = 85; + num_gs_stack_entries = 85; + num_es_stack_entries = 85; + num_hs_stack_entries = 85; + num_ls_stack_entries = 85; + break; + case CHIP_BARTS: + num_ps_gprs = 93; + num_vs_gprs = 46; + num_temp_gprs = 4; + num_gs_gprs = 31; + num_es_gprs = 31; + num_hs_gprs = 23; + num_ls_gprs = 23; + num_ps_threads = 128; + num_vs_threads = 20; + num_gs_threads = 20; + num_es_threads = 20; + num_hs_threads = 20; + num_ls_threads = 20; + num_ps_stack_entries = 85; + num_vs_stack_entries = 85; + num_gs_stack_entries = 85; + num_es_stack_entries = 85; + num_hs_stack_entries = 85; + num_ls_stack_entries = 85; + break; + case CHIP_TURKS: + num_ps_gprs = 93; + num_vs_gprs = 46; + num_temp_gprs = 4; + num_gs_gprs = 31; + num_es_gprs = 31; + num_hs_gprs = 23; + num_ls_gprs = 23; + num_ps_threads = 128; + num_vs_threads = 20; + num_gs_threads = 20; + num_es_threads = 20; + num_hs_threads = 20; + num_ls_threads = 20; + num_ps_stack_entries = 42; + num_vs_stack_entries = 42; + num_gs_stack_entries = 42; + num_es_stack_entries = 42; + num_hs_stack_entries = 42; + num_ls_stack_entries = 42; + break; + case CHIP_CAICOS: + num_ps_gprs = 93; + num_vs_gprs = 46; + num_temp_gprs = 4; + num_gs_gprs = 31; + num_es_gprs = 31; + num_hs_gprs = 23; + num_ls_gprs = 23; + num_ps_threads = 128; + num_vs_threads = 10; + num_gs_threads = 10; + num_es_threads = 10; + num_hs_threads = 10; + num_ls_threads = 10; + num_ps_stack_entries = 42; + num_vs_stack_entries = 42; + num_gs_stack_entries = 42; + num_es_stack_entries = 42; + num_hs_stack_entries = 42; + num_ls_stack_entries = 42; + break; + } + + if ((rdev->family == CHIP_CEDAR) || + (rdev->family == CHIP_PALM) || + (rdev->family == CHIP_SUMO) || + (rdev->family == CHIP_SUMO2) || + (rdev->family == CHIP_CAICOS)) + sq_config = 0; + else + sq_config = VC_ENABLE; + + sq_config |= (EXPORT_SRC_C | + CS_PRIO(0) | + LS_PRIO(0) | + HS_PRIO(0) | + PS_PRIO(0) | + VS_PRIO(1) | + GS_PRIO(2) | + ES_PRIO(3)); + + sq_gpr_resource_mgmt_1 = (NUM_PS_GPRS(num_ps_gprs) | + NUM_VS_GPRS(num_vs_gprs) | + NUM_CLAUSE_TEMP_GPRS(num_temp_gprs)); + sq_gpr_resource_mgmt_2 = (NUM_GS_GPRS(num_gs_gprs) | + NUM_ES_GPRS(num_es_gprs)); + sq_gpr_resource_mgmt_3 = (NUM_HS_GPRS(num_hs_gprs) | + NUM_LS_GPRS(num_ls_gprs)); + sq_thread_resource_mgmt = (NUM_PS_THREADS(num_ps_threads) | + NUM_VS_THREADS(num_vs_threads) | + NUM_GS_THREADS(num_gs_threads) | + NUM_ES_THREADS(num_es_threads)); + sq_thread_resource_mgmt_2 = (NUM_HS_THREADS(num_hs_threads) | + NUM_LS_THREADS(num_ls_threads)); + sq_stack_resource_mgmt_1 = (NUM_PS_STACK_ENTRIES(num_ps_stack_entries) | + NUM_VS_STACK_ENTRIES(num_vs_stack_entries)); + sq_stack_resource_mgmt_2 = (NUM_GS_STACK_ENTRIES(num_gs_stack_entries) | + NUM_ES_STACK_ENTRIES(num_es_stack_entries)); + sq_stack_resource_mgmt_3 = (NUM_HS_STACK_ENTRIES(num_hs_stack_entries) | + NUM_LS_STACK_ENTRIES(num_ls_stack_entries)); + + /* disable dyn gprs */ + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONFIG_REG, 1)); + radeon_ring_write(rdev, (SQ_DYN_GPR_CNTL_PS_FLUSH_REQ - PACKET3_SET_CONFIG_REG_START) >> 2); + radeon_ring_write(rdev, 0); + + /* setup LDS */ + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONFIG_REG, 1)); + radeon_ring_write(rdev, (SQ_LDS_RESOURCE_MGMT - PACKET3_SET_CONFIG_REG_START) >> 2); + radeon_ring_write(rdev, 0x10001000); + + /* SQ config */ + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONFIG_REG, 11)); + radeon_ring_write(rdev, (SQ_CONFIG - PACKET3_SET_CONFIG_REG_START) >> 2); + radeon_ring_write(rdev, sq_config); + radeon_ring_write(rdev, sq_gpr_resource_mgmt_1); + radeon_ring_write(rdev, sq_gpr_resource_mgmt_2); + radeon_ring_write(rdev, sq_gpr_resource_mgmt_3); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, sq_thread_resource_mgmt); + radeon_ring_write(rdev, sq_thread_resource_mgmt_2); + radeon_ring_write(rdev, sq_stack_resource_mgmt_1); + radeon_ring_write(rdev, sq_stack_resource_mgmt_2); + radeon_ring_write(rdev, sq_stack_resource_mgmt_3); + } + + /* CONTEXT_CONTROL */ + radeon_ring_write(rdev, 0xc0012800); + radeon_ring_write(rdev, 0x80000000); + radeon_ring_write(rdev, 0x80000000); + + /* SQ_VTX_BASE_VTX_LOC */ + radeon_ring_write(rdev, 0xc0026f00); + radeon_ring_write(rdev, 0x00000000); + radeon_ring_write(rdev, 0x00000000); + radeon_ring_write(rdev, 0x00000000); + + /* SET_SAMPLER */ + radeon_ring_write(rdev, 0xc0036e00); + radeon_ring_write(rdev, 0x00000000); + radeon_ring_write(rdev, 0x00000012); + radeon_ring_write(rdev, 0x00000000); + radeon_ring_write(rdev, 0x00000000); + + /* set to DX10/11 mode */ + radeon_ring_write(rdev, PACKET3(PACKET3_MODE_CONTROL, 0)); + radeon_ring_write(rdev, 1); + + /* emit an IB pointing at default state */ + dwords = ALIGN(rdev->r600_blit.state_len, 0x10); + gpu_addr = rdev->r600_blit.shader_gpu_addr + rdev->r600_blit.state_offset; + radeon_ring_write(rdev, PACKET3(PACKET3_INDIRECT_BUFFER, 2)); + radeon_ring_write(rdev, gpu_addr & 0xFFFFFFFC); + radeon_ring_write(rdev, upper_32_bits(gpu_addr) & 0xFF); + radeon_ring_write(rdev, dwords); + +} + +static inline uint32_t i2f(uint32_t input) +{ + u32 result, i, exponent, fraction; + + if ((input & 0x3fff) == 0) + result = 0; /* 0 is a special case */ + else { + exponent = 140; /* exponent biased by 127; */ + fraction = (input & 0x3fff) << 10; /* cheat and only + handle numbers below 2^^15 */ + for (i = 0; i < 14; i++) { + if (fraction & 0x800000) + break; + else { + fraction = fraction << 1; /* keep + shifting left until top bit = 1 */ + exponent = exponent - 1; + } + } + result = exponent << 23 | (fraction & 0x7fffff); /* mask + off top bit; assumed 1 */ + } + return result; +} + +int evergreen_blit_init(struct radeon_device *rdev) +{ + u32 obj_size; + int i, r, dwords; + void *ptr; + u32 packet2s[16]; + int num_packet2s = 0; + + /* pin copy shader into vram if already initialized */ + if (rdev->r600_blit.shader_obj) + goto done; + + mutex_init(&rdev->r600_blit.mutex); + rdev->r600_blit.state_offset = 0; + + if (rdev->family < CHIP_CAYMAN) + rdev->r600_blit.state_len = evergreen_default_size; + else + rdev->r600_blit.state_len = cayman_default_size; + + dwords = rdev->r600_blit.state_len; + while (dwords & 0xf) { + packet2s[num_packet2s++] = cpu_to_le32(PACKET2(0)); + dwords++; + } + + obj_size = dwords * 4; + obj_size = ALIGN(obj_size, 256); + + rdev->r600_blit.vs_offset = obj_size; + if (rdev->family < CHIP_CAYMAN) + obj_size += evergreen_vs_size * 4; + else + obj_size += cayman_vs_size * 4; + obj_size = ALIGN(obj_size, 256); + + rdev->r600_blit.ps_offset = obj_size; + if (rdev->family < CHIP_CAYMAN) + obj_size += evergreen_ps_size * 4; + else + obj_size += cayman_ps_size * 4; + obj_size = ALIGN(obj_size, 256); + + r = radeon_bo_create(rdev, obj_size, PAGE_SIZE, true, RADEON_GEM_DOMAIN_VRAM, + &rdev->r600_blit.shader_obj); + if (r) { + DRM_ERROR("evergreen failed to allocate shader\n"); + return r; + } + + DRM_DEBUG("evergreen blit allocated bo %08x vs %08x ps %08x\n", + obj_size, + rdev->r600_blit.vs_offset, rdev->r600_blit.ps_offset); + + r = radeon_bo_reserve(rdev->r600_blit.shader_obj, false); + if (unlikely(r != 0)) + return r; + r = radeon_bo_kmap(rdev->r600_blit.shader_obj, &ptr); + if (r) { + DRM_ERROR("failed to map blit object %d\n", r); + return r; + } + + if (rdev->family < CHIP_CAYMAN) { + memcpy_toio(ptr + rdev->r600_blit.state_offset, + evergreen_default_state, rdev->r600_blit.state_len * 4); + + if (num_packet2s) + memcpy_toio(ptr + rdev->r600_blit.state_offset + (rdev->r600_blit.state_len * 4), + packet2s, num_packet2s * 4); + for (i = 0; i < evergreen_vs_size; i++) + *(u32 *)((unsigned long)ptr + rdev->r600_blit.vs_offset + i * 4) = cpu_to_le32(evergreen_vs[i]); + for (i = 0; i < evergreen_ps_size; i++) + *(u32 *)((unsigned long)ptr + rdev->r600_blit.ps_offset + i * 4) = cpu_to_le32(evergreen_ps[i]); + } else { + memcpy_toio(ptr + rdev->r600_blit.state_offset, + cayman_default_state, rdev->r600_blit.state_len * 4); + + if (num_packet2s) + memcpy_toio(ptr + rdev->r600_blit.state_offset + (rdev->r600_blit.state_len * 4), + packet2s, num_packet2s * 4); + for (i = 0; i < cayman_vs_size; i++) + *(u32 *)((unsigned long)ptr + rdev->r600_blit.vs_offset + i * 4) = cpu_to_le32(cayman_vs[i]); + for (i = 0; i < cayman_ps_size; i++) + *(u32 *)((unsigned long)ptr + rdev->r600_blit.ps_offset + i * 4) = cpu_to_le32(cayman_ps[i]); + } + radeon_bo_kunmap(rdev->r600_blit.shader_obj); + radeon_bo_unreserve(rdev->r600_blit.shader_obj); + +done: + r = radeon_bo_reserve(rdev->r600_blit.shader_obj, false); + if (unlikely(r != 0)) + return r; + r = radeon_bo_pin(rdev->r600_blit.shader_obj, RADEON_GEM_DOMAIN_VRAM, + &rdev->r600_blit.shader_gpu_addr); + radeon_bo_unreserve(rdev->r600_blit.shader_obj); + if (r) { + dev_err(rdev->dev, "(%d) pin blit object failed\n", r); + return r; + } + radeon_ttm_set_active_vram_size(rdev, rdev->mc.real_vram_size); + return 0; +} + +void evergreen_blit_fini(struct radeon_device *rdev) +{ + int r; + + radeon_ttm_set_active_vram_size(rdev, rdev->mc.visible_vram_size); + if (rdev->r600_blit.shader_obj == NULL) + return; + /* If we can't reserve the bo, unref should be enough to destroy + * it when it becomes idle. + */ + r = radeon_bo_reserve(rdev->r600_blit.shader_obj, false); + if (!r) { + radeon_bo_unpin(rdev->r600_blit.shader_obj); + radeon_bo_unreserve(rdev->r600_blit.shader_obj); + } + radeon_bo_unref(&rdev->r600_blit.shader_obj); +} + +static int evergreen_vb_ib_get(struct radeon_device *rdev) +{ + int r; + r = radeon_ib_get(rdev, &rdev->r600_blit.vb_ib); + if (r) { + DRM_ERROR("failed to get IB for vertex buffer\n"); + return r; + } + + rdev->r600_blit.vb_total = 64*1024; + rdev->r600_blit.vb_used = 0; + return 0; +} + +static void evergreen_vb_ib_put(struct radeon_device *rdev) +{ + radeon_fence_emit(rdev, rdev->r600_blit.vb_ib->fence); + radeon_ib_free(rdev, &rdev->r600_blit.vb_ib); +} + +int evergreen_blit_prepare_copy(struct radeon_device *rdev, int size_bytes) +{ + int r; + int ring_size, line_size; + int max_size; + /* loops of emits + fence emit possible */ + int dwords_per_loop = 74, num_loops; + + r = evergreen_vb_ib_get(rdev); + if (r) + return r; + + /* 8 bpp vs 32 bpp for xfer unit */ + if (size_bytes & 3) + line_size = 8192; + else + line_size = 8192 * 4; + + max_size = 8192 * line_size; + + /* major loops cover the max size transfer */ + num_loops = ((size_bytes + max_size) / max_size); + /* minor loops cover the extra non aligned bits */ + num_loops += ((size_bytes % line_size) ? 1 : 0); + /* calculate number of loops correctly */ + ring_size = num_loops * dwords_per_loop; + /* set default + shaders */ + ring_size += 55; /* shaders + def state */ + ring_size += 10; /* fence emit for VB IB */ + ring_size += 5; /* done copy */ + ring_size += 10; /* fence emit for done copy */ + r = radeon_ring_lock(rdev, ring_size); + if (r) + return r; + + set_default_state(rdev); /* 36 */ + set_shaders(rdev); /* 16 */ + return 0; +} + +void evergreen_blit_done_copy(struct radeon_device *rdev, struct radeon_fence *fence) +{ + int r; + + if (rdev->r600_blit.vb_ib) + evergreen_vb_ib_put(rdev); + + if (fence) + r = radeon_fence_emit(rdev, fence); + + radeon_ring_unlock_commit(rdev); +} + +void evergreen_kms_blit_copy(struct radeon_device *rdev, + u64 src_gpu_addr, u64 dst_gpu_addr, + int size_bytes) +{ + int max_bytes; + u64 vb_gpu_addr; + u32 *vb; + + DRM_DEBUG("emitting copy %16llx %16llx %d %d\n", src_gpu_addr, dst_gpu_addr, + size_bytes, rdev->r600_blit.vb_used); + vb = (u32 *)(rdev->r600_blit.vb_ib->ptr + rdev->r600_blit.vb_used); + if ((size_bytes & 3) || (src_gpu_addr & 3) || (dst_gpu_addr & 3)) { + max_bytes = 8192; + + while (size_bytes) { + int cur_size = size_bytes; + int src_x = src_gpu_addr & 255; + int dst_x = dst_gpu_addr & 255; + int h = 1; + src_gpu_addr = src_gpu_addr & ~255ULL; + dst_gpu_addr = dst_gpu_addr & ~255ULL; + + if (!src_x && !dst_x) { + h = (cur_size / max_bytes); + if (h > 8192) + h = 8192; + if (h == 0) + h = 1; + else + cur_size = max_bytes; + } else { + if (cur_size > max_bytes) + cur_size = max_bytes; + if (cur_size > (max_bytes - dst_x)) + cur_size = (max_bytes - dst_x); + if (cur_size > (max_bytes - src_x)) + cur_size = (max_bytes - src_x); + } + + if ((rdev->r600_blit.vb_used + 48) > rdev->r600_blit.vb_total) { + WARN_ON(1); + } + + vb[0] = i2f(dst_x); + vb[1] = 0; + vb[2] = i2f(src_x); + vb[3] = 0; + + vb[4] = i2f(dst_x); + vb[5] = i2f(h); + vb[6] = i2f(src_x); + vb[7] = i2f(h); + + vb[8] = i2f(dst_x + cur_size); + vb[9] = i2f(h); + vb[10] = i2f(src_x + cur_size); + vb[11] = i2f(h); + + /* src 10 */ + set_tex_resource(rdev, FMT_8, + src_x + cur_size, h, src_x + cur_size, + src_gpu_addr); + + /* 5 */ + cp_set_surface_sync(rdev, + PACKET3_TC_ACTION_ENA, (src_x + cur_size * h), src_gpu_addr); + + + /* dst 17 */ + set_render_target(rdev, COLOR_8, + dst_x + cur_size, h, + dst_gpu_addr); + + /* scissors 12 */ + set_scissors(rdev, dst_x, 0, dst_x + cur_size, h); + + /* 15 */ + vb_gpu_addr = rdev->r600_blit.vb_ib->gpu_addr + rdev->r600_blit.vb_used; + set_vtx_resource(rdev, vb_gpu_addr); + + /* draw 10 */ + draw_auto(rdev); + + /* 5 */ + cp_set_surface_sync(rdev, + PACKET3_CB_ACTION_ENA | PACKET3_CB0_DEST_BASE_ENA, + cur_size * h, dst_gpu_addr); + + vb += 12; + rdev->r600_blit.vb_used += 12 * 4; + + src_gpu_addr += cur_size * h; + dst_gpu_addr += cur_size * h; + size_bytes -= cur_size * h; + } + } else { + max_bytes = 8192 * 4; + + while (size_bytes) { + int cur_size = size_bytes; + int src_x = (src_gpu_addr & 255); + int dst_x = (dst_gpu_addr & 255); + int h = 1; + src_gpu_addr = src_gpu_addr & ~255ULL; + dst_gpu_addr = dst_gpu_addr & ~255ULL; + + if (!src_x && !dst_x) { + h = (cur_size / max_bytes); + if (h > 8192) + h = 8192; + if (h == 0) + h = 1; + else + cur_size = max_bytes; + } else { + if (cur_size > max_bytes) + cur_size = max_bytes; + if (cur_size > (max_bytes - dst_x)) + cur_size = (max_bytes - dst_x); + if (cur_size > (max_bytes - src_x)) + cur_size = (max_bytes - src_x); + } + + if ((rdev->r600_blit.vb_used + 48) > rdev->r600_blit.vb_total) { + WARN_ON(1); + } + + vb[0] = i2f(dst_x / 4); + vb[1] = 0; + vb[2] = i2f(src_x / 4); + vb[3] = 0; + + vb[4] = i2f(dst_x / 4); + vb[5] = i2f(h); + vb[6] = i2f(src_x / 4); + vb[7] = i2f(h); + + vb[8] = i2f((dst_x + cur_size) / 4); + vb[9] = i2f(h); + vb[10] = i2f((src_x + cur_size) / 4); + vb[11] = i2f(h); + + /* src 10 */ + set_tex_resource(rdev, FMT_8_8_8_8, + (src_x + cur_size) / 4, + h, (src_x + cur_size) / 4, + src_gpu_addr); + /* 5 */ + cp_set_surface_sync(rdev, + PACKET3_TC_ACTION_ENA, (src_x + cur_size * h), src_gpu_addr); + + /* dst 17 */ + set_render_target(rdev, COLOR_8_8_8_8, + (dst_x + cur_size) / 4, h, + dst_gpu_addr); + + /* scissors 12 */ + set_scissors(rdev, (dst_x / 4), 0, (dst_x + cur_size / 4), h); + + /* Vertex buffer setup 15 */ + vb_gpu_addr = rdev->r600_blit.vb_ib->gpu_addr + rdev->r600_blit.vb_used; + set_vtx_resource(rdev, vb_gpu_addr); + + /* draw 10 */ + draw_auto(rdev); + + /* 5 */ + cp_set_surface_sync(rdev, + PACKET3_CB_ACTION_ENA | PACKET3_CB0_DEST_BASE_ENA, + cur_size * h, dst_gpu_addr); + + /* 74 ring dwords per loop */ + vb += 12; + rdev->r600_blit.vb_used += 12 * 4; + + src_gpu_addr += cur_size * h; + dst_gpu_addr += cur_size * h; + size_bytes -= cur_size * h; + } + } +} + diff --git a/drivers/gpu/drm/radeon/evergreen_blit_shaders.c b/drivers/gpu/drm/radeon/evergreen_blit_shaders.c new file mode 100644 index 00000000..3a10399e --- /dev/null +++ b/drivers/gpu/drm/radeon/evergreen_blit_shaders.c @@ -0,0 +1,356 @@ +/* + * Copyright 2010 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Alex Deucher + */ + +#include +#include + +/* + * evergreen cards need to use the 3D engine to blit data which requires + * quite a bit of hw state setup. Rather than pull the whole 3D driver + * (which normally generates the 3D state) into the DRM, we opt to use + * statically generated state tables. The regsiter state and shaders + * were hand generated to support blitting functionality. See the 3D + * driver or documentation for descriptions of the registers and + * shader instructions. + */ + +const u32 evergreen_default_state[] = +{ + 0xc0016900, + 0x0000023b, + 0x00000000, /* SQ_LDS_ALLOC_PS */ + + 0xc0066900, + 0x00000240, + 0x00000000, /* SQ_ESGS_RING_ITEMSIZE */ + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + + 0xc0046900, + 0x00000247, + 0x00000000, /* SQ_GS_VERT_ITEMSIZE */ + 0x00000000, + 0x00000000, + 0x00000000, + + 0xc0026900, + 0x00000010, + 0x00000000, /* DB_Z_INFO */ + 0x00000000, /* DB_STENCIL_INFO */ + + 0xc0016900, + 0x00000200, + 0x00000000, /* DB_DEPTH_CONTROL */ + + 0xc0066900, + 0x00000000, + 0x00000060, /* DB_RENDER_CONTROL */ + 0x00000000, /* DB_COUNT_CONTROL */ + 0x00000000, /* DB_DEPTH_VIEW */ + 0x0000002a, /* DB_RENDER_OVERRIDE */ + 0x00000000, /* DB_RENDER_OVERRIDE2 */ + 0x00000000, /* DB_HTILE_DATA_BASE */ + + 0xc0026900, + 0x0000000a, + 0x00000000, /* DB_STENCIL_CLEAR */ + 0x00000000, /* DB_DEPTH_CLEAR */ + + 0xc0016900, + 0x000002dc, + 0x0000aa00, /* DB_ALPHA_TO_MASK */ + + 0xc0016900, + 0x00000080, + 0x00000000, /* PA_SC_WINDOW_OFFSET */ + + 0xc00d6900, + 0x00000083, + 0x0000ffff, /* PA_SC_CLIPRECT_RULE */ + 0x00000000, /* PA_SC_CLIPRECT_0_TL */ + 0x20002000, /* PA_SC_CLIPRECT_0_BR */ + 0x00000000, + 0x20002000, + 0x00000000, + 0x20002000, + 0x00000000, + 0x20002000, + 0xaaaaaaaa, /* PA_SC_EDGERULE */ + 0x00000000, /* PA_SU_HARDWARE_SCREEN_OFFSET */ + 0x0000000f, /* CB_TARGET_MASK */ + 0x0000000f, /* CB_SHADER_MASK */ + + 0xc0226900, + 0x00000094, + 0x80000000, /* PA_SC_VPORT_SCISSOR_0_TL */ + 0x20002000, /* PA_SC_VPORT_SCISSOR_0_BR */ + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x00000000, /* PA_SC_VPORT_ZMIN_0 */ + 0x3f800000, /* PA_SC_VPORT_ZMAX_0 */ + + 0xc0016900, + 0x000000d4, + 0x00000000, /* SX_MISC */ + + 0xc0026900, + 0x00000292, + 0x00000000, /* PA_SC_MODE_CNTL_0 */ + 0x00000000, /* PA_SC_MODE_CNTL_1 */ + + 0xc0106900, + 0x00000300, + 0x00000000, /* PA_SC_LINE_CNTL */ + 0x00000000, /* PA_SC_AA_CONFIG */ + 0x00000005, /* PA_SU_VTX_CNTL */ + 0x3f800000, /* PA_CL_GB_VERT_CLIP_ADJ */ + 0x3f800000, /* PA_CL_GB_VERT_DISC_ADJ */ + 0x3f800000, /* PA_CL_GB_HORZ_CLIP_ADJ */ + 0x3f800000, /* PA_CL_GB_HORZ_DISC_ADJ */ + 0x00000000, /* PA_SC_AA_SAMPLE_LOCS_0 */ + 0x00000000, /* */ + 0x00000000, /* */ + 0x00000000, /* */ + 0x00000000, /* */ + 0x00000000, /* */ + 0x00000000, /* */ + 0x00000000, /* PA_SC_AA_SAMPLE_LOCS_7 */ + 0xffffffff, /* PA_SC_AA_MASK */ + + 0xc00d6900, + 0x00000202, + 0x00cc0010, /* CB_COLOR_CONTROL */ + 0x00000210, /* DB_SHADER_CONTROL */ + 0x00010000, /* PA_CL_CLIP_CNTL */ + 0x00000004, /* PA_SU_SC_MODE_CNTL */ + 0x00000100, /* PA_CL_VTE_CNTL */ + 0x00000000, /* PA_CL_VS_OUT_CNTL */ + 0x00000000, /* PA_CL_NANINF_CNTL */ + 0x00000000, /* PA_SU_LINE_STIPPLE_CNTL */ + 0x00000000, /* PA_SU_LINE_STIPPLE_SCALE */ + 0x00000000, /* PA_SU_PRIM_FILTER_CNTL */ + 0x00000000, /* */ + 0x00000000, /* */ + 0x00000000, /* SQ_DYN_GPR_RESOURCE_LIMIT_1 */ + + 0xc0066900, + 0x000002de, + 0x00000000, /* PA_SU_POLY_OFFSET_DB_FMT_CNTL */ + 0x00000000, /* */ + 0x00000000, /* */ + 0x00000000, /* */ + 0x00000000, /* */ + 0x00000000, /* */ + + 0xc0016900, + 0x00000229, + 0x00000000, /* SQ_PGM_START_FS */ + + 0xc0016900, + 0x0000022a, + 0x00000000, /* SQ_PGM_RESOURCES_FS */ + + 0xc0096900, + 0x00000100, + 0x00ffffff, /* VGT_MAX_VTX_INDX */ + 0x00000000, /* */ + 0x00000000, /* */ + 0x00000000, /* */ + 0x00000000, /* SX_ALPHA_TEST_CONTROL */ + 0x00000000, /* CB_BLEND_RED */ + 0x00000000, /* CB_BLEND_GREEN */ + 0x00000000, /* CB_BLEND_BLUE */ + 0x00000000, /* CB_BLEND_ALPHA */ + + 0xc0026900, + 0x000002a8, + 0x00000000, /* VGT_INSTANCE_STEP_RATE_0 */ + 0x00000000, /* */ + + 0xc0026900, + 0x000002ad, + 0x00000000, /* VGT_REUSE_OFF */ + 0x00000000, /* */ + + 0xc0116900, + 0x00000280, + 0x00000000, /* PA_SU_POINT_SIZE */ + 0x00000000, /* PA_SU_POINT_MINMAX */ + 0x00000008, /* PA_SU_LINE_CNTL */ + 0x00000000, /* PA_SC_LINE_STIPPLE */ + 0x00000000, /* VGT_OUTPUT_PATH_CNTL */ + 0x00000000, /* VGT_HOS_CNTL */ + 0x00000000, /* */ + 0x00000000, /* */ + 0x00000000, /* */ + 0x00000000, /* */ + 0x00000000, /* */ + 0x00000000, /* */ + 0x00000000, /* */ + 0x00000000, /* */ + 0x00000000, /* */ + 0x00000000, /* */ + 0x00000000, /* VGT_GS_MODE */ + + 0xc0016900, + 0x000002a1, + 0x00000000, /* VGT_PRIMITIVEID_EN */ + + 0xc0016900, + 0x000002a5, + 0x00000000, /* VGT_MULTI_PRIM_IB_RESET_EN */ + + 0xc0016900, + 0x000002d5, + 0x00000000, /* VGT_SHADER_STAGES_EN */ + + 0xc0026900, + 0x000002e5, + 0x00000000, /* VGT_STRMOUT_CONFIG */ + 0x00000000, /* */ + + 0xc0016900, + 0x000001e0, + 0x00000000, /* CB_BLEND0_CONTROL */ + + 0xc0016900, + 0x000001b1, + 0x00000000, /* SPI_VS_OUT_CONFIG */ + + 0xc0016900, + 0x00000187, + 0x00000000, /* SPI_VS_OUT_ID_0 */ + + 0xc0016900, + 0x00000191, + 0x00000100, /* SPI_PS_INPUT_CNTL_0 */ + + 0xc00b6900, + 0x000001b3, + 0x20000001, /* SPI_PS_IN_CONTROL_0 */ + 0x00000000, /* SPI_PS_IN_CONTROL_1 */ + 0x00000000, /* SPI_INTERP_CONTROL_0 */ + 0x00000000, /* SPI_INPUT_Z */ + 0x00000000, /* SPI_FOG_CNTL */ + 0x00100000, /* SPI_BARYC_CNTL */ + 0x00000000, /* SPI_PS_IN_CONTROL_2 */ + 0x00000000, /* */ + 0x00000000, /* */ + 0x00000000, /* */ + 0x00000000, /* */ + + 0xc0026900, + 0x00000316, + 0x0000000e, /* VGT_VERTEX_REUSE_BLOCK_CNTL */ + 0x00000010, /* */ +}; + +const u32 evergreen_vs[] = +{ + 0x00000004, + 0x80800400, + 0x0000a03c, + 0x95000688, + 0x00004000, + 0x15200688, + 0x00000000, + 0x00000000, + 0x3c000000, + 0x67961001, +#ifdef __BIG_ENDIAN + 0x000a0000, +#else + 0x00080000, +#endif + 0x00000000, + 0x1c000000, + 0x67961000, +#ifdef __BIG_ENDIAN + 0x00020008, +#else + 0x00000008, +#endif + 0x00000000, +}; + +const u32 evergreen_ps[] = +{ + 0x00000003, + 0xa00c0000, + 0x00000008, + 0x80400000, + 0x00000000, + 0x95200688, + 0x00380400, + 0x00146b10, + 0x00380000, + 0x20146b10, + 0x00380400, + 0x40146b00, + 0x80380000, + 0x60146b00, + 0x00000000, + 0x00000000, + 0x00000010, + 0x000d1000, + 0xb0800000, + 0x00000000, +}; + +const u32 evergreen_ps_size = ARRAY_SIZE(evergreen_ps); +const u32 evergreen_vs_size = ARRAY_SIZE(evergreen_vs); +const u32 evergreen_default_size = ARRAY_SIZE(evergreen_default_state); diff --git a/drivers/gpu/drm/radeon/evergreen_blit_shaders.h b/drivers/gpu/drm/radeon/evergreen_blit_shaders.h new file mode 100644 index 00000000..bb8d6c75 --- /dev/null +++ b/drivers/gpu/drm/radeon/evergreen_blit_shaders.h @@ -0,0 +1,35 @@ +/* + * Copyright 2009 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef EVERGREEN_BLIT_SHADERS_H +#define EVERGREEN_BLIT_SHADERS_H + +extern const u32 evergreen_ps[]; +extern const u32 evergreen_vs[]; +extern const u32 evergreen_default_state[]; + +extern const u32 evergreen_ps_size, evergreen_vs_size; +extern const u32 evergreen_default_size; + +#endif diff --git a/drivers/gpu/drm/radeon/evergreen_cs.c b/drivers/gpu/drm/radeon/evergreen_cs.c new file mode 100644 index 00000000..23d36417 --- /dev/null +++ b/drivers/gpu/drm/radeon/evergreen_cs.c @@ -0,0 +1,1442 @@ +/* + * Copyright 2010 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#include "drmP.h" +#include "radeon.h" +#include "evergreend.h" +#include "evergreen_reg_safe.h" +#include "cayman_reg_safe.h" + +static int evergreen_cs_packet_next_reloc(struct radeon_cs_parser *p, + struct radeon_cs_reloc **cs_reloc); + +struct evergreen_cs_track { + u32 group_size; + u32 nbanks; + u32 npipes; + /* value we track */ + u32 nsamples; + u32 cb_color_base_last[12]; + struct radeon_bo *cb_color_bo[12]; + u32 cb_color_bo_offset[12]; + struct radeon_bo *cb_color_fmask_bo[8]; + struct radeon_bo *cb_color_cmask_bo[8]; + u32 cb_color_info[12]; + u32 cb_color_view[12]; + u32 cb_color_pitch_idx[12]; + u32 cb_color_slice_idx[12]; + u32 cb_color_dim_idx[12]; + u32 cb_color_dim[12]; + u32 cb_color_pitch[12]; + u32 cb_color_slice[12]; + u32 cb_color_cmask_slice[8]; + u32 cb_color_fmask_slice[8]; + u32 cb_target_mask; + u32 cb_shader_mask; + u32 vgt_strmout_config; + u32 vgt_strmout_buffer_config; + u32 db_depth_control; + u32 db_depth_view; + u32 db_depth_size; + u32 db_depth_size_idx; + u32 db_z_info; + u32 db_z_idx; + u32 db_z_read_offset; + u32 db_z_write_offset; + struct radeon_bo *db_z_read_bo; + struct radeon_bo *db_z_write_bo; + u32 db_s_info; + u32 db_s_idx; + u32 db_s_read_offset; + u32 db_s_write_offset; + struct radeon_bo *db_s_read_bo; + struct radeon_bo *db_s_write_bo; +}; + +static void evergreen_cs_track_init(struct evergreen_cs_track *track) +{ + int i; + + for (i = 0; i < 8; i++) { + track->cb_color_fmask_bo[i] = NULL; + track->cb_color_cmask_bo[i] = NULL; + track->cb_color_cmask_slice[i] = 0; + track->cb_color_fmask_slice[i] = 0; + } + + for (i = 0; i < 12; i++) { + track->cb_color_base_last[i] = 0; + track->cb_color_bo[i] = NULL; + track->cb_color_bo_offset[i] = 0xFFFFFFFF; + track->cb_color_info[i] = 0; + track->cb_color_view[i] = 0; + track->cb_color_pitch_idx[i] = 0; + track->cb_color_slice_idx[i] = 0; + track->cb_color_dim[i] = 0; + track->cb_color_pitch[i] = 0; + track->cb_color_slice[i] = 0; + track->cb_color_dim[i] = 0; + } + track->cb_target_mask = 0xFFFFFFFF; + track->cb_shader_mask = 0xFFFFFFFF; + + track->db_depth_view = 0xFFFFC000; + track->db_depth_size = 0xFFFFFFFF; + track->db_depth_size_idx = 0; + track->db_depth_control = 0xFFFFFFFF; + track->db_z_info = 0xFFFFFFFF; + track->db_z_idx = 0xFFFFFFFF; + track->db_z_read_offset = 0xFFFFFFFF; + track->db_z_write_offset = 0xFFFFFFFF; + track->db_z_read_bo = NULL; + track->db_z_write_bo = NULL; + track->db_s_info = 0xFFFFFFFF; + track->db_s_idx = 0xFFFFFFFF; + track->db_s_read_offset = 0xFFFFFFFF; + track->db_s_write_offset = 0xFFFFFFFF; + track->db_s_read_bo = NULL; + track->db_s_write_bo = NULL; +} + +static inline int evergreen_cs_track_validate_cb(struct radeon_cs_parser *p, int i) +{ + /* XXX fill in */ + return 0; +} + +static int evergreen_cs_track_check(struct radeon_cs_parser *p) +{ + struct evergreen_cs_track *track = p->track; + + /* we don't support stream out buffer yet */ + if (track->vgt_strmout_config || track->vgt_strmout_buffer_config) { + dev_warn(p->dev, "this kernel doesn't support SMX output buffer\n"); + return -EINVAL; + } + + /* XXX fill in */ + return 0; +} + +/** + * evergreen_cs_packet_parse() - parse cp packet and point ib index to next packet + * @parser: parser structure holding parsing context. + * @pkt: where to store packet informations + * + * Assume that chunk_ib_index is properly set. Will return -EINVAL + * if packet is bigger than remaining ib size. or if packets is unknown. + **/ +int evergreen_cs_packet_parse(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt, + unsigned idx) +{ + struct radeon_cs_chunk *ib_chunk = &p->chunks[p->chunk_ib_idx]; + uint32_t header; + + if (idx >= ib_chunk->length_dw) { + DRM_ERROR("Can not parse packet at %d after CS end %d !\n", + idx, ib_chunk->length_dw); + return -EINVAL; + } + header = radeon_get_ib_value(p, idx); + pkt->idx = idx; + pkt->type = CP_PACKET_GET_TYPE(header); + pkt->count = CP_PACKET_GET_COUNT(header); + pkt->one_reg_wr = 0; + switch (pkt->type) { + case PACKET_TYPE0: + pkt->reg = CP_PACKET0_GET_REG(header); + break; + case PACKET_TYPE3: + pkt->opcode = CP_PACKET3_GET_OPCODE(header); + break; + case PACKET_TYPE2: + pkt->count = -1; + break; + default: + DRM_ERROR("Unknown packet type %d at %d !\n", pkt->type, idx); + return -EINVAL; + } + if ((pkt->count + 1 + pkt->idx) >= ib_chunk->length_dw) { + DRM_ERROR("Packet (%d:%d:%d) end after CS buffer (%d) !\n", + pkt->idx, pkt->type, pkt->count, ib_chunk->length_dw); + return -EINVAL; + } + return 0; +} + +/** + * evergreen_cs_packet_next_reloc() - parse next packet which should be reloc packet3 + * @parser: parser structure holding parsing context. + * @data: pointer to relocation data + * @offset_start: starting offset + * @offset_mask: offset mask (to align start offset on) + * @reloc: reloc informations + * + * Check next packet is relocation packet3, do bo validation and compute + * GPU offset using the provided start. + **/ +static int evergreen_cs_packet_next_reloc(struct radeon_cs_parser *p, + struct radeon_cs_reloc **cs_reloc) +{ + struct radeon_cs_chunk *relocs_chunk; + struct radeon_cs_packet p3reloc; + unsigned idx; + int r; + + if (p->chunk_relocs_idx == -1) { + DRM_ERROR("No relocation chunk !\n"); + return -EINVAL; + } + *cs_reloc = NULL; + relocs_chunk = &p->chunks[p->chunk_relocs_idx]; + r = evergreen_cs_packet_parse(p, &p3reloc, p->idx); + if (r) { + return r; + } + p->idx += p3reloc.count + 2; + if (p3reloc.type != PACKET_TYPE3 || p3reloc.opcode != PACKET3_NOP) { + DRM_ERROR("No packet3 for relocation for packet at %d.\n", + p3reloc.idx); + return -EINVAL; + } + idx = radeon_get_ib_value(p, p3reloc.idx + 1); + if (idx >= relocs_chunk->length_dw) { + DRM_ERROR("Relocs at %d after relocations chunk end %d !\n", + idx, relocs_chunk->length_dw); + return -EINVAL; + } + /* FIXME: we assume reloc size is 4 dwords */ + *cs_reloc = p->relocs_ptr[(idx / 4)]; + return 0; +} + +/** + * evergreen_cs_packet_next_is_pkt3_nop() - test if next packet is packet3 nop for reloc + * @parser: parser structure holding parsing context. + * + * Check next packet is relocation packet3, do bo validation and compute + * GPU offset using the provided start. + **/ +static inline int evergreen_cs_packet_next_is_pkt3_nop(struct radeon_cs_parser *p) +{ + struct radeon_cs_packet p3reloc; + int r; + + r = evergreen_cs_packet_parse(p, &p3reloc, p->idx); + if (r) { + return 0; + } + if (p3reloc.type != PACKET_TYPE3 || p3reloc.opcode != PACKET3_NOP) { + return 0; + } + return 1; +} + +/** + * evergreen_cs_packet_next_vline() - parse userspace VLINE packet + * @parser: parser structure holding parsing context. + * + * Userspace sends a special sequence for VLINE waits. + * PACKET0 - VLINE_START_END + value + * PACKET3 - WAIT_REG_MEM poll vline status reg + * RELOC (P3) - crtc_id in reloc. + * + * This function parses this and relocates the VLINE START END + * and WAIT_REG_MEM packets to the correct crtc. + * It also detects a switched off crtc and nulls out the + * wait in that case. + */ +static int evergreen_cs_packet_parse_vline(struct radeon_cs_parser *p) +{ + struct drm_mode_object *obj; + struct drm_crtc *crtc; + struct radeon_crtc *radeon_crtc; + struct radeon_cs_packet p3reloc, wait_reg_mem; + int crtc_id; + int r; + uint32_t header, h_idx, reg, wait_reg_mem_info; + volatile uint32_t *ib; + + ib = p->ib->ptr; + + /* parse the WAIT_REG_MEM */ + r = evergreen_cs_packet_parse(p, &wait_reg_mem, p->idx); + if (r) + return r; + + /* check its a WAIT_REG_MEM */ + if (wait_reg_mem.type != PACKET_TYPE3 || + wait_reg_mem.opcode != PACKET3_WAIT_REG_MEM) { + DRM_ERROR("vline wait missing WAIT_REG_MEM segment\n"); + return -EINVAL; + } + + wait_reg_mem_info = radeon_get_ib_value(p, wait_reg_mem.idx + 1); + /* bit 4 is reg (0) or mem (1) */ + if (wait_reg_mem_info & 0x10) { + DRM_ERROR("vline WAIT_REG_MEM waiting on MEM rather than REG\n"); + return -EINVAL; + } + /* waiting for value to be equal */ + if ((wait_reg_mem_info & 0x7) != 0x3) { + DRM_ERROR("vline WAIT_REG_MEM function not equal\n"); + return -EINVAL; + } + if ((radeon_get_ib_value(p, wait_reg_mem.idx + 2) << 2) != EVERGREEN_VLINE_STATUS) { + DRM_ERROR("vline WAIT_REG_MEM bad reg\n"); + return -EINVAL; + } + + if (radeon_get_ib_value(p, wait_reg_mem.idx + 5) != EVERGREEN_VLINE_STAT) { + DRM_ERROR("vline WAIT_REG_MEM bad bit mask\n"); + return -EINVAL; + } + + /* jump over the NOP */ + r = evergreen_cs_packet_parse(p, &p3reloc, p->idx + wait_reg_mem.count + 2); + if (r) + return r; + + h_idx = p->idx - 2; + p->idx += wait_reg_mem.count + 2; + p->idx += p3reloc.count + 2; + + header = radeon_get_ib_value(p, h_idx); + crtc_id = radeon_get_ib_value(p, h_idx + 2 + 7 + 1); + reg = CP_PACKET0_GET_REG(header); + obj = drm_mode_object_find(p->rdev->ddev, crtc_id, DRM_MODE_OBJECT_CRTC); + if (!obj) { + DRM_ERROR("cannot find crtc %d\n", crtc_id); + return -EINVAL; + } + crtc = obj_to_crtc(obj); + radeon_crtc = to_radeon_crtc(crtc); + crtc_id = radeon_crtc->crtc_id; + + if (!crtc->enabled) { + /* if the CRTC isn't enabled - we need to nop out the WAIT_REG_MEM */ + ib[h_idx + 2] = PACKET2(0); + ib[h_idx + 3] = PACKET2(0); + ib[h_idx + 4] = PACKET2(0); + ib[h_idx + 5] = PACKET2(0); + ib[h_idx + 6] = PACKET2(0); + ib[h_idx + 7] = PACKET2(0); + ib[h_idx + 8] = PACKET2(0); + } else { + switch (reg) { + case EVERGREEN_VLINE_START_END: + header &= ~R600_CP_PACKET0_REG_MASK; + header |= (EVERGREEN_VLINE_START_END + radeon_crtc->crtc_offset) >> 2; + ib[h_idx] = header; + ib[h_idx + 4] = (EVERGREEN_VLINE_STATUS + radeon_crtc->crtc_offset) >> 2; + break; + default: + DRM_ERROR("unknown crtc reloc\n"); + return -EINVAL; + } + } + return 0; +} + +static int evergreen_packet0_check(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt, + unsigned idx, unsigned reg) +{ + int r; + + switch (reg) { + case EVERGREEN_VLINE_START_END: + r = evergreen_cs_packet_parse_vline(p); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + return r; + } + break; + default: + printk(KERN_ERR "Forbidden register 0x%04X in cs at %d\n", + reg, idx); + return -EINVAL; + } + return 0; +} + +static int evergreen_cs_parse_packet0(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt) +{ + unsigned reg, i; + unsigned idx; + int r; + + idx = pkt->idx + 1; + reg = pkt->reg; + for (i = 0; i <= pkt->count; i++, idx++, reg += 4) { + r = evergreen_packet0_check(p, pkt, idx, reg); + if (r) { + return r; + } + } + return 0; +} + +/** + * evergreen_cs_check_reg() - check if register is authorized or not + * @parser: parser structure holding parsing context + * @reg: register we are testing + * @idx: index into the cs buffer + * + * This function will test against evergreen_reg_safe_bm and return 0 + * if register is safe. If register is not flag as safe this function + * will test it against a list of register needind special handling. + */ +static inline int evergreen_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) +{ + struct evergreen_cs_track *track = (struct evergreen_cs_track *)p->track; + struct radeon_cs_reloc *reloc; + u32 last_reg; + u32 m, i, tmp, *ib; + int r; + + if (p->rdev->family >= CHIP_CAYMAN) + last_reg = ARRAY_SIZE(cayman_reg_safe_bm); + else + last_reg = ARRAY_SIZE(evergreen_reg_safe_bm); + + i = (reg >> 7); + if (i > last_reg) { + dev_warn(p->dev, "forbidden register 0x%08x at %d\n", reg, idx); + return -EINVAL; + } + m = 1 << ((reg >> 2) & 31); + if (p->rdev->family >= CHIP_CAYMAN) { + if (!(cayman_reg_safe_bm[i] & m)) + return 0; + } else { + if (!(evergreen_reg_safe_bm[i] & m)) + return 0; + } + ib = p->ib->ptr; + switch (reg) { + /* force following reg to 0 in an attempt to disable out buffer + * which will need us to better understand how it works to perform + * security check on it (Jerome) + */ + case SQ_ESGS_RING_SIZE: + case SQ_GSVS_RING_SIZE: + case SQ_ESTMP_RING_SIZE: + case SQ_GSTMP_RING_SIZE: + case SQ_HSTMP_RING_SIZE: + case SQ_LSTMP_RING_SIZE: + case SQ_PSTMP_RING_SIZE: + case SQ_VSTMP_RING_SIZE: + case SQ_ESGS_RING_ITEMSIZE: + case SQ_ESTMP_RING_ITEMSIZE: + case SQ_GSTMP_RING_ITEMSIZE: + case SQ_GSVS_RING_ITEMSIZE: + case SQ_GS_VERT_ITEMSIZE: + case SQ_GS_VERT_ITEMSIZE_1: + case SQ_GS_VERT_ITEMSIZE_2: + case SQ_GS_VERT_ITEMSIZE_3: + case SQ_GSVS_RING_OFFSET_1: + case SQ_GSVS_RING_OFFSET_2: + case SQ_GSVS_RING_OFFSET_3: + case SQ_HSTMP_RING_ITEMSIZE: + case SQ_LSTMP_RING_ITEMSIZE: + case SQ_PSTMP_RING_ITEMSIZE: + case SQ_VSTMP_RING_ITEMSIZE: + case VGT_TF_RING_SIZE: + /* get value to populate the IB don't remove */ + /*tmp =radeon_get_ib_value(p, idx); + ib[idx] = 0;*/ + break; + case SQ_ESGS_RING_BASE: + case SQ_GSVS_RING_BASE: + case SQ_ESTMP_RING_BASE: + case SQ_GSTMP_RING_BASE: + case SQ_HSTMP_RING_BASE: + case SQ_LSTMP_RING_BASE: + case SQ_PSTMP_RING_BASE: + case SQ_VSTMP_RING_BASE: + r = evergreen_cs_packet_next_reloc(p, &reloc); + if (r) { + dev_warn(p->dev, "bad SET_CONTEXT_REG " + "0x%04X\n", reg); + return -EINVAL; + } + ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + break; + case DB_DEPTH_CONTROL: + track->db_depth_control = radeon_get_ib_value(p, idx); + break; + case CAYMAN_DB_EQAA: + if (p->rdev->family < CHIP_CAYMAN) { + dev_warn(p->dev, "bad SET_CONTEXT_REG " + "0x%04X\n", reg); + return -EINVAL; + } + break; + case CAYMAN_DB_DEPTH_INFO: + if (p->rdev->family < CHIP_CAYMAN) { + dev_warn(p->dev, "bad SET_CONTEXT_REG " + "0x%04X\n", reg); + return -EINVAL; + } + break; + case DB_Z_INFO: + r = evergreen_cs_packet_next_reloc(p, &reloc); + if (r) { + dev_warn(p->dev, "bad SET_CONTEXT_REG " + "0x%04X\n", reg); + return -EINVAL; + } + track->db_z_info = radeon_get_ib_value(p, idx); + ib[idx] &= ~Z_ARRAY_MODE(0xf); + track->db_z_info &= ~Z_ARRAY_MODE(0xf); + if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) { + ib[idx] |= Z_ARRAY_MODE(ARRAY_2D_TILED_THIN1); + track->db_z_info |= Z_ARRAY_MODE(ARRAY_2D_TILED_THIN1); + } else { + ib[idx] |= Z_ARRAY_MODE(ARRAY_1D_TILED_THIN1); + track->db_z_info |= Z_ARRAY_MODE(ARRAY_1D_TILED_THIN1); + } + break; + case DB_STENCIL_INFO: + track->db_s_info = radeon_get_ib_value(p, idx); + break; + case DB_DEPTH_VIEW: + track->db_depth_view = radeon_get_ib_value(p, idx); + break; + case DB_DEPTH_SIZE: + track->db_depth_size = radeon_get_ib_value(p, idx); + track->db_depth_size_idx = idx; + break; + case DB_Z_READ_BASE: + r = evergreen_cs_packet_next_reloc(p, &reloc); + if (r) { + dev_warn(p->dev, "bad SET_CONTEXT_REG " + "0x%04X\n", reg); + return -EINVAL; + } + track->db_z_read_offset = radeon_get_ib_value(p, idx); + ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + track->db_z_read_bo = reloc->robj; + break; + case DB_Z_WRITE_BASE: + r = evergreen_cs_packet_next_reloc(p, &reloc); + if (r) { + dev_warn(p->dev, "bad SET_CONTEXT_REG " + "0x%04X\n", reg); + return -EINVAL; + } + track->db_z_write_offset = radeon_get_ib_value(p, idx); + ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + track->db_z_write_bo = reloc->robj; + break; + case DB_STENCIL_READ_BASE: + r = evergreen_cs_packet_next_reloc(p, &reloc); + if (r) { + dev_warn(p->dev, "bad SET_CONTEXT_REG " + "0x%04X\n", reg); + return -EINVAL; + } + track->db_s_read_offset = radeon_get_ib_value(p, idx); + ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + track->db_s_read_bo = reloc->robj; + break; + case DB_STENCIL_WRITE_BASE: + r = evergreen_cs_packet_next_reloc(p, &reloc); + if (r) { + dev_warn(p->dev, "bad SET_CONTEXT_REG " + "0x%04X\n", reg); + return -EINVAL; + } + track->db_s_write_offset = radeon_get_ib_value(p, idx); + ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + track->db_s_write_bo = reloc->robj; + break; + case VGT_STRMOUT_CONFIG: + track->vgt_strmout_config = radeon_get_ib_value(p, idx); + break; + case VGT_STRMOUT_BUFFER_CONFIG: + track->vgt_strmout_buffer_config = radeon_get_ib_value(p, idx); + break; + case CB_TARGET_MASK: + track->cb_target_mask = radeon_get_ib_value(p, idx); + break; + case CB_SHADER_MASK: + track->cb_shader_mask = radeon_get_ib_value(p, idx); + break; + case PA_SC_AA_CONFIG: + if (p->rdev->family >= CHIP_CAYMAN) { + dev_warn(p->dev, "bad SET_CONTEXT_REG " + "0x%04X\n", reg); + return -EINVAL; + } + tmp = radeon_get_ib_value(p, idx) & MSAA_NUM_SAMPLES_MASK; + track->nsamples = 1 << tmp; + break; + case CAYMAN_PA_SC_AA_CONFIG: + if (p->rdev->family < CHIP_CAYMAN) { + dev_warn(p->dev, "bad SET_CONTEXT_REG " + "0x%04X\n", reg); + return -EINVAL; + } + tmp = radeon_get_ib_value(p, idx) & CAYMAN_MSAA_NUM_SAMPLES_MASK; + track->nsamples = 1 << tmp; + break; + case CB_COLOR0_VIEW: + case CB_COLOR1_VIEW: + case CB_COLOR2_VIEW: + case CB_COLOR3_VIEW: + case CB_COLOR4_VIEW: + case CB_COLOR5_VIEW: + case CB_COLOR6_VIEW: + case CB_COLOR7_VIEW: + tmp = (reg - CB_COLOR0_VIEW) / 0x3c; + track->cb_color_view[tmp] = radeon_get_ib_value(p, idx); + break; + case CB_COLOR8_VIEW: + case CB_COLOR9_VIEW: + case CB_COLOR10_VIEW: + case CB_COLOR11_VIEW: + tmp = ((reg - CB_COLOR8_VIEW) / 0x1c) + 8; + track->cb_color_view[tmp] = radeon_get_ib_value(p, idx); + break; + case CB_COLOR0_INFO: + case CB_COLOR1_INFO: + case CB_COLOR2_INFO: + case CB_COLOR3_INFO: + case CB_COLOR4_INFO: + case CB_COLOR5_INFO: + case CB_COLOR6_INFO: + case CB_COLOR7_INFO: + r = evergreen_cs_packet_next_reloc(p, &reloc); + if (r) { + dev_warn(p->dev, "bad SET_CONTEXT_REG " + "0x%04X\n", reg); + return -EINVAL; + } + tmp = (reg - CB_COLOR0_INFO) / 0x3c; + track->cb_color_info[tmp] = radeon_get_ib_value(p, idx); + if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) { + ib[idx] |= CB_ARRAY_MODE(ARRAY_2D_TILED_THIN1); + track->cb_color_info[tmp] |= CB_ARRAY_MODE(ARRAY_2D_TILED_THIN1); + } else if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) { + ib[idx] |= CB_ARRAY_MODE(ARRAY_1D_TILED_THIN1); + track->cb_color_info[tmp] |= CB_ARRAY_MODE(ARRAY_1D_TILED_THIN1); + } + break; + case CB_COLOR8_INFO: + case CB_COLOR9_INFO: + case CB_COLOR10_INFO: + case CB_COLOR11_INFO: + r = evergreen_cs_packet_next_reloc(p, &reloc); + if (r) { + dev_warn(p->dev, "bad SET_CONTEXT_REG " + "0x%04X\n", reg); + return -EINVAL; + } + tmp = ((reg - CB_COLOR8_INFO) / 0x1c) + 8; + track->cb_color_info[tmp] = radeon_get_ib_value(p, idx); + if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) { + ib[idx] |= CB_ARRAY_MODE(ARRAY_2D_TILED_THIN1); + track->cb_color_info[tmp] |= CB_ARRAY_MODE(ARRAY_2D_TILED_THIN1); + } else if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) { + ib[idx] |= CB_ARRAY_MODE(ARRAY_1D_TILED_THIN1); + track->cb_color_info[tmp] |= CB_ARRAY_MODE(ARRAY_1D_TILED_THIN1); + } + break; + case CB_COLOR0_PITCH: + case CB_COLOR1_PITCH: + case CB_COLOR2_PITCH: + case CB_COLOR3_PITCH: + case CB_COLOR4_PITCH: + case CB_COLOR5_PITCH: + case CB_COLOR6_PITCH: + case CB_COLOR7_PITCH: + tmp = (reg - CB_COLOR0_PITCH) / 0x3c; + track->cb_color_pitch[tmp] = radeon_get_ib_value(p, idx); + track->cb_color_pitch_idx[tmp] = idx; + break; + case CB_COLOR8_PITCH: + case CB_COLOR9_PITCH: + case CB_COLOR10_PITCH: + case CB_COLOR11_PITCH: + tmp = ((reg - CB_COLOR8_PITCH) / 0x1c) + 8; + track->cb_color_pitch[tmp] = radeon_get_ib_value(p, idx); + track->cb_color_pitch_idx[tmp] = idx; + break; + case CB_COLOR0_SLICE: + case CB_COLOR1_SLICE: + case CB_COLOR2_SLICE: + case CB_COLOR3_SLICE: + case CB_COLOR4_SLICE: + case CB_COLOR5_SLICE: + case CB_COLOR6_SLICE: + case CB_COLOR7_SLICE: + tmp = (reg - CB_COLOR0_SLICE) / 0x3c; + track->cb_color_slice[tmp] = radeon_get_ib_value(p, idx); + track->cb_color_slice_idx[tmp] = idx; + break; + case CB_COLOR8_SLICE: + case CB_COLOR9_SLICE: + case CB_COLOR10_SLICE: + case CB_COLOR11_SLICE: + tmp = ((reg - CB_COLOR8_SLICE) / 0x1c) + 8; + track->cb_color_slice[tmp] = radeon_get_ib_value(p, idx); + track->cb_color_slice_idx[tmp] = idx; + break; + case CB_COLOR0_ATTRIB: + case CB_COLOR1_ATTRIB: + case CB_COLOR2_ATTRIB: + case CB_COLOR3_ATTRIB: + case CB_COLOR4_ATTRIB: + case CB_COLOR5_ATTRIB: + case CB_COLOR6_ATTRIB: + case CB_COLOR7_ATTRIB: + case CB_COLOR8_ATTRIB: + case CB_COLOR9_ATTRIB: + case CB_COLOR10_ATTRIB: + case CB_COLOR11_ATTRIB: + break; + case CB_COLOR0_DIM: + case CB_COLOR1_DIM: + case CB_COLOR2_DIM: + case CB_COLOR3_DIM: + case CB_COLOR4_DIM: + case CB_COLOR5_DIM: + case CB_COLOR6_DIM: + case CB_COLOR7_DIM: + tmp = (reg - CB_COLOR0_DIM) / 0x3c; + track->cb_color_dim[tmp] = radeon_get_ib_value(p, idx); + track->cb_color_dim_idx[tmp] = idx; + break; + case CB_COLOR8_DIM: + case CB_COLOR9_DIM: + case CB_COLOR10_DIM: + case CB_COLOR11_DIM: + tmp = ((reg - CB_COLOR8_DIM) / 0x1c) + 8; + track->cb_color_dim[tmp] = radeon_get_ib_value(p, idx); + track->cb_color_dim_idx[tmp] = idx; + break; + case CB_COLOR0_FMASK: + case CB_COLOR1_FMASK: + case CB_COLOR2_FMASK: + case CB_COLOR3_FMASK: + case CB_COLOR4_FMASK: + case CB_COLOR5_FMASK: + case CB_COLOR6_FMASK: + case CB_COLOR7_FMASK: + tmp = (reg - CB_COLOR0_FMASK) / 0x3c; + r = evergreen_cs_packet_next_reloc(p, &reloc); + if (r) { + dev_err(p->dev, "bad SET_CONTEXT_REG 0x%04X\n", reg); + return -EINVAL; + } + ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + track->cb_color_fmask_bo[tmp] = reloc->robj; + break; + case CB_COLOR0_CMASK: + case CB_COLOR1_CMASK: + case CB_COLOR2_CMASK: + case CB_COLOR3_CMASK: + case CB_COLOR4_CMASK: + case CB_COLOR5_CMASK: + case CB_COLOR6_CMASK: + case CB_COLOR7_CMASK: + tmp = (reg - CB_COLOR0_CMASK) / 0x3c; + r = evergreen_cs_packet_next_reloc(p, &reloc); + if (r) { + dev_err(p->dev, "bad SET_CONTEXT_REG 0x%04X\n", reg); + return -EINVAL; + } + ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + track->cb_color_cmask_bo[tmp] = reloc->robj; + break; + case CB_COLOR0_FMASK_SLICE: + case CB_COLOR1_FMASK_SLICE: + case CB_COLOR2_FMASK_SLICE: + case CB_COLOR3_FMASK_SLICE: + case CB_COLOR4_FMASK_SLICE: + case CB_COLOR5_FMASK_SLICE: + case CB_COLOR6_FMASK_SLICE: + case CB_COLOR7_FMASK_SLICE: + tmp = (reg - CB_COLOR0_FMASK_SLICE) / 0x3c; + track->cb_color_fmask_slice[tmp] = radeon_get_ib_value(p, idx); + break; + case CB_COLOR0_CMASK_SLICE: + case CB_COLOR1_CMASK_SLICE: + case CB_COLOR2_CMASK_SLICE: + case CB_COLOR3_CMASK_SLICE: + case CB_COLOR4_CMASK_SLICE: + case CB_COLOR5_CMASK_SLICE: + case CB_COLOR6_CMASK_SLICE: + case CB_COLOR7_CMASK_SLICE: + tmp = (reg - CB_COLOR0_CMASK_SLICE) / 0x3c; + track->cb_color_cmask_slice[tmp] = radeon_get_ib_value(p, idx); + break; + case CB_COLOR0_BASE: + case CB_COLOR1_BASE: + case CB_COLOR2_BASE: + case CB_COLOR3_BASE: + case CB_COLOR4_BASE: + case CB_COLOR5_BASE: + case CB_COLOR6_BASE: + case CB_COLOR7_BASE: + r = evergreen_cs_packet_next_reloc(p, &reloc); + if (r) { + dev_warn(p->dev, "bad SET_CONTEXT_REG " + "0x%04X\n", reg); + return -EINVAL; + } + tmp = (reg - CB_COLOR0_BASE) / 0x3c; + track->cb_color_bo_offset[tmp] = radeon_get_ib_value(p, idx); + ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + track->cb_color_base_last[tmp] = ib[idx]; + track->cb_color_bo[tmp] = reloc->robj; + break; + case CB_COLOR8_BASE: + case CB_COLOR9_BASE: + case CB_COLOR10_BASE: + case CB_COLOR11_BASE: + r = evergreen_cs_packet_next_reloc(p, &reloc); + if (r) { + dev_warn(p->dev, "bad SET_CONTEXT_REG " + "0x%04X\n", reg); + return -EINVAL; + } + tmp = ((reg - CB_COLOR8_BASE) / 0x1c) + 8; + track->cb_color_bo_offset[tmp] = radeon_get_ib_value(p, idx); + ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + track->cb_color_base_last[tmp] = ib[idx]; + track->cb_color_bo[tmp] = reloc->robj; + break; + case CB_IMMED0_BASE: + case CB_IMMED1_BASE: + case CB_IMMED2_BASE: + case CB_IMMED3_BASE: + case CB_IMMED4_BASE: + case CB_IMMED5_BASE: + case CB_IMMED6_BASE: + case CB_IMMED7_BASE: + case CB_IMMED8_BASE: + case CB_IMMED9_BASE: + case CB_IMMED10_BASE: + case CB_IMMED11_BASE: + case DB_HTILE_DATA_BASE: + case SQ_PGM_START_FS: + case SQ_PGM_START_ES: + case SQ_PGM_START_VS: + case SQ_PGM_START_GS: + case SQ_PGM_START_PS: + case SQ_PGM_START_HS: + case SQ_PGM_START_LS: + case GDS_ADDR_BASE: + case SQ_CONST_MEM_BASE: + case SQ_ALU_CONST_CACHE_GS_0: + case SQ_ALU_CONST_CACHE_GS_1: + case SQ_ALU_CONST_CACHE_GS_2: + case SQ_ALU_CONST_CACHE_GS_3: + case SQ_ALU_CONST_CACHE_GS_4: + case SQ_ALU_CONST_CACHE_GS_5: + case SQ_ALU_CONST_CACHE_GS_6: + case SQ_ALU_CONST_CACHE_GS_7: + case SQ_ALU_CONST_CACHE_GS_8: + case SQ_ALU_CONST_CACHE_GS_9: + case SQ_ALU_CONST_CACHE_GS_10: + case SQ_ALU_CONST_CACHE_GS_11: + case SQ_ALU_CONST_CACHE_GS_12: + case SQ_ALU_CONST_CACHE_GS_13: + case SQ_ALU_CONST_CACHE_GS_14: + case SQ_ALU_CONST_CACHE_GS_15: + case SQ_ALU_CONST_CACHE_PS_0: + case SQ_ALU_CONST_CACHE_PS_1: + case SQ_ALU_CONST_CACHE_PS_2: + case SQ_ALU_CONST_CACHE_PS_3: + case SQ_ALU_CONST_CACHE_PS_4: + case SQ_ALU_CONST_CACHE_PS_5: + case SQ_ALU_CONST_CACHE_PS_6: + case SQ_ALU_CONST_CACHE_PS_7: + case SQ_ALU_CONST_CACHE_PS_8: + case SQ_ALU_CONST_CACHE_PS_9: + case SQ_ALU_CONST_CACHE_PS_10: + case SQ_ALU_CONST_CACHE_PS_11: + case SQ_ALU_CONST_CACHE_PS_12: + case SQ_ALU_CONST_CACHE_PS_13: + case SQ_ALU_CONST_CACHE_PS_14: + case SQ_ALU_CONST_CACHE_PS_15: + case SQ_ALU_CONST_CACHE_VS_0: + case SQ_ALU_CONST_CACHE_VS_1: + case SQ_ALU_CONST_CACHE_VS_2: + case SQ_ALU_CONST_CACHE_VS_3: + case SQ_ALU_CONST_CACHE_VS_4: + case SQ_ALU_CONST_CACHE_VS_5: + case SQ_ALU_CONST_CACHE_VS_6: + case SQ_ALU_CONST_CACHE_VS_7: + case SQ_ALU_CONST_CACHE_VS_8: + case SQ_ALU_CONST_CACHE_VS_9: + case SQ_ALU_CONST_CACHE_VS_10: + case SQ_ALU_CONST_CACHE_VS_11: + case SQ_ALU_CONST_CACHE_VS_12: + case SQ_ALU_CONST_CACHE_VS_13: + case SQ_ALU_CONST_CACHE_VS_14: + case SQ_ALU_CONST_CACHE_VS_15: + case SQ_ALU_CONST_CACHE_HS_0: + case SQ_ALU_CONST_CACHE_HS_1: + case SQ_ALU_CONST_CACHE_HS_2: + case SQ_ALU_CONST_CACHE_HS_3: + case SQ_ALU_CONST_CACHE_HS_4: + case SQ_ALU_CONST_CACHE_HS_5: + case SQ_ALU_CONST_CACHE_HS_6: + case SQ_ALU_CONST_CACHE_HS_7: + case SQ_ALU_CONST_CACHE_HS_8: + case SQ_ALU_CONST_CACHE_HS_9: + case SQ_ALU_CONST_CACHE_HS_10: + case SQ_ALU_CONST_CACHE_HS_11: + case SQ_ALU_CONST_CACHE_HS_12: + case SQ_ALU_CONST_CACHE_HS_13: + case SQ_ALU_CONST_CACHE_HS_14: + case SQ_ALU_CONST_CACHE_HS_15: + case SQ_ALU_CONST_CACHE_LS_0: + case SQ_ALU_CONST_CACHE_LS_1: + case SQ_ALU_CONST_CACHE_LS_2: + case SQ_ALU_CONST_CACHE_LS_3: + case SQ_ALU_CONST_CACHE_LS_4: + case SQ_ALU_CONST_CACHE_LS_5: + case SQ_ALU_CONST_CACHE_LS_6: + case SQ_ALU_CONST_CACHE_LS_7: + case SQ_ALU_CONST_CACHE_LS_8: + case SQ_ALU_CONST_CACHE_LS_9: + case SQ_ALU_CONST_CACHE_LS_10: + case SQ_ALU_CONST_CACHE_LS_11: + case SQ_ALU_CONST_CACHE_LS_12: + case SQ_ALU_CONST_CACHE_LS_13: + case SQ_ALU_CONST_CACHE_LS_14: + case SQ_ALU_CONST_CACHE_LS_15: + r = evergreen_cs_packet_next_reloc(p, &reloc); + if (r) { + dev_warn(p->dev, "bad SET_CONTEXT_REG " + "0x%04X\n", reg); + return -EINVAL; + } + ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + break; + default: + dev_warn(p->dev, "forbidden register 0x%08x at %d\n", reg, idx); + return -EINVAL; + } + return 0; +} + +/** + * evergreen_check_texture_resource() - check if register is authorized or not + * @p: parser structure holding parsing context + * @idx: index into the cs buffer + * @texture: texture's bo structure + * @mipmap: mipmap's bo structure + * + * This function will check that the resource has valid field and that + * the texture and mipmap bo object are big enough to cover this resource. + */ +static inline int evergreen_check_texture_resource(struct radeon_cs_parser *p, u32 idx, + struct radeon_bo *texture, + struct radeon_bo *mipmap) +{ + /* XXX fill in */ + return 0; +} + +static int evergreen_packet3_check(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt) +{ + struct radeon_cs_reloc *reloc; + struct evergreen_cs_track *track; + volatile u32 *ib; + unsigned idx; + unsigned i; + unsigned start_reg, end_reg, reg; + int r; + u32 idx_value; + + track = (struct evergreen_cs_track *)p->track; + ib = p->ib->ptr; + idx = pkt->idx + 1; + idx_value = radeon_get_ib_value(p, idx); + + switch (pkt->opcode) { + case PACKET3_SET_PREDICATION: + { + int pred_op; + int tmp; + if (pkt->count != 1) { + DRM_ERROR("bad SET PREDICATION\n"); + return -EINVAL; + } + + tmp = radeon_get_ib_value(p, idx + 1); + pred_op = (tmp >> 16) & 0x7; + + /* for the clear predicate operation */ + if (pred_op == 0) + return 0; + + if (pred_op > 2) { + DRM_ERROR("bad SET PREDICATION operation %d\n", pred_op); + return -EINVAL; + } + + r = evergreen_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("bad SET PREDICATION\n"); + return -EINVAL; + } + + ib[idx + 0] = idx_value + (u32)(reloc->lobj.gpu_offset & 0xffffffff); + ib[idx + 1] = tmp + (upper_32_bits(reloc->lobj.gpu_offset) & 0xff); + } + break; + case PACKET3_CONTEXT_CONTROL: + if (pkt->count != 1) { + DRM_ERROR("bad CONTEXT_CONTROL\n"); + return -EINVAL; + } + break; + case PACKET3_INDEX_TYPE: + case PACKET3_NUM_INSTANCES: + case PACKET3_CLEAR_STATE: + if (pkt->count) { + DRM_ERROR("bad INDEX_TYPE/NUM_INSTANCES/CLEAR_STATE\n"); + return -EINVAL; + } + break; + case CAYMAN_PACKET3_DEALLOC_STATE: + if (p->rdev->family < CHIP_CAYMAN) { + DRM_ERROR("bad PACKET3_DEALLOC_STATE\n"); + return -EINVAL; + } + if (pkt->count) { + DRM_ERROR("bad INDEX_TYPE/NUM_INSTANCES/CLEAR_STATE\n"); + return -EINVAL; + } + break; + case PACKET3_INDEX_BASE: + if (pkt->count != 1) { + DRM_ERROR("bad INDEX_BASE\n"); + return -EINVAL; + } + r = evergreen_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("bad INDEX_BASE\n"); + return -EINVAL; + } + ib[idx+0] = idx_value + (u32)(reloc->lobj.gpu_offset & 0xffffffff); + ib[idx+1] += upper_32_bits(reloc->lobj.gpu_offset) & 0xff; + r = evergreen_cs_track_check(p); + if (r) { + dev_warn(p->dev, "%s:%d invalid cmd stream\n", __func__, __LINE__); + return r; + } + break; + case PACKET3_DRAW_INDEX: + if (pkt->count != 3) { + DRM_ERROR("bad DRAW_INDEX\n"); + return -EINVAL; + } + r = evergreen_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("bad DRAW_INDEX\n"); + return -EINVAL; + } + ib[idx+0] = idx_value + (u32)(reloc->lobj.gpu_offset & 0xffffffff); + ib[idx+1] += upper_32_bits(reloc->lobj.gpu_offset) & 0xff; + r = evergreen_cs_track_check(p); + if (r) { + dev_warn(p->dev, "%s:%d invalid cmd stream\n", __func__, __LINE__); + return r; + } + break; + case PACKET3_DRAW_INDEX_2: + if (pkt->count != 4) { + DRM_ERROR("bad DRAW_INDEX_2\n"); + return -EINVAL; + } + r = evergreen_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("bad DRAW_INDEX_2\n"); + return -EINVAL; + } + ib[idx+1] = idx_value + (u32)(reloc->lobj.gpu_offset & 0xffffffff); + ib[idx+2] += upper_32_bits(reloc->lobj.gpu_offset) & 0xff; + r = evergreen_cs_track_check(p); + if (r) { + dev_warn(p->dev, "%s:%d invalid cmd stream\n", __func__, __LINE__); + return r; + } + break; + case PACKET3_DRAW_INDEX_AUTO: + if (pkt->count != 1) { + DRM_ERROR("bad DRAW_INDEX_AUTO\n"); + return -EINVAL; + } + r = evergreen_cs_track_check(p); + if (r) { + dev_warn(p->dev, "%s:%d invalid cmd stream %d\n", __func__, __LINE__, idx); + return r; + } + break; + case PACKET3_DRAW_INDEX_MULTI_AUTO: + if (pkt->count != 2) { + DRM_ERROR("bad DRAW_INDEX_MULTI_AUTO\n"); + return -EINVAL; + } + r = evergreen_cs_track_check(p); + if (r) { + dev_warn(p->dev, "%s:%d invalid cmd stream %d\n", __func__, __LINE__, idx); + return r; + } + break; + case PACKET3_DRAW_INDEX_IMMD: + if (pkt->count < 2) { + DRM_ERROR("bad DRAW_INDEX_IMMD\n"); + return -EINVAL; + } + r = evergreen_cs_track_check(p); + if (r) { + dev_warn(p->dev, "%s:%d invalid cmd stream\n", __func__, __LINE__); + return r; + } + break; + case PACKET3_DRAW_INDEX_OFFSET: + if (pkt->count != 2) { + DRM_ERROR("bad DRAW_INDEX_OFFSET\n"); + return -EINVAL; + } + r = evergreen_cs_track_check(p); + if (r) { + dev_warn(p->dev, "%s:%d invalid cmd stream\n", __func__, __LINE__); + return r; + } + break; + case PACKET3_DRAW_INDEX_OFFSET_2: + if (pkt->count != 3) { + DRM_ERROR("bad DRAW_INDEX_OFFSET_2\n"); + return -EINVAL; + } + r = evergreen_cs_track_check(p); + if (r) { + dev_warn(p->dev, "%s:%d invalid cmd stream\n", __func__, __LINE__); + return r; + } + break; + case PACKET3_WAIT_REG_MEM: + if (pkt->count != 5) { + DRM_ERROR("bad WAIT_REG_MEM\n"); + return -EINVAL; + } + /* bit 4 is reg (0) or mem (1) */ + if (idx_value & 0x10) { + r = evergreen_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("bad WAIT_REG_MEM\n"); + return -EINVAL; + } + ib[idx+1] += (u32)(reloc->lobj.gpu_offset & 0xffffffff); + ib[idx+2] += upper_32_bits(reloc->lobj.gpu_offset) & 0xff; + } + break; + case PACKET3_SURFACE_SYNC: + if (pkt->count != 3) { + DRM_ERROR("bad SURFACE_SYNC\n"); + return -EINVAL; + } + /* 0xffffffff/0x0 is flush all cache flag */ + if (radeon_get_ib_value(p, idx + 1) != 0xffffffff || + radeon_get_ib_value(p, idx + 2) != 0) { + r = evergreen_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("bad SURFACE_SYNC\n"); + return -EINVAL; + } + ib[idx+2] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + } + break; + case PACKET3_EVENT_WRITE: + if (pkt->count != 2 && pkt->count != 0) { + DRM_ERROR("bad EVENT_WRITE\n"); + return -EINVAL; + } + if (pkt->count) { + r = evergreen_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("bad EVENT_WRITE\n"); + return -EINVAL; + } + ib[idx+1] += (u32)(reloc->lobj.gpu_offset & 0xffffffff); + ib[idx+2] += upper_32_bits(reloc->lobj.gpu_offset) & 0xff; + } + break; + case PACKET3_EVENT_WRITE_EOP: + if (pkt->count != 4) { + DRM_ERROR("bad EVENT_WRITE_EOP\n"); + return -EINVAL; + } + r = evergreen_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("bad EVENT_WRITE_EOP\n"); + return -EINVAL; + } + ib[idx+1] += (u32)(reloc->lobj.gpu_offset & 0xffffffff); + ib[idx+2] += upper_32_bits(reloc->lobj.gpu_offset) & 0xff; + break; + case PACKET3_EVENT_WRITE_EOS: + if (pkt->count != 3) { + DRM_ERROR("bad EVENT_WRITE_EOS\n"); + return -EINVAL; + } + r = evergreen_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("bad EVENT_WRITE_EOS\n"); + return -EINVAL; + } + ib[idx+1] += (u32)(reloc->lobj.gpu_offset & 0xffffffff); + ib[idx+2] += upper_32_bits(reloc->lobj.gpu_offset) & 0xff; + break; + case PACKET3_SET_CONFIG_REG: + start_reg = (idx_value << 2) + PACKET3_SET_CONFIG_REG_START; + end_reg = 4 * pkt->count + start_reg - 4; + if ((start_reg < PACKET3_SET_CONFIG_REG_START) || + (start_reg >= PACKET3_SET_CONFIG_REG_END) || + (end_reg >= PACKET3_SET_CONFIG_REG_END)) { + DRM_ERROR("bad PACKET3_SET_CONFIG_REG\n"); + return -EINVAL; + } + for (i = 0; i < pkt->count; i++) { + reg = start_reg + (4 * i); + r = evergreen_cs_check_reg(p, reg, idx+1+i); + if (r) + return r; + } + break; + case PACKET3_SET_CONTEXT_REG: + start_reg = (idx_value << 2) + PACKET3_SET_CONTEXT_REG_START; + end_reg = 4 * pkt->count + start_reg - 4; + if ((start_reg < PACKET3_SET_CONTEXT_REG_START) || + (start_reg >= PACKET3_SET_CONTEXT_REG_END) || + (end_reg >= PACKET3_SET_CONTEXT_REG_END)) { + DRM_ERROR("bad PACKET3_SET_CONTEXT_REG\n"); + return -EINVAL; + } + for (i = 0; i < pkt->count; i++) { + reg = start_reg + (4 * i); + r = evergreen_cs_check_reg(p, reg, idx+1+i); + if (r) + return r; + } + break; + case PACKET3_SET_RESOURCE: + if (pkt->count % 8) { + DRM_ERROR("bad SET_RESOURCE\n"); + return -EINVAL; + } + start_reg = (idx_value << 2) + PACKET3_SET_RESOURCE_START; + end_reg = 4 * pkt->count + start_reg - 4; + if ((start_reg < PACKET3_SET_RESOURCE_START) || + (start_reg >= PACKET3_SET_RESOURCE_END) || + (end_reg >= PACKET3_SET_RESOURCE_END)) { + DRM_ERROR("bad SET_RESOURCE\n"); + return -EINVAL; + } + for (i = 0; i < (pkt->count / 8); i++) { + struct radeon_bo *texture, *mipmap; + u32 size, offset; + + switch (G__SQ_CONSTANT_TYPE(radeon_get_ib_value(p, idx+1+(i*8)+7))) { + case SQ_TEX_VTX_VALID_TEXTURE: + /* tex base */ + r = evergreen_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("bad SET_RESOURCE (tex)\n"); + return -EINVAL; + } + ib[idx+1+(i*8)+2] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) + ib[idx+1+(i*8)+1] |= TEX_ARRAY_MODE(ARRAY_2D_TILED_THIN1); + else if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) + ib[idx+1+(i*8)+1] |= TEX_ARRAY_MODE(ARRAY_1D_TILED_THIN1); + texture = reloc->robj; + /* tex mip base */ + r = evergreen_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("bad SET_RESOURCE (tex)\n"); + return -EINVAL; + } + ib[idx+1+(i*8)+3] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + mipmap = reloc->robj; + r = evergreen_check_texture_resource(p, idx+1+(i*8), + texture, mipmap); + if (r) + return r; + break; + case SQ_TEX_VTX_VALID_BUFFER: + /* vtx base */ + r = evergreen_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("bad SET_RESOURCE (vtx)\n"); + return -EINVAL; + } + offset = radeon_get_ib_value(p, idx+1+(i*8)+0); + size = radeon_get_ib_value(p, idx+1+(i*8)+1); + if (p->rdev && (size + offset) > radeon_bo_size(reloc->robj)) { + /* force size to size of the buffer */ + dev_warn(p->dev, "vbo resource seems too big for the bo\n"); + ib[idx+1+(i*8)+1] = radeon_bo_size(reloc->robj); + } + ib[idx+1+(i*8)+0] += (u32)((reloc->lobj.gpu_offset) & 0xffffffff); + ib[idx+1+(i*8)+2] += upper_32_bits(reloc->lobj.gpu_offset) & 0xff; + break; + case SQ_TEX_VTX_INVALID_TEXTURE: + case SQ_TEX_VTX_INVALID_BUFFER: + default: + DRM_ERROR("bad SET_RESOURCE\n"); + return -EINVAL; + } + } + break; + case PACKET3_SET_ALU_CONST: + /* XXX fix me ALU const buffers only */ + break; + case PACKET3_SET_BOOL_CONST: + start_reg = (idx_value << 2) + PACKET3_SET_BOOL_CONST_START; + end_reg = 4 * pkt->count + start_reg - 4; + if ((start_reg < PACKET3_SET_BOOL_CONST_START) || + (start_reg >= PACKET3_SET_BOOL_CONST_END) || + (end_reg >= PACKET3_SET_BOOL_CONST_END)) { + DRM_ERROR("bad SET_BOOL_CONST\n"); + return -EINVAL; + } + break; + case PACKET3_SET_LOOP_CONST: + start_reg = (idx_value << 2) + PACKET3_SET_LOOP_CONST_START; + end_reg = 4 * pkt->count + start_reg - 4; + if ((start_reg < PACKET3_SET_LOOP_CONST_START) || + (start_reg >= PACKET3_SET_LOOP_CONST_END) || + (end_reg >= PACKET3_SET_LOOP_CONST_END)) { + DRM_ERROR("bad SET_LOOP_CONST\n"); + return -EINVAL; + } + break; + case PACKET3_SET_CTL_CONST: + start_reg = (idx_value << 2) + PACKET3_SET_CTL_CONST_START; + end_reg = 4 * pkt->count + start_reg - 4; + if ((start_reg < PACKET3_SET_CTL_CONST_START) || + (start_reg >= PACKET3_SET_CTL_CONST_END) || + (end_reg >= PACKET3_SET_CTL_CONST_END)) { + DRM_ERROR("bad SET_CTL_CONST\n"); + return -EINVAL; + } + break; + case PACKET3_SET_SAMPLER: + if (pkt->count % 3) { + DRM_ERROR("bad SET_SAMPLER\n"); + return -EINVAL; + } + start_reg = (idx_value << 2) + PACKET3_SET_SAMPLER_START; + end_reg = 4 * pkt->count + start_reg - 4; + if ((start_reg < PACKET3_SET_SAMPLER_START) || + (start_reg >= PACKET3_SET_SAMPLER_END) || + (end_reg >= PACKET3_SET_SAMPLER_END)) { + DRM_ERROR("bad SET_SAMPLER\n"); + return -EINVAL; + } + break; + case PACKET3_NOP: + break; + default: + DRM_ERROR("Packet3 opcode %x not supported\n", pkt->opcode); + return -EINVAL; + } + return 0; +} + +int evergreen_cs_parse(struct radeon_cs_parser *p) +{ + struct radeon_cs_packet pkt; + struct evergreen_cs_track *track; + int r; + + if (p->track == NULL) { + /* initialize tracker, we are in kms */ + track = kzalloc(sizeof(*track), GFP_KERNEL); + if (track == NULL) + return -ENOMEM; + evergreen_cs_track_init(track); + track->npipes = p->rdev->config.evergreen.tiling_npipes; + track->nbanks = p->rdev->config.evergreen.tiling_nbanks; + track->group_size = p->rdev->config.evergreen.tiling_group_size; + p->track = track; + } + do { + r = evergreen_cs_packet_parse(p, &pkt, p->idx); + if (r) { + kfree(p->track); + p->track = NULL; + return r; + } + p->idx += pkt.count + 2; + switch (pkt.type) { + case PACKET_TYPE0: + r = evergreen_cs_parse_packet0(p, &pkt); + break; + case PACKET_TYPE2: + break; + case PACKET_TYPE3: + r = evergreen_packet3_check(p, &pkt); + break; + default: + DRM_ERROR("Unknown packet type %d !\n", pkt.type); + kfree(p->track); + p->track = NULL; + return -EINVAL; + } + if (r) { + kfree(p->track); + p->track = NULL; + return r; + } + } while (p->idx < p->chunks[p->chunk_ib_idx].length_dw); +#if 0 + for (r = 0; r < p->ib->length_dw; r++) { + printk(KERN_INFO "%05d 0x%08X\n", r, p->ib->ptr[r]); + mdelay(1); + } +#endif + kfree(p->track); + p->track = NULL; + return 0; +} + diff --git a/drivers/gpu/drm/radeon/evergreen_reg.h b/drivers/gpu/drm/radeon/evergreen_reg.h new file mode 100644 index 00000000..c781c92c --- /dev/null +++ b/drivers/gpu/drm/radeon/evergreen_reg.h @@ -0,0 +1,194 @@ +/* + * Copyright 2010 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Alex Deucher + */ +#ifndef __EVERGREEN_REG_H__ +#define __EVERGREEN_REG_H__ + +/* evergreen */ +#define EVERGREEN_VGA_MEMORY_BASE_ADDRESS 0x310 +#define EVERGREEN_VGA_MEMORY_BASE_ADDRESS_HIGH 0x324 +#define EVERGREEN_D3VGA_CONTROL 0x3e0 +#define EVERGREEN_D4VGA_CONTROL 0x3e4 +#define EVERGREEN_D5VGA_CONTROL 0x3e8 +#define EVERGREEN_D6VGA_CONTROL 0x3ec + +#define EVERGREEN_P1PLL_SS_CNTL 0x414 +#define EVERGREEN_P2PLL_SS_CNTL 0x454 +# define EVERGREEN_PxPLL_SS_EN (1 << 12) +/* GRPH blocks at 0x6800, 0x7400, 0x10000, 0x10c00, 0x11800, 0x12400 */ +#define EVERGREEN_GRPH_ENABLE 0x6800 +#define EVERGREEN_GRPH_CONTROL 0x6804 +# define EVERGREEN_GRPH_DEPTH(x) (((x) & 0x3) << 0) +# define EVERGREEN_GRPH_DEPTH_8BPP 0 +# define EVERGREEN_GRPH_DEPTH_16BPP 1 +# define EVERGREEN_GRPH_DEPTH_32BPP 2 +# define EVERGREEN_GRPH_FORMAT(x) (((x) & 0x7) << 8) +/* 8 BPP */ +# define EVERGREEN_GRPH_FORMAT_INDEXED 0 +/* 16 BPP */ +# define EVERGREEN_GRPH_FORMAT_ARGB1555 0 +# define EVERGREEN_GRPH_FORMAT_ARGB565 1 +# define EVERGREEN_GRPH_FORMAT_ARGB4444 2 +# define EVERGREEN_GRPH_FORMAT_AI88 3 +# define EVERGREEN_GRPH_FORMAT_MONO16 4 +# define EVERGREEN_GRPH_FORMAT_BGRA5551 5 +/* 32 BPP */ +# define EVERGREEN_GRPH_FORMAT_ARGB8888 0 +# define EVERGREEN_GRPH_FORMAT_ARGB2101010 1 +# define EVERGREEN_GRPH_FORMAT_32BPP_DIG 2 +# define EVERGREEN_GRPH_FORMAT_8B_ARGB2101010 3 +# define EVERGREEN_GRPH_FORMAT_BGRA1010102 4 +# define EVERGREEN_GRPH_FORMAT_8B_BGRA1010102 5 +# define EVERGREEN_GRPH_FORMAT_RGB111110 6 +# define EVERGREEN_GRPH_FORMAT_BGR101111 7 +# define EVERGREEN_GRPH_ARRAY_MODE(x) (((x) & 0x7) << 20) +# define EVERGREEN_GRPH_ARRAY_LINEAR_GENERAL 0 +# define EVERGREEN_GRPH_ARRAY_LINEAR_ALIGNED 1 +# define EVERGREEN_GRPH_ARRAY_1D_TILED_THIN1 2 +# define EVERGREEN_GRPH_ARRAY_2D_TILED_THIN1 4 +#define EVERGREEN_GRPH_SWAP_CONTROL 0x680c +# define EVERGREEN_GRPH_ENDIAN_SWAP(x) (((x) & 0x3) << 0) +# define EVERGREEN_GRPH_ENDIAN_NONE 0 +# define EVERGREEN_GRPH_ENDIAN_8IN16 1 +# define EVERGREEN_GRPH_ENDIAN_8IN32 2 +# define EVERGREEN_GRPH_ENDIAN_8IN64 3 +# define EVERGREEN_GRPH_RED_CROSSBAR(x) (((x) & 0x3) << 4) +# define EVERGREEN_GRPH_RED_SEL_R 0 +# define EVERGREEN_GRPH_RED_SEL_G 1 +# define EVERGREEN_GRPH_RED_SEL_B 2 +# define EVERGREEN_GRPH_RED_SEL_A 3 +# define EVERGREEN_GRPH_GREEN_CROSSBAR(x) (((x) & 0x3) << 6) +# define EVERGREEN_GRPH_GREEN_SEL_G 0 +# define EVERGREEN_GRPH_GREEN_SEL_B 1 +# define EVERGREEN_GRPH_GREEN_SEL_A 2 +# define EVERGREEN_GRPH_GREEN_SEL_R 3 +# define EVERGREEN_GRPH_BLUE_CROSSBAR(x) (((x) & 0x3) << 8) +# define EVERGREEN_GRPH_BLUE_SEL_B 0 +# define EVERGREEN_GRPH_BLUE_SEL_A 1 +# define EVERGREEN_GRPH_BLUE_SEL_R 2 +# define EVERGREEN_GRPH_BLUE_SEL_G 3 +# define EVERGREEN_GRPH_ALPHA_CROSSBAR(x) (((x) & 0x3) << 10) +# define EVERGREEN_GRPH_ALPHA_SEL_A 0 +# define EVERGREEN_GRPH_ALPHA_SEL_R 1 +# define EVERGREEN_GRPH_ALPHA_SEL_G 2 +# define EVERGREEN_GRPH_ALPHA_SEL_B 3 +#define EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS 0x6810 +#define EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS 0x6814 +# define EVERGREEN_GRPH_DFQ_ENABLE (1 << 0) +# define EVERGREEN_GRPH_SURFACE_ADDRESS_MASK 0xffffff00 +#define EVERGREEN_GRPH_PITCH 0x6818 +#define EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS_HIGH 0x681c +#define EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS_HIGH 0x6820 +#define EVERGREEN_GRPH_SURFACE_OFFSET_X 0x6824 +#define EVERGREEN_GRPH_SURFACE_OFFSET_Y 0x6828 +#define EVERGREEN_GRPH_X_START 0x682c +#define EVERGREEN_GRPH_Y_START 0x6830 +#define EVERGREEN_GRPH_X_END 0x6834 +#define EVERGREEN_GRPH_Y_END 0x6838 +#define EVERGREEN_GRPH_UPDATE 0x6844 +# define EVERGREEN_GRPH_SURFACE_UPDATE_PENDING (1 << 2) +# define EVERGREEN_GRPH_UPDATE_LOCK (1 << 16) +#define EVERGREEN_GRPH_FLIP_CONTROL 0x6848 +# define EVERGREEN_GRPH_SURFACE_UPDATE_H_RETRACE_EN (1 << 0) + +/* CUR blocks at 0x6998, 0x7598, 0x10198, 0x10d98, 0x11998, 0x12598 */ +#define EVERGREEN_CUR_CONTROL 0x6998 +# define EVERGREEN_CURSOR_EN (1 << 0) +# define EVERGREEN_CURSOR_MODE(x) (((x) & 0x3) << 8) +# define EVERGREEN_CURSOR_MONO 0 +# define EVERGREEN_CURSOR_24_1 1 +# define EVERGREEN_CURSOR_24_8_PRE_MULT 2 +# define EVERGREEN_CURSOR_24_8_UNPRE_MULT 3 +# define EVERGREEN_CURSOR_2X_MAGNIFY (1 << 16) +# define EVERGREEN_CURSOR_FORCE_MC_ON (1 << 20) +# define EVERGREEN_CURSOR_URGENT_CONTROL(x) (((x) & 0x7) << 24) +# define EVERGREEN_CURSOR_URGENT_ALWAYS 0 +# define EVERGREEN_CURSOR_URGENT_1_8 1 +# define EVERGREEN_CURSOR_URGENT_1_4 2 +# define EVERGREEN_CURSOR_URGENT_3_8 3 +# define EVERGREEN_CURSOR_URGENT_1_2 4 +#define EVERGREEN_CUR_SURFACE_ADDRESS 0x699c +# define EVERGREEN_CUR_SURFACE_ADDRESS_MASK 0xfffff000 +#define EVERGREEN_CUR_SIZE 0x69a0 +#define EVERGREEN_CUR_SURFACE_ADDRESS_HIGH 0x69a4 +#define EVERGREEN_CUR_POSITION 0x69a8 +#define EVERGREEN_CUR_HOT_SPOT 0x69ac +#define EVERGREEN_CUR_COLOR1 0x69b0 +#define EVERGREEN_CUR_COLOR2 0x69b4 +#define EVERGREEN_CUR_UPDATE 0x69b8 +# define EVERGREEN_CURSOR_UPDATE_PENDING (1 << 0) +# define EVERGREEN_CURSOR_UPDATE_TAKEN (1 << 1) +# define EVERGREEN_CURSOR_UPDATE_LOCK (1 << 16) +# define EVERGREEN_CURSOR_DISABLE_MULTIPLE_UPDATE (1 << 24) + +/* LUT blocks at 0x69e0, 0x75e0, 0x101e0, 0x10de0, 0x119e0, 0x125e0 */ +#define EVERGREEN_DC_LUT_RW_MODE 0x69e0 +#define EVERGREEN_DC_LUT_RW_INDEX 0x69e4 +#define EVERGREEN_DC_LUT_SEQ_COLOR 0x69e8 +#define EVERGREEN_DC_LUT_PWL_DATA 0x69ec +#define EVERGREEN_DC_LUT_30_COLOR 0x69f0 +#define EVERGREEN_DC_LUT_VGA_ACCESS_ENABLE 0x69f4 +#define EVERGREEN_DC_LUT_WRITE_EN_MASK 0x69f8 +#define EVERGREEN_DC_LUT_AUTOFILL 0x69fc +#define EVERGREEN_DC_LUT_CONTROL 0x6a00 +#define EVERGREEN_DC_LUT_BLACK_OFFSET_BLUE 0x6a04 +#define EVERGREEN_DC_LUT_BLACK_OFFSET_GREEN 0x6a08 +#define EVERGREEN_DC_LUT_BLACK_OFFSET_RED 0x6a0c +#define EVERGREEN_DC_LUT_WHITE_OFFSET_BLUE 0x6a10 +#define EVERGREEN_DC_LUT_WHITE_OFFSET_GREEN 0x6a14 +#define EVERGREEN_DC_LUT_WHITE_OFFSET_RED 0x6a18 + +#define EVERGREEN_DATA_FORMAT 0x6b00 +# define EVERGREEN_INTERLEAVE_EN (1 << 0) +#define EVERGREEN_DESKTOP_HEIGHT 0x6b04 +#define EVERGREEN_VLINE_START_END 0x6b08 +#define EVERGREEN_VLINE_STATUS 0x6bb8 +# define EVERGREEN_VLINE_STAT (1 << 12) + +#define EVERGREEN_VIEWPORT_START 0x6d70 +#define EVERGREEN_VIEWPORT_SIZE 0x6d74 + +/* display controller offsets used for crtc/cur/lut/grph/viewport/etc. */ +#define EVERGREEN_CRTC0_REGISTER_OFFSET (0x6df0 - 0x6df0) +#define EVERGREEN_CRTC1_REGISTER_OFFSET (0x79f0 - 0x6df0) +#define EVERGREEN_CRTC2_REGISTER_OFFSET (0x105f0 - 0x6df0) +#define EVERGREEN_CRTC3_REGISTER_OFFSET (0x111f0 - 0x6df0) +#define EVERGREEN_CRTC4_REGISTER_OFFSET (0x11df0 - 0x6df0) +#define EVERGREEN_CRTC5_REGISTER_OFFSET (0x129f0 - 0x6df0) + +/* CRTC blocks at 0x6df0, 0x79f0, 0x105f0, 0x111f0, 0x11df0, 0x129f0 */ +#define EVERGREEN_CRTC_V_BLANK_START_END 0x6e34 +#define EVERGREEN_CRTC_CONTROL 0x6e70 +# define EVERGREEN_CRTC_MASTER_EN (1 << 0) +# define EVERGREEN_CRTC_DISP_READ_REQUEST_DISABLE (1 << 24) +#define EVERGREEN_CRTC_STATUS 0x6e8c +#define EVERGREEN_CRTC_STATUS_POSITION 0x6e90 +#define EVERGREEN_MASTER_UPDATE_MODE 0x6ef8 +#define EVERGREEN_CRTC_UPDATE_LOCK 0x6ed4 + +#define EVERGREEN_DC_GPIO_HPD_MASK 0x64b0 +#define EVERGREEN_DC_GPIO_HPD_A 0x64b4 +#define EVERGREEN_DC_GPIO_HPD_EN 0x64b8 +#define EVERGREEN_DC_GPIO_HPD_Y 0x64bc + +#endif diff --git a/drivers/gpu/drm/radeon/evergreend.h b/drivers/gpu/drm/radeon/evergreend.h new file mode 100644 index 00000000..6078ae4c --- /dev/null +++ b/drivers/gpu/drm/radeon/evergreend.h @@ -0,0 +1,1129 @@ +/* + * Copyright 2010 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Alex Deucher + */ +#ifndef EVERGREEND_H +#define EVERGREEND_H + +#define EVERGREEN_MAX_SH_GPRS 256 +#define EVERGREEN_MAX_TEMP_GPRS 16 +#define EVERGREEN_MAX_SH_THREADS 256 +#define EVERGREEN_MAX_SH_STACK_ENTRIES 4096 +#define EVERGREEN_MAX_FRC_EOV_CNT 16384 +#define EVERGREEN_MAX_BACKENDS 8 +#define EVERGREEN_MAX_BACKENDS_MASK 0xFF +#define EVERGREEN_MAX_SIMDS 16 +#define EVERGREEN_MAX_SIMDS_MASK 0xFFFF +#define EVERGREEN_MAX_PIPES 8 +#define EVERGREEN_MAX_PIPES_MASK 0xFF +#define EVERGREEN_MAX_LDS_NUM 0xFFFF + +/* Registers */ + +#define RCU_IND_INDEX 0x100 +#define RCU_IND_DATA 0x104 + +#define GRBM_GFX_INDEX 0x802C +#define INSTANCE_INDEX(x) ((x) << 0) +#define SE_INDEX(x) ((x) << 16) +#define INSTANCE_BROADCAST_WRITES (1 << 30) +#define SE_BROADCAST_WRITES (1 << 31) +#define RLC_GFX_INDEX 0x3fC4 +#define CC_GC_SHADER_PIPE_CONFIG 0x8950 +#define WRITE_DIS (1 << 0) +#define CC_RB_BACKEND_DISABLE 0x98F4 +#define BACKEND_DISABLE(x) ((x) << 16) +#define GB_ADDR_CONFIG 0x98F8 +#define NUM_PIPES(x) ((x) << 0) +#define PIPE_INTERLEAVE_SIZE(x) ((x) << 4) +#define BANK_INTERLEAVE_SIZE(x) ((x) << 8) +#define NUM_SHADER_ENGINES(x) ((x) << 12) +#define SHADER_ENGINE_TILE_SIZE(x) ((x) << 16) +#define NUM_GPUS(x) ((x) << 20) +#define MULTI_GPU_TILE_SIZE(x) ((x) << 24) +#define ROW_SIZE(x) ((x) << 28) +#define GB_BACKEND_MAP 0x98FC +#define DMIF_ADDR_CONFIG 0xBD4 +#define HDP_ADDR_CONFIG 0x2F48 +#define HDP_MISC_CNTL 0x2F4C +#define HDP_FLUSH_INVALIDATE_CACHE (1 << 0) + +#define CC_SYS_RB_BACKEND_DISABLE 0x3F88 +#define GC_USER_RB_BACKEND_DISABLE 0x9B7C + +#define CGTS_SYS_TCC_DISABLE 0x3F90 +#define CGTS_TCC_DISABLE 0x9148 +#define CGTS_USER_SYS_TCC_DISABLE 0x3F94 +#define CGTS_USER_TCC_DISABLE 0x914C + +#define CONFIG_MEMSIZE 0x5428 + +#define CP_ME_CNTL 0x86D8 +#define CP_ME_HALT (1 << 28) +#define CP_PFP_HALT (1 << 26) +#define CP_ME_RAM_DATA 0xC160 +#define CP_ME_RAM_RADDR 0xC158 +#define CP_ME_RAM_WADDR 0xC15C +#define CP_MEQ_THRESHOLDS 0x8764 +#define STQ_SPLIT(x) ((x) << 0) +#define CP_PERFMON_CNTL 0x87FC +#define CP_PFP_UCODE_ADDR 0xC150 +#define CP_PFP_UCODE_DATA 0xC154 +#define CP_QUEUE_THRESHOLDS 0x8760 +#define ROQ_IB1_START(x) ((x) << 0) +#define ROQ_IB2_START(x) ((x) << 8) +#define CP_RB_BASE 0xC100 +#define CP_RB_CNTL 0xC104 +#define RB_BUFSZ(x) ((x) << 0) +#define RB_BLKSZ(x) ((x) << 8) +#define RB_NO_UPDATE (1 << 27) +#define RB_RPTR_WR_ENA (1 << 31) +#define BUF_SWAP_32BIT (2 << 16) +#define CP_RB_RPTR 0x8700 +#define CP_RB_RPTR_ADDR 0xC10C +#define RB_RPTR_SWAP(x) ((x) << 0) +#define CP_RB_RPTR_ADDR_HI 0xC110 +#define CP_RB_RPTR_WR 0xC108 +#define CP_RB_WPTR 0xC114 +#define CP_RB_WPTR_ADDR 0xC118 +#define CP_RB_WPTR_ADDR_HI 0xC11C +#define CP_RB_WPTR_DELAY 0x8704 +#define CP_SEM_WAIT_TIMER 0x85BC +#define CP_DEBUG 0xC1FC + + +#define GC_USER_SHADER_PIPE_CONFIG 0x8954 +#define INACTIVE_QD_PIPES(x) ((x) << 8) +#define INACTIVE_QD_PIPES_MASK 0x0000FF00 +#define INACTIVE_SIMDS(x) ((x) << 16) +#define INACTIVE_SIMDS_MASK 0x00FF0000 + +#define GRBM_CNTL 0x8000 +#define GRBM_READ_TIMEOUT(x) ((x) << 0) +#define GRBM_SOFT_RESET 0x8020 +#define SOFT_RESET_CP (1 << 0) +#define SOFT_RESET_CB (1 << 1) +#define SOFT_RESET_DB (1 << 3) +#define SOFT_RESET_PA (1 << 5) +#define SOFT_RESET_SC (1 << 6) +#define SOFT_RESET_SPI (1 << 8) +#define SOFT_RESET_SH (1 << 9) +#define SOFT_RESET_SX (1 << 10) +#define SOFT_RESET_TC (1 << 11) +#define SOFT_RESET_TA (1 << 12) +#define SOFT_RESET_VC (1 << 13) +#define SOFT_RESET_VGT (1 << 14) + +#define GRBM_STATUS 0x8010 +#define CMDFIFO_AVAIL_MASK 0x0000000F +#define SRBM_RQ_PENDING (1 << 5) +#define CF_RQ_PENDING (1 << 7) +#define PF_RQ_PENDING (1 << 8) +#define GRBM_EE_BUSY (1 << 10) +#define SX_CLEAN (1 << 11) +#define DB_CLEAN (1 << 12) +#define CB_CLEAN (1 << 13) +#define TA_BUSY (1 << 14) +#define VGT_BUSY_NO_DMA (1 << 16) +#define VGT_BUSY (1 << 17) +#define SX_BUSY (1 << 20) +#define SH_BUSY (1 << 21) +#define SPI_BUSY (1 << 22) +#define SC_BUSY (1 << 24) +#define PA_BUSY (1 << 25) +#define DB_BUSY (1 << 26) +#define CP_COHERENCY_BUSY (1 << 28) +#define CP_BUSY (1 << 29) +#define CB_BUSY (1 << 30) +#define GUI_ACTIVE (1 << 31) +#define GRBM_STATUS_SE0 0x8014 +#define GRBM_STATUS_SE1 0x8018 +#define SE_SX_CLEAN (1 << 0) +#define SE_DB_CLEAN (1 << 1) +#define SE_CB_CLEAN (1 << 2) +#define SE_TA_BUSY (1 << 25) +#define SE_SX_BUSY (1 << 26) +#define SE_SPI_BUSY (1 << 27) +#define SE_SH_BUSY (1 << 28) +#define SE_SC_BUSY (1 << 29) +#define SE_DB_BUSY (1 << 30) +#define SE_CB_BUSY (1 << 31) +/* evergreen */ +#define CG_THERMAL_CTRL 0x72c +#define TOFFSET_MASK 0x00003FE0 +#define TOFFSET_SHIFT 5 +#define CG_MULT_THERMAL_STATUS 0x740 +#define ASIC_T(x) ((x) << 16) +#define ASIC_T_MASK 0x07FF0000 +#define ASIC_T_SHIFT 16 +#define CG_TS0_STATUS 0x760 +#define TS0_ADC_DOUT_MASK 0x000003FF +#define TS0_ADC_DOUT_SHIFT 0 +/* APU */ +#define CG_THERMAL_STATUS 0x678 + +#define HDP_HOST_PATH_CNTL 0x2C00 +#define HDP_NONSURFACE_BASE 0x2C04 +#define HDP_NONSURFACE_INFO 0x2C08 +#define HDP_NONSURFACE_SIZE 0x2C0C +#define HDP_MEM_COHERENCY_FLUSH_CNTL 0x5480 +#define HDP_REG_COHERENCY_FLUSH_CNTL 0x54A0 +#define HDP_TILING_CONFIG 0x2F3C + +#define MC_SHARED_CHMAP 0x2004 +#define NOOFCHAN_SHIFT 12 +#define NOOFCHAN_MASK 0x00003000 +#define MC_SHARED_CHREMAP 0x2008 + +#define MC_ARB_RAMCFG 0x2760 +#define NOOFBANK_SHIFT 0 +#define NOOFBANK_MASK 0x00000003 +#define NOOFRANK_SHIFT 2 +#define NOOFRANK_MASK 0x00000004 +#define NOOFROWS_SHIFT 3 +#define NOOFROWS_MASK 0x00000038 +#define NOOFCOLS_SHIFT 6 +#define NOOFCOLS_MASK 0x000000C0 +#define CHANSIZE_SHIFT 8 +#define CHANSIZE_MASK 0x00000100 +#define BURSTLENGTH_SHIFT 9 +#define BURSTLENGTH_MASK 0x00000200 +#define CHANSIZE_OVERRIDE (1 << 11) +#define FUS_MC_ARB_RAMCFG 0x2768 +#define MC_VM_AGP_TOP 0x2028 +#define MC_VM_AGP_BOT 0x202C +#define MC_VM_AGP_BASE 0x2030 +#define MC_VM_FB_LOCATION 0x2024 +#define MC_FUS_VM_FB_OFFSET 0x2898 +#define MC_VM_MB_L1_TLB0_CNTL 0x2234 +#define MC_VM_MB_L1_TLB1_CNTL 0x2238 +#define MC_VM_MB_L1_TLB2_CNTL 0x223C +#define MC_VM_MB_L1_TLB3_CNTL 0x2240 +#define ENABLE_L1_TLB (1 << 0) +#define ENABLE_L1_FRAGMENT_PROCESSING (1 << 1) +#define SYSTEM_ACCESS_MODE_PA_ONLY (0 << 3) +#define SYSTEM_ACCESS_MODE_USE_SYS_MAP (1 << 3) +#define SYSTEM_ACCESS_MODE_IN_SYS (2 << 3) +#define SYSTEM_ACCESS_MODE_NOT_IN_SYS (3 << 3) +#define SYSTEM_APERTURE_UNMAPPED_ACCESS_PASS_THRU (0 << 5) +#define EFFECTIVE_L1_TLB_SIZE(x) ((x)<<15) +#define EFFECTIVE_L1_QUEUE_SIZE(x) ((x)<<18) +#define MC_VM_MD_L1_TLB0_CNTL 0x2654 +#define MC_VM_MD_L1_TLB1_CNTL 0x2658 +#define MC_VM_MD_L1_TLB2_CNTL 0x265C +#define MC_VM_MD_L1_TLB3_CNTL 0x2698 + +#define FUS_MC_VM_MD_L1_TLB0_CNTL 0x265C +#define FUS_MC_VM_MD_L1_TLB1_CNTL 0x2660 +#define FUS_MC_VM_MD_L1_TLB2_CNTL 0x2664 + +#define MC_VM_SYSTEM_APERTURE_DEFAULT_ADDR 0x203C +#define MC_VM_SYSTEM_APERTURE_HIGH_ADDR 0x2038 +#define MC_VM_SYSTEM_APERTURE_LOW_ADDR 0x2034 + +#define PA_CL_ENHANCE 0x8A14 +#define CLIP_VTX_REORDER_ENA (1 << 0) +#define NUM_CLIP_SEQ(x) ((x) << 1) +#define PA_SC_AA_CONFIG 0x28C04 +#define MSAA_NUM_SAMPLES_SHIFT 0 +#define MSAA_NUM_SAMPLES_MASK 0x3 +#define PA_SC_CLIPRECT_RULE 0x2820C +#define PA_SC_EDGERULE 0x28230 +#define PA_SC_FIFO_SIZE 0x8BCC +#define SC_PRIM_FIFO_SIZE(x) ((x) << 0) +#define SC_HIZ_TILE_FIFO_SIZE(x) ((x) << 12) +#define SC_EARLYZ_TILE_FIFO_SIZE(x) ((x) << 20) +#define PA_SC_FORCE_EOV_MAX_CNTS 0x8B24 +#define FORCE_EOV_MAX_CLK_CNT(x) ((x) << 0) +#define FORCE_EOV_MAX_REZ_CNT(x) ((x) << 16) +#define PA_SC_LINE_STIPPLE 0x28A0C +#define PA_SU_LINE_STIPPLE_VALUE 0x8A60 +#define PA_SC_LINE_STIPPLE_STATE 0x8B10 + +#define SCRATCH_REG0 0x8500 +#define SCRATCH_REG1 0x8504 +#define SCRATCH_REG2 0x8508 +#define SCRATCH_REG3 0x850C +#define SCRATCH_REG4 0x8510 +#define SCRATCH_REG5 0x8514 +#define SCRATCH_REG6 0x8518 +#define SCRATCH_REG7 0x851C +#define SCRATCH_UMSK 0x8540 +#define SCRATCH_ADDR 0x8544 + +#define SMX_DC_CTL0 0xA020 +#define USE_HASH_FUNCTION (1 << 0) +#define NUMBER_OF_SETS(x) ((x) << 1) +#define FLUSH_ALL_ON_EVENT (1 << 10) +#define STALL_ON_EVENT (1 << 11) +#define SMX_EVENT_CTL 0xA02C +#define ES_FLUSH_CTL(x) ((x) << 0) +#define GS_FLUSH_CTL(x) ((x) << 3) +#define ACK_FLUSH_CTL(x) ((x) << 6) +#define SYNC_FLUSH_CTL (1 << 8) + +#define SPI_CONFIG_CNTL 0x9100 +#define GPR_WRITE_PRIORITY(x) ((x) << 0) +#define SPI_CONFIG_CNTL_1 0x913C +#define VTX_DONE_DELAY(x) ((x) << 0) +#define INTERP_ONE_PRIM_PER_ROW (1 << 4) +#define SPI_INPUT_Z 0x286D8 +#define SPI_PS_IN_CONTROL_0 0x286CC +#define NUM_INTERP(x) ((x)<<0) +#define POSITION_ENA (1<<8) +#define POSITION_CENTROID (1<<9) +#define POSITION_ADDR(x) ((x)<<10) +#define PARAM_GEN(x) ((x)<<15) +#define PARAM_GEN_ADDR(x) ((x)<<19) +#define BARYC_SAMPLE_CNTL(x) ((x)<<26) +#define PERSP_GRADIENT_ENA (1<<28) +#define LINEAR_GRADIENT_ENA (1<<29) +#define POSITION_SAMPLE (1<<30) +#define BARYC_AT_SAMPLE_ENA (1<<31) + +#define SQ_CONFIG 0x8C00 +#define VC_ENABLE (1 << 0) +#define EXPORT_SRC_C (1 << 1) +#define CS_PRIO(x) ((x) << 18) +#define LS_PRIO(x) ((x) << 20) +#define HS_PRIO(x) ((x) << 22) +#define PS_PRIO(x) ((x) << 24) +#define VS_PRIO(x) ((x) << 26) +#define GS_PRIO(x) ((x) << 28) +#define ES_PRIO(x) ((x) << 30) +#define SQ_GPR_RESOURCE_MGMT_1 0x8C04 +#define NUM_PS_GPRS(x) ((x) << 0) +#define NUM_VS_GPRS(x) ((x) << 16) +#define NUM_CLAUSE_TEMP_GPRS(x) ((x) << 28) +#define SQ_GPR_RESOURCE_MGMT_2 0x8C08 +#define NUM_GS_GPRS(x) ((x) << 0) +#define NUM_ES_GPRS(x) ((x) << 16) +#define SQ_GPR_RESOURCE_MGMT_3 0x8C0C +#define NUM_HS_GPRS(x) ((x) << 0) +#define NUM_LS_GPRS(x) ((x) << 16) +#define SQ_THREAD_RESOURCE_MGMT 0x8C18 +#define NUM_PS_THREADS(x) ((x) << 0) +#define NUM_VS_THREADS(x) ((x) << 8) +#define NUM_GS_THREADS(x) ((x) << 16) +#define NUM_ES_THREADS(x) ((x) << 24) +#define SQ_THREAD_RESOURCE_MGMT_2 0x8C1C +#define NUM_HS_THREADS(x) ((x) << 0) +#define NUM_LS_THREADS(x) ((x) << 8) +#define SQ_STACK_RESOURCE_MGMT_1 0x8C20 +#define NUM_PS_STACK_ENTRIES(x) ((x) << 0) +#define NUM_VS_STACK_ENTRIES(x) ((x) << 16) +#define SQ_STACK_RESOURCE_MGMT_2 0x8C24 +#define NUM_GS_STACK_ENTRIES(x) ((x) << 0) +#define NUM_ES_STACK_ENTRIES(x) ((x) << 16) +#define SQ_STACK_RESOURCE_MGMT_3 0x8C28 +#define NUM_HS_STACK_ENTRIES(x) ((x) << 0) +#define NUM_LS_STACK_ENTRIES(x) ((x) << 16) +#define SQ_DYN_GPR_CNTL_PS_FLUSH_REQ 0x8D8C +#define SQ_LDS_RESOURCE_MGMT 0x8E2C + +#define SQ_MS_FIFO_SIZES 0x8CF0 +#define CACHE_FIFO_SIZE(x) ((x) << 0) +#define FETCH_FIFO_HIWATER(x) ((x) << 8) +#define DONE_FIFO_HIWATER(x) ((x) << 16) +#define ALU_UPDATE_FIFO_HIWATER(x) ((x) << 24) + +#define SX_DEBUG_1 0x9058 +#define ENABLE_NEW_SMX_ADDRESS (1 << 16) +#define SX_EXPORT_BUFFER_SIZES 0x900C +#define COLOR_BUFFER_SIZE(x) ((x) << 0) +#define POSITION_BUFFER_SIZE(x) ((x) << 8) +#define SMX_BUFFER_SIZE(x) ((x) << 16) +#define SX_MISC 0x28350 + +#define CB_PERF_CTR0_SEL_0 0x9A20 +#define CB_PERF_CTR0_SEL_1 0x9A24 +#define CB_PERF_CTR1_SEL_0 0x9A28 +#define CB_PERF_CTR1_SEL_1 0x9A2C +#define CB_PERF_CTR2_SEL_0 0x9A30 +#define CB_PERF_CTR2_SEL_1 0x9A34 +#define CB_PERF_CTR3_SEL_0 0x9A38 +#define CB_PERF_CTR3_SEL_1 0x9A3C + +#define TA_CNTL_AUX 0x9508 +#define DISABLE_CUBE_WRAP (1 << 0) +#define DISABLE_CUBE_ANISO (1 << 1) +#define SYNC_GRADIENT (1 << 24) +#define SYNC_WALKER (1 << 25) +#define SYNC_ALIGNER (1 << 26) + +#define TCP_CHAN_STEER_LO 0x960c +#define TCP_CHAN_STEER_HI 0x9610 + +#define VGT_CACHE_INVALIDATION 0x88C4 +#define CACHE_INVALIDATION(x) ((x) << 0) +#define VC_ONLY 0 +#define TC_ONLY 1 +#define VC_AND_TC 2 +#define AUTO_INVLD_EN(x) ((x) << 6) +#define NO_AUTO 0 +#define ES_AUTO 1 +#define GS_AUTO 2 +#define ES_AND_GS_AUTO 3 +#define VGT_GS_VERTEX_REUSE 0x88D4 +#define VGT_NUM_INSTANCES 0x8974 +#define VGT_OUT_DEALLOC_CNTL 0x28C5C +#define DEALLOC_DIST_MASK 0x0000007F +#define VGT_VERTEX_REUSE_BLOCK_CNTL 0x28C58 +#define VTX_REUSE_DEPTH_MASK 0x000000FF + +#define VM_CONTEXT0_CNTL 0x1410 +#define ENABLE_CONTEXT (1 << 0) +#define PAGE_TABLE_DEPTH(x) (((x) & 3) << 1) +#define RANGE_PROTECTION_FAULT_ENABLE_DEFAULT (1 << 4) +#define VM_CONTEXT1_CNTL 0x1414 +#define VM_CONTEXT0_PAGE_TABLE_BASE_ADDR 0x153C +#define VM_CONTEXT0_PAGE_TABLE_END_ADDR 0x157C +#define VM_CONTEXT0_PAGE_TABLE_START_ADDR 0x155C +#define VM_CONTEXT0_PROTECTION_FAULT_DEFAULT_ADDR 0x1518 +#define VM_CONTEXT0_REQUEST_RESPONSE 0x1470 +#define REQUEST_TYPE(x) (((x) & 0xf) << 0) +#define RESPONSE_TYPE_MASK 0x000000F0 +#define RESPONSE_TYPE_SHIFT 4 +#define VM_L2_CNTL 0x1400 +#define ENABLE_L2_CACHE (1 << 0) +#define ENABLE_L2_FRAGMENT_PROCESSING (1 << 1) +#define ENABLE_L2_PTE_CACHE_LRU_UPDATE_BY_WRITE (1 << 9) +#define EFFECTIVE_L2_QUEUE_SIZE(x) (((x) & 7) << 14) +#define VM_L2_CNTL2 0x1404 +#define INVALIDATE_ALL_L1_TLBS (1 << 0) +#define INVALIDATE_L2_CACHE (1 << 1) +#define VM_L2_CNTL3 0x1408 +#define BANK_SELECT(x) ((x) << 0) +#define CACHE_UPDATE_MODE(x) ((x) << 6) +#define VM_L2_STATUS 0x140C +#define L2_BUSY (1 << 0) + +#define WAIT_UNTIL 0x8040 + +#define SRBM_STATUS 0x0E50 +#define SRBM_SOFT_RESET 0x0E60 +#define SRBM_SOFT_RESET_ALL_MASK 0x00FEEFA6 +#define SOFT_RESET_BIF (1 << 1) +#define SOFT_RESET_CG (1 << 2) +#define SOFT_RESET_DC (1 << 5) +#define SOFT_RESET_GRBM (1 << 8) +#define SOFT_RESET_HDP (1 << 9) +#define SOFT_RESET_IH (1 << 10) +#define SOFT_RESET_MC (1 << 11) +#define SOFT_RESET_RLC (1 << 13) +#define SOFT_RESET_ROM (1 << 14) +#define SOFT_RESET_SEM (1 << 15) +#define SOFT_RESET_VMC (1 << 17) +#define SOFT_RESET_TST (1 << 21) +#define SOFT_RESET_REGBB (1 << 22) +#define SOFT_RESET_ORB (1 << 23) + +/* display watermarks */ +#define DC_LB_MEMORY_SPLIT 0x6b0c +#define PRIORITY_A_CNT 0x6b18 +#define PRIORITY_MARK_MASK 0x7fff +#define PRIORITY_OFF (1 << 16) +#define PRIORITY_ALWAYS_ON (1 << 20) +#define PRIORITY_B_CNT 0x6b1c +#define PIPE0_ARBITRATION_CONTROL3 0x0bf0 +# define LATENCY_WATERMARK_MASK(x) ((x) << 16) +#define PIPE0_LATENCY_CONTROL 0x0bf4 +# define LATENCY_LOW_WATERMARK(x) ((x) << 0) +# define LATENCY_HIGH_WATERMARK(x) ((x) << 16) + +#define IH_RB_CNTL 0x3e00 +# define IH_RB_ENABLE (1 << 0) +# define IH_IB_SIZE(x) ((x) << 1) /* log2 */ +# define IH_RB_FULL_DRAIN_ENABLE (1 << 6) +# define IH_WPTR_WRITEBACK_ENABLE (1 << 8) +# define IH_WPTR_WRITEBACK_TIMER(x) ((x) << 9) /* log2 */ +# define IH_WPTR_OVERFLOW_ENABLE (1 << 16) +# define IH_WPTR_OVERFLOW_CLEAR (1 << 31) +#define IH_RB_BASE 0x3e04 +#define IH_RB_RPTR 0x3e08 +#define IH_RB_WPTR 0x3e0c +# define RB_OVERFLOW (1 << 0) +# define WPTR_OFFSET_MASK 0x3fffc +#define IH_RB_WPTR_ADDR_HI 0x3e10 +#define IH_RB_WPTR_ADDR_LO 0x3e14 +#define IH_CNTL 0x3e18 +# define ENABLE_INTR (1 << 0) +# define IH_MC_SWAP(x) ((x) << 1) +# define IH_MC_SWAP_NONE 0 +# define IH_MC_SWAP_16BIT 1 +# define IH_MC_SWAP_32BIT 2 +# define IH_MC_SWAP_64BIT 3 +# define RPTR_REARM (1 << 4) +# define MC_WRREQ_CREDIT(x) ((x) << 15) +# define MC_WR_CLEAN_CNT(x) ((x) << 20) + +#define CP_INT_CNTL 0xc124 +# define CNTX_BUSY_INT_ENABLE (1 << 19) +# define CNTX_EMPTY_INT_ENABLE (1 << 20) +# define SCRATCH_INT_ENABLE (1 << 25) +# define TIME_STAMP_INT_ENABLE (1 << 26) +# define IB2_INT_ENABLE (1 << 29) +# define IB1_INT_ENABLE (1 << 30) +# define RB_INT_ENABLE (1 << 31) +#define CP_INT_STATUS 0xc128 +# define SCRATCH_INT_STAT (1 << 25) +# define TIME_STAMP_INT_STAT (1 << 26) +# define IB2_INT_STAT (1 << 29) +# define IB1_INT_STAT (1 << 30) +# define RB_INT_STAT (1 << 31) + +#define GRBM_INT_CNTL 0x8060 +# define RDERR_INT_ENABLE (1 << 0) +# define GUI_IDLE_INT_ENABLE (1 << 19) + +/* 0x6e98, 0x7a98, 0x10698, 0x11298, 0x11e98, 0x12a98 */ +#define CRTC_STATUS_FRAME_COUNT 0x6e98 + +/* 0x6bb8, 0x77b8, 0x103b8, 0x10fb8, 0x11bb8, 0x127b8 */ +#define VLINE_STATUS 0x6bb8 +# define VLINE_OCCURRED (1 << 0) +# define VLINE_ACK (1 << 4) +# define VLINE_STAT (1 << 12) +# define VLINE_INTERRUPT (1 << 16) +# define VLINE_INTERRUPT_TYPE (1 << 17) +/* 0x6bbc, 0x77bc, 0x103bc, 0x10fbc, 0x11bbc, 0x127bc */ +#define VBLANK_STATUS 0x6bbc +# define VBLANK_OCCURRED (1 << 0) +# define VBLANK_ACK (1 << 4) +# define VBLANK_STAT (1 << 12) +# define VBLANK_INTERRUPT (1 << 16) +# define VBLANK_INTERRUPT_TYPE (1 << 17) + +/* 0x6b40, 0x7740, 0x10340, 0x10f40, 0x11b40, 0x12740 */ +#define INT_MASK 0x6b40 +# define VBLANK_INT_MASK (1 << 0) +# define VLINE_INT_MASK (1 << 4) + +#define DISP_INTERRUPT_STATUS 0x60f4 +# define LB_D1_VLINE_INTERRUPT (1 << 2) +# define LB_D1_VBLANK_INTERRUPT (1 << 3) +# define DC_HPD1_INTERRUPT (1 << 17) +# define DC_HPD1_RX_INTERRUPT (1 << 18) +# define DACA_AUTODETECT_INTERRUPT (1 << 22) +# define DACB_AUTODETECT_INTERRUPT (1 << 23) +# define DC_I2C_SW_DONE_INTERRUPT (1 << 24) +# define DC_I2C_HW_DONE_INTERRUPT (1 << 25) +#define DISP_INTERRUPT_STATUS_CONTINUE 0x60f8 +# define LB_D2_VLINE_INTERRUPT (1 << 2) +# define LB_D2_VBLANK_INTERRUPT (1 << 3) +# define DC_HPD2_INTERRUPT (1 << 17) +# define DC_HPD2_RX_INTERRUPT (1 << 18) +# define DISP_TIMER_INTERRUPT (1 << 24) +#define DISP_INTERRUPT_STATUS_CONTINUE2 0x60fc +# define LB_D3_VLINE_INTERRUPT (1 << 2) +# define LB_D3_VBLANK_INTERRUPT (1 << 3) +# define DC_HPD3_INTERRUPT (1 << 17) +# define DC_HPD3_RX_INTERRUPT (1 << 18) +#define DISP_INTERRUPT_STATUS_CONTINUE3 0x6100 +# define LB_D4_VLINE_INTERRUPT (1 << 2) +# define LB_D4_VBLANK_INTERRUPT (1 << 3) +# define DC_HPD4_INTERRUPT (1 << 17) +# define DC_HPD4_RX_INTERRUPT (1 << 18) +#define DISP_INTERRUPT_STATUS_CONTINUE4 0x614c +# define LB_D5_VLINE_INTERRUPT (1 << 2) +# define LB_D5_VBLANK_INTERRUPT (1 << 3) +# define DC_HPD5_INTERRUPT (1 << 17) +# define DC_HPD5_RX_INTERRUPT (1 << 18) +#define DISP_INTERRUPT_STATUS_CONTINUE5 0x6150 +# define LB_D6_VLINE_INTERRUPT (1 << 2) +# define LB_D6_VBLANK_INTERRUPT (1 << 3) +# define DC_HPD6_INTERRUPT (1 << 17) +# define DC_HPD6_RX_INTERRUPT (1 << 18) + +/* 0x6858, 0x7458, 0x10058, 0x10c58, 0x11858, 0x12458 */ +#define GRPH_INT_STATUS 0x6858 +# define GRPH_PFLIP_INT_OCCURRED (1 << 0) +# define GRPH_PFLIP_INT_CLEAR (1 << 8) +/* 0x685c, 0x745c, 0x1005c, 0x10c5c, 0x1185c, 0x1245c */ +#define GRPH_INT_CONTROL 0x685c +# define GRPH_PFLIP_INT_MASK (1 << 0) +# define GRPH_PFLIP_INT_TYPE (1 << 8) + +#define DACA_AUTODETECT_INT_CONTROL 0x66c8 +#define DACB_AUTODETECT_INT_CONTROL 0x67c8 + +#define DC_HPD1_INT_STATUS 0x601c +#define DC_HPD2_INT_STATUS 0x6028 +#define DC_HPD3_INT_STATUS 0x6034 +#define DC_HPD4_INT_STATUS 0x6040 +#define DC_HPD5_INT_STATUS 0x604c +#define DC_HPD6_INT_STATUS 0x6058 +# define DC_HPDx_INT_STATUS (1 << 0) +# define DC_HPDx_SENSE (1 << 1) +# define DC_HPDx_RX_INT_STATUS (1 << 8) + +#define DC_HPD1_INT_CONTROL 0x6020 +#define DC_HPD2_INT_CONTROL 0x602c +#define DC_HPD3_INT_CONTROL 0x6038 +#define DC_HPD4_INT_CONTROL 0x6044 +#define DC_HPD5_INT_CONTROL 0x6050 +#define DC_HPD6_INT_CONTROL 0x605c +# define DC_HPDx_INT_ACK (1 << 0) +# define DC_HPDx_INT_POLARITY (1 << 8) +# define DC_HPDx_INT_EN (1 << 16) +# define DC_HPDx_RX_INT_ACK (1 << 20) +# define DC_HPDx_RX_INT_EN (1 << 24) + +#define DC_HPD1_CONTROL 0x6024 +#define DC_HPD2_CONTROL 0x6030 +#define DC_HPD3_CONTROL 0x603c +#define DC_HPD4_CONTROL 0x6048 +#define DC_HPD5_CONTROL 0x6054 +#define DC_HPD6_CONTROL 0x6060 +# define DC_HPDx_CONNECTION_TIMER(x) ((x) << 0) +# define DC_HPDx_RX_INT_TIMER(x) ((x) << 16) +# define DC_HPDx_EN (1 << 28) + +/* PCIE link stuff */ +#define PCIE_LC_TRAINING_CNTL 0xa1 /* PCIE_P */ +#define PCIE_LC_LINK_WIDTH_CNTL 0xa2 /* PCIE_P */ +# define LC_LINK_WIDTH_SHIFT 0 +# define LC_LINK_WIDTH_MASK 0x7 +# define LC_LINK_WIDTH_X0 0 +# define LC_LINK_WIDTH_X1 1 +# define LC_LINK_WIDTH_X2 2 +# define LC_LINK_WIDTH_X4 3 +# define LC_LINK_WIDTH_X8 4 +# define LC_LINK_WIDTH_X16 6 +# define LC_LINK_WIDTH_RD_SHIFT 4 +# define LC_LINK_WIDTH_RD_MASK 0x70 +# define LC_RECONFIG_ARC_MISSING_ESCAPE (1 << 7) +# define LC_RECONFIG_NOW (1 << 8) +# define LC_RENEGOTIATION_SUPPORT (1 << 9) +# define LC_RENEGOTIATE_EN (1 << 10) +# define LC_SHORT_RECONFIG_EN (1 << 11) +# define LC_UPCONFIGURE_SUPPORT (1 << 12) +# define LC_UPCONFIGURE_DIS (1 << 13) +#define PCIE_LC_SPEED_CNTL 0xa4 /* PCIE_P */ +# define LC_GEN2_EN_STRAP (1 << 0) +# define LC_TARGET_LINK_SPEED_OVERRIDE_EN (1 << 1) +# define LC_FORCE_EN_HW_SPEED_CHANGE (1 << 5) +# define LC_FORCE_DIS_HW_SPEED_CHANGE (1 << 6) +# define LC_SPEED_CHANGE_ATTEMPTS_ALLOWED_MASK (0x3 << 8) +# define LC_SPEED_CHANGE_ATTEMPTS_ALLOWED_SHIFT 3 +# define LC_CURRENT_DATA_RATE (1 << 11) +# define LC_VOLTAGE_TIMER_SEL_MASK (0xf << 14) +# define LC_CLR_FAILED_SPD_CHANGE_CNT (1 << 21) +# define LC_OTHER_SIDE_EVER_SENT_GEN2 (1 << 23) +# define LC_OTHER_SIDE_SUPPORTS_GEN2 (1 << 24) +#define MM_CFGREGS_CNTL 0x544c +# define MM_WR_TO_CFG_EN (1 << 3) +#define LINK_CNTL2 0x88 /* F0 */ +# define TARGET_LINK_SPEED_MASK (0xf << 0) +# define SELECTABLE_DEEMPHASIS (1 << 6) + +/* + * PM4 + */ +#define PACKET_TYPE0 0 +#define PACKET_TYPE1 1 +#define PACKET_TYPE2 2 +#define PACKET_TYPE3 3 + +#define CP_PACKET_GET_TYPE(h) (((h) >> 30) & 3) +#define CP_PACKET_GET_COUNT(h) (((h) >> 16) & 0x3FFF) +#define CP_PACKET0_GET_REG(h) (((h) & 0xFFFF) << 2) +#define CP_PACKET3_GET_OPCODE(h) (((h) >> 8) & 0xFF) +#define PACKET0(reg, n) ((PACKET_TYPE0 << 30) | \ + (((reg) >> 2) & 0xFFFF) | \ + ((n) & 0x3FFF) << 16) +#define CP_PACKET2 0x80000000 +#define PACKET2_PAD_SHIFT 0 +#define PACKET2_PAD_MASK (0x3fffffff << 0) + +#define PACKET2(v) (CP_PACKET2 | REG_SET(PACKET2_PAD, (v))) + +#define PACKET3(op, n) ((PACKET_TYPE3 << 30) | \ + (((op) & 0xFF) << 8) | \ + ((n) & 0x3FFF) << 16) + +/* Packet 3 types */ +#define PACKET3_NOP 0x10 +#define PACKET3_SET_BASE 0x11 +#define PACKET3_CLEAR_STATE 0x12 +#define PACKET3_INDEX_BUFFER_SIZE 0x13 +#define PACKET3_DISPATCH_DIRECT 0x15 +#define PACKET3_DISPATCH_INDIRECT 0x16 +#define PACKET3_INDIRECT_BUFFER_END 0x17 +#define PACKET3_MODE_CONTROL 0x18 +#define PACKET3_SET_PREDICATION 0x20 +#define PACKET3_REG_RMW 0x21 +#define PACKET3_COND_EXEC 0x22 +#define PACKET3_PRED_EXEC 0x23 +#define PACKET3_DRAW_INDIRECT 0x24 +#define PACKET3_DRAW_INDEX_INDIRECT 0x25 +#define PACKET3_INDEX_BASE 0x26 +#define PACKET3_DRAW_INDEX_2 0x27 +#define PACKET3_CONTEXT_CONTROL 0x28 +#define PACKET3_DRAW_INDEX_OFFSET 0x29 +#define PACKET3_INDEX_TYPE 0x2A +#define PACKET3_DRAW_INDEX 0x2B +#define PACKET3_DRAW_INDEX_AUTO 0x2D +#define PACKET3_DRAW_INDEX_IMMD 0x2E +#define PACKET3_NUM_INSTANCES 0x2F +#define PACKET3_DRAW_INDEX_MULTI_AUTO 0x30 +#define PACKET3_STRMOUT_BUFFER_UPDATE 0x34 +#define PACKET3_DRAW_INDEX_OFFSET_2 0x35 +#define PACKET3_DRAW_INDEX_MULTI_ELEMENT 0x36 +#define PACKET3_MEM_SEMAPHORE 0x39 +#define PACKET3_MPEG_INDEX 0x3A +#define PACKET3_WAIT_REG_MEM 0x3C +#define PACKET3_MEM_WRITE 0x3D +#define PACKET3_INDIRECT_BUFFER 0x32 +#define PACKET3_SURFACE_SYNC 0x43 +# define PACKET3_CB0_DEST_BASE_ENA (1 << 6) +# define PACKET3_CB1_DEST_BASE_ENA (1 << 7) +# define PACKET3_CB2_DEST_BASE_ENA (1 << 8) +# define PACKET3_CB3_DEST_BASE_ENA (1 << 9) +# define PACKET3_CB4_DEST_BASE_ENA (1 << 10) +# define PACKET3_CB5_DEST_BASE_ENA (1 << 11) +# define PACKET3_CB6_DEST_BASE_ENA (1 << 12) +# define PACKET3_CB7_DEST_BASE_ENA (1 << 13) +# define PACKET3_DB_DEST_BASE_ENA (1 << 14) +# define PACKET3_CB8_DEST_BASE_ENA (1 << 15) +# define PACKET3_CB9_DEST_BASE_ENA (1 << 16) +# define PACKET3_CB10_DEST_BASE_ENA (1 << 17) +# define PACKET3_CB11_DEST_BASE_ENA (1 << 18) +# define PACKET3_FULL_CACHE_ENA (1 << 20) +# define PACKET3_TC_ACTION_ENA (1 << 23) +# define PACKET3_VC_ACTION_ENA (1 << 24) +# define PACKET3_CB_ACTION_ENA (1 << 25) +# define PACKET3_DB_ACTION_ENA (1 << 26) +# define PACKET3_SH_ACTION_ENA (1 << 27) +# define PACKET3_SX_ACTION_ENA (1 << 28) +#define PACKET3_ME_INITIALIZE 0x44 +#define PACKET3_ME_INITIALIZE_DEVICE_ID(x) ((x) << 16) +#define PACKET3_COND_WRITE 0x45 +#define PACKET3_EVENT_WRITE 0x46 +#define PACKET3_EVENT_WRITE_EOP 0x47 +#define PACKET3_EVENT_WRITE_EOS 0x48 +#define PACKET3_PREAMBLE_CNTL 0x4A +# define PACKET3_PREAMBLE_BEGIN_CLEAR_STATE (2 << 28) +# define PACKET3_PREAMBLE_END_CLEAR_STATE (3 << 28) +#define PACKET3_RB_OFFSET 0x4B +#define PACKET3_ALU_PS_CONST_BUFFER_COPY 0x4C +#define PACKET3_ALU_VS_CONST_BUFFER_COPY 0x4D +#define PACKET3_ALU_PS_CONST_UPDATE 0x4E +#define PACKET3_ALU_VS_CONST_UPDATE 0x4F +#define PACKET3_ONE_REG_WRITE 0x57 +#define PACKET3_SET_CONFIG_REG 0x68 +#define PACKET3_SET_CONFIG_REG_START 0x00008000 +#define PACKET3_SET_CONFIG_REG_END 0x0000ac00 +#define PACKET3_SET_CONTEXT_REG 0x69 +#define PACKET3_SET_CONTEXT_REG_START 0x00028000 +#define PACKET3_SET_CONTEXT_REG_END 0x00029000 +#define PACKET3_SET_ALU_CONST 0x6A +/* alu const buffers only; no reg file */ +#define PACKET3_SET_BOOL_CONST 0x6B +#define PACKET3_SET_BOOL_CONST_START 0x0003a500 +#define PACKET3_SET_BOOL_CONST_END 0x0003a518 +#define PACKET3_SET_LOOP_CONST 0x6C +#define PACKET3_SET_LOOP_CONST_START 0x0003a200 +#define PACKET3_SET_LOOP_CONST_END 0x0003a500 +#define PACKET3_SET_RESOURCE 0x6D +#define PACKET3_SET_RESOURCE_START 0x00030000 +#define PACKET3_SET_RESOURCE_END 0x00038000 +#define PACKET3_SET_SAMPLER 0x6E +#define PACKET3_SET_SAMPLER_START 0x0003c000 +#define PACKET3_SET_SAMPLER_END 0x0003c600 +#define PACKET3_SET_CTL_CONST 0x6F +#define PACKET3_SET_CTL_CONST_START 0x0003cff0 +#define PACKET3_SET_CTL_CONST_END 0x0003ff0c +#define PACKET3_SET_RESOURCE_OFFSET 0x70 +#define PACKET3_SET_ALU_CONST_VS 0x71 +#define PACKET3_SET_ALU_CONST_DI 0x72 +#define PACKET3_SET_CONTEXT_REG_INDIRECT 0x73 +#define PACKET3_SET_RESOURCE_INDIRECT 0x74 +#define PACKET3_SET_APPEND_CNT 0x75 + +#define SQ_RESOURCE_CONSTANT_WORD7_0 0x3001c +#define S__SQ_CONSTANT_TYPE(x) (((x) & 3) << 30) +#define G__SQ_CONSTANT_TYPE(x) (((x) >> 30) & 3) +#define SQ_TEX_VTX_INVALID_TEXTURE 0x0 +#define SQ_TEX_VTX_INVALID_BUFFER 0x1 +#define SQ_TEX_VTX_VALID_TEXTURE 0x2 +#define SQ_TEX_VTX_VALID_BUFFER 0x3 + +#define SQ_CONST_MEM_BASE 0x8df8 + +#define SQ_ESGS_RING_BASE 0x8c40 +#define SQ_ESGS_RING_SIZE 0x8c44 +#define SQ_GSVS_RING_BASE 0x8c48 +#define SQ_GSVS_RING_SIZE 0x8c4c +#define SQ_ESTMP_RING_BASE 0x8c50 +#define SQ_ESTMP_RING_SIZE 0x8c54 +#define SQ_GSTMP_RING_BASE 0x8c58 +#define SQ_GSTMP_RING_SIZE 0x8c5c +#define SQ_VSTMP_RING_BASE 0x8c60 +#define SQ_VSTMP_RING_SIZE 0x8c64 +#define SQ_PSTMP_RING_BASE 0x8c68 +#define SQ_PSTMP_RING_SIZE 0x8c6c +#define SQ_LSTMP_RING_BASE 0x8e10 +#define SQ_LSTMP_RING_SIZE 0x8e14 +#define SQ_HSTMP_RING_BASE 0x8e18 +#define SQ_HSTMP_RING_SIZE 0x8e1c +#define VGT_TF_RING_SIZE 0x8988 + +#define SQ_ESGS_RING_ITEMSIZE 0x28900 +#define SQ_GSVS_RING_ITEMSIZE 0x28904 +#define SQ_ESTMP_RING_ITEMSIZE 0x28908 +#define SQ_GSTMP_RING_ITEMSIZE 0x2890c +#define SQ_VSTMP_RING_ITEMSIZE 0x28910 +#define SQ_PSTMP_RING_ITEMSIZE 0x28914 +#define SQ_LSTMP_RING_ITEMSIZE 0x28830 +#define SQ_HSTMP_RING_ITEMSIZE 0x28834 + +#define SQ_GS_VERT_ITEMSIZE 0x2891c +#define SQ_GS_VERT_ITEMSIZE_1 0x28920 +#define SQ_GS_VERT_ITEMSIZE_2 0x28924 +#define SQ_GS_VERT_ITEMSIZE_3 0x28928 +#define SQ_GSVS_RING_OFFSET_1 0x2892c +#define SQ_GSVS_RING_OFFSET_2 0x28930 +#define SQ_GSVS_RING_OFFSET_3 0x28934 + +#define SQ_ALU_CONST_BUFFER_SIZE_PS_0 0x28140 +#define SQ_ALU_CONST_BUFFER_SIZE_HS_0 0x28f80 + +#define SQ_ALU_CONST_CACHE_PS_0 0x28940 +#define SQ_ALU_CONST_CACHE_PS_1 0x28944 +#define SQ_ALU_CONST_CACHE_PS_2 0x28948 +#define SQ_ALU_CONST_CACHE_PS_3 0x2894c +#define SQ_ALU_CONST_CACHE_PS_4 0x28950 +#define SQ_ALU_CONST_CACHE_PS_5 0x28954 +#define SQ_ALU_CONST_CACHE_PS_6 0x28958 +#define SQ_ALU_CONST_CACHE_PS_7 0x2895c +#define SQ_ALU_CONST_CACHE_PS_8 0x28960 +#define SQ_ALU_CONST_CACHE_PS_9 0x28964 +#define SQ_ALU_CONST_CACHE_PS_10 0x28968 +#define SQ_ALU_CONST_CACHE_PS_11 0x2896c +#define SQ_ALU_CONST_CACHE_PS_12 0x28970 +#define SQ_ALU_CONST_CACHE_PS_13 0x28974 +#define SQ_ALU_CONST_CACHE_PS_14 0x28978 +#define SQ_ALU_CONST_CACHE_PS_15 0x2897c +#define SQ_ALU_CONST_CACHE_VS_0 0x28980 +#define SQ_ALU_CONST_CACHE_VS_1 0x28984 +#define SQ_ALU_CONST_CACHE_VS_2 0x28988 +#define SQ_ALU_CONST_CACHE_VS_3 0x2898c +#define SQ_ALU_CONST_CACHE_VS_4 0x28990 +#define SQ_ALU_CONST_CACHE_VS_5 0x28994 +#define SQ_ALU_CONST_CACHE_VS_6 0x28998 +#define SQ_ALU_CONST_CACHE_VS_7 0x2899c +#define SQ_ALU_CONST_CACHE_VS_8 0x289a0 +#define SQ_ALU_CONST_CACHE_VS_9 0x289a4 +#define SQ_ALU_CONST_CACHE_VS_10 0x289a8 +#define SQ_ALU_CONST_CACHE_VS_11 0x289ac +#define SQ_ALU_CONST_CACHE_VS_12 0x289b0 +#define SQ_ALU_CONST_CACHE_VS_13 0x289b4 +#define SQ_ALU_CONST_CACHE_VS_14 0x289b8 +#define SQ_ALU_CONST_CACHE_VS_15 0x289bc +#define SQ_ALU_CONST_CACHE_GS_0 0x289c0 +#define SQ_ALU_CONST_CACHE_GS_1 0x289c4 +#define SQ_ALU_CONST_CACHE_GS_2 0x289c8 +#define SQ_ALU_CONST_CACHE_GS_3 0x289cc +#define SQ_ALU_CONST_CACHE_GS_4 0x289d0 +#define SQ_ALU_CONST_CACHE_GS_5 0x289d4 +#define SQ_ALU_CONST_CACHE_GS_6 0x289d8 +#define SQ_ALU_CONST_CACHE_GS_7 0x289dc +#define SQ_ALU_CONST_CACHE_GS_8 0x289e0 +#define SQ_ALU_CONST_CACHE_GS_9 0x289e4 +#define SQ_ALU_CONST_CACHE_GS_10 0x289e8 +#define SQ_ALU_CONST_CACHE_GS_11 0x289ec +#define SQ_ALU_CONST_CACHE_GS_12 0x289f0 +#define SQ_ALU_CONST_CACHE_GS_13 0x289f4 +#define SQ_ALU_CONST_CACHE_GS_14 0x289f8 +#define SQ_ALU_CONST_CACHE_GS_15 0x289fc +#define SQ_ALU_CONST_CACHE_HS_0 0x28f00 +#define SQ_ALU_CONST_CACHE_HS_1 0x28f04 +#define SQ_ALU_CONST_CACHE_HS_2 0x28f08 +#define SQ_ALU_CONST_CACHE_HS_3 0x28f0c +#define SQ_ALU_CONST_CACHE_HS_4 0x28f10 +#define SQ_ALU_CONST_CACHE_HS_5 0x28f14 +#define SQ_ALU_CONST_CACHE_HS_6 0x28f18 +#define SQ_ALU_CONST_CACHE_HS_7 0x28f1c +#define SQ_ALU_CONST_CACHE_HS_8 0x28f20 +#define SQ_ALU_CONST_CACHE_HS_9 0x28f24 +#define SQ_ALU_CONST_CACHE_HS_10 0x28f28 +#define SQ_ALU_CONST_CACHE_HS_11 0x28f2c +#define SQ_ALU_CONST_CACHE_HS_12 0x28f30 +#define SQ_ALU_CONST_CACHE_HS_13 0x28f34 +#define SQ_ALU_CONST_CACHE_HS_14 0x28f38 +#define SQ_ALU_CONST_CACHE_HS_15 0x28f3c +#define SQ_ALU_CONST_CACHE_LS_0 0x28f40 +#define SQ_ALU_CONST_CACHE_LS_1 0x28f44 +#define SQ_ALU_CONST_CACHE_LS_2 0x28f48 +#define SQ_ALU_CONST_CACHE_LS_3 0x28f4c +#define SQ_ALU_CONST_CACHE_LS_4 0x28f50 +#define SQ_ALU_CONST_CACHE_LS_5 0x28f54 +#define SQ_ALU_CONST_CACHE_LS_6 0x28f58 +#define SQ_ALU_CONST_CACHE_LS_7 0x28f5c +#define SQ_ALU_CONST_CACHE_LS_8 0x28f60 +#define SQ_ALU_CONST_CACHE_LS_9 0x28f64 +#define SQ_ALU_CONST_CACHE_LS_10 0x28f68 +#define SQ_ALU_CONST_CACHE_LS_11 0x28f6c +#define SQ_ALU_CONST_CACHE_LS_12 0x28f70 +#define SQ_ALU_CONST_CACHE_LS_13 0x28f74 +#define SQ_ALU_CONST_CACHE_LS_14 0x28f78 +#define SQ_ALU_CONST_CACHE_LS_15 0x28f7c + +#define PA_SC_SCREEN_SCISSOR_TL 0x28030 +#define PA_SC_GENERIC_SCISSOR_TL 0x28240 +#define PA_SC_WINDOW_SCISSOR_TL 0x28204 +#define VGT_PRIMITIVE_TYPE 0x8958 + +#define DB_DEPTH_CONTROL 0x28800 +#define DB_DEPTH_VIEW 0x28008 +#define DB_HTILE_DATA_BASE 0x28014 +#define DB_Z_INFO 0x28040 +# define Z_ARRAY_MODE(x) ((x) << 4) +#define DB_STENCIL_INFO 0x28044 +#define DB_Z_READ_BASE 0x28048 +#define DB_STENCIL_READ_BASE 0x2804c +#define DB_Z_WRITE_BASE 0x28050 +#define DB_STENCIL_WRITE_BASE 0x28054 +#define DB_DEPTH_SIZE 0x28058 + +#define SQ_PGM_START_PS 0x28840 +#define SQ_PGM_START_VS 0x2885c +#define SQ_PGM_START_GS 0x28874 +#define SQ_PGM_START_ES 0x2888c +#define SQ_PGM_START_FS 0x288a4 +#define SQ_PGM_START_HS 0x288b8 +#define SQ_PGM_START_LS 0x288d0 + +#define VGT_STRMOUT_CONFIG 0x28b94 +#define VGT_STRMOUT_BUFFER_CONFIG 0x28b98 + +#define CB_TARGET_MASK 0x28238 +#define CB_SHADER_MASK 0x2823c + +#define GDS_ADDR_BASE 0x28720 + +#define CB_IMMED0_BASE 0x28b9c +#define CB_IMMED1_BASE 0x28ba0 +#define CB_IMMED2_BASE 0x28ba4 +#define CB_IMMED3_BASE 0x28ba8 +#define CB_IMMED4_BASE 0x28bac +#define CB_IMMED5_BASE 0x28bb0 +#define CB_IMMED6_BASE 0x28bb4 +#define CB_IMMED7_BASE 0x28bb8 +#define CB_IMMED8_BASE 0x28bbc +#define CB_IMMED9_BASE 0x28bc0 +#define CB_IMMED10_BASE 0x28bc4 +#define CB_IMMED11_BASE 0x28bc8 + +/* all 12 CB blocks have these regs */ +#define CB_COLOR0_BASE 0x28c60 +#define CB_COLOR0_PITCH 0x28c64 +#define CB_COLOR0_SLICE 0x28c68 +#define CB_COLOR0_VIEW 0x28c6c +#define CB_COLOR0_INFO 0x28c70 +# define CB_ARRAY_MODE(x) ((x) << 8) +# define ARRAY_LINEAR_GENERAL 0 +# define ARRAY_LINEAR_ALIGNED 1 +# define ARRAY_1D_TILED_THIN1 2 +# define ARRAY_2D_TILED_THIN1 4 +#define CB_COLOR0_ATTRIB 0x28c74 +#define CB_COLOR0_DIM 0x28c78 +/* only CB0-7 blocks have these regs */ +#define CB_COLOR0_CMASK 0x28c7c +#define CB_COLOR0_CMASK_SLICE 0x28c80 +#define CB_COLOR0_FMASK 0x28c84 +#define CB_COLOR0_FMASK_SLICE 0x28c88 +#define CB_COLOR0_CLEAR_WORD0 0x28c8c +#define CB_COLOR0_CLEAR_WORD1 0x28c90 +#define CB_COLOR0_CLEAR_WORD2 0x28c94 +#define CB_COLOR0_CLEAR_WORD3 0x28c98 + +#define CB_COLOR1_BASE 0x28c9c +#define CB_COLOR2_BASE 0x28cd8 +#define CB_COLOR3_BASE 0x28d14 +#define CB_COLOR4_BASE 0x28d50 +#define CB_COLOR5_BASE 0x28d8c +#define CB_COLOR6_BASE 0x28dc8 +#define CB_COLOR7_BASE 0x28e04 +#define CB_COLOR8_BASE 0x28e40 +#define CB_COLOR9_BASE 0x28e5c +#define CB_COLOR10_BASE 0x28e78 +#define CB_COLOR11_BASE 0x28e94 + +#define CB_COLOR1_PITCH 0x28ca0 +#define CB_COLOR2_PITCH 0x28cdc +#define CB_COLOR3_PITCH 0x28d18 +#define CB_COLOR4_PITCH 0x28d54 +#define CB_COLOR5_PITCH 0x28d90 +#define CB_COLOR6_PITCH 0x28dcc +#define CB_COLOR7_PITCH 0x28e08 +#define CB_COLOR8_PITCH 0x28e44 +#define CB_COLOR9_PITCH 0x28e60 +#define CB_COLOR10_PITCH 0x28e7c +#define CB_COLOR11_PITCH 0x28e98 + +#define CB_COLOR1_SLICE 0x28ca4 +#define CB_COLOR2_SLICE 0x28ce0 +#define CB_COLOR3_SLICE 0x28d1c +#define CB_COLOR4_SLICE 0x28d58 +#define CB_COLOR5_SLICE 0x28d94 +#define CB_COLOR6_SLICE 0x28dd0 +#define CB_COLOR7_SLICE 0x28e0c +#define CB_COLOR8_SLICE 0x28e48 +#define CB_COLOR9_SLICE 0x28e64 +#define CB_COLOR10_SLICE 0x28e80 +#define CB_COLOR11_SLICE 0x28e9c + +#define CB_COLOR1_VIEW 0x28ca8 +#define CB_COLOR2_VIEW 0x28ce4 +#define CB_COLOR3_VIEW 0x28d20 +#define CB_COLOR4_VIEW 0x28d5c +#define CB_COLOR5_VIEW 0x28d98 +#define CB_COLOR6_VIEW 0x28dd4 +#define CB_COLOR7_VIEW 0x28e10 +#define CB_COLOR8_VIEW 0x28e4c +#define CB_COLOR9_VIEW 0x28e68 +#define CB_COLOR10_VIEW 0x28e84 +#define CB_COLOR11_VIEW 0x28ea0 + +#define CB_COLOR1_INFO 0x28cac +#define CB_COLOR2_INFO 0x28ce8 +#define CB_COLOR3_INFO 0x28d24 +#define CB_COLOR4_INFO 0x28d60 +#define CB_COLOR5_INFO 0x28d9c +#define CB_COLOR6_INFO 0x28dd8 +#define CB_COLOR7_INFO 0x28e14 +#define CB_COLOR8_INFO 0x28e50 +#define CB_COLOR9_INFO 0x28e6c +#define CB_COLOR10_INFO 0x28e88 +#define CB_COLOR11_INFO 0x28ea4 + +#define CB_COLOR1_ATTRIB 0x28cb0 +#define CB_COLOR2_ATTRIB 0x28cec +#define CB_COLOR3_ATTRIB 0x28d28 +#define CB_COLOR4_ATTRIB 0x28d64 +#define CB_COLOR5_ATTRIB 0x28da0 +#define CB_COLOR6_ATTRIB 0x28ddc +#define CB_COLOR7_ATTRIB 0x28e18 +#define CB_COLOR8_ATTRIB 0x28e54 +#define CB_COLOR9_ATTRIB 0x28e70 +#define CB_COLOR10_ATTRIB 0x28e8c +#define CB_COLOR11_ATTRIB 0x28ea8 + +#define CB_COLOR1_DIM 0x28cb4 +#define CB_COLOR2_DIM 0x28cf0 +#define CB_COLOR3_DIM 0x28d2c +#define CB_COLOR4_DIM 0x28d68 +#define CB_COLOR5_DIM 0x28da4 +#define CB_COLOR6_DIM 0x28de0 +#define CB_COLOR7_DIM 0x28e1c +#define CB_COLOR8_DIM 0x28e58 +#define CB_COLOR9_DIM 0x28e74 +#define CB_COLOR10_DIM 0x28e90 +#define CB_COLOR11_DIM 0x28eac + +#define CB_COLOR1_CMASK 0x28cb8 +#define CB_COLOR2_CMASK 0x28cf4 +#define CB_COLOR3_CMASK 0x28d30 +#define CB_COLOR4_CMASK 0x28d6c +#define CB_COLOR5_CMASK 0x28da8 +#define CB_COLOR6_CMASK 0x28de4 +#define CB_COLOR7_CMASK 0x28e20 + +#define CB_COLOR1_CMASK_SLICE 0x28cbc +#define CB_COLOR2_CMASK_SLICE 0x28cf8 +#define CB_COLOR3_CMASK_SLICE 0x28d34 +#define CB_COLOR4_CMASK_SLICE 0x28d70 +#define CB_COLOR5_CMASK_SLICE 0x28dac +#define CB_COLOR6_CMASK_SLICE 0x28de8 +#define CB_COLOR7_CMASK_SLICE 0x28e24 + +#define CB_COLOR1_FMASK 0x28cc0 +#define CB_COLOR2_FMASK 0x28cfc +#define CB_COLOR3_FMASK 0x28d38 +#define CB_COLOR4_FMASK 0x28d74 +#define CB_COLOR5_FMASK 0x28db0 +#define CB_COLOR6_FMASK 0x28dec +#define CB_COLOR7_FMASK 0x28e28 + +#define CB_COLOR1_FMASK_SLICE 0x28cc4 +#define CB_COLOR2_FMASK_SLICE 0x28d00 +#define CB_COLOR3_FMASK_SLICE 0x28d3c +#define CB_COLOR4_FMASK_SLICE 0x28d78 +#define CB_COLOR5_FMASK_SLICE 0x28db4 +#define CB_COLOR6_FMASK_SLICE 0x28df0 +#define CB_COLOR7_FMASK_SLICE 0x28e2c + +#define CB_COLOR1_CLEAR_WORD0 0x28cc8 +#define CB_COLOR2_CLEAR_WORD0 0x28d04 +#define CB_COLOR3_CLEAR_WORD0 0x28d40 +#define CB_COLOR4_CLEAR_WORD0 0x28d7c +#define CB_COLOR5_CLEAR_WORD0 0x28db8 +#define CB_COLOR6_CLEAR_WORD0 0x28df4 +#define CB_COLOR7_CLEAR_WORD0 0x28e30 + +#define CB_COLOR1_CLEAR_WORD1 0x28ccc +#define CB_COLOR2_CLEAR_WORD1 0x28d08 +#define CB_COLOR3_CLEAR_WORD1 0x28d44 +#define CB_COLOR4_CLEAR_WORD1 0x28d80 +#define CB_COLOR5_CLEAR_WORD1 0x28dbc +#define CB_COLOR6_CLEAR_WORD1 0x28df8 +#define CB_COLOR7_CLEAR_WORD1 0x28e34 + +#define CB_COLOR1_CLEAR_WORD2 0x28cd0 +#define CB_COLOR2_CLEAR_WORD2 0x28d0c +#define CB_COLOR3_CLEAR_WORD2 0x28d48 +#define CB_COLOR4_CLEAR_WORD2 0x28d84 +#define CB_COLOR5_CLEAR_WORD2 0x28dc0 +#define CB_COLOR6_CLEAR_WORD2 0x28dfc +#define CB_COLOR7_CLEAR_WORD2 0x28e38 + +#define CB_COLOR1_CLEAR_WORD3 0x28cd4 +#define CB_COLOR2_CLEAR_WORD3 0x28d10 +#define CB_COLOR3_CLEAR_WORD3 0x28d4c +#define CB_COLOR4_CLEAR_WORD3 0x28d88 +#define CB_COLOR5_CLEAR_WORD3 0x28dc4 +#define CB_COLOR6_CLEAR_WORD3 0x28e00 +#define CB_COLOR7_CLEAR_WORD3 0x28e3c + +#define SQ_TEX_RESOURCE_WORD0_0 0x30000 +#define SQ_TEX_RESOURCE_WORD1_0 0x30004 +# define TEX_ARRAY_MODE(x) ((x) << 28) +#define SQ_TEX_RESOURCE_WORD2_0 0x30008 +#define SQ_TEX_RESOURCE_WORD3_0 0x3000C +#define SQ_TEX_RESOURCE_WORD4_0 0x30010 +#define SQ_TEX_RESOURCE_WORD5_0 0x30014 +#define SQ_TEX_RESOURCE_WORD6_0 0x30018 +#define SQ_TEX_RESOURCE_WORD7_0 0x3001c + +/* cayman 3D regs */ +#define CAYMAN_VGT_OFFCHIP_LDS_BASE 0x89B0 +#define CAYMAN_DB_EQAA 0x28804 +#define CAYMAN_DB_DEPTH_INFO 0x2803C +#define CAYMAN_PA_SC_AA_CONFIG 0x28BE0 +#define CAYMAN_MSAA_NUM_SAMPLES_SHIFT 0 +#define CAYMAN_MSAA_NUM_SAMPLES_MASK 0x7 +/* cayman packet3 addition */ +#define CAYMAN_PACKET3_DEALLOC_STATE 0x14 + +#endif diff --git a/drivers/gpu/drm/radeon/mkregtable.c b/drivers/gpu/drm/radeon/mkregtable.c new file mode 100644 index 00000000..5a82b6b7 --- /dev/null +++ b/drivers/gpu/drm/radeon/mkregtable.c @@ -0,0 +1,725 @@ +/* utility to create the register check tables + * this includes inlined list.h safe for userspace. + * + * Copyright 2009 Jerome Glisse + * Copyright 2009 Red Hat Inc. + * + * Authors: + * Jerome Glisse + * Dave Airlie + */ + +#include +#include +#include +#include +#include +#include + +#define offsetof(TYPE, MEMBER) ((size_t) &((TYPE *)0)->MEMBER) +/** + * container_of - cast a member of a structure out to the containing structure + * @ptr: the pointer to the member. + * @type: the type of the container struct this is embedded in. + * @member: the name of the member within the struct. + * + */ +#define container_of(ptr, type, member) ({ \ + const typeof(((type *)0)->member)*__mptr = (ptr); \ + (type *)((char *)__mptr - offsetof(type, member)); }) + +/* + * Simple doubly linked list implementation. + * + * Some of the internal functions ("__xxx") are useful when + * manipulating whole lists rather than single entries, as + * sometimes we already know the next/prev entries and we can + * generate better code by using them directly rather than + * using the generic single-entry routines. + */ + +struct list_head { + struct list_head *next, *prev; +}; + +#define LIST_HEAD_INIT(name) { &(name), &(name) } + +#define LIST_HEAD(name) \ + struct list_head name = LIST_HEAD_INIT(name) + +static inline void INIT_LIST_HEAD(struct list_head *list) +{ + list->next = list; + list->prev = list; +} + +/* + * Insert a new entry between two known consecutive entries. + * + * This is only for internal list manipulation where we know + * the prev/next entries already! + */ +#ifndef CONFIG_DEBUG_LIST +static inline void __list_add(struct list_head *new, + struct list_head *prev, struct list_head *next) +{ + next->prev = new; + new->next = next; + new->prev = prev; + prev->next = new; +} +#else +extern void __list_add(struct list_head *new, + struct list_head *prev, struct list_head *next); +#endif + +/** + * list_add - add a new entry + * @new: new entry to be added + * @head: list head to add it after + * + * Insert a new entry after the specified head. + * This is good for implementing stacks. + */ +static inline void list_add(struct list_head *new, struct list_head *head) +{ + __list_add(new, head, head->next); +} + +/** + * list_add_tail - add a new entry + * @new: new entry to be added + * @head: list head to add it before + * + * Insert a new entry before the specified head. + * This is useful for implementing queues. + */ +static inline void list_add_tail(struct list_head *new, struct list_head *head) +{ + __list_add(new, head->prev, head); +} + +/* + * Delete a list entry by making the prev/next entries + * point to each other. + * + * This is only for internal list manipulation where we know + * the prev/next entries already! + */ +static inline void __list_del(struct list_head *prev, struct list_head *next) +{ + next->prev = prev; + prev->next = next; +} + +/** + * list_del - deletes entry from list. + * @entry: the element to delete from the list. + * Note: list_empty() on entry does not return true after this, the entry is + * in an undefined state. + */ +#ifndef CONFIG_DEBUG_LIST +static inline void list_del(struct list_head *entry) +{ + __list_del(entry->prev, entry->next); + entry->next = (void *)0xDEADBEEF; + entry->prev = (void *)0xBEEFDEAD; +} +#else +extern void list_del(struct list_head *entry); +#endif + +/** + * list_replace - replace old entry by new one + * @old : the element to be replaced + * @new : the new element to insert + * + * If @old was empty, it will be overwritten. + */ +static inline void list_replace(struct list_head *old, struct list_head *new) +{ + new->next = old->next; + new->next->prev = new; + new->prev = old->prev; + new->prev->next = new; +} + +static inline void list_replace_init(struct list_head *old, + struct list_head *new) +{ + list_replace(old, new); + INIT_LIST_HEAD(old); +} + +/** + * list_del_init - deletes entry from list and reinitialize it. + * @entry: the element to delete from the list. + */ +static inline void list_del_init(struct list_head *entry) +{ + __list_del(entry->prev, entry->next); + INIT_LIST_HEAD(entry); +} + +/** + * list_move - delete from one list and add as another's head + * @list: the entry to move + * @head: the head that will precede our entry + */ +static inline void list_move(struct list_head *list, struct list_head *head) +{ + __list_del(list->prev, list->next); + list_add(list, head); +} + +/** + * list_move_tail - delete from one list and add as another's tail + * @list: the entry to move + * @head: the head that will follow our entry + */ +static inline void list_move_tail(struct list_head *list, + struct list_head *head) +{ + __list_del(list->prev, list->next); + list_add_tail(list, head); +} + +/** + * list_is_last - tests whether @list is the last entry in list @head + * @list: the entry to test + * @head: the head of the list + */ +static inline int list_is_last(const struct list_head *list, + const struct list_head *head) +{ + return list->next == head; +} + +/** + * list_empty - tests whether a list is empty + * @head: the list to test. + */ +static inline int list_empty(const struct list_head *head) +{ + return head->next == head; +} + +/** + * list_empty_careful - tests whether a list is empty and not being modified + * @head: the list to test + * + * Description: + * tests whether a list is empty _and_ checks that no other CPU might be + * in the process of modifying either member (next or prev) + * + * NOTE: using list_empty_careful() without synchronization + * can only be safe if the only activity that can happen + * to the list entry is list_del_init(). Eg. it cannot be used + * if another CPU could re-list_add() it. + */ +static inline int list_empty_careful(const struct list_head *head) +{ + struct list_head *next = head->next; + return (next == head) && (next == head->prev); +} + +/** + * list_is_singular - tests whether a list has just one entry. + * @head: the list to test. + */ +static inline int list_is_singular(const struct list_head *head) +{ + return !list_empty(head) && (head->next == head->prev); +} + +static inline void __list_cut_position(struct list_head *list, + struct list_head *head, + struct list_head *entry) +{ + struct list_head *new_first = entry->next; + list->next = head->next; + list->next->prev = list; + list->prev = entry; + entry->next = list; + head->next = new_first; + new_first->prev = head; +} + +/** + * list_cut_position - cut a list into two + * @list: a new list to add all removed entries + * @head: a list with entries + * @entry: an entry within head, could be the head itself + * and if so we won't cut the list + * + * This helper moves the initial part of @head, up to and + * including @entry, from @head to @list. You should + * pass on @entry an element you know is on @head. @list + * should be an empty list or a list you do not care about + * losing its data. + * + */ +static inline void list_cut_position(struct list_head *list, + struct list_head *head, + struct list_head *entry) +{ + if (list_empty(head)) + return; + if (list_is_singular(head) && (head->next != entry && head != entry)) + return; + if (entry == head) + INIT_LIST_HEAD(list); + else + __list_cut_position(list, head, entry); +} + +static inline void __list_splice(const struct list_head *list, + struct list_head *prev, struct list_head *next) +{ + struct list_head *first = list->next; + struct list_head *last = list->prev; + + first->prev = prev; + prev->next = first; + + last->next = next; + next->prev = last; +} + +/** + * list_splice - join two lists, this is designed for stacks + * @list: the new list to add. + * @head: the place to add it in the first list. + */ +static inline void list_splice(const struct list_head *list, + struct list_head *head) +{ + if (!list_empty(list)) + __list_splice(list, head, head->next); +} + +/** + * list_splice_tail - join two lists, each list being a queue + * @list: the new list to add. + * @head: the place to add it in the first list. + */ +static inline void list_splice_tail(struct list_head *list, + struct list_head *head) +{ + if (!list_empty(list)) + __list_splice(list, head->prev, head); +} + +/** + * list_splice_init - join two lists and reinitialise the emptied list. + * @list: the new list to add. + * @head: the place to add it in the first list. + * + * The list at @list is reinitialised + */ +static inline void list_splice_init(struct list_head *list, + struct list_head *head) +{ + if (!list_empty(list)) { + __list_splice(list, head, head->next); + INIT_LIST_HEAD(list); + } +} + +/** + * list_splice_tail_init - join two lists and reinitialise the emptied list + * @list: the new list to add. + * @head: the place to add it in the first list. + * + * Each of the lists is a queue. + * The list at @list is reinitialised + */ +static inline void list_splice_tail_init(struct list_head *list, + struct list_head *head) +{ + if (!list_empty(list)) { + __list_splice(list, head->prev, head); + INIT_LIST_HEAD(list); + } +} + +/** + * list_entry - get the struct for this entry + * @ptr: the &struct list_head pointer. + * @type: the type of the struct this is embedded in. + * @member: the name of the list_struct within the struct. + */ +#define list_entry(ptr, type, member) \ + container_of(ptr, type, member) + +/** + * list_first_entry - get the first element from a list + * @ptr: the list head to take the element from. + * @type: the type of the struct this is embedded in. + * @member: the name of the list_struct within the struct. + * + * Note, that list is expected to be not empty. + */ +#define list_first_entry(ptr, type, member) \ + list_entry((ptr)->next, type, member) + +/** + * list_for_each - iterate over a list + * @pos: the &struct list_head to use as a loop cursor. + * @head: the head for your list. + */ +#define list_for_each(pos, head) \ + for (pos = (head)->next; prefetch(pos->next), pos != (head); \ + pos = pos->next) + +/** + * __list_for_each - iterate over a list + * @pos: the &struct list_head to use as a loop cursor. + * @head: the head for your list. + * + * This variant differs from list_for_each() in that it's the + * simplest possible list iteration code, no prefetching is done. + * Use this for code that knows the list to be very short (empty + * or 1 entry) most of the time. + */ +#define __list_for_each(pos, head) \ + for (pos = (head)->next; pos != (head); pos = pos->next) + +/** + * list_for_each_prev - iterate over a list backwards + * @pos: the &struct list_head to use as a loop cursor. + * @head: the head for your list. + */ +#define list_for_each_prev(pos, head) \ + for (pos = (head)->prev; prefetch(pos->prev), pos != (head); \ + pos = pos->prev) + +/** + * list_for_each_safe - iterate over a list safe against removal of list entry + * @pos: the &struct list_head to use as a loop cursor. + * @n: another &struct list_head to use as temporary storage + * @head: the head for your list. + */ +#define list_for_each_safe(pos, n, head) \ + for (pos = (head)->next, n = pos->next; pos != (head); \ + pos = n, n = pos->next) + +/** + * list_for_each_prev_safe - iterate over a list backwards safe against removal of list entry + * @pos: the &struct list_head to use as a loop cursor. + * @n: another &struct list_head to use as temporary storage + * @head: the head for your list. + */ +#define list_for_each_prev_safe(pos, n, head) \ + for (pos = (head)->prev, n = pos->prev; \ + prefetch(pos->prev), pos != (head); \ + pos = n, n = pos->prev) + +/** + * list_for_each_entry - iterate over list of given type + * @pos: the type * to use as a loop cursor. + * @head: the head for your list. + * @member: the name of the list_struct within the struct. + */ +#define list_for_each_entry(pos, head, member) \ + for (pos = list_entry((head)->next, typeof(*pos), member); \ + &pos->member != (head); \ + pos = list_entry(pos->member.next, typeof(*pos), member)) + +/** + * list_for_each_entry_reverse - iterate backwards over list of given type. + * @pos: the type * to use as a loop cursor. + * @head: the head for your list. + * @member: the name of the list_struct within the struct. + */ +#define list_for_each_entry_reverse(pos, head, member) \ + for (pos = list_entry((head)->prev, typeof(*pos), member); \ + prefetch(pos->member.prev), &pos->member != (head); \ + pos = list_entry(pos->member.prev, typeof(*pos), member)) + +/** + * list_prepare_entry - prepare a pos entry for use in list_for_each_entry_continue() + * @pos: the type * to use as a start point + * @head: the head of the list + * @member: the name of the list_struct within the struct. + * + * Prepares a pos entry for use as a start point in list_for_each_entry_continue(). + */ +#define list_prepare_entry(pos, head, member) \ + ((pos) ? : list_entry(head, typeof(*pos), member)) + +/** + * list_for_each_entry_continue - continue iteration over list of given type + * @pos: the type * to use as a loop cursor. + * @head: the head for your list. + * @member: the name of the list_struct within the struct. + * + * Continue to iterate over list of given type, continuing after + * the current position. + */ +#define list_for_each_entry_continue(pos, head, member) \ + for (pos = list_entry(pos->member.next, typeof(*pos), member); \ + prefetch(pos->member.next), &pos->member != (head); \ + pos = list_entry(pos->member.next, typeof(*pos), member)) + +/** + * list_for_each_entry_continue_reverse - iterate backwards from the given point + * @pos: the type * to use as a loop cursor. + * @head: the head for your list. + * @member: the name of the list_struct within the struct. + * + * Start to iterate over list of given type backwards, continuing after + * the current position. + */ +#define list_for_each_entry_continue_reverse(pos, head, member) \ + for (pos = list_entry(pos->member.prev, typeof(*pos), member); \ + prefetch(pos->member.prev), &pos->member != (head); \ + pos = list_entry(pos->member.prev, typeof(*pos), member)) + +/** + * list_for_each_entry_from - iterate over list of given type from the current point + * @pos: the type * to use as a loop cursor. + * @head: the head for your list. + * @member: the name of the list_struct within the struct. + * + * Iterate over list of given type, continuing from current position. + */ +#define list_for_each_entry_from(pos, head, member) \ + for (; prefetch(pos->member.next), &pos->member != (head); \ + pos = list_entry(pos->member.next, typeof(*pos), member)) + +/** + * list_for_each_entry_safe - iterate over list of given type safe against removal of list entry + * @pos: the type * to use as a loop cursor. + * @n: another type * to use as temporary storage + * @head: the head for your list. + * @member: the name of the list_struct within the struct. + */ +#define list_for_each_entry_safe(pos, n, head, member) \ + for (pos = list_entry((head)->next, typeof(*pos), member), \ + n = list_entry(pos->member.next, typeof(*pos), member); \ + &pos->member != (head); \ + pos = n, n = list_entry(n->member.next, typeof(*n), member)) + +/** + * list_for_each_entry_safe_continue + * @pos: the type * to use as a loop cursor. + * @n: another type * to use as temporary storage + * @head: the head for your list. + * @member: the name of the list_struct within the struct. + * + * Iterate over list of given type, continuing after current point, + * safe against removal of list entry. + */ +#define list_for_each_entry_safe_continue(pos, n, head, member) \ + for (pos = list_entry(pos->member.next, typeof(*pos), member), \ + n = list_entry(pos->member.next, typeof(*pos), member); \ + &pos->member != (head); \ + pos = n, n = list_entry(n->member.next, typeof(*n), member)) + +/** + * list_for_each_entry_safe_from + * @pos: the type * to use as a loop cursor. + * @n: another type * to use as temporary storage + * @head: the head for your list. + * @member: the name of the list_struct within the struct. + * + * Iterate over list of given type from current point, safe against + * removal of list entry. + */ +#define list_for_each_entry_safe_from(pos, n, head, member) \ + for (n = list_entry(pos->member.next, typeof(*pos), member); \ + &pos->member != (head); \ + pos = n, n = list_entry(n->member.next, typeof(*n), member)) + +/** + * list_for_each_entry_safe_reverse + * @pos: the type * to use as a loop cursor. + * @n: another type * to use as temporary storage + * @head: the head for your list. + * @member: the name of the list_struct within the struct. + * + * Iterate backwards over list of given type, safe against removal + * of list entry. + */ +#define list_for_each_entry_safe_reverse(pos, n, head, member) \ + for (pos = list_entry((head)->prev, typeof(*pos), member), \ + n = list_entry(pos->member.prev, typeof(*pos), member); \ + &pos->member != (head); \ + pos = n, n = list_entry(n->member.prev, typeof(*n), member)) + +struct offset { + struct list_head list; + unsigned offset; +}; + +struct table { + struct list_head offsets; + unsigned offset_max; + unsigned nentry; + unsigned *table; + char *gpu_prefix; +}; + +static struct offset *offset_new(unsigned o) +{ + struct offset *offset; + + offset = (struct offset *)malloc(sizeof(struct offset)); + if (offset) { + INIT_LIST_HEAD(&offset->list); + offset->offset = o; + } + return offset; +} + +static void table_offset_add(struct table *t, struct offset *offset) +{ + list_add_tail(&offset->list, &t->offsets); +} + +static void table_init(struct table *t) +{ + INIT_LIST_HEAD(&t->offsets); + t->offset_max = 0; + t->nentry = 0; + t->table = NULL; +} + +static void table_print(struct table *t) +{ + unsigned nlloop, i, j, n, c, id; + + nlloop = (t->nentry + 3) / 4; + c = t->nentry; + printf("static const unsigned %s_reg_safe_bm[%d] = {\n", t->gpu_prefix, + t->nentry); + for (i = 0, id = 0; i < nlloop; i++) { + n = 4; + if (n > c) + n = c; + c -= n; + for (j = 0; j < n; j++) { + if (j == 0) + printf("\t"); + else + printf(" "); + printf("0x%08X,", t->table[id++]); + } + printf("\n"); + } + printf("};\n"); +} + +static int table_build(struct table *t) +{ + struct offset *offset; + unsigned i, m; + + t->nentry = ((t->offset_max >> 2) + 31) / 32; + t->table = (unsigned *)malloc(sizeof(unsigned) * t->nentry); + if (t->table == NULL) + return -1; + memset(t->table, 0xff, sizeof(unsigned) * t->nentry); + list_for_each_entry(offset, &t->offsets, list) { + i = (offset->offset >> 2) / 32; + m = (offset->offset >> 2) & 31; + m = 1 << m; + t->table[i] ^= m; + } + return 0; +} + +static char gpu_name[10]; +static int parser_auth(struct table *t, const char *filename) +{ + FILE *file; + regex_t mask_rex; + regmatch_t match[4]; + char buf[1024]; + size_t end; + int len; + int done = 0; + int r; + unsigned o; + struct offset *offset; + char last_reg_s[10]; + int last_reg; + + if (regcomp + (&mask_rex, "(0x[0-9a-fA-F]*) *([_a-zA-Z0-9]*)", REG_EXTENDED)) { + fprintf(stderr, "Failed to compile regular expression\n"); + return -1; + } + file = fopen(filename, "r"); + if (file == NULL) { + fprintf(stderr, "Failed to open: %s\n", filename); + return -1; + } + fseek(file, 0, SEEK_END); + end = ftell(file); + fseek(file, 0, SEEK_SET); + + /* get header */ + if (fgets(buf, 1024, file) == NULL) { + fclose(file); + return -1; + } + + /* first line will contain the last register + * and gpu name */ + sscanf(buf, "%s %s", gpu_name, last_reg_s); + t->gpu_prefix = gpu_name; + last_reg = strtol(last_reg_s, NULL, 16); + + do { + if (fgets(buf, 1024, file) == NULL) { + fclose(file); + return -1; + } + len = strlen(buf); + if (ftell(file) == end) + done = 1; + if (len) { + r = regexec(&mask_rex, buf, 4, match, 0); + if (r == REG_NOMATCH) { + } else if (r) { + fprintf(stderr, + "Error matching regular expression %d in %s\n", + r, filename); + fclose(file); + return -1; + } else { + buf[match[0].rm_eo] = 0; + buf[match[1].rm_eo] = 0; + buf[match[2].rm_eo] = 0; + o = strtol(&buf[match[1].rm_so], NULL, 16); + offset = offset_new(o); + table_offset_add(t, offset); + if (o > t->offset_max) + t->offset_max = o; + } + } + } while (!done); + fclose(file); + if (t->offset_max < last_reg) + t->offset_max = last_reg; + return table_build(t); +} + +int main(int argc, char *argv[]) +{ + struct table t; + + if (argc != 2) { + fprintf(stderr, "Usage: %s \n", argv[0]); + exit(1); + } + table_init(&t); + if (parser_auth(&t, argv[1])) { + fprintf(stderr, "Failed to parse file %s\n", argv[1]); + return -1; + } + table_print(&t); + return 0; +} diff --git a/drivers/gpu/drm/radeon/ni.c b/drivers/gpu/drm/radeon/ni.c new file mode 100644 index 00000000..0c460c40 --- /dev/null +++ b/drivers/gpu/drm/radeon/ni.c @@ -0,0 +1,1566 @@ +/* + * Copyright 2010 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Alex Deucher + */ +#include +#include +#include +#include "drmP.h" +#include "radeon.h" +#include "radeon_asic.h" +#include "radeon_drm.h" +#include "nid.h" +#include "atom.h" +#include "ni_reg.h" +#include "cayman_blit_shaders.h" + +extern void evergreen_mc_stop(struct radeon_device *rdev, struct evergreen_mc_save *save); +extern void evergreen_mc_resume(struct radeon_device *rdev, struct evergreen_mc_save *save); +extern int evergreen_mc_wait_for_idle(struct radeon_device *rdev); +extern void evergreen_mc_program(struct radeon_device *rdev); +extern void evergreen_irq_suspend(struct radeon_device *rdev); +extern int evergreen_mc_init(struct radeon_device *rdev); +extern void evergreen_fix_pci_max_read_req_size(struct radeon_device *rdev); + +#define EVERGREEN_PFP_UCODE_SIZE 1120 +#define EVERGREEN_PM4_UCODE_SIZE 1376 +#define EVERGREEN_RLC_UCODE_SIZE 768 +#define BTC_MC_UCODE_SIZE 6024 + +#define CAYMAN_PFP_UCODE_SIZE 2176 +#define CAYMAN_PM4_UCODE_SIZE 2176 +#define CAYMAN_RLC_UCODE_SIZE 1024 +#define CAYMAN_MC_UCODE_SIZE 6037 + +/* Firmware Names */ +MODULE_FIRMWARE("radeon/BARTS_pfp.bin"); +MODULE_FIRMWARE("radeon/BARTS_me.bin"); +MODULE_FIRMWARE("radeon/BARTS_mc.bin"); +MODULE_FIRMWARE("radeon/BTC_rlc.bin"); +MODULE_FIRMWARE("radeon/TURKS_pfp.bin"); +MODULE_FIRMWARE("radeon/TURKS_me.bin"); +MODULE_FIRMWARE("radeon/TURKS_mc.bin"); +MODULE_FIRMWARE("radeon/CAICOS_pfp.bin"); +MODULE_FIRMWARE("radeon/CAICOS_me.bin"); +MODULE_FIRMWARE("radeon/CAICOS_mc.bin"); +MODULE_FIRMWARE("radeon/CAYMAN_pfp.bin"); +MODULE_FIRMWARE("radeon/CAYMAN_me.bin"); +MODULE_FIRMWARE("radeon/CAYMAN_mc.bin"); +MODULE_FIRMWARE("radeon/CAYMAN_rlc.bin"); + +#define BTC_IO_MC_REGS_SIZE 29 + +static const u32 barts_io_mc_regs[BTC_IO_MC_REGS_SIZE][2] = { + {0x00000077, 0xff010100}, + {0x00000078, 0x00000000}, + {0x00000079, 0x00001434}, + {0x0000007a, 0xcc08ec08}, + {0x0000007b, 0x00040000}, + {0x0000007c, 0x000080c0}, + {0x0000007d, 0x09000000}, + {0x0000007e, 0x00210404}, + {0x00000081, 0x08a8e800}, + {0x00000082, 0x00030444}, + {0x00000083, 0x00000000}, + {0x00000085, 0x00000001}, + {0x00000086, 0x00000002}, + {0x00000087, 0x48490000}, + {0x00000088, 0x20244647}, + {0x00000089, 0x00000005}, + {0x0000008b, 0x66030000}, + {0x0000008c, 0x00006603}, + {0x0000008d, 0x00000100}, + {0x0000008f, 0x00001c0a}, + {0x00000090, 0xff000001}, + {0x00000094, 0x00101101}, + {0x00000095, 0x00000fff}, + {0x00000096, 0x00116fff}, + {0x00000097, 0x60010000}, + {0x00000098, 0x10010000}, + {0x00000099, 0x00006000}, + {0x0000009a, 0x00001000}, + {0x0000009f, 0x00946a00} +}; + +static const u32 turks_io_mc_regs[BTC_IO_MC_REGS_SIZE][2] = { + {0x00000077, 0xff010100}, + {0x00000078, 0x00000000}, + {0x00000079, 0x00001434}, + {0x0000007a, 0xcc08ec08}, + {0x0000007b, 0x00040000}, + {0x0000007c, 0x000080c0}, + {0x0000007d, 0x09000000}, + {0x0000007e, 0x00210404}, + {0x00000081, 0x08a8e800}, + {0x00000082, 0x00030444}, + {0x00000083, 0x00000000}, + {0x00000085, 0x00000001}, + {0x00000086, 0x00000002}, + {0x00000087, 0x48490000}, + {0x00000088, 0x20244647}, + {0x00000089, 0x00000005}, + {0x0000008b, 0x66030000}, + {0x0000008c, 0x00006603}, + {0x0000008d, 0x00000100}, + {0x0000008f, 0x00001c0a}, + {0x00000090, 0xff000001}, + {0x00000094, 0x00101101}, + {0x00000095, 0x00000fff}, + {0x00000096, 0x00116fff}, + {0x00000097, 0x60010000}, + {0x00000098, 0x10010000}, + {0x00000099, 0x00006000}, + {0x0000009a, 0x00001000}, + {0x0000009f, 0x00936a00} +}; + +static const u32 caicos_io_mc_regs[BTC_IO_MC_REGS_SIZE][2] = { + {0x00000077, 0xff010100}, + {0x00000078, 0x00000000}, + {0x00000079, 0x00001434}, + {0x0000007a, 0xcc08ec08}, + {0x0000007b, 0x00040000}, + {0x0000007c, 0x000080c0}, + {0x0000007d, 0x09000000}, + {0x0000007e, 0x00210404}, + {0x00000081, 0x08a8e800}, + {0x00000082, 0x00030444}, + {0x00000083, 0x00000000}, + {0x00000085, 0x00000001}, + {0x00000086, 0x00000002}, + {0x00000087, 0x48490000}, + {0x00000088, 0x20244647}, + {0x00000089, 0x00000005}, + {0x0000008b, 0x66030000}, + {0x0000008c, 0x00006603}, + {0x0000008d, 0x00000100}, + {0x0000008f, 0x00001c0a}, + {0x00000090, 0xff000001}, + {0x00000094, 0x00101101}, + {0x00000095, 0x00000fff}, + {0x00000096, 0x00116fff}, + {0x00000097, 0x60010000}, + {0x00000098, 0x10010000}, + {0x00000099, 0x00006000}, + {0x0000009a, 0x00001000}, + {0x0000009f, 0x00916a00} +}; + +static const u32 cayman_io_mc_regs[BTC_IO_MC_REGS_SIZE][2] = { + {0x00000077, 0xff010100}, + {0x00000078, 0x00000000}, + {0x00000079, 0x00001434}, + {0x0000007a, 0xcc08ec08}, + {0x0000007b, 0x00040000}, + {0x0000007c, 0x000080c0}, + {0x0000007d, 0x09000000}, + {0x0000007e, 0x00210404}, + {0x00000081, 0x08a8e800}, + {0x00000082, 0x00030444}, + {0x00000083, 0x00000000}, + {0x00000085, 0x00000001}, + {0x00000086, 0x00000002}, + {0x00000087, 0x48490000}, + {0x00000088, 0x20244647}, + {0x00000089, 0x00000005}, + {0x0000008b, 0x66030000}, + {0x0000008c, 0x00006603}, + {0x0000008d, 0x00000100}, + {0x0000008f, 0x00001c0a}, + {0x00000090, 0xff000001}, + {0x00000094, 0x00101101}, + {0x00000095, 0x00000fff}, + {0x00000096, 0x00116fff}, + {0x00000097, 0x60010000}, + {0x00000098, 0x10010000}, + {0x00000099, 0x00006000}, + {0x0000009a, 0x00001000}, + {0x0000009f, 0x00976b00} +}; + +int ni_mc_load_microcode(struct radeon_device *rdev) +{ + const __be32 *fw_data; + u32 mem_type, running, blackout = 0; + u32 *io_mc_regs; + int i, ucode_size, regs_size; + + if (!rdev->mc_fw) + return -EINVAL; + + switch (rdev->family) { + case CHIP_BARTS: + io_mc_regs = (u32 *)&barts_io_mc_regs; + ucode_size = BTC_MC_UCODE_SIZE; + regs_size = BTC_IO_MC_REGS_SIZE; + break; + case CHIP_TURKS: + io_mc_regs = (u32 *)&turks_io_mc_regs; + ucode_size = BTC_MC_UCODE_SIZE; + regs_size = BTC_IO_MC_REGS_SIZE; + break; + case CHIP_CAICOS: + default: + io_mc_regs = (u32 *)&caicos_io_mc_regs; + ucode_size = BTC_MC_UCODE_SIZE; + regs_size = BTC_IO_MC_REGS_SIZE; + break; + case CHIP_CAYMAN: + io_mc_regs = (u32 *)&cayman_io_mc_regs; + ucode_size = CAYMAN_MC_UCODE_SIZE; + regs_size = BTC_IO_MC_REGS_SIZE; + break; + } + + mem_type = (RREG32(MC_SEQ_MISC0) & MC_SEQ_MISC0_GDDR5_MASK) >> MC_SEQ_MISC0_GDDR5_SHIFT; + running = RREG32(MC_SEQ_SUP_CNTL) & RUN_MASK; + + if ((mem_type == MC_SEQ_MISC0_GDDR5_VALUE) && (running == 0)) { + if (running) { + blackout = RREG32(MC_SHARED_BLACKOUT_CNTL); + WREG32(MC_SHARED_BLACKOUT_CNTL, 1); + } + + /* reset the engine and set to writable */ + WREG32(MC_SEQ_SUP_CNTL, 0x00000008); + WREG32(MC_SEQ_SUP_CNTL, 0x00000010); + + /* load mc io regs */ + for (i = 0; i < regs_size; i++) { + WREG32(MC_SEQ_IO_DEBUG_INDEX, io_mc_regs[(i << 1)]); + WREG32(MC_SEQ_IO_DEBUG_DATA, io_mc_regs[(i << 1) + 1]); + } + /* load the MC ucode */ + fw_data = (const __be32 *)rdev->mc_fw->data; + for (i = 0; i < ucode_size; i++) + WREG32(MC_SEQ_SUP_PGM, be32_to_cpup(fw_data++)); + + /* put the engine back into the active state */ + WREG32(MC_SEQ_SUP_CNTL, 0x00000008); + WREG32(MC_SEQ_SUP_CNTL, 0x00000004); + WREG32(MC_SEQ_SUP_CNTL, 0x00000001); + + /* wait for training to complete */ + while (!(RREG32(MC_IO_PAD_CNTL_D0) & MEM_FALL_OUT_CMD)) + udelay(10); + + if (running) + WREG32(MC_SHARED_BLACKOUT_CNTL, blackout); + } + + return 0; +} + +int ni_init_microcode(struct radeon_device *rdev) +{ + struct platform_device *pdev; + const char *chip_name; + const char *rlc_chip_name; + size_t pfp_req_size, me_req_size, rlc_req_size, mc_req_size; + char fw_name[30]; + int err; + + DRM_DEBUG("\n"); + + pdev = platform_device_register_simple("radeon_cp", 0, NULL, 0); + err = IS_ERR(pdev); + if (err) { + printk(KERN_ERR "radeon_cp: Failed to register firmware\n"); + return -EINVAL; + } + + switch (rdev->family) { + case CHIP_BARTS: + chip_name = "BARTS"; + rlc_chip_name = "BTC"; + pfp_req_size = EVERGREEN_PFP_UCODE_SIZE * 4; + me_req_size = EVERGREEN_PM4_UCODE_SIZE * 4; + rlc_req_size = EVERGREEN_RLC_UCODE_SIZE * 4; + mc_req_size = BTC_MC_UCODE_SIZE * 4; + break; + case CHIP_TURKS: + chip_name = "TURKS"; + rlc_chip_name = "BTC"; + pfp_req_size = EVERGREEN_PFP_UCODE_SIZE * 4; + me_req_size = EVERGREEN_PM4_UCODE_SIZE * 4; + rlc_req_size = EVERGREEN_RLC_UCODE_SIZE * 4; + mc_req_size = BTC_MC_UCODE_SIZE * 4; + break; + case CHIP_CAICOS: + chip_name = "CAICOS"; + rlc_chip_name = "BTC"; + pfp_req_size = EVERGREEN_PFP_UCODE_SIZE * 4; + me_req_size = EVERGREEN_PM4_UCODE_SIZE * 4; + rlc_req_size = EVERGREEN_RLC_UCODE_SIZE * 4; + mc_req_size = BTC_MC_UCODE_SIZE * 4; + break; + case CHIP_CAYMAN: + chip_name = "CAYMAN"; + rlc_chip_name = "CAYMAN"; + pfp_req_size = CAYMAN_PFP_UCODE_SIZE * 4; + me_req_size = CAYMAN_PM4_UCODE_SIZE * 4; + rlc_req_size = CAYMAN_RLC_UCODE_SIZE * 4; + mc_req_size = CAYMAN_MC_UCODE_SIZE * 4; + break; + default: BUG(); + } + + DRM_INFO("Loading %s Microcode\n", chip_name); + + snprintf(fw_name, sizeof(fw_name), "radeon/%s_pfp.bin", chip_name); + err = request_firmware(&rdev->pfp_fw, fw_name, &pdev->dev); + if (err) + goto out; + if (rdev->pfp_fw->size != pfp_req_size) { + printk(KERN_ERR + "ni_cp: Bogus length %zu in firmware \"%s\"\n", + rdev->pfp_fw->size, fw_name); + err = -EINVAL; + goto out; + } + + snprintf(fw_name, sizeof(fw_name), "radeon/%s_me.bin", chip_name); + err = request_firmware(&rdev->me_fw, fw_name, &pdev->dev); + if (err) + goto out; + if (rdev->me_fw->size != me_req_size) { + printk(KERN_ERR + "ni_cp: Bogus length %zu in firmware \"%s\"\n", + rdev->me_fw->size, fw_name); + err = -EINVAL; + } + + snprintf(fw_name, sizeof(fw_name), "radeon/%s_rlc.bin", rlc_chip_name); + err = request_firmware(&rdev->rlc_fw, fw_name, &pdev->dev); + if (err) + goto out; + if (rdev->rlc_fw->size != rlc_req_size) { + printk(KERN_ERR + "ni_rlc: Bogus length %zu in firmware \"%s\"\n", + rdev->rlc_fw->size, fw_name); + err = -EINVAL; + } + + snprintf(fw_name, sizeof(fw_name), "radeon/%s_mc.bin", chip_name); + err = request_firmware(&rdev->mc_fw, fw_name, &pdev->dev); + if (err) + goto out; + if (rdev->mc_fw->size != mc_req_size) { + printk(KERN_ERR + "ni_mc: Bogus length %zu in firmware \"%s\"\n", + rdev->mc_fw->size, fw_name); + err = -EINVAL; + } +out: + platform_device_unregister(pdev); + + if (err) { + if (err != -EINVAL) + printk(KERN_ERR + "ni_cp: Failed to load firmware \"%s\"\n", + fw_name); + release_firmware(rdev->pfp_fw); + rdev->pfp_fw = NULL; + release_firmware(rdev->me_fw); + rdev->me_fw = NULL; + release_firmware(rdev->rlc_fw); + rdev->rlc_fw = NULL; + release_firmware(rdev->mc_fw); + rdev->mc_fw = NULL; + } + return err; +} + +/* + * Core functions + */ +static u32 cayman_get_tile_pipe_to_backend_map(struct radeon_device *rdev, + u32 num_tile_pipes, + u32 num_backends_per_asic, + u32 *backend_disable_mask_per_asic, + u32 num_shader_engines) +{ + u32 backend_map = 0; + u32 enabled_backends_mask = 0; + u32 enabled_backends_count = 0; + u32 num_backends_per_se; + u32 cur_pipe; + u32 swizzle_pipe[CAYMAN_MAX_PIPES]; + u32 cur_backend = 0; + u32 i; + bool force_no_swizzle; + + /* force legal values */ + if (num_tile_pipes < 1) + num_tile_pipes = 1; + if (num_tile_pipes > rdev->config.cayman.max_tile_pipes) + num_tile_pipes = rdev->config.cayman.max_tile_pipes; + if (num_shader_engines < 1) + num_shader_engines = 1; + if (num_shader_engines > rdev->config.cayman.max_shader_engines) + num_shader_engines = rdev->config.cayman.max_shader_engines; + if (num_backends_per_asic < num_shader_engines) + num_backends_per_asic = num_shader_engines; + if (num_backends_per_asic > (rdev->config.cayman.max_backends_per_se * num_shader_engines)) + num_backends_per_asic = rdev->config.cayman.max_backends_per_se * num_shader_engines; + + /* make sure we have the same number of backends per se */ + num_backends_per_asic = ALIGN(num_backends_per_asic, num_shader_engines); + /* set up the number of backends per se */ + num_backends_per_se = num_backends_per_asic / num_shader_engines; + if (num_backends_per_se > rdev->config.cayman.max_backends_per_se) { + num_backends_per_se = rdev->config.cayman.max_backends_per_se; + num_backends_per_asic = num_backends_per_se * num_shader_engines; + } + + /* create enable mask and count for enabled backends */ + for (i = 0; i < CAYMAN_MAX_BACKENDS; ++i) { + if (((*backend_disable_mask_per_asic >> i) & 1) == 0) { + enabled_backends_mask |= (1 << i); + ++enabled_backends_count; + } + if (enabled_backends_count == num_backends_per_asic) + break; + } + + /* force the backends mask to match the current number of backends */ + if (enabled_backends_count != num_backends_per_asic) { + u32 this_backend_enabled; + u32 shader_engine; + u32 backend_per_se; + + enabled_backends_mask = 0; + enabled_backends_count = 0; + *backend_disable_mask_per_asic = CAYMAN_MAX_BACKENDS_MASK; + for (i = 0; i < CAYMAN_MAX_BACKENDS; ++i) { + /* calc the current se */ + shader_engine = i / rdev->config.cayman.max_backends_per_se; + /* calc the backend per se */ + backend_per_se = i % rdev->config.cayman.max_backends_per_se; + /* default to not enabled */ + this_backend_enabled = 0; + if ((shader_engine < num_shader_engines) && + (backend_per_se < num_backends_per_se)) + this_backend_enabled = 1; + if (this_backend_enabled) { + enabled_backends_mask |= (1 << i); + *backend_disable_mask_per_asic &= ~(1 << i); + ++enabled_backends_count; + } + } + } + + + memset((uint8_t *)&swizzle_pipe[0], 0, sizeof(u32) * CAYMAN_MAX_PIPES); + switch (rdev->family) { + case CHIP_CAYMAN: + force_no_swizzle = true; + break; + default: + force_no_swizzle = false; + break; + } + if (force_no_swizzle) { + bool last_backend_enabled = false; + + force_no_swizzle = false; + for (i = 0; i < CAYMAN_MAX_BACKENDS; ++i) { + if (((enabled_backends_mask >> i) & 1) == 1) { + if (last_backend_enabled) + force_no_swizzle = true; + last_backend_enabled = true; + } else + last_backend_enabled = false; + } + } + + switch (num_tile_pipes) { + case 1: + case 3: + case 5: + case 7: + DRM_ERROR("odd number of pipes!\n"); + break; + case 2: + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 1; + break; + case 4: + if (force_no_swizzle) { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 1; + swizzle_pipe[2] = 2; + swizzle_pipe[3] = 3; + } else { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 2; + swizzle_pipe[2] = 1; + swizzle_pipe[3] = 3; + } + break; + case 6: + if (force_no_swizzle) { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 1; + swizzle_pipe[2] = 2; + swizzle_pipe[3] = 3; + swizzle_pipe[4] = 4; + swizzle_pipe[5] = 5; + } else { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 2; + swizzle_pipe[2] = 4; + swizzle_pipe[3] = 1; + swizzle_pipe[4] = 3; + swizzle_pipe[5] = 5; + } + break; + case 8: + if (force_no_swizzle) { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 1; + swizzle_pipe[2] = 2; + swizzle_pipe[3] = 3; + swizzle_pipe[4] = 4; + swizzle_pipe[5] = 5; + swizzle_pipe[6] = 6; + swizzle_pipe[7] = 7; + } else { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 2; + swizzle_pipe[2] = 4; + swizzle_pipe[3] = 6; + swizzle_pipe[4] = 1; + swizzle_pipe[5] = 3; + swizzle_pipe[6] = 5; + swizzle_pipe[7] = 7; + } + break; + } + + for (cur_pipe = 0; cur_pipe < num_tile_pipes; ++cur_pipe) { + while (((1 << cur_backend) & enabled_backends_mask) == 0) + cur_backend = (cur_backend + 1) % CAYMAN_MAX_BACKENDS; + + backend_map |= (((cur_backend & 0xf) << (swizzle_pipe[cur_pipe] * 4))); + + cur_backend = (cur_backend + 1) % CAYMAN_MAX_BACKENDS; + } + + return backend_map; +} + +static u32 cayman_get_disable_mask_per_asic(struct radeon_device *rdev, + u32 disable_mask_per_se, + u32 max_disable_mask_per_se, + u32 num_shader_engines) +{ + u32 disable_field_width_per_se = r600_count_pipe_bits(disable_mask_per_se); + u32 disable_mask_per_asic = disable_mask_per_se & max_disable_mask_per_se; + + if (num_shader_engines == 1) + return disable_mask_per_asic; + else if (num_shader_engines == 2) + return disable_mask_per_asic | (disable_mask_per_asic << disable_field_width_per_se); + else + return 0xffffffff; +} + +static void cayman_gpu_init(struct radeon_device *rdev) +{ + u32 cc_rb_backend_disable = 0; + u32 cc_gc_shader_pipe_config; + u32 gb_addr_config = 0; + u32 mc_shared_chmap, mc_arb_ramcfg; + u32 gb_backend_map; + u32 cgts_tcc_disable; + u32 sx_debug_1; + u32 smx_dc_ctl0; + u32 gc_user_shader_pipe_config; + u32 gc_user_rb_backend_disable; + u32 cgts_user_tcc_disable; + u32 cgts_sm_ctrl_reg; + u32 hdp_host_path_cntl; + u32 tmp; + int i, j; + + switch (rdev->family) { + case CHIP_CAYMAN: + default: + rdev->config.cayman.max_shader_engines = 2; + rdev->config.cayman.max_pipes_per_simd = 4; + rdev->config.cayman.max_tile_pipes = 8; + rdev->config.cayman.max_simds_per_se = 12; + rdev->config.cayman.max_backends_per_se = 4; + rdev->config.cayman.max_texture_channel_caches = 8; + rdev->config.cayman.max_gprs = 256; + rdev->config.cayman.max_threads = 256; + rdev->config.cayman.max_gs_threads = 32; + rdev->config.cayman.max_stack_entries = 512; + rdev->config.cayman.sx_num_of_sets = 8; + rdev->config.cayman.sx_max_export_size = 256; + rdev->config.cayman.sx_max_export_pos_size = 64; + rdev->config.cayman.sx_max_export_smx_size = 192; + rdev->config.cayman.max_hw_contexts = 8; + rdev->config.cayman.sq_num_cf_insts = 2; + + rdev->config.cayman.sc_prim_fifo_size = 0x100; + rdev->config.cayman.sc_hiz_tile_fifo_size = 0x30; + rdev->config.cayman.sc_earlyz_tile_fifo_size = 0x130; + break; + } + + /* Initialize HDP */ + for (i = 0, j = 0; i < 32; i++, j += 0x18) { + WREG32((0x2c14 + j), 0x00000000); + WREG32((0x2c18 + j), 0x00000000); + WREG32((0x2c1c + j), 0x00000000); + WREG32((0x2c20 + j), 0x00000000); + WREG32((0x2c24 + j), 0x00000000); + } + + WREG32(GRBM_CNTL, GRBM_READ_TIMEOUT(0xff)); + + evergreen_fix_pci_max_read_req_size(rdev); + + mc_shared_chmap = RREG32(MC_SHARED_CHMAP); + mc_arb_ramcfg = RREG32(MC_ARB_RAMCFG); + + cc_rb_backend_disable = RREG32(CC_RB_BACKEND_DISABLE); + cc_gc_shader_pipe_config = RREG32(CC_GC_SHADER_PIPE_CONFIG); + cgts_tcc_disable = 0xff000000; + gc_user_rb_backend_disable = RREG32(GC_USER_RB_BACKEND_DISABLE); + gc_user_shader_pipe_config = RREG32(GC_USER_SHADER_PIPE_CONFIG); + cgts_user_tcc_disable = RREG32(CGTS_USER_TCC_DISABLE); + + rdev->config.cayman.num_shader_engines = rdev->config.cayman.max_shader_engines; + tmp = ((~gc_user_shader_pipe_config) & INACTIVE_QD_PIPES_MASK) >> INACTIVE_QD_PIPES_SHIFT; + rdev->config.cayman.num_shader_pipes_per_simd = r600_count_pipe_bits(tmp); + rdev->config.cayman.num_tile_pipes = rdev->config.cayman.max_tile_pipes; + tmp = ((~gc_user_shader_pipe_config) & INACTIVE_SIMDS_MASK) >> INACTIVE_SIMDS_SHIFT; + rdev->config.cayman.num_simds_per_se = r600_count_pipe_bits(tmp); + tmp = ((~gc_user_rb_backend_disable) & BACKEND_DISABLE_MASK) >> BACKEND_DISABLE_SHIFT; + rdev->config.cayman.num_backends_per_se = r600_count_pipe_bits(tmp); + tmp = (gc_user_rb_backend_disable & BACKEND_DISABLE_MASK) >> BACKEND_DISABLE_SHIFT; + rdev->config.cayman.backend_disable_mask_per_asic = + cayman_get_disable_mask_per_asic(rdev, tmp, CAYMAN_MAX_BACKENDS_PER_SE_MASK, + rdev->config.cayman.num_shader_engines); + rdev->config.cayman.backend_map = + cayman_get_tile_pipe_to_backend_map(rdev, rdev->config.cayman.num_tile_pipes, + rdev->config.cayman.num_backends_per_se * + rdev->config.cayman.num_shader_engines, + &rdev->config.cayman.backend_disable_mask_per_asic, + rdev->config.cayman.num_shader_engines); + tmp = ((~cgts_user_tcc_disable) & TCC_DISABLE_MASK) >> TCC_DISABLE_SHIFT; + rdev->config.cayman.num_texture_channel_caches = r600_count_pipe_bits(tmp); + tmp = (mc_arb_ramcfg & BURSTLENGTH_MASK) >> BURSTLENGTH_SHIFT; + rdev->config.cayman.mem_max_burst_length_bytes = (tmp + 1) * 256; + if (rdev->config.cayman.mem_max_burst_length_bytes > 512) + rdev->config.cayman.mem_max_burst_length_bytes = 512; + tmp = (mc_arb_ramcfg & NOOFCOLS_MASK) >> NOOFCOLS_SHIFT; + rdev->config.cayman.mem_row_size_in_kb = (4 * (1 << (8 + tmp))) / 1024; + if (rdev->config.cayman.mem_row_size_in_kb > 4) + rdev->config.cayman.mem_row_size_in_kb = 4; + /* XXX use MC settings? */ + rdev->config.cayman.shader_engine_tile_size = 32; + rdev->config.cayman.num_gpus = 1; + rdev->config.cayman.multi_gpu_tile_size = 64; + + //gb_addr_config = 0x02011003 +#if 0 + gb_addr_config = RREG32(GB_ADDR_CONFIG); +#else + gb_addr_config = 0; + switch (rdev->config.cayman.num_tile_pipes) { + case 1: + default: + gb_addr_config |= NUM_PIPES(0); + break; + case 2: + gb_addr_config |= NUM_PIPES(1); + break; + case 4: + gb_addr_config |= NUM_PIPES(2); + break; + case 8: + gb_addr_config |= NUM_PIPES(3); + break; + } + + tmp = (rdev->config.cayman.mem_max_burst_length_bytes / 256) - 1; + gb_addr_config |= PIPE_INTERLEAVE_SIZE(tmp); + gb_addr_config |= NUM_SHADER_ENGINES(rdev->config.cayman.num_shader_engines - 1); + tmp = (rdev->config.cayman.shader_engine_tile_size / 16) - 1; + gb_addr_config |= SHADER_ENGINE_TILE_SIZE(tmp); + switch (rdev->config.cayman.num_gpus) { + case 1: + default: + gb_addr_config |= NUM_GPUS(0); + break; + case 2: + gb_addr_config |= NUM_GPUS(1); + break; + case 4: + gb_addr_config |= NUM_GPUS(2); + break; + } + switch (rdev->config.cayman.multi_gpu_tile_size) { + case 16: + gb_addr_config |= MULTI_GPU_TILE_SIZE(0); + break; + case 32: + default: + gb_addr_config |= MULTI_GPU_TILE_SIZE(1); + break; + case 64: + gb_addr_config |= MULTI_GPU_TILE_SIZE(2); + break; + case 128: + gb_addr_config |= MULTI_GPU_TILE_SIZE(3); + break; + } + switch (rdev->config.cayman.mem_row_size_in_kb) { + case 1: + default: + gb_addr_config |= ROW_SIZE(0); + break; + case 2: + gb_addr_config |= ROW_SIZE(1); + break; + case 4: + gb_addr_config |= ROW_SIZE(2); + break; + } +#endif + + tmp = (gb_addr_config & NUM_PIPES_MASK) >> NUM_PIPES_SHIFT; + rdev->config.cayman.num_tile_pipes = (1 << tmp); + tmp = (gb_addr_config & PIPE_INTERLEAVE_SIZE_MASK) >> PIPE_INTERLEAVE_SIZE_SHIFT; + rdev->config.cayman.mem_max_burst_length_bytes = (tmp + 1) * 256; + tmp = (gb_addr_config & NUM_SHADER_ENGINES_MASK) >> NUM_SHADER_ENGINES_SHIFT; + rdev->config.cayman.num_shader_engines = tmp + 1; + tmp = (gb_addr_config & NUM_GPUS_MASK) >> NUM_GPUS_SHIFT; + rdev->config.cayman.num_gpus = tmp + 1; + tmp = (gb_addr_config & MULTI_GPU_TILE_SIZE_MASK) >> MULTI_GPU_TILE_SIZE_SHIFT; + rdev->config.cayman.multi_gpu_tile_size = 1 << tmp; + tmp = (gb_addr_config & ROW_SIZE_MASK) >> ROW_SIZE_SHIFT; + rdev->config.cayman.mem_row_size_in_kb = 1 << tmp; + + //gb_backend_map = 0x76541032; +#if 0 + gb_backend_map = RREG32(GB_BACKEND_MAP); +#else + gb_backend_map = + cayman_get_tile_pipe_to_backend_map(rdev, rdev->config.cayman.num_tile_pipes, + rdev->config.cayman.num_backends_per_se * + rdev->config.cayman.num_shader_engines, + &rdev->config.cayman.backend_disable_mask_per_asic, + rdev->config.cayman.num_shader_engines); +#endif + /* setup tiling info dword. gb_addr_config is not adequate since it does + * not have bank info, so create a custom tiling dword. + * bits 3:0 num_pipes + * bits 7:4 num_banks + * bits 11:8 group_size + * bits 15:12 row_size + */ + rdev->config.cayman.tile_config = 0; + switch (rdev->config.cayman.num_tile_pipes) { + case 1: + default: + rdev->config.cayman.tile_config |= (0 << 0); + break; + case 2: + rdev->config.cayman.tile_config |= (1 << 0); + break; + case 4: + rdev->config.cayman.tile_config |= (2 << 0); + break; + case 8: + rdev->config.cayman.tile_config |= (3 << 0); + break; + } + rdev->config.cayman.tile_config |= + ((mc_arb_ramcfg & NOOFBANK_MASK) >> NOOFBANK_SHIFT) << 4; + rdev->config.cayman.tile_config |= + ((gb_addr_config & PIPE_INTERLEAVE_SIZE_MASK) >> PIPE_INTERLEAVE_SIZE_SHIFT) << 8; + rdev->config.cayman.tile_config |= + ((gb_addr_config & ROW_SIZE_MASK) >> ROW_SIZE_SHIFT) << 12; + + WREG32(GB_BACKEND_MAP, gb_backend_map); + WREG32(GB_ADDR_CONFIG, gb_addr_config); + WREG32(DMIF_ADDR_CONFIG, gb_addr_config); + WREG32(HDP_ADDR_CONFIG, gb_addr_config); + + /* primary versions */ + WREG32(CC_RB_BACKEND_DISABLE, cc_rb_backend_disable); + WREG32(CC_SYS_RB_BACKEND_DISABLE, cc_rb_backend_disable); + WREG32(CC_GC_SHADER_PIPE_CONFIG, cc_gc_shader_pipe_config); + + WREG32(CGTS_TCC_DISABLE, cgts_tcc_disable); + WREG32(CGTS_SYS_TCC_DISABLE, cgts_tcc_disable); + + /* user versions */ + WREG32(GC_USER_RB_BACKEND_DISABLE, cc_rb_backend_disable); + WREG32(GC_USER_SYS_RB_BACKEND_DISABLE, cc_rb_backend_disable); + WREG32(GC_USER_SHADER_PIPE_CONFIG, cc_gc_shader_pipe_config); + + WREG32(CGTS_USER_SYS_TCC_DISABLE, cgts_tcc_disable); + WREG32(CGTS_USER_TCC_DISABLE, cgts_tcc_disable); + + /* reprogram the shader complex */ + cgts_sm_ctrl_reg = RREG32(CGTS_SM_CTRL_REG); + for (i = 0; i < 16; i++) + WREG32(CGTS_SM_CTRL_REG, OVERRIDE); + WREG32(CGTS_SM_CTRL_REG, cgts_sm_ctrl_reg); + + /* set HW defaults for 3D engine */ + WREG32(CP_MEQ_THRESHOLDS, MEQ1_START(0x30) | MEQ2_START(0x60)); + + sx_debug_1 = RREG32(SX_DEBUG_1); + sx_debug_1 |= ENABLE_NEW_SMX_ADDRESS; + WREG32(SX_DEBUG_1, sx_debug_1); + + smx_dc_ctl0 = RREG32(SMX_DC_CTL0); + smx_dc_ctl0 &= ~NUMBER_OF_SETS(0x1ff); + smx_dc_ctl0 |= NUMBER_OF_SETS(rdev->config.cayman.sx_num_of_sets); + WREG32(SMX_DC_CTL0, smx_dc_ctl0); + + WREG32(SPI_CONFIG_CNTL_1, VTX_DONE_DELAY(4) | CRC_SIMD_ID_WADDR_DISABLE); + + /* need to be explicitly zero-ed */ + WREG32(VGT_OFFCHIP_LDS_BASE, 0); + WREG32(SQ_LSTMP_RING_BASE, 0); + WREG32(SQ_HSTMP_RING_BASE, 0); + WREG32(SQ_ESTMP_RING_BASE, 0); + WREG32(SQ_GSTMP_RING_BASE, 0); + WREG32(SQ_VSTMP_RING_BASE, 0); + WREG32(SQ_PSTMP_RING_BASE, 0); + + WREG32(TA_CNTL_AUX, DISABLE_CUBE_ANISO); + + WREG32(SX_EXPORT_BUFFER_SIZES, (COLOR_BUFFER_SIZE((rdev->config.cayman.sx_max_export_size / 4) - 1) | + POSITION_BUFFER_SIZE((rdev->config.cayman.sx_max_export_pos_size / 4) - 1) | + SMX_BUFFER_SIZE((rdev->config.cayman.sx_max_export_smx_size / 4) - 1))); + + WREG32(PA_SC_FIFO_SIZE, (SC_PRIM_FIFO_SIZE(rdev->config.cayman.sc_prim_fifo_size) | + SC_HIZ_TILE_FIFO_SIZE(rdev->config.cayman.sc_hiz_tile_fifo_size) | + SC_EARLYZ_TILE_FIFO_SIZE(rdev->config.cayman.sc_earlyz_tile_fifo_size))); + + + WREG32(VGT_NUM_INSTANCES, 1); + + WREG32(CP_PERFMON_CNTL, 0); + + WREG32(SQ_MS_FIFO_SIZES, (CACHE_FIFO_SIZE(16 * rdev->config.cayman.sq_num_cf_insts) | + FETCH_FIFO_HIWATER(0x4) | + DONE_FIFO_HIWATER(0xe0) | + ALU_UPDATE_FIFO_HIWATER(0x8))); + + WREG32(SQ_GPR_RESOURCE_MGMT_1, NUM_CLAUSE_TEMP_GPRS(4)); + WREG32(SQ_CONFIG, (VC_ENABLE | + EXPORT_SRC_C | + GFX_PRIO(0) | + CS1_PRIO(0) | + CS2_PRIO(1))); + WREG32(SQ_DYN_GPR_CNTL_PS_FLUSH_REQ, DYN_GPR_ENABLE); + + WREG32(PA_SC_FORCE_EOV_MAX_CNTS, (FORCE_EOV_MAX_CLK_CNT(4095) | + FORCE_EOV_MAX_REZ_CNT(255))); + + WREG32(VGT_CACHE_INVALIDATION, CACHE_INVALIDATION(VC_AND_TC) | + AUTO_INVLD_EN(ES_AND_GS_AUTO)); + + WREG32(VGT_GS_VERTEX_REUSE, 16); + WREG32(PA_SC_LINE_STIPPLE_STATE, 0); + + WREG32(CB_PERF_CTR0_SEL_0, 0); + WREG32(CB_PERF_CTR0_SEL_1, 0); + WREG32(CB_PERF_CTR1_SEL_0, 0); + WREG32(CB_PERF_CTR1_SEL_1, 0); + WREG32(CB_PERF_CTR2_SEL_0, 0); + WREG32(CB_PERF_CTR2_SEL_1, 0); + WREG32(CB_PERF_CTR3_SEL_0, 0); + WREG32(CB_PERF_CTR3_SEL_1, 0); + + tmp = RREG32(HDP_MISC_CNTL); + tmp |= HDP_FLUSH_INVALIDATE_CACHE; + WREG32(HDP_MISC_CNTL, tmp); + + hdp_host_path_cntl = RREG32(HDP_HOST_PATH_CNTL); + WREG32(HDP_HOST_PATH_CNTL, hdp_host_path_cntl); + + WREG32(PA_CL_ENHANCE, CLIP_VTX_REORDER_ENA | NUM_CLIP_SEQ(3)); + + udelay(50); +} + +/* + * GART + */ +void cayman_pcie_gart_tlb_flush(struct radeon_device *rdev) +{ + /* flush hdp cache */ + WREG32(HDP_MEM_COHERENCY_FLUSH_CNTL, 0x1); + + /* bits 0-7 are the VM contexts0-7 */ + WREG32(VM_INVALIDATE_REQUEST, 1); +} + +int cayman_pcie_gart_enable(struct radeon_device *rdev) +{ + int r; + + if (rdev->gart.table.vram.robj == NULL) { + dev_err(rdev->dev, "No VRAM object for PCIE GART.\n"); + return -EINVAL; + } + r = radeon_gart_table_vram_pin(rdev); + if (r) + return r; + radeon_gart_restore(rdev); + /* Setup TLB control */ + WREG32(MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB | + ENABLE_L1_FRAGMENT_PROCESSING | + SYSTEM_ACCESS_MODE_NOT_IN_SYS | + SYSTEM_APERTURE_UNMAPPED_ACCESS_PASS_THRU); + /* Setup L2 cache */ + WREG32(VM_L2_CNTL, ENABLE_L2_CACHE | + ENABLE_L2_PTE_CACHE_LRU_UPDATE_BY_WRITE | + ENABLE_L2_PDE0_CACHE_LRU_UPDATE_BY_WRITE | + EFFECTIVE_L2_QUEUE_SIZE(7) | + CONTEXT1_IDENTITY_ACCESS_MODE(1)); + WREG32(VM_L2_CNTL2, INVALIDATE_ALL_L1_TLBS | INVALIDATE_L2_CACHE); + WREG32(VM_L2_CNTL3, L2_CACHE_BIGK_ASSOCIATIVITY | + L2_CACHE_BIGK_FRAGMENT_SIZE(6)); + /* setup context0 */ + WREG32(VM_CONTEXT0_PAGE_TABLE_START_ADDR, rdev->mc.gtt_start >> 12); + WREG32(VM_CONTEXT0_PAGE_TABLE_END_ADDR, rdev->mc.gtt_end >> 12); + WREG32(VM_CONTEXT0_PAGE_TABLE_BASE_ADDR, rdev->gart.table_addr >> 12); + WREG32(VM_CONTEXT0_PROTECTION_FAULT_DEFAULT_ADDR, + (u32)(rdev->dummy_page.addr >> 12)); + WREG32(VM_CONTEXT0_CNTL2, 0); + WREG32(VM_CONTEXT0_CNTL, ENABLE_CONTEXT | PAGE_TABLE_DEPTH(0) | + RANGE_PROTECTION_FAULT_ENABLE_DEFAULT); + /* disable context1-7 */ + WREG32(VM_CONTEXT1_CNTL2, 0); + WREG32(VM_CONTEXT1_CNTL, 0); + + cayman_pcie_gart_tlb_flush(rdev); + rdev->gart.ready = true; + return 0; +} + +void cayman_pcie_gart_disable(struct radeon_device *rdev) +{ + int r; + + /* Disable all tables */ + WREG32(VM_CONTEXT0_CNTL, 0); + WREG32(VM_CONTEXT1_CNTL, 0); + /* Setup TLB control */ + WREG32(MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_FRAGMENT_PROCESSING | + SYSTEM_ACCESS_MODE_NOT_IN_SYS | + SYSTEM_APERTURE_UNMAPPED_ACCESS_PASS_THRU); + /* Setup L2 cache */ + WREG32(VM_L2_CNTL, ENABLE_L2_PTE_CACHE_LRU_UPDATE_BY_WRITE | + ENABLE_L2_PDE0_CACHE_LRU_UPDATE_BY_WRITE | + EFFECTIVE_L2_QUEUE_SIZE(7) | + CONTEXT1_IDENTITY_ACCESS_MODE(1)); + WREG32(VM_L2_CNTL2, 0); + WREG32(VM_L2_CNTL3, L2_CACHE_BIGK_ASSOCIATIVITY | + L2_CACHE_BIGK_FRAGMENT_SIZE(6)); + if (rdev->gart.table.vram.robj) { + r = radeon_bo_reserve(rdev->gart.table.vram.robj, false); + if (likely(r == 0)) { + radeon_bo_kunmap(rdev->gart.table.vram.robj); + radeon_bo_unpin(rdev->gart.table.vram.robj); + radeon_bo_unreserve(rdev->gart.table.vram.robj); + } + } +} + +void cayman_pcie_gart_fini(struct radeon_device *rdev) +{ + cayman_pcie_gart_disable(rdev); + radeon_gart_table_vram_free(rdev); + radeon_gart_fini(rdev); +} + +/* + * CP. + */ +static void cayman_cp_enable(struct radeon_device *rdev, bool enable) +{ + if (enable) + WREG32(CP_ME_CNTL, 0); + else { + radeon_ttm_set_active_vram_size(rdev, rdev->mc.visible_vram_size); + WREG32(CP_ME_CNTL, (CP_ME_HALT | CP_PFP_HALT)); + WREG32(SCRATCH_UMSK, 0); + } +} + +static int cayman_cp_load_microcode(struct radeon_device *rdev) +{ + const __be32 *fw_data; + int i; + + if (!rdev->me_fw || !rdev->pfp_fw) + return -EINVAL; + + cayman_cp_enable(rdev, false); + + fw_data = (const __be32 *)rdev->pfp_fw->data; + WREG32(CP_PFP_UCODE_ADDR, 0); + for (i = 0; i < CAYMAN_PFP_UCODE_SIZE; i++) + WREG32(CP_PFP_UCODE_DATA, be32_to_cpup(fw_data++)); + WREG32(CP_PFP_UCODE_ADDR, 0); + + fw_data = (const __be32 *)rdev->me_fw->data; + WREG32(CP_ME_RAM_WADDR, 0); + for (i = 0; i < CAYMAN_PM4_UCODE_SIZE; i++) + WREG32(CP_ME_RAM_DATA, be32_to_cpup(fw_data++)); + + WREG32(CP_PFP_UCODE_ADDR, 0); + WREG32(CP_ME_RAM_WADDR, 0); + WREG32(CP_ME_RAM_RADDR, 0); + return 0; +} + +static int cayman_cp_start(struct radeon_device *rdev) +{ + int r, i; + + r = radeon_ring_lock(rdev, 7); + if (r) { + DRM_ERROR("radeon: cp failed to lock ring (%d).\n", r); + return r; + } + radeon_ring_write(rdev, PACKET3(PACKET3_ME_INITIALIZE, 5)); + radeon_ring_write(rdev, 0x1); + radeon_ring_write(rdev, 0x0); + radeon_ring_write(rdev, rdev->config.cayman.max_hw_contexts - 1); + radeon_ring_write(rdev, PACKET3_ME_INITIALIZE_DEVICE_ID(1)); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, 0); + radeon_ring_unlock_commit(rdev); + + cayman_cp_enable(rdev, true); + + r = radeon_ring_lock(rdev, cayman_default_size + 19); + if (r) { + DRM_ERROR("radeon: cp failed to lock ring (%d).\n", r); + return r; + } + + /* setup clear context state */ + radeon_ring_write(rdev, PACKET3(PACKET3_PREAMBLE_CNTL, 0)); + radeon_ring_write(rdev, PACKET3_PREAMBLE_BEGIN_CLEAR_STATE); + + for (i = 0; i < cayman_default_size; i++) + radeon_ring_write(rdev, cayman_default_state[i]); + + radeon_ring_write(rdev, PACKET3(PACKET3_PREAMBLE_CNTL, 0)); + radeon_ring_write(rdev, PACKET3_PREAMBLE_END_CLEAR_STATE); + + /* set clear context state */ + radeon_ring_write(rdev, PACKET3(PACKET3_CLEAR_STATE, 0)); + radeon_ring_write(rdev, 0); + + /* SQ_VTX_BASE_VTX_LOC */ + radeon_ring_write(rdev, 0xc0026f00); + radeon_ring_write(rdev, 0x00000000); + radeon_ring_write(rdev, 0x00000000); + radeon_ring_write(rdev, 0x00000000); + + /* Clear consts */ + radeon_ring_write(rdev, 0xc0036f00); + radeon_ring_write(rdev, 0x00000bc4); + radeon_ring_write(rdev, 0xffffffff); + radeon_ring_write(rdev, 0xffffffff); + radeon_ring_write(rdev, 0xffffffff); + + radeon_ring_write(rdev, 0xc0026900); + radeon_ring_write(rdev, 0x00000316); + radeon_ring_write(rdev, 0x0000000e); /* VGT_VERTEX_REUSE_BLOCK_CNTL */ + radeon_ring_write(rdev, 0x00000010); /* */ + + radeon_ring_unlock_commit(rdev); + + /* XXX init other rings */ + + return 0; +} + +static void cayman_cp_fini(struct radeon_device *rdev) +{ + cayman_cp_enable(rdev, false); + radeon_ring_fini(rdev); +} + +int cayman_cp_resume(struct radeon_device *rdev) +{ + u32 tmp; + u32 rb_bufsz; + int r; + + /* Reset cp; if cp is reset, then PA, SH, VGT also need to be reset */ + WREG32(GRBM_SOFT_RESET, (SOFT_RESET_CP | + SOFT_RESET_PA | + SOFT_RESET_SH | + SOFT_RESET_VGT | + SOFT_RESET_SPI | + SOFT_RESET_SX)); + RREG32(GRBM_SOFT_RESET); + mdelay(15); + WREG32(GRBM_SOFT_RESET, 0); + RREG32(GRBM_SOFT_RESET); + + WREG32(CP_SEM_WAIT_TIMER, 0x4); + + /* Set the write pointer delay */ + WREG32(CP_RB_WPTR_DELAY, 0); + + WREG32(CP_DEBUG, (1 << 27)); + + /* ring 0 - compute and gfx */ + /* Set ring buffer size */ + rb_bufsz = drm_order(rdev->cp.ring_size / 8); + tmp = (drm_order(RADEON_GPU_PAGE_SIZE/8) << 8) | rb_bufsz; +#ifdef __BIG_ENDIAN + tmp |= BUF_SWAP_32BIT; +#endif + WREG32(CP_RB0_CNTL, tmp); + + /* Initialize the ring buffer's read and write pointers */ + WREG32(CP_RB0_CNTL, tmp | RB_RPTR_WR_ENA); + rdev->cp.wptr = 0; + WREG32(CP_RB0_WPTR, rdev->cp.wptr); + + /* set the wb address wether it's enabled or not */ + WREG32(CP_RB0_RPTR_ADDR, (rdev->wb.gpu_addr + RADEON_WB_CP_RPTR_OFFSET) & 0xFFFFFFFC); + WREG32(CP_RB0_RPTR_ADDR_HI, upper_32_bits(rdev->wb.gpu_addr + RADEON_WB_CP_RPTR_OFFSET) & 0xFF); + WREG32(SCRATCH_ADDR, ((rdev->wb.gpu_addr + RADEON_WB_SCRATCH_OFFSET) >> 8) & 0xFFFFFFFF); + + if (rdev->wb.enabled) + WREG32(SCRATCH_UMSK, 0xff); + else { + tmp |= RB_NO_UPDATE; + WREG32(SCRATCH_UMSK, 0); + } + + mdelay(1); + WREG32(CP_RB0_CNTL, tmp); + + WREG32(CP_RB0_BASE, rdev->cp.gpu_addr >> 8); + + rdev->cp.rptr = RREG32(CP_RB0_RPTR); + + /* ring1 - compute only */ + /* Set ring buffer size */ + rb_bufsz = drm_order(rdev->cp1.ring_size / 8); + tmp = (drm_order(RADEON_GPU_PAGE_SIZE/8) << 8) | rb_bufsz; +#ifdef __BIG_ENDIAN + tmp |= BUF_SWAP_32BIT; +#endif + WREG32(CP_RB1_CNTL, tmp); + + /* Initialize the ring buffer's read and write pointers */ + WREG32(CP_RB1_CNTL, tmp | RB_RPTR_WR_ENA); + rdev->cp1.wptr = 0; + WREG32(CP_RB1_WPTR, rdev->cp1.wptr); + + /* set the wb address wether it's enabled or not */ + WREG32(CP_RB1_RPTR_ADDR, (rdev->wb.gpu_addr + RADEON_WB_CP1_RPTR_OFFSET) & 0xFFFFFFFC); + WREG32(CP_RB1_RPTR_ADDR_HI, upper_32_bits(rdev->wb.gpu_addr + RADEON_WB_CP1_RPTR_OFFSET) & 0xFF); + + mdelay(1); + WREG32(CP_RB1_CNTL, tmp); + + WREG32(CP_RB1_BASE, rdev->cp1.gpu_addr >> 8); + + rdev->cp1.rptr = RREG32(CP_RB1_RPTR); + + /* ring2 - compute only */ + /* Set ring buffer size */ + rb_bufsz = drm_order(rdev->cp2.ring_size / 8); + tmp = (drm_order(RADEON_GPU_PAGE_SIZE/8) << 8) | rb_bufsz; +#ifdef __BIG_ENDIAN + tmp |= BUF_SWAP_32BIT; +#endif + WREG32(CP_RB2_CNTL, tmp); + + /* Initialize the ring buffer's read and write pointers */ + WREG32(CP_RB2_CNTL, tmp | RB_RPTR_WR_ENA); + rdev->cp2.wptr = 0; + WREG32(CP_RB2_WPTR, rdev->cp2.wptr); + + /* set the wb address wether it's enabled or not */ + WREG32(CP_RB2_RPTR_ADDR, (rdev->wb.gpu_addr + RADEON_WB_CP2_RPTR_OFFSET) & 0xFFFFFFFC); + WREG32(CP_RB2_RPTR_ADDR_HI, upper_32_bits(rdev->wb.gpu_addr + RADEON_WB_CP2_RPTR_OFFSET) & 0xFF); + + mdelay(1); + WREG32(CP_RB2_CNTL, tmp); + + WREG32(CP_RB2_BASE, rdev->cp2.gpu_addr >> 8); + + rdev->cp2.rptr = RREG32(CP_RB2_RPTR); + + /* start the rings */ + cayman_cp_start(rdev); + rdev->cp.ready = true; + rdev->cp1.ready = true; + rdev->cp2.ready = true; + /* this only test cp0 */ + r = radeon_ring_test(rdev); + if (r) { + rdev->cp.ready = false; + rdev->cp1.ready = false; + rdev->cp2.ready = false; + return r; + } + + return 0; +} + +bool cayman_gpu_is_lockup(struct radeon_device *rdev) +{ + u32 srbm_status; + u32 grbm_status; + u32 grbm_status_se0, grbm_status_se1; + struct r100_gpu_lockup *lockup = &rdev->config.cayman.lockup; + int r; + + srbm_status = RREG32(SRBM_STATUS); + grbm_status = RREG32(GRBM_STATUS); + grbm_status_se0 = RREG32(GRBM_STATUS_SE0); + grbm_status_se1 = RREG32(GRBM_STATUS_SE1); + if (!(grbm_status & GUI_ACTIVE)) { + r100_gpu_lockup_update(lockup, &rdev->cp); + return false; + } + /* force CP activities */ + r = radeon_ring_lock(rdev, 2); + if (!r) { + /* PACKET2 NOP */ + radeon_ring_write(rdev, 0x80000000); + radeon_ring_write(rdev, 0x80000000); + radeon_ring_unlock_commit(rdev); + } + /* XXX deal with CP0,1,2 */ + rdev->cp.rptr = RREG32(CP_RB0_RPTR); + return r100_gpu_cp_is_lockup(rdev, lockup, &rdev->cp); +} + +static int cayman_gpu_soft_reset(struct radeon_device *rdev) +{ + struct evergreen_mc_save save; + u32 grbm_reset = 0; + + if (!(RREG32(GRBM_STATUS) & GUI_ACTIVE)) + return 0; + + dev_info(rdev->dev, "GPU softreset \n"); + dev_info(rdev->dev, " GRBM_STATUS=0x%08X\n", + RREG32(GRBM_STATUS)); + dev_info(rdev->dev, " GRBM_STATUS_SE0=0x%08X\n", + RREG32(GRBM_STATUS_SE0)); + dev_info(rdev->dev, " GRBM_STATUS_SE1=0x%08X\n", + RREG32(GRBM_STATUS_SE1)); + dev_info(rdev->dev, " SRBM_STATUS=0x%08X\n", + RREG32(SRBM_STATUS)); + evergreen_mc_stop(rdev, &save); + if (evergreen_mc_wait_for_idle(rdev)) { + dev_warn(rdev->dev, "Wait for MC idle timedout !\n"); + } + /* Disable CP parsing/prefetching */ + WREG32(CP_ME_CNTL, CP_ME_HALT | CP_PFP_HALT); + + /* reset all the gfx blocks */ + grbm_reset = (SOFT_RESET_CP | + SOFT_RESET_CB | + SOFT_RESET_DB | + SOFT_RESET_GDS | + SOFT_RESET_PA | + SOFT_RESET_SC | + SOFT_RESET_SPI | + SOFT_RESET_SH | + SOFT_RESET_SX | + SOFT_RESET_TC | + SOFT_RESET_TA | + SOFT_RESET_VGT | + SOFT_RESET_IA); + + dev_info(rdev->dev, " GRBM_SOFT_RESET=0x%08X\n", grbm_reset); + WREG32(GRBM_SOFT_RESET, grbm_reset); + (void)RREG32(GRBM_SOFT_RESET); + udelay(50); + WREG32(GRBM_SOFT_RESET, 0); + (void)RREG32(GRBM_SOFT_RESET); + /* Wait a little for things to settle down */ + udelay(50); + dev_info(rdev->dev, " GRBM_STATUS=0x%08X\n", + RREG32(GRBM_STATUS)); + dev_info(rdev->dev, " GRBM_STATUS_SE0=0x%08X\n", + RREG32(GRBM_STATUS_SE0)); + dev_info(rdev->dev, " GRBM_STATUS_SE1=0x%08X\n", + RREG32(GRBM_STATUS_SE1)); + dev_info(rdev->dev, " SRBM_STATUS=0x%08X\n", + RREG32(SRBM_STATUS)); + evergreen_mc_resume(rdev, &save); + return 0; +} + +int cayman_asic_reset(struct radeon_device *rdev) +{ + return cayman_gpu_soft_reset(rdev); +} + +static int cayman_startup(struct radeon_device *rdev) +{ + int r; + + if (!rdev->me_fw || !rdev->pfp_fw || !rdev->rlc_fw || !rdev->mc_fw) { + r = ni_init_microcode(rdev); + if (r) { + DRM_ERROR("Failed to load firmware!\n"); + return r; + } + } + r = ni_mc_load_microcode(rdev); + if (r) { + DRM_ERROR("Failed to load MC firmware!\n"); + return r; + } + + evergreen_mc_program(rdev); + r = cayman_pcie_gart_enable(rdev); + if (r) + return r; + cayman_gpu_init(rdev); + + r = evergreen_blit_init(rdev); + if (r) { + evergreen_blit_fini(rdev); + rdev->asic->copy = NULL; + dev_warn(rdev->dev, "failed blitter (%d) falling back to memcpy\n", r); + } + + /* allocate wb buffer */ + r = radeon_wb_init(rdev); + if (r) + return r; + + /* Enable IRQ */ + r = r600_irq_init(rdev); + if (r) { + DRM_ERROR("radeon: IH init failed (%d).\n", r); + radeon_irq_kms_fini(rdev); + return r; + } + evergreen_irq_set(rdev); + + r = radeon_ring_init(rdev, rdev->cp.ring_size); + if (r) + return r; + r = cayman_cp_load_microcode(rdev); + if (r) + return r; + r = cayman_cp_resume(rdev); + if (r) + return r; + + return 0; +} + +int cayman_resume(struct radeon_device *rdev) +{ + int r; + + /* Do not reset GPU before posting, on rv770 hw unlike on r500 hw, + * posting will perform necessary task to bring back GPU into good + * shape. + */ + /* post card */ + atom_asic_init(rdev->mode_info.atom_context); + + r = cayman_startup(rdev); + if (r) { + DRM_ERROR("cayman startup failed on resume\n"); + return r; + } + + r = r600_ib_test(rdev); + if (r) { + DRM_ERROR("radeon: failled testing IB (%d).\n", r); + return r; + } + + return r; + +} + +int cayman_suspend(struct radeon_device *rdev) +{ + int r; + + /* FIXME: we should wait for ring to be empty */ + cayman_cp_enable(rdev, false); + rdev->cp.ready = false; + evergreen_irq_suspend(rdev); + radeon_wb_disable(rdev); + cayman_pcie_gart_disable(rdev); + + /* unpin shaders bo */ + r = radeon_bo_reserve(rdev->r600_blit.shader_obj, false); + if (likely(r == 0)) { + radeon_bo_unpin(rdev->r600_blit.shader_obj); + radeon_bo_unreserve(rdev->r600_blit.shader_obj); + } + + return 0; +} + +/* Plan is to move initialization in that function and use + * helper function so that radeon_device_init pretty much + * do nothing more than calling asic specific function. This + * should also allow to remove a bunch of callback function + * like vram_info. + */ +int cayman_init(struct radeon_device *rdev) +{ + int r; + + /* This don't do much */ + r = radeon_gem_init(rdev); + if (r) + return r; + /* Read BIOS */ + if (!radeon_get_bios(rdev)) { + if (ASIC_IS_AVIVO(rdev)) + return -EINVAL; + } + /* Must be an ATOMBIOS */ + if (!rdev->is_atom_bios) { + dev_err(rdev->dev, "Expecting atombios for cayman GPU\n"); + return -EINVAL; + } + r = radeon_atombios_init(rdev); + if (r) + return r; + + /* Post card if necessary */ + if (!radeon_card_posted(rdev)) { + if (!rdev->bios) { + dev_err(rdev->dev, "Card not posted and no BIOS - ignoring\n"); + return -EINVAL; + } + DRM_INFO("GPU not posted. posting now...\n"); + atom_asic_init(rdev->mode_info.atom_context); + } + /* Initialize scratch registers */ + r600_scratch_init(rdev); + /* Initialize surface registers */ + radeon_surface_init(rdev); + /* Initialize clocks */ + radeon_get_clock_info(rdev->ddev); + /* Fence driver */ + r = radeon_fence_driver_init(rdev); + if (r) + return r; + /* initialize memory controller */ + r = evergreen_mc_init(rdev); + if (r) + return r; + /* Memory manager */ + r = radeon_bo_init(rdev); + if (r) + return r; + + r = radeon_irq_kms_init(rdev); + if (r) + return r; + + rdev->cp.ring_obj = NULL; + r600_ring_init(rdev, 1024 * 1024); + + rdev->ih.ring_obj = NULL; + r600_ih_ring_init(rdev, 64 * 1024); + + r = r600_pcie_gart_init(rdev); + if (r) + return r; + + rdev->accel_working = true; + r = cayman_startup(rdev); + if (r) { + dev_err(rdev->dev, "disabling GPU acceleration\n"); + cayman_cp_fini(rdev); + r600_irq_fini(rdev); + radeon_wb_fini(rdev); + radeon_irq_kms_fini(rdev); + cayman_pcie_gart_fini(rdev); + rdev->accel_working = false; + } + if (rdev->accel_working) { + r = radeon_ib_pool_init(rdev); + if (r) { + DRM_ERROR("radeon: failed initializing IB pool (%d).\n", r); + rdev->accel_working = false; + } + r = r600_ib_test(rdev); + if (r) { + DRM_ERROR("radeon: failed testing IB (%d).\n", r); + rdev->accel_working = false; + } + } + + /* Don't start up if the MC ucode is missing. + * The default clocks and voltages before the MC ucode + * is loaded are not suffient for advanced operations. + */ + if (!rdev->mc_fw) { + DRM_ERROR("radeon: MC ucode required for NI+.\n"); + return -EINVAL; + } + + return 0; +} + +void cayman_fini(struct radeon_device *rdev) +{ + evergreen_blit_fini(rdev); + cayman_cp_fini(rdev); + r600_irq_fini(rdev); + radeon_wb_fini(rdev); + radeon_ib_pool_fini(rdev); + radeon_irq_kms_fini(rdev); + cayman_pcie_gart_fini(rdev); + radeon_gem_fini(rdev); + radeon_fence_driver_fini(rdev); + radeon_bo_fini(rdev); + radeon_atombios_fini(rdev); + kfree(rdev->bios); + rdev->bios = NULL; +} + diff --git a/drivers/gpu/drm/radeon/ni_reg.h b/drivers/gpu/drm/radeon/ni_reg.h new file mode 100644 index 00000000..5db7b7d6 --- /dev/null +++ b/drivers/gpu/drm/radeon/ni_reg.h @@ -0,0 +1,86 @@ +/* + * Copyright 2010 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Alex Deucher + */ +#ifndef __NI_REG_H__ +#define __NI_REG_H__ + +/* northern islands - DCE5 */ + +#define NI_INPUT_GAMMA_CONTROL 0x6840 +# define NI_GRPH_INPUT_GAMMA_MODE(x) (((x) & 0x3) << 0) +# define NI_INPUT_GAMMA_USE_LUT 0 +# define NI_INPUT_GAMMA_BYPASS 1 +# define NI_INPUT_GAMMA_SRGB_24 2 +# define NI_INPUT_GAMMA_XVYCC_222 3 +# define NI_OVL_INPUT_GAMMA_MODE(x) (((x) & 0x3) << 4) + +#define NI_PRESCALE_GRPH_CONTROL 0x68b4 +# define NI_GRPH_PRESCALE_BYPASS (1 << 4) + +#define NI_PRESCALE_OVL_CONTROL 0x68c4 +# define NI_OVL_PRESCALE_BYPASS (1 << 4) + +#define NI_INPUT_CSC_CONTROL 0x68d4 +# define NI_INPUT_CSC_GRPH_MODE(x) (((x) & 0x3) << 0) +# define NI_INPUT_CSC_BYPASS 0 +# define NI_INPUT_CSC_PROG_COEFF 1 +# define NI_INPUT_CSC_PROG_SHARED_MATRIXA 2 +# define NI_INPUT_CSC_OVL_MODE(x) (((x) & 0x3) << 4) + +#define NI_OUTPUT_CSC_CONTROL 0x68f0 +# define NI_OUTPUT_CSC_GRPH_MODE(x) (((x) & 0x7) << 0) +# define NI_OUTPUT_CSC_BYPASS 0 +# define NI_OUTPUT_CSC_TV_RGB 1 +# define NI_OUTPUT_CSC_YCBCR_601 2 +# define NI_OUTPUT_CSC_YCBCR_709 3 +# define NI_OUTPUT_CSC_PROG_COEFF 4 +# define NI_OUTPUT_CSC_PROG_SHARED_MATRIXB 5 +# define NI_OUTPUT_CSC_OVL_MODE(x) (((x) & 0x7) << 4) + +#define NI_DEGAMMA_CONTROL 0x6960 +# define NI_GRPH_DEGAMMA_MODE(x) (((x) & 0x3) << 0) +# define NI_DEGAMMA_BYPASS 0 +# define NI_DEGAMMA_SRGB_24 1 +# define NI_DEGAMMA_XVYCC_222 2 +# define NI_OVL_DEGAMMA_MODE(x) (((x) & 0x3) << 4) +# define NI_ICON_DEGAMMA_MODE(x) (((x) & 0x3) << 8) +# define NI_CURSOR_DEGAMMA_MODE(x) (((x) & 0x3) << 12) + +#define NI_GAMUT_REMAP_CONTROL 0x6964 +# define NI_GRPH_GAMUT_REMAP_MODE(x) (((x) & 0x3) << 0) +# define NI_GAMUT_REMAP_BYPASS 0 +# define NI_GAMUT_REMAP_PROG_COEFF 1 +# define NI_GAMUT_REMAP_PROG_SHARED_MATRIXA 2 +# define NI_GAMUT_REMAP_PROG_SHARED_MATRIXB 3 +# define NI_OVL_GAMUT_REMAP_MODE(x) (((x) & 0x3) << 4) + +#define NI_REGAMMA_CONTROL 0x6a80 +# define NI_GRPH_REGAMMA_MODE(x) (((x) & 0x7) << 0) +# define NI_REGAMMA_BYPASS 0 +# define NI_REGAMMA_SRGB_24 1 +# define NI_REGAMMA_XVYCC_222 2 +# define NI_REGAMMA_PROG_A 3 +# define NI_REGAMMA_PROG_B 4 +# define NI_OVL_REGAMMA_MODE(x) (((x) & 0x7) << 4) + +#endif diff --git a/drivers/gpu/drm/radeon/nid.h b/drivers/gpu/drm/radeon/nid.h new file mode 100644 index 00000000..4672869c --- /dev/null +++ b/drivers/gpu/drm/radeon/nid.h @@ -0,0 +1,538 @@ +/* + * Copyright 2010 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Alex Deucher + */ +#ifndef NI_H +#define NI_H + +#define CAYMAN_MAX_SH_GPRS 256 +#define CAYMAN_MAX_TEMP_GPRS 16 +#define CAYMAN_MAX_SH_THREADS 256 +#define CAYMAN_MAX_SH_STACK_ENTRIES 4096 +#define CAYMAN_MAX_FRC_EOV_CNT 16384 +#define CAYMAN_MAX_BACKENDS 8 +#define CAYMAN_MAX_BACKENDS_MASK 0xFF +#define CAYMAN_MAX_BACKENDS_PER_SE_MASK 0xF +#define CAYMAN_MAX_SIMDS 16 +#define CAYMAN_MAX_SIMDS_MASK 0xFFFF +#define CAYMAN_MAX_SIMDS_PER_SE_MASK 0xFFF +#define CAYMAN_MAX_PIPES 8 +#define CAYMAN_MAX_PIPES_MASK 0xFF +#define CAYMAN_MAX_LDS_NUM 0xFFFF +#define CAYMAN_MAX_TCC 16 +#define CAYMAN_MAX_TCC_MASK 0xFF + +#define DMIF_ADDR_CONFIG 0xBD4 +#define SRBM_STATUS 0x0E50 + +#define VM_CONTEXT0_REQUEST_RESPONSE 0x1470 +#define REQUEST_TYPE(x) (((x) & 0xf) << 0) +#define RESPONSE_TYPE_MASK 0x000000F0 +#define RESPONSE_TYPE_SHIFT 4 +#define VM_L2_CNTL 0x1400 +#define ENABLE_L2_CACHE (1 << 0) +#define ENABLE_L2_FRAGMENT_PROCESSING (1 << 1) +#define ENABLE_L2_PTE_CACHE_LRU_UPDATE_BY_WRITE (1 << 9) +#define ENABLE_L2_PDE0_CACHE_LRU_UPDATE_BY_WRITE (1 << 10) +#define EFFECTIVE_L2_QUEUE_SIZE(x) (((x) & 7) << 14) +#define CONTEXT1_IDENTITY_ACCESS_MODE(x) (((x) & 3) << 18) +/* CONTEXT1_IDENTITY_ACCESS_MODE + * 0 physical = logical + * 1 logical via context1 page table + * 2 inside identity aperture use translation, outside physical = logical + * 3 inside identity aperture physical = logical, outside use translation + */ +#define VM_L2_CNTL2 0x1404 +#define INVALIDATE_ALL_L1_TLBS (1 << 0) +#define INVALIDATE_L2_CACHE (1 << 1) +#define VM_L2_CNTL3 0x1408 +#define BANK_SELECT(x) ((x) << 0) +#define CACHE_UPDATE_MODE(x) ((x) << 6) +#define L2_CACHE_BIGK_ASSOCIATIVITY (1 << 20) +#define L2_CACHE_BIGK_FRAGMENT_SIZE(x) ((x) << 15) +#define VM_L2_STATUS 0x140C +#define L2_BUSY (1 << 0) +#define VM_CONTEXT0_CNTL 0x1410 +#define ENABLE_CONTEXT (1 << 0) +#define PAGE_TABLE_DEPTH(x) (((x) & 3) << 1) +#define RANGE_PROTECTION_FAULT_ENABLE_DEFAULT (1 << 4) +#define VM_CONTEXT1_CNTL 0x1414 +#define VM_CONTEXT0_CNTL2 0x1430 +#define VM_CONTEXT1_CNTL2 0x1434 +#define VM_INVALIDATE_REQUEST 0x1478 +#define VM_INVALIDATE_RESPONSE 0x147c +#define VM_CONTEXT0_PROTECTION_FAULT_DEFAULT_ADDR 0x1518 +#define VM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR 0x151c +#define VM_CONTEXT0_PAGE_TABLE_BASE_ADDR 0x153C +#define VM_CONTEXT0_PAGE_TABLE_START_ADDR 0x155C +#define VM_CONTEXT0_PAGE_TABLE_END_ADDR 0x157C + +#define MC_SHARED_CHMAP 0x2004 +#define NOOFCHAN_SHIFT 12 +#define NOOFCHAN_MASK 0x00003000 +#define MC_SHARED_CHREMAP 0x2008 + +#define MC_VM_SYSTEM_APERTURE_LOW_ADDR 0x2034 +#define MC_VM_SYSTEM_APERTURE_HIGH_ADDR 0x2038 +#define MC_VM_SYSTEM_APERTURE_DEFAULT_ADDR 0x203C +#define MC_VM_MX_L1_TLB_CNTL 0x2064 +#define ENABLE_L1_TLB (1 << 0) +#define ENABLE_L1_FRAGMENT_PROCESSING (1 << 1) +#define SYSTEM_ACCESS_MODE_PA_ONLY (0 << 3) +#define SYSTEM_ACCESS_MODE_USE_SYS_MAP (1 << 3) +#define SYSTEM_ACCESS_MODE_IN_SYS (2 << 3) +#define SYSTEM_ACCESS_MODE_NOT_IN_SYS (3 << 3) +#define SYSTEM_APERTURE_UNMAPPED_ACCESS_PASS_THRU (0 << 5) +#define ENABLE_ADVANCED_DRIVER_MODEL (1 << 6) + +#define MC_SHARED_BLACKOUT_CNTL 0x20ac +#define MC_ARB_RAMCFG 0x2760 +#define NOOFBANK_SHIFT 0 +#define NOOFBANK_MASK 0x00000003 +#define NOOFRANK_SHIFT 2 +#define NOOFRANK_MASK 0x00000004 +#define NOOFROWS_SHIFT 3 +#define NOOFROWS_MASK 0x00000038 +#define NOOFCOLS_SHIFT 6 +#define NOOFCOLS_MASK 0x000000C0 +#define CHANSIZE_SHIFT 8 +#define CHANSIZE_MASK 0x00000100 +#define BURSTLENGTH_SHIFT 9 +#define BURSTLENGTH_MASK 0x00000200 +#define CHANSIZE_OVERRIDE (1 << 11) +#define MC_SEQ_SUP_CNTL 0x28c8 +#define RUN_MASK (1 << 0) +#define MC_SEQ_SUP_PGM 0x28cc +#define MC_IO_PAD_CNTL_D0 0x29d0 +#define MEM_FALL_OUT_CMD (1 << 8) +#define MC_SEQ_MISC0 0x2a00 +#define MC_SEQ_MISC0_GDDR5_SHIFT 28 +#define MC_SEQ_MISC0_GDDR5_MASK 0xf0000000 +#define MC_SEQ_MISC0_GDDR5_VALUE 5 +#define MC_SEQ_IO_DEBUG_INDEX 0x2a44 +#define MC_SEQ_IO_DEBUG_DATA 0x2a48 + +#define HDP_HOST_PATH_CNTL 0x2C00 +#define HDP_NONSURFACE_BASE 0x2C04 +#define HDP_NONSURFACE_INFO 0x2C08 +#define HDP_NONSURFACE_SIZE 0x2C0C +#define HDP_ADDR_CONFIG 0x2F48 +#define HDP_MISC_CNTL 0x2F4C +#define HDP_FLUSH_INVALIDATE_CACHE (1 << 0) + +#define CC_SYS_RB_BACKEND_DISABLE 0x3F88 +#define GC_USER_SYS_RB_BACKEND_DISABLE 0x3F8C +#define CGTS_SYS_TCC_DISABLE 0x3F90 +#define CGTS_USER_SYS_TCC_DISABLE 0x3F94 + +#define CONFIG_MEMSIZE 0x5428 + +#define HDP_MEM_COHERENCY_FLUSH_CNTL 0x5480 +#define HDP_REG_COHERENCY_FLUSH_CNTL 0x54A0 + +#define GRBM_CNTL 0x8000 +#define GRBM_READ_TIMEOUT(x) ((x) << 0) +#define GRBM_STATUS 0x8010 +#define CMDFIFO_AVAIL_MASK 0x0000000F +#define RING2_RQ_PENDING (1 << 4) +#define SRBM_RQ_PENDING (1 << 5) +#define RING1_RQ_PENDING (1 << 6) +#define CF_RQ_PENDING (1 << 7) +#define PF_RQ_PENDING (1 << 8) +#define GDS_DMA_RQ_PENDING (1 << 9) +#define GRBM_EE_BUSY (1 << 10) +#define SX_CLEAN (1 << 11) +#define DB_CLEAN (1 << 12) +#define CB_CLEAN (1 << 13) +#define TA_BUSY (1 << 14) +#define GDS_BUSY (1 << 15) +#define VGT_BUSY_NO_DMA (1 << 16) +#define VGT_BUSY (1 << 17) +#define IA_BUSY_NO_DMA (1 << 18) +#define IA_BUSY (1 << 19) +#define SX_BUSY (1 << 20) +#define SH_BUSY (1 << 21) +#define SPI_BUSY (1 << 22) +#define SC_BUSY (1 << 24) +#define PA_BUSY (1 << 25) +#define DB_BUSY (1 << 26) +#define CP_COHERENCY_BUSY (1 << 28) +#define CP_BUSY (1 << 29) +#define CB_BUSY (1 << 30) +#define GUI_ACTIVE (1 << 31) +#define GRBM_STATUS_SE0 0x8014 +#define GRBM_STATUS_SE1 0x8018 +#define SE_SX_CLEAN (1 << 0) +#define SE_DB_CLEAN (1 << 1) +#define SE_CB_CLEAN (1 << 2) +#define SE_VGT_BUSY (1 << 23) +#define SE_PA_BUSY (1 << 24) +#define SE_TA_BUSY (1 << 25) +#define SE_SX_BUSY (1 << 26) +#define SE_SPI_BUSY (1 << 27) +#define SE_SH_BUSY (1 << 28) +#define SE_SC_BUSY (1 << 29) +#define SE_DB_BUSY (1 << 30) +#define SE_CB_BUSY (1 << 31) +#define GRBM_SOFT_RESET 0x8020 +#define SOFT_RESET_CP (1 << 0) +#define SOFT_RESET_CB (1 << 1) +#define SOFT_RESET_DB (1 << 3) +#define SOFT_RESET_GDS (1 << 4) +#define SOFT_RESET_PA (1 << 5) +#define SOFT_RESET_SC (1 << 6) +#define SOFT_RESET_SPI (1 << 8) +#define SOFT_RESET_SH (1 << 9) +#define SOFT_RESET_SX (1 << 10) +#define SOFT_RESET_TC (1 << 11) +#define SOFT_RESET_TA (1 << 12) +#define SOFT_RESET_VGT (1 << 14) +#define SOFT_RESET_IA (1 << 15) + +#define SCRATCH_REG0 0x8500 +#define SCRATCH_REG1 0x8504 +#define SCRATCH_REG2 0x8508 +#define SCRATCH_REG3 0x850C +#define SCRATCH_REG4 0x8510 +#define SCRATCH_REG5 0x8514 +#define SCRATCH_REG6 0x8518 +#define SCRATCH_REG7 0x851C +#define SCRATCH_UMSK 0x8540 +#define SCRATCH_ADDR 0x8544 +#define CP_SEM_WAIT_TIMER 0x85BC +#define CP_ME_CNTL 0x86D8 +#define CP_ME_HALT (1 << 28) +#define CP_PFP_HALT (1 << 26) +#define CP_RB2_RPTR 0x86f8 +#define CP_RB1_RPTR 0x86fc +#define CP_RB0_RPTR 0x8700 +#define CP_RB_WPTR_DELAY 0x8704 +#define CP_MEQ_THRESHOLDS 0x8764 +#define MEQ1_START(x) ((x) << 0) +#define MEQ2_START(x) ((x) << 8) +#define CP_PERFMON_CNTL 0x87FC + +#define VGT_CACHE_INVALIDATION 0x88C4 +#define CACHE_INVALIDATION(x) ((x) << 0) +#define VC_ONLY 0 +#define TC_ONLY 1 +#define VC_AND_TC 2 +#define AUTO_INVLD_EN(x) ((x) << 6) +#define NO_AUTO 0 +#define ES_AUTO 1 +#define GS_AUTO 2 +#define ES_AND_GS_AUTO 3 +#define VGT_GS_VERTEX_REUSE 0x88D4 + +#define CC_GC_SHADER_PIPE_CONFIG 0x8950 +#define GC_USER_SHADER_PIPE_CONFIG 0x8954 +#define INACTIVE_QD_PIPES(x) ((x) << 8) +#define INACTIVE_QD_PIPES_MASK 0x0000FF00 +#define INACTIVE_QD_PIPES_SHIFT 8 +#define INACTIVE_SIMDS(x) ((x) << 16) +#define INACTIVE_SIMDS_MASK 0xFFFF0000 +#define INACTIVE_SIMDS_SHIFT 16 + +#define VGT_PRIMITIVE_TYPE 0x8958 +#define VGT_NUM_INSTANCES 0x8974 +#define VGT_TF_RING_SIZE 0x8988 +#define VGT_OFFCHIP_LDS_BASE 0x89b4 + +#define PA_SC_LINE_STIPPLE_STATE 0x8B10 +#define PA_CL_ENHANCE 0x8A14 +#define CLIP_VTX_REORDER_ENA (1 << 0) +#define NUM_CLIP_SEQ(x) ((x) << 1) +#define PA_SC_FIFO_SIZE 0x8BCC +#define SC_PRIM_FIFO_SIZE(x) ((x) << 0) +#define SC_HIZ_TILE_FIFO_SIZE(x) ((x) << 12) +#define SC_EARLYZ_TILE_FIFO_SIZE(x) ((x) << 20) +#define PA_SC_FORCE_EOV_MAX_CNTS 0x8B24 +#define FORCE_EOV_MAX_CLK_CNT(x) ((x) << 0) +#define FORCE_EOV_MAX_REZ_CNT(x) ((x) << 16) + +#define SQ_CONFIG 0x8C00 +#define VC_ENABLE (1 << 0) +#define EXPORT_SRC_C (1 << 1) +#define GFX_PRIO(x) ((x) << 2) +#define CS1_PRIO(x) ((x) << 4) +#define CS2_PRIO(x) ((x) << 6) +#define SQ_GPR_RESOURCE_MGMT_1 0x8C04 +#define NUM_PS_GPRS(x) ((x) << 0) +#define NUM_VS_GPRS(x) ((x) << 16) +#define NUM_CLAUSE_TEMP_GPRS(x) ((x) << 28) +#define SQ_ESGS_RING_SIZE 0x8c44 +#define SQ_GSVS_RING_SIZE 0x8c4c +#define SQ_ESTMP_RING_BASE 0x8c50 +#define SQ_ESTMP_RING_SIZE 0x8c54 +#define SQ_GSTMP_RING_BASE 0x8c58 +#define SQ_GSTMP_RING_SIZE 0x8c5c +#define SQ_VSTMP_RING_BASE 0x8c60 +#define SQ_VSTMP_RING_SIZE 0x8c64 +#define SQ_PSTMP_RING_BASE 0x8c68 +#define SQ_PSTMP_RING_SIZE 0x8c6c +#define SQ_MS_FIFO_SIZES 0x8CF0 +#define CACHE_FIFO_SIZE(x) ((x) << 0) +#define FETCH_FIFO_HIWATER(x) ((x) << 8) +#define DONE_FIFO_HIWATER(x) ((x) << 16) +#define ALU_UPDATE_FIFO_HIWATER(x) ((x) << 24) +#define SQ_LSTMP_RING_BASE 0x8e10 +#define SQ_LSTMP_RING_SIZE 0x8e14 +#define SQ_HSTMP_RING_BASE 0x8e18 +#define SQ_HSTMP_RING_SIZE 0x8e1c +#define SQ_DYN_GPR_CNTL_PS_FLUSH_REQ 0x8D8C +#define DYN_GPR_ENABLE (1 << 8) +#define SQ_CONST_MEM_BASE 0x8df8 + +#define SX_EXPORT_BUFFER_SIZES 0x900C +#define COLOR_BUFFER_SIZE(x) ((x) << 0) +#define POSITION_BUFFER_SIZE(x) ((x) << 8) +#define SMX_BUFFER_SIZE(x) ((x) << 16) +#define SX_DEBUG_1 0x9058 +#define ENABLE_NEW_SMX_ADDRESS (1 << 16) + +#define SPI_CONFIG_CNTL 0x9100 +#define GPR_WRITE_PRIORITY(x) ((x) << 0) +#define SPI_CONFIG_CNTL_1 0x913C +#define VTX_DONE_DELAY(x) ((x) << 0) +#define INTERP_ONE_PRIM_PER_ROW (1 << 4) +#define CRC_SIMD_ID_WADDR_DISABLE (1 << 8) + +#define CGTS_TCC_DISABLE 0x9148 +#define CGTS_USER_TCC_DISABLE 0x914C +#define TCC_DISABLE_MASK 0xFFFF0000 +#define TCC_DISABLE_SHIFT 16 +#define CGTS_SM_CTRL_REG 0x9150 +#define OVERRIDE (1 << 21) + +#define TA_CNTL_AUX 0x9508 +#define DISABLE_CUBE_WRAP (1 << 0) +#define DISABLE_CUBE_ANISO (1 << 1) + +#define TCP_CHAN_STEER_LO 0x960c +#define TCP_CHAN_STEER_HI 0x9610 + +#define CC_RB_BACKEND_DISABLE 0x98F4 +#define BACKEND_DISABLE(x) ((x) << 16) +#define GB_ADDR_CONFIG 0x98F8 +#define NUM_PIPES(x) ((x) << 0) +#define NUM_PIPES_MASK 0x00000007 +#define NUM_PIPES_SHIFT 0 +#define PIPE_INTERLEAVE_SIZE(x) ((x) << 4) +#define PIPE_INTERLEAVE_SIZE_MASK 0x00000070 +#define PIPE_INTERLEAVE_SIZE_SHIFT 4 +#define BANK_INTERLEAVE_SIZE(x) ((x) << 8) +#define NUM_SHADER_ENGINES(x) ((x) << 12) +#define NUM_SHADER_ENGINES_MASK 0x00003000 +#define NUM_SHADER_ENGINES_SHIFT 12 +#define SHADER_ENGINE_TILE_SIZE(x) ((x) << 16) +#define SHADER_ENGINE_TILE_SIZE_MASK 0x00070000 +#define SHADER_ENGINE_TILE_SIZE_SHIFT 16 +#define NUM_GPUS(x) ((x) << 20) +#define NUM_GPUS_MASK 0x00700000 +#define NUM_GPUS_SHIFT 20 +#define MULTI_GPU_TILE_SIZE(x) ((x) << 24) +#define MULTI_GPU_TILE_SIZE_MASK 0x03000000 +#define MULTI_GPU_TILE_SIZE_SHIFT 24 +#define ROW_SIZE(x) ((x) << 28) +#define ROW_SIZE_MASK 0x30000000 +#define ROW_SIZE_SHIFT 28 +#define NUM_LOWER_PIPES(x) ((x) << 30) +#define NUM_LOWER_PIPES_MASK 0x40000000 +#define NUM_LOWER_PIPES_SHIFT 30 +#define GB_BACKEND_MAP 0x98FC + +#define CB_PERF_CTR0_SEL_0 0x9A20 +#define CB_PERF_CTR0_SEL_1 0x9A24 +#define CB_PERF_CTR1_SEL_0 0x9A28 +#define CB_PERF_CTR1_SEL_1 0x9A2C +#define CB_PERF_CTR2_SEL_0 0x9A30 +#define CB_PERF_CTR2_SEL_1 0x9A34 +#define CB_PERF_CTR3_SEL_0 0x9A38 +#define CB_PERF_CTR3_SEL_1 0x9A3C + +#define GC_USER_RB_BACKEND_DISABLE 0x9B7C +#define BACKEND_DISABLE_MASK 0x00FF0000 +#define BACKEND_DISABLE_SHIFT 16 + +#define SMX_DC_CTL0 0xA020 +#define USE_HASH_FUNCTION (1 << 0) +#define NUMBER_OF_SETS(x) ((x) << 1) +#define FLUSH_ALL_ON_EVENT (1 << 10) +#define STALL_ON_EVENT (1 << 11) +#define SMX_EVENT_CTL 0xA02C +#define ES_FLUSH_CTL(x) ((x) << 0) +#define GS_FLUSH_CTL(x) ((x) << 3) +#define ACK_FLUSH_CTL(x) ((x) << 6) +#define SYNC_FLUSH_CTL (1 << 8) + +#define CP_RB0_BASE 0xC100 +#define CP_RB0_CNTL 0xC104 +#define RB_BUFSZ(x) ((x) << 0) +#define RB_BLKSZ(x) ((x) << 8) +#define RB_NO_UPDATE (1 << 27) +#define RB_RPTR_WR_ENA (1 << 31) +#define BUF_SWAP_32BIT (2 << 16) +#define CP_RB0_RPTR_ADDR 0xC10C +#define CP_RB0_RPTR_ADDR_HI 0xC110 +#define CP_RB0_WPTR 0xC114 +#define CP_RB1_BASE 0xC180 +#define CP_RB1_CNTL 0xC184 +#define CP_RB1_RPTR_ADDR 0xC188 +#define CP_RB1_RPTR_ADDR_HI 0xC18C +#define CP_RB1_WPTR 0xC190 +#define CP_RB2_BASE 0xC194 +#define CP_RB2_CNTL 0xC198 +#define CP_RB2_RPTR_ADDR 0xC19C +#define CP_RB2_RPTR_ADDR_HI 0xC1A0 +#define CP_RB2_WPTR 0xC1A4 +#define CP_PFP_UCODE_ADDR 0xC150 +#define CP_PFP_UCODE_DATA 0xC154 +#define CP_ME_RAM_RADDR 0xC158 +#define CP_ME_RAM_WADDR 0xC15C +#define CP_ME_RAM_DATA 0xC160 +#define CP_DEBUG 0xC1FC + +/* + * PM4 + */ +#define PACKET_TYPE0 0 +#define PACKET_TYPE1 1 +#define PACKET_TYPE2 2 +#define PACKET_TYPE3 3 + +#define CP_PACKET_GET_TYPE(h) (((h) >> 30) & 3) +#define CP_PACKET_GET_COUNT(h) (((h) >> 16) & 0x3FFF) +#define CP_PACKET0_GET_REG(h) (((h) & 0xFFFF) << 2) +#define CP_PACKET3_GET_OPCODE(h) (((h) >> 8) & 0xFF) +#define PACKET0(reg, n) ((PACKET_TYPE0 << 30) | \ + (((reg) >> 2) & 0xFFFF) | \ + ((n) & 0x3FFF) << 16) +#define CP_PACKET2 0x80000000 +#define PACKET2_PAD_SHIFT 0 +#define PACKET2_PAD_MASK (0x3fffffff << 0) + +#define PACKET2(v) (CP_PACKET2 | REG_SET(PACKET2_PAD, (v))) + +#define PACKET3(op, n) ((PACKET_TYPE3 << 30) | \ + (((op) & 0xFF) << 8) | \ + ((n) & 0x3FFF) << 16) + +/* Packet 3 types */ +#define PACKET3_NOP 0x10 +#define PACKET3_SET_BASE 0x11 +#define PACKET3_CLEAR_STATE 0x12 +#define PACKET3_INDEX_BUFFER_SIZE 0x13 +#define PACKET3_DEALLOC_STATE 0x14 +#define PACKET3_DISPATCH_DIRECT 0x15 +#define PACKET3_DISPATCH_INDIRECT 0x16 +#define PACKET3_INDIRECT_BUFFER_END 0x17 +#define PACKET3_SET_PREDICATION 0x20 +#define PACKET3_REG_RMW 0x21 +#define PACKET3_COND_EXEC 0x22 +#define PACKET3_PRED_EXEC 0x23 +#define PACKET3_DRAW_INDIRECT 0x24 +#define PACKET3_DRAW_INDEX_INDIRECT 0x25 +#define PACKET3_INDEX_BASE 0x26 +#define PACKET3_DRAW_INDEX_2 0x27 +#define PACKET3_CONTEXT_CONTROL 0x28 +#define PACKET3_DRAW_INDEX_OFFSET 0x29 +#define PACKET3_INDEX_TYPE 0x2A +#define PACKET3_DRAW_INDEX 0x2B +#define PACKET3_DRAW_INDEX_AUTO 0x2D +#define PACKET3_DRAW_INDEX_IMMD 0x2E +#define PACKET3_NUM_INSTANCES 0x2F +#define PACKET3_DRAW_INDEX_MULTI_AUTO 0x30 +#define PACKET3_INDIRECT_BUFFER 0x32 +#define PACKET3_STRMOUT_BUFFER_UPDATE 0x34 +#define PACKET3_DRAW_INDEX_OFFSET_2 0x35 +#define PACKET3_DRAW_INDEX_MULTI_ELEMENT 0x36 +#define PACKET3_WRITE_DATA 0x37 +#define PACKET3_MEM_SEMAPHORE 0x39 +#define PACKET3_MPEG_INDEX 0x3A +#define PACKET3_WAIT_REG_MEM 0x3C +#define PACKET3_MEM_WRITE 0x3D +#define PACKET3_SURFACE_SYNC 0x43 +# define PACKET3_CB0_DEST_BASE_ENA (1 << 6) +# define PACKET3_CB1_DEST_BASE_ENA (1 << 7) +# define PACKET3_CB2_DEST_BASE_ENA (1 << 8) +# define PACKET3_CB3_DEST_BASE_ENA (1 << 9) +# define PACKET3_CB4_DEST_BASE_ENA (1 << 10) +# define PACKET3_CB5_DEST_BASE_ENA (1 << 11) +# define PACKET3_CB6_DEST_BASE_ENA (1 << 12) +# define PACKET3_CB7_DEST_BASE_ENA (1 << 13) +# define PACKET3_DB_DEST_BASE_ENA (1 << 14) +# define PACKET3_CB8_DEST_BASE_ENA (1 << 15) +# define PACKET3_CB9_DEST_BASE_ENA (1 << 16) +# define PACKET3_CB10_DEST_BASE_ENA (1 << 17) +# define PACKET3_CB11_DEST_BASE_ENA (1 << 18) +# define PACKET3_FULL_CACHE_ENA (1 << 20) +# define PACKET3_TC_ACTION_ENA (1 << 23) +# define PACKET3_CB_ACTION_ENA (1 << 25) +# define PACKET3_DB_ACTION_ENA (1 << 26) +# define PACKET3_SH_ACTION_ENA (1 << 27) +# define PACKET3_SX_ACTION_ENA (1 << 28) +#define PACKET3_ME_INITIALIZE 0x44 +#define PACKET3_ME_INITIALIZE_DEVICE_ID(x) ((x) << 16) +#define PACKET3_COND_WRITE 0x45 +#define PACKET3_EVENT_WRITE 0x46 +#define PACKET3_EVENT_WRITE_EOP 0x47 +#define PACKET3_EVENT_WRITE_EOS 0x48 +#define PACKET3_PREAMBLE_CNTL 0x4A +# define PACKET3_PREAMBLE_BEGIN_CLEAR_STATE (2 << 28) +# define PACKET3_PREAMBLE_END_CLEAR_STATE (3 << 28) +#define PACKET3_ALU_PS_CONST_BUFFER_COPY 0x4C +#define PACKET3_ALU_VS_CONST_BUFFER_COPY 0x4D +#define PACKET3_ALU_PS_CONST_UPDATE 0x4E +#define PACKET3_ALU_VS_CONST_UPDATE 0x4F +#define PACKET3_ONE_REG_WRITE 0x57 +#define PACKET3_SET_CONFIG_REG 0x68 +#define PACKET3_SET_CONFIG_REG_START 0x00008000 +#define PACKET3_SET_CONFIG_REG_END 0x0000ac00 +#define PACKET3_SET_CONTEXT_REG 0x69 +#define PACKET3_SET_CONTEXT_REG_START 0x00028000 +#define PACKET3_SET_CONTEXT_REG_END 0x00029000 +#define PACKET3_SET_ALU_CONST 0x6A +/* alu const buffers only; no reg file */ +#define PACKET3_SET_BOOL_CONST 0x6B +#define PACKET3_SET_BOOL_CONST_START 0x0003a500 +#define PACKET3_SET_BOOL_CONST_END 0x0003a518 +#define PACKET3_SET_LOOP_CONST 0x6C +#define PACKET3_SET_LOOP_CONST_START 0x0003a200 +#define PACKET3_SET_LOOP_CONST_END 0x0003a500 +#define PACKET3_SET_RESOURCE 0x6D +#define PACKET3_SET_RESOURCE_START 0x00030000 +#define PACKET3_SET_RESOURCE_END 0x00038000 +#define PACKET3_SET_SAMPLER 0x6E +#define PACKET3_SET_SAMPLER_START 0x0003c000 +#define PACKET3_SET_SAMPLER_END 0x0003c600 +#define PACKET3_SET_CTL_CONST 0x6F +#define PACKET3_SET_CTL_CONST_START 0x0003cff0 +#define PACKET3_SET_CTL_CONST_END 0x0003ff0c +#define PACKET3_SET_RESOURCE_OFFSET 0x70 +#define PACKET3_SET_ALU_CONST_VS 0x71 +#define PACKET3_SET_ALU_CONST_DI 0x72 +#define PACKET3_SET_CONTEXT_REG_INDIRECT 0x73 +#define PACKET3_SET_RESOURCE_INDIRECT 0x74 +#define PACKET3_SET_APPEND_CNT 0x75 + +#endif + diff --git a/drivers/gpu/drm/radeon/r100.c b/drivers/gpu/drm/radeon/r100.c new file mode 100644 index 00000000..d94f440f --- /dev/null +++ b/drivers/gpu/drm/radeon/r100.c @@ -0,0 +1,3972 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#include +#include +#include "drmP.h" +#include "drm.h" +#include "radeon_drm.h" +#include "radeon_reg.h" +#include "radeon.h" +#include "radeon_asic.h" +#include "r100d.h" +#include "rs100d.h" +#include "rv200d.h" +#include "rv250d.h" +#include "atom.h" + +#include +#include + +#include "r100_reg_safe.h" +#include "rn50_reg_safe.h" + +/* Firmware Names */ +#define FIRMWARE_R100 "radeon/R100_cp.bin" +#define FIRMWARE_R200 "radeon/R200_cp.bin" +#define FIRMWARE_R300 "radeon/R300_cp.bin" +#define FIRMWARE_R420 "radeon/R420_cp.bin" +#define FIRMWARE_RS690 "radeon/RS690_cp.bin" +#define FIRMWARE_RS600 "radeon/RS600_cp.bin" +#define FIRMWARE_R520 "radeon/R520_cp.bin" + +MODULE_FIRMWARE(FIRMWARE_R100); +MODULE_FIRMWARE(FIRMWARE_R200); +MODULE_FIRMWARE(FIRMWARE_R300); +MODULE_FIRMWARE(FIRMWARE_R420); +MODULE_FIRMWARE(FIRMWARE_RS690); +MODULE_FIRMWARE(FIRMWARE_RS600); +MODULE_FIRMWARE(FIRMWARE_R520); + +#include "r100_track.h" + +/* This files gather functions specifics to: + * r100,rv100,rs100,rv200,rs200,r200,rv250,rs300,rv280 + */ + +void r100_pre_page_flip(struct radeon_device *rdev, int crtc) +{ + /* enable the pflip int */ + radeon_irq_kms_pflip_irq_get(rdev, crtc); +} + +void r100_post_page_flip(struct radeon_device *rdev, int crtc) +{ + /* disable the pflip int */ + radeon_irq_kms_pflip_irq_put(rdev, crtc); +} + +u32 r100_page_flip(struct radeon_device *rdev, int crtc_id, u64 crtc_base) +{ + struct radeon_crtc *radeon_crtc = rdev->mode_info.crtcs[crtc_id]; + u32 tmp = ((u32)crtc_base) | RADEON_CRTC_OFFSET__OFFSET_LOCK; + int i; + + /* Lock the graphics update lock */ + /* update the scanout addresses */ + WREG32(RADEON_CRTC_OFFSET + radeon_crtc->crtc_offset, tmp); + + /* Wait for update_pending to go high. */ + for (i = 0; i < rdev->usec_timeout; i++) { + if (RREG32(RADEON_CRTC_OFFSET + radeon_crtc->crtc_offset) & RADEON_CRTC_OFFSET__GUI_TRIG_OFFSET) + break; + udelay(1); + } + DRM_DEBUG("Update pending now high. Unlocking vupdate_lock.\n"); + + /* Unlock the lock, so double-buffering can take place inside vblank */ + tmp &= ~RADEON_CRTC_OFFSET__OFFSET_LOCK; + WREG32(RADEON_CRTC_OFFSET + radeon_crtc->crtc_offset, tmp); + + /* Return current update_pending status: */ + return RREG32(RADEON_CRTC_OFFSET + radeon_crtc->crtc_offset) & RADEON_CRTC_OFFSET__GUI_TRIG_OFFSET; +} + +void r100_pm_get_dynpm_state(struct radeon_device *rdev) +{ + int i; + rdev->pm.dynpm_can_upclock = true; + rdev->pm.dynpm_can_downclock = true; + + switch (rdev->pm.dynpm_planned_action) { + case DYNPM_ACTION_MINIMUM: + rdev->pm.requested_power_state_index = 0; + rdev->pm.dynpm_can_downclock = false; + break; + case DYNPM_ACTION_DOWNCLOCK: + if (rdev->pm.current_power_state_index == 0) { + rdev->pm.requested_power_state_index = rdev->pm.current_power_state_index; + rdev->pm.dynpm_can_downclock = false; + } else { + if (rdev->pm.active_crtc_count > 1) { + for (i = 0; i < rdev->pm.num_power_states; i++) { + if (rdev->pm.power_state[i].flags & RADEON_PM_STATE_SINGLE_DISPLAY_ONLY) + continue; + else if (i >= rdev->pm.current_power_state_index) { + rdev->pm.requested_power_state_index = rdev->pm.current_power_state_index; + break; + } else { + rdev->pm.requested_power_state_index = i; + break; + } + } + } else + rdev->pm.requested_power_state_index = + rdev->pm.current_power_state_index - 1; + } + /* don't use the power state if crtcs are active and no display flag is set */ + if ((rdev->pm.active_crtc_count > 0) && + (rdev->pm.power_state[rdev->pm.requested_power_state_index].clock_info[0].flags & + RADEON_PM_MODE_NO_DISPLAY)) { + rdev->pm.requested_power_state_index++; + } + break; + case DYNPM_ACTION_UPCLOCK: + if (rdev->pm.current_power_state_index == (rdev->pm.num_power_states - 1)) { + rdev->pm.requested_power_state_index = rdev->pm.current_power_state_index; + rdev->pm.dynpm_can_upclock = false; + } else { + if (rdev->pm.active_crtc_count > 1) { + for (i = (rdev->pm.num_power_states - 1); i >= 0; i--) { + if (rdev->pm.power_state[i].flags & RADEON_PM_STATE_SINGLE_DISPLAY_ONLY) + continue; + else if (i <= rdev->pm.current_power_state_index) { + rdev->pm.requested_power_state_index = rdev->pm.current_power_state_index; + break; + } else { + rdev->pm.requested_power_state_index = i; + break; + } + } + } else + rdev->pm.requested_power_state_index = + rdev->pm.current_power_state_index + 1; + } + break; + case DYNPM_ACTION_DEFAULT: + rdev->pm.requested_power_state_index = rdev->pm.default_power_state_index; + rdev->pm.dynpm_can_upclock = false; + break; + case DYNPM_ACTION_NONE: + default: + DRM_ERROR("Requested mode for not defined action\n"); + return; + } + /* only one clock mode per power state */ + rdev->pm.requested_clock_mode_index = 0; + + DRM_DEBUG_DRIVER("Requested: e: %d m: %d p: %d\n", + rdev->pm.power_state[rdev->pm.requested_power_state_index]. + clock_info[rdev->pm.requested_clock_mode_index].sclk, + rdev->pm.power_state[rdev->pm.requested_power_state_index]. + clock_info[rdev->pm.requested_clock_mode_index].mclk, + rdev->pm.power_state[rdev->pm.requested_power_state_index]. + pcie_lanes); +} + +void r100_pm_init_profile(struct radeon_device *rdev) +{ + /* default */ + rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_cm_idx = 0; + /* low sh */ + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_ps_idx = 0; + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_ps_idx = 0; + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_cm_idx = 0; + /* mid sh */ + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_ps_idx = 0; + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_ps_idx = 0; + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_cm_idx = 0; + /* high sh */ + rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_ps_idx = 0; + rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_cm_idx = 0; + /* low mh */ + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_ps_idx = 0; + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_cm_idx = 0; + /* mid mh */ + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_ps_idx = 0; + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_cm_idx = 0; + /* high mh */ + rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_ps_idx = 0; + rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_cm_idx = 0; +} + +void r100_pm_misc(struct radeon_device *rdev) +{ + int requested_index = rdev->pm.requested_power_state_index; + struct radeon_power_state *ps = &rdev->pm.power_state[requested_index]; + struct radeon_voltage *voltage = &ps->clock_info[0].voltage; + u32 tmp, sclk_cntl, sclk_cntl2, sclk_more_cntl; + + if ((voltage->type == VOLTAGE_GPIO) && (voltage->gpio.valid)) { + if (ps->misc & ATOM_PM_MISCINFO_VOLTAGE_DROP_SUPPORT) { + tmp = RREG32(voltage->gpio.reg); + if (voltage->active_high) + tmp |= voltage->gpio.mask; + else + tmp &= ~(voltage->gpio.mask); + WREG32(voltage->gpio.reg, tmp); + if (voltage->delay) + udelay(voltage->delay); + } else { + tmp = RREG32(voltage->gpio.reg); + if (voltage->active_high) + tmp &= ~voltage->gpio.mask; + else + tmp |= voltage->gpio.mask; + WREG32(voltage->gpio.reg, tmp); + if (voltage->delay) + udelay(voltage->delay); + } + } + + sclk_cntl = RREG32_PLL(SCLK_CNTL); + sclk_cntl2 = RREG32_PLL(SCLK_CNTL2); + sclk_cntl2 &= ~REDUCED_SPEED_SCLK_SEL(3); + sclk_more_cntl = RREG32_PLL(SCLK_MORE_CNTL); + sclk_more_cntl &= ~VOLTAGE_DELAY_SEL(3); + if (ps->misc & ATOM_PM_MISCINFO_ASIC_REDUCED_SPEED_SCLK_EN) { + sclk_more_cntl |= REDUCED_SPEED_SCLK_EN; + if (ps->misc & ATOM_PM_MISCINFO_DYN_CLK_3D_IDLE) + sclk_cntl2 |= REDUCED_SPEED_SCLK_MODE; + else + sclk_cntl2 &= ~REDUCED_SPEED_SCLK_MODE; + if (ps->misc & ATOM_PM_MISCINFO_DYNAMIC_CLOCK_DIVIDER_BY_2) + sclk_cntl2 |= REDUCED_SPEED_SCLK_SEL(0); + else if (ps->misc & ATOM_PM_MISCINFO_DYNAMIC_CLOCK_DIVIDER_BY_4) + sclk_cntl2 |= REDUCED_SPEED_SCLK_SEL(2); + } else + sclk_more_cntl &= ~REDUCED_SPEED_SCLK_EN; + + if (ps->misc & ATOM_PM_MISCINFO_ASIC_DYNAMIC_VOLTAGE_EN) { + sclk_more_cntl |= IO_CG_VOLTAGE_DROP; + if (voltage->delay) { + sclk_more_cntl |= VOLTAGE_DROP_SYNC; + switch (voltage->delay) { + case 33: + sclk_more_cntl |= VOLTAGE_DELAY_SEL(0); + break; + case 66: + sclk_more_cntl |= VOLTAGE_DELAY_SEL(1); + break; + case 99: + sclk_more_cntl |= VOLTAGE_DELAY_SEL(2); + break; + case 132: + sclk_more_cntl |= VOLTAGE_DELAY_SEL(3); + break; + } + } else + sclk_more_cntl &= ~VOLTAGE_DROP_SYNC; + } else + sclk_more_cntl &= ~IO_CG_VOLTAGE_DROP; + + if (ps->misc & ATOM_PM_MISCINFO_DYNAMIC_HDP_BLOCK_EN) + sclk_cntl &= ~FORCE_HDP; + else + sclk_cntl |= FORCE_HDP; + + WREG32_PLL(SCLK_CNTL, sclk_cntl); + WREG32_PLL(SCLK_CNTL2, sclk_cntl2); + WREG32_PLL(SCLK_MORE_CNTL, sclk_more_cntl); + + /* set pcie lanes */ + if ((rdev->flags & RADEON_IS_PCIE) && + !(rdev->flags & RADEON_IS_IGP) && + rdev->asic->set_pcie_lanes && + (ps->pcie_lanes != + rdev->pm.power_state[rdev->pm.current_power_state_index].pcie_lanes)) { + radeon_set_pcie_lanes(rdev, + ps->pcie_lanes); + DRM_DEBUG_DRIVER("Setting: p: %d\n", ps->pcie_lanes); + } +} + +void r100_pm_prepare(struct radeon_device *rdev) +{ + struct drm_device *ddev = rdev->ddev; + struct drm_crtc *crtc; + struct radeon_crtc *radeon_crtc; + u32 tmp; + + /* disable any active CRTCs */ + list_for_each_entry(crtc, &ddev->mode_config.crtc_list, head) { + radeon_crtc = to_radeon_crtc(crtc); + if (radeon_crtc->enabled) { + if (radeon_crtc->crtc_id) { + tmp = RREG32(RADEON_CRTC2_GEN_CNTL); + tmp |= RADEON_CRTC2_DISP_REQ_EN_B; + WREG32(RADEON_CRTC2_GEN_CNTL, tmp); + } else { + tmp = RREG32(RADEON_CRTC_GEN_CNTL); + tmp |= RADEON_CRTC_DISP_REQ_EN_B; + WREG32(RADEON_CRTC_GEN_CNTL, tmp); + } + } + } +} + +void r100_pm_finish(struct radeon_device *rdev) +{ + struct drm_device *ddev = rdev->ddev; + struct drm_crtc *crtc; + struct radeon_crtc *radeon_crtc; + u32 tmp; + + /* enable any active CRTCs */ + list_for_each_entry(crtc, &ddev->mode_config.crtc_list, head) { + radeon_crtc = to_radeon_crtc(crtc); + if (radeon_crtc->enabled) { + if (radeon_crtc->crtc_id) { + tmp = RREG32(RADEON_CRTC2_GEN_CNTL); + tmp &= ~RADEON_CRTC2_DISP_REQ_EN_B; + WREG32(RADEON_CRTC2_GEN_CNTL, tmp); + } else { + tmp = RREG32(RADEON_CRTC_GEN_CNTL); + tmp &= ~RADEON_CRTC_DISP_REQ_EN_B; + WREG32(RADEON_CRTC_GEN_CNTL, tmp); + } + } + } +} + +bool r100_gui_idle(struct radeon_device *rdev) +{ + if (RREG32(RADEON_RBBM_STATUS) & RADEON_RBBM_ACTIVE) + return false; + else + return true; +} + +/* hpd for digital panel detect/disconnect */ +bool r100_hpd_sense(struct radeon_device *rdev, enum radeon_hpd_id hpd) +{ + bool connected = false; + + switch (hpd) { + case RADEON_HPD_1: + if (RREG32(RADEON_FP_GEN_CNTL) & RADEON_FP_DETECT_SENSE) + connected = true; + break; + case RADEON_HPD_2: + if (RREG32(RADEON_FP2_GEN_CNTL) & RADEON_FP2_DETECT_SENSE) + connected = true; + break; + default: + break; + } + return connected; +} + +void r100_hpd_set_polarity(struct radeon_device *rdev, + enum radeon_hpd_id hpd) +{ + u32 tmp; + bool connected = r100_hpd_sense(rdev, hpd); + + switch (hpd) { + case RADEON_HPD_1: + tmp = RREG32(RADEON_FP_GEN_CNTL); + if (connected) + tmp &= ~RADEON_FP_DETECT_INT_POL; + else + tmp |= RADEON_FP_DETECT_INT_POL; + WREG32(RADEON_FP_GEN_CNTL, tmp); + break; + case RADEON_HPD_2: + tmp = RREG32(RADEON_FP2_GEN_CNTL); + if (connected) + tmp &= ~RADEON_FP2_DETECT_INT_POL; + else + tmp |= RADEON_FP2_DETECT_INT_POL; + WREG32(RADEON_FP2_GEN_CNTL, tmp); + break; + default: + break; + } +} + +void r100_hpd_init(struct radeon_device *rdev) +{ + struct drm_device *dev = rdev->ddev; + struct drm_connector *connector; + + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + switch (radeon_connector->hpd.hpd) { + case RADEON_HPD_1: + rdev->irq.hpd[0] = true; + break; + case RADEON_HPD_2: + rdev->irq.hpd[1] = true; + break; + default: + break; + } + radeon_hpd_set_polarity(rdev, radeon_connector->hpd.hpd); + } + if (rdev->irq.installed) + r100_irq_set(rdev); +} + +void r100_hpd_fini(struct radeon_device *rdev) +{ + struct drm_device *dev = rdev->ddev; + struct drm_connector *connector; + + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + switch (radeon_connector->hpd.hpd) { + case RADEON_HPD_1: + rdev->irq.hpd[0] = false; + break; + case RADEON_HPD_2: + rdev->irq.hpd[1] = false; + break; + default: + break; + } + } +} + +/* + * PCI GART + */ +void r100_pci_gart_tlb_flush(struct radeon_device *rdev) +{ + /* TODO: can we do somethings here ? */ + /* It seems hw only cache one entry so we should discard this + * entry otherwise if first GPU GART read hit this entry it + * could end up in wrong address. */ +} + +int r100_pci_gart_init(struct radeon_device *rdev) +{ + int r; + + if (rdev->gart.table.ram.ptr) { + WARN(1, "R100 PCI GART already initialized\n"); + return 0; + } + /* Initialize common gart structure */ + r = radeon_gart_init(rdev); + if (r) + return r; + rdev->gart.table_size = rdev->gart.num_gpu_pages * 4; + rdev->asic->gart_tlb_flush = &r100_pci_gart_tlb_flush; + rdev->asic->gart_set_page = &r100_pci_gart_set_page; + return radeon_gart_table_ram_alloc(rdev); +} + +/* required on r1xx, r2xx, r300, r(v)350, r420/r481, rs400/rs480 */ +void r100_enable_bm(struct radeon_device *rdev) +{ + uint32_t tmp; + /* Enable bus mastering */ + tmp = RREG32(RADEON_BUS_CNTL) & ~RADEON_BUS_MASTER_DIS; + WREG32(RADEON_BUS_CNTL, tmp); +} + +int r100_pci_gart_enable(struct radeon_device *rdev) +{ + uint32_t tmp; + + radeon_gart_restore(rdev); + /* discard memory request outside of configured range */ + tmp = RREG32(RADEON_AIC_CNTL) | RADEON_DIS_OUT_OF_PCI_GART_ACCESS; + WREG32(RADEON_AIC_CNTL, tmp); + /* set address range for PCI address translate */ + WREG32(RADEON_AIC_LO_ADDR, rdev->mc.gtt_start); + WREG32(RADEON_AIC_HI_ADDR, rdev->mc.gtt_end); + /* set PCI GART page-table base address */ + WREG32(RADEON_AIC_PT_BASE, rdev->gart.table_addr); + tmp = RREG32(RADEON_AIC_CNTL) | RADEON_PCIGART_TRANSLATE_EN; + WREG32(RADEON_AIC_CNTL, tmp); + r100_pci_gart_tlb_flush(rdev); + rdev->gart.ready = true; + return 0; +} + +void r100_pci_gart_disable(struct radeon_device *rdev) +{ + uint32_t tmp; + + /* discard memory request outside of configured range */ + tmp = RREG32(RADEON_AIC_CNTL) | RADEON_DIS_OUT_OF_PCI_GART_ACCESS; + WREG32(RADEON_AIC_CNTL, tmp & ~RADEON_PCIGART_TRANSLATE_EN); + WREG32(RADEON_AIC_LO_ADDR, 0); + WREG32(RADEON_AIC_HI_ADDR, 0); +} + +int r100_pci_gart_set_page(struct radeon_device *rdev, int i, uint64_t addr) +{ + if (i < 0 || i > rdev->gart.num_gpu_pages) { + return -EINVAL; + } + rdev->gart.table.ram.ptr[i] = cpu_to_le32(lower_32_bits(addr)); + return 0; +} + +void r100_pci_gart_fini(struct radeon_device *rdev) +{ + radeon_gart_fini(rdev); + r100_pci_gart_disable(rdev); + radeon_gart_table_ram_free(rdev); +} + +int r100_irq_set(struct radeon_device *rdev) +{ + uint32_t tmp = 0; + + if (!rdev->irq.installed) { + WARN(1, "Can't enable IRQ/MSI because no handler is installed\n"); + WREG32(R_000040_GEN_INT_CNTL, 0); + return -EINVAL; + } + if (rdev->irq.sw_int) { + tmp |= RADEON_SW_INT_ENABLE; + } + if (rdev->irq.gui_idle) { + tmp |= RADEON_GUI_IDLE_MASK; + } + if (rdev->irq.crtc_vblank_int[0] || + rdev->irq.pflip[0]) { + tmp |= RADEON_CRTC_VBLANK_MASK; + } + if (rdev->irq.crtc_vblank_int[1] || + rdev->irq.pflip[1]) { + tmp |= RADEON_CRTC2_VBLANK_MASK; + } + if (rdev->irq.hpd[0]) { + tmp |= RADEON_FP_DETECT_MASK; + } + if (rdev->irq.hpd[1]) { + tmp |= RADEON_FP2_DETECT_MASK; + } + WREG32(RADEON_GEN_INT_CNTL, tmp); + return 0; +} + +void r100_irq_disable(struct radeon_device *rdev) +{ + u32 tmp; + + WREG32(R_000040_GEN_INT_CNTL, 0); + /* Wait and acknowledge irq */ + mdelay(1); + tmp = RREG32(R_000044_GEN_INT_STATUS); + WREG32(R_000044_GEN_INT_STATUS, tmp); +} + +static inline uint32_t r100_irq_ack(struct radeon_device *rdev) +{ + uint32_t irqs = RREG32(RADEON_GEN_INT_STATUS); + uint32_t irq_mask = RADEON_SW_INT_TEST | + RADEON_CRTC_VBLANK_STAT | RADEON_CRTC2_VBLANK_STAT | + RADEON_FP_DETECT_STAT | RADEON_FP2_DETECT_STAT; + + /* the interrupt works, but the status bit is permanently asserted */ + if (rdev->irq.gui_idle && radeon_gui_idle(rdev)) { + if (!rdev->irq.gui_idle_acked) + irq_mask |= RADEON_GUI_IDLE_STAT; + } + + if (irqs) { + WREG32(RADEON_GEN_INT_STATUS, irqs); + } + return irqs & irq_mask; +} + +int r100_irq_process(struct radeon_device *rdev) +{ + uint32_t status, msi_rearm; + bool queue_hotplug = false; + + /* reset gui idle ack. the status bit is broken */ + rdev->irq.gui_idle_acked = false; + + status = r100_irq_ack(rdev); + if (!status) { + return IRQ_NONE; + } + if (rdev->shutdown) { + return IRQ_NONE; + } + while (status) { + /* SW interrupt */ + if (status & RADEON_SW_INT_TEST) { + radeon_fence_process(rdev); + } + /* gui idle interrupt */ + if (status & RADEON_GUI_IDLE_STAT) { + rdev->irq.gui_idle_acked = true; + rdev->pm.gui_idle = true; + wake_up(&rdev->irq.idle_queue); + } + /* Vertical blank interrupts */ + if (status & RADEON_CRTC_VBLANK_STAT) { + if (rdev->irq.crtc_vblank_int[0]) { + drm_handle_vblank(rdev->ddev, 0); + rdev->pm.vblank_sync = true; + wake_up(&rdev->irq.vblank_queue); + } + if (rdev->irq.pflip[0]) + radeon_crtc_handle_flip(rdev, 0); + } + if (status & RADEON_CRTC2_VBLANK_STAT) { + if (rdev->irq.crtc_vblank_int[1]) { + drm_handle_vblank(rdev->ddev, 1); + rdev->pm.vblank_sync = true; + wake_up(&rdev->irq.vblank_queue); + } + if (rdev->irq.pflip[1]) + radeon_crtc_handle_flip(rdev, 1); + } + if (status & RADEON_FP_DETECT_STAT) { + queue_hotplug = true; + DRM_DEBUG("HPD1\n"); + } + if (status & RADEON_FP2_DETECT_STAT) { + queue_hotplug = true; + DRM_DEBUG("HPD2\n"); + } + status = r100_irq_ack(rdev); + } + /* reset gui idle ack. the status bit is broken */ + rdev->irq.gui_idle_acked = false; + if (queue_hotplug) + schedule_work(&rdev->hotplug_work); + if (rdev->msi_enabled) { + switch (rdev->family) { + case CHIP_RS400: + case CHIP_RS480: + msi_rearm = RREG32(RADEON_AIC_CNTL) & ~RS400_MSI_REARM; + WREG32(RADEON_AIC_CNTL, msi_rearm); + WREG32(RADEON_AIC_CNTL, msi_rearm | RS400_MSI_REARM); + break; + default: + WREG32(RADEON_MSI_REARM_EN, RV370_MSI_REARM_EN); + break; + } + } + return IRQ_HANDLED; +} + +u32 r100_get_vblank_counter(struct radeon_device *rdev, int crtc) +{ + if (crtc == 0) + return RREG32(RADEON_CRTC_CRNT_FRAME); + else + return RREG32(RADEON_CRTC2_CRNT_FRAME); +} + +/* Who ever call radeon_fence_emit should call ring_lock and ask + * for enough space (today caller are ib schedule and buffer move) */ +void r100_fence_ring_emit(struct radeon_device *rdev, + struct radeon_fence *fence) +{ + /* We have to make sure that caches are flushed before + * CPU might read something from VRAM. */ + radeon_ring_write(rdev, PACKET0(RADEON_RB3D_DSTCACHE_CTLSTAT, 0)); + radeon_ring_write(rdev, RADEON_RB3D_DC_FLUSH_ALL); + radeon_ring_write(rdev, PACKET0(RADEON_RB3D_ZCACHE_CTLSTAT, 0)); + radeon_ring_write(rdev, RADEON_RB3D_ZC_FLUSH_ALL); + /* Wait until IDLE & CLEAN */ + radeon_ring_write(rdev, PACKET0(RADEON_WAIT_UNTIL, 0)); + radeon_ring_write(rdev, RADEON_WAIT_2D_IDLECLEAN | RADEON_WAIT_3D_IDLECLEAN); + radeon_ring_write(rdev, PACKET0(RADEON_HOST_PATH_CNTL, 0)); + radeon_ring_write(rdev, rdev->config.r100.hdp_cntl | + RADEON_HDP_READ_BUFFER_INVALIDATE); + radeon_ring_write(rdev, PACKET0(RADEON_HOST_PATH_CNTL, 0)); + radeon_ring_write(rdev, rdev->config.r100.hdp_cntl); + /* Emit fence sequence & fire IRQ */ + radeon_ring_write(rdev, PACKET0(rdev->fence_drv.scratch_reg, 0)); + radeon_ring_write(rdev, fence->seq); + radeon_ring_write(rdev, PACKET0(RADEON_GEN_INT_STATUS, 0)); + radeon_ring_write(rdev, RADEON_SW_INT_FIRE); +} + +int r100_copy_blit(struct radeon_device *rdev, + uint64_t src_offset, + uint64_t dst_offset, + unsigned num_gpu_pages, + struct radeon_fence *fence) +{ + uint32_t cur_pages; + uint32_t stride_bytes = RADEON_GPU_PAGE_SIZE; + uint32_t pitch; + uint32_t stride_pixels; + unsigned ndw; + int num_loops; + int r = 0; + + /* radeon limited to 16k stride */ + stride_bytes &= 0x3fff; + /* radeon pitch is /64 */ + pitch = stride_bytes / 64; + stride_pixels = stride_bytes / 4; + num_loops = DIV_ROUND_UP(num_gpu_pages, 8191); + + /* Ask for enough room for blit + flush + fence */ + ndw = 64 + (10 * num_loops); + r = radeon_ring_lock(rdev, ndw); + if (r) { + DRM_ERROR("radeon: moving bo (%d) asking for %u dw.\n", r, ndw); + return -EINVAL; + } + while (num_gpu_pages > 0) { + cur_pages = num_gpu_pages; + if (cur_pages > 8191) { + cur_pages = 8191; + } + num_gpu_pages -= cur_pages; + + /* pages are in Y direction - height + page width in X direction - width */ + radeon_ring_write(rdev, PACKET3(PACKET3_BITBLT_MULTI, 8)); + radeon_ring_write(rdev, + RADEON_GMC_SRC_PITCH_OFFSET_CNTL | + RADEON_GMC_DST_PITCH_OFFSET_CNTL | + RADEON_GMC_SRC_CLIPPING | + RADEON_GMC_DST_CLIPPING | + RADEON_GMC_BRUSH_NONE | + (RADEON_COLOR_FORMAT_ARGB8888 << 8) | + RADEON_GMC_SRC_DATATYPE_COLOR | + RADEON_ROP3_S | + RADEON_DP_SRC_SOURCE_MEMORY | + RADEON_GMC_CLR_CMP_CNTL_DIS | + RADEON_GMC_WR_MSK_DIS); + radeon_ring_write(rdev, (pitch << 22) | (src_offset >> 10)); + radeon_ring_write(rdev, (pitch << 22) | (dst_offset >> 10)); + radeon_ring_write(rdev, (0x1fff) | (0x1fff << 16)); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, (0x1fff) | (0x1fff << 16)); + radeon_ring_write(rdev, num_gpu_pages); + radeon_ring_write(rdev, num_gpu_pages); + radeon_ring_write(rdev, cur_pages | (stride_pixels << 16)); + } + radeon_ring_write(rdev, PACKET0(RADEON_DSTCACHE_CTLSTAT, 0)); + radeon_ring_write(rdev, RADEON_RB2D_DC_FLUSH_ALL); + radeon_ring_write(rdev, PACKET0(RADEON_WAIT_UNTIL, 0)); + radeon_ring_write(rdev, + RADEON_WAIT_2D_IDLECLEAN | + RADEON_WAIT_HOST_IDLECLEAN | + RADEON_WAIT_DMA_GUI_IDLE); + if (fence) { + r = radeon_fence_emit(rdev, fence); + } + radeon_ring_unlock_commit(rdev); + return r; +} + +static int r100_cp_wait_for_idle(struct radeon_device *rdev) +{ + unsigned i; + u32 tmp; + + for (i = 0; i < rdev->usec_timeout; i++) { + tmp = RREG32(R_000E40_RBBM_STATUS); + if (!G_000E40_CP_CMDSTRM_BUSY(tmp)) { + return 0; + } + udelay(1); + } + return -1; +} + +void r100_ring_start(struct radeon_device *rdev) +{ + int r; + + r = radeon_ring_lock(rdev, 2); + if (r) { + return; + } + radeon_ring_write(rdev, PACKET0(RADEON_ISYNC_CNTL, 0)); + radeon_ring_write(rdev, + RADEON_ISYNC_ANY2D_IDLE3D | + RADEON_ISYNC_ANY3D_IDLE2D | + RADEON_ISYNC_WAIT_IDLEGUI | + RADEON_ISYNC_CPSCRATCH_IDLEGUI); + radeon_ring_unlock_commit(rdev); +} + + +/* Load the microcode for the CP */ +static int r100_cp_init_microcode(struct radeon_device *rdev) +{ + struct platform_device *pdev; + const char *fw_name = NULL; + int err; + + DRM_DEBUG_KMS("\n"); + + pdev = platform_device_register_simple("radeon_cp", 0, NULL, 0); + err = IS_ERR(pdev); + if (err) { + printk(KERN_ERR "radeon_cp: Failed to register firmware\n"); + return -EINVAL; + } + if ((rdev->family == CHIP_R100) || (rdev->family == CHIP_RV100) || + (rdev->family == CHIP_RV200) || (rdev->family == CHIP_RS100) || + (rdev->family == CHIP_RS200)) { + DRM_INFO("Loading R100 Microcode\n"); + fw_name = FIRMWARE_R100; + } else if ((rdev->family == CHIP_R200) || + (rdev->family == CHIP_RV250) || + (rdev->family == CHIP_RV280) || + (rdev->family == CHIP_RS300)) { + DRM_INFO("Loading R200 Microcode\n"); + fw_name = FIRMWARE_R200; + } else if ((rdev->family == CHIP_R300) || + (rdev->family == CHIP_R350) || + (rdev->family == CHIP_RV350) || + (rdev->family == CHIP_RV380) || + (rdev->family == CHIP_RS400) || + (rdev->family == CHIP_RS480)) { + DRM_INFO("Loading R300 Microcode\n"); + fw_name = FIRMWARE_R300; + } else if ((rdev->family == CHIP_R420) || + (rdev->family == CHIP_R423) || + (rdev->family == CHIP_RV410)) { + DRM_INFO("Loading R400 Microcode\n"); + fw_name = FIRMWARE_R420; + } else if ((rdev->family == CHIP_RS690) || + (rdev->family == CHIP_RS740)) { + DRM_INFO("Loading RS690/RS740 Microcode\n"); + fw_name = FIRMWARE_RS690; + } else if (rdev->family == CHIP_RS600) { + DRM_INFO("Loading RS600 Microcode\n"); + fw_name = FIRMWARE_RS600; + } else if ((rdev->family == CHIP_RV515) || + (rdev->family == CHIP_R520) || + (rdev->family == CHIP_RV530) || + (rdev->family == CHIP_R580) || + (rdev->family == CHIP_RV560) || + (rdev->family == CHIP_RV570)) { + DRM_INFO("Loading R500 Microcode\n"); + fw_name = FIRMWARE_R520; + } + + err = request_firmware(&rdev->me_fw, fw_name, &pdev->dev); + platform_device_unregister(pdev); + if (err) { + printk(KERN_ERR "radeon_cp: Failed to load firmware \"%s\"\n", + fw_name); + } else if (rdev->me_fw->size % 8) { + printk(KERN_ERR + "radeon_cp: Bogus length %zu in firmware \"%s\"\n", + rdev->me_fw->size, fw_name); + err = -EINVAL; + release_firmware(rdev->me_fw); + rdev->me_fw = NULL; + } + return err; +} + +static void r100_cp_load_microcode(struct radeon_device *rdev) +{ + const __be32 *fw_data; + int i, size; + + if (r100_gui_wait_for_idle(rdev)) { + printk(KERN_WARNING "Failed to wait GUI idle while " + "programming pipes. Bad things might happen.\n"); + } + + if (rdev->me_fw) { + size = rdev->me_fw->size / 4; + fw_data = (const __be32 *)&rdev->me_fw->data[0]; + WREG32(RADEON_CP_ME_RAM_ADDR, 0); + for (i = 0; i < size; i += 2) { + WREG32(RADEON_CP_ME_RAM_DATAH, + be32_to_cpup(&fw_data[i])); + WREG32(RADEON_CP_ME_RAM_DATAL, + be32_to_cpup(&fw_data[i + 1])); + } + } +} + +int r100_cp_init(struct radeon_device *rdev, unsigned ring_size) +{ + unsigned rb_bufsz; + unsigned rb_blksz; + unsigned max_fetch; + unsigned pre_write_timer; + unsigned pre_write_limit; + unsigned indirect2_start; + unsigned indirect1_start; + uint32_t tmp; + int r; + + if (r100_debugfs_cp_init(rdev)) { + DRM_ERROR("Failed to register debugfs file for CP !\n"); + } + if (!rdev->me_fw) { + r = r100_cp_init_microcode(rdev); + if (r) { + DRM_ERROR("Failed to load firmware!\n"); + return r; + } + } + + /* Align ring size */ + rb_bufsz = drm_order(ring_size / 8); + ring_size = (1 << (rb_bufsz + 1)) * 4; + r100_cp_load_microcode(rdev); + r = radeon_ring_init(rdev, ring_size); + if (r) { + return r; + } + /* Each time the cp read 1024 bytes (16 dword/quadword) update + * the rptr copy in system ram */ + rb_blksz = 9; + /* cp will read 128bytes at a time (4 dwords) */ + max_fetch = 1; + rdev->cp.align_mask = 16 - 1; + /* Write to CP_RB_WPTR will be delayed for pre_write_timer clocks */ + pre_write_timer = 64; + /* Force CP_RB_WPTR write if written more than one time before the + * delay expire + */ + pre_write_limit = 0; + /* Setup the cp cache like this (cache size is 96 dwords) : + * RING 0 to 15 + * INDIRECT1 16 to 79 + * INDIRECT2 80 to 95 + * So ring cache size is 16dwords (> (2 * max_fetch = 2 * 4dwords)) + * indirect1 cache size is 64dwords (> (2 * max_fetch = 2 * 4dwords)) + * indirect2 cache size is 16dwords (> (2 * max_fetch = 2 * 4dwords)) + * Idea being that most of the gpu cmd will be through indirect1 buffer + * so it gets the bigger cache. + */ + indirect2_start = 80; + indirect1_start = 16; + /* cp setup */ + WREG32(0x718, pre_write_timer | (pre_write_limit << 28)); + tmp = (REG_SET(RADEON_RB_BUFSZ, rb_bufsz) | + REG_SET(RADEON_RB_BLKSZ, rb_blksz) | + REG_SET(RADEON_MAX_FETCH, max_fetch)); +#ifdef __BIG_ENDIAN + tmp |= RADEON_BUF_SWAP_32BIT; +#endif + WREG32(RADEON_CP_RB_CNTL, tmp | RADEON_RB_NO_UPDATE); + + /* Set ring address */ + DRM_INFO("radeon: ring at 0x%016lX\n", (unsigned long)rdev->cp.gpu_addr); + WREG32(RADEON_CP_RB_BASE, rdev->cp.gpu_addr); + /* Force read & write ptr to 0 */ + WREG32(RADEON_CP_RB_CNTL, tmp | RADEON_RB_RPTR_WR_ENA | RADEON_RB_NO_UPDATE); + WREG32(RADEON_CP_RB_RPTR_WR, 0); + rdev->cp.wptr = 0; + WREG32(RADEON_CP_RB_WPTR, rdev->cp.wptr); + + /* set the wb address whether it's enabled or not */ + WREG32(R_00070C_CP_RB_RPTR_ADDR, + S_00070C_RB_RPTR_ADDR((rdev->wb.gpu_addr + RADEON_WB_CP_RPTR_OFFSET) >> 2)); + WREG32(R_000774_SCRATCH_ADDR, rdev->wb.gpu_addr + RADEON_WB_SCRATCH_OFFSET); + + if (rdev->wb.enabled) + WREG32(R_000770_SCRATCH_UMSK, 0xff); + else { + tmp |= RADEON_RB_NO_UPDATE; + WREG32(R_000770_SCRATCH_UMSK, 0); + } + + WREG32(RADEON_CP_RB_CNTL, tmp); + udelay(10); + rdev->cp.rptr = RREG32(RADEON_CP_RB_RPTR); + /* Set cp mode to bus mastering & enable cp*/ + WREG32(RADEON_CP_CSQ_MODE, + REG_SET(RADEON_INDIRECT2_START, indirect2_start) | + REG_SET(RADEON_INDIRECT1_START, indirect1_start)); + WREG32(RADEON_CP_RB_WPTR_DELAY, 0); + WREG32(RADEON_CP_CSQ_MODE, 0x00004D4D); + WREG32(RADEON_CP_CSQ_CNTL, RADEON_CSQ_PRIBM_INDBM); + radeon_ring_start(rdev); + r = radeon_ring_test(rdev); + if (r) { + DRM_ERROR("radeon: cp isn't working (%d).\n", r); + return r; + } + rdev->cp.ready = true; + radeon_ttm_set_active_vram_size(rdev, rdev->mc.real_vram_size); + return 0; +} + +void r100_cp_fini(struct radeon_device *rdev) +{ + if (r100_cp_wait_for_idle(rdev)) { + DRM_ERROR("Wait for CP idle timeout, shutting down CP.\n"); + } + /* Disable ring */ + r100_cp_disable(rdev); + radeon_ring_fini(rdev); + DRM_INFO("radeon: cp finalized\n"); +} + +void r100_cp_disable(struct radeon_device *rdev) +{ + /* Disable ring */ + radeon_ttm_set_active_vram_size(rdev, rdev->mc.visible_vram_size); + rdev->cp.ready = false; + WREG32(RADEON_CP_CSQ_MODE, 0); + WREG32(RADEON_CP_CSQ_CNTL, 0); + WREG32(R_000770_SCRATCH_UMSK, 0); + if (r100_gui_wait_for_idle(rdev)) { + printk(KERN_WARNING "Failed to wait GUI idle while " + "programming pipes. Bad things might happen.\n"); + } +} + +void r100_cp_commit(struct radeon_device *rdev) +{ + WREG32(RADEON_CP_RB_WPTR, rdev->cp.wptr); + (void)RREG32(RADEON_CP_RB_WPTR); +} + + +/* + * CS functions + */ +int r100_cs_parse_packet0(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt, + const unsigned *auth, unsigned n, + radeon_packet0_check_t check) +{ + unsigned reg; + unsigned i, j, m; + unsigned idx; + int r; + + idx = pkt->idx + 1; + reg = pkt->reg; + /* Check that register fall into register range + * determined by the number of entry (n) in the + * safe register bitmap. + */ + if (pkt->one_reg_wr) { + if ((reg >> 7) > n) { + return -EINVAL; + } + } else { + if (((reg + (pkt->count << 2)) >> 7) > n) { + return -EINVAL; + } + } + for (i = 0; i <= pkt->count; i++, idx++) { + j = (reg >> 7); + m = 1 << ((reg >> 2) & 31); + if (auth[j] & m) { + r = check(p, pkt, idx, reg); + if (r) { + return r; + } + } + if (pkt->one_reg_wr) { + if (!(auth[j] & m)) { + break; + } + } else { + reg += 4; + } + } + return 0; +} + +void r100_cs_dump_packet(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt) +{ + volatile uint32_t *ib; + unsigned i; + unsigned idx; + + ib = p->ib->ptr; + idx = pkt->idx; + for (i = 0; i <= (pkt->count + 1); i++, idx++) { + DRM_INFO("ib[%d]=0x%08X\n", idx, ib[idx]); + } +} + +/** + * r100_cs_packet_parse() - parse cp packet and point ib index to next packet + * @parser: parser structure holding parsing context. + * @pkt: where to store packet informations + * + * Assume that chunk_ib_index is properly set. Will return -EINVAL + * if packet is bigger than remaining ib size. or if packets is unknown. + **/ +int r100_cs_packet_parse(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt, + unsigned idx) +{ + struct radeon_cs_chunk *ib_chunk = &p->chunks[p->chunk_ib_idx]; + uint32_t header; + + if (idx >= ib_chunk->length_dw) { + DRM_ERROR("Can not parse packet at %d after CS end %d !\n", + idx, ib_chunk->length_dw); + return -EINVAL; + } + header = radeon_get_ib_value(p, idx); + pkt->idx = idx; + pkt->type = CP_PACKET_GET_TYPE(header); + pkt->count = CP_PACKET_GET_COUNT(header); + switch (pkt->type) { + case PACKET_TYPE0: + pkt->reg = CP_PACKET0_GET_REG(header); + pkt->one_reg_wr = CP_PACKET0_GET_ONE_REG_WR(header); + break; + case PACKET_TYPE3: + pkt->opcode = CP_PACKET3_GET_OPCODE(header); + break; + case PACKET_TYPE2: + pkt->count = -1; + break; + default: + DRM_ERROR("Unknown packet type %d at %d !\n", pkt->type, idx); + return -EINVAL; + } + if ((pkt->count + 1 + pkt->idx) >= ib_chunk->length_dw) { + DRM_ERROR("Packet (%d:%d:%d) end after CS buffer (%d) !\n", + pkt->idx, pkt->type, pkt->count, ib_chunk->length_dw); + return -EINVAL; + } + return 0; +} + +/** + * r100_cs_packet_next_vline() - parse userspace VLINE packet + * @parser: parser structure holding parsing context. + * + * Userspace sends a special sequence for VLINE waits. + * PACKET0 - VLINE_START_END + value + * PACKET0 - WAIT_UNTIL +_value + * RELOC (P3) - crtc_id in reloc. + * + * This function parses this and relocates the VLINE START END + * and WAIT UNTIL packets to the correct crtc. + * It also detects a switched off crtc and nulls out the + * wait in that case. + */ +int r100_cs_packet_parse_vline(struct radeon_cs_parser *p) +{ + struct drm_mode_object *obj; + struct drm_crtc *crtc; + struct radeon_crtc *radeon_crtc; + struct radeon_cs_packet p3reloc, waitreloc; + int crtc_id; + int r; + uint32_t header, h_idx, reg; + volatile uint32_t *ib; + + ib = p->ib->ptr; + + /* parse the wait until */ + r = r100_cs_packet_parse(p, &waitreloc, p->idx); + if (r) + return r; + + /* check its a wait until and only 1 count */ + if (waitreloc.reg != RADEON_WAIT_UNTIL || + waitreloc.count != 0) { + DRM_ERROR("vline wait had illegal wait until segment\n"); + return -EINVAL; + } + + if (radeon_get_ib_value(p, waitreloc.idx + 1) != RADEON_WAIT_CRTC_VLINE) { + DRM_ERROR("vline wait had illegal wait until\n"); + return -EINVAL; + } + + /* jump over the NOP */ + r = r100_cs_packet_parse(p, &p3reloc, p->idx + waitreloc.count + 2); + if (r) + return r; + + h_idx = p->idx - 2; + p->idx += waitreloc.count + 2; + p->idx += p3reloc.count + 2; + + header = radeon_get_ib_value(p, h_idx); + crtc_id = radeon_get_ib_value(p, h_idx + 5); + reg = CP_PACKET0_GET_REG(header); + obj = drm_mode_object_find(p->rdev->ddev, crtc_id, DRM_MODE_OBJECT_CRTC); + if (!obj) { + DRM_ERROR("cannot find crtc %d\n", crtc_id); + return -EINVAL; + } + crtc = obj_to_crtc(obj); + radeon_crtc = to_radeon_crtc(crtc); + crtc_id = radeon_crtc->crtc_id; + + if (!crtc->enabled) { + /* if the CRTC isn't enabled - we need to nop out the wait until */ + ib[h_idx + 2] = PACKET2(0); + ib[h_idx + 3] = PACKET2(0); + } else if (crtc_id == 1) { + switch (reg) { + case AVIVO_D1MODE_VLINE_START_END: + header &= ~R300_CP_PACKET0_REG_MASK; + header |= AVIVO_D2MODE_VLINE_START_END >> 2; + break; + case RADEON_CRTC_GUI_TRIG_VLINE: + header &= ~R300_CP_PACKET0_REG_MASK; + header |= RADEON_CRTC2_GUI_TRIG_VLINE >> 2; + break; + default: + DRM_ERROR("unknown crtc reloc\n"); + return -EINVAL; + } + ib[h_idx] = header; + ib[h_idx + 3] |= RADEON_ENG_DISPLAY_SELECT_CRTC1; + } + + return 0; +} + +/** + * r100_cs_packet_next_reloc() - parse next packet which should be reloc packet3 + * @parser: parser structure holding parsing context. + * @data: pointer to relocation data + * @offset_start: starting offset + * @offset_mask: offset mask (to align start offset on) + * @reloc: reloc informations + * + * Check next packet is relocation packet3, do bo validation and compute + * GPU offset using the provided start. + **/ +int r100_cs_packet_next_reloc(struct radeon_cs_parser *p, + struct radeon_cs_reloc **cs_reloc) +{ + struct radeon_cs_chunk *relocs_chunk; + struct radeon_cs_packet p3reloc; + unsigned idx; + int r; + + if (p->chunk_relocs_idx == -1) { + DRM_ERROR("No relocation chunk !\n"); + return -EINVAL; + } + *cs_reloc = NULL; + relocs_chunk = &p->chunks[p->chunk_relocs_idx]; + r = r100_cs_packet_parse(p, &p3reloc, p->idx); + if (r) { + return r; + } + p->idx += p3reloc.count + 2; + if (p3reloc.type != PACKET_TYPE3 || p3reloc.opcode != PACKET3_NOP) { + DRM_ERROR("No packet3 for relocation for packet at %d.\n", + p3reloc.idx); + r100_cs_dump_packet(p, &p3reloc); + return -EINVAL; + } + idx = radeon_get_ib_value(p, p3reloc.idx + 1); + if (idx >= relocs_chunk->length_dw) { + DRM_ERROR("Relocs at %d after relocations chunk end %d !\n", + idx, relocs_chunk->length_dw); + r100_cs_dump_packet(p, &p3reloc); + return -EINVAL; + } + /* FIXME: we assume reloc size is 4 dwords */ + *cs_reloc = p->relocs_ptr[(idx / 4)]; + return 0; +} + +static int r100_get_vtx_size(uint32_t vtx_fmt) +{ + int vtx_size; + vtx_size = 2; + /* ordered according to bits in spec */ + if (vtx_fmt & RADEON_SE_VTX_FMT_W0) + vtx_size++; + if (vtx_fmt & RADEON_SE_VTX_FMT_FPCOLOR) + vtx_size += 3; + if (vtx_fmt & RADEON_SE_VTX_FMT_FPALPHA) + vtx_size++; + if (vtx_fmt & RADEON_SE_VTX_FMT_PKCOLOR) + vtx_size++; + if (vtx_fmt & RADEON_SE_VTX_FMT_FPSPEC) + vtx_size += 3; + if (vtx_fmt & RADEON_SE_VTX_FMT_FPFOG) + vtx_size++; + if (vtx_fmt & RADEON_SE_VTX_FMT_PKSPEC) + vtx_size++; + if (vtx_fmt & RADEON_SE_VTX_FMT_ST0) + vtx_size += 2; + if (vtx_fmt & RADEON_SE_VTX_FMT_ST1) + vtx_size += 2; + if (vtx_fmt & RADEON_SE_VTX_FMT_Q1) + vtx_size++; + if (vtx_fmt & RADEON_SE_VTX_FMT_ST2) + vtx_size += 2; + if (vtx_fmt & RADEON_SE_VTX_FMT_Q2) + vtx_size++; + if (vtx_fmt & RADEON_SE_VTX_FMT_ST3) + vtx_size += 2; + if (vtx_fmt & RADEON_SE_VTX_FMT_Q3) + vtx_size++; + if (vtx_fmt & RADEON_SE_VTX_FMT_Q0) + vtx_size++; + /* blend weight */ + if (vtx_fmt & (0x7 << 15)) + vtx_size += (vtx_fmt >> 15) & 0x7; + if (vtx_fmt & RADEON_SE_VTX_FMT_N0) + vtx_size += 3; + if (vtx_fmt & RADEON_SE_VTX_FMT_XY1) + vtx_size += 2; + if (vtx_fmt & RADEON_SE_VTX_FMT_Z1) + vtx_size++; + if (vtx_fmt & RADEON_SE_VTX_FMT_W1) + vtx_size++; + if (vtx_fmt & RADEON_SE_VTX_FMT_N1) + vtx_size++; + if (vtx_fmt & RADEON_SE_VTX_FMT_Z) + vtx_size++; + return vtx_size; +} + +static int r100_packet0_check(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt, + unsigned idx, unsigned reg) +{ + struct radeon_cs_reloc *reloc; + struct r100_cs_track *track; + volatile uint32_t *ib; + uint32_t tmp; + int r; + int i, face; + u32 tile_flags = 0; + u32 idx_value; + + ib = p->ib->ptr; + track = (struct r100_cs_track *)p->track; + + idx_value = radeon_get_ib_value(p, idx); + + switch (reg) { + case RADEON_CRTC_GUI_TRIG_VLINE: + r = r100_cs_packet_parse_vline(p); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + break; + /* FIXME: only allow PACKET3 blit? easier to check for out of + * range access */ + case RADEON_DST_PITCH_OFFSET: + case RADEON_SRC_PITCH_OFFSET: + r = r100_reloc_pitch_offset(p, pkt, idx, reg); + if (r) + return r; + break; + case RADEON_RB3D_DEPTHOFFSET: + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + track->zb.robj = reloc->robj; + track->zb.offset = idx_value; + track->zb_dirty = true; + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + break; + case RADEON_RB3D_COLOROFFSET: + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + track->cb[0].robj = reloc->robj; + track->cb[0].offset = idx_value; + track->cb_dirty = true; + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + break; + case RADEON_PP_TXOFFSET_0: + case RADEON_PP_TXOFFSET_1: + case RADEON_PP_TXOFFSET_2: + i = (reg - RADEON_PP_TXOFFSET_0) / 24; + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + track->textures[i].robj = reloc->robj; + track->tex_dirty = true; + break; + case RADEON_PP_CUBIC_OFFSET_T0_0: + case RADEON_PP_CUBIC_OFFSET_T0_1: + case RADEON_PP_CUBIC_OFFSET_T0_2: + case RADEON_PP_CUBIC_OFFSET_T0_3: + case RADEON_PP_CUBIC_OFFSET_T0_4: + i = (reg - RADEON_PP_CUBIC_OFFSET_T0_0) / 4; + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + track->textures[0].cube_info[i].offset = idx_value; + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + track->textures[0].cube_info[i].robj = reloc->robj; + track->tex_dirty = true; + break; + case RADEON_PP_CUBIC_OFFSET_T1_0: + case RADEON_PP_CUBIC_OFFSET_T1_1: + case RADEON_PP_CUBIC_OFFSET_T1_2: + case RADEON_PP_CUBIC_OFFSET_T1_3: + case RADEON_PP_CUBIC_OFFSET_T1_4: + i = (reg - RADEON_PP_CUBIC_OFFSET_T1_0) / 4; + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + track->textures[1].cube_info[i].offset = idx_value; + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + track->textures[1].cube_info[i].robj = reloc->robj; + track->tex_dirty = true; + break; + case RADEON_PP_CUBIC_OFFSET_T2_0: + case RADEON_PP_CUBIC_OFFSET_T2_1: + case RADEON_PP_CUBIC_OFFSET_T2_2: + case RADEON_PP_CUBIC_OFFSET_T2_3: + case RADEON_PP_CUBIC_OFFSET_T2_4: + i = (reg - RADEON_PP_CUBIC_OFFSET_T2_0) / 4; + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + track->textures[2].cube_info[i].offset = idx_value; + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + track->textures[2].cube_info[i].robj = reloc->robj; + track->tex_dirty = true; + break; + case RADEON_RE_WIDTH_HEIGHT: + track->maxy = ((idx_value >> 16) & 0x7FF); + track->cb_dirty = true; + track->zb_dirty = true; + break; + case RADEON_RB3D_COLORPITCH: + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + + if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) + tile_flags |= RADEON_COLOR_TILE_ENABLE; + if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) + tile_flags |= RADEON_COLOR_MICROTILE_ENABLE; + + tmp = idx_value & ~(0x7 << 16); + tmp |= tile_flags; + ib[idx] = tmp; + + track->cb[0].pitch = idx_value & RADEON_COLORPITCH_MASK; + track->cb_dirty = true; + break; + case RADEON_RB3D_DEPTHPITCH: + track->zb.pitch = idx_value & RADEON_DEPTHPITCH_MASK; + track->zb_dirty = true; + break; + case RADEON_RB3D_CNTL: + switch ((idx_value >> RADEON_RB3D_COLOR_FORMAT_SHIFT) & 0x1f) { + case 7: + case 8: + case 9: + case 11: + case 12: + track->cb[0].cpp = 1; + break; + case 3: + case 4: + case 15: + track->cb[0].cpp = 2; + break; + case 6: + track->cb[0].cpp = 4; + break; + default: + DRM_ERROR("Invalid color buffer format (%d) !\n", + ((idx_value >> RADEON_RB3D_COLOR_FORMAT_SHIFT) & 0x1f)); + return -EINVAL; + } + track->z_enabled = !!(idx_value & RADEON_Z_ENABLE); + track->cb_dirty = true; + track->zb_dirty = true; + break; + case RADEON_RB3D_ZSTENCILCNTL: + switch (idx_value & 0xf) { + case 0: + track->zb.cpp = 2; + break; + case 2: + case 3: + case 4: + case 5: + case 9: + case 11: + track->zb.cpp = 4; + break; + default: + break; + } + track->zb_dirty = true; + break; + case RADEON_RB3D_ZPASS_ADDR: + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + break; + case RADEON_PP_CNTL: + { + uint32_t temp = idx_value >> 4; + for (i = 0; i < track->num_texture; i++) + track->textures[i].enabled = !!(temp & (1 << i)); + track->tex_dirty = true; + } + break; + case RADEON_SE_VF_CNTL: + track->vap_vf_cntl = idx_value; + break; + case RADEON_SE_VTX_FMT: + track->vtx_size = r100_get_vtx_size(idx_value); + break; + case RADEON_PP_TEX_SIZE_0: + case RADEON_PP_TEX_SIZE_1: + case RADEON_PP_TEX_SIZE_2: + i = (reg - RADEON_PP_TEX_SIZE_0) / 8; + track->textures[i].width = (idx_value & RADEON_TEX_USIZE_MASK) + 1; + track->textures[i].height = ((idx_value & RADEON_TEX_VSIZE_MASK) >> RADEON_TEX_VSIZE_SHIFT) + 1; + track->tex_dirty = true; + break; + case RADEON_PP_TEX_PITCH_0: + case RADEON_PP_TEX_PITCH_1: + case RADEON_PP_TEX_PITCH_2: + i = (reg - RADEON_PP_TEX_PITCH_0) / 8; + track->textures[i].pitch = idx_value + 32; + track->tex_dirty = true; + break; + case RADEON_PP_TXFILTER_0: + case RADEON_PP_TXFILTER_1: + case RADEON_PP_TXFILTER_2: + i = (reg - RADEON_PP_TXFILTER_0) / 24; + track->textures[i].num_levels = ((idx_value & RADEON_MAX_MIP_LEVEL_MASK) + >> RADEON_MAX_MIP_LEVEL_SHIFT); + tmp = (idx_value >> 23) & 0x7; + if (tmp == 2 || tmp == 6) + track->textures[i].roundup_w = false; + tmp = (idx_value >> 27) & 0x7; + if (tmp == 2 || tmp == 6) + track->textures[i].roundup_h = false; + track->tex_dirty = true; + break; + case RADEON_PP_TXFORMAT_0: + case RADEON_PP_TXFORMAT_1: + case RADEON_PP_TXFORMAT_2: + i = (reg - RADEON_PP_TXFORMAT_0) / 24; + if (idx_value & RADEON_TXFORMAT_NON_POWER2) { + track->textures[i].use_pitch = 1; + } else { + track->textures[i].use_pitch = 0; + track->textures[i].width = 1 << ((idx_value >> RADEON_TXFORMAT_WIDTH_SHIFT) & RADEON_TXFORMAT_WIDTH_MASK); + track->textures[i].height = 1 << ((idx_value >> RADEON_TXFORMAT_HEIGHT_SHIFT) & RADEON_TXFORMAT_HEIGHT_MASK); + } + if (idx_value & RADEON_TXFORMAT_CUBIC_MAP_ENABLE) + track->textures[i].tex_coord_type = 2; + switch ((idx_value & RADEON_TXFORMAT_FORMAT_MASK)) { + case RADEON_TXFORMAT_I8: + case RADEON_TXFORMAT_RGB332: + case RADEON_TXFORMAT_Y8: + track->textures[i].cpp = 1; + track->textures[i].compress_format = R100_TRACK_COMP_NONE; + break; + case RADEON_TXFORMAT_AI88: + case RADEON_TXFORMAT_ARGB1555: + case RADEON_TXFORMAT_RGB565: + case RADEON_TXFORMAT_ARGB4444: + case RADEON_TXFORMAT_VYUY422: + case RADEON_TXFORMAT_YVYU422: + case RADEON_TXFORMAT_SHADOW16: + case RADEON_TXFORMAT_LDUDV655: + case RADEON_TXFORMAT_DUDV88: + track->textures[i].cpp = 2; + track->textures[i].compress_format = R100_TRACK_COMP_NONE; + break; + case RADEON_TXFORMAT_ARGB8888: + case RADEON_TXFORMAT_RGBA8888: + case RADEON_TXFORMAT_SHADOW32: + case RADEON_TXFORMAT_LDUDUV8888: + track->textures[i].cpp = 4; + track->textures[i].compress_format = R100_TRACK_COMP_NONE; + break; + case RADEON_TXFORMAT_DXT1: + track->textures[i].cpp = 1; + track->textures[i].compress_format = R100_TRACK_COMP_DXT1; + break; + case RADEON_TXFORMAT_DXT23: + case RADEON_TXFORMAT_DXT45: + track->textures[i].cpp = 1; + track->textures[i].compress_format = R100_TRACK_COMP_DXT35; + break; + } + track->textures[i].cube_info[4].width = 1 << ((idx_value >> 16) & 0xf); + track->textures[i].cube_info[4].height = 1 << ((idx_value >> 20) & 0xf); + track->tex_dirty = true; + break; + case RADEON_PP_CUBIC_FACES_0: + case RADEON_PP_CUBIC_FACES_1: + case RADEON_PP_CUBIC_FACES_2: + tmp = idx_value; + i = (reg - RADEON_PP_CUBIC_FACES_0) / 4; + for (face = 0; face < 4; face++) { + track->textures[i].cube_info[face].width = 1 << ((tmp >> (face * 8)) & 0xf); + track->textures[i].cube_info[face].height = 1 << ((tmp >> ((face * 8) + 4)) & 0xf); + } + track->tex_dirty = true; + break; + default: + printk(KERN_ERR "Forbidden register 0x%04X in cs at %d\n", + reg, idx); + return -EINVAL; + } + return 0; +} + +int r100_cs_track_check_pkt3_indx_buffer(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt, + struct radeon_bo *robj) +{ + unsigned idx; + u32 value; + idx = pkt->idx + 1; + value = radeon_get_ib_value(p, idx + 2); + if ((value + 1) > radeon_bo_size(robj)) { + DRM_ERROR("[drm] Buffer too small for PACKET3 INDX_BUFFER " + "(need %u have %lu) !\n", + value + 1, + radeon_bo_size(robj)); + return -EINVAL; + } + return 0; +} + +static int r100_packet3_check(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt) +{ + struct radeon_cs_reloc *reloc; + struct r100_cs_track *track; + unsigned idx; + volatile uint32_t *ib; + int r; + + ib = p->ib->ptr; + idx = pkt->idx + 1; + track = (struct r100_cs_track *)p->track; + switch (pkt->opcode) { + case PACKET3_3D_LOAD_VBPNTR: + r = r100_packet3_load_vbpntr(p, pkt, idx); + if (r) + return r; + break; + case PACKET3_INDX_BUFFER: + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for packet3 %d\n", pkt->opcode); + r100_cs_dump_packet(p, pkt); + return r; + } + ib[idx+1] = radeon_get_ib_value(p, idx+1) + ((u32)reloc->lobj.gpu_offset); + r = r100_cs_track_check_pkt3_indx_buffer(p, pkt, reloc->robj); + if (r) { + return r; + } + break; + case 0x23: + /* 3D_RNDR_GEN_INDX_PRIM on r100/r200 */ + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for packet3 %d\n", pkt->opcode); + r100_cs_dump_packet(p, pkt); + return r; + } + ib[idx] = radeon_get_ib_value(p, idx) + ((u32)reloc->lobj.gpu_offset); + track->num_arrays = 1; + track->vtx_size = r100_get_vtx_size(radeon_get_ib_value(p, idx + 2)); + + track->arrays[0].robj = reloc->robj; + track->arrays[0].esize = track->vtx_size; + + track->max_indx = radeon_get_ib_value(p, idx+1); + + track->vap_vf_cntl = radeon_get_ib_value(p, idx+3); + track->immd_dwords = pkt->count - 1; + r = r100_cs_track_check(p->rdev, track); + if (r) + return r; + break; + case PACKET3_3D_DRAW_IMMD: + if (((radeon_get_ib_value(p, idx + 1) >> 4) & 0x3) != 3) { + DRM_ERROR("PRIM_WALK must be 3 for IMMD draw\n"); + return -EINVAL; + } + track->vtx_size = r100_get_vtx_size(radeon_get_ib_value(p, idx + 0)); + track->vap_vf_cntl = radeon_get_ib_value(p, idx + 1); + track->immd_dwords = pkt->count - 1; + r = r100_cs_track_check(p->rdev, track); + if (r) + return r; + break; + /* triggers drawing using in-packet vertex data */ + case PACKET3_3D_DRAW_IMMD_2: + if (((radeon_get_ib_value(p, idx) >> 4) & 0x3) != 3) { + DRM_ERROR("PRIM_WALK must be 3 for IMMD draw\n"); + return -EINVAL; + } + track->vap_vf_cntl = radeon_get_ib_value(p, idx); + track->immd_dwords = pkt->count; + r = r100_cs_track_check(p->rdev, track); + if (r) + return r; + break; + /* triggers drawing using in-packet vertex data */ + case PACKET3_3D_DRAW_VBUF_2: + track->vap_vf_cntl = radeon_get_ib_value(p, idx); + r = r100_cs_track_check(p->rdev, track); + if (r) + return r; + break; + /* triggers drawing of vertex buffers setup elsewhere */ + case PACKET3_3D_DRAW_INDX_2: + track->vap_vf_cntl = radeon_get_ib_value(p, idx); + r = r100_cs_track_check(p->rdev, track); + if (r) + return r; + break; + /* triggers drawing using indices to vertex buffer */ + case PACKET3_3D_DRAW_VBUF: + track->vap_vf_cntl = radeon_get_ib_value(p, idx + 1); + r = r100_cs_track_check(p->rdev, track); + if (r) + return r; + break; + /* triggers drawing of vertex buffers setup elsewhere */ + case PACKET3_3D_DRAW_INDX: + track->vap_vf_cntl = radeon_get_ib_value(p, idx + 1); + r = r100_cs_track_check(p->rdev, track); + if (r) + return r; + break; + /* triggers drawing using indices to vertex buffer */ + case PACKET3_3D_CLEAR_HIZ: + case PACKET3_3D_CLEAR_ZMASK: + if (p->rdev->hyperz_filp != p->filp) + return -EINVAL; + break; + case PACKET3_NOP: + break; + default: + DRM_ERROR("Packet3 opcode %x not supported\n", pkt->opcode); + return -EINVAL; + } + return 0; +} + +int r100_cs_parse(struct radeon_cs_parser *p) +{ + struct radeon_cs_packet pkt; + struct r100_cs_track *track; + int r; + + track = kzalloc(sizeof(*track), GFP_KERNEL); + r100_cs_track_clear(p->rdev, track); + p->track = track; + do { + r = r100_cs_packet_parse(p, &pkt, p->idx); + if (r) { + return r; + } + p->idx += pkt.count + 2; + switch (pkt.type) { + case PACKET_TYPE0: + if (p->rdev->family >= CHIP_R200) + r = r100_cs_parse_packet0(p, &pkt, + p->rdev->config.r100.reg_safe_bm, + p->rdev->config.r100.reg_safe_bm_size, + &r200_packet0_check); + else + r = r100_cs_parse_packet0(p, &pkt, + p->rdev->config.r100.reg_safe_bm, + p->rdev->config.r100.reg_safe_bm_size, + &r100_packet0_check); + break; + case PACKET_TYPE2: + break; + case PACKET_TYPE3: + r = r100_packet3_check(p, &pkt); + break; + default: + DRM_ERROR("Unknown packet type %d !\n", + pkt.type); + return -EINVAL; + } + if (r) { + return r; + } + } while (p->idx < p->chunks[p->chunk_ib_idx].length_dw); + return 0; +} + + +/* + * Global GPU functions + */ +void r100_errata(struct radeon_device *rdev) +{ + rdev->pll_errata = 0; + + if (rdev->family == CHIP_RV200 || rdev->family == CHIP_RS200) { + rdev->pll_errata |= CHIP_ERRATA_PLL_DUMMYREADS; + } + + if (rdev->family == CHIP_RV100 || + rdev->family == CHIP_RS100 || + rdev->family == CHIP_RS200) { + rdev->pll_errata |= CHIP_ERRATA_PLL_DELAY; + } +} + +/* Wait for vertical sync on primary CRTC */ +void r100_gpu_wait_for_vsync(struct radeon_device *rdev) +{ + uint32_t crtc_gen_cntl, tmp; + int i; + + crtc_gen_cntl = RREG32(RADEON_CRTC_GEN_CNTL); + if ((crtc_gen_cntl & RADEON_CRTC_DISP_REQ_EN_B) || + !(crtc_gen_cntl & RADEON_CRTC_EN)) { + return; + } + /* Clear the CRTC_VBLANK_SAVE bit */ + WREG32(RADEON_CRTC_STATUS, RADEON_CRTC_VBLANK_SAVE_CLEAR); + for (i = 0; i < rdev->usec_timeout; i++) { + tmp = RREG32(RADEON_CRTC_STATUS); + if (tmp & RADEON_CRTC_VBLANK_SAVE) { + return; + } + DRM_UDELAY(1); + } +} + +/* Wait for vertical sync on secondary CRTC */ +void r100_gpu_wait_for_vsync2(struct radeon_device *rdev) +{ + uint32_t crtc2_gen_cntl, tmp; + int i; + + crtc2_gen_cntl = RREG32(RADEON_CRTC2_GEN_CNTL); + if ((crtc2_gen_cntl & RADEON_CRTC2_DISP_REQ_EN_B) || + !(crtc2_gen_cntl & RADEON_CRTC2_EN)) + return; + + /* Clear the CRTC_VBLANK_SAVE bit */ + WREG32(RADEON_CRTC2_STATUS, RADEON_CRTC2_VBLANK_SAVE_CLEAR); + for (i = 0; i < rdev->usec_timeout; i++) { + tmp = RREG32(RADEON_CRTC2_STATUS); + if (tmp & RADEON_CRTC2_VBLANK_SAVE) { + return; + } + DRM_UDELAY(1); + } +} + +int r100_rbbm_fifo_wait_for_entry(struct radeon_device *rdev, unsigned n) +{ + unsigned i; + uint32_t tmp; + + for (i = 0; i < rdev->usec_timeout; i++) { + tmp = RREG32(RADEON_RBBM_STATUS) & RADEON_RBBM_FIFOCNT_MASK; + if (tmp >= n) { + return 0; + } + DRM_UDELAY(1); + } + return -1; +} + +int r100_gui_wait_for_idle(struct radeon_device *rdev) +{ + unsigned i; + uint32_t tmp; + + if (r100_rbbm_fifo_wait_for_entry(rdev, 64)) { + printk(KERN_WARNING "radeon: wait for empty RBBM fifo failed !" + " Bad things might happen.\n"); + } + for (i = 0; i < rdev->usec_timeout; i++) { + tmp = RREG32(RADEON_RBBM_STATUS); + if (!(tmp & RADEON_RBBM_ACTIVE)) { + return 0; + } + DRM_UDELAY(1); + } + return -1; +} + +int r100_mc_wait_for_idle(struct radeon_device *rdev) +{ + unsigned i; + uint32_t tmp; + + for (i = 0; i < rdev->usec_timeout; i++) { + /* read MC_STATUS */ + tmp = RREG32(RADEON_MC_STATUS); + if (tmp & RADEON_MC_IDLE) { + return 0; + } + DRM_UDELAY(1); + } + return -1; +} + +void r100_gpu_lockup_update(struct r100_gpu_lockup *lockup, struct radeon_cp *cp) +{ + lockup->last_cp_rptr = cp->rptr; + lockup->last_jiffies = jiffies; +} + +/** + * r100_gpu_cp_is_lockup() - check if CP is lockup by recording information + * @rdev: radeon device structure + * @lockup: r100_gpu_lockup structure holding CP lockup tracking informations + * @cp: radeon_cp structure holding CP information + * + * We don't need to initialize the lockup tracking information as we will either + * have CP rptr to a different value of jiffies wrap around which will force + * initialization of the lockup tracking informations. + * + * A possible false positivie is if we get call after while and last_cp_rptr == + * the current CP rptr, even if it's unlikely it might happen. To avoid this + * if the elapsed time since last call is bigger than 2 second than we return + * false and update the tracking information. Due to this the caller must call + * r100_gpu_cp_is_lockup several time in less than 2sec for lockup to be reported + * the fencing code should be cautious about that. + * + * Caller should write to the ring to force CP to do something so we don't get + * false positive when CP is just gived nothing to do. + * + **/ +bool r100_gpu_cp_is_lockup(struct radeon_device *rdev, struct r100_gpu_lockup *lockup, struct radeon_cp *cp) +{ + unsigned long cjiffies, elapsed; + + cjiffies = jiffies; + if (!time_after(cjiffies, lockup->last_jiffies)) { + /* likely a wrap around */ + lockup->last_cp_rptr = cp->rptr; + lockup->last_jiffies = jiffies; + return false; + } + if (cp->rptr != lockup->last_cp_rptr) { + /* CP is still working no lockup */ + lockup->last_cp_rptr = cp->rptr; + lockup->last_jiffies = jiffies; + return false; + } + elapsed = jiffies_to_msecs(cjiffies - lockup->last_jiffies); + if (elapsed >= 10000) { + dev_err(rdev->dev, "GPU lockup CP stall for more than %lumsec\n", elapsed); + return true; + } + /* give a chance to the GPU ... */ + return false; +} + +bool r100_gpu_is_lockup(struct radeon_device *rdev) +{ + u32 rbbm_status; + int r; + + rbbm_status = RREG32(R_000E40_RBBM_STATUS); + if (!G_000E40_GUI_ACTIVE(rbbm_status)) { + r100_gpu_lockup_update(&rdev->config.r100.lockup, &rdev->cp); + return false; + } + /* force CP activities */ + r = radeon_ring_lock(rdev, 2); + if (!r) { + /* PACKET2 NOP */ + radeon_ring_write(rdev, 0x80000000); + radeon_ring_write(rdev, 0x80000000); + radeon_ring_unlock_commit(rdev); + } + rdev->cp.rptr = RREG32(RADEON_CP_RB_RPTR); + return r100_gpu_cp_is_lockup(rdev, &rdev->config.r100.lockup, &rdev->cp); +} + +void r100_bm_disable(struct radeon_device *rdev) +{ + u32 tmp; + u16 tmp16; + + /* disable bus mastering */ + tmp = RREG32(R_000030_BUS_CNTL); + WREG32(R_000030_BUS_CNTL, (tmp & 0xFFFFFFFF) | 0x00000044); + mdelay(1); + WREG32(R_000030_BUS_CNTL, (tmp & 0xFFFFFFFF) | 0x00000042); + mdelay(1); + WREG32(R_000030_BUS_CNTL, (tmp & 0xFFFFFFFF) | 0x00000040); + tmp = RREG32(RADEON_BUS_CNTL); + mdelay(1); + pci_read_config_word(rdev->pdev, 0x4, &tmp16); + pci_write_config_word(rdev->pdev, 0x4, tmp16 & 0xFFFB); + mdelay(1); +} + +int r100_asic_reset(struct radeon_device *rdev) +{ + struct r100_mc_save save; + u32 status, tmp; + int ret = 0; + + status = RREG32(R_000E40_RBBM_STATUS); + if (!G_000E40_GUI_ACTIVE(status)) { + return 0; + } + r100_mc_stop(rdev, &save); + status = RREG32(R_000E40_RBBM_STATUS); + dev_info(rdev->dev, "(%s:%d) RBBM_STATUS=0x%08X\n", __func__, __LINE__, status); + /* stop CP */ + WREG32(RADEON_CP_CSQ_CNTL, 0); + tmp = RREG32(RADEON_CP_RB_CNTL); + WREG32(RADEON_CP_RB_CNTL, tmp | RADEON_RB_RPTR_WR_ENA); + WREG32(RADEON_CP_RB_RPTR_WR, 0); + WREG32(RADEON_CP_RB_WPTR, 0); + WREG32(RADEON_CP_RB_CNTL, tmp); + /* save PCI state */ + pci_save_state(rdev->pdev); + /* disable bus mastering */ + r100_bm_disable(rdev); + WREG32(R_0000F0_RBBM_SOFT_RESET, S_0000F0_SOFT_RESET_SE(1) | + S_0000F0_SOFT_RESET_RE(1) | + S_0000F0_SOFT_RESET_PP(1) | + S_0000F0_SOFT_RESET_RB(1)); + RREG32(R_0000F0_RBBM_SOFT_RESET); + mdelay(500); + WREG32(R_0000F0_RBBM_SOFT_RESET, 0); + mdelay(1); + status = RREG32(R_000E40_RBBM_STATUS); + dev_info(rdev->dev, "(%s:%d) RBBM_STATUS=0x%08X\n", __func__, __LINE__, status); + /* reset CP */ + WREG32(R_0000F0_RBBM_SOFT_RESET, S_0000F0_SOFT_RESET_CP(1)); + RREG32(R_0000F0_RBBM_SOFT_RESET); + mdelay(500); + WREG32(R_0000F0_RBBM_SOFT_RESET, 0); + mdelay(1); + status = RREG32(R_000E40_RBBM_STATUS); + dev_info(rdev->dev, "(%s:%d) RBBM_STATUS=0x%08X\n", __func__, __LINE__, status); + /* restore PCI & busmastering */ + pci_restore_state(rdev->pdev); + r100_enable_bm(rdev); + /* Check if GPU is idle */ + if (G_000E40_SE_BUSY(status) || G_000E40_RE_BUSY(status) || + G_000E40_TAM_BUSY(status) || G_000E40_PB_BUSY(status)) { + dev_err(rdev->dev, "failed to reset GPU\n"); + rdev->gpu_lockup = true; + ret = -1; + } else + dev_info(rdev->dev, "GPU reset succeed\n"); + r100_mc_resume(rdev, &save); + return ret; +} + +void r100_set_common_regs(struct radeon_device *rdev) +{ + struct drm_device *dev = rdev->ddev; + bool force_dac2 = false; + u32 tmp; + + /* set these so they don't interfere with anything */ + WREG32(RADEON_OV0_SCALE_CNTL, 0); + WREG32(RADEON_SUBPIC_CNTL, 0); + WREG32(RADEON_VIPH_CONTROL, 0); + WREG32(RADEON_I2C_CNTL_1, 0); + WREG32(RADEON_DVI_I2C_CNTL_1, 0); + WREG32(RADEON_CAP0_TRIG_CNTL, 0); + WREG32(RADEON_CAP1_TRIG_CNTL, 0); + + /* always set up dac2 on rn50 and some rv100 as lots + * of servers seem to wire it up to a VGA port but + * don't report it in the bios connector + * table. + */ + switch (dev->pdev->device) { + /* RN50 */ + case 0x515e: + case 0x5969: + force_dac2 = true; + break; + /* RV100*/ + case 0x5159: + case 0x515a: + /* DELL triple head servers */ + if ((dev->pdev->subsystem_vendor == 0x1028 /* DELL */) && + ((dev->pdev->subsystem_device == 0x016c) || + (dev->pdev->subsystem_device == 0x016d) || + (dev->pdev->subsystem_device == 0x016e) || + (dev->pdev->subsystem_device == 0x016f) || + (dev->pdev->subsystem_device == 0x0170) || + (dev->pdev->subsystem_device == 0x017d) || + (dev->pdev->subsystem_device == 0x017e) || + (dev->pdev->subsystem_device == 0x0183) || + (dev->pdev->subsystem_device == 0x018a) || + (dev->pdev->subsystem_device == 0x019a))) + force_dac2 = true; + break; + } + + if (force_dac2) { + u32 disp_hw_debug = RREG32(RADEON_DISP_HW_DEBUG); + u32 tv_dac_cntl = RREG32(RADEON_TV_DAC_CNTL); + u32 dac2_cntl = RREG32(RADEON_DAC_CNTL2); + + /* For CRT on DAC2, don't turn it on if BIOS didn't + enable it, even it's detected. + */ + + /* force it to crtc0 */ + dac2_cntl &= ~RADEON_DAC2_DAC_CLK_SEL; + dac2_cntl |= RADEON_DAC2_DAC2_CLK_SEL; + disp_hw_debug |= RADEON_CRT2_DISP1_SEL; + + /* set up the TV DAC */ + tv_dac_cntl &= ~(RADEON_TV_DAC_PEDESTAL | + RADEON_TV_DAC_STD_MASK | + RADEON_TV_DAC_RDACPD | + RADEON_TV_DAC_GDACPD | + RADEON_TV_DAC_BDACPD | + RADEON_TV_DAC_BGADJ_MASK | + RADEON_TV_DAC_DACADJ_MASK); + tv_dac_cntl |= (RADEON_TV_DAC_NBLANK | + RADEON_TV_DAC_NHOLD | + RADEON_TV_DAC_STD_PS2 | + (0x58 << 16)); + + WREG32(RADEON_TV_DAC_CNTL, tv_dac_cntl); + WREG32(RADEON_DISP_HW_DEBUG, disp_hw_debug); + WREG32(RADEON_DAC_CNTL2, dac2_cntl); + } + + /* switch PM block to ACPI mode */ + tmp = RREG32_PLL(RADEON_PLL_PWRMGT_CNTL); + tmp &= ~RADEON_PM_MODE_SEL; + WREG32_PLL(RADEON_PLL_PWRMGT_CNTL, tmp); + +} + +/* + * VRAM info + */ +static void r100_vram_get_type(struct radeon_device *rdev) +{ + uint32_t tmp; + + rdev->mc.vram_is_ddr = false; + if (rdev->flags & RADEON_IS_IGP) + rdev->mc.vram_is_ddr = true; + else if (RREG32(RADEON_MEM_SDRAM_MODE_REG) & RADEON_MEM_CFG_TYPE_DDR) + rdev->mc.vram_is_ddr = true; + if ((rdev->family == CHIP_RV100) || + (rdev->family == CHIP_RS100) || + (rdev->family == CHIP_RS200)) { + tmp = RREG32(RADEON_MEM_CNTL); + if (tmp & RV100_HALF_MODE) { + rdev->mc.vram_width = 32; + } else { + rdev->mc.vram_width = 64; + } + if (rdev->flags & RADEON_SINGLE_CRTC) { + rdev->mc.vram_width /= 4; + rdev->mc.vram_is_ddr = true; + } + } else if (rdev->family <= CHIP_RV280) { + tmp = RREG32(RADEON_MEM_CNTL); + if (tmp & RADEON_MEM_NUM_CHANNELS_MASK) { + rdev->mc.vram_width = 128; + } else { + rdev->mc.vram_width = 64; + } + } else { + /* newer IGPs */ + rdev->mc.vram_width = 128; + } +} + +static u32 r100_get_accessible_vram(struct radeon_device *rdev) +{ + u32 aper_size; + u8 byte; + + aper_size = RREG32(RADEON_CONFIG_APER_SIZE); + + /* Set HDP_APER_CNTL only on cards that are known not to be broken, + * that is has the 2nd generation multifunction PCI interface + */ + if (rdev->family == CHIP_RV280 || + rdev->family >= CHIP_RV350) { + WREG32_P(RADEON_HOST_PATH_CNTL, RADEON_HDP_APER_CNTL, + ~RADEON_HDP_APER_CNTL); + DRM_INFO("Generation 2 PCI interface, using max accessible memory\n"); + return aper_size * 2; + } + + /* Older cards have all sorts of funny issues to deal with. First + * check if it's a multifunction card by reading the PCI config + * header type... Limit those to one aperture size + */ + pci_read_config_byte(rdev->pdev, 0xe, &byte); + if (byte & 0x80) { + DRM_INFO("Generation 1 PCI interface in multifunction mode\n"); + DRM_INFO("Limiting VRAM to one aperture\n"); + return aper_size; + } + + /* Single function older card. We read HDP_APER_CNTL to see how the BIOS + * have set it up. We don't write this as it's broken on some ASICs but + * we expect the BIOS to have done the right thing (might be too optimistic...) + */ + if (RREG32(RADEON_HOST_PATH_CNTL) & RADEON_HDP_APER_CNTL) + return aper_size * 2; + return aper_size; +} + +void r100_vram_init_sizes(struct radeon_device *rdev) +{ + u64 config_aper_size; + + /* work out accessible VRAM */ + rdev->mc.aper_base = pci_resource_start(rdev->pdev, 0); + rdev->mc.aper_size = pci_resource_len(rdev->pdev, 0); + rdev->mc.visible_vram_size = r100_get_accessible_vram(rdev); + /* FIXME we don't use the second aperture yet when we could use it */ + if (rdev->mc.visible_vram_size > rdev->mc.aper_size) + rdev->mc.visible_vram_size = rdev->mc.aper_size; + config_aper_size = RREG32(RADEON_CONFIG_APER_SIZE); + if (rdev->flags & RADEON_IS_IGP) { + uint32_t tom; + /* read NB_TOM to get the amount of ram stolen for the GPU */ + tom = RREG32(RADEON_NB_TOM); + rdev->mc.real_vram_size = (((tom >> 16) - (tom & 0xffff) + 1) << 16); + WREG32(RADEON_CONFIG_MEMSIZE, rdev->mc.real_vram_size); + rdev->mc.mc_vram_size = rdev->mc.real_vram_size; + } else { + rdev->mc.real_vram_size = RREG32(RADEON_CONFIG_MEMSIZE); + /* Some production boards of m6 will report 0 + * if it's 8 MB + */ + if (rdev->mc.real_vram_size == 0) { + rdev->mc.real_vram_size = 8192 * 1024; + WREG32(RADEON_CONFIG_MEMSIZE, rdev->mc.real_vram_size); + } + /* Fix for RN50, M6, M7 with 8/16/32(??) MBs of VRAM - + * Novell bug 204882 + along with lots of ubuntu ones + */ + if (rdev->mc.aper_size > config_aper_size) + config_aper_size = rdev->mc.aper_size; + + if (config_aper_size > rdev->mc.real_vram_size) + rdev->mc.mc_vram_size = config_aper_size; + else + rdev->mc.mc_vram_size = rdev->mc.real_vram_size; + } +} + +void r100_vga_set_state(struct radeon_device *rdev, bool state) +{ + uint32_t temp; + + temp = RREG32(RADEON_CONFIG_CNTL); + if (state == false) { + temp &= ~RADEON_CFG_VGA_RAM_EN; + temp |= RADEON_CFG_VGA_IO_DIS; + } else { + temp &= ~RADEON_CFG_VGA_IO_DIS; + } + WREG32(RADEON_CONFIG_CNTL, temp); +} + +void r100_mc_init(struct radeon_device *rdev) +{ + u64 base; + + r100_vram_get_type(rdev); + r100_vram_init_sizes(rdev); + base = rdev->mc.aper_base; + if (rdev->flags & RADEON_IS_IGP) + base = (RREG32(RADEON_NB_TOM) & 0xffff) << 16; + radeon_vram_location(rdev, &rdev->mc, base); + rdev->mc.gtt_base_align = 0; + if (!(rdev->flags & RADEON_IS_AGP)) + radeon_gtt_location(rdev, &rdev->mc); + radeon_update_bandwidth_info(rdev); +} + + +/* + * Indirect registers accessor + */ +void r100_pll_errata_after_index(struct radeon_device *rdev) +{ + if (rdev->pll_errata & CHIP_ERRATA_PLL_DUMMYREADS) { + (void)RREG32(RADEON_CLOCK_CNTL_DATA); + (void)RREG32(RADEON_CRTC_GEN_CNTL); + } +} + +static void r100_pll_errata_after_data(struct radeon_device *rdev) +{ + /* This workarounds is necessary on RV100, RS100 and RS200 chips + * or the chip could hang on a subsequent access + */ + if (rdev->pll_errata & CHIP_ERRATA_PLL_DELAY) { + udelay(5000); + } + + /* This function is required to workaround a hardware bug in some (all?) + * revisions of the R300. This workaround should be called after every + * CLOCK_CNTL_INDEX register access. If not, register reads afterward + * may not be correct. + */ + if (rdev->pll_errata & CHIP_ERRATA_R300_CG) { + uint32_t save, tmp; + + save = RREG32(RADEON_CLOCK_CNTL_INDEX); + tmp = save & ~(0x3f | RADEON_PLL_WR_EN); + WREG32(RADEON_CLOCK_CNTL_INDEX, tmp); + tmp = RREG32(RADEON_CLOCK_CNTL_DATA); + WREG32(RADEON_CLOCK_CNTL_INDEX, save); + } +} + +uint32_t r100_pll_rreg(struct radeon_device *rdev, uint32_t reg) +{ + uint32_t data; + + WREG8(RADEON_CLOCK_CNTL_INDEX, reg & 0x3f); + r100_pll_errata_after_index(rdev); + data = RREG32(RADEON_CLOCK_CNTL_DATA); + r100_pll_errata_after_data(rdev); + return data; +} + +void r100_pll_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v) +{ + WREG8(RADEON_CLOCK_CNTL_INDEX, ((reg & 0x3f) | RADEON_PLL_WR_EN)); + r100_pll_errata_after_index(rdev); + WREG32(RADEON_CLOCK_CNTL_DATA, v); + r100_pll_errata_after_data(rdev); +} + +void r100_set_safe_registers(struct radeon_device *rdev) +{ + if (ASIC_IS_RN50(rdev)) { + rdev->config.r100.reg_safe_bm = rn50_reg_safe_bm; + rdev->config.r100.reg_safe_bm_size = ARRAY_SIZE(rn50_reg_safe_bm); + } else if (rdev->family < CHIP_R200) { + rdev->config.r100.reg_safe_bm = r100_reg_safe_bm; + rdev->config.r100.reg_safe_bm_size = ARRAY_SIZE(r100_reg_safe_bm); + } else { + r200_set_safe_registers(rdev); + } +} + +/* + * Debugfs info + */ +#if defined(CONFIG_DEBUG_FS) +static int r100_debugfs_rbbm_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + struct radeon_device *rdev = dev->dev_private; + uint32_t reg, value; + unsigned i; + + seq_printf(m, "RBBM_STATUS 0x%08x\n", RREG32(RADEON_RBBM_STATUS)); + seq_printf(m, "RBBM_CMDFIFO_STAT 0x%08x\n", RREG32(0xE7C)); + seq_printf(m, "CP_STAT 0x%08x\n", RREG32(RADEON_CP_STAT)); + for (i = 0; i < 64; i++) { + WREG32(RADEON_RBBM_CMDFIFO_ADDR, i | 0x100); + reg = (RREG32(RADEON_RBBM_CMDFIFO_DATA) - 1) >> 2; + WREG32(RADEON_RBBM_CMDFIFO_ADDR, i); + value = RREG32(RADEON_RBBM_CMDFIFO_DATA); + seq_printf(m, "[0x%03X] 0x%04X=0x%08X\n", i, reg, value); + } + return 0; +} + +static int r100_debugfs_cp_ring_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + struct radeon_device *rdev = dev->dev_private; + uint32_t rdp, wdp; + unsigned count, i, j; + + radeon_ring_free_size(rdev); + rdp = RREG32(RADEON_CP_RB_RPTR); + wdp = RREG32(RADEON_CP_RB_WPTR); + count = (rdp + rdev->cp.ring_size - wdp) & rdev->cp.ptr_mask; + seq_printf(m, "CP_STAT 0x%08x\n", RREG32(RADEON_CP_STAT)); + seq_printf(m, "CP_RB_WPTR 0x%08x\n", wdp); + seq_printf(m, "CP_RB_RPTR 0x%08x\n", rdp); + seq_printf(m, "%u free dwords in ring\n", rdev->cp.ring_free_dw); + seq_printf(m, "%u dwords in ring\n", count); + for (j = 0; j <= count; j++) { + i = (rdp + j) & rdev->cp.ptr_mask; + seq_printf(m, "r[%04d]=0x%08x\n", i, rdev->cp.ring[i]); + } + return 0; +} + + +static int r100_debugfs_cp_csq_fifo(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + struct radeon_device *rdev = dev->dev_private; + uint32_t csq_stat, csq2_stat, tmp; + unsigned r_rptr, r_wptr, ib1_rptr, ib1_wptr, ib2_rptr, ib2_wptr; + unsigned i; + + seq_printf(m, "CP_STAT 0x%08x\n", RREG32(RADEON_CP_STAT)); + seq_printf(m, "CP_CSQ_MODE 0x%08x\n", RREG32(RADEON_CP_CSQ_MODE)); + csq_stat = RREG32(RADEON_CP_CSQ_STAT); + csq2_stat = RREG32(RADEON_CP_CSQ2_STAT); + r_rptr = (csq_stat >> 0) & 0x3ff; + r_wptr = (csq_stat >> 10) & 0x3ff; + ib1_rptr = (csq_stat >> 20) & 0x3ff; + ib1_wptr = (csq2_stat >> 0) & 0x3ff; + ib2_rptr = (csq2_stat >> 10) & 0x3ff; + ib2_wptr = (csq2_stat >> 20) & 0x3ff; + seq_printf(m, "CP_CSQ_STAT 0x%08x\n", csq_stat); + seq_printf(m, "CP_CSQ2_STAT 0x%08x\n", csq2_stat); + seq_printf(m, "Ring rptr %u\n", r_rptr); + seq_printf(m, "Ring wptr %u\n", r_wptr); + seq_printf(m, "Indirect1 rptr %u\n", ib1_rptr); + seq_printf(m, "Indirect1 wptr %u\n", ib1_wptr); + seq_printf(m, "Indirect2 rptr %u\n", ib2_rptr); + seq_printf(m, "Indirect2 wptr %u\n", ib2_wptr); + /* FIXME: 0, 128, 640 depends on fifo setup see cp_init_kms + * 128 = indirect1_start * 8 & 640 = indirect2_start * 8 */ + seq_printf(m, "Ring fifo:\n"); + for (i = 0; i < 256; i++) { + WREG32(RADEON_CP_CSQ_ADDR, i << 2); + tmp = RREG32(RADEON_CP_CSQ_DATA); + seq_printf(m, "rfifo[%04d]=0x%08X\n", i, tmp); + } + seq_printf(m, "Indirect1 fifo:\n"); + for (i = 256; i <= 512; i++) { + WREG32(RADEON_CP_CSQ_ADDR, i << 2); + tmp = RREG32(RADEON_CP_CSQ_DATA); + seq_printf(m, "ib1fifo[%04d]=0x%08X\n", i, tmp); + } + seq_printf(m, "Indirect2 fifo:\n"); + for (i = 640; i < ib1_wptr; i++) { + WREG32(RADEON_CP_CSQ_ADDR, i << 2); + tmp = RREG32(RADEON_CP_CSQ_DATA); + seq_printf(m, "ib2fifo[%04d]=0x%08X\n", i, tmp); + } + return 0; +} + +static int r100_debugfs_mc_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + struct radeon_device *rdev = dev->dev_private; + uint32_t tmp; + + tmp = RREG32(RADEON_CONFIG_MEMSIZE); + seq_printf(m, "CONFIG_MEMSIZE 0x%08x\n", tmp); + tmp = RREG32(RADEON_MC_FB_LOCATION); + seq_printf(m, "MC_FB_LOCATION 0x%08x\n", tmp); + tmp = RREG32(RADEON_BUS_CNTL); + seq_printf(m, "BUS_CNTL 0x%08x\n", tmp); + tmp = RREG32(RADEON_MC_AGP_LOCATION); + seq_printf(m, "MC_AGP_LOCATION 0x%08x\n", tmp); + tmp = RREG32(RADEON_AGP_BASE); + seq_printf(m, "AGP_BASE 0x%08x\n", tmp); + tmp = RREG32(RADEON_HOST_PATH_CNTL); + seq_printf(m, "HOST_PATH_CNTL 0x%08x\n", tmp); + tmp = RREG32(0x01D0); + seq_printf(m, "AIC_CTRL 0x%08x\n", tmp); + tmp = RREG32(RADEON_AIC_LO_ADDR); + seq_printf(m, "AIC_LO_ADDR 0x%08x\n", tmp); + tmp = RREG32(RADEON_AIC_HI_ADDR); + seq_printf(m, "AIC_HI_ADDR 0x%08x\n", tmp); + tmp = RREG32(0x01E4); + seq_printf(m, "AIC_TLB_ADDR 0x%08x\n", tmp); + return 0; +} + +static struct drm_info_list r100_debugfs_rbbm_list[] = { + {"r100_rbbm_info", r100_debugfs_rbbm_info, 0, NULL}, +}; + +static struct drm_info_list r100_debugfs_cp_list[] = { + {"r100_cp_ring_info", r100_debugfs_cp_ring_info, 0, NULL}, + {"r100_cp_csq_fifo", r100_debugfs_cp_csq_fifo, 0, NULL}, +}; + +static struct drm_info_list r100_debugfs_mc_info_list[] = { + {"r100_mc_info", r100_debugfs_mc_info, 0, NULL}, +}; +#endif + +int r100_debugfs_rbbm_init(struct radeon_device *rdev) +{ +#if defined(CONFIG_DEBUG_FS) + return radeon_debugfs_add_files(rdev, r100_debugfs_rbbm_list, 1); +#else + return 0; +#endif +} + +int r100_debugfs_cp_init(struct radeon_device *rdev) +{ +#if defined(CONFIG_DEBUG_FS) + return radeon_debugfs_add_files(rdev, r100_debugfs_cp_list, 2); +#else + return 0; +#endif +} + +int r100_debugfs_mc_info_init(struct radeon_device *rdev) +{ +#if defined(CONFIG_DEBUG_FS) + return radeon_debugfs_add_files(rdev, r100_debugfs_mc_info_list, 1); +#else + return 0; +#endif +} + +int r100_set_surface_reg(struct radeon_device *rdev, int reg, + uint32_t tiling_flags, uint32_t pitch, + uint32_t offset, uint32_t obj_size) +{ + int surf_index = reg * 16; + int flags = 0; + + if (rdev->family <= CHIP_RS200) { + if ((tiling_flags & (RADEON_TILING_MACRO|RADEON_TILING_MICRO)) + == (RADEON_TILING_MACRO|RADEON_TILING_MICRO)) + flags |= RADEON_SURF_TILE_COLOR_BOTH; + if (tiling_flags & RADEON_TILING_MACRO) + flags |= RADEON_SURF_TILE_COLOR_MACRO; + } else if (rdev->family <= CHIP_RV280) { + if (tiling_flags & (RADEON_TILING_MACRO)) + flags |= R200_SURF_TILE_COLOR_MACRO; + if (tiling_flags & RADEON_TILING_MICRO) + flags |= R200_SURF_TILE_COLOR_MICRO; + } else { + if (tiling_flags & RADEON_TILING_MACRO) + flags |= R300_SURF_TILE_MACRO; + if (tiling_flags & RADEON_TILING_MICRO) + flags |= R300_SURF_TILE_MICRO; + } + + if (tiling_flags & RADEON_TILING_SWAP_16BIT) + flags |= RADEON_SURF_AP0_SWP_16BPP | RADEON_SURF_AP1_SWP_16BPP; + if (tiling_flags & RADEON_TILING_SWAP_32BIT) + flags |= RADEON_SURF_AP0_SWP_32BPP | RADEON_SURF_AP1_SWP_32BPP; + + /* when we aren't tiling the pitch seems to needs to be furtherdivided down. - tested on power5 + rn50 server */ + if (tiling_flags & (RADEON_TILING_SWAP_16BIT | RADEON_TILING_SWAP_32BIT)) { + if (!(tiling_flags & (RADEON_TILING_MACRO | RADEON_TILING_MICRO))) + if (ASIC_IS_RN50(rdev)) + pitch /= 16; + } + + /* r100/r200 divide by 16 */ + if (rdev->family < CHIP_R300) + flags |= pitch / 16; + else + flags |= pitch / 8; + + + DRM_DEBUG_KMS("writing surface %d %d %x %x\n", reg, flags, offset, offset+obj_size-1); + WREG32(RADEON_SURFACE0_INFO + surf_index, flags); + WREG32(RADEON_SURFACE0_LOWER_BOUND + surf_index, offset); + WREG32(RADEON_SURFACE0_UPPER_BOUND + surf_index, offset + obj_size - 1); + return 0; +} + +void r100_clear_surface_reg(struct radeon_device *rdev, int reg) +{ + int surf_index = reg * 16; + WREG32(RADEON_SURFACE0_INFO + surf_index, 0); +} + +void r100_bandwidth_update(struct radeon_device *rdev) +{ + fixed20_12 trcd_ff, trp_ff, tras_ff, trbs_ff, tcas_ff; + fixed20_12 sclk_ff, mclk_ff, sclk_eff_ff, sclk_delay_ff; + fixed20_12 peak_disp_bw, mem_bw, pix_clk, pix_clk2, temp_ff, crit_point_ff; + uint32_t temp, data, mem_trcd, mem_trp, mem_tras; + fixed20_12 memtcas_ff[8] = { + dfixed_init(1), + dfixed_init(2), + dfixed_init(3), + dfixed_init(0), + dfixed_init_half(1), + dfixed_init_half(2), + dfixed_init(0), + }; + fixed20_12 memtcas_rs480_ff[8] = { + dfixed_init(0), + dfixed_init(1), + dfixed_init(2), + dfixed_init(3), + dfixed_init(0), + dfixed_init_half(1), + dfixed_init_half(2), + dfixed_init_half(3), + }; + fixed20_12 memtcas2_ff[8] = { + dfixed_init(0), + dfixed_init(1), + dfixed_init(2), + dfixed_init(3), + dfixed_init(4), + dfixed_init(5), + dfixed_init(6), + dfixed_init(7), + }; + fixed20_12 memtrbs[8] = { + dfixed_init(1), + dfixed_init_half(1), + dfixed_init(2), + dfixed_init_half(2), + dfixed_init(3), + dfixed_init_half(3), + dfixed_init(4), + dfixed_init_half(4) + }; + fixed20_12 memtrbs_r4xx[8] = { + dfixed_init(4), + dfixed_init(5), + dfixed_init(6), + dfixed_init(7), + dfixed_init(8), + dfixed_init(9), + dfixed_init(10), + dfixed_init(11) + }; + fixed20_12 min_mem_eff; + fixed20_12 mc_latency_sclk, mc_latency_mclk, k1; + fixed20_12 cur_latency_mclk, cur_latency_sclk; + fixed20_12 disp_latency, disp_latency_overhead, disp_drain_rate, + disp_drain_rate2, read_return_rate; + fixed20_12 time_disp1_drop_priority; + int c; + int cur_size = 16; /* in octawords */ + int critical_point = 0, critical_point2; +/* uint32_t read_return_rate, time_disp1_drop_priority; */ + int stop_req, max_stop_req; + struct drm_display_mode *mode1 = NULL; + struct drm_display_mode *mode2 = NULL; + uint32_t pixel_bytes1 = 0; + uint32_t pixel_bytes2 = 0; + + radeon_update_display_priority(rdev); + + if (rdev->mode_info.crtcs[0]->base.enabled) { + mode1 = &rdev->mode_info.crtcs[0]->base.mode; + pixel_bytes1 = rdev->mode_info.crtcs[0]->base.fb->bits_per_pixel / 8; + } + if (!(rdev->flags & RADEON_SINGLE_CRTC)) { + if (rdev->mode_info.crtcs[1]->base.enabled) { + mode2 = &rdev->mode_info.crtcs[1]->base.mode; + pixel_bytes2 = rdev->mode_info.crtcs[1]->base.fb->bits_per_pixel / 8; + } + } + + min_mem_eff.full = dfixed_const_8(0); + /* get modes */ + if ((rdev->disp_priority == 2) && ASIC_IS_R300(rdev)) { + uint32_t mc_init_misc_lat_timer = RREG32(R300_MC_INIT_MISC_LAT_TIMER); + mc_init_misc_lat_timer &= ~(R300_MC_DISP1R_INIT_LAT_MASK << R300_MC_DISP1R_INIT_LAT_SHIFT); + mc_init_misc_lat_timer &= ~(R300_MC_DISP0R_INIT_LAT_MASK << R300_MC_DISP0R_INIT_LAT_SHIFT); + /* check crtc enables */ + if (mode2) + mc_init_misc_lat_timer |= (1 << R300_MC_DISP1R_INIT_LAT_SHIFT); + if (mode1) + mc_init_misc_lat_timer |= (1 << R300_MC_DISP0R_INIT_LAT_SHIFT); + WREG32(R300_MC_INIT_MISC_LAT_TIMER, mc_init_misc_lat_timer); + } + + /* + * determine is there is enough bw for current mode + */ + sclk_ff = rdev->pm.sclk; + mclk_ff = rdev->pm.mclk; + + temp = (rdev->mc.vram_width / 8) * (rdev->mc.vram_is_ddr ? 2 : 1); + temp_ff.full = dfixed_const(temp); + mem_bw.full = dfixed_mul(mclk_ff, temp_ff); + + pix_clk.full = 0; + pix_clk2.full = 0; + peak_disp_bw.full = 0; + if (mode1) { + temp_ff.full = dfixed_const(1000); + pix_clk.full = dfixed_const(mode1->clock); /* convert to fixed point */ + pix_clk.full = dfixed_div(pix_clk, temp_ff); + temp_ff.full = dfixed_const(pixel_bytes1); + peak_disp_bw.full += dfixed_mul(pix_clk, temp_ff); + } + if (mode2) { + temp_ff.full = dfixed_const(1000); + pix_clk2.full = dfixed_const(mode2->clock); /* convert to fixed point */ + pix_clk2.full = dfixed_div(pix_clk2, temp_ff); + temp_ff.full = dfixed_const(pixel_bytes2); + peak_disp_bw.full += dfixed_mul(pix_clk2, temp_ff); + } + + mem_bw.full = dfixed_mul(mem_bw, min_mem_eff); + if (peak_disp_bw.full >= mem_bw.full) { + DRM_ERROR("You may not have enough display bandwidth for current mode\n" + "If you have flickering problem, try to lower resolution, refresh rate, or color depth\n"); + } + + /* Get values from the EXT_MEM_CNTL register...converting its contents. */ + temp = RREG32(RADEON_MEM_TIMING_CNTL); + if ((rdev->family == CHIP_RV100) || (rdev->flags & RADEON_IS_IGP)) { /* RV100, M6, IGPs */ + mem_trcd = ((temp >> 2) & 0x3) + 1; + mem_trp = ((temp & 0x3)) + 1; + mem_tras = ((temp & 0x70) >> 4) + 1; + } else if (rdev->family == CHIP_R300 || + rdev->family == CHIP_R350) { /* r300, r350 */ + mem_trcd = (temp & 0x7) + 1; + mem_trp = ((temp >> 8) & 0x7) + 1; + mem_tras = ((temp >> 11) & 0xf) + 4; + } else if (rdev->family == CHIP_RV350 || + rdev->family <= CHIP_RV380) { + /* rv3x0 */ + mem_trcd = (temp & 0x7) + 3; + mem_trp = ((temp >> 8) & 0x7) + 3; + mem_tras = ((temp >> 11) & 0xf) + 6; + } else if (rdev->family == CHIP_R420 || + rdev->family == CHIP_R423 || + rdev->family == CHIP_RV410) { + /* r4xx */ + mem_trcd = (temp & 0xf) + 3; + if (mem_trcd > 15) + mem_trcd = 15; + mem_trp = ((temp >> 8) & 0xf) + 3; + if (mem_trp > 15) + mem_trp = 15; + mem_tras = ((temp >> 12) & 0x1f) + 6; + if (mem_tras > 31) + mem_tras = 31; + } else { /* RV200, R200 */ + mem_trcd = (temp & 0x7) + 1; + mem_trp = ((temp >> 8) & 0x7) + 1; + mem_tras = ((temp >> 12) & 0xf) + 4; + } + /* convert to FF */ + trcd_ff.full = dfixed_const(mem_trcd); + trp_ff.full = dfixed_const(mem_trp); + tras_ff.full = dfixed_const(mem_tras); + + /* Get values from the MEM_SDRAM_MODE_REG register...converting its */ + temp = RREG32(RADEON_MEM_SDRAM_MODE_REG); + data = (temp & (7 << 20)) >> 20; + if ((rdev->family == CHIP_RV100) || rdev->flags & RADEON_IS_IGP) { + if (rdev->family == CHIP_RS480) /* don't think rs400 */ + tcas_ff = memtcas_rs480_ff[data]; + else + tcas_ff = memtcas_ff[data]; + } else + tcas_ff = memtcas2_ff[data]; + + if (rdev->family == CHIP_RS400 || + rdev->family == CHIP_RS480) { + /* extra cas latency stored in bits 23-25 0-4 clocks */ + data = (temp >> 23) & 0x7; + if (data < 5) + tcas_ff.full += dfixed_const(data); + } + + if (ASIC_IS_R300(rdev) && !(rdev->flags & RADEON_IS_IGP)) { + /* on the R300, Tcas is included in Trbs. + */ + temp = RREG32(RADEON_MEM_CNTL); + data = (R300_MEM_NUM_CHANNELS_MASK & temp); + if (data == 1) { + if (R300_MEM_USE_CD_CH_ONLY & temp) { + temp = RREG32(R300_MC_IND_INDEX); + temp &= ~R300_MC_IND_ADDR_MASK; + temp |= R300_MC_READ_CNTL_CD_mcind; + WREG32(R300_MC_IND_INDEX, temp); + temp = RREG32(R300_MC_IND_DATA); + data = (R300_MEM_RBS_POSITION_C_MASK & temp); + } else { + temp = RREG32(R300_MC_READ_CNTL_AB); + data = (R300_MEM_RBS_POSITION_A_MASK & temp); + } + } else { + temp = RREG32(R300_MC_READ_CNTL_AB); + data = (R300_MEM_RBS_POSITION_A_MASK & temp); + } + if (rdev->family == CHIP_RV410 || + rdev->family == CHIP_R420 || + rdev->family == CHIP_R423) + trbs_ff = memtrbs_r4xx[data]; + else + trbs_ff = memtrbs[data]; + tcas_ff.full += trbs_ff.full; + } + + sclk_eff_ff.full = sclk_ff.full; + + if (rdev->flags & RADEON_IS_AGP) { + fixed20_12 agpmode_ff; + agpmode_ff.full = dfixed_const(radeon_agpmode); + temp_ff.full = dfixed_const_666(16); + sclk_eff_ff.full -= dfixed_mul(agpmode_ff, temp_ff); + } + /* TODO PCIE lanes may affect this - agpmode == 16?? */ + + if (ASIC_IS_R300(rdev)) { + sclk_delay_ff.full = dfixed_const(250); + } else { + if ((rdev->family == CHIP_RV100) || + rdev->flags & RADEON_IS_IGP) { + if (rdev->mc.vram_is_ddr) + sclk_delay_ff.full = dfixed_const(41); + else + sclk_delay_ff.full = dfixed_const(33); + } else { + if (rdev->mc.vram_width == 128) + sclk_delay_ff.full = dfixed_const(57); + else + sclk_delay_ff.full = dfixed_const(41); + } + } + + mc_latency_sclk.full = dfixed_div(sclk_delay_ff, sclk_eff_ff); + + if (rdev->mc.vram_is_ddr) { + if (rdev->mc.vram_width == 32) { + k1.full = dfixed_const(40); + c = 3; + } else { + k1.full = dfixed_const(20); + c = 1; + } + } else { + k1.full = dfixed_const(40); + c = 3; + } + + temp_ff.full = dfixed_const(2); + mc_latency_mclk.full = dfixed_mul(trcd_ff, temp_ff); + temp_ff.full = dfixed_const(c); + mc_latency_mclk.full += dfixed_mul(tcas_ff, temp_ff); + temp_ff.full = dfixed_const(4); + mc_latency_mclk.full += dfixed_mul(tras_ff, temp_ff); + mc_latency_mclk.full += dfixed_mul(trp_ff, temp_ff); + mc_latency_mclk.full += k1.full; + + mc_latency_mclk.full = dfixed_div(mc_latency_mclk, mclk_ff); + mc_latency_mclk.full += dfixed_div(temp_ff, sclk_eff_ff); + + /* + HW cursor time assuming worst case of full size colour cursor. + */ + temp_ff.full = dfixed_const((2 * (cur_size - (rdev->mc.vram_is_ddr + 1)))); + temp_ff.full += trcd_ff.full; + if (temp_ff.full < tras_ff.full) + temp_ff.full = tras_ff.full; + cur_latency_mclk.full = dfixed_div(temp_ff, mclk_ff); + + temp_ff.full = dfixed_const(cur_size); + cur_latency_sclk.full = dfixed_div(temp_ff, sclk_eff_ff); + /* + Find the total latency for the display data. + */ + disp_latency_overhead.full = dfixed_const(8); + disp_latency_overhead.full = dfixed_div(disp_latency_overhead, sclk_ff); + mc_latency_mclk.full += disp_latency_overhead.full + cur_latency_mclk.full; + mc_latency_sclk.full += disp_latency_overhead.full + cur_latency_sclk.full; + + if (mc_latency_mclk.full > mc_latency_sclk.full) + disp_latency.full = mc_latency_mclk.full; + else + disp_latency.full = mc_latency_sclk.full; + + /* setup Max GRPH_STOP_REQ default value */ + if (ASIC_IS_RV100(rdev)) + max_stop_req = 0x5c; + else + max_stop_req = 0x7c; + + if (mode1) { + /* CRTC1 + Set GRPH_BUFFER_CNTL register using h/w defined optimal values. + GRPH_STOP_REQ <= MIN[ 0x7C, (CRTC_H_DISP + 1) * (bit depth) / 0x10 ] + */ + stop_req = mode1->hdisplay * pixel_bytes1 / 16; + + if (stop_req > max_stop_req) + stop_req = max_stop_req; + + /* + Find the drain rate of the display buffer. + */ + temp_ff.full = dfixed_const((16/pixel_bytes1)); + disp_drain_rate.full = dfixed_div(pix_clk, temp_ff); + + /* + Find the critical point of the display buffer. + */ + crit_point_ff.full = dfixed_mul(disp_drain_rate, disp_latency); + crit_point_ff.full += dfixed_const_half(0); + + critical_point = dfixed_trunc(crit_point_ff); + + if (rdev->disp_priority == 2) { + critical_point = 0; + } + + /* + The critical point should never be above max_stop_req-4. Setting + GRPH_CRITICAL_CNTL = 0 will thus force high priority all the time. + */ + if (max_stop_req - critical_point < 4) + critical_point = 0; + + if (critical_point == 0 && mode2 && rdev->family == CHIP_R300) { + /* some R300 cards have problem with this set to 0, when CRTC2 is enabled.*/ + critical_point = 0x10; + } + + temp = RREG32(RADEON_GRPH_BUFFER_CNTL); + temp &= ~(RADEON_GRPH_STOP_REQ_MASK); + temp |= (stop_req << RADEON_GRPH_STOP_REQ_SHIFT); + temp &= ~(RADEON_GRPH_START_REQ_MASK); + if ((rdev->family == CHIP_R350) && + (stop_req > 0x15)) { + stop_req -= 0x10; + } + temp |= (stop_req << RADEON_GRPH_START_REQ_SHIFT); + temp |= RADEON_GRPH_BUFFER_SIZE; + temp &= ~(RADEON_GRPH_CRITICAL_CNTL | + RADEON_GRPH_CRITICAL_AT_SOF | + RADEON_GRPH_STOP_CNTL); + /* + Write the result into the register. + */ + WREG32(RADEON_GRPH_BUFFER_CNTL, ((temp & ~RADEON_GRPH_CRITICAL_POINT_MASK) | + (critical_point << RADEON_GRPH_CRITICAL_POINT_SHIFT))); + +#if 0 + if ((rdev->family == CHIP_RS400) || + (rdev->family == CHIP_RS480)) { + /* attempt to program RS400 disp regs correctly ??? */ + temp = RREG32(RS400_DISP1_REG_CNTL); + temp &= ~(RS400_DISP1_START_REQ_LEVEL_MASK | + RS400_DISP1_STOP_REQ_LEVEL_MASK); + WREG32(RS400_DISP1_REQ_CNTL1, (temp | + (critical_point << RS400_DISP1_START_REQ_LEVEL_SHIFT) | + (critical_point << RS400_DISP1_STOP_REQ_LEVEL_SHIFT))); + temp = RREG32(RS400_DMIF_MEM_CNTL1); + temp &= ~(RS400_DISP1_CRITICAL_POINT_START_MASK | + RS400_DISP1_CRITICAL_POINT_STOP_MASK); + WREG32(RS400_DMIF_MEM_CNTL1, (temp | + (critical_point << RS400_DISP1_CRITICAL_POINT_START_SHIFT) | + (critical_point << RS400_DISP1_CRITICAL_POINT_STOP_SHIFT))); + } +#endif + + DRM_DEBUG_KMS("GRPH_BUFFER_CNTL from to %x\n", + /* (unsigned int)info->SavedReg->grph_buffer_cntl, */ + (unsigned int)RREG32(RADEON_GRPH_BUFFER_CNTL)); + } + + if (mode2) { + u32 grph2_cntl; + stop_req = mode2->hdisplay * pixel_bytes2 / 16; + + if (stop_req > max_stop_req) + stop_req = max_stop_req; + + /* + Find the drain rate of the display buffer. + */ + temp_ff.full = dfixed_const((16/pixel_bytes2)); + disp_drain_rate2.full = dfixed_div(pix_clk2, temp_ff); + + grph2_cntl = RREG32(RADEON_GRPH2_BUFFER_CNTL); + grph2_cntl &= ~(RADEON_GRPH_STOP_REQ_MASK); + grph2_cntl |= (stop_req << RADEON_GRPH_STOP_REQ_SHIFT); + grph2_cntl &= ~(RADEON_GRPH_START_REQ_MASK); + if ((rdev->family == CHIP_R350) && + (stop_req > 0x15)) { + stop_req -= 0x10; + } + grph2_cntl |= (stop_req << RADEON_GRPH_START_REQ_SHIFT); + grph2_cntl |= RADEON_GRPH_BUFFER_SIZE; + grph2_cntl &= ~(RADEON_GRPH_CRITICAL_CNTL | + RADEON_GRPH_CRITICAL_AT_SOF | + RADEON_GRPH_STOP_CNTL); + + if ((rdev->family == CHIP_RS100) || + (rdev->family == CHIP_RS200)) + critical_point2 = 0; + else { + temp = (rdev->mc.vram_width * rdev->mc.vram_is_ddr + 1)/128; + temp_ff.full = dfixed_const(temp); + temp_ff.full = dfixed_mul(mclk_ff, temp_ff); + if (sclk_ff.full < temp_ff.full) + temp_ff.full = sclk_ff.full; + + read_return_rate.full = temp_ff.full; + + if (mode1) { + temp_ff.full = read_return_rate.full - disp_drain_rate.full; + time_disp1_drop_priority.full = dfixed_div(crit_point_ff, temp_ff); + } else { + time_disp1_drop_priority.full = 0; + } + crit_point_ff.full = disp_latency.full + time_disp1_drop_priority.full + disp_latency.full; + crit_point_ff.full = dfixed_mul(crit_point_ff, disp_drain_rate2); + crit_point_ff.full += dfixed_const_half(0); + + critical_point2 = dfixed_trunc(crit_point_ff); + + if (rdev->disp_priority == 2) { + critical_point2 = 0; + } + + if (max_stop_req - critical_point2 < 4) + critical_point2 = 0; + + } + + if (critical_point2 == 0 && rdev->family == CHIP_R300) { + /* some R300 cards have problem with this set to 0 */ + critical_point2 = 0x10; + } + + WREG32(RADEON_GRPH2_BUFFER_CNTL, ((grph2_cntl & ~RADEON_GRPH_CRITICAL_POINT_MASK) | + (critical_point2 << RADEON_GRPH_CRITICAL_POINT_SHIFT))); + + if ((rdev->family == CHIP_RS400) || + (rdev->family == CHIP_RS480)) { +#if 0 + /* attempt to program RS400 disp2 regs correctly ??? */ + temp = RREG32(RS400_DISP2_REQ_CNTL1); + temp &= ~(RS400_DISP2_START_REQ_LEVEL_MASK | + RS400_DISP2_STOP_REQ_LEVEL_MASK); + WREG32(RS400_DISP2_REQ_CNTL1, (temp | + (critical_point2 << RS400_DISP1_START_REQ_LEVEL_SHIFT) | + (critical_point2 << RS400_DISP1_STOP_REQ_LEVEL_SHIFT))); + temp = RREG32(RS400_DISP2_REQ_CNTL2); + temp &= ~(RS400_DISP2_CRITICAL_POINT_START_MASK | + RS400_DISP2_CRITICAL_POINT_STOP_MASK); + WREG32(RS400_DISP2_REQ_CNTL2, (temp | + (critical_point2 << RS400_DISP2_CRITICAL_POINT_START_SHIFT) | + (critical_point2 << RS400_DISP2_CRITICAL_POINT_STOP_SHIFT))); +#endif + WREG32(RS400_DISP2_REQ_CNTL1, 0x105DC1CC); + WREG32(RS400_DISP2_REQ_CNTL2, 0x2749D000); + WREG32(RS400_DMIF_MEM_CNTL1, 0x29CA71DC); + WREG32(RS400_DISP1_REQ_CNTL1, 0x28FBC3AC); + } + + DRM_DEBUG_KMS("GRPH2_BUFFER_CNTL from to %x\n", + (unsigned int)RREG32(RADEON_GRPH2_BUFFER_CNTL)); + } +} + +static inline void r100_cs_track_texture_print(struct r100_cs_track_texture *t) +{ + DRM_ERROR("pitch %d\n", t->pitch); + DRM_ERROR("use_pitch %d\n", t->use_pitch); + DRM_ERROR("width %d\n", t->width); + DRM_ERROR("width_11 %d\n", t->width_11); + DRM_ERROR("height %d\n", t->height); + DRM_ERROR("height_11 %d\n", t->height_11); + DRM_ERROR("num levels %d\n", t->num_levels); + DRM_ERROR("depth %d\n", t->txdepth); + DRM_ERROR("bpp %d\n", t->cpp); + DRM_ERROR("coordinate type %d\n", t->tex_coord_type); + DRM_ERROR("width round to power of 2 %d\n", t->roundup_w); + DRM_ERROR("height round to power of 2 %d\n", t->roundup_h); + DRM_ERROR("compress format %d\n", t->compress_format); +} + +static int r100_track_compress_size(int compress_format, int w, int h) +{ + int block_width, block_height, block_bytes; + int wblocks, hblocks; + int min_wblocks; + int sz; + + block_width = 4; + block_height = 4; + + switch (compress_format) { + case R100_TRACK_COMP_DXT1: + block_bytes = 8; + min_wblocks = 4; + break; + default: + case R100_TRACK_COMP_DXT35: + block_bytes = 16; + min_wblocks = 2; + break; + } + + hblocks = (h + block_height - 1) / block_height; + wblocks = (w + block_width - 1) / block_width; + if (wblocks < min_wblocks) + wblocks = min_wblocks; + sz = wblocks * hblocks * block_bytes; + return sz; +} + +static int r100_cs_track_cube(struct radeon_device *rdev, + struct r100_cs_track *track, unsigned idx) +{ + unsigned face, w, h; + struct radeon_bo *cube_robj; + unsigned long size; + unsigned compress_format = track->textures[idx].compress_format; + + for (face = 0; face < 5; face++) { + cube_robj = track->textures[idx].cube_info[face].robj; + w = track->textures[idx].cube_info[face].width; + h = track->textures[idx].cube_info[face].height; + + if (compress_format) { + size = r100_track_compress_size(compress_format, w, h); + } else + size = w * h; + size *= track->textures[idx].cpp; + + size += track->textures[idx].cube_info[face].offset; + + if (size > radeon_bo_size(cube_robj)) { + DRM_ERROR("Cube texture offset greater than object size %lu %lu\n", + size, radeon_bo_size(cube_robj)); + r100_cs_track_texture_print(&track->textures[idx]); + return -1; + } + } + return 0; +} + +static int r100_cs_track_texture_check(struct radeon_device *rdev, + struct r100_cs_track *track) +{ + struct radeon_bo *robj; + unsigned long size; + unsigned u, i, w, h, d; + int ret; + + for (u = 0; u < track->num_texture; u++) { + if (!track->textures[u].enabled) + continue; + if (track->textures[u].lookup_disable) + continue; + robj = track->textures[u].robj; + if (robj == NULL) { + DRM_ERROR("No texture bound to unit %u\n", u); + return -EINVAL; + } + size = 0; + for (i = 0; i <= track->textures[u].num_levels; i++) { + if (track->textures[u].use_pitch) { + if (rdev->family < CHIP_R300) + w = (track->textures[u].pitch / track->textures[u].cpp) / (1 << i); + else + w = track->textures[u].pitch / (1 << i); + } else { + w = track->textures[u].width; + if (rdev->family >= CHIP_RV515) + w |= track->textures[u].width_11; + w = w / (1 << i); + if (track->textures[u].roundup_w) + w = roundup_pow_of_two(w); + } + h = track->textures[u].height; + if (rdev->family >= CHIP_RV515) + h |= track->textures[u].height_11; + h = h / (1 << i); + if (track->textures[u].roundup_h) + h = roundup_pow_of_two(h); + if (track->textures[u].tex_coord_type == 1) { + d = (1 << track->textures[u].txdepth) / (1 << i); + if (!d) + d = 1; + } else { + d = 1; + } + if (track->textures[u].compress_format) { + + size += r100_track_compress_size(track->textures[u].compress_format, w, h) * d; + /* compressed textures are block based */ + } else + size += w * h * d; + } + size *= track->textures[u].cpp; + + switch (track->textures[u].tex_coord_type) { + case 0: + case 1: + break; + case 2: + if (track->separate_cube) { + ret = r100_cs_track_cube(rdev, track, u); + if (ret) + return ret; + } else + size *= 6; + break; + default: + DRM_ERROR("Invalid texture coordinate type %u for unit " + "%u\n", track->textures[u].tex_coord_type, u); + return -EINVAL; + } + if (size > radeon_bo_size(robj)) { + DRM_ERROR("Texture of unit %u needs %lu bytes but is " + "%lu\n", u, size, radeon_bo_size(robj)); + r100_cs_track_texture_print(&track->textures[u]); + return -EINVAL; + } + } + return 0; +} + +int r100_cs_track_check(struct radeon_device *rdev, struct r100_cs_track *track) +{ + unsigned i; + unsigned long size; + unsigned prim_walk; + unsigned nverts; + unsigned num_cb = track->cb_dirty ? track->num_cb : 0; + + if (num_cb && !track->zb_cb_clear && !track->color_channel_mask && + !track->blend_read_enable) + num_cb = 0; + + for (i = 0; i < num_cb; i++) { + if (track->cb[i].robj == NULL) { + DRM_ERROR("[drm] No buffer for color buffer %d !\n", i); + return -EINVAL; + } + size = track->cb[i].pitch * track->cb[i].cpp * track->maxy; + size += track->cb[i].offset; + if (size > radeon_bo_size(track->cb[i].robj)) { + DRM_ERROR("[drm] Buffer too small for color buffer %d " + "(need %lu have %lu) !\n", i, size, + radeon_bo_size(track->cb[i].robj)); + DRM_ERROR("[drm] color buffer %d (%u %u %u %u)\n", + i, track->cb[i].pitch, track->cb[i].cpp, + track->cb[i].offset, track->maxy); + return -EINVAL; + } + } + track->cb_dirty = false; + + if (track->zb_dirty && track->z_enabled) { + if (track->zb.robj == NULL) { + DRM_ERROR("[drm] No buffer for z buffer !\n"); + return -EINVAL; + } + size = track->zb.pitch * track->zb.cpp * track->maxy; + size += track->zb.offset; + if (size > radeon_bo_size(track->zb.robj)) { + DRM_ERROR("[drm] Buffer too small for z buffer " + "(need %lu have %lu) !\n", size, + radeon_bo_size(track->zb.robj)); + DRM_ERROR("[drm] zbuffer (%u %u %u %u)\n", + track->zb.pitch, track->zb.cpp, + track->zb.offset, track->maxy); + return -EINVAL; + } + } + track->zb_dirty = false; + + if (track->aa_dirty && track->aaresolve) { + if (track->aa.robj == NULL) { + DRM_ERROR("[drm] No buffer for AA resolve buffer %d !\n", i); + return -EINVAL; + } + /* I believe the format comes from colorbuffer0. */ + size = track->aa.pitch * track->cb[0].cpp * track->maxy; + size += track->aa.offset; + if (size > radeon_bo_size(track->aa.robj)) { + DRM_ERROR("[drm] Buffer too small for AA resolve buffer %d " + "(need %lu have %lu) !\n", i, size, + radeon_bo_size(track->aa.robj)); + DRM_ERROR("[drm] AA resolve buffer %d (%u %u %u %u)\n", + i, track->aa.pitch, track->cb[0].cpp, + track->aa.offset, track->maxy); + return -EINVAL; + } + } + track->aa_dirty = false; + + prim_walk = (track->vap_vf_cntl >> 4) & 0x3; + if (track->vap_vf_cntl & (1 << 14)) { + nverts = track->vap_alt_nverts; + } else { + nverts = (track->vap_vf_cntl >> 16) & 0xFFFF; + } + switch (prim_walk) { + case 1: + for (i = 0; i < track->num_arrays; i++) { + size = track->arrays[i].esize * track->max_indx * 4; + if (track->arrays[i].robj == NULL) { + DRM_ERROR("(PW %u) Vertex array %u no buffer " + "bound\n", prim_walk, i); + return -EINVAL; + } + if (size > radeon_bo_size(track->arrays[i].robj)) { + dev_err(rdev->dev, "(PW %u) Vertex array %u " + "need %lu dwords have %lu dwords\n", + prim_walk, i, size >> 2, + radeon_bo_size(track->arrays[i].robj) + >> 2); + DRM_ERROR("Max indices %u\n", track->max_indx); + return -EINVAL; + } + } + break; + case 2: + for (i = 0; i < track->num_arrays; i++) { + size = track->arrays[i].esize * (nverts - 1) * 4; + if (track->arrays[i].robj == NULL) { + DRM_ERROR("(PW %u) Vertex array %u no buffer " + "bound\n", prim_walk, i); + return -EINVAL; + } + if (size > radeon_bo_size(track->arrays[i].robj)) { + dev_err(rdev->dev, "(PW %u) Vertex array %u " + "need %lu dwords have %lu dwords\n", + prim_walk, i, size >> 2, + radeon_bo_size(track->arrays[i].robj) + >> 2); + return -EINVAL; + } + } + break; + case 3: + size = track->vtx_size * nverts; + if (size != track->immd_dwords) { + DRM_ERROR("IMMD draw %u dwors but needs %lu dwords\n", + track->immd_dwords, size); + DRM_ERROR("VAP_VF_CNTL.NUM_VERTICES %u, VTX_SIZE %u\n", + nverts, track->vtx_size); + return -EINVAL; + } + break; + default: + DRM_ERROR("[drm] Invalid primitive walk %d for VAP_VF_CNTL\n", + prim_walk); + return -EINVAL; + } + + if (track->tex_dirty) { + track->tex_dirty = false; + return r100_cs_track_texture_check(rdev, track); + } + return 0; +} + +void r100_cs_track_clear(struct radeon_device *rdev, struct r100_cs_track *track) +{ + unsigned i, face; + + track->cb_dirty = true; + track->zb_dirty = true; + track->tex_dirty = true; + track->aa_dirty = true; + + if (rdev->family < CHIP_R300) { + track->num_cb = 1; + if (rdev->family <= CHIP_RS200) + track->num_texture = 3; + else + track->num_texture = 6; + track->maxy = 2048; + track->separate_cube = 1; + } else { + track->num_cb = 4; + track->num_texture = 16; + track->maxy = 4096; + track->separate_cube = 0; + track->aaresolve = false; + track->aa.robj = NULL; + } + + for (i = 0; i < track->num_cb; i++) { + track->cb[i].robj = NULL; + track->cb[i].pitch = 8192; + track->cb[i].cpp = 16; + track->cb[i].offset = 0; + } + track->z_enabled = true; + track->zb.robj = NULL; + track->zb.pitch = 8192; + track->zb.cpp = 4; + track->zb.offset = 0; + track->vtx_size = 0x7F; + track->immd_dwords = 0xFFFFFFFFUL; + track->num_arrays = 11; + track->max_indx = 0x00FFFFFFUL; + for (i = 0; i < track->num_arrays; i++) { + track->arrays[i].robj = NULL; + track->arrays[i].esize = 0x7F; + } + for (i = 0; i < track->num_texture; i++) { + track->textures[i].compress_format = R100_TRACK_COMP_NONE; + track->textures[i].pitch = 16536; + track->textures[i].width = 16536; + track->textures[i].height = 16536; + track->textures[i].width_11 = 1 << 11; + track->textures[i].height_11 = 1 << 11; + track->textures[i].num_levels = 12; + if (rdev->family <= CHIP_RS200) { + track->textures[i].tex_coord_type = 0; + track->textures[i].txdepth = 0; + } else { + track->textures[i].txdepth = 16; + track->textures[i].tex_coord_type = 1; + } + track->textures[i].cpp = 64; + track->textures[i].robj = NULL; + /* CS IB emission code makes sure texture unit are disabled */ + track->textures[i].enabled = false; + track->textures[i].lookup_disable = false; + track->textures[i].roundup_w = true; + track->textures[i].roundup_h = true; + if (track->separate_cube) + for (face = 0; face < 5; face++) { + track->textures[i].cube_info[face].robj = NULL; + track->textures[i].cube_info[face].width = 16536; + track->textures[i].cube_info[face].height = 16536; + track->textures[i].cube_info[face].offset = 0; + } + } +} + +int r100_ring_test(struct radeon_device *rdev) +{ + uint32_t scratch; + uint32_t tmp = 0; + unsigned i; + int r; + + r = radeon_scratch_get(rdev, &scratch); + if (r) { + DRM_ERROR("radeon: cp failed to get scratch reg (%d).\n", r); + return r; + } + WREG32(scratch, 0xCAFEDEAD); + r = radeon_ring_lock(rdev, 2); + if (r) { + DRM_ERROR("radeon: cp failed to lock ring (%d).\n", r); + radeon_scratch_free(rdev, scratch); + return r; + } + radeon_ring_write(rdev, PACKET0(scratch, 0)); + radeon_ring_write(rdev, 0xDEADBEEF); + radeon_ring_unlock_commit(rdev); + for (i = 0; i < rdev->usec_timeout; i++) { + tmp = RREG32(scratch); + if (tmp == 0xDEADBEEF) { + break; + } + DRM_UDELAY(1); + } + if (i < rdev->usec_timeout) { + DRM_INFO("ring test succeeded in %d usecs\n", i); + } else { + DRM_ERROR("radeon: ring test failed (scratch(0x%04X)=0x%08X)\n", + scratch, tmp); + r = -EINVAL; + } + radeon_scratch_free(rdev, scratch); + return r; +} + +void r100_ring_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib) +{ + radeon_ring_write(rdev, PACKET0(RADEON_CP_IB_BASE, 1)); + radeon_ring_write(rdev, ib->gpu_addr); + radeon_ring_write(rdev, ib->length_dw); +} + +int r100_ib_test(struct radeon_device *rdev) +{ + struct radeon_ib *ib; + uint32_t scratch; + uint32_t tmp = 0; + unsigned i; + int r; + + r = radeon_scratch_get(rdev, &scratch); + if (r) { + DRM_ERROR("radeon: failed to get scratch reg (%d).\n", r); + return r; + } + WREG32(scratch, 0xCAFEDEAD); + r = radeon_ib_get(rdev, &ib); + if (r) { + return r; + } + ib->ptr[0] = PACKET0(scratch, 0); + ib->ptr[1] = 0xDEADBEEF; + ib->ptr[2] = PACKET2(0); + ib->ptr[3] = PACKET2(0); + ib->ptr[4] = PACKET2(0); + ib->ptr[5] = PACKET2(0); + ib->ptr[6] = PACKET2(0); + ib->ptr[7] = PACKET2(0); + ib->length_dw = 8; + r = radeon_ib_schedule(rdev, ib); + if (r) { + radeon_scratch_free(rdev, scratch); + radeon_ib_free(rdev, &ib); + return r; + } + r = radeon_fence_wait(ib->fence, false); + if (r) { + return r; + } + for (i = 0; i < rdev->usec_timeout; i++) { + tmp = RREG32(scratch); + if (tmp == 0xDEADBEEF) { + break; + } + DRM_UDELAY(1); + } + if (i < rdev->usec_timeout) { + DRM_INFO("ib test succeeded in %u usecs\n", i); + } else { + DRM_ERROR("radeon: ib test failed (scratch(0x%04X)=0x%08X)\n", + scratch, tmp); + r = -EINVAL; + } + radeon_scratch_free(rdev, scratch); + radeon_ib_free(rdev, &ib); + return r; +} + +void r100_ib_fini(struct radeon_device *rdev) +{ + radeon_ib_pool_fini(rdev); +} + +int r100_ib_init(struct radeon_device *rdev) +{ + int r; + + r = radeon_ib_pool_init(rdev); + if (r) { + dev_err(rdev->dev, "failed initializing IB pool (%d).\n", r); + r100_ib_fini(rdev); + return r; + } + r = r100_ib_test(rdev); + if (r) { + dev_err(rdev->dev, "failed testing IB (%d).\n", r); + r100_ib_fini(rdev); + return r; + } + return 0; +} + +void r100_mc_stop(struct radeon_device *rdev, struct r100_mc_save *save) +{ + /* Shutdown CP we shouldn't need to do that but better be safe than + * sorry + */ + rdev->cp.ready = false; + WREG32(R_000740_CP_CSQ_CNTL, 0); + + /* Save few CRTC registers */ + save->GENMO_WT = RREG8(R_0003C2_GENMO_WT); + save->CRTC_EXT_CNTL = RREG32(R_000054_CRTC_EXT_CNTL); + save->CRTC_GEN_CNTL = RREG32(R_000050_CRTC_GEN_CNTL); + save->CUR_OFFSET = RREG32(R_000260_CUR_OFFSET); + if (!(rdev->flags & RADEON_SINGLE_CRTC)) { + save->CRTC2_GEN_CNTL = RREG32(R_0003F8_CRTC2_GEN_CNTL); + save->CUR2_OFFSET = RREG32(R_000360_CUR2_OFFSET); + } + + /* Disable VGA aperture access */ + WREG8(R_0003C2_GENMO_WT, C_0003C2_VGA_RAM_EN & save->GENMO_WT); + /* Disable cursor, overlay, crtc */ + WREG32(R_000260_CUR_OFFSET, save->CUR_OFFSET | S_000260_CUR_LOCK(1)); + WREG32(R_000054_CRTC_EXT_CNTL, save->CRTC_EXT_CNTL | + S_000054_CRTC_DISPLAY_DIS(1)); + WREG32(R_000050_CRTC_GEN_CNTL, + (C_000050_CRTC_CUR_EN & save->CRTC_GEN_CNTL) | + S_000050_CRTC_DISP_REQ_EN_B(1)); + WREG32(R_000420_OV0_SCALE_CNTL, + C_000420_OV0_OVERLAY_EN & RREG32(R_000420_OV0_SCALE_CNTL)); + WREG32(R_000260_CUR_OFFSET, C_000260_CUR_LOCK & save->CUR_OFFSET); + if (!(rdev->flags & RADEON_SINGLE_CRTC)) { + WREG32(R_000360_CUR2_OFFSET, save->CUR2_OFFSET | + S_000360_CUR2_LOCK(1)); + WREG32(R_0003F8_CRTC2_GEN_CNTL, + (C_0003F8_CRTC2_CUR_EN & save->CRTC2_GEN_CNTL) | + S_0003F8_CRTC2_DISPLAY_DIS(1) | + S_0003F8_CRTC2_DISP_REQ_EN_B(1)); + WREG32(R_000360_CUR2_OFFSET, + C_000360_CUR2_LOCK & save->CUR2_OFFSET); + } +} + +void r100_mc_resume(struct radeon_device *rdev, struct r100_mc_save *save) +{ + /* Update base address for crtc */ + WREG32(R_00023C_DISPLAY_BASE_ADDR, rdev->mc.vram_start); + if (!(rdev->flags & RADEON_SINGLE_CRTC)) { + WREG32(R_00033C_CRTC2_DISPLAY_BASE_ADDR, rdev->mc.vram_start); + } + /* Restore CRTC registers */ + WREG8(R_0003C2_GENMO_WT, save->GENMO_WT); + WREG32(R_000054_CRTC_EXT_CNTL, save->CRTC_EXT_CNTL); + WREG32(R_000050_CRTC_GEN_CNTL, save->CRTC_GEN_CNTL); + if (!(rdev->flags & RADEON_SINGLE_CRTC)) { + WREG32(R_0003F8_CRTC2_GEN_CNTL, save->CRTC2_GEN_CNTL); + } +} + +void r100_vga_render_disable(struct radeon_device *rdev) +{ + u32 tmp; + + tmp = RREG8(R_0003C2_GENMO_WT); + WREG8(R_0003C2_GENMO_WT, C_0003C2_VGA_RAM_EN & tmp); +} + +static void r100_debugfs(struct radeon_device *rdev) +{ + int r; + + r = r100_debugfs_mc_info_init(rdev); + if (r) + dev_warn(rdev->dev, "Failed to create r100_mc debugfs file.\n"); +} + +static void r100_mc_program(struct radeon_device *rdev) +{ + struct r100_mc_save save; + + /* Stops all mc clients */ + r100_mc_stop(rdev, &save); + if (rdev->flags & RADEON_IS_AGP) { + WREG32(R_00014C_MC_AGP_LOCATION, + S_00014C_MC_AGP_START(rdev->mc.gtt_start >> 16) | + S_00014C_MC_AGP_TOP(rdev->mc.gtt_end >> 16)); + WREG32(R_000170_AGP_BASE, lower_32_bits(rdev->mc.agp_base)); + if (rdev->family > CHIP_RV200) + WREG32(R_00015C_AGP_BASE_2, + upper_32_bits(rdev->mc.agp_base) & 0xff); + } else { + WREG32(R_00014C_MC_AGP_LOCATION, 0x0FFFFFFF); + WREG32(R_000170_AGP_BASE, 0); + if (rdev->family > CHIP_RV200) + WREG32(R_00015C_AGP_BASE_2, 0); + } + /* Wait for mc idle */ + if (r100_mc_wait_for_idle(rdev)) + dev_warn(rdev->dev, "Wait for MC idle timeout.\n"); + /* Program MC, should be a 32bits limited address space */ + WREG32(R_000148_MC_FB_LOCATION, + S_000148_MC_FB_START(rdev->mc.vram_start >> 16) | + S_000148_MC_FB_TOP(rdev->mc.vram_end >> 16)); + r100_mc_resume(rdev, &save); +} + +void r100_clock_startup(struct radeon_device *rdev) +{ + u32 tmp; + + if (radeon_dynclks != -1 && radeon_dynclks) + radeon_legacy_set_clock_gating(rdev, 1); + /* We need to force on some of the block */ + tmp = RREG32_PLL(R_00000D_SCLK_CNTL); + tmp |= S_00000D_FORCE_CP(1) | S_00000D_FORCE_VIP(1); + if ((rdev->family == CHIP_RV250) || (rdev->family == CHIP_RV280)) + tmp |= S_00000D_FORCE_DISP1(1) | S_00000D_FORCE_DISP2(1); + WREG32_PLL(R_00000D_SCLK_CNTL, tmp); +} + +static int r100_startup(struct radeon_device *rdev) +{ + int r; + + /* set common regs */ + r100_set_common_regs(rdev); + /* program mc */ + r100_mc_program(rdev); + /* Resume clock */ + r100_clock_startup(rdev); + /* Initialize GART (initialize after TTM so we can allocate + * memory through TTM but finalize after TTM) */ + r100_enable_bm(rdev); + if (rdev->flags & RADEON_IS_PCI) { + r = r100_pci_gart_enable(rdev); + if (r) + return r; + } + + /* allocate wb buffer */ + r = radeon_wb_init(rdev); + if (r) + return r; + + /* Enable IRQ */ + r100_irq_set(rdev); + rdev->config.r100.hdp_cntl = RREG32(RADEON_HOST_PATH_CNTL); + /* 1M ring buffer */ + r = r100_cp_init(rdev, 1024 * 1024); + if (r) { + dev_err(rdev->dev, "failed initializing CP (%d).\n", r); + return r; + } + r = r100_ib_init(rdev); + if (r) { + dev_err(rdev->dev, "failed initializing IB (%d).\n", r); + return r; + } + return 0; +} + +int r100_resume(struct radeon_device *rdev) +{ + /* Make sur GART are not working */ + if (rdev->flags & RADEON_IS_PCI) + r100_pci_gart_disable(rdev); + /* Resume clock before doing reset */ + r100_clock_startup(rdev); + /* Reset gpu before posting otherwise ATOM will enter infinite loop */ + if (radeon_asic_reset(rdev)) { + dev_warn(rdev->dev, "GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n", + RREG32(R_000E40_RBBM_STATUS), + RREG32(R_0007C0_CP_STAT)); + } + /* post */ + radeon_combios_asic_init(rdev->ddev); + /* Resume clock after posting */ + r100_clock_startup(rdev); + /* Initialize surface registers */ + radeon_surface_init(rdev); + return r100_startup(rdev); +} + +int r100_suspend(struct radeon_device *rdev) +{ + r100_cp_disable(rdev); + radeon_wb_disable(rdev); + r100_irq_disable(rdev); + if (rdev->flags & RADEON_IS_PCI) + r100_pci_gart_disable(rdev); + return 0; +} + +void r100_fini(struct radeon_device *rdev) +{ + r100_cp_fini(rdev); + radeon_wb_fini(rdev); + r100_ib_fini(rdev); + radeon_gem_fini(rdev); + if (rdev->flags & RADEON_IS_PCI) + r100_pci_gart_fini(rdev); + radeon_agp_fini(rdev); + radeon_irq_kms_fini(rdev); + radeon_fence_driver_fini(rdev); + radeon_bo_fini(rdev); + radeon_atombios_fini(rdev); + kfree(rdev->bios); + rdev->bios = NULL; +} + +/* + * Due to how kexec works, it can leave the hw fully initialised when it + * boots the new kernel. However doing our init sequence with the CP and + * WB stuff setup causes GPU hangs on the RN50 at least. So at startup + * do some quick sanity checks and restore sane values to avoid this + * problem. + */ +void r100_restore_sanity(struct radeon_device *rdev) +{ + u32 tmp; + + tmp = RREG32(RADEON_CP_CSQ_CNTL); + if (tmp) { + WREG32(RADEON_CP_CSQ_CNTL, 0); + } + tmp = RREG32(RADEON_CP_RB_CNTL); + if (tmp) { + WREG32(RADEON_CP_RB_CNTL, 0); + } + tmp = RREG32(RADEON_SCRATCH_UMSK); + if (tmp) { + WREG32(RADEON_SCRATCH_UMSK, 0); + } +} + +int r100_init(struct radeon_device *rdev) +{ + int r; + + /* Register debugfs file specific to this group of asics */ + r100_debugfs(rdev); + /* Disable VGA */ + r100_vga_render_disable(rdev); + /* Initialize scratch registers */ + radeon_scratch_init(rdev); + /* Initialize surface registers */ + radeon_surface_init(rdev); + /* sanity check some register to avoid hangs like after kexec */ + r100_restore_sanity(rdev); + /* TODO: disable VGA need to use VGA request */ + /* BIOS*/ + if (!radeon_get_bios(rdev)) { + if (ASIC_IS_AVIVO(rdev)) + return -EINVAL; + } + if (rdev->is_atom_bios) { + dev_err(rdev->dev, "Expecting combios for RS400/RS480 GPU\n"); + return -EINVAL; + } else { + r = radeon_combios_init(rdev); + if (r) + return r; + } + /* Reset gpu before posting otherwise ATOM will enter infinite loop */ + if (radeon_asic_reset(rdev)) { + dev_warn(rdev->dev, + "GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n", + RREG32(R_000E40_RBBM_STATUS), + RREG32(R_0007C0_CP_STAT)); + } + /* check if cards are posted or not */ + if (radeon_boot_test_post_card(rdev) == false) + return -EINVAL; + /* Set asic errata */ + r100_errata(rdev); + /* Initialize clocks */ + radeon_get_clock_info(rdev->ddev); + /* initialize AGP */ + if (rdev->flags & RADEON_IS_AGP) { + r = radeon_agp_init(rdev); + if (r) { + radeon_agp_disable(rdev); + } + } + /* initialize VRAM */ + r100_mc_init(rdev); + /* Fence driver */ + r = radeon_fence_driver_init(rdev); + if (r) + return r; + r = radeon_irq_kms_init(rdev); + if (r) + return r; + /* Memory manager */ + r = radeon_bo_init(rdev); + if (r) + return r; + if (rdev->flags & RADEON_IS_PCI) { + r = r100_pci_gart_init(rdev); + if (r) + return r; + } + r100_set_safe_registers(rdev); + rdev->accel_working = true; + r = r100_startup(rdev); + if (r) { + /* Somethings want wront with the accel init stop accel */ + dev_err(rdev->dev, "Disabling GPU acceleration\n"); + r100_cp_fini(rdev); + radeon_wb_fini(rdev); + r100_ib_fini(rdev); + radeon_irq_kms_fini(rdev); + if (rdev->flags & RADEON_IS_PCI) + r100_pci_gart_fini(rdev); + rdev->accel_working = false; + } + return 0; +} diff --git a/drivers/gpu/drm/radeon/r100_track.h b/drivers/gpu/drm/radeon/r100_track.h new file mode 100644 index 00000000..686f9dc5 --- /dev/null +++ b/drivers/gpu/drm/radeon/r100_track.h @@ -0,0 +1,197 @@ + +#define R100_TRACK_MAX_TEXTURE 3 +#define R200_TRACK_MAX_TEXTURE 6 +#define R300_TRACK_MAX_TEXTURE 16 + +#define R100_MAX_CB 1 +#define R300_MAX_CB 4 + +/* + * CS functions + */ +struct r100_cs_track_cb { + struct radeon_bo *robj; + unsigned pitch; + unsigned cpp; + unsigned offset; +}; + +struct r100_cs_track_array { + struct radeon_bo *robj; + unsigned esize; +}; + +struct r100_cs_cube_info { + struct radeon_bo *robj; + unsigned offset; + unsigned width; + unsigned height; +}; + +#define R100_TRACK_COMP_NONE 0 +#define R100_TRACK_COMP_DXT1 1 +#define R100_TRACK_COMP_DXT35 2 + +struct r100_cs_track_texture { + struct radeon_bo *robj; + struct r100_cs_cube_info cube_info[5]; /* info for 5 non-primary faces */ + unsigned pitch; + unsigned width; + unsigned height; + unsigned num_levels; + unsigned cpp; + unsigned tex_coord_type; + unsigned txdepth; + unsigned width_11; + unsigned height_11; + bool use_pitch; + bool enabled; + bool lookup_disable; + bool roundup_w; + bool roundup_h; + unsigned compress_format; +}; + +struct r100_cs_track { + unsigned num_cb; + unsigned num_texture; + unsigned maxy; + unsigned vtx_size; + unsigned vap_vf_cntl; + unsigned vap_alt_nverts; + unsigned immd_dwords; + unsigned num_arrays; + unsigned max_indx; + unsigned color_channel_mask; + struct r100_cs_track_array arrays[16]; + struct r100_cs_track_cb cb[R300_MAX_CB]; + struct r100_cs_track_cb zb; + struct r100_cs_track_cb aa; + struct r100_cs_track_texture textures[R300_TRACK_MAX_TEXTURE]; + bool z_enabled; + bool separate_cube; + bool zb_cb_clear; + bool blend_read_enable; + bool cb_dirty; + bool zb_dirty; + bool tex_dirty; + bool aa_dirty; + bool aaresolve; +}; + +int r100_cs_track_check(struct radeon_device *rdev, struct r100_cs_track *track); +void r100_cs_track_clear(struct radeon_device *rdev, struct r100_cs_track *track); +int r100_cs_packet_next_reloc(struct radeon_cs_parser *p, + struct radeon_cs_reloc **cs_reloc); +void r100_cs_dump_packet(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt); + +int r100_cs_packet_parse_vline(struct radeon_cs_parser *p); + +int r200_packet0_check(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt, + unsigned idx, unsigned reg); + + + +static inline int r100_reloc_pitch_offset(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt, + unsigned idx, + unsigned reg) +{ + int r; + u32 tile_flags = 0; + u32 tmp; + struct radeon_cs_reloc *reloc; + u32 value; + + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + value = radeon_get_ib_value(p, idx); + tmp = value & 0x003fffff; + tmp += (((u32)reloc->lobj.gpu_offset) >> 10); + + if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) + tile_flags |= RADEON_DST_TILE_MACRO; + if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) { + if (reg == RADEON_SRC_PITCH_OFFSET) { + DRM_ERROR("Cannot src blit from microtiled surface\n"); + r100_cs_dump_packet(p, pkt); + return -EINVAL; + } + tile_flags |= RADEON_DST_TILE_MICRO; + } + + tmp |= tile_flags; + p->ib->ptr[idx] = (value & 0x3fc00000) | tmp; + return 0; +} + +static inline int r100_packet3_load_vbpntr(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt, + int idx) +{ + unsigned c, i; + struct radeon_cs_reloc *reloc; + struct r100_cs_track *track; + int r = 0; + volatile uint32_t *ib; + u32 idx_value; + + ib = p->ib->ptr; + track = (struct r100_cs_track *)p->track; + c = radeon_get_ib_value(p, idx++) & 0x1F; + if (c > 16) { + DRM_ERROR("Only 16 vertex buffers are allowed %d\n", + pkt->opcode); + r100_cs_dump_packet(p, pkt); + return -EINVAL; + } + track->num_arrays = c; + for (i = 0; i < (c - 1); i+=2, idx+=3) { + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for packet3 %d\n", + pkt->opcode); + r100_cs_dump_packet(p, pkt); + return r; + } + idx_value = radeon_get_ib_value(p, idx); + ib[idx+1] = radeon_get_ib_value(p, idx + 1) + ((u32)reloc->lobj.gpu_offset); + + track->arrays[i + 0].esize = idx_value >> 8; + track->arrays[i + 0].robj = reloc->robj; + track->arrays[i + 0].esize &= 0x7F; + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for packet3 %d\n", + pkt->opcode); + r100_cs_dump_packet(p, pkt); + return r; + } + ib[idx+2] = radeon_get_ib_value(p, idx + 2) + ((u32)reloc->lobj.gpu_offset); + track->arrays[i + 1].robj = reloc->robj; + track->arrays[i + 1].esize = idx_value >> 24; + track->arrays[i + 1].esize &= 0x7F; + } + if (c & 1) { + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for packet3 %d\n", + pkt->opcode); + r100_cs_dump_packet(p, pkt); + return r; + } + idx_value = radeon_get_ib_value(p, idx); + ib[idx+1] = radeon_get_ib_value(p, idx + 1) + ((u32)reloc->lobj.gpu_offset); + track->arrays[i + 0].robj = reloc->robj; + track->arrays[i + 0].esize = idx_value >> 8; + track->arrays[i + 0].esize &= 0x7F; + } + return r; +} diff --git a/drivers/gpu/drm/radeon/r100d.h b/drivers/gpu/drm/radeon/r100d.h new file mode 100644 index 00000000..eab91760 --- /dev/null +++ b/drivers/gpu/drm/radeon/r100d.h @@ -0,0 +1,880 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#ifndef __R100D_H__ +#define __R100D_H__ + +#define CP_PACKET0 0x00000000 +#define PACKET0_BASE_INDEX_SHIFT 0 +#define PACKET0_BASE_INDEX_MASK (0x1ffff << 0) +#define PACKET0_COUNT_SHIFT 16 +#define PACKET0_COUNT_MASK (0x3fff << 16) +#define CP_PACKET1 0x40000000 +#define CP_PACKET2 0x80000000 +#define PACKET2_PAD_SHIFT 0 +#define PACKET2_PAD_MASK (0x3fffffff << 0) +#define CP_PACKET3 0xC0000000 +#define PACKET3_IT_OPCODE_SHIFT 8 +#define PACKET3_IT_OPCODE_MASK (0xff << 8) +#define PACKET3_COUNT_SHIFT 16 +#define PACKET3_COUNT_MASK (0x3fff << 16) +/* PACKET3 op code */ +#define PACKET3_NOP 0x10 +#define PACKET3_3D_DRAW_VBUF 0x28 +#define PACKET3_3D_DRAW_IMMD 0x29 +#define PACKET3_3D_DRAW_INDX 0x2A +#define PACKET3_3D_LOAD_VBPNTR 0x2F +#define PACKET3_3D_CLEAR_ZMASK 0x32 +#define PACKET3_INDX_BUFFER 0x33 +#define PACKET3_3D_DRAW_VBUF_2 0x34 +#define PACKET3_3D_DRAW_IMMD_2 0x35 +#define PACKET3_3D_DRAW_INDX_2 0x36 +#define PACKET3_3D_CLEAR_HIZ 0x37 +#define PACKET3_BITBLT_MULTI 0x9B + +#define PACKET0(reg, n) (CP_PACKET0 | \ + REG_SET(PACKET0_BASE_INDEX, (reg) >> 2) | \ + REG_SET(PACKET0_COUNT, (n))) +#define PACKET2(v) (CP_PACKET2 | REG_SET(PACKET2_PAD, (v))) +#define PACKET3(op, n) (CP_PACKET3 | \ + REG_SET(PACKET3_IT_OPCODE, (op)) | \ + REG_SET(PACKET3_COUNT, (n))) + +#define PACKET_TYPE0 0 +#define PACKET_TYPE1 1 +#define PACKET_TYPE2 2 +#define PACKET_TYPE3 3 + +#define CP_PACKET_GET_TYPE(h) (((h) >> 30) & 3) +#define CP_PACKET_GET_COUNT(h) (((h) >> 16) & 0x3FFF) +#define CP_PACKET0_GET_REG(h) (((h) & 0x1FFF) << 2) +#define CP_PACKET0_GET_ONE_REG_WR(h) (((h) >> 15) & 1) +#define CP_PACKET3_GET_OPCODE(h) (((h) >> 8) & 0xFF) + +/* Registers */ +#define R_0000F0_RBBM_SOFT_RESET 0x0000F0 +#define S_0000F0_SOFT_RESET_CP(x) (((x) & 0x1) << 0) +#define G_0000F0_SOFT_RESET_CP(x) (((x) >> 0) & 0x1) +#define C_0000F0_SOFT_RESET_CP 0xFFFFFFFE +#define S_0000F0_SOFT_RESET_HI(x) (((x) & 0x1) << 1) +#define G_0000F0_SOFT_RESET_HI(x) (((x) >> 1) & 0x1) +#define C_0000F0_SOFT_RESET_HI 0xFFFFFFFD +#define S_0000F0_SOFT_RESET_SE(x) (((x) & 0x1) << 2) +#define G_0000F0_SOFT_RESET_SE(x) (((x) >> 2) & 0x1) +#define C_0000F0_SOFT_RESET_SE 0xFFFFFFFB +#define S_0000F0_SOFT_RESET_RE(x) (((x) & 0x1) << 3) +#define G_0000F0_SOFT_RESET_RE(x) (((x) >> 3) & 0x1) +#define C_0000F0_SOFT_RESET_RE 0xFFFFFFF7 +#define S_0000F0_SOFT_RESET_PP(x) (((x) & 0x1) << 4) +#define G_0000F0_SOFT_RESET_PP(x) (((x) >> 4) & 0x1) +#define C_0000F0_SOFT_RESET_PP 0xFFFFFFEF +#define S_0000F0_SOFT_RESET_E2(x) (((x) & 0x1) << 5) +#define G_0000F0_SOFT_RESET_E2(x) (((x) >> 5) & 0x1) +#define C_0000F0_SOFT_RESET_E2 0xFFFFFFDF +#define S_0000F0_SOFT_RESET_RB(x) (((x) & 0x1) << 6) +#define G_0000F0_SOFT_RESET_RB(x) (((x) >> 6) & 0x1) +#define C_0000F0_SOFT_RESET_RB 0xFFFFFFBF +#define S_0000F0_SOFT_RESET_HDP(x) (((x) & 0x1) << 7) +#define G_0000F0_SOFT_RESET_HDP(x) (((x) >> 7) & 0x1) +#define C_0000F0_SOFT_RESET_HDP 0xFFFFFF7F +#define S_0000F0_SOFT_RESET_MC(x) (((x) & 0x1) << 8) +#define G_0000F0_SOFT_RESET_MC(x) (((x) >> 8) & 0x1) +#define C_0000F0_SOFT_RESET_MC 0xFFFFFEFF +#define S_0000F0_SOFT_RESET_AIC(x) (((x) & 0x1) << 9) +#define G_0000F0_SOFT_RESET_AIC(x) (((x) >> 9) & 0x1) +#define C_0000F0_SOFT_RESET_AIC 0xFFFFFDFF +#define S_0000F0_SOFT_RESET_VIP(x) (((x) & 0x1) << 10) +#define G_0000F0_SOFT_RESET_VIP(x) (((x) >> 10) & 0x1) +#define C_0000F0_SOFT_RESET_VIP 0xFFFFFBFF +#define S_0000F0_SOFT_RESET_DISP(x) (((x) & 0x1) << 11) +#define G_0000F0_SOFT_RESET_DISP(x) (((x) >> 11) & 0x1) +#define C_0000F0_SOFT_RESET_DISP 0xFFFFF7FF +#define S_0000F0_SOFT_RESET_CG(x) (((x) & 0x1) << 12) +#define G_0000F0_SOFT_RESET_CG(x) (((x) >> 12) & 0x1) +#define C_0000F0_SOFT_RESET_CG 0xFFFFEFFF +#define R_000030_BUS_CNTL 0x000030 +#define S_000030_BUS_DBL_RESYNC(x) (((x) & 0x1) << 0) +#define G_000030_BUS_DBL_RESYNC(x) (((x) >> 0) & 0x1) +#define C_000030_BUS_DBL_RESYNC 0xFFFFFFFE +#define S_000030_BUS_MSTR_RESET(x) (((x) & 0x1) << 1) +#define G_000030_BUS_MSTR_RESET(x) (((x) >> 1) & 0x1) +#define C_000030_BUS_MSTR_RESET 0xFFFFFFFD +#define S_000030_BUS_FLUSH_BUF(x) (((x) & 0x1) << 2) +#define G_000030_BUS_FLUSH_BUF(x) (((x) >> 2) & 0x1) +#define C_000030_BUS_FLUSH_BUF 0xFFFFFFFB +#define S_000030_BUS_STOP_REQ_DIS(x) (((x) & 0x1) << 3) +#define G_000030_BUS_STOP_REQ_DIS(x) (((x) >> 3) & 0x1) +#define C_000030_BUS_STOP_REQ_DIS 0xFFFFFFF7 +#define S_000030_BUS_PM4_READ_COMBINE_EN(x) (((x) & 0x1) << 4) +#define G_000030_BUS_PM4_READ_COMBINE_EN(x) (((x) >> 4) & 0x1) +#define C_000030_BUS_PM4_READ_COMBINE_EN 0xFFFFFFEF +#define S_000030_BUS_WRT_COMBINE_EN(x) (((x) & 0x1) << 5) +#define G_000030_BUS_WRT_COMBINE_EN(x) (((x) >> 5) & 0x1) +#define C_000030_BUS_WRT_COMBINE_EN 0xFFFFFFDF +#define S_000030_BUS_MASTER_DIS(x) (((x) & 0x1) << 6) +#define G_000030_BUS_MASTER_DIS(x) (((x) >> 6) & 0x1) +#define C_000030_BUS_MASTER_DIS 0xFFFFFFBF +#define S_000030_BIOS_ROM_WRT_EN(x) (((x) & 0x1) << 7) +#define G_000030_BIOS_ROM_WRT_EN(x) (((x) >> 7) & 0x1) +#define C_000030_BIOS_ROM_WRT_EN 0xFFFFFF7F +#define S_000030_BM_DAC_CRIPPLE(x) (((x) & 0x1) << 8) +#define G_000030_BM_DAC_CRIPPLE(x) (((x) >> 8) & 0x1) +#define C_000030_BM_DAC_CRIPPLE 0xFFFFFEFF +#define S_000030_BUS_NON_PM4_READ_COMBINE_EN(x) (((x) & 0x1) << 9) +#define G_000030_BUS_NON_PM4_READ_COMBINE_EN(x) (((x) >> 9) & 0x1) +#define C_000030_BUS_NON_PM4_READ_COMBINE_EN 0xFFFFFDFF +#define S_000030_BUS_XFERD_DISCARD_EN(x) (((x) & 0x1) << 10) +#define G_000030_BUS_XFERD_DISCARD_EN(x) (((x) >> 10) & 0x1) +#define C_000030_BUS_XFERD_DISCARD_EN 0xFFFFFBFF +#define S_000030_BUS_SGL_READ_DISABLE(x) (((x) & 0x1) << 11) +#define G_000030_BUS_SGL_READ_DISABLE(x) (((x) >> 11) & 0x1) +#define C_000030_BUS_SGL_READ_DISABLE 0xFFFFF7FF +#define S_000030_BIOS_DIS_ROM(x) (((x) & 0x1) << 12) +#define G_000030_BIOS_DIS_ROM(x) (((x) >> 12) & 0x1) +#define C_000030_BIOS_DIS_ROM 0xFFFFEFFF +#define S_000030_BUS_PCI_READ_RETRY_EN(x) (((x) & 0x1) << 13) +#define G_000030_BUS_PCI_READ_RETRY_EN(x) (((x) >> 13) & 0x1) +#define C_000030_BUS_PCI_READ_RETRY_EN 0xFFFFDFFF +#define S_000030_BUS_AGP_AD_STEPPING_EN(x) (((x) & 0x1) << 14) +#define G_000030_BUS_AGP_AD_STEPPING_EN(x) (((x) >> 14) & 0x1) +#define C_000030_BUS_AGP_AD_STEPPING_EN 0xFFFFBFFF +#define S_000030_BUS_PCI_WRT_RETRY_EN(x) (((x) & 0x1) << 15) +#define G_000030_BUS_PCI_WRT_RETRY_EN(x) (((x) >> 15) & 0x1) +#define C_000030_BUS_PCI_WRT_RETRY_EN 0xFFFF7FFF +#define S_000030_BUS_RETRY_WS(x) (((x) & 0xF) << 16) +#define G_000030_BUS_RETRY_WS(x) (((x) >> 16) & 0xF) +#define C_000030_BUS_RETRY_WS 0xFFF0FFFF +#define S_000030_BUS_MSTR_RD_MULT(x) (((x) & 0x1) << 20) +#define G_000030_BUS_MSTR_RD_MULT(x) (((x) >> 20) & 0x1) +#define C_000030_BUS_MSTR_RD_MULT 0xFFEFFFFF +#define S_000030_BUS_MSTR_RD_LINE(x) (((x) & 0x1) << 21) +#define G_000030_BUS_MSTR_RD_LINE(x) (((x) >> 21) & 0x1) +#define C_000030_BUS_MSTR_RD_LINE 0xFFDFFFFF +#define S_000030_BUS_SUSPEND(x) (((x) & 0x1) << 22) +#define G_000030_BUS_SUSPEND(x) (((x) >> 22) & 0x1) +#define C_000030_BUS_SUSPEND 0xFFBFFFFF +#define S_000030_LAT_16X(x) (((x) & 0x1) << 23) +#define G_000030_LAT_16X(x) (((x) >> 23) & 0x1) +#define C_000030_LAT_16X 0xFF7FFFFF +#define S_000030_BUS_RD_DISCARD_EN(x) (((x) & 0x1) << 24) +#define G_000030_BUS_RD_DISCARD_EN(x) (((x) >> 24) & 0x1) +#define C_000030_BUS_RD_DISCARD_EN 0xFEFFFFFF +#define S_000030_ENFRCWRDY(x) (((x) & 0x1) << 25) +#define G_000030_ENFRCWRDY(x) (((x) >> 25) & 0x1) +#define C_000030_ENFRCWRDY 0xFDFFFFFF +#define S_000030_BUS_MSTR_WS(x) (((x) & 0x1) << 26) +#define G_000030_BUS_MSTR_WS(x) (((x) >> 26) & 0x1) +#define C_000030_BUS_MSTR_WS 0xFBFFFFFF +#define S_000030_BUS_PARKING_DIS(x) (((x) & 0x1) << 27) +#define G_000030_BUS_PARKING_DIS(x) (((x) >> 27) & 0x1) +#define C_000030_BUS_PARKING_DIS 0xF7FFFFFF +#define S_000030_BUS_MSTR_DISCONNECT_EN(x) (((x) & 0x1) << 28) +#define G_000030_BUS_MSTR_DISCONNECT_EN(x) (((x) >> 28) & 0x1) +#define C_000030_BUS_MSTR_DISCONNECT_EN 0xEFFFFFFF +#define S_000030_SERR_EN(x) (((x) & 0x1) << 29) +#define G_000030_SERR_EN(x) (((x) >> 29) & 0x1) +#define C_000030_SERR_EN 0xDFFFFFFF +#define S_000030_BUS_READ_BURST(x) (((x) & 0x1) << 30) +#define G_000030_BUS_READ_BURST(x) (((x) >> 30) & 0x1) +#define C_000030_BUS_READ_BURST 0xBFFFFFFF +#define S_000030_BUS_RDY_READ_DLY(x) (((x) & 0x1) << 31) +#define G_000030_BUS_RDY_READ_DLY(x) (((x) >> 31) & 0x1) +#define C_000030_BUS_RDY_READ_DLY 0x7FFFFFFF +#define R_000040_GEN_INT_CNTL 0x000040 +#define S_000040_CRTC_VBLANK(x) (((x) & 0x1) << 0) +#define G_000040_CRTC_VBLANK(x) (((x) >> 0) & 0x1) +#define C_000040_CRTC_VBLANK 0xFFFFFFFE +#define S_000040_CRTC_VLINE(x) (((x) & 0x1) << 1) +#define G_000040_CRTC_VLINE(x) (((x) >> 1) & 0x1) +#define C_000040_CRTC_VLINE 0xFFFFFFFD +#define S_000040_CRTC_VSYNC(x) (((x) & 0x1) << 2) +#define G_000040_CRTC_VSYNC(x) (((x) >> 2) & 0x1) +#define C_000040_CRTC_VSYNC 0xFFFFFFFB +#define S_000040_SNAPSHOT(x) (((x) & 0x1) << 3) +#define G_000040_SNAPSHOT(x) (((x) >> 3) & 0x1) +#define C_000040_SNAPSHOT 0xFFFFFFF7 +#define S_000040_FP_DETECT(x) (((x) & 0x1) << 4) +#define G_000040_FP_DETECT(x) (((x) >> 4) & 0x1) +#define C_000040_FP_DETECT 0xFFFFFFEF +#define S_000040_CRTC2_VLINE(x) (((x) & 0x1) << 5) +#define G_000040_CRTC2_VLINE(x) (((x) >> 5) & 0x1) +#define C_000040_CRTC2_VLINE 0xFFFFFFDF +#define S_000040_DMA_VIPH0_INT_EN(x) (((x) & 0x1) << 12) +#define G_000040_DMA_VIPH0_INT_EN(x) (((x) >> 12) & 0x1) +#define C_000040_DMA_VIPH0_INT_EN 0xFFFFEFFF +#define S_000040_CRTC2_VSYNC(x) (((x) & 0x1) << 6) +#define G_000040_CRTC2_VSYNC(x) (((x) >> 6) & 0x1) +#define C_000040_CRTC2_VSYNC 0xFFFFFFBF +#define S_000040_SNAPSHOT2(x) (((x) & 0x1) << 7) +#define G_000040_SNAPSHOT2(x) (((x) >> 7) & 0x1) +#define C_000040_SNAPSHOT2 0xFFFFFF7F +#define S_000040_CRTC2_VBLANK(x) (((x) & 0x1) << 9) +#define G_000040_CRTC2_VBLANK(x) (((x) >> 9) & 0x1) +#define C_000040_CRTC2_VBLANK 0xFFFFFDFF +#define S_000040_FP2_DETECT(x) (((x) & 0x1) << 10) +#define G_000040_FP2_DETECT(x) (((x) >> 10) & 0x1) +#define C_000040_FP2_DETECT 0xFFFFFBFF +#define S_000040_VSYNC_DIFF_OVER_LIMIT(x) (((x) & 0x1) << 11) +#define G_000040_VSYNC_DIFF_OVER_LIMIT(x) (((x) >> 11) & 0x1) +#define C_000040_VSYNC_DIFF_OVER_LIMIT 0xFFFFF7FF +#define S_000040_DMA_VIPH1_INT_EN(x) (((x) & 0x1) << 13) +#define G_000040_DMA_VIPH1_INT_EN(x) (((x) >> 13) & 0x1) +#define C_000040_DMA_VIPH1_INT_EN 0xFFFFDFFF +#define S_000040_DMA_VIPH2_INT_EN(x) (((x) & 0x1) << 14) +#define G_000040_DMA_VIPH2_INT_EN(x) (((x) >> 14) & 0x1) +#define C_000040_DMA_VIPH2_INT_EN 0xFFFFBFFF +#define S_000040_DMA_VIPH3_INT_EN(x) (((x) & 0x1) << 15) +#define G_000040_DMA_VIPH3_INT_EN(x) (((x) >> 15) & 0x1) +#define C_000040_DMA_VIPH3_INT_EN 0xFFFF7FFF +#define S_000040_I2C_INT_EN(x) (((x) & 0x1) << 17) +#define G_000040_I2C_INT_EN(x) (((x) >> 17) & 0x1) +#define C_000040_I2C_INT_EN 0xFFFDFFFF +#define S_000040_GUI_IDLE(x) (((x) & 0x1) << 19) +#define G_000040_GUI_IDLE(x) (((x) >> 19) & 0x1) +#define C_000040_GUI_IDLE 0xFFF7FFFF +#define S_000040_VIPH_INT_EN(x) (((x) & 0x1) << 24) +#define G_000040_VIPH_INT_EN(x) (((x) >> 24) & 0x1) +#define C_000040_VIPH_INT_EN 0xFEFFFFFF +#define S_000040_SW_INT_EN(x) (((x) & 0x1) << 25) +#define G_000040_SW_INT_EN(x) (((x) >> 25) & 0x1) +#define C_000040_SW_INT_EN 0xFDFFFFFF +#define S_000040_GEYSERVILLE(x) (((x) & 0x1) << 27) +#define G_000040_GEYSERVILLE(x) (((x) >> 27) & 0x1) +#define C_000040_GEYSERVILLE 0xF7FFFFFF +#define S_000040_HDCP_AUTHORIZED_INT(x) (((x) & 0x1) << 28) +#define G_000040_HDCP_AUTHORIZED_INT(x) (((x) >> 28) & 0x1) +#define C_000040_HDCP_AUTHORIZED_INT 0xEFFFFFFF +#define S_000040_DVI_I2C_INT(x) (((x) & 0x1) << 29) +#define G_000040_DVI_I2C_INT(x) (((x) >> 29) & 0x1) +#define C_000040_DVI_I2C_INT 0xDFFFFFFF +#define S_000040_GUIDMA(x) (((x) & 0x1) << 30) +#define G_000040_GUIDMA(x) (((x) >> 30) & 0x1) +#define C_000040_GUIDMA 0xBFFFFFFF +#define S_000040_VIDDMA(x) (((x) & 0x1) << 31) +#define G_000040_VIDDMA(x) (((x) >> 31) & 0x1) +#define C_000040_VIDDMA 0x7FFFFFFF +#define R_000044_GEN_INT_STATUS 0x000044 +#define S_000044_CRTC_VBLANK_STAT(x) (((x) & 0x1) << 0) +#define G_000044_CRTC_VBLANK_STAT(x) (((x) >> 0) & 0x1) +#define C_000044_CRTC_VBLANK_STAT 0xFFFFFFFE +#define S_000044_CRTC_VBLANK_STAT_AK(x) (((x) & 0x1) << 0) +#define G_000044_CRTC_VBLANK_STAT_AK(x) (((x) >> 0) & 0x1) +#define C_000044_CRTC_VBLANK_STAT_AK 0xFFFFFFFE +#define S_000044_CRTC_VLINE_STAT(x) (((x) & 0x1) << 1) +#define G_000044_CRTC_VLINE_STAT(x) (((x) >> 1) & 0x1) +#define C_000044_CRTC_VLINE_STAT 0xFFFFFFFD +#define S_000044_CRTC_VLINE_STAT_AK(x) (((x) & 0x1) << 1) +#define G_000044_CRTC_VLINE_STAT_AK(x) (((x) >> 1) & 0x1) +#define C_000044_CRTC_VLINE_STAT_AK 0xFFFFFFFD +#define S_000044_CRTC_VSYNC_STAT(x) (((x) & 0x1) << 2) +#define G_000044_CRTC_VSYNC_STAT(x) (((x) >> 2) & 0x1) +#define C_000044_CRTC_VSYNC_STAT 0xFFFFFFFB +#define S_000044_CRTC_VSYNC_STAT_AK(x) (((x) & 0x1) << 2) +#define G_000044_CRTC_VSYNC_STAT_AK(x) (((x) >> 2) & 0x1) +#define C_000044_CRTC_VSYNC_STAT_AK 0xFFFFFFFB +#define S_000044_SNAPSHOT_STAT(x) (((x) & 0x1) << 3) +#define G_000044_SNAPSHOT_STAT(x) (((x) >> 3) & 0x1) +#define C_000044_SNAPSHOT_STAT 0xFFFFFFF7 +#define S_000044_SNAPSHOT_STAT_AK(x) (((x) & 0x1) << 3) +#define G_000044_SNAPSHOT_STAT_AK(x) (((x) >> 3) & 0x1) +#define C_000044_SNAPSHOT_STAT_AK 0xFFFFFFF7 +#define S_000044_FP_DETECT_STAT(x) (((x) & 0x1) << 4) +#define G_000044_FP_DETECT_STAT(x) (((x) >> 4) & 0x1) +#define C_000044_FP_DETECT_STAT 0xFFFFFFEF +#define S_000044_FP_DETECT_STAT_AK(x) (((x) & 0x1) << 4) +#define G_000044_FP_DETECT_STAT_AK(x) (((x) >> 4) & 0x1) +#define C_000044_FP_DETECT_STAT_AK 0xFFFFFFEF +#define S_000044_CRTC2_VLINE_STAT(x) (((x) & 0x1) << 5) +#define G_000044_CRTC2_VLINE_STAT(x) (((x) >> 5) & 0x1) +#define C_000044_CRTC2_VLINE_STAT 0xFFFFFFDF +#define S_000044_CRTC2_VLINE_STAT_AK(x) (((x) & 0x1) << 5) +#define G_000044_CRTC2_VLINE_STAT_AK(x) (((x) >> 5) & 0x1) +#define C_000044_CRTC2_VLINE_STAT_AK 0xFFFFFFDF +#define S_000044_CRTC2_VSYNC_STAT(x) (((x) & 0x1) << 6) +#define G_000044_CRTC2_VSYNC_STAT(x) (((x) >> 6) & 0x1) +#define C_000044_CRTC2_VSYNC_STAT 0xFFFFFFBF +#define S_000044_CRTC2_VSYNC_STAT_AK(x) (((x) & 0x1) << 6) +#define G_000044_CRTC2_VSYNC_STAT_AK(x) (((x) >> 6) & 0x1) +#define C_000044_CRTC2_VSYNC_STAT_AK 0xFFFFFFBF +#define S_000044_SNAPSHOT2_STAT(x) (((x) & 0x1) << 7) +#define G_000044_SNAPSHOT2_STAT(x) (((x) >> 7) & 0x1) +#define C_000044_SNAPSHOT2_STAT 0xFFFFFF7F +#define S_000044_SNAPSHOT2_STAT_AK(x) (((x) & 0x1) << 7) +#define G_000044_SNAPSHOT2_STAT_AK(x) (((x) >> 7) & 0x1) +#define C_000044_SNAPSHOT2_STAT_AK 0xFFFFFF7F +#define S_000044_CAP0_INT_ACTIVE(x) (((x) & 0x1) << 8) +#define G_000044_CAP0_INT_ACTIVE(x) (((x) >> 8) & 0x1) +#define C_000044_CAP0_INT_ACTIVE 0xFFFFFEFF +#define S_000044_CRTC2_VBLANK_STAT(x) (((x) & 0x1) << 9) +#define G_000044_CRTC2_VBLANK_STAT(x) (((x) >> 9) & 0x1) +#define C_000044_CRTC2_VBLANK_STAT 0xFFFFFDFF +#define S_000044_CRTC2_VBLANK_STAT_AK(x) (((x) & 0x1) << 9) +#define G_000044_CRTC2_VBLANK_STAT_AK(x) (((x) >> 9) & 0x1) +#define C_000044_CRTC2_VBLANK_STAT_AK 0xFFFFFDFF +#define S_000044_FP2_DETECT_STAT(x) (((x) & 0x1) << 10) +#define G_000044_FP2_DETECT_STAT(x) (((x) >> 10) & 0x1) +#define C_000044_FP2_DETECT_STAT 0xFFFFFBFF +#define S_000044_FP2_DETECT_STAT_AK(x) (((x) & 0x1) << 10) +#define G_000044_FP2_DETECT_STAT_AK(x) (((x) >> 10) & 0x1) +#define C_000044_FP2_DETECT_STAT_AK 0xFFFFFBFF +#define S_000044_VSYNC_DIFF_OVER_LIMIT_STAT(x) (((x) & 0x1) << 11) +#define G_000044_VSYNC_DIFF_OVER_LIMIT_STAT(x) (((x) >> 11) & 0x1) +#define C_000044_VSYNC_DIFF_OVER_LIMIT_STAT 0xFFFFF7FF +#define S_000044_VSYNC_DIFF_OVER_LIMIT_STAT_AK(x) (((x) & 0x1) << 11) +#define G_000044_VSYNC_DIFF_OVER_LIMIT_STAT_AK(x) (((x) >> 11) & 0x1) +#define C_000044_VSYNC_DIFF_OVER_LIMIT_STAT_AK 0xFFFFF7FF +#define S_000044_DMA_VIPH0_INT(x) (((x) & 0x1) << 12) +#define G_000044_DMA_VIPH0_INT(x) (((x) >> 12) & 0x1) +#define C_000044_DMA_VIPH0_INT 0xFFFFEFFF +#define S_000044_DMA_VIPH0_INT_AK(x) (((x) & 0x1) << 12) +#define G_000044_DMA_VIPH0_INT_AK(x) (((x) >> 12) & 0x1) +#define C_000044_DMA_VIPH0_INT_AK 0xFFFFEFFF +#define S_000044_DMA_VIPH1_INT(x) (((x) & 0x1) << 13) +#define G_000044_DMA_VIPH1_INT(x) (((x) >> 13) & 0x1) +#define C_000044_DMA_VIPH1_INT 0xFFFFDFFF +#define S_000044_DMA_VIPH1_INT_AK(x) (((x) & 0x1) << 13) +#define G_000044_DMA_VIPH1_INT_AK(x) (((x) >> 13) & 0x1) +#define C_000044_DMA_VIPH1_INT_AK 0xFFFFDFFF +#define S_000044_DMA_VIPH2_INT(x) (((x) & 0x1) << 14) +#define G_000044_DMA_VIPH2_INT(x) (((x) >> 14) & 0x1) +#define C_000044_DMA_VIPH2_INT 0xFFFFBFFF +#define S_000044_DMA_VIPH2_INT_AK(x) (((x) & 0x1) << 14) +#define G_000044_DMA_VIPH2_INT_AK(x) (((x) >> 14) & 0x1) +#define C_000044_DMA_VIPH2_INT_AK 0xFFFFBFFF +#define S_000044_DMA_VIPH3_INT(x) (((x) & 0x1) << 15) +#define G_000044_DMA_VIPH3_INT(x) (((x) >> 15) & 0x1) +#define C_000044_DMA_VIPH3_INT 0xFFFF7FFF +#define S_000044_DMA_VIPH3_INT_AK(x) (((x) & 0x1) << 15) +#define G_000044_DMA_VIPH3_INT_AK(x) (((x) >> 15) & 0x1) +#define C_000044_DMA_VIPH3_INT_AK 0xFFFF7FFF +#define S_000044_I2C_INT(x) (((x) & 0x1) << 17) +#define G_000044_I2C_INT(x) (((x) >> 17) & 0x1) +#define C_000044_I2C_INT 0xFFFDFFFF +#define S_000044_I2C_INT_AK(x) (((x) & 0x1) << 17) +#define G_000044_I2C_INT_AK(x) (((x) >> 17) & 0x1) +#define C_000044_I2C_INT_AK 0xFFFDFFFF +#define S_000044_GUI_IDLE_STAT(x) (((x) & 0x1) << 19) +#define G_000044_GUI_IDLE_STAT(x) (((x) >> 19) & 0x1) +#define C_000044_GUI_IDLE_STAT 0xFFF7FFFF +#define S_000044_GUI_IDLE_STAT_AK(x) (((x) & 0x1) << 19) +#define G_000044_GUI_IDLE_STAT_AK(x) (((x) >> 19) & 0x1) +#define C_000044_GUI_IDLE_STAT_AK 0xFFF7FFFF +#define S_000044_VIPH_INT(x) (((x) & 0x1) << 24) +#define G_000044_VIPH_INT(x) (((x) >> 24) & 0x1) +#define C_000044_VIPH_INT 0xFEFFFFFF +#define S_000044_SW_INT(x) (((x) & 0x1) << 25) +#define G_000044_SW_INT(x) (((x) >> 25) & 0x1) +#define C_000044_SW_INT 0xFDFFFFFF +#define S_000044_SW_INT_AK(x) (((x) & 0x1) << 25) +#define G_000044_SW_INT_AK(x) (((x) >> 25) & 0x1) +#define C_000044_SW_INT_AK 0xFDFFFFFF +#define S_000044_SW_INT_SET(x) (((x) & 0x1) << 26) +#define G_000044_SW_INT_SET(x) (((x) >> 26) & 0x1) +#define C_000044_SW_INT_SET 0xFBFFFFFF +#define S_000044_GEYSERVILLE_STAT(x) (((x) & 0x1) << 27) +#define G_000044_GEYSERVILLE_STAT(x) (((x) >> 27) & 0x1) +#define C_000044_GEYSERVILLE_STAT 0xF7FFFFFF +#define S_000044_GEYSERVILLE_STAT_AK(x) (((x) & 0x1) << 27) +#define G_000044_GEYSERVILLE_STAT_AK(x) (((x) >> 27) & 0x1) +#define C_000044_GEYSERVILLE_STAT_AK 0xF7FFFFFF +#define S_000044_HDCP_AUTHORIZED_INT_STAT(x) (((x) & 0x1) << 28) +#define G_000044_HDCP_AUTHORIZED_INT_STAT(x) (((x) >> 28) & 0x1) +#define C_000044_HDCP_AUTHORIZED_INT_STAT 0xEFFFFFFF +#define S_000044_HDCP_AUTHORIZED_INT_AK(x) (((x) & 0x1) << 28) +#define G_000044_HDCP_AUTHORIZED_INT_AK(x) (((x) >> 28) & 0x1) +#define C_000044_HDCP_AUTHORIZED_INT_AK 0xEFFFFFFF +#define S_000044_DVI_I2C_INT_STAT(x) (((x) & 0x1) << 29) +#define G_000044_DVI_I2C_INT_STAT(x) (((x) >> 29) & 0x1) +#define C_000044_DVI_I2C_INT_STAT 0xDFFFFFFF +#define S_000044_DVI_I2C_INT_AK(x) (((x) & 0x1) << 29) +#define G_000044_DVI_I2C_INT_AK(x) (((x) >> 29) & 0x1) +#define C_000044_DVI_I2C_INT_AK 0xDFFFFFFF +#define S_000044_GUIDMA_STAT(x) (((x) & 0x1) << 30) +#define G_000044_GUIDMA_STAT(x) (((x) >> 30) & 0x1) +#define C_000044_GUIDMA_STAT 0xBFFFFFFF +#define S_000044_GUIDMA_AK(x) (((x) & 0x1) << 30) +#define G_000044_GUIDMA_AK(x) (((x) >> 30) & 0x1) +#define C_000044_GUIDMA_AK 0xBFFFFFFF +#define S_000044_VIDDMA_STAT(x) (((x) & 0x1) << 31) +#define G_000044_VIDDMA_STAT(x) (((x) >> 31) & 0x1) +#define C_000044_VIDDMA_STAT 0x7FFFFFFF +#define S_000044_VIDDMA_AK(x) (((x) & 0x1) << 31) +#define G_000044_VIDDMA_AK(x) (((x) >> 31) & 0x1) +#define C_000044_VIDDMA_AK 0x7FFFFFFF +#define R_000050_CRTC_GEN_CNTL 0x000050 +#define S_000050_CRTC_DBL_SCAN_EN(x) (((x) & 0x1) << 0) +#define G_000050_CRTC_DBL_SCAN_EN(x) (((x) >> 0) & 0x1) +#define C_000050_CRTC_DBL_SCAN_EN 0xFFFFFFFE +#define S_000050_CRTC_INTERLACE_EN(x) (((x) & 0x1) << 1) +#define G_000050_CRTC_INTERLACE_EN(x) (((x) >> 1) & 0x1) +#define C_000050_CRTC_INTERLACE_EN 0xFFFFFFFD +#define S_000050_CRTC_C_SYNC_EN(x) (((x) & 0x1) << 4) +#define G_000050_CRTC_C_SYNC_EN(x) (((x) >> 4) & 0x1) +#define C_000050_CRTC_C_SYNC_EN 0xFFFFFFEF +#define S_000050_CRTC_PIX_WIDTH(x) (((x) & 0xF) << 8) +#define G_000050_CRTC_PIX_WIDTH(x) (((x) >> 8) & 0xF) +#define C_000050_CRTC_PIX_WIDTH 0xFFFFF0FF +#define S_000050_CRTC_ICON_EN(x) (((x) & 0x1) << 15) +#define G_000050_CRTC_ICON_EN(x) (((x) >> 15) & 0x1) +#define C_000050_CRTC_ICON_EN 0xFFFF7FFF +#define S_000050_CRTC_CUR_EN(x) (((x) & 0x1) << 16) +#define G_000050_CRTC_CUR_EN(x) (((x) >> 16) & 0x1) +#define C_000050_CRTC_CUR_EN 0xFFFEFFFF +#define S_000050_CRTC_VSTAT_MODE(x) (((x) & 0x3) << 17) +#define G_000050_CRTC_VSTAT_MODE(x) (((x) >> 17) & 0x3) +#define C_000050_CRTC_VSTAT_MODE 0xFFF9FFFF +#define S_000050_CRTC_CUR_MODE(x) (((x) & 0x7) << 20) +#define G_000050_CRTC_CUR_MODE(x) (((x) >> 20) & 0x7) +#define C_000050_CRTC_CUR_MODE 0xFF8FFFFF +#define S_000050_CRTC_EXT_DISP_EN(x) (((x) & 0x1) << 24) +#define G_000050_CRTC_EXT_DISP_EN(x) (((x) >> 24) & 0x1) +#define C_000050_CRTC_EXT_DISP_EN 0xFEFFFFFF +#define S_000050_CRTC_EN(x) (((x) & 0x1) << 25) +#define G_000050_CRTC_EN(x) (((x) >> 25) & 0x1) +#define C_000050_CRTC_EN 0xFDFFFFFF +#define S_000050_CRTC_DISP_REQ_EN_B(x) (((x) & 0x1) << 26) +#define G_000050_CRTC_DISP_REQ_EN_B(x) (((x) >> 26) & 0x1) +#define C_000050_CRTC_DISP_REQ_EN_B 0xFBFFFFFF +#define R_000054_CRTC_EXT_CNTL 0x000054 +#define S_000054_CRTC_VGA_XOVERSCAN(x) (((x) & 0x1) << 0) +#define G_000054_CRTC_VGA_XOVERSCAN(x) (((x) >> 0) & 0x1) +#define C_000054_CRTC_VGA_XOVERSCAN 0xFFFFFFFE +#define S_000054_VGA_BLINK_RATE(x) (((x) & 0x3) << 1) +#define G_000054_VGA_BLINK_RATE(x) (((x) >> 1) & 0x3) +#define C_000054_VGA_BLINK_RATE 0xFFFFFFF9 +#define S_000054_VGA_ATI_LINEAR(x) (((x) & 0x1) << 3) +#define G_000054_VGA_ATI_LINEAR(x) (((x) >> 3) & 0x1) +#define C_000054_VGA_ATI_LINEAR 0xFFFFFFF7 +#define S_000054_VGA_128KAP_PAGING(x) (((x) & 0x1) << 4) +#define G_000054_VGA_128KAP_PAGING(x) (((x) >> 4) & 0x1) +#define C_000054_VGA_128KAP_PAGING 0xFFFFFFEF +#define S_000054_VGA_TEXT_132(x) (((x) & 0x1) << 5) +#define G_000054_VGA_TEXT_132(x) (((x) >> 5) & 0x1) +#define C_000054_VGA_TEXT_132 0xFFFFFFDF +#define S_000054_VGA_XCRT_CNT_EN(x) (((x) & 0x1) << 6) +#define G_000054_VGA_XCRT_CNT_EN(x) (((x) >> 6) & 0x1) +#define C_000054_VGA_XCRT_CNT_EN 0xFFFFFFBF +#define S_000054_CRTC_HSYNC_DIS(x) (((x) & 0x1) << 8) +#define G_000054_CRTC_HSYNC_DIS(x) (((x) >> 8) & 0x1) +#define C_000054_CRTC_HSYNC_DIS 0xFFFFFEFF +#define S_000054_CRTC_VSYNC_DIS(x) (((x) & 0x1) << 9) +#define G_000054_CRTC_VSYNC_DIS(x) (((x) >> 9) & 0x1) +#define C_000054_CRTC_VSYNC_DIS 0xFFFFFDFF +#define S_000054_CRTC_DISPLAY_DIS(x) (((x) & 0x1) << 10) +#define G_000054_CRTC_DISPLAY_DIS(x) (((x) >> 10) & 0x1) +#define C_000054_CRTC_DISPLAY_DIS 0xFFFFFBFF +#define S_000054_CRTC_SYNC_TRISTATE(x) (((x) & 0x1) << 11) +#define G_000054_CRTC_SYNC_TRISTATE(x) (((x) >> 11) & 0x1) +#define C_000054_CRTC_SYNC_TRISTATE 0xFFFFF7FF +#define S_000054_CRTC_HSYNC_TRISTATE(x) (((x) & 0x1) << 12) +#define G_000054_CRTC_HSYNC_TRISTATE(x) (((x) >> 12) & 0x1) +#define C_000054_CRTC_HSYNC_TRISTATE 0xFFFFEFFF +#define S_000054_CRTC_VSYNC_TRISTATE(x) (((x) & 0x1) << 13) +#define G_000054_CRTC_VSYNC_TRISTATE(x) (((x) >> 13) & 0x1) +#define C_000054_CRTC_VSYNC_TRISTATE 0xFFFFDFFF +#define S_000054_CRT_ON(x) (((x) & 0x1) << 15) +#define G_000054_CRT_ON(x) (((x) >> 15) & 0x1) +#define C_000054_CRT_ON 0xFFFF7FFF +#define S_000054_VGA_CUR_B_TEST(x) (((x) & 0x1) << 17) +#define G_000054_VGA_CUR_B_TEST(x) (((x) >> 17) & 0x1) +#define C_000054_VGA_CUR_B_TEST 0xFFFDFFFF +#define S_000054_VGA_PACK_DIS(x) (((x) & 0x1) << 18) +#define G_000054_VGA_PACK_DIS(x) (((x) >> 18) & 0x1) +#define C_000054_VGA_PACK_DIS 0xFFFBFFFF +#define S_000054_VGA_MEM_PS_EN(x) (((x) & 0x1) << 19) +#define G_000054_VGA_MEM_PS_EN(x) (((x) >> 19) & 0x1) +#define C_000054_VGA_MEM_PS_EN 0xFFF7FFFF +#define S_000054_VCRTC_IDX_MASTER(x) (((x) & 0x7F) << 24) +#define G_000054_VCRTC_IDX_MASTER(x) (((x) >> 24) & 0x7F) +#define C_000054_VCRTC_IDX_MASTER 0x80FFFFFF +#define R_000148_MC_FB_LOCATION 0x000148 +#define S_000148_MC_FB_START(x) (((x) & 0xFFFF) << 0) +#define G_000148_MC_FB_START(x) (((x) >> 0) & 0xFFFF) +#define C_000148_MC_FB_START 0xFFFF0000 +#define S_000148_MC_FB_TOP(x) (((x) & 0xFFFF) << 16) +#define G_000148_MC_FB_TOP(x) (((x) >> 16) & 0xFFFF) +#define C_000148_MC_FB_TOP 0x0000FFFF +#define R_00014C_MC_AGP_LOCATION 0x00014C +#define S_00014C_MC_AGP_START(x) (((x) & 0xFFFF) << 0) +#define G_00014C_MC_AGP_START(x) (((x) >> 0) & 0xFFFF) +#define C_00014C_MC_AGP_START 0xFFFF0000 +#define S_00014C_MC_AGP_TOP(x) (((x) & 0xFFFF) << 16) +#define G_00014C_MC_AGP_TOP(x) (((x) >> 16) & 0xFFFF) +#define C_00014C_MC_AGP_TOP 0x0000FFFF +#define R_000170_AGP_BASE 0x000170 +#define S_000170_AGP_BASE_ADDR(x) (((x) & 0xFFFFFFFF) << 0) +#define G_000170_AGP_BASE_ADDR(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_000170_AGP_BASE_ADDR 0x00000000 +#define R_00023C_DISPLAY_BASE_ADDR 0x00023C +#define S_00023C_DISPLAY_BASE_ADDR(x) (((x) & 0xFFFFFFFF) << 0) +#define G_00023C_DISPLAY_BASE_ADDR(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_00023C_DISPLAY_BASE_ADDR 0x00000000 +#define R_000260_CUR_OFFSET 0x000260 +#define S_000260_CUR_OFFSET(x) (((x) & 0x7FFFFFF) << 0) +#define G_000260_CUR_OFFSET(x) (((x) >> 0) & 0x7FFFFFF) +#define C_000260_CUR_OFFSET 0xF8000000 +#define S_000260_CUR_LOCK(x) (((x) & 0x1) << 31) +#define G_000260_CUR_LOCK(x) (((x) >> 31) & 0x1) +#define C_000260_CUR_LOCK 0x7FFFFFFF +#define R_00033C_CRTC2_DISPLAY_BASE_ADDR 0x00033C +#define S_00033C_CRTC2_DISPLAY_BASE_ADDR(x) (((x) & 0xFFFFFFFF) << 0) +#define G_00033C_CRTC2_DISPLAY_BASE_ADDR(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_00033C_CRTC2_DISPLAY_BASE_ADDR 0x00000000 +#define R_000360_CUR2_OFFSET 0x000360 +#define S_000360_CUR2_OFFSET(x) (((x) & 0x7FFFFFF) << 0) +#define G_000360_CUR2_OFFSET(x) (((x) >> 0) & 0x7FFFFFF) +#define C_000360_CUR2_OFFSET 0xF8000000 +#define S_000360_CUR2_LOCK(x) (((x) & 0x1) << 31) +#define G_000360_CUR2_LOCK(x) (((x) >> 31) & 0x1) +#define C_000360_CUR2_LOCK 0x7FFFFFFF +#define R_0003C2_GENMO_WT 0x0003C2 +#define S_0003C2_GENMO_MONO_ADDRESS_B(x) (((x) & 0x1) << 0) +#define G_0003C2_GENMO_MONO_ADDRESS_B(x) (((x) >> 0) & 0x1) +#define C_0003C2_GENMO_MONO_ADDRESS_B 0xFE +#define S_0003C2_VGA_RAM_EN(x) (((x) & 0x1) << 1) +#define G_0003C2_VGA_RAM_EN(x) (((x) >> 1) & 0x1) +#define C_0003C2_VGA_RAM_EN 0xFD +#define S_0003C2_VGA_CKSEL(x) (((x) & 0x3) << 2) +#define G_0003C2_VGA_CKSEL(x) (((x) >> 2) & 0x3) +#define C_0003C2_VGA_CKSEL 0xF3 +#define S_0003C2_ODD_EVEN_MD_PGSEL(x) (((x) & 0x1) << 5) +#define G_0003C2_ODD_EVEN_MD_PGSEL(x) (((x) >> 5) & 0x1) +#define C_0003C2_ODD_EVEN_MD_PGSEL 0xDF +#define S_0003C2_VGA_HSYNC_POL(x) (((x) & 0x1) << 6) +#define G_0003C2_VGA_HSYNC_POL(x) (((x) >> 6) & 0x1) +#define C_0003C2_VGA_HSYNC_POL 0xBF +#define S_0003C2_VGA_VSYNC_POL(x) (((x) & 0x1) << 7) +#define G_0003C2_VGA_VSYNC_POL(x) (((x) >> 7) & 0x1) +#define C_0003C2_VGA_VSYNC_POL 0x7F +#define R_0003F8_CRTC2_GEN_CNTL 0x0003F8 +#define S_0003F8_CRTC2_DBL_SCAN_EN(x) (((x) & 0x1) << 0) +#define G_0003F8_CRTC2_DBL_SCAN_EN(x) (((x) >> 0) & 0x1) +#define C_0003F8_CRTC2_DBL_SCAN_EN 0xFFFFFFFE +#define S_0003F8_CRTC2_INTERLACE_EN(x) (((x) & 0x1) << 1) +#define G_0003F8_CRTC2_INTERLACE_EN(x) (((x) >> 1) & 0x1) +#define C_0003F8_CRTC2_INTERLACE_EN 0xFFFFFFFD +#define S_0003F8_CRTC2_SYNC_TRISTATE(x) (((x) & 0x1) << 4) +#define G_0003F8_CRTC2_SYNC_TRISTATE(x) (((x) >> 4) & 0x1) +#define C_0003F8_CRTC2_SYNC_TRISTATE 0xFFFFFFEF +#define S_0003F8_CRTC2_HSYNC_TRISTATE(x) (((x) & 0x1) << 5) +#define G_0003F8_CRTC2_HSYNC_TRISTATE(x) (((x) >> 5) & 0x1) +#define C_0003F8_CRTC2_HSYNC_TRISTATE 0xFFFFFFDF +#define S_0003F8_CRTC2_VSYNC_TRISTATE(x) (((x) & 0x1) << 6) +#define G_0003F8_CRTC2_VSYNC_TRISTATE(x) (((x) >> 6) & 0x1) +#define C_0003F8_CRTC2_VSYNC_TRISTATE 0xFFFFFFBF +#define S_0003F8_CRT2_ON(x) (((x) & 0x1) << 7) +#define G_0003F8_CRT2_ON(x) (((x) >> 7) & 0x1) +#define C_0003F8_CRT2_ON 0xFFFFFF7F +#define S_0003F8_CRTC2_PIX_WIDTH(x) (((x) & 0xF) << 8) +#define G_0003F8_CRTC2_PIX_WIDTH(x) (((x) >> 8) & 0xF) +#define C_0003F8_CRTC2_PIX_WIDTH 0xFFFFF0FF +#define S_0003F8_CRTC2_ICON_EN(x) (((x) & 0x1) << 15) +#define G_0003F8_CRTC2_ICON_EN(x) (((x) >> 15) & 0x1) +#define C_0003F8_CRTC2_ICON_EN 0xFFFF7FFF +#define S_0003F8_CRTC2_CUR_EN(x) (((x) & 0x1) << 16) +#define G_0003F8_CRTC2_CUR_EN(x) (((x) >> 16) & 0x1) +#define C_0003F8_CRTC2_CUR_EN 0xFFFEFFFF +#define S_0003F8_CRTC2_CUR_MODE(x) (((x) & 0x7) << 20) +#define G_0003F8_CRTC2_CUR_MODE(x) (((x) >> 20) & 0x7) +#define C_0003F8_CRTC2_CUR_MODE 0xFF8FFFFF +#define S_0003F8_CRTC2_DISPLAY_DIS(x) (((x) & 0x1) << 23) +#define G_0003F8_CRTC2_DISPLAY_DIS(x) (((x) >> 23) & 0x1) +#define C_0003F8_CRTC2_DISPLAY_DIS 0xFF7FFFFF +#define S_0003F8_CRTC2_EN(x) (((x) & 0x1) << 25) +#define G_0003F8_CRTC2_EN(x) (((x) >> 25) & 0x1) +#define C_0003F8_CRTC2_EN 0xFDFFFFFF +#define S_0003F8_CRTC2_DISP_REQ_EN_B(x) (((x) & 0x1) << 26) +#define G_0003F8_CRTC2_DISP_REQ_EN_B(x) (((x) >> 26) & 0x1) +#define C_0003F8_CRTC2_DISP_REQ_EN_B 0xFBFFFFFF +#define S_0003F8_CRTC2_C_SYNC_EN(x) (((x) & 0x1) << 27) +#define G_0003F8_CRTC2_C_SYNC_EN(x) (((x) >> 27) & 0x1) +#define C_0003F8_CRTC2_C_SYNC_EN 0xF7FFFFFF +#define S_0003F8_CRTC2_HSYNC_DIS(x) (((x) & 0x1) << 28) +#define G_0003F8_CRTC2_HSYNC_DIS(x) (((x) >> 28) & 0x1) +#define C_0003F8_CRTC2_HSYNC_DIS 0xEFFFFFFF +#define S_0003F8_CRTC2_VSYNC_DIS(x) (((x) & 0x1) << 29) +#define G_0003F8_CRTC2_VSYNC_DIS(x) (((x) >> 29) & 0x1) +#define C_0003F8_CRTC2_VSYNC_DIS 0xDFFFFFFF +#define R_000420_OV0_SCALE_CNTL 0x000420 +#define S_000420_OV0_NO_READ_BEHIND_SCAN(x) (((x) & 0x1) << 1) +#define G_000420_OV0_NO_READ_BEHIND_SCAN(x) (((x) >> 1) & 0x1) +#define C_000420_OV0_NO_READ_BEHIND_SCAN 0xFFFFFFFD +#define S_000420_OV0_HORZ_PICK_NEAREST(x) (((x) & 0x1) << 2) +#define G_000420_OV0_HORZ_PICK_NEAREST(x) (((x) >> 2) & 0x1) +#define C_000420_OV0_HORZ_PICK_NEAREST 0xFFFFFFFB +#define S_000420_OV0_VERT_PICK_NEAREST(x) (((x) & 0x1) << 3) +#define G_000420_OV0_VERT_PICK_NEAREST(x) (((x) >> 3) & 0x1) +#define C_000420_OV0_VERT_PICK_NEAREST 0xFFFFFFF7 +#define S_000420_OV0_SIGNED_UV(x) (((x) & 0x1) << 4) +#define G_000420_OV0_SIGNED_UV(x) (((x) >> 4) & 0x1) +#define C_000420_OV0_SIGNED_UV 0xFFFFFFEF +#define S_000420_OV0_GAMMA_SEL(x) (((x) & 0x7) << 5) +#define G_000420_OV0_GAMMA_SEL(x) (((x) >> 5) & 0x7) +#define C_000420_OV0_GAMMA_SEL 0xFFFFFF1F +#define S_000420_OV0_SURFACE_FORMAT(x) (((x) & 0xF) << 8) +#define G_000420_OV0_SURFACE_FORMAT(x) (((x) >> 8) & 0xF) +#define C_000420_OV0_SURFACE_FORMAT 0xFFFFF0FF +#define S_000420_OV0_ADAPTIVE_DEINT(x) (((x) & 0x1) << 12) +#define G_000420_OV0_ADAPTIVE_DEINT(x) (((x) >> 12) & 0x1) +#define C_000420_OV0_ADAPTIVE_DEINT 0xFFFFEFFF +#define S_000420_OV0_CRTC_SEL(x) (((x) & 0x1) << 14) +#define G_000420_OV0_CRTC_SEL(x) (((x) >> 14) & 0x1) +#define C_000420_OV0_CRTC_SEL 0xFFFFBFFF +#define S_000420_OV0_BURST_PER_PLANE(x) (((x) & 0x7F) << 16) +#define G_000420_OV0_BURST_PER_PLANE(x) (((x) >> 16) & 0x7F) +#define C_000420_OV0_BURST_PER_PLANE 0xFF80FFFF +#define S_000420_OV0_DOUBLE_BUFFER_REGS(x) (((x) & 0x1) << 24) +#define G_000420_OV0_DOUBLE_BUFFER_REGS(x) (((x) >> 24) & 0x1) +#define C_000420_OV0_DOUBLE_BUFFER_REGS 0xFEFFFFFF +#define S_000420_OV0_BANDWIDTH(x) (((x) & 0x1) << 26) +#define G_000420_OV0_BANDWIDTH(x) (((x) >> 26) & 0x1) +#define C_000420_OV0_BANDWIDTH 0xFBFFFFFF +#define S_000420_OV0_LIN_TRANS_BYPASS(x) (((x) & 0x1) << 28) +#define G_000420_OV0_LIN_TRANS_BYPASS(x) (((x) >> 28) & 0x1) +#define C_000420_OV0_LIN_TRANS_BYPASS 0xEFFFFFFF +#define S_000420_OV0_INT_EMU(x) (((x) & 0x1) << 29) +#define G_000420_OV0_INT_EMU(x) (((x) >> 29) & 0x1) +#define C_000420_OV0_INT_EMU 0xDFFFFFFF +#define S_000420_OV0_OVERLAY_EN(x) (((x) & 0x1) << 30) +#define G_000420_OV0_OVERLAY_EN(x) (((x) >> 30) & 0x1) +#define C_000420_OV0_OVERLAY_EN 0xBFFFFFFF +#define S_000420_OV0_SOFT_RESET(x) (((x) & 0x1) << 31) +#define G_000420_OV0_SOFT_RESET(x) (((x) >> 31) & 0x1) +#define C_000420_OV0_SOFT_RESET 0x7FFFFFFF +#define R_00070C_CP_RB_RPTR_ADDR 0x00070C +#define S_00070C_RB_RPTR_SWAP(x) (((x) & 0x3) << 0) +#define G_00070C_RB_RPTR_SWAP(x) (((x) >> 0) & 0x3) +#define C_00070C_RB_RPTR_SWAP 0xFFFFFFFC +#define S_00070C_RB_RPTR_ADDR(x) (((x) & 0x3FFFFFFF) << 2) +#define G_00070C_RB_RPTR_ADDR(x) (((x) >> 2) & 0x3FFFFFFF) +#define C_00070C_RB_RPTR_ADDR 0x00000003 +#define R_000740_CP_CSQ_CNTL 0x000740 +#define S_000740_CSQ_CNT_PRIMARY(x) (((x) & 0xFF) << 0) +#define G_000740_CSQ_CNT_PRIMARY(x) (((x) >> 0) & 0xFF) +#define C_000740_CSQ_CNT_PRIMARY 0xFFFFFF00 +#define S_000740_CSQ_CNT_INDIRECT(x) (((x) & 0xFF) << 8) +#define G_000740_CSQ_CNT_INDIRECT(x) (((x) >> 8) & 0xFF) +#define C_000740_CSQ_CNT_INDIRECT 0xFFFF00FF +#define S_000740_CSQ_MODE(x) (((x) & 0xF) << 28) +#define G_000740_CSQ_MODE(x) (((x) >> 28) & 0xF) +#define C_000740_CSQ_MODE 0x0FFFFFFF +#define R_000770_SCRATCH_UMSK 0x000770 +#define S_000770_SCRATCH_UMSK(x) (((x) & 0x3F) << 0) +#define G_000770_SCRATCH_UMSK(x) (((x) >> 0) & 0x3F) +#define C_000770_SCRATCH_UMSK 0xFFFFFFC0 +#define S_000770_SCRATCH_SWAP(x) (((x) & 0x3) << 16) +#define G_000770_SCRATCH_SWAP(x) (((x) >> 16) & 0x3) +#define C_000770_SCRATCH_SWAP 0xFFFCFFFF +#define R_000774_SCRATCH_ADDR 0x000774 +#define S_000774_SCRATCH_ADDR(x) (((x) & 0x7FFFFFF) << 5) +#define G_000774_SCRATCH_ADDR(x) (((x) >> 5) & 0x7FFFFFF) +#define C_000774_SCRATCH_ADDR 0x0000001F +#define R_0007C0_CP_STAT 0x0007C0 +#define S_0007C0_MRU_BUSY(x) (((x) & 0x1) << 0) +#define G_0007C0_MRU_BUSY(x) (((x) >> 0) & 0x1) +#define C_0007C0_MRU_BUSY 0xFFFFFFFE +#define S_0007C0_MWU_BUSY(x) (((x) & 0x1) << 1) +#define G_0007C0_MWU_BUSY(x) (((x) >> 1) & 0x1) +#define C_0007C0_MWU_BUSY 0xFFFFFFFD +#define S_0007C0_RSIU_BUSY(x) (((x) & 0x1) << 2) +#define G_0007C0_RSIU_BUSY(x) (((x) >> 2) & 0x1) +#define C_0007C0_RSIU_BUSY 0xFFFFFFFB +#define S_0007C0_RCIU_BUSY(x) (((x) & 0x1) << 3) +#define G_0007C0_RCIU_BUSY(x) (((x) >> 3) & 0x1) +#define C_0007C0_RCIU_BUSY 0xFFFFFFF7 +#define S_0007C0_CSF_PRIMARY_BUSY(x) (((x) & 0x1) << 9) +#define G_0007C0_CSF_PRIMARY_BUSY(x) (((x) >> 9) & 0x1) +#define C_0007C0_CSF_PRIMARY_BUSY 0xFFFFFDFF +#define S_0007C0_CSF_INDIRECT_BUSY(x) (((x) & 0x1) << 10) +#define G_0007C0_CSF_INDIRECT_BUSY(x) (((x) >> 10) & 0x1) +#define C_0007C0_CSF_INDIRECT_BUSY 0xFFFFFBFF +#define S_0007C0_CSQ_PRIMARY_BUSY(x) (((x) & 0x1) << 11) +#define G_0007C0_CSQ_PRIMARY_BUSY(x) (((x) >> 11) & 0x1) +#define C_0007C0_CSQ_PRIMARY_BUSY 0xFFFFF7FF +#define S_0007C0_CSQ_INDIRECT_BUSY(x) (((x) & 0x1) << 12) +#define G_0007C0_CSQ_INDIRECT_BUSY(x) (((x) >> 12) & 0x1) +#define C_0007C0_CSQ_INDIRECT_BUSY 0xFFFFEFFF +#define S_0007C0_CSI_BUSY(x) (((x) & 0x1) << 13) +#define G_0007C0_CSI_BUSY(x) (((x) >> 13) & 0x1) +#define C_0007C0_CSI_BUSY 0xFFFFDFFF +#define S_0007C0_GUIDMA_BUSY(x) (((x) & 0x1) << 28) +#define G_0007C0_GUIDMA_BUSY(x) (((x) >> 28) & 0x1) +#define C_0007C0_GUIDMA_BUSY 0xEFFFFFFF +#define S_0007C0_VIDDMA_BUSY(x) (((x) & 0x1) << 29) +#define G_0007C0_VIDDMA_BUSY(x) (((x) >> 29) & 0x1) +#define C_0007C0_VIDDMA_BUSY 0xDFFFFFFF +#define S_0007C0_CMDSTRM_BUSY(x) (((x) & 0x1) << 30) +#define G_0007C0_CMDSTRM_BUSY(x) (((x) >> 30) & 0x1) +#define C_0007C0_CMDSTRM_BUSY 0xBFFFFFFF +#define S_0007C0_CP_BUSY(x) (((x) & 0x1) << 31) +#define G_0007C0_CP_BUSY(x) (((x) >> 31) & 0x1) +#define C_0007C0_CP_BUSY 0x7FFFFFFF +#define R_000E40_RBBM_STATUS 0x000E40 +#define S_000E40_CMDFIFO_AVAIL(x) (((x) & 0x7F) << 0) +#define G_000E40_CMDFIFO_AVAIL(x) (((x) >> 0) & 0x7F) +#define C_000E40_CMDFIFO_AVAIL 0xFFFFFF80 +#define S_000E40_HIRQ_ON_RBB(x) (((x) & 0x1) << 8) +#define G_000E40_HIRQ_ON_RBB(x) (((x) >> 8) & 0x1) +#define C_000E40_HIRQ_ON_RBB 0xFFFFFEFF +#define S_000E40_CPRQ_ON_RBB(x) (((x) & 0x1) << 9) +#define G_000E40_CPRQ_ON_RBB(x) (((x) >> 9) & 0x1) +#define C_000E40_CPRQ_ON_RBB 0xFFFFFDFF +#define S_000E40_CFRQ_ON_RBB(x) (((x) & 0x1) << 10) +#define G_000E40_CFRQ_ON_RBB(x) (((x) >> 10) & 0x1) +#define C_000E40_CFRQ_ON_RBB 0xFFFFFBFF +#define S_000E40_HIRQ_IN_RTBUF(x) (((x) & 0x1) << 11) +#define G_000E40_HIRQ_IN_RTBUF(x) (((x) >> 11) & 0x1) +#define C_000E40_HIRQ_IN_RTBUF 0xFFFFF7FF +#define S_000E40_CPRQ_IN_RTBUF(x) (((x) & 0x1) << 12) +#define G_000E40_CPRQ_IN_RTBUF(x) (((x) >> 12) & 0x1) +#define C_000E40_CPRQ_IN_RTBUF 0xFFFFEFFF +#define S_000E40_CFRQ_IN_RTBUF(x) (((x) & 0x1) << 13) +#define G_000E40_CFRQ_IN_RTBUF(x) (((x) >> 13) & 0x1) +#define C_000E40_CFRQ_IN_RTBUF 0xFFFFDFFF +#define S_000E40_CF_PIPE_BUSY(x) (((x) & 0x1) << 14) +#define G_000E40_CF_PIPE_BUSY(x) (((x) >> 14) & 0x1) +#define C_000E40_CF_PIPE_BUSY 0xFFFFBFFF +#define S_000E40_ENG_EV_BUSY(x) (((x) & 0x1) << 15) +#define G_000E40_ENG_EV_BUSY(x) (((x) >> 15) & 0x1) +#define C_000E40_ENG_EV_BUSY 0xFFFF7FFF +#define S_000E40_CP_CMDSTRM_BUSY(x) (((x) & 0x1) << 16) +#define G_000E40_CP_CMDSTRM_BUSY(x) (((x) >> 16) & 0x1) +#define C_000E40_CP_CMDSTRM_BUSY 0xFFFEFFFF +#define S_000E40_E2_BUSY(x) (((x) & 0x1) << 17) +#define G_000E40_E2_BUSY(x) (((x) >> 17) & 0x1) +#define C_000E40_E2_BUSY 0xFFFDFFFF +#define S_000E40_RB2D_BUSY(x) (((x) & 0x1) << 18) +#define G_000E40_RB2D_BUSY(x) (((x) >> 18) & 0x1) +#define C_000E40_RB2D_BUSY 0xFFFBFFFF +#define S_000E40_RB3D_BUSY(x) (((x) & 0x1) << 19) +#define G_000E40_RB3D_BUSY(x) (((x) >> 19) & 0x1) +#define C_000E40_RB3D_BUSY 0xFFF7FFFF +#define S_000E40_SE_BUSY(x) (((x) & 0x1) << 20) +#define G_000E40_SE_BUSY(x) (((x) >> 20) & 0x1) +#define C_000E40_SE_BUSY 0xFFEFFFFF +#define S_000E40_RE_BUSY(x) (((x) & 0x1) << 21) +#define G_000E40_RE_BUSY(x) (((x) >> 21) & 0x1) +#define C_000E40_RE_BUSY 0xFFDFFFFF +#define S_000E40_TAM_BUSY(x) (((x) & 0x1) << 22) +#define G_000E40_TAM_BUSY(x) (((x) >> 22) & 0x1) +#define C_000E40_TAM_BUSY 0xFFBFFFFF +#define S_000E40_TDM_BUSY(x) (((x) & 0x1) << 23) +#define G_000E40_TDM_BUSY(x) (((x) >> 23) & 0x1) +#define C_000E40_TDM_BUSY 0xFF7FFFFF +#define S_000E40_PB_BUSY(x) (((x) & 0x1) << 24) +#define G_000E40_PB_BUSY(x) (((x) >> 24) & 0x1) +#define C_000E40_PB_BUSY 0xFEFFFFFF +#define S_000E40_GUI_ACTIVE(x) (((x) & 0x1) << 31) +#define G_000E40_GUI_ACTIVE(x) (((x) >> 31) & 0x1) +#define C_000E40_GUI_ACTIVE 0x7FFFFFFF + + +#define R_00000D_SCLK_CNTL 0x00000D +#define S_00000D_SCLK_SRC_SEL(x) (((x) & 0x7) << 0) +#define G_00000D_SCLK_SRC_SEL(x) (((x) >> 0) & 0x7) +#define C_00000D_SCLK_SRC_SEL 0xFFFFFFF8 +#define S_00000D_TCLK_SRC_SEL(x) (((x) & 0x7) << 8) +#define G_00000D_TCLK_SRC_SEL(x) (((x) >> 8) & 0x7) +#define C_00000D_TCLK_SRC_SEL 0xFFFFF8FF +#define S_00000D_FORCE_CP(x) (((x) & 0x1) << 16) +#define G_00000D_FORCE_CP(x) (((x) >> 16) & 0x1) +#define C_00000D_FORCE_CP 0xFFFEFFFF +#define S_00000D_FORCE_HDP(x) (((x) & 0x1) << 17) +#define G_00000D_FORCE_HDP(x) (((x) >> 17) & 0x1) +#define C_00000D_FORCE_HDP 0xFFFDFFFF +#define S_00000D_FORCE_DISP(x) (((x) & 0x1) << 18) +#define G_00000D_FORCE_DISP(x) (((x) >> 18) & 0x1) +#define C_00000D_FORCE_DISP 0xFFFBFFFF +#define S_00000D_FORCE_TOP(x) (((x) & 0x1) << 19) +#define G_00000D_FORCE_TOP(x) (((x) >> 19) & 0x1) +#define C_00000D_FORCE_TOP 0xFFF7FFFF +#define S_00000D_FORCE_E2(x) (((x) & 0x1) << 20) +#define G_00000D_FORCE_E2(x) (((x) >> 20) & 0x1) +#define C_00000D_FORCE_E2 0xFFEFFFFF +#define S_00000D_FORCE_SE(x) (((x) & 0x1) << 21) +#define G_00000D_FORCE_SE(x) (((x) >> 21) & 0x1) +#define C_00000D_FORCE_SE 0xFFDFFFFF +#define S_00000D_FORCE_IDCT(x) (((x) & 0x1) << 22) +#define G_00000D_FORCE_IDCT(x) (((x) >> 22) & 0x1) +#define C_00000D_FORCE_IDCT 0xFFBFFFFF +#define S_00000D_FORCE_VIP(x) (((x) & 0x1) << 23) +#define G_00000D_FORCE_VIP(x) (((x) >> 23) & 0x1) +#define C_00000D_FORCE_VIP 0xFF7FFFFF +#define S_00000D_FORCE_RE(x) (((x) & 0x1) << 24) +#define G_00000D_FORCE_RE(x) (((x) >> 24) & 0x1) +#define C_00000D_FORCE_RE 0xFEFFFFFF +#define S_00000D_FORCE_PB(x) (((x) & 0x1) << 25) +#define G_00000D_FORCE_PB(x) (((x) >> 25) & 0x1) +#define C_00000D_FORCE_PB 0xFDFFFFFF +#define S_00000D_FORCE_TAM(x) (((x) & 0x1) << 26) +#define G_00000D_FORCE_TAM(x) (((x) >> 26) & 0x1) +#define C_00000D_FORCE_TAM 0xFBFFFFFF +#define S_00000D_FORCE_TDM(x) (((x) & 0x1) << 27) +#define G_00000D_FORCE_TDM(x) (((x) >> 27) & 0x1) +#define C_00000D_FORCE_TDM 0xF7FFFFFF +#define S_00000D_FORCE_RB(x) (((x) & 0x1) << 28) +#define G_00000D_FORCE_RB(x) (((x) >> 28) & 0x1) +#define C_00000D_FORCE_RB 0xEFFFFFFF + +/* PLL regs */ +#define SCLK_CNTL 0xd +#define FORCE_HDP (1 << 17) +#define CLK_PWRMGT_CNTL 0x14 +#define GLOBAL_PMAN_EN (1 << 10) +#define DISP_PM (1 << 20) +#define PLL_PWRMGT_CNTL 0x15 +#define MPLL_TURNOFF (1 << 0) +#define SPLL_TURNOFF (1 << 1) +#define PPLL_TURNOFF (1 << 2) +#define P2PLL_TURNOFF (1 << 3) +#define TVPLL_TURNOFF (1 << 4) +#define MOBILE_SU (1 << 16) +#define SU_SCLK_USE_BCLK (1 << 17) +#define SCLK_CNTL2 0x1e +#define REDUCED_SPEED_SCLK_MODE (1 << 16) +#define REDUCED_SPEED_SCLK_SEL(x) ((x) << 17) +#define MCLK_MISC 0x1f +#define EN_MCLK_TRISTATE_IN_SUSPEND (1 << 18) +#define SCLK_MORE_CNTL 0x35 +#define REDUCED_SPEED_SCLK_EN (1 << 16) +#define IO_CG_VOLTAGE_DROP (1 << 17) +#define VOLTAGE_DELAY_SEL(x) ((x) << 20) +#define VOLTAGE_DROP_SYNC (1 << 19) + +/* mmreg */ +#define DISP_PWR_MAN 0xd08 +#define DISP_D3_GRPH_RST (1 << 18) +#define DISP_D3_SUBPIC_RST (1 << 19) +#define DISP_D3_OV0_RST (1 << 20) +#define DISP_D1D2_GRPH_RST (1 << 21) +#define DISP_D1D2_SUBPIC_RST (1 << 22) +#define DISP_D1D2_OV0_RST (1 << 23) +#define DISP_DVO_ENABLE_RST (1 << 24) +#define TV_ENABLE_RST (1 << 25) +#define AUTO_PWRUP_EN (1 << 26) + +#endif diff --git a/drivers/gpu/drm/radeon/r200.c b/drivers/gpu/drm/radeon/r200.c new file mode 100644 index 00000000..a1f3ba06 --- /dev/null +++ b/drivers/gpu/drm/radeon/r200.c @@ -0,0 +1,535 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#include "drmP.h" +#include "drm.h" +#include "radeon_drm.h" +#include "radeon_reg.h" +#include "radeon.h" +#include "radeon_asic.h" + +#include "r100d.h" +#include "r200_reg_safe.h" + +#include "r100_track.h" + +static int r200_get_vtx_size_0(uint32_t vtx_fmt_0) +{ + int vtx_size, i; + vtx_size = 2; + + if (vtx_fmt_0 & R200_VTX_Z0) + vtx_size++; + if (vtx_fmt_0 & R200_VTX_W0) + vtx_size++; + /* blend weight */ + if (vtx_fmt_0 & (0x7 << R200_VTX_WEIGHT_COUNT_SHIFT)) + vtx_size += (vtx_fmt_0 >> R200_VTX_WEIGHT_COUNT_SHIFT) & 0x7; + if (vtx_fmt_0 & R200_VTX_PV_MATRIX_SEL) + vtx_size++; + if (vtx_fmt_0 & R200_VTX_N0) + vtx_size += 3; + if (vtx_fmt_0 & R200_VTX_POINT_SIZE) + vtx_size++; + if (vtx_fmt_0 & R200_VTX_DISCRETE_FOG) + vtx_size++; + if (vtx_fmt_0 & R200_VTX_SHININESS_0) + vtx_size++; + if (vtx_fmt_0 & R200_VTX_SHININESS_1) + vtx_size++; + for (i = 0; i < 8; i++) { + int color_size = (vtx_fmt_0 >> (11 + 2*i)) & 0x3; + switch (color_size) { + case 0: break; + case 1: vtx_size++; break; + case 2: vtx_size += 3; break; + case 3: vtx_size += 4; break; + } + } + if (vtx_fmt_0 & R200_VTX_XY1) + vtx_size += 2; + if (vtx_fmt_0 & R200_VTX_Z1) + vtx_size++; + if (vtx_fmt_0 & R200_VTX_W1) + vtx_size++; + if (vtx_fmt_0 & R200_VTX_N1) + vtx_size += 3; + return vtx_size; +} + +int r200_copy_dma(struct radeon_device *rdev, + uint64_t src_offset, + uint64_t dst_offset, + unsigned num_gpu_pages, + struct radeon_fence *fence) +{ + uint32_t size; + uint32_t cur_size; + int i, num_loops; + int r = 0; + + /* radeon pitch is /64 */ + size = num_gpu_pages << RADEON_GPU_PAGE_SHIFT; + num_loops = DIV_ROUND_UP(size, 0x1FFFFF); + r = radeon_ring_lock(rdev, num_loops * 4 + 64); + if (r) { + DRM_ERROR("radeon: moving bo (%d).\n", r); + return r; + } + /* Must wait for 2D idle & clean before DMA or hangs might happen */ + radeon_ring_write(rdev, PACKET0(RADEON_WAIT_UNTIL, 0)); + radeon_ring_write(rdev, (1 << 16)); + for (i = 0; i < num_loops; i++) { + cur_size = size; + if (cur_size > 0x1FFFFF) { + cur_size = 0x1FFFFF; + } + size -= cur_size; + radeon_ring_write(rdev, PACKET0(0x720, 2)); + radeon_ring_write(rdev, src_offset); + radeon_ring_write(rdev, dst_offset); + radeon_ring_write(rdev, cur_size | (1 << 31) | (1 << 30)); + src_offset += cur_size; + dst_offset += cur_size; + } + radeon_ring_write(rdev, PACKET0(RADEON_WAIT_UNTIL, 0)); + radeon_ring_write(rdev, RADEON_WAIT_DMA_GUI_IDLE); + if (fence) { + r = radeon_fence_emit(rdev, fence); + } + radeon_ring_unlock_commit(rdev); + return r; +} + + +static int r200_get_vtx_size_1(uint32_t vtx_fmt_1) +{ + int vtx_size, i, tex_size; + vtx_size = 0; + for (i = 0; i < 6; i++) { + tex_size = (vtx_fmt_1 >> (i * 3)) & 0x7; + if (tex_size > 4) + continue; + vtx_size += tex_size; + } + return vtx_size; +} + +int r200_packet0_check(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt, + unsigned idx, unsigned reg) +{ + struct radeon_cs_reloc *reloc; + struct r100_cs_track *track; + volatile uint32_t *ib; + uint32_t tmp; + int r; + int i; + int face; + u32 tile_flags = 0; + u32 idx_value; + + ib = p->ib->ptr; + track = (struct r100_cs_track *)p->track; + idx_value = radeon_get_ib_value(p, idx); + switch (reg) { + case RADEON_CRTC_GUI_TRIG_VLINE: + r = r100_cs_packet_parse_vline(p); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + break; + /* FIXME: only allow PACKET3 blit? easier to check for out of + * range access */ + case RADEON_DST_PITCH_OFFSET: + case RADEON_SRC_PITCH_OFFSET: + r = r100_reloc_pitch_offset(p, pkt, idx, reg); + if (r) + return r; + break; + case RADEON_RB3D_DEPTHOFFSET: + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + track->zb.robj = reloc->robj; + track->zb.offset = idx_value; + track->zb_dirty = true; + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + break; + case RADEON_RB3D_COLOROFFSET: + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + track->cb[0].robj = reloc->robj; + track->cb[0].offset = idx_value; + track->cb_dirty = true; + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + break; + case R200_PP_TXOFFSET_0: + case R200_PP_TXOFFSET_1: + case R200_PP_TXOFFSET_2: + case R200_PP_TXOFFSET_3: + case R200_PP_TXOFFSET_4: + case R200_PP_TXOFFSET_5: + i = (reg - R200_PP_TXOFFSET_0) / 24; + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + track->textures[i].robj = reloc->robj; + track->tex_dirty = true; + break; + case R200_PP_CUBIC_OFFSET_F1_0: + case R200_PP_CUBIC_OFFSET_F2_0: + case R200_PP_CUBIC_OFFSET_F3_0: + case R200_PP_CUBIC_OFFSET_F4_0: + case R200_PP_CUBIC_OFFSET_F5_0: + case R200_PP_CUBIC_OFFSET_F1_1: + case R200_PP_CUBIC_OFFSET_F2_1: + case R200_PP_CUBIC_OFFSET_F3_1: + case R200_PP_CUBIC_OFFSET_F4_1: + case R200_PP_CUBIC_OFFSET_F5_1: + case R200_PP_CUBIC_OFFSET_F1_2: + case R200_PP_CUBIC_OFFSET_F2_2: + case R200_PP_CUBIC_OFFSET_F3_2: + case R200_PP_CUBIC_OFFSET_F4_2: + case R200_PP_CUBIC_OFFSET_F5_2: + case R200_PP_CUBIC_OFFSET_F1_3: + case R200_PP_CUBIC_OFFSET_F2_3: + case R200_PP_CUBIC_OFFSET_F3_3: + case R200_PP_CUBIC_OFFSET_F4_3: + case R200_PP_CUBIC_OFFSET_F5_3: + case R200_PP_CUBIC_OFFSET_F1_4: + case R200_PP_CUBIC_OFFSET_F2_4: + case R200_PP_CUBIC_OFFSET_F3_4: + case R200_PP_CUBIC_OFFSET_F4_4: + case R200_PP_CUBIC_OFFSET_F5_4: + case R200_PP_CUBIC_OFFSET_F1_5: + case R200_PP_CUBIC_OFFSET_F2_5: + case R200_PP_CUBIC_OFFSET_F3_5: + case R200_PP_CUBIC_OFFSET_F4_5: + case R200_PP_CUBIC_OFFSET_F5_5: + i = (reg - R200_PP_TXOFFSET_0) / 24; + face = (reg - ((i * 24) + R200_PP_TXOFFSET_0)) / 4; + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + track->textures[i].cube_info[face - 1].offset = idx_value; + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + track->textures[i].cube_info[face - 1].robj = reloc->robj; + track->tex_dirty = true; + break; + case RADEON_RE_WIDTH_HEIGHT: + track->maxy = ((idx_value >> 16) & 0x7FF); + track->cb_dirty = true; + track->zb_dirty = true; + break; + case RADEON_RB3D_COLORPITCH: + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + + if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) + tile_flags |= RADEON_COLOR_TILE_ENABLE; + if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) + tile_flags |= RADEON_COLOR_MICROTILE_ENABLE; + + tmp = idx_value & ~(0x7 << 16); + tmp |= tile_flags; + ib[idx] = tmp; + + track->cb[0].pitch = idx_value & RADEON_COLORPITCH_MASK; + track->cb_dirty = true; + break; + case RADEON_RB3D_DEPTHPITCH: + track->zb.pitch = idx_value & RADEON_DEPTHPITCH_MASK; + track->zb_dirty = true; + break; + case RADEON_RB3D_CNTL: + switch ((idx_value >> RADEON_RB3D_COLOR_FORMAT_SHIFT) & 0x1f) { + case 7: + case 8: + case 9: + case 11: + case 12: + track->cb[0].cpp = 1; + break; + case 3: + case 4: + case 15: + track->cb[0].cpp = 2; + break; + case 6: + track->cb[0].cpp = 4; + break; + default: + DRM_ERROR("Invalid color buffer format (%d) !\n", + ((idx_value >> RADEON_RB3D_COLOR_FORMAT_SHIFT) & 0x1f)); + return -EINVAL; + } + if (idx_value & RADEON_DEPTHXY_OFFSET_ENABLE) { + DRM_ERROR("No support for depth xy offset in kms\n"); + return -EINVAL; + } + + track->z_enabled = !!(idx_value & RADEON_Z_ENABLE); + track->cb_dirty = true; + track->zb_dirty = true; + break; + case RADEON_RB3D_ZSTENCILCNTL: + switch (idx_value & 0xf) { + case 0: + track->zb.cpp = 2; + break; + case 2: + case 3: + case 4: + case 5: + case 9: + case 11: + track->zb.cpp = 4; + break; + default: + break; + } + track->zb_dirty = true; + break; + case RADEON_RB3D_ZPASS_ADDR: + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + break; + case RADEON_PP_CNTL: + { + uint32_t temp = idx_value >> 4; + for (i = 0; i < track->num_texture; i++) + track->textures[i].enabled = !!(temp & (1 << i)); + track->tex_dirty = true; + } + break; + case RADEON_SE_VF_CNTL: + track->vap_vf_cntl = idx_value; + break; + case 0x210c: + /* VAP_VF_MAX_VTX_INDX */ + track->max_indx = idx_value & 0x00FFFFFFUL; + break; + case R200_SE_VTX_FMT_0: + track->vtx_size = r200_get_vtx_size_0(idx_value); + break; + case R200_SE_VTX_FMT_1: + track->vtx_size += r200_get_vtx_size_1(idx_value); + break; + case R200_PP_TXSIZE_0: + case R200_PP_TXSIZE_1: + case R200_PP_TXSIZE_2: + case R200_PP_TXSIZE_3: + case R200_PP_TXSIZE_4: + case R200_PP_TXSIZE_5: + i = (reg - R200_PP_TXSIZE_0) / 32; + track->textures[i].width = (idx_value & RADEON_TEX_USIZE_MASK) + 1; + track->textures[i].height = ((idx_value & RADEON_TEX_VSIZE_MASK) >> RADEON_TEX_VSIZE_SHIFT) + 1; + track->tex_dirty = true; + break; + case R200_PP_TXPITCH_0: + case R200_PP_TXPITCH_1: + case R200_PP_TXPITCH_2: + case R200_PP_TXPITCH_3: + case R200_PP_TXPITCH_4: + case R200_PP_TXPITCH_5: + i = (reg - R200_PP_TXPITCH_0) / 32; + track->textures[i].pitch = idx_value + 32; + track->tex_dirty = true; + break; + case R200_PP_TXFILTER_0: + case R200_PP_TXFILTER_1: + case R200_PP_TXFILTER_2: + case R200_PP_TXFILTER_3: + case R200_PP_TXFILTER_4: + case R200_PP_TXFILTER_5: + i = (reg - R200_PP_TXFILTER_0) / 32; + track->textures[i].num_levels = ((idx_value & R200_MAX_MIP_LEVEL_MASK) + >> R200_MAX_MIP_LEVEL_SHIFT); + tmp = (idx_value >> 23) & 0x7; + if (tmp == 2 || tmp == 6) + track->textures[i].roundup_w = false; + tmp = (idx_value >> 27) & 0x7; + if (tmp == 2 || tmp == 6) + track->textures[i].roundup_h = false; + track->tex_dirty = true; + break; + case R200_PP_TXMULTI_CTL_0: + case R200_PP_TXMULTI_CTL_1: + case R200_PP_TXMULTI_CTL_2: + case R200_PP_TXMULTI_CTL_3: + case R200_PP_TXMULTI_CTL_4: + case R200_PP_TXMULTI_CTL_5: + i = (reg - R200_PP_TXMULTI_CTL_0) / 32; + break; + case R200_PP_TXFORMAT_X_0: + case R200_PP_TXFORMAT_X_1: + case R200_PP_TXFORMAT_X_2: + case R200_PP_TXFORMAT_X_3: + case R200_PP_TXFORMAT_X_4: + case R200_PP_TXFORMAT_X_5: + i = (reg - R200_PP_TXFORMAT_X_0) / 32; + track->textures[i].txdepth = idx_value & 0x7; + tmp = (idx_value >> 16) & 0x3; + /* 2D, 3D, CUBE */ + switch (tmp) { + case 0: + case 3: + case 4: + case 5: + case 6: + case 7: + /* 1D/2D */ + track->textures[i].tex_coord_type = 0; + break; + case 1: + /* CUBE */ + track->textures[i].tex_coord_type = 2; + break; + case 2: + /* 3D */ + track->textures[i].tex_coord_type = 1; + break; + } + track->tex_dirty = true; + break; + case R200_PP_TXFORMAT_0: + case R200_PP_TXFORMAT_1: + case R200_PP_TXFORMAT_2: + case R200_PP_TXFORMAT_3: + case R200_PP_TXFORMAT_4: + case R200_PP_TXFORMAT_5: + i = (reg - R200_PP_TXFORMAT_0) / 32; + if (idx_value & R200_TXFORMAT_NON_POWER2) { + track->textures[i].use_pitch = 1; + } else { + track->textures[i].use_pitch = 0; + track->textures[i].width = 1 << ((idx_value >> RADEON_TXFORMAT_WIDTH_SHIFT) & RADEON_TXFORMAT_WIDTH_MASK); + track->textures[i].height = 1 << ((idx_value >> RADEON_TXFORMAT_HEIGHT_SHIFT) & RADEON_TXFORMAT_HEIGHT_MASK); + } + if (idx_value & R200_TXFORMAT_LOOKUP_DISABLE) + track->textures[i].lookup_disable = true; + switch ((idx_value & RADEON_TXFORMAT_FORMAT_MASK)) { + case R200_TXFORMAT_I8: + case R200_TXFORMAT_RGB332: + case R200_TXFORMAT_Y8: + track->textures[i].cpp = 1; + track->textures[i].compress_format = R100_TRACK_COMP_NONE; + break; + case R200_TXFORMAT_AI88: + case R200_TXFORMAT_ARGB1555: + case R200_TXFORMAT_RGB565: + case R200_TXFORMAT_ARGB4444: + case R200_TXFORMAT_VYUY422: + case R200_TXFORMAT_YVYU422: + case R200_TXFORMAT_LDVDU655: + case R200_TXFORMAT_DVDU88: + case R200_TXFORMAT_AVYU4444: + track->textures[i].cpp = 2; + track->textures[i].compress_format = R100_TRACK_COMP_NONE; + break; + case R200_TXFORMAT_ARGB8888: + case R200_TXFORMAT_RGBA8888: + case R200_TXFORMAT_ABGR8888: + case R200_TXFORMAT_BGR111110: + case R200_TXFORMAT_LDVDU8888: + track->textures[i].cpp = 4; + track->textures[i].compress_format = R100_TRACK_COMP_NONE; + break; + case R200_TXFORMAT_DXT1: + track->textures[i].cpp = 1; + track->textures[i].compress_format = R100_TRACK_COMP_DXT1; + break; + case R200_TXFORMAT_DXT23: + case R200_TXFORMAT_DXT45: + track->textures[i].cpp = 1; + track->textures[i].compress_format = R100_TRACK_COMP_DXT1; + break; + } + track->textures[i].cube_info[4].width = 1 << ((idx_value >> 16) & 0xf); + track->textures[i].cube_info[4].height = 1 << ((idx_value >> 20) & 0xf); + track->tex_dirty = true; + break; + case R200_PP_CUBIC_FACES_0: + case R200_PP_CUBIC_FACES_1: + case R200_PP_CUBIC_FACES_2: + case R200_PP_CUBIC_FACES_3: + case R200_PP_CUBIC_FACES_4: + case R200_PP_CUBIC_FACES_5: + tmp = idx_value; + i = (reg - R200_PP_CUBIC_FACES_0) / 32; + for (face = 0; face < 4; face++) { + track->textures[i].cube_info[face].width = 1 << ((tmp >> (face * 8)) & 0xf); + track->textures[i].cube_info[face].height = 1 << ((tmp >> ((face * 8) + 4)) & 0xf); + } + track->tex_dirty = true; + break; + default: + printk(KERN_ERR "Forbidden register 0x%04X in cs at %d\n", + reg, idx); + return -EINVAL; + } + return 0; +} + +void r200_set_safe_registers(struct radeon_device *rdev) +{ + rdev->config.r100.reg_safe_bm = r200_reg_safe_bm; + rdev->config.r100.reg_safe_bm_size = ARRAY_SIZE(r200_reg_safe_bm); +} diff --git a/drivers/gpu/drm/radeon/r300.c b/drivers/gpu/drm/radeon/r300.c new file mode 100644 index 00000000..55a7f190 --- /dev/null +++ b/drivers/gpu/drm/radeon/r300.c @@ -0,0 +1,1558 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#include +#include +#include +#include +#include +#include "radeon_reg.h" +#include "radeon.h" +#include "radeon_asic.h" +#include "radeon_drm.h" +#include "r100_track.h" +#include "r300d.h" +#include "rv350d.h" +#include "r300_reg_safe.h" + +/* This files gather functions specifics to: r300,r350,rv350,rv370,rv380 + * + * GPU Errata: + * - HOST_PATH_CNTL: r300 family seems to dislike write to HOST_PATH_CNTL + * using MMIO to flush host path read cache, this lead to HARDLOCKUP. + * However, scheduling such write to the ring seems harmless, i suspect + * the CP read collide with the flush somehow, or maybe the MC, hard to + * tell. (Jerome Glisse) + */ + +/* + * rv370,rv380 PCIE GART + */ +static int rv370_debugfs_pcie_gart_info_init(struct radeon_device *rdev); + +void rv370_pcie_gart_tlb_flush(struct radeon_device *rdev) +{ + uint32_t tmp; + int i; + + /* Workaround HW bug do flush 2 times */ + for (i = 0; i < 2; i++) { + tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_CNTL); + WREG32_PCIE(RADEON_PCIE_TX_GART_CNTL, tmp | RADEON_PCIE_TX_GART_INVALIDATE_TLB); + (void)RREG32_PCIE(RADEON_PCIE_TX_GART_CNTL); + WREG32_PCIE(RADEON_PCIE_TX_GART_CNTL, tmp); + } + mb(); +} + +#define R300_PTE_WRITEABLE (1 << 2) +#define R300_PTE_READABLE (1 << 3) + +int rv370_pcie_gart_set_page(struct radeon_device *rdev, int i, uint64_t addr) +{ + void __iomem *ptr = (void *)rdev->gart.table.vram.ptr; + + if (i < 0 || i > rdev->gart.num_gpu_pages) { + return -EINVAL; + } + addr = (lower_32_bits(addr) >> 8) | + ((upper_32_bits(addr) & 0xff) << 24) | + R300_PTE_WRITEABLE | R300_PTE_READABLE; + /* on x86 we want this to be CPU endian, on powerpc + * on powerpc without HW swappers, it'll get swapped on way + * into VRAM - so no need for cpu_to_le32 on VRAM tables */ + writel(addr, ((void __iomem *)ptr) + (i * 4)); + return 0; +} + +int rv370_pcie_gart_init(struct radeon_device *rdev) +{ + int r; + + if (rdev->gart.table.vram.robj) { + WARN(1, "RV370 PCIE GART already initialized\n"); + return 0; + } + /* Initialize common gart structure */ + r = radeon_gart_init(rdev); + if (r) + return r; + r = rv370_debugfs_pcie_gart_info_init(rdev); + if (r) + DRM_ERROR("Failed to register debugfs file for PCIE gart !\n"); + rdev->gart.table_size = rdev->gart.num_gpu_pages * 4; + rdev->asic->gart_tlb_flush = &rv370_pcie_gart_tlb_flush; + rdev->asic->gart_set_page = &rv370_pcie_gart_set_page; + return radeon_gart_table_vram_alloc(rdev); +} + +int rv370_pcie_gart_enable(struct radeon_device *rdev) +{ + uint32_t table_addr; + uint32_t tmp; + int r; + + if (rdev->gart.table.vram.robj == NULL) { + dev_err(rdev->dev, "No VRAM object for PCIE GART.\n"); + return -EINVAL; + } + r = radeon_gart_table_vram_pin(rdev); + if (r) + return r; + radeon_gart_restore(rdev); + /* discard memory request outside of configured range */ + tmp = RADEON_PCIE_TX_GART_UNMAPPED_ACCESS_DISCARD; + WREG32_PCIE(RADEON_PCIE_TX_GART_CNTL, tmp); + WREG32_PCIE(RADEON_PCIE_TX_GART_START_LO, rdev->mc.gtt_start); + tmp = rdev->mc.gtt_end & ~RADEON_GPU_PAGE_MASK; + WREG32_PCIE(RADEON_PCIE_TX_GART_END_LO, tmp); + WREG32_PCIE(RADEON_PCIE_TX_GART_START_HI, 0); + WREG32_PCIE(RADEON_PCIE_TX_GART_END_HI, 0); + table_addr = rdev->gart.table_addr; + WREG32_PCIE(RADEON_PCIE_TX_GART_BASE, table_addr); + /* FIXME: setup default page */ + WREG32_PCIE(RADEON_PCIE_TX_DISCARD_RD_ADDR_LO, rdev->mc.vram_start); + WREG32_PCIE(RADEON_PCIE_TX_DISCARD_RD_ADDR_HI, 0); + /* Clear error */ + WREG32_PCIE(RADEON_PCIE_TX_GART_ERROR, 0); + tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_CNTL); + tmp |= RADEON_PCIE_TX_GART_EN; + tmp |= RADEON_PCIE_TX_GART_UNMAPPED_ACCESS_DISCARD; + WREG32_PCIE(RADEON_PCIE_TX_GART_CNTL, tmp); + rv370_pcie_gart_tlb_flush(rdev); + DRM_INFO("PCIE GART of %uM enabled (table at 0x%08X).\n", + (unsigned)(rdev->mc.gtt_size >> 20), table_addr); + rdev->gart.ready = true; + return 0; +} + +void rv370_pcie_gart_disable(struct radeon_device *rdev) +{ + u32 tmp; + int r; + + WREG32_PCIE(RADEON_PCIE_TX_GART_START_LO, 0); + WREG32_PCIE(RADEON_PCIE_TX_GART_END_LO, 0); + WREG32_PCIE(RADEON_PCIE_TX_GART_START_HI, 0); + WREG32_PCIE(RADEON_PCIE_TX_GART_END_HI, 0); + tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_CNTL); + tmp |= RADEON_PCIE_TX_GART_UNMAPPED_ACCESS_DISCARD; + WREG32_PCIE(RADEON_PCIE_TX_GART_CNTL, tmp & ~RADEON_PCIE_TX_GART_EN); + if (rdev->gart.table.vram.robj) { + r = radeon_bo_reserve(rdev->gart.table.vram.robj, false); + if (likely(r == 0)) { + radeon_bo_kunmap(rdev->gart.table.vram.robj); + radeon_bo_unpin(rdev->gart.table.vram.robj); + radeon_bo_unreserve(rdev->gart.table.vram.robj); + } + } +} + +void rv370_pcie_gart_fini(struct radeon_device *rdev) +{ + radeon_gart_fini(rdev); + rv370_pcie_gart_disable(rdev); + radeon_gart_table_vram_free(rdev); +} + +void r300_fence_ring_emit(struct radeon_device *rdev, + struct radeon_fence *fence) +{ + /* Who ever call radeon_fence_emit should call ring_lock and ask + * for enough space (today caller are ib schedule and buffer move) */ + /* Write SC register so SC & US assert idle */ + radeon_ring_write(rdev, PACKET0(R300_RE_SCISSORS_TL, 0)); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, PACKET0(R300_RE_SCISSORS_BR, 0)); + radeon_ring_write(rdev, 0); + /* Flush 3D cache */ + radeon_ring_write(rdev, PACKET0(R300_RB3D_DSTCACHE_CTLSTAT, 0)); + radeon_ring_write(rdev, R300_RB3D_DC_FLUSH); + radeon_ring_write(rdev, PACKET0(R300_RB3D_ZCACHE_CTLSTAT, 0)); + radeon_ring_write(rdev, R300_ZC_FLUSH); + /* Wait until IDLE & CLEAN */ + radeon_ring_write(rdev, PACKET0(RADEON_WAIT_UNTIL, 0)); + radeon_ring_write(rdev, (RADEON_WAIT_3D_IDLECLEAN | + RADEON_WAIT_2D_IDLECLEAN | + RADEON_WAIT_DMA_GUI_IDLE)); + radeon_ring_write(rdev, PACKET0(RADEON_HOST_PATH_CNTL, 0)); + radeon_ring_write(rdev, rdev->config.r300.hdp_cntl | + RADEON_HDP_READ_BUFFER_INVALIDATE); + radeon_ring_write(rdev, PACKET0(RADEON_HOST_PATH_CNTL, 0)); + radeon_ring_write(rdev, rdev->config.r300.hdp_cntl); + /* Emit fence sequence & fire IRQ */ + radeon_ring_write(rdev, PACKET0(rdev->fence_drv.scratch_reg, 0)); + radeon_ring_write(rdev, fence->seq); + radeon_ring_write(rdev, PACKET0(RADEON_GEN_INT_STATUS, 0)); + radeon_ring_write(rdev, RADEON_SW_INT_FIRE); +} + +void r300_ring_start(struct radeon_device *rdev) +{ + unsigned gb_tile_config; + int r; + + /* Sub pixel 1/12 so we can have 4K rendering according to doc */ + gb_tile_config = (R300_ENABLE_TILING | R300_TILE_SIZE_16); + switch(rdev->num_gb_pipes) { + case 2: + gb_tile_config |= R300_PIPE_COUNT_R300; + break; + case 3: + gb_tile_config |= R300_PIPE_COUNT_R420_3P; + break; + case 4: + gb_tile_config |= R300_PIPE_COUNT_R420; + break; + case 1: + default: + gb_tile_config |= R300_PIPE_COUNT_RV350; + break; + } + + r = radeon_ring_lock(rdev, 64); + if (r) { + return; + } + radeon_ring_write(rdev, PACKET0(RADEON_ISYNC_CNTL, 0)); + radeon_ring_write(rdev, + RADEON_ISYNC_ANY2D_IDLE3D | + RADEON_ISYNC_ANY3D_IDLE2D | + RADEON_ISYNC_WAIT_IDLEGUI | + RADEON_ISYNC_CPSCRATCH_IDLEGUI); + radeon_ring_write(rdev, PACKET0(R300_GB_TILE_CONFIG, 0)); + radeon_ring_write(rdev, gb_tile_config); + radeon_ring_write(rdev, PACKET0(RADEON_WAIT_UNTIL, 0)); + radeon_ring_write(rdev, + RADEON_WAIT_2D_IDLECLEAN | + RADEON_WAIT_3D_IDLECLEAN); + radeon_ring_write(rdev, PACKET0(R300_DST_PIPE_CONFIG, 0)); + radeon_ring_write(rdev, R300_PIPE_AUTO_CONFIG); + radeon_ring_write(rdev, PACKET0(R300_GB_SELECT, 0)); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, PACKET0(R300_GB_ENABLE, 0)); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, PACKET0(R300_RB3D_DSTCACHE_CTLSTAT, 0)); + radeon_ring_write(rdev, R300_RB3D_DC_FLUSH | R300_RB3D_DC_FREE); + radeon_ring_write(rdev, PACKET0(R300_RB3D_ZCACHE_CTLSTAT, 0)); + radeon_ring_write(rdev, R300_ZC_FLUSH | R300_ZC_FREE); + radeon_ring_write(rdev, PACKET0(RADEON_WAIT_UNTIL, 0)); + radeon_ring_write(rdev, + RADEON_WAIT_2D_IDLECLEAN | + RADEON_WAIT_3D_IDLECLEAN); + radeon_ring_write(rdev, PACKET0(R300_GB_AA_CONFIG, 0)); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, PACKET0(R300_RB3D_DSTCACHE_CTLSTAT, 0)); + radeon_ring_write(rdev, R300_RB3D_DC_FLUSH | R300_RB3D_DC_FREE); + radeon_ring_write(rdev, PACKET0(R300_RB3D_ZCACHE_CTLSTAT, 0)); + radeon_ring_write(rdev, R300_ZC_FLUSH | R300_ZC_FREE); + radeon_ring_write(rdev, PACKET0(R300_GB_MSPOS0, 0)); + radeon_ring_write(rdev, + ((6 << R300_MS_X0_SHIFT) | + (6 << R300_MS_Y0_SHIFT) | + (6 << R300_MS_X1_SHIFT) | + (6 << R300_MS_Y1_SHIFT) | + (6 << R300_MS_X2_SHIFT) | + (6 << R300_MS_Y2_SHIFT) | + (6 << R300_MSBD0_Y_SHIFT) | + (6 << R300_MSBD0_X_SHIFT))); + radeon_ring_write(rdev, PACKET0(R300_GB_MSPOS1, 0)); + radeon_ring_write(rdev, + ((6 << R300_MS_X3_SHIFT) | + (6 << R300_MS_Y3_SHIFT) | + (6 << R300_MS_X4_SHIFT) | + (6 << R300_MS_Y4_SHIFT) | + (6 << R300_MS_X5_SHIFT) | + (6 << R300_MS_Y5_SHIFT) | + (6 << R300_MSBD1_SHIFT))); + radeon_ring_write(rdev, PACKET0(R300_GA_ENHANCE, 0)); + radeon_ring_write(rdev, R300_GA_DEADLOCK_CNTL | R300_GA_FASTSYNC_CNTL); + radeon_ring_write(rdev, PACKET0(R300_GA_POLY_MODE, 0)); + radeon_ring_write(rdev, + R300_FRONT_PTYPE_TRIANGE | R300_BACK_PTYPE_TRIANGE); + radeon_ring_write(rdev, PACKET0(R300_GA_ROUND_MODE, 0)); + radeon_ring_write(rdev, + R300_GEOMETRY_ROUND_NEAREST | + R300_COLOR_ROUND_NEAREST); + radeon_ring_unlock_commit(rdev); +} + +void r300_errata(struct radeon_device *rdev) +{ + rdev->pll_errata = 0; + + if (rdev->family == CHIP_R300 && + (RREG32(RADEON_CONFIG_CNTL) & RADEON_CFG_ATI_REV_ID_MASK) == RADEON_CFG_ATI_REV_A11) { + rdev->pll_errata |= CHIP_ERRATA_R300_CG; + } +} + +int r300_mc_wait_for_idle(struct radeon_device *rdev) +{ + unsigned i; + uint32_t tmp; + + for (i = 0; i < rdev->usec_timeout; i++) { + /* read MC_STATUS */ + tmp = RREG32(RADEON_MC_STATUS); + if (tmp & R300_MC_IDLE) { + return 0; + } + DRM_UDELAY(1); + } + return -1; +} + +void r300_gpu_init(struct radeon_device *rdev) +{ + uint32_t gb_tile_config, tmp; + + if ((rdev->family == CHIP_R300 && rdev->pdev->device != 0x4144) || + (rdev->family == CHIP_R350 && rdev->pdev->device != 0x4148)) { + /* r300,r350 */ + rdev->num_gb_pipes = 2; + } else { + /* rv350,rv370,rv380,r300 AD, r350 AH */ + rdev->num_gb_pipes = 1; + } + rdev->num_z_pipes = 1; + gb_tile_config = (R300_ENABLE_TILING | R300_TILE_SIZE_16); + switch (rdev->num_gb_pipes) { + case 2: + gb_tile_config |= R300_PIPE_COUNT_R300; + break; + case 3: + gb_tile_config |= R300_PIPE_COUNT_R420_3P; + break; + case 4: + gb_tile_config |= R300_PIPE_COUNT_R420; + break; + default: + case 1: + gb_tile_config |= R300_PIPE_COUNT_RV350; + break; + } + WREG32(R300_GB_TILE_CONFIG, gb_tile_config); + + if (r100_gui_wait_for_idle(rdev)) { + printk(KERN_WARNING "Failed to wait GUI idle while " + "programming pipes. Bad things might happen.\n"); + } + + tmp = RREG32(R300_DST_PIPE_CONFIG); + WREG32(R300_DST_PIPE_CONFIG, tmp | R300_PIPE_AUTO_CONFIG); + + WREG32(R300_RB2D_DSTCACHE_MODE, + R300_DC_AUTOFLUSH_ENABLE | + R300_DC_DC_DISABLE_IGNORE_PE); + + if (r100_gui_wait_for_idle(rdev)) { + printk(KERN_WARNING "Failed to wait GUI idle while " + "programming pipes. Bad things might happen.\n"); + } + if (r300_mc_wait_for_idle(rdev)) { + printk(KERN_WARNING "Failed to wait MC idle while " + "programming pipes. Bad things might happen.\n"); + } + DRM_INFO("radeon: %d quad pipes, %d Z pipes initialized.\n", + rdev->num_gb_pipes, rdev->num_z_pipes); +} + +bool r300_gpu_is_lockup(struct radeon_device *rdev) +{ + u32 rbbm_status; + int r; + + rbbm_status = RREG32(R_000E40_RBBM_STATUS); + if (!G_000E40_GUI_ACTIVE(rbbm_status)) { + r100_gpu_lockup_update(&rdev->config.r300.lockup, &rdev->cp); + return false; + } + /* force CP activities */ + r = radeon_ring_lock(rdev, 2); + if (!r) { + /* PACKET2 NOP */ + radeon_ring_write(rdev, 0x80000000); + radeon_ring_write(rdev, 0x80000000); + radeon_ring_unlock_commit(rdev); + } + rdev->cp.rptr = RREG32(RADEON_CP_RB_RPTR); + return r100_gpu_cp_is_lockup(rdev, &rdev->config.r300.lockup, &rdev->cp); +} + +int r300_asic_reset(struct radeon_device *rdev) +{ + struct r100_mc_save save; + u32 status, tmp; + int ret = 0; + + status = RREG32(R_000E40_RBBM_STATUS); + if (!G_000E40_GUI_ACTIVE(status)) { + return 0; + } + r100_mc_stop(rdev, &save); + status = RREG32(R_000E40_RBBM_STATUS); + dev_info(rdev->dev, "(%s:%d) RBBM_STATUS=0x%08X\n", __func__, __LINE__, status); + /* stop CP */ + WREG32(RADEON_CP_CSQ_CNTL, 0); + tmp = RREG32(RADEON_CP_RB_CNTL); + WREG32(RADEON_CP_RB_CNTL, tmp | RADEON_RB_RPTR_WR_ENA); + WREG32(RADEON_CP_RB_RPTR_WR, 0); + WREG32(RADEON_CP_RB_WPTR, 0); + WREG32(RADEON_CP_RB_CNTL, tmp); + /* save PCI state */ + pci_save_state(rdev->pdev); + /* disable bus mastering */ + r100_bm_disable(rdev); + WREG32(R_0000F0_RBBM_SOFT_RESET, S_0000F0_SOFT_RESET_VAP(1) | + S_0000F0_SOFT_RESET_GA(1)); + RREG32(R_0000F0_RBBM_SOFT_RESET); + mdelay(500); + WREG32(R_0000F0_RBBM_SOFT_RESET, 0); + mdelay(1); + status = RREG32(R_000E40_RBBM_STATUS); + dev_info(rdev->dev, "(%s:%d) RBBM_STATUS=0x%08X\n", __func__, __LINE__, status); + /* resetting the CP seems to be problematic sometimes it end up + * hard locking the computer, but it's necessary for successful + * reset more test & playing is needed on R3XX/R4XX to find a + * reliable (if any solution) + */ + WREG32(R_0000F0_RBBM_SOFT_RESET, S_0000F0_SOFT_RESET_CP(1)); + RREG32(R_0000F0_RBBM_SOFT_RESET); + mdelay(500); + WREG32(R_0000F0_RBBM_SOFT_RESET, 0); + mdelay(1); + status = RREG32(R_000E40_RBBM_STATUS); + dev_info(rdev->dev, "(%s:%d) RBBM_STATUS=0x%08X\n", __func__, __LINE__, status); + /* restore PCI & busmastering */ + pci_restore_state(rdev->pdev); + r100_enable_bm(rdev); + /* Check if GPU is idle */ + if (G_000E40_GA_BUSY(status) || G_000E40_VAP_BUSY(status)) { + dev_err(rdev->dev, "failed to reset GPU\n"); + rdev->gpu_lockup = true; + ret = -1; + } else + dev_info(rdev->dev, "GPU reset succeed\n"); + r100_mc_resume(rdev, &save); + return ret; +} + +/* + * r300,r350,rv350,rv380 VRAM info + */ +void r300_mc_init(struct radeon_device *rdev) +{ + u64 base; + u32 tmp; + + /* DDR for all card after R300 & IGP */ + rdev->mc.vram_is_ddr = true; + tmp = RREG32(RADEON_MEM_CNTL); + tmp &= R300_MEM_NUM_CHANNELS_MASK; + switch (tmp) { + case 0: rdev->mc.vram_width = 64; break; + case 1: rdev->mc.vram_width = 128; break; + case 2: rdev->mc.vram_width = 256; break; + default: rdev->mc.vram_width = 128; break; + } + r100_vram_init_sizes(rdev); + base = rdev->mc.aper_base; + if (rdev->flags & RADEON_IS_IGP) + base = (RREG32(RADEON_NB_TOM) & 0xffff) << 16; + radeon_vram_location(rdev, &rdev->mc, base); + rdev->mc.gtt_base_align = 0; + if (!(rdev->flags & RADEON_IS_AGP)) + radeon_gtt_location(rdev, &rdev->mc); + radeon_update_bandwidth_info(rdev); +} + +void rv370_set_pcie_lanes(struct radeon_device *rdev, int lanes) +{ + uint32_t link_width_cntl, mask; + + if (rdev->flags & RADEON_IS_IGP) + return; + + if (!(rdev->flags & RADEON_IS_PCIE)) + return; + + /* FIXME wait for idle */ + + switch (lanes) { + case 0: + mask = RADEON_PCIE_LC_LINK_WIDTH_X0; + break; + case 1: + mask = RADEON_PCIE_LC_LINK_WIDTH_X1; + break; + case 2: + mask = RADEON_PCIE_LC_LINK_WIDTH_X2; + break; + case 4: + mask = RADEON_PCIE_LC_LINK_WIDTH_X4; + break; + case 8: + mask = RADEON_PCIE_LC_LINK_WIDTH_X8; + break; + case 12: + mask = RADEON_PCIE_LC_LINK_WIDTH_X12; + break; + case 16: + default: + mask = RADEON_PCIE_LC_LINK_WIDTH_X16; + break; + } + + link_width_cntl = RREG32_PCIE(RADEON_PCIE_LC_LINK_WIDTH_CNTL); + + if ((link_width_cntl & RADEON_PCIE_LC_LINK_WIDTH_RD_MASK) == + (mask << RADEON_PCIE_LC_LINK_WIDTH_RD_SHIFT)) + return; + + link_width_cntl &= ~(RADEON_PCIE_LC_LINK_WIDTH_MASK | + RADEON_PCIE_LC_RECONFIG_NOW | + RADEON_PCIE_LC_RECONFIG_LATER | + RADEON_PCIE_LC_SHORT_RECONFIG_EN); + link_width_cntl |= mask; + WREG32_PCIE(RADEON_PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl); + WREG32_PCIE(RADEON_PCIE_LC_LINK_WIDTH_CNTL, (link_width_cntl | + RADEON_PCIE_LC_RECONFIG_NOW)); + + /* wait for lane set to complete */ + link_width_cntl = RREG32_PCIE(RADEON_PCIE_LC_LINK_WIDTH_CNTL); + while (link_width_cntl == 0xffffffff) + link_width_cntl = RREG32_PCIE(RADEON_PCIE_LC_LINK_WIDTH_CNTL); + +} + +int rv370_get_pcie_lanes(struct radeon_device *rdev) +{ + u32 link_width_cntl; + + if (rdev->flags & RADEON_IS_IGP) + return 0; + + if (!(rdev->flags & RADEON_IS_PCIE)) + return 0; + + /* FIXME wait for idle */ + + link_width_cntl = RREG32_PCIE(RADEON_PCIE_LC_LINK_WIDTH_CNTL); + + switch ((link_width_cntl & RADEON_PCIE_LC_LINK_WIDTH_RD_MASK) >> RADEON_PCIE_LC_LINK_WIDTH_RD_SHIFT) { + case RADEON_PCIE_LC_LINK_WIDTH_X0: + return 0; + case RADEON_PCIE_LC_LINK_WIDTH_X1: + return 1; + case RADEON_PCIE_LC_LINK_WIDTH_X2: + return 2; + case RADEON_PCIE_LC_LINK_WIDTH_X4: + return 4; + case RADEON_PCIE_LC_LINK_WIDTH_X8: + return 8; + case RADEON_PCIE_LC_LINK_WIDTH_X16: + default: + return 16; + } +} + +#if defined(CONFIG_DEBUG_FS) +static int rv370_debugfs_pcie_gart_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + struct radeon_device *rdev = dev->dev_private; + uint32_t tmp; + + tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_CNTL); + seq_printf(m, "PCIE_TX_GART_CNTL 0x%08x\n", tmp); + tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_BASE); + seq_printf(m, "PCIE_TX_GART_BASE 0x%08x\n", tmp); + tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_START_LO); + seq_printf(m, "PCIE_TX_GART_START_LO 0x%08x\n", tmp); + tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_START_HI); + seq_printf(m, "PCIE_TX_GART_START_HI 0x%08x\n", tmp); + tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_END_LO); + seq_printf(m, "PCIE_TX_GART_END_LO 0x%08x\n", tmp); + tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_END_HI); + seq_printf(m, "PCIE_TX_GART_END_HI 0x%08x\n", tmp); + tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_ERROR); + seq_printf(m, "PCIE_TX_GART_ERROR 0x%08x\n", tmp); + return 0; +} + +static struct drm_info_list rv370_pcie_gart_info_list[] = { + {"rv370_pcie_gart_info", rv370_debugfs_pcie_gart_info, 0, NULL}, +}; +#endif + +static int rv370_debugfs_pcie_gart_info_init(struct radeon_device *rdev) +{ +#if defined(CONFIG_DEBUG_FS) + return radeon_debugfs_add_files(rdev, rv370_pcie_gart_info_list, 1); +#else + return 0; +#endif +} + +static int r300_packet0_check(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt, + unsigned idx, unsigned reg) +{ + struct radeon_cs_reloc *reloc; + struct r100_cs_track *track; + volatile uint32_t *ib; + uint32_t tmp, tile_flags = 0; + unsigned i; + int r; + u32 idx_value; + + ib = p->ib->ptr; + track = (struct r100_cs_track *)p->track; + idx_value = radeon_get_ib_value(p, idx); + + switch(reg) { + case AVIVO_D1MODE_VLINE_START_END: + case RADEON_CRTC_GUI_TRIG_VLINE: + r = r100_cs_packet_parse_vline(p); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + break; + case RADEON_DST_PITCH_OFFSET: + case RADEON_SRC_PITCH_OFFSET: + r = r100_reloc_pitch_offset(p, pkt, idx, reg); + if (r) + return r; + break; + case R300_RB3D_COLOROFFSET0: + case R300_RB3D_COLOROFFSET1: + case R300_RB3D_COLOROFFSET2: + case R300_RB3D_COLOROFFSET3: + i = (reg - R300_RB3D_COLOROFFSET0) >> 2; + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + track->cb[i].robj = reloc->robj; + track->cb[i].offset = idx_value; + track->cb_dirty = true; + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + break; + case R300_ZB_DEPTHOFFSET: + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + track->zb.robj = reloc->robj; + track->zb.offset = idx_value; + track->zb_dirty = true; + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + break; + case R300_TX_OFFSET_0: + case R300_TX_OFFSET_0+4: + case R300_TX_OFFSET_0+8: + case R300_TX_OFFSET_0+12: + case R300_TX_OFFSET_0+16: + case R300_TX_OFFSET_0+20: + case R300_TX_OFFSET_0+24: + case R300_TX_OFFSET_0+28: + case R300_TX_OFFSET_0+32: + case R300_TX_OFFSET_0+36: + case R300_TX_OFFSET_0+40: + case R300_TX_OFFSET_0+44: + case R300_TX_OFFSET_0+48: + case R300_TX_OFFSET_0+52: + case R300_TX_OFFSET_0+56: + case R300_TX_OFFSET_0+60: + i = (reg - R300_TX_OFFSET_0) >> 2; + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + + if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) + tile_flags |= R300_TXO_MACRO_TILE; + if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) + tile_flags |= R300_TXO_MICRO_TILE; + else if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO_SQUARE) + tile_flags |= R300_TXO_MICRO_TILE_SQUARE; + + tmp = idx_value + ((u32)reloc->lobj.gpu_offset); + tmp |= tile_flags; + ib[idx] = tmp; + track->textures[i].robj = reloc->robj; + track->tex_dirty = true; + break; + /* Tracked registers */ + case 0x2084: + /* VAP_VF_CNTL */ + track->vap_vf_cntl = idx_value; + break; + case 0x20B4: + /* VAP_VTX_SIZE */ + track->vtx_size = idx_value & 0x7F; + break; + case 0x2134: + /* VAP_VF_MAX_VTX_INDX */ + track->max_indx = idx_value & 0x00FFFFFFUL; + break; + case 0x2088: + /* VAP_ALT_NUM_VERTICES - only valid on r500 */ + if (p->rdev->family < CHIP_RV515) + goto fail; + track->vap_alt_nverts = idx_value & 0xFFFFFF; + break; + case 0x43E4: + /* SC_SCISSOR1 */ + track->maxy = ((idx_value >> 13) & 0x1FFF) + 1; + if (p->rdev->family < CHIP_RV515) { + track->maxy -= 1440; + } + track->cb_dirty = true; + track->zb_dirty = true; + break; + case 0x4E00: + /* RB3D_CCTL */ + if ((idx_value & (1 << 10)) && /* CMASK_ENABLE */ + p->rdev->cmask_filp != p->filp) { + DRM_ERROR("Invalid RB3D_CCTL: Cannot enable CMASK.\n"); + return -EINVAL; + } + track->num_cb = ((idx_value >> 5) & 0x3) + 1; + track->cb_dirty = true; + break; + case 0x4E38: + case 0x4E3C: + case 0x4E40: + case 0x4E44: + /* RB3D_COLORPITCH0 */ + /* RB3D_COLORPITCH1 */ + /* RB3D_COLORPITCH2 */ + /* RB3D_COLORPITCH3 */ + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + + if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) + tile_flags |= R300_COLOR_TILE_ENABLE; + if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) + tile_flags |= R300_COLOR_MICROTILE_ENABLE; + else if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO_SQUARE) + tile_flags |= R300_COLOR_MICROTILE_SQUARE_ENABLE; + + tmp = idx_value & ~(0x7 << 16); + tmp |= tile_flags; + ib[idx] = tmp; + i = (reg - 0x4E38) >> 2; + track->cb[i].pitch = idx_value & 0x3FFE; + switch (((idx_value >> 21) & 0xF)) { + case 9: + case 11: + case 12: + track->cb[i].cpp = 1; + break; + case 3: + case 4: + case 13: + case 15: + track->cb[i].cpp = 2; + break; + case 5: + if (p->rdev->family < CHIP_RV515) { + DRM_ERROR("Invalid color buffer format (%d)!\n", + ((idx_value >> 21) & 0xF)); + return -EINVAL; + } + /* Pass through. */ + case 6: + track->cb[i].cpp = 4; + break; + case 10: + track->cb[i].cpp = 8; + break; + case 7: + track->cb[i].cpp = 16; + break; + default: + DRM_ERROR("Invalid color buffer format (%d) !\n", + ((idx_value >> 21) & 0xF)); + return -EINVAL; + } + track->cb_dirty = true; + break; + case 0x4F00: + /* ZB_CNTL */ + if (idx_value & 2) { + track->z_enabled = true; + } else { + track->z_enabled = false; + } + track->zb_dirty = true; + break; + case 0x4F10: + /* ZB_FORMAT */ + switch ((idx_value & 0xF)) { + case 0: + case 1: + track->zb.cpp = 2; + break; + case 2: + track->zb.cpp = 4; + break; + default: + DRM_ERROR("Invalid z buffer format (%d) !\n", + (idx_value & 0xF)); + return -EINVAL; + } + track->zb_dirty = true; + break; + case 0x4F24: + /* ZB_DEPTHPITCH */ + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + + if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) + tile_flags |= R300_DEPTHMACROTILE_ENABLE; + if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) + tile_flags |= R300_DEPTHMICROTILE_TILED; + else if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO_SQUARE) + tile_flags |= R300_DEPTHMICROTILE_TILED_SQUARE; + + tmp = idx_value & ~(0x7 << 16); + tmp |= tile_flags; + ib[idx] = tmp; + + track->zb.pitch = idx_value & 0x3FFC; + track->zb_dirty = true; + break; + case 0x4104: + /* TX_ENABLE */ + for (i = 0; i < 16; i++) { + bool enabled; + + enabled = !!(idx_value & (1 << i)); + track->textures[i].enabled = enabled; + } + track->tex_dirty = true; + break; + case 0x44C0: + case 0x44C4: + case 0x44C8: + case 0x44CC: + case 0x44D0: + case 0x44D4: + case 0x44D8: + case 0x44DC: + case 0x44E0: + case 0x44E4: + case 0x44E8: + case 0x44EC: + case 0x44F0: + case 0x44F4: + case 0x44F8: + case 0x44FC: + /* TX_FORMAT1_[0-15] */ + i = (reg - 0x44C0) >> 2; + tmp = (idx_value >> 25) & 0x3; + track->textures[i].tex_coord_type = tmp; + switch ((idx_value & 0x1F)) { + case R300_TX_FORMAT_X8: + case R300_TX_FORMAT_Y4X4: + case R300_TX_FORMAT_Z3Y3X2: + track->textures[i].cpp = 1; + track->textures[i].compress_format = R100_TRACK_COMP_NONE; + break; + case R300_TX_FORMAT_X16: + case R300_TX_FORMAT_FL_I16: + case R300_TX_FORMAT_Y8X8: + case R300_TX_FORMAT_Z5Y6X5: + case R300_TX_FORMAT_Z6Y5X5: + case R300_TX_FORMAT_W4Z4Y4X4: + case R300_TX_FORMAT_W1Z5Y5X5: + case R300_TX_FORMAT_D3DMFT_CxV8U8: + case R300_TX_FORMAT_B8G8_B8G8: + case R300_TX_FORMAT_G8R8_G8B8: + track->textures[i].cpp = 2; + track->textures[i].compress_format = R100_TRACK_COMP_NONE; + break; + case R300_TX_FORMAT_Y16X16: + case R300_TX_FORMAT_FL_I16A16: + case R300_TX_FORMAT_Z11Y11X10: + case R300_TX_FORMAT_Z10Y11X11: + case R300_TX_FORMAT_W8Z8Y8X8: + case R300_TX_FORMAT_W2Z10Y10X10: + case 0x17: + case R300_TX_FORMAT_FL_I32: + case 0x1e: + track->textures[i].cpp = 4; + track->textures[i].compress_format = R100_TRACK_COMP_NONE; + break; + case R300_TX_FORMAT_W16Z16Y16X16: + case R300_TX_FORMAT_FL_R16G16B16A16: + case R300_TX_FORMAT_FL_I32A32: + track->textures[i].cpp = 8; + track->textures[i].compress_format = R100_TRACK_COMP_NONE; + break; + case R300_TX_FORMAT_FL_R32G32B32A32: + track->textures[i].cpp = 16; + track->textures[i].compress_format = R100_TRACK_COMP_NONE; + break; + case R300_TX_FORMAT_DXT1: + track->textures[i].cpp = 1; + track->textures[i].compress_format = R100_TRACK_COMP_DXT1; + break; + case R300_TX_FORMAT_ATI2N: + if (p->rdev->family < CHIP_R420) { + DRM_ERROR("Invalid texture format %u\n", + (idx_value & 0x1F)); + return -EINVAL; + } + /* The same rules apply as for DXT3/5. */ + /* Pass through. */ + case R300_TX_FORMAT_DXT3: + case R300_TX_FORMAT_DXT5: + track->textures[i].cpp = 1; + track->textures[i].compress_format = R100_TRACK_COMP_DXT35; + break; + default: + DRM_ERROR("Invalid texture format %u\n", + (idx_value & 0x1F)); + return -EINVAL; + } + track->tex_dirty = true; + break; + case 0x4400: + case 0x4404: + case 0x4408: + case 0x440C: + case 0x4410: + case 0x4414: + case 0x4418: + case 0x441C: + case 0x4420: + case 0x4424: + case 0x4428: + case 0x442C: + case 0x4430: + case 0x4434: + case 0x4438: + case 0x443C: + /* TX_FILTER0_[0-15] */ + i = (reg - 0x4400) >> 2; + tmp = idx_value & 0x7; + if (tmp == 2 || tmp == 4 || tmp == 6) { + track->textures[i].roundup_w = false; + } + tmp = (idx_value >> 3) & 0x7; + if (tmp == 2 || tmp == 4 || tmp == 6) { + track->textures[i].roundup_h = false; + } + track->tex_dirty = true; + break; + case 0x4500: + case 0x4504: + case 0x4508: + case 0x450C: + case 0x4510: + case 0x4514: + case 0x4518: + case 0x451C: + case 0x4520: + case 0x4524: + case 0x4528: + case 0x452C: + case 0x4530: + case 0x4534: + case 0x4538: + case 0x453C: + /* TX_FORMAT2_[0-15] */ + i = (reg - 0x4500) >> 2; + tmp = idx_value & 0x3FFF; + track->textures[i].pitch = tmp + 1; + if (p->rdev->family >= CHIP_RV515) { + tmp = ((idx_value >> 15) & 1) << 11; + track->textures[i].width_11 = tmp; + tmp = ((idx_value >> 16) & 1) << 11; + track->textures[i].height_11 = tmp; + + /* ATI1N */ + if (idx_value & (1 << 14)) { + /* The same rules apply as for DXT1. */ + track->textures[i].compress_format = + R100_TRACK_COMP_DXT1; + } + } else if (idx_value & (1 << 14)) { + DRM_ERROR("Forbidden bit TXFORMAT_MSB\n"); + return -EINVAL; + } + track->tex_dirty = true; + break; + case 0x4480: + case 0x4484: + case 0x4488: + case 0x448C: + case 0x4490: + case 0x4494: + case 0x4498: + case 0x449C: + case 0x44A0: + case 0x44A4: + case 0x44A8: + case 0x44AC: + case 0x44B0: + case 0x44B4: + case 0x44B8: + case 0x44BC: + /* TX_FORMAT0_[0-15] */ + i = (reg - 0x4480) >> 2; + tmp = idx_value & 0x7FF; + track->textures[i].width = tmp + 1; + tmp = (idx_value >> 11) & 0x7FF; + track->textures[i].height = tmp + 1; + tmp = (idx_value >> 26) & 0xF; + track->textures[i].num_levels = tmp; + tmp = idx_value & (1 << 31); + track->textures[i].use_pitch = !!tmp; + tmp = (idx_value >> 22) & 0xF; + track->textures[i].txdepth = tmp; + track->tex_dirty = true; + break; + case R300_ZB_ZPASS_ADDR: + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + break; + case 0x4e0c: + /* RB3D_COLOR_CHANNEL_MASK */ + track->color_channel_mask = idx_value; + track->cb_dirty = true; + break; + case 0x43a4: + /* SC_HYPERZ_EN */ + /* r300c emits this register - we need to disable hyperz for it + * without complaining */ + if (p->rdev->hyperz_filp != p->filp) { + if (idx_value & 0x1) + ib[idx] = idx_value & ~1; + } + break; + case 0x4f1c: + /* ZB_BW_CNTL */ + track->zb_cb_clear = !!(idx_value & (1 << 5)); + track->cb_dirty = true; + track->zb_dirty = true; + if (p->rdev->hyperz_filp != p->filp) { + if (idx_value & (R300_HIZ_ENABLE | + R300_RD_COMP_ENABLE | + R300_WR_COMP_ENABLE | + R300_FAST_FILL_ENABLE)) + goto fail; + } + break; + case 0x4e04: + /* RB3D_BLENDCNTL */ + track->blend_read_enable = !!(idx_value & (1 << 2)); + track->cb_dirty = true; + break; + case R300_RB3D_AARESOLVE_OFFSET: + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + track->aa.robj = reloc->robj; + track->aa.offset = idx_value; + track->aa_dirty = true; + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + break; + case R300_RB3D_AARESOLVE_PITCH: + track->aa.pitch = idx_value & 0x3FFE; + track->aa_dirty = true; + break; + case R300_RB3D_AARESOLVE_CTL: + track->aaresolve = idx_value & 0x1; + track->aa_dirty = true; + break; + case 0x4f30: /* ZB_MASK_OFFSET */ + case 0x4f34: /* ZB_ZMASK_PITCH */ + case 0x4f44: /* ZB_HIZ_OFFSET */ + case 0x4f54: /* ZB_HIZ_PITCH */ + if (idx_value && (p->rdev->hyperz_filp != p->filp)) + goto fail; + break; + case 0x4028: + if (idx_value && (p->rdev->hyperz_filp != p->filp)) + goto fail; + /* GB_Z_PEQ_CONFIG */ + if (p->rdev->family >= CHIP_RV350) + break; + goto fail; + break; + case 0x4be8: + /* valid register only on RV530 */ + if (p->rdev->family == CHIP_RV530) + break; + /* fallthrough do not move */ + default: + goto fail; + } + return 0; +fail: + printk(KERN_ERR "Forbidden register 0x%04X in cs at %d (val=%08x)\n", + reg, idx, idx_value); + return -EINVAL; +} + +static int r300_packet3_check(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt) +{ + struct radeon_cs_reloc *reloc; + struct r100_cs_track *track; + volatile uint32_t *ib; + unsigned idx; + int r; + + ib = p->ib->ptr; + idx = pkt->idx + 1; + track = (struct r100_cs_track *)p->track; + switch(pkt->opcode) { + case PACKET3_3D_LOAD_VBPNTR: + r = r100_packet3_load_vbpntr(p, pkt, idx); + if (r) + return r; + break; + case PACKET3_INDX_BUFFER: + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for packet3 %d\n", pkt->opcode); + r100_cs_dump_packet(p, pkt); + return r; + } + ib[idx+1] = radeon_get_ib_value(p, idx + 1) + ((u32)reloc->lobj.gpu_offset); + r = r100_cs_track_check_pkt3_indx_buffer(p, pkt, reloc->robj); + if (r) { + return r; + } + break; + /* Draw packet */ + case PACKET3_3D_DRAW_IMMD: + /* Number of dwords is vtx_size * (num_vertices - 1) + * PRIM_WALK must be equal to 3 vertex data in embedded + * in cmd stream */ + if (((radeon_get_ib_value(p, idx + 1) >> 4) & 0x3) != 3) { + DRM_ERROR("PRIM_WALK must be 3 for IMMD draw\n"); + return -EINVAL; + } + track->vap_vf_cntl = radeon_get_ib_value(p, idx + 1); + track->immd_dwords = pkt->count - 1; + r = r100_cs_track_check(p->rdev, track); + if (r) { + return r; + } + break; + case PACKET3_3D_DRAW_IMMD_2: + /* Number of dwords is vtx_size * (num_vertices - 1) + * PRIM_WALK must be equal to 3 vertex data in embedded + * in cmd stream */ + if (((radeon_get_ib_value(p, idx) >> 4) & 0x3) != 3) { + DRM_ERROR("PRIM_WALK must be 3 for IMMD draw\n"); + return -EINVAL; + } + track->vap_vf_cntl = radeon_get_ib_value(p, idx); + track->immd_dwords = pkt->count; + r = r100_cs_track_check(p->rdev, track); + if (r) { + return r; + } + break; + case PACKET3_3D_DRAW_VBUF: + track->vap_vf_cntl = radeon_get_ib_value(p, idx + 1); + r = r100_cs_track_check(p->rdev, track); + if (r) { + return r; + } + break; + case PACKET3_3D_DRAW_VBUF_2: + track->vap_vf_cntl = radeon_get_ib_value(p, idx); + r = r100_cs_track_check(p->rdev, track); + if (r) { + return r; + } + break; + case PACKET3_3D_DRAW_INDX: + track->vap_vf_cntl = radeon_get_ib_value(p, idx + 1); + r = r100_cs_track_check(p->rdev, track); + if (r) { + return r; + } + break; + case PACKET3_3D_DRAW_INDX_2: + track->vap_vf_cntl = radeon_get_ib_value(p, idx); + r = r100_cs_track_check(p->rdev, track); + if (r) { + return r; + } + break; + case PACKET3_3D_CLEAR_HIZ: + case PACKET3_3D_CLEAR_ZMASK: + if (p->rdev->hyperz_filp != p->filp) + return -EINVAL; + break; + case PACKET3_3D_CLEAR_CMASK: + if (p->rdev->cmask_filp != p->filp) + return -EINVAL; + break; + case PACKET3_NOP: + break; + default: + DRM_ERROR("Packet3 opcode %x not supported\n", pkt->opcode); + return -EINVAL; + } + return 0; +} + +int r300_cs_parse(struct radeon_cs_parser *p) +{ + struct radeon_cs_packet pkt; + struct r100_cs_track *track; + int r; + + track = kzalloc(sizeof(*track), GFP_KERNEL); + if (track == NULL) + return -ENOMEM; + r100_cs_track_clear(p->rdev, track); + p->track = track; + do { + r = r100_cs_packet_parse(p, &pkt, p->idx); + if (r) { + return r; + } + p->idx += pkt.count + 2; + switch (pkt.type) { + case PACKET_TYPE0: + r = r100_cs_parse_packet0(p, &pkt, + p->rdev->config.r300.reg_safe_bm, + p->rdev->config.r300.reg_safe_bm_size, + &r300_packet0_check); + break; + case PACKET_TYPE2: + break; + case PACKET_TYPE3: + r = r300_packet3_check(p, &pkt); + break; + default: + DRM_ERROR("Unknown packet type %d !\n", pkt.type); + return -EINVAL; + } + if (r) { + return r; + } + } while (p->idx < p->chunks[p->chunk_ib_idx].length_dw); + return 0; +} + +void r300_set_reg_safe(struct radeon_device *rdev) +{ + rdev->config.r300.reg_safe_bm = r300_reg_safe_bm; + rdev->config.r300.reg_safe_bm_size = ARRAY_SIZE(r300_reg_safe_bm); +} + +void r300_mc_program(struct radeon_device *rdev) +{ + struct r100_mc_save save; + int r; + + r = r100_debugfs_mc_info_init(rdev); + if (r) { + dev_err(rdev->dev, "Failed to create r100_mc debugfs file.\n"); + } + + /* Stops all mc clients */ + r100_mc_stop(rdev, &save); + if (rdev->flags & RADEON_IS_AGP) { + WREG32(R_00014C_MC_AGP_LOCATION, + S_00014C_MC_AGP_START(rdev->mc.gtt_start >> 16) | + S_00014C_MC_AGP_TOP(rdev->mc.gtt_end >> 16)); + WREG32(R_000170_AGP_BASE, lower_32_bits(rdev->mc.agp_base)); + WREG32(R_00015C_AGP_BASE_2, + upper_32_bits(rdev->mc.agp_base) & 0xff); + } else { + WREG32(R_00014C_MC_AGP_LOCATION, 0x0FFFFFFF); + WREG32(R_000170_AGP_BASE, 0); + WREG32(R_00015C_AGP_BASE_2, 0); + } + /* Wait for mc idle */ + if (r300_mc_wait_for_idle(rdev)) + DRM_INFO("Failed to wait MC idle before programming MC.\n"); + /* Program MC, should be a 32bits limited address space */ + WREG32(R_000148_MC_FB_LOCATION, + S_000148_MC_FB_START(rdev->mc.vram_start >> 16) | + S_000148_MC_FB_TOP(rdev->mc.vram_end >> 16)); + r100_mc_resume(rdev, &save); +} + +void r300_clock_startup(struct radeon_device *rdev) +{ + u32 tmp; + + if (radeon_dynclks != -1 && radeon_dynclks) + radeon_legacy_set_clock_gating(rdev, 1); + /* We need to force on some of the block */ + tmp = RREG32_PLL(R_00000D_SCLK_CNTL); + tmp |= S_00000D_FORCE_CP(1) | S_00000D_FORCE_VIP(1); + if ((rdev->family == CHIP_RV350) || (rdev->family == CHIP_RV380)) + tmp |= S_00000D_FORCE_VAP(1); + WREG32_PLL(R_00000D_SCLK_CNTL, tmp); +} + +static int r300_startup(struct radeon_device *rdev) +{ + int r; + + /* set common regs */ + r100_set_common_regs(rdev); + /* program mc */ + r300_mc_program(rdev); + /* Resume clock */ + r300_clock_startup(rdev); + /* Initialize GPU configuration (# pipes, ...) */ + r300_gpu_init(rdev); + /* Initialize GART (initialize after TTM so we can allocate + * memory through TTM but finalize after TTM) */ + if (rdev->flags & RADEON_IS_PCIE) { + r = rv370_pcie_gart_enable(rdev); + if (r) + return r; + } + + if (rdev->family == CHIP_R300 || + rdev->family == CHIP_R350 || + rdev->family == CHIP_RV350) + r100_enable_bm(rdev); + + if (rdev->flags & RADEON_IS_PCI) { + r = r100_pci_gart_enable(rdev); + if (r) + return r; + } + + /* allocate wb buffer */ + r = radeon_wb_init(rdev); + if (r) + return r; + + /* Enable IRQ */ + r100_irq_set(rdev); + rdev->config.r300.hdp_cntl = RREG32(RADEON_HOST_PATH_CNTL); + /* 1M ring buffer */ + r = r100_cp_init(rdev, 1024 * 1024); + if (r) { + dev_err(rdev->dev, "failed initializing CP (%d).\n", r); + return r; + } + r = r100_ib_init(rdev); + if (r) { + dev_err(rdev->dev, "failed initializing IB (%d).\n", r); + return r; + } + return 0; +} + +int r300_resume(struct radeon_device *rdev) +{ + /* Make sur GART are not working */ + if (rdev->flags & RADEON_IS_PCIE) + rv370_pcie_gart_disable(rdev); + if (rdev->flags & RADEON_IS_PCI) + r100_pci_gart_disable(rdev); + /* Resume clock before doing reset */ + r300_clock_startup(rdev); + /* Reset gpu before posting otherwise ATOM will enter infinite loop */ + if (radeon_asic_reset(rdev)) { + dev_warn(rdev->dev, "GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n", + RREG32(R_000E40_RBBM_STATUS), + RREG32(R_0007C0_CP_STAT)); + } + /* post */ + radeon_combios_asic_init(rdev->ddev); + /* Resume clock after posting */ + r300_clock_startup(rdev); + /* Initialize surface registers */ + radeon_surface_init(rdev); + return r300_startup(rdev); +} + +int r300_suspend(struct radeon_device *rdev) +{ + r100_cp_disable(rdev); + radeon_wb_disable(rdev); + r100_irq_disable(rdev); + if (rdev->flags & RADEON_IS_PCIE) + rv370_pcie_gart_disable(rdev); + if (rdev->flags & RADEON_IS_PCI) + r100_pci_gart_disable(rdev); + return 0; +} + +void r300_fini(struct radeon_device *rdev) +{ + r100_cp_fini(rdev); + radeon_wb_fini(rdev); + r100_ib_fini(rdev); + radeon_gem_fini(rdev); + if (rdev->flags & RADEON_IS_PCIE) + rv370_pcie_gart_fini(rdev); + if (rdev->flags & RADEON_IS_PCI) + r100_pci_gart_fini(rdev); + radeon_agp_fini(rdev); + radeon_irq_kms_fini(rdev); + radeon_fence_driver_fini(rdev); + radeon_bo_fini(rdev); + radeon_atombios_fini(rdev); + kfree(rdev->bios); + rdev->bios = NULL; +} + +int r300_init(struct radeon_device *rdev) +{ + int r; + + /* Disable VGA */ + r100_vga_render_disable(rdev); + /* Initialize scratch registers */ + radeon_scratch_init(rdev); + /* Initialize surface registers */ + radeon_surface_init(rdev); + /* TODO: disable VGA need to use VGA request */ + /* restore some register to sane defaults */ + r100_restore_sanity(rdev); + /* BIOS*/ + if (!radeon_get_bios(rdev)) { + if (ASIC_IS_AVIVO(rdev)) + return -EINVAL; + } + if (rdev->is_atom_bios) { + dev_err(rdev->dev, "Expecting combios for RS400/RS480 GPU\n"); + return -EINVAL; + } else { + r = radeon_combios_init(rdev); + if (r) + return r; + } + /* Reset gpu before posting otherwise ATOM will enter infinite loop */ + if (radeon_asic_reset(rdev)) { + dev_warn(rdev->dev, + "GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n", + RREG32(R_000E40_RBBM_STATUS), + RREG32(R_0007C0_CP_STAT)); + } + /* check if cards are posted or not */ + if (radeon_boot_test_post_card(rdev) == false) + return -EINVAL; + /* Set asic errata */ + r300_errata(rdev); + /* Initialize clocks */ + radeon_get_clock_info(rdev->ddev); + /* initialize AGP */ + if (rdev->flags & RADEON_IS_AGP) { + r = radeon_agp_init(rdev); + if (r) { + radeon_agp_disable(rdev); + } + } + /* initialize memory controller */ + r300_mc_init(rdev); + /* Fence driver */ + r = radeon_fence_driver_init(rdev); + if (r) + return r; + r = radeon_irq_kms_init(rdev); + if (r) + return r; + /* Memory manager */ + r = radeon_bo_init(rdev); + if (r) + return r; + if (rdev->flags & RADEON_IS_PCIE) { + r = rv370_pcie_gart_init(rdev); + if (r) + return r; + } + if (rdev->flags & RADEON_IS_PCI) { + r = r100_pci_gart_init(rdev); + if (r) + return r; + } + r300_set_reg_safe(rdev); + rdev->accel_working = true; + r = r300_startup(rdev); + if (r) { + /* Somethings want wront with the accel init stop accel */ + dev_err(rdev->dev, "Disabling GPU acceleration\n"); + r100_cp_fini(rdev); + radeon_wb_fini(rdev); + r100_ib_fini(rdev); + radeon_irq_kms_fini(rdev); + if (rdev->flags & RADEON_IS_PCIE) + rv370_pcie_gart_fini(rdev); + if (rdev->flags & RADEON_IS_PCI) + r100_pci_gart_fini(rdev); + radeon_agp_fini(rdev); + rdev->accel_working = false; + } + return 0; +} diff --git a/drivers/gpu/drm/radeon/r300_cmdbuf.c b/drivers/gpu/drm/radeon/r300_cmdbuf.c new file mode 100644 index 00000000..c5c2742e --- /dev/null +++ b/drivers/gpu/drm/radeon/r300_cmdbuf.c @@ -0,0 +1,1185 @@ +/* r300_cmdbuf.c -- Command buffer emission for R300 -*- linux-c -*- + * + * Copyright (C) The Weather Channel, Inc. 2002. + * Copyright (C) 2004 Nicolai Haehnle. + * All Rights Reserved. + * + * The Weather Channel (TM) funded Tungsten Graphics to develop the + * initial release of the Radeon 8500 driver under the XFree86 license. + * This notice must be preserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Nicolai Haehnle + */ + +#include "drmP.h" +#include "drm.h" +#include "drm_buffer.h" +#include "radeon_drm.h" +#include "radeon_drv.h" +#include "r300_reg.h" + +#include + +#define R300_SIMULTANEOUS_CLIPRECTS 4 + +/* Values for R300_RE_CLIPRECT_CNTL depending on the number of cliprects + */ +static const int r300_cliprect_cntl[4] = { + 0xAAAA, + 0xEEEE, + 0xFEFE, + 0xFFFE +}; + +/** + * Emit up to R300_SIMULTANEOUS_CLIPRECTS cliprects from the given command + * buffer, starting with index n. + */ +static int r300_emit_cliprects(drm_radeon_private_t *dev_priv, + drm_radeon_kcmd_buffer_t *cmdbuf, int n) +{ + struct drm_clip_rect box; + int nr; + int i; + RING_LOCALS; + + nr = cmdbuf->nbox - n; + if (nr > R300_SIMULTANEOUS_CLIPRECTS) + nr = R300_SIMULTANEOUS_CLIPRECTS; + + DRM_DEBUG("%i cliprects\n", nr); + + if (nr) { + BEGIN_RING(6 + nr * 2); + OUT_RING(CP_PACKET0(R300_RE_CLIPRECT_TL_0, nr * 2 - 1)); + + for (i = 0; i < nr; ++i) { + if (DRM_COPY_FROM_USER_UNCHECKED + (&box, &cmdbuf->boxes[n + i], sizeof(box))) { + DRM_ERROR("copy cliprect faulted\n"); + return -EFAULT; + } + + box.x2--; /* Hardware expects inclusive bottom-right corner */ + box.y2--; + + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_RV515) { + box.x1 = (box.x1) & + R300_CLIPRECT_MASK; + box.y1 = (box.y1) & + R300_CLIPRECT_MASK; + box.x2 = (box.x2) & + R300_CLIPRECT_MASK; + box.y2 = (box.y2) & + R300_CLIPRECT_MASK; + } else { + box.x1 = (box.x1 + R300_CLIPRECT_OFFSET) & + R300_CLIPRECT_MASK; + box.y1 = (box.y1 + R300_CLIPRECT_OFFSET) & + R300_CLIPRECT_MASK; + box.x2 = (box.x2 + R300_CLIPRECT_OFFSET) & + R300_CLIPRECT_MASK; + box.y2 = (box.y2 + R300_CLIPRECT_OFFSET) & + R300_CLIPRECT_MASK; + } + + OUT_RING((box.x1 << R300_CLIPRECT_X_SHIFT) | + (box.y1 << R300_CLIPRECT_Y_SHIFT)); + OUT_RING((box.x2 << R300_CLIPRECT_X_SHIFT) | + (box.y2 << R300_CLIPRECT_Y_SHIFT)); + + } + + OUT_RING_REG(R300_RE_CLIPRECT_CNTL, r300_cliprect_cntl[nr - 1]); + + /* TODO/SECURITY: Force scissors to a safe value, otherwise the + * client might be able to trample over memory. + * The impact should be very limited, but I'd rather be safe than + * sorry. + */ + OUT_RING(CP_PACKET0(R300_RE_SCISSORS_TL, 1)); + OUT_RING(0); + OUT_RING(R300_SCISSORS_X_MASK | R300_SCISSORS_Y_MASK); + ADVANCE_RING(); + } else { + /* Why we allow zero cliprect rendering: + * There are some commands in a command buffer that must be submitted + * even when there are no cliprects, e.g. DMA buffer discard + * or state setting (though state setting could be avoided by + * simulating a loss of context). + * + * Now since the cmdbuf interface is so chaotic right now (and is + * bound to remain that way for a bit until things settle down), + * it is basically impossible to filter out the commands that are + * necessary and those that aren't. + * + * So I choose the safe way and don't do any filtering at all; + * instead, I simply set up the engine so that all rendering + * can't produce any fragments. + */ + BEGIN_RING(2); + OUT_RING_REG(R300_RE_CLIPRECT_CNTL, 0); + ADVANCE_RING(); + } + + /* flus cache and wait idle clean after cliprect change */ + BEGIN_RING(2); + OUT_RING(CP_PACKET0(R300_RB3D_DSTCACHE_CTLSTAT, 0)); + OUT_RING(R300_RB3D_DC_FLUSH); + ADVANCE_RING(); + BEGIN_RING(2); + OUT_RING(CP_PACKET0(RADEON_WAIT_UNTIL, 0)); + OUT_RING(RADEON_WAIT_3D_IDLECLEAN); + ADVANCE_RING(); + /* set flush flag */ + dev_priv->track_flush |= RADEON_FLUSH_EMITED; + + return 0; +} + +static u8 r300_reg_flags[0x10000 >> 2]; + +void r300_init_reg_flags(struct drm_device *dev) +{ + int i; + drm_radeon_private_t *dev_priv = dev->dev_private; + + memset(r300_reg_flags, 0, 0x10000 >> 2); +#define ADD_RANGE_MARK(reg, count,mark) \ + for(i=((reg)>>2);i<((reg)>>2)+(count);i++)\ + r300_reg_flags[i]|=(mark); + +#define MARK_SAFE 1 +#define MARK_CHECK_OFFSET 2 + +#define ADD_RANGE(reg, count) ADD_RANGE_MARK(reg, count, MARK_SAFE) + + /* these match cmducs() command in r300_driver/r300/r300_cmdbuf.c */ + ADD_RANGE(R300_SE_VPORT_XSCALE, 6); + ADD_RANGE(R300_VAP_CNTL, 1); + ADD_RANGE(R300_SE_VTE_CNTL, 2); + ADD_RANGE(0x2134, 2); + ADD_RANGE(R300_VAP_CNTL_STATUS, 1); + ADD_RANGE(R300_VAP_INPUT_CNTL_0, 2); + ADD_RANGE(0x21DC, 1); + ADD_RANGE(R300_VAP_UNKNOWN_221C, 1); + ADD_RANGE(R300_VAP_CLIP_X_0, 4); + ADD_RANGE(R300_VAP_PVS_STATE_FLUSH_REG, 1); + ADD_RANGE(R300_VAP_UNKNOWN_2288, 1); + ADD_RANGE(R300_VAP_OUTPUT_VTX_FMT_0, 2); + ADD_RANGE(R300_VAP_PVS_CNTL_1, 3); + ADD_RANGE(R300_GB_ENABLE, 1); + ADD_RANGE(R300_GB_MSPOS0, 5); + ADD_RANGE(R300_TX_INVALTAGS, 1); + ADD_RANGE(R300_TX_ENABLE, 1); + ADD_RANGE(0x4200, 4); + ADD_RANGE(0x4214, 1); + ADD_RANGE(R300_RE_POINTSIZE, 1); + ADD_RANGE(0x4230, 3); + ADD_RANGE(R300_RE_LINE_CNT, 1); + ADD_RANGE(R300_RE_UNK4238, 1); + ADD_RANGE(0x4260, 3); + ADD_RANGE(R300_RE_SHADE, 4); + ADD_RANGE(R300_RE_POLYGON_MODE, 5); + ADD_RANGE(R300_RE_ZBIAS_CNTL, 1); + ADD_RANGE(R300_RE_ZBIAS_T_FACTOR, 4); + ADD_RANGE(R300_RE_OCCLUSION_CNTL, 1); + ADD_RANGE(R300_RE_CULL_CNTL, 1); + ADD_RANGE(0x42C0, 2); + ADD_RANGE(R300_RS_CNTL_0, 2); + + ADD_RANGE(R300_SU_REG_DEST, 1); + if ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV530) + ADD_RANGE(RV530_FG_ZBREG_DEST, 1); + + ADD_RANGE(R300_SC_HYPERZ, 2); + ADD_RANGE(0x43E8, 1); + + ADD_RANGE(0x46A4, 5); + + ADD_RANGE(R300_RE_FOG_STATE, 1); + ADD_RANGE(R300_FOG_COLOR_R, 3); + ADD_RANGE(R300_PP_ALPHA_TEST, 2); + ADD_RANGE(0x4BD8, 1); + ADD_RANGE(R300_PFS_PARAM_0_X, 64); + ADD_RANGE(0x4E00, 1); + ADD_RANGE(R300_RB3D_CBLEND, 2); + ADD_RANGE(R300_RB3D_COLORMASK, 1); + ADD_RANGE(R300_RB3D_BLEND_COLOR, 3); + ADD_RANGE_MARK(R300_RB3D_COLOROFFSET0, 1, MARK_CHECK_OFFSET); /* check offset */ + ADD_RANGE(R300_RB3D_COLORPITCH0, 1); + ADD_RANGE(0x4E50, 9); + ADD_RANGE(0x4E88, 1); + ADD_RANGE(0x4EA0, 2); + ADD_RANGE(R300_ZB_CNTL, 3); + ADD_RANGE(R300_ZB_FORMAT, 4); + ADD_RANGE_MARK(R300_ZB_DEPTHOFFSET, 1, MARK_CHECK_OFFSET); /* check offset */ + ADD_RANGE(R300_ZB_DEPTHPITCH, 1); + ADD_RANGE(R300_ZB_DEPTHCLEARVALUE, 1); + ADD_RANGE(R300_ZB_ZMASK_OFFSET, 13); + ADD_RANGE(R300_ZB_ZPASS_DATA, 2); /* ZB_ZPASS_DATA, ZB_ZPASS_ADDR */ + + ADD_RANGE(R300_TX_FILTER_0, 16); + ADD_RANGE(R300_TX_FILTER1_0, 16); + ADD_RANGE(R300_TX_SIZE_0, 16); + ADD_RANGE(R300_TX_FORMAT_0, 16); + ADD_RANGE(R300_TX_PITCH_0, 16); + /* Texture offset is dangerous and needs more checking */ + ADD_RANGE_MARK(R300_TX_OFFSET_0, 16, MARK_CHECK_OFFSET); + ADD_RANGE(R300_TX_CHROMA_KEY_0, 16); + ADD_RANGE(R300_TX_BORDER_COLOR_0, 16); + + /* Sporadic registers used as primitives are emitted */ + ADD_RANGE(R300_ZB_ZCACHE_CTLSTAT, 1); + ADD_RANGE(R300_RB3D_DSTCACHE_CTLSTAT, 1); + ADD_RANGE(R300_VAP_INPUT_ROUTE_0_0, 8); + ADD_RANGE(R300_VAP_INPUT_ROUTE_1_0, 8); + + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_RV515) { + ADD_RANGE(R500_VAP_INDEX_OFFSET, 1); + ADD_RANGE(R500_US_CONFIG, 2); + ADD_RANGE(R500_US_CODE_ADDR, 3); + ADD_RANGE(R500_US_FC_CTRL, 1); + ADD_RANGE(R500_RS_IP_0, 16); + ADD_RANGE(R500_RS_INST_0, 16); + ADD_RANGE(R500_RB3D_COLOR_CLEAR_VALUE_AR, 2); + ADD_RANGE(R500_RB3D_CONSTANT_COLOR_AR, 2); + ADD_RANGE(R500_ZB_FIFO_SIZE, 2); + } else { + ADD_RANGE(R300_PFS_CNTL_0, 3); + ADD_RANGE(R300_PFS_NODE_0, 4); + ADD_RANGE(R300_PFS_TEXI_0, 64); + ADD_RANGE(R300_PFS_INSTR0_0, 64); + ADD_RANGE(R300_PFS_INSTR1_0, 64); + ADD_RANGE(R300_PFS_INSTR2_0, 64); + ADD_RANGE(R300_PFS_INSTR3_0, 64); + ADD_RANGE(R300_RS_INTERP_0, 8); + ADD_RANGE(R300_RS_ROUTE_0, 8); + + } +} + +static __inline__ int r300_check_range(unsigned reg, int count) +{ + int i; + if (reg & ~0xffff) + return -1; + for (i = (reg >> 2); i < (reg >> 2) + count; i++) + if (r300_reg_flags[i] != MARK_SAFE) + return 1; + return 0; +} + +static __inline__ int r300_emit_carefully_checked_packet0(drm_radeon_private_t * + dev_priv, + drm_radeon_kcmd_buffer_t + * cmdbuf, + drm_r300_cmd_header_t + header) +{ + int reg; + int sz; + int i; + u32 *value; + RING_LOCALS; + + sz = header.packet0.count; + reg = (header.packet0.reghi << 8) | header.packet0.reglo; + + if ((sz > 64) || (sz < 0)) { + DRM_ERROR("Cannot emit more than 64 values at a time (reg=%04x sz=%d)\n", + reg, sz); + return -EINVAL; + } + + for (i = 0; i < sz; i++) { + switch (r300_reg_flags[(reg >> 2) + i]) { + case MARK_SAFE: + break; + case MARK_CHECK_OFFSET: + value = drm_buffer_pointer_to_dword(cmdbuf->buffer, i); + if (!radeon_check_offset(dev_priv, *value)) { + DRM_ERROR("Offset failed range check (reg=%04x sz=%d)\n", + reg, sz); + return -EINVAL; + } + break; + default: + DRM_ERROR("Register %04x failed check as flag=%02x\n", + reg + i * 4, r300_reg_flags[(reg >> 2) + i]); + return -EINVAL; + } + } + + BEGIN_RING(1 + sz); + OUT_RING(CP_PACKET0(reg, sz - 1)); + OUT_RING_DRM_BUFFER(cmdbuf->buffer, sz); + ADVANCE_RING(); + + return 0; +} + +/** + * Emits a packet0 setting arbitrary registers. + * Called by r300_do_cp_cmdbuf. + * + * Note that checks are performed on contents and addresses of the registers + */ +static __inline__ int r300_emit_packet0(drm_radeon_private_t *dev_priv, + drm_radeon_kcmd_buffer_t *cmdbuf, + drm_r300_cmd_header_t header) +{ + int reg; + int sz; + RING_LOCALS; + + sz = header.packet0.count; + reg = (header.packet0.reghi << 8) | header.packet0.reglo; + + if (!sz) + return 0; + + if (sz * 4 > drm_buffer_unprocessed(cmdbuf->buffer)) + return -EINVAL; + + if (reg + sz * 4 >= 0x10000) { + DRM_ERROR("No such registers in hardware reg=%04x sz=%d\n", reg, + sz); + return -EINVAL; + } + + if (r300_check_range(reg, sz)) { + /* go and check everything */ + return r300_emit_carefully_checked_packet0(dev_priv, cmdbuf, + header); + } + /* the rest of the data is safe to emit, whatever the values the user passed */ + + BEGIN_RING(1 + sz); + OUT_RING(CP_PACKET0(reg, sz - 1)); + OUT_RING_DRM_BUFFER(cmdbuf->buffer, sz); + ADVANCE_RING(); + + return 0; +} + +/** + * Uploads user-supplied vertex program instructions or parameters onto + * the graphics card. + * Called by r300_do_cp_cmdbuf. + */ +static __inline__ int r300_emit_vpu(drm_radeon_private_t *dev_priv, + drm_radeon_kcmd_buffer_t *cmdbuf, + drm_r300_cmd_header_t header) +{ + int sz; + int addr; + RING_LOCALS; + + sz = header.vpu.count; + addr = (header.vpu.adrhi << 8) | header.vpu.adrlo; + + if (!sz) + return 0; + if (sz * 16 > drm_buffer_unprocessed(cmdbuf->buffer)) + return -EINVAL; + + /* VAP is very sensitive so we purge cache before we program it + * and we also flush its state before & after */ + BEGIN_RING(6); + OUT_RING(CP_PACKET0(R300_RB3D_DSTCACHE_CTLSTAT, 0)); + OUT_RING(R300_RB3D_DC_FLUSH); + OUT_RING(CP_PACKET0(RADEON_WAIT_UNTIL, 0)); + OUT_RING(RADEON_WAIT_3D_IDLECLEAN); + OUT_RING(CP_PACKET0(R300_VAP_PVS_STATE_FLUSH_REG, 0)); + OUT_RING(0); + ADVANCE_RING(); + /* set flush flag */ + dev_priv->track_flush |= RADEON_FLUSH_EMITED; + + BEGIN_RING(3 + sz * 4); + OUT_RING_REG(R300_VAP_PVS_UPLOAD_ADDRESS, addr); + OUT_RING(CP_PACKET0_TABLE(R300_VAP_PVS_UPLOAD_DATA, sz * 4 - 1)); + OUT_RING_DRM_BUFFER(cmdbuf->buffer, sz * 4); + ADVANCE_RING(); + + BEGIN_RING(2); + OUT_RING(CP_PACKET0(R300_VAP_PVS_STATE_FLUSH_REG, 0)); + OUT_RING(0); + ADVANCE_RING(); + + return 0; +} + +/** + * Emit a clear packet from userspace. + * Called by r300_emit_packet3. + */ +static __inline__ int r300_emit_clear(drm_radeon_private_t *dev_priv, + drm_radeon_kcmd_buffer_t *cmdbuf) +{ + RING_LOCALS; + + if (8 * 4 > drm_buffer_unprocessed(cmdbuf->buffer)) + return -EINVAL; + + BEGIN_RING(10); + OUT_RING(CP_PACKET3(R200_3D_DRAW_IMMD_2, 8)); + OUT_RING(R300_PRIM_TYPE_POINT | R300_PRIM_WALK_RING | + (1 << R300_PRIM_NUM_VERTICES_SHIFT)); + OUT_RING_DRM_BUFFER(cmdbuf->buffer, 8); + ADVANCE_RING(); + + BEGIN_RING(4); + OUT_RING(CP_PACKET0(R300_RB3D_DSTCACHE_CTLSTAT, 0)); + OUT_RING(R300_RB3D_DC_FLUSH); + OUT_RING(CP_PACKET0(RADEON_WAIT_UNTIL, 0)); + OUT_RING(RADEON_WAIT_3D_IDLECLEAN); + ADVANCE_RING(); + /* set flush flag */ + dev_priv->track_flush |= RADEON_FLUSH_EMITED; + + return 0; +} + +static __inline__ int r300_emit_3d_load_vbpntr(drm_radeon_private_t *dev_priv, + drm_radeon_kcmd_buffer_t *cmdbuf, + u32 header) +{ + int count, i, k; +#define MAX_ARRAY_PACKET 64 + u32 *data; + u32 narrays; + RING_LOCALS; + + count = (header & RADEON_CP_PACKET_COUNT_MASK) >> 16; + + if ((count + 1) > MAX_ARRAY_PACKET) { + DRM_ERROR("Too large payload in 3D_LOAD_VBPNTR (count=%d)\n", + count); + return -EINVAL; + } + /* carefully check packet contents */ + + /* We have already read the header so advance the buffer. */ + drm_buffer_advance(cmdbuf->buffer, 4); + + narrays = *(u32 *)drm_buffer_pointer_to_dword(cmdbuf->buffer, 0); + k = 0; + i = 1; + while ((k < narrays) && (i < (count + 1))) { + i++; /* skip attribute field */ + data = drm_buffer_pointer_to_dword(cmdbuf->buffer, i); + if (!radeon_check_offset(dev_priv, *data)) { + DRM_ERROR + ("Offset failed range check (k=%d i=%d) while processing 3D_LOAD_VBPNTR packet.\n", + k, i); + return -EINVAL; + } + k++; + i++; + if (k == narrays) + break; + /* have one more to process, they come in pairs */ + data = drm_buffer_pointer_to_dword(cmdbuf->buffer, i); + if (!radeon_check_offset(dev_priv, *data)) { + DRM_ERROR + ("Offset failed range check (k=%d i=%d) while processing 3D_LOAD_VBPNTR packet.\n", + k, i); + return -EINVAL; + } + k++; + i++; + } + /* do the counts match what we expect ? */ + if ((k != narrays) || (i != (count + 1))) { + DRM_ERROR + ("Malformed 3D_LOAD_VBPNTR packet (k=%d i=%d narrays=%d count+1=%d).\n", + k, i, narrays, count + 1); + return -EINVAL; + } + + /* all clear, output packet */ + + BEGIN_RING(count + 2); + OUT_RING(header); + OUT_RING_DRM_BUFFER(cmdbuf->buffer, count + 1); + ADVANCE_RING(); + + return 0; +} + +static __inline__ int r300_emit_bitblt_multi(drm_radeon_private_t *dev_priv, + drm_radeon_kcmd_buffer_t *cmdbuf) +{ + u32 *cmd = drm_buffer_pointer_to_dword(cmdbuf->buffer, 0); + int count, ret; + RING_LOCALS; + + + count = (*cmd & RADEON_CP_PACKET_COUNT_MASK) >> 16; + + if (*cmd & 0x8000) { + u32 offset; + u32 *cmd1 = drm_buffer_pointer_to_dword(cmdbuf->buffer, 1); + if (*cmd1 & (RADEON_GMC_SRC_PITCH_OFFSET_CNTL + | RADEON_GMC_DST_PITCH_OFFSET_CNTL)) { + + u32 *cmd2 = drm_buffer_pointer_to_dword(cmdbuf->buffer, 2); + offset = *cmd2 << 10; + ret = !radeon_check_offset(dev_priv, offset); + if (ret) { + DRM_ERROR("Invalid bitblt first offset is %08X\n", offset); + return -EINVAL; + } + } + + if ((*cmd1 & RADEON_GMC_SRC_PITCH_OFFSET_CNTL) && + (*cmd1 & RADEON_GMC_DST_PITCH_OFFSET_CNTL)) { + u32 *cmd3 = drm_buffer_pointer_to_dword(cmdbuf->buffer, 3); + offset = *cmd3 << 10; + ret = !radeon_check_offset(dev_priv, offset); + if (ret) { + DRM_ERROR("Invalid bitblt second offset is %08X\n", offset); + return -EINVAL; + } + + } + } + + BEGIN_RING(count+2); + OUT_RING_DRM_BUFFER(cmdbuf->buffer, count + 2); + ADVANCE_RING(); + + return 0; +} + +static __inline__ int r300_emit_draw_indx_2(drm_radeon_private_t *dev_priv, + drm_radeon_kcmd_buffer_t *cmdbuf) +{ + u32 *cmd = drm_buffer_pointer_to_dword(cmdbuf->buffer, 0); + u32 *cmd1 = drm_buffer_pointer_to_dword(cmdbuf->buffer, 1); + int count; + int expected_count; + RING_LOCALS; + + count = (*cmd & RADEON_CP_PACKET_COUNT_MASK) >> 16; + + expected_count = *cmd1 >> 16; + if (!(*cmd1 & R300_VAP_VF_CNTL__INDEX_SIZE_32bit)) + expected_count = (expected_count+1)/2; + + if (count && count != expected_count) { + DRM_ERROR("3D_DRAW_INDX_2: packet size %i, expected %i\n", + count, expected_count); + return -EINVAL; + } + + BEGIN_RING(count+2); + OUT_RING_DRM_BUFFER(cmdbuf->buffer, count + 2); + ADVANCE_RING(); + + if (!count) { + drm_r300_cmd_header_t stack_header, *header; + u32 *cmd1, *cmd2, *cmd3; + + if (drm_buffer_unprocessed(cmdbuf->buffer) + < 4*4 + sizeof(stack_header)) { + DRM_ERROR("3D_DRAW_INDX_2: expect subsequent INDX_BUFFER, but stream is too short.\n"); + return -EINVAL; + } + + header = drm_buffer_read_object(cmdbuf->buffer, + sizeof(stack_header), &stack_header); + + cmd = drm_buffer_pointer_to_dword(cmdbuf->buffer, 0); + cmd1 = drm_buffer_pointer_to_dword(cmdbuf->buffer, 1); + cmd2 = drm_buffer_pointer_to_dword(cmdbuf->buffer, 2); + cmd3 = drm_buffer_pointer_to_dword(cmdbuf->buffer, 3); + + if (header->header.cmd_type != R300_CMD_PACKET3 || + header->packet3.packet != R300_CMD_PACKET3_RAW || + *cmd != CP_PACKET3(RADEON_CP_INDX_BUFFER, 2)) { + DRM_ERROR("3D_DRAW_INDX_2: expect subsequent INDX_BUFFER.\n"); + return -EINVAL; + } + + if ((*cmd1 & 0x8000ffff) != 0x80000810) { + DRM_ERROR("Invalid indx_buffer reg address %08X\n", + *cmd1); + return -EINVAL; + } + if (!radeon_check_offset(dev_priv, *cmd2)) { + DRM_ERROR("Invalid indx_buffer offset is %08X\n", + *cmd2); + return -EINVAL; + } + if (*cmd3 != expected_count) { + DRM_ERROR("INDX_BUFFER: buffer size %i, expected %i\n", + *cmd3, expected_count); + return -EINVAL; + } + + BEGIN_RING(4); + OUT_RING_DRM_BUFFER(cmdbuf->buffer, 4); + ADVANCE_RING(); + } + + return 0; +} + +static __inline__ int r300_emit_raw_packet3(drm_radeon_private_t *dev_priv, + drm_radeon_kcmd_buffer_t *cmdbuf) +{ + u32 *header; + int count; + RING_LOCALS; + + if (4 > drm_buffer_unprocessed(cmdbuf->buffer)) + return -EINVAL; + + /* Fixme !! This simply emits a packet without much checking. + We need to be smarter. */ + + /* obtain first word - actual packet3 header */ + header = drm_buffer_pointer_to_dword(cmdbuf->buffer, 0); + + /* Is it packet 3 ? */ + if ((*header >> 30) != 0x3) { + DRM_ERROR("Not a packet3 header (0x%08x)\n", *header); + return -EINVAL; + } + + count = (*header >> 16) & 0x3fff; + + /* Check again now that we know how much data to expect */ + if ((count + 2) * 4 > drm_buffer_unprocessed(cmdbuf->buffer)) { + DRM_ERROR + ("Expected packet3 of length %d but have only %d bytes left\n", + (count + 2) * 4, drm_buffer_unprocessed(cmdbuf->buffer)); + return -EINVAL; + } + + /* Is it a packet type we know about ? */ + switch (*header & 0xff00) { + case RADEON_3D_LOAD_VBPNTR: /* load vertex array pointers */ + return r300_emit_3d_load_vbpntr(dev_priv, cmdbuf, *header); + + case RADEON_CNTL_BITBLT_MULTI: + return r300_emit_bitblt_multi(dev_priv, cmdbuf); + + case RADEON_CP_INDX_BUFFER: + DRM_ERROR("packet3 INDX_BUFFER without preceding 3D_DRAW_INDX_2 is illegal.\n"); + return -EINVAL; + case RADEON_CP_3D_DRAW_IMMD_2: + /* triggers drawing using in-packet vertex data */ + case RADEON_CP_3D_DRAW_VBUF_2: + /* triggers drawing of vertex buffers setup elsewhere */ + dev_priv->track_flush &= ~(RADEON_FLUSH_EMITED | + RADEON_PURGE_EMITED); + break; + case RADEON_CP_3D_DRAW_INDX_2: + /* triggers drawing using indices to vertex buffer */ + /* whenever we send vertex we clear flush & purge */ + dev_priv->track_flush &= ~(RADEON_FLUSH_EMITED | + RADEON_PURGE_EMITED); + return r300_emit_draw_indx_2(dev_priv, cmdbuf); + case RADEON_WAIT_FOR_IDLE: + case RADEON_CP_NOP: + /* these packets are safe */ + break; + default: + DRM_ERROR("Unknown packet3 header (0x%08x)\n", *header); + return -EINVAL; + } + + BEGIN_RING(count + 2); + OUT_RING_DRM_BUFFER(cmdbuf->buffer, count + 2); + ADVANCE_RING(); + + return 0; +} + +/** + * Emit a rendering packet3 from userspace. + * Called by r300_do_cp_cmdbuf. + */ +static __inline__ int r300_emit_packet3(drm_radeon_private_t *dev_priv, + drm_radeon_kcmd_buffer_t *cmdbuf, + drm_r300_cmd_header_t header) +{ + int n; + int ret; + int orig_iter = cmdbuf->buffer->iterator; + + /* This is a do-while-loop so that we run the interior at least once, + * even if cmdbuf->nbox is 0. Compare r300_emit_cliprects for rationale. + */ + n = 0; + do { + if (cmdbuf->nbox > R300_SIMULTANEOUS_CLIPRECTS) { + ret = r300_emit_cliprects(dev_priv, cmdbuf, n); + if (ret) + return ret; + + cmdbuf->buffer->iterator = orig_iter; + } + + switch (header.packet3.packet) { + case R300_CMD_PACKET3_CLEAR: + DRM_DEBUG("R300_CMD_PACKET3_CLEAR\n"); + ret = r300_emit_clear(dev_priv, cmdbuf); + if (ret) { + DRM_ERROR("r300_emit_clear failed\n"); + return ret; + } + break; + + case R300_CMD_PACKET3_RAW: + DRM_DEBUG("R300_CMD_PACKET3_RAW\n"); + ret = r300_emit_raw_packet3(dev_priv, cmdbuf); + if (ret) { + DRM_ERROR("r300_emit_raw_packet3 failed\n"); + return ret; + } + break; + + default: + DRM_ERROR("bad packet3 type %i at byte %d\n", + header.packet3.packet, + cmdbuf->buffer->iterator - (int)sizeof(header)); + return -EINVAL; + } + + n += R300_SIMULTANEOUS_CLIPRECTS; + } while (n < cmdbuf->nbox); + + return 0; +} + +/* Some of the R300 chips seem to be extremely touchy about the two registers + * that are configured in r300_pacify. + * Among the worst offenders seems to be the R300 ND (0x4E44): When userspace + * sends a command buffer that contains only state setting commands and a + * vertex program/parameter upload sequence, this will eventually lead to a + * lockup, unless the sequence is bracketed by calls to r300_pacify. + * So we should take great care to *always* call r300_pacify before + * *anything* 3D related, and again afterwards. This is what the + * call bracket in r300_do_cp_cmdbuf is for. + */ + +/** + * Emit the sequence to pacify R300. + */ +static __inline__ void r300_pacify(drm_radeon_private_t *dev_priv) +{ + uint32_t cache_z, cache_3d, cache_2d; + RING_LOCALS; + + cache_z = R300_ZC_FLUSH; + cache_2d = R300_RB2D_DC_FLUSH; + cache_3d = R300_RB3D_DC_FLUSH; + if (!(dev_priv->track_flush & RADEON_PURGE_EMITED)) { + /* we can purge, primitive where draw since last purge */ + cache_z |= R300_ZC_FREE; + cache_2d |= R300_RB2D_DC_FREE; + cache_3d |= R300_RB3D_DC_FREE; + } + + /* flush & purge zbuffer */ + BEGIN_RING(2); + OUT_RING(CP_PACKET0(R300_ZB_ZCACHE_CTLSTAT, 0)); + OUT_RING(cache_z); + ADVANCE_RING(); + /* flush & purge 3d */ + BEGIN_RING(2); + OUT_RING(CP_PACKET0(R300_RB3D_DSTCACHE_CTLSTAT, 0)); + OUT_RING(cache_3d); + ADVANCE_RING(); + /* flush & purge texture */ + BEGIN_RING(2); + OUT_RING(CP_PACKET0(R300_TX_INVALTAGS, 0)); + OUT_RING(0); + ADVANCE_RING(); + /* FIXME: is this one really needed ? */ + BEGIN_RING(2); + OUT_RING(CP_PACKET0(R300_RB3D_AARESOLVE_CTL, 0)); + OUT_RING(0); + ADVANCE_RING(); + BEGIN_RING(2); + OUT_RING(CP_PACKET0(RADEON_WAIT_UNTIL, 0)); + OUT_RING(RADEON_WAIT_3D_IDLECLEAN); + ADVANCE_RING(); + /* flush & purge 2d through E2 as RB2D will trigger lockup */ + BEGIN_RING(4); + OUT_RING(CP_PACKET0(R300_DSTCACHE_CTLSTAT, 0)); + OUT_RING(cache_2d); + OUT_RING(CP_PACKET0(RADEON_WAIT_UNTIL, 0)); + OUT_RING(RADEON_WAIT_2D_IDLECLEAN | + RADEON_WAIT_HOST_IDLECLEAN); + ADVANCE_RING(); + /* set flush & purge flags */ + dev_priv->track_flush |= RADEON_FLUSH_EMITED | RADEON_PURGE_EMITED; +} + +/** + * Called by r300_do_cp_cmdbuf to update the internal buffer age and state. + * The actual age emit is done by r300_do_cp_cmdbuf, which is why you must + * be careful about how this function is called. + */ +static void r300_discard_buffer(struct drm_device *dev, struct drm_master *master, struct drm_buf *buf) +{ + drm_radeon_buf_priv_t *buf_priv = buf->dev_private; + struct drm_radeon_master_private *master_priv = master->driver_priv; + + buf_priv->age = ++master_priv->sarea_priv->last_dispatch; + buf->pending = 1; + buf->used = 0; +} + +static void r300_cmd_wait(drm_radeon_private_t * dev_priv, + drm_r300_cmd_header_t header) +{ + u32 wait_until; + RING_LOCALS; + + if (!header.wait.flags) + return; + + wait_until = 0; + + switch(header.wait.flags) { + case R300_WAIT_2D: + wait_until = RADEON_WAIT_2D_IDLE; + break; + case R300_WAIT_3D: + wait_until = RADEON_WAIT_3D_IDLE; + break; + case R300_NEW_WAIT_2D_3D: + wait_until = RADEON_WAIT_2D_IDLE|RADEON_WAIT_3D_IDLE; + break; + case R300_NEW_WAIT_2D_2D_CLEAN: + wait_until = RADEON_WAIT_2D_IDLE|RADEON_WAIT_2D_IDLECLEAN; + break; + case R300_NEW_WAIT_3D_3D_CLEAN: + wait_until = RADEON_WAIT_3D_IDLE|RADEON_WAIT_3D_IDLECLEAN; + break; + case R300_NEW_WAIT_2D_2D_CLEAN_3D_3D_CLEAN: + wait_until = RADEON_WAIT_2D_IDLE|RADEON_WAIT_2D_IDLECLEAN; + wait_until |= RADEON_WAIT_3D_IDLE|RADEON_WAIT_3D_IDLECLEAN; + break; + default: + return; + } + + BEGIN_RING(2); + OUT_RING(CP_PACKET0(RADEON_WAIT_UNTIL, 0)); + OUT_RING(wait_until); + ADVANCE_RING(); +} + +static int r300_scratch(drm_radeon_private_t *dev_priv, + drm_radeon_kcmd_buffer_t *cmdbuf, + drm_r300_cmd_header_t header) +{ + u32 *ref_age_base; + u32 i, *buf_idx, h_pending; + u64 *ptr_addr; + u64 stack_ptr_addr; + RING_LOCALS; + + if (drm_buffer_unprocessed(cmdbuf->buffer) < + (sizeof(u64) + header.scratch.n_bufs * sizeof(*buf_idx))) { + return -EINVAL; + } + + if (header.scratch.reg >= 5) { + return -EINVAL; + } + + dev_priv->scratch_ages[header.scratch.reg]++; + + ptr_addr = drm_buffer_read_object(cmdbuf->buffer, + sizeof(stack_ptr_addr), &stack_ptr_addr); + ref_age_base = (u32 *)(unsigned long)get_unaligned(ptr_addr); + + for (i=0; i < header.scratch.n_bufs; i++) { + buf_idx = drm_buffer_pointer_to_dword(cmdbuf->buffer, 0); + *buf_idx *= 2; /* 8 bytes per buf */ + + if (DRM_COPY_TO_USER(ref_age_base + *buf_idx, + &dev_priv->scratch_ages[header.scratch.reg], + sizeof(u32))) + return -EINVAL; + + if (DRM_COPY_FROM_USER(&h_pending, + ref_age_base + *buf_idx + 1, + sizeof(u32))) + return -EINVAL; + + if (h_pending == 0) + return -EINVAL; + + h_pending--; + + if (DRM_COPY_TO_USER(ref_age_base + *buf_idx + 1, + &h_pending, + sizeof(u32))) + return -EINVAL; + + drm_buffer_advance(cmdbuf->buffer, sizeof(*buf_idx)); + } + + BEGIN_RING(2); + OUT_RING( CP_PACKET0( RADEON_SCRATCH_REG0 + header.scratch.reg * 4, 0 ) ); + OUT_RING( dev_priv->scratch_ages[header.scratch.reg] ); + ADVANCE_RING(); + + return 0; +} + +/** + * Uploads user-supplied vertex program instructions or parameters onto + * the graphics card. + * Called by r300_do_cp_cmdbuf. + */ +static inline int r300_emit_r500fp(drm_radeon_private_t *dev_priv, + drm_radeon_kcmd_buffer_t *cmdbuf, + drm_r300_cmd_header_t header) +{ + int sz; + int addr; + int type; + int isclamp; + int stride; + RING_LOCALS; + + sz = header.r500fp.count; + /* address is 9 bits 0 - 8, bit 1 of flags is part of address */ + addr = ((header.r500fp.adrhi_flags & 1) << 8) | header.r500fp.adrlo; + + type = !!(header.r500fp.adrhi_flags & R500FP_CONSTANT_TYPE); + isclamp = !!(header.r500fp.adrhi_flags & R500FP_CONSTANT_CLAMP); + + addr |= (type << 16); + addr |= (isclamp << 17); + + stride = type ? 4 : 6; + + DRM_DEBUG("r500fp %d %d type: %d\n", sz, addr, type); + if (!sz) + return 0; + if (sz * stride * 4 > drm_buffer_unprocessed(cmdbuf->buffer)) + return -EINVAL; + + BEGIN_RING(3 + sz * stride); + OUT_RING_REG(R500_GA_US_VECTOR_INDEX, addr); + OUT_RING(CP_PACKET0_TABLE(R500_GA_US_VECTOR_DATA, sz * stride - 1)); + OUT_RING_DRM_BUFFER(cmdbuf->buffer, sz * stride); + + ADVANCE_RING(); + + return 0; +} + + +/** + * Parses and validates a user-supplied command buffer and emits appropriate + * commands on the DMA ring buffer. + * Called by the ioctl handler function radeon_cp_cmdbuf. + */ +int r300_do_cp_cmdbuf(struct drm_device *dev, + struct drm_file *file_priv, + drm_radeon_kcmd_buffer_t *cmdbuf) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + struct drm_radeon_master_private *master_priv = file_priv->master->driver_priv; + struct drm_device_dma *dma = dev->dma; + struct drm_buf *buf = NULL; + int emit_dispatch_age = 0; + int ret = 0; + + DRM_DEBUG("\n"); + + /* pacify */ + r300_pacify(dev_priv); + + if (cmdbuf->nbox <= R300_SIMULTANEOUS_CLIPRECTS) { + ret = r300_emit_cliprects(dev_priv, cmdbuf, 0); + if (ret) + goto cleanup; + } + + while (drm_buffer_unprocessed(cmdbuf->buffer) + >= sizeof(drm_r300_cmd_header_t)) { + int idx; + drm_r300_cmd_header_t *header, stack_header; + + header = drm_buffer_read_object(cmdbuf->buffer, + sizeof(stack_header), &stack_header); + + switch (header->header.cmd_type) { + case R300_CMD_PACKET0: + DRM_DEBUG("R300_CMD_PACKET0\n"); + ret = r300_emit_packet0(dev_priv, cmdbuf, *header); + if (ret) { + DRM_ERROR("r300_emit_packet0 failed\n"); + goto cleanup; + } + break; + + case R300_CMD_VPU: + DRM_DEBUG("R300_CMD_VPU\n"); + ret = r300_emit_vpu(dev_priv, cmdbuf, *header); + if (ret) { + DRM_ERROR("r300_emit_vpu failed\n"); + goto cleanup; + } + break; + + case R300_CMD_PACKET3: + DRM_DEBUG("R300_CMD_PACKET3\n"); + ret = r300_emit_packet3(dev_priv, cmdbuf, *header); + if (ret) { + DRM_ERROR("r300_emit_packet3 failed\n"); + goto cleanup; + } + break; + + case R300_CMD_END3D: + DRM_DEBUG("R300_CMD_END3D\n"); + /* TODO: + Ideally userspace driver should not need to issue this call, + i.e. the drm driver should issue it automatically and prevent + lockups. + + In practice, we do not understand why this call is needed and what + it does (except for some vague guesses that it has to do with cache + coherence) and so the user space driver does it. + + Once we are sure which uses prevent lockups the code could be moved + into the kernel and the userspace driver will not + need to use this command. + + Note that issuing this command does not hurt anything + except, possibly, performance */ + r300_pacify(dev_priv); + break; + + case R300_CMD_CP_DELAY: + /* simple enough, we can do it here */ + DRM_DEBUG("R300_CMD_CP_DELAY\n"); + { + int i; + RING_LOCALS; + + BEGIN_RING(header->delay.count); + for (i = 0; i < header->delay.count; i++) + OUT_RING(RADEON_CP_PACKET2); + ADVANCE_RING(); + } + break; + + case R300_CMD_DMA_DISCARD: + DRM_DEBUG("RADEON_CMD_DMA_DISCARD\n"); + idx = header->dma.buf_idx; + if (idx < 0 || idx >= dma->buf_count) { + DRM_ERROR("buffer index %d (of %d max)\n", + idx, dma->buf_count - 1); + ret = -EINVAL; + goto cleanup; + } + + buf = dma->buflist[idx]; + if (buf->file_priv != file_priv || buf->pending) { + DRM_ERROR("bad buffer %p %p %d\n", + buf->file_priv, file_priv, + buf->pending); + ret = -EINVAL; + goto cleanup; + } + + emit_dispatch_age = 1; + r300_discard_buffer(dev, file_priv->master, buf); + break; + + case R300_CMD_WAIT: + DRM_DEBUG("R300_CMD_WAIT\n"); + r300_cmd_wait(dev_priv, *header); + break; + + case R300_CMD_SCRATCH: + DRM_DEBUG("R300_CMD_SCRATCH\n"); + ret = r300_scratch(dev_priv, cmdbuf, *header); + if (ret) { + DRM_ERROR("r300_scratch failed\n"); + goto cleanup; + } + break; + + case R300_CMD_R500FP: + if ((dev_priv->flags & RADEON_FAMILY_MASK) < CHIP_RV515) { + DRM_ERROR("Calling r500 command on r300 card\n"); + ret = -EINVAL; + goto cleanup; + } + DRM_DEBUG("R300_CMD_R500FP\n"); + ret = r300_emit_r500fp(dev_priv, cmdbuf, *header); + if (ret) { + DRM_ERROR("r300_emit_r500fp failed\n"); + goto cleanup; + } + break; + default: + DRM_ERROR("bad cmd_type %i at byte %d\n", + header->header.cmd_type, + cmdbuf->buffer->iterator - (int)sizeof(*header)); + ret = -EINVAL; + goto cleanup; + } + } + + DRM_DEBUG("END\n"); + + cleanup: + r300_pacify(dev_priv); + + /* We emit the vertex buffer age here, outside the pacifier "brackets" + * for two reasons: + * (1) This may coalesce multiple age emissions into a single one and + * (2) more importantly, some chips lock up hard when scratch registers + * are written inside the pacifier bracket. + */ + if (emit_dispatch_age) { + RING_LOCALS; + + /* Emit the vertex buffer age */ + BEGIN_RING(2); + RADEON_DISPATCH_AGE(master_priv->sarea_priv->last_dispatch); + ADVANCE_RING(); + } + + COMMIT_RING(); + + return ret; +} diff --git a/drivers/gpu/drm/radeon/r300_reg.h b/drivers/gpu/drm/radeon/r300_reg.h new file mode 100644 index 00000000..00c0d2ba --- /dev/null +++ b/drivers/gpu/drm/radeon/r300_reg.h @@ -0,0 +1,1789 @@ +/* + * Copyright 2005 Nicolai Haehnle et al. + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Nicolai Haehnle + * Jerome Glisse + */ +#ifndef _R300_REG_H_ +#define _R300_REG_H_ + +#define R300_SURF_TILE_MACRO (1<<16) +#define R300_SURF_TILE_MICRO (2<<16) +#define R300_SURF_TILE_BOTH (3<<16) + + +#define R300_MC_INIT_MISC_LAT_TIMER 0x180 +# define R300_MC_MISC__MC_CPR_INIT_LAT_SHIFT 0 +# define R300_MC_MISC__MC_VF_INIT_LAT_SHIFT 4 +# define R300_MC_MISC__MC_DISP0R_INIT_LAT_SHIFT 8 +# define R300_MC_MISC__MC_DISP1R_INIT_LAT_SHIFT 12 +# define R300_MC_MISC__MC_FIXED_INIT_LAT_SHIFT 16 +# define R300_MC_MISC__MC_E2R_INIT_LAT_SHIFT 20 +# define R300_MC_MISC__MC_SAME_PAGE_PRIO_SHIFT 24 +# define R300_MC_MISC__MC_GLOBW_INIT_LAT_SHIFT 28 + +#define R300_MC_INIT_GFX_LAT_TIMER 0x154 +# define R300_MC_MISC__MC_G3D0R_INIT_LAT_SHIFT 0 +# define R300_MC_MISC__MC_G3D1R_INIT_LAT_SHIFT 4 +# define R300_MC_MISC__MC_G3D2R_INIT_LAT_SHIFT 8 +# define R300_MC_MISC__MC_G3D3R_INIT_LAT_SHIFT 12 +# define R300_MC_MISC__MC_TX0R_INIT_LAT_SHIFT 16 +# define R300_MC_MISC__MC_TX1R_INIT_LAT_SHIFT 20 +# define R300_MC_MISC__MC_GLOBR_INIT_LAT_SHIFT 24 +# define R300_MC_MISC__MC_GLOBW_FULL_LAT_SHIFT 28 + +/* + * This file contains registers and constants for the R300. They have been + * found mostly by examining command buffers captured using glxtest, as well + * as by extrapolating some known registers and constants from the R200. + * I am fairly certain that they are correct unless stated otherwise + * in comments. + */ + +#define R300_SE_VPORT_XSCALE 0x1D98 +#define R300_SE_VPORT_XOFFSET 0x1D9C +#define R300_SE_VPORT_YSCALE 0x1DA0 +#define R300_SE_VPORT_YOFFSET 0x1DA4 +#define R300_SE_VPORT_ZSCALE 0x1DA8 +#define R300_SE_VPORT_ZOFFSET 0x1DAC + + +/* + * Vertex Array Processing (VAP) Control + * Stolen from r200 code from Christoph Brill (It's a guess!) + */ +#define R300_VAP_CNTL 0x2080 + +/* This register is written directly and also starts data section + * in many 3d CP_PACKET3's + */ +#define R300_VAP_VF_CNTL 0x2084 +# define R300_VAP_VF_CNTL__PRIM_TYPE__SHIFT 0 +# define R300_VAP_VF_CNTL__PRIM_NONE (0<<0) +# define R300_VAP_VF_CNTL__PRIM_POINTS (1<<0) +# define R300_VAP_VF_CNTL__PRIM_LINES (2<<0) +# define R300_VAP_VF_CNTL__PRIM_LINE_STRIP (3<<0) +# define R300_VAP_VF_CNTL__PRIM_TRIANGLES (4<<0) +# define R300_VAP_VF_CNTL__PRIM_TRIANGLE_FAN (5<<0) +# define R300_VAP_VF_CNTL__PRIM_TRIANGLE_STRIP (6<<0) +# define R300_VAP_VF_CNTL__PRIM_LINE_LOOP (12<<0) +# define R300_VAP_VF_CNTL__PRIM_QUADS (13<<0) +# define R300_VAP_VF_CNTL__PRIM_QUAD_STRIP (14<<0) +# define R300_VAP_VF_CNTL__PRIM_POLYGON (15<<0) + +# define R300_VAP_VF_CNTL__PRIM_WALK__SHIFT 4 + /* State based - direct writes to registers trigger vertex + generation */ +# define R300_VAP_VF_CNTL__PRIM_WALK_STATE_BASED (0<<4) +# define R300_VAP_VF_CNTL__PRIM_WALK_INDICES (1<<4) +# define R300_VAP_VF_CNTL__PRIM_WALK_VERTEX_LIST (2<<4) +# define R300_VAP_VF_CNTL__PRIM_WALK_VERTEX_EMBEDDED (3<<4) + + /* I don't think I saw these three used.. */ +# define R300_VAP_VF_CNTL__COLOR_ORDER__SHIFT 6 +# define R300_VAP_VF_CNTL__TCL_OUTPUT_CTL_ENA__SHIFT 9 +# define R300_VAP_VF_CNTL__PROG_STREAM_ENA__SHIFT 10 + + /* index size - when not set the indices are assumed to be 16 bit */ +# define R300_VAP_VF_CNTL__INDEX_SIZE_32bit (1<<11) + /* number of vertices */ +# define R300_VAP_VF_CNTL__NUM_VERTICES__SHIFT 16 + +/* BEGIN: Wild guesses */ +#define R300_VAP_OUTPUT_VTX_FMT_0 0x2090 +# define R300_VAP_OUTPUT_VTX_FMT_0__POS_PRESENT (1<<0) +# define R300_VAP_OUTPUT_VTX_FMT_0__COLOR_PRESENT (1<<1) +# define R300_VAP_OUTPUT_VTX_FMT_0__COLOR_1_PRESENT (1<<2) /* GUESS */ +# define R300_VAP_OUTPUT_VTX_FMT_0__COLOR_2_PRESENT (1<<3) /* GUESS */ +# define R300_VAP_OUTPUT_VTX_FMT_0__COLOR_3_PRESENT (1<<4) /* GUESS */ +# define R300_VAP_OUTPUT_VTX_FMT_0__PT_SIZE_PRESENT (1<<16) /* GUESS */ + +#define R300_VAP_OUTPUT_VTX_FMT_1 0x2094 + /* each of the following is 3 bits wide, specifies number + of components */ +# define R300_VAP_OUTPUT_VTX_FMT_1__TEX_0_COMP_CNT_SHIFT 0 +# define R300_VAP_OUTPUT_VTX_FMT_1__TEX_1_COMP_CNT_SHIFT 3 +# define R300_VAP_OUTPUT_VTX_FMT_1__TEX_2_COMP_CNT_SHIFT 6 +# define R300_VAP_OUTPUT_VTX_FMT_1__TEX_3_COMP_CNT_SHIFT 9 +# define R300_VAP_OUTPUT_VTX_FMT_1__TEX_4_COMP_CNT_SHIFT 12 +# define R300_VAP_OUTPUT_VTX_FMT_1__TEX_5_COMP_CNT_SHIFT 15 +# define R300_VAP_OUTPUT_VTX_FMT_1__TEX_6_COMP_CNT_SHIFT 18 +# define R300_VAP_OUTPUT_VTX_FMT_1__TEX_7_COMP_CNT_SHIFT 21 +/* END: Wild guesses */ + +#define R300_SE_VTE_CNTL 0x20b0 +# define R300_VPORT_X_SCALE_ENA 0x00000001 +# define R300_VPORT_X_OFFSET_ENA 0x00000002 +# define R300_VPORT_Y_SCALE_ENA 0x00000004 +# define R300_VPORT_Y_OFFSET_ENA 0x00000008 +# define R300_VPORT_Z_SCALE_ENA 0x00000010 +# define R300_VPORT_Z_OFFSET_ENA 0x00000020 +# define R300_VTX_XY_FMT 0x00000100 +# define R300_VTX_Z_FMT 0x00000200 +# define R300_VTX_W0_FMT 0x00000400 +# define R300_VTX_W0_NORMALIZE 0x00000800 +# define R300_VTX_ST_DENORMALIZED 0x00001000 + +/* BEGIN: Vertex data assembly - lots of uncertainties */ + +/* gap */ + +#define R300_VAP_CNTL_STATUS 0x2140 +# define R300_VC_NO_SWAP (0 << 0) +# define R300_VC_16BIT_SWAP (1 << 0) +# define R300_VC_32BIT_SWAP (2 << 0) +# define R300_VAP_TCL_BYPASS (1 << 8) + +/* gap */ + +/* Where do we get our vertex data? + * + * Vertex data either comes either from immediate mode registers or from + * vertex arrays. + * There appears to be no mixed mode (though we can force the pitch of + * vertex arrays to 0, effectively reusing the same element over and over + * again). + * + * Immediate mode is controlled by the INPUT_CNTL registers. I am not sure + * if these registers influence vertex array processing. + * + * Vertex arrays are controlled via the 3D_LOAD_VBPNTR packet3. + * + * In both cases, vertex attributes are then passed through INPUT_ROUTE. + * + * Beginning with INPUT_ROUTE_0_0 is a list of WORDs that route vertex data + * into the vertex processor's input registers. + * The first word routes the first input, the second word the second, etc. + * The corresponding input is routed into the register with the given index. + * The list is ended by a word with INPUT_ROUTE_END set. + * + * Always set COMPONENTS_4 in immediate mode. + */ + +#define R300_VAP_INPUT_ROUTE_0_0 0x2150 +# define R300_INPUT_ROUTE_COMPONENTS_1 (0 << 0) +# define R300_INPUT_ROUTE_COMPONENTS_2 (1 << 0) +# define R300_INPUT_ROUTE_COMPONENTS_3 (2 << 0) +# define R300_INPUT_ROUTE_COMPONENTS_4 (3 << 0) +# define R300_INPUT_ROUTE_COMPONENTS_RGBA (4 << 0) /* GUESS */ +# define R300_VAP_INPUT_ROUTE_IDX_SHIFT 8 +# define R300_VAP_INPUT_ROUTE_IDX_MASK (31 << 8) /* GUESS */ +# define R300_VAP_INPUT_ROUTE_END (1 << 13) +# define R300_INPUT_ROUTE_IMMEDIATE_MODE (0 << 14) /* GUESS */ +# define R300_INPUT_ROUTE_FLOAT (1 << 14) /* GUESS */ +# define R300_INPUT_ROUTE_UNSIGNED_BYTE (2 << 14) /* GUESS */ +# define R300_INPUT_ROUTE_FLOAT_COLOR (3 << 14) /* GUESS */ +#define R300_VAP_INPUT_ROUTE_0_1 0x2154 +#define R300_VAP_INPUT_ROUTE_0_2 0x2158 +#define R300_VAP_INPUT_ROUTE_0_3 0x215C +#define R300_VAP_INPUT_ROUTE_0_4 0x2160 +#define R300_VAP_INPUT_ROUTE_0_5 0x2164 +#define R300_VAP_INPUT_ROUTE_0_6 0x2168 +#define R300_VAP_INPUT_ROUTE_0_7 0x216C + +/* gap */ + +/* Notes: + * - always set up to produce at least two attributes: + * if vertex program uses only position, fglrx will set normal, too + * - INPUT_CNTL_0_COLOR and INPUT_CNTL_COLOR bits are always equal. + */ +#define R300_VAP_INPUT_CNTL_0 0x2180 +# define R300_INPUT_CNTL_0_COLOR 0x00000001 +#define R300_VAP_INPUT_CNTL_1 0x2184 +# define R300_INPUT_CNTL_POS 0x00000001 +# define R300_INPUT_CNTL_NORMAL 0x00000002 +# define R300_INPUT_CNTL_COLOR 0x00000004 +# define R300_INPUT_CNTL_TC0 0x00000400 +# define R300_INPUT_CNTL_TC1 0x00000800 +# define R300_INPUT_CNTL_TC2 0x00001000 /* GUESS */ +# define R300_INPUT_CNTL_TC3 0x00002000 /* GUESS */ +# define R300_INPUT_CNTL_TC4 0x00004000 /* GUESS */ +# define R300_INPUT_CNTL_TC5 0x00008000 /* GUESS */ +# define R300_INPUT_CNTL_TC6 0x00010000 /* GUESS */ +# define R300_INPUT_CNTL_TC7 0x00020000 /* GUESS */ + +/* gap */ + +/* Words parallel to INPUT_ROUTE_0; All words that are active in INPUT_ROUTE_0 + * are set to a swizzling bit pattern, other words are 0. + * + * In immediate mode, the pattern is always set to xyzw. In vertex array + * mode, the swizzling pattern is e.g. used to set zw components in texture + * coordinates with only tweo components. + */ +#define R300_VAP_INPUT_ROUTE_1_0 0x21E0 +# define R300_INPUT_ROUTE_SELECT_X 0 +# define R300_INPUT_ROUTE_SELECT_Y 1 +# define R300_INPUT_ROUTE_SELECT_Z 2 +# define R300_INPUT_ROUTE_SELECT_W 3 +# define R300_INPUT_ROUTE_SELECT_ZERO 4 +# define R300_INPUT_ROUTE_SELECT_ONE 5 +# define R300_INPUT_ROUTE_SELECT_MASK 7 +# define R300_INPUT_ROUTE_X_SHIFT 0 +# define R300_INPUT_ROUTE_Y_SHIFT 3 +# define R300_INPUT_ROUTE_Z_SHIFT 6 +# define R300_INPUT_ROUTE_W_SHIFT 9 +# define R300_INPUT_ROUTE_ENABLE (15 << 12) +#define R300_VAP_INPUT_ROUTE_1_1 0x21E4 +#define R300_VAP_INPUT_ROUTE_1_2 0x21E8 +#define R300_VAP_INPUT_ROUTE_1_3 0x21EC +#define R300_VAP_INPUT_ROUTE_1_4 0x21F0 +#define R300_VAP_INPUT_ROUTE_1_5 0x21F4 +#define R300_VAP_INPUT_ROUTE_1_6 0x21F8 +#define R300_VAP_INPUT_ROUTE_1_7 0x21FC + +/* END: Vertex data assembly */ + +/* gap */ + +/* BEGIN: Upload vertex program and data */ + +/* + * The programmable vertex shader unit has a memory bank of unknown size + * that can be written to in 16 byte units by writing the address into + * UPLOAD_ADDRESS, followed by data in UPLOAD_DATA (multiples of 4 DWORDs). + * + * Pointers into the memory bank are always in multiples of 16 bytes. + * + * The memory bank is divided into areas with fixed meaning. + * + * Starting at address UPLOAD_PROGRAM: Vertex program instructions. + * Native limits reported by drivers from ATI suggest size 256 (i.e. 4KB), + * whereas the difference between known addresses suggests size 512. + * + * Starting at address UPLOAD_PARAMETERS: Vertex program parameters. + * Native reported limits and the VPI layout suggest size 256, whereas + * difference between known addresses suggests size 512. + * + * At address UPLOAD_POINTSIZE is a vector (0, 0, ps, 0), where ps is the + * floating point pointsize. The exact purpose of this state is uncertain, + * as there is also the R300_RE_POINTSIZE register. + * + * Multiple vertex programs and parameter sets can be loaded at once, + * which could explain the size discrepancy. + */ +#define R300_VAP_PVS_UPLOAD_ADDRESS 0x2200 +# define R300_PVS_UPLOAD_PROGRAM 0x00000000 +# define R300_PVS_UPLOAD_PARAMETERS 0x00000200 +# define R300_PVS_UPLOAD_POINTSIZE 0x00000406 + +/* gap */ + +#define R300_VAP_PVS_UPLOAD_DATA 0x2208 + +/* END: Upload vertex program and data */ + +/* gap */ + +/* I do not know the purpose of this register. However, I do know that + * it is set to 221C_CLEAR for clear operations and to 221C_NORMAL + * for normal rendering. + */ +#define R300_VAP_UNKNOWN_221C 0x221C +# define R300_221C_NORMAL 0x00000000 +# define R300_221C_CLEAR 0x0001C000 + +/* These seem to be per-pixel and per-vertex X and Y clipping planes. The first + * plane is per-pixel and the second plane is per-vertex. + * + * This was determined by experimentation alone but I believe it is correct. + * + * These registers are called X_QUAD0_1_FL to X_QUAD0_4_FL by glxtest. + */ +#define R300_VAP_CLIP_X_0 0x2220 +#define R300_VAP_CLIP_X_1 0x2224 +#define R300_VAP_CLIP_Y_0 0x2228 +#define R300_VAP_CLIP_Y_1 0x2230 + +/* gap */ + +/* Sometimes, END_OF_PKT and 0x2284=0 are the only commands sent between + * rendering commands and overwriting vertex program parameters. + * Therefore, I suspect writing zero to 0x2284 synchronizes the engine and + * avoids bugs caused by still running shaders reading bad data from memory. + */ +#define R300_VAP_PVS_STATE_FLUSH_REG 0x2284 + +/* Absolutely no clue what this register is about. */ +#define R300_VAP_UNKNOWN_2288 0x2288 +# define R300_2288_R300 0x00750000 /* -- nh */ +# define R300_2288_RV350 0x0000FFFF /* -- Vladimir */ + +/* gap */ + +/* Addresses are relative to the vertex program instruction area of the + * memory bank. PROGRAM_END points to the last instruction of the active + * program + * + * The meaning of the two UNKNOWN fields is obviously not known. However, + * experiments so far have shown that both *must* point to an instruction + * inside the vertex program, otherwise the GPU locks up. + * + * fglrx usually sets CNTL_3_UNKNOWN to the end of the program and + * R300_PVS_CNTL_1_POS_END_SHIFT points to instruction where last write to + * position takes place. + * + * Most likely this is used to ignore rest of the program in cases + * where group of verts arent visible. For some reason this "section" + * is sometimes accepted other instruction that have no relationship with + * position calculations. + */ +#define R300_VAP_PVS_CNTL_1 0x22D0 +# define R300_PVS_CNTL_1_PROGRAM_START_SHIFT 0 +# define R300_PVS_CNTL_1_POS_END_SHIFT 10 +# define R300_PVS_CNTL_1_PROGRAM_END_SHIFT 20 +/* Addresses are relative the the vertex program parameters area. */ +#define R300_VAP_PVS_CNTL_2 0x22D4 +# define R300_PVS_CNTL_2_PARAM_OFFSET_SHIFT 0 +# define R300_PVS_CNTL_2_PARAM_COUNT_SHIFT 16 +#define R300_VAP_PVS_CNTL_3 0x22D8 +# define R300_PVS_CNTL_3_PROGRAM_UNKNOWN_SHIFT 10 +# define R300_PVS_CNTL_3_PROGRAM_UNKNOWN2_SHIFT 0 + +/* The entire range from 0x2300 to 0x2AC inclusive seems to be used for + * immediate vertices + */ +#define R300_VAP_VTX_COLOR_R 0x2464 +#define R300_VAP_VTX_COLOR_G 0x2468 +#define R300_VAP_VTX_COLOR_B 0x246C +#define R300_VAP_VTX_POS_0_X_1 0x2490 /* used for glVertex2*() */ +#define R300_VAP_VTX_POS_0_Y_1 0x2494 +#define R300_VAP_VTX_COLOR_PKD 0x249C /* RGBA */ +#define R300_VAP_VTX_POS_0_X_2 0x24A0 /* used for glVertex3*() */ +#define R300_VAP_VTX_POS_0_Y_2 0x24A4 +#define R300_VAP_VTX_POS_0_Z_2 0x24A8 +/* write 0 to indicate end of packet? */ +#define R300_VAP_VTX_END_OF_PKT 0x24AC + +/* gap */ + +/* These are values from r300_reg/r300_reg.h - they are known to be correct + * and are here so we can use one register file instead of several + * - Vladimir + */ +#define R300_GB_VAP_RASTER_VTX_FMT_0 0x4000 +# define R300_GB_VAP_RASTER_VTX_FMT_0__POS_PRESENT (1<<0) +# define R300_GB_VAP_RASTER_VTX_FMT_0__COLOR_0_PRESENT (1<<1) +# define R300_GB_VAP_RASTER_VTX_FMT_0__COLOR_1_PRESENT (1<<2) +# define R300_GB_VAP_RASTER_VTX_FMT_0__COLOR_2_PRESENT (1<<3) +# define R300_GB_VAP_RASTER_VTX_FMT_0__COLOR_3_PRESENT (1<<4) +# define R300_GB_VAP_RASTER_VTX_FMT_0__COLOR_SPACE (0xf<<5) +# define R300_GB_VAP_RASTER_VTX_FMT_0__PT_SIZE_PRESENT (0x1<<16) + +#define R300_GB_VAP_RASTER_VTX_FMT_1 0x4004 + /* each of the following is 3 bits wide, specifies number + of components */ +# define R300_GB_VAP_RASTER_VTX_FMT_1__TEX_0_COMP_CNT_SHIFT 0 +# define R300_GB_VAP_RASTER_VTX_FMT_1__TEX_1_COMP_CNT_SHIFT 3 +# define R300_GB_VAP_RASTER_VTX_FMT_1__TEX_2_COMP_CNT_SHIFT 6 +# define R300_GB_VAP_RASTER_VTX_FMT_1__TEX_3_COMP_CNT_SHIFT 9 +# define R300_GB_VAP_RASTER_VTX_FMT_1__TEX_4_COMP_CNT_SHIFT 12 +# define R300_GB_VAP_RASTER_VTX_FMT_1__TEX_5_COMP_CNT_SHIFT 15 +# define R300_GB_VAP_RASTER_VTX_FMT_1__TEX_6_COMP_CNT_SHIFT 18 +# define R300_GB_VAP_RASTER_VTX_FMT_1__TEX_7_COMP_CNT_SHIFT 21 + +/* UNK30 seems to enables point to quad transformation on textures + * (or something closely related to that). + * This bit is rather fatal at the time being due to lackings at pixel + * shader side + */ +#define R300_GB_ENABLE 0x4008 +# define R300_GB_POINT_STUFF_ENABLE (1<<0) +# define R300_GB_LINE_STUFF_ENABLE (1<<1) +# define R300_GB_TRIANGLE_STUFF_ENABLE (1<<2) +# define R300_GB_STENCIL_AUTO_ENABLE (1<<4) +# define R300_GB_UNK31 (1<<31) + /* each of the following is 2 bits wide */ +#define R300_GB_TEX_REPLICATE 0 +#define R300_GB_TEX_ST 1 +#define R300_GB_TEX_STR 2 +# define R300_GB_TEX0_SOURCE_SHIFT 16 +# define R300_GB_TEX1_SOURCE_SHIFT 18 +# define R300_GB_TEX2_SOURCE_SHIFT 20 +# define R300_GB_TEX3_SOURCE_SHIFT 22 +# define R300_GB_TEX4_SOURCE_SHIFT 24 +# define R300_GB_TEX5_SOURCE_SHIFT 26 +# define R300_GB_TEX6_SOURCE_SHIFT 28 +# define R300_GB_TEX7_SOURCE_SHIFT 30 + +/* MSPOS - positions for multisample antialiasing (?) */ +#define R300_GB_MSPOS0 0x4010 + /* shifts - each of the fields is 4 bits */ +# define R300_GB_MSPOS0__MS_X0_SHIFT 0 +# define R300_GB_MSPOS0__MS_Y0_SHIFT 4 +# define R300_GB_MSPOS0__MS_X1_SHIFT 8 +# define R300_GB_MSPOS0__MS_Y1_SHIFT 12 +# define R300_GB_MSPOS0__MS_X2_SHIFT 16 +# define R300_GB_MSPOS0__MS_Y2_SHIFT 20 +# define R300_GB_MSPOS0__MSBD0_Y 24 +# define R300_GB_MSPOS0__MSBD0_X 28 + +#define R300_GB_MSPOS1 0x4014 +# define R300_GB_MSPOS1__MS_X3_SHIFT 0 +# define R300_GB_MSPOS1__MS_Y3_SHIFT 4 +# define R300_GB_MSPOS1__MS_X4_SHIFT 8 +# define R300_GB_MSPOS1__MS_Y4_SHIFT 12 +# define R300_GB_MSPOS1__MS_X5_SHIFT 16 +# define R300_GB_MSPOS1__MS_Y5_SHIFT 20 +# define R300_GB_MSPOS1__MSBD1 24 + + +#define R300_GB_TILE_CONFIG 0x4018 +# define R300_GB_TILE_ENABLE (1<<0) +# define R300_GB_TILE_PIPE_COUNT_RV300 0 +# define R300_GB_TILE_PIPE_COUNT_R300 (3<<1) +# define R300_GB_TILE_PIPE_COUNT_R420 (7<<1) +# define R300_GB_TILE_PIPE_COUNT_RV410 (3<<1) +# define R300_GB_TILE_SIZE_8 0 +# define R300_GB_TILE_SIZE_16 (1<<4) +# define R300_GB_TILE_SIZE_32 (2<<4) +# define R300_GB_SUPER_SIZE_1 (0<<6) +# define R300_GB_SUPER_SIZE_2 (1<<6) +# define R300_GB_SUPER_SIZE_4 (2<<6) +# define R300_GB_SUPER_SIZE_8 (3<<6) +# define R300_GB_SUPER_SIZE_16 (4<<6) +# define R300_GB_SUPER_SIZE_32 (5<<6) +# define R300_GB_SUPER_SIZE_64 (6<<6) +# define R300_GB_SUPER_SIZE_128 (7<<6) +# define R300_GB_SUPER_X_SHIFT 9 /* 3 bits wide */ +# define R300_GB_SUPER_Y_SHIFT 12 /* 3 bits wide */ +# define R300_GB_SUPER_TILE_A 0 +# define R300_GB_SUPER_TILE_B (1<<15) +# define R300_GB_SUBPIXEL_1_12 0 +# define R300_GB_SUBPIXEL_1_16 (1<<16) + +#define R300_GB_FIFO_SIZE 0x4024 + /* each of the following is 2 bits wide */ +#define R300_GB_FIFO_SIZE_32 0 +#define R300_GB_FIFO_SIZE_64 1 +#define R300_GB_FIFO_SIZE_128 2 +#define R300_GB_FIFO_SIZE_256 3 +# define R300_SC_IFIFO_SIZE_SHIFT 0 +# define R300_SC_TZFIFO_SIZE_SHIFT 2 +# define R300_SC_BFIFO_SIZE_SHIFT 4 + +# define R300_US_OFIFO_SIZE_SHIFT 12 +# define R300_US_WFIFO_SIZE_SHIFT 14 + /* the following use the same constants as above, but meaning is + is times 2 (i.e. instead of 32 words it means 64 */ +# define R300_RS_TFIFO_SIZE_SHIFT 6 +# define R300_RS_CFIFO_SIZE_SHIFT 8 +# define R300_US_RAM_SIZE_SHIFT 10 + /* watermarks, 3 bits wide */ +# define R300_RS_HIGHWATER_COL_SHIFT 16 +# define R300_RS_HIGHWATER_TEX_SHIFT 19 +# define R300_OFIFO_HIGHWATER_SHIFT 22 /* two bits only */ +# define R300_CUBE_FIFO_HIGHWATER_COL_SHIFT 24 + +#define R300_GB_SELECT 0x401C +# define R300_GB_FOG_SELECT_C0A 0 +# define R300_GB_FOG_SELECT_C1A 1 +# define R300_GB_FOG_SELECT_C2A 2 +# define R300_GB_FOG_SELECT_C3A 3 +# define R300_GB_FOG_SELECT_1_1_W 4 +# define R300_GB_FOG_SELECT_Z 5 +# define R300_GB_DEPTH_SELECT_Z 0 +# define R300_GB_DEPTH_SELECT_1_1_W (1<<3) +# define R300_GB_W_SELECT_1_W 0 +# define R300_GB_W_SELECT_1 (1<<4) + +#define R300_GB_AA_CONFIG 0x4020 +# define R300_AA_DISABLE 0x00 +# define R300_AA_ENABLE 0x01 +# define R300_AA_SUBSAMPLES_2 0 +# define R300_AA_SUBSAMPLES_3 (1<<1) +# define R300_AA_SUBSAMPLES_4 (2<<1) +# define R300_AA_SUBSAMPLES_6 (3<<1) + +/* gap */ + +/* Zero to flush caches. */ +#define R300_TX_INVALTAGS 0x4100 +#define R300_TX_FLUSH 0x0 + +/* The upper enable bits are guessed, based on fglrx reported limits. */ +#define R300_TX_ENABLE 0x4104 +# define R300_TX_ENABLE_0 (1 << 0) +# define R300_TX_ENABLE_1 (1 << 1) +# define R300_TX_ENABLE_2 (1 << 2) +# define R300_TX_ENABLE_3 (1 << 3) +# define R300_TX_ENABLE_4 (1 << 4) +# define R300_TX_ENABLE_5 (1 << 5) +# define R300_TX_ENABLE_6 (1 << 6) +# define R300_TX_ENABLE_7 (1 << 7) +# define R300_TX_ENABLE_8 (1 << 8) +# define R300_TX_ENABLE_9 (1 << 9) +# define R300_TX_ENABLE_10 (1 << 10) +# define R300_TX_ENABLE_11 (1 << 11) +# define R300_TX_ENABLE_12 (1 << 12) +# define R300_TX_ENABLE_13 (1 << 13) +# define R300_TX_ENABLE_14 (1 << 14) +# define R300_TX_ENABLE_15 (1 << 15) + +/* The pointsize is given in multiples of 6. The pointsize can be + * enormous: Clear() renders a single point that fills the entire + * framebuffer. + */ +#define R300_RE_POINTSIZE 0x421C +# define R300_POINTSIZE_Y_SHIFT 0 +# define R300_POINTSIZE_Y_MASK (0xFFFF << 0) /* GUESS */ +# define R300_POINTSIZE_X_SHIFT 16 +# define R300_POINTSIZE_X_MASK (0xFFFF << 16) /* GUESS */ +# define R300_POINTSIZE_MAX (R300_POINTSIZE_Y_MASK / 6) + +/* The line width is given in multiples of 6. + * In default mode lines are classified as vertical lines. + * HO: horizontal + * VE: vertical or horizontal + * HO & VE: no classification + */ +#define R300_RE_LINE_CNT 0x4234 +# define R300_LINESIZE_SHIFT 0 +# define R300_LINESIZE_MASK (0xFFFF << 0) /* GUESS */ +# define R300_LINESIZE_MAX (R300_LINESIZE_MASK / 6) +# define R300_LINE_CNT_HO (1 << 16) +# define R300_LINE_CNT_VE (1 << 17) + +/* Some sort of scale or clamp value for texcoordless textures. */ +#define R300_RE_UNK4238 0x4238 + +/* Something shade related */ +#define R300_RE_SHADE 0x4274 + +#define R300_RE_SHADE_MODEL 0x4278 +# define R300_RE_SHADE_MODEL_SMOOTH 0x3aaaa +# define R300_RE_SHADE_MODEL_FLAT 0x39595 + +/* Dangerous */ +#define R300_RE_POLYGON_MODE 0x4288 +# define R300_PM_ENABLED (1 << 0) +# define R300_PM_FRONT_POINT (0 << 0) +# define R300_PM_BACK_POINT (0 << 0) +# define R300_PM_FRONT_LINE (1 << 4) +# define R300_PM_FRONT_FILL (1 << 5) +# define R300_PM_BACK_LINE (1 << 7) +# define R300_PM_BACK_FILL (1 << 8) + +/* Fog parameters */ +#define R300_RE_FOG_SCALE 0x4294 +#define R300_RE_FOG_START 0x4298 + +/* Not sure why there are duplicate of factor and constant values. + * My best guess so far is that there are separate zbiases for test and write. + * Ordering might be wrong. + * Some of the tests indicate that fgl has a fallback implementation of zbias + * via pixel shaders. + */ +#define R300_RE_ZBIAS_CNTL 0x42A0 /* GUESS */ +#define R300_RE_ZBIAS_T_FACTOR 0x42A4 +#define R300_RE_ZBIAS_T_CONSTANT 0x42A8 +#define R300_RE_ZBIAS_W_FACTOR 0x42AC +#define R300_RE_ZBIAS_W_CONSTANT 0x42B0 + +/* This register needs to be set to (1<<1) for RV350 to correctly + * perform depth test (see --vb-triangles in r300_demo) + * Don't know about other chips. - Vladimir + * This is set to 3 when GL_POLYGON_OFFSET_FILL is on. + * My guess is that there are two bits for each zbias primitive + * (FILL, LINE, POINT). + * One to enable depth test and one for depth write. + * Yet this doesn't explain why depth writes work ... + */ +#define R300_RE_OCCLUSION_CNTL 0x42B4 +# define R300_OCCLUSION_ON (1<<1) + +#define R300_RE_CULL_CNTL 0x42B8 +# define R300_CULL_FRONT (1 << 0) +# define R300_CULL_BACK (1 << 1) +# define R300_FRONT_FACE_CCW (0 << 2) +# define R300_FRONT_FACE_CW (1 << 2) + + +/* BEGIN: Rasterization / Interpolators - many guesses */ + +/* 0_UNKNOWN_18 has always been set except for clear operations. + * TC_CNT is the number of incoming texture coordinate sets (i.e. it depends + * on the vertex program, *not* the fragment program) + */ +#define R300_RS_CNTL_0 0x4300 +# define R300_RS_CNTL_TC_CNT_SHIFT 2 +# define R300_RS_CNTL_TC_CNT_MASK (7 << 2) + /* number of color interpolators used */ +# define R300_RS_CNTL_CI_CNT_SHIFT 7 +# define R300_RS_CNTL_0_UNKNOWN_18 (1 << 18) + /* Guess: RS_CNTL_1 holds the index of the highest used RS_ROUTE_n + register. */ +#define R300_RS_CNTL_1 0x4304 + +/* gap */ + +/* Only used for texture coordinates. + * Use the source field to route texture coordinate input from the + * vertex program to the desired interpolator. Note that the source + * field is relative to the outputs the vertex program *actually* + * writes. If a vertex program only writes texcoord[1], this will + * be source index 0. + * Set INTERP_USED on all interpolators that produce data used by + * the fragment program. INTERP_USED looks like a swizzling mask, + * but I haven't seen it used that way. + * + * Note: The _UNKNOWN constants are always set in their respective + * register. I don't know if this is necessary. + */ +#define R300_RS_INTERP_0 0x4310 +#define R300_RS_INTERP_1 0x4314 +# define R300_RS_INTERP_1_UNKNOWN 0x40 +#define R300_RS_INTERP_2 0x4318 +# define R300_RS_INTERP_2_UNKNOWN 0x80 +#define R300_RS_INTERP_3 0x431C +# define R300_RS_INTERP_3_UNKNOWN 0xC0 +#define R300_RS_INTERP_4 0x4320 +#define R300_RS_INTERP_5 0x4324 +#define R300_RS_INTERP_6 0x4328 +#define R300_RS_INTERP_7 0x432C +# define R300_RS_INTERP_SRC_SHIFT 2 +# define R300_RS_INTERP_SRC_MASK (7 << 2) +# define R300_RS_INTERP_USED 0x00D10000 + +/* These DWORDs control how vertex data is routed into fragment program + * registers, after interpolators. + */ +#define R300_RS_ROUTE_0 0x4330 +#define R300_RS_ROUTE_1 0x4334 +#define R300_RS_ROUTE_2 0x4338 +#define R300_RS_ROUTE_3 0x433C /* GUESS */ +#define R300_RS_ROUTE_4 0x4340 /* GUESS */ +#define R300_RS_ROUTE_5 0x4344 /* GUESS */ +#define R300_RS_ROUTE_6 0x4348 /* GUESS */ +#define R300_RS_ROUTE_7 0x434C /* GUESS */ +# define R300_RS_ROUTE_SOURCE_INTERP_0 0 +# define R300_RS_ROUTE_SOURCE_INTERP_1 1 +# define R300_RS_ROUTE_SOURCE_INTERP_2 2 +# define R300_RS_ROUTE_SOURCE_INTERP_3 3 +# define R300_RS_ROUTE_SOURCE_INTERP_4 4 +# define R300_RS_ROUTE_SOURCE_INTERP_5 5 /* GUESS */ +# define R300_RS_ROUTE_SOURCE_INTERP_6 6 /* GUESS */ +# define R300_RS_ROUTE_SOURCE_INTERP_7 7 /* GUESS */ +# define R300_RS_ROUTE_ENABLE (1 << 3) /* GUESS */ +# define R300_RS_ROUTE_DEST_SHIFT 6 +# define R300_RS_ROUTE_DEST_MASK (31 << 6) /* GUESS */ + +/* Special handling for color: When the fragment program uses color, + * the ROUTE_0_COLOR bit is set and ROUTE_0_COLOR_DEST contains the + * color register index. + * + * Apperently you may set the R300_RS_ROUTE_0_COLOR bit, but not provide any + * R300_RS_ROUTE_0_COLOR_DEST value; this setup is used for clearing the state. + * See r300_ioctl.c:r300EmitClearState. I'm not sure if this setup is strictly + * correct or not. - Oliver. + */ +# define R300_RS_ROUTE_0_COLOR (1 << 14) +# define R300_RS_ROUTE_0_COLOR_DEST_SHIFT 17 +# define R300_RS_ROUTE_0_COLOR_DEST_MASK (31 << 17) /* GUESS */ +/* As above, but for secondary color */ +# define R300_RS_ROUTE_1_COLOR1 (1 << 14) +# define R300_RS_ROUTE_1_COLOR1_DEST_SHIFT 17 +# define R300_RS_ROUTE_1_COLOR1_DEST_MASK (31 << 17) +# define R300_RS_ROUTE_1_UNKNOWN11 (1 << 11) +/* END: Rasterization / Interpolators - many guesses */ + +/* Hierarchical Z Enable */ +#define R300_SC_HYPERZ 0x43a4 +# define R300_SC_HYPERZ_DISABLE (0 << 0) +# define R300_SC_HYPERZ_ENABLE (1 << 0) +# define R300_SC_HYPERZ_MIN (0 << 1) +# define R300_SC_HYPERZ_MAX (1 << 1) +# define R300_SC_HYPERZ_ADJ_256 (0 << 2) +# define R300_SC_HYPERZ_ADJ_128 (1 << 2) +# define R300_SC_HYPERZ_ADJ_64 (2 << 2) +# define R300_SC_HYPERZ_ADJ_32 (3 << 2) +# define R300_SC_HYPERZ_ADJ_16 (4 << 2) +# define R300_SC_HYPERZ_ADJ_8 (5 << 2) +# define R300_SC_HYPERZ_ADJ_4 (6 << 2) +# define R300_SC_HYPERZ_ADJ_2 (7 << 2) +# define R300_SC_HYPERZ_HZ_Z0MIN_NO (0 << 5) +# define R300_SC_HYPERZ_HZ_Z0MIN (1 << 5) +# define R300_SC_HYPERZ_HZ_Z0MAX_NO (0 << 6) +# define R300_SC_HYPERZ_HZ_Z0MAX (1 << 6) + +#define R300_SC_EDGERULE 0x43a8 + +/* BEGIN: Scissors and cliprects */ + +/* There are four clipping rectangles. Their corner coordinates are inclusive. + * Every pixel is assigned a number from 0 and 15 by setting bits 0-3 depending + * on whether the pixel is inside cliprects 0-3, respectively. For example, + * if a pixel is inside cliprects 0 and 1, but outside 2 and 3, it is assigned + * the number 3 (binary 0011). + * Iff the bit corresponding to the pixel's number in RE_CLIPRECT_CNTL is set, + * the pixel is rasterized. + * + * In addition to this, there is a scissors rectangle. Only pixels inside the + * scissors rectangle are drawn. (coordinates are inclusive) + * + * For some reason, the top-left corner of the framebuffer is at (1440, 1440) + * for the purpose of clipping and scissors. + */ +#define R300_RE_CLIPRECT_TL_0 0x43B0 +#define R300_RE_CLIPRECT_BR_0 0x43B4 +#define R300_RE_CLIPRECT_TL_1 0x43B8 +#define R300_RE_CLIPRECT_BR_1 0x43BC +#define R300_RE_CLIPRECT_TL_2 0x43C0 +#define R300_RE_CLIPRECT_BR_2 0x43C4 +#define R300_RE_CLIPRECT_TL_3 0x43C8 +#define R300_RE_CLIPRECT_BR_3 0x43CC +# define R300_CLIPRECT_OFFSET 1440 +# define R300_CLIPRECT_MASK 0x1FFF +# define R300_CLIPRECT_X_SHIFT 0 +# define R300_CLIPRECT_X_MASK (0x1FFF << 0) +# define R300_CLIPRECT_Y_SHIFT 13 +# define R300_CLIPRECT_Y_MASK (0x1FFF << 13) +#define R300_RE_CLIPRECT_CNTL 0x43D0 +# define R300_CLIP_OUT (1 << 0) +# define R300_CLIP_0 (1 << 1) +# define R300_CLIP_1 (1 << 2) +# define R300_CLIP_10 (1 << 3) +# define R300_CLIP_2 (1 << 4) +# define R300_CLIP_20 (1 << 5) +# define R300_CLIP_21 (1 << 6) +# define R300_CLIP_210 (1 << 7) +# define R300_CLIP_3 (1 << 8) +# define R300_CLIP_30 (1 << 9) +# define R300_CLIP_31 (1 << 10) +# define R300_CLIP_310 (1 << 11) +# define R300_CLIP_32 (1 << 12) +# define R300_CLIP_320 (1 << 13) +# define R300_CLIP_321 (1 << 14) +# define R300_CLIP_3210 (1 << 15) + +/* gap */ + +#define R300_RE_SCISSORS_TL 0x43E0 +#define R300_RE_SCISSORS_BR 0x43E4 +# define R300_SCISSORS_OFFSET 1440 +# define R300_SCISSORS_X_SHIFT 0 +# define R300_SCISSORS_X_MASK (0x1FFF << 0) +# define R300_SCISSORS_Y_SHIFT 13 +# define R300_SCISSORS_Y_MASK (0x1FFF << 13) +/* END: Scissors and cliprects */ + +/* BEGIN: Texture specification */ + +/* + * The texture specification dwords are grouped by meaning and not by texture + * unit. This means that e.g. the offset for texture image unit N is found in + * register TX_OFFSET_0 + (4*N) + */ +#define R300_TX_FILTER_0 0x4400 +# define R300_TX_REPEAT 0 +# define R300_TX_MIRRORED 1 +# define R300_TX_CLAMP 4 +# define R300_TX_CLAMP_TO_EDGE 2 +# define R300_TX_CLAMP_TO_BORDER 6 +# define R300_TX_WRAP_S_SHIFT 0 +# define R300_TX_WRAP_S_MASK (7 << 0) +# define R300_TX_WRAP_T_SHIFT 3 +# define R300_TX_WRAP_T_MASK (7 << 3) +# define R300_TX_WRAP_Q_SHIFT 6 +# define R300_TX_WRAP_Q_MASK (7 << 6) +# define R300_TX_MAG_FILTER_NEAREST (1 << 9) +# define R300_TX_MAG_FILTER_LINEAR (2 << 9) +# define R300_TX_MAG_FILTER_MASK (3 << 9) +# define R300_TX_MIN_FILTER_NEAREST (1 << 11) +# define R300_TX_MIN_FILTER_LINEAR (2 << 11) +# define R300_TX_MIN_FILTER_NEAREST_MIP_NEAREST (5 << 11) +# define R300_TX_MIN_FILTER_NEAREST_MIP_LINEAR (9 << 11) +# define R300_TX_MIN_FILTER_LINEAR_MIP_NEAREST (6 << 11) +# define R300_TX_MIN_FILTER_LINEAR_MIP_LINEAR (10 << 11) + +/* NOTE: NEAREST doesn't seem to exist. + * Im not seting MAG_FILTER_MASK and (3 << 11) on for all + * anisotropy modes because that would void selected mag filter + */ +# define R300_TX_MIN_FILTER_ANISO_NEAREST (0 << 13) +# define R300_TX_MIN_FILTER_ANISO_LINEAR (0 << 13) +# define R300_TX_MIN_FILTER_ANISO_NEAREST_MIP_NEAREST (1 << 13) +# define R300_TX_MIN_FILTER_ANISO_NEAREST_MIP_LINEAR (2 << 13) +# define R300_TX_MIN_FILTER_MASK ( (15 << 11) | (3 << 13) ) +# define R300_TX_MAX_ANISO_1_TO_1 (0 << 21) +# define R300_TX_MAX_ANISO_2_TO_1 (2 << 21) +# define R300_TX_MAX_ANISO_4_TO_1 (4 << 21) +# define R300_TX_MAX_ANISO_8_TO_1 (6 << 21) +# define R300_TX_MAX_ANISO_16_TO_1 (8 << 21) +# define R300_TX_MAX_ANISO_MASK (14 << 21) + +#define R300_TX_FILTER1_0 0x4440 +# define R300_CHROMA_KEY_MODE_DISABLE 0 +# define R300_CHROMA_KEY_FORCE 1 +# define R300_CHROMA_KEY_BLEND 2 +# define R300_MC_ROUND_NORMAL (0<<2) +# define R300_MC_ROUND_MPEG4 (1<<2) +# define R300_LOD_BIAS_MASK 0x1fff +# define R300_EDGE_ANISO_EDGE_DIAG (0<<13) +# define R300_EDGE_ANISO_EDGE_ONLY (1<<13) +# define R300_MC_COORD_TRUNCATE_DISABLE (0<<14) +# define R300_MC_COORD_TRUNCATE_MPEG (1<<14) +# define R300_TX_TRI_PERF_0_8 (0<<15) +# define R300_TX_TRI_PERF_1_8 (1<<15) +# define R300_TX_TRI_PERF_1_4 (2<<15) +# define R300_TX_TRI_PERF_3_8 (3<<15) +# define R300_ANISO_THRESHOLD_MASK (7<<17) + +#define R300_TX_SIZE_0 0x4480 +# define R300_TX_WIDTHMASK_SHIFT 0 +# define R300_TX_WIDTHMASK_MASK (2047 << 0) +# define R300_TX_HEIGHTMASK_SHIFT 11 +# define R300_TX_HEIGHTMASK_MASK (2047 << 11) +# define R300_TX_UNK23 (1 << 23) +# define R300_TX_MAX_MIP_LEVEL_SHIFT 26 +# define R300_TX_MAX_MIP_LEVEL_MASK (0xf << 26) +# define R300_TX_SIZE_PROJECTED (1<<30) +# define R300_TX_SIZE_TXPITCH_EN (1<<31) +#define R300_TX_FORMAT_0 0x44C0 + /* The interpretation of the format word by Wladimir van der Laan */ + /* The X, Y, Z and W refer to the layout of the components. + They are given meanings as R, G, B and Alpha by the swizzle + specification */ +# define R300_TX_FORMAT_X8 0x0 +# define R300_TX_FORMAT_X16 0x1 +# define R300_TX_FORMAT_Y4X4 0x2 +# define R300_TX_FORMAT_Y8X8 0x3 +# define R300_TX_FORMAT_Y16X16 0x4 +# define R300_TX_FORMAT_Z3Y3X2 0x5 +# define R300_TX_FORMAT_Z5Y6X5 0x6 +# define R300_TX_FORMAT_Z6Y5X5 0x7 +# define R300_TX_FORMAT_Z11Y11X10 0x8 +# define R300_TX_FORMAT_Z10Y11X11 0x9 +# define R300_TX_FORMAT_W4Z4Y4X4 0xA +# define R300_TX_FORMAT_W1Z5Y5X5 0xB +# define R300_TX_FORMAT_W8Z8Y8X8 0xC +# define R300_TX_FORMAT_W2Z10Y10X10 0xD +# define R300_TX_FORMAT_W16Z16Y16X16 0xE +# define R300_TX_FORMAT_DXT1 0xF +# define R300_TX_FORMAT_DXT3 0x10 +# define R300_TX_FORMAT_DXT5 0x11 +# define R300_TX_FORMAT_D3DMFT_CxV8U8 0x12 /* no swizzle */ +# define R300_TX_FORMAT_A8R8G8B8 0x13 /* no swizzle */ +# define R300_TX_FORMAT_B8G8_B8G8 0x14 /* no swizzle */ +# define R300_TX_FORMAT_G8R8_G8B8 0x15 /* no swizzle */ + /* 0x16 - some 16 bit green format.. ?? */ +# define R300_TX_FORMAT_UNK25 (1 << 25) /* no swizzle */ +# define R300_TX_FORMAT_CUBIC_MAP (1 << 26) + + /* gap */ + /* Floating point formats */ + /* Note - hardware supports both 16 and 32 bit floating point */ +# define R300_TX_FORMAT_FL_I16 0x18 +# define R300_TX_FORMAT_FL_I16A16 0x19 +# define R300_TX_FORMAT_FL_R16G16B16A16 0x1A +# define R300_TX_FORMAT_FL_I32 0x1B +# define R300_TX_FORMAT_FL_I32A32 0x1C +# define R300_TX_FORMAT_FL_R32G32B32A32 0x1D +# define R300_TX_FORMAT_ATI2N 0x1F + /* alpha modes, convenience mostly */ + /* if you have alpha, pick constant appropriate to the + number of channels (1 for I8, 2 for I8A8, 4 for R8G8B8A8, etc */ +# define R300_TX_FORMAT_ALPHA_1CH 0x000 +# define R300_TX_FORMAT_ALPHA_2CH 0x200 +# define R300_TX_FORMAT_ALPHA_4CH 0x600 +# define R300_TX_FORMAT_ALPHA_NONE 0xA00 + /* Swizzling */ + /* constants */ +# define R300_TX_FORMAT_X 0 +# define R300_TX_FORMAT_Y 1 +# define R300_TX_FORMAT_Z 2 +# define R300_TX_FORMAT_W 3 +# define R300_TX_FORMAT_ZERO 4 +# define R300_TX_FORMAT_ONE 5 + /* 2.0*Z, everything above 1.0 is set to 0.0 */ +# define R300_TX_FORMAT_CUT_Z 6 + /* 2.0*W, everything above 1.0 is set to 0.0 */ +# define R300_TX_FORMAT_CUT_W 7 + +# define R300_TX_FORMAT_B_SHIFT 18 +# define R300_TX_FORMAT_G_SHIFT 15 +# define R300_TX_FORMAT_R_SHIFT 12 +# define R300_TX_FORMAT_A_SHIFT 9 + /* Convenience macro to take care of layout and swizzling */ +# define R300_EASY_TX_FORMAT(B, G, R, A, FMT) ( \ + ((R300_TX_FORMAT_##B)< 0.5, return ARG0, else return ARG1 + * - CMP: If ARG2 < 0, return ARG1, else return ARG0 + * - FLR: use FRC+MAD + * - XPD: use MAD+MAD + * - SGE, SLT: use MAD+CMP + * - RSQ: use ABS modifier for argument + * - Use OUTC_REPL_ALPHA to write results of an alpha-only operation + * (e.g. RCP) into color register + * - apparently, there's no quick DST operation + * - fglrx set FPI2_UNKNOWN_31 on a "MAD fragment.color, tmp0, tmp1, tmp2" + * - fglrx set FPI2_UNKNOWN_31 on a "MAX r2, r1, c0" + * - fglrx once set FPI0_UNKNOWN_31 on a "FRC r1, r1" + * + * Operand selection + * First stage selects three sources from the available registers and + * constant parameters. This is defined in INSTR1 (color) and INSTR3 (alpha). + * fglrx sorts the three source fields: Registers before constants, + * lower indices before higher indices; I do not know whether this is + * necessary. + * + * fglrx fills unused sources with "read constant 0" + * According to specs, you cannot select more than two different constants. + * + * Second stage selects the operands from the sources. This is defined in + * INSTR0 (color) and INSTR2 (alpha). You can also select the special constants + * zero and one. + * Swizzling and negation happens in this stage, as well. + * + * Important: Color and alpha seem to be mostly separate, i.e. their sources + * selection appears to be fully independent (the register storage is probably + * physically split into a color and an alpha section). + * However (because of the apparent physical split), there is some interaction + * WRT swizzling. If, for example, you want to load an R component into an + * Alpha operand, this R component is taken from a *color* source, not from + * an alpha source. The corresponding register doesn't even have to appear in + * the alpha sources list. (I hope this all makes sense to you) + * + * Destination selection + * The destination register index is in FPI1 (color) and FPI3 (alpha) + * together with enable bits. + * There are separate enable bits for writing into temporary registers + * (DSTC_REG_* /DSTA_REG) and and program output registers (DSTC_OUTPUT_* + * /DSTA_OUTPUT). You can write to both at once, or not write at all (the + * same index must be used for both). + * + * Note: There is a special form for LRP + * - Argument order is the same as in ARB_fragment_program. + * - Operation is MAD + * - ARG1 is set to ARGC_SRC1C_LRP/ARGC_SRC1A_LRP + * - Set FPI0/FPI2_SPECIAL_LRP + * Arbitrary LRP (including support for swizzling) requires vanilla MAD+MAD + */ +#define R300_PFS_INSTR1_0 0x46C0 +# define R300_FPI1_SRC0C_SHIFT 0 +# define R300_FPI1_SRC0C_MASK (31 << 0) +# define R300_FPI1_SRC0C_CONST (1 << 5) +# define R300_FPI1_SRC1C_SHIFT 6 +# define R300_FPI1_SRC1C_MASK (31 << 6) +# define R300_FPI1_SRC1C_CONST (1 << 11) +# define R300_FPI1_SRC2C_SHIFT 12 +# define R300_FPI1_SRC2C_MASK (31 << 12) +# define R300_FPI1_SRC2C_CONST (1 << 17) +# define R300_FPI1_SRC_MASK 0x0003ffff +# define R300_FPI1_DSTC_SHIFT 18 +# define R300_FPI1_DSTC_MASK (31 << 18) +# define R300_FPI1_DSTC_REG_MASK_SHIFT 23 +# define R300_FPI1_DSTC_REG_X (1 << 23) +# define R300_FPI1_DSTC_REG_Y (1 << 24) +# define R300_FPI1_DSTC_REG_Z (1 << 25) +# define R300_FPI1_DSTC_OUTPUT_MASK_SHIFT 26 +# define R300_FPI1_DSTC_OUTPUT_X (1 << 26) +# define R300_FPI1_DSTC_OUTPUT_Y (1 << 27) +# define R300_FPI1_DSTC_OUTPUT_Z (1 << 28) + +#define R300_PFS_INSTR3_0 0x47C0 +# define R300_FPI3_SRC0A_SHIFT 0 +# define R300_FPI3_SRC0A_MASK (31 << 0) +# define R300_FPI3_SRC0A_CONST (1 << 5) +# define R300_FPI3_SRC1A_SHIFT 6 +# define R300_FPI3_SRC1A_MASK (31 << 6) +# define R300_FPI3_SRC1A_CONST (1 << 11) +# define R300_FPI3_SRC2A_SHIFT 12 +# define R300_FPI3_SRC2A_MASK (31 << 12) +# define R300_FPI3_SRC2A_CONST (1 << 17) +# define R300_FPI3_SRC_MASK 0x0003ffff +# define R300_FPI3_DSTA_SHIFT 18 +# define R300_FPI3_DSTA_MASK (31 << 18) +# define R300_FPI3_DSTA_REG (1 << 23) +# define R300_FPI3_DSTA_OUTPUT (1 << 24) +# define R300_FPI3_DSTA_DEPTH (1 << 27) + +#define R300_PFS_INSTR0_0 0x48C0 +# define R300_FPI0_ARGC_SRC0C_XYZ 0 +# define R300_FPI0_ARGC_SRC0C_XXX 1 +# define R300_FPI0_ARGC_SRC0C_YYY 2 +# define R300_FPI0_ARGC_SRC0C_ZZZ 3 +# define R300_FPI0_ARGC_SRC1C_XYZ 4 +# define R300_FPI0_ARGC_SRC1C_XXX 5 +# define R300_FPI0_ARGC_SRC1C_YYY 6 +# define R300_FPI0_ARGC_SRC1C_ZZZ 7 +# define R300_FPI0_ARGC_SRC2C_XYZ 8 +# define R300_FPI0_ARGC_SRC2C_XXX 9 +# define R300_FPI0_ARGC_SRC2C_YYY 10 +# define R300_FPI0_ARGC_SRC2C_ZZZ 11 +# define R300_FPI0_ARGC_SRC0A 12 +# define R300_FPI0_ARGC_SRC1A 13 +# define R300_FPI0_ARGC_SRC2A 14 +# define R300_FPI0_ARGC_SRC1C_LRP 15 +# define R300_FPI0_ARGC_ZERO 20 +# define R300_FPI0_ARGC_ONE 21 + /* GUESS */ +# define R300_FPI0_ARGC_HALF 22 +# define R300_FPI0_ARGC_SRC0C_YZX 23 +# define R300_FPI0_ARGC_SRC1C_YZX 24 +# define R300_FPI0_ARGC_SRC2C_YZX 25 +# define R300_FPI0_ARGC_SRC0C_ZXY 26 +# define R300_FPI0_ARGC_SRC1C_ZXY 27 +# define R300_FPI0_ARGC_SRC2C_ZXY 28 +# define R300_FPI0_ARGC_SRC0CA_WZY 29 +# define R300_FPI0_ARGC_SRC1CA_WZY 30 +# define R300_FPI0_ARGC_SRC2CA_WZY 31 + +# define R300_FPI0_ARG0C_SHIFT 0 +# define R300_FPI0_ARG0C_MASK (31 << 0) +# define R300_FPI0_ARG0C_NEG (1 << 5) +# define R300_FPI0_ARG0C_ABS (1 << 6) +# define R300_FPI0_ARG1C_SHIFT 7 +# define R300_FPI0_ARG1C_MASK (31 << 7) +# define R300_FPI0_ARG1C_NEG (1 << 12) +# define R300_FPI0_ARG1C_ABS (1 << 13) +# define R300_FPI0_ARG2C_SHIFT 14 +# define R300_FPI0_ARG2C_MASK (31 << 14) +# define R300_FPI0_ARG2C_NEG (1 << 19) +# define R300_FPI0_ARG2C_ABS (1 << 20) +# define R300_FPI0_SPECIAL_LRP (1 << 21) +# define R300_FPI0_OUTC_MAD (0 << 23) +# define R300_FPI0_OUTC_DP3 (1 << 23) +# define R300_FPI0_OUTC_DP4 (2 << 23) +# define R300_FPI0_OUTC_MIN (4 << 23) +# define R300_FPI0_OUTC_MAX (5 << 23) +# define R300_FPI0_OUTC_CMPH (7 << 23) +# define R300_FPI0_OUTC_CMP (8 << 23) +# define R300_FPI0_OUTC_FRC (9 << 23) +# define R300_FPI0_OUTC_REPL_ALPHA (10 << 23) +# define R300_FPI0_OUTC_SAT (1 << 30) +# define R300_FPI0_INSERT_NOP (1 << 31) + +#define R300_PFS_INSTR2_0 0x49C0 +# define R300_FPI2_ARGA_SRC0C_X 0 +# define R300_FPI2_ARGA_SRC0C_Y 1 +# define R300_FPI2_ARGA_SRC0C_Z 2 +# define R300_FPI2_ARGA_SRC1C_X 3 +# define R300_FPI2_ARGA_SRC1C_Y 4 +# define R300_FPI2_ARGA_SRC1C_Z 5 +# define R300_FPI2_ARGA_SRC2C_X 6 +# define R300_FPI2_ARGA_SRC2C_Y 7 +# define R300_FPI2_ARGA_SRC2C_Z 8 +# define R300_FPI2_ARGA_SRC0A 9 +# define R300_FPI2_ARGA_SRC1A 10 +# define R300_FPI2_ARGA_SRC2A 11 +# define R300_FPI2_ARGA_SRC1A_LRP 15 +# define R300_FPI2_ARGA_ZERO 16 +# define R300_FPI2_ARGA_ONE 17 + /* GUESS */ +# define R300_FPI2_ARGA_HALF 18 +# define R300_FPI2_ARG0A_SHIFT 0 +# define R300_FPI2_ARG0A_MASK (31 << 0) +# define R300_FPI2_ARG0A_NEG (1 << 5) + /* GUESS */ +# define R300_FPI2_ARG0A_ABS (1 << 6) +# define R300_FPI2_ARG1A_SHIFT 7 +# define R300_FPI2_ARG1A_MASK (31 << 7) +# define R300_FPI2_ARG1A_NEG (1 << 12) + /* GUESS */ +# define R300_FPI2_ARG1A_ABS (1 << 13) +# define R300_FPI2_ARG2A_SHIFT 14 +# define R300_FPI2_ARG2A_MASK (31 << 14) +# define R300_FPI2_ARG2A_NEG (1 << 19) + /* GUESS */ +# define R300_FPI2_ARG2A_ABS (1 << 20) +# define R300_FPI2_SPECIAL_LRP (1 << 21) +# define R300_FPI2_OUTA_MAD (0 << 23) +# define R300_FPI2_OUTA_DP4 (1 << 23) +# define R300_FPI2_OUTA_MIN (2 << 23) +# define R300_FPI2_OUTA_MAX (3 << 23) +# define R300_FPI2_OUTA_CMP (6 << 23) +# define R300_FPI2_OUTA_FRC (7 << 23) +# define R300_FPI2_OUTA_EX2 (8 << 23) +# define R300_FPI2_OUTA_LG2 (9 << 23) +# define R300_FPI2_OUTA_RCP (10 << 23) +# define R300_FPI2_OUTA_RSQ (11 << 23) +# define R300_FPI2_OUTA_SAT (1 << 30) +# define R300_FPI2_UNKNOWN_31 (1 << 31) +/* END: Fragment program instruction set */ + +/* Fog state and color */ +#define R300_RE_FOG_STATE 0x4BC0 +# define R300_FOG_ENABLE (1 << 0) +# define R300_FOG_MODE_LINEAR (0 << 1) +# define R300_FOG_MODE_EXP (1 << 1) +# define R300_FOG_MODE_EXP2 (2 << 1) +# define R300_FOG_MODE_MASK (3 << 1) +#define R300_FOG_COLOR_R 0x4BC8 +#define R300_FOG_COLOR_G 0x4BCC +#define R300_FOG_COLOR_B 0x4BD0 + +#define R300_PP_ALPHA_TEST 0x4BD4 +# define R300_REF_ALPHA_MASK 0x000000ff +# define R300_ALPHA_TEST_FAIL (0 << 8) +# define R300_ALPHA_TEST_LESS (1 << 8) +# define R300_ALPHA_TEST_LEQUAL (3 << 8) +# define R300_ALPHA_TEST_EQUAL (2 << 8) +# define R300_ALPHA_TEST_GEQUAL (6 << 8) +# define R300_ALPHA_TEST_GREATER (4 << 8) +# define R300_ALPHA_TEST_NEQUAL (5 << 8) +# define R300_ALPHA_TEST_PASS (7 << 8) +# define R300_ALPHA_TEST_OP_MASK (7 << 8) +# define R300_ALPHA_TEST_ENABLE (1 << 11) + +/* gap */ + +/* Fragment program parameters in 7.16 floating point */ +#define R300_PFS_PARAM_0_X 0x4C00 +#define R300_PFS_PARAM_0_Y 0x4C04 +#define R300_PFS_PARAM_0_Z 0x4C08 +#define R300_PFS_PARAM_0_W 0x4C0C +/* GUESS: PARAM_31 is last, based on native limits reported by fglrx */ +#define R300_PFS_PARAM_31_X 0x4DF0 +#define R300_PFS_PARAM_31_Y 0x4DF4 +#define R300_PFS_PARAM_31_Z 0x4DF8 +#define R300_PFS_PARAM_31_W 0x4DFC + +/* Notes: + * - AFAIK fglrx always sets BLEND_UNKNOWN when blending is used in + * the application + * - AFAIK fglrx always sets BLEND_NO_SEPARATE when CBLEND and ABLEND + * are set to the same + * function (both registers are always set up completely in any case) + * - Most blend flags are simply copied from R200 and not tested yet + */ +#define R300_RB3D_CBLEND 0x4E04 +#define R300_RB3D_ABLEND 0x4E08 +/* the following only appear in CBLEND */ +# define R300_BLEND_ENABLE (1 << 0) +# define R300_BLEND_UNKNOWN (3 << 1) +# define R300_BLEND_NO_SEPARATE (1 << 3) +/* the following are shared between CBLEND and ABLEND */ +# define R300_FCN_MASK (3 << 12) +# define R300_COMB_FCN_ADD_CLAMP (0 << 12) +# define R300_COMB_FCN_ADD_NOCLAMP (1 << 12) +# define R300_COMB_FCN_SUB_CLAMP (2 << 12) +# define R300_COMB_FCN_SUB_NOCLAMP (3 << 12) +# define R300_COMB_FCN_MIN (4 << 12) +# define R300_COMB_FCN_MAX (5 << 12) +# define R300_COMB_FCN_RSUB_CLAMP (6 << 12) +# define R300_COMB_FCN_RSUB_NOCLAMP (7 << 12) +# define R300_BLEND_GL_ZERO (32) +# define R300_BLEND_GL_ONE (33) +# define R300_BLEND_GL_SRC_COLOR (34) +# define R300_BLEND_GL_ONE_MINUS_SRC_COLOR (35) +# define R300_BLEND_GL_DST_COLOR (36) +# define R300_BLEND_GL_ONE_MINUS_DST_COLOR (37) +# define R300_BLEND_GL_SRC_ALPHA (38) +# define R300_BLEND_GL_ONE_MINUS_SRC_ALPHA (39) +# define R300_BLEND_GL_DST_ALPHA (40) +# define R300_BLEND_GL_ONE_MINUS_DST_ALPHA (41) +# define R300_BLEND_GL_SRC_ALPHA_SATURATE (42) +# define R300_BLEND_GL_CONST_COLOR (43) +# define R300_BLEND_GL_ONE_MINUS_CONST_COLOR (44) +# define R300_BLEND_GL_CONST_ALPHA (45) +# define R300_BLEND_GL_ONE_MINUS_CONST_ALPHA (46) +# define R300_BLEND_MASK (63) +# define R300_SRC_BLEND_SHIFT (16) +# define R300_DST_BLEND_SHIFT (24) +#define R300_RB3D_BLEND_COLOR 0x4E10 +#define R300_RB3D_COLORMASK 0x4E0C +# define R300_COLORMASK0_B (1<<0) +# define R300_COLORMASK0_G (1<<1) +# define R300_COLORMASK0_R (1<<2) +# define R300_COLORMASK0_A (1<<3) + +/* gap */ + +#define R300_RB3D_COLOROFFSET0 0x4E28 +# define R300_COLOROFFSET_MASK 0xFFFFFFF0 /* GUESS */ +#define R300_RB3D_COLOROFFSET1 0x4E2C /* GUESS */ +#define R300_RB3D_COLOROFFSET2 0x4E30 /* GUESS */ +#define R300_RB3D_COLOROFFSET3 0x4E34 /* GUESS */ + +/* gap */ + +/* Bit 16: Larger tiles + * Bit 17: 4x2 tiles + * Bit 18: Extremely weird tile like, but some pixels duplicated? + */ +#define R300_RB3D_COLORPITCH0 0x4E38 +# define R300_COLORPITCH_MASK 0x00001FF8 /* GUESS */ +# define R300_COLOR_TILE_ENABLE (1 << 16) /* GUESS */ +# define R300_COLOR_MICROTILE_ENABLE (1 << 17) /* GUESS */ +# define R300_COLOR_MICROTILE_SQUARE_ENABLE (2 << 17) +# define R300_COLOR_ENDIAN_NO_SWAP (0 << 18) /* GUESS */ +# define R300_COLOR_ENDIAN_WORD_SWAP (1 << 18) /* GUESS */ +# define R300_COLOR_ENDIAN_DWORD_SWAP (2 << 18) /* GUESS */ +# define R300_COLOR_FORMAT_RGB565 (2 << 22) +# define R300_COLOR_FORMAT_ARGB8888 (3 << 22) +#define R300_RB3D_COLORPITCH1 0x4E3C /* GUESS */ +#define R300_RB3D_COLORPITCH2 0x4E40 /* GUESS */ +#define R300_RB3D_COLORPITCH3 0x4E44 /* GUESS */ + +#define R300_RB3D_AARESOLVE_OFFSET 0x4E80 +#define R300_RB3D_AARESOLVE_PITCH 0x4E84 +#define R300_RB3D_AARESOLVE_CTL 0x4E88 +/* gap */ + +/* Guess by Vladimir. + * Set to 0A before 3D operations, set to 02 afterwards. + */ +/*#define R300_RB3D_DSTCACHE_CTLSTAT 0x4E4C*/ +# define R300_RB3D_DSTCACHE_UNKNOWN_02 0x00000002 +# define R300_RB3D_DSTCACHE_UNKNOWN_0A 0x0000000A + +/* gap */ +/* There seems to be no "write only" setting, so use Z-test = ALWAYS + * for this. + * Bit (1<<8) is the "test" bit. so plain write is 6 - vd + */ +#define R300_ZB_CNTL 0x4F00 +# define R300_STENCIL_ENABLE (1 << 0) +# define R300_Z_ENABLE (1 << 1) +# define R300_Z_WRITE_ENABLE (1 << 2) +# define R300_Z_SIGNED_COMPARE (1 << 3) +# define R300_STENCIL_FRONT_BACK (1 << 4) + +#define R300_ZB_ZSTENCILCNTL 0x4f04 + /* functions */ +# define R300_ZS_NEVER 0 +# define R300_ZS_LESS 1 +# define R300_ZS_LEQUAL 2 +# define R300_ZS_EQUAL 3 +# define R300_ZS_GEQUAL 4 +# define R300_ZS_GREATER 5 +# define R300_ZS_NOTEQUAL 6 +# define R300_ZS_ALWAYS 7 +# define R300_ZS_MASK 7 + /* operations */ +# define R300_ZS_KEEP 0 +# define R300_ZS_ZERO 1 +# define R300_ZS_REPLACE 2 +# define R300_ZS_INCR 3 +# define R300_ZS_DECR 4 +# define R300_ZS_INVERT 5 +# define R300_ZS_INCR_WRAP 6 +# define R300_ZS_DECR_WRAP 7 +# define R300_Z_FUNC_SHIFT 0 + /* front and back refer to operations done for front + and back faces, i.e. separate stencil function support */ +# define R300_S_FRONT_FUNC_SHIFT 3 +# define R300_S_FRONT_SFAIL_OP_SHIFT 6 +# define R300_S_FRONT_ZPASS_OP_SHIFT 9 +# define R300_S_FRONT_ZFAIL_OP_SHIFT 12 +# define R300_S_BACK_FUNC_SHIFT 15 +# define R300_S_BACK_SFAIL_OP_SHIFT 18 +# define R300_S_BACK_ZPASS_OP_SHIFT 21 +# define R300_S_BACK_ZFAIL_OP_SHIFT 24 + +#define R300_ZB_STENCILREFMASK 0x4f08 +# define R300_STENCILREF_SHIFT 0 +# define R300_STENCILREF_MASK 0x000000ff +# define R300_STENCILMASK_SHIFT 8 +# define R300_STENCILMASK_MASK 0x0000ff00 +# define R300_STENCILWRITEMASK_SHIFT 16 +# define R300_STENCILWRITEMASK_MASK 0x00ff0000 + +/* gap */ + +#define R300_ZB_FORMAT 0x4f10 +# define R300_DEPTHFORMAT_16BIT_INT_Z (0 << 0) +# define R300_DEPTHFORMAT_16BIT_13E3 (1 << 0) +# define R300_DEPTHFORMAT_24BIT_INT_Z_8BIT_STENCIL (2 << 0) +/* reserved up to (15 << 0) */ +# define R300_INVERT_13E3_LEADING_ONES (0 << 4) +# define R300_INVERT_13E3_LEADING_ZEROS (1 << 4) + +#define R300_ZB_ZTOP 0x4F14 +# define R300_ZTOP_DISABLE (0 << 0) +# define R300_ZTOP_ENABLE (1 << 0) + +/* gap */ + +#define R300_ZB_ZCACHE_CTLSTAT 0x4f18 +# define R300_ZB_ZCACHE_CTLSTAT_ZC_FLUSH_NO_EFFECT (0 << 0) +# define R300_ZB_ZCACHE_CTLSTAT_ZC_FLUSH_FLUSH_AND_FREE (1 << 0) +# define R300_ZB_ZCACHE_CTLSTAT_ZC_FREE_NO_EFFECT (0 << 1) +# define R300_ZB_ZCACHE_CTLSTAT_ZC_FREE_FREE (1 << 1) +# define R300_ZB_ZCACHE_CTLSTAT_ZC_BUSY_IDLE (0 << 31) +# define R300_ZB_ZCACHE_CTLSTAT_ZC_BUSY_BUSY (1 << 31) + +#define R300_ZB_BW_CNTL 0x4f1c +# define R300_HIZ_DISABLE (0 << 0) +# define R300_HIZ_ENABLE (1 << 0) +# define R300_HIZ_MIN (0 << 1) +# define R300_HIZ_MAX (1 << 1) +# define R300_FAST_FILL_DISABLE (0 << 2) +# define R300_FAST_FILL_ENABLE (1 << 2) +# define R300_RD_COMP_DISABLE (0 << 3) +# define R300_RD_COMP_ENABLE (1 << 3) +# define R300_WR_COMP_DISABLE (0 << 4) +# define R300_WR_COMP_ENABLE (1 << 4) +# define R300_ZB_CB_CLEAR_RMW (0 << 5) +# define R300_ZB_CB_CLEAR_CACHE_LINEAR (1 << 5) +# define R300_FORCE_COMPRESSED_STENCIL_VALUE_DISABLE (0 << 6) +# define R300_FORCE_COMPRESSED_STENCIL_VALUE_ENABLE (1 << 6) + +# define R500_ZEQUAL_OPTIMIZE_ENABLE (0 << 7) +# define R500_ZEQUAL_OPTIMIZE_DISABLE (1 << 7) +# define R500_SEQUAL_OPTIMIZE_ENABLE (0 << 8) +# define R500_SEQUAL_OPTIMIZE_DISABLE (1 << 8) + +# define R500_BMASK_ENABLE (0 << 10) +# define R500_BMASK_DISABLE (1 << 10) +# define R500_HIZ_EQUAL_REJECT_DISABLE (0 << 11) +# define R500_HIZ_EQUAL_REJECT_ENABLE (1 << 11) +# define R500_HIZ_FP_EXP_BITS_DISABLE (0 << 12) +# define R500_HIZ_FP_EXP_BITS_1 (1 << 12) +# define R500_HIZ_FP_EXP_BITS_2 (2 << 12) +# define R500_HIZ_FP_EXP_BITS_3 (3 << 12) +# define R500_HIZ_FP_EXP_BITS_4 (4 << 12) +# define R500_HIZ_FP_EXP_BITS_5 (5 << 12) +# define R500_HIZ_FP_INVERT_LEADING_ONES (0 << 15) +# define R500_HIZ_FP_INVERT_LEADING_ZEROS (1 << 15) +# define R500_TILE_OVERWRITE_RECOMPRESSION_ENABLE (0 << 16) +# define R500_TILE_OVERWRITE_RECOMPRESSION_DISABLE (1 << 16) +# define R500_CONTIGUOUS_6XAA_SAMPLES_ENABLE (0 << 17) +# define R500_CONTIGUOUS_6XAA_SAMPLES_DISABLE (1 << 17) +# define R500_PEQ_PACKING_DISABLE (0 << 18) +# define R500_PEQ_PACKING_ENABLE (1 << 18) +# define R500_COVERED_PTR_MASKING_DISABLE (0 << 18) +# define R500_COVERED_PTR_MASKING_ENABLE (1 << 18) + + +/* gap */ + +/* Z Buffer Address Offset. + * Bits 31 to 5 are used for aligned Z buffer address offset for macro tiles. + */ +#define R300_ZB_DEPTHOFFSET 0x4f20 + +/* Z Buffer Pitch and Endian Control */ +#define R300_ZB_DEPTHPITCH 0x4f24 +# define R300_DEPTHPITCH_MASK 0x00003FFC +# define R300_DEPTHMACROTILE_DISABLE (0 << 16) +# define R300_DEPTHMACROTILE_ENABLE (1 << 16) +# define R300_DEPTHMICROTILE_LINEAR (0 << 17) +# define R300_DEPTHMICROTILE_TILED (1 << 17) +# define R300_DEPTHMICROTILE_TILED_SQUARE (2 << 17) +# define R300_DEPTHENDIAN_NO_SWAP (0 << 18) +# define R300_DEPTHENDIAN_WORD_SWAP (1 << 18) +# define R300_DEPTHENDIAN_DWORD_SWAP (2 << 18) +# define R300_DEPTHENDIAN_HALF_DWORD_SWAP (3 << 18) + +/* Z Buffer Clear Value */ +#define R300_ZB_DEPTHCLEARVALUE 0x4f28 + +#define R300_ZB_ZMASK_OFFSET 0x4f30 +#define R300_ZB_ZMASK_PITCH 0x4f34 +#define R300_ZB_ZMASK_WRINDEX 0x4f38 +#define R300_ZB_ZMASK_DWORD 0x4f3c +#define R300_ZB_ZMASK_RDINDEX 0x4f40 + +/* Hierarchical Z Memory Offset */ +#define R300_ZB_HIZ_OFFSET 0x4f44 + +/* Hierarchical Z Write Index */ +#define R300_ZB_HIZ_WRINDEX 0x4f48 + +/* Hierarchical Z Data */ +#define R300_ZB_HIZ_DWORD 0x4f4c + +/* Hierarchical Z Read Index */ +#define R300_ZB_HIZ_RDINDEX 0x4f50 + +/* Hierarchical Z Pitch */ +#define R300_ZB_HIZ_PITCH 0x4f54 + +/* Z Buffer Z Pass Counter Data */ +#define R300_ZB_ZPASS_DATA 0x4f58 + +/* Z Buffer Z Pass Counter Address */ +#define R300_ZB_ZPASS_ADDR 0x4f5c + +/* Depth buffer X and Y coordinate offset */ +#define R300_ZB_DEPTHXY_OFFSET 0x4f60 +# define R300_DEPTHX_OFFSET_SHIFT 1 +# define R300_DEPTHX_OFFSET_MASK 0x000007FE +# define R300_DEPTHY_OFFSET_SHIFT 17 +# define R300_DEPTHY_OFFSET_MASK 0x07FE0000 + +/* Sets the fifo sizes */ +#define R500_ZB_FIFO_SIZE 0x4fd0 +# define R500_OP_FIFO_SIZE_FULL (0 << 0) +# define R500_OP_FIFO_SIZE_HALF (1 << 0) +# define R500_OP_FIFO_SIZE_QUATER (2 << 0) +# define R500_OP_FIFO_SIZE_EIGTHS (4 << 0) + +/* Stencil Reference Value and Mask for backfacing quads */ +/* R300_ZB_STENCILREFMASK handles front face */ +#define R500_ZB_STENCILREFMASK_BF 0x4fd4 +# define R500_STENCILREF_SHIFT 0 +# define R500_STENCILREF_MASK 0x000000ff +# define R500_STENCILMASK_SHIFT 8 +# define R500_STENCILMASK_MASK 0x0000ff00 +# define R500_STENCILWRITEMASK_SHIFT 16 +# define R500_STENCILWRITEMASK_MASK 0x00ff0000 + +/* BEGIN: Vertex program instruction set */ + +/* Every instruction is four dwords long: + * DWORD 0: output and opcode + * DWORD 1: first argument + * DWORD 2: second argument + * DWORD 3: third argument + * + * Notes: + * - ABS r, a is implemented as MAX r, a, -a + * - MOV is implemented as ADD to zero + * - XPD is implemented as MUL + MAD + * - FLR is implemented as FRC + ADD + * - apparently, fglrx tries to schedule instructions so that there is at + * least one instruction between the write to a temporary and the first + * read from said temporary; however, violations of this scheduling are + * allowed + * - register indices seem to be unrelated with OpenGL aliasing to + * conventional state + * - only one attribute and one parameter can be loaded at a time; however, + * the same attribute/parameter can be used for more than one argument + * - the second software argument for POW is the third hardware argument + * (no idea why) + * - MAD with only temporaries as input seems to use VPI_OUT_SELECT_MAD_2 + * + * There is some magic surrounding LIT: + * The single argument is replicated across all three inputs, but swizzled: + * First argument: xyzy + * Second argument: xyzx + * Third argument: xyzw + * Whenever the result is used later in the fragment program, fglrx forces + * x and w to be 1.0 in the input selection; I don't know whether this is + * strictly necessary + */ +#define R300_VPI_OUT_OP_DOT (1 << 0) +#define R300_VPI_OUT_OP_MUL (2 << 0) +#define R300_VPI_OUT_OP_ADD (3 << 0) +#define R300_VPI_OUT_OP_MAD (4 << 0) +#define R300_VPI_OUT_OP_DST (5 << 0) +#define R300_VPI_OUT_OP_FRC (6 << 0) +#define R300_VPI_OUT_OP_MAX (7 << 0) +#define R300_VPI_OUT_OP_MIN (8 << 0) +#define R300_VPI_OUT_OP_SGE (9 << 0) +#define R300_VPI_OUT_OP_SLT (10 << 0) + /* Used in GL_POINT_DISTANCE_ATTENUATION_ARB, vector(scalar, vector) */ +#define R300_VPI_OUT_OP_UNK12 (12 << 0) +#define R300_VPI_OUT_OP_ARL (13 << 0) +#define R300_VPI_OUT_OP_EXP (65 << 0) +#define R300_VPI_OUT_OP_LOG (66 << 0) + /* Used in fog computations, scalar(scalar) */ +#define R300_VPI_OUT_OP_UNK67 (67 << 0) +#define R300_VPI_OUT_OP_LIT (68 << 0) +#define R300_VPI_OUT_OP_POW (69 << 0) +#define R300_VPI_OUT_OP_RCP (70 << 0) +#define R300_VPI_OUT_OP_RSQ (72 << 0) + /* Used in GL_POINT_DISTANCE_ATTENUATION_ARB, scalar(scalar) */ +#define R300_VPI_OUT_OP_UNK73 (73 << 0) +#define R300_VPI_OUT_OP_EX2 (75 << 0) +#define R300_VPI_OUT_OP_LG2 (76 << 0) +#define R300_VPI_OUT_OP_MAD_2 (128 << 0) + /* all temps, vector(scalar, vector, vector) */ +#define R300_VPI_OUT_OP_UNK129 (129 << 0) + +#define R300_VPI_OUT_REG_CLASS_TEMPORARY (0 << 8) +#define R300_VPI_OUT_REG_CLASS_ADDR (1 << 8) +#define R300_VPI_OUT_REG_CLASS_RESULT (2 << 8) +#define R300_VPI_OUT_REG_CLASS_MASK (31 << 8) + +#define R300_VPI_OUT_REG_INDEX_SHIFT 13 + /* GUESS based on fglrx native limits */ +#define R300_VPI_OUT_REG_INDEX_MASK (31 << 13) + +#define R300_VPI_OUT_WRITE_X (1 << 20) +#define R300_VPI_OUT_WRITE_Y (1 << 21) +#define R300_VPI_OUT_WRITE_Z (1 << 22) +#define R300_VPI_OUT_WRITE_W (1 << 23) + +#define R300_VPI_IN_REG_CLASS_TEMPORARY (0 << 0) +#define R300_VPI_IN_REG_CLASS_ATTRIBUTE (1 << 0) +#define R300_VPI_IN_REG_CLASS_PARAMETER (2 << 0) +#define R300_VPI_IN_REG_CLASS_NONE (9 << 0) +#define R300_VPI_IN_REG_CLASS_MASK (31 << 0) + +#define R300_VPI_IN_REG_INDEX_SHIFT 5 + /* GUESS based on fglrx native limits */ +#define R300_VPI_IN_REG_INDEX_MASK (255 << 5) + +/* The R300 can select components from the input register arbitrarily. + * Use the following constants, shifted by the component shift you + * want to select + */ +#define R300_VPI_IN_SELECT_X 0 +#define R300_VPI_IN_SELECT_Y 1 +#define R300_VPI_IN_SELECT_Z 2 +#define R300_VPI_IN_SELECT_W 3 +#define R300_VPI_IN_SELECT_ZERO 4 +#define R300_VPI_IN_SELECT_ONE 5 +#define R300_VPI_IN_SELECT_MASK 7 + +#define R300_VPI_IN_X_SHIFT 13 +#define R300_VPI_IN_Y_SHIFT 16 +#define R300_VPI_IN_Z_SHIFT 19 +#define R300_VPI_IN_W_SHIFT 22 + +#define R300_VPI_IN_NEG_X (1 << 25) +#define R300_VPI_IN_NEG_Y (1 << 26) +#define R300_VPI_IN_NEG_Z (1 << 27) +#define R300_VPI_IN_NEG_W (1 << 28) +/* END: Vertex program instruction set */ + +/* BEGIN: Packet 3 commands */ + +/* A primitive emission dword. */ +#define R300_PRIM_TYPE_NONE (0 << 0) +#define R300_PRIM_TYPE_POINT (1 << 0) +#define R300_PRIM_TYPE_LINE (2 << 0) +#define R300_PRIM_TYPE_LINE_STRIP (3 << 0) +#define R300_PRIM_TYPE_TRI_LIST (4 << 0) +#define R300_PRIM_TYPE_TRI_FAN (5 << 0) +#define R300_PRIM_TYPE_TRI_STRIP (6 << 0) +#define R300_PRIM_TYPE_TRI_TYPE2 (7 << 0) +#define R300_PRIM_TYPE_RECT_LIST (8 << 0) +#define R300_PRIM_TYPE_3VRT_POINT_LIST (9 << 0) +#define R300_PRIM_TYPE_3VRT_LINE_LIST (10 << 0) + /* GUESS (based on r200) */ +#define R300_PRIM_TYPE_POINT_SPRITES (11 << 0) +#define R300_PRIM_TYPE_LINE_LOOP (12 << 0) +#define R300_PRIM_TYPE_QUADS (13 << 0) +#define R300_PRIM_TYPE_QUAD_STRIP (14 << 0) +#define R300_PRIM_TYPE_POLYGON (15 << 0) +#define R300_PRIM_TYPE_MASK 0xF +#define R300_PRIM_WALK_IND (1 << 4) +#define R300_PRIM_WALK_LIST (2 << 4) +#define R300_PRIM_WALK_RING (3 << 4) +#define R300_PRIM_WALK_MASK (3 << 4) + /* GUESS (based on r200) */ +#define R300_PRIM_COLOR_ORDER_BGRA (0 << 6) +#define R300_PRIM_COLOR_ORDER_RGBA (1 << 6) +#define R300_PRIM_NUM_VERTICES_SHIFT 16 +#define R300_PRIM_NUM_VERTICES_MASK 0xffff + +/* Draw a primitive from vertex data in arrays loaded via 3D_LOAD_VBPNTR. + * Two parameter dwords: + * 0. The first parameter appears to be always 0 + * 1. The second parameter is a standard primitive emission dword. + */ +#define R300_PACKET3_3D_DRAW_VBUF 0x00002800 + +/* Specify the full set of vertex arrays as (address, stride). + * The first parameter is the number of vertex arrays specified. + * The rest of the command is a variable length list of blocks, where + * each block is three dwords long and specifies two arrays. + * The first dword of a block is split into two words, the lower significant + * word refers to the first array, the more significant word to the second + * array in the block. + * The low byte of each word contains the size of an array entry in dwords, + * the high byte contains the stride of the array. + * The second dword of a block contains the pointer to the first array, + * the third dword of a block contains the pointer to the second array. + * Note that if the total number of arrays is odd, the third dword of + * the last block is omitted. + */ +#define R300_PACKET3_3D_LOAD_VBPNTR 0x00002F00 + +#define R300_PACKET3_INDX_BUFFER 0x00003300 +# define R300_EB_UNK1_SHIFT 24 +# define R300_EB_UNK1 (0x80<<24) +# define R300_EB_UNK2 0x0810 +#define R300_PACKET3_3D_DRAW_VBUF_2 0x00003400 +#define R300_PACKET3_3D_DRAW_INDX_2 0x00003600 + +/* END: Packet 3 commands */ + + +/* Color formats for 2d packets + */ +#define R300_CP_COLOR_FORMAT_CI8 2 +#define R300_CP_COLOR_FORMAT_ARGB1555 3 +#define R300_CP_COLOR_FORMAT_RGB565 4 +#define R300_CP_COLOR_FORMAT_ARGB8888 6 +#define R300_CP_COLOR_FORMAT_RGB332 7 +#define R300_CP_COLOR_FORMAT_RGB8 9 +#define R300_CP_COLOR_FORMAT_ARGB4444 15 + +/* + * CP type-3 packets + */ +#define R300_CP_CMD_BITBLT_MULTI 0xC0009B00 + +#define R500_VAP_INDEX_OFFSET 0x208c + +#define R500_GA_US_VECTOR_INDEX 0x4250 +#define R500_GA_US_VECTOR_DATA 0x4254 + +#define R500_RS_IP_0 0x4074 +#define R500_RS_INST_0 0x4320 + +#define R500_US_CONFIG 0x4600 + +#define R500_US_FC_CTRL 0x4624 +#define R500_US_CODE_ADDR 0x4630 + +#define R500_RB3D_COLOR_CLEAR_VALUE_AR 0x46c0 +#define R500_RB3D_CONSTANT_COLOR_AR 0x4ef8 + +#define R300_SU_REG_DEST 0x42c8 +#define RV530_FG_ZBREG_DEST 0x4be8 +#define R300_ZB_ZPASS_DATA 0x4f58 +#define R300_ZB_ZPASS_ADDR 0x4f5c + +#endif /* _R300_REG_H */ diff --git a/drivers/gpu/drm/radeon/r300d.h b/drivers/gpu/drm/radeon/r300d.h new file mode 100644 index 00000000..1f519a5f --- /dev/null +++ b/drivers/gpu/drm/radeon/r300d.h @@ -0,0 +1,354 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#ifndef __R300D_H__ +#define __R300D_H__ + +#define CP_PACKET0 0x00000000 +#define PACKET0_BASE_INDEX_SHIFT 0 +#define PACKET0_BASE_INDEX_MASK (0x1ffff << 0) +#define PACKET0_COUNT_SHIFT 16 +#define PACKET0_COUNT_MASK (0x3fff << 16) +#define CP_PACKET1 0x40000000 +#define CP_PACKET2 0x80000000 +#define PACKET2_PAD_SHIFT 0 +#define PACKET2_PAD_MASK (0x3fffffff << 0) +#define CP_PACKET3 0xC0000000 +#define PACKET3_IT_OPCODE_SHIFT 8 +#define PACKET3_IT_OPCODE_MASK (0xff << 8) +#define PACKET3_COUNT_SHIFT 16 +#define PACKET3_COUNT_MASK (0x3fff << 16) +/* PACKET3 op code */ +#define PACKET3_NOP 0x10 +#define PACKET3_3D_DRAW_VBUF 0x28 +#define PACKET3_3D_DRAW_IMMD 0x29 +#define PACKET3_3D_DRAW_INDX 0x2A +#define PACKET3_3D_LOAD_VBPNTR 0x2F +#define PACKET3_3D_CLEAR_ZMASK 0x32 +#define PACKET3_INDX_BUFFER 0x33 +#define PACKET3_3D_DRAW_VBUF_2 0x34 +#define PACKET3_3D_DRAW_IMMD_2 0x35 +#define PACKET3_3D_DRAW_INDX_2 0x36 +#define PACKET3_3D_CLEAR_HIZ 0x37 +#define PACKET3_3D_CLEAR_CMASK 0x38 +#define PACKET3_BITBLT_MULTI 0x9B + +#define PACKET0(reg, n) (CP_PACKET0 | \ + REG_SET(PACKET0_BASE_INDEX, (reg) >> 2) | \ + REG_SET(PACKET0_COUNT, (n))) +#define PACKET2(v) (CP_PACKET2 | REG_SET(PACKET2_PAD, (v))) +#define PACKET3(op, n) (CP_PACKET3 | \ + REG_SET(PACKET3_IT_OPCODE, (op)) | \ + REG_SET(PACKET3_COUNT, (n))) + +#define PACKET_TYPE0 0 +#define PACKET_TYPE1 1 +#define PACKET_TYPE2 2 +#define PACKET_TYPE3 3 + +#define CP_PACKET_GET_TYPE(h) (((h) >> 30) & 3) +#define CP_PACKET_GET_COUNT(h) (((h) >> 16) & 0x3FFF) +#define CP_PACKET0_GET_REG(h) (((h) & 0x1FFF) << 2) +#define CP_PACKET0_GET_ONE_REG_WR(h) (((h) >> 15) & 1) +#define CP_PACKET3_GET_OPCODE(h) (((h) >> 8) & 0xFF) + +/* Registers */ +#define R_000148_MC_FB_LOCATION 0x000148 +#define S_000148_MC_FB_START(x) (((x) & 0xFFFF) << 0) +#define G_000148_MC_FB_START(x) (((x) >> 0) & 0xFFFF) +#define C_000148_MC_FB_START 0xFFFF0000 +#define S_000148_MC_FB_TOP(x) (((x) & 0xFFFF) << 16) +#define G_000148_MC_FB_TOP(x) (((x) >> 16) & 0xFFFF) +#define C_000148_MC_FB_TOP 0x0000FFFF +#define R_00014C_MC_AGP_LOCATION 0x00014C +#define S_00014C_MC_AGP_START(x) (((x) & 0xFFFF) << 0) +#define G_00014C_MC_AGP_START(x) (((x) >> 0) & 0xFFFF) +#define C_00014C_MC_AGP_START 0xFFFF0000 +#define S_00014C_MC_AGP_TOP(x) (((x) & 0xFFFF) << 16) +#define G_00014C_MC_AGP_TOP(x) (((x) >> 16) & 0xFFFF) +#define C_00014C_MC_AGP_TOP 0x0000FFFF +#define R_00015C_AGP_BASE_2 0x00015C +#define S_00015C_AGP_BASE_ADDR_2(x) (((x) & 0xF) << 0) +#define G_00015C_AGP_BASE_ADDR_2(x) (((x) >> 0) & 0xF) +#define C_00015C_AGP_BASE_ADDR_2 0xFFFFFFF0 +#define R_000170_AGP_BASE 0x000170 +#define S_000170_AGP_BASE_ADDR(x) (((x) & 0xFFFFFFFF) << 0) +#define G_000170_AGP_BASE_ADDR(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_000170_AGP_BASE_ADDR 0x00000000 +#define R_0007C0_CP_STAT 0x0007C0 +#define S_0007C0_MRU_BUSY(x) (((x) & 0x1) << 0) +#define G_0007C0_MRU_BUSY(x) (((x) >> 0) & 0x1) +#define C_0007C0_MRU_BUSY 0xFFFFFFFE +#define S_0007C0_MWU_BUSY(x) (((x) & 0x1) << 1) +#define G_0007C0_MWU_BUSY(x) (((x) >> 1) & 0x1) +#define C_0007C0_MWU_BUSY 0xFFFFFFFD +#define S_0007C0_RSIU_BUSY(x) (((x) & 0x1) << 2) +#define G_0007C0_RSIU_BUSY(x) (((x) >> 2) & 0x1) +#define C_0007C0_RSIU_BUSY 0xFFFFFFFB +#define S_0007C0_RCIU_BUSY(x) (((x) & 0x1) << 3) +#define G_0007C0_RCIU_BUSY(x) (((x) >> 3) & 0x1) +#define C_0007C0_RCIU_BUSY 0xFFFFFFF7 +#define S_0007C0_CSF_PRIMARY_BUSY(x) (((x) & 0x1) << 9) +#define G_0007C0_CSF_PRIMARY_BUSY(x) (((x) >> 9) & 0x1) +#define C_0007C0_CSF_PRIMARY_BUSY 0xFFFFFDFF +#define S_0007C0_CSF_INDIRECT_BUSY(x) (((x) & 0x1) << 10) +#define G_0007C0_CSF_INDIRECT_BUSY(x) (((x) >> 10) & 0x1) +#define C_0007C0_CSF_INDIRECT_BUSY 0xFFFFFBFF +#define S_0007C0_CSQ_PRIMARY_BUSY(x) (((x) & 0x1) << 11) +#define G_0007C0_CSQ_PRIMARY_BUSY(x) (((x) >> 11) & 0x1) +#define C_0007C0_CSQ_PRIMARY_BUSY 0xFFFFF7FF +#define S_0007C0_CSQ_INDIRECT_BUSY(x) (((x) & 0x1) << 12) +#define G_0007C0_CSQ_INDIRECT_BUSY(x) (((x) >> 12) & 0x1) +#define C_0007C0_CSQ_INDIRECT_BUSY 0xFFFFEFFF +#define S_0007C0_CSI_BUSY(x) (((x) & 0x1) << 13) +#define G_0007C0_CSI_BUSY(x) (((x) >> 13) & 0x1) +#define C_0007C0_CSI_BUSY 0xFFFFDFFF +#define S_0007C0_CSF_INDIRECT2_BUSY(x) (((x) & 0x1) << 14) +#define G_0007C0_CSF_INDIRECT2_BUSY(x) (((x) >> 14) & 0x1) +#define C_0007C0_CSF_INDIRECT2_BUSY 0xFFFFBFFF +#define S_0007C0_CSQ_INDIRECT2_BUSY(x) (((x) & 0x1) << 15) +#define G_0007C0_CSQ_INDIRECT2_BUSY(x) (((x) >> 15) & 0x1) +#define C_0007C0_CSQ_INDIRECT2_BUSY 0xFFFF7FFF +#define S_0007C0_GUIDMA_BUSY(x) (((x) & 0x1) << 28) +#define G_0007C0_GUIDMA_BUSY(x) (((x) >> 28) & 0x1) +#define C_0007C0_GUIDMA_BUSY 0xEFFFFFFF +#define S_0007C0_VIDDMA_BUSY(x) (((x) & 0x1) << 29) +#define G_0007C0_VIDDMA_BUSY(x) (((x) >> 29) & 0x1) +#define C_0007C0_VIDDMA_BUSY 0xDFFFFFFF +#define S_0007C0_CMDSTRM_BUSY(x) (((x) & 0x1) << 30) +#define G_0007C0_CMDSTRM_BUSY(x) (((x) >> 30) & 0x1) +#define C_0007C0_CMDSTRM_BUSY 0xBFFFFFFF +#define S_0007C0_CP_BUSY(x) (((x) & 0x1) << 31) +#define G_0007C0_CP_BUSY(x) (((x) >> 31) & 0x1) +#define C_0007C0_CP_BUSY 0x7FFFFFFF +#define R_000E40_RBBM_STATUS 0x000E40 +#define S_000E40_CMDFIFO_AVAIL(x) (((x) & 0x7F) << 0) +#define G_000E40_CMDFIFO_AVAIL(x) (((x) >> 0) & 0x7F) +#define C_000E40_CMDFIFO_AVAIL 0xFFFFFF80 +#define S_000E40_HIRQ_ON_RBB(x) (((x) & 0x1) << 8) +#define G_000E40_HIRQ_ON_RBB(x) (((x) >> 8) & 0x1) +#define C_000E40_HIRQ_ON_RBB 0xFFFFFEFF +#define S_000E40_CPRQ_ON_RBB(x) (((x) & 0x1) << 9) +#define G_000E40_CPRQ_ON_RBB(x) (((x) >> 9) & 0x1) +#define C_000E40_CPRQ_ON_RBB 0xFFFFFDFF +#define S_000E40_CFRQ_ON_RBB(x) (((x) & 0x1) << 10) +#define G_000E40_CFRQ_ON_RBB(x) (((x) >> 10) & 0x1) +#define C_000E40_CFRQ_ON_RBB 0xFFFFFBFF +#define S_000E40_HIRQ_IN_RTBUF(x) (((x) & 0x1) << 11) +#define G_000E40_HIRQ_IN_RTBUF(x) (((x) >> 11) & 0x1) +#define C_000E40_HIRQ_IN_RTBUF 0xFFFFF7FF +#define S_000E40_CPRQ_IN_RTBUF(x) (((x) & 0x1) << 12) +#define G_000E40_CPRQ_IN_RTBUF(x) (((x) >> 12) & 0x1) +#define C_000E40_CPRQ_IN_RTBUF 0xFFFFEFFF +#define S_000E40_CFRQ_IN_RTBUF(x) (((x) & 0x1) << 13) +#define G_000E40_CFRQ_IN_RTBUF(x) (((x) >> 13) & 0x1) +#define C_000E40_CFRQ_IN_RTBUF 0xFFFFDFFF +#define S_000E40_CF_PIPE_BUSY(x) (((x) & 0x1) << 14) +#define G_000E40_CF_PIPE_BUSY(x) (((x) >> 14) & 0x1) +#define C_000E40_CF_PIPE_BUSY 0xFFFFBFFF +#define S_000E40_ENG_EV_BUSY(x) (((x) & 0x1) << 15) +#define G_000E40_ENG_EV_BUSY(x) (((x) >> 15) & 0x1) +#define C_000E40_ENG_EV_BUSY 0xFFFF7FFF +#define S_000E40_CP_CMDSTRM_BUSY(x) (((x) & 0x1) << 16) +#define G_000E40_CP_CMDSTRM_BUSY(x) (((x) >> 16) & 0x1) +#define C_000E40_CP_CMDSTRM_BUSY 0xFFFEFFFF +#define S_000E40_E2_BUSY(x) (((x) & 0x1) << 17) +#define G_000E40_E2_BUSY(x) (((x) >> 17) & 0x1) +#define C_000E40_E2_BUSY 0xFFFDFFFF +#define S_000E40_RB2D_BUSY(x) (((x) & 0x1) << 18) +#define G_000E40_RB2D_BUSY(x) (((x) >> 18) & 0x1) +#define C_000E40_RB2D_BUSY 0xFFFBFFFF +#define S_000E40_RB3D_BUSY(x) (((x) & 0x1) << 19) +#define G_000E40_RB3D_BUSY(x) (((x) >> 19) & 0x1) +#define C_000E40_RB3D_BUSY 0xFFF7FFFF +#define S_000E40_VAP_BUSY(x) (((x) & 0x1) << 20) +#define G_000E40_VAP_BUSY(x) (((x) >> 20) & 0x1) +#define C_000E40_VAP_BUSY 0xFFEFFFFF +#define S_000E40_RE_BUSY(x) (((x) & 0x1) << 21) +#define G_000E40_RE_BUSY(x) (((x) >> 21) & 0x1) +#define C_000E40_RE_BUSY 0xFFDFFFFF +#define S_000E40_TAM_BUSY(x) (((x) & 0x1) << 22) +#define G_000E40_TAM_BUSY(x) (((x) >> 22) & 0x1) +#define C_000E40_TAM_BUSY 0xFFBFFFFF +#define S_000E40_TDM_BUSY(x) (((x) & 0x1) << 23) +#define G_000E40_TDM_BUSY(x) (((x) >> 23) & 0x1) +#define C_000E40_TDM_BUSY 0xFF7FFFFF +#define S_000E40_PB_BUSY(x) (((x) & 0x1) << 24) +#define G_000E40_PB_BUSY(x) (((x) >> 24) & 0x1) +#define C_000E40_PB_BUSY 0xFEFFFFFF +#define S_000E40_TIM_BUSY(x) (((x) & 0x1) << 25) +#define G_000E40_TIM_BUSY(x) (((x) >> 25) & 0x1) +#define C_000E40_TIM_BUSY 0xFDFFFFFF +#define S_000E40_GA_BUSY(x) (((x) & 0x1) << 26) +#define G_000E40_GA_BUSY(x) (((x) >> 26) & 0x1) +#define C_000E40_GA_BUSY 0xFBFFFFFF +#define S_000E40_CBA2D_BUSY(x) (((x) & 0x1) << 27) +#define G_000E40_CBA2D_BUSY(x) (((x) >> 27) & 0x1) +#define C_000E40_CBA2D_BUSY 0xF7FFFFFF +#define S_000E40_GUI_ACTIVE(x) (((x) & 0x1) << 31) +#define G_000E40_GUI_ACTIVE(x) (((x) >> 31) & 0x1) +#define C_000E40_GUI_ACTIVE 0x7FFFFFFF +#define R_0000F0_RBBM_SOFT_RESET 0x0000F0 +#define S_0000F0_SOFT_RESET_CP(x) (((x) & 0x1) << 0) +#define G_0000F0_SOFT_RESET_CP(x) (((x) >> 0) & 0x1) +#define C_0000F0_SOFT_RESET_CP 0xFFFFFFFE +#define S_0000F0_SOFT_RESET_HI(x) (((x) & 0x1) << 1) +#define G_0000F0_SOFT_RESET_HI(x) (((x) >> 1) & 0x1) +#define C_0000F0_SOFT_RESET_HI 0xFFFFFFFD +#define S_0000F0_SOFT_RESET_VAP(x) (((x) & 0x1) << 2) +#define G_0000F0_SOFT_RESET_VAP(x) (((x) >> 2) & 0x1) +#define C_0000F0_SOFT_RESET_VAP 0xFFFFFFFB +#define S_0000F0_SOFT_RESET_RE(x) (((x) & 0x1) << 3) +#define G_0000F0_SOFT_RESET_RE(x) (((x) >> 3) & 0x1) +#define C_0000F0_SOFT_RESET_RE 0xFFFFFFF7 +#define S_0000F0_SOFT_RESET_PP(x) (((x) & 0x1) << 4) +#define G_0000F0_SOFT_RESET_PP(x) (((x) >> 4) & 0x1) +#define C_0000F0_SOFT_RESET_PP 0xFFFFFFEF +#define S_0000F0_SOFT_RESET_E2(x) (((x) & 0x1) << 5) +#define G_0000F0_SOFT_RESET_E2(x) (((x) >> 5) & 0x1) +#define C_0000F0_SOFT_RESET_E2 0xFFFFFFDF +#define S_0000F0_SOFT_RESET_RB(x) (((x) & 0x1) << 6) +#define G_0000F0_SOFT_RESET_RB(x) (((x) >> 6) & 0x1) +#define C_0000F0_SOFT_RESET_RB 0xFFFFFFBF +#define S_0000F0_SOFT_RESET_HDP(x) (((x) & 0x1) << 7) +#define G_0000F0_SOFT_RESET_HDP(x) (((x) >> 7) & 0x1) +#define C_0000F0_SOFT_RESET_HDP 0xFFFFFF7F +#define S_0000F0_SOFT_RESET_MC(x) (((x) & 0x1) << 8) +#define G_0000F0_SOFT_RESET_MC(x) (((x) >> 8) & 0x1) +#define C_0000F0_SOFT_RESET_MC 0xFFFFFEFF +#define S_0000F0_SOFT_RESET_AIC(x) (((x) & 0x1) << 9) +#define G_0000F0_SOFT_RESET_AIC(x) (((x) >> 9) & 0x1) +#define C_0000F0_SOFT_RESET_AIC 0xFFFFFDFF +#define S_0000F0_SOFT_RESET_VIP(x) (((x) & 0x1) << 10) +#define G_0000F0_SOFT_RESET_VIP(x) (((x) >> 10) & 0x1) +#define C_0000F0_SOFT_RESET_VIP 0xFFFFFBFF +#define S_0000F0_SOFT_RESET_DISP(x) (((x) & 0x1) << 11) +#define G_0000F0_SOFT_RESET_DISP(x) (((x) >> 11) & 0x1) +#define C_0000F0_SOFT_RESET_DISP 0xFFFFF7FF +#define S_0000F0_SOFT_RESET_CG(x) (((x) & 0x1) << 12) +#define G_0000F0_SOFT_RESET_CG(x) (((x) >> 12) & 0x1) +#define C_0000F0_SOFT_RESET_CG 0xFFFFEFFF +#define S_0000F0_SOFT_RESET_GA(x) (((x) & 0x1) << 13) +#define G_0000F0_SOFT_RESET_GA(x) (((x) >> 13) & 0x1) +#define C_0000F0_SOFT_RESET_GA 0xFFFFDFFF +#define S_0000F0_SOFT_RESET_IDCT(x) (((x) & 0x1) << 14) +#define G_0000F0_SOFT_RESET_IDCT(x) (((x) >> 14) & 0x1) +#define C_0000F0_SOFT_RESET_IDCT 0xFFFFBFFF + +#define R_00000D_SCLK_CNTL 0x00000D +#define S_00000D_SCLK_SRC_SEL(x) (((x) & 0x7) << 0) +#define G_00000D_SCLK_SRC_SEL(x) (((x) >> 0) & 0x7) +#define C_00000D_SCLK_SRC_SEL 0xFFFFFFF8 +#define S_00000D_CP_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 3) +#define G_00000D_CP_MAX_DYN_STOP_LAT(x) (((x) >> 3) & 0x1) +#define C_00000D_CP_MAX_DYN_STOP_LAT 0xFFFFFFF7 +#define S_00000D_HDP_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 4) +#define G_00000D_HDP_MAX_DYN_STOP_LAT(x) (((x) >> 4) & 0x1) +#define C_00000D_HDP_MAX_DYN_STOP_LAT 0xFFFFFFEF +#define S_00000D_TV_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 5) +#define G_00000D_TV_MAX_DYN_STOP_LAT(x) (((x) >> 5) & 0x1) +#define C_00000D_TV_MAX_DYN_STOP_LAT 0xFFFFFFDF +#define S_00000D_E2_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 6) +#define G_00000D_E2_MAX_DYN_STOP_LAT(x) (((x) >> 6) & 0x1) +#define C_00000D_E2_MAX_DYN_STOP_LAT 0xFFFFFFBF +#define S_00000D_SE_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 7) +#define G_00000D_SE_MAX_DYN_STOP_LAT(x) (((x) >> 7) & 0x1) +#define C_00000D_SE_MAX_DYN_STOP_LAT 0xFFFFFF7F +#define S_00000D_IDCT_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 8) +#define G_00000D_IDCT_MAX_DYN_STOP_LAT(x) (((x) >> 8) & 0x1) +#define C_00000D_IDCT_MAX_DYN_STOP_LAT 0xFFFFFEFF +#define S_00000D_VIP_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 9) +#define G_00000D_VIP_MAX_DYN_STOP_LAT(x) (((x) >> 9) & 0x1) +#define C_00000D_VIP_MAX_DYN_STOP_LAT 0xFFFFFDFF +#define S_00000D_RE_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 10) +#define G_00000D_RE_MAX_DYN_STOP_LAT(x) (((x) >> 10) & 0x1) +#define C_00000D_RE_MAX_DYN_STOP_LAT 0xFFFFFBFF +#define S_00000D_PB_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 11) +#define G_00000D_PB_MAX_DYN_STOP_LAT(x) (((x) >> 11) & 0x1) +#define C_00000D_PB_MAX_DYN_STOP_LAT 0xFFFFF7FF +#define S_00000D_TAM_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 12) +#define G_00000D_TAM_MAX_DYN_STOP_LAT(x) (((x) >> 12) & 0x1) +#define C_00000D_TAM_MAX_DYN_STOP_LAT 0xFFFFEFFF +#define S_00000D_TDM_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 13) +#define G_00000D_TDM_MAX_DYN_STOP_LAT(x) (((x) >> 13) & 0x1) +#define C_00000D_TDM_MAX_DYN_STOP_LAT 0xFFFFDFFF +#define S_00000D_RB_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 14) +#define G_00000D_RB_MAX_DYN_STOP_LAT(x) (((x) >> 14) & 0x1) +#define C_00000D_RB_MAX_DYN_STOP_LAT 0xFFFFBFFF +#define S_00000D_FORCE_DISP2(x) (((x) & 0x1) << 15) +#define G_00000D_FORCE_DISP2(x) (((x) >> 15) & 0x1) +#define C_00000D_FORCE_DISP2 0xFFFF7FFF +#define S_00000D_FORCE_CP(x) (((x) & 0x1) << 16) +#define G_00000D_FORCE_CP(x) (((x) >> 16) & 0x1) +#define C_00000D_FORCE_CP 0xFFFEFFFF +#define S_00000D_FORCE_HDP(x) (((x) & 0x1) << 17) +#define G_00000D_FORCE_HDP(x) (((x) >> 17) & 0x1) +#define C_00000D_FORCE_HDP 0xFFFDFFFF +#define S_00000D_FORCE_DISP1(x) (((x) & 0x1) << 18) +#define G_00000D_FORCE_DISP1(x) (((x) >> 18) & 0x1) +#define C_00000D_FORCE_DISP1 0xFFFBFFFF +#define S_00000D_FORCE_TOP(x) (((x) & 0x1) << 19) +#define G_00000D_FORCE_TOP(x) (((x) >> 19) & 0x1) +#define C_00000D_FORCE_TOP 0xFFF7FFFF +#define S_00000D_FORCE_E2(x) (((x) & 0x1) << 20) +#define G_00000D_FORCE_E2(x) (((x) >> 20) & 0x1) +#define C_00000D_FORCE_E2 0xFFEFFFFF +#define S_00000D_FORCE_SE(x) (((x) & 0x1) << 21) +#define G_00000D_FORCE_SE(x) (((x) >> 21) & 0x1) +#define C_00000D_FORCE_SE 0xFFDFFFFF +#define S_00000D_FORCE_IDCT(x) (((x) & 0x1) << 22) +#define G_00000D_FORCE_IDCT(x) (((x) >> 22) & 0x1) +#define C_00000D_FORCE_IDCT 0xFFBFFFFF +#define S_00000D_FORCE_VIP(x) (((x) & 0x1) << 23) +#define G_00000D_FORCE_VIP(x) (((x) >> 23) & 0x1) +#define C_00000D_FORCE_VIP 0xFF7FFFFF +#define S_00000D_FORCE_RE(x) (((x) & 0x1) << 24) +#define G_00000D_FORCE_RE(x) (((x) >> 24) & 0x1) +#define C_00000D_FORCE_RE 0xFEFFFFFF +#define S_00000D_FORCE_PB(x) (((x) & 0x1) << 25) +#define G_00000D_FORCE_PB(x) (((x) >> 25) & 0x1) +#define C_00000D_FORCE_PB 0xFDFFFFFF +#define S_00000D_FORCE_TAM(x) (((x) & 0x1) << 26) +#define G_00000D_FORCE_TAM(x) (((x) >> 26) & 0x1) +#define C_00000D_FORCE_TAM 0xFBFFFFFF +#define S_00000D_FORCE_TDM(x) (((x) & 0x1) << 27) +#define G_00000D_FORCE_TDM(x) (((x) >> 27) & 0x1) +#define C_00000D_FORCE_TDM 0xF7FFFFFF +#define S_00000D_FORCE_RB(x) (((x) & 0x1) << 28) +#define G_00000D_FORCE_RB(x) (((x) >> 28) & 0x1) +#define C_00000D_FORCE_RB 0xEFFFFFFF +#define S_00000D_FORCE_TV_SCLK(x) (((x) & 0x1) << 29) +#define G_00000D_FORCE_TV_SCLK(x) (((x) >> 29) & 0x1) +#define C_00000D_FORCE_TV_SCLK 0xDFFFFFFF +#define S_00000D_FORCE_SUBPIC(x) (((x) & 0x1) << 30) +#define G_00000D_FORCE_SUBPIC(x) (((x) >> 30) & 0x1) +#define C_00000D_FORCE_SUBPIC 0xBFFFFFFF +#define S_00000D_FORCE_OV0(x) (((x) & 0x1) << 31) +#define G_00000D_FORCE_OV0(x) (((x) >> 31) & 0x1) +#define C_00000D_FORCE_OV0 0x7FFFFFFF + +#endif diff --git a/drivers/gpu/drm/radeon/r420.c b/drivers/gpu/drm/radeon/r420.c new file mode 100644 index 00000000..417fab81 --- /dev/null +++ b/drivers/gpu/drm/radeon/r420.c @@ -0,0 +1,468 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#include +#include +#include "drmP.h" +#include "radeon_reg.h" +#include "radeon.h" +#include "radeon_asic.h" +#include "atom.h" +#include "r100d.h" +#include "r420d.h" +#include "r420_reg_safe.h" + +void r420_pm_init_profile(struct radeon_device *rdev) +{ + /* default */ + rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_cm_idx = 0; + /* low sh */ + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_ps_idx = 0; + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_ps_idx = 0; + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_cm_idx = 0; + /* mid sh */ + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_ps_idx = 0; + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_ps_idx = 1; + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_cm_idx = 0; + /* high sh */ + rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_ps_idx = 0; + rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_cm_idx = 0; + /* low mh */ + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_ps_idx = 0; + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_cm_idx = 0; + /* mid mh */ + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_ps_idx = 0; + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_cm_idx = 0; + /* high mh */ + rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_ps_idx = 0; + rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_cm_idx = 0; +} + +static void r420_set_reg_safe(struct radeon_device *rdev) +{ + rdev->config.r300.reg_safe_bm = r420_reg_safe_bm; + rdev->config.r300.reg_safe_bm_size = ARRAY_SIZE(r420_reg_safe_bm); +} + +void r420_pipes_init(struct radeon_device *rdev) +{ + unsigned tmp; + unsigned gb_pipe_select; + unsigned num_pipes; + + /* GA_ENHANCE workaround TCL deadlock issue */ + WREG32(R300_GA_ENHANCE, R300_GA_DEADLOCK_CNTL | R300_GA_FASTSYNC_CNTL | + (1 << 2) | (1 << 3)); + /* add idle wait as per freedesktop.org bug 24041 */ + if (r100_gui_wait_for_idle(rdev)) { + printk(KERN_WARNING "Failed to wait GUI idle while " + "programming pipes. Bad things might happen.\n"); + } + /* get max number of pipes */ + gb_pipe_select = RREG32(R400_GB_PIPE_SELECT); + num_pipes = ((gb_pipe_select >> 12) & 3) + 1; + + /* SE chips have 1 pipe */ + if ((rdev->pdev->device == 0x5e4c) || + (rdev->pdev->device == 0x5e4f)) + num_pipes = 1; + + rdev->num_gb_pipes = num_pipes; + tmp = 0; + switch (num_pipes) { + default: + /* force to 1 pipe */ + num_pipes = 1; + case 1: + tmp = (0 << 1); + break; + case 2: + tmp = (3 << 1); + break; + case 3: + tmp = (6 << 1); + break; + case 4: + tmp = (7 << 1); + break; + } + WREG32(R500_SU_REG_DEST, (1 << num_pipes) - 1); + /* Sub pixel 1/12 so we can have 4K rendering according to doc */ + tmp |= R300_TILE_SIZE_16 | R300_ENABLE_TILING; + WREG32(R300_GB_TILE_CONFIG, tmp); + if (r100_gui_wait_for_idle(rdev)) { + printk(KERN_WARNING "Failed to wait GUI idle while " + "programming pipes. Bad things might happen.\n"); + } + + tmp = RREG32(R300_DST_PIPE_CONFIG); + WREG32(R300_DST_PIPE_CONFIG, tmp | R300_PIPE_AUTO_CONFIG); + + WREG32(R300_RB2D_DSTCACHE_MODE, + RREG32(R300_RB2D_DSTCACHE_MODE) | + R300_DC_AUTOFLUSH_ENABLE | + R300_DC_DC_DISABLE_IGNORE_PE); + + if (r100_gui_wait_for_idle(rdev)) { + printk(KERN_WARNING "Failed to wait GUI idle while " + "programming pipes. Bad things might happen.\n"); + } + + if (rdev->family == CHIP_RV530) { + tmp = RREG32(RV530_GB_PIPE_SELECT2); + if ((tmp & 3) == 3) + rdev->num_z_pipes = 2; + else + rdev->num_z_pipes = 1; + } else + rdev->num_z_pipes = 1; + + DRM_INFO("radeon: %d quad pipes, %d z pipes initialized.\n", + rdev->num_gb_pipes, rdev->num_z_pipes); +} + +u32 r420_mc_rreg(struct radeon_device *rdev, u32 reg) +{ + u32 r; + + WREG32(R_0001F8_MC_IND_INDEX, S_0001F8_MC_IND_ADDR(reg)); + r = RREG32(R_0001FC_MC_IND_DATA); + return r; +} + +void r420_mc_wreg(struct radeon_device *rdev, u32 reg, u32 v) +{ + WREG32(R_0001F8_MC_IND_INDEX, S_0001F8_MC_IND_ADDR(reg) | + S_0001F8_MC_IND_WR_EN(1)); + WREG32(R_0001FC_MC_IND_DATA, v); +} + +static void r420_debugfs(struct radeon_device *rdev) +{ + if (r100_debugfs_rbbm_init(rdev)) { + DRM_ERROR("Failed to register debugfs file for RBBM !\n"); + } + if (r420_debugfs_pipes_info_init(rdev)) { + DRM_ERROR("Failed to register debugfs file for pipes !\n"); + } +} + +static void r420_clock_resume(struct radeon_device *rdev) +{ + u32 sclk_cntl; + + if (radeon_dynclks != -1 && radeon_dynclks) + radeon_atom_set_clock_gating(rdev, 1); + sclk_cntl = RREG32_PLL(R_00000D_SCLK_CNTL); + sclk_cntl |= S_00000D_FORCE_CP(1) | S_00000D_FORCE_VIP(1); + if (rdev->family == CHIP_R420) + sclk_cntl |= S_00000D_FORCE_PX(1) | S_00000D_FORCE_TX(1); + WREG32_PLL(R_00000D_SCLK_CNTL, sclk_cntl); +} + +static void r420_cp_errata_init(struct radeon_device *rdev) +{ + /* RV410 and R420 can lock up if CP DMA to host memory happens + * while the 2D engine is busy. + * + * The proper workaround is to queue a RESYNC at the beginning + * of the CP init, apparently. + */ + radeon_scratch_get(rdev, &rdev->config.r300.resync_scratch); + radeon_ring_lock(rdev, 8); + radeon_ring_write(rdev, PACKET0(R300_CP_RESYNC_ADDR, 1)); + radeon_ring_write(rdev, rdev->config.r300.resync_scratch); + radeon_ring_write(rdev, 0xDEADBEEF); + radeon_ring_unlock_commit(rdev); +} + +static void r420_cp_errata_fini(struct radeon_device *rdev) +{ + /* Catch the RESYNC we dispatched all the way back, + * at the very beginning of the CP init. + */ + radeon_ring_lock(rdev, 8); + radeon_ring_write(rdev, PACKET0(R300_RB3D_DSTCACHE_CTLSTAT, 0)); + radeon_ring_write(rdev, R300_RB3D_DC_FINISH); + radeon_ring_unlock_commit(rdev); + radeon_scratch_free(rdev, rdev->config.r300.resync_scratch); +} + +static int r420_startup(struct radeon_device *rdev) +{ + int r; + + /* set common regs */ + r100_set_common_regs(rdev); + /* program mc */ + r300_mc_program(rdev); + /* Resume clock */ + r420_clock_resume(rdev); + /* Initialize GART (initialize after TTM so we can allocate + * memory through TTM but finalize after TTM) */ + if (rdev->flags & RADEON_IS_PCIE) { + r = rv370_pcie_gart_enable(rdev); + if (r) + return r; + } + if (rdev->flags & RADEON_IS_PCI) { + r = r100_pci_gart_enable(rdev); + if (r) + return r; + } + r420_pipes_init(rdev); + + /* allocate wb buffer */ + r = radeon_wb_init(rdev); + if (r) + return r; + + /* Enable IRQ */ + r100_irq_set(rdev); + rdev->config.r300.hdp_cntl = RREG32(RADEON_HOST_PATH_CNTL); + /* 1M ring buffer */ + r = r100_cp_init(rdev, 1024 * 1024); + if (r) { + dev_err(rdev->dev, "failed initializing CP (%d).\n", r); + return r; + } + r420_cp_errata_init(rdev); + r = r100_ib_init(rdev); + if (r) { + dev_err(rdev->dev, "failed initializing IB (%d).\n", r); + return r; + } + return 0; +} + +int r420_resume(struct radeon_device *rdev) +{ + /* Make sur GART are not working */ + if (rdev->flags & RADEON_IS_PCIE) + rv370_pcie_gart_disable(rdev); + if (rdev->flags & RADEON_IS_PCI) + r100_pci_gart_disable(rdev); + /* Resume clock before doing reset */ + r420_clock_resume(rdev); + /* Reset gpu before posting otherwise ATOM will enter infinite loop */ + if (radeon_asic_reset(rdev)) { + dev_warn(rdev->dev, "GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n", + RREG32(R_000E40_RBBM_STATUS), + RREG32(R_0007C0_CP_STAT)); + } + /* check if cards are posted or not */ + if (rdev->is_atom_bios) { + atom_asic_init(rdev->mode_info.atom_context); + } else { + radeon_combios_asic_init(rdev->ddev); + } + /* Resume clock after posting */ + r420_clock_resume(rdev); + /* Initialize surface registers */ + radeon_surface_init(rdev); + return r420_startup(rdev); +} + +int r420_suspend(struct radeon_device *rdev) +{ + r420_cp_errata_fini(rdev); + r100_cp_disable(rdev); + radeon_wb_disable(rdev); + r100_irq_disable(rdev); + if (rdev->flags & RADEON_IS_PCIE) + rv370_pcie_gart_disable(rdev); + if (rdev->flags & RADEON_IS_PCI) + r100_pci_gart_disable(rdev); + return 0; +} + +void r420_fini(struct radeon_device *rdev) +{ + r100_cp_fini(rdev); + radeon_wb_fini(rdev); + r100_ib_fini(rdev); + radeon_gem_fini(rdev); + if (rdev->flags & RADEON_IS_PCIE) + rv370_pcie_gart_fini(rdev); + if (rdev->flags & RADEON_IS_PCI) + r100_pci_gart_fini(rdev); + radeon_agp_fini(rdev); + radeon_irq_kms_fini(rdev); + radeon_fence_driver_fini(rdev); + radeon_bo_fini(rdev); + if (rdev->is_atom_bios) { + radeon_atombios_fini(rdev); + } else { + radeon_combios_fini(rdev); + } + kfree(rdev->bios); + rdev->bios = NULL; +} + +int r420_init(struct radeon_device *rdev) +{ + int r; + + /* Initialize scratch registers */ + radeon_scratch_init(rdev); + /* Initialize surface registers */ + radeon_surface_init(rdev); + /* TODO: disable VGA need to use VGA request */ + /* restore some register to sane defaults */ + r100_restore_sanity(rdev); + /* BIOS*/ + if (!radeon_get_bios(rdev)) { + if (ASIC_IS_AVIVO(rdev)) + return -EINVAL; + } + if (rdev->is_atom_bios) { + r = radeon_atombios_init(rdev); + if (r) { + return r; + } + } else { + r = radeon_combios_init(rdev); + if (r) { + return r; + } + } + /* Reset gpu before posting otherwise ATOM will enter infinite loop */ + if (radeon_asic_reset(rdev)) { + dev_warn(rdev->dev, + "GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n", + RREG32(R_000E40_RBBM_STATUS), + RREG32(R_0007C0_CP_STAT)); + } + /* check if cards are posted or not */ + if (radeon_boot_test_post_card(rdev) == false) + return -EINVAL; + + /* Initialize clocks */ + radeon_get_clock_info(rdev->ddev); + /* initialize AGP */ + if (rdev->flags & RADEON_IS_AGP) { + r = radeon_agp_init(rdev); + if (r) { + radeon_agp_disable(rdev); + } + } + /* initialize memory controller */ + r300_mc_init(rdev); + r420_debugfs(rdev); + /* Fence driver */ + r = radeon_fence_driver_init(rdev); + if (r) { + return r; + } + r = radeon_irq_kms_init(rdev); + if (r) { + return r; + } + /* Memory manager */ + r = radeon_bo_init(rdev); + if (r) { + return r; + } + if (rdev->family == CHIP_R420) + r100_enable_bm(rdev); + + if (rdev->flags & RADEON_IS_PCIE) { + r = rv370_pcie_gart_init(rdev); + if (r) + return r; + } + if (rdev->flags & RADEON_IS_PCI) { + r = r100_pci_gart_init(rdev); + if (r) + return r; + } + r420_set_reg_safe(rdev); + rdev->accel_working = true; + r = r420_startup(rdev); + if (r) { + /* Somethings want wront with the accel init stop accel */ + dev_err(rdev->dev, "Disabling GPU acceleration\n"); + r100_cp_fini(rdev); + radeon_wb_fini(rdev); + r100_ib_fini(rdev); + radeon_irq_kms_fini(rdev); + if (rdev->flags & RADEON_IS_PCIE) + rv370_pcie_gart_fini(rdev); + if (rdev->flags & RADEON_IS_PCI) + r100_pci_gart_fini(rdev); + radeon_agp_fini(rdev); + rdev->accel_working = false; + } + return 0; +} + +/* + * Debugfs info + */ +#if defined(CONFIG_DEBUG_FS) +static int r420_debugfs_pipes_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + struct radeon_device *rdev = dev->dev_private; + uint32_t tmp; + + tmp = RREG32(R400_GB_PIPE_SELECT); + seq_printf(m, "GB_PIPE_SELECT 0x%08x\n", tmp); + tmp = RREG32(R300_GB_TILE_CONFIG); + seq_printf(m, "GB_TILE_CONFIG 0x%08x\n", tmp); + tmp = RREG32(R300_DST_PIPE_CONFIG); + seq_printf(m, "DST_PIPE_CONFIG 0x%08x\n", tmp); + return 0; +} + +static struct drm_info_list r420_pipes_info_list[] = { + {"r420_pipes_info", r420_debugfs_pipes_info, 0, NULL}, +}; +#endif + +int r420_debugfs_pipes_info_init(struct radeon_device *rdev) +{ +#if defined(CONFIG_DEBUG_FS) + return radeon_debugfs_add_files(rdev, r420_pipes_info_list, 1); +#else + return 0; +#endif +} diff --git a/drivers/gpu/drm/radeon/r420d.h b/drivers/gpu/drm/radeon/r420d.h new file mode 100644 index 00000000..fc78d31a --- /dev/null +++ b/drivers/gpu/drm/radeon/r420d.h @@ -0,0 +1,249 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#ifndef R420D_H +#define R420D_H + +#define R_0001F8_MC_IND_INDEX 0x0001F8 +#define S_0001F8_MC_IND_ADDR(x) (((x) & 0x7F) << 0) +#define G_0001F8_MC_IND_ADDR(x) (((x) >> 0) & 0x7F) +#define C_0001F8_MC_IND_ADDR 0xFFFFFF80 +#define S_0001F8_MC_IND_WR_EN(x) (((x) & 0x1) << 8) +#define G_0001F8_MC_IND_WR_EN(x) (((x) >> 8) & 0x1) +#define C_0001F8_MC_IND_WR_EN 0xFFFFFEFF +#define R_0001FC_MC_IND_DATA 0x0001FC +#define S_0001FC_MC_IND_DATA(x) (((x) & 0xFFFFFFFF) << 0) +#define G_0001FC_MC_IND_DATA(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_0001FC_MC_IND_DATA 0x00000000 +#define R_0007C0_CP_STAT 0x0007C0 +#define S_0007C0_MRU_BUSY(x) (((x) & 0x1) << 0) +#define G_0007C0_MRU_BUSY(x) (((x) >> 0) & 0x1) +#define C_0007C0_MRU_BUSY 0xFFFFFFFE +#define S_0007C0_MWU_BUSY(x) (((x) & 0x1) << 1) +#define G_0007C0_MWU_BUSY(x) (((x) >> 1) & 0x1) +#define C_0007C0_MWU_BUSY 0xFFFFFFFD +#define S_0007C0_RSIU_BUSY(x) (((x) & 0x1) << 2) +#define G_0007C0_RSIU_BUSY(x) (((x) >> 2) & 0x1) +#define C_0007C0_RSIU_BUSY 0xFFFFFFFB +#define S_0007C0_RCIU_BUSY(x) (((x) & 0x1) << 3) +#define G_0007C0_RCIU_BUSY(x) (((x) >> 3) & 0x1) +#define C_0007C0_RCIU_BUSY 0xFFFFFFF7 +#define S_0007C0_CSF_PRIMARY_BUSY(x) (((x) & 0x1) << 9) +#define G_0007C0_CSF_PRIMARY_BUSY(x) (((x) >> 9) & 0x1) +#define C_0007C0_CSF_PRIMARY_BUSY 0xFFFFFDFF +#define S_0007C0_CSF_INDIRECT_BUSY(x) (((x) & 0x1) << 10) +#define G_0007C0_CSF_INDIRECT_BUSY(x) (((x) >> 10) & 0x1) +#define C_0007C0_CSF_INDIRECT_BUSY 0xFFFFFBFF +#define S_0007C0_CSQ_PRIMARY_BUSY(x) (((x) & 0x1) << 11) +#define G_0007C0_CSQ_PRIMARY_BUSY(x) (((x) >> 11) & 0x1) +#define C_0007C0_CSQ_PRIMARY_BUSY 0xFFFFF7FF +#define S_0007C0_CSQ_INDIRECT_BUSY(x) (((x) & 0x1) << 12) +#define G_0007C0_CSQ_INDIRECT_BUSY(x) (((x) >> 12) & 0x1) +#define C_0007C0_CSQ_INDIRECT_BUSY 0xFFFFEFFF +#define S_0007C0_CSI_BUSY(x) (((x) & 0x1) << 13) +#define G_0007C0_CSI_BUSY(x) (((x) >> 13) & 0x1) +#define C_0007C0_CSI_BUSY 0xFFFFDFFF +#define S_0007C0_CSF_INDIRECT2_BUSY(x) (((x) & 0x1) << 14) +#define G_0007C0_CSF_INDIRECT2_BUSY(x) (((x) >> 14) & 0x1) +#define C_0007C0_CSF_INDIRECT2_BUSY 0xFFFFBFFF +#define S_0007C0_CSQ_INDIRECT2_BUSY(x) (((x) & 0x1) << 15) +#define G_0007C0_CSQ_INDIRECT2_BUSY(x) (((x) >> 15) & 0x1) +#define C_0007C0_CSQ_INDIRECT2_BUSY 0xFFFF7FFF +#define S_0007C0_GUIDMA_BUSY(x) (((x) & 0x1) << 28) +#define G_0007C0_GUIDMA_BUSY(x) (((x) >> 28) & 0x1) +#define C_0007C0_GUIDMA_BUSY 0xEFFFFFFF +#define S_0007C0_VIDDMA_BUSY(x) (((x) & 0x1) << 29) +#define G_0007C0_VIDDMA_BUSY(x) (((x) >> 29) & 0x1) +#define C_0007C0_VIDDMA_BUSY 0xDFFFFFFF +#define S_0007C0_CMDSTRM_BUSY(x) (((x) & 0x1) << 30) +#define G_0007C0_CMDSTRM_BUSY(x) (((x) >> 30) & 0x1) +#define C_0007C0_CMDSTRM_BUSY 0xBFFFFFFF +#define S_0007C0_CP_BUSY(x) (((x) & 0x1) << 31) +#define G_0007C0_CP_BUSY(x) (((x) >> 31) & 0x1) +#define C_0007C0_CP_BUSY 0x7FFFFFFF +#define R_000E40_RBBM_STATUS 0x000E40 +#define S_000E40_CMDFIFO_AVAIL(x) (((x) & 0x7F) << 0) +#define G_000E40_CMDFIFO_AVAIL(x) (((x) >> 0) & 0x7F) +#define C_000E40_CMDFIFO_AVAIL 0xFFFFFF80 +#define S_000E40_HIRQ_ON_RBB(x) (((x) & 0x1) << 8) +#define G_000E40_HIRQ_ON_RBB(x) (((x) >> 8) & 0x1) +#define C_000E40_HIRQ_ON_RBB 0xFFFFFEFF +#define S_000E40_CPRQ_ON_RBB(x) (((x) & 0x1) << 9) +#define G_000E40_CPRQ_ON_RBB(x) (((x) >> 9) & 0x1) +#define C_000E40_CPRQ_ON_RBB 0xFFFFFDFF +#define S_000E40_CFRQ_ON_RBB(x) (((x) & 0x1) << 10) +#define G_000E40_CFRQ_ON_RBB(x) (((x) >> 10) & 0x1) +#define C_000E40_CFRQ_ON_RBB 0xFFFFFBFF +#define S_000E40_HIRQ_IN_RTBUF(x) (((x) & 0x1) << 11) +#define G_000E40_HIRQ_IN_RTBUF(x) (((x) >> 11) & 0x1) +#define C_000E40_HIRQ_IN_RTBUF 0xFFFFF7FF +#define S_000E40_CPRQ_IN_RTBUF(x) (((x) & 0x1) << 12) +#define G_000E40_CPRQ_IN_RTBUF(x) (((x) >> 12) & 0x1) +#define C_000E40_CPRQ_IN_RTBUF 0xFFFFEFFF +#define S_000E40_CFRQ_IN_RTBUF(x) (((x) & 0x1) << 13) +#define G_000E40_CFRQ_IN_RTBUF(x) (((x) >> 13) & 0x1) +#define C_000E40_CFRQ_IN_RTBUF 0xFFFFDFFF +#define S_000E40_CF_PIPE_BUSY(x) (((x) & 0x1) << 14) +#define G_000E40_CF_PIPE_BUSY(x) (((x) >> 14) & 0x1) +#define C_000E40_CF_PIPE_BUSY 0xFFFFBFFF +#define S_000E40_ENG_EV_BUSY(x) (((x) & 0x1) << 15) +#define G_000E40_ENG_EV_BUSY(x) (((x) >> 15) & 0x1) +#define C_000E40_ENG_EV_BUSY 0xFFFF7FFF +#define S_000E40_CP_CMDSTRM_BUSY(x) (((x) & 0x1) << 16) +#define G_000E40_CP_CMDSTRM_BUSY(x) (((x) >> 16) & 0x1) +#define C_000E40_CP_CMDSTRM_BUSY 0xFFFEFFFF +#define S_000E40_E2_BUSY(x) (((x) & 0x1) << 17) +#define G_000E40_E2_BUSY(x) (((x) >> 17) & 0x1) +#define C_000E40_E2_BUSY 0xFFFDFFFF +#define S_000E40_RB2D_BUSY(x) (((x) & 0x1) << 18) +#define G_000E40_RB2D_BUSY(x) (((x) >> 18) & 0x1) +#define C_000E40_RB2D_BUSY 0xFFFBFFFF +#define S_000E40_RB3D_BUSY(x) (((x) & 0x1) << 19) +#define G_000E40_RB3D_BUSY(x) (((x) >> 19) & 0x1) +#define C_000E40_RB3D_BUSY 0xFFF7FFFF +#define S_000E40_VAP_BUSY(x) (((x) & 0x1) << 20) +#define G_000E40_VAP_BUSY(x) (((x) >> 20) & 0x1) +#define C_000E40_VAP_BUSY 0xFFEFFFFF +#define S_000E40_RE_BUSY(x) (((x) & 0x1) << 21) +#define G_000E40_RE_BUSY(x) (((x) >> 21) & 0x1) +#define C_000E40_RE_BUSY 0xFFDFFFFF +#define S_000E40_TAM_BUSY(x) (((x) & 0x1) << 22) +#define G_000E40_TAM_BUSY(x) (((x) >> 22) & 0x1) +#define C_000E40_TAM_BUSY 0xFFBFFFFF +#define S_000E40_TDM_BUSY(x) (((x) & 0x1) << 23) +#define G_000E40_TDM_BUSY(x) (((x) >> 23) & 0x1) +#define C_000E40_TDM_BUSY 0xFF7FFFFF +#define S_000E40_PB_BUSY(x) (((x) & 0x1) << 24) +#define G_000E40_PB_BUSY(x) (((x) >> 24) & 0x1) +#define C_000E40_PB_BUSY 0xFEFFFFFF +#define S_000E40_TIM_BUSY(x) (((x) & 0x1) << 25) +#define G_000E40_TIM_BUSY(x) (((x) >> 25) & 0x1) +#define C_000E40_TIM_BUSY 0xFDFFFFFF +#define S_000E40_GA_BUSY(x) (((x) & 0x1) << 26) +#define G_000E40_GA_BUSY(x) (((x) >> 26) & 0x1) +#define C_000E40_GA_BUSY 0xFBFFFFFF +#define S_000E40_CBA2D_BUSY(x) (((x) & 0x1) << 27) +#define G_000E40_CBA2D_BUSY(x) (((x) >> 27) & 0x1) +#define C_000E40_CBA2D_BUSY 0xF7FFFFFF +#define S_000E40_GUI_ACTIVE(x) (((x) & 0x1) << 31) +#define G_000E40_GUI_ACTIVE(x) (((x) >> 31) & 0x1) +#define C_000E40_GUI_ACTIVE 0x7FFFFFFF + +/* CLK registers */ +#define R_00000D_SCLK_CNTL 0x00000D +#define S_00000D_SCLK_SRC_SEL(x) (((x) & 0x7) << 0) +#define G_00000D_SCLK_SRC_SEL(x) (((x) >> 0) & 0x7) +#define C_00000D_SCLK_SRC_SEL 0xFFFFFFF8 +#define S_00000D_CP_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 3) +#define G_00000D_CP_MAX_DYN_STOP_LAT(x) (((x) >> 3) & 0x1) +#define C_00000D_CP_MAX_DYN_STOP_LAT 0xFFFFFFF7 +#define S_00000D_HDP_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 4) +#define G_00000D_HDP_MAX_DYN_STOP_LAT(x) (((x) >> 4) & 0x1) +#define C_00000D_HDP_MAX_DYN_STOP_LAT 0xFFFFFFEF +#define S_00000D_TV_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 5) +#define G_00000D_TV_MAX_DYN_STOP_LAT(x) (((x) >> 5) & 0x1) +#define C_00000D_TV_MAX_DYN_STOP_LAT 0xFFFFFFDF +#define S_00000D_E2_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 6) +#define G_00000D_E2_MAX_DYN_STOP_LAT(x) (((x) >> 6) & 0x1) +#define C_00000D_E2_MAX_DYN_STOP_LAT 0xFFFFFFBF +#define S_00000D_SE_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 7) +#define G_00000D_SE_MAX_DYN_STOP_LAT(x) (((x) >> 7) & 0x1) +#define C_00000D_SE_MAX_DYN_STOP_LAT 0xFFFFFF7F +#define S_00000D_IDCT_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 8) +#define G_00000D_IDCT_MAX_DYN_STOP_LAT(x) (((x) >> 8) & 0x1) +#define C_00000D_IDCT_MAX_DYN_STOP_LAT 0xFFFFFEFF +#define S_00000D_VIP_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 9) +#define G_00000D_VIP_MAX_DYN_STOP_LAT(x) (((x) >> 9) & 0x1) +#define C_00000D_VIP_MAX_DYN_STOP_LAT 0xFFFFFDFF +#define S_00000D_RE_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 10) +#define G_00000D_RE_MAX_DYN_STOP_LAT(x) (((x) >> 10) & 0x1) +#define C_00000D_RE_MAX_DYN_STOP_LAT 0xFFFFFBFF +#define S_00000D_PB_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 11) +#define G_00000D_PB_MAX_DYN_STOP_LAT(x) (((x) >> 11) & 0x1) +#define C_00000D_PB_MAX_DYN_STOP_LAT 0xFFFFF7FF +#define S_00000D_TAM_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 12) +#define G_00000D_TAM_MAX_DYN_STOP_LAT(x) (((x) >> 12) & 0x1) +#define C_00000D_TAM_MAX_DYN_STOP_LAT 0xFFFFEFFF +#define S_00000D_TDM_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 13) +#define G_00000D_TDM_MAX_DYN_STOP_LAT(x) (((x) >> 13) & 0x1) +#define C_00000D_TDM_MAX_DYN_STOP_LAT 0xFFFFDFFF +#define S_00000D_RB_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 14) +#define G_00000D_RB_MAX_DYN_STOP_LAT(x) (((x) >> 14) & 0x1) +#define C_00000D_RB_MAX_DYN_STOP_LAT 0xFFFFBFFF +#define S_00000D_FORCE_DISP2(x) (((x) & 0x1) << 15) +#define G_00000D_FORCE_DISP2(x) (((x) >> 15) & 0x1) +#define C_00000D_FORCE_DISP2 0xFFFF7FFF +#define S_00000D_FORCE_CP(x) (((x) & 0x1) << 16) +#define G_00000D_FORCE_CP(x) (((x) >> 16) & 0x1) +#define C_00000D_FORCE_CP 0xFFFEFFFF +#define S_00000D_FORCE_HDP(x) (((x) & 0x1) << 17) +#define G_00000D_FORCE_HDP(x) (((x) >> 17) & 0x1) +#define C_00000D_FORCE_HDP 0xFFFDFFFF +#define S_00000D_FORCE_DISP1(x) (((x) & 0x1) << 18) +#define G_00000D_FORCE_DISP1(x) (((x) >> 18) & 0x1) +#define C_00000D_FORCE_DISP1 0xFFFBFFFF +#define S_00000D_FORCE_TOP(x) (((x) & 0x1) << 19) +#define G_00000D_FORCE_TOP(x) (((x) >> 19) & 0x1) +#define C_00000D_FORCE_TOP 0xFFF7FFFF +#define S_00000D_FORCE_E2(x) (((x) & 0x1) << 20) +#define G_00000D_FORCE_E2(x) (((x) >> 20) & 0x1) +#define C_00000D_FORCE_E2 0xFFEFFFFF +#define S_00000D_FORCE_VAP(x) (((x) & 0x1) << 21) +#define G_00000D_FORCE_VAP(x) (((x) >> 21) & 0x1) +#define C_00000D_FORCE_VAP 0xFFDFFFFF +#define S_00000D_FORCE_IDCT(x) (((x) & 0x1) << 22) +#define G_00000D_FORCE_IDCT(x) (((x) >> 22) & 0x1) +#define C_00000D_FORCE_IDCT 0xFFBFFFFF +#define S_00000D_FORCE_VIP(x) (((x) & 0x1) << 23) +#define G_00000D_FORCE_VIP(x) (((x) >> 23) & 0x1) +#define C_00000D_FORCE_VIP 0xFF7FFFFF +#define S_00000D_FORCE_RE(x) (((x) & 0x1) << 24) +#define G_00000D_FORCE_RE(x) (((x) >> 24) & 0x1) +#define C_00000D_FORCE_RE 0xFEFFFFFF +#define S_00000D_FORCE_SR(x) (((x) & 0x1) << 25) +#define G_00000D_FORCE_SR(x) (((x) >> 25) & 0x1) +#define C_00000D_FORCE_SR 0xFDFFFFFF +#define S_00000D_FORCE_PX(x) (((x) & 0x1) << 26) +#define G_00000D_FORCE_PX(x) (((x) >> 26) & 0x1) +#define C_00000D_FORCE_PX 0xFBFFFFFF +#define S_00000D_FORCE_TX(x) (((x) & 0x1) << 27) +#define G_00000D_FORCE_TX(x) (((x) >> 27) & 0x1) +#define C_00000D_FORCE_TX 0xF7FFFFFF +#define S_00000D_FORCE_US(x) (((x) & 0x1) << 28) +#define G_00000D_FORCE_US(x) (((x) >> 28) & 0x1) +#define C_00000D_FORCE_US 0xEFFFFFFF +#define S_00000D_FORCE_TV_SCLK(x) (((x) & 0x1) << 29) +#define G_00000D_FORCE_TV_SCLK(x) (((x) >> 29) & 0x1) +#define C_00000D_FORCE_TV_SCLK 0xDFFFFFFF +#define S_00000D_FORCE_SU(x) (((x) & 0x1) << 30) +#define G_00000D_FORCE_SU(x) (((x) >> 30) & 0x1) +#define C_00000D_FORCE_SU 0xBFFFFFFF +#define S_00000D_FORCE_OV0(x) (((x) & 0x1) << 31) +#define G_00000D_FORCE_OV0(x) (((x) >> 31) & 0x1) +#define C_00000D_FORCE_OV0 0x7FFFFFFF + +#endif diff --git a/drivers/gpu/drm/radeon/r500_reg.h b/drivers/gpu/drm/radeon/r500_reg.h new file mode 100644 index 00000000..fc437059 --- /dev/null +++ b/drivers/gpu/drm/radeon/r500_reg.h @@ -0,0 +1,793 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#ifndef __R500_REG_H__ +#define __R500_REG_H__ + +/* pipe config regs */ +#define R300_GA_POLY_MODE 0x4288 +# define R300_FRONT_PTYPE_POINT (0 << 4) +# define R300_FRONT_PTYPE_LINE (1 << 4) +# define R300_FRONT_PTYPE_TRIANGE (2 << 4) +# define R300_BACK_PTYPE_POINT (0 << 7) +# define R300_BACK_PTYPE_LINE (1 << 7) +# define R300_BACK_PTYPE_TRIANGE (2 << 7) +#define R300_GA_ROUND_MODE 0x428c +# define R300_GEOMETRY_ROUND_TRUNC (0 << 0) +# define R300_GEOMETRY_ROUND_NEAREST (1 << 0) +# define R300_COLOR_ROUND_TRUNC (0 << 2) +# define R300_COLOR_ROUND_NEAREST (1 << 2) +#define R300_GB_MSPOS0 0x4010 +# define R300_MS_X0_SHIFT 0 +# define R300_MS_Y0_SHIFT 4 +# define R300_MS_X1_SHIFT 8 +# define R300_MS_Y1_SHIFT 12 +# define R300_MS_X2_SHIFT 16 +# define R300_MS_Y2_SHIFT 20 +# define R300_MSBD0_Y_SHIFT 24 +# define R300_MSBD0_X_SHIFT 28 +#define R300_GB_MSPOS1 0x4014 +# define R300_MS_X3_SHIFT 0 +# define R300_MS_Y3_SHIFT 4 +# define R300_MS_X4_SHIFT 8 +# define R300_MS_Y4_SHIFT 12 +# define R300_MS_X5_SHIFT 16 +# define R300_MS_Y5_SHIFT 20 +# define R300_MSBD1_SHIFT 24 + +#define R300_GA_ENHANCE 0x4274 +# define R300_GA_DEADLOCK_CNTL (1 << 0) +# define R300_GA_FASTSYNC_CNTL (1 << 1) +#define R300_RB3D_DSTCACHE_CTLSTAT 0x4e4c +# define R300_RB3D_DC_FLUSH (2 << 0) +# define R300_RB3D_DC_FREE (2 << 2) +# define R300_RB3D_DC_FINISH (1 << 4) +#define R300_RB3D_ZCACHE_CTLSTAT 0x4f18 +# define R300_ZC_FLUSH (1 << 0) +# define R300_ZC_FREE (1 << 1) +# define R300_ZC_FLUSH_ALL 0x3 +#define R400_GB_PIPE_SELECT 0x402c +#define R500_DYN_SCLK_PWMEM_PIPE 0x000d /* PLL */ +#define R500_SU_REG_DEST 0x42c8 +#define R300_GB_TILE_CONFIG 0x4018 +# define R300_ENABLE_TILING (1 << 0) +# define R300_PIPE_COUNT_RV350 (0 << 1) +# define R300_PIPE_COUNT_R300 (3 << 1) +# define R300_PIPE_COUNT_R420_3P (6 << 1) +# define R300_PIPE_COUNT_R420 (7 << 1) +# define R300_TILE_SIZE_8 (0 << 4) +# define R300_TILE_SIZE_16 (1 << 4) +# define R300_TILE_SIZE_32 (2 << 4) +# define R300_SUBPIXEL_1_12 (0 << 16) +# define R300_SUBPIXEL_1_16 (1 << 16) +#define R300_DST_PIPE_CONFIG 0x170c +# define R300_PIPE_AUTO_CONFIG (1 << 31) +#define R300_RB2D_DSTCACHE_MODE 0x3428 +# define R300_DC_AUTOFLUSH_ENABLE (1 << 8) +# define R300_DC_DC_DISABLE_IGNORE_PE (1 << 17) + +#define RADEON_CP_STAT 0x7C0 +#define RADEON_RBBM_CMDFIFO_ADDR 0xE70 +#define RADEON_RBBM_CMDFIFO_DATA 0xE74 +#define RADEON_ISYNC_CNTL 0x1724 +# define RADEON_ISYNC_ANY2D_IDLE3D (1 << 0) +# define RADEON_ISYNC_ANY3D_IDLE2D (1 << 1) +# define RADEON_ISYNC_TRIG2D_IDLE3D (1 << 2) +# define RADEON_ISYNC_TRIG3D_IDLE2D (1 << 3) +# define RADEON_ISYNC_WAIT_IDLEGUI (1 << 4) +# define RADEON_ISYNC_CPSCRATCH_IDLEGUI (1 << 5) + +#define RS480_NB_MC_INDEX 0x168 +# define RS480_NB_MC_IND_WR_EN (1 << 8) +#define RS480_NB_MC_DATA 0x16c + +/* + * RS690 + */ +#define RS690_MCCFG_FB_LOCATION 0x100 +#define RS690_MC_FB_START_MASK 0x0000FFFF +#define RS690_MC_FB_START_SHIFT 0 +#define RS690_MC_FB_TOP_MASK 0xFFFF0000 +#define RS690_MC_FB_TOP_SHIFT 16 +#define RS690_MCCFG_AGP_LOCATION 0x101 +#define RS690_MC_AGP_START_MASK 0x0000FFFF +#define RS690_MC_AGP_START_SHIFT 0 +#define RS690_MC_AGP_TOP_MASK 0xFFFF0000 +#define RS690_MC_AGP_TOP_SHIFT 16 +#define RS690_MCCFG_AGP_BASE 0x102 +#define RS690_MCCFG_AGP_BASE_2 0x103 +#define RS690_MC_INIT_MISC_LAT_TIMER 0x104 +#define RS690_HDP_FB_LOCATION 0x0134 +#define RS690_MC_INDEX 0x78 +# define RS690_MC_INDEX_MASK 0x1ff +# define RS690_MC_INDEX_WR_EN (1 << 9) +# define RS690_MC_INDEX_WR_ACK 0x7f +#define RS690_MC_DATA 0x7c +#define RS690_MC_STATUS 0x90 +#define RS690_MC_STATUS_IDLE (1 << 0) +#define RS480_AGP_BASE_2 0x0164 +#define RS480_MC_MISC_CNTL 0x18 +# define RS480_DISABLE_GTW (1 << 1) +# define RS480_GART_INDEX_REG_EN (1 << 12) +# define RS690_BLOCK_GFX_D3_EN (1 << 14) +#define RS480_GART_FEATURE_ID 0x2b +# define RS480_HANG_EN (1 << 11) +# define RS480_TLB_ENABLE (1 << 18) +# define RS480_P2P_ENABLE (1 << 19) +# define RS480_GTW_LAC_EN (1 << 25) +# define RS480_2LEVEL_GART (0 << 30) +# define RS480_1LEVEL_GART (1 << 30) +# define RS480_PDC_EN (1 << 31) +#define RS480_GART_BASE 0x2c +#define RS480_GART_CACHE_CNTRL 0x2e +# define RS480_GART_CACHE_INVALIDATE (1 << 0) /* wait for it to clear */ +#define RS480_AGP_ADDRESS_SPACE_SIZE 0x38 +# define RS480_GART_EN (1 << 0) +# define RS480_VA_SIZE_32MB (0 << 1) +# define RS480_VA_SIZE_64MB (1 << 1) +# define RS480_VA_SIZE_128MB (2 << 1) +# define RS480_VA_SIZE_256MB (3 << 1) +# define RS480_VA_SIZE_512MB (4 << 1) +# define RS480_VA_SIZE_1GB (5 << 1) +# define RS480_VA_SIZE_2GB (6 << 1) +#define RS480_AGP_MODE_CNTL 0x39 +# define RS480_POST_GART_Q_SIZE (1 << 18) +# define RS480_NONGART_SNOOP (1 << 19) +# define RS480_AGP_RD_BUF_SIZE (1 << 20) +# define RS480_REQ_TYPE_SNOOP_SHIFT 22 +# define RS480_REQ_TYPE_SNOOP_MASK 0x3 +# define RS480_REQ_TYPE_SNOOP_DIS (1 << 24) + +#define RS690_AIC_CTRL_SCRATCH 0x3A +# define RS690_DIS_OUT_OF_PCI_GART_ACCESS (1 << 1) + +/* + * RS600 + */ +#define RS600_MC_STATUS 0x0 +#define RS600_MC_STATUS_IDLE (1 << 0) +#define RS600_MC_INDEX 0x70 +# define RS600_MC_ADDR_MASK 0xffff +# define RS600_MC_IND_SEQ_RBS_0 (1 << 16) +# define RS600_MC_IND_SEQ_RBS_1 (1 << 17) +# define RS600_MC_IND_SEQ_RBS_2 (1 << 18) +# define RS600_MC_IND_SEQ_RBS_3 (1 << 19) +# define RS600_MC_IND_AIC_RBS (1 << 20) +# define RS600_MC_IND_CITF_ARB0 (1 << 21) +# define RS600_MC_IND_CITF_ARB1 (1 << 22) +# define RS600_MC_IND_WR_EN (1 << 23) +#define RS600_MC_DATA 0x74 +#define RS600_MC_STATUS 0x0 +# define RS600_MC_IDLE (1 << 1) +#define RS600_MC_FB_LOCATION 0x4 +#define RS600_MC_FB_START_MASK 0x0000FFFF +#define RS600_MC_FB_START_SHIFT 0 +#define RS600_MC_FB_TOP_MASK 0xFFFF0000 +#define RS600_MC_FB_TOP_SHIFT 16 +#define RS600_MC_AGP_LOCATION 0x5 +#define RS600_MC_AGP_START_MASK 0x0000FFFF +#define RS600_MC_AGP_START_SHIFT 0 +#define RS600_MC_AGP_TOP_MASK 0xFFFF0000 +#define RS600_MC_AGP_TOP_SHIFT 16 +#define RS600_MC_AGP_BASE 0x6 +#define RS600_MC_AGP_BASE_2 0x7 +#define RS600_MC_CNTL1 0x9 +# define RS600_ENABLE_PAGE_TABLES (1 << 26) +#define RS600_MC_PT0_CNTL 0x100 +# define RS600_ENABLE_PT (1 << 0) +# define RS600_EFFECTIVE_L2_CACHE_SIZE(x) ((x) << 15) +# define RS600_EFFECTIVE_L2_QUEUE_SIZE(x) ((x) << 21) +# define RS600_INVALIDATE_ALL_L1_TLBS (1 << 28) +# define RS600_INVALIDATE_L2_CACHE (1 << 29) +#define RS600_MC_PT0_CONTEXT0_CNTL 0x102 +# define RS600_ENABLE_PAGE_TABLE (1 << 0) +# define RS600_PAGE_TABLE_TYPE_FLAT (0 << 1) +#define RS600_MC_PT0_SYSTEM_APERTURE_LOW_ADDR 0x112 +#define RS600_MC_PT0_SYSTEM_APERTURE_HIGH_ADDR 0x114 +#define RS600_MC_PT0_CONTEXT0_DEFAULT_READ_ADDR 0x11c +#define RS600_MC_PT0_CONTEXT0_FLAT_BASE_ADDR 0x12c +#define RS600_MC_PT0_CONTEXT0_FLAT_START_ADDR 0x13c +#define RS600_MC_PT0_CONTEXT0_FLAT_END_ADDR 0x14c +#define RS600_MC_PT0_CLIENT0_CNTL 0x16c +# define RS600_ENABLE_TRANSLATION_MODE_OVERRIDE (1 << 0) +# define RS600_TRANSLATION_MODE_OVERRIDE (1 << 1) +# define RS600_SYSTEM_ACCESS_MODE_MASK (3 << 8) +# define RS600_SYSTEM_ACCESS_MODE_PA_ONLY (0 << 8) +# define RS600_SYSTEM_ACCESS_MODE_USE_SYS_MAP (1 << 8) +# define RS600_SYSTEM_ACCESS_MODE_IN_SYS (2 << 8) +# define RS600_SYSTEM_ACCESS_MODE_NOT_IN_SYS (3 << 8) +# define RS600_SYSTEM_APERTURE_UNMAPPED_ACCESS_PASSTHROUGH (0 << 10) +# define RS600_SYSTEM_APERTURE_UNMAPPED_ACCESS_DEFAULT_PAGE (1 << 10) +# define RS600_EFFECTIVE_L1_CACHE_SIZE(x) ((x) << 11) +# define RS600_ENABLE_FRAGMENT_PROCESSING (1 << 14) +# define RS600_EFFECTIVE_L1_QUEUE_SIZE(x) ((x) << 15) +# define RS600_INVALIDATE_L1_TLB (1 << 20) +/* rs600/rs690/rs740 */ +# define RS600_BUS_MASTER_DIS (1 << 14) +# define RS600_MSI_REARM (1 << 20) +/* see RS400_MSI_REARM in AIC_CNTL for rs480 */ + + + +#define RV515_MC_FB_LOCATION 0x01 +#define RV515_MC_FB_START_MASK 0x0000FFFF +#define RV515_MC_FB_START_SHIFT 0 +#define RV515_MC_FB_TOP_MASK 0xFFFF0000 +#define RV515_MC_FB_TOP_SHIFT 16 +#define RV515_MC_AGP_LOCATION 0x02 +#define RV515_MC_AGP_START_MASK 0x0000FFFF +#define RV515_MC_AGP_START_SHIFT 0 +#define RV515_MC_AGP_TOP_MASK 0xFFFF0000 +#define RV515_MC_AGP_TOP_SHIFT 16 +#define RV515_MC_AGP_BASE 0x03 +#define RV515_MC_AGP_BASE_2 0x04 + +#define R520_MC_FB_LOCATION 0x04 +#define R520_MC_FB_START_MASK 0x0000FFFF +#define R520_MC_FB_START_SHIFT 0 +#define R520_MC_FB_TOP_MASK 0xFFFF0000 +#define R520_MC_FB_TOP_SHIFT 16 +#define R520_MC_AGP_LOCATION 0x05 +#define R520_MC_AGP_START_MASK 0x0000FFFF +#define R520_MC_AGP_START_SHIFT 0 +#define R520_MC_AGP_TOP_MASK 0xFFFF0000 +#define R520_MC_AGP_TOP_SHIFT 16 +#define R520_MC_AGP_BASE 0x06 +#define R520_MC_AGP_BASE_2 0x07 + + +#define AVIVO_MC_INDEX 0x0070 +#define R520_MC_STATUS 0x00 +#define R520_MC_STATUS_IDLE (1<<1) +#define RV515_MC_STATUS 0x08 +#define RV515_MC_STATUS_IDLE (1<<4) +#define RV515_MC_INIT_MISC_LAT_TIMER 0x09 +#define AVIVO_MC_DATA 0x0074 + +#define R520_MC_IND_INDEX 0x70 +#define R520_MC_IND_WR_EN (1 << 24) +#define R520_MC_IND_DATA 0x74 + +#define RV515_MC_CNTL 0x5 +# define RV515_MEM_NUM_CHANNELS_MASK 0x3 +#define R520_MC_CNTL0 0x8 +# define R520_MEM_NUM_CHANNELS_MASK (0x3 << 24) +# define R520_MEM_NUM_CHANNELS_SHIFT 24 +# define R520_MC_CHANNEL_SIZE (1 << 23) + +#define AVIVO_CP_DYN_CNTL 0x000f /* PLL */ +# define AVIVO_CP_FORCEON (1 << 0) +#define AVIVO_E2_DYN_CNTL 0x0011 /* PLL */ +# define AVIVO_E2_FORCEON (1 << 0) +#define AVIVO_IDCT_DYN_CNTL 0x0013 /* PLL */ +# define AVIVO_IDCT_FORCEON (1 << 0) + +#define AVIVO_HDP_FB_LOCATION 0x134 + +#define AVIVO_VGA_RENDER_CONTROL 0x0300 +# define AVIVO_VGA_VSTATUS_CNTL_MASK (3 << 16) +#define AVIVO_D1VGA_CONTROL 0x0330 +# define AVIVO_DVGA_CONTROL_MODE_ENABLE (1<<0) +# define AVIVO_DVGA_CONTROL_TIMING_SELECT (1<<8) +# define AVIVO_DVGA_CONTROL_SYNC_POLARITY_SELECT (1<<9) +# define AVIVO_DVGA_CONTROL_OVERSCAN_TIMING_SELECT (1<<10) +# define AVIVO_DVGA_CONTROL_OVERSCAN_COLOR_EN (1<<16) +# define AVIVO_DVGA_CONTROL_ROTATE (1<<24) +#define AVIVO_D2VGA_CONTROL 0x0338 + +#define AVIVO_EXT1_PPLL_REF_DIV_SRC 0x400 +#define AVIVO_EXT1_PPLL_REF_DIV 0x404 +#define AVIVO_EXT1_PPLL_UPDATE_LOCK 0x408 +#define AVIVO_EXT1_PPLL_UPDATE_CNTL 0x40c + +#define AVIVO_EXT2_PPLL_REF_DIV_SRC 0x410 +#define AVIVO_EXT2_PPLL_REF_DIV 0x414 +#define AVIVO_EXT2_PPLL_UPDATE_LOCK 0x418 +#define AVIVO_EXT2_PPLL_UPDATE_CNTL 0x41c + +#define AVIVO_EXT1_PPLL_FB_DIV 0x430 +#define AVIVO_EXT2_PPLL_FB_DIV 0x434 + +#define AVIVO_EXT1_PPLL_POST_DIV_SRC 0x438 +#define AVIVO_EXT1_PPLL_POST_DIV 0x43c + +#define AVIVO_EXT2_PPLL_POST_DIV_SRC 0x440 +#define AVIVO_EXT2_PPLL_POST_DIV 0x444 + +#define AVIVO_EXT1_PPLL_CNTL 0x448 +#define AVIVO_EXT2_PPLL_CNTL 0x44c + +#define AVIVO_P1PLL_CNTL 0x450 +#define AVIVO_P2PLL_CNTL 0x454 +#define AVIVO_P1PLL_INT_SS_CNTL 0x458 +#define AVIVO_P2PLL_INT_SS_CNTL 0x45c +#define AVIVO_P1PLL_TMDSA_CNTL 0x460 +#define AVIVO_P2PLL_LVTMA_CNTL 0x464 + +#define AVIVO_PCLK_CRTC1_CNTL 0x480 +#define AVIVO_PCLK_CRTC2_CNTL 0x484 + +#define AVIVO_D1CRTC_H_TOTAL 0x6000 +#define AVIVO_D1CRTC_H_BLANK_START_END 0x6004 +#define AVIVO_D1CRTC_H_SYNC_A 0x6008 +#define AVIVO_D1CRTC_H_SYNC_A_CNTL 0x600c +#define AVIVO_D1CRTC_H_SYNC_B 0x6010 +#define AVIVO_D1CRTC_H_SYNC_B_CNTL 0x6014 + +#define AVIVO_D1CRTC_V_TOTAL 0x6020 +#define AVIVO_D1CRTC_V_BLANK_START_END 0x6024 +#define AVIVO_D1CRTC_V_SYNC_A 0x6028 +#define AVIVO_D1CRTC_V_SYNC_A_CNTL 0x602c +#define AVIVO_D1CRTC_V_SYNC_B 0x6030 +#define AVIVO_D1CRTC_V_SYNC_B_CNTL 0x6034 + +#define AVIVO_D1CRTC_CONTROL 0x6080 +# define AVIVO_CRTC_EN (1 << 0) +# define AVIVO_CRTC_DISP_READ_REQUEST_DISABLE (1 << 24) +#define AVIVO_D1CRTC_BLANK_CONTROL 0x6084 +#define AVIVO_D1CRTC_INTERLACE_CONTROL 0x6088 +#define AVIVO_D1CRTC_INTERLACE_STATUS 0x608c +#define AVIVO_D1CRTC_STATUS_POSITION 0x60a0 +#define AVIVO_D1CRTC_FRAME_COUNT 0x60a4 +#define AVIVO_D1CRTC_STEREO_CONTROL 0x60c4 + +#define AVIVO_D1MODE_MASTER_UPDATE_MODE 0x60e4 + +/* master controls */ +#define AVIVO_DC_CRTC_MASTER_EN 0x60f8 +#define AVIVO_DC_CRTC_TV_CONTROL 0x60fc + +#define AVIVO_D1GRPH_ENABLE 0x6100 +#define AVIVO_D1GRPH_CONTROL 0x6104 +# define AVIVO_D1GRPH_CONTROL_DEPTH_8BPP (0 << 0) +# define AVIVO_D1GRPH_CONTROL_DEPTH_16BPP (1 << 0) +# define AVIVO_D1GRPH_CONTROL_DEPTH_32BPP (2 << 0) +# define AVIVO_D1GRPH_CONTROL_DEPTH_64BPP (3 << 0) + +# define AVIVO_D1GRPH_CONTROL_8BPP_INDEXED (0 << 8) + +# define AVIVO_D1GRPH_CONTROL_16BPP_ARGB1555 (0 << 8) +# define AVIVO_D1GRPH_CONTROL_16BPP_RGB565 (1 << 8) +# define AVIVO_D1GRPH_CONTROL_16BPP_ARGB4444 (2 << 8) +# define AVIVO_D1GRPH_CONTROL_16BPP_AI88 (3 << 8) +# define AVIVO_D1GRPH_CONTROL_16BPP_MONO16 (4 << 8) + +# define AVIVO_D1GRPH_CONTROL_32BPP_ARGB8888 (0 << 8) +# define AVIVO_D1GRPH_CONTROL_32BPP_ARGB2101010 (1 << 8) +# define AVIVO_D1GRPH_CONTROL_32BPP_DIGITAL (2 << 8) +# define AVIVO_D1GRPH_CONTROL_32BPP_8B_ARGB2101010 (3 << 8) + + +# define AVIVO_D1GRPH_CONTROL_64BPP_ARGB16161616 (0 << 8) + +# define AVIVO_D1GRPH_SWAP_RB (1 << 16) +# define AVIVO_D1GRPH_TILED (1 << 20) +# define AVIVO_D1GRPH_MACRO_ADDRESS_MODE (1 << 21) + +# define R600_D1GRPH_ARRAY_MODE_LINEAR_GENERAL (0 << 20) +# define R600_D1GRPH_ARRAY_MODE_LINEAR_ALIGNED (1 << 20) +# define R600_D1GRPH_ARRAY_MODE_1D_TILED_THIN1 (2 << 20) +# define R600_D1GRPH_ARRAY_MODE_2D_TILED_THIN1 (4 << 20) + +/* The R7xx *_HIGH surface regs are backwards; the D1 regs are in the D2 + * block and vice versa. This applies to GRPH, CUR, etc. + */ +#define AVIVO_D1GRPH_LUT_SEL 0x6108 +#define AVIVO_D1GRPH_PRIMARY_SURFACE_ADDRESS 0x6110 +#define R700_D1GRPH_PRIMARY_SURFACE_ADDRESS_HIGH 0x6914 +#define R700_D2GRPH_PRIMARY_SURFACE_ADDRESS_HIGH 0x6114 +#define AVIVO_D1GRPH_SECONDARY_SURFACE_ADDRESS 0x6118 +#define R700_D1GRPH_SECONDARY_SURFACE_ADDRESS_HIGH 0x691c +#define R700_D2GRPH_SECONDARY_SURFACE_ADDRESS_HIGH 0x611c +#define AVIVO_D1GRPH_PITCH 0x6120 +#define AVIVO_D1GRPH_SURFACE_OFFSET_X 0x6124 +#define AVIVO_D1GRPH_SURFACE_OFFSET_Y 0x6128 +#define AVIVO_D1GRPH_X_START 0x612c +#define AVIVO_D1GRPH_Y_START 0x6130 +#define AVIVO_D1GRPH_X_END 0x6134 +#define AVIVO_D1GRPH_Y_END 0x6138 +#define AVIVO_D1GRPH_UPDATE 0x6144 +# define AVIVO_D1GRPH_SURFACE_UPDATE_PENDING (1 << 2) +# define AVIVO_D1GRPH_UPDATE_LOCK (1 << 16) +#define AVIVO_D1GRPH_FLIP_CONTROL 0x6148 +# define AVIVO_D1GRPH_SURFACE_UPDATE_H_RETRACE_EN (1 << 0) + +#define AVIVO_D1CUR_CONTROL 0x6400 +# define AVIVO_D1CURSOR_EN (1 << 0) +# define AVIVO_D1CURSOR_MODE_SHIFT 8 +# define AVIVO_D1CURSOR_MODE_MASK (3 << 8) +# define AVIVO_D1CURSOR_MODE_24BPP 2 +#define AVIVO_D1CUR_SURFACE_ADDRESS 0x6408 +#define R700_D1CUR_SURFACE_ADDRESS_HIGH 0x6c0c +#define R700_D2CUR_SURFACE_ADDRESS_HIGH 0x640c +#define AVIVO_D1CUR_SIZE 0x6410 +#define AVIVO_D1CUR_POSITION 0x6414 +#define AVIVO_D1CUR_HOT_SPOT 0x6418 +#define AVIVO_D1CUR_UPDATE 0x6424 +# define AVIVO_D1CURSOR_UPDATE_LOCK (1 << 16) + +#define AVIVO_DC_LUT_RW_SELECT 0x6480 +#define AVIVO_DC_LUT_RW_MODE 0x6484 +#define AVIVO_DC_LUT_RW_INDEX 0x6488 +#define AVIVO_DC_LUT_SEQ_COLOR 0x648c +#define AVIVO_DC_LUT_PWL_DATA 0x6490 +#define AVIVO_DC_LUT_30_COLOR 0x6494 +#define AVIVO_DC_LUT_READ_PIPE_SELECT 0x6498 +#define AVIVO_DC_LUT_WRITE_EN_MASK 0x649c +#define AVIVO_DC_LUT_AUTOFILL 0x64a0 + +#define AVIVO_DC_LUTA_CONTROL 0x64c0 +#define AVIVO_DC_LUTA_BLACK_OFFSET_BLUE 0x64c4 +#define AVIVO_DC_LUTA_BLACK_OFFSET_GREEN 0x64c8 +#define AVIVO_DC_LUTA_BLACK_OFFSET_RED 0x64cc +#define AVIVO_DC_LUTA_WHITE_OFFSET_BLUE 0x64d0 +#define AVIVO_DC_LUTA_WHITE_OFFSET_GREEN 0x64d4 +#define AVIVO_DC_LUTA_WHITE_OFFSET_RED 0x64d8 + +#define AVIVO_DC_LB_MEMORY_SPLIT 0x6520 +# define AVIVO_DC_LB_MEMORY_SPLIT_MASK 0x3 +# define AVIVO_DC_LB_MEMORY_SPLIT_SHIFT 0 +# define AVIVO_DC_LB_MEMORY_SPLIT_D1HALF_D2HALF 0 +# define AVIVO_DC_LB_MEMORY_SPLIT_D1_3Q_D2_1Q 1 +# define AVIVO_DC_LB_MEMORY_SPLIT_D1_ONLY 2 +# define AVIVO_DC_LB_MEMORY_SPLIT_D1_1Q_D2_3Q 3 +# define AVIVO_DC_LB_MEMORY_SPLIT_SHIFT_MODE (1 << 2) +# define AVIVO_DC_LB_DISP1_END_ADR_SHIFT 4 +# define AVIVO_DC_LB_DISP1_END_ADR_MASK 0x7ff + +#define AVIVO_D1MODE_DATA_FORMAT 0x6528 +# define AVIVO_D1MODE_INTERLEAVE_EN (1 << 0) +#define AVIVO_D1MODE_DESKTOP_HEIGHT 0x652C +#define AVIVO_D1MODE_VBLANK_STATUS 0x6534 +# define AVIVO_VBLANK_ACK (1 << 4) +#define AVIVO_D1MODE_VLINE_START_END 0x6538 +#define AVIVO_D1MODE_VLINE_STATUS 0x653c +# define AVIVO_D1MODE_VLINE_STAT (1 << 12) +#define AVIVO_DxMODE_INT_MASK 0x6540 +# define AVIVO_D1MODE_INT_MASK (1 << 0) +# define AVIVO_D2MODE_INT_MASK (1 << 8) +#define AVIVO_D1MODE_VIEWPORT_START 0x6580 +#define AVIVO_D1MODE_VIEWPORT_SIZE 0x6584 +#define AVIVO_D1MODE_EXT_OVERSCAN_LEFT_RIGHT 0x6588 +#define AVIVO_D1MODE_EXT_OVERSCAN_TOP_BOTTOM 0x658c + +#define AVIVO_D1SCL_SCALER_ENABLE 0x6590 +#define AVIVO_D1SCL_SCALER_TAP_CONTROL 0x6594 +#define AVIVO_D1SCL_UPDATE 0x65cc +# define AVIVO_D1SCL_UPDATE_LOCK (1 << 16) + +/* second crtc */ +#define AVIVO_D2CRTC_H_TOTAL 0x6800 +#define AVIVO_D2CRTC_H_BLANK_START_END 0x6804 +#define AVIVO_D2CRTC_H_SYNC_A 0x6808 +#define AVIVO_D2CRTC_H_SYNC_A_CNTL 0x680c +#define AVIVO_D2CRTC_H_SYNC_B 0x6810 +#define AVIVO_D2CRTC_H_SYNC_B_CNTL 0x6814 + +#define AVIVO_D2CRTC_V_TOTAL 0x6820 +#define AVIVO_D2CRTC_V_BLANK_START_END 0x6824 +#define AVIVO_D2CRTC_V_SYNC_A 0x6828 +#define AVIVO_D2CRTC_V_SYNC_A_CNTL 0x682c +#define AVIVO_D2CRTC_V_SYNC_B 0x6830 +#define AVIVO_D2CRTC_V_SYNC_B_CNTL 0x6834 + +#define AVIVO_D2CRTC_CONTROL 0x6880 +#define AVIVO_D2CRTC_BLANK_CONTROL 0x6884 +#define AVIVO_D2CRTC_INTERLACE_CONTROL 0x6888 +#define AVIVO_D2CRTC_INTERLACE_STATUS 0x688c +#define AVIVO_D2CRTC_STATUS_POSITION 0x68a0 +#define AVIVO_D2CRTC_FRAME_COUNT 0x68a4 +#define AVIVO_D2CRTC_STEREO_CONTROL 0x68c4 + +#define AVIVO_D2GRPH_ENABLE 0x6900 +#define AVIVO_D2GRPH_CONTROL 0x6904 +#define AVIVO_D2GRPH_LUT_SEL 0x6908 +#define AVIVO_D2GRPH_PRIMARY_SURFACE_ADDRESS 0x6910 +#define AVIVO_D2GRPH_SECONDARY_SURFACE_ADDRESS 0x6918 +#define AVIVO_D2GRPH_PITCH 0x6920 +#define AVIVO_D2GRPH_SURFACE_OFFSET_X 0x6924 +#define AVIVO_D2GRPH_SURFACE_OFFSET_Y 0x6928 +#define AVIVO_D2GRPH_X_START 0x692c +#define AVIVO_D2GRPH_Y_START 0x6930 +#define AVIVO_D2GRPH_X_END 0x6934 +#define AVIVO_D2GRPH_Y_END 0x6938 +#define AVIVO_D2GRPH_UPDATE 0x6944 +#define AVIVO_D2GRPH_FLIP_CONTROL 0x6948 + +#define AVIVO_D2CUR_CONTROL 0x6c00 +#define AVIVO_D2CUR_SURFACE_ADDRESS 0x6c08 +#define AVIVO_D2CUR_SIZE 0x6c10 +#define AVIVO_D2CUR_POSITION 0x6c14 + +#define AVIVO_D2MODE_VBLANK_STATUS 0x6d34 +#define AVIVO_D2MODE_VLINE_START_END 0x6d38 +#define AVIVO_D2MODE_VLINE_STATUS 0x6d3c +#define AVIVO_D2MODE_VIEWPORT_START 0x6d80 +#define AVIVO_D2MODE_VIEWPORT_SIZE 0x6d84 +#define AVIVO_D2MODE_EXT_OVERSCAN_LEFT_RIGHT 0x6d88 +#define AVIVO_D2MODE_EXT_OVERSCAN_TOP_BOTTOM 0x6d8c + +#define AVIVO_D2SCL_SCALER_ENABLE 0x6d90 +#define AVIVO_D2SCL_SCALER_TAP_CONTROL 0x6d94 + +#define AVIVO_DDIA_BIT_DEPTH_CONTROL 0x7214 + +#define AVIVO_DACA_ENABLE 0x7800 +# define AVIVO_DAC_ENABLE (1 << 0) +#define AVIVO_DACA_SOURCE_SELECT 0x7804 +# define AVIVO_DAC_SOURCE_CRTC1 (0 << 0) +# define AVIVO_DAC_SOURCE_CRTC2 (1 << 0) +# define AVIVO_DAC_SOURCE_TV (2 << 0) + +#define AVIVO_DACA_FORCE_OUTPUT_CNTL 0x783c +# define AVIVO_DACA_FORCE_OUTPUT_CNTL_FORCE_DATA_EN (1 << 0) +# define AVIVO_DACA_FORCE_OUTPUT_CNTL_DATA_SEL_SHIFT (8) +# define AVIVO_DACA_FORCE_OUTPUT_CNTL_DATA_SEL_BLUE (1 << 0) +# define AVIVO_DACA_FORCE_OUTPUT_CNTL_DATA_SEL_GREEN (1 << 1) +# define AVIVO_DACA_FORCE_OUTPUT_CNTL_DATA_SEL_RED (1 << 2) +# define AVIVO_DACA_FORCE_OUTPUT_CNTL_DATA_ON_BLANKB_ONLY (1 << 24) +#define AVIVO_DACA_POWERDOWN 0x7850 +# define AVIVO_DACA_POWERDOWN_POWERDOWN (1 << 0) +# define AVIVO_DACA_POWERDOWN_BLUE (1 << 8) +# define AVIVO_DACA_POWERDOWN_GREEN (1 << 16) +# define AVIVO_DACA_POWERDOWN_RED (1 << 24) + +#define AVIVO_DACB_ENABLE 0x7a00 +#define AVIVO_DACB_SOURCE_SELECT 0x7a04 +#define AVIVO_DACB_FORCE_OUTPUT_CNTL 0x7a3c +# define AVIVO_DACB_FORCE_OUTPUT_CNTL_FORCE_DATA_EN (1 << 0) +# define AVIVO_DACB_FORCE_OUTPUT_CNTL_DATA_SEL_SHIFT (8) +# define AVIVO_DACB_FORCE_OUTPUT_CNTL_DATA_SEL_BLUE (1 << 0) +# define AVIVO_DACB_FORCE_OUTPUT_CNTL_DATA_SEL_GREEN (1 << 1) +# define AVIVO_DACB_FORCE_OUTPUT_CNTL_DATA_SEL_RED (1 << 2) +# define AVIVO_DACB_FORCE_OUTPUT_CNTL_DATA_ON_BLANKB_ONLY (1 << 24) +#define AVIVO_DACB_POWERDOWN 0x7a50 +# define AVIVO_DACB_POWERDOWN_POWERDOWN (1 << 0) +# define AVIVO_DACB_POWERDOWN_BLUE (1 << 8) +# define AVIVO_DACB_POWERDOWN_GREEN (1 << 16) +# define AVIVO_DACB_POWERDOWN_RED + +#define AVIVO_TMDSA_CNTL 0x7880 +# define AVIVO_TMDSA_CNTL_ENABLE (1 << 0) +# define AVIVO_TMDSA_CNTL_HPD_MASK (1 << 4) +# define AVIVO_TMDSA_CNTL_HPD_SELECT (1 << 8) +# define AVIVO_TMDSA_CNTL_SYNC_PHASE (1 << 12) +# define AVIVO_TMDSA_CNTL_PIXEL_ENCODING (1 << 16) +# define AVIVO_TMDSA_CNTL_DUAL_LINK_ENABLE (1 << 24) +# define AVIVO_TMDSA_CNTL_SWAP (1 << 28) +#define AVIVO_TMDSA_SOURCE_SELECT 0x7884 +/* 78a8 appears to be some kind of (reasonably tolerant) clock? + * 78d0 definitely hits the transmitter, definitely clock. */ +/* MYSTERY1 This appears to control dithering? */ +#define AVIVO_TMDSA_BIT_DEPTH_CONTROL 0x7894 +# define AVIVO_TMDS_BIT_DEPTH_CONTROL_TRUNCATE_EN (1 << 0) +# define AVIVO_TMDS_BIT_DEPTH_CONTROL_TRUNCATE_DEPTH (1 << 4) +# define AVIVO_TMDS_BIT_DEPTH_CONTROL_SPATIAL_DITHER_EN (1 << 8) +# define AVIVO_TMDS_BIT_DEPTH_CONTROL_SPATIAL_DITHER_DEPTH (1 << 12) +# define AVIVO_TMDS_BIT_DEPTH_CONTROL_TEMPORAL_DITHER_EN (1 << 16) +# define AVIVO_TMDS_BIT_DEPTH_CONTROL_TEMPORAL_DITHER_DEPTH (1 << 20) +# define AVIVO_TMDS_BIT_DEPTH_CONTROL_TEMPORAL_LEVEL (1 << 24) +# define AVIVO_TMDS_BIT_DEPTH_CONTROL_TEMPORAL_DITHER_RESET (1 << 26) +#define AVIVO_TMDSA_DCBALANCER_CONTROL 0x78d0 +# define AVIVO_TMDSA_DCBALANCER_CONTROL_EN (1 << 0) +# define AVIVO_TMDSA_DCBALANCER_CONTROL_TEST_EN (1 << 8) +# define AVIVO_TMDSA_DCBALANCER_CONTROL_TEST_IN_SHIFT (16) +# define AVIVO_TMDSA_DCBALANCER_CONTROL_FORCE (1 << 24) +#define AVIVO_TMDSA_DATA_SYNCHRONIZATION 0x78d8 +# define AVIVO_TMDSA_DATA_SYNCHRONIZATION_DSYNSEL (1 << 0) +# define AVIVO_TMDSA_DATA_SYNCHRONIZATION_PFREQCHG (1 << 8) +#define AVIVO_TMDSA_CLOCK_ENABLE 0x7900 +#define AVIVO_TMDSA_TRANSMITTER_ENABLE 0x7904 +# define AVIVO_TMDSA_TRANSMITTER_ENABLE_TX0_ENABLE (1 << 0) +# define AVIVO_TMDSA_TRANSMITTER_ENABLE_LNKC0EN (1 << 1) +# define AVIVO_TMDSA_TRANSMITTER_ENABLE_LNKD00EN (1 << 2) +# define AVIVO_TMDSA_TRANSMITTER_ENABLE_LNKD01EN (1 << 3) +# define AVIVO_TMDSA_TRANSMITTER_ENABLE_LNKD02EN (1 << 4) +# define AVIVO_TMDSA_TRANSMITTER_ENABLE_TX1_ENABLE (1 << 8) +# define AVIVO_TMDSA_TRANSMITTER_ENABLE_LNKD10EN (1 << 10) +# define AVIVO_TMDSA_TRANSMITTER_ENABLE_LNKD11EN (1 << 11) +# define AVIVO_TMDSA_TRANSMITTER_ENABLE_LNKD12EN (1 << 12) +# define AVIVO_TMDSA_TRANSMITTER_ENABLE_TX_ENABLE_HPD_MASK (1 << 16) +# define AVIVO_TMDSA_TRANSMITTER_ENABLE_LNKCEN_HPD_MASK (1 << 17) +# define AVIVO_TMDSA_TRANSMITTER_ENABLE_LNKDEN_HPD_MASK (1 << 18) + +#define AVIVO_TMDSA_TRANSMITTER_CONTROL 0x7910 +# define AVIVO_TMDSA_TRANSMITTER_CONTROL_PLL_ENABLE (1 << 0) +# define AVIVO_TMDSA_TRANSMITTER_CONTROL_PLL_RESET (1 << 1) +# define AVIVO_TMDSA_TRANSMITTER_CONTROL_PLL_HPD_MASK_SHIFT (2) +# define AVIVO_TMDSA_TRANSMITTER_CONTROL_IDSCKSEL (1 << 4) +# define AVIVO_TMDSA_TRANSMITTER_CONTROL_BGSLEEP (1 << 5) +# define AVIVO_TMDSA_TRANSMITTER_CONTROL_PLL_PWRUP_SEQ_EN (1 << 6) +# define AVIVO_TMDSA_TRANSMITTER_CONTROL_TMCLK (1 << 8) +# define AVIVO_TMDSA_TRANSMITTER_CONTROL_TMCLK_FROM_PADS (1 << 13) +# define AVIVO_TMDSA_TRANSMITTER_CONTROL_TDCLK (1 << 14) +# define AVIVO_TMDSA_TRANSMITTER_CONTROL_TDCLK_FROM_PADS (1 << 15) +# define AVIVO_TMDSA_TRANSMITTER_CONTROL_CLK_PATTERN_SHIFT (16) +# define AVIVO_TMDSA_TRANSMITTER_CONTROL_BYPASS_PLL (1 << 28) +# define AVIVO_TMDSA_TRANSMITTER_CONTROL_USE_CLK_DATA (1 << 29) +# define AVIVO_TMDSA_TRANSMITTER_CONTROL_INPUT_TEST_CLK_SEL (1 << 31) + +#define AVIVO_LVTMA_CNTL 0x7a80 +# define AVIVO_LVTMA_CNTL_ENABLE (1 << 0) +# define AVIVO_LVTMA_CNTL_HPD_MASK (1 << 4) +# define AVIVO_LVTMA_CNTL_HPD_SELECT (1 << 8) +# define AVIVO_LVTMA_CNTL_SYNC_PHASE (1 << 12) +# define AVIVO_LVTMA_CNTL_PIXEL_ENCODING (1 << 16) +# define AVIVO_LVTMA_CNTL_DUAL_LINK_ENABLE (1 << 24) +# define AVIVO_LVTMA_CNTL_SWAP (1 << 28) +#define AVIVO_LVTMA_SOURCE_SELECT 0x7a84 +#define AVIVO_LVTMA_COLOR_FORMAT 0x7a88 +#define AVIVO_LVTMA_BIT_DEPTH_CONTROL 0x7a94 +# define AVIVO_LVTMA_BIT_DEPTH_CONTROL_TRUNCATE_EN (1 << 0) +# define AVIVO_LVTMA_BIT_DEPTH_CONTROL_TRUNCATE_DEPTH (1 << 4) +# define AVIVO_LVTMA_BIT_DEPTH_CONTROL_SPATIAL_DITHER_EN (1 << 8) +# define AVIVO_LVTMA_BIT_DEPTH_CONTROL_SPATIAL_DITHER_DEPTH (1 << 12) +# define AVIVO_LVTMA_BIT_DEPTH_CONTROL_TEMPORAL_DITHER_EN (1 << 16) +# define AVIVO_LVTMA_BIT_DEPTH_CONTROL_TEMPORAL_DITHER_DEPTH (1 << 20) +# define AVIVO_LVTMA_BIT_DEPTH_CONTROL_TEMPORAL_LEVEL (1 << 24) +# define AVIVO_LVTMA_BIT_DEPTH_CONTROL_TEMPORAL_DITHER_RESET (1 << 26) + + + +#define AVIVO_LVTMA_DCBALANCER_CONTROL 0x7ad0 +# define AVIVO_LVTMA_DCBALANCER_CONTROL_EN (1 << 0) +# define AVIVO_LVTMA_DCBALANCER_CONTROL_TEST_EN (1 << 8) +# define AVIVO_LVTMA_DCBALANCER_CONTROL_TEST_IN_SHIFT (16) +# define AVIVO_LVTMA_DCBALANCER_CONTROL_FORCE (1 << 24) + +#define AVIVO_LVTMA_DATA_SYNCHRONIZATION 0x78d8 +# define AVIVO_LVTMA_DATA_SYNCHRONIZATION_DSYNSEL (1 << 0) +# define AVIVO_LVTMA_DATA_SYNCHRONIZATION_PFREQCHG (1 << 8) +#define R500_LVTMA_CLOCK_ENABLE 0x7b00 +#define R600_LVTMA_CLOCK_ENABLE 0x7b04 + +#define R500_LVTMA_TRANSMITTER_ENABLE 0x7b04 +#define R600_LVTMA_TRANSMITTER_ENABLE 0x7b08 +# define AVIVO_LVTMA_TRANSMITTER_ENABLE_LNKC0EN (1 << 1) +# define AVIVO_LVTMA_TRANSMITTER_ENABLE_LNKD00EN (1 << 2) +# define AVIVO_LVTMA_TRANSMITTER_ENABLE_LNKD01EN (1 << 3) +# define AVIVO_LVTMA_TRANSMITTER_ENABLE_LNKD02EN (1 << 4) +# define AVIVO_LVTMA_TRANSMITTER_ENABLE_LNKD03EN (1 << 5) +# define AVIVO_LVTMA_TRANSMITTER_ENABLE_LNKC1EN (1 << 9) +# define AVIVO_LVTMA_TRANSMITTER_ENABLE_LNKD10EN (1 << 10) +# define AVIVO_LVTMA_TRANSMITTER_ENABLE_LNKD11EN (1 << 11) +# define AVIVO_LVTMA_TRANSMITTER_ENABLE_LNKD12EN (1 << 12) +# define AVIVO_LVTMA_TRANSMITTER_ENABLE_LNKCEN_HPD_MASK (1 << 17) +# define AVIVO_LVTMA_TRANSMITTER_ENABLE_LNKDEN_HPD_MASK (1 << 18) + +#define R500_LVTMA_TRANSMITTER_CONTROL 0x7b10 +#define R600_LVTMA_TRANSMITTER_CONTROL 0x7b14 +# define AVIVO_LVTMA_TRANSMITTER_CONTROL_PLL_ENABLE (1 << 0) +# define AVIVO_LVTMA_TRANSMITTER_CONTROL_PLL_RESET (1 << 1) +# define AVIVO_LVTMA_TRANSMITTER_CONTROL_PLL_HPD_MASK_SHIFT (2) +# define AVIVO_LVTMA_TRANSMITTER_CONTROL_IDSCKSEL (1 << 4) +# define AVIVO_LVTMA_TRANSMITTER_CONTROL_BGSLEEP (1 << 5) +# define AVIVO_LVTMA_TRANSMITTER_CONTROL_PLL_PWRUP_SEQ_EN (1 << 6) +# define AVIVO_LVTMA_TRANSMITTER_CONTROL_TMCLK (1 << 8) +# define AVIVO_LVTMA_TRANSMITTER_CONTROL_TMCLK_FROM_PADS (1 << 13) +# define AVIVO_LVTMA_TRANSMITTER_CONTROL_TDCLK (1 << 14) +# define AVIVO_LVTMA_TRANSMITTER_CONTROL_TDCLK_FROM_PADS (1 << 15) +# define AVIVO_LVTMA_TRANSMITTER_CONTROL_CLK_PATTERN_SHIFT (16) +# define AVIVO_LVTMA_TRANSMITTER_CONTROL_BYPASS_PLL (1 << 28) +# define AVIVO_LVTMA_TRANSMITTER_CONTROL_USE_CLK_DATA (1 << 29) +# define AVIVO_LVTMA_TRANSMITTER_CONTROL_INPUT_TEST_CLK_SEL (1 << 31) + +#define R500_LVTMA_PWRSEQ_CNTL 0x7af0 +#define R600_LVTMA_PWRSEQ_CNTL 0x7af4 +# define AVIVO_LVTMA_PWRSEQ_EN (1 << 0) +# define AVIVO_LVTMA_PWRSEQ_PLL_ENABLE_MASK (1 << 2) +# define AVIVO_LVTMA_PWRSEQ_PLL_RESET_MASK (1 << 3) +# define AVIVO_LVTMA_PWRSEQ_TARGET_STATE (1 << 4) +# define AVIVO_LVTMA_SYNCEN (1 << 8) +# define AVIVO_LVTMA_SYNCEN_OVRD (1 << 9) +# define AVIVO_LVTMA_SYNCEN_POL (1 << 10) +# define AVIVO_LVTMA_DIGON (1 << 16) +# define AVIVO_LVTMA_DIGON_OVRD (1 << 17) +# define AVIVO_LVTMA_DIGON_POL (1 << 18) +# define AVIVO_LVTMA_BLON (1 << 24) +# define AVIVO_LVTMA_BLON_OVRD (1 << 25) +# define AVIVO_LVTMA_BLON_POL (1 << 26) + +#define R500_LVTMA_PWRSEQ_STATE 0x7af4 +#define R600_LVTMA_PWRSEQ_STATE 0x7af8 +# define AVIVO_LVTMA_PWRSEQ_STATE_TARGET_STATE_R (1 << 0) +# define AVIVO_LVTMA_PWRSEQ_STATE_DIGON (1 << 1) +# define AVIVO_LVTMA_PWRSEQ_STATE_SYNCEN (1 << 2) +# define AVIVO_LVTMA_PWRSEQ_STATE_BLON (1 << 3) +# define AVIVO_LVTMA_PWRSEQ_STATE_DONE (1 << 4) +# define AVIVO_LVTMA_PWRSEQ_STATE_STATUS_SHIFT (8) + +#define AVIVO_LVDS_BACKLIGHT_CNTL 0x7af8 +# define AVIVO_LVDS_BACKLIGHT_CNTL_EN (1 << 0) +# define AVIVO_LVDS_BACKLIGHT_LEVEL_MASK 0x0000ff00 +# define AVIVO_LVDS_BACKLIGHT_LEVEL_SHIFT 8 + +#define AVIVO_DVOA_BIT_DEPTH_CONTROL 0x7988 + +#define AVIVO_DC_GPIO_HPD_A 0x7e94 +#define AVIVO_DC_GPIO_HPD_Y 0x7e9c + +#define AVIVO_DC_I2C_STATUS1 0x7d30 +# define AVIVO_DC_I2C_DONE (1 << 0) +# define AVIVO_DC_I2C_NACK (1 << 1) +# define AVIVO_DC_I2C_HALT (1 << 2) +# define AVIVO_DC_I2C_GO (1 << 3) +#define AVIVO_DC_I2C_RESET 0x7d34 +# define AVIVO_DC_I2C_SOFT_RESET (1 << 0) +# define AVIVO_DC_I2C_ABORT (1 << 8) +#define AVIVO_DC_I2C_CONTROL1 0x7d38 +# define AVIVO_DC_I2C_START (1 << 0) +# define AVIVO_DC_I2C_STOP (1 << 1) +# define AVIVO_DC_I2C_RECEIVE (1 << 2) +# define AVIVO_DC_I2C_EN (1 << 8) +# define AVIVO_DC_I2C_PIN_SELECT(x) ((x) << 16) +# define AVIVO_SEL_DDC1 0 +# define AVIVO_SEL_DDC2 1 +# define AVIVO_SEL_DDC3 2 +#define AVIVO_DC_I2C_CONTROL2 0x7d3c +# define AVIVO_DC_I2C_ADDR_COUNT(x) ((x) << 0) +# define AVIVO_DC_I2C_DATA_COUNT(x) ((x) << 8) +#define AVIVO_DC_I2C_CONTROL3 0x7d40 +# define AVIVO_DC_I2C_DATA_DRIVE_EN (1 << 0) +# define AVIVO_DC_I2C_DATA_DRIVE_SEL (1 << 1) +# define AVIVO_DC_I2C_CLK_DRIVE_EN (1 << 7) +# define AVIVO_DC_I2C_RD_INTRA_BYTE_DELAY(x) ((x) << 8) +# define AVIVO_DC_I2C_WR_INTRA_BYTE_DELAY(x) ((x) << 16) +# define AVIVO_DC_I2C_TIME_LIMIT(x) ((x) << 24) +#define AVIVO_DC_I2C_DATA 0x7d44 +#define AVIVO_DC_I2C_INTERRUPT_CONTROL 0x7d48 +# define AVIVO_DC_I2C_INTERRUPT_STATUS (1 << 0) +# define AVIVO_DC_I2C_INTERRUPT_AK (1 << 8) +# define AVIVO_DC_I2C_INTERRUPT_ENABLE (1 << 16) +#define AVIVO_DC_I2C_ARBITRATION 0x7d50 +# define AVIVO_DC_I2C_SW_WANTS_TO_USE_I2C (1 << 0) +# define AVIVO_DC_I2C_SW_CAN_USE_I2C (1 << 1) +# define AVIVO_DC_I2C_SW_DONE_USING_I2C (1 << 8) +# define AVIVO_DC_I2C_HW_NEEDS_I2C (1 << 9) +# define AVIVO_DC_I2C_ABORT_HDCP_I2C (1 << 16) +# define AVIVO_DC_I2C_HW_USING_I2C (1 << 17) + +#define AVIVO_DC_GPIO_DDC1_MASK 0x7e40 +#define AVIVO_DC_GPIO_DDC1_A 0x7e44 +#define AVIVO_DC_GPIO_DDC1_EN 0x7e48 +#define AVIVO_DC_GPIO_DDC1_Y 0x7e4c + +#define AVIVO_DC_GPIO_DDC2_MASK 0x7e50 +#define AVIVO_DC_GPIO_DDC2_A 0x7e54 +#define AVIVO_DC_GPIO_DDC2_EN 0x7e58 +#define AVIVO_DC_GPIO_DDC2_Y 0x7e5c + +#define AVIVO_DC_GPIO_DDC3_MASK 0x7e60 +#define AVIVO_DC_GPIO_DDC3_A 0x7e64 +#define AVIVO_DC_GPIO_DDC3_EN 0x7e68 +#define AVIVO_DC_GPIO_DDC3_Y 0x7e6c + +#define AVIVO_DISP_INTERRUPT_STATUS 0x7edc +# define AVIVO_D1_VBLANK_INTERRUPT (1 << 4) +# define AVIVO_D2_VBLANK_INTERRUPT (1 << 5) + +#endif diff --git a/drivers/gpu/drm/radeon/r520.c b/drivers/gpu/drm/radeon/r520.c new file mode 100644 index 00000000..3081d07f --- /dev/null +++ b/drivers/gpu/drm/radeon/r520.c @@ -0,0 +1,309 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#include "drmP.h" +#include "radeon.h" +#include "radeon_asic.h" +#include "atom.h" +#include "r520d.h" + +/* This files gather functions specifics to: r520,rv530,rv560,rv570,r580 */ + +static int r520_mc_wait_for_idle(struct radeon_device *rdev) +{ + unsigned i; + uint32_t tmp; + + for (i = 0; i < rdev->usec_timeout; i++) { + /* read MC_STATUS */ + tmp = RREG32_MC(R520_MC_STATUS); + if (tmp & R520_MC_STATUS_IDLE) { + return 0; + } + DRM_UDELAY(1); + } + return -1; +} + +static void r520_gpu_init(struct radeon_device *rdev) +{ + unsigned pipe_select_current, gb_pipe_select, tmp; + + rv515_vga_render_disable(rdev); + /* + * DST_PIPE_CONFIG 0x170C + * GB_TILE_CONFIG 0x4018 + * GB_FIFO_SIZE 0x4024 + * GB_PIPE_SELECT 0x402C + * GB_PIPE_SELECT2 0x4124 + * Z_PIPE_SHIFT 0 + * Z_PIPE_MASK 0x000000003 + * GB_FIFO_SIZE2 0x4128 + * SC_SFIFO_SIZE_SHIFT 0 + * SC_SFIFO_SIZE_MASK 0x000000003 + * SC_MFIFO_SIZE_SHIFT 2 + * SC_MFIFO_SIZE_MASK 0x00000000C + * FG_SFIFO_SIZE_SHIFT 4 + * FG_SFIFO_SIZE_MASK 0x000000030 + * ZB_MFIFO_SIZE_SHIFT 6 + * ZB_MFIFO_SIZE_MASK 0x0000000C0 + * GA_ENHANCE 0x4274 + * SU_REG_DEST 0x42C8 + */ + /* workaround for RV530 */ + if (rdev->family == CHIP_RV530) { + WREG32(0x4128, 0xFF); + } + r420_pipes_init(rdev); + gb_pipe_select = RREG32(R400_GB_PIPE_SELECT); + tmp = RREG32(R300_DST_PIPE_CONFIG); + pipe_select_current = (tmp >> 2) & 3; + tmp = (1 << pipe_select_current) | + (((gb_pipe_select >> 8) & 0xF) << 4); + WREG32_PLL(0x000D, tmp); + if (r520_mc_wait_for_idle(rdev)) { + printk(KERN_WARNING "Failed to wait MC idle while " + "programming pipes. Bad things might happen.\n"); + } +} + +static void r520_vram_get_type(struct radeon_device *rdev) +{ + uint32_t tmp; + + rdev->mc.vram_width = 128; + rdev->mc.vram_is_ddr = true; + tmp = RREG32_MC(R520_MC_CNTL0); + switch ((tmp & R520_MEM_NUM_CHANNELS_MASK) >> R520_MEM_NUM_CHANNELS_SHIFT) { + case 0: + rdev->mc.vram_width = 32; + break; + case 1: + rdev->mc.vram_width = 64; + break; + case 2: + rdev->mc.vram_width = 128; + break; + case 3: + rdev->mc.vram_width = 256; + break; + default: + rdev->mc.vram_width = 128; + break; + } + if (tmp & R520_MC_CHANNEL_SIZE) + rdev->mc.vram_width *= 2; +} + +void r520_mc_init(struct radeon_device *rdev) +{ + + r520_vram_get_type(rdev); + r100_vram_init_sizes(rdev); + radeon_vram_location(rdev, &rdev->mc, 0); + rdev->mc.gtt_base_align = 0; + if (!(rdev->flags & RADEON_IS_AGP)) + radeon_gtt_location(rdev, &rdev->mc); + radeon_update_bandwidth_info(rdev); +} + +void r520_mc_program(struct radeon_device *rdev) +{ + struct rv515_mc_save save; + + /* Stops all mc clients */ + rv515_mc_stop(rdev, &save); + + /* Wait for mc idle */ + if (r520_mc_wait_for_idle(rdev)) + dev_warn(rdev->dev, "Wait MC idle timeout before updating MC.\n"); + /* Write VRAM size in case we are limiting it */ + WREG32(R_0000F8_CONFIG_MEMSIZE, rdev->mc.real_vram_size); + /* Program MC, should be a 32bits limited address space */ + WREG32_MC(R_000004_MC_FB_LOCATION, + S_000004_MC_FB_START(rdev->mc.vram_start >> 16) | + S_000004_MC_FB_TOP(rdev->mc.vram_end >> 16)); + WREG32(R_000134_HDP_FB_LOCATION, + S_000134_HDP_FB_START(rdev->mc.vram_start >> 16)); + if (rdev->flags & RADEON_IS_AGP) { + WREG32_MC(R_000005_MC_AGP_LOCATION, + S_000005_MC_AGP_START(rdev->mc.gtt_start >> 16) | + S_000005_MC_AGP_TOP(rdev->mc.gtt_end >> 16)); + WREG32_MC(R_000006_AGP_BASE, lower_32_bits(rdev->mc.agp_base)); + WREG32_MC(R_000007_AGP_BASE_2, + S_000007_AGP_BASE_ADDR_2(upper_32_bits(rdev->mc.agp_base))); + } else { + WREG32_MC(R_000005_MC_AGP_LOCATION, 0xFFFFFFFF); + WREG32_MC(R_000006_AGP_BASE, 0); + WREG32_MC(R_000007_AGP_BASE_2, 0); + } + + rv515_mc_resume(rdev, &save); +} + +static int r520_startup(struct radeon_device *rdev) +{ + int r; + + r520_mc_program(rdev); + /* Resume clock */ + rv515_clock_startup(rdev); + /* Initialize GPU configuration (# pipes, ...) */ + r520_gpu_init(rdev); + /* Initialize GART (initialize after TTM so we can allocate + * memory through TTM but finalize after TTM) */ + if (rdev->flags & RADEON_IS_PCIE) { + r = rv370_pcie_gart_enable(rdev); + if (r) + return r; + } + + /* allocate wb buffer */ + r = radeon_wb_init(rdev); + if (r) + return r; + + /* Enable IRQ */ + rs600_irq_set(rdev); + rdev->config.r300.hdp_cntl = RREG32(RADEON_HOST_PATH_CNTL); + /* 1M ring buffer */ + r = r100_cp_init(rdev, 1024 * 1024); + if (r) { + dev_err(rdev->dev, "failed initializing CP (%d).\n", r); + return r; + } + r = r100_ib_init(rdev); + if (r) { + dev_err(rdev->dev, "failed initializing IB (%d).\n", r); + return r; + } + return 0; +} + +int r520_resume(struct radeon_device *rdev) +{ + /* Make sur GART are not working */ + if (rdev->flags & RADEON_IS_PCIE) + rv370_pcie_gart_disable(rdev); + /* Resume clock before doing reset */ + rv515_clock_startup(rdev); + /* Reset gpu before posting otherwise ATOM will enter infinite loop */ + if (radeon_asic_reset(rdev)) { + dev_warn(rdev->dev, "GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n", + RREG32(R_000E40_RBBM_STATUS), + RREG32(R_0007C0_CP_STAT)); + } + /* post */ + atom_asic_init(rdev->mode_info.atom_context); + /* Resume clock after posting */ + rv515_clock_startup(rdev); + /* Initialize surface registers */ + radeon_surface_init(rdev); + return r520_startup(rdev); +} + +int r520_init(struct radeon_device *rdev) +{ + int r; + + /* Initialize scratch registers */ + radeon_scratch_init(rdev); + /* Initialize surface registers */ + radeon_surface_init(rdev); + /* restore some register to sane defaults */ + r100_restore_sanity(rdev); + /* TODO: disable VGA need to use VGA request */ + /* BIOS*/ + if (!radeon_get_bios(rdev)) { + if (ASIC_IS_AVIVO(rdev)) + return -EINVAL; + } + if (rdev->is_atom_bios) { + r = radeon_atombios_init(rdev); + if (r) + return r; + } else { + dev_err(rdev->dev, "Expecting atombios for RV515 GPU\n"); + return -EINVAL; + } + /* Reset gpu before posting otherwise ATOM will enter infinite loop */ + if (radeon_asic_reset(rdev)) { + dev_warn(rdev->dev, + "GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n", + RREG32(R_000E40_RBBM_STATUS), + RREG32(R_0007C0_CP_STAT)); + } + /* check if cards are posted or not */ + if (radeon_boot_test_post_card(rdev) == false) + return -EINVAL; + + if (!radeon_card_posted(rdev) && rdev->bios) { + DRM_INFO("GPU not posted. posting now...\n"); + atom_asic_init(rdev->mode_info.atom_context); + } + /* Initialize clocks */ + radeon_get_clock_info(rdev->ddev); + /* initialize AGP */ + if (rdev->flags & RADEON_IS_AGP) { + r = radeon_agp_init(rdev); + if (r) { + radeon_agp_disable(rdev); + } + } + /* initialize memory controller */ + r520_mc_init(rdev); + rv515_debugfs(rdev); + /* Fence driver */ + r = radeon_fence_driver_init(rdev); + if (r) + return r; + r = radeon_irq_kms_init(rdev); + if (r) + return r; + /* Memory manager */ + r = radeon_bo_init(rdev); + if (r) + return r; + r = rv370_pcie_gart_init(rdev); + if (r) + return r; + rv515_set_safe_registers(rdev); + rdev->accel_working = true; + r = r520_startup(rdev); + if (r) { + /* Somethings want wront with the accel init stop accel */ + dev_err(rdev->dev, "Disabling GPU acceleration\n"); + r100_cp_fini(rdev); + radeon_wb_fini(rdev); + r100_ib_fini(rdev); + radeon_irq_kms_fini(rdev); + rv370_pcie_gart_fini(rdev); + radeon_agp_fini(rdev); + rdev->accel_working = false; + } + return 0; +} diff --git a/drivers/gpu/drm/radeon/r520d.h b/drivers/gpu/drm/radeon/r520d.h new file mode 100644 index 00000000..61af61f6 --- /dev/null +++ b/drivers/gpu/drm/radeon/r520d.h @@ -0,0 +1,187 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#ifndef __R520D_H__ +#define __R520D_H__ + +/* Registers */ +#define R_0000F8_CONFIG_MEMSIZE 0x0000F8 +#define S_0000F8_CONFIG_MEMSIZE(x) (((x) & 0xFFFFFFFF) << 0) +#define G_0000F8_CONFIG_MEMSIZE(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_0000F8_CONFIG_MEMSIZE 0x00000000 +#define R_000134_HDP_FB_LOCATION 0x000134 +#define S_000134_HDP_FB_START(x) (((x) & 0xFFFF) << 0) +#define G_000134_HDP_FB_START(x) (((x) >> 0) & 0xFFFF) +#define C_000134_HDP_FB_START 0xFFFF0000 +#define R_0007C0_CP_STAT 0x0007C0 +#define S_0007C0_MRU_BUSY(x) (((x) & 0x1) << 0) +#define G_0007C0_MRU_BUSY(x) (((x) >> 0) & 0x1) +#define C_0007C0_MRU_BUSY 0xFFFFFFFE +#define S_0007C0_MWU_BUSY(x) (((x) & 0x1) << 1) +#define G_0007C0_MWU_BUSY(x) (((x) >> 1) & 0x1) +#define C_0007C0_MWU_BUSY 0xFFFFFFFD +#define S_0007C0_RSIU_BUSY(x) (((x) & 0x1) << 2) +#define G_0007C0_RSIU_BUSY(x) (((x) >> 2) & 0x1) +#define C_0007C0_RSIU_BUSY 0xFFFFFFFB +#define S_0007C0_RCIU_BUSY(x) (((x) & 0x1) << 3) +#define G_0007C0_RCIU_BUSY(x) (((x) >> 3) & 0x1) +#define C_0007C0_RCIU_BUSY 0xFFFFFFF7 +#define S_0007C0_CSF_PRIMARY_BUSY(x) (((x) & 0x1) << 9) +#define G_0007C0_CSF_PRIMARY_BUSY(x) (((x) >> 9) & 0x1) +#define C_0007C0_CSF_PRIMARY_BUSY 0xFFFFFDFF +#define S_0007C0_CSF_INDIRECT_BUSY(x) (((x) & 0x1) << 10) +#define G_0007C0_CSF_INDIRECT_BUSY(x) (((x) >> 10) & 0x1) +#define C_0007C0_CSF_INDIRECT_BUSY 0xFFFFFBFF +#define S_0007C0_CSQ_PRIMARY_BUSY(x) (((x) & 0x1) << 11) +#define G_0007C0_CSQ_PRIMARY_BUSY(x) (((x) >> 11) & 0x1) +#define C_0007C0_CSQ_PRIMARY_BUSY 0xFFFFF7FF +#define S_0007C0_CSQ_INDIRECT_BUSY(x) (((x) & 0x1) << 12) +#define G_0007C0_CSQ_INDIRECT_BUSY(x) (((x) >> 12) & 0x1) +#define C_0007C0_CSQ_INDIRECT_BUSY 0xFFFFEFFF +#define S_0007C0_CSI_BUSY(x) (((x) & 0x1) << 13) +#define G_0007C0_CSI_BUSY(x) (((x) >> 13) & 0x1) +#define C_0007C0_CSI_BUSY 0xFFFFDFFF +#define S_0007C0_CSF_INDIRECT2_BUSY(x) (((x) & 0x1) << 14) +#define G_0007C0_CSF_INDIRECT2_BUSY(x) (((x) >> 14) & 0x1) +#define C_0007C0_CSF_INDIRECT2_BUSY 0xFFFFBFFF +#define S_0007C0_CSQ_INDIRECT2_BUSY(x) (((x) & 0x1) << 15) +#define G_0007C0_CSQ_INDIRECT2_BUSY(x) (((x) >> 15) & 0x1) +#define C_0007C0_CSQ_INDIRECT2_BUSY 0xFFFF7FFF +#define S_0007C0_GUIDMA_BUSY(x) (((x) & 0x1) << 28) +#define G_0007C0_GUIDMA_BUSY(x) (((x) >> 28) & 0x1) +#define C_0007C0_GUIDMA_BUSY 0xEFFFFFFF +#define S_0007C0_VIDDMA_BUSY(x) (((x) & 0x1) << 29) +#define G_0007C0_VIDDMA_BUSY(x) (((x) >> 29) & 0x1) +#define C_0007C0_VIDDMA_BUSY 0xDFFFFFFF +#define S_0007C0_CMDSTRM_BUSY(x) (((x) & 0x1) << 30) +#define G_0007C0_CMDSTRM_BUSY(x) (((x) >> 30) & 0x1) +#define C_0007C0_CMDSTRM_BUSY 0xBFFFFFFF +#define S_0007C0_CP_BUSY(x) (((x) & 0x1) << 31) +#define G_0007C0_CP_BUSY(x) (((x) >> 31) & 0x1) +#define C_0007C0_CP_BUSY 0x7FFFFFFF +#define R_000E40_RBBM_STATUS 0x000E40 +#define S_000E40_CMDFIFO_AVAIL(x) (((x) & 0x7F) << 0) +#define G_000E40_CMDFIFO_AVAIL(x) (((x) >> 0) & 0x7F) +#define C_000E40_CMDFIFO_AVAIL 0xFFFFFF80 +#define S_000E40_HIRQ_ON_RBB(x) (((x) & 0x1) << 8) +#define G_000E40_HIRQ_ON_RBB(x) (((x) >> 8) & 0x1) +#define C_000E40_HIRQ_ON_RBB 0xFFFFFEFF +#define S_000E40_CPRQ_ON_RBB(x) (((x) & 0x1) << 9) +#define G_000E40_CPRQ_ON_RBB(x) (((x) >> 9) & 0x1) +#define C_000E40_CPRQ_ON_RBB 0xFFFFFDFF +#define S_000E40_CFRQ_ON_RBB(x) (((x) & 0x1) << 10) +#define G_000E40_CFRQ_ON_RBB(x) (((x) >> 10) & 0x1) +#define C_000E40_CFRQ_ON_RBB 0xFFFFFBFF +#define S_000E40_HIRQ_IN_RTBUF(x) (((x) & 0x1) << 11) +#define G_000E40_HIRQ_IN_RTBUF(x) (((x) >> 11) & 0x1) +#define C_000E40_HIRQ_IN_RTBUF 0xFFFFF7FF +#define S_000E40_CPRQ_IN_RTBUF(x) (((x) & 0x1) << 12) +#define G_000E40_CPRQ_IN_RTBUF(x) (((x) >> 12) & 0x1) +#define C_000E40_CPRQ_IN_RTBUF 0xFFFFEFFF +#define S_000E40_CFRQ_IN_RTBUF(x) (((x) & 0x1) << 13) +#define G_000E40_CFRQ_IN_RTBUF(x) (((x) >> 13) & 0x1) +#define C_000E40_CFRQ_IN_RTBUF 0xFFFFDFFF +#define S_000E40_CF_PIPE_BUSY(x) (((x) & 0x1) << 14) +#define G_000E40_CF_PIPE_BUSY(x) (((x) >> 14) & 0x1) +#define C_000E40_CF_PIPE_BUSY 0xFFFFBFFF +#define S_000E40_ENG_EV_BUSY(x) (((x) & 0x1) << 15) +#define G_000E40_ENG_EV_BUSY(x) (((x) >> 15) & 0x1) +#define C_000E40_ENG_EV_BUSY 0xFFFF7FFF +#define S_000E40_CP_CMDSTRM_BUSY(x) (((x) & 0x1) << 16) +#define G_000E40_CP_CMDSTRM_BUSY(x) (((x) >> 16) & 0x1) +#define C_000E40_CP_CMDSTRM_BUSY 0xFFFEFFFF +#define S_000E40_E2_BUSY(x) (((x) & 0x1) << 17) +#define G_000E40_E2_BUSY(x) (((x) >> 17) & 0x1) +#define C_000E40_E2_BUSY 0xFFFDFFFF +#define S_000E40_RB2D_BUSY(x) (((x) & 0x1) << 18) +#define G_000E40_RB2D_BUSY(x) (((x) >> 18) & 0x1) +#define C_000E40_RB2D_BUSY 0xFFFBFFFF +#define S_000E40_RB3D_BUSY(x) (((x) & 0x1) << 19) +#define G_000E40_RB3D_BUSY(x) (((x) >> 19) & 0x1) +#define C_000E40_RB3D_BUSY 0xFFF7FFFF +#define S_000E40_VAP_BUSY(x) (((x) & 0x1) << 20) +#define G_000E40_VAP_BUSY(x) (((x) >> 20) & 0x1) +#define C_000E40_VAP_BUSY 0xFFEFFFFF +#define S_000E40_RE_BUSY(x) (((x) & 0x1) << 21) +#define G_000E40_RE_BUSY(x) (((x) >> 21) & 0x1) +#define C_000E40_RE_BUSY 0xFFDFFFFF +#define S_000E40_TAM_BUSY(x) (((x) & 0x1) << 22) +#define G_000E40_TAM_BUSY(x) (((x) >> 22) & 0x1) +#define C_000E40_TAM_BUSY 0xFFBFFFFF +#define S_000E40_TDM_BUSY(x) (((x) & 0x1) << 23) +#define G_000E40_TDM_BUSY(x) (((x) >> 23) & 0x1) +#define C_000E40_TDM_BUSY 0xFF7FFFFF +#define S_000E40_PB_BUSY(x) (((x) & 0x1) << 24) +#define G_000E40_PB_BUSY(x) (((x) >> 24) & 0x1) +#define C_000E40_PB_BUSY 0xFEFFFFFF +#define S_000E40_TIM_BUSY(x) (((x) & 0x1) << 25) +#define G_000E40_TIM_BUSY(x) (((x) >> 25) & 0x1) +#define C_000E40_TIM_BUSY 0xFDFFFFFF +#define S_000E40_GA_BUSY(x) (((x) & 0x1) << 26) +#define G_000E40_GA_BUSY(x) (((x) >> 26) & 0x1) +#define C_000E40_GA_BUSY 0xFBFFFFFF +#define S_000E40_CBA2D_BUSY(x) (((x) & 0x1) << 27) +#define G_000E40_CBA2D_BUSY(x) (((x) >> 27) & 0x1) +#define C_000E40_CBA2D_BUSY 0xF7FFFFFF +#define S_000E40_RBBM_HIBUSY(x) (((x) & 0x1) << 28) +#define G_000E40_RBBM_HIBUSY(x) (((x) >> 28) & 0x1) +#define C_000E40_RBBM_HIBUSY 0xEFFFFFFF +#define S_000E40_SKID_CFBUSY(x) (((x) & 0x1) << 29) +#define G_000E40_SKID_CFBUSY(x) (((x) >> 29) & 0x1) +#define C_000E40_SKID_CFBUSY 0xDFFFFFFF +#define S_000E40_VAP_VF_BUSY(x) (((x) & 0x1) << 30) +#define G_000E40_VAP_VF_BUSY(x) (((x) >> 30) & 0x1) +#define C_000E40_VAP_VF_BUSY 0xBFFFFFFF +#define S_000E40_GUI_ACTIVE(x) (((x) & 0x1) << 31) +#define G_000E40_GUI_ACTIVE(x) (((x) >> 31) & 0x1) +#define C_000E40_GUI_ACTIVE 0x7FFFFFFF + + +#define R_000004_MC_FB_LOCATION 0x000004 +#define S_000004_MC_FB_START(x) (((x) & 0xFFFF) << 0) +#define G_000004_MC_FB_START(x) (((x) >> 0) & 0xFFFF) +#define C_000004_MC_FB_START 0xFFFF0000 +#define S_000004_MC_FB_TOP(x) (((x) & 0xFFFF) << 16) +#define G_000004_MC_FB_TOP(x) (((x) >> 16) & 0xFFFF) +#define C_000004_MC_FB_TOP 0x0000FFFF +#define R_000005_MC_AGP_LOCATION 0x000005 +#define S_000005_MC_AGP_START(x) (((x) & 0xFFFF) << 0) +#define G_000005_MC_AGP_START(x) (((x) >> 0) & 0xFFFF) +#define C_000005_MC_AGP_START 0xFFFF0000 +#define S_000005_MC_AGP_TOP(x) (((x) & 0xFFFF) << 16) +#define G_000005_MC_AGP_TOP(x) (((x) >> 16) & 0xFFFF) +#define C_000005_MC_AGP_TOP 0x0000FFFF +#define R_000006_AGP_BASE 0x000006 +#define S_000006_AGP_BASE_ADDR(x) (((x) & 0xFFFFFFFF) << 0) +#define G_000006_AGP_BASE_ADDR(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_000006_AGP_BASE_ADDR 0x00000000 +#define R_000007_AGP_BASE_2 0x000007 +#define S_000007_AGP_BASE_ADDR_2(x) (((x) & 0xF) << 0) +#define G_000007_AGP_BASE_ADDR_2(x) (((x) >> 0) & 0xF) +#define C_000007_AGP_BASE_ADDR_2 0xFFFFFFF0 + +#endif diff --git a/drivers/gpu/drm/radeon/r600.c b/drivers/gpu/drm/radeon/r600.c new file mode 100644 index 00000000..1a4ed433 --- /dev/null +++ b/drivers/gpu/drm/radeon/r600.c @@ -0,0 +1,3776 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#include +#include +#include +#include +#include "drmP.h" +#include "radeon_drm.h" +#include "radeon.h" +#include "radeon_asic.h" +#include "radeon_mode.h" +#include "r600d.h" +#include "atom.h" +#include "avivod.h" + +#define PFP_UCODE_SIZE 576 +#define PM4_UCODE_SIZE 1792 +#define RLC_UCODE_SIZE 768 +#define R700_PFP_UCODE_SIZE 848 +#define R700_PM4_UCODE_SIZE 1360 +#define R700_RLC_UCODE_SIZE 1024 +#define EVERGREEN_PFP_UCODE_SIZE 1120 +#define EVERGREEN_PM4_UCODE_SIZE 1376 +#define EVERGREEN_RLC_UCODE_SIZE 768 +#define CAYMAN_RLC_UCODE_SIZE 1024 + +/* Firmware Names */ +MODULE_FIRMWARE("radeon/R600_pfp.bin"); +MODULE_FIRMWARE("radeon/R600_me.bin"); +MODULE_FIRMWARE("radeon/RV610_pfp.bin"); +MODULE_FIRMWARE("radeon/RV610_me.bin"); +MODULE_FIRMWARE("radeon/RV630_pfp.bin"); +MODULE_FIRMWARE("radeon/RV630_me.bin"); +MODULE_FIRMWARE("radeon/RV620_pfp.bin"); +MODULE_FIRMWARE("radeon/RV620_me.bin"); +MODULE_FIRMWARE("radeon/RV635_pfp.bin"); +MODULE_FIRMWARE("radeon/RV635_me.bin"); +MODULE_FIRMWARE("radeon/RV670_pfp.bin"); +MODULE_FIRMWARE("radeon/RV670_me.bin"); +MODULE_FIRMWARE("radeon/RS780_pfp.bin"); +MODULE_FIRMWARE("radeon/RS780_me.bin"); +MODULE_FIRMWARE("radeon/RV770_pfp.bin"); +MODULE_FIRMWARE("radeon/RV770_me.bin"); +MODULE_FIRMWARE("radeon/RV730_pfp.bin"); +MODULE_FIRMWARE("radeon/RV730_me.bin"); +MODULE_FIRMWARE("radeon/RV710_pfp.bin"); +MODULE_FIRMWARE("radeon/RV710_me.bin"); +MODULE_FIRMWARE("radeon/R600_rlc.bin"); +MODULE_FIRMWARE("radeon/R700_rlc.bin"); +MODULE_FIRMWARE("radeon/CEDAR_pfp.bin"); +MODULE_FIRMWARE("radeon/CEDAR_me.bin"); +MODULE_FIRMWARE("radeon/CEDAR_rlc.bin"); +MODULE_FIRMWARE("radeon/REDWOOD_pfp.bin"); +MODULE_FIRMWARE("radeon/REDWOOD_me.bin"); +MODULE_FIRMWARE("radeon/REDWOOD_rlc.bin"); +MODULE_FIRMWARE("radeon/JUNIPER_pfp.bin"); +MODULE_FIRMWARE("radeon/JUNIPER_me.bin"); +MODULE_FIRMWARE("radeon/JUNIPER_rlc.bin"); +MODULE_FIRMWARE("radeon/CYPRESS_pfp.bin"); +MODULE_FIRMWARE("radeon/CYPRESS_me.bin"); +MODULE_FIRMWARE("radeon/CYPRESS_rlc.bin"); +MODULE_FIRMWARE("radeon/PALM_pfp.bin"); +MODULE_FIRMWARE("radeon/PALM_me.bin"); +MODULE_FIRMWARE("radeon/SUMO_rlc.bin"); +MODULE_FIRMWARE("radeon/SUMO_pfp.bin"); +MODULE_FIRMWARE("radeon/SUMO_me.bin"); +MODULE_FIRMWARE("radeon/SUMO2_pfp.bin"); +MODULE_FIRMWARE("radeon/SUMO2_me.bin"); + +int r600_debugfs_mc_info_init(struct radeon_device *rdev); + +/* r600,rv610,rv630,rv620,rv635,rv670 */ +int r600_mc_wait_for_idle(struct radeon_device *rdev); +void r600_gpu_init(struct radeon_device *rdev); +void r600_fini(struct radeon_device *rdev); +void r600_irq_disable(struct radeon_device *rdev); +static void r600_pcie_gen2_enable(struct radeon_device *rdev); + +/* get temperature in millidegrees */ +int rv6xx_get_temp(struct radeon_device *rdev) +{ + u32 temp = (RREG32(CG_THERMAL_STATUS) & ASIC_T_MASK) >> + ASIC_T_SHIFT; + int actual_temp = temp & 0xff; + + if (temp & 0x100) + actual_temp -= 256; + + return actual_temp * 1000; +} + +void r600_pm_get_dynpm_state(struct radeon_device *rdev) +{ + int i; + + rdev->pm.dynpm_can_upclock = true; + rdev->pm.dynpm_can_downclock = true; + + /* power state array is low to high, default is first */ + if ((rdev->flags & RADEON_IS_IGP) || (rdev->family == CHIP_R600)) { + int min_power_state_index = 0; + + if (rdev->pm.num_power_states > 2) + min_power_state_index = 1; + + switch (rdev->pm.dynpm_planned_action) { + case DYNPM_ACTION_MINIMUM: + rdev->pm.requested_power_state_index = min_power_state_index; + rdev->pm.requested_clock_mode_index = 0; + rdev->pm.dynpm_can_downclock = false; + break; + case DYNPM_ACTION_DOWNCLOCK: + if (rdev->pm.current_power_state_index == min_power_state_index) { + rdev->pm.requested_power_state_index = rdev->pm.current_power_state_index; + rdev->pm.dynpm_can_downclock = false; + } else { + if (rdev->pm.active_crtc_count > 1) { + for (i = 0; i < rdev->pm.num_power_states; i++) { + if (rdev->pm.power_state[i].flags & RADEON_PM_STATE_SINGLE_DISPLAY_ONLY) + continue; + else if (i >= rdev->pm.current_power_state_index) { + rdev->pm.requested_power_state_index = + rdev->pm.current_power_state_index; + break; + } else { + rdev->pm.requested_power_state_index = i; + break; + } + } + } else { + if (rdev->pm.current_power_state_index == 0) + rdev->pm.requested_power_state_index = + rdev->pm.num_power_states - 1; + else + rdev->pm.requested_power_state_index = + rdev->pm.current_power_state_index - 1; + } + } + rdev->pm.requested_clock_mode_index = 0; + /* don't use the power state if crtcs are active and no display flag is set */ + if ((rdev->pm.active_crtc_count > 0) && + (rdev->pm.power_state[rdev->pm.requested_power_state_index]. + clock_info[rdev->pm.requested_clock_mode_index].flags & + RADEON_PM_MODE_NO_DISPLAY)) { + rdev->pm.requested_power_state_index++; + } + break; + case DYNPM_ACTION_UPCLOCK: + if (rdev->pm.current_power_state_index == (rdev->pm.num_power_states - 1)) { + rdev->pm.requested_power_state_index = rdev->pm.current_power_state_index; + rdev->pm.dynpm_can_upclock = false; + } else { + if (rdev->pm.active_crtc_count > 1) { + for (i = (rdev->pm.num_power_states - 1); i >= 0; i--) { + if (rdev->pm.power_state[i].flags & RADEON_PM_STATE_SINGLE_DISPLAY_ONLY) + continue; + else if (i <= rdev->pm.current_power_state_index) { + rdev->pm.requested_power_state_index = + rdev->pm.current_power_state_index; + break; + } else { + rdev->pm.requested_power_state_index = i; + break; + } + } + } else + rdev->pm.requested_power_state_index = + rdev->pm.current_power_state_index + 1; + } + rdev->pm.requested_clock_mode_index = 0; + break; + case DYNPM_ACTION_DEFAULT: + rdev->pm.requested_power_state_index = rdev->pm.default_power_state_index; + rdev->pm.requested_clock_mode_index = 0; + rdev->pm.dynpm_can_upclock = false; + break; + case DYNPM_ACTION_NONE: + default: + DRM_ERROR("Requested mode for not defined action\n"); + return; + } + } else { + /* XXX select a power state based on AC/DC, single/dualhead, etc. */ + /* for now just select the first power state and switch between clock modes */ + /* power state array is low to high, default is first (0) */ + if (rdev->pm.active_crtc_count > 1) { + rdev->pm.requested_power_state_index = -1; + /* start at 1 as we don't want the default mode */ + for (i = 1; i < rdev->pm.num_power_states; i++) { + if (rdev->pm.power_state[i].flags & RADEON_PM_STATE_SINGLE_DISPLAY_ONLY) + continue; + else if ((rdev->pm.power_state[i].type == POWER_STATE_TYPE_PERFORMANCE) || + (rdev->pm.power_state[i].type == POWER_STATE_TYPE_BATTERY)) { + rdev->pm.requested_power_state_index = i; + break; + } + } + /* if nothing selected, grab the default state. */ + if (rdev->pm.requested_power_state_index == -1) + rdev->pm.requested_power_state_index = 0; + } else + rdev->pm.requested_power_state_index = 1; + + switch (rdev->pm.dynpm_planned_action) { + case DYNPM_ACTION_MINIMUM: + rdev->pm.requested_clock_mode_index = 0; + rdev->pm.dynpm_can_downclock = false; + break; + case DYNPM_ACTION_DOWNCLOCK: + if (rdev->pm.requested_power_state_index == rdev->pm.current_power_state_index) { + if (rdev->pm.current_clock_mode_index == 0) { + rdev->pm.requested_clock_mode_index = 0; + rdev->pm.dynpm_can_downclock = false; + } else + rdev->pm.requested_clock_mode_index = + rdev->pm.current_clock_mode_index - 1; + } else { + rdev->pm.requested_clock_mode_index = 0; + rdev->pm.dynpm_can_downclock = false; + } + /* don't use the power state if crtcs are active and no display flag is set */ + if ((rdev->pm.active_crtc_count > 0) && + (rdev->pm.power_state[rdev->pm.requested_power_state_index]. + clock_info[rdev->pm.requested_clock_mode_index].flags & + RADEON_PM_MODE_NO_DISPLAY)) { + rdev->pm.requested_clock_mode_index++; + } + break; + case DYNPM_ACTION_UPCLOCK: + if (rdev->pm.requested_power_state_index == rdev->pm.current_power_state_index) { + if (rdev->pm.current_clock_mode_index == + (rdev->pm.power_state[rdev->pm.requested_power_state_index].num_clock_modes - 1)) { + rdev->pm.requested_clock_mode_index = rdev->pm.current_clock_mode_index; + rdev->pm.dynpm_can_upclock = false; + } else + rdev->pm.requested_clock_mode_index = + rdev->pm.current_clock_mode_index + 1; + } else { + rdev->pm.requested_clock_mode_index = + rdev->pm.power_state[rdev->pm.requested_power_state_index].num_clock_modes - 1; + rdev->pm.dynpm_can_upclock = false; + } + break; + case DYNPM_ACTION_DEFAULT: + rdev->pm.requested_power_state_index = rdev->pm.default_power_state_index; + rdev->pm.requested_clock_mode_index = 0; + rdev->pm.dynpm_can_upclock = false; + break; + case DYNPM_ACTION_NONE: + default: + DRM_ERROR("Requested mode for not defined action\n"); + return; + } + } + + DRM_DEBUG_DRIVER("Requested: e: %d m: %d p: %d\n", + rdev->pm.power_state[rdev->pm.requested_power_state_index]. + clock_info[rdev->pm.requested_clock_mode_index].sclk, + rdev->pm.power_state[rdev->pm.requested_power_state_index]. + clock_info[rdev->pm.requested_clock_mode_index].mclk, + rdev->pm.power_state[rdev->pm.requested_power_state_index]. + pcie_lanes); +} + +static int r600_pm_get_type_index(struct radeon_device *rdev, + enum radeon_pm_state_type ps_type, + int instance) +{ + int i; + int found_instance = -1; + + for (i = 0; i < rdev->pm.num_power_states; i++) { + if (rdev->pm.power_state[i].type == ps_type) { + found_instance++; + if (found_instance == instance) + return i; + } + } + /* return default if no match */ + return rdev->pm.default_power_state_index; +} + +void rs780_pm_init_profile(struct radeon_device *rdev) +{ + if (rdev->pm.num_power_states == 2) { + /* default */ + rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_cm_idx = 0; + /* low sh */ + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_ps_idx = 0; + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_ps_idx = 0; + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_cm_idx = 0; + /* mid sh */ + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_ps_idx = 0; + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_ps_idx = 0; + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_cm_idx = 0; + /* high sh */ + rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_ps_idx = 0; + rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_ps_idx = 1; + rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_cm_idx = 0; + /* low mh */ + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_ps_idx = 0; + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_ps_idx = 0; + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_cm_idx = 0; + /* mid mh */ + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_ps_idx = 0; + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_ps_idx = 0; + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_cm_idx = 0; + /* high mh */ + rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_ps_idx = 0; + rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_ps_idx = 1; + rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_cm_idx = 0; + } else if (rdev->pm.num_power_states == 3) { + /* default */ + rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_cm_idx = 0; + /* low sh */ + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_ps_idx = 1; + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_ps_idx = 1; + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_cm_idx = 0; + /* mid sh */ + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_ps_idx = 1; + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_ps_idx = 1; + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_cm_idx = 0; + /* high sh */ + rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_ps_idx = 1; + rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_ps_idx = 2; + rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_cm_idx = 0; + /* low mh */ + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_ps_idx = 1; + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_ps_idx = 1; + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_cm_idx = 0; + /* mid mh */ + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_ps_idx = 1; + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_ps_idx = 1; + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_cm_idx = 0; + /* high mh */ + rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_ps_idx = 1; + rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_ps_idx = 2; + rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_cm_idx = 0; + } else { + /* default */ + rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_cm_idx = 0; + /* low sh */ + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_ps_idx = 2; + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_ps_idx = 2; + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_cm_idx = 0; + /* mid sh */ + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_ps_idx = 2; + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_ps_idx = 2; + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_cm_idx = 0; + /* high sh */ + rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_ps_idx = 2; + rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_ps_idx = 3; + rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_cm_idx = 0; + /* low mh */ + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_ps_idx = 2; + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_ps_idx = 0; + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_cm_idx = 0; + /* mid mh */ + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_ps_idx = 2; + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_ps_idx = 0; + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_cm_idx = 0; + /* high mh */ + rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_ps_idx = 2; + rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_ps_idx = 3; + rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_cm_idx = 0; + } +} + +void r600_pm_init_profile(struct radeon_device *rdev) +{ + if (rdev->family == CHIP_R600) { + /* XXX */ + /* default */ + rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_cm_idx = 0; + /* low sh */ + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_cm_idx = 0; + /* mid sh */ + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_cm_idx = 0; + /* high sh */ + rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_cm_idx = 0; + /* low mh */ + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_cm_idx = 0; + /* mid mh */ + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_cm_idx = 0; + /* high mh */ + rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_cm_idx = 0; + } else { + if (rdev->pm.num_power_states < 4) { + /* default */ + rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_cm_idx = 2; + /* low sh */ + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_ps_idx = 1; + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_ps_idx = 1; + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_cm_idx = 0; + /* mid sh */ + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_ps_idx = 1; + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_ps_idx = 1; + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_cm_idx = 1; + /* high sh */ + rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_ps_idx = 1; + rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_ps_idx = 1; + rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_cm_idx = 2; + /* low mh */ + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_ps_idx = 2; + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_ps_idx = 2; + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_cm_idx = 0; + /* low mh */ + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_ps_idx = 2; + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_ps_idx = 2; + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_cm_idx = 1; + /* high mh */ + rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_ps_idx = 2; + rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_ps_idx = 2; + rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_cm_idx = 2; + } else { + /* default */ + rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_ps_idx = rdev->pm.default_power_state_index; + rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_DEFAULT_IDX].dpms_on_cm_idx = 2; + /* low sh */ + if (rdev->flags & RADEON_IS_MOBILITY) { + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_ps_idx = + r600_pm_get_type_index(rdev, POWER_STATE_TYPE_BATTERY, 0); + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_ps_idx = + r600_pm_get_type_index(rdev, POWER_STATE_TYPE_BATTERY, 0); + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_cm_idx = 0; + } else { + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_ps_idx = + r600_pm_get_type_index(rdev, POWER_STATE_TYPE_PERFORMANCE, 0); + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_ps_idx = + r600_pm_get_type_index(rdev, POWER_STATE_TYPE_PERFORMANCE, 0); + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_LOW_SH_IDX].dpms_on_cm_idx = 0; + } + /* mid sh */ + if (rdev->flags & RADEON_IS_MOBILITY) { + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_ps_idx = + r600_pm_get_type_index(rdev, POWER_STATE_TYPE_BATTERY, 0); + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_ps_idx = + r600_pm_get_type_index(rdev, POWER_STATE_TYPE_BATTERY, 0); + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_cm_idx = 1; + } else { + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_ps_idx = + r600_pm_get_type_index(rdev, POWER_STATE_TYPE_PERFORMANCE, 0); + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_ps_idx = + r600_pm_get_type_index(rdev, POWER_STATE_TYPE_PERFORMANCE, 0); + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_MID_SH_IDX].dpms_on_cm_idx = 1; + } + /* high sh */ + rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_ps_idx = + r600_pm_get_type_index(rdev, POWER_STATE_TYPE_PERFORMANCE, 0); + rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_ps_idx = + r600_pm_get_type_index(rdev, POWER_STATE_TYPE_PERFORMANCE, 0); + rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_HIGH_SH_IDX].dpms_on_cm_idx = 2; + /* low mh */ + if (rdev->flags & RADEON_IS_MOBILITY) { + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_ps_idx = + r600_pm_get_type_index(rdev, POWER_STATE_TYPE_BATTERY, 1); + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_ps_idx = + r600_pm_get_type_index(rdev, POWER_STATE_TYPE_BATTERY, 1); + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_cm_idx = 0; + } else { + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_ps_idx = + r600_pm_get_type_index(rdev, POWER_STATE_TYPE_PERFORMANCE, 1); + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_ps_idx = + r600_pm_get_type_index(rdev, POWER_STATE_TYPE_PERFORMANCE, 1); + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_LOW_MH_IDX].dpms_on_cm_idx = 0; + } + /* mid mh */ + if (rdev->flags & RADEON_IS_MOBILITY) { + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_ps_idx = + r600_pm_get_type_index(rdev, POWER_STATE_TYPE_BATTERY, 1); + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_ps_idx = + r600_pm_get_type_index(rdev, POWER_STATE_TYPE_BATTERY, 1); + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_cm_idx = 1; + } else { + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_ps_idx = + r600_pm_get_type_index(rdev, POWER_STATE_TYPE_PERFORMANCE, 1); + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_ps_idx = + r600_pm_get_type_index(rdev, POWER_STATE_TYPE_PERFORMANCE, 1); + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_MID_MH_IDX].dpms_on_cm_idx = 1; + } + /* high mh */ + rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_ps_idx = + r600_pm_get_type_index(rdev, POWER_STATE_TYPE_PERFORMANCE, 1); + rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_ps_idx = + r600_pm_get_type_index(rdev, POWER_STATE_TYPE_PERFORMANCE, 1); + rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_off_cm_idx = 0; + rdev->pm.profiles[PM_PROFILE_HIGH_MH_IDX].dpms_on_cm_idx = 2; + } + } +} + +void r600_pm_misc(struct radeon_device *rdev) +{ + int req_ps_idx = rdev->pm.requested_power_state_index; + int req_cm_idx = rdev->pm.requested_clock_mode_index; + struct radeon_power_state *ps = &rdev->pm.power_state[req_ps_idx]; + struct radeon_voltage *voltage = &ps->clock_info[req_cm_idx].voltage; + + if ((voltage->type == VOLTAGE_SW) && voltage->voltage) { + /* 0xff01 is a flag rather then an actual voltage */ + if (voltage->voltage == 0xff01) + return; + if (voltage->voltage != rdev->pm.current_vddc) { + radeon_atom_set_voltage(rdev, voltage->voltage, SET_VOLTAGE_TYPE_ASIC_VDDC); + rdev->pm.current_vddc = voltage->voltage; + DRM_DEBUG_DRIVER("Setting: v: %d\n", voltage->voltage); + } + } +} + +bool r600_gui_idle(struct radeon_device *rdev) +{ + if (RREG32(GRBM_STATUS) & GUI_ACTIVE) + return false; + else + return true; +} + +/* hpd for digital panel detect/disconnect */ +bool r600_hpd_sense(struct radeon_device *rdev, enum radeon_hpd_id hpd) +{ + bool connected = false; + + if (ASIC_IS_DCE3(rdev)) { + switch (hpd) { + case RADEON_HPD_1: + if (RREG32(DC_HPD1_INT_STATUS) & DC_HPDx_SENSE) + connected = true; + break; + case RADEON_HPD_2: + if (RREG32(DC_HPD2_INT_STATUS) & DC_HPDx_SENSE) + connected = true; + break; + case RADEON_HPD_3: + if (RREG32(DC_HPD3_INT_STATUS) & DC_HPDx_SENSE) + connected = true; + break; + case RADEON_HPD_4: + if (RREG32(DC_HPD4_INT_STATUS) & DC_HPDx_SENSE) + connected = true; + break; + /* DCE 3.2 */ + case RADEON_HPD_5: + if (RREG32(DC_HPD5_INT_STATUS) & DC_HPDx_SENSE) + connected = true; + break; + case RADEON_HPD_6: + if (RREG32(DC_HPD6_INT_STATUS) & DC_HPDx_SENSE) + connected = true; + break; + default: + break; + } + } else { + switch (hpd) { + case RADEON_HPD_1: + if (RREG32(DC_HOT_PLUG_DETECT1_INT_STATUS) & DC_HOT_PLUG_DETECTx_SENSE) + connected = true; + break; + case RADEON_HPD_2: + if (RREG32(DC_HOT_PLUG_DETECT2_INT_STATUS) & DC_HOT_PLUG_DETECTx_SENSE) + connected = true; + break; + case RADEON_HPD_3: + if (RREG32(DC_HOT_PLUG_DETECT3_INT_STATUS) & DC_HOT_PLUG_DETECTx_SENSE) + connected = true; + break; + default: + break; + } + } + return connected; +} + +void r600_hpd_set_polarity(struct radeon_device *rdev, + enum radeon_hpd_id hpd) +{ + u32 tmp; + bool connected = r600_hpd_sense(rdev, hpd); + + if (ASIC_IS_DCE3(rdev)) { + switch (hpd) { + case RADEON_HPD_1: + tmp = RREG32(DC_HPD1_INT_CONTROL); + if (connected) + tmp &= ~DC_HPDx_INT_POLARITY; + else + tmp |= DC_HPDx_INT_POLARITY; + WREG32(DC_HPD1_INT_CONTROL, tmp); + break; + case RADEON_HPD_2: + tmp = RREG32(DC_HPD2_INT_CONTROL); + if (connected) + tmp &= ~DC_HPDx_INT_POLARITY; + else + tmp |= DC_HPDx_INT_POLARITY; + WREG32(DC_HPD2_INT_CONTROL, tmp); + break; + case RADEON_HPD_3: + tmp = RREG32(DC_HPD3_INT_CONTROL); + if (connected) + tmp &= ~DC_HPDx_INT_POLARITY; + else + tmp |= DC_HPDx_INT_POLARITY; + WREG32(DC_HPD3_INT_CONTROL, tmp); + break; + case RADEON_HPD_4: + tmp = RREG32(DC_HPD4_INT_CONTROL); + if (connected) + tmp &= ~DC_HPDx_INT_POLARITY; + else + tmp |= DC_HPDx_INT_POLARITY; + WREG32(DC_HPD4_INT_CONTROL, tmp); + break; + case RADEON_HPD_5: + tmp = RREG32(DC_HPD5_INT_CONTROL); + if (connected) + tmp &= ~DC_HPDx_INT_POLARITY; + else + tmp |= DC_HPDx_INT_POLARITY; + WREG32(DC_HPD5_INT_CONTROL, tmp); + break; + /* DCE 3.2 */ + case RADEON_HPD_6: + tmp = RREG32(DC_HPD6_INT_CONTROL); + if (connected) + tmp &= ~DC_HPDx_INT_POLARITY; + else + tmp |= DC_HPDx_INT_POLARITY; + WREG32(DC_HPD6_INT_CONTROL, tmp); + break; + default: + break; + } + } else { + switch (hpd) { + case RADEON_HPD_1: + tmp = RREG32(DC_HOT_PLUG_DETECT1_INT_CONTROL); + if (connected) + tmp &= ~DC_HOT_PLUG_DETECTx_INT_POLARITY; + else + tmp |= DC_HOT_PLUG_DETECTx_INT_POLARITY; + WREG32(DC_HOT_PLUG_DETECT1_INT_CONTROL, tmp); + break; + case RADEON_HPD_2: + tmp = RREG32(DC_HOT_PLUG_DETECT2_INT_CONTROL); + if (connected) + tmp &= ~DC_HOT_PLUG_DETECTx_INT_POLARITY; + else + tmp |= DC_HOT_PLUG_DETECTx_INT_POLARITY; + WREG32(DC_HOT_PLUG_DETECT2_INT_CONTROL, tmp); + break; + case RADEON_HPD_3: + tmp = RREG32(DC_HOT_PLUG_DETECT3_INT_CONTROL); + if (connected) + tmp &= ~DC_HOT_PLUG_DETECTx_INT_POLARITY; + else + tmp |= DC_HOT_PLUG_DETECTx_INT_POLARITY; + WREG32(DC_HOT_PLUG_DETECT3_INT_CONTROL, tmp); + break; + default: + break; + } + } +} + +void r600_hpd_init(struct radeon_device *rdev) +{ + struct drm_device *dev = rdev->ddev; + struct drm_connector *connector; + + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + + if (ASIC_IS_DCE3(rdev)) { + u32 tmp = DC_HPDx_CONNECTION_TIMER(0x9c4) | DC_HPDx_RX_INT_TIMER(0xfa); + if (ASIC_IS_DCE32(rdev)) + tmp |= DC_HPDx_EN; + + switch (radeon_connector->hpd.hpd) { + case RADEON_HPD_1: + WREG32(DC_HPD1_CONTROL, tmp); + rdev->irq.hpd[0] = true; + break; + case RADEON_HPD_2: + WREG32(DC_HPD2_CONTROL, tmp); + rdev->irq.hpd[1] = true; + break; + case RADEON_HPD_3: + WREG32(DC_HPD3_CONTROL, tmp); + rdev->irq.hpd[2] = true; + break; + case RADEON_HPD_4: + WREG32(DC_HPD4_CONTROL, tmp); + rdev->irq.hpd[3] = true; + break; + /* DCE 3.2 */ + case RADEON_HPD_5: + WREG32(DC_HPD5_CONTROL, tmp); + rdev->irq.hpd[4] = true; + break; + case RADEON_HPD_6: + WREG32(DC_HPD6_CONTROL, tmp); + rdev->irq.hpd[5] = true; + break; + default: + break; + } + } else { + switch (radeon_connector->hpd.hpd) { + case RADEON_HPD_1: + WREG32(DC_HOT_PLUG_DETECT1_CONTROL, DC_HOT_PLUG_DETECTx_EN); + rdev->irq.hpd[0] = true; + break; + case RADEON_HPD_2: + WREG32(DC_HOT_PLUG_DETECT2_CONTROL, DC_HOT_PLUG_DETECTx_EN); + rdev->irq.hpd[1] = true; + break; + case RADEON_HPD_3: + WREG32(DC_HOT_PLUG_DETECT3_CONTROL, DC_HOT_PLUG_DETECTx_EN); + rdev->irq.hpd[2] = true; + break; + default: + break; + } + } + radeon_hpd_set_polarity(rdev, radeon_connector->hpd.hpd); + } + if (rdev->irq.installed) + r600_irq_set(rdev); +} + +void r600_hpd_fini(struct radeon_device *rdev) +{ + struct drm_device *dev = rdev->ddev; + struct drm_connector *connector; + + if (ASIC_IS_DCE3(rdev)) { + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + switch (radeon_connector->hpd.hpd) { + case RADEON_HPD_1: + WREG32(DC_HPD1_CONTROL, 0); + rdev->irq.hpd[0] = false; + break; + case RADEON_HPD_2: + WREG32(DC_HPD2_CONTROL, 0); + rdev->irq.hpd[1] = false; + break; + case RADEON_HPD_3: + WREG32(DC_HPD3_CONTROL, 0); + rdev->irq.hpd[2] = false; + break; + case RADEON_HPD_4: + WREG32(DC_HPD4_CONTROL, 0); + rdev->irq.hpd[3] = false; + break; + /* DCE 3.2 */ + case RADEON_HPD_5: + WREG32(DC_HPD5_CONTROL, 0); + rdev->irq.hpd[4] = false; + break; + case RADEON_HPD_6: + WREG32(DC_HPD6_CONTROL, 0); + rdev->irq.hpd[5] = false; + break; + default: + break; + } + } + } else { + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + switch (radeon_connector->hpd.hpd) { + case RADEON_HPD_1: + WREG32(DC_HOT_PLUG_DETECT1_CONTROL, 0); + rdev->irq.hpd[0] = false; + break; + case RADEON_HPD_2: + WREG32(DC_HOT_PLUG_DETECT2_CONTROL, 0); + rdev->irq.hpd[1] = false; + break; + case RADEON_HPD_3: + WREG32(DC_HOT_PLUG_DETECT3_CONTROL, 0); + rdev->irq.hpd[2] = false; + break; + default: + break; + } + } + } +} + +/* + * R600 PCIE GART + */ +void r600_pcie_gart_tlb_flush(struct radeon_device *rdev) +{ + unsigned i; + u32 tmp; + + /* flush hdp cache so updates hit vram */ + if ((rdev->family >= CHIP_RV770) && (rdev->family <= CHIP_RV740) && + !(rdev->flags & RADEON_IS_AGP)) { + void __iomem *ptr = (void *)rdev->gart.table.vram.ptr; + u32 tmp; + + /* r7xx hw bug. write to HDP_DEBUG1 followed by fb read + * rather than write to HDP_REG_COHERENCY_FLUSH_CNTL + * This seems to cause problems on some AGP cards. Just use the old + * method for them. + */ + WREG32(HDP_DEBUG1, 0); + tmp = readl((void __iomem *)ptr); + } else + WREG32(R_005480_HDP_MEM_COHERENCY_FLUSH_CNTL, 0x1); + + WREG32(VM_CONTEXT0_INVALIDATION_LOW_ADDR, rdev->mc.gtt_start >> 12); + WREG32(VM_CONTEXT0_INVALIDATION_HIGH_ADDR, (rdev->mc.gtt_end - 1) >> 12); + WREG32(VM_CONTEXT0_REQUEST_RESPONSE, REQUEST_TYPE(1)); + for (i = 0; i < rdev->usec_timeout; i++) { + /* read MC_STATUS */ + tmp = RREG32(VM_CONTEXT0_REQUEST_RESPONSE); + tmp = (tmp & RESPONSE_TYPE_MASK) >> RESPONSE_TYPE_SHIFT; + if (tmp == 2) { + printk(KERN_WARNING "[drm] r600 flush TLB failed\n"); + return; + } + if (tmp) { + return; + } + udelay(1); + } +} + +int r600_pcie_gart_init(struct radeon_device *rdev) +{ + int r; + + if (rdev->gart.table.vram.robj) { + WARN(1, "R600 PCIE GART already initialized\n"); + return 0; + } + /* Initialize common gart structure */ + r = radeon_gart_init(rdev); + if (r) + return r; + rdev->gart.table_size = rdev->gart.num_gpu_pages * 8; + return radeon_gart_table_vram_alloc(rdev); +} + +int r600_pcie_gart_enable(struct radeon_device *rdev) +{ + u32 tmp; + int r, i; + + if (rdev->gart.table.vram.robj == NULL) { + dev_err(rdev->dev, "No VRAM object for PCIE GART.\n"); + return -EINVAL; + } + r = radeon_gart_table_vram_pin(rdev); + if (r) + return r; + radeon_gart_restore(rdev); + + /* Setup L2 cache */ + WREG32(VM_L2_CNTL, ENABLE_L2_CACHE | ENABLE_L2_FRAGMENT_PROCESSING | + ENABLE_L2_PTE_CACHE_LRU_UPDATE_BY_WRITE | + EFFECTIVE_L2_QUEUE_SIZE(7)); + WREG32(VM_L2_CNTL2, 0); + WREG32(VM_L2_CNTL3, BANK_SELECT_0(0) | BANK_SELECT_1(1)); + /* Setup TLB control */ + tmp = ENABLE_L1_TLB | ENABLE_L1_FRAGMENT_PROCESSING | + SYSTEM_ACCESS_MODE_NOT_IN_SYS | + EFFECTIVE_L1_TLB_SIZE(5) | EFFECTIVE_L1_QUEUE_SIZE(5) | + ENABLE_WAIT_L2_QUERY; + WREG32(MC_VM_L1_TLB_MCB_RD_SYS_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCB_WR_SYS_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCB_RD_HDP_CNTL, tmp | ENABLE_L1_STRICT_ORDERING); + WREG32(MC_VM_L1_TLB_MCB_WR_HDP_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCD_RD_A_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCD_WR_A_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCD_RD_B_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCD_WR_B_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCB_RD_GFX_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCB_WR_GFX_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCB_RD_PDMA_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCB_WR_PDMA_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCB_RD_SEM_CNTL, tmp | ENABLE_SEMAPHORE_MODE); + WREG32(MC_VM_L1_TLB_MCB_WR_SEM_CNTL, tmp | ENABLE_SEMAPHORE_MODE); + WREG32(VM_CONTEXT0_PAGE_TABLE_START_ADDR, rdev->mc.gtt_start >> 12); + WREG32(VM_CONTEXT0_PAGE_TABLE_END_ADDR, rdev->mc.gtt_end >> 12); + WREG32(VM_CONTEXT0_PAGE_TABLE_BASE_ADDR, rdev->gart.table_addr >> 12); + WREG32(VM_CONTEXT0_CNTL, ENABLE_CONTEXT | PAGE_TABLE_DEPTH(0) | + RANGE_PROTECTION_FAULT_ENABLE_DEFAULT); + WREG32(VM_CONTEXT0_PROTECTION_FAULT_DEFAULT_ADDR, + (u32)(rdev->dummy_page.addr >> 12)); + for (i = 1; i < 7; i++) + WREG32(VM_CONTEXT0_CNTL + (i * 4), 0); + + r600_pcie_gart_tlb_flush(rdev); + rdev->gart.ready = true; + return 0; +} + +void r600_pcie_gart_disable(struct radeon_device *rdev) +{ + u32 tmp; + int i, r; + + /* Disable all tables */ + for (i = 0; i < 7; i++) + WREG32(VM_CONTEXT0_CNTL + (i * 4), 0); + + /* Disable L2 cache */ + WREG32(VM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING | + EFFECTIVE_L2_QUEUE_SIZE(7)); + WREG32(VM_L2_CNTL3, BANK_SELECT_0(0) | BANK_SELECT_1(1)); + /* Setup L1 TLB control */ + tmp = EFFECTIVE_L1_TLB_SIZE(5) | EFFECTIVE_L1_QUEUE_SIZE(5) | + ENABLE_WAIT_L2_QUERY; + WREG32(MC_VM_L1_TLB_MCD_RD_A_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCD_WR_A_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCD_RD_B_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCD_WR_B_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCB_RD_GFX_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCB_WR_GFX_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCB_RD_PDMA_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCB_WR_PDMA_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCB_RD_SEM_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCB_WR_SEM_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCB_RD_SYS_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCB_WR_SYS_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCB_RD_HDP_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCB_WR_HDP_CNTL, tmp); + if (rdev->gart.table.vram.robj) { + r = radeon_bo_reserve(rdev->gart.table.vram.robj, false); + if (likely(r == 0)) { + radeon_bo_kunmap(rdev->gart.table.vram.robj); + radeon_bo_unpin(rdev->gart.table.vram.robj); + radeon_bo_unreserve(rdev->gart.table.vram.robj); + } + } +} + +void r600_pcie_gart_fini(struct radeon_device *rdev) +{ + radeon_gart_fini(rdev); + r600_pcie_gart_disable(rdev); + radeon_gart_table_vram_free(rdev); +} + +void r600_agp_enable(struct radeon_device *rdev) +{ + u32 tmp; + int i; + + /* Setup L2 cache */ + WREG32(VM_L2_CNTL, ENABLE_L2_CACHE | ENABLE_L2_FRAGMENT_PROCESSING | + ENABLE_L2_PTE_CACHE_LRU_UPDATE_BY_WRITE | + EFFECTIVE_L2_QUEUE_SIZE(7)); + WREG32(VM_L2_CNTL2, 0); + WREG32(VM_L2_CNTL3, BANK_SELECT_0(0) | BANK_SELECT_1(1)); + /* Setup TLB control */ + tmp = ENABLE_L1_TLB | ENABLE_L1_FRAGMENT_PROCESSING | + SYSTEM_ACCESS_MODE_NOT_IN_SYS | + EFFECTIVE_L1_TLB_SIZE(5) | EFFECTIVE_L1_QUEUE_SIZE(5) | + ENABLE_WAIT_L2_QUERY; + WREG32(MC_VM_L1_TLB_MCB_RD_SYS_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCB_WR_SYS_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCB_RD_HDP_CNTL, tmp | ENABLE_L1_STRICT_ORDERING); + WREG32(MC_VM_L1_TLB_MCB_WR_HDP_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCD_RD_A_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCD_WR_A_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCD_RD_B_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCD_WR_B_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCB_RD_GFX_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCB_WR_GFX_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCB_RD_PDMA_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCB_WR_PDMA_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCB_RD_SEM_CNTL, tmp | ENABLE_SEMAPHORE_MODE); + WREG32(MC_VM_L1_TLB_MCB_WR_SEM_CNTL, tmp | ENABLE_SEMAPHORE_MODE); + for (i = 0; i < 7; i++) + WREG32(VM_CONTEXT0_CNTL + (i * 4), 0); +} + +int r600_mc_wait_for_idle(struct radeon_device *rdev) +{ + unsigned i; + u32 tmp; + + for (i = 0; i < rdev->usec_timeout; i++) { + /* read MC_STATUS */ + tmp = RREG32(R_000E50_SRBM_STATUS) & 0x3F00; + if (!tmp) + return 0; + udelay(1); + } + return -1; +} + +static void r600_mc_program(struct radeon_device *rdev) +{ + struct rv515_mc_save save; + u32 tmp; + int i, j; + + /* Initialize HDP */ + for (i = 0, j = 0; i < 32; i++, j += 0x18) { + WREG32((0x2c14 + j), 0x00000000); + WREG32((0x2c18 + j), 0x00000000); + WREG32((0x2c1c + j), 0x00000000); + WREG32((0x2c20 + j), 0x00000000); + WREG32((0x2c24 + j), 0x00000000); + } + WREG32(HDP_REG_COHERENCY_FLUSH_CNTL, 0); + + rv515_mc_stop(rdev, &save); + if (r600_mc_wait_for_idle(rdev)) { + dev_warn(rdev->dev, "Wait for MC idle timedout !\n"); + } + /* Lockout access through VGA aperture (doesn't exist before R600) */ + WREG32(VGA_HDP_CONTROL, VGA_MEMORY_DISABLE); + /* Update configuration */ + if (rdev->flags & RADEON_IS_AGP) { + if (rdev->mc.vram_start < rdev->mc.gtt_start) { + /* VRAM before AGP */ + WREG32(MC_VM_SYSTEM_APERTURE_LOW_ADDR, + rdev->mc.vram_start >> 12); + WREG32(MC_VM_SYSTEM_APERTURE_HIGH_ADDR, + rdev->mc.gtt_end >> 12); + } else { + /* VRAM after AGP */ + WREG32(MC_VM_SYSTEM_APERTURE_LOW_ADDR, + rdev->mc.gtt_start >> 12); + WREG32(MC_VM_SYSTEM_APERTURE_HIGH_ADDR, + rdev->mc.vram_end >> 12); + } + } else { + WREG32(MC_VM_SYSTEM_APERTURE_LOW_ADDR, rdev->mc.vram_start >> 12); + WREG32(MC_VM_SYSTEM_APERTURE_HIGH_ADDR, rdev->mc.vram_end >> 12); + } + WREG32(MC_VM_SYSTEM_APERTURE_DEFAULT_ADDR, 0); + tmp = ((rdev->mc.vram_end >> 24) & 0xFFFF) << 16; + tmp |= ((rdev->mc.vram_start >> 24) & 0xFFFF); + WREG32(MC_VM_FB_LOCATION, tmp); + WREG32(HDP_NONSURFACE_BASE, (rdev->mc.vram_start >> 8)); + WREG32(HDP_NONSURFACE_INFO, (2 << 7)); + WREG32(HDP_NONSURFACE_SIZE, 0x3FFFFFFF); + if (rdev->flags & RADEON_IS_AGP) { + WREG32(MC_VM_AGP_TOP, rdev->mc.gtt_end >> 22); + WREG32(MC_VM_AGP_BOT, rdev->mc.gtt_start >> 22); + WREG32(MC_VM_AGP_BASE, rdev->mc.agp_base >> 22); + } else { + WREG32(MC_VM_AGP_BASE, 0); + WREG32(MC_VM_AGP_TOP, 0x0FFFFFFF); + WREG32(MC_VM_AGP_BOT, 0x0FFFFFFF); + } + if (r600_mc_wait_for_idle(rdev)) { + dev_warn(rdev->dev, "Wait for MC idle timedout !\n"); + } + rv515_mc_resume(rdev, &save); + /* we need to own VRAM, so turn off the VGA renderer here + * to stop it overwriting our objects */ + rv515_vga_render_disable(rdev); +} + +/** + * r600_vram_gtt_location - try to find VRAM & GTT location + * @rdev: radeon device structure holding all necessary informations + * @mc: memory controller structure holding memory informations + * + * Function will place try to place VRAM at same place as in CPU (PCI) + * address space as some GPU seems to have issue when we reprogram at + * different address space. + * + * If there is not enough space to fit the unvisible VRAM after the + * aperture then we limit the VRAM size to the aperture. + * + * If we are using AGP then place VRAM adjacent to AGP aperture are we need + * them to be in one from GPU point of view so that we can program GPU to + * catch access outside them (weird GPU policy see ??). + * + * This function will never fails, worst case are limiting VRAM or GTT. + * + * Note: GTT start, end, size should be initialized before calling this + * function on AGP platform. + */ +static void r600_vram_gtt_location(struct radeon_device *rdev, struct radeon_mc *mc) +{ + u64 size_bf, size_af; + + if (mc->mc_vram_size > 0xE0000000) { + /* leave room for at least 512M GTT */ + dev_warn(rdev->dev, "limiting VRAM\n"); + mc->real_vram_size = 0xE0000000; + mc->mc_vram_size = 0xE0000000; + } + if (rdev->flags & RADEON_IS_AGP) { + size_bf = mc->gtt_start; + size_af = 0xFFFFFFFF - mc->gtt_end + 1; + if (size_bf > size_af) { + if (mc->mc_vram_size > size_bf) { + dev_warn(rdev->dev, "limiting VRAM\n"); + mc->real_vram_size = size_bf; + mc->mc_vram_size = size_bf; + } + mc->vram_start = mc->gtt_start - mc->mc_vram_size; + } else { + if (mc->mc_vram_size > size_af) { + dev_warn(rdev->dev, "limiting VRAM\n"); + mc->real_vram_size = size_af; + mc->mc_vram_size = size_af; + } + mc->vram_start = mc->gtt_end; + } + mc->vram_end = mc->vram_start + mc->mc_vram_size - 1; + dev_info(rdev->dev, "VRAM: %lluM 0x%08llX - 0x%08llX (%lluM used)\n", + mc->mc_vram_size >> 20, mc->vram_start, + mc->vram_end, mc->real_vram_size >> 20); + } else { + u64 base = 0; + if (rdev->flags & RADEON_IS_IGP) { + base = RREG32(MC_VM_FB_LOCATION) & 0xFFFF; + base <<= 24; + } + radeon_vram_location(rdev, &rdev->mc, base); + rdev->mc.gtt_base_align = 0; + radeon_gtt_location(rdev, mc); + } +} + +int r600_mc_init(struct radeon_device *rdev) +{ + u32 tmp; + int chansize, numchan; + + /* Get VRAM informations */ + rdev->mc.vram_is_ddr = true; + tmp = RREG32(RAMCFG); + if (tmp & CHANSIZE_OVERRIDE) { + chansize = 16; + } else if (tmp & CHANSIZE_MASK) { + chansize = 64; + } else { + chansize = 32; + } + tmp = RREG32(CHMAP); + switch ((tmp & NOOFCHAN_MASK) >> NOOFCHAN_SHIFT) { + case 0: + default: + numchan = 1; + break; + case 1: + numchan = 2; + break; + case 2: + numchan = 4; + break; + case 3: + numchan = 8; + break; + } + rdev->mc.vram_width = numchan * chansize; + /* Could aper size report 0 ? */ + rdev->mc.aper_base = pci_resource_start(rdev->pdev, 0); + rdev->mc.aper_size = pci_resource_len(rdev->pdev, 0); + /* Setup GPU memory space */ + rdev->mc.mc_vram_size = RREG32(CONFIG_MEMSIZE); + rdev->mc.real_vram_size = RREG32(CONFIG_MEMSIZE); + rdev->mc.visible_vram_size = rdev->mc.aper_size; + r600_vram_gtt_location(rdev, &rdev->mc); + + if (rdev->flags & RADEON_IS_IGP) { + rs690_pm_info(rdev); + rdev->mc.igp_sideport_enabled = radeon_atombios_sideport_present(rdev); + } + radeon_update_bandwidth_info(rdev); + return 0; +} + +/* We doesn't check that the GPU really needs a reset we simply do the + * reset, it's up to the caller to determine if the GPU needs one. We + * might add an helper function to check that. + */ +int r600_gpu_soft_reset(struct radeon_device *rdev) +{ + struct rv515_mc_save save; + u32 grbm_busy_mask = S_008010_VC_BUSY(1) | S_008010_VGT_BUSY_NO_DMA(1) | + S_008010_VGT_BUSY(1) | S_008010_TA03_BUSY(1) | + S_008010_TC_BUSY(1) | S_008010_SX_BUSY(1) | + S_008010_SH_BUSY(1) | S_008010_SPI03_BUSY(1) | + S_008010_SMX_BUSY(1) | S_008010_SC_BUSY(1) | + S_008010_PA_BUSY(1) | S_008010_DB03_BUSY(1) | + S_008010_CR_BUSY(1) | S_008010_CB03_BUSY(1) | + S_008010_GUI_ACTIVE(1); + u32 grbm2_busy_mask = S_008014_SPI0_BUSY(1) | S_008014_SPI1_BUSY(1) | + S_008014_SPI2_BUSY(1) | S_008014_SPI3_BUSY(1) | + S_008014_TA0_BUSY(1) | S_008014_TA1_BUSY(1) | + S_008014_TA2_BUSY(1) | S_008014_TA3_BUSY(1) | + S_008014_DB0_BUSY(1) | S_008014_DB1_BUSY(1) | + S_008014_DB2_BUSY(1) | S_008014_DB3_BUSY(1) | + S_008014_CB0_BUSY(1) | S_008014_CB1_BUSY(1) | + S_008014_CB2_BUSY(1) | S_008014_CB3_BUSY(1); + u32 tmp; + + if (!(RREG32(GRBM_STATUS) & GUI_ACTIVE)) + return 0; + + dev_info(rdev->dev, "GPU softreset \n"); + dev_info(rdev->dev, " R_008010_GRBM_STATUS=0x%08X\n", + RREG32(R_008010_GRBM_STATUS)); + dev_info(rdev->dev, " R_008014_GRBM_STATUS2=0x%08X\n", + RREG32(R_008014_GRBM_STATUS2)); + dev_info(rdev->dev, " R_000E50_SRBM_STATUS=0x%08X\n", + RREG32(R_000E50_SRBM_STATUS)); + rv515_mc_stop(rdev, &save); + if (r600_mc_wait_for_idle(rdev)) { + dev_warn(rdev->dev, "Wait for MC idle timedout !\n"); + } + /* Disable CP parsing/prefetching */ + WREG32(R_0086D8_CP_ME_CNTL, S_0086D8_CP_ME_HALT(1)); + /* Check if any of the rendering block is busy and reset it */ + if ((RREG32(R_008010_GRBM_STATUS) & grbm_busy_mask) || + (RREG32(R_008014_GRBM_STATUS2) & grbm2_busy_mask)) { + tmp = S_008020_SOFT_RESET_CR(1) | + S_008020_SOFT_RESET_DB(1) | + S_008020_SOFT_RESET_CB(1) | + S_008020_SOFT_RESET_PA(1) | + S_008020_SOFT_RESET_SC(1) | + S_008020_SOFT_RESET_SMX(1) | + S_008020_SOFT_RESET_SPI(1) | + S_008020_SOFT_RESET_SX(1) | + S_008020_SOFT_RESET_SH(1) | + S_008020_SOFT_RESET_TC(1) | + S_008020_SOFT_RESET_TA(1) | + S_008020_SOFT_RESET_VC(1) | + S_008020_SOFT_RESET_VGT(1); + dev_info(rdev->dev, " R_008020_GRBM_SOFT_RESET=0x%08X\n", tmp); + WREG32(R_008020_GRBM_SOFT_RESET, tmp); + RREG32(R_008020_GRBM_SOFT_RESET); + mdelay(15); + WREG32(R_008020_GRBM_SOFT_RESET, 0); + } + /* Reset CP (we always reset CP) */ + tmp = S_008020_SOFT_RESET_CP(1); + dev_info(rdev->dev, "R_008020_GRBM_SOFT_RESET=0x%08X\n", tmp); + WREG32(R_008020_GRBM_SOFT_RESET, tmp); + RREG32(R_008020_GRBM_SOFT_RESET); + mdelay(15); + WREG32(R_008020_GRBM_SOFT_RESET, 0); + /* Wait a little for things to settle down */ + mdelay(1); + dev_info(rdev->dev, " R_008010_GRBM_STATUS=0x%08X\n", + RREG32(R_008010_GRBM_STATUS)); + dev_info(rdev->dev, " R_008014_GRBM_STATUS2=0x%08X\n", + RREG32(R_008014_GRBM_STATUS2)); + dev_info(rdev->dev, " R_000E50_SRBM_STATUS=0x%08X\n", + RREG32(R_000E50_SRBM_STATUS)); + rv515_mc_resume(rdev, &save); + return 0; +} + +bool r600_gpu_is_lockup(struct radeon_device *rdev) +{ + u32 srbm_status; + u32 grbm_status; + u32 grbm_status2; + struct r100_gpu_lockup *lockup; + int r; + + if (rdev->family >= CHIP_RV770) + lockup = &rdev->config.rv770.lockup; + else + lockup = &rdev->config.r600.lockup; + + srbm_status = RREG32(R_000E50_SRBM_STATUS); + grbm_status = RREG32(R_008010_GRBM_STATUS); + grbm_status2 = RREG32(R_008014_GRBM_STATUS2); + if (!G_008010_GUI_ACTIVE(grbm_status)) { + r100_gpu_lockup_update(lockup, &rdev->cp); + return false; + } + /* force CP activities */ + r = radeon_ring_lock(rdev, 2); + if (!r) { + /* PACKET2 NOP */ + radeon_ring_write(rdev, 0x80000000); + radeon_ring_write(rdev, 0x80000000); + radeon_ring_unlock_commit(rdev); + } + rdev->cp.rptr = RREG32(R600_CP_RB_RPTR); + return r100_gpu_cp_is_lockup(rdev, lockup, &rdev->cp); +} + +int r600_asic_reset(struct radeon_device *rdev) +{ + return r600_gpu_soft_reset(rdev); +} + +static u32 r600_get_tile_pipe_to_backend_map(u32 num_tile_pipes, + u32 num_backends, + u32 backend_disable_mask) +{ + u32 backend_map = 0; + u32 enabled_backends_mask; + u32 enabled_backends_count; + u32 cur_pipe; + u32 swizzle_pipe[R6XX_MAX_PIPES]; + u32 cur_backend; + u32 i; + + if (num_tile_pipes > R6XX_MAX_PIPES) + num_tile_pipes = R6XX_MAX_PIPES; + if (num_tile_pipes < 1) + num_tile_pipes = 1; + if (num_backends > R6XX_MAX_BACKENDS) + num_backends = R6XX_MAX_BACKENDS; + if (num_backends < 1) + num_backends = 1; + + enabled_backends_mask = 0; + enabled_backends_count = 0; + for (i = 0; i < R6XX_MAX_BACKENDS; ++i) { + if (((backend_disable_mask >> i) & 1) == 0) { + enabled_backends_mask |= (1 << i); + ++enabled_backends_count; + } + if (enabled_backends_count == num_backends) + break; + } + + if (enabled_backends_count == 0) { + enabled_backends_mask = 1; + enabled_backends_count = 1; + } + + if (enabled_backends_count != num_backends) + num_backends = enabled_backends_count; + + memset((uint8_t *)&swizzle_pipe[0], 0, sizeof(u32) * R6XX_MAX_PIPES); + switch (num_tile_pipes) { + case 1: + swizzle_pipe[0] = 0; + break; + case 2: + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 1; + break; + case 3: + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 1; + swizzle_pipe[2] = 2; + break; + case 4: + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 1; + swizzle_pipe[2] = 2; + swizzle_pipe[3] = 3; + break; + case 5: + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 1; + swizzle_pipe[2] = 2; + swizzle_pipe[3] = 3; + swizzle_pipe[4] = 4; + break; + case 6: + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 2; + swizzle_pipe[2] = 4; + swizzle_pipe[3] = 5; + swizzle_pipe[4] = 1; + swizzle_pipe[5] = 3; + break; + case 7: + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 2; + swizzle_pipe[2] = 4; + swizzle_pipe[3] = 6; + swizzle_pipe[4] = 1; + swizzle_pipe[5] = 3; + swizzle_pipe[6] = 5; + break; + case 8: + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 2; + swizzle_pipe[2] = 4; + swizzle_pipe[3] = 6; + swizzle_pipe[4] = 1; + swizzle_pipe[5] = 3; + swizzle_pipe[6] = 5; + swizzle_pipe[7] = 7; + break; + } + + cur_backend = 0; + for (cur_pipe = 0; cur_pipe < num_tile_pipes; ++cur_pipe) { + while (((1 << cur_backend) & enabled_backends_mask) == 0) + cur_backend = (cur_backend + 1) % R6XX_MAX_BACKENDS; + + backend_map |= (u32)(((cur_backend & 3) << (swizzle_pipe[cur_pipe] * 2))); + + cur_backend = (cur_backend + 1) % R6XX_MAX_BACKENDS; + } + + return backend_map; +} + +int r600_count_pipe_bits(uint32_t val) +{ + int i, ret = 0; + + for (i = 0; i < 32; i++) { + ret += val & 1; + val >>= 1; + } + return ret; +} + +void r600_gpu_init(struct radeon_device *rdev) +{ + u32 tiling_config; + u32 ramcfg; + u32 backend_map; + u32 cc_rb_backend_disable; + u32 cc_gc_shader_pipe_config; + u32 tmp; + int i, j; + u32 sq_config; + u32 sq_gpr_resource_mgmt_1 = 0; + u32 sq_gpr_resource_mgmt_2 = 0; + u32 sq_thread_resource_mgmt = 0; + u32 sq_stack_resource_mgmt_1 = 0; + u32 sq_stack_resource_mgmt_2 = 0; + + /* FIXME: implement */ + switch (rdev->family) { + case CHIP_R600: + rdev->config.r600.max_pipes = 4; + rdev->config.r600.max_tile_pipes = 8; + rdev->config.r600.max_simds = 4; + rdev->config.r600.max_backends = 4; + rdev->config.r600.max_gprs = 256; + rdev->config.r600.max_threads = 192; + rdev->config.r600.max_stack_entries = 256; + rdev->config.r600.max_hw_contexts = 8; + rdev->config.r600.max_gs_threads = 16; + rdev->config.r600.sx_max_export_size = 128; + rdev->config.r600.sx_max_export_pos_size = 16; + rdev->config.r600.sx_max_export_smx_size = 128; + rdev->config.r600.sq_num_cf_insts = 2; + break; + case CHIP_RV630: + case CHIP_RV635: + rdev->config.r600.max_pipes = 2; + rdev->config.r600.max_tile_pipes = 2; + rdev->config.r600.max_simds = 3; + rdev->config.r600.max_backends = 1; + rdev->config.r600.max_gprs = 128; + rdev->config.r600.max_threads = 192; + rdev->config.r600.max_stack_entries = 128; + rdev->config.r600.max_hw_contexts = 8; + rdev->config.r600.max_gs_threads = 4; + rdev->config.r600.sx_max_export_size = 128; + rdev->config.r600.sx_max_export_pos_size = 16; + rdev->config.r600.sx_max_export_smx_size = 128; + rdev->config.r600.sq_num_cf_insts = 2; + break; + case CHIP_RV610: + case CHIP_RV620: + case CHIP_RS780: + case CHIP_RS880: + rdev->config.r600.max_pipes = 1; + rdev->config.r600.max_tile_pipes = 1; + rdev->config.r600.max_simds = 2; + rdev->config.r600.max_backends = 1; + rdev->config.r600.max_gprs = 128; + rdev->config.r600.max_threads = 192; + rdev->config.r600.max_stack_entries = 128; + rdev->config.r600.max_hw_contexts = 4; + rdev->config.r600.max_gs_threads = 4; + rdev->config.r600.sx_max_export_size = 128; + rdev->config.r600.sx_max_export_pos_size = 16; + rdev->config.r600.sx_max_export_smx_size = 128; + rdev->config.r600.sq_num_cf_insts = 1; + break; + case CHIP_RV670: + rdev->config.r600.max_pipes = 4; + rdev->config.r600.max_tile_pipes = 4; + rdev->config.r600.max_simds = 4; + rdev->config.r600.max_backends = 4; + rdev->config.r600.max_gprs = 192; + rdev->config.r600.max_threads = 192; + rdev->config.r600.max_stack_entries = 256; + rdev->config.r600.max_hw_contexts = 8; + rdev->config.r600.max_gs_threads = 16; + rdev->config.r600.sx_max_export_size = 128; + rdev->config.r600.sx_max_export_pos_size = 16; + rdev->config.r600.sx_max_export_smx_size = 128; + rdev->config.r600.sq_num_cf_insts = 2; + break; + default: + break; + } + + /* Initialize HDP */ + for (i = 0, j = 0; i < 32; i++, j += 0x18) { + WREG32((0x2c14 + j), 0x00000000); + WREG32((0x2c18 + j), 0x00000000); + WREG32((0x2c1c + j), 0x00000000); + WREG32((0x2c20 + j), 0x00000000); + WREG32((0x2c24 + j), 0x00000000); + } + + WREG32(GRBM_CNTL, GRBM_READ_TIMEOUT(0xff)); + + /* Setup tiling */ + tiling_config = 0; + ramcfg = RREG32(RAMCFG); + switch (rdev->config.r600.max_tile_pipes) { + case 1: + tiling_config |= PIPE_TILING(0); + break; + case 2: + tiling_config |= PIPE_TILING(1); + break; + case 4: + tiling_config |= PIPE_TILING(2); + break; + case 8: + tiling_config |= PIPE_TILING(3); + break; + default: + break; + } + rdev->config.r600.tiling_npipes = rdev->config.r600.max_tile_pipes; + rdev->config.r600.tiling_nbanks = 4 << ((ramcfg & NOOFBANK_MASK) >> NOOFBANK_SHIFT); + tiling_config |= BANK_TILING((ramcfg & NOOFBANK_MASK) >> NOOFBANK_SHIFT); + tiling_config |= GROUP_SIZE((ramcfg & BURSTLENGTH_MASK) >> BURSTLENGTH_SHIFT); + if ((ramcfg & BURSTLENGTH_MASK) >> BURSTLENGTH_SHIFT) + rdev->config.r600.tiling_group_size = 512; + else + rdev->config.r600.tiling_group_size = 256; + tmp = (ramcfg & NOOFROWS_MASK) >> NOOFROWS_SHIFT; + if (tmp > 3) { + tiling_config |= ROW_TILING(3); + tiling_config |= SAMPLE_SPLIT(3); + } else { + tiling_config |= ROW_TILING(tmp); + tiling_config |= SAMPLE_SPLIT(tmp); + } + tiling_config |= BANK_SWAPS(1); + + cc_rb_backend_disable = RREG32(CC_RB_BACKEND_DISABLE) & 0x00ff0000; + cc_rb_backend_disable |= + BACKEND_DISABLE((R6XX_MAX_BACKENDS_MASK << rdev->config.r600.max_backends) & R6XX_MAX_BACKENDS_MASK); + + cc_gc_shader_pipe_config = RREG32(CC_GC_SHADER_PIPE_CONFIG) & 0xffffff00; + cc_gc_shader_pipe_config |= + INACTIVE_QD_PIPES((R6XX_MAX_PIPES_MASK << rdev->config.r600.max_pipes) & R6XX_MAX_PIPES_MASK); + cc_gc_shader_pipe_config |= + INACTIVE_SIMDS((R6XX_MAX_SIMDS_MASK << rdev->config.r600.max_simds) & R6XX_MAX_SIMDS_MASK); + + backend_map = r600_get_tile_pipe_to_backend_map(rdev->config.r600.max_tile_pipes, + (R6XX_MAX_BACKENDS - + r600_count_pipe_bits((cc_rb_backend_disable & + R6XX_MAX_BACKENDS_MASK) >> 16)), + (cc_rb_backend_disable >> 16)); + rdev->config.r600.tile_config = tiling_config; + tiling_config |= BACKEND_MAP(backend_map); + WREG32(GB_TILING_CONFIG, tiling_config); + WREG32(DCP_TILING_CONFIG, tiling_config & 0xffff); + WREG32(HDP_TILING_CONFIG, tiling_config & 0xffff); + + /* Setup pipes */ + WREG32(CC_RB_BACKEND_DISABLE, cc_rb_backend_disable); + WREG32(CC_GC_SHADER_PIPE_CONFIG, cc_gc_shader_pipe_config); + WREG32(GC_USER_SHADER_PIPE_CONFIG, cc_gc_shader_pipe_config); + + tmp = R6XX_MAX_PIPES - r600_count_pipe_bits((cc_gc_shader_pipe_config & INACTIVE_QD_PIPES_MASK) >> 8); + WREG32(VGT_OUT_DEALLOC_CNTL, (tmp * 4) & DEALLOC_DIST_MASK); + WREG32(VGT_VERTEX_REUSE_BLOCK_CNTL, ((tmp * 4) - 2) & VTX_REUSE_DEPTH_MASK); + + /* Setup some CP states */ + WREG32(CP_QUEUE_THRESHOLDS, (ROQ_IB1_START(0x16) | ROQ_IB2_START(0x2b))); + WREG32(CP_MEQ_THRESHOLDS, (MEQ_END(0x40) | ROQ_END(0x40))); + + WREG32(TA_CNTL_AUX, (DISABLE_CUBE_ANISO | SYNC_GRADIENT | + SYNC_WALKER | SYNC_ALIGNER)); + /* Setup various GPU states */ + if (rdev->family == CHIP_RV670) + WREG32(ARB_GDEC_RD_CNTL, 0x00000021); + + tmp = RREG32(SX_DEBUG_1); + tmp |= SMX_EVENT_RELEASE; + if ((rdev->family > CHIP_R600)) + tmp |= ENABLE_NEW_SMX_ADDRESS; + WREG32(SX_DEBUG_1, tmp); + + if (((rdev->family) == CHIP_R600) || + ((rdev->family) == CHIP_RV630) || + ((rdev->family) == CHIP_RV610) || + ((rdev->family) == CHIP_RV620) || + ((rdev->family) == CHIP_RS780) || + ((rdev->family) == CHIP_RS880)) { + WREG32(DB_DEBUG, PREZ_MUST_WAIT_FOR_POSTZ_DONE); + } else { + WREG32(DB_DEBUG, 0); + } + WREG32(DB_WATERMARKS, (DEPTH_FREE(4) | DEPTH_CACHELINE_FREE(16) | + DEPTH_FLUSH(16) | DEPTH_PENDING_FREE(4))); + + WREG32(PA_SC_MULTI_CHIP_CNTL, 0); + WREG32(VGT_NUM_INSTANCES, 0); + + WREG32(SPI_CONFIG_CNTL, GPR_WRITE_PRIORITY(0)); + WREG32(SPI_CONFIG_CNTL_1, VTX_DONE_DELAY(0)); + + tmp = RREG32(SQ_MS_FIFO_SIZES); + if (((rdev->family) == CHIP_RV610) || + ((rdev->family) == CHIP_RV620) || + ((rdev->family) == CHIP_RS780) || + ((rdev->family) == CHIP_RS880)) { + tmp = (CACHE_FIFO_SIZE(0xa) | + FETCH_FIFO_HIWATER(0xa) | + DONE_FIFO_HIWATER(0xe0) | + ALU_UPDATE_FIFO_HIWATER(0x8)); + } else if (((rdev->family) == CHIP_R600) || + ((rdev->family) == CHIP_RV630)) { + tmp &= ~DONE_FIFO_HIWATER(0xff); + tmp |= DONE_FIFO_HIWATER(0x4); + } + WREG32(SQ_MS_FIFO_SIZES, tmp); + + /* SQ_CONFIG, SQ_GPR_RESOURCE_MGMT, SQ_THREAD_RESOURCE_MGMT, SQ_STACK_RESOURCE_MGMT + * should be adjusted as needed by the 2D/3D drivers. This just sets default values + */ + sq_config = RREG32(SQ_CONFIG); + sq_config &= ~(PS_PRIO(3) | + VS_PRIO(3) | + GS_PRIO(3) | + ES_PRIO(3)); + sq_config |= (DX9_CONSTS | + VC_ENABLE | + PS_PRIO(0) | + VS_PRIO(1) | + GS_PRIO(2) | + ES_PRIO(3)); + + if ((rdev->family) == CHIP_R600) { + sq_gpr_resource_mgmt_1 = (NUM_PS_GPRS(124) | + NUM_VS_GPRS(124) | + NUM_CLAUSE_TEMP_GPRS(4)); + sq_gpr_resource_mgmt_2 = (NUM_GS_GPRS(0) | + NUM_ES_GPRS(0)); + sq_thread_resource_mgmt = (NUM_PS_THREADS(136) | + NUM_VS_THREADS(48) | + NUM_GS_THREADS(4) | + NUM_ES_THREADS(4)); + sq_stack_resource_mgmt_1 = (NUM_PS_STACK_ENTRIES(128) | + NUM_VS_STACK_ENTRIES(128)); + sq_stack_resource_mgmt_2 = (NUM_GS_STACK_ENTRIES(0) | + NUM_ES_STACK_ENTRIES(0)); + } else if (((rdev->family) == CHIP_RV610) || + ((rdev->family) == CHIP_RV620) || + ((rdev->family) == CHIP_RS780) || + ((rdev->family) == CHIP_RS880)) { + /* no vertex cache */ + sq_config &= ~VC_ENABLE; + + sq_gpr_resource_mgmt_1 = (NUM_PS_GPRS(44) | + NUM_VS_GPRS(44) | + NUM_CLAUSE_TEMP_GPRS(2)); + sq_gpr_resource_mgmt_2 = (NUM_GS_GPRS(17) | + NUM_ES_GPRS(17)); + sq_thread_resource_mgmt = (NUM_PS_THREADS(79) | + NUM_VS_THREADS(78) | + NUM_GS_THREADS(4) | + NUM_ES_THREADS(31)); + sq_stack_resource_mgmt_1 = (NUM_PS_STACK_ENTRIES(40) | + NUM_VS_STACK_ENTRIES(40)); + sq_stack_resource_mgmt_2 = (NUM_GS_STACK_ENTRIES(32) | + NUM_ES_STACK_ENTRIES(16)); + } else if (((rdev->family) == CHIP_RV630) || + ((rdev->family) == CHIP_RV635)) { + sq_gpr_resource_mgmt_1 = (NUM_PS_GPRS(44) | + NUM_VS_GPRS(44) | + NUM_CLAUSE_TEMP_GPRS(2)); + sq_gpr_resource_mgmt_2 = (NUM_GS_GPRS(18) | + NUM_ES_GPRS(18)); + sq_thread_resource_mgmt = (NUM_PS_THREADS(79) | + NUM_VS_THREADS(78) | + NUM_GS_THREADS(4) | + NUM_ES_THREADS(31)); + sq_stack_resource_mgmt_1 = (NUM_PS_STACK_ENTRIES(40) | + NUM_VS_STACK_ENTRIES(40)); + sq_stack_resource_mgmt_2 = (NUM_GS_STACK_ENTRIES(32) | + NUM_ES_STACK_ENTRIES(16)); + } else if ((rdev->family) == CHIP_RV670) { + sq_gpr_resource_mgmt_1 = (NUM_PS_GPRS(44) | + NUM_VS_GPRS(44) | + NUM_CLAUSE_TEMP_GPRS(2)); + sq_gpr_resource_mgmt_2 = (NUM_GS_GPRS(17) | + NUM_ES_GPRS(17)); + sq_thread_resource_mgmt = (NUM_PS_THREADS(79) | + NUM_VS_THREADS(78) | + NUM_GS_THREADS(4) | + NUM_ES_THREADS(31)); + sq_stack_resource_mgmt_1 = (NUM_PS_STACK_ENTRIES(64) | + NUM_VS_STACK_ENTRIES(64)); + sq_stack_resource_mgmt_2 = (NUM_GS_STACK_ENTRIES(64) | + NUM_ES_STACK_ENTRIES(64)); + } + + WREG32(SQ_CONFIG, sq_config); + WREG32(SQ_GPR_RESOURCE_MGMT_1, sq_gpr_resource_mgmt_1); + WREG32(SQ_GPR_RESOURCE_MGMT_2, sq_gpr_resource_mgmt_2); + WREG32(SQ_THREAD_RESOURCE_MGMT, sq_thread_resource_mgmt); + WREG32(SQ_STACK_RESOURCE_MGMT_1, sq_stack_resource_mgmt_1); + WREG32(SQ_STACK_RESOURCE_MGMT_2, sq_stack_resource_mgmt_2); + + if (((rdev->family) == CHIP_RV610) || + ((rdev->family) == CHIP_RV620) || + ((rdev->family) == CHIP_RS780) || + ((rdev->family) == CHIP_RS880)) { + WREG32(VGT_CACHE_INVALIDATION, CACHE_INVALIDATION(TC_ONLY)); + } else { + WREG32(VGT_CACHE_INVALIDATION, CACHE_INVALIDATION(VC_AND_TC)); + } + + /* More default values. 2D/3D driver should adjust as needed */ + WREG32(PA_SC_AA_SAMPLE_LOCS_2S, (S0_X(0xc) | S0_Y(0x4) | + S1_X(0x4) | S1_Y(0xc))); + WREG32(PA_SC_AA_SAMPLE_LOCS_4S, (S0_X(0xe) | S0_Y(0xe) | + S1_X(0x2) | S1_Y(0x2) | + S2_X(0xa) | S2_Y(0x6) | + S3_X(0x6) | S3_Y(0xa))); + WREG32(PA_SC_AA_SAMPLE_LOCS_8S_WD0, (S0_X(0xe) | S0_Y(0xb) | + S1_X(0x4) | S1_Y(0xc) | + S2_X(0x1) | S2_Y(0x6) | + S3_X(0xa) | S3_Y(0xe))); + WREG32(PA_SC_AA_SAMPLE_LOCS_8S_WD1, (S4_X(0x6) | S4_Y(0x1) | + S5_X(0x0) | S5_Y(0x0) | + S6_X(0xb) | S6_Y(0x4) | + S7_X(0x7) | S7_Y(0x8))); + + WREG32(VGT_STRMOUT_EN, 0); + tmp = rdev->config.r600.max_pipes * 16; + switch (rdev->family) { + case CHIP_RV610: + case CHIP_RV620: + case CHIP_RS780: + case CHIP_RS880: + tmp += 32; + break; + case CHIP_RV670: + tmp += 128; + break; + default: + break; + } + if (tmp > 256) { + tmp = 256; + } + WREG32(VGT_ES_PER_GS, 128); + WREG32(VGT_GS_PER_ES, tmp); + WREG32(VGT_GS_PER_VS, 2); + WREG32(VGT_GS_VERTEX_REUSE, 16); + + /* more default values. 2D/3D driver should adjust as needed */ + WREG32(PA_SC_LINE_STIPPLE_STATE, 0); + WREG32(VGT_STRMOUT_EN, 0); + WREG32(SX_MISC, 0); + WREG32(PA_SC_MODE_CNTL, 0); + WREG32(PA_SC_AA_CONFIG, 0); + WREG32(PA_SC_LINE_STIPPLE, 0); + WREG32(SPI_INPUT_Z, 0); + WREG32(SPI_PS_IN_CONTROL_0, NUM_INTERP(2)); + WREG32(CB_COLOR7_FRAG, 0); + + /* Clear render buffer base addresses */ + WREG32(CB_COLOR0_BASE, 0); + WREG32(CB_COLOR1_BASE, 0); + WREG32(CB_COLOR2_BASE, 0); + WREG32(CB_COLOR3_BASE, 0); + WREG32(CB_COLOR4_BASE, 0); + WREG32(CB_COLOR5_BASE, 0); + WREG32(CB_COLOR6_BASE, 0); + WREG32(CB_COLOR7_BASE, 0); + WREG32(CB_COLOR7_FRAG, 0); + + switch (rdev->family) { + case CHIP_RV610: + case CHIP_RV620: + case CHIP_RS780: + case CHIP_RS880: + tmp = TC_L2_SIZE(8); + break; + case CHIP_RV630: + case CHIP_RV635: + tmp = TC_L2_SIZE(4); + break; + case CHIP_R600: + tmp = TC_L2_SIZE(0) | L2_DISABLE_LATE_HIT; + break; + default: + tmp = TC_L2_SIZE(0); + break; + } + WREG32(TC_CNTL, tmp); + + tmp = RREG32(HDP_HOST_PATH_CNTL); + WREG32(HDP_HOST_PATH_CNTL, tmp); + + tmp = RREG32(ARB_POP); + tmp |= ENABLE_TC128; + WREG32(ARB_POP, tmp); + + WREG32(PA_SC_MULTI_CHIP_CNTL, 0); + WREG32(PA_CL_ENHANCE, (CLIP_VTX_REORDER_ENA | + NUM_CLIP_SEQ(3))); + WREG32(PA_SC_ENHANCE, FORCE_EOV_MAX_CLK_CNT(4095)); +} + + +/* + * Indirect registers accessor + */ +u32 r600_pciep_rreg(struct radeon_device *rdev, u32 reg) +{ + u32 r; + + WREG32(PCIE_PORT_INDEX, ((reg) & 0xff)); + (void)RREG32(PCIE_PORT_INDEX); + r = RREG32(PCIE_PORT_DATA); + return r; +} + +void r600_pciep_wreg(struct radeon_device *rdev, u32 reg, u32 v) +{ + WREG32(PCIE_PORT_INDEX, ((reg) & 0xff)); + (void)RREG32(PCIE_PORT_INDEX); + WREG32(PCIE_PORT_DATA, (v)); + (void)RREG32(PCIE_PORT_DATA); +} + +/* + * CP & Ring + */ +void r600_cp_stop(struct radeon_device *rdev) +{ + radeon_ttm_set_active_vram_size(rdev, rdev->mc.visible_vram_size); + WREG32(R_0086D8_CP_ME_CNTL, S_0086D8_CP_ME_HALT(1)); + WREG32(SCRATCH_UMSK, 0); +} + +int r600_init_microcode(struct radeon_device *rdev) +{ + struct platform_device *pdev; + const char *chip_name; + const char *rlc_chip_name; + size_t pfp_req_size, me_req_size, rlc_req_size; + char fw_name[30]; + int err; + + DRM_DEBUG("\n"); + + pdev = platform_device_register_simple("radeon_cp", 0, NULL, 0); + err = IS_ERR(pdev); + if (err) { + printk(KERN_ERR "radeon_cp: Failed to register firmware\n"); + return -EINVAL; + } + + switch (rdev->family) { + case CHIP_R600: + chip_name = "R600"; + rlc_chip_name = "R600"; + break; + case CHIP_RV610: + chip_name = "RV610"; + rlc_chip_name = "R600"; + break; + case CHIP_RV630: + chip_name = "RV630"; + rlc_chip_name = "R600"; + break; + case CHIP_RV620: + chip_name = "RV620"; + rlc_chip_name = "R600"; + break; + case CHIP_RV635: + chip_name = "RV635"; + rlc_chip_name = "R600"; + break; + case CHIP_RV670: + chip_name = "RV670"; + rlc_chip_name = "R600"; + break; + case CHIP_RS780: + case CHIP_RS880: + chip_name = "RS780"; + rlc_chip_name = "R600"; + break; + case CHIP_RV770: + chip_name = "RV770"; + rlc_chip_name = "R700"; + break; + case CHIP_RV730: + case CHIP_RV740: + chip_name = "RV730"; + rlc_chip_name = "R700"; + break; + case CHIP_RV710: + chip_name = "RV710"; + rlc_chip_name = "R700"; + break; + case CHIP_CEDAR: + chip_name = "CEDAR"; + rlc_chip_name = "CEDAR"; + break; + case CHIP_REDWOOD: + chip_name = "REDWOOD"; + rlc_chip_name = "REDWOOD"; + break; + case CHIP_JUNIPER: + chip_name = "JUNIPER"; + rlc_chip_name = "JUNIPER"; + break; + case CHIP_CYPRESS: + case CHIP_HEMLOCK: + chip_name = "CYPRESS"; + rlc_chip_name = "CYPRESS"; + break; + case CHIP_PALM: + chip_name = "PALM"; + rlc_chip_name = "SUMO"; + break; + case CHIP_SUMO: + chip_name = "SUMO"; + rlc_chip_name = "SUMO"; + break; + case CHIP_SUMO2: + chip_name = "SUMO2"; + rlc_chip_name = "SUMO"; + break; + default: BUG(); + } + + if (rdev->family >= CHIP_CEDAR) { + pfp_req_size = EVERGREEN_PFP_UCODE_SIZE * 4; + me_req_size = EVERGREEN_PM4_UCODE_SIZE * 4; + rlc_req_size = EVERGREEN_RLC_UCODE_SIZE * 4; + } else if (rdev->family >= CHIP_RV770) { + pfp_req_size = R700_PFP_UCODE_SIZE * 4; + me_req_size = R700_PM4_UCODE_SIZE * 4; + rlc_req_size = R700_RLC_UCODE_SIZE * 4; + } else { + pfp_req_size = PFP_UCODE_SIZE * 4; + me_req_size = PM4_UCODE_SIZE * 12; + rlc_req_size = RLC_UCODE_SIZE * 4; + } + + DRM_INFO("Loading %s Microcode\n", chip_name); + + snprintf(fw_name, sizeof(fw_name), "radeon/%s_pfp.bin", chip_name); + err = request_firmware(&rdev->pfp_fw, fw_name, &pdev->dev); + if (err) + goto out; + if (rdev->pfp_fw->size != pfp_req_size) { + printk(KERN_ERR + "r600_cp: Bogus length %zu in firmware \"%s\"\n", + rdev->pfp_fw->size, fw_name); + err = -EINVAL; + goto out; + } + + snprintf(fw_name, sizeof(fw_name), "radeon/%s_me.bin", chip_name); + err = request_firmware(&rdev->me_fw, fw_name, &pdev->dev); + if (err) + goto out; + if (rdev->me_fw->size != me_req_size) { + printk(KERN_ERR + "r600_cp: Bogus length %zu in firmware \"%s\"\n", + rdev->me_fw->size, fw_name); + err = -EINVAL; + } + + snprintf(fw_name, sizeof(fw_name), "radeon/%s_rlc.bin", rlc_chip_name); + err = request_firmware(&rdev->rlc_fw, fw_name, &pdev->dev); + if (err) + goto out; + if (rdev->rlc_fw->size != rlc_req_size) { + printk(KERN_ERR + "r600_rlc: Bogus length %zu in firmware \"%s\"\n", + rdev->rlc_fw->size, fw_name); + err = -EINVAL; + } + +out: + platform_device_unregister(pdev); + + if (err) { + if (err != -EINVAL) + printk(KERN_ERR + "r600_cp: Failed to load firmware \"%s\"\n", + fw_name); + release_firmware(rdev->pfp_fw); + rdev->pfp_fw = NULL; + release_firmware(rdev->me_fw); + rdev->me_fw = NULL; + release_firmware(rdev->rlc_fw); + rdev->rlc_fw = NULL; + } + return err; +} + +static int r600_cp_load_microcode(struct radeon_device *rdev) +{ + const __be32 *fw_data; + int i; + + if (!rdev->me_fw || !rdev->pfp_fw) + return -EINVAL; + + r600_cp_stop(rdev); + + WREG32(CP_RB_CNTL, +#ifdef __BIG_ENDIAN + BUF_SWAP_32BIT | +#endif + RB_NO_UPDATE | RB_BLKSZ(15) | RB_BUFSZ(3)); + + /* Reset cp */ + WREG32(GRBM_SOFT_RESET, SOFT_RESET_CP); + RREG32(GRBM_SOFT_RESET); + mdelay(15); + WREG32(GRBM_SOFT_RESET, 0); + + WREG32(CP_ME_RAM_WADDR, 0); + + fw_data = (const __be32 *)rdev->me_fw->data; + WREG32(CP_ME_RAM_WADDR, 0); + for (i = 0; i < PM4_UCODE_SIZE * 3; i++) + WREG32(CP_ME_RAM_DATA, + be32_to_cpup(fw_data++)); + + fw_data = (const __be32 *)rdev->pfp_fw->data; + WREG32(CP_PFP_UCODE_ADDR, 0); + for (i = 0; i < PFP_UCODE_SIZE; i++) + WREG32(CP_PFP_UCODE_DATA, + be32_to_cpup(fw_data++)); + + WREG32(CP_PFP_UCODE_ADDR, 0); + WREG32(CP_ME_RAM_WADDR, 0); + WREG32(CP_ME_RAM_RADDR, 0); + return 0; +} + +int r600_cp_start(struct radeon_device *rdev) +{ + int r; + uint32_t cp_me; + + r = radeon_ring_lock(rdev, 7); + if (r) { + DRM_ERROR("radeon: cp failed to lock ring (%d).\n", r); + return r; + } + radeon_ring_write(rdev, PACKET3(PACKET3_ME_INITIALIZE, 5)); + radeon_ring_write(rdev, 0x1); + if (rdev->family >= CHIP_RV770) { + radeon_ring_write(rdev, 0x0); + radeon_ring_write(rdev, rdev->config.rv770.max_hw_contexts - 1); + } else { + radeon_ring_write(rdev, 0x3); + radeon_ring_write(rdev, rdev->config.r600.max_hw_contexts - 1); + } + radeon_ring_write(rdev, PACKET3_ME_INITIALIZE_DEVICE_ID(1)); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, 0); + radeon_ring_unlock_commit(rdev); + + cp_me = 0xff; + WREG32(R_0086D8_CP_ME_CNTL, cp_me); + return 0; +} + +int r600_cp_resume(struct radeon_device *rdev) +{ + u32 tmp; + u32 rb_bufsz; + int r; + + /* Reset cp */ + WREG32(GRBM_SOFT_RESET, SOFT_RESET_CP); + RREG32(GRBM_SOFT_RESET); + mdelay(15); + WREG32(GRBM_SOFT_RESET, 0); + + /* Set ring buffer size */ + rb_bufsz = drm_order(rdev->cp.ring_size / 8); + tmp = (drm_order(RADEON_GPU_PAGE_SIZE/8) << 8) | rb_bufsz; +#ifdef __BIG_ENDIAN + tmp |= BUF_SWAP_32BIT; +#endif + WREG32(CP_RB_CNTL, tmp); + WREG32(CP_SEM_WAIT_TIMER, 0x4); + + /* Set the write pointer delay */ + WREG32(CP_RB_WPTR_DELAY, 0); + + /* Initialize the ring buffer's read and write pointers */ + WREG32(CP_RB_CNTL, tmp | RB_RPTR_WR_ENA); + WREG32(CP_RB_RPTR_WR, 0); + rdev->cp.wptr = 0; + WREG32(CP_RB_WPTR, rdev->cp.wptr); + + /* set the wb address whether it's enabled or not */ + WREG32(CP_RB_RPTR_ADDR, +#ifdef __BIG_ENDIAN + RB_RPTR_SWAP(2) | +#endif + ((rdev->wb.gpu_addr + RADEON_WB_CP_RPTR_OFFSET) & 0xFFFFFFFC)); + WREG32(CP_RB_RPTR_ADDR_HI, upper_32_bits(rdev->wb.gpu_addr + RADEON_WB_CP_RPTR_OFFSET) & 0xFF); + WREG32(SCRATCH_ADDR, ((rdev->wb.gpu_addr + RADEON_WB_SCRATCH_OFFSET) >> 8) & 0xFFFFFFFF); + + if (rdev->wb.enabled) + WREG32(SCRATCH_UMSK, 0xff); + else { + tmp |= RB_NO_UPDATE; + WREG32(SCRATCH_UMSK, 0); + } + + mdelay(1); + WREG32(CP_RB_CNTL, tmp); + + WREG32(CP_RB_BASE, rdev->cp.gpu_addr >> 8); + WREG32(CP_DEBUG, (1 << 27) | (1 << 28)); + + rdev->cp.rptr = RREG32(CP_RB_RPTR); + + r600_cp_start(rdev); + rdev->cp.ready = true; + r = radeon_ring_test(rdev); + if (r) { + rdev->cp.ready = false; + return r; + } + return 0; +} + +void r600_cp_commit(struct radeon_device *rdev) +{ + WREG32(CP_RB_WPTR, rdev->cp.wptr); + (void)RREG32(CP_RB_WPTR); +} + +void r600_ring_init(struct radeon_device *rdev, unsigned ring_size) +{ + u32 rb_bufsz; + + /* Align ring size */ + rb_bufsz = drm_order(ring_size / 8); + ring_size = (1 << (rb_bufsz + 1)) * 4; + rdev->cp.ring_size = ring_size; + rdev->cp.align_mask = 16 - 1; +} + +void r600_cp_fini(struct radeon_device *rdev) +{ + r600_cp_stop(rdev); + radeon_ring_fini(rdev); +} + + +/* + * GPU scratch registers helpers function. + */ +void r600_scratch_init(struct radeon_device *rdev) +{ + int i; + + rdev->scratch.num_reg = 7; + rdev->scratch.reg_base = SCRATCH_REG0; + for (i = 0; i < rdev->scratch.num_reg; i++) { + rdev->scratch.free[i] = true; + rdev->scratch.reg[i] = rdev->scratch.reg_base + (i * 4); + } +} + +int r600_ring_test(struct radeon_device *rdev) +{ + uint32_t scratch; + uint32_t tmp = 0; + unsigned i; + int r; + + r = radeon_scratch_get(rdev, &scratch); + if (r) { + DRM_ERROR("radeon: cp failed to get scratch reg (%d).\n", r); + return r; + } + WREG32(scratch, 0xCAFEDEAD); + r = radeon_ring_lock(rdev, 3); + if (r) { + DRM_ERROR("radeon: cp failed to lock ring (%d).\n", r); + radeon_scratch_free(rdev, scratch); + return r; + } + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONFIG_REG, 1)); + radeon_ring_write(rdev, ((scratch - PACKET3_SET_CONFIG_REG_OFFSET) >> 2)); + radeon_ring_write(rdev, 0xDEADBEEF); + radeon_ring_unlock_commit(rdev); + for (i = 0; i < rdev->usec_timeout; i++) { + tmp = RREG32(scratch); + if (tmp == 0xDEADBEEF) + break; + DRM_UDELAY(1); + } + if (i < rdev->usec_timeout) { + DRM_INFO("ring test succeeded in %d usecs\n", i); + } else { + DRM_ERROR("radeon: ring test failed (scratch(0x%04X)=0x%08X)\n", + scratch, tmp); + r = -EINVAL; + } + radeon_scratch_free(rdev, scratch); + return r; +} + +void r600_fence_ring_emit(struct radeon_device *rdev, + struct radeon_fence *fence) +{ + if (rdev->wb.use_event) { + u64 addr = rdev->wb.gpu_addr + R600_WB_EVENT_OFFSET + + (u64)(rdev->fence_drv.scratch_reg - rdev->scratch.reg_base); + /* EVENT_WRITE_EOP - flush caches, send int */ + radeon_ring_write(rdev, PACKET3(PACKET3_EVENT_WRITE_EOP, 4)); + radeon_ring_write(rdev, EVENT_TYPE(CACHE_FLUSH_AND_INV_EVENT_TS) | EVENT_INDEX(5)); + radeon_ring_write(rdev, addr & 0xffffffff); + radeon_ring_write(rdev, (upper_32_bits(addr) & 0xff) | DATA_SEL(1) | INT_SEL(2)); + radeon_ring_write(rdev, fence->seq); + radeon_ring_write(rdev, 0); + } else { + radeon_ring_write(rdev, PACKET3(PACKET3_EVENT_WRITE, 0)); + radeon_ring_write(rdev, EVENT_TYPE(CACHE_FLUSH_AND_INV_EVENT) | EVENT_INDEX(0)); + /* wait for 3D idle clean */ + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONFIG_REG, 1)); + radeon_ring_write(rdev, (WAIT_UNTIL - PACKET3_SET_CONFIG_REG_OFFSET) >> 2); + radeon_ring_write(rdev, WAIT_3D_IDLE_bit | WAIT_3D_IDLECLEAN_bit); + /* Emit fence sequence & fire IRQ */ + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONFIG_REG, 1)); + radeon_ring_write(rdev, ((rdev->fence_drv.scratch_reg - PACKET3_SET_CONFIG_REG_OFFSET) >> 2)); + radeon_ring_write(rdev, fence->seq); + /* CP_INTERRUPT packet 3 no longer exists, use packet 0 */ + radeon_ring_write(rdev, PACKET0(CP_INT_STATUS, 0)); + radeon_ring_write(rdev, RB_INT_STAT); + } +} + +int r600_copy_blit(struct radeon_device *rdev, + uint64_t src_offset, + uint64_t dst_offset, + unsigned num_gpu_pages, + struct radeon_fence *fence) +{ + int r; + + mutex_lock(&rdev->r600_blit.mutex); + rdev->r600_blit.vb_ib = NULL; + r = r600_blit_prepare_copy(rdev, num_gpu_pages * RADEON_GPU_PAGE_SIZE); + if (r) { + if (rdev->r600_blit.vb_ib) + radeon_ib_free(rdev, &rdev->r600_blit.vb_ib); + mutex_unlock(&rdev->r600_blit.mutex); + return r; + } + r600_kms_blit_copy(rdev, src_offset, dst_offset, num_gpu_pages * RADEON_GPU_PAGE_SIZE); + r600_blit_done_copy(rdev, fence); + mutex_unlock(&rdev->r600_blit.mutex); + return 0; +} + +int r600_set_surface_reg(struct radeon_device *rdev, int reg, + uint32_t tiling_flags, uint32_t pitch, + uint32_t offset, uint32_t obj_size) +{ + /* FIXME: implement */ + return 0; +} + +void r600_clear_surface_reg(struct radeon_device *rdev, int reg) +{ + /* FIXME: implement */ +} + +int r600_startup(struct radeon_device *rdev) +{ + int r; + + /* enable pcie gen2 link */ + r600_pcie_gen2_enable(rdev); + + if (!rdev->me_fw || !rdev->pfp_fw || !rdev->rlc_fw) { + r = r600_init_microcode(rdev); + if (r) { + DRM_ERROR("Failed to load firmware!\n"); + return r; + } + } + + r600_mc_program(rdev); + if (rdev->flags & RADEON_IS_AGP) { + r600_agp_enable(rdev); + } else { + r = r600_pcie_gart_enable(rdev); + if (r) + return r; + } + r600_gpu_init(rdev); + r = r600_blit_init(rdev); + if (r) { + r600_blit_fini(rdev); + rdev->asic->copy = NULL; + dev_warn(rdev->dev, "failed blitter (%d) falling back to memcpy\n", r); + } + + /* allocate wb buffer */ + r = radeon_wb_init(rdev); + if (r) + return r; + + /* Enable IRQ */ + r = r600_irq_init(rdev); + if (r) { + DRM_ERROR("radeon: IH init failed (%d).\n", r); + radeon_irq_kms_fini(rdev); + return r; + } + r600_irq_set(rdev); + + r = radeon_ring_init(rdev, rdev->cp.ring_size); + if (r) + return r; + r = r600_cp_load_microcode(rdev); + if (r) + return r; + r = r600_cp_resume(rdev); + if (r) + return r; + + return 0; +} + +void r600_vga_set_state(struct radeon_device *rdev, bool state) +{ + uint32_t temp; + + temp = RREG32(CONFIG_CNTL); + if (state == false) { + temp &= ~(1<<0); + temp |= (1<<1); + } else { + temp &= ~(1<<1); + } + WREG32(CONFIG_CNTL, temp); +} + +int r600_resume(struct radeon_device *rdev) +{ + int r; + + /* Do not reset GPU before posting, on r600 hw unlike on r500 hw, + * posting will perform necessary task to bring back GPU into good + * shape. + */ + /* post card */ + atom_asic_init(rdev->mode_info.atom_context); + + r = r600_startup(rdev); + if (r) { + DRM_ERROR("r600 startup failed on resume\n"); + return r; + } + + r = r600_ib_test(rdev); + if (r) { + DRM_ERROR("radeon: failed testing IB (%d).\n", r); + return r; + } + + r = r600_audio_init(rdev); + if (r) { + DRM_ERROR("radeon: audio resume failed\n"); + return r; + } + + return r; +} + +int r600_suspend(struct radeon_device *rdev) +{ + int r; + + r600_audio_fini(rdev); + /* FIXME: we should wait for ring to be empty */ + r600_cp_stop(rdev); + rdev->cp.ready = false; + r600_irq_suspend(rdev); + radeon_wb_disable(rdev); + r600_pcie_gart_disable(rdev); + /* unpin shaders bo */ + if (rdev->r600_blit.shader_obj) { + r = radeon_bo_reserve(rdev->r600_blit.shader_obj, false); + if (!r) { + radeon_bo_unpin(rdev->r600_blit.shader_obj); + radeon_bo_unreserve(rdev->r600_blit.shader_obj); + } + } + return 0; +} + +/* Plan is to move initialization in that function and use + * helper function so that radeon_device_init pretty much + * do nothing more than calling asic specific function. This + * should also allow to remove a bunch of callback function + * like vram_info. + */ +int r600_init(struct radeon_device *rdev) +{ + int r; + + if (r600_debugfs_mc_info_init(rdev)) { + DRM_ERROR("Failed to register debugfs file for mc !\n"); + } + /* This don't do much */ + r = radeon_gem_init(rdev); + if (r) + return r; + /* Read BIOS */ + if (!radeon_get_bios(rdev)) { + if (ASIC_IS_AVIVO(rdev)) + return -EINVAL; + } + /* Must be an ATOMBIOS */ + if (!rdev->is_atom_bios) { + dev_err(rdev->dev, "Expecting atombios for R600 GPU\n"); + return -EINVAL; + } + r = radeon_atombios_init(rdev); + if (r) + return r; + /* Post card if necessary */ + if (!radeon_card_posted(rdev)) { + if (!rdev->bios) { + dev_err(rdev->dev, "Card not posted and no BIOS - ignoring\n"); + return -EINVAL; + } + DRM_INFO("GPU not posted. posting now...\n"); + atom_asic_init(rdev->mode_info.atom_context); + } + /* Initialize scratch registers */ + r600_scratch_init(rdev); + /* Initialize surface registers */ + radeon_surface_init(rdev); + /* Initialize clocks */ + radeon_get_clock_info(rdev->ddev); + /* Fence driver */ + r = radeon_fence_driver_init(rdev); + if (r) + return r; + if (rdev->flags & RADEON_IS_AGP) { + r = radeon_agp_init(rdev); + if (r) + radeon_agp_disable(rdev); + } + r = r600_mc_init(rdev); + if (r) + return r; + /* Memory manager */ + r = radeon_bo_init(rdev); + if (r) + return r; + + r = radeon_irq_kms_init(rdev); + if (r) + return r; + + rdev->cp.ring_obj = NULL; + r600_ring_init(rdev, 1024 * 1024); + + rdev->ih.ring_obj = NULL; + r600_ih_ring_init(rdev, 64 * 1024); + + r = r600_pcie_gart_init(rdev); + if (r) + return r; + + rdev->accel_working = true; + r = r600_startup(rdev); + if (r) { + dev_err(rdev->dev, "disabling GPU acceleration\n"); + r600_cp_fini(rdev); + r600_irq_fini(rdev); + radeon_wb_fini(rdev); + radeon_irq_kms_fini(rdev); + r600_pcie_gart_fini(rdev); + rdev->accel_working = false; + } + if (rdev->accel_working) { + r = radeon_ib_pool_init(rdev); + if (r) { + dev_err(rdev->dev, "IB initialization failed (%d).\n", r); + rdev->accel_working = false; + } else { + r = r600_ib_test(rdev); + if (r) { + dev_err(rdev->dev, "IB test failed (%d).\n", r); + rdev->accel_working = false; + } + } + } + + r = r600_audio_init(rdev); + if (r) + return r; /* TODO error handling */ + return 0; +} + +void r600_fini(struct radeon_device *rdev) +{ + r600_audio_fini(rdev); + r600_blit_fini(rdev); + r600_cp_fini(rdev); + r600_irq_fini(rdev); + radeon_wb_fini(rdev); + radeon_ib_pool_fini(rdev); + radeon_irq_kms_fini(rdev); + r600_pcie_gart_fini(rdev); + radeon_agp_fini(rdev); + radeon_gem_fini(rdev); + radeon_fence_driver_fini(rdev); + radeon_bo_fini(rdev); + radeon_atombios_fini(rdev); + kfree(rdev->bios); + rdev->bios = NULL; +} + + +/* + * CS stuff + */ +void r600_ring_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib) +{ + /* FIXME: implement */ + radeon_ring_write(rdev, PACKET3(PACKET3_INDIRECT_BUFFER, 2)); + radeon_ring_write(rdev, +#ifdef __BIG_ENDIAN + (2 << 0) | +#endif + (ib->gpu_addr & 0xFFFFFFFC)); + radeon_ring_write(rdev, upper_32_bits(ib->gpu_addr) & 0xFF); + radeon_ring_write(rdev, ib->length_dw); +} + +int r600_ib_test(struct radeon_device *rdev) +{ + struct radeon_ib *ib; + uint32_t scratch; + uint32_t tmp = 0; + unsigned i; + int r; + + r = radeon_scratch_get(rdev, &scratch); + if (r) { + DRM_ERROR("radeon: failed to get scratch reg (%d).\n", r); + return r; + } + WREG32(scratch, 0xCAFEDEAD); + r = radeon_ib_get(rdev, &ib); + if (r) { + DRM_ERROR("radeon: failed to get ib (%d).\n", r); + return r; + } + ib->ptr[0] = PACKET3(PACKET3_SET_CONFIG_REG, 1); + ib->ptr[1] = ((scratch - PACKET3_SET_CONFIG_REG_OFFSET) >> 2); + ib->ptr[2] = 0xDEADBEEF; + ib->ptr[3] = PACKET2(0); + ib->ptr[4] = PACKET2(0); + ib->ptr[5] = PACKET2(0); + ib->ptr[6] = PACKET2(0); + ib->ptr[7] = PACKET2(0); + ib->ptr[8] = PACKET2(0); + ib->ptr[9] = PACKET2(0); + ib->ptr[10] = PACKET2(0); + ib->ptr[11] = PACKET2(0); + ib->ptr[12] = PACKET2(0); + ib->ptr[13] = PACKET2(0); + ib->ptr[14] = PACKET2(0); + ib->ptr[15] = PACKET2(0); + ib->length_dw = 16; + r = radeon_ib_schedule(rdev, ib); + if (r) { + radeon_scratch_free(rdev, scratch); + radeon_ib_free(rdev, &ib); + DRM_ERROR("radeon: failed to schedule ib (%d).\n", r); + return r; + } + r = radeon_fence_wait(ib->fence, false); + if (r) { + DRM_ERROR("radeon: fence wait failed (%d).\n", r); + return r; + } + for (i = 0; i < rdev->usec_timeout; i++) { + tmp = RREG32(scratch); + if (tmp == 0xDEADBEEF) + break; + DRM_UDELAY(1); + } + if (i < rdev->usec_timeout) { + DRM_INFO("ib test succeeded in %u usecs\n", i); + } else { + DRM_ERROR("radeon: ib test failed (scratch(0x%04X)=0x%08X)\n", + scratch, tmp); + r = -EINVAL; + } + radeon_scratch_free(rdev, scratch); + radeon_ib_free(rdev, &ib); + return r; +} + +/* + * Interrupts + * + * Interrupts use a ring buffer on r6xx/r7xx hardware. It works pretty + * the same as the CP ring buffer, but in reverse. Rather than the CPU + * writing to the ring and the GPU consuming, the GPU writes to the ring + * and host consumes. As the host irq handler processes interrupts, it + * increments the rptr. When the rptr catches up with the wptr, all the + * current interrupts have been processed. + */ + +void r600_ih_ring_init(struct radeon_device *rdev, unsigned ring_size) +{ + u32 rb_bufsz; + + /* Align ring size */ + rb_bufsz = drm_order(ring_size / 4); + ring_size = (1 << rb_bufsz) * 4; + rdev->ih.ring_size = ring_size; + rdev->ih.ptr_mask = rdev->ih.ring_size - 1; + rdev->ih.rptr = 0; +} + +static int r600_ih_ring_alloc(struct radeon_device *rdev) +{ + int r; + + /* Allocate ring buffer */ + if (rdev->ih.ring_obj == NULL) { + r = radeon_bo_create(rdev, rdev->ih.ring_size, + PAGE_SIZE, true, + RADEON_GEM_DOMAIN_GTT, + &rdev->ih.ring_obj); + if (r) { + DRM_ERROR("radeon: failed to create ih ring buffer (%d).\n", r); + return r; + } + r = radeon_bo_reserve(rdev->ih.ring_obj, false); + if (unlikely(r != 0)) + return r; + r = radeon_bo_pin(rdev->ih.ring_obj, + RADEON_GEM_DOMAIN_GTT, + &rdev->ih.gpu_addr); + if (r) { + radeon_bo_unreserve(rdev->ih.ring_obj); + DRM_ERROR("radeon: failed to pin ih ring buffer (%d).\n", r); + return r; + } + r = radeon_bo_kmap(rdev->ih.ring_obj, + (void **)&rdev->ih.ring); + radeon_bo_unreserve(rdev->ih.ring_obj); + if (r) { + DRM_ERROR("radeon: failed to map ih ring buffer (%d).\n", r); + return r; + } + } + return 0; +} + +static void r600_ih_ring_fini(struct radeon_device *rdev) +{ + int r; + if (rdev->ih.ring_obj) { + r = radeon_bo_reserve(rdev->ih.ring_obj, false); + if (likely(r == 0)) { + radeon_bo_kunmap(rdev->ih.ring_obj); + radeon_bo_unpin(rdev->ih.ring_obj); + radeon_bo_unreserve(rdev->ih.ring_obj); + } + radeon_bo_unref(&rdev->ih.ring_obj); + rdev->ih.ring = NULL; + rdev->ih.ring_obj = NULL; + } +} + +void r600_rlc_stop(struct radeon_device *rdev) +{ + + if ((rdev->family >= CHIP_RV770) && + (rdev->family <= CHIP_RV740)) { + /* r7xx asics need to soft reset RLC before halting */ + WREG32(SRBM_SOFT_RESET, SOFT_RESET_RLC); + RREG32(SRBM_SOFT_RESET); + udelay(15000); + WREG32(SRBM_SOFT_RESET, 0); + RREG32(SRBM_SOFT_RESET); + } + + WREG32(RLC_CNTL, 0); +} + +static void r600_rlc_start(struct radeon_device *rdev) +{ + WREG32(RLC_CNTL, RLC_ENABLE); +} + +static int r600_rlc_init(struct radeon_device *rdev) +{ + u32 i; + const __be32 *fw_data; + + if (!rdev->rlc_fw) + return -EINVAL; + + r600_rlc_stop(rdev); + + WREG32(RLC_HB_BASE, 0); + WREG32(RLC_HB_CNTL, 0); + WREG32(RLC_HB_RPTR, 0); + WREG32(RLC_HB_WPTR, 0); + if (rdev->family <= CHIP_CAICOS) { + WREG32(RLC_HB_WPTR_LSB_ADDR, 0); + WREG32(RLC_HB_WPTR_MSB_ADDR, 0); + } + WREG32(RLC_MC_CNTL, 0); + WREG32(RLC_UCODE_CNTL, 0); + + fw_data = (const __be32 *)rdev->rlc_fw->data; + if (rdev->family >= CHIP_CAYMAN) { + for (i = 0; i < CAYMAN_RLC_UCODE_SIZE; i++) { + WREG32(RLC_UCODE_ADDR, i); + WREG32(RLC_UCODE_DATA, be32_to_cpup(fw_data++)); + } + } else if (rdev->family >= CHIP_CEDAR) { + for (i = 0; i < EVERGREEN_RLC_UCODE_SIZE; i++) { + WREG32(RLC_UCODE_ADDR, i); + WREG32(RLC_UCODE_DATA, be32_to_cpup(fw_data++)); + } + } else if (rdev->family >= CHIP_RV770) { + for (i = 0; i < R700_RLC_UCODE_SIZE; i++) { + WREG32(RLC_UCODE_ADDR, i); + WREG32(RLC_UCODE_DATA, be32_to_cpup(fw_data++)); + } + } else { + for (i = 0; i < RLC_UCODE_SIZE; i++) { + WREG32(RLC_UCODE_ADDR, i); + WREG32(RLC_UCODE_DATA, be32_to_cpup(fw_data++)); + } + } + WREG32(RLC_UCODE_ADDR, 0); + + r600_rlc_start(rdev); + + return 0; +} + +static void r600_enable_interrupts(struct radeon_device *rdev) +{ + u32 ih_cntl = RREG32(IH_CNTL); + u32 ih_rb_cntl = RREG32(IH_RB_CNTL); + + ih_cntl |= ENABLE_INTR; + ih_rb_cntl |= IH_RB_ENABLE; + WREG32(IH_CNTL, ih_cntl); + WREG32(IH_RB_CNTL, ih_rb_cntl); + rdev->ih.enabled = true; +} + +void r600_disable_interrupts(struct radeon_device *rdev) +{ + u32 ih_rb_cntl = RREG32(IH_RB_CNTL); + u32 ih_cntl = RREG32(IH_CNTL); + + ih_rb_cntl &= ~IH_RB_ENABLE; + ih_cntl &= ~ENABLE_INTR; + WREG32(IH_RB_CNTL, ih_rb_cntl); + WREG32(IH_CNTL, ih_cntl); + /* set rptr, wptr to 0 */ + WREG32(IH_RB_RPTR, 0); + WREG32(IH_RB_WPTR, 0); + rdev->ih.enabled = false; + rdev->ih.wptr = 0; + rdev->ih.rptr = 0; +} + +static void r600_disable_interrupt_state(struct radeon_device *rdev) +{ + u32 tmp; + + WREG32(CP_INT_CNTL, CNTX_BUSY_INT_ENABLE | CNTX_EMPTY_INT_ENABLE); + WREG32(GRBM_INT_CNTL, 0); + WREG32(DxMODE_INT_MASK, 0); + WREG32(D1GRPH_INTERRUPT_CONTROL, 0); + WREG32(D2GRPH_INTERRUPT_CONTROL, 0); + if (ASIC_IS_DCE3(rdev)) { + WREG32(DCE3_DACA_AUTODETECT_INT_CONTROL, 0); + WREG32(DCE3_DACB_AUTODETECT_INT_CONTROL, 0); + tmp = RREG32(DC_HPD1_INT_CONTROL) & DC_HPDx_INT_POLARITY; + WREG32(DC_HPD1_INT_CONTROL, tmp); + tmp = RREG32(DC_HPD2_INT_CONTROL) & DC_HPDx_INT_POLARITY; + WREG32(DC_HPD2_INT_CONTROL, tmp); + tmp = RREG32(DC_HPD3_INT_CONTROL) & DC_HPDx_INT_POLARITY; + WREG32(DC_HPD3_INT_CONTROL, tmp); + tmp = RREG32(DC_HPD4_INT_CONTROL) & DC_HPDx_INT_POLARITY; + WREG32(DC_HPD4_INT_CONTROL, tmp); + if (ASIC_IS_DCE32(rdev)) { + tmp = RREG32(DC_HPD5_INT_CONTROL) & DC_HPDx_INT_POLARITY; + WREG32(DC_HPD5_INT_CONTROL, tmp); + tmp = RREG32(DC_HPD6_INT_CONTROL) & DC_HPDx_INT_POLARITY; + WREG32(DC_HPD6_INT_CONTROL, tmp); + } + } else { + WREG32(DACA_AUTODETECT_INT_CONTROL, 0); + WREG32(DACB_AUTODETECT_INT_CONTROL, 0); + tmp = RREG32(DC_HOT_PLUG_DETECT1_INT_CONTROL) & DC_HOT_PLUG_DETECTx_INT_POLARITY; + WREG32(DC_HOT_PLUG_DETECT1_INT_CONTROL, tmp); + tmp = RREG32(DC_HOT_PLUG_DETECT2_INT_CONTROL) & DC_HOT_PLUG_DETECTx_INT_POLARITY; + WREG32(DC_HOT_PLUG_DETECT2_INT_CONTROL, tmp); + tmp = RREG32(DC_HOT_PLUG_DETECT3_INT_CONTROL) & DC_HOT_PLUG_DETECTx_INT_POLARITY; + WREG32(DC_HOT_PLUG_DETECT3_INT_CONTROL, tmp); + } +} + +int r600_irq_init(struct radeon_device *rdev) +{ + int ret = 0; + int rb_bufsz; + u32 interrupt_cntl, ih_cntl, ih_rb_cntl; + + /* allocate ring */ + ret = r600_ih_ring_alloc(rdev); + if (ret) + return ret; + + /* disable irqs */ + r600_disable_interrupts(rdev); + + /* init rlc */ + ret = r600_rlc_init(rdev); + if (ret) { + r600_ih_ring_fini(rdev); + return ret; + } + + /* setup interrupt control */ + /* set dummy read address to ring address */ + WREG32(INTERRUPT_CNTL2, rdev->ih.gpu_addr >> 8); + interrupt_cntl = RREG32(INTERRUPT_CNTL); + /* IH_DUMMY_RD_OVERRIDE=0 - dummy read disabled with msi, enabled without msi + * IH_DUMMY_RD_OVERRIDE=1 - dummy read controlled by IH_DUMMY_RD_EN + */ + interrupt_cntl &= ~IH_DUMMY_RD_OVERRIDE; + /* IH_REQ_NONSNOOP_EN=1 if ring is in non-cacheable memory, e.g., vram */ + interrupt_cntl &= ~IH_REQ_NONSNOOP_EN; + WREG32(INTERRUPT_CNTL, interrupt_cntl); + + WREG32(IH_RB_BASE, rdev->ih.gpu_addr >> 8); + rb_bufsz = drm_order(rdev->ih.ring_size / 4); + + ih_rb_cntl = (IH_WPTR_OVERFLOW_ENABLE | + IH_WPTR_OVERFLOW_CLEAR | + (rb_bufsz << 1)); + + if (rdev->wb.enabled) + ih_rb_cntl |= IH_WPTR_WRITEBACK_ENABLE; + + /* set the writeback address whether it's enabled or not */ + WREG32(IH_RB_WPTR_ADDR_LO, (rdev->wb.gpu_addr + R600_WB_IH_WPTR_OFFSET) & 0xFFFFFFFC); + WREG32(IH_RB_WPTR_ADDR_HI, upper_32_bits(rdev->wb.gpu_addr + R600_WB_IH_WPTR_OFFSET) & 0xFF); + + WREG32(IH_RB_CNTL, ih_rb_cntl); + + /* set rptr, wptr to 0 */ + WREG32(IH_RB_RPTR, 0); + WREG32(IH_RB_WPTR, 0); + + /* Default settings for IH_CNTL (disabled at first) */ + ih_cntl = MC_WRREQ_CREDIT(0x10) | MC_WR_CLEAN_CNT(0x10); + /* RPTR_REARM only works if msi's are enabled */ + if (rdev->msi_enabled) + ih_cntl |= RPTR_REARM; + +#ifdef __BIG_ENDIAN + ih_cntl |= IH_MC_SWAP(IH_MC_SWAP_32BIT); +#endif + WREG32(IH_CNTL, ih_cntl); + + /* force the active interrupt state to all disabled */ + if (rdev->family >= CHIP_CEDAR) + evergreen_disable_interrupt_state(rdev); + else + r600_disable_interrupt_state(rdev); + + /* enable irqs */ + r600_enable_interrupts(rdev); + + return ret; +} + +void r600_irq_suspend(struct radeon_device *rdev) +{ + r600_irq_disable(rdev); + r600_rlc_stop(rdev); +} + +void r600_irq_fini(struct radeon_device *rdev) +{ + r600_irq_suspend(rdev); + r600_ih_ring_fini(rdev); +} + +int r600_irq_set(struct radeon_device *rdev) +{ + u32 cp_int_cntl = CNTX_BUSY_INT_ENABLE | CNTX_EMPTY_INT_ENABLE; + u32 mode_int = 0; + u32 hpd1, hpd2, hpd3, hpd4 = 0, hpd5 = 0, hpd6 = 0; + u32 grbm_int_cntl = 0; + u32 hdmi1, hdmi2; + u32 d1grph = 0, d2grph = 0; + + if (!rdev->irq.installed) { + WARN(1, "Can't enable IRQ/MSI because no handler is installed\n"); + return -EINVAL; + } + /* don't enable anything if the ih is disabled */ + if (!rdev->ih.enabled) { + r600_disable_interrupts(rdev); + /* force the active interrupt state to all disabled */ + r600_disable_interrupt_state(rdev); + return 0; + } + + hdmi1 = RREG32(R600_HDMI_BLOCK1 + R600_HDMI_CNTL) & ~R600_HDMI_INT_EN; + if (ASIC_IS_DCE3(rdev)) { + hdmi2 = RREG32(R600_HDMI_BLOCK3 + R600_HDMI_CNTL) & ~R600_HDMI_INT_EN; + hpd1 = RREG32(DC_HPD1_INT_CONTROL) & ~DC_HPDx_INT_EN; + hpd2 = RREG32(DC_HPD2_INT_CONTROL) & ~DC_HPDx_INT_EN; + hpd3 = RREG32(DC_HPD3_INT_CONTROL) & ~DC_HPDx_INT_EN; + hpd4 = RREG32(DC_HPD4_INT_CONTROL) & ~DC_HPDx_INT_EN; + if (ASIC_IS_DCE32(rdev)) { + hpd5 = RREG32(DC_HPD5_INT_CONTROL) & ~DC_HPDx_INT_EN; + hpd6 = RREG32(DC_HPD6_INT_CONTROL) & ~DC_HPDx_INT_EN; + } + } else { + hdmi2 = RREG32(R600_HDMI_BLOCK2 + R600_HDMI_CNTL) & ~R600_HDMI_INT_EN; + hpd1 = RREG32(DC_HOT_PLUG_DETECT1_INT_CONTROL) & ~DC_HPDx_INT_EN; + hpd2 = RREG32(DC_HOT_PLUG_DETECT2_INT_CONTROL) & ~DC_HPDx_INT_EN; + hpd3 = RREG32(DC_HOT_PLUG_DETECT3_INT_CONTROL) & ~DC_HPDx_INT_EN; + } + + if (rdev->irq.sw_int) { + DRM_DEBUG("r600_irq_set: sw int\n"); + cp_int_cntl |= RB_INT_ENABLE; + cp_int_cntl |= TIME_STAMP_INT_ENABLE; + } + if (rdev->irq.crtc_vblank_int[0] || + rdev->irq.pflip[0]) { + DRM_DEBUG("r600_irq_set: vblank 0\n"); + mode_int |= D1MODE_VBLANK_INT_MASK; + } + if (rdev->irq.crtc_vblank_int[1] || + rdev->irq.pflip[1]) { + DRM_DEBUG("r600_irq_set: vblank 1\n"); + mode_int |= D2MODE_VBLANK_INT_MASK; + } + if (rdev->irq.hpd[0]) { + DRM_DEBUG("r600_irq_set: hpd 1\n"); + hpd1 |= DC_HPDx_INT_EN; + } + if (rdev->irq.hpd[1]) { + DRM_DEBUG("r600_irq_set: hpd 2\n"); + hpd2 |= DC_HPDx_INT_EN; + } + if (rdev->irq.hpd[2]) { + DRM_DEBUG("r600_irq_set: hpd 3\n"); + hpd3 |= DC_HPDx_INT_EN; + } + if (rdev->irq.hpd[3]) { + DRM_DEBUG("r600_irq_set: hpd 4\n"); + hpd4 |= DC_HPDx_INT_EN; + } + if (rdev->irq.hpd[4]) { + DRM_DEBUG("r600_irq_set: hpd 5\n"); + hpd5 |= DC_HPDx_INT_EN; + } + if (rdev->irq.hpd[5]) { + DRM_DEBUG("r600_irq_set: hpd 6\n"); + hpd6 |= DC_HPDx_INT_EN; + } + if (rdev->irq.hdmi[0]) { + DRM_DEBUG("r600_irq_set: hdmi 1\n"); + hdmi1 |= R600_HDMI_INT_EN; + } + if (rdev->irq.hdmi[1]) { + DRM_DEBUG("r600_irq_set: hdmi 2\n"); + hdmi2 |= R600_HDMI_INT_EN; + } + if (rdev->irq.gui_idle) { + DRM_DEBUG("gui idle\n"); + grbm_int_cntl |= GUI_IDLE_INT_ENABLE; + } + + WREG32(CP_INT_CNTL, cp_int_cntl); + WREG32(DxMODE_INT_MASK, mode_int); + WREG32(D1GRPH_INTERRUPT_CONTROL, d1grph); + WREG32(D2GRPH_INTERRUPT_CONTROL, d2grph); + WREG32(GRBM_INT_CNTL, grbm_int_cntl); + WREG32(R600_HDMI_BLOCK1 + R600_HDMI_CNTL, hdmi1); + if (ASIC_IS_DCE3(rdev)) { + WREG32(R600_HDMI_BLOCK3 + R600_HDMI_CNTL, hdmi2); + WREG32(DC_HPD1_INT_CONTROL, hpd1); + WREG32(DC_HPD2_INT_CONTROL, hpd2); + WREG32(DC_HPD3_INT_CONTROL, hpd3); + WREG32(DC_HPD4_INT_CONTROL, hpd4); + if (ASIC_IS_DCE32(rdev)) { + WREG32(DC_HPD5_INT_CONTROL, hpd5); + WREG32(DC_HPD6_INT_CONTROL, hpd6); + } + } else { + WREG32(R600_HDMI_BLOCK2 + R600_HDMI_CNTL, hdmi2); + WREG32(DC_HOT_PLUG_DETECT1_INT_CONTROL, hpd1); + WREG32(DC_HOT_PLUG_DETECT2_INT_CONTROL, hpd2); + WREG32(DC_HOT_PLUG_DETECT3_INT_CONTROL, hpd3); + } + + return 0; +} + +static inline void r600_irq_ack(struct radeon_device *rdev) +{ + u32 tmp; + + if (ASIC_IS_DCE3(rdev)) { + rdev->irq.stat_regs.r600.disp_int = RREG32(DCE3_DISP_INTERRUPT_STATUS); + rdev->irq.stat_regs.r600.disp_int_cont = RREG32(DCE3_DISP_INTERRUPT_STATUS_CONTINUE); + rdev->irq.stat_regs.r600.disp_int_cont2 = RREG32(DCE3_DISP_INTERRUPT_STATUS_CONTINUE2); + } else { + rdev->irq.stat_regs.r600.disp_int = RREG32(DISP_INTERRUPT_STATUS); + rdev->irq.stat_regs.r600.disp_int_cont = RREG32(DISP_INTERRUPT_STATUS_CONTINUE); + rdev->irq.stat_regs.r600.disp_int_cont2 = 0; + } + rdev->irq.stat_regs.r600.d1grph_int = RREG32(D1GRPH_INTERRUPT_STATUS); + rdev->irq.stat_regs.r600.d2grph_int = RREG32(D2GRPH_INTERRUPT_STATUS); + + if (rdev->irq.stat_regs.r600.d1grph_int & DxGRPH_PFLIP_INT_OCCURRED) + WREG32(D1GRPH_INTERRUPT_STATUS, DxGRPH_PFLIP_INT_CLEAR); + if (rdev->irq.stat_regs.r600.d2grph_int & DxGRPH_PFLIP_INT_OCCURRED) + WREG32(D2GRPH_INTERRUPT_STATUS, DxGRPH_PFLIP_INT_CLEAR); + if (rdev->irq.stat_regs.r600.disp_int & LB_D1_VBLANK_INTERRUPT) + WREG32(D1MODE_VBLANK_STATUS, DxMODE_VBLANK_ACK); + if (rdev->irq.stat_regs.r600.disp_int & LB_D1_VLINE_INTERRUPT) + WREG32(D1MODE_VLINE_STATUS, DxMODE_VLINE_ACK); + if (rdev->irq.stat_regs.r600.disp_int & LB_D2_VBLANK_INTERRUPT) + WREG32(D2MODE_VBLANK_STATUS, DxMODE_VBLANK_ACK); + if (rdev->irq.stat_regs.r600.disp_int & LB_D2_VLINE_INTERRUPT) + WREG32(D2MODE_VLINE_STATUS, DxMODE_VLINE_ACK); + if (rdev->irq.stat_regs.r600.disp_int & DC_HPD1_INTERRUPT) { + if (ASIC_IS_DCE3(rdev)) { + tmp = RREG32(DC_HPD1_INT_CONTROL); + tmp |= DC_HPDx_INT_ACK; + WREG32(DC_HPD1_INT_CONTROL, tmp); + } else { + tmp = RREG32(DC_HOT_PLUG_DETECT1_INT_CONTROL); + tmp |= DC_HPDx_INT_ACK; + WREG32(DC_HOT_PLUG_DETECT1_INT_CONTROL, tmp); + } + } + if (rdev->irq.stat_regs.r600.disp_int & DC_HPD2_INTERRUPT) { + if (ASIC_IS_DCE3(rdev)) { + tmp = RREG32(DC_HPD2_INT_CONTROL); + tmp |= DC_HPDx_INT_ACK; + WREG32(DC_HPD2_INT_CONTROL, tmp); + } else { + tmp = RREG32(DC_HOT_PLUG_DETECT2_INT_CONTROL); + tmp |= DC_HPDx_INT_ACK; + WREG32(DC_HOT_PLUG_DETECT2_INT_CONTROL, tmp); + } + } + if (rdev->irq.stat_regs.r600.disp_int_cont & DC_HPD3_INTERRUPT) { + if (ASIC_IS_DCE3(rdev)) { + tmp = RREG32(DC_HPD3_INT_CONTROL); + tmp |= DC_HPDx_INT_ACK; + WREG32(DC_HPD3_INT_CONTROL, tmp); + } else { + tmp = RREG32(DC_HOT_PLUG_DETECT3_INT_CONTROL); + tmp |= DC_HPDx_INT_ACK; + WREG32(DC_HOT_PLUG_DETECT3_INT_CONTROL, tmp); + } + } + if (rdev->irq.stat_regs.r600.disp_int_cont & DC_HPD4_INTERRUPT) { + tmp = RREG32(DC_HPD4_INT_CONTROL); + tmp |= DC_HPDx_INT_ACK; + WREG32(DC_HPD4_INT_CONTROL, tmp); + } + if (ASIC_IS_DCE32(rdev)) { + if (rdev->irq.stat_regs.r600.disp_int_cont2 & DC_HPD5_INTERRUPT) { + tmp = RREG32(DC_HPD5_INT_CONTROL); + tmp |= DC_HPDx_INT_ACK; + WREG32(DC_HPD5_INT_CONTROL, tmp); + } + if (rdev->irq.stat_regs.r600.disp_int_cont2 & DC_HPD6_INTERRUPT) { + tmp = RREG32(DC_HPD5_INT_CONTROL); + tmp |= DC_HPDx_INT_ACK; + WREG32(DC_HPD6_INT_CONTROL, tmp); + } + } + if (RREG32(R600_HDMI_BLOCK1 + R600_HDMI_STATUS) & R600_HDMI_INT_PENDING) { + WREG32_P(R600_HDMI_BLOCK1 + R600_HDMI_CNTL, R600_HDMI_INT_ACK, ~R600_HDMI_INT_ACK); + } + if (ASIC_IS_DCE3(rdev)) { + if (RREG32(R600_HDMI_BLOCK3 + R600_HDMI_STATUS) & R600_HDMI_INT_PENDING) { + WREG32_P(R600_HDMI_BLOCK3 + R600_HDMI_CNTL, R600_HDMI_INT_ACK, ~R600_HDMI_INT_ACK); + } + } else { + if (RREG32(R600_HDMI_BLOCK2 + R600_HDMI_STATUS) & R600_HDMI_INT_PENDING) { + WREG32_P(R600_HDMI_BLOCK2 + R600_HDMI_CNTL, R600_HDMI_INT_ACK, ~R600_HDMI_INT_ACK); + } + } +} + +void r600_irq_disable(struct radeon_device *rdev) +{ + r600_disable_interrupts(rdev); + /* Wait and acknowledge irq */ + mdelay(1); + r600_irq_ack(rdev); + r600_disable_interrupt_state(rdev); +} + +static inline u32 r600_get_ih_wptr(struct radeon_device *rdev) +{ + u32 wptr, tmp; + + if (rdev->wb.enabled) + wptr = le32_to_cpu(rdev->wb.wb[R600_WB_IH_WPTR_OFFSET/4]); + else + wptr = RREG32(IH_RB_WPTR); + + if (wptr & RB_OVERFLOW) { + /* When a ring buffer overflow happen start parsing interrupt + * from the last not overwritten vector (wptr + 16). Hopefully + * this should allow us to catchup. + */ + dev_warn(rdev->dev, "IH ring buffer overflow (0x%08X, %d, %d)\n", + wptr, rdev->ih.rptr, (wptr + 16) + rdev->ih.ptr_mask); + rdev->ih.rptr = (wptr + 16) & rdev->ih.ptr_mask; + tmp = RREG32(IH_RB_CNTL); + tmp |= IH_WPTR_OVERFLOW_CLEAR; + WREG32(IH_RB_CNTL, tmp); + } + return (wptr & rdev->ih.ptr_mask); +} + +/* r600 IV Ring + * Each IV ring entry is 128 bits: + * [7:0] - interrupt source id + * [31:8] - reserved + * [59:32] - interrupt source data + * [127:60] - reserved + * + * The basic interrupt vector entries + * are decoded as follows: + * src_id src_data description + * 1 0 D1 Vblank + * 1 1 D1 Vline + * 5 0 D2 Vblank + * 5 1 D2 Vline + * 19 0 FP Hot plug detection A + * 19 1 FP Hot plug detection B + * 19 2 DAC A auto-detection + * 19 3 DAC B auto-detection + * 21 4 HDMI block A + * 21 5 HDMI block B + * 176 - CP_INT RB + * 177 - CP_INT IB1 + * 178 - CP_INT IB2 + * 181 - EOP Interrupt + * 233 - GUI Idle + * + * Note, these are based on r600 and may need to be + * adjusted or added to on newer asics + */ + +int r600_irq_process(struct radeon_device *rdev) +{ + u32 wptr; + u32 rptr; + u32 src_id, src_data; + u32 ring_index; + unsigned long flags; + bool queue_hotplug = false; + + if (!rdev->ih.enabled || rdev->shutdown) + return IRQ_NONE; + + wptr = r600_get_ih_wptr(rdev); + rptr = rdev->ih.rptr; + DRM_DEBUG("r600_irq_process start: rptr %d, wptr %d\n", rptr, wptr); + + spin_lock_irqsave(&rdev->ih.lock, flags); + + if (rptr == wptr) { + spin_unlock_irqrestore(&rdev->ih.lock, flags); + return IRQ_NONE; + } + +restart_ih: + /* display interrupts */ + r600_irq_ack(rdev); + + rdev->ih.wptr = wptr; + while (rptr != wptr) { + /* wptr/rptr are in bytes! */ + ring_index = rptr / 4; + src_id = le32_to_cpu(rdev->ih.ring[ring_index]) & 0xff; + src_data = le32_to_cpu(rdev->ih.ring[ring_index + 1]) & 0xfffffff; + + switch (src_id) { + case 1: /* D1 vblank/vline */ + switch (src_data) { + case 0: /* D1 vblank */ + if (rdev->irq.stat_regs.r600.disp_int & LB_D1_VBLANK_INTERRUPT) { + if (rdev->irq.crtc_vblank_int[0]) { + drm_handle_vblank(rdev->ddev, 0); + rdev->pm.vblank_sync = true; + wake_up(&rdev->irq.vblank_queue); + } + if (rdev->irq.pflip[0]) + radeon_crtc_handle_flip(rdev, 0); + rdev->irq.stat_regs.r600.disp_int &= ~LB_D1_VBLANK_INTERRUPT; + DRM_DEBUG("IH: D1 vblank\n"); + } + break; + case 1: /* D1 vline */ + if (rdev->irq.stat_regs.r600.disp_int & LB_D1_VLINE_INTERRUPT) { + rdev->irq.stat_regs.r600.disp_int &= ~LB_D1_VLINE_INTERRUPT; + DRM_DEBUG("IH: D1 vline\n"); + } + break; + default: + DRM_DEBUG("Unhandled interrupt: %d %d\n", src_id, src_data); + break; + } + break; + case 5: /* D2 vblank/vline */ + switch (src_data) { + case 0: /* D2 vblank */ + if (rdev->irq.stat_regs.r600.disp_int & LB_D2_VBLANK_INTERRUPT) { + if (rdev->irq.crtc_vblank_int[1]) { + drm_handle_vblank(rdev->ddev, 1); + rdev->pm.vblank_sync = true; + wake_up(&rdev->irq.vblank_queue); + } + if (rdev->irq.pflip[1]) + radeon_crtc_handle_flip(rdev, 1); + rdev->irq.stat_regs.r600.disp_int &= ~LB_D2_VBLANK_INTERRUPT; + DRM_DEBUG("IH: D2 vblank\n"); + } + break; + case 1: /* D1 vline */ + if (rdev->irq.stat_regs.r600.disp_int & LB_D2_VLINE_INTERRUPT) { + rdev->irq.stat_regs.r600.disp_int &= ~LB_D2_VLINE_INTERRUPT; + DRM_DEBUG("IH: D2 vline\n"); + } + break; + default: + DRM_DEBUG("Unhandled interrupt: %d %d\n", src_id, src_data); + break; + } + break; + case 19: /* HPD/DAC hotplug */ + switch (src_data) { + case 0: + if (rdev->irq.stat_regs.r600.disp_int & DC_HPD1_INTERRUPT) { + rdev->irq.stat_regs.r600.disp_int &= ~DC_HPD1_INTERRUPT; + queue_hotplug = true; + DRM_DEBUG("IH: HPD1\n"); + } + break; + case 1: + if (rdev->irq.stat_regs.r600.disp_int & DC_HPD2_INTERRUPT) { + rdev->irq.stat_regs.r600.disp_int &= ~DC_HPD2_INTERRUPT; + queue_hotplug = true; + DRM_DEBUG("IH: HPD2\n"); + } + break; + case 4: + if (rdev->irq.stat_regs.r600.disp_int_cont & DC_HPD3_INTERRUPT) { + rdev->irq.stat_regs.r600.disp_int_cont &= ~DC_HPD3_INTERRUPT; + queue_hotplug = true; + DRM_DEBUG("IH: HPD3\n"); + } + break; + case 5: + if (rdev->irq.stat_regs.r600.disp_int_cont & DC_HPD4_INTERRUPT) { + rdev->irq.stat_regs.r600.disp_int_cont &= ~DC_HPD4_INTERRUPT; + queue_hotplug = true; + DRM_DEBUG("IH: HPD4\n"); + } + break; + case 10: + if (rdev->irq.stat_regs.r600.disp_int_cont2 & DC_HPD5_INTERRUPT) { + rdev->irq.stat_regs.r600.disp_int_cont2 &= ~DC_HPD5_INTERRUPT; + queue_hotplug = true; + DRM_DEBUG("IH: HPD5\n"); + } + break; + case 12: + if (rdev->irq.stat_regs.r600.disp_int_cont2 & DC_HPD6_INTERRUPT) { + rdev->irq.stat_regs.r600.disp_int_cont2 &= ~DC_HPD6_INTERRUPT; + queue_hotplug = true; + DRM_DEBUG("IH: HPD6\n"); + } + break; + default: + DRM_DEBUG("Unhandled interrupt: %d %d\n", src_id, src_data); + break; + } + break; + case 21: /* HDMI */ + DRM_DEBUG("IH: HDMI: 0x%x\n", src_data); + r600_audio_schedule_polling(rdev); + break; + case 176: /* CP_INT in ring buffer */ + case 177: /* CP_INT in IB1 */ + case 178: /* CP_INT in IB2 */ + DRM_DEBUG("IH: CP int: 0x%08x\n", src_data); + radeon_fence_process(rdev); + break; + case 181: /* CP EOP event */ + DRM_DEBUG("IH: CP EOP\n"); + radeon_fence_process(rdev); + break; + case 233: /* GUI IDLE */ + DRM_DEBUG("IH: GUI idle\n"); + rdev->pm.gui_idle = true; + wake_up(&rdev->irq.idle_queue); + break; + default: + DRM_DEBUG("Unhandled interrupt: %d %d\n", src_id, src_data); + break; + } + + /* wptr/rptr are in bytes! */ + rptr += 16; + rptr &= rdev->ih.ptr_mask; + } + /* make sure wptr hasn't changed while processing */ + wptr = r600_get_ih_wptr(rdev); + if (wptr != rdev->ih.wptr) + goto restart_ih; + if (queue_hotplug) + schedule_work(&rdev->hotplug_work); + rdev->ih.rptr = rptr; + WREG32(IH_RB_RPTR, rdev->ih.rptr); + spin_unlock_irqrestore(&rdev->ih.lock, flags); + return IRQ_HANDLED; +} + +/* + * Debugfs info + */ +#if defined(CONFIG_DEBUG_FS) + +static int r600_debugfs_cp_ring_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + struct radeon_device *rdev = dev->dev_private; + unsigned count, i, j; + + radeon_ring_free_size(rdev); + count = (rdev->cp.ring_size / 4) - rdev->cp.ring_free_dw; + seq_printf(m, "CP_STAT 0x%08x\n", RREG32(CP_STAT)); + seq_printf(m, "CP_RB_WPTR 0x%08x\n", RREG32(CP_RB_WPTR)); + seq_printf(m, "CP_RB_RPTR 0x%08x\n", RREG32(CP_RB_RPTR)); + seq_printf(m, "driver's copy of the CP_RB_WPTR 0x%08x\n", rdev->cp.wptr); + seq_printf(m, "driver's copy of the CP_RB_RPTR 0x%08x\n", rdev->cp.rptr); + seq_printf(m, "%u free dwords in ring\n", rdev->cp.ring_free_dw); + seq_printf(m, "%u dwords in ring\n", count); + i = rdev->cp.rptr; + for (j = 0; j <= count; j++) { + seq_printf(m, "r[%04d]=0x%08x\n", i, rdev->cp.ring[i]); + i = (i + 1) & rdev->cp.ptr_mask; + } + return 0; +} + +static int r600_debugfs_mc_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + struct radeon_device *rdev = dev->dev_private; + + DREG32_SYS(m, rdev, R_000E50_SRBM_STATUS); + DREG32_SYS(m, rdev, VM_L2_STATUS); + return 0; +} + +static struct drm_info_list r600_mc_info_list[] = { + {"r600_mc_info", r600_debugfs_mc_info, 0, NULL}, + {"r600_ring_info", r600_debugfs_cp_ring_info, 0, NULL}, +}; +#endif + +int r600_debugfs_mc_info_init(struct radeon_device *rdev) +{ +#if defined(CONFIG_DEBUG_FS) + return radeon_debugfs_add_files(rdev, r600_mc_info_list, ARRAY_SIZE(r600_mc_info_list)); +#else + return 0; +#endif +} + +/** + * r600_ioctl_wait_idle - flush host path cache on wait idle ioctl + * rdev: radeon device structure + * bo: buffer object struct which userspace is waiting for idle + * + * Some R6XX/R7XX doesn't seems to take into account HDP flush performed + * through ring buffer, this leads to corruption in rendering, see + * http://bugzilla.kernel.org/show_bug.cgi?id=15186 to avoid this we + * directly perform HDP flush by writing register through MMIO. + */ +void r600_ioctl_wait_idle(struct radeon_device *rdev, struct radeon_bo *bo) +{ + /* r7xx hw bug. write to HDP_DEBUG1 followed by fb read + * rather than write to HDP_REG_COHERENCY_FLUSH_CNTL. + * This seems to cause problems on some AGP cards. Just use the old + * method for them. + */ + if ((rdev->family >= CHIP_RV770) && (rdev->family <= CHIP_RV740) && + rdev->vram_scratch.ptr && !(rdev->flags & RADEON_IS_AGP)) { + void __iomem *ptr = (void *)rdev->vram_scratch.ptr; + u32 tmp; + + WREG32(HDP_DEBUG1, 0); + tmp = readl((void __iomem *)ptr); + } else + WREG32(R_005480_HDP_MEM_COHERENCY_FLUSH_CNTL, 0x1); +} + +void r600_set_pcie_lanes(struct radeon_device *rdev, int lanes) +{ + u32 link_width_cntl, mask, target_reg; + + if (rdev->flags & RADEON_IS_IGP) + return; + + if (!(rdev->flags & RADEON_IS_PCIE)) + return; + + /* x2 cards have a special sequence */ + if (ASIC_IS_X2(rdev)) + return; + + /* FIXME wait for idle */ + + switch (lanes) { + case 0: + mask = RADEON_PCIE_LC_LINK_WIDTH_X0; + break; + case 1: + mask = RADEON_PCIE_LC_LINK_WIDTH_X1; + break; + case 2: + mask = RADEON_PCIE_LC_LINK_WIDTH_X2; + break; + case 4: + mask = RADEON_PCIE_LC_LINK_WIDTH_X4; + break; + case 8: + mask = RADEON_PCIE_LC_LINK_WIDTH_X8; + break; + case 12: + mask = RADEON_PCIE_LC_LINK_WIDTH_X12; + break; + case 16: + default: + mask = RADEON_PCIE_LC_LINK_WIDTH_X16; + break; + } + + link_width_cntl = RREG32_PCIE_P(RADEON_PCIE_LC_LINK_WIDTH_CNTL); + + if ((link_width_cntl & RADEON_PCIE_LC_LINK_WIDTH_RD_MASK) == + (mask << RADEON_PCIE_LC_LINK_WIDTH_RD_SHIFT)) + return; + + if (link_width_cntl & R600_PCIE_LC_UPCONFIGURE_DIS) + return; + + link_width_cntl &= ~(RADEON_PCIE_LC_LINK_WIDTH_MASK | + RADEON_PCIE_LC_RECONFIG_NOW | + R600_PCIE_LC_RENEGOTIATE_EN | + R600_PCIE_LC_RECONFIG_ARC_MISSING_ESCAPE); + link_width_cntl |= mask; + + WREG32_PCIE_P(RADEON_PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl); + + /* some northbridges can renegotiate the link rather than requiring + * a complete re-config. + * e.g., AMD 780/790 northbridges (pci ids: 0x5956, 0x5957, 0x5958, etc.) + */ + if (link_width_cntl & R600_PCIE_LC_RENEGOTIATION_SUPPORT) + link_width_cntl |= R600_PCIE_LC_RENEGOTIATE_EN | R600_PCIE_LC_UPCONFIGURE_SUPPORT; + else + link_width_cntl |= R600_PCIE_LC_RECONFIG_ARC_MISSING_ESCAPE; + + WREG32_PCIE_P(RADEON_PCIE_LC_LINK_WIDTH_CNTL, (link_width_cntl | + RADEON_PCIE_LC_RECONFIG_NOW)); + + if (rdev->family >= CHIP_RV770) + target_reg = R700_TARGET_AND_CURRENT_PROFILE_INDEX; + else + target_reg = R600_TARGET_AND_CURRENT_PROFILE_INDEX; + + /* wait for lane set to complete */ + link_width_cntl = RREG32(target_reg); + while (link_width_cntl == 0xffffffff) + link_width_cntl = RREG32(target_reg); + +} + +int r600_get_pcie_lanes(struct radeon_device *rdev) +{ + u32 link_width_cntl; + + if (rdev->flags & RADEON_IS_IGP) + return 0; + + if (!(rdev->flags & RADEON_IS_PCIE)) + return 0; + + /* x2 cards have a special sequence */ + if (ASIC_IS_X2(rdev)) + return 0; + + /* FIXME wait for idle */ + + link_width_cntl = RREG32_PCIE_P(RADEON_PCIE_LC_LINK_WIDTH_CNTL); + + switch ((link_width_cntl & RADEON_PCIE_LC_LINK_WIDTH_RD_MASK) >> RADEON_PCIE_LC_LINK_WIDTH_RD_SHIFT) { + case RADEON_PCIE_LC_LINK_WIDTH_X0: + return 0; + case RADEON_PCIE_LC_LINK_WIDTH_X1: + return 1; + case RADEON_PCIE_LC_LINK_WIDTH_X2: + return 2; + case RADEON_PCIE_LC_LINK_WIDTH_X4: + return 4; + case RADEON_PCIE_LC_LINK_WIDTH_X8: + return 8; + case RADEON_PCIE_LC_LINK_WIDTH_X16: + default: + return 16; + } +} + +static void r600_pcie_gen2_enable(struct radeon_device *rdev) +{ + u32 link_width_cntl, lanes, speed_cntl, training_cntl, tmp; + u16 link_cntl2; + + if (radeon_pcie_gen2 == 0) + return; + + if (rdev->flags & RADEON_IS_IGP) + return; + + if (!(rdev->flags & RADEON_IS_PCIE)) + return; + + /* x2 cards have a special sequence */ + if (ASIC_IS_X2(rdev)) + return; + + /* only RV6xx+ chips are supported */ + if (rdev->family <= CHIP_R600) + return; + + /* 55 nm r6xx asics */ + if ((rdev->family == CHIP_RV670) || + (rdev->family == CHIP_RV620) || + (rdev->family == CHIP_RV635)) { + /* advertise upconfig capability */ + link_width_cntl = RREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL); + link_width_cntl &= ~LC_UPCONFIGURE_DIS; + WREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl); + link_width_cntl = RREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL); + if (link_width_cntl & LC_RENEGOTIATION_SUPPORT) { + lanes = (link_width_cntl & LC_LINK_WIDTH_RD_MASK) >> LC_LINK_WIDTH_RD_SHIFT; + link_width_cntl &= ~(LC_LINK_WIDTH_MASK | + LC_RECONFIG_ARC_MISSING_ESCAPE); + link_width_cntl |= lanes | LC_RECONFIG_NOW | LC_RENEGOTIATE_EN; + WREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl); + } else { + link_width_cntl |= LC_UPCONFIGURE_DIS; + WREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl); + } + } + + speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL); + if ((speed_cntl & LC_OTHER_SIDE_EVER_SENT_GEN2) && + (speed_cntl & LC_OTHER_SIDE_SUPPORTS_GEN2)) { + + /* 55 nm r6xx asics */ + if ((rdev->family == CHIP_RV670) || + (rdev->family == CHIP_RV620) || + (rdev->family == CHIP_RV635)) { + WREG32(MM_CFGREGS_CNTL, 0x8); + link_cntl2 = RREG32(0x4088); + WREG32(MM_CFGREGS_CNTL, 0); + /* not supported yet */ + if (link_cntl2 & SELECTABLE_DEEMPHASIS) + return; + } + + speed_cntl &= ~LC_SPEED_CHANGE_ATTEMPTS_ALLOWED_MASK; + speed_cntl |= (0x3 << LC_SPEED_CHANGE_ATTEMPTS_ALLOWED_SHIFT); + speed_cntl &= ~LC_VOLTAGE_TIMER_SEL_MASK; + speed_cntl &= ~LC_FORCE_DIS_HW_SPEED_CHANGE; + speed_cntl |= LC_FORCE_EN_HW_SPEED_CHANGE; + WREG32_PCIE_P(PCIE_LC_SPEED_CNTL, speed_cntl); + + tmp = RREG32(0x541c); + WREG32(0x541c, tmp | 0x8); + WREG32(MM_CFGREGS_CNTL, MM_WR_TO_CFG_EN); + link_cntl2 = RREG16(0x4088); + link_cntl2 &= ~TARGET_LINK_SPEED_MASK; + link_cntl2 |= 0x2; + WREG16(0x4088, link_cntl2); + WREG32(MM_CFGREGS_CNTL, 0); + + if ((rdev->family == CHIP_RV670) || + (rdev->family == CHIP_RV620) || + (rdev->family == CHIP_RV635)) { + training_cntl = RREG32_PCIE_P(PCIE_LC_TRAINING_CNTL); + training_cntl &= ~LC_POINT_7_PLUS_EN; + WREG32_PCIE_P(PCIE_LC_TRAINING_CNTL, training_cntl); + } else { + speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL); + speed_cntl &= ~LC_TARGET_LINK_SPEED_OVERRIDE_EN; + WREG32_PCIE_P(PCIE_LC_SPEED_CNTL, speed_cntl); + } + + speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL); + speed_cntl |= LC_GEN2_EN_STRAP; + WREG32_PCIE_P(PCIE_LC_SPEED_CNTL, speed_cntl); + + } else { + link_width_cntl = RREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL); + /* XXX: only disable it if gen1 bridge vendor == 0x111d or 0x1106 */ + if (1) + link_width_cntl |= LC_UPCONFIGURE_DIS; + else + link_width_cntl &= ~LC_UPCONFIGURE_DIS; + WREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl); + } +} diff --git a/drivers/gpu/drm/radeon/r600_audio.c b/drivers/gpu/drm/radeon/r600_audio.c new file mode 100644 index 00000000..846fae57 --- /dev/null +++ b/drivers/gpu/drm/radeon/r600_audio.c @@ -0,0 +1,282 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Christian König. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Christian König + */ +#include "drmP.h" +#include "radeon.h" +#include "radeon_reg.h" +#include "radeon_asic.h" +#include "atom.h" + +#define AUDIO_TIMER_INTERVALL 100 /* 1/10 sekund should be enough */ + +/* + * check if the chipset is supported + */ +static int r600_audio_chipset_supported(struct radeon_device *rdev) +{ + return (rdev->family >= CHIP_R600 && rdev->family < CHIP_CEDAR) + || rdev->family == CHIP_RS600 + || rdev->family == CHIP_RS690 + || rdev->family == CHIP_RS740; +} + +/* + * current number of channels + */ +int r600_audio_channels(struct radeon_device *rdev) +{ + return (RREG32(R600_AUDIO_RATE_BPS_CHANNEL) & 0x7) + 1; +} + +/* + * current bits per sample + */ +int r600_audio_bits_per_sample(struct radeon_device *rdev) +{ + uint32_t value = (RREG32(R600_AUDIO_RATE_BPS_CHANNEL) & 0xF0) >> 4; + switch (value) { + case 0x0: return 8; + case 0x1: return 16; + case 0x2: return 20; + case 0x3: return 24; + case 0x4: return 32; + } + + dev_err(rdev->dev, "Unknown bits per sample 0x%x using 16 instead\n", + (int)value); + + return 16; +} + +/* + * current sampling rate in HZ + */ +int r600_audio_rate(struct radeon_device *rdev) +{ + uint32_t value = RREG32(R600_AUDIO_RATE_BPS_CHANNEL); + uint32_t result; + + if (value & 0x4000) + result = 44100; + else + result = 48000; + + result *= ((value >> 11) & 0x7) + 1; + result /= ((value >> 8) & 0x7) + 1; + + return result; +} + +/* + * iec 60958 status bits + */ +uint8_t r600_audio_status_bits(struct radeon_device *rdev) +{ + return RREG32(R600_AUDIO_STATUS_BITS) & 0xff; +} + +/* + * iec 60958 category code + */ +uint8_t r600_audio_category_code(struct radeon_device *rdev) +{ + return (RREG32(R600_AUDIO_STATUS_BITS) >> 8) & 0xff; +} + +/* + * schedule next audio update event + */ +void r600_audio_schedule_polling(struct radeon_device *rdev) +{ + mod_timer(&rdev->audio_timer, + jiffies + msecs_to_jiffies(AUDIO_TIMER_INTERVALL)); +} + +/* + * update all hdmi interfaces with current audio parameters + */ +static void r600_audio_update_hdmi(unsigned long param) +{ + struct radeon_device *rdev = (struct radeon_device *)param; + struct drm_device *dev = rdev->ddev; + + int channels = r600_audio_channels(rdev); + int rate = r600_audio_rate(rdev); + int bps = r600_audio_bits_per_sample(rdev); + uint8_t status_bits = r600_audio_status_bits(rdev); + uint8_t category_code = r600_audio_category_code(rdev); + + struct drm_encoder *encoder; + int changes = 0, still_going = 0; + + changes |= channels != rdev->audio_channels; + changes |= rate != rdev->audio_rate; + changes |= bps != rdev->audio_bits_per_sample; + changes |= status_bits != rdev->audio_status_bits; + changes |= category_code != rdev->audio_category_code; + + if (changes) { + rdev->audio_channels = channels; + rdev->audio_rate = rate; + rdev->audio_bits_per_sample = bps; + rdev->audio_status_bits = status_bits; + rdev->audio_category_code = category_code; + } + + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + still_going |= radeon_encoder->audio_polling_active; + if (changes || r600_hdmi_buffer_status_changed(encoder)) + r600_hdmi_update_audio_settings(encoder); + } + + if (still_going) + r600_audio_schedule_polling(rdev); +} + +/* + * turn on/off audio engine + */ +static void r600_audio_engine_enable(struct radeon_device *rdev, bool enable) +{ + DRM_INFO("%s audio support\n", enable ? "Enabling" : "Disabling"); + WREG32_P(R600_AUDIO_ENABLE, enable ? 0x81000000 : 0x0, ~0x81000000); + rdev->audio_enabled = enable; +} + +/* + * initialize the audio vars and register the update timer + */ +int r600_audio_init(struct radeon_device *rdev) +{ + if (!radeon_audio || !r600_audio_chipset_supported(rdev)) + return 0; + + r600_audio_engine_enable(rdev, true); + + rdev->audio_channels = -1; + rdev->audio_rate = -1; + rdev->audio_bits_per_sample = -1; + rdev->audio_status_bits = 0; + rdev->audio_category_code = 0; + + setup_timer( + &rdev->audio_timer, + r600_audio_update_hdmi, + (unsigned long)rdev); + + return 0; +} + +/* + * enable the polling timer, to check for status changes + */ +void r600_audio_enable_polling(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + + DRM_DEBUG("r600_audio_enable_polling: %d\n", + radeon_encoder->audio_polling_active); + if (radeon_encoder->audio_polling_active) + return; + + radeon_encoder->audio_polling_active = 1; + if (rdev->audio_enabled) + mod_timer(&rdev->audio_timer, jiffies + 1); +} + +/* + * disable the polling timer, so we get no more status updates + */ +void r600_audio_disable_polling(struct drm_encoder *encoder) +{ + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + DRM_DEBUG("r600_audio_disable_polling: %d\n", + radeon_encoder->audio_polling_active); + radeon_encoder->audio_polling_active = 0; +} + +/* + * atach the audio codec to the clock source of the encoder + */ +void r600_audio_set_clock(struct drm_encoder *encoder, int clock) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct radeon_encoder_atom_dig *dig = radeon_encoder->enc_priv; + int base_rate = 48000; + + switch (radeon_encoder->encoder_id) { + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_TMDS1: + case ENCODER_OBJECT_ID_INTERNAL_LVTM1: + WREG32_P(R600_AUDIO_TIMING, 0, ~0x301); + break; + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY: + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1: + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA: + WREG32_P(R600_AUDIO_TIMING, 0x100, ~0x301); + break; + default: + dev_err(rdev->dev, "Unsupported encoder type 0x%02X\n", + radeon_encoder->encoder_id); + return; + } + + switch (dig->dig_encoder) { + case 0: + WREG32(R600_AUDIO_PLL1_MUL, base_rate * 50); + WREG32(R600_AUDIO_PLL1_DIV, clock * 100); + WREG32(R600_AUDIO_CLK_SRCSEL, 0); + break; + + case 1: + WREG32(R600_AUDIO_PLL2_MUL, base_rate * 50); + WREG32(R600_AUDIO_PLL2_DIV, clock * 100); + WREG32(R600_AUDIO_CLK_SRCSEL, 1); + break; + default: + dev_err(rdev->dev, "Unsupported DIG on encoder 0x%02X\n", + radeon_encoder->encoder_id); + return; + } +} + +/* + * release the audio timer + * TODO: How to do this correctly on SMP systems? + */ +void r600_audio_fini(struct radeon_device *rdev) +{ + if (!rdev->audio_enabled) + return; + + del_timer(&rdev->audio_timer); + + r600_audio_engine_enable(rdev, false); +} diff --git a/drivers/gpu/drm/radeon/r600_blit.c b/drivers/gpu/drm/radeon/r600_blit.c new file mode 100644 index 00000000..7f104344 --- /dev/null +++ b/drivers/gpu/drm/radeon/r600_blit.c @@ -0,0 +1,868 @@ +/* + * Copyright 2009 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Alex Deucher + */ +#include "drmP.h" +#include "drm.h" +#include "radeon_drm.h" +#include "radeon_drv.h" + +#include "r600_blit_shaders.h" + +#define DI_PT_RECTLIST 0x11 +#define DI_INDEX_SIZE_16_BIT 0x0 +#define DI_SRC_SEL_AUTO_INDEX 0x2 + +#define FMT_8 0x1 +#define FMT_5_6_5 0x8 +#define FMT_8_8_8_8 0x1a +#define COLOR_8 0x1 +#define COLOR_5_6_5 0x8 +#define COLOR_8_8_8_8 0x1a + +static inline void +set_render_target(drm_radeon_private_t *dev_priv, int format, int w, int h, u64 gpu_addr) +{ + u32 cb_color_info; + int pitch, slice; + RING_LOCALS; + DRM_DEBUG("\n"); + + h = ALIGN(h, 8); + if (h < 8) + h = 8; + + cb_color_info = ((format << 2) | (1 << 27)); + pitch = (w / 8) - 1; + slice = ((w * h) / 64) - 1; + + if (((dev_priv->flags & RADEON_FAMILY_MASK) > CHIP_R600) && + ((dev_priv->flags & RADEON_FAMILY_MASK) < CHIP_RV770)) { + BEGIN_RING(21 + 2); + OUT_RING(CP_PACKET3(R600_IT_SET_CONTEXT_REG, 1)); + OUT_RING((R600_CB_COLOR0_BASE - R600_SET_CONTEXT_REG_OFFSET) >> 2); + OUT_RING(gpu_addr >> 8); + OUT_RING(CP_PACKET3(R600_IT_SURFACE_BASE_UPDATE, 0)); + OUT_RING(2 << 0); + } else { + BEGIN_RING(21); + OUT_RING(CP_PACKET3(R600_IT_SET_CONTEXT_REG, 1)); + OUT_RING((R600_CB_COLOR0_BASE - R600_SET_CONTEXT_REG_OFFSET) >> 2); + OUT_RING(gpu_addr >> 8); + } + + OUT_RING(CP_PACKET3(R600_IT_SET_CONTEXT_REG, 1)); + OUT_RING((R600_CB_COLOR0_SIZE - R600_SET_CONTEXT_REG_OFFSET) >> 2); + OUT_RING((pitch << 0) | (slice << 10)); + + OUT_RING(CP_PACKET3(R600_IT_SET_CONTEXT_REG, 1)); + OUT_RING((R600_CB_COLOR0_VIEW - R600_SET_CONTEXT_REG_OFFSET) >> 2); + OUT_RING(0); + + OUT_RING(CP_PACKET3(R600_IT_SET_CONTEXT_REG, 1)); + OUT_RING((R600_CB_COLOR0_INFO - R600_SET_CONTEXT_REG_OFFSET) >> 2); + OUT_RING(cb_color_info); + + OUT_RING(CP_PACKET3(R600_IT_SET_CONTEXT_REG, 1)); + OUT_RING((R600_CB_COLOR0_TILE - R600_SET_CONTEXT_REG_OFFSET) >> 2); + OUT_RING(0); + + OUT_RING(CP_PACKET3(R600_IT_SET_CONTEXT_REG, 1)); + OUT_RING((R600_CB_COLOR0_FRAG - R600_SET_CONTEXT_REG_OFFSET) >> 2); + OUT_RING(0); + + OUT_RING(CP_PACKET3(R600_IT_SET_CONTEXT_REG, 1)); + OUT_RING((R600_CB_COLOR0_MASK - R600_SET_CONTEXT_REG_OFFSET) >> 2); + OUT_RING(0); + + ADVANCE_RING(); +} + +static inline void +cp_set_surface_sync(drm_radeon_private_t *dev_priv, + u32 sync_type, u32 size, u64 mc_addr) +{ + u32 cp_coher_size; + RING_LOCALS; + DRM_DEBUG("\n"); + + if (size == 0xffffffff) + cp_coher_size = 0xffffffff; + else + cp_coher_size = ((size + 255) >> 8); + + BEGIN_RING(5); + OUT_RING(CP_PACKET3(R600_IT_SURFACE_SYNC, 3)); + OUT_RING(sync_type); + OUT_RING(cp_coher_size); + OUT_RING((mc_addr >> 8)); + OUT_RING(10); /* poll interval */ + ADVANCE_RING(); +} + +static inline void +set_shaders(struct drm_device *dev) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + u64 gpu_addr; + int i; + u32 *vs, *ps; + uint32_t sq_pgm_resources; + RING_LOCALS; + DRM_DEBUG("\n"); + + /* load shaders */ + vs = (u32 *) ((char *)dev->agp_buffer_map->handle + dev_priv->blit_vb->offset); + ps = (u32 *) ((char *)dev->agp_buffer_map->handle + dev_priv->blit_vb->offset + 256); + + for (i = 0; i < r6xx_vs_size; i++) + vs[i] = cpu_to_le32(r6xx_vs[i]); + for (i = 0; i < r6xx_ps_size; i++) + ps[i] = cpu_to_le32(r6xx_ps[i]); + + dev_priv->blit_vb->used = 512; + + gpu_addr = dev_priv->gart_buffers_offset + dev_priv->blit_vb->offset; + + /* setup shader regs */ + sq_pgm_resources = (1 << 0); + + BEGIN_RING(9 + 12); + /* VS */ + OUT_RING(CP_PACKET3(R600_IT_SET_CONTEXT_REG, 1)); + OUT_RING((R600_SQ_PGM_START_VS - R600_SET_CONTEXT_REG_OFFSET) >> 2); + OUT_RING(gpu_addr >> 8); + + OUT_RING(CP_PACKET3(R600_IT_SET_CONTEXT_REG, 1)); + OUT_RING((R600_SQ_PGM_RESOURCES_VS - R600_SET_CONTEXT_REG_OFFSET) >> 2); + OUT_RING(sq_pgm_resources); + + OUT_RING(CP_PACKET3(R600_IT_SET_CONTEXT_REG, 1)); + OUT_RING((R600_SQ_PGM_CF_OFFSET_VS - R600_SET_CONTEXT_REG_OFFSET) >> 2); + OUT_RING(0); + + /* PS */ + OUT_RING(CP_PACKET3(R600_IT_SET_CONTEXT_REG, 1)); + OUT_RING((R600_SQ_PGM_START_PS - R600_SET_CONTEXT_REG_OFFSET) >> 2); + OUT_RING((gpu_addr + 256) >> 8); + + OUT_RING(CP_PACKET3(R600_IT_SET_CONTEXT_REG, 1)); + OUT_RING((R600_SQ_PGM_RESOURCES_PS - R600_SET_CONTEXT_REG_OFFSET) >> 2); + OUT_RING(sq_pgm_resources | (1 << 28)); + + OUT_RING(CP_PACKET3(R600_IT_SET_CONTEXT_REG, 1)); + OUT_RING((R600_SQ_PGM_EXPORTS_PS - R600_SET_CONTEXT_REG_OFFSET) >> 2); + OUT_RING(2); + + OUT_RING(CP_PACKET3(R600_IT_SET_CONTEXT_REG, 1)); + OUT_RING((R600_SQ_PGM_CF_OFFSET_PS - R600_SET_CONTEXT_REG_OFFSET) >> 2); + OUT_RING(0); + ADVANCE_RING(); + + cp_set_surface_sync(dev_priv, + R600_SH_ACTION_ENA, 512, gpu_addr); +} + +static inline void +set_vtx_resource(drm_radeon_private_t *dev_priv, u64 gpu_addr) +{ + uint32_t sq_vtx_constant_word2; + RING_LOCALS; + DRM_DEBUG("\n"); + + sq_vtx_constant_word2 = (((gpu_addr >> 32) & 0xff) | (16 << 8)); +#ifdef __BIG_ENDIAN + sq_vtx_constant_word2 |= (2 << 30); +#endif + + BEGIN_RING(9); + OUT_RING(CP_PACKET3(R600_IT_SET_RESOURCE, 7)); + OUT_RING(0x460); + OUT_RING(gpu_addr & 0xffffffff); + OUT_RING(48 - 1); + OUT_RING(sq_vtx_constant_word2); + OUT_RING(1 << 0); + OUT_RING(0); + OUT_RING(0); + OUT_RING(R600_SQ_TEX_VTX_VALID_BUFFER << 30); + ADVANCE_RING(); + + if (((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV610) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV620) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS780) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS880) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV710)) + cp_set_surface_sync(dev_priv, + R600_TC_ACTION_ENA, 48, gpu_addr); + else + cp_set_surface_sync(dev_priv, + R600_VC_ACTION_ENA, 48, gpu_addr); +} + +static inline void +set_tex_resource(drm_radeon_private_t *dev_priv, + int format, int w, int h, int pitch, u64 gpu_addr) +{ + uint32_t sq_tex_resource_word0, sq_tex_resource_word1, sq_tex_resource_word4; + RING_LOCALS; + DRM_DEBUG("\n"); + + if (h < 1) + h = 1; + + sq_tex_resource_word0 = (1 << 0); + sq_tex_resource_word0 |= ((((pitch >> 3) - 1) << 8) | + ((w - 1) << 19)); + + sq_tex_resource_word1 = (format << 26); + sq_tex_resource_word1 |= ((h - 1) << 0); + + sq_tex_resource_word4 = ((1 << 14) | + (0 << 16) | + (1 << 19) | + (2 << 22) | + (3 << 25)); + + BEGIN_RING(9); + OUT_RING(CP_PACKET3(R600_IT_SET_RESOURCE, 7)); + OUT_RING(0); + OUT_RING(sq_tex_resource_word0); + OUT_RING(sq_tex_resource_word1); + OUT_RING(gpu_addr >> 8); + OUT_RING(gpu_addr >> 8); + OUT_RING(sq_tex_resource_word4); + OUT_RING(0); + OUT_RING(R600_SQ_TEX_VTX_VALID_TEXTURE << 30); + ADVANCE_RING(); + +} + +static inline void +set_scissors(drm_radeon_private_t *dev_priv, int x1, int y1, int x2, int y2) +{ + RING_LOCALS; + DRM_DEBUG("\n"); + + BEGIN_RING(12); + OUT_RING(CP_PACKET3(R600_IT_SET_CONTEXT_REG, 2)); + OUT_RING((R600_PA_SC_SCREEN_SCISSOR_TL - R600_SET_CONTEXT_REG_OFFSET) >> 2); + OUT_RING((x1 << 0) | (y1 << 16)); + OUT_RING((x2 << 0) | (y2 << 16)); + + OUT_RING(CP_PACKET3(R600_IT_SET_CONTEXT_REG, 2)); + OUT_RING((R600_PA_SC_GENERIC_SCISSOR_TL - R600_SET_CONTEXT_REG_OFFSET) >> 2); + OUT_RING((x1 << 0) | (y1 << 16) | (1 << 31)); + OUT_RING((x2 << 0) | (y2 << 16)); + + OUT_RING(CP_PACKET3(R600_IT_SET_CONTEXT_REG, 2)); + OUT_RING((R600_PA_SC_WINDOW_SCISSOR_TL - R600_SET_CONTEXT_REG_OFFSET) >> 2); + OUT_RING((x1 << 0) | (y1 << 16) | (1 << 31)); + OUT_RING((x2 << 0) | (y2 << 16)); + ADVANCE_RING(); +} + +static inline void +draw_auto(drm_radeon_private_t *dev_priv) +{ + RING_LOCALS; + DRM_DEBUG("\n"); + + BEGIN_RING(10); + OUT_RING(CP_PACKET3(R600_IT_SET_CONFIG_REG, 1)); + OUT_RING((R600_VGT_PRIMITIVE_TYPE - R600_SET_CONFIG_REG_OFFSET) >> 2); + OUT_RING(DI_PT_RECTLIST); + + OUT_RING(CP_PACKET3(R600_IT_INDEX_TYPE, 0)); +#ifdef __BIG_ENDIAN + OUT_RING((2 << 2) | DI_INDEX_SIZE_16_BIT); +#else + OUT_RING(DI_INDEX_SIZE_16_BIT); +#endif + + OUT_RING(CP_PACKET3(R600_IT_NUM_INSTANCES, 0)); + OUT_RING(1); + + OUT_RING(CP_PACKET3(R600_IT_DRAW_INDEX_AUTO, 1)); + OUT_RING(3); + OUT_RING(DI_SRC_SEL_AUTO_INDEX); + + ADVANCE_RING(); + COMMIT_RING(); +} + +static inline void +set_default_state(drm_radeon_private_t *dev_priv) +{ + int i; + u32 sq_config, sq_gpr_resource_mgmt_1, sq_gpr_resource_mgmt_2; + u32 sq_thread_resource_mgmt, sq_stack_resource_mgmt_1, sq_stack_resource_mgmt_2; + int num_ps_gprs, num_vs_gprs, num_temp_gprs, num_gs_gprs, num_es_gprs; + int num_ps_threads, num_vs_threads, num_gs_threads, num_es_threads; + int num_ps_stack_entries, num_vs_stack_entries, num_gs_stack_entries, num_es_stack_entries; + RING_LOCALS; + + switch ((dev_priv->flags & RADEON_FAMILY_MASK)) { + case CHIP_R600: + num_ps_gprs = 192; + num_vs_gprs = 56; + num_temp_gprs = 4; + num_gs_gprs = 0; + num_es_gprs = 0; + num_ps_threads = 136; + num_vs_threads = 48; + num_gs_threads = 4; + num_es_threads = 4; + num_ps_stack_entries = 128; + num_vs_stack_entries = 128; + num_gs_stack_entries = 0; + num_es_stack_entries = 0; + break; + case CHIP_RV630: + case CHIP_RV635: + num_ps_gprs = 84; + num_vs_gprs = 36; + num_temp_gprs = 4; + num_gs_gprs = 0; + num_es_gprs = 0; + num_ps_threads = 144; + num_vs_threads = 40; + num_gs_threads = 4; + num_es_threads = 4; + num_ps_stack_entries = 40; + num_vs_stack_entries = 40; + num_gs_stack_entries = 32; + num_es_stack_entries = 16; + break; + case CHIP_RV610: + case CHIP_RV620: + case CHIP_RS780: + case CHIP_RS880: + default: + num_ps_gprs = 84; + num_vs_gprs = 36; + num_temp_gprs = 4; + num_gs_gprs = 0; + num_es_gprs = 0; + num_ps_threads = 136; + num_vs_threads = 48; + num_gs_threads = 4; + num_es_threads = 4; + num_ps_stack_entries = 40; + num_vs_stack_entries = 40; + num_gs_stack_entries = 32; + num_es_stack_entries = 16; + break; + case CHIP_RV670: + num_ps_gprs = 144; + num_vs_gprs = 40; + num_temp_gprs = 4; + num_gs_gprs = 0; + num_es_gprs = 0; + num_ps_threads = 136; + num_vs_threads = 48; + num_gs_threads = 4; + num_es_threads = 4; + num_ps_stack_entries = 40; + num_vs_stack_entries = 40; + num_gs_stack_entries = 32; + num_es_stack_entries = 16; + break; + case CHIP_RV770: + num_ps_gprs = 192; + num_vs_gprs = 56; + num_temp_gprs = 4; + num_gs_gprs = 0; + num_es_gprs = 0; + num_ps_threads = 188; + num_vs_threads = 60; + num_gs_threads = 0; + num_es_threads = 0; + num_ps_stack_entries = 256; + num_vs_stack_entries = 256; + num_gs_stack_entries = 0; + num_es_stack_entries = 0; + break; + case CHIP_RV730: + case CHIP_RV740: + num_ps_gprs = 84; + num_vs_gprs = 36; + num_temp_gprs = 4; + num_gs_gprs = 0; + num_es_gprs = 0; + num_ps_threads = 188; + num_vs_threads = 60; + num_gs_threads = 0; + num_es_threads = 0; + num_ps_stack_entries = 128; + num_vs_stack_entries = 128; + num_gs_stack_entries = 0; + num_es_stack_entries = 0; + break; + case CHIP_RV710: + num_ps_gprs = 192; + num_vs_gprs = 56; + num_temp_gprs = 4; + num_gs_gprs = 0; + num_es_gprs = 0; + num_ps_threads = 144; + num_vs_threads = 48; + num_gs_threads = 0; + num_es_threads = 0; + num_ps_stack_entries = 128; + num_vs_stack_entries = 128; + num_gs_stack_entries = 0; + num_es_stack_entries = 0; + break; + } + + if (((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV610) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV620) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS780) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS880) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV710)) + sq_config = 0; + else + sq_config = R600_VC_ENABLE; + + sq_config |= (R600_DX9_CONSTS | + R600_ALU_INST_PREFER_VECTOR | + R600_PS_PRIO(0) | + R600_VS_PRIO(1) | + R600_GS_PRIO(2) | + R600_ES_PRIO(3)); + + sq_gpr_resource_mgmt_1 = (R600_NUM_PS_GPRS(num_ps_gprs) | + R600_NUM_VS_GPRS(num_vs_gprs) | + R600_NUM_CLAUSE_TEMP_GPRS(num_temp_gprs)); + sq_gpr_resource_mgmt_2 = (R600_NUM_GS_GPRS(num_gs_gprs) | + R600_NUM_ES_GPRS(num_es_gprs)); + sq_thread_resource_mgmt = (R600_NUM_PS_THREADS(num_ps_threads) | + R600_NUM_VS_THREADS(num_vs_threads) | + R600_NUM_GS_THREADS(num_gs_threads) | + R600_NUM_ES_THREADS(num_es_threads)); + sq_stack_resource_mgmt_1 = (R600_NUM_PS_STACK_ENTRIES(num_ps_stack_entries) | + R600_NUM_VS_STACK_ENTRIES(num_vs_stack_entries)); + sq_stack_resource_mgmt_2 = (R600_NUM_GS_STACK_ENTRIES(num_gs_stack_entries) | + R600_NUM_ES_STACK_ENTRIES(num_es_stack_entries)); + + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_RV770) { + BEGIN_RING(r7xx_default_size + 10); + for (i = 0; i < r7xx_default_size; i++) + OUT_RING(r7xx_default_state[i]); + } else { + BEGIN_RING(r6xx_default_size + 10); + for (i = 0; i < r6xx_default_size; i++) + OUT_RING(r6xx_default_state[i]); + } + OUT_RING(CP_PACKET3(R600_IT_EVENT_WRITE, 0)); + OUT_RING(R600_CACHE_FLUSH_AND_INV_EVENT); + /* SQ config */ + OUT_RING(CP_PACKET3(R600_IT_SET_CONFIG_REG, 6)); + OUT_RING((R600_SQ_CONFIG - R600_SET_CONFIG_REG_OFFSET) >> 2); + OUT_RING(sq_config); + OUT_RING(sq_gpr_resource_mgmt_1); + OUT_RING(sq_gpr_resource_mgmt_2); + OUT_RING(sq_thread_resource_mgmt); + OUT_RING(sq_stack_resource_mgmt_1); + OUT_RING(sq_stack_resource_mgmt_2); + ADVANCE_RING(); +} + +static inline uint32_t i2f(uint32_t input) +{ + u32 result, i, exponent, fraction; + + if ((input & 0x3fff) == 0) + result = 0; /* 0 is a special case */ + else { + exponent = 140; /* exponent biased by 127; */ + fraction = (input & 0x3fff) << 10; /* cheat and only + handle numbers below 2^^15 */ + for (i = 0; i < 14; i++) { + if (fraction & 0x800000) + break; + else { + fraction = fraction << 1; /* keep + shifting left until top bit = 1 */ + exponent = exponent - 1; + } + } + result = exponent << 23 | (fraction & 0x7fffff); /* mask + off top bit; assumed 1 */ + } + return result; +} + + +static inline int r600_nomm_get_vb(struct drm_device *dev) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + dev_priv->blit_vb = radeon_freelist_get(dev); + if (!dev_priv->blit_vb) { + DRM_ERROR("Unable to allocate vertex buffer for blit\n"); + return -EAGAIN; + } + return 0; +} + +static inline void r600_nomm_put_vb(struct drm_device *dev) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + + dev_priv->blit_vb->used = 0; + radeon_cp_discard_buffer(dev, dev_priv->blit_vb->file_priv->master, dev_priv->blit_vb); +} + +static inline void *r600_nomm_get_vb_ptr(struct drm_device *dev) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + return (((char *)dev->agp_buffer_map->handle + + dev_priv->blit_vb->offset + dev_priv->blit_vb->used)); +} + +int +r600_prepare_blit_copy(struct drm_device *dev, struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + int ret; + DRM_DEBUG("\n"); + + ret = r600_nomm_get_vb(dev); + if (ret) + return ret; + + dev_priv->blit_vb->file_priv = file_priv; + + set_default_state(dev_priv); + set_shaders(dev); + + return 0; +} + + +void +r600_done_blit_copy(struct drm_device *dev) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + RING_LOCALS; + DRM_DEBUG("\n"); + + BEGIN_RING(5); + OUT_RING(CP_PACKET3(R600_IT_EVENT_WRITE, 0)); + OUT_RING(R600_CACHE_FLUSH_AND_INV_EVENT); + /* wait for 3D idle clean */ + OUT_RING(CP_PACKET3(R600_IT_SET_CONFIG_REG, 1)); + OUT_RING((R600_WAIT_UNTIL - R600_SET_CONFIG_REG_OFFSET) >> 2); + OUT_RING(RADEON_WAIT_3D_IDLE | RADEON_WAIT_3D_IDLECLEAN); + + ADVANCE_RING(); + COMMIT_RING(); + + r600_nomm_put_vb(dev); +} + +void +r600_blit_copy(struct drm_device *dev, + uint64_t src_gpu_addr, uint64_t dst_gpu_addr, + int size_bytes) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + int max_bytes; + u64 vb_addr; + u32 *vb; + + vb = r600_nomm_get_vb_ptr(dev); + + if ((size_bytes & 3) || (src_gpu_addr & 3) || (dst_gpu_addr & 3)) { + max_bytes = 8192; + + while (size_bytes) { + int cur_size = size_bytes; + int src_x = src_gpu_addr & 255; + int dst_x = dst_gpu_addr & 255; + int h = 1; + src_gpu_addr = src_gpu_addr & ~255; + dst_gpu_addr = dst_gpu_addr & ~255; + + if (!src_x && !dst_x) { + h = (cur_size / max_bytes); + if (h > 8192) + h = 8192; + if (h == 0) + h = 1; + else + cur_size = max_bytes; + } else { + if (cur_size > max_bytes) + cur_size = max_bytes; + if (cur_size > (max_bytes - dst_x)) + cur_size = (max_bytes - dst_x); + if (cur_size > (max_bytes - src_x)) + cur_size = (max_bytes - src_x); + } + + if ((dev_priv->blit_vb->used + 48) > dev_priv->blit_vb->total) { + + r600_nomm_put_vb(dev); + r600_nomm_get_vb(dev); + if (!dev_priv->blit_vb) + return; + set_shaders(dev); + vb = r600_nomm_get_vb_ptr(dev); + } + + vb[0] = i2f(dst_x); + vb[1] = 0; + vb[2] = i2f(src_x); + vb[3] = 0; + + vb[4] = i2f(dst_x); + vb[5] = i2f(h); + vb[6] = i2f(src_x); + vb[7] = i2f(h); + + vb[8] = i2f(dst_x + cur_size); + vb[9] = i2f(h); + vb[10] = i2f(src_x + cur_size); + vb[11] = i2f(h); + + /* src */ + set_tex_resource(dev_priv, FMT_8, + src_x + cur_size, h, src_x + cur_size, + src_gpu_addr); + + cp_set_surface_sync(dev_priv, + R600_TC_ACTION_ENA, (src_x + cur_size * h), src_gpu_addr); + + /* dst */ + set_render_target(dev_priv, COLOR_8, + dst_x + cur_size, h, + dst_gpu_addr); + + /* scissors */ + set_scissors(dev_priv, dst_x, 0, dst_x + cur_size, h); + + /* Vertex buffer setup */ + vb_addr = dev_priv->gart_buffers_offset + + dev_priv->blit_vb->offset + + dev_priv->blit_vb->used; + set_vtx_resource(dev_priv, vb_addr); + + /* draw */ + draw_auto(dev_priv); + + cp_set_surface_sync(dev_priv, + R600_CB_ACTION_ENA | R600_CB0_DEST_BASE_ENA, + cur_size * h, dst_gpu_addr); + + vb += 12; + dev_priv->blit_vb->used += 12 * 4; + + src_gpu_addr += cur_size * h; + dst_gpu_addr += cur_size * h; + size_bytes -= cur_size * h; + } + } else { + max_bytes = 8192 * 4; + + while (size_bytes) { + int cur_size = size_bytes; + int src_x = (src_gpu_addr & 255); + int dst_x = (dst_gpu_addr & 255); + int h = 1; + src_gpu_addr = src_gpu_addr & ~255; + dst_gpu_addr = dst_gpu_addr & ~255; + + if (!src_x && !dst_x) { + h = (cur_size / max_bytes); + if (h > 8192) + h = 8192; + if (h == 0) + h = 1; + else + cur_size = max_bytes; + } else { + if (cur_size > max_bytes) + cur_size = max_bytes; + if (cur_size > (max_bytes - dst_x)) + cur_size = (max_bytes - dst_x); + if (cur_size > (max_bytes - src_x)) + cur_size = (max_bytes - src_x); + } + + if ((dev_priv->blit_vb->used + 48) > dev_priv->blit_vb->total) { + r600_nomm_put_vb(dev); + r600_nomm_get_vb(dev); + if (!dev_priv->blit_vb) + return; + + set_shaders(dev); + vb = r600_nomm_get_vb_ptr(dev); + } + + vb[0] = i2f(dst_x / 4); + vb[1] = 0; + vb[2] = i2f(src_x / 4); + vb[3] = 0; + + vb[4] = i2f(dst_x / 4); + vb[5] = i2f(h); + vb[6] = i2f(src_x / 4); + vb[7] = i2f(h); + + vb[8] = i2f((dst_x + cur_size) / 4); + vb[9] = i2f(h); + vb[10] = i2f((src_x + cur_size) / 4); + vb[11] = i2f(h); + + /* src */ + set_tex_resource(dev_priv, FMT_8_8_8_8, + (src_x + cur_size) / 4, + h, (src_x + cur_size) / 4, + src_gpu_addr); + + cp_set_surface_sync(dev_priv, + R600_TC_ACTION_ENA, (src_x + cur_size * h), src_gpu_addr); + + /* dst */ + set_render_target(dev_priv, COLOR_8_8_8_8, + (dst_x + cur_size) / 4, h, + dst_gpu_addr); + + /* scissors */ + set_scissors(dev_priv, (dst_x / 4), 0, (dst_x + cur_size / 4), h); + + /* Vertex buffer setup */ + vb_addr = dev_priv->gart_buffers_offset + + dev_priv->blit_vb->offset + + dev_priv->blit_vb->used; + set_vtx_resource(dev_priv, vb_addr); + + /* draw */ + draw_auto(dev_priv); + + cp_set_surface_sync(dev_priv, + R600_CB_ACTION_ENA | R600_CB0_DEST_BASE_ENA, + cur_size * h, dst_gpu_addr); + + vb += 12; + dev_priv->blit_vb->used += 12 * 4; + + src_gpu_addr += cur_size * h; + dst_gpu_addr += cur_size * h; + size_bytes -= cur_size * h; + } + } +} + +void +r600_blit_swap(struct drm_device *dev, + uint64_t src_gpu_addr, uint64_t dst_gpu_addr, + int sx, int sy, int dx, int dy, + int w, int h, int src_pitch, int dst_pitch, int cpp) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + int cb_format, tex_format; + int sx2, sy2, dx2, dy2; + u64 vb_addr; + u32 *vb; + + if ((dev_priv->blit_vb->used + 48) > dev_priv->blit_vb->total) { + + r600_nomm_put_vb(dev); + r600_nomm_get_vb(dev); + if (!dev_priv->blit_vb) + return; + + set_shaders(dev); + } + vb = r600_nomm_get_vb_ptr(dev); + + sx2 = sx + w; + sy2 = sy + h; + dx2 = dx + w; + dy2 = dy + h; + + vb[0] = i2f(dx); + vb[1] = i2f(dy); + vb[2] = i2f(sx); + vb[3] = i2f(sy); + + vb[4] = i2f(dx); + vb[5] = i2f(dy2); + vb[6] = i2f(sx); + vb[7] = i2f(sy2); + + vb[8] = i2f(dx2); + vb[9] = i2f(dy2); + vb[10] = i2f(sx2); + vb[11] = i2f(sy2); + + switch(cpp) { + case 4: + cb_format = COLOR_8_8_8_8; + tex_format = FMT_8_8_8_8; + break; + case 2: + cb_format = COLOR_5_6_5; + tex_format = FMT_5_6_5; + break; + default: + cb_format = COLOR_8; + tex_format = FMT_8; + break; + } + + /* src */ + set_tex_resource(dev_priv, tex_format, + src_pitch / cpp, + sy2, src_pitch / cpp, + src_gpu_addr); + + cp_set_surface_sync(dev_priv, + R600_TC_ACTION_ENA, src_pitch * sy2, src_gpu_addr); + + /* dst */ + set_render_target(dev_priv, cb_format, + dst_pitch / cpp, dy2, + dst_gpu_addr); + + /* scissors */ + set_scissors(dev_priv, dx, dy, dx2, dy2); + + /* Vertex buffer setup */ + vb_addr = dev_priv->gart_buffers_offset + + dev_priv->blit_vb->offset + + dev_priv->blit_vb->used; + set_vtx_resource(dev_priv, vb_addr); + + /* draw */ + draw_auto(dev_priv); + + cp_set_surface_sync(dev_priv, + R600_CB_ACTION_ENA | R600_CB0_DEST_BASE_ENA, + dst_pitch * dy2, dst_gpu_addr); + + dev_priv->blit_vb->used += 12 * 4; +} diff --git a/drivers/gpu/drm/radeon/r600_blit_kms.c b/drivers/gpu/drm/radeon/r600_blit_kms.c new file mode 100644 index 00000000..9aa74c3f --- /dev/null +++ b/drivers/gpu/drm/radeon/r600_blit_kms.c @@ -0,0 +1,840 @@ +/* + * Copyright 2009 Advanced Micro Devices, Inc. + * Copyright 2009 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + */ + +#include "drmP.h" +#include "drm.h" +#include "radeon_drm.h" +#include "radeon.h" + +#include "r600d.h" +#include "r600_blit_shaders.h" + +#define DI_PT_RECTLIST 0x11 +#define DI_INDEX_SIZE_16_BIT 0x0 +#define DI_SRC_SEL_AUTO_INDEX 0x2 + +#define FMT_8 0x1 +#define FMT_5_6_5 0x8 +#define FMT_8_8_8_8 0x1a +#define COLOR_8 0x1 +#define COLOR_5_6_5 0x8 +#define COLOR_8_8_8_8 0x1a + +/* emits 21 on rv770+, 23 on r600 */ +static void +set_render_target(struct radeon_device *rdev, int format, + int w, int h, u64 gpu_addr) +{ + u32 cb_color_info; + int pitch, slice; + + h = ALIGN(h, 8); + if (h < 8) + h = 8; + + cb_color_info = ((format << 2) | (1 << 27) | (1 << 8)); + pitch = (w / 8) - 1; + slice = ((w * h) / 64) - 1; + + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONTEXT_REG, 1)); + radeon_ring_write(rdev, (CB_COLOR0_BASE - PACKET3_SET_CONTEXT_REG_OFFSET) >> 2); + radeon_ring_write(rdev, gpu_addr >> 8); + + if (rdev->family > CHIP_R600 && rdev->family < CHIP_RV770) { + radeon_ring_write(rdev, PACKET3(PACKET3_SURFACE_BASE_UPDATE, 0)); + radeon_ring_write(rdev, 2 << 0); + } + + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONTEXT_REG, 1)); + radeon_ring_write(rdev, (CB_COLOR0_SIZE - PACKET3_SET_CONTEXT_REG_OFFSET) >> 2); + radeon_ring_write(rdev, (pitch << 0) | (slice << 10)); + + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONTEXT_REG, 1)); + radeon_ring_write(rdev, (CB_COLOR0_VIEW - PACKET3_SET_CONTEXT_REG_OFFSET) >> 2); + radeon_ring_write(rdev, 0); + + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONTEXT_REG, 1)); + radeon_ring_write(rdev, (CB_COLOR0_INFO - PACKET3_SET_CONTEXT_REG_OFFSET) >> 2); + radeon_ring_write(rdev, cb_color_info); + + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONTEXT_REG, 1)); + radeon_ring_write(rdev, (CB_COLOR0_TILE - PACKET3_SET_CONTEXT_REG_OFFSET) >> 2); + radeon_ring_write(rdev, 0); + + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONTEXT_REG, 1)); + radeon_ring_write(rdev, (CB_COLOR0_FRAG - PACKET3_SET_CONTEXT_REG_OFFSET) >> 2); + radeon_ring_write(rdev, 0); + + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONTEXT_REG, 1)); + radeon_ring_write(rdev, (CB_COLOR0_MASK - PACKET3_SET_CONTEXT_REG_OFFSET) >> 2); + radeon_ring_write(rdev, 0); +} + +/* emits 5dw */ +static void +cp_set_surface_sync(struct radeon_device *rdev, + u32 sync_type, u32 size, + u64 mc_addr) +{ + u32 cp_coher_size; + + if (size == 0xffffffff) + cp_coher_size = 0xffffffff; + else + cp_coher_size = ((size + 255) >> 8); + + radeon_ring_write(rdev, PACKET3(PACKET3_SURFACE_SYNC, 3)); + radeon_ring_write(rdev, sync_type); + radeon_ring_write(rdev, cp_coher_size); + radeon_ring_write(rdev, mc_addr >> 8); + radeon_ring_write(rdev, 10); /* poll interval */ +} + +/* emits 21dw + 1 surface sync = 26dw */ +static void +set_shaders(struct radeon_device *rdev) +{ + u64 gpu_addr; + u32 sq_pgm_resources; + + /* setup shader regs */ + sq_pgm_resources = (1 << 0); + + /* VS */ + gpu_addr = rdev->r600_blit.shader_gpu_addr + rdev->r600_blit.vs_offset; + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONTEXT_REG, 1)); + radeon_ring_write(rdev, (SQ_PGM_START_VS - PACKET3_SET_CONTEXT_REG_OFFSET) >> 2); + radeon_ring_write(rdev, gpu_addr >> 8); + + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONTEXT_REG, 1)); + radeon_ring_write(rdev, (SQ_PGM_RESOURCES_VS - PACKET3_SET_CONTEXT_REG_OFFSET) >> 2); + radeon_ring_write(rdev, sq_pgm_resources); + + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONTEXT_REG, 1)); + radeon_ring_write(rdev, (SQ_PGM_CF_OFFSET_VS - PACKET3_SET_CONTEXT_REG_OFFSET) >> 2); + radeon_ring_write(rdev, 0); + + /* PS */ + gpu_addr = rdev->r600_blit.shader_gpu_addr + rdev->r600_blit.ps_offset; + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONTEXT_REG, 1)); + radeon_ring_write(rdev, (SQ_PGM_START_PS - PACKET3_SET_CONTEXT_REG_OFFSET) >> 2); + radeon_ring_write(rdev, gpu_addr >> 8); + + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONTEXT_REG, 1)); + radeon_ring_write(rdev, (SQ_PGM_RESOURCES_PS - PACKET3_SET_CONTEXT_REG_OFFSET) >> 2); + radeon_ring_write(rdev, sq_pgm_resources | (1 << 28)); + + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONTEXT_REG, 1)); + radeon_ring_write(rdev, (SQ_PGM_EXPORTS_PS - PACKET3_SET_CONTEXT_REG_OFFSET) >> 2); + radeon_ring_write(rdev, 2); + + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONTEXT_REG, 1)); + radeon_ring_write(rdev, (SQ_PGM_CF_OFFSET_PS - PACKET3_SET_CONTEXT_REG_OFFSET) >> 2); + radeon_ring_write(rdev, 0); + + gpu_addr = rdev->r600_blit.shader_gpu_addr + rdev->r600_blit.vs_offset; + cp_set_surface_sync(rdev, PACKET3_SH_ACTION_ENA, 512, gpu_addr); +} + +/* emits 9 + 1 sync (5) = 14*/ +static void +set_vtx_resource(struct radeon_device *rdev, u64 gpu_addr) +{ + u32 sq_vtx_constant_word2; + + sq_vtx_constant_word2 = ((upper_32_bits(gpu_addr) & 0xff) | (16 << 8)); +#ifdef __BIG_ENDIAN + sq_vtx_constant_word2 |= (2 << 30); +#endif + + radeon_ring_write(rdev, PACKET3(PACKET3_SET_RESOURCE, 7)); + radeon_ring_write(rdev, 0x460); + radeon_ring_write(rdev, gpu_addr & 0xffffffff); + radeon_ring_write(rdev, 48 - 1); + radeon_ring_write(rdev, sq_vtx_constant_word2); + radeon_ring_write(rdev, 1 << 0); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, SQ_TEX_VTX_VALID_BUFFER << 30); + + if ((rdev->family == CHIP_RV610) || + (rdev->family == CHIP_RV620) || + (rdev->family == CHIP_RS780) || + (rdev->family == CHIP_RS880) || + (rdev->family == CHIP_RV710)) + cp_set_surface_sync(rdev, + PACKET3_TC_ACTION_ENA, 48, gpu_addr); + else + cp_set_surface_sync(rdev, + PACKET3_VC_ACTION_ENA, 48, gpu_addr); +} + +/* emits 9 */ +static void +set_tex_resource(struct radeon_device *rdev, + int format, int w, int h, int pitch, + u64 gpu_addr) +{ + uint32_t sq_tex_resource_word0, sq_tex_resource_word1, sq_tex_resource_word4; + + if (h < 1) + h = 1; + + sq_tex_resource_word0 = (1 << 0) | (1 << 3); + sq_tex_resource_word0 |= ((((pitch >> 3) - 1) << 8) | + ((w - 1) << 19)); + + sq_tex_resource_word1 = (format << 26); + sq_tex_resource_word1 |= ((h - 1) << 0); + + sq_tex_resource_word4 = ((1 << 14) | + (0 << 16) | + (1 << 19) | + (2 << 22) | + (3 << 25)); + + radeon_ring_write(rdev, PACKET3(PACKET3_SET_RESOURCE, 7)); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, sq_tex_resource_word0); + radeon_ring_write(rdev, sq_tex_resource_word1); + radeon_ring_write(rdev, gpu_addr >> 8); + radeon_ring_write(rdev, gpu_addr >> 8); + radeon_ring_write(rdev, sq_tex_resource_word4); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, SQ_TEX_VTX_VALID_TEXTURE << 30); +} + +/* emits 12 */ +static void +set_scissors(struct radeon_device *rdev, int x1, int y1, + int x2, int y2) +{ + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONTEXT_REG, 2)); + radeon_ring_write(rdev, (PA_SC_SCREEN_SCISSOR_TL - PACKET3_SET_CONTEXT_REG_OFFSET) >> 2); + radeon_ring_write(rdev, (x1 << 0) | (y1 << 16)); + radeon_ring_write(rdev, (x2 << 0) | (y2 << 16)); + + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONTEXT_REG, 2)); + radeon_ring_write(rdev, (PA_SC_GENERIC_SCISSOR_TL - PACKET3_SET_CONTEXT_REG_OFFSET) >> 2); + radeon_ring_write(rdev, (x1 << 0) | (y1 << 16) | (1 << 31)); + radeon_ring_write(rdev, (x2 << 0) | (y2 << 16)); + + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONTEXT_REG, 2)); + radeon_ring_write(rdev, (PA_SC_WINDOW_SCISSOR_TL - PACKET3_SET_CONTEXT_REG_OFFSET) >> 2); + radeon_ring_write(rdev, (x1 << 0) | (y1 << 16) | (1 << 31)); + radeon_ring_write(rdev, (x2 << 0) | (y2 << 16)); +} + +/* emits 10 */ +static void +draw_auto(struct radeon_device *rdev) +{ + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONFIG_REG, 1)); + radeon_ring_write(rdev, (VGT_PRIMITIVE_TYPE - PACKET3_SET_CONFIG_REG_OFFSET) >> 2); + radeon_ring_write(rdev, DI_PT_RECTLIST); + + radeon_ring_write(rdev, PACKET3(PACKET3_INDEX_TYPE, 0)); + radeon_ring_write(rdev, +#ifdef __BIG_ENDIAN + (2 << 2) | +#endif + DI_INDEX_SIZE_16_BIT); + + radeon_ring_write(rdev, PACKET3(PACKET3_NUM_INSTANCES, 0)); + radeon_ring_write(rdev, 1); + + radeon_ring_write(rdev, PACKET3(PACKET3_DRAW_INDEX_AUTO, 1)); + radeon_ring_write(rdev, 3); + radeon_ring_write(rdev, DI_SRC_SEL_AUTO_INDEX); + +} + +/* emits 14 */ +static void +set_default_state(struct radeon_device *rdev) +{ + u32 sq_config, sq_gpr_resource_mgmt_1, sq_gpr_resource_mgmt_2; + u32 sq_thread_resource_mgmt, sq_stack_resource_mgmt_1, sq_stack_resource_mgmt_2; + int num_ps_gprs, num_vs_gprs, num_temp_gprs, num_gs_gprs, num_es_gprs; + int num_ps_threads, num_vs_threads, num_gs_threads, num_es_threads; + int num_ps_stack_entries, num_vs_stack_entries, num_gs_stack_entries, num_es_stack_entries; + u64 gpu_addr; + int dwords; + + switch (rdev->family) { + case CHIP_R600: + num_ps_gprs = 192; + num_vs_gprs = 56; + num_temp_gprs = 4; + num_gs_gprs = 0; + num_es_gprs = 0; + num_ps_threads = 136; + num_vs_threads = 48; + num_gs_threads = 4; + num_es_threads = 4; + num_ps_stack_entries = 128; + num_vs_stack_entries = 128; + num_gs_stack_entries = 0; + num_es_stack_entries = 0; + break; + case CHIP_RV630: + case CHIP_RV635: + num_ps_gprs = 84; + num_vs_gprs = 36; + num_temp_gprs = 4; + num_gs_gprs = 0; + num_es_gprs = 0; + num_ps_threads = 144; + num_vs_threads = 40; + num_gs_threads = 4; + num_es_threads = 4; + num_ps_stack_entries = 40; + num_vs_stack_entries = 40; + num_gs_stack_entries = 32; + num_es_stack_entries = 16; + break; + case CHIP_RV610: + case CHIP_RV620: + case CHIP_RS780: + case CHIP_RS880: + default: + num_ps_gprs = 84; + num_vs_gprs = 36; + num_temp_gprs = 4; + num_gs_gprs = 0; + num_es_gprs = 0; + num_ps_threads = 136; + num_vs_threads = 48; + num_gs_threads = 4; + num_es_threads = 4; + num_ps_stack_entries = 40; + num_vs_stack_entries = 40; + num_gs_stack_entries = 32; + num_es_stack_entries = 16; + break; + case CHIP_RV670: + num_ps_gprs = 144; + num_vs_gprs = 40; + num_temp_gprs = 4; + num_gs_gprs = 0; + num_es_gprs = 0; + num_ps_threads = 136; + num_vs_threads = 48; + num_gs_threads = 4; + num_es_threads = 4; + num_ps_stack_entries = 40; + num_vs_stack_entries = 40; + num_gs_stack_entries = 32; + num_es_stack_entries = 16; + break; + case CHIP_RV770: + num_ps_gprs = 192; + num_vs_gprs = 56; + num_temp_gprs = 4; + num_gs_gprs = 0; + num_es_gprs = 0; + num_ps_threads = 188; + num_vs_threads = 60; + num_gs_threads = 0; + num_es_threads = 0; + num_ps_stack_entries = 256; + num_vs_stack_entries = 256; + num_gs_stack_entries = 0; + num_es_stack_entries = 0; + break; + case CHIP_RV730: + case CHIP_RV740: + num_ps_gprs = 84; + num_vs_gprs = 36; + num_temp_gprs = 4; + num_gs_gprs = 0; + num_es_gprs = 0; + num_ps_threads = 188; + num_vs_threads = 60; + num_gs_threads = 0; + num_es_threads = 0; + num_ps_stack_entries = 128; + num_vs_stack_entries = 128; + num_gs_stack_entries = 0; + num_es_stack_entries = 0; + break; + case CHIP_RV710: + num_ps_gprs = 192; + num_vs_gprs = 56; + num_temp_gprs = 4; + num_gs_gprs = 0; + num_es_gprs = 0; + num_ps_threads = 144; + num_vs_threads = 48; + num_gs_threads = 0; + num_es_threads = 0; + num_ps_stack_entries = 128; + num_vs_stack_entries = 128; + num_gs_stack_entries = 0; + num_es_stack_entries = 0; + break; + } + + if ((rdev->family == CHIP_RV610) || + (rdev->family == CHIP_RV620) || + (rdev->family == CHIP_RS780) || + (rdev->family == CHIP_RS880) || + (rdev->family == CHIP_RV710)) + sq_config = 0; + else + sq_config = VC_ENABLE; + + sq_config |= (DX9_CONSTS | + ALU_INST_PREFER_VECTOR | + PS_PRIO(0) | + VS_PRIO(1) | + GS_PRIO(2) | + ES_PRIO(3)); + + sq_gpr_resource_mgmt_1 = (NUM_PS_GPRS(num_ps_gprs) | + NUM_VS_GPRS(num_vs_gprs) | + NUM_CLAUSE_TEMP_GPRS(num_temp_gprs)); + sq_gpr_resource_mgmt_2 = (NUM_GS_GPRS(num_gs_gprs) | + NUM_ES_GPRS(num_es_gprs)); + sq_thread_resource_mgmt = (NUM_PS_THREADS(num_ps_threads) | + NUM_VS_THREADS(num_vs_threads) | + NUM_GS_THREADS(num_gs_threads) | + NUM_ES_THREADS(num_es_threads)); + sq_stack_resource_mgmt_1 = (NUM_PS_STACK_ENTRIES(num_ps_stack_entries) | + NUM_VS_STACK_ENTRIES(num_vs_stack_entries)); + sq_stack_resource_mgmt_2 = (NUM_GS_STACK_ENTRIES(num_gs_stack_entries) | + NUM_ES_STACK_ENTRIES(num_es_stack_entries)); + + /* emit an IB pointing at default state */ + dwords = ALIGN(rdev->r600_blit.state_len, 0x10); + gpu_addr = rdev->r600_blit.shader_gpu_addr + rdev->r600_blit.state_offset; + radeon_ring_write(rdev, PACKET3(PACKET3_INDIRECT_BUFFER, 2)); + radeon_ring_write(rdev, +#ifdef __BIG_ENDIAN + (2 << 0) | +#endif + (gpu_addr & 0xFFFFFFFC)); + radeon_ring_write(rdev, upper_32_bits(gpu_addr) & 0xFF); + radeon_ring_write(rdev, dwords); + + /* SQ config */ + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONFIG_REG, 6)); + radeon_ring_write(rdev, (SQ_CONFIG - PACKET3_SET_CONFIG_REG_OFFSET) >> 2); + radeon_ring_write(rdev, sq_config); + radeon_ring_write(rdev, sq_gpr_resource_mgmt_1); + radeon_ring_write(rdev, sq_gpr_resource_mgmt_2); + radeon_ring_write(rdev, sq_thread_resource_mgmt); + radeon_ring_write(rdev, sq_stack_resource_mgmt_1); + radeon_ring_write(rdev, sq_stack_resource_mgmt_2); +} + +static inline uint32_t i2f(uint32_t input) +{ + u32 result, i, exponent, fraction; + + if ((input & 0x3fff) == 0) + result = 0; /* 0 is a special case */ + else { + exponent = 140; /* exponent biased by 127; */ + fraction = (input & 0x3fff) << 10; /* cheat and only + handle numbers below 2^^15 */ + for (i = 0; i < 14; i++) { + if (fraction & 0x800000) + break; + else { + fraction = fraction << 1; /* keep + shifting left until top bit = 1 */ + exponent = exponent - 1; + } + } + result = exponent << 23 | (fraction & 0x7fffff); /* mask + off top bit; assumed 1 */ + } + return result; +} + +int r600_blit_init(struct radeon_device *rdev) +{ + u32 obj_size; + int i, r, dwords; + void *ptr; + u32 packet2s[16]; + int num_packet2s = 0; + + /* pin copy shader into vram if already initialized */ + if (rdev->r600_blit.shader_obj) + goto done; + + mutex_init(&rdev->r600_blit.mutex); + rdev->r600_blit.state_offset = 0; + + if (rdev->family >= CHIP_RV770) + rdev->r600_blit.state_len = r7xx_default_size; + else + rdev->r600_blit.state_len = r6xx_default_size; + + dwords = rdev->r600_blit.state_len; + while (dwords & 0xf) { + packet2s[num_packet2s++] = cpu_to_le32(PACKET2(0)); + dwords++; + } + + obj_size = dwords * 4; + obj_size = ALIGN(obj_size, 256); + + rdev->r600_blit.vs_offset = obj_size; + obj_size += r6xx_vs_size * 4; + obj_size = ALIGN(obj_size, 256); + + rdev->r600_blit.ps_offset = obj_size; + obj_size += r6xx_ps_size * 4; + obj_size = ALIGN(obj_size, 256); + + r = radeon_bo_create(rdev, obj_size, PAGE_SIZE, true, RADEON_GEM_DOMAIN_VRAM, + &rdev->r600_blit.shader_obj); + if (r) { + DRM_ERROR("r600 failed to allocate shader\n"); + return r; + } + + DRM_DEBUG("r6xx blit allocated bo %08x vs %08x ps %08x\n", + obj_size, + rdev->r600_blit.vs_offset, rdev->r600_blit.ps_offset); + + r = radeon_bo_reserve(rdev->r600_blit.shader_obj, false); + if (unlikely(r != 0)) + return r; + r = radeon_bo_kmap(rdev->r600_blit.shader_obj, &ptr); + if (r) { + DRM_ERROR("failed to map blit object %d\n", r); + return r; + } + if (rdev->family >= CHIP_RV770) + memcpy_toio(ptr + rdev->r600_blit.state_offset, + r7xx_default_state, rdev->r600_blit.state_len * 4); + else + memcpy_toio(ptr + rdev->r600_blit.state_offset, + r6xx_default_state, rdev->r600_blit.state_len * 4); + if (num_packet2s) + memcpy_toio(ptr + rdev->r600_blit.state_offset + (rdev->r600_blit.state_len * 4), + packet2s, num_packet2s * 4); + for (i = 0; i < r6xx_vs_size; i++) + *(u32 *)((unsigned long)ptr + rdev->r600_blit.vs_offset + i * 4) = cpu_to_le32(r6xx_vs[i]); + for (i = 0; i < r6xx_ps_size; i++) + *(u32 *)((unsigned long)ptr + rdev->r600_blit.ps_offset + i * 4) = cpu_to_le32(r6xx_ps[i]); + radeon_bo_kunmap(rdev->r600_blit.shader_obj); + radeon_bo_unreserve(rdev->r600_blit.shader_obj); + +done: + r = radeon_bo_reserve(rdev->r600_blit.shader_obj, false); + if (unlikely(r != 0)) + return r; + r = radeon_bo_pin(rdev->r600_blit.shader_obj, RADEON_GEM_DOMAIN_VRAM, + &rdev->r600_blit.shader_gpu_addr); + radeon_bo_unreserve(rdev->r600_blit.shader_obj); + if (r) { + dev_err(rdev->dev, "(%d) pin blit object failed\n", r); + return r; + } + radeon_ttm_set_active_vram_size(rdev, rdev->mc.real_vram_size); + return 0; +} + +void r600_blit_fini(struct radeon_device *rdev) +{ + int r; + + radeon_ttm_set_active_vram_size(rdev, rdev->mc.visible_vram_size); + if (rdev->r600_blit.shader_obj == NULL) + return; + /* If we can't reserve the bo, unref should be enough to destroy + * it when it becomes idle. + */ + r = radeon_bo_reserve(rdev->r600_blit.shader_obj, false); + if (!r) { + radeon_bo_unpin(rdev->r600_blit.shader_obj); + radeon_bo_unreserve(rdev->r600_blit.shader_obj); + } + radeon_bo_unref(&rdev->r600_blit.shader_obj); +} + +static int r600_vb_ib_get(struct radeon_device *rdev) +{ + int r; + r = radeon_ib_get(rdev, &rdev->r600_blit.vb_ib); + if (r) { + DRM_ERROR("failed to get IB for vertex buffer\n"); + return r; + } + + rdev->r600_blit.vb_total = 64*1024; + rdev->r600_blit.vb_used = 0; + return 0; +} + +static void r600_vb_ib_put(struct radeon_device *rdev) +{ + radeon_fence_emit(rdev, rdev->r600_blit.vb_ib->fence); + radeon_ib_free(rdev, &rdev->r600_blit.vb_ib); +} + +int r600_blit_prepare_copy(struct radeon_device *rdev, int size_bytes) +{ + int r; + int ring_size, line_size; + int max_size; + /* loops of emits 64 + fence emit possible */ + int dwords_per_loop = 76, num_loops; + + r = r600_vb_ib_get(rdev); + if (r) + return r; + + /* set_render_target emits 2 extra dwords on rv6xx */ + if (rdev->family > CHIP_R600 && rdev->family < CHIP_RV770) + dwords_per_loop += 2; + + /* 8 bpp vs 32 bpp for xfer unit */ + if (size_bytes & 3) + line_size = 8192; + else + line_size = 8192*4; + + max_size = 8192 * line_size; + + /* major loops cover the max size transfer */ + num_loops = ((size_bytes + max_size) / max_size); + /* minor loops cover the extra non aligned bits */ + num_loops += ((size_bytes % line_size) ? 1 : 0); + /* calculate number of loops correctly */ + ring_size = num_loops * dwords_per_loop; + /* set default + shaders */ + ring_size += 40; /* shaders + def state */ + ring_size += 10; /* fence emit for VB IB */ + ring_size += 5; /* done copy */ + ring_size += 10; /* fence emit for done copy */ + r = radeon_ring_lock(rdev, ring_size); + if (r) + return r; + + set_default_state(rdev); /* 14 */ + set_shaders(rdev); /* 26 */ + return 0; +} + +void r600_blit_done_copy(struct radeon_device *rdev, struct radeon_fence *fence) +{ + int r; + + if (rdev->r600_blit.vb_ib) + r600_vb_ib_put(rdev); + + if (fence) + r = radeon_fence_emit(rdev, fence); + + radeon_ring_unlock_commit(rdev); +} + +void r600_kms_blit_copy(struct radeon_device *rdev, + u64 src_gpu_addr, u64 dst_gpu_addr, + int size_bytes) +{ + int max_bytes; + u64 vb_gpu_addr; + u32 *vb; + + DRM_DEBUG("emitting copy %16llx %16llx %d %d\n", src_gpu_addr, dst_gpu_addr, + size_bytes, rdev->r600_blit.vb_used); + vb = (u32 *)(rdev->r600_blit.vb_ib->ptr + rdev->r600_blit.vb_used); + if ((size_bytes & 3) || (src_gpu_addr & 3) || (dst_gpu_addr & 3)) { + max_bytes = 8192; + + while (size_bytes) { + int cur_size = size_bytes; + int src_x = src_gpu_addr & 255; + int dst_x = dst_gpu_addr & 255; + int h = 1; + src_gpu_addr = src_gpu_addr & ~255ULL; + dst_gpu_addr = dst_gpu_addr & ~255ULL; + + if (!src_x && !dst_x) { + h = (cur_size / max_bytes); + if (h > 8192) + h = 8192; + if (h == 0) + h = 1; + else + cur_size = max_bytes; + } else { + if (cur_size > max_bytes) + cur_size = max_bytes; + if (cur_size > (max_bytes - dst_x)) + cur_size = (max_bytes - dst_x); + if (cur_size > (max_bytes - src_x)) + cur_size = (max_bytes - src_x); + } + + if ((rdev->r600_blit.vb_used + 48) > rdev->r600_blit.vb_total) { + WARN_ON(1); + } + + vb[0] = i2f(dst_x); + vb[1] = 0; + vb[2] = i2f(src_x); + vb[3] = 0; + + vb[4] = i2f(dst_x); + vb[5] = i2f(h); + vb[6] = i2f(src_x); + vb[7] = i2f(h); + + vb[8] = i2f(dst_x + cur_size); + vb[9] = i2f(h); + vb[10] = i2f(src_x + cur_size); + vb[11] = i2f(h); + + /* src 9 */ + set_tex_resource(rdev, FMT_8, + src_x + cur_size, h, src_x + cur_size, + src_gpu_addr); + + /* 5 */ + cp_set_surface_sync(rdev, + PACKET3_TC_ACTION_ENA, (src_x + cur_size * h), src_gpu_addr); + + /* dst 23 */ + set_render_target(rdev, COLOR_8, + dst_x + cur_size, h, + dst_gpu_addr); + + /* scissors 12 */ + set_scissors(rdev, dst_x, 0, dst_x + cur_size, h); + + /* 14 */ + vb_gpu_addr = rdev->r600_blit.vb_ib->gpu_addr + rdev->r600_blit.vb_used; + set_vtx_resource(rdev, vb_gpu_addr); + + /* draw 10 */ + draw_auto(rdev); + + /* 5 */ + cp_set_surface_sync(rdev, + PACKET3_CB_ACTION_ENA | PACKET3_CB0_DEST_BASE_ENA, + cur_size * h, dst_gpu_addr); + + vb += 12; + rdev->r600_blit.vb_used += 12 * 4; + + src_gpu_addr += cur_size * h; + dst_gpu_addr += cur_size * h; + size_bytes -= cur_size * h; + } + } else { + max_bytes = 8192 * 4; + + while (size_bytes) { + int cur_size = size_bytes; + int src_x = (src_gpu_addr & 255); + int dst_x = (dst_gpu_addr & 255); + int h = 1; + src_gpu_addr = src_gpu_addr & ~255ULL; + dst_gpu_addr = dst_gpu_addr & ~255ULL; + + if (!src_x && !dst_x) { + h = (cur_size / max_bytes); + if (h > 8192) + h = 8192; + if (h == 0) + h = 1; + else + cur_size = max_bytes; + } else { + if (cur_size > max_bytes) + cur_size = max_bytes; + if (cur_size > (max_bytes - dst_x)) + cur_size = (max_bytes - dst_x); + if (cur_size > (max_bytes - src_x)) + cur_size = (max_bytes - src_x); + } + + if ((rdev->r600_blit.vb_used + 48) > rdev->r600_blit.vb_total) { + WARN_ON(1); + } + + vb[0] = i2f(dst_x / 4); + vb[1] = 0; + vb[2] = i2f(src_x / 4); + vb[3] = 0; + + vb[4] = i2f(dst_x / 4); + vb[5] = i2f(h); + vb[6] = i2f(src_x / 4); + vb[7] = i2f(h); + + vb[8] = i2f((dst_x + cur_size) / 4); + vb[9] = i2f(h); + vb[10] = i2f((src_x + cur_size) / 4); + vb[11] = i2f(h); + + /* src 9 */ + set_tex_resource(rdev, FMT_8_8_8_8, + (src_x + cur_size) / 4, + h, (src_x + cur_size) / 4, + src_gpu_addr); + /* 5 */ + cp_set_surface_sync(rdev, + PACKET3_TC_ACTION_ENA, (src_x + cur_size * h), src_gpu_addr); + + /* dst 23 */ + set_render_target(rdev, COLOR_8_8_8_8, + (dst_x + cur_size) / 4, h, + dst_gpu_addr); + + /* scissors 12 */ + set_scissors(rdev, (dst_x / 4), 0, (dst_x + cur_size / 4), h); + + /* Vertex buffer setup 14 */ + vb_gpu_addr = rdev->r600_blit.vb_ib->gpu_addr + rdev->r600_blit.vb_used; + set_vtx_resource(rdev, vb_gpu_addr); + + /* draw 10 */ + draw_auto(rdev); + + /* 5 */ + cp_set_surface_sync(rdev, + PACKET3_CB_ACTION_ENA | PACKET3_CB0_DEST_BASE_ENA, + cur_size * h, dst_gpu_addr); + + /* 78 ring dwords per loop */ + vb += 12; + rdev->r600_blit.vb_used += 12 * 4; + + src_gpu_addr += cur_size * h; + dst_gpu_addr += cur_size * h; + size_bytes -= cur_size * h; + } + } +} + diff --git a/drivers/gpu/drm/radeon/r600_blit_shaders.c b/drivers/gpu/drm/radeon/r600_blit_shaders.c new file mode 100644 index 00000000..73e2c7c6 --- /dev/null +++ b/drivers/gpu/drm/radeon/r600_blit_shaders.c @@ -0,0 +1,718 @@ +/* + * Copyright 2009 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Alex Deucher + */ + +#include +#include + +/* + * R6xx+ cards need to use the 3D engine to blit data which requires + * quite a bit of hw state setup. Rather than pull the whole 3D driver + * (which normally generates the 3D state) into the DRM, we opt to use + * statically generated state tables. The regsiter state and shaders + * were hand generated to support blitting functionality. See the 3D + * driver or documentation for descriptions of the registers and + * shader instructions. + */ + +const u32 r6xx_default_state[] = +{ + 0xc0002400, /* START_3D_CMDBUF */ + 0x00000000, + + 0xc0012800, /* CONTEXT_CONTROL */ + 0x80000000, + 0x80000000, + + 0xc0016800, + 0x00000010, + 0x00008000, /* WAIT_UNTIL */ + + 0xc0016800, + 0x00000542, + 0x07000003, /* TA_CNTL_AUX */ + + 0xc0016800, + 0x000005c5, + 0x00000000, /* VC_ENHANCE */ + + 0xc0016800, + 0x00000363, + 0x00000000, /* SQ_DYN_GPR_CNTL_PS_FLUSH_REQ */ + + 0xc0016800, + 0x0000060c, + 0x82000000, /* DB_DEBUG */ + + 0xc0016800, + 0x0000060e, + 0x01020204, /* DB_WATERMARKS */ + + 0xc0026f00, + 0x00000000, + 0x00000000, /* SQ_VTX_BASE_VTX_LOC */ + 0x00000000, /* SQ_VTX_START_INST_LOC */ + + 0xc0096900, + 0x0000022a, + 0x00000000, /* SQ_ESGS_RING_ITEMSIZE */ + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + + 0xc0016900, + 0x00000004, + 0x00000000, /* DB_DEPTH_INFO */ + + 0xc0026900, + 0x0000000a, + 0x00000000, /* DB_STENCIL_CLEAR */ + 0x00000000, /* DB_DEPTH_CLEAR */ + + 0xc0016900, + 0x00000200, + 0x00000000, /* DB_DEPTH_CONTROL */ + + 0xc0026900, + 0x00000343, + 0x00000060, /* DB_RENDER_CONTROL */ + 0x00000040, /* DB_RENDER_OVERRIDE */ + + 0xc0016900, + 0x00000351, + 0x0000aa00, /* DB_ALPHA_TO_MASK */ + + 0xc00f6900, + 0x00000100, + 0x00000800, /* VGT_MAX_VTX_INDX */ + 0x00000000, /* VGT_MIN_VTX_INDX */ + 0x00000000, /* VGT_INDX_OFFSET */ + 0x00000000, /* VGT_MULTI_PRIM_IB_RESET_INDX */ + 0x00000000, /* SX_ALPHA_TEST_CONTROL */ + 0x00000000, /* CB_BLEND_RED */ + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, /* CB_FOG_RED */ + 0x00000000, + 0x00000000, + 0x00000000, /* DB_STENCILREFMASK */ + 0x00000000, /* DB_STENCILREFMASK_BF */ + 0x00000000, /* SX_ALPHA_REF */ + + 0xc0046900, + 0x0000030c, + 0x01000000, /* CB_CLRCMP_CNTL */ + 0x00000000, + 0x00000000, + 0x00000000, + + 0xc0046900, + 0x00000048, + 0x3f800000, /* CB_CLEAR_RED */ + 0x00000000, + 0x3f800000, + 0x3f800000, + + 0xc0016900, + 0x00000080, + 0x00000000, /* PA_SC_WINDOW_OFFSET */ + + 0xc00a6900, + 0x00000083, + 0x0000ffff, /* PA_SC_CLIP_RECT_RULE */ + 0x00000000, /* PA_SC_CLIPRECT_0_TL */ + 0x20002000, + 0x00000000, + 0x20002000, + 0x00000000, + 0x20002000, + 0x00000000, + 0x20002000, + 0x00000000, /* PA_SC_EDGERULE */ + + 0xc0406900, + 0x00000094, + 0x80000000, /* PA_SC_VPORT_SCISSOR_0_TL */ + 0x20002000, /* PA_SC_VPORT_SCISSOR_0_BR */ + 0x80000000, /* PA_SC_VPORT_SCISSOR_1_TL */ + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x00000000, /* PA_SC_VPORT_ZMIN_0 */ + 0x3f800000, + 0x00000000, + 0x3f800000, + 0x00000000, + 0x3f800000, + 0x00000000, + 0x3f800000, + 0x00000000, + 0x3f800000, + 0x00000000, + 0x3f800000, + 0x00000000, + 0x3f800000, + 0x00000000, + 0x3f800000, + 0x00000000, + 0x3f800000, + 0x00000000, + 0x3f800000, + 0x00000000, + 0x3f800000, + 0x00000000, + 0x3f800000, + 0x00000000, + 0x3f800000, + 0x00000000, + 0x3f800000, + 0x00000000, + 0x3f800000, + 0x00000000, + 0x3f800000, + + 0xc0026900, + 0x00000292, + 0x00000000, /* PA_SC_MPASS_PS_CNTL */ + 0x00004010, /* PA_SC_MODE_CNTL */ + + 0xc0096900, + 0x00000300, + 0x00000000, /* PA_SC_LINE_CNTL */ + 0x00000000, /* PA_SC_AA_CONFIG */ + 0x0000002d, /* PA_SU_VTX_CNTL */ + 0x3f800000, /* PA_CL_GB_VERT_CLIP_ADJ */ + 0x3f800000, + 0x3f800000, + 0x3f800000, + 0x00000000, /* PA_SC_SAMPLE_LOCS_MCTX */ + 0x00000000, + + 0xc0016900, + 0x00000312, + 0xffffffff, /* PA_SC_AA_MASK */ + + 0xc0066900, + 0x0000037e, + 0x00000000, /* PA_SU_POLY_OFFSET_DB_FMT_CNTL */ + 0x00000000, /* PA_SU_POLY_OFFSET_CLAMP */ + 0x00000000, /* PA_SU_POLY_OFFSET_FRONT_SCALE */ + 0x00000000, /* PA_SU_POLY_OFFSET_FRONT_OFFSET */ + 0x00000000, /* PA_SU_POLY_OFFSET_BACK_SCALE */ + 0x00000000, /* PA_SU_POLY_OFFSET_BACK_OFFSET */ + + 0xc0046900, + 0x000001b6, + 0x00000000, /* SPI_INPUT_Z */ + 0x00000000, /* SPI_FOG_CNTL */ + 0x00000000, /* SPI_FOG_FUNC_SCALE */ + 0x00000000, /* SPI_FOG_FUNC_BIAS */ + + 0xc0016900, + 0x00000225, + 0x00000000, /* SQ_PGM_START_FS */ + + 0xc0016900, + 0x00000229, + 0x00000000, /* SQ_PGM_RESOURCES_FS */ + + 0xc0016900, + 0x00000237, + 0x00000000, /* SQ_PGM_CF_OFFSET_FS */ + + 0xc0026900, + 0x000002a8, + 0x00000000, /* VGT_INSTANCE_STEP_RATE_0 */ + 0x00000000, /* VGT_INSTANCE_STEP_RATE_1 */ + + 0xc0116900, + 0x00000280, + 0x00000000, /* PA_SU_POINT_SIZE */ + 0x00000000, /* PA_SU_POINT_MINMAX */ + 0x00000008, /* PA_SU_LINE_CNTL */ + 0x00000000, /* PA_SC_LINE_STIPPLE */ + 0x00000000, /* VGT_OUTPUT_PATH_CNTL */ + 0x00000000, /* VGT_HOS_CNTL */ + 0x00000000, /* VGT_HOS_MAX_TESS_LEVEL */ + 0x00000000, /* VGT_HOS_MIN_TESS_LEVEL */ + 0x00000000, /* VGT_HOS_REUSE_DEPTH */ + 0x00000000, /* VGT_GROUP_PRIM_TYPE */ + 0x00000000, /* VGT_GROUP_FIRST_DECR */ + 0x00000000, /* VGT_GROUP_DECR */ + 0x00000000, /* VGT_GROUP_VECT_0_CNTL */ + 0x00000000, /* VGT_GROUP_VECT_1_CNTL */ + 0x00000000, /* VGT_GROUP_VECT_0_FMT_CNTL */ + 0x00000000, /* VGT_GROUP_VECT_1_FMT_CNTL */ + 0x00000000, /* VGT_GS_MODE */ + + 0xc0016900, + 0x000002a1, + 0x00000000, /* VGT_PRIMITIVEID_EN */ + + 0xc0016900, + 0x000002a5, + 0x00000000, /* VGT_MULTI_PRIM_ID_RESET_EN */ + + 0xc0036900, + 0x000002ac, + 0x00000000, /* VGT_STRMOUT_EN */ + 0x00000000, /* VGT_REUSE_OFF */ + 0x00000000, /* VGT_VTX_CNT_EN */ + + 0xc0016900, + 0x000000d4, + 0x00000000, /* SX_MISC */ + + 0xc0016900, + 0x000002c8, + 0x00000000, /* VGT_STRMOUT_BUFFER_EN */ + + 0xc0076900, + 0x00000202, + 0x00cc0000, /* CB_COLOR_CONTROL */ + 0x00000210, /* DB_SHADER_CNTL */ + 0x00010000, /* PA_CL_CLIP_CNTL */ + 0x00000244, /* PA_SU_SC_MODE_CNTL */ + 0x00000100, /* PA_CL_VTE_CNTL */ + 0x00000000, /* PA_CL_VS_OUT_CNTL */ + 0x00000000, /* PA_CL_NANINF_CNTL */ + + 0xc0026900, + 0x0000008e, + 0x0000000f, /* CB_TARGET_MASK */ + 0x0000000f, /* CB_SHADER_MASK */ + + 0xc0016900, + 0x000001e8, + 0x00000001, /* CB_SHADER_CONTROL */ + + 0xc0016900, + 0x00000185, + 0x00000000, /* SPI_VS_OUT_ID_0 */ + + 0xc0016900, + 0x00000191, + 0x00000b00, /* SPI_PS_INPUT_CNTL_0 */ + + 0xc0056900, + 0x000001b1, + 0x00000000, /* SPI_VS_OUT_CONFIG */ + 0x00000000, /* SPI_THREAD_GROUPING */ + 0x00000001, /* SPI_PS_IN_CONTROL_0 */ + 0x00000000, /* SPI_PS_IN_CONTROL_1 */ + 0x00000000, /* SPI_INTERP_CONTROL_0 */ + + 0xc0036e00, /* SET_SAMPLER */ + 0x00000000, + 0x00000012, + 0x00000000, + 0x00000000, +}; + +const u32 r7xx_default_state[] = +{ + 0xc0012800, /* CONTEXT_CONTROL */ + 0x80000000, + 0x80000000, + + 0xc0016800, + 0x00000010, + 0x00008000, /* WAIT_UNTIL */ + + 0xc0016800, + 0x00000542, + 0x07000002, /* TA_CNTL_AUX */ + + 0xc0016800, + 0x000005c5, + 0x00000000, /* VC_ENHANCE */ + + 0xc0016800, + 0x00000363, + 0x00004000, /* SQ_DYN_GPR_CNTL_PS_FLUSH_REQ */ + + 0xc0016800, + 0x0000060c, + 0x00000000, /* DB_DEBUG */ + + 0xc0016800, + 0x0000060e, + 0x00420204, /* DB_WATERMARKS */ + + 0xc0026f00, + 0x00000000, + 0x00000000, /* SQ_VTX_BASE_VTX_LOC */ + 0x00000000, /* SQ_VTX_START_INST_LOC */ + + 0xc0096900, + 0x0000022a, + 0x00000000, /* SQ_ESGS_RING_ITEMSIZE */ + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + + 0xc0016900, + 0x00000004, + 0x00000000, /* DB_DEPTH_INFO */ + + 0xc0026900, + 0x0000000a, + 0x00000000, /* DB_STENCIL_CLEAR */ + 0x00000000, /* DB_DEPTH_CLEAR */ + + 0xc0016900, + 0x00000200, + 0x00000000, /* DB_DEPTH_CONTROL */ + + 0xc0026900, + 0x00000343, + 0x00000060, /* DB_RENDER_CONTROL */ + 0x00000000, /* DB_RENDER_OVERRIDE */ + + 0xc0016900, + 0x00000351, + 0x0000aa00, /* DB_ALPHA_TO_MASK */ + + 0xc0096900, + 0x00000100, + 0x00000800, /* VGT_MAX_VTX_INDX */ + 0x00000000, /* VGT_MIN_VTX_INDX */ + 0x00000000, /* VGT_INDX_OFFSET */ + 0x00000000, /* VGT_MULTI_PRIM_IB_RESET_INDX */ + 0x00000000, /* SX_ALPHA_TEST_CONTROL */ + 0x00000000, /* CB_BLEND_RED */ + 0x00000000, + 0x00000000, + 0x00000000, + + 0xc0036900, + 0x0000010c, + 0x00000000, /* DB_STENCILREFMASK */ + 0x00000000, /* DB_STENCILREFMASK_BF */ + 0x00000000, /* SX_ALPHA_REF */ + + 0xc0046900, + 0x0000030c, /* CB_CLRCMP_CNTL */ + 0x01000000, + 0x00000000, + 0x00000000, + 0x00000000, + + 0xc0016900, + 0x00000080, + 0x00000000, /* PA_SC_WINDOW_OFFSET */ + + 0xc00a6900, + 0x00000083, + 0x0000ffff, /* PA_SC_CLIP_RECT_RULE */ + 0x00000000, /* PA_SC_CLIPRECT_0_TL */ + 0x20002000, + 0x00000000, + 0x20002000, + 0x00000000, + 0x20002000, + 0x00000000, + 0x20002000, + 0xaaaaaaaa, /* PA_SC_EDGERULE */ + + 0xc0406900, + 0x00000094, + 0x80000000, /* PA_SC_VPORT_SCISSOR_0_TL */ + 0x20002000, /* PA_SC_VPORT_SCISSOR_0_BR */ + 0x80000000, /* PA_SC_VPORT_SCISSOR_1_TL */ + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x00000000, /* PA_SC_VPORT_ZMIN_0 */ + 0x3f800000, + 0x00000000, + 0x3f800000, + 0x00000000, + 0x3f800000, + 0x00000000, + 0x3f800000, + 0x00000000, + 0x3f800000, + 0x00000000, + 0x3f800000, + 0x00000000, + 0x3f800000, + 0x00000000, + 0x3f800000, + 0x00000000, + 0x3f800000, + 0x00000000, + 0x3f800000, + 0x00000000, + 0x3f800000, + 0x00000000, + 0x3f800000, + 0x00000000, + 0x3f800000, + 0x00000000, + 0x3f800000, + 0x00000000, + 0x3f800000, + 0x00000000, + 0x3f800000, + + 0xc0026900, + 0x00000292, + 0x00000000, /* PA_SC_MPASS_PS_CNTL */ + 0x00514000, /* PA_SC_MODE_CNTL */ + + 0xc0096900, + 0x00000300, + 0x00000000, /* PA_SC_LINE_CNTL */ + 0x00000000, /* PA_SC_AA_CONFIG */ + 0x0000002d, /* PA_SU_VTX_CNTL */ + 0x3f800000, /* PA_CL_GB_VERT_CLIP_ADJ */ + 0x3f800000, + 0x3f800000, + 0x3f800000, + 0x00000000, /* PA_SC_SAMPLE_LOCS_MCTX */ + 0x00000000, + + 0xc0016900, + 0x00000312, + 0xffffffff, /* PA_SC_AA_MASK */ + + 0xc0066900, + 0x0000037e, + 0x00000000, /* PA_SU_POLY_OFFSET_DB_FMT_CNTL */ + 0x00000000, /* PA_SU_POLY_OFFSET_CLAMP */ + 0x00000000, /* PA_SU_POLY_OFFSET_FRONT_SCALE */ + 0x00000000, /* PA_SU_POLY_OFFSET_FRONT_OFFSET */ + 0x00000000, /* PA_SU_POLY_OFFSET_BACK_SCALE */ + 0x00000000, /* PA_SU_POLY_OFFSET_BACK_OFFSET */ + + 0xc0046900, + 0x000001b6, + 0x00000000, /* SPI_INPUT_Z */ + 0x00000000, /* SPI_FOG_CNTL */ + 0x00000000, /* SPI_FOG_FUNC_SCALE */ + 0x00000000, /* SPI_FOG_FUNC_BIAS */ + + 0xc0016900, + 0x00000225, + 0x00000000, /* SQ_PGM_START_FS */ + + 0xc0016900, + 0x00000229, + 0x00000000, /* SQ_PGM_RESOURCES_FS */ + + 0xc0016900, + 0x00000237, + 0x00000000, /* SQ_PGM_CF_OFFSET_FS */ + + 0xc0026900, + 0x000002a8, + 0x00000000, /* VGT_INSTANCE_STEP_RATE_0 */ + 0x00000000, /* VGT_INSTANCE_STEP_RATE_1 */ + + 0xc0116900, + 0x00000280, + 0x00000000, /* PA_SU_POINT_SIZE */ + 0x00000000, /* PA_SU_POINT_MINMAX */ + 0x00000008, /* PA_SU_LINE_CNTL */ + 0x00000000, /* PA_SC_LINE_STIPPLE */ + 0x00000000, /* VGT_OUTPUT_PATH_CNTL */ + 0x00000000, /* VGT_HOS_CNTL */ + 0x00000000, /* VGT_HOS_MAX_TESS_LEVEL */ + 0x00000000, /* VGT_HOS_MIN_TESS_LEVEL */ + 0x00000000, /* VGT_HOS_REUSE_DEPTH */ + 0x00000000, /* VGT_GROUP_PRIM_TYPE */ + 0x00000000, /* VGT_GROUP_FIRST_DECR */ + 0x00000000, /* VGT_GROUP_DECR */ + 0x00000000, /* VGT_GROUP_VECT_0_CNTL */ + 0x00000000, /* VGT_GROUP_VECT_1_CNTL */ + 0x00000000, /* VGT_GROUP_VECT_0_FMT_CNTL */ + 0x00000000, /* VGT_GROUP_VECT_1_FMT_CNTL */ + 0x00000000, /* VGT_GS_MODE */ + + 0xc0016900, + 0x000002a1, + 0x00000000, /* VGT_PRIMITIVEID_EN */ + + 0xc0016900, + 0x000002a5, + 0x00000000, /* VGT_MULTI_PRIM_ID_RESET_EN */ + + 0xc0036900, + 0x000002ac, + 0x00000000, /* VGT_STRMOUT_EN */ + 0x00000000, /* VGT_REUSE_OFF */ + 0x00000000, /* VGT_VTX_CNT_EN */ + + 0xc0016900, + 0x000000d4, + 0x00000000, /* SX_MISC */ + + 0xc0016900, + 0x000002c8, + 0x00000000, /* VGT_STRMOUT_BUFFER_EN */ + + 0xc0076900, + 0x00000202, + 0x00cc0000, /* CB_COLOR_CONTROL */ + 0x00000210, /* DB_SHADER_CNTL */ + 0x00010000, /* PA_CL_CLIP_CNTL */ + 0x00000244, /* PA_SU_SC_MODE_CNTL */ + 0x00000100, /* PA_CL_VTE_CNTL */ + 0x00000000, /* PA_CL_VS_OUT_CNTL */ + 0x00000000, /* PA_CL_NANINF_CNTL */ + + 0xc0026900, + 0x0000008e, + 0x0000000f, /* CB_TARGET_MASK */ + 0x0000000f, /* CB_SHADER_MASK */ + + 0xc0016900, + 0x000001e8, + 0x00000001, /* CB_SHADER_CONTROL */ + + 0xc0016900, + 0x00000185, + 0x00000000, /* SPI_VS_OUT_ID_0 */ + + 0xc0016900, + 0x00000191, + 0x00000b00, /* SPI_PS_INPUT_CNTL_0 */ + + 0xc0056900, + 0x000001b1, + 0x00000000, /* SPI_VS_OUT_CONFIG */ + 0x00000001, /* SPI_THREAD_GROUPING */ + 0x00000001, /* SPI_PS_IN_CONTROL_0 */ + 0x00000000, /* SPI_PS_IN_CONTROL_1 */ + 0x00000000, /* SPI_INTERP_CONTROL_0 */ + + 0xc0036e00, /* SET_SAMPLER */ + 0x00000000, + 0x00000012, + 0x00000000, + 0x00000000, +}; + +/* same for r6xx/r7xx */ +const u32 r6xx_vs[] = +{ + 0x00000004, + 0x81000000, + 0x0000203c, + 0x94000b08, + 0x00004000, + 0x14200b1a, + 0x00000000, + 0x00000000, + 0x3c000000, + 0x68cd1000, +#ifdef __BIG_ENDIAN + 0x000a0000, +#else + 0x00080000, +#endif + 0x00000000, +}; + +const u32 r6xx_ps[] = +{ + 0x00000002, + 0x80800000, + 0x00000000, + 0x94200688, + 0x00000010, + 0x000d1000, + 0xb0800000, + 0x00000000, +}; + +const u32 r6xx_ps_size = ARRAY_SIZE(r6xx_ps); +const u32 r6xx_vs_size = ARRAY_SIZE(r6xx_vs); +const u32 r6xx_default_size = ARRAY_SIZE(r6xx_default_state); +const u32 r7xx_default_size = ARRAY_SIZE(r7xx_default_state); diff --git a/drivers/gpu/drm/radeon/r600_blit_shaders.h b/drivers/gpu/drm/radeon/r600_blit_shaders.h new file mode 100644 index 00000000..f437d36d --- /dev/null +++ b/drivers/gpu/drm/radeon/r600_blit_shaders.h @@ -0,0 +1,38 @@ +/* + * Copyright 2009 Advanced Micro Devices, Inc. + * Copyright 2009 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef R600_BLIT_SHADERS_H +#define R600_BLIT_SHADERS_H + +extern const u32 r6xx_ps[]; +extern const u32 r6xx_vs[]; +extern const u32 r7xx_default_state[]; +extern const u32 r6xx_default_state[]; + + +extern const u32 r6xx_ps_size, r6xx_vs_size; +extern const u32 r6xx_default_size, r7xx_default_size; + +#endif diff --git a/drivers/gpu/drm/radeon/r600_cp.c b/drivers/gpu/drm/radeon/r600_cp.c new file mode 100644 index 00000000..c3ab959b --- /dev/null +++ b/drivers/gpu/drm/radeon/r600_cp.c @@ -0,0 +1,2667 @@ +/* + * Copyright 2008-2009 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Dave Airlie + * Alex Deucher + */ + +#include "drmP.h" +#include "drm.h" +#include "radeon_drm.h" +#include "radeon_drv.h" + +#define PFP_UCODE_SIZE 576 +#define PM4_UCODE_SIZE 1792 +#define R700_PFP_UCODE_SIZE 848 +#define R700_PM4_UCODE_SIZE 1360 + +/* Firmware Names */ +MODULE_FIRMWARE("radeon/R600_pfp.bin"); +MODULE_FIRMWARE("radeon/R600_me.bin"); +MODULE_FIRMWARE("radeon/RV610_pfp.bin"); +MODULE_FIRMWARE("radeon/RV610_me.bin"); +MODULE_FIRMWARE("radeon/RV630_pfp.bin"); +MODULE_FIRMWARE("radeon/RV630_me.bin"); +MODULE_FIRMWARE("radeon/RV620_pfp.bin"); +MODULE_FIRMWARE("radeon/RV620_me.bin"); +MODULE_FIRMWARE("radeon/RV635_pfp.bin"); +MODULE_FIRMWARE("radeon/RV635_me.bin"); +MODULE_FIRMWARE("radeon/RV670_pfp.bin"); +MODULE_FIRMWARE("radeon/RV670_me.bin"); +MODULE_FIRMWARE("radeon/RS780_pfp.bin"); +MODULE_FIRMWARE("radeon/RS780_me.bin"); +MODULE_FIRMWARE("radeon/RV770_pfp.bin"); +MODULE_FIRMWARE("radeon/RV770_me.bin"); +MODULE_FIRMWARE("radeon/RV730_pfp.bin"); +MODULE_FIRMWARE("radeon/RV730_me.bin"); +MODULE_FIRMWARE("radeon/RV710_pfp.bin"); +MODULE_FIRMWARE("radeon/RV710_me.bin"); + + +int r600_cs_legacy(struct drm_device *dev, void *data, struct drm_file *filp, + unsigned family, u32 *ib, int *l); +void r600_cs_legacy_init(void); + + +# define ATI_PCIGART_PAGE_SIZE 4096 /**< PCI GART page size */ +# define ATI_PCIGART_PAGE_MASK (~(ATI_PCIGART_PAGE_SIZE-1)) + +#define R600_PTE_VALID (1 << 0) +#define R600_PTE_SYSTEM (1 << 1) +#define R600_PTE_SNOOPED (1 << 2) +#define R600_PTE_READABLE (1 << 5) +#define R600_PTE_WRITEABLE (1 << 6) + +/* MAX values used for gfx init */ +#define R6XX_MAX_SH_GPRS 256 +#define R6XX_MAX_TEMP_GPRS 16 +#define R6XX_MAX_SH_THREADS 256 +#define R6XX_MAX_SH_STACK_ENTRIES 4096 +#define R6XX_MAX_BACKENDS 8 +#define R6XX_MAX_BACKENDS_MASK 0xff +#define R6XX_MAX_SIMDS 8 +#define R6XX_MAX_SIMDS_MASK 0xff +#define R6XX_MAX_PIPES 8 +#define R6XX_MAX_PIPES_MASK 0xff + +#define R7XX_MAX_SH_GPRS 256 +#define R7XX_MAX_TEMP_GPRS 16 +#define R7XX_MAX_SH_THREADS 256 +#define R7XX_MAX_SH_STACK_ENTRIES 4096 +#define R7XX_MAX_BACKENDS 8 +#define R7XX_MAX_BACKENDS_MASK 0xff +#define R7XX_MAX_SIMDS 16 +#define R7XX_MAX_SIMDS_MASK 0xffff +#define R7XX_MAX_PIPES 8 +#define R7XX_MAX_PIPES_MASK 0xff + +static int r600_do_wait_for_fifo(drm_radeon_private_t *dev_priv, int entries) +{ + int i; + + dev_priv->stats.boxes |= RADEON_BOX_WAIT_IDLE; + + for (i = 0; i < dev_priv->usec_timeout; i++) { + int slots; + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_RV770) + slots = (RADEON_READ(R600_GRBM_STATUS) + & R700_CMDFIFO_AVAIL_MASK); + else + slots = (RADEON_READ(R600_GRBM_STATUS) + & R600_CMDFIFO_AVAIL_MASK); + if (slots >= entries) + return 0; + DRM_UDELAY(1); + } + DRM_INFO("wait for fifo failed status : 0x%08X 0x%08X\n", + RADEON_READ(R600_GRBM_STATUS), + RADEON_READ(R600_GRBM_STATUS2)); + + return -EBUSY; +} + +static int r600_do_wait_for_idle(drm_radeon_private_t *dev_priv) +{ + int i, ret; + + dev_priv->stats.boxes |= RADEON_BOX_WAIT_IDLE; + + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_RV770) + ret = r600_do_wait_for_fifo(dev_priv, 8); + else + ret = r600_do_wait_for_fifo(dev_priv, 16); + if (ret) + return ret; + for (i = 0; i < dev_priv->usec_timeout; i++) { + if (!(RADEON_READ(R600_GRBM_STATUS) & R600_GUI_ACTIVE)) + return 0; + DRM_UDELAY(1); + } + DRM_INFO("wait idle failed status : 0x%08X 0x%08X\n", + RADEON_READ(R600_GRBM_STATUS), + RADEON_READ(R600_GRBM_STATUS2)); + + return -EBUSY; +} + +void r600_page_table_cleanup(struct drm_device *dev, struct drm_ati_pcigart_info *gart_info) +{ + struct drm_sg_mem *entry = dev->sg; + int max_pages; + int pages; + int i; + + if (!entry) + return; + + if (gart_info->bus_addr) { + max_pages = (gart_info->table_size / sizeof(u64)); + pages = (entry->pages <= max_pages) + ? entry->pages : max_pages; + + for (i = 0; i < pages; i++) { + if (!entry->busaddr[i]) + break; + pci_unmap_page(dev->pdev, entry->busaddr[i], + PAGE_SIZE, PCI_DMA_BIDIRECTIONAL); + } + if (gart_info->gart_table_location == DRM_ATI_GART_MAIN) + gart_info->bus_addr = 0; + } +} + +/* R600 has page table setup */ +int r600_page_table_init(struct drm_device *dev) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + struct drm_ati_pcigart_info *gart_info = &dev_priv->gart_info; + struct drm_local_map *map = &gart_info->mapping; + struct drm_sg_mem *entry = dev->sg; + int ret = 0; + int i, j; + int pages; + u64 page_base; + dma_addr_t entry_addr; + int max_ati_pages, max_real_pages, gart_idx; + + /* okay page table is available - lets rock */ + max_ati_pages = (gart_info->table_size / sizeof(u64)); + max_real_pages = max_ati_pages / (PAGE_SIZE / ATI_PCIGART_PAGE_SIZE); + + pages = (entry->pages <= max_real_pages) ? + entry->pages : max_real_pages; + + memset_io((void __iomem *)map->handle, 0, max_ati_pages * sizeof(u64)); + + gart_idx = 0; + for (i = 0; i < pages; i++) { + entry->busaddr[i] = pci_map_page(dev->pdev, + entry->pagelist[i], 0, + PAGE_SIZE, + PCI_DMA_BIDIRECTIONAL); + if (pci_dma_mapping_error(dev->pdev, entry->busaddr[i])) { + DRM_ERROR("unable to map PCIGART pages!\n"); + r600_page_table_cleanup(dev, gart_info); + goto done; + } + entry_addr = entry->busaddr[i]; + for (j = 0; j < (PAGE_SIZE / ATI_PCIGART_PAGE_SIZE); j++) { + page_base = (u64) entry_addr & ATI_PCIGART_PAGE_MASK; + page_base |= R600_PTE_VALID | R600_PTE_SYSTEM | R600_PTE_SNOOPED; + page_base |= R600_PTE_READABLE | R600_PTE_WRITEABLE; + + DRM_WRITE64(map, gart_idx * sizeof(u64), page_base); + + gart_idx++; + + if ((i % 128) == 0) + DRM_DEBUG("page entry %d: 0x%016llx\n", + i, (unsigned long long)page_base); + entry_addr += ATI_PCIGART_PAGE_SIZE; + } + } + ret = 1; +done: + return ret; +} + +static void r600_vm_flush_gart_range(struct drm_device *dev) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + u32 resp, countdown = 1000; + RADEON_WRITE(R600_VM_CONTEXT0_INVALIDATION_LOW_ADDR, dev_priv->gart_vm_start >> 12); + RADEON_WRITE(R600_VM_CONTEXT0_INVALIDATION_HIGH_ADDR, (dev_priv->gart_vm_start + dev_priv->gart_size - 1) >> 12); + RADEON_WRITE(R600_VM_CONTEXT0_REQUEST_RESPONSE, 2); + + do { + resp = RADEON_READ(R600_VM_CONTEXT0_REQUEST_RESPONSE); + countdown--; + DRM_UDELAY(1); + } while (((resp & 0xf0) == 0) && countdown); +} + +static void r600_vm_init(struct drm_device *dev) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + /* initialise the VM to use the page table we constructed up there */ + u32 vm_c0, i; + u32 mc_rd_a; + u32 vm_l2_cntl, vm_l2_cntl3; + /* okay set up the PCIE aperture type thingo */ + RADEON_WRITE(R600_MC_VM_SYSTEM_APERTURE_LOW_ADDR, dev_priv->gart_vm_start >> 12); + RADEON_WRITE(R600_MC_VM_SYSTEM_APERTURE_HIGH_ADDR, (dev_priv->gart_vm_start + dev_priv->gart_size - 1) >> 12); + RADEON_WRITE(R600_MC_VM_SYSTEM_APERTURE_DEFAULT_ADDR, 0); + + /* setup MC RD a */ + mc_rd_a = R600_MCD_L1_TLB | R600_MCD_L1_FRAG_PROC | R600_MCD_SYSTEM_ACCESS_MODE_IN_SYS | + R600_MCD_SYSTEM_APERTURE_UNMAPPED_ACCESS_PASS_THRU | R600_MCD_EFFECTIVE_L1_TLB_SIZE(5) | + R600_MCD_EFFECTIVE_L1_QUEUE_SIZE(5) | R600_MCD_WAIT_L2_QUERY; + + RADEON_WRITE(R600_MCD_RD_A_CNTL, mc_rd_a); + RADEON_WRITE(R600_MCD_RD_B_CNTL, mc_rd_a); + + RADEON_WRITE(R600_MCD_WR_A_CNTL, mc_rd_a); + RADEON_WRITE(R600_MCD_WR_B_CNTL, mc_rd_a); + + RADEON_WRITE(R600_MCD_RD_GFX_CNTL, mc_rd_a); + RADEON_WRITE(R600_MCD_WR_GFX_CNTL, mc_rd_a); + + RADEON_WRITE(R600_MCD_RD_SYS_CNTL, mc_rd_a); + RADEON_WRITE(R600_MCD_WR_SYS_CNTL, mc_rd_a); + + RADEON_WRITE(R600_MCD_RD_HDP_CNTL, mc_rd_a | R600_MCD_L1_STRICT_ORDERING); + RADEON_WRITE(R600_MCD_WR_HDP_CNTL, mc_rd_a /*| R600_MCD_L1_STRICT_ORDERING*/); + + RADEON_WRITE(R600_MCD_RD_PDMA_CNTL, mc_rd_a); + RADEON_WRITE(R600_MCD_WR_PDMA_CNTL, mc_rd_a); + + RADEON_WRITE(R600_MCD_RD_SEM_CNTL, mc_rd_a | R600_MCD_SEMAPHORE_MODE); + RADEON_WRITE(R600_MCD_WR_SEM_CNTL, mc_rd_a); + + vm_l2_cntl = R600_VM_L2_CACHE_EN | R600_VM_L2_FRAG_PROC | R600_VM_ENABLE_PTE_CACHE_LRU_W; + vm_l2_cntl |= R600_VM_L2_CNTL_QUEUE_SIZE(7); + RADEON_WRITE(R600_VM_L2_CNTL, vm_l2_cntl); + + RADEON_WRITE(R600_VM_L2_CNTL2, 0); + vm_l2_cntl3 = (R600_VM_L2_CNTL3_BANK_SELECT_0(0) | + R600_VM_L2_CNTL3_BANK_SELECT_1(1) | + R600_VM_L2_CNTL3_CACHE_UPDATE_MODE(2)); + RADEON_WRITE(R600_VM_L2_CNTL3, vm_l2_cntl3); + + vm_c0 = R600_VM_ENABLE_CONTEXT | R600_VM_PAGE_TABLE_DEPTH_FLAT; + + RADEON_WRITE(R600_VM_CONTEXT0_CNTL, vm_c0); + + vm_c0 &= ~R600_VM_ENABLE_CONTEXT; + + /* disable all other contexts */ + for (i = 1; i < 8; i++) + RADEON_WRITE(R600_VM_CONTEXT0_CNTL + (i * 4), vm_c0); + + RADEON_WRITE(R600_VM_CONTEXT0_PAGE_TABLE_BASE_ADDR, dev_priv->gart_info.bus_addr >> 12); + RADEON_WRITE(R600_VM_CONTEXT0_PAGE_TABLE_START_ADDR, dev_priv->gart_vm_start >> 12); + RADEON_WRITE(R600_VM_CONTEXT0_PAGE_TABLE_END_ADDR, (dev_priv->gart_vm_start + dev_priv->gart_size - 1) >> 12); + + r600_vm_flush_gart_range(dev); +} + +static int r600_cp_init_microcode(drm_radeon_private_t *dev_priv) +{ + struct platform_device *pdev; + const char *chip_name; + size_t pfp_req_size, me_req_size; + char fw_name[30]; + int err; + + pdev = platform_device_register_simple("r600_cp", 0, NULL, 0); + err = IS_ERR(pdev); + if (err) { + printk(KERN_ERR "r600_cp: Failed to register firmware\n"); + return -EINVAL; + } + + switch (dev_priv->flags & RADEON_FAMILY_MASK) { + case CHIP_R600: chip_name = "R600"; break; + case CHIP_RV610: chip_name = "RV610"; break; + case CHIP_RV630: chip_name = "RV630"; break; + case CHIP_RV620: chip_name = "RV620"; break; + case CHIP_RV635: chip_name = "RV635"; break; + case CHIP_RV670: chip_name = "RV670"; break; + case CHIP_RS780: + case CHIP_RS880: chip_name = "RS780"; break; + case CHIP_RV770: chip_name = "RV770"; break; + case CHIP_RV730: + case CHIP_RV740: chip_name = "RV730"; break; + case CHIP_RV710: chip_name = "RV710"; break; + default: BUG(); + } + + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_RV770) { + pfp_req_size = R700_PFP_UCODE_SIZE * 4; + me_req_size = R700_PM4_UCODE_SIZE * 4; + } else { + pfp_req_size = PFP_UCODE_SIZE * 4; + me_req_size = PM4_UCODE_SIZE * 12; + } + + DRM_INFO("Loading %s CP Microcode\n", chip_name); + + snprintf(fw_name, sizeof(fw_name), "radeon/%s_pfp.bin", chip_name); + err = request_firmware(&dev_priv->pfp_fw, fw_name, &pdev->dev); + if (err) + goto out; + if (dev_priv->pfp_fw->size != pfp_req_size) { + printk(KERN_ERR + "r600_cp: Bogus length %zu in firmware \"%s\"\n", + dev_priv->pfp_fw->size, fw_name); + err = -EINVAL; + goto out; + } + + snprintf(fw_name, sizeof(fw_name), "radeon/%s_me.bin", chip_name); + err = request_firmware(&dev_priv->me_fw, fw_name, &pdev->dev); + if (err) + goto out; + if (dev_priv->me_fw->size != me_req_size) { + printk(KERN_ERR + "r600_cp: Bogus length %zu in firmware \"%s\"\n", + dev_priv->me_fw->size, fw_name); + err = -EINVAL; + } +out: + platform_device_unregister(pdev); + + if (err) { + if (err != -EINVAL) + printk(KERN_ERR + "r600_cp: Failed to load firmware \"%s\"\n", + fw_name); + release_firmware(dev_priv->pfp_fw); + dev_priv->pfp_fw = NULL; + release_firmware(dev_priv->me_fw); + dev_priv->me_fw = NULL; + } + return err; +} + +static void r600_cp_load_microcode(drm_radeon_private_t *dev_priv) +{ + const __be32 *fw_data; + int i; + + if (!dev_priv->me_fw || !dev_priv->pfp_fw) + return; + + r600_do_cp_stop(dev_priv); + + RADEON_WRITE(R600_CP_RB_CNTL, +#ifdef __BIG_ENDIAN + R600_BUF_SWAP_32BIT | +#endif + R600_RB_NO_UPDATE | + R600_RB_BLKSZ(15) | + R600_RB_BUFSZ(3)); + + RADEON_WRITE(R600_GRBM_SOFT_RESET, R600_SOFT_RESET_CP); + RADEON_READ(R600_GRBM_SOFT_RESET); + DRM_UDELAY(15000); + RADEON_WRITE(R600_GRBM_SOFT_RESET, 0); + + fw_data = (const __be32 *)dev_priv->me_fw->data; + RADEON_WRITE(R600_CP_ME_RAM_WADDR, 0); + for (i = 0; i < PM4_UCODE_SIZE * 3; i++) + RADEON_WRITE(R600_CP_ME_RAM_DATA, + be32_to_cpup(fw_data++)); + + fw_data = (const __be32 *)dev_priv->pfp_fw->data; + RADEON_WRITE(R600_CP_PFP_UCODE_ADDR, 0); + for (i = 0; i < PFP_UCODE_SIZE; i++) + RADEON_WRITE(R600_CP_PFP_UCODE_DATA, + be32_to_cpup(fw_data++)); + + RADEON_WRITE(R600_CP_PFP_UCODE_ADDR, 0); + RADEON_WRITE(R600_CP_ME_RAM_WADDR, 0); + RADEON_WRITE(R600_CP_ME_RAM_RADDR, 0); + +} + +static void r700_vm_init(struct drm_device *dev) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + /* initialise the VM to use the page table we constructed up there */ + u32 vm_c0, i; + u32 mc_vm_md_l1; + u32 vm_l2_cntl, vm_l2_cntl3; + /* okay set up the PCIE aperture type thingo */ + RADEON_WRITE(R700_MC_VM_SYSTEM_APERTURE_LOW_ADDR, dev_priv->gart_vm_start >> 12); + RADEON_WRITE(R700_MC_VM_SYSTEM_APERTURE_HIGH_ADDR, (dev_priv->gart_vm_start + dev_priv->gart_size - 1) >> 12); + RADEON_WRITE(R700_MC_VM_SYSTEM_APERTURE_DEFAULT_ADDR, 0); + + mc_vm_md_l1 = R700_ENABLE_L1_TLB | + R700_ENABLE_L1_FRAGMENT_PROCESSING | + R700_SYSTEM_ACCESS_MODE_IN_SYS | + R700_SYSTEM_APERTURE_UNMAPPED_ACCESS_PASS_THRU | + R700_EFFECTIVE_L1_TLB_SIZE(5) | + R700_EFFECTIVE_L1_QUEUE_SIZE(5); + + RADEON_WRITE(R700_MC_VM_MD_L1_TLB0_CNTL, mc_vm_md_l1); + RADEON_WRITE(R700_MC_VM_MD_L1_TLB1_CNTL, mc_vm_md_l1); + RADEON_WRITE(R700_MC_VM_MD_L1_TLB2_CNTL, mc_vm_md_l1); + RADEON_WRITE(R700_MC_VM_MB_L1_TLB0_CNTL, mc_vm_md_l1); + RADEON_WRITE(R700_MC_VM_MB_L1_TLB1_CNTL, mc_vm_md_l1); + RADEON_WRITE(R700_MC_VM_MB_L1_TLB2_CNTL, mc_vm_md_l1); + RADEON_WRITE(R700_MC_VM_MB_L1_TLB3_CNTL, mc_vm_md_l1); + + vm_l2_cntl = R600_VM_L2_CACHE_EN | R600_VM_L2_FRAG_PROC | R600_VM_ENABLE_PTE_CACHE_LRU_W; + vm_l2_cntl |= R700_VM_L2_CNTL_QUEUE_SIZE(7); + RADEON_WRITE(R600_VM_L2_CNTL, vm_l2_cntl); + + RADEON_WRITE(R600_VM_L2_CNTL2, 0); + vm_l2_cntl3 = R700_VM_L2_CNTL3_BANK_SELECT(0) | R700_VM_L2_CNTL3_CACHE_UPDATE_MODE(2); + RADEON_WRITE(R600_VM_L2_CNTL3, vm_l2_cntl3); + + vm_c0 = R600_VM_ENABLE_CONTEXT | R600_VM_PAGE_TABLE_DEPTH_FLAT; + + RADEON_WRITE(R600_VM_CONTEXT0_CNTL, vm_c0); + + vm_c0 &= ~R600_VM_ENABLE_CONTEXT; + + /* disable all other contexts */ + for (i = 1; i < 8; i++) + RADEON_WRITE(R600_VM_CONTEXT0_CNTL + (i * 4), vm_c0); + + RADEON_WRITE(R700_VM_CONTEXT0_PAGE_TABLE_BASE_ADDR, dev_priv->gart_info.bus_addr >> 12); + RADEON_WRITE(R700_VM_CONTEXT0_PAGE_TABLE_START_ADDR, dev_priv->gart_vm_start >> 12); + RADEON_WRITE(R700_VM_CONTEXT0_PAGE_TABLE_END_ADDR, (dev_priv->gart_vm_start + dev_priv->gart_size - 1) >> 12); + + r600_vm_flush_gart_range(dev); +} + +static void r700_cp_load_microcode(drm_radeon_private_t *dev_priv) +{ + const __be32 *fw_data; + int i; + + if (!dev_priv->me_fw || !dev_priv->pfp_fw) + return; + + r600_do_cp_stop(dev_priv); + + RADEON_WRITE(R600_CP_RB_CNTL, +#ifdef __BIG_ENDIAN + R600_BUF_SWAP_32BIT | +#endif + R600_RB_NO_UPDATE | + R600_RB_BLKSZ(15) | + R600_RB_BUFSZ(3)); + + RADEON_WRITE(R600_GRBM_SOFT_RESET, R600_SOFT_RESET_CP); + RADEON_READ(R600_GRBM_SOFT_RESET); + DRM_UDELAY(15000); + RADEON_WRITE(R600_GRBM_SOFT_RESET, 0); + + fw_data = (const __be32 *)dev_priv->pfp_fw->data; + RADEON_WRITE(R600_CP_PFP_UCODE_ADDR, 0); + for (i = 0; i < R700_PFP_UCODE_SIZE; i++) + RADEON_WRITE(R600_CP_PFP_UCODE_DATA, be32_to_cpup(fw_data++)); + RADEON_WRITE(R600_CP_PFP_UCODE_ADDR, 0); + + fw_data = (const __be32 *)dev_priv->me_fw->data; + RADEON_WRITE(R600_CP_ME_RAM_WADDR, 0); + for (i = 0; i < R700_PM4_UCODE_SIZE; i++) + RADEON_WRITE(R600_CP_ME_RAM_DATA, be32_to_cpup(fw_data++)); + RADEON_WRITE(R600_CP_ME_RAM_WADDR, 0); + + RADEON_WRITE(R600_CP_PFP_UCODE_ADDR, 0); + RADEON_WRITE(R600_CP_ME_RAM_WADDR, 0); + RADEON_WRITE(R600_CP_ME_RAM_RADDR, 0); + +} + +static void r600_test_writeback(drm_radeon_private_t *dev_priv) +{ + u32 tmp; + + /* Start with assuming that writeback doesn't work */ + dev_priv->writeback_works = 0; + + /* Writeback doesn't seem to work everywhere, test it here and possibly + * enable it if it appears to work + */ + radeon_write_ring_rptr(dev_priv, R600_SCRATCHOFF(1), 0); + + RADEON_WRITE(R600_SCRATCH_REG1, 0xdeadbeef); + + for (tmp = 0; tmp < dev_priv->usec_timeout; tmp++) { + u32 val; + + val = radeon_read_ring_rptr(dev_priv, R600_SCRATCHOFF(1)); + if (val == 0xdeadbeef) + break; + DRM_UDELAY(1); + } + + if (tmp < dev_priv->usec_timeout) { + dev_priv->writeback_works = 1; + DRM_INFO("writeback test succeeded in %d usecs\n", tmp); + } else { + dev_priv->writeback_works = 0; + DRM_INFO("writeback test failed\n"); + } + if (radeon_no_wb == 1) { + dev_priv->writeback_works = 0; + DRM_INFO("writeback forced off\n"); + } + + if (!dev_priv->writeback_works) { + /* Disable writeback to avoid unnecessary bus master transfer */ + RADEON_WRITE(R600_CP_RB_CNTL, +#ifdef __BIG_ENDIAN + R600_BUF_SWAP_32BIT | +#endif + RADEON_READ(R600_CP_RB_CNTL) | + R600_RB_NO_UPDATE); + RADEON_WRITE(R600_SCRATCH_UMSK, 0); + } +} + +int r600_do_engine_reset(struct drm_device *dev) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + u32 cp_ptr, cp_me_cntl, cp_rb_cntl; + + DRM_INFO("Resetting GPU\n"); + + cp_ptr = RADEON_READ(R600_CP_RB_WPTR); + cp_me_cntl = RADEON_READ(R600_CP_ME_CNTL); + RADEON_WRITE(R600_CP_ME_CNTL, R600_CP_ME_HALT); + + RADEON_WRITE(R600_GRBM_SOFT_RESET, 0x7fff); + RADEON_READ(R600_GRBM_SOFT_RESET); + DRM_UDELAY(50); + RADEON_WRITE(R600_GRBM_SOFT_RESET, 0); + RADEON_READ(R600_GRBM_SOFT_RESET); + + RADEON_WRITE(R600_CP_RB_WPTR_DELAY, 0); + cp_rb_cntl = RADEON_READ(R600_CP_RB_CNTL); + RADEON_WRITE(R600_CP_RB_CNTL, +#ifdef __BIG_ENDIAN + R600_BUF_SWAP_32BIT | +#endif + R600_RB_RPTR_WR_ENA); + + RADEON_WRITE(R600_CP_RB_RPTR_WR, cp_ptr); + RADEON_WRITE(R600_CP_RB_WPTR, cp_ptr); + RADEON_WRITE(R600_CP_RB_CNTL, cp_rb_cntl); + RADEON_WRITE(R600_CP_ME_CNTL, cp_me_cntl); + + /* Reset the CP ring */ + r600_do_cp_reset(dev_priv); + + /* The CP is no longer running after an engine reset */ + dev_priv->cp_running = 0; + + /* Reset any pending vertex, indirect buffers */ + radeon_freelist_reset(dev); + + return 0; + +} + +static u32 r600_get_tile_pipe_to_backend_map(u32 num_tile_pipes, + u32 num_backends, + u32 backend_disable_mask) +{ + u32 backend_map = 0; + u32 enabled_backends_mask; + u32 enabled_backends_count; + u32 cur_pipe; + u32 swizzle_pipe[R6XX_MAX_PIPES]; + u32 cur_backend; + u32 i; + + if (num_tile_pipes > R6XX_MAX_PIPES) + num_tile_pipes = R6XX_MAX_PIPES; + if (num_tile_pipes < 1) + num_tile_pipes = 1; + if (num_backends > R6XX_MAX_BACKENDS) + num_backends = R6XX_MAX_BACKENDS; + if (num_backends < 1) + num_backends = 1; + + enabled_backends_mask = 0; + enabled_backends_count = 0; + for (i = 0; i < R6XX_MAX_BACKENDS; ++i) { + if (((backend_disable_mask >> i) & 1) == 0) { + enabled_backends_mask |= (1 << i); + ++enabled_backends_count; + } + if (enabled_backends_count == num_backends) + break; + } + + if (enabled_backends_count == 0) { + enabled_backends_mask = 1; + enabled_backends_count = 1; + } + + if (enabled_backends_count != num_backends) + num_backends = enabled_backends_count; + + memset((uint8_t *)&swizzle_pipe[0], 0, sizeof(u32) * R6XX_MAX_PIPES); + switch (num_tile_pipes) { + case 1: + swizzle_pipe[0] = 0; + break; + case 2: + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 1; + break; + case 3: + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 1; + swizzle_pipe[2] = 2; + break; + case 4: + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 1; + swizzle_pipe[2] = 2; + swizzle_pipe[3] = 3; + break; + case 5: + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 1; + swizzle_pipe[2] = 2; + swizzle_pipe[3] = 3; + swizzle_pipe[4] = 4; + break; + case 6: + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 2; + swizzle_pipe[2] = 4; + swizzle_pipe[3] = 5; + swizzle_pipe[4] = 1; + swizzle_pipe[5] = 3; + break; + case 7: + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 2; + swizzle_pipe[2] = 4; + swizzle_pipe[3] = 6; + swizzle_pipe[4] = 1; + swizzle_pipe[5] = 3; + swizzle_pipe[6] = 5; + break; + case 8: + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 2; + swizzle_pipe[2] = 4; + swizzle_pipe[3] = 6; + swizzle_pipe[4] = 1; + swizzle_pipe[5] = 3; + swizzle_pipe[6] = 5; + swizzle_pipe[7] = 7; + break; + } + + cur_backend = 0; + for (cur_pipe = 0; cur_pipe < num_tile_pipes; ++cur_pipe) { + while (((1 << cur_backend) & enabled_backends_mask) == 0) + cur_backend = (cur_backend + 1) % R6XX_MAX_BACKENDS; + + backend_map |= (u32)(((cur_backend & 3) << (swizzle_pipe[cur_pipe] * 2))); + + cur_backend = (cur_backend + 1) % R6XX_MAX_BACKENDS; + } + + return backend_map; +} + +static int r600_count_pipe_bits(uint32_t val) +{ + int i, ret = 0; + for (i = 0; i < 32; i++) { + ret += val & 1; + val >>= 1; + } + return ret; +} + +static void r600_gfx_init(struct drm_device *dev, + drm_radeon_private_t *dev_priv) +{ + int i, j, num_qd_pipes; + u32 sx_debug_1; + u32 tc_cntl; + u32 arb_pop; + u32 num_gs_verts_per_thread; + u32 vgt_gs_per_es; + u32 gs_prim_buffer_depth = 0; + u32 sq_ms_fifo_sizes; + u32 sq_config; + u32 sq_gpr_resource_mgmt_1 = 0; + u32 sq_gpr_resource_mgmt_2 = 0; + u32 sq_thread_resource_mgmt = 0; + u32 sq_stack_resource_mgmt_1 = 0; + u32 sq_stack_resource_mgmt_2 = 0; + u32 hdp_host_path_cntl; + u32 backend_map; + u32 gb_tiling_config = 0; + u32 cc_rb_backend_disable; + u32 cc_gc_shader_pipe_config; + u32 ramcfg; + + /* setup chip specs */ + switch (dev_priv->flags & RADEON_FAMILY_MASK) { + case CHIP_R600: + dev_priv->r600_max_pipes = 4; + dev_priv->r600_max_tile_pipes = 8; + dev_priv->r600_max_simds = 4; + dev_priv->r600_max_backends = 4; + dev_priv->r600_max_gprs = 256; + dev_priv->r600_max_threads = 192; + dev_priv->r600_max_stack_entries = 256; + dev_priv->r600_max_hw_contexts = 8; + dev_priv->r600_max_gs_threads = 16; + dev_priv->r600_sx_max_export_size = 128; + dev_priv->r600_sx_max_export_pos_size = 16; + dev_priv->r600_sx_max_export_smx_size = 128; + dev_priv->r600_sq_num_cf_insts = 2; + break; + case CHIP_RV630: + case CHIP_RV635: + dev_priv->r600_max_pipes = 2; + dev_priv->r600_max_tile_pipes = 2; + dev_priv->r600_max_simds = 3; + dev_priv->r600_max_backends = 1; + dev_priv->r600_max_gprs = 128; + dev_priv->r600_max_threads = 192; + dev_priv->r600_max_stack_entries = 128; + dev_priv->r600_max_hw_contexts = 8; + dev_priv->r600_max_gs_threads = 4; + dev_priv->r600_sx_max_export_size = 128; + dev_priv->r600_sx_max_export_pos_size = 16; + dev_priv->r600_sx_max_export_smx_size = 128; + dev_priv->r600_sq_num_cf_insts = 2; + break; + case CHIP_RV610: + case CHIP_RS780: + case CHIP_RS880: + case CHIP_RV620: + dev_priv->r600_max_pipes = 1; + dev_priv->r600_max_tile_pipes = 1; + dev_priv->r600_max_simds = 2; + dev_priv->r600_max_backends = 1; + dev_priv->r600_max_gprs = 128; + dev_priv->r600_max_threads = 192; + dev_priv->r600_max_stack_entries = 128; + dev_priv->r600_max_hw_contexts = 4; + dev_priv->r600_max_gs_threads = 4; + dev_priv->r600_sx_max_export_size = 128; + dev_priv->r600_sx_max_export_pos_size = 16; + dev_priv->r600_sx_max_export_smx_size = 128; + dev_priv->r600_sq_num_cf_insts = 1; + break; + case CHIP_RV670: + dev_priv->r600_max_pipes = 4; + dev_priv->r600_max_tile_pipes = 4; + dev_priv->r600_max_simds = 4; + dev_priv->r600_max_backends = 4; + dev_priv->r600_max_gprs = 192; + dev_priv->r600_max_threads = 192; + dev_priv->r600_max_stack_entries = 256; + dev_priv->r600_max_hw_contexts = 8; + dev_priv->r600_max_gs_threads = 16; + dev_priv->r600_sx_max_export_size = 128; + dev_priv->r600_sx_max_export_pos_size = 16; + dev_priv->r600_sx_max_export_smx_size = 128; + dev_priv->r600_sq_num_cf_insts = 2; + break; + default: + break; + } + + /* Initialize HDP */ + j = 0; + for (i = 0; i < 32; i++) { + RADEON_WRITE((0x2c14 + j), 0x00000000); + RADEON_WRITE((0x2c18 + j), 0x00000000); + RADEON_WRITE((0x2c1c + j), 0x00000000); + RADEON_WRITE((0x2c20 + j), 0x00000000); + RADEON_WRITE((0x2c24 + j), 0x00000000); + j += 0x18; + } + + RADEON_WRITE(R600_GRBM_CNTL, R600_GRBM_READ_TIMEOUT(0xff)); + + /* setup tiling, simd, pipe config */ + ramcfg = RADEON_READ(R600_RAMCFG); + + switch (dev_priv->r600_max_tile_pipes) { + case 1: + gb_tiling_config |= R600_PIPE_TILING(0); + break; + case 2: + gb_tiling_config |= R600_PIPE_TILING(1); + break; + case 4: + gb_tiling_config |= R600_PIPE_TILING(2); + break; + case 8: + gb_tiling_config |= R600_PIPE_TILING(3); + break; + default: + break; + } + + gb_tiling_config |= R600_BANK_TILING((ramcfg >> R600_NOOFBANK_SHIFT) & R600_NOOFBANK_MASK); + + gb_tiling_config |= R600_GROUP_SIZE(0); + + if (((ramcfg >> R600_NOOFROWS_SHIFT) & R600_NOOFROWS_MASK) > 3) { + gb_tiling_config |= R600_ROW_TILING(3); + gb_tiling_config |= R600_SAMPLE_SPLIT(3); + } else { + gb_tiling_config |= + R600_ROW_TILING(((ramcfg >> R600_NOOFROWS_SHIFT) & R600_NOOFROWS_MASK)); + gb_tiling_config |= + R600_SAMPLE_SPLIT(((ramcfg >> R600_NOOFROWS_SHIFT) & R600_NOOFROWS_MASK)); + } + + gb_tiling_config |= R600_BANK_SWAPS(1); + + cc_rb_backend_disable = RADEON_READ(R600_CC_RB_BACKEND_DISABLE) & 0x00ff0000; + cc_rb_backend_disable |= + R600_BACKEND_DISABLE((R6XX_MAX_BACKENDS_MASK << dev_priv->r600_max_backends) & R6XX_MAX_BACKENDS_MASK); + + cc_gc_shader_pipe_config = RADEON_READ(R600_CC_GC_SHADER_PIPE_CONFIG) & 0xffffff00; + cc_gc_shader_pipe_config |= + R600_INACTIVE_QD_PIPES((R6XX_MAX_PIPES_MASK << dev_priv->r600_max_pipes) & R6XX_MAX_PIPES_MASK); + cc_gc_shader_pipe_config |= + R600_INACTIVE_SIMDS((R6XX_MAX_SIMDS_MASK << dev_priv->r600_max_simds) & R6XX_MAX_SIMDS_MASK); + + backend_map = r600_get_tile_pipe_to_backend_map(dev_priv->r600_max_tile_pipes, + (R6XX_MAX_BACKENDS - + r600_count_pipe_bits((cc_rb_backend_disable & + R6XX_MAX_BACKENDS_MASK) >> 16)), + (cc_rb_backend_disable >> 16)); + gb_tiling_config |= R600_BACKEND_MAP(backend_map); + + RADEON_WRITE(R600_GB_TILING_CONFIG, gb_tiling_config); + RADEON_WRITE(R600_DCP_TILING_CONFIG, (gb_tiling_config & 0xffff)); + RADEON_WRITE(R600_HDP_TILING_CONFIG, (gb_tiling_config & 0xffff)); + if (gb_tiling_config & 0xc0) { + dev_priv->r600_group_size = 512; + } else { + dev_priv->r600_group_size = 256; + } + dev_priv->r600_npipes = 1 << ((gb_tiling_config >> 1) & 0x7); + if (gb_tiling_config & 0x30) { + dev_priv->r600_nbanks = 8; + } else { + dev_priv->r600_nbanks = 4; + } + + RADEON_WRITE(R600_CC_RB_BACKEND_DISABLE, cc_rb_backend_disable); + RADEON_WRITE(R600_CC_GC_SHADER_PIPE_CONFIG, cc_gc_shader_pipe_config); + RADEON_WRITE(R600_GC_USER_SHADER_PIPE_CONFIG, cc_gc_shader_pipe_config); + + num_qd_pipes = + R6XX_MAX_PIPES - r600_count_pipe_bits((cc_gc_shader_pipe_config & R600_INACTIVE_QD_PIPES_MASK) >> 8); + RADEON_WRITE(R600_VGT_OUT_DEALLOC_CNTL, (num_qd_pipes * 4) & R600_DEALLOC_DIST_MASK); + RADEON_WRITE(R600_VGT_VERTEX_REUSE_BLOCK_CNTL, ((num_qd_pipes * 4) - 2) & R600_VTX_REUSE_DEPTH_MASK); + + /* set HW defaults for 3D engine */ + RADEON_WRITE(R600_CP_QUEUE_THRESHOLDS, (R600_ROQ_IB1_START(0x16) | + R600_ROQ_IB2_START(0x2b))); + + RADEON_WRITE(R600_CP_MEQ_THRESHOLDS, (R600_MEQ_END(0x40) | + R600_ROQ_END(0x40))); + + RADEON_WRITE(R600_TA_CNTL_AUX, (R600_DISABLE_CUBE_ANISO | + R600_SYNC_GRADIENT | + R600_SYNC_WALKER | + R600_SYNC_ALIGNER)); + + if ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV670) + RADEON_WRITE(R600_ARB_GDEC_RD_CNTL, 0x00000021); + + sx_debug_1 = RADEON_READ(R600_SX_DEBUG_1); + sx_debug_1 |= R600_SMX_EVENT_RELEASE; + if (((dev_priv->flags & RADEON_FAMILY_MASK) > CHIP_R600)) + sx_debug_1 |= R600_ENABLE_NEW_SMX_ADDRESS; + RADEON_WRITE(R600_SX_DEBUG_1, sx_debug_1); + + if (((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_R600) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV630) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV610) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV620) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS780) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS880)) + RADEON_WRITE(R600_DB_DEBUG, R600_PREZ_MUST_WAIT_FOR_POSTZ_DONE); + else + RADEON_WRITE(R600_DB_DEBUG, 0); + + RADEON_WRITE(R600_DB_WATERMARKS, (R600_DEPTH_FREE(4) | + R600_DEPTH_FLUSH(16) | + R600_DEPTH_PENDING_FREE(4) | + R600_DEPTH_CACHELINE_FREE(16))); + RADEON_WRITE(R600_PA_SC_MULTI_CHIP_CNTL, 0); + RADEON_WRITE(R600_VGT_NUM_INSTANCES, 0); + + RADEON_WRITE(R600_SPI_CONFIG_CNTL, R600_GPR_WRITE_PRIORITY(0)); + RADEON_WRITE(R600_SPI_CONFIG_CNTL_1, R600_VTX_DONE_DELAY(0)); + + sq_ms_fifo_sizes = RADEON_READ(R600_SQ_MS_FIFO_SIZES); + if (((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV610) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV620) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS780) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS880)) { + sq_ms_fifo_sizes = (R600_CACHE_FIFO_SIZE(0xa) | + R600_FETCH_FIFO_HIWATER(0xa) | + R600_DONE_FIFO_HIWATER(0xe0) | + R600_ALU_UPDATE_FIFO_HIWATER(0x8)); + } else if (((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_R600) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV630)) { + sq_ms_fifo_sizes &= ~R600_DONE_FIFO_HIWATER(0xff); + sq_ms_fifo_sizes |= R600_DONE_FIFO_HIWATER(0x4); + } + RADEON_WRITE(R600_SQ_MS_FIFO_SIZES, sq_ms_fifo_sizes); + + /* SQ_CONFIG, SQ_GPR_RESOURCE_MGMT, SQ_THREAD_RESOURCE_MGMT, SQ_STACK_RESOURCE_MGMT + * should be adjusted as needed by the 2D/3D drivers. This just sets default values + */ + sq_config = RADEON_READ(R600_SQ_CONFIG); + sq_config &= ~(R600_PS_PRIO(3) | + R600_VS_PRIO(3) | + R600_GS_PRIO(3) | + R600_ES_PRIO(3)); + sq_config |= (R600_DX9_CONSTS | + R600_VC_ENABLE | + R600_PS_PRIO(0) | + R600_VS_PRIO(1) | + R600_GS_PRIO(2) | + R600_ES_PRIO(3)); + + if ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_R600) { + sq_gpr_resource_mgmt_1 = (R600_NUM_PS_GPRS(124) | + R600_NUM_VS_GPRS(124) | + R600_NUM_CLAUSE_TEMP_GPRS(4)); + sq_gpr_resource_mgmt_2 = (R600_NUM_GS_GPRS(0) | + R600_NUM_ES_GPRS(0)); + sq_thread_resource_mgmt = (R600_NUM_PS_THREADS(136) | + R600_NUM_VS_THREADS(48) | + R600_NUM_GS_THREADS(4) | + R600_NUM_ES_THREADS(4)); + sq_stack_resource_mgmt_1 = (R600_NUM_PS_STACK_ENTRIES(128) | + R600_NUM_VS_STACK_ENTRIES(128)); + sq_stack_resource_mgmt_2 = (R600_NUM_GS_STACK_ENTRIES(0) | + R600_NUM_ES_STACK_ENTRIES(0)); + } else if (((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV610) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV620) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS780) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS880)) { + /* no vertex cache */ + sq_config &= ~R600_VC_ENABLE; + + sq_gpr_resource_mgmt_1 = (R600_NUM_PS_GPRS(44) | + R600_NUM_VS_GPRS(44) | + R600_NUM_CLAUSE_TEMP_GPRS(2)); + sq_gpr_resource_mgmt_2 = (R600_NUM_GS_GPRS(17) | + R600_NUM_ES_GPRS(17)); + sq_thread_resource_mgmt = (R600_NUM_PS_THREADS(79) | + R600_NUM_VS_THREADS(78) | + R600_NUM_GS_THREADS(4) | + R600_NUM_ES_THREADS(31)); + sq_stack_resource_mgmt_1 = (R600_NUM_PS_STACK_ENTRIES(40) | + R600_NUM_VS_STACK_ENTRIES(40)); + sq_stack_resource_mgmt_2 = (R600_NUM_GS_STACK_ENTRIES(32) | + R600_NUM_ES_STACK_ENTRIES(16)); + } else if (((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV630) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV635)) { + sq_gpr_resource_mgmt_1 = (R600_NUM_PS_GPRS(44) | + R600_NUM_VS_GPRS(44) | + R600_NUM_CLAUSE_TEMP_GPRS(2)); + sq_gpr_resource_mgmt_2 = (R600_NUM_GS_GPRS(18) | + R600_NUM_ES_GPRS(18)); + sq_thread_resource_mgmt = (R600_NUM_PS_THREADS(79) | + R600_NUM_VS_THREADS(78) | + R600_NUM_GS_THREADS(4) | + R600_NUM_ES_THREADS(31)); + sq_stack_resource_mgmt_1 = (R600_NUM_PS_STACK_ENTRIES(40) | + R600_NUM_VS_STACK_ENTRIES(40)); + sq_stack_resource_mgmt_2 = (R600_NUM_GS_STACK_ENTRIES(32) | + R600_NUM_ES_STACK_ENTRIES(16)); + } else if ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV670) { + sq_gpr_resource_mgmt_1 = (R600_NUM_PS_GPRS(44) | + R600_NUM_VS_GPRS(44) | + R600_NUM_CLAUSE_TEMP_GPRS(2)); + sq_gpr_resource_mgmt_2 = (R600_NUM_GS_GPRS(17) | + R600_NUM_ES_GPRS(17)); + sq_thread_resource_mgmt = (R600_NUM_PS_THREADS(79) | + R600_NUM_VS_THREADS(78) | + R600_NUM_GS_THREADS(4) | + R600_NUM_ES_THREADS(31)); + sq_stack_resource_mgmt_1 = (R600_NUM_PS_STACK_ENTRIES(64) | + R600_NUM_VS_STACK_ENTRIES(64)); + sq_stack_resource_mgmt_2 = (R600_NUM_GS_STACK_ENTRIES(64) | + R600_NUM_ES_STACK_ENTRIES(64)); + } + + RADEON_WRITE(R600_SQ_CONFIG, sq_config); + RADEON_WRITE(R600_SQ_GPR_RESOURCE_MGMT_1, sq_gpr_resource_mgmt_1); + RADEON_WRITE(R600_SQ_GPR_RESOURCE_MGMT_2, sq_gpr_resource_mgmt_2); + RADEON_WRITE(R600_SQ_THREAD_RESOURCE_MGMT, sq_thread_resource_mgmt); + RADEON_WRITE(R600_SQ_STACK_RESOURCE_MGMT_1, sq_stack_resource_mgmt_1); + RADEON_WRITE(R600_SQ_STACK_RESOURCE_MGMT_2, sq_stack_resource_mgmt_2); + + if (((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV610) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV620) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS780) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS880)) + RADEON_WRITE(R600_VGT_CACHE_INVALIDATION, R600_CACHE_INVALIDATION(R600_TC_ONLY)); + else + RADEON_WRITE(R600_VGT_CACHE_INVALIDATION, R600_CACHE_INVALIDATION(R600_VC_AND_TC)); + + RADEON_WRITE(R600_PA_SC_AA_SAMPLE_LOCS_2S, (R600_S0_X(0xc) | + R600_S0_Y(0x4) | + R600_S1_X(0x4) | + R600_S1_Y(0xc))); + RADEON_WRITE(R600_PA_SC_AA_SAMPLE_LOCS_4S, (R600_S0_X(0xe) | + R600_S0_Y(0xe) | + R600_S1_X(0x2) | + R600_S1_Y(0x2) | + R600_S2_X(0xa) | + R600_S2_Y(0x6) | + R600_S3_X(0x6) | + R600_S3_Y(0xa))); + RADEON_WRITE(R600_PA_SC_AA_SAMPLE_LOCS_8S_WD0, (R600_S0_X(0xe) | + R600_S0_Y(0xb) | + R600_S1_X(0x4) | + R600_S1_Y(0xc) | + R600_S2_X(0x1) | + R600_S2_Y(0x6) | + R600_S3_X(0xa) | + R600_S3_Y(0xe))); + RADEON_WRITE(R600_PA_SC_AA_SAMPLE_LOCS_8S_WD1, (R600_S4_X(0x6) | + R600_S4_Y(0x1) | + R600_S5_X(0x0) | + R600_S5_Y(0x0) | + R600_S6_X(0xb) | + R600_S6_Y(0x4) | + R600_S7_X(0x7) | + R600_S7_Y(0x8))); + + + switch (dev_priv->flags & RADEON_FAMILY_MASK) { + case CHIP_R600: + case CHIP_RV630: + case CHIP_RV635: + gs_prim_buffer_depth = 0; + break; + case CHIP_RV610: + case CHIP_RS780: + case CHIP_RS880: + case CHIP_RV620: + gs_prim_buffer_depth = 32; + break; + case CHIP_RV670: + gs_prim_buffer_depth = 128; + break; + default: + break; + } + + num_gs_verts_per_thread = dev_priv->r600_max_pipes * 16; + vgt_gs_per_es = gs_prim_buffer_depth + num_gs_verts_per_thread; + /* Max value for this is 256 */ + if (vgt_gs_per_es > 256) + vgt_gs_per_es = 256; + + RADEON_WRITE(R600_VGT_ES_PER_GS, 128); + RADEON_WRITE(R600_VGT_GS_PER_ES, vgt_gs_per_es); + RADEON_WRITE(R600_VGT_GS_PER_VS, 2); + RADEON_WRITE(R600_VGT_GS_VERTEX_REUSE, 16); + + /* more default values. 2D/3D driver should adjust as needed */ + RADEON_WRITE(R600_PA_SC_LINE_STIPPLE_STATE, 0); + RADEON_WRITE(R600_VGT_STRMOUT_EN, 0); + RADEON_WRITE(R600_SX_MISC, 0); + RADEON_WRITE(R600_PA_SC_MODE_CNTL, 0); + RADEON_WRITE(R600_PA_SC_AA_CONFIG, 0); + RADEON_WRITE(R600_PA_SC_LINE_STIPPLE, 0); + RADEON_WRITE(R600_SPI_INPUT_Z, 0); + RADEON_WRITE(R600_SPI_PS_IN_CONTROL_0, R600_NUM_INTERP(2)); + RADEON_WRITE(R600_CB_COLOR7_FRAG, 0); + + /* clear render buffer base addresses */ + RADEON_WRITE(R600_CB_COLOR0_BASE, 0); + RADEON_WRITE(R600_CB_COLOR1_BASE, 0); + RADEON_WRITE(R600_CB_COLOR2_BASE, 0); + RADEON_WRITE(R600_CB_COLOR3_BASE, 0); + RADEON_WRITE(R600_CB_COLOR4_BASE, 0); + RADEON_WRITE(R600_CB_COLOR5_BASE, 0); + RADEON_WRITE(R600_CB_COLOR6_BASE, 0); + RADEON_WRITE(R600_CB_COLOR7_BASE, 0); + + switch (dev_priv->flags & RADEON_FAMILY_MASK) { + case CHIP_RV610: + case CHIP_RS780: + case CHIP_RS880: + case CHIP_RV620: + tc_cntl = R600_TC_L2_SIZE(8); + break; + case CHIP_RV630: + case CHIP_RV635: + tc_cntl = R600_TC_L2_SIZE(4); + break; + case CHIP_R600: + tc_cntl = R600_TC_L2_SIZE(0) | R600_L2_DISABLE_LATE_HIT; + break; + default: + tc_cntl = R600_TC_L2_SIZE(0); + break; + } + + RADEON_WRITE(R600_TC_CNTL, tc_cntl); + + hdp_host_path_cntl = RADEON_READ(R600_HDP_HOST_PATH_CNTL); + RADEON_WRITE(R600_HDP_HOST_PATH_CNTL, hdp_host_path_cntl); + + arb_pop = RADEON_READ(R600_ARB_POP); + arb_pop |= R600_ENABLE_TC128; + RADEON_WRITE(R600_ARB_POP, arb_pop); + + RADEON_WRITE(R600_PA_SC_MULTI_CHIP_CNTL, 0); + RADEON_WRITE(R600_PA_CL_ENHANCE, (R600_CLIP_VTX_REORDER_ENA | + R600_NUM_CLIP_SEQ(3))); + RADEON_WRITE(R600_PA_SC_ENHANCE, R600_FORCE_EOV_MAX_CLK_CNT(4095)); + +} + +static u32 r700_get_tile_pipe_to_backend_map(drm_radeon_private_t *dev_priv, + u32 num_tile_pipes, + u32 num_backends, + u32 backend_disable_mask) +{ + u32 backend_map = 0; + u32 enabled_backends_mask; + u32 enabled_backends_count; + u32 cur_pipe; + u32 swizzle_pipe[R7XX_MAX_PIPES]; + u32 cur_backend; + u32 i; + bool force_no_swizzle; + + if (num_tile_pipes > R7XX_MAX_PIPES) + num_tile_pipes = R7XX_MAX_PIPES; + if (num_tile_pipes < 1) + num_tile_pipes = 1; + if (num_backends > R7XX_MAX_BACKENDS) + num_backends = R7XX_MAX_BACKENDS; + if (num_backends < 1) + num_backends = 1; + + enabled_backends_mask = 0; + enabled_backends_count = 0; + for (i = 0; i < R7XX_MAX_BACKENDS; ++i) { + if (((backend_disable_mask >> i) & 1) == 0) { + enabled_backends_mask |= (1 << i); + ++enabled_backends_count; + } + if (enabled_backends_count == num_backends) + break; + } + + if (enabled_backends_count == 0) { + enabled_backends_mask = 1; + enabled_backends_count = 1; + } + + if (enabled_backends_count != num_backends) + num_backends = enabled_backends_count; + + switch (dev_priv->flags & RADEON_FAMILY_MASK) { + case CHIP_RV770: + case CHIP_RV730: + force_no_swizzle = false; + break; + case CHIP_RV710: + case CHIP_RV740: + default: + force_no_swizzle = true; + break; + } + + memset((uint8_t *)&swizzle_pipe[0], 0, sizeof(u32) * R7XX_MAX_PIPES); + switch (num_tile_pipes) { + case 1: + swizzle_pipe[0] = 0; + break; + case 2: + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 1; + break; + case 3: + if (force_no_swizzle) { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 1; + swizzle_pipe[2] = 2; + } else { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 2; + swizzle_pipe[2] = 1; + } + break; + case 4: + if (force_no_swizzle) { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 1; + swizzle_pipe[2] = 2; + swizzle_pipe[3] = 3; + } else { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 2; + swizzle_pipe[2] = 3; + swizzle_pipe[3] = 1; + } + break; + case 5: + if (force_no_swizzle) { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 1; + swizzle_pipe[2] = 2; + swizzle_pipe[3] = 3; + swizzle_pipe[4] = 4; + } else { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 2; + swizzle_pipe[2] = 4; + swizzle_pipe[3] = 1; + swizzle_pipe[4] = 3; + } + break; + case 6: + if (force_no_swizzle) { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 1; + swizzle_pipe[2] = 2; + swizzle_pipe[3] = 3; + swizzle_pipe[4] = 4; + swizzle_pipe[5] = 5; + } else { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 2; + swizzle_pipe[2] = 4; + swizzle_pipe[3] = 5; + swizzle_pipe[4] = 3; + swizzle_pipe[5] = 1; + } + break; + case 7: + if (force_no_swizzle) { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 1; + swizzle_pipe[2] = 2; + swizzle_pipe[3] = 3; + swizzle_pipe[4] = 4; + swizzle_pipe[5] = 5; + swizzle_pipe[6] = 6; + } else { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 2; + swizzle_pipe[2] = 4; + swizzle_pipe[3] = 6; + swizzle_pipe[4] = 3; + swizzle_pipe[5] = 1; + swizzle_pipe[6] = 5; + } + break; + case 8: + if (force_no_swizzle) { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 1; + swizzle_pipe[2] = 2; + swizzle_pipe[3] = 3; + swizzle_pipe[4] = 4; + swizzle_pipe[5] = 5; + swizzle_pipe[6] = 6; + swizzle_pipe[7] = 7; + } else { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 2; + swizzle_pipe[2] = 4; + swizzle_pipe[3] = 6; + swizzle_pipe[4] = 3; + swizzle_pipe[5] = 1; + swizzle_pipe[6] = 7; + swizzle_pipe[7] = 5; + } + break; + } + + cur_backend = 0; + for (cur_pipe = 0; cur_pipe < num_tile_pipes; ++cur_pipe) { + while (((1 << cur_backend) & enabled_backends_mask) == 0) + cur_backend = (cur_backend + 1) % R7XX_MAX_BACKENDS; + + backend_map |= (u32)(((cur_backend & 3) << (swizzle_pipe[cur_pipe] * 2))); + + cur_backend = (cur_backend + 1) % R7XX_MAX_BACKENDS; + } + + return backend_map; +} + +static void r700_gfx_init(struct drm_device *dev, + drm_radeon_private_t *dev_priv) +{ + int i, j, num_qd_pipes; + u32 ta_aux_cntl; + u32 sx_debug_1; + u32 smx_dc_ctl0; + u32 db_debug3; + u32 num_gs_verts_per_thread; + u32 vgt_gs_per_es; + u32 gs_prim_buffer_depth = 0; + u32 sq_ms_fifo_sizes; + u32 sq_config; + u32 sq_thread_resource_mgmt; + u32 hdp_host_path_cntl; + u32 sq_dyn_gpr_size_simd_ab_0; + u32 backend_map; + u32 gb_tiling_config = 0; + u32 cc_rb_backend_disable; + u32 cc_gc_shader_pipe_config; + u32 mc_arb_ramcfg; + u32 db_debug4; + + /* setup chip specs */ + switch (dev_priv->flags & RADEON_FAMILY_MASK) { + case CHIP_RV770: + dev_priv->r600_max_pipes = 4; + dev_priv->r600_max_tile_pipes = 8; + dev_priv->r600_max_simds = 10; + dev_priv->r600_max_backends = 4; + dev_priv->r600_max_gprs = 256; + dev_priv->r600_max_threads = 248; + dev_priv->r600_max_stack_entries = 512; + dev_priv->r600_max_hw_contexts = 8; + dev_priv->r600_max_gs_threads = 16 * 2; + dev_priv->r600_sx_max_export_size = 128; + dev_priv->r600_sx_max_export_pos_size = 16; + dev_priv->r600_sx_max_export_smx_size = 112; + dev_priv->r600_sq_num_cf_insts = 2; + + dev_priv->r700_sx_num_of_sets = 7; + dev_priv->r700_sc_prim_fifo_size = 0xF9; + dev_priv->r700_sc_hiz_tile_fifo_size = 0x30; + dev_priv->r700_sc_earlyz_tile_fifo_fize = 0x130; + break; + case CHIP_RV730: + dev_priv->r600_max_pipes = 2; + dev_priv->r600_max_tile_pipes = 4; + dev_priv->r600_max_simds = 8; + dev_priv->r600_max_backends = 2; + dev_priv->r600_max_gprs = 128; + dev_priv->r600_max_threads = 248; + dev_priv->r600_max_stack_entries = 256; + dev_priv->r600_max_hw_contexts = 8; + dev_priv->r600_max_gs_threads = 16 * 2; + dev_priv->r600_sx_max_export_size = 256; + dev_priv->r600_sx_max_export_pos_size = 32; + dev_priv->r600_sx_max_export_smx_size = 224; + dev_priv->r600_sq_num_cf_insts = 2; + + dev_priv->r700_sx_num_of_sets = 7; + dev_priv->r700_sc_prim_fifo_size = 0xf9; + dev_priv->r700_sc_hiz_tile_fifo_size = 0x30; + dev_priv->r700_sc_earlyz_tile_fifo_fize = 0x130; + if (dev_priv->r600_sx_max_export_pos_size > 16) { + dev_priv->r600_sx_max_export_pos_size -= 16; + dev_priv->r600_sx_max_export_smx_size += 16; + } + break; + case CHIP_RV710: + dev_priv->r600_max_pipes = 2; + dev_priv->r600_max_tile_pipes = 2; + dev_priv->r600_max_simds = 2; + dev_priv->r600_max_backends = 1; + dev_priv->r600_max_gprs = 256; + dev_priv->r600_max_threads = 192; + dev_priv->r600_max_stack_entries = 256; + dev_priv->r600_max_hw_contexts = 4; + dev_priv->r600_max_gs_threads = 8 * 2; + dev_priv->r600_sx_max_export_size = 128; + dev_priv->r600_sx_max_export_pos_size = 16; + dev_priv->r600_sx_max_export_smx_size = 112; + dev_priv->r600_sq_num_cf_insts = 1; + + dev_priv->r700_sx_num_of_sets = 7; + dev_priv->r700_sc_prim_fifo_size = 0x40; + dev_priv->r700_sc_hiz_tile_fifo_size = 0x30; + dev_priv->r700_sc_earlyz_tile_fifo_fize = 0x130; + break; + case CHIP_RV740: + dev_priv->r600_max_pipes = 4; + dev_priv->r600_max_tile_pipes = 4; + dev_priv->r600_max_simds = 8; + dev_priv->r600_max_backends = 4; + dev_priv->r600_max_gprs = 256; + dev_priv->r600_max_threads = 248; + dev_priv->r600_max_stack_entries = 512; + dev_priv->r600_max_hw_contexts = 8; + dev_priv->r600_max_gs_threads = 16 * 2; + dev_priv->r600_sx_max_export_size = 256; + dev_priv->r600_sx_max_export_pos_size = 32; + dev_priv->r600_sx_max_export_smx_size = 224; + dev_priv->r600_sq_num_cf_insts = 2; + + dev_priv->r700_sx_num_of_sets = 7; + dev_priv->r700_sc_prim_fifo_size = 0x100; + dev_priv->r700_sc_hiz_tile_fifo_size = 0x30; + dev_priv->r700_sc_earlyz_tile_fifo_fize = 0x130; + + if (dev_priv->r600_sx_max_export_pos_size > 16) { + dev_priv->r600_sx_max_export_pos_size -= 16; + dev_priv->r600_sx_max_export_smx_size += 16; + } + break; + default: + break; + } + + /* Initialize HDP */ + j = 0; + for (i = 0; i < 32; i++) { + RADEON_WRITE((0x2c14 + j), 0x00000000); + RADEON_WRITE((0x2c18 + j), 0x00000000); + RADEON_WRITE((0x2c1c + j), 0x00000000); + RADEON_WRITE((0x2c20 + j), 0x00000000); + RADEON_WRITE((0x2c24 + j), 0x00000000); + j += 0x18; + } + + RADEON_WRITE(R600_GRBM_CNTL, R600_GRBM_READ_TIMEOUT(0xff)); + + /* setup tiling, simd, pipe config */ + mc_arb_ramcfg = RADEON_READ(R700_MC_ARB_RAMCFG); + + switch (dev_priv->r600_max_tile_pipes) { + case 1: + gb_tiling_config |= R600_PIPE_TILING(0); + break; + case 2: + gb_tiling_config |= R600_PIPE_TILING(1); + break; + case 4: + gb_tiling_config |= R600_PIPE_TILING(2); + break; + case 8: + gb_tiling_config |= R600_PIPE_TILING(3); + break; + default: + break; + } + + if ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV770) + gb_tiling_config |= R600_BANK_TILING(1); + else + gb_tiling_config |= R600_BANK_TILING((mc_arb_ramcfg >> R700_NOOFBANK_SHIFT) & R700_NOOFBANK_MASK); + + gb_tiling_config |= R600_GROUP_SIZE(0); + + if (((mc_arb_ramcfg >> R700_NOOFROWS_SHIFT) & R700_NOOFROWS_MASK) > 3) { + gb_tiling_config |= R600_ROW_TILING(3); + gb_tiling_config |= R600_SAMPLE_SPLIT(3); + } else { + gb_tiling_config |= + R600_ROW_TILING(((mc_arb_ramcfg >> R700_NOOFROWS_SHIFT) & R700_NOOFROWS_MASK)); + gb_tiling_config |= + R600_SAMPLE_SPLIT(((mc_arb_ramcfg >> R700_NOOFROWS_SHIFT) & R700_NOOFROWS_MASK)); + } + + gb_tiling_config |= R600_BANK_SWAPS(1); + + cc_rb_backend_disable = RADEON_READ(R600_CC_RB_BACKEND_DISABLE) & 0x00ff0000; + cc_rb_backend_disable |= + R600_BACKEND_DISABLE((R7XX_MAX_BACKENDS_MASK << dev_priv->r600_max_backends) & R7XX_MAX_BACKENDS_MASK); + + cc_gc_shader_pipe_config = RADEON_READ(R600_CC_GC_SHADER_PIPE_CONFIG) & 0xffffff00; + cc_gc_shader_pipe_config |= + R600_INACTIVE_QD_PIPES((R7XX_MAX_PIPES_MASK << dev_priv->r600_max_pipes) & R7XX_MAX_PIPES_MASK); + cc_gc_shader_pipe_config |= + R600_INACTIVE_SIMDS((R7XX_MAX_SIMDS_MASK << dev_priv->r600_max_simds) & R7XX_MAX_SIMDS_MASK); + + if ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV740) + backend_map = 0x28; + else + backend_map = r700_get_tile_pipe_to_backend_map(dev_priv, + dev_priv->r600_max_tile_pipes, + (R7XX_MAX_BACKENDS - + r600_count_pipe_bits((cc_rb_backend_disable & + R7XX_MAX_BACKENDS_MASK) >> 16)), + (cc_rb_backend_disable >> 16)); + gb_tiling_config |= R600_BACKEND_MAP(backend_map); + + RADEON_WRITE(R600_GB_TILING_CONFIG, gb_tiling_config); + RADEON_WRITE(R600_DCP_TILING_CONFIG, (gb_tiling_config & 0xffff)); + RADEON_WRITE(R600_HDP_TILING_CONFIG, (gb_tiling_config & 0xffff)); + if (gb_tiling_config & 0xc0) { + dev_priv->r600_group_size = 512; + } else { + dev_priv->r600_group_size = 256; + } + dev_priv->r600_npipes = 1 << ((gb_tiling_config >> 1) & 0x7); + if (gb_tiling_config & 0x30) { + dev_priv->r600_nbanks = 8; + } else { + dev_priv->r600_nbanks = 4; + } + + RADEON_WRITE(R600_CC_RB_BACKEND_DISABLE, cc_rb_backend_disable); + RADEON_WRITE(R600_CC_GC_SHADER_PIPE_CONFIG, cc_gc_shader_pipe_config); + RADEON_WRITE(R600_GC_USER_SHADER_PIPE_CONFIG, cc_gc_shader_pipe_config); + + RADEON_WRITE(R700_CC_SYS_RB_BACKEND_DISABLE, cc_rb_backend_disable); + RADEON_WRITE(R700_CGTS_SYS_TCC_DISABLE, 0); + RADEON_WRITE(R700_CGTS_TCC_DISABLE, 0); + RADEON_WRITE(R700_CGTS_USER_SYS_TCC_DISABLE, 0); + RADEON_WRITE(R700_CGTS_USER_TCC_DISABLE, 0); + + num_qd_pipes = + R7XX_MAX_PIPES - r600_count_pipe_bits((cc_gc_shader_pipe_config & R600_INACTIVE_QD_PIPES_MASK) >> 8); + RADEON_WRITE(R600_VGT_OUT_DEALLOC_CNTL, (num_qd_pipes * 4) & R600_DEALLOC_DIST_MASK); + RADEON_WRITE(R600_VGT_VERTEX_REUSE_BLOCK_CNTL, ((num_qd_pipes * 4) - 2) & R600_VTX_REUSE_DEPTH_MASK); + + /* set HW defaults for 3D engine */ + RADEON_WRITE(R600_CP_QUEUE_THRESHOLDS, (R600_ROQ_IB1_START(0x16) | + R600_ROQ_IB2_START(0x2b))); + + RADEON_WRITE(R600_CP_MEQ_THRESHOLDS, R700_STQ_SPLIT(0x30)); + + ta_aux_cntl = RADEON_READ(R600_TA_CNTL_AUX); + RADEON_WRITE(R600_TA_CNTL_AUX, ta_aux_cntl | R600_DISABLE_CUBE_ANISO); + + sx_debug_1 = RADEON_READ(R700_SX_DEBUG_1); + sx_debug_1 |= R700_ENABLE_NEW_SMX_ADDRESS; + RADEON_WRITE(R700_SX_DEBUG_1, sx_debug_1); + + smx_dc_ctl0 = RADEON_READ(R600_SMX_DC_CTL0); + smx_dc_ctl0 &= ~R700_CACHE_DEPTH(0x1ff); + smx_dc_ctl0 |= R700_CACHE_DEPTH((dev_priv->r700_sx_num_of_sets * 64) - 1); + RADEON_WRITE(R600_SMX_DC_CTL0, smx_dc_ctl0); + + if ((dev_priv->flags & RADEON_FAMILY_MASK) != CHIP_RV740) + RADEON_WRITE(R700_SMX_EVENT_CTL, (R700_ES_FLUSH_CTL(4) | + R700_GS_FLUSH_CTL(4) | + R700_ACK_FLUSH_CTL(3) | + R700_SYNC_FLUSH_CTL)); + + db_debug3 = RADEON_READ(R700_DB_DEBUG3); + db_debug3 &= ~R700_DB_CLK_OFF_DELAY(0x1f); + switch (dev_priv->flags & RADEON_FAMILY_MASK) { + case CHIP_RV770: + case CHIP_RV740: + db_debug3 |= R700_DB_CLK_OFF_DELAY(0x1f); + break; + case CHIP_RV710: + case CHIP_RV730: + default: + db_debug3 |= R700_DB_CLK_OFF_DELAY(2); + break; + } + RADEON_WRITE(R700_DB_DEBUG3, db_debug3); + + if ((dev_priv->flags & RADEON_FAMILY_MASK) != CHIP_RV770) { + db_debug4 = RADEON_READ(RV700_DB_DEBUG4); + db_debug4 |= RV700_DISABLE_TILE_COVERED_FOR_PS_ITER; + RADEON_WRITE(RV700_DB_DEBUG4, db_debug4); + } + + RADEON_WRITE(R600_SX_EXPORT_BUFFER_SIZES, (R600_COLOR_BUFFER_SIZE((dev_priv->r600_sx_max_export_size / 4) - 1) | + R600_POSITION_BUFFER_SIZE((dev_priv->r600_sx_max_export_pos_size / 4) - 1) | + R600_SMX_BUFFER_SIZE((dev_priv->r600_sx_max_export_smx_size / 4) - 1))); + + RADEON_WRITE(R700_PA_SC_FIFO_SIZE_R7XX, (R700_SC_PRIM_FIFO_SIZE(dev_priv->r700_sc_prim_fifo_size) | + R700_SC_HIZ_TILE_FIFO_SIZE(dev_priv->r700_sc_hiz_tile_fifo_size) | + R700_SC_EARLYZ_TILE_FIFO_SIZE(dev_priv->r700_sc_earlyz_tile_fifo_fize))); + + RADEON_WRITE(R600_PA_SC_MULTI_CHIP_CNTL, 0); + + RADEON_WRITE(R600_VGT_NUM_INSTANCES, 1); + + RADEON_WRITE(R600_SPI_CONFIG_CNTL, R600_GPR_WRITE_PRIORITY(0)); + + RADEON_WRITE(R600_SPI_CONFIG_CNTL_1, R600_VTX_DONE_DELAY(4)); + + RADEON_WRITE(R600_CP_PERFMON_CNTL, 0); + + sq_ms_fifo_sizes = (R600_CACHE_FIFO_SIZE(16 * dev_priv->r600_sq_num_cf_insts) | + R600_DONE_FIFO_HIWATER(0xe0) | + R600_ALU_UPDATE_FIFO_HIWATER(0x8)); + switch (dev_priv->flags & RADEON_FAMILY_MASK) { + case CHIP_RV770: + case CHIP_RV730: + case CHIP_RV710: + sq_ms_fifo_sizes |= R600_FETCH_FIFO_HIWATER(0x1); + break; + case CHIP_RV740: + default: + sq_ms_fifo_sizes |= R600_FETCH_FIFO_HIWATER(0x4); + break; + } + RADEON_WRITE(R600_SQ_MS_FIFO_SIZES, sq_ms_fifo_sizes); + + /* SQ_CONFIG, SQ_GPR_RESOURCE_MGMT, SQ_THREAD_RESOURCE_MGMT, SQ_STACK_RESOURCE_MGMT + * should be adjusted as needed by the 2D/3D drivers. This just sets default values + */ + sq_config = RADEON_READ(R600_SQ_CONFIG); + sq_config &= ~(R600_PS_PRIO(3) | + R600_VS_PRIO(3) | + R600_GS_PRIO(3) | + R600_ES_PRIO(3)); + sq_config |= (R600_DX9_CONSTS | + R600_VC_ENABLE | + R600_EXPORT_SRC_C | + R600_PS_PRIO(0) | + R600_VS_PRIO(1) | + R600_GS_PRIO(2) | + R600_ES_PRIO(3)); + if ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV710) + /* no vertex cache */ + sq_config &= ~R600_VC_ENABLE; + + RADEON_WRITE(R600_SQ_CONFIG, sq_config); + + RADEON_WRITE(R600_SQ_GPR_RESOURCE_MGMT_1, (R600_NUM_PS_GPRS((dev_priv->r600_max_gprs * 24)/64) | + R600_NUM_VS_GPRS((dev_priv->r600_max_gprs * 24)/64) | + R600_NUM_CLAUSE_TEMP_GPRS(((dev_priv->r600_max_gprs * 24)/64)/2))); + + RADEON_WRITE(R600_SQ_GPR_RESOURCE_MGMT_2, (R600_NUM_GS_GPRS((dev_priv->r600_max_gprs * 7)/64) | + R600_NUM_ES_GPRS((dev_priv->r600_max_gprs * 7)/64))); + + sq_thread_resource_mgmt = (R600_NUM_PS_THREADS((dev_priv->r600_max_threads * 4)/8) | + R600_NUM_VS_THREADS((dev_priv->r600_max_threads * 2)/8) | + R600_NUM_ES_THREADS((dev_priv->r600_max_threads * 1)/8)); + if (((dev_priv->r600_max_threads * 1) / 8) > dev_priv->r600_max_gs_threads) + sq_thread_resource_mgmt |= R600_NUM_GS_THREADS(dev_priv->r600_max_gs_threads); + else + sq_thread_resource_mgmt |= R600_NUM_GS_THREADS((dev_priv->r600_max_gs_threads * 1)/8); + RADEON_WRITE(R600_SQ_THREAD_RESOURCE_MGMT, sq_thread_resource_mgmt); + + RADEON_WRITE(R600_SQ_STACK_RESOURCE_MGMT_1, (R600_NUM_PS_STACK_ENTRIES((dev_priv->r600_max_stack_entries * 1)/4) | + R600_NUM_VS_STACK_ENTRIES((dev_priv->r600_max_stack_entries * 1)/4))); + + RADEON_WRITE(R600_SQ_STACK_RESOURCE_MGMT_2, (R600_NUM_GS_STACK_ENTRIES((dev_priv->r600_max_stack_entries * 1)/4) | + R600_NUM_ES_STACK_ENTRIES((dev_priv->r600_max_stack_entries * 1)/4))); + + sq_dyn_gpr_size_simd_ab_0 = (R700_SIMDA_RING0((dev_priv->r600_max_gprs * 38)/64) | + R700_SIMDA_RING1((dev_priv->r600_max_gprs * 38)/64) | + R700_SIMDB_RING0((dev_priv->r600_max_gprs * 38)/64) | + R700_SIMDB_RING1((dev_priv->r600_max_gprs * 38)/64)); + + RADEON_WRITE(R700_SQ_DYN_GPR_SIZE_SIMD_AB_0, sq_dyn_gpr_size_simd_ab_0); + RADEON_WRITE(R700_SQ_DYN_GPR_SIZE_SIMD_AB_1, sq_dyn_gpr_size_simd_ab_0); + RADEON_WRITE(R700_SQ_DYN_GPR_SIZE_SIMD_AB_2, sq_dyn_gpr_size_simd_ab_0); + RADEON_WRITE(R700_SQ_DYN_GPR_SIZE_SIMD_AB_3, sq_dyn_gpr_size_simd_ab_0); + RADEON_WRITE(R700_SQ_DYN_GPR_SIZE_SIMD_AB_4, sq_dyn_gpr_size_simd_ab_0); + RADEON_WRITE(R700_SQ_DYN_GPR_SIZE_SIMD_AB_5, sq_dyn_gpr_size_simd_ab_0); + RADEON_WRITE(R700_SQ_DYN_GPR_SIZE_SIMD_AB_6, sq_dyn_gpr_size_simd_ab_0); + RADEON_WRITE(R700_SQ_DYN_GPR_SIZE_SIMD_AB_7, sq_dyn_gpr_size_simd_ab_0); + + RADEON_WRITE(R700_PA_SC_FORCE_EOV_MAX_CNTS, (R700_FORCE_EOV_MAX_CLK_CNT(4095) | + R700_FORCE_EOV_MAX_REZ_CNT(255))); + + if ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV710) + RADEON_WRITE(R600_VGT_CACHE_INVALIDATION, (R600_CACHE_INVALIDATION(R600_TC_ONLY) | + R700_AUTO_INVLD_EN(R700_ES_AND_GS_AUTO))); + else + RADEON_WRITE(R600_VGT_CACHE_INVALIDATION, (R600_CACHE_INVALIDATION(R600_VC_AND_TC) | + R700_AUTO_INVLD_EN(R700_ES_AND_GS_AUTO))); + + switch (dev_priv->flags & RADEON_FAMILY_MASK) { + case CHIP_RV770: + case CHIP_RV730: + case CHIP_RV740: + gs_prim_buffer_depth = 384; + break; + case CHIP_RV710: + gs_prim_buffer_depth = 128; + break; + default: + break; + } + + num_gs_verts_per_thread = dev_priv->r600_max_pipes * 16; + vgt_gs_per_es = gs_prim_buffer_depth + num_gs_verts_per_thread; + /* Max value for this is 256 */ + if (vgt_gs_per_es > 256) + vgt_gs_per_es = 256; + + RADEON_WRITE(R600_VGT_ES_PER_GS, 128); + RADEON_WRITE(R600_VGT_GS_PER_ES, vgt_gs_per_es); + RADEON_WRITE(R600_VGT_GS_PER_VS, 2); + + /* more default values. 2D/3D driver should adjust as needed */ + RADEON_WRITE(R600_VGT_GS_VERTEX_REUSE, 16); + RADEON_WRITE(R600_PA_SC_LINE_STIPPLE_STATE, 0); + RADEON_WRITE(R600_VGT_STRMOUT_EN, 0); + RADEON_WRITE(R600_SX_MISC, 0); + RADEON_WRITE(R600_PA_SC_MODE_CNTL, 0); + RADEON_WRITE(R700_PA_SC_EDGERULE, 0xaaaaaaaa); + RADEON_WRITE(R600_PA_SC_AA_CONFIG, 0); + RADEON_WRITE(R600_PA_SC_CLIPRECT_RULE, 0xffff); + RADEON_WRITE(R600_PA_SC_LINE_STIPPLE, 0); + RADEON_WRITE(R600_SPI_INPUT_Z, 0); + RADEON_WRITE(R600_SPI_PS_IN_CONTROL_0, R600_NUM_INTERP(2)); + RADEON_WRITE(R600_CB_COLOR7_FRAG, 0); + + /* clear render buffer base addresses */ + RADEON_WRITE(R600_CB_COLOR0_BASE, 0); + RADEON_WRITE(R600_CB_COLOR1_BASE, 0); + RADEON_WRITE(R600_CB_COLOR2_BASE, 0); + RADEON_WRITE(R600_CB_COLOR3_BASE, 0); + RADEON_WRITE(R600_CB_COLOR4_BASE, 0); + RADEON_WRITE(R600_CB_COLOR5_BASE, 0); + RADEON_WRITE(R600_CB_COLOR6_BASE, 0); + RADEON_WRITE(R600_CB_COLOR7_BASE, 0); + + RADEON_WRITE(R700_TCP_CNTL, 0); + + hdp_host_path_cntl = RADEON_READ(R600_HDP_HOST_PATH_CNTL); + RADEON_WRITE(R600_HDP_HOST_PATH_CNTL, hdp_host_path_cntl); + + RADEON_WRITE(R600_PA_SC_MULTI_CHIP_CNTL, 0); + + RADEON_WRITE(R600_PA_CL_ENHANCE, (R600_CLIP_VTX_REORDER_ENA | + R600_NUM_CLIP_SEQ(3))); + +} + +static void r600_cp_init_ring_buffer(struct drm_device *dev, + drm_radeon_private_t *dev_priv, + struct drm_file *file_priv) +{ + struct drm_radeon_master_private *master_priv; + u32 ring_start; + u64 rptr_addr; + + if (((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_RV770)) + r700_gfx_init(dev, dev_priv); + else + r600_gfx_init(dev, dev_priv); + + RADEON_WRITE(R600_GRBM_SOFT_RESET, R600_SOFT_RESET_CP); + RADEON_READ(R600_GRBM_SOFT_RESET); + DRM_UDELAY(15000); + RADEON_WRITE(R600_GRBM_SOFT_RESET, 0); + + + /* Set ring buffer size */ +#ifdef __BIG_ENDIAN + RADEON_WRITE(R600_CP_RB_CNTL, + RADEON_BUF_SWAP_32BIT | + RADEON_RB_NO_UPDATE | + (dev_priv->ring.rptr_update_l2qw << 8) | + dev_priv->ring.size_l2qw); +#else + RADEON_WRITE(R600_CP_RB_CNTL, + RADEON_RB_NO_UPDATE | + (dev_priv->ring.rptr_update_l2qw << 8) | + dev_priv->ring.size_l2qw); +#endif + + RADEON_WRITE(R600_CP_SEM_WAIT_TIMER, 0x4); + + /* Set the write pointer delay */ + RADEON_WRITE(R600_CP_RB_WPTR_DELAY, 0); + +#ifdef __BIG_ENDIAN + RADEON_WRITE(R600_CP_RB_CNTL, + RADEON_BUF_SWAP_32BIT | + RADEON_RB_NO_UPDATE | + RADEON_RB_RPTR_WR_ENA | + (dev_priv->ring.rptr_update_l2qw << 8) | + dev_priv->ring.size_l2qw); +#else + RADEON_WRITE(R600_CP_RB_CNTL, + RADEON_RB_NO_UPDATE | + RADEON_RB_RPTR_WR_ENA | + (dev_priv->ring.rptr_update_l2qw << 8) | + dev_priv->ring.size_l2qw); +#endif + + /* Initialize the ring buffer's read and write pointers */ + RADEON_WRITE(R600_CP_RB_RPTR_WR, 0); + RADEON_WRITE(R600_CP_RB_WPTR, 0); + SET_RING_HEAD(dev_priv, 0); + dev_priv->ring.tail = 0; + +#if __OS_HAS_AGP + if (dev_priv->flags & RADEON_IS_AGP) { + rptr_addr = dev_priv->ring_rptr->offset + - dev->agp->base + + dev_priv->gart_vm_start; + } else +#endif + { + rptr_addr = dev_priv->ring_rptr->offset + - ((unsigned long) dev->sg->virtual) + + dev_priv->gart_vm_start; + } + RADEON_WRITE(R600_CP_RB_RPTR_ADDR, +#ifdef __BIG_ENDIAN + (2 << 0) | +#endif + (rptr_addr & 0xfffffffc)); + RADEON_WRITE(R600_CP_RB_RPTR_ADDR_HI, + upper_32_bits(rptr_addr)); + +#ifdef __BIG_ENDIAN + RADEON_WRITE(R600_CP_RB_CNTL, + RADEON_BUF_SWAP_32BIT | + (dev_priv->ring.rptr_update_l2qw << 8) | + dev_priv->ring.size_l2qw); +#else + RADEON_WRITE(R600_CP_RB_CNTL, + (dev_priv->ring.rptr_update_l2qw << 8) | + dev_priv->ring.size_l2qw); +#endif + +#if __OS_HAS_AGP + if (dev_priv->flags & RADEON_IS_AGP) { + /* XXX */ + radeon_write_agp_base(dev_priv, dev->agp->base); + + /* XXX */ + radeon_write_agp_location(dev_priv, + (((dev_priv->gart_vm_start - 1 + + dev_priv->gart_size) & 0xffff0000) | + (dev_priv->gart_vm_start >> 16))); + + ring_start = (dev_priv->cp_ring->offset + - dev->agp->base + + dev_priv->gart_vm_start); + } else +#endif + ring_start = (dev_priv->cp_ring->offset + - (unsigned long)dev->sg->virtual + + dev_priv->gart_vm_start); + + RADEON_WRITE(R600_CP_RB_BASE, ring_start >> 8); + + RADEON_WRITE(R600_CP_ME_CNTL, 0xff); + + RADEON_WRITE(R600_CP_DEBUG, (1 << 27) | (1 << 28)); + + /* Initialize the scratch register pointer. This will cause + * the scratch register values to be written out to memory + * whenever they are updated. + * + * We simply put this behind the ring read pointer, this works + * with PCI GART as well as (whatever kind of) AGP GART + */ + { + u64 scratch_addr; + + scratch_addr = RADEON_READ(R600_CP_RB_RPTR_ADDR) & 0xFFFFFFFC; + scratch_addr |= ((u64)RADEON_READ(R600_CP_RB_RPTR_ADDR_HI)) << 32; + scratch_addr += R600_SCRATCH_REG_OFFSET; + scratch_addr >>= 8; + scratch_addr &= 0xffffffff; + + RADEON_WRITE(R600_SCRATCH_ADDR, (uint32_t)scratch_addr); + } + + RADEON_WRITE(R600_SCRATCH_UMSK, 0x7); + + /* Turn on bus mastering */ + radeon_enable_bm(dev_priv); + + radeon_write_ring_rptr(dev_priv, R600_SCRATCHOFF(0), 0); + RADEON_WRITE(R600_LAST_FRAME_REG, 0); + + radeon_write_ring_rptr(dev_priv, R600_SCRATCHOFF(1), 0); + RADEON_WRITE(R600_LAST_DISPATCH_REG, 0); + + radeon_write_ring_rptr(dev_priv, R600_SCRATCHOFF(2), 0); + RADEON_WRITE(R600_LAST_CLEAR_REG, 0); + + /* reset sarea copies of these */ + master_priv = file_priv->master->driver_priv; + if (master_priv->sarea_priv) { + master_priv->sarea_priv->last_frame = 0; + master_priv->sarea_priv->last_dispatch = 0; + master_priv->sarea_priv->last_clear = 0; + } + + r600_do_wait_for_idle(dev_priv); + +} + +int r600_do_cleanup_cp(struct drm_device *dev) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + DRM_DEBUG("\n"); + + /* Make sure interrupts are disabled here because the uninstall ioctl + * may not have been called from userspace and after dev_private + * is freed, it's too late. + */ + if (dev->irq_enabled) + drm_irq_uninstall(dev); + +#if __OS_HAS_AGP + if (dev_priv->flags & RADEON_IS_AGP) { + if (dev_priv->cp_ring != NULL) { + drm_core_ioremapfree(dev_priv->cp_ring, dev); + dev_priv->cp_ring = NULL; + } + if (dev_priv->ring_rptr != NULL) { + drm_core_ioremapfree(dev_priv->ring_rptr, dev); + dev_priv->ring_rptr = NULL; + } + if (dev->agp_buffer_map != NULL) { + drm_core_ioremapfree(dev->agp_buffer_map, dev); + dev->agp_buffer_map = NULL; + } + } else +#endif + { + + if (dev_priv->gart_info.bus_addr) + r600_page_table_cleanup(dev, &dev_priv->gart_info); + + if (dev_priv->gart_info.gart_table_location == DRM_ATI_GART_FB) { + drm_core_ioremapfree(&dev_priv->gart_info.mapping, dev); + dev_priv->gart_info.addr = NULL; + } + } + /* only clear to the start of flags */ + memset(dev_priv, 0, offsetof(drm_radeon_private_t, flags)); + + return 0; +} + +int r600_do_init_cp(struct drm_device *dev, drm_radeon_init_t *init, + struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + struct drm_radeon_master_private *master_priv = file_priv->master->driver_priv; + + DRM_DEBUG("\n"); + + mutex_init(&dev_priv->cs_mutex); + r600_cs_legacy_init(); + /* if we require new memory map but we don't have it fail */ + if ((dev_priv->flags & RADEON_NEW_MEMMAP) && !dev_priv->new_memmap) { + DRM_ERROR("Cannot initialise DRM on this card\nThis card requires a new X.org DDX for 3D\n"); + r600_do_cleanup_cp(dev); + return -EINVAL; + } + + if (init->is_pci && (dev_priv->flags & RADEON_IS_AGP)) { + DRM_DEBUG("Forcing AGP card to PCI mode\n"); + dev_priv->flags &= ~RADEON_IS_AGP; + /* The writeback test succeeds, but when writeback is enabled, + * the ring buffer read ptr update fails after first 128 bytes. + */ + radeon_no_wb = 1; + } else if (!(dev_priv->flags & (RADEON_IS_AGP | RADEON_IS_PCI | RADEON_IS_PCIE)) + && !init->is_pci) { + DRM_DEBUG("Restoring AGP flag\n"); + dev_priv->flags |= RADEON_IS_AGP; + } + + dev_priv->usec_timeout = init->usec_timeout; + if (dev_priv->usec_timeout < 1 || + dev_priv->usec_timeout > RADEON_MAX_USEC_TIMEOUT) { + DRM_DEBUG("TIMEOUT problem!\n"); + r600_do_cleanup_cp(dev); + return -EINVAL; + } + + /* Enable vblank on CRTC1 for older X servers + */ + dev_priv->vblank_crtc = DRM_RADEON_VBLANK_CRTC1; + dev_priv->do_boxes = 0; + dev_priv->cp_mode = init->cp_mode; + + /* We don't support anything other than bus-mastering ring mode, + * but the ring can be in either AGP or PCI space for the ring + * read pointer. + */ + if ((init->cp_mode != RADEON_CSQ_PRIBM_INDDIS) && + (init->cp_mode != RADEON_CSQ_PRIBM_INDBM)) { + DRM_DEBUG("BAD cp_mode (%x)!\n", init->cp_mode); + r600_do_cleanup_cp(dev); + return -EINVAL; + } + + switch (init->fb_bpp) { + case 16: + dev_priv->color_fmt = RADEON_COLOR_FORMAT_RGB565; + break; + case 32: + default: + dev_priv->color_fmt = RADEON_COLOR_FORMAT_ARGB8888; + break; + } + dev_priv->front_offset = init->front_offset; + dev_priv->front_pitch = init->front_pitch; + dev_priv->back_offset = init->back_offset; + dev_priv->back_pitch = init->back_pitch; + + dev_priv->ring_offset = init->ring_offset; + dev_priv->ring_rptr_offset = init->ring_rptr_offset; + dev_priv->buffers_offset = init->buffers_offset; + dev_priv->gart_textures_offset = init->gart_textures_offset; + + master_priv->sarea = drm_getsarea(dev); + if (!master_priv->sarea) { + DRM_ERROR("could not find sarea!\n"); + r600_do_cleanup_cp(dev); + return -EINVAL; + } + + dev_priv->cp_ring = drm_core_findmap(dev, init->ring_offset); + if (!dev_priv->cp_ring) { + DRM_ERROR("could not find cp ring region!\n"); + r600_do_cleanup_cp(dev); + return -EINVAL; + } + dev_priv->ring_rptr = drm_core_findmap(dev, init->ring_rptr_offset); + if (!dev_priv->ring_rptr) { + DRM_ERROR("could not find ring read pointer!\n"); + r600_do_cleanup_cp(dev); + return -EINVAL; + } + dev->agp_buffer_token = init->buffers_offset; + dev->agp_buffer_map = drm_core_findmap(dev, init->buffers_offset); + if (!dev->agp_buffer_map) { + DRM_ERROR("could not find dma buffer region!\n"); + r600_do_cleanup_cp(dev); + return -EINVAL; + } + + if (init->gart_textures_offset) { + dev_priv->gart_textures = + drm_core_findmap(dev, init->gart_textures_offset); + if (!dev_priv->gart_textures) { + DRM_ERROR("could not find GART texture region!\n"); + r600_do_cleanup_cp(dev); + return -EINVAL; + } + } + +#if __OS_HAS_AGP + /* XXX */ + if (dev_priv->flags & RADEON_IS_AGP) { + drm_core_ioremap_wc(dev_priv->cp_ring, dev); + drm_core_ioremap_wc(dev_priv->ring_rptr, dev); + drm_core_ioremap_wc(dev->agp_buffer_map, dev); + if (!dev_priv->cp_ring->handle || + !dev_priv->ring_rptr->handle || + !dev->agp_buffer_map->handle) { + DRM_ERROR("could not find ioremap agp regions!\n"); + r600_do_cleanup_cp(dev); + return -EINVAL; + } + } else +#endif + { + dev_priv->cp_ring->handle = (void *)(unsigned long)dev_priv->cp_ring->offset; + dev_priv->ring_rptr->handle = + (void *)(unsigned long)dev_priv->ring_rptr->offset; + dev->agp_buffer_map->handle = + (void *)(unsigned long)dev->agp_buffer_map->offset; + + DRM_DEBUG("dev_priv->cp_ring->handle %p\n", + dev_priv->cp_ring->handle); + DRM_DEBUG("dev_priv->ring_rptr->handle %p\n", + dev_priv->ring_rptr->handle); + DRM_DEBUG("dev->agp_buffer_map->handle %p\n", + dev->agp_buffer_map->handle); + } + + dev_priv->fb_location = (radeon_read_fb_location(dev_priv) & 0xffff) << 24; + dev_priv->fb_size = + (((radeon_read_fb_location(dev_priv) & 0xffff0000u) << 8) + 0x1000000) + - dev_priv->fb_location; + + dev_priv->front_pitch_offset = (((dev_priv->front_pitch / 64) << 22) | + ((dev_priv->front_offset + + dev_priv->fb_location) >> 10)); + + dev_priv->back_pitch_offset = (((dev_priv->back_pitch / 64) << 22) | + ((dev_priv->back_offset + + dev_priv->fb_location) >> 10)); + + dev_priv->depth_pitch_offset = (((dev_priv->depth_pitch / 64) << 22) | + ((dev_priv->depth_offset + + dev_priv->fb_location) >> 10)); + + dev_priv->gart_size = init->gart_size; + + /* New let's set the memory map ... */ + if (dev_priv->new_memmap) { + u32 base = 0; + + DRM_INFO("Setting GART location based on new memory map\n"); + + /* If using AGP, try to locate the AGP aperture at the same + * location in the card and on the bus, though we have to + * align it down. + */ +#if __OS_HAS_AGP + /* XXX */ + if (dev_priv->flags & RADEON_IS_AGP) { + base = dev->agp->base; + /* Check if valid */ + if ((base + dev_priv->gart_size - 1) >= dev_priv->fb_location && + base < (dev_priv->fb_location + dev_priv->fb_size - 1)) { + DRM_INFO("Can't use AGP base @0x%08lx, won't fit\n", + dev->agp->base); + base = 0; + } + } +#endif + /* If not or if AGP is at 0 (Macs), try to put it elsewhere */ + if (base == 0) { + base = dev_priv->fb_location + dev_priv->fb_size; + if (base < dev_priv->fb_location || + ((base + dev_priv->gart_size) & 0xfffffffful) < base) + base = dev_priv->fb_location + - dev_priv->gart_size; + } + dev_priv->gart_vm_start = base & 0xffc00000u; + if (dev_priv->gart_vm_start != base) + DRM_INFO("GART aligned down from 0x%08x to 0x%08x\n", + base, dev_priv->gart_vm_start); + } + +#if __OS_HAS_AGP + /* XXX */ + if (dev_priv->flags & RADEON_IS_AGP) + dev_priv->gart_buffers_offset = (dev->agp_buffer_map->offset + - dev->agp->base + + dev_priv->gart_vm_start); + else +#endif + dev_priv->gart_buffers_offset = (dev->agp_buffer_map->offset + - (unsigned long)dev->sg->virtual + + dev_priv->gart_vm_start); + + DRM_DEBUG("fb 0x%08x size %d\n", + (unsigned int) dev_priv->fb_location, + (unsigned int) dev_priv->fb_size); + DRM_DEBUG("dev_priv->gart_size %d\n", dev_priv->gart_size); + DRM_DEBUG("dev_priv->gart_vm_start 0x%08x\n", + (unsigned int) dev_priv->gart_vm_start); + DRM_DEBUG("dev_priv->gart_buffers_offset 0x%08lx\n", + dev_priv->gart_buffers_offset); + + dev_priv->ring.start = (u32 *) dev_priv->cp_ring->handle; + dev_priv->ring.end = ((u32 *) dev_priv->cp_ring->handle + + init->ring_size / sizeof(u32)); + dev_priv->ring.size = init->ring_size; + dev_priv->ring.size_l2qw = drm_order(init->ring_size / 8); + + dev_priv->ring.rptr_update = /* init->rptr_update */ 4096; + dev_priv->ring.rptr_update_l2qw = drm_order(/* init->rptr_update */ 4096 / 8); + + dev_priv->ring.fetch_size = /* init->fetch_size */ 32; + dev_priv->ring.fetch_size_l2ow = drm_order(/* init->fetch_size */ 32 / 16); + + dev_priv->ring.tail_mask = (dev_priv->ring.size / sizeof(u32)) - 1; + + dev_priv->ring.high_mark = RADEON_RING_HIGH_MARK; + +#if __OS_HAS_AGP + if (dev_priv->flags & RADEON_IS_AGP) { + /* XXX turn off pcie gart */ + } else +#endif + { + dev_priv->gart_info.table_mask = DMA_BIT_MASK(32); + /* if we have an offset set from userspace */ + if (!dev_priv->pcigart_offset_set) { + DRM_ERROR("Need gart offset from userspace\n"); + r600_do_cleanup_cp(dev); + return -EINVAL; + } + + DRM_DEBUG("Using gart offset 0x%08lx\n", dev_priv->pcigart_offset); + + dev_priv->gart_info.bus_addr = + dev_priv->pcigart_offset + dev_priv->fb_location; + dev_priv->gart_info.mapping.offset = + dev_priv->pcigart_offset + dev_priv->fb_aper_offset; + dev_priv->gart_info.mapping.size = + dev_priv->gart_info.table_size; + + drm_core_ioremap_wc(&dev_priv->gart_info.mapping, dev); + if (!dev_priv->gart_info.mapping.handle) { + DRM_ERROR("ioremap failed.\n"); + r600_do_cleanup_cp(dev); + return -EINVAL; + } + + dev_priv->gart_info.addr = + dev_priv->gart_info.mapping.handle; + + DRM_DEBUG("Setting phys_pci_gart to %p %08lX\n", + dev_priv->gart_info.addr, + dev_priv->pcigart_offset); + + if (!r600_page_table_init(dev)) { + DRM_ERROR("Failed to init GART table\n"); + r600_do_cleanup_cp(dev); + return -EINVAL; + } + + if (((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_RV770)) + r700_vm_init(dev); + else + r600_vm_init(dev); + } + + if (!dev_priv->me_fw || !dev_priv->pfp_fw) { + int err = r600_cp_init_microcode(dev_priv); + if (err) { + DRM_ERROR("Failed to load firmware!\n"); + r600_do_cleanup_cp(dev); + return err; + } + } + if (((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_RV770)) + r700_cp_load_microcode(dev_priv); + else + r600_cp_load_microcode(dev_priv); + + r600_cp_init_ring_buffer(dev, dev_priv, file_priv); + + dev_priv->last_buf = 0; + + r600_do_engine_reset(dev); + r600_test_writeback(dev_priv); + + return 0; +} + +int r600_do_resume_cp(struct drm_device *dev, struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + + DRM_DEBUG("\n"); + if (((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_RV770)) { + r700_vm_init(dev); + r700_cp_load_microcode(dev_priv); + } else { + r600_vm_init(dev); + r600_cp_load_microcode(dev_priv); + } + r600_cp_init_ring_buffer(dev, dev_priv, file_priv); + r600_do_engine_reset(dev); + + return 0; +} + +/* Wait for the CP to go idle. + */ +int r600_do_cp_idle(drm_radeon_private_t *dev_priv) +{ + RING_LOCALS; + DRM_DEBUG("\n"); + + BEGIN_RING(5); + OUT_RING(CP_PACKET3(R600_IT_EVENT_WRITE, 0)); + OUT_RING(R600_CACHE_FLUSH_AND_INV_EVENT); + /* wait for 3D idle clean */ + OUT_RING(CP_PACKET3(R600_IT_SET_CONFIG_REG, 1)); + OUT_RING((R600_WAIT_UNTIL - R600_SET_CONFIG_REG_OFFSET) >> 2); + OUT_RING(RADEON_WAIT_3D_IDLE | RADEON_WAIT_3D_IDLECLEAN); + + ADVANCE_RING(); + COMMIT_RING(); + + return r600_do_wait_for_idle(dev_priv); +} + +/* Start the Command Processor. + */ +void r600_do_cp_start(drm_radeon_private_t *dev_priv) +{ + u32 cp_me; + RING_LOCALS; + DRM_DEBUG("\n"); + + BEGIN_RING(7); + OUT_RING(CP_PACKET3(R600_IT_ME_INITIALIZE, 5)); + OUT_RING(0x00000001); + if (((dev_priv->flags & RADEON_FAMILY_MASK) < CHIP_RV770)) + OUT_RING(0x00000003); + else + OUT_RING(0x00000000); + OUT_RING((dev_priv->r600_max_hw_contexts - 1)); + OUT_RING(R600_ME_INITIALIZE_DEVICE_ID(1)); + OUT_RING(0x00000000); + OUT_RING(0x00000000); + ADVANCE_RING(); + COMMIT_RING(); + + /* set the mux and reset the halt bit */ + cp_me = 0xff; + RADEON_WRITE(R600_CP_ME_CNTL, cp_me); + + dev_priv->cp_running = 1; + +} + +void r600_do_cp_reset(drm_radeon_private_t *dev_priv) +{ + u32 cur_read_ptr; + DRM_DEBUG("\n"); + + cur_read_ptr = RADEON_READ(R600_CP_RB_RPTR); + RADEON_WRITE(R600_CP_RB_WPTR, cur_read_ptr); + SET_RING_HEAD(dev_priv, cur_read_ptr); + dev_priv->ring.tail = cur_read_ptr; +} + +void r600_do_cp_stop(drm_radeon_private_t *dev_priv) +{ + uint32_t cp_me; + + DRM_DEBUG("\n"); + + cp_me = 0xff | R600_CP_ME_HALT; + + RADEON_WRITE(R600_CP_ME_CNTL, cp_me); + + dev_priv->cp_running = 0; +} + +int r600_cp_dispatch_indirect(struct drm_device *dev, + struct drm_buf *buf, int start, int end) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + RING_LOCALS; + + if (start != end) { + unsigned long offset = (dev_priv->gart_buffers_offset + + buf->offset + start); + int dwords = (end - start + 3) / sizeof(u32); + + DRM_DEBUG("dwords:%d\n", dwords); + DRM_DEBUG("offset 0x%lx\n", offset); + + + /* Indirect buffer data must be a multiple of 16 dwords. + * pad the data with a Type-2 CP packet. + */ + while (dwords & 0xf) { + u32 *data = (u32 *) + ((char *)dev->agp_buffer_map->handle + + buf->offset + start); + data[dwords++] = RADEON_CP_PACKET2; + } + + /* Fire off the indirect buffer */ + BEGIN_RING(4); + OUT_RING(CP_PACKET3(R600_IT_INDIRECT_BUFFER, 2)); + OUT_RING((offset & 0xfffffffc)); + OUT_RING((upper_32_bits(offset) & 0xff)); + OUT_RING(dwords); + ADVANCE_RING(); + } + + return 0; +} + +void r600_cp_dispatch_swap(struct drm_device *dev, struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + struct drm_master *master = file_priv->master; + struct drm_radeon_master_private *master_priv = master->driver_priv; + drm_radeon_sarea_t *sarea_priv = master_priv->sarea_priv; + int nbox = sarea_priv->nbox; + struct drm_clip_rect *pbox = sarea_priv->boxes; + int i, cpp, src_pitch, dst_pitch; + uint64_t src, dst; + RING_LOCALS; + DRM_DEBUG("\n"); + + if (dev_priv->color_fmt == RADEON_COLOR_FORMAT_ARGB8888) + cpp = 4; + else + cpp = 2; + + if (sarea_priv->pfCurrentPage == 0) { + src_pitch = dev_priv->back_pitch; + dst_pitch = dev_priv->front_pitch; + src = dev_priv->back_offset + dev_priv->fb_location; + dst = dev_priv->front_offset + dev_priv->fb_location; + } else { + src_pitch = dev_priv->front_pitch; + dst_pitch = dev_priv->back_pitch; + src = dev_priv->front_offset + dev_priv->fb_location; + dst = dev_priv->back_offset + dev_priv->fb_location; + } + + if (r600_prepare_blit_copy(dev, file_priv)) { + DRM_ERROR("unable to allocate vertex buffer for swap buffer\n"); + return; + } + for (i = 0; i < nbox; i++) { + int x = pbox[i].x1; + int y = pbox[i].y1; + int w = pbox[i].x2 - x; + int h = pbox[i].y2 - y; + + DRM_DEBUG("%d,%d-%d,%d\n", x, y, w, h); + + r600_blit_swap(dev, + src, dst, + x, y, x, y, w, h, + src_pitch, dst_pitch, cpp); + } + r600_done_blit_copy(dev); + + /* Increment the frame counter. The client-side 3D driver must + * throttle the framerate by waiting for this value before + * performing the swapbuffer ioctl. + */ + sarea_priv->last_frame++; + + BEGIN_RING(3); + R600_FRAME_AGE(sarea_priv->last_frame); + ADVANCE_RING(); +} + +int r600_cp_dispatch_texture(struct drm_device *dev, + struct drm_file *file_priv, + drm_radeon_texture_t *tex, + drm_radeon_tex_image_t *image) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + struct drm_buf *buf; + u32 *buffer; + const u8 __user *data; + int size, pass_size; + u64 src_offset, dst_offset; + + if (!radeon_check_offset(dev_priv, tex->offset)) { + DRM_ERROR("Invalid destination offset\n"); + return -EINVAL; + } + + /* this might fail for zero-sized uploads - are those illegal? */ + if (!radeon_check_offset(dev_priv, tex->offset + tex->height * tex->pitch - 1)) { + DRM_ERROR("Invalid final destination offset\n"); + return -EINVAL; + } + + size = tex->height * tex->pitch; + + if (size == 0) + return 0; + + dst_offset = tex->offset; + + if (r600_prepare_blit_copy(dev, file_priv)) { + DRM_ERROR("unable to allocate vertex buffer for swap buffer\n"); + return -EAGAIN; + } + do { + data = (const u8 __user *)image->data; + pass_size = size; + + buf = radeon_freelist_get(dev); + if (!buf) { + DRM_DEBUG("EAGAIN\n"); + if (DRM_COPY_TO_USER(tex->image, image, sizeof(*image))) + return -EFAULT; + return -EAGAIN; + } + + if (pass_size > buf->total) + pass_size = buf->total; + + /* Dispatch the indirect buffer. + */ + buffer = + (u32 *) ((char *)dev->agp_buffer_map->handle + buf->offset); + + if (DRM_COPY_FROM_USER(buffer, data, pass_size)) { + DRM_ERROR("EFAULT on pad, %d bytes\n", pass_size); + return -EFAULT; + } + + buf->file_priv = file_priv; + buf->used = pass_size; + src_offset = dev_priv->gart_buffers_offset + buf->offset; + + r600_blit_copy(dev, src_offset, dst_offset, pass_size); + + radeon_cp_discard_buffer(dev, file_priv->master, buf); + + /* Update the input parameters for next time */ + image->data = (const u8 __user *)image->data + pass_size; + dst_offset += pass_size; + size -= pass_size; + } while (size > 0); + r600_done_blit_copy(dev); + + return 0; +} + +/* + * Legacy cs ioctl + */ +static u32 radeon_cs_id_get(struct drm_radeon_private *radeon) +{ + /* FIXME: check if wrap affect last reported wrap & sequence */ + radeon->cs_id_scnt = (radeon->cs_id_scnt + 1) & 0x00FFFFFF; + if (!radeon->cs_id_scnt) { + /* increment wrap counter */ + radeon->cs_id_wcnt += 0x01000000; + /* valid sequence counter start at 1 */ + radeon->cs_id_scnt = 1; + } + return (radeon->cs_id_scnt | radeon->cs_id_wcnt); +} + +static void r600_cs_id_emit(drm_radeon_private_t *dev_priv, u32 *id) +{ + RING_LOCALS; + + *id = radeon_cs_id_get(dev_priv); + + /* SCRATCH 2 */ + BEGIN_RING(3); + R600_CLEAR_AGE(*id); + ADVANCE_RING(); + COMMIT_RING(); +} + +static int r600_ib_get(struct drm_device *dev, + struct drm_file *fpriv, + struct drm_buf **buffer) +{ + struct drm_buf *buf; + + *buffer = NULL; + buf = radeon_freelist_get(dev); + if (!buf) { + return -EBUSY; + } + buf->file_priv = fpriv; + *buffer = buf; + return 0; +} + +static void r600_ib_free(struct drm_device *dev, struct drm_buf *buf, + struct drm_file *fpriv, int l, int r) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + + if (buf) { + if (!r) + r600_cp_dispatch_indirect(dev, buf, 0, l * 4); + radeon_cp_discard_buffer(dev, fpriv->master, buf); + COMMIT_RING(); + } +} + +int r600_cs_legacy_ioctl(struct drm_device *dev, void *data, struct drm_file *fpriv) +{ + struct drm_radeon_private *dev_priv = dev->dev_private; + struct drm_radeon_cs *cs = data; + struct drm_buf *buf; + unsigned family; + int l, r = 0; + u32 *ib, cs_id = 0; + + if (dev_priv == NULL) { + DRM_ERROR("called with no initialization\n"); + return -EINVAL; + } + family = dev_priv->flags & RADEON_FAMILY_MASK; + if (family < CHIP_R600) { + DRM_ERROR("cs ioctl valid only for R6XX & R7XX in legacy mode\n"); + return -EINVAL; + } + mutex_lock(&dev_priv->cs_mutex); + /* get ib */ + r = r600_ib_get(dev, fpriv, &buf); + if (r) { + DRM_ERROR("ib_get failed\n"); + goto out; + } + ib = dev->agp_buffer_map->handle + buf->offset; + /* now parse command stream */ + r = r600_cs_legacy(dev, data, fpriv, family, ib, &l); + if (r) { + goto out; + } + +out: + r600_ib_free(dev, buf, fpriv, l, r); + /* emit cs id sequence */ + r600_cs_id_emit(dev_priv, &cs_id); + cs->cs_id = cs_id; + mutex_unlock(&dev_priv->cs_mutex); + return r; +} + +void r600_cs_legacy_get_tiling_conf(struct drm_device *dev, u32 *npipes, u32 *nbanks, u32 *group_size) +{ + struct drm_radeon_private *dev_priv = dev->dev_private; + + *npipes = dev_priv->r600_npipes; + *nbanks = dev_priv->r600_nbanks; + *group_size = dev_priv->r600_group_size; +} diff --git a/drivers/gpu/drm/radeon/r600_cs.c b/drivers/gpu/drm/radeon/r600_cs.c new file mode 100644 index 00000000..909bda8d --- /dev/null +++ b/drivers/gpu/drm/radeon/r600_cs.c @@ -0,0 +1,1917 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#include +#include "drmP.h" +#include "radeon.h" +#include "r600d.h" +#include "r600_reg_safe.h" + +static int r600_cs_packet_next_reloc_mm(struct radeon_cs_parser *p, + struct radeon_cs_reloc **cs_reloc); +static int r600_cs_packet_next_reloc_nomm(struct radeon_cs_parser *p, + struct radeon_cs_reloc **cs_reloc); +typedef int (*next_reloc_t)(struct radeon_cs_parser*, struct radeon_cs_reloc**); +static next_reloc_t r600_cs_packet_next_reloc = &r600_cs_packet_next_reloc_mm; +extern void r600_cs_legacy_get_tiling_conf(struct drm_device *dev, u32 *npipes, u32 *nbanks, u32 *group_size); + + +struct r600_cs_track { + /* configuration we miror so that we use same code btw kms/ums */ + u32 group_size; + u32 nbanks; + u32 npipes; + /* value we track */ + u32 sq_config; + u32 nsamples; + u32 cb_color_base_last[8]; + struct radeon_bo *cb_color_bo[8]; + u64 cb_color_bo_mc[8]; + u32 cb_color_bo_offset[8]; + struct radeon_bo *cb_color_frag_bo[8]; + struct radeon_bo *cb_color_tile_bo[8]; + u32 cb_color_info[8]; + u32 cb_color_size_idx[8]; + u32 cb_target_mask; + u32 cb_shader_mask; + u32 cb_color_size[8]; + u32 vgt_strmout_en; + u32 vgt_strmout_buffer_en; + u32 db_depth_control; + u32 db_depth_info; + u32 db_depth_size_idx; + u32 db_depth_view; + u32 db_depth_size; + u32 db_offset; + struct radeon_bo *db_bo; + u64 db_bo_mc; +}; + +#define FMT_8_BIT(fmt, vc) [fmt] = { 1, 1, 1, vc, CHIP_R600 } +#define FMT_16_BIT(fmt, vc) [fmt] = { 1, 1, 2, vc, CHIP_R600 } +#define FMT_24_BIT(fmt) [fmt] = { 1, 1, 3, 0, CHIP_R600 } +#define FMT_32_BIT(fmt, vc) [fmt] = { 1, 1, 4, vc, CHIP_R600 } +#define FMT_48_BIT(fmt) [fmt] = { 1, 1, 6, 0, CHIP_R600 } +#define FMT_64_BIT(fmt, vc) [fmt] = { 1, 1, 8, vc, CHIP_R600 } +#define FMT_96_BIT(fmt) [fmt] = { 1, 1, 12, 0, CHIP_R600 } +#define FMT_128_BIT(fmt, vc) [fmt] = { 1, 1, 16,vc, CHIP_R600 } + +struct gpu_formats { + unsigned blockwidth; + unsigned blockheight; + unsigned blocksize; + unsigned valid_color; + enum radeon_family min_family; +}; + +static const struct gpu_formats color_formats_table[] = { + /* 8 bit */ + FMT_8_BIT(V_038004_COLOR_8, 1), + FMT_8_BIT(V_038004_COLOR_4_4, 1), + FMT_8_BIT(V_038004_COLOR_3_3_2, 1), + FMT_8_BIT(V_038004_FMT_1, 0), + + /* 16-bit */ + FMT_16_BIT(V_038004_COLOR_16, 1), + FMT_16_BIT(V_038004_COLOR_16_FLOAT, 1), + FMT_16_BIT(V_038004_COLOR_8_8, 1), + FMT_16_BIT(V_038004_COLOR_5_6_5, 1), + FMT_16_BIT(V_038004_COLOR_6_5_5, 1), + FMT_16_BIT(V_038004_COLOR_1_5_5_5, 1), + FMT_16_BIT(V_038004_COLOR_4_4_4_4, 1), + FMT_16_BIT(V_038004_COLOR_5_5_5_1, 1), + + /* 24-bit */ + FMT_24_BIT(V_038004_FMT_8_8_8), + + /* 32-bit */ + FMT_32_BIT(V_038004_COLOR_32, 1), + FMT_32_BIT(V_038004_COLOR_32_FLOAT, 1), + FMT_32_BIT(V_038004_COLOR_16_16, 1), + FMT_32_BIT(V_038004_COLOR_16_16_FLOAT, 1), + FMT_32_BIT(V_038004_COLOR_8_24, 1), + FMT_32_BIT(V_038004_COLOR_8_24_FLOAT, 1), + FMT_32_BIT(V_038004_COLOR_24_8, 1), + FMT_32_BIT(V_038004_COLOR_24_8_FLOAT, 1), + FMT_32_BIT(V_038004_COLOR_10_11_11, 1), + FMT_32_BIT(V_038004_COLOR_10_11_11_FLOAT, 1), + FMT_32_BIT(V_038004_COLOR_11_11_10, 1), + FMT_32_BIT(V_038004_COLOR_11_11_10_FLOAT, 1), + FMT_32_BIT(V_038004_COLOR_2_10_10_10, 1), + FMT_32_BIT(V_038004_COLOR_8_8_8_8, 1), + FMT_32_BIT(V_038004_COLOR_10_10_10_2, 1), + FMT_32_BIT(V_038004_FMT_5_9_9_9_SHAREDEXP, 0), + FMT_32_BIT(V_038004_FMT_32_AS_8, 0), + FMT_32_BIT(V_038004_FMT_32_AS_8_8, 0), + + /* 48-bit */ + FMT_48_BIT(V_038004_FMT_16_16_16), + FMT_48_BIT(V_038004_FMT_16_16_16_FLOAT), + + /* 64-bit */ + FMT_64_BIT(V_038004_COLOR_X24_8_32_FLOAT, 1), + FMT_64_BIT(V_038004_COLOR_32_32, 1), + FMT_64_BIT(V_038004_COLOR_32_32_FLOAT, 1), + FMT_64_BIT(V_038004_COLOR_16_16_16_16, 1), + FMT_64_BIT(V_038004_COLOR_16_16_16_16_FLOAT, 1), + + FMT_96_BIT(V_038004_FMT_32_32_32), + FMT_96_BIT(V_038004_FMT_32_32_32_FLOAT), + + /* 128-bit */ + FMT_128_BIT(V_038004_COLOR_32_32_32_32, 1), + FMT_128_BIT(V_038004_COLOR_32_32_32_32_FLOAT, 1), + + [V_038004_FMT_GB_GR] = { 2, 1, 4, 0 }, + [V_038004_FMT_BG_RG] = { 2, 1, 4, 0 }, + + /* block compressed formats */ + [V_038004_FMT_BC1] = { 4, 4, 8, 0 }, + [V_038004_FMT_BC2] = { 4, 4, 16, 0 }, + [V_038004_FMT_BC3] = { 4, 4, 16, 0 }, + [V_038004_FMT_BC4] = { 4, 4, 8, 0 }, + [V_038004_FMT_BC5] = { 4, 4, 16, 0}, + [V_038004_FMT_BC6] = { 4, 4, 16, 0, CHIP_CEDAR}, /* Evergreen-only */ + [V_038004_FMT_BC7] = { 4, 4, 16, 0, CHIP_CEDAR}, /* Evergreen-only */ + + /* The other Evergreen formats */ + [V_038004_FMT_32_AS_32_32_32_32] = { 1, 1, 4, 0, CHIP_CEDAR}, +}; + +static inline bool fmt_is_valid_color(u32 format) +{ + if (format >= ARRAY_SIZE(color_formats_table)) + return false; + + if (color_formats_table[format].valid_color) + return true; + + return false; +} + +static inline bool fmt_is_valid_texture(u32 format, enum radeon_family family) +{ + if (format >= ARRAY_SIZE(color_formats_table)) + return false; + + if (family < color_formats_table[format].min_family) + return false; + + if (color_formats_table[format].blockwidth > 0) + return true; + + return false; +} + +static inline int fmt_get_blocksize(u32 format) +{ + if (format >= ARRAY_SIZE(color_formats_table)) + return 0; + + return color_formats_table[format].blocksize; +} + +static inline int fmt_get_nblocksx(u32 format, u32 w) +{ + unsigned bw; + + if (format >= ARRAY_SIZE(color_formats_table)) + return 0; + + bw = color_formats_table[format].blockwidth; + if (bw == 0) + return 0; + + return (w + bw - 1) / bw; +} + +static inline int fmt_get_nblocksy(u32 format, u32 h) +{ + unsigned bh; + + if (format >= ARRAY_SIZE(color_formats_table)) + return 0; + + bh = color_formats_table[format].blockheight; + if (bh == 0) + return 0; + + return (h + bh - 1) / bh; +} + +static inline int r600_bpe_from_format(u32 *bpe, u32 format) +{ + unsigned res; + + if (format >= ARRAY_SIZE(color_formats_table)) + goto fail; + + res = color_formats_table[format].blocksize; + if (res == 0) + goto fail; + + *bpe = res; + return 0; + +fail: + *bpe = 16; + return -EINVAL; +} + +struct array_mode_checker { + int array_mode; + u32 group_size; + u32 nbanks; + u32 npipes; + u32 nsamples; + u32 blocksize; +}; + +/* returns alignment in pixels for pitch/height/depth and bytes for base */ +static inline int r600_get_array_mode_alignment(struct array_mode_checker *values, + u32 *pitch_align, + u32 *height_align, + u32 *depth_align, + u64 *base_align) +{ + u32 tile_width = 8; + u32 tile_height = 8; + u32 macro_tile_width = values->nbanks; + u32 macro_tile_height = values->npipes; + u32 tile_bytes = tile_width * tile_height * values->blocksize * values->nsamples; + u32 macro_tile_bytes = macro_tile_width * macro_tile_height * tile_bytes; + + switch (values->array_mode) { + case ARRAY_LINEAR_GENERAL: + /* technically tile_width/_height for pitch/height */ + *pitch_align = 1; /* tile_width */ + *height_align = 1; /* tile_height */ + *depth_align = 1; + *base_align = 1; + break; + case ARRAY_LINEAR_ALIGNED: + *pitch_align = max((u32)64, (u32)(values->group_size / values->blocksize)); + *height_align = tile_height; + *depth_align = 1; + *base_align = values->group_size; + break; + case ARRAY_1D_TILED_THIN1: + *pitch_align = max((u32)tile_width, + (u32)(values->group_size / + (tile_height * values->blocksize * values->nsamples))); + *height_align = tile_height; + *depth_align = 1; + *base_align = values->group_size; + break; + case ARRAY_2D_TILED_THIN1: + *pitch_align = max((u32)macro_tile_width, + (u32)(((values->group_size / tile_height) / + (values->blocksize * values->nsamples)) * + values->nbanks)) * tile_width; + *height_align = macro_tile_height * tile_height; + *depth_align = 1; + *base_align = max(macro_tile_bytes, + (*pitch_align) * values->blocksize * (*height_align) * values->nsamples); + break; + default: + return -EINVAL; + } + + return 0; +} + +static void r600_cs_track_init(struct r600_cs_track *track) +{ + int i; + + /* assume DX9 mode */ + track->sq_config = DX9_CONSTS; + for (i = 0; i < 8; i++) { + track->cb_color_base_last[i] = 0; + track->cb_color_size[i] = 0; + track->cb_color_size_idx[i] = 0; + track->cb_color_info[i] = 0; + track->cb_color_bo[i] = NULL; + track->cb_color_bo_offset[i] = 0xFFFFFFFF; + track->cb_color_bo_mc[i] = 0xFFFFFFFF; + } + track->cb_target_mask = 0xFFFFFFFF; + track->cb_shader_mask = 0xFFFFFFFF; + track->db_bo = NULL; + track->db_bo_mc = 0xFFFFFFFF; + /* assume the biggest format and that htile is enabled */ + track->db_depth_info = 7 | (1 << 25); + track->db_depth_view = 0xFFFFC000; + track->db_depth_size = 0xFFFFFFFF; + track->db_depth_size_idx = 0; + track->db_depth_control = 0xFFFFFFFF; +} + +static inline int r600_cs_track_validate_cb(struct radeon_cs_parser *p, int i) +{ + struct r600_cs_track *track = p->track; + u32 slice_tile_max, size, tmp; + u32 height, height_align, pitch, pitch_align, depth_align; + u64 base_offset, base_align; + struct array_mode_checker array_check; + volatile u32 *ib = p->ib->ptr; + unsigned array_mode; + u32 format; + if (G_0280A0_TILE_MODE(track->cb_color_info[i])) { + dev_warn(p->dev, "FMASK or CMASK buffer are not supported by this kernel\n"); + return -EINVAL; + } + size = radeon_bo_size(track->cb_color_bo[i]) - track->cb_color_bo_offset[i]; + format = G_0280A0_FORMAT(track->cb_color_info[i]); + if (!fmt_is_valid_color(format)) { + dev_warn(p->dev, "%s:%d cb invalid format %d for %d (0x%08X)\n", + __func__, __LINE__, format, + i, track->cb_color_info[i]); + return -EINVAL; + } + /* pitch in pixels */ + pitch = (G_028060_PITCH_TILE_MAX(track->cb_color_size[i]) + 1) * 8; + slice_tile_max = G_028060_SLICE_TILE_MAX(track->cb_color_size[i]) + 1; + slice_tile_max *= 64; + height = slice_tile_max / pitch; + if (height > 8192) + height = 8192; + array_mode = G_0280A0_ARRAY_MODE(track->cb_color_info[i]); + + base_offset = track->cb_color_bo_mc[i] + track->cb_color_bo_offset[i]; + array_check.array_mode = array_mode; + array_check.group_size = track->group_size; + array_check.nbanks = track->nbanks; + array_check.npipes = track->npipes; + array_check.nsamples = track->nsamples; + array_check.blocksize = fmt_get_blocksize(format); + if (r600_get_array_mode_alignment(&array_check, + &pitch_align, &height_align, &depth_align, &base_align)) { + dev_warn(p->dev, "%s invalid tiling %d for %d (0x%08X)\n", __func__, + G_0280A0_ARRAY_MODE(track->cb_color_info[i]), i, + track->cb_color_info[i]); + return -EINVAL; + } + switch (array_mode) { + case V_0280A0_ARRAY_LINEAR_GENERAL: + break; + case V_0280A0_ARRAY_LINEAR_ALIGNED: + break; + case V_0280A0_ARRAY_1D_TILED_THIN1: + /* avoid breaking userspace */ + if (height > 7) + height &= ~0x7; + break; + case V_0280A0_ARRAY_2D_TILED_THIN1: + break; + default: + dev_warn(p->dev, "%s invalid tiling %d for %d (0x%08X)\n", __func__, + G_0280A0_ARRAY_MODE(track->cb_color_info[i]), i, + track->cb_color_info[i]); + return -EINVAL; + } + + if (!IS_ALIGNED(pitch, pitch_align)) { + dev_warn(p->dev, "%s:%d cb pitch (%d, 0x%x, %d) invalid\n", + __func__, __LINE__, pitch, pitch_align, array_mode); + return -EINVAL; + } + if (!IS_ALIGNED(height, height_align)) { + dev_warn(p->dev, "%s:%d cb height (%d, 0x%x, %d) invalid\n", + __func__, __LINE__, height, height_align, array_mode); + return -EINVAL; + } + if (!IS_ALIGNED(base_offset, base_align)) { + dev_warn(p->dev, "%s offset[%d] 0x%llx 0x%llx, %d not aligned\n", __func__, i, + base_offset, base_align, array_mode); + return -EINVAL; + } + + /* check offset */ + tmp = fmt_get_nblocksy(format, height) * fmt_get_nblocksx(format, pitch) * fmt_get_blocksize(format); + if ((tmp + track->cb_color_bo_offset[i]) > radeon_bo_size(track->cb_color_bo[i])) { + if (array_mode == V_0280A0_ARRAY_LINEAR_GENERAL) { + /* the initial DDX does bad things with the CB size occasionally */ + /* it rounds up height too far for slice tile max but the BO is smaller */ + /* r600c,g also seem to flush at bad times in some apps resulting in + * bogus values here. So for linear just allow anything to avoid breaking + * broken userspace. + */ + } else { + dev_warn(p->dev, "%s offset[%d] %d %d %d %lu too big\n", __func__, i, + array_mode, + track->cb_color_bo_offset[i], tmp, + radeon_bo_size(track->cb_color_bo[i])); + return -EINVAL; + } + } + /* limit max tile */ + tmp = (height * pitch) >> 6; + if (tmp < slice_tile_max) + slice_tile_max = tmp; + tmp = S_028060_PITCH_TILE_MAX((pitch / 8) - 1) | + S_028060_SLICE_TILE_MAX(slice_tile_max - 1); + ib[track->cb_color_size_idx[i]] = tmp; + return 0; +} + +static int r600_cs_track_check(struct radeon_cs_parser *p) +{ + struct r600_cs_track *track = p->track; + u32 tmp; + int r, i; + volatile u32 *ib = p->ib->ptr; + + /* on legacy kernel we don't perform advanced check */ + if (p->rdev == NULL) + return 0; + /* we don't support out buffer yet */ + if (track->vgt_strmout_en || track->vgt_strmout_buffer_en) { + dev_warn(p->dev, "this kernel doesn't support SMX output buffer\n"); + return -EINVAL; + } + /* check that we have a cb for each enabled target, we don't check + * shader_mask because it seems mesa isn't always setting it :( + */ + tmp = track->cb_target_mask; + for (i = 0; i < 8; i++) { + if ((tmp >> (i * 4)) & 0xF) { + /* at least one component is enabled */ + if (track->cb_color_bo[i] == NULL) { + dev_warn(p->dev, "%s:%d mask 0x%08X | 0x%08X no cb for %d\n", + __func__, __LINE__, track->cb_target_mask, track->cb_shader_mask, i); + return -EINVAL; + } + /* perform rewrite of CB_COLOR[0-7]_SIZE */ + r = r600_cs_track_validate_cb(p, i); + if (r) + return r; + } + } + /* Check depth buffer */ + if (G_028800_STENCIL_ENABLE(track->db_depth_control) || + G_028800_Z_ENABLE(track->db_depth_control)) { + u32 nviews, bpe, ntiles, size, slice_tile_max; + u32 height, height_align, pitch, pitch_align, depth_align; + u64 base_offset, base_align; + struct array_mode_checker array_check; + int array_mode; + + if (track->db_bo == NULL) { + dev_warn(p->dev, "z/stencil with no depth buffer\n"); + return -EINVAL; + } + if (G_028010_TILE_SURFACE_ENABLE(track->db_depth_info)) { + dev_warn(p->dev, "this kernel doesn't support z/stencil htile\n"); + return -EINVAL; + } + switch (G_028010_FORMAT(track->db_depth_info)) { + case V_028010_DEPTH_16: + bpe = 2; + break; + case V_028010_DEPTH_X8_24: + case V_028010_DEPTH_8_24: + case V_028010_DEPTH_X8_24_FLOAT: + case V_028010_DEPTH_8_24_FLOAT: + case V_028010_DEPTH_32_FLOAT: + bpe = 4; + break; + case V_028010_DEPTH_X24_8_32_FLOAT: + bpe = 8; + break; + default: + dev_warn(p->dev, "z/stencil with invalid format %d\n", G_028010_FORMAT(track->db_depth_info)); + return -EINVAL; + } + if ((track->db_depth_size & 0xFFFFFC00) == 0xFFFFFC00) { + if (!track->db_depth_size_idx) { + dev_warn(p->dev, "z/stencil buffer size not set\n"); + return -EINVAL; + } + tmp = radeon_bo_size(track->db_bo) - track->db_offset; + tmp = (tmp / bpe) >> 6; + if (!tmp) { + dev_warn(p->dev, "z/stencil buffer too small (0x%08X %d %d %ld)\n", + track->db_depth_size, bpe, track->db_offset, + radeon_bo_size(track->db_bo)); + return -EINVAL; + } + ib[track->db_depth_size_idx] = S_028000_SLICE_TILE_MAX(tmp - 1) | (track->db_depth_size & 0x3FF); + } else { + size = radeon_bo_size(track->db_bo); + /* pitch in pixels */ + pitch = (G_028000_PITCH_TILE_MAX(track->db_depth_size) + 1) * 8; + slice_tile_max = G_028000_SLICE_TILE_MAX(track->db_depth_size) + 1; + slice_tile_max *= 64; + height = slice_tile_max / pitch; + if (height > 8192) + height = 8192; + base_offset = track->db_bo_mc + track->db_offset; + array_mode = G_028010_ARRAY_MODE(track->db_depth_info); + array_check.array_mode = array_mode; + array_check.group_size = track->group_size; + array_check.nbanks = track->nbanks; + array_check.npipes = track->npipes; + array_check.nsamples = track->nsamples; + array_check.blocksize = bpe; + if (r600_get_array_mode_alignment(&array_check, + &pitch_align, &height_align, &depth_align, &base_align)) { + dev_warn(p->dev, "%s invalid tiling %d (0x%08X)\n", __func__, + G_028010_ARRAY_MODE(track->db_depth_info), + track->db_depth_info); + return -EINVAL; + } + switch (array_mode) { + case V_028010_ARRAY_1D_TILED_THIN1: + /* don't break userspace */ + height &= ~0x7; + break; + case V_028010_ARRAY_2D_TILED_THIN1: + break; + default: + dev_warn(p->dev, "%s invalid tiling %d (0x%08X)\n", __func__, + G_028010_ARRAY_MODE(track->db_depth_info), + track->db_depth_info); + return -EINVAL; + } + + if (!IS_ALIGNED(pitch, pitch_align)) { + dev_warn(p->dev, "%s:%d db pitch (%d, 0x%x, %d) invalid\n", + __func__, __LINE__, pitch, pitch_align, array_mode); + return -EINVAL; + } + if (!IS_ALIGNED(height, height_align)) { + dev_warn(p->dev, "%s:%d db height (%d, 0x%x, %d) invalid\n", + __func__, __LINE__, height, height_align, array_mode); + return -EINVAL; + } + if (!IS_ALIGNED(base_offset, base_align)) { + dev_warn(p->dev, "%s offset[%d] 0x%llx, 0x%llx, %d not aligned\n", __func__, i, + base_offset, base_align, array_mode); + return -EINVAL; + } + + ntiles = G_028000_SLICE_TILE_MAX(track->db_depth_size) + 1; + nviews = G_028004_SLICE_MAX(track->db_depth_view) + 1; + tmp = ntiles * bpe * 64 * nviews; + if ((tmp + track->db_offset) > radeon_bo_size(track->db_bo)) { + dev_warn(p->dev, "z/stencil buffer (%d) too small (0x%08X %d %d %d -> %u have %lu)\n", + array_mode, + track->db_depth_size, ntiles, nviews, bpe, tmp + track->db_offset, + radeon_bo_size(track->db_bo)); + return -EINVAL; + } + } + } + return 0; +} + +/** + * r600_cs_packet_parse() - parse cp packet and point ib index to next packet + * @parser: parser structure holding parsing context. + * @pkt: where to store packet informations + * + * Assume that chunk_ib_index is properly set. Will return -EINVAL + * if packet is bigger than remaining ib size. or if packets is unknown. + **/ +int r600_cs_packet_parse(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt, + unsigned idx) +{ + struct radeon_cs_chunk *ib_chunk = &p->chunks[p->chunk_ib_idx]; + uint32_t header; + + if (idx >= ib_chunk->length_dw) { + DRM_ERROR("Can not parse packet at %d after CS end %d !\n", + idx, ib_chunk->length_dw); + return -EINVAL; + } + header = radeon_get_ib_value(p, idx); + pkt->idx = idx; + pkt->type = CP_PACKET_GET_TYPE(header); + pkt->count = CP_PACKET_GET_COUNT(header); + pkt->one_reg_wr = 0; + switch (pkt->type) { + case PACKET_TYPE0: + pkt->reg = CP_PACKET0_GET_REG(header); + break; + case PACKET_TYPE3: + pkt->opcode = CP_PACKET3_GET_OPCODE(header); + break; + case PACKET_TYPE2: + pkt->count = -1; + break; + default: + DRM_ERROR("Unknown packet type %d at %d !\n", pkt->type, idx); + return -EINVAL; + } + if ((pkt->count + 1 + pkt->idx) >= ib_chunk->length_dw) { + DRM_ERROR("Packet (%d:%d:%d) end after CS buffer (%d) !\n", + pkt->idx, pkt->type, pkt->count, ib_chunk->length_dw); + return -EINVAL; + } + return 0; +} + +/** + * r600_cs_packet_next_reloc_mm() - parse next packet which should be reloc packet3 + * @parser: parser structure holding parsing context. + * @data: pointer to relocation data + * @offset_start: starting offset + * @offset_mask: offset mask (to align start offset on) + * @reloc: reloc informations + * + * Check next packet is relocation packet3, do bo validation and compute + * GPU offset using the provided start. + **/ +static int r600_cs_packet_next_reloc_mm(struct radeon_cs_parser *p, + struct radeon_cs_reloc **cs_reloc) +{ + struct radeon_cs_chunk *relocs_chunk; + struct radeon_cs_packet p3reloc; + unsigned idx; + int r; + + if (p->chunk_relocs_idx == -1) { + DRM_ERROR("No relocation chunk !\n"); + return -EINVAL; + } + *cs_reloc = NULL; + relocs_chunk = &p->chunks[p->chunk_relocs_idx]; + r = r600_cs_packet_parse(p, &p3reloc, p->idx); + if (r) { + return r; + } + p->idx += p3reloc.count + 2; + if (p3reloc.type != PACKET_TYPE3 || p3reloc.opcode != PACKET3_NOP) { + DRM_ERROR("No packet3 for relocation for packet at %d.\n", + p3reloc.idx); + return -EINVAL; + } + idx = radeon_get_ib_value(p, p3reloc.idx + 1); + if (idx >= relocs_chunk->length_dw) { + DRM_ERROR("Relocs at %d after relocations chunk end %d !\n", + idx, relocs_chunk->length_dw); + return -EINVAL; + } + /* FIXME: we assume reloc size is 4 dwords */ + *cs_reloc = p->relocs_ptr[(idx / 4)]; + return 0; +} + +/** + * r600_cs_packet_next_reloc_nomm() - parse next packet which should be reloc packet3 + * @parser: parser structure holding parsing context. + * @data: pointer to relocation data + * @offset_start: starting offset + * @offset_mask: offset mask (to align start offset on) + * @reloc: reloc informations + * + * Check next packet is relocation packet3, do bo validation and compute + * GPU offset using the provided start. + **/ +static int r600_cs_packet_next_reloc_nomm(struct radeon_cs_parser *p, + struct radeon_cs_reloc **cs_reloc) +{ + struct radeon_cs_chunk *relocs_chunk; + struct radeon_cs_packet p3reloc; + unsigned idx; + int r; + + if (p->chunk_relocs_idx == -1) { + DRM_ERROR("No relocation chunk !\n"); + return -EINVAL; + } + *cs_reloc = NULL; + relocs_chunk = &p->chunks[p->chunk_relocs_idx]; + r = r600_cs_packet_parse(p, &p3reloc, p->idx); + if (r) { + return r; + } + p->idx += p3reloc.count + 2; + if (p3reloc.type != PACKET_TYPE3 || p3reloc.opcode != PACKET3_NOP) { + DRM_ERROR("No packet3 for relocation for packet at %d.\n", + p3reloc.idx); + return -EINVAL; + } + idx = radeon_get_ib_value(p, p3reloc.idx + 1); + if (idx >= relocs_chunk->length_dw) { + DRM_ERROR("Relocs at %d after relocations chunk end %d !\n", + idx, relocs_chunk->length_dw); + return -EINVAL; + } + *cs_reloc = p->relocs; + (*cs_reloc)->lobj.gpu_offset = (u64)relocs_chunk->kdata[idx + 3] << 32; + (*cs_reloc)->lobj.gpu_offset |= relocs_chunk->kdata[idx + 0]; + return 0; +} + +/** + * r600_cs_packet_next_is_pkt3_nop() - test if next packet is packet3 nop for reloc + * @parser: parser structure holding parsing context. + * + * Check next packet is relocation packet3, do bo validation and compute + * GPU offset using the provided start. + **/ +static inline int r600_cs_packet_next_is_pkt3_nop(struct radeon_cs_parser *p) +{ + struct radeon_cs_packet p3reloc; + int r; + + r = r600_cs_packet_parse(p, &p3reloc, p->idx); + if (r) { + return 0; + } + if (p3reloc.type != PACKET_TYPE3 || p3reloc.opcode != PACKET3_NOP) { + return 0; + } + return 1; +} + +/** + * r600_cs_packet_next_vline() - parse userspace VLINE packet + * @parser: parser structure holding parsing context. + * + * Userspace sends a special sequence for VLINE waits. + * PACKET0 - VLINE_START_END + value + * PACKET3 - WAIT_REG_MEM poll vline status reg + * RELOC (P3) - crtc_id in reloc. + * + * This function parses this and relocates the VLINE START END + * and WAIT_REG_MEM packets to the correct crtc. + * It also detects a switched off crtc and nulls out the + * wait in that case. + */ +static int r600_cs_packet_parse_vline(struct radeon_cs_parser *p) +{ + struct drm_mode_object *obj; + struct drm_crtc *crtc; + struct radeon_crtc *radeon_crtc; + struct radeon_cs_packet p3reloc, wait_reg_mem; + int crtc_id; + int r; + uint32_t header, h_idx, reg, wait_reg_mem_info; + volatile uint32_t *ib; + + ib = p->ib->ptr; + + /* parse the WAIT_REG_MEM */ + r = r600_cs_packet_parse(p, &wait_reg_mem, p->idx); + if (r) + return r; + + /* check its a WAIT_REG_MEM */ + if (wait_reg_mem.type != PACKET_TYPE3 || + wait_reg_mem.opcode != PACKET3_WAIT_REG_MEM) { + DRM_ERROR("vline wait missing WAIT_REG_MEM segment\n"); + return -EINVAL; + } + + wait_reg_mem_info = radeon_get_ib_value(p, wait_reg_mem.idx + 1); + /* bit 4 is reg (0) or mem (1) */ + if (wait_reg_mem_info & 0x10) { + DRM_ERROR("vline WAIT_REG_MEM waiting on MEM rather than REG\n"); + return -EINVAL; + } + /* waiting for value to be equal */ + if ((wait_reg_mem_info & 0x7) != 0x3) { + DRM_ERROR("vline WAIT_REG_MEM function not equal\n"); + return -EINVAL; + } + if ((radeon_get_ib_value(p, wait_reg_mem.idx + 2) << 2) != AVIVO_D1MODE_VLINE_STATUS) { + DRM_ERROR("vline WAIT_REG_MEM bad reg\n"); + return -EINVAL; + } + + if (radeon_get_ib_value(p, wait_reg_mem.idx + 5) != AVIVO_D1MODE_VLINE_STAT) { + DRM_ERROR("vline WAIT_REG_MEM bad bit mask\n"); + return -EINVAL; + } + + /* jump over the NOP */ + r = r600_cs_packet_parse(p, &p3reloc, p->idx + wait_reg_mem.count + 2); + if (r) + return r; + + h_idx = p->idx - 2; + p->idx += wait_reg_mem.count + 2; + p->idx += p3reloc.count + 2; + + header = radeon_get_ib_value(p, h_idx); + crtc_id = radeon_get_ib_value(p, h_idx + 2 + 7 + 1); + reg = CP_PACKET0_GET_REG(header); + + obj = drm_mode_object_find(p->rdev->ddev, crtc_id, DRM_MODE_OBJECT_CRTC); + if (!obj) { + DRM_ERROR("cannot find crtc %d\n", crtc_id); + return -EINVAL; + } + crtc = obj_to_crtc(obj); + radeon_crtc = to_radeon_crtc(crtc); + crtc_id = radeon_crtc->crtc_id; + + if (!crtc->enabled) { + /* if the CRTC isn't enabled - we need to nop out the WAIT_REG_MEM */ + ib[h_idx + 2] = PACKET2(0); + ib[h_idx + 3] = PACKET2(0); + ib[h_idx + 4] = PACKET2(0); + ib[h_idx + 5] = PACKET2(0); + ib[h_idx + 6] = PACKET2(0); + ib[h_idx + 7] = PACKET2(0); + ib[h_idx + 8] = PACKET2(0); + } else if (crtc_id == 1) { + switch (reg) { + case AVIVO_D1MODE_VLINE_START_END: + header &= ~R600_CP_PACKET0_REG_MASK; + header |= AVIVO_D2MODE_VLINE_START_END >> 2; + break; + default: + DRM_ERROR("unknown crtc reloc\n"); + return -EINVAL; + } + ib[h_idx] = header; + ib[h_idx + 4] = AVIVO_D2MODE_VLINE_STATUS >> 2; + } + + return 0; +} + +static int r600_packet0_check(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt, + unsigned idx, unsigned reg) +{ + int r; + + switch (reg) { + case AVIVO_D1MODE_VLINE_START_END: + r = r600_cs_packet_parse_vline(p); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + return r; + } + break; + default: + printk(KERN_ERR "Forbidden register 0x%04X in cs at %d\n", + reg, idx); + return -EINVAL; + } + return 0; +} + +static int r600_cs_parse_packet0(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt) +{ + unsigned reg, i; + unsigned idx; + int r; + + idx = pkt->idx + 1; + reg = pkt->reg; + for (i = 0; i <= pkt->count; i++, idx++, reg += 4) { + r = r600_packet0_check(p, pkt, idx, reg); + if (r) { + return r; + } + } + return 0; +} + +/** + * r600_cs_check_reg() - check if register is authorized or not + * @parser: parser structure holding parsing context + * @reg: register we are testing + * @idx: index into the cs buffer + * + * This function will test against r600_reg_safe_bm and return 0 + * if register is safe. If register is not flag as safe this function + * will test it against a list of register needind special handling. + */ +static inline int r600_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) +{ + struct r600_cs_track *track = (struct r600_cs_track *)p->track; + struct radeon_cs_reloc *reloc; + u32 last_reg = ARRAY_SIZE(r600_reg_safe_bm); + u32 m, i, tmp, *ib; + int r; + + i = (reg >> 7); + if (i > last_reg) { + dev_warn(p->dev, "forbidden register 0x%08x at %d\n", reg, idx); + return -EINVAL; + } + m = 1 << ((reg >> 2) & 31); + if (!(r600_reg_safe_bm[i] & m)) + return 0; + ib = p->ib->ptr; + switch (reg) { + /* force following reg to 0 in an attempt to disable out buffer + * which will need us to better understand how it works to perform + * security check on it (Jerome) + */ + case R_0288A8_SQ_ESGS_RING_ITEMSIZE: + case R_008C44_SQ_ESGS_RING_SIZE: + case R_0288B0_SQ_ESTMP_RING_ITEMSIZE: + case R_008C54_SQ_ESTMP_RING_SIZE: + case R_0288C0_SQ_FBUF_RING_ITEMSIZE: + case R_008C74_SQ_FBUF_RING_SIZE: + case R_0288B4_SQ_GSTMP_RING_ITEMSIZE: + case R_008C5C_SQ_GSTMP_RING_SIZE: + case R_0288AC_SQ_GSVS_RING_ITEMSIZE: + case R_008C4C_SQ_GSVS_RING_SIZE: + case R_0288BC_SQ_PSTMP_RING_ITEMSIZE: + case R_008C6C_SQ_PSTMP_RING_SIZE: + case R_0288C4_SQ_REDUC_RING_ITEMSIZE: + case R_008C7C_SQ_REDUC_RING_SIZE: + case R_0288B8_SQ_VSTMP_RING_ITEMSIZE: + case R_008C64_SQ_VSTMP_RING_SIZE: + case R_0288C8_SQ_GS_VERT_ITEMSIZE: + /* get value to populate the IB don't remove */ + tmp =radeon_get_ib_value(p, idx); + ib[idx] = 0; + break; + case SQ_CONFIG: + track->sq_config = radeon_get_ib_value(p, idx); + break; + case R_028800_DB_DEPTH_CONTROL: + track->db_depth_control = radeon_get_ib_value(p, idx); + break; + case R_028010_DB_DEPTH_INFO: + if (r600_cs_packet_next_is_pkt3_nop(p)) { + r = r600_cs_packet_next_reloc(p, &reloc); + if (r) { + dev_warn(p->dev, "bad SET_CONTEXT_REG " + "0x%04X\n", reg); + return -EINVAL; + } + track->db_depth_info = radeon_get_ib_value(p, idx); + ib[idx] &= C_028010_ARRAY_MODE; + track->db_depth_info &= C_028010_ARRAY_MODE; + if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) { + ib[idx] |= S_028010_ARRAY_MODE(V_028010_ARRAY_2D_TILED_THIN1); + track->db_depth_info |= S_028010_ARRAY_MODE(V_028010_ARRAY_2D_TILED_THIN1); + } else { + ib[idx] |= S_028010_ARRAY_MODE(V_028010_ARRAY_1D_TILED_THIN1); + track->db_depth_info |= S_028010_ARRAY_MODE(V_028010_ARRAY_1D_TILED_THIN1); + } + } else + track->db_depth_info = radeon_get_ib_value(p, idx); + break; + case R_028004_DB_DEPTH_VIEW: + track->db_depth_view = radeon_get_ib_value(p, idx); + break; + case R_028000_DB_DEPTH_SIZE: + track->db_depth_size = radeon_get_ib_value(p, idx); + track->db_depth_size_idx = idx; + break; + case R_028AB0_VGT_STRMOUT_EN: + track->vgt_strmout_en = radeon_get_ib_value(p, idx); + break; + case R_028B20_VGT_STRMOUT_BUFFER_EN: + track->vgt_strmout_buffer_en = radeon_get_ib_value(p, idx); + break; + case R_028238_CB_TARGET_MASK: + track->cb_target_mask = radeon_get_ib_value(p, idx); + break; + case R_02823C_CB_SHADER_MASK: + track->cb_shader_mask = radeon_get_ib_value(p, idx); + break; + case R_028C04_PA_SC_AA_CONFIG: + tmp = G_028C04_MSAA_NUM_SAMPLES(radeon_get_ib_value(p, idx)); + track->nsamples = 1 << tmp; + break; + case R_0280A0_CB_COLOR0_INFO: + case R_0280A4_CB_COLOR1_INFO: + case R_0280A8_CB_COLOR2_INFO: + case R_0280AC_CB_COLOR3_INFO: + case R_0280B0_CB_COLOR4_INFO: + case R_0280B4_CB_COLOR5_INFO: + case R_0280B8_CB_COLOR6_INFO: + case R_0280BC_CB_COLOR7_INFO: + if (r600_cs_packet_next_is_pkt3_nop(p)) { + r = r600_cs_packet_next_reloc(p, &reloc); + if (r) { + dev_err(p->dev, "bad SET_CONTEXT_REG 0x%04X\n", reg); + return -EINVAL; + } + tmp = (reg - R_0280A0_CB_COLOR0_INFO) / 4; + track->cb_color_info[tmp] = radeon_get_ib_value(p, idx); + if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) { + ib[idx] |= S_0280A0_ARRAY_MODE(V_0280A0_ARRAY_2D_TILED_THIN1); + track->cb_color_info[tmp] |= S_0280A0_ARRAY_MODE(V_0280A0_ARRAY_2D_TILED_THIN1); + } else if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) { + ib[idx] |= S_0280A0_ARRAY_MODE(V_0280A0_ARRAY_1D_TILED_THIN1); + track->cb_color_info[tmp] |= S_0280A0_ARRAY_MODE(V_0280A0_ARRAY_1D_TILED_THIN1); + } + } else { + tmp = (reg - R_0280A0_CB_COLOR0_INFO) / 4; + track->cb_color_info[tmp] = radeon_get_ib_value(p, idx); + } + break; + case R_028060_CB_COLOR0_SIZE: + case R_028064_CB_COLOR1_SIZE: + case R_028068_CB_COLOR2_SIZE: + case R_02806C_CB_COLOR3_SIZE: + case R_028070_CB_COLOR4_SIZE: + case R_028074_CB_COLOR5_SIZE: + case R_028078_CB_COLOR6_SIZE: + case R_02807C_CB_COLOR7_SIZE: + tmp = (reg - R_028060_CB_COLOR0_SIZE) / 4; + track->cb_color_size[tmp] = radeon_get_ib_value(p, idx); + track->cb_color_size_idx[tmp] = idx; + break; + /* This register were added late, there is userspace + * which does provide relocation for those but set + * 0 offset. In order to avoid breaking old userspace + * we detect this and set address to point to last + * CB_COLOR0_BASE, note that if userspace doesn't set + * CB_COLOR0_BASE before this register we will report + * error. Old userspace always set CB_COLOR0_BASE + * before any of this. + */ + case R_0280E0_CB_COLOR0_FRAG: + case R_0280E4_CB_COLOR1_FRAG: + case R_0280E8_CB_COLOR2_FRAG: + case R_0280EC_CB_COLOR3_FRAG: + case R_0280F0_CB_COLOR4_FRAG: + case R_0280F4_CB_COLOR5_FRAG: + case R_0280F8_CB_COLOR6_FRAG: + case R_0280FC_CB_COLOR7_FRAG: + tmp = (reg - R_0280E0_CB_COLOR0_FRAG) / 4; + if (!r600_cs_packet_next_is_pkt3_nop(p)) { + if (!track->cb_color_base_last[tmp]) { + dev_err(p->dev, "Broken old userspace ? no cb_color0_base supplied before trying to write 0x%08X\n", reg); + return -EINVAL; + } + ib[idx] = track->cb_color_base_last[tmp]; + track->cb_color_frag_bo[tmp] = track->cb_color_bo[tmp]; + } else { + r = r600_cs_packet_next_reloc(p, &reloc); + if (r) { + dev_err(p->dev, "bad SET_CONTEXT_REG 0x%04X\n", reg); + return -EINVAL; + } + ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + track->cb_color_frag_bo[tmp] = reloc->robj; + } + break; + case R_0280C0_CB_COLOR0_TILE: + case R_0280C4_CB_COLOR1_TILE: + case R_0280C8_CB_COLOR2_TILE: + case R_0280CC_CB_COLOR3_TILE: + case R_0280D0_CB_COLOR4_TILE: + case R_0280D4_CB_COLOR5_TILE: + case R_0280D8_CB_COLOR6_TILE: + case R_0280DC_CB_COLOR7_TILE: + tmp = (reg - R_0280C0_CB_COLOR0_TILE) / 4; + if (!r600_cs_packet_next_is_pkt3_nop(p)) { + if (!track->cb_color_base_last[tmp]) { + dev_err(p->dev, "Broken old userspace ? no cb_color0_base supplied before trying to write 0x%08X\n", reg); + return -EINVAL; + } + ib[idx] = track->cb_color_base_last[tmp]; + track->cb_color_tile_bo[tmp] = track->cb_color_bo[tmp]; + } else { + r = r600_cs_packet_next_reloc(p, &reloc); + if (r) { + dev_err(p->dev, "bad SET_CONTEXT_REG 0x%04X\n", reg); + return -EINVAL; + } + ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + track->cb_color_tile_bo[tmp] = reloc->robj; + } + break; + case CB_COLOR0_BASE: + case CB_COLOR1_BASE: + case CB_COLOR2_BASE: + case CB_COLOR3_BASE: + case CB_COLOR4_BASE: + case CB_COLOR5_BASE: + case CB_COLOR6_BASE: + case CB_COLOR7_BASE: + r = r600_cs_packet_next_reloc(p, &reloc); + if (r) { + dev_warn(p->dev, "bad SET_CONTEXT_REG " + "0x%04X\n", reg); + return -EINVAL; + } + tmp = (reg - CB_COLOR0_BASE) / 4; + track->cb_color_bo_offset[tmp] = radeon_get_ib_value(p, idx) << 8; + ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + track->cb_color_base_last[tmp] = ib[idx]; + track->cb_color_bo[tmp] = reloc->robj; + track->cb_color_bo_mc[tmp] = reloc->lobj.gpu_offset; + break; + case DB_DEPTH_BASE: + r = r600_cs_packet_next_reloc(p, &reloc); + if (r) { + dev_warn(p->dev, "bad SET_CONTEXT_REG " + "0x%04X\n", reg); + return -EINVAL; + } + track->db_offset = radeon_get_ib_value(p, idx) << 8; + ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + track->db_bo = reloc->robj; + track->db_bo_mc = reloc->lobj.gpu_offset; + break; + case DB_HTILE_DATA_BASE: + case SQ_PGM_START_FS: + case SQ_PGM_START_ES: + case SQ_PGM_START_VS: + case SQ_PGM_START_GS: + case SQ_PGM_START_PS: + case SQ_ALU_CONST_CACHE_GS_0: + case SQ_ALU_CONST_CACHE_GS_1: + case SQ_ALU_CONST_CACHE_GS_2: + case SQ_ALU_CONST_CACHE_GS_3: + case SQ_ALU_CONST_CACHE_GS_4: + case SQ_ALU_CONST_CACHE_GS_5: + case SQ_ALU_CONST_CACHE_GS_6: + case SQ_ALU_CONST_CACHE_GS_7: + case SQ_ALU_CONST_CACHE_GS_8: + case SQ_ALU_CONST_CACHE_GS_9: + case SQ_ALU_CONST_CACHE_GS_10: + case SQ_ALU_CONST_CACHE_GS_11: + case SQ_ALU_CONST_CACHE_GS_12: + case SQ_ALU_CONST_CACHE_GS_13: + case SQ_ALU_CONST_CACHE_GS_14: + case SQ_ALU_CONST_CACHE_GS_15: + case SQ_ALU_CONST_CACHE_PS_0: + case SQ_ALU_CONST_CACHE_PS_1: + case SQ_ALU_CONST_CACHE_PS_2: + case SQ_ALU_CONST_CACHE_PS_3: + case SQ_ALU_CONST_CACHE_PS_4: + case SQ_ALU_CONST_CACHE_PS_5: + case SQ_ALU_CONST_CACHE_PS_6: + case SQ_ALU_CONST_CACHE_PS_7: + case SQ_ALU_CONST_CACHE_PS_8: + case SQ_ALU_CONST_CACHE_PS_9: + case SQ_ALU_CONST_CACHE_PS_10: + case SQ_ALU_CONST_CACHE_PS_11: + case SQ_ALU_CONST_CACHE_PS_12: + case SQ_ALU_CONST_CACHE_PS_13: + case SQ_ALU_CONST_CACHE_PS_14: + case SQ_ALU_CONST_CACHE_PS_15: + case SQ_ALU_CONST_CACHE_VS_0: + case SQ_ALU_CONST_CACHE_VS_1: + case SQ_ALU_CONST_CACHE_VS_2: + case SQ_ALU_CONST_CACHE_VS_3: + case SQ_ALU_CONST_CACHE_VS_4: + case SQ_ALU_CONST_CACHE_VS_5: + case SQ_ALU_CONST_CACHE_VS_6: + case SQ_ALU_CONST_CACHE_VS_7: + case SQ_ALU_CONST_CACHE_VS_8: + case SQ_ALU_CONST_CACHE_VS_9: + case SQ_ALU_CONST_CACHE_VS_10: + case SQ_ALU_CONST_CACHE_VS_11: + case SQ_ALU_CONST_CACHE_VS_12: + case SQ_ALU_CONST_CACHE_VS_13: + case SQ_ALU_CONST_CACHE_VS_14: + case SQ_ALU_CONST_CACHE_VS_15: + r = r600_cs_packet_next_reloc(p, &reloc); + if (r) { + dev_warn(p->dev, "bad SET_CONTEXT_REG " + "0x%04X\n", reg); + return -EINVAL; + } + ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + break; + default: + dev_warn(p->dev, "forbidden register 0x%08x at %d\n", reg, idx); + return -EINVAL; + } + return 0; +} + +static inline unsigned mip_minify(unsigned size, unsigned level) +{ + unsigned val; + + val = max(1U, size >> level); + if (level > 0) + val = roundup_pow_of_two(val); + return val; +} + +static void r600_texture_size(unsigned nfaces, unsigned blevel, unsigned llevel, + unsigned w0, unsigned h0, unsigned d0, unsigned format, + unsigned block_align, unsigned height_align, unsigned base_align, + unsigned *l0_size, unsigned *mipmap_size) +{ + unsigned offset, i, level; + unsigned width, height, depth, size; + unsigned blocksize; + unsigned nbx, nby; + unsigned nlevels = llevel - blevel + 1; + + *l0_size = -1; + blocksize = fmt_get_blocksize(format); + + w0 = mip_minify(w0, 0); + h0 = mip_minify(h0, 0); + d0 = mip_minify(d0, 0); + for(i = 0, offset = 0, level = blevel; i < nlevels; i++, level++) { + width = mip_minify(w0, i); + nbx = fmt_get_nblocksx(format, width); + + nbx = round_up(nbx, block_align); + + height = mip_minify(h0, i); + nby = fmt_get_nblocksy(format, height); + nby = round_up(nby, height_align); + + depth = mip_minify(d0, i); + + size = nbx * nby * blocksize; + if (nfaces) + size *= nfaces; + else + size *= depth; + + if (i == 0) + *l0_size = size; + + if (i == 0 || i == 1) + offset = round_up(offset, base_align); + + offset += size; + } + *mipmap_size = offset; + if (llevel == 0) + *mipmap_size = *l0_size; + if (!blevel) + *mipmap_size -= *l0_size; +} + +/** + * r600_check_texture_resource() - check if register is authorized or not + * @p: parser structure holding parsing context + * @idx: index into the cs buffer + * @texture: texture's bo structure + * @mipmap: mipmap's bo structure + * + * This function will check that the resource has valid field and that + * the texture and mipmap bo object are big enough to cover this resource. + */ +static inline int r600_check_texture_resource(struct radeon_cs_parser *p, u32 idx, + struct radeon_bo *texture, + struct radeon_bo *mipmap, + u64 base_offset, + u64 mip_offset, + u32 tiling_flags) +{ + struct r600_cs_track *track = p->track; + u32 nfaces, llevel, blevel, w0, h0, d0; + u32 word0, word1, l0_size, mipmap_size, word2, word3; + u32 height_align, pitch, pitch_align, depth_align; + u32 array, barray, larray; + u64 base_align; + struct array_mode_checker array_check; + u32 format; + + /* on legacy kernel we don't perform advanced check */ + if (p->rdev == NULL) + return 0; + + /* convert to bytes */ + base_offset <<= 8; + mip_offset <<= 8; + + word0 = radeon_get_ib_value(p, idx + 0); + if (tiling_flags & RADEON_TILING_MACRO) + word0 |= S_038000_TILE_MODE(V_038000_ARRAY_2D_TILED_THIN1); + else if (tiling_flags & RADEON_TILING_MICRO) + word0 |= S_038000_TILE_MODE(V_038000_ARRAY_1D_TILED_THIN1); + word1 = radeon_get_ib_value(p, idx + 1); + w0 = G_038000_TEX_WIDTH(word0) + 1; + h0 = G_038004_TEX_HEIGHT(word1) + 1; + d0 = G_038004_TEX_DEPTH(word1); + nfaces = 1; + switch (G_038000_DIM(word0)) { + case V_038000_SQ_TEX_DIM_1D: + case V_038000_SQ_TEX_DIM_2D: + case V_038000_SQ_TEX_DIM_3D: + break; + case V_038000_SQ_TEX_DIM_CUBEMAP: + if (p->family >= CHIP_RV770) + nfaces = 8; + else + nfaces = 6; + break; + case V_038000_SQ_TEX_DIM_1D_ARRAY: + case V_038000_SQ_TEX_DIM_2D_ARRAY: + array = 1; + break; + case V_038000_SQ_TEX_DIM_2D_MSAA: + case V_038000_SQ_TEX_DIM_2D_ARRAY_MSAA: + default: + dev_warn(p->dev, "this kernel doesn't support %d texture dim\n", G_038000_DIM(word0)); + return -EINVAL; + } + format = G_038004_DATA_FORMAT(word1); + if (!fmt_is_valid_texture(format, p->family)) { + dev_warn(p->dev, "%s:%d texture invalid format %d\n", + __func__, __LINE__, format); + return -EINVAL; + } + + /* pitch in texels */ + pitch = (G_038000_PITCH(word0) + 1) * 8; + array_check.array_mode = G_038000_TILE_MODE(word0); + array_check.group_size = track->group_size; + array_check.nbanks = track->nbanks; + array_check.npipes = track->npipes; + array_check.nsamples = 1; + array_check.blocksize = fmt_get_blocksize(format); + if (r600_get_array_mode_alignment(&array_check, + &pitch_align, &height_align, &depth_align, &base_align)) { + dev_warn(p->dev, "%s:%d tex array mode (%d) invalid\n", + __func__, __LINE__, G_038000_TILE_MODE(word0)); + return -EINVAL; + } + + /* XXX check height as well... */ + + if (!IS_ALIGNED(pitch, pitch_align)) { + dev_warn(p->dev, "%s:%d tex pitch (%d, 0x%x, %d) invalid\n", + __func__, __LINE__, pitch, pitch_align, G_038000_TILE_MODE(word0)); + return -EINVAL; + } + if (!IS_ALIGNED(base_offset, base_align)) { + dev_warn(p->dev, "%s:%d tex base offset (0x%llx, 0x%llx, %d) invalid\n", + __func__, __LINE__, base_offset, base_align, G_038000_TILE_MODE(word0)); + return -EINVAL; + } + if (!IS_ALIGNED(mip_offset, base_align)) { + dev_warn(p->dev, "%s:%d tex mip offset (0x%llx, 0x%llx, %d) invalid\n", + __func__, __LINE__, mip_offset, base_align, G_038000_TILE_MODE(word0)); + return -EINVAL; + } + + word2 = radeon_get_ib_value(p, idx + 2) << 8; + word3 = radeon_get_ib_value(p, idx + 3) << 8; + + word0 = radeon_get_ib_value(p, idx + 4); + word1 = radeon_get_ib_value(p, idx + 5); + blevel = G_038010_BASE_LEVEL(word0); + llevel = G_038014_LAST_LEVEL(word1); + if (array == 1) { + barray = G_038014_BASE_ARRAY(word1); + larray = G_038014_LAST_ARRAY(word1); + + nfaces = larray - barray + 1; + } + r600_texture_size(nfaces, blevel, llevel, w0, h0, d0, format, + pitch_align, height_align, base_align, + &l0_size, &mipmap_size); + /* using get ib will give us the offset into the texture bo */ + if ((l0_size + word2) > radeon_bo_size(texture)) { + dev_warn(p->dev, "texture bo too small (%d %d %d %d -> %d have %ld)\n", + w0, h0, format, word2, l0_size, radeon_bo_size(texture)); + dev_warn(p->dev, "alignments %d %d %d %lld\n", pitch, pitch_align, height_align, base_align); + return -EINVAL; + } + /* using get ib will give us the offset into the mipmap bo */ + word3 = radeon_get_ib_value(p, idx + 3) << 8; + if ((mipmap_size + word3) > radeon_bo_size(mipmap)) { + /*dev_warn(p->dev, "mipmap bo too small (%d %d %d %d %d %d -> %d have %ld)\n", + w0, h0, format, blevel, nlevels, word3, mipmap_size, radeon_bo_size(texture));*/ + } + return 0; +} + +static int r600_packet3_check(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt) +{ + struct radeon_cs_reloc *reloc; + struct r600_cs_track *track; + volatile u32 *ib; + unsigned idx; + unsigned i; + unsigned start_reg, end_reg, reg; + int r; + u32 idx_value; + + track = (struct r600_cs_track *)p->track; + ib = p->ib->ptr; + idx = pkt->idx + 1; + idx_value = radeon_get_ib_value(p, idx); + + switch (pkt->opcode) { + case PACKET3_SET_PREDICATION: + { + int pred_op; + int tmp; + if (pkt->count != 1) { + DRM_ERROR("bad SET PREDICATION\n"); + return -EINVAL; + } + + tmp = radeon_get_ib_value(p, idx + 1); + pred_op = (tmp >> 16) & 0x7; + + /* for the clear predicate operation */ + if (pred_op == 0) + return 0; + + if (pred_op > 2) { + DRM_ERROR("bad SET PREDICATION operation %d\n", pred_op); + return -EINVAL; + } + + r = r600_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("bad SET PREDICATION\n"); + return -EINVAL; + } + + ib[idx + 0] = idx_value + (u32)(reloc->lobj.gpu_offset & 0xffffffff); + ib[idx + 1] = tmp + (upper_32_bits(reloc->lobj.gpu_offset) & 0xff); + } + break; + + case PACKET3_START_3D_CMDBUF: + if (p->family >= CHIP_RV770 || pkt->count) { + DRM_ERROR("bad START_3D\n"); + return -EINVAL; + } + break; + case PACKET3_CONTEXT_CONTROL: + if (pkt->count != 1) { + DRM_ERROR("bad CONTEXT_CONTROL\n"); + return -EINVAL; + } + break; + case PACKET3_INDEX_TYPE: + case PACKET3_NUM_INSTANCES: + if (pkt->count) { + DRM_ERROR("bad INDEX_TYPE/NUM_INSTANCES\n"); + return -EINVAL; + } + break; + case PACKET3_DRAW_INDEX: + if (pkt->count != 3) { + DRM_ERROR("bad DRAW_INDEX\n"); + return -EINVAL; + } + r = r600_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("bad DRAW_INDEX\n"); + return -EINVAL; + } + ib[idx+0] = idx_value + (u32)(reloc->lobj.gpu_offset & 0xffffffff); + ib[idx+1] += upper_32_bits(reloc->lobj.gpu_offset) & 0xff; + r = r600_cs_track_check(p); + if (r) { + dev_warn(p->dev, "%s:%d invalid cmd stream\n", __func__, __LINE__); + return r; + } + break; + case PACKET3_DRAW_INDEX_AUTO: + if (pkt->count != 1) { + DRM_ERROR("bad DRAW_INDEX_AUTO\n"); + return -EINVAL; + } + r = r600_cs_track_check(p); + if (r) { + dev_warn(p->dev, "%s:%d invalid cmd stream %d\n", __func__, __LINE__, idx); + return r; + } + break; + case PACKET3_DRAW_INDEX_IMMD_BE: + case PACKET3_DRAW_INDEX_IMMD: + if (pkt->count < 2) { + DRM_ERROR("bad DRAW_INDEX_IMMD\n"); + return -EINVAL; + } + r = r600_cs_track_check(p); + if (r) { + dev_warn(p->dev, "%s:%d invalid cmd stream\n", __func__, __LINE__); + return r; + } + break; + case PACKET3_WAIT_REG_MEM: + if (pkt->count != 5) { + DRM_ERROR("bad WAIT_REG_MEM\n"); + return -EINVAL; + } + /* bit 4 is reg (0) or mem (1) */ + if (idx_value & 0x10) { + r = r600_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("bad WAIT_REG_MEM\n"); + return -EINVAL; + } + ib[idx+1] += (u32)(reloc->lobj.gpu_offset & 0xffffffff); + ib[idx+2] += upper_32_bits(reloc->lobj.gpu_offset) & 0xff; + } + break; + case PACKET3_SURFACE_SYNC: + if (pkt->count != 3) { + DRM_ERROR("bad SURFACE_SYNC\n"); + return -EINVAL; + } + /* 0xffffffff/0x0 is flush all cache flag */ + if (radeon_get_ib_value(p, idx + 1) != 0xffffffff || + radeon_get_ib_value(p, idx + 2) != 0) { + r = r600_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("bad SURFACE_SYNC\n"); + return -EINVAL; + } + ib[idx+2] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + } + break; + case PACKET3_EVENT_WRITE: + if (pkt->count != 2 && pkt->count != 0) { + DRM_ERROR("bad EVENT_WRITE\n"); + return -EINVAL; + } + if (pkt->count) { + r = r600_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("bad EVENT_WRITE\n"); + return -EINVAL; + } + ib[idx+1] += (u32)(reloc->lobj.gpu_offset & 0xffffffff); + ib[idx+2] += upper_32_bits(reloc->lobj.gpu_offset) & 0xff; + } + break; + case PACKET3_EVENT_WRITE_EOP: + if (pkt->count != 4) { + DRM_ERROR("bad EVENT_WRITE_EOP\n"); + return -EINVAL; + } + r = r600_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("bad EVENT_WRITE\n"); + return -EINVAL; + } + ib[idx+1] += (u32)(reloc->lobj.gpu_offset & 0xffffffff); + ib[idx+2] += upper_32_bits(reloc->lobj.gpu_offset) & 0xff; + break; + case PACKET3_SET_CONFIG_REG: + start_reg = (idx_value << 2) + PACKET3_SET_CONFIG_REG_OFFSET; + end_reg = 4 * pkt->count + start_reg - 4; + if ((start_reg < PACKET3_SET_CONFIG_REG_OFFSET) || + (start_reg >= PACKET3_SET_CONFIG_REG_END) || + (end_reg >= PACKET3_SET_CONFIG_REG_END)) { + DRM_ERROR("bad PACKET3_SET_CONFIG_REG\n"); + return -EINVAL; + } + for (i = 0; i < pkt->count; i++) { + reg = start_reg + (4 * i); + r = r600_cs_check_reg(p, reg, idx+1+i); + if (r) + return r; + } + break; + case PACKET3_SET_CONTEXT_REG: + start_reg = (idx_value << 2) + PACKET3_SET_CONTEXT_REG_OFFSET; + end_reg = 4 * pkt->count + start_reg - 4; + if ((start_reg < PACKET3_SET_CONTEXT_REG_OFFSET) || + (start_reg >= PACKET3_SET_CONTEXT_REG_END) || + (end_reg >= PACKET3_SET_CONTEXT_REG_END)) { + DRM_ERROR("bad PACKET3_SET_CONTEXT_REG\n"); + return -EINVAL; + } + for (i = 0; i < pkt->count; i++) { + reg = start_reg + (4 * i); + r = r600_cs_check_reg(p, reg, idx+1+i); + if (r) + return r; + } + break; + case PACKET3_SET_RESOURCE: + if (pkt->count % 7) { + DRM_ERROR("bad SET_RESOURCE\n"); + return -EINVAL; + } + start_reg = (idx_value << 2) + PACKET3_SET_RESOURCE_OFFSET; + end_reg = 4 * pkt->count + start_reg - 4; + if ((start_reg < PACKET3_SET_RESOURCE_OFFSET) || + (start_reg >= PACKET3_SET_RESOURCE_END) || + (end_reg >= PACKET3_SET_RESOURCE_END)) { + DRM_ERROR("bad SET_RESOURCE\n"); + return -EINVAL; + } + for (i = 0; i < (pkt->count / 7); i++) { + struct radeon_bo *texture, *mipmap; + u32 size, offset, base_offset, mip_offset; + + switch (G__SQ_VTX_CONSTANT_TYPE(radeon_get_ib_value(p, idx+(i*7)+6+1))) { + case SQ_TEX_VTX_VALID_TEXTURE: + /* tex base */ + r = r600_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("bad SET_RESOURCE\n"); + return -EINVAL; + } + base_offset = (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) + ib[idx+1+(i*7)+0] |= S_038000_TILE_MODE(V_038000_ARRAY_2D_TILED_THIN1); + else if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) + ib[idx+1+(i*7)+0] |= S_038000_TILE_MODE(V_038000_ARRAY_1D_TILED_THIN1); + texture = reloc->robj; + /* tex mip base */ + r = r600_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("bad SET_RESOURCE\n"); + return -EINVAL; + } + mip_offset = (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + mipmap = reloc->robj; + r = r600_check_texture_resource(p, idx+(i*7)+1, + texture, mipmap, + base_offset + radeon_get_ib_value(p, idx+1+(i*7)+2), + mip_offset + radeon_get_ib_value(p, idx+1+(i*7)+3), + reloc->lobj.tiling_flags); + if (r) + return r; + ib[idx+1+(i*7)+2] += base_offset; + ib[idx+1+(i*7)+3] += mip_offset; + break; + case SQ_TEX_VTX_VALID_BUFFER: + /* vtx base */ + r = r600_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("bad SET_RESOURCE\n"); + return -EINVAL; + } + offset = radeon_get_ib_value(p, idx+1+(i*7)+0); + size = radeon_get_ib_value(p, idx+1+(i*7)+1) + 1; + if (p->rdev && (size + offset) > radeon_bo_size(reloc->robj)) { + /* force size to size of the buffer */ + dev_warn(p->dev, "vbo resource seems too big (%d) for the bo (%ld)\n", + size + offset, radeon_bo_size(reloc->robj)); + ib[idx+1+(i*7)+1] = radeon_bo_size(reloc->robj); + } + ib[idx+1+(i*7)+0] += (u32)((reloc->lobj.gpu_offset) & 0xffffffff); + ib[idx+1+(i*7)+2] += upper_32_bits(reloc->lobj.gpu_offset) & 0xff; + break; + case SQ_TEX_VTX_INVALID_TEXTURE: + case SQ_TEX_VTX_INVALID_BUFFER: + default: + DRM_ERROR("bad SET_RESOURCE\n"); + return -EINVAL; + } + } + break; + case PACKET3_SET_ALU_CONST: + if (track->sq_config & DX9_CONSTS) { + start_reg = (idx_value << 2) + PACKET3_SET_ALU_CONST_OFFSET; + end_reg = 4 * pkt->count + start_reg - 4; + if ((start_reg < PACKET3_SET_ALU_CONST_OFFSET) || + (start_reg >= PACKET3_SET_ALU_CONST_END) || + (end_reg >= PACKET3_SET_ALU_CONST_END)) { + DRM_ERROR("bad SET_ALU_CONST\n"); + return -EINVAL; + } + } + break; + case PACKET3_SET_BOOL_CONST: + start_reg = (idx_value << 2) + PACKET3_SET_BOOL_CONST_OFFSET; + end_reg = 4 * pkt->count + start_reg - 4; + if ((start_reg < PACKET3_SET_BOOL_CONST_OFFSET) || + (start_reg >= PACKET3_SET_BOOL_CONST_END) || + (end_reg >= PACKET3_SET_BOOL_CONST_END)) { + DRM_ERROR("bad SET_BOOL_CONST\n"); + return -EINVAL; + } + break; + case PACKET3_SET_LOOP_CONST: + start_reg = (idx_value << 2) + PACKET3_SET_LOOP_CONST_OFFSET; + end_reg = 4 * pkt->count + start_reg - 4; + if ((start_reg < PACKET3_SET_LOOP_CONST_OFFSET) || + (start_reg >= PACKET3_SET_LOOP_CONST_END) || + (end_reg >= PACKET3_SET_LOOP_CONST_END)) { + DRM_ERROR("bad SET_LOOP_CONST\n"); + return -EINVAL; + } + break; + case PACKET3_SET_CTL_CONST: + start_reg = (idx_value << 2) + PACKET3_SET_CTL_CONST_OFFSET; + end_reg = 4 * pkt->count + start_reg - 4; + if ((start_reg < PACKET3_SET_CTL_CONST_OFFSET) || + (start_reg >= PACKET3_SET_CTL_CONST_END) || + (end_reg >= PACKET3_SET_CTL_CONST_END)) { + DRM_ERROR("bad SET_CTL_CONST\n"); + return -EINVAL; + } + break; + case PACKET3_SET_SAMPLER: + if (pkt->count % 3) { + DRM_ERROR("bad SET_SAMPLER\n"); + return -EINVAL; + } + start_reg = (idx_value << 2) + PACKET3_SET_SAMPLER_OFFSET; + end_reg = 4 * pkt->count + start_reg - 4; + if ((start_reg < PACKET3_SET_SAMPLER_OFFSET) || + (start_reg >= PACKET3_SET_SAMPLER_END) || + (end_reg >= PACKET3_SET_SAMPLER_END)) { + DRM_ERROR("bad SET_SAMPLER\n"); + return -EINVAL; + } + break; + case PACKET3_SURFACE_BASE_UPDATE: + if (p->family >= CHIP_RV770 || p->family == CHIP_R600) { + DRM_ERROR("bad SURFACE_BASE_UPDATE\n"); + return -EINVAL; + } + if (pkt->count) { + DRM_ERROR("bad SURFACE_BASE_UPDATE\n"); + return -EINVAL; + } + break; + case PACKET3_NOP: + break; + default: + DRM_ERROR("Packet3 opcode %x not supported\n", pkt->opcode); + return -EINVAL; + } + return 0; +} + +int r600_cs_parse(struct radeon_cs_parser *p) +{ + struct radeon_cs_packet pkt; + struct r600_cs_track *track; + int r; + + if (p->track == NULL) { + /* initialize tracker, we are in kms */ + track = kzalloc(sizeof(*track), GFP_KERNEL); + if (track == NULL) + return -ENOMEM; + r600_cs_track_init(track); + if (p->rdev->family < CHIP_RV770) { + track->npipes = p->rdev->config.r600.tiling_npipes; + track->nbanks = p->rdev->config.r600.tiling_nbanks; + track->group_size = p->rdev->config.r600.tiling_group_size; + } else if (p->rdev->family <= CHIP_RV740) { + track->npipes = p->rdev->config.rv770.tiling_npipes; + track->nbanks = p->rdev->config.rv770.tiling_nbanks; + track->group_size = p->rdev->config.rv770.tiling_group_size; + } + p->track = track; + } + do { + r = r600_cs_packet_parse(p, &pkt, p->idx); + if (r) { + kfree(p->track); + p->track = NULL; + return r; + } + p->idx += pkt.count + 2; + switch (pkt.type) { + case PACKET_TYPE0: + r = r600_cs_parse_packet0(p, &pkt); + break; + case PACKET_TYPE2: + break; + case PACKET_TYPE3: + r = r600_packet3_check(p, &pkt); + break; + default: + DRM_ERROR("Unknown packet type %d !\n", pkt.type); + kfree(p->track); + p->track = NULL; + return -EINVAL; + } + if (r) { + kfree(p->track); + p->track = NULL; + return r; + } + } while (p->idx < p->chunks[p->chunk_ib_idx].length_dw); +#if 0 + for (r = 0; r < p->ib->length_dw; r++) { + printk(KERN_INFO "%05d 0x%08X\n", r, p->ib->ptr[r]); + mdelay(1); + } +#endif + kfree(p->track); + p->track = NULL; + return 0; +} + +static int r600_cs_parser_relocs_legacy(struct radeon_cs_parser *p) +{ + if (p->chunk_relocs_idx == -1) { + return 0; + } + p->relocs = kzalloc(sizeof(struct radeon_cs_reloc), GFP_KERNEL); + if (p->relocs == NULL) { + return -ENOMEM; + } + return 0; +} + +/** + * cs_parser_fini() - clean parser states + * @parser: parser structure holding parsing context. + * @error: error number + * + * If error is set than unvalidate buffer, otherwise just free memory + * used by parsing context. + **/ +static void r600_cs_parser_fini(struct radeon_cs_parser *parser, int error) +{ + unsigned i; + + kfree(parser->relocs); + for (i = 0; i < parser->nchunks; i++) { + kfree(parser->chunks[i].kdata); + kfree(parser->chunks[i].kpage[0]); + kfree(parser->chunks[i].kpage[1]); + } + kfree(parser->chunks); + kfree(parser->chunks_array); +} + +int r600_cs_legacy(struct drm_device *dev, void *data, struct drm_file *filp, + unsigned family, u32 *ib, int *l) +{ + struct radeon_cs_parser parser; + struct radeon_cs_chunk *ib_chunk; + struct radeon_ib fake_ib; + struct r600_cs_track *track; + int r; + + /* initialize tracker */ + track = kzalloc(sizeof(*track), GFP_KERNEL); + if (track == NULL) + return -ENOMEM; + r600_cs_track_init(track); + r600_cs_legacy_get_tiling_conf(dev, &track->npipes, &track->nbanks, &track->group_size); + /* initialize parser */ + memset(&parser, 0, sizeof(struct radeon_cs_parser)); + parser.filp = filp; + parser.dev = &dev->pdev->dev; + parser.rdev = NULL; + parser.family = family; + parser.ib = &fake_ib; + parser.track = track; + fake_ib.ptr = ib; + r = radeon_cs_parser_init(&parser, data); + if (r) { + DRM_ERROR("Failed to initialize parser !\n"); + r600_cs_parser_fini(&parser, r); + return r; + } + r = r600_cs_parser_relocs_legacy(&parser); + if (r) { + DRM_ERROR("Failed to parse relocation !\n"); + r600_cs_parser_fini(&parser, r); + return r; + } + /* Copy the packet into the IB, the parser will read from the + * input memory (cached) and write to the IB (which can be + * uncached). */ + ib_chunk = &parser.chunks[parser.chunk_ib_idx]; + parser.ib->length_dw = ib_chunk->length_dw; + *l = parser.ib->length_dw; + r = r600_cs_parse(&parser); + if (r) { + DRM_ERROR("Invalid command stream !\n"); + r600_cs_parser_fini(&parser, r); + return r; + } + r = radeon_cs_finish_pages(&parser); + if (r) { + DRM_ERROR("Invalid command stream !\n"); + r600_cs_parser_fini(&parser, r); + return r; + } + r600_cs_parser_fini(&parser, r); + return r; +} + +void r600_cs_legacy_init(void) +{ + r600_cs_packet_next_reloc = &r600_cs_packet_next_reloc_nomm; +} diff --git a/drivers/gpu/drm/radeon/r600_hdmi.c b/drivers/gpu/drm/radeon/r600_hdmi.c new file mode 100644 index 00000000..c45d9219 --- /dev/null +++ b/drivers/gpu/drm/radeon/r600_hdmi.c @@ -0,0 +1,593 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Christian König. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Christian König + */ +#include "drmP.h" +#include "radeon_drm.h" +#include "radeon.h" +#include "radeon_asic.h" +#include "atom.h" + +/* + * HDMI color format + */ +enum r600_hdmi_color_format { + RGB = 0, + YCC_422 = 1, + YCC_444 = 2 +}; + +/* + * IEC60958 status bits + */ +enum r600_hdmi_iec_status_bits { + AUDIO_STATUS_DIG_ENABLE = 0x01, + AUDIO_STATUS_V = 0x02, + AUDIO_STATUS_VCFG = 0x04, + AUDIO_STATUS_EMPHASIS = 0x08, + AUDIO_STATUS_COPYRIGHT = 0x10, + AUDIO_STATUS_NONAUDIO = 0x20, + AUDIO_STATUS_PROFESSIONAL = 0x40, + AUDIO_STATUS_LEVEL = 0x80 +}; + +struct { + uint32_t Clock; + + int N_32kHz; + int CTS_32kHz; + + int N_44_1kHz; + int CTS_44_1kHz; + + int N_48kHz; + int CTS_48kHz; + +} r600_hdmi_ACR[] = { + /* 32kHz 44.1kHz 48kHz */ + /* Clock N CTS N CTS N CTS */ + { 25174, 4576, 28125, 7007, 31250, 6864, 28125 }, /* 25,20/1.001 MHz */ + { 25200, 4096, 25200, 6272, 28000, 6144, 25200 }, /* 25.20 MHz */ + { 27000, 4096, 27000, 6272, 30000, 6144, 27000 }, /* 27.00 MHz */ + { 27027, 4096, 27027, 6272, 30030, 6144, 27027 }, /* 27.00*1.001 MHz */ + { 54000, 4096, 54000, 6272, 60000, 6144, 54000 }, /* 54.00 MHz */ + { 54054, 4096, 54054, 6272, 60060, 6144, 54054 }, /* 54.00*1.001 MHz */ + { 74175, 11648, 210937, 17836, 234375, 11648, 140625 }, /* 74.25/1.001 MHz */ + { 74250, 4096, 74250, 6272, 82500, 6144, 74250 }, /* 74.25 MHz */ + { 148351, 11648, 421875, 8918, 234375, 5824, 140625 }, /* 148.50/1.001 MHz */ + { 148500, 4096, 148500, 6272, 165000, 6144, 148500 }, /* 148.50 MHz */ + { 0, 4096, 0, 6272, 0, 6144, 0 } /* Other */ +}; + +/* + * calculate CTS value if it's not found in the table + */ +static void r600_hdmi_calc_CTS(uint32_t clock, int *CTS, int N, int freq) +{ + if (*CTS == 0) + *CTS = clock * N / (128 * freq) * 1000; + DRM_DEBUG("Using ACR timing N=%d CTS=%d for frequency %d\n", + N, *CTS, freq); +} + +/* + * update the N and CTS parameters for a given pixel clock rate + */ +static void r600_hdmi_update_ACR(struct drm_encoder *encoder, uint32_t clock) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + uint32_t offset = to_radeon_encoder(encoder)->hdmi_offset; + int CTS; + int N; + int i; + + for (i = 0; r600_hdmi_ACR[i].Clock != clock && r600_hdmi_ACR[i].Clock != 0; i++); + + CTS = r600_hdmi_ACR[i].CTS_32kHz; + N = r600_hdmi_ACR[i].N_32kHz; + r600_hdmi_calc_CTS(clock, &CTS, N, 32000); + WREG32(offset+R600_HDMI_32kHz_CTS, CTS << 12); + WREG32(offset+R600_HDMI_32kHz_N, N); + + CTS = r600_hdmi_ACR[i].CTS_44_1kHz; + N = r600_hdmi_ACR[i].N_44_1kHz; + r600_hdmi_calc_CTS(clock, &CTS, N, 44100); + WREG32(offset+R600_HDMI_44_1kHz_CTS, CTS << 12); + WREG32(offset+R600_HDMI_44_1kHz_N, N); + + CTS = r600_hdmi_ACR[i].CTS_48kHz; + N = r600_hdmi_ACR[i].N_48kHz; + r600_hdmi_calc_CTS(clock, &CTS, N, 48000); + WREG32(offset+R600_HDMI_48kHz_CTS, CTS << 12); + WREG32(offset+R600_HDMI_48kHz_N, N); +} + +/* + * calculate the crc for a given info frame + */ +static void r600_hdmi_infoframe_checksum(uint8_t packetType, + uint8_t versionNumber, + uint8_t length, + uint8_t *frame) +{ + int i; + frame[0] = packetType + versionNumber + length; + for (i = 1; i <= length; i++) + frame[0] += frame[i]; + frame[0] = 0x100 - frame[0]; +} + +/* + * build a HDMI Video Info Frame + */ +static void r600_hdmi_videoinfoframe( + struct drm_encoder *encoder, + enum r600_hdmi_color_format color_format, + int active_information_present, + uint8_t active_format_aspect_ratio, + uint8_t scan_information, + uint8_t colorimetry, + uint8_t ex_colorimetry, + uint8_t quantization, + int ITC, + uint8_t picture_aspect_ratio, + uint8_t video_format_identification, + uint8_t pixel_repetition, + uint8_t non_uniform_picture_scaling, + uint8_t bar_info_data_valid, + uint16_t top_bar, + uint16_t bottom_bar, + uint16_t left_bar, + uint16_t right_bar +) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + uint32_t offset = to_radeon_encoder(encoder)->hdmi_offset; + + uint8_t frame[14]; + + frame[0x0] = 0; + frame[0x1] = + (scan_information & 0x3) | + ((bar_info_data_valid & 0x3) << 2) | + ((active_information_present & 0x1) << 4) | + ((color_format & 0x3) << 5); + frame[0x2] = + (active_format_aspect_ratio & 0xF) | + ((picture_aspect_ratio & 0x3) << 4) | + ((colorimetry & 0x3) << 6); + frame[0x3] = + (non_uniform_picture_scaling & 0x3) | + ((quantization & 0x3) << 2) | + ((ex_colorimetry & 0x7) << 4) | + ((ITC & 0x1) << 7); + frame[0x4] = (video_format_identification & 0x7F); + frame[0x5] = (pixel_repetition & 0xF); + frame[0x6] = (top_bar & 0xFF); + frame[0x7] = (top_bar >> 8); + frame[0x8] = (bottom_bar & 0xFF); + frame[0x9] = (bottom_bar >> 8); + frame[0xA] = (left_bar & 0xFF); + frame[0xB] = (left_bar >> 8); + frame[0xC] = (right_bar & 0xFF); + frame[0xD] = (right_bar >> 8); + + r600_hdmi_infoframe_checksum(0x82, 0x02, 0x0D, frame); + /* Our header values (type, version, length) should be alright, Intel + * is using the same. Checksum function also seems to be OK, it works + * fine for audio infoframe. However calculated value is always lower + * by 2 in comparison to fglrx. It breaks displaying anything in case + * of TVs that strictly check the checksum. Hack it manually here to + * workaround this issue. */ + frame[0x0] += 2; + + WREG32(offset+R600_HDMI_VIDEOINFOFRAME_0, + frame[0x0] | (frame[0x1] << 8) | (frame[0x2] << 16) | (frame[0x3] << 24)); + WREG32(offset+R600_HDMI_VIDEOINFOFRAME_1, + frame[0x4] | (frame[0x5] << 8) | (frame[0x6] << 16) | (frame[0x7] << 24)); + WREG32(offset+R600_HDMI_VIDEOINFOFRAME_2, + frame[0x8] | (frame[0x9] << 8) | (frame[0xA] << 16) | (frame[0xB] << 24)); + WREG32(offset+R600_HDMI_VIDEOINFOFRAME_3, + frame[0xC] | (frame[0xD] << 8)); +} + +/* + * build a Audio Info Frame + */ +static void r600_hdmi_audioinfoframe( + struct drm_encoder *encoder, + uint8_t channel_count, + uint8_t coding_type, + uint8_t sample_size, + uint8_t sample_frequency, + uint8_t format, + uint8_t channel_allocation, + uint8_t level_shift, + int downmix_inhibit +) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + uint32_t offset = to_radeon_encoder(encoder)->hdmi_offset; + + uint8_t frame[11]; + + frame[0x0] = 0; + frame[0x1] = (channel_count & 0x7) | ((coding_type & 0xF) << 4); + frame[0x2] = (sample_size & 0x3) | ((sample_frequency & 0x7) << 2); + frame[0x3] = format; + frame[0x4] = channel_allocation; + frame[0x5] = ((level_shift & 0xF) << 3) | ((downmix_inhibit & 0x1) << 7); + frame[0x6] = 0; + frame[0x7] = 0; + frame[0x8] = 0; + frame[0x9] = 0; + frame[0xA] = 0; + + r600_hdmi_infoframe_checksum(0x84, 0x01, 0x0A, frame); + + WREG32(offset+R600_HDMI_AUDIOINFOFRAME_0, + frame[0x0] | (frame[0x1] << 8) | (frame[0x2] << 16) | (frame[0x3] << 24)); + WREG32(offset+R600_HDMI_AUDIOINFOFRAME_1, + frame[0x4] | (frame[0x5] << 8) | (frame[0x6] << 16) | (frame[0x8] << 24)); +} + +/* + * test if audio buffer is filled enough to start playing + */ +static int r600_hdmi_is_audio_buffer_filled(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + uint32_t offset = to_radeon_encoder(encoder)->hdmi_offset; + + return (RREG32(offset+R600_HDMI_STATUS) & 0x10) != 0; +} + +/* + * have buffer status changed since last call? + */ +int r600_hdmi_buffer_status_changed(struct drm_encoder *encoder) +{ + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + int status, result; + + if (!radeon_encoder->hdmi_offset) + return 0; + + status = r600_hdmi_is_audio_buffer_filled(encoder); + result = radeon_encoder->hdmi_buffer_status != status; + radeon_encoder->hdmi_buffer_status = status; + + return result; +} + +/* + * write the audio workaround status to the hardware + */ +void r600_hdmi_audio_workaround(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + uint32_t offset = radeon_encoder->hdmi_offset; + + if (!offset) + return; + + if (!radeon_encoder->hdmi_audio_workaround || + r600_hdmi_is_audio_buffer_filled(encoder)) { + + /* disable audio workaround */ + WREG32_P(offset+R600_HDMI_CNTL, 0x00000001, ~0x00001001); + + } else { + /* enable audio workaround */ + WREG32_P(offset+R600_HDMI_CNTL, 0x00001001, ~0x00001001); + } +} + + +/* + * update the info frames with the data from the current display mode + */ +void r600_hdmi_setmode(struct drm_encoder *encoder, struct drm_display_mode *mode) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + uint32_t offset = to_radeon_encoder(encoder)->hdmi_offset; + + if (ASIC_IS_DCE4(rdev)) + return; + + if (!offset) + return; + + r600_audio_set_clock(encoder, mode->clock); + + WREG32(offset+R600_HDMI_UNKNOWN_0, 0x1000); + WREG32(offset+R600_HDMI_UNKNOWN_1, 0x0); + WREG32(offset+R600_HDMI_UNKNOWN_2, 0x1000); + + r600_hdmi_update_ACR(encoder, mode->clock); + + WREG32(offset+R600_HDMI_VIDEOCNTL, 0x13); + + WREG32(offset+R600_HDMI_VERSION, 0x202); + + r600_hdmi_videoinfoframe(encoder, RGB, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0); + + /* it's unknown what these bits do excatly, but it's indeed quite useful for debugging */ + WREG32(offset+R600_HDMI_AUDIO_DEBUG_0, 0x00FFFFFF); + WREG32(offset+R600_HDMI_AUDIO_DEBUG_1, 0x007FFFFF); + WREG32(offset+R600_HDMI_AUDIO_DEBUG_2, 0x00000001); + WREG32(offset+R600_HDMI_AUDIO_DEBUG_3, 0x00000001); + + r600_hdmi_audio_workaround(encoder); + + /* audio packets per line, does anyone know how to calc this ? */ + WREG32_P(offset+R600_HDMI_CNTL, 0x00040000, ~0x001F0000); +} + +/* + * update settings with current parameters from audio engine + */ +void r600_hdmi_update_audio_settings(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + uint32_t offset = to_radeon_encoder(encoder)->hdmi_offset; + + int channels = r600_audio_channels(rdev); + int rate = r600_audio_rate(rdev); + int bps = r600_audio_bits_per_sample(rdev); + uint8_t status_bits = r600_audio_status_bits(rdev); + uint8_t category_code = r600_audio_category_code(rdev); + + uint32_t iec; + + if (!offset) + return; + + DRM_DEBUG("%s with %d channels, %d Hz sampling rate, %d bits per sample,\n", + r600_hdmi_is_audio_buffer_filled(encoder) ? "playing" : "stopped", + channels, rate, bps); + DRM_DEBUG("0x%02X IEC60958 status bits and 0x%02X category code\n", + (int)status_bits, (int)category_code); + + iec = 0; + if (status_bits & AUDIO_STATUS_PROFESSIONAL) + iec |= 1 << 0; + if (status_bits & AUDIO_STATUS_NONAUDIO) + iec |= 1 << 1; + if (status_bits & AUDIO_STATUS_COPYRIGHT) + iec |= 1 << 2; + if (status_bits & AUDIO_STATUS_EMPHASIS) + iec |= 1 << 3; + + iec |= category_code << 8; + + switch (rate) { + case 32000: iec |= 0x3 << 24; break; + case 44100: iec |= 0x0 << 24; break; + case 88200: iec |= 0x8 << 24; break; + case 176400: iec |= 0xc << 24; break; + case 48000: iec |= 0x2 << 24; break; + case 96000: iec |= 0xa << 24; break; + case 192000: iec |= 0xe << 24; break; + } + + WREG32(offset+R600_HDMI_IEC60958_1, iec); + + iec = 0; + switch (bps) { + case 16: iec |= 0x2; break; + case 20: iec |= 0x3; break; + case 24: iec |= 0xb; break; + } + if (status_bits & AUDIO_STATUS_V) + iec |= 0x5 << 16; + + WREG32_P(offset+R600_HDMI_IEC60958_2, iec, ~0x5000f); + + /* 0x021 or 0x031 sets the audio frame length */ + WREG32(offset+R600_HDMI_AUDIOCNTL, 0x31); + r600_hdmi_audioinfoframe(encoder, channels-1, 0, 0, 0, 0, 0, 0, 0); + + r600_hdmi_audio_workaround(encoder); +} + +static int r600_hdmi_find_free_block(struct drm_device *dev) +{ + struct radeon_device *rdev = dev->dev_private; + struct drm_encoder *encoder; + struct radeon_encoder *radeon_encoder; + bool free_blocks[3] = { true, true, true }; + + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + radeon_encoder = to_radeon_encoder(encoder); + switch (radeon_encoder->hdmi_offset) { + case R600_HDMI_BLOCK1: + free_blocks[0] = false; + break; + case R600_HDMI_BLOCK2: + free_blocks[1] = false; + break; + case R600_HDMI_BLOCK3: + free_blocks[2] = false; + break; + } + } + + if (rdev->family == CHIP_RS600 || rdev->family == CHIP_RS690 || + rdev->family == CHIP_RS740) { + return free_blocks[0] ? R600_HDMI_BLOCK1 : 0; + } else if (rdev->family >= CHIP_R600) { + if (free_blocks[0]) + return R600_HDMI_BLOCK1; + else if (free_blocks[1]) + return R600_HDMI_BLOCK2; + } + return 0; +} + +static void r600_hdmi_assign_block(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct radeon_encoder_atom_dig *dig = radeon_encoder->enc_priv; + + if (!dig) { + dev_err(rdev->dev, "Enabling HDMI on non-dig encoder\n"); + return; + } + + if (ASIC_IS_DCE4(rdev)) { + /* TODO */ + } else if (ASIC_IS_DCE3(rdev)) { + radeon_encoder->hdmi_offset = dig->dig_encoder ? + R600_HDMI_BLOCK3 : R600_HDMI_BLOCK1; + if (ASIC_IS_DCE32(rdev)) + radeon_encoder->hdmi_config_offset = dig->dig_encoder ? + R600_HDMI_CONFIG2 : R600_HDMI_CONFIG1; + } else if (rdev->family >= CHIP_R600 || rdev->family == CHIP_RS600 || + rdev->family == CHIP_RS690 || rdev->family == CHIP_RS740) { + radeon_encoder->hdmi_offset = r600_hdmi_find_free_block(dev); + } +} + +/* + * enable the HDMI engine + */ +void r600_hdmi_enable(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + uint32_t offset; + + if (ASIC_IS_DCE4(rdev)) + return; + + if (!radeon_encoder->hdmi_offset) { + r600_hdmi_assign_block(encoder); + if (!radeon_encoder->hdmi_offset) { + dev_warn(rdev->dev, "Could not find HDMI block for " + "0x%x encoder\n", radeon_encoder->encoder_id); + return; + } + } + + offset = radeon_encoder->hdmi_offset; + if (ASIC_IS_DCE32(rdev) && !ASIC_IS_DCE4(rdev)) { + WREG32_P(radeon_encoder->hdmi_config_offset + 0x4, 0x1, ~0x1); + } else if (rdev->family >= CHIP_R600 && !ASIC_IS_DCE3(rdev)) { + switch (radeon_encoder->encoder_id) { + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_TMDS1: + WREG32_P(AVIVO_TMDSA_CNTL, 0x4, ~0x4); + WREG32(offset + R600_HDMI_ENABLE, 0x101); + break; + case ENCODER_OBJECT_ID_INTERNAL_LVTM1: + WREG32_P(AVIVO_LVTMA_CNTL, 0x4, ~0x4); + WREG32(offset + R600_HDMI_ENABLE, 0x105); + break; + default: + dev_err(rdev->dev, "Unknown HDMI output type\n"); + break; + } + } + + if (rdev->irq.installed + && rdev->family != CHIP_RS600 + && rdev->family != CHIP_RS690 + && rdev->family != CHIP_RS740) { + + /* if irq is available use it */ + rdev->irq.hdmi[offset == R600_HDMI_BLOCK1 ? 0 : 1] = true; + radeon_irq_set(rdev); + + r600_audio_disable_polling(encoder); + } else { + /* if not fallback to polling */ + r600_audio_enable_polling(encoder); + } + + DRM_DEBUG("Enabling HDMI interface @ 0x%04X for encoder 0x%x\n", + radeon_encoder->hdmi_offset, radeon_encoder->encoder_id); +} + +/* + * disable the HDMI engine + */ +void r600_hdmi_disable(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + uint32_t offset; + + if (ASIC_IS_DCE4(rdev)) + return; + + offset = radeon_encoder->hdmi_offset; + if (!offset) { + dev_err(rdev->dev, "Disabling not enabled HDMI\n"); + return; + } + + DRM_DEBUG("Disabling HDMI interface @ 0x%04X for encoder 0x%x\n", + offset, radeon_encoder->encoder_id); + + /* disable irq */ + rdev->irq.hdmi[offset == R600_HDMI_BLOCK1 ? 0 : 1] = false; + radeon_irq_set(rdev); + + /* disable polling */ + r600_audio_disable_polling(encoder); + + if (ASIC_IS_DCE32(rdev) && !ASIC_IS_DCE4(rdev)) { + WREG32_P(radeon_encoder->hdmi_config_offset + 0x4, 0, ~0x1); + } else if (rdev->family >= CHIP_R600 && !ASIC_IS_DCE3(rdev)) { + switch (radeon_encoder->encoder_id) { + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_TMDS1: + WREG32_P(AVIVO_TMDSA_CNTL, 0, ~0x4); + WREG32(offset + R600_HDMI_ENABLE, 0); + break; + case ENCODER_OBJECT_ID_INTERNAL_LVTM1: + WREG32_P(AVIVO_LVTMA_CNTL, 0, ~0x4); + WREG32(offset + R600_HDMI_ENABLE, 0); + break; + default: + dev_err(rdev->dev, "Unknown HDMI output type\n"); + break; + } + } + + radeon_encoder->hdmi_offset = 0; + radeon_encoder->hdmi_config_offset = 0; +} diff --git a/drivers/gpu/drm/radeon/r600_reg.h b/drivers/gpu/drm/radeon/r600_reg.h new file mode 100644 index 00000000..f869897c --- /dev/null +++ b/drivers/gpu/drm/radeon/r600_reg.h @@ -0,0 +1,200 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#ifndef __R600_REG_H__ +#define __R600_REG_H__ + +#define R600_PCIE_PORT_INDEX 0x0038 +#define R600_PCIE_PORT_DATA 0x003c + +#define R600_MC_VM_FB_LOCATION 0x2180 +#define R600_MC_FB_BASE_MASK 0x0000FFFF +#define R600_MC_FB_BASE_SHIFT 0 +#define R600_MC_FB_TOP_MASK 0xFFFF0000 +#define R600_MC_FB_TOP_SHIFT 16 +#define R600_MC_VM_AGP_TOP 0x2184 +#define R600_MC_AGP_TOP_MASK 0x0003FFFF +#define R600_MC_AGP_TOP_SHIFT 0 +#define R600_MC_VM_AGP_BOT 0x2188 +#define R600_MC_AGP_BOT_MASK 0x0003FFFF +#define R600_MC_AGP_BOT_SHIFT 0 +#define R600_MC_VM_AGP_BASE 0x218c +#define R600_MC_VM_SYSTEM_APERTURE_LOW_ADDR 0x2190 +#define R600_LOGICAL_PAGE_NUMBER_MASK 0x000FFFFF +#define R600_LOGICAL_PAGE_NUMBER_SHIFT 0 +#define R600_MC_VM_SYSTEM_APERTURE_HIGH_ADDR 0x2194 +#define R600_MC_VM_SYSTEM_APERTURE_DEFAULT_ADDR 0x2198 + +#define R700_MC_VM_FB_LOCATION 0x2024 +#define R700_MC_FB_BASE_MASK 0x0000FFFF +#define R700_MC_FB_BASE_SHIFT 0 +#define R700_MC_FB_TOP_MASK 0xFFFF0000 +#define R700_MC_FB_TOP_SHIFT 16 +#define R700_MC_VM_AGP_TOP 0x2028 +#define R700_MC_AGP_TOP_MASK 0x0003FFFF +#define R700_MC_AGP_TOP_SHIFT 0 +#define R700_MC_VM_AGP_BOT 0x202c +#define R700_MC_AGP_BOT_MASK 0x0003FFFF +#define R700_MC_AGP_BOT_SHIFT 0 +#define R700_MC_VM_AGP_BASE 0x2030 +#define R700_MC_VM_SYSTEM_APERTURE_LOW_ADDR 0x2034 +#define R700_LOGICAL_PAGE_NUMBER_MASK 0x000FFFFF +#define R700_LOGICAL_PAGE_NUMBER_SHIFT 0 +#define R700_MC_VM_SYSTEM_APERTURE_HIGH_ADDR 0x2038 +#define R700_MC_VM_SYSTEM_APERTURE_DEFAULT_ADDR 0x203c + +#define R600_RAMCFG 0x2408 +# define R600_CHANSIZE (1 << 7) +# define R600_CHANSIZE_OVERRIDE (1 << 10) + + +#define R600_GENERAL_PWRMGT 0x618 +# define R600_OPEN_DRAIN_PADS (1 << 11) + +#define R600_LOWER_GPIO_ENABLE 0x710 +#define R600_CTXSW_VID_LOWER_GPIO_CNTL 0x718 +#define R600_HIGH_VID_LOWER_GPIO_CNTL 0x71c +#define R600_MEDIUM_VID_LOWER_GPIO_CNTL 0x720 +#define R600_LOW_VID_LOWER_GPIO_CNTL 0x724 + +#define R600_D1GRPH_SWAP_CONTROL 0x610C +# define R600_D1GRPH_SWAP_ENDIAN_NONE (0 << 0) +# define R600_D1GRPH_SWAP_ENDIAN_16BIT (1 << 0) +# define R600_D1GRPH_SWAP_ENDIAN_32BIT (2 << 0) +# define R600_D1GRPH_SWAP_ENDIAN_64BIT (3 << 0) + +#define R600_HDP_NONSURFACE_BASE 0x2c04 + +#define R600_BUS_CNTL 0x5420 +# define R600_BIOS_ROM_DIS (1 << 1) +#define R600_CONFIG_CNTL 0x5424 +#define R600_CONFIG_MEMSIZE 0x5428 +#define R600_CONFIG_F0_BASE 0x542C +#define R600_CONFIG_APER_SIZE 0x5430 + +#define R600_ROM_CNTL 0x1600 +# define R600_SCK_OVERWRITE (1 << 1) +# define R600_SCK_PRESCALE_CRYSTAL_CLK_SHIFT 28 +# define R600_SCK_PRESCALE_CRYSTAL_CLK_MASK (0xf << 28) + +#define R600_CG_SPLL_FUNC_CNTL 0x600 +# define R600_SPLL_BYPASS_EN (1 << 3) +#define R600_CG_SPLL_STATUS 0x60c +# define R600_SPLL_CHG_STATUS (1 << 1) + +#define R600_BIOS_0_SCRATCH 0x1724 +#define R600_BIOS_1_SCRATCH 0x1728 +#define R600_BIOS_2_SCRATCH 0x172c +#define R600_BIOS_3_SCRATCH 0x1730 +#define R600_BIOS_4_SCRATCH 0x1734 +#define R600_BIOS_5_SCRATCH 0x1738 +#define R600_BIOS_6_SCRATCH 0x173c +#define R600_BIOS_7_SCRATCH 0x1740 + +/* Audio, these regs were reverse enginered, + * so the chance is high that the naming is wrong + * R6xx+ ??? */ + +/* Audio clocks */ +#define R600_AUDIO_PLL1_MUL 0x0514 +#define R600_AUDIO_PLL1_DIV 0x0518 +#define R600_AUDIO_PLL2_MUL 0x0524 +#define R600_AUDIO_PLL2_DIV 0x0528 +#define R600_AUDIO_CLK_SRCSEL 0x0534 + +/* Audio general */ +#define R600_AUDIO_ENABLE 0x7300 +#define R600_AUDIO_TIMING 0x7344 + +/* Audio params */ +#define R600_AUDIO_VENDOR_ID 0x7380 +#define R600_AUDIO_REVISION_ID 0x7384 +#define R600_AUDIO_ROOT_NODE_COUNT 0x7388 +#define R600_AUDIO_NID1_NODE_COUNT 0x738c +#define R600_AUDIO_NID1_TYPE 0x7390 +#define R600_AUDIO_SUPPORTED_SIZE_RATE 0x7394 +#define R600_AUDIO_SUPPORTED_CODEC 0x7398 +#define R600_AUDIO_SUPPORTED_POWER_STATES 0x739c +#define R600_AUDIO_NID2_CAPS 0x73a0 +#define R600_AUDIO_NID3_CAPS 0x73a4 +#define R600_AUDIO_NID3_PIN_CAPS 0x73a8 + +/* Audio conn list */ +#define R600_AUDIO_CONN_LIST_LEN 0x73ac +#define R600_AUDIO_CONN_LIST 0x73b0 + +/* Audio verbs */ +#define R600_AUDIO_RATE_BPS_CHANNEL 0x73c0 +#define R600_AUDIO_PLAYING 0x73c4 +#define R600_AUDIO_IMPLEMENTATION_ID 0x73c8 +#define R600_AUDIO_CONFIG_DEFAULT 0x73cc +#define R600_AUDIO_PIN_SENSE 0x73d0 +#define R600_AUDIO_PIN_WIDGET_CNTL 0x73d4 +#define R600_AUDIO_STATUS_BITS 0x73d8 + +/* HDMI base register addresses */ +#define R600_HDMI_BLOCK1 0x7400 +#define R600_HDMI_BLOCK2 0x7700 +#define R600_HDMI_BLOCK3 0x7800 + +/* HDMI registers */ +#define R600_HDMI_ENABLE 0x00 +#define R600_HDMI_STATUS 0x04 +# define R600_HDMI_INT_PENDING (1 << 29) +#define R600_HDMI_CNTL 0x08 +# define R600_HDMI_INT_EN (1 << 28) +# define R600_HDMI_INT_ACK (1 << 29) +#define R600_HDMI_UNKNOWN_0 0x0C +#define R600_HDMI_AUDIOCNTL 0x10 +#define R600_HDMI_VIDEOCNTL 0x14 +#define R600_HDMI_VERSION 0x18 +#define R600_HDMI_UNKNOWN_1 0x28 +#define R600_HDMI_VIDEOINFOFRAME_0 0x54 +#define R600_HDMI_VIDEOINFOFRAME_1 0x58 +#define R600_HDMI_VIDEOINFOFRAME_2 0x5c +#define R600_HDMI_VIDEOINFOFRAME_3 0x60 +#define R600_HDMI_32kHz_CTS 0xac +#define R600_HDMI_32kHz_N 0xb0 +#define R600_HDMI_44_1kHz_CTS 0xb4 +#define R600_HDMI_44_1kHz_N 0xb8 +#define R600_HDMI_48kHz_CTS 0xbc +#define R600_HDMI_48kHz_N 0xc0 +#define R600_HDMI_AUDIOINFOFRAME_0 0xcc +#define R600_HDMI_AUDIOINFOFRAME_1 0xd0 +#define R600_HDMI_IEC60958_1 0xd4 +#define R600_HDMI_IEC60958_2 0xd8 +#define R600_HDMI_UNKNOWN_2 0xdc +#define R600_HDMI_AUDIO_DEBUG_0 0xe0 +#define R600_HDMI_AUDIO_DEBUG_1 0xe4 +#define R600_HDMI_AUDIO_DEBUG_2 0xe8 +#define R600_HDMI_AUDIO_DEBUG_3 0xec + +/* HDMI additional config base register addresses */ +#define R600_HDMI_CONFIG1 0x7600 +#define R600_HDMI_CONFIG2 0x7a00 + +#endif diff --git a/drivers/gpu/drm/radeon/r600d.h b/drivers/gpu/drm/radeon/r600d.h new file mode 100644 index 00000000..0245ae6c --- /dev/null +++ b/drivers/gpu/drm/radeon/r600d.h @@ -0,0 +1,1520 @@ +/* + * Copyright 2009 Advanced Micro Devices, Inc. + * Copyright 2009 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#ifndef R600D_H +#define R600D_H + +#define CP_PACKET2 0x80000000 +#define PACKET2_PAD_SHIFT 0 +#define PACKET2_PAD_MASK (0x3fffffff << 0) + +#define PACKET2(v) (CP_PACKET2 | REG_SET(PACKET2_PAD, (v))) + +#define R6XX_MAX_SH_GPRS 256 +#define R6XX_MAX_TEMP_GPRS 16 +#define R6XX_MAX_SH_THREADS 256 +#define R6XX_MAX_SH_STACK_ENTRIES 4096 +#define R6XX_MAX_BACKENDS 8 +#define R6XX_MAX_BACKENDS_MASK 0xff +#define R6XX_MAX_SIMDS 8 +#define R6XX_MAX_SIMDS_MASK 0xff +#define R6XX_MAX_PIPES 8 +#define R6XX_MAX_PIPES_MASK 0xff + +/* PTE flags */ +#define PTE_VALID (1 << 0) +#define PTE_SYSTEM (1 << 1) +#define PTE_SNOOPED (1 << 2) +#define PTE_READABLE (1 << 5) +#define PTE_WRITEABLE (1 << 6) + +/* tiling bits */ +#define ARRAY_LINEAR_GENERAL 0x00000000 +#define ARRAY_LINEAR_ALIGNED 0x00000001 +#define ARRAY_1D_TILED_THIN1 0x00000002 +#define ARRAY_2D_TILED_THIN1 0x00000004 + +/* Registers */ +#define ARB_POP 0x2418 +#define ENABLE_TC128 (1 << 30) +#define ARB_GDEC_RD_CNTL 0x246C + +#define CC_GC_SHADER_PIPE_CONFIG 0x8950 +#define CC_RB_BACKEND_DISABLE 0x98F4 +#define BACKEND_DISABLE(x) ((x) << 16) + +#define CB_COLOR0_BASE 0x28040 +#define CB_COLOR1_BASE 0x28044 +#define CB_COLOR2_BASE 0x28048 +#define CB_COLOR3_BASE 0x2804C +#define CB_COLOR4_BASE 0x28050 +#define CB_COLOR5_BASE 0x28054 +#define CB_COLOR6_BASE 0x28058 +#define CB_COLOR7_BASE 0x2805C +#define CB_COLOR7_FRAG 0x280FC + +#define CB_COLOR0_SIZE 0x28060 +#define CB_COLOR0_VIEW 0x28080 +#define CB_COLOR0_INFO 0x280a0 +#define CB_COLOR0_TILE 0x280c0 +#define CB_COLOR0_FRAG 0x280e0 +#define CB_COLOR0_MASK 0x28100 + +#define SQ_ALU_CONST_CACHE_PS_0 0x28940 +#define SQ_ALU_CONST_CACHE_PS_1 0x28944 +#define SQ_ALU_CONST_CACHE_PS_2 0x28948 +#define SQ_ALU_CONST_CACHE_PS_3 0x2894c +#define SQ_ALU_CONST_CACHE_PS_4 0x28950 +#define SQ_ALU_CONST_CACHE_PS_5 0x28954 +#define SQ_ALU_CONST_CACHE_PS_6 0x28958 +#define SQ_ALU_CONST_CACHE_PS_7 0x2895c +#define SQ_ALU_CONST_CACHE_PS_8 0x28960 +#define SQ_ALU_CONST_CACHE_PS_9 0x28964 +#define SQ_ALU_CONST_CACHE_PS_10 0x28968 +#define SQ_ALU_CONST_CACHE_PS_11 0x2896c +#define SQ_ALU_CONST_CACHE_PS_12 0x28970 +#define SQ_ALU_CONST_CACHE_PS_13 0x28974 +#define SQ_ALU_CONST_CACHE_PS_14 0x28978 +#define SQ_ALU_CONST_CACHE_PS_15 0x2897c +#define SQ_ALU_CONST_CACHE_VS_0 0x28980 +#define SQ_ALU_CONST_CACHE_VS_1 0x28984 +#define SQ_ALU_CONST_CACHE_VS_2 0x28988 +#define SQ_ALU_CONST_CACHE_VS_3 0x2898c +#define SQ_ALU_CONST_CACHE_VS_4 0x28990 +#define SQ_ALU_CONST_CACHE_VS_5 0x28994 +#define SQ_ALU_CONST_CACHE_VS_6 0x28998 +#define SQ_ALU_CONST_CACHE_VS_7 0x2899c +#define SQ_ALU_CONST_CACHE_VS_8 0x289a0 +#define SQ_ALU_CONST_CACHE_VS_9 0x289a4 +#define SQ_ALU_CONST_CACHE_VS_10 0x289a8 +#define SQ_ALU_CONST_CACHE_VS_11 0x289ac +#define SQ_ALU_CONST_CACHE_VS_12 0x289b0 +#define SQ_ALU_CONST_CACHE_VS_13 0x289b4 +#define SQ_ALU_CONST_CACHE_VS_14 0x289b8 +#define SQ_ALU_CONST_CACHE_VS_15 0x289bc +#define SQ_ALU_CONST_CACHE_GS_0 0x289c0 +#define SQ_ALU_CONST_CACHE_GS_1 0x289c4 +#define SQ_ALU_CONST_CACHE_GS_2 0x289c8 +#define SQ_ALU_CONST_CACHE_GS_3 0x289cc +#define SQ_ALU_CONST_CACHE_GS_4 0x289d0 +#define SQ_ALU_CONST_CACHE_GS_5 0x289d4 +#define SQ_ALU_CONST_CACHE_GS_6 0x289d8 +#define SQ_ALU_CONST_CACHE_GS_7 0x289dc +#define SQ_ALU_CONST_CACHE_GS_8 0x289e0 +#define SQ_ALU_CONST_CACHE_GS_9 0x289e4 +#define SQ_ALU_CONST_CACHE_GS_10 0x289e8 +#define SQ_ALU_CONST_CACHE_GS_11 0x289ec +#define SQ_ALU_CONST_CACHE_GS_12 0x289f0 +#define SQ_ALU_CONST_CACHE_GS_13 0x289f4 +#define SQ_ALU_CONST_CACHE_GS_14 0x289f8 +#define SQ_ALU_CONST_CACHE_GS_15 0x289fc + +#define CONFIG_MEMSIZE 0x5428 +#define CONFIG_CNTL 0x5424 +#define CP_STAT 0x8680 +#define CP_COHER_BASE 0x85F8 +#define CP_DEBUG 0xC1FC +#define R_0086D8_CP_ME_CNTL 0x86D8 +#define S_0086D8_CP_ME_HALT(x) (((x) & 1)<<28) +#define C_0086D8_CP_ME_HALT(x) ((x) & 0xEFFFFFFF) +#define CP_ME_RAM_DATA 0xC160 +#define CP_ME_RAM_RADDR 0xC158 +#define CP_ME_RAM_WADDR 0xC15C +#define CP_MEQ_THRESHOLDS 0x8764 +#define MEQ_END(x) ((x) << 16) +#define ROQ_END(x) ((x) << 24) +#define CP_PERFMON_CNTL 0x87FC +#define CP_PFP_UCODE_ADDR 0xC150 +#define CP_PFP_UCODE_DATA 0xC154 +#define CP_QUEUE_THRESHOLDS 0x8760 +#define ROQ_IB1_START(x) ((x) << 0) +#define ROQ_IB2_START(x) ((x) << 8) +#define CP_RB_BASE 0xC100 +#define CP_RB_CNTL 0xC104 +#define RB_BUFSZ(x) ((x) << 0) +#define RB_BLKSZ(x) ((x) << 8) +#define RB_NO_UPDATE (1 << 27) +#define RB_RPTR_WR_ENA (1 << 31) +#define BUF_SWAP_32BIT (2 << 16) +#define CP_RB_RPTR 0x8700 +#define CP_RB_RPTR_ADDR 0xC10C +#define RB_RPTR_SWAP(x) ((x) << 0) +#define CP_RB_RPTR_ADDR_HI 0xC110 +#define CP_RB_RPTR_WR 0xC108 +#define CP_RB_WPTR 0xC114 +#define CP_RB_WPTR_ADDR 0xC118 +#define CP_RB_WPTR_ADDR_HI 0xC11C +#define CP_RB_WPTR_DELAY 0x8704 +#define CP_ROQ_IB1_STAT 0x8784 +#define CP_ROQ_IB2_STAT 0x8788 +#define CP_SEM_WAIT_TIMER 0x85BC + +#define DB_DEBUG 0x9830 +#define PREZ_MUST_WAIT_FOR_POSTZ_DONE (1 << 31) +#define DB_DEPTH_BASE 0x2800C +#define DB_HTILE_DATA_BASE 0x28014 +#define DB_WATERMARKS 0x9838 +#define DEPTH_FREE(x) ((x) << 0) +#define DEPTH_FLUSH(x) ((x) << 5) +#define DEPTH_PENDING_FREE(x) ((x) << 15) +#define DEPTH_CACHELINE_FREE(x) ((x) << 20) + +#define DCP_TILING_CONFIG 0x6CA0 +#define PIPE_TILING(x) ((x) << 1) +#define BANK_TILING(x) ((x) << 4) +#define GROUP_SIZE(x) ((x) << 6) +#define ROW_TILING(x) ((x) << 8) +#define BANK_SWAPS(x) ((x) << 11) +#define SAMPLE_SPLIT(x) ((x) << 14) +#define BACKEND_MAP(x) ((x) << 16) + +#define GB_TILING_CONFIG 0x98F0 + +#define GC_USER_SHADER_PIPE_CONFIG 0x8954 +#define INACTIVE_QD_PIPES(x) ((x) << 8) +#define INACTIVE_QD_PIPES_MASK 0x0000FF00 +#define INACTIVE_SIMDS(x) ((x) << 16) +#define INACTIVE_SIMDS_MASK 0x00FF0000 + +#define SQ_CONFIG 0x8c00 +# define VC_ENABLE (1 << 0) +# define EXPORT_SRC_C (1 << 1) +# define DX9_CONSTS (1 << 2) +# define ALU_INST_PREFER_VECTOR (1 << 3) +# define DX10_CLAMP (1 << 4) +# define CLAUSE_SEQ_PRIO(x) ((x) << 8) +# define PS_PRIO(x) ((x) << 24) +# define VS_PRIO(x) ((x) << 26) +# define GS_PRIO(x) ((x) << 28) +# define ES_PRIO(x) ((x) << 30) +#define SQ_GPR_RESOURCE_MGMT_1 0x8c04 +# define NUM_PS_GPRS(x) ((x) << 0) +# define NUM_VS_GPRS(x) ((x) << 16) +# define NUM_CLAUSE_TEMP_GPRS(x) ((x) << 28) +#define SQ_GPR_RESOURCE_MGMT_2 0x8c08 +# define NUM_GS_GPRS(x) ((x) << 0) +# define NUM_ES_GPRS(x) ((x) << 16) +#define SQ_THREAD_RESOURCE_MGMT 0x8c0c +# define NUM_PS_THREADS(x) ((x) << 0) +# define NUM_VS_THREADS(x) ((x) << 8) +# define NUM_GS_THREADS(x) ((x) << 16) +# define NUM_ES_THREADS(x) ((x) << 24) +#define SQ_STACK_RESOURCE_MGMT_1 0x8c10 +# define NUM_PS_STACK_ENTRIES(x) ((x) << 0) +# define NUM_VS_STACK_ENTRIES(x) ((x) << 16) +#define SQ_STACK_RESOURCE_MGMT_2 0x8c14 +# define NUM_GS_STACK_ENTRIES(x) ((x) << 0) +# define NUM_ES_STACK_ENTRIES(x) ((x) << 16) +#define SQ_ESGS_RING_BASE 0x8c40 +#define SQ_GSVS_RING_BASE 0x8c48 +#define SQ_ESTMP_RING_BASE 0x8c50 +#define SQ_GSTMP_RING_BASE 0x8c58 +#define SQ_VSTMP_RING_BASE 0x8c60 +#define SQ_PSTMP_RING_BASE 0x8c68 +#define SQ_FBUF_RING_BASE 0x8c70 +#define SQ_REDUC_RING_BASE 0x8c78 + +#define GRBM_CNTL 0x8000 +# define GRBM_READ_TIMEOUT(x) ((x) << 0) +#define GRBM_STATUS 0x8010 +#define CMDFIFO_AVAIL_MASK 0x0000001F +#define GUI_ACTIVE (1<<31) +#define GRBM_STATUS2 0x8014 +#define GRBM_SOFT_RESET 0x8020 +#define SOFT_RESET_CP (1<<0) + +#define CG_THERMAL_STATUS 0x7F4 +#define ASIC_T(x) ((x) << 0) +#define ASIC_T_MASK 0x1FF +#define ASIC_T_SHIFT 0 + +#define HDP_HOST_PATH_CNTL 0x2C00 +#define HDP_NONSURFACE_BASE 0x2C04 +#define HDP_NONSURFACE_INFO 0x2C08 +#define HDP_NONSURFACE_SIZE 0x2C0C +#define HDP_REG_COHERENCY_FLUSH_CNTL 0x54A0 +#define HDP_TILING_CONFIG 0x2F3C +#define HDP_DEBUG1 0x2F34 + +#define MC_VM_AGP_TOP 0x2184 +#define MC_VM_AGP_BOT 0x2188 +#define MC_VM_AGP_BASE 0x218C +#define MC_VM_FB_LOCATION 0x2180 +#define MC_VM_L1_TLB_MCD_RD_A_CNTL 0x219C +#define ENABLE_L1_TLB (1 << 0) +#define ENABLE_L1_FRAGMENT_PROCESSING (1 << 1) +#define ENABLE_L1_STRICT_ORDERING (1 << 2) +#define SYSTEM_ACCESS_MODE_MASK 0x000000C0 +#define SYSTEM_ACCESS_MODE_SHIFT 6 +#define SYSTEM_ACCESS_MODE_PA_ONLY (0 << 6) +#define SYSTEM_ACCESS_MODE_USE_SYS_MAP (1 << 6) +#define SYSTEM_ACCESS_MODE_IN_SYS (2 << 6) +#define SYSTEM_ACCESS_MODE_NOT_IN_SYS (3 << 6) +#define SYSTEM_APERTURE_UNMAPPED_ACCESS_PASS_THRU (0 << 8) +#define SYSTEM_APERTURE_UNMAPPED_ACCESS_DEFAULT_PAGE (1 << 8) +#define ENABLE_SEMAPHORE_MODE (1 << 10) +#define ENABLE_WAIT_L2_QUERY (1 << 11) +#define EFFECTIVE_L1_TLB_SIZE(x) (((x) & 7) << 12) +#define EFFECTIVE_L1_TLB_SIZE_MASK 0x00007000 +#define EFFECTIVE_L1_TLB_SIZE_SHIFT 12 +#define EFFECTIVE_L1_QUEUE_SIZE(x) (((x) & 7) << 15) +#define EFFECTIVE_L1_QUEUE_SIZE_MASK 0x00038000 +#define EFFECTIVE_L1_QUEUE_SIZE_SHIFT 15 +#define MC_VM_L1_TLB_MCD_RD_B_CNTL 0x21A0 +#define MC_VM_L1_TLB_MCB_RD_GFX_CNTL 0x21FC +#define MC_VM_L1_TLB_MCB_RD_HDP_CNTL 0x2204 +#define MC_VM_L1_TLB_MCB_RD_PDMA_CNTL 0x2208 +#define MC_VM_L1_TLB_MCB_RD_SEM_CNTL 0x220C +#define MC_VM_L1_TLB_MCB_RD_SYS_CNTL 0x2200 +#define MC_VM_L1_TLB_MCD_WR_A_CNTL 0x21A4 +#define MC_VM_L1_TLB_MCD_WR_B_CNTL 0x21A8 +#define MC_VM_L1_TLB_MCB_WR_GFX_CNTL 0x2210 +#define MC_VM_L1_TLB_MCB_WR_HDP_CNTL 0x2218 +#define MC_VM_L1_TLB_MCB_WR_PDMA_CNTL 0x221C +#define MC_VM_L1_TLB_MCB_WR_SEM_CNTL 0x2220 +#define MC_VM_L1_TLB_MCB_WR_SYS_CNTL 0x2214 +#define MC_VM_SYSTEM_APERTURE_LOW_ADDR 0x2190 +#define LOGICAL_PAGE_NUMBER_MASK 0x000FFFFF +#define LOGICAL_PAGE_NUMBER_SHIFT 0 +#define MC_VM_SYSTEM_APERTURE_HIGH_ADDR 0x2194 +#define MC_VM_SYSTEM_APERTURE_DEFAULT_ADDR 0x2198 + +#define PA_CL_ENHANCE 0x8A14 +#define CLIP_VTX_REORDER_ENA (1 << 0) +#define NUM_CLIP_SEQ(x) ((x) << 1) +#define PA_SC_AA_CONFIG 0x28C04 +#define PA_SC_AA_SAMPLE_LOCS_2S 0x8B40 +#define PA_SC_AA_SAMPLE_LOCS_4S 0x8B44 +#define PA_SC_AA_SAMPLE_LOCS_8S_WD0 0x8B48 +#define PA_SC_AA_SAMPLE_LOCS_8S_WD1 0x8B4C +#define S0_X(x) ((x) << 0) +#define S0_Y(x) ((x) << 4) +#define S1_X(x) ((x) << 8) +#define S1_Y(x) ((x) << 12) +#define S2_X(x) ((x) << 16) +#define S2_Y(x) ((x) << 20) +#define S3_X(x) ((x) << 24) +#define S3_Y(x) ((x) << 28) +#define S4_X(x) ((x) << 0) +#define S4_Y(x) ((x) << 4) +#define S5_X(x) ((x) << 8) +#define S5_Y(x) ((x) << 12) +#define S6_X(x) ((x) << 16) +#define S6_Y(x) ((x) << 20) +#define S7_X(x) ((x) << 24) +#define S7_Y(x) ((x) << 28) +#define PA_SC_CLIPRECT_RULE 0x2820c +#define PA_SC_ENHANCE 0x8BF0 +#define FORCE_EOV_MAX_CLK_CNT(x) ((x) << 0) +#define FORCE_EOV_MAX_TILE_CNT(x) ((x) << 12) +#define PA_SC_LINE_STIPPLE 0x28A0C +#define PA_SC_LINE_STIPPLE_STATE 0x8B10 +#define PA_SC_MODE_CNTL 0x28A4C +#define PA_SC_MULTI_CHIP_CNTL 0x8B20 + +#define PA_SC_SCREEN_SCISSOR_TL 0x28030 +#define PA_SC_GENERIC_SCISSOR_TL 0x28240 +#define PA_SC_WINDOW_SCISSOR_TL 0x28204 + +#define PCIE_PORT_INDEX 0x0038 +#define PCIE_PORT_DATA 0x003C + +#define CHMAP 0x2004 +#define NOOFCHAN_SHIFT 12 +#define NOOFCHAN_MASK 0x00003000 + +#define RAMCFG 0x2408 +#define NOOFBANK_SHIFT 0 +#define NOOFBANK_MASK 0x00000001 +#define NOOFRANK_SHIFT 1 +#define NOOFRANK_MASK 0x00000002 +#define NOOFROWS_SHIFT 2 +#define NOOFROWS_MASK 0x0000001C +#define NOOFCOLS_SHIFT 5 +#define NOOFCOLS_MASK 0x00000060 +#define CHANSIZE_SHIFT 7 +#define CHANSIZE_MASK 0x00000080 +#define BURSTLENGTH_SHIFT 8 +#define BURSTLENGTH_MASK 0x00000100 +#define CHANSIZE_OVERRIDE (1 << 10) + +#define SCRATCH_REG0 0x8500 +#define SCRATCH_REG1 0x8504 +#define SCRATCH_REG2 0x8508 +#define SCRATCH_REG3 0x850C +#define SCRATCH_REG4 0x8510 +#define SCRATCH_REG5 0x8514 +#define SCRATCH_REG6 0x8518 +#define SCRATCH_REG7 0x851C +#define SCRATCH_UMSK 0x8540 +#define SCRATCH_ADDR 0x8544 + +#define SPI_CONFIG_CNTL 0x9100 +#define GPR_WRITE_PRIORITY(x) ((x) << 0) +#define DISABLE_INTERP_1 (1 << 5) +#define SPI_CONFIG_CNTL_1 0x913C +#define VTX_DONE_DELAY(x) ((x) << 0) +#define INTERP_ONE_PRIM_PER_ROW (1 << 4) +#define SPI_INPUT_Z 0x286D8 +#define SPI_PS_IN_CONTROL_0 0x286CC +#define NUM_INTERP(x) ((x)<<0) +#define POSITION_ENA (1<<8) +#define POSITION_CENTROID (1<<9) +#define POSITION_ADDR(x) ((x)<<10) +#define PARAM_GEN(x) ((x)<<15) +#define PARAM_GEN_ADDR(x) ((x)<<19) +#define BARYC_SAMPLE_CNTL(x) ((x)<<26) +#define PERSP_GRADIENT_ENA (1<<28) +#define LINEAR_GRADIENT_ENA (1<<29) +#define POSITION_SAMPLE (1<<30) +#define BARYC_AT_SAMPLE_ENA (1<<31) +#define SPI_PS_IN_CONTROL_1 0x286D0 +#define GEN_INDEX_PIX (1<<0) +#define GEN_INDEX_PIX_ADDR(x) ((x)<<1) +#define FRONT_FACE_ENA (1<<8) +#define FRONT_FACE_CHAN(x) ((x)<<9) +#define FRONT_FACE_ALL_BITS (1<<11) +#define FRONT_FACE_ADDR(x) ((x)<<12) +#define FOG_ADDR(x) ((x)<<17) +#define FIXED_PT_POSITION_ENA (1<<24) +#define FIXED_PT_POSITION_ADDR(x) ((x)<<25) + +#define SQ_MS_FIFO_SIZES 0x8CF0 +#define CACHE_FIFO_SIZE(x) ((x) << 0) +#define FETCH_FIFO_HIWATER(x) ((x) << 8) +#define DONE_FIFO_HIWATER(x) ((x) << 16) +#define ALU_UPDATE_FIFO_HIWATER(x) ((x) << 24) +#define SQ_PGM_START_ES 0x28880 +#define SQ_PGM_START_FS 0x28894 +#define SQ_PGM_START_GS 0x2886C +#define SQ_PGM_START_PS 0x28840 +#define SQ_PGM_RESOURCES_PS 0x28850 +#define SQ_PGM_EXPORTS_PS 0x28854 +#define SQ_PGM_CF_OFFSET_PS 0x288cc +#define SQ_PGM_START_VS 0x28858 +#define SQ_PGM_RESOURCES_VS 0x28868 +#define SQ_PGM_CF_OFFSET_VS 0x288d0 +#define SQ_VTX_CONSTANT_WORD6_0 0x38018 +#define S__SQ_VTX_CONSTANT_TYPE(x) (((x) & 3) << 30) +#define G__SQ_VTX_CONSTANT_TYPE(x) (((x) >> 30) & 3) +#define SQ_TEX_VTX_INVALID_TEXTURE 0x0 +#define SQ_TEX_VTX_INVALID_BUFFER 0x1 +#define SQ_TEX_VTX_VALID_TEXTURE 0x2 +#define SQ_TEX_VTX_VALID_BUFFER 0x3 + + +#define SX_MISC 0x28350 +#define SX_MEMORY_EXPORT_BASE 0x9010 +#define SX_DEBUG_1 0x9054 +#define SMX_EVENT_RELEASE (1 << 0) +#define ENABLE_NEW_SMX_ADDRESS (1 << 16) + +#define TA_CNTL_AUX 0x9508 +#define DISABLE_CUBE_WRAP (1 << 0) +#define DISABLE_CUBE_ANISO (1 << 1) +#define SYNC_GRADIENT (1 << 24) +#define SYNC_WALKER (1 << 25) +#define SYNC_ALIGNER (1 << 26) +#define BILINEAR_PRECISION_6_BIT (0 << 31) +#define BILINEAR_PRECISION_8_BIT (1 << 31) + +#define TC_CNTL 0x9608 +#define TC_L2_SIZE(x) ((x)<<5) +#define L2_DISABLE_LATE_HIT (1<<9) + + +#define VGT_CACHE_INVALIDATION 0x88C4 +#define CACHE_INVALIDATION(x) ((x)<<0) +#define VC_ONLY 0 +#define TC_ONLY 1 +#define VC_AND_TC 2 +#define VGT_DMA_BASE 0x287E8 +#define VGT_DMA_BASE_HI 0x287E4 +#define VGT_ES_PER_GS 0x88CC +#define VGT_GS_PER_ES 0x88C8 +#define VGT_GS_PER_VS 0x88E8 +#define VGT_GS_VERTEX_REUSE 0x88D4 +#define VGT_PRIMITIVE_TYPE 0x8958 +#define VGT_NUM_INSTANCES 0x8974 +#define VGT_OUT_DEALLOC_CNTL 0x28C5C +#define DEALLOC_DIST_MASK 0x0000007F +#define VGT_STRMOUT_BASE_OFFSET_0 0x28B10 +#define VGT_STRMOUT_BASE_OFFSET_1 0x28B14 +#define VGT_STRMOUT_BASE_OFFSET_2 0x28B18 +#define VGT_STRMOUT_BASE_OFFSET_3 0x28B1c +#define VGT_STRMOUT_BASE_OFFSET_HI_0 0x28B44 +#define VGT_STRMOUT_BASE_OFFSET_HI_1 0x28B48 +#define VGT_STRMOUT_BASE_OFFSET_HI_2 0x28B4c +#define VGT_STRMOUT_BASE_OFFSET_HI_3 0x28B50 +#define VGT_STRMOUT_BUFFER_BASE_0 0x28AD8 +#define VGT_STRMOUT_BUFFER_BASE_1 0x28AE8 +#define VGT_STRMOUT_BUFFER_BASE_2 0x28AF8 +#define VGT_STRMOUT_BUFFER_BASE_3 0x28B08 +#define VGT_STRMOUT_BUFFER_OFFSET_0 0x28ADC +#define VGT_STRMOUT_BUFFER_OFFSET_1 0x28AEC +#define VGT_STRMOUT_BUFFER_OFFSET_2 0x28AFC +#define VGT_STRMOUT_BUFFER_OFFSET_3 0x28B0C +#define VGT_STRMOUT_EN 0x28AB0 +#define VGT_VERTEX_REUSE_BLOCK_CNTL 0x28C58 +#define VTX_REUSE_DEPTH_MASK 0x000000FF +#define VGT_EVENT_INITIATOR 0x28a90 +# define CACHE_FLUSH_AND_INV_EVENT_TS (0x14 << 0) +# define CACHE_FLUSH_AND_INV_EVENT (0x16 << 0) + +#define VM_CONTEXT0_CNTL 0x1410 +#define ENABLE_CONTEXT (1 << 0) +#define PAGE_TABLE_DEPTH(x) (((x) & 3) << 1) +#define RANGE_PROTECTION_FAULT_ENABLE_DEFAULT (1 << 4) +#define VM_CONTEXT0_INVALIDATION_LOW_ADDR 0x1490 +#define VM_CONTEXT0_INVALIDATION_HIGH_ADDR 0x14B0 +#define VM_CONTEXT0_PAGE_TABLE_BASE_ADDR 0x1574 +#define VM_CONTEXT0_PAGE_TABLE_START_ADDR 0x1594 +#define VM_CONTEXT0_PAGE_TABLE_END_ADDR 0x15B4 +#define VM_CONTEXT0_PROTECTION_FAULT_DEFAULT_ADDR 0x1554 +#define VM_CONTEXT0_REQUEST_RESPONSE 0x1470 +#define REQUEST_TYPE(x) (((x) & 0xf) << 0) +#define RESPONSE_TYPE_MASK 0x000000F0 +#define RESPONSE_TYPE_SHIFT 4 +#define VM_L2_CNTL 0x1400 +#define ENABLE_L2_CACHE (1 << 0) +#define ENABLE_L2_FRAGMENT_PROCESSING (1 << 1) +#define ENABLE_L2_PTE_CACHE_LRU_UPDATE_BY_WRITE (1 << 9) +#define EFFECTIVE_L2_QUEUE_SIZE(x) (((x) & 7) << 13) +#define VM_L2_CNTL2 0x1404 +#define INVALIDATE_ALL_L1_TLBS (1 << 0) +#define INVALIDATE_L2_CACHE (1 << 1) +#define VM_L2_CNTL3 0x1408 +#define BANK_SELECT_0(x) (((x) & 0x1f) << 0) +#define BANK_SELECT_1(x) (((x) & 0x1f) << 5) +#define L2_CACHE_UPDATE_MODE(x) (((x) & 3) << 10) +#define VM_L2_STATUS 0x140C +#define L2_BUSY (1 << 0) + +#define WAIT_UNTIL 0x8040 +#define WAIT_2D_IDLE_bit (1 << 14) +#define WAIT_3D_IDLE_bit (1 << 15) +#define WAIT_2D_IDLECLEAN_bit (1 << 16) +#define WAIT_3D_IDLECLEAN_bit (1 << 17) + +#define IH_RB_CNTL 0x3e00 +# define IH_RB_ENABLE (1 << 0) +# define IH_IB_SIZE(x) ((x) << 1) /* log2 */ +# define IH_RB_FULL_DRAIN_ENABLE (1 << 6) +# define IH_WPTR_WRITEBACK_ENABLE (1 << 8) +# define IH_WPTR_WRITEBACK_TIMER(x) ((x) << 9) /* log2 */ +# define IH_WPTR_OVERFLOW_ENABLE (1 << 16) +# define IH_WPTR_OVERFLOW_CLEAR (1 << 31) +#define IH_RB_BASE 0x3e04 +#define IH_RB_RPTR 0x3e08 +#define IH_RB_WPTR 0x3e0c +# define RB_OVERFLOW (1 << 0) +# define WPTR_OFFSET_MASK 0x3fffc +#define IH_RB_WPTR_ADDR_HI 0x3e10 +#define IH_RB_WPTR_ADDR_LO 0x3e14 +#define IH_CNTL 0x3e18 +# define ENABLE_INTR (1 << 0) +# define IH_MC_SWAP(x) ((x) << 1) +# define IH_MC_SWAP_NONE 0 +# define IH_MC_SWAP_16BIT 1 +# define IH_MC_SWAP_32BIT 2 +# define IH_MC_SWAP_64BIT 3 +# define RPTR_REARM (1 << 4) +# define MC_WRREQ_CREDIT(x) ((x) << 15) +# define MC_WR_CLEAN_CNT(x) ((x) << 20) + +#define RLC_CNTL 0x3f00 +# define RLC_ENABLE (1 << 0) +#define RLC_HB_BASE 0x3f10 +#define RLC_HB_CNTL 0x3f0c +#define RLC_HB_RPTR 0x3f20 +#define RLC_HB_WPTR 0x3f1c +#define RLC_HB_WPTR_LSB_ADDR 0x3f14 +#define RLC_HB_WPTR_MSB_ADDR 0x3f18 +#define RLC_MC_CNTL 0x3f44 +#define RLC_UCODE_CNTL 0x3f48 +#define RLC_UCODE_ADDR 0x3f2c +#define RLC_UCODE_DATA 0x3f30 + +#define SRBM_SOFT_RESET 0xe60 +# define SOFT_RESET_RLC (1 << 13) + +#define CP_INT_CNTL 0xc124 +# define CNTX_BUSY_INT_ENABLE (1 << 19) +# define CNTX_EMPTY_INT_ENABLE (1 << 20) +# define SCRATCH_INT_ENABLE (1 << 25) +# define TIME_STAMP_INT_ENABLE (1 << 26) +# define IB2_INT_ENABLE (1 << 29) +# define IB1_INT_ENABLE (1 << 30) +# define RB_INT_ENABLE (1 << 31) +#define CP_INT_STATUS 0xc128 +# define SCRATCH_INT_STAT (1 << 25) +# define TIME_STAMP_INT_STAT (1 << 26) +# define IB2_INT_STAT (1 << 29) +# define IB1_INT_STAT (1 << 30) +# define RB_INT_STAT (1 << 31) + +#define GRBM_INT_CNTL 0x8060 +# define RDERR_INT_ENABLE (1 << 0) +# define WAIT_COUNT_TIMEOUT_INT_ENABLE (1 << 1) +# define GUI_IDLE_INT_ENABLE (1 << 19) + +#define INTERRUPT_CNTL 0x5468 +# define IH_DUMMY_RD_OVERRIDE (1 << 0) +# define IH_DUMMY_RD_EN (1 << 1) +# define IH_REQ_NONSNOOP_EN (1 << 3) +# define GEN_IH_INT_EN (1 << 8) +#define INTERRUPT_CNTL2 0x546c + +#define D1MODE_VBLANK_STATUS 0x6534 +#define D2MODE_VBLANK_STATUS 0x6d34 +# define DxMODE_VBLANK_OCCURRED (1 << 0) +# define DxMODE_VBLANK_ACK (1 << 4) +# define DxMODE_VBLANK_STAT (1 << 12) +# define DxMODE_VBLANK_INTERRUPT (1 << 16) +# define DxMODE_VBLANK_INTERRUPT_TYPE (1 << 17) +#define D1MODE_VLINE_STATUS 0x653c +#define D2MODE_VLINE_STATUS 0x6d3c +# define DxMODE_VLINE_OCCURRED (1 << 0) +# define DxMODE_VLINE_ACK (1 << 4) +# define DxMODE_VLINE_STAT (1 << 12) +# define DxMODE_VLINE_INTERRUPT (1 << 16) +# define DxMODE_VLINE_INTERRUPT_TYPE (1 << 17) +#define DxMODE_INT_MASK 0x6540 +# define D1MODE_VBLANK_INT_MASK (1 << 0) +# define D1MODE_VLINE_INT_MASK (1 << 4) +# define D2MODE_VBLANK_INT_MASK (1 << 8) +# define D2MODE_VLINE_INT_MASK (1 << 12) +#define DCE3_DISP_INTERRUPT_STATUS 0x7ddc +# define DC_HPD1_INTERRUPT (1 << 18) +# define DC_HPD2_INTERRUPT (1 << 19) +#define DISP_INTERRUPT_STATUS 0x7edc +# define LB_D1_VLINE_INTERRUPT (1 << 2) +# define LB_D2_VLINE_INTERRUPT (1 << 3) +# define LB_D1_VBLANK_INTERRUPT (1 << 4) +# define LB_D2_VBLANK_INTERRUPT (1 << 5) +# define DACA_AUTODETECT_INTERRUPT (1 << 16) +# define DACB_AUTODETECT_INTERRUPT (1 << 17) +# define DC_HOT_PLUG_DETECT1_INTERRUPT (1 << 18) +# define DC_HOT_PLUG_DETECT2_INTERRUPT (1 << 19) +# define DC_I2C_SW_DONE_INTERRUPT (1 << 20) +# define DC_I2C_HW_DONE_INTERRUPT (1 << 21) +#define DISP_INTERRUPT_STATUS_CONTINUE 0x7ee8 +#define DCE3_DISP_INTERRUPT_STATUS_CONTINUE 0x7de8 +# define DC_HPD4_INTERRUPT (1 << 14) +# define DC_HPD4_RX_INTERRUPT (1 << 15) +# define DC_HPD3_INTERRUPT (1 << 28) +# define DC_HPD1_RX_INTERRUPT (1 << 29) +# define DC_HPD2_RX_INTERRUPT (1 << 30) +#define DCE3_DISP_INTERRUPT_STATUS_CONTINUE2 0x7dec +# define DC_HPD3_RX_INTERRUPT (1 << 0) +# define DIGA_DP_VID_STREAM_DISABLE_INTERRUPT (1 << 1) +# define DIGA_DP_STEER_FIFO_OVERFLOW_INTERRUPT (1 << 2) +# define DIGB_DP_VID_STREAM_DISABLE_INTERRUPT (1 << 3) +# define DIGB_DP_STEER_FIFO_OVERFLOW_INTERRUPT (1 << 4) +# define AUX1_SW_DONE_INTERRUPT (1 << 5) +# define AUX1_LS_DONE_INTERRUPT (1 << 6) +# define AUX2_SW_DONE_INTERRUPT (1 << 7) +# define AUX2_LS_DONE_INTERRUPT (1 << 8) +# define AUX3_SW_DONE_INTERRUPT (1 << 9) +# define AUX3_LS_DONE_INTERRUPT (1 << 10) +# define AUX4_SW_DONE_INTERRUPT (1 << 11) +# define AUX4_LS_DONE_INTERRUPT (1 << 12) +# define DIGA_DP_FAST_TRAINING_COMPLETE_INTERRUPT (1 << 13) +# define DIGB_DP_FAST_TRAINING_COMPLETE_INTERRUPT (1 << 14) +/* DCE 3.2 */ +# define AUX5_SW_DONE_INTERRUPT (1 << 15) +# define AUX5_LS_DONE_INTERRUPT (1 << 16) +# define AUX6_SW_DONE_INTERRUPT (1 << 17) +# define AUX6_LS_DONE_INTERRUPT (1 << 18) +# define DC_HPD5_INTERRUPT (1 << 19) +# define DC_HPD5_RX_INTERRUPT (1 << 20) +# define DC_HPD6_INTERRUPT (1 << 21) +# define DC_HPD6_RX_INTERRUPT (1 << 22) + +#define DACA_AUTO_DETECT_CONTROL 0x7828 +#define DACB_AUTO_DETECT_CONTROL 0x7a28 +#define DCE3_DACA_AUTO_DETECT_CONTROL 0x7028 +#define DCE3_DACB_AUTO_DETECT_CONTROL 0x7128 +# define DACx_AUTODETECT_MODE(x) ((x) << 0) +# define DACx_AUTODETECT_MODE_NONE 0 +# define DACx_AUTODETECT_MODE_CONNECT 1 +# define DACx_AUTODETECT_MODE_DISCONNECT 2 +# define DACx_AUTODETECT_FRAME_TIME_COUNTER(x) ((x) << 8) +/* bit 18 = R/C, 17 = G/Y, 16 = B/Comp */ +# define DACx_AUTODETECT_CHECK_MASK(x) ((x) << 16) + +#define DCE3_DACA_AUTODETECT_INT_CONTROL 0x7038 +#define DCE3_DACB_AUTODETECT_INT_CONTROL 0x7138 +#define DACA_AUTODETECT_INT_CONTROL 0x7838 +#define DACB_AUTODETECT_INT_CONTROL 0x7a38 +# define DACx_AUTODETECT_ACK (1 << 0) +# define DACx_AUTODETECT_INT_ENABLE (1 << 16) + +#define DC_HOT_PLUG_DETECT1_CONTROL 0x7d00 +#define DC_HOT_PLUG_DETECT2_CONTROL 0x7d10 +#define DC_HOT_PLUG_DETECT3_CONTROL 0x7d24 +# define DC_HOT_PLUG_DETECTx_EN (1 << 0) + +#define DC_HOT_PLUG_DETECT1_INT_STATUS 0x7d04 +#define DC_HOT_PLUG_DETECT2_INT_STATUS 0x7d14 +#define DC_HOT_PLUG_DETECT3_INT_STATUS 0x7d28 +# define DC_HOT_PLUG_DETECTx_INT_STATUS (1 << 0) +# define DC_HOT_PLUG_DETECTx_SENSE (1 << 1) + +/* DCE 3.0 */ +#define DC_HPD1_INT_STATUS 0x7d00 +#define DC_HPD2_INT_STATUS 0x7d0c +#define DC_HPD3_INT_STATUS 0x7d18 +#define DC_HPD4_INT_STATUS 0x7d24 +/* DCE 3.2 */ +#define DC_HPD5_INT_STATUS 0x7dc0 +#define DC_HPD6_INT_STATUS 0x7df4 +# define DC_HPDx_INT_STATUS (1 << 0) +# define DC_HPDx_SENSE (1 << 1) +# define DC_HPDx_RX_INT_STATUS (1 << 8) + +#define DC_HOT_PLUG_DETECT1_INT_CONTROL 0x7d08 +#define DC_HOT_PLUG_DETECT2_INT_CONTROL 0x7d18 +#define DC_HOT_PLUG_DETECT3_INT_CONTROL 0x7d2c +# define DC_HOT_PLUG_DETECTx_INT_ACK (1 << 0) +# define DC_HOT_PLUG_DETECTx_INT_POLARITY (1 << 8) +# define DC_HOT_PLUG_DETECTx_INT_EN (1 << 16) +/* DCE 3.0 */ +#define DC_HPD1_INT_CONTROL 0x7d04 +#define DC_HPD2_INT_CONTROL 0x7d10 +#define DC_HPD3_INT_CONTROL 0x7d1c +#define DC_HPD4_INT_CONTROL 0x7d28 +/* DCE 3.2 */ +#define DC_HPD5_INT_CONTROL 0x7dc4 +#define DC_HPD6_INT_CONTROL 0x7df8 +# define DC_HPDx_INT_ACK (1 << 0) +# define DC_HPDx_INT_POLARITY (1 << 8) +# define DC_HPDx_INT_EN (1 << 16) +# define DC_HPDx_RX_INT_ACK (1 << 20) +# define DC_HPDx_RX_INT_EN (1 << 24) + +/* DCE 3.0 */ +#define DC_HPD1_CONTROL 0x7d08 +#define DC_HPD2_CONTROL 0x7d14 +#define DC_HPD3_CONTROL 0x7d20 +#define DC_HPD4_CONTROL 0x7d2c +/* DCE 3.2 */ +#define DC_HPD5_CONTROL 0x7dc8 +#define DC_HPD6_CONTROL 0x7dfc +# define DC_HPDx_CONNECTION_TIMER(x) ((x) << 0) +# define DC_HPDx_RX_INT_TIMER(x) ((x) << 16) +/* DCE 3.2 */ +# define DC_HPDx_EN (1 << 28) + +#define D1GRPH_INTERRUPT_STATUS 0x6158 +#define D2GRPH_INTERRUPT_STATUS 0x6958 +# define DxGRPH_PFLIP_INT_OCCURRED (1 << 0) +# define DxGRPH_PFLIP_INT_CLEAR (1 << 8) +#define D1GRPH_INTERRUPT_CONTROL 0x615c +#define D2GRPH_INTERRUPT_CONTROL 0x695c +# define DxGRPH_PFLIP_INT_MASK (1 << 0) +# define DxGRPH_PFLIP_INT_TYPE (1 << 8) + +/* PCIE link stuff */ +#define PCIE_LC_TRAINING_CNTL 0xa1 /* PCIE_P */ +# define LC_POINT_7_PLUS_EN (1 << 6) +#define PCIE_LC_LINK_WIDTH_CNTL 0xa2 /* PCIE_P */ +# define LC_LINK_WIDTH_SHIFT 0 +# define LC_LINK_WIDTH_MASK 0x7 +# define LC_LINK_WIDTH_X0 0 +# define LC_LINK_WIDTH_X1 1 +# define LC_LINK_WIDTH_X2 2 +# define LC_LINK_WIDTH_X4 3 +# define LC_LINK_WIDTH_X8 4 +# define LC_LINK_WIDTH_X16 6 +# define LC_LINK_WIDTH_RD_SHIFT 4 +# define LC_LINK_WIDTH_RD_MASK 0x70 +# define LC_RECONFIG_ARC_MISSING_ESCAPE (1 << 7) +# define LC_RECONFIG_NOW (1 << 8) +# define LC_RENEGOTIATION_SUPPORT (1 << 9) +# define LC_RENEGOTIATE_EN (1 << 10) +# define LC_SHORT_RECONFIG_EN (1 << 11) +# define LC_UPCONFIGURE_SUPPORT (1 << 12) +# define LC_UPCONFIGURE_DIS (1 << 13) +#define PCIE_LC_SPEED_CNTL 0xa4 /* PCIE_P */ +# define LC_GEN2_EN_STRAP (1 << 0) +# define LC_TARGET_LINK_SPEED_OVERRIDE_EN (1 << 1) +# define LC_FORCE_EN_HW_SPEED_CHANGE (1 << 5) +# define LC_FORCE_DIS_HW_SPEED_CHANGE (1 << 6) +# define LC_SPEED_CHANGE_ATTEMPTS_ALLOWED_MASK (0x3 << 8) +# define LC_SPEED_CHANGE_ATTEMPTS_ALLOWED_SHIFT 3 +# define LC_CURRENT_DATA_RATE (1 << 11) +# define LC_VOLTAGE_TIMER_SEL_MASK (0xf << 14) +# define LC_CLR_FAILED_SPD_CHANGE_CNT (1 << 21) +# define LC_OTHER_SIDE_EVER_SENT_GEN2 (1 << 23) +# define LC_OTHER_SIDE_SUPPORTS_GEN2 (1 << 24) +#define MM_CFGREGS_CNTL 0x544c +# define MM_WR_TO_CFG_EN (1 << 3) +#define LINK_CNTL2 0x88 /* F0 */ +# define TARGET_LINK_SPEED_MASK (0xf << 0) +# define SELECTABLE_DEEMPHASIS (1 << 6) + +/* + * PM4 + */ +#define PACKET_TYPE0 0 +#define PACKET_TYPE1 1 +#define PACKET_TYPE2 2 +#define PACKET_TYPE3 3 + +#define CP_PACKET_GET_TYPE(h) (((h) >> 30) & 3) +#define CP_PACKET_GET_COUNT(h) (((h) >> 16) & 0x3FFF) +#define CP_PACKET0_GET_REG(h) (((h) & 0xFFFF) << 2) +#define CP_PACKET3_GET_OPCODE(h) (((h) >> 8) & 0xFF) +#define PACKET0(reg, n) ((PACKET_TYPE0 << 30) | \ + (((reg) >> 2) & 0xFFFF) | \ + ((n) & 0x3FFF) << 16) +#define PACKET3(op, n) ((PACKET_TYPE3 << 30) | \ + (((op) & 0xFF) << 8) | \ + ((n) & 0x3FFF) << 16) + +/* Packet 3 types */ +#define PACKET3_NOP 0x10 +#define PACKET3_INDIRECT_BUFFER_END 0x17 +#define PACKET3_SET_PREDICATION 0x20 +#define PACKET3_REG_RMW 0x21 +#define PACKET3_COND_EXEC 0x22 +#define PACKET3_PRED_EXEC 0x23 +#define PACKET3_START_3D_CMDBUF 0x24 +#define PACKET3_DRAW_INDEX_2 0x27 +#define PACKET3_CONTEXT_CONTROL 0x28 +#define PACKET3_DRAW_INDEX_IMMD_BE 0x29 +#define PACKET3_INDEX_TYPE 0x2A +#define PACKET3_DRAW_INDEX 0x2B +#define PACKET3_DRAW_INDEX_AUTO 0x2D +#define PACKET3_DRAW_INDEX_IMMD 0x2E +#define PACKET3_NUM_INSTANCES 0x2F +#define PACKET3_STRMOUT_BUFFER_UPDATE 0x34 +#define PACKET3_INDIRECT_BUFFER_MP 0x38 +#define PACKET3_MEM_SEMAPHORE 0x39 +#define PACKET3_MPEG_INDEX 0x3A +#define PACKET3_WAIT_REG_MEM 0x3C +#define PACKET3_MEM_WRITE 0x3D +#define PACKET3_INDIRECT_BUFFER 0x32 +#define PACKET3_SURFACE_SYNC 0x43 +# define PACKET3_CB0_DEST_BASE_ENA (1 << 6) +# define PACKET3_TC_ACTION_ENA (1 << 23) +# define PACKET3_VC_ACTION_ENA (1 << 24) +# define PACKET3_CB_ACTION_ENA (1 << 25) +# define PACKET3_DB_ACTION_ENA (1 << 26) +# define PACKET3_SH_ACTION_ENA (1 << 27) +# define PACKET3_SMX_ACTION_ENA (1 << 28) +#define PACKET3_ME_INITIALIZE 0x44 +#define PACKET3_ME_INITIALIZE_DEVICE_ID(x) ((x) << 16) +#define PACKET3_COND_WRITE 0x45 +#define PACKET3_EVENT_WRITE 0x46 +#define EVENT_TYPE(x) ((x) << 0) +#define EVENT_INDEX(x) ((x) << 8) + /* 0 - any non-TS event + * 1 - ZPASS_DONE + * 2 - SAMPLE_PIPELINESTAT + * 3 - SAMPLE_STREAMOUTSTAT* + * 4 - *S_PARTIAL_FLUSH + * 5 - TS events + */ +#define PACKET3_EVENT_WRITE_EOP 0x47 +#define DATA_SEL(x) ((x) << 29) + /* 0 - discard + * 1 - send low 32bit data + * 2 - send 64bit data + * 3 - send 64bit counter value + */ +#define INT_SEL(x) ((x) << 24) + /* 0 - none + * 1 - interrupt only (DATA_SEL = 0) + * 2 - interrupt when data write is confirmed + */ +#define PACKET3_ONE_REG_WRITE 0x57 +#define PACKET3_SET_CONFIG_REG 0x68 +#define PACKET3_SET_CONFIG_REG_OFFSET 0x00008000 +#define PACKET3_SET_CONFIG_REG_END 0x0000ac00 +#define PACKET3_SET_CONTEXT_REG 0x69 +#define PACKET3_SET_CONTEXT_REG_OFFSET 0x00028000 +#define PACKET3_SET_CONTEXT_REG_END 0x00029000 +#define PACKET3_SET_ALU_CONST 0x6A +#define PACKET3_SET_ALU_CONST_OFFSET 0x00030000 +#define PACKET3_SET_ALU_CONST_END 0x00032000 +#define PACKET3_SET_BOOL_CONST 0x6B +#define PACKET3_SET_BOOL_CONST_OFFSET 0x0003e380 +#define PACKET3_SET_BOOL_CONST_END 0x00040000 +#define PACKET3_SET_LOOP_CONST 0x6C +#define PACKET3_SET_LOOP_CONST_OFFSET 0x0003e200 +#define PACKET3_SET_LOOP_CONST_END 0x0003e380 +#define PACKET3_SET_RESOURCE 0x6D +#define PACKET3_SET_RESOURCE_OFFSET 0x00038000 +#define PACKET3_SET_RESOURCE_END 0x0003c000 +#define PACKET3_SET_SAMPLER 0x6E +#define PACKET3_SET_SAMPLER_OFFSET 0x0003c000 +#define PACKET3_SET_SAMPLER_END 0x0003cff0 +#define PACKET3_SET_CTL_CONST 0x6F +#define PACKET3_SET_CTL_CONST_OFFSET 0x0003cff0 +#define PACKET3_SET_CTL_CONST_END 0x0003e200 +#define PACKET3_SURFACE_BASE_UPDATE 0x73 + + +#define R_008020_GRBM_SOFT_RESET 0x8020 +#define S_008020_SOFT_RESET_CP(x) (((x) & 1) << 0) +#define S_008020_SOFT_RESET_CB(x) (((x) & 1) << 1) +#define S_008020_SOFT_RESET_CR(x) (((x) & 1) << 2) +#define S_008020_SOFT_RESET_DB(x) (((x) & 1) << 3) +#define S_008020_SOFT_RESET_PA(x) (((x) & 1) << 5) +#define S_008020_SOFT_RESET_SC(x) (((x) & 1) << 6) +#define S_008020_SOFT_RESET_SMX(x) (((x) & 1) << 7) +#define S_008020_SOFT_RESET_SPI(x) (((x) & 1) << 8) +#define S_008020_SOFT_RESET_SH(x) (((x) & 1) << 9) +#define S_008020_SOFT_RESET_SX(x) (((x) & 1) << 10) +#define S_008020_SOFT_RESET_TC(x) (((x) & 1) << 11) +#define S_008020_SOFT_RESET_TA(x) (((x) & 1) << 12) +#define S_008020_SOFT_RESET_VC(x) (((x) & 1) << 13) +#define S_008020_SOFT_RESET_VGT(x) (((x) & 1) << 14) +#define R_008010_GRBM_STATUS 0x8010 +#define S_008010_CMDFIFO_AVAIL(x) (((x) & 0x1F) << 0) +#define S_008010_CP_RQ_PENDING(x) (((x) & 1) << 6) +#define S_008010_CF_RQ_PENDING(x) (((x) & 1) << 7) +#define S_008010_PF_RQ_PENDING(x) (((x) & 1) << 8) +#define S_008010_GRBM_EE_BUSY(x) (((x) & 1) << 10) +#define S_008010_VC_BUSY(x) (((x) & 1) << 11) +#define S_008010_DB03_CLEAN(x) (((x) & 1) << 12) +#define S_008010_CB03_CLEAN(x) (((x) & 1) << 13) +#define S_008010_VGT_BUSY_NO_DMA(x) (((x) & 1) << 16) +#define S_008010_VGT_BUSY(x) (((x) & 1) << 17) +#define S_008010_TA03_BUSY(x) (((x) & 1) << 18) +#define S_008010_TC_BUSY(x) (((x) & 1) << 19) +#define S_008010_SX_BUSY(x) (((x) & 1) << 20) +#define S_008010_SH_BUSY(x) (((x) & 1) << 21) +#define S_008010_SPI03_BUSY(x) (((x) & 1) << 22) +#define S_008010_SMX_BUSY(x) (((x) & 1) << 23) +#define S_008010_SC_BUSY(x) (((x) & 1) << 24) +#define S_008010_PA_BUSY(x) (((x) & 1) << 25) +#define S_008010_DB03_BUSY(x) (((x) & 1) << 26) +#define S_008010_CR_BUSY(x) (((x) & 1) << 27) +#define S_008010_CP_COHERENCY_BUSY(x) (((x) & 1) << 28) +#define S_008010_CP_BUSY(x) (((x) & 1) << 29) +#define S_008010_CB03_BUSY(x) (((x) & 1) << 30) +#define S_008010_GUI_ACTIVE(x) (((x) & 1) << 31) +#define G_008010_CMDFIFO_AVAIL(x) (((x) >> 0) & 0x1F) +#define G_008010_CP_RQ_PENDING(x) (((x) >> 6) & 1) +#define G_008010_CF_RQ_PENDING(x) (((x) >> 7) & 1) +#define G_008010_PF_RQ_PENDING(x) (((x) >> 8) & 1) +#define G_008010_GRBM_EE_BUSY(x) (((x) >> 10) & 1) +#define G_008010_VC_BUSY(x) (((x) >> 11) & 1) +#define G_008010_DB03_CLEAN(x) (((x) >> 12) & 1) +#define G_008010_CB03_CLEAN(x) (((x) >> 13) & 1) +#define G_008010_VGT_BUSY_NO_DMA(x) (((x) >> 16) & 1) +#define G_008010_VGT_BUSY(x) (((x) >> 17) & 1) +#define G_008010_TA03_BUSY(x) (((x) >> 18) & 1) +#define G_008010_TC_BUSY(x) (((x) >> 19) & 1) +#define G_008010_SX_BUSY(x) (((x) >> 20) & 1) +#define G_008010_SH_BUSY(x) (((x) >> 21) & 1) +#define G_008010_SPI03_BUSY(x) (((x) >> 22) & 1) +#define G_008010_SMX_BUSY(x) (((x) >> 23) & 1) +#define G_008010_SC_BUSY(x) (((x) >> 24) & 1) +#define G_008010_PA_BUSY(x) (((x) >> 25) & 1) +#define G_008010_DB03_BUSY(x) (((x) >> 26) & 1) +#define G_008010_CR_BUSY(x) (((x) >> 27) & 1) +#define G_008010_CP_COHERENCY_BUSY(x) (((x) >> 28) & 1) +#define G_008010_CP_BUSY(x) (((x) >> 29) & 1) +#define G_008010_CB03_BUSY(x) (((x) >> 30) & 1) +#define G_008010_GUI_ACTIVE(x) (((x) >> 31) & 1) +#define R_008014_GRBM_STATUS2 0x8014 +#define S_008014_CR_CLEAN(x) (((x) & 1) << 0) +#define S_008014_SMX_CLEAN(x) (((x) & 1) << 1) +#define S_008014_SPI0_BUSY(x) (((x) & 1) << 8) +#define S_008014_SPI1_BUSY(x) (((x) & 1) << 9) +#define S_008014_SPI2_BUSY(x) (((x) & 1) << 10) +#define S_008014_SPI3_BUSY(x) (((x) & 1) << 11) +#define S_008014_TA0_BUSY(x) (((x) & 1) << 12) +#define S_008014_TA1_BUSY(x) (((x) & 1) << 13) +#define S_008014_TA2_BUSY(x) (((x) & 1) << 14) +#define S_008014_TA3_BUSY(x) (((x) & 1) << 15) +#define S_008014_DB0_BUSY(x) (((x) & 1) << 16) +#define S_008014_DB1_BUSY(x) (((x) & 1) << 17) +#define S_008014_DB2_BUSY(x) (((x) & 1) << 18) +#define S_008014_DB3_BUSY(x) (((x) & 1) << 19) +#define S_008014_CB0_BUSY(x) (((x) & 1) << 20) +#define S_008014_CB1_BUSY(x) (((x) & 1) << 21) +#define S_008014_CB2_BUSY(x) (((x) & 1) << 22) +#define S_008014_CB3_BUSY(x) (((x) & 1) << 23) +#define G_008014_CR_CLEAN(x) (((x) >> 0) & 1) +#define G_008014_SMX_CLEAN(x) (((x) >> 1) & 1) +#define G_008014_SPI0_BUSY(x) (((x) >> 8) & 1) +#define G_008014_SPI1_BUSY(x) (((x) >> 9) & 1) +#define G_008014_SPI2_BUSY(x) (((x) >> 10) & 1) +#define G_008014_SPI3_BUSY(x) (((x) >> 11) & 1) +#define G_008014_TA0_BUSY(x) (((x) >> 12) & 1) +#define G_008014_TA1_BUSY(x) (((x) >> 13) & 1) +#define G_008014_TA2_BUSY(x) (((x) >> 14) & 1) +#define G_008014_TA3_BUSY(x) (((x) >> 15) & 1) +#define G_008014_DB0_BUSY(x) (((x) >> 16) & 1) +#define G_008014_DB1_BUSY(x) (((x) >> 17) & 1) +#define G_008014_DB2_BUSY(x) (((x) >> 18) & 1) +#define G_008014_DB3_BUSY(x) (((x) >> 19) & 1) +#define G_008014_CB0_BUSY(x) (((x) >> 20) & 1) +#define G_008014_CB1_BUSY(x) (((x) >> 21) & 1) +#define G_008014_CB2_BUSY(x) (((x) >> 22) & 1) +#define G_008014_CB3_BUSY(x) (((x) >> 23) & 1) +#define R_000E50_SRBM_STATUS 0x0E50 +#define G_000E50_RLC_RQ_PENDING(x) (((x) >> 3) & 1) +#define G_000E50_RCU_RQ_PENDING(x) (((x) >> 4) & 1) +#define G_000E50_GRBM_RQ_PENDING(x) (((x) >> 5) & 1) +#define G_000E50_HI_RQ_PENDING(x) (((x) >> 6) & 1) +#define G_000E50_IO_EXTERN_SIGNAL(x) (((x) >> 7) & 1) +#define G_000E50_VMC_BUSY(x) (((x) >> 8) & 1) +#define G_000E50_MCB_BUSY(x) (((x) >> 9) & 1) +#define G_000E50_MCDZ_BUSY(x) (((x) >> 10) & 1) +#define G_000E50_MCDY_BUSY(x) (((x) >> 11) & 1) +#define G_000E50_MCDX_BUSY(x) (((x) >> 12) & 1) +#define G_000E50_MCDW_BUSY(x) (((x) >> 13) & 1) +#define G_000E50_SEM_BUSY(x) (((x) >> 14) & 1) +#define G_000E50_RLC_BUSY(x) (((x) >> 15) & 1) +#define G_000E50_BIF_BUSY(x) (((x) >> 29) & 1) +#define R_000E60_SRBM_SOFT_RESET 0x0E60 +#define S_000E60_SOFT_RESET_BIF(x) (((x) & 1) << 1) +#define S_000E60_SOFT_RESET_CG(x) (((x) & 1) << 2) +#define S_000E60_SOFT_RESET_CMC(x) (((x) & 1) << 3) +#define S_000E60_SOFT_RESET_CSC(x) (((x) & 1) << 4) +#define S_000E60_SOFT_RESET_DC(x) (((x) & 1) << 5) +#define S_000E60_SOFT_RESET_GRBM(x) (((x) & 1) << 8) +#define S_000E60_SOFT_RESET_HDP(x) (((x) & 1) << 9) +#define S_000E60_SOFT_RESET_IH(x) (((x) & 1) << 10) +#define S_000E60_SOFT_RESET_MC(x) (((x) & 1) << 11) +#define S_000E60_SOFT_RESET_RLC(x) (((x) & 1) << 13) +#define S_000E60_SOFT_RESET_ROM(x) (((x) & 1) << 14) +#define S_000E60_SOFT_RESET_SEM(x) (((x) & 1) << 15) +#define S_000E60_SOFT_RESET_TSC(x) (((x) & 1) << 16) +#define S_000E60_SOFT_RESET_VMC(x) (((x) & 1) << 17) + +#define R_005480_HDP_MEM_COHERENCY_FLUSH_CNTL 0x5480 + +#define R_028C04_PA_SC_AA_CONFIG 0x028C04 +#define S_028C04_MSAA_NUM_SAMPLES(x) (((x) & 0x3) << 0) +#define G_028C04_MSAA_NUM_SAMPLES(x) (((x) >> 0) & 0x3) +#define C_028C04_MSAA_NUM_SAMPLES 0xFFFFFFFC +#define S_028C04_AA_MASK_CENTROID_DTMN(x) (((x) & 0x1) << 4) +#define G_028C04_AA_MASK_CENTROID_DTMN(x) (((x) >> 4) & 0x1) +#define C_028C04_AA_MASK_CENTROID_DTMN 0xFFFFFFEF +#define S_028C04_MAX_SAMPLE_DIST(x) (((x) & 0xF) << 13) +#define G_028C04_MAX_SAMPLE_DIST(x) (((x) >> 13) & 0xF) +#define C_028C04_MAX_SAMPLE_DIST 0xFFFE1FFF +#define R_0280E0_CB_COLOR0_FRAG 0x0280E0 +#define S_0280E0_BASE_256B(x) (((x) & 0xFFFFFFFF) << 0) +#define G_0280E0_BASE_256B(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_0280E0_BASE_256B 0x00000000 +#define R_0280E4_CB_COLOR1_FRAG 0x0280E4 +#define R_0280E8_CB_COLOR2_FRAG 0x0280E8 +#define R_0280EC_CB_COLOR3_FRAG 0x0280EC +#define R_0280F0_CB_COLOR4_FRAG 0x0280F0 +#define R_0280F4_CB_COLOR5_FRAG 0x0280F4 +#define R_0280F8_CB_COLOR6_FRAG 0x0280F8 +#define R_0280FC_CB_COLOR7_FRAG 0x0280FC +#define R_0280C0_CB_COLOR0_TILE 0x0280C0 +#define S_0280C0_BASE_256B(x) (((x) & 0xFFFFFFFF) << 0) +#define G_0280C0_BASE_256B(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_0280C0_BASE_256B 0x00000000 +#define R_0280C4_CB_COLOR1_TILE 0x0280C4 +#define R_0280C8_CB_COLOR2_TILE 0x0280C8 +#define R_0280CC_CB_COLOR3_TILE 0x0280CC +#define R_0280D0_CB_COLOR4_TILE 0x0280D0 +#define R_0280D4_CB_COLOR5_TILE 0x0280D4 +#define R_0280D8_CB_COLOR6_TILE 0x0280D8 +#define R_0280DC_CB_COLOR7_TILE 0x0280DC +#define R_0280A0_CB_COLOR0_INFO 0x0280A0 +#define S_0280A0_ENDIAN(x) (((x) & 0x3) << 0) +#define G_0280A0_ENDIAN(x) (((x) >> 0) & 0x3) +#define C_0280A0_ENDIAN 0xFFFFFFFC +#define S_0280A0_FORMAT(x) (((x) & 0x3F) << 2) +#define G_0280A0_FORMAT(x) (((x) >> 2) & 0x3F) +#define C_0280A0_FORMAT 0xFFFFFF03 +#define V_0280A0_COLOR_INVALID 0x00000000 +#define V_0280A0_COLOR_8 0x00000001 +#define V_0280A0_COLOR_4_4 0x00000002 +#define V_0280A0_COLOR_3_3_2 0x00000003 +#define V_0280A0_COLOR_16 0x00000005 +#define V_0280A0_COLOR_16_FLOAT 0x00000006 +#define V_0280A0_COLOR_8_8 0x00000007 +#define V_0280A0_COLOR_5_6_5 0x00000008 +#define V_0280A0_COLOR_6_5_5 0x00000009 +#define V_0280A0_COLOR_1_5_5_5 0x0000000A +#define V_0280A0_COLOR_4_4_4_4 0x0000000B +#define V_0280A0_COLOR_5_5_5_1 0x0000000C +#define V_0280A0_COLOR_32 0x0000000D +#define V_0280A0_COLOR_32_FLOAT 0x0000000E +#define V_0280A0_COLOR_16_16 0x0000000F +#define V_0280A0_COLOR_16_16_FLOAT 0x00000010 +#define V_0280A0_COLOR_8_24 0x00000011 +#define V_0280A0_COLOR_8_24_FLOAT 0x00000012 +#define V_0280A0_COLOR_24_8 0x00000013 +#define V_0280A0_COLOR_24_8_FLOAT 0x00000014 +#define V_0280A0_COLOR_10_11_11 0x00000015 +#define V_0280A0_COLOR_10_11_11_FLOAT 0x00000016 +#define V_0280A0_COLOR_11_11_10 0x00000017 +#define V_0280A0_COLOR_11_11_10_FLOAT 0x00000018 +#define V_0280A0_COLOR_2_10_10_10 0x00000019 +#define V_0280A0_COLOR_8_8_8_8 0x0000001A +#define V_0280A0_COLOR_10_10_10_2 0x0000001B +#define V_0280A0_COLOR_X24_8_32_FLOAT 0x0000001C +#define V_0280A0_COLOR_32_32 0x0000001D +#define V_0280A0_COLOR_32_32_FLOAT 0x0000001E +#define V_0280A0_COLOR_16_16_16_16 0x0000001F +#define V_0280A0_COLOR_16_16_16_16_FLOAT 0x00000020 +#define V_0280A0_COLOR_32_32_32_32 0x00000022 +#define V_0280A0_COLOR_32_32_32_32_FLOAT 0x00000023 +#define S_0280A0_ARRAY_MODE(x) (((x) & 0xF) << 8) +#define G_0280A0_ARRAY_MODE(x) (((x) >> 8) & 0xF) +#define C_0280A0_ARRAY_MODE 0xFFFFF0FF +#define V_0280A0_ARRAY_LINEAR_GENERAL 0x00000000 +#define V_0280A0_ARRAY_LINEAR_ALIGNED 0x00000001 +#define V_0280A0_ARRAY_1D_TILED_THIN1 0x00000002 +#define V_0280A0_ARRAY_2D_TILED_THIN1 0x00000004 +#define S_0280A0_NUMBER_TYPE(x) (((x) & 0x7) << 12) +#define G_0280A0_NUMBER_TYPE(x) (((x) >> 12) & 0x7) +#define C_0280A0_NUMBER_TYPE 0xFFFF8FFF +#define S_0280A0_READ_SIZE(x) (((x) & 0x1) << 15) +#define G_0280A0_READ_SIZE(x) (((x) >> 15) & 0x1) +#define C_0280A0_READ_SIZE 0xFFFF7FFF +#define S_0280A0_COMP_SWAP(x) (((x) & 0x3) << 16) +#define G_0280A0_COMP_SWAP(x) (((x) >> 16) & 0x3) +#define C_0280A0_COMP_SWAP 0xFFFCFFFF +#define S_0280A0_TILE_MODE(x) (((x) & 0x3) << 18) +#define G_0280A0_TILE_MODE(x) (((x) >> 18) & 0x3) +#define C_0280A0_TILE_MODE 0xFFF3FFFF +#define S_0280A0_BLEND_CLAMP(x) (((x) & 0x1) << 20) +#define G_0280A0_BLEND_CLAMP(x) (((x) >> 20) & 0x1) +#define C_0280A0_BLEND_CLAMP 0xFFEFFFFF +#define S_0280A0_CLEAR_COLOR(x) (((x) & 0x1) << 21) +#define G_0280A0_CLEAR_COLOR(x) (((x) >> 21) & 0x1) +#define C_0280A0_CLEAR_COLOR 0xFFDFFFFF +#define S_0280A0_BLEND_BYPASS(x) (((x) & 0x1) << 22) +#define G_0280A0_BLEND_BYPASS(x) (((x) >> 22) & 0x1) +#define C_0280A0_BLEND_BYPASS 0xFFBFFFFF +#define S_0280A0_BLEND_FLOAT32(x) (((x) & 0x1) << 23) +#define G_0280A0_BLEND_FLOAT32(x) (((x) >> 23) & 0x1) +#define C_0280A0_BLEND_FLOAT32 0xFF7FFFFF +#define S_0280A0_SIMPLE_FLOAT(x) (((x) & 0x1) << 24) +#define G_0280A0_SIMPLE_FLOAT(x) (((x) >> 24) & 0x1) +#define C_0280A0_SIMPLE_FLOAT 0xFEFFFFFF +#define S_0280A0_ROUND_MODE(x) (((x) & 0x1) << 25) +#define G_0280A0_ROUND_MODE(x) (((x) >> 25) & 0x1) +#define C_0280A0_ROUND_MODE 0xFDFFFFFF +#define S_0280A0_TILE_COMPACT(x) (((x) & 0x1) << 26) +#define G_0280A0_TILE_COMPACT(x) (((x) >> 26) & 0x1) +#define C_0280A0_TILE_COMPACT 0xFBFFFFFF +#define S_0280A0_SOURCE_FORMAT(x) (((x) & 0x1) << 27) +#define G_0280A0_SOURCE_FORMAT(x) (((x) >> 27) & 0x1) +#define C_0280A0_SOURCE_FORMAT 0xF7FFFFFF +#define R_0280A4_CB_COLOR1_INFO 0x0280A4 +#define R_0280A8_CB_COLOR2_INFO 0x0280A8 +#define R_0280AC_CB_COLOR3_INFO 0x0280AC +#define R_0280B0_CB_COLOR4_INFO 0x0280B0 +#define R_0280B4_CB_COLOR5_INFO 0x0280B4 +#define R_0280B8_CB_COLOR6_INFO 0x0280B8 +#define R_0280BC_CB_COLOR7_INFO 0x0280BC +#define R_028060_CB_COLOR0_SIZE 0x028060 +#define S_028060_PITCH_TILE_MAX(x) (((x) & 0x3FF) << 0) +#define G_028060_PITCH_TILE_MAX(x) (((x) >> 0) & 0x3FF) +#define C_028060_PITCH_TILE_MAX 0xFFFFFC00 +#define S_028060_SLICE_TILE_MAX(x) (((x) & 0xFFFFF) << 10) +#define G_028060_SLICE_TILE_MAX(x) (((x) >> 10) & 0xFFFFF) +#define C_028060_SLICE_TILE_MAX 0xC00003FF +#define R_028064_CB_COLOR1_SIZE 0x028064 +#define R_028068_CB_COLOR2_SIZE 0x028068 +#define R_02806C_CB_COLOR3_SIZE 0x02806C +#define R_028070_CB_COLOR4_SIZE 0x028070 +#define R_028074_CB_COLOR5_SIZE 0x028074 +#define R_028078_CB_COLOR6_SIZE 0x028078 +#define R_02807C_CB_COLOR7_SIZE 0x02807C +#define R_028238_CB_TARGET_MASK 0x028238 +#define S_028238_TARGET0_ENABLE(x) (((x) & 0xF) << 0) +#define G_028238_TARGET0_ENABLE(x) (((x) >> 0) & 0xF) +#define C_028238_TARGET0_ENABLE 0xFFFFFFF0 +#define S_028238_TARGET1_ENABLE(x) (((x) & 0xF) << 4) +#define G_028238_TARGET1_ENABLE(x) (((x) >> 4) & 0xF) +#define C_028238_TARGET1_ENABLE 0xFFFFFF0F +#define S_028238_TARGET2_ENABLE(x) (((x) & 0xF) << 8) +#define G_028238_TARGET2_ENABLE(x) (((x) >> 8) & 0xF) +#define C_028238_TARGET2_ENABLE 0xFFFFF0FF +#define S_028238_TARGET3_ENABLE(x) (((x) & 0xF) << 12) +#define G_028238_TARGET3_ENABLE(x) (((x) >> 12) & 0xF) +#define C_028238_TARGET3_ENABLE 0xFFFF0FFF +#define S_028238_TARGET4_ENABLE(x) (((x) & 0xF) << 16) +#define G_028238_TARGET4_ENABLE(x) (((x) >> 16) & 0xF) +#define C_028238_TARGET4_ENABLE 0xFFF0FFFF +#define S_028238_TARGET5_ENABLE(x) (((x) & 0xF) << 20) +#define G_028238_TARGET5_ENABLE(x) (((x) >> 20) & 0xF) +#define C_028238_TARGET5_ENABLE 0xFF0FFFFF +#define S_028238_TARGET6_ENABLE(x) (((x) & 0xF) << 24) +#define G_028238_TARGET6_ENABLE(x) (((x) >> 24) & 0xF) +#define C_028238_TARGET6_ENABLE 0xF0FFFFFF +#define S_028238_TARGET7_ENABLE(x) (((x) & 0xF) << 28) +#define G_028238_TARGET7_ENABLE(x) (((x) >> 28) & 0xF) +#define C_028238_TARGET7_ENABLE 0x0FFFFFFF +#define R_02823C_CB_SHADER_MASK 0x02823C +#define S_02823C_OUTPUT0_ENABLE(x) (((x) & 0xF) << 0) +#define G_02823C_OUTPUT0_ENABLE(x) (((x) >> 0) & 0xF) +#define C_02823C_OUTPUT0_ENABLE 0xFFFFFFF0 +#define S_02823C_OUTPUT1_ENABLE(x) (((x) & 0xF) << 4) +#define G_02823C_OUTPUT1_ENABLE(x) (((x) >> 4) & 0xF) +#define C_02823C_OUTPUT1_ENABLE 0xFFFFFF0F +#define S_02823C_OUTPUT2_ENABLE(x) (((x) & 0xF) << 8) +#define G_02823C_OUTPUT2_ENABLE(x) (((x) >> 8) & 0xF) +#define C_02823C_OUTPUT2_ENABLE 0xFFFFF0FF +#define S_02823C_OUTPUT3_ENABLE(x) (((x) & 0xF) << 12) +#define G_02823C_OUTPUT3_ENABLE(x) (((x) >> 12) & 0xF) +#define C_02823C_OUTPUT3_ENABLE 0xFFFF0FFF +#define S_02823C_OUTPUT4_ENABLE(x) (((x) & 0xF) << 16) +#define G_02823C_OUTPUT4_ENABLE(x) (((x) >> 16) & 0xF) +#define C_02823C_OUTPUT4_ENABLE 0xFFF0FFFF +#define S_02823C_OUTPUT5_ENABLE(x) (((x) & 0xF) << 20) +#define G_02823C_OUTPUT5_ENABLE(x) (((x) >> 20) & 0xF) +#define C_02823C_OUTPUT5_ENABLE 0xFF0FFFFF +#define S_02823C_OUTPUT6_ENABLE(x) (((x) & 0xF) << 24) +#define G_02823C_OUTPUT6_ENABLE(x) (((x) >> 24) & 0xF) +#define C_02823C_OUTPUT6_ENABLE 0xF0FFFFFF +#define S_02823C_OUTPUT7_ENABLE(x) (((x) & 0xF) << 28) +#define G_02823C_OUTPUT7_ENABLE(x) (((x) >> 28) & 0xF) +#define C_02823C_OUTPUT7_ENABLE 0x0FFFFFFF +#define R_028AB0_VGT_STRMOUT_EN 0x028AB0 +#define S_028AB0_STREAMOUT(x) (((x) & 0x1) << 0) +#define G_028AB0_STREAMOUT(x) (((x) >> 0) & 0x1) +#define C_028AB0_STREAMOUT 0xFFFFFFFE +#define R_028B20_VGT_STRMOUT_BUFFER_EN 0x028B20 +#define S_028B20_BUFFER_0_EN(x) (((x) & 0x1) << 0) +#define G_028B20_BUFFER_0_EN(x) (((x) >> 0) & 0x1) +#define C_028B20_BUFFER_0_EN 0xFFFFFFFE +#define S_028B20_BUFFER_1_EN(x) (((x) & 0x1) << 1) +#define G_028B20_BUFFER_1_EN(x) (((x) >> 1) & 0x1) +#define C_028B20_BUFFER_1_EN 0xFFFFFFFD +#define S_028B20_BUFFER_2_EN(x) (((x) & 0x1) << 2) +#define G_028B20_BUFFER_2_EN(x) (((x) >> 2) & 0x1) +#define C_028B20_BUFFER_2_EN 0xFFFFFFFB +#define S_028B20_BUFFER_3_EN(x) (((x) & 0x1) << 3) +#define G_028B20_BUFFER_3_EN(x) (((x) >> 3) & 0x1) +#define C_028B20_BUFFER_3_EN 0xFFFFFFF7 +#define S_028B20_SIZE(x) (((x) & 0xFFFFFFFF) << 0) +#define G_028B20_SIZE(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_028B20_SIZE 0x00000000 +#define R_038000_SQ_TEX_RESOURCE_WORD0_0 0x038000 +#define S_038000_DIM(x) (((x) & 0x7) << 0) +#define G_038000_DIM(x) (((x) >> 0) & 0x7) +#define C_038000_DIM 0xFFFFFFF8 +#define V_038000_SQ_TEX_DIM_1D 0x00000000 +#define V_038000_SQ_TEX_DIM_2D 0x00000001 +#define V_038000_SQ_TEX_DIM_3D 0x00000002 +#define V_038000_SQ_TEX_DIM_CUBEMAP 0x00000003 +#define V_038000_SQ_TEX_DIM_1D_ARRAY 0x00000004 +#define V_038000_SQ_TEX_DIM_2D_ARRAY 0x00000005 +#define V_038000_SQ_TEX_DIM_2D_MSAA 0x00000006 +#define V_038000_SQ_TEX_DIM_2D_ARRAY_MSAA 0x00000007 +#define S_038000_TILE_MODE(x) (((x) & 0xF) << 3) +#define G_038000_TILE_MODE(x) (((x) >> 3) & 0xF) +#define C_038000_TILE_MODE 0xFFFFFF87 +#define V_038000_ARRAY_LINEAR_GENERAL 0x00000000 +#define V_038000_ARRAY_LINEAR_ALIGNED 0x00000001 +#define V_038000_ARRAY_1D_TILED_THIN1 0x00000002 +#define V_038000_ARRAY_2D_TILED_THIN1 0x00000004 +#define S_038000_TILE_TYPE(x) (((x) & 0x1) << 7) +#define G_038000_TILE_TYPE(x) (((x) >> 7) & 0x1) +#define C_038000_TILE_TYPE 0xFFFFFF7F +#define S_038000_PITCH(x) (((x) & 0x7FF) << 8) +#define G_038000_PITCH(x) (((x) >> 8) & 0x7FF) +#define C_038000_PITCH 0xFFF800FF +#define S_038000_TEX_WIDTH(x) (((x) & 0x1FFF) << 19) +#define G_038000_TEX_WIDTH(x) (((x) >> 19) & 0x1FFF) +#define C_038000_TEX_WIDTH 0x0007FFFF +#define R_038004_SQ_TEX_RESOURCE_WORD1_0 0x038004 +#define S_038004_TEX_HEIGHT(x) (((x) & 0x1FFF) << 0) +#define G_038004_TEX_HEIGHT(x) (((x) >> 0) & 0x1FFF) +#define C_038004_TEX_HEIGHT 0xFFFFE000 +#define S_038004_TEX_DEPTH(x) (((x) & 0x1FFF) << 13) +#define G_038004_TEX_DEPTH(x) (((x) >> 13) & 0x1FFF) +#define C_038004_TEX_DEPTH 0xFC001FFF +#define S_038004_DATA_FORMAT(x) (((x) & 0x3F) << 26) +#define G_038004_DATA_FORMAT(x) (((x) >> 26) & 0x3F) +#define C_038004_DATA_FORMAT 0x03FFFFFF +#define V_038004_COLOR_INVALID 0x00000000 +#define V_038004_COLOR_8 0x00000001 +#define V_038004_COLOR_4_4 0x00000002 +#define V_038004_COLOR_3_3_2 0x00000003 +#define V_038004_COLOR_16 0x00000005 +#define V_038004_COLOR_16_FLOAT 0x00000006 +#define V_038004_COLOR_8_8 0x00000007 +#define V_038004_COLOR_5_6_5 0x00000008 +#define V_038004_COLOR_6_5_5 0x00000009 +#define V_038004_COLOR_1_5_5_5 0x0000000A +#define V_038004_COLOR_4_4_4_4 0x0000000B +#define V_038004_COLOR_5_5_5_1 0x0000000C +#define V_038004_COLOR_32 0x0000000D +#define V_038004_COLOR_32_FLOAT 0x0000000E +#define V_038004_COLOR_16_16 0x0000000F +#define V_038004_COLOR_16_16_FLOAT 0x00000010 +#define V_038004_COLOR_8_24 0x00000011 +#define V_038004_COLOR_8_24_FLOAT 0x00000012 +#define V_038004_COLOR_24_8 0x00000013 +#define V_038004_COLOR_24_8_FLOAT 0x00000014 +#define V_038004_COLOR_10_11_11 0x00000015 +#define V_038004_COLOR_10_11_11_FLOAT 0x00000016 +#define V_038004_COLOR_11_11_10 0x00000017 +#define V_038004_COLOR_11_11_10_FLOAT 0x00000018 +#define V_038004_COLOR_2_10_10_10 0x00000019 +#define V_038004_COLOR_8_8_8_8 0x0000001A +#define V_038004_COLOR_10_10_10_2 0x0000001B +#define V_038004_COLOR_X24_8_32_FLOAT 0x0000001C +#define V_038004_COLOR_32_32 0x0000001D +#define V_038004_COLOR_32_32_FLOAT 0x0000001E +#define V_038004_COLOR_16_16_16_16 0x0000001F +#define V_038004_COLOR_16_16_16_16_FLOAT 0x00000020 +#define V_038004_COLOR_32_32_32_32 0x00000022 +#define V_038004_COLOR_32_32_32_32_FLOAT 0x00000023 +#define V_038004_FMT_1 0x00000025 +#define V_038004_FMT_GB_GR 0x00000027 +#define V_038004_FMT_BG_RG 0x00000028 +#define V_038004_FMT_32_AS_8 0x00000029 +#define V_038004_FMT_32_AS_8_8 0x0000002A +#define V_038004_FMT_5_9_9_9_SHAREDEXP 0x0000002B +#define V_038004_FMT_8_8_8 0x0000002C +#define V_038004_FMT_16_16_16 0x0000002D +#define V_038004_FMT_16_16_16_FLOAT 0x0000002E +#define V_038004_FMT_32_32_32 0x0000002F +#define V_038004_FMT_32_32_32_FLOAT 0x00000030 +#define V_038004_FMT_BC1 0x00000031 +#define V_038004_FMT_BC2 0x00000032 +#define V_038004_FMT_BC3 0x00000033 +#define V_038004_FMT_BC4 0x00000034 +#define V_038004_FMT_BC5 0x00000035 +#define V_038004_FMT_BC6 0x00000036 +#define V_038004_FMT_BC7 0x00000037 +#define V_038004_FMT_32_AS_32_32_32_32 0x00000038 +#define R_038010_SQ_TEX_RESOURCE_WORD4_0 0x038010 +#define S_038010_FORMAT_COMP_X(x) (((x) & 0x3) << 0) +#define G_038010_FORMAT_COMP_X(x) (((x) >> 0) & 0x3) +#define C_038010_FORMAT_COMP_X 0xFFFFFFFC +#define S_038010_FORMAT_COMP_Y(x) (((x) & 0x3) << 2) +#define G_038010_FORMAT_COMP_Y(x) (((x) >> 2) & 0x3) +#define C_038010_FORMAT_COMP_Y 0xFFFFFFF3 +#define S_038010_FORMAT_COMP_Z(x) (((x) & 0x3) << 4) +#define G_038010_FORMAT_COMP_Z(x) (((x) >> 4) & 0x3) +#define C_038010_FORMAT_COMP_Z 0xFFFFFFCF +#define S_038010_FORMAT_COMP_W(x) (((x) & 0x3) << 6) +#define G_038010_FORMAT_COMP_W(x) (((x) >> 6) & 0x3) +#define C_038010_FORMAT_COMP_W 0xFFFFFF3F +#define S_038010_NUM_FORMAT_ALL(x) (((x) & 0x3) << 8) +#define G_038010_NUM_FORMAT_ALL(x) (((x) >> 8) & 0x3) +#define C_038010_NUM_FORMAT_ALL 0xFFFFFCFF +#define S_038010_SRF_MODE_ALL(x) (((x) & 0x1) << 10) +#define G_038010_SRF_MODE_ALL(x) (((x) >> 10) & 0x1) +#define C_038010_SRF_MODE_ALL 0xFFFFFBFF +#define S_038010_FORCE_DEGAMMA(x) (((x) & 0x1) << 11) +#define G_038010_FORCE_DEGAMMA(x) (((x) >> 11) & 0x1) +#define C_038010_FORCE_DEGAMMA 0xFFFFF7FF +#define S_038010_ENDIAN_SWAP(x) (((x) & 0x3) << 12) +#define G_038010_ENDIAN_SWAP(x) (((x) >> 12) & 0x3) +#define C_038010_ENDIAN_SWAP 0xFFFFCFFF +#define S_038010_REQUEST_SIZE(x) (((x) & 0x3) << 14) +#define G_038010_REQUEST_SIZE(x) (((x) >> 14) & 0x3) +#define C_038010_REQUEST_SIZE 0xFFFF3FFF +#define S_038010_DST_SEL_X(x) (((x) & 0x7) << 16) +#define G_038010_DST_SEL_X(x) (((x) >> 16) & 0x7) +#define C_038010_DST_SEL_X 0xFFF8FFFF +#define S_038010_DST_SEL_Y(x) (((x) & 0x7) << 19) +#define G_038010_DST_SEL_Y(x) (((x) >> 19) & 0x7) +#define C_038010_DST_SEL_Y 0xFFC7FFFF +#define S_038010_DST_SEL_Z(x) (((x) & 0x7) << 22) +#define G_038010_DST_SEL_Z(x) (((x) >> 22) & 0x7) +#define C_038010_DST_SEL_Z 0xFE3FFFFF +#define S_038010_DST_SEL_W(x) (((x) & 0x7) << 25) +#define G_038010_DST_SEL_W(x) (((x) >> 25) & 0x7) +#define C_038010_DST_SEL_W 0xF1FFFFFF +#define S_038010_BASE_LEVEL(x) (((x) & 0xF) << 28) +#define G_038010_BASE_LEVEL(x) (((x) >> 28) & 0xF) +#define C_038010_BASE_LEVEL 0x0FFFFFFF +#define R_038014_SQ_TEX_RESOURCE_WORD5_0 0x038014 +#define S_038014_LAST_LEVEL(x) (((x) & 0xF) << 0) +#define G_038014_LAST_LEVEL(x) (((x) >> 0) & 0xF) +#define C_038014_LAST_LEVEL 0xFFFFFFF0 +#define S_038014_BASE_ARRAY(x) (((x) & 0x1FFF) << 4) +#define G_038014_BASE_ARRAY(x) (((x) >> 4) & 0x1FFF) +#define C_038014_BASE_ARRAY 0xFFFE000F +#define S_038014_LAST_ARRAY(x) (((x) & 0x1FFF) << 17) +#define G_038014_LAST_ARRAY(x) (((x) >> 17) & 0x1FFF) +#define C_038014_LAST_ARRAY 0xC001FFFF +#define R_0288A8_SQ_ESGS_RING_ITEMSIZE 0x0288A8 +#define S_0288A8_ITEMSIZE(x) (((x) & 0x7FFF) << 0) +#define G_0288A8_ITEMSIZE(x) (((x) >> 0) & 0x7FFF) +#define C_0288A8_ITEMSIZE 0xFFFF8000 +#define R_008C44_SQ_ESGS_RING_SIZE 0x008C44 +#define S_008C44_MEM_SIZE(x) (((x) & 0xFFFFFFFF) << 0) +#define G_008C44_MEM_SIZE(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_008C44_MEM_SIZE 0x00000000 +#define R_0288B0_SQ_ESTMP_RING_ITEMSIZE 0x0288B0 +#define S_0288B0_ITEMSIZE(x) (((x) & 0x7FFF) << 0) +#define G_0288B0_ITEMSIZE(x) (((x) >> 0) & 0x7FFF) +#define C_0288B0_ITEMSIZE 0xFFFF8000 +#define R_008C54_SQ_ESTMP_RING_SIZE 0x008C54 +#define S_008C54_MEM_SIZE(x) (((x) & 0xFFFFFFFF) << 0) +#define G_008C54_MEM_SIZE(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_008C54_MEM_SIZE 0x00000000 +#define R_0288C0_SQ_FBUF_RING_ITEMSIZE 0x0288C0 +#define S_0288C0_ITEMSIZE(x) (((x) & 0x7FFF) << 0) +#define G_0288C0_ITEMSIZE(x) (((x) >> 0) & 0x7FFF) +#define C_0288C0_ITEMSIZE 0xFFFF8000 +#define R_008C74_SQ_FBUF_RING_SIZE 0x008C74 +#define S_008C74_MEM_SIZE(x) (((x) & 0xFFFFFFFF) << 0) +#define G_008C74_MEM_SIZE(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_008C74_MEM_SIZE 0x00000000 +#define R_0288B4_SQ_GSTMP_RING_ITEMSIZE 0x0288B4 +#define S_0288B4_ITEMSIZE(x) (((x) & 0x7FFF) << 0) +#define G_0288B4_ITEMSIZE(x) (((x) >> 0) & 0x7FFF) +#define C_0288B4_ITEMSIZE 0xFFFF8000 +#define R_008C5C_SQ_GSTMP_RING_SIZE 0x008C5C +#define S_008C5C_MEM_SIZE(x) (((x) & 0xFFFFFFFF) << 0) +#define G_008C5C_MEM_SIZE(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_008C5C_MEM_SIZE 0x00000000 +#define R_0288AC_SQ_GSVS_RING_ITEMSIZE 0x0288AC +#define S_0288AC_ITEMSIZE(x) (((x) & 0x7FFF) << 0) +#define G_0288AC_ITEMSIZE(x) (((x) >> 0) & 0x7FFF) +#define C_0288AC_ITEMSIZE 0xFFFF8000 +#define R_008C4C_SQ_GSVS_RING_SIZE 0x008C4C +#define S_008C4C_MEM_SIZE(x) (((x) & 0xFFFFFFFF) << 0) +#define G_008C4C_MEM_SIZE(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_008C4C_MEM_SIZE 0x00000000 +#define R_0288BC_SQ_PSTMP_RING_ITEMSIZE 0x0288BC +#define S_0288BC_ITEMSIZE(x) (((x) & 0x7FFF) << 0) +#define G_0288BC_ITEMSIZE(x) (((x) >> 0) & 0x7FFF) +#define C_0288BC_ITEMSIZE 0xFFFF8000 +#define R_008C6C_SQ_PSTMP_RING_SIZE 0x008C6C +#define S_008C6C_MEM_SIZE(x) (((x) & 0xFFFFFFFF) << 0) +#define G_008C6C_MEM_SIZE(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_008C6C_MEM_SIZE 0x00000000 +#define R_0288C4_SQ_REDUC_RING_ITEMSIZE 0x0288C4 +#define S_0288C4_ITEMSIZE(x) (((x) & 0x7FFF) << 0) +#define G_0288C4_ITEMSIZE(x) (((x) >> 0) & 0x7FFF) +#define C_0288C4_ITEMSIZE 0xFFFF8000 +#define R_008C7C_SQ_REDUC_RING_SIZE 0x008C7C +#define S_008C7C_MEM_SIZE(x) (((x) & 0xFFFFFFFF) << 0) +#define G_008C7C_MEM_SIZE(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_008C7C_MEM_SIZE 0x00000000 +#define R_0288B8_SQ_VSTMP_RING_ITEMSIZE 0x0288B8 +#define S_0288B8_ITEMSIZE(x) (((x) & 0x7FFF) << 0) +#define G_0288B8_ITEMSIZE(x) (((x) >> 0) & 0x7FFF) +#define C_0288B8_ITEMSIZE 0xFFFF8000 +#define R_008C64_SQ_VSTMP_RING_SIZE 0x008C64 +#define S_008C64_MEM_SIZE(x) (((x) & 0xFFFFFFFF) << 0) +#define G_008C64_MEM_SIZE(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_008C64_MEM_SIZE 0x00000000 +#define R_0288C8_SQ_GS_VERT_ITEMSIZE 0x0288C8 +#define S_0288C8_ITEMSIZE(x) (((x) & 0x7FFF) << 0) +#define G_0288C8_ITEMSIZE(x) (((x) >> 0) & 0x7FFF) +#define C_0288C8_ITEMSIZE 0xFFFF8000 +#define R_028010_DB_DEPTH_INFO 0x028010 +#define S_028010_FORMAT(x) (((x) & 0x7) << 0) +#define G_028010_FORMAT(x) (((x) >> 0) & 0x7) +#define C_028010_FORMAT 0xFFFFFFF8 +#define V_028010_DEPTH_INVALID 0x00000000 +#define V_028010_DEPTH_16 0x00000001 +#define V_028010_DEPTH_X8_24 0x00000002 +#define V_028010_DEPTH_8_24 0x00000003 +#define V_028010_DEPTH_X8_24_FLOAT 0x00000004 +#define V_028010_DEPTH_8_24_FLOAT 0x00000005 +#define V_028010_DEPTH_32_FLOAT 0x00000006 +#define V_028010_DEPTH_X24_8_32_FLOAT 0x00000007 +#define S_028010_READ_SIZE(x) (((x) & 0x1) << 3) +#define G_028010_READ_SIZE(x) (((x) >> 3) & 0x1) +#define C_028010_READ_SIZE 0xFFFFFFF7 +#define S_028010_ARRAY_MODE(x) (((x) & 0xF) << 15) +#define G_028010_ARRAY_MODE(x) (((x) >> 15) & 0xF) +#define C_028010_ARRAY_MODE 0xFFF87FFF +#define V_028010_ARRAY_1D_TILED_THIN1 0x00000002 +#define V_028010_ARRAY_2D_TILED_THIN1 0x00000004 +#define S_028010_TILE_SURFACE_ENABLE(x) (((x) & 0x1) << 25) +#define G_028010_TILE_SURFACE_ENABLE(x) (((x) >> 25) & 0x1) +#define C_028010_TILE_SURFACE_ENABLE 0xFDFFFFFF +#define S_028010_TILE_COMPACT(x) (((x) & 0x1) << 26) +#define G_028010_TILE_COMPACT(x) (((x) >> 26) & 0x1) +#define C_028010_TILE_COMPACT 0xFBFFFFFF +#define S_028010_ZRANGE_PRECISION(x) (((x) & 0x1) << 31) +#define G_028010_ZRANGE_PRECISION(x) (((x) >> 31) & 0x1) +#define C_028010_ZRANGE_PRECISION 0x7FFFFFFF +#define R_028000_DB_DEPTH_SIZE 0x028000 +#define S_028000_PITCH_TILE_MAX(x) (((x) & 0x3FF) << 0) +#define G_028000_PITCH_TILE_MAX(x) (((x) >> 0) & 0x3FF) +#define C_028000_PITCH_TILE_MAX 0xFFFFFC00 +#define S_028000_SLICE_TILE_MAX(x) (((x) & 0xFFFFF) << 10) +#define G_028000_SLICE_TILE_MAX(x) (((x) >> 10) & 0xFFFFF) +#define C_028000_SLICE_TILE_MAX 0xC00003FF +#define R_028004_DB_DEPTH_VIEW 0x028004 +#define S_028004_SLICE_START(x) (((x) & 0x7FF) << 0) +#define G_028004_SLICE_START(x) (((x) >> 0) & 0x7FF) +#define C_028004_SLICE_START 0xFFFFF800 +#define S_028004_SLICE_MAX(x) (((x) & 0x7FF) << 13) +#define G_028004_SLICE_MAX(x) (((x) >> 13) & 0x7FF) +#define C_028004_SLICE_MAX 0xFF001FFF +#define R_028800_DB_DEPTH_CONTROL 0x028800 +#define S_028800_STENCIL_ENABLE(x) (((x) & 0x1) << 0) +#define G_028800_STENCIL_ENABLE(x) (((x) >> 0) & 0x1) +#define C_028800_STENCIL_ENABLE 0xFFFFFFFE +#define S_028800_Z_ENABLE(x) (((x) & 0x1) << 1) +#define G_028800_Z_ENABLE(x) (((x) >> 1) & 0x1) +#define C_028800_Z_ENABLE 0xFFFFFFFD +#define S_028800_Z_WRITE_ENABLE(x) (((x) & 0x1) << 2) +#define G_028800_Z_WRITE_ENABLE(x) (((x) >> 2) & 0x1) +#define C_028800_Z_WRITE_ENABLE 0xFFFFFFFB +#define S_028800_ZFUNC(x) (((x) & 0x7) << 4) +#define G_028800_ZFUNC(x) (((x) >> 4) & 0x7) +#define C_028800_ZFUNC 0xFFFFFF8F +#define S_028800_BACKFACE_ENABLE(x) (((x) & 0x1) << 7) +#define G_028800_BACKFACE_ENABLE(x) (((x) >> 7) & 0x1) +#define C_028800_BACKFACE_ENABLE 0xFFFFFF7F +#define S_028800_STENCILFUNC(x) (((x) & 0x7) << 8) +#define G_028800_STENCILFUNC(x) (((x) >> 8) & 0x7) +#define C_028800_STENCILFUNC 0xFFFFF8FF +#define S_028800_STENCILFAIL(x) (((x) & 0x7) << 11) +#define G_028800_STENCILFAIL(x) (((x) >> 11) & 0x7) +#define C_028800_STENCILFAIL 0xFFFFC7FF +#define S_028800_STENCILZPASS(x) (((x) & 0x7) << 14) +#define G_028800_STENCILZPASS(x) (((x) >> 14) & 0x7) +#define C_028800_STENCILZPASS 0xFFFE3FFF +#define S_028800_STENCILZFAIL(x) (((x) & 0x7) << 17) +#define G_028800_STENCILZFAIL(x) (((x) >> 17) & 0x7) +#define C_028800_STENCILZFAIL 0xFFF1FFFF +#define S_028800_STENCILFUNC_BF(x) (((x) & 0x7) << 20) +#define G_028800_STENCILFUNC_BF(x) (((x) >> 20) & 0x7) +#define C_028800_STENCILFUNC_BF 0xFF8FFFFF +#define S_028800_STENCILFAIL_BF(x) (((x) & 0x7) << 23) +#define G_028800_STENCILFAIL_BF(x) (((x) >> 23) & 0x7) +#define C_028800_STENCILFAIL_BF 0xFC7FFFFF +#define S_028800_STENCILZPASS_BF(x) (((x) & 0x7) << 26) +#define G_028800_STENCILZPASS_BF(x) (((x) >> 26) & 0x7) +#define C_028800_STENCILZPASS_BF 0xE3FFFFFF +#define S_028800_STENCILZFAIL_BF(x) (((x) & 0x7) << 29) +#define G_028800_STENCILZFAIL_BF(x) (((x) >> 29) & 0x7) +#define C_028800_STENCILZFAIL_BF 0x1FFFFFFF + +#endif diff --git a/drivers/gpu/drm/radeon/radeon.h b/drivers/gpu/drm/radeon/radeon.h new file mode 100644 index 00000000..59d72d0f --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon.h @@ -0,0 +1,1523 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#ifndef __RADEON_H__ +#define __RADEON_H__ + +/* TODO: Here are things that needs to be done : + * - surface allocator & initializer : (bit like scratch reg) should + * initialize HDP_ stuff on RS600, R600, R700 hw, well anythings + * related to surface + * - WB : write back stuff (do it bit like scratch reg things) + * - Vblank : look at Jesse's rework and what we should do + * - r600/r700: gart & cp + * - cs : clean cs ioctl use bitmap & things like that. + * - power management stuff + * - Barrier in gart code + * - Unmappabled vram ? + * - TESTING, TESTING, TESTING + */ + +/* Initialization path: + * We expect that acceleration initialization might fail for various + * reasons even thought we work hard to make it works on most + * configurations. In order to still have a working userspace in such + * situation the init path must succeed up to the memory controller + * initialization point. Failure before this point are considered as + * fatal error. Here is the init callchain : + * radeon_device_init perform common structure, mutex initialization + * asic_init setup the GPU memory layout and perform all + * one time initialization (failure in this + * function are considered fatal) + * asic_startup setup the GPU acceleration, in order to + * follow guideline the first thing this + * function should do is setting the GPU + * memory controller (only MC setup failure + * are considered as fatal) + */ + +#include +#include +#include +#include + +#include +#include +#include +#include +#include + +#include "radeon_family.h" +#include "radeon_mode.h" +#include "radeon_reg.h" + +/* + * Modules parameters. + */ +extern int radeon_no_wb; +extern int radeon_modeset; +extern int radeon_dynclks; +extern int radeon_r4xx_atom; +extern int radeon_agpmode; +extern int radeon_vram_limit; +extern int radeon_gart_size; +extern int radeon_benchmarking; +extern int radeon_testing; +extern int radeon_connector_table; +extern int radeon_tv; +extern int radeon_audio; +extern int radeon_disp_priority; +extern int radeon_hw_i2c; +extern int radeon_pcie_gen2; +extern int radeon_msi; + +/* + * Copy from radeon_drv.h so we don't have to include both and have conflicting + * symbol; + */ +#define RADEON_MAX_USEC_TIMEOUT 100000 /* 100 ms */ +#define RADEON_FENCE_JIFFIES_TIMEOUT (HZ / 2) +/* RADEON_IB_POOL_SIZE must be a power of 2 */ +#define RADEON_IB_POOL_SIZE 16 +#define RADEON_DEBUGFS_MAX_NUM_FILES 32 +#define RADEONFB_CONN_LIMIT 4 +#define RADEON_BIOS_NUM_SCRATCH 8 + +/* + * Errata workarounds. + */ +enum radeon_pll_errata { + CHIP_ERRATA_R300_CG = 0x00000001, + CHIP_ERRATA_PLL_DUMMYREADS = 0x00000002, + CHIP_ERRATA_PLL_DELAY = 0x00000004 +}; + + +struct radeon_device; + + +/* + * BIOS. + */ +#define ATRM_BIOS_PAGE 4096 + +#if defined(CONFIG_VGA_SWITCHEROO) +bool radeon_atrm_supported(struct pci_dev *pdev); +int radeon_atrm_get_bios_chunk(uint8_t *bios, int offset, int len); +#else +static inline bool radeon_atrm_supported(struct pci_dev *pdev) +{ + return false; +} + +static inline int radeon_atrm_get_bios_chunk(uint8_t *bios, int offset, int len){ + return -EINVAL; +} +#endif +bool radeon_get_bios(struct radeon_device *rdev); + + +/* + * Dummy page + */ +struct radeon_dummy_page { + struct page *page; + dma_addr_t addr; +}; +int radeon_dummy_page_init(struct radeon_device *rdev); +void radeon_dummy_page_fini(struct radeon_device *rdev); + + +/* + * Clocks + */ +struct radeon_clock { + struct radeon_pll p1pll; + struct radeon_pll p2pll; + struct radeon_pll dcpll; + struct radeon_pll spll; + struct radeon_pll mpll; + /* 10 Khz units */ + uint32_t default_mclk; + uint32_t default_sclk; + uint32_t default_dispclk; + uint32_t dp_extclk; + uint32_t max_pixel_clock; +}; + +/* + * Power management + */ +int radeon_pm_init(struct radeon_device *rdev); +void radeon_pm_fini(struct radeon_device *rdev); +void radeon_pm_compute_clocks(struct radeon_device *rdev); +void radeon_pm_suspend(struct radeon_device *rdev); +void radeon_pm_resume(struct radeon_device *rdev); +void radeon_combios_get_power_modes(struct radeon_device *rdev); +void radeon_atombios_get_power_modes(struct radeon_device *rdev); +void radeon_atom_set_voltage(struct radeon_device *rdev, u16 voltage_level, u8 voltage_type); +int radeon_atom_get_max_vddc(struct radeon_device *rdev, u16 *voltage); +void rs690_pm_info(struct radeon_device *rdev); +extern int rv6xx_get_temp(struct radeon_device *rdev); +extern int rv770_get_temp(struct radeon_device *rdev); +extern int evergreen_get_temp(struct radeon_device *rdev); +extern int sumo_get_temp(struct radeon_device *rdev); + +/* + * Fences. + */ +struct radeon_fence_driver { + uint32_t scratch_reg; + atomic_t seq; + uint32_t last_seq; + unsigned long last_jiffies; + unsigned long last_timeout; + wait_queue_head_t queue; + rwlock_t lock; + struct list_head created; + struct list_head emited; + struct list_head signaled; + bool initialized; +}; + +struct radeon_fence { + struct radeon_device *rdev; + struct kref kref; + struct list_head list; + /* protected by radeon_fence.lock */ + uint32_t seq; + bool emited; + bool signaled; +}; + +int radeon_fence_driver_init(struct radeon_device *rdev); +void radeon_fence_driver_fini(struct radeon_device *rdev); +int radeon_fence_create(struct radeon_device *rdev, struct radeon_fence **fence); +int radeon_fence_emit(struct radeon_device *rdev, struct radeon_fence *fence); +void radeon_fence_process(struct radeon_device *rdev); +bool radeon_fence_signaled(struct radeon_fence *fence); +int radeon_fence_wait(struct radeon_fence *fence, bool interruptible); +int radeon_fence_wait_next(struct radeon_device *rdev); +int radeon_fence_wait_last(struct radeon_device *rdev); +struct radeon_fence *radeon_fence_ref(struct radeon_fence *fence); +void radeon_fence_unref(struct radeon_fence **fence); + +/* + * Tiling registers + */ +struct radeon_surface_reg { + struct radeon_bo *bo; +}; + +#define RADEON_GEM_MAX_SURFACES 8 + +/* + * TTM. + */ +struct radeon_mman { + struct ttm_bo_global_ref bo_global_ref; + struct drm_global_reference mem_global_ref; + struct ttm_bo_device bdev; + bool mem_global_referenced; + bool initialized; +}; + +struct radeon_bo { + /* Protected by gem.mutex */ + struct list_head list; + /* Protected by tbo.reserved */ + u32 placements[3]; + struct ttm_placement placement; + struct ttm_buffer_object tbo; + struct ttm_bo_kmap_obj kmap; + unsigned pin_count; + void *kptr; + u32 tiling_flags; + u32 pitch; + int surface_reg; + /* Constant after initialization */ + struct radeon_device *rdev; + struct drm_gem_object gem_base; +}; +#define gem_to_radeon_bo(gobj) container_of((gobj), struct radeon_bo, gem_base) + +struct radeon_bo_list { + struct ttm_validate_buffer tv; + struct radeon_bo *bo; + uint64_t gpu_offset; + unsigned rdomain; + unsigned wdomain; + u32 tiling_flags; +}; + +/* + * GEM objects. + */ +struct radeon_gem { + struct mutex mutex; + struct list_head objects; +}; + +int radeon_gem_init(struct radeon_device *rdev); +void radeon_gem_fini(struct radeon_device *rdev); +int radeon_gem_object_create(struct radeon_device *rdev, int size, + int alignment, int initial_domain, + bool discardable, bool kernel, + struct drm_gem_object **obj); +int radeon_gem_object_pin(struct drm_gem_object *obj, uint32_t pin_domain, + uint64_t *gpu_addr); +void radeon_gem_object_unpin(struct drm_gem_object *obj); + +int radeon_mode_dumb_create(struct drm_file *file_priv, + struct drm_device *dev, + struct drm_mode_create_dumb *args); +int radeon_mode_dumb_mmap(struct drm_file *filp, + struct drm_device *dev, + uint32_t handle, uint64_t *offset_p); +int radeon_mode_dumb_destroy(struct drm_file *file_priv, + struct drm_device *dev, + uint32_t handle); + +/* + * GART structures, functions & helpers + */ +struct radeon_mc; + +struct radeon_gart_table_ram { + volatile uint32_t *ptr; +}; + +struct radeon_gart_table_vram { + struct radeon_bo *robj; + volatile uint32_t *ptr; +}; + +union radeon_gart_table { + struct radeon_gart_table_ram ram; + struct radeon_gart_table_vram vram; +}; + +#define RADEON_GPU_PAGE_SIZE 4096 +#define RADEON_GPU_PAGE_MASK (RADEON_GPU_PAGE_SIZE - 1) +#define RADEON_GPU_PAGE_SHIFT 12 + +struct radeon_gart { + dma_addr_t table_addr; + unsigned num_gpu_pages; + unsigned num_cpu_pages; + unsigned table_size; + union radeon_gart_table table; + struct page **pages; + dma_addr_t *pages_addr; + bool *ttm_alloced; + bool ready; +}; + +int radeon_gart_table_ram_alloc(struct radeon_device *rdev); +void radeon_gart_table_ram_free(struct radeon_device *rdev); +int radeon_gart_table_vram_alloc(struct radeon_device *rdev); +void radeon_gart_table_vram_free(struct radeon_device *rdev); +int radeon_gart_init(struct radeon_device *rdev); +void radeon_gart_fini(struct radeon_device *rdev); +void radeon_gart_unbind(struct radeon_device *rdev, unsigned offset, + int pages); +int radeon_gart_bind(struct radeon_device *rdev, unsigned offset, + int pages, struct page **pagelist, + dma_addr_t *dma_addr); + + +/* + * GPU MC structures, functions & helpers + */ +struct radeon_mc { + resource_size_t aper_size; + resource_size_t aper_base; + resource_size_t agp_base; + /* for some chips with <= 32MB we need to lie + * about vram size near mc fb location */ + u64 mc_vram_size; + u64 visible_vram_size; + u64 gtt_size; + u64 gtt_start; + u64 gtt_end; + u64 vram_start; + u64 vram_end; + unsigned vram_width; + u64 real_vram_size; + int vram_mtrr; + bool vram_is_ddr; + bool igp_sideport_enabled; + u64 gtt_base_align; +}; + +bool radeon_combios_sideport_present(struct radeon_device *rdev); +bool radeon_atombios_sideport_present(struct radeon_device *rdev); + +/* + * GPU scratch registers structures, functions & helpers + */ +struct radeon_scratch { + unsigned num_reg; + uint32_t reg_base; + bool free[32]; + uint32_t reg[32]; +}; + +int radeon_scratch_get(struct radeon_device *rdev, uint32_t *reg); +void radeon_scratch_free(struct radeon_device *rdev, uint32_t reg); + + +/* + * IRQS. + */ + +struct radeon_unpin_work { + struct work_struct work; + struct radeon_device *rdev; + int crtc_id; + struct radeon_fence *fence; + struct drm_pending_vblank_event *event; + struct radeon_bo *old_rbo; + u64 new_crtc_base; +}; + +struct r500_irq_stat_regs { + u32 disp_int; +}; + +struct r600_irq_stat_regs { + u32 disp_int; + u32 disp_int_cont; + u32 disp_int_cont2; + u32 d1grph_int; + u32 d2grph_int; +}; + +struct evergreen_irq_stat_regs { + u32 disp_int; + u32 disp_int_cont; + u32 disp_int_cont2; + u32 disp_int_cont3; + u32 disp_int_cont4; + u32 disp_int_cont5; + u32 d1grph_int; + u32 d2grph_int; + u32 d3grph_int; + u32 d4grph_int; + u32 d5grph_int; + u32 d6grph_int; +}; + +union radeon_irq_stat_regs { + struct r500_irq_stat_regs r500; + struct r600_irq_stat_regs r600; + struct evergreen_irq_stat_regs evergreen; +}; + +struct radeon_irq { + bool installed; + bool sw_int; + /* FIXME: use a define max crtc rather than hardcode it */ + bool crtc_vblank_int[6]; + bool pflip[6]; + wait_queue_head_t vblank_queue; + /* FIXME: use defines for max hpd/dacs */ + bool hpd[6]; + bool gui_idle; + bool gui_idle_acked; + wait_queue_head_t idle_queue; + /* FIXME: use defines for max HDMI blocks */ + bool hdmi[2]; + spinlock_t sw_lock; + int sw_refcount; + union radeon_irq_stat_regs stat_regs; + spinlock_t pflip_lock[6]; + int pflip_refcount[6]; +}; + +int radeon_irq_kms_init(struct radeon_device *rdev); +void radeon_irq_kms_fini(struct radeon_device *rdev); +void radeon_irq_kms_sw_irq_get(struct radeon_device *rdev); +void radeon_irq_kms_sw_irq_put(struct radeon_device *rdev); +void radeon_irq_kms_pflip_irq_get(struct radeon_device *rdev, int crtc); +void radeon_irq_kms_pflip_irq_put(struct radeon_device *rdev, int crtc); + +/* + * CP & ring. + */ +struct radeon_ib { + struct list_head list; + unsigned idx; + uint64_t gpu_addr; + struct radeon_fence *fence; + uint32_t *ptr; + uint32_t length_dw; + bool free; +}; + +/* + * locking - + * mutex protects scheduled_ibs, ready, alloc_bm + */ +struct radeon_ib_pool { + struct mutex mutex; + struct radeon_bo *robj; + struct list_head bogus_ib; + struct radeon_ib ibs[RADEON_IB_POOL_SIZE]; + bool ready; + unsigned head_id; +}; + +struct radeon_cp { + struct radeon_bo *ring_obj; + volatile uint32_t *ring; + unsigned rptr; + unsigned wptr; + unsigned wptr_old; + unsigned ring_size; + unsigned ring_free_dw; + int count_dw; + uint64_t gpu_addr; + uint32_t align_mask; + uint32_t ptr_mask; + struct mutex mutex; + bool ready; +}; + +/* + * R6xx+ IH ring + */ +struct r600_ih { + struct radeon_bo *ring_obj; + volatile uint32_t *ring; + unsigned rptr; + unsigned wptr; + unsigned wptr_old; + unsigned ring_size; + uint64_t gpu_addr; + uint32_t ptr_mask; + spinlock_t lock; + bool enabled; +}; + +struct r600_blit { + struct mutex mutex; + struct radeon_bo *shader_obj; + u64 shader_gpu_addr; + u32 vs_offset, ps_offset; + u32 state_offset; + u32 state_len; + u32 vb_used, vb_total; + struct radeon_ib *vb_ib; +}; + +int radeon_ib_get(struct radeon_device *rdev, struct radeon_ib **ib); +void radeon_ib_free(struct radeon_device *rdev, struct radeon_ib **ib); +int radeon_ib_schedule(struct radeon_device *rdev, struct radeon_ib *ib); +int radeon_ib_pool_init(struct radeon_device *rdev); +void radeon_ib_pool_fini(struct radeon_device *rdev); +int radeon_ib_test(struct radeon_device *rdev); +extern void radeon_ib_bogus_add(struct radeon_device *rdev, struct radeon_ib *ib); +/* Ring access between begin & end cannot sleep */ +void radeon_ring_free_size(struct radeon_device *rdev); +int radeon_ring_alloc(struct radeon_device *rdev, unsigned ndw); +int radeon_ring_lock(struct radeon_device *rdev, unsigned ndw); +void radeon_ring_commit(struct radeon_device *rdev); +void radeon_ring_unlock_commit(struct radeon_device *rdev); +void radeon_ring_unlock_undo(struct radeon_device *rdev); +int radeon_ring_test(struct radeon_device *rdev); +int radeon_ring_init(struct radeon_device *rdev, unsigned ring_size); +void radeon_ring_fini(struct radeon_device *rdev); + + +/* + * CS. + */ +struct radeon_cs_reloc { + struct drm_gem_object *gobj; + struct radeon_bo *robj; + struct radeon_bo_list lobj; + uint32_t handle; + uint32_t flags; +}; + +struct radeon_cs_chunk { + uint32_t chunk_id; + uint32_t length_dw; + int kpage_idx[2]; + uint32_t *kpage[2]; + uint32_t *kdata; + void __user *user_ptr; + int last_copied_page; + int last_page_index; +}; + +struct radeon_cs_parser { + struct device *dev; + struct radeon_device *rdev; + struct drm_file *filp; + /* chunks */ + unsigned nchunks; + struct radeon_cs_chunk *chunks; + uint64_t *chunks_array; + /* IB */ + unsigned idx; + /* relocations */ + unsigned nrelocs; + struct radeon_cs_reloc *relocs; + struct radeon_cs_reloc **relocs_ptr; + struct list_head validated; + /* indices of various chunks */ + int chunk_ib_idx; + int chunk_relocs_idx; + struct radeon_ib *ib; + void *track; + unsigned family; + int parser_error; +}; + +extern int radeon_cs_update_pages(struct radeon_cs_parser *p, int pg_idx); +extern int radeon_cs_finish_pages(struct radeon_cs_parser *p); + + +static inline u32 radeon_get_ib_value(struct radeon_cs_parser *p, int idx) +{ + struct radeon_cs_chunk *ibc = &p->chunks[p->chunk_ib_idx]; + u32 pg_idx, pg_offset; + u32 idx_value = 0; + int new_page; + + pg_idx = (idx * 4) / PAGE_SIZE; + pg_offset = (idx * 4) % PAGE_SIZE; + + if (ibc->kpage_idx[0] == pg_idx) + return ibc->kpage[0][pg_offset/4]; + if (ibc->kpage_idx[1] == pg_idx) + return ibc->kpage[1][pg_offset/4]; + + new_page = radeon_cs_update_pages(p, pg_idx); + if (new_page < 0) { + p->parser_error = new_page; + return 0; + } + + idx_value = ibc->kpage[new_page][pg_offset/4]; + return idx_value; +} + +struct radeon_cs_packet { + unsigned idx; + unsigned type; + unsigned reg; + unsigned opcode; + int count; + unsigned one_reg_wr; +}; + +typedef int (*radeon_packet0_check_t)(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt, + unsigned idx, unsigned reg); +typedef int (*radeon_packet3_check_t)(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt); + + +/* + * AGP + */ +int radeon_agp_init(struct radeon_device *rdev); +void radeon_agp_resume(struct radeon_device *rdev); +void radeon_agp_suspend(struct radeon_device *rdev); +void radeon_agp_fini(struct radeon_device *rdev); + + +/* + * Writeback + */ +struct radeon_wb { + struct radeon_bo *wb_obj; + volatile uint32_t *wb; + uint64_t gpu_addr; + bool enabled; + bool use_event; +}; + +#define RADEON_WB_SCRATCH_OFFSET 0 +#define RADEON_WB_CP_RPTR_OFFSET 1024 +#define RADEON_WB_CP1_RPTR_OFFSET 1280 +#define RADEON_WB_CP2_RPTR_OFFSET 1536 +#define R600_WB_IH_WPTR_OFFSET 2048 +#define R600_WB_EVENT_OFFSET 3072 + +/** + * struct radeon_pm - power management datas + * @max_bandwidth: maximum bandwidth the gpu has (MByte/s) + * @igp_sideport_mclk: sideport memory clock Mhz (rs690,rs740,rs780,rs880) + * @igp_system_mclk: system clock Mhz (rs690,rs740,rs780,rs880) + * @igp_ht_link_clk: ht link clock Mhz (rs690,rs740,rs780,rs880) + * @igp_ht_link_width: ht link width in bits (rs690,rs740,rs780,rs880) + * @k8_bandwidth: k8 bandwidth the gpu has (MByte/s) (IGP) + * @sideport_bandwidth: sideport bandwidth the gpu has (MByte/s) (IGP) + * @ht_bandwidth: ht bandwidth the gpu has (MByte/s) (IGP) + * @core_bandwidth: core GPU bandwidth the gpu has (MByte/s) (IGP) + * @sclk: GPU clock Mhz (core bandwidth depends of this clock) + * @needed_bandwidth: current bandwidth needs + * + * It keeps track of various data needed to take powermanagement decision. + * Bandwidth need is used to determine minimun clock of the GPU and memory. + * Equation between gpu/memory clock and available bandwidth is hw dependent + * (type of memory, bus size, efficiency, ...) + */ + +enum radeon_pm_method { + PM_METHOD_PROFILE, + PM_METHOD_DYNPM, +}; + +enum radeon_dynpm_state { + DYNPM_STATE_DISABLED, + DYNPM_STATE_MINIMUM, + DYNPM_STATE_PAUSED, + DYNPM_STATE_ACTIVE, + DYNPM_STATE_SUSPENDED, +}; +enum radeon_dynpm_action { + DYNPM_ACTION_NONE, + DYNPM_ACTION_MINIMUM, + DYNPM_ACTION_DOWNCLOCK, + DYNPM_ACTION_UPCLOCK, + DYNPM_ACTION_DEFAULT +}; + +enum radeon_voltage_type { + VOLTAGE_NONE = 0, + VOLTAGE_GPIO, + VOLTAGE_VDDC, + VOLTAGE_SW +}; + +enum radeon_pm_state_type { + POWER_STATE_TYPE_DEFAULT, + POWER_STATE_TYPE_POWERSAVE, + POWER_STATE_TYPE_BATTERY, + POWER_STATE_TYPE_BALANCED, + POWER_STATE_TYPE_PERFORMANCE, +}; + +enum radeon_pm_profile_type { + PM_PROFILE_DEFAULT, + PM_PROFILE_AUTO, + PM_PROFILE_LOW, + PM_PROFILE_MID, + PM_PROFILE_HIGH, +}; + +#define PM_PROFILE_DEFAULT_IDX 0 +#define PM_PROFILE_LOW_SH_IDX 1 +#define PM_PROFILE_MID_SH_IDX 2 +#define PM_PROFILE_HIGH_SH_IDX 3 +#define PM_PROFILE_LOW_MH_IDX 4 +#define PM_PROFILE_MID_MH_IDX 5 +#define PM_PROFILE_HIGH_MH_IDX 6 +#define PM_PROFILE_MAX 7 + +struct radeon_pm_profile { + int dpms_off_ps_idx; + int dpms_on_ps_idx; + int dpms_off_cm_idx; + int dpms_on_cm_idx; +}; + +enum radeon_int_thermal_type { + THERMAL_TYPE_NONE, + THERMAL_TYPE_RV6XX, + THERMAL_TYPE_RV770, + THERMAL_TYPE_EVERGREEN, + THERMAL_TYPE_SUMO, + THERMAL_TYPE_NI, +}; + +struct radeon_voltage { + enum radeon_voltage_type type; + /* gpio voltage */ + struct radeon_gpio_rec gpio; + u32 delay; /* delay in usec from voltage drop to sclk change */ + bool active_high; /* voltage drop is active when bit is high */ + /* VDDC voltage */ + u8 vddc_id; /* index into vddc voltage table */ + u8 vddci_id; /* index into vddci voltage table */ + bool vddci_enabled; + /* r6xx+ sw */ + u16 voltage; + /* evergreen+ vddci */ + u16 vddci; +}; + +/* clock mode flags */ +#define RADEON_PM_MODE_NO_DISPLAY (1 << 0) + +struct radeon_pm_clock_info { + /* memory clock */ + u32 mclk; + /* engine clock */ + u32 sclk; + /* voltage info */ + struct radeon_voltage voltage; + /* standardized clock flags */ + u32 flags; +}; + +/* state flags */ +#define RADEON_PM_STATE_SINGLE_DISPLAY_ONLY (1 << 0) + +struct radeon_power_state { + enum radeon_pm_state_type type; + /* XXX: use a define for num clock modes */ + struct radeon_pm_clock_info clock_info[8]; + /* number of valid clock modes in this power state */ + int num_clock_modes; + struct radeon_pm_clock_info *default_clock_mode; + /* standardized state flags */ + u32 flags; + u32 misc; /* vbios specific flags */ + u32 misc2; /* vbios specific flags */ + int pcie_lanes; /* pcie lanes */ +}; + +/* + * Some modes are overclocked by very low value, accept them + */ +#define RADEON_MODE_OVERCLOCK_MARGIN 500 /* 5 MHz */ + +struct radeon_pm { + struct mutex mutex; + u32 active_crtcs; + int active_crtc_count; + int req_vblank; + bool vblank_sync; + bool gui_idle; + fixed20_12 max_bandwidth; + fixed20_12 igp_sideport_mclk; + fixed20_12 igp_system_mclk; + fixed20_12 igp_ht_link_clk; + fixed20_12 igp_ht_link_width; + fixed20_12 k8_bandwidth; + fixed20_12 sideport_bandwidth; + fixed20_12 ht_bandwidth; + fixed20_12 core_bandwidth; + fixed20_12 sclk; + fixed20_12 mclk; + fixed20_12 needed_bandwidth; + struct radeon_power_state *power_state; + /* number of valid power states */ + int num_power_states; + int current_power_state_index; + int current_clock_mode_index; + int requested_power_state_index; + int requested_clock_mode_index; + int default_power_state_index; + u32 current_sclk; + u32 current_mclk; + u16 current_vddc; + u16 current_vddci; + u32 default_sclk; + u32 default_mclk; + u16 default_vddc; + u16 default_vddci; + struct radeon_i2c_chan *i2c_bus; + /* selected pm method */ + enum radeon_pm_method pm_method; + /* dynpm power management */ + struct delayed_work dynpm_idle_work; + enum radeon_dynpm_state dynpm_state; + enum radeon_dynpm_action dynpm_planned_action; + unsigned long dynpm_action_timeout; + bool dynpm_can_upclock; + bool dynpm_can_downclock; + /* profile-based power management */ + enum radeon_pm_profile_type profile; + int profile_index; + struct radeon_pm_profile profiles[PM_PROFILE_MAX]; + /* internal thermal controller on rv6xx+ */ + enum radeon_int_thermal_type int_thermal_type; + struct device *int_hwmon_dev; +}; + + +/* + * Benchmarking + */ +void radeon_benchmark(struct radeon_device *rdev); + + +/* + * Testing + */ +void radeon_test_moves(struct radeon_device *rdev); + + +/* + * Debugfs + */ +int radeon_debugfs_add_files(struct radeon_device *rdev, + struct drm_info_list *files, + unsigned nfiles); +int radeon_debugfs_fence_init(struct radeon_device *rdev); + + +/* + * ASIC specific functions. + */ +struct radeon_asic { + int (*init)(struct radeon_device *rdev); + void (*fini)(struct radeon_device *rdev); + int (*resume)(struct radeon_device *rdev); + int (*suspend)(struct radeon_device *rdev); + void (*vga_set_state)(struct radeon_device *rdev, bool state); + bool (*gpu_is_lockup)(struct radeon_device *rdev); + int (*asic_reset)(struct radeon_device *rdev); + void (*gart_tlb_flush)(struct radeon_device *rdev); + int (*gart_set_page)(struct radeon_device *rdev, int i, uint64_t addr); + int (*cp_init)(struct radeon_device *rdev, unsigned ring_size); + void (*cp_fini)(struct radeon_device *rdev); + void (*cp_disable)(struct radeon_device *rdev); + void (*cp_commit)(struct radeon_device *rdev); + void (*ring_start)(struct radeon_device *rdev); + int (*ring_test)(struct radeon_device *rdev); + void (*ring_ib_execute)(struct radeon_device *rdev, struct radeon_ib *ib); + int (*irq_set)(struct radeon_device *rdev); + int (*irq_process)(struct radeon_device *rdev); + u32 (*get_vblank_counter)(struct radeon_device *rdev, int crtc); + void (*fence_ring_emit)(struct radeon_device *rdev, struct radeon_fence *fence); + int (*cs_parse)(struct radeon_cs_parser *p); + int (*copy_blit)(struct radeon_device *rdev, + uint64_t src_offset, + uint64_t dst_offset, + unsigned num_gpu_pages, + struct radeon_fence *fence); + int (*copy_dma)(struct radeon_device *rdev, + uint64_t src_offset, + uint64_t dst_offset, + unsigned num_gpu_pages, + struct radeon_fence *fence); + int (*copy)(struct radeon_device *rdev, + uint64_t src_offset, + uint64_t dst_offset, + unsigned num_gpu_pages, + struct radeon_fence *fence); + uint32_t (*get_engine_clock)(struct radeon_device *rdev); + void (*set_engine_clock)(struct radeon_device *rdev, uint32_t eng_clock); + uint32_t (*get_memory_clock)(struct radeon_device *rdev); + void (*set_memory_clock)(struct radeon_device *rdev, uint32_t mem_clock); + int (*get_pcie_lanes)(struct radeon_device *rdev); + void (*set_pcie_lanes)(struct radeon_device *rdev, int lanes); + void (*set_clock_gating)(struct radeon_device *rdev, int enable); + int (*set_surface_reg)(struct radeon_device *rdev, int reg, + uint32_t tiling_flags, uint32_t pitch, + uint32_t offset, uint32_t obj_size); + void (*clear_surface_reg)(struct radeon_device *rdev, int reg); + void (*bandwidth_update)(struct radeon_device *rdev); + void (*hpd_init)(struct radeon_device *rdev); + void (*hpd_fini)(struct radeon_device *rdev); + bool (*hpd_sense)(struct radeon_device *rdev, enum radeon_hpd_id hpd); + void (*hpd_set_polarity)(struct radeon_device *rdev, enum radeon_hpd_id hpd); + /* ioctl hw specific callback. Some hw might want to perform special + * operation on specific ioctl. For instance on wait idle some hw + * might want to perform and HDP flush through MMIO as it seems that + * some R6XX/R7XX hw doesn't take HDP flush into account if programmed + * through ring. + */ + void (*ioctl_wait_idle)(struct radeon_device *rdev, struct radeon_bo *bo); + bool (*gui_idle)(struct radeon_device *rdev); + /* power management */ + void (*pm_misc)(struct radeon_device *rdev); + void (*pm_prepare)(struct radeon_device *rdev); + void (*pm_finish)(struct radeon_device *rdev); + void (*pm_init_profile)(struct radeon_device *rdev); + void (*pm_get_dynpm_state)(struct radeon_device *rdev); + /* pageflipping */ + void (*pre_page_flip)(struct radeon_device *rdev, int crtc); + u32 (*page_flip)(struct radeon_device *rdev, int crtc, u64 crtc_base); + void (*post_page_flip)(struct radeon_device *rdev, int crtc); +}; + +/* + * Asic structures + */ +struct r100_gpu_lockup { + unsigned long last_jiffies; + u32 last_cp_rptr; +}; + +struct r100_asic { + const unsigned *reg_safe_bm; + unsigned reg_safe_bm_size; + u32 hdp_cntl; + struct r100_gpu_lockup lockup; +}; + +struct r300_asic { + const unsigned *reg_safe_bm; + unsigned reg_safe_bm_size; + u32 resync_scratch; + u32 hdp_cntl; + struct r100_gpu_lockup lockup; +}; + +struct r600_asic { + unsigned max_pipes; + unsigned max_tile_pipes; + unsigned max_simds; + unsigned max_backends; + unsigned max_gprs; + unsigned max_threads; + unsigned max_stack_entries; + unsigned max_hw_contexts; + unsigned max_gs_threads; + unsigned sx_max_export_size; + unsigned sx_max_export_pos_size; + unsigned sx_max_export_smx_size; + unsigned sq_num_cf_insts; + unsigned tiling_nbanks; + unsigned tiling_npipes; + unsigned tiling_group_size; + unsigned tile_config; + struct r100_gpu_lockup lockup; +}; + +struct rv770_asic { + unsigned max_pipes; + unsigned max_tile_pipes; + unsigned max_simds; + unsigned max_backends; + unsigned max_gprs; + unsigned max_threads; + unsigned max_stack_entries; + unsigned max_hw_contexts; + unsigned max_gs_threads; + unsigned sx_max_export_size; + unsigned sx_max_export_pos_size; + unsigned sx_max_export_smx_size; + unsigned sq_num_cf_insts; + unsigned sx_num_of_sets; + unsigned sc_prim_fifo_size; + unsigned sc_hiz_tile_fifo_size; + unsigned sc_earlyz_tile_fifo_fize; + unsigned tiling_nbanks; + unsigned tiling_npipes; + unsigned tiling_group_size; + unsigned tile_config; + struct r100_gpu_lockup lockup; +}; + +struct evergreen_asic { + unsigned num_ses; + unsigned max_pipes; + unsigned max_tile_pipes; + unsigned max_simds; + unsigned max_backends; + unsigned max_gprs; + unsigned max_threads; + unsigned max_stack_entries; + unsigned max_hw_contexts; + unsigned max_gs_threads; + unsigned sx_max_export_size; + unsigned sx_max_export_pos_size; + unsigned sx_max_export_smx_size; + unsigned sq_num_cf_insts; + unsigned sx_num_of_sets; + unsigned sc_prim_fifo_size; + unsigned sc_hiz_tile_fifo_size; + unsigned sc_earlyz_tile_fifo_size; + unsigned tiling_nbanks; + unsigned tiling_npipes; + unsigned tiling_group_size; + unsigned tile_config; + struct r100_gpu_lockup lockup; +}; + +struct cayman_asic { + unsigned max_shader_engines; + unsigned max_pipes_per_simd; + unsigned max_tile_pipes; + unsigned max_simds_per_se; + unsigned max_backends_per_se; + unsigned max_texture_channel_caches; + unsigned max_gprs; + unsigned max_threads; + unsigned max_gs_threads; + unsigned max_stack_entries; + unsigned sx_num_of_sets; + unsigned sx_max_export_size; + unsigned sx_max_export_pos_size; + unsigned sx_max_export_smx_size; + unsigned max_hw_contexts; + unsigned sq_num_cf_insts; + unsigned sc_prim_fifo_size; + unsigned sc_hiz_tile_fifo_size; + unsigned sc_earlyz_tile_fifo_size; + + unsigned num_shader_engines; + unsigned num_shader_pipes_per_simd; + unsigned num_tile_pipes; + unsigned num_simds_per_se; + unsigned num_backends_per_se; + unsigned backend_disable_mask_per_asic; + unsigned backend_map; + unsigned num_texture_channel_caches; + unsigned mem_max_burst_length_bytes; + unsigned mem_row_size_in_kb; + unsigned shader_engine_tile_size; + unsigned num_gpus; + unsigned multi_gpu_tile_size; + + unsigned tile_config; + struct r100_gpu_lockup lockup; +}; + +union radeon_asic_config { + struct r300_asic r300; + struct r100_asic r100; + struct r600_asic r600; + struct rv770_asic rv770; + struct evergreen_asic evergreen; + struct cayman_asic cayman; +}; + +/* + * asic initizalization from radeon_asic.c + */ +void radeon_agp_disable(struct radeon_device *rdev); +int radeon_asic_init(struct radeon_device *rdev); + + +/* + * IOCTL. + */ +int radeon_gem_info_ioctl(struct drm_device *dev, void *data, + struct drm_file *filp); +int radeon_gem_create_ioctl(struct drm_device *dev, void *data, + struct drm_file *filp); +int radeon_gem_pin_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv); +int radeon_gem_unpin_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv); +int radeon_gem_pwrite_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv); +int radeon_gem_pread_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv); +int radeon_gem_set_domain_ioctl(struct drm_device *dev, void *data, + struct drm_file *filp); +int radeon_gem_mmap_ioctl(struct drm_device *dev, void *data, + struct drm_file *filp); +int radeon_gem_busy_ioctl(struct drm_device *dev, void *data, + struct drm_file *filp); +int radeon_gem_wait_idle_ioctl(struct drm_device *dev, void *data, + struct drm_file *filp); +int radeon_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp); +int radeon_gem_set_tiling_ioctl(struct drm_device *dev, void *data, + struct drm_file *filp); +int radeon_gem_get_tiling_ioctl(struct drm_device *dev, void *data, + struct drm_file *filp); + +/* VRAM scratch page for HDP bug */ +struct r700_vram_scratch { + struct radeon_bo *robj; + volatile uint32_t *ptr; +}; + +/* + * Core structure, functions and helpers. + */ +typedef uint32_t (*radeon_rreg_t)(struct radeon_device*, uint32_t); +typedef void (*radeon_wreg_t)(struct radeon_device*, uint32_t, uint32_t); + +struct radeon_device { + struct device *dev; + struct drm_device *ddev; + struct pci_dev *pdev; + /* ASIC */ + union radeon_asic_config config; + enum radeon_family family; + unsigned long flags; + int usec_timeout; + enum radeon_pll_errata pll_errata; + int num_gb_pipes; + int num_z_pipes; + int disp_priority; + /* BIOS */ + uint8_t *bios; + bool is_atom_bios; + uint16_t bios_header_start; + struct radeon_bo *stollen_vga_memory; + /* Register mmio */ + resource_size_t rmmio_base; + resource_size_t rmmio_size; + void *rmmio; + radeon_rreg_t mc_rreg; + radeon_wreg_t mc_wreg; + radeon_rreg_t pll_rreg; + radeon_wreg_t pll_wreg; + uint32_t pcie_reg_mask; + radeon_rreg_t pciep_rreg; + radeon_wreg_t pciep_wreg; + /* io port */ + void __iomem *rio_mem; + resource_size_t rio_mem_size; + struct radeon_clock clock; + struct radeon_mc mc; + struct radeon_gart gart; + struct radeon_mode_info mode_info; + struct radeon_scratch scratch; + struct radeon_mman mman; + struct radeon_fence_driver fence_drv; + struct radeon_cp cp; + /* cayman compute rings */ + struct radeon_cp cp1; + struct radeon_cp cp2; + struct radeon_ib_pool ib_pool; + struct radeon_irq irq; + struct radeon_asic *asic; + struct radeon_gem gem; + struct radeon_pm pm; + uint32_t bios_scratch[RADEON_BIOS_NUM_SCRATCH]; + struct mutex cs_mutex; + struct radeon_wb wb; + struct radeon_dummy_page dummy_page; + bool gpu_lockup; + bool shutdown; + bool suspend; + bool need_dma32; + bool accel_working; + struct radeon_surface_reg surface_regs[RADEON_GEM_MAX_SURFACES]; + const struct firmware *me_fw; /* all family ME firmware */ + const struct firmware *pfp_fw; /* r6/700 PFP firmware */ + const struct firmware *rlc_fw; /* r6/700 RLC firmware */ + const struct firmware *mc_fw; /* NI MC firmware */ + struct r600_blit r600_blit; + struct r700_vram_scratch vram_scratch; + int msi_enabled; /* msi enabled */ + struct r600_ih ih; /* r6/700 interrupt ring */ + struct work_struct hotplug_work; + int num_crtc; /* number of crtcs */ + struct mutex dc_hw_i2c_mutex; /* display controller hw i2c mutex */ + struct mutex vram_mutex; + + /* audio stuff */ + bool audio_enabled; + struct timer_list audio_timer; + int audio_channels; + int audio_rate; + int audio_bits_per_sample; + uint8_t audio_status_bits; + uint8_t audio_category_code; + + struct notifier_block acpi_nb; + /* only one userspace can use Hyperz features or CMASK at a time */ + struct drm_file *hyperz_filp; + struct drm_file *cmask_filp; + /* i2c buses */ + struct radeon_i2c_chan *i2c_bus[RADEON_MAX_I2C_BUS]; +}; + +int radeon_device_init(struct radeon_device *rdev, + struct drm_device *ddev, + struct pci_dev *pdev, + uint32_t flags); +void radeon_device_fini(struct radeon_device *rdev); +int radeon_gpu_wait_for_idle(struct radeon_device *rdev); + +static inline uint32_t r100_mm_rreg(struct radeon_device *rdev, uint32_t reg) +{ + if (reg < rdev->rmmio_size) + return readl(((void __iomem *)rdev->rmmio) + reg); + else { + writel(reg, ((void __iomem *)rdev->rmmio) + RADEON_MM_INDEX); + return readl(((void __iomem *)rdev->rmmio) + RADEON_MM_DATA); + } +} + +static inline void r100_mm_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v) +{ + if (reg < rdev->rmmio_size) + writel(v, ((void __iomem *)rdev->rmmio) + reg); + else { + writel(reg, ((void __iomem *)rdev->rmmio) + RADEON_MM_INDEX); + writel(v, ((void __iomem *)rdev->rmmio) + RADEON_MM_DATA); + } +} + +static inline u32 r100_io_rreg(struct radeon_device *rdev, u32 reg) +{ + if (reg < rdev->rio_mem_size) + return ioread32(rdev->rio_mem + reg); + else { + iowrite32(reg, rdev->rio_mem + RADEON_MM_INDEX); + return ioread32(rdev->rio_mem + RADEON_MM_DATA); + } +} + +static inline void r100_io_wreg(struct radeon_device *rdev, u32 reg, u32 v) +{ + if (reg < rdev->rio_mem_size) + iowrite32(v, rdev->rio_mem + reg); + else { + iowrite32(reg, rdev->rio_mem + RADEON_MM_INDEX); + iowrite32(v, rdev->rio_mem + RADEON_MM_DATA); + } +} + +/* + * Cast helper + */ +#define to_radeon_fence(p) ((struct radeon_fence *)(p)) + +/* + * Registers read & write functions. + */ +#define RREG8(reg) readb(((void __iomem *)rdev->rmmio) + (reg)) +#define WREG8(reg, v) writeb(v, ((void __iomem *)rdev->rmmio) + (reg)) +#define RREG16(reg) readw(((void __iomem *)rdev->rmmio) + (reg)) +#define WREG16(reg, v) writew(v, ((void __iomem *)rdev->rmmio) + (reg)) +#define RREG32(reg) r100_mm_rreg(rdev, (reg)) +#define DREG32(reg) printk(KERN_INFO "REGISTER: " #reg " : 0x%08X\n", r100_mm_rreg(rdev, (reg))) +#define WREG32(reg, v) r100_mm_wreg(rdev, (reg), (v)) +#define REG_SET(FIELD, v) (((v) << FIELD##_SHIFT) & FIELD##_MASK) +#define REG_GET(FIELD, v) (((v) << FIELD##_SHIFT) & FIELD##_MASK) +#define RREG32_PLL(reg) rdev->pll_rreg(rdev, (reg)) +#define WREG32_PLL(reg, v) rdev->pll_wreg(rdev, (reg), (v)) +#define RREG32_MC(reg) rdev->mc_rreg(rdev, (reg)) +#define WREG32_MC(reg, v) rdev->mc_wreg(rdev, (reg), (v)) +#define RREG32_PCIE(reg) rv370_pcie_rreg(rdev, (reg)) +#define WREG32_PCIE(reg, v) rv370_pcie_wreg(rdev, (reg), (v)) +#define RREG32_PCIE_P(reg) rdev->pciep_rreg(rdev, (reg)) +#define WREG32_PCIE_P(reg, v) rdev->pciep_wreg(rdev, (reg), (v)) +#define WREG32_P(reg, val, mask) \ + do { \ + uint32_t tmp_ = RREG32(reg); \ + tmp_ &= (mask); \ + tmp_ |= ((val) & ~(mask)); \ + WREG32(reg, tmp_); \ + } while (0) +#define WREG32_PLL_P(reg, val, mask) \ + do { \ + uint32_t tmp_ = RREG32_PLL(reg); \ + tmp_ &= (mask); \ + tmp_ |= ((val) & ~(mask)); \ + WREG32_PLL(reg, tmp_); \ + } while (0) +#define DREG32_SYS(sqf, rdev, reg) seq_printf((sqf), #reg " : 0x%08X\n", r100_mm_rreg((rdev), (reg))) +#define RREG32_IO(reg) r100_io_rreg(rdev, (reg)) +#define WREG32_IO(reg, v) r100_io_wreg(rdev, (reg), (v)) + +/* + * Indirect registers accessor + */ +static inline uint32_t rv370_pcie_rreg(struct radeon_device *rdev, uint32_t reg) +{ + uint32_t r; + + WREG32(RADEON_PCIE_INDEX, ((reg) & rdev->pcie_reg_mask)); + r = RREG32(RADEON_PCIE_DATA); + return r; +} + +static inline void rv370_pcie_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v) +{ + WREG32(RADEON_PCIE_INDEX, ((reg) & rdev->pcie_reg_mask)); + WREG32(RADEON_PCIE_DATA, (v)); +} + +void r100_pll_errata_after_index(struct radeon_device *rdev); + + +/* + * ASICs helpers. + */ +#define ASIC_IS_RN50(rdev) ((rdev->pdev->device == 0x515e) || \ + (rdev->pdev->device == 0x5969)) +#define ASIC_IS_RV100(rdev) ((rdev->family == CHIP_RV100) || \ + (rdev->family == CHIP_RV200) || \ + (rdev->family == CHIP_RS100) || \ + (rdev->family == CHIP_RS200) || \ + (rdev->family == CHIP_RV250) || \ + (rdev->family == CHIP_RV280) || \ + (rdev->family == CHIP_RS300)) +#define ASIC_IS_R300(rdev) ((rdev->family == CHIP_R300) || \ + (rdev->family == CHIP_RV350) || \ + (rdev->family == CHIP_R350) || \ + (rdev->family == CHIP_RV380) || \ + (rdev->family == CHIP_R420) || \ + (rdev->family == CHIP_R423) || \ + (rdev->family == CHIP_RV410) || \ + (rdev->family == CHIP_RS400) || \ + (rdev->family == CHIP_RS480)) +#define ASIC_IS_X2(rdev) ((rdev->ddev->pdev->device == 0x9441) || \ + (rdev->ddev->pdev->device == 0x9443) || \ + (rdev->ddev->pdev->device == 0x944B) || \ + (rdev->ddev->pdev->device == 0x9506) || \ + (rdev->ddev->pdev->device == 0x9509) || \ + (rdev->ddev->pdev->device == 0x950F) || \ + (rdev->ddev->pdev->device == 0x689C) || \ + (rdev->ddev->pdev->device == 0x689D)) +#define ASIC_IS_AVIVO(rdev) ((rdev->family >= CHIP_RS600)) +#define ASIC_IS_DCE2(rdev) ((rdev->family == CHIP_RS600) || \ + (rdev->family == CHIP_RS690) || \ + (rdev->family == CHIP_RS740) || \ + (rdev->family >= CHIP_R600)) +#define ASIC_IS_DCE3(rdev) ((rdev->family >= CHIP_RV620)) +#define ASIC_IS_DCE32(rdev) ((rdev->family >= CHIP_RV730)) +#define ASIC_IS_DCE4(rdev) ((rdev->family >= CHIP_CEDAR)) +#define ASIC_IS_DCE41(rdev) ((rdev->family >= CHIP_PALM) && \ + (rdev->flags & RADEON_IS_IGP)) +#define ASIC_IS_DCE5(rdev) ((rdev->family >= CHIP_BARTS)) + +/* + * BIOS helpers. + */ +#define RBIOS8(i) (rdev->bios[i]) +#define RBIOS16(i) (RBIOS8(i) | (RBIOS8((i)+1) << 8)) +#define RBIOS32(i) ((RBIOS16(i)) | (RBIOS16((i)+2) << 16)) + +int radeon_combios_init(struct radeon_device *rdev); +void radeon_combios_fini(struct radeon_device *rdev); +int radeon_atombios_init(struct radeon_device *rdev); +void radeon_atombios_fini(struct radeon_device *rdev); + + +/* + * RING helpers. + */ +static inline void radeon_ring_write(struct radeon_device *rdev, uint32_t v) +{ +#if DRM_DEBUG_CODE + if (rdev->cp.count_dw <= 0) { + DRM_ERROR("radeon: writting more dword to ring than expected !\n"); + } +#endif + rdev->cp.ring[rdev->cp.wptr++] = v; + rdev->cp.wptr &= rdev->cp.ptr_mask; + rdev->cp.count_dw--; + rdev->cp.ring_free_dw--; +} + + +/* + * ASICs macro. + */ +#define radeon_init(rdev) (rdev)->asic->init((rdev)) +#define radeon_fini(rdev) (rdev)->asic->fini((rdev)) +#define radeon_resume(rdev) (rdev)->asic->resume((rdev)) +#define radeon_suspend(rdev) (rdev)->asic->suspend((rdev)) +#define radeon_cs_parse(p) rdev->asic->cs_parse((p)) +#define radeon_vga_set_state(rdev, state) (rdev)->asic->vga_set_state((rdev), (state)) +#define radeon_gpu_is_lockup(rdev) (rdev)->asic->gpu_is_lockup((rdev)) +#define radeon_asic_reset(rdev) (rdev)->asic->asic_reset((rdev)) +#define radeon_gart_tlb_flush(rdev) (rdev)->asic->gart_tlb_flush((rdev)) +#define radeon_gart_set_page(rdev, i, p) (rdev)->asic->gart_set_page((rdev), (i), (p)) +#define radeon_cp_commit(rdev) (rdev)->asic->cp_commit((rdev)) +#define radeon_ring_start(rdev) (rdev)->asic->ring_start((rdev)) +#define radeon_ring_test(rdev) (rdev)->asic->ring_test((rdev)) +#define radeon_ring_ib_execute(rdev, ib) (rdev)->asic->ring_ib_execute((rdev), (ib)) +#define radeon_irq_set(rdev) (rdev)->asic->irq_set((rdev)) +#define radeon_irq_process(rdev) (rdev)->asic->irq_process((rdev)) +#define radeon_get_vblank_counter(rdev, crtc) (rdev)->asic->get_vblank_counter((rdev), (crtc)) +#define radeon_fence_ring_emit(rdev, fence) (rdev)->asic->fence_ring_emit((rdev), (fence)) +#define radeon_copy_blit(rdev, s, d, np, f) (rdev)->asic->copy_blit((rdev), (s), (d), (np), (f)) +#define radeon_copy_dma(rdev, s, d, np, f) (rdev)->asic->copy_dma((rdev), (s), (d), (np), (f)) +#define radeon_copy(rdev, s, d, np, f) (rdev)->asic->copy((rdev), (s), (d), (np), (f)) +#define radeon_get_engine_clock(rdev) (rdev)->asic->get_engine_clock((rdev)) +#define radeon_set_engine_clock(rdev, e) (rdev)->asic->set_engine_clock((rdev), (e)) +#define radeon_get_memory_clock(rdev) (rdev)->asic->get_memory_clock((rdev)) +#define radeon_set_memory_clock(rdev, e) (rdev)->asic->set_memory_clock((rdev), (e)) +#define radeon_get_pcie_lanes(rdev) (rdev)->asic->get_pcie_lanes((rdev)) +#define radeon_set_pcie_lanes(rdev, l) (rdev)->asic->set_pcie_lanes((rdev), (l)) +#define radeon_set_clock_gating(rdev, e) (rdev)->asic->set_clock_gating((rdev), (e)) +#define radeon_set_surface_reg(rdev, r, f, p, o, s) ((rdev)->asic->set_surface_reg((rdev), (r), (f), (p), (o), (s))) +#define radeon_clear_surface_reg(rdev, r) ((rdev)->asic->clear_surface_reg((rdev), (r))) +#define radeon_bandwidth_update(rdev) (rdev)->asic->bandwidth_update((rdev)) +#define radeon_hpd_init(rdev) (rdev)->asic->hpd_init((rdev)) +#define radeon_hpd_fini(rdev) (rdev)->asic->hpd_fini((rdev)) +#define radeon_hpd_sense(rdev, hpd) (rdev)->asic->hpd_sense((rdev), (hpd)) +#define radeon_hpd_set_polarity(rdev, hpd) (rdev)->asic->hpd_set_polarity((rdev), (hpd)) +#define radeon_gui_idle(rdev) (rdev)->asic->gui_idle((rdev)) +#define radeon_pm_misc(rdev) (rdev)->asic->pm_misc((rdev)) +#define radeon_pm_prepare(rdev) (rdev)->asic->pm_prepare((rdev)) +#define radeon_pm_finish(rdev) (rdev)->asic->pm_finish((rdev)) +#define radeon_pm_init_profile(rdev) (rdev)->asic->pm_init_profile((rdev)) +#define radeon_pm_get_dynpm_state(rdev) (rdev)->asic->pm_get_dynpm_state((rdev)) +#define radeon_pre_page_flip(rdev, crtc) rdev->asic->pre_page_flip((rdev), (crtc)) +#define radeon_page_flip(rdev, crtc, base) rdev->asic->page_flip((rdev), (crtc), (base)) +#define radeon_post_page_flip(rdev, crtc) rdev->asic->post_page_flip((rdev), (crtc)) + +/* Common functions */ +/* AGP */ +extern int radeon_gpu_reset(struct radeon_device *rdev); +extern void radeon_agp_disable(struct radeon_device *rdev); +extern int radeon_gart_table_vram_pin(struct radeon_device *rdev); +extern void radeon_gart_restore(struct radeon_device *rdev); +extern int radeon_modeset_init(struct radeon_device *rdev); +extern void radeon_modeset_fini(struct radeon_device *rdev); +extern bool radeon_card_posted(struct radeon_device *rdev); +extern void radeon_update_bandwidth_info(struct radeon_device *rdev); +extern void radeon_update_display_priority(struct radeon_device *rdev); +extern bool radeon_boot_test_post_card(struct radeon_device *rdev); +extern void radeon_scratch_init(struct radeon_device *rdev); +extern void radeon_wb_fini(struct radeon_device *rdev); +extern int radeon_wb_init(struct radeon_device *rdev); +extern void radeon_wb_disable(struct radeon_device *rdev); +extern void radeon_surface_init(struct radeon_device *rdev); +extern int radeon_cs_parser_init(struct radeon_cs_parser *p, void *data); +extern void radeon_legacy_set_clock_gating(struct radeon_device *rdev, int enable); +extern void radeon_atom_set_clock_gating(struct radeon_device *rdev, int enable); +extern void radeon_ttm_placement_from_domain(struct radeon_bo *rbo, u32 domain); +extern bool radeon_ttm_bo_is_radeon_bo(struct ttm_buffer_object *bo); +extern void radeon_vram_location(struct radeon_device *rdev, struct radeon_mc *mc, u64 base); +extern void radeon_gtt_location(struct radeon_device *rdev, struct radeon_mc *mc); +extern int radeon_resume_kms(struct drm_device *dev); +extern int radeon_suspend_kms(struct drm_device *dev, pm_message_t state); +extern void radeon_ttm_set_active_vram_size(struct radeon_device *rdev, u64 size); + +/* + * r600 functions used by radeon_encoder.c + */ +extern void r600_hdmi_enable(struct drm_encoder *encoder); +extern void r600_hdmi_disable(struct drm_encoder *encoder); +extern void r600_hdmi_setmode(struct drm_encoder *encoder, struct drm_display_mode *mode); + +extern int ni_init_microcode(struct radeon_device *rdev); +extern int ni_mc_load_microcode(struct radeon_device *rdev); + +/* radeon_acpi.c */ +#if defined(CONFIG_ACPI) +extern int radeon_acpi_init(struct radeon_device *rdev); +#else +static inline int radeon_acpi_init(struct radeon_device *rdev) { return 0; } +#endif + +#include "radeon_object.h" + +#endif diff --git a/drivers/gpu/drm/radeon/radeon_acpi.c b/drivers/gpu/drm/radeon/radeon_acpi.c new file mode 100644 index 00000000..3f6636bb --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_acpi.c @@ -0,0 +1,67 @@ +#include +#include +#include +#include +#include + +#include "drmP.h" +#include "drm.h" +#include "drm_sarea.h" +#include "drm_crtc_helper.h" +#include "radeon.h" + +#include + +/* Call the ATIF method + * + * Note: currently we discard the output + */ +static int radeon_atif_call(acpi_handle handle) +{ + acpi_status status; + union acpi_object atif_arg_elements[2]; + struct acpi_object_list atif_arg; + struct acpi_buffer buffer = { ACPI_ALLOCATE_BUFFER, NULL}; + + atif_arg.count = 2; + atif_arg.pointer = &atif_arg_elements[0]; + + atif_arg_elements[0].type = ACPI_TYPE_INTEGER; + atif_arg_elements[0].integer.value = 0; + atif_arg_elements[1].type = ACPI_TYPE_INTEGER; + atif_arg_elements[1].integer.value = 0; + + status = acpi_evaluate_object(handle, "ATIF", &atif_arg, &buffer); + + /* Fail only if calling the method fails and ATIF is supported */ + if (ACPI_FAILURE(status) && status != AE_NOT_FOUND) { + printk(KERN_DEBUG "failed to evaluate ATIF got %s\n", acpi_format_exception(status)); + kfree(buffer.pointer); + return 1; + } + + kfree(buffer.pointer); + return 0; +} + +/* Call all ACPI methods here */ +int radeon_acpi_init(struct radeon_device *rdev) +{ + acpi_handle handle; + int ret; + + /* No need to proceed if we're sure that ATIF is not supported */ + if (!ASIC_IS_AVIVO(rdev) || !rdev->bios) + return 0; + + /* Get the device handle */ + handle = DEVICE_ACPI_HANDLE(&rdev->pdev->dev); + + /* Call the ATIF method */ + ret = radeon_atif_call(handle); + if (ret) + return ret; + + return 0; +} + diff --git a/drivers/gpu/drm/radeon/radeon_agp.c b/drivers/gpu/drm/radeon/radeon_agp.c new file mode 100644 index 00000000..bd2f33e5 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_agp.c @@ -0,0 +1,283 @@ +/* + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: + * Dave Airlie + * Jerome Glisse + */ +#include "drmP.h" +#include "drm.h" +#include "radeon.h" +#include "radeon_drm.h" + +#if __OS_HAS_AGP + +struct radeon_agpmode_quirk { + u32 hostbridge_vendor; + u32 hostbridge_device; + u32 chip_vendor; + u32 chip_device; + u32 subsys_vendor; + u32 subsys_device; + u32 default_mode; +}; + +static struct radeon_agpmode_quirk radeon_agpmode_quirk_list[] = { + /* Intel E7505 Memory Controller Hub / RV350 AR [Radeon 9600XT] Needs AGPMode 4 (deb #515326) */ + { PCI_VENDOR_ID_INTEL, 0x2550, PCI_VENDOR_ID_ATI, 0x4152, 0x1458, 0x4038, 4}, + /* Intel 82865G/PE/P DRAM Controller/Host-Hub / Mobility 9800 Needs AGPMode 4 (deb #462590) */ + { PCI_VENDOR_ID_INTEL, 0x2570, PCI_VENDOR_ID_ATI, 0x4a4e, PCI_VENDOR_ID_DELL, 0x5106, 4}, + /* Intel 82865G/PE/P DRAM Controller/Host-Hub / RV280 [Radeon 9200 SE] Needs AGPMode 4 (lp #300304) */ + { PCI_VENDOR_ID_INTEL, 0x2570, PCI_VENDOR_ID_ATI, 0x5964, + 0x148c, 0x2073, 4}, + /* Intel 82855PM Processor to I/O Controller / Mobility M6 LY Needs AGPMode 1 (deb #467235) */ + { PCI_VENDOR_ID_INTEL, 0x3340, PCI_VENDOR_ID_ATI, 0x4c59, + PCI_VENDOR_ID_IBM, 0x052f, 1}, + /* Intel 82855PM host bridge / Mobility 9600 M10 RV350 Needs AGPMode 1 (lp #195051) */ + { PCI_VENDOR_ID_INTEL, 0x3340, PCI_VENDOR_ID_ATI, 0x4e50, + PCI_VENDOR_ID_IBM, 0x0550, 1}, + /* Intel 82855PM host bridge / Mobility M7 needs AGPMode 1 */ + { PCI_VENDOR_ID_INTEL, 0x3340, PCI_VENDOR_ID_ATI, 0x4c57, + PCI_VENDOR_ID_IBM, 0x0530, 1}, + /* Intel 82855PM host bridge / FireGL Mobility T2 RV350 Needs AGPMode 2 (fdo #20647) */ + { PCI_VENDOR_ID_INTEL, 0x3340, PCI_VENDOR_ID_ATI, 0x4e54, + PCI_VENDOR_ID_IBM, 0x054f, 2}, + /* Intel 82855PM host bridge / Mobility M9+ / VaioPCG-V505DX Needs AGPMode 2 (fdo #17928) */ + { PCI_VENDOR_ID_INTEL, 0x3340, PCI_VENDOR_ID_ATI, 0x5c61, + PCI_VENDOR_ID_SONY, 0x816b, 2}, + /* Intel 82855PM Processor to I/O Controller / Mobility M9+ Needs AGPMode 8 (phoronix forum) */ + { PCI_VENDOR_ID_INTEL, 0x3340, PCI_VENDOR_ID_ATI, 0x5c61, + PCI_VENDOR_ID_SONY, 0x8195, 8}, + /* Intel 82830 830 Chipset Host Bridge / Mobility M6 LY Needs AGPMode 2 (fdo #17360)*/ + { PCI_VENDOR_ID_INTEL, 0x3575, PCI_VENDOR_ID_ATI, 0x4c59, + PCI_VENDOR_ID_DELL, 0x00e3, 2}, + /* Intel 82852/82855 host bridge / Mobility FireGL 9000 R250 Needs AGPMode 1 (lp #296617) */ + { PCI_VENDOR_ID_INTEL, 0x3580, PCI_VENDOR_ID_ATI, 0x4c66, + PCI_VENDOR_ID_DELL, 0x0149, 1}, + /* Intel 82852/82855 host bridge / Mobility 9600 M10 RV350 Needs AGPMode 1 (deb #467460) */ + { PCI_VENDOR_ID_INTEL, 0x3580, PCI_VENDOR_ID_ATI, 0x4e50, + 0x1025, 0x0061, 1}, + /* Intel 82852/82855 host bridge / Mobility 9600 M10 RV350 Needs AGPMode 1 (lp #203007) */ + { PCI_VENDOR_ID_INTEL, 0x3580, PCI_VENDOR_ID_ATI, 0x4e50, + 0x1025, 0x0064, 1}, + /* Intel 82852/82855 host bridge / Mobility 9600 M10 RV350 Needs AGPMode 1 (lp #141551) */ + { PCI_VENDOR_ID_INTEL, 0x3580, PCI_VENDOR_ID_ATI, 0x4e50, + PCI_VENDOR_ID_ASUSTEK, 0x1942, 1}, + /* Intel 82852/82855 host bridge / Mobility 9600/9700 Needs AGPMode 1 (deb #510208) */ + { PCI_VENDOR_ID_INTEL, 0x3580, PCI_VENDOR_ID_ATI, 0x4e50, + 0x10cf, 0x127f, 1}, + /* ASRock K7VT4A+ AGP 8x / ATI Radeon 9250 AGP Needs AGPMode 4 (lp #133192) */ + { 0x1849, 0x3189, PCI_VENDOR_ID_ATI, 0x5960, + 0x1787, 0x5960, 4}, + /* VIA K8M800 Host Bridge / RV280 [Radeon 9200 PRO] Needs AGPMode 4 (fdo #12544) */ + { PCI_VENDOR_ID_VIA, 0x0204, PCI_VENDOR_ID_ATI, 0x5960, + 0x17af, 0x2020, 4}, + /* VIA KT880 Host Bridge / RV350 [Radeon 9550] Needs AGPMode 4 (fdo #19981) */ + { PCI_VENDOR_ID_VIA, 0x0269, PCI_VENDOR_ID_ATI, 0x4153, + PCI_VENDOR_ID_ASUSTEK, 0x003c, 4}, + /* VIA VT8363 Host Bridge / R200 QL [Radeon 8500] Needs AGPMode 2 (lp #141551) */ + { PCI_VENDOR_ID_VIA, 0x0305, PCI_VENDOR_ID_ATI, 0x514c, + PCI_VENDOR_ID_ATI, 0x013a, 2}, + /* VIA VT82C693A Host Bridge / RV280 [Radeon 9200 PRO] Needs AGPMode 2 (deb #515512) */ + { PCI_VENDOR_ID_VIA, 0x0691, PCI_VENDOR_ID_ATI, 0x5960, + PCI_VENDOR_ID_ASUSTEK, 0x004c, 2}, + /* VIA VT82C693A Host Bridge / RV280 [Radeon 9200 PRO] Needs AGPMode 2 */ + { PCI_VENDOR_ID_VIA, 0x0691, PCI_VENDOR_ID_ATI, 0x5960, + PCI_VENDOR_ID_ASUSTEK, 0x0054, 2}, + /* VIA VT8377 Host Bridge / R200 QM [Radeon 9100] Needs AGPMode 4 (deb #461144) */ + { PCI_VENDOR_ID_VIA, 0x3189, PCI_VENDOR_ID_ATI, 0x514d, + 0x174b, 0x7149, 4}, + /* VIA VT8377 Host Bridge / RV280 [Radeon 9200 PRO] Needs AGPMode 4 (lp #312693) */ + { PCI_VENDOR_ID_VIA, 0x3189, PCI_VENDOR_ID_ATI, 0x5960, + 0x1462, 0x0380, 4}, + /* VIA VT8377 Host Bridge / RV280 Needs AGPMode 4 (ati ML) */ + { PCI_VENDOR_ID_VIA, 0x3189, PCI_VENDOR_ID_ATI, 0x5964, + 0x148c, 0x2073, 4}, + /* ATI Host Bridge / RV280 [M9+] Needs AGPMode 1 (phoronix forum) */ + { PCI_VENDOR_ID_ATI, 0xcbb2, PCI_VENDOR_ID_ATI, 0x5c61, + PCI_VENDOR_ID_SONY, 0x8175, 1}, + /* HP Host Bridge / R300 [FireGL X1] Needs AGPMode 2 (fdo #7770) */ + { PCI_VENDOR_ID_HP, 0x122e, PCI_VENDOR_ID_ATI, 0x4e47, + PCI_VENDOR_ID_ATI, 0x0152, 2}, + { 0, 0, 0, 0, 0, 0, 0 }, +}; +#endif + +int radeon_agp_init(struct radeon_device *rdev) +{ +#if __OS_HAS_AGP + struct radeon_agpmode_quirk *p = radeon_agpmode_quirk_list; + struct drm_agp_mode mode; + struct drm_agp_info info; + uint32_t agp_status; + int default_mode; + bool is_v3; + int ret; + + /* Acquire AGP. */ + ret = drm_agp_acquire(rdev->ddev); + if (ret) { + DRM_ERROR("Unable to acquire AGP: %d\n", ret); + return ret; + } + + ret = drm_agp_info(rdev->ddev, &info); + if (ret) { + drm_agp_release(rdev->ddev); + DRM_ERROR("Unable to get AGP info: %d\n", ret); + return ret; + } + + if (rdev->ddev->agp->agp_info.aper_size < 32) { + drm_agp_release(rdev->ddev); + dev_warn(rdev->dev, "AGP aperture too small (%zuM) " + "need at least 32M, disabling AGP\n", + rdev->ddev->agp->agp_info.aper_size); + return -EINVAL; + } + + mode.mode = info.mode; + /* chips with the agp to pcie bridge don't have the AGP_STATUS register + * Just use the whatever mode the host sets up. + */ + if (rdev->family <= CHIP_RV350) + agp_status = (RREG32(RADEON_AGP_STATUS) | RADEON_AGPv3_MODE) & mode.mode; + else + agp_status = mode.mode; + is_v3 = !!(agp_status & RADEON_AGPv3_MODE); + + if (is_v3) { + default_mode = (agp_status & RADEON_AGPv3_8X_MODE) ? 8 : 4; + } else { + if (agp_status & RADEON_AGP_4X_MODE) { + default_mode = 4; + } else if (agp_status & RADEON_AGP_2X_MODE) { + default_mode = 2; + } else { + default_mode = 1; + } + } + + /* Apply AGPMode Quirks */ + while (p && p->chip_device != 0) { + if (info.id_vendor == p->hostbridge_vendor && + info.id_device == p->hostbridge_device && + rdev->pdev->vendor == p->chip_vendor && + rdev->pdev->device == p->chip_device && + rdev->pdev->subsystem_vendor == p->subsys_vendor && + rdev->pdev->subsystem_device == p->subsys_device) { + default_mode = p->default_mode; + } + ++p; + } + + if (radeon_agpmode > 0) { + if ((radeon_agpmode < (is_v3 ? 4 : 1)) || + (radeon_agpmode > (is_v3 ? 8 : 4)) || + (radeon_agpmode & (radeon_agpmode - 1))) { + DRM_ERROR("Illegal AGP Mode: %d (valid %s), leaving at %d\n", + radeon_agpmode, is_v3 ? "4, 8" : "1, 2, 4", + default_mode); + radeon_agpmode = default_mode; + } else { + DRM_INFO("AGP mode requested: %d\n", radeon_agpmode); + } + } else { + radeon_agpmode = default_mode; + } + + mode.mode &= ~RADEON_AGP_MODE_MASK; + if (is_v3) { + switch (radeon_agpmode) { + case 8: + mode.mode |= RADEON_AGPv3_8X_MODE; + break; + case 4: + default: + mode.mode |= RADEON_AGPv3_4X_MODE; + break; + } + } else { + switch (radeon_agpmode) { + case 4: + mode.mode |= RADEON_AGP_4X_MODE; + break; + case 2: + mode.mode |= RADEON_AGP_2X_MODE; + break; + case 1: + default: + mode.mode |= RADEON_AGP_1X_MODE; + break; + } + } + + mode.mode &= ~RADEON_AGP_FW_MODE; /* disable fw */ + ret = drm_agp_enable(rdev->ddev, mode); + if (ret) { + DRM_ERROR("Unable to enable AGP (mode = 0x%lx)\n", mode.mode); + drm_agp_release(rdev->ddev); + return ret; + } + + rdev->mc.agp_base = rdev->ddev->agp->agp_info.aper_base; + rdev->mc.gtt_size = rdev->ddev->agp->agp_info.aper_size << 20; + rdev->mc.gtt_start = rdev->mc.agp_base; + rdev->mc.gtt_end = rdev->mc.gtt_start + rdev->mc.gtt_size - 1; + dev_info(rdev->dev, "GTT: %lluM 0x%08llX - 0x%08llX\n", + rdev->mc.gtt_size >> 20, rdev->mc.gtt_start, rdev->mc.gtt_end); + + /* workaround some hw issues */ + if (rdev->family < CHIP_R200) { + WREG32(RADEON_AGP_CNTL, RREG32(RADEON_AGP_CNTL) | 0x000e0000); + } + return 0; +#else + return 0; +#endif +} + +void radeon_agp_resume(struct radeon_device *rdev) +{ +#if __OS_HAS_AGP + int r; + if (rdev->flags & RADEON_IS_AGP) { + r = radeon_agp_init(rdev); + if (r) + dev_warn(rdev->dev, "radeon AGP reinit failed\n"); + } +#endif +} + +void radeon_agp_fini(struct radeon_device *rdev) +{ +#if __OS_HAS_AGP + if (rdev->ddev->agp && rdev->ddev->agp->acquired) { + drm_agp_release(rdev->ddev); + } +#endif +} + +void radeon_agp_suspend(struct radeon_device *rdev) +{ + radeon_agp_fini(rdev); +} diff --git a/drivers/gpu/drm/radeon/radeon_asic.c b/drivers/gpu/drm/radeon/radeon_asic.c new file mode 100644 index 00000000..b2449629 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_asic.c @@ -0,0 +1,1066 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ + +#include +#include +#include +#include +#include +#include +#include "radeon_reg.h" +#include "radeon.h" +#include "radeon_asic.h" +#include "atom.h" + +/* + * Registers accessors functions. + */ +static uint32_t radeon_invalid_rreg(struct radeon_device *rdev, uint32_t reg) +{ + DRM_ERROR("Invalid callback to read register 0x%04X\n", reg); + BUG_ON(1); + return 0; +} + +static void radeon_invalid_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v) +{ + DRM_ERROR("Invalid callback to write register 0x%04X with 0x%08X\n", + reg, v); + BUG_ON(1); +} + +static void radeon_register_accessor_init(struct radeon_device *rdev) +{ + rdev->mc_rreg = &radeon_invalid_rreg; + rdev->mc_wreg = &radeon_invalid_wreg; + rdev->pll_rreg = &radeon_invalid_rreg; + rdev->pll_wreg = &radeon_invalid_wreg; + rdev->pciep_rreg = &radeon_invalid_rreg; + rdev->pciep_wreg = &radeon_invalid_wreg; + + /* Don't change order as we are overridding accessor. */ + if (rdev->family < CHIP_RV515) { + rdev->pcie_reg_mask = 0xff; + } else { + rdev->pcie_reg_mask = 0x7ff; + } + /* FIXME: not sure here */ + if (rdev->family <= CHIP_R580) { + rdev->pll_rreg = &r100_pll_rreg; + rdev->pll_wreg = &r100_pll_wreg; + } + if (rdev->family >= CHIP_R420) { + rdev->mc_rreg = &r420_mc_rreg; + rdev->mc_wreg = &r420_mc_wreg; + } + if (rdev->family >= CHIP_RV515) { + rdev->mc_rreg = &rv515_mc_rreg; + rdev->mc_wreg = &rv515_mc_wreg; + } + if (rdev->family == CHIP_RS400 || rdev->family == CHIP_RS480) { + rdev->mc_rreg = &rs400_mc_rreg; + rdev->mc_wreg = &rs400_mc_wreg; + } + if (rdev->family == CHIP_RS690 || rdev->family == CHIP_RS740) { + rdev->mc_rreg = &rs690_mc_rreg; + rdev->mc_wreg = &rs690_mc_wreg; + } + if (rdev->family == CHIP_RS600) { + rdev->mc_rreg = &rs600_mc_rreg; + rdev->mc_wreg = &rs600_mc_wreg; + } + if (rdev->family >= CHIP_R600) { + rdev->pciep_rreg = &r600_pciep_rreg; + rdev->pciep_wreg = &r600_pciep_wreg; + } +} + + +/* helper to disable agp */ +void radeon_agp_disable(struct radeon_device *rdev) +{ + rdev->flags &= ~RADEON_IS_AGP; + if (rdev->family >= CHIP_R600) { + DRM_INFO("Forcing AGP to PCIE mode\n"); + rdev->flags |= RADEON_IS_PCIE; + } else if (rdev->family >= CHIP_RV515 || + rdev->family == CHIP_RV380 || + rdev->family == CHIP_RV410 || + rdev->family == CHIP_R423) { + DRM_INFO("Forcing AGP to PCIE mode\n"); + rdev->flags |= RADEON_IS_PCIE; + rdev->asic->gart_tlb_flush = &rv370_pcie_gart_tlb_flush; + rdev->asic->gart_set_page = &rv370_pcie_gart_set_page; + } else { + DRM_INFO("Forcing AGP to PCI mode\n"); + rdev->flags |= RADEON_IS_PCI; + rdev->asic->gart_tlb_flush = &r100_pci_gart_tlb_flush; + rdev->asic->gart_set_page = &r100_pci_gart_set_page; + } + rdev->mc.gtt_size = radeon_gart_size * 1024 * 1024; +} + +/* + * ASIC + */ +static struct radeon_asic r100_asic = { + .init = &r100_init, + .fini = &r100_fini, + .suspend = &r100_suspend, + .resume = &r100_resume, + .vga_set_state = &r100_vga_set_state, + .gpu_is_lockup = &r100_gpu_is_lockup, + .asic_reset = &r100_asic_reset, + .gart_tlb_flush = &r100_pci_gart_tlb_flush, + .gart_set_page = &r100_pci_gart_set_page, + .cp_commit = &r100_cp_commit, + .ring_start = &r100_ring_start, + .ring_test = &r100_ring_test, + .ring_ib_execute = &r100_ring_ib_execute, + .irq_set = &r100_irq_set, + .irq_process = &r100_irq_process, + .get_vblank_counter = &r100_get_vblank_counter, + .fence_ring_emit = &r100_fence_ring_emit, + .cs_parse = &r100_cs_parse, + .copy_blit = &r100_copy_blit, + .copy_dma = NULL, + .copy = &r100_copy_blit, + .get_engine_clock = &radeon_legacy_get_engine_clock, + .set_engine_clock = &radeon_legacy_set_engine_clock, + .get_memory_clock = &radeon_legacy_get_memory_clock, + .set_memory_clock = NULL, + .get_pcie_lanes = NULL, + .set_pcie_lanes = NULL, + .set_clock_gating = &radeon_legacy_set_clock_gating, + .set_surface_reg = r100_set_surface_reg, + .clear_surface_reg = r100_clear_surface_reg, + .bandwidth_update = &r100_bandwidth_update, + .hpd_init = &r100_hpd_init, + .hpd_fini = &r100_hpd_fini, + .hpd_sense = &r100_hpd_sense, + .hpd_set_polarity = &r100_hpd_set_polarity, + .ioctl_wait_idle = NULL, + .gui_idle = &r100_gui_idle, + .pm_misc = &r100_pm_misc, + .pm_prepare = &r100_pm_prepare, + .pm_finish = &r100_pm_finish, + .pm_init_profile = &r100_pm_init_profile, + .pm_get_dynpm_state = &r100_pm_get_dynpm_state, + .pre_page_flip = &r100_pre_page_flip, + .page_flip = &r100_page_flip, + .post_page_flip = &r100_post_page_flip, +}; + +static struct radeon_asic r200_asic = { + .init = &r100_init, + .fini = &r100_fini, + .suspend = &r100_suspend, + .resume = &r100_resume, + .vga_set_state = &r100_vga_set_state, + .gpu_is_lockup = &r100_gpu_is_lockup, + .asic_reset = &r100_asic_reset, + .gart_tlb_flush = &r100_pci_gart_tlb_flush, + .gart_set_page = &r100_pci_gart_set_page, + .cp_commit = &r100_cp_commit, + .ring_start = &r100_ring_start, + .ring_test = &r100_ring_test, + .ring_ib_execute = &r100_ring_ib_execute, + .irq_set = &r100_irq_set, + .irq_process = &r100_irq_process, + .get_vblank_counter = &r100_get_vblank_counter, + .fence_ring_emit = &r100_fence_ring_emit, + .cs_parse = &r100_cs_parse, + .copy_blit = &r100_copy_blit, + .copy_dma = &r200_copy_dma, + .copy = &r100_copy_blit, + .get_engine_clock = &radeon_legacy_get_engine_clock, + .set_engine_clock = &radeon_legacy_set_engine_clock, + .get_memory_clock = &radeon_legacy_get_memory_clock, + .set_memory_clock = NULL, + .set_pcie_lanes = NULL, + .set_clock_gating = &radeon_legacy_set_clock_gating, + .set_surface_reg = r100_set_surface_reg, + .clear_surface_reg = r100_clear_surface_reg, + .bandwidth_update = &r100_bandwidth_update, + .hpd_init = &r100_hpd_init, + .hpd_fini = &r100_hpd_fini, + .hpd_sense = &r100_hpd_sense, + .hpd_set_polarity = &r100_hpd_set_polarity, + .ioctl_wait_idle = NULL, + .gui_idle = &r100_gui_idle, + .pm_misc = &r100_pm_misc, + .pm_prepare = &r100_pm_prepare, + .pm_finish = &r100_pm_finish, + .pm_init_profile = &r100_pm_init_profile, + .pm_get_dynpm_state = &r100_pm_get_dynpm_state, + .pre_page_flip = &r100_pre_page_flip, + .page_flip = &r100_page_flip, + .post_page_flip = &r100_post_page_flip, +}; + +static struct radeon_asic r300_asic = { + .init = &r300_init, + .fini = &r300_fini, + .suspend = &r300_suspend, + .resume = &r300_resume, + .vga_set_state = &r100_vga_set_state, + .gpu_is_lockup = &r300_gpu_is_lockup, + .asic_reset = &r300_asic_reset, + .gart_tlb_flush = &r100_pci_gart_tlb_flush, + .gart_set_page = &r100_pci_gart_set_page, + .cp_commit = &r100_cp_commit, + .ring_start = &r300_ring_start, + .ring_test = &r100_ring_test, + .ring_ib_execute = &r100_ring_ib_execute, + .irq_set = &r100_irq_set, + .irq_process = &r100_irq_process, + .get_vblank_counter = &r100_get_vblank_counter, + .fence_ring_emit = &r300_fence_ring_emit, + .cs_parse = &r300_cs_parse, + .copy_blit = &r100_copy_blit, + .copy_dma = &r200_copy_dma, + .copy = &r100_copy_blit, + .get_engine_clock = &radeon_legacy_get_engine_clock, + .set_engine_clock = &radeon_legacy_set_engine_clock, + .get_memory_clock = &radeon_legacy_get_memory_clock, + .set_memory_clock = NULL, + .get_pcie_lanes = &rv370_get_pcie_lanes, + .set_pcie_lanes = &rv370_set_pcie_lanes, + .set_clock_gating = &radeon_legacy_set_clock_gating, + .set_surface_reg = r100_set_surface_reg, + .clear_surface_reg = r100_clear_surface_reg, + .bandwidth_update = &r100_bandwidth_update, + .hpd_init = &r100_hpd_init, + .hpd_fini = &r100_hpd_fini, + .hpd_sense = &r100_hpd_sense, + .hpd_set_polarity = &r100_hpd_set_polarity, + .ioctl_wait_idle = NULL, + .gui_idle = &r100_gui_idle, + .pm_misc = &r100_pm_misc, + .pm_prepare = &r100_pm_prepare, + .pm_finish = &r100_pm_finish, + .pm_init_profile = &r100_pm_init_profile, + .pm_get_dynpm_state = &r100_pm_get_dynpm_state, + .pre_page_flip = &r100_pre_page_flip, + .page_flip = &r100_page_flip, + .post_page_flip = &r100_post_page_flip, +}; + +static struct radeon_asic r300_asic_pcie = { + .init = &r300_init, + .fini = &r300_fini, + .suspend = &r300_suspend, + .resume = &r300_resume, + .vga_set_state = &r100_vga_set_state, + .gpu_is_lockup = &r300_gpu_is_lockup, + .asic_reset = &r300_asic_reset, + .gart_tlb_flush = &rv370_pcie_gart_tlb_flush, + .gart_set_page = &rv370_pcie_gart_set_page, + .cp_commit = &r100_cp_commit, + .ring_start = &r300_ring_start, + .ring_test = &r100_ring_test, + .ring_ib_execute = &r100_ring_ib_execute, + .irq_set = &r100_irq_set, + .irq_process = &r100_irq_process, + .get_vblank_counter = &r100_get_vblank_counter, + .fence_ring_emit = &r300_fence_ring_emit, + .cs_parse = &r300_cs_parse, + .copy_blit = &r100_copy_blit, + .copy_dma = &r200_copy_dma, + .copy = &r100_copy_blit, + .get_engine_clock = &radeon_legacy_get_engine_clock, + .set_engine_clock = &radeon_legacy_set_engine_clock, + .get_memory_clock = &radeon_legacy_get_memory_clock, + .set_memory_clock = NULL, + .set_pcie_lanes = &rv370_set_pcie_lanes, + .set_clock_gating = &radeon_legacy_set_clock_gating, + .set_surface_reg = r100_set_surface_reg, + .clear_surface_reg = r100_clear_surface_reg, + .bandwidth_update = &r100_bandwidth_update, + .hpd_init = &r100_hpd_init, + .hpd_fini = &r100_hpd_fini, + .hpd_sense = &r100_hpd_sense, + .hpd_set_polarity = &r100_hpd_set_polarity, + .ioctl_wait_idle = NULL, + .gui_idle = &r100_gui_idle, + .pm_misc = &r100_pm_misc, + .pm_prepare = &r100_pm_prepare, + .pm_finish = &r100_pm_finish, + .pm_init_profile = &r100_pm_init_profile, + .pm_get_dynpm_state = &r100_pm_get_dynpm_state, + .pre_page_flip = &r100_pre_page_flip, + .page_flip = &r100_page_flip, + .post_page_flip = &r100_post_page_flip, +}; + +static struct radeon_asic r420_asic = { + .init = &r420_init, + .fini = &r420_fini, + .suspend = &r420_suspend, + .resume = &r420_resume, + .vga_set_state = &r100_vga_set_state, + .gpu_is_lockup = &r300_gpu_is_lockup, + .asic_reset = &r300_asic_reset, + .gart_tlb_flush = &rv370_pcie_gart_tlb_flush, + .gart_set_page = &rv370_pcie_gart_set_page, + .cp_commit = &r100_cp_commit, + .ring_start = &r300_ring_start, + .ring_test = &r100_ring_test, + .ring_ib_execute = &r100_ring_ib_execute, + .irq_set = &r100_irq_set, + .irq_process = &r100_irq_process, + .get_vblank_counter = &r100_get_vblank_counter, + .fence_ring_emit = &r300_fence_ring_emit, + .cs_parse = &r300_cs_parse, + .copy_blit = &r100_copy_blit, + .copy_dma = &r200_copy_dma, + .copy = &r100_copy_blit, + .get_engine_clock = &radeon_atom_get_engine_clock, + .set_engine_clock = &radeon_atom_set_engine_clock, + .get_memory_clock = &radeon_atom_get_memory_clock, + .set_memory_clock = &radeon_atom_set_memory_clock, + .get_pcie_lanes = &rv370_get_pcie_lanes, + .set_pcie_lanes = &rv370_set_pcie_lanes, + .set_clock_gating = &radeon_atom_set_clock_gating, + .set_surface_reg = r100_set_surface_reg, + .clear_surface_reg = r100_clear_surface_reg, + .bandwidth_update = &r100_bandwidth_update, + .hpd_init = &r100_hpd_init, + .hpd_fini = &r100_hpd_fini, + .hpd_sense = &r100_hpd_sense, + .hpd_set_polarity = &r100_hpd_set_polarity, + .ioctl_wait_idle = NULL, + .gui_idle = &r100_gui_idle, + .pm_misc = &r100_pm_misc, + .pm_prepare = &r100_pm_prepare, + .pm_finish = &r100_pm_finish, + .pm_init_profile = &r420_pm_init_profile, + .pm_get_dynpm_state = &r100_pm_get_dynpm_state, + .pre_page_flip = &r100_pre_page_flip, + .page_flip = &r100_page_flip, + .post_page_flip = &r100_post_page_flip, +}; + +static struct radeon_asic rs400_asic = { + .init = &rs400_init, + .fini = &rs400_fini, + .suspend = &rs400_suspend, + .resume = &rs400_resume, + .vga_set_state = &r100_vga_set_state, + .gpu_is_lockup = &r300_gpu_is_lockup, + .asic_reset = &r300_asic_reset, + .gart_tlb_flush = &rs400_gart_tlb_flush, + .gart_set_page = &rs400_gart_set_page, + .cp_commit = &r100_cp_commit, + .ring_start = &r300_ring_start, + .ring_test = &r100_ring_test, + .ring_ib_execute = &r100_ring_ib_execute, + .irq_set = &r100_irq_set, + .irq_process = &r100_irq_process, + .get_vblank_counter = &r100_get_vblank_counter, + .fence_ring_emit = &r300_fence_ring_emit, + .cs_parse = &r300_cs_parse, + .copy_blit = &r100_copy_blit, + .copy_dma = &r200_copy_dma, + .copy = &r100_copy_blit, + .get_engine_clock = &radeon_legacy_get_engine_clock, + .set_engine_clock = &radeon_legacy_set_engine_clock, + .get_memory_clock = &radeon_legacy_get_memory_clock, + .set_memory_clock = NULL, + .get_pcie_lanes = NULL, + .set_pcie_lanes = NULL, + .set_clock_gating = &radeon_legacy_set_clock_gating, + .set_surface_reg = r100_set_surface_reg, + .clear_surface_reg = r100_clear_surface_reg, + .bandwidth_update = &r100_bandwidth_update, + .hpd_init = &r100_hpd_init, + .hpd_fini = &r100_hpd_fini, + .hpd_sense = &r100_hpd_sense, + .hpd_set_polarity = &r100_hpd_set_polarity, + .ioctl_wait_idle = NULL, + .gui_idle = &r100_gui_idle, + .pm_misc = &r100_pm_misc, + .pm_prepare = &r100_pm_prepare, + .pm_finish = &r100_pm_finish, + .pm_init_profile = &r100_pm_init_profile, + .pm_get_dynpm_state = &r100_pm_get_dynpm_state, + .pre_page_flip = &r100_pre_page_flip, + .page_flip = &r100_page_flip, + .post_page_flip = &r100_post_page_flip, +}; + +static struct radeon_asic rs600_asic = { + .init = &rs600_init, + .fini = &rs600_fini, + .suspend = &rs600_suspend, + .resume = &rs600_resume, + .vga_set_state = &r100_vga_set_state, + .gpu_is_lockup = &r300_gpu_is_lockup, + .asic_reset = &rs600_asic_reset, + .gart_tlb_flush = &rs600_gart_tlb_flush, + .gart_set_page = &rs600_gart_set_page, + .cp_commit = &r100_cp_commit, + .ring_start = &r300_ring_start, + .ring_test = &r100_ring_test, + .ring_ib_execute = &r100_ring_ib_execute, + .irq_set = &rs600_irq_set, + .irq_process = &rs600_irq_process, + .get_vblank_counter = &rs600_get_vblank_counter, + .fence_ring_emit = &r300_fence_ring_emit, + .cs_parse = &r300_cs_parse, + .copy_blit = &r100_copy_blit, + .copy_dma = &r200_copy_dma, + .copy = &r100_copy_blit, + .get_engine_clock = &radeon_atom_get_engine_clock, + .set_engine_clock = &radeon_atom_set_engine_clock, + .get_memory_clock = &radeon_atom_get_memory_clock, + .set_memory_clock = &radeon_atom_set_memory_clock, + .get_pcie_lanes = NULL, + .set_pcie_lanes = NULL, + .set_clock_gating = &radeon_atom_set_clock_gating, + .set_surface_reg = r100_set_surface_reg, + .clear_surface_reg = r100_clear_surface_reg, + .bandwidth_update = &rs600_bandwidth_update, + .hpd_init = &rs600_hpd_init, + .hpd_fini = &rs600_hpd_fini, + .hpd_sense = &rs600_hpd_sense, + .hpd_set_polarity = &rs600_hpd_set_polarity, + .ioctl_wait_idle = NULL, + .gui_idle = &r100_gui_idle, + .pm_misc = &rs600_pm_misc, + .pm_prepare = &rs600_pm_prepare, + .pm_finish = &rs600_pm_finish, + .pm_init_profile = &r420_pm_init_profile, + .pm_get_dynpm_state = &r100_pm_get_dynpm_state, + .pre_page_flip = &rs600_pre_page_flip, + .page_flip = &rs600_page_flip, + .post_page_flip = &rs600_post_page_flip, +}; + +static struct radeon_asic rs690_asic = { + .init = &rs690_init, + .fini = &rs690_fini, + .suspend = &rs690_suspend, + .resume = &rs690_resume, + .vga_set_state = &r100_vga_set_state, + .gpu_is_lockup = &r300_gpu_is_lockup, + .asic_reset = &rs600_asic_reset, + .gart_tlb_flush = &rs400_gart_tlb_flush, + .gart_set_page = &rs400_gart_set_page, + .cp_commit = &r100_cp_commit, + .ring_start = &r300_ring_start, + .ring_test = &r100_ring_test, + .ring_ib_execute = &r100_ring_ib_execute, + .irq_set = &rs600_irq_set, + .irq_process = &rs600_irq_process, + .get_vblank_counter = &rs600_get_vblank_counter, + .fence_ring_emit = &r300_fence_ring_emit, + .cs_parse = &r300_cs_parse, + .copy_blit = &r100_copy_blit, + .copy_dma = &r200_copy_dma, + .copy = &r200_copy_dma, + .get_engine_clock = &radeon_atom_get_engine_clock, + .set_engine_clock = &radeon_atom_set_engine_clock, + .get_memory_clock = &radeon_atom_get_memory_clock, + .set_memory_clock = &radeon_atom_set_memory_clock, + .get_pcie_lanes = NULL, + .set_pcie_lanes = NULL, + .set_clock_gating = &radeon_atom_set_clock_gating, + .set_surface_reg = r100_set_surface_reg, + .clear_surface_reg = r100_clear_surface_reg, + .bandwidth_update = &rs690_bandwidth_update, + .hpd_init = &rs600_hpd_init, + .hpd_fini = &rs600_hpd_fini, + .hpd_sense = &rs600_hpd_sense, + .hpd_set_polarity = &rs600_hpd_set_polarity, + .ioctl_wait_idle = NULL, + .gui_idle = &r100_gui_idle, + .pm_misc = &rs600_pm_misc, + .pm_prepare = &rs600_pm_prepare, + .pm_finish = &rs600_pm_finish, + .pm_init_profile = &r420_pm_init_profile, + .pm_get_dynpm_state = &r100_pm_get_dynpm_state, + .pre_page_flip = &rs600_pre_page_flip, + .page_flip = &rs600_page_flip, + .post_page_flip = &rs600_post_page_flip, +}; + +static struct radeon_asic rv515_asic = { + .init = &rv515_init, + .fini = &rv515_fini, + .suspend = &rv515_suspend, + .resume = &rv515_resume, + .vga_set_state = &r100_vga_set_state, + .gpu_is_lockup = &r300_gpu_is_lockup, + .asic_reset = &rs600_asic_reset, + .gart_tlb_flush = &rv370_pcie_gart_tlb_flush, + .gart_set_page = &rv370_pcie_gart_set_page, + .cp_commit = &r100_cp_commit, + .ring_start = &rv515_ring_start, + .ring_test = &r100_ring_test, + .ring_ib_execute = &r100_ring_ib_execute, + .irq_set = &rs600_irq_set, + .irq_process = &rs600_irq_process, + .get_vblank_counter = &rs600_get_vblank_counter, + .fence_ring_emit = &r300_fence_ring_emit, + .cs_parse = &r300_cs_parse, + .copy_blit = &r100_copy_blit, + .copy_dma = &r200_copy_dma, + .copy = &r100_copy_blit, + .get_engine_clock = &radeon_atom_get_engine_clock, + .set_engine_clock = &radeon_atom_set_engine_clock, + .get_memory_clock = &radeon_atom_get_memory_clock, + .set_memory_clock = &radeon_atom_set_memory_clock, + .get_pcie_lanes = &rv370_get_pcie_lanes, + .set_pcie_lanes = &rv370_set_pcie_lanes, + .set_clock_gating = &radeon_atom_set_clock_gating, + .set_surface_reg = r100_set_surface_reg, + .clear_surface_reg = r100_clear_surface_reg, + .bandwidth_update = &rv515_bandwidth_update, + .hpd_init = &rs600_hpd_init, + .hpd_fini = &rs600_hpd_fini, + .hpd_sense = &rs600_hpd_sense, + .hpd_set_polarity = &rs600_hpd_set_polarity, + .ioctl_wait_idle = NULL, + .gui_idle = &r100_gui_idle, + .pm_misc = &rs600_pm_misc, + .pm_prepare = &rs600_pm_prepare, + .pm_finish = &rs600_pm_finish, + .pm_init_profile = &r420_pm_init_profile, + .pm_get_dynpm_state = &r100_pm_get_dynpm_state, + .pre_page_flip = &rs600_pre_page_flip, + .page_flip = &rs600_page_flip, + .post_page_flip = &rs600_post_page_flip, +}; + +static struct radeon_asic r520_asic = { + .init = &r520_init, + .fini = &rv515_fini, + .suspend = &rv515_suspend, + .resume = &r520_resume, + .vga_set_state = &r100_vga_set_state, + .gpu_is_lockup = &r300_gpu_is_lockup, + .asic_reset = &rs600_asic_reset, + .gart_tlb_flush = &rv370_pcie_gart_tlb_flush, + .gart_set_page = &rv370_pcie_gart_set_page, + .cp_commit = &r100_cp_commit, + .ring_start = &rv515_ring_start, + .ring_test = &r100_ring_test, + .ring_ib_execute = &r100_ring_ib_execute, + .irq_set = &rs600_irq_set, + .irq_process = &rs600_irq_process, + .get_vblank_counter = &rs600_get_vblank_counter, + .fence_ring_emit = &r300_fence_ring_emit, + .cs_parse = &r300_cs_parse, + .copy_blit = &r100_copy_blit, + .copy_dma = &r200_copy_dma, + .copy = &r100_copy_blit, + .get_engine_clock = &radeon_atom_get_engine_clock, + .set_engine_clock = &radeon_atom_set_engine_clock, + .get_memory_clock = &radeon_atom_get_memory_clock, + .set_memory_clock = &radeon_atom_set_memory_clock, + .get_pcie_lanes = &rv370_get_pcie_lanes, + .set_pcie_lanes = &rv370_set_pcie_lanes, + .set_clock_gating = &radeon_atom_set_clock_gating, + .set_surface_reg = r100_set_surface_reg, + .clear_surface_reg = r100_clear_surface_reg, + .bandwidth_update = &rv515_bandwidth_update, + .hpd_init = &rs600_hpd_init, + .hpd_fini = &rs600_hpd_fini, + .hpd_sense = &rs600_hpd_sense, + .hpd_set_polarity = &rs600_hpd_set_polarity, + .ioctl_wait_idle = NULL, + .gui_idle = &r100_gui_idle, + .pm_misc = &rs600_pm_misc, + .pm_prepare = &rs600_pm_prepare, + .pm_finish = &rs600_pm_finish, + .pm_init_profile = &r420_pm_init_profile, + .pm_get_dynpm_state = &r100_pm_get_dynpm_state, + .pre_page_flip = &rs600_pre_page_flip, + .page_flip = &rs600_page_flip, + .post_page_flip = &rs600_post_page_flip, +}; + +static struct radeon_asic r600_asic = { + .init = &r600_init, + .fini = &r600_fini, + .suspend = &r600_suspend, + .resume = &r600_resume, + .cp_commit = &r600_cp_commit, + .vga_set_state = &r600_vga_set_state, + .gpu_is_lockup = &r600_gpu_is_lockup, + .asic_reset = &r600_asic_reset, + .gart_tlb_flush = &r600_pcie_gart_tlb_flush, + .gart_set_page = &rs600_gart_set_page, + .ring_test = &r600_ring_test, + .ring_ib_execute = &r600_ring_ib_execute, + .irq_set = &r600_irq_set, + .irq_process = &r600_irq_process, + .get_vblank_counter = &rs600_get_vblank_counter, + .fence_ring_emit = &r600_fence_ring_emit, + .cs_parse = &r600_cs_parse, + .copy_blit = &r600_copy_blit, + .copy_dma = &r600_copy_blit, + .copy = &r600_copy_blit, + .get_engine_clock = &radeon_atom_get_engine_clock, + .set_engine_clock = &radeon_atom_set_engine_clock, + .get_memory_clock = &radeon_atom_get_memory_clock, + .set_memory_clock = &radeon_atom_set_memory_clock, + .get_pcie_lanes = &r600_get_pcie_lanes, + .set_pcie_lanes = &r600_set_pcie_lanes, + .set_clock_gating = NULL, + .set_surface_reg = r600_set_surface_reg, + .clear_surface_reg = r600_clear_surface_reg, + .bandwidth_update = &rv515_bandwidth_update, + .hpd_init = &r600_hpd_init, + .hpd_fini = &r600_hpd_fini, + .hpd_sense = &r600_hpd_sense, + .hpd_set_polarity = &r600_hpd_set_polarity, + .ioctl_wait_idle = r600_ioctl_wait_idle, + .gui_idle = &r600_gui_idle, + .pm_misc = &r600_pm_misc, + .pm_prepare = &rs600_pm_prepare, + .pm_finish = &rs600_pm_finish, + .pm_init_profile = &r600_pm_init_profile, + .pm_get_dynpm_state = &r600_pm_get_dynpm_state, + .pre_page_flip = &rs600_pre_page_flip, + .page_flip = &rs600_page_flip, + .post_page_flip = &rs600_post_page_flip, +}; + +static struct radeon_asic rs780_asic = { + .init = &r600_init, + .fini = &r600_fini, + .suspend = &r600_suspend, + .resume = &r600_resume, + .cp_commit = &r600_cp_commit, + .gpu_is_lockup = &r600_gpu_is_lockup, + .vga_set_state = &r600_vga_set_state, + .asic_reset = &r600_asic_reset, + .gart_tlb_flush = &r600_pcie_gart_tlb_flush, + .gart_set_page = &rs600_gart_set_page, + .ring_test = &r600_ring_test, + .ring_ib_execute = &r600_ring_ib_execute, + .irq_set = &r600_irq_set, + .irq_process = &r600_irq_process, + .get_vblank_counter = &rs600_get_vblank_counter, + .fence_ring_emit = &r600_fence_ring_emit, + .cs_parse = &r600_cs_parse, + .copy_blit = &r600_copy_blit, + .copy_dma = &r600_copy_blit, + .copy = &r600_copy_blit, + .get_engine_clock = &radeon_atom_get_engine_clock, + .set_engine_clock = &radeon_atom_set_engine_clock, + .get_memory_clock = NULL, + .set_memory_clock = NULL, + .get_pcie_lanes = NULL, + .set_pcie_lanes = NULL, + .set_clock_gating = NULL, + .set_surface_reg = r600_set_surface_reg, + .clear_surface_reg = r600_clear_surface_reg, + .bandwidth_update = &rs690_bandwidth_update, + .hpd_init = &r600_hpd_init, + .hpd_fini = &r600_hpd_fini, + .hpd_sense = &r600_hpd_sense, + .hpd_set_polarity = &r600_hpd_set_polarity, + .ioctl_wait_idle = r600_ioctl_wait_idle, + .gui_idle = &r600_gui_idle, + .pm_misc = &r600_pm_misc, + .pm_prepare = &rs600_pm_prepare, + .pm_finish = &rs600_pm_finish, + .pm_init_profile = &rs780_pm_init_profile, + .pm_get_dynpm_state = &r600_pm_get_dynpm_state, + .pre_page_flip = &rs600_pre_page_flip, + .page_flip = &rs600_page_flip, + .post_page_flip = &rs600_post_page_flip, +}; + +static struct radeon_asic rv770_asic = { + .init = &rv770_init, + .fini = &rv770_fini, + .suspend = &rv770_suspend, + .resume = &rv770_resume, + .cp_commit = &r600_cp_commit, + .asic_reset = &r600_asic_reset, + .gpu_is_lockup = &r600_gpu_is_lockup, + .vga_set_state = &r600_vga_set_state, + .gart_tlb_flush = &r600_pcie_gart_tlb_flush, + .gart_set_page = &rs600_gart_set_page, + .ring_test = &r600_ring_test, + .ring_ib_execute = &r600_ring_ib_execute, + .irq_set = &r600_irq_set, + .irq_process = &r600_irq_process, + .get_vblank_counter = &rs600_get_vblank_counter, + .fence_ring_emit = &r600_fence_ring_emit, + .cs_parse = &r600_cs_parse, + .copy_blit = &r600_copy_blit, + .copy_dma = &r600_copy_blit, + .copy = &r600_copy_blit, + .get_engine_clock = &radeon_atom_get_engine_clock, + .set_engine_clock = &radeon_atom_set_engine_clock, + .get_memory_clock = &radeon_atom_get_memory_clock, + .set_memory_clock = &radeon_atom_set_memory_clock, + .get_pcie_lanes = &r600_get_pcie_lanes, + .set_pcie_lanes = &r600_set_pcie_lanes, + .set_clock_gating = &radeon_atom_set_clock_gating, + .set_surface_reg = r600_set_surface_reg, + .clear_surface_reg = r600_clear_surface_reg, + .bandwidth_update = &rv515_bandwidth_update, + .hpd_init = &r600_hpd_init, + .hpd_fini = &r600_hpd_fini, + .hpd_sense = &r600_hpd_sense, + .hpd_set_polarity = &r600_hpd_set_polarity, + .ioctl_wait_idle = r600_ioctl_wait_idle, + .gui_idle = &r600_gui_idle, + .pm_misc = &rv770_pm_misc, + .pm_prepare = &rs600_pm_prepare, + .pm_finish = &rs600_pm_finish, + .pm_init_profile = &r600_pm_init_profile, + .pm_get_dynpm_state = &r600_pm_get_dynpm_state, + .pre_page_flip = &rs600_pre_page_flip, + .page_flip = &rv770_page_flip, + .post_page_flip = &rs600_post_page_flip, +}; + +static struct radeon_asic evergreen_asic = { + .init = &evergreen_init, + .fini = &evergreen_fini, + .suspend = &evergreen_suspend, + .resume = &evergreen_resume, + .cp_commit = &r600_cp_commit, + .gpu_is_lockup = &evergreen_gpu_is_lockup, + .asic_reset = &evergreen_asic_reset, + .vga_set_state = &r600_vga_set_state, + .gart_tlb_flush = &evergreen_pcie_gart_tlb_flush, + .gart_set_page = &rs600_gart_set_page, + .ring_test = &r600_ring_test, + .ring_ib_execute = &evergreen_ring_ib_execute, + .irq_set = &evergreen_irq_set, + .irq_process = &evergreen_irq_process, + .get_vblank_counter = &evergreen_get_vblank_counter, + .fence_ring_emit = &r600_fence_ring_emit, + .cs_parse = &evergreen_cs_parse, + .copy_blit = &evergreen_copy_blit, + .copy_dma = &evergreen_copy_blit, + .copy = &evergreen_copy_blit, + .get_engine_clock = &radeon_atom_get_engine_clock, + .set_engine_clock = &radeon_atom_set_engine_clock, + .get_memory_clock = &radeon_atom_get_memory_clock, + .set_memory_clock = &radeon_atom_set_memory_clock, + .get_pcie_lanes = &r600_get_pcie_lanes, + .set_pcie_lanes = &r600_set_pcie_lanes, + .set_clock_gating = NULL, + .set_surface_reg = r600_set_surface_reg, + .clear_surface_reg = r600_clear_surface_reg, + .bandwidth_update = &evergreen_bandwidth_update, + .hpd_init = &evergreen_hpd_init, + .hpd_fini = &evergreen_hpd_fini, + .hpd_sense = &evergreen_hpd_sense, + .hpd_set_polarity = &evergreen_hpd_set_polarity, + .ioctl_wait_idle = r600_ioctl_wait_idle, + .gui_idle = &r600_gui_idle, + .pm_misc = &evergreen_pm_misc, + .pm_prepare = &evergreen_pm_prepare, + .pm_finish = &evergreen_pm_finish, + .pm_init_profile = &r600_pm_init_profile, + .pm_get_dynpm_state = &r600_pm_get_dynpm_state, + .pre_page_flip = &evergreen_pre_page_flip, + .page_flip = &evergreen_page_flip, + .post_page_flip = &evergreen_post_page_flip, +}; + +static struct radeon_asic sumo_asic = { + .init = &evergreen_init, + .fini = &evergreen_fini, + .suspend = &evergreen_suspend, + .resume = &evergreen_resume, + .cp_commit = &r600_cp_commit, + .gpu_is_lockup = &evergreen_gpu_is_lockup, + .asic_reset = &evergreen_asic_reset, + .vga_set_state = &r600_vga_set_state, + .gart_tlb_flush = &evergreen_pcie_gart_tlb_flush, + .gart_set_page = &rs600_gart_set_page, + .ring_test = &r600_ring_test, + .ring_ib_execute = &evergreen_ring_ib_execute, + .irq_set = &evergreen_irq_set, + .irq_process = &evergreen_irq_process, + .get_vblank_counter = &evergreen_get_vblank_counter, + .fence_ring_emit = &r600_fence_ring_emit, + .cs_parse = &evergreen_cs_parse, + .copy_blit = &evergreen_copy_blit, + .copy_dma = &evergreen_copy_blit, + .copy = &evergreen_copy_blit, + .get_engine_clock = &radeon_atom_get_engine_clock, + .set_engine_clock = &radeon_atom_set_engine_clock, + .get_memory_clock = NULL, + .set_memory_clock = NULL, + .get_pcie_lanes = NULL, + .set_pcie_lanes = NULL, + .set_clock_gating = NULL, + .set_surface_reg = r600_set_surface_reg, + .clear_surface_reg = r600_clear_surface_reg, + .bandwidth_update = &evergreen_bandwidth_update, + .hpd_init = &evergreen_hpd_init, + .hpd_fini = &evergreen_hpd_fini, + .hpd_sense = &evergreen_hpd_sense, + .hpd_set_polarity = &evergreen_hpd_set_polarity, + .ioctl_wait_idle = r600_ioctl_wait_idle, + .gui_idle = &r600_gui_idle, + .pm_misc = &evergreen_pm_misc, + .pm_prepare = &evergreen_pm_prepare, + .pm_finish = &evergreen_pm_finish, + .pm_init_profile = &rs780_pm_init_profile, + .pm_get_dynpm_state = &r600_pm_get_dynpm_state, + .pre_page_flip = &evergreen_pre_page_flip, + .page_flip = &evergreen_page_flip, + .post_page_flip = &evergreen_post_page_flip, +}; + +static struct radeon_asic btc_asic = { + .init = &evergreen_init, + .fini = &evergreen_fini, + .suspend = &evergreen_suspend, + .resume = &evergreen_resume, + .cp_commit = &r600_cp_commit, + .gpu_is_lockup = &evergreen_gpu_is_lockup, + .asic_reset = &evergreen_asic_reset, + .vga_set_state = &r600_vga_set_state, + .gart_tlb_flush = &evergreen_pcie_gart_tlb_flush, + .gart_set_page = &rs600_gart_set_page, + .ring_test = &r600_ring_test, + .ring_ib_execute = &evergreen_ring_ib_execute, + .irq_set = &evergreen_irq_set, + .irq_process = &evergreen_irq_process, + .get_vblank_counter = &evergreen_get_vblank_counter, + .fence_ring_emit = &r600_fence_ring_emit, + .cs_parse = &evergreen_cs_parse, + .copy_blit = &evergreen_copy_blit, + .copy_dma = &evergreen_copy_blit, + .copy = &evergreen_copy_blit, + .get_engine_clock = &radeon_atom_get_engine_clock, + .set_engine_clock = &radeon_atom_set_engine_clock, + .get_memory_clock = &radeon_atom_get_memory_clock, + .set_memory_clock = &radeon_atom_set_memory_clock, + .get_pcie_lanes = NULL, + .set_pcie_lanes = NULL, + .set_clock_gating = NULL, + .set_surface_reg = r600_set_surface_reg, + .clear_surface_reg = r600_clear_surface_reg, + .bandwidth_update = &evergreen_bandwidth_update, + .hpd_init = &evergreen_hpd_init, + .hpd_fini = &evergreen_hpd_fini, + .hpd_sense = &evergreen_hpd_sense, + .hpd_set_polarity = &evergreen_hpd_set_polarity, + .ioctl_wait_idle = r600_ioctl_wait_idle, + .gui_idle = &r600_gui_idle, + .pm_misc = &evergreen_pm_misc, + .pm_prepare = &evergreen_pm_prepare, + .pm_finish = &evergreen_pm_finish, + .pm_init_profile = &r600_pm_init_profile, + .pm_get_dynpm_state = &r600_pm_get_dynpm_state, + .pre_page_flip = &evergreen_pre_page_flip, + .page_flip = &evergreen_page_flip, + .post_page_flip = &evergreen_post_page_flip, +}; + +static struct radeon_asic cayman_asic = { + .init = &cayman_init, + .fini = &cayman_fini, + .suspend = &cayman_suspend, + .resume = &cayman_resume, + .cp_commit = &r600_cp_commit, + .gpu_is_lockup = &cayman_gpu_is_lockup, + .asic_reset = &cayman_asic_reset, + .vga_set_state = &r600_vga_set_state, + .gart_tlb_flush = &cayman_pcie_gart_tlb_flush, + .gart_set_page = &rs600_gart_set_page, + .ring_test = &r600_ring_test, + .ring_ib_execute = &evergreen_ring_ib_execute, + .irq_set = &evergreen_irq_set, + .irq_process = &evergreen_irq_process, + .get_vblank_counter = &evergreen_get_vblank_counter, + .fence_ring_emit = &r600_fence_ring_emit, + .cs_parse = &evergreen_cs_parse, + .copy_blit = &evergreen_copy_blit, + .copy_dma = &evergreen_copy_blit, + .copy = &evergreen_copy_blit, + .get_engine_clock = &radeon_atom_get_engine_clock, + .set_engine_clock = &radeon_atom_set_engine_clock, + .get_memory_clock = &radeon_atom_get_memory_clock, + .set_memory_clock = &radeon_atom_set_memory_clock, + .get_pcie_lanes = NULL, + .set_pcie_lanes = NULL, + .set_clock_gating = NULL, + .set_surface_reg = r600_set_surface_reg, + .clear_surface_reg = r600_clear_surface_reg, + .bandwidth_update = &evergreen_bandwidth_update, + .hpd_init = &evergreen_hpd_init, + .hpd_fini = &evergreen_hpd_fini, + .hpd_sense = &evergreen_hpd_sense, + .hpd_set_polarity = &evergreen_hpd_set_polarity, + .ioctl_wait_idle = r600_ioctl_wait_idle, + .gui_idle = &r600_gui_idle, + .pm_misc = &evergreen_pm_misc, + .pm_prepare = &evergreen_pm_prepare, + .pm_finish = &evergreen_pm_finish, + .pm_init_profile = &r600_pm_init_profile, + .pm_get_dynpm_state = &r600_pm_get_dynpm_state, + .pre_page_flip = &evergreen_pre_page_flip, + .page_flip = &evergreen_page_flip, + .post_page_flip = &evergreen_post_page_flip, +}; + +int radeon_asic_init(struct radeon_device *rdev) +{ + radeon_register_accessor_init(rdev); + + /* set the number of crtcs */ + if (rdev->flags & RADEON_SINGLE_CRTC) + rdev->num_crtc = 1; + else + rdev->num_crtc = 2; + + switch (rdev->family) { + case CHIP_R100: + case CHIP_RV100: + case CHIP_RS100: + case CHIP_RV200: + case CHIP_RS200: + rdev->asic = &r100_asic; + break; + case CHIP_R200: + case CHIP_RV250: + case CHIP_RS300: + case CHIP_RV280: + rdev->asic = &r200_asic; + break; + case CHIP_R300: + case CHIP_R350: + case CHIP_RV350: + case CHIP_RV380: + if (rdev->flags & RADEON_IS_PCIE) + rdev->asic = &r300_asic_pcie; + else + rdev->asic = &r300_asic; + break; + case CHIP_R420: + case CHIP_R423: + case CHIP_RV410: + rdev->asic = &r420_asic; + /* handle macs */ + if (rdev->bios == NULL) { + rdev->asic->get_engine_clock = &radeon_legacy_get_engine_clock; + rdev->asic->set_engine_clock = &radeon_legacy_set_engine_clock; + rdev->asic->get_memory_clock = &radeon_legacy_get_memory_clock; + rdev->asic->set_memory_clock = NULL; + } + break; + case CHIP_RS400: + case CHIP_RS480: + rdev->asic = &rs400_asic; + break; + case CHIP_RS600: + rdev->asic = &rs600_asic; + break; + case CHIP_RS690: + case CHIP_RS740: + rdev->asic = &rs690_asic; + break; + case CHIP_RV515: + rdev->asic = &rv515_asic; + break; + case CHIP_R520: + case CHIP_RV530: + case CHIP_RV560: + case CHIP_RV570: + case CHIP_R580: + rdev->asic = &r520_asic; + break; + case CHIP_R600: + case CHIP_RV610: + case CHIP_RV630: + case CHIP_RV620: + case CHIP_RV635: + case CHIP_RV670: + rdev->asic = &r600_asic; + break; + case CHIP_RS780: + case CHIP_RS880: + rdev->asic = &rs780_asic; + break; + case CHIP_RV770: + case CHIP_RV730: + case CHIP_RV710: + case CHIP_RV740: + rdev->asic = &rv770_asic; + break; + case CHIP_CEDAR: + case CHIP_REDWOOD: + case CHIP_JUNIPER: + case CHIP_CYPRESS: + case CHIP_HEMLOCK: + /* set num crtcs */ + if (rdev->family == CHIP_CEDAR) + rdev->num_crtc = 4; + else + rdev->num_crtc = 6; + rdev->asic = &evergreen_asic; + break; + case CHIP_PALM: + case CHIP_SUMO: + case CHIP_SUMO2: + rdev->asic = &sumo_asic; + break; + case CHIP_BARTS: + case CHIP_TURKS: + case CHIP_CAICOS: + /* set num crtcs */ + if (rdev->family == CHIP_CAICOS) + rdev->num_crtc = 4; + else + rdev->num_crtc = 6; + rdev->asic = &btc_asic; + break; + case CHIP_CAYMAN: + rdev->asic = &cayman_asic; + /* set num crtcs */ + rdev->num_crtc = 6; + break; + default: + /* FIXME: not supported yet */ + return -EINVAL; + } + + if (rdev->flags & RADEON_IS_IGP) { + rdev->asic->get_memory_clock = NULL; + rdev->asic->set_memory_clock = NULL; + } + + return 0; +} + diff --git a/drivers/gpu/drm/radeon/radeon_asic.h b/drivers/gpu/drm/radeon/radeon_asic.h new file mode 100644 index 00000000..3dedaa07 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_asic.h @@ -0,0 +1,443 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#ifndef __RADEON_ASIC_H__ +#define __RADEON_ASIC_H__ + +/* + * common functions + */ +uint32_t radeon_legacy_get_engine_clock(struct radeon_device *rdev); +void radeon_legacy_set_engine_clock(struct radeon_device *rdev, uint32_t eng_clock); +uint32_t radeon_legacy_get_memory_clock(struct radeon_device *rdev); +void radeon_legacy_set_clock_gating(struct radeon_device *rdev, int enable); + +uint32_t radeon_atom_get_engine_clock(struct radeon_device *rdev); +void radeon_atom_set_engine_clock(struct radeon_device *rdev, uint32_t eng_clock); +uint32_t radeon_atom_get_memory_clock(struct radeon_device *rdev); +void radeon_atom_set_memory_clock(struct radeon_device *rdev, uint32_t mem_clock); +void radeon_atom_set_clock_gating(struct radeon_device *rdev, int enable); + +/* + * r100,rv100,rs100,rv200,rs200 + */ +struct r100_mc_save { + u32 GENMO_WT; + u32 CRTC_EXT_CNTL; + u32 CRTC_GEN_CNTL; + u32 CRTC2_GEN_CNTL; + u32 CUR_OFFSET; + u32 CUR2_OFFSET; +}; +int r100_init(struct radeon_device *rdev); +void r100_fini(struct radeon_device *rdev); +int r100_suspend(struct radeon_device *rdev); +int r100_resume(struct radeon_device *rdev); +void r100_vga_set_state(struct radeon_device *rdev, bool state); +bool r100_gpu_is_lockup(struct radeon_device *rdev); +int r100_asic_reset(struct radeon_device *rdev); +u32 r100_get_vblank_counter(struct radeon_device *rdev, int crtc); +void r100_pci_gart_tlb_flush(struct radeon_device *rdev); +int r100_pci_gart_set_page(struct radeon_device *rdev, int i, uint64_t addr); +void r100_cp_commit(struct radeon_device *rdev); +void r100_ring_start(struct radeon_device *rdev); +int r100_irq_set(struct radeon_device *rdev); +int r100_irq_process(struct radeon_device *rdev); +void r100_fence_ring_emit(struct radeon_device *rdev, + struct radeon_fence *fence); +int r100_cs_parse(struct radeon_cs_parser *p); +void r100_pll_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v); +uint32_t r100_pll_rreg(struct radeon_device *rdev, uint32_t reg); +int r100_copy_blit(struct radeon_device *rdev, + uint64_t src_offset, + uint64_t dst_offset, + unsigned num_gpu_pages, + struct radeon_fence *fence); +int r100_set_surface_reg(struct radeon_device *rdev, int reg, + uint32_t tiling_flags, uint32_t pitch, + uint32_t offset, uint32_t obj_size); +void r100_clear_surface_reg(struct radeon_device *rdev, int reg); +void r100_bandwidth_update(struct radeon_device *rdev); +void r100_ring_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib); +int r100_ring_test(struct radeon_device *rdev); +void r100_hpd_init(struct radeon_device *rdev); +void r100_hpd_fini(struct radeon_device *rdev); +bool r100_hpd_sense(struct radeon_device *rdev, enum radeon_hpd_id hpd); +void r100_hpd_set_polarity(struct radeon_device *rdev, + enum radeon_hpd_id hpd); +int r100_debugfs_rbbm_init(struct radeon_device *rdev); +int r100_debugfs_cp_init(struct radeon_device *rdev); +void r100_cp_disable(struct radeon_device *rdev); +int r100_cp_init(struct radeon_device *rdev, unsigned ring_size); +void r100_cp_fini(struct radeon_device *rdev); +int r100_pci_gart_init(struct radeon_device *rdev); +void r100_pci_gart_fini(struct radeon_device *rdev); +int r100_pci_gart_enable(struct radeon_device *rdev); +void r100_pci_gart_disable(struct radeon_device *rdev); +int r100_debugfs_mc_info_init(struct radeon_device *rdev); +int r100_gui_wait_for_idle(struct radeon_device *rdev); +void r100_gpu_lockup_update(struct r100_gpu_lockup *lockup, + struct radeon_cp *cp); +bool r100_gpu_cp_is_lockup(struct radeon_device *rdev, + struct r100_gpu_lockup *lockup, + struct radeon_cp *cp); +void r100_ib_fini(struct radeon_device *rdev); +int r100_ib_init(struct radeon_device *rdev); +void r100_irq_disable(struct radeon_device *rdev); +void r100_mc_stop(struct radeon_device *rdev, struct r100_mc_save *save); +void r100_mc_resume(struct radeon_device *rdev, struct r100_mc_save *save); +void r100_vram_init_sizes(struct radeon_device *rdev); +int r100_cp_reset(struct radeon_device *rdev); +void r100_vga_render_disable(struct radeon_device *rdev); +void r100_restore_sanity(struct radeon_device *rdev); +int r100_cs_track_check_pkt3_indx_buffer(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt, + struct radeon_bo *robj); +int r100_cs_parse_packet0(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt, + const unsigned *auth, unsigned n, + radeon_packet0_check_t check); +int r100_cs_packet_parse(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt, + unsigned idx); +void r100_enable_bm(struct radeon_device *rdev); +void r100_set_common_regs(struct radeon_device *rdev); +void r100_bm_disable(struct radeon_device *rdev); +extern bool r100_gui_idle(struct radeon_device *rdev); +extern void r100_pm_misc(struct radeon_device *rdev); +extern void r100_pm_prepare(struct radeon_device *rdev); +extern void r100_pm_finish(struct radeon_device *rdev); +extern void r100_pm_init_profile(struct radeon_device *rdev); +extern void r100_pm_get_dynpm_state(struct radeon_device *rdev); +extern void r100_pre_page_flip(struct radeon_device *rdev, int crtc); +extern u32 r100_page_flip(struct radeon_device *rdev, int crtc, u64 crtc_base); +extern void r100_post_page_flip(struct radeon_device *rdev, int crtc); + +/* + * r200,rv250,rs300,rv280 + */ +extern int r200_copy_dma(struct radeon_device *rdev, + uint64_t src_offset, + uint64_t dst_offset, + unsigned num_gpu_pages, + struct radeon_fence *fence); +void r200_set_safe_registers(struct radeon_device *rdev); + +/* + * r300,r350,rv350,rv380 + */ +extern int r300_init(struct radeon_device *rdev); +extern void r300_fini(struct radeon_device *rdev); +extern int r300_suspend(struct radeon_device *rdev); +extern int r300_resume(struct radeon_device *rdev); +extern bool r300_gpu_is_lockup(struct radeon_device *rdev); +extern int r300_asic_reset(struct radeon_device *rdev); +extern void r300_ring_start(struct radeon_device *rdev); +extern void r300_fence_ring_emit(struct radeon_device *rdev, + struct radeon_fence *fence); +extern int r300_cs_parse(struct radeon_cs_parser *p); +extern void rv370_pcie_gart_tlb_flush(struct radeon_device *rdev); +extern int rv370_pcie_gart_set_page(struct radeon_device *rdev, int i, uint64_t addr); +extern void rv370_set_pcie_lanes(struct radeon_device *rdev, int lanes); +extern int rv370_get_pcie_lanes(struct radeon_device *rdev); +extern void r300_set_reg_safe(struct radeon_device *rdev); +extern void r300_mc_program(struct radeon_device *rdev); +extern void r300_mc_init(struct radeon_device *rdev); +extern void r300_clock_startup(struct radeon_device *rdev); +extern int r300_mc_wait_for_idle(struct radeon_device *rdev); +extern int rv370_pcie_gart_init(struct radeon_device *rdev); +extern void rv370_pcie_gart_fini(struct radeon_device *rdev); +extern int rv370_pcie_gart_enable(struct radeon_device *rdev); +extern void rv370_pcie_gart_disable(struct radeon_device *rdev); + +/* + * r420,r423,rv410 + */ +extern int r420_init(struct radeon_device *rdev); +extern void r420_fini(struct radeon_device *rdev); +extern int r420_suspend(struct radeon_device *rdev); +extern int r420_resume(struct radeon_device *rdev); +extern void r420_pm_init_profile(struct radeon_device *rdev); +extern u32 r420_mc_rreg(struct radeon_device *rdev, u32 reg); +extern void r420_mc_wreg(struct radeon_device *rdev, u32 reg, u32 v); +extern int r420_debugfs_pipes_info_init(struct radeon_device *rdev); +extern void r420_pipes_init(struct radeon_device *rdev); + +/* + * rs400,rs480 + */ +extern int rs400_init(struct radeon_device *rdev); +extern void rs400_fini(struct radeon_device *rdev); +extern int rs400_suspend(struct radeon_device *rdev); +extern int rs400_resume(struct radeon_device *rdev); +void rs400_gart_tlb_flush(struct radeon_device *rdev); +int rs400_gart_set_page(struct radeon_device *rdev, int i, uint64_t addr); +uint32_t rs400_mc_rreg(struct radeon_device *rdev, uint32_t reg); +void rs400_mc_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v); +int rs400_gart_init(struct radeon_device *rdev); +int rs400_gart_enable(struct radeon_device *rdev); +void rs400_gart_adjust_size(struct radeon_device *rdev); +void rs400_gart_disable(struct radeon_device *rdev); +void rs400_gart_fini(struct radeon_device *rdev); + +/* + * rs600. + */ +extern int rs600_asic_reset(struct radeon_device *rdev); +extern int rs600_init(struct radeon_device *rdev); +extern void rs600_fini(struct radeon_device *rdev); +extern int rs600_suspend(struct radeon_device *rdev); +extern int rs600_resume(struct radeon_device *rdev); +int rs600_irq_set(struct radeon_device *rdev); +int rs600_irq_process(struct radeon_device *rdev); +void rs600_irq_disable(struct radeon_device *rdev); +u32 rs600_get_vblank_counter(struct radeon_device *rdev, int crtc); +void rs600_gart_tlb_flush(struct radeon_device *rdev); +int rs600_gart_set_page(struct radeon_device *rdev, int i, uint64_t addr); +uint32_t rs600_mc_rreg(struct radeon_device *rdev, uint32_t reg); +void rs600_mc_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v); +void rs600_bandwidth_update(struct radeon_device *rdev); +void rs600_hpd_init(struct radeon_device *rdev); +void rs600_hpd_fini(struct radeon_device *rdev); +bool rs600_hpd_sense(struct radeon_device *rdev, enum radeon_hpd_id hpd); +void rs600_hpd_set_polarity(struct radeon_device *rdev, + enum radeon_hpd_id hpd); +extern void rs600_pm_misc(struct radeon_device *rdev); +extern void rs600_pm_prepare(struct radeon_device *rdev); +extern void rs600_pm_finish(struct radeon_device *rdev); +extern void rs600_pre_page_flip(struct radeon_device *rdev, int crtc); +extern u32 rs600_page_flip(struct radeon_device *rdev, int crtc, u64 crtc_base); +extern void rs600_post_page_flip(struct radeon_device *rdev, int crtc); +void rs600_set_safe_registers(struct radeon_device *rdev); + + +/* + * rs690,rs740 + */ +int rs690_init(struct radeon_device *rdev); +void rs690_fini(struct radeon_device *rdev); +int rs690_resume(struct radeon_device *rdev); +int rs690_suspend(struct radeon_device *rdev); +uint32_t rs690_mc_rreg(struct radeon_device *rdev, uint32_t reg); +void rs690_mc_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v); +void rs690_bandwidth_update(struct radeon_device *rdev); +void rs690_line_buffer_adjust(struct radeon_device *rdev, + struct drm_display_mode *mode1, + struct drm_display_mode *mode2); + +/* + * rv515 + */ +struct rv515_mc_save { + u32 d1vga_control; + u32 d2vga_control; + u32 vga_render_control; + u32 vga_hdp_control; + u32 d1crtc_control; + u32 d2crtc_control; +}; +int rv515_init(struct radeon_device *rdev); +void rv515_fini(struct radeon_device *rdev); +uint32_t rv515_mc_rreg(struct radeon_device *rdev, uint32_t reg); +void rv515_mc_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v); +void rv515_ring_start(struct radeon_device *rdev); +void rv515_bandwidth_update(struct radeon_device *rdev); +int rv515_resume(struct radeon_device *rdev); +int rv515_suspend(struct radeon_device *rdev); +void rv515_bandwidth_avivo_update(struct radeon_device *rdev); +void rv515_vga_render_disable(struct radeon_device *rdev); +void rv515_set_safe_registers(struct radeon_device *rdev); +void rv515_mc_stop(struct radeon_device *rdev, struct rv515_mc_save *save); +void rv515_mc_resume(struct radeon_device *rdev, struct rv515_mc_save *save); +void rv515_clock_startup(struct radeon_device *rdev); +void rv515_debugfs(struct radeon_device *rdev); + + +/* + * r520,rv530,rv560,rv570,r580 + */ +int r520_init(struct radeon_device *rdev); +int r520_resume(struct radeon_device *rdev); + +/* + * r600,rv610,rv630,rv620,rv635,rv670,rs780,rs880 + */ +int r600_init(struct radeon_device *rdev); +void r600_fini(struct radeon_device *rdev); +int r600_suspend(struct radeon_device *rdev); +int r600_resume(struct radeon_device *rdev); +void r600_vga_set_state(struct radeon_device *rdev, bool state); +int r600_wb_init(struct radeon_device *rdev); +void r600_wb_fini(struct radeon_device *rdev); +void r600_cp_commit(struct radeon_device *rdev); +void r600_pcie_gart_tlb_flush(struct radeon_device *rdev); +uint32_t r600_pciep_rreg(struct radeon_device *rdev, uint32_t reg); +void r600_pciep_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v); +int r600_cs_parse(struct radeon_cs_parser *p); +void r600_fence_ring_emit(struct radeon_device *rdev, + struct radeon_fence *fence); +bool r600_gpu_is_lockup(struct radeon_device *rdev); +int r600_asic_reset(struct radeon_device *rdev); +int r600_set_surface_reg(struct radeon_device *rdev, int reg, + uint32_t tiling_flags, uint32_t pitch, + uint32_t offset, uint32_t obj_size); +void r600_clear_surface_reg(struct radeon_device *rdev, int reg); +int r600_ib_test(struct radeon_device *rdev); +void r600_ring_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib); +int r600_ring_test(struct radeon_device *rdev); +int r600_copy_blit(struct radeon_device *rdev, + uint64_t src_offset, uint64_t dst_offset, + unsigned num_gpu_pages, struct radeon_fence *fence); +void r600_hpd_init(struct radeon_device *rdev); +void r600_hpd_fini(struct radeon_device *rdev); +bool r600_hpd_sense(struct radeon_device *rdev, enum radeon_hpd_id hpd); +void r600_hpd_set_polarity(struct radeon_device *rdev, + enum radeon_hpd_id hpd); +extern void r600_ioctl_wait_idle(struct radeon_device *rdev, struct radeon_bo *bo); +extern bool r600_gui_idle(struct radeon_device *rdev); +extern void r600_pm_misc(struct radeon_device *rdev); +extern void r600_pm_init_profile(struct radeon_device *rdev); +extern void rs780_pm_init_profile(struct radeon_device *rdev); +extern void r600_pm_get_dynpm_state(struct radeon_device *rdev); +extern void r600_set_pcie_lanes(struct radeon_device *rdev, int lanes); +extern int r600_get_pcie_lanes(struct radeon_device *rdev); +bool r600_card_posted(struct radeon_device *rdev); +void r600_cp_stop(struct radeon_device *rdev); +int r600_cp_start(struct radeon_device *rdev); +void r600_ring_init(struct radeon_device *rdev, unsigned ring_size); +int r600_cp_resume(struct radeon_device *rdev); +void r600_cp_fini(struct radeon_device *rdev); +int r600_count_pipe_bits(uint32_t val); +int r600_mc_wait_for_idle(struct radeon_device *rdev); +int r600_pcie_gart_init(struct radeon_device *rdev); +void r600_scratch_init(struct radeon_device *rdev); +int r600_blit_init(struct radeon_device *rdev); +void r600_blit_fini(struct radeon_device *rdev); +int r600_init_microcode(struct radeon_device *rdev); +/* r600 irq */ +int r600_irq_process(struct radeon_device *rdev); +int r600_irq_init(struct radeon_device *rdev); +void r600_irq_fini(struct radeon_device *rdev); +void r600_ih_ring_init(struct radeon_device *rdev, unsigned ring_size); +int r600_irq_set(struct radeon_device *rdev); +void r600_irq_suspend(struct radeon_device *rdev); +void r600_disable_interrupts(struct radeon_device *rdev); +void r600_rlc_stop(struct radeon_device *rdev); +/* r600 audio */ +int r600_audio_init(struct radeon_device *rdev); +int r600_audio_tmds_index(struct drm_encoder *encoder); +void r600_audio_set_clock(struct drm_encoder *encoder, int clock); +int r600_audio_channels(struct radeon_device *rdev); +int r600_audio_bits_per_sample(struct radeon_device *rdev); +int r600_audio_rate(struct radeon_device *rdev); +uint8_t r600_audio_status_bits(struct radeon_device *rdev); +uint8_t r600_audio_category_code(struct radeon_device *rdev); +void r600_audio_schedule_polling(struct radeon_device *rdev); +void r600_audio_enable_polling(struct drm_encoder *encoder); +void r600_audio_disable_polling(struct drm_encoder *encoder); +void r600_audio_fini(struct radeon_device *rdev); +void r600_hdmi_init(struct drm_encoder *encoder); +int r600_hdmi_buffer_status_changed(struct drm_encoder *encoder); +void r600_hdmi_update_audio_settings(struct drm_encoder *encoder); +/* r600 blit */ +int r600_blit_prepare_copy(struct radeon_device *rdev, int size_bytes); +void r600_blit_done_copy(struct radeon_device *rdev, struct radeon_fence *fence); +void r600_kms_blit_copy(struct radeon_device *rdev, + u64 src_gpu_addr, u64 dst_gpu_addr, + int size_bytes); + +/* + * rv770,rv730,rv710,rv740 + */ +int rv770_init(struct radeon_device *rdev); +void rv770_fini(struct radeon_device *rdev); +int rv770_suspend(struct radeon_device *rdev); +int rv770_resume(struct radeon_device *rdev); +void rv770_pm_misc(struct radeon_device *rdev); +u32 rv770_page_flip(struct radeon_device *rdev, int crtc, u64 crtc_base); +void r700_vram_gtt_location(struct radeon_device *rdev, struct radeon_mc *mc); +void r700_cp_stop(struct radeon_device *rdev); +void r700_cp_fini(struct radeon_device *rdev); + +/* + * evergreen + */ +struct evergreen_mc_save { + u32 vga_control[6]; + u32 vga_render_control; + u32 vga_hdp_control; + u32 crtc_control[6]; +}; +void evergreen_pcie_gart_tlb_flush(struct radeon_device *rdev); +int evergreen_init(struct radeon_device *rdev); +void evergreen_fini(struct radeon_device *rdev); +int evergreen_suspend(struct radeon_device *rdev); +int evergreen_resume(struct radeon_device *rdev); +bool evergreen_gpu_is_lockup(struct radeon_device *rdev); +int evergreen_asic_reset(struct radeon_device *rdev); +void evergreen_bandwidth_update(struct radeon_device *rdev); +void evergreen_ring_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib); +int evergreen_copy_blit(struct radeon_device *rdev, + uint64_t src_offset, uint64_t dst_offset, + unsigned num_gpu_pages, struct radeon_fence *fence); +void evergreen_hpd_init(struct radeon_device *rdev); +void evergreen_hpd_fini(struct radeon_device *rdev); +bool evergreen_hpd_sense(struct radeon_device *rdev, enum radeon_hpd_id hpd); +void evergreen_hpd_set_polarity(struct radeon_device *rdev, + enum radeon_hpd_id hpd); +u32 evergreen_get_vblank_counter(struct radeon_device *rdev, int crtc); +int evergreen_irq_set(struct radeon_device *rdev); +int evergreen_irq_process(struct radeon_device *rdev); +extern int evergreen_cs_parse(struct radeon_cs_parser *p); +extern void evergreen_pm_misc(struct radeon_device *rdev); +extern void evergreen_pm_prepare(struct radeon_device *rdev); +extern void evergreen_pm_finish(struct radeon_device *rdev); +extern void evergreen_pre_page_flip(struct radeon_device *rdev, int crtc); +extern u32 evergreen_page_flip(struct radeon_device *rdev, int crtc, u64 crtc_base); +extern void evergreen_post_page_flip(struct radeon_device *rdev, int crtc); +void evergreen_disable_interrupt_state(struct radeon_device *rdev); +int evergreen_blit_init(struct radeon_device *rdev); +void evergreen_blit_fini(struct radeon_device *rdev); +/* evergreen blit */ +int evergreen_blit_prepare_copy(struct radeon_device *rdev, int size_bytes); +void evergreen_blit_done_copy(struct radeon_device *rdev, struct radeon_fence *fence); +void evergreen_kms_blit_copy(struct radeon_device *rdev, + u64 src_gpu_addr, u64 dst_gpu_addr, + int size_bytes); + +/* + * cayman + */ +void cayman_pcie_gart_tlb_flush(struct radeon_device *rdev); +int cayman_init(struct radeon_device *rdev); +void cayman_fini(struct radeon_device *rdev); +int cayman_suspend(struct radeon_device *rdev); +int cayman_resume(struct radeon_device *rdev); +bool cayman_gpu_is_lockup(struct radeon_device *rdev); +int cayman_asic_reset(struct radeon_device *rdev); + +#endif diff --git a/drivers/gpu/drm/radeon/radeon_atombios.c b/drivers/gpu/drm/radeon/radeon_atombios.c new file mode 100644 index 00000000..ef6b426b --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_atombios.c @@ -0,0 +1,3100 @@ +/* + * Copyright 2007-8 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + */ +#include "drmP.h" +#include "radeon_drm.h" +#include "radeon.h" + +#include "atom.h" +#include "atom-bits.h" + +/* from radeon_encoder.c */ +extern uint32_t +radeon_get_encoder_enum(struct drm_device *dev, uint32_t supported_device, + uint8_t dac); +extern void radeon_link_encoder_connector(struct drm_device *dev); +extern void +radeon_add_atom_encoder(struct drm_device *dev, uint32_t encoder_enum, + uint32_t supported_device, u16 caps); + +/* from radeon_connector.c */ +extern void +radeon_add_atom_connector(struct drm_device *dev, + uint32_t connector_id, + uint32_t supported_device, + int connector_type, + struct radeon_i2c_bus_rec *i2c_bus, + uint32_t igp_lane_info, + uint16_t connector_object_id, + struct radeon_hpd *hpd, + struct radeon_router *router); + +/* from radeon_legacy_encoder.c */ +extern void +radeon_add_legacy_encoder(struct drm_device *dev, uint32_t encoder_enum, + uint32_t supported_device); + +union atom_supported_devices { + struct _ATOM_SUPPORTED_DEVICES_INFO info; + struct _ATOM_SUPPORTED_DEVICES_INFO_2 info_2; + struct _ATOM_SUPPORTED_DEVICES_INFO_2d1 info_2d1; +}; + +static inline struct radeon_i2c_bus_rec radeon_lookup_i2c_gpio(struct radeon_device *rdev, + uint8_t id) +{ + struct atom_context *ctx = rdev->mode_info.atom_context; + ATOM_GPIO_I2C_ASSIGMENT *gpio; + struct radeon_i2c_bus_rec i2c; + int index = GetIndexIntoMasterTable(DATA, GPIO_I2C_Info); + struct _ATOM_GPIO_I2C_INFO *i2c_info; + uint16_t data_offset, size; + int i, num_indices; + + memset(&i2c, 0, sizeof(struct radeon_i2c_bus_rec)); + i2c.valid = false; + + if (atom_parse_data_header(ctx, index, &size, NULL, NULL, &data_offset)) { + i2c_info = (struct _ATOM_GPIO_I2C_INFO *)(ctx->bios + data_offset); + + num_indices = (size - sizeof(ATOM_COMMON_TABLE_HEADER)) / + sizeof(ATOM_GPIO_I2C_ASSIGMENT); + + for (i = 0; i < num_indices; i++) { + gpio = &i2c_info->asGPIO_Info[i]; + + /* r4xx mask is technically not used by the hw, so patch in the legacy mask bits */ + if ((rdev->family == CHIP_R420) || + (rdev->family == CHIP_R423) || + (rdev->family == CHIP_RV410)) { + if ((le16_to_cpu(gpio->usClkMaskRegisterIndex) == 0x0018) || + (le16_to_cpu(gpio->usClkMaskRegisterIndex) == 0x0019) || + (le16_to_cpu(gpio->usClkMaskRegisterIndex) == 0x001a)) { + gpio->ucClkMaskShift = 0x19; + gpio->ucDataMaskShift = 0x18; + } + } + + /* some evergreen boards have bad data for this entry */ + if (ASIC_IS_DCE4(rdev)) { + if ((i == 7) && + (le16_to_cpu(gpio->usClkMaskRegisterIndex) == 0x1936) && + (gpio->sucI2cId.ucAccess == 0)) { + gpio->sucI2cId.ucAccess = 0x97; + gpio->ucDataMaskShift = 8; + gpio->ucDataEnShift = 8; + gpio->ucDataY_Shift = 8; + gpio->ucDataA_Shift = 8; + } + } + + /* some DCE3 boards have bad data for this entry */ + if (ASIC_IS_DCE3(rdev)) { + if ((i == 4) && + (le16_to_cpu(gpio->usClkMaskRegisterIndex) == 0x1fda) && + (gpio->sucI2cId.ucAccess == 0x94)) + gpio->sucI2cId.ucAccess = 0x14; + } + + if (gpio->sucI2cId.ucAccess == id) { + i2c.mask_clk_reg = le16_to_cpu(gpio->usClkMaskRegisterIndex) * 4; + i2c.mask_data_reg = le16_to_cpu(gpio->usDataMaskRegisterIndex) * 4; + i2c.en_clk_reg = le16_to_cpu(gpio->usClkEnRegisterIndex) * 4; + i2c.en_data_reg = le16_to_cpu(gpio->usDataEnRegisterIndex) * 4; + i2c.y_clk_reg = le16_to_cpu(gpio->usClkY_RegisterIndex) * 4; + i2c.y_data_reg = le16_to_cpu(gpio->usDataY_RegisterIndex) * 4; + i2c.a_clk_reg = le16_to_cpu(gpio->usClkA_RegisterIndex) * 4; + i2c.a_data_reg = le16_to_cpu(gpio->usDataA_RegisterIndex) * 4; + i2c.mask_clk_mask = (1 << gpio->ucClkMaskShift); + i2c.mask_data_mask = (1 << gpio->ucDataMaskShift); + i2c.en_clk_mask = (1 << gpio->ucClkEnShift); + i2c.en_data_mask = (1 << gpio->ucDataEnShift); + i2c.y_clk_mask = (1 << gpio->ucClkY_Shift); + i2c.y_data_mask = (1 << gpio->ucDataY_Shift); + i2c.a_clk_mask = (1 << gpio->ucClkA_Shift); + i2c.a_data_mask = (1 << gpio->ucDataA_Shift); + + if (gpio->sucI2cId.sbfAccess.bfHW_Capable) + i2c.hw_capable = true; + else + i2c.hw_capable = false; + + if (gpio->sucI2cId.ucAccess == 0xa0) + i2c.mm_i2c = true; + else + i2c.mm_i2c = false; + + i2c.i2c_id = gpio->sucI2cId.ucAccess; + + if (i2c.mask_clk_reg) + i2c.valid = true; + break; + } + } + } + + return i2c; +} + +void radeon_atombios_i2c_init(struct radeon_device *rdev) +{ + struct atom_context *ctx = rdev->mode_info.atom_context; + ATOM_GPIO_I2C_ASSIGMENT *gpio; + struct radeon_i2c_bus_rec i2c; + int index = GetIndexIntoMasterTable(DATA, GPIO_I2C_Info); + struct _ATOM_GPIO_I2C_INFO *i2c_info; + uint16_t data_offset, size; + int i, num_indices; + char stmp[32]; + + memset(&i2c, 0, sizeof(struct radeon_i2c_bus_rec)); + + if (atom_parse_data_header(ctx, index, &size, NULL, NULL, &data_offset)) { + i2c_info = (struct _ATOM_GPIO_I2C_INFO *)(ctx->bios + data_offset); + + num_indices = (size - sizeof(ATOM_COMMON_TABLE_HEADER)) / + sizeof(ATOM_GPIO_I2C_ASSIGMENT); + + for (i = 0; i < num_indices; i++) { + gpio = &i2c_info->asGPIO_Info[i]; + i2c.valid = false; + + /* r4xx mask is technically not used by the hw, so patch in the legacy mask bits */ + if ((rdev->family == CHIP_R420) || + (rdev->family == CHIP_R423) || + (rdev->family == CHIP_RV410)) { + if ((le16_to_cpu(gpio->usClkMaskRegisterIndex) == 0x0018) || + (le16_to_cpu(gpio->usClkMaskRegisterIndex) == 0x0019) || + (le16_to_cpu(gpio->usClkMaskRegisterIndex) == 0x001a)) { + gpio->ucClkMaskShift = 0x19; + gpio->ucDataMaskShift = 0x18; + } + } + + /* some evergreen boards have bad data for this entry */ + if (ASIC_IS_DCE4(rdev)) { + if ((i == 7) && + (le16_to_cpu(gpio->usClkMaskRegisterIndex) == 0x1936) && + (gpio->sucI2cId.ucAccess == 0)) { + gpio->sucI2cId.ucAccess = 0x97; + gpio->ucDataMaskShift = 8; + gpio->ucDataEnShift = 8; + gpio->ucDataY_Shift = 8; + gpio->ucDataA_Shift = 8; + } + } + + /* some DCE3 boards have bad data for this entry */ + if (ASIC_IS_DCE3(rdev)) { + if ((i == 4) && + (le16_to_cpu(gpio->usClkMaskRegisterIndex) == 0x1fda) && + (gpio->sucI2cId.ucAccess == 0x94)) + gpio->sucI2cId.ucAccess = 0x14; + } + + i2c.mask_clk_reg = le16_to_cpu(gpio->usClkMaskRegisterIndex) * 4; + i2c.mask_data_reg = le16_to_cpu(gpio->usDataMaskRegisterIndex) * 4; + i2c.en_clk_reg = le16_to_cpu(gpio->usClkEnRegisterIndex) * 4; + i2c.en_data_reg = le16_to_cpu(gpio->usDataEnRegisterIndex) * 4; + i2c.y_clk_reg = le16_to_cpu(gpio->usClkY_RegisterIndex) * 4; + i2c.y_data_reg = le16_to_cpu(gpio->usDataY_RegisterIndex) * 4; + i2c.a_clk_reg = le16_to_cpu(gpio->usClkA_RegisterIndex) * 4; + i2c.a_data_reg = le16_to_cpu(gpio->usDataA_RegisterIndex) * 4; + i2c.mask_clk_mask = (1 << gpio->ucClkMaskShift); + i2c.mask_data_mask = (1 << gpio->ucDataMaskShift); + i2c.en_clk_mask = (1 << gpio->ucClkEnShift); + i2c.en_data_mask = (1 << gpio->ucDataEnShift); + i2c.y_clk_mask = (1 << gpio->ucClkY_Shift); + i2c.y_data_mask = (1 << gpio->ucDataY_Shift); + i2c.a_clk_mask = (1 << gpio->ucClkA_Shift); + i2c.a_data_mask = (1 << gpio->ucDataA_Shift); + + if (gpio->sucI2cId.sbfAccess.bfHW_Capable) + i2c.hw_capable = true; + else + i2c.hw_capable = false; + + if (gpio->sucI2cId.ucAccess == 0xa0) + i2c.mm_i2c = true; + else + i2c.mm_i2c = false; + + i2c.i2c_id = gpio->sucI2cId.ucAccess; + + if (i2c.mask_clk_reg) { + i2c.valid = true; + sprintf(stmp, "0x%x", i2c.i2c_id); + rdev->i2c_bus[i] = radeon_i2c_create(rdev->ddev, &i2c, stmp); + } + } + } +} + +static inline struct radeon_gpio_rec radeon_lookup_gpio(struct radeon_device *rdev, + u8 id) +{ + struct atom_context *ctx = rdev->mode_info.atom_context; + struct radeon_gpio_rec gpio; + int index = GetIndexIntoMasterTable(DATA, GPIO_Pin_LUT); + struct _ATOM_GPIO_PIN_LUT *gpio_info; + ATOM_GPIO_PIN_ASSIGNMENT *pin; + u16 data_offset, size; + int i, num_indices; + + memset(&gpio, 0, sizeof(struct radeon_gpio_rec)); + gpio.valid = false; + + if (atom_parse_data_header(ctx, index, &size, NULL, NULL, &data_offset)) { + gpio_info = (struct _ATOM_GPIO_PIN_LUT *)(ctx->bios + data_offset); + + num_indices = (size - sizeof(ATOM_COMMON_TABLE_HEADER)) / + sizeof(ATOM_GPIO_PIN_ASSIGNMENT); + + for (i = 0; i < num_indices; i++) { + pin = &gpio_info->asGPIO_Pin[i]; + if (id == pin->ucGPIO_ID) { + gpio.id = pin->ucGPIO_ID; + gpio.reg = le16_to_cpu(pin->usGpioPin_AIndex) * 4; + gpio.mask = (1 << pin->ucGpioPinBitShift); + gpio.valid = true; + break; + } + } + } + + return gpio; +} + +static struct radeon_hpd radeon_atom_get_hpd_info_from_gpio(struct radeon_device *rdev, + struct radeon_gpio_rec *gpio) +{ + struct radeon_hpd hpd; + u32 reg; + + memset(&hpd, 0, sizeof(struct radeon_hpd)); + + if (ASIC_IS_DCE4(rdev)) + reg = EVERGREEN_DC_GPIO_HPD_A; + else + reg = AVIVO_DC_GPIO_HPD_A; + + hpd.gpio = *gpio; + if (gpio->reg == reg) { + switch(gpio->mask) { + case (1 << 0): + hpd.hpd = RADEON_HPD_1; + break; + case (1 << 8): + hpd.hpd = RADEON_HPD_2; + break; + case (1 << 16): + hpd.hpd = RADEON_HPD_3; + break; + case (1 << 24): + hpd.hpd = RADEON_HPD_4; + break; + case (1 << 26): + hpd.hpd = RADEON_HPD_5; + break; + case (1 << 28): + hpd.hpd = RADEON_HPD_6; + break; + default: + hpd.hpd = RADEON_HPD_NONE; + break; + } + } else + hpd.hpd = RADEON_HPD_NONE; + return hpd; +} + +static bool radeon_atom_apply_quirks(struct drm_device *dev, + uint32_t supported_device, + int *connector_type, + struct radeon_i2c_bus_rec *i2c_bus, + uint16_t *line_mux, + struct radeon_hpd *hpd) +{ + + /* Asus M2A-VM HDMI board lists the DVI port as HDMI */ + if ((dev->pdev->device == 0x791e) && + (dev->pdev->subsystem_vendor == 0x1043) && + (dev->pdev->subsystem_device == 0x826d)) { + if ((*connector_type == DRM_MODE_CONNECTOR_HDMIA) && + (supported_device == ATOM_DEVICE_DFP3_SUPPORT)) + *connector_type = DRM_MODE_CONNECTOR_DVID; + } + + /* Asrock RS600 board lists the DVI port as HDMI */ + if ((dev->pdev->device == 0x7941) && + (dev->pdev->subsystem_vendor == 0x1849) && + (dev->pdev->subsystem_device == 0x7941)) { + if ((*connector_type == DRM_MODE_CONNECTOR_HDMIA) && + (supported_device == ATOM_DEVICE_DFP3_SUPPORT)) + *connector_type = DRM_MODE_CONNECTOR_DVID; + } + + /* MSI K9A2GM V2/V3 board has no HDMI or DVI */ + if ((dev->pdev->device == 0x796e) && + (dev->pdev->subsystem_vendor == 0x1462) && + (dev->pdev->subsystem_device == 0x7302)) { + if ((supported_device == ATOM_DEVICE_DFP2_SUPPORT) || + (supported_device == ATOM_DEVICE_DFP3_SUPPORT)) + return false; + } + + /* a-bit f-i90hd - ciaranm on #radeonhd - this board has no DVI */ + if ((dev->pdev->device == 0x7941) && + (dev->pdev->subsystem_vendor == 0x147b) && + (dev->pdev->subsystem_device == 0x2412)) { + if (*connector_type == DRM_MODE_CONNECTOR_DVII) + return false; + } + + /* Falcon NW laptop lists vga ddc line for LVDS */ + if ((dev->pdev->device == 0x5653) && + (dev->pdev->subsystem_vendor == 0x1462) && + (dev->pdev->subsystem_device == 0x0291)) { + if (*connector_type == DRM_MODE_CONNECTOR_LVDS) { + i2c_bus->valid = false; + *line_mux = 53; + } + } + + /* HIS X1300 is DVI+VGA, not DVI+DVI */ + if ((dev->pdev->device == 0x7146) && + (dev->pdev->subsystem_vendor == 0x17af) && + (dev->pdev->subsystem_device == 0x2058)) { + if (supported_device == ATOM_DEVICE_DFP1_SUPPORT) + return false; + } + + /* Gigabyte X1300 is DVI+VGA, not DVI+DVI */ + if ((dev->pdev->device == 0x7142) && + (dev->pdev->subsystem_vendor == 0x1458) && + (dev->pdev->subsystem_device == 0x2134)) { + if (supported_device == ATOM_DEVICE_DFP1_SUPPORT) + return false; + } + + + /* Funky macbooks */ + if ((dev->pdev->device == 0x71C5) && + (dev->pdev->subsystem_vendor == 0x106b) && + (dev->pdev->subsystem_device == 0x0080)) { + if ((supported_device == ATOM_DEVICE_CRT1_SUPPORT) || + (supported_device == ATOM_DEVICE_DFP2_SUPPORT)) + return false; + if (supported_device == ATOM_DEVICE_CRT2_SUPPORT) + *line_mux = 0x90; + } + + /* mac rv630, rv730, others */ + if ((supported_device == ATOM_DEVICE_TV1_SUPPORT) && + (*connector_type == DRM_MODE_CONNECTOR_DVII)) { + *connector_type = DRM_MODE_CONNECTOR_9PinDIN; + *line_mux = CONNECTOR_7PIN_DIN_ENUM_ID1; + } + + /* ASUS HD 3600 XT board lists the DVI port as HDMI */ + if ((dev->pdev->device == 0x9598) && + (dev->pdev->subsystem_vendor == 0x1043) && + (dev->pdev->subsystem_device == 0x01da)) { + if (*connector_type == DRM_MODE_CONNECTOR_HDMIA) { + *connector_type = DRM_MODE_CONNECTOR_DVII; + } + } + + /* ASUS HD 3600 board lists the DVI port as HDMI */ + if ((dev->pdev->device == 0x9598) && + (dev->pdev->subsystem_vendor == 0x1043) && + (dev->pdev->subsystem_device == 0x01e4)) { + if (*connector_type == DRM_MODE_CONNECTOR_HDMIA) { + *connector_type = DRM_MODE_CONNECTOR_DVII; + } + } + + /* ASUS HD 3450 board lists the DVI port as HDMI */ + if ((dev->pdev->device == 0x95C5) && + (dev->pdev->subsystem_vendor == 0x1043) && + (dev->pdev->subsystem_device == 0x01e2)) { + if (*connector_type == DRM_MODE_CONNECTOR_HDMIA) { + *connector_type = DRM_MODE_CONNECTOR_DVII; + } + } + + /* some BIOSes seem to report DAC on HDMI - usually this is a board with + * HDMI + VGA reporting as HDMI + */ + if (*connector_type == DRM_MODE_CONNECTOR_HDMIA) { + if (supported_device & (ATOM_DEVICE_CRT_SUPPORT)) { + *connector_type = DRM_MODE_CONNECTOR_VGA; + *line_mux = 0; + } + } + + /* Acer laptop (Acer TravelMate 5730/5730G) has an HDMI port + * on the laptop and a DVI port on the docking station and + * both share the same encoder, hpd pin, and ddc line. + * So while the bios table is technically correct, + * we drop the DVI port here since xrandr has no concept of + * encoders and will try and drive both connectors + * with different crtcs which isn't possible on the hardware + * side and leaves no crtcs for LVDS or VGA. + */ + if (((dev->pdev->device == 0x95c4) || (dev->pdev->device == 0x9591)) && + (dev->pdev->subsystem_vendor == 0x1025) && + (dev->pdev->subsystem_device == 0x013c)) { + if ((*connector_type == DRM_MODE_CONNECTOR_DVII) && + (supported_device == ATOM_DEVICE_DFP1_SUPPORT)) { + /* actually it's a DVI-D port not DVI-I */ + *connector_type = DRM_MODE_CONNECTOR_DVID; + return false; + } + } + + /* XFX Pine Group device rv730 reports no VGA DDC lines + * even though they are wired up to record 0x93 + */ + if ((dev->pdev->device == 0x9498) && + (dev->pdev->subsystem_vendor == 0x1682) && + (dev->pdev->subsystem_device == 0x2452) && + (i2c_bus->valid == false) && + !(supported_device & (ATOM_DEVICE_TV_SUPPORT | ATOM_DEVICE_CV_SUPPORT))) { + struct radeon_device *rdev = dev->dev_private; + *i2c_bus = radeon_lookup_i2c_gpio(rdev, 0x93); + } + + /* Fujitsu D3003-S2 board lists DVI-I as DVI-D and VGA */ + if ((dev->pdev->device == 0x9802) && + (dev->pdev->subsystem_vendor == 0x1734) && + (dev->pdev->subsystem_device == 0x11bd)) { + if (*connector_type == DRM_MODE_CONNECTOR_VGA) { + *connector_type = DRM_MODE_CONNECTOR_DVII; + *line_mux = 0x3103; + } else if (*connector_type == DRM_MODE_CONNECTOR_DVID) { + *connector_type = DRM_MODE_CONNECTOR_DVII; + } + } + + + return true; +} + +const int supported_devices_connector_convert[] = { + DRM_MODE_CONNECTOR_Unknown, + DRM_MODE_CONNECTOR_VGA, + DRM_MODE_CONNECTOR_DVII, + DRM_MODE_CONNECTOR_DVID, + DRM_MODE_CONNECTOR_DVIA, + DRM_MODE_CONNECTOR_SVIDEO, + DRM_MODE_CONNECTOR_Composite, + DRM_MODE_CONNECTOR_LVDS, + DRM_MODE_CONNECTOR_Unknown, + DRM_MODE_CONNECTOR_Unknown, + DRM_MODE_CONNECTOR_HDMIA, + DRM_MODE_CONNECTOR_HDMIB, + DRM_MODE_CONNECTOR_Unknown, + DRM_MODE_CONNECTOR_Unknown, + DRM_MODE_CONNECTOR_9PinDIN, + DRM_MODE_CONNECTOR_DisplayPort +}; + +const uint16_t supported_devices_connector_object_id_convert[] = { + CONNECTOR_OBJECT_ID_NONE, + CONNECTOR_OBJECT_ID_VGA, + CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_I, /* not all boards support DL */ + CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_D, /* not all boards support DL */ + CONNECTOR_OBJECT_ID_VGA, /* technically DVI-A */ + CONNECTOR_OBJECT_ID_COMPOSITE, + CONNECTOR_OBJECT_ID_SVIDEO, + CONNECTOR_OBJECT_ID_LVDS, + CONNECTOR_OBJECT_ID_9PIN_DIN, + CONNECTOR_OBJECT_ID_9PIN_DIN, + CONNECTOR_OBJECT_ID_DISPLAYPORT, + CONNECTOR_OBJECT_ID_HDMI_TYPE_A, + CONNECTOR_OBJECT_ID_HDMI_TYPE_B, + CONNECTOR_OBJECT_ID_SVIDEO +}; + +const int object_connector_convert[] = { + DRM_MODE_CONNECTOR_Unknown, + DRM_MODE_CONNECTOR_DVII, + DRM_MODE_CONNECTOR_DVII, + DRM_MODE_CONNECTOR_DVID, + DRM_MODE_CONNECTOR_DVID, + DRM_MODE_CONNECTOR_VGA, + DRM_MODE_CONNECTOR_Composite, + DRM_MODE_CONNECTOR_SVIDEO, + DRM_MODE_CONNECTOR_Unknown, + DRM_MODE_CONNECTOR_Unknown, + DRM_MODE_CONNECTOR_9PinDIN, + DRM_MODE_CONNECTOR_Unknown, + DRM_MODE_CONNECTOR_HDMIA, + DRM_MODE_CONNECTOR_HDMIB, + DRM_MODE_CONNECTOR_LVDS, + DRM_MODE_CONNECTOR_9PinDIN, + DRM_MODE_CONNECTOR_Unknown, + DRM_MODE_CONNECTOR_Unknown, + DRM_MODE_CONNECTOR_Unknown, + DRM_MODE_CONNECTOR_DisplayPort, + DRM_MODE_CONNECTOR_eDP, + DRM_MODE_CONNECTOR_Unknown +}; + +bool radeon_get_atom_connector_info_from_object_table(struct drm_device *dev) +{ + struct radeon_device *rdev = dev->dev_private; + struct radeon_mode_info *mode_info = &rdev->mode_info; + struct atom_context *ctx = mode_info->atom_context; + int index = GetIndexIntoMasterTable(DATA, Object_Header); + u16 size, data_offset; + u8 frev, crev; + ATOM_CONNECTOR_OBJECT_TABLE *con_obj; + ATOM_ENCODER_OBJECT_TABLE *enc_obj; + ATOM_OBJECT_TABLE *router_obj; + ATOM_DISPLAY_OBJECT_PATH_TABLE *path_obj; + ATOM_OBJECT_HEADER *obj_header; + int i, j, k, path_size, device_support; + int connector_type; + u16 igp_lane_info, conn_id, connector_object_id; + struct radeon_i2c_bus_rec ddc_bus; + struct radeon_router router; + struct radeon_gpio_rec gpio; + struct radeon_hpd hpd; + + if (!atom_parse_data_header(ctx, index, &size, &frev, &crev, &data_offset)) + return false; + + if (crev < 2) + return false; + + obj_header = (ATOM_OBJECT_HEADER *) (ctx->bios + data_offset); + path_obj = (ATOM_DISPLAY_OBJECT_PATH_TABLE *) + (ctx->bios + data_offset + + le16_to_cpu(obj_header->usDisplayPathTableOffset)); + con_obj = (ATOM_CONNECTOR_OBJECT_TABLE *) + (ctx->bios + data_offset + + le16_to_cpu(obj_header->usConnectorObjectTableOffset)); + enc_obj = (ATOM_ENCODER_OBJECT_TABLE *) + (ctx->bios + data_offset + + le16_to_cpu(obj_header->usEncoderObjectTableOffset)); + router_obj = (ATOM_OBJECT_TABLE *) + (ctx->bios + data_offset + + le16_to_cpu(obj_header->usRouterObjectTableOffset)); + device_support = le16_to_cpu(obj_header->usDeviceSupport); + + path_size = 0; + for (i = 0; i < path_obj->ucNumOfDispPath; i++) { + uint8_t *addr = (uint8_t *) path_obj->asDispPath; + ATOM_DISPLAY_OBJECT_PATH *path; + addr += path_size; + path = (ATOM_DISPLAY_OBJECT_PATH *) addr; + path_size += le16_to_cpu(path->usSize); + + if (device_support & le16_to_cpu(path->usDeviceTag)) { + uint8_t con_obj_id, con_obj_num, con_obj_type; + + con_obj_id = + (le16_to_cpu(path->usConnObjectId) & OBJECT_ID_MASK) + >> OBJECT_ID_SHIFT; + con_obj_num = + (le16_to_cpu(path->usConnObjectId) & ENUM_ID_MASK) + >> ENUM_ID_SHIFT; + con_obj_type = + (le16_to_cpu(path->usConnObjectId) & + OBJECT_TYPE_MASK) >> OBJECT_TYPE_SHIFT; + + /* TODO CV support */ + if (le16_to_cpu(path->usDeviceTag) == + ATOM_DEVICE_CV_SUPPORT) + continue; + + /* IGP chips */ + if ((rdev->flags & RADEON_IS_IGP) && + (con_obj_id == + CONNECTOR_OBJECT_ID_PCIE_CONNECTOR)) { + uint16_t igp_offset = 0; + ATOM_INTEGRATED_SYSTEM_INFO_V2 *igp_obj; + + index = + GetIndexIntoMasterTable(DATA, + IntegratedSystemInfo); + + if (atom_parse_data_header(ctx, index, &size, &frev, + &crev, &igp_offset)) { + + if (crev >= 2) { + igp_obj = + (ATOM_INTEGRATED_SYSTEM_INFO_V2 + *) (ctx->bios + igp_offset); + + if (igp_obj) { + uint32_t slot_config, ct; + + if (con_obj_num == 1) + slot_config = + igp_obj-> + ulDDISlot1Config; + else + slot_config = + igp_obj-> + ulDDISlot2Config; + + ct = (slot_config >> 16) & 0xff; + connector_type = + object_connector_convert + [ct]; + connector_object_id = ct; + igp_lane_info = + slot_config & 0xffff; + } else + continue; + } else + continue; + } else { + igp_lane_info = 0; + connector_type = + object_connector_convert[con_obj_id]; + connector_object_id = con_obj_id; + } + } else { + igp_lane_info = 0; + connector_type = + object_connector_convert[con_obj_id]; + connector_object_id = con_obj_id; + } + + if (connector_type == DRM_MODE_CONNECTOR_Unknown) + continue; + + router.ddc_valid = false; + router.cd_valid = false; + for (j = 0; j < ((le16_to_cpu(path->usSize) - 8) / 2); j++) { + uint8_t grph_obj_id, grph_obj_num, grph_obj_type; + + grph_obj_id = + (le16_to_cpu(path->usGraphicObjIds[j]) & + OBJECT_ID_MASK) >> OBJECT_ID_SHIFT; + grph_obj_num = + (le16_to_cpu(path->usGraphicObjIds[j]) & + ENUM_ID_MASK) >> ENUM_ID_SHIFT; + grph_obj_type = + (le16_to_cpu(path->usGraphicObjIds[j]) & + OBJECT_TYPE_MASK) >> OBJECT_TYPE_SHIFT; + + if (grph_obj_type == GRAPH_OBJECT_TYPE_ENCODER) { + for (k = 0; k < enc_obj->ucNumberOfObjects; k++) { + u16 encoder_obj = le16_to_cpu(enc_obj->asObjects[k].usObjectID); + if (le16_to_cpu(path->usGraphicObjIds[j]) == encoder_obj) { + ATOM_COMMON_RECORD_HEADER *record = (ATOM_COMMON_RECORD_HEADER *) + (ctx->bios + data_offset + + le16_to_cpu(enc_obj->asObjects[k].usRecordOffset)); + ATOM_ENCODER_CAP_RECORD *cap_record; + u16 caps = 0; + + while (record->ucRecordSize > 0 && + record->ucRecordType > 0 && + record->ucRecordType <= ATOM_MAX_OBJECT_RECORD_NUMBER) { + switch (record->ucRecordType) { + case ATOM_ENCODER_CAP_RECORD_TYPE: + cap_record =(ATOM_ENCODER_CAP_RECORD *) + record; + caps = le16_to_cpu(cap_record->usEncoderCap); + break; + } + record = (ATOM_COMMON_RECORD_HEADER *) + ((char *)record + record->ucRecordSize); + } + radeon_add_atom_encoder(dev, + encoder_obj, + le16_to_cpu + (path-> + usDeviceTag), + caps); + } + } + } else if (grph_obj_type == GRAPH_OBJECT_TYPE_ROUTER) { + for (k = 0; k < router_obj->ucNumberOfObjects; k++) { + u16 router_obj_id = le16_to_cpu(router_obj->asObjects[k].usObjectID); + if (le16_to_cpu(path->usGraphicObjIds[j]) == router_obj_id) { + ATOM_COMMON_RECORD_HEADER *record = (ATOM_COMMON_RECORD_HEADER *) + (ctx->bios + data_offset + + le16_to_cpu(router_obj->asObjects[k].usRecordOffset)); + ATOM_I2C_RECORD *i2c_record; + ATOM_I2C_ID_CONFIG_ACCESS *i2c_config; + ATOM_ROUTER_DDC_PATH_SELECT_RECORD *ddc_path; + ATOM_ROUTER_DATA_CLOCK_PATH_SELECT_RECORD *cd_path; + ATOM_SRC_DST_TABLE_FOR_ONE_OBJECT *router_src_dst_table = + (ATOM_SRC_DST_TABLE_FOR_ONE_OBJECT *) + (ctx->bios + data_offset + + le16_to_cpu(router_obj->asObjects[k].usSrcDstTableOffset)); + int enum_id; + + router.router_id = router_obj_id; + for (enum_id = 0; enum_id < router_src_dst_table->ucNumberOfDst; + enum_id++) { + if (le16_to_cpu(path->usConnObjectId) == + le16_to_cpu(router_src_dst_table->usDstObjectID[enum_id])) + break; + } + + while (record->ucRecordSize > 0 && + record->ucRecordType > 0 && + record->ucRecordType <= ATOM_MAX_OBJECT_RECORD_NUMBER) { + switch (record->ucRecordType) { + case ATOM_I2C_RECORD_TYPE: + i2c_record = + (ATOM_I2C_RECORD *) + record; + i2c_config = + (ATOM_I2C_ID_CONFIG_ACCESS *) + &i2c_record->sucI2cId; + router.i2c_info = + radeon_lookup_i2c_gpio(rdev, + i2c_config-> + ucAccess); + router.i2c_addr = i2c_record->ucI2CAddr >> 1; + break; + case ATOM_ROUTER_DDC_PATH_SELECT_RECORD_TYPE: + ddc_path = (ATOM_ROUTER_DDC_PATH_SELECT_RECORD *) + record; + router.ddc_valid = true; + router.ddc_mux_type = ddc_path->ucMuxType; + router.ddc_mux_control_pin = ddc_path->ucMuxControlPin; + router.ddc_mux_state = ddc_path->ucMuxState[enum_id]; + break; + case ATOM_ROUTER_DATA_CLOCK_PATH_SELECT_RECORD_TYPE: + cd_path = (ATOM_ROUTER_DATA_CLOCK_PATH_SELECT_RECORD *) + record; + router.cd_valid = true; + router.cd_mux_type = cd_path->ucMuxType; + router.cd_mux_control_pin = cd_path->ucMuxControlPin; + router.cd_mux_state = cd_path->ucMuxState[enum_id]; + break; + } + record = (ATOM_COMMON_RECORD_HEADER *) + ((char *)record + record->ucRecordSize); + } + } + } + } + } + + /* look up gpio for ddc, hpd */ + ddc_bus.valid = false; + hpd.hpd = RADEON_HPD_NONE; + if ((le16_to_cpu(path->usDeviceTag) & + (ATOM_DEVICE_TV_SUPPORT | ATOM_DEVICE_CV_SUPPORT)) == 0) { + for (j = 0; j < con_obj->ucNumberOfObjects; j++) { + if (le16_to_cpu(path->usConnObjectId) == + le16_to_cpu(con_obj->asObjects[j]. + usObjectID)) { + ATOM_COMMON_RECORD_HEADER + *record = + (ATOM_COMMON_RECORD_HEADER + *) + (ctx->bios + data_offset + + le16_to_cpu(con_obj-> + asObjects[j]. + usRecordOffset)); + ATOM_I2C_RECORD *i2c_record; + ATOM_HPD_INT_RECORD *hpd_record; + ATOM_I2C_ID_CONFIG_ACCESS *i2c_config; + + while (record->ucRecordSize > 0 && + record->ucRecordType > 0 && + record->ucRecordType <= ATOM_MAX_OBJECT_RECORD_NUMBER) { + switch (record->ucRecordType) { + case ATOM_I2C_RECORD_TYPE: + i2c_record = + (ATOM_I2C_RECORD *) + record; + i2c_config = + (ATOM_I2C_ID_CONFIG_ACCESS *) + &i2c_record->sucI2cId; + ddc_bus = radeon_lookup_i2c_gpio(rdev, + i2c_config-> + ucAccess); + break; + case ATOM_HPD_INT_RECORD_TYPE: + hpd_record = + (ATOM_HPD_INT_RECORD *) + record; + gpio = radeon_lookup_gpio(rdev, + hpd_record->ucHPDIntGPIOID); + hpd = radeon_atom_get_hpd_info_from_gpio(rdev, &gpio); + hpd.plugged_state = hpd_record->ucPlugged_PinState; + break; + } + record = + (ATOM_COMMON_RECORD_HEADER + *) ((char *)record + + + record-> + ucRecordSize); + } + break; + } + } + } + + /* needed for aux chan transactions */ + ddc_bus.hpd = hpd.hpd; + + conn_id = le16_to_cpu(path->usConnObjectId); + + if (!radeon_atom_apply_quirks + (dev, le16_to_cpu(path->usDeviceTag), &connector_type, + &ddc_bus, &conn_id, &hpd)) + continue; + + radeon_add_atom_connector(dev, + conn_id, + le16_to_cpu(path-> + usDeviceTag), + connector_type, &ddc_bus, + igp_lane_info, + connector_object_id, + &hpd, + &router); + + } + } + + radeon_link_encoder_connector(dev); + + return true; +} + +static uint16_t atombios_get_connector_object_id(struct drm_device *dev, + int connector_type, + uint16_t devices) +{ + struct radeon_device *rdev = dev->dev_private; + + if (rdev->flags & RADEON_IS_IGP) { + return supported_devices_connector_object_id_convert + [connector_type]; + } else if (((connector_type == DRM_MODE_CONNECTOR_DVII) || + (connector_type == DRM_MODE_CONNECTOR_DVID)) && + (devices & ATOM_DEVICE_DFP2_SUPPORT)) { + struct radeon_mode_info *mode_info = &rdev->mode_info; + struct atom_context *ctx = mode_info->atom_context; + int index = GetIndexIntoMasterTable(DATA, XTMDS_Info); + uint16_t size, data_offset; + uint8_t frev, crev; + ATOM_XTMDS_INFO *xtmds; + + if (atom_parse_data_header(ctx, index, &size, &frev, &crev, &data_offset)) { + xtmds = (ATOM_XTMDS_INFO *)(ctx->bios + data_offset); + + if (xtmds->ucSupportedLink & ATOM_XTMDS_SUPPORTED_DUALLINK) { + if (connector_type == DRM_MODE_CONNECTOR_DVII) + return CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_I; + else + return CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_D; + } else { + if (connector_type == DRM_MODE_CONNECTOR_DVII) + return CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I; + else + return CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D; + } + } else + return supported_devices_connector_object_id_convert + [connector_type]; + } else { + return supported_devices_connector_object_id_convert + [connector_type]; + } +} + +struct bios_connector { + bool valid; + uint16_t line_mux; + uint16_t devices; + int connector_type; + struct radeon_i2c_bus_rec ddc_bus; + struct radeon_hpd hpd; +}; + +bool radeon_get_atom_connector_info_from_supported_devices_table(struct + drm_device + *dev) +{ + struct radeon_device *rdev = dev->dev_private; + struct radeon_mode_info *mode_info = &rdev->mode_info; + struct atom_context *ctx = mode_info->atom_context; + int index = GetIndexIntoMasterTable(DATA, SupportedDevicesInfo); + uint16_t size, data_offset; + uint8_t frev, crev; + uint16_t device_support; + uint8_t dac; + union atom_supported_devices *supported_devices; + int i, j, max_device; + struct bios_connector *bios_connectors; + size_t bc_size = sizeof(*bios_connectors) * ATOM_MAX_SUPPORTED_DEVICE; + struct radeon_router router; + + router.ddc_valid = false; + router.cd_valid = false; + + bios_connectors = kzalloc(bc_size, GFP_KERNEL); + if (!bios_connectors) + return false; + + if (!atom_parse_data_header(ctx, index, &size, &frev, &crev, + &data_offset)) { + kfree(bios_connectors); + return false; + } + + supported_devices = + (union atom_supported_devices *)(ctx->bios + data_offset); + + device_support = le16_to_cpu(supported_devices->info.usDeviceSupport); + + if (frev > 1) + max_device = ATOM_MAX_SUPPORTED_DEVICE; + else + max_device = ATOM_MAX_SUPPORTED_DEVICE_INFO; + + for (i = 0; i < max_device; i++) { + ATOM_CONNECTOR_INFO_I2C ci = + supported_devices->info.asConnInfo[i]; + + bios_connectors[i].valid = false; + + if (!(device_support & (1 << i))) { + continue; + } + + if (i == ATOM_DEVICE_CV_INDEX) { + DRM_DEBUG_KMS("Skipping Component Video\n"); + continue; + } + + bios_connectors[i].connector_type = + supported_devices_connector_convert[ci.sucConnectorInfo. + sbfAccess. + bfConnectorType]; + + if (bios_connectors[i].connector_type == + DRM_MODE_CONNECTOR_Unknown) + continue; + + dac = ci.sucConnectorInfo.sbfAccess.bfAssociatedDAC; + + bios_connectors[i].line_mux = + ci.sucI2cId.ucAccess; + + /* give tv unique connector ids */ + if (i == ATOM_DEVICE_TV1_INDEX) { + bios_connectors[i].ddc_bus.valid = false; + bios_connectors[i].line_mux = 50; + } else if (i == ATOM_DEVICE_TV2_INDEX) { + bios_connectors[i].ddc_bus.valid = false; + bios_connectors[i].line_mux = 51; + } else if (i == ATOM_DEVICE_CV_INDEX) { + bios_connectors[i].ddc_bus.valid = false; + bios_connectors[i].line_mux = 52; + } else + bios_connectors[i].ddc_bus = + radeon_lookup_i2c_gpio(rdev, + bios_connectors[i].line_mux); + + if ((crev > 1) && (frev > 1)) { + u8 isb = supported_devices->info_2d1.asIntSrcInfo[i].ucIntSrcBitmap; + switch (isb) { + case 0x4: + bios_connectors[i].hpd.hpd = RADEON_HPD_1; + break; + case 0xa: + bios_connectors[i].hpd.hpd = RADEON_HPD_2; + break; + default: + bios_connectors[i].hpd.hpd = RADEON_HPD_NONE; + break; + } + } else { + if (i == ATOM_DEVICE_DFP1_INDEX) + bios_connectors[i].hpd.hpd = RADEON_HPD_1; + else if (i == ATOM_DEVICE_DFP2_INDEX) + bios_connectors[i].hpd.hpd = RADEON_HPD_2; + else + bios_connectors[i].hpd.hpd = RADEON_HPD_NONE; + } + + /* Always set the connector type to VGA for CRT1/CRT2. if they are + * shared with a DVI port, we'll pick up the DVI connector when we + * merge the outputs. Some bioses incorrectly list VGA ports as DVI. + */ + if (i == ATOM_DEVICE_CRT1_INDEX || i == ATOM_DEVICE_CRT2_INDEX) + bios_connectors[i].connector_type = + DRM_MODE_CONNECTOR_VGA; + + if (!radeon_atom_apply_quirks + (dev, (1 << i), &bios_connectors[i].connector_type, + &bios_connectors[i].ddc_bus, &bios_connectors[i].line_mux, + &bios_connectors[i].hpd)) + continue; + + bios_connectors[i].valid = true; + bios_connectors[i].devices = (1 << i); + + if (ASIC_IS_AVIVO(rdev) || radeon_r4xx_atom) + radeon_add_atom_encoder(dev, + radeon_get_encoder_enum(dev, + (1 << i), + dac), + (1 << i), + 0); + else + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + (1 << i), + dac), + (1 << i)); + } + + /* combine shared connectors */ + for (i = 0; i < max_device; i++) { + if (bios_connectors[i].valid) { + for (j = 0; j < max_device; j++) { + if (bios_connectors[j].valid && (i != j)) { + if (bios_connectors[i].line_mux == + bios_connectors[j].line_mux) { + /* make sure not to combine LVDS */ + if (bios_connectors[i].devices & (ATOM_DEVICE_LCD_SUPPORT)) { + bios_connectors[i].line_mux = 53; + bios_connectors[i].ddc_bus.valid = false; + continue; + } + if (bios_connectors[j].devices & (ATOM_DEVICE_LCD_SUPPORT)) { + bios_connectors[j].line_mux = 53; + bios_connectors[j].ddc_bus.valid = false; + continue; + } + /* combine analog and digital for DVI-I */ + if (((bios_connectors[i].devices & (ATOM_DEVICE_DFP_SUPPORT)) && + (bios_connectors[j].devices & (ATOM_DEVICE_CRT_SUPPORT))) || + ((bios_connectors[j].devices & (ATOM_DEVICE_DFP_SUPPORT)) && + (bios_connectors[i].devices & (ATOM_DEVICE_CRT_SUPPORT)))) { + bios_connectors[i].devices |= + bios_connectors[j].devices; + bios_connectors[i].connector_type = + DRM_MODE_CONNECTOR_DVII; + if (bios_connectors[j].devices & (ATOM_DEVICE_DFP_SUPPORT)) + bios_connectors[i].hpd = + bios_connectors[j].hpd; + bios_connectors[j].valid = false; + } + } + } + } + } + } + + /* add the connectors */ + for (i = 0; i < max_device; i++) { + if (bios_connectors[i].valid) { + uint16_t connector_object_id = + atombios_get_connector_object_id(dev, + bios_connectors[i].connector_type, + bios_connectors[i].devices); + radeon_add_atom_connector(dev, + bios_connectors[i].line_mux, + bios_connectors[i].devices, + bios_connectors[i]. + connector_type, + &bios_connectors[i].ddc_bus, + 0, + connector_object_id, + &bios_connectors[i].hpd, + &router); + } + } + + radeon_link_encoder_connector(dev); + + kfree(bios_connectors); + return true; +} + +union firmware_info { + ATOM_FIRMWARE_INFO info; + ATOM_FIRMWARE_INFO_V1_2 info_12; + ATOM_FIRMWARE_INFO_V1_3 info_13; + ATOM_FIRMWARE_INFO_V1_4 info_14; + ATOM_FIRMWARE_INFO_V2_1 info_21; + ATOM_FIRMWARE_INFO_V2_2 info_22; +}; + +bool radeon_atom_get_clock_info(struct drm_device *dev) +{ + struct radeon_device *rdev = dev->dev_private; + struct radeon_mode_info *mode_info = &rdev->mode_info; + int index = GetIndexIntoMasterTable(DATA, FirmwareInfo); + union firmware_info *firmware_info; + uint8_t frev, crev; + struct radeon_pll *p1pll = &rdev->clock.p1pll; + struct radeon_pll *p2pll = &rdev->clock.p2pll; + struct radeon_pll *dcpll = &rdev->clock.dcpll; + struct radeon_pll *spll = &rdev->clock.spll; + struct radeon_pll *mpll = &rdev->clock.mpll; + uint16_t data_offset; + + if (atom_parse_data_header(mode_info->atom_context, index, NULL, + &frev, &crev, &data_offset)) { + firmware_info = + (union firmware_info *)(mode_info->atom_context->bios + + data_offset); + /* pixel clocks */ + p1pll->reference_freq = + le16_to_cpu(firmware_info->info.usReferenceClock); + p1pll->reference_div = 0; + + if (crev < 2) + p1pll->pll_out_min = + le16_to_cpu(firmware_info->info.usMinPixelClockPLL_Output); + else + p1pll->pll_out_min = + le32_to_cpu(firmware_info->info_12.ulMinPixelClockPLL_Output); + p1pll->pll_out_max = + le32_to_cpu(firmware_info->info.ulMaxPixelClockPLL_Output); + + if (crev >= 4) { + p1pll->lcd_pll_out_min = + le16_to_cpu(firmware_info->info_14.usLcdMinPixelClockPLL_Output) * 100; + if (p1pll->lcd_pll_out_min == 0) + p1pll->lcd_pll_out_min = p1pll->pll_out_min; + p1pll->lcd_pll_out_max = + le16_to_cpu(firmware_info->info_14.usLcdMaxPixelClockPLL_Output) * 100; + if (p1pll->lcd_pll_out_max == 0) + p1pll->lcd_pll_out_max = p1pll->pll_out_max; + } else { + p1pll->lcd_pll_out_min = p1pll->pll_out_min; + p1pll->lcd_pll_out_max = p1pll->pll_out_max; + } + + if (p1pll->pll_out_min == 0) { + if (ASIC_IS_AVIVO(rdev)) + p1pll->pll_out_min = 64800; + else + p1pll->pll_out_min = 20000; + } + + p1pll->pll_in_min = + le16_to_cpu(firmware_info->info.usMinPixelClockPLL_Input); + p1pll->pll_in_max = + le16_to_cpu(firmware_info->info.usMaxPixelClockPLL_Input); + + *p2pll = *p1pll; + + /* system clock */ + if (ASIC_IS_DCE4(rdev)) + spll->reference_freq = + le16_to_cpu(firmware_info->info_21.usCoreReferenceClock); + else + spll->reference_freq = + le16_to_cpu(firmware_info->info.usReferenceClock); + spll->reference_div = 0; + + spll->pll_out_min = + le16_to_cpu(firmware_info->info.usMinEngineClockPLL_Output); + spll->pll_out_max = + le32_to_cpu(firmware_info->info.ulMaxEngineClockPLL_Output); + + /* ??? */ + if (spll->pll_out_min == 0) { + if (ASIC_IS_AVIVO(rdev)) + spll->pll_out_min = 64800; + else + spll->pll_out_min = 20000; + } + + spll->pll_in_min = + le16_to_cpu(firmware_info->info.usMinEngineClockPLL_Input); + spll->pll_in_max = + le16_to_cpu(firmware_info->info.usMaxEngineClockPLL_Input); + + /* memory clock */ + if (ASIC_IS_DCE4(rdev)) + mpll->reference_freq = + le16_to_cpu(firmware_info->info_21.usMemoryReferenceClock); + else + mpll->reference_freq = + le16_to_cpu(firmware_info->info.usReferenceClock); + mpll->reference_div = 0; + + mpll->pll_out_min = + le16_to_cpu(firmware_info->info.usMinMemoryClockPLL_Output); + mpll->pll_out_max = + le32_to_cpu(firmware_info->info.ulMaxMemoryClockPLL_Output); + + /* ??? */ + if (mpll->pll_out_min == 0) { + if (ASIC_IS_AVIVO(rdev)) + mpll->pll_out_min = 64800; + else + mpll->pll_out_min = 20000; + } + + mpll->pll_in_min = + le16_to_cpu(firmware_info->info.usMinMemoryClockPLL_Input); + mpll->pll_in_max = + le16_to_cpu(firmware_info->info.usMaxMemoryClockPLL_Input); + + rdev->clock.default_sclk = + le32_to_cpu(firmware_info->info.ulDefaultEngineClock); + rdev->clock.default_mclk = + le32_to_cpu(firmware_info->info.ulDefaultMemoryClock); + + if (ASIC_IS_DCE4(rdev)) { + rdev->clock.default_dispclk = + le32_to_cpu(firmware_info->info_21.ulDefaultDispEngineClkFreq); + if (rdev->clock.default_dispclk == 0) { + if (ASIC_IS_DCE5(rdev)) + rdev->clock.default_dispclk = 54000; /* 540 Mhz */ + else + rdev->clock.default_dispclk = 60000; /* 600 Mhz */ + } + rdev->clock.dp_extclk = + le16_to_cpu(firmware_info->info_21.usUniphyDPModeExtClkFreq); + } + *dcpll = *p1pll; + + rdev->clock.max_pixel_clock = le16_to_cpu(firmware_info->info.usMaxPixelClock); + if (rdev->clock.max_pixel_clock == 0) + rdev->clock.max_pixel_clock = 40000; + + return true; + } + + return false; +} + +union igp_info { + struct _ATOM_INTEGRATED_SYSTEM_INFO info; + struct _ATOM_INTEGRATED_SYSTEM_INFO_V2 info_2; +}; + +bool radeon_atombios_sideport_present(struct radeon_device *rdev) +{ + struct radeon_mode_info *mode_info = &rdev->mode_info; + int index = GetIndexIntoMasterTable(DATA, IntegratedSystemInfo); + union igp_info *igp_info; + u8 frev, crev; + u16 data_offset; + + /* sideport is AMD only */ + if (rdev->family == CHIP_RS600) + return false; + + if (atom_parse_data_header(mode_info->atom_context, index, NULL, + &frev, &crev, &data_offset)) { + igp_info = (union igp_info *)(mode_info->atom_context->bios + + data_offset); + switch (crev) { + case 1: + if (le32_to_cpu(igp_info->info.ulBootUpMemoryClock)) + return true; + break; + case 2: + if (le32_to_cpu(igp_info->info_2.ulBootUpSidePortClock)) + return true; + break; + default: + DRM_ERROR("Unsupported IGP table: %d %d\n", frev, crev); + break; + } + } + return false; +} + +bool radeon_atombios_get_tmds_info(struct radeon_encoder *encoder, + struct radeon_encoder_int_tmds *tmds) +{ + struct drm_device *dev = encoder->base.dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_mode_info *mode_info = &rdev->mode_info; + int index = GetIndexIntoMasterTable(DATA, TMDS_Info); + uint16_t data_offset; + struct _ATOM_TMDS_INFO *tmds_info; + uint8_t frev, crev; + uint16_t maxfreq; + int i; + + if (atom_parse_data_header(mode_info->atom_context, index, NULL, + &frev, &crev, &data_offset)) { + tmds_info = + (struct _ATOM_TMDS_INFO *)(mode_info->atom_context->bios + + data_offset); + + maxfreq = le16_to_cpu(tmds_info->usMaxFrequency); + for (i = 0; i < 4; i++) { + tmds->tmds_pll[i].freq = + le16_to_cpu(tmds_info->asMiscInfo[i].usFrequency); + tmds->tmds_pll[i].value = + tmds_info->asMiscInfo[i].ucPLL_ChargePump & 0x3f; + tmds->tmds_pll[i].value |= + (tmds_info->asMiscInfo[i]. + ucPLL_VCO_Gain & 0x3f) << 6; + tmds->tmds_pll[i].value |= + (tmds_info->asMiscInfo[i]. + ucPLL_DutyCycle & 0xf) << 12; + tmds->tmds_pll[i].value |= + (tmds_info->asMiscInfo[i]. + ucPLL_VoltageSwing & 0xf) << 16; + + DRM_DEBUG_KMS("TMDS PLL From ATOMBIOS %u %x\n", + tmds->tmds_pll[i].freq, + tmds->tmds_pll[i].value); + + if (maxfreq == tmds->tmds_pll[i].freq) { + tmds->tmds_pll[i].freq = 0xffffffff; + break; + } + } + return true; + } + return false; +} + +bool radeon_atombios_get_ppll_ss_info(struct radeon_device *rdev, + struct radeon_atom_ss *ss, + int id) +{ + struct radeon_mode_info *mode_info = &rdev->mode_info; + int index = GetIndexIntoMasterTable(DATA, PPLL_SS_Info); + uint16_t data_offset, size; + struct _ATOM_SPREAD_SPECTRUM_INFO *ss_info; + uint8_t frev, crev; + int i, num_indices; + + memset(ss, 0, sizeof(struct radeon_atom_ss)); + if (atom_parse_data_header(mode_info->atom_context, index, &size, + &frev, &crev, &data_offset)) { + ss_info = + (struct _ATOM_SPREAD_SPECTRUM_INFO *)(mode_info->atom_context->bios + data_offset); + + num_indices = (size - sizeof(ATOM_COMMON_TABLE_HEADER)) / + sizeof(ATOM_SPREAD_SPECTRUM_ASSIGNMENT); + + for (i = 0; i < num_indices; i++) { + if (ss_info->asSS_Info[i].ucSS_Id == id) { + ss->percentage = + le16_to_cpu(ss_info->asSS_Info[i].usSpreadSpectrumPercentage); + ss->type = ss_info->asSS_Info[i].ucSpreadSpectrumType; + ss->step = ss_info->asSS_Info[i].ucSS_Step; + ss->delay = ss_info->asSS_Info[i].ucSS_Delay; + ss->range = ss_info->asSS_Info[i].ucSS_Range; + ss->refdiv = ss_info->asSS_Info[i].ucRecommendedRef_Div; + return true; + } + } + } + return false; +} + +static void radeon_atombios_get_igp_ss_overrides(struct radeon_device *rdev, + struct radeon_atom_ss *ss, + int id) +{ + struct radeon_mode_info *mode_info = &rdev->mode_info; + int index = GetIndexIntoMasterTable(DATA, IntegratedSystemInfo); + u16 data_offset, size; + struct _ATOM_INTEGRATED_SYSTEM_INFO_V6 *igp_info; + u8 frev, crev; + u16 percentage = 0, rate = 0; + + /* get any igp specific overrides */ + if (atom_parse_data_header(mode_info->atom_context, index, &size, + &frev, &crev, &data_offset)) { + igp_info = (struct _ATOM_INTEGRATED_SYSTEM_INFO_V6 *) + (mode_info->atom_context->bios + data_offset); + switch (id) { + case ASIC_INTERNAL_SS_ON_TMDS: + percentage = le16_to_cpu(igp_info->usDVISSPercentage); + rate = le16_to_cpu(igp_info->usDVISSpreadRateIn10Hz); + break; + case ASIC_INTERNAL_SS_ON_HDMI: + percentage = le16_to_cpu(igp_info->usHDMISSPercentage); + rate = le16_to_cpu(igp_info->usHDMISSpreadRateIn10Hz); + break; + case ASIC_INTERNAL_SS_ON_LVDS: + percentage = le16_to_cpu(igp_info->usLvdsSSPercentage); + rate = le16_to_cpu(igp_info->usLvdsSSpreadRateIn10Hz); + break; + } + if (percentage) + ss->percentage = percentage; + if (rate) + ss->rate = rate; + } +} + +union asic_ss_info { + struct _ATOM_ASIC_INTERNAL_SS_INFO info; + struct _ATOM_ASIC_INTERNAL_SS_INFO_V2 info_2; + struct _ATOM_ASIC_INTERNAL_SS_INFO_V3 info_3; +}; + +bool radeon_atombios_get_asic_ss_info(struct radeon_device *rdev, + struct radeon_atom_ss *ss, + int id, u32 clock) +{ + struct radeon_mode_info *mode_info = &rdev->mode_info; + int index = GetIndexIntoMasterTable(DATA, ASIC_InternalSS_Info); + uint16_t data_offset, size; + union asic_ss_info *ss_info; + uint8_t frev, crev; + int i, num_indices; + + memset(ss, 0, sizeof(struct radeon_atom_ss)); + if (atom_parse_data_header(mode_info->atom_context, index, &size, + &frev, &crev, &data_offset)) { + + ss_info = + (union asic_ss_info *)(mode_info->atom_context->bios + data_offset); + + switch (frev) { + case 1: + num_indices = (size - sizeof(ATOM_COMMON_TABLE_HEADER)) / + sizeof(ATOM_ASIC_SS_ASSIGNMENT); + + for (i = 0; i < num_indices; i++) { + if ((ss_info->info.asSpreadSpectrum[i].ucClockIndication == id) && + (clock <= le32_to_cpu(ss_info->info.asSpreadSpectrum[i].ulTargetClockRange))) { + ss->percentage = + le16_to_cpu(ss_info->info.asSpreadSpectrum[i].usSpreadSpectrumPercentage); + ss->type = ss_info->info.asSpreadSpectrum[i].ucSpreadSpectrumMode; + ss->rate = le16_to_cpu(ss_info->info.asSpreadSpectrum[i].usSpreadRateInKhz); + return true; + } + } + break; + case 2: + num_indices = (size - sizeof(ATOM_COMMON_TABLE_HEADER)) / + sizeof(ATOM_ASIC_SS_ASSIGNMENT_V2); + for (i = 0; i < num_indices; i++) { + if ((ss_info->info_2.asSpreadSpectrum[i].ucClockIndication == id) && + (clock <= le32_to_cpu(ss_info->info_2.asSpreadSpectrum[i].ulTargetClockRange))) { + ss->percentage = + le16_to_cpu(ss_info->info_2.asSpreadSpectrum[i].usSpreadSpectrumPercentage); + ss->type = ss_info->info_2.asSpreadSpectrum[i].ucSpreadSpectrumMode; + ss->rate = le16_to_cpu(ss_info->info_2.asSpreadSpectrum[i].usSpreadRateIn10Hz); + return true; + } + } + break; + case 3: + num_indices = (size - sizeof(ATOM_COMMON_TABLE_HEADER)) / + sizeof(ATOM_ASIC_SS_ASSIGNMENT_V3); + for (i = 0; i < num_indices; i++) { + if ((ss_info->info_3.asSpreadSpectrum[i].ucClockIndication == id) && + (clock <= le32_to_cpu(ss_info->info_3.asSpreadSpectrum[i].ulTargetClockRange))) { + ss->percentage = + le16_to_cpu(ss_info->info_3.asSpreadSpectrum[i].usSpreadSpectrumPercentage); + ss->type = ss_info->info_3.asSpreadSpectrum[i].ucSpreadSpectrumMode; + ss->rate = le16_to_cpu(ss_info->info_3.asSpreadSpectrum[i].usSpreadRateIn10Hz); + if (rdev->flags & RADEON_IS_IGP) + radeon_atombios_get_igp_ss_overrides(rdev, ss, id); + return true; + } + } + break; + default: + DRM_ERROR("Unsupported ASIC_InternalSS_Info table: %d %d\n", frev, crev); + break; + } + + } + return false; +} + +union lvds_info { + struct _ATOM_LVDS_INFO info; + struct _ATOM_LVDS_INFO_V12 info_12; +}; + +struct radeon_encoder_atom_dig *radeon_atombios_get_lvds_info(struct + radeon_encoder + *encoder) +{ + struct drm_device *dev = encoder->base.dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_mode_info *mode_info = &rdev->mode_info; + int index = GetIndexIntoMasterTable(DATA, LVDS_Info); + uint16_t data_offset, misc; + union lvds_info *lvds_info; + uint8_t frev, crev; + struct radeon_encoder_atom_dig *lvds = NULL; + int encoder_enum = (encoder->encoder_enum & ENUM_ID_MASK) >> ENUM_ID_SHIFT; + + if (atom_parse_data_header(mode_info->atom_context, index, NULL, + &frev, &crev, &data_offset)) { + lvds_info = + (union lvds_info *)(mode_info->atom_context->bios + data_offset); + lvds = + kzalloc(sizeof(struct radeon_encoder_atom_dig), GFP_KERNEL); + + if (!lvds) + return NULL; + + lvds->native_mode.clock = + le16_to_cpu(lvds_info->info.sLCDTiming.usPixClk) * 10; + lvds->native_mode.hdisplay = + le16_to_cpu(lvds_info->info.sLCDTiming.usHActive); + lvds->native_mode.vdisplay = + le16_to_cpu(lvds_info->info.sLCDTiming.usVActive); + lvds->native_mode.htotal = lvds->native_mode.hdisplay + + le16_to_cpu(lvds_info->info.sLCDTiming.usHBlanking_Time); + lvds->native_mode.hsync_start = lvds->native_mode.hdisplay + + le16_to_cpu(lvds_info->info.sLCDTiming.usHSyncOffset); + lvds->native_mode.hsync_end = lvds->native_mode.hsync_start + + le16_to_cpu(lvds_info->info.sLCDTiming.usHSyncWidth); + lvds->native_mode.vtotal = lvds->native_mode.vdisplay + + le16_to_cpu(lvds_info->info.sLCDTiming.usVBlanking_Time); + lvds->native_mode.vsync_start = lvds->native_mode.vdisplay + + le16_to_cpu(lvds_info->info.sLCDTiming.usVSyncOffset); + lvds->native_mode.vsync_end = lvds->native_mode.vsync_start + + le16_to_cpu(lvds_info->info.sLCDTiming.usVSyncWidth); + lvds->panel_pwr_delay = + le16_to_cpu(lvds_info->info.usOffDelayInMs); + lvds->lcd_misc = lvds_info->info.ucLVDS_Misc; + + misc = le16_to_cpu(lvds_info->info.sLCDTiming.susModeMiscInfo.usAccess); + if (misc & ATOM_VSYNC_POLARITY) + lvds->native_mode.flags |= DRM_MODE_FLAG_NVSYNC; + if (misc & ATOM_HSYNC_POLARITY) + lvds->native_mode.flags |= DRM_MODE_FLAG_NHSYNC; + if (misc & ATOM_COMPOSITESYNC) + lvds->native_mode.flags |= DRM_MODE_FLAG_CSYNC; + if (misc & ATOM_INTERLACE) + lvds->native_mode.flags |= DRM_MODE_FLAG_INTERLACE; + if (misc & ATOM_DOUBLE_CLOCK_MODE) + lvds->native_mode.flags |= DRM_MODE_FLAG_DBLSCAN; + + lvds->native_mode.width_mm = le16_to_cpu(lvds_info->info.sLCDTiming.usImageHSize); + lvds->native_mode.height_mm = le16_to_cpu(lvds_info->info.sLCDTiming.usImageVSize); + + /* set crtc values */ + drm_mode_set_crtcinfo(&lvds->native_mode, CRTC_INTERLACE_HALVE_V); + + lvds->lcd_ss_id = lvds_info->info.ucSS_Id; + + encoder->native_mode = lvds->native_mode; + + if (encoder_enum == 2) + lvds->linkb = true; + else + lvds->linkb = false; + + /* parse the lcd record table */ + if (le16_to_cpu(lvds_info->info.usModePatchTableOffset)) { + ATOM_FAKE_EDID_PATCH_RECORD *fake_edid_record; + ATOM_PANEL_RESOLUTION_PATCH_RECORD *panel_res_record; + bool bad_record = false; + u8 *record; + + if ((frev == 1) && (crev < 2)) + /* absolute */ + record = (u8 *)(mode_info->atom_context->bios + + le16_to_cpu(lvds_info->info.usModePatchTableOffset)); + else + /* relative */ + record = (u8 *)(mode_info->atom_context->bios + + data_offset + + le16_to_cpu(lvds_info->info.usModePatchTableOffset)); + while (*record != ATOM_RECORD_END_TYPE) { + switch (*record) { + case LCD_MODE_PATCH_RECORD_MODE_TYPE: + record += sizeof(ATOM_PATCH_RECORD_MODE); + break; + case LCD_RTS_RECORD_TYPE: + record += sizeof(ATOM_LCD_RTS_RECORD); + break; + case LCD_CAP_RECORD_TYPE: + record += sizeof(ATOM_LCD_MODE_CONTROL_CAP); + break; + case LCD_FAKE_EDID_PATCH_RECORD_TYPE: + fake_edid_record = (ATOM_FAKE_EDID_PATCH_RECORD *)record; + if (fake_edid_record->ucFakeEDIDLength) { + struct edid *edid; + int edid_size = + max((int)EDID_LENGTH, (int)fake_edid_record->ucFakeEDIDLength); + edid = kmalloc(edid_size, GFP_KERNEL); + if (edid) { + memcpy((u8 *)edid, (u8 *)&fake_edid_record->ucFakeEDIDString[0], + fake_edid_record->ucFakeEDIDLength); + + if (drm_edid_is_valid(edid)) { + rdev->mode_info.bios_hardcoded_edid = edid; + rdev->mode_info.bios_hardcoded_edid_size = edid_size; + } else + kfree(edid); + } + } + record += sizeof(ATOM_FAKE_EDID_PATCH_RECORD); + break; + case LCD_PANEL_RESOLUTION_RECORD_TYPE: + panel_res_record = (ATOM_PANEL_RESOLUTION_PATCH_RECORD *)record; + lvds->native_mode.width_mm = panel_res_record->usHSize; + lvds->native_mode.height_mm = panel_res_record->usVSize; + record += sizeof(ATOM_PANEL_RESOLUTION_PATCH_RECORD); + break; + default: + DRM_ERROR("Bad LCD record %d\n", *record); + bad_record = true; + break; + } + if (bad_record) + break; + } + } + } + return lvds; +} + +struct radeon_encoder_primary_dac * +radeon_atombios_get_primary_dac_info(struct radeon_encoder *encoder) +{ + struct drm_device *dev = encoder->base.dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_mode_info *mode_info = &rdev->mode_info; + int index = GetIndexIntoMasterTable(DATA, CompassionateData); + uint16_t data_offset; + struct _COMPASSIONATE_DATA *dac_info; + uint8_t frev, crev; + uint8_t bg, dac; + struct radeon_encoder_primary_dac *p_dac = NULL; + + if (atom_parse_data_header(mode_info->atom_context, index, NULL, + &frev, &crev, &data_offset)) { + dac_info = (struct _COMPASSIONATE_DATA *) + (mode_info->atom_context->bios + data_offset); + + p_dac = kzalloc(sizeof(struct radeon_encoder_primary_dac), GFP_KERNEL); + + if (!p_dac) + return NULL; + + bg = dac_info->ucDAC1_BG_Adjustment; + dac = dac_info->ucDAC1_DAC_Adjustment; + p_dac->ps2_pdac_adj = (bg << 8) | (dac); + + } + return p_dac; +} + +bool radeon_atom_get_tv_timings(struct radeon_device *rdev, int index, + struct drm_display_mode *mode) +{ + struct radeon_mode_info *mode_info = &rdev->mode_info; + ATOM_ANALOG_TV_INFO *tv_info; + ATOM_ANALOG_TV_INFO_V1_2 *tv_info_v1_2; + ATOM_DTD_FORMAT *dtd_timings; + int data_index = GetIndexIntoMasterTable(DATA, AnalogTV_Info); + u8 frev, crev; + u16 data_offset, misc; + + if (!atom_parse_data_header(mode_info->atom_context, data_index, NULL, + &frev, &crev, &data_offset)) + return false; + + switch (crev) { + case 1: + tv_info = (ATOM_ANALOG_TV_INFO *)(mode_info->atom_context->bios + data_offset); + if (index >= MAX_SUPPORTED_TV_TIMING) + return false; + + mode->crtc_htotal = le16_to_cpu(tv_info->aModeTimings[index].usCRTC_H_Total); + mode->crtc_hdisplay = le16_to_cpu(tv_info->aModeTimings[index].usCRTC_H_Disp); + mode->crtc_hsync_start = le16_to_cpu(tv_info->aModeTimings[index].usCRTC_H_SyncStart); + mode->crtc_hsync_end = le16_to_cpu(tv_info->aModeTimings[index].usCRTC_H_SyncStart) + + le16_to_cpu(tv_info->aModeTimings[index].usCRTC_H_SyncWidth); + + mode->crtc_vtotal = le16_to_cpu(tv_info->aModeTimings[index].usCRTC_V_Total); + mode->crtc_vdisplay = le16_to_cpu(tv_info->aModeTimings[index].usCRTC_V_Disp); + mode->crtc_vsync_start = le16_to_cpu(tv_info->aModeTimings[index].usCRTC_V_SyncStart); + mode->crtc_vsync_end = le16_to_cpu(tv_info->aModeTimings[index].usCRTC_V_SyncStart) + + le16_to_cpu(tv_info->aModeTimings[index].usCRTC_V_SyncWidth); + + mode->flags = 0; + misc = le16_to_cpu(tv_info->aModeTimings[index].susModeMiscInfo.usAccess); + if (misc & ATOM_VSYNC_POLARITY) + mode->flags |= DRM_MODE_FLAG_NVSYNC; + if (misc & ATOM_HSYNC_POLARITY) + mode->flags |= DRM_MODE_FLAG_NHSYNC; + if (misc & ATOM_COMPOSITESYNC) + mode->flags |= DRM_MODE_FLAG_CSYNC; + if (misc & ATOM_INTERLACE) + mode->flags |= DRM_MODE_FLAG_INTERLACE; + if (misc & ATOM_DOUBLE_CLOCK_MODE) + mode->flags |= DRM_MODE_FLAG_DBLSCAN; + + mode->clock = le16_to_cpu(tv_info->aModeTimings[index].usPixelClock) * 10; + + if (index == 1) { + /* PAL timings appear to have wrong values for totals */ + mode->crtc_htotal -= 1; + mode->crtc_vtotal -= 1; + } + break; + case 2: + tv_info_v1_2 = (ATOM_ANALOG_TV_INFO_V1_2 *)(mode_info->atom_context->bios + data_offset); + if (index >= MAX_SUPPORTED_TV_TIMING_V1_2) + return false; + + dtd_timings = &tv_info_v1_2->aModeTimings[index]; + mode->crtc_htotal = le16_to_cpu(dtd_timings->usHActive) + + le16_to_cpu(dtd_timings->usHBlanking_Time); + mode->crtc_hdisplay = le16_to_cpu(dtd_timings->usHActive); + mode->crtc_hsync_start = le16_to_cpu(dtd_timings->usHActive) + + le16_to_cpu(dtd_timings->usHSyncOffset); + mode->crtc_hsync_end = mode->crtc_hsync_start + + le16_to_cpu(dtd_timings->usHSyncWidth); + + mode->crtc_vtotal = le16_to_cpu(dtd_timings->usVActive) + + le16_to_cpu(dtd_timings->usVBlanking_Time); + mode->crtc_vdisplay = le16_to_cpu(dtd_timings->usVActive); + mode->crtc_vsync_start = le16_to_cpu(dtd_timings->usVActive) + + le16_to_cpu(dtd_timings->usVSyncOffset); + mode->crtc_vsync_end = mode->crtc_vsync_start + + le16_to_cpu(dtd_timings->usVSyncWidth); + + mode->flags = 0; + misc = le16_to_cpu(dtd_timings->susModeMiscInfo.usAccess); + if (misc & ATOM_VSYNC_POLARITY) + mode->flags |= DRM_MODE_FLAG_NVSYNC; + if (misc & ATOM_HSYNC_POLARITY) + mode->flags |= DRM_MODE_FLAG_NHSYNC; + if (misc & ATOM_COMPOSITESYNC) + mode->flags |= DRM_MODE_FLAG_CSYNC; + if (misc & ATOM_INTERLACE) + mode->flags |= DRM_MODE_FLAG_INTERLACE; + if (misc & ATOM_DOUBLE_CLOCK_MODE) + mode->flags |= DRM_MODE_FLAG_DBLSCAN; + + mode->clock = le16_to_cpu(dtd_timings->usPixClk) * 10; + break; + } + return true; +} + +enum radeon_tv_std +radeon_atombios_get_tv_info(struct radeon_device *rdev) +{ + struct radeon_mode_info *mode_info = &rdev->mode_info; + int index = GetIndexIntoMasterTable(DATA, AnalogTV_Info); + uint16_t data_offset; + uint8_t frev, crev; + struct _ATOM_ANALOG_TV_INFO *tv_info; + enum radeon_tv_std tv_std = TV_STD_NTSC; + + if (atom_parse_data_header(mode_info->atom_context, index, NULL, + &frev, &crev, &data_offset)) { + + tv_info = (struct _ATOM_ANALOG_TV_INFO *) + (mode_info->atom_context->bios + data_offset); + + switch (tv_info->ucTV_BootUpDefaultStandard) { + case ATOM_TV_NTSC: + tv_std = TV_STD_NTSC; + DRM_DEBUG_KMS("Default TV standard: NTSC\n"); + break; + case ATOM_TV_NTSCJ: + tv_std = TV_STD_NTSC_J; + DRM_DEBUG_KMS("Default TV standard: NTSC-J\n"); + break; + case ATOM_TV_PAL: + tv_std = TV_STD_PAL; + DRM_DEBUG_KMS("Default TV standard: PAL\n"); + break; + case ATOM_TV_PALM: + tv_std = TV_STD_PAL_M; + DRM_DEBUG_KMS("Default TV standard: PAL-M\n"); + break; + case ATOM_TV_PALN: + tv_std = TV_STD_PAL_N; + DRM_DEBUG_KMS("Default TV standard: PAL-N\n"); + break; + case ATOM_TV_PALCN: + tv_std = TV_STD_PAL_CN; + DRM_DEBUG_KMS("Default TV standard: PAL-CN\n"); + break; + case ATOM_TV_PAL60: + tv_std = TV_STD_PAL_60; + DRM_DEBUG_KMS("Default TV standard: PAL-60\n"); + break; + case ATOM_TV_SECAM: + tv_std = TV_STD_SECAM; + DRM_DEBUG_KMS("Default TV standard: SECAM\n"); + break; + default: + tv_std = TV_STD_NTSC; + DRM_DEBUG_KMS("Unknown TV standard; defaulting to NTSC\n"); + break; + } + } + return tv_std; +} + +struct radeon_encoder_tv_dac * +radeon_atombios_get_tv_dac_info(struct radeon_encoder *encoder) +{ + struct drm_device *dev = encoder->base.dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_mode_info *mode_info = &rdev->mode_info; + int index = GetIndexIntoMasterTable(DATA, CompassionateData); + uint16_t data_offset; + struct _COMPASSIONATE_DATA *dac_info; + uint8_t frev, crev; + uint8_t bg, dac; + struct radeon_encoder_tv_dac *tv_dac = NULL; + + if (atom_parse_data_header(mode_info->atom_context, index, NULL, + &frev, &crev, &data_offset)) { + + dac_info = (struct _COMPASSIONATE_DATA *) + (mode_info->atom_context->bios + data_offset); + + tv_dac = kzalloc(sizeof(struct radeon_encoder_tv_dac), GFP_KERNEL); + + if (!tv_dac) + return NULL; + + bg = dac_info->ucDAC2_CRT2_BG_Adjustment; + dac = dac_info->ucDAC2_CRT2_DAC_Adjustment; + tv_dac->ps2_tvdac_adj = (bg << 16) | (dac << 20); + + bg = dac_info->ucDAC2_PAL_BG_Adjustment; + dac = dac_info->ucDAC2_PAL_DAC_Adjustment; + tv_dac->pal_tvdac_adj = (bg << 16) | (dac << 20); + + bg = dac_info->ucDAC2_NTSC_BG_Adjustment; + dac = dac_info->ucDAC2_NTSC_DAC_Adjustment; + tv_dac->ntsc_tvdac_adj = (bg << 16) | (dac << 20); + + tv_dac->tv_std = radeon_atombios_get_tv_info(rdev); + } + return tv_dac; +} + +static const char *thermal_controller_names[] = { + "NONE", + "lm63", + "adm1032", + "adm1030", + "max6649", + "lm64", + "f75375", + "asc7xxx", +}; + +static const char *pp_lib_thermal_controller_names[] = { + "NONE", + "lm63", + "adm1032", + "adm1030", + "max6649", + "lm64", + "f75375", + "RV6xx", + "RV770", + "adt7473", + "NONE", + "External GPIO", + "Evergreen", + "emc2103", + "Sumo", + "Northern Islands", +}; + +union power_info { + struct _ATOM_POWERPLAY_INFO info; + struct _ATOM_POWERPLAY_INFO_V2 info_2; + struct _ATOM_POWERPLAY_INFO_V3 info_3; + struct _ATOM_PPLIB_POWERPLAYTABLE pplib; + struct _ATOM_PPLIB_POWERPLAYTABLE2 pplib2; + struct _ATOM_PPLIB_POWERPLAYTABLE3 pplib3; +}; + +union pplib_clock_info { + struct _ATOM_PPLIB_R600_CLOCK_INFO r600; + struct _ATOM_PPLIB_RS780_CLOCK_INFO rs780; + struct _ATOM_PPLIB_EVERGREEN_CLOCK_INFO evergreen; + struct _ATOM_PPLIB_SUMO_CLOCK_INFO sumo; +}; + +union pplib_power_state { + struct _ATOM_PPLIB_STATE v1; + struct _ATOM_PPLIB_STATE_V2 v2; +}; + +static void radeon_atombios_parse_misc_flags_1_3(struct radeon_device *rdev, + int state_index, + u32 misc, u32 misc2) +{ + rdev->pm.power_state[state_index].misc = misc; + rdev->pm.power_state[state_index].misc2 = misc2; + /* order matters! */ + if (misc & ATOM_PM_MISCINFO_POWER_SAVING_MODE) + rdev->pm.power_state[state_index].type = + POWER_STATE_TYPE_POWERSAVE; + if (misc & ATOM_PM_MISCINFO_DEFAULT_DC_STATE_ENTRY_TRUE) + rdev->pm.power_state[state_index].type = + POWER_STATE_TYPE_BATTERY; + if (misc & ATOM_PM_MISCINFO_DEFAULT_LOW_DC_STATE_ENTRY_TRUE) + rdev->pm.power_state[state_index].type = + POWER_STATE_TYPE_BATTERY; + if (misc & ATOM_PM_MISCINFO_LOAD_BALANCE_EN) + rdev->pm.power_state[state_index].type = + POWER_STATE_TYPE_BALANCED; + if (misc & ATOM_PM_MISCINFO_3D_ACCELERATION_EN) { + rdev->pm.power_state[state_index].type = + POWER_STATE_TYPE_PERFORMANCE; + rdev->pm.power_state[state_index].flags &= + ~RADEON_PM_STATE_SINGLE_DISPLAY_ONLY; + } + if (misc2 & ATOM_PM_MISCINFO2_SYSTEM_AC_LITE_MODE) + rdev->pm.power_state[state_index].type = + POWER_STATE_TYPE_BALANCED; + if (misc & ATOM_PM_MISCINFO_DRIVER_DEFAULT_MODE) { + rdev->pm.power_state[state_index].type = + POWER_STATE_TYPE_DEFAULT; + rdev->pm.default_power_state_index = state_index; + rdev->pm.power_state[state_index].default_clock_mode = + &rdev->pm.power_state[state_index].clock_info[0]; + } else if (state_index == 0) { + rdev->pm.power_state[state_index].clock_info[0].flags |= + RADEON_PM_MODE_NO_DISPLAY; + } +} + +static int radeon_atombios_parse_power_table_1_3(struct radeon_device *rdev) +{ + struct radeon_mode_info *mode_info = &rdev->mode_info; + u32 misc, misc2 = 0; + int num_modes = 0, i; + int state_index = 0; + struct radeon_i2c_bus_rec i2c_bus; + union power_info *power_info; + int index = GetIndexIntoMasterTable(DATA, PowerPlayInfo); + u16 data_offset; + u8 frev, crev; + + if (!atom_parse_data_header(mode_info->atom_context, index, NULL, + &frev, &crev, &data_offset)) + return state_index; + power_info = (union power_info *)(mode_info->atom_context->bios + data_offset); + + /* add the i2c bus for thermal/fan chip */ + if (power_info->info.ucOverdriveThermalController > 0) { + DRM_INFO("Possible %s thermal controller at 0x%02x\n", + thermal_controller_names[power_info->info.ucOverdriveThermalController], + power_info->info.ucOverdriveControllerAddress >> 1); + i2c_bus = radeon_lookup_i2c_gpio(rdev, power_info->info.ucOverdriveI2cLine); + rdev->pm.i2c_bus = radeon_i2c_lookup(rdev, &i2c_bus); + if (rdev->pm.i2c_bus) { + struct i2c_board_info info = { }; + const char *name = thermal_controller_names[power_info->info. + ucOverdriveThermalController]; + info.addr = power_info->info.ucOverdriveControllerAddress >> 1; + strlcpy(info.type, name, sizeof(info.type)); + i2c_new_device(&rdev->pm.i2c_bus->adapter, &info); + } + } + num_modes = power_info->info.ucNumOfPowerModeEntries; + if (num_modes > ATOM_MAX_NUMBEROF_POWER_BLOCK) + num_modes = ATOM_MAX_NUMBEROF_POWER_BLOCK; + rdev->pm.power_state = kzalloc(sizeof(struct radeon_power_state) * num_modes, GFP_KERNEL); + if (!rdev->pm.power_state) + return state_index; + /* last mode is usually default, array is low to high */ + for (i = 0; i < num_modes; i++) { + rdev->pm.power_state[state_index].clock_info[0].voltage.type = VOLTAGE_NONE; + switch (frev) { + case 1: + rdev->pm.power_state[state_index].num_clock_modes = 1; + rdev->pm.power_state[state_index].clock_info[0].mclk = + le16_to_cpu(power_info->info.asPowerPlayInfo[i].usMemoryClock); + rdev->pm.power_state[state_index].clock_info[0].sclk = + le16_to_cpu(power_info->info.asPowerPlayInfo[i].usEngineClock); + /* skip invalid modes */ + if ((rdev->pm.power_state[state_index].clock_info[0].mclk == 0) || + (rdev->pm.power_state[state_index].clock_info[0].sclk == 0)) + continue; + rdev->pm.power_state[state_index].pcie_lanes = + power_info->info.asPowerPlayInfo[i].ucNumPciELanes; + misc = le32_to_cpu(power_info->info.asPowerPlayInfo[i].ulMiscInfo); + if ((misc & ATOM_PM_MISCINFO_VOLTAGE_DROP_SUPPORT) || + (misc & ATOM_PM_MISCINFO_VOLTAGE_DROP_ACTIVE_HIGH)) { + rdev->pm.power_state[state_index].clock_info[0].voltage.type = + VOLTAGE_GPIO; + rdev->pm.power_state[state_index].clock_info[0].voltage.gpio = + radeon_lookup_gpio(rdev, + power_info->info.asPowerPlayInfo[i].ucVoltageDropIndex); + if (misc & ATOM_PM_MISCINFO_VOLTAGE_DROP_ACTIVE_HIGH) + rdev->pm.power_state[state_index].clock_info[0].voltage.active_high = + true; + else + rdev->pm.power_state[state_index].clock_info[0].voltage.active_high = + false; + } else if (misc & ATOM_PM_MISCINFO_PROGRAM_VOLTAGE) { + rdev->pm.power_state[state_index].clock_info[0].voltage.type = + VOLTAGE_VDDC; + rdev->pm.power_state[state_index].clock_info[0].voltage.vddc_id = + power_info->info.asPowerPlayInfo[i].ucVoltageDropIndex; + } + rdev->pm.power_state[state_index].flags = RADEON_PM_STATE_SINGLE_DISPLAY_ONLY; + radeon_atombios_parse_misc_flags_1_3(rdev, state_index, misc, 0); + state_index++; + break; + case 2: + rdev->pm.power_state[state_index].num_clock_modes = 1; + rdev->pm.power_state[state_index].clock_info[0].mclk = + le32_to_cpu(power_info->info_2.asPowerPlayInfo[i].ulMemoryClock); + rdev->pm.power_state[state_index].clock_info[0].sclk = + le32_to_cpu(power_info->info_2.asPowerPlayInfo[i].ulEngineClock); + /* skip invalid modes */ + if ((rdev->pm.power_state[state_index].clock_info[0].mclk == 0) || + (rdev->pm.power_state[state_index].clock_info[0].sclk == 0)) + continue; + rdev->pm.power_state[state_index].pcie_lanes = + power_info->info_2.asPowerPlayInfo[i].ucNumPciELanes; + misc = le32_to_cpu(power_info->info_2.asPowerPlayInfo[i].ulMiscInfo); + misc2 = le32_to_cpu(power_info->info_2.asPowerPlayInfo[i].ulMiscInfo2); + if ((misc & ATOM_PM_MISCINFO_VOLTAGE_DROP_SUPPORT) || + (misc & ATOM_PM_MISCINFO_VOLTAGE_DROP_ACTIVE_HIGH)) { + rdev->pm.power_state[state_index].clock_info[0].voltage.type = + VOLTAGE_GPIO; + rdev->pm.power_state[state_index].clock_info[0].voltage.gpio = + radeon_lookup_gpio(rdev, + power_info->info_2.asPowerPlayInfo[i].ucVoltageDropIndex); + if (misc & ATOM_PM_MISCINFO_VOLTAGE_DROP_ACTIVE_HIGH) + rdev->pm.power_state[state_index].clock_info[0].voltage.active_high = + true; + else + rdev->pm.power_state[state_index].clock_info[0].voltage.active_high = + false; + } else if (misc & ATOM_PM_MISCINFO_PROGRAM_VOLTAGE) { + rdev->pm.power_state[state_index].clock_info[0].voltage.type = + VOLTAGE_VDDC; + rdev->pm.power_state[state_index].clock_info[0].voltage.vddc_id = + power_info->info_2.asPowerPlayInfo[i].ucVoltageDropIndex; + } + rdev->pm.power_state[state_index].flags = RADEON_PM_STATE_SINGLE_DISPLAY_ONLY; + radeon_atombios_parse_misc_flags_1_3(rdev, state_index, misc, misc2); + state_index++; + break; + case 3: + rdev->pm.power_state[state_index].num_clock_modes = 1; + rdev->pm.power_state[state_index].clock_info[0].mclk = + le32_to_cpu(power_info->info_3.asPowerPlayInfo[i].ulMemoryClock); + rdev->pm.power_state[state_index].clock_info[0].sclk = + le32_to_cpu(power_info->info_3.asPowerPlayInfo[i].ulEngineClock); + /* skip invalid modes */ + if ((rdev->pm.power_state[state_index].clock_info[0].mclk == 0) || + (rdev->pm.power_state[state_index].clock_info[0].sclk == 0)) + continue; + rdev->pm.power_state[state_index].pcie_lanes = + power_info->info_3.asPowerPlayInfo[i].ucNumPciELanes; + misc = le32_to_cpu(power_info->info_3.asPowerPlayInfo[i].ulMiscInfo); + misc2 = le32_to_cpu(power_info->info_3.asPowerPlayInfo[i].ulMiscInfo2); + if ((misc & ATOM_PM_MISCINFO_VOLTAGE_DROP_SUPPORT) || + (misc & ATOM_PM_MISCINFO_VOLTAGE_DROP_ACTIVE_HIGH)) { + rdev->pm.power_state[state_index].clock_info[0].voltage.type = + VOLTAGE_GPIO; + rdev->pm.power_state[state_index].clock_info[0].voltage.gpio = + radeon_lookup_gpio(rdev, + power_info->info_3.asPowerPlayInfo[i].ucVoltageDropIndex); + if (misc & ATOM_PM_MISCINFO_VOLTAGE_DROP_ACTIVE_HIGH) + rdev->pm.power_state[state_index].clock_info[0].voltage.active_high = + true; + else + rdev->pm.power_state[state_index].clock_info[0].voltage.active_high = + false; + } else if (misc & ATOM_PM_MISCINFO_PROGRAM_VOLTAGE) { + rdev->pm.power_state[state_index].clock_info[0].voltage.type = + VOLTAGE_VDDC; + rdev->pm.power_state[state_index].clock_info[0].voltage.vddc_id = + power_info->info_3.asPowerPlayInfo[i].ucVoltageDropIndex; + if (misc2 & ATOM_PM_MISCINFO2_VDDCI_DYNAMIC_VOLTAGE_EN) { + rdev->pm.power_state[state_index].clock_info[0].voltage.vddci_enabled = + true; + rdev->pm.power_state[state_index].clock_info[0].voltage.vddci_id = + power_info->info_3.asPowerPlayInfo[i].ucVDDCI_VoltageDropIndex; + } + } + rdev->pm.power_state[state_index].flags = RADEON_PM_STATE_SINGLE_DISPLAY_ONLY; + radeon_atombios_parse_misc_flags_1_3(rdev, state_index, misc, misc2); + state_index++; + break; + } + } + /* last mode is usually default */ + if (rdev->pm.default_power_state_index == -1) { + rdev->pm.power_state[state_index - 1].type = + POWER_STATE_TYPE_DEFAULT; + rdev->pm.default_power_state_index = state_index - 1; + rdev->pm.power_state[state_index - 1].default_clock_mode = + &rdev->pm.power_state[state_index - 1].clock_info[0]; + rdev->pm.power_state[state_index].flags &= + ~RADEON_PM_STATE_SINGLE_DISPLAY_ONLY; + rdev->pm.power_state[state_index].misc = 0; + rdev->pm.power_state[state_index].misc2 = 0; + } + return state_index; +} + +static void radeon_atombios_add_pplib_thermal_controller(struct radeon_device *rdev, + ATOM_PPLIB_THERMALCONTROLLER *controller) +{ + struct radeon_i2c_bus_rec i2c_bus; + + /* add the i2c bus for thermal/fan chip */ + if (controller->ucType > 0) { + if (controller->ucType == ATOM_PP_THERMALCONTROLLER_RV6xx) { + DRM_INFO("Internal thermal controller %s fan control\n", + (controller->ucFanParameters & + ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with"); + rdev->pm.int_thermal_type = THERMAL_TYPE_RV6XX; + } else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_RV770) { + DRM_INFO("Internal thermal controller %s fan control\n", + (controller->ucFanParameters & + ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with"); + rdev->pm.int_thermal_type = THERMAL_TYPE_RV770; + } else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_EVERGREEN) { + DRM_INFO("Internal thermal controller %s fan control\n", + (controller->ucFanParameters & + ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with"); + rdev->pm.int_thermal_type = THERMAL_TYPE_EVERGREEN; + } else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_SUMO) { + DRM_INFO("Internal thermal controller %s fan control\n", + (controller->ucFanParameters & + ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with"); + rdev->pm.int_thermal_type = THERMAL_TYPE_SUMO; + } else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_NISLANDS) { + DRM_INFO("Internal thermal controller %s fan control\n", + (controller->ucFanParameters & + ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with"); + rdev->pm.int_thermal_type = THERMAL_TYPE_NI; + } else if ((controller->ucType == + ATOM_PP_THERMALCONTROLLER_EXTERNAL_GPIO) || + (controller->ucType == + ATOM_PP_THERMALCONTROLLER_ADT7473_WITH_INTERNAL) || + (controller->ucType == + ATOM_PP_THERMALCONTROLLER_EMC2103_WITH_INTERNAL)) { + DRM_INFO("Special thermal controller config\n"); + } else { + DRM_INFO("Possible %s thermal controller at 0x%02x %s fan control\n", + pp_lib_thermal_controller_names[controller->ucType], + controller->ucI2cAddress >> 1, + (controller->ucFanParameters & + ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with"); + i2c_bus = radeon_lookup_i2c_gpio(rdev, controller->ucI2cLine); + rdev->pm.i2c_bus = radeon_i2c_lookup(rdev, &i2c_bus); + if (rdev->pm.i2c_bus) { + struct i2c_board_info info = { }; + const char *name = pp_lib_thermal_controller_names[controller->ucType]; + info.addr = controller->ucI2cAddress >> 1; + strlcpy(info.type, name, sizeof(info.type)); + i2c_new_device(&rdev->pm.i2c_bus->adapter, &info); + } + } + } +} + +static void radeon_atombios_get_default_voltages(struct radeon_device *rdev, + u16 *vddc, u16 *vddci) +{ + struct radeon_mode_info *mode_info = &rdev->mode_info; + int index = GetIndexIntoMasterTable(DATA, FirmwareInfo); + u8 frev, crev; + u16 data_offset; + union firmware_info *firmware_info; + + *vddc = 0; + *vddci = 0; + + if (atom_parse_data_header(mode_info->atom_context, index, NULL, + &frev, &crev, &data_offset)) { + firmware_info = + (union firmware_info *)(mode_info->atom_context->bios + + data_offset); + *vddc = le16_to_cpu(firmware_info->info_14.usBootUpVDDCVoltage); + if ((frev == 2) && (crev >= 2)) + *vddci = le16_to_cpu(firmware_info->info_22.usBootUpVDDCIVoltage); + } +} + +static void radeon_atombios_parse_pplib_non_clock_info(struct radeon_device *rdev, + int state_index, int mode_index, + struct _ATOM_PPLIB_NONCLOCK_INFO *non_clock_info) +{ + int j; + u32 misc = le32_to_cpu(non_clock_info->ulCapsAndSettings); + u32 misc2 = le16_to_cpu(non_clock_info->usClassification); + u16 vddc, vddci; + + radeon_atombios_get_default_voltages(rdev, &vddc, &vddci); + + rdev->pm.power_state[state_index].misc = misc; + rdev->pm.power_state[state_index].misc2 = misc2; + rdev->pm.power_state[state_index].pcie_lanes = + ((misc & ATOM_PPLIB_PCIE_LINK_WIDTH_MASK) >> + ATOM_PPLIB_PCIE_LINK_WIDTH_SHIFT) + 1; + switch (misc2 & ATOM_PPLIB_CLASSIFICATION_UI_MASK) { + case ATOM_PPLIB_CLASSIFICATION_UI_BATTERY: + rdev->pm.power_state[state_index].type = + POWER_STATE_TYPE_BATTERY; + break; + case ATOM_PPLIB_CLASSIFICATION_UI_BALANCED: + rdev->pm.power_state[state_index].type = + POWER_STATE_TYPE_BALANCED; + break; + case ATOM_PPLIB_CLASSIFICATION_UI_PERFORMANCE: + rdev->pm.power_state[state_index].type = + POWER_STATE_TYPE_PERFORMANCE; + break; + case ATOM_PPLIB_CLASSIFICATION_UI_NONE: + if (misc2 & ATOM_PPLIB_CLASSIFICATION_3DPERFORMANCE) + rdev->pm.power_state[state_index].type = + POWER_STATE_TYPE_PERFORMANCE; + break; + } + rdev->pm.power_state[state_index].flags = 0; + if (misc & ATOM_PPLIB_SINGLE_DISPLAY_ONLY) + rdev->pm.power_state[state_index].flags |= + RADEON_PM_STATE_SINGLE_DISPLAY_ONLY; + if (misc2 & ATOM_PPLIB_CLASSIFICATION_BOOT) { + rdev->pm.power_state[state_index].type = + POWER_STATE_TYPE_DEFAULT; + rdev->pm.default_power_state_index = state_index; + rdev->pm.power_state[state_index].default_clock_mode = + &rdev->pm.power_state[state_index].clock_info[mode_index - 1]; + if (ASIC_IS_DCE5(rdev)) { + /* NI chips post without MC ucode, so default clocks are strobe mode only */ + rdev->pm.default_sclk = rdev->pm.power_state[state_index].clock_info[0].sclk; + rdev->pm.default_mclk = rdev->pm.power_state[state_index].clock_info[0].mclk; + rdev->pm.default_vddc = rdev->pm.power_state[state_index].clock_info[0].voltage.voltage; + rdev->pm.default_vddci = rdev->pm.power_state[state_index].clock_info[0].voltage.vddci; + } else { + /* patch the table values with the default slck/mclk from firmware info */ + for (j = 0; j < mode_index; j++) { + rdev->pm.power_state[state_index].clock_info[j].mclk = + rdev->clock.default_mclk; + rdev->pm.power_state[state_index].clock_info[j].sclk = + rdev->clock.default_sclk; + if (vddc) + rdev->pm.power_state[state_index].clock_info[j].voltage.voltage = + vddc; + } + } + } +} + +static bool radeon_atombios_parse_pplib_clock_info(struct radeon_device *rdev, + int state_index, int mode_index, + union pplib_clock_info *clock_info) +{ + u32 sclk, mclk; + + if (rdev->flags & RADEON_IS_IGP) { + if (rdev->family >= CHIP_PALM) { + sclk = le16_to_cpu(clock_info->sumo.usEngineClockLow); + sclk |= clock_info->sumo.ucEngineClockHigh << 16; + rdev->pm.power_state[state_index].clock_info[mode_index].sclk = sclk; + } else { + sclk = le16_to_cpu(clock_info->rs780.usLowEngineClockLow); + sclk |= clock_info->rs780.ucLowEngineClockHigh << 16; + rdev->pm.power_state[state_index].clock_info[mode_index].sclk = sclk; + } + } else if (ASIC_IS_DCE4(rdev)) { + sclk = le16_to_cpu(clock_info->evergreen.usEngineClockLow); + sclk |= clock_info->evergreen.ucEngineClockHigh << 16; + mclk = le16_to_cpu(clock_info->evergreen.usMemoryClockLow); + mclk |= clock_info->evergreen.ucMemoryClockHigh << 16; + rdev->pm.power_state[state_index].clock_info[mode_index].mclk = mclk; + rdev->pm.power_state[state_index].clock_info[mode_index].sclk = sclk; + rdev->pm.power_state[state_index].clock_info[mode_index].voltage.type = + VOLTAGE_SW; + rdev->pm.power_state[state_index].clock_info[mode_index].voltage.voltage = + le16_to_cpu(clock_info->evergreen.usVDDC); + rdev->pm.power_state[state_index].clock_info[mode_index].voltage.vddci = + le16_to_cpu(clock_info->evergreen.usVDDCI); + } else { + sclk = le16_to_cpu(clock_info->r600.usEngineClockLow); + sclk |= clock_info->r600.ucEngineClockHigh << 16; + mclk = le16_to_cpu(clock_info->r600.usMemoryClockLow); + mclk |= clock_info->r600.ucMemoryClockHigh << 16; + rdev->pm.power_state[state_index].clock_info[mode_index].mclk = mclk; + rdev->pm.power_state[state_index].clock_info[mode_index].sclk = sclk; + rdev->pm.power_state[state_index].clock_info[mode_index].voltage.type = + VOLTAGE_SW; + rdev->pm.power_state[state_index].clock_info[mode_index].voltage.voltage = + le16_to_cpu(clock_info->r600.usVDDC); + } + + /* patch up vddc if necessary */ + if (rdev->pm.power_state[state_index].clock_info[mode_index].voltage.voltage == 0xff01) { + u16 vddc; + + if (radeon_atom_get_max_vddc(rdev, &vddc) == 0) + rdev->pm.power_state[state_index].clock_info[mode_index].voltage.voltage = vddc; + } + + if (rdev->flags & RADEON_IS_IGP) { + /* skip invalid modes */ + if (rdev->pm.power_state[state_index].clock_info[mode_index].sclk == 0) + return false; + } else { + /* skip invalid modes */ + if ((rdev->pm.power_state[state_index].clock_info[mode_index].mclk == 0) || + (rdev->pm.power_state[state_index].clock_info[mode_index].sclk == 0)) + return false; + } + return true; +} + +static int radeon_atombios_parse_power_table_4_5(struct radeon_device *rdev) +{ + struct radeon_mode_info *mode_info = &rdev->mode_info; + struct _ATOM_PPLIB_NONCLOCK_INFO *non_clock_info; + union pplib_power_state *power_state; + int i, j; + int state_index = 0, mode_index = 0; + union pplib_clock_info *clock_info; + bool valid; + union power_info *power_info; + int index = GetIndexIntoMasterTable(DATA, PowerPlayInfo); + u16 data_offset; + u8 frev, crev; + + if (!atom_parse_data_header(mode_info->atom_context, index, NULL, + &frev, &crev, &data_offset)) + return state_index; + power_info = (union power_info *)(mode_info->atom_context->bios + data_offset); + + radeon_atombios_add_pplib_thermal_controller(rdev, &power_info->pplib.sThermalController); + rdev->pm.power_state = kzalloc(sizeof(struct radeon_power_state) * + power_info->pplib.ucNumStates, GFP_KERNEL); + if (!rdev->pm.power_state) + return state_index; + /* first mode is usually default, followed by low to high */ + for (i = 0; i < power_info->pplib.ucNumStates; i++) { + mode_index = 0; + power_state = (union pplib_power_state *) + (mode_info->atom_context->bios + data_offset + + le16_to_cpu(power_info->pplib.usStateArrayOffset) + + i * power_info->pplib.ucStateEntrySize); + non_clock_info = (struct _ATOM_PPLIB_NONCLOCK_INFO *) + (mode_info->atom_context->bios + data_offset + + le16_to_cpu(power_info->pplib.usNonClockInfoArrayOffset) + + (power_state->v1.ucNonClockStateIndex * + power_info->pplib.ucNonClockSize)); + for (j = 0; j < (power_info->pplib.ucStateEntrySize - 1); j++) { + clock_info = (union pplib_clock_info *) + (mode_info->atom_context->bios + data_offset + + le16_to_cpu(power_info->pplib.usClockInfoArrayOffset) + + (power_state->v1.ucClockStateIndices[j] * + power_info->pplib.ucClockInfoSize)); + valid = radeon_atombios_parse_pplib_clock_info(rdev, + state_index, mode_index, + clock_info); + if (valid) + mode_index++; + } + rdev->pm.power_state[state_index].num_clock_modes = mode_index; + if (mode_index) { + radeon_atombios_parse_pplib_non_clock_info(rdev, state_index, mode_index, + non_clock_info); + state_index++; + } + } + /* if multiple clock modes, mark the lowest as no display */ + for (i = 0; i < state_index; i++) { + if (rdev->pm.power_state[i].num_clock_modes > 1) + rdev->pm.power_state[i].clock_info[0].flags |= + RADEON_PM_MODE_NO_DISPLAY; + } + /* first mode is usually default */ + if (rdev->pm.default_power_state_index == -1) { + rdev->pm.power_state[0].type = + POWER_STATE_TYPE_DEFAULT; + rdev->pm.default_power_state_index = 0; + rdev->pm.power_state[0].default_clock_mode = + &rdev->pm.power_state[0].clock_info[0]; + } + return state_index; +} + +static int radeon_atombios_parse_power_table_6(struct radeon_device *rdev) +{ + struct radeon_mode_info *mode_info = &rdev->mode_info; + struct _ATOM_PPLIB_NONCLOCK_INFO *non_clock_info; + union pplib_power_state *power_state; + int i, j, non_clock_array_index, clock_array_index; + int state_index = 0, mode_index = 0; + union pplib_clock_info *clock_info; + struct StateArray *state_array; + struct ClockInfoArray *clock_info_array; + struct NonClockInfoArray *non_clock_info_array; + bool valid; + union power_info *power_info; + int index = GetIndexIntoMasterTable(DATA, PowerPlayInfo); + u16 data_offset; + u8 frev, crev; + + if (!atom_parse_data_header(mode_info->atom_context, index, NULL, + &frev, &crev, &data_offset)) + return state_index; + power_info = (union power_info *)(mode_info->atom_context->bios + data_offset); + + radeon_atombios_add_pplib_thermal_controller(rdev, &power_info->pplib.sThermalController); + state_array = (struct StateArray *) + (mode_info->atom_context->bios + data_offset + + le16_to_cpu(power_info->pplib.usStateArrayOffset)); + clock_info_array = (struct ClockInfoArray *) + (mode_info->atom_context->bios + data_offset + + le16_to_cpu(power_info->pplib.usClockInfoArrayOffset)); + non_clock_info_array = (struct NonClockInfoArray *) + (mode_info->atom_context->bios + data_offset + + le16_to_cpu(power_info->pplib.usNonClockInfoArrayOffset)); + rdev->pm.power_state = kzalloc(sizeof(struct radeon_power_state) * + state_array->ucNumEntries, GFP_KERNEL); + if (!rdev->pm.power_state) + return state_index; + for (i = 0; i < state_array->ucNumEntries; i++) { + mode_index = 0; + power_state = (union pplib_power_state *)&state_array->states[i]; + /* XXX this might be an inagua bug... */ + non_clock_array_index = i; /* power_state->v2.nonClockInfoIndex */ + non_clock_info = (struct _ATOM_PPLIB_NONCLOCK_INFO *) + &non_clock_info_array->nonClockInfo[non_clock_array_index]; + for (j = 0; j < power_state->v2.ucNumDPMLevels; j++) { + clock_array_index = power_state->v2.clockInfoIndex[j]; + /* XXX this might be an inagua bug... */ + if (clock_array_index >= clock_info_array->ucNumEntries) + continue; + clock_info = (union pplib_clock_info *) + &clock_info_array->clockInfo[clock_array_index]; + valid = radeon_atombios_parse_pplib_clock_info(rdev, + state_index, mode_index, + clock_info); + if (valid) + mode_index++; + } + rdev->pm.power_state[state_index].num_clock_modes = mode_index; + if (mode_index) { + radeon_atombios_parse_pplib_non_clock_info(rdev, state_index, mode_index, + non_clock_info); + state_index++; + } + } + /* if multiple clock modes, mark the lowest as no display */ + for (i = 0; i < state_index; i++) { + if (rdev->pm.power_state[i].num_clock_modes > 1) + rdev->pm.power_state[i].clock_info[0].flags |= + RADEON_PM_MODE_NO_DISPLAY; + } + /* first mode is usually default */ + if (rdev->pm.default_power_state_index == -1) { + rdev->pm.power_state[0].type = + POWER_STATE_TYPE_DEFAULT; + rdev->pm.default_power_state_index = 0; + rdev->pm.power_state[0].default_clock_mode = + &rdev->pm.power_state[0].clock_info[0]; + } + return state_index; +} + +void radeon_atombios_get_power_modes(struct radeon_device *rdev) +{ + struct radeon_mode_info *mode_info = &rdev->mode_info; + int index = GetIndexIntoMasterTable(DATA, PowerPlayInfo); + u16 data_offset; + u8 frev, crev; + int state_index = 0; + + rdev->pm.default_power_state_index = -1; + + if (atom_parse_data_header(mode_info->atom_context, index, NULL, + &frev, &crev, &data_offset)) { + switch (frev) { + case 1: + case 2: + case 3: + state_index = radeon_atombios_parse_power_table_1_3(rdev); + break; + case 4: + case 5: + state_index = radeon_atombios_parse_power_table_4_5(rdev); + break; + case 6: + state_index = radeon_atombios_parse_power_table_6(rdev); + break; + default: + break; + } + } else { + rdev->pm.power_state = kzalloc(sizeof(struct radeon_power_state), GFP_KERNEL); + if (rdev->pm.power_state) { + /* add the default mode */ + rdev->pm.power_state[state_index].type = + POWER_STATE_TYPE_DEFAULT; + rdev->pm.power_state[state_index].num_clock_modes = 1; + rdev->pm.power_state[state_index].clock_info[0].mclk = rdev->clock.default_mclk; + rdev->pm.power_state[state_index].clock_info[0].sclk = rdev->clock.default_sclk; + rdev->pm.power_state[state_index].default_clock_mode = + &rdev->pm.power_state[state_index].clock_info[0]; + rdev->pm.power_state[state_index].clock_info[0].voltage.type = VOLTAGE_NONE; + rdev->pm.power_state[state_index].pcie_lanes = 16; + rdev->pm.default_power_state_index = state_index; + rdev->pm.power_state[state_index].flags = 0; + state_index++; + } + } + + rdev->pm.num_power_states = state_index; + + rdev->pm.current_power_state_index = rdev->pm.default_power_state_index; + rdev->pm.current_clock_mode_index = 0; + if (rdev->pm.default_power_state_index >= 0) + rdev->pm.current_vddc = + rdev->pm.power_state[rdev->pm.default_power_state_index].clock_info[0].voltage.voltage; + else + rdev->pm.current_vddc = 0; +} + +void radeon_atom_set_clock_gating(struct radeon_device *rdev, int enable) +{ + DYNAMIC_CLOCK_GATING_PS_ALLOCATION args; + int index = GetIndexIntoMasterTable(COMMAND, DynamicClockGating); + + args.ucEnable = enable; + + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); +} + +uint32_t radeon_atom_get_engine_clock(struct radeon_device *rdev) +{ + GET_ENGINE_CLOCK_PS_ALLOCATION args; + int index = GetIndexIntoMasterTable(COMMAND, GetEngineClock); + + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); + return le32_to_cpu(args.ulReturnEngineClock); +} + +uint32_t radeon_atom_get_memory_clock(struct radeon_device *rdev) +{ + GET_MEMORY_CLOCK_PS_ALLOCATION args; + int index = GetIndexIntoMasterTable(COMMAND, GetMemoryClock); + + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); + return le32_to_cpu(args.ulReturnMemoryClock); +} + +void radeon_atom_set_engine_clock(struct radeon_device *rdev, + uint32_t eng_clock) +{ + SET_ENGINE_CLOCK_PS_ALLOCATION args; + int index = GetIndexIntoMasterTable(COMMAND, SetEngineClock); + + args.ulTargetEngineClock = cpu_to_le32(eng_clock); /* 10 khz */ + + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); +} + +void radeon_atom_set_memory_clock(struct radeon_device *rdev, + uint32_t mem_clock) +{ + SET_MEMORY_CLOCK_PS_ALLOCATION args; + int index = GetIndexIntoMasterTable(COMMAND, SetMemoryClock); + + if (rdev->flags & RADEON_IS_IGP) + return; + + args.ulTargetMemoryClock = cpu_to_le32(mem_clock); /* 10 khz */ + + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); +} + +union set_voltage { + struct _SET_VOLTAGE_PS_ALLOCATION alloc; + struct _SET_VOLTAGE_PARAMETERS v1; + struct _SET_VOLTAGE_PARAMETERS_V2 v2; +}; + +void radeon_atom_set_voltage(struct radeon_device *rdev, u16 voltage_level, u8 voltage_type) +{ + union set_voltage args; + int index = GetIndexIntoMasterTable(COMMAND, SetVoltage); + u8 frev, crev, volt_index = voltage_level; + + if (!atom_parse_cmd_header(rdev->mode_info.atom_context, index, &frev, &crev)) + return; + + /* 0xff01 is a flag rather then an actual voltage */ + if (voltage_level == 0xff01) + return; + + switch (crev) { + case 1: + args.v1.ucVoltageType = voltage_type; + args.v1.ucVoltageMode = SET_ASIC_VOLTAGE_MODE_ALL_SOURCE; + args.v1.ucVoltageIndex = volt_index; + break; + case 2: + args.v2.ucVoltageType = voltage_type; + args.v2.ucVoltageMode = SET_ASIC_VOLTAGE_MODE_SET_VOLTAGE; + args.v2.usVoltageLevel = cpu_to_le16(voltage_level); + break; + default: + DRM_ERROR("Unknown table version %d, %d\n", frev, crev); + return; + } + + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); +} + +int radeon_atom_get_max_vddc(struct radeon_device *rdev, + u16 *voltage) +{ + union set_voltage args; + int index = GetIndexIntoMasterTable(COMMAND, SetVoltage); + u8 frev, crev; + + if (!atom_parse_cmd_header(rdev->mode_info.atom_context, index, &frev, &crev)) + return -EINVAL; + + switch (crev) { + case 1: + return -EINVAL; + case 2: + args.v2.ucVoltageType = SET_VOLTAGE_GET_MAX_VOLTAGE; + args.v2.ucVoltageMode = 0; + args.v2.usVoltageLevel = 0; + + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); + + *voltage = le16_to_cpu(args.v2.usVoltageLevel); + break; + default: + DRM_ERROR("Unknown table version %d, %d\n", frev, crev); + return -EINVAL; + } + + return 0; +} + +void radeon_atom_initialize_bios_scratch_regs(struct drm_device *dev) +{ + struct radeon_device *rdev = dev->dev_private; + uint32_t bios_2_scratch, bios_6_scratch; + + if (rdev->family >= CHIP_R600) { + bios_2_scratch = RREG32(R600_BIOS_2_SCRATCH); + bios_6_scratch = RREG32(R600_BIOS_6_SCRATCH); + } else { + bios_2_scratch = RREG32(RADEON_BIOS_2_SCRATCH); + bios_6_scratch = RREG32(RADEON_BIOS_6_SCRATCH); + } + + /* let the bios control the backlight */ + bios_2_scratch &= ~ATOM_S2_VRI_BRIGHT_ENABLE; + + /* tell the bios not to handle mode switching */ + bios_6_scratch |= ATOM_S6_ACC_BLOCK_DISPLAY_SWITCH; + + if (rdev->family >= CHIP_R600) { + WREG32(R600_BIOS_2_SCRATCH, bios_2_scratch); + WREG32(R600_BIOS_6_SCRATCH, bios_6_scratch); + } else { + WREG32(RADEON_BIOS_2_SCRATCH, bios_2_scratch); + WREG32(RADEON_BIOS_6_SCRATCH, bios_6_scratch); + } + +} + +void radeon_save_bios_scratch_regs(struct radeon_device *rdev) +{ + uint32_t scratch_reg; + int i; + + if (rdev->family >= CHIP_R600) + scratch_reg = R600_BIOS_0_SCRATCH; + else + scratch_reg = RADEON_BIOS_0_SCRATCH; + + for (i = 0; i < RADEON_BIOS_NUM_SCRATCH; i++) + rdev->bios_scratch[i] = RREG32(scratch_reg + (i * 4)); +} + +void radeon_restore_bios_scratch_regs(struct radeon_device *rdev) +{ + uint32_t scratch_reg; + int i; + + if (rdev->family >= CHIP_R600) + scratch_reg = R600_BIOS_0_SCRATCH; + else + scratch_reg = RADEON_BIOS_0_SCRATCH; + + for (i = 0; i < RADEON_BIOS_NUM_SCRATCH; i++) + WREG32(scratch_reg + (i * 4), rdev->bios_scratch[i]); +} + +void radeon_atom_output_lock(struct drm_encoder *encoder, bool lock) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + uint32_t bios_6_scratch; + + if (rdev->family >= CHIP_R600) + bios_6_scratch = RREG32(R600_BIOS_6_SCRATCH); + else + bios_6_scratch = RREG32(RADEON_BIOS_6_SCRATCH); + + if (lock) { + bios_6_scratch |= ATOM_S6_CRITICAL_STATE; + bios_6_scratch &= ~ATOM_S6_ACC_MODE; + } else { + bios_6_scratch &= ~ATOM_S6_CRITICAL_STATE; + bios_6_scratch |= ATOM_S6_ACC_MODE; + } + + if (rdev->family >= CHIP_R600) + WREG32(R600_BIOS_6_SCRATCH, bios_6_scratch); + else + WREG32(RADEON_BIOS_6_SCRATCH, bios_6_scratch); +} + +/* at some point we may want to break this out into individual functions */ +void +radeon_atombios_connected_scratch_regs(struct drm_connector *connector, + struct drm_encoder *encoder, + bool connected) +{ + struct drm_device *dev = connector->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_connector *radeon_connector = + to_radeon_connector(connector); + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + uint32_t bios_0_scratch, bios_3_scratch, bios_6_scratch; + + if (rdev->family >= CHIP_R600) { + bios_0_scratch = RREG32(R600_BIOS_0_SCRATCH); + bios_3_scratch = RREG32(R600_BIOS_3_SCRATCH); + bios_6_scratch = RREG32(R600_BIOS_6_SCRATCH); + } else { + bios_0_scratch = RREG32(RADEON_BIOS_0_SCRATCH); + bios_3_scratch = RREG32(RADEON_BIOS_3_SCRATCH); + bios_6_scratch = RREG32(RADEON_BIOS_6_SCRATCH); + } + + if ((radeon_encoder->devices & ATOM_DEVICE_TV1_SUPPORT) && + (radeon_connector->devices & ATOM_DEVICE_TV1_SUPPORT)) { + if (connected) { + DRM_DEBUG_KMS("TV1 connected\n"); + bios_3_scratch |= ATOM_S3_TV1_ACTIVE; + bios_6_scratch |= ATOM_S6_ACC_REQ_TV1; + } else { + DRM_DEBUG_KMS("TV1 disconnected\n"); + bios_0_scratch &= ~ATOM_S0_TV1_MASK; + bios_3_scratch &= ~ATOM_S3_TV1_ACTIVE; + bios_6_scratch &= ~ATOM_S6_ACC_REQ_TV1; + } + } + if ((radeon_encoder->devices & ATOM_DEVICE_CV_SUPPORT) && + (radeon_connector->devices & ATOM_DEVICE_CV_SUPPORT)) { + if (connected) { + DRM_DEBUG_KMS("CV connected\n"); + bios_3_scratch |= ATOM_S3_CV_ACTIVE; + bios_6_scratch |= ATOM_S6_ACC_REQ_CV; + } else { + DRM_DEBUG_KMS("CV disconnected\n"); + bios_0_scratch &= ~ATOM_S0_CV_MASK; + bios_3_scratch &= ~ATOM_S3_CV_ACTIVE; + bios_6_scratch &= ~ATOM_S6_ACC_REQ_CV; + } + } + if ((radeon_encoder->devices & ATOM_DEVICE_LCD1_SUPPORT) && + (radeon_connector->devices & ATOM_DEVICE_LCD1_SUPPORT)) { + if (connected) { + DRM_DEBUG_KMS("LCD1 connected\n"); + bios_0_scratch |= ATOM_S0_LCD1; + bios_3_scratch |= ATOM_S3_LCD1_ACTIVE; + bios_6_scratch |= ATOM_S6_ACC_REQ_LCD1; + } else { + DRM_DEBUG_KMS("LCD1 disconnected\n"); + bios_0_scratch &= ~ATOM_S0_LCD1; + bios_3_scratch &= ~ATOM_S3_LCD1_ACTIVE; + bios_6_scratch &= ~ATOM_S6_ACC_REQ_LCD1; + } + } + if ((radeon_encoder->devices & ATOM_DEVICE_CRT1_SUPPORT) && + (radeon_connector->devices & ATOM_DEVICE_CRT1_SUPPORT)) { + if (connected) { + DRM_DEBUG_KMS("CRT1 connected\n"); + bios_0_scratch |= ATOM_S0_CRT1_COLOR; + bios_3_scratch |= ATOM_S3_CRT1_ACTIVE; + bios_6_scratch |= ATOM_S6_ACC_REQ_CRT1; + } else { + DRM_DEBUG_KMS("CRT1 disconnected\n"); + bios_0_scratch &= ~ATOM_S0_CRT1_MASK; + bios_3_scratch &= ~ATOM_S3_CRT1_ACTIVE; + bios_6_scratch &= ~ATOM_S6_ACC_REQ_CRT1; + } + } + if ((radeon_encoder->devices & ATOM_DEVICE_CRT2_SUPPORT) && + (radeon_connector->devices & ATOM_DEVICE_CRT2_SUPPORT)) { + if (connected) { + DRM_DEBUG_KMS("CRT2 connected\n"); + bios_0_scratch |= ATOM_S0_CRT2_COLOR; + bios_3_scratch |= ATOM_S3_CRT2_ACTIVE; + bios_6_scratch |= ATOM_S6_ACC_REQ_CRT2; + } else { + DRM_DEBUG_KMS("CRT2 disconnected\n"); + bios_0_scratch &= ~ATOM_S0_CRT2_MASK; + bios_3_scratch &= ~ATOM_S3_CRT2_ACTIVE; + bios_6_scratch &= ~ATOM_S6_ACC_REQ_CRT2; + } + } + if ((radeon_encoder->devices & ATOM_DEVICE_DFP1_SUPPORT) && + (radeon_connector->devices & ATOM_DEVICE_DFP1_SUPPORT)) { + if (connected) { + DRM_DEBUG_KMS("DFP1 connected\n"); + bios_0_scratch |= ATOM_S0_DFP1; + bios_3_scratch |= ATOM_S3_DFP1_ACTIVE; + bios_6_scratch |= ATOM_S6_ACC_REQ_DFP1; + } else { + DRM_DEBUG_KMS("DFP1 disconnected\n"); + bios_0_scratch &= ~ATOM_S0_DFP1; + bios_3_scratch &= ~ATOM_S3_DFP1_ACTIVE; + bios_6_scratch &= ~ATOM_S6_ACC_REQ_DFP1; + } + } + if ((radeon_encoder->devices & ATOM_DEVICE_DFP2_SUPPORT) && + (radeon_connector->devices & ATOM_DEVICE_DFP2_SUPPORT)) { + if (connected) { + DRM_DEBUG_KMS("DFP2 connected\n"); + bios_0_scratch |= ATOM_S0_DFP2; + bios_3_scratch |= ATOM_S3_DFP2_ACTIVE; + bios_6_scratch |= ATOM_S6_ACC_REQ_DFP2; + } else { + DRM_DEBUG_KMS("DFP2 disconnected\n"); + bios_0_scratch &= ~ATOM_S0_DFP2; + bios_3_scratch &= ~ATOM_S3_DFP2_ACTIVE; + bios_6_scratch &= ~ATOM_S6_ACC_REQ_DFP2; + } + } + if ((radeon_encoder->devices & ATOM_DEVICE_DFP3_SUPPORT) && + (radeon_connector->devices & ATOM_DEVICE_DFP3_SUPPORT)) { + if (connected) { + DRM_DEBUG_KMS("DFP3 connected\n"); + bios_0_scratch |= ATOM_S0_DFP3; + bios_3_scratch |= ATOM_S3_DFP3_ACTIVE; + bios_6_scratch |= ATOM_S6_ACC_REQ_DFP3; + } else { + DRM_DEBUG_KMS("DFP3 disconnected\n"); + bios_0_scratch &= ~ATOM_S0_DFP3; + bios_3_scratch &= ~ATOM_S3_DFP3_ACTIVE; + bios_6_scratch &= ~ATOM_S6_ACC_REQ_DFP3; + } + } + if ((radeon_encoder->devices & ATOM_DEVICE_DFP4_SUPPORT) && + (radeon_connector->devices & ATOM_DEVICE_DFP4_SUPPORT)) { + if (connected) { + DRM_DEBUG_KMS("DFP4 connected\n"); + bios_0_scratch |= ATOM_S0_DFP4; + bios_3_scratch |= ATOM_S3_DFP4_ACTIVE; + bios_6_scratch |= ATOM_S6_ACC_REQ_DFP4; + } else { + DRM_DEBUG_KMS("DFP4 disconnected\n"); + bios_0_scratch &= ~ATOM_S0_DFP4; + bios_3_scratch &= ~ATOM_S3_DFP4_ACTIVE; + bios_6_scratch &= ~ATOM_S6_ACC_REQ_DFP4; + } + } + if ((radeon_encoder->devices & ATOM_DEVICE_DFP5_SUPPORT) && + (radeon_connector->devices & ATOM_DEVICE_DFP5_SUPPORT)) { + if (connected) { + DRM_DEBUG_KMS("DFP5 connected\n"); + bios_0_scratch |= ATOM_S0_DFP5; + bios_3_scratch |= ATOM_S3_DFP5_ACTIVE; + bios_6_scratch |= ATOM_S6_ACC_REQ_DFP5; + } else { + DRM_DEBUG_KMS("DFP5 disconnected\n"); + bios_0_scratch &= ~ATOM_S0_DFP5; + bios_3_scratch &= ~ATOM_S3_DFP5_ACTIVE; + bios_6_scratch &= ~ATOM_S6_ACC_REQ_DFP5; + } + } + + if (rdev->family >= CHIP_R600) { + WREG32(R600_BIOS_0_SCRATCH, bios_0_scratch); + WREG32(R600_BIOS_3_SCRATCH, bios_3_scratch); + WREG32(R600_BIOS_6_SCRATCH, bios_6_scratch); + } else { + WREG32(RADEON_BIOS_0_SCRATCH, bios_0_scratch); + WREG32(RADEON_BIOS_3_SCRATCH, bios_3_scratch); + WREG32(RADEON_BIOS_6_SCRATCH, bios_6_scratch); + } +} + +void +radeon_atombios_encoder_crtc_scratch_regs(struct drm_encoder *encoder, int crtc) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + uint32_t bios_3_scratch; + + if (rdev->family >= CHIP_R600) + bios_3_scratch = RREG32(R600_BIOS_3_SCRATCH); + else + bios_3_scratch = RREG32(RADEON_BIOS_3_SCRATCH); + + if (radeon_encoder->devices & ATOM_DEVICE_TV1_SUPPORT) { + bios_3_scratch &= ~ATOM_S3_TV1_CRTC_ACTIVE; + bios_3_scratch |= (crtc << 18); + } + if (radeon_encoder->devices & ATOM_DEVICE_CV_SUPPORT) { + bios_3_scratch &= ~ATOM_S3_CV_CRTC_ACTIVE; + bios_3_scratch |= (crtc << 24); + } + if (radeon_encoder->devices & ATOM_DEVICE_CRT1_SUPPORT) { + bios_3_scratch &= ~ATOM_S3_CRT1_CRTC_ACTIVE; + bios_3_scratch |= (crtc << 16); + } + if (radeon_encoder->devices & ATOM_DEVICE_CRT2_SUPPORT) { + bios_3_scratch &= ~ATOM_S3_CRT2_CRTC_ACTIVE; + bios_3_scratch |= (crtc << 20); + } + if (radeon_encoder->devices & ATOM_DEVICE_LCD1_SUPPORT) { + bios_3_scratch &= ~ATOM_S3_LCD1_CRTC_ACTIVE; + bios_3_scratch |= (crtc << 17); + } + if (radeon_encoder->devices & ATOM_DEVICE_DFP1_SUPPORT) { + bios_3_scratch &= ~ATOM_S3_DFP1_CRTC_ACTIVE; + bios_3_scratch |= (crtc << 19); + } + if (radeon_encoder->devices & ATOM_DEVICE_DFP2_SUPPORT) { + bios_3_scratch &= ~ATOM_S3_DFP2_CRTC_ACTIVE; + bios_3_scratch |= (crtc << 23); + } + if (radeon_encoder->devices & ATOM_DEVICE_DFP3_SUPPORT) { + bios_3_scratch &= ~ATOM_S3_DFP3_CRTC_ACTIVE; + bios_3_scratch |= (crtc << 25); + } + + if (rdev->family >= CHIP_R600) + WREG32(R600_BIOS_3_SCRATCH, bios_3_scratch); + else + WREG32(RADEON_BIOS_3_SCRATCH, bios_3_scratch); +} + +void +radeon_atombios_encoder_dpms_scratch_regs(struct drm_encoder *encoder, bool on) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + uint32_t bios_2_scratch; + + if (rdev->family >= CHIP_R600) + bios_2_scratch = RREG32(R600_BIOS_2_SCRATCH); + else + bios_2_scratch = RREG32(RADEON_BIOS_2_SCRATCH); + + if (radeon_encoder->devices & ATOM_DEVICE_TV1_SUPPORT) { + if (on) + bios_2_scratch &= ~ATOM_S2_TV1_DPMS_STATE; + else + bios_2_scratch |= ATOM_S2_TV1_DPMS_STATE; + } + if (radeon_encoder->devices & ATOM_DEVICE_CV_SUPPORT) { + if (on) + bios_2_scratch &= ~ATOM_S2_CV_DPMS_STATE; + else + bios_2_scratch |= ATOM_S2_CV_DPMS_STATE; + } + if (radeon_encoder->devices & ATOM_DEVICE_CRT1_SUPPORT) { + if (on) + bios_2_scratch &= ~ATOM_S2_CRT1_DPMS_STATE; + else + bios_2_scratch |= ATOM_S2_CRT1_DPMS_STATE; + } + if (radeon_encoder->devices & ATOM_DEVICE_CRT2_SUPPORT) { + if (on) + bios_2_scratch &= ~ATOM_S2_CRT2_DPMS_STATE; + else + bios_2_scratch |= ATOM_S2_CRT2_DPMS_STATE; + } + if (radeon_encoder->devices & ATOM_DEVICE_LCD1_SUPPORT) { + if (on) + bios_2_scratch &= ~ATOM_S2_LCD1_DPMS_STATE; + else + bios_2_scratch |= ATOM_S2_LCD1_DPMS_STATE; + } + if (radeon_encoder->devices & ATOM_DEVICE_DFP1_SUPPORT) { + if (on) + bios_2_scratch &= ~ATOM_S2_DFP1_DPMS_STATE; + else + bios_2_scratch |= ATOM_S2_DFP1_DPMS_STATE; + } + if (radeon_encoder->devices & ATOM_DEVICE_DFP2_SUPPORT) { + if (on) + bios_2_scratch &= ~ATOM_S2_DFP2_DPMS_STATE; + else + bios_2_scratch |= ATOM_S2_DFP2_DPMS_STATE; + } + if (radeon_encoder->devices & ATOM_DEVICE_DFP3_SUPPORT) { + if (on) + bios_2_scratch &= ~ATOM_S2_DFP3_DPMS_STATE; + else + bios_2_scratch |= ATOM_S2_DFP3_DPMS_STATE; + } + if (radeon_encoder->devices & ATOM_DEVICE_DFP4_SUPPORT) { + if (on) + bios_2_scratch &= ~ATOM_S2_DFP4_DPMS_STATE; + else + bios_2_scratch |= ATOM_S2_DFP4_DPMS_STATE; + } + if (radeon_encoder->devices & ATOM_DEVICE_DFP5_SUPPORT) { + if (on) + bios_2_scratch &= ~ATOM_S2_DFP5_DPMS_STATE; + else + bios_2_scratch |= ATOM_S2_DFP5_DPMS_STATE; + } + + if (rdev->family >= CHIP_R600) + WREG32(R600_BIOS_2_SCRATCH, bios_2_scratch); + else + WREG32(RADEON_BIOS_2_SCRATCH, bios_2_scratch); +} diff --git a/drivers/gpu/drm/radeon/radeon_atpx_handler.c b/drivers/gpu/drm/radeon/radeon_atpx_handler.c new file mode 100644 index 00000000..9d95792b --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_atpx_handler.c @@ -0,0 +1,283 @@ +/* + * Copyright (c) 2010 Red Hat Inc. + * Author : Dave Airlie + * + * Licensed under GPLv2 + * + * ATPX support for both Intel/ATI + */ +#include +#include +#include +#include +#include + +#define ATPX_VERSION 0 +#define ATPX_GPU_PWR 2 +#define ATPX_MUX_SELECT 3 +#define ATPX_I2C_MUX_SELECT 4 +#define ATPX_SWITCH_START 5 +#define ATPX_SWITCH_END 6 + +#define ATPX_INTEGRATED 0 +#define ATPX_DISCRETE 1 + +#define ATPX_MUX_IGD 0 +#define ATPX_MUX_DISCRETE 1 + +static struct radeon_atpx_priv { + bool atpx_detected; + /* handle for device - and atpx */ + acpi_handle dhandle; + acpi_handle atpx_handle; + acpi_handle atrm_handle; +} radeon_atpx_priv; + +/* retrieve the ROM in 4k blocks */ +static int radeon_atrm_call(acpi_handle atrm_handle, uint8_t *bios, + int offset, int len) +{ + acpi_status status; + union acpi_object atrm_arg_elements[2], *obj; + struct acpi_object_list atrm_arg; + struct acpi_buffer buffer = { ACPI_ALLOCATE_BUFFER, NULL}; + + atrm_arg.count = 2; + atrm_arg.pointer = &atrm_arg_elements[0]; + + atrm_arg_elements[0].type = ACPI_TYPE_INTEGER; + atrm_arg_elements[0].integer.value = offset; + + atrm_arg_elements[1].type = ACPI_TYPE_INTEGER; + atrm_arg_elements[1].integer.value = len; + + status = acpi_evaluate_object(atrm_handle, NULL, &atrm_arg, &buffer); + if (ACPI_FAILURE(status)) { + printk("failed to evaluate ATRM got %s\n", acpi_format_exception(status)); + return -ENODEV; + } + + obj = (union acpi_object *)buffer.pointer; + memcpy(bios+offset, obj->buffer.pointer, len); + kfree(buffer.pointer); + return len; +} + +bool radeon_atrm_supported(struct pci_dev *pdev) +{ + /* get the discrete ROM only via ATRM */ + if (!radeon_atpx_priv.atpx_detected) + return false; + + if (radeon_atpx_priv.dhandle == DEVICE_ACPI_HANDLE(&pdev->dev)) + return false; + return true; +} + + +int radeon_atrm_get_bios_chunk(uint8_t *bios, int offset, int len) +{ + return radeon_atrm_call(radeon_atpx_priv.atrm_handle, bios, offset, len); +} + +static int radeon_atpx_get_version(acpi_handle handle) +{ + acpi_status status; + union acpi_object atpx_arg_elements[2], *obj; + struct acpi_object_list atpx_arg; + struct acpi_buffer buffer = { ACPI_ALLOCATE_BUFFER, NULL }; + + atpx_arg.count = 2; + atpx_arg.pointer = &atpx_arg_elements[0]; + + atpx_arg_elements[0].type = ACPI_TYPE_INTEGER; + atpx_arg_elements[0].integer.value = ATPX_VERSION; + + atpx_arg_elements[1].type = ACPI_TYPE_INTEGER; + atpx_arg_elements[1].integer.value = ATPX_VERSION; + + status = acpi_evaluate_object(handle, NULL, &atpx_arg, &buffer); + if (ACPI_FAILURE(status)) { + printk("%s: failed to call ATPX: %s\n", __func__, acpi_format_exception(status)); + return -ENOSYS; + } + obj = (union acpi_object *)buffer.pointer; + if (obj && (obj->type == ACPI_TYPE_BUFFER)) + printk(KERN_INFO "radeon atpx: version is %d\n", *((u8 *)(obj->buffer.pointer) + 2)); + kfree(buffer.pointer); + return 0; +} + +static int radeon_atpx_execute(acpi_handle handle, int cmd_id, u16 value) +{ + acpi_status status; + union acpi_object atpx_arg_elements[2]; + struct acpi_object_list atpx_arg; + struct acpi_buffer buffer = { ACPI_ALLOCATE_BUFFER, NULL }; + uint8_t buf[4] = {0}; + + if (!handle) + return -EINVAL; + + atpx_arg.count = 2; + atpx_arg.pointer = &atpx_arg_elements[0]; + + atpx_arg_elements[0].type = ACPI_TYPE_INTEGER; + atpx_arg_elements[0].integer.value = cmd_id; + + buf[2] = value & 0xff; + buf[3] = (value >> 8) & 0xff; + + atpx_arg_elements[1].type = ACPI_TYPE_BUFFER; + atpx_arg_elements[1].buffer.length = 4; + atpx_arg_elements[1].buffer.pointer = buf; + + status = acpi_evaluate_object(handle, NULL, &atpx_arg, &buffer); + if (ACPI_FAILURE(status)) { + printk("%s: failed to call ATPX: %s\n", __func__, acpi_format_exception(status)); + return -ENOSYS; + } + kfree(buffer.pointer); + + return 0; +} + +static int radeon_atpx_set_discrete_state(acpi_handle handle, int state) +{ + return radeon_atpx_execute(handle, ATPX_GPU_PWR, state); +} + +static int radeon_atpx_switch_mux(acpi_handle handle, int mux_id) +{ + return radeon_atpx_execute(handle, ATPX_MUX_SELECT, mux_id); +} + +static int radeon_atpx_switch_i2c_mux(acpi_handle handle, int mux_id) +{ + return radeon_atpx_execute(handle, ATPX_I2C_MUX_SELECT, mux_id); +} + +static int radeon_atpx_switch_start(acpi_handle handle, int gpu_id) +{ + return radeon_atpx_execute(handle, ATPX_SWITCH_START, gpu_id); +} + +static int radeon_atpx_switch_end(acpi_handle handle, int gpu_id) +{ + return radeon_atpx_execute(handle, ATPX_SWITCH_END, gpu_id); +} + +static int radeon_atpx_switchto(enum vga_switcheroo_client_id id) +{ + int gpu_id; + + if (id == VGA_SWITCHEROO_IGD) + gpu_id = ATPX_INTEGRATED; + else + gpu_id = ATPX_DISCRETE; + + radeon_atpx_switch_start(radeon_atpx_priv.atpx_handle, gpu_id); + radeon_atpx_switch_mux(radeon_atpx_priv.atpx_handle, gpu_id); + radeon_atpx_switch_i2c_mux(radeon_atpx_priv.atpx_handle, gpu_id); + radeon_atpx_switch_end(radeon_atpx_priv.atpx_handle, gpu_id); + + return 0; +} + +static int radeon_atpx_power_state(enum vga_switcheroo_client_id id, + enum vga_switcheroo_state state) +{ + /* on w500 ACPI can't change intel gpu state */ + if (id == VGA_SWITCHEROO_IGD) + return 0; + + radeon_atpx_set_discrete_state(radeon_atpx_priv.atpx_handle, state); + return 0; +} + +static bool radeon_atpx_pci_probe_handle(struct pci_dev *pdev) +{ + acpi_handle dhandle, atpx_handle, atrm_handle; + acpi_status status; + + dhandle = DEVICE_ACPI_HANDLE(&pdev->dev); + if (!dhandle) + return false; + + status = acpi_get_handle(dhandle, "ATPX", &atpx_handle); + if (ACPI_FAILURE(status)) + return false; + + status = acpi_get_handle(dhandle, "ATRM", &atrm_handle); + if (ACPI_FAILURE(status)) + return false; + + radeon_atpx_priv.dhandle = dhandle; + radeon_atpx_priv.atpx_handle = atpx_handle; + radeon_atpx_priv.atrm_handle = atrm_handle; + return true; +} + +static int radeon_atpx_init(void) +{ + /* set up the ATPX handle */ + + radeon_atpx_get_version(radeon_atpx_priv.atpx_handle); + return 0; +} + +static int radeon_atpx_get_client_id(struct pci_dev *pdev) +{ + if (radeon_atpx_priv.dhandle == DEVICE_ACPI_HANDLE(&pdev->dev)) + return VGA_SWITCHEROO_IGD; + else + return VGA_SWITCHEROO_DIS; +} + +static struct vga_switcheroo_handler radeon_atpx_handler = { + .switchto = radeon_atpx_switchto, + .power_state = radeon_atpx_power_state, + .init = radeon_atpx_init, + .get_client_id = radeon_atpx_get_client_id, +}; + +static bool radeon_atpx_detect(void) +{ + char acpi_method_name[255] = { 0 }; + struct acpi_buffer buffer = {sizeof(acpi_method_name), acpi_method_name}; + struct pci_dev *pdev = NULL; + bool has_atpx = false; + int vga_count = 0; + + while ((pdev = pci_get_class(PCI_CLASS_DISPLAY_VGA << 8, pdev)) != NULL) { + vga_count++; + + has_atpx |= (radeon_atpx_pci_probe_handle(pdev) == true); + } + + if (has_atpx && vga_count == 2) { + acpi_get_name(radeon_atpx_priv.atpx_handle, ACPI_FULL_PATHNAME, &buffer); + printk(KERN_INFO "VGA switcheroo: detected switching method %s handle\n", + acpi_method_name); + radeon_atpx_priv.atpx_detected = true; + return true; + } + return false; +} + +void radeon_register_atpx_handler(void) +{ + bool r; + + /* detect if we have any ATPX + 2 VGA in the system */ + r = radeon_atpx_detect(); + if (!r) + return; + + vga_switcheroo_register_handler(&radeon_atpx_handler); +} + +void radeon_unregister_atpx_handler(void) +{ + vga_switcheroo_unregister_handler(); +} diff --git a/drivers/gpu/drm/radeon/radeon_benchmark.c b/drivers/gpu/drm/radeon/radeon_benchmark.c new file mode 100644 index 00000000..10191d93 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_benchmark.c @@ -0,0 +1,160 @@ +/* + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Jerome Glisse + */ +#include +#include +#include "radeon_reg.h" +#include "radeon.h" + +void radeon_benchmark_move(struct radeon_device *rdev, unsigned bsize, + unsigned sdomain, unsigned ddomain) +{ + struct radeon_bo *dobj = NULL; + struct radeon_bo *sobj = NULL; + struct radeon_fence *fence = NULL; + uint64_t saddr, daddr; + unsigned long start_jiffies; + unsigned long end_jiffies; + unsigned long time; + unsigned i, n, size; + int r; + + size = bsize; + n = 1024; + r = radeon_bo_create(rdev, size, PAGE_SIZE, true, sdomain, &sobj); + if (r) { + goto out_cleanup; + } + r = radeon_bo_reserve(sobj, false); + if (unlikely(r != 0)) + goto out_cleanup; + r = radeon_bo_pin(sobj, sdomain, &saddr); + radeon_bo_unreserve(sobj); + if (r) { + goto out_cleanup; + } + r = radeon_bo_create(rdev, size, PAGE_SIZE, true, ddomain, &dobj); + if (r) { + goto out_cleanup; + } + r = radeon_bo_reserve(dobj, false); + if (unlikely(r != 0)) + goto out_cleanup; + r = radeon_bo_pin(dobj, ddomain, &daddr); + radeon_bo_unreserve(dobj); + if (r) { + goto out_cleanup; + } + + /* r100 doesn't have dma engine so skip the test */ + if (rdev->asic->copy_dma) { + + start_jiffies = jiffies; + for (i = 0; i < n; i++) { + r = radeon_fence_create(rdev, &fence); + if (r) { + goto out_cleanup; + } + + r = radeon_copy_dma(rdev, saddr, daddr, + size / RADEON_GPU_PAGE_SIZE, fence); + + if (r) { + goto out_cleanup; + } + r = radeon_fence_wait(fence, false); + if (r) { + goto out_cleanup; + } + radeon_fence_unref(&fence); + } + end_jiffies = jiffies; + time = end_jiffies - start_jiffies; + time = jiffies_to_msecs(time); + if (time > 0) { + i = ((n * size) >> 10) / time; + printk(KERN_INFO "radeon: dma %u bo moves of %ukb from" + " %d to %d in %lums (%ukb/ms %ukb/s %uM/s)\n", + n, size >> 10, + sdomain, ddomain, time, + i, i * 1000, (i * 1000) / 1024); + } + } + + start_jiffies = jiffies; + for (i = 0; i < n; i++) { + r = radeon_fence_create(rdev, &fence); + if (r) { + goto out_cleanup; + } + r = radeon_copy_blit(rdev, saddr, daddr, size / RADEON_GPU_PAGE_SIZE, fence); + if (r) { + goto out_cleanup; + } + r = radeon_fence_wait(fence, false); + if (r) { + goto out_cleanup; + } + radeon_fence_unref(&fence); + } + end_jiffies = jiffies; + time = end_jiffies - start_jiffies; + time = jiffies_to_msecs(time); + if (time > 0) { + i = ((n * size) >> 10) / time; + printk(KERN_INFO "radeon: blit %u bo moves of %ukb from %d to %d" + " in %lums (%ukb/ms %ukb/s %uM/s)\n", n, size >> 10, + sdomain, ddomain, time, i, i * 1000, (i * 1000) / 1024); + } +out_cleanup: + if (sobj) { + r = radeon_bo_reserve(sobj, false); + if (likely(r == 0)) { + radeon_bo_unpin(sobj); + radeon_bo_unreserve(sobj); + } + radeon_bo_unref(&sobj); + } + if (dobj) { + r = radeon_bo_reserve(dobj, false); + if (likely(r == 0)) { + radeon_bo_unpin(dobj); + radeon_bo_unreserve(dobj); + } + radeon_bo_unref(&dobj); + } + if (fence) { + radeon_fence_unref(&fence); + } + if (r) { + printk(KERN_WARNING "Error while benchmarking BO move.\n"); + } +} + +void radeon_benchmark(struct radeon_device *rdev) +{ + radeon_benchmark_move(rdev, 1024*1024, RADEON_GEM_DOMAIN_GTT, + RADEON_GEM_DOMAIN_VRAM); + radeon_benchmark_move(rdev, 1024*1024, RADEON_GEM_DOMAIN_VRAM, + RADEON_GEM_DOMAIN_GTT); +} diff --git a/drivers/gpu/drm/radeon/radeon_bios.c b/drivers/gpu/drm/radeon/radeon_bios.c new file mode 100644 index 00000000..229a20f1 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_bios.c @@ -0,0 +1,527 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#include "drmP.h" +#include "radeon_reg.h" +#include "radeon.h" +#include "atom.h" + +#include +#include +/* + * BIOS. + */ + +/* If you boot an IGP board with a discrete card as the primary, + * the IGP rom is not accessible via the rom bar as the IGP rom is + * part of the system bios. On boot, the system bios puts a + * copy of the igp rom at the start of vram if a discrete card is + * present. + */ +static bool igp_read_bios_from_vram(struct radeon_device *rdev) +{ + uint8_t __iomem *bios; + resource_size_t vram_base; + resource_size_t size = 256 * 1024; /* ??? */ + + if (!(rdev->flags & RADEON_IS_IGP)) + if (!radeon_card_posted(rdev)) + return false; + + rdev->bios = NULL; + vram_base = pci_resource_start(rdev->pdev, 0); + bios = ioremap(vram_base, size); + if (!bios) { + return false; + } + + if (size == 0 || bios[0] != 0x55 || bios[1] != 0xaa) { + iounmap(bios); + return false; + } + rdev->bios = kmalloc(size, GFP_KERNEL); + if (rdev->bios == NULL) { + iounmap(bios); + return false; + } + memcpy_fromio(rdev->bios, bios, size); + iounmap(bios); + return true; +} + +static bool radeon_read_bios(struct radeon_device *rdev) +{ + uint8_t __iomem *bios; + size_t size; + + rdev->bios = NULL; + /* XXX: some cards may return 0 for rom size? ddx has a workaround */ + bios = pci_map_rom(rdev->pdev, &size); + if (!bios) { + return false; + } + + if (size == 0 || bios[0] != 0x55 || bios[1] != 0xaa) { + pci_unmap_rom(rdev->pdev, bios); + return false; + } + rdev->bios = kmemdup(bios, size, GFP_KERNEL); + if (rdev->bios == NULL) { + pci_unmap_rom(rdev->pdev, bios); + return false; + } + pci_unmap_rom(rdev->pdev, bios); + return true; +} + +/* ATRM is used to get the BIOS on the discrete cards in + * dual-gpu systems. + */ +static bool radeon_atrm_get_bios(struct radeon_device *rdev) +{ + int ret; + int size = 256 * 1024; + int i; + + if (!radeon_atrm_supported(rdev->pdev)) + return false; + + rdev->bios = kmalloc(size, GFP_KERNEL); + if (!rdev->bios) { + DRM_ERROR("Unable to allocate bios\n"); + return false; + } + + for (i = 0; i < size / ATRM_BIOS_PAGE; i++) { + ret = radeon_atrm_get_bios_chunk(rdev->bios, + (i * ATRM_BIOS_PAGE), + ATRM_BIOS_PAGE); + if (ret <= 0) + break; + } + + if (i == 0 || rdev->bios[0] != 0x55 || rdev->bios[1] != 0xaa) { + kfree(rdev->bios); + return false; + } + return true; +} + +static bool ni_read_disabled_bios(struct radeon_device *rdev) +{ + u32 bus_cntl; + u32 d1vga_control; + u32 d2vga_control; + u32 vga_render_control; + u32 rom_cntl; + bool r; + + bus_cntl = RREG32(R600_BUS_CNTL); + d1vga_control = RREG32(AVIVO_D1VGA_CONTROL); + d2vga_control = RREG32(AVIVO_D2VGA_CONTROL); + vga_render_control = RREG32(AVIVO_VGA_RENDER_CONTROL); + rom_cntl = RREG32(R600_ROM_CNTL); + + /* enable the rom */ + WREG32(R600_BUS_CNTL, (bus_cntl & ~R600_BIOS_ROM_DIS)); + /* Disable VGA mode */ + WREG32(AVIVO_D1VGA_CONTROL, + (d1vga_control & ~(AVIVO_DVGA_CONTROL_MODE_ENABLE | + AVIVO_DVGA_CONTROL_TIMING_SELECT))); + WREG32(AVIVO_D2VGA_CONTROL, + (d2vga_control & ~(AVIVO_DVGA_CONTROL_MODE_ENABLE | + AVIVO_DVGA_CONTROL_TIMING_SELECT))); + WREG32(AVIVO_VGA_RENDER_CONTROL, + (vga_render_control & ~AVIVO_VGA_VSTATUS_CNTL_MASK)); + WREG32(R600_ROM_CNTL, rom_cntl | R600_SCK_OVERWRITE); + + r = radeon_read_bios(rdev); + + /* restore regs */ + WREG32(R600_BUS_CNTL, bus_cntl); + WREG32(AVIVO_D1VGA_CONTROL, d1vga_control); + WREG32(AVIVO_D2VGA_CONTROL, d2vga_control); + WREG32(AVIVO_VGA_RENDER_CONTROL, vga_render_control); + WREG32(R600_ROM_CNTL, rom_cntl); + return r; +} + +static bool r700_read_disabled_bios(struct radeon_device *rdev) +{ + uint32_t viph_control; + uint32_t bus_cntl; + uint32_t d1vga_control; + uint32_t d2vga_control; + uint32_t vga_render_control; + uint32_t rom_cntl; + uint32_t cg_spll_func_cntl = 0; + uint32_t cg_spll_status; + bool r; + + viph_control = RREG32(RADEON_VIPH_CONTROL); + bus_cntl = RREG32(R600_BUS_CNTL); + d1vga_control = RREG32(AVIVO_D1VGA_CONTROL); + d2vga_control = RREG32(AVIVO_D2VGA_CONTROL); + vga_render_control = RREG32(AVIVO_VGA_RENDER_CONTROL); + rom_cntl = RREG32(R600_ROM_CNTL); + + /* disable VIP */ + WREG32(RADEON_VIPH_CONTROL, (viph_control & ~RADEON_VIPH_EN)); + /* enable the rom */ + WREG32(R600_BUS_CNTL, (bus_cntl & ~R600_BIOS_ROM_DIS)); + /* Disable VGA mode */ + WREG32(AVIVO_D1VGA_CONTROL, + (d1vga_control & ~(AVIVO_DVGA_CONTROL_MODE_ENABLE | + AVIVO_DVGA_CONTROL_TIMING_SELECT))); + WREG32(AVIVO_D2VGA_CONTROL, + (d2vga_control & ~(AVIVO_DVGA_CONTROL_MODE_ENABLE | + AVIVO_DVGA_CONTROL_TIMING_SELECT))); + WREG32(AVIVO_VGA_RENDER_CONTROL, + (vga_render_control & ~AVIVO_VGA_VSTATUS_CNTL_MASK)); + + if (rdev->family == CHIP_RV730) { + cg_spll_func_cntl = RREG32(R600_CG_SPLL_FUNC_CNTL); + + /* enable bypass mode */ + WREG32(R600_CG_SPLL_FUNC_CNTL, (cg_spll_func_cntl | + R600_SPLL_BYPASS_EN)); + + /* wait for SPLL_CHG_STATUS to change to 1 */ + cg_spll_status = 0; + while (!(cg_spll_status & R600_SPLL_CHG_STATUS)) + cg_spll_status = RREG32(R600_CG_SPLL_STATUS); + + WREG32(R600_ROM_CNTL, (rom_cntl & ~R600_SCK_OVERWRITE)); + } else + WREG32(R600_ROM_CNTL, (rom_cntl | R600_SCK_OVERWRITE)); + + r = radeon_read_bios(rdev); + + /* restore regs */ + if (rdev->family == CHIP_RV730) { + WREG32(R600_CG_SPLL_FUNC_CNTL, cg_spll_func_cntl); + + /* wait for SPLL_CHG_STATUS to change to 1 */ + cg_spll_status = 0; + while (!(cg_spll_status & R600_SPLL_CHG_STATUS)) + cg_spll_status = RREG32(R600_CG_SPLL_STATUS); + } + WREG32(RADEON_VIPH_CONTROL, viph_control); + WREG32(R600_BUS_CNTL, bus_cntl); + WREG32(AVIVO_D1VGA_CONTROL, d1vga_control); + WREG32(AVIVO_D2VGA_CONTROL, d2vga_control); + WREG32(AVIVO_VGA_RENDER_CONTROL, vga_render_control); + WREG32(R600_ROM_CNTL, rom_cntl); + return r; +} + +static bool r600_read_disabled_bios(struct radeon_device *rdev) +{ + uint32_t viph_control; + uint32_t bus_cntl; + uint32_t d1vga_control; + uint32_t d2vga_control; + uint32_t vga_render_control; + uint32_t rom_cntl; + uint32_t general_pwrmgt; + uint32_t low_vid_lower_gpio_cntl; + uint32_t medium_vid_lower_gpio_cntl; + uint32_t high_vid_lower_gpio_cntl; + uint32_t ctxsw_vid_lower_gpio_cntl; + uint32_t lower_gpio_enable; + bool r; + + viph_control = RREG32(RADEON_VIPH_CONTROL); + bus_cntl = RREG32(R600_BUS_CNTL); + d1vga_control = RREG32(AVIVO_D1VGA_CONTROL); + d2vga_control = RREG32(AVIVO_D2VGA_CONTROL); + vga_render_control = RREG32(AVIVO_VGA_RENDER_CONTROL); + rom_cntl = RREG32(R600_ROM_CNTL); + general_pwrmgt = RREG32(R600_GENERAL_PWRMGT); + low_vid_lower_gpio_cntl = RREG32(R600_LOW_VID_LOWER_GPIO_CNTL); + medium_vid_lower_gpio_cntl = RREG32(R600_MEDIUM_VID_LOWER_GPIO_CNTL); + high_vid_lower_gpio_cntl = RREG32(R600_HIGH_VID_LOWER_GPIO_CNTL); + ctxsw_vid_lower_gpio_cntl = RREG32(R600_CTXSW_VID_LOWER_GPIO_CNTL); + lower_gpio_enable = RREG32(R600_LOWER_GPIO_ENABLE); + + /* disable VIP */ + WREG32(RADEON_VIPH_CONTROL, (viph_control & ~RADEON_VIPH_EN)); + /* enable the rom */ + WREG32(R600_BUS_CNTL, (bus_cntl & ~R600_BIOS_ROM_DIS)); + /* Disable VGA mode */ + WREG32(AVIVO_D1VGA_CONTROL, + (d1vga_control & ~(AVIVO_DVGA_CONTROL_MODE_ENABLE | + AVIVO_DVGA_CONTROL_TIMING_SELECT))); + WREG32(AVIVO_D2VGA_CONTROL, + (d2vga_control & ~(AVIVO_DVGA_CONTROL_MODE_ENABLE | + AVIVO_DVGA_CONTROL_TIMING_SELECT))); + WREG32(AVIVO_VGA_RENDER_CONTROL, + (vga_render_control & ~AVIVO_VGA_VSTATUS_CNTL_MASK)); + + WREG32(R600_ROM_CNTL, + ((rom_cntl & ~R600_SCK_PRESCALE_CRYSTAL_CLK_MASK) | + (1 << R600_SCK_PRESCALE_CRYSTAL_CLK_SHIFT) | + R600_SCK_OVERWRITE)); + + WREG32(R600_GENERAL_PWRMGT, (general_pwrmgt & ~R600_OPEN_DRAIN_PADS)); + WREG32(R600_LOW_VID_LOWER_GPIO_CNTL, + (low_vid_lower_gpio_cntl & ~0x400)); + WREG32(R600_MEDIUM_VID_LOWER_GPIO_CNTL, + (medium_vid_lower_gpio_cntl & ~0x400)); + WREG32(R600_HIGH_VID_LOWER_GPIO_CNTL, + (high_vid_lower_gpio_cntl & ~0x400)); + WREG32(R600_CTXSW_VID_LOWER_GPIO_CNTL, + (ctxsw_vid_lower_gpio_cntl & ~0x400)); + WREG32(R600_LOWER_GPIO_ENABLE, (lower_gpio_enable | 0x400)); + + r = radeon_read_bios(rdev); + + /* restore regs */ + WREG32(RADEON_VIPH_CONTROL, viph_control); + WREG32(R600_BUS_CNTL, bus_cntl); + WREG32(AVIVO_D1VGA_CONTROL, d1vga_control); + WREG32(AVIVO_D2VGA_CONTROL, d2vga_control); + WREG32(AVIVO_VGA_RENDER_CONTROL, vga_render_control); + WREG32(R600_ROM_CNTL, rom_cntl); + WREG32(R600_GENERAL_PWRMGT, general_pwrmgt); + WREG32(R600_LOW_VID_LOWER_GPIO_CNTL, low_vid_lower_gpio_cntl); + WREG32(R600_MEDIUM_VID_LOWER_GPIO_CNTL, medium_vid_lower_gpio_cntl); + WREG32(R600_HIGH_VID_LOWER_GPIO_CNTL, high_vid_lower_gpio_cntl); + WREG32(R600_CTXSW_VID_LOWER_GPIO_CNTL, ctxsw_vid_lower_gpio_cntl); + WREG32(R600_LOWER_GPIO_ENABLE, lower_gpio_enable); + return r; +} + +static bool avivo_read_disabled_bios(struct radeon_device *rdev) +{ + uint32_t seprom_cntl1; + uint32_t viph_control; + uint32_t bus_cntl; + uint32_t d1vga_control; + uint32_t d2vga_control; + uint32_t vga_render_control; + uint32_t gpiopad_a; + uint32_t gpiopad_en; + uint32_t gpiopad_mask; + bool r; + + seprom_cntl1 = RREG32(RADEON_SEPROM_CNTL1); + viph_control = RREG32(RADEON_VIPH_CONTROL); + bus_cntl = RREG32(RV370_BUS_CNTL); + d1vga_control = RREG32(AVIVO_D1VGA_CONTROL); + d2vga_control = RREG32(AVIVO_D2VGA_CONTROL); + vga_render_control = RREG32(AVIVO_VGA_RENDER_CONTROL); + gpiopad_a = RREG32(RADEON_GPIOPAD_A); + gpiopad_en = RREG32(RADEON_GPIOPAD_EN); + gpiopad_mask = RREG32(RADEON_GPIOPAD_MASK); + + WREG32(RADEON_SEPROM_CNTL1, + ((seprom_cntl1 & ~RADEON_SCK_PRESCALE_MASK) | + (0xc << RADEON_SCK_PRESCALE_SHIFT))); + WREG32(RADEON_GPIOPAD_A, 0); + WREG32(RADEON_GPIOPAD_EN, 0); + WREG32(RADEON_GPIOPAD_MASK, 0); + + /* disable VIP */ + WREG32(RADEON_VIPH_CONTROL, (viph_control & ~RADEON_VIPH_EN)); + + /* enable the rom */ + WREG32(RV370_BUS_CNTL, (bus_cntl & ~RV370_BUS_BIOS_DIS_ROM)); + + /* Disable VGA mode */ + WREG32(AVIVO_D1VGA_CONTROL, + (d1vga_control & ~(AVIVO_DVGA_CONTROL_MODE_ENABLE | + AVIVO_DVGA_CONTROL_TIMING_SELECT))); + WREG32(AVIVO_D2VGA_CONTROL, + (d2vga_control & ~(AVIVO_DVGA_CONTROL_MODE_ENABLE | + AVIVO_DVGA_CONTROL_TIMING_SELECT))); + WREG32(AVIVO_VGA_RENDER_CONTROL, + (vga_render_control & ~AVIVO_VGA_VSTATUS_CNTL_MASK)); + + r = radeon_read_bios(rdev); + + /* restore regs */ + WREG32(RADEON_SEPROM_CNTL1, seprom_cntl1); + WREG32(RADEON_VIPH_CONTROL, viph_control); + WREG32(RV370_BUS_CNTL, bus_cntl); + WREG32(AVIVO_D1VGA_CONTROL, d1vga_control); + WREG32(AVIVO_D2VGA_CONTROL, d2vga_control); + WREG32(AVIVO_VGA_RENDER_CONTROL, vga_render_control); + WREG32(RADEON_GPIOPAD_A, gpiopad_a); + WREG32(RADEON_GPIOPAD_EN, gpiopad_en); + WREG32(RADEON_GPIOPAD_MASK, gpiopad_mask); + return r; +} + +static bool legacy_read_disabled_bios(struct radeon_device *rdev) +{ + uint32_t seprom_cntl1; + uint32_t viph_control; + uint32_t bus_cntl; + uint32_t crtc_gen_cntl; + uint32_t crtc2_gen_cntl; + uint32_t crtc_ext_cntl; + uint32_t fp2_gen_cntl; + bool r; + + seprom_cntl1 = RREG32(RADEON_SEPROM_CNTL1); + viph_control = RREG32(RADEON_VIPH_CONTROL); + if (rdev->flags & RADEON_IS_PCIE) + bus_cntl = RREG32(RV370_BUS_CNTL); + else + bus_cntl = RREG32(RADEON_BUS_CNTL); + crtc_gen_cntl = RREG32(RADEON_CRTC_GEN_CNTL); + crtc2_gen_cntl = 0; + crtc_ext_cntl = RREG32(RADEON_CRTC_EXT_CNTL); + fp2_gen_cntl = 0; + + if (rdev->ddev->pci_device == PCI_DEVICE_ID_ATI_RADEON_QY) { + fp2_gen_cntl = RREG32(RADEON_FP2_GEN_CNTL); + } + + if (!(rdev->flags & RADEON_SINGLE_CRTC)) { + crtc2_gen_cntl = RREG32(RADEON_CRTC2_GEN_CNTL); + } + + WREG32(RADEON_SEPROM_CNTL1, + ((seprom_cntl1 & ~RADEON_SCK_PRESCALE_MASK) | + (0xc << RADEON_SCK_PRESCALE_SHIFT))); + + /* disable VIP */ + WREG32(RADEON_VIPH_CONTROL, (viph_control & ~RADEON_VIPH_EN)); + + /* enable the rom */ + if (rdev->flags & RADEON_IS_PCIE) + WREG32(RV370_BUS_CNTL, (bus_cntl & ~RV370_BUS_BIOS_DIS_ROM)); + else + WREG32(RADEON_BUS_CNTL, (bus_cntl & ~RADEON_BUS_BIOS_DIS_ROM)); + + /* Turn off mem requests and CRTC for both controllers */ + WREG32(RADEON_CRTC_GEN_CNTL, + ((crtc_gen_cntl & ~RADEON_CRTC_EN) | + (RADEON_CRTC_DISP_REQ_EN_B | + RADEON_CRTC_EXT_DISP_EN))); + if (!(rdev->flags & RADEON_SINGLE_CRTC)) { + WREG32(RADEON_CRTC2_GEN_CNTL, + ((crtc2_gen_cntl & ~RADEON_CRTC2_EN) | + RADEON_CRTC2_DISP_REQ_EN_B)); + } + /* Turn off CRTC */ + WREG32(RADEON_CRTC_EXT_CNTL, + ((crtc_ext_cntl & ~RADEON_CRTC_CRT_ON) | + (RADEON_CRTC_SYNC_TRISTAT | + RADEON_CRTC_DISPLAY_DIS))); + + if (rdev->ddev->pci_device == PCI_DEVICE_ID_ATI_RADEON_QY) { + WREG32(RADEON_FP2_GEN_CNTL, (fp2_gen_cntl & ~RADEON_FP2_ON)); + } + + r = radeon_read_bios(rdev); + + /* restore regs */ + WREG32(RADEON_SEPROM_CNTL1, seprom_cntl1); + WREG32(RADEON_VIPH_CONTROL, viph_control); + if (rdev->flags & RADEON_IS_PCIE) + WREG32(RV370_BUS_CNTL, bus_cntl); + else + WREG32(RADEON_BUS_CNTL, bus_cntl); + WREG32(RADEON_CRTC_GEN_CNTL, crtc_gen_cntl); + if (!(rdev->flags & RADEON_SINGLE_CRTC)) { + WREG32(RADEON_CRTC2_GEN_CNTL, crtc2_gen_cntl); + } + WREG32(RADEON_CRTC_EXT_CNTL, crtc_ext_cntl); + if (rdev->ddev->pci_device == PCI_DEVICE_ID_ATI_RADEON_QY) { + WREG32(RADEON_FP2_GEN_CNTL, fp2_gen_cntl); + } + return r; +} + +static bool radeon_read_disabled_bios(struct radeon_device *rdev) +{ + if (rdev->flags & RADEON_IS_IGP) + return igp_read_bios_from_vram(rdev); + else if (rdev->family >= CHIP_BARTS) + return ni_read_disabled_bios(rdev); + else if (rdev->family >= CHIP_RV770) + return r700_read_disabled_bios(rdev); + else if (rdev->family >= CHIP_R600) + return r600_read_disabled_bios(rdev); + else if (rdev->family >= CHIP_RS600) + return avivo_read_disabled_bios(rdev); + else + return legacy_read_disabled_bios(rdev); +} + + +bool radeon_get_bios(struct radeon_device *rdev) +{ + bool r; + uint16_t tmp; + + r = radeon_atrm_get_bios(rdev); + if (r == false) + r = igp_read_bios_from_vram(rdev); + if (r == false) + r = radeon_read_bios(rdev); + if (r == false) { + r = radeon_read_disabled_bios(rdev); + } + if (r == false || rdev->bios == NULL) { + DRM_ERROR("Unable to locate a BIOS ROM\n"); + rdev->bios = NULL; + return false; + } + if (rdev->bios[0] != 0x55 || rdev->bios[1] != 0xaa) { + printk("BIOS signature incorrect %x %x\n", rdev->bios[0], rdev->bios[1]); + goto free_bios; + } + + tmp = RBIOS16(0x18); + if (RBIOS8(tmp + 0x14) != 0x0) { + DRM_INFO("Not an x86 BIOS ROM, not using.\n"); + goto free_bios; + } + + rdev->bios_header_start = RBIOS16(0x48); + if (!rdev->bios_header_start) { + goto free_bios; + } + tmp = rdev->bios_header_start + 4; + if (!memcmp(rdev->bios + tmp, "ATOM", 4) || + !memcmp(rdev->bios + tmp, "MOTA", 4)) { + rdev->is_atom_bios = true; + } else { + rdev->is_atom_bios = false; + } + + DRM_DEBUG("%sBIOS detected\n", rdev->is_atom_bios ? "ATOM" : "COM"); + return true; +free_bios: + kfree(rdev->bios); + rdev->bios = NULL; + return false; +} diff --git a/drivers/gpu/drm/radeon/radeon_clocks.c b/drivers/gpu/drm/radeon/radeon_clocks.c new file mode 100644 index 00000000..b956cf1f --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_clocks.c @@ -0,0 +1,912 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#include "drmP.h" +#include "radeon_drm.h" +#include "radeon_reg.h" +#include "radeon.h" +#include "atom.h" + +/* 10 khz */ +uint32_t radeon_legacy_get_engine_clock(struct radeon_device *rdev) +{ + struct radeon_pll *spll = &rdev->clock.spll; + uint32_t fb_div, ref_div, post_div, sclk; + + fb_div = RREG32_PLL(RADEON_M_SPLL_REF_FB_DIV); + fb_div = (fb_div >> RADEON_SPLL_FB_DIV_SHIFT) & RADEON_SPLL_FB_DIV_MASK; + fb_div <<= 1; + fb_div *= spll->reference_freq; + + ref_div = + RREG32_PLL(RADEON_M_SPLL_REF_FB_DIV) & RADEON_M_SPLL_REF_DIV_MASK; + + if (ref_div == 0) + return 0; + + sclk = fb_div / ref_div; + + post_div = RREG32_PLL(RADEON_SCLK_CNTL) & RADEON_SCLK_SRC_SEL_MASK; + if (post_div == 2) + sclk >>= 1; + else if (post_div == 3) + sclk >>= 2; + else if (post_div == 4) + sclk >>= 3; + + return sclk; +} + +/* 10 khz */ +uint32_t radeon_legacy_get_memory_clock(struct radeon_device *rdev) +{ + struct radeon_pll *mpll = &rdev->clock.mpll; + uint32_t fb_div, ref_div, post_div, mclk; + + fb_div = RREG32_PLL(RADEON_M_SPLL_REF_FB_DIV); + fb_div = (fb_div >> RADEON_MPLL_FB_DIV_SHIFT) & RADEON_MPLL_FB_DIV_MASK; + fb_div <<= 1; + fb_div *= mpll->reference_freq; + + ref_div = + RREG32_PLL(RADEON_M_SPLL_REF_FB_DIV) & RADEON_M_SPLL_REF_DIV_MASK; + + if (ref_div == 0) + return 0; + + mclk = fb_div / ref_div; + + post_div = RREG32_PLL(RADEON_MCLK_CNTL) & 0x7; + if (post_div == 2) + mclk >>= 1; + else if (post_div == 3) + mclk >>= 2; + else if (post_div == 4) + mclk >>= 3; + + return mclk; +} + +#ifdef CONFIG_OF +/* + * Read XTAL (ref clock), SCLK and MCLK from Open Firmware device + * tree. Hopefully, ATI OF driver is kind enough to fill these + */ +static bool __devinit radeon_read_clocks_OF(struct drm_device *dev) +{ + struct radeon_device *rdev = dev->dev_private; + struct device_node *dp = rdev->pdev->dev.of_node; + const u32 *val; + struct radeon_pll *p1pll = &rdev->clock.p1pll; + struct radeon_pll *p2pll = &rdev->clock.p2pll; + struct radeon_pll *spll = &rdev->clock.spll; + struct radeon_pll *mpll = &rdev->clock.mpll; + + if (dp == NULL) + return false; + val = of_get_property(dp, "ATY,RefCLK", NULL); + if (!val || !*val) { + printk(KERN_WARNING "radeonfb: No ATY,RefCLK property !\n"); + return false; + } + p1pll->reference_freq = p2pll->reference_freq = (*val) / 10; + p1pll->reference_div = RREG32_PLL(RADEON_PPLL_REF_DIV) & 0x3ff; + if (p1pll->reference_div < 2) + p1pll->reference_div = 12; + p2pll->reference_div = p1pll->reference_div; + + /* These aren't in the device-tree */ + if (rdev->family >= CHIP_R420) { + p1pll->pll_in_min = 100; + p1pll->pll_in_max = 1350; + p1pll->pll_out_min = 20000; + p1pll->pll_out_max = 50000; + p2pll->pll_in_min = 100; + p2pll->pll_in_max = 1350; + p2pll->pll_out_min = 20000; + p2pll->pll_out_max = 50000; + } else { + p1pll->pll_in_min = 40; + p1pll->pll_in_max = 500; + p1pll->pll_out_min = 12500; + p1pll->pll_out_max = 35000; + p2pll->pll_in_min = 40; + p2pll->pll_in_max = 500; + p2pll->pll_out_min = 12500; + p2pll->pll_out_max = 35000; + } + /* not sure what the max should be in all cases */ + rdev->clock.max_pixel_clock = 35000; + + spll->reference_freq = mpll->reference_freq = p1pll->reference_freq; + spll->reference_div = mpll->reference_div = + RREG32_PLL(RADEON_M_SPLL_REF_FB_DIV) & + RADEON_M_SPLL_REF_DIV_MASK; + + val = of_get_property(dp, "ATY,SCLK", NULL); + if (val && *val) + rdev->clock.default_sclk = (*val) / 10; + else + rdev->clock.default_sclk = + radeon_legacy_get_engine_clock(rdev); + + val = of_get_property(dp, "ATY,MCLK", NULL); + if (val && *val) + rdev->clock.default_mclk = (*val) / 10; + else + rdev->clock.default_mclk = + radeon_legacy_get_memory_clock(rdev); + + DRM_INFO("Using device-tree clock info\n"); + + return true; +} +#else +static bool __devinit radeon_read_clocks_OF(struct drm_device *dev) +{ + return false; +} +#endif /* CONFIG_OF */ + +void radeon_get_clock_info(struct drm_device *dev) +{ + struct radeon_device *rdev = dev->dev_private; + struct radeon_pll *p1pll = &rdev->clock.p1pll; + struct radeon_pll *p2pll = &rdev->clock.p2pll; + struct radeon_pll *dcpll = &rdev->clock.dcpll; + struct radeon_pll *spll = &rdev->clock.spll; + struct radeon_pll *mpll = &rdev->clock.mpll; + int ret; + + if (rdev->is_atom_bios) + ret = radeon_atom_get_clock_info(dev); + else + ret = radeon_combios_get_clock_info(dev); + if (!ret) + ret = radeon_read_clocks_OF(dev); + + if (ret) { + if (p1pll->reference_div < 2) { + if (!ASIC_IS_AVIVO(rdev)) { + u32 tmp = RREG32_PLL(RADEON_PPLL_REF_DIV); + if (ASIC_IS_R300(rdev)) + p1pll->reference_div = + (tmp & R300_PPLL_REF_DIV_ACC_MASK) >> R300_PPLL_REF_DIV_ACC_SHIFT; + else + p1pll->reference_div = tmp & RADEON_PPLL_REF_DIV_MASK; + if (p1pll->reference_div < 2) + p1pll->reference_div = 12; + } else + p1pll->reference_div = 12; + } + if (p2pll->reference_div < 2) + p2pll->reference_div = 12; + if (rdev->family < CHIP_RS600) { + if (spll->reference_div < 2) + spll->reference_div = + RREG32_PLL(RADEON_M_SPLL_REF_FB_DIV) & + RADEON_M_SPLL_REF_DIV_MASK; + } + if (mpll->reference_div < 2) + mpll->reference_div = spll->reference_div; + } else { + if (ASIC_IS_AVIVO(rdev)) { + /* TODO FALLBACK */ + } else { + DRM_INFO("Using generic clock info\n"); + + /* may need to be per card */ + rdev->clock.max_pixel_clock = 35000; + + if (rdev->flags & RADEON_IS_IGP) { + p1pll->reference_freq = 1432; + p2pll->reference_freq = 1432; + spll->reference_freq = 1432; + mpll->reference_freq = 1432; + } else { + p1pll->reference_freq = 2700; + p2pll->reference_freq = 2700; + spll->reference_freq = 2700; + mpll->reference_freq = 2700; + } + p1pll->reference_div = + RREG32_PLL(RADEON_PPLL_REF_DIV) & 0x3ff; + if (p1pll->reference_div < 2) + p1pll->reference_div = 12; + p2pll->reference_div = p1pll->reference_div; + + if (rdev->family >= CHIP_R420) { + p1pll->pll_in_min = 100; + p1pll->pll_in_max = 1350; + p1pll->pll_out_min = 20000; + p1pll->pll_out_max = 50000; + p2pll->pll_in_min = 100; + p2pll->pll_in_max = 1350; + p2pll->pll_out_min = 20000; + p2pll->pll_out_max = 50000; + } else { + p1pll->pll_in_min = 40; + p1pll->pll_in_max = 500; + p1pll->pll_out_min = 12500; + p1pll->pll_out_max = 35000; + p2pll->pll_in_min = 40; + p2pll->pll_in_max = 500; + p2pll->pll_out_min = 12500; + p2pll->pll_out_max = 35000; + } + + spll->reference_div = + RREG32_PLL(RADEON_M_SPLL_REF_FB_DIV) & + RADEON_M_SPLL_REF_DIV_MASK; + mpll->reference_div = spll->reference_div; + rdev->clock.default_sclk = + radeon_legacy_get_engine_clock(rdev); + rdev->clock.default_mclk = + radeon_legacy_get_memory_clock(rdev); + } + } + + /* pixel clocks */ + if (ASIC_IS_AVIVO(rdev)) { + p1pll->min_post_div = 2; + p1pll->max_post_div = 0x7f; + p1pll->min_frac_feedback_div = 0; + p1pll->max_frac_feedback_div = 9; + p2pll->min_post_div = 2; + p2pll->max_post_div = 0x7f; + p2pll->min_frac_feedback_div = 0; + p2pll->max_frac_feedback_div = 9; + } else { + p1pll->min_post_div = 1; + p1pll->max_post_div = 16; + p1pll->min_frac_feedback_div = 0; + p1pll->max_frac_feedback_div = 0; + p2pll->min_post_div = 1; + p2pll->max_post_div = 12; + p2pll->min_frac_feedback_div = 0; + p2pll->max_frac_feedback_div = 0; + } + + /* dcpll is DCE4 only */ + dcpll->min_post_div = 2; + dcpll->max_post_div = 0x7f; + dcpll->min_frac_feedback_div = 0; + dcpll->max_frac_feedback_div = 9; + dcpll->min_ref_div = 2; + dcpll->max_ref_div = 0x3ff; + dcpll->min_feedback_div = 4; + dcpll->max_feedback_div = 0xfff; + dcpll->best_vco = 0; + + p1pll->min_ref_div = 2; + p1pll->max_ref_div = 0x3ff; + p1pll->min_feedback_div = 4; + p1pll->max_feedback_div = 0x7ff; + p1pll->best_vco = 0; + + p2pll->min_ref_div = 2; + p2pll->max_ref_div = 0x3ff; + p2pll->min_feedback_div = 4; + p2pll->max_feedback_div = 0x7ff; + p2pll->best_vco = 0; + + /* system clock */ + spll->min_post_div = 1; + spll->max_post_div = 1; + spll->min_ref_div = 2; + spll->max_ref_div = 0xff; + spll->min_feedback_div = 4; + spll->max_feedback_div = 0xff; + spll->best_vco = 0; + + /* memory clock */ + mpll->min_post_div = 1; + mpll->max_post_div = 1; + mpll->min_ref_div = 2; + mpll->max_ref_div = 0xff; + mpll->min_feedback_div = 4; + mpll->max_feedback_div = 0xff; + mpll->best_vco = 0; + + if (!rdev->clock.default_sclk) + rdev->clock.default_sclk = radeon_get_engine_clock(rdev); + if ((!rdev->clock.default_mclk) && rdev->asic->get_memory_clock) + rdev->clock.default_mclk = radeon_get_memory_clock(rdev); + + rdev->pm.current_sclk = rdev->clock.default_sclk; + rdev->pm.current_mclk = rdev->clock.default_mclk; + +} + +/* 10 khz */ +static uint32_t calc_eng_mem_clock(struct radeon_device *rdev, + uint32_t req_clock, + int *fb_div, int *post_div) +{ + struct radeon_pll *spll = &rdev->clock.spll; + int ref_div = spll->reference_div; + + if (!ref_div) + ref_div = + RREG32_PLL(RADEON_M_SPLL_REF_FB_DIV) & + RADEON_M_SPLL_REF_DIV_MASK; + + if (req_clock < 15000) { + *post_div = 8; + req_clock *= 8; + } else if (req_clock < 30000) { + *post_div = 4; + req_clock *= 4; + } else if (req_clock < 60000) { + *post_div = 2; + req_clock *= 2; + } else + *post_div = 1; + + req_clock *= ref_div; + req_clock += spll->reference_freq; + req_clock /= (2 * spll->reference_freq); + + *fb_div = req_clock & 0xff; + + req_clock = (req_clock & 0xffff) << 1; + req_clock *= spll->reference_freq; + req_clock /= ref_div; + req_clock /= *post_div; + + return req_clock; +} + +/* 10 khz */ +void radeon_legacy_set_engine_clock(struct radeon_device *rdev, + uint32_t eng_clock) +{ + uint32_t tmp; + int fb_div, post_div; + + /* XXX: wait for idle */ + + eng_clock = calc_eng_mem_clock(rdev, eng_clock, &fb_div, &post_div); + + tmp = RREG32_PLL(RADEON_CLK_PIN_CNTL); + tmp &= ~RADEON_DONT_USE_XTALIN; + WREG32_PLL(RADEON_CLK_PIN_CNTL, tmp); + + tmp = RREG32_PLL(RADEON_SCLK_CNTL); + tmp &= ~RADEON_SCLK_SRC_SEL_MASK; + WREG32_PLL(RADEON_SCLK_CNTL, tmp); + + udelay(10); + + tmp = RREG32_PLL(RADEON_SPLL_CNTL); + tmp |= RADEON_SPLL_SLEEP; + WREG32_PLL(RADEON_SPLL_CNTL, tmp); + + udelay(2); + + tmp = RREG32_PLL(RADEON_SPLL_CNTL); + tmp |= RADEON_SPLL_RESET; + WREG32_PLL(RADEON_SPLL_CNTL, tmp); + + udelay(200); + + tmp = RREG32_PLL(RADEON_M_SPLL_REF_FB_DIV); + tmp &= ~(RADEON_SPLL_FB_DIV_MASK << RADEON_SPLL_FB_DIV_SHIFT); + tmp |= (fb_div & RADEON_SPLL_FB_DIV_MASK) << RADEON_SPLL_FB_DIV_SHIFT; + WREG32_PLL(RADEON_M_SPLL_REF_FB_DIV, tmp); + + /* XXX: verify on different asics */ + tmp = RREG32_PLL(RADEON_SPLL_CNTL); + tmp &= ~RADEON_SPLL_PVG_MASK; + if ((eng_clock * post_div) >= 90000) + tmp |= (0x7 << RADEON_SPLL_PVG_SHIFT); + else + tmp |= (0x4 << RADEON_SPLL_PVG_SHIFT); + WREG32_PLL(RADEON_SPLL_CNTL, tmp); + + tmp = RREG32_PLL(RADEON_SPLL_CNTL); + tmp &= ~RADEON_SPLL_SLEEP; + WREG32_PLL(RADEON_SPLL_CNTL, tmp); + + udelay(2); + + tmp = RREG32_PLL(RADEON_SPLL_CNTL); + tmp &= ~RADEON_SPLL_RESET; + WREG32_PLL(RADEON_SPLL_CNTL, tmp); + + udelay(200); + + tmp = RREG32_PLL(RADEON_SCLK_CNTL); + tmp &= ~RADEON_SCLK_SRC_SEL_MASK; + switch (post_div) { + case 1: + default: + tmp |= 1; + break; + case 2: + tmp |= 2; + break; + case 4: + tmp |= 3; + break; + case 8: + tmp |= 4; + break; + } + WREG32_PLL(RADEON_SCLK_CNTL, tmp); + + udelay(20); + + tmp = RREG32_PLL(RADEON_CLK_PIN_CNTL); + tmp |= RADEON_DONT_USE_XTALIN; + WREG32_PLL(RADEON_CLK_PIN_CNTL, tmp); + + udelay(10); +} + +void radeon_legacy_set_clock_gating(struct radeon_device *rdev, int enable) +{ + uint32_t tmp; + + if (enable) { + if (rdev->flags & RADEON_SINGLE_CRTC) { + tmp = RREG32_PLL(RADEON_SCLK_CNTL); + if ((RREG32(RADEON_CONFIG_CNTL) & + RADEON_CFG_ATI_REV_ID_MASK) > + RADEON_CFG_ATI_REV_A13) { + tmp &= + ~(RADEON_SCLK_FORCE_CP | + RADEON_SCLK_FORCE_RB); + } + tmp &= + ~(RADEON_SCLK_FORCE_HDP | RADEON_SCLK_FORCE_DISP1 | + RADEON_SCLK_FORCE_TOP | RADEON_SCLK_FORCE_SE | + RADEON_SCLK_FORCE_IDCT | RADEON_SCLK_FORCE_RE | + RADEON_SCLK_FORCE_PB | RADEON_SCLK_FORCE_TAM | + RADEON_SCLK_FORCE_TDM); + WREG32_PLL(RADEON_SCLK_CNTL, tmp); + } else if (ASIC_IS_R300(rdev)) { + if ((rdev->family == CHIP_RS400) || + (rdev->family == CHIP_RS480)) { + tmp = RREG32_PLL(RADEON_SCLK_CNTL); + tmp &= + ~(RADEON_SCLK_FORCE_DISP2 | + RADEON_SCLK_FORCE_CP | + RADEON_SCLK_FORCE_HDP | + RADEON_SCLK_FORCE_DISP1 | + RADEON_SCLK_FORCE_TOP | + RADEON_SCLK_FORCE_E2 | R300_SCLK_FORCE_VAP + | RADEON_SCLK_FORCE_IDCT | + RADEON_SCLK_FORCE_VIP | R300_SCLK_FORCE_SR + | R300_SCLK_FORCE_PX | R300_SCLK_FORCE_TX + | R300_SCLK_FORCE_US | + RADEON_SCLK_FORCE_TV_SCLK | + R300_SCLK_FORCE_SU | + RADEON_SCLK_FORCE_OV0); + tmp |= RADEON_DYN_STOP_LAT_MASK; + tmp |= + RADEON_SCLK_FORCE_TOP | + RADEON_SCLK_FORCE_VIP; + WREG32_PLL(RADEON_SCLK_CNTL, tmp); + + tmp = RREG32_PLL(RADEON_SCLK_MORE_CNTL); + tmp &= ~RADEON_SCLK_MORE_FORCEON; + tmp |= RADEON_SCLK_MORE_MAX_DYN_STOP_LAT; + WREG32_PLL(RADEON_SCLK_MORE_CNTL, tmp); + + tmp = RREG32_PLL(RADEON_VCLK_ECP_CNTL); + tmp |= (RADEON_PIXCLK_ALWAYS_ONb | + RADEON_PIXCLK_DAC_ALWAYS_ONb); + WREG32_PLL(RADEON_VCLK_ECP_CNTL, tmp); + + tmp = RREG32_PLL(RADEON_PIXCLKS_CNTL); + tmp |= (RADEON_PIX2CLK_ALWAYS_ONb | + RADEON_PIX2CLK_DAC_ALWAYS_ONb | + RADEON_DISP_TVOUT_PIXCLK_TV_ALWAYS_ONb | + R300_DVOCLK_ALWAYS_ONb | + RADEON_PIXCLK_BLEND_ALWAYS_ONb | + RADEON_PIXCLK_GV_ALWAYS_ONb | + R300_PIXCLK_DVO_ALWAYS_ONb | + RADEON_PIXCLK_LVDS_ALWAYS_ONb | + RADEON_PIXCLK_TMDS_ALWAYS_ONb | + R300_PIXCLK_TRANS_ALWAYS_ONb | + R300_PIXCLK_TVO_ALWAYS_ONb | + R300_P2G2CLK_ALWAYS_ONb | + R300_P2G2CLK_DAC_ALWAYS_ONb); + WREG32_PLL(RADEON_PIXCLKS_CNTL, tmp); + } else if (rdev->family >= CHIP_RV350) { + tmp = RREG32_PLL(R300_SCLK_CNTL2); + tmp &= ~(R300_SCLK_FORCE_TCL | + R300_SCLK_FORCE_GA | + R300_SCLK_FORCE_CBA); + tmp |= (R300_SCLK_TCL_MAX_DYN_STOP_LAT | + R300_SCLK_GA_MAX_DYN_STOP_LAT | + R300_SCLK_CBA_MAX_DYN_STOP_LAT); + WREG32_PLL(R300_SCLK_CNTL2, tmp); + + tmp = RREG32_PLL(RADEON_SCLK_CNTL); + tmp &= + ~(RADEON_SCLK_FORCE_DISP2 | + RADEON_SCLK_FORCE_CP | + RADEON_SCLK_FORCE_HDP | + RADEON_SCLK_FORCE_DISP1 | + RADEON_SCLK_FORCE_TOP | + RADEON_SCLK_FORCE_E2 | R300_SCLK_FORCE_VAP + | RADEON_SCLK_FORCE_IDCT | + RADEON_SCLK_FORCE_VIP | R300_SCLK_FORCE_SR + | R300_SCLK_FORCE_PX | R300_SCLK_FORCE_TX + | R300_SCLK_FORCE_US | + RADEON_SCLK_FORCE_TV_SCLK | + R300_SCLK_FORCE_SU | + RADEON_SCLK_FORCE_OV0); + tmp |= RADEON_DYN_STOP_LAT_MASK; + WREG32_PLL(RADEON_SCLK_CNTL, tmp); + + tmp = RREG32_PLL(RADEON_SCLK_MORE_CNTL); + tmp &= ~RADEON_SCLK_MORE_FORCEON; + tmp |= RADEON_SCLK_MORE_MAX_DYN_STOP_LAT; + WREG32_PLL(RADEON_SCLK_MORE_CNTL, tmp); + + tmp = RREG32_PLL(RADEON_VCLK_ECP_CNTL); + tmp |= (RADEON_PIXCLK_ALWAYS_ONb | + RADEON_PIXCLK_DAC_ALWAYS_ONb); + WREG32_PLL(RADEON_VCLK_ECP_CNTL, tmp); + + tmp = RREG32_PLL(RADEON_PIXCLKS_CNTL); + tmp |= (RADEON_PIX2CLK_ALWAYS_ONb | + RADEON_PIX2CLK_DAC_ALWAYS_ONb | + RADEON_DISP_TVOUT_PIXCLK_TV_ALWAYS_ONb | + R300_DVOCLK_ALWAYS_ONb | + RADEON_PIXCLK_BLEND_ALWAYS_ONb | + RADEON_PIXCLK_GV_ALWAYS_ONb | + R300_PIXCLK_DVO_ALWAYS_ONb | + RADEON_PIXCLK_LVDS_ALWAYS_ONb | + RADEON_PIXCLK_TMDS_ALWAYS_ONb | + R300_PIXCLK_TRANS_ALWAYS_ONb | + R300_PIXCLK_TVO_ALWAYS_ONb | + R300_P2G2CLK_ALWAYS_ONb | + R300_P2G2CLK_DAC_ALWAYS_ONb); + WREG32_PLL(RADEON_PIXCLKS_CNTL, tmp); + + tmp = RREG32_PLL(RADEON_MCLK_MISC); + tmp |= (RADEON_MC_MCLK_DYN_ENABLE | + RADEON_IO_MCLK_DYN_ENABLE); + WREG32_PLL(RADEON_MCLK_MISC, tmp); + + tmp = RREG32_PLL(RADEON_MCLK_CNTL); + tmp |= (RADEON_FORCEON_MCLKA | + RADEON_FORCEON_MCLKB); + + tmp &= ~(RADEON_FORCEON_YCLKA | + RADEON_FORCEON_YCLKB | + RADEON_FORCEON_MC); + + /* Some releases of vbios have set DISABLE_MC_MCLKA + and DISABLE_MC_MCLKB bits in the vbios table. Setting these + bits will cause H/W hang when reading video memory with dynamic clocking + enabled. */ + if ((tmp & R300_DISABLE_MC_MCLKA) && + (tmp & R300_DISABLE_MC_MCLKB)) { + /* If both bits are set, then check the active channels */ + tmp = RREG32_PLL(RADEON_MCLK_CNTL); + if (rdev->mc.vram_width == 64) { + if (RREG32(RADEON_MEM_CNTL) & + R300_MEM_USE_CD_CH_ONLY) + tmp &= + ~R300_DISABLE_MC_MCLKB; + else + tmp &= + ~R300_DISABLE_MC_MCLKA; + } else { + tmp &= ~(R300_DISABLE_MC_MCLKA | + R300_DISABLE_MC_MCLKB); + } + } + + WREG32_PLL(RADEON_MCLK_CNTL, tmp); + } else { + tmp = RREG32_PLL(RADEON_SCLK_CNTL); + tmp &= ~(R300_SCLK_FORCE_VAP); + tmp |= RADEON_SCLK_FORCE_CP; + WREG32_PLL(RADEON_SCLK_CNTL, tmp); + udelay(15000); + + tmp = RREG32_PLL(R300_SCLK_CNTL2); + tmp &= ~(R300_SCLK_FORCE_TCL | + R300_SCLK_FORCE_GA | + R300_SCLK_FORCE_CBA); + WREG32_PLL(R300_SCLK_CNTL2, tmp); + } + } else { + tmp = RREG32_PLL(RADEON_CLK_PWRMGT_CNTL); + + tmp &= ~(RADEON_ACTIVE_HILO_LAT_MASK | + RADEON_DISP_DYN_STOP_LAT_MASK | + RADEON_DYN_STOP_MODE_MASK); + + tmp |= (RADEON_ENGIN_DYNCLK_MODE | + (0x01 << RADEON_ACTIVE_HILO_LAT_SHIFT)); + WREG32_PLL(RADEON_CLK_PWRMGT_CNTL, tmp); + udelay(15000); + + tmp = RREG32_PLL(RADEON_CLK_PIN_CNTL); + tmp |= RADEON_SCLK_DYN_START_CNTL; + WREG32_PLL(RADEON_CLK_PIN_CNTL, tmp); + udelay(15000); + + /* When DRI is enabled, setting DYN_STOP_LAT to zero can cause some R200 + to lockup randomly, leave them as set by BIOS. + */ + tmp = RREG32_PLL(RADEON_SCLK_CNTL); + /*tmp &= RADEON_SCLK_SRC_SEL_MASK; */ + tmp &= ~RADEON_SCLK_FORCEON_MASK; + + /*RAGE_6::A11 A12 A12N1 A13, RV250::A11 A12, R300 */ + if (((rdev->family == CHIP_RV250) && + ((RREG32(RADEON_CONFIG_CNTL) & + RADEON_CFG_ATI_REV_ID_MASK) < + RADEON_CFG_ATI_REV_A13)) + || ((rdev->family == CHIP_RV100) + && + ((RREG32(RADEON_CONFIG_CNTL) & + RADEON_CFG_ATI_REV_ID_MASK) <= + RADEON_CFG_ATI_REV_A13))) { + tmp |= RADEON_SCLK_FORCE_CP; + tmp |= RADEON_SCLK_FORCE_VIP; + } + + WREG32_PLL(RADEON_SCLK_CNTL, tmp); + + if ((rdev->family == CHIP_RV200) || + (rdev->family == CHIP_RV250) || + (rdev->family == CHIP_RV280)) { + tmp = RREG32_PLL(RADEON_SCLK_MORE_CNTL); + tmp &= ~RADEON_SCLK_MORE_FORCEON; + + /* RV200::A11 A12 RV250::A11 A12 */ + if (((rdev->family == CHIP_RV200) || + (rdev->family == CHIP_RV250)) && + ((RREG32(RADEON_CONFIG_CNTL) & + RADEON_CFG_ATI_REV_ID_MASK) < + RADEON_CFG_ATI_REV_A13)) { + tmp |= RADEON_SCLK_MORE_FORCEON; + } + WREG32_PLL(RADEON_SCLK_MORE_CNTL, tmp); + udelay(15000); + } + + /* RV200::A11 A12, RV250::A11 A12 */ + if (((rdev->family == CHIP_RV200) || + (rdev->family == CHIP_RV250)) && + ((RREG32(RADEON_CONFIG_CNTL) & + RADEON_CFG_ATI_REV_ID_MASK) < + RADEON_CFG_ATI_REV_A13)) { + tmp = RREG32_PLL(RADEON_PLL_PWRMGT_CNTL); + tmp |= RADEON_TCL_BYPASS_DISABLE; + WREG32_PLL(RADEON_PLL_PWRMGT_CNTL, tmp); + } + udelay(15000); + + /*enable dynamic mode for display clocks (PIXCLK and PIX2CLK) */ + tmp = RREG32_PLL(RADEON_PIXCLKS_CNTL); + tmp |= (RADEON_PIX2CLK_ALWAYS_ONb | + RADEON_PIX2CLK_DAC_ALWAYS_ONb | + RADEON_PIXCLK_BLEND_ALWAYS_ONb | + RADEON_PIXCLK_GV_ALWAYS_ONb | + RADEON_PIXCLK_DIG_TMDS_ALWAYS_ONb | + RADEON_PIXCLK_LVDS_ALWAYS_ONb | + RADEON_PIXCLK_TMDS_ALWAYS_ONb); + + WREG32_PLL(RADEON_PIXCLKS_CNTL, tmp); + udelay(15000); + + tmp = RREG32_PLL(RADEON_VCLK_ECP_CNTL); + tmp |= (RADEON_PIXCLK_ALWAYS_ONb | + RADEON_PIXCLK_DAC_ALWAYS_ONb); + + WREG32_PLL(RADEON_VCLK_ECP_CNTL, tmp); + udelay(15000); + } + } else { + /* Turn everything OFF (ForceON to everything) */ + if (rdev->flags & RADEON_SINGLE_CRTC) { + tmp = RREG32_PLL(RADEON_SCLK_CNTL); + tmp |= (RADEON_SCLK_FORCE_CP | RADEON_SCLK_FORCE_HDP | + RADEON_SCLK_FORCE_DISP1 | RADEON_SCLK_FORCE_TOP + | RADEON_SCLK_FORCE_E2 | RADEON_SCLK_FORCE_SE | + RADEON_SCLK_FORCE_IDCT | RADEON_SCLK_FORCE_VIP | + RADEON_SCLK_FORCE_RE | RADEON_SCLK_FORCE_PB | + RADEON_SCLK_FORCE_TAM | RADEON_SCLK_FORCE_TDM | + RADEON_SCLK_FORCE_RB); + WREG32_PLL(RADEON_SCLK_CNTL, tmp); + } else if ((rdev->family == CHIP_RS400) || + (rdev->family == CHIP_RS480)) { + tmp = RREG32_PLL(RADEON_SCLK_CNTL); + tmp |= (RADEON_SCLK_FORCE_DISP2 | RADEON_SCLK_FORCE_CP | + RADEON_SCLK_FORCE_HDP | RADEON_SCLK_FORCE_DISP1 + | RADEON_SCLK_FORCE_TOP | RADEON_SCLK_FORCE_E2 | + R300_SCLK_FORCE_VAP | RADEON_SCLK_FORCE_IDCT | + RADEON_SCLK_FORCE_VIP | R300_SCLK_FORCE_SR | + R300_SCLK_FORCE_PX | R300_SCLK_FORCE_TX | + R300_SCLK_FORCE_US | RADEON_SCLK_FORCE_TV_SCLK | + R300_SCLK_FORCE_SU | RADEON_SCLK_FORCE_OV0); + WREG32_PLL(RADEON_SCLK_CNTL, tmp); + + tmp = RREG32_PLL(RADEON_SCLK_MORE_CNTL); + tmp |= RADEON_SCLK_MORE_FORCEON; + WREG32_PLL(RADEON_SCLK_MORE_CNTL, tmp); + + tmp = RREG32_PLL(RADEON_VCLK_ECP_CNTL); + tmp &= ~(RADEON_PIXCLK_ALWAYS_ONb | + RADEON_PIXCLK_DAC_ALWAYS_ONb | + R300_DISP_DAC_PIXCLK_DAC_BLANK_OFF); + WREG32_PLL(RADEON_VCLK_ECP_CNTL, tmp); + + tmp = RREG32_PLL(RADEON_PIXCLKS_CNTL); + tmp &= ~(RADEON_PIX2CLK_ALWAYS_ONb | + RADEON_PIX2CLK_DAC_ALWAYS_ONb | + RADEON_DISP_TVOUT_PIXCLK_TV_ALWAYS_ONb | + R300_DVOCLK_ALWAYS_ONb | + RADEON_PIXCLK_BLEND_ALWAYS_ONb | + RADEON_PIXCLK_GV_ALWAYS_ONb | + R300_PIXCLK_DVO_ALWAYS_ONb | + RADEON_PIXCLK_LVDS_ALWAYS_ONb | + RADEON_PIXCLK_TMDS_ALWAYS_ONb | + R300_PIXCLK_TRANS_ALWAYS_ONb | + R300_PIXCLK_TVO_ALWAYS_ONb | + R300_P2G2CLK_ALWAYS_ONb | + R300_P2G2CLK_DAC_ALWAYS_ONb | + R300_DISP_DAC_PIXCLK_DAC2_BLANK_OFF); + WREG32_PLL(RADEON_PIXCLKS_CNTL, tmp); + } else if (rdev->family >= CHIP_RV350) { + /* for RV350/M10, no delays are required. */ + tmp = RREG32_PLL(R300_SCLK_CNTL2); + tmp |= (R300_SCLK_FORCE_TCL | + R300_SCLK_FORCE_GA | R300_SCLK_FORCE_CBA); + WREG32_PLL(R300_SCLK_CNTL2, tmp); + + tmp = RREG32_PLL(RADEON_SCLK_CNTL); + tmp |= (RADEON_SCLK_FORCE_DISP2 | RADEON_SCLK_FORCE_CP | + RADEON_SCLK_FORCE_HDP | RADEON_SCLK_FORCE_DISP1 + | RADEON_SCLK_FORCE_TOP | RADEON_SCLK_FORCE_E2 | + R300_SCLK_FORCE_VAP | RADEON_SCLK_FORCE_IDCT | + RADEON_SCLK_FORCE_VIP | R300_SCLK_FORCE_SR | + R300_SCLK_FORCE_PX | R300_SCLK_FORCE_TX | + R300_SCLK_FORCE_US | RADEON_SCLK_FORCE_TV_SCLK | + R300_SCLK_FORCE_SU | RADEON_SCLK_FORCE_OV0); + WREG32_PLL(RADEON_SCLK_CNTL, tmp); + + tmp = RREG32_PLL(RADEON_SCLK_MORE_CNTL); + tmp |= RADEON_SCLK_MORE_FORCEON; + WREG32_PLL(RADEON_SCLK_MORE_CNTL, tmp); + + tmp = RREG32_PLL(RADEON_MCLK_CNTL); + tmp |= (RADEON_FORCEON_MCLKA | + RADEON_FORCEON_MCLKB | + RADEON_FORCEON_YCLKA | + RADEON_FORCEON_YCLKB | RADEON_FORCEON_MC); + WREG32_PLL(RADEON_MCLK_CNTL, tmp); + + tmp = RREG32_PLL(RADEON_VCLK_ECP_CNTL); + tmp &= ~(RADEON_PIXCLK_ALWAYS_ONb | + RADEON_PIXCLK_DAC_ALWAYS_ONb | + R300_DISP_DAC_PIXCLK_DAC_BLANK_OFF); + WREG32_PLL(RADEON_VCLK_ECP_CNTL, tmp); + + tmp = RREG32_PLL(RADEON_PIXCLKS_CNTL); + tmp &= ~(RADEON_PIX2CLK_ALWAYS_ONb | + RADEON_PIX2CLK_DAC_ALWAYS_ONb | + RADEON_DISP_TVOUT_PIXCLK_TV_ALWAYS_ONb | + R300_DVOCLK_ALWAYS_ONb | + RADEON_PIXCLK_BLEND_ALWAYS_ONb | + RADEON_PIXCLK_GV_ALWAYS_ONb | + R300_PIXCLK_DVO_ALWAYS_ONb | + RADEON_PIXCLK_LVDS_ALWAYS_ONb | + RADEON_PIXCLK_TMDS_ALWAYS_ONb | + R300_PIXCLK_TRANS_ALWAYS_ONb | + R300_PIXCLK_TVO_ALWAYS_ONb | + R300_P2G2CLK_ALWAYS_ONb | + R300_P2G2CLK_DAC_ALWAYS_ONb | + R300_DISP_DAC_PIXCLK_DAC2_BLANK_OFF); + WREG32_PLL(RADEON_PIXCLKS_CNTL, tmp); + } else { + tmp = RREG32_PLL(RADEON_SCLK_CNTL); + tmp |= (RADEON_SCLK_FORCE_CP | RADEON_SCLK_FORCE_E2); + tmp |= RADEON_SCLK_FORCE_SE; + + if (rdev->flags & RADEON_SINGLE_CRTC) { + tmp |= (RADEON_SCLK_FORCE_RB | + RADEON_SCLK_FORCE_TDM | + RADEON_SCLK_FORCE_TAM | + RADEON_SCLK_FORCE_PB | + RADEON_SCLK_FORCE_RE | + RADEON_SCLK_FORCE_VIP | + RADEON_SCLK_FORCE_IDCT | + RADEON_SCLK_FORCE_TOP | + RADEON_SCLK_FORCE_DISP1 | + RADEON_SCLK_FORCE_DISP2 | + RADEON_SCLK_FORCE_HDP); + } else if ((rdev->family == CHIP_R300) || + (rdev->family == CHIP_R350)) { + tmp |= (RADEON_SCLK_FORCE_HDP | + RADEON_SCLK_FORCE_DISP1 | + RADEON_SCLK_FORCE_DISP2 | + RADEON_SCLK_FORCE_TOP | + RADEON_SCLK_FORCE_IDCT | + RADEON_SCLK_FORCE_VIP); + } + WREG32_PLL(RADEON_SCLK_CNTL, tmp); + + udelay(16000); + + if ((rdev->family == CHIP_R300) || + (rdev->family == CHIP_R350)) { + tmp = RREG32_PLL(R300_SCLK_CNTL2); + tmp |= (R300_SCLK_FORCE_TCL | + R300_SCLK_FORCE_GA | + R300_SCLK_FORCE_CBA); + WREG32_PLL(R300_SCLK_CNTL2, tmp); + udelay(16000); + } + + if (rdev->flags & RADEON_IS_IGP) { + tmp = RREG32_PLL(RADEON_MCLK_CNTL); + tmp &= ~(RADEON_FORCEON_MCLKA | + RADEON_FORCEON_YCLKA); + WREG32_PLL(RADEON_MCLK_CNTL, tmp); + udelay(16000); + } + + if ((rdev->family == CHIP_RV200) || + (rdev->family == CHIP_RV250) || + (rdev->family == CHIP_RV280)) { + tmp = RREG32_PLL(RADEON_SCLK_MORE_CNTL); + tmp |= RADEON_SCLK_MORE_FORCEON; + WREG32_PLL(RADEON_SCLK_MORE_CNTL, tmp); + udelay(16000); + } + + tmp = RREG32_PLL(RADEON_PIXCLKS_CNTL); + tmp &= ~(RADEON_PIX2CLK_ALWAYS_ONb | + RADEON_PIX2CLK_DAC_ALWAYS_ONb | + RADEON_PIXCLK_BLEND_ALWAYS_ONb | + RADEON_PIXCLK_GV_ALWAYS_ONb | + RADEON_PIXCLK_DIG_TMDS_ALWAYS_ONb | + RADEON_PIXCLK_LVDS_ALWAYS_ONb | + RADEON_PIXCLK_TMDS_ALWAYS_ONb); + + WREG32_PLL(RADEON_PIXCLKS_CNTL, tmp); + udelay(16000); + + tmp = RREG32_PLL(RADEON_VCLK_ECP_CNTL); + tmp &= ~(RADEON_PIXCLK_ALWAYS_ONb | + RADEON_PIXCLK_DAC_ALWAYS_ONb); + WREG32_PLL(RADEON_VCLK_ECP_CNTL, tmp); + } + } +} + diff --git a/drivers/gpu/drm/radeon/radeon_combios.c b/drivers/gpu/drm/radeon/radeon_combios.c new file mode 100644 index 00000000..859df6b5 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_combios.c @@ -0,0 +1,3508 @@ +/* + * Copyright 2004 ATI Technologies Inc., Markham, Ontario + * Copyright 2007-8 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + */ +#include "drmP.h" +#include "radeon_drm.h" +#include "radeon.h" +#include "atom.h" + +#ifdef CONFIG_PPC_PMAC +/* not sure which of these are needed */ +#include +#include +#include +#include +#endif /* CONFIG_PPC_PMAC */ + +/* from radeon_encoder.c */ +extern uint32_t +radeon_get_encoder_enum(struct drm_device *dev, uint32_t supported_device, + uint8_t dac); +extern void radeon_link_encoder_connector(struct drm_device *dev); + +/* from radeon_connector.c */ +extern void +radeon_add_legacy_connector(struct drm_device *dev, + uint32_t connector_id, + uint32_t supported_device, + int connector_type, + struct radeon_i2c_bus_rec *i2c_bus, + uint16_t connector_object_id, + struct radeon_hpd *hpd); + +/* from radeon_legacy_encoder.c */ +extern void +radeon_add_legacy_encoder(struct drm_device *dev, uint32_t encoder_enum, + uint32_t supported_device); + +/* old legacy ATI BIOS routines */ + +/* COMBIOS table offsets */ +enum radeon_combios_table_offset { + /* absolute offset tables */ + COMBIOS_ASIC_INIT_1_TABLE, + COMBIOS_BIOS_SUPPORT_TABLE, + COMBIOS_DAC_PROGRAMMING_TABLE, + COMBIOS_MAX_COLOR_DEPTH_TABLE, + COMBIOS_CRTC_INFO_TABLE, + COMBIOS_PLL_INFO_TABLE, + COMBIOS_TV_INFO_TABLE, + COMBIOS_DFP_INFO_TABLE, + COMBIOS_HW_CONFIG_INFO_TABLE, + COMBIOS_MULTIMEDIA_INFO_TABLE, + COMBIOS_TV_STD_PATCH_TABLE, + COMBIOS_LCD_INFO_TABLE, + COMBIOS_MOBILE_INFO_TABLE, + COMBIOS_PLL_INIT_TABLE, + COMBIOS_MEM_CONFIG_TABLE, + COMBIOS_SAVE_MASK_TABLE, + COMBIOS_HARDCODED_EDID_TABLE, + COMBIOS_ASIC_INIT_2_TABLE, + COMBIOS_CONNECTOR_INFO_TABLE, + COMBIOS_DYN_CLK_1_TABLE, + COMBIOS_RESERVED_MEM_TABLE, + COMBIOS_EXT_TMDS_INFO_TABLE, + COMBIOS_MEM_CLK_INFO_TABLE, + COMBIOS_EXT_DAC_INFO_TABLE, + COMBIOS_MISC_INFO_TABLE, + COMBIOS_CRT_INFO_TABLE, + COMBIOS_INTEGRATED_SYSTEM_INFO_TABLE, + COMBIOS_COMPONENT_VIDEO_INFO_TABLE, + COMBIOS_FAN_SPEED_INFO_TABLE, + COMBIOS_OVERDRIVE_INFO_TABLE, + COMBIOS_OEM_INFO_TABLE, + COMBIOS_DYN_CLK_2_TABLE, + COMBIOS_POWER_CONNECTOR_INFO_TABLE, + COMBIOS_I2C_INFO_TABLE, + /* relative offset tables */ + COMBIOS_ASIC_INIT_3_TABLE, /* offset from misc info */ + COMBIOS_ASIC_INIT_4_TABLE, /* offset from misc info */ + COMBIOS_DETECTED_MEM_TABLE, /* offset from misc info */ + COMBIOS_ASIC_INIT_5_TABLE, /* offset from misc info */ + COMBIOS_RAM_RESET_TABLE, /* offset from mem config */ + COMBIOS_POWERPLAY_INFO_TABLE, /* offset from mobile info */ + COMBIOS_GPIO_INFO_TABLE, /* offset from mobile info */ + COMBIOS_LCD_DDC_INFO_TABLE, /* offset from mobile info */ + COMBIOS_TMDS_POWER_TABLE, /* offset from mobile info */ + COMBIOS_TMDS_POWER_ON_TABLE, /* offset from tmds power */ + COMBIOS_TMDS_POWER_OFF_TABLE, /* offset from tmds power */ +}; + +enum radeon_combios_ddc { + DDC_NONE_DETECTED, + DDC_MONID, + DDC_DVI, + DDC_VGA, + DDC_CRT2, + DDC_LCD, + DDC_GPIO, +}; + +enum radeon_combios_connector { + CONNECTOR_NONE_LEGACY, + CONNECTOR_PROPRIETARY_LEGACY, + CONNECTOR_CRT_LEGACY, + CONNECTOR_DVI_I_LEGACY, + CONNECTOR_DVI_D_LEGACY, + CONNECTOR_CTV_LEGACY, + CONNECTOR_STV_LEGACY, + CONNECTOR_UNSUPPORTED_LEGACY +}; + +const int legacy_connector_convert[] = { + DRM_MODE_CONNECTOR_Unknown, + DRM_MODE_CONNECTOR_DVID, + DRM_MODE_CONNECTOR_VGA, + DRM_MODE_CONNECTOR_DVII, + DRM_MODE_CONNECTOR_DVID, + DRM_MODE_CONNECTOR_Composite, + DRM_MODE_CONNECTOR_SVIDEO, + DRM_MODE_CONNECTOR_Unknown, +}; + +static uint16_t combios_get_table_offset(struct drm_device *dev, + enum radeon_combios_table_offset table) +{ + struct radeon_device *rdev = dev->dev_private; + int rev; + uint16_t offset = 0, check_offset; + + if (!rdev->bios) + return 0; + + switch (table) { + /* absolute offset tables */ + case COMBIOS_ASIC_INIT_1_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0xc); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_BIOS_SUPPORT_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x14); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_DAC_PROGRAMMING_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x2a); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_MAX_COLOR_DEPTH_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x2c); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_CRTC_INFO_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x2e); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_PLL_INFO_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x30); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_TV_INFO_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x32); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_DFP_INFO_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x34); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_HW_CONFIG_INFO_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x36); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_MULTIMEDIA_INFO_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x38); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_TV_STD_PATCH_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x3e); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_LCD_INFO_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x40); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_MOBILE_INFO_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x42); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_PLL_INIT_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x46); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_MEM_CONFIG_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x48); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_SAVE_MASK_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x4a); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_HARDCODED_EDID_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x4c); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_ASIC_INIT_2_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x4e); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_CONNECTOR_INFO_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x50); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_DYN_CLK_1_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x52); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_RESERVED_MEM_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x54); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_EXT_TMDS_INFO_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x58); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_MEM_CLK_INFO_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x5a); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_EXT_DAC_INFO_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x5c); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_MISC_INFO_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x5e); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_CRT_INFO_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x60); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_INTEGRATED_SYSTEM_INFO_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x62); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_COMPONENT_VIDEO_INFO_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x64); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_FAN_SPEED_INFO_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x66); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_OVERDRIVE_INFO_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x68); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_OEM_INFO_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x6a); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_DYN_CLK_2_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x6c); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_POWER_CONNECTOR_INFO_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x6e); + if (check_offset) + offset = check_offset; + break; + case COMBIOS_I2C_INFO_TABLE: + check_offset = RBIOS16(rdev->bios_header_start + 0x70); + if (check_offset) + offset = check_offset; + break; + /* relative offset tables */ + case COMBIOS_ASIC_INIT_3_TABLE: /* offset from misc info */ + check_offset = + combios_get_table_offset(dev, COMBIOS_MISC_INFO_TABLE); + if (check_offset) { + rev = RBIOS8(check_offset); + if (rev > 0) { + check_offset = RBIOS16(check_offset + 0x3); + if (check_offset) + offset = check_offset; + } + } + break; + case COMBIOS_ASIC_INIT_4_TABLE: /* offset from misc info */ + check_offset = + combios_get_table_offset(dev, COMBIOS_MISC_INFO_TABLE); + if (check_offset) { + rev = RBIOS8(check_offset); + if (rev > 0) { + check_offset = RBIOS16(check_offset + 0x5); + if (check_offset) + offset = check_offset; + } + } + break; + case COMBIOS_DETECTED_MEM_TABLE: /* offset from misc info */ + check_offset = + combios_get_table_offset(dev, COMBIOS_MISC_INFO_TABLE); + if (check_offset) { + rev = RBIOS8(check_offset); + if (rev > 0) { + check_offset = RBIOS16(check_offset + 0x7); + if (check_offset) + offset = check_offset; + } + } + break; + case COMBIOS_ASIC_INIT_5_TABLE: /* offset from misc info */ + check_offset = + combios_get_table_offset(dev, COMBIOS_MISC_INFO_TABLE); + if (check_offset) { + rev = RBIOS8(check_offset); + if (rev == 2) { + check_offset = RBIOS16(check_offset + 0x9); + if (check_offset) + offset = check_offset; + } + } + break; + case COMBIOS_RAM_RESET_TABLE: /* offset from mem config */ + check_offset = + combios_get_table_offset(dev, COMBIOS_MEM_CONFIG_TABLE); + if (check_offset) { + while (RBIOS8(check_offset++)); + check_offset += 2; + if (check_offset) + offset = check_offset; + } + break; + case COMBIOS_POWERPLAY_INFO_TABLE: /* offset from mobile info */ + check_offset = + combios_get_table_offset(dev, COMBIOS_MOBILE_INFO_TABLE); + if (check_offset) { + check_offset = RBIOS16(check_offset + 0x11); + if (check_offset) + offset = check_offset; + } + break; + case COMBIOS_GPIO_INFO_TABLE: /* offset from mobile info */ + check_offset = + combios_get_table_offset(dev, COMBIOS_MOBILE_INFO_TABLE); + if (check_offset) { + check_offset = RBIOS16(check_offset + 0x13); + if (check_offset) + offset = check_offset; + } + break; + case COMBIOS_LCD_DDC_INFO_TABLE: /* offset from mobile info */ + check_offset = + combios_get_table_offset(dev, COMBIOS_MOBILE_INFO_TABLE); + if (check_offset) { + check_offset = RBIOS16(check_offset + 0x15); + if (check_offset) + offset = check_offset; + } + break; + case COMBIOS_TMDS_POWER_TABLE: /* offset from mobile info */ + check_offset = + combios_get_table_offset(dev, COMBIOS_MOBILE_INFO_TABLE); + if (check_offset) { + check_offset = RBIOS16(check_offset + 0x17); + if (check_offset) + offset = check_offset; + } + break; + case COMBIOS_TMDS_POWER_ON_TABLE: /* offset from tmds power */ + check_offset = + combios_get_table_offset(dev, COMBIOS_TMDS_POWER_TABLE); + if (check_offset) { + check_offset = RBIOS16(check_offset + 0x2); + if (check_offset) + offset = check_offset; + } + break; + case COMBIOS_TMDS_POWER_OFF_TABLE: /* offset from tmds power */ + check_offset = + combios_get_table_offset(dev, COMBIOS_TMDS_POWER_TABLE); + if (check_offset) { + check_offset = RBIOS16(check_offset + 0x4); + if (check_offset) + offset = check_offset; + } + break; + default: + break; + } + + return offset; + +} + +bool radeon_combios_check_hardcoded_edid(struct radeon_device *rdev) +{ + int edid_info, size; + struct edid *edid; + unsigned char *raw; + edid_info = combios_get_table_offset(rdev->ddev, COMBIOS_HARDCODED_EDID_TABLE); + if (!edid_info) + return false; + + raw = rdev->bios + edid_info; + size = EDID_LENGTH * (raw[0x7e] + 1); + edid = kmalloc(size, GFP_KERNEL); + if (edid == NULL) + return false; + + memcpy((unsigned char *)edid, raw, size); + + if (!drm_edid_is_valid(edid)) { + kfree(edid); + return false; + } + + rdev->mode_info.bios_hardcoded_edid = edid; + rdev->mode_info.bios_hardcoded_edid_size = size; + return true; +} + +/* this is used for atom LCDs as well */ +struct edid * +radeon_bios_get_hardcoded_edid(struct radeon_device *rdev) +{ + struct edid *edid; + + if (rdev->mode_info.bios_hardcoded_edid) { + edid = kmalloc(rdev->mode_info.bios_hardcoded_edid_size, GFP_KERNEL); + if (edid) { + memcpy((unsigned char *)edid, + (unsigned char *)rdev->mode_info.bios_hardcoded_edid, + rdev->mode_info.bios_hardcoded_edid_size); + return edid; + } + } + return NULL; +} + +static struct radeon_i2c_bus_rec combios_setup_i2c_bus(struct radeon_device *rdev, + enum radeon_combios_ddc ddc, + u32 clk_mask, + u32 data_mask) +{ + struct radeon_i2c_bus_rec i2c; + int ddc_line = 0; + + /* ddc id = mask reg + * DDC_NONE_DETECTED = none + * DDC_DVI = RADEON_GPIO_DVI_DDC + * DDC_VGA = RADEON_GPIO_VGA_DDC + * DDC_LCD = RADEON_GPIOPAD_MASK + * DDC_GPIO = RADEON_MDGPIO_MASK + * r1xx + * DDC_MONID = RADEON_GPIO_MONID + * DDC_CRT2 = RADEON_GPIO_CRT2_DDC + * r200 + * DDC_MONID = RADEON_GPIO_MONID + * DDC_CRT2 = RADEON_GPIO_DVI_DDC + * r300/r350 + * DDC_MONID = RADEON_GPIO_DVI_DDC + * DDC_CRT2 = RADEON_GPIO_DVI_DDC + * rv2xx/rv3xx + * DDC_MONID = RADEON_GPIO_MONID + * DDC_CRT2 = RADEON_GPIO_MONID + * rs3xx/rs4xx + * DDC_MONID = RADEON_GPIOPAD_MASK + * DDC_CRT2 = RADEON_GPIO_MONID + */ + switch (ddc) { + case DDC_NONE_DETECTED: + default: + ddc_line = 0; + break; + case DDC_DVI: + ddc_line = RADEON_GPIO_DVI_DDC; + break; + case DDC_VGA: + ddc_line = RADEON_GPIO_VGA_DDC; + break; + case DDC_LCD: + ddc_line = RADEON_GPIOPAD_MASK; + break; + case DDC_GPIO: + ddc_line = RADEON_MDGPIO_MASK; + break; + case DDC_MONID: + if (rdev->family == CHIP_RS300 || + rdev->family == CHIP_RS400 || + rdev->family == CHIP_RS480) + ddc_line = RADEON_GPIOPAD_MASK; + else if (rdev->family == CHIP_R300 || + rdev->family == CHIP_R350) { + ddc_line = RADEON_GPIO_DVI_DDC; + ddc = DDC_DVI; + } else + ddc_line = RADEON_GPIO_MONID; + break; + case DDC_CRT2: + if (rdev->family == CHIP_R200 || + rdev->family == CHIP_R300 || + rdev->family == CHIP_R350) { + ddc_line = RADEON_GPIO_DVI_DDC; + ddc = DDC_DVI; + } else if (rdev->family == CHIP_RS300 || + rdev->family == CHIP_RS400 || + rdev->family == CHIP_RS480) + ddc_line = RADEON_GPIO_MONID; + else if (rdev->family >= CHIP_RV350) { + ddc_line = RADEON_GPIO_MONID; + ddc = DDC_MONID; + } else + ddc_line = RADEON_GPIO_CRT2_DDC; + break; + } + + if (ddc_line == RADEON_GPIOPAD_MASK) { + i2c.mask_clk_reg = RADEON_GPIOPAD_MASK; + i2c.mask_data_reg = RADEON_GPIOPAD_MASK; + i2c.a_clk_reg = RADEON_GPIOPAD_A; + i2c.a_data_reg = RADEON_GPIOPAD_A; + i2c.en_clk_reg = RADEON_GPIOPAD_EN; + i2c.en_data_reg = RADEON_GPIOPAD_EN; + i2c.y_clk_reg = RADEON_GPIOPAD_Y; + i2c.y_data_reg = RADEON_GPIOPAD_Y; + } else if (ddc_line == RADEON_MDGPIO_MASK) { + i2c.mask_clk_reg = RADEON_MDGPIO_MASK; + i2c.mask_data_reg = RADEON_MDGPIO_MASK; + i2c.a_clk_reg = RADEON_MDGPIO_A; + i2c.a_data_reg = RADEON_MDGPIO_A; + i2c.en_clk_reg = RADEON_MDGPIO_EN; + i2c.en_data_reg = RADEON_MDGPIO_EN; + i2c.y_clk_reg = RADEON_MDGPIO_Y; + i2c.y_data_reg = RADEON_MDGPIO_Y; + } else { + i2c.mask_clk_reg = ddc_line; + i2c.mask_data_reg = ddc_line; + i2c.a_clk_reg = ddc_line; + i2c.a_data_reg = ddc_line; + i2c.en_clk_reg = ddc_line; + i2c.en_data_reg = ddc_line; + i2c.y_clk_reg = ddc_line; + i2c.y_data_reg = ddc_line; + } + + if (clk_mask && data_mask) { + /* system specific masks */ + i2c.mask_clk_mask = clk_mask; + i2c.mask_data_mask = data_mask; + i2c.a_clk_mask = clk_mask; + i2c.a_data_mask = data_mask; + i2c.en_clk_mask = clk_mask; + i2c.en_data_mask = data_mask; + i2c.y_clk_mask = clk_mask; + i2c.y_data_mask = data_mask; + } else if ((ddc_line == RADEON_GPIOPAD_MASK) || + (ddc_line == RADEON_MDGPIO_MASK)) { + /* default gpiopad masks */ + i2c.mask_clk_mask = (0x20 << 8); + i2c.mask_data_mask = 0x80; + i2c.a_clk_mask = (0x20 << 8); + i2c.a_data_mask = 0x80; + i2c.en_clk_mask = (0x20 << 8); + i2c.en_data_mask = 0x80; + i2c.y_clk_mask = (0x20 << 8); + i2c.y_data_mask = 0x80; + } else { + /* default masks for ddc pads */ + i2c.mask_clk_mask = RADEON_GPIO_MASK_1; + i2c.mask_data_mask = RADEON_GPIO_MASK_0; + i2c.a_clk_mask = RADEON_GPIO_A_1; + i2c.a_data_mask = RADEON_GPIO_A_0; + i2c.en_clk_mask = RADEON_GPIO_EN_1; + i2c.en_data_mask = RADEON_GPIO_EN_0; + i2c.y_clk_mask = RADEON_GPIO_Y_1; + i2c.y_data_mask = RADEON_GPIO_Y_0; + } + + switch (rdev->family) { + case CHIP_R100: + case CHIP_RV100: + case CHIP_RS100: + case CHIP_RV200: + case CHIP_RS200: + case CHIP_RS300: + switch (ddc_line) { + case RADEON_GPIO_DVI_DDC: + i2c.hw_capable = true; + break; + default: + i2c.hw_capable = false; + break; + } + break; + case CHIP_R200: + switch (ddc_line) { + case RADEON_GPIO_DVI_DDC: + case RADEON_GPIO_MONID: + i2c.hw_capable = true; + break; + default: + i2c.hw_capable = false; + break; + } + break; + case CHIP_RV250: + case CHIP_RV280: + switch (ddc_line) { + case RADEON_GPIO_VGA_DDC: + case RADEON_GPIO_DVI_DDC: + case RADEON_GPIO_CRT2_DDC: + i2c.hw_capable = true; + break; + default: + i2c.hw_capable = false; + break; + } + break; + case CHIP_R300: + case CHIP_R350: + switch (ddc_line) { + case RADEON_GPIO_VGA_DDC: + case RADEON_GPIO_DVI_DDC: + i2c.hw_capable = true; + break; + default: + i2c.hw_capable = false; + break; + } + break; + case CHIP_RV350: + case CHIP_RV380: + case CHIP_RS400: + case CHIP_RS480: + switch (ddc_line) { + case RADEON_GPIO_VGA_DDC: + case RADEON_GPIO_DVI_DDC: + i2c.hw_capable = true; + break; + case RADEON_GPIO_MONID: + /* hw i2c on RADEON_GPIO_MONID doesn't seem to work + * reliably on some pre-r4xx hardware; not sure why. + */ + i2c.hw_capable = false; + break; + default: + i2c.hw_capable = false; + break; + } + break; + default: + i2c.hw_capable = false; + break; + } + i2c.mm_i2c = false; + + i2c.i2c_id = ddc; + i2c.hpd = RADEON_HPD_NONE; + + if (ddc_line) + i2c.valid = true; + else + i2c.valid = false; + + return i2c; +} + +void radeon_combios_i2c_init(struct radeon_device *rdev) +{ + struct drm_device *dev = rdev->ddev; + struct radeon_i2c_bus_rec i2c; + + /* actual hw pads + * r1xx/rs2xx/rs3xx + * 0x60, 0x64, 0x68, 0x6c, gpiopads, mm + * r200 + * 0x60, 0x64, 0x68, mm + * r300/r350 + * 0x60, 0x64, mm + * rv2xx/rv3xx/rs4xx + * 0x60, 0x64, 0x68, gpiopads, mm + */ + + /* 0x60 */ + i2c = combios_setup_i2c_bus(rdev, DDC_DVI, 0, 0); + rdev->i2c_bus[0] = radeon_i2c_create(dev, &i2c, "DVI_DDC"); + /* 0x64 */ + i2c = combios_setup_i2c_bus(rdev, DDC_VGA, 0, 0); + rdev->i2c_bus[1] = radeon_i2c_create(dev, &i2c, "VGA_DDC"); + + /* mm i2c */ + i2c.valid = true; + i2c.hw_capable = true; + i2c.mm_i2c = true; + i2c.i2c_id = 0xa0; + rdev->i2c_bus[2] = radeon_i2c_create(dev, &i2c, "MM_I2C"); + + if (rdev->family == CHIP_R300 || + rdev->family == CHIP_R350) { + /* only 2 sw i2c pads */ + } else if (rdev->family == CHIP_RS300 || + rdev->family == CHIP_RS400 || + rdev->family == CHIP_RS480) { + u16 offset; + u8 id, blocks, clk, data; + int i; + + /* 0x68 */ + i2c = combios_setup_i2c_bus(rdev, DDC_CRT2, 0, 0); + rdev->i2c_bus[3] = radeon_i2c_create(dev, &i2c, "MONID"); + + offset = combios_get_table_offset(dev, COMBIOS_I2C_INFO_TABLE); + if (offset) { + blocks = RBIOS8(offset + 2); + for (i = 0; i < blocks; i++) { + id = RBIOS8(offset + 3 + (i * 5) + 0); + if (id == 136) { + clk = RBIOS8(offset + 3 + (i * 5) + 3); + data = RBIOS8(offset + 3 + (i * 5) + 4); + /* gpiopad */ + i2c = combios_setup_i2c_bus(rdev, DDC_MONID, + (1 << clk), (1 << data)); + rdev->i2c_bus[4] = radeon_i2c_create(dev, &i2c, "GPIOPAD_MASK"); + break; + } + } + } + } else if ((rdev->family == CHIP_R200) || + (rdev->family >= CHIP_R300)) { + /* 0x68 */ + i2c = combios_setup_i2c_bus(rdev, DDC_MONID, 0, 0); + rdev->i2c_bus[3] = radeon_i2c_create(dev, &i2c, "MONID"); + } else { + /* 0x68 */ + i2c = combios_setup_i2c_bus(rdev, DDC_MONID, 0, 0); + rdev->i2c_bus[3] = radeon_i2c_create(dev, &i2c, "MONID"); + /* 0x6c */ + i2c = combios_setup_i2c_bus(rdev, DDC_CRT2, 0, 0); + rdev->i2c_bus[4] = radeon_i2c_create(dev, &i2c, "CRT2_DDC"); + } +} + +bool radeon_combios_get_clock_info(struct drm_device *dev) +{ + struct radeon_device *rdev = dev->dev_private; + uint16_t pll_info; + struct radeon_pll *p1pll = &rdev->clock.p1pll; + struct radeon_pll *p2pll = &rdev->clock.p2pll; + struct radeon_pll *spll = &rdev->clock.spll; + struct radeon_pll *mpll = &rdev->clock.mpll; + int8_t rev; + uint16_t sclk, mclk; + + pll_info = combios_get_table_offset(dev, COMBIOS_PLL_INFO_TABLE); + if (pll_info) { + rev = RBIOS8(pll_info); + + /* pixel clocks */ + p1pll->reference_freq = RBIOS16(pll_info + 0xe); + p1pll->reference_div = RBIOS16(pll_info + 0x10); + p1pll->pll_out_min = RBIOS32(pll_info + 0x12); + p1pll->pll_out_max = RBIOS32(pll_info + 0x16); + p1pll->lcd_pll_out_min = p1pll->pll_out_min; + p1pll->lcd_pll_out_max = p1pll->pll_out_max; + + if (rev > 9) { + p1pll->pll_in_min = RBIOS32(pll_info + 0x36); + p1pll->pll_in_max = RBIOS32(pll_info + 0x3a); + } else { + p1pll->pll_in_min = 40; + p1pll->pll_in_max = 500; + } + *p2pll = *p1pll; + + /* system clock */ + spll->reference_freq = RBIOS16(pll_info + 0x1a); + spll->reference_div = RBIOS16(pll_info + 0x1c); + spll->pll_out_min = RBIOS32(pll_info + 0x1e); + spll->pll_out_max = RBIOS32(pll_info + 0x22); + + if (rev > 10) { + spll->pll_in_min = RBIOS32(pll_info + 0x48); + spll->pll_in_max = RBIOS32(pll_info + 0x4c); + } else { + /* ??? */ + spll->pll_in_min = 40; + spll->pll_in_max = 500; + } + + /* memory clock */ + mpll->reference_freq = RBIOS16(pll_info + 0x26); + mpll->reference_div = RBIOS16(pll_info + 0x28); + mpll->pll_out_min = RBIOS32(pll_info + 0x2a); + mpll->pll_out_max = RBIOS32(pll_info + 0x2e); + + if (rev > 10) { + mpll->pll_in_min = RBIOS32(pll_info + 0x5a); + mpll->pll_in_max = RBIOS32(pll_info + 0x5e); + } else { + /* ??? */ + mpll->pll_in_min = 40; + mpll->pll_in_max = 500; + } + + /* default sclk/mclk */ + sclk = RBIOS16(pll_info + 0xa); + mclk = RBIOS16(pll_info + 0x8); + if (sclk == 0) + sclk = 200 * 100; + if (mclk == 0) + mclk = 200 * 100; + + rdev->clock.default_sclk = sclk; + rdev->clock.default_mclk = mclk; + + if (RBIOS32(pll_info + 0x16)) + rdev->clock.max_pixel_clock = RBIOS32(pll_info + 0x16); + else + rdev->clock.max_pixel_clock = 35000; /* might need something asic specific */ + + return true; + } + return false; +} + +bool radeon_combios_sideport_present(struct radeon_device *rdev) +{ + struct drm_device *dev = rdev->ddev; + u16 igp_info; + + /* sideport is AMD only */ + if (rdev->family == CHIP_RS400) + return false; + + igp_info = combios_get_table_offset(dev, COMBIOS_INTEGRATED_SYSTEM_INFO_TABLE); + + if (igp_info) { + if (RBIOS16(igp_info + 0x4)) + return true; + } + return false; +} + +static const uint32_t default_primarydac_adj[CHIP_LAST] = { + 0x00000808, /* r100 */ + 0x00000808, /* rv100 */ + 0x00000808, /* rs100 */ + 0x00000808, /* rv200 */ + 0x00000808, /* rs200 */ + 0x00000808, /* r200 */ + 0x00000808, /* rv250 */ + 0x00000000, /* rs300 */ + 0x00000808, /* rv280 */ + 0x00000808, /* r300 */ + 0x00000808, /* r350 */ + 0x00000808, /* rv350 */ + 0x00000808, /* rv380 */ + 0x00000808, /* r420 */ + 0x00000808, /* r423 */ + 0x00000808, /* rv410 */ + 0x00000000, /* rs400 */ + 0x00000000, /* rs480 */ +}; + +static void radeon_legacy_get_primary_dac_info_from_table(struct radeon_device *rdev, + struct radeon_encoder_primary_dac *p_dac) +{ + p_dac->ps2_pdac_adj = default_primarydac_adj[rdev->family]; + return; +} + +struct radeon_encoder_primary_dac *radeon_combios_get_primary_dac_info(struct + radeon_encoder + *encoder) +{ + struct drm_device *dev = encoder->base.dev; + struct radeon_device *rdev = dev->dev_private; + uint16_t dac_info; + uint8_t rev, bg, dac; + struct radeon_encoder_primary_dac *p_dac = NULL; + int found = 0; + + p_dac = kzalloc(sizeof(struct radeon_encoder_primary_dac), + GFP_KERNEL); + + if (!p_dac) + return NULL; + + /* check CRT table */ + dac_info = combios_get_table_offset(dev, COMBIOS_CRT_INFO_TABLE); + if (dac_info) { + rev = RBIOS8(dac_info) & 0x3; + if (rev < 2) { + bg = RBIOS8(dac_info + 0x2) & 0xf; + dac = (RBIOS8(dac_info + 0x2) >> 4) & 0xf; + p_dac->ps2_pdac_adj = (bg << 8) | (dac); + } else { + bg = RBIOS8(dac_info + 0x2) & 0xf; + dac = RBIOS8(dac_info + 0x3) & 0xf; + p_dac->ps2_pdac_adj = (bg << 8) | (dac); + } + /* if the values are all zeros, use the table */ + if (p_dac->ps2_pdac_adj) + found = 1; + } + + if (!found) /* fallback to defaults */ + radeon_legacy_get_primary_dac_info_from_table(rdev, p_dac); + + return p_dac; +} + +enum radeon_tv_std +radeon_combios_get_tv_info(struct radeon_device *rdev) +{ + struct drm_device *dev = rdev->ddev; + uint16_t tv_info; + enum radeon_tv_std tv_std = TV_STD_NTSC; + + tv_info = combios_get_table_offset(dev, COMBIOS_TV_INFO_TABLE); + if (tv_info) { + if (RBIOS8(tv_info + 6) == 'T') { + switch (RBIOS8(tv_info + 7) & 0xf) { + case 1: + tv_std = TV_STD_NTSC; + DRM_DEBUG_KMS("Default TV standard: NTSC\n"); + break; + case 2: + tv_std = TV_STD_PAL; + DRM_DEBUG_KMS("Default TV standard: PAL\n"); + break; + case 3: + tv_std = TV_STD_PAL_M; + DRM_DEBUG_KMS("Default TV standard: PAL-M\n"); + break; + case 4: + tv_std = TV_STD_PAL_60; + DRM_DEBUG_KMS("Default TV standard: PAL-60\n"); + break; + case 5: + tv_std = TV_STD_NTSC_J; + DRM_DEBUG_KMS("Default TV standard: NTSC-J\n"); + break; + case 6: + tv_std = TV_STD_SCART_PAL; + DRM_DEBUG_KMS("Default TV standard: SCART-PAL\n"); + break; + default: + tv_std = TV_STD_NTSC; + DRM_DEBUG_KMS + ("Unknown TV standard; defaulting to NTSC\n"); + break; + } + + switch ((RBIOS8(tv_info + 9) >> 2) & 0x3) { + case 0: + DRM_DEBUG_KMS("29.498928713 MHz TV ref clk\n"); + break; + case 1: + DRM_DEBUG_KMS("28.636360000 MHz TV ref clk\n"); + break; + case 2: + DRM_DEBUG_KMS("14.318180000 MHz TV ref clk\n"); + break; + case 3: + DRM_DEBUG_KMS("27.000000000 MHz TV ref clk\n"); + break; + default: + break; + } + } + } + return tv_std; +} + +static const uint32_t default_tvdac_adj[CHIP_LAST] = { + 0x00000000, /* r100 */ + 0x00280000, /* rv100 */ + 0x00000000, /* rs100 */ + 0x00880000, /* rv200 */ + 0x00000000, /* rs200 */ + 0x00000000, /* r200 */ + 0x00770000, /* rv250 */ + 0x00290000, /* rs300 */ + 0x00560000, /* rv280 */ + 0x00780000, /* r300 */ + 0x00770000, /* r350 */ + 0x00780000, /* rv350 */ + 0x00780000, /* rv380 */ + 0x01080000, /* r420 */ + 0x01080000, /* r423 */ + 0x01080000, /* rv410 */ + 0x00780000, /* rs400 */ + 0x00780000, /* rs480 */ +}; + +static void radeon_legacy_get_tv_dac_info_from_table(struct radeon_device *rdev, + struct radeon_encoder_tv_dac *tv_dac) +{ + tv_dac->ps2_tvdac_adj = default_tvdac_adj[rdev->family]; + if ((rdev->flags & RADEON_IS_MOBILITY) && (rdev->family == CHIP_RV250)) + tv_dac->ps2_tvdac_adj = 0x00880000; + tv_dac->pal_tvdac_adj = tv_dac->ps2_tvdac_adj; + tv_dac->ntsc_tvdac_adj = tv_dac->ps2_tvdac_adj; + return; +} + +struct radeon_encoder_tv_dac *radeon_combios_get_tv_dac_info(struct + radeon_encoder + *encoder) +{ + struct drm_device *dev = encoder->base.dev; + struct radeon_device *rdev = dev->dev_private; + uint16_t dac_info; + uint8_t rev, bg, dac; + struct radeon_encoder_tv_dac *tv_dac = NULL; + int found = 0; + + tv_dac = kzalloc(sizeof(struct radeon_encoder_tv_dac), GFP_KERNEL); + if (!tv_dac) + return NULL; + + /* first check TV table */ + dac_info = combios_get_table_offset(dev, COMBIOS_TV_INFO_TABLE); + if (dac_info) { + rev = RBIOS8(dac_info + 0x3); + if (rev > 4) { + bg = RBIOS8(dac_info + 0xc) & 0xf; + dac = RBIOS8(dac_info + 0xd) & 0xf; + tv_dac->ps2_tvdac_adj = (bg << 16) | (dac << 20); + + bg = RBIOS8(dac_info + 0xe) & 0xf; + dac = RBIOS8(dac_info + 0xf) & 0xf; + tv_dac->pal_tvdac_adj = (bg << 16) | (dac << 20); + + bg = RBIOS8(dac_info + 0x10) & 0xf; + dac = RBIOS8(dac_info + 0x11) & 0xf; + tv_dac->ntsc_tvdac_adj = (bg << 16) | (dac << 20); + /* if the values are all zeros, use the table */ + if (tv_dac->ps2_tvdac_adj) + found = 1; + } else if (rev > 1) { + bg = RBIOS8(dac_info + 0xc) & 0xf; + dac = (RBIOS8(dac_info + 0xc) >> 4) & 0xf; + tv_dac->ps2_tvdac_adj = (bg << 16) | (dac << 20); + + bg = RBIOS8(dac_info + 0xd) & 0xf; + dac = (RBIOS8(dac_info + 0xd) >> 4) & 0xf; + tv_dac->pal_tvdac_adj = (bg << 16) | (dac << 20); + + bg = RBIOS8(dac_info + 0xe) & 0xf; + dac = (RBIOS8(dac_info + 0xe) >> 4) & 0xf; + tv_dac->ntsc_tvdac_adj = (bg << 16) | (dac << 20); + /* if the values are all zeros, use the table */ + if (tv_dac->ps2_tvdac_adj) + found = 1; + } + tv_dac->tv_std = radeon_combios_get_tv_info(rdev); + } + if (!found) { + /* then check CRT table */ + dac_info = + combios_get_table_offset(dev, COMBIOS_CRT_INFO_TABLE); + if (dac_info) { + rev = RBIOS8(dac_info) & 0x3; + if (rev < 2) { + bg = RBIOS8(dac_info + 0x3) & 0xf; + dac = (RBIOS8(dac_info + 0x3) >> 4) & 0xf; + tv_dac->ps2_tvdac_adj = + (bg << 16) | (dac << 20); + tv_dac->pal_tvdac_adj = tv_dac->ps2_tvdac_adj; + tv_dac->ntsc_tvdac_adj = tv_dac->ps2_tvdac_adj; + /* if the values are all zeros, use the table */ + if (tv_dac->ps2_tvdac_adj) + found = 1; + } else { + bg = RBIOS8(dac_info + 0x4) & 0xf; + dac = RBIOS8(dac_info + 0x5) & 0xf; + tv_dac->ps2_tvdac_adj = + (bg << 16) | (dac << 20); + tv_dac->pal_tvdac_adj = tv_dac->ps2_tvdac_adj; + tv_dac->ntsc_tvdac_adj = tv_dac->ps2_tvdac_adj; + /* if the values are all zeros, use the table */ + if (tv_dac->ps2_tvdac_adj) + found = 1; + } + } else { + DRM_INFO("No TV DAC info found in BIOS\n"); + } + } + + if (!found) /* fallback to defaults */ + radeon_legacy_get_tv_dac_info_from_table(rdev, tv_dac); + + return tv_dac; +} + +static struct radeon_encoder_lvds *radeon_legacy_get_lvds_info_from_regs(struct + radeon_device + *rdev) +{ + struct radeon_encoder_lvds *lvds = NULL; + uint32_t fp_vert_stretch, fp_horz_stretch; + uint32_t ppll_div_sel, ppll_val; + uint32_t lvds_ss_gen_cntl = RREG32(RADEON_LVDS_SS_GEN_CNTL); + + lvds = kzalloc(sizeof(struct radeon_encoder_lvds), GFP_KERNEL); + + if (!lvds) + return NULL; + + fp_vert_stretch = RREG32(RADEON_FP_VERT_STRETCH); + fp_horz_stretch = RREG32(RADEON_FP_HORZ_STRETCH); + + /* These should be fail-safe defaults, fingers crossed */ + lvds->panel_pwr_delay = 200; + lvds->panel_vcc_delay = 2000; + + lvds->lvds_gen_cntl = RREG32(RADEON_LVDS_GEN_CNTL); + lvds->panel_digon_delay = (lvds_ss_gen_cntl >> RADEON_LVDS_PWRSEQ_DELAY1_SHIFT) & 0xf; + lvds->panel_blon_delay = (lvds_ss_gen_cntl >> RADEON_LVDS_PWRSEQ_DELAY2_SHIFT) & 0xf; + + if (fp_vert_stretch & RADEON_VERT_STRETCH_ENABLE) + lvds->native_mode.vdisplay = + ((fp_vert_stretch & RADEON_VERT_PANEL_SIZE) >> + RADEON_VERT_PANEL_SHIFT) + 1; + else + lvds->native_mode.vdisplay = + (RREG32(RADEON_CRTC_V_TOTAL_DISP) >> 16) + 1; + + if (fp_horz_stretch & RADEON_HORZ_STRETCH_ENABLE) + lvds->native_mode.hdisplay = + (((fp_horz_stretch & RADEON_HORZ_PANEL_SIZE) >> + RADEON_HORZ_PANEL_SHIFT) + 1) * 8; + else + lvds->native_mode.hdisplay = + ((RREG32(RADEON_CRTC_H_TOTAL_DISP) >> 16) + 1) * 8; + + if ((lvds->native_mode.hdisplay < 640) || + (lvds->native_mode.vdisplay < 480)) { + lvds->native_mode.hdisplay = 640; + lvds->native_mode.vdisplay = 480; + } + + ppll_div_sel = RREG8(RADEON_CLOCK_CNTL_INDEX + 1) & 0x3; + ppll_val = RREG32_PLL(RADEON_PPLL_DIV_0 + ppll_div_sel); + if ((ppll_val & 0x000707ff) == 0x1bb) + lvds->use_bios_dividers = false; + else { + lvds->panel_ref_divider = + RREG32_PLL(RADEON_PPLL_REF_DIV) & 0x3ff; + lvds->panel_post_divider = (ppll_val >> 16) & 0x7; + lvds->panel_fb_divider = ppll_val & 0x7ff; + + if ((lvds->panel_ref_divider != 0) && + (lvds->panel_fb_divider > 3)) + lvds->use_bios_dividers = true; + } + lvds->panel_vcc_delay = 200; + + DRM_INFO("Panel info derived from registers\n"); + DRM_INFO("Panel Size %dx%d\n", lvds->native_mode.hdisplay, + lvds->native_mode.vdisplay); + + return lvds; +} + +struct radeon_encoder_lvds *radeon_combios_get_lvds_info(struct radeon_encoder + *encoder) +{ + struct drm_device *dev = encoder->base.dev; + struct radeon_device *rdev = dev->dev_private; + uint16_t lcd_info; + uint32_t panel_setup; + char stmp[30]; + int tmp, i; + struct radeon_encoder_lvds *lvds = NULL; + + lcd_info = combios_get_table_offset(dev, COMBIOS_LCD_INFO_TABLE); + + if (lcd_info) { + lvds = kzalloc(sizeof(struct radeon_encoder_lvds), GFP_KERNEL); + + if (!lvds) + return NULL; + + for (i = 0; i < 24; i++) + stmp[i] = RBIOS8(lcd_info + i + 1); + stmp[24] = 0; + + DRM_INFO("Panel ID String: %s\n", stmp); + + lvds->native_mode.hdisplay = RBIOS16(lcd_info + 0x19); + lvds->native_mode.vdisplay = RBIOS16(lcd_info + 0x1b); + + DRM_INFO("Panel Size %dx%d\n", lvds->native_mode.hdisplay, + lvds->native_mode.vdisplay); + + lvds->panel_vcc_delay = RBIOS16(lcd_info + 0x2c); + lvds->panel_vcc_delay = min_t(u16, lvds->panel_vcc_delay, 2000); + + lvds->panel_pwr_delay = RBIOS8(lcd_info + 0x24); + lvds->panel_digon_delay = RBIOS16(lcd_info + 0x38) & 0xf; + lvds->panel_blon_delay = (RBIOS16(lcd_info + 0x38) >> 4) & 0xf; + + lvds->panel_ref_divider = RBIOS16(lcd_info + 0x2e); + lvds->panel_post_divider = RBIOS8(lcd_info + 0x30); + lvds->panel_fb_divider = RBIOS16(lcd_info + 0x31); + if ((lvds->panel_ref_divider != 0) && + (lvds->panel_fb_divider > 3)) + lvds->use_bios_dividers = true; + + panel_setup = RBIOS32(lcd_info + 0x39); + lvds->lvds_gen_cntl = 0xff00; + if (panel_setup & 0x1) + lvds->lvds_gen_cntl |= RADEON_LVDS_PANEL_FORMAT; + + if ((panel_setup >> 4) & 0x1) + lvds->lvds_gen_cntl |= RADEON_LVDS_PANEL_TYPE; + + switch ((panel_setup >> 8) & 0x7) { + case 0: + lvds->lvds_gen_cntl |= RADEON_LVDS_NO_FM; + break; + case 1: + lvds->lvds_gen_cntl |= RADEON_LVDS_2_GREY; + break; + case 2: + lvds->lvds_gen_cntl |= RADEON_LVDS_4_GREY; + break; + default: + break; + } + + if ((panel_setup >> 16) & 0x1) + lvds->lvds_gen_cntl |= RADEON_LVDS_FP_POL_LOW; + + if ((panel_setup >> 17) & 0x1) + lvds->lvds_gen_cntl |= RADEON_LVDS_LP_POL_LOW; + + if ((panel_setup >> 18) & 0x1) + lvds->lvds_gen_cntl |= RADEON_LVDS_DTM_POL_LOW; + + if ((panel_setup >> 23) & 0x1) + lvds->lvds_gen_cntl |= RADEON_LVDS_BL_CLK_SEL; + + lvds->lvds_gen_cntl |= (panel_setup & 0xf0000000); + + for (i = 0; i < 32; i++) { + tmp = RBIOS16(lcd_info + 64 + i * 2); + if (tmp == 0) + break; + + if ((RBIOS16(tmp) == lvds->native_mode.hdisplay) && + (RBIOS16(tmp + 2) == lvds->native_mode.vdisplay)) { + lvds->native_mode.htotal = lvds->native_mode.hdisplay + + (RBIOS16(tmp + 17) - RBIOS16(tmp + 19)) * 8; + lvds->native_mode.hsync_start = lvds->native_mode.hdisplay + + (RBIOS16(tmp + 21) - RBIOS16(tmp + 19) - 1) * 8; + lvds->native_mode.hsync_end = lvds->native_mode.hsync_start + + (RBIOS8(tmp + 23) * 8); + + lvds->native_mode.vtotal = lvds->native_mode.vdisplay + + (RBIOS16(tmp + 24) - RBIOS16(tmp + 26)); + lvds->native_mode.vsync_start = lvds->native_mode.vdisplay + + ((RBIOS16(tmp + 28) & 0x7ff) - RBIOS16(tmp + 26)); + lvds->native_mode.vsync_end = lvds->native_mode.vsync_start + + ((RBIOS16(tmp + 28) & 0xf800) >> 11); + + lvds->native_mode.clock = RBIOS16(tmp + 9) * 10; + lvds->native_mode.flags = 0; + /* set crtc values */ + drm_mode_set_crtcinfo(&lvds->native_mode, CRTC_INTERLACE_HALVE_V); + + } + } + } else { + DRM_INFO("No panel info found in BIOS\n"); + lvds = radeon_legacy_get_lvds_info_from_regs(rdev); + } + + if (lvds) + encoder->native_mode = lvds->native_mode; + return lvds; +} + +static const struct radeon_tmds_pll default_tmds_pll[CHIP_LAST][4] = { + {{12000, 0xa1b}, {0xffffffff, 0xa3f}, {0, 0}, {0, 0}}, /* CHIP_R100 */ + {{12000, 0xa1b}, {0xffffffff, 0xa3f}, {0, 0}, {0, 0}}, /* CHIP_RV100 */ + {{0, 0}, {0, 0}, {0, 0}, {0, 0}}, /* CHIP_RS100 */ + {{15000, 0xa1b}, {0xffffffff, 0xa3f}, {0, 0}, {0, 0}}, /* CHIP_RV200 */ + {{12000, 0xa1b}, {0xffffffff, 0xa3f}, {0, 0}, {0, 0}}, /* CHIP_RS200 */ + {{15000, 0xa1b}, {0xffffffff, 0xa3f}, {0, 0}, {0, 0}}, /* CHIP_R200 */ + {{15500, 0x81b}, {0xffffffff, 0x83f}, {0, 0}, {0, 0}}, /* CHIP_RV250 */ + {{0, 0}, {0, 0}, {0, 0}, {0, 0}}, /* CHIP_RS300 */ + {{13000, 0x400f4}, {15000, 0x400f7}, {0xffffffff, 0x40111}, {0, 0}}, /* CHIP_RV280 */ + {{0xffffffff, 0xb01cb}, {0, 0}, {0, 0}, {0, 0}}, /* CHIP_R300 */ + {{0xffffffff, 0xb01cb}, {0, 0}, {0, 0}, {0, 0}}, /* CHIP_R350 */ + {{15000, 0xb0155}, {0xffffffff, 0xb01cb}, {0, 0}, {0, 0}}, /* CHIP_RV350 */ + {{15000, 0xb0155}, {0xffffffff, 0xb01cb}, {0, 0}, {0, 0}}, /* CHIP_RV380 */ + {{0xffffffff, 0xb01cb}, {0, 0}, {0, 0}, {0, 0}}, /* CHIP_R420 */ + {{0xffffffff, 0xb01cb}, {0, 0}, {0, 0}, {0, 0}}, /* CHIP_R423 */ + {{0xffffffff, 0xb01cb}, {0, 0}, {0, 0}, {0, 0}}, /* CHIP_RV410 */ + { {0, 0}, {0, 0}, {0, 0}, {0, 0} }, /* CHIP_RS400 */ + { {0, 0}, {0, 0}, {0, 0}, {0, 0} }, /* CHIP_RS480 */ +}; + +bool radeon_legacy_get_tmds_info_from_table(struct radeon_encoder *encoder, + struct radeon_encoder_int_tmds *tmds) +{ + struct drm_device *dev = encoder->base.dev; + struct radeon_device *rdev = dev->dev_private; + int i; + + for (i = 0; i < 4; i++) { + tmds->tmds_pll[i].value = + default_tmds_pll[rdev->family][i].value; + tmds->tmds_pll[i].freq = default_tmds_pll[rdev->family][i].freq; + } + + return true; +} + +bool radeon_legacy_get_tmds_info_from_combios(struct radeon_encoder *encoder, + struct radeon_encoder_int_tmds *tmds) +{ + struct drm_device *dev = encoder->base.dev; + struct radeon_device *rdev = dev->dev_private; + uint16_t tmds_info; + int i, n; + uint8_t ver; + + tmds_info = combios_get_table_offset(dev, COMBIOS_DFP_INFO_TABLE); + + if (tmds_info) { + ver = RBIOS8(tmds_info); + DRM_DEBUG_KMS("DFP table revision: %d\n", ver); + if (ver == 3) { + n = RBIOS8(tmds_info + 5) + 1; + if (n > 4) + n = 4; + for (i = 0; i < n; i++) { + tmds->tmds_pll[i].value = + RBIOS32(tmds_info + i * 10 + 0x08); + tmds->tmds_pll[i].freq = + RBIOS16(tmds_info + i * 10 + 0x10); + DRM_DEBUG_KMS("TMDS PLL From COMBIOS %u %x\n", + tmds->tmds_pll[i].freq, + tmds->tmds_pll[i].value); + } + } else if (ver == 4) { + int stride = 0; + n = RBIOS8(tmds_info + 5) + 1; + if (n > 4) + n = 4; + for (i = 0; i < n; i++) { + tmds->tmds_pll[i].value = + RBIOS32(tmds_info + stride + 0x08); + tmds->tmds_pll[i].freq = + RBIOS16(tmds_info + stride + 0x10); + if (i == 0) + stride += 10; + else + stride += 6; + DRM_DEBUG_KMS("TMDS PLL From COMBIOS %u %x\n", + tmds->tmds_pll[i].freq, + tmds->tmds_pll[i].value); + } + } + } else { + DRM_INFO("No TMDS info found in BIOS\n"); + return false; + } + return true; +} + +bool radeon_legacy_get_ext_tmds_info_from_table(struct radeon_encoder *encoder, + struct radeon_encoder_ext_tmds *tmds) +{ + struct drm_device *dev = encoder->base.dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_i2c_bus_rec i2c_bus; + + /* default for macs */ + i2c_bus = combios_setup_i2c_bus(rdev, DDC_MONID, 0, 0); + tmds->i2c_bus = radeon_i2c_lookup(rdev, &i2c_bus); + + /* XXX some macs have duallink chips */ + switch (rdev->mode_info.connector_table) { + case CT_POWERBOOK_EXTERNAL: + case CT_MINI_EXTERNAL: + default: + tmds->dvo_chip = DVO_SIL164; + tmds->slave_addr = 0x70 >> 1; /* 7 bit addressing */ + break; + } + + return true; +} + +bool radeon_legacy_get_ext_tmds_info_from_combios(struct radeon_encoder *encoder, + struct radeon_encoder_ext_tmds *tmds) +{ + struct drm_device *dev = encoder->base.dev; + struct radeon_device *rdev = dev->dev_private; + uint16_t offset; + uint8_t ver; + enum radeon_combios_ddc gpio; + struct radeon_i2c_bus_rec i2c_bus; + + tmds->i2c_bus = NULL; + if (rdev->flags & RADEON_IS_IGP) { + i2c_bus = combios_setup_i2c_bus(rdev, DDC_MONID, 0, 0); + tmds->i2c_bus = radeon_i2c_lookup(rdev, &i2c_bus); + tmds->dvo_chip = DVO_SIL164; + tmds->slave_addr = 0x70 >> 1; /* 7 bit addressing */ + } else { + offset = combios_get_table_offset(dev, COMBIOS_EXT_TMDS_INFO_TABLE); + if (offset) { + ver = RBIOS8(offset); + DRM_DEBUG_KMS("External TMDS Table revision: %d\n", ver); + tmds->slave_addr = RBIOS8(offset + 4 + 2); + tmds->slave_addr >>= 1; /* 7 bit addressing */ + gpio = RBIOS8(offset + 4 + 3); + if (gpio == DDC_LCD) { + /* MM i2c */ + i2c_bus.valid = true; + i2c_bus.hw_capable = true; + i2c_bus.mm_i2c = true; + i2c_bus.i2c_id = 0xa0; + } else + i2c_bus = combios_setup_i2c_bus(rdev, gpio, 0, 0); + tmds->i2c_bus = radeon_i2c_lookup(rdev, &i2c_bus); + } + } + + if (!tmds->i2c_bus) { + DRM_INFO("No valid Ext TMDS info found in BIOS\n"); + return false; + } + + return true; +} + +bool radeon_get_legacy_connector_info_from_table(struct drm_device *dev) +{ + struct radeon_device *rdev = dev->dev_private; + struct radeon_i2c_bus_rec ddc_i2c; + struct radeon_hpd hpd; + + rdev->mode_info.connector_table = radeon_connector_table; + if (rdev->mode_info.connector_table == CT_NONE) { +#ifdef CONFIG_PPC_PMAC + if (of_machine_is_compatible("PowerBook3,3")) { + /* powerbook with VGA */ + rdev->mode_info.connector_table = CT_POWERBOOK_VGA; + } else if (of_machine_is_compatible("PowerBook3,4") || + of_machine_is_compatible("PowerBook3,5")) { + /* powerbook with internal tmds */ + rdev->mode_info.connector_table = CT_POWERBOOK_INTERNAL; + } else if (of_machine_is_compatible("PowerBook5,1") || + of_machine_is_compatible("PowerBook5,2") || + of_machine_is_compatible("PowerBook5,3") || + of_machine_is_compatible("PowerBook5,4") || + of_machine_is_compatible("PowerBook5,5")) { + /* powerbook with external single link tmds (sil164) */ + rdev->mode_info.connector_table = CT_POWERBOOK_EXTERNAL; + } else if (of_machine_is_compatible("PowerBook5,6")) { + /* powerbook with external dual or single link tmds */ + rdev->mode_info.connector_table = CT_POWERBOOK_EXTERNAL; + } else if (of_machine_is_compatible("PowerBook5,7") || + of_machine_is_compatible("PowerBook5,8") || + of_machine_is_compatible("PowerBook5,9")) { + /* PowerBook6,2 ? */ + /* powerbook with external dual link tmds (sil1178?) */ + rdev->mode_info.connector_table = CT_POWERBOOK_EXTERNAL; + } else if (of_machine_is_compatible("PowerBook4,1") || + of_machine_is_compatible("PowerBook4,2") || + of_machine_is_compatible("PowerBook4,3") || + of_machine_is_compatible("PowerBook6,3") || + of_machine_is_compatible("PowerBook6,5") || + of_machine_is_compatible("PowerBook6,7")) { + /* ibook */ + rdev->mode_info.connector_table = CT_IBOOK; + } else if (of_machine_is_compatible("PowerMac4,4")) { + /* emac */ + rdev->mode_info.connector_table = CT_EMAC; + } else if (of_machine_is_compatible("PowerMac10,1")) { + /* mini with internal tmds */ + rdev->mode_info.connector_table = CT_MINI_INTERNAL; + } else if (of_machine_is_compatible("PowerMac10,2")) { + /* mini with external tmds */ + rdev->mode_info.connector_table = CT_MINI_EXTERNAL; + } else if (of_machine_is_compatible("PowerMac12,1")) { + /* PowerMac8,1 ? */ + /* imac g5 isight */ + rdev->mode_info.connector_table = CT_IMAC_G5_ISIGHT; + } else if ((rdev->pdev->device == 0x4a48) && + (rdev->pdev->subsystem_vendor == 0x1002) && + (rdev->pdev->subsystem_device == 0x4a48)) { + /* Mac X800 */ + rdev->mode_info.connector_table = CT_MAC_X800; + } else if ((of_machine_is_compatible("PowerMac7,2") || + of_machine_is_compatible("PowerMac7,3")) && + (rdev->pdev->device == 0x4150) && + (rdev->pdev->subsystem_vendor == 0x1002) && + (rdev->pdev->subsystem_device == 0x4150)) { + /* Mac G5 tower 9600 */ + rdev->mode_info.connector_table = CT_MAC_G5_9600; + } else +#endif /* CONFIG_PPC_PMAC */ +#ifdef CONFIG_PPC64 + if (ASIC_IS_RN50(rdev)) + rdev->mode_info.connector_table = CT_RN50_POWER; + else +#endif + rdev->mode_info.connector_table = CT_GENERIC; + } + + switch (rdev->mode_info.connector_table) { + case CT_GENERIC: + DRM_INFO("Connector Table: %d (generic)\n", + rdev->mode_info.connector_table); + /* these are the most common settings */ + if (rdev->flags & RADEON_SINGLE_CRTC) { + /* VGA - primary dac */ + ddc_i2c = combios_setup_i2c_bus(rdev, DDC_VGA, 0, 0); + hpd.hpd = RADEON_HPD_NONE; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_CRT1_SUPPORT, + 1), + ATOM_DEVICE_CRT1_SUPPORT); + radeon_add_legacy_connector(dev, 0, + ATOM_DEVICE_CRT1_SUPPORT, + DRM_MODE_CONNECTOR_VGA, + &ddc_i2c, + CONNECTOR_OBJECT_ID_VGA, + &hpd); + } else if (rdev->flags & RADEON_IS_MOBILITY) { + /* LVDS */ + ddc_i2c = combios_setup_i2c_bus(rdev, DDC_NONE_DETECTED, 0, 0); + hpd.hpd = RADEON_HPD_NONE; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_LCD1_SUPPORT, + 0), + ATOM_DEVICE_LCD1_SUPPORT); + radeon_add_legacy_connector(dev, 0, + ATOM_DEVICE_LCD1_SUPPORT, + DRM_MODE_CONNECTOR_LVDS, + &ddc_i2c, + CONNECTOR_OBJECT_ID_LVDS, + &hpd); + + /* VGA - primary dac */ + ddc_i2c = combios_setup_i2c_bus(rdev, DDC_VGA, 0, 0); + hpd.hpd = RADEON_HPD_NONE; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_CRT1_SUPPORT, + 1), + ATOM_DEVICE_CRT1_SUPPORT); + radeon_add_legacy_connector(dev, 1, + ATOM_DEVICE_CRT1_SUPPORT, + DRM_MODE_CONNECTOR_VGA, + &ddc_i2c, + CONNECTOR_OBJECT_ID_VGA, + &hpd); + } else { + /* DVI-I - tv dac, int tmds */ + ddc_i2c = combios_setup_i2c_bus(rdev, DDC_DVI, 0, 0); + hpd.hpd = RADEON_HPD_1; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_DFP1_SUPPORT, + 0), + ATOM_DEVICE_DFP1_SUPPORT); + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_CRT2_SUPPORT, + 2), + ATOM_DEVICE_CRT2_SUPPORT); + radeon_add_legacy_connector(dev, 0, + ATOM_DEVICE_DFP1_SUPPORT | + ATOM_DEVICE_CRT2_SUPPORT, + DRM_MODE_CONNECTOR_DVII, + &ddc_i2c, + CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I, + &hpd); + + /* VGA - primary dac */ + ddc_i2c = combios_setup_i2c_bus(rdev, DDC_VGA, 0, 0); + hpd.hpd = RADEON_HPD_NONE; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_CRT1_SUPPORT, + 1), + ATOM_DEVICE_CRT1_SUPPORT); + radeon_add_legacy_connector(dev, 1, + ATOM_DEVICE_CRT1_SUPPORT, + DRM_MODE_CONNECTOR_VGA, + &ddc_i2c, + CONNECTOR_OBJECT_ID_VGA, + &hpd); + } + + if (rdev->family != CHIP_R100 && rdev->family != CHIP_R200) { + /* TV - tv dac */ + ddc_i2c.valid = false; + hpd.hpd = RADEON_HPD_NONE; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_TV1_SUPPORT, + 2), + ATOM_DEVICE_TV1_SUPPORT); + radeon_add_legacy_connector(dev, 2, + ATOM_DEVICE_TV1_SUPPORT, + DRM_MODE_CONNECTOR_SVIDEO, + &ddc_i2c, + CONNECTOR_OBJECT_ID_SVIDEO, + &hpd); + } + break; + case CT_IBOOK: + DRM_INFO("Connector Table: %d (ibook)\n", + rdev->mode_info.connector_table); + /* LVDS */ + ddc_i2c = combios_setup_i2c_bus(rdev, DDC_DVI, 0, 0); + hpd.hpd = RADEON_HPD_NONE; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_LCD1_SUPPORT, + 0), + ATOM_DEVICE_LCD1_SUPPORT); + radeon_add_legacy_connector(dev, 0, ATOM_DEVICE_LCD1_SUPPORT, + DRM_MODE_CONNECTOR_LVDS, &ddc_i2c, + CONNECTOR_OBJECT_ID_LVDS, + &hpd); + /* VGA - TV DAC */ + ddc_i2c = combios_setup_i2c_bus(rdev, DDC_VGA, 0, 0); + hpd.hpd = RADEON_HPD_NONE; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_CRT2_SUPPORT, + 2), + ATOM_DEVICE_CRT2_SUPPORT); + radeon_add_legacy_connector(dev, 1, ATOM_DEVICE_CRT2_SUPPORT, + DRM_MODE_CONNECTOR_VGA, &ddc_i2c, + CONNECTOR_OBJECT_ID_VGA, + &hpd); + /* TV - TV DAC */ + ddc_i2c.valid = false; + hpd.hpd = RADEON_HPD_NONE; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_TV1_SUPPORT, + 2), + ATOM_DEVICE_TV1_SUPPORT); + radeon_add_legacy_connector(dev, 2, ATOM_DEVICE_TV1_SUPPORT, + DRM_MODE_CONNECTOR_SVIDEO, + &ddc_i2c, + CONNECTOR_OBJECT_ID_SVIDEO, + &hpd); + break; + case CT_POWERBOOK_EXTERNAL: + DRM_INFO("Connector Table: %d (powerbook external tmds)\n", + rdev->mode_info.connector_table); + /* LVDS */ + ddc_i2c = combios_setup_i2c_bus(rdev, DDC_DVI, 0, 0); + hpd.hpd = RADEON_HPD_NONE; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_LCD1_SUPPORT, + 0), + ATOM_DEVICE_LCD1_SUPPORT); + radeon_add_legacy_connector(dev, 0, ATOM_DEVICE_LCD1_SUPPORT, + DRM_MODE_CONNECTOR_LVDS, &ddc_i2c, + CONNECTOR_OBJECT_ID_LVDS, + &hpd); + /* DVI-I - primary dac, ext tmds */ + ddc_i2c = combios_setup_i2c_bus(rdev, DDC_VGA, 0, 0); + hpd.hpd = RADEON_HPD_2; /* ??? */ + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_DFP2_SUPPORT, + 0), + ATOM_DEVICE_DFP2_SUPPORT); + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_CRT1_SUPPORT, + 1), + ATOM_DEVICE_CRT1_SUPPORT); + /* XXX some are SL */ + radeon_add_legacy_connector(dev, 1, + ATOM_DEVICE_DFP2_SUPPORT | + ATOM_DEVICE_CRT1_SUPPORT, + DRM_MODE_CONNECTOR_DVII, &ddc_i2c, + CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_I, + &hpd); + /* TV - TV DAC */ + ddc_i2c.valid = false; + hpd.hpd = RADEON_HPD_NONE; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_TV1_SUPPORT, + 2), + ATOM_DEVICE_TV1_SUPPORT); + radeon_add_legacy_connector(dev, 2, ATOM_DEVICE_TV1_SUPPORT, + DRM_MODE_CONNECTOR_SVIDEO, + &ddc_i2c, + CONNECTOR_OBJECT_ID_SVIDEO, + &hpd); + break; + case CT_POWERBOOK_INTERNAL: + DRM_INFO("Connector Table: %d (powerbook internal tmds)\n", + rdev->mode_info.connector_table); + /* LVDS */ + ddc_i2c = combios_setup_i2c_bus(rdev, DDC_DVI, 0, 0); + hpd.hpd = RADEON_HPD_NONE; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_LCD1_SUPPORT, + 0), + ATOM_DEVICE_LCD1_SUPPORT); + radeon_add_legacy_connector(dev, 0, ATOM_DEVICE_LCD1_SUPPORT, + DRM_MODE_CONNECTOR_LVDS, &ddc_i2c, + CONNECTOR_OBJECT_ID_LVDS, + &hpd); + /* DVI-I - primary dac, int tmds */ + ddc_i2c = combios_setup_i2c_bus(rdev, DDC_VGA, 0, 0); + hpd.hpd = RADEON_HPD_1; /* ??? */ + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_DFP1_SUPPORT, + 0), + ATOM_DEVICE_DFP1_SUPPORT); + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_CRT1_SUPPORT, + 1), + ATOM_DEVICE_CRT1_SUPPORT); + radeon_add_legacy_connector(dev, 1, + ATOM_DEVICE_DFP1_SUPPORT | + ATOM_DEVICE_CRT1_SUPPORT, + DRM_MODE_CONNECTOR_DVII, &ddc_i2c, + CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I, + &hpd); + /* TV - TV DAC */ + ddc_i2c.valid = false; + hpd.hpd = RADEON_HPD_NONE; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_TV1_SUPPORT, + 2), + ATOM_DEVICE_TV1_SUPPORT); + radeon_add_legacy_connector(dev, 2, ATOM_DEVICE_TV1_SUPPORT, + DRM_MODE_CONNECTOR_SVIDEO, + &ddc_i2c, + CONNECTOR_OBJECT_ID_SVIDEO, + &hpd); + break; + case CT_POWERBOOK_VGA: + DRM_INFO("Connector Table: %d (powerbook vga)\n", + rdev->mode_info.connector_table); + /* LVDS */ + ddc_i2c = combios_setup_i2c_bus(rdev, DDC_DVI, 0, 0); + hpd.hpd = RADEON_HPD_NONE; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_LCD1_SUPPORT, + 0), + ATOM_DEVICE_LCD1_SUPPORT); + radeon_add_legacy_connector(dev, 0, ATOM_DEVICE_LCD1_SUPPORT, + DRM_MODE_CONNECTOR_LVDS, &ddc_i2c, + CONNECTOR_OBJECT_ID_LVDS, + &hpd); + /* VGA - primary dac */ + ddc_i2c = combios_setup_i2c_bus(rdev, DDC_VGA, 0, 0); + hpd.hpd = RADEON_HPD_NONE; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_CRT1_SUPPORT, + 1), + ATOM_DEVICE_CRT1_SUPPORT); + radeon_add_legacy_connector(dev, 1, ATOM_DEVICE_CRT1_SUPPORT, + DRM_MODE_CONNECTOR_VGA, &ddc_i2c, + CONNECTOR_OBJECT_ID_VGA, + &hpd); + /* TV - TV DAC */ + ddc_i2c.valid = false; + hpd.hpd = RADEON_HPD_NONE; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_TV1_SUPPORT, + 2), + ATOM_DEVICE_TV1_SUPPORT); + radeon_add_legacy_connector(dev, 2, ATOM_DEVICE_TV1_SUPPORT, + DRM_MODE_CONNECTOR_SVIDEO, + &ddc_i2c, + CONNECTOR_OBJECT_ID_SVIDEO, + &hpd); + break; + case CT_MINI_EXTERNAL: + DRM_INFO("Connector Table: %d (mini external tmds)\n", + rdev->mode_info.connector_table); + /* DVI-I - tv dac, ext tmds */ + ddc_i2c = combios_setup_i2c_bus(rdev, DDC_CRT2, 0, 0); + hpd.hpd = RADEON_HPD_2; /* ??? */ + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_DFP2_SUPPORT, + 0), + ATOM_DEVICE_DFP2_SUPPORT); + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_CRT2_SUPPORT, + 2), + ATOM_DEVICE_CRT2_SUPPORT); + /* XXX are any DL? */ + radeon_add_legacy_connector(dev, 0, + ATOM_DEVICE_DFP2_SUPPORT | + ATOM_DEVICE_CRT2_SUPPORT, + DRM_MODE_CONNECTOR_DVII, &ddc_i2c, + CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I, + &hpd); + /* TV - TV DAC */ + ddc_i2c.valid = false; + hpd.hpd = RADEON_HPD_NONE; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_TV1_SUPPORT, + 2), + ATOM_DEVICE_TV1_SUPPORT); + radeon_add_legacy_connector(dev, 1, ATOM_DEVICE_TV1_SUPPORT, + DRM_MODE_CONNECTOR_SVIDEO, + &ddc_i2c, + CONNECTOR_OBJECT_ID_SVIDEO, + &hpd); + break; + case CT_MINI_INTERNAL: + DRM_INFO("Connector Table: %d (mini internal tmds)\n", + rdev->mode_info.connector_table); + /* DVI-I - tv dac, int tmds */ + ddc_i2c = combios_setup_i2c_bus(rdev, DDC_CRT2, 0, 0); + hpd.hpd = RADEON_HPD_1; /* ??? */ + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_DFP1_SUPPORT, + 0), + ATOM_DEVICE_DFP1_SUPPORT); + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_CRT2_SUPPORT, + 2), + ATOM_DEVICE_CRT2_SUPPORT); + radeon_add_legacy_connector(dev, 0, + ATOM_DEVICE_DFP1_SUPPORT | + ATOM_DEVICE_CRT2_SUPPORT, + DRM_MODE_CONNECTOR_DVII, &ddc_i2c, + CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I, + &hpd); + /* TV - TV DAC */ + ddc_i2c.valid = false; + hpd.hpd = RADEON_HPD_NONE; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_TV1_SUPPORT, + 2), + ATOM_DEVICE_TV1_SUPPORT); + radeon_add_legacy_connector(dev, 1, ATOM_DEVICE_TV1_SUPPORT, + DRM_MODE_CONNECTOR_SVIDEO, + &ddc_i2c, + CONNECTOR_OBJECT_ID_SVIDEO, + &hpd); + break; + case CT_IMAC_G5_ISIGHT: + DRM_INFO("Connector Table: %d (imac g5 isight)\n", + rdev->mode_info.connector_table); + /* DVI-D - int tmds */ + ddc_i2c = combios_setup_i2c_bus(rdev, DDC_MONID, 0, 0); + hpd.hpd = RADEON_HPD_1; /* ??? */ + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_DFP1_SUPPORT, + 0), + ATOM_DEVICE_DFP1_SUPPORT); + radeon_add_legacy_connector(dev, 0, ATOM_DEVICE_DFP1_SUPPORT, + DRM_MODE_CONNECTOR_DVID, &ddc_i2c, + CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D, + &hpd); + /* VGA - tv dac */ + ddc_i2c = combios_setup_i2c_bus(rdev, DDC_DVI, 0, 0); + hpd.hpd = RADEON_HPD_NONE; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_CRT2_SUPPORT, + 2), + ATOM_DEVICE_CRT2_SUPPORT); + radeon_add_legacy_connector(dev, 1, ATOM_DEVICE_CRT2_SUPPORT, + DRM_MODE_CONNECTOR_VGA, &ddc_i2c, + CONNECTOR_OBJECT_ID_VGA, + &hpd); + /* TV - TV DAC */ + ddc_i2c.valid = false; + hpd.hpd = RADEON_HPD_NONE; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_TV1_SUPPORT, + 2), + ATOM_DEVICE_TV1_SUPPORT); + radeon_add_legacy_connector(dev, 2, ATOM_DEVICE_TV1_SUPPORT, + DRM_MODE_CONNECTOR_SVIDEO, + &ddc_i2c, + CONNECTOR_OBJECT_ID_SVIDEO, + &hpd); + break; + case CT_EMAC: + DRM_INFO("Connector Table: %d (emac)\n", + rdev->mode_info.connector_table); + /* VGA - primary dac */ + ddc_i2c = combios_setup_i2c_bus(rdev, DDC_VGA, 0, 0); + hpd.hpd = RADEON_HPD_NONE; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_CRT1_SUPPORT, + 1), + ATOM_DEVICE_CRT1_SUPPORT); + radeon_add_legacy_connector(dev, 0, ATOM_DEVICE_CRT1_SUPPORT, + DRM_MODE_CONNECTOR_VGA, &ddc_i2c, + CONNECTOR_OBJECT_ID_VGA, + &hpd); + /* VGA - tv dac */ + ddc_i2c = combios_setup_i2c_bus(rdev, DDC_CRT2, 0, 0); + hpd.hpd = RADEON_HPD_NONE; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_CRT2_SUPPORT, + 2), + ATOM_DEVICE_CRT2_SUPPORT); + radeon_add_legacy_connector(dev, 1, ATOM_DEVICE_CRT2_SUPPORT, + DRM_MODE_CONNECTOR_VGA, &ddc_i2c, + CONNECTOR_OBJECT_ID_VGA, + &hpd); + /* TV - TV DAC */ + ddc_i2c.valid = false; + hpd.hpd = RADEON_HPD_NONE; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_TV1_SUPPORT, + 2), + ATOM_DEVICE_TV1_SUPPORT); + radeon_add_legacy_connector(dev, 2, ATOM_DEVICE_TV1_SUPPORT, + DRM_MODE_CONNECTOR_SVIDEO, + &ddc_i2c, + CONNECTOR_OBJECT_ID_SVIDEO, + &hpd); + break; + case CT_RN50_POWER: + DRM_INFO("Connector Table: %d (rn50-power)\n", + rdev->mode_info.connector_table); + /* VGA - primary dac */ + ddc_i2c = combios_setup_i2c_bus(rdev, DDC_VGA, 0, 0); + hpd.hpd = RADEON_HPD_NONE; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_CRT1_SUPPORT, + 1), + ATOM_DEVICE_CRT1_SUPPORT); + radeon_add_legacy_connector(dev, 0, ATOM_DEVICE_CRT1_SUPPORT, + DRM_MODE_CONNECTOR_VGA, &ddc_i2c, + CONNECTOR_OBJECT_ID_VGA, + &hpd); + ddc_i2c = combios_setup_i2c_bus(rdev, DDC_CRT2, 0, 0); + hpd.hpd = RADEON_HPD_NONE; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_CRT2_SUPPORT, + 2), + ATOM_DEVICE_CRT2_SUPPORT); + radeon_add_legacy_connector(dev, 1, ATOM_DEVICE_CRT2_SUPPORT, + DRM_MODE_CONNECTOR_VGA, &ddc_i2c, + CONNECTOR_OBJECT_ID_VGA, + &hpd); + break; + case CT_MAC_X800: + DRM_INFO("Connector Table: %d (mac x800)\n", + rdev->mode_info.connector_table); + /* DVI - primary dac, internal tmds */ + ddc_i2c = combios_setup_i2c_bus(rdev, DDC_DVI, 0, 0); + hpd.hpd = RADEON_HPD_1; /* ??? */ + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_DFP1_SUPPORT, + 0), + ATOM_DEVICE_DFP1_SUPPORT); + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_CRT1_SUPPORT, + 1), + ATOM_DEVICE_CRT1_SUPPORT); + radeon_add_legacy_connector(dev, 0, + ATOM_DEVICE_DFP1_SUPPORT | + ATOM_DEVICE_CRT1_SUPPORT, + DRM_MODE_CONNECTOR_DVII, &ddc_i2c, + CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I, + &hpd); + /* DVI - tv dac, dvo */ + ddc_i2c = combios_setup_i2c_bus(rdev, DDC_MONID, 0, 0); + hpd.hpd = RADEON_HPD_2; /* ??? */ + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_DFP2_SUPPORT, + 0), + ATOM_DEVICE_DFP2_SUPPORT); + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_CRT2_SUPPORT, + 2), + ATOM_DEVICE_CRT2_SUPPORT); + radeon_add_legacy_connector(dev, 1, + ATOM_DEVICE_DFP2_SUPPORT | + ATOM_DEVICE_CRT2_SUPPORT, + DRM_MODE_CONNECTOR_DVII, &ddc_i2c, + CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_I, + &hpd); + break; + case CT_MAC_G5_9600: + DRM_INFO("Connector Table: %d (mac g5 9600)\n", + rdev->mode_info.connector_table); + /* DVI - tv dac, dvo */ + ddc_i2c = combios_setup_i2c_bus(rdev, DDC_DVI, 0, 0); + hpd.hpd = RADEON_HPD_1; /* ??? */ + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_DFP2_SUPPORT, + 0), + ATOM_DEVICE_DFP2_SUPPORT); + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_CRT2_SUPPORT, + 2), + ATOM_DEVICE_CRT2_SUPPORT); + radeon_add_legacy_connector(dev, 0, + ATOM_DEVICE_DFP2_SUPPORT | + ATOM_DEVICE_CRT2_SUPPORT, + DRM_MODE_CONNECTOR_DVII, &ddc_i2c, + CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I, + &hpd); + /* ADC - primary dac, internal tmds */ + ddc_i2c = combios_setup_i2c_bus(rdev, DDC_VGA, 0, 0); + hpd.hpd = RADEON_HPD_2; /* ??? */ + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_DFP1_SUPPORT, + 0), + ATOM_DEVICE_DFP1_SUPPORT); + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_CRT1_SUPPORT, + 1), + ATOM_DEVICE_CRT1_SUPPORT); + radeon_add_legacy_connector(dev, 1, + ATOM_DEVICE_DFP1_SUPPORT | + ATOM_DEVICE_CRT1_SUPPORT, + DRM_MODE_CONNECTOR_DVII, &ddc_i2c, + CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I, + &hpd); + /* TV - TV DAC */ + ddc_i2c.valid = false; + hpd.hpd = RADEON_HPD_NONE; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_TV1_SUPPORT, + 2), + ATOM_DEVICE_TV1_SUPPORT); + radeon_add_legacy_connector(dev, 2, ATOM_DEVICE_TV1_SUPPORT, + DRM_MODE_CONNECTOR_SVIDEO, + &ddc_i2c, + CONNECTOR_OBJECT_ID_SVIDEO, + &hpd); + break; + default: + DRM_INFO("Connector table: %d (invalid)\n", + rdev->mode_info.connector_table); + return false; + } + + radeon_link_encoder_connector(dev); + + return true; +} + +static bool radeon_apply_legacy_quirks(struct drm_device *dev, + int bios_index, + enum radeon_combios_connector + *legacy_connector, + struct radeon_i2c_bus_rec *ddc_i2c, + struct radeon_hpd *hpd) +{ + + /* Certain IBM chipset RN50s have a BIOS reporting two VGAs, + one with VGA DDC and one with CRT2 DDC. - kill the CRT2 DDC one */ + if (dev->pdev->device == 0x515e && + dev->pdev->subsystem_vendor == 0x1014) { + if (*legacy_connector == CONNECTOR_CRT_LEGACY && + ddc_i2c->mask_clk_reg == RADEON_GPIO_CRT2_DDC) + return false; + } + + /* X300 card with extra non-existent DVI port */ + if (dev->pdev->device == 0x5B60 && + dev->pdev->subsystem_vendor == 0x17af && + dev->pdev->subsystem_device == 0x201e && bios_index == 2) { + if (*legacy_connector == CONNECTOR_DVI_I_LEGACY) + return false; + } + + return true; +} + +static bool radeon_apply_legacy_tv_quirks(struct drm_device *dev) +{ + /* Acer 5102 has non-existent TV port */ + if (dev->pdev->device == 0x5975 && + dev->pdev->subsystem_vendor == 0x1025 && + dev->pdev->subsystem_device == 0x009f) + return false; + + /* HP dc5750 has non-existent TV port */ + if (dev->pdev->device == 0x5974 && + dev->pdev->subsystem_vendor == 0x103c && + dev->pdev->subsystem_device == 0x280a) + return false; + + /* MSI S270 has non-existent TV port */ + if (dev->pdev->device == 0x5955 && + dev->pdev->subsystem_vendor == 0x1462 && + dev->pdev->subsystem_device == 0x0131) + return false; + + return true; +} + +static uint16_t combios_check_dl_dvi(struct drm_device *dev, int is_dvi_d) +{ + struct radeon_device *rdev = dev->dev_private; + uint32_t ext_tmds_info; + + if (rdev->flags & RADEON_IS_IGP) { + if (is_dvi_d) + return CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D; + else + return CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I; + } + ext_tmds_info = combios_get_table_offset(dev, COMBIOS_EXT_TMDS_INFO_TABLE); + if (ext_tmds_info) { + uint8_t rev = RBIOS8(ext_tmds_info); + uint8_t flags = RBIOS8(ext_tmds_info + 4 + 5); + if (rev >= 3) { + if (is_dvi_d) + return CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_D; + else + return CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_I; + } else { + if (flags & 1) { + if (is_dvi_d) + return CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_D; + else + return CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_I; + } + } + } + if (is_dvi_d) + return CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D; + else + return CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I; +} + +bool radeon_get_legacy_connector_info_from_bios(struct drm_device *dev) +{ + struct radeon_device *rdev = dev->dev_private; + uint32_t conn_info, entry, devices; + uint16_t tmp, connector_object_id; + enum radeon_combios_ddc ddc_type; + enum radeon_combios_connector connector; + int i = 0; + struct radeon_i2c_bus_rec ddc_i2c; + struct radeon_hpd hpd; + + conn_info = combios_get_table_offset(dev, COMBIOS_CONNECTOR_INFO_TABLE); + if (conn_info) { + for (i = 0; i < 4; i++) { + entry = conn_info + 2 + i * 2; + + if (!RBIOS16(entry)) + break; + + tmp = RBIOS16(entry); + + connector = (tmp >> 12) & 0xf; + + ddc_type = (tmp >> 8) & 0xf; + ddc_i2c = combios_setup_i2c_bus(rdev, ddc_type, 0, 0); + + switch (connector) { + case CONNECTOR_PROPRIETARY_LEGACY: + case CONNECTOR_DVI_I_LEGACY: + case CONNECTOR_DVI_D_LEGACY: + if ((tmp >> 4) & 0x1) + hpd.hpd = RADEON_HPD_2; + else + hpd.hpd = RADEON_HPD_1; + break; + default: + hpd.hpd = RADEON_HPD_NONE; + break; + } + + if (!radeon_apply_legacy_quirks(dev, i, &connector, + &ddc_i2c, &hpd)) + continue; + + switch (connector) { + case CONNECTOR_PROPRIETARY_LEGACY: + if ((tmp >> 4) & 0x1) + devices = ATOM_DEVICE_DFP2_SUPPORT; + else + devices = ATOM_DEVICE_DFP1_SUPPORT; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum + (dev, devices, 0), + devices); + radeon_add_legacy_connector(dev, i, devices, + legacy_connector_convert + [connector], + &ddc_i2c, + CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D, + &hpd); + break; + case CONNECTOR_CRT_LEGACY: + if (tmp & 0x1) { + devices = ATOM_DEVICE_CRT2_SUPPORT; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum + (dev, + ATOM_DEVICE_CRT2_SUPPORT, + 2), + ATOM_DEVICE_CRT2_SUPPORT); + } else { + devices = ATOM_DEVICE_CRT1_SUPPORT; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum + (dev, + ATOM_DEVICE_CRT1_SUPPORT, + 1), + ATOM_DEVICE_CRT1_SUPPORT); + } + radeon_add_legacy_connector(dev, + i, + devices, + legacy_connector_convert + [connector], + &ddc_i2c, + CONNECTOR_OBJECT_ID_VGA, + &hpd); + break; + case CONNECTOR_DVI_I_LEGACY: + devices = 0; + if (tmp & 0x1) { + devices |= ATOM_DEVICE_CRT2_SUPPORT; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum + (dev, + ATOM_DEVICE_CRT2_SUPPORT, + 2), + ATOM_DEVICE_CRT2_SUPPORT); + } else { + devices |= ATOM_DEVICE_CRT1_SUPPORT; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum + (dev, + ATOM_DEVICE_CRT1_SUPPORT, + 1), + ATOM_DEVICE_CRT1_SUPPORT); + } + if ((tmp >> 4) & 0x1) { + devices |= ATOM_DEVICE_DFP2_SUPPORT; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum + (dev, + ATOM_DEVICE_DFP2_SUPPORT, + 0), + ATOM_DEVICE_DFP2_SUPPORT); + connector_object_id = combios_check_dl_dvi(dev, 0); + } else { + devices |= ATOM_DEVICE_DFP1_SUPPORT; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum + (dev, + ATOM_DEVICE_DFP1_SUPPORT, + 0), + ATOM_DEVICE_DFP1_SUPPORT); + connector_object_id = CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I; + } + radeon_add_legacy_connector(dev, + i, + devices, + legacy_connector_convert + [connector], + &ddc_i2c, + connector_object_id, + &hpd); + break; + case CONNECTOR_DVI_D_LEGACY: + if ((tmp >> 4) & 0x1) { + devices = ATOM_DEVICE_DFP2_SUPPORT; + connector_object_id = combios_check_dl_dvi(dev, 1); + } else { + devices = ATOM_DEVICE_DFP1_SUPPORT; + connector_object_id = CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I; + } + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum + (dev, devices, 0), + devices); + radeon_add_legacy_connector(dev, i, devices, + legacy_connector_convert + [connector], + &ddc_i2c, + connector_object_id, + &hpd); + break; + case CONNECTOR_CTV_LEGACY: + case CONNECTOR_STV_LEGACY: + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum + (dev, + ATOM_DEVICE_TV1_SUPPORT, + 2), + ATOM_DEVICE_TV1_SUPPORT); + radeon_add_legacy_connector(dev, i, + ATOM_DEVICE_TV1_SUPPORT, + legacy_connector_convert + [connector], + &ddc_i2c, + CONNECTOR_OBJECT_ID_SVIDEO, + &hpd); + break; + default: + DRM_ERROR("Unknown connector type: %d\n", + connector); + continue; + } + + } + } else { + uint16_t tmds_info = + combios_get_table_offset(dev, COMBIOS_DFP_INFO_TABLE); + if (tmds_info) { + DRM_DEBUG_KMS("Found DFP table, assuming DVI connector\n"); + + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_CRT1_SUPPORT, + 1), + ATOM_DEVICE_CRT1_SUPPORT); + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_DFP1_SUPPORT, + 0), + ATOM_DEVICE_DFP1_SUPPORT); + + ddc_i2c = combios_setup_i2c_bus(rdev, DDC_DVI, 0, 0); + hpd.hpd = RADEON_HPD_1; + radeon_add_legacy_connector(dev, + 0, + ATOM_DEVICE_CRT1_SUPPORT | + ATOM_DEVICE_DFP1_SUPPORT, + DRM_MODE_CONNECTOR_DVII, + &ddc_i2c, + CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I, + &hpd); + } else { + uint16_t crt_info = + combios_get_table_offset(dev, COMBIOS_CRT_INFO_TABLE); + DRM_DEBUG_KMS("Found CRT table, assuming VGA connector\n"); + if (crt_info) { + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_CRT1_SUPPORT, + 1), + ATOM_DEVICE_CRT1_SUPPORT); + ddc_i2c = combios_setup_i2c_bus(rdev, DDC_VGA, 0, 0); + hpd.hpd = RADEON_HPD_NONE; + radeon_add_legacy_connector(dev, + 0, + ATOM_DEVICE_CRT1_SUPPORT, + DRM_MODE_CONNECTOR_VGA, + &ddc_i2c, + CONNECTOR_OBJECT_ID_VGA, + &hpd); + } else { + DRM_DEBUG_KMS("No connector info found\n"); + return false; + } + } + } + + if (rdev->flags & RADEON_IS_MOBILITY || rdev->flags & RADEON_IS_IGP) { + uint16_t lcd_info = + combios_get_table_offset(dev, COMBIOS_LCD_INFO_TABLE); + if (lcd_info) { + uint16_t lcd_ddc_info = + combios_get_table_offset(dev, + COMBIOS_LCD_DDC_INFO_TABLE); + + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum(dev, + ATOM_DEVICE_LCD1_SUPPORT, + 0), + ATOM_DEVICE_LCD1_SUPPORT); + + if (lcd_ddc_info) { + ddc_type = RBIOS8(lcd_ddc_info + 2); + switch (ddc_type) { + case DDC_LCD: + ddc_i2c = + combios_setup_i2c_bus(rdev, + DDC_LCD, + RBIOS32(lcd_ddc_info + 3), + RBIOS32(lcd_ddc_info + 7)); + radeon_i2c_add(rdev, &ddc_i2c, "LCD"); + break; + case DDC_GPIO: + ddc_i2c = + combios_setup_i2c_bus(rdev, + DDC_GPIO, + RBIOS32(lcd_ddc_info + 3), + RBIOS32(lcd_ddc_info + 7)); + radeon_i2c_add(rdev, &ddc_i2c, "LCD"); + break; + default: + ddc_i2c = + combios_setup_i2c_bus(rdev, ddc_type, 0, 0); + break; + } + DRM_DEBUG_KMS("LCD DDC Info Table found!\n"); + } else + ddc_i2c.valid = false; + + hpd.hpd = RADEON_HPD_NONE; + radeon_add_legacy_connector(dev, + 5, + ATOM_DEVICE_LCD1_SUPPORT, + DRM_MODE_CONNECTOR_LVDS, + &ddc_i2c, + CONNECTOR_OBJECT_ID_LVDS, + &hpd); + } + } + + /* check TV table */ + if (rdev->family != CHIP_R100 && rdev->family != CHIP_R200) { + uint32_t tv_info = + combios_get_table_offset(dev, COMBIOS_TV_INFO_TABLE); + if (tv_info) { + if (RBIOS8(tv_info + 6) == 'T') { + if (radeon_apply_legacy_tv_quirks(dev)) { + hpd.hpd = RADEON_HPD_NONE; + ddc_i2c.valid = false; + radeon_add_legacy_encoder(dev, + radeon_get_encoder_enum + (dev, + ATOM_DEVICE_TV1_SUPPORT, + 2), + ATOM_DEVICE_TV1_SUPPORT); + radeon_add_legacy_connector(dev, 6, + ATOM_DEVICE_TV1_SUPPORT, + DRM_MODE_CONNECTOR_SVIDEO, + &ddc_i2c, + CONNECTOR_OBJECT_ID_SVIDEO, + &hpd); + } + } + } + } + + radeon_link_encoder_connector(dev); + + return true; +} + +static const char *thermal_controller_names[] = { + "NONE", + "lm63", + "adm1032", +}; + +void radeon_combios_get_power_modes(struct radeon_device *rdev) +{ + struct drm_device *dev = rdev->ddev; + u16 offset, misc, misc2 = 0; + u8 rev, blocks, tmp; + int state_index = 0; + + rdev->pm.default_power_state_index = -1; + + /* allocate 2 power states */ + rdev->pm.power_state = kzalloc(sizeof(struct radeon_power_state) * 2, GFP_KERNEL); + if (!rdev->pm.power_state) { + rdev->pm.default_power_state_index = state_index; + rdev->pm.num_power_states = 0; + + rdev->pm.current_power_state_index = rdev->pm.default_power_state_index; + rdev->pm.current_clock_mode_index = 0; + return; + } + + /* check for a thermal chip */ + offset = combios_get_table_offset(dev, COMBIOS_OVERDRIVE_INFO_TABLE); + if (offset) { + u8 thermal_controller = 0, gpio = 0, i2c_addr = 0, clk_bit = 0, data_bit = 0; + struct radeon_i2c_bus_rec i2c_bus; + + rev = RBIOS8(offset); + + if (rev == 0) { + thermal_controller = RBIOS8(offset + 3); + gpio = RBIOS8(offset + 4) & 0x3f; + i2c_addr = RBIOS8(offset + 5); + } else if (rev == 1) { + thermal_controller = RBIOS8(offset + 4); + gpio = RBIOS8(offset + 5) & 0x3f; + i2c_addr = RBIOS8(offset + 6); + } else if (rev == 2) { + thermal_controller = RBIOS8(offset + 4); + gpio = RBIOS8(offset + 5) & 0x3f; + i2c_addr = RBIOS8(offset + 6); + clk_bit = RBIOS8(offset + 0xa); + data_bit = RBIOS8(offset + 0xb); + } + if ((thermal_controller > 0) && (thermal_controller < 3)) { + DRM_INFO("Possible %s thermal controller at 0x%02x\n", + thermal_controller_names[thermal_controller], + i2c_addr >> 1); + if (gpio == DDC_LCD) { + /* MM i2c */ + i2c_bus.valid = true; + i2c_bus.hw_capable = true; + i2c_bus.mm_i2c = true; + i2c_bus.i2c_id = 0xa0; + } else if (gpio == DDC_GPIO) + i2c_bus = combios_setup_i2c_bus(rdev, gpio, 1 << clk_bit, 1 << data_bit); + else + i2c_bus = combios_setup_i2c_bus(rdev, gpio, 0, 0); + rdev->pm.i2c_bus = radeon_i2c_lookup(rdev, &i2c_bus); + if (rdev->pm.i2c_bus) { + struct i2c_board_info info = { }; + const char *name = thermal_controller_names[thermal_controller]; + info.addr = i2c_addr >> 1; + strlcpy(info.type, name, sizeof(info.type)); + i2c_new_device(&rdev->pm.i2c_bus->adapter, &info); + } + } + } + + if (rdev->flags & RADEON_IS_MOBILITY) { + offset = combios_get_table_offset(dev, COMBIOS_POWERPLAY_INFO_TABLE); + if (offset) { + rev = RBIOS8(offset); + blocks = RBIOS8(offset + 0x2); + /* power mode 0 tends to be the only valid one */ + rdev->pm.power_state[state_index].num_clock_modes = 1; + rdev->pm.power_state[state_index].clock_info[0].mclk = RBIOS32(offset + 0x5 + 0x2); + rdev->pm.power_state[state_index].clock_info[0].sclk = RBIOS32(offset + 0x5 + 0x6); + if ((rdev->pm.power_state[state_index].clock_info[0].mclk == 0) || + (rdev->pm.power_state[state_index].clock_info[0].sclk == 0)) + goto default_mode; + rdev->pm.power_state[state_index].type = + POWER_STATE_TYPE_BATTERY; + misc = RBIOS16(offset + 0x5 + 0x0); + if (rev > 4) + misc2 = RBIOS16(offset + 0x5 + 0xe); + rdev->pm.power_state[state_index].misc = misc; + rdev->pm.power_state[state_index].misc2 = misc2; + if (misc & 0x4) { + rdev->pm.power_state[state_index].clock_info[0].voltage.type = VOLTAGE_GPIO; + if (misc & 0x8) + rdev->pm.power_state[state_index].clock_info[0].voltage.active_high = + true; + else + rdev->pm.power_state[state_index].clock_info[0].voltage.active_high = + false; + rdev->pm.power_state[state_index].clock_info[0].voltage.gpio.valid = true; + if (rev < 6) { + rdev->pm.power_state[state_index].clock_info[0].voltage.gpio.reg = + RBIOS16(offset + 0x5 + 0xb) * 4; + tmp = RBIOS8(offset + 0x5 + 0xd); + rdev->pm.power_state[state_index].clock_info[0].voltage.gpio.mask = (1 << tmp); + } else { + u8 entries = RBIOS8(offset + 0x5 + 0xb); + u16 voltage_table_offset = RBIOS16(offset + 0x5 + 0xc); + if (entries && voltage_table_offset) { + rdev->pm.power_state[state_index].clock_info[0].voltage.gpio.reg = + RBIOS16(voltage_table_offset) * 4; + tmp = RBIOS8(voltage_table_offset + 0x2); + rdev->pm.power_state[state_index].clock_info[0].voltage.gpio.mask = (1 << tmp); + } else + rdev->pm.power_state[state_index].clock_info[0].voltage.gpio.valid = false; + } + switch ((misc2 & 0x700) >> 8) { + case 0: + default: + rdev->pm.power_state[state_index].clock_info[0].voltage.delay = 0; + break; + case 1: + rdev->pm.power_state[state_index].clock_info[0].voltage.delay = 33; + break; + case 2: + rdev->pm.power_state[state_index].clock_info[0].voltage.delay = 66; + break; + case 3: + rdev->pm.power_state[state_index].clock_info[0].voltage.delay = 99; + break; + case 4: + rdev->pm.power_state[state_index].clock_info[0].voltage.delay = 132; + break; + } + } else + rdev->pm.power_state[state_index].clock_info[0].voltage.type = VOLTAGE_NONE; + if (rev > 6) + rdev->pm.power_state[state_index].pcie_lanes = + RBIOS8(offset + 0x5 + 0x10); + rdev->pm.power_state[state_index].flags = RADEON_PM_STATE_SINGLE_DISPLAY_ONLY; + state_index++; + } else { + /* XXX figure out some good default low power mode for mobility cards w/out power tables */ + } + } else { + /* XXX figure out some good default low power mode for desktop cards */ + } + +default_mode: + /* add the default mode */ + rdev->pm.power_state[state_index].type = + POWER_STATE_TYPE_DEFAULT; + rdev->pm.power_state[state_index].num_clock_modes = 1; + rdev->pm.power_state[state_index].clock_info[0].mclk = rdev->clock.default_mclk; + rdev->pm.power_state[state_index].clock_info[0].sclk = rdev->clock.default_sclk; + rdev->pm.power_state[state_index].default_clock_mode = &rdev->pm.power_state[state_index].clock_info[0]; + if ((state_index > 0) && + (rdev->pm.power_state[0].clock_info[0].voltage.type == VOLTAGE_GPIO)) + rdev->pm.power_state[state_index].clock_info[0].voltage = + rdev->pm.power_state[0].clock_info[0].voltage; + else + rdev->pm.power_state[state_index].clock_info[0].voltage.type = VOLTAGE_NONE; + rdev->pm.power_state[state_index].pcie_lanes = 16; + rdev->pm.power_state[state_index].flags = 0; + rdev->pm.default_power_state_index = state_index; + rdev->pm.num_power_states = state_index + 1; + + rdev->pm.current_power_state_index = rdev->pm.default_power_state_index; + rdev->pm.current_clock_mode_index = 0; +} + +void radeon_external_tmds_setup(struct drm_encoder *encoder) +{ + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct radeon_encoder_ext_tmds *tmds = radeon_encoder->enc_priv; + + if (!tmds) + return; + + switch (tmds->dvo_chip) { + case DVO_SIL164: + /* sil 164 */ + radeon_i2c_put_byte(tmds->i2c_bus, + tmds->slave_addr, + 0x08, 0x30); + radeon_i2c_put_byte(tmds->i2c_bus, + tmds->slave_addr, + 0x09, 0x00); + radeon_i2c_put_byte(tmds->i2c_bus, + tmds->slave_addr, + 0x0a, 0x90); + radeon_i2c_put_byte(tmds->i2c_bus, + tmds->slave_addr, + 0x0c, 0x89); + radeon_i2c_put_byte(tmds->i2c_bus, + tmds->slave_addr, + 0x08, 0x3b); + break; + case DVO_SIL1178: + /* sil 1178 - untested */ + /* + * 0x0f, 0x44 + * 0x0f, 0x4c + * 0x0e, 0x01 + * 0x0a, 0x80 + * 0x09, 0x30 + * 0x0c, 0xc9 + * 0x0d, 0x70 + * 0x08, 0x32 + * 0x08, 0x33 + */ + break; + default: + break; + } + +} + +bool radeon_combios_external_tmds_setup(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + uint16_t offset; + uint8_t blocks, slave_addr, rev; + uint32_t index, id; + uint32_t reg, val, and_mask, or_mask; + struct radeon_encoder_ext_tmds *tmds = radeon_encoder->enc_priv; + + if (!tmds) + return false; + + if (rdev->flags & RADEON_IS_IGP) { + offset = combios_get_table_offset(dev, COMBIOS_TMDS_POWER_ON_TABLE); + rev = RBIOS8(offset); + if (offset) { + rev = RBIOS8(offset); + if (rev > 1) { + blocks = RBIOS8(offset + 3); + index = offset + 4; + while (blocks > 0) { + id = RBIOS16(index); + index += 2; + switch (id >> 13) { + case 0: + reg = (id & 0x1fff) * 4; + val = RBIOS32(index); + index += 4; + WREG32(reg, val); + break; + case 2: + reg = (id & 0x1fff) * 4; + and_mask = RBIOS32(index); + index += 4; + or_mask = RBIOS32(index); + index += 4; + val = RREG32(reg); + val = (val & and_mask) | or_mask; + WREG32(reg, val); + break; + case 3: + val = RBIOS16(index); + index += 2; + udelay(val); + break; + case 4: + val = RBIOS16(index); + index += 2; + udelay(val * 1000); + break; + case 6: + slave_addr = id & 0xff; + slave_addr >>= 1; /* 7 bit addressing */ + index++; + reg = RBIOS8(index); + index++; + val = RBIOS8(index); + index++; + radeon_i2c_put_byte(tmds->i2c_bus, + slave_addr, + reg, val); + break; + default: + DRM_ERROR("Unknown id %d\n", id >> 13); + break; + } + blocks--; + } + return true; + } + } + } else { + offset = combios_get_table_offset(dev, COMBIOS_EXT_TMDS_INFO_TABLE); + if (offset) { + index = offset + 10; + id = RBIOS16(index); + while (id != 0xffff) { + index += 2; + switch (id >> 13) { + case 0: + reg = (id & 0x1fff) * 4; + val = RBIOS32(index); + WREG32(reg, val); + break; + case 2: + reg = (id & 0x1fff) * 4; + and_mask = RBIOS32(index); + index += 4; + or_mask = RBIOS32(index); + index += 4; + val = RREG32(reg); + val = (val & and_mask) | or_mask; + WREG32(reg, val); + break; + case 4: + val = RBIOS16(index); + index += 2; + udelay(val); + break; + case 5: + reg = id & 0x1fff; + and_mask = RBIOS32(index); + index += 4; + or_mask = RBIOS32(index); + index += 4; + val = RREG32_PLL(reg); + val = (val & and_mask) | or_mask; + WREG32_PLL(reg, val); + break; + case 6: + reg = id & 0x1fff; + val = RBIOS8(index); + index += 1; + radeon_i2c_put_byte(tmds->i2c_bus, + tmds->slave_addr, + reg, val); + break; + default: + DRM_ERROR("Unknown id %d\n", id >> 13); + break; + } + id = RBIOS16(index); + } + return true; + } + } + return false; +} + +static void combios_parse_mmio_table(struct drm_device *dev, uint16_t offset) +{ + struct radeon_device *rdev = dev->dev_private; + + if (offset) { + while (RBIOS16(offset)) { + uint16_t cmd = ((RBIOS16(offset) & 0xe000) >> 13); + uint32_t addr = (RBIOS16(offset) & 0x1fff); + uint32_t val, and_mask, or_mask; + uint32_t tmp; + + offset += 2; + switch (cmd) { + case 0: + val = RBIOS32(offset); + offset += 4; + WREG32(addr, val); + break; + case 1: + val = RBIOS32(offset); + offset += 4; + WREG32(addr, val); + break; + case 2: + and_mask = RBIOS32(offset); + offset += 4; + or_mask = RBIOS32(offset); + offset += 4; + tmp = RREG32(addr); + tmp &= and_mask; + tmp |= or_mask; + WREG32(addr, tmp); + break; + case 3: + and_mask = RBIOS32(offset); + offset += 4; + or_mask = RBIOS32(offset); + offset += 4; + tmp = RREG32(addr); + tmp &= and_mask; + tmp |= or_mask; + WREG32(addr, tmp); + break; + case 4: + val = RBIOS16(offset); + offset += 2; + udelay(val); + break; + case 5: + val = RBIOS16(offset); + offset += 2; + switch (addr) { + case 8: + while (val--) { + if (! + (RREG32_PLL + (RADEON_CLK_PWRMGT_CNTL) & + RADEON_MC_BUSY)) + break; + } + break; + case 9: + while (val--) { + if ((RREG32(RADEON_MC_STATUS) & + RADEON_MC_IDLE)) + break; + } + break; + default: + break; + } + break; + default: + break; + } + } + } +} + +static void combios_parse_pll_table(struct drm_device *dev, uint16_t offset) +{ + struct radeon_device *rdev = dev->dev_private; + + if (offset) { + while (RBIOS8(offset)) { + uint8_t cmd = ((RBIOS8(offset) & 0xc0) >> 6); + uint8_t addr = (RBIOS8(offset) & 0x3f); + uint32_t val, shift, tmp; + uint32_t and_mask, or_mask; + + offset++; + switch (cmd) { + case 0: + val = RBIOS32(offset); + offset += 4; + WREG32_PLL(addr, val); + break; + case 1: + shift = RBIOS8(offset) * 8; + offset++; + and_mask = RBIOS8(offset) << shift; + and_mask |= ~(0xff << shift); + offset++; + or_mask = RBIOS8(offset) << shift; + offset++; + tmp = RREG32_PLL(addr); + tmp &= and_mask; + tmp |= or_mask; + WREG32_PLL(addr, tmp); + break; + case 2: + case 3: + tmp = 1000; + switch (addr) { + case 1: + udelay(150); + break; + case 2: + udelay(1000); + break; + case 3: + while (tmp--) { + if (! + (RREG32_PLL + (RADEON_CLK_PWRMGT_CNTL) & + RADEON_MC_BUSY)) + break; + } + break; + case 4: + while (tmp--) { + if (RREG32_PLL + (RADEON_CLK_PWRMGT_CNTL) & + RADEON_DLL_READY) + break; + } + break; + case 5: + tmp = + RREG32_PLL(RADEON_CLK_PWRMGT_CNTL); + if (tmp & RADEON_CG_NO1_DEBUG_0) { +#if 0 + uint32_t mclk_cntl = + RREG32_PLL + (RADEON_MCLK_CNTL); + mclk_cntl &= 0xffff0000; + /*mclk_cntl |= 0x00001111;*//* ??? */ + WREG32_PLL(RADEON_MCLK_CNTL, + mclk_cntl); + udelay(10000); +#endif + WREG32_PLL + (RADEON_CLK_PWRMGT_CNTL, + tmp & + ~RADEON_CG_NO1_DEBUG_0); + udelay(10000); + } + break; + default: + break; + } + break; + default: + break; + } + } + } +} + +static void combios_parse_ram_reset_table(struct drm_device *dev, + uint16_t offset) +{ + struct radeon_device *rdev = dev->dev_private; + uint32_t tmp; + + if (offset) { + uint8_t val = RBIOS8(offset); + while (val != 0xff) { + offset++; + + if (val == 0x0f) { + uint32_t channel_complete_mask; + + if (ASIC_IS_R300(rdev)) + channel_complete_mask = + R300_MEM_PWRUP_COMPLETE; + else + channel_complete_mask = + RADEON_MEM_PWRUP_COMPLETE; + tmp = 20000; + while (tmp--) { + if ((RREG32(RADEON_MEM_STR_CNTL) & + channel_complete_mask) == + channel_complete_mask) + break; + } + } else { + uint32_t or_mask = RBIOS16(offset); + offset += 2; + + tmp = RREG32(RADEON_MEM_SDRAM_MODE_REG); + tmp &= RADEON_SDRAM_MODE_MASK; + tmp |= or_mask; + WREG32(RADEON_MEM_SDRAM_MODE_REG, tmp); + + or_mask = val << 24; + tmp = RREG32(RADEON_MEM_SDRAM_MODE_REG); + tmp &= RADEON_B3MEM_RESET_MASK; + tmp |= or_mask; + WREG32(RADEON_MEM_SDRAM_MODE_REG, tmp); + } + val = RBIOS8(offset); + } + } +} + +static uint32_t combios_detect_ram(struct drm_device *dev, int ram, + int mem_addr_mapping) +{ + struct radeon_device *rdev = dev->dev_private; + uint32_t mem_cntl; + uint32_t mem_size; + uint32_t addr = 0; + + mem_cntl = RREG32(RADEON_MEM_CNTL); + if (mem_cntl & RV100_HALF_MODE) + ram /= 2; + mem_size = ram; + mem_cntl &= ~(0xff << 8); + mem_cntl |= (mem_addr_mapping & 0xff) << 8; + WREG32(RADEON_MEM_CNTL, mem_cntl); + RREG32(RADEON_MEM_CNTL); + + /* sdram reset ? */ + + /* something like this???? */ + while (ram--) { + addr = ram * 1024 * 1024; + /* write to each page */ + WREG32(RADEON_MM_INDEX, (addr) | RADEON_MM_APER); + WREG32(RADEON_MM_DATA, 0xdeadbeef); + /* read back and verify */ + WREG32(RADEON_MM_INDEX, (addr) | RADEON_MM_APER); + if (RREG32(RADEON_MM_DATA) != 0xdeadbeef) + return 0; + } + + return mem_size; +} + +static void combios_write_ram_size(struct drm_device *dev) +{ + struct radeon_device *rdev = dev->dev_private; + uint8_t rev; + uint16_t offset; + uint32_t mem_size = 0; + uint32_t mem_cntl = 0; + + /* should do something smarter here I guess... */ + if (rdev->flags & RADEON_IS_IGP) + return; + + /* first check detected mem table */ + offset = combios_get_table_offset(dev, COMBIOS_DETECTED_MEM_TABLE); + if (offset) { + rev = RBIOS8(offset); + if (rev < 3) { + mem_cntl = RBIOS32(offset + 1); + mem_size = RBIOS16(offset + 5); + if ((rdev->family < CHIP_R200) && + !ASIC_IS_RN50(rdev)) + WREG32(RADEON_MEM_CNTL, mem_cntl); + } + } + + if (!mem_size) { + offset = + combios_get_table_offset(dev, COMBIOS_MEM_CONFIG_TABLE); + if (offset) { + rev = RBIOS8(offset - 1); + if (rev < 1) { + if ((rdev->family < CHIP_R200) + && !ASIC_IS_RN50(rdev)) { + int ram = 0; + int mem_addr_mapping = 0; + + while (RBIOS8(offset)) { + ram = RBIOS8(offset); + mem_addr_mapping = + RBIOS8(offset + 1); + if (mem_addr_mapping != 0x25) + ram *= 2; + mem_size = + combios_detect_ram(dev, ram, + mem_addr_mapping); + if (mem_size) + break; + offset += 2; + } + } else + mem_size = RBIOS8(offset); + } else { + mem_size = RBIOS8(offset); + mem_size *= 2; /* convert to MB */ + } + } + } + + mem_size *= (1024 * 1024); /* convert to bytes */ + WREG32(RADEON_CONFIG_MEMSIZE, mem_size); +} + +void radeon_combios_dyn_clk_setup(struct drm_device *dev, int enable) +{ + uint16_t dyn_clk_info = + combios_get_table_offset(dev, COMBIOS_DYN_CLK_1_TABLE); + + if (dyn_clk_info) + combios_parse_pll_table(dev, dyn_clk_info); +} + +void radeon_combios_asic_init(struct drm_device *dev) +{ + struct radeon_device *rdev = dev->dev_private; + uint16_t table; + + /* port hardcoded mac stuff from radeonfb */ + if (rdev->bios == NULL) + return; + + /* ASIC INIT 1 */ + table = combios_get_table_offset(dev, COMBIOS_ASIC_INIT_1_TABLE); + if (table) + combios_parse_mmio_table(dev, table); + + /* PLL INIT */ + table = combios_get_table_offset(dev, COMBIOS_PLL_INIT_TABLE); + if (table) + combios_parse_pll_table(dev, table); + + /* ASIC INIT 2 */ + table = combios_get_table_offset(dev, COMBIOS_ASIC_INIT_2_TABLE); + if (table) + combios_parse_mmio_table(dev, table); + + if (!(rdev->flags & RADEON_IS_IGP)) { + /* ASIC INIT 4 */ + table = + combios_get_table_offset(dev, COMBIOS_ASIC_INIT_4_TABLE); + if (table) + combios_parse_mmio_table(dev, table); + + /* RAM RESET */ + table = combios_get_table_offset(dev, COMBIOS_RAM_RESET_TABLE); + if (table) + combios_parse_ram_reset_table(dev, table); + + /* ASIC INIT 3 */ + table = + combios_get_table_offset(dev, COMBIOS_ASIC_INIT_3_TABLE); + if (table) + combios_parse_mmio_table(dev, table); + + /* write CONFIG_MEMSIZE */ + combios_write_ram_size(dev); + } + + /* quirk for rs4xx HP nx6125 laptop to make it resume + * - it hangs on resume inside the dynclk 1 table. + */ + if (rdev->family == CHIP_RS480 && + rdev->pdev->subsystem_vendor == 0x103c && + rdev->pdev->subsystem_device == 0x308b) + return; + + /* quirk for rs4xx HP dv5000 laptop to make it resume + * - it hangs on resume inside the dynclk 1 table. + */ + if (rdev->family == CHIP_RS480 && + rdev->pdev->subsystem_vendor == 0x103c && + rdev->pdev->subsystem_device == 0x30a4) + return; + + /* quirk for rs4xx Compaq Presario V5245EU laptop to make it resume + * - it hangs on resume inside the dynclk 1 table. + */ + if (rdev->family == CHIP_RS480 && + rdev->pdev->subsystem_vendor == 0x103c && + rdev->pdev->subsystem_device == 0x30ae) + return; + + /* DYN CLK 1 */ + table = combios_get_table_offset(dev, COMBIOS_DYN_CLK_1_TABLE); + if (table) + combios_parse_pll_table(dev, table); + +} + +void radeon_combios_initialize_bios_scratch_regs(struct drm_device *dev) +{ + struct radeon_device *rdev = dev->dev_private; + uint32_t bios_0_scratch, bios_6_scratch, bios_7_scratch; + + bios_0_scratch = RREG32(RADEON_BIOS_0_SCRATCH); + bios_6_scratch = RREG32(RADEON_BIOS_6_SCRATCH); + bios_7_scratch = RREG32(RADEON_BIOS_7_SCRATCH); + + /* let the bios control the backlight */ + bios_0_scratch &= ~RADEON_DRIVER_BRIGHTNESS_EN; + + /* tell the bios not to handle mode switching */ + bios_6_scratch |= (RADEON_DISPLAY_SWITCHING_DIS | + RADEON_ACC_MODE_CHANGE); + + /* tell the bios a driver is loaded */ + bios_7_scratch |= RADEON_DRV_LOADED; + + WREG32(RADEON_BIOS_0_SCRATCH, bios_0_scratch); + WREG32(RADEON_BIOS_6_SCRATCH, bios_6_scratch); + WREG32(RADEON_BIOS_7_SCRATCH, bios_7_scratch); +} + +void radeon_combios_output_lock(struct drm_encoder *encoder, bool lock) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + uint32_t bios_6_scratch; + + bios_6_scratch = RREG32(RADEON_BIOS_6_SCRATCH); + + if (lock) + bios_6_scratch |= RADEON_DRIVER_CRITICAL; + else + bios_6_scratch &= ~RADEON_DRIVER_CRITICAL; + + WREG32(RADEON_BIOS_6_SCRATCH, bios_6_scratch); +} + +void +radeon_combios_connected_scratch_regs(struct drm_connector *connector, + struct drm_encoder *encoder, + bool connected) +{ + struct drm_device *dev = connector->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_connector *radeon_connector = + to_radeon_connector(connector); + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + uint32_t bios_4_scratch = RREG32(RADEON_BIOS_4_SCRATCH); + uint32_t bios_5_scratch = RREG32(RADEON_BIOS_5_SCRATCH); + + if ((radeon_encoder->devices & ATOM_DEVICE_TV1_SUPPORT) && + (radeon_connector->devices & ATOM_DEVICE_TV1_SUPPORT)) { + if (connected) { + DRM_DEBUG_KMS("TV1 connected\n"); + /* fix me */ + bios_4_scratch |= RADEON_TV1_ATTACHED_SVIDEO; + /*save->bios_4_scratch |= RADEON_TV1_ATTACHED_COMP; */ + bios_5_scratch |= RADEON_TV1_ON; + bios_5_scratch |= RADEON_ACC_REQ_TV1; + } else { + DRM_DEBUG_KMS("TV1 disconnected\n"); + bios_4_scratch &= ~RADEON_TV1_ATTACHED_MASK; + bios_5_scratch &= ~RADEON_TV1_ON; + bios_5_scratch &= ~RADEON_ACC_REQ_TV1; + } + } + if ((radeon_encoder->devices & ATOM_DEVICE_LCD1_SUPPORT) && + (radeon_connector->devices & ATOM_DEVICE_LCD1_SUPPORT)) { + if (connected) { + DRM_DEBUG_KMS("LCD1 connected\n"); + bios_4_scratch |= RADEON_LCD1_ATTACHED; + bios_5_scratch |= RADEON_LCD1_ON; + bios_5_scratch |= RADEON_ACC_REQ_LCD1; + } else { + DRM_DEBUG_KMS("LCD1 disconnected\n"); + bios_4_scratch &= ~RADEON_LCD1_ATTACHED; + bios_5_scratch &= ~RADEON_LCD1_ON; + bios_5_scratch &= ~RADEON_ACC_REQ_LCD1; + } + } + if ((radeon_encoder->devices & ATOM_DEVICE_CRT1_SUPPORT) && + (radeon_connector->devices & ATOM_DEVICE_CRT1_SUPPORT)) { + if (connected) { + DRM_DEBUG_KMS("CRT1 connected\n"); + bios_4_scratch |= RADEON_CRT1_ATTACHED_COLOR; + bios_5_scratch |= RADEON_CRT1_ON; + bios_5_scratch |= RADEON_ACC_REQ_CRT1; + } else { + DRM_DEBUG_KMS("CRT1 disconnected\n"); + bios_4_scratch &= ~RADEON_CRT1_ATTACHED_MASK; + bios_5_scratch &= ~RADEON_CRT1_ON; + bios_5_scratch &= ~RADEON_ACC_REQ_CRT1; + } + } + if ((radeon_encoder->devices & ATOM_DEVICE_CRT2_SUPPORT) && + (radeon_connector->devices & ATOM_DEVICE_CRT2_SUPPORT)) { + if (connected) { + DRM_DEBUG_KMS("CRT2 connected\n"); + bios_4_scratch |= RADEON_CRT2_ATTACHED_COLOR; + bios_5_scratch |= RADEON_CRT2_ON; + bios_5_scratch |= RADEON_ACC_REQ_CRT2; + } else { + DRM_DEBUG_KMS("CRT2 disconnected\n"); + bios_4_scratch &= ~RADEON_CRT2_ATTACHED_MASK; + bios_5_scratch &= ~RADEON_CRT2_ON; + bios_5_scratch &= ~RADEON_ACC_REQ_CRT2; + } + } + if ((radeon_encoder->devices & ATOM_DEVICE_DFP1_SUPPORT) && + (radeon_connector->devices & ATOM_DEVICE_DFP1_SUPPORT)) { + if (connected) { + DRM_DEBUG_KMS("DFP1 connected\n"); + bios_4_scratch |= RADEON_DFP1_ATTACHED; + bios_5_scratch |= RADEON_DFP1_ON; + bios_5_scratch |= RADEON_ACC_REQ_DFP1; + } else { + DRM_DEBUG_KMS("DFP1 disconnected\n"); + bios_4_scratch &= ~RADEON_DFP1_ATTACHED; + bios_5_scratch &= ~RADEON_DFP1_ON; + bios_5_scratch &= ~RADEON_ACC_REQ_DFP1; + } + } + if ((radeon_encoder->devices & ATOM_DEVICE_DFP2_SUPPORT) && + (radeon_connector->devices & ATOM_DEVICE_DFP2_SUPPORT)) { + if (connected) { + DRM_DEBUG_KMS("DFP2 connected\n"); + bios_4_scratch |= RADEON_DFP2_ATTACHED; + bios_5_scratch |= RADEON_DFP2_ON; + bios_5_scratch |= RADEON_ACC_REQ_DFP2; + } else { + DRM_DEBUG_KMS("DFP2 disconnected\n"); + bios_4_scratch &= ~RADEON_DFP2_ATTACHED; + bios_5_scratch &= ~RADEON_DFP2_ON; + bios_5_scratch &= ~RADEON_ACC_REQ_DFP2; + } + } + WREG32(RADEON_BIOS_4_SCRATCH, bios_4_scratch); + WREG32(RADEON_BIOS_5_SCRATCH, bios_5_scratch); +} + +void +radeon_combios_encoder_crtc_scratch_regs(struct drm_encoder *encoder, int crtc) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + uint32_t bios_5_scratch = RREG32(RADEON_BIOS_5_SCRATCH); + + if (radeon_encoder->devices & ATOM_DEVICE_TV1_SUPPORT) { + bios_5_scratch &= ~RADEON_TV1_CRTC_MASK; + bios_5_scratch |= (crtc << RADEON_TV1_CRTC_SHIFT); + } + if (radeon_encoder->devices & ATOM_DEVICE_CRT1_SUPPORT) { + bios_5_scratch &= ~RADEON_CRT1_CRTC_MASK; + bios_5_scratch |= (crtc << RADEON_CRT1_CRTC_SHIFT); + } + if (radeon_encoder->devices & ATOM_DEVICE_CRT2_SUPPORT) { + bios_5_scratch &= ~RADEON_CRT2_CRTC_MASK; + bios_5_scratch |= (crtc << RADEON_CRT2_CRTC_SHIFT); + } + if (radeon_encoder->devices & ATOM_DEVICE_LCD1_SUPPORT) { + bios_5_scratch &= ~RADEON_LCD1_CRTC_MASK; + bios_5_scratch |= (crtc << RADEON_LCD1_CRTC_SHIFT); + } + if (radeon_encoder->devices & ATOM_DEVICE_DFP1_SUPPORT) { + bios_5_scratch &= ~RADEON_DFP1_CRTC_MASK; + bios_5_scratch |= (crtc << RADEON_DFP1_CRTC_SHIFT); + } + if (radeon_encoder->devices & ATOM_DEVICE_DFP2_SUPPORT) { + bios_5_scratch &= ~RADEON_DFP2_CRTC_MASK; + bios_5_scratch |= (crtc << RADEON_DFP2_CRTC_SHIFT); + } + WREG32(RADEON_BIOS_5_SCRATCH, bios_5_scratch); +} + +void +radeon_combios_encoder_dpms_scratch_regs(struct drm_encoder *encoder, bool on) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + uint32_t bios_6_scratch = RREG32(RADEON_BIOS_6_SCRATCH); + + if (radeon_encoder->devices & (ATOM_DEVICE_TV_SUPPORT)) { + if (on) + bios_6_scratch |= RADEON_TV_DPMS_ON; + else + bios_6_scratch &= ~RADEON_TV_DPMS_ON; + } + if (radeon_encoder->devices & (ATOM_DEVICE_CRT_SUPPORT)) { + if (on) + bios_6_scratch |= RADEON_CRT_DPMS_ON; + else + bios_6_scratch &= ~RADEON_CRT_DPMS_ON; + } + if (radeon_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) { + if (on) + bios_6_scratch |= RADEON_LCD_DPMS_ON; + else + bios_6_scratch &= ~RADEON_LCD_DPMS_ON; + } + if (radeon_encoder->devices & (ATOM_DEVICE_DFP_SUPPORT)) { + if (on) + bios_6_scratch |= RADEON_DFP_DPMS_ON; + else + bios_6_scratch &= ~RADEON_DFP_DPMS_ON; + } + WREG32(RADEON_BIOS_6_SCRATCH, bios_6_scratch); +} diff --git a/drivers/gpu/drm/radeon/radeon_connectors.c b/drivers/gpu/drm/radeon/radeon_connectors.c new file mode 100644 index 00000000..1f6a0f55 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_connectors.c @@ -0,0 +1,1959 @@ +/* + * Copyright 2007-8 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + */ +#include "drmP.h" +#include "drm_edid.h" +#include "drm_crtc_helper.h" +#include "drm_fb_helper.h" +#include "radeon_drm.h" +#include "radeon.h" +#include "atom.h" + +extern void +radeon_combios_connected_scratch_regs(struct drm_connector *connector, + struct drm_encoder *encoder, + bool connected); +extern void +radeon_atombios_connected_scratch_regs(struct drm_connector *connector, + struct drm_encoder *encoder, + bool connected); + +extern void +radeon_legacy_backlight_init(struct radeon_encoder *radeon_encoder, + struct drm_connector *drm_connector); + +bool radeon_connector_encoder_is_dp_bridge(struct drm_connector *connector); + +void radeon_connector_hotplug(struct drm_connector *connector) +{ + struct drm_device *dev = connector->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + + /* bail if the connector does not have hpd pin, e.g., + * VGA, TV, etc. + */ + if (radeon_connector->hpd.hpd == RADEON_HPD_NONE) + return; + + radeon_hpd_set_polarity(rdev, radeon_connector->hpd.hpd); + + /* if the connector is already off, don't turn it back on */ + if (connector->dpms != DRM_MODE_DPMS_ON) + return; + + /* just deal with DP (not eDP) here. */ + if (connector->connector_type == DRM_MODE_CONNECTOR_DisplayPort) { + int saved_dpms = connector->dpms; + + /* Only turn off the display it it's physically disconnected */ + if (!radeon_hpd_sense(rdev, radeon_connector->hpd.hpd)) + drm_helper_connector_dpms(connector, DRM_MODE_DPMS_OFF); + else if (radeon_dp_needs_link_train(radeon_connector)) + drm_helper_connector_dpms(connector, DRM_MODE_DPMS_ON); + connector->dpms = saved_dpms; + } +} + +static void radeon_property_change_mode(struct drm_encoder *encoder) +{ + struct drm_crtc *crtc = encoder->crtc; + + if (crtc && crtc->enabled) { + drm_crtc_helper_set_mode(crtc, &crtc->mode, + crtc->x, crtc->y, crtc->fb); + } +} +static void +radeon_connector_update_scratch_regs(struct drm_connector *connector, enum drm_connector_status status) +{ + struct drm_device *dev = connector->dev; + struct radeon_device *rdev = dev->dev_private; + struct drm_encoder *best_encoder = NULL; + struct drm_encoder *encoder = NULL; + struct drm_connector_helper_funcs *connector_funcs = connector->helper_private; + struct drm_mode_object *obj; + bool connected; + int i; + + best_encoder = connector_funcs->best_encoder(connector); + + for (i = 0; i < DRM_CONNECTOR_MAX_ENCODER; i++) { + if (connector->encoder_ids[i] == 0) + break; + + obj = drm_mode_object_find(connector->dev, + connector->encoder_ids[i], + DRM_MODE_OBJECT_ENCODER); + if (!obj) + continue; + + encoder = obj_to_encoder(obj); + + if ((encoder == best_encoder) && (status == connector_status_connected)) + connected = true; + else + connected = false; + + if (rdev->is_atom_bios) + radeon_atombios_connected_scratch_regs(connector, encoder, connected); + else + radeon_combios_connected_scratch_regs(connector, encoder, connected); + + } +} + +struct drm_encoder *radeon_find_encoder(struct drm_connector *connector, int encoder_type) +{ + struct drm_mode_object *obj; + struct drm_encoder *encoder; + int i; + + for (i = 0; i < DRM_CONNECTOR_MAX_ENCODER; i++) { + if (connector->encoder_ids[i] == 0) + break; + + obj = drm_mode_object_find(connector->dev, connector->encoder_ids[i], DRM_MODE_OBJECT_ENCODER); + if (!obj) + continue; + + encoder = obj_to_encoder(obj); + if (encoder->encoder_type == encoder_type) + return encoder; + } + return NULL; +} + +struct drm_encoder *radeon_best_single_encoder(struct drm_connector *connector) +{ + int enc_id = connector->encoder_ids[0]; + struct drm_mode_object *obj; + struct drm_encoder *encoder; + + /* pick the encoder ids */ + if (enc_id) { + obj = drm_mode_object_find(connector->dev, enc_id, DRM_MODE_OBJECT_ENCODER); + if (!obj) + return NULL; + encoder = obj_to_encoder(obj); + return encoder; + } + return NULL; +} + +/* + * radeon_connector_analog_encoder_conflict_solve + * - search for other connectors sharing this encoder + * if priority is true, then set them disconnected if this is connected + * if priority is false, set us disconnected if they are connected + */ +static enum drm_connector_status +radeon_connector_analog_encoder_conflict_solve(struct drm_connector *connector, + struct drm_encoder *encoder, + enum drm_connector_status current_status, + bool priority) +{ + struct drm_device *dev = connector->dev; + struct drm_connector *conflict; + struct radeon_connector *radeon_conflict; + int i; + + list_for_each_entry(conflict, &dev->mode_config.connector_list, head) { + if (conflict == connector) + continue; + + radeon_conflict = to_radeon_connector(conflict); + for (i = 0; i < DRM_CONNECTOR_MAX_ENCODER; i++) { + if (conflict->encoder_ids[i] == 0) + break; + + /* if the IDs match */ + if (conflict->encoder_ids[i] == encoder->base.id) { + if (conflict->status != connector_status_connected) + continue; + + if (radeon_conflict->use_digital) + continue; + + if (priority == true) { + DRM_DEBUG_KMS("1: conflicting encoders switching off %s\n", drm_get_connector_name(conflict)); + DRM_DEBUG_KMS("in favor of %s\n", drm_get_connector_name(connector)); + conflict->status = connector_status_disconnected; + radeon_connector_update_scratch_regs(conflict, connector_status_disconnected); + } else { + DRM_DEBUG_KMS("2: conflicting encoders switching off %s\n", drm_get_connector_name(connector)); + DRM_DEBUG_KMS("in favor of %s\n", drm_get_connector_name(conflict)); + current_status = connector_status_disconnected; + } + break; + } + } + } + return current_status; + +} + +static struct drm_display_mode *radeon_fp_native_mode(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct drm_display_mode *mode = NULL; + struct drm_display_mode *native_mode = &radeon_encoder->native_mode; + + if (native_mode->hdisplay != 0 && + native_mode->vdisplay != 0 && + native_mode->clock != 0) { + mode = drm_mode_duplicate(dev, native_mode); + mode->type = DRM_MODE_TYPE_PREFERRED | DRM_MODE_TYPE_DRIVER; + drm_mode_set_name(mode); + + DRM_DEBUG_KMS("Adding native panel mode %s\n", mode->name); + } else if (native_mode->hdisplay != 0 && + native_mode->vdisplay != 0) { + /* mac laptops without an edid */ + /* Note that this is not necessarily the exact panel mode, + * but an approximation based on the cvt formula. For these + * systems we should ideally read the mode info out of the + * registers or add a mode table, but this works and is much + * simpler. + */ + mode = drm_cvt_mode(dev, native_mode->hdisplay, native_mode->vdisplay, 60, true, false, false); + mode->type = DRM_MODE_TYPE_PREFERRED | DRM_MODE_TYPE_DRIVER; + DRM_DEBUG_KMS("Adding cvt approximation of native panel mode %s\n", mode->name); + } + return mode; +} + +static void radeon_add_common_modes(struct drm_encoder *encoder, struct drm_connector *connector) +{ + struct drm_device *dev = encoder->dev; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct drm_display_mode *mode = NULL; + struct drm_display_mode *native_mode = &radeon_encoder->native_mode; + int i; + struct mode_size { + int w; + int h; + } common_modes[17] = { + { 640, 480}, + { 720, 480}, + { 800, 600}, + { 848, 480}, + {1024, 768}, + {1152, 768}, + {1280, 720}, + {1280, 800}, + {1280, 854}, + {1280, 960}, + {1280, 1024}, + {1440, 900}, + {1400, 1050}, + {1680, 1050}, + {1600, 1200}, + {1920, 1080}, + {1920, 1200} + }; + + for (i = 0; i < 17; i++) { + if (radeon_encoder->devices & (ATOM_DEVICE_TV_SUPPORT)) { + if (common_modes[i].w > 1024 || + common_modes[i].h > 768) + continue; + } + if (radeon_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) { + if (common_modes[i].w > native_mode->hdisplay || + common_modes[i].h > native_mode->vdisplay || + (common_modes[i].w == native_mode->hdisplay && + common_modes[i].h == native_mode->vdisplay)) + continue; + } + if (common_modes[i].w < 320 || common_modes[i].h < 200) + continue; + + mode = drm_cvt_mode(dev, common_modes[i].w, common_modes[i].h, 60, false, false, false); + drm_mode_probed_add(connector, mode); + } +} + +int radeon_connector_set_property(struct drm_connector *connector, struct drm_property *property, + uint64_t val) +{ + struct drm_device *dev = connector->dev; + struct radeon_device *rdev = dev->dev_private; + struct drm_encoder *encoder; + struct radeon_encoder *radeon_encoder; + + if (property == rdev->mode_info.coherent_mode_property) { + struct radeon_encoder_atom_dig *dig; + bool new_coherent_mode; + + /* need to find digital encoder on connector */ + encoder = radeon_find_encoder(connector, DRM_MODE_ENCODER_TMDS); + if (!encoder) + return 0; + + radeon_encoder = to_radeon_encoder(encoder); + + if (!radeon_encoder->enc_priv) + return 0; + + dig = radeon_encoder->enc_priv; + new_coherent_mode = val ? true : false; + if (dig->coherent_mode != new_coherent_mode) { + dig->coherent_mode = new_coherent_mode; + radeon_property_change_mode(&radeon_encoder->base); + } + } + + if (property == rdev->mode_info.underscan_property) { + /* need to find digital encoder on connector */ + encoder = radeon_find_encoder(connector, DRM_MODE_ENCODER_TMDS); + if (!encoder) + return 0; + + radeon_encoder = to_radeon_encoder(encoder); + + if (radeon_encoder->underscan_type != val) { + radeon_encoder->underscan_type = val; + radeon_property_change_mode(&radeon_encoder->base); + } + } + + if (property == rdev->mode_info.underscan_hborder_property) { + /* need to find digital encoder on connector */ + encoder = radeon_find_encoder(connector, DRM_MODE_ENCODER_TMDS); + if (!encoder) + return 0; + + radeon_encoder = to_radeon_encoder(encoder); + + if (radeon_encoder->underscan_hborder != val) { + radeon_encoder->underscan_hborder = val; + radeon_property_change_mode(&radeon_encoder->base); + } + } + + if (property == rdev->mode_info.underscan_vborder_property) { + /* need to find digital encoder on connector */ + encoder = radeon_find_encoder(connector, DRM_MODE_ENCODER_TMDS); + if (!encoder) + return 0; + + radeon_encoder = to_radeon_encoder(encoder); + + if (radeon_encoder->underscan_vborder != val) { + radeon_encoder->underscan_vborder = val; + radeon_property_change_mode(&radeon_encoder->base); + } + } + + if (property == rdev->mode_info.tv_std_property) { + encoder = radeon_find_encoder(connector, DRM_MODE_ENCODER_TVDAC); + if (!encoder) { + encoder = radeon_find_encoder(connector, DRM_MODE_ENCODER_DAC); + } + + if (!encoder) + return 0; + + radeon_encoder = to_radeon_encoder(encoder); + if (!radeon_encoder->enc_priv) + return 0; + if (ASIC_IS_AVIVO(rdev) || radeon_r4xx_atom) { + struct radeon_encoder_atom_dac *dac_int; + dac_int = radeon_encoder->enc_priv; + dac_int->tv_std = val; + } else { + struct radeon_encoder_tv_dac *dac_int; + dac_int = radeon_encoder->enc_priv; + dac_int->tv_std = val; + } + radeon_property_change_mode(&radeon_encoder->base); + } + + if (property == rdev->mode_info.load_detect_property) { + struct radeon_connector *radeon_connector = + to_radeon_connector(connector); + + if (val == 0) + radeon_connector->dac_load_detect = false; + else + radeon_connector->dac_load_detect = true; + } + + if (property == rdev->mode_info.tmds_pll_property) { + struct radeon_encoder_int_tmds *tmds = NULL; + bool ret = false; + /* need to find digital encoder on connector */ + encoder = radeon_find_encoder(connector, DRM_MODE_ENCODER_TMDS); + if (!encoder) + return 0; + + radeon_encoder = to_radeon_encoder(encoder); + + tmds = radeon_encoder->enc_priv; + if (!tmds) + return 0; + + if (val == 0) { + if (rdev->is_atom_bios) + ret = radeon_atombios_get_tmds_info(radeon_encoder, tmds); + else + ret = radeon_legacy_get_tmds_info_from_combios(radeon_encoder, tmds); + } + if (val == 1 || ret == false) { + radeon_legacy_get_tmds_info_from_table(radeon_encoder, tmds); + } + radeon_property_change_mode(&radeon_encoder->base); + } + + return 0; +} + +/* + * Some integrated ATI Radeon chipset implementations (e. g. + * Asus M2A-VM HDMI) may indicate the availability of a DDC, + * even when there's no monitor connected. For these connectors + * following DDC probe extension will be applied: check also for the + * availability of EDID with at least a correct EDID header. Only then, + * DDC is assumed to be available. This prevents drm_get_edid() and + * drm_edid_block_valid() from periodically dumping data and kernel + * errors into the logs and onto the terminal. + */ +static bool radeon_connector_needs_extended_probe(struct radeon_device *dev, + uint32_t supported_device, + int connector_type) +{ + /* Asus M2A-VM HDMI board sends data to i2c bus even, + * if HDMI add-on card is not plugged in or HDMI is disabled in + * BIOS. Valid DDC can only be assumed, if also a valid EDID header + * can be retrieved via i2c bus during DDC probe */ + if ((dev->pdev->device == 0x791e) && + (dev->pdev->subsystem_vendor == 0x1043) && + (dev->pdev->subsystem_device == 0x826d)) { + if ((connector_type == DRM_MODE_CONNECTOR_HDMIA) && + (supported_device == ATOM_DEVICE_DFP2_SUPPORT)) + return true; + } + /* ECS A740GM-M with ATI RADEON 2100 sends data to i2c bus + * for a DVI connector that is not implemented */ + if ((dev->pdev->device == 0x796e) && + (dev->pdev->subsystem_vendor == 0x1019) && + (dev->pdev->subsystem_device == 0x2615)) { + if ((connector_type == DRM_MODE_CONNECTOR_DVID) && + (supported_device == ATOM_DEVICE_DFP2_SUPPORT)) + return true; + } + /* TOSHIBA Satellite L300D with ATI Mobility Radeon x1100 + * (RS690M) sends data to i2c bus for a HDMI connector that + * is not implemented */ + if ((dev->pdev->device == 0x791f) && + (dev->pdev->subsystem_vendor == 0x1179) && + (dev->pdev->subsystem_device == 0xff68)) { + if ((connector_type == DRM_MODE_CONNECTOR_HDMIA) && + (supported_device == ATOM_DEVICE_DFP2_SUPPORT)) + return true; + } + + /* Default: no EDID header probe required for DDC probing */ + return false; +} + +static void radeon_fixup_lvds_native_mode(struct drm_encoder *encoder, + struct drm_connector *connector) +{ + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct drm_display_mode *native_mode = &radeon_encoder->native_mode; + + /* Try to get native mode details from EDID if necessary */ + if (!native_mode->clock) { + struct drm_display_mode *t, *mode; + + list_for_each_entry_safe(mode, t, &connector->probed_modes, head) { + if (mode->hdisplay == native_mode->hdisplay && + mode->vdisplay == native_mode->vdisplay) { + *native_mode = *mode; + drm_mode_set_crtcinfo(native_mode, CRTC_INTERLACE_HALVE_V); + DRM_DEBUG_KMS("Determined LVDS native mode details from EDID\n"); + break; + } + } + } + if (!native_mode->clock) { + DRM_DEBUG_KMS("No LVDS native mode details, disabling RMX\n"); + radeon_encoder->rmx_type = RMX_OFF; + } +} + +static int radeon_lvds_get_modes(struct drm_connector *connector) +{ + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + struct drm_encoder *encoder; + int ret = 0; + struct drm_display_mode *mode; + + if (radeon_connector->ddc_bus) { + ret = radeon_ddc_get_modes(radeon_connector); + if (ret > 0) { + encoder = radeon_best_single_encoder(connector); + if (encoder) { + radeon_fixup_lvds_native_mode(encoder, connector); + /* add scaled modes */ + radeon_add_common_modes(encoder, connector); + } + return ret; + } + } + + encoder = radeon_best_single_encoder(connector); + if (!encoder) + return 0; + + /* we have no EDID modes */ + mode = radeon_fp_native_mode(encoder); + if (mode) { + ret = 1; + drm_mode_probed_add(connector, mode); + /* add the width/height from vbios tables if available */ + connector->display_info.width_mm = mode->width_mm; + connector->display_info.height_mm = mode->height_mm; + /* add scaled modes */ + radeon_add_common_modes(encoder, connector); + } + + return ret; +} + +static int radeon_lvds_mode_valid(struct drm_connector *connector, + struct drm_display_mode *mode) +{ + struct drm_encoder *encoder = radeon_best_single_encoder(connector); + + if ((mode->hdisplay < 320) || (mode->vdisplay < 240)) + return MODE_PANEL; + + if (encoder) { + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct drm_display_mode *native_mode = &radeon_encoder->native_mode; + + /* AVIVO hardware supports downscaling modes larger than the panel + * to the panel size, but I'm not sure this is desirable. + */ + if ((mode->hdisplay > native_mode->hdisplay) || + (mode->vdisplay > native_mode->vdisplay)) + return MODE_PANEL; + + /* if scaling is disabled, block non-native modes */ + if (radeon_encoder->rmx_type == RMX_OFF) { + if ((mode->hdisplay != native_mode->hdisplay) || + (mode->vdisplay != native_mode->vdisplay)) + return MODE_PANEL; + } + } + + return MODE_OK; +} + +static enum drm_connector_status +radeon_lvds_detect(struct drm_connector *connector, bool force) +{ + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + struct drm_encoder *encoder = radeon_best_single_encoder(connector); + enum drm_connector_status ret = connector_status_disconnected; + + if (encoder) { + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct drm_display_mode *native_mode = &radeon_encoder->native_mode; + + /* check if panel is valid */ + if (native_mode->hdisplay >= 320 && native_mode->vdisplay >= 240) + ret = connector_status_connected; + + } + + /* check for edid as well */ + if (radeon_connector->edid) + ret = connector_status_connected; + else { + if (radeon_connector->ddc_bus) { + radeon_connector->edid = drm_get_edid(&radeon_connector->base, + &radeon_connector->ddc_bus->adapter); + if (radeon_connector->edid) + ret = connector_status_connected; + } + } + /* check acpi lid status ??? */ + + radeon_connector_update_scratch_regs(connector, ret); + return ret; +} + +static void radeon_connector_destroy(struct drm_connector *connector) +{ + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + + if (radeon_connector->edid) + kfree(radeon_connector->edid); + kfree(radeon_connector->con_priv); + drm_sysfs_connector_remove(connector); + drm_connector_cleanup(connector); + kfree(connector); +} + +static int radeon_lvds_set_property(struct drm_connector *connector, + struct drm_property *property, + uint64_t value) +{ + struct drm_device *dev = connector->dev; + struct radeon_encoder *radeon_encoder; + enum radeon_rmx_type rmx_type; + + DRM_DEBUG_KMS("\n"); + if (property != dev->mode_config.scaling_mode_property) + return 0; + + if (connector->encoder) + radeon_encoder = to_radeon_encoder(connector->encoder); + else { + struct drm_connector_helper_funcs *connector_funcs = connector->helper_private; + radeon_encoder = to_radeon_encoder(connector_funcs->best_encoder(connector)); + } + + switch (value) { + case DRM_MODE_SCALE_NONE: rmx_type = RMX_OFF; break; + case DRM_MODE_SCALE_CENTER: rmx_type = RMX_CENTER; break; + case DRM_MODE_SCALE_ASPECT: rmx_type = RMX_ASPECT; break; + default: + case DRM_MODE_SCALE_FULLSCREEN: rmx_type = RMX_FULL; break; + } + if (radeon_encoder->rmx_type == rmx_type) + return 0; + + radeon_encoder->rmx_type = rmx_type; + + radeon_property_change_mode(&radeon_encoder->base); + return 0; +} + + +struct drm_connector_helper_funcs radeon_lvds_connector_helper_funcs = { + .get_modes = radeon_lvds_get_modes, + .mode_valid = radeon_lvds_mode_valid, + .best_encoder = radeon_best_single_encoder, +}; + +struct drm_connector_funcs radeon_lvds_connector_funcs = { + .dpms = drm_helper_connector_dpms, + .detect = radeon_lvds_detect, + .fill_modes = drm_helper_probe_single_connector_modes, + .destroy = radeon_connector_destroy, + .set_property = radeon_lvds_set_property, +}; + +static int radeon_vga_get_modes(struct drm_connector *connector) +{ + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + int ret; + + ret = radeon_ddc_get_modes(radeon_connector); + + return ret; +} + +static int radeon_vga_mode_valid(struct drm_connector *connector, + struct drm_display_mode *mode) +{ + struct drm_device *dev = connector->dev; + struct radeon_device *rdev = dev->dev_private; + + /* XXX check mode bandwidth */ + + if ((mode->clock / 10) > rdev->clock.max_pixel_clock) + return MODE_CLOCK_HIGH; + + return MODE_OK; +} + +static enum drm_connector_status +radeon_vga_detect(struct drm_connector *connector, bool force) +{ + struct drm_device *dev = connector->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + struct drm_encoder *encoder; + struct drm_encoder_helper_funcs *encoder_funcs; + bool dret = false; + enum drm_connector_status ret = connector_status_disconnected; + + encoder = radeon_best_single_encoder(connector); + if (!encoder) + ret = connector_status_disconnected; + + if (radeon_connector->ddc_bus) + dret = radeon_ddc_probe(radeon_connector, + radeon_connector->requires_extended_probe); + if (dret) { + radeon_connector->detected_by_load = false; + if (radeon_connector->edid) { + kfree(radeon_connector->edid); + radeon_connector->edid = NULL; + } + radeon_connector->edid = drm_get_edid(&radeon_connector->base, &radeon_connector->ddc_bus->adapter); + + if (!radeon_connector->edid) { + DRM_ERROR("%s: probed a monitor but no|invalid EDID\n", + drm_get_connector_name(connector)); + ret = connector_status_connected; + } else { + radeon_connector->use_digital = !!(radeon_connector->edid->input & DRM_EDID_INPUT_DIGITAL); + + /* some oems have boards with separate digital and analog connectors + * with a shared ddc line (often vga + hdmi) + */ + if (radeon_connector->use_digital && radeon_connector->shared_ddc) { + kfree(radeon_connector->edid); + radeon_connector->edid = NULL; + ret = connector_status_disconnected; + } else + ret = connector_status_connected; + } + } else { + + /* if we aren't forcing don't do destructive polling */ + if (!force) { + /* only return the previous status if we last + * detected a monitor via load. + */ + if (radeon_connector->detected_by_load) + return connector->status; + else + return ret; + } + + if (radeon_connector->dac_load_detect && encoder) { + encoder_funcs = encoder->helper_private; + ret = encoder_funcs->detect(encoder, connector); + if (ret != connector_status_disconnected) + radeon_connector->detected_by_load = true; + } + } + + if (ret == connector_status_connected) + ret = radeon_connector_analog_encoder_conflict_solve(connector, encoder, ret, true); + + /* RN50 and some RV100 asics in servers often have a hardcoded EDID in the + * vbios to deal with KVMs. If we have one and are not able to detect a monitor + * by other means, assume the CRT is connected and use that EDID. + */ + if ((!rdev->is_atom_bios) && + (ret == connector_status_disconnected) && + rdev->mode_info.bios_hardcoded_edid_size) { + ret = connector_status_connected; + } + + radeon_connector_update_scratch_regs(connector, ret); + return ret; +} + +struct drm_connector_helper_funcs radeon_vga_connector_helper_funcs = { + .get_modes = radeon_vga_get_modes, + .mode_valid = radeon_vga_mode_valid, + .best_encoder = radeon_best_single_encoder, +}; + +struct drm_connector_funcs radeon_vga_connector_funcs = { + .dpms = drm_helper_connector_dpms, + .detect = radeon_vga_detect, + .fill_modes = drm_helper_probe_single_connector_modes, + .destroy = radeon_connector_destroy, + .set_property = radeon_connector_set_property, +}; + +static int radeon_tv_get_modes(struct drm_connector *connector) +{ + struct drm_device *dev = connector->dev; + struct radeon_device *rdev = dev->dev_private; + struct drm_display_mode *tv_mode; + struct drm_encoder *encoder; + + encoder = radeon_best_single_encoder(connector); + if (!encoder) + return 0; + + /* avivo chips can scale any mode */ + if (rdev->family >= CHIP_RS600) + /* add scaled modes */ + radeon_add_common_modes(encoder, connector); + else { + /* only 800x600 is supported right now on pre-avivo chips */ + tv_mode = drm_cvt_mode(dev, 800, 600, 60, false, false, false); + tv_mode->type = DRM_MODE_TYPE_DRIVER | DRM_MODE_TYPE_PREFERRED; + drm_mode_probed_add(connector, tv_mode); + } + return 1; +} + +static int radeon_tv_mode_valid(struct drm_connector *connector, + struct drm_display_mode *mode) +{ + if ((mode->hdisplay > 1024) || (mode->vdisplay > 768)) + return MODE_CLOCK_RANGE; + return MODE_OK; +} + +static enum drm_connector_status +radeon_tv_detect(struct drm_connector *connector, bool force) +{ + struct drm_encoder *encoder; + struct drm_encoder_helper_funcs *encoder_funcs; + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + enum drm_connector_status ret = connector_status_disconnected; + + if (!radeon_connector->dac_load_detect) + return ret; + + encoder = radeon_best_single_encoder(connector); + if (!encoder) + ret = connector_status_disconnected; + else { + encoder_funcs = encoder->helper_private; + ret = encoder_funcs->detect(encoder, connector); + } + if (ret == connector_status_connected) + ret = radeon_connector_analog_encoder_conflict_solve(connector, encoder, ret, false); + radeon_connector_update_scratch_regs(connector, ret); + return ret; +} + +struct drm_connector_helper_funcs radeon_tv_connector_helper_funcs = { + .get_modes = radeon_tv_get_modes, + .mode_valid = radeon_tv_mode_valid, + .best_encoder = radeon_best_single_encoder, +}; + +struct drm_connector_funcs radeon_tv_connector_funcs = { + .dpms = drm_helper_connector_dpms, + .detect = radeon_tv_detect, + .fill_modes = drm_helper_probe_single_connector_modes, + .destroy = radeon_connector_destroy, + .set_property = radeon_connector_set_property, +}; + +static int radeon_dvi_get_modes(struct drm_connector *connector) +{ + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + int ret; + + ret = radeon_ddc_get_modes(radeon_connector); + return ret; +} + +/* + * DVI is complicated + * Do a DDC probe, if DDC probe passes, get the full EDID so + * we can do analog/digital monitor detection at this point. + * If the monitor is an analog monitor or we got no DDC, + * we need to find the DAC encoder object for this connector. + * If we got no DDC, we do load detection on the DAC encoder object. + * If we got analog DDC or load detection passes on the DAC encoder + * we have to check if this analog encoder is shared with anyone else (TV) + * if its shared we have to set the other connector to disconnected. + */ +static enum drm_connector_status +radeon_dvi_detect(struct drm_connector *connector, bool force) +{ + struct drm_device *dev = connector->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + struct drm_encoder *encoder = NULL; + struct drm_encoder_helper_funcs *encoder_funcs; + struct drm_mode_object *obj; + int i; + enum drm_connector_status ret = connector_status_disconnected; + bool dret = false; + + if (radeon_connector->ddc_bus) + dret = radeon_ddc_probe(radeon_connector, + radeon_connector->requires_extended_probe); + if (dret) { + radeon_connector->detected_by_load = false; + if (radeon_connector->edid) { + kfree(radeon_connector->edid); + radeon_connector->edid = NULL; + } + radeon_connector->edid = drm_get_edid(&radeon_connector->base, &radeon_connector->ddc_bus->adapter); + + if (!radeon_connector->edid) { + DRM_ERROR("%s: probed a monitor but no|invalid EDID\n", + drm_get_connector_name(connector)); + /* rs690 seems to have a problem with connectors not existing and always + * return a block of 0's. If we see this just stop polling on this output */ + if ((rdev->family == CHIP_RS690 || rdev->family == CHIP_RS740) && radeon_connector->base.null_edid_counter) { + ret = connector_status_disconnected; + DRM_ERROR("%s: detected RS690 floating bus bug, stopping ddc detect\n", drm_get_connector_name(connector)); + radeon_connector->ddc_bus = NULL; + } + } else { + radeon_connector->use_digital = !!(radeon_connector->edid->input & DRM_EDID_INPUT_DIGITAL); + + /* some oems have boards with separate digital and analog connectors + * with a shared ddc line (often vga + hdmi) + */ + if ((!radeon_connector->use_digital) && radeon_connector->shared_ddc) { + kfree(radeon_connector->edid); + radeon_connector->edid = NULL; + ret = connector_status_disconnected; + } else + ret = connector_status_connected; + + /* This gets complicated. We have boards with VGA + HDMI with a + * shared DDC line and we have boards with DVI-D + HDMI with a shared + * DDC line. The latter is more complex because with DVI<->HDMI adapters + * you don't really know what's connected to which port as both are digital. + */ + if (radeon_connector->shared_ddc && (ret == connector_status_connected)) { + struct drm_connector *list_connector; + struct radeon_connector *list_radeon_connector; + list_for_each_entry(list_connector, &dev->mode_config.connector_list, head) { + if (connector == list_connector) + continue; + list_radeon_connector = to_radeon_connector(list_connector); + if (list_radeon_connector->shared_ddc && + (list_radeon_connector->ddc_bus->rec.i2c_id == + radeon_connector->ddc_bus->rec.i2c_id)) { + /* cases where both connectors are digital */ + if (list_connector->connector_type != DRM_MODE_CONNECTOR_VGA) { + /* hpd is our only option in this case */ + if (!radeon_hpd_sense(rdev, radeon_connector->hpd.hpd)) { + kfree(radeon_connector->edid); + radeon_connector->edid = NULL; + ret = connector_status_disconnected; + } + } + } + } + } + } + } + + if ((ret == connector_status_connected) && (radeon_connector->use_digital == true)) + goto out; + + /* DVI-D and HDMI-A are digital only */ + if ((connector->connector_type == DRM_MODE_CONNECTOR_DVID) || + (connector->connector_type == DRM_MODE_CONNECTOR_HDMIA)) + goto out; + + /* if we aren't forcing don't do destructive polling */ + if (!force) { + /* only return the previous status if we last + * detected a monitor via load. + */ + if (radeon_connector->detected_by_load) + ret = connector->status; + goto out; + } + + /* find analog encoder */ + if (radeon_connector->dac_load_detect) { + for (i = 0; i < DRM_CONNECTOR_MAX_ENCODER; i++) { + if (connector->encoder_ids[i] == 0) + break; + + obj = drm_mode_object_find(connector->dev, + connector->encoder_ids[i], + DRM_MODE_OBJECT_ENCODER); + if (!obj) + continue; + + encoder = obj_to_encoder(obj); + + if (encoder->encoder_type != DRM_MODE_ENCODER_DAC && + encoder->encoder_type != DRM_MODE_ENCODER_TVDAC) + continue; + + encoder_funcs = encoder->helper_private; + if (encoder_funcs->detect) { + if (ret != connector_status_connected) { + ret = encoder_funcs->detect(encoder, connector); + if (ret == connector_status_connected) { + radeon_connector->use_digital = false; + } + if (ret != connector_status_disconnected) + radeon_connector->detected_by_load = true; + } + break; + } + } + } + + if ((ret == connector_status_connected) && (radeon_connector->use_digital == false) && + encoder) { + ret = radeon_connector_analog_encoder_conflict_solve(connector, encoder, ret, true); + } + + /* RN50 and some RV100 asics in servers often have a hardcoded EDID in the + * vbios to deal with KVMs. If we have one and are not able to detect a monitor + * by other means, assume the DFP is connected and use that EDID. In most + * cases the DVI port is actually a virtual KVM port connected to the service + * processor. + */ +out: + if ((!rdev->is_atom_bios) && + (ret == connector_status_disconnected) && + rdev->mode_info.bios_hardcoded_edid_size) { + radeon_connector->use_digital = true; + ret = connector_status_connected; + } + + /* updated in get modes as well since we need to know if it's analog or digital */ + radeon_connector_update_scratch_regs(connector, ret); + return ret; +} + +/* okay need to be smart in here about which encoder to pick */ +struct drm_encoder *radeon_dvi_encoder(struct drm_connector *connector) +{ + int enc_id = connector->encoder_ids[0]; + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + struct drm_mode_object *obj; + struct drm_encoder *encoder; + int i; + for (i = 0; i < DRM_CONNECTOR_MAX_ENCODER; i++) { + if (connector->encoder_ids[i] == 0) + break; + + obj = drm_mode_object_find(connector->dev, connector->encoder_ids[i], DRM_MODE_OBJECT_ENCODER); + if (!obj) + continue; + + encoder = obj_to_encoder(obj); + + if (radeon_connector->use_digital == true) { + if (encoder->encoder_type == DRM_MODE_ENCODER_TMDS) + return encoder; + } else { + if (encoder->encoder_type == DRM_MODE_ENCODER_DAC || + encoder->encoder_type == DRM_MODE_ENCODER_TVDAC) + return encoder; + } + } + + /* see if we have a default encoder TODO */ + + /* then check use digitial */ + /* pick the first one */ + if (enc_id) { + obj = drm_mode_object_find(connector->dev, enc_id, DRM_MODE_OBJECT_ENCODER); + if (!obj) + return NULL; + encoder = obj_to_encoder(obj); + return encoder; + } + return NULL; +} + +static void radeon_dvi_force(struct drm_connector *connector) +{ + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + if (connector->force == DRM_FORCE_ON) + radeon_connector->use_digital = false; + if (connector->force == DRM_FORCE_ON_DIGITAL) + radeon_connector->use_digital = true; +} + +static int radeon_dvi_mode_valid(struct drm_connector *connector, + struct drm_display_mode *mode) +{ + struct drm_device *dev = connector->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + + /* XXX check mode bandwidth */ + + /* clocks over 135 MHz have heat issues with DVI on RV100 */ + if (radeon_connector->use_digital && + (rdev->family == CHIP_RV100) && + (mode->clock > 135000)) + return MODE_CLOCK_HIGH; + + if (radeon_connector->use_digital && (mode->clock > 165000)) { + if ((radeon_connector->connector_object_id == CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_I) || + (radeon_connector->connector_object_id == CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_D) || + (radeon_connector->connector_object_id == CONNECTOR_OBJECT_ID_HDMI_TYPE_B)) + return MODE_OK; + else if (radeon_connector->connector_object_id == CONNECTOR_OBJECT_ID_HDMI_TYPE_A) { + if (ASIC_IS_DCE3(rdev)) { + /* HDMI 1.3+ supports max clock of 340 Mhz */ + if (mode->clock > 340000) + return MODE_CLOCK_HIGH; + else + return MODE_OK; + } else + return MODE_CLOCK_HIGH; + } else + return MODE_CLOCK_HIGH; + } + + /* check against the max pixel clock */ + if ((mode->clock / 10) > rdev->clock.max_pixel_clock) + return MODE_CLOCK_HIGH; + + return MODE_OK; +} + +struct drm_connector_helper_funcs radeon_dvi_connector_helper_funcs = { + .get_modes = radeon_dvi_get_modes, + .mode_valid = radeon_dvi_mode_valid, + .best_encoder = radeon_dvi_encoder, +}; + +struct drm_connector_funcs radeon_dvi_connector_funcs = { + .dpms = drm_helper_connector_dpms, + .detect = radeon_dvi_detect, + .fill_modes = drm_helper_probe_single_connector_modes, + .set_property = radeon_connector_set_property, + .destroy = radeon_connector_destroy, + .force = radeon_dvi_force, +}; + +static void radeon_dp_connector_destroy(struct drm_connector *connector) +{ + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + struct radeon_connector_atom_dig *radeon_dig_connector = radeon_connector->con_priv; + + if (radeon_connector->edid) + kfree(radeon_connector->edid); + if (radeon_dig_connector->dp_i2c_bus) + radeon_i2c_destroy(radeon_dig_connector->dp_i2c_bus); + kfree(radeon_connector->con_priv); + drm_sysfs_connector_remove(connector); + drm_connector_cleanup(connector); + kfree(connector); +} + +static int radeon_dp_get_modes(struct drm_connector *connector) +{ + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + struct radeon_connector_atom_dig *radeon_dig_connector = radeon_connector->con_priv; + struct drm_encoder *encoder = radeon_best_single_encoder(connector); + int ret; + + if ((connector->connector_type == DRM_MODE_CONNECTOR_eDP) || + (connector->connector_type == DRM_MODE_CONNECTOR_LVDS)) { + struct drm_display_mode *mode; + + if (!radeon_dig_connector->edp_on) + atombios_set_edp_panel_power(connector, + ATOM_TRANSMITTER_ACTION_POWER_ON); + ret = radeon_ddc_get_modes(radeon_connector); + if (!radeon_dig_connector->edp_on) + atombios_set_edp_panel_power(connector, + ATOM_TRANSMITTER_ACTION_POWER_OFF); + + if (ret > 0) { + if (encoder) { + radeon_fixup_lvds_native_mode(encoder, connector); + /* add scaled modes */ + radeon_add_common_modes(encoder, connector); + } + return ret; + } + + encoder = radeon_best_single_encoder(connector); + if (!encoder) + return 0; + + /* we have no EDID modes */ + mode = radeon_fp_native_mode(encoder); + if (mode) { + ret = 1; + drm_mode_probed_add(connector, mode); + /* add the width/height from vbios tables if available */ + connector->display_info.width_mm = mode->width_mm; + connector->display_info.height_mm = mode->height_mm; + /* add scaled modes */ + radeon_add_common_modes(encoder, connector); + } + } else { + /* need to setup ddc on the bridge */ + if (radeon_connector_encoder_is_dp_bridge(connector)) { + if (encoder) + radeon_atom_ext_encoder_setup_ddc(encoder); + } + ret = radeon_ddc_get_modes(radeon_connector); + } + + return ret; +} + +bool radeon_connector_encoder_is_dp_bridge(struct drm_connector *connector) +{ + struct drm_mode_object *obj; + struct drm_encoder *encoder; + struct radeon_encoder *radeon_encoder; + int i; + bool found = false; + + for (i = 0; i < DRM_CONNECTOR_MAX_ENCODER; i++) { + if (connector->encoder_ids[i] == 0) + break; + + obj = drm_mode_object_find(connector->dev, connector->encoder_ids[i], DRM_MODE_OBJECT_ENCODER); + if (!obj) + continue; + + encoder = obj_to_encoder(obj); + radeon_encoder = to_radeon_encoder(encoder); + + switch (radeon_encoder->encoder_id) { + case ENCODER_OBJECT_ID_TRAVIS: + case ENCODER_OBJECT_ID_NUTMEG: + found = true; + break; + default: + break; + } + } + + return found; +} + +bool radeon_connector_encoder_is_hbr2(struct drm_connector *connector) +{ + struct drm_mode_object *obj; + struct drm_encoder *encoder; + struct radeon_encoder *radeon_encoder; + int i; + bool found = false; + + for (i = 0; i < DRM_CONNECTOR_MAX_ENCODER; i++) { + if (connector->encoder_ids[i] == 0) + break; + + obj = drm_mode_object_find(connector->dev, connector->encoder_ids[i], DRM_MODE_OBJECT_ENCODER); + if (!obj) + continue; + + encoder = obj_to_encoder(obj); + radeon_encoder = to_radeon_encoder(encoder); + if (radeon_encoder->caps & ATOM_ENCODER_CAP_RECORD_HBR2) + found = true; + } + + return found; +} + +bool radeon_connector_is_dp12_capable(struct drm_connector *connector) +{ + struct drm_device *dev = connector->dev; + struct radeon_device *rdev = dev->dev_private; + + if (ASIC_IS_DCE5(rdev) && + (rdev->clock.dp_extclk >= 53900) && + radeon_connector_encoder_is_hbr2(connector)) { + return true; + } + + return false; +} + +static enum drm_connector_status +radeon_dp_detect(struct drm_connector *connector, bool force) +{ + struct drm_device *dev = connector->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + enum drm_connector_status ret = connector_status_disconnected; + struct radeon_connector_atom_dig *radeon_dig_connector = radeon_connector->con_priv; + struct drm_encoder *encoder = radeon_best_single_encoder(connector); + + if (radeon_connector->edid) { + kfree(radeon_connector->edid); + radeon_connector->edid = NULL; + } + + if ((connector->connector_type == DRM_MODE_CONNECTOR_eDP) || + (connector->connector_type == DRM_MODE_CONNECTOR_LVDS)) { + if (encoder) { + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct drm_display_mode *native_mode = &radeon_encoder->native_mode; + + /* check if panel is valid */ + if (native_mode->hdisplay >= 320 && native_mode->vdisplay >= 240) + ret = connector_status_connected; + } + /* eDP is always DP */ + radeon_dig_connector->dp_sink_type = CONNECTOR_OBJECT_ID_DISPLAYPORT; + if (!radeon_dig_connector->edp_on) + atombios_set_edp_panel_power(connector, + ATOM_TRANSMITTER_ACTION_POWER_ON); + if (radeon_dp_getdpcd(radeon_connector)) + ret = connector_status_connected; + if (!radeon_dig_connector->edp_on) + atombios_set_edp_panel_power(connector, + ATOM_TRANSMITTER_ACTION_POWER_OFF); + } else { + /* need to setup ddc on the bridge */ + if (radeon_connector_encoder_is_dp_bridge(connector)) { + if (encoder) + radeon_atom_ext_encoder_setup_ddc(encoder); + } + radeon_dig_connector->dp_sink_type = radeon_dp_getsinktype(radeon_connector); + if (radeon_hpd_sense(rdev, radeon_connector->hpd.hpd)) { + ret = connector_status_connected; + if (radeon_dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT) + radeon_dp_getdpcd(radeon_connector); + } else { + if (radeon_dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT) { + if (radeon_dp_getdpcd(radeon_connector)) + ret = connector_status_connected; + } else { + if (radeon_ddc_probe(radeon_connector, + radeon_connector->requires_extended_probe)) + ret = connector_status_connected; + } + } + + if ((ret == connector_status_disconnected) && + radeon_connector->dac_load_detect) { + struct drm_encoder *encoder = radeon_best_single_encoder(connector); + struct drm_encoder_helper_funcs *encoder_funcs; + if (encoder) { + encoder_funcs = encoder->helper_private; + ret = encoder_funcs->detect(encoder, connector); + } + } + } + + radeon_connector_update_scratch_regs(connector, ret); + return ret; +} + +static int radeon_dp_mode_valid(struct drm_connector *connector, + struct drm_display_mode *mode) +{ + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + struct radeon_connector_atom_dig *radeon_dig_connector = radeon_connector->con_priv; + + /* XXX check mode bandwidth */ + + if ((connector->connector_type == DRM_MODE_CONNECTOR_eDP) || + (connector->connector_type == DRM_MODE_CONNECTOR_LVDS)) { + struct drm_encoder *encoder = radeon_best_single_encoder(connector); + + if ((mode->hdisplay < 320) || (mode->vdisplay < 240)) + return MODE_PANEL; + + if (encoder) { + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct drm_display_mode *native_mode = &radeon_encoder->native_mode; + + /* AVIVO hardware supports downscaling modes larger than the panel + * to the panel size, but I'm not sure this is desirable. + */ + if ((mode->hdisplay > native_mode->hdisplay) || + (mode->vdisplay > native_mode->vdisplay)) + return MODE_PANEL; + + /* if scaling is disabled, block non-native modes */ + if (radeon_encoder->rmx_type == RMX_OFF) { + if ((mode->hdisplay != native_mode->hdisplay) || + (mode->vdisplay != native_mode->vdisplay)) + return MODE_PANEL; + } + } + return MODE_OK; + } else { + if ((radeon_dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT) || + (radeon_dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_eDP)) + return radeon_dp_mode_valid_helper(connector, mode); + else + return MODE_OK; + } +} + +struct drm_connector_helper_funcs radeon_dp_connector_helper_funcs = { + .get_modes = radeon_dp_get_modes, + .mode_valid = radeon_dp_mode_valid, + .best_encoder = radeon_dvi_encoder, +}; + +struct drm_connector_funcs radeon_dp_connector_funcs = { + .dpms = drm_helper_connector_dpms, + .detect = radeon_dp_detect, + .fill_modes = drm_helper_probe_single_connector_modes, + .set_property = radeon_connector_set_property, + .destroy = radeon_dp_connector_destroy, + .force = radeon_dvi_force, +}; + +void +radeon_add_atom_connector(struct drm_device *dev, + uint32_t connector_id, + uint32_t supported_device, + int connector_type, + struct radeon_i2c_bus_rec *i2c_bus, + uint32_t igp_lane_info, + uint16_t connector_object_id, + struct radeon_hpd *hpd, + struct radeon_router *router) +{ + struct radeon_device *rdev = dev->dev_private; + struct drm_connector *connector; + struct radeon_connector *radeon_connector; + struct radeon_connector_atom_dig *radeon_dig_connector; + struct drm_encoder *encoder; + struct radeon_encoder *radeon_encoder; + uint32_t subpixel_order = SubPixelNone; + bool shared_ddc = false; + bool is_dp_bridge = false; + + if (connector_type == DRM_MODE_CONNECTOR_Unknown) + return; + + /* if the user selected tv=0 don't try and add the connector */ + if (((connector_type == DRM_MODE_CONNECTOR_SVIDEO) || + (connector_type == DRM_MODE_CONNECTOR_Composite) || + (connector_type == DRM_MODE_CONNECTOR_9PinDIN)) && + (radeon_tv == 0)) + return; + + /* see if we already added it */ + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + radeon_connector = to_radeon_connector(connector); + if (radeon_connector->connector_id == connector_id) { + radeon_connector->devices |= supported_device; + return; + } + if (radeon_connector->ddc_bus && i2c_bus->valid) { + if (radeon_connector->ddc_bus->rec.i2c_id == i2c_bus->i2c_id) { + radeon_connector->shared_ddc = true; + shared_ddc = true; + } + if (radeon_connector->router_bus && router->ddc_valid && + (radeon_connector->router.router_id == router->router_id)) { + radeon_connector->shared_ddc = false; + shared_ddc = false; + } + } + } + + /* check if it's a dp bridge */ + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + radeon_encoder = to_radeon_encoder(encoder); + if (radeon_encoder->devices & supported_device) { + switch (radeon_encoder->encoder_id) { + case ENCODER_OBJECT_ID_TRAVIS: + case ENCODER_OBJECT_ID_NUTMEG: + is_dp_bridge = true; + break; + default: + break; + } + } + } + + radeon_connector = kzalloc(sizeof(struct radeon_connector), GFP_KERNEL); + if (!radeon_connector) + return; + + connector = &radeon_connector->base; + + radeon_connector->connector_id = connector_id; + radeon_connector->devices = supported_device; + radeon_connector->shared_ddc = shared_ddc; + radeon_connector->connector_object_id = connector_object_id; + radeon_connector->hpd = *hpd; + radeon_connector->requires_extended_probe = + radeon_connector_needs_extended_probe(rdev, supported_device, + connector_type); + radeon_connector->router = *router; + if (router->ddc_valid || router->cd_valid) { + radeon_connector->router_bus = radeon_i2c_lookup(rdev, &router->i2c_info); + if (!radeon_connector->router_bus) + DRM_ERROR("Failed to assign router i2c bus! Check dmesg for i2c errors.\n"); + } + + if (is_dp_bridge) { + radeon_dig_connector = kzalloc(sizeof(struct radeon_connector_atom_dig), GFP_KERNEL); + if (!radeon_dig_connector) + goto failed; + radeon_dig_connector->igp_lane_info = igp_lane_info; + radeon_connector->con_priv = radeon_dig_connector; + drm_connector_init(dev, &radeon_connector->base, &radeon_dp_connector_funcs, connector_type); + drm_connector_helper_add(&radeon_connector->base, &radeon_dp_connector_helper_funcs); + if (i2c_bus->valid) { + /* add DP i2c bus */ + if (connector_type == DRM_MODE_CONNECTOR_eDP) + radeon_dig_connector->dp_i2c_bus = radeon_i2c_create_dp(dev, i2c_bus, "eDP-auxch"); + else + radeon_dig_connector->dp_i2c_bus = radeon_i2c_create_dp(dev, i2c_bus, "DP-auxch"); + if (!radeon_dig_connector->dp_i2c_bus) + DRM_ERROR("DP: Failed to assign dp ddc bus! Check dmesg for i2c errors.\n"); + radeon_connector->ddc_bus = radeon_i2c_lookup(rdev, i2c_bus); + if (!radeon_connector->ddc_bus) + DRM_ERROR("DP: Failed to assign ddc bus! Check dmesg for i2c errors.\n"); + } + switch (connector_type) { + case DRM_MODE_CONNECTOR_VGA: + case DRM_MODE_CONNECTOR_DVIA: + default: + connector->interlace_allowed = true; + connector->doublescan_allowed = true; + radeon_connector->dac_load_detect = true; + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.load_detect_property, + 1); + break; + case DRM_MODE_CONNECTOR_DVII: + case DRM_MODE_CONNECTOR_DVID: + case DRM_MODE_CONNECTOR_HDMIA: + case DRM_MODE_CONNECTOR_HDMIB: + case DRM_MODE_CONNECTOR_DisplayPort: + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.underscan_property, + UNDERSCAN_OFF); + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.underscan_hborder_property, + 0); + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.underscan_vborder_property, + 0); + subpixel_order = SubPixelHorizontalRGB; + connector->interlace_allowed = true; + if (connector_type == DRM_MODE_CONNECTOR_HDMIB) + connector->doublescan_allowed = true; + else + connector->doublescan_allowed = false; + if (connector_type == DRM_MODE_CONNECTOR_DVII) { + radeon_connector->dac_load_detect = true; + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.load_detect_property, + 1); + } + break; + case DRM_MODE_CONNECTOR_LVDS: + case DRM_MODE_CONNECTOR_eDP: + drm_connector_attach_property(&radeon_connector->base, + dev->mode_config.scaling_mode_property, + DRM_MODE_SCALE_FULLSCREEN); + subpixel_order = SubPixelHorizontalRGB; + connector->interlace_allowed = false; + connector->doublescan_allowed = false; + break; + } + } else { + switch (connector_type) { + case DRM_MODE_CONNECTOR_VGA: + drm_connector_init(dev, &radeon_connector->base, &radeon_vga_connector_funcs, connector_type); + drm_connector_helper_add(&radeon_connector->base, &radeon_vga_connector_helper_funcs); + if (i2c_bus->valid) { + radeon_connector->ddc_bus = radeon_i2c_lookup(rdev, i2c_bus); + if (!radeon_connector->ddc_bus) + DRM_ERROR("VGA: Failed to assign ddc bus! Check dmesg for i2c errors.\n"); + } + radeon_connector->dac_load_detect = true; + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.load_detect_property, + 1); + /* no HPD on analog connectors */ + radeon_connector->hpd.hpd = RADEON_HPD_NONE; + connector->polled = DRM_CONNECTOR_POLL_CONNECT; + connector->interlace_allowed = true; + connector->doublescan_allowed = true; + break; + case DRM_MODE_CONNECTOR_DVIA: + drm_connector_init(dev, &radeon_connector->base, &radeon_vga_connector_funcs, connector_type); + drm_connector_helper_add(&radeon_connector->base, &radeon_vga_connector_helper_funcs); + if (i2c_bus->valid) { + radeon_connector->ddc_bus = radeon_i2c_lookup(rdev, i2c_bus); + if (!radeon_connector->ddc_bus) + DRM_ERROR("DVIA: Failed to assign ddc bus! Check dmesg for i2c errors.\n"); + } + radeon_connector->dac_load_detect = true; + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.load_detect_property, + 1); + /* no HPD on analog connectors */ + radeon_connector->hpd.hpd = RADEON_HPD_NONE; + connector->interlace_allowed = true; + connector->doublescan_allowed = true; + break; + case DRM_MODE_CONNECTOR_DVII: + case DRM_MODE_CONNECTOR_DVID: + radeon_dig_connector = kzalloc(sizeof(struct radeon_connector_atom_dig), GFP_KERNEL); + if (!radeon_dig_connector) + goto failed; + radeon_dig_connector->igp_lane_info = igp_lane_info; + radeon_connector->con_priv = radeon_dig_connector; + drm_connector_init(dev, &radeon_connector->base, &radeon_dvi_connector_funcs, connector_type); + drm_connector_helper_add(&radeon_connector->base, &radeon_dvi_connector_helper_funcs); + if (i2c_bus->valid) { + radeon_connector->ddc_bus = radeon_i2c_lookup(rdev, i2c_bus); + if (!radeon_connector->ddc_bus) + DRM_ERROR("DVI: Failed to assign ddc bus! Check dmesg for i2c errors.\n"); + } + subpixel_order = SubPixelHorizontalRGB; + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.coherent_mode_property, + 1); + if (ASIC_IS_AVIVO(rdev)) { + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.underscan_property, + UNDERSCAN_OFF); + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.underscan_hborder_property, + 0); + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.underscan_vborder_property, + 0); + } + if (connector_type == DRM_MODE_CONNECTOR_DVII) { + radeon_connector->dac_load_detect = true; + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.load_detect_property, + 1); + } + connector->interlace_allowed = true; + if (connector_type == DRM_MODE_CONNECTOR_DVII) + connector->doublescan_allowed = true; + else + connector->doublescan_allowed = false; + break; + case DRM_MODE_CONNECTOR_HDMIA: + case DRM_MODE_CONNECTOR_HDMIB: + radeon_dig_connector = kzalloc(sizeof(struct radeon_connector_atom_dig), GFP_KERNEL); + if (!radeon_dig_connector) + goto failed; + radeon_dig_connector->igp_lane_info = igp_lane_info; + radeon_connector->con_priv = radeon_dig_connector; + drm_connector_init(dev, &radeon_connector->base, &radeon_dvi_connector_funcs, connector_type); + drm_connector_helper_add(&radeon_connector->base, &radeon_dvi_connector_helper_funcs); + if (i2c_bus->valid) { + radeon_connector->ddc_bus = radeon_i2c_lookup(rdev, i2c_bus); + if (!radeon_connector->ddc_bus) + DRM_ERROR("HDMI: Failed to assign ddc bus! Check dmesg for i2c errors.\n"); + } + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.coherent_mode_property, + 1); + if (ASIC_IS_AVIVO(rdev)) { + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.underscan_property, + UNDERSCAN_OFF); + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.underscan_hborder_property, + 0); + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.underscan_vborder_property, + 0); + } + subpixel_order = SubPixelHorizontalRGB; + connector->interlace_allowed = true; + if (connector_type == DRM_MODE_CONNECTOR_HDMIB) + connector->doublescan_allowed = true; + else + connector->doublescan_allowed = false; + break; + case DRM_MODE_CONNECTOR_DisplayPort: + radeon_dig_connector = kzalloc(sizeof(struct radeon_connector_atom_dig), GFP_KERNEL); + if (!radeon_dig_connector) + goto failed; + radeon_dig_connector->igp_lane_info = igp_lane_info; + radeon_connector->con_priv = radeon_dig_connector; + drm_connector_init(dev, &radeon_connector->base, &radeon_dp_connector_funcs, connector_type); + drm_connector_helper_add(&radeon_connector->base, &radeon_dp_connector_helper_funcs); + if (i2c_bus->valid) { + /* add DP i2c bus */ + radeon_dig_connector->dp_i2c_bus = radeon_i2c_create_dp(dev, i2c_bus, "DP-auxch"); + if (!radeon_dig_connector->dp_i2c_bus) + DRM_ERROR("DP: Failed to assign dp ddc bus! Check dmesg for i2c errors.\n"); + radeon_connector->ddc_bus = radeon_i2c_lookup(rdev, i2c_bus); + if (!radeon_connector->ddc_bus) + DRM_ERROR("DP: Failed to assign ddc bus! Check dmesg for i2c errors.\n"); + } + subpixel_order = SubPixelHorizontalRGB; + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.coherent_mode_property, + 1); + if (ASIC_IS_AVIVO(rdev)) { + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.underscan_property, + UNDERSCAN_OFF); + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.underscan_hborder_property, + 0); + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.underscan_vborder_property, + 0); + } + connector->interlace_allowed = true; + /* in theory with a DP to VGA converter... */ + connector->doublescan_allowed = false; + break; + case DRM_MODE_CONNECTOR_eDP: + radeon_dig_connector = kzalloc(sizeof(struct radeon_connector_atom_dig), GFP_KERNEL); + if (!radeon_dig_connector) + goto failed; + radeon_dig_connector->igp_lane_info = igp_lane_info; + radeon_connector->con_priv = radeon_dig_connector; + drm_connector_init(dev, &radeon_connector->base, &radeon_dp_connector_funcs, connector_type); + drm_connector_helper_add(&radeon_connector->base, &radeon_dp_connector_helper_funcs); + if (i2c_bus->valid) { + /* add DP i2c bus */ + radeon_dig_connector->dp_i2c_bus = radeon_i2c_create_dp(dev, i2c_bus, "eDP-auxch"); + if (!radeon_dig_connector->dp_i2c_bus) + DRM_ERROR("DP: Failed to assign dp ddc bus! Check dmesg for i2c errors.\n"); + radeon_connector->ddc_bus = radeon_i2c_lookup(rdev, i2c_bus); + if (!radeon_connector->ddc_bus) + DRM_ERROR("DP: Failed to assign ddc bus! Check dmesg for i2c errors.\n"); + } + drm_connector_attach_property(&radeon_connector->base, + dev->mode_config.scaling_mode_property, + DRM_MODE_SCALE_FULLSCREEN); + subpixel_order = SubPixelHorizontalRGB; + connector->interlace_allowed = false; + connector->doublescan_allowed = false; + break; + case DRM_MODE_CONNECTOR_SVIDEO: + case DRM_MODE_CONNECTOR_Composite: + case DRM_MODE_CONNECTOR_9PinDIN: + drm_connector_init(dev, &radeon_connector->base, &radeon_tv_connector_funcs, connector_type); + drm_connector_helper_add(&radeon_connector->base, &radeon_tv_connector_helper_funcs); + radeon_connector->dac_load_detect = true; + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.load_detect_property, + 1); + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.tv_std_property, + radeon_atombios_get_tv_info(rdev)); + /* no HPD on analog connectors */ + radeon_connector->hpd.hpd = RADEON_HPD_NONE; + connector->interlace_allowed = false; + connector->doublescan_allowed = false; + break; + case DRM_MODE_CONNECTOR_LVDS: + radeon_dig_connector = kzalloc(sizeof(struct radeon_connector_atom_dig), GFP_KERNEL); + if (!radeon_dig_connector) + goto failed; + radeon_dig_connector->igp_lane_info = igp_lane_info; + radeon_connector->con_priv = radeon_dig_connector; + drm_connector_init(dev, &radeon_connector->base, &radeon_lvds_connector_funcs, connector_type); + drm_connector_helper_add(&radeon_connector->base, &radeon_lvds_connector_helper_funcs); + if (i2c_bus->valid) { + radeon_connector->ddc_bus = radeon_i2c_lookup(rdev, i2c_bus); + if (!radeon_connector->ddc_bus) + DRM_ERROR("LVDS: Failed to assign ddc bus! Check dmesg for i2c errors.\n"); + } + drm_connector_attach_property(&radeon_connector->base, + dev->mode_config.scaling_mode_property, + DRM_MODE_SCALE_FULLSCREEN); + subpixel_order = SubPixelHorizontalRGB; + connector->interlace_allowed = false; + connector->doublescan_allowed = false; + break; + } + } + + if (radeon_connector->hpd.hpd == RADEON_HPD_NONE) { + if (i2c_bus->valid) + connector->polled = DRM_CONNECTOR_POLL_CONNECT; + } else + connector->polled = DRM_CONNECTOR_POLL_HPD; + + connector->display_info.subpixel_order = subpixel_order; + drm_sysfs_connector_add(connector); + return; + +failed: + drm_connector_cleanup(connector); + kfree(connector); +} + +void +radeon_add_legacy_connector(struct drm_device *dev, + uint32_t connector_id, + uint32_t supported_device, + int connector_type, + struct radeon_i2c_bus_rec *i2c_bus, + uint16_t connector_object_id, + struct radeon_hpd *hpd) +{ + struct radeon_device *rdev = dev->dev_private; + struct drm_connector *connector; + struct radeon_connector *radeon_connector; + uint32_t subpixel_order = SubPixelNone; + + if (connector_type == DRM_MODE_CONNECTOR_Unknown) + return; + + /* if the user selected tv=0 don't try and add the connector */ + if (((connector_type == DRM_MODE_CONNECTOR_SVIDEO) || + (connector_type == DRM_MODE_CONNECTOR_Composite) || + (connector_type == DRM_MODE_CONNECTOR_9PinDIN)) && + (radeon_tv == 0)) + return; + + /* see if we already added it */ + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + radeon_connector = to_radeon_connector(connector); + if (radeon_connector->connector_id == connector_id) { + radeon_connector->devices |= supported_device; + return; + } + } + + radeon_connector = kzalloc(sizeof(struct radeon_connector), GFP_KERNEL); + if (!radeon_connector) + return; + + connector = &radeon_connector->base; + + radeon_connector->connector_id = connector_id; + radeon_connector->devices = supported_device; + radeon_connector->connector_object_id = connector_object_id; + radeon_connector->hpd = *hpd; + radeon_connector->requires_extended_probe = + radeon_connector_needs_extended_probe(rdev, supported_device, + connector_type); + switch (connector_type) { + case DRM_MODE_CONNECTOR_VGA: + drm_connector_init(dev, &radeon_connector->base, &radeon_vga_connector_funcs, connector_type); + drm_connector_helper_add(&radeon_connector->base, &radeon_vga_connector_helper_funcs); + if (i2c_bus->valid) { + radeon_connector->ddc_bus = radeon_i2c_lookup(rdev, i2c_bus); + if (!radeon_connector->ddc_bus) + DRM_ERROR("VGA: Failed to assign ddc bus! Check dmesg for i2c errors.\n"); + } + radeon_connector->dac_load_detect = true; + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.load_detect_property, + 1); + /* no HPD on analog connectors */ + radeon_connector->hpd.hpd = RADEON_HPD_NONE; + connector->polled = DRM_CONNECTOR_POLL_CONNECT; + connector->interlace_allowed = true; + connector->doublescan_allowed = true; + break; + case DRM_MODE_CONNECTOR_DVIA: + drm_connector_init(dev, &radeon_connector->base, &radeon_vga_connector_funcs, connector_type); + drm_connector_helper_add(&radeon_connector->base, &radeon_vga_connector_helper_funcs); + if (i2c_bus->valid) { + radeon_connector->ddc_bus = radeon_i2c_lookup(rdev, i2c_bus); + if (!radeon_connector->ddc_bus) + DRM_ERROR("DVIA: Failed to assign ddc bus! Check dmesg for i2c errors.\n"); + } + radeon_connector->dac_load_detect = true; + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.load_detect_property, + 1); + /* no HPD on analog connectors */ + radeon_connector->hpd.hpd = RADEON_HPD_NONE; + connector->interlace_allowed = true; + connector->doublescan_allowed = true; + break; + case DRM_MODE_CONNECTOR_DVII: + case DRM_MODE_CONNECTOR_DVID: + drm_connector_init(dev, &radeon_connector->base, &radeon_dvi_connector_funcs, connector_type); + drm_connector_helper_add(&radeon_connector->base, &radeon_dvi_connector_helper_funcs); + if (i2c_bus->valid) { + radeon_connector->ddc_bus = radeon_i2c_lookup(rdev, i2c_bus); + if (!radeon_connector->ddc_bus) + DRM_ERROR("DVI: Failed to assign ddc bus! Check dmesg for i2c errors.\n"); + } + if (connector_type == DRM_MODE_CONNECTOR_DVII) { + radeon_connector->dac_load_detect = true; + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.load_detect_property, + 1); + } + subpixel_order = SubPixelHorizontalRGB; + connector->interlace_allowed = true; + if (connector_type == DRM_MODE_CONNECTOR_DVII) + connector->doublescan_allowed = true; + else + connector->doublescan_allowed = false; + break; + case DRM_MODE_CONNECTOR_SVIDEO: + case DRM_MODE_CONNECTOR_Composite: + case DRM_MODE_CONNECTOR_9PinDIN: + drm_connector_init(dev, &radeon_connector->base, &radeon_tv_connector_funcs, connector_type); + drm_connector_helper_add(&radeon_connector->base, &radeon_tv_connector_helper_funcs); + radeon_connector->dac_load_detect = true; + /* RS400,RC410,RS480 chipset seems to report a lot + * of false positive on load detect, we haven't yet + * found a way to make load detect reliable on those + * chipset, thus just disable it for TV. + */ + if (rdev->family == CHIP_RS400 || rdev->family == CHIP_RS480) + radeon_connector->dac_load_detect = false; + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.load_detect_property, + radeon_connector->dac_load_detect); + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.tv_std_property, + radeon_combios_get_tv_info(rdev)); + /* no HPD on analog connectors */ + radeon_connector->hpd.hpd = RADEON_HPD_NONE; + connector->interlace_allowed = false; + connector->doublescan_allowed = false; + break; + case DRM_MODE_CONNECTOR_LVDS: + drm_connector_init(dev, &radeon_connector->base, &radeon_lvds_connector_funcs, connector_type); + drm_connector_helper_add(&radeon_connector->base, &radeon_lvds_connector_helper_funcs); + if (i2c_bus->valid) { + radeon_connector->ddc_bus = radeon_i2c_lookup(rdev, i2c_bus); + if (!radeon_connector->ddc_bus) + DRM_ERROR("LVDS: Failed to assign ddc bus! Check dmesg for i2c errors.\n"); + } + drm_connector_attach_property(&radeon_connector->base, + dev->mode_config.scaling_mode_property, + DRM_MODE_SCALE_FULLSCREEN); + subpixel_order = SubPixelHorizontalRGB; + connector->interlace_allowed = false; + connector->doublescan_allowed = false; + break; + } + + if (radeon_connector->hpd.hpd == RADEON_HPD_NONE) { + if (i2c_bus->valid) + connector->polled = DRM_CONNECTOR_POLL_CONNECT; + } else + connector->polled = DRM_CONNECTOR_POLL_HPD; + connector->display_info.subpixel_order = subpixel_order; + drm_sysfs_connector_add(connector); + if (connector_type == DRM_MODE_CONNECTOR_LVDS) { + struct drm_encoder *drm_encoder; + + list_for_each_entry(drm_encoder, &dev->mode_config.encoder_list, head) { + struct radeon_encoder *radeon_encoder; + + radeon_encoder = to_radeon_encoder(drm_encoder); + if (radeon_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_LVDS) + radeon_legacy_backlight_init(radeon_encoder, connector); + } + } +} diff --git a/drivers/gpu/drm/radeon/radeon_cp.c b/drivers/gpu/drm/radeon/radeon_cp.c new file mode 100644 index 00000000..75867792 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_cp.c @@ -0,0 +1,2257 @@ +/* radeon_cp.c -- CP support for Radeon -*- linux-c -*- */ +/* + * Copyright 2000 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Fremont, California. + * Copyright 2007 Advanced Micro Devices, Inc. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Kevin E. Martin + * Gareth Hughes + */ + +#include "drmP.h" +#include "drm.h" +#include "drm_sarea.h" +#include "radeon_drm.h" +#include "radeon_drv.h" +#include "r300_reg.h" + +#define RADEON_FIFO_DEBUG 0 + +/* Firmware Names */ +#define FIRMWARE_R100 "radeon/R100_cp.bin" +#define FIRMWARE_R200 "radeon/R200_cp.bin" +#define FIRMWARE_R300 "radeon/R300_cp.bin" +#define FIRMWARE_R420 "radeon/R420_cp.bin" +#define FIRMWARE_RS690 "radeon/RS690_cp.bin" +#define FIRMWARE_RS600 "radeon/RS600_cp.bin" +#define FIRMWARE_R520 "radeon/R520_cp.bin" + +MODULE_FIRMWARE(FIRMWARE_R100); +MODULE_FIRMWARE(FIRMWARE_R200); +MODULE_FIRMWARE(FIRMWARE_R300); +MODULE_FIRMWARE(FIRMWARE_R420); +MODULE_FIRMWARE(FIRMWARE_RS690); +MODULE_FIRMWARE(FIRMWARE_RS600); +MODULE_FIRMWARE(FIRMWARE_R520); + +static int radeon_do_cleanup_cp(struct drm_device * dev); +static void radeon_do_cp_start(drm_radeon_private_t * dev_priv); + +u32 radeon_read_ring_rptr(drm_radeon_private_t *dev_priv, u32 off) +{ + u32 val; + + if (dev_priv->flags & RADEON_IS_AGP) { + val = DRM_READ32(dev_priv->ring_rptr, off); + } else { + val = *(((volatile u32 *) + dev_priv->ring_rptr->handle) + + (off / sizeof(u32))); + val = le32_to_cpu(val); + } + return val; +} + +u32 radeon_get_ring_head(drm_radeon_private_t *dev_priv) +{ + if (dev_priv->writeback_works) + return radeon_read_ring_rptr(dev_priv, 0); + else { + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) + return RADEON_READ(R600_CP_RB_RPTR); + else + return RADEON_READ(RADEON_CP_RB_RPTR); + } +} + +void radeon_write_ring_rptr(drm_radeon_private_t *dev_priv, u32 off, u32 val) +{ + if (dev_priv->flags & RADEON_IS_AGP) + DRM_WRITE32(dev_priv->ring_rptr, off, val); + else + *(((volatile u32 *) dev_priv->ring_rptr->handle) + + (off / sizeof(u32))) = cpu_to_le32(val); +} + +void radeon_set_ring_head(drm_radeon_private_t *dev_priv, u32 val) +{ + radeon_write_ring_rptr(dev_priv, 0, val); +} + +u32 radeon_get_scratch(drm_radeon_private_t *dev_priv, int index) +{ + if (dev_priv->writeback_works) { + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) + return radeon_read_ring_rptr(dev_priv, + R600_SCRATCHOFF(index)); + else + return radeon_read_ring_rptr(dev_priv, + RADEON_SCRATCHOFF(index)); + } else { + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) + return RADEON_READ(R600_SCRATCH_REG0 + 4*index); + else + return RADEON_READ(RADEON_SCRATCH_REG0 + 4*index); + } +} + +u32 RADEON_READ_MM(drm_radeon_private_t *dev_priv, int addr) +{ + u32 ret; + + if (addr < 0x10000) + ret = DRM_READ32(dev_priv->mmio, addr); + else { + DRM_WRITE32(dev_priv->mmio, RADEON_MM_INDEX, addr); + ret = DRM_READ32(dev_priv->mmio, RADEON_MM_DATA); + } + + return ret; +} + +static u32 R500_READ_MCIND(drm_radeon_private_t *dev_priv, int addr) +{ + u32 ret; + RADEON_WRITE(R520_MC_IND_INDEX, 0x7f0000 | (addr & 0xff)); + ret = RADEON_READ(R520_MC_IND_DATA); + RADEON_WRITE(R520_MC_IND_INDEX, 0); + return ret; +} + +static u32 RS480_READ_MCIND(drm_radeon_private_t *dev_priv, int addr) +{ + u32 ret; + RADEON_WRITE(RS480_NB_MC_INDEX, addr & 0xff); + ret = RADEON_READ(RS480_NB_MC_DATA); + RADEON_WRITE(RS480_NB_MC_INDEX, 0xff); + return ret; +} + +static u32 RS690_READ_MCIND(drm_radeon_private_t *dev_priv, int addr) +{ + u32 ret; + RADEON_WRITE(RS690_MC_INDEX, (addr & RS690_MC_INDEX_MASK)); + ret = RADEON_READ(RS690_MC_DATA); + RADEON_WRITE(RS690_MC_INDEX, RS690_MC_INDEX_MASK); + return ret; +} + +static u32 RS600_READ_MCIND(drm_radeon_private_t *dev_priv, int addr) +{ + u32 ret; + RADEON_WRITE(RS600_MC_INDEX, ((addr & RS600_MC_ADDR_MASK) | + RS600_MC_IND_CITF_ARB0)); + ret = RADEON_READ(RS600_MC_DATA); + return ret; +} + +static u32 IGP_READ_MCIND(drm_radeon_private_t *dev_priv, int addr) +{ + if (((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS690) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS740)) + return RS690_READ_MCIND(dev_priv, addr); + else if ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS600) + return RS600_READ_MCIND(dev_priv, addr); + else + return RS480_READ_MCIND(dev_priv, addr); +} + +u32 radeon_read_fb_location(drm_radeon_private_t *dev_priv) +{ + + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_RV770) + return RADEON_READ(R700_MC_VM_FB_LOCATION); + else if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) + return RADEON_READ(R600_MC_VM_FB_LOCATION); + else if ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV515) + return R500_READ_MCIND(dev_priv, RV515_MC_FB_LOCATION); + else if (((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS690) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS740)) + return RS690_READ_MCIND(dev_priv, RS690_MC_FB_LOCATION); + else if ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS600) + return RS600_READ_MCIND(dev_priv, RS600_MC_FB_LOCATION); + else if ((dev_priv->flags & RADEON_FAMILY_MASK) > CHIP_RV515) + return R500_READ_MCIND(dev_priv, R520_MC_FB_LOCATION); + else + return RADEON_READ(RADEON_MC_FB_LOCATION); +} + +static void radeon_write_fb_location(drm_radeon_private_t *dev_priv, u32 fb_loc) +{ + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_RV770) + RADEON_WRITE(R700_MC_VM_FB_LOCATION, fb_loc); + else if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) + RADEON_WRITE(R600_MC_VM_FB_LOCATION, fb_loc); + else if ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV515) + R500_WRITE_MCIND(RV515_MC_FB_LOCATION, fb_loc); + else if (((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS690) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS740)) + RS690_WRITE_MCIND(RS690_MC_FB_LOCATION, fb_loc); + else if ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS600) + RS600_WRITE_MCIND(RS600_MC_FB_LOCATION, fb_loc); + else if ((dev_priv->flags & RADEON_FAMILY_MASK) > CHIP_RV515) + R500_WRITE_MCIND(R520_MC_FB_LOCATION, fb_loc); + else + RADEON_WRITE(RADEON_MC_FB_LOCATION, fb_loc); +} + +void radeon_write_agp_location(drm_radeon_private_t *dev_priv, u32 agp_loc) +{ + /*R6xx/R7xx: AGP_TOP and BOT are actually 18 bits each */ + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_RV770) { + RADEON_WRITE(R700_MC_VM_AGP_BOT, agp_loc & 0xffff); /* FIX ME */ + RADEON_WRITE(R700_MC_VM_AGP_TOP, (agp_loc >> 16) & 0xffff); + } else if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) { + RADEON_WRITE(R600_MC_VM_AGP_BOT, agp_loc & 0xffff); /* FIX ME */ + RADEON_WRITE(R600_MC_VM_AGP_TOP, (agp_loc >> 16) & 0xffff); + } else if ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV515) + R500_WRITE_MCIND(RV515_MC_AGP_LOCATION, agp_loc); + else if (((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS690) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS740)) + RS690_WRITE_MCIND(RS690_MC_AGP_LOCATION, agp_loc); + else if ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS600) + RS600_WRITE_MCIND(RS600_MC_AGP_LOCATION, agp_loc); + else if ((dev_priv->flags & RADEON_FAMILY_MASK) > CHIP_RV515) + R500_WRITE_MCIND(R520_MC_AGP_LOCATION, agp_loc); + else + RADEON_WRITE(RADEON_MC_AGP_LOCATION, agp_loc); +} + +void radeon_write_agp_base(drm_radeon_private_t *dev_priv, u64 agp_base) +{ + u32 agp_base_hi = upper_32_bits(agp_base); + u32 agp_base_lo = agp_base & 0xffffffff; + u32 r6xx_agp_base = (agp_base >> 22) & 0x3ffff; + + /* R6xx/R7xx must be aligned to a 4MB boundary */ + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_RV770) + RADEON_WRITE(R700_MC_VM_AGP_BASE, r6xx_agp_base); + else if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) + RADEON_WRITE(R600_MC_VM_AGP_BASE, r6xx_agp_base); + else if ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV515) { + R500_WRITE_MCIND(RV515_MC_AGP_BASE, agp_base_lo); + R500_WRITE_MCIND(RV515_MC_AGP_BASE_2, agp_base_hi); + } else if (((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS690) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS740)) { + RS690_WRITE_MCIND(RS690_MC_AGP_BASE, agp_base_lo); + RS690_WRITE_MCIND(RS690_MC_AGP_BASE_2, agp_base_hi); + } else if ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS600) { + RS600_WRITE_MCIND(RS600_AGP_BASE, agp_base_lo); + RS600_WRITE_MCIND(RS600_AGP_BASE_2, agp_base_hi); + } else if ((dev_priv->flags & RADEON_FAMILY_MASK) > CHIP_RV515) { + R500_WRITE_MCIND(R520_MC_AGP_BASE, agp_base_lo); + R500_WRITE_MCIND(R520_MC_AGP_BASE_2, agp_base_hi); + } else if (((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS400) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS480)) { + RADEON_WRITE(RADEON_AGP_BASE, agp_base_lo); + RADEON_WRITE(RS480_AGP_BASE_2, agp_base_hi); + } else { + RADEON_WRITE(RADEON_AGP_BASE, agp_base_lo); + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R200) + RADEON_WRITE(RADEON_AGP_BASE_2, agp_base_hi); + } +} + +void radeon_enable_bm(struct drm_radeon_private *dev_priv) +{ + u32 tmp; + /* Turn on bus mastering */ + if (((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS690) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS740)) { + /* rs600/rs690/rs740 */ + tmp = RADEON_READ(RADEON_BUS_CNTL) & ~RS600_BUS_MASTER_DIS; + RADEON_WRITE(RADEON_BUS_CNTL, tmp); + } else if (((dev_priv->flags & RADEON_FAMILY_MASK) <= CHIP_RV350) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_R420) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS400) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS480)) { + /* r1xx, r2xx, r300, r(v)350, r420/r481, rs400/rs480 */ + tmp = RADEON_READ(RADEON_BUS_CNTL) & ~RADEON_BUS_MASTER_DIS; + RADEON_WRITE(RADEON_BUS_CNTL, tmp); + } /* PCIE cards appears to not need this */ +} + +static int RADEON_READ_PLL(struct drm_device * dev, int addr) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + + RADEON_WRITE8(RADEON_CLOCK_CNTL_INDEX, addr & 0x1f); + return RADEON_READ(RADEON_CLOCK_CNTL_DATA); +} + +static u32 RADEON_READ_PCIE(drm_radeon_private_t *dev_priv, int addr) +{ + RADEON_WRITE8(RADEON_PCIE_INDEX, addr & 0xff); + return RADEON_READ(RADEON_PCIE_DATA); +} + +#if RADEON_FIFO_DEBUG +static void radeon_status(drm_radeon_private_t * dev_priv) +{ + printk("%s:\n", __func__); + printk("RBBM_STATUS = 0x%08x\n", + (unsigned int)RADEON_READ(RADEON_RBBM_STATUS)); + printk("CP_RB_RTPR = 0x%08x\n", + (unsigned int)RADEON_READ(RADEON_CP_RB_RPTR)); + printk("CP_RB_WTPR = 0x%08x\n", + (unsigned int)RADEON_READ(RADEON_CP_RB_WPTR)); + printk("AIC_CNTL = 0x%08x\n", + (unsigned int)RADEON_READ(RADEON_AIC_CNTL)); + printk("AIC_STAT = 0x%08x\n", + (unsigned int)RADEON_READ(RADEON_AIC_STAT)); + printk("AIC_PT_BASE = 0x%08x\n", + (unsigned int)RADEON_READ(RADEON_AIC_PT_BASE)); + printk("TLB_ADDR = 0x%08x\n", + (unsigned int)RADEON_READ(RADEON_AIC_TLB_ADDR)); + printk("TLB_DATA = 0x%08x\n", + (unsigned int)RADEON_READ(RADEON_AIC_TLB_DATA)); +} +#endif + +/* ================================================================ + * Engine, FIFO control + */ + +static int radeon_do_pixcache_flush(drm_radeon_private_t * dev_priv) +{ + u32 tmp; + int i; + + dev_priv->stats.boxes |= RADEON_BOX_WAIT_IDLE; + + if ((dev_priv->flags & RADEON_FAMILY_MASK) <= CHIP_RV280) { + tmp = RADEON_READ(RADEON_RB3D_DSTCACHE_CTLSTAT); + tmp |= RADEON_RB3D_DC_FLUSH_ALL; + RADEON_WRITE(RADEON_RB3D_DSTCACHE_CTLSTAT, tmp); + + for (i = 0; i < dev_priv->usec_timeout; i++) { + if (!(RADEON_READ(RADEON_RB3D_DSTCACHE_CTLSTAT) + & RADEON_RB3D_DC_BUSY)) { + return 0; + } + DRM_UDELAY(1); + } + } else { + /* don't flush or purge cache here or lockup */ + return 0; + } + +#if RADEON_FIFO_DEBUG + DRM_ERROR("failed!\n"); + radeon_status(dev_priv); +#endif + return -EBUSY; +} + +static int radeon_do_wait_for_fifo(drm_radeon_private_t * dev_priv, int entries) +{ + int i; + + dev_priv->stats.boxes |= RADEON_BOX_WAIT_IDLE; + + for (i = 0; i < dev_priv->usec_timeout; i++) { + int slots = (RADEON_READ(RADEON_RBBM_STATUS) + & RADEON_RBBM_FIFOCNT_MASK); + if (slots >= entries) + return 0; + DRM_UDELAY(1); + } + DRM_DEBUG("wait for fifo failed status : 0x%08X 0x%08X\n", + RADEON_READ(RADEON_RBBM_STATUS), + RADEON_READ(R300_VAP_CNTL_STATUS)); + +#if RADEON_FIFO_DEBUG + DRM_ERROR("failed!\n"); + radeon_status(dev_priv); +#endif + return -EBUSY; +} + +static int radeon_do_wait_for_idle(drm_radeon_private_t * dev_priv) +{ + int i, ret; + + dev_priv->stats.boxes |= RADEON_BOX_WAIT_IDLE; + + ret = radeon_do_wait_for_fifo(dev_priv, 64); + if (ret) + return ret; + + for (i = 0; i < dev_priv->usec_timeout; i++) { + if (!(RADEON_READ(RADEON_RBBM_STATUS) + & RADEON_RBBM_ACTIVE)) { + radeon_do_pixcache_flush(dev_priv); + return 0; + } + DRM_UDELAY(1); + } + DRM_DEBUG("wait idle failed status : 0x%08X 0x%08X\n", + RADEON_READ(RADEON_RBBM_STATUS), + RADEON_READ(R300_VAP_CNTL_STATUS)); + +#if RADEON_FIFO_DEBUG + DRM_ERROR("failed!\n"); + radeon_status(dev_priv); +#endif + return -EBUSY; +} + +static void radeon_init_pipes(struct drm_device *dev) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + uint32_t gb_tile_config, gb_pipe_sel = 0; + + if ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV530) { + uint32_t z_pipe_sel = RADEON_READ(RV530_GB_PIPE_SELECT2); + if ((z_pipe_sel & 3) == 3) + dev_priv->num_z_pipes = 2; + else + dev_priv->num_z_pipes = 1; + } else + dev_priv->num_z_pipes = 1; + + /* RS4xx/RS6xx/R4xx/R5xx */ + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R420) { + gb_pipe_sel = RADEON_READ(R400_GB_PIPE_SELECT); + dev_priv->num_gb_pipes = ((gb_pipe_sel >> 12) & 0x3) + 1; + /* SE cards have 1 pipe */ + if ((dev->pdev->device == 0x5e4c) || + (dev->pdev->device == 0x5e4f)) + dev_priv->num_gb_pipes = 1; + } else { + /* R3xx */ + if (((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_R300 && + dev->pdev->device != 0x4144) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_R350 && + dev->pdev->device != 0x4148)) { + dev_priv->num_gb_pipes = 2; + } else { + /* RV3xx/R300 AD/R350 AH */ + dev_priv->num_gb_pipes = 1; + } + } + DRM_INFO("Num pipes: %d\n", dev_priv->num_gb_pipes); + + gb_tile_config = (R300_ENABLE_TILING | R300_TILE_SIZE_16 /*| R300_SUBPIXEL_1_16*/); + + switch (dev_priv->num_gb_pipes) { + case 2: gb_tile_config |= R300_PIPE_COUNT_R300; break; + case 3: gb_tile_config |= R300_PIPE_COUNT_R420_3P; break; + case 4: gb_tile_config |= R300_PIPE_COUNT_R420; break; + default: + case 1: gb_tile_config |= R300_PIPE_COUNT_RV350; break; + } + + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_RV515) { + RADEON_WRITE_PLL(R500_DYN_SCLK_PWMEM_PIPE, (1 | ((gb_pipe_sel >> 8) & 0xf) << 4)); + RADEON_WRITE(R300_SU_REG_DEST, ((1 << dev_priv->num_gb_pipes) - 1)); + } + RADEON_WRITE(R300_GB_TILE_CONFIG, gb_tile_config); + radeon_do_wait_for_idle(dev_priv); + RADEON_WRITE(R300_DST_PIPE_CONFIG, RADEON_READ(R300_DST_PIPE_CONFIG) | R300_PIPE_AUTO_CONFIG); + RADEON_WRITE(R300_RB2D_DSTCACHE_MODE, (RADEON_READ(R300_RB2D_DSTCACHE_MODE) | + R300_DC_AUTOFLUSH_ENABLE | + R300_DC_DC_DISABLE_IGNORE_PE)); + + +} + +/* ================================================================ + * CP control, initialization + */ + +/* Load the microcode for the CP */ +static int radeon_cp_init_microcode(drm_radeon_private_t *dev_priv) +{ + struct platform_device *pdev; + const char *fw_name = NULL; + int err; + + DRM_DEBUG("\n"); + + pdev = platform_device_register_simple("radeon_cp", 0, NULL, 0); + err = IS_ERR(pdev); + if (err) { + printk(KERN_ERR "radeon_cp: Failed to register firmware\n"); + return -EINVAL; + } + + if (((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_R100) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV100) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV200) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS100) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS200)) { + DRM_INFO("Loading R100 Microcode\n"); + fw_name = FIRMWARE_R100; + } else if (((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_R200) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV250) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV280) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS300)) { + DRM_INFO("Loading R200 Microcode\n"); + fw_name = FIRMWARE_R200; + } else if (((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_R300) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_R350) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV350) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV380) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS400) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS480)) { + DRM_INFO("Loading R300 Microcode\n"); + fw_name = FIRMWARE_R300; + } else if (((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_R420) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_R423) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV410)) { + DRM_INFO("Loading R400 Microcode\n"); + fw_name = FIRMWARE_R420; + } else if (((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS690) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS740)) { + DRM_INFO("Loading RS690/RS740 Microcode\n"); + fw_name = FIRMWARE_RS690; + } else if ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS600) { + DRM_INFO("Loading RS600 Microcode\n"); + fw_name = FIRMWARE_RS600; + } else if (((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV515) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_R520) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV530) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_R580) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV560) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV570)) { + DRM_INFO("Loading R500 Microcode\n"); + fw_name = FIRMWARE_R520; + } + + err = request_firmware(&dev_priv->me_fw, fw_name, &pdev->dev); + platform_device_unregister(pdev); + if (err) { + printk(KERN_ERR "radeon_cp: Failed to load firmware \"%s\"\n", + fw_name); + } else if (dev_priv->me_fw->size % 8) { + printk(KERN_ERR + "radeon_cp: Bogus length %zu in firmware \"%s\"\n", + dev_priv->me_fw->size, fw_name); + err = -EINVAL; + release_firmware(dev_priv->me_fw); + dev_priv->me_fw = NULL; + } + return err; +} + +static void radeon_cp_load_microcode(drm_radeon_private_t *dev_priv) +{ + const __be32 *fw_data; + int i, size; + + radeon_do_wait_for_idle(dev_priv); + + if (dev_priv->me_fw) { + size = dev_priv->me_fw->size / 4; + fw_data = (const __be32 *)&dev_priv->me_fw->data[0]; + RADEON_WRITE(RADEON_CP_ME_RAM_ADDR, 0); + for (i = 0; i < size; i += 2) { + RADEON_WRITE(RADEON_CP_ME_RAM_DATAH, + be32_to_cpup(&fw_data[i])); + RADEON_WRITE(RADEON_CP_ME_RAM_DATAL, + be32_to_cpup(&fw_data[i + 1])); + } + } +} + +/* Flush any pending commands to the CP. This should only be used just + * prior to a wait for idle, as it informs the engine that the command + * stream is ending. + */ +static void radeon_do_cp_flush(drm_radeon_private_t * dev_priv) +{ + DRM_DEBUG("\n"); +#if 0 + u32 tmp; + + tmp = RADEON_READ(RADEON_CP_RB_WPTR) | (1 << 31); + RADEON_WRITE(RADEON_CP_RB_WPTR, tmp); +#endif +} + +/* Wait for the CP to go idle. + */ +int radeon_do_cp_idle(drm_radeon_private_t * dev_priv) +{ + RING_LOCALS; + DRM_DEBUG("\n"); + + BEGIN_RING(6); + + RADEON_PURGE_CACHE(); + RADEON_PURGE_ZCACHE(); + RADEON_WAIT_UNTIL_IDLE(); + + ADVANCE_RING(); + COMMIT_RING(); + + return radeon_do_wait_for_idle(dev_priv); +} + +/* Start the Command Processor. + */ +static void radeon_do_cp_start(drm_radeon_private_t * dev_priv) +{ + RING_LOCALS; + DRM_DEBUG("\n"); + + radeon_do_wait_for_idle(dev_priv); + + RADEON_WRITE(RADEON_CP_CSQ_CNTL, dev_priv->cp_mode); + + dev_priv->cp_running = 1; + + /* on r420, any DMA from CP to system memory while 2D is active + * can cause a hang. workaround is to queue a CP RESYNC token + */ + if ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_R420) { + BEGIN_RING(3); + OUT_RING(CP_PACKET0(R300_CP_RESYNC_ADDR, 1)); + OUT_RING(5); /* scratch reg 5 */ + OUT_RING(0xdeadbeef); + ADVANCE_RING(); + COMMIT_RING(); + } + + BEGIN_RING(8); + /* isync can only be written through cp on r5xx write it here */ + OUT_RING(CP_PACKET0(RADEON_ISYNC_CNTL, 0)); + OUT_RING(RADEON_ISYNC_ANY2D_IDLE3D | + RADEON_ISYNC_ANY3D_IDLE2D | + RADEON_ISYNC_WAIT_IDLEGUI | + RADEON_ISYNC_CPSCRATCH_IDLEGUI); + RADEON_PURGE_CACHE(); + RADEON_PURGE_ZCACHE(); + RADEON_WAIT_UNTIL_IDLE(); + ADVANCE_RING(); + COMMIT_RING(); + + dev_priv->track_flush |= RADEON_FLUSH_EMITED | RADEON_PURGE_EMITED; +} + +/* Reset the Command Processor. This will not flush any pending + * commands, so you must wait for the CP command stream to complete + * before calling this routine. + */ +static void radeon_do_cp_reset(drm_radeon_private_t * dev_priv) +{ + u32 cur_read_ptr; + DRM_DEBUG("\n"); + + cur_read_ptr = RADEON_READ(RADEON_CP_RB_RPTR); + RADEON_WRITE(RADEON_CP_RB_WPTR, cur_read_ptr); + SET_RING_HEAD(dev_priv, cur_read_ptr); + dev_priv->ring.tail = cur_read_ptr; +} + +/* Stop the Command Processor. This will not flush any pending + * commands, so you must flush the command stream and wait for the CP + * to go idle before calling this routine. + */ +static void radeon_do_cp_stop(drm_radeon_private_t * dev_priv) +{ + RING_LOCALS; + DRM_DEBUG("\n"); + + /* finish the pending CP_RESYNC token */ + if ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_R420) { + BEGIN_RING(2); + OUT_RING(CP_PACKET0(R300_RB3D_DSTCACHE_CTLSTAT, 0)); + OUT_RING(R300_RB3D_DC_FINISH); + ADVANCE_RING(); + COMMIT_RING(); + radeon_do_wait_for_idle(dev_priv); + } + + RADEON_WRITE(RADEON_CP_CSQ_CNTL, RADEON_CSQ_PRIDIS_INDDIS); + + dev_priv->cp_running = 0; +} + +/* Reset the engine. This will stop the CP if it is running. + */ +static int radeon_do_engine_reset(struct drm_device * dev) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + u32 clock_cntl_index = 0, mclk_cntl = 0, rbbm_soft_reset; + DRM_DEBUG("\n"); + + radeon_do_pixcache_flush(dev_priv); + + if ((dev_priv->flags & RADEON_FAMILY_MASK) <= CHIP_RV410) { + /* may need something similar for newer chips */ + clock_cntl_index = RADEON_READ(RADEON_CLOCK_CNTL_INDEX); + mclk_cntl = RADEON_READ_PLL(dev, RADEON_MCLK_CNTL); + + RADEON_WRITE_PLL(RADEON_MCLK_CNTL, (mclk_cntl | + RADEON_FORCEON_MCLKA | + RADEON_FORCEON_MCLKB | + RADEON_FORCEON_YCLKA | + RADEON_FORCEON_YCLKB | + RADEON_FORCEON_MC | + RADEON_FORCEON_AIC)); + } + + rbbm_soft_reset = RADEON_READ(RADEON_RBBM_SOFT_RESET); + + RADEON_WRITE(RADEON_RBBM_SOFT_RESET, (rbbm_soft_reset | + RADEON_SOFT_RESET_CP | + RADEON_SOFT_RESET_HI | + RADEON_SOFT_RESET_SE | + RADEON_SOFT_RESET_RE | + RADEON_SOFT_RESET_PP | + RADEON_SOFT_RESET_E2 | + RADEON_SOFT_RESET_RB)); + RADEON_READ(RADEON_RBBM_SOFT_RESET); + RADEON_WRITE(RADEON_RBBM_SOFT_RESET, (rbbm_soft_reset & + ~(RADEON_SOFT_RESET_CP | + RADEON_SOFT_RESET_HI | + RADEON_SOFT_RESET_SE | + RADEON_SOFT_RESET_RE | + RADEON_SOFT_RESET_PP | + RADEON_SOFT_RESET_E2 | + RADEON_SOFT_RESET_RB))); + RADEON_READ(RADEON_RBBM_SOFT_RESET); + + if ((dev_priv->flags & RADEON_FAMILY_MASK) <= CHIP_RV410) { + RADEON_WRITE_PLL(RADEON_MCLK_CNTL, mclk_cntl); + RADEON_WRITE(RADEON_CLOCK_CNTL_INDEX, clock_cntl_index); + RADEON_WRITE(RADEON_RBBM_SOFT_RESET, rbbm_soft_reset); + } + + /* setup the raster pipes */ + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R300) + radeon_init_pipes(dev); + + /* Reset the CP ring */ + radeon_do_cp_reset(dev_priv); + + /* The CP is no longer running after an engine reset */ + dev_priv->cp_running = 0; + + /* Reset any pending vertex, indirect buffers */ + radeon_freelist_reset(dev); + + return 0; +} + +static void radeon_cp_init_ring_buffer(struct drm_device * dev, + drm_radeon_private_t *dev_priv, + struct drm_file *file_priv) +{ + struct drm_radeon_master_private *master_priv; + u32 ring_start, cur_read_ptr; + + /* Initialize the memory controller. With new memory map, the fb location + * is not changed, it should have been properly initialized already. Part + * of the problem is that the code below is bogus, assuming the GART is + * always appended to the fb which is not necessarily the case + */ + if (!dev_priv->new_memmap) + radeon_write_fb_location(dev_priv, + ((dev_priv->gart_vm_start - 1) & 0xffff0000) + | (dev_priv->fb_location >> 16)); + +#if __OS_HAS_AGP + if (dev_priv->flags & RADEON_IS_AGP) { + radeon_write_agp_base(dev_priv, dev->agp->base); + + radeon_write_agp_location(dev_priv, + (((dev_priv->gart_vm_start - 1 + + dev_priv->gart_size) & 0xffff0000) | + (dev_priv->gart_vm_start >> 16))); + + ring_start = (dev_priv->cp_ring->offset + - dev->agp->base + + dev_priv->gart_vm_start); + } else +#endif + ring_start = (dev_priv->cp_ring->offset + - (unsigned long)dev->sg->virtual + + dev_priv->gart_vm_start); + + RADEON_WRITE(RADEON_CP_RB_BASE, ring_start); + + /* Set the write pointer delay */ + RADEON_WRITE(RADEON_CP_RB_WPTR_DELAY, 0); + + /* Initialize the ring buffer's read and write pointers */ + cur_read_ptr = RADEON_READ(RADEON_CP_RB_RPTR); + RADEON_WRITE(RADEON_CP_RB_WPTR, cur_read_ptr); + SET_RING_HEAD(dev_priv, cur_read_ptr); + dev_priv->ring.tail = cur_read_ptr; + +#if __OS_HAS_AGP + if (dev_priv->flags & RADEON_IS_AGP) { + RADEON_WRITE(RADEON_CP_RB_RPTR_ADDR, + dev_priv->ring_rptr->offset + - dev->agp->base + dev_priv->gart_vm_start); + } else +#endif + { + RADEON_WRITE(RADEON_CP_RB_RPTR_ADDR, + dev_priv->ring_rptr->offset + - ((unsigned long) dev->sg->virtual) + + dev_priv->gart_vm_start); + } + + /* Set ring buffer size */ +#ifdef __BIG_ENDIAN + RADEON_WRITE(RADEON_CP_RB_CNTL, + RADEON_BUF_SWAP_32BIT | + (dev_priv->ring.fetch_size_l2ow << 18) | + (dev_priv->ring.rptr_update_l2qw << 8) | + dev_priv->ring.size_l2qw); +#else + RADEON_WRITE(RADEON_CP_RB_CNTL, + (dev_priv->ring.fetch_size_l2ow << 18) | + (dev_priv->ring.rptr_update_l2qw << 8) | + dev_priv->ring.size_l2qw); +#endif + + + /* Initialize the scratch register pointer. This will cause + * the scratch register values to be written out to memory + * whenever they are updated. + * + * We simply put this behind the ring read pointer, this works + * with PCI GART as well as (whatever kind of) AGP GART + */ + RADEON_WRITE(RADEON_SCRATCH_ADDR, RADEON_READ(RADEON_CP_RB_RPTR_ADDR) + + RADEON_SCRATCH_REG_OFFSET); + + RADEON_WRITE(RADEON_SCRATCH_UMSK, 0x7); + + radeon_enable_bm(dev_priv); + + radeon_write_ring_rptr(dev_priv, RADEON_SCRATCHOFF(0), 0); + RADEON_WRITE(RADEON_LAST_FRAME_REG, 0); + + radeon_write_ring_rptr(dev_priv, RADEON_SCRATCHOFF(1), 0); + RADEON_WRITE(RADEON_LAST_DISPATCH_REG, 0); + + radeon_write_ring_rptr(dev_priv, RADEON_SCRATCHOFF(2), 0); + RADEON_WRITE(RADEON_LAST_CLEAR_REG, 0); + + /* reset sarea copies of these */ + master_priv = file_priv->master->driver_priv; + if (master_priv->sarea_priv) { + master_priv->sarea_priv->last_frame = 0; + master_priv->sarea_priv->last_dispatch = 0; + master_priv->sarea_priv->last_clear = 0; + } + + radeon_do_wait_for_idle(dev_priv); + + /* Sync everything up */ + RADEON_WRITE(RADEON_ISYNC_CNTL, + (RADEON_ISYNC_ANY2D_IDLE3D | + RADEON_ISYNC_ANY3D_IDLE2D | + RADEON_ISYNC_WAIT_IDLEGUI | + RADEON_ISYNC_CPSCRATCH_IDLEGUI)); + +} + +static void radeon_test_writeback(drm_radeon_private_t * dev_priv) +{ + u32 tmp; + + /* Start with assuming that writeback doesn't work */ + dev_priv->writeback_works = 0; + + /* Writeback doesn't seem to work everywhere, test it here and possibly + * enable it if it appears to work + */ + radeon_write_ring_rptr(dev_priv, RADEON_SCRATCHOFF(1), 0); + + RADEON_WRITE(RADEON_SCRATCH_REG1, 0xdeadbeef); + + for (tmp = 0; tmp < dev_priv->usec_timeout; tmp++) { + u32 val; + + val = radeon_read_ring_rptr(dev_priv, RADEON_SCRATCHOFF(1)); + if (val == 0xdeadbeef) + break; + DRM_UDELAY(1); + } + + if (tmp < dev_priv->usec_timeout) { + dev_priv->writeback_works = 1; + DRM_INFO("writeback test succeeded in %d usecs\n", tmp); + } else { + dev_priv->writeback_works = 0; + DRM_INFO("writeback test failed\n"); + } + if (radeon_no_wb == 1) { + dev_priv->writeback_works = 0; + DRM_INFO("writeback forced off\n"); + } + + if (!dev_priv->writeback_works) { + /* Disable writeback to avoid unnecessary bus master transfer */ + RADEON_WRITE(RADEON_CP_RB_CNTL, RADEON_READ(RADEON_CP_RB_CNTL) | + RADEON_RB_NO_UPDATE); + RADEON_WRITE(RADEON_SCRATCH_UMSK, 0); + } +} + +/* Enable or disable IGP GART on the chip */ +static void radeon_set_igpgart(drm_radeon_private_t * dev_priv, int on) +{ + u32 temp; + + if (on) { + DRM_DEBUG("programming igp gart %08X %08lX %08X\n", + dev_priv->gart_vm_start, + (long)dev_priv->gart_info.bus_addr, + dev_priv->gart_size); + + temp = IGP_READ_MCIND(dev_priv, RS480_MC_MISC_CNTL); + if (((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS690) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS740)) + IGP_WRITE_MCIND(RS480_MC_MISC_CNTL, (RS480_GART_INDEX_REG_EN | + RS690_BLOCK_GFX_D3_EN)); + else + IGP_WRITE_MCIND(RS480_MC_MISC_CNTL, RS480_GART_INDEX_REG_EN); + + IGP_WRITE_MCIND(RS480_AGP_ADDRESS_SPACE_SIZE, (RS480_GART_EN | + RS480_VA_SIZE_32MB)); + + temp = IGP_READ_MCIND(dev_priv, RS480_GART_FEATURE_ID); + IGP_WRITE_MCIND(RS480_GART_FEATURE_ID, (RS480_HANG_EN | + RS480_TLB_ENABLE | + RS480_GTW_LAC_EN | + RS480_1LEVEL_GART)); + + temp = dev_priv->gart_info.bus_addr & 0xfffff000; + temp |= (upper_32_bits(dev_priv->gart_info.bus_addr) & 0xff) << 4; + IGP_WRITE_MCIND(RS480_GART_BASE, temp); + + temp = IGP_READ_MCIND(dev_priv, RS480_AGP_MODE_CNTL); + IGP_WRITE_MCIND(RS480_AGP_MODE_CNTL, ((1 << RS480_REQ_TYPE_SNOOP_SHIFT) | + RS480_REQ_TYPE_SNOOP_DIS)); + + radeon_write_agp_base(dev_priv, dev_priv->gart_vm_start); + + dev_priv->gart_size = 32*1024*1024; + temp = (((dev_priv->gart_vm_start - 1 + dev_priv->gart_size) & + 0xffff0000) | (dev_priv->gart_vm_start >> 16)); + + radeon_write_agp_location(dev_priv, temp); + + temp = IGP_READ_MCIND(dev_priv, RS480_AGP_ADDRESS_SPACE_SIZE); + IGP_WRITE_MCIND(RS480_AGP_ADDRESS_SPACE_SIZE, (RS480_GART_EN | + RS480_VA_SIZE_32MB)); + + do { + temp = IGP_READ_MCIND(dev_priv, RS480_GART_CACHE_CNTRL); + if ((temp & RS480_GART_CACHE_INVALIDATE) == 0) + break; + DRM_UDELAY(1); + } while (1); + + IGP_WRITE_MCIND(RS480_GART_CACHE_CNTRL, + RS480_GART_CACHE_INVALIDATE); + + do { + temp = IGP_READ_MCIND(dev_priv, RS480_GART_CACHE_CNTRL); + if ((temp & RS480_GART_CACHE_INVALIDATE) == 0) + break; + DRM_UDELAY(1); + } while (1); + + IGP_WRITE_MCIND(RS480_GART_CACHE_CNTRL, 0); + } else { + IGP_WRITE_MCIND(RS480_AGP_ADDRESS_SPACE_SIZE, 0); + } +} + +/* Enable or disable IGP GART on the chip */ +static void rs600_set_igpgart(drm_radeon_private_t *dev_priv, int on) +{ + u32 temp; + int i; + + if (on) { + DRM_DEBUG("programming igp gart %08X %08lX %08X\n", + dev_priv->gart_vm_start, + (long)dev_priv->gart_info.bus_addr, + dev_priv->gart_size); + + IGP_WRITE_MCIND(RS600_MC_PT0_CNTL, (RS600_EFFECTIVE_L2_CACHE_SIZE(6) | + RS600_EFFECTIVE_L2_QUEUE_SIZE(6))); + + for (i = 0; i < 19; i++) + IGP_WRITE_MCIND(RS600_MC_PT0_CLIENT0_CNTL + i, + (RS600_ENABLE_TRANSLATION_MODE_OVERRIDE | + RS600_SYSTEM_ACCESS_MODE_IN_SYS | + RS600_SYSTEM_APERTURE_UNMAPPED_ACCESS_PASSTHROUGH | + RS600_EFFECTIVE_L1_CACHE_SIZE(3) | + RS600_ENABLE_FRAGMENT_PROCESSING | + RS600_EFFECTIVE_L1_QUEUE_SIZE(3))); + + IGP_WRITE_MCIND(RS600_MC_PT0_CONTEXT0_CNTL, (RS600_ENABLE_PAGE_TABLE | + RS600_PAGE_TABLE_TYPE_FLAT)); + + /* disable all other contexts */ + for (i = 1; i < 8; i++) + IGP_WRITE_MCIND(RS600_MC_PT0_CONTEXT0_CNTL + i, 0); + + /* setup the page table aperture */ + IGP_WRITE_MCIND(RS600_MC_PT0_CONTEXT0_FLAT_BASE_ADDR, + dev_priv->gart_info.bus_addr); + IGP_WRITE_MCIND(RS600_MC_PT0_CONTEXT0_FLAT_START_ADDR, + dev_priv->gart_vm_start); + IGP_WRITE_MCIND(RS600_MC_PT0_CONTEXT0_FLAT_END_ADDR, + (dev_priv->gart_vm_start + dev_priv->gart_size - 1)); + IGP_WRITE_MCIND(RS600_MC_PT0_CONTEXT0_DEFAULT_READ_ADDR, 0); + + /* setup the system aperture */ + IGP_WRITE_MCIND(RS600_MC_PT0_SYSTEM_APERTURE_LOW_ADDR, + dev_priv->gart_vm_start); + IGP_WRITE_MCIND(RS600_MC_PT0_SYSTEM_APERTURE_HIGH_ADDR, + (dev_priv->gart_vm_start + dev_priv->gart_size - 1)); + + /* enable page tables */ + temp = IGP_READ_MCIND(dev_priv, RS600_MC_PT0_CNTL); + IGP_WRITE_MCIND(RS600_MC_PT0_CNTL, (temp | RS600_ENABLE_PT)); + + temp = IGP_READ_MCIND(dev_priv, RS600_MC_CNTL1); + IGP_WRITE_MCIND(RS600_MC_CNTL1, (temp | RS600_ENABLE_PAGE_TABLES)); + + /* invalidate the cache */ + temp = IGP_READ_MCIND(dev_priv, RS600_MC_PT0_CNTL); + + temp &= ~(RS600_INVALIDATE_ALL_L1_TLBS | RS600_INVALIDATE_L2_CACHE); + IGP_WRITE_MCIND(RS600_MC_PT0_CNTL, temp); + temp = IGP_READ_MCIND(dev_priv, RS600_MC_PT0_CNTL); + + temp |= RS600_INVALIDATE_ALL_L1_TLBS | RS600_INVALIDATE_L2_CACHE; + IGP_WRITE_MCIND(RS600_MC_PT0_CNTL, temp); + temp = IGP_READ_MCIND(dev_priv, RS600_MC_PT0_CNTL); + + temp &= ~(RS600_INVALIDATE_ALL_L1_TLBS | RS600_INVALIDATE_L2_CACHE); + IGP_WRITE_MCIND(RS600_MC_PT0_CNTL, temp); + temp = IGP_READ_MCIND(dev_priv, RS600_MC_PT0_CNTL); + + } else { + IGP_WRITE_MCIND(RS600_MC_PT0_CNTL, 0); + temp = IGP_READ_MCIND(dev_priv, RS600_MC_CNTL1); + temp &= ~RS600_ENABLE_PAGE_TABLES; + IGP_WRITE_MCIND(RS600_MC_CNTL1, temp); + } +} + +static void radeon_set_pciegart(drm_radeon_private_t * dev_priv, int on) +{ + u32 tmp = RADEON_READ_PCIE(dev_priv, RADEON_PCIE_TX_GART_CNTL); + if (on) { + + DRM_DEBUG("programming pcie %08X %08lX %08X\n", + dev_priv->gart_vm_start, + (long)dev_priv->gart_info.bus_addr, + dev_priv->gart_size); + RADEON_WRITE_PCIE(RADEON_PCIE_TX_DISCARD_RD_ADDR_LO, + dev_priv->gart_vm_start); + RADEON_WRITE_PCIE(RADEON_PCIE_TX_GART_BASE, + dev_priv->gart_info.bus_addr); + RADEON_WRITE_PCIE(RADEON_PCIE_TX_GART_START_LO, + dev_priv->gart_vm_start); + RADEON_WRITE_PCIE(RADEON_PCIE_TX_GART_END_LO, + dev_priv->gart_vm_start + + dev_priv->gart_size - 1); + + radeon_write_agp_location(dev_priv, 0xffffffc0); /* ?? */ + + RADEON_WRITE_PCIE(RADEON_PCIE_TX_GART_CNTL, + RADEON_PCIE_TX_GART_EN); + } else { + RADEON_WRITE_PCIE(RADEON_PCIE_TX_GART_CNTL, + tmp & ~RADEON_PCIE_TX_GART_EN); + } +} + +/* Enable or disable PCI GART on the chip */ +static void radeon_set_pcigart(drm_radeon_private_t * dev_priv, int on) +{ + u32 tmp; + + if (((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS690) || + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS740) || + (dev_priv->flags & RADEON_IS_IGPGART)) { + radeon_set_igpgart(dev_priv, on); + return; + } + + if ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS600) { + rs600_set_igpgart(dev_priv, on); + return; + } + + if (dev_priv->flags & RADEON_IS_PCIE) { + radeon_set_pciegart(dev_priv, on); + return; + } + + tmp = RADEON_READ(RADEON_AIC_CNTL); + + if (on) { + RADEON_WRITE(RADEON_AIC_CNTL, + tmp | RADEON_PCIGART_TRANSLATE_EN); + + /* set PCI GART page-table base address + */ + RADEON_WRITE(RADEON_AIC_PT_BASE, dev_priv->gart_info.bus_addr); + + /* set address range for PCI address translate + */ + RADEON_WRITE(RADEON_AIC_LO_ADDR, dev_priv->gart_vm_start); + RADEON_WRITE(RADEON_AIC_HI_ADDR, dev_priv->gart_vm_start + + dev_priv->gart_size - 1); + + /* Turn off AGP aperture -- is this required for PCI GART? + */ + radeon_write_agp_location(dev_priv, 0xffffffc0); + RADEON_WRITE(RADEON_AGP_COMMAND, 0); /* clear AGP_COMMAND */ + } else { + RADEON_WRITE(RADEON_AIC_CNTL, + tmp & ~RADEON_PCIGART_TRANSLATE_EN); + } +} + +static int radeon_setup_pcigart_surface(drm_radeon_private_t *dev_priv) +{ + struct drm_ati_pcigart_info *gart_info = &dev_priv->gart_info; + struct radeon_virt_surface *vp; + int i; + + for (i = 0; i < RADEON_MAX_SURFACES * 2; i++) { + if (!dev_priv->virt_surfaces[i].file_priv || + dev_priv->virt_surfaces[i].file_priv == PCIGART_FILE_PRIV) + break; + } + if (i >= 2 * RADEON_MAX_SURFACES) + return -ENOMEM; + vp = &dev_priv->virt_surfaces[i]; + + for (i = 0; i < RADEON_MAX_SURFACES; i++) { + struct radeon_surface *sp = &dev_priv->surfaces[i]; + if (sp->refcount) + continue; + + vp->surface_index = i; + vp->lower = gart_info->bus_addr; + vp->upper = vp->lower + gart_info->table_size; + vp->flags = 0; + vp->file_priv = PCIGART_FILE_PRIV; + + sp->refcount = 1; + sp->lower = vp->lower; + sp->upper = vp->upper; + sp->flags = 0; + + RADEON_WRITE(RADEON_SURFACE0_INFO + 16 * i, sp->flags); + RADEON_WRITE(RADEON_SURFACE0_LOWER_BOUND + 16 * i, sp->lower); + RADEON_WRITE(RADEON_SURFACE0_UPPER_BOUND + 16 * i, sp->upper); + return 0; + } + + return -ENOMEM; +} + +static int radeon_do_init_cp(struct drm_device *dev, drm_radeon_init_t *init, + struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + struct drm_radeon_master_private *master_priv = file_priv->master->driver_priv; + + DRM_DEBUG("\n"); + + /* if we require new memory map but we don't have it fail */ + if ((dev_priv->flags & RADEON_NEW_MEMMAP) && !dev_priv->new_memmap) { + DRM_ERROR("Cannot initialise DRM on this card\nThis card requires a new X.org DDX for 3D\n"); + radeon_do_cleanup_cp(dev); + return -EINVAL; + } + + if (init->is_pci && (dev_priv->flags & RADEON_IS_AGP)) { + DRM_DEBUG("Forcing AGP card to PCI mode\n"); + dev_priv->flags &= ~RADEON_IS_AGP; + } else if (!(dev_priv->flags & (RADEON_IS_AGP | RADEON_IS_PCI | RADEON_IS_PCIE)) + && !init->is_pci) { + DRM_DEBUG("Restoring AGP flag\n"); + dev_priv->flags |= RADEON_IS_AGP; + } + + if ((!(dev_priv->flags & RADEON_IS_AGP)) && !dev->sg) { + DRM_ERROR("PCI GART memory not allocated!\n"); + radeon_do_cleanup_cp(dev); + return -EINVAL; + } + + dev_priv->usec_timeout = init->usec_timeout; + if (dev_priv->usec_timeout < 1 || + dev_priv->usec_timeout > RADEON_MAX_USEC_TIMEOUT) { + DRM_DEBUG("TIMEOUT problem!\n"); + radeon_do_cleanup_cp(dev); + return -EINVAL; + } + + /* Enable vblank on CRTC1 for older X servers + */ + dev_priv->vblank_crtc = DRM_RADEON_VBLANK_CRTC1; + + switch(init->func) { + case RADEON_INIT_R200_CP: + dev_priv->microcode_version = UCODE_R200; + break; + case RADEON_INIT_R300_CP: + dev_priv->microcode_version = UCODE_R300; + break; + default: + dev_priv->microcode_version = UCODE_R100; + } + + dev_priv->do_boxes = 0; + dev_priv->cp_mode = init->cp_mode; + + /* We don't support anything other than bus-mastering ring mode, + * but the ring can be in either AGP or PCI space for the ring + * read pointer. + */ + if ((init->cp_mode != RADEON_CSQ_PRIBM_INDDIS) && + (init->cp_mode != RADEON_CSQ_PRIBM_INDBM)) { + DRM_DEBUG("BAD cp_mode (%x)!\n", init->cp_mode); + radeon_do_cleanup_cp(dev); + return -EINVAL; + } + + switch (init->fb_bpp) { + case 16: + dev_priv->color_fmt = RADEON_COLOR_FORMAT_RGB565; + break; + case 32: + default: + dev_priv->color_fmt = RADEON_COLOR_FORMAT_ARGB8888; + break; + } + dev_priv->front_offset = init->front_offset; + dev_priv->front_pitch = init->front_pitch; + dev_priv->back_offset = init->back_offset; + dev_priv->back_pitch = init->back_pitch; + + switch (init->depth_bpp) { + case 16: + dev_priv->depth_fmt = RADEON_DEPTH_FORMAT_16BIT_INT_Z; + break; + case 32: + default: + dev_priv->depth_fmt = RADEON_DEPTH_FORMAT_24BIT_INT_Z; + break; + } + dev_priv->depth_offset = init->depth_offset; + dev_priv->depth_pitch = init->depth_pitch; + + /* Hardware state for depth clears. Remove this if/when we no + * longer clear the depth buffer with a 3D rectangle. Hard-code + * all values to prevent unwanted 3D state from slipping through + * and screwing with the clear operation. + */ + dev_priv->depth_clear.rb3d_cntl = (RADEON_PLANE_MASK_ENABLE | + (dev_priv->color_fmt << 10) | + (dev_priv->microcode_version == + UCODE_R100 ? RADEON_ZBLOCK16 : 0)); + + dev_priv->depth_clear.rb3d_zstencilcntl = + (dev_priv->depth_fmt | + RADEON_Z_TEST_ALWAYS | + RADEON_STENCIL_TEST_ALWAYS | + RADEON_STENCIL_S_FAIL_REPLACE | + RADEON_STENCIL_ZPASS_REPLACE | + RADEON_STENCIL_ZFAIL_REPLACE | RADEON_Z_WRITE_ENABLE); + + dev_priv->depth_clear.se_cntl = (RADEON_FFACE_CULL_CW | + RADEON_BFACE_SOLID | + RADEON_FFACE_SOLID | + RADEON_FLAT_SHADE_VTX_LAST | + RADEON_DIFFUSE_SHADE_FLAT | + RADEON_ALPHA_SHADE_FLAT | + RADEON_SPECULAR_SHADE_FLAT | + RADEON_FOG_SHADE_FLAT | + RADEON_VTX_PIX_CENTER_OGL | + RADEON_ROUND_MODE_TRUNC | + RADEON_ROUND_PREC_8TH_PIX); + + + dev_priv->ring_offset = init->ring_offset; + dev_priv->ring_rptr_offset = init->ring_rptr_offset; + dev_priv->buffers_offset = init->buffers_offset; + dev_priv->gart_textures_offset = init->gart_textures_offset; + + master_priv->sarea = drm_getsarea(dev); + if (!master_priv->sarea) { + DRM_ERROR("could not find sarea!\n"); + radeon_do_cleanup_cp(dev); + return -EINVAL; + } + + dev_priv->cp_ring = drm_core_findmap(dev, init->ring_offset); + if (!dev_priv->cp_ring) { + DRM_ERROR("could not find cp ring region!\n"); + radeon_do_cleanup_cp(dev); + return -EINVAL; + } + dev_priv->ring_rptr = drm_core_findmap(dev, init->ring_rptr_offset); + if (!dev_priv->ring_rptr) { + DRM_ERROR("could not find ring read pointer!\n"); + radeon_do_cleanup_cp(dev); + return -EINVAL; + } + dev->agp_buffer_token = init->buffers_offset; + dev->agp_buffer_map = drm_core_findmap(dev, init->buffers_offset); + if (!dev->agp_buffer_map) { + DRM_ERROR("could not find dma buffer region!\n"); + radeon_do_cleanup_cp(dev); + return -EINVAL; + } + + if (init->gart_textures_offset) { + dev_priv->gart_textures = + drm_core_findmap(dev, init->gart_textures_offset); + if (!dev_priv->gart_textures) { + DRM_ERROR("could not find GART texture region!\n"); + radeon_do_cleanup_cp(dev); + return -EINVAL; + } + } + +#if __OS_HAS_AGP + if (dev_priv->flags & RADEON_IS_AGP) { + drm_core_ioremap_wc(dev_priv->cp_ring, dev); + drm_core_ioremap_wc(dev_priv->ring_rptr, dev); + drm_core_ioremap_wc(dev->agp_buffer_map, dev); + if (!dev_priv->cp_ring->handle || + !dev_priv->ring_rptr->handle || + !dev->agp_buffer_map->handle) { + DRM_ERROR("could not find ioremap agp regions!\n"); + radeon_do_cleanup_cp(dev); + return -EINVAL; + } + } else +#endif + { + dev_priv->cp_ring->handle = + (void *)(unsigned long)dev_priv->cp_ring->offset; + dev_priv->ring_rptr->handle = + (void *)(unsigned long)dev_priv->ring_rptr->offset; + dev->agp_buffer_map->handle = + (void *)(unsigned long)dev->agp_buffer_map->offset; + + DRM_DEBUG("dev_priv->cp_ring->handle %p\n", + dev_priv->cp_ring->handle); + DRM_DEBUG("dev_priv->ring_rptr->handle %p\n", + dev_priv->ring_rptr->handle); + DRM_DEBUG("dev->agp_buffer_map->handle %p\n", + dev->agp_buffer_map->handle); + } + + dev_priv->fb_location = (radeon_read_fb_location(dev_priv) & 0xffff) << 16; + dev_priv->fb_size = + ((radeon_read_fb_location(dev_priv) & 0xffff0000u) + 0x10000) + - dev_priv->fb_location; + + dev_priv->front_pitch_offset = (((dev_priv->front_pitch / 64) << 22) | + ((dev_priv->front_offset + + dev_priv->fb_location) >> 10)); + + dev_priv->back_pitch_offset = (((dev_priv->back_pitch / 64) << 22) | + ((dev_priv->back_offset + + dev_priv->fb_location) >> 10)); + + dev_priv->depth_pitch_offset = (((dev_priv->depth_pitch / 64) << 22) | + ((dev_priv->depth_offset + + dev_priv->fb_location) >> 10)); + + dev_priv->gart_size = init->gart_size; + + /* New let's set the memory map ... */ + if (dev_priv->new_memmap) { + u32 base = 0; + + DRM_INFO("Setting GART location based on new memory map\n"); + + /* If using AGP, try to locate the AGP aperture at the same + * location in the card and on the bus, though we have to + * align it down. + */ +#if __OS_HAS_AGP + if (dev_priv->flags & RADEON_IS_AGP) { + base = dev->agp->base; + /* Check if valid */ + if ((base + dev_priv->gart_size - 1) >= dev_priv->fb_location && + base < (dev_priv->fb_location + dev_priv->fb_size - 1)) { + DRM_INFO("Can't use AGP base @0x%08lx, won't fit\n", + dev->agp->base); + base = 0; + } + } +#endif + /* If not or if AGP is at 0 (Macs), try to put it elsewhere */ + if (base == 0) { + base = dev_priv->fb_location + dev_priv->fb_size; + if (base < dev_priv->fb_location || + ((base + dev_priv->gart_size) & 0xfffffffful) < base) + base = dev_priv->fb_location + - dev_priv->gart_size; + } + dev_priv->gart_vm_start = base & 0xffc00000u; + if (dev_priv->gart_vm_start != base) + DRM_INFO("GART aligned down from 0x%08x to 0x%08x\n", + base, dev_priv->gart_vm_start); + } else { + DRM_INFO("Setting GART location based on old memory map\n"); + dev_priv->gart_vm_start = dev_priv->fb_location + + RADEON_READ(RADEON_CONFIG_APER_SIZE); + } + +#if __OS_HAS_AGP + if (dev_priv->flags & RADEON_IS_AGP) + dev_priv->gart_buffers_offset = (dev->agp_buffer_map->offset + - dev->agp->base + + dev_priv->gart_vm_start); + else +#endif + dev_priv->gart_buffers_offset = (dev->agp_buffer_map->offset + - (unsigned long)dev->sg->virtual + + dev_priv->gart_vm_start); + + DRM_DEBUG("dev_priv->gart_size %d\n", dev_priv->gart_size); + DRM_DEBUG("dev_priv->gart_vm_start 0x%x\n", dev_priv->gart_vm_start); + DRM_DEBUG("dev_priv->gart_buffers_offset 0x%lx\n", + dev_priv->gart_buffers_offset); + + dev_priv->ring.start = (u32 *) dev_priv->cp_ring->handle; + dev_priv->ring.end = ((u32 *) dev_priv->cp_ring->handle + + init->ring_size / sizeof(u32)); + dev_priv->ring.size = init->ring_size; + dev_priv->ring.size_l2qw = drm_order(init->ring_size / 8); + + dev_priv->ring.rptr_update = /* init->rptr_update */ 4096; + dev_priv->ring.rptr_update_l2qw = drm_order( /* init->rptr_update */ 4096 / 8); + + dev_priv->ring.fetch_size = /* init->fetch_size */ 32; + dev_priv->ring.fetch_size_l2ow = drm_order( /* init->fetch_size */ 32 / 16); + dev_priv->ring.tail_mask = (dev_priv->ring.size / sizeof(u32)) - 1; + + dev_priv->ring.high_mark = RADEON_RING_HIGH_MARK; + +#if __OS_HAS_AGP + if (dev_priv->flags & RADEON_IS_AGP) { + /* Turn off PCI GART */ + radeon_set_pcigart(dev_priv, 0); + } else +#endif + { + u32 sctrl; + int ret; + + dev_priv->gart_info.table_mask = DMA_BIT_MASK(32); + /* if we have an offset set from userspace */ + if (dev_priv->pcigart_offset_set) { + dev_priv->gart_info.bus_addr = + (resource_size_t)dev_priv->pcigart_offset + dev_priv->fb_location; + dev_priv->gart_info.mapping.offset = + dev_priv->pcigart_offset + dev_priv->fb_aper_offset; + dev_priv->gart_info.mapping.size = + dev_priv->gart_info.table_size; + + drm_core_ioremap_wc(&dev_priv->gart_info.mapping, dev); + dev_priv->gart_info.addr = + dev_priv->gart_info.mapping.handle; + + if (dev_priv->flags & RADEON_IS_PCIE) + dev_priv->gart_info.gart_reg_if = DRM_ATI_GART_PCIE; + else + dev_priv->gart_info.gart_reg_if = DRM_ATI_GART_PCI; + dev_priv->gart_info.gart_table_location = + DRM_ATI_GART_FB; + + DRM_DEBUG("Setting phys_pci_gart to %p %08lX\n", + dev_priv->gart_info.addr, + dev_priv->pcigart_offset); + } else { + if (dev_priv->flags & RADEON_IS_IGPGART) + dev_priv->gart_info.gart_reg_if = DRM_ATI_GART_IGP; + else + dev_priv->gart_info.gart_reg_if = DRM_ATI_GART_PCI; + dev_priv->gart_info.gart_table_location = + DRM_ATI_GART_MAIN; + dev_priv->gart_info.addr = NULL; + dev_priv->gart_info.bus_addr = 0; + if (dev_priv->flags & RADEON_IS_PCIE) { + DRM_ERROR + ("Cannot use PCI Express without GART in FB memory\n"); + radeon_do_cleanup_cp(dev); + return -EINVAL; + } + } + + sctrl = RADEON_READ(RADEON_SURFACE_CNTL); + RADEON_WRITE(RADEON_SURFACE_CNTL, 0); + if ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS600) + ret = r600_page_table_init(dev); + else + ret = drm_ati_pcigart_init(dev, &dev_priv->gart_info); + RADEON_WRITE(RADEON_SURFACE_CNTL, sctrl); + + if (!ret) { + DRM_ERROR("failed to init PCI GART!\n"); + radeon_do_cleanup_cp(dev); + return -ENOMEM; + } + + ret = radeon_setup_pcigart_surface(dev_priv); + if (ret) { + DRM_ERROR("failed to setup GART surface!\n"); + if ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS600) + r600_page_table_cleanup(dev, &dev_priv->gart_info); + else + drm_ati_pcigart_cleanup(dev, &dev_priv->gart_info); + radeon_do_cleanup_cp(dev); + return ret; + } + + /* Turn on PCI GART */ + radeon_set_pcigart(dev_priv, 1); + } + + if (!dev_priv->me_fw) { + int err = radeon_cp_init_microcode(dev_priv); + if (err) { + DRM_ERROR("Failed to load firmware!\n"); + radeon_do_cleanup_cp(dev); + return err; + } + } + radeon_cp_load_microcode(dev_priv); + radeon_cp_init_ring_buffer(dev, dev_priv, file_priv); + + dev_priv->last_buf = 0; + + radeon_do_engine_reset(dev); + radeon_test_writeback(dev_priv); + + return 0; +} + +static int radeon_do_cleanup_cp(struct drm_device * dev) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + DRM_DEBUG("\n"); + + /* Make sure interrupts are disabled here because the uninstall ioctl + * may not have been called from userspace and after dev_private + * is freed, it's too late. + */ + if (dev->irq_enabled) + drm_irq_uninstall(dev); + +#if __OS_HAS_AGP + if (dev_priv->flags & RADEON_IS_AGP) { + if (dev_priv->cp_ring != NULL) { + drm_core_ioremapfree(dev_priv->cp_ring, dev); + dev_priv->cp_ring = NULL; + } + if (dev_priv->ring_rptr != NULL) { + drm_core_ioremapfree(dev_priv->ring_rptr, dev); + dev_priv->ring_rptr = NULL; + } + if (dev->agp_buffer_map != NULL) { + drm_core_ioremapfree(dev->agp_buffer_map, dev); + dev->agp_buffer_map = NULL; + } + } else +#endif + { + + if (dev_priv->gart_info.bus_addr) { + /* Turn off PCI GART */ + radeon_set_pcigart(dev_priv, 0); + if ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS600) + r600_page_table_cleanup(dev, &dev_priv->gart_info); + else { + if (!drm_ati_pcigart_cleanup(dev, &dev_priv->gart_info)) + DRM_ERROR("failed to cleanup PCI GART!\n"); + } + } + + if (dev_priv->gart_info.gart_table_location == DRM_ATI_GART_FB) + { + drm_core_ioremapfree(&dev_priv->gart_info.mapping, dev); + dev_priv->gart_info.addr = NULL; + } + } + /* only clear to the start of flags */ + memset(dev_priv, 0, offsetof(drm_radeon_private_t, flags)); + + return 0; +} + +/* This code will reinit the Radeon CP hardware after a resume from disc. + * AFAIK, it would be very difficult to pickle the state at suspend time, so + * here we make sure that all Radeon hardware initialisation is re-done without + * affecting running applications. + * + * Charl P. Botha + */ +static int radeon_do_resume_cp(struct drm_device *dev, struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + + if (!dev_priv) { + DRM_ERROR("Called with no initialization\n"); + return -EINVAL; + } + + DRM_DEBUG("Starting radeon_do_resume_cp()\n"); + +#if __OS_HAS_AGP + if (dev_priv->flags & RADEON_IS_AGP) { + /* Turn off PCI GART */ + radeon_set_pcigart(dev_priv, 0); + } else +#endif + { + /* Turn on PCI GART */ + radeon_set_pcigart(dev_priv, 1); + } + + radeon_cp_load_microcode(dev_priv); + radeon_cp_init_ring_buffer(dev, dev_priv, file_priv); + + dev_priv->have_z_offset = 0; + radeon_do_engine_reset(dev); + radeon_irq_set_state(dev, RADEON_SW_INT_ENABLE, 1); + + DRM_DEBUG("radeon_do_resume_cp() complete\n"); + + return 0; +} + +int radeon_cp_init(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + drm_radeon_init_t *init = data; + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + if (init->func == RADEON_INIT_R300_CP) + r300_init_reg_flags(dev); + + switch (init->func) { + case RADEON_INIT_CP: + case RADEON_INIT_R200_CP: + case RADEON_INIT_R300_CP: + return radeon_do_init_cp(dev, init, file_priv); + case RADEON_INIT_R600_CP: + return r600_do_init_cp(dev, init, file_priv); + case RADEON_CLEANUP_CP: + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) + return r600_do_cleanup_cp(dev); + else + return radeon_do_cleanup_cp(dev); + } + + return -EINVAL; +} + +int radeon_cp_start(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + DRM_DEBUG("\n"); + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + if (dev_priv->cp_running) { + DRM_DEBUG("while CP running\n"); + return 0; + } + if (dev_priv->cp_mode == RADEON_CSQ_PRIDIS_INDDIS) { + DRM_DEBUG("called with bogus CP mode (%d)\n", + dev_priv->cp_mode); + return 0; + } + + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) + r600_do_cp_start(dev_priv); + else + radeon_do_cp_start(dev_priv); + + return 0; +} + +/* Stop the CP. The engine must have been idled before calling this + * routine. + */ +int radeon_cp_stop(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + drm_radeon_cp_stop_t *stop = data; + int ret; + DRM_DEBUG("\n"); + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + if (!dev_priv->cp_running) + return 0; + + /* Flush any pending CP commands. This ensures any outstanding + * commands are exectuted by the engine before we turn it off. + */ + if (stop->flush) { + radeon_do_cp_flush(dev_priv); + } + + /* If we fail to make the engine go idle, we return an error + * code so that the DRM ioctl wrapper can try again. + */ + if (stop->idle) { + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) + ret = r600_do_cp_idle(dev_priv); + else + ret = radeon_do_cp_idle(dev_priv); + if (ret) + return ret; + } + + /* Finally, we can turn off the CP. If the engine isn't idle, + * we will get some dropped triangles as they won't be fully + * rendered before the CP is shut down. + */ + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) + r600_do_cp_stop(dev_priv); + else + radeon_do_cp_stop(dev_priv); + + /* Reset the engine */ + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) + r600_do_engine_reset(dev); + else + radeon_do_engine_reset(dev); + + return 0; +} + +void radeon_do_release(struct drm_device * dev) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + int i, ret; + + if (dev_priv) { + if (dev_priv->cp_running) { + /* Stop the cp */ + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) { + while ((ret = r600_do_cp_idle(dev_priv)) != 0) { + DRM_DEBUG("radeon_do_cp_idle %d\n", ret); +#ifdef __linux__ + schedule(); +#else + tsleep(&ret, PZERO, "rdnrel", 1); +#endif + } + } else { + while ((ret = radeon_do_cp_idle(dev_priv)) != 0) { + DRM_DEBUG("radeon_do_cp_idle %d\n", ret); +#ifdef __linux__ + schedule(); +#else + tsleep(&ret, PZERO, "rdnrel", 1); +#endif + } + } + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) { + r600_do_cp_stop(dev_priv); + r600_do_engine_reset(dev); + } else { + radeon_do_cp_stop(dev_priv); + radeon_do_engine_reset(dev); + } + } + + if ((dev_priv->flags & RADEON_FAMILY_MASK) < CHIP_R600) { + /* Disable *all* interrupts */ + if (dev_priv->mmio) /* remove this after permanent addmaps */ + RADEON_WRITE(RADEON_GEN_INT_CNTL, 0); + + if (dev_priv->mmio) { /* remove all surfaces */ + for (i = 0; i < RADEON_MAX_SURFACES; i++) { + RADEON_WRITE(RADEON_SURFACE0_INFO + 16 * i, 0); + RADEON_WRITE(RADEON_SURFACE0_LOWER_BOUND + + 16 * i, 0); + RADEON_WRITE(RADEON_SURFACE0_UPPER_BOUND + + 16 * i, 0); + } + } + } + + /* Free memory heap structures */ + radeon_mem_takedown(&(dev_priv->gart_heap)); + radeon_mem_takedown(&(dev_priv->fb_heap)); + + /* deallocate kernel resources */ + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) + r600_do_cleanup_cp(dev); + else + radeon_do_cleanup_cp(dev); + if (dev_priv->me_fw) { + release_firmware(dev_priv->me_fw); + dev_priv->me_fw = NULL; + } + if (dev_priv->pfp_fw) { + release_firmware(dev_priv->pfp_fw); + dev_priv->pfp_fw = NULL; + } + } +} + +/* Just reset the CP ring. Called as part of an X Server engine reset. + */ +int radeon_cp_reset(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + DRM_DEBUG("\n"); + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + if (!dev_priv) { + DRM_DEBUG("called before init done\n"); + return -EINVAL; + } + + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) + r600_do_cp_reset(dev_priv); + else + radeon_do_cp_reset(dev_priv); + + /* The CP is no longer running after an engine reset */ + dev_priv->cp_running = 0; + + return 0; +} + +int radeon_cp_idle(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + DRM_DEBUG("\n"); + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) + return r600_do_cp_idle(dev_priv); + else + return radeon_do_cp_idle(dev_priv); +} + +/* Added by Charl P. Botha to call radeon_do_resume_cp(). + */ +int radeon_cp_resume(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + DRM_DEBUG("\n"); + + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) + return r600_do_resume_cp(dev, file_priv); + else + return radeon_do_resume_cp(dev, file_priv); +} + +int radeon_engine_reset(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + DRM_DEBUG("\n"); + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) + return r600_do_engine_reset(dev); + else + return radeon_do_engine_reset(dev); +} + +/* ================================================================ + * Fullscreen mode + */ + +/* KW: Deprecated to say the least: + */ +int radeon_fullscreen(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + return 0; +} + +/* ================================================================ + * Freelist management + */ + +/* Original comment: FIXME: ROTATE_BUFS is a hack to cycle through + * bufs until freelist code is used. Note this hides a problem with + * the scratch register * (used to keep track of last buffer + * completed) being written to before * the last buffer has actually + * completed rendering. + * + * KW: It's also a good way to find free buffers quickly. + * + * KW: Ideally this loop wouldn't exist, and freelist_get wouldn't + * sleep. However, bugs in older versions of radeon_accel.c mean that + * we essentially have to do this, else old clients will break. + * + * However, it does leave open a potential deadlock where all the + * buffers are held by other clients, which can't release them because + * they can't get the lock. + */ + +struct drm_buf *radeon_freelist_get(struct drm_device * dev) +{ + struct drm_device_dma *dma = dev->dma; + drm_radeon_private_t *dev_priv = dev->dev_private; + drm_radeon_buf_priv_t *buf_priv; + struct drm_buf *buf; + int i, t; + int start; + + if (++dev_priv->last_buf >= dma->buf_count) + dev_priv->last_buf = 0; + + start = dev_priv->last_buf; + + for (t = 0; t < dev_priv->usec_timeout; t++) { + u32 done_age = GET_SCRATCH(dev_priv, 1); + DRM_DEBUG("done_age = %d\n", done_age); + for (i = 0; i < dma->buf_count; i++) { + buf = dma->buflist[start]; + buf_priv = buf->dev_private; + if (buf->file_priv == NULL || (buf->pending && + buf_priv->age <= + done_age)) { + dev_priv->stats.requested_bufs++; + buf->pending = 0; + return buf; + } + if (++start >= dma->buf_count) + start = 0; + } + + if (t) { + DRM_UDELAY(1); + dev_priv->stats.freelist_loops++; + } + } + + return NULL; +} + +void radeon_freelist_reset(struct drm_device * dev) +{ + struct drm_device_dma *dma = dev->dma; + drm_radeon_private_t *dev_priv = dev->dev_private; + int i; + + dev_priv->last_buf = 0; + for (i = 0; i < dma->buf_count; i++) { + struct drm_buf *buf = dma->buflist[i]; + drm_radeon_buf_priv_t *buf_priv = buf->dev_private; + buf_priv->age = 0; + } +} + +/* ================================================================ + * CP command submission + */ + +int radeon_wait_ring(drm_radeon_private_t * dev_priv, int n) +{ + drm_radeon_ring_buffer_t *ring = &dev_priv->ring; + int i; + u32 last_head = GET_RING_HEAD(dev_priv); + + for (i = 0; i < dev_priv->usec_timeout; i++) { + u32 head = GET_RING_HEAD(dev_priv); + + ring->space = (head - ring->tail) * sizeof(u32); + if (ring->space <= 0) + ring->space += ring->size; + if (ring->space > n) + return 0; + + dev_priv->stats.boxes |= RADEON_BOX_WAIT_IDLE; + + if (head != last_head) + i = 0; + last_head = head; + + DRM_UDELAY(1); + } + + /* FIXME: This return value is ignored in the BEGIN_RING macro! */ +#if RADEON_FIFO_DEBUG + radeon_status(dev_priv); + DRM_ERROR("failed!\n"); +#endif + return -EBUSY; +} + +static int radeon_cp_get_buffers(struct drm_device *dev, + struct drm_file *file_priv, + struct drm_dma * d) +{ + int i; + struct drm_buf *buf; + + for (i = d->granted_count; i < d->request_count; i++) { + buf = radeon_freelist_get(dev); + if (!buf) + return -EBUSY; /* NOTE: broken client */ + + buf->file_priv = file_priv; + + if (DRM_COPY_TO_USER(&d->request_indices[i], &buf->idx, + sizeof(buf->idx))) + return -EFAULT; + if (DRM_COPY_TO_USER(&d->request_sizes[i], &buf->total, + sizeof(buf->total))) + return -EFAULT; + + d->granted_count++; + } + return 0; +} + +int radeon_cp_buffers(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + struct drm_device_dma *dma = dev->dma; + int ret = 0; + struct drm_dma *d = data; + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + /* Please don't send us buffers. + */ + if (d->send_count != 0) { + DRM_ERROR("Process %d trying to send %d buffers via drmDMA\n", + DRM_CURRENTPID, d->send_count); + return -EINVAL; + } + + /* We'll send you buffers. + */ + if (d->request_count < 0 || d->request_count > dma->buf_count) { + DRM_ERROR("Process %d trying to get %d buffers (of %d max)\n", + DRM_CURRENTPID, d->request_count, dma->buf_count); + return -EINVAL; + } + + d->granted_count = 0; + + if (d->request_count) { + ret = radeon_cp_get_buffers(dev, file_priv, d); + } + + return ret; +} + +int radeon_driver_load(struct drm_device *dev, unsigned long flags) +{ + drm_radeon_private_t *dev_priv; + int ret = 0; + + dev_priv = kzalloc(sizeof(drm_radeon_private_t), GFP_KERNEL); + if (dev_priv == NULL) + return -ENOMEM; + + dev->dev_private = (void *)dev_priv; + dev_priv->flags = flags; + + switch (flags & RADEON_FAMILY_MASK) { + case CHIP_R100: + case CHIP_RV200: + case CHIP_R200: + case CHIP_R300: + case CHIP_R350: + case CHIP_R420: + case CHIP_R423: + case CHIP_RV410: + case CHIP_RV515: + case CHIP_R520: + case CHIP_RV570: + case CHIP_R580: + dev_priv->flags |= RADEON_HAS_HIERZ; + break; + default: + /* all other chips have no hierarchical z buffer */ + break; + } + + if (drm_pci_device_is_agp(dev)) + dev_priv->flags |= RADEON_IS_AGP; + else if (drm_pci_device_is_pcie(dev)) + dev_priv->flags |= RADEON_IS_PCIE; + else + dev_priv->flags |= RADEON_IS_PCI; + + ret = drm_addmap(dev, pci_resource_start(dev->pdev, 2), + pci_resource_len(dev->pdev, 2), _DRM_REGISTERS, + _DRM_READ_ONLY | _DRM_DRIVER, &dev_priv->mmio); + if (ret != 0) + return ret; + + ret = drm_vblank_init(dev, 2); + if (ret) { + radeon_driver_unload(dev); + return ret; + } + + DRM_DEBUG("%s card detected\n", + ((dev_priv->flags & RADEON_IS_AGP) ? "AGP" : (((dev_priv->flags & RADEON_IS_PCIE) ? "PCIE" : "PCI")))); + return ret; +} + +int radeon_master_create(struct drm_device *dev, struct drm_master *master) +{ + struct drm_radeon_master_private *master_priv; + unsigned long sareapage; + int ret; + + master_priv = kzalloc(sizeof(*master_priv), GFP_KERNEL); + if (!master_priv) + return -ENOMEM; + + /* prebuild the SAREA */ + sareapage = max_t(unsigned long, SAREA_MAX, PAGE_SIZE); + ret = drm_addmap(dev, 0, sareapage, _DRM_SHM, _DRM_CONTAINS_LOCK, + &master_priv->sarea); + if (ret) { + DRM_ERROR("SAREA setup failed\n"); + kfree(master_priv); + return ret; + } + master_priv->sarea_priv = master_priv->sarea->handle + sizeof(struct drm_sarea); + master_priv->sarea_priv->pfCurrentPage = 0; + + master->driver_priv = master_priv; + return 0; +} + +void radeon_master_destroy(struct drm_device *dev, struct drm_master *master) +{ + struct drm_radeon_master_private *master_priv = master->driver_priv; + + if (!master_priv) + return; + + if (master_priv->sarea_priv && + master_priv->sarea_priv->pfCurrentPage != 0) + radeon_cp_dispatch_flip(dev, master); + + master_priv->sarea_priv = NULL; + if (master_priv->sarea) + drm_rmmap_locked(dev, master_priv->sarea); + + kfree(master_priv); + + master->driver_priv = NULL; +} + +/* Create mappings for registers and framebuffer so userland doesn't necessarily + * have to find them. + */ +int radeon_driver_firstopen(struct drm_device *dev) +{ + int ret; + drm_local_map_t *map; + drm_radeon_private_t *dev_priv = dev->dev_private; + + dev_priv->gart_info.table_size = RADEON_PCIGART_TABLE_SIZE; + + dev_priv->fb_aper_offset = pci_resource_start(dev->pdev, 0); + ret = drm_addmap(dev, dev_priv->fb_aper_offset, + pci_resource_len(dev->pdev, 0), _DRM_FRAME_BUFFER, + _DRM_WRITE_COMBINING, &map); + if (ret != 0) + return ret; + + return 0; +} + +int radeon_driver_unload(struct drm_device *dev) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + + DRM_DEBUG("\n"); + + drm_rmmap(dev, dev_priv->mmio); + + kfree(dev_priv); + + dev->dev_private = NULL; + return 0; +} + +void radeon_commit_ring(drm_radeon_private_t *dev_priv) +{ + int i; + u32 *ring; + int tail_aligned; + + /* check if the ring is padded out to 16-dword alignment */ + + tail_aligned = dev_priv->ring.tail & (RADEON_RING_ALIGN-1); + if (tail_aligned) { + int num_p2 = RADEON_RING_ALIGN - tail_aligned; + + ring = dev_priv->ring.start; + /* pad with some CP_PACKET2 */ + for (i = 0; i < num_p2; i++) + ring[dev_priv->ring.tail + i] = CP_PACKET2(); + + dev_priv->ring.tail += i; + + dev_priv->ring.space -= num_p2 * sizeof(u32); + } + + dev_priv->ring.tail &= dev_priv->ring.tail_mask; + + DRM_MEMORYBARRIER(); + GET_RING_HEAD( dev_priv ); + + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) { + RADEON_WRITE(R600_CP_RB_WPTR, dev_priv->ring.tail); + /* read from PCI bus to ensure correct posting */ + RADEON_READ(R600_CP_RB_RPTR); + } else { + RADEON_WRITE(RADEON_CP_RB_WPTR, dev_priv->ring.tail); + /* read from PCI bus to ensure correct posting */ + RADEON_READ(RADEON_CP_RB_RPTR); + } +} diff --git a/drivers/gpu/drm/radeon/radeon_cs.c b/drivers/gpu/drm/radeon/radeon_cs.c new file mode 100644 index 00000000..fae00c0d --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_cs.c @@ -0,0 +1,341 @@ +/* + * Copyright 2008 Jerome Glisse. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Jerome Glisse + */ +#include "drmP.h" +#include "radeon_drm.h" +#include "radeon_reg.h" +#include "radeon.h" + +void r100_cs_dump_packet(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt); + +int radeon_cs_parser_relocs(struct radeon_cs_parser *p) +{ + struct drm_device *ddev = p->rdev->ddev; + struct radeon_cs_chunk *chunk; + unsigned i, j; + bool duplicate; + + if (p->chunk_relocs_idx == -1) { + return 0; + } + chunk = &p->chunks[p->chunk_relocs_idx]; + /* FIXME: we assume that each relocs use 4 dwords */ + p->nrelocs = chunk->length_dw / 4; + p->relocs_ptr = kcalloc(p->nrelocs, sizeof(void *), GFP_KERNEL); + if (p->relocs_ptr == NULL) { + return -ENOMEM; + } + p->relocs = kcalloc(p->nrelocs, sizeof(struct radeon_cs_reloc), GFP_KERNEL); + if (p->relocs == NULL) { + return -ENOMEM; + } + for (i = 0; i < p->nrelocs; i++) { + struct drm_radeon_cs_reloc *r; + + duplicate = false; + r = (struct drm_radeon_cs_reloc *)&chunk->kdata[i*4]; + for (j = 0; j < p->nrelocs; j++) { + if (r->handle == p->relocs[j].handle) { + p->relocs_ptr[i] = &p->relocs[j]; + duplicate = true; + break; + } + } + if (!duplicate) { + p->relocs[i].gobj = drm_gem_object_lookup(ddev, + p->filp, + r->handle); + if (p->relocs[i].gobj == NULL) { + DRM_ERROR("gem object lookup failed 0x%x\n", + r->handle); + return -ENOENT; + } + p->relocs_ptr[i] = &p->relocs[i]; + p->relocs[i].robj = gem_to_radeon_bo(p->relocs[i].gobj); + p->relocs[i].lobj.bo = p->relocs[i].robj; + p->relocs[i].lobj.wdomain = r->write_domain; + p->relocs[i].lobj.rdomain = r->read_domains; + p->relocs[i].lobj.tv.bo = &p->relocs[i].robj->tbo; + p->relocs[i].handle = r->handle; + p->relocs[i].flags = r->flags; + radeon_bo_list_add_object(&p->relocs[i].lobj, + &p->validated); + } + } + return radeon_bo_list_validate(&p->validated); +} + +int radeon_cs_parser_init(struct radeon_cs_parser *p, void *data) +{ + struct drm_radeon_cs *cs = data; + uint64_t *chunk_array_ptr; + unsigned size, i; + + if (!cs->num_chunks) { + return 0; + } + /* get chunks */ + INIT_LIST_HEAD(&p->validated); + p->idx = 0; + p->chunk_ib_idx = -1; + p->chunk_relocs_idx = -1; + p->chunks_array = kcalloc(cs->num_chunks, sizeof(uint64_t), GFP_KERNEL); + if (p->chunks_array == NULL) { + return -ENOMEM; + } + chunk_array_ptr = (uint64_t *)(unsigned long)(cs->chunks); + if (DRM_COPY_FROM_USER(p->chunks_array, chunk_array_ptr, + sizeof(uint64_t)*cs->num_chunks)) { + return -EFAULT; + } + p->nchunks = cs->num_chunks; + p->chunks = kcalloc(p->nchunks, sizeof(struct radeon_cs_chunk), GFP_KERNEL); + if (p->chunks == NULL) { + return -ENOMEM; + } + for (i = 0; i < p->nchunks; i++) { + struct drm_radeon_cs_chunk __user **chunk_ptr = NULL; + struct drm_radeon_cs_chunk user_chunk; + uint32_t __user *cdata; + + chunk_ptr = (void __user*)(unsigned long)p->chunks_array[i]; + if (DRM_COPY_FROM_USER(&user_chunk, chunk_ptr, + sizeof(struct drm_radeon_cs_chunk))) { + return -EFAULT; + } + p->chunks[i].length_dw = user_chunk.length_dw; + p->chunks[i].kdata = NULL; + p->chunks[i].chunk_id = user_chunk.chunk_id; + + if (p->chunks[i].chunk_id == RADEON_CHUNK_ID_RELOCS) { + p->chunk_relocs_idx = i; + } + if (p->chunks[i].chunk_id == RADEON_CHUNK_ID_IB) { + p->chunk_ib_idx = i; + /* zero length IB isn't useful */ + if (p->chunks[i].length_dw == 0) + return -EINVAL; + } + + p->chunks[i].length_dw = user_chunk.length_dw; + p->chunks[i].user_ptr = (void __user *)(unsigned long)user_chunk.chunk_data; + + cdata = (uint32_t *)(unsigned long)user_chunk.chunk_data; + if (p->chunks[i].chunk_id != RADEON_CHUNK_ID_IB) { + size = p->chunks[i].length_dw * sizeof(uint32_t); + p->chunks[i].kdata = kmalloc(size, GFP_KERNEL); + if (p->chunks[i].kdata == NULL) { + return -ENOMEM; + } + if (DRM_COPY_FROM_USER(p->chunks[i].kdata, + p->chunks[i].user_ptr, size)) { + return -EFAULT; + } + } else { + p->chunks[i].kpage[0] = kmalloc(PAGE_SIZE, GFP_KERNEL); + p->chunks[i].kpage[1] = kmalloc(PAGE_SIZE, GFP_KERNEL); + if (p->chunks[i].kpage[0] == NULL || p->chunks[i].kpage[1] == NULL) { + kfree(p->chunks[i].kpage[0]); + kfree(p->chunks[i].kpage[1]); + return -ENOMEM; + } + p->chunks[i].kpage_idx[0] = -1; + p->chunks[i].kpage_idx[1] = -1; + p->chunks[i].last_copied_page = -1; + p->chunks[i].last_page_index = ((p->chunks[i].length_dw * 4) - 1) / PAGE_SIZE; + } + } + if (p->chunks[p->chunk_ib_idx].length_dw > (16 * 1024)) { + DRM_ERROR("cs IB too big: %d\n", + p->chunks[p->chunk_ib_idx].length_dw); + return -EINVAL; + } + return 0; +} + +/** + * cs_parser_fini() - clean parser states + * @parser: parser structure holding parsing context. + * @error: error number + * + * If error is set than unvalidate buffer, otherwise just free memory + * used by parsing context. + **/ +static void radeon_cs_parser_fini(struct radeon_cs_parser *parser, int error) +{ + unsigned i; + + + if (!error && parser->ib) + ttm_eu_fence_buffer_objects(&parser->validated, + parser->ib->fence); + else + ttm_eu_backoff_reservation(&parser->validated); + + if (parser->relocs != NULL) { + for (i = 0; i < parser->nrelocs; i++) { + if (parser->relocs[i].gobj) + drm_gem_object_unreference_unlocked(parser->relocs[i].gobj); + } + } + kfree(parser->track); + kfree(parser->relocs); + kfree(parser->relocs_ptr); + for (i = 0; i < parser->nchunks; i++) { + kfree(parser->chunks[i].kdata); + kfree(parser->chunks[i].kpage[0]); + kfree(parser->chunks[i].kpage[1]); + } + kfree(parser->chunks); + kfree(parser->chunks_array); + radeon_ib_free(parser->rdev, &parser->ib); +} + +int radeon_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp) +{ + struct radeon_device *rdev = dev->dev_private; + struct radeon_cs_parser parser; + struct radeon_cs_chunk *ib_chunk; + int r; + + mutex_lock(&rdev->cs_mutex); + /* initialize parser */ + memset(&parser, 0, sizeof(struct radeon_cs_parser)); + parser.filp = filp; + parser.rdev = rdev; + parser.dev = rdev->dev; + parser.family = rdev->family; + r = radeon_cs_parser_init(&parser, data); + if (r) { + DRM_ERROR("Failed to initialize parser !\n"); + radeon_cs_parser_fini(&parser, r); + mutex_unlock(&rdev->cs_mutex); + return r; + } + r = radeon_ib_get(rdev, &parser.ib); + if (r) { + DRM_ERROR("Failed to get ib !\n"); + radeon_cs_parser_fini(&parser, r); + mutex_unlock(&rdev->cs_mutex); + return r; + } + r = radeon_cs_parser_relocs(&parser); + if (r) { + if (r != -ERESTARTSYS) + DRM_ERROR("Failed to parse relocation %d!\n", r); + radeon_cs_parser_fini(&parser, r); + mutex_unlock(&rdev->cs_mutex); + return r; + } + /* Copy the packet into the IB, the parser will read from the + * input memory (cached) and write to the IB (which can be + * uncached). */ + ib_chunk = &parser.chunks[parser.chunk_ib_idx]; + parser.ib->length_dw = ib_chunk->length_dw; + r = radeon_cs_parse(&parser); + if (r || parser.parser_error) { + DRM_ERROR("Invalid command stream !\n"); + radeon_cs_parser_fini(&parser, r); + mutex_unlock(&rdev->cs_mutex); + return r; + } + r = radeon_cs_finish_pages(&parser); + if (r) { + DRM_ERROR("Invalid command stream !\n"); + radeon_cs_parser_fini(&parser, r); + mutex_unlock(&rdev->cs_mutex); + return r; + } + r = radeon_ib_schedule(rdev, parser.ib); + if (r) { + DRM_ERROR("Failed to schedule IB !\n"); + } + radeon_cs_parser_fini(&parser, r); + mutex_unlock(&rdev->cs_mutex); + return r; +} + +int radeon_cs_finish_pages(struct radeon_cs_parser *p) +{ + struct radeon_cs_chunk *ibc = &p->chunks[p->chunk_ib_idx]; + int i; + int size = PAGE_SIZE; + + for (i = ibc->last_copied_page + 1; i <= ibc->last_page_index; i++) { + if (i == ibc->last_page_index) { + size = (ibc->length_dw * 4) % PAGE_SIZE; + if (size == 0) + size = PAGE_SIZE; + } + + if (DRM_COPY_FROM_USER(p->ib->ptr + (i * (PAGE_SIZE/4)), + ibc->user_ptr + (i * PAGE_SIZE), + size)) + return -EFAULT; + } + return 0; +} + +int radeon_cs_update_pages(struct radeon_cs_parser *p, int pg_idx) +{ + int new_page; + struct radeon_cs_chunk *ibc = &p->chunks[p->chunk_ib_idx]; + int i; + int size = PAGE_SIZE; + + for (i = ibc->last_copied_page + 1; i < pg_idx; i++) { + if (DRM_COPY_FROM_USER(p->ib->ptr + (i * (PAGE_SIZE/4)), + ibc->user_ptr + (i * PAGE_SIZE), + PAGE_SIZE)) { + p->parser_error = -EFAULT; + return 0; + } + } + + new_page = ibc->kpage_idx[0] < ibc->kpage_idx[1] ? 0 : 1; + + if (pg_idx == ibc->last_page_index) { + size = (ibc->length_dw * 4) % PAGE_SIZE; + if (size == 0) + size = PAGE_SIZE; + } + + if (DRM_COPY_FROM_USER(ibc->kpage[new_page], + ibc->user_ptr + (pg_idx * PAGE_SIZE), + size)) { + p->parser_error = -EFAULT; + return 0; + } + + /* copy to IB here */ + memcpy((void *)(p->ib->ptr+(pg_idx*(PAGE_SIZE/4))), ibc->kpage[new_page], size); + + ibc->last_copied_page = pg_idx; + ibc->kpage_idx[new_page] = pg_idx; + + return new_page; +} diff --git a/drivers/gpu/drm/radeon/radeon_cursor.c b/drivers/gpu/drm/radeon/radeon_cursor.c new file mode 100644 index 00000000..3fb22261 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_cursor.c @@ -0,0 +1,299 @@ +/* + * Copyright 2007-8 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + */ +#include "drmP.h" +#include "radeon_drm.h" +#include "radeon.h" + +#define CURSOR_WIDTH 64 +#define CURSOR_HEIGHT 64 + +static void radeon_lock_cursor(struct drm_crtc *crtc, bool lock) +{ + struct radeon_device *rdev = crtc->dev->dev_private; + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + uint32_t cur_lock; + + if (ASIC_IS_DCE4(rdev)) { + cur_lock = RREG32(EVERGREEN_CUR_UPDATE + radeon_crtc->crtc_offset); + if (lock) + cur_lock |= EVERGREEN_CURSOR_UPDATE_LOCK; + else + cur_lock &= ~EVERGREEN_CURSOR_UPDATE_LOCK; + WREG32(EVERGREEN_CUR_UPDATE + radeon_crtc->crtc_offset, cur_lock); + } else if (ASIC_IS_AVIVO(rdev)) { + cur_lock = RREG32(AVIVO_D1CUR_UPDATE + radeon_crtc->crtc_offset); + if (lock) + cur_lock |= AVIVO_D1CURSOR_UPDATE_LOCK; + else + cur_lock &= ~AVIVO_D1CURSOR_UPDATE_LOCK; + WREG32(AVIVO_D1CUR_UPDATE + radeon_crtc->crtc_offset, cur_lock); + } else { + cur_lock = RREG32(RADEON_CUR_OFFSET + radeon_crtc->crtc_offset); + if (lock) + cur_lock |= RADEON_CUR_LOCK; + else + cur_lock &= ~RADEON_CUR_LOCK; + WREG32(RADEON_CUR_OFFSET + radeon_crtc->crtc_offset, cur_lock); + } +} + +static void radeon_hide_cursor(struct drm_crtc *crtc) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct radeon_device *rdev = crtc->dev->dev_private; + + if (ASIC_IS_DCE4(rdev)) { + WREG32(RADEON_MM_INDEX, EVERGREEN_CUR_CONTROL + radeon_crtc->crtc_offset); + WREG32(RADEON_MM_DATA, EVERGREEN_CURSOR_MODE(EVERGREEN_CURSOR_24_8_PRE_MULT)); + } else if (ASIC_IS_AVIVO(rdev)) { + WREG32(RADEON_MM_INDEX, AVIVO_D1CUR_CONTROL + radeon_crtc->crtc_offset); + WREG32(RADEON_MM_DATA, (AVIVO_D1CURSOR_MODE_24BPP << AVIVO_D1CURSOR_MODE_SHIFT)); + } else { + switch (radeon_crtc->crtc_id) { + case 0: + WREG32(RADEON_MM_INDEX, RADEON_CRTC_GEN_CNTL); + break; + case 1: + WREG32(RADEON_MM_INDEX, RADEON_CRTC2_GEN_CNTL); + break; + default: + return; + } + WREG32_P(RADEON_MM_DATA, 0, ~RADEON_CRTC_CUR_EN); + } +} + +static void radeon_show_cursor(struct drm_crtc *crtc) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct radeon_device *rdev = crtc->dev->dev_private; + + if (ASIC_IS_DCE4(rdev)) { + WREG32(RADEON_MM_INDEX, EVERGREEN_CUR_CONTROL + radeon_crtc->crtc_offset); + WREG32(RADEON_MM_DATA, EVERGREEN_CURSOR_EN | + EVERGREEN_CURSOR_MODE(EVERGREEN_CURSOR_24_8_PRE_MULT)); + } else if (ASIC_IS_AVIVO(rdev)) { + WREG32(RADEON_MM_INDEX, AVIVO_D1CUR_CONTROL + radeon_crtc->crtc_offset); + WREG32(RADEON_MM_DATA, AVIVO_D1CURSOR_EN | + (AVIVO_D1CURSOR_MODE_24BPP << AVIVO_D1CURSOR_MODE_SHIFT)); + } else { + switch (radeon_crtc->crtc_id) { + case 0: + WREG32(RADEON_MM_INDEX, RADEON_CRTC_GEN_CNTL); + break; + case 1: + WREG32(RADEON_MM_INDEX, RADEON_CRTC2_GEN_CNTL); + break; + default: + return; + } + + WREG32_P(RADEON_MM_DATA, (RADEON_CRTC_CUR_EN | + (RADEON_CRTC_CUR_MODE_24BPP << RADEON_CRTC_CUR_MODE_SHIFT)), + ~(RADEON_CRTC_CUR_EN | RADEON_CRTC_CUR_MODE_MASK)); + } +} + +static void radeon_set_cursor(struct drm_crtc *crtc, struct drm_gem_object *obj, + uint64_t gpu_addr) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct radeon_device *rdev = crtc->dev->dev_private; + + if (ASIC_IS_DCE4(rdev)) { + WREG32(EVERGREEN_CUR_SURFACE_ADDRESS_HIGH + radeon_crtc->crtc_offset, + upper_32_bits(gpu_addr)); + WREG32(EVERGREEN_CUR_SURFACE_ADDRESS + radeon_crtc->crtc_offset, + gpu_addr & 0xffffffff); + } else if (ASIC_IS_AVIVO(rdev)) { + if (rdev->family >= CHIP_RV770) { + if (radeon_crtc->crtc_id) + WREG32(R700_D2CUR_SURFACE_ADDRESS_HIGH, upper_32_bits(gpu_addr)); + else + WREG32(R700_D1CUR_SURFACE_ADDRESS_HIGH, upper_32_bits(gpu_addr)); + } + WREG32(AVIVO_D1CUR_SURFACE_ADDRESS + radeon_crtc->crtc_offset, + gpu_addr & 0xffffffff); + } else { + radeon_crtc->legacy_cursor_offset = gpu_addr - radeon_crtc->legacy_display_base_addr; + /* offset is from DISP(2)_BASE_ADDRESS */ + WREG32(RADEON_CUR_OFFSET + radeon_crtc->crtc_offset, radeon_crtc->legacy_cursor_offset); + } +} + +int radeon_crtc_cursor_set(struct drm_crtc *crtc, + struct drm_file *file_priv, + uint32_t handle, + uint32_t width, + uint32_t height) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct radeon_device *rdev = crtc->dev->dev_private; + struct drm_gem_object *obj; + struct radeon_bo *robj; + uint64_t gpu_addr; + int ret; + + if (!handle) { + /* turn off cursor */ + radeon_hide_cursor(crtc); + obj = NULL; + goto unpin; + } + + if ((width > CURSOR_WIDTH) || (height > CURSOR_HEIGHT)) { + DRM_ERROR("bad cursor width or height %d x %d\n", width, height); + return -EINVAL; + } + + obj = drm_gem_object_lookup(crtc->dev, file_priv, handle); + if (!obj) { + DRM_ERROR("Cannot find cursor object %x for crtc %d\n", handle, radeon_crtc->crtc_id); + return -ENOENT; + } + + robj = gem_to_radeon_bo(obj); + ret = radeon_bo_reserve(robj, false); + if (unlikely(ret != 0)) + goto fail; + /* Only 27 bit offset for legacy cursor */ + ret = radeon_bo_pin_restricted(robj, RADEON_GEM_DOMAIN_VRAM, + ASIC_IS_AVIVO(rdev) ? 0 : 1 << 27, + &gpu_addr); + radeon_bo_unreserve(robj); + if (ret) + goto fail; + + radeon_crtc->cursor_width = width; + radeon_crtc->cursor_height = height; + + radeon_lock_cursor(crtc, true); + radeon_set_cursor(crtc, obj, gpu_addr); + radeon_show_cursor(crtc); + radeon_lock_cursor(crtc, false); + +unpin: + if (radeon_crtc->cursor_bo) { + radeon_gem_object_unpin(radeon_crtc->cursor_bo); + drm_gem_object_unreference_unlocked(radeon_crtc->cursor_bo); + } + + radeon_crtc->cursor_bo = obj; + return 0; +fail: + drm_gem_object_unreference_unlocked(obj); + + return ret; +} + +int radeon_crtc_cursor_move(struct drm_crtc *crtc, + int x, int y) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct radeon_device *rdev = crtc->dev->dev_private; + int xorigin = 0, yorigin = 0; + int w = radeon_crtc->cursor_width; + + if (ASIC_IS_AVIVO(rdev)) { + /* avivo cursor are offset into the total surface */ + x += crtc->x; + y += crtc->y; + } + DRM_DEBUG("x %d y %d c->x %d c->y %d\n", x, y, crtc->x, crtc->y); + + if (x < 0) + xorigin = -x + 1; + if (y < 0) + yorigin = -y + 1; + if (xorigin >= CURSOR_WIDTH) + xorigin = CURSOR_WIDTH - 1; + if (yorigin >= CURSOR_HEIGHT) + yorigin = CURSOR_HEIGHT - 1; + + if (ASIC_IS_AVIVO(rdev)) { + int i = 0; + struct drm_crtc *crtc_p; + + /* avivo cursor image can't end on 128 pixel boundary or + * go past the end of the frame if both crtcs are enabled + */ + list_for_each_entry(crtc_p, &crtc->dev->mode_config.crtc_list, head) { + if (crtc_p->enabled) + i++; + } + if (i > 1) { + int cursor_end, frame_end; + + cursor_end = x - xorigin + w; + frame_end = crtc->x + crtc->mode.crtc_hdisplay; + if (cursor_end >= frame_end) { + w = w - (cursor_end - frame_end); + if (!(frame_end & 0x7f)) + w--; + } else { + if (!(cursor_end & 0x7f)) + w--; + } + if (w <= 0) + w = 1; + } + } + + radeon_lock_cursor(crtc, true); + if (ASIC_IS_DCE4(rdev)) { + WREG32(EVERGREEN_CUR_POSITION + radeon_crtc->crtc_offset, + ((xorigin ? 0 : x) << 16) | + (yorigin ? 0 : y)); + WREG32(EVERGREEN_CUR_HOT_SPOT + radeon_crtc->crtc_offset, (xorigin << 16) | yorigin); + WREG32(EVERGREEN_CUR_SIZE + radeon_crtc->crtc_offset, + ((w - 1) << 16) | (radeon_crtc->cursor_height - 1)); + } else if (ASIC_IS_AVIVO(rdev)) { + WREG32(AVIVO_D1CUR_POSITION + radeon_crtc->crtc_offset, + ((xorigin ? 0 : x) << 16) | + (yorigin ? 0 : y)); + WREG32(AVIVO_D1CUR_HOT_SPOT + radeon_crtc->crtc_offset, (xorigin << 16) | yorigin); + WREG32(AVIVO_D1CUR_SIZE + radeon_crtc->crtc_offset, + ((w - 1) << 16) | (radeon_crtc->cursor_height - 1)); + } else { + if (crtc->mode.flags & DRM_MODE_FLAG_DBLSCAN) + y *= 2; + + WREG32(RADEON_CUR_HORZ_VERT_OFF + radeon_crtc->crtc_offset, + (RADEON_CUR_LOCK + | (xorigin << 16) + | yorigin)); + WREG32(RADEON_CUR_HORZ_VERT_POSN + radeon_crtc->crtc_offset, + (RADEON_CUR_LOCK + | ((xorigin ? 0 : x) << 16) + | (yorigin ? 0 : y))); + /* offset is from DISP(2)_BASE_ADDRESS */ + WREG32(RADEON_CUR_OFFSET + radeon_crtc->crtc_offset, (radeon_crtc->legacy_cursor_offset + + (yorigin * 256))); + } + radeon_lock_cursor(crtc, false); + + return 0; +} diff --git a/drivers/gpu/drm/radeon/radeon_device.c b/drivers/gpu/drm/radeon/radeon_device.c new file mode 100644 index 00000000..e87893c2 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_device.c @@ -0,0 +1,1034 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#include +#include +#include +#include +#include +#include +#include +#include "radeon_reg.h" +#include "radeon.h" +#include "atom.h" + +static const char radeon_family_name[][16] = { + "R100", + "RV100", + "RS100", + "RV200", + "RS200", + "R200", + "RV250", + "RS300", + "RV280", + "R300", + "R350", + "RV350", + "RV380", + "R420", + "R423", + "RV410", + "RS400", + "RS480", + "RS600", + "RS690", + "RS740", + "RV515", + "R520", + "RV530", + "RV560", + "RV570", + "R580", + "R600", + "RV610", + "RV630", + "RV670", + "RV620", + "RV635", + "RS780", + "RS880", + "RV770", + "RV730", + "RV710", + "RV740", + "CEDAR", + "REDWOOD", + "JUNIPER", + "CYPRESS", + "HEMLOCK", + "PALM", + "SUMO", + "SUMO2", + "BARTS", + "TURKS", + "CAICOS", + "CAYMAN", + "LAST", +}; + +/* + * Clear GPU surface registers. + */ +void radeon_surface_init(struct radeon_device *rdev) +{ + /* FIXME: check this out */ + if (rdev->family < CHIP_R600) { + int i; + + for (i = 0; i < RADEON_GEM_MAX_SURFACES; i++) { + if (rdev->surface_regs[i].bo) + radeon_bo_get_surface_reg(rdev->surface_regs[i].bo); + else + radeon_clear_surface_reg(rdev, i); + } + /* enable surfaces */ + WREG32(RADEON_SURFACE_CNTL, 0); + } +} + +/* + * GPU scratch registers helpers function. + */ +void radeon_scratch_init(struct radeon_device *rdev) +{ + int i; + + /* FIXME: check this out */ + if (rdev->family < CHIP_R300) { + rdev->scratch.num_reg = 5; + } else { + rdev->scratch.num_reg = 7; + } + rdev->scratch.reg_base = RADEON_SCRATCH_REG0; + for (i = 0; i < rdev->scratch.num_reg; i++) { + rdev->scratch.free[i] = true; + rdev->scratch.reg[i] = rdev->scratch.reg_base + (i * 4); + } +} + +int radeon_scratch_get(struct radeon_device *rdev, uint32_t *reg) +{ + int i; + + for (i = 0; i < rdev->scratch.num_reg; i++) { + if (rdev->scratch.free[i]) { + rdev->scratch.free[i] = false; + *reg = rdev->scratch.reg[i]; + return 0; + } + } + return -EINVAL; +} + +void radeon_scratch_free(struct radeon_device *rdev, uint32_t reg) +{ + int i; + + for (i = 0; i < rdev->scratch.num_reg; i++) { + if (rdev->scratch.reg[i] == reg) { + rdev->scratch.free[i] = true; + return; + } + } +} + +void radeon_wb_disable(struct radeon_device *rdev) +{ + int r; + + if (rdev->wb.wb_obj) { + r = radeon_bo_reserve(rdev->wb.wb_obj, false); + if (unlikely(r != 0)) + return; + radeon_bo_kunmap(rdev->wb.wb_obj); + radeon_bo_unpin(rdev->wb.wb_obj); + radeon_bo_unreserve(rdev->wb.wb_obj); + } + rdev->wb.enabled = false; +} + +void radeon_wb_fini(struct radeon_device *rdev) +{ + radeon_wb_disable(rdev); + if (rdev->wb.wb_obj) { + radeon_bo_unref(&rdev->wb.wb_obj); + rdev->wb.wb = NULL; + rdev->wb.wb_obj = NULL; + } +} + +int radeon_wb_init(struct radeon_device *rdev) +{ + int r; + + if (rdev->wb.wb_obj == NULL) { + r = radeon_bo_create(rdev, RADEON_GPU_PAGE_SIZE, PAGE_SIZE, true, + RADEON_GEM_DOMAIN_GTT, &rdev->wb.wb_obj); + if (r) { + dev_warn(rdev->dev, "(%d) create WB bo failed\n", r); + return r; + } + } + r = radeon_bo_reserve(rdev->wb.wb_obj, false); + if (unlikely(r != 0)) { + radeon_wb_fini(rdev); + return r; + } + r = radeon_bo_pin(rdev->wb.wb_obj, RADEON_GEM_DOMAIN_GTT, + &rdev->wb.gpu_addr); + if (r) { + radeon_bo_unreserve(rdev->wb.wb_obj); + dev_warn(rdev->dev, "(%d) pin WB bo failed\n", r); + radeon_wb_fini(rdev); + return r; + } + r = radeon_bo_kmap(rdev->wb.wb_obj, (void **)&rdev->wb.wb); + radeon_bo_unreserve(rdev->wb.wb_obj); + if (r) { + dev_warn(rdev->dev, "(%d) map WB bo failed\n", r); + radeon_wb_fini(rdev); + return r; + } + + /* clear wb memory */ + memset((char *)rdev->wb.wb, 0, RADEON_GPU_PAGE_SIZE); + /* disable event_write fences */ + rdev->wb.use_event = false; + /* disabled via module param */ + if (radeon_no_wb == 1) + rdev->wb.enabled = false; + else { + if (rdev->flags & RADEON_IS_AGP) { + /* often unreliable on AGP */ + rdev->wb.enabled = false; + } else if (rdev->family < CHIP_R300) { + /* often unreliable on pre-r300 */ + rdev->wb.enabled = false; + } else { + rdev->wb.enabled = true; + /* event_write fences are only available on r600+ */ + if (rdev->family >= CHIP_R600) + rdev->wb.use_event = true; + } + } + /* always use writeback/events on NI */ + if (ASIC_IS_DCE5(rdev)) { + rdev->wb.enabled = true; + rdev->wb.use_event = true; + } + + dev_info(rdev->dev, "WB %sabled\n", rdev->wb.enabled ? "en" : "dis"); + + return 0; +} + +/** + * radeon_vram_location - try to find VRAM location + * @rdev: radeon device structure holding all necessary informations + * @mc: memory controller structure holding memory informations + * @base: base address at which to put VRAM + * + * Function will place try to place VRAM at base address provided + * as parameter (which is so far either PCI aperture address or + * for IGP TOM base address). + * + * If there is not enough space to fit the unvisible VRAM in the 32bits + * address space then we limit the VRAM size to the aperture. + * + * If we are using AGP and if the AGP aperture doesn't allow us to have + * room for all the VRAM than we restrict the VRAM to the PCI aperture + * size and print a warning. + * + * This function will never fails, worst case are limiting VRAM. + * + * Note: GTT start, end, size should be initialized before calling this + * function on AGP platform. + * + * Note: We don't explicitly enforce VRAM start to be aligned on VRAM size, + * this shouldn't be a problem as we are using the PCI aperture as a reference. + * Otherwise this would be needed for rv280, all r3xx, and all r4xx, but + * not IGP. + * + * Note: we use mc_vram_size as on some board we need to program the mc to + * cover the whole aperture even if VRAM size is inferior to aperture size + * Novell bug 204882 + along with lots of ubuntu ones + * + * Note: when limiting vram it's safe to overwritte real_vram_size because + * we are not in case where real_vram_size is inferior to mc_vram_size (ie + * note afected by bogus hw of Novell bug 204882 + along with lots of ubuntu + * ones) + * + * Note: IGP TOM addr should be the same as the aperture addr, we don't + * explicitly check for that thought. + * + * FIXME: when reducing VRAM size align new size on power of 2. + */ +void radeon_vram_location(struct radeon_device *rdev, struct radeon_mc *mc, u64 base) +{ + mc->vram_start = base; + if (mc->mc_vram_size > (0xFFFFFFFF - base + 1)) { + dev_warn(rdev->dev, "limiting VRAM to PCI aperture size\n"); + mc->real_vram_size = mc->aper_size; + mc->mc_vram_size = mc->aper_size; + } + mc->vram_end = mc->vram_start + mc->mc_vram_size - 1; + if (rdev->flags & RADEON_IS_AGP && mc->vram_end > mc->gtt_start && mc->vram_start <= mc->gtt_end) { + dev_warn(rdev->dev, "limiting VRAM to PCI aperture size\n"); + mc->real_vram_size = mc->aper_size; + mc->mc_vram_size = mc->aper_size; + } + mc->vram_end = mc->vram_start + mc->mc_vram_size - 1; + dev_info(rdev->dev, "VRAM: %lluM 0x%016llX - 0x%016llX (%lluM used)\n", + mc->mc_vram_size >> 20, mc->vram_start, + mc->vram_end, mc->real_vram_size >> 20); +} + +/** + * radeon_gtt_location - try to find GTT location + * @rdev: radeon device structure holding all necessary informations + * @mc: memory controller structure holding memory informations + * + * Function will place try to place GTT before or after VRAM. + * + * If GTT size is bigger than space left then we ajust GTT size. + * Thus function will never fails. + * + * FIXME: when reducing GTT size align new size on power of 2. + */ +void radeon_gtt_location(struct radeon_device *rdev, struct radeon_mc *mc) +{ + u64 size_af, size_bf; + + size_af = ((0xFFFFFFFF - mc->vram_end) + mc->gtt_base_align) & ~mc->gtt_base_align; + size_bf = mc->vram_start & ~mc->gtt_base_align; + if (size_bf > size_af) { + if (mc->gtt_size > size_bf) { + dev_warn(rdev->dev, "limiting GTT\n"); + mc->gtt_size = size_bf; + } + mc->gtt_start = (mc->vram_start & ~mc->gtt_base_align) - mc->gtt_size; + } else { + if (mc->gtt_size > size_af) { + dev_warn(rdev->dev, "limiting GTT\n"); + mc->gtt_size = size_af; + } + mc->gtt_start = (mc->vram_end + 1 + mc->gtt_base_align) & ~mc->gtt_base_align; + } + mc->gtt_end = mc->gtt_start + mc->gtt_size - 1; + dev_info(rdev->dev, "GTT: %lluM 0x%016llX - 0x%016llX\n", + mc->gtt_size >> 20, mc->gtt_start, mc->gtt_end); +} + +/* + * GPU helpers function. + */ +bool radeon_card_posted(struct radeon_device *rdev) +{ + uint32_t reg; + + /* first check CRTCs */ + if (ASIC_IS_DCE41(rdev)) { + reg = RREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC0_REGISTER_OFFSET) | + RREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC1_REGISTER_OFFSET); + if (reg & EVERGREEN_CRTC_MASTER_EN) + return true; + } else if (ASIC_IS_DCE4(rdev)) { + reg = RREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC0_REGISTER_OFFSET) | + RREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC1_REGISTER_OFFSET) | + RREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC2_REGISTER_OFFSET) | + RREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC3_REGISTER_OFFSET) | + RREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC4_REGISTER_OFFSET) | + RREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC5_REGISTER_OFFSET); + if (reg & EVERGREEN_CRTC_MASTER_EN) + return true; + } else if (ASIC_IS_AVIVO(rdev)) { + reg = RREG32(AVIVO_D1CRTC_CONTROL) | + RREG32(AVIVO_D2CRTC_CONTROL); + if (reg & AVIVO_CRTC_EN) { + return true; + } + } else { + reg = RREG32(RADEON_CRTC_GEN_CNTL) | + RREG32(RADEON_CRTC2_GEN_CNTL); + if (reg & RADEON_CRTC_EN) { + return true; + } + } + + /* then check MEM_SIZE, in case the crtcs are off */ + if (rdev->family >= CHIP_R600) + reg = RREG32(R600_CONFIG_MEMSIZE); + else + reg = RREG32(RADEON_CONFIG_MEMSIZE); + + if (reg) + return true; + + return false; + +} + +void radeon_update_bandwidth_info(struct radeon_device *rdev) +{ + fixed20_12 a; + u32 sclk = rdev->pm.current_sclk; + u32 mclk = rdev->pm.current_mclk; + + /* sclk/mclk in Mhz */ + a.full = dfixed_const(100); + rdev->pm.sclk.full = dfixed_const(sclk); + rdev->pm.sclk.full = dfixed_div(rdev->pm.sclk, a); + rdev->pm.mclk.full = dfixed_const(mclk); + rdev->pm.mclk.full = dfixed_div(rdev->pm.mclk, a); + + if (rdev->flags & RADEON_IS_IGP) { + a.full = dfixed_const(16); + /* core_bandwidth = sclk(Mhz) * 16 */ + rdev->pm.core_bandwidth.full = dfixed_div(rdev->pm.sclk, a); + } +} + +bool radeon_boot_test_post_card(struct radeon_device *rdev) +{ + if (radeon_card_posted(rdev)) + return true; + + if (rdev->bios) { + DRM_INFO("GPU not posted. posting now...\n"); + if (rdev->is_atom_bios) + atom_asic_init(rdev->mode_info.atom_context); + else + radeon_combios_asic_init(rdev->ddev); + return true; + } else { + dev_err(rdev->dev, "Card not posted and no BIOS - ignoring\n"); + return false; + } +} + +int radeon_dummy_page_init(struct radeon_device *rdev) +{ + if (rdev->dummy_page.page) + return 0; + rdev->dummy_page.page = alloc_page(GFP_DMA32 | GFP_KERNEL | __GFP_ZERO); + if (rdev->dummy_page.page == NULL) + return -ENOMEM; + rdev->dummy_page.addr = pci_map_page(rdev->pdev, rdev->dummy_page.page, + 0, PAGE_SIZE, PCI_DMA_BIDIRECTIONAL); + if (pci_dma_mapping_error(rdev->pdev, rdev->dummy_page.addr)) { + dev_err(&rdev->pdev->dev, "Failed to DMA MAP the dummy page\n"); + __free_page(rdev->dummy_page.page); + rdev->dummy_page.page = NULL; + return -ENOMEM; + } + return 0; +} + +void radeon_dummy_page_fini(struct radeon_device *rdev) +{ + if (rdev->dummy_page.page == NULL) + return; + pci_unmap_page(rdev->pdev, rdev->dummy_page.addr, + PAGE_SIZE, PCI_DMA_BIDIRECTIONAL); + __free_page(rdev->dummy_page.page); + rdev->dummy_page.page = NULL; +} + + +/* ATOM accessor methods */ +static uint32_t cail_pll_read(struct card_info *info, uint32_t reg) +{ + struct radeon_device *rdev = info->dev->dev_private; + uint32_t r; + + r = rdev->pll_rreg(rdev, reg); + return r; +} + +static void cail_pll_write(struct card_info *info, uint32_t reg, uint32_t val) +{ + struct radeon_device *rdev = info->dev->dev_private; + + rdev->pll_wreg(rdev, reg, val); +} + +static uint32_t cail_mc_read(struct card_info *info, uint32_t reg) +{ + struct radeon_device *rdev = info->dev->dev_private; + uint32_t r; + + r = rdev->mc_rreg(rdev, reg); + return r; +} + +static void cail_mc_write(struct card_info *info, uint32_t reg, uint32_t val) +{ + struct radeon_device *rdev = info->dev->dev_private; + + rdev->mc_wreg(rdev, reg, val); +} + +static void cail_reg_write(struct card_info *info, uint32_t reg, uint32_t val) +{ + struct radeon_device *rdev = info->dev->dev_private; + + WREG32(reg*4, val); +} + +static uint32_t cail_reg_read(struct card_info *info, uint32_t reg) +{ + struct radeon_device *rdev = info->dev->dev_private; + uint32_t r; + + r = RREG32(reg*4); + return r; +} + +static void cail_ioreg_write(struct card_info *info, uint32_t reg, uint32_t val) +{ + struct radeon_device *rdev = info->dev->dev_private; + + WREG32_IO(reg*4, val); +} + +static uint32_t cail_ioreg_read(struct card_info *info, uint32_t reg) +{ + struct radeon_device *rdev = info->dev->dev_private; + uint32_t r; + + r = RREG32_IO(reg*4); + return r; +} + +int radeon_atombios_init(struct radeon_device *rdev) +{ + struct card_info *atom_card_info = + kzalloc(sizeof(struct card_info), GFP_KERNEL); + + if (!atom_card_info) + return -ENOMEM; + + rdev->mode_info.atom_card_info = atom_card_info; + atom_card_info->dev = rdev->ddev; + atom_card_info->reg_read = cail_reg_read; + atom_card_info->reg_write = cail_reg_write; + /* needed for iio ops */ + if (rdev->rio_mem) { + atom_card_info->ioreg_read = cail_ioreg_read; + atom_card_info->ioreg_write = cail_ioreg_write; + } else { + DRM_ERROR("Unable to find PCI I/O BAR; using MMIO for ATOM IIO\n"); + atom_card_info->ioreg_read = cail_reg_read; + atom_card_info->ioreg_write = cail_reg_write; + } + atom_card_info->mc_read = cail_mc_read; + atom_card_info->mc_write = cail_mc_write; + atom_card_info->pll_read = cail_pll_read; + atom_card_info->pll_write = cail_pll_write; + + rdev->mode_info.atom_context = atom_parse(atom_card_info, rdev->bios); + mutex_init(&rdev->mode_info.atom_context->mutex); + radeon_atom_initialize_bios_scratch_regs(rdev->ddev); + atom_allocate_fb_scratch(rdev->mode_info.atom_context); + return 0; +} + +void radeon_atombios_fini(struct radeon_device *rdev) +{ + if (rdev->mode_info.atom_context) { + kfree(rdev->mode_info.atom_context->scratch); + kfree(rdev->mode_info.atom_context); + } + kfree(rdev->mode_info.atom_card_info); +} + +int radeon_combios_init(struct radeon_device *rdev) +{ + radeon_combios_initialize_bios_scratch_regs(rdev->ddev); + return 0; +} + +void radeon_combios_fini(struct radeon_device *rdev) +{ +} + +/* if we get transitioned to only one device, tak VGA back */ +static unsigned int radeon_vga_set_decode(void *cookie, bool state) +{ + struct radeon_device *rdev = cookie; + radeon_vga_set_state(rdev, state); + if (state) + return VGA_RSRC_LEGACY_IO | VGA_RSRC_LEGACY_MEM | + VGA_RSRC_NORMAL_IO | VGA_RSRC_NORMAL_MEM; + else + return VGA_RSRC_NORMAL_IO | VGA_RSRC_NORMAL_MEM; +} + +void radeon_check_arguments(struct radeon_device *rdev) +{ + /* vramlimit must be a power of two */ + switch (radeon_vram_limit) { + case 0: + case 4: + case 8: + case 16: + case 32: + case 64: + case 128: + case 256: + case 512: + case 1024: + case 2048: + case 4096: + break; + default: + dev_warn(rdev->dev, "vram limit (%d) must be a power of 2\n", + radeon_vram_limit); + radeon_vram_limit = 0; + break; + } + radeon_vram_limit = radeon_vram_limit << 20; + /* gtt size must be power of two and greater or equal to 32M */ + switch (radeon_gart_size) { + case 4: + case 8: + case 16: + dev_warn(rdev->dev, "gart size (%d) too small forcing to 512M\n", + radeon_gart_size); + radeon_gart_size = 512; + break; + case 32: + case 64: + case 128: + case 256: + case 512: + case 1024: + case 2048: + case 4096: + break; + default: + dev_warn(rdev->dev, "gart size (%d) must be a power of 2\n", + radeon_gart_size); + radeon_gart_size = 512; + break; + } + rdev->mc.gtt_size = radeon_gart_size * 1024 * 1024; + /* AGP mode can only be -1, 1, 2, 4, 8 */ + switch (radeon_agpmode) { + case -1: + case 0: + case 1: + case 2: + case 4: + case 8: + break; + default: + dev_warn(rdev->dev, "invalid AGP mode %d (valid mode: " + "-1, 0, 1, 2, 4, 8)\n", radeon_agpmode); + radeon_agpmode = 0; + break; + } +} + +static void radeon_switcheroo_set_state(struct pci_dev *pdev, enum vga_switcheroo_state state) +{ + struct drm_device *dev = pci_get_drvdata(pdev); + pm_message_t pmm = { .event = PM_EVENT_SUSPEND }; + if (state == VGA_SWITCHEROO_ON) { + printk(KERN_INFO "radeon: switched on\n"); + /* don't suspend or resume card normally */ + dev->switch_power_state = DRM_SWITCH_POWER_CHANGING; + radeon_resume_kms(dev); + dev->switch_power_state = DRM_SWITCH_POWER_ON; + drm_kms_helper_poll_enable(dev); + } else { + printk(KERN_INFO "radeon: switched off\n"); + drm_kms_helper_poll_disable(dev); + dev->switch_power_state = DRM_SWITCH_POWER_CHANGING; + radeon_suspend_kms(dev, pmm); + dev->switch_power_state = DRM_SWITCH_POWER_OFF; + } +} + +static bool radeon_switcheroo_can_switch(struct pci_dev *pdev) +{ + struct drm_device *dev = pci_get_drvdata(pdev); + bool can_switch; + + spin_lock(&dev->count_lock); + can_switch = (dev->open_count == 0); + spin_unlock(&dev->count_lock); + return can_switch; +} + + +int radeon_device_init(struct radeon_device *rdev, + struct drm_device *ddev, + struct pci_dev *pdev, + uint32_t flags) +{ + int r, i; + int dma_bits; + + rdev->shutdown = false; + rdev->dev = &pdev->dev; + rdev->ddev = ddev; + rdev->pdev = pdev; + rdev->flags = flags; + rdev->family = flags & RADEON_FAMILY_MASK; + rdev->is_atom_bios = false; + rdev->usec_timeout = RADEON_MAX_USEC_TIMEOUT; + rdev->mc.gtt_size = radeon_gart_size * 1024 * 1024; + rdev->gpu_lockup = false; + rdev->accel_working = false; + + DRM_INFO("initializing kernel modesetting (%s 0x%04X:0x%04X 0x%04X:0x%04X).\n", + radeon_family_name[rdev->family], pdev->vendor, pdev->device, + pdev->subsystem_vendor, pdev->subsystem_device); + + /* mutex initialization are all done here so we + * can recall function without having locking issues */ + mutex_init(&rdev->cs_mutex); + mutex_init(&rdev->ib_pool.mutex); + mutex_init(&rdev->cp.mutex); + mutex_init(&rdev->dc_hw_i2c_mutex); + if (rdev->family >= CHIP_R600) + spin_lock_init(&rdev->ih.lock); + mutex_init(&rdev->gem.mutex); + mutex_init(&rdev->pm.mutex); + mutex_init(&rdev->vram_mutex); + rwlock_init(&rdev->fence_drv.lock); + INIT_LIST_HEAD(&rdev->gem.objects); + init_waitqueue_head(&rdev->irq.vblank_queue); + init_waitqueue_head(&rdev->irq.idle_queue); + + /* Set asic functions */ + r = radeon_asic_init(rdev); + if (r) + return r; + radeon_check_arguments(rdev); + + /* all of the newer IGP chips have an internal gart + * However some rs4xx report as AGP, so remove that here. + */ + if ((rdev->family >= CHIP_RS400) && + (rdev->flags & RADEON_IS_IGP)) { + rdev->flags &= ~RADEON_IS_AGP; + } + + if (rdev->flags & RADEON_IS_AGP && radeon_agpmode == -1) { + radeon_agp_disable(rdev); + } + + /* set DMA mask + need_dma32 flags. + * PCIE - can handle 40-bits. + * IGP - can handle 40-bits (in theory) + * AGP - generally dma32 is safest + * PCI - only dma32 + */ + rdev->need_dma32 = false; + if (rdev->flags & RADEON_IS_AGP) + rdev->need_dma32 = true; + if (rdev->flags & RADEON_IS_PCI) + rdev->need_dma32 = true; + + dma_bits = rdev->need_dma32 ? 32 : 40; + r = pci_set_dma_mask(rdev->pdev, DMA_BIT_MASK(dma_bits)); + if (r) { + rdev->need_dma32 = true; + printk(KERN_WARNING "radeon: No suitable DMA available.\n"); + } + + /* Registers mapping */ + /* TODO: block userspace mapping of io register */ + rdev->rmmio_base = pci_resource_start(rdev->pdev, 2); + rdev->rmmio_size = pci_resource_len(rdev->pdev, 2); + rdev->rmmio = ioremap(rdev->rmmio_base, rdev->rmmio_size); + if (rdev->rmmio == NULL) { + return -ENOMEM; + } + DRM_INFO("register mmio base: 0x%08X\n", (uint32_t)rdev->rmmio_base); + DRM_INFO("register mmio size: %u\n", (unsigned)rdev->rmmio_size); + + /* io port mapping */ + for (i = 0; i < DEVICE_COUNT_RESOURCE; i++) { + if (pci_resource_flags(rdev->pdev, i) & IORESOURCE_IO) { + rdev->rio_mem_size = pci_resource_len(rdev->pdev, i); + rdev->rio_mem = pci_iomap(rdev->pdev, i, rdev->rio_mem_size); + break; + } + } + if (rdev->rio_mem == NULL) + DRM_ERROR("Unable to find PCI I/O BAR\n"); + + /* if we have > 1 VGA cards, then disable the radeon VGA resources */ + /* this will fail for cards that aren't VGA class devices, just + * ignore it */ + vga_client_register(rdev->pdev, rdev, NULL, radeon_vga_set_decode); + vga_switcheroo_register_client(rdev->pdev, + radeon_switcheroo_set_state, + NULL, + radeon_switcheroo_can_switch); + + r = radeon_init(rdev); + if (r) + return r; + + if (rdev->flags & RADEON_IS_AGP && !rdev->accel_working) { + /* Acceleration not working on AGP card try again + * with fallback to PCI or PCIE GART + */ + radeon_asic_reset(rdev); + radeon_fini(rdev); + radeon_agp_disable(rdev); + r = radeon_init(rdev); + if (r) + return r; + } + if (radeon_testing) { + radeon_test_moves(rdev); + } + if (radeon_benchmarking) { + radeon_benchmark(rdev); + } + return 0; +} + +void radeon_device_fini(struct radeon_device *rdev) +{ + DRM_INFO("radeon: finishing device.\n"); + rdev->shutdown = true; + /* evict vram memory */ + radeon_bo_evict_vram(rdev); + radeon_fini(rdev); + vga_switcheroo_unregister_client(rdev->pdev); + vga_client_register(rdev->pdev, NULL, NULL, NULL); + if (rdev->rio_mem) + pci_iounmap(rdev->pdev, rdev->rio_mem); + rdev->rio_mem = NULL; + iounmap(rdev->rmmio); + rdev->rmmio = NULL; +} + + +/* + * Suspend & resume. + */ +int radeon_suspend_kms(struct drm_device *dev, pm_message_t state) +{ + struct radeon_device *rdev; + struct drm_crtc *crtc; + struct drm_connector *connector; + int r; + + if (dev == NULL || dev->dev_private == NULL) { + return -ENODEV; + } + if (state.event == PM_EVENT_PRETHAW) { + return 0; + } + rdev = dev->dev_private; + + if (dev->switch_power_state == DRM_SWITCH_POWER_OFF) + return 0; + + drm_kms_helper_poll_disable(dev); + + /* turn off display hw */ + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + drm_helper_connector_dpms(connector, DRM_MODE_DPMS_OFF); + } + + /* unpin the front buffers */ + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { + struct radeon_framebuffer *rfb = to_radeon_framebuffer(crtc->fb); + struct radeon_bo *robj; + + if (rfb == NULL || rfb->obj == NULL) { + continue; + } + robj = gem_to_radeon_bo(rfb->obj); + /* don't unpin kernel fb objects */ + if (!radeon_fbdev_robj_is_fb(rdev, robj)) { + r = radeon_bo_reserve(robj, false); + if (r == 0) { + radeon_bo_unpin(robj); + radeon_bo_unreserve(robj); + } + } + } + /* evict vram memory */ + radeon_bo_evict_vram(rdev); + /* wait for gpu to finish processing current batch */ + radeon_fence_wait_last(rdev); + + radeon_save_bios_scratch_regs(rdev); + + radeon_pm_suspend(rdev); + radeon_suspend(rdev); + radeon_hpd_fini(rdev); + /* evict remaining vram memory */ + radeon_bo_evict_vram(rdev); + + radeon_agp_suspend(rdev); + + pci_save_state(dev->pdev); + if (state.event == PM_EVENT_SUSPEND) { + /* Shut down the device */ + pci_disable_device(dev->pdev); + pci_set_power_state(dev->pdev, PCI_D3hot); + } + console_lock(); + radeon_fbdev_set_suspend(rdev, 1); + console_unlock(); + return 0; +} + +int radeon_resume_kms(struct drm_device *dev) +{ + struct drm_connector *connector; + struct radeon_device *rdev = dev->dev_private; + + if (dev->switch_power_state == DRM_SWITCH_POWER_OFF) + return 0; + + console_lock(); + pci_set_power_state(dev->pdev, PCI_D0); + pci_restore_state(dev->pdev); + if (pci_enable_device(dev->pdev)) { + console_unlock(); + return -1; + } + pci_set_master(dev->pdev); + /* resume AGP if in use */ + radeon_agp_resume(rdev); + radeon_resume(rdev); + radeon_pm_resume(rdev); + radeon_restore_bios_scratch_regs(rdev); + + radeon_fbdev_set_suspend(rdev, 0); + console_unlock(); + + /* init dig PHYs */ + if (rdev->is_atom_bios) + radeon_atom_encoder_init(rdev); + /* reset hpd state */ + radeon_hpd_init(rdev); + /* blat the mode back in */ + drm_helper_resume_force_mode(dev); + /* turn on display hw */ + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + drm_helper_connector_dpms(connector, DRM_MODE_DPMS_ON); + } + + drm_kms_helper_poll_enable(dev); + return 0; +} + +int radeon_gpu_reset(struct radeon_device *rdev) +{ + int r; + int resched; + + radeon_save_bios_scratch_regs(rdev); + /* block TTM */ + resched = ttm_bo_lock_delayed_workqueue(&rdev->mman.bdev); + radeon_suspend(rdev); + + r = radeon_asic_reset(rdev); + if (!r) { + dev_info(rdev->dev, "GPU reset succeed\n"); + radeon_resume(rdev); + radeon_restore_bios_scratch_regs(rdev); + drm_helper_resume_force_mode(rdev->ddev); + ttm_bo_unlock_delayed_workqueue(&rdev->mman.bdev, resched); + return 0; + } + /* bad news, how to tell it to userspace ? */ + dev_info(rdev->dev, "GPU reset failed\n"); + return r; +} + + +/* + * Debugfs + */ +struct radeon_debugfs { + struct drm_info_list *files; + unsigned num_files; +}; +static struct radeon_debugfs _radeon_debugfs[RADEON_DEBUGFS_MAX_NUM_FILES]; +static unsigned _radeon_debugfs_count = 0; + +int radeon_debugfs_add_files(struct radeon_device *rdev, + struct drm_info_list *files, + unsigned nfiles) +{ + unsigned i; + + for (i = 0; i < _radeon_debugfs_count; i++) { + if (_radeon_debugfs[i].files == files) { + /* Already registered */ + return 0; + } + } + if ((_radeon_debugfs_count + nfiles) > RADEON_DEBUGFS_MAX_NUM_FILES) { + DRM_ERROR("Reached maximum number of debugfs files.\n"); + DRM_ERROR("Report so we increase RADEON_DEBUGFS_MAX_NUM_FILES.\n"); + return -EINVAL; + } + _radeon_debugfs[_radeon_debugfs_count].files = files; + _radeon_debugfs[_radeon_debugfs_count].num_files = nfiles; + _radeon_debugfs_count++; +#if defined(CONFIG_DEBUG_FS) + drm_debugfs_create_files(files, nfiles, + rdev->ddev->control->debugfs_root, + rdev->ddev->control); + drm_debugfs_create_files(files, nfiles, + rdev->ddev->primary->debugfs_root, + rdev->ddev->primary); +#endif + return 0; +} + +#if defined(CONFIG_DEBUG_FS) +int radeon_debugfs_init(struct drm_minor *minor) +{ + return 0; +} + +void radeon_debugfs_cleanup(struct drm_minor *minor) +{ + unsigned i; + + for (i = 0; i < _radeon_debugfs_count; i++) { + drm_debugfs_remove_files(_radeon_debugfs[i].files, + _radeon_debugfs[i].num_files, minor); + } +} +#endif diff --git a/drivers/gpu/drm/radeon/radeon_display.c b/drivers/gpu/drm/radeon/radeon_display.c new file mode 100644 index 00000000..ed085ce9 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_display.c @@ -0,0 +1,1651 @@ +/* + * Copyright 2007-8 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + */ +#include "drmP.h" +#include "radeon_drm.h" +#include "radeon.h" + +#include "atom.h" +#include + +#include "drm_crtc_helper.h" +#include "drm_edid.h" + +static int radeon_ddc_dump(struct drm_connector *connector); + +static void avivo_crtc_load_lut(struct drm_crtc *crtc) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + int i; + + DRM_DEBUG_KMS("%d\n", radeon_crtc->crtc_id); + WREG32(AVIVO_DC_LUTA_CONTROL + radeon_crtc->crtc_offset, 0); + + WREG32(AVIVO_DC_LUTA_BLACK_OFFSET_BLUE + radeon_crtc->crtc_offset, 0); + WREG32(AVIVO_DC_LUTA_BLACK_OFFSET_GREEN + radeon_crtc->crtc_offset, 0); + WREG32(AVIVO_DC_LUTA_BLACK_OFFSET_RED + radeon_crtc->crtc_offset, 0); + + WREG32(AVIVO_DC_LUTA_WHITE_OFFSET_BLUE + radeon_crtc->crtc_offset, 0xffff); + WREG32(AVIVO_DC_LUTA_WHITE_OFFSET_GREEN + radeon_crtc->crtc_offset, 0xffff); + WREG32(AVIVO_DC_LUTA_WHITE_OFFSET_RED + radeon_crtc->crtc_offset, 0xffff); + + WREG32(AVIVO_DC_LUT_RW_SELECT, radeon_crtc->crtc_id); + WREG32(AVIVO_DC_LUT_RW_MODE, 0); + WREG32(AVIVO_DC_LUT_WRITE_EN_MASK, 0x0000003f); + + WREG8(AVIVO_DC_LUT_RW_INDEX, 0); + for (i = 0; i < 256; i++) { + WREG32(AVIVO_DC_LUT_30_COLOR, + (radeon_crtc->lut_r[i] << 20) | + (radeon_crtc->lut_g[i] << 10) | + (radeon_crtc->lut_b[i] << 0)); + } + + WREG32(AVIVO_D1GRPH_LUT_SEL + radeon_crtc->crtc_offset, radeon_crtc->crtc_id); +} + +static void dce4_crtc_load_lut(struct drm_crtc *crtc) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + int i; + + DRM_DEBUG_KMS("%d\n", radeon_crtc->crtc_id); + WREG32(EVERGREEN_DC_LUT_CONTROL + radeon_crtc->crtc_offset, 0); + + WREG32(EVERGREEN_DC_LUT_BLACK_OFFSET_BLUE + radeon_crtc->crtc_offset, 0); + WREG32(EVERGREEN_DC_LUT_BLACK_OFFSET_GREEN + radeon_crtc->crtc_offset, 0); + WREG32(EVERGREEN_DC_LUT_BLACK_OFFSET_RED + radeon_crtc->crtc_offset, 0); + + WREG32(EVERGREEN_DC_LUT_WHITE_OFFSET_BLUE + radeon_crtc->crtc_offset, 0xffff); + WREG32(EVERGREEN_DC_LUT_WHITE_OFFSET_GREEN + radeon_crtc->crtc_offset, 0xffff); + WREG32(EVERGREEN_DC_LUT_WHITE_OFFSET_RED + radeon_crtc->crtc_offset, 0xffff); + + WREG32(EVERGREEN_DC_LUT_RW_MODE + radeon_crtc->crtc_offset, 0); + WREG32(EVERGREEN_DC_LUT_WRITE_EN_MASK + radeon_crtc->crtc_offset, 0x00000007); + + WREG32(EVERGREEN_DC_LUT_RW_INDEX + radeon_crtc->crtc_offset, 0); + for (i = 0; i < 256; i++) { + WREG32(EVERGREEN_DC_LUT_30_COLOR + radeon_crtc->crtc_offset, + (radeon_crtc->lut_r[i] << 20) | + (radeon_crtc->lut_g[i] << 10) | + (radeon_crtc->lut_b[i] << 0)); + } +} + +static void dce5_crtc_load_lut(struct drm_crtc *crtc) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + int i; + + DRM_DEBUG_KMS("%d\n", radeon_crtc->crtc_id); + + WREG32(NI_INPUT_CSC_CONTROL + radeon_crtc->crtc_offset, + (NI_INPUT_CSC_GRPH_MODE(NI_INPUT_CSC_BYPASS) | + NI_INPUT_CSC_OVL_MODE(NI_INPUT_CSC_BYPASS))); + WREG32(NI_PRESCALE_GRPH_CONTROL + radeon_crtc->crtc_offset, + NI_GRPH_PRESCALE_BYPASS); + WREG32(NI_PRESCALE_OVL_CONTROL + radeon_crtc->crtc_offset, + NI_OVL_PRESCALE_BYPASS); + WREG32(NI_INPUT_GAMMA_CONTROL + radeon_crtc->crtc_offset, + (NI_GRPH_INPUT_GAMMA_MODE(NI_INPUT_GAMMA_USE_LUT) | + NI_OVL_INPUT_GAMMA_MODE(NI_INPUT_GAMMA_USE_LUT))); + + WREG32(EVERGREEN_DC_LUT_CONTROL + radeon_crtc->crtc_offset, 0); + + WREG32(EVERGREEN_DC_LUT_BLACK_OFFSET_BLUE + radeon_crtc->crtc_offset, 0); + WREG32(EVERGREEN_DC_LUT_BLACK_OFFSET_GREEN + radeon_crtc->crtc_offset, 0); + WREG32(EVERGREEN_DC_LUT_BLACK_OFFSET_RED + radeon_crtc->crtc_offset, 0); + + WREG32(EVERGREEN_DC_LUT_WHITE_OFFSET_BLUE + radeon_crtc->crtc_offset, 0xffff); + WREG32(EVERGREEN_DC_LUT_WHITE_OFFSET_GREEN + radeon_crtc->crtc_offset, 0xffff); + WREG32(EVERGREEN_DC_LUT_WHITE_OFFSET_RED + radeon_crtc->crtc_offset, 0xffff); + + WREG32(EVERGREEN_DC_LUT_RW_MODE + radeon_crtc->crtc_offset, 0); + WREG32(EVERGREEN_DC_LUT_WRITE_EN_MASK + radeon_crtc->crtc_offset, 0x00000007); + + WREG32(EVERGREEN_DC_LUT_RW_INDEX + radeon_crtc->crtc_offset, 0); + for (i = 0; i < 256; i++) { + WREG32(EVERGREEN_DC_LUT_30_COLOR + radeon_crtc->crtc_offset, + (radeon_crtc->lut_r[i] << 20) | + (radeon_crtc->lut_g[i] << 10) | + (radeon_crtc->lut_b[i] << 0)); + } + + WREG32(NI_DEGAMMA_CONTROL + radeon_crtc->crtc_offset, + (NI_GRPH_DEGAMMA_MODE(NI_DEGAMMA_BYPASS) | + NI_OVL_DEGAMMA_MODE(NI_DEGAMMA_BYPASS) | + NI_ICON_DEGAMMA_MODE(NI_DEGAMMA_BYPASS) | + NI_CURSOR_DEGAMMA_MODE(NI_DEGAMMA_BYPASS))); + WREG32(NI_GAMUT_REMAP_CONTROL + radeon_crtc->crtc_offset, + (NI_GRPH_GAMUT_REMAP_MODE(NI_GAMUT_REMAP_BYPASS) | + NI_OVL_GAMUT_REMAP_MODE(NI_GAMUT_REMAP_BYPASS))); + WREG32(NI_REGAMMA_CONTROL + radeon_crtc->crtc_offset, + (NI_GRPH_REGAMMA_MODE(NI_REGAMMA_BYPASS) | + NI_OVL_REGAMMA_MODE(NI_REGAMMA_BYPASS))); + WREG32(NI_OUTPUT_CSC_CONTROL + radeon_crtc->crtc_offset, + (NI_OUTPUT_CSC_GRPH_MODE(NI_OUTPUT_CSC_BYPASS) | + NI_OUTPUT_CSC_OVL_MODE(NI_OUTPUT_CSC_BYPASS))); + /* XXX match this to the depth of the crtc fmt block, move to modeset? */ + WREG32(0x6940 + radeon_crtc->crtc_offset, 0); + +} + +static void legacy_crtc_load_lut(struct drm_crtc *crtc) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + int i; + uint32_t dac2_cntl; + + dac2_cntl = RREG32(RADEON_DAC_CNTL2); + if (radeon_crtc->crtc_id == 0) + dac2_cntl &= (uint32_t)~RADEON_DAC2_PALETTE_ACC_CTL; + else + dac2_cntl |= RADEON_DAC2_PALETTE_ACC_CTL; + WREG32(RADEON_DAC_CNTL2, dac2_cntl); + + WREG8(RADEON_PALETTE_INDEX, 0); + for (i = 0; i < 256; i++) { + WREG32(RADEON_PALETTE_30_DATA, + (radeon_crtc->lut_r[i] << 20) | + (radeon_crtc->lut_g[i] << 10) | + (radeon_crtc->lut_b[i] << 0)); + } +} + +void radeon_crtc_load_lut(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + + if (!crtc->enabled) + return; + + if (ASIC_IS_DCE5(rdev)) + dce5_crtc_load_lut(crtc); + else if (ASIC_IS_DCE4(rdev)) + dce4_crtc_load_lut(crtc); + else if (ASIC_IS_AVIVO(rdev)) + avivo_crtc_load_lut(crtc); + else + legacy_crtc_load_lut(crtc); +} + +/** Sets the color ramps on behalf of fbcon */ +void radeon_crtc_fb_gamma_set(struct drm_crtc *crtc, u16 red, u16 green, + u16 blue, int regno) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + + radeon_crtc->lut_r[regno] = red >> 6; + radeon_crtc->lut_g[regno] = green >> 6; + radeon_crtc->lut_b[regno] = blue >> 6; +} + +/** Gets the color ramps on behalf of fbcon */ +void radeon_crtc_fb_gamma_get(struct drm_crtc *crtc, u16 *red, u16 *green, + u16 *blue, int regno) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + + *red = radeon_crtc->lut_r[regno] << 6; + *green = radeon_crtc->lut_g[regno] << 6; + *blue = radeon_crtc->lut_b[regno] << 6; +} + +static void radeon_crtc_gamma_set(struct drm_crtc *crtc, u16 *red, u16 *green, + u16 *blue, uint32_t start, uint32_t size) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + int end = (start + size > 256) ? 256 : start + size, i; + + /* userspace palettes are always correct as is */ + for (i = start; i < end; i++) { + radeon_crtc->lut_r[i] = red[i] >> 6; + radeon_crtc->lut_g[i] = green[i] >> 6; + radeon_crtc->lut_b[i] = blue[i] >> 6; + } + radeon_crtc_load_lut(crtc); +} + +static void radeon_crtc_destroy(struct drm_crtc *crtc) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + + drm_crtc_cleanup(crtc); + kfree(radeon_crtc); +} + +/* + * Handle unpin events outside the interrupt handler proper. + */ +static void radeon_unpin_work_func(struct work_struct *__work) +{ + struct radeon_unpin_work *work = + container_of(__work, struct radeon_unpin_work, work); + int r; + + /* unpin of the old buffer */ + r = radeon_bo_reserve(work->old_rbo, false); + if (likely(r == 0)) { + r = radeon_bo_unpin(work->old_rbo); + if (unlikely(r != 0)) { + DRM_ERROR("failed to unpin buffer after flip\n"); + } + radeon_bo_unreserve(work->old_rbo); + } else + DRM_ERROR("failed to reserve buffer after flip\n"); + + drm_gem_object_unreference_unlocked(&work->old_rbo->gem_base); + kfree(work); +} + +void radeon_crtc_handle_flip(struct radeon_device *rdev, int crtc_id) +{ + struct radeon_crtc *radeon_crtc = rdev->mode_info.crtcs[crtc_id]; + struct radeon_unpin_work *work; + struct drm_pending_vblank_event *e; + struct timeval now; + unsigned long flags; + u32 update_pending; + int vpos, hpos; + + spin_lock_irqsave(&rdev->ddev->event_lock, flags); + work = radeon_crtc->unpin_work; + if (work == NULL || + !radeon_fence_signaled(work->fence)) { + spin_unlock_irqrestore(&rdev->ddev->event_lock, flags); + return; + } + /* New pageflip, or just completion of a previous one? */ + if (!radeon_crtc->deferred_flip_completion) { + /* do the flip (mmio) */ + update_pending = radeon_page_flip(rdev, crtc_id, work->new_crtc_base); + } else { + /* This is just a completion of a flip queued in crtc + * at last invocation. Make sure we go directly to + * completion routine. + */ + update_pending = 0; + radeon_crtc->deferred_flip_completion = 0; + } + + /* Has the pageflip already completed in crtc, or is it certain + * to complete in this vblank? + */ + if (update_pending && + (DRM_SCANOUTPOS_VALID & radeon_get_crtc_scanoutpos(rdev->ddev, crtc_id, + &vpos, &hpos)) && + (vpos >=0) && + (vpos < (99 * rdev->mode_info.crtcs[crtc_id]->base.hwmode.crtc_vdisplay)/100)) { + /* crtc didn't flip in this target vblank interval, + * but flip is pending in crtc. It will complete it + * in next vblank interval, so complete the flip at + * next vblank irq. + */ + radeon_crtc->deferred_flip_completion = 1; + spin_unlock_irqrestore(&rdev->ddev->event_lock, flags); + return; + } + + /* Pageflip (will be) certainly completed in this vblank. Clean up. */ + radeon_crtc->unpin_work = NULL; + + /* wakeup userspace */ + if (work->event) { + e = work->event; + e->event.sequence = drm_vblank_count_and_time(rdev->ddev, crtc_id, &now); + e->event.tv_sec = now.tv_sec; + e->event.tv_usec = now.tv_usec; + list_add_tail(&e->base.link, &e->base.file_priv->event_list); + wake_up_interruptible(&e->base.file_priv->event_wait); + } + spin_unlock_irqrestore(&rdev->ddev->event_lock, flags); + + drm_vblank_put(rdev->ddev, radeon_crtc->crtc_id); + radeon_fence_unref(&work->fence); + radeon_post_page_flip(work->rdev, work->crtc_id); + schedule_work(&work->work); +} + +static int radeon_crtc_page_flip(struct drm_crtc *crtc, + struct drm_framebuffer *fb, + struct drm_pending_vblank_event *event) +{ + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct radeon_framebuffer *old_radeon_fb; + struct radeon_framebuffer *new_radeon_fb; + struct drm_gem_object *obj; + struct radeon_bo *rbo; + struct radeon_fence *fence; + struct radeon_unpin_work *work; + unsigned long flags; + u32 tiling_flags, pitch_pixels; + u64 base; + int r; + + work = kzalloc(sizeof *work, GFP_KERNEL); + if (work == NULL) + return -ENOMEM; + + r = radeon_fence_create(rdev, &fence); + if (unlikely(r != 0)) { + kfree(work); + DRM_ERROR("flip queue: failed to create fence.\n"); + return -ENOMEM; + } + work->event = event; + work->rdev = rdev; + work->crtc_id = radeon_crtc->crtc_id; + work->fence = radeon_fence_ref(fence); + old_radeon_fb = to_radeon_framebuffer(crtc->fb); + new_radeon_fb = to_radeon_framebuffer(fb); + /* schedule unpin of the old buffer */ + obj = old_radeon_fb->obj; + /* take a reference to the old object */ + drm_gem_object_reference(obj); + rbo = gem_to_radeon_bo(obj); + work->old_rbo = rbo; + INIT_WORK(&work->work, radeon_unpin_work_func); + + /* We borrow the event spin lock for protecting unpin_work */ + spin_lock_irqsave(&dev->event_lock, flags); + if (radeon_crtc->unpin_work) { + DRM_DEBUG_DRIVER("flip queue: crtc already busy\n"); + r = -EBUSY; + goto unlock_free; + } + radeon_crtc->unpin_work = work; + radeon_crtc->deferred_flip_completion = 0; + spin_unlock_irqrestore(&dev->event_lock, flags); + + /* pin the new buffer */ + obj = new_radeon_fb->obj; + rbo = gem_to_radeon_bo(obj); + + DRM_DEBUG_DRIVER("flip-ioctl() cur_fbo = %p, cur_bbo = %p\n", + work->old_rbo, rbo); + + r = radeon_bo_reserve(rbo, false); + if (unlikely(r != 0)) { + DRM_ERROR("failed to reserve new rbo buffer before flip\n"); + goto pflip_cleanup; + } + r = radeon_bo_pin(rbo, RADEON_GEM_DOMAIN_VRAM, &base); + if (unlikely(r != 0)) { + radeon_bo_unreserve(rbo); + r = -EINVAL; + DRM_ERROR("failed to pin new rbo buffer before flip\n"); + goto pflip_cleanup; + } + radeon_bo_get_tiling_flags(rbo, &tiling_flags, NULL); + radeon_bo_unreserve(rbo); + + if (!ASIC_IS_AVIVO(rdev)) { + /* crtc offset is from display base addr not FB location */ + base -= radeon_crtc->legacy_display_base_addr; + pitch_pixels = fb->pitch / (fb->bits_per_pixel / 8); + + if (tiling_flags & RADEON_TILING_MACRO) { + if (ASIC_IS_R300(rdev)) { + base &= ~0x7ff; + } else { + int byteshift = fb->bits_per_pixel >> 4; + int tile_addr = (((crtc->y >> 3) * pitch_pixels + crtc->x) >> (8 - byteshift)) << 11; + base += tile_addr + ((crtc->x << byteshift) % 256) + ((crtc->y % 8) << 8); + } + } else { + int offset = crtc->y * pitch_pixels + crtc->x; + switch (fb->bits_per_pixel) { + case 8: + default: + offset *= 1; + break; + case 15: + case 16: + offset *= 2; + break; + case 24: + offset *= 3; + break; + case 32: + offset *= 4; + break; + } + base += offset; + } + base &= ~7; + } + + spin_lock_irqsave(&dev->event_lock, flags); + work->new_crtc_base = base; + spin_unlock_irqrestore(&dev->event_lock, flags); + + /* update crtc fb */ + crtc->fb = fb; + + r = drm_vblank_get(dev, radeon_crtc->crtc_id); + if (r) { + DRM_ERROR("failed to get vblank before flip\n"); + goto pflip_cleanup1; + } + + /* 32 ought to cover us */ + r = radeon_ring_lock(rdev, 32); + if (r) { + DRM_ERROR("failed to lock the ring before flip\n"); + goto pflip_cleanup2; + } + + /* emit the fence */ + radeon_fence_emit(rdev, fence); + /* set the proper interrupt */ + radeon_pre_page_flip(rdev, radeon_crtc->crtc_id); + /* fire the ring */ + radeon_ring_unlock_commit(rdev); + + return 0; + +pflip_cleanup2: + drm_vblank_put(dev, radeon_crtc->crtc_id); + +pflip_cleanup1: + r = radeon_bo_reserve(rbo, false); + if (unlikely(r != 0)) { + DRM_ERROR("failed to reserve new rbo in error path\n"); + goto pflip_cleanup; + } + r = radeon_bo_unpin(rbo); + if (unlikely(r != 0)) { + radeon_bo_unreserve(rbo); + r = -EINVAL; + DRM_ERROR("failed to unpin new rbo in error path\n"); + goto pflip_cleanup; + } + radeon_bo_unreserve(rbo); + +pflip_cleanup: + spin_lock_irqsave(&dev->event_lock, flags); + radeon_crtc->unpin_work = NULL; +unlock_free: + drm_gem_object_unreference_unlocked(old_radeon_fb->obj); + spin_unlock_irqrestore(&dev->event_lock, flags); + radeon_fence_unref(&fence); + kfree(work); + + return r; +} + +static const struct drm_crtc_funcs radeon_crtc_funcs = { + .cursor_set = radeon_crtc_cursor_set, + .cursor_move = radeon_crtc_cursor_move, + .gamma_set = radeon_crtc_gamma_set, + .set_config = drm_crtc_helper_set_config, + .destroy = radeon_crtc_destroy, + .page_flip = radeon_crtc_page_flip, +}; + +static void radeon_crtc_init(struct drm_device *dev, int index) +{ + struct radeon_device *rdev = dev->dev_private; + struct radeon_crtc *radeon_crtc; + int i; + + radeon_crtc = kzalloc(sizeof(struct radeon_crtc) + (RADEONFB_CONN_LIMIT * sizeof(struct drm_connector *)), GFP_KERNEL); + if (radeon_crtc == NULL) + return; + + drm_crtc_init(dev, &radeon_crtc->base, &radeon_crtc_funcs); + + drm_mode_crtc_set_gamma_size(&radeon_crtc->base, 256); + radeon_crtc->crtc_id = index; + rdev->mode_info.crtcs[index] = radeon_crtc; + +#if 0 + radeon_crtc->mode_set.crtc = &radeon_crtc->base; + radeon_crtc->mode_set.connectors = (struct drm_connector **)(radeon_crtc + 1); + radeon_crtc->mode_set.num_connectors = 0; +#endif + + for (i = 0; i < 256; i++) { + radeon_crtc->lut_r[i] = i << 2; + radeon_crtc->lut_g[i] = i << 2; + radeon_crtc->lut_b[i] = i << 2; + } + + if (rdev->is_atom_bios && (ASIC_IS_AVIVO(rdev) || radeon_r4xx_atom)) + radeon_atombios_init_crtc(dev, radeon_crtc); + else + radeon_legacy_init_crtc(dev, radeon_crtc); +} + +static const char *encoder_names[36] = { + "NONE", + "INTERNAL_LVDS", + "INTERNAL_TMDS1", + "INTERNAL_TMDS2", + "INTERNAL_DAC1", + "INTERNAL_DAC2", + "INTERNAL_SDVOA", + "INTERNAL_SDVOB", + "SI170B", + "CH7303", + "CH7301", + "INTERNAL_DVO1", + "EXTERNAL_SDVOA", + "EXTERNAL_SDVOB", + "TITFP513", + "INTERNAL_LVTM1", + "VT1623", + "HDMI_SI1930", + "HDMI_INTERNAL", + "INTERNAL_KLDSCP_TMDS1", + "INTERNAL_KLDSCP_DVO1", + "INTERNAL_KLDSCP_DAC1", + "INTERNAL_KLDSCP_DAC2", + "SI178", + "MVPU_FPGA", + "INTERNAL_DDI", + "VT1625", + "HDMI_SI1932", + "DP_AN9801", + "DP_DP501", + "INTERNAL_UNIPHY", + "INTERNAL_KLDSCP_LVTMA", + "INTERNAL_UNIPHY1", + "INTERNAL_UNIPHY2", + "NUTMEG", + "TRAVIS", +}; + +static const char *connector_names[15] = { + "Unknown", + "VGA", + "DVI-I", + "DVI-D", + "DVI-A", + "Composite", + "S-video", + "LVDS", + "Component", + "DIN", + "DisplayPort", + "HDMI-A", + "HDMI-B", + "TV", + "eDP", +}; + +static const char *hpd_names[6] = { + "HPD1", + "HPD2", + "HPD3", + "HPD4", + "HPD5", + "HPD6", +}; + +static void radeon_print_display_setup(struct drm_device *dev) +{ + struct drm_connector *connector; + struct radeon_connector *radeon_connector; + struct drm_encoder *encoder; + struct radeon_encoder *radeon_encoder; + uint32_t devices; + int i = 0; + + DRM_INFO("Radeon Display Connectors\n"); + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + radeon_connector = to_radeon_connector(connector); + DRM_INFO("Connector %d:\n", i); + DRM_INFO(" %s\n", connector_names[connector->connector_type]); + if (radeon_connector->hpd.hpd != RADEON_HPD_NONE) + DRM_INFO(" %s\n", hpd_names[radeon_connector->hpd.hpd]); + if (radeon_connector->ddc_bus) { + DRM_INFO(" DDC: 0x%x 0x%x 0x%x 0x%x 0x%x 0x%x 0x%x 0x%x\n", + radeon_connector->ddc_bus->rec.mask_clk_reg, + radeon_connector->ddc_bus->rec.mask_data_reg, + radeon_connector->ddc_bus->rec.a_clk_reg, + radeon_connector->ddc_bus->rec.a_data_reg, + radeon_connector->ddc_bus->rec.en_clk_reg, + radeon_connector->ddc_bus->rec.en_data_reg, + radeon_connector->ddc_bus->rec.y_clk_reg, + radeon_connector->ddc_bus->rec.y_data_reg); + if (radeon_connector->router.ddc_valid) + DRM_INFO(" DDC Router 0x%x/0x%x\n", + radeon_connector->router.ddc_mux_control_pin, + radeon_connector->router.ddc_mux_state); + if (radeon_connector->router.cd_valid) + DRM_INFO(" Clock/Data Router 0x%x/0x%x\n", + radeon_connector->router.cd_mux_control_pin, + radeon_connector->router.cd_mux_state); + } else { + if (connector->connector_type == DRM_MODE_CONNECTOR_VGA || + connector->connector_type == DRM_MODE_CONNECTOR_DVII || + connector->connector_type == DRM_MODE_CONNECTOR_DVID || + connector->connector_type == DRM_MODE_CONNECTOR_DVIA || + connector->connector_type == DRM_MODE_CONNECTOR_HDMIA || + connector->connector_type == DRM_MODE_CONNECTOR_HDMIB) + DRM_INFO(" DDC: no ddc bus - possible BIOS bug - please report to xorg-driver-ati@lists.x.org\n"); + } + DRM_INFO(" Encoders:\n"); + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + radeon_encoder = to_radeon_encoder(encoder); + devices = radeon_encoder->devices & radeon_connector->devices; + if (devices) { + if (devices & ATOM_DEVICE_CRT1_SUPPORT) + DRM_INFO(" CRT1: %s\n", encoder_names[radeon_encoder->encoder_id]); + if (devices & ATOM_DEVICE_CRT2_SUPPORT) + DRM_INFO(" CRT2: %s\n", encoder_names[radeon_encoder->encoder_id]); + if (devices & ATOM_DEVICE_LCD1_SUPPORT) + DRM_INFO(" LCD1: %s\n", encoder_names[radeon_encoder->encoder_id]); + if (devices & ATOM_DEVICE_DFP1_SUPPORT) + DRM_INFO(" DFP1: %s\n", encoder_names[radeon_encoder->encoder_id]); + if (devices & ATOM_DEVICE_DFP2_SUPPORT) + DRM_INFO(" DFP2: %s\n", encoder_names[radeon_encoder->encoder_id]); + if (devices & ATOM_DEVICE_DFP3_SUPPORT) + DRM_INFO(" DFP3: %s\n", encoder_names[radeon_encoder->encoder_id]); + if (devices & ATOM_DEVICE_DFP4_SUPPORT) + DRM_INFO(" DFP4: %s\n", encoder_names[radeon_encoder->encoder_id]); + if (devices & ATOM_DEVICE_DFP5_SUPPORT) + DRM_INFO(" DFP5: %s\n", encoder_names[radeon_encoder->encoder_id]); + if (devices & ATOM_DEVICE_DFP6_SUPPORT) + DRM_INFO(" DFP6: %s\n", encoder_names[radeon_encoder->encoder_id]); + if (devices & ATOM_DEVICE_TV1_SUPPORT) + DRM_INFO(" TV1: %s\n", encoder_names[radeon_encoder->encoder_id]); + if (devices & ATOM_DEVICE_CV_SUPPORT) + DRM_INFO(" CV: %s\n", encoder_names[radeon_encoder->encoder_id]); + } + } + i++; + } +} + +static bool radeon_setup_enc_conn(struct drm_device *dev) +{ + struct radeon_device *rdev = dev->dev_private; + struct drm_connector *drm_connector; + bool ret = false; + + if (rdev->bios) { + if (rdev->is_atom_bios) { + ret = radeon_get_atom_connector_info_from_supported_devices_table(dev); + if (ret == false) + ret = radeon_get_atom_connector_info_from_object_table(dev); + } else { + ret = radeon_get_legacy_connector_info_from_bios(dev); + if (ret == false) + ret = radeon_get_legacy_connector_info_from_table(dev); + } + } else { + if (!ASIC_IS_AVIVO(rdev)) + ret = radeon_get_legacy_connector_info_from_table(dev); + } + if (ret) { + radeon_setup_encoder_clones(dev); + radeon_print_display_setup(dev); + list_for_each_entry(drm_connector, &dev->mode_config.connector_list, head) + radeon_ddc_dump(drm_connector); + } + + return ret; +} + +int radeon_ddc_get_modes(struct radeon_connector *radeon_connector) +{ + struct drm_device *dev = radeon_connector->base.dev; + struct radeon_device *rdev = dev->dev_private; + int ret = 0; + + /* on hw with routers, select right port */ + if (radeon_connector->router.ddc_valid) + radeon_router_select_ddc_port(radeon_connector); + + if ((radeon_connector->base.connector_type == DRM_MODE_CONNECTOR_DisplayPort) || + (radeon_connector->base.connector_type == DRM_MODE_CONNECTOR_eDP)) { + struct radeon_connector_atom_dig *dig = radeon_connector->con_priv; + if ((dig->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT || + dig->dp_sink_type == CONNECTOR_OBJECT_ID_eDP) && dig->dp_i2c_bus) + radeon_connector->edid = drm_get_edid(&radeon_connector->base, &dig->dp_i2c_bus->adapter); + } + if (!radeon_connector->ddc_bus) + return -1; + if (!radeon_connector->edid) { + radeon_connector->edid = drm_get_edid(&radeon_connector->base, &radeon_connector->ddc_bus->adapter); + } + + if (!radeon_connector->edid) { + if (rdev->is_atom_bios) { + /* some laptops provide a hardcoded edid in rom for LCDs */ + if (((radeon_connector->base.connector_type == DRM_MODE_CONNECTOR_LVDS) || + (radeon_connector->base.connector_type == DRM_MODE_CONNECTOR_eDP))) + radeon_connector->edid = radeon_bios_get_hardcoded_edid(rdev); + } else + /* some servers provide a hardcoded edid in rom for KVMs */ + radeon_connector->edid = radeon_bios_get_hardcoded_edid(rdev); + } + if (radeon_connector->edid) { + drm_mode_connector_update_edid_property(&radeon_connector->base, radeon_connector->edid); + ret = drm_add_edid_modes(&radeon_connector->base, radeon_connector->edid); + return ret; + } + drm_mode_connector_update_edid_property(&radeon_connector->base, NULL); + return 0; +} + +static int radeon_ddc_dump(struct drm_connector *connector) +{ + struct edid *edid; + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + int ret = 0; + + /* on hw with routers, select right port */ + if (radeon_connector->router.ddc_valid) + radeon_router_select_ddc_port(radeon_connector); + + if (!radeon_connector->ddc_bus) + return -1; + edid = drm_get_edid(connector, &radeon_connector->ddc_bus->adapter); + /* Log EDID retrieval status here. In particular with regard to + * connectors with requires_extended_probe flag set, that will prevent + * function radeon_dvi_detect() to fetch EDID on this connector, + * as long as there is no valid EDID header found */ + if (edid) { + DRM_INFO("Radeon display connector %s: Found valid EDID", + drm_get_connector_name(connector)); + kfree(edid); + } else { + DRM_INFO("Radeon display connector %s: No monitor connected or invalid EDID", + drm_get_connector_name(connector)); + } + return ret; +} + +/* avivo */ +static void avivo_get_fb_div(struct radeon_pll *pll, + u32 target_clock, + u32 post_div, + u32 ref_div, + u32 *fb_div, + u32 *frac_fb_div) +{ + u32 tmp = post_div * ref_div; + + tmp *= target_clock; + *fb_div = tmp / pll->reference_freq; + *frac_fb_div = tmp % pll->reference_freq; + + if (*fb_div > pll->max_feedback_div) + *fb_div = pll->max_feedback_div; + else if (*fb_div < pll->min_feedback_div) + *fb_div = pll->min_feedback_div; +} + +static u32 avivo_get_post_div(struct radeon_pll *pll, + u32 target_clock) +{ + u32 vco, post_div, tmp; + + if (pll->flags & RADEON_PLL_USE_POST_DIV) + return pll->post_div; + + if (pll->flags & RADEON_PLL_PREFER_MINM_OVER_MAXP) { + if (pll->flags & RADEON_PLL_IS_LCD) + vco = pll->lcd_pll_out_min; + else + vco = pll->pll_out_min; + } else { + if (pll->flags & RADEON_PLL_IS_LCD) + vco = pll->lcd_pll_out_max; + else + vco = pll->pll_out_max; + } + + post_div = vco / target_clock; + tmp = vco % target_clock; + + if (pll->flags & RADEON_PLL_PREFER_MINM_OVER_MAXP) { + if (tmp) + post_div++; + } else { + if (!tmp) + post_div--; + } + + if (post_div > pll->max_post_div) + post_div = pll->max_post_div; + else if (post_div < pll->min_post_div) + post_div = pll->min_post_div; + + return post_div; +} + +#define MAX_TOLERANCE 10 + +void radeon_compute_pll_avivo(struct radeon_pll *pll, + u32 freq, + u32 *dot_clock_p, + u32 *fb_div_p, + u32 *frac_fb_div_p, + u32 *ref_div_p, + u32 *post_div_p) +{ + u32 target_clock = freq / 10; + u32 post_div = avivo_get_post_div(pll, target_clock); + u32 ref_div = pll->min_ref_div; + u32 fb_div = 0, frac_fb_div = 0, tmp; + + if (pll->flags & RADEON_PLL_USE_REF_DIV) + ref_div = pll->reference_div; + + if (pll->flags & RADEON_PLL_USE_FRAC_FB_DIV) { + avivo_get_fb_div(pll, target_clock, post_div, ref_div, &fb_div, &frac_fb_div); + frac_fb_div = (100 * frac_fb_div) / pll->reference_freq; + if (frac_fb_div >= 5) { + frac_fb_div -= 5; + frac_fb_div = frac_fb_div / 10; + frac_fb_div++; + } + if (frac_fb_div >= 10) { + fb_div++; + frac_fb_div = 0; + } + } else { + while (ref_div <= pll->max_ref_div) { + avivo_get_fb_div(pll, target_clock, post_div, ref_div, + &fb_div, &frac_fb_div); + if (frac_fb_div >= (pll->reference_freq / 2)) + fb_div++; + frac_fb_div = 0; + tmp = (pll->reference_freq * fb_div) / (post_div * ref_div); + tmp = (tmp * 10000) / target_clock; + + if (tmp > (10000 + MAX_TOLERANCE)) + ref_div++; + else if (tmp >= (10000 - MAX_TOLERANCE)) + break; + else + ref_div++; + } + } + + *dot_clock_p = ((pll->reference_freq * fb_div * 10) + (pll->reference_freq * frac_fb_div)) / + (ref_div * post_div * 10); + *fb_div_p = fb_div; + *frac_fb_div_p = frac_fb_div; + *ref_div_p = ref_div; + *post_div_p = post_div; + DRM_DEBUG_KMS("%d, pll dividers - fb: %d.%d ref: %d, post %d\n", + *dot_clock_p, fb_div, frac_fb_div, ref_div, post_div); +} + +/* pre-avivo */ +static inline uint32_t radeon_div(uint64_t n, uint32_t d) +{ + uint64_t mod; + + n += d / 2; + + mod = do_div(n, d); + return n; +} + +void radeon_compute_pll_legacy(struct radeon_pll *pll, + uint64_t freq, + uint32_t *dot_clock_p, + uint32_t *fb_div_p, + uint32_t *frac_fb_div_p, + uint32_t *ref_div_p, + uint32_t *post_div_p) +{ + uint32_t min_ref_div = pll->min_ref_div; + uint32_t max_ref_div = pll->max_ref_div; + uint32_t min_post_div = pll->min_post_div; + uint32_t max_post_div = pll->max_post_div; + uint32_t min_fractional_feed_div = 0; + uint32_t max_fractional_feed_div = 0; + uint32_t best_vco = pll->best_vco; + uint32_t best_post_div = 1; + uint32_t best_ref_div = 1; + uint32_t best_feedback_div = 1; + uint32_t best_frac_feedback_div = 0; + uint32_t best_freq = -1; + uint32_t best_error = 0xffffffff; + uint32_t best_vco_diff = 1; + uint32_t post_div; + u32 pll_out_min, pll_out_max; + + DRM_DEBUG_KMS("PLL freq %llu %u %u\n", freq, pll->min_ref_div, pll->max_ref_div); + freq = freq * 1000; + + if (pll->flags & RADEON_PLL_IS_LCD) { + pll_out_min = pll->lcd_pll_out_min; + pll_out_max = pll->lcd_pll_out_max; + } else { + pll_out_min = pll->pll_out_min; + pll_out_max = pll->pll_out_max; + } + + if (pll_out_min > 64800) + pll_out_min = 64800; + + if (pll->flags & RADEON_PLL_USE_REF_DIV) + min_ref_div = max_ref_div = pll->reference_div; + else { + while (min_ref_div < max_ref_div-1) { + uint32_t mid = (min_ref_div + max_ref_div) / 2; + uint32_t pll_in = pll->reference_freq / mid; + if (pll_in < pll->pll_in_min) + max_ref_div = mid; + else if (pll_in > pll->pll_in_max) + min_ref_div = mid; + else + break; + } + } + + if (pll->flags & RADEON_PLL_USE_POST_DIV) + min_post_div = max_post_div = pll->post_div; + + if (pll->flags & RADEON_PLL_USE_FRAC_FB_DIV) { + min_fractional_feed_div = pll->min_frac_feedback_div; + max_fractional_feed_div = pll->max_frac_feedback_div; + } + + for (post_div = max_post_div; post_div >= min_post_div; --post_div) { + uint32_t ref_div; + + if ((pll->flags & RADEON_PLL_NO_ODD_POST_DIV) && (post_div & 1)) + continue; + + /* legacy radeons only have a few post_divs */ + if (pll->flags & RADEON_PLL_LEGACY) { + if ((post_div == 5) || + (post_div == 7) || + (post_div == 9) || + (post_div == 10) || + (post_div == 11) || + (post_div == 13) || + (post_div == 14) || + (post_div == 15)) + continue; + } + + for (ref_div = min_ref_div; ref_div <= max_ref_div; ++ref_div) { + uint32_t feedback_div, current_freq = 0, error, vco_diff; + uint32_t pll_in = pll->reference_freq / ref_div; + uint32_t min_feed_div = pll->min_feedback_div; + uint32_t max_feed_div = pll->max_feedback_div + 1; + + if (pll_in < pll->pll_in_min || pll_in > pll->pll_in_max) + continue; + + while (min_feed_div < max_feed_div) { + uint32_t vco; + uint32_t min_frac_feed_div = min_fractional_feed_div; + uint32_t max_frac_feed_div = max_fractional_feed_div + 1; + uint32_t frac_feedback_div; + uint64_t tmp; + + feedback_div = (min_feed_div + max_feed_div) / 2; + + tmp = (uint64_t)pll->reference_freq * feedback_div; + vco = radeon_div(tmp, ref_div); + + if (vco < pll_out_min) { + min_feed_div = feedback_div + 1; + continue; + } else if (vco > pll_out_max) { + max_feed_div = feedback_div; + continue; + } + + while (min_frac_feed_div < max_frac_feed_div) { + frac_feedback_div = (min_frac_feed_div + max_frac_feed_div) / 2; + tmp = (uint64_t)pll->reference_freq * 10000 * feedback_div; + tmp += (uint64_t)pll->reference_freq * 1000 * frac_feedback_div; + current_freq = radeon_div(tmp, ref_div * post_div); + + if (pll->flags & RADEON_PLL_PREFER_CLOSEST_LOWER) { + if (freq < current_freq) + error = 0xffffffff; + else + error = freq - current_freq; + } else + error = abs(current_freq - freq); + vco_diff = abs(vco - best_vco); + + if ((best_vco == 0 && error < best_error) || + (best_vco != 0 && + ((best_error > 100 && error < best_error - 100) || + (abs(error - best_error) < 100 && vco_diff < best_vco_diff)))) { + best_post_div = post_div; + best_ref_div = ref_div; + best_feedback_div = feedback_div; + best_frac_feedback_div = frac_feedback_div; + best_freq = current_freq; + best_error = error; + best_vco_diff = vco_diff; + } else if (current_freq == freq) { + if (best_freq == -1) { + best_post_div = post_div; + best_ref_div = ref_div; + best_feedback_div = feedback_div; + best_frac_feedback_div = frac_feedback_div; + best_freq = current_freq; + best_error = error; + best_vco_diff = vco_diff; + } else if (((pll->flags & RADEON_PLL_PREFER_LOW_REF_DIV) && (ref_div < best_ref_div)) || + ((pll->flags & RADEON_PLL_PREFER_HIGH_REF_DIV) && (ref_div > best_ref_div)) || + ((pll->flags & RADEON_PLL_PREFER_LOW_FB_DIV) && (feedback_div < best_feedback_div)) || + ((pll->flags & RADEON_PLL_PREFER_HIGH_FB_DIV) && (feedback_div > best_feedback_div)) || + ((pll->flags & RADEON_PLL_PREFER_LOW_POST_DIV) && (post_div < best_post_div)) || + ((pll->flags & RADEON_PLL_PREFER_HIGH_POST_DIV) && (post_div > best_post_div))) { + best_post_div = post_div; + best_ref_div = ref_div; + best_feedback_div = feedback_div; + best_frac_feedback_div = frac_feedback_div; + best_freq = current_freq; + best_error = error; + best_vco_diff = vco_diff; + } + } + if (current_freq < freq) + min_frac_feed_div = frac_feedback_div + 1; + else + max_frac_feed_div = frac_feedback_div; + } + if (current_freq < freq) + min_feed_div = feedback_div + 1; + else + max_feed_div = feedback_div; + } + } + } + + *dot_clock_p = best_freq / 10000; + *fb_div_p = best_feedback_div; + *frac_fb_div_p = best_frac_feedback_div; + *ref_div_p = best_ref_div; + *post_div_p = best_post_div; + DRM_DEBUG_KMS("%lld %d, pll dividers - fb: %d.%d ref: %d, post %d\n", + (long long)freq, + best_freq / 1000, best_feedback_div, best_frac_feedback_div, + best_ref_div, best_post_div); + +} + +static void radeon_user_framebuffer_destroy(struct drm_framebuffer *fb) +{ + struct radeon_framebuffer *radeon_fb = to_radeon_framebuffer(fb); + + if (radeon_fb->obj) { + drm_gem_object_unreference_unlocked(radeon_fb->obj); + } + drm_framebuffer_cleanup(fb); + kfree(radeon_fb); +} + +static int radeon_user_framebuffer_create_handle(struct drm_framebuffer *fb, + struct drm_file *file_priv, + unsigned int *handle) +{ + struct radeon_framebuffer *radeon_fb = to_radeon_framebuffer(fb); + + return drm_gem_handle_create(file_priv, radeon_fb->obj, handle); +} + +static const struct drm_framebuffer_funcs radeon_fb_funcs = { + .destroy = radeon_user_framebuffer_destroy, + .create_handle = radeon_user_framebuffer_create_handle, +}; + +void +radeon_framebuffer_init(struct drm_device *dev, + struct radeon_framebuffer *rfb, + struct drm_mode_fb_cmd *mode_cmd, + struct drm_gem_object *obj) +{ + rfb->obj = obj; + drm_framebuffer_init(dev, &rfb->base, &radeon_fb_funcs); + drm_helper_mode_fill_fb_struct(&rfb->base, mode_cmd); +} + +static struct drm_framebuffer * +radeon_user_framebuffer_create(struct drm_device *dev, + struct drm_file *file_priv, + struct drm_mode_fb_cmd *mode_cmd) +{ + struct drm_gem_object *obj; + struct radeon_framebuffer *radeon_fb; + + obj = drm_gem_object_lookup(dev, file_priv, mode_cmd->handle); + if (obj == NULL) { + dev_err(&dev->pdev->dev, "No GEM object associated to handle 0x%08X, " + "can't create framebuffer\n", mode_cmd->handle); + return ERR_PTR(-ENOENT); + } + + radeon_fb = kzalloc(sizeof(*radeon_fb), GFP_KERNEL); + if (radeon_fb == NULL) + return ERR_PTR(-ENOMEM); + + radeon_framebuffer_init(dev, radeon_fb, mode_cmd, obj); + + return &radeon_fb->base; +} + +static void radeon_output_poll_changed(struct drm_device *dev) +{ + struct radeon_device *rdev = dev->dev_private; + radeon_fb_output_poll_changed(rdev); +} + +static const struct drm_mode_config_funcs radeon_mode_funcs = { + .fb_create = radeon_user_framebuffer_create, + .output_poll_changed = radeon_output_poll_changed +}; + +struct drm_prop_enum_list { + int type; + char *name; +}; + +static struct drm_prop_enum_list radeon_tmds_pll_enum_list[] = +{ { 0, "driver" }, + { 1, "bios" }, +}; + +static struct drm_prop_enum_list radeon_tv_std_enum_list[] = +{ { TV_STD_NTSC, "ntsc" }, + { TV_STD_PAL, "pal" }, + { TV_STD_PAL_M, "pal-m" }, + { TV_STD_PAL_60, "pal-60" }, + { TV_STD_NTSC_J, "ntsc-j" }, + { TV_STD_SCART_PAL, "scart-pal" }, + { TV_STD_PAL_CN, "pal-cn" }, + { TV_STD_SECAM, "secam" }, +}; + +static struct drm_prop_enum_list radeon_underscan_enum_list[] = +{ { UNDERSCAN_OFF, "off" }, + { UNDERSCAN_ON, "on" }, + { UNDERSCAN_AUTO, "auto" }, +}; + +static int radeon_modeset_create_props(struct radeon_device *rdev) +{ + int i, sz; + + if (rdev->is_atom_bios) { + rdev->mode_info.coherent_mode_property = + drm_property_create(rdev->ddev, + DRM_MODE_PROP_RANGE, + "coherent", 2); + if (!rdev->mode_info.coherent_mode_property) + return -ENOMEM; + + rdev->mode_info.coherent_mode_property->values[0] = 0; + rdev->mode_info.coherent_mode_property->values[1] = 1; + } + + if (!ASIC_IS_AVIVO(rdev)) { + sz = ARRAY_SIZE(radeon_tmds_pll_enum_list); + rdev->mode_info.tmds_pll_property = + drm_property_create(rdev->ddev, + DRM_MODE_PROP_ENUM, + "tmds_pll", sz); + for (i = 0; i < sz; i++) { + drm_property_add_enum(rdev->mode_info.tmds_pll_property, + i, + radeon_tmds_pll_enum_list[i].type, + radeon_tmds_pll_enum_list[i].name); + } + } + + rdev->mode_info.load_detect_property = + drm_property_create(rdev->ddev, + DRM_MODE_PROP_RANGE, + "load detection", 2); + if (!rdev->mode_info.load_detect_property) + return -ENOMEM; + rdev->mode_info.load_detect_property->values[0] = 0; + rdev->mode_info.load_detect_property->values[1] = 1; + + drm_mode_create_scaling_mode_property(rdev->ddev); + + sz = ARRAY_SIZE(radeon_tv_std_enum_list); + rdev->mode_info.tv_std_property = + drm_property_create(rdev->ddev, + DRM_MODE_PROP_ENUM, + "tv standard", sz); + for (i = 0; i < sz; i++) { + drm_property_add_enum(rdev->mode_info.tv_std_property, + i, + radeon_tv_std_enum_list[i].type, + radeon_tv_std_enum_list[i].name); + } + + sz = ARRAY_SIZE(radeon_underscan_enum_list); + rdev->mode_info.underscan_property = + drm_property_create(rdev->ddev, + DRM_MODE_PROP_ENUM, + "underscan", sz); + for (i = 0; i < sz; i++) { + drm_property_add_enum(rdev->mode_info.underscan_property, + i, + radeon_underscan_enum_list[i].type, + radeon_underscan_enum_list[i].name); + } + + rdev->mode_info.underscan_hborder_property = + drm_property_create(rdev->ddev, + DRM_MODE_PROP_RANGE, + "underscan hborder", 2); + if (!rdev->mode_info.underscan_hborder_property) + return -ENOMEM; + rdev->mode_info.underscan_hborder_property->values[0] = 0; + rdev->mode_info.underscan_hborder_property->values[1] = 128; + + rdev->mode_info.underscan_vborder_property = + drm_property_create(rdev->ddev, + DRM_MODE_PROP_RANGE, + "underscan vborder", 2); + if (!rdev->mode_info.underscan_vborder_property) + return -ENOMEM; + rdev->mode_info.underscan_vborder_property->values[0] = 0; + rdev->mode_info.underscan_vborder_property->values[1] = 128; + + return 0; +} + +void radeon_update_display_priority(struct radeon_device *rdev) +{ + /* adjustment options for the display watermarks */ + if ((radeon_disp_priority == 0) || (radeon_disp_priority > 2)) { + /* set display priority to high for r3xx, rv515 chips + * this avoids flickering due to underflow to the + * display controllers during heavy acceleration. + * Don't force high on rs4xx igp chips as it seems to + * affect the sound card. See kernel bug 15982. + */ + if ((ASIC_IS_R300(rdev) || (rdev->family == CHIP_RV515)) && + !(rdev->flags & RADEON_IS_IGP)) + rdev->disp_priority = 2; + else + rdev->disp_priority = 0; + } else + rdev->disp_priority = radeon_disp_priority; + +} + +int radeon_modeset_init(struct radeon_device *rdev) +{ + int i; + int ret; + + drm_mode_config_init(rdev->ddev); + rdev->mode_info.mode_config_initialized = true; + + rdev->ddev->mode_config.funcs = (void *)&radeon_mode_funcs; + + if (ASIC_IS_DCE5(rdev)) { + rdev->ddev->mode_config.max_width = 16384; + rdev->ddev->mode_config.max_height = 16384; + } else if (ASIC_IS_AVIVO(rdev)) { + rdev->ddev->mode_config.max_width = 8192; + rdev->ddev->mode_config.max_height = 8192; + } else { + rdev->ddev->mode_config.max_width = 4096; + rdev->ddev->mode_config.max_height = 4096; + } + + rdev->ddev->mode_config.fb_base = rdev->mc.aper_base; + + ret = radeon_modeset_create_props(rdev); + if (ret) { + return ret; + } + + /* init i2c buses */ + radeon_i2c_init(rdev); + + /* check combios for a valid hardcoded EDID - Sun servers */ + if (!rdev->is_atom_bios) { + /* check for hardcoded EDID in BIOS */ + radeon_combios_check_hardcoded_edid(rdev); + } + + /* allocate crtcs */ + for (i = 0; i < rdev->num_crtc; i++) { + radeon_crtc_init(rdev->ddev, i); + } + + /* okay we should have all the bios connectors */ + ret = radeon_setup_enc_conn(rdev->ddev); + if (!ret) { + return ret; + } + + /* init dig PHYs */ + if (rdev->is_atom_bios) + radeon_atom_encoder_init(rdev); + + /* initialize hpd */ + radeon_hpd_init(rdev); + + /* Initialize power management */ + radeon_pm_init(rdev); + + radeon_fbdev_init(rdev); + drm_kms_helper_poll_init(rdev->ddev); + + return 0; +} + +void radeon_modeset_fini(struct radeon_device *rdev) +{ + radeon_fbdev_fini(rdev); + kfree(rdev->mode_info.bios_hardcoded_edid); + radeon_pm_fini(rdev); + + if (rdev->mode_info.mode_config_initialized) { + drm_kms_helper_poll_fini(rdev->ddev); + radeon_hpd_fini(rdev); + drm_mode_config_cleanup(rdev->ddev); + rdev->mode_info.mode_config_initialized = false; + } + /* free i2c buses */ + radeon_i2c_fini(rdev); +} + +static bool is_hdtv_mode(struct drm_display_mode *mode) +{ + /* try and guess if this is a tv or a monitor */ + if ((mode->vdisplay == 480 && mode->hdisplay == 720) || /* 480p */ + (mode->vdisplay == 576) || /* 576p */ + (mode->vdisplay == 720) || /* 720p */ + (mode->vdisplay == 1080)) /* 1080p */ + return true; + else + return false; +} + +bool radeon_crtc_scaling_mode_fixup(struct drm_crtc *crtc, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + struct drm_encoder *encoder; + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct radeon_encoder *radeon_encoder; + struct drm_connector *connector; + struct radeon_connector *radeon_connector; + bool first = true; + u32 src_v = 1, dst_v = 1; + u32 src_h = 1, dst_h = 1; + + radeon_crtc->h_border = 0; + radeon_crtc->v_border = 0; + + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + if (encoder->crtc != crtc) + continue; + radeon_encoder = to_radeon_encoder(encoder); + connector = radeon_get_connector_for_encoder(encoder); + radeon_connector = to_radeon_connector(connector); + + if (first) { + /* set scaling */ + if (radeon_encoder->rmx_type == RMX_OFF) + radeon_crtc->rmx_type = RMX_OFF; + else if (mode->hdisplay < radeon_encoder->native_mode.hdisplay || + mode->vdisplay < radeon_encoder->native_mode.vdisplay) + radeon_crtc->rmx_type = radeon_encoder->rmx_type; + else + radeon_crtc->rmx_type = RMX_OFF; + /* copy native mode */ + memcpy(&radeon_crtc->native_mode, + &radeon_encoder->native_mode, + sizeof(struct drm_display_mode)); + src_v = crtc->mode.vdisplay; + dst_v = radeon_crtc->native_mode.vdisplay; + src_h = crtc->mode.hdisplay; + dst_h = radeon_crtc->native_mode.hdisplay; + + /* fix up for overscan on hdmi */ + if (ASIC_IS_AVIVO(rdev) && + (!(mode->flags & DRM_MODE_FLAG_INTERLACE)) && + ((radeon_encoder->underscan_type == UNDERSCAN_ON) || + ((radeon_encoder->underscan_type == UNDERSCAN_AUTO) && + drm_detect_hdmi_monitor(radeon_connector->edid) && + is_hdtv_mode(mode)))) { + if (radeon_encoder->underscan_hborder != 0) + radeon_crtc->h_border = radeon_encoder->underscan_hborder; + else + radeon_crtc->h_border = (mode->hdisplay >> 5) + 16; + if (radeon_encoder->underscan_vborder != 0) + radeon_crtc->v_border = radeon_encoder->underscan_vborder; + else + radeon_crtc->v_border = (mode->vdisplay >> 5) + 16; + radeon_crtc->rmx_type = RMX_FULL; + src_v = crtc->mode.vdisplay; + dst_v = crtc->mode.vdisplay - (radeon_crtc->v_border * 2); + src_h = crtc->mode.hdisplay; + dst_h = crtc->mode.hdisplay - (radeon_crtc->h_border * 2); + } + first = false; + } else { + if (radeon_crtc->rmx_type != radeon_encoder->rmx_type) { + /* WARNING: Right now this can't happen but + * in the future we need to check that scaling + * are consistent across different encoder + * (ie all encoder can work with the same + * scaling). + */ + DRM_ERROR("Scaling not consistent across encoder.\n"); + return false; + } + } + } + if (radeon_crtc->rmx_type != RMX_OFF) { + fixed20_12 a, b; + a.full = dfixed_const(src_v); + b.full = dfixed_const(dst_v); + radeon_crtc->vsc.full = dfixed_div(a, b); + a.full = dfixed_const(src_h); + b.full = dfixed_const(dst_h); + radeon_crtc->hsc.full = dfixed_div(a, b); + } else { + radeon_crtc->vsc.full = dfixed_const(1); + radeon_crtc->hsc.full = dfixed_const(1); + } + return true; +} + +/* + * Retrieve current video scanout position of crtc on a given gpu. + * + * \param dev Device to query. + * \param crtc Crtc to query. + * \param *vpos Location where vertical scanout position should be stored. + * \param *hpos Location where horizontal scanout position should go. + * + * Returns vpos as a positive number while in active scanout area. + * Returns vpos as a negative number inside vblank, counting the number + * of scanlines to go until end of vblank, e.g., -1 means "one scanline + * until start of active scanout / end of vblank." + * + * \return Flags, or'ed together as follows: + * + * DRM_SCANOUTPOS_VALID = Query successful. + * DRM_SCANOUTPOS_INVBL = Inside vblank. + * DRM_SCANOUTPOS_ACCURATE = Returned position is accurate. A lack of + * this flag means that returned position may be offset by a constant but + * unknown small number of scanlines wrt. real scanout position. + * + */ +int radeon_get_crtc_scanoutpos(struct drm_device *dev, int crtc, int *vpos, int *hpos) +{ + u32 stat_crtc = 0, vbl = 0, position = 0; + int vbl_start, vbl_end, vtotal, ret = 0; + bool in_vbl = true; + + struct radeon_device *rdev = dev->dev_private; + + if (ASIC_IS_DCE4(rdev)) { + if (crtc == 0) { + vbl = RREG32(EVERGREEN_CRTC_V_BLANK_START_END + + EVERGREEN_CRTC0_REGISTER_OFFSET); + position = RREG32(EVERGREEN_CRTC_STATUS_POSITION + + EVERGREEN_CRTC0_REGISTER_OFFSET); + ret |= DRM_SCANOUTPOS_VALID; + } + if (crtc == 1) { + vbl = RREG32(EVERGREEN_CRTC_V_BLANK_START_END + + EVERGREEN_CRTC1_REGISTER_OFFSET); + position = RREG32(EVERGREEN_CRTC_STATUS_POSITION + + EVERGREEN_CRTC1_REGISTER_OFFSET); + ret |= DRM_SCANOUTPOS_VALID; + } + if (crtc == 2) { + vbl = RREG32(EVERGREEN_CRTC_V_BLANK_START_END + + EVERGREEN_CRTC2_REGISTER_OFFSET); + position = RREG32(EVERGREEN_CRTC_STATUS_POSITION + + EVERGREEN_CRTC2_REGISTER_OFFSET); + ret |= DRM_SCANOUTPOS_VALID; + } + if (crtc == 3) { + vbl = RREG32(EVERGREEN_CRTC_V_BLANK_START_END + + EVERGREEN_CRTC3_REGISTER_OFFSET); + position = RREG32(EVERGREEN_CRTC_STATUS_POSITION + + EVERGREEN_CRTC3_REGISTER_OFFSET); + ret |= DRM_SCANOUTPOS_VALID; + } + if (crtc == 4) { + vbl = RREG32(EVERGREEN_CRTC_V_BLANK_START_END + + EVERGREEN_CRTC4_REGISTER_OFFSET); + position = RREG32(EVERGREEN_CRTC_STATUS_POSITION + + EVERGREEN_CRTC4_REGISTER_OFFSET); + ret |= DRM_SCANOUTPOS_VALID; + } + if (crtc == 5) { + vbl = RREG32(EVERGREEN_CRTC_V_BLANK_START_END + + EVERGREEN_CRTC5_REGISTER_OFFSET); + position = RREG32(EVERGREEN_CRTC_STATUS_POSITION + + EVERGREEN_CRTC5_REGISTER_OFFSET); + ret |= DRM_SCANOUTPOS_VALID; + } + } else if (ASIC_IS_AVIVO(rdev)) { + if (crtc == 0) { + vbl = RREG32(AVIVO_D1CRTC_V_BLANK_START_END); + position = RREG32(AVIVO_D1CRTC_STATUS_POSITION); + ret |= DRM_SCANOUTPOS_VALID; + } + if (crtc == 1) { + vbl = RREG32(AVIVO_D2CRTC_V_BLANK_START_END); + position = RREG32(AVIVO_D2CRTC_STATUS_POSITION); + ret |= DRM_SCANOUTPOS_VALID; + } + } else { + /* Pre-AVIVO: Different encoding of scanout pos and vblank interval. */ + if (crtc == 0) { + /* Assume vbl_end == 0, get vbl_start from + * upper 16 bits. + */ + vbl = (RREG32(RADEON_CRTC_V_TOTAL_DISP) & + RADEON_CRTC_V_DISP) >> RADEON_CRTC_V_DISP_SHIFT; + /* Only retrieve vpos from upper 16 bits, set hpos == 0. */ + position = (RREG32(RADEON_CRTC_VLINE_CRNT_VLINE) >> 16) & RADEON_CRTC_V_TOTAL; + stat_crtc = RREG32(RADEON_CRTC_STATUS); + if (!(stat_crtc & 1)) + in_vbl = false; + + ret |= DRM_SCANOUTPOS_VALID; + } + if (crtc == 1) { + vbl = (RREG32(RADEON_CRTC2_V_TOTAL_DISP) & + RADEON_CRTC_V_DISP) >> RADEON_CRTC_V_DISP_SHIFT; + position = (RREG32(RADEON_CRTC2_VLINE_CRNT_VLINE) >> 16) & RADEON_CRTC_V_TOTAL; + stat_crtc = RREG32(RADEON_CRTC2_STATUS); + if (!(stat_crtc & 1)) + in_vbl = false; + + ret |= DRM_SCANOUTPOS_VALID; + } + } + + /* Decode into vertical and horizontal scanout position. */ + *vpos = position & 0x1fff; + *hpos = (position >> 16) & 0x1fff; + + /* Valid vblank area boundaries from gpu retrieved? */ + if (vbl > 0) { + /* Yes: Decode. */ + ret |= DRM_SCANOUTPOS_ACCURATE; + vbl_start = vbl & 0x1fff; + vbl_end = (vbl >> 16) & 0x1fff; + } + else { + /* No: Fake something reasonable which gives at least ok results. */ + vbl_start = rdev->mode_info.crtcs[crtc]->base.hwmode.crtc_vdisplay; + vbl_end = 0; + } + + /* Test scanout position against vblank region. */ + if ((*vpos < vbl_start) && (*vpos >= vbl_end)) + in_vbl = false; + + /* Check if inside vblank area and apply corrective offsets: + * vpos will then be >=0 in video scanout area, but negative + * within vblank area, counting down the number of lines until + * start of scanout. + */ + + /* Inside "upper part" of vblank area? Apply corrective offset if so: */ + if (in_vbl && (*vpos >= vbl_start)) { + vtotal = rdev->mode_info.crtcs[crtc]->base.hwmode.crtc_vtotal; + *vpos = *vpos - vtotal; + } + + /* Correct for shifted end of vbl at vbl_end. */ + *vpos = *vpos - vbl_end; + + /* In vblank? */ + if (in_vbl) + ret |= DRM_SCANOUTPOS_INVBL; + + return ret; +} diff --git a/drivers/gpu/drm/radeon/radeon_drv.c b/drivers/gpu/drm/radeon/radeon_drv.c new file mode 100644 index 00000000..60e16057 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_drv.c @@ -0,0 +1,426 @@ +/** + * \file radeon_drv.c + * ATI Radeon driver + * + * \author Gareth Hughes + */ + +/* + * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * VA LINUX SYSTEMS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#include "drmP.h" +#include "drm.h" +#include "radeon_drm.h" +#include "radeon_drv.h" + +#include "drm_pciids.h" +#include + + +/* + * KMS wrapper. + * - 2.0.0 - initial interface + * - 2.1.0 - add square tiling interface + * - 2.2.0 - add r6xx/r7xx const buffer support + * - 2.3.0 - add MSPOS + 3D texture + r500 VAP regs + * - 2.4.0 - add crtc id query + * - 2.5.0 - add get accel 2 to work around ddx breakage for evergreen + * - 2.6.0 - add tiling config query (r6xx+), add initial HiZ support (r300->r500) + * 2.7.0 - fixups for r600 2D tiling support. (no external ABI change), add eg dyn gpr regs + * 2.8.0 - pageflip support, r500 US_FORMAT regs. r500 ARGB2101010 colorbuf, r300->r500 CMASK, clock crystal query + * 2.9.0 - r600 tiling (s3tc,rgtc) working, SET_PREDICATION packet 3 on r600 + eg, backend query + * 2.10.0 - fusion 2D tiling + */ +#define KMS_DRIVER_MAJOR 2 +#define KMS_DRIVER_MINOR 10 +#define KMS_DRIVER_PATCHLEVEL 0 +int radeon_driver_load_kms(struct drm_device *dev, unsigned long flags); +int radeon_driver_unload_kms(struct drm_device *dev); +int radeon_driver_firstopen_kms(struct drm_device *dev); +void radeon_driver_lastclose_kms(struct drm_device *dev); +int radeon_driver_open_kms(struct drm_device *dev, struct drm_file *file_priv); +void radeon_driver_postclose_kms(struct drm_device *dev, + struct drm_file *file_priv); +void radeon_driver_preclose_kms(struct drm_device *dev, + struct drm_file *file_priv); +int radeon_suspend_kms(struct drm_device *dev, pm_message_t state); +int radeon_resume_kms(struct drm_device *dev); +u32 radeon_get_vblank_counter_kms(struct drm_device *dev, int crtc); +int radeon_enable_vblank_kms(struct drm_device *dev, int crtc); +void radeon_disable_vblank_kms(struct drm_device *dev, int crtc); +int radeon_get_vblank_timestamp_kms(struct drm_device *dev, int crtc, + int *max_error, + struct timeval *vblank_time, + unsigned flags); +void radeon_driver_irq_preinstall_kms(struct drm_device *dev); +int radeon_driver_irq_postinstall_kms(struct drm_device *dev); +void radeon_driver_irq_uninstall_kms(struct drm_device *dev); +irqreturn_t radeon_driver_irq_handler_kms(DRM_IRQ_ARGS); +int radeon_dma_ioctl_kms(struct drm_device *dev, void *data, + struct drm_file *file_priv); +int radeon_gem_object_init(struct drm_gem_object *obj); +void radeon_gem_object_free(struct drm_gem_object *obj); +extern int radeon_get_crtc_scanoutpos(struct drm_device *dev, int crtc, + int *vpos, int *hpos); +extern struct drm_ioctl_desc radeon_ioctls_kms[]; +extern int radeon_max_kms_ioctl; +int radeon_mmap(struct file *filp, struct vm_area_struct *vma); +int radeon_mode_dumb_mmap(struct drm_file *filp, + struct drm_device *dev, + uint32_t handle, uint64_t *offset_p); +int radeon_mode_dumb_create(struct drm_file *file_priv, + struct drm_device *dev, + struct drm_mode_create_dumb *args); +int radeon_mode_dumb_destroy(struct drm_file *file_priv, + struct drm_device *dev, + uint32_t handle); + +#if defined(CONFIG_DEBUG_FS) +int radeon_debugfs_init(struct drm_minor *minor); +void radeon_debugfs_cleanup(struct drm_minor *minor); +#endif + + +int radeon_no_wb; +int radeon_modeset = -1; +int radeon_dynclks = -1; +int radeon_r4xx_atom = 0; +int radeon_agpmode = 0; +int radeon_vram_limit = 0; +int radeon_gart_size = 512; /* default gart size */ +int radeon_benchmarking = 0; +int radeon_testing = 0; +int radeon_connector_table = 0; +int radeon_tv = 1; +int radeon_audio = 0; +int radeon_disp_priority = 0; +int radeon_hw_i2c = 0; +int radeon_pcie_gen2 = 0; +int radeon_msi = -1; + +MODULE_PARM_DESC(no_wb, "Disable AGP writeback for scratch registers"); +module_param_named(no_wb, radeon_no_wb, int, 0444); + +MODULE_PARM_DESC(modeset, "Disable/Enable modesetting"); +module_param_named(modeset, radeon_modeset, int, 0400); + +MODULE_PARM_DESC(dynclks, "Disable/Enable dynamic clocks"); +module_param_named(dynclks, radeon_dynclks, int, 0444); + +MODULE_PARM_DESC(r4xx_atom, "Enable ATOMBIOS modesetting for R4xx"); +module_param_named(r4xx_atom, radeon_r4xx_atom, int, 0444); + +MODULE_PARM_DESC(vramlimit, "Restrict VRAM for testing"); +module_param_named(vramlimit, radeon_vram_limit, int, 0600); + +MODULE_PARM_DESC(agpmode, "AGP Mode (-1 == PCI)"); +module_param_named(agpmode, radeon_agpmode, int, 0444); + +MODULE_PARM_DESC(gartsize, "Size of PCIE/IGP gart to setup in megabytes (32,64, etc)\n"); +module_param_named(gartsize, radeon_gart_size, int, 0600); + +MODULE_PARM_DESC(benchmark, "Run benchmark"); +module_param_named(benchmark, radeon_benchmarking, int, 0444); + +MODULE_PARM_DESC(test, "Run tests"); +module_param_named(test, radeon_testing, int, 0444); + +MODULE_PARM_DESC(connector_table, "Force connector table"); +module_param_named(connector_table, radeon_connector_table, int, 0444); + +MODULE_PARM_DESC(tv, "TV enable (0 = disable)"); +module_param_named(tv, radeon_tv, int, 0444); + +MODULE_PARM_DESC(audio, "Audio enable (1 = enable)"); +module_param_named(audio, radeon_audio, int, 0444); + +MODULE_PARM_DESC(disp_priority, "Display Priority (0 = auto, 1 = normal, 2 = high)"); +module_param_named(disp_priority, radeon_disp_priority, int, 0444); + +MODULE_PARM_DESC(hw_i2c, "hw i2c engine enable (0 = disable)"); +module_param_named(hw_i2c, radeon_hw_i2c, int, 0444); + +MODULE_PARM_DESC(pcie_gen2, "PCIE Gen2 mode (1 = enable)"); +module_param_named(pcie_gen2, radeon_pcie_gen2, int, 0444); + +MODULE_PARM_DESC(msi, "MSI support (1 = enable, 0 = disable, -1 = auto)"); +module_param_named(msi, radeon_msi, int, 0444); + +static int radeon_suspend(struct drm_device *dev, pm_message_t state) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) + return 0; + + /* Disable *all* interrupts */ + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_RS600) + RADEON_WRITE(R500_DxMODE_INT_MASK, 0); + RADEON_WRITE(RADEON_GEN_INT_CNTL, 0); + return 0; +} + +static int radeon_resume(struct drm_device *dev) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) + return 0; + + /* Restore interrupt registers */ + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_RS600) + RADEON_WRITE(R500_DxMODE_INT_MASK, dev_priv->r500_disp_irq_reg); + RADEON_WRITE(RADEON_GEN_INT_CNTL, dev_priv->irq_enable_reg); + return 0; +} + +static struct pci_device_id pciidlist[] = { + radeon_PCI_IDS +}; + +#if defined(CONFIG_DRM_RADEON_KMS) +MODULE_DEVICE_TABLE(pci, pciidlist); +#endif + +static struct drm_driver driver_old = { + .driver_features = + DRIVER_USE_AGP | DRIVER_USE_MTRR | DRIVER_PCI_DMA | DRIVER_SG | + DRIVER_HAVE_IRQ | DRIVER_HAVE_DMA | DRIVER_IRQ_SHARED, + .dev_priv_size = sizeof(drm_radeon_buf_priv_t), + .load = radeon_driver_load, + .firstopen = radeon_driver_firstopen, + .open = radeon_driver_open, + .preclose = radeon_driver_preclose, + .postclose = radeon_driver_postclose, + .lastclose = radeon_driver_lastclose, + .unload = radeon_driver_unload, + .suspend = radeon_suspend, + .resume = radeon_resume, + .get_vblank_counter = radeon_get_vblank_counter, + .enable_vblank = radeon_enable_vblank, + .disable_vblank = radeon_disable_vblank, + .master_create = radeon_master_create, + .master_destroy = radeon_master_destroy, + .irq_preinstall = radeon_driver_irq_preinstall, + .irq_postinstall = radeon_driver_irq_postinstall, + .irq_uninstall = radeon_driver_irq_uninstall, + .irq_handler = radeon_driver_irq_handler, + .reclaim_buffers = drm_core_reclaim_buffers, + .ioctls = radeon_ioctls, + .dma_ioctl = radeon_cp_buffers, + .fops = { + .owner = THIS_MODULE, + .open = drm_open, + .release = drm_release, + .unlocked_ioctl = drm_ioctl, + .mmap = drm_mmap, + .poll = drm_poll, + .fasync = drm_fasync, + .read = drm_read, +#ifdef CONFIG_COMPAT + .compat_ioctl = radeon_compat_ioctl, +#endif + .llseek = noop_llseek, + }, + + .name = DRIVER_NAME, + .desc = DRIVER_DESC, + .date = DRIVER_DATE, + .major = DRIVER_MAJOR, + .minor = DRIVER_MINOR, + .patchlevel = DRIVER_PATCHLEVEL, +}; + +static struct drm_driver kms_driver; + +static void radeon_kick_out_firmware_fb(struct pci_dev *pdev) +{ + struct apertures_struct *ap; + bool primary = false; + + ap = alloc_apertures(1); + ap->ranges[0].base = pci_resource_start(pdev, 0); + ap->ranges[0].size = pci_resource_len(pdev, 0); + +#ifdef CONFIG_X86 + primary = pdev->resource[PCI_ROM_RESOURCE].flags & IORESOURCE_ROM_SHADOW; +#endif + remove_conflicting_framebuffers(ap, "radeondrmfb", primary); + kfree(ap); +} + +static int __devinit +radeon_pci_probe(struct pci_dev *pdev, const struct pci_device_id *ent) +{ + /* Get rid of things like offb */ + radeon_kick_out_firmware_fb(pdev); + + return drm_get_pci_dev(pdev, ent, &kms_driver); +} + +static void +radeon_pci_remove(struct pci_dev *pdev) +{ + struct drm_device *dev = pci_get_drvdata(pdev); + + drm_put_dev(dev); +} + +static int +radeon_pci_suspend(struct pci_dev *pdev, pm_message_t state) +{ + struct drm_device *dev = pci_get_drvdata(pdev); + return radeon_suspend_kms(dev, state); +} + +static int +radeon_pci_resume(struct pci_dev *pdev) +{ + struct drm_device *dev = pci_get_drvdata(pdev); + return radeon_resume_kms(dev); +} + +static struct drm_driver kms_driver = { + .driver_features = + DRIVER_USE_AGP | DRIVER_USE_MTRR | DRIVER_PCI_DMA | DRIVER_SG | + DRIVER_HAVE_IRQ | DRIVER_HAVE_DMA | DRIVER_IRQ_SHARED | DRIVER_GEM, + .dev_priv_size = 0, + .load = radeon_driver_load_kms, + .firstopen = radeon_driver_firstopen_kms, + .open = radeon_driver_open_kms, + .preclose = radeon_driver_preclose_kms, + .postclose = radeon_driver_postclose_kms, + .lastclose = radeon_driver_lastclose_kms, + .unload = radeon_driver_unload_kms, + .suspend = radeon_suspend_kms, + .resume = radeon_resume_kms, + .get_vblank_counter = radeon_get_vblank_counter_kms, + .enable_vblank = radeon_enable_vblank_kms, + .disable_vblank = radeon_disable_vblank_kms, + .get_vblank_timestamp = radeon_get_vblank_timestamp_kms, + .get_scanout_position = radeon_get_crtc_scanoutpos, +#if defined(CONFIG_DEBUG_FS) + .debugfs_init = radeon_debugfs_init, + .debugfs_cleanup = radeon_debugfs_cleanup, +#endif + .irq_preinstall = radeon_driver_irq_preinstall_kms, + .irq_postinstall = radeon_driver_irq_postinstall_kms, + .irq_uninstall = radeon_driver_irq_uninstall_kms, + .irq_handler = radeon_driver_irq_handler_kms, + .reclaim_buffers = drm_core_reclaim_buffers, + .ioctls = radeon_ioctls_kms, + .gem_init_object = radeon_gem_object_init, + .gem_free_object = radeon_gem_object_free, + .dma_ioctl = radeon_dma_ioctl_kms, + .dumb_create = radeon_mode_dumb_create, + .dumb_map_offset = radeon_mode_dumb_mmap, + .dumb_destroy = radeon_mode_dumb_destroy, + .fops = { + .owner = THIS_MODULE, + .open = drm_open, + .release = drm_release, + .unlocked_ioctl = drm_ioctl, + .mmap = radeon_mmap, + .poll = drm_poll, + .fasync = drm_fasync, + .read = drm_read, +#ifdef CONFIG_COMPAT + .compat_ioctl = radeon_kms_compat_ioctl, +#endif + }, + + .name = DRIVER_NAME, + .desc = DRIVER_DESC, + .date = DRIVER_DATE, + .major = KMS_DRIVER_MAJOR, + .minor = KMS_DRIVER_MINOR, + .patchlevel = KMS_DRIVER_PATCHLEVEL, +}; + +static struct drm_driver *driver; +static struct pci_driver *pdriver; + +static struct pci_driver radeon_pci_driver = { + .name = DRIVER_NAME, + .id_table = pciidlist, +}; + +static struct pci_driver radeon_kms_pci_driver = { + .name = DRIVER_NAME, + .id_table = pciidlist, + .probe = radeon_pci_probe, + .remove = radeon_pci_remove, + .suspend = radeon_pci_suspend, + .resume = radeon_pci_resume, +}; + +static int __init radeon_init(void) +{ + driver = &driver_old; + pdriver = &radeon_pci_driver; + driver->num_ioctls = radeon_max_ioctl; +#ifdef CONFIG_VGA_CONSOLE + if (vgacon_text_force() && radeon_modeset == -1) { + DRM_INFO("VGACON disable radeon kernel modesetting.\n"); + driver = &driver_old; + pdriver = &radeon_pci_driver; + driver->driver_features &= ~DRIVER_MODESET; + radeon_modeset = 0; + } +#endif + /* if enabled by default */ + if (radeon_modeset == -1) { +#ifdef CONFIG_DRM_RADEON_KMS + DRM_INFO("radeon defaulting to kernel modesetting.\n"); + radeon_modeset = 1; +#else + DRM_INFO("radeon defaulting to userspace modesetting.\n"); + radeon_modeset = 0; +#endif + } + if (radeon_modeset == 1) { + DRM_INFO("radeon kernel modesetting enabled.\n"); + driver = &kms_driver; + pdriver = &radeon_kms_pci_driver; + driver->driver_features |= DRIVER_MODESET; + driver->num_ioctls = radeon_max_kms_ioctl; + radeon_register_atpx_handler(); + } + /* if the vga console setting is enabled still + * let modprobe override it */ + return drm_pci_init(driver, pdriver); +} + +static void __exit radeon_exit(void) +{ + drm_pci_exit(driver, pdriver); + radeon_unregister_atpx_handler(); +} + +module_init(radeon_init); +module_exit(radeon_exit); + +MODULE_AUTHOR(DRIVER_AUTHOR); +MODULE_DESCRIPTION(DRIVER_DESC); +MODULE_LICENSE("GPL and additional rights"); diff --git a/drivers/gpu/drm/radeon/radeon_drv.h b/drivers/gpu/drm/radeon/radeon_drv.h new file mode 100644 index 00000000..a1b59ca9 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_drv.h @@ -0,0 +1,2171 @@ +/* radeon_drv.h -- Private header for radeon driver -*- linux-c -*- + * + * Copyright 1999 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Fremont, California. + * All rights reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Kevin E. Martin + * Gareth Hughes + */ + +#ifndef __RADEON_DRV_H__ +#define __RADEON_DRV_H__ + +#include +#include + +#include "radeon_family.h" + +/* General customization: + */ + +#define DRIVER_AUTHOR "Gareth Hughes, Keith Whitwell, others." + +#define DRIVER_NAME "radeon" +#define DRIVER_DESC "ATI Radeon" +#define DRIVER_DATE "20080528" + +/* Interface history: + * + * 1.1 - ?? + * 1.2 - Add vertex2 ioctl (keith) + * - Add stencil capability to clear ioctl (gareth, keith) + * - Increase MAX_TEXTURE_LEVELS (brian) + * 1.3 - Add cmdbuf ioctl (keith) + * - Add support for new radeon packets (keith) + * - Add getparam ioctl (keith) + * - Add flip-buffers ioctl, deprecate fullscreen foo (keith). + * 1.4 - Add scratch registers to get_param ioctl. + * 1.5 - Add r200 packets to cmdbuf ioctl + * - Add r200 function to init ioctl + * - Add 'scalar2' instruction to cmdbuf + * 1.6 - Add static GART memory manager + * Add irq handler (won't be turned on unless X server knows to) + * Add irq ioctls and irq_active getparam. + * Add wait command for cmdbuf ioctl + * Add GART offset query for getparam + * 1.7 - Add support for cube map registers: R200_PP_CUBIC_FACES_[0..5] + * and R200_PP_CUBIC_OFFSET_F1_[0..5]. + * Added packets R200_EMIT_PP_CUBIC_FACES_[0..5] and + * R200_EMIT_PP_CUBIC_OFFSETS_[0..5]. (brian) + * 1.8 - Remove need to call cleanup ioctls on last client exit (keith) + * Add 'GET' queries for starting additional clients on different VT's. + * 1.9 - Add DRM_IOCTL_RADEON_CP_RESUME ioctl. + * Add texture rectangle support for r100. + * 1.10- Add SETPARAM ioctl; first parameter to set is FB_LOCATION, which + * clients use to tell the DRM where they think the framebuffer is + * located in the card's address space + * 1.11- Add packet R200_EMIT_RB3D_BLENDCOLOR to support GL_EXT_blend_color + * and GL_EXT_blend_[func|equation]_separate on r200 + * 1.12- Add R300 CP microcode support - this just loads the CP on r300 + * (No 3D support yet - just microcode loading). + * 1.13- Add packet R200_EMIT_TCL_POINT_SPRITE_CNTL for ARB_point_parameters + * - Add hyperz support, add hyperz flags to clear ioctl. + * 1.14- Add support for color tiling + * - Add R100/R200 surface allocation/free support + * 1.15- Add support for texture micro tiling + * - Add support for r100 cube maps + * 1.16- Add R200_EMIT_PP_TRI_PERF_CNTL packet to support brilinear + * texture filtering on r200 + * 1.17- Add initial support for R300 (3D). + * 1.18- Add support for GL_ATI_fragment_shader, new packets + * R200_EMIT_PP_AFS_0/1, R200_EMIT_PP_TXCTLALL_0-5 (replaces + * R200_EMIT_PP_TXFILTER_0-5, 2 more regs) and R200_EMIT_ATF_TFACTOR + * (replaces R200_EMIT_TFACTOR_0 (8 consts instead of 6) + * 1.19- Add support for gart table in FB memory and PCIE r300 + * 1.20- Add support for r300 texrect + * 1.21- Add support for card type getparam + * 1.22- Add support for texture cache flushes (R300_TX_CNTL) + * 1.23- Add new radeon memory map work from benh + * 1.24- Add general-purpose packet for manipulating scratch registers (r300) + * 1.25- Add support for r200 vertex programs (R200_EMIT_VAP_PVS_CNTL, + * new packet type) + * 1.26- Add support for variable size PCI(E) gart aperture + * 1.27- Add support for IGP GART + * 1.28- Add support for VBL on CRTC2 + * 1.29- R500 3D cmd buffer support + * 1.30- Add support for occlusion queries + * 1.31- Add support for num Z pipes from GET_PARAM + * 1.32- fixes for rv740 setup + * 1.33- Add r6xx/r7xx const buffer support + */ +#define DRIVER_MAJOR 1 +#define DRIVER_MINOR 33 +#define DRIVER_PATCHLEVEL 0 + +enum radeon_cp_microcode_version { + UCODE_R100, + UCODE_R200, + UCODE_R300, +}; + +typedef struct drm_radeon_freelist { + unsigned int age; + struct drm_buf *buf; + struct drm_radeon_freelist *next; + struct drm_radeon_freelist *prev; +} drm_radeon_freelist_t; + +typedef struct drm_radeon_ring_buffer { + u32 *start; + u32 *end; + int size; + int size_l2qw; + + int rptr_update; /* Double Words */ + int rptr_update_l2qw; /* log2 Quad Words */ + + int fetch_size; /* Double Words */ + int fetch_size_l2ow; /* log2 Oct Words */ + + u32 tail; + u32 tail_mask; + int space; + + int high_mark; +} drm_radeon_ring_buffer_t; + +typedef struct drm_radeon_depth_clear_t { + u32 rb3d_cntl; + u32 rb3d_zstencilcntl; + u32 se_cntl; +} drm_radeon_depth_clear_t; + +struct drm_radeon_driver_file_fields { + int64_t radeon_fb_delta; +}; + +struct mem_block { + struct mem_block *next; + struct mem_block *prev; + int start; + int size; + struct drm_file *file_priv; /* NULL: free, -1: heap, other: real files */ +}; + +struct radeon_surface { + int refcount; + u32 lower; + u32 upper; + u32 flags; +}; + +struct radeon_virt_surface { + int surface_index; + u32 lower; + u32 upper; + u32 flags; + struct drm_file *file_priv; +#define PCIGART_FILE_PRIV ((void *) -1L) +}; + +#define RADEON_FLUSH_EMITED (1 << 0) +#define RADEON_PURGE_EMITED (1 << 1) + +struct drm_radeon_master_private { + drm_local_map_t *sarea; + drm_radeon_sarea_t *sarea_priv; +}; + +typedef struct drm_radeon_private { + drm_radeon_ring_buffer_t ring; + + u32 fb_location; + u32 fb_size; + int new_memmap; + + int gart_size; + u32 gart_vm_start; + unsigned long gart_buffers_offset; + + int cp_mode; + int cp_running; + + drm_radeon_freelist_t *head; + drm_radeon_freelist_t *tail; + int last_buf; + int writeback_works; + + int usec_timeout; + + int microcode_version; + + struct { + u32 boxes; + int freelist_timeouts; + int freelist_loops; + int requested_bufs; + int last_frame_reads; + int last_clear_reads; + int clears; + int texture_uploads; + } stats; + + int do_boxes; + int page_flipping; + + u32 color_fmt; + unsigned int front_offset; + unsigned int front_pitch; + unsigned int back_offset; + unsigned int back_pitch; + + u32 depth_fmt; + unsigned int depth_offset; + unsigned int depth_pitch; + + u32 front_pitch_offset; + u32 back_pitch_offset; + u32 depth_pitch_offset; + + drm_radeon_depth_clear_t depth_clear; + + unsigned long ring_offset; + unsigned long ring_rptr_offset; + unsigned long buffers_offset; + unsigned long gart_textures_offset; + + drm_local_map_t *sarea; + drm_local_map_t *cp_ring; + drm_local_map_t *ring_rptr; + drm_local_map_t *gart_textures; + + struct mem_block *gart_heap; + struct mem_block *fb_heap; + + /* SW interrupt */ + wait_queue_head_t swi_queue; + atomic_t swi_emitted; + int vblank_crtc; + uint32_t irq_enable_reg; + uint32_t r500_disp_irq_reg; + + struct radeon_surface surfaces[RADEON_MAX_SURFACES]; + struct radeon_virt_surface virt_surfaces[2 * RADEON_MAX_SURFACES]; + + unsigned long pcigart_offset; + unsigned int pcigart_offset_set; + struct drm_ati_pcigart_info gart_info; + + u32 scratch_ages[5]; + + int have_z_offset; + + /* starting from here on, data is preserved across an open */ + uint32_t flags; /* see radeon_chip_flags */ + resource_size_t fb_aper_offset; + + int num_gb_pipes; + int num_z_pipes; + int track_flush; + drm_local_map_t *mmio; + + /* r6xx/r7xx pipe/shader config */ + int r600_max_pipes; + int r600_max_tile_pipes; + int r600_max_simds; + int r600_max_backends; + int r600_max_gprs; + int r600_max_threads; + int r600_max_stack_entries; + int r600_max_hw_contexts; + int r600_max_gs_threads; + int r600_sx_max_export_size; + int r600_sx_max_export_pos_size; + int r600_sx_max_export_smx_size; + int r600_sq_num_cf_insts; + int r700_sx_num_of_sets; + int r700_sc_prim_fifo_size; + int r700_sc_hiz_tile_fifo_size; + int r700_sc_earlyz_tile_fifo_fize; + int r600_group_size; + int r600_npipes; + int r600_nbanks; + + struct mutex cs_mutex; + u32 cs_id_scnt; + u32 cs_id_wcnt; + /* r6xx/r7xx drm blit vertex buffer */ + struct drm_buf *blit_vb; + + /* firmware */ + const struct firmware *me_fw, *pfp_fw; +} drm_radeon_private_t; + +typedef struct drm_radeon_buf_priv { + u32 age; +} drm_radeon_buf_priv_t; + +struct drm_buffer; + +typedef struct drm_radeon_kcmd_buffer { + int bufsz; + struct drm_buffer *buffer; + int nbox; + struct drm_clip_rect __user *boxes; +} drm_radeon_kcmd_buffer_t; + +extern int radeon_no_wb; +extern struct drm_ioctl_desc radeon_ioctls[]; +extern int radeon_max_ioctl; + +extern u32 radeon_get_ring_head(drm_radeon_private_t *dev_priv); +extern void radeon_set_ring_head(drm_radeon_private_t *dev_priv, u32 val); + +#define GET_RING_HEAD(dev_priv) radeon_get_ring_head(dev_priv) +#define SET_RING_HEAD(dev_priv, val) radeon_set_ring_head(dev_priv, val) + +/* Check whether the given hardware address is inside the framebuffer or the + * GART area. + */ +static __inline__ int radeon_check_offset(drm_radeon_private_t *dev_priv, + u64 off) +{ + u32 fb_start = dev_priv->fb_location; + u32 fb_end = fb_start + dev_priv->fb_size - 1; + u32 gart_start = dev_priv->gart_vm_start; + u32 gart_end = gart_start + dev_priv->gart_size - 1; + + return ((off >= fb_start && off <= fb_end) || + (off >= gart_start && off <= gart_end)); +} + +/* radeon_state.c */ +extern void radeon_cp_discard_buffer(struct drm_device *dev, struct drm_master *master, struct drm_buf *buf); + + /* radeon_cp.c */ +extern int radeon_cp_init(struct drm_device *dev, void *data, struct drm_file *file_priv); +extern int radeon_cp_start(struct drm_device *dev, void *data, struct drm_file *file_priv); +extern int radeon_cp_stop(struct drm_device *dev, void *data, struct drm_file *file_priv); +extern int radeon_cp_reset(struct drm_device *dev, void *data, struct drm_file *file_priv); +extern int radeon_cp_idle(struct drm_device *dev, void *data, struct drm_file *file_priv); +extern int radeon_cp_resume(struct drm_device *dev, void *data, struct drm_file *file_priv); +extern int radeon_engine_reset(struct drm_device *dev, void *data, struct drm_file *file_priv); +extern int radeon_fullscreen(struct drm_device *dev, void *data, struct drm_file *file_priv); +extern int radeon_cp_buffers(struct drm_device *dev, void *data, struct drm_file *file_priv); +extern u32 radeon_read_fb_location(drm_radeon_private_t *dev_priv); +extern void radeon_write_agp_location(drm_radeon_private_t *dev_priv, u32 agp_loc); +extern void radeon_write_agp_base(drm_radeon_private_t *dev_priv, u64 agp_base); +extern u32 RADEON_READ_MM(drm_radeon_private_t *dev_priv, int addr); + +extern void radeon_freelist_reset(struct drm_device * dev); +extern struct drm_buf *radeon_freelist_get(struct drm_device * dev); + +extern int radeon_wait_ring(drm_radeon_private_t * dev_priv, int n); + +extern int radeon_do_cp_idle(drm_radeon_private_t * dev_priv); + +extern int radeon_driver_preinit(struct drm_device *dev, unsigned long flags); +extern int radeon_presetup(struct drm_device *dev); +extern int radeon_driver_postcleanup(struct drm_device *dev); + +extern int radeon_mem_alloc(struct drm_device *dev, void *data, struct drm_file *file_priv); +extern int radeon_mem_free(struct drm_device *dev, void *data, struct drm_file *file_priv); +extern int radeon_mem_init_heap(struct drm_device *dev, void *data, struct drm_file *file_priv); +extern void radeon_mem_takedown(struct mem_block **heap); +extern void radeon_mem_release(struct drm_file *file_priv, + struct mem_block *heap); + +extern void radeon_enable_bm(struct drm_radeon_private *dev_priv); +extern u32 radeon_read_ring_rptr(drm_radeon_private_t *dev_priv, u32 off); +extern void radeon_write_ring_rptr(drm_radeon_private_t *dev_priv, u32 off, u32 val); + + /* radeon_irq.c */ +extern void radeon_irq_set_state(struct drm_device *dev, u32 mask, int state); +extern int radeon_irq_emit(struct drm_device *dev, void *data, struct drm_file *file_priv); +extern int radeon_irq_wait(struct drm_device *dev, void *data, struct drm_file *file_priv); + +extern void radeon_do_release(struct drm_device * dev); +extern u32 radeon_get_vblank_counter(struct drm_device *dev, int crtc); +extern int radeon_enable_vblank(struct drm_device *dev, int crtc); +extern void radeon_disable_vblank(struct drm_device *dev, int crtc); +extern irqreturn_t radeon_driver_irq_handler(DRM_IRQ_ARGS); +extern void radeon_driver_irq_preinstall(struct drm_device * dev); +extern int radeon_driver_irq_postinstall(struct drm_device *dev); +extern void radeon_driver_irq_uninstall(struct drm_device * dev); +extern void radeon_enable_interrupt(struct drm_device *dev); +extern int radeon_vblank_crtc_get(struct drm_device *dev); +extern int radeon_vblank_crtc_set(struct drm_device *dev, int64_t value); + +extern int radeon_driver_load(struct drm_device *dev, unsigned long flags); +extern int radeon_driver_unload(struct drm_device *dev); +extern int radeon_driver_firstopen(struct drm_device *dev); +extern void radeon_driver_preclose(struct drm_device *dev, + struct drm_file *file_priv); +extern void radeon_driver_postclose(struct drm_device *dev, + struct drm_file *file_priv); +extern void radeon_driver_lastclose(struct drm_device * dev); +extern int radeon_driver_open(struct drm_device *dev, + struct drm_file *file_priv); +extern long radeon_compat_ioctl(struct file *filp, unsigned int cmd, + unsigned long arg); +extern long radeon_kms_compat_ioctl(struct file *filp, unsigned int cmd, + unsigned long arg); + +extern int radeon_master_create(struct drm_device *dev, struct drm_master *master); +extern void radeon_master_destroy(struct drm_device *dev, struct drm_master *master); +extern void radeon_cp_dispatch_flip(struct drm_device *dev, struct drm_master *master); +/* r300_cmdbuf.c */ +extern void r300_init_reg_flags(struct drm_device *dev); + +extern int r300_do_cp_cmdbuf(struct drm_device *dev, + struct drm_file *file_priv, + drm_radeon_kcmd_buffer_t *cmdbuf); + +/* r600_cp.c */ +extern int r600_do_engine_reset(struct drm_device *dev); +extern int r600_do_cleanup_cp(struct drm_device *dev); +extern int r600_do_init_cp(struct drm_device *dev, drm_radeon_init_t *init, + struct drm_file *file_priv); +extern int r600_do_resume_cp(struct drm_device *dev, struct drm_file *file_priv); +extern int r600_do_cp_idle(drm_radeon_private_t *dev_priv); +extern void r600_do_cp_start(drm_radeon_private_t *dev_priv); +extern void r600_do_cp_reset(drm_radeon_private_t *dev_priv); +extern void r600_do_cp_stop(drm_radeon_private_t *dev_priv); +extern int r600_cp_dispatch_indirect(struct drm_device *dev, + struct drm_buf *buf, int start, int end); +extern int r600_page_table_init(struct drm_device *dev); +extern void r600_page_table_cleanup(struct drm_device *dev, struct drm_ati_pcigart_info *gart_info); +extern int r600_cs_legacy_ioctl(struct drm_device *dev, void *data, struct drm_file *fpriv); +extern void r600_cp_dispatch_swap(struct drm_device *dev, struct drm_file *file_priv); +extern int r600_cp_dispatch_texture(struct drm_device *dev, + struct drm_file *file_priv, + drm_radeon_texture_t *tex, + drm_radeon_tex_image_t *image); +/* r600_blit.c */ +extern int r600_prepare_blit_copy(struct drm_device *dev, struct drm_file *file_priv); +extern void r600_done_blit_copy(struct drm_device *dev); +extern void r600_blit_copy(struct drm_device *dev, + uint64_t src_gpu_addr, uint64_t dst_gpu_addr, + int size_bytes); +extern void r600_blit_swap(struct drm_device *dev, + uint64_t src_gpu_addr, uint64_t dst_gpu_addr, + int sx, int sy, int dx, int dy, + int w, int h, int src_pitch, int dst_pitch, int cpp); + +/* atpx handler */ +#if defined(CONFIG_VGA_SWITCHEROO) +void radeon_register_atpx_handler(void); +void radeon_unregister_atpx_handler(void); +#else +static inline void radeon_register_atpx_handler(void) {} +static inline void radeon_unregister_atpx_handler(void) {} +#endif + +/* Flags for stats.boxes + */ +#define RADEON_BOX_DMA_IDLE 0x1 +#define RADEON_BOX_RING_FULL 0x2 +#define RADEON_BOX_FLIP 0x4 +#define RADEON_BOX_WAIT_IDLE 0x8 +#define RADEON_BOX_TEXTURE_LOAD 0x10 + +/* Register definitions, register access macros and drmAddMap constants + * for Radeon kernel driver. + */ +#define RADEON_MM_INDEX 0x0000 +#define RADEON_MM_DATA 0x0004 + +#define RADEON_AGP_COMMAND 0x0f60 +#define RADEON_AGP_COMMAND_PCI_CONFIG 0x0060 /* offset in PCI config */ +# define RADEON_AGP_ENABLE (1<<8) +#define RADEON_AUX_SCISSOR_CNTL 0x26f0 +# define RADEON_EXCLUSIVE_SCISSOR_0 (1 << 24) +# define RADEON_EXCLUSIVE_SCISSOR_1 (1 << 25) +# define RADEON_EXCLUSIVE_SCISSOR_2 (1 << 26) +# define RADEON_SCISSOR_0_ENABLE (1 << 28) +# define RADEON_SCISSOR_1_ENABLE (1 << 29) +# define RADEON_SCISSOR_2_ENABLE (1 << 30) + +/* + * PCIE radeons (rv370/rv380, rv410, r423/r430/r480, r5xx) + * don't have an explicit bus mastering disable bit. It's handled + * by the PCI D-states. PMI_BM_DIS disables D-state bus master + * handling, not bus mastering itself. + */ +#define RADEON_BUS_CNTL 0x0030 +/* r1xx, r2xx, r300, r(v)350, r420/r481, rs400/rs480 */ +# define RADEON_BUS_MASTER_DIS (1 << 6) +/* rs600/rs690/rs740 */ +# define RS600_BUS_MASTER_DIS (1 << 14) +# define RS600_MSI_REARM (1 << 20) +/* see RS400_MSI_REARM in AIC_CNTL for rs480 */ + +#define RADEON_BUS_CNTL1 0x0034 +# define RADEON_PMI_BM_DIS (1 << 2) +# define RADEON_PMI_INT_DIS (1 << 3) + +#define RV370_BUS_CNTL 0x004c +# define RV370_PMI_BM_DIS (1 << 5) +# define RV370_PMI_INT_DIS (1 << 6) + +#define RADEON_MSI_REARM_EN 0x0160 +/* rv370/rv380, rv410, r423/r430/r480, r5xx */ +# define RV370_MSI_REARM_EN (1 << 0) + +#define RADEON_CLOCK_CNTL_DATA 0x000c +# define RADEON_PLL_WR_EN (1 << 7) +#define RADEON_CLOCK_CNTL_INDEX 0x0008 +#define RADEON_CONFIG_APER_SIZE 0x0108 +#define RADEON_CONFIG_MEMSIZE 0x00f8 +#define RADEON_CRTC_OFFSET 0x0224 +#define RADEON_CRTC_OFFSET_CNTL 0x0228 +# define RADEON_CRTC_TILE_EN (1 << 15) +# define RADEON_CRTC_OFFSET_FLIP_CNTL (1 << 16) +#define RADEON_CRTC2_OFFSET 0x0324 +#define RADEON_CRTC2_OFFSET_CNTL 0x0328 + +#define RADEON_PCIE_INDEX 0x0030 +#define RADEON_PCIE_DATA 0x0034 +#define RADEON_PCIE_TX_GART_CNTL 0x10 +# define RADEON_PCIE_TX_GART_EN (1 << 0) +# define RADEON_PCIE_TX_GART_UNMAPPED_ACCESS_PASS_THRU (0 << 1) +# define RADEON_PCIE_TX_GART_UNMAPPED_ACCESS_CLAMP_LO (1 << 1) +# define RADEON_PCIE_TX_GART_UNMAPPED_ACCESS_DISCARD (3 << 1) +# define RADEON_PCIE_TX_GART_MODE_32_128_CACHE (0 << 3) +# define RADEON_PCIE_TX_GART_MODE_8_4_128_CACHE (1 << 3) +# define RADEON_PCIE_TX_GART_CHK_RW_VALID_EN (1 << 5) +# define RADEON_PCIE_TX_GART_INVALIDATE_TLB (1 << 8) +#define RADEON_PCIE_TX_DISCARD_RD_ADDR_LO 0x11 +#define RADEON_PCIE_TX_DISCARD_RD_ADDR_HI 0x12 +#define RADEON_PCIE_TX_GART_BASE 0x13 +#define RADEON_PCIE_TX_GART_START_LO 0x14 +#define RADEON_PCIE_TX_GART_START_HI 0x15 +#define RADEON_PCIE_TX_GART_END_LO 0x16 +#define RADEON_PCIE_TX_GART_END_HI 0x17 + +#define RS480_NB_MC_INDEX 0x168 +# define RS480_NB_MC_IND_WR_EN (1 << 8) +#define RS480_NB_MC_DATA 0x16c + +#define RS690_MC_INDEX 0x78 +# define RS690_MC_INDEX_MASK 0x1ff +# define RS690_MC_INDEX_WR_EN (1 << 9) +# define RS690_MC_INDEX_WR_ACK 0x7f +#define RS690_MC_DATA 0x7c + +/* MC indirect registers */ +#define RS480_MC_MISC_CNTL 0x18 +# define RS480_DISABLE_GTW (1 << 1) +/* switch between MCIND GART and MM GART registers. 0 = mmgart, 1 = mcind gart */ +# define RS480_GART_INDEX_REG_EN (1 << 12) +# define RS690_BLOCK_GFX_D3_EN (1 << 14) +#define RS480_K8_FB_LOCATION 0x1e +#define RS480_GART_FEATURE_ID 0x2b +# define RS480_HANG_EN (1 << 11) +# define RS480_TLB_ENABLE (1 << 18) +# define RS480_P2P_ENABLE (1 << 19) +# define RS480_GTW_LAC_EN (1 << 25) +# define RS480_2LEVEL_GART (0 << 30) +# define RS480_1LEVEL_GART (1 << 30) +# define RS480_PDC_EN (1 << 31) +#define RS480_GART_BASE 0x2c +#define RS480_GART_CACHE_CNTRL 0x2e +# define RS480_GART_CACHE_INVALIDATE (1 << 0) /* wait for it to clear */ +#define RS480_AGP_ADDRESS_SPACE_SIZE 0x38 +# define RS480_GART_EN (1 << 0) +# define RS480_VA_SIZE_32MB (0 << 1) +# define RS480_VA_SIZE_64MB (1 << 1) +# define RS480_VA_SIZE_128MB (2 << 1) +# define RS480_VA_SIZE_256MB (3 << 1) +# define RS480_VA_SIZE_512MB (4 << 1) +# define RS480_VA_SIZE_1GB (5 << 1) +# define RS480_VA_SIZE_2GB (6 << 1) +#define RS480_AGP_MODE_CNTL 0x39 +# define RS480_POST_GART_Q_SIZE (1 << 18) +# define RS480_NONGART_SNOOP (1 << 19) +# define RS480_AGP_RD_BUF_SIZE (1 << 20) +# define RS480_REQ_TYPE_SNOOP_SHIFT 22 +# define RS480_REQ_TYPE_SNOOP_MASK 0x3 +# define RS480_REQ_TYPE_SNOOP_DIS (1 << 24) +#define RS480_MC_MISC_UMA_CNTL 0x5f +#define RS480_MC_MCLK_CNTL 0x7a +#define RS480_MC_UMA_DUALCH_CNTL 0x86 + +#define RS690_MC_FB_LOCATION 0x100 +#define RS690_MC_AGP_LOCATION 0x101 +#define RS690_MC_AGP_BASE 0x102 +#define RS690_MC_AGP_BASE_2 0x103 + +#define RS600_MC_INDEX 0x70 +# define RS600_MC_ADDR_MASK 0xffff +# define RS600_MC_IND_SEQ_RBS_0 (1 << 16) +# define RS600_MC_IND_SEQ_RBS_1 (1 << 17) +# define RS600_MC_IND_SEQ_RBS_2 (1 << 18) +# define RS600_MC_IND_SEQ_RBS_3 (1 << 19) +# define RS600_MC_IND_AIC_RBS (1 << 20) +# define RS600_MC_IND_CITF_ARB0 (1 << 21) +# define RS600_MC_IND_CITF_ARB1 (1 << 22) +# define RS600_MC_IND_WR_EN (1 << 23) +#define RS600_MC_DATA 0x74 + +#define RS600_MC_STATUS 0x0 +# define RS600_MC_IDLE (1 << 1) +#define RS600_MC_FB_LOCATION 0x4 +#define RS600_MC_AGP_LOCATION 0x5 +#define RS600_AGP_BASE 0x6 +#define RS600_AGP_BASE_2 0x7 +#define RS600_MC_CNTL1 0x9 +# define RS600_ENABLE_PAGE_TABLES (1 << 26) +#define RS600_MC_PT0_CNTL 0x100 +# define RS600_ENABLE_PT (1 << 0) +# define RS600_EFFECTIVE_L2_CACHE_SIZE(x) ((x) << 15) +# define RS600_EFFECTIVE_L2_QUEUE_SIZE(x) ((x) << 21) +# define RS600_INVALIDATE_ALL_L1_TLBS (1 << 28) +# define RS600_INVALIDATE_L2_CACHE (1 << 29) +#define RS600_MC_PT0_CONTEXT0_CNTL 0x102 +# define RS600_ENABLE_PAGE_TABLE (1 << 0) +# define RS600_PAGE_TABLE_TYPE_FLAT (0 << 1) +#define RS600_MC_PT0_SYSTEM_APERTURE_LOW_ADDR 0x112 +#define RS600_MC_PT0_SYSTEM_APERTURE_HIGH_ADDR 0x114 +#define RS600_MC_PT0_CONTEXT0_DEFAULT_READ_ADDR 0x11c +#define RS600_MC_PT0_CONTEXT0_FLAT_BASE_ADDR 0x12c +#define RS600_MC_PT0_CONTEXT0_FLAT_START_ADDR 0x13c +#define RS600_MC_PT0_CONTEXT0_FLAT_END_ADDR 0x14c +#define RS600_MC_PT0_CLIENT0_CNTL 0x16c +# define RS600_ENABLE_TRANSLATION_MODE_OVERRIDE (1 << 0) +# define RS600_TRANSLATION_MODE_OVERRIDE (1 << 1) +# define RS600_SYSTEM_ACCESS_MODE_MASK (3 << 8) +# define RS600_SYSTEM_ACCESS_MODE_PA_ONLY (0 << 8) +# define RS600_SYSTEM_ACCESS_MODE_USE_SYS_MAP (1 << 8) +# define RS600_SYSTEM_ACCESS_MODE_IN_SYS (2 << 8) +# define RS600_SYSTEM_ACCESS_MODE_NOT_IN_SYS (3 << 8) +# define RS600_SYSTEM_APERTURE_UNMAPPED_ACCESS_PASSTHROUGH (0 << 10) +# define RS600_SYSTEM_APERTURE_UNMAPPED_ACCESS_DEFAULT_PAGE (1 << 10) +# define RS600_EFFECTIVE_L1_CACHE_SIZE(x) ((x) << 11) +# define RS600_ENABLE_FRAGMENT_PROCESSING (1 << 14) +# define RS600_EFFECTIVE_L1_QUEUE_SIZE(x) ((x) << 15) +# define RS600_INVALIDATE_L1_TLB (1 << 20) + +#define R520_MC_IND_INDEX 0x70 +#define R520_MC_IND_WR_EN (1 << 24) +#define R520_MC_IND_DATA 0x74 + +#define RV515_MC_FB_LOCATION 0x01 +#define RV515_MC_AGP_LOCATION 0x02 +#define RV515_MC_AGP_BASE 0x03 +#define RV515_MC_AGP_BASE_2 0x04 + +#define R520_MC_FB_LOCATION 0x04 +#define R520_MC_AGP_LOCATION 0x05 +#define R520_MC_AGP_BASE 0x06 +#define R520_MC_AGP_BASE_2 0x07 + +#define RADEON_MPP_TB_CONFIG 0x01c0 +#define RADEON_MEM_CNTL 0x0140 +#define RADEON_MEM_SDRAM_MODE_REG 0x0158 +#define RADEON_AGP_BASE_2 0x015c /* r200+ only */ +#define RS480_AGP_BASE_2 0x0164 +#define RADEON_AGP_BASE 0x0170 + +/* pipe config regs */ +#define R400_GB_PIPE_SELECT 0x402c +#define RV530_GB_PIPE_SELECT2 0x4124 +#define R500_DYN_SCLK_PWMEM_PIPE 0x000d /* PLL */ +#define R300_GB_TILE_CONFIG 0x4018 +# define R300_ENABLE_TILING (1 << 0) +# define R300_PIPE_COUNT_RV350 (0 << 1) +# define R300_PIPE_COUNT_R300 (3 << 1) +# define R300_PIPE_COUNT_R420_3P (6 << 1) +# define R300_PIPE_COUNT_R420 (7 << 1) +# define R300_TILE_SIZE_8 (0 << 4) +# define R300_TILE_SIZE_16 (1 << 4) +# define R300_TILE_SIZE_32 (2 << 4) +# define R300_SUBPIXEL_1_12 (0 << 16) +# define R300_SUBPIXEL_1_16 (1 << 16) +#define R300_DST_PIPE_CONFIG 0x170c +# define R300_PIPE_AUTO_CONFIG (1 << 31) +#define R300_RB2D_DSTCACHE_MODE 0x3428 +# define R300_DC_AUTOFLUSH_ENABLE (1 << 8) +# define R300_DC_DC_DISABLE_IGNORE_PE (1 << 17) + +#define RADEON_RB3D_COLOROFFSET 0x1c40 +#define RADEON_RB3D_COLORPITCH 0x1c48 + +#define RADEON_SRC_X_Y 0x1590 + +#define RADEON_DP_GUI_MASTER_CNTL 0x146c +# define RADEON_GMC_SRC_PITCH_OFFSET_CNTL (1 << 0) +# define RADEON_GMC_DST_PITCH_OFFSET_CNTL (1 << 1) +# define RADEON_GMC_BRUSH_SOLID_COLOR (13 << 4) +# define RADEON_GMC_BRUSH_NONE (15 << 4) +# define RADEON_GMC_DST_16BPP (4 << 8) +# define RADEON_GMC_DST_24BPP (5 << 8) +# define RADEON_GMC_DST_32BPP (6 << 8) +# define RADEON_GMC_DST_DATATYPE_SHIFT 8 +# define RADEON_GMC_SRC_DATATYPE_COLOR (3 << 12) +# define RADEON_DP_SRC_SOURCE_MEMORY (2 << 24) +# define RADEON_DP_SRC_SOURCE_HOST_DATA (3 << 24) +# define RADEON_GMC_CLR_CMP_CNTL_DIS (1 << 28) +# define RADEON_GMC_WR_MSK_DIS (1 << 30) +# define RADEON_ROP3_S 0x00cc0000 +# define RADEON_ROP3_P 0x00f00000 +#define RADEON_DP_WRITE_MASK 0x16cc +#define RADEON_SRC_PITCH_OFFSET 0x1428 +#define RADEON_DST_PITCH_OFFSET 0x142c +#define RADEON_DST_PITCH_OFFSET_C 0x1c80 +# define RADEON_DST_TILE_LINEAR (0 << 30) +# define RADEON_DST_TILE_MACRO (1 << 30) +# define RADEON_DST_TILE_MICRO (2 << 30) +# define RADEON_DST_TILE_BOTH (3 << 30) + +#define RADEON_SCRATCH_REG0 0x15e0 +#define RADEON_SCRATCH_REG1 0x15e4 +#define RADEON_SCRATCH_REG2 0x15e8 +#define RADEON_SCRATCH_REG3 0x15ec +#define RADEON_SCRATCH_REG4 0x15f0 +#define RADEON_SCRATCH_REG5 0x15f4 +#define RADEON_SCRATCH_UMSK 0x0770 +#define RADEON_SCRATCH_ADDR 0x0774 + +#define RADEON_SCRATCHOFF( x ) (RADEON_SCRATCH_REG_OFFSET + 4*(x)) + +extern u32 radeon_get_scratch(drm_radeon_private_t *dev_priv, int index); + +#define GET_SCRATCH(dev_priv, x) radeon_get_scratch(dev_priv, x) + +#define R600_SCRATCH_REG0 0x8500 +#define R600_SCRATCH_REG1 0x8504 +#define R600_SCRATCH_REG2 0x8508 +#define R600_SCRATCH_REG3 0x850c +#define R600_SCRATCH_REG4 0x8510 +#define R600_SCRATCH_REG5 0x8514 +#define R600_SCRATCH_REG6 0x8518 +#define R600_SCRATCH_REG7 0x851c +#define R600_SCRATCH_UMSK 0x8540 +#define R600_SCRATCH_ADDR 0x8544 + +#define R600_SCRATCHOFF(x) (R600_SCRATCH_REG_OFFSET + 4*(x)) + +#define RADEON_GEN_INT_CNTL 0x0040 +# define RADEON_CRTC_VBLANK_MASK (1 << 0) +# define RADEON_CRTC2_VBLANK_MASK (1 << 9) +# define RADEON_GUI_IDLE_INT_ENABLE (1 << 19) +# define RADEON_SW_INT_ENABLE (1 << 25) + +#define RADEON_GEN_INT_STATUS 0x0044 +# define RADEON_CRTC_VBLANK_STAT (1 << 0) +# define RADEON_CRTC_VBLANK_STAT_ACK (1 << 0) +# define RADEON_CRTC2_VBLANK_STAT (1 << 9) +# define RADEON_CRTC2_VBLANK_STAT_ACK (1 << 9) +# define RADEON_GUI_IDLE_INT_TEST_ACK (1 << 19) +# define RADEON_SW_INT_TEST (1 << 25) +# define RADEON_SW_INT_TEST_ACK (1 << 25) +# define RADEON_SW_INT_FIRE (1 << 26) +# define R500_DISPLAY_INT_STATUS (1 << 0) + +#define RADEON_HOST_PATH_CNTL 0x0130 +# define RADEON_HDP_SOFT_RESET (1 << 26) +# define RADEON_HDP_WC_TIMEOUT_MASK (7 << 28) +# define RADEON_HDP_WC_TIMEOUT_28BCLK (7 << 28) + +#define RADEON_ISYNC_CNTL 0x1724 +# define RADEON_ISYNC_ANY2D_IDLE3D (1 << 0) +# define RADEON_ISYNC_ANY3D_IDLE2D (1 << 1) +# define RADEON_ISYNC_TRIG2D_IDLE3D (1 << 2) +# define RADEON_ISYNC_TRIG3D_IDLE2D (1 << 3) +# define RADEON_ISYNC_WAIT_IDLEGUI (1 << 4) +# define RADEON_ISYNC_CPSCRATCH_IDLEGUI (1 << 5) + +#define RADEON_RBBM_GUICNTL 0x172c +# define RADEON_HOST_DATA_SWAP_NONE (0 << 0) +# define RADEON_HOST_DATA_SWAP_16BIT (1 << 0) +# define RADEON_HOST_DATA_SWAP_32BIT (2 << 0) +# define RADEON_HOST_DATA_SWAP_HDW (3 << 0) + +#define RADEON_MC_AGP_LOCATION 0x014c +#define RADEON_MC_FB_LOCATION 0x0148 +#define RADEON_MCLK_CNTL 0x0012 +# define RADEON_FORCEON_MCLKA (1 << 16) +# define RADEON_FORCEON_MCLKB (1 << 17) +# define RADEON_FORCEON_YCLKA (1 << 18) +# define RADEON_FORCEON_YCLKB (1 << 19) +# define RADEON_FORCEON_MC (1 << 20) +# define RADEON_FORCEON_AIC (1 << 21) + +#define RADEON_PP_BORDER_COLOR_0 0x1d40 +#define RADEON_PP_BORDER_COLOR_1 0x1d44 +#define RADEON_PP_BORDER_COLOR_2 0x1d48 +#define RADEON_PP_CNTL 0x1c38 +# define RADEON_SCISSOR_ENABLE (1 << 1) +#define RADEON_PP_LUM_MATRIX 0x1d00 +#define RADEON_PP_MISC 0x1c14 +#define RADEON_PP_ROT_MATRIX_0 0x1d58 +#define RADEON_PP_TXFILTER_0 0x1c54 +#define RADEON_PP_TXOFFSET_0 0x1c5c +#define RADEON_PP_TXFILTER_1 0x1c6c +#define RADEON_PP_TXFILTER_2 0x1c84 + +#define R300_RB2D_DSTCACHE_CTLSTAT 0x342c /* use R300_DSTCACHE_CTLSTAT */ +#define R300_DSTCACHE_CTLSTAT 0x1714 +# define R300_RB2D_DC_FLUSH (3 << 0) +# define R300_RB2D_DC_FREE (3 << 2) +# define R300_RB2D_DC_FLUSH_ALL 0xf +# define R300_RB2D_DC_BUSY (1 << 31) +#define RADEON_RB3D_CNTL 0x1c3c +# define RADEON_ALPHA_BLEND_ENABLE (1 << 0) +# define RADEON_PLANE_MASK_ENABLE (1 << 1) +# define RADEON_DITHER_ENABLE (1 << 2) +# define RADEON_ROUND_ENABLE (1 << 3) +# define RADEON_SCALE_DITHER_ENABLE (1 << 4) +# define RADEON_DITHER_INIT (1 << 5) +# define RADEON_ROP_ENABLE (1 << 6) +# define RADEON_STENCIL_ENABLE (1 << 7) +# define RADEON_Z_ENABLE (1 << 8) +# define RADEON_ZBLOCK16 (1 << 15) +#define RADEON_RB3D_DEPTHOFFSET 0x1c24 +#define RADEON_RB3D_DEPTHCLEARVALUE 0x3230 +#define RADEON_RB3D_DEPTHPITCH 0x1c28 +#define RADEON_RB3D_PLANEMASK 0x1d84 +#define RADEON_RB3D_STENCILREFMASK 0x1d7c +#define RADEON_RB3D_ZCACHE_MODE 0x3250 +#define RADEON_RB3D_ZCACHE_CTLSTAT 0x3254 +# define RADEON_RB3D_ZC_FLUSH (1 << 0) +# define RADEON_RB3D_ZC_FREE (1 << 2) +# define RADEON_RB3D_ZC_FLUSH_ALL 0x5 +# define RADEON_RB3D_ZC_BUSY (1 << 31) +#define R300_ZB_ZCACHE_CTLSTAT 0x4f18 +# define R300_ZC_FLUSH (1 << 0) +# define R300_ZC_FREE (1 << 1) +# define R300_ZC_BUSY (1 << 31) +#define RADEON_RB3D_DSTCACHE_CTLSTAT 0x325c +# define RADEON_RB3D_DC_FLUSH (3 << 0) +# define RADEON_RB3D_DC_FREE (3 << 2) +# define RADEON_RB3D_DC_FLUSH_ALL 0xf +# define RADEON_RB3D_DC_BUSY (1 << 31) +#define R300_RB3D_DSTCACHE_CTLSTAT 0x4e4c +# define R300_RB3D_DC_FLUSH (2 << 0) +# define R300_RB3D_DC_FREE (2 << 2) +# define R300_RB3D_DC_FINISH (1 << 4) +#define RADEON_RB3D_ZSTENCILCNTL 0x1c2c +# define RADEON_Z_TEST_MASK (7 << 4) +# define RADEON_Z_TEST_ALWAYS (7 << 4) +# define RADEON_Z_HIERARCHY_ENABLE (1 << 8) +# define RADEON_STENCIL_TEST_ALWAYS (7 << 12) +# define RADEON_STENCIL_S_FAIL_REPLACE (2 << 16) +# define RADEON_STENCIL_ZPASS_REPLACE (2 << 20) +# define RADEON_STENCIL_ZFAIL_REPLACE (2 << 24) +# define RADEON_Z_COMPRESSION_ENABLE (1 << 28) +# define RADEON_FORCE_Z_DIRTY (1 << 29) +# define RADEON_Z_WRITE_ENABLE (1 << 30) +# define RADEON_Z_DECOMPRESSION_ENABLE (1 << 31) +#define RADEON_RBBM_SOFT_RESET 0x00f0 +# define RADEON_SOFT_RESET_CP (1 << 0) +# define RADEON_SOFT_RESET_HI (1 << 1) +# define RADEON_SOFT_RESET_SE (1 << 2) +# define RADEON_SOFT_RESET_RE (1 << 3) +# define RADEON_SOFT_RESET_PP (1 << 4) +# define RADEON_SOFT_RESET_E2 (1 << 5) +# define RADEON_SOFT_RESET_RB (1 << 6) +# define RADEON_SOFT_RESET_HDP (1 << 7) +/* + * 6:0 Available slots in the FIFO + * 8 Host Interface active + * 9 CP request active + * 10 FIFO request active + * 11 Host Interface retry active + * 12 CP retry active + * 13 FIFO retry active + * 14 FIFO pipeline busy + * 15 Event engine busy + * 16 CP command stream busy + * 17 2D engine busy + * 18 2D portion of render backend busy + * 20 3D setup engine busy + * 26 GA engine busy + * 27 CBA 2D engine busy + * 31 2D engine busy or 3D engine busy or FIFO not empty or CP busy or + * command stream queue not empty or Ring Buffer not empty + */ +#define RADEON_RBBM_STATUS 0x0e40 +/* Same as the previous RADEON_RBBM_STATUS; this is a mirror of that register. */ +/* #define RADEON_RBBM_STATUS 0x1740 */ +/* bits 6:0 are dword slots available in the cmd fifo */ +# define RADEON_RBBM_FIFOCNT_MASK 0x007f +# define RADEON_HIRQ_ON_RBB (1 << 8) +# define RADEON_CPRQ_ON_RBB (1 << 9) +# define RADEON_CFRQ_ON_RBB (1 << 10) +# define RADEON_HIRQ_IN_RTBUF (1 << 11) +# define RADEON_CPRQ_IN_RTBUF (1 << 12) +# define RADEON_CFRQ_IN_RTBUF (1 << 13) +# define RADEON_PIPE_BUSY (1 << 14) +# define RADEON_ENG_EV_BUSY (1 << 15) +# define RADEON_CP_CMDSTRM_BUSY (1 << 16) +# define RADEON_E2_BUSY (1 << 17) +# define RADEON_RB2D_BUSY (1 << 18) +# define RADEON_RB3D_BUSY (1 << 19) /* not used on r300 */ +# define RADEON_VAP_BUSY (1 << 20) +# define RADEON_RE_BUSY (1 << 21) /* not used on r300 */ +# define RADEON_TAM_BUSY (1 << 22) /* not used on r300 */ +# define RADEON_TDM_BUSY (1 << 23) /* not used on r300 */ +# define RADEON_PB_BUSY (1 << 24) /* not used on r300 */ +# define RADEON_TIM_BUSY (1 << 25) /* not used on r300 */ +# define RADEON_GA_BUSY (1 << 26) +# define RADEON_CBA2D_BUSY (1 << 27) +# define RADEON_RBBM_ACTIVE (1 << 31) +#define RADEON_RE_LINE_PATTERN 0x1cd0 +#define RADEON_RE_MISC 0x26c4 +#define RADEON_RE_TOP_LEFT 0x26c0 +#define RADEON_RE_WIDTH_HEIGHT 0x1c44 +#define RADEON_RE_STIPPLE_ADDR 0x1cc8 +#define RADEON_RE_STIPPLE_DATA 0x1ccc + +#define RADEON_SCISSOR_TL_0 0x1cd8 +#define RADEON_SCISSOR_BR_0 0x1cdc +#define RADEON_SCISSOR_TL_1 0x1ce0 +#define RADEON_SCISSOR_BR_1 0x1ce4 +#define RADEON_SCISSOR_TL_2 0x1ce8 +#define RADEON_SCISSOR_BR_2 0x1cec +#define RADEON_SE_COORD_FMT 0x1c50 +#define RADEON_SE_CNTL 0x1c4c +# define RADEON_FFACE_CULL_CW (0 << 0) +# define RADEON_BFACE_SOLID (3 << 1) +# define RADEON_FFACE_SOLID (3 << 3) +# define RADEON_FLAT_SHADE_VTX_LAST (3 << 6) +# define RADEON_DIFFUSE_SHADE_FLAT (1 << 8) +# define RADEON_DIFFUSE_SHADE_GOURAUD (2 << 8) +# define RADEON_ALPHA_SHADE_FLAT (1 << 10) +# define RADEON_ALPHA_SHADE_GOURAUD (2 << 10) +# define RADEON_SPECULAR_SHADE_FLAT (1 << 12) +# define RADEON_SPECULAR_SHADE_GOURAUD (2 << 12) +# define RADEON_FOG_SHADE_FLAT (1 << 14) +# define RADEON_FOG_SHADE_GOURAUD (2 << 14) +# define RADEON_VPORT_XY_XFORM_ENABLE (1 << 24) +# define RADEON_VPORT_Z_XFORM_ENABLE (1 << 25) +# define RADEON_VTX_PIX_CENTER_OGL (1 << 27) +# define RADEON_ROUND_MODE_TRUNC (0 << 28) +# define RADEON_ROUND_PREC_8TH_PIX (1 << 30) +#define RADEON_SE_CNTL_STATUS 0x2140 +#define RADEON_SE_LINE_WIDTH 0x1db8 +#define RADEON_SE_VPORT_XSCALE 0x1d98 +#define RADEON_SE_ZBIAS_FACTOR 0x1db0 +#define RADEON_SE_TCL_MATERIAL_EMMISSIVE_RED 0x2210 +#define RADEON_SE_TCL_OUTPUT_VTX_FMT 0x2254 +#define RADEON_SE_TCL_VECTOR_INDX_REG 0x2200 +# define RADEON_VEC_INDX_OCTWORD_STRIDE_SHIFT 16 +# define RADEON_VEC_INDX_DWORD_COUNT_SHIFT 28 +#define RADEON_SE_TCL_VECTOR_DATA_REG 0x2204 +#define RADEON_SE_TCL_SCALAR_INDX_REG 0x2208 +# define RADEON_SCAL_INDX_DWORD_STRIDE_SHIFT 16 +#define RADEON_SE_TCL_SCALAR_DATA_REG 0x220C +#define RADEON_SURFACE_ACCESS_FLAGS 0x0bf8 +#define RADEON_SURFACE_ACCESS_CLR 0x0bfc +#define RADEON_SURFACE_CNTL 0x0b00 +# define RADEON_SURF_TRANSLATION_DIS (1 << 8) +# define RADEON_NONSURF_AP0_SWP_MASK (3 << 20) +# define RADEON_NONSURF_AP0_SWP_LITTLE (0 << 20) +# define RADEON_NONSURF_AP0_SWP_BIG16 (1 << 20) +# define RADEON_NONSURF_AP0_SWP_BIG32 (2 << 20) +# define RADEON_NONSURF_AP1_SWP_MASK (3 << 22) +# define RADEON_NONSURF_AP1_SWP_LITTLE (0 << 22) +# define RADEON_NONSURF_AP1_SWP_BIG16 (1 << 22) +# define RADEON_NONSURF_AP1_SWP_BIG32 (2 << 22) +#define RADEON_SURFACE0_INFO 0x0b0c +# define RADEON_SURF_PITCHSEL_MASK (0x1ff << 0) +# define RADEON_SURF_TILE_MODE_MASK (3 << 16) +# define RADEON_SURF_TILE_MODE_MACRO (0 << 16) +# define RADEON_SURF_TILE_MODE_MICRO (1 << 16) +# define RADEON_SURF_TILE_MODE_32BIT_Z (2 << 16) +# define RADEON_SURF_TILE_MODE_16BIT_Z (3 << 16) +#define RADEON_SURFACE0_LOWER_BOUND 0x0b04 +#define RADEON_SURFACE0_UPPER_BOUND 0x0b08 +# define RADEON_SURF_ADDRESS_FIXED_MASK (0x3ff << 0) +#define RADEON_SURFACE1_INFO 0x0b1c +#define RADEON_SURFACE1_LOWER_BOUND 0x0b14 +#define RADEON_SURFACE1_UPPER_BOUND 0x0b18 +#define RADEON_SURFACE2_INFO 0x0b2c +#define RADEON_SURFACE2_LOWER_BOUND 0x0b24 +#define RADEON_SURFACE2_UPPER_BOUND 0x0b28 +#define RADEON_SURFACE3_INFO 0x0b3c +#define RADEON_SURFACE3_LOWER_BOUND 0x0b34 +#define RADEON_SURFACE3_UPPER_BOUND 0x0b38 +#define RADEON_SURFACE4_INFO 0x0b4c +#define RADEON_SURFACE4_LOWER_BOUND 0x0b44 +#define RADEON_SURFACE4_UPPER_BOUND 0x0b48 +#define RADEON_SURFACE5_INFO 0x0b5c +#define RADEON_SURFACE5_LOWER_BOUND 0x0b54 +#define RADEON_SURFACE5_UPPER_BOUND 0x0b58 +#define RADEON_SURFACE6_INFO 0x0b6c +#define RADEON_SURFACE6_LOWER_BOUND 0x0b64 +#define RADEON_SURFACE6_UPPER_BOUND 0x0b68 +#define RADEON_SURFACE7_INFO 0x0b7c +#define RADEON_SURFACE7_LOWER_BOUND 0x0b74 +#define RADEON_SURFACE7_UPPER_BOUND 0x0b78 +#define RADEON_SW_SEMAPHORE 0x013c + +#define RADEON_WAIT_UNTIL 0x1720 +# define RADEON_WAIT_CRTC_PFLIP (1 << 0) +# define RADEON_WAIT_2D_IDLE (1 << 14) +# define RADEON_WAIT_3D_IDLE (1 << 15) +# define RADEON_WAIT_2D_IDLECLEAN (1 << 16) +# define RADEON_WAIT_3D_IDLECLEAN (1 << 17) +# define RADEON_WAIT_HOST_IDLECLEAN (1 << 18) + +#define RADEON_RB3D_ZMASKOFFSET 0x3234 +#define RADEON_RB3D_ZSTENCILCNTL 0x1c2c +# define RADEON_DEPTH_FORMAT_16BIT_INT_Z (0 << 0) +# define RADEON_DEPTH_FORMAT_24BIT_INT_Z (2 << 0) + +/* CP registers */ +#define RADEON_CP_ME_RAM_ADDR 0x07d4 +#define RADEON_CP_ME_RAM_RADDR 0x07d8 +#define RADEON_CP_ME_RAM_DATAH 0x07dc +#define RADEON_CP_ME_RAM_DATAL 0x07e0 + +#define RADEON_CP_RB_BASE 0x0700 +#define RADEON_CP_RB_CNTL 0x0704 +# define RADEON_BUF_SWAP_32BIT (2 << 16) +# define RADEON_RB_NO_UPDATE (1 << 27) +# define RADEON_RB_RPTR_WR_ENA (1 << 31) +#define RADEON_CP_RB_RPTR_ADDR 0x070c +#define RADEON_CP_RB_RPTR 0x0710 +#define RADEON_CP_RB_WPTR 0x0714 + +#define RADEON_CP_RB_WPTR_DELAY 0x0718 +# define RADEON_PRE_WRITE_TIMER_SHIFT 0 +# define RADEON_PRE_WRITE_LIMIT_SHIFT 23 + +#define RADEON_CP_IB_BASE 0x0738 + +#define RADEON_CP_CSQ_CNTL 0x0740 +# define RADEON_CSQ_CNT_PRIMARY_MASK (0xff << 0) +# define RADEON_CSQ_PRIDIS_INDDIS (0 << 28) +# define RADEON_CSQ_PRIPIO_INDDIS (1 << 28) +# define RADEON_CSQ_PRIBM_INDDIS (2 << 28) +# define RADEON_CSQ_PRIPIO_INDBM (3 << 28) +# define RADEON_CSQ_PRIBM_INDBM (4 << 28) +# define RADEON_CSQ_PRIPIO_INDPIO (15 << 28) + +#define R300_CP_RESYNC_ADDR 0x0778 +#define R300_CP_RESYNC_DATA 0x077c + +#define RADEON_AIC_CNTL 0x01d0 +# define RADEON_PCIGART_TRANSLATE_EN (1 << 0) +# define RS400_MSI_REARM (1 << 3) +#define RADEON_AIC_STAT 0x01d4 +#define RADEON_AIC_PT_BASE 0x01d8 +#define RADEON_AIC_LO_ADDR 0x01dc +#define RADEON_AIC_HI_ADDR 0x01e0 +#define RADEON_AIC_TLB_ADDR 0x01e4 +#define RADEON_AIC_TLB_DATA 0x01e8 + +/* CP command packets */ +#define RADEON_CP_PACKET0 0x00000000 +# define RADEON_ONE_REG_WR (1 << 15) +#define RADEON_CP_PACKET1 0x40000000 +#define RADEON_CP_PACKET2 0x80000000 +#define RADEON_CP_PACKET3 0xC0000000 +# define RADEON_CP_NOP 0x00001000 +# define RADEON_CP_NEXT_CHAR 0x00001900 +# define RADEON_CP_PLY_NEXTSCAN 0x00001D00 +# define RADEON_CP_SET_SCISSORS 0x00001E00 + /* GEN_INDX_PRIM is unsupported starting with R300 */ +# define RADEON_3D_RNDR_GEN_INDX_PRIM 0x00002300 +# define RADEON_WAIT_FOR_IDLE 0x00002600 +# define RADEON_3D_DRAW_VBUF 0x00002800 +# define RADEON_3D_DRAW_IMMD 0x00002900 +# define RADEON_3D_DRAW_INDX 0x00002A00 +# define RADEON_CP_LOAD_PALETTE 0x00002C00 +# define RADEON_3D_LOAD_VBPNTR 0x00002F00 +# define RADEON_MPEG_IDCT_MACROBLOCK 0x00003000 +# define RADEON_MPEG_IDCT_MACROBLOCK_REV 0x00003100 +# define RADEON_3D_CLEAR_ZMASK 0x00003200 +# define RADEON_CP_INDX_BUFFER 0x00003300 +# define RADEON_CP_3D_DRAW_VBUF_2 0x00003400 +# define RADEON_CP_3D_DRAW_IMMD_2 0x00003500 +# define RADEON_CP_3D_DRAW_INDX_2 0x00003600 +# define RADEON_3D_CLEAR_HIZ 0x00003700 +# define RADEON_CP_3D_CLEAR_CMASK 0x00003802 +# define RADEON_CNTL_HOSTDATA_BLT 0x00009400 +# define RADEON_CNTL_PAINT_MULTI 0x00009A00 +# define RADEON_CNTL_BITBLT_MULTI 0x00009B00 +# define RADEON_CNTL_SET_SCISSORS 0xC0001E00 + +# define R600_IT_INDIRECT_BUFFER_END 0x00001700 +# define R600_IT_SET_PREDICATION 0x00002000 +# define R600_IT_REG_RMW 0x00002100 +# define R600_IT_COND_EXEC 0x00002200 +# define R600_IT_PRED_EXEC 0x00002300 +# define R600_IT_START_3D_CMDBUF 0x00002400 +# define R600_IT_DRAW_INDEX_2 0x00002700 +# define R600_IT_CONTEXT_CONTROL 0x00002800 +# define R600_IT_DRAW_INDEX_IMMD_BE 0x00002900 +# define R600_IT_INDEX_TYPE 0x00002A00 +# define R600_IT_DRAW_INDEX 0x00002B00 +# define R600_IT_DRAW_INDEX_AUTO 0x00002D00 +# define R600_IT_DRAW_INDEX_IMMD 0x00002E00 +# define R600_IT_NUM_INSTANCES 0x00002F00 +# define R600_IT_STRMOUT_BUFFER_UPDATE 0x00003400 +# define R600_IT_INDIRECT_BUFFER_MP 0x00003800 +# define R600_IT_MEM_SEMAPHORE 0x00003900 +# define R600_IT_MPEG_INDEX 0x00003A00 +# define R600_IT_WAIT_REG_MEM 0x00003C00 +# define R600_IT_MEM_WRITE 0x00003D00 +# define R600_IT_INDIRECT_BUFFER 0x00003200 +# define R600_IT_SURFACE_SYNC 0x00004300 +# define R600_CB0_DEST_BASE_ENA (1 << 6) +# define R600_TC_ACTION_ENA (1 << 23) +# define R600_VC_ACTION_ENA (1 << 24) +# define R600_CB_ACTION_ENA (1 << 25) +# define R600_DB_ACTION_ENA (1 << 26) +# define R600_SH_ACTION_ENA (1 << 27) +# define R600_SMX_ACTION_ENA (1 << 28) +# define R600_IT_ME_INITIALIZE 0x00004400 +# define R600_ME_INITIALIZE_DEVICE_ID(x) ((x) << 16) +# define R600_IT_COND_WRITE 0x00004500 +# define R600_IT_EVENT_WRITE 0x00004600 +# define R600_IT_EVENT_WRITE_EOP 0x00004700 +# define R600_IT_ONE_REG_WRITE 0x00005700 +# define R600_IT_SET_CONFIG_REG 0x00006800 +# define R600_SET_CONFIG_REG_OFFSET 0x00008000 +# define R600_SET_CONFIG_REG_END 0x0000ac00 +# define R600_IT_SET_CONTEXT_REG 0x00006900 +# define R600_SET_CONTEXT_REG_OFFSET 0x00028000 +# define R600_SET_CONTEXT_REG_END 0x00029000 +# define R600_IT_SET_ALU_CONST 0x00006A00 +# define R600_SET_ALU_CONST_OFFSET 0x00030000 +# define R600_SET_ALU_CONST_END 0x00032000 +# define R600_IT_SET_BOOL_CONST 0x00006B00 +# define R600_SET_BOOL_CONST_OFFSET 0x0003e380 +# define R600_SET_BOOL_CONST_END 0x00040000 +# define R600_IT_SET_LOOP_CONST 0x00006C00 +# define R600_SET_LOOP_CONST_OFFSET 0x0003e200 +# define R600_SET_LOOP_CONST_END 0x0003e380 +# define R600_IT_SET_RESOURCE 0x00006D00 +# define R600_SET_RESOURCE_OFFSET 0x00038000 +# define R600_SET_RESOURCE_END 0x0003c000 +# define R600_SQ_TEX_VTX_INVALID_TEXTURE 0x0 +# define R600_SQ_TEX_VTX_INVALID_BUFFER 0x1 +# define R600_SQ_TEX_VTX_VALID_TEXTURE 0x2 +# define R600_SQ_TEX_VTX_VALID_BUFFER 0x3 +# define R600_IT_SET_SAMPLER 0x00006E00 +# define R600_SET_SAMPLER_OFFSET 0x0003c000 +# define R600_SET_SAMPLER_END 0x0003cff0 +# define R600_IT_SET_CTL_CONST 0x00006F00 +# define R600_SET_CTL_CONST_OFFSET 0x0003cff0 +# define R600_SET_CTL_CONST_END 0x0003e200 +# define R600_IT_SURFACE_BASE_UPDATE 0x00007300 + +#define RADEON_CP_PACKET_MASK 0xC0000000 +#define RADEON_CP_PACKET_COUNT_MASK 0x3fff0000 +#define RADEON_CP_PACKET0_REG_MASK 0x000007ff +#define RADEON_CP_PACKET1_REG0_MASK 0x000007ff +#define RADEON_CP_PACKET1_REG1_MASK 0x003ff800 + +#define RADEON_VTX_Z_PRESENT (1 << 31) +#define RADEON_VTX_PKCOLOR_PRESENT (1 << 3) + +#define RADEON_PRIM_TYPE_NONE (0 << 0) +#define RADEON_PRIM_TYPE_POINT (1 << 0) +#define RADEON_PRIM_TYPE_LINE (2 << 0) +#define RADEON_PRIM_TYPE_LINE_STRIP (3 << 0) +#define RADEON_PRIM_TYPE_TRI_LIST (4 << 0) +#define RADEON_PRIM_TYPE_TRI_FAN (5 << 0) +#define RADEON_PRIM_TYPE_TRI_STRIP (6 << 0) +#define RADEON_PRIM_TYPE_TRI_TYPE2 (7 << 0) +#define RADEON_PRIM_TYPE_RECT_LIST (8 << 0) +#define RADEON_PRIM_TYPE_3VRT_POINT_LIST (9 << 0) +#define RADEON_PRIM_TYPE_3VRT_LINE_LIST (10 << 0) +#define RADEON_PRIM_TYPE_MASK 0xf +#define RADEON_PRIM_WALK_IND (1 << 4) +#define RADEON_PRIM_WALK_LIST (2 << 4) +#define RADEON_PRIM_WALK_RING (3 << 4) +#define RADEON_COLOR_ORDER_BGRA (0 << 6) +#define RADEON_COLOR_ORDER_RGBA (1 << 6) +#define RADEON_MAOS_ENABLE (1 << 7) +#define RADEON_VTX_FMT_R128_MODE (0 << 8) +#define RADEON_VTX_FMT_RADEON_MODE (1 << 8) +#define RADEON_NUM_VERTICES_SHIFT 16 + +#define RADEON_COLOR_FORMAT_CI8 2 +#define RADEON_COLOR_FORMAT_ARGB1555 3 +#define RADEON_COLOR_FORMAT_RGB565 4 +#define RADEON_COLOR_FORMAT_ARGB8888 6 +#define RADEON_COLOR_FORMAT_RGB332 7 +#define RADEON_COLOR_FORMAT_RGB8 9 +#define RADEON_COLOR_FORMAT_ARGB4444 15 + +#define RADEON_TXFORMAT_I8 0 +#define RADEON_TXFORMAT_AI88 1 +#define RADEON_TXFORMAT_RGB332 2 +#define RADEON_TXFORMAT_ARGB1555 3 +#define RADEON_TXFORMAT_RGB565 4 +#define RADEON_TXFORMAT_ARGB4444 5 +#define RADEON_TXFORMAT_ARGB8888 6 +#define RADEON_TXFORMAT_RGBA8888 7 +#define RADEON_TXFORMAT_Y8 8 +#define RADEON_TXFORMAT_VYUY422 10 +#define RADEON_TXFORMAT_YVYU422 11 +#define RADEON_TXFORMAT_DXT1 12 +#define RADEON_TXFORMAT_DXT23 14 +#define RADEON_TXFORMAT_DXT45 15 + +#define R200_PP_TXCBLEND_0 0x2f00 +#define R200_PP_TXCBLEND_1 0x2f10 +#define R200_PP_TXCBLEND_2 0x2f20 +#define R200_PP_TXCBLEND_3 0x2f30 +#define R200_PP_TXCBLEND_4 0x2f40 +#define R200_PP_TXCBLEND_5 0x2f50 +#define R200_PP_TXCBLEND_6 0x2f60 +#define R200_PP_TXCBLEND_7 0x2f70 +#define R200_SE_TCL_LIGHT_MODEL_CTL_0 0x2268 +#define R200_PP_TFACTOR_0 0x2ee0 +#define R200_SE_VTX_FMT_0 0x2088 +#define R200_SE_VAP_CNTL 0x2080 +#define R200_SE_TCL_MATRIX_SEL_0 0x2230 +#define R200_SE_TCL_TEX_PROC_CTL_2 0x22a8 +#define R200_SE_TCL_UCP_VERT_BLEND_CTL 0x22c0 +#define R200_PP_TXFILTER_5 0x2ca0 +#define R200_PP_TXFILTER_4 0x2c80 +#define R200_PP_TXFILTER_3 0x2c60 +#define R200_PP_TXFILTER_2 0x2c40 +#define R200_PP_TXFILTER_1 0x2c20 +#define R200_PP_TXFILTER_0 0x2c00 +#define R200_PP_TXOFFSET_5 0x2d78 +#define R200_PP_TXOFFSET_4 0x2d60 +#define R200_PP_TXOFFSET_3 0x2d48 +#define R200_PP_TXOFFSET_2 0x2d30 +#define R200_PP_TXOFFSET_1 0x2d18 +#define R200_PP_TXOFFSET_0 0x2d00 + +#define R200_PP_CUBIC_FACES_0 0x2c18 +#define R200_PP_CUBIC_FACES_1 0x2c38 +#define R200_PP_CUBIC_FACES_2 0x2c58 +#define R200_PP_CUBIC_FACES_3 0x2c78 +#define R200_PP_CUBIC_FACES_4 0x2c98 +#define R200_PP_CUBIC_FACES_5 0x2cb8 +#define R200_PP_CUBIC_OFFSET_F1_0 0x2d04 +#define R200_PP_CUBIC_OFFSET_F2_0 0x2d08 +#define R200_PP_CUBIC_OFFSET_F3_0 0x2d0c +#define R200_PP_CUBIC_OFFSET_F4_0 0x2d10 +#define R200_PP_CUBIC_OFFSET_F5_0 0x2d14 +#define R200_PP_CUBIC_OFFSET_F1_1 0x2d1c +#define R200_PP_CUBIC_OFFSET_F2_1 0x2d20 +#define R200_PP_CUBIC_OFFSET_F3_1 0x2d24 +#define R200_PP_CUBIC_OFFSET_F4_1 0x2d28 +#define R200_PP_CUBIC_OFFSET_F5_1 0x2d2c +#define R200_PP_CUBIC_OFFSET_F1_2 0x2d34 +#define R200_PP_CUBIC_OFFSET_F2_2 0x2d38 +#define R200_PP_CUBIC_OFFSET_F3_2 0x2d3c +#define R200_PP_CUBIC_OFFSET_F4_2 0x2d40 +#define R200_PP_CUBIC_OFFSET_F5_2 0x2d44 +#define R200_PP_CUBIC_OFFSET_F1_3 0x2d4c +#define R200_PP_CUBIC_OFFSET_F2_3 0x2d50 +#define R200_PP_CUBIC_OFFSET_F3_3 0x2d54 +#define R200_PP_CUBIC_OFFSET_F4_3 0x2d58 +#define R200_PP_CUBIC_OFFSET_F5_3 0x2d5c +#define R200_PP_CUBIC_OFFSET_F1_4 0x2d64 +#define R200_PP_CUBIC_OFFSET_F2_4 0x2d68 +#define R200_PP_CUBIC_OFFSET_F3_4 0x2d6c +#define R200_PP_CUBIC_OFFSET_F4_4 0x2d70 +#define R200_PP_CUBIC_OFFSET_F5_4 0x2d74 +#define R200_PP_CUBIC_OFFSET_F1_5 0x2d7c +#define R200_PP_CUBIC_OFFSET_F2_5 0x2d80 +#define R200_PP_CUBIC_OFFSET_F3_5 0x2d84 +#define R200_PP_CUBIC_OFFSET_F4_5 0x2d88 +#define R200_PP_CUBIC_OFFSET_F5_5 0x2d8c + +#define R200_RE_AUX_SCISSOR_CNTL 0x26f0 +#define R200_SE_VTE_CNTL 0x20b0 +#define R200_SE_TCL_OUTPUT_VTX_COMP_SEL 0x2250 +#define R200_PP_TAM_DEBUG3 0x2d9c +#define R200_PP_CNTL_X 0x2cc4 +#define R200_SE_VAP_CNTL_STATUS 0x2140 +#define R200_RE_SCISSOR_TL_0 0x1cd8 +#define R200_RE_SCISSOR_TL_1 0x1ce0 +#define R200_RE_SCISSOR_TL_2 0x1ce8 +#define R200_RB3D_DEPTHXY_OFFSET 0x1d60 +#define R200_RE_AUX_SCISSOR_CNTL 0x26f0 +#define R200_SE_VTX_STATE_CNTL 0x2180 +#define R200_RE_POINTSIZE 0x2648 +#define R200_SE_TCL_INPUT_VTX_VECTOR_ADDR_0 0x2254 + +#define RADEON_PP_TEX_SIZE_0 0x1d04 /* NPOT */ +#define RADEON_PP_TEX_SIZE_1 0x1d0c +#define RADEON_PP_TEX_SIZE_2 0x1d14 + +#define RADEON_PP_CUBIC_FACES_0 0x1d24 +#define RADEON_PP_CUBIC_FACES_1 0x1d28 +#define RADEON_PP_CUBIC_FACES_2 0x1d2c +#define RADEON_PP_CUBIC_OFFSET_T0_0 0x1dd0 /* bits [31:5] */ +#define RADEON_PP_CUBIC_OFFSET_T1_0 0x1e00 +#define RADEON_PP_CUBIC_OFFSET_T2_0 0x1e14 + +#define RADEON_SE_TCL_STATE_FLUSH 0x2284 + +#define SE_VAP_CNTL__TCL_ENA_MASK 0x00000001 +#define SE_VAP_CNTL__FORCE_W_TO_ONE_MASK 0x00010000 +#define SE_VAP_CNTL__VF_MAX_VTX_NUM__SHIFT 0x00000012 +#define SE_VTE_CNTL__VTX_XY_FMT_MASK 0x00000100 +#define SE_VTE_CNTL__VTX_Z_FMT_MASK 0x00000200 +#define SE_VTX_FMT_0__VTX_Z0_PRESENT_MASK 0x00000001 +#define SE_VTX_FMT_0__VTX_W0_PRESENT_MASK 0x00000002 +#define SE_VTX_FMT_0__VTX_COLOR_0_FMT__SHIFT 0x0000000b +#define R200_3D_DRAW_IMMD_2 0xC0003500 +#define R200_SE_VTX_FMT_1 0x208c +#define R200_RE_CNTL 0x1c50 + +#define R200_RB3D_BLENDCOLOR 0x3218 + +#define R200_SE_TCL_POINT_SPRITE_CNTL 0x22c4 + +#define R200_PP_TRI_PERF 0x2cf8 + +#define R200_PP_AFS_0 0x2f80 +#define R200_PP_AFS_1 0x2f00 /* same as txcblend_0 */ + +#define R200_VAP_PVS_CNTL_1 0x22D0 + +#define RADEON_CRTC_CRNT_FRAME 0x0214 +#define RADEON_CRTC2_CRNT_FRAME 0x0314 + +#define R500_D1CRTC_STATUS 0x609c +#define R500_D2CRTC_STATUS 0x689c +#define R500_CRTC_V_BLANK (1<<0) + +#define R500_D1CRTC_FRAME_COUNT 0x60a4 +#define R500_D2CRTC_FRAME_COUNT 0x68a4 + +#define R500_D1MODE_V_COUNTER 0x6530 +#define R500_D2MODE_V_COUNTER 0x6d30 + +#define R500_D1MODE_VBLANK_STATUS 0x6534 +#define R500_D2MODE_VBLANK_STATUS 0x6d34 +#define R500_VBLANK_OCCURED (1<<0) +#define R500_VBLANK_ACK (1<<4) +#define R500_VBLANK_STAT (1<<12) +#define R500_VBLANK_INT (1<<16) + +#define R500_DxMODE_INT_MASK 0x6540 +#define R500_D1MODE_INT_MASK (1<<0) +#define R500_D2MODE_INT_MASK (1<<8) + +#define R500_DISP_INTERRUPT_STATUS 0x7edc +#define R500_D1_VBLANK_INTERRUPT (1 << 4) +#define R500_D2_VBLANK_INTERRUPT (1 << 5) + +/* R6xx/R7xx registers */ +#define R600_MC_VM_FB_LOCATION 0x2180 +#define R600_MC_VM_AGP_TOP 0x2184 +#define R600_MC_VM_AGP_BOT 0x2188 +#define R600_MC_VM_AGP_BASE 0x218c +#define R600_MC_VM_SYSTEM_APERTURE_LOW_ADDR 0x2190 +#define R600_MC_VM_SYSTEM_APERTURE_HIGH_ADDR 0x2194 +#define R600_MC_VM_SYSTEM_APERTURE_DEFAULT_ADDR 0x2198 + +#define R700_MC_VM_FB_LOCATION 0x2024 +#define R700_MC_VM_AGP_TOP 0x2028 +#define R700_MC_VM_AGP_BOT 0x202c +#define R700_MC_VM_AGP_BASE 0x2030 +#define R700_MC_VM_SYSTEM_APERTURE_LOW_ADDR 0x2034 +#define R700_MC_VM_SYSTEM_APERTURE_HIGH_ADDR 0x2038 +#define R700_MC_VM_SYSTEM_APERTURE_DEFAULT_ADDR 0x203c + +#define R600_MCD_RD_A_CNTL 0x219c +#define R600_MCD_RD_B_CNTL 0x21a0 + +#define R600_MCD_WR_A_CNTL 0x21a4 +#define R600_MCD_WR_B_CNTL 0x21a8 + +#define R600_MCD_RD_SYS_CNTL 0x2200 +#define R600_MCD_WR_SYS_CNTL 0x2214 + +#define R600_MCD_RD_GFX_CNTL 0x21fc +#define R600_MCD_RD_HDP_CNTL 0x2204 +#define R600_MCD_RD_PDMA_CNTL 0x2208 +#define R600_MCD_RD_SEM_CNTL 0x220c +#define R600_MCD_WR_GFX_CNTL 0x2210 +#define R600_MCD_WR_HDP_CNTL 0x2218 +#define R600_MCD_WR_PDMA_CNTL 0x221c +#define R600_MCD_WR_SEM_CNTL 0x2220 + +# define R600_MCD_L1_TLB (1 << 0) +# define R600_MCD_L1_FRAG_PROC (1 << 1) +# define R600_MCD_L1_STRICT_ORDERING (1 << 2) + +# define R600_MCD_SYSTEM_ACCESS_MODE_MASK (3 << 6) +# define R600_MCD_SYSTEM_ACCESS_MODE_PA_ONLY (0 << 6) +# define R600_MCD_SYSTEM_ACCESS_MODE_USE_SYS_MAP (1 << 6) +# define R600_MCD_SYSTEM_ACCESS_MODE_IN_SYS (2 << 6) +# define R600_MCD_SYSTEM_ACCESS_MODE_NOT_IN_SYS (3 << 6) + +# define R600_MCD_SYSTEM_APERTURE_UNMAPPED_ACCESS_PASS_THRU (0 << 8) +# define R600_MCD_SYSTEM_APERTURE_UNMAPPED_ACCESS_DEFAULT_PAGE (1 << 8) + +# define R600_MCD_SEMAPHORE_MODE (1 << 10) +# define R600_MCD_WAIT_L2_QUERY (1 << 11) +# define R600_MCD_EFFECTIVE_L1_TLB_SIZE(x) ((x) << 12) +# define R600_MCD_EFFECTIVE_L1_QUEUE_SIZE(x) ((x) << 15) + +#define R700_MC_VM_MD_L1_TLB0_CNTL 0x2654 +#define R700_MC_VM_MD_L1_TLB1_CNTL 0x2658 +#define R700_MC_VM_MD_L1_TLB2_CNTL 0x265c + +#define R700_MC_VM_MB_L1_TLB0_CNTL 0x2234 +#define R700_MC_VM_MB_L1_TLB1_CNTL 0x2238 +#define R700_MC_VM_MB_L1_TLB2_CNTL 0x223c +#define R700_MC_VM_MB_L1_TLB3_CNTL 0x2240 + +# define R700_ENABLE_L1_TLB (1 << 0) +# define R700_ENABLE_L1_FRAGMENT_PROCESSING (1 << 1) +# define R700_SYSTEM_ACCESS_MODE_IN_SYS (2 << 3) +# define R700_SYSTEM_APERTURE_UNMAPPED_ACCESS_PASS_THRU (0 << 5) +# define R700_EFFECTIVE_L1_TLB_SIZE(x) ((x) << 15) +# define R700_EFFECTIVE_L1_QUEUE_SIZE(x) ((x) << 18) + +#define R700_MC_ARB_RAMCFG 0x2760 +# define R700_NOOFBANK_SHIFT 0 +# define R700_NOOFBANK_MASK 0x3 +# define R700_NOOFRANK_SHIFT 2 +# define R700_NOOFRANK_MASK 0x1 +# define R700_NOOFROWS_SHIFT 3 +# define R700_NOOFROWS_MASK 0x7 +# define R700_NOOFCOLS_SHIFT 6 +# define R700_NOOFCOLS_MASK 0x3 +# define R700_CHANSIZE_SHIFT 8 +# define R700_CHANSIZE_MASK 0x1 +# define R700_BURSTLENGTH_SHIFT 9 +# define R700_BURSTLENGTH_MASK 0x1 +#define R600_RAMCFG 0x2408 +# define R600_NOOFBANK_SHIFT 0 +# define R600_NOOFBANK_MASK 0x1 +# define R600_NOOFRANK_SHIFT 1 +# define R600_NOOFRANK_MASK 0x1 +# define R600_NOOFROWS_SHIFT 2 +# define R600_NOOFROWS_MASK 0x7 +# define R600_NOOFCOLS_SHIFT 5 +# define R600_NOOFCOLS_MASK 0x3 +# define R600_CHANSIZE_SHIFT 7 +# define R600_CHANSIZE_MASK 0x1 +# define R600_BURSTLENGTH_SHIFT 8 +# define R600_BURSTLENGTH_MASK 0x1 + +#define R600_VM_L2_CNTL 0x1400 +# define R600_VM_L2_CACHE_EN (1 << 0) +# define R600_VM_L2_FRAG_PROC (1 << 1) +# define R600_VM_ENABLE_PTE_CACHE_LRU_W (1 << 9) +# define R600_VM_L2_CNTL_QUEUE_SIZE(x) ((x) << 13) +# define R700_VM_L2_CNTL_QUEUE_SIZE(x) ((x) << 14) + +#define R600_VM_L2_CNTL2 0x1404 +# define R600_VM_L2_CNTL2_INVALIDATE_ALL_L1_TLBS (1 << 0) +# define R600_VM_L2_CNTL2_INVALIDATE_L2_CACHE (1 << 1) +#define R600_VM_L2_CNTL3 0x1408 +# define R600_VM_L2_CNTL3_BANK_SELECT_0(x) ((x) << 0) +# define R600_VM_L2_CNTL3_BANK_SELECT_1(x) ((x) << 5) +# define R600_VM_L2_CNTL3_CACHE_UPDATE_MODE(x) ((x) << 10) +# define R700_VM_L2_CNTL3_BANK_SELECT(x) ((x) << 0) +# define R700_VM_L2_CNTL3_CACHE_UPDATE_MODE(x) ((x) << 6) + +#define R600_VM_L2_STATUS 0x140c + +#define R600_VM_CONTEXT0_CNTL 0x1410 +# define R600_VM_ENABLE_CONTEXT (1 << 0) +# define R600_VM_PAGE_TABLE_DEPTH_FLAT (0 << 1) + +#define R600_VM_CONTEXT0_CNTL2 0x1430 +#define R600_VM_CONTEXT0_REQUEST_RESPONSE 0x1470 +#define R600_VM_CONTEXT0_INVALIDATION_LOW_ADDR 0x1490 +#define R600_VM_CONTEXT0_INVALIDATION_HIGH_ADDR 0x14b0 +#define R600_VM_CONTEXT0_PAGE_TABLE_BASE_ADDR 0x1574 +#define R600_VM_CONTEXT0_PAGE_TABLE_START_ADDR 0x1594 +#define R600_VM_CONTEXT0_PAGE_TABLE_END_ADDR 0x15b4 + +#define R700_VM_CONTEXT0_PAGE_TABLE_BASE_ADDR 0x153c +#define R700_VM_CONTEXT0_PAGE_TABLE_START_ADDR 0x155c +#define R700_VM_CONTEXT0_PAGE_TABLE_END_ADDR 0x157c + +#define R600_HDP_HOST_PATH_CNTL 0x2c00 + +#define R600_GRBM_CNTL 0x8000 +# define R600_GRBM_READ_TIMEOUT(x) ((x) << 0) + +#define R600_GRBM_STATUS 0x8010 +# define R600_CMDFIFO_AVAIL_MASK 0x1f +# define R700_CMDFIFO_AVAIL_MASK 0xf +# define R600_GUI_ACTIVE (1 << 31) +#define R600_GRBM_STATUS2 0x8014 +#define R600_GRBM_SOFT_RESET 0x8020 +# define R600_SOFT_RESET_CP (1 << 0) +#define R600_WAIT_UNTIL 0x8040 + +#define R600_CP_SEM_WAIT_TIMER 0x85bc +#define R600_CP_ME_CNTL 0x86d8 +# define R600_CP_ME_HALT (1 << 28) +#define R600_CP_QUEUE_THRESHOLDS 0x8760 +# define R600_ROQ_IB1_START(x) ((x) << 0) +# define R600_ROQ_IB2_START(x) ((x) << 8) +#define R600_CP_MEQ_THRESHOLDS 0x8764 +# define R700_STQ_SPLIT(x) ((x) << 0) +# define R600_MEQ_END(x) ((x) << 16) +# define R600_ROQ_END(x) ((x) << 24) +#define R600_CP_PERFMON_CNTL 0x87fc +#define R600_CP_RB_BASE 0xc100 +#define R600_CP_RB_CNTL 0xc104 +# define R600_RB_BUFSZ(x) ((x) << 0) +# define R600_RB_BLKSZ(x) ((x) << 8) +# define R600_BUF_SWAP_32BIT (2 << 16) +# define R600_RB_NO_UPDATE (1 << 27) +# define R600_RB_RPTR_WR_ENA (1 << 31) +#define R600_CP_RB_RPTR_WR 0xc108 +#define R600_CP_RB_RPTR_ADDR 0xc10c +#define R600_CP_RB_RPTR_ADDR_HI 0xc110 +#define R600_CP_RB_WPTR 0xc114 +#define R600_CP_RB_WPTR_ADDR 0xc118 +#define R600_CP_RB_WPTR_ADDR_HI 0xc11c +#define R600_CP_RB_RPTR 0x8700 +#define R600_CP_RB_WPTR_DELAY 0x8704 +#define R600_CP_PFP_UCODE_ADDR 0xc150 +#define R600_CP_PFP_UCODE_DATA 0xc154 +#define R600_CP_ME_RAM_RADDR 0xc158 +#define R600_CP_ME_RAM_WADDR 0xc15c +#define R600_CP_ME_RAM_DATA 0xc160 +#define R600_CP_DEBUG 0xc1fc + +#define R600_PA_CL_ENHANCE 0x8a14 +# define R600_CLIP_VTX_REORDER_ENA (1 << 0) +# define R600_NUM_CLIP_SEQ(x) ((x) << 1) +#define R600_PA_SC_LINE_STIPPLE_STATE 0x8b10 +#define R600_PA_SC_MULTI_CHIP_CNTL 0x8b20 +#define R700_PA_SC_FORCE_EOV_MAX_CNTS 0x8b24 +# define R700_FORCE_EOV_MAX_CLK_CNT(x) ((x) << 0) +# define R700_FORCE_EOV_MAX_REZ_CNT(x) ((x) << 16) +#define R600_PA_SC_AA_SAMPLE_LOCS_2S 0x8b40 +#define R600_PA_SC_AA_SAMPLE_LOCS_4S 0x8b44 +#define R600_PA_SC_AA_SAMPLE_LOCS_8S_WD0 0x8b48 +#define R600_PA_SC_AA_SAMPLE_LOCS_8S_WD1 0x8b4c +# define R600_S0_X(x) ((x) << 0) +# define R600_S0_Y(x) ((x) << 4) +# define R600_S1_X(x) ((x) << 8) +# define R600_S1_Y(x) ((x) << 12) +# define R600_S2_X(x) ((x) << 16) +# define R600_S2_Y(x) ((x) << 20) +# define R600_S3_X(x) ((x) << 24) +# define R600_S3_Y(x) ((x) << 28) +# define R600_S4_X(x) ((x) << 0) +# define R600_S4_Y(x) ((x) << 4) +# define R600_S5_X(x) ((x) << 8) +# define R600_S5_Y(x) ((x) << 12) +# define R600_S6_X(x) ((x) << 16) +# define R600_S6_Y(x) ((x) << 20) +# define R600_S7_X(x) ((x) << 24) +# define R600_S7_Y(x) ((x) << 28) +#define R600_PA_SC_FIFO_SIZE 0x8bd0 +# define R600_SC_PRIM_FIFO_SIZE(x) ((x) << 0) +# define R600_SC_HIZ_TILE_FIFO_SIZE(x) ((x) << 8) +# define R600_SC_EARLYZ_TILE_FIFO_SIZE(x) ((x) << 16) +#define R700_PA_SC_FIFO_SIZE_R7XX 0x8bcc +# define R700_SC_PRIM_FIFO_SIZE(x) ((x) << 0) +# define R700_SC_HIZ_TILE_FIFO_SIZE(x) ((x) << 12) +# define R700_SC_EARLYZ_TILE_FIFO_SIZE(x) ((x) << 20) +#define R600_PA_SC_ENHANCE 0x8bf0 +# define R600_FORCE_EOV_MAX_CLK_CNT(x) ((x) << 0) +# define R600_FORCE_EOV_MAX_TILE_CNT(x) ((x) << 12) +#define R600_PA_SC_CLIPRECT_RULE 0x2820c +#define R700_PA_SC_EDGERULE 0x28230 +#define R600_PA_SC_LINE_STIPPLE 0x28a0c +#define R600_PA_SC_MODE_CNTL 0x28a4c +#define R600_PA_SC_AA_CONFIG 0x28c04 + +#define R600_SX_EXPORT_BUFFER_SIZES 0x900c +# define R600_COLOR_BUFFER_SIZE(x) ((x) << 0) +# define R600_POSITION_BUFFER_SIZE(x) ((x) << 8) +# define R600_SMX_BUFFER_SIZE(x) ((x) << 16) +#define R600_SX_DEBUG_1 0x9054 +# define R600_SMX_EVENT_RELEASE (1 << 0) +# define R600_ENABLE_NEW_SMX_ADDRESS (1 << 16) +#define R700_SX_DEBUG_1 0x9058 +# define R700_ENABLE_NEW_SMX_ADDRESS (1 << 16) +#define R600_SX_MISC 0x28350 + +#define R600_DB_DEBUG 0x9830 +# define R600_PREZ_MUST_WAIT_FOR_POSTZ_DONE (1 << 31) +#define R600_DB_WATERMARKS 0x9838 +# define R600_DEPTH_FREE(x) ((x) << 0) +# define R600_DEPTH_FLUSH(x) ((x) << 5) +# define R600_DEPTH_PENDING_FREE(x) ((x) << 15) +# define R600_DEPTH_CACHELINE_FREE(x) ((x) << 20) +#define R700_DB_DEBUG3 0x98b0 +# define R700_DB_CLK_OFF_DELAY(x) ((x) << 11) +#define RV700_DB_DEBUG4 0x9b8c +# define RV700_DISABLE_TILE_COVERED_FOR_PS_ITER (1 << 6) + +#define R600_VGT_CACHE_INVALIDATION 0x88c4 +# define R600_CACHE_INVALIDATION(x) ((x) << 0) +# define R600_VC_ONLY 0 +# define R600_TC_ONLY 1 +# define R600_VC_AND_TC 2 +# define R700_AUTO_INVLD_EN(x) ((x) << 6) +# define R700_NO_AUTO 0 +# define R700_ES_AUTO 1 +# define R700_GS_AUTO 2 +# define R700_ES_AND_GS_AUTO 3 +#define R600_VGT_GS_PER_ES 0x88c8 +#define R600_VGT_ES_PER_GS 0x88cc +#define R600_VGT_GS_PER_VS 0x88e8 +#define R600_VGT_GS_VERTEX_REUSE 0x88d4 +#define R600_VGT_NUM_INSTANCES 0x8974 +#define R600_VGT_STRMOUT_EN 0x28ab0 +#define R600_VGT_EVENT_INITIATOR 0x28a90 +# define R600_CACHE_FLUSH_AND_INV_EVENT (0x16 << 0) +#define R600_VGT_VERTEX_REUSE_BLOCK_CNTL 0x28c58 +# define R600_VTX_REUSE_DEPTH_MASK 0xff +#define R600_VGT_OUT_DEALLOC_CNTL 0x28c5c +# define R600_DEALLOC_DIST_MASK 0x7f + +#define R600_CB_COLOR0_BASE 0x28040 +#define R600_CB_COLOR1_BASE 0x28044 +#define R600_CB_COLOR2_BASE 0x28048 +#define R600_CB_COLOR3_BASE 0x2804c +#define R600_CB_COLOR4_BASE 0x28050 +#define R600_CB_COLOR5_BASE 0x28054 +#define R600_CB_COLOR6_BASE 0x28058 +#define R600_CB_COLOR7_BASE 0x2805c +#define R600_CB_COLOR7_FRAG 0x280fc + +#define R600_CB_COLOR0_SIZE 0x28060 +#define R600_CB_COLOR0_VIEW 0x28080 +#define R600_CB_COLOR0_INFO 0x280a0 +#define R600_CB_COLOR0_TILE 0x280c0 +#define R600_CB_COLOR0_FRAG 0x280e0 +#define R600_CB_COLOR0_MASK 0x28100 + +#define AVIVO_D1MODE_VLINE_START_END 0x6538 +#define AVIVO_D2MODE_VLINE_START_END 0x6d38 +#define R600_CP_COHER_BASE 0x85f8 +#define R600_DB_DEPTH_BASE 0x2800c +#define R600_SQ_PGM_START_FS 0x28894 +#define R600_SQ_PGM_START_ES 0x28880 +#define R600_SQ_PGM_START_VS 0x28858 +#define R600_SQ_PGM_RESOURCES_VS 0x28868 +#define R600_SQ_PGM_CF_OFFSET_VS 0x288d0 +#define R600_SQ_PGM_START_GS 0x2886c +#define R600_SQ_PGM_START_PS 0x28840 +#define R600_SQ_PGM_RESOURCES_PS 0x28850 +#define R600_SQ_PGM_EXPORTS_PS 0x28854 +#define R600_SQ_PGM_CF_OFFSET_PS 0x288cc +#define R600_VGT_DMA_BASE 0x287e8 +#define R600_VGT_DMA_BASE_HI 0x287e4 +#define R600_VGT_STRMOUT_BASE_OFFSET_0 0x28b10 +#define R600_VGT_STRMOUT_BASE_OFFSET_1 0x28b14 +#define R600_VGT_STRMOUT_BASE_OFFSET_2 0x28b18 +#define R600_VGT_STRMOUT_BASE_OFFSET_3 0x28b1c +#define R600_VGT_STRMOUT_BASE_OFFSET_HI_0 0x28b44 +#define R600_VGT_STRMOUT_BASE_OFFSET_HI_1 0x28b48 +#define R600_VGT_STRMOUT_BASE_OFFSET_HI_2 0x28b4c +#define R600_VGT_STRMOUT_BASE_OFFSET_HI_3 0x28b50 +#define R600_VGT_STRMOUT_BUFFER_BASE_0 0x28ad8 +#define R600_VGT_STRMOUT_BUFFER_BASE_1 0x28ae8 +#define R600_VGT_STRMOUT_BUFFER_BASE_2 0x28af8 +#define R600_VGT_STRMOUT_BUFFER_BASE_3 0x28b08 +#define R600_VGT_STRMOUT_BUFFER_OFFSET_0 0x28adc +#define R600_VGT_STRMOUT_BUFFER_OFFSET_1 0x28aec +#define R600_VGT_STRMOUT_BUFFER_OFFSET_2 0x28afc +#define R600_VGT_STRMOUT_BUFFER_OFFSET_3 0x28b0c + +#define R600_VGT_PRIMITIVE_TYPE 0x8958 + +#define R600_PA_SC_SCREEN_SCISSOR_TL 0x28030 +#define R600_PA_SC_GENERIC_SCISSOR_TL 0x28240 +#define R600_PA_SC_WINDOW_SCISSOR_TL 0x28204 + +#define R600_TC_CNTL 0x9608 +# define R600_TC_L2_SIZE(x) ((x) << 5) +# define R600_L2_DISABLE_LATE_HIT (1 << 9) + +#define R600_ARB_POP 0x2418 +# define R600_ENABLE_TC128 (1 << 30) +#define R600_ARB_GDEC_RD_CNTL 0x246c + +#define R600_TA_CNTL_AUX 0x9508 +# define R600_DISABLE_CUBE_WRAP (1 << 0) +# define R600_DISABLE_CUBE_ANISO (1 << 1) +# define R700_GETLOD_SELECT(x) ((x) << 2) +# define R600_SYNC_GRADIENT (1 << 24) +# define R600_SYNC_WALKER (1 << 25) +# define R600_SYNC_ALIGNER (1 << 26) +# define R600_BILINEAR_PRECISION_6_BIT (0 << 31) +# define R600_BILINEAR_PRECISION_8_BIT (1 << 31) + +#define R700_TCP_CNTL 0x9610 + +#define R600_SMX_DC_CTL0 0xa020 +# define R700_USE_HASH_FUNCTION (1 << 0) +# define R700_CACHE_DEPTH(x) ((x) << 1) +# define R700_FLUSH_ALL_ON_EVENT (1 << 10) +# define R700_STALL_ON_EVENT (1 << 11) +#define R700_SMX_EVENT_CTL 0xa02c +# define R700_ES_FLUSH_CTL(x) ((x) << 0) +# define R700_GS_FLUSH_CTL(x) ((x) << 3) +# define R700_ACK_FLUSH_CTL(x) ((x) << 6) +# define R700_SYNC_FLUSH_CTL (1 << 8) + +#define R600_SQ_CONFIG 0x8c00 +# define R600_VC_ENABLE (1 << 0) +# define R600_EXPORT_SRC_C (1 << 1) +# define R600_DX9_CONSTS (1 << 2) +# define R600_ALU_INST_PREFER_VECTOR (1 << 3) +# define R600_DX10_CLAMP (1 << 4) +# define R600_CLAUSE_SEQ_PRIO(x) ((x) << 8) +# define R600_PS_PRIO(x) ((x) << 24) +# define R600_VS_PRIO(x) ((x) << 26) +# define R600_GS_PRIO(x) ((x) << 28) +# define R600_ES_PRIO(x) ((x) << 30) +#define R600_SQ_GPR_RESOURCE_MGMT_1 0x8c04 +# define R600_NUM_PS_GPRS(x) ((x) << 0) +# define R600_NUM_VS_GPRS(x) ((x) << 16) +# define R700_DYN_GPR_ENABLE (1 << 27) +# define R600_NUM_CLAUSE_TEMP_GPRS(x) ((x) << 28) +#define R600_SQ_GPR_RESOURCE_MGMT_2 0x8c08 +# define R600_NUM_GS_GPRS(x) ((x) << 0) +# define R600_NUM_ES_GPRS(x) ((x) << 16) +#define R600_SQ_THREAD_RESOURCE_MGMT 0x8c0c +# define R600_NUM_PS_THREADS(x) ((x) << 0) +# define R600_NUM_VS_THREADS(x) ((x) << 8) +# define R600_NUM_GS_THREADS(x) ((x) << 16) +# define R600_NUM_ES_THREADS(x) ((x) << 24) +#define R600_SQ_STACK_RESOURCE_MGMT_1 0x8c10 +# define R600_NUM_PS_STACK_ENTRIES(x) ((x) << 0) +# define R600_NUM_VS_STACK_ENTRIES(x) ((x) << 16) +#define R600_SQ_STACK_RESOURCE_MGMT_2 0x8c14 +# define R600_NUM_GS_STACK_ENTRIES(x) ((x) << 0) +# define R600_NUM_ES_STACK_ENTRIES(x) ((x) << 16) +#define R600_SQ_MS_FIFO_SIZES 0x8cf0 +# define R600_CACHE_FIFO_SIZE(x) ((x) << 0) +# define R600_FETCH_FIFO_HIWATER(x) ((x) << 8) +# define R600_DONE_FIFO_HIWATER(x) ((x) << 16) +# define R600_ALU_UPDATE_FIFO_HIWATER(x) ((x) << 24) +#define R700_SQ_DYN_GPR_SIZE_SIMD_AB_0 0x8db0 +# define R700_SIMDA_RING0(x) ((x) << 0) +# define R700_SIMDA_RING1(x) ((x) << 8) +# define R700_SIMDB_RING0(x) ((x) << 16) +# define R700_SIMDB_RING1(x) ((x) << 24) +#define R700_SQ_DYN_GPR_SIZE_SIMD_AB_1 0x8db4 +#define R700_SQ_DYN_GPR_SIZE_SIMD_AB_2 0x8db8 +#define R700_SQ_DYN_GPR_SIZE_SIMD_AB_3 0x8dbc +#define R700_SQ_DYN_GPR_SIZE_SIMD_AB_4 0x8dc0 +#define R700_SQ_DYN_GPR_SIZE_SIMD_AB_5 0x8dc4 +#define R700_SQ_DYN_GPR_SIZE_SIMD_AB_6 0x8dc8 +#define R700_SQ_DYN_GPR_SIZE_SIMD_AB_7 0x8dcc + +#define R600_SPI_PS_IN_CONTROL_0 0x286cc +# define R600_NUM_INTERP(x) ((x) << 0) +# define R600_POSITION_ENA (1 << 8) +# define R600_POSITION_CENTROID (1 << 9) +# define R600_POSITION_ADDR(x) ((x) << 10) +# define R600_PARAM_GEN(x) ((x) << 15) +# define R600_PARAM_GEN_ADDR(x) ((x) << 19) +# define R600_BARYC_SAMPLE_CNTL(x) ((x) << 26) +# define R600_PERSP_GRADIENT_ENA (1 << 28) +# define R600_LINEAR_GRADIENT_ENA (1 << 29) +# define R600_POSITION_SAMPLE (1 << 30) +# define R600_BARYC_AT_SAMPLE_ENA (1 << 31) +#define R600_SPI_PS_IN_CONTROL_1 0x286d0 +# define R600_GEN_INDEX_PIX (1 << 0) +# define R600_GEN_INDEX_PIX_ADDR(x) ((x) << 1) +# define R600_FRONT_FACE_ENA (1 << 8) +# define R600_FRONT_FACE_CHAN(x) ((x) << 9) +# define R600_FRONT_FACE_ALL_BITS (1 << 11) +# define R600_FRONT_FACE_ADDR(x) ((x) << 12) +# define R600_FOG_ADDR(x) ((x) << 17) +# define R600_FIXED_PT_POSITION_ENA (1 << 24) +# define R600_FIXED_PT_POSITION_ADDR(x) ((x) << 25) +# define R700_POSITION_ULC (1 << 30) +#define R600_SPI_INPUT_Z 0x286d8 + +#define R600_SPI_CONFIG_CNTL 0x9100 +# define R600_GPR_WRITE_PRIORITY(x) ((x) << 0) +# define R600_DISABLE_INTERP_1 (1 << 5) +#define R600_SPI_CONFIG_CNTL_1 0x913c +# define R600_VTX_DONE_DELAY(x) ((x) << 0) +# define R600_INTERP_ONE_PRIM_PER_ROW (1 << 4) + +#define R600_GB_TILING_CONFIG 0x98f0 +# define R600_PIPE_TILING(x) ((x) << 1) +# define R600_BANK_TILING(x) ((x) << 4) +# define R600_GROUP_SIZE(x) ((x) << 6) +# define R600_ROW_TILING(x) ((x) << 8) +# define R600_BANK_SWAPS(x) ((x) << 11) +# define R600_SAMPLE_SPLIT(x) ((x) << 14) +# define R600_BACKEND_MAP(x) ((x) << 16) +#define R600_DCP_TILING_CONFIG 0x6ca0 +#define R600_HDP_TILING_CONFIG 0x2f3c + +#define R600_CC_RB_BACKEND_DISABLE 0x98f4 +#define R700_CC_SYS_RB_BACKEND_DISABLE 0x3f88 +# define R600_BACKEND_DISABLE(x) ((x) << 16) + +#define R600_CC_GC_SHADER_PIPE_CONFIG 0x8950 +#define R600_GC_USER_SHADER_PIPE_CONFIG 0x8954 +# define R600_INACTIVE_QD_PIPES(x) ((x) << 8) +# define R600_INACTIVE_QD_PIPES_MASK (0xff << 8) +# define R600_INACTIVE_SIMDS(x) ((x) << 16) +# define R600_INACTIVE_SIMDS_MASK (0xff << 16) + +#define R700_CGTS_SYS_TCC_DISABLE 0x3f90 +#define R700_CGTS_USER_SYS_TCC_DISABLE 0x3f94 +#define R700_CGTS_TCC_DISABLE 0x9148 +#define R700_CGTS_USER_TCC_DISABLE 0x914c + +/* Constants */ +#define RADEON_MAX_USEC_TIMEOUT 100000 /* 100 ms */ + +#define RADEON_LAST_FRAME_REG RADEON_SCRATCH_REG0 +#define RADEON_LAST_DISPATCH_REG RADEON_SCRATCH_REG1 +#define RADEON_LAST_CLEAR_REG RADEON_SCRATCH_REG2 +#define RADEON_LAST_SWI_REG RADEON_SCRATCH_REG3 +#define RADEON_LAST_DISPATCH 1 + +#define R600_LAST_FRAME_REG R600_SCRATCH_REG0 +#define R600_LAST_DISPATCH_REG R600_SCRATCH_REG1 +#define R600_LAST_CLEAR_REG R600_SCRATCH_REG2 +#define R600_LAST_SWI_REG R600_SCRATCH_REG3 + +#define RADEON_MAX_VB_AGE 0x7fffffff +#define RADEON_MAX_VB_VERTS (0xffff) + +#define RADEON_RING_HIGH_MARK 128 + +#define RADEON_PCIGART_TABLE_SIZE (32*1024) + +#define RADEON_READ(reg) DRM_READ32( dev_priv->mmio, (reg) ) +#define RADEON_WRITE(reg, val) \ +do { \ + if (reg < 0x10000) { \ + DRM_WRITE32(dev_priv->mmio, (reg), (val)); \ + } else { \ + DRM_WRITE32(dev_priv->mmio, RADEON_MM_INDEX, (reg)); \ + DRM_WRITE32(dev_priv->mmio, RADEON_MM_DATA, (val)); \ + } \ +} while (0) +#define RADEON_READ8(reg) DRM_READ8( dev_priv->mmio, (reg) ) +#define RADEON_WRITE8(reg,val) DRM_WRITE8( dev_priv->mmio, (reg), (val) ) + +#define RADEON_WRITE_PLL(addr, val) \ +do { \ + RADEON_WRITE8(RADEON_CLOCK_CNTL_INDEX, \ + ((addr) & 0x1f) | RADEON_PLL_WR_EN ); \ + RADEON_WRITE(RADEON_CLOCK_CNTL_DATA, (val)); \ +} while (0) + +#define RADEON_WRITE_PCIE(addr, val) \ +do { \ + RADEON_WRITE8(RADEON_PCIE_INDEX, \ + ((addr) & 0xff)); \ + RADEON_WRITE(RADEON_PCIE_DATA, (val)); \ +} while (0) + +#define R500_WRITE_MCIND(addr, val) \ +do { \ + RADEON_WRITE(R520_MC_IND_INDEX, 0xff0000 | ((addr) & 0xff)); \ + RADEON_WRITE(R520_MC_IND_DATA, (val)); \ + RADEON_WRITE(R520_MC_IND_INDEX, 0); \ +} while (0) + +#define RS480_WRITE_MCIND(addr, val) \ +do { \ + RADEON_WRITE(RS480_NB_MC_INDEX, \ + ((addr) & 0xff) | RS480_NB_MC_IND_WR_EN); \ + RADEON_WRITE(RS480_NB_MC_DATA, (val)); \ + RADEON_WRITE(RS480_NB_MC_INDEX, 0xff); \ +} while (0) + +#define RS690_WRITE_MCIND(addr, val) \ +do { \ + RADEON_WRITE(RS690_MC_INDEX, RS690_MC_INDEX_WR_EN | ((addr) & RS690_MC_INDEX_MASK)); \ + RADEON_WRITE(RS690_MC_DATA, val); \ + RADEON_WRITE(RS690_MC_INDEX, RS690_MC_INDEX_WR_ACK); \ +} while (0) + +#define RS600_WRITE_MCIND(addr, val) \ +do { \ + RADEON_WRITE(RS600_MC_INDEX, RS600_MC_IND_WR_EN | RS600_MC_IND_CITF_ARB0 | ((addr) & RS600_MC_ADDR_MASK)); \ + RADEON_WRITE(RS600_MC_DATA, val); \ +} while (0) + +#define IGP_WRITE_MCIND(addr, val) \ +do { \ + if (((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS690) || \ + ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS740)) \ + RS690_WRITE_MCIND(addr, val); \ + else if ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS600) \ + RS600_WRITE_MCIND(addr, val); \ + else \ + RS480_WRITE_MCIND(addr, val); \ +} while (0) + +#define CP_PACKET0( reg, n ) \ + (RADEON_CP_PACKET0 | ((n) << 16) | ((reg) >> 2)) +#define CP_PACKET0_TABLE( reg, n ) \ + (RADEON_CP_PACKET0 | RADEON_ONE_REG_WR | ((n) << 16) | ((reg) >> 2)) +#define CP_PACKET1( reg0, reg1 ) \ + (RADEON_CP_PACKET1 | (((reg1) >> 2) << 15) | ((reg0) >> 2)) +#define CP_PACKET2() \ + (RADEON_CP_PACKET2) +#define CP_PACKET3( pkt, n ) \ + (RADEON_CP_PACKET3 | (pkt) | ((n) << 16)) + +/* ================================================================ + * Engine control helper macros + */ + +#define RADEON_WAIT_UNTIL_2D_IDLE() do { \ + OUT_RING( CP_PACKET0( RADEON_WAIT_UNTIL, 0 ) ); \ + OUT_RING( (RADEON_WAIT_2D_IDLECLEAN | \ + RADEON_WAIT_HOST_IDLECLEAN) ); \ +} while (0) + +#define RADEON_WAIT_UNTIL_3D_IDLE() do { \ + OUT_RING( CP_PACKET0( RADEON_WAIT_UNTIL, 0 ) ); \ + OUT_RING( (RADEON_WAIT_3D_IDLECLEAN | \ + RADEON_WAIT_HOST_IDLECLEAN) ); \ +} while (0) + +#define RADEON_WAIT_UNTIL_IDLE() do { \ + OUT_RING( CP_PACKET0( RADEON_WAIT_UNTIL, 0 ) ); \ + OUT_RING( (RADEON_WAIT_2D_IDLECLEAN | \ + RADEON_WAIT_3D_IDLECLEAN | \ + RADEON_WAIT_HOST_IDLECLEAN) ); \ +} while (0) + +#define RADEON_WAIT_UNTIL_PAGE_FLIPPED() do { \ + OUT_RING( CP_PACKET0( RADEON_WAIT_UNTIL, 0 ) ); \ + OUT_RING( RADEON_WAIT_CRTC_PFLIP ); \ +} while (0) + +#define RADEON_FLUSH_CACHE() do { \ + if ((dev_priv->flags & RADEON_FAMILY_MASK) <= CHIP_RV280) { \ + OUT_RING(CP_PACKET0(RADEON_RB3D_DSTCACHE_CTLSTAT, 0)); \ + OUT_RING(RADEON_RB3D_DC_FLUSH); \ + } else { \ + OUT_RING(CP_PACKET0(R300_RB3D_DSTCACHE_CTLSTAT, 0)); \ + OUT_RING(R300_RB3D_DC_FLUSH); \ + } \ +} while (0) + +#define RADEON_PURGE_CACHE() do { \ + if ((dev_priv->flags & RADEON_FAMILY_MASK) <= CHIP_RV280) { \ + OUT_RING(CP_PACKET0(RADEON_RB3D_DSTCACHE_CTLSTAT, 0)); \ + OUT_RING(RADEON_RB3D_DC_FLUSH | RADEON_RB3D_DC_FREE); \ + } else { \ + OUT_RING(CP_PACKET0(R300_RB3D_DSTCACHE_CTLSTAT, 0)); \ + OUT_RING(R300_RB3D_DC_FLUSH | R300_RB3D_DC_FREE); \ + } \ +} while (0) + +#define RADEON_FLUSH_ZCACHE() do { \ + if ((dev_priv->flags & RADEON_FAMILY_MASK) <= CHIP_RV280) { \ + OUT_RING(CP_PACKET0(RADEON_RB3D_ZCACHE_CTLSTAT, 0)); \ + OUT_RING(RADEON_RB3D_ZC_FLUSH); \ + } else { \ + OUT_RING(CP_PACKET0(R300_ZB_ZCACHE_CTLSTAT, 0)); \ + OUT_RING(R300_ZC_FLUSH); \ + } \ +} while (0) + +#define RADEON_PURGE_ZCACHE() do { \ + if ((dev_priv->flags & RADEON_FAMILY_MASK) <= CHIP_RV280) { \ + OUT_RING(CP_PACKET0(RADEON_RB3D_ZCACHE_CTLSTAT, 0)); \ + OUT_RING(RADEON_RB3D_ZC_FLUSH | RADEON_RB3D_ZC_FREE); \ + } else { \ + OUT_RING(CP_PACKET0(R300_ZB_ZCACHE_CTLSTAT, 0)); \ + OUT_RING(R300_ZC_FLUSH | R300_ZC_FREE); \ + } \ +} while (0) + +/* ================================================================ + * Misc helper macros + */ + +/* Perfbox functionality only. + */ +#define RING_SPACE_TEST_WITH_RETURN( dev_priv ) \ +do { \ + if (!(dev_priv->stats.boxes & RADEON_BOX_DMA_IDLE)) { \ + u32 head = GET_RING_HEAD( dev_priv ); \ + if (head == dev_priv->ring.tail) \ + dev_priv->stats.boxes |= RADEON_BOX_DMA_IDLE; \ + } \ +} while (0) + +#define VB_AGE_TEST_WITH_RETURN( dev_priv ) \ +do { \ + struct drm_radeon_master_private *master_priv = file_priv->master->driver_priv; \ + drm_radeon_sarea_t *sarea_priv = master_priv->sarea_priv; \ + if ( sarea_priv->last_dispatch >= RADEON_MAX_VB_AGE ) { \ + int __ret; \ + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) \ + __ret = r600_do_cp_idle(dev_priv); \ + else \ + __ret = radeon_do_cp_idle(dev_priv); \ + if ( __ret ) return __ret; \ + sarea_priv->last_dispatch = 0; \ + radeon_freelist_reset( dev ); \ + } \ +} while (0) + +#define RADEON_DISPATCH_AGE( age ) do { \ + OUT_RING( CP_PACKET0( RADEON_LAST_DISPATCH_REG, 0 ) ); \ + OUT_RING( age ); \ +} while (0) + +#define RADEON_FRAME_AGE( age ) do { \ + OUT_RING( CP_PACKET0( RADEON_LAST_FRAME_REG, 0 ) ); \ + OUT_RING( age ); \ +} while (0) + +#define RADEON_CLEAR_AGE( age ) do { \ + OUT_RING( CP_PACKET0( RADEON_LAST_CLEAR_REG, 0 ) ); \ + OUT_RING( age ); \ +} while (0) + +#define R600_DISPATCH_AGE(age) do { \ + OUT_RING(CP_PACKET3(R600_IT_SET_CONFIG_REG, 1)); \ + OUT_RING((R600_LAST_DISPATCH_REG - R600_SET_CONFIG_REG_OFFSET) >> 2); \ + OUT_RING(age); \ +} while (0) + +#define R600_FRAME_AGE(age) do { \ + OUT_RING(CP_PACKET3(R600_IT_SET_CONFIG_REG, 1)); \ + OUT_RING((R600_LAST_FRAME_REG - R600_SET_CONFIG_REG_OFFSET) >> 2); \ + OUT_RING(age); \ +} while (0) + +#define R600_CLEAR_AGE(age) do { \ + OUT_RING(CP_PACKET3(R600_IT_SET_CONFIG_REG, 1)); \ + OUT_RING((R600_LAST_CLEAR_REG - R600_SET_CONFIG_REG_OFFSET) >> 2); \ + OUT_RING(age); \ +} while (0) + +/* ================================================================ + * Ring control + */ + +#define RADEON_VERBOSE 0 + +#define RING_LOCALS int write, _nr, _align_nr; unsigned int mask; u32 *ring; + +#define RADEON_RING_ALIGN 16 + +#define BEGIN_RING( n ) do { \ + if ( RADEON_VERBOSE ) { \ + DRM_INFO( "BEGIN_RING( %d )\n", (n)); \ + } \ + _align_nr = RADEON_RING_ALIGN - ((dev_priv->ring.tail + n) & (RADEON_RING_ALIGN-1)); \ + _align_nr += n; \ + if (dev_priv->ring.space <= (_align_nr * sizeof(u32))) { \ + COMMIT_RING(); \ + radeon_wait_ring( dev_priv, _align_nr * sizeof(u32)); \ + } \ + _nr = n; dev_priv->ring.space -= (n) * sizeof(u32); \ + ring = dev_priv->ring.start; \ + write = dev_priv->ring.tail; \ + mask = dev_priv->ring.tail_mask; \ +} while (0) + +#define ADVANCE_RING() do { \ + if ( RADEON_VERBOSE ) { \ + DRM_INFO( "ADVANCE_RING() wr=0x%06x tail=0x%06x\n", \ + write, dev_priv->ring.tail ); \ + } \ + if (((dev_priv->ring.tail + _nr) & mask) != write) { \ + DRM_ERROR( \ + "ADVANCE_RING(): mismatch: nr: %x write: %x line: %d\n", \ + ((dev_priv->ring.tail + _nr) & mask), \ + write, __LINE__); \ + } else \ + dev_priv->ring.tail = write; \ +} while (0) + +extern void radeon_commit_ring(drm_radeon_private_t *dev_priv); + +#define COMMIT_RING() do { \ + radeon_commit_ring(dev_priv); \ + } while(0) + +#define OUT_RING( x ) do { \ + if ( RADEON_VERBOSE ) { \ + DRM_INFO( " OUT_RING( 0x%08x ) at 0x%x\n", \ + (unsigned int)(x), write ); \ + } \ + ring[write++] = (x); \ + write &= mask; \ +} while (0) + +#define OUT_RING_REG( reg, val ) do { \ + OUT_RING( CP_PACKET0( reg, 0 ) ); \ + OUT_RING( val ); \ +} while (0) + +#define OUT_RING_TABLE( tab, sz ) do { \ + int _size = (sz); \ + int *_tab = (int *)(tab); \ + \ + if (write + _size > mask) { \ + int _i = (mask+1) - write; \ + _size -= _i; \ + while (_i > 0 ) { \ + *(int *)(ring + write) = *_tab++; \ + write++; \ + _i--; \ + } \ + write = 0; \ + _tab += _i; \ + } \ + while (_size > 0) { \ + *(ring + write) = *_tab++; \ + write++; \ + _size--; \ + } \ + write &= mask; \ +} while (0) + +/** + * Copy given number of dwords from drm buffer to the ring buffer. + */ +#define OUT_RING_DRM_BUFFER(buf, sz) do { \ + int _size = (sz) * 4; \ + struct drm_buffer *_buf = (buf); \ + int _part_size; \ + while (_size > 0) { \ + _part_size = _size; \ + \ + if (write + _part_size/4 > mask) \ + _part_size = ((mask + 1) - write)*4; \ + \ + if (drm_buffer_index(_buf) + _part_size > PAGE_SIZE) \ + _part_size = PAGE_SIZE - drm_buffer_index(_buf);\ + \ + \ + \ + memcpy(ring + write, &_buf->data[drm_buffer_page(_buf)] \ + [drm_buffer_index(_buf)], _part_size); \ + \ + _size -= _part_size; \ + write = (write + _part_size/4) & mask; \ + drm_buffer_advance(_buf, _part_size); \ + } \ +} while (0) + + +#endif /* __RADEON_DRV_H__ */ diff --git a/drivers/gpu/drm/radeon/radeon_encoders.c b/drivers/gpu/drm/radeon/radeon_encoders.c new file mode 100644 index 00000000..8a171b21 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_encoders.c @@ -0,0 +1,2421 @@ +/* + * Copyright 2007-8 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + */ +#include "drmP.h" +#include "drm_crtc_helper.h" +#include "radeon_drm.h" +#include "radeon.h" +#include "atom.h" + +extern int atom_debug; + +/* evil but including atombios.h is much worse */ +bool radeon_atom_get_tv_timings(struct radeon_device *rdev, int index, + struct drm_display_mode *mode); + +static uint32_t radeon_encoder_clones(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct drm_encoder *clone_encoder; + uint32_t index_mask = 0; + int count; + + /* DIG routing gets problematic */ + if (rdev->family >= CHIP_R600) + return index_mask; + /* LVDS/TV are too wacky */ + if (radeon_encoder->devices & ATOM_DEVICE_LCD_SUPPORT) + return index_mask; + /* DVO requires 2x ppll clocks depending on tmds chip */ + if (radeon_encoder->devices & ATOM_DEVICE_DFP2_SUPPORT) + return index_mask; + + count = -1; + list_for_each_entry(clone_encoder, &dev->mode_config.encoder_list, head) { + struct radeon_encoder *radeon_clone = to_radeon_encoder(clone_encoder); + count++; + + if (clone_encoder == encoder) + continue; + if (radeon_clone->devices & (ATOM_DEVICE_LCD_SUPPORT)) + continue; + if (radeon_clone->devices & ATOM_DEVICE_DFP2_SUPPORT) + continue; + else + index_mask |= (1 << count); + } + return index_mask; +} + +void radeon_setup_encoder_clones(struct drm_device *dev) +{ + struct drm_encoder *encoder; + + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + encoder->possible_clones = radeon_encoder_clones(encoder); + } +} + +uint32_t +radeon_get_encoder_enum(struct drm_device *dev, uint32_t supported_device, uint8_t dac) +{ + struct radeon_device *rdev = dev->dev_private; + uint32_t ret = 0; + + switch (supported_device) { + case ATOM_DEVICE_CRT1_SUPPORT: + case ATOM_DEVICE_TV1_SUPPORT: + case ATOM_DEVICE_TV2_SUPPORT: + case ATOM_DEVICE_CRT2_SUPPORT: + case ATOM_DEVICE_CV_SUPPORT: + switch (dac) { + case 1: /* dac a */ + if ((rdev->family == CHIP_RS300) || + (rdev->family == CHIP_RS400) || + (rdev->family == CHIP_RS480)) + ret = ENCODER_INTERNAL_DAC2_ENUM_ID1; + else if (ASIC_IS_AVIVO(rdev)) + ret = ENCODER_INTERNAL_KLDSCP_DAC1_ENUM_ID1; + else + ret = ENCODER_INTERNAL_DAC1_ENUM_ID1; + break; + case 2: /* dac b */ + if (ASIC_IS_AVIVO(rdev)) + ret = ENCODER_INTERNAL_KLDSCP_DAC2_ENUM_ID1; + else { + /*if (rdev->family == CHIP_R200) + ret = ENCODER_INTERNAL_DVO1_ENUM_ID1; + else*/ + ret = ENCODER_INTERNAL_DAC2_ENUM_ID1; + } + break; + case 3: /* external dac */ + if (ASIC_IS_AVIVO(rdev)) + ret = ENCODER_INTERNAL_KLDSCP_DVO1_ENUM_ID1; + else + ret = ENCODER_INTERNAL_DVO1_ENUM_ID1; + break; + } + break; + case ATOM_DEVICE_LCD1_SUPPORT: + if (ASIC_IS_AVIVO(rdev)) + ret = ENCODER_INTERNAL_LVTM1_ENUM_ID1; + else + ret = ENCODER_INTERNAL_LVDS_ENUM_ID1; + break; + case ATOM_DEVICE_DFP1_SUPPORT: + if ((rdev->family == CHIP_RS300) || + (rdev->family == CHIP_RS400) || + (rdev->family == CHIP_RS480)) + ret = ENCODER_INTERNAL_DVO1_ENUM_ID1; + else if (ASIC_IS_AVIVO(rdev)) + ret = ENCODER_INTERNAL_KLDSCP_TMDS1_ENUM_ID1; + else + ret = ENCODER_INTERNAL_TMDS1_ENUM_ID1; + break; + case ATOM_DEVICE_LCD2_SUPPORT: + case ATOM_DEVICE_DFP2_SUPPORT: + if ((rdev->family == CHIP_RS600) || + (rdev->family == CHIP_RS690) || + (rdev->family == CHIP_RS740)) + ret = ENCODER_INTERNAL_DDI_ENUM_ID1; + else if (ASIC_IS_AVIVO(rdev)) + ret = ENCODER_INTERNAL_KLDSCP_DVO1_ENUM_ID1; + else + ret = ENCODER_INTERNAL_DVO1_ENUM_ID1; + break; + case ATOM_DEVICE_DFP3_SUPPORT: + ret = ENCODER_INTERNAL_LVTM1_ENUM_ID1; + break; + } + + return ret; +} + +static inline bool radeon_encoder_is_digital(struct drm_encoder *encoder) +{ + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + switch (radeon_encoder->encoder_id) { + case ENCODER_OBJECT_ID_INTERNAL_LVDS: + case ENCODER_OBJECT_ID_INTERNAL_TMDS1: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_TMDS1: + case ENCODER_OBJECT_ID_INTERNAL_LVTM1: + case ENCODER_OBJECT_ID_INTERNAL_DVO1: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1: + case ENCODER_OBJECT_ID_INTERNAL_DDI: + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA: + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1: + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2: + return true; + default: + return false; + } +} + +void +radeon_link_encoder_connector(struct drm_device *dev) +{ + struct drm_connector *connector; + struct radeon_connector *radeon_connector; + struct drm_encoder *encoder; + struct radeon_encoder *radeon_encoder; + + /* walk the list and link encoders to connectors */ + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + radeon_connector = to_radeon_connector(connector); + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + radeon_encoder = to_radeon_encoder(encoder); + if (radeon_encoder->devices & radeon_connector->devices) + drm_mode_connector_attach_encoder(connector, encoder); + } + } +} + +void radeon_encoder_set_active_device(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct drm_connector *connector; + + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + if (connector->encoder == encoder) { + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + radeon_encoder->active_device = radeon_encoder->devices & radeon_connector->devices; + DRM_DEBUG_KMS("setting active device to %08x from %08x %08x for encoder %d\n", + radeon_encoder->active_device, radeon_encoder->devices, + radeon_connector->devices, encoder->encoder_type); + } + } +} + +struct drm_connector * +radeon_get_connector_for_encoder(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct drm_connector *connector; + struct radeon_connector *radeon_connector; + + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + radeon_connector = to_radeon_connector(connector); + if (radeon_encoder->active_device & radeon_connector->devices) + return connector; + } + return NULL; +} + +static struct drm_connector * +radeon_get_connector_for_encoder_init(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct drm_connector *connector; + struct radeon_connector *radeon_connector; + + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + radeon_connector = to_radeon_connector(connector); + if (radeon_encoder->devices & radeon_connector->devices) + return connector; + } + return NULL; +} + +struct drm_encoder *radeon_atom_get_external_encoder(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct drm_encoder *other_encoder; + struct radeon_encoder *other_radeon_encoder; + + if (radeon_encoder->is_ext_encoder) + return NULL; + + list_for_each_entry(other_encoder, &dev->mode_config.encoder_list, head) { + if (other_encoder == encoder) + continue; + other_radeon_encoder = to_radeon_encoder(other_encoder); + if (other_radeon_encoder->is_ext_encoder && + (radeon_encoder->devices & other_radeon_encoder->devices)) + return other_encoder; + } + return NULL; +} + +bool radeon_encoder_is_dp_bridge(struct drm_encoder *encoder) +{ + struct drm_encoder *other_encoder = radeon_atom_get_external_encoder(encoder); + + if (other_encoder) { + struct radeon_encoder *radeon_encoder = to_radeon_encoder(other_encoder); + + switch (radeon_encoder->encoder_id) { + case ENCODER_OBJECT_ID_TRAVIS: + case ENCODER_OBJECT_ID_NUTMEG: + return true; + default: + return false; + } + } + + return false; +} + +void radeon_panel_mode_fixup(struct drm_encoder *encoder, + struct drm_display_mode *adjusted_mode) +{ + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct drm_display_mode *native_mode = &radeon_encoder->native_mode; + unsigned hblank = native_mode->htotal - native_mode->hdisplay; + unsigned vblank = native_mode->vtotal - native_mode->vdisplay; + unsigned hover = native_mode->hsync_start - native_mode->hdisplay; + unsigned vover = native_mode->vsync_start - native_mode->vdisplay; + unsigned hsync_width = native_mode->hsync_end - native_mode->hsync_start; + unsigned vsync_width = native_mode->vsync_end - native_mode->vsync_start; + + adjusted_mode->clock = native_mode->clock; + adjusted_mode->flags = native_mode->flags; + + if (ASIC_IS_AVIVO(rdev)) { + adjusted_mode->hdisplay = native_mode->hdisplay; + adjusted_mode->vdisplay = native_mode->vdisplay; + } + + adjusted_mode->htotal = native_mode->hdisplay + hblank; + adjusted_mode->hsync_start = native_mode->hdisplay + hover; + adjusted_mode->hsync_end = adjusted_mode->hsync_start + hsync_width; + + adjusted_mode->vtotal = native_mode->vdisplay + vblank; + adjusted_mode->vsync_start = native_mode->vdisplay + vover; + adjusted_mode->vsync_end = adjusted_mode->vsync_start + vsync_width; + + drm_mode_set_crtcinfo(adjusted_mode, CRTC_INTERLACE_HALVE_V); + + if (ASIC_IS_AVIVO(rdev)) { + adjusted_mode->crtc_hdisplay = native_mode->hdisplay; + adjusted_mode->crtc_vdisplay = native_mode->vdisplay; + } + + adjusted_mode->crtc_htotal = adjusted_mode->crtc_hdisplay + hblank; + adjusted_mode->crtc_hsync_start = adjusted_mode->crtc_hdisplay + hover; + adjusted_mode->crtc_hsync_end = adjusted_mode->crtc_hsync_start + hsync_width; + + adjusted_mode->crtc_vtotal = adjusted_mode->crtc_vdisplay + vblank; + adjusted_mode->crtc_vsync_start = adjusted_mode->crtc_vdisplay + vover; + adjusted_mode->crtc_vsync_end = adjusted_mode->crtc_vsync_start + vsync_width; + +} + +static bool radeon_atom_mode_fixup(struct drm_encoder *encoder, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + + /* set the active encoder to connector routing */ + radeon_encoder_set_active_device(encoder); + drm_mode_set_crtcinfo(adjusted_mode, 0); + + /* hw bug */ + if ((mode->flags & DRM_MODE_FLAG_INTERLACE) + && (mode->crtc_vsync_start < (mode->crtc_vdisplay + 2))) + adjusted_mode->crtc_vsync_start = adjusted_mode->crtc_vdisplay + 2; + + /* get the native mode for LVDS */ + if (radeon_encoder->active_device & (ATOM_DEVICE_LCD_SUPPORT)) + radeon_panel_mode_fixup(encoder, adjusted_mode); + + /* get the native mode for TV */ + if (radeon_encoder->active_device & (ATOM_DEVICE_TV_SUPPORT)) { + struct radeon_encoder_atom_dac *tv_dac = radeon_encoder->enc_priv; + if (tv_dac) { + if (tv_dac->tv_std == TV_STD_NTSC || + tv_dac->tv_std == TV_STD_NTSC_J || + tv_dac->tv_std == TV_STD_PAL_M) + radeon_atom_get_tv_timings(rdev, 0, adjusted_mode); + else + radeon_atom_get_tv_timings(rdev, 1, adjusted_mode); + } + } + + if (ASIC_IS_DCE3(rdev) && + ((radeon_encoder->active_device & (ATOM_DEVICE_DFP_SUPPORT | ATOM_DEVICE_LCD_SUPPORT)) || + radeon_encoder_is_dp_bridge(encoder))) { + struct drm_connector *connector = radeon_get_connector_for_encoder(encoder); + radeon_dp_set_link_config(connector, mode); + } + + return true; +} + +static void +atombios_dac_setup(struct drm_encoder *encoder, int action) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + DAC_ENCODER_CONTROL_PS_ALLOCATION args; + int index = 0; + struct radeon_encoder_atom_dac *dac_info = radeon_encoder->enc_priv; + + memset(&args, 0, sizeof(args)); + + switch (radeon_encoder->encoder_id) { + case ENCODER_OBJECT_ID_INTERNAL_DAC1: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1: + index = GetIndexIntoMasterTable(COMMAND, DAC1EncoderControl); + break; + case ENCODER_OBJECT_ID_INTERNAL_DAC2: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2: + index = GetIndexIntoMasterTable(COMMAND, DAC2EncoderControl); + break; + } + + args.ucAction = action; + + if (radeon_encoder->active_device & (ATOM_DEVICE_CRT_SUPPORT)) + args.ucDacStandard = ATOM_DAC1_PS2; + else if (radeon_encoder->active_device & (ATOM_DEVICE_CV_SUPPORT)) + args.ucDacStandard = ATOM_DAC1_CV; + else { + switch (dac_info->tv_std) { + case TV_STD_PAL: + case TV_STD_PAL_M: + case TV_STD_SCART_PAL: + case TV_STD_SECAM: + case TV_STD_PAL_CN: + args.ucDacStandard = ATOM_DAC1_PAL; + break; + case TV_STD_NTSC: + case TV_STD_NTSC_J: + case TV_STD_PAL_60: + default: + args.ucDacStandard = ATOM_DAC1_NTSC; + break; + } + } + args.usPixelClock = cpu_to_le16(radeon_encoder->pixel_clock / 10); + + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); + +} + +static void +atombios_tv_setup(struct drm_encoder *encoder, int action) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + TV_ENCODER_CONTROL_PS_ALLOCATION args; + int index = 0; + struct radeon_encoder_atom_dac *dac_info = radeon_encoder->enc_priv; + + memset(&args, 0, sizeof(args)); + + index = GetIndexIntoMasterTable(COMMAND, TVEncoderControl); + + args.sTVEncoder.ucAction = action; + + if (radeon_encoder->active_device & (ATOM_DEVICE_CV_SUPPORT)) + args.sTVEncoder.ucTvStandard = ATOM_TV_CV; + else { + switch (dac_info->tv_std) { + case TV_STD_NTSC: + args.sTVEncoder.ucTvStandard = ATOM_TV_NTSC; + break; + case TV_STD_PAL: + args.sTVEncoder.ucTvStandard = ATOM_TV_PAL; + break; + case TV_STD_PAL_M: + args.sTVEncoder.ucTvStandard = ATOM_TV_PALM; + break; + case TV_STD_PAL_60: + args.sTVEncoder.ucTvStandard = ATOM_TV_PAL60; + break; + case TV_STD_NTSC_J: + args.sTVEncoder.ucTvStandard = ATOM_TV_NTSCJ; + break; + case TV_STD_SCART_PAL: + args.sTVEncoder.ucTvStandard = ATOM_TV_PAL; /* ??? */ + break; + case TV_STD_SECAM: + args.sTVEncoder.ucTvStandard = ATOM_TV_SECAM; + break; + case TV_STD_PAL_CN: + args.sTVEncoder.ucTvStandard = ATOM_TV_PALCN; + break; + default: + args.sTVEncoder.ucTvStandard = ATOM_TV_NTSC; + break; + } + } + + args.sTVEncoder.usPixelClock = cpu_to_le16(radeon_encoder->pixel_clock / 10); + + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); + +} + +union dvo_encoder_control { + ENABLE_EXTERNAL_TMDS_ENCODER_PS_ALLOCATION ext_tmds; + DVO_ENCODER_CONTROL_PS_ALLOCATION dvo; + DVO_ENCODER_CONTROL_PS_ALLOCATION_V3 dvo_v3; +}; + +void +atombios_dvo_setup(struct drm_encoder *encoder, int action) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + union dvo_encoder_control args; + int index = GetIndexIntoMasterTable(COMMAND, DVOEncoderControl); + + memset(&args, 0, sizeof(args)); + + if (ASIC_IS_DCE3(rdev)) { + /* DCE3+ */ + args.dvo_v3.ucAction = action; + args.dvo_v3.usPixelClock = cpu_to_le16(radeon_encoder->pixel_clock / 10); + args.dvo_v3.ucDVOConfig = 0; /* XXX */ + } else if (ASIC_IS_DCE2(rdev)) { + /* DCE2 (pre-DCE3 R6xx, RS600/690/740 */ + args.dvo.sDVOEncoder.ucAction = action; + args.dvo.sDVOEncoder.usPixelClock = cpu_to_le16(radeon_encoder->pixel_clock / 10); + /* DFP1, CRT1, TV1 depending on the type of port */ + args.dvo.sDVOEncoder.ucDeviceType = ATOM_DEVICE_DFP1_INDEX; + + if (radeon_encoder->pixel_clock > 165000) + args.dvo.sDVOEncoder.usDevAttr.sDigAttrib.ucAttribute |= PANEL_ENCODER_MISC_DUAL; + } else { + /* R4xx, R5xx */ + args.ext_tmds.sXTmdsEncoder.ucEnable = action; + + if (radeon_encoder->pixel_clock > 165000) + args.ext_tmds.sXTmdsEncoder.ucMisc |= PANEL_ENCODER_MISC_DUAL; + + /*if (pScrn->rgbBits == 8)*/ + args.ext_tmds.sXTmdsEncoder.ucMisc |= ATOM_PANEL_MISC_888RGB; + } + + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); +} + +union lvds_encoder_control { + LVDS_ENCODER_CONTROL_PS_ALLOCATION v1; + LVDS_ENCODER_CONTROL_PS_ALLOCATION_V2 v2; +}; + +void +atombios_digital_setup(struct drm_encoder *encoder, int action) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct radeon_encoder_atom_dig *dig = radeon_encoder->enc_priv; + union lvds_encoder_control args; + int index = 0; + int hdmi_detected = 0; + uint8_t frev, crev; + + if (!dig) + return; + + if (atombios_get_encoder_mode(encoder) == ATOM_ENCODER_MODE_HDMI) + hdmi_detected = 1; + + memset(&args, 0, sizeof(args)); + + switch (radeon_encoder->encoder_id) { + case ENCODER_OBJECT_ID_INTERNAL_LVDS: + index = GetIndexIntoMasterTable(COMMAND, LVDSEncoderControl); + break; + case ENCODER_OBJECT_ID_INTERNAL_TMDS1: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_TMDS1: + index = GetIndexIntoMasterTable(COMMAND, TMDS1EncoderControl); + break; + case ENCODER_OBJECT_ID_INTERNAL_LVTM1: + if (radeon_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) + index = GetIndexIntoMasterTable(COMMAND, LVDSEncoderControl); + else + index = GetIndexIntoMasterTable(COMMAND, TMDS2EncoderControl); + break; + } + + if (!atom_parse_cmd_header(rdev->mode_info.atom_context, index, &frev, &crev)) + return; + + switch (frev) { + case 1: + case 2: + switch (crev) { + case 1: + args.v1.ucMisc = 0; + args.v1.ucAction = action; + if (hdmi_detected) + args.v1.ucMisc |= PANEL_ENCODER_MISC_HDMI_TYPE; + args.v1.usPixelClock = cpu_to_le16(radeon_encoder->pixel_clock / 10); + if (radeon_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) { + if (dig->lcd_misc & ATOM_PANEL_MISC_DUAL) + args.v1.ucMisc |= PANEL_ENCODER_MISC_DUAL; + if (dig->lcd_misc & ATOM_PANEL_MISC_888RGB) + args.v1.ucMisc |= ATOM_PANEL_MISC_888RGB; + } else { + if (dig->linkb) + args.v1.ucMisc |= PANEL_ENCODER_MISC_TMDS_LINKB; + if (radeon_encoder->pixel_clock > 165000) + args.v1.ucMisc |= PANEL_ENCODER_MISC_DUAL; + /*if (pScrn->rgbBits == 8) */ + args.v1.ucMisc |= ATOM_PANEL_MISC_888RGB; + } + break; + case 2: + case 3: + args.v2.ucMisc = 0; + args.v2.ucAction = action; + if (crev == 3) { + if (dig->coherent_mode) + args.v2.ucMisc |= PANEL_ENCODER_MISC_COHERENT; + } + if (hdmi_detected) + args.v2.ucMisc |= PANEL_ENCODER_MISC_HDMI_TYPE; + args.v2.usPixelClock = cpu_to_le16(radeon_encoder->pixel_clock / 10); + args.v2.ucTruncate = 0; + args.v2.ucSpatial = 0; + args.v2.ucTemporal = 0; + args.v2.ucFRC = 0; + if (radeon_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) { + if (dig->lcd_misc & ATOM_PANEL_MISC_DUAL) + args.v2.ucMisc |= PANEL_ENCODER_MISC_DUAL; + if (dig->lcd_misc & ATOM_PANEL_MISC_SPATIAL) { + args.v2.ucSpatial = PANEL_ENCODER_SPATIAL_DITHER_EN; + if (dig->lcd_misc & ATOM_PANEL_MISC_888RGB) + args.v2.ucSpatial |= PANEL_ENCODER_SPATIAL_DITHER_DEPTH; + } + if (dig->lcd_misc & ATOM_PANEL_MISC_TEMPORAL) { + args.v2.ucTemporal = PANEL_ENCODER_TEMPORAL_DITHER_EN; + if (dig->lcd_misc & ATOM_PANEL_MISC_888RGB) + args.v2.ucTemporal |= PANEL_ENCODER_TEMPORAL_DITHER_DEPTH; + if (((dig->lcd_misc >> ATOM_PANEL_MISC_GREY_LEVEL_SHIFT) & 0x3) == 2) + args.v2.ucTemporal |= PANEL_ENCODER_TEMPORAL_LEVEL_4; + } + } else { + if (dig->linkb) + args.v2.ucMisc |= PANEL_ENCODER_MISC_TMDS_LINKB; + if (radeon_encoder->pixel_clock > 165000) + args.v2.ucMisc |= PANEL_ENCODER_MISC_DUAL; + } + break; + default: + DRM_ERROR("Unknown table version %d, %d\n", frev, crev); + break; + } + break; + default: + DRM_ERROR("Unknown table version %d, %d\n", frev, crev); + break; + } + + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); +} + +int +atombios_get_encoder_mode(struct drm_encoder *encoder) +{ + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct drm_connector *connector; + struct radeon_connector *radeon_connector; + struct radeon_connector_atom_dig *dig_connector; + + /* dp bridges are always DP */ + if (radeon_encoder_is_dp_bridge(encoder)) + return ATOM_ENCODER_MODE_DP; + + /* DVO is always DVO */ + if (radeon_encoder->encoder_id == ATOM_ENCODER_MODE_DVO) + return ATOM_ENCODER_MODE_DVO; + + connector = radeon_get_connector_for_encoder(encoder); + /* if we don't have an active device yet, just use one of + * the connectors tied to the encoder. + */ + if (!connector) + connector = radeon_get_connector_for_encoder_init(encoder); + radeon_connector = to_radeon_connector(connector); + + switch (connector->connector_type) { + case DRM_MODE_CONNECTOR_DVII: + case DRM_MODE_CONNECTOR_HDMIB: /* HDMI-B is basically DL-DVI; analog works fine */ + if (drm_detect_monitor_audio(radeon_connector->edid) && radeon_audio) { + /* fix me */ + if (ASIC_IS_DCE4(rdev)) + return ATOM_ENCODER_MODE_DVI; + else + return ATOM_ENCODER_MODE_HDMI; + } else if (radeon_connector->use_digital) + return ATOM_ENCODER_MODE_DVI; + else + return ATOM_ENCODER_MODE_CRT; + break; + case DRM_MODE_CONNECTOR_DVID: + case DRM_MODE_CONNECTOR_HDMIA: + default: + if (drm_detect_monitor_audio(radeon_connector->edid) && radeon_audio) { + /* fix me */ + if (ASIC_IS_DCE4(rdev)) + return ATOM_ENCODER_MODE_DVI; + else + return ATOM_ENCODER_MODE_HDMI; + } else + return ATOM_ENCODER_MODE_DVI; + break; + case DRM_MODE_CONNECTOR_LVDS: + return ATOM_ENCODER_MODE_LVDS; + break; + case DRM_MODE_CONNECTOR_DisplayPort: + dig_connector = radeon_connector->con_priv; + if ((dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT) || + (dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_eDP)) + return ATOM_ENCODER_MODE_DP; + else if (drm_detect_monitor_audio(radeon_connector->edid) && radeon_audio) { + /* fix me */ + if (ASIC_IS_DCE4(rdev)) + return ATOM_ENCODER_MODE_DVI; + else + return ATOM_ENCODER_MODE_HDMI; + } else + return ATOM_ENCODER_MODE_DVI; + break; + case DRM_MODE_CONNECTOR_eDP: + return ATOM_ENCODER_MODE_DP; + case DRM_MODE_CONNECTOR_DVIA: + case DRM_MODE_CONNECTOR_VGA: + return ATOM_ENCODER_MODE_CRT; + break; + case DRM_MODE_CONNECTOR_Composite: + case DRM_MODE_CONNECTOR_SVIDEO: + case DRM_MODE_CONNECTOR_9PinDIN: + /* fix me */ + return ATOM_ENCODER_MODE_TV; + /*return ATOM_ENCODER_MODE_CV;*/ + break; + } +} + +/* + * DIG Encoder/Transmitter Setup + * + * DCE 3.0/3.1 + * - 2 DIG transmitter blocks. UNIPHY (links A and B) and LVTMA. + * Supports up to 3 digital outputs + * - 2 DIG encoder blocks. + * DIG1 can drive UNIPHY link A or link B + * DIG2 can drive UNIPHY link B or LVTMA + * + * DCE 3.2 + * - 3 DIG transmitter blocks. UNIPHY0/1/2 (links A and B). + * Supports up to 5 digital outputs + * - 2 DIG encoder blocks. + * DIG1/2 can drive UNIPHY0/1/2 link A or link B + * + * DCE 4.0/5.0 + * - 3 DIG transmitter blocks UNIPHY0/1/2 (links A and B). + * Supports up to 6 digital outputs + * - 6 DIG encoder blocks. + * - DIG to PHY mapping is hardcoded + * DIG1 drives UNIPHY0 link A, A+B + * DIG2 drives UNIPHY0 link B + * DIG3 drives UNIPHY1 link A, A+B + * DIG4 drives UNIPHY1 link B + * DIG5 drives UNIPHY2 link A, A+B + * DIG6 drives UNIPHY2 link B + * + * DCE 4.1 + * - 3 DIG transmitter blocks UNIPHY0/1/2 (links A and B). + * Supports up to 6 digital outputs + * - 2 DIG encoder blocks. + * DIG1/2 can drive UNIPHY0/1/2 link A or link B + * + * Routing + * crtc -> dig encoder -> UNIPHY/LVTMA (1 or 2 links) + * Examples: + * crtc0 -> dig2 -> LVTMA links A+B -> TMDS/HDMI + * crtc1 -> dig1 -> UNIPHY0 link B -> DP + * crtc0 -> dig1 -> UNIPHY2 link A -> LVDS + * crtc1 -> dig2 -> UNIPHY1 link B+A -> TMDS/HDMI + */ + +union dig_encoder_control { + DIG_ENCODER_CONTROL_PS_ALLOCATION v1; + DIG_ENCODER_CONTROL_PARAMETERS_V2 v2; + DIG_ENCODER_CONTROL_PARAMETERS_V3 v3; + DIG_ENCODER_CONTROL_PARAMETERS_V4 v4; +}; + +void +atombios_dig_encoder_setup(struct drm_encoder *encoder, int action, int panel_mode) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct radeon_encoder_atom_dig *dig = radeon_encoder->enc_priv; + struct drm_connector *connector = radeon_get_connector_for_encoder(encoder); + union dig_encoder_control args; + int index = 0; + uint8_t frev, crev; + int dp_clock = 0; + int dp_lane_count = 0; + int hpd_id = RADEON_HPD_NONE; + int bpc = 8; + + if (connector) { + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + struct radeon_connector_atom_dig *dig_connector = + radeon_connector->con_priv; + + dp_clock = dig_connector->dp_clock; + dp_lane_count = dig_connector->dp_lane_count; + hpd_id = radeon_connector->hpd.hpd; + bpc = connector->display_info.bpc; + } + + /* no dig encoder assigned */ + if (dig->dig_encoder == -1) + return; + + memset(&args, 0, sizeof(args)); + + if (ASIC_IS_DCE4(rdev)) + index = GetIndexIntoMasterTable(COMMAND, DIGxEncoderControl); + else { + if (dig->dig_encoder) + index = GetIndexIntoMasterTable(COMMAND, DIG2EncoderControl); + else + index = GetIndexIntoMasterTable(COMMAND, DIG1EncoderControl); + } + + if (!atom_parse_cmd_header(rdev->mode_info.atom_context, index, &frev, &crev)) + return; + + args.v1.ucAction = action; + args.v1.usPixelClock = cpu_to_le16(radeon_encoder->pixel_clock / 10); + if (action == ATOM_ENCODER_CMD_SETUP_PANEL_MODE) + args.v3.ucPanelMode = panel_mode; + else + args.v1.ucEncoderMode = atombios_get_encoder_mode(encoder); + + if ((args.v1.ucEncoderMode == ATOM_ENCODER_MODE_DP) || + (args.v1.ucEncoderMode == ATOM_ENCODER_MODE_DP_MST)) + args.v1.ucLaneNum = dp_lane_count; + else if (radeon_encoder->pixel_clock > 165000) + args.v1.ucLaneNum = 8; + else + args.v1.ucLaneNum = 4; + + if (ASIC_IS_DCE5(rdev)) { + if ((args.v1.ucEncoderMode == ATOM_ENCODER_MODE_DP) || + (args.v1.ucEncoderMode == ATOM_ENCODER_MODE_DP_MST)) { + if (dp_clock == 270000) + args.v1.ucConfig |= ATOM_ENCODER_CONFIG_V4_DPLINKRATE_2_70GHZ; + else if (dp_clock == 540000) + args.v1.ucConfig |= ATOM_ENCODER_CONFIG_V4_DPLINKRATE_5_40GHZ; + } + args.v4.acConfig.ucDigSel = dig->dig_encoder; + switch (bpc) { + case 0: + args.v4.ucBitPerColor = PANEL_BPC_UNDEFINE; + break; + case 6: + args.v4.ucBitPerColor = PANEL_6BIT_PER_COLOR; + break; + case 8: + default: + args.v4.ucBitPerColor = PANEL_8BIT_PER_COLOR; + break; + case 10: + args.v4.ucBitPerColor = PANEL_10BIT_PER_COLOR; + break; + case 12: + args.v4.ucBitPerColor = PANEL_12BIT_PER_COLOR; + break; + case 16: + args.v4.ucBitPerColor = PANEL_16BIT_PER_COLOR; + break; + } + if (hpd_id == RADEON_HPD_NONE) + args.v4.ucHPD_ID = 0; + else + args.v4.ucHPD_ID = hpd_id + 1; + } else if (ASIC_IS_DCE4(rdev)) { + if ((args.v1.ucEncoderMode == ATOM_ENCODER_MODE_DP) && (dp_clock == 270000)) + args.v1.ucConfig |= ATOM_ENCODER_CONFIG_V3_DPLINKRATE_2_70GHZ; + args.v3.acConfig.ucDigSel = dig->dig_encoder; + switch (bpc) { + case 0: + args.v3.ucBitPerColor = PANEL_BPC_UNDEFINE; + break; + case 6: + args.v3.ucBitPerColor = PANEL_6BIT_PER_COLOR; + break; + case 8: + default: + args.v3.ucBitPerColor = PANEL_8BIT_PER_COLOR; + break; + case 10: + args.v3.ucBitPerColor = PANEL_10BIT_PER_COLOR; + break; + case 12: + args.v3.ucBitPerColor = PANEL_12BIT_PER_COLOR; + break; + case 16: + args.v3.ucBitPerColor = PANEL_16BIT_PER_COLOR; + break; + } + } else { + if ((args.v1.ucEncoderMode == ATOM_ENCODER_MODE_DP) && (dp_clock == 270000)) + args.v1.ucConfig |= ATOM_ENCODER_CONFIG_DPLINKRATE_2_70GHZ; + switch (radeon_encoder->encoder_id) { + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY: + args.v1.ucConfig = ATOM_ENCODER_CONFIG_V2_TRANSMITTER1; + break; + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA: + args.v1.ucConfig = ATOM_ENCODER_CONFIG_V2_TRANSMITTER2; + break; + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2: + args.v1.ucConfig = ATOM_ENCODER_CONFIG_V2_TRANSMITTER3; + break; + } + if (dig->linkb) + args.v1.ucConfig |= ATOM_ENCODER_CONFIG_LINKB; + else + args.v1.ucConfig |= ATOM_ENCODER_CONFIG_LINKA; + } + + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); + +} + +union dig_transmitter_control { + DIG_TRANSMITTER_CONTROL_PS_ALLOCATION v1; + DIG_TRANSMITTER_CONTROL_PARAMETERS_V2 v2; + DIG_TRANSMITTER_CONTROL_PARAMETERS_V3 v3; + DIG_TRANSMITTER_CONTROL_PARAMETERS_V4 v4; +}; + +void +atombios_dig_transmitter_setup(struct drm_encoder *encoder, int action, uint8_t lane_num, uint8_t lane_set) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct radeon_encoder_atom_dig *dig = radeon_encoder->enc_priv; + struct drm_connector *connector; + union dig_transmitter_control args; + int index = 0; + uint8_t frev, crev; + bool is_dp = false; + int pll_id = 0; + int dp_clock = 0; + int dp_lane_count = 0; + int connector_object_id = 0; + int igp_lane_info = 0; + int dig_encoder = dig->dig_encoder; + + if (action == ATOM_TRANSMITTER_ACTION_INIT) { + connector = radeon_get_connector_for_encoder_init(encoder); + /* just needed to avoid bailing in the encoder check. the encoder + * isn't used for init + */ + dig_encoder = 0; + } else + connector = radeon_get_connector_for_encoder(encoder); + + if (connector) { + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + struct radeon_connector_atom_dig *dig_connector = + radeon_connector->con_priv; + + dp_clock = dig_connector->dp_clock; + dp_lane_count = dig_connector->dp_lane_count; + connector_object_id = + (radeon_connector->connector_object_id & OBJECT_ID_MASK) >> OBJECT_ID_SHIFT; + igp_lane_info = dig_connector->igp_lane_info; + } + + /* no dig encoder assigned */ + if (dig_encoder == -1) + return; + + if (atombios_get_encoder_mode(encoder) == ATOM_ENCODER_MODE_DP) + is_dp = true; + + memset(&args, 0, sizeof(args)); + + switch (radeon_encoder->encoder_id) { + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1: + index = GetIndexIntoMasterTable(COMMAND, DVOOutputControl); + break; + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY: + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1: + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2: + index = GetIndexIntoMasterTable(COMMAND, UNIPHYTransmitterControl); + break; + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA: + index = GetIndexIntoMasterTable(COMMAND, LVTMATransmitterControl); + break; + } + + if (!atom_parse_cmd_header(rdev->mode_info.atom_context, index, &frev, &crev)) + return; + + args.v1.ucAction = action; + if (action == ATOM_TRANSMITTER_ACTION_INIT) { + args.v1.usInitInfo = cpu_to_le16(connector_object_id); + } else if (action == ATOM_TRANSMITTER_ACTION_SETUP_VSEMPH) { + args.v1.asMode.ucLaneSel = lane_num; + args.v1.asMode.ucLaneSet = lane_set; + } else { + if (is_dp) + args.v1.usPixelClock = + cpu_to_le16(dp_clock / 10); + else if (radeon_encoder->pixel_clock > 165000) + args.v1.usPixelClock = cpu_to_le16((radeon_encoder->pixel_clock / 2) / 10); + else + args.v1.usPixelClock = cpu_to_le16(radeon_encoder->pixel_clock / 10); + } + if (ASIC_IS_DCE4(rdev)) { + if (is_dp) + args.v3.ucLaneNum = dp_lane_count; + else if (radeon_encoder->pixel_clock > 165000) + args.v3.ucLaneNum = 8; + else + args.v3.ucLaneNum = 4; + + if (dig->linkb) + args.v3.acConfig.ucLinkSel = 1; + if (dig_encoder & 1) + args.v3.acConfig.ucEncoderSel = 1; + + /* Select the PLL for the PHY + * DP PHY should be clocked from external src if there is + * one. + */ + if (encoder->crtc) { + struct radeon_crtc *radeon_crtc = to_radeon_crtc(encoder->crtc); + pll_id = radeon_crtc->pll_id; + } + + if (ASIC_IS_DCE5(rdev)) { + /* On DCE5 DCPLL usually generates the DP ref clock */ + if (is_dp) { + if (rdev->clock.dp_extclk) + args.v4.acConfig.ucRefClkSource = ENCODER_REFCLK_SRC_EXTCLK; + else + args.v4.acConfig.ucRefClkSource = ENCODER_REFCLK_SRC_DCPLL; + } else + args.v4.acConfig.ucRefClkSource = pll_id; + } else { + /* On DCE4, if there is an external clock, it generates the DP ref clock */ + if (is_dp && rdev->clock.dp_extclk) + args.v3.acConfig.ucRefClkSource = 2; /* external src */ + else + args.v3.acConfig.ucRefClkSource = pll_id; + } + + switch (radeon_encoder->encoder_id) { + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY: + args.v3.acConfig.ucTransmitterSel = 0; + break; + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1: + args.v3.acConfig.ucTransmitterSel = 1; + break; + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2: + args.v3.acConfig.ucTransmitterSel = 2; + break; + } + + if (is_dp) + args.v3.acConfig.fCoherentMode = 1; /* DP requires coherent */ + else if (radeon_encoder->devices & (ATOM_DEVICE_DFP_SUPPORT)) { + if (dig->coherent_mode) + args.v3.acConfig.fCoherentMode = 1; + if (radeon_encoder->pixel_clock > 165000) + args.v3.acConfig.fDualLinkConnector = 1; + } + } else if (ASIC_IS_DCE32(rdev)) { + args.v2.acConfig.ucEncoderSel = dig_encoder; + if (dig->linkb) + args.v2.acConfig.ucLinkSel = 1; + + switch (radeon_encoder->encoder_id) { + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY: + args.v2.acConfig.ucTransmitterSel = 0; + break; + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1: + args.v2.acConfig.ucTransmitterSel = 1; + break; + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2: + args.v2.acConfig.ucTransmitterSel = 2; + break; + } + + if (is_dp) { + args.v2.acConfig.fCoherentMode = 1; + args.v2.acConfig.fDPConnector = 1; + } else if (radeon_encoder->devices & (ATOM_DEVICE_DFP_SUPPORT)) { + if (dig->coherent_mode) + args.v2.acConfig.fCoherentMode = 1; + if (radeon_encoder->pixel_clock > 165000) + args.v2.acConfig.fDualLinkConnector = 1; + } + } else { + args.v1.ucConfig = ATOM_TRANSMITTER_CONFIG_CLKSRC_PPLL; + + if (dig_encoder) + args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_DIG2_ENCODER; + else + args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_DIG1_ENCODER; + + if ((rdev->flags & RADEON_IS_IGP) && + (radeon_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_UNIPHY)) { + if (is_dp || (radeon_encoder->pixel_clock <= 165000)) { + if (igp_lane_info & 0x1) + args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_LANE_0_3; + else if (igp_lane_info & 0x2) + args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_LANE_4_7; + else if (igp_lane_info & 0x4) + args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_LANE_8_11; + else if (igp_lane_info & 0x8) + args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_LANE_12_15; + } else { + if (igp_lane_info & 0x3) + args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_LANE_0_7; + else if (igp_lane_info & 0xc) + args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_LANE_8_15; + } + } + + if (dig->linkb) + args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_LINKB; + else + args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_LINKA; + + if (is_dp) + args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_COHERENT; + else if (radeon_encoder->devices & (ATOM_DEVICE_DFP_SUPPORT)) { + if (dig->coherent_mode) + args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_COHERENT; + if (radeon_encoder->pixel_clock > 165000) + args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_8LANE_LINK; + } + } + + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); +} + +bool +atombios_set_edp_panel_power(struct drm_connector *connector, int action) +{ + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + struct drm_device *dev = radeon_connector->base.dev; + struct radeon_device *rdev = dev->dev_private; + union dig_transmitter_control args; + int index = GetIndexIntoMasterTable(COMMAND, UNIPHYTransmitterControl); + uint8_t frev, crev; + + if (connector->connector_type != DRM_MODE_CONNECTOR_eDP) + goto done; + + if (!ASIC_IS_DCE4(rdev)) + goto done; + + if ((action != ATOM_TRANSMITTER_ACTION_POWER_ON) && + (action != ATOM_TRANSMITTER_ACTION_POWER_OFF)) + goto done; + + if (!atom_parse_cmd_header(rdev->mode_info.atom_context, index, &frev, &crev)) + goto done; + + memset(&args, 0, sizeof(args)); + + args.v1.ucAction = action; + + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); + + /* wait for the panel to power up */ + if (action == ATOM_TRANSMITTER_ACTION_POWER_ON) { + int i; + + for (i = 0; i < 300; i++) { + if (radeon_hpd_sense(rdev, radeon_connector->hpd.hpd)) + return true; + mdelay(1); + } + return false; + } +done: + return true; +} + +union external_encoder_control { + EXTERNAL_ENCODER_CONTROL_PS_ALLOCATION v1; + EXTERNAL_ENCODER_CONTROL_PS_ALLOCATION_V3 v3; +}; + +static void +atombios_external_encoder_setup(struct drm_encoder *encoder, + struct drm_encoder *ext_encoder, + int action) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct radeon_encoder *ext_radeon_encoder = to_radeon_encoder(ext_encoder); + union external_encoder_control args; + struct drm_connector *connector; + int index = GetIndexIntoMasterTable(COMMAND, ExternalEncoderControl); + u8 frev, crev; + int dp_clock = 0; + int dp_lane_count = 0; + int connector_object_id = 0; + u32 ext_enum = (ext_radeon_encoder->encoder_enum & ENUM_ID_MASK) >> ENUM_ID_SHIFT; + int bpc = 8; + + if (action == EXTERNAL_ENCODER_ACTION_V3_ENCODER_INIT) + connector = radeon_get_connector_for_encoder_init(encoder); + else + connector = radeon_get_connector_for_encoder(encoder); + + if (connector) { + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + struct radeon_connector_atom_dig *dig_connector = + radeon_connector->con_priv; + + dp_clock = dig_connector->dp_clock; + dp_lane_count = dig_connector->dp_lane_count; + connector_object_id = + (radeon_connector->connector_object_id & OBJECT_ID_MASK) >> OBJECT_ID_SHIFT; + bpc = connector->display_info.bpc; + } + + memset(&args, 0, sizeof(args)); + + if (!atom_parse_cmd_header(rdev->mode_info.atom_context, index, &frev, &crev)) + return; + + switch (frev) { + case 1: + /* no params on frev 1 */ + break; + case 2: + switch (crev) { + case 1: + case 2: + args.v1.sDigEncoder.ucAction = action; + args.v1.sDigEncoder.usPixelClock = cpu_to_le16(radeon_encoder->pixel_clock / 10); + args.v1.sDigEncoder.ucEncoderMode = atombios_get_encoder_mode(encoder); + + if (args.v1.sDigEncoder.ucEncoderMode == ATOM_ENCODER_MODE_DP) { + if (dp_clock == 270000) + args.v1.sDigEncoder.ucConfig |= ATOM_ENCODER_CONFIG_DPLINKRATE_2_70GHZ; + args.v1.sDigEncoder.ucLaneNum = dp_lane_count; + } else if (radeon_encoder->pixel_clock > 165000) + args.v1.sDigEncoder.ucLaneNum = 8; + else + args.v1.sDigEncoder.ucLaneNum = 4; + break; + case 3: + args.v3.sExtEncoder.ucAction = action; + if (action == EXTERNAL_ENCODER_ACTION_V3_ENCODER_INIT) + args.v3.sExtEncoder.usConnectorId = cpu_to_le16(connector_object_id); + else + args.v3.sExtEncoder.usPixelClock = cpu_to_le16(radeon_encoder->pixel_clock / 10); + args.v3.sExtEncoder.ucEncoderMode = atombios_get_encoder_mode(encoder); + + if (args.v3.sExtEncoder.ucEncoderMode == ATOM_ENCODER_MODE_DP) { + if (dp_clock == 270000) + args.v3.sExtEncoder.ucConfig |= EXTERNAL_ENCODER_CONFIG_V3_DPLINKRATE_2_70GHZ; + else if (dp_clock == 540000) + args.v3.sExtEncoder.ucConfig |= EXTERNAL_ENCODER_CONFIG_V3_DPLINKRATE_5_40GHZ; + args.v3.sExtEncoder.ucLaneNum = dp_lane_count; + } else if (radeon_encoder->pixel_clock > 165000) + args.v3.sExtEncoder.ucLaneNum = 8; + else + args.v3.sExtEncoder.ucLaneNum = 4; + switch (ext_enum) { + case GRAPH_OBJECT_ENUM_ID1: + args.v3.sExtEncoder.ucConfig |= EXTERNAL_ENCODER_CONFIG_V3_ENCODER1; + break; + case GRAPH_OBJECT_ENUM_ID2: + args.v3.sExtEncoder.ucConfig |= EXTERNAL_ENCODER_CONFIG_V3_ENCODER2; + break; + case GRAPH_OBJECT_ENUM_ID3: + args.v3.sExtEncoder.ucConfig |= EXTERNAL_ENCODER_CONFIG_V3_ENCODER3; + break; + } + switch (bpc) { + case 0: + args.v3.sExtEncoder.ucBitPerColor = PANEL_BPC_UNDEFINE; + break; + case 6: + args.v3.sExtEncoder.ucBitPerColor = PANEL_6BIT_PER_COLOR; + break; + case 8: + default: + args.v3.sExtEncoder.ucBitPerColor = PANEL_8BIT_PER_COLOR; + break; + case 10: + args.v3.sExtEncoder.ucBitPerColor = PANEL_10BIT_PER_COLOR; + break; + case 12: + args.v3.sExtEncoder.ucBitPerColor = PANEL_12BIT_PER_COLOR; + break; + case 16: + args.v3.sExtEncoder.ucBitPerColor = PANEL_16BIT_PER_COLOR; + break; + } + break; + default: + DRM_ERROR("Unknown table version: %d, %d\n", frev, crev); + return; + } + break; + default: + DRM_ERROR("Unknown table version: %d, %d\n", frev, crev); + return; + } + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); +} + +static void +atombios_yuv_setup(struct drm_encoder *encoder, bool enable) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct radeon_crtc *radeon_crtc = to_radeon_crtc(encoder->crtc); + ENABLE_YUV_PS_ALLOCATION args; + int index = GetIndexIntoMasterTable(COMMAND, EnableYUV); + uint32_t temp, reg; + + memset(&args, 0, sizeof(args)); + + if (rdev->family >= CHIP_R600) + reg = R600_BIOS_3_SCRATCH; + else + reg = RADEON_BIOS_3_SCRATCH; + + /* XXX: fix up scratch reg handling */ + temp = RREG32(reg); + if (radeon_encoder->active_device & (ATOM_DEVICE_TV_SUPPORT)) + WREG32(reg, (ATOM_S3_TV1_ACTIVE | + (radeon_crtc->crtc_id << 18))); + else if (radeon_encoder->active_device & (ATOM_DEVICE_CV_SUPPORT)) + WREG32(reg, (ATOM_S3_CV_ACTIVE | (radeon_crtc->crtc_id << 24))); + else + WREG32(reg, 0); + + if (enable) + args.ucEnable = ATOM_ENABLE; + args.ucCRTC = radeon_crtc->crtc_id; + + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); + + WREG32(reg, temp); +} + +static void +radeon_atom_encoder_dpms(struct drm_encoder *encoder, int mode) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct drm_encoder *ext_encoder = radeon_atom_get_external_encoder(encoder); + DISPLAY_DEVICE_OUTPUT_CONTROL_PS_ALLOCATION args; + int index = 0; + bool is_dig = false; + bool is_dce5_dac = false; + bool is_dce5_dvo = false; + + memset(&args, 0, sizeof(args)); + + DRM_DEBUG_KMS("encoder dpms %d to mode %d, devices %08x, active_devices %08x\n", + radeon_encoder->encoder_id, mode, radeon_encoder->devices, + radeon_encoder->active_device); + switch (radeon_encoder->encoder_id) { + case ENCODER_OBJECT_ID_INTERNAL_TMDS1: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_TMDS1: + index = GetIndexIntoMasterTable(COMMAND, TMDSAOutputControl); + break; + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY: + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1: + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA: + is_dig = true; + break; + case ENCODER_OBJECT_ID_INTERNAL_DVO1: + case ENCODER_OBJECT_ID_INTERNAL_DDI: + index = GetIndexIntoMasterTable(COMMAND, DVOOutputControl); + break; + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1: + if (ASIC_IS_DCE5(rdev)) + is_dce5_dvo = true; + else if (ASIC_IS_DCE3(rdev)) + is_dig = true; + else + index = GetIndexIntoMasterTable(COMMAND, DVOOutputControl); + break; + case ENCODER_OBJECT_ID_INTERNAL_LVDS: + index = GetIndexIntoMasterTable(COMMAND, LCD1OutputControl); + break; + case ENCODER_OBJECT_ID_INTERNAL_LVTM1: + if (radeon_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) + index = GetIndexIntoMasterTable(COMMAND, LCD1OutputControl); + else + index = GetIndexIntoMasterTable(COMMAND, LVTMAOutputControl); + break; + case ENCODER_OBJECT_ID_INTERNAL_DAC1: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1: + if (ASIC_IS_DCE5(rdev)) + is_dce5_dac = true; + else { + if (radeon_encoder->active_device & (ATOM_DEVICE_TV_SUPPORT)) + index = GetIndexIntoMasterTable(COMMAND, TV1OutputControl); + else if (radeon_encoder->active_device & (ATOM_DEVICE_CV_SUPPORT)) + index = GetIndexIntoMasterTable(COMMAND, CV1OutputControl); + else + index = GetIndexIntoMasterTable(COMMAND, DAC1OutputControl); + } + break; + case ENCODER_OBJECT_ID_INTERNAL_DAC2: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2: + if (radeon_encoder->active_device & (ATOM_DEVICE_TV_SUPPORT)) + index = GetIndexIntoMasterTable(COMMAND, TV1OutputControl); + else if (radeon_encoder->active_device & (ATOM_DEVICE_CV_SUPPORT)) + index = GetIndexIntoMasterTable(COMMAND, CV1OutputControl); + else + index = GetIndexIntoMasterTable(COMMAND, DAC2OutputControl); + break; + } + + if (is_dig) { + switch (mode) { + case DRM_MODE_DPMS_ON: + /* some early dce3.2 boards have a bug in their transmitter control table */ + if ((rdev->family == CHIP_RV710) || (rdev->family == CHIP_RV730)) + atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_ENABLE, 0, 0); + else + atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_ENABLE_OUTPUT, 0, 0); + if (atombios_get_encoder_mode(encoder) == ATOM_ENCODER_MODE_DP) { + struct drm_connector *connector = radeon_get_connector_for_encoder(encoder); + + if (connector && + (connector->connector_type == DRM_MODE_CONNECTOR_eDP)) { + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + struct radeon_connector_atom_dig *radeon_dig_connector = + radeon_connector->con_priv; + atombios_set_edp_panel_power(connector, + ATOM_TRANSMITTER_ACTION_POWER_ON); + radeon_dig_connector->edp_on = true; + } + if (ASIC_IS_DCE4(rdev)) + atombios_dig_encoder_setup(encoder, ATOM_ENCODER_CMD_DP_VIDEO_OFF, 0); + radeon_dp_link_train(encoder, connector); + if (ASIC_IS_DCE4(rdev)) + atombios_dig_encoder_setup(encoder, ATOM_ENCODER_CMD_DP_VIDEO_ON, 0); + } + if (radeon_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) + atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_LCD_BLON, 0, 0); + break; + case DRM_MODE_DPMS_STANDBY: + case DRM_MODE_DPMS_SUSPEND: + case DRM_MODE_DPMS_OFF: + atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_DISABLE_OUTPUT, 0, 0); + if (atombios_get_encoder_mode(encoder) == ATOM_ENCODER_MODE_DP) { + struct drm_connector *connector = radeon_get_connector_for_encoder(encoder); + + if (ASIC_IS_DCE4(rdev)) + atombios_dig_encoder_setup(encoder, ATOM_ENCODER_CMD_DP_VIDEO_OFF, 0); + if (connector && + (connector->connector_type == DRM_MODE_CONNECTOR_eDP)) { + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + struct radeon_connector_atom_dig *radeon_dig_connector = + radeon_connector->con_priv; + atombios_set_edp_panel_power(connector, + ATOM_TRANSMITTER_ACTION_POWER_OFF); + radeon_dig_connector->edp_on = false; + } + } + if (radeon_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) + atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_LCD_BLOFF, 0, 0); + break; + } + } else if (is_dce5_dac) { + switch (mode) { + case DRM_MODE_DPMS_ON: + atombios_dac_setup(encoder, ATOM_ENABLE); + break; + case DRM_MODE_DPMS_STANDBY: + case DRM_MODE_DPMS_SUSPEND: + case DRM_MODE_DPMS_OFF: + atombios_dac_setup(encoder, ATOM_DISABLE); + break; + } + } else if (is_dce5_dvo) { + switch (mode) { + case DRM_MODE_DPMS_ON: + atombios_dvo_setup(encoder, ATOM_ENABLE); + break; + case DRM_MODE_DPMS_STANDBY: + case DRM_MODE_DPMS_SUSPEND: + case DRM_MODE_DPMS_OFF: + atombios_dvo_setup(encoder, ATOM_DISABLE); + break; + } + } else { + switch (mode) { + case DRM_MODE_DPMS_ON: + args.ucAction = ATOM_ENABLE; + /* workaround for DVOOutputControl on some RS690 systems */ + if (radeon_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_DDI) { + u32 reg = RREG32(RADEON_BIOS_3_SCRATCH); + WREG32(RADEON_BIOS_3_SCRATCH, reg & ~ATOM_S3_DFP2I_ACTIVE); + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); + WREG32(RADEON_BIOS_3_SCRATCH, reg); + } else + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); + if (radeon_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) { + args.ucAction = ATOM_LCD_BLON; + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); + } + break; + case DRM_MODE_DPMS_STANDBY: + case DRM_MODE_DPMS_SUSPEND: + case DRM_MODE_DPMS_OFF: + args.ucAction = ATOM_DISABLE; + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); + if (radeon_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) { + args.ucAction = ATOM_LCD_BLOFF; + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); + } + break; + } + } + + if (ext_encoder) { + switch (mode) { + case DRM_MODE_DPMS_ON: + default: + if (ASIC_IS_DCE41(rdev)) { + atombios_external_encoder_setup(encoder, ext_encoder, + EXTERNAL_ENCODER_ACTION_V3_ENABLE_OUTPUT); + atombios_external_encoder_setup(encoder, ext_encoder, + EXTERNAL_ENCODER_ACTION_V3_ENCODER_BLANKING_OFF); + } else + atombios_external_encoder_setup(encoder, ext_encoder, ATOM_ENABLE); + break; + case DRM_MODE_DPMS_STANDBY: + case DRM_MODE_DPMS_SUSPEND: + case DRM_MODE_DPMS_OFF: + if (ASIC_IS_DCE41(rdev)) { + atombios_external_encoder_setup(encoder, ext_encoder, + EXTERNAL_ENCODER_ACTION_V3_ENCODER_BLANKING); + atombios_external_encoder_setup(encoder, ext_encoder, + EXTERNAL_ENCODER_ACTION_V3_DISABLE_OUTPUT); + } else + atombios_external_encoder_setup(encoder, ext_encoder, ATOM_DISABLE); + break; + } + } + + radeon_atombios_encoder_dpms_scratch_regs(encoder, (mode == DRM_MODE_DPMS_ON) ? true : false); + +} + +union crtc_source_param { + SELECT_CRTC_SOURCE_PS_ALLOCATION v1; + SELECT_CRTC_SOURCE_PARAMETERS_V2 v2; +}; + +static void +atombios_set_encoder_crtc_source(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct radeon_crtc *radeon_crtc = to_radeon_crtc(encoder->crtc); + union crtc_source_param args; + int index = GetIndexIntoMasterTable(COMMAND, SelectCRTC_Source); + uint8_t frev, crev; + struct radeon_encoder_atom_dig *dig; + + memset(&args, 0, sizeof(args)); + + if (!atom_parse_cmd_header(rdev->mode_info.atom_context, index, &frev, &crev)) + return; + + switch (frev) { + case 1: + switch (crev) { + case 1: + default: + if (ASIC_IS_AVIVO(rdev)) + args.v1.ucCRTC = radeon_crtc->crtc_id; + else { + if (radeon_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_DAC1) { + args.v1.ucCRTC = radeon_crtc->crtc_id; + } else { + args.v1.ucCRTC = radeon_crtc->crtc_id << 2; + } + } + switch (radeon_encoder->encoder_id) { + case ENCODER_OBJECT_ID_INTERNAL_TMDS1: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_TMDS1: + args.v1.ucDevice = ATOM_DEVICE_DFP1_INDEX; + break; + case ENCODER_OBJECT_ID_INTERNAL_LVDS: + case ENCODER_OBJECT_ID_INTERNAL_LVTM1: + if (radeon_encoder->devices & ATOM_DEVICE_LCD1_SUPPORT) + args.v1.ucDevice = ATOM_DEVICE_LCD1_INDEX; + else + args.v1.ucDevice = ATOM_DEVICE_DFP3_INDEX; + break; + case ENCODER_OBJECT_ID_INTERNAL_DVO1: + case ENCODER_OBJECT_ID_INTERNAL_DDI: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1: + args.v1.ucDevice = ATOM_DEVICE_DFP2_INDEX; + break; + case ENCODER_OBJECT_ID_INTERNAL_DAC1: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1: + if (radeon_encoder->active_device & (ATOM_DEVICE_TV_SUPPORT)) + args.v1.ucDevice = ATOM_DEVICE_TV1_INDEX; + else if (radeon_encoder->active_device & (ATOM_DEVICE_CV_SUPPORT)) + args.v1.ucDevice = ATOM_DEVICE_CV_INDEX; + else + args.v1.ucDevice = ATOM_DEVICE_CRT1_INDEX; + break; + case ENCODER_OBJECT_ID_INTERNAL_DAC2: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2: + if (radeon_encoder->active_device & (ATOM_DEVICE_TV_SUPPORT)) + args.v1.ucDevice = ATOM_DEVICE_TV1_INDEX; + else if (radeon_encoder->active_device & (ATOM_DEVICE_CV_SUPPORT)) + args.v1.ucDevice = ATOM_DEVICE_CV_INDEX; + else + args.v1.ucDevice = ATOM_DEVICE_CRT2_INDEX; + break; + } + break; + case 2: + args.v2.ucCRTC = radeon_crtc->crtc_id; + args.v2.ucEncodeMode = atombios_get_encoder_mode(encoder); + switch (radeon_encoder->encoder_id) { + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY: + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1: + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA: + dig = radeon_encoder->enc_priv; + switch (dig->dig_encoder) { + case 0: + args.v2.ucEncoderID = ASIC_INT_DIG1_ENCODER_ID; + break; + case 1: + args.v2.ucEncoderID = ASIC_INT_DIG2_ENCODER_ID; + break; + case 2: + args.v2.ucEncoderID = ASIC_INT_DIG3_ENCODER_ID; + break; + case 3: + args.v2.ucEncoderID = ASIC_INT_DIG4_ENCODER_ID; + break; + case 4: + args.v2.ucEncoderID = ASIC_INT_DIG5_ENCODER_ID; + break; + case 5: + args.v2.ucEncoderID = ASIC_INT_DIG6_ENCODER_ID; + break; + } + break; + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1: + args.v2.ucEncoderID = ASIC_INT_DVO_ENCODER_ID; + break; + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1: + if (radeon_encoder->active_device & (ATOM_DEVICE_TV_SUPPORT)) + args.v2.ucEncoderID = ASIC_INT_TV_ENCODER_ID; + else if (radeon_encoder->active_device & (ATOM_DEVICE_CV_SUPPORT)) + args.v2.ucEncoderID = ASIC_INT_TV_ENCODER_ID; + else + args.v2.ucEncoderID = ASIC_INT_DAC1_ENCODER_ID; + break; + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2: + if (radeon_encoder->active_device & (ATOM_DEVICE_TV_SUPPORT)) + args.v2.ucEncoderID = ASIC_INT_TV_ENCODER_ID; + else if (radeon_encoder->active_device & (ATOM_DEVICE_CV_SUPPORT)) + args.v2.ucEncoderID = ASIC_INT_TV_ENCODER_ID; + else + args.v2.ucEncoderID = ASIC_INT_DAC2_ENCODER_ID; + break; + } + break; + } + break; + default: + DRM_ERROR("Unknown table version: %d, %d\n", frev, crev); + return; + } + + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); + + /* update scratch regs with new routing */ + radeon_atombios_encoder_crtc_scratch_regs(encoder, radeon_crtc->crtc_id); +} + +static void +atombios_apply_encoder_quirks(struct drm_encoder *encoder, + struct drm_display_mode *mode) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct radeon_crtc *radeon_crtc = to_radeon_crtc(encoder->crtc); + + /* Funky macbooks */ + if ((dev->pdev->device == 0x71C5) && + (dev->pdev->subsystem_vendor == 0x106b) && + (dev->pdev->subsystem_device == 0x0080)) { + if (radeon_encoder->devices & ATOM_DEVICE_LCD1_SUPPORT) { + uint32_t lvtma_bit_depth_control = RREG32(AVIVO_LVTMA_BIT_DEPTH_CONTROL); + + lvtma_bit_depth_control &= ~AVIVO_LVTMA_BIT_DEPTH_CONTROL_TRUNCATE_EN; + lvtma_bit_depth_control &= ~AVIVO_LVTMA_BIT_DEPTH_CONTROL_SPATIAL_DITHER_EN; + + WREG32(AVIVO_LVTMA_BIT_DEPTH_CONTROL, lvtma_bit_depth_control); + } + } + + /* set scaler clears this on some chips */ + if (ASIC_IS_AVIVO(rdev) && + (!(radeon_encoder->active_device & (ATOM_DEVICE_TV_SUPPORT)))) { + if (ASIC_IS_DCE4(rdev)) { + if (mode->flags & DRM_MODE_FLAG_INTERLACE) + WREG32(EVERGREEN_DATA_FORMAT + radeon_crtc->crtc_offset, + EVERGREEN_INTERLEAVE_EN); + else + WREG32(EVERGREEN_DATA_FORMAT + radeon_crtc->crtc_offset, 0); + } else { + if (mode->flags & DRM_MODE_FLAG_INTERLACE) + WREG32(AVIVO_D1MODE_DATA_FORMAT + radeon_crtc->crtc_offset, + AVIVO_D1MODE_INTERLEAVE_EN); + else + WREG32(AVIVO_D1MODE_DATA_FORMAT + radeon_crtc->crtc_offset, 0); + } + } +} + +static int radeon_atom_pick_dig_encoder(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_crtc *radeon_crtc = to_radeon_crtc(encoder->crtc); + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct drm_encoder *test_encoder; + struct radeon_encoder_atom_dig *dig; + uint32_t dig_enc_in_use = 0; + + /* DCE4/5 */ + if (ASIC_IS_DCE4(rdev)) { + dig = radeon_encoder->enc_priv; + if (ASIC_IS_DCE41(rdev)) { + if (dig->linkb) + return 1; + else + return 0; + } else { + switch (radeon_encoder->encoder_id) { + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY: + if (dig->linkb) + return 1; + else + return 0; + break; + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1: + if (dig->linkb) + return 3; + else + return 2; + break; + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2: + if (dig->linkb) + return 5; + else + return 4; + break; + } + } + } + + /* on DCE32 and encoder can driver any block so just crtc id */ + if (ASIC_IS_DCE32(rdev)) { + return radeon_crtc->crtc_id; + } + + /* on DCE3 - LVTMA can only be driven by DIGB */ + list_for_each_entry(test_encoder, &dev->mode_config.encoder_list, head) { + struct radeon_encoder *radeon_test_encoder; + + if (encoder == test_encoder) + continue; + + if (!radeon_encoder_is_digital(test_encoder)) + continue; + + radeon_test_encoder = to_radeon_encoder(test_encoder); + dig = radeon_test_encoder->enc_priv; + + if (dig->dig_encoder >= 0) + dig_enc_in_use |= (1 << dig->dig_encoder); + } + + if (radeon_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA) { + if (dig_enc_in_use & 0x2) + DRM_ERROR("LVDS required digital encoder 2 but it was in use - stealing\n"); + return 1; + } + if (!(dig_enc_in_use & 1)) + return 0; + return 1; +} + +/* This only needs to be called once at startup */ +void +radeon_atom_encoder_init(struct radeon_device *rdev) +{ + struct drm_device *dev = rdev->ddev; + struct drm_encoder *encoder; + + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct drm_encoder *ext_encoder = radeon_atom_get_external_encoder(encoder); + + switch (radeon_encoder->encoder_id) { + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY: + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1: + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA: + atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_INIT, 0, 0); + break; + default: + break; + } + + if (ext_encoder && ASIC_IS_DCE41(rdev)) + atombios_external_encoder_setup(encoder, ext_encoder, + EXTERNAL_ENCODER_ACTION_V3_ENCODER_INIT); + } +} + +static void +radeon_atom_encoder_mode_set(struct drm_encoder *encoder, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct drm_encoder *ext_encoder = radeon_atom_get_external_encoder(encoder); + + radeon_encoder->pixel_clock = adjusted_mode->clock; + + if (ASIC_IS_AVIVO(rdev) && !ASIC_IS_DCE4(rdev)) { + if (radeon_encoder->active_device & (ATOM_DEVICE_CV_SUPPORT | ATOM_DEVICE_TV_SUPPORT)) + atombios_yuv_setup(encoder, true); + else + atombios_yuv_setup(encoder, false); + } + + switch (radeon_encoder->encoder_id) { + case ENCODER_OBJECT_ID_INTERNAL_TMDS1: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_TMDS1: + case ENCODER_OBJECT_ID_INTERNAL_LVDS: + case ENCODER_OBJECT_ID_INTERNAL_LVTM1: + atombios_digital_setup(encoder, PANEL_ENCODER_ACTION_ENABLE); + break; + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY: + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1: + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA: + if (ASIC_IS_DCE4(rdev)) { + /* disable the transmitter */ + atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_DISABLE, 0, 0); + /* setup and enable the encoder */ + atombios_dig_encoder_setup(encoder, ATOM_ENCODER_CMD_SETUP, 0); + + /* enable the transmitter */ + atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_ENABLE, 0, 0); + } else { + /* disable the encoder and transmitter */ + atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_DISABLE, 0, 0); + atombios_dig_encoder_setup(encoder, ATOM_DISABLE, 0); + + /* setup and enable the encoder and transmitter */ + atombios_dig_encoder_setup(encoder, ATOM_ENABLE, 0); + atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_SETUP, 0, 0); + atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_ENABLE, 0, 0); + } + break; + case ENCODER_OBJECT_ID_INTERNAL_DDI: + case ENCODER_OBJECT_ID_INTERNAL_DVO1: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1: + atombios_dvo_setup(encoder, ATOM_ENABLE); + break; + case ENCODER_OBJECT_ID_INTERNAL_DAC1: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1: + case ENCODER_OBJECT_ID_INTERNAL_DAC2: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2: + atombios_dac_setup(encoder, ATOM_ENABLE); + if (radeon_encoder->devices & (ATOM_DEVICE_TV_SUPPORT | ATOM_DEVICE_CV_SUPPORT)) { + if (radeon_encoder->active_device & (ATOM_DEVICE_TV_SUPPORT | ATOM_DEVICE_CV_SUPPORT)) + atombios_tv_setup(encoder, ATOM_ENABLE); + else + atombios_tv_setup(encoder, ATOM_DISABLE); + } + break; + } + + if (ext_encoder) { + if (ASIC_IS_DCE41(rdev)) + atombios_external_encoder_setup(encoder, ext_encoder, + EXTERNAL_ENCODER_ACTION_V3_ENCODER_SETUP); + else + atombios_external_encoder_setup(encoder, ext_encoder, ATOM_ENABLE); + } + + atombios_apply_encoder_quirks(encoder, adjusted_mode); + + if (atombios_get_encoder_mode(encoder) == ATOM_ENCODER_MODE_HDMI) { + r600_hdmi_enable(encoder); + r600_hdmi_setmode(encoder, adjusted_mode); + } +} + +static bool +atombios_dac_load_detect(struct drm_encoder *encoder, struct drm_connector *connector) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + + if (radeon_encoder->devices & (ATOM_DEVICE_TV_SUPPORT | + ATOM_DEVICE_CV_SUPPORT | + ATOM_DEVICE_CRT_SUPPORT)) { + DAC_LOAD_DETECTION_PS_ALLOCATION args; + int index = GetIndexIntoMasterTable(COMMAND, DAC_LoadDetection); + uint8_t frev, crev; + + memset(&args, 0, sizeof(args)); + + if (!atom_parse_cmd_header(rdev->mode_info.atom_context, index, &frev, &crev)) + return false; + + args.sDacload.ucMisc = 0; + + if ((radeon_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_DAC1) || + (radeon_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1)) + args.sDacload.ucDacType = ATOM_DAC_A; + else + args.sDacload.ucDacType = ATOM_DAC_B; + + if (radeon_connector->devices & ATOM_DEVICE_CRT1_SUPPORT) + args.sDacload.usDeviceID = cpu_to_le16(ATOM_DEVICE_CRT1_SUPPORT); + else if (radeon_connector->devices & ATOM_DEVICE_CRT2_SUPPORT) + args.sDacload.usDeviceID = cpu_to_le16(ATOM_DEVICE_CRT2_SUPPORT); + else if (radeon_connector->devices & ATOM_DEVICE_CV_SUPPORT) { + args.sDacload.usDeviceID = cpu_to_le16(ATOM_DEVICE_CV_SUPPORT); + if (crev >= 3) + args.sDacload.ucMisc = DAC_LOAD_MISC_YPrPb; + } else if (radeon_connector->devices & ATOM_DEVICE_TV1_SUPPORT) { + args.sDacload.usDeviceID = cpu_to_le16(ATOM_DEVICE_TV1_SUPPORT); + if (crev >= 3) + args.sDacload.ucMisc = DAC_LOAD_MISC_YPrPb; + } + + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); + + return true; + } else + return false; +} + +static enum drm_connector_status +radeon_atom_dac_detect(struct drm_encoder *encoder, struct drm_connector *connector) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + uint32_t bios_0_scratch; + + if (!atombios_dac_load_detect(encoder, connector)) { + DRM_DEBUG_KMS("detect returned false \n"); + return connector_status_unknown; + } + + if (rdev->family >= CHIP_R600) + bios_0_scratch = RREG32(R600_BIOS_0_SCRATCH); + else + bios_0_scratch = RREG32(RADEON_BIOS_0_SCRATCH); + + DRM_DEBUG_KMS("Bios 0 scratch %x %08x\n", bios_0_scratch, radeon_encoder->devices); + if (radeon_connector->devices & ATOM_DEVICE_CRT1_SUPPORT) { + if (bios_0_scratch & ATOM_S0_CRT1_MASK) + return connector_status_connected; + } + if (radeon_connector->devices & ATOM_DEVICE_CRT2_SUPPORT) { + if (bios_0_scratch & ATOM_S0_CRT2_MASK) + return connector_status_connected; + } + if (radeon_connector->devices & ATOM_DEVICE_CV_SUPPORT) { + if (bios_0_scratch & (ATOM_S0_CV_MASK|ATOM_S0_CV_MASK_A)) + return connector_status_connected; + } + if (radeon_connector->devices & ATOM_DEVICE_TV1_SUPPORT) { + if (bios_0_scratch & (ATOM_S0_TV1_COMPOSITE | ATOM_S0_TV1_COMPOSITE_A)) + return connector_status_connected; /* CTV */ + else if (bios_0_scratch & (ATOM_S0_TV1_SVIDEO | ATOM_S0_TV1_SVIDEO_A)) + return connector_status_connected; /* STV */ + } + return connector_status_disconnected; +} + +static enum drm_connector_status +radeon_atom_dig_detect(struct drm_encoder *encoder, struct drm_connector *connector) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + struct drm_encoder *ext_encoder = radeon_atom_get_external_encoder(encoder); + u32 bios_0_scratch; + + if (!ASIC_IS_DCE4(rdev)) + return connector_status_unknown; + + if (!ext_encoder) + return connector_status_unknown; + + if ((radeon_connector->devices & ATOM_DEVICE_CRT_SUPPORT) == 0) + return connector_status_unknown; + + /* load detect on the dp bridge */ + atombios_external_encoder_setup(encoder, ext_encoder, + EXTERNAL_ENCODER_ACTION_V3_DACLOAD_DETECTION); + + bios_0_scratch = RREG32(R600_BIOS_0_SCRATCH); + + DRM_DEBUG_KMS("Bios 0 scratch %x %08x\n", bios_0_scratch, radeon_encoder->devices); + if (radeon_connector->devices & ATOM_DEVICE_CRT1_SUPPORT) { + if (bios_0_scratch & ATOM_S0_CRT1_MASK) + return connector_status_connected; + } + if (radeon_connector->devices & ATOM_DEVICE_CRT2_SUPPORT) { + if (bios_0_scratch & ATOM_S0_CRT2_MASK) + return connector_status_connected; + } + if (radeon_connector->devices & ATOM_DEVICE_CV_SUPPORT) { + if (bios_0_scratch & (ATOM_S0_CV_MASK|ATOM_S0_CV_MASK_A)) + return connector_status_connected; + } + if (radeon_connector->devices & ATOM_DEVICE_TV1_SUPPORT) { + if (bios_0_scratch & (ATOM_S0_TV1_COMPOSITE | ATOM_S0_TV1_COMPOSITE_A)) + return connector_status_connected; /* CTV */ + else if (bios_0_scratch & (ATOM_S0_TV1_SVIDEO | ATOM_S0_TV1_SVIDEO_A)) + return connector_status_connected; /* STV */ + } + return connector_status_disconnected; +} + +void +radeon_atom_ext_encoder_setup_ddc(struct drm_encoder *encoder) +{ + struct drm_encoder *ext_encoder = radeon_atom_get_external_encoder(encoder); + + if (ext_encoder) + /* ddc_setup on the dp bridge */ + atombios_external_encoder_setup(encoder, ext_encoder, + EXTERNAL_ENCODER_ACTION_V3_DDC_SETUP); + +} + +static void radeon_atom_encoder_prepare(struct drm_encoder *encoder) +{ + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct drm_connector *connector = radeon_get_connector_for_encoder(encoder); + + if ((radeon_encoder->active_device & + (ATOM_DEVICE_DFP_SUPPORT | ATOM_DEVICE_LCD_SUPPORT)) || + radeon_encoder_is_dp_bridge(encoder)) { + struct radeon_encoder_atom_dig *dig = radeon_encoder->enc_priv; + if (dig) + dig->dig_encoder = radeon_atom_pick_dig_encoder(encoder); + } + + radeon_atom_output_lock(encoder, true); + radeon_atom_encoder_dpms(encoder, DRM_MODE_DPMS_OFF); + + if (connector) { + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + + /* select the clock/data port if it uses a router */ + if (radeon_connector->router.cd_valid) + radeon_router_select_cd_port(radeon_connector); + + /* turn eDP panel on for mode set */ + if (connector->connector_type == DRM_MODE_CONNECTOR_eDP) + atombios_set_edp_panel_power(connector, + ATOM_TRANSMITTER_ACTION_POWER_ON); + } + + /* this is needed for the pll/ss setup to work correctly in some cases */ + atombios_set_encoder_crtc_source(encoder); +} + +static void radeon_atom_encoder_commit(struct drm_encoder *encoder) +{ + radeon_atom_encoder_dpms(encoder, DRM_MODE_DPMS_ON); + radeon_atom_output_lock(encoder, false); +} + +static void radeon_atom_encoder_disable(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct radeon_encoder_atom_dig *dig; + + /* check for pre-DCE3 cards with shared encoders; + * can't really use the links individually, so don't disable + * the encoder if it's in use by another connector + */ + if (!ASIC_IS_DCE3(rdev)) { + struct drm_encoder *other_encoder; + struct radeon_encoder *other_radeon_encoder; + + list_for_each_entry(other_encoder, &dev->mode_config.encoder_list, head) { + other_radeon_encoder = to_radeon_encoder(other_encoder); + if ((radeon_encoder->encoder_id == other_radeon_encoder->encoder_id) && + drm_helper_encoder_in_use(other_encoder)) + goto disable_done; + } + } + + radeon_atom_encoder_dpms(encoder, DRM_MODE_DPMS_OFF); + + switch (radeon_encoder->encoder_id) { + case ENCODER_OBJECT_ID_INTERNAL_TMDS1: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_TMDS1: + case ENCODER_OBJECT_ID_INTERNAL_LVDS: + case ENCODER_OBJECT_ID_INTERNAL_LVTM1: + atombios_digital_setup(encoder, PANEL_ENCODER_ACTION_DISABLE); + break; + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY: + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1: + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA: + if (ASIC_IS_DCE4(rdev)) + /* disable the transmitter */ + atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_DISABLE, 0, 0); + else { + /* disable the encoder and transmitter */ + atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_DISABLE, 0, 0); + atombios_dig_encoder_setup(encoder, ATOM_DISABLE, 0); + } + break; + case ENCODER_OBJECT_ID_INTERNAL_DDI: + case ENCODER_OBJECT_ID_INTERNAL_DVO1: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1: + atombios_dvo_setup(encoder, ATOM_DISABLE); + break; + case ENCODER_OBJECT_ID_INTERNAL_DAC1: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1: + case ENCODER_OBJECT_ID_INTERNAL_DAC2: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2: + atombios_dac_setup(encoder, ATOM_DISABLE); + if (radeon_encoder->devices & (ATOM_DEVICE_TV_SUPPORT | ATOM_DEVICE_CV_SUPPORT)) + atombios_tv_setup(encoder, ATOM_DISABLE); + break; + } + +disable_done: + if (radeon_encoder_is_digital(encoder)) { + if (atombios_get_encoder_mode(encoder) == ATOM_ENCODER_MODE_HDMI) + r600_hdmi_disable(encoder); + dig = radeon_encoder->enc_priv; + dig->dig_encoder = -1; + } + radeon_encoder->active_device = 0; +} + +/* these are handled by the primary encoders */ +static void radeon_atom_ext_prepare(struct drm_encoder *encoder) +{ + +} + +static void radeon_atom_ext_commit(struct drm_encoder *encoder) +{ + +} + +static void +radeon_atom_ext_mode_set(struct drm_encoder *encoder, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + +} + +static void radeon_atom_ext_disable(struct drm_encoder *encoder) +{ + +} + +static void +radeon_atom_ext_dpms(struct drm_encoder *encoder, int mode) +{ + +} + +static bool radeon_atom_ext_mode_fixup(struct drm_encoder *encoder, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + return true; +} + +static const struct drm_encoder_helper_funcs radeon_atom_ext_helper_funcs = { + .dpms = radeon_atom_ext_dpms, + .mode_fixup = radeon_atom_ext_mode_fixup, + .prepare = radeon_atom_ext_prepare, + .mode_set = radeon_atom_ext_mode_set, + .commit = radeon_atom_ext_commit, + .disable = radeon_atom_ext_disable, + /* no detect for TMDS/LVDS yet */ +}; + +static const struct drm_encoder_helper_funcs radeon_atom_dig_helper_funcs = { + .dpms = radeon_atom_encoder_dpms, + .mode_fixup = radeon_atom_mode_fixup, + .prepare = radeon_atom_encoder_prepare, + .mode_set = radeon_atom_encoder_mode_set, + .commit = radeon_atom_encoder_commit, + .disable = radeon_atom_encoder_disable, + .detect = radeon_atom_dig_detect, +}; + +static const struct drm_encoder_helper_funcs radeon_atom_dac_helper_funcs = { + .dpms = radeon_atom_encoder_dpms, + .mode_fixup = radeon_atom_mode_fixup, + .prepare = radeon_atom_encoder_prepare, + .mode_set = radeon_atom_encoder_mode_set, + .commit = radeon_atom_encoder_commit, + .detect = radeon_atom_dac_detect, +}; + +void radeon_enc_destroy(struct drm_encoder *encoder) +{ + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + kfree(radeon_encoder->enc_priv); + drm_encoder_cleanup(encoder); + kfree(radeon_encoder); +} + +static const struct drm_encoder_funcs radeon_atom_enc_funcs = { + .destroy = radeon_enc_destroy, +}; + +struct radeon_encoder_atom_dac * +radeon_atombios_set_dac_info(struct radeon_encoder *radeon_encoder) +{ + struct drm_device *dev = radeon_encoder->base.dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder_atom_dac *dac = kzalloc(sizeof(struct radeon_encoder_atom_dac), GFP_KERNEL); + + if (!dac) + return NULL; + + dac->tv_std = radeon_atombios_get_tv_info(rdev); + return dac; +} + +struct radeon_encoder_atom_dig * +radeon_atombios_set_dig_info(struct radeon_encoder *radeon_encoder) +{ + int encoder_enum = (radeon_encoder->encoder_enum & ENUM_ID_MASK) >> ENUM_ID_SHIFT; + struct radeon_encoder_atom_dig *dig = kzalloc(sizeof(struct radeon_encoder_atom_dig), GFP_KERNEL); + + if (!dig) + return NULL; + + /* coherent mode by default */ + dig->coherent_mode = true; + dig->dig_encoder = -1; + + if (encoder_enum == 2) + dig->linkb = true; + else + dig->linkb = false; + + return dig; +} + +void +radeon_add_atom_encoder(struct drm_device *dev, + uint32_t encoder_enum, + uint32_t supported_device, + u16 caps) +{ + struct radeon_device *rdev = dev->dev_private; + struct drm_encoder *encoder; + struct radeon_encoder *radeon_encoder; + + /* see if we already added it */ + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + radeon_encoder = to_radeon_encoder(encoder); + if (radeon_encoder->encoder_enum == encoder_enum) { + radeon_encoder->devices |= supported_device; + return; + } + + } + + /* add a new one */ + radeon_encoder = kzalloc(sizeof(struct radeon_encoder), GFP_KERNEL); + if (!radeon_encoder) + return; + + encoder = &radeon_encoder->base; + switch (rdev->num_crtc) { + case 1: + encoder->possible_crtcs = 0x1; + break; + case 2: + default: + encoder->possible_crtcs = 0x3; + break; + case 4: + encoder->possible_crtcs = 0xf; + break; + case 6: + encoder->possible_crtcs = 0x3f; + break; + } + + radeon_encoder->enc_priv = NULL; + + radeon_encoder->encoder_enum = encoder_enum; + radeon_encoder->encoder_id = (encoder_enum & OBJECT_ID_MASK) >> OBJECT_ID_SHIFT; + radeon_encoder->devices = supported_device; + radeon_encoder->rmx_type = RMX_OFF; + radeon_encoder->underscan_type = UNDERSCAN_OFF; + radeon_encoder->is_ext_encoder = false; + radeon_encoder->caps = caps; + + switch (radeon_encoder->encoder_id) { + case ENCODER_OBJECT_ID_INTERNAL_LVDS: + case ENCODER_OBJECT_ID_INTERNAL_TMDS1: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_TMDS1: + case ENCODER_OBJECT_ID_INTERNAL_LVTM1: + if (radeon_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) { + radeon_encoder->rmx_type = RMX_FULL; + drm_encoder_init(dev, encoder, &radeon_atom_enc_funcs, DRM_MODE_ENCODER_LVDS); + radeon_encoder->enc_priv = radeon_atombios_get_lvds_info(radeon_encoder); + } else { + drm_encoder_init(dev, encoder, &radeon_atom_enc_funcs, DRM_MODE_ENCODER_TMDS); + radeon_encoder->enc_priv = radeon_atombios_set_dig_info(radeon_encoder); + } + drm_encoder_helper_add(encoder, &radeon_atom_dig_helper_funcs); + break; + case ENCODER_OBJECT_ID_INTERNAL_DAC1: + drm_encoder_init(dev, encoder, &radeon_atom_enc_funcs, DRM_MODE_ENCODER_DAC); + radeon_encoder->enc_priv = radeon_atombios_set_dac_info(radeon_encoder); + drm_encoder_helper_add(encoder, &radeon_atom_dac_helper_funcs); + break; + case ENCODER_OBJECT_ID_INTERNAL_DAC2: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2: + drm_encoder_init(dev, encoder, &radeon_atom_enc_funcs, DRM_MODE_ENCODER_TVDAC); + radeon_encoder->enc_priv = radeon_atombios_set_dac_info(radeon_encoder); + drm_encoder_helper_add(encoder, &radeon_atom_dac_helper_funcs); + break; + case ENCODER_OBJECT_ID_INTERNAL_DVO1: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1: + case ENCODER_OBJECT_ID_INTERNAL_DDI: + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA: + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1: + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2: + if (radeon_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) { + radeon_encoder->rmx_type = RMX_FULL; + drm_encoder_init(dev, encoder, &radeon_atom_enc_funcs, DRM_MODE_ENCODER_LVDS); + radeon_encoder->enc_priv = radeon_atombios_get_lvds_info(radeon_encoder); + } else if (radeon_encoder->devices & (ATOM_DEVICE_CRT_SUPPORT)) { + drm_encoder_init(dev, encoder, &radeon_atom_enc_funcs, DRM_MODE_ENCODER_DAC); + radeon_encoder->enc_priv = radeon_atombios_set_dig_info(radeon_encoder); + } else { + drm_encoder_init(dev, encoder, &radeon_atom_enc_funcs, DRM_MODE_ENCODER_TMDS); + radeon_encoder->enc_priv = radeon_atombios_set_dig_info(radeon_encoder); + } + drm_encoder_helper_add(encoder, &radeon_atom_dig_helper_funcs); + break; + case ENCODER_OBJECT_ID_SI170B: + case ENCODER_OBJECT_ID_CH7303: + case ENCODER_OBJECT_ID_EXTERNAL_SDVOA: + case ENCODER_OBJECT_ID_EXTERNAL_SDVOB: + case ENCODER_OBJECT_ID_TITFP513: + case ENCODER_OBJECT_ID_VT1623: + case ENCODER_OBJECT_ID_HDMI_SI1930: + case ENCODER_OBJECT_ID_TRAVIS: + case ENCODER_OBJECT_ID_NUTMEG: + /* these are handled by the primary encoders */ + radeon_encoder->is_ext_encoder = true; + if (radeon_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) + drm_encoder_init(dev, encoder, &radeon_atom_enc_funcs, DRM_MODE_ENCODER_LVDS); + else if (radeon_encoder->devices & (ATOM_DEVICE_CRT_SUPPORT)) + drm_encoder_init(dev, encoder, &radeon_atom_enc_funcs, DRM_MODE_ENCODER_DAC); + else + drm_encoder_init(dev, encoder, &radeon_atom_enc_funcs, DRM_MODE_ENCODER_TMDS); + drm_encoder_helper_add(encoder, &radeon_atom_ext_helper_funcs); + break; + } +} diff --git a/drivers/gpu/drm/radeon/radeon_family.h b/drivers/gpu/drm/radeon/radeon_family.h new file mode 100644 index 00000000..ec2f1ea8 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_family.h @@ -0,0 +1,110 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ + +/* this file defines the CHIP_ and family flags used in the pciids, + * its is common between kms and non-kms because duplicating it and + * changing one place is fail. + */ +#ifndef RADEON_FAMILY_H +#define RADEON_FAMILY_H +/* + * Radeon chip families + */ +enum radeon_family { + CHIP_R100 = 0, + CHIP_RV100, + CHIP_RS100, + CHIP_RV200, + CHIP_RS200, + CHIP_R200, + CHIP_RV250, + CHIP_RS300, + CHIP_RV280, + CHIP_R300, + CHIP_R350, + CHIP_RV350, + CHIP_RV380, + CHIP_R420, + CHIP_R423, + CHIP_RV410, + CHIP_RS400, + CHIP_RS480, + CHIP_RS600, + CHIP_RS690, + CHIP_RS740, + CHIP_RV515, + CHIP_R520, + CHIP_RV530, + CHIP_RV560, + CHIP_RV570, + CHIP_R580, + CHIP_R600, + CHIP_RV610, + CHIP_RV630, + CHIP_RV670, + CHIP_RV620, + CHIP_RV635, + CHIP_RS780, + CHIP_RS880, + CHIP_RV770, + CHIP_RV730, + CHIP_RV710, + CHIP_RV740, + CHIP_CEDAR, + CHIP_REDWOOD, + CHIP_JUNIPER, + CHIP_CYPRESS, + CHIP_HEMLOCK, + CHIP_PALM, + CHIP_SUMO, + CHIP_SUMO2, + CHIP_BARTS, + CHIP_TURKS, + CHIP_CAICOS, + CHIP_CAYMAN, + CHIP_LAST, +}; + +/* + * Chip flags + */ +enum radeon_chip_flags { + RADEON_FAMILY_MASK = 0x0000ffffUL, + RADEON_FLAGS_MASK = 0xffff0000UL, + RADEON_IS_MOBILITY = 0x00010000UL, + RADEON_IS_IGP = 0x00020000UL, + RADEON_SINGLE_CRTC = 0x00040000UL, + RADEON_IS_AGP = 0x00080000UL, + RADEON_HAS_HIERZ = 0x00100000UL, + RADEON_IS_PCIE = 0x00200000UL, + RADEON_NEW_MEMMAP = 0x00400000UL, + RADEON_IS_PCI = 0x00800000UL, + RADEON_IS_IGPGART = 0x01000000UL, +}; + +#endif diff --git a/drivers/gpu/drm/radeon/radeon_fb.c b/drivers/gpu/drm/radeon/radeon_fb.c new file mode 100644 index 00000000..0b7b486c --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_fb.c @@ -0,0 +1,419 @@ +/* + * Copyright © 2007 David Airlie + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * David Airlie + */ +#include +#include +#include + +#include "drmP.h" +#include "drm.h" +#include "drm_crtc.h" +#include "drm_crtc_helper.h" +#include "radeon_drm.h" +#include "radeon.h" + +#include "drm_fb_helper.h" + +#include + +/* object hierarchy - + this contains a helper + a radeon fb + the helper contains a pointer to radeon framebuffer baseclass. +*/ +struct radeon_fbdev { + struct drm_fb_helper helper; + struct radeon_framebuffer rfb; + struct list_head fbdev_list; + struct radeon_device *rdev; +}; + +static struct fb_ops radeonfb_ops = { + .owner = THIS_MODULE, + .fb_check_var = drm_fb_helper_check_var, + .fb_set_par = drm_fb_helper_set_par, + .fb_fillrect = cfb_fillrect, + .fb_copyarea = cfb_copyarea, + .fb_imageblit = cfb_imageblit, + .fb_pan_display = drm_fb_helper_pan_display, + .fb_blank = drm_fb_helper_blank, + .fb_setcmap = drm_fb_helper_setcmap, + .fb_debug_enter = drm_fb_helper_debug_enter, + .fb_debug_leave = drm_fb_helper_debug_leave, +}; + + +int radeon_align_pitch(struct radeon_device *rdev, int width, int bpp, bool tiled) +{ + int aligned = width; + int align_large = (ASIC_IS_AVIVO(rdev)) || tiled; + int pitch_mask = 0; + + switch (bpp / 8) { + case 1: + pitch_mask = align_large ? 255 : 127; + break; + case 2: + pitch_mask = align_large ? 127 : 31; + break; + case 3: + case 4: + pitch_mask = align_large ? 63 : 15; + break; + } + + aligned += pitch_mask; + aligned &= ~pitch_mask; + return aligned; +} + +static void radeonfb_destroy_pinned_object(struct drm_gem_object *gobj) +{ + struct radeon_bo *rbo = gem_to_radeon_bo(gobj); + int ret; + + ret = radeon_bo_reserve(rbo, false); + if (likely(ret == 0)) { + radeon_bo_kunmap(rbo); + radeon_bo_unpin(rbo); + radeon_bo_unreserve(rbo); + } + drm_gem_object_unreference_unlocked(gobj); +} + +static int radeonfb_create_pinned_object(struct radeon_fbdev *rfbdev, + struct drm_mode_fb_cmd *mode_cmd, + struct drm_gem_object **gobj_p) +{ + struct radeon_device *rdev = rfbdev->rdev; + struct drm_gem_object *gobj = NULL; + struct radeon_bo *rbo = NULL; + bool fb_tiled = false; /* useful for testing */ + u32 tiling_flags = 0; + int ret; + int aligned_size, size; + int height = mode_cmd->height; + + /* need to align pitch with crtc limits */ + mode_cmd->pitch = radeon_align_pitch(rdev, mode_cmd->width, mode_cmd->bpp, fb_tiled) * ((mode_cmd->bpp + 1) / 8); + + if (rdev->family >= CHIP_R600) + height = ALIGN(mode_cmd->height, 8); + size = mode_cmd->pitch * height; + aligned_size = ALIGN(size, PAGE_SIZE); + ret = radeon_gem_object_create(rdev, aligned_size, 0, + RADEON_GEM_DOMAIN_VRAM, + false, true, + &gobj); + if (ret) { + printk(KERN_ERR "failed to allocate framebuffer (%d)\n", + aligned_size); + return -ENOMEM; + } + rbo = gem_to_radeon_bo(gobj); + + if (fb_tiled) + tiling_flags = RADEON_TILING_MACRO; + +#ifdef __BIG_ENDIAN + switch (mode_cmd->bpp) { + case 32: + tiling_flags |= RADEON_TILING_SWAP_32BIT; + break; + case 16: + tiling_flags |= RADEON_TILING_SWAP_16BIT; + default: + break; + } +#endif + + if (tiling_flags) { + ret = radeon_bo_set_tiling_flags(rbo, + tiling_flags | RADEON_TILING_SURFACE, + mode_cmd->pitch); + if (ret) + dev_err(rdev->dev, "FB failed to set tiling flags\n"); + } + + + ret = radeon_bo_reserve(rbo, false); + if (unlikely(ret != 0)) + goto out_unref; + ret = radeon_bo_pin(rbo, RADEON_GEM_DOMAIN_VRAM, NULL); + if (ret) { + radeon_bo_unreserve(rbo); + goto out_unref; + } + if (fb_tiled) + radeon_bo_check_tiling(rbo, 0, 0); + ret = radeon_bo_kmap(rbo, NULL); + radeon_bo_unreserve(rbo); + if (ret) { + goto out_unref; + } + + *gobj_p = gobj; + return 0; +out_unref: + radeonfb_destroy_pinned_object(gobj); + *gobj_p = NULL; + return ret; +} + +static int radeonfb_create(struct radeon_fbdev *rfbdev, + struct drm_fb_helper_surface_size *sizes) +{ + struct radeon_device *rdev = rfbdev->rdev; + struct fb_info *info; + struct drm_framebuffer *fb = NULL; + struct drm_mode_fb_cmd mode_cmd; + struct drm_gem_object *gobj = NULL; + struct radeon_bo *rbo = NULL; + struct device *device = &rdev->pdev->dev; + int ret; + unsigned long tmp; + + mode_cmd.width = sizes->surface_width; + mode_cmd.height = sizes->surface_height; + + /* avivo can't scanout real 24bpp */ + if ((sizes->surface_bpp == 24) && ASIC_IS_AVIVO(rdev)) + sizes->surface_bpp = 32; + + mode_cmd.bpp = sizes->surface_bpp; + mode_cmd.depth = sizes->surface_depth; + + ret = radeonfb_create_pinned_object(rfbdev, &mode_cmd, &gobj); + rbo = gem_to_radeon_bo(gobj); + + /* okay we have an object now allocate the framebuffer */ + info = framebuffer_alloc(0, device); + if (info == NULL) { + ret = -ENOMEM; + goto out_unref; + } + + info->par = rfbdev; + + radeon_framebuffer_init(rdev->ddev, &rfbdev->rfb, &mode_cmd, gobj); + + fb = &rfbdev->rfb.base; + + /* setup helper */ + rfbdev->helper.fb = fb; + rfbdev->helper.fbdev = info; + + memset_io(rbo->kptr, 0x0, radeon_bo_size(rbo)); + + strcpy(info->fix.id, "radeondrmfb"); + + drm_fb_helper_fill_fix(info, fb->pitch, fb->depth); + + info->flags = FBINFO_DEFAULT | FBINFO_CAN_FORCE_OUTPUT; + info->fbops = &radeonfb_ops; + + tmp = radeon_bo_gpu_offset(rbo) - rdev->mc.vram_start; + info->fix.smem_start = rdev->mc.aper_base + tmp; + info->fix.smem_len = radeon_bo_size(rbo); + info->screen_base = rbo->kptr; + info->screen_size = radeon_bo_size(rbo); + + drm_fb_helper_fill_var(info, &rfbdev->helper, sizes->fb_width, sizes->fb_height); + + /* setup aperture base/size for vesafb takeover */ + info->apertures = alloc_apertures(1); + if (!info->apertures) { + ret = -ENOMEM; + goto out_unref; + } + info->apertures->ranges[0].base = rdev->ddev->mode_config.fb_base; + info->apertures->ranges[0].size = rdev->mc.aper_size; + + info->pixmap.size = 64*1024; + info->pixmap.buf_align = 8; + info->pixmap.access_align = 32; + info->pixmap.flags = FB_PIXMAP_SYSTEM; + info->pixmap.scan_align = 1; + + if (info->screen_base == NULL) { + ret = -ENOSPC; + goto out_unref; + } + + ret = fb_alloc_cmap(&info->cmap, 256, 0); + if (ret) { + ret = -ENOMEM; + goto out_unref; + } + + DRM_INFO("fb mappable at 0x%lX\n", info->fix.smem_start); + DRM_INFO("vram apper at 0x%lX\n", (unsigned long)rdev->mc.aper_base); + DRM_INFO("size %lu\n", (unsigned long)radeon_bo_size(rbo)); + DRM_INFO("fb depth is %d\n", fb->depth); + DRM_INFO(" pitch is %d\n", fb->pitch); + + vga_switcheroo_client_fb_set(rdev->ddev->pdev, info); + return 0; + +out_unref: + if (rbo) { + + } + if (fb && ret) { + drm_gem_object_unreference(gobj); + drm_framebuffer_cleanup(fb); + kfree(fb); + } + return ret; +} + +static int radeon_fb_find_or_create_single(struct drm_fb_helper *helper, + struct drm_fb_helper_surface_size *sizes) +{ + struct radeon_fbdev *rfbdev = (struct radeon_fbdev *)helper; + int new_fb = 0; + int ret; + + if (!helper->fb) { + ret = radeonfb_create(rfbdev, sizes); + if (ret) + return ret; + new_fb = 1; + } + return new_fb; +} + +static char *mode_option; +int radeon_parse_options(char *options) +{ + char *this_opt; + + if (!options || !*options) + return 0; + + while ((this_opt = strsep(&options, ",")) != NULL) { + if (!*this_opt) + continue; + mode_option = this_opt; + } + return 0; +} + +void radeon_fb_output_poll_changed(struct radeon_device *rdev) +{ + drm_fb_helper_hotplug_event(&rdev->mode_info.rfbdev->helper); +} + +static int radeon_fbdev_destroy(struct drm_device *dev, struct radeon_fbdev *rfbdev) +{ + struct fb_info *info; + struct radeon_framebuffer *rfb = &rfbdev->rfb; + + if (rfbdev->helper.fbdev) { + info = rfbdev->helper.fbdev; + + unregister_framebuffer(info); + if (info->cmap.len) + fb_dealloc_cmap(&info->cmap); + framebuffer_release(info); + } + + if (rfb->obj) { + radeonfb_destroy_pinned_object(rfb->obj); + rfb->obj = NULL; + } + drm_fb_helper_fini(&rfbdev->helper); + drm_framebuffer_cleanup(&rfb->base); + + return 0; +} + +static struct drm_fb_helper_funcs radeon_fb_helper_funcs = { + .gamma_set = radeon_crtc_fb_gamma_set, + .gamma_get = radeon_crtc_fb_gamma_get, + .fb_probe = radeon_fb_find_or_create_single, +}; + +int radeon_fbdev_init(struct radeon_device *rdev) +{ + struct radeon_fbdev *rfbdev; + int bpp_sel = 32; + int ret; + + /* select 8 bpp console on RN50 or 16MB cards */ + if (ASIC_IS_RN50(rdev) || rdev->mc.real_vram_size <= (32*1024*1024)) + bpp_sel = 8; + + rfbdev = kzalloc(sizeof(struct radeon_fbdev), GFP_KERNEL); + if (!rfbdev) + return -ENOMEM; + + rfbdev->rdev = rdev; + rdev->mode_info.rfbdev = rfbdev; + rfbdev->helper.funcs = &radeon_fb_helper_funcs; + + ret = drm_fb_helper_init(rdev->ddev, &rfbdev->helper, + rdev->num_crtc, + RADEONFB_CONN_LIMIT); + if (ret) { + kfree(rfbdev); + return ret; + } + + drm_fb_helper_single_add_all_connectors(&rfbdev->helper); + drm_fb_helper_initial_config(&rfbdev->helper, bpp_sel); + return 0; +} + +void radeon_fbdev_fini(struct radeon_device *rdev) +{ + if (!rdev->mode_info.rfbdev) + return; + + radeon_fbdev_destroy(rdev->ddev, rdev->mode_info.rfbdev); + kfree(rdev->mode_info.rfbdev); + rdev->mode_info.rfbdev = NULL; +} + +void radeon_fbdev_set_suspend(struct radeon_device *rdev, int state) +{ + fb_set_suspend(rdev->mode_info.rfbdev->helper.fbdev, state); +} + +int radeon_fbdev_total_size(struct radeon_device *rdev) +{ + struct radeon_bo *robj; + int size = 0; + + robj = gem_to_radeon_bo(rdev->mode_info.rfbdev->rfb.obj); + size += radeon_bo_size(robj); + return size; +} + +bool radeon_fbdev_robj_is_fb(struct radeon_device *rdev, struct radeon_bo *robj) +{ + if (robj == gem_to_radeon_bo(rdev->mode_info.rfbdev->rfb.obj)) + return true; + return false; +} diff --git a/drivers/gpu/drm/radeon/radeon_fence.c b/drivers/gpu/drm/radeon/radeon_fence.c new file mode 100644 index 00000000..021d2b6b --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_fence.c @@ -0,0 +1,437 @@ +/* + * Copyright 2009 Jerome Glisse. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + */ +/* + * Authors: + * Jerome Glisse + * Dave Airlie + */ +#include +#include +#include +#include +#include +#include +#include "drmP.h" +#include "drm.h" +#include "radeon_reg.h" +#include "radeon.h" +#include "radeon_trace.h" + +static void radeon_fence_write(struct radeon_device *rdev, u32 seq) +{ + if (rdev->wb.enabled) { + u32 scratch_index; + if (rdev->wb.use_event) + scratch_index = R600_WB_EVENT_OFFSET + rdev->fence_drv.scratch_reg - rdev->scratch.reg_base; + else + scratch_index = RADEON_WB_SCRATCH_OFFSET + rdev->fence_drv.scratch_reg - rdev->scratch.reg_base; + rdev->wb.wb[scratch_index/4] = cpu_to_le32(seq);; + } else + WREG32(rdev->fence_drv.scratch_reg, seq); +} + +static u32 radeon_fence_read(struct radeon_device *rdev) +{ + u32 seq; + + if (rdev->wb.enabled) { + u32 scratch_index; + if (rdev->wb.use_event) + scratch_index = R600_WB_EVENT_OFFSET + rdev->fence_drv.scratch_reg - rdev->scratch.reg_base; + else + scratch_index = RADEON_WB_SCRATCH_OFFSET + rdev->fence_drv.scratch_reg - rdev->scratch.reg_base; + seq = le32_to_cpu(rdev->wb.wb[scratch_index/4]); + } else + seq = RREG32(rdev->fence_drv.scratch_reg); + return seq; +} + +int radeon_fence_emit(struct radeon_device *rdev, struct radeon_fence *fence) +{ + unsigned long irq_flags; + + write_lock_irqsave(&rdev->fence_drv.lock, irq_flags); + if (fence->emited) { + write_unlock_irqrestore(&rdev->fence_drv.lock, irq_flags); + return 0; + } + fence->seq = atomic_add_return(1, &rdev->fence_drv.seq); + if (!rdev->cp.ready) + /* FIXME: cp is not running assume everythings is done right + * away + */ + radeon_fence_write(rdev, fence->seq); + else + radeon_fence_ring_emit(rdev, fence); + + trace_radeon_fence_emit(rdev->ddev, fence->seq); + fence->emited = true; + list_move_tail(&fence->list, &rdev->fence_drv.emited); + write_unlock_irqrestore(&rdev->fence_drv.lock, irq_flags); + return 0; +} + +static bool radeon_fence_poll_locked(struct radeon_device *rdev) +{ + struct radeon_fence *fence; + struct list_head *i, *n; + uint32_t seq; + bool wake = false; + unsigned long cjiffies; + + seq = radeon_fence_read(rdev); + if (seq != rdev->fence_drv.last_seq) { + rdev->fence_drv.last_seq = seq; + rdev->fence_drv.last_jiffies = jiffies; + rdev->fence_drv.last_timeout = RADEON_FENCE_JIFFIES_TIMEOUT; + } else { + cjiffies = jiffies; + if (time_after(cjiffies, rdev->fence_drv.last_jiffies)) { + cjiffies -= rdev->fence_drv.last_jiffies; + if (time_after(rdev->fence_drv.last_timeout, cjiffies)) { + /* update the timeout */ + rdev->fence_drv.last_timeout -= cjiffies; + } else { + /* the 500ms timeout is elapsed we should test + * for GPU lockup + */ + rdev->fence_drv.last_timeout = 1; + } + } else { + /* wrap around update last jiffies, we will just wait + * a little longer + */ + rdev->fence_drv.last_jiffies = cjiffies; + } + return false; + } + n = NULL; + list_for_each(i, &rdev->fence_drv.emited) { + fence = list_entry(i, struct radeon_fence, list); + if (fence->seq == seq) { + n = i; + break; + } + } + /* all fence previous to this one are considered as signaled */ + if (n) { + i = n; + do { + n = i->prev; + list_move_tail(i, &rdev->fence_drv.signaled); + fence = list_entry(i, struct radeon_fence, list); + fence->signaled = true; + i = n; + } while (i != &rdev->fence_drv.emited); + wake = true; + } + return wake; +} + +static void radeon_fence_destroy(struct kref *kref) +{ + unsigned long irq_flags; + struct radeon_fence *fence; + + fence = container_of(kref, struct radeon_fence, kref); + write_lock_irqsave(&fence->rdev->fence_drv.lock, irq_flags); + list_del(&fence->list); + fence->emited = false; + write_unlock_irqrestore(&fence->rdev->fence_drv.lock, irq_flags); + kfree(fence); +} + +int radeon_fence_create(struct radeon_device *rdev, struct radeon_fence **fence) +{ + unsigned long irq_flags; + + *fence = kmalloc(sizeof(struct radeon_fence), GFP_KERNEL); + if ((*fence) == NULL) { + return -ENOMEM; + } + kref_init(&((*fence)->kref)); + (*fence)->rdev = rdev; + (*fence)->emited = false; + (*fence)->signaled = false; + (*fence)->seq = 0; + INIT_LIST_HEAD(&(*fence)->list); + + write_lock_irqsave(&rdev->fence_drv.lock, irq_flags); + list_add_tail(&(*fence)->list, &rdev->fence_drv.created); + write_unlock_irqrestore(&rdev->fence_drv.lock, irq_flags); + return 0; +} + + +bool radeon_fence_signaled(struct radeon_fence *fence) +{ + unsigned long irq_flags; + bool signaled = false; + + if (!fence) + return true; + + if (fence->rdev->gpu_lockup) + return true; + + write_lock_irqsave(&fence->rdev->fence_drv.lock, irq_flags); + signaled = fence->signaled; + /* if we are shuting down report all fence as signaled */ + if (fence->rdev->shutdown) { + signaled = true; + } + if (!fence->emited) { + WARN(1, "Querying an unemited fence : %p !\n", fence); + signaled = true; + } + if (!signaled) { + radeon_fence_poll_locked(fence->rdev); + signaled = fence->signaled; + } + write_unlock_irqrestore(&fence->rdev->fence_drv.lock, irq_flags); + return signaled; +} + +int radeon_fence_wait(struct radeon_fence *fence, bool intr) +{ + struct radeon_device *rdev; + unsigned long irq_flags, timeout; + u32 seq; + int r; + + if (fence == NULL) { + WARN(1, "Querying an invalid fence : %p !\n", fence); + return 0; + } + rdev = fence->rdev; + if (radeon_fence_signaled(fence)) { + return 0; + } + timeout = rdev->fence_drv.last_timeout; +retry: + /* save current sequence used to check for GPU lockup */ + seq = rdev->fence_drv.last_seq; + trace_radeon_fence_wait_begin(rdev->ddev, seq); + if (intr) { + radeon_irq_kms_sw_irq_get(rdev); + r = wait_event_interruptible_timeout(rdev->fence_drv.queue, + radeon_fence_signaled(fence), timeout); + radeon_irq_kms_sw_irq_put(rdev); + if (unlikely(r < 0)) { + return r; + } + } else { + radeon_irq_kms_sw_irq_get(rdev); + r = wait_event_timeout(rdev->fence_drv.queue, + radeon_fence_signaled(fence), timeout); + radeon_irq_kms_sw_irq_put(rdev); + } + trace_radeon_fence_wait_end(rdev->ddev, seq); + if (unlikely(!radeon_fence_signaled(fence))) { + /* we were interrupted for some reason and fence isn't + * isn't signaled yet, resume wait + */ + if (r) { + timeout = r; + goto retry; + } + /* don't protect read access to rdev->fence_drv.last_seq + * if we experiencing a lockup the value doesn't change + */ + if (seq == rdev->fence_drv.last_seq && radeon_gpu_is_lockup(rdev)) { + /* good news we believe it's a lockup */ + WARN(1, "GPU lockup (waiting for 0x%08X last fence id 0x%08X)\n", + fence->seq, seq); + /* FIXME: what should we do ? marking everyone + * as signaled for now + */ + rdev->gpu_lockup = true; + r = radeon_gpu_reset(rdev); + if (r) + return r; + radeon_fence_write(rdev, fence->seq); + rdev->gpu_lockup = false; + } + timeout = RADEON_FENCE_JIFFIES_TIMEOUT; + write_lock_irqsave(&rdev->fence_drv.lock, irq_flags); + rdev->fence_drv.last_timeout = RADEON_FENCE_JIFFIES_TIMEOUT; + rdev->fence_drv.last_jiffies = jiffies; + write_unlock_irqrestore(&rdev->fence_drv.lock, irq_flags); + goto retry; + } + return 0; +} + +int radeon_fence_wait_next(struct radeon_device *rdev) +{ + unsigned long irq_flags; + struct radeon_fence *fence; + int r; + + if (rdev->gpu_lockup) { + return 0; + } + write_lock_irqsave(&rdev->fence_drv.lock, irq_flags); + if (list_empty(&rdev->fence_drv.emited)) { + write_unlock_irqrestore(&rdev->fence_drv.lock, irq_flags); + return 0; + } + fence = list_entry(rdev->fence_drv.emited.next, + struct radeon_fence, list); + radeon_fence_ref(fence); + write_unlock_irqrestore(&rdev->fence_drv.lock, irq_flags); + r = radeon_fence_wait(fence, false); + radeon_fence_unref(&fence); + return r; +} + +int radeon_fence_wait_last(struct radeon_device *rdev) +{ + unsigned long irq_flags; + struct radeon_fence *fence; + int r; + + if (rdev->gpu_lockup) { + return 0; + } + write_lock_irqsave(&rdev->fence_drv.lock, irq_flags); + if (list_empty(&rdev->fence_drv.emited)) { + write_unlock_irqrestore(&rdev->fence_drv.lock, irq_flags); + return 0; + } + fence = list_entry(rdev->fence_drv.emited.prev, + struct radeon_fence, list); + radeon_fence_ref(fence); + write_unlock_irqrestore(&rdev->fence_drv.lock, irq_flags); + r = radeon_fence_wait(fence, false); + radeon_fence_unref(&fence); + return r; +} + +struct radeon_fence *radeon_fence_ref(struct radeon_fence *fence) +{ + kref_get(&fence->kref); + return fence; +} + +void radeon_fence_unref(struct radeon_fence **fence) +{ + struct radeon_fence *tmp = *fence; + + *fence = NULL; + if (tmp) { + kref_put(&tmp->kref, radeon_fence_destroy); + } +} + +void radeon_fence_process(struct radeon_device *rdev) +{ + unsigned long irq_flags; + bool wake; + + write_lock_irqsave(&rdev->fence_drv.lock, irq_flags); + wake = radeon_fence_poll_locked(rdev); + write_unlock_irqrestore(&rdev->fence_drv.lock, irq_flags); + if (wake) { + wake_up_all(&rdev->fence_drv.queue); + } +} + +int radeon_fence_driver_init(struct radeon_device *rdev) +{ + unsigned long irq_flags; + int r; + + write_lock_irqsave(&rdev->fence_drv.lock, irq_flags); + r = radeon_scratch_get(rdev, &rdev->fence_drv.scratch_reg); + if (r) { + dev_err(rdev->dev, "fence failed to get scratch register\n"); + write_unlock_irqrestore(&rdev->fence_drv.lock, irq_flags); + return r; + } + radeon_fence_write(rdev, 0); + atomic_set(&rdev->fence_drv.seq, 0); + INIT_LIST_HEAD(&rdev->fence_drv.created); + INIT_LIST_HEAD(&rdev->fence_drv.emited); + INIT_LIST_HEAD(&rdev->fence_drv.signaled); + init_waitqueue_head(&rdev->fence_drv.queue); + rdev->fence_drv.initialized = true; + write_unlock_irqrestore(&rdev->fence_drv.lock, irq_flags); + if (radeon_debugfs_fence_init(rdev)) { + dev_err(rdev->dev, "fence debugfs file creation failed\n"); + } + return 0; +} + +void radeon_fence_driver_fini(struct radeon_device *rdev) +{ + unsigned long irq_flags; + + if (!rdev->fence_drv.initialized) + return; + wake_up_all(&rdev->fence_drv.queue); + write_lock_irqsave(&rdev->fence_drv.lock, irq_flags); + radeon_scratch_free(rdev, rdev->fence_drv.scratch_reg); + write_unlock_irqrestore(&rdev->fence_drv.lock, irq_flags); + rdev->fence_drv.initialized = false; +} + + +/* + * Fence debugfs + */ +#if defined(CONFIG_DEBUG_FS) +static int radeon_debugfs_fence_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *)m->private; + struct drm_device *dev = node->minor->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_fence *fence; + + seq_printf(m, "Last signaled fence 0x%08X\n", + radeon_fence_read(rdev)); + if (!list_empty(&rdev->fence_drv.emited)) { + fence = list_entry(rdev->fence_drv.emited.prev, + struct radeon_fence, list); + seq_printf(m, "Last emited fence %p with 0x%08X\n", + fence, fence->seq); + } + return 0; +} + +static struct drm_info_list radeon_debugfs_fence_list[] = { + {"radeon_fence_info", &radeon_debugfs_fence_info, 0, NULL}, +}; +#endif + +int radeon_debugfs_fence_init(struct radeon_device *rdev) +{ +#if defined(CONFIG_DEBUG_FS) + return radeon_debugfs_add_files(rdev, radeon_debugfs_fence_list, 1); +#else + return 0; +#endif +} diff --git a/drivers/gpu/drm/radeon/radeon_gart.c b/drivers/gpu/drm/radeon/radeon_gart.c new file mode 100644 index 00000000..a533f52f --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_gart.c @@ -0,0 +1,290 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#include "drmP.h" +#include "radeon_drm.h" +#include "radeon.h" +#include "radeon_reg.h" + +/* + * Common GART table functions. + */ +int radeon_gart_table_ram_alloc(struct radeon_device *rdev) +{ + void *ptr; + + ptr = pci_alloc_consistent(rdev->pdev, rdev->gart.table_size, + &rdev->gart.table_addr); + if (ptr == NULL) { + return -ENOMEM; + } +#ifdef CONFIG_X86 + if (rdev->family == CHIP_RS400 || rdev->family == CHIP_RS480 || + rdev->family == CHIP_RS690 || rdev->family == CHIP_RS740) { + set_memory_uc((unsigned long)ptr, + rdev->gart.table_size >> PAGE_SHIFT); + } +#endif + rdev->gart.table.ram.ptr = ptr; + memset((void *)rdev->gart.table.ram.ptr, 0, rdev->gart.table_size); + return 0; +} + +void radeon_gart_table_ram_free(struct radeon_device *rdev) +{ + if (rdev->gart.table.ram.ptr == NULL) { + return; + } +#ifdef CONFIG_X86 + if (rdev->family == CHIP_RS400 || rdev->family == CHIP_RS480 || + rdev->family == CHIP_RS690 || rdev->family == CHIP_RS740) { + set_memory_wb((unsigned long)rdev->gart.table.ram.ptr, + rdev->gart.table_size >> PAGE_SHIFT); + } +#endif + pci_free_consistent(rdev->pdev, rdev->gart.table_size, + (void *)rdev->gart.table.ram.ptr, + rdev->gart.table_addr); + rdev->gart.table.ram.ptr = NULL; + rdev->gart.table_addr = 0; +} + +int radeon_gart_table_vram_alloc(struct radeon_device *rdev) +{ + int r; + + if (rdev->gart.table.vram.robj == NULL) { + r = radeon_bo_create(rdev, rdev->gart.table_size, + PAGE_SIZE, true, RADEON_GEM_DOMAIN_VRAM, + &rdev->gart.table.vram.robj); + if (r) { + return r; + } + } + return 0; +} + +int radeon_gart_table_vram_pin(struct radeon_device *rdev) +{ + uint64_t gpu_addr; + int r; + + r = radeon_bo_reserve(rdev->gart.table.vram.robj, false); + if (unlikely(r != 0)) + return r; + r = radeon_bo_pin(rdev->gart.table.vram.robj, + RADEON_GEM_DOMAIN_VRAM, &gpu_addr); + if (r) { + radeon_bo_unreserve(rdev->gart.table.vram.robj); + return r; + } + r = radeon_bo_kmap(rdev->gart.table.vram.robj, + (void **)&rdev->gart.table.vram.ptr); + if (r) + radeon_bo_unpin(rdev->gart.table.vram.robj); + radeon_bo_unreserve(rdev->gart.table.vram.robj); + rdev->gart.table_addr = gpu_addr; + return r; +} + +void radeon_gart_table_vram_free(struct radeon_device *rdev) +{ + int r; + + if (rdev->gart.table.vram.robj == NULL) { + return; + } + r = radeon_bo_reserve(rdev->gart.table.vram.robj, false); + if (likely(r == 0)) { + radeon_bo_kunmap(rdev->gart.table.vram.robj); + radeon_bo_unpin(rdev->gart.table.vram.robj); + radeon_bo_unreserve(rdev->gart.table.vram.robj); + } + radeon_bo_unref(&rdev->gart.table.vram.robj); +} + + + + +/* + * Common gart functions. + */ +void radeon_gart_unbind(struct radeon_device *rdev, unsigned offset, + int pages) +{ + unsigned t; + unsigned p; + int i, j; + u64 page_base; + + if (!rdev->gart.ready) { + WARN(1, "trying to unbind memory to unitialized GART !\n"); + return; + } + t = offset / RADEON_GPU_PAGE_SIZE; + p = t / (PAGE_SIZE / RADEON_GPU_PAGE_SIZE); + for (i = 0; i < pages; i++, p++) { + if (rdev->gart.pages[p]) { + if (!rdev->gart.ttm_alloced[p]) + pci_unmap_page(rdev->pdev, rdev->gart.pages_addr[p], + PAGE_SIZE, PCI_DMA_BIDIRECTIONAL); + rdev->gart.pages[p] = NULL; + rdev->gart.pages_addr[p] = rdev->dummy_page.addr; + page_base = rdev->gart.pages_addr[p]; + for (j = 0; j < (PAGE_SIZE / RADEON_GPU_PAGE_SIZE); j++, t++) { + radeon_gart_set_page(rdev, t, page_base); + page_base += RADEON_GPU_PAGE_SIZE; + } + } + } + mb(); + radeon_gart_tlb_flush(rdev); +} + +int radeon_gart_bind(struct radeon_device *rdev, unsigned offset, + int pages, struct page **pagelist, dma_addr_t *dma_addr) +{ + unsigned t; + unsigned p; + uint64_t page_base; + int i, j; + + if (!rdev->gart.ready) { + WARN(1, "trying to bind memory to unitialized GART !\n"); + return -EINVAL; + } + t = offset / RADEON_GPU_PAGE_SIZE; + p = t / (PAGE_SIZE / RADEON_GPU_PAGE_SIZE); + + for (i = 0; i < pages; i++, p++) { + /* we reverted the patch using dma_addr in TTM for now but this + * code stops building on alpha so just comment it out for now */ + if (0) { /*dma_addr[i] != DMA_ERROR_CODE) */ + rdev->gart.ttm_alloced[p] = true; + rdev->gart.pages_addr[p] = dma_addr[i]; + } else { + /* we need to support large memory configurations */ + /* assume that unbind have already been call on the range */ + rdev->gart.pages_addr[p] = pci_map_page(rdev->pdev, pagelist[i], + 0, PAGE_SIZE, + PCI_DMA_BIDIRECTIONAL); + if (pci_dma_mapping_error(rdev->pdev, rdev->gart.pages_addr[p])) { + /* FIXME: failed to map page (return -ENOMEM?) */ + radeon_gart_unbind(rdev, offset, pages); + return -ENOMEM; + } + } + rdev->gart.pages[p] = pagelist[i]; + page_base = rdev->gart.pages_addr[p]; + for (j = 0; j < (PAGE_SIZE / RADEON_GPU_PAGE_SIZE); j++, t++) { + radeon_gart_set_page(rdev, t, page_base); + page_base += RADEON_GPU_PAGE_SIZE; + } + } + mb(); + radeon_gart_tlb_flush(rdev); + return 0; +} + +void radeon_gart_restore(struct radeon_device *rdev) +{ + int i, j, t; + u64 page_base; + + for (i = 0, t = 0; i < rdev->gart.num_cpu_pages; i++) { + page_base = rdev->gart.pages_addr[i]; + for (j = 0; j < (PAGE_SIZE / RADEON_GPU_PAGE_SIZE); j++, t++) { + radeon_gart_set_page(rdev, t, page_base); + page_base += RADEON_GPU_PAGE_SIZE; + } + } + mb(); + radeon_gart_tlb_flush(rdev); +} + +int radeon_gart_init(struct radeon_device *rdev) +{ + int r, i; + + if (rdev->gart.pages) { + return 0; + } + /* We need PAGE_SIZE >= RADEON_GPU_PAGE_SIZE */ + if (PAGE_SIZE < RADEON_GPU_PAGE_SIZE) { + DRM_ERROR("Page size is smaller than GPU page size!\n"); + return -EINVAL; + } + r = radeon_dummy_page_init(rdev); + if (r) + return r; + /* Compute table size */ + rdev->gart.num_cpu_pages = rdev->mc.gtt_size / PAGE_SIZE; + rdev->gart.num_gpu_pages = rdev->mc.gtt_size / RADEON_GPU_PAGE_SIZE; + DRM_INFO("GART: num cpu pages %u, num gpu pages %u\n", + rdev->gart.num_cpu_pages, rdev->gart.num_gpu_pages); + /* Allocate pages table */ + rdev->gart.pages = kzalloc(sizeof(void *) * rdev->gart.num_cpu_pages, + GFP_KERNEL); + if (rdev->gart.pages == NULL) { + radeon_gart_fini(rdev); + return -ENOMEM; + } + rdev->gart.pages_addr = kzalloc(sizeof(dma_addr_t) * + rdev->gart.num_cpu_pages, GFP_KERNEL); + if (rdev->gart.pages_addr == NULL) { + radeon_gart_fini(rdev); + return -ENOMEM; + } + rdev->gart.ttm_alloced = kzalloc(sizeof(bool) * + rdev->gart.num_cpu_pages, GFP_KERNEL); + if (rdev->gart.ttm_alloced == NULL) { + radeon_gart_fini(rdev); + return -ENOMEM; + } + /* set GART entry to point to the dummy page by default */ + for (i = 0; i < rdev->gart.num_cpu_pages; i++) { + rdev->gart.pages_addr[i] = rdev->dummy_page.addr; + } + return 0; +} + +void radeon_gart_fini(struct radeon_device *rdev) +{ + if (rdev->gart.pages && rdev->gart.pages_addr && rdev->gart.ready) { + /* unbind pages */ + radeon_gart_unbind(rdev, 0, rdev->gart.num_cpu_pages); + } + rdev->gart.ready = false; + kfree(rdev->gart.pages); + kfree(rdev->gart.pages_addr); + kfree(rdev->gart.ttm_alloced); + rdev->gart.pages = NULL; + rdev->gart.pages_addr = NULL; + rdev->gart.ttm_alloced = NULL; + + radeon_dummy_page_fini(rdev); +} diff --git a/drivers/gpu/drm/radeon/radeon_gem.c b/drivers/gpu/drm/radeon/radeon_gem.c new file mode 100644 index 00000000..aa1ca2de --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_gem.c @@ -0,0 +1,390 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#include "drmP.h" +#include "drm.h" +#include "radeon_drm.h" +#include "radeon.h" + +int radeon_gem_object_init(struct drm_gem_object *obj) +{ + BUG(); + + return 0; +} + +void radeon_gem_object_free(struct drm_gem_object *gobj) +{ + struct radeon_bo *robj = gem_to_radeon_bo(gobj); + + if (robj) { + radeon_bo_unref(&robj); + } +} + +int radeon_gem_object_create(struct radeon_device *rdev, int size, + int alignment, int initial_domain, + bool discardable, bool kernel, + struct drm_gem_object **obj) +{ + struct radeon_bo *robj; + int r; + + *obj = NULL; + /* At least align on page size */ + if (alignment < PAGE_SIZE) { + alignment = PAGE_SIZE; + } + r = radeon_bo_create(rdev, size, alignment, kernel, initial_domain, &robj); + if (r) { + if (r != -ERESTARTSYS) + DRM_ERROR("Failed to allocate GEM object (%d, %d, %u, %d)\n", + size, initial_domain, alignment, r); + return r; + } + *obj = &robj->gem_base; + + mutex_lock(&rdev->gem.mutex); + list_add_tail(&robj->list, &rdev->gem.objects); + mutex_unlock(&rdev->gem.mutex); + + return 0; +} + +int radeon_gem_object_pin(struct drm_gem_object *obj, uint32_t pin_domain, + uint64_t *gpu_addr) +{ + struct radeon_bo *robj = gem_to_radeon_bo(obj); + int r; + + r = radeon_bo_reserve(robj, false); + if (unlikely(r != 0)) + return r; + r = radeon_bo_pin(robj, pin_domain, gpu_addr); + radeon_bo_unreserve(robj); + return r; +} + +void radeon_gem_object_unpin(struct drm_gem_object *obj) +{ + struct radeon_bo *robj = gem_to_radeon_bo(obj); + int r; + + r = radeon_bo_reserve(robj, false); + if (likely(r == 0)) { + radeon_bo_unpin(robj); + radeon_bo_unreserve(robj); + } +} + +int radeon_gem_set_domain(struct drm_gem_object *gobj, + uint32_t rdomain, uint32_t wdomain) +{ + struct radeon_bo *robj; + uint32_t domain; + int r; + + /* FIXME: reeimplement */ + robj = gem_to_radeon_bo(gobj); + /* work out where to validate the buffer to */ + domain = wdomain; + if (!domain) { + domain = rdomain; + } + if (!domain) { + /* Do nothings */ + printk(KERN_WARNING "Set domain withou domain !\n"); + return 0; + } + if (domain == RADEON_GEM_DOMAIN_CPU) { + /* Asking for cpu access wait for object idle */ + r = radeon_bo_wait(robj, NULL, false); + if (r) { + printk(KERN_ERR "Failed to wait for object !\n"); + return r; + } + } + return 0; +} + +int radeon_gem_init(struct radeon_device *rdev) +{ + INIT_LIST_HEAD(&rdev->gem.objects); + return 0; +} + +void radeon_gem_fini(struct radeon_device *rdev) +{ + radeon_bo_force_delete(rdev); +} + + +/* + * GEM ioctls. + */ +int radeon_gem_info_ioctl(struct drm_device *dev, void *data, + struct drm_file *filp) +{ + struct radeon_device *rdev = dev->dev_private; + struct drm_radeon_gem_info *args = data; + struct ttm_mem_type_manager *man; + + man = &rdev->mman.bdev.man[TTM_PL_VRAM]; + + args->vram_size = rdev->mc.real_vram_size; + args->vram_visible = (u64)man->size << PAGE_SHIFT; + if (rdev->stollen_vga_memory) + args->vram_visible -= radeon_bo_size(rdev->stollen_vga_memory); + args->vram_visible -= radeon_fbdev_total_size(rdev); + args->gart_size = rdev->mc.gtt_size - rdev->cp.ring_size - 4096 - + RADEON_IB_POOL_SIZE*64*1024; + return 0; +} + +int radeon_gem_pread_ioctl(struct drm_device *dev, void *data, + struct drm_file *filp) +{ + /* TODO: implement */ + DRM_ERROR("unimplemented %s\n", __func__); + return -ENOSYS; +} + +int radeon_gem_pwrite_ioctl(struct drm_device *dev, void *data, + struct drm_file *filp) +{ + /* TODO: implement */ + DRM_ERROR("unimplemented %s\n", __func__); + return -ENOSYS; +} + +int radeon_gem_create_ioctl(struct drm_device *dev, void *data, + struct drm_file *filp) +{ + struct radeon_device *rdev = dev->dev_private; + struct drm_radeon_gem_create *args = data; + struct drm_gem_object *gobj; + uint32_t handle; + int r; + + /* create a gem object to contain this object in */ + args->size = roundup(args->size, PAGE_SIZE); + r = radeon_gem_object_create(rdev, args->size, args->alignment, + args->initial_domain, false, + false, &gobj); + if (r) { + return r; + } + r = drm_gem_handle_create(filp, gobj, &handle); + /* drop reference from allocate - handle holds it now */ + drm_gem_object_unreference_unlocked(gobj); + if (r) { + return r; + } + args->handle = handle; + return 0; +} + +int radeon_gem_set_domain_ioctl(struct drm_device *dev, void *data, + struct drm_file *filp) +{ + /* transition the BO to a domain - + * just validate the BO into a certain domain */ + struct drm_radeon_gem_set_domain *args = data; + struct drm_gem_object *gobj; + struct radeon_bo *robj; + int r; + + /* for now if someone requests domain CPU - + * just make sure the buffer is finished with */ + + /* just do a BO wait for now */ + gobj = drm_gem_object_lookup(dev, filp, args->handle); + if (gobj == NULL) { + return -ENOENT; + } + robj = gem_to_radeon_bo(gobj); + + r = radeon_gem_set_domain(gobj, args->read_domains, args->write_domain); + + drm_gem_object_unreference_unlocked(gobj); + return r; +} + +int radeon_mode_dumb_mmap(struct drm_file *filp, + struct drm_device *dev, + uint32_t handle, uint64_t *offset_p) +{ + struct drm_gem_object *gobj; + struct radeon_bo *robj; + + gobj = drm_gem_object_lookup(dev, filp, handle); + if (gobj == NULL) { + return -ENOENT; + } + robj = gem_to_radeon_bo(gobj); + *offset_p = radeon_bo_mmap_offset(robj); + drm_gem_object_unreference_unlocked(gobj); + return 0; +} + +int radeon_gem_mmap_ioctl(struct drm_device *dev, void *data, + struct drm_file *filp) +{ + struct drm_radeon_gem_mmap *args = data; + + return radeon_mode_dumb_mmap(filp, dev, args->handle, &args->addr_ptr); +} + +int radeon_gem_busy_ioctl(struct drm_device *dev, void *data, + struct drm_file *filp) +{ + struct drm_radeon_gem_busy *args = data; + struct drm_gem_object *gobj; + struct radeon_bo *robj; + int r; + uint32_t cur_placement = 0; + + gobj = drm_gem_object_lookup(dev, filp, args->handle); + if (gobj == NULL) { + return -ENOENT; + } + robj = gem_to_radeon_bo(gobj); + r = radeon_bo_wait(robj, &cur_placement, true); + switch (cur_placement) { + case TTM_PL_VRAM: + args->domain = RADEON_GEM_DOMAIN_VRAM; + break; + case TTM_PL_TT: + args->domain = RADEON_GEM_DOMAIN_GTT; + break; + case TTM_PL_SYSTEM: + args->domain = RADEON_GEM_DOMAIN_CPU; + default: + break; + } + drm_gem_object_unreference_unlocked(gobj); + return r; +} + +int radeon_gem_wait_idle_ioctl(struct drm_device *dev, void *data, + struct drm_file *filp) +{ + struct drm_radeon_gem_wait_idle *args = data; + struct drm_gem_object *gobj; + struct radeon_bo *robj; + int r; + + gobj = drm_gem_object_lookup(dev, filp, args->handle); + if (gobj == NULL) { + return -ENOENT; + } + robj = gem_to_radeon_bo(gobj); + r = radeon_bo_wait(robj, NULL, false); + /* callback hw specific functions if any */ + if (robj->rdev->asic->ioctl_wait_idle) + robj->rdev->asic->ioctl_wait_idle(robj->rdev, robj); + drm_gem_object_unreference_unlocked(gobj); + return r; +} + +int radeon_gem_set_tiling_ioctl(struct drm_device *dev, void *data, + struct drm_file *filp) +{ + struct drm_radeon_gem_set_tiling *args = data; + struct drm_gem_object *gobj; + struct radeon_bo *robj; + int r = 0; + + DRM_DEBUG("%d \n", args->handle); + gobj = drm_gem_object_lookup(dev, filp, args->handle); + if (gobj == NULL) + return -ENOENT; + robj = gem_to_radeon_bo(gobj); + r = radeon_bo_set_tiling_flags(robj, args->tiling_flags, args->pitch); + drm_gem_object_unreference_unlocked(gobj); + return r; +} + +int radeon_gem_get_tiling_ioctl(struct drm_device *dev, void *data, + struct drm_file *filp) +{ + struct drm_radeon_gem_get_tiling *args = data; + struct drm_gem_object *gobj; + struct radeon_bo *rbo; + int r = 0; + + DRM_DEBUG("\n"); + gobj = drm_gem_object_lookup(dev, filp, args->handle); + if (gobj == NULL) + return -ENOENT; + rbo = gem_to_radeon_bo(gobj); + r = radeon_bo_reserve(rbo, false); + if (unlikely(r != 0)) + goto out; + radeon_bo_get_tiling_flags(rbo, &args->tiling_flags, &args->pitch); + radeon_bo_unreserve(rbo); +out: + drm_gem_object_unreference_unlocked(gobj); + return r; +} + +int radeon_mode_dumb_create(struct drm_file *file_priv, + struct drm_device *dev, + struct drm_mode_create_dumb *args) +{ + struct radeon_device *rdev = dev->dev_private; + struct drm_gem_object *gobj; + uint32_t handle; + int r; + + args->pitch = radeon_align_pitch(rdev, args->width, args->bpp, 0) * ((args->bpp + 1) / 8); + args->size = args->pitch * args->height; + args->size = ALIGN(args->size, PAGE_SIZE); + + r = radeon_gem_object_create(rdev, args->size, 0, + RADEON_GEM_DOMAIN_VRAM, + false, ttm_bo_type_device, + &gobj); + if (r) + return -ENOMEM; + + r = drm_gem_handle_create(file_priv, gobj, &handle); + /* drop reference from allocate - handle holds it now */ + drm_gem_object_unreference_unlocked(gobj); + if (r) { + return r; + } + args->handle = handle; + return 0; +} + +int radeon_mode_dumb_destroy(struct drm_file *file_priv, + struct drm_device *dev, + uint32_t handle) +{ + return drm_gem_handle_delete(file_priv, handle); +} diff --git a/drivers/gpu/drm/radeon/radeon_i2c.c b/drivers/gpu/drm/radeon/radeon_i2c.c new file mode 100644 index 00000000..c90425c4 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_i2c.c @@ -0,0 +1,1168 @@ +/* + * Copyright 2007-8 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + */ +#include "drmP.h" +#include "radeon_drm.h" +#include "radeon.h" +#include "atom.h" + +/** + * radeon_ddc_probe + * + */ +bool radeon_ddc_probe(struct radeon_connector *radeon_connector, bool requires_extended_probe) +{ + u8 out = 0x0; + u8 buf[8]; + int ret; + struct i2c_msg msgs[] = { + { + .addr = 0x50, + .flags = 0, + .len = 1, + .buf = &out, + }, + { + .addr = 0x50, + .flags = I2C_M_RD, + .len = 1, + .buf = buf, + } + }; + + /* Read 8 bytes from i2c for extended probe of EDID header */ + if (requires_extended_probe) + msgs[1].len = 8; + + /* on hw with routers, select right port */ + if (radeon_connector->router.ddc_valid) + radeon_router_select_ddc_port(radeon_connector); + + ret = i2c_transfer(&radeon_connector->ddc_bus->adapter, msgs, 2); + if (ret != 2) + /* Couldn't find an accessible DDC on this connector */ + return false; + if (requires_extended_probe) { + /* Probe also for valid EDID header + * EDID header starts with: + * 0x00,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0x00. + * Only the first 6 bytes must be valid as + * drm_edid_block_valid() can fix the last 2 bytes */ + if (drm_edid_header_is_valid(buf) < 6) { + /* Couldn't find an accessible EDID on this + * connector */ + return false; + } + } + return true; +} + +/* bit banging i2c */ + +static void radeon_i2c_do_lock(struct radeon_i2c_chan *i2c, int lock_state) +{ + struct radeon_device *rdev = i2c->dev->dev_private; + struct radeon_i2c_bus_rec *rec = &i2c->rec; + uint32_t temp; + + /* RV410 appears to have a bug where the hw i2c in reset + * holds the i2c port in a bad state - switch hw i2c away before + * doing DDC - do this for all r200s/r300s/r400s for safety sake + */ + if (rec->hw_capable) { + if ((rdev->family >= CHIP_R200) && !ASIC_IS_AVIVO(rdev)) { + u32 reg; + + if (rdev->family >= CHIP_RV350) + reg = RADEON_GPIO_MONID; + else if ((rdev->family == CHIP_R300) || + (rdev->family == CHIP_R350)) + reg = RADEON_GPIO_DVI_DDC; + else + reg = RADEON_GPIO_CRT2_DDC; + + mutex_lock(&rdev->dc_hw_i2c_mutex); + if (rec->a_clk_reg == reg) { + WREG32(RADEON_DVI_I2C_CNTL_0, (RADEON_I2C_SOFT_RST | + R200_DVI_I2C_PIN_SEL(R200_SEL_DDC1))); + } else { + WREG32(RADEON_DVI_I2C_CNTL_0, (RADEON_I2C_SOFT_RST | + R200_DVI_I2C_PIN_SEL(R200_SEL_DDC3))); + } + mutex_unlock(&rdev->dc_hw_i2c_mutex); + } + } + + /* switch the pads to ddc mode */ + if (ASIC_IS_DCE3(rdev) && rec->hw_capable) { + temp = RREG32(rec->mask_clk_reg); + temp &= ~(1 << 16); + WREG32(rec->mask_clk_reg, temp); + } + + /* clear the output pin values */ + temp = RREG32(rec->a_clk_reg) & ~rec->a_clk_mask; + WREG32(rec->a_clk_reg, temp); + + temp = RREG32(rec->a_data_reg) & ~rec->a_data_mask; + WREG32(rec->a_data_reg, temp); + + /* set the pins to input */ + temp = RREG32(rec->en_clk_reg) & ~rec->en_clk_mask; + WREG32(rec->en_clk_reg, temp); + + temp = RREG32(rec->en_data_reg) & ~rec->en_data_mask; + WREG32(rec->en_data_reg, temp); + + /* mask the gpio pins for software use */ + temp = RREG32(rec->mask_clk_reg); + if (lock_state) + temp |= rec->mask_clk_mask; + else + temp &= ~rec->mask_clk_mask; + WREG32(rec->mask_clk_reg, temp); + temp = RREG32(rec->mask_clk_reg); + + temp = RREG32(rec->mask_data_reg); + if (lock_state) + temp |= rec->mask_data_mask; + else + temp &= ~rec->mask_data_mask; + WREG32(rec->mask_data_reg, temp); + temp = RREG32(rec->mask_data_reg); +} + +static int get_clock(void *i2c_priv) +{ + struct radeon_i2c_chan *i2c = i2c_priv; + struct radeon_device *rdev = i2c->dev->dev_private; + struct radeon_i2c_bus_rec *rec = &i2c->rec; + uint32_t val; + + /* read the value off the pin */ + val = RREG32(rec->y_clk_reg); + val &= rec->y_clk_mask; + + return (val != 0); +} + + +static int get_data(void *i2c_priv) +{ + struct radeon_i2c_chan *i2c = i2c_priv; + struct radeon_device *rdev = i2c->dev->dev_private; + struct radeon_i2c_bus_rec *rec = &i2c->rec; + uint32_t val; + + /* read the value off the pin */ + val = RREG32(rec->y_data_reg); + val &= rec->y_data_mask; + + return (val != 0); +} + +static void set_clock(void *i2c_priv, int clock) +{ + struct radeon_i2c_chan *i2c = i2c_priv; + struct radeon_device *rdev = i2c->dev->dev_private; + struct radeon_i2c_bus_rec *rec = &i2c->rec; + uint32_t val; + + /* set pin direction */ + val = RREG32(rec->en_clk_reg) & ~rec->en_clk_mask; + val |= clock ? 0 : rec->en_clk_mask; + WREG32(rec->en_clk_reg, val); +} + +static void set_data(void *i2c_priv, int data) +{ + struct radeon_i2c_chan *i2c = i2c_priv; + struct radeon_device *rdev = i2c->dev->dev_private; + struct radeon_i2c_bus_rec *rec = &i2c->rec; + uint32_t val; + + /* set pin direction */ + val = RREG32(rec->en_data_reg) & ~rec->en_data_mask; + val |= data ? 0 : rec->en_data_mask; + WREG32(rec->en_data_reg, val); +} + +static int pre_xfer(struct i2c_adapter *i2c_adap) +{ + struct radeon_i2c_chan *i2c = i2c_get_adapdata(i2c_adap); + + radeon_i2c_do_lock(i2c, 1); + + return 0; +} + +static void post_xfer(struct i2c_adapter *i2c_adap) +{ + struct radeon_i2c_chan *i2c = i2c_get_adapdata(i2c_adap); + + radeon_i2c_do_lock(i2c, 0); +} + +/* hw i2c */ + +static u32 radeon_get_i2c_prescale(struct radeon_device *rdev) +{ + u32 sclk = rdev->pm.current_sclk; + u32 prescale = 0; + u32 nm; + u8 n, m, loop; + int i2c_clock; + + switch (rdev->family) { + case CHIP_R100: + case CHIP_RV100: + case CHIP_RS100: + case CHIP_RV200: + case CHIP_RS200: + case CHIP_R200: + case CHIP_RV250: + case CHIP_RS300: + case CHIP_RV280: + case CHIP_R300: + case CHIP_R350: + case CHIP_RV350: + i2c_clock = 60; + nm = (sclk * 10) / (i2c_clock * 4); + for (loop = 1; loop < 255; loop++) { + if ((nm / loop) < loop) + break; + } + n = loop - 1; + m = loop - 2; + prescale = m | (n << 8); + break; + case CHIP_RV380: + case CHIP_RS400: + case CHIP_RS480: + case CHIP_R420: + case CHIP_R423: + case CHIP_RV410: + prescale = (((sclk * 10)/(4 * 128 * 100) + 1) << 8) + 128; + break; + case CHIP_RS600: + case CHIP_RS690: + case CHIP_RS740: + /* todo */ + break; + case CHIP_RV515: + case CHIP_R520: + case CHIP_RV530: + case CHIP_RV560: + case CHIP_RV570: + case CHIP_R580: + i2c_clock = 50; + if (rdev->family == CHIP_R520) + prescale = (127 << 8) + ((sclk * 10) / (4 * 127 * i2c_clock)); + else + prescale = (((sclk * 10)/(4 * 128 * 100) + 1) << 8) + 128; + break; + case CHIP_R600: + case CHIP_RV610: + case CHIP_RV630: + case CHIP_RV670: + /* todo */ + break; + case CHIP_RV620: + case CHIP_RV635: + case CHIP_RS780: + case CHIP_RS880: + case CHIP_RV770: + case CHIP_RV730: + case CHIP_RV710: + case CHIP_RV740: + /* todo */ + break; + case CHIP_CEDAR: + case CHIP_REDWOOD: + case CHIP_JUNIPER: + case CHIP_CYPRESS: + case CHIP_HEMLOCK: + /* todo */ + break; + default: + DRM_ERROR("i2c: unhandled radeon chip\n"); + break; + } + return prescale; +} + + +/* hw i2c engine for r1xx-4xx hardware + * hw can buffer up to 15 bytes + */ +static int r100_hw_i2c_xfer(struct i2c_adapter *i2c_adap, + struct i2c_msg *msgs, int num) +{ + struct radeon_i2c_chan *i2c = i2c_get_adapdata(i2c_adap); + struct radeon_device *rdev = i2c->dev->dev_private; + struct radeon_i2c_bus_rec *rec = &i2c->rec; + struct i2c_msg *p; + int i, j, k, ret = num; + u32 prescale; + u32 i2c_cntl_0, i2c_cntl_1, i2c_data; + u32 tmp, reg; + + mutex_lock(&rdev->dc_hw_i2c_mutex); + /* take the pm lock since we need a constant sclk */ + mutex_lock(&rdev->pm.mutex); + + prescale = radeon_get_i2c_prescale(rdev); + + reg = ((prescale << RADEON_I2C_PRESCALE_SHIFT) | + RADEON_I2C_DRIVE_EN | + RADEON_I2C_START | + RADEON_I2C_STOP | + RADEON_I2C_GO); + + if (rdev->is_atom_bios) { + tmp = RREG32(RADEON_BIOS_6_SCRATCH); + WREG32(RADEON_BIOS_6_SCRATCH, tmp | ATOM_S6_HW_I2C_BUSY_STATE); + } + + if (rec->mm_i2c) { + i2c_cntl_0 = RADEON_I2C_CNTL_0; + i2c_cntl_1 = RADEON_I2C_CNTL_1; + i2c_data = RADEON_I2C_DATA; + } else { + i2c_cntl_0 = RADEON_DVI_I2C_CNTL_0; + i2c_cntl_1 = RADEON_DVI_I2C_CNTL_1; + i2c_data = RADEON_DVI_I2C_DATA; + + switch (rdev->family) { + case CHIP_R100: + case CHIP_RV100: + case CHIP_RS100: + case CHIP_RV200: + case CHIP_RS200: + case CHIP_RS300: + switch (rec->mask_clk_reg) { + case RADEON_GPIO_DVI_DDC: + /* no gpio select bit */ + break; + default: + DRM_ERROR("gpio not supported with hw i2c\n"); + ret = -EINVAL; + goto done; + } + break; + case CHIP_R200: + /* only bit 4 on r200 */ + switch (rec->mask_clk_reg) { + case RADEON_GPIO_DVI_DDC: + reg |= R200_DVI_I2C_PIN_SEL(R200_SEL_DDC1); + break; + case RADEON_GPIO_MONID: + reg |= R200_DVI_I2C_PIN_SEL(R200_SEL_DDC3); + break; + default: + DRM_ERROR("gpio not supported with hw i2c\n"); + ret = -EINVAL; + goto done; + } + break; + case CHIP_RV250: + case CHIP_RV280: + /* bits 3 and 4 */ + switch (rec->mask_clk_reg) { + case RADEON_GPIO_DVI_DDC: + reg |= R200_DVI_I2C_PIN_SEL(R200_SEL_DDC1); + break; + case RADEON_GPIO_VGA_DDC: + reg |= R200_DVI_I2C_PIN_SEL(R200_SEL_DDC2); + break; + case RADEON_GPIO_CRT2_DDC: + reg |= R200_DVI_I2C_PIN_SEL(R200_SEL_DDC3); + break; + default: + DRM_ERROR("gpio not supported with hw i2c\n"); + ret = -EINVAL; + goto done; + } + break; + case CHIP_R300: + case CHIP_R350: + /* only bit 4 on r300/r350 */ + switch (rec->mask_clk_reg) { + case RADEON_GPIO_VGA_DDC: + reg |= R200_DVI_I2C_PIN_SEL(R200_SEL_DDC1); + break; + case RADEON_GPIO_DVI_DDC: + reg |= R200_DVI_I2C_PIN_SEL(R200_SEL_DDC3); + break; + default: + DRM_ERROR("gpio not supported with hw i2c\n"); + ret = -EINVAL; + goto done; + } + break; + case CHIP_RV350: + case CHIP_RV380: + case CHIP_R420: + case CHIP_R423: + case CHIP_RV410: + case CHIP_RS400: + case CHIP_RS480: + /* bits 3 and 4 */ + switch (rec->mask_clk_reg) { + case RADEON_GPIO_VGA_DDC: + reg |= R200_DVI_I2C_PIN_SEL(R200_SEL_DDC1); + break; + case RADEON_GPIO_DVI_DDC: + reg |= R200_DVI_I2C_PIN_SEL(R200_SEL_DDC2); + break; + case RADEON_GPIO_MONID: + reg |= R200_DVI_I2C_PIN_SEL(R200_SEL_DDC3); + break; + default: + DRM_ERROR("gpio not supported with hw i2c\n"); + ret = -EINVAL; + goto done; + } + break; + default: + DRM_ERROR("unsupported asic\n"); + ret = -EINVAL; + goto done; + break; + } + } + + /* check for bus probe */ + p = &msgs[0]; + if ((num == 1) && (p->len == 0)) { + WREG32(i2c_cntl_0, (RADEON_I2C_DONE | + RADEON_I2C_NACK | + RADEON_I2C_HALT | + RADEON_I2C_SOFT_RST)); + WREG32(i2c_data, (p->addr << 1) & 0xff); + WREG32(i2c_data, 0); + WREG32(i2c_cntl_1, ((1 << RADEON_I2C_DATA_COUNT_SHIFT) | + (1 << RADEON_I2C_ADDR_COUNT_SHIFT) | + RADEON_I2C_EN | + (48 << RADEON_I2C_TIME_LIMIT_SHIFT))); + WREG32(i2c_cntl_0, reg); + for (k = 0; k < 32; k++) { + udelay(10); + tmp = RREG32(i2c_cntl_0); + if (tmp & RADEON_I2C_GO) + continue; + tmp = RREG32(i2c_cntl_0); + if (tmp & RADEON_I2C_DONE) + break; + else { + DRM_DEBUG("i2c write error 0x%08x\n", tmp); + WREG32(i2c_cntl_0, tmp | RADEON_I2C_ABORT); + ret = -EIO; + goto done; + } + } + goto done; + } + + for (i = 0; i < num; i++) { + p = &msgs[i]; + for (j = 0; j < p->len; j++) { + if (p->flags & I2C_M_RD) { + WREG32(i2c_cntl_0, (RADEON_I2C_DONE | + RADEON_I2C_NACK | + RADEON_I2C_HALT | + RADEON_I2C_SOFT_RST)); + WREG32(i2c_data, ((p->addr << 1) & 0xff) | 0x1); + WREG32(i2c_cntl_1, ((1 << RADEON_I2C_DATA_COUNT_SHIFT) | + (1 << RADEON_I2C_ADDR_COUNT_SHIFT) | + RADEON_I2C_EN | + (48 << RADEON_I2C_TIME_LIMIT_SHIFT))); + WREG32(i2c_cntl_0, reg | RADEON_I2C_RECEIVE); + for (k = 0; k < 32; k++) { + udelay(10); + tmp = RREG32(i2c_cntl_0); + if (tmp & RADEON_I2C_GO) + continue; + tmp = RREG32(i2c_cntl_0); + if (tmp & RADEON_I2C_DONE) + break; + else { + DRM_DEBUG("i2c read error 0x%08x\n", tmp); + WREG32(i2c_cntl_0, tmp | RADEON_I2C_ABORT); + ret = -EIO; + goto done; + } + } + p->buf[j] = RREG32(i2c_data) & 0xff; + } else { + WREG32(i2c_cntl_0, (RADEON_I2C_DONE | + RADEON_I2C_NACK | + RADEON_I2C_HALT | + RADEON_I2C_SOFT_RST)); + WREG32(i2c_data, (p->addr << 1) & 0xff); + WREG32(i2c_data, p->buf[j]); + WREG32(i2c_cntl_1, ((1 << RADEON_I2C_DATA_COUNT_SHIFT) | + (1 << RADEON_I2C_ADDR_COUNT_SHIFT) | + RADEON_I2C_EN | + (48 << RADEON_I2C_TIME_LIMIT_SHIFT))); + WREG32(i2c_cntl_0, reg); + for (k = 0; k < 32; k++) { + udelay(10); + tmp = RREG32(i2c_cntl_0); + if (tmp & RADEON_I2C_GO) + continue; + tmp = RREG32(i2c_cntl_0); + if (tmp & RADEON_I2C_DONE) + break; + else { + DRM_DEBUG("i2c write error 0x%08x\n", tmp); + WREG32(i2c_cntl_0, tmp | RADEON_I2C_ABORT); + ret = -EIO; + goto done; + } + } + } + } + } + +done: + WREG32(i2c_cntl_0, 0); + WREG32(i2c_cntl_1, 0); + WREG32(i2c_cntl_0, (RADEON_I2C_DONE | + RADEON_I2C_NACK | + RADEON_I2C_HALT | + RADEON_I2C_SOFT_RST)); + + if (rdev->is_atom_bios) { + tmp = RREG32(RADEON_BIOS_6_SCRATCH); + tmp &= ~ATOM_S6_HW_I2C_BUSY_STATE; + WREG32(RADEON_BIOS_6_SCRATCH, tmp); + } + + mutex_unlock(&rdev->pm.mutex); + mutex_unlock(&rdev->dc_hw_i2c_mutex); + + return ret; +} + +/* hw i2c engine for r5xx hardware + * hw can buffer up to 15 bytes + */ +static int r500_hw_i2c_xfer(struct i2c_adapter *i2c_adap, + struct i2c_msg *msgs, int num) +{ + struct radeon_i2c_chan *i2c = i2c_get_adapdata(i2c_adap); + struct radeon_device *rdev = i2c->dev->dev_private; + struct radeon_i2c_bus_rec *rec = &i2c->rec; + struct i2c_msg *p; + int i, j, remaining, current_count, buffer_offset, ret = num; + u32 prescale; + u32 tmp, reg; + u32 saved1, saved2; + + mutex_lock(&rdev->dc_hw_i2c_mutex); + /* take the pm lock since we need a constant sclk */ + mutex_lock(&rdev->pm.mutex); + + prescale = radeon_get_i2c_prescale(rdev); + + /* clear gpio mask bits */ + tmp = RREG32(rec->mask_clk_reg); + tmp &= ~rec->mask_clk_mask; + WREG32(rec->mask_clk_reg, tmp); + tmp = RREG32(rec->mask_clk_reg); + + tmp = RREG32(rec->mask_data_reg); + tmp &= ~rec->mask_data_mask; + WREG32(rec->mask_data_reg, tmp); + tmp = RREG32(rec->mask_data_reg); + + /* clear pin values */ + tmp = RREG32(rec->a_clk_reg); + tmp &= ~rec->a_clk_mask; + WREG32(rec->a_clk_reg, tmp); + tmp = RREG32(rec->a_clk_reg); + + tmp = RREG32(rec->a_data_reg); + tmp &= ~rec->a_data_mask; + WREG32(rec->a_data_reg, tmp); + tmp = RREG32(rec->a_data_reg); + + /* set the pins to input */ + tmp = RREG32(rec->en_clk_reg); + tmp &= ~rec->en_clk_mask; + WREG32(rec->en_clk_reg, tmp); + tmp = RREG32(rec->en_clk_reg); + + tmp = RREG32(rec->en_data_reg); + tmp &= ~rec->en_data_mask; + WREG32(rec->en_data_reg, tmp); + tmp = RREG32(rec->en_data_reg); + + /* */ + tmp = RREG32(RADEON_BIOS_6_SCRATCH); + WREG32(RADEON_BIOS_6_SCRATCH, tmp | ATOM_S6_HW_I2C_BUSY_STATE); + saved1 = RREG32(AVIVO_DC_I2C_CONTROL1); + saved2 = RREG32(0x494); + WREG32(0x494, saved2 | 0x1); + + WREG32(AVIVO_DC_I2C_ARBITRATION, AVIVO_DC_I2C_SW_WANTS_TO_USE_I2C); + for (i = 0; i < 50; i++) { + udelay(1); + if (RREG32(AVIVO_DC_I2C_ARBITRATION) & AVIVO_DC_I2C_SW_CAN_USE_I2C) + break; + } + if (i == 50) { + DRM_ERROR("failed to get i2c bus\n"); + ret = -EBUSY; + goto done; + } + + reg = AVIVO_DC_I2C_START | AVIVO_DC_I2C_STOP | AVIVO_DC_I2C_EN; + switch (rec->mask_clk_reg) { + case AVIVO_DC_GPIO_DDC1_MASK: + reg |= AVIVO_DC_I2C_PIN_SELECT(AVIVO_SEL_DDC1); + break; + case AVIVO_DC_GPIO_DDC2_MASK: + reg |= AVIVO_DC_I2C_PIN_SELECT(AVIVO_SEL_DDC2); + break; + case AVIVO_DC_GPIO_DDC3_MASK: + reg |= AVIVO_DC_I2C_PIN_SELECT(AVIVO_SEL_DDC3); + break; + default: + DRM_ERROR("gpio not supported with hw i2c\n"); + ret = -EINVAL; + goto done; + } + + /* check for bus probe */ + p = &msgs[0]; + if ((num == 1) && (p->len == 0)) { + WREG32(AVIVO_DC_I2C_STATUS1, (AVIVO_DC_I2C_DONE | + AVIVO_DC_I2C_NACK | + AVIVO_DC_I2C_HALT)); + WREG32(AVIVO_DC_I2C_RESET, AVIVO_DC_I2C_SOFT_RESET); + udelay(1); + WREG32(AVIVO_DC_I2C_RESET, 0); + + WREG32(AVIVO_DC_I2C_DATA, (p->addr << 1) & 0xff); + WREG32(AVIVO_DC_I2C_DATA, 0); + + WREG32(AVIVO_DC_I2C_CONTROL3, AVIVO_DC_I2C_TIME_LIMIT(48)); + WREG32(AVIVO_DC_I2C_CONTROL2, (AVIVO_DC_I2C_ADDR_COUNT(1) | + AVIVO_DC_I2C_DATA_COUNT(1) | + (prescale << 16))); + WREG32(AVIVO_DC_I2C_CONTROL1, reg); + WREG32(AVIVO_DC_I2C_STATUS1, AVIVO_DC_I2C_GO); + for (j = 0; j < 200; j++) { + udelay(50); + tmp = RREG32(AVIVO_DC_I2C_STATUS1); + if (tmp & AVIVO_DC_I2C_GO) + continue; + tmp = RREG32(AVIVO_DC_I2C_STATUS1); + if (tmp & AVIVO_DC_I2C_DONE) + break; + else { + DRM_DEBUG("i2c write error 0x%08x\n", tmp); + WREG32(AVIVO_DC_I2C_RESET, AVIVO_DC_I2C_ABORT); + ret = -EIO; + goto done; + } + } + goto done; + } + + for (i = 0; i < num; i++) { + p = &msgs[i]; + remaining = p->len; + buffer_offset = 0; + if (p->flags & I2C_M_RD) { + while (remaining) { + if (remaining > 15) + current_count = 15; + else + current_count = remaining; + WREG32(AVIVO_DC_I2C_STATUS1, (AVIVO_DC_I2C_DONE | + AVIVO_DC_I2C_NACK | + AVIVO_DC_I2C_HALT)); + WREG32(AVIVO_DC_I2C_RESET, AVIVO_DC_I2C_SOFT_RESET); + udelay(1); + WREG32(AVIVO_DC_I2C_RESET, 0); + + WREG32(AVIVO_DC_I2C_DATA, ((p->addr << 1) & 0xff) | 0x1); + WREG32(AVIVO_DC_I2C_CONTROL3, AVIVO_DC_I2C_TIME_LIMIT(48)); + WREG32(AVIVO_DC_I2C_CONTROL2, (AVIVO_DC_I2C_ADDR_COUNT(1) | + AVIVO_DC_I2C_DATA_COUNT(current_count) | + (prescale << 16))); + WREG32(AVIVO_DC_I2C_CONTROL1, reg | AVIVO_DC_I2C_RECEIVE); + WREG32(AVIVO_DC_I2C_STATUS1, AVIVO_DC_I2C_GO); + for (j = 0; j < 200; j++) { + udelay(50); + tmp = RREG32(AVIVO_DC_I2C_STATUS1); + if (tmp & AVIVO_DC_I2C_GO) + continue; + tmp = RREG32(AVIVO_DC_I2C_STATUS1); + if (tmp & AVIVO_DC_I2C_DONE) + break; + else { + DRM_DEBUG("i2c read error 0x%08x\n", tmp); + WREG32(AVIVO_DC_I2C_RESET, AVIVO_DC_I2C_ABORT); + ret = -EIO; + goto done; + } + } + for (j = 0; j < current_count; j++) + p->buf[buffer_offset + j] = RREG32(AVIVO_DC_I2C_DATA) & 0xff; + remaining -= current_count; + buffer_offset += current_count; + } + } else { + while (remaining) { + if (remaining > 15) + current_count = 15; + else + current_count = remaining; + WREG32(AVIVO_DC_I2C_STATUS1, (AVIVO_DC_I2C_DONE | + AVIVO_DC_I2C_NACK | + AVIVO_DC_I2C_HALT)); + WREG32(AVIVO_DC_I2C_RESET, AVIVO_DC_I2C_SOFT_RESET); + udelay(1); + WREG32(AVIVO_DC_I2C_RESET, 0); + + WREG32(AVIVO_DC_I2C_DATA, (p->addr << 1) & 0xff); + for (j = 0; j < current_count; j++) + WREG32(AVIVO_DC_I2C_DATA, p->buf[buffer_offset + j]); + + WREG32(AVIVO_DC_I2C_CONTROL3, AVIVO_DC_I2C_TIME_LIMIT(48)); + WREG32(AVIVO_DC_I2C_CONTROL2, (AVIVO_DC_I2C_ADDR_COUNT(1) | + AVIVO_DC_I2C_DATA_COUNT(current_count) | + (prescale << 16))); + WREG32(AVIVO_DC_I2C_CONTROL1, reg); + WREG32(AVIVO_DC_I2C_STATUS1, AVIVO_DC_I2C_GO); + for (j = 0; j < 200; j++) { + udelay(50); + tmp = RREG32(AVIVO_DC_I2C_STATUS1); + if (tmp & AVIVO_DC_I2C_GO) + continue; + tmp = RREG32(AVIVO_DC_I2C_STATUS1); + if (tmp & AVIVO_DC_I2C_DONE) + break; + else { + DRM_DEBUG("i2c write error 0x%08x\n", tmp); + WREG32(AVIVO_DC_I2C_RESET, AVIVO_DC_I2C_ABORT); + ret = -EIO; + goto done; + } + } + remaining -= current_count; + buffer_offset += current_count; + } + } + } + +done: + WREG32(AVIVO_DC_I2C_STATUS1, (AVIVO_DC_I2C_DONE | + AVIVO_DC_I2C_NACK | + AVIVO_DC_I2C_HALT)); + WREG32(AVIVO_DC_I2C_RESET, AVIVO_DC_I2C_SOFT_RESET); + udelay(1); + WREG32(AVIVO_DC_I2C_RESET, 0); + + WREG32(AVIVO_DC_I2C_ARBITRATION, AVIVO_DC_I2C_SW_DONE_USING_I2C); + WREG32(AVIVO_DC_I2C_CONTROL1, saved1); + WREG32(0x494, saved2); + tmp = RREG32(RADEON_BIOS_6_SCRATCH); + tmp &= ~ATOM_S6_HW_I2C_BUSY_STATE; + WREG32(RADEON_BIOS_6_SCRATCH, tmp); + + mutex_unlock(&rdev->pm.mutex); + mutex_unlock(&rdev->dc_hw_i2c_mutex); + + return ret; +} + +static int radeon_hw_i2c_xfer(struct i2c_adapter *i2c_adap, + struct i2c_msg *msgs, int num) +{ + struct radeon_i2c_chan *i2c = i2c_get_adapdata(i2c_adap); + struct radeon_device *rdev = i2c->dev->dev_private; + struct radeon_i2c_bus_rec *rec = &i2c->rec; + int ret = 0; + + switch (rdev->family) { + case CHIP_R100: + case CHIP_RV100: + case CHIP_RS100: + case CHIP_RV200: + case CHIP_RS200: + case CHIP_R200: + case CHIP_RV250: + case CHIP_RS300: + case CHIP_RV280: + case CHIP_R300: + case CHIP_R350: + case CHIP_RV350: + case CHIP_RV380: + case CHIP_R420: + case CHIP_R423: + case CHIP_RV410: + case CHIP_RS400: + case CHIP_RS480: + ret = r100_hw_i2c_xfer(i2c_adap, msgs, num); + break; + case CHIP_RS600: + case CHIP_RS690: + case CHIP_RS740: + /* XXX fill in hw i2c implementation */ + break; + case CHIP_RV515: + case CHIP_R520: + case CHIP_RV530: + case CHIP_RV560: + case CHIP_RV570: + case CHIP_R580: + if (rec->mm_i2c) + ret = r100_hw_i2c_xfer(i2c_adap, msgs, num); + else + ret = r500_hw_i2c_xfer(i2c_adap, msgs, num); + break; + case CHIP_R600: + case CHIP_RV610: + case CHIP_RV630: + case CHIP_RV670: + /* XXX fill in hw i2c implementation */ + break; + case CHIP_RV620: + case CHIP_RV635: + case CHIP_RS780: + case CHIP_RS880: + case CHIP_RV770: + case CHIP_RV730: + case CHIP_RV710: + case CHIP_RV740: + /* XXX fill in hw i2c implementation */ + break; + case CHIP_CEDAR: + case CHIP_REDWOOD: + case CHIP_JUNIPER: + case CHIP_CYPRESS: + case CHIP_HEMLOCK: + /* XXX fill in hw i2c implementation */ + break; + default: + DRM_ERROR("i2c: unhandled radeon chip\n"); + ret = -EIO; + break; + } + + return ret; +} + +static u32 radeon_hw_i2c_func(struct i2c_adapter *adap) +{ + return I2C_FUNC_I2C | I2C_FUNC_SMBUS_EMUL; +} + +static const struct i2c_algorithm radeon_i2c_algo = { + .master_xfer = radeon_hw_i2c_xfer, + .functionality = radeon_hw_i2c_func, +}; + +struct radeon_i2c_chan *radeon_i2c_create(struct drm_device *dev, + struct radeon_i2c_bus_rec *rec, + const char *name) +{ + struct radeon_device *rdev = dev->dev_private; + struct radeon_i2c_chan *i2c; + int ret; + + /* don't add the mm_i2c bus unless hw_i2c is enabled */ + if (rec->mm_i2c && (radeon_hw_i2c == 0)) + return NULL; + + i2c = kzalloc(sizeof(struct radeon_i2c_chan), GFP_KERNEL); + if (i2c == NULL) + return NULL; + + i2c->rec = *rec; + i2c->adapter.owner = THIS_MODULE; + i2c->adapter.class = I2C_CLASS_DDC; + i2c->dev = dev; + i2c_set_adapdata(&i2c->adapter, i2c); + if (rec->mm_i2c || + (rec->hw_capable && + radeon_hw_i2c && + ((rdev->family <= CHIP_RS480) || + ((rdev->family >= CHIP_RV515) && (rdev->family <= CHIP_R580))))) { + /* set the radeon hw i2c adapter */ + snprintf(i2c->adapter.name, sizeof(i2c->adapter.name), + "Radeon i2c hw bus %s", name); + i2c->adapter.algo = &radeon_i2c_algo; + ret = i2c_add_adapter(&i2c->adapter); + if (ret) { + DRM_ERROR("Failed to register hw i2c %s\n", name); + goto out_free; + } + } else { + /* set the radeon bit adapter */ + snprintf(i2c->adapter.name, sizeof(i2c->adapter.name), + "Radeon i2c bit bus %s", name); + i2c->adapter.algo_data = &i2c->algo.bit; + i2c->algo.bit.pre_xfer = pre_xfer; + i2c->algo.bit.post_xfer = post_xfer; + i2c->algo.bit.setsda = set_data; + i2c->algo.bit.setscl = set_clock; + i2c->algo.bit.getsda = get_data; + i2c->algo.bit.getscl = get_clock; + i2c->algo.bit.udelay = 20; + /* vesa says 2.2 ms is enough, 1 jiffy doesn't seem to always + * make this, 2 jiffies is a lot more reliable */ + i2c->algo.bit.timeout = 2; + i2c->algo.bit.data = i2c; + ret = i2c_bit_add_bus(&i2c->adapter); + if (ret) { + DRM_ERROR("Failed to register bit i2c %s\n", name); + goto out_free; + } + } + + return i2c; +out_free: + kfree(i2c); + return NULL; + +} + +struct radeon_i2c_chan *radeon_i2c_create_dp(struct drm_device *dev, + struct radeon_i2c_bus_rec *rec, + const char *name) +{ + struct radeon_i2c_chan *i2c; + int ret; + + i2c = kzalloc(sizeof(struct radeon_i2c_chan), GFP_KERNEL); + if (i2c == NULL) + return NULL; + + i2c->rec = *rec; + i2c->adapter.owner = THIS_MODULE; + i2c->adapter.class = I2C_CLASS_DDC; + i2c->dev = dev; + snprintf(i2c->adapter.name, sizeof(i2c->adapter.name), + "Radeon aux bus %s", name); + i2c_set_adapdata(&i2c->adapter, i2c); + i2c->adapter.algo_data = &i2c->algo.dp; + i2c->algo.dp.aux_ch = radeon_dp_i2c_aux_ch; + i2c->algo.dp.address = 0; + ret = i2c_dp_aux_add_bus(&i2c->adapter); + if (ret) { + DRM_INFO("Failed to register i2c %s\n", name); + goto out_free; + } + + return i2c; +out_free: + kfree(i2c); + return NULL; + +} + +void radeon_i2c_destroy(struct radeon_i2c_chan *i2c) +{ + if (!i2c) + return; + i2c_del_adapter(&i2c->adapter); + kfree(i2c); +} + +/* Add the default buses */ +void radeon_i2c_init(struct radeon_device *rdev) +{ + if (rdev->is_atom_bios) + radeon_atombios_i2c_init(rdev); + else + radeon_combios_i2c_init(rdev); +} + +/* remove all the buses */ +void radeon_i2c_fini(struct radeon_device *rdev) +{ + int i; + + for (i = 0; i < RADEON_MAX_I2C_BUS; i++) { + if (rdev->i2c_bus[i]) { + radeon_i2c_destroy(rdev->i2c_bus[i]); + rdev->i2c_bus[i] = NULL; + } + } +} + +/* Add additional buses */ +void radeon_i2c_add(struct radeon_device *rdev, + struct radeon_i2c_bus_rec *rec, + const char *name) +{ + struct drm_device *dev = rdev->ddev; + int i; + + for (i = 0; i < RADEON_MAX_I2C_BUS; i++) { + if (!rdev->i2c_bus[i]) { + rdev->i2c_bus[i] = radeon_i2c_create(dev, rec, name); + return; + } + } +} + +/* looks up bus based on id */ +struct radeon_i2c_chan *radeon_i2c_lookup(struct radeon_device *rdev, + struct radeon_i2c_bus_rec *i2c_bus) +{ + int i; + + for (i = 0; i < RADEON_MAX_I2C_BUS; i++) { + if (rdev->i2c_bus[i] && + (rdev->i2c_bus[i]->rec.i2c_id == i2c_bus->i2c_id)) { + return rdev->i2c_bus[i]; + } + } + return NULL; +} + +struct drm_encoder *radeon_best_encoder(struct drm_connector *connector) +{ + return NULL; +} + +void radeon_i2c_get_byte(struct radeon_i2c_chan *i2c_bus, + u8 slave_addr, + u8 addr, + u8 *val) +{ + u8 out_buf[2]; + u8 in_buf[2]; + struct i2c_msg msgs[] = { + { + .addr = slave_addr, + .flags = 0, + .len = 1, + .buf = out_buf, + }, + { + .addr = slave_addr, + .flags = I2C_M_RD, + .len = 1, + .buf = in_buf, + } + }; + + out_buf[0] = addr; + out_buf[1] = 0; + + if (i2c_transfer(&i2c_bus->adapter, msgs, 2) == 2) { + *val = in_buf[0]; + DRM_DEBUG("val = 0x%02x\n", *val); + } else { + DRM_DEBUG("i2c 0x%02x 0x%02x read failed\n", + addr, *val); + } +} + +void radeon_i2c_put_byte(struct radeon_i2c_chan *i2c_bus, + u8 slave_addr, + u8 addr, + u8 val) +{ + uint8_t out_buf[2]; + struct i2c_msg msg = { + .addr = slave_addr, + .flags = 0, + .len = 2, + .buf = out_buf, + }; + + out_buf[0] = addr; + out_buf[1] = val; + + if (i2c_transfer(&i2c_bus->adapter, &msg, 1) != 1) + DRM_DEBUG("i2c 0x%02x 0x%02x write failed\n", + addr, val); +} + +/* ddc router switching */ +void radeon_router_select_ddc_port(struct radeon_connector *radeon_connector) +{ + u8 val; + + if (!radeon_connector->router.ddc_valid) + return; + + if (!radeon_connector->router_bus) + return; + + radeon_i2c_get_byte(radeon_connector->router_bus, + radeon_connector->router.i2c_addr, + 0x3, &val); + val &= ~radeon_connector->router.ddc_mux_control_pin; + radeon_i2c_put_byte(radeon_connector->router_bus, + radeon_connector->router.i2c_addr, + 0x3, val); + radeon_i2c_get_byte(radeon_connector->router_bus, + radeon_connector->router.i2c_addr, + 0x1, &val); + val &= ~radeon_connector->router.ddc_mux_control_pin; + val |= radeon_connector->router.ddc_mux_state; + radeon_i2c_put_byte(radeon_connector->router_bus, + radeon_connector->router.i2c_addr, + 0x1, val); +} + +/* clock/data router switching */ +void radeon_router_select_cd_port(struct radeon_connector *radeon_connector) +{ + u8 val; + + if (!radeon_connector->router.cd_valid) + return; + + if (!radeon_connector->router_bus) + return; + + radeon_i2c_get_byte(radeon_connector->router_bus, + radeon_connector->router.i2c_addr, + 0x3, &val); + val &= ~radeon_connector->router.cd_mux_control_pin; + radeon_i2c_put_byte(radeon_connector->router_bus, + radeon_connector->router.i2c_addr, + 0x3, val); + radeon_i2c_get_byte(radeon_connector->router_bus, + radeon_connector->router.i2c_addr, + 0x1, &val); + val &= ~radeon_connector->router.cd_mux_control_pin; + val |= radeon_connector->router.cd_mux_state; + radeon_i2c_put_byte(radeon_connector->router_bus, + radeon_connector->router.i2c_addr, + 0x1, val); +} + diff --git a/drivers/gpu/drm/radeon/radeon_ioc32.c b/drivers/gpu/drm/radeon/radeon_ioc32.c new file mode 100644 index 00000000..48b7cea3 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_ioc32.c @@ -0,0 +1,425 @@ +/** + * \file radeon_ioc32.c + * + * 32-bit ioctl compatibility routines for the Radeon DRM. + * + * \author Paul Mackerras + * + * Copyright (C) Paul Mackerras 2005 + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHOR BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS + * IN THE SOFTWARE. + */ +#include + +#include "drmP.h" +#include "drm.h" +#include "radeon_drm.h" +#include "radeon_drv.h" + +typedef struct drm_radeon_init32 { + int func; + u32 sarea_priv_offset; + int is_pci; + int cp_mode; + int gart_size; + int ring_size; + int usec_timeout; + + unsigned int fb_bpp; + unsigned int front_offset, front_pitch; + unsigned int back_offset, back_pitch; + unsigned int depth_bpp; + unsigned int depth_offset, depth_pitch; + + u32 fb_offset; + u32 mmio_offset; + u32 ring_offset; + u32 ring_rptr_offset; + u32 buffers_offset; + u32 gart_textures_offset; +} drm_radeon_init32_t; + +static int compat_radeon_cp_init(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_radeon_init32_t init32; + drm_radeon_init_t __user *init; + + if (copy_from_user(&init32, (void __user *)arg, sizeof(init32))) + return -EFAULT; + + init = compat_alloc_user_space(sizeof(*init)); + if (!access_ok(VERIFY_WRITE, init, sizeof(*init)) + || __put_user(init32.func, &init->func) + || __put_user(init32.sarea_priv_offset, &init->sarea_priv_offset) + || __put_user(init32.is_pci, &init->is_pci) + || __put_user(init32.cp_mode, &init->cp_mode) + || __put_user(init32.gart_size, &init->gart_size) + || __put_user(init32.ring_size, &init->ring_size) + || __put_user(init32.usec_timeout, &init->usec_timeout) + || __put_user(init32.fb_bpp, &init->fb_bpp) + || __put_user(init32.front_offset, &init->front_offset) + || __put_user(init32.front_pitch, &init->front_pitch) + || __put_user(init32.back_offset, &init->back_offset) + || __put_user(init32.back_pitch, &init->back_pitch) + || __put_user(init32.depth_bpp, &init->depth_bpp) + || __put_user(init32.depth_offset, &init->depth_offset) + || __put_user(init32.depth_pitch, &init->depth_pitch) + || __put_user(init32.fb_offset, &init->fb_offset) + || __put_user(init32.mmio_offset, &init->mmio_offset) + || __put_user(init32.ring_offset, &init->ring_offset) + || __put_user(init32.ring_rptr_offset, &init->ring_rptr_offset) + || __put_user(init32.buffers_offset, &init->buffers_offset) + || __put_user(init32.gart_textures_offset, + &init->gart_textures_offset)) + return -EFAULT; + + return drm_ioctl(file, DRM_IOCTL_RADEON_CP_INIT, (unsigned long)init); +} + +typedef struct drm_radeon_clear32 { + unsigned int flags; + unsigned int clear_color; + unsigned int clear_depth; + unsigned int color_mask; + unsigned int depth_mask; /* misnamed field: should be stencil */ + u32 depth_boxes; +} drm_radeon_clear32_t; + +static int compat_radeon_cp_clear(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_radeon_clear32_t clr32; + drm_radeon_clear_t __user *clr; + + if (copy_from_user(&clr32, (void __user *)arg, sizeof(clr32))) + return -EFAULT; + + clr = compat_alloc_user_space(sizeof(*clr)); + if (!access_ok(VERIFY_WRITE, clr, sizeof(*clr)) + || __put_user(clr32.flags, &clr->flags) + || __put_user(clr32.clear_color, &clr->clear_color) + || __put_user(clr32.clear_depth, &clr->clear_depth) + || __put_user(clr32.color_mask, &clr->color_mask) + || __put_user(clr32.depth_mask, &clr->depth_mask) + || __put_user((void __user *)(unsigned long)clr32.depth_boxes, + &clr->depth_boxes)) + return -EFAULT; + + return drm_ioctl(file, DRM_IOCTL_RADEON_CLEAR, (unsigned long)clr); +} + +typedef struct drm_radeon_stipple32 { + u32 mask; +} drm_radeon_stipple32_t; + +static int compat_radeon_cp_stipple(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_radeon_stipple32_t __user *argp = (void __user *)arg; + drm_radeon_stipple_t __user *request; + u32 mask; + + if (get_user(mask, &argp->mask)) + return -EFAULT; + + request = compat_alloc_user_space(sizeof(*request)); + if (!access_ok(VERIFY_WRITE, request, sizeof(*request)) + || __put_user((unsigned int __user *)(unsigned long)mask, + &request->mask)) + return -EFAULT; + + return drm_ioctl(file, DRM_IOCTL_RADEON_STIPPLE, (unsigned long)request); +} + +typedef struct drm_radeon_tex_image32 { + unsigned int x, y; /* Blit coordinates */ + unsigned int width, height; + u32 data; +} drm_radeon_tex_image32_t; + +typedef struct drm_radeon_texture32 { + unsigned int offset; + int pitch; + int format; + int width; /* Texture image coordinates */ + int height; + u32 image; +} drm_radeon_texture32_t; + +static int compat_radeon_cp_texture(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_radeon_texture32_t req32; + drm_radeon_texture_t __user *request; + drm_radeon_tex_image32_t img32; + drm_radeon_tex_image_t __user *image; + + if (copy_from_user(&req32, (void __user *)arg, sizeof(req32))) + return -EFAULT; + if (req32.image == 0) + return -EINVAL; + if (copy_from_user(&img32, (void __user *)(unsigned long)req32.image, + sizeof(img32))) + return -EFAULT; + + request = compat_alloc_user_space(sizeof(*request) + sizeof(*image)); + if (!access_ok(VERIFY_WRITE, request, + sizeof(*request) + sizeof(*image))) + return -EFAULT; + image = (drm_radeon_tex_image_t __user *) (request + 1); + + if (__put_user(req32.offset, &request->offset) + || __put_user(req32.pitch, &request->pitch) + || __put_user(req32.format, &request->format) + || __put_user(req32.width, &request->width) + || __put_user(req32.height, &request->height) + || __put_user(image, &request->image) + || __put_user(img32.x, &image->x) + || __put_user(img32.y, &image->y) + || __put_user(img32.width, &image->width) + || __put_user(img32.height, &image->height) + || __put_user((const void __user *)(unsigned long)img32.data, + &image->data)) + return -EFAULT; + + return drm_ioctl(file, DRM_IOCTL_RADEON_TEXTURE, (unsigned long)request); +} + +typedef struct drm_radeon_vertex2_32 { + int idx; /* Index of vertex buffer */ + int discard; /* Client finished with buffer? */ + int nr_states; + u32 state; + int nr_prims; + u32 prim; +} drm_radeon_vertex2_32_t; + +static int compat_radeon_cp_vertex2(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_radeon_vertex2_32_t req32; + drm_radeon_vertex2_t __user *request; + + if (copy_from_user(&req32, (void __user *)arg, sizeof(req32))) + return -EFAULT; + + request = compat_alloc_user_space(sizeof(*request)); + if (!access_ok(VERIFY_WRITE, request, sizeof(*request)) + || __put_user(req32.idx, &request->idx) + || __put_user(req32.discard, &request->discard) + || __put_user(req32.nr_states, &request->nr_states) + || __put_user((void __user *)(unsigned long)req32.state, + &request->state) + || __put_user(req32.nr_prims, &request->nr_prims) + || __put_user((void __user *)(unsigned long)req32.prim, + &request->prim)) + return -EFAULT; + + return drm_ioctl(file, DRM_IOCTL_RADEON_VERTEX2, (unsigned long)request); +} + +typedef struct drm_radeon_cmd_buffer32 { + int bufsz; + u32 buf; + int nbox; + u32 boxes; +} drm_radeon_cmd_buffer32_t; + +static int compat_radeon_cp_cmdbuf(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_radeon_cmd_buffer32_t req32; + drm_radeon_cmd_buffer_t __user *request; + + if (copy_from_user(&req32, (void __user *)arg, sizeof(req32))) + return -EFAULT; + + request = compat_alloc_user_space(sizeof(*request)); + if (!access_ok(VERIFY_WRITE, request, sizeof(*request)) + || __put_user(req32.bufsz, &request->bufsz) + || __put_user((void __user *)(unsigned long)req32.buf, + &request->buf) + || __put_user(req32.nbox, &request->nbox) + || __put_user((void __user *)(unsigned long)req32.boxes, + &request->boxes)) + return -EFAULT; + + return drm_ioctl(file, DRM_IOCTL_RADEON_CMDBUF, (unsigned long)request); +} + +typedef struct drm_radeon_getparam32 { + int param; + u32 value; +} drm_radeon_getparam32_t; + +static int compat_radeon_cp_getparam(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_radeon_getparam32_t req32; + drm_radeon_getparam_t __user *request; + + if (copy_from_user(&req32, (void __user *)arg, sizeof(req32))) + return -EFAULT; + + request = compat_alloc_user_space(sizeof(*request)); + if (!access_ok(VERIFY_WRITE, request, sizeof(*request)) + || __put_user(req32.param, &request->param) + || __put_user((void __user *)(unsigned long)req32.value, + &request->value)) + return -EFAULT; + + return drm_ioctl(file, DRM_IOCTL_RADEON_GETPARAM, (unsigned long)request); +} + +typedef struct drm_radeon_mem_alloc32 { + int region; + int alignment; + int size; + u32 region_offset; /* offset from start of fb or GART */ +} drm_radeon_mem_alloc32_t; + +static int compat_radeon_mem_alloc(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_radeon_mem_alloc32_t req32; + drm_radeon_mem_alloc_t __user *request; + + if (copy_from_user(&req32, (void __user *)arg, sizeof(req32))) + return -EFAULT; + + request = compat_alloc_user_space(sizeof(*request)); + if (!access_ok(VERIFY_WRITE, request, sizeof(*request)) + || __put_user(req32.region, &request->region) + || __put_user(req32.alignment, &request->alignment) + || __put_user(req32.size, &request->size) + || __put_user((int __user *)(unsigned long)req32.region_offset, + &request->region_offset)) + return -EFAULT; + + return drm_ioctl(file, DRM_IOCTL_RADEON_ALLOC, (unsigned long)request); +} + +typedef struct drm_radeon_irq_emit32 { + u32 irq_seq; +} drm_radeon_irq_emit32_t; + +static int compat_radeon_irq_emit(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_radeon_irq_emit32_t req32; + drm_radeon_irq_emit_t __user *request; + + if (copy_from_user(&req32, (void __user *)arg, sizeof(req32))) + return -EFAULT; + + request = compat_alloc_user_space(sizeof(*request)); + if (!access_ok(VERIFY_WRITE, request, sizeof(*request)) + || __put_user((int __user *)(unsigned long)req32.irq_seq, + &request->irq_seq)) + return -EFAULT; + + return drm_ioctl(file, DRM_IOCTL_RADEON_IRQ_EMIT, (unsigned long)request); +} + +/* The two 64-bit arches where alignof(u64)==4 in 32-bit code */ +#if defined (CONFIG_X86_64) || defined(CONFIG_IA64) +typedef struct drm_radeon_setparam32 { + int param; + u64 value; +} __attribute__((packed)) drm_radeon_setparam32_t; + +static int compat_radeon_cp_setparam(struct file *file, unsigned int cmd, + unsigned long arg) +{ + drm_radeon_setparam32_t req32; + drm_radeon_setparam_t __user *request; + + if (copy_from_user(&req32, (void __user *) arg, sizeof(req32))) + return -EFAULT; + + request = compat_alloc_user_space(sizeof(*request)); + if (!access_ok(VERIFY_WRITE, request, sizeof(*request)) + || __put_user(req32.param, &request->param) + || __put_user((void __user *)(unsigned long)req32.value, + &request->value)) + return -EFAULT; + + return drm_ioctl(file, DRM_IOCTL_RADEON_SETPARAM, (unsigned long) request); +} +#else +#define compat_radeon_cp_setparam NULL +#endif /* X86_64 || IA64 */ + +drm_ioctl_compat_t *radeon_compat_ioctls[] = { + [DRM_RADEON_CP_INIT] = compat_radeon_cp_init, + [DRM_RADEON_CLEAR] = compat_radeon_cp_clear, + [DRM_RADEON_STIPPLE] = compat_radeon_cp_stipple, + [DRM_RADEON_TEXTURE] = compat_radeon_cp_texture, + [DRM_RADEON_VERTEX2] = compat_radeon_cp_vertex2, + [DRM_RADEON_CMDBUF] = compat_radeon_cp_cmdbuf, + [DRM_RADEON_GETPARAM] = compat_radeon_cp_getparam, + [DRM_RADEON_SETPARAM] = compat_radeon_cp_setparam, + [DRM_RADEON_ALLOC] = compat_radeon_mem_alloc, + [DRM_RADEON_IRQ_EMIT] = compat_radeon_irq_emit, +}; + +/** + * Called whenever a 32-bit process running under a 64-bit kernel + * performs an ioctl on /dev/dri/card. + * + * \param filp file pointer. + * \param cmd command. + * \param arg user argument. + * \return zero on success or negative number on failure. + */ +long radeon_compat_ioctl(struct file *filp, unsigned int cmd, unsigned long arg) +{ + unsigned int nr = DRM_IOCTL_NR(cmd); + drm_ioctl_compat_t *fn = NULL; + int ret; + + if (nr < DRM_COMMAND_BASE) + return drm_compat_ioctl(filp, cmd, arg); + + if (nr < DRM_COMMAND_BASE + DRM_ARRAY_SIZE(radeon_compat_ioctls)) + fn = radeon_compat_ioctls[nr - DRM_COMMAND_BASE]; + + if (fn != NULL) + ret = (*fn) (filp, cmd, arg); + else + ret = drm_ioctl(filp, cmd, arg); + + return ret; +} + +long radeon_kms_compat_ioctl(struct file *filp, unsigned int cmd, unsigned long arg) +{ + unsigned int nr = DRM_IOCTL_NR(cmd); + int ret; + + if (nr < DRM_COMMAND_BASE) + return drm_compat_ioctl(filp, cmd, arg); + + ret = drm_ioctl(filp, cmd, arg); + + return ret; +} diff --git a/drivers/gpu/drm/radeon/radeon_irq.c b/drivers/gpu/drm/radeon/radeon_irq.c new file mode 100644 index 00000000..465746bd --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_irq.c @@ -0,0 +1,401 @@ +/* radeon_irq.c -- IRQ handling for radeon -*- linux-c -*- */ +/* + * Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + * + * The Weather Channel (TM) funded Tungsten Graphics to develop the + * initial release of the Radeon 8500 driver under the XFree86 license. + * This notice must be preserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Keith Whitwell + * Michel D�zer + */ + +#include "drmP.h" +#include "drm.h" +#include "radeon_drm.h" +#include "radeon_drv.h" + +void radeon_irq_set_state(struct drm_device *dev, u32 mask, int state) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + + if (state) + dev_priv->irq_enable_reg |= mask; + else + dev_priv->irq_enable_reg &= ~mask; + + if (dev->irq_enabled) + RADEON_WRITE(RADEON_GEN_INT_CNTL, dev_priv->irq_enable_reg); +} + +static void r500_vbl_irq_set_state(struct drm_device *dev, u32 mask, int state) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + + if (state) + dev_priv->r500_disp_irq_reg |= mask; + else + dev_priv->r500_disp_irq_reg &= ~mask; + + if (dev->irq_enabled) + RADEON_WRITE(R500_DxMODE_INT_MASK, dev_priv->r500_disp_irq_reg); +} + +int radeon_enable_vblank(struct drm_device *dev, int crtc) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_RS600) { + switch (crtc) { + case 0: + r500_vbl_irq_set_state(dev, R500_D1MODE_INT_MASK, 1); + break; + case 1: + r500_vbl_irq_set_state(dev, R500_D2MODE_INT_MASK, 1); + break; + default: + DRM_ERROR("tried to enable vblank on non-existent crtc %d\n", + crtc); + return -EINVAL; + } + } else { + switch (crtc) { + case 0: + radeon_irq_set_state(dev, RADEON_CRTC_VBLANK_MASK, 1); + break; + case 1: + radeon_irq_set_state(dev, RADEON_CRTC2_VBLANK_MASK, 1); + break; + default: + DRM_ERROR("tried to enable vblank on non-existent crtc %d\n", + crtc); + return -EINVAL; + } + } + + return 0; +} + +void radeon_disable_vblank(struct drm_device *dev, int crtc) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_RS600) { + switch (crtc) { + case 0: + r500_vbl_irq_set_state(dev, R500_D1MODE_INT_MASK, 0); + break; + case 1: + r500_vbl_irq_set_state(dev, R500_D2MODE_INT_MASK, 0); + break; + default: + DRM_ERROR("tried to enable vblank on non-existent crtc %d\n", + crtc); + break; + } + } else { + switch (crtc) { + case 0: + radeon_irq_set_state(dev, RADEON_CRTC_VBLANK_MASK, 0); + break; + case 1: + radeon_irq_set_state(dev, RADEON_CRTC2_VBLANK_MASK, 0); + break; + default: + DRM_ERROR("tried to enable vblank on non-existent crtc %d\n", + crtc); + break; + } + } +} + +static inline u32 radeon_acknowledge_irqs(drm_radeon_private_t *dev_priv, u32 *r500_disp_int) +{ + u32 irqs = RADEON_READ(RADEON_GEN_INT_STATUS); + u32 irq_mask = RADEON_SW_INT_TEST; + + *r500_disp_int = 0; + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_RS600) { + /* vbl interrupts in a different place */ + + if (irqs & R500_DISPLAY_INT_STATUS) { + /* if a display interrupt */ + u32 disp_irq; + + disp_irq = RADEON_READ(R500_DISP_INTERRUPT_STATUS); + + *r500_disp_int = disp_irq; + if (disp_irq & R500_D1_VBLANK_INTERRUPT) + RADEON_WRITE(R500_D1MODE_VBLANK_STATUS, R500_VBLANK_ACK); + if (disp_irq & R500_D2_VBLANK_INTERRUPT) + RADEON_WRITE(R500_D2MODE_VBLANK_STATUS, R500_VBLANK_ACK); + } + irq_mask |= R500_DISPLAY_INT_STATUS; + } else + irq_mask |= RADEON_CRTC_VBLANK_STAT | RADEON_CRTC2_VBLANK_STAT; + + irqs &= irq_mask; + + if (irqs) + RADEON_WRITE(RADEON_GEN_INT_STATUS, irqs); + + return irqs; +} + +/* Interrupts - Used for device synchronization and flushing in the + * following circumstances: + * + * - Exclusive FB access with hw idle: + * - Wait for GUI Idle (?) interrupt, then do normal flush. + * + * - Frame throttling, NV_fence: + * - Drop marker irq's into command stream ahead of time. + * - Wait on irq's with lock *not held* + * - Check each for termination condition + * + * - Internally in cp_getbuffer, etc: + * - as above, but wait with lock held??? + * + * NOTE: These functions are misleadingly named -- the irq's aren't + * tied to dma at all, this is just a hangover from dri prehistory. + */ + +irqreturn_t radeon_driver_irq_handler(DRM_IRQ_ARGS) +{ + struct drm_device *dev = (struct drm_device *) arg; + drm_radeon_private_t *dev_priv = + (drm_radeon_private_t *) dev->dev_private; + u32 stat; + u32 r500_disp_int; + + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) + return IRQ_NONE; + + /* Only consider the bits we're interested in - others could be used + * outside the DRM + */ + stat = radeon_acknowledge_irqs(dev_priv, &r500_disp_int); + if (!stat) + return IRQ_NONE; + + stat &= dev_priv->irq_enable_reg; + + /* SW interrupt */ + if (stat & RADEON_SW_INT_TEST) + DRM_WAKEUP(&dev_priv->swi_queue); + + /* VBLANK interrupt */ + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_RS600) { + if (r500_disp_int & R500_D1_VBLANK_INTERRUPT) + drm_handle_vblank(dev, 0); + if (r500_disp_int & R500_D2_VBLANK_INTERRUPT) + drm_handle_vblank(dev, 1); + } else { + if (stat & RADEON_CRTC_VBLANK_STAT) + drm_handle_vblank(dev, 0); + if (stat & RADEON_CRTC2_VBLANK_STAT) + drm_handle_vblank(dev, 1); + } + return IRQ_HANDLED; +} + +static int radeon_emit_irq(struct drm_device * dev) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + unsigned int ret; + RING_LOCALS; + + atomic_inc(&dev_priv->swi_emitted); + ret = atomic_read(&dev_priv->swi_emitted); + + BEGIN_RING(4); + OUT_RING_REG(RADEON_LAST_SWI_REG, ret); + OUT_RING_REG(RADEON_GEN_INT_STATUS, RADEON_SW_INT_FIRE); + ADVANCE_RING(); + COMMIT_RING(); + + return ret; +} + +static int radeon_wait_irq(struct drm_device * dev, int swi_nr) +{ + drm_radeon_private_t *dev_priv = + (drm_radeon_private_t *) dev->dev_private; + int ret = 0; + + if (RADEON_READ(RADEON_LAST_SWI_REG) >= swi_nr) + return 0; + + dev_priv->stats.boxes |= RADEON_BOX_WAIT_IDLE; + + DRM_WAIT_ON(ret, dev_priv->swi_queue, 3 * DRM_HZ, + RADEON_READ(RADEON_LAST_SWI_REG) >= swi_nr); + + return ret; +} + +u32 radeon_get_vblank_counter(struct drm_device *dev, int crtc) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + + if (!dev_priv) { + DRM_ERROR("called with no initialization\n"); + return -EINVAL; + } + + if (crtc < 0 || crtc > 1) { + DRM_ERROR("Invalid crtc %d\n", crtc); + return -EINVAL; + } + + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_RS600) { + if (crtc == 0) + return RADEON_READ(R500_D1CRTC_FRAME_COUNT); + else + return RADEON_READ(R500_D2CRTC_FRAME_COUNT); + } else { + if (crtc == 0) + return RADEON_READ(RADEON_CRTC_CRNT_FRAME); + else + return RADEON_READ(RADEON_CRTC2_CRNT_FRAME); + } +} + +/* Needs the lock as it touches the ring. + */ +int radeon_irq_emit(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + drm_radeon_irq_emit_t *emit = data; + int result; + + if (!dev_priv) { + DRM_ERROR("called with no initialization\n"); + return -EINVAL; + } + + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) + return -EINVAL; + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + result = radeon_emit_irq(dev); + + if (DRM_COPY_TO_USER(emit->irq_seq, &result, sizeof(int))) { + DRM_ERROR("copy_to_user\n"); + return -EFAULT; + } + + return 0; +} + +/* Doesn't need the hardware lock. + */ +int radeon_irq_wait(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + drm_radeon_irq_wait_t *irqwait = data; + + if (!dev_priv) { + DRM_ERROR("called with no initialization\n"); + return -EINVAL; + } + + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) + return -EINVAL; + + return radeon_wait_irq(dev, irqwait->irq_seq); +} + +/* drm_dma.h hooks +*/ +void radeon_driver_irq_preinstall(struct drm_device * dev) +{ + drm_radeon_private_t *dev_priv = + (drm_radeon_private_t *) dev->dev_private; + u32 dummy; + + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) + return; + + /* Disable *all* interrupts */ + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_RS600) + RADEON_WRITE(R500_DxMODE_INT_MASK, 0); + RADEON_WRITE(RADEON_GEN_INT_CNTL, 0); + + /* Clear bits if they're already high */ + radeon_acknowledge_irqs(dev_priv, &dummy); +} + +int radeon_driver_irq_postinstall(struct drm_device *dev) +{ + drm_radeon_private_t *dev_priv = + (drm_radeon_private_t *) dev->dev_private; + + atomic_set(&dev_priv->swi_emitted, 0); + DRM_INIT_WAITQUEUE(&dev_priv->swi_queue); + + dev->max_vblank_count = 0x001fffff; + + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) + return 0; + + radeon_irq_set_state(dev, RADEON_SW_INT_ENABLE, 1); + + return 0; +} + +void radeon_driver_irq_uninstall(struct drm_device * dev) +{ + drm_radeon_private_t *dev_priv = + (drm_radeon_private_t *) dev->dev_private; + if (!dev_priv) + return; + + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) + return; + + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_RS600) + RADEON_WRITE(R500_DxMODE_INT_MASK, 0); + /* Disable *all* interrupts */ + RADEON_WRITE(RADEON_GEN_INT_CNTL, 0); +} + + +int radeon_vblank_crtc_get(struct drm_device *dev) +{ + drm_radeon_private_t *dev_priv = (drm_radeon_private_t *) dev->dev_private; + + return dev_priv->vblank_crtc; +} + +int radeon_vblank_crtc_set(struct drm_device *dev, int64_t value) +{ + drm_radeon_private_t *dev_priv = (drm_radeon_private_t *) dev->dev_private; + if (value & ~(DRM_RADEON_VBLANK_CRTC1 | DRM_RADEON_VBLANK_CRTC2)) { + DRM_ERROR("called with invalid crtc 0x%x\n", (unsigned int)value); + return -EINVAL; + } + dev_priv->vblank_crtc = (unsigned int)value; + return 0; +} diff --git a/drivers/gpu/drm/radeon/radeon_irq_kms.c b/drivers/gpu/drm/radeon/radeon_irq_kms.c new file mode 100644 index 00000000..eb6fe79c --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_irq_kms.c @@ -0,0 +1,264 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#include "drmP.h" +#include "drm_crtc_helper.h" +#include "radeon_drm.h" +#include "radeon_reg.h" +#include "radeon.h" +#include "atom.h" + +irqreturn_t radeon_driver_irq_handler_kms(DRM_IRQ_ARGS) +{ + struct drm_device *dev = (struct drm_device *) arg; + struct radeon_device *rdev = dev->dev_private; + + return radeon_irq_process(rdev); +} + +/* + * Handle hotplug events outside the interrupt handler proper. + */ +static void radeon_hotplug_work_func(struct work_struct *work) +{ + struct radeon_device *rdev = container_of(work, struct radeon_device, + hotplug_work); + struct drm_device *dev = rdev->ddev; + struct drm_mode_config *mode_config = &dev->mode_config; + struct drm_connector *connector; + + if (mode_config->num_connector) { + list_for_each_entry(connector, &mode_config->connector_list, head) + radeon_connector_hotplug(connector); + } + /* Just fire off a uevent and let userspace tell us what to do */ + drm_helper_hpd_irq_event(dev); +} + +void radeon_driver_irq_preinstall_kms(struct drm_device *dev) +{ + struct radeon_device *rdev = dev->dev_private; + unsigned i; + + /* Disable *all* interrupts */ + rdev->irq.sw_int = false; + rdev->irq.gui_idle = false; + for (i = 0; i < rdev->num_crtc; i++) + rdev->irq.crtc_vblank_int[i] = false; + for (i = 0; i < 6; i++) { + rdev->irq.hpd[i] = false; + rdev->irq.pflip[i] = false; + } + radeon_irq_set(rdev); + /* Clear bits */ + radeon_irq_process(rdev); +} + +int radeon_driver_irq_postinstall_kms(struct drm_device *dev) +{ + struct radeon_device *rdev = dev->dev_private; + + dev->max_vblank_count = 0x001fffff; + rdev->irq.sw_int = true; + radeon_irq_set(rdev); + return 0; +} + +void radeon_driver_irq_uninstall_kms(struct drm_device *dev) +{ + struct radeon_device *rdev = dev->dev_private; + unsigned i; + + if (rdev == NULL) { + return; + } + /* Disable *all* interrupts */ + rdev->irq.sw_int = false; + rdev->irq.gui_idle = false; + for (i = 0; i < rdev->num_crtc; i++) + rdev->irq.crtc_vblank_int[i] = false; + for (i = 0; i < 6; i++) { + rdev->irq.hpd[i] = false; + rdev->irq.pflip[i] = false; + } + radeon_irq_set(rdev); +} + +static bool radeon_msi_ok(struct radeon_device *rdev) +{ + /* RV370/RV380 was first asic with MSI support */ + if (rdev->family < CHIP_RV380) + return false; + + /* MSIs don't work on AGP */ + if (rdev->flags & RADEON_IS_AGP) + return false; + + /* force MSI on */ + if (radeon_msi == 1) + return true; + else if (radeon_msi == 0) + return false; + + /* Quirks */ + /* HP RS690 only seems to work with MSIs. */ + if ((rdev->pdev->device == 0x791f) && + (rdev->pdev->subsystem_vendor == 0x103c) && + (rdev->pdev->subsystem_device == 0x30c2)) + return true; + + /* Dell RS690 only seems to work with MSIs. */ + if ((rdev->pdev->device == 0x791f) && + (rdev->pdev->subsystem_vendor == 0x1028) && + (rdev->pdev->subsystem_device == 0x01fc)) + return true; + + /* Dell RS690 only seems to work with MSIs. */ + if ((rdev->pdev->device == 0x791f) && + (rdev->pdev->subsystem_vendor == 0x1028) && + (rdev->pdev->subsystem_device == 0x01fd)) + return true; + + /* RV515 seems to have MSI issues where it loses + * MSI rearms occasionally. This leads to lockups and freezes. + * disable it by default. + */ + if (rdev->family == CHIP_RV515) + return false; + if (rdev->flags & RADEON_IS_IGP) { + /* APUs work fine with MSIs */ + if (rdev->family >= CHIP_PALM) + return true; + /* lots of IGPs have problems with MSIs */ + return false; + } + + return true; +} + +int radeon_irq_kms_init(struct radeon_device *rdev) +{ + int i; + int r = 0; + + INIT_WORK(&rdev->hotplug_work, radeon_hotplug_work_func); + + spin_lock_init(&rdev->irq.sw_lock); + for (i = 0; i < rdev->num_crtc; i++) + spin_lock_init(&rdev->irq.pflip_lock[i]); + r = drm_vblank_init(rdev->ddev, rdev->num_crtc); + if (r) { + return r; + } + /* enable msi */ + rdev->msi_enabled = 0; + + if (radeon_msi_ok(rdev)) { + int ret = pci_enable_msi(rdev->pdev); + if (!ret) { + rdev->msi_enabled = 1; + dev_info(rdev->dev, "radeon: using MSI.\n"); + } + } + rdev->irq.installed = true; + r = drm_irq_install(rdev->ddev); + if (r) { + rdev->irq.installed = false; + return r; + } + DRM_INFO("radeon: irq initialized.\n"); + return 0; +} + +void radeon_irq_kms_fini(struct radeon_device *rdev) +{ + drm_vblank_cleanup(rdev->ddev); + if (rdev->irq.installed) { + drm_irq_uninstall(rdev->ddev); + rdev->irq.installed = false; + if (rdev->msi_enabled) + pci_disable_msi(rdev->pdev); + } + flush_work_sync(&rdev->hotplug_work); +} + +void radeon_irq_kms_sw_irq_get(struct radeon_device *rdev) +{ + unsigned long irqflags; + + spin_lock_irqsave(&rdev->irq.sw_lock, irqflags); + if (rdev->ddev->irq_enabled && (++rdev->irq.sw_refcount == 1)) { + rdev->irq.sw_int = true; + radeon_irq_set(rdev); + } + spin_unlock_irqrestore(&rdev->irq.sw_lock, irqflags); +} + +void radeon_irq_kms_sw_irq_put(struct radeon_device *rdev) +{ + unsigned long irqflags; + + spin_lock_irqsave(&rdev->irq.sw_lock, irqflags); + BUG_ON(rdev->ddev->irq_enabled && rdev->irq.sw_refcount <= 0); + if (rdev->ddev->irq_enabled && (--rdev->irq.sw_refcount == 0)) { + rdev->irq.sw_int = false; + radeon_irq_set(rdev); + } + spin_unlock_irqrestore(&rdev->irq.sw_lock, irqflags); +} + +void radeon_irq_kms_pflip_irq_get(struct radeon_device *rdev, int crtc) +{ + unsigned long irqflags; + + if (crtc < 0 || crtc >= rdev->num_crtc) + return; + + spin_lock_irqsave(&rdev->irq.pflip_lock[crtc], irqflags); + if (rdev->ddev->irq_enabled && (++rdev->irq.pflip_refcount[crtc] == 1)) { + rdev->irq.pflip[crtc] = true; + radeon_irq_set(rdev); + } + spin_unlock_irqrestore(&rdev->irq.pflip_lock[crtc], irqflags); +} + +void radeon_irq_kms_pflip_irq_put(struct radeon_device *rdev, int crtc) +{ + unsigned long irqflags; + + if (crtc < 0 || crtc >= rdev->num_crtc) + return; + + spin_lock_irqsave(&rdev->irq.pflip_lock[crtc], irqflags); + BUG_ON(rdev->ddev->irq_enabled && rdev->irq.pflip_refcount[crtc] <= 0); + if (rdev->ddev->irq_enabled && (--rdev->irq.pflip_refcount[crtc] == 0)) { + rdev->irq.pflip[crtc] = false; + radeon_irq_set(rdev); + } + spin_unlock_irqrestore(&rdev->irq.pflip_lock[crtc], irqflags); +} + diff --git a/drivers/gpu/drm/radeon/radeon_kms.c b/drivers/gpu/drm/radeon/radeon_kms.c new file mode 100644 index 00000000..bd58af65 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_kms.c @@ -0,0 +1,442 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#include "drmP.h" +#include "drm_sarea.h" +#include "radeon.h" +#include "radeon_drm.h" + +#include +#include + +int radeon_driver_unload_kms(struct drm_device *dev) +{ + struct radeon_device *rdev = dev->dev_private; + + if (rdev == NULL) + return 0; + radeon_modeset_fini(rdev); + radeon_device_fini(rdev); + kfree(rdev); + dev->dev_private = NULL; + return 0; +} + +int radeon_driver_load_kms(struct drm_device *dev, unsigned long flags) +{ + struct radeon_device *rdev; + int r, acpi_status; + + rdev = kzalloc(sizeof(struct radeon_device), GFP_KERNEL); + if (rdev == NULL) { + return -ENOMEM; + } + dev->dev_private = (void *)rdev; + + /* update BUS flag */ + if (drm_pci_device_is_agp(dev)) { + flags |= RADEON_IS_AGP; + } else if (drm_pci_device_is_pcie(dev)) { + flags |= RADEON_IS_PCIE; + } else { + flags |= RADEON_IS_PCI; + } + + /* radeon_device_init should report only fatal error + * like memory allocation failure or iomapping failure, + * or memory manager initialization failure, it must + * properly initialize the GPU MC controller and permit + * VRAM allocation + */ + r = radeon_device_init(rdev, dev, dev->pdev, flags); + if (r) { + dev_err(&dev->pdev->dev, "Fatal error during GPU init\n"); + goto out; + } + + /* Call ACPI methods */ + acpi_status = radeon_acpi_init(rdev); + if (acpi_status) + dev_dbg(&dev->pdev->dev, "Error during ACPI methods call\n"); + + /* Again modeset_init should fail only on fatal error + * otherwise it should provide enough functionalities + * for shadowfb to run + */ + r = radeon_modeset_init(rdev); + if (r) + dev_err(&dev->pdev->dev, "Fatal error during modeset init\n"); +out: + if (r) + radeon_driver_unload_kms(dev); + return r; +} + +static void radeon_set_filp_rights(struct drm_device *dev, + struct drm_file **owner, + struct drm_file *applier, + uint32_t *value) +{ + mutex_lock(&dev->struct_mutex); + if (*value == 1) { + /* wants rights */ + if (!*owner) + *owner = applier; + } else if (*value == 0) { + /* revokes rights */ + if (*owner == applier) + *owner = NULL; + } + *value = *owner == applier ? 1 : 0; + mutex_unlock(&dev->struct_mutex); +} + +/* + * Userspace get information ioctl + */ +int radeon_info_ioctl(struct drm_device *dev, void *data, struct drm_file *filp) +{ + struct radeon_device *rdev = dev->dev_private; + struct drm_radeon_info *info; + struct radeon_mode_info *minfo = &rdev->mode_info; + uint32_t *value_ptr; + uint32_t value; + struct drm_crtc *crtc; + int i, found; + + info = data; + value_ptr = (uint32_t *)((unsigned long)info->value); + if (DRM_COPY_FROM_USER(&value, value_ptr, sizeof(value))) + return -EFAULT; + + switch (info->request) { + case RADEON_INFO_DEVICE_ID: + value = dev->pci_device; + break; + case RADEON_INFO_NUM_GB_PIPES: + value = rdev->num_gb_pipes; + break; + case RADEON_INFO_NUM_Z_PIPES: + value = rdev->num_z_pipes; + break; + case RADEON_INFO_ACCEL_WORKING: + /* xf86-video-ati 6.13.0 relies on this being false for evergreen */ + if ((rdev->family >= CHIP_CEDAR) && (rdev->family <= CHIP_HEMLOCK)) + value = false; + else + value = rdev->accel_working; + break; + case RADEON_INFO_CRTC_FROM_ID: + for (i = 0, found = 0; i < rdev->num_crtc; i++) { + crtc = (struct drm_crtc *)minfo->crtcs[i]; + if (crtc && crtc->base.id == value) { + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + value = radeon_crtc->crtc_id; + found = 1; + break; + } + } + if (!found) { + DRM_DEBUG_KMS("unknown crtc id %d\n", value); + return -EINVAL; + } + break; + case RADEON_INFO_ACCEL_WORKING2: + value = rdev->accel_working; + break; + case RADEON_INFO_TILING_CONFIG: + if (rdev->family >= CHIP_CAYMAN) + value = rdev->config.cayman.tile_config; + else if (rdev->family >= CHIP_CEDAR) + value = rdev->config.evergreen.tile_config; + else if (rdev->family >= CHIP_RV770) + value = rdev->config.rv770.tile_config; + else if (rdev->family >= CHIP_R600) + value = rdev->config.r600.tile_config; + else { + DRM_DEBUG_KMS("tiling config is r6xx+ only!\n"); + return -EINVAL; + } + break; + case RADEON_INFO_WANT_HYPERZ: + /* The "value" here is both an input and output parameter. + * If the input value is 1, filp requests hyper-z access. + * If the input value is 0, filp revokes its hyper-z access. + * + * When returning, the value is 1 if filp owns hyper-z access, + * 0 otherwise. */ + if (value >= 2) { + DRM_DEBUG_KMS("WANT_HYPERZ: invalid value %d\n", value); + return -EINVAL; + } + radeon_set_filp_rights(dev, &rdev->hyperz_filp, filp, &value); + break; + case RADEON_INFO_WANT_CMASK: + /* The same logic as Hyper-Z. */ + if (value >= 2) { + DRM_DEBUG_KMS("WANT_CMASK: invalid value %d\n", value); + return -EINVAL; + } + radeon_set_filp_rights(dev, &rdev->cmask_filp, filp, &value); + break; + case RADEON_INFO_CLOCK_CRYSTAL_FREQ: + /* return clock value in KHz */ + value = rdev->clock.spll.reference_freq * 10; + break; + case RADEON_INFO_NUM_BACKENDS: + if (rdev->family >= CHIP_CAYMAN) + value = rdev->config.cayman.max_backends_per_se * + rdev->config.cayman.max_shader_engines; + else if (rdev->family >= CHIP_CEDAR) + value = rdev->config.evergreen.max_backends; + else if (rdev->family >= CHIP_RV770) + value = rdev->config.rv770.max_backends; + else if (rdev->family >= CHIP_R600) + value = rdev->config.r600.max_backends; + else { + return -EINVAL; + } + break; + case RADEON_INFO_NUM_TILE_PIPES: + if (rdev->family >= CHIP_CAYMAN) + value = rdev->config.cayman.max_tile_pipes; + else if (rdev->family >= CHIP_CEDAR) + value = rdev->config.evergreen.max_tile_pipes; + else if (rdev->family >= CHIP_RV770) + value = rdev->config.rv770.max_tile_pipes; + else if (rdev->family >= CHIP_R600) + value = rdev->config.r600.max_tile_pipes; + else { + return -EINVAL; + } + break; + case RADEON_INFO_FUSION_GART_WORKING: + value = 1; + break; + default: + DRM_DEBUG_KMS("Invalid request %d\n", info->request); + return -EINVAL; + } + if (DRM_COPY_TO_USER(value_ptr, &value, sizeof(uint32_t))) { + DRM_ERROR("copy_to_user\n"); + return -EFAULT; + } + return 0; +} + + +/* + * Outdated mess for old drm with Xorg being in charge (void function now). + */ +int radeon_driver_firstopen_kms(struct drm_device *dev) +{ + return 0; +} + + +void radeon_driver_lastclose_kms(struct drm_device *dev) +{ + vga_switcheroo_process_delayed_switch(); +} + +int radeon_driver_open_kms(struct drm_device *dev, struct drm_file *file_priv) +{ + return 0; +} + +void radeon_driver_postclose_kms(struct drm_device *dev, + struct drm_file *file_priv) +{ +} + +void radeon_driver_preclose_kms(struct drm_device *dev, + struct drm_file *file_priv) +{ + struct radeon_device *rdev = dev->dev_private; + if (rdev->hyperz_filp == file_priv) + rdev->hyperz_filp = NULL; + if (rdev->cmask_filp == file_priv) + rdev->cmask_filp = NULL; +} + +/* + * VBlank related functions. + */ +u32 radeon_get_vblank_counter_kms(struct drm_device *dev, int crtc) +{ + struct radeon_device *rdev = dev->dev_private; + + if (crtc < 0 || crtc >= rdev->num_crtc) { + DRM_ERROR("Invalid crtc %d\n", crtc); + return -EINVAL; + } + + return radeon_get_vblank_counter(rdev, crtc); +} + +int radeon_enable_vblank_kms(struct drm_device *dev, int crtc) +{ + struct radeon_device *rdev = dev->dev_private; + + if (crtc < 0 || crtc >= rdev->num_crtc) { + DRM_ERROR("Invalid crtc %d\n", crtc); + return -EINVAL; + } + + rdev->irq.crtc_vblank_int[crtc] = true; + + return radeon_irq_set(rdev); +} + +void radeon_disable_vblank_kms(struct drm_device *dev, int crtc) +{ + struct radeon_device *rdev = dev->dev_private; + + if (crtc < 0 || crtc >= rdev->num_crtc) { + DRM_ERROR("Invalid crtc %d\n", crtc); + return; + } + + rdev->irq.crtc_vblank_int[crtc] = false; + + radeon_irq_set(rdev); +} + +int radeon_get_vblank_timestamp_kms(struct drm_device *dev, int crtc, + int *max_error, + struct timeval *vblank_time, + unsigned flags) +{ + struct drm_crtc *drmcrtc; + struct radeon_device *rdev = dev->dev_private; + + if (crtc < 0 || crtc >= dev->num_crtcs) { + DRM_ERROR("Invalid crtc %d\n", crtc); + return -EINVAL; + } + + /* Get associated drm_crtc: */ + drmcrtc = &rdev->mode_info.crtcs[crtc]->base; + + /* Helper routine in DRM core does all the work: */ + return drm_calc_vbltimestamp_from_scanoutpos(dev, crtc, max_error, + vblank_time, flags, + drmcrtc); +} + +/* + * IOCTL. + */ +int radeon_dma_ioctl_kms(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + /* Not valid in KMS. */ + return -EINVAL; +} + +#define KMS_INVALID_IOCTL(name) \ +int name(struct drm_device *dev, void *data, struct drm_file *file_priv)\ +{ \ + DRM_ERROR("invalid ioctl with kms %s\n", __func__); \ + return -EINVAL; \ +} + +/* + * All these ioctls are invalid in kms world. + */ +KMS_INVALID_IOCTL(radeon_cp_init_kms) +KMS_INVALID_IOCTL(radeon_cp_start_kms) +KMS_INVALID_IOCTL(radeon_cp_stop_kms) +KMS_INVALID_IOCTL(radeon_cp_reset_kms) +KMS_INVALID_IOCTL(radeon_cp_idle_kms) +KMS_INVALID_IOCTL(radeon_cp_resume_kms) +KMS_INVALID_IOCTL(radeon_engine_reset_kms) +KMS_INVALID_IOCTL(radeon_fullscreen_kms) +KMS_INVALID_IOCTL(radeon_cp_swap_kms) +KMS_INVALID_IOCTL(radeon_cp_clear_kms) +KMS_INVALID_IOCTL(radeon_cp_vertex_kms) +KMS_INVALID_IOCTL(radeon_cp_indices_kms) +KMS_INVALID_IOCTL(radeon_cp_texture_kms) +KMS_INVALID_IOCTL(radeon_cp_stipple_kms) +KMS_INVALID_IOCTL(radeon_cp_indirect_kms) +KMS_INVALID_IOCTL(radeon_cp_vertex2_kms) +KMS_INVALID_IOCTL(radeon_cp_cmdbuf_kms) +KMS_INVALID_IOCTL(radeon_cp_getparam_kms) +KMS_INVALID_IOCTL(radeon_cp_flip_kms) +KMS_INVALID_IOCTL(radeon_mem_alloc_kms) +KMS_INVALID_IOCTL(radeon_mem_free_kms) +KMS_INVALID_IOCTL(radeon_mem_init_heap_kms) +KMS_INVALID_IOCTL(radeon_irq_emit_kms) +KMS_INVALID_IOCTL(radeon_irq_wait_kms) +KMS_INVALID_IOCTL(radeon_cp_setparam_kms) +KMS_INVALID_IOCTL(radeon_surface_alloc_kms) +KMS_INVALID_IOCTL(radeon_surface_free_kms) + + +struct drm_ioctl_desc radeon_ioctls_kms[] = { + DRM_IOCTL_DEF_DRV(RADEON_CP_INIT, radeon_cp_init_kms, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF_DRV(RADEON_CP_START, radeon_cp_start_kms, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF_DRV(RADEON_CP_STOP, radeon_cp_stop_kms, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF_DRV(RADEON_CP_RESET, radeon_cp_reset_kms, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF_DRV(RADEON_CP_IDLE, radeon_cp_idle_kms, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_CP_RESUME, radeon_cp_resume_kms, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_RESET, radeon_engine_reset_kms, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_FULLSCREEN, radeon_fullscreen_kms, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_SWAP, radeon_cp_swap_kms, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_CLEAR, radeon_cp_clear_kms, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_VERTEX, radeon_cp_vertex_kms, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_INDICES, radeon_cp_indices_kms, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_TEXTURE, radeon_cp_texture_kms, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_STIPPLE, radeon_cp_stipple_kms, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_INDIRECT, radeon_cp_indirect_kms, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF_DRV(RADEON_VERTEX2, radeon_cp_vertex2_kms, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_CMDBUF, radeon_cp_cmdbuf_kms, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_GETPARAM, radeon_cp_getparam_kms, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_FLIP, radeon_cp_flip_kms, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_ALLOC, radeon_mem_alloc_kms, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_FREE, radeon_mem_free_kms, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_INIT_HEAP, radeon_mem_init_heap_kms, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF_DRV(RADEON_IRQ_EMIT, radeon_irq_emit_kms, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_IRQ_WAIT, radeon_irq_wait_kms, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_SETPARAM, radeon_cp_setparam_kms, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_SURF_ALLOC, radeon_surface_alloc_kms, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_SURF_FREE, radeon_surface_free_kms, DRM_AUTH), + /* KMS */ + DRM_IOCTL_DEF_DRV(RADEON_GEM_INFO, radeon_gem_info_ioctl, DRM_AUTH|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(RADEON_GEM_CREATE, radeon_gem_create_ioctl, DRM_AUTH|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(RADEON_GEM_MMAP, radeon_gem_mmap_ioctl, DRM_AUTH|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(RADEON_GEM_SET_DOMAIN, radeon_gem_set_domain_ioctl, DRM_AUTH|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(RADEON_GEM_PREAD, radeon_gem_pread_ioctl, DRM_AUTH|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(RADEON_GEM_PWRITE, radeon_gem_pwrite_ioctl, DRM_AUTH|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(RADEON_GEM_WAIT_IDLE, radeon_gem_wait_idle_ioctl, DRM_AUTH|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(RADEON_CS, radeon_cs_ioctl, DRM_AUTH|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(RADEON_INFO, radeon_info_ioctl, DRM_AUTH|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(RADEON_GEM_SET_TILING, radeon_gem_set_tiling_ioctl, DRM_AUTH|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(RADEON_GEM_GET_TILING, radeon_gem_get_tiling_ioctl, DRM_AUTH|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(RADEON_GEM_BUSY, radeon_gem_busy_ioctl, DRM_AUTH|DRM_UNLOCKED), +}; +int radeon_max_kms_ioctl = DRM_ARRAY_SIZE(radeon_ioctls_kms); diff --git a/drivers/gpu/drm/radeon/radeon_legacy_crtc.c b/drivers/gpu/drm/radeon/radeon_legacy_crtc.c new file mode 100644 index 00000000..41a5d48e --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_legacy_crtc.c @@ -0,0 +1,1075 @@ +/* + * Copyright 2007-8 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + */ +#include +#include +#include +#include +#include "radeon.h" +#include "atom.h" + +static void radeon_overscan_setup(struct drm_crtc *crtc, + struct drm_display_mode *mode) +{ + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + + WREG32(RADEON_OVR_CLR + radeon_crtc->crtc_offset, 0); + WREG32(RADEON_OVR_WID_LEFT_RIGHT + radeon_crtc->crtc_offset, 0); + WREG32(RADEON_OVR_WID_TOP_BOTTOM + radeon_crtc->crtc_offset, 0); +} + +static void radeon_legacy_rmx_mode_set(struct drm_crtc *crtc, + struct drm_display_mode *mode) +{ + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + int xres = mode->hdisplay; + int yres = mode->vdisplay; + bool hscale = true, vscale = true; + int hsync_wid; + int vsync_wid; + int hsync_start; + int blank_width; + u32 scale, inc, crtc_more_cntl; + u32 fp_horz_stretch, fp_vert_stretch, fp_horz_vert_active; + u32 fp_h_sync_strt_wid, fp_crtc_h_total_disp; + u32 fp_v_sync_strt_wid, fp_crtc_v_total_disp; + struct drm_display_mode *native_mode = &radeon_crtc->native_mode; + + fp_vert_stretch = RREG32(RADEON_FP_VERT_STRETCH) & + (RADEON_VERT_STRETCH_RESERVED | + RADEON_VERT_AUTO_RATIO_INC); + fp_horz_stretch = RREG32(RADEON_FP_HORZ_STRETCH) & + (RADEON_HORZ_FP_LOOP_STRETCH | + RADEON_HORZ_AUTO_RATIO_INC); + + crtc_more_cntl = 0; + if ((rdev->family == CHIP_RS100) || + (rdev->family == CHIP_RS200)) { + /* This is to workaround the asic bug for RMX, some versions + of BIOS dosen't have this register initialized correctly. */ + crtc_more_cntl |= RADEON_CRTC_H_CUTOFF_ACTIVE_EN; + } + + + fp_crtc_h_total_disp = ((((mode->crtc_htotal / 8) - 1) & 0x3ff) + | ((((mode->crtc_hdisplay / 8) - 1) & 0x1ff) << 16)); + + hsync_wid = (mode->crtc_hsync_end - mode->crtc_hsync_start) / 8; + if (!hsync_wid) + hsync_wid = 1; + hsync_start = mode->crtc_hsync_start - 8; + + fp_h_sync_strt_wid = ((hsync_start & 0x1fff) + | ((hsync_wid & 0x3f) << 16) + | ((mode->flags & DRM_MODE_FLAG_NHSYNC) + ? RADEON_CRTC_H_SYNC_POL + : 0)); + + fp_crtc_v_total_disp = (((mode->crtc_vtotal - 1) & 0xffff) + | ((mode->crtc_vdisplay - 1) << 16)); + + vsync_wid = mode->crtc_vsync_end - mode->crtc_vsync_start; + if (!vsync_wid) + vsync_wid = 1; + + fp_v_sync_strt_wid = (((mode->crtc_vsync_start - 1) & 0xfff) + | ((vsync_wid & 0x1f) << 16) + | ((mode->flags & DRM_MODE_FLAG_NVSYNC) + ? RADEON_CRTC_V_SYNC_POL + : 0)); + + fp_horz_vert_active = 0; + + if (native_mode->hdisplay == 0 || + native_mode->vdisplay == 0) { + hscale = false; + vscale = false; + } else { + if (xres > native_mode->hdisplay) + xres = native_mode->hdisplay; + if (yres > native_mode->vdisplay) + yres = native_mode->vdisplay; + + if (xres == native_mode->hdisplay) + hscale = false; + if (yres == native_mode->vdisplay) + vscale = false; + } + + switch (radeon_crtc->rmx_type) { + case RMX_FULL: + case RMX_ASPECT: + if (!hscale) + fp_horz_stretch |= ((xres/8-1) << 16); + else { + inc = (fp_horz_stretch & RADEON_HORZ_AUTO_RATIO_INC) ? 1 : 0; + scale = ((xres + inc) * RADEON_HORZ_STRETCH_RATIO_MAX) + / native_mode->hdisplay + 1; + fp_horz_stretch |= (((scale) & RADEON_HORZ_STRETCH_RATIO_MASK) | + RADEON_HORZ_STRETCH_BLEND | + RADEON_HORZ_STRETCH_ENABLE | + ((native_mode->hdisplay/8-1) << 16)); + } + + if (!vscale) + fp_vert_stretch |= ((yres-1) << 12); + else { + inc = (fp_vert_stretch & RADEON_VERT_AUTO_RATIO_INC) ? 1 : 0; + scale = ((yres + inc) * RADEON_VERT_STRETCH_RATIO_MAX) + / native_mode->vdisplay + 1; + fp_vert_stretch |= (((scale) & RADEON_VERT_STRETCH_RATIO_MASK) | + RADEON_VERT_STRETCH_ENABLE | + RADEON_VERT_STRETCH_BLEND | + ((native_mode->vdisplay-1) << 12)); + } + break; + case RMX_CENTER: + fp_horz_stretch |= ((xres/8-1) << 16); + fp_vert_stretch |= ((yres-1) << 12); + + crtc_more_cntl |= (RADEON_CRTC_AUTO_HORZ_CENTER_EN | + RADEON_CRTC_AUTO_VERT_CENTER_EN); + + blank_width = (mode->crtc_hblank_end - mode->crtc_hblank_start) / 8; + if (blank_width > 110) + blank_width = 110; + + fp_crtc_h_total_disp = (((blank_width) & 0x3ff) + | ((((mode->crtc_hdisplay / 8) - 1) & 0x1ff) << 16)); + + hsync_wid = (mode->crtc_hsync_end - mode->crtc_hsync_start) / 8; + if (!hsync_wid) + hsync_wid = 1; + + fp_h_sync_strt_wid = ((((mode->crtc_hsync_start - mode->crtc_hblank_start) / 8) & 0x1fff) + | ((hsync_wid & 0x3f) << 16) + | ((mode->flags & DRM_MODE_FLAG_NHSYNC) + ? RADEON_CRTC_H_SYNC_POL + : 0)); + + fp_crtc_v_total_disp = (((mode->crtc_vblank_end - mode->crtc_vblank_start) & 0xffff) + | ((mode->crtc_vdisplay - 1) << 16)); + + vsync_wid = mode->crtc_vsync_end - mode->crtc_vsync_start; + if (!vsync_wid) + vsync_wid = 1; + + fp_v_sync_strt_wid = ((((mode->crtc_vsync_start - mode->crtc_vblank_start) & 0xfff) + | ((vsync_wid & 0x1f) << 16) + | ((mode->flags & DRM_MODE_FLAG_NVSYNC) + ? RADEON_CRTC_V_SYNC_POL + : 0))); + + fp_horz_vert_active = (((native_mode->vdisplay) & 0xfff) | + (((native_mode->hdisplay / 8) & 0x1ff) << 16)); + break; + case RMX_OFF: + default: + fp_horz_stretch |= ((xres/8-1) << 16); + fp_vert_stretch |= ((yres-1) << 12); + break; + } + + WREG32(RADEON_FP_HORZ_STRETCH, fp_horz_stretch); + WREG32(RADEON_FP_VERT_STRETCH, fp_vert_stretch); + WREG32(RADEON_CRTC_MORE_CNTL, crtc_more_cntl); + WREG32(RADEON_FP_HORZ_VERT_ACTIVE, fp_horz_vert_active); + WREG32(RADEON_FP_H_SYNC_STRT_WID, fp_h_sync_strt_wid); + WREG32(RADEON_FP_V_SYNC_STRT_WID, fp_v_sync_strt_wid); + WREG32(RADEON_FP_CRTC_H_TOTAL_DISP, fp_crtc_h_total_disp); + WREG32(RADEON_FP_CRTC_V_TOTAL_DISP, fp_crtc_v_total_disp); +} + +void radeon_restore_common_regs(struct drm_device *dev) +{ + /* don't need this yet */ +} + +static void radeon_pll_wait_for_read_update_complete(struct drm_device *dev) +{ + struct radeon_device *rdev = dev->dev_private; + int i = 0; + + /* FIXME: Certain revisions of R300 can't recover here. Not sure of + the cause yet, but this workaround will mask the problem for now. + Other chips usually will pass at the very first test, so the + workaround shouldn't have any effect on them. */ + for (i = 0; + (i < 10000 && + RREG32_PLL(RADEON_PPLL_REF_DIV) & RADEON_PPLL_ATOMIC_UPDATE_R); + i++); +} + +static void radeon_pll_write_update(struct drm_device *dev) +{ + struct radeon_device *rdev = dev->dev_private; + + while (RREG32_PLL(RADEON_PPLL_REF_DIV) & RADEON_PPLL_ATOMIC_UPDATE_R); + + WREG32_PLL_P(RADEON_PPLL_REF_DIV, + RADEON_PPLL_ATOMIC_UPDATE_W, + ~(RADEON_PPLL_ATOMIC_UPDATE_W)); +} + +static void radeon_pll2_wait_for_read_update_complete(struct drm_device *dev) +{ + struct radeon_device *rdev = dev->dev_private; + int i = 0; + + + /* FIXME: Certain revisions of R300 can't recover here. Not sure of + the cause yet, but this workaround will mask the problem for now. + Other chips usually will pass at the very first test, so the + workaround shouldn't have any effect on them. */ + for (i = 0; + (i < 10000 && + RREG32_PLL(RADEON_P2PLL_REF_DIV) & RADEON_P2PLL_ATOMIC_UPDATE_R); + i++); +} + +static void radeon_pll2_write_update(struct drm_device *dev) +{ + struct radeon_device *rdev = dev->dev_private; + + while (RREG32_PLL(RADEON_P2PLL_REF_DIV) & RADEON_P2PLL_ATOMIC_UPDATE_R); + + WREG32_PLL_P(RADEON_P2PLL_REF_DIV, + RADEON_P2PLL_ATOMIC_UPDATE_W, + ~(RADEON_P2PLL_ATOMIC_UPDATE_W)); +} + +static uint8_t radeon_compute_pll_gain(uint16_t ref_freq, uint16_t ref_div, + uint16_t fb_div) +{ + unsigned int vcoFreq; + + if (!ref_div) + return 1; + + vcoFreq = ((unsigned)ref_freq * fb_div) / ref_div; + + /* + * This is horribly crude: the VCO frequency range is divided into + * 3 parts, each part having a fixed PLL gain value. + */ + if (vcoFreq >= 30000) + /* + * [300..max] MHz : 7 + */ + return 7; + else if (vcoFreq >= 18000) + /* + * [180..300) MHz : 4 + */ + return 4; + else + /* + * [0..180) MHz : 1 + */ + return 1; +} + +void radeon_crtc_dpms(struct drm_crtc *crtc, int mode) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + uint32_t mask; + + if (radeon_crtc->crtc_id) + mask = (RADEON_CRTC2_DISP_DIS | + RADEON_CRTC2_VSYNC_DIS | + RADEON_CRTC2_HSYNC_DIS | + RADEON_CRTC2_DISP_REQ_EN_B); + else + mask = (RADEON_CRTC_DISPLAY_DIS | + RADEON_CRTC_VSYNC_DIS | + RADEON_CRTC_HSYNC_DIS); + + switch (mode) { + case DRM_MODE_DPMS_ON: + radeon_crtc->enabled = true; + /* adjust pm to dpms changes BEFORE enabling crtcs */ + radeon_pm_compute_clocks(rdev); + if (radeon_crtc->crtc_id) + WREG32_P(RADEON_CRTC2_GEN_CNTL, RADEON_CRTC2_EN, ~(RADEON_CRTC2_EN | mask)); + else { + WREG32_P(RADEON_CRTC_GEN_CNTL, RADEON_CRTC_EN, ~(RADEON_CRTC_EN | + RADEON_CRTC_DISP_REQ_EN_B)); + WREG32_P(RADEON_CRTC_EXT_CNTL, 0, ~mask); + } + drm_vblank_post_modeset(dev, radeon_crtc->crtc_id); + radeon_crtc_load_lut(crtc); + break; + case DRM_MODE_DPMS_STANDBY: + case DRM_MODE_DPMS_SUSPEND: + case DRM_MODE_DPMS_OFF: + drm_vblank_pre_modeset(dev, radeon_crtc->crtc_id); + if (radeon_crtc->crtc_id) + WREG32_P(RADEON_CRTC2_GEN_CNTL, mask, ~(RADEON_CRTC2_EN | mask)); + else { + WREG32_P(RADEON_CRTC_GEN_CNTL, RADEON_CRTC_DISP_REQ_EN_B, ~(RADEON_CRTC_EN | + RADEON_CRTC_DISP_REQ_EN_B)); + WREG32_P(RADEON_CRTC_EXT_CNTL, mask, ~mask); + } + radeon_crtc->enabled = false; + /* adjust pm to dpms changes AFTER disabling crtcs */ + radeon_pm_compute_clocks(rdev); + break; + } +} + +int radeon_crtc_set_base(struct drm_crtc *crtc, int x, int y, + struct drm_framebuffer *old_fb) +{ + return radeon_crtc_do_set_base(crtc, old_fb, x, y, 0); +} + +int radeon_crtc_set_base_atomic(struct drm_crtc *crtc, + struct drm_framebuffer *fb, + int x, int y, enum mode_set_atomic state) +{ + return radeon_crtc_do_set_base(crtc, fb, x, y, 1); +} + +int radeon_crtc_do_set_base(struct drm_crtc *crtc, + struct drm_framebuffer *fb, + int x, int y, int atomic) +{ + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct radeon_framebuffer *radeon_fb; + struct drm_framebuffer *target_fb; + struct drm_gem_object *obj; + struct radeon_bo *rbo; + uint64_t base; + uint32_t crtc_offset, crtc_offset_cntl, crtc_tile_x0_y0 = 0; + uint32_t crtc_pitch, pitch_pixels; + uint32_t tiling_flags; + int format; + uint32_t gen_cntl_reg, gen_cntl_val; + int r; + + DRM_DEBUG_KMS("\n"); + /* no fb bound */ + if (!atomic && !crtc->fb) { + DRM_DEBUG_KMS("No FB bound\n"); + return 0; + } + + if (atomic) { + radeon_fb = to_radeon_framebuffer(fb); + target_fb = fb; + } + else { + radeon_fb = to_radeon_framebuffer(crtc->fb); + target_fb = crtc->fb; + } + + switch (target_fb->bits_per_pixel) { + case 8: + format = 2; + break; + case 15: /* 555 */ + format = 3; + break; + case 16: /* 565 */ + format = 4; + break; + case 24: /* RGB */ + format = 5; + break; + case 32: /* xRGB */ + format = 6; + break; + default: + return false; + } + + /* Pin framebuffer & get tilling informations */ + obj = radeon_fb->obj; + rbo = gem_to_radeon_bo(obj); + r = radeon_bo_reserve(rbo, false); + if (unlikely(r != 0)) + return r; + r = radeon_bo_pin(rbo, RADEON_GEM_DOMAIN_VRAM, &base); + if (unlikely(r != 0)) { + radeon_bo_unreserve(rbo); + return -EINVAL; + } + radeon_bo_get_tiling_flags(rbo, &tiling_flags, NULL); + radeon_bo_unreserve(rbo); + if (tiling_flags & RADEON_TILING_MICRO) + DRM_ERROR("trying to scanout microtiled buffer\n"); + + /* if scanout was in GTT this really wouldn't work */ + /* crtc offset is from display base addr not FB location */ + radeon_crtc->legacy_display_base_addr = rdev->mc.vram_start; + + base -= radeon_crtc->legacy_display_base_addr; + + crtc_offset_cntl = 0; + + pitch_pixels = target_fb->pitch / (target_fb->bits_per_pixel / 8); + crtc_pitch = (((pitch_pixels * target_fb->bits_per_pixel) + + ((target_fb->bits_per_pixel * 8) - 1)) / + (target_fb->bits_per_pixel * 8)); + crtc_pitch |= crtc_pitch << 16; + + crtc_offset_cntl |= RADEON_CRTC_GUI_TRIG_OFFSET_LEFT_EN; + if (tiling_flags & RADEON_TILING_MACRO) { + if (ASIC_IS_R300(rdev)) + crtc_offset_cntl |= (R300_CRTC_X_Y_MODE_EN | + R300_CRTC_MICRO_TILE_BUFFER_DIS | + R300_CRTC_MACRO_TILE_EN); + else + crtc_offset_cntl |= RADEON_CRTC_TILE_EN; + } else { + if (ASIC_IS_R300(rdev)) + crtc_offset_cntl &= ~(R300_CRTC_X_Y_MODE_EN | + R300_CRTC_MICRO_TILE_BUFFER_DIS | + R300_CRTC_MACRO_TILE_EN); + else + crtc_offset_cntl &= ~RADEON_CRTC_TILE_EN; + } + + if (tiling_flags & RADEON_TILING_MACRO) { + if (ASIC_IS_R300(rdev)) { + crtc_tile_x0_y0 = x | (y << 16); + base &= ~0x7ff; + } else { + int byteshift = target_fb->bits_per_pixel >> 4; + int tile_addr = (((y >> 3) * pitch_pixels + x) >> (8 - byteshift)) << 11; + base += tile_addr + ((x << byteshift) % 256) + ((y % 8) << 8); + crtc_offset_cntl |= (y % 16); + } + } else { + int offset = y * pitch_pixels + x; + switch (target_fb->bits_per_pixel) { + case 8: + offset *= 1; + break; + case 15: + case 16: + offset *= 2; + break; + case 24: + offset *= 3; + break; + case 32: + offset *= 4; + break; + default: + return false; + } + base += offset; + } + + base &= ~7; + + if (radeon_crtc->crtc_id == 1) + gen_cntl_reg = RADEON_CRTC2_GEN_CNTL; + else + gen_cntl_reg = RADEON_CRTC_GEN_CNTL; + + gen_cntl_val = RREG32(gen_cntl_reg); + gen_cntl_val &= ~(0xf << 8); + gen_cntl_val |= (format << 8); + gen_cntl_val &= ~RADEON_CRTC_VSTAT_MODE_MASK; + WREG32(gen_cntl_reg, gen_cntl_val); + + crtc_offset = (u32)base; + + WREG32(RADEON_DISPLAY_BASE_ADDR + radeon_crtc->crtc_offset, radeon_crtc->legacy_display_base_addr); + + if (ASIC_IS_R300(rdev)) { + if (radeon_crtc->crtc_id) + WREG32(R300_CRTC2_TILE_X0_Y0, crtc_tile_x0_y0); + else + WREG32(R300_CRTC_TILE_X0_Y0, crtc_tile_x0_y0); + } + WREG32(RADEON_CRTC_OFFSET_CNTL + radeon_crtc->crtc_offset, crtc_offset_cntl); + WREG32(RADEON_CRTC_OFFSET + radeon_crtc->crtc_offset, crtc_offset); + WREG32(RADEON_CRTC_PITCH + radeon_crtc->crtc_offset, crtc_pitch); + + if (!atomic && fb && fb != crtc->fb) { + radeon_fb = to_radeon_framebuffer(fb); + rbo = gem_to_radeon_bo(radeon_fb->obj); + r = radeon_bo_reserve(rbo, false); + if (unlikely(r != 0)) + return r; + radeon_bo_unpin(rbo); + radeon_bo_unreserve(rbo); + } + + /* Bytes per pixel may have changed */ + radeon_bandwidth_update(rdev); + + return 0; +} + +static bool radeon_set_crtc_timing(struct drm_crtc *crtc, struct drm_display_mode *mode) +{ + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct drm_encoder *encoder; + int format; + int hsync_start; + int hsync_wid; + int vsync_wid; + uint32_t crtc_h_total_disp; + uint32_t crtc_h_sync_strt_wid; + uint32_t crtc_v_total_disp; + uint32_t crtc_v_sync_strt_wid; + bool is_tv = false; + + DRM_DEBUG_KMS("\n"); + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + if (encoder->crtc == crtc) { + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + if (radeon_encoder->active_device & ATOM_DEVICE_TV_SUPPORT) { + is_tv = true; + DRM_INFO("crtc %d is connected to a TV\n", radeon_crtc->crtc_id); + break; + } + } + } + + switch (crtc->fb->bits_per_pixel) { + case 8: + format = 2; + break; + case 15: /* 555 */ + format = 3; + break; + case 16: /* 565 */ + format = 4; + break; + case 24: /* RGB */ + format = 5; + break; + case 32: /* xRGB */ + format = 6; + break; + default: + return false; + } + + crtc_h_total_disp = ((((mode->crtc_htotal / 8) - 1) & 0x3ff) + | ((((mode->crtc_hdisplay / 8) - 1) & 0x1ff) << 16)); + + hsync_wid = (mode->crtc_hsync_end - mode->crtc_hsync_start) / 8; + if (!hsync_wid) + hsync_wid = 1; + hsync_start = mode->crtc_hsync_start - 8; + + crtc_h_sync_strt_wid = ((hsync_start & 0x1fff) + | ((hsync_wid & 0x3f) << 16) + | ((mode->flags & DRM_MODE_FLAG_NHSYNC) + ? RADEON_CRTC_H_SYNC_POL + : 0)); + + /* This works for double scan mode. */ + crtc_v_total_disp = (((mode->crtc_vtotal - 1) & 0xffff) + | ((mode->crtc_vdisplay - 1) << 16)); + + vsync_wid = mode->crtc_vsync_end - mode->crtc_vsync_start; + if (!vsync_wid) + vsync_wid = 1; + + crtc_v_sync_strt_wid = (((mode->crtc_vsync_start - 1) & 0xfff) + | ((vsync_wid & 0x1f) << 16) + | ((mode->flags & DRM_MODE_FLAG_NVSYNC) + ? RADEON_CRTC_V_SYNC_POL + : 0)); + + if (radeon_crtc->crtc_id) { + uint32_t crtc2_gen_cntl; + uint32_t disp2_merge_cntl; + + /* if TV DAC is enabled for another crtc and keep it enabled */ + crtc2_gen_cntl = RREG32(RADEON_CRTC2_GEN_CNTL) & 0x00718080; + crtc2_gen_cntl |= ((format << 8) + | RADEON_CRTC2_VSYNC_DIS + | RADEON_CRTC2_HSYNC_DIS + | RADEON_CRTC2_DISP_DIS + | RADEON_CRTC2_DISP_REQ_EN_B + | ((mode->flags & DRM_MODE_FLAG_DBLSCAN) + ? RADEON_CRTC2_DBL_SCAN_EN + : 0) + | ((mode->flags & DRM_MODE_FLAG_CSYNC) + ? RADEON_CRTC2_CSYNC_EN + : 0) + | ((mode->flags & DRM_MODE_FLAG_INTERLACE) + ? RADEON_CRTC2_INTERLACE_EN + : 0)); + + /* rs4xx chips seem to like to have the crtc enabled when the timing is set */ + if ((rdev->family == CHIP_RS400) || (rdev->family == CHIP_RS480)) + crtc2_gen_cntl |= RADEON_CRTC2_EN; + + disp2_merge_cntl = RREG32(RADEON_DISP2_MERGE_CNTL); + disp2_merge_cntl &= ~RADEON_DISP2_RGB_OFFSET_EN; + + WREG32(RADEON_DISP2_MERGE_CNTL, disp2_merge_cntl); + WREG32(RADEON_CRTC2_GEN_CNTL, crtc2_gen_cntl); + + WREG32(RADEON_FP_H2_SYNC_STRT_WID, crtc_h_sync_strt_wid); + WREG32(RADEON_FP_V2_SYNC_STRT_WID, crtc_v_sync_strt_wid); + } else { + uint32_t crtc_gen_cntl; + uint32_t crtc_ext_cntl; + uint32_t disp_merge_cntl; + + crtc_gen_cntl = RREG32(RADEON_CRTC_GEN_CNTL) & 0x00718000; + crtc_gen_cntl |= (RADEON_CRTC_EXT_DISP_EN + | (format << 8) + | RADEON_CRTC_DISP_REQ_EN_B + | ((mode->flags & DRM_MODE_FLAG_DBLSCAN) + ? RADEON_CRTC_DBL_SCAN_EN + : 0) + | ((mode->flags & DRM_MODE_FLAG_CSYNC) + ? RADEON_CRTC_CSYNC_EN + : 0) + | ((mode->flags & DRM_MODE_FLAG_INTERLACE) + ? RADEON_CRTC_INTERLACE_EN + : 0)); + + /* rs4xx chips seem to like to have the crtc enabled when the timing is set */ + if ((rdev->family == CHIP_RS400) || (rdev->family == CHIP_RS480)) + crtc_gen_cntl |= RADEON_CRTC_EN; + + crtc_ext_cntl = RREG32(RADEON_CRTC_EXT_CNTL); + crtc_ext_cntl |= (RADEON_XCRT_CNT_EN | + RADEON_CRTC_VSYNC_DIS | + RADEON_CRTC_HSYNC_DIS | + RADEON_CRTC_DISPLAY_DIS); + + disp_merge_cntl = RREG32(RADEON_DISP_MERGE_CNTL); + disp_merge_cntl &= ~RADEON_DISP_RGB_OFFSET_EN; + + WREG32(RADEON_DISP_MERGE_CNTL, disp_merge_cntl); + WREG32(RADEON_CRTC_GEN_CNTL, crtc_gen_cntl); + WREG32(RADEON_CRTC_EXT_CNTL, crtc_ext_cntl); + } + + if (is_tv) + radeon_legacy_tv_adjust_crtc_reg(encoder, &crtc_h_total_disp, + &crtc_h_sync_strt_wid, &crtc_v_total_disp, + &crtc_v_sync_strt_wid); + + WREG32(RADEON_CRTC_H_TOTAL_DISP + radeon_crtc->crtc_offset, crtc_h_total_disp); + WREG32(RADEON_CRTC_H_SYNC_STRT_WID + radeon_crtc->crtc_offset, crtc_h_sync_strt_wid); + WREG32(RADEON_CRTC_V_TOTAL_DISP + radeon_crtc->crtc_offset, crtc_v_total_disp); + WREG32(RADEON_CRTC_V_SYNC_STRT_WID + radeon_crtc->crtc_offset, crtc_v_sync_strt_wid); + + return true; +} + +static void radeon_set_pll(struct drm_crtc *crtc, struct drm_display_mode *mode) +{ + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct drm_encoder *encoder; + uint32_t feedback_div = 0; + uint32_t frac_fb_div = 0; + uint32_t reference_div = 0; + uint32_t post_divider = 0; + uint32_t freq = 0; + uint8_t pll_gain; + bool use_bios_divs = false; + /* PLL registers */ + uint32_t pll_ref_div = 0; + uint32_t pll_fb_post_div = 0; + uint32_t htotal_cntl = 0; + bool is_tv = false; + struct radeon_pll *pll; + + struct { + int divider; + int bitvalue; + } *post_div, post_divs[] = { + /* From RAGE 128 VR/RAGE 128 GL Register + * Reference Manual (Technical Reference + * Manual P/N RRG-G04100-C Rev. 0.04), page + * 3-17 (PLL_DIV_[3:0]). + */ + { 1, 0 }, /* VCLK_SRC */ + { 2, 1 }, /* VCLK_SRC/2 */ + { 4, 2 }, /* VCLK_SRC/4 */ + { 8, 3 }, /* VCLK_SRC/8 */ + { 3, 4 }, /* VCLK_SRC/3 */ + { 16, 5 }, /* VCLK_SRC/16 */ + { 6, 6 }, /* VCLK_SRC/6 */ + { 12, 7 }, /* VCLK_SRC/12 */ + { 0, 0 } + }; + + if (radeon_crtc->crtc_id) + pll = &rdev->clock.p2pll; + else + pll = &rdev->clock.p1pll; + + pll->flags = RADEON_PLL_LEGACY; + + if (mode->clock > 200000) /* range limits??? */ + pll->flags |= RADEON_PLL_PREFER_HIGH_FB_DIV; + else + pll->flags |= RADEON_PLL_PREFER_LOW_REF_DIV; + + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + if (encoder->crtc == crtc) { + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + + if (radeon_encoder->active_device & ATOM_DEVICE_TV_SUPPORT) { + is_tv = true; + break; + } + + if (encoder->encoder_type != DRM_MODE_ENCODER_DAC) + pll->flags |= RADEON_PLL_NO_ODD_POST_DIV; + if (encoder->encoder_type == DRM_MODE_ENCODER_LVDS) { + if (!rdev->is_atom_bios) { + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct radeon_encoder_lvds *lvds = (struct radeon_encoder_lvds *)radeon_encoder->enc_priv; + if (lvds) { + if (lvds->use_bios_dividers) { + pll_ref_div = lvds->panel_ref_divider; + pll_fb_post_div = (lvds->panel_fb_divider | + (lvds->panel_post_divider << 16)); + htotal_cntl = 0; + use_bios_divs = true; + } + } + } + pll->flags |= RADEON_PLL_USE_REF_DIV; + } + } + } + + DRM_DEBUG_KMS("\n"); + + if (!use_bios_divs) { + radeon_compute_pll_legacy(pll, mode->clock, + &freq, &feedback_div, &frac_fb_div, + &reference_div, &post_divider); + + for (post_div = &post_divs[0]; post_div->divider; ++post_div) { + if (post_div->divider == post_divider) + break; + } + + if (!post_div->divider) + post_div = &post_divs[0]; + + DRM_DEBUG_KMS("dc=%u, fd=%d, rd=%d, pd=%d\n", + (unsigned)freq, + feedback_div, + reference_div, + post_divider); + + pll_ref_div = reference_div; +#if defined(__powerpc__) && (0) /* TODO */ + /* apparently programming this otherwise causes a hang??? */ + if (info->MacModel == RADEON_MAC_IBOOK) + pll_fb_post_div = 0x000600ad; + else +#endif + pll_fb_post_div = (feedback_div | (post_div->bitvalue << 16)); + + htotal_cntl = mode->htotal & 0x7; + + } + + pll_gain = radeon_compute_pll_gain(pll->reference_freq, + pll_ref_div & 0x3ff, + pll_fb_post_div & 0x7ff); + + if (radeon_crtc->crtc_id) { + uint32_t pixclks_cntl = ((RREG32_PLL(RADEON_PIXCLKS_CNTL) & + ~(RADEON_PIX2CLK_SRC_SEL_MASK)) | + RADEON_PIX2CLK_SRC_SEL_P2PLLCLK); + + if (is_tv) { + radeon_legacy_tv_adjust_pll2(encoder, &htotal_cntl, + &pll_ref_div, &pll_fb_post_div, + &pixclks_cntl); + } + + WREG32_PLL_P(RADEON_PIXCLKS_CNTL, + RADEON_PIX2CLK_SRC_SEL_CPUCLK, + ~(RADEON_PIX2CLK_SRC_SEL_MASK)); + + WREG32_PLL_P(RADEON_P2PLL_CNTL, + RADEON_P2PLL_RESET + | RADEON_P2PLL_ATOMIC_UPDATE_EN + | ((uint32_t)pll_gain << RADEON_P2PLL_PVG_SHIFT), + ~(RADEON_P2PLL_RESET + | RADEON_P2PLL_ATOMIC_UPDATE_EN + | RADEON_P2PLL_PVG_MASK)); + + WREG32_PLL_P(RADEON_P2PLL_REF_DIV, + pll_ref_div, + ~RADEON_P2PLL_REF_DIV_MASK); + + WREG32_PLL_P(RADEON_P2PLL_DIV_0, + pll_fb_post_div, + ~RADEON_P2PLL_FB0_DIV_MASK); + + WREG32_PLL_P(RADEON_P2PLL_DIV_0, + pll_fb_post_div, + ~RADEON_P2PLL_POST0_DIV_MASK); + + radeon_pll2_write_update(dev); + radeon_pll2_wait_for_read_update_complete(dev); + + WREG32_PLL(RADEON_HTOTAL2_CNTL, htotal_cntl); + + WREG32_PLL_P(RADEON_P2PLL_CNTL, + 0, + ~(RADEON_P2PLL_RESET + | RADEON_P2PLL_SLEEP + | RADEON_P2PLL_ATOMIC_UPDATE_EN)); + + DRM_DEBUG_KMS("Wrote2: 0x%08x 0x%08x 0x%08x (0x%08x)\n", + (unsigned)pll_ref_div, + (unsigned)pll_fb_post_div, + (unsigned)htotal_cntl, + RREG32_PLL(RADEON_P2PLL_CNTL)); + DRM_DEBUG_KMS("Wrote2: rd=%u, fd=%u, pd=%u\n", + (unsigned)pll_ref_div & RADEON_P2PLL_REF_DIV_MASK, + (unsigned)pll_fb_post_div & RADEON_P2PLL_FB0_DIV_MASK, + (unsigned)((pll_fb_post_div & + RADEON_P2PLL_POST0_DIV_MASK) >> 16)); + + mdelay(50); /* Let the clock to lock */ + + WREG32_PLL_P(RADEON_PIXCLKS_CNTL, + RADEON_PIX2CLK_SRC_SEL_P2PLLCLK, + ~(RADEON_PIX2CLK_SRC_SEL_MASK)); + + WREG32_PLL(RADEON_PIXCLKS_CNTL, pixclks_cntl); + } else { + uint32_t pixclks_cntl; + + + if (is_tv) { + pixclks_cntl = RREG32_PLL(RADEON_PIXCLKS_CNTL); + radeon_legacy_tv_adjust_pll1(encoder, &htotal_cntl, &pll_ref_div, + &pll_fb_post_div, &pixclks_cntl); + } + + if (rdev->flags & RADEON_IS_MOBILITY) { + /* A temporal workaround for the occasional blanking on certain laptop panels. + This appears to related to the PLL divider registers (fail to lock?). + It occurs even when all dividers are the same with their old settings. + In this case we really don't need to fiddle with PLL registers. + By doing this we can avoid the blanking problem with some panels. + */ + if ((pll_ref_div == (RREG32_PLL(RADEON_PPLL_REF_DIV) & RADEON_PPLL_REF_DIV_MASK)) && + (pll_fb_post_div == (RREG32_PLL(RADEON_PPLL_DIV_3) & + (RADEON_PPLL_POST3_DIV_MASK | RADEON_PPLL_FB3_DIV_MASK)))) { + WREG32_P(RADEON_CLOCK_CNTL_INDEX, + RADEON_PLL_DIV_SEL, + ~(RADEON_PLL_DIV_SEL)); + r100_pll_errata_after_index(rdev); + return; + } + } + + WREG32_PLL_P(RADEON_VCLK_ECP_CNTL, + RADEON_VCLK_SRC_SEL_CPUCLK, + ~(RADEON_VCLK_SRC_SEL_MASK)); + WREG32_PLL_P(RADEON_PPLL_CNTL, + RADEON_PPLL_RESET + | RADEON_PPLL_ATOMIC_UPDATE_EN + | RADEON_PPLL_VGA_ATOMIC_UPDATE_EN + | ((uint32_t)pll_gain << RADEON_PPLL_PVG_SHIFT), + ~(RADEON_PPLL_RESET + | RADEON_PPLL_ATOMIC_UPDATE_EN + | RADEON_PPLL_VGA_ATOMIC_UPDATE_EN + | RADEON_PPLL_PVG_MASK)); + + WREG32_P(RADEON_CLOCK_CNTL_INDEX, + RADEON_PLL_DIV_SEL, + ~(RADEON_PLL_DIV_SEL)); + r100_pll_errata_after_index(rdev); + + if (ASIC_IS_R300(rdev) || + (rdev->family == CHIP_RS300) || + (rdev->family == CHIP_RS400) || + (rdev->family == CHIP_RS480)) { + if (pll_ref_div & R300_PPLL_REF_DIV_ACC_MASK) { + /* When restoring console mode, use saved PPLL_REF_DIV + * setting. + */ + WREG32_PLL_P(RADEON_PPLL_REF_DIV, + pll_ref_div, + 0); + } else { + /* R300 uses ref_div_acc field as real ref divider */ + WREG32_PLL_P(RADEON_PPLL_REF_DIV, + (pll_ref_div << R300_PPLL_REF_DIV_ACC_SHIFT), + ~R300_PPLL_REF_DIV_ACC_MASK); + } + } else + WREG32_PLL_P(RADEON_PPLL_REF_DIV, + pll_ref_div, + ~RADEON_PPLL_REF_DIV_MASK); + + WREG32_PLL_P(RADEON_PPLL_DIV_3, + pll_fb_post_div, + ~RADEON_PPLL_FB3_DIV_MASK); + + WREG32_PLL_P(RADEON_PPLL_DIV_3, + pll_fb_post_div, + ~RADEON_PPLL_POST3_DIV_MASK); + + radeon_pll_write_update(dev); + radeon_pll_wait_for_read_update_complete(dev); + + WREG32_PLL(RADEON_HTOTAL_CNTL, htotal_cntl); + + WREG32_PLL_P(RADEON_PPLL_CNTL, + 0, + ~(RADEON_PPLL_RESET + | RADEON_PPLL_SLEEP + | RADEON_PPLL_ATOMIC_UPDATE_EN + | RADEON_PPLL_VGA_ATOMIC_UPDATE_EN)); + + DRM_DEBUG_KMS("Wrote: 0x%08x 0x%08x 0x%08x (0x%08x)\n", + pll_ref_div, + pll_fb_post_div, + (unsigned)htotal_cntl, + RREG32_PLL(RADEON_PPLL_CNTL)); + DRM_DEBUG_KMS("Wrote: rd=%d, fd=%d, pd=%d\n", + pll_ref_div & RADEON_PPLL_REF_DIV_MASK, + pll_fb_post_div & RADEON_PPLL_FB3_DIV_MASK, + (pll_fb_post_div & RADEON_PPLL_POST3_DIV_MASK) >> 16); + + mdelay(50); /* Let the clock to lock */ + + WREG32_PLL_P(RADEON_VCLK_ECP_CNTL, + RADEON_VCLK_SRC_SEL_PPLLCLK, + ~(RADEON_VCLK_SRC_SEL_MASK)); + + if (is_tv) + WREG32_PLL(RADEON_PIXCLKS_CNTL, pixclks_cntl); + } +} + +static bool radeon_crtc_mode_fixup(struct drm_crtc *crtc, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + + /* adjust pm to upcoming mode change */ + radeon_pm_compute_clocks(rdev); + + if (!radeon_crtc_scaling_mode_fixup(crtc, mode, adjusted_mode)) + return false; + return true; +} + +static int radeon_crtc_mode_set(struct drm_crtc *crtc, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode, + int x, int y, struct drm_framebuffer *old_fb) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + + /* TODO TV */ + radeon_crtc_set_base(crtc, x, y, old_fb); + radeon_set_crtc_timing(crtc, adjusted_mode); + radeon_set_pll(crtc, adjusted_mode); + radeon_overscan_setup(crtc, adjusted_mode); + if (radeon_crtc->crtc_id == 0) { + radeon_legacy_rmx_mode_set(crtc, adjusted_mode); + } else { + if (radeon_crtc->rmx_type != RMX_OFF) { + /* FIXME: only first crtc has rmx what should we + * do ? + */ + DRM_ERROR("Mode need scaling but only first crtc can do that.\n"); + } + } + return 0; +} + +static void radeon_crtc_prepare(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + struct drm_crtc *crtci; + + /* + * The hardware wedges sometimes if you reconfigure one CRTC + * whilst another is running (see fdo bug #24611). + */ + list_for_each_entry(crtci, &dev->mode_config.crtc_list, head) + radeon_crtc_dpms(crtci, DRM_MODE_DPMS_OFF); +} + +static void radeon_crtc_commit(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + struct drm_crtc *crtci; + + /* + * Reenable the CRTCs that should be running. + */ + list_for_each_entry(crtci, &dev->mode_config.crtc_list, head) { + if (crtci->enabled) + radeon_crtc_dpms(crtci, DRM_MODE_DPMS_ON); + } +} + +static const struct drm_crtc_helper_funcs legacy_helper_funcs = { + .dpms = radeon_crtc_dpms, + .mode_fixup = radeon_crtc_mode_fixup, + .mode_set = radeon_crtc_mode_set, + .mode_set_base = radeon_crtc_set_base, + .mode_set_base_atomic = radeon_crtc_set_base_atomic, + .prepare = radeon_crtc_prepare, + .commit = radeon_crtc_commit, + .load_lut = radeon_crtc_load_lut, +}; + + +void radeon_legacy_init_crtc(struct drm_device *dev, + struct radeon_crtc *radeon_crtc) +{ + if (radeon_crtc->crtc_id == 1) + radeon_crtc->crtc_offset = RADEON_CRTC2_H_TOTAL_DISP - RADEON_CRTC_H_TOTAL_DISP; + drm_crtc_helper_add(&radeon_crtc->base, &legacy_helper_funcs); +} diff --git a/drivers/gpu/drm/radeon/radeon_legacy_encoders.c b/drivers/gpu/drm/radeon/radeon_legacy_encoders.c new file mode 100644 index 00000000..2f46e0c8 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_legacy_encoders.c @@ -0,0 +1,1666 @@ +/* + * Copyright 2007-8 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + */ +#include "drmP.h" +#include "drm_crtc_helper.h" +#include "radeon_drm.h" +#include "radeon.h" +#include "atom.h" +#include +#ifdef CONFIG_PMAC_BACKLIGHT +#include +#endif + +static void radeon_legacy_encoder_disable(struct drm_encoder *encoder) +{ + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct drm_encoder_helper_funcs *encoder_funcs; + + encoder_funcs = encoder->helper_private; + encoder_funcs->dpms(encoder, DRM_MODE_DPMS_OFF); + radeon_encoder->active_device = 0; +} + +static void radeon_legacy_lvds_update(struct drm_encoder *encoder, int mode) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + uint32_t lvds_gen_cntl, lvds_pll_cntl, pixclks_cntl, disp_pwr_man; + int panel_pwr_delay = 2000; + bool is_mac = false; + uint8_t backlight_level; + DRM_DEBUG_KMS("\n"); + + lvds_gen_cntl = RREG32(RADEON_LVDS_GEN_CNTL); + backlight_level = (lvds_gen_cntl >> RADEON_LVDS_BL_MOD_LEVEL_SHIFT) & 0xff; + + if (radeon_encoder->enc_priv) { + if (rdev->is_atom_bios) { + struct radeon_encoder_atom_dig *lvds = radeon_encoder->enc_priv; + panel_pwr_delay = lvds->panel_pwr_delay; + if (lvds->bl_dev) + backlight_level = lvds->backlight_level; + } else { + struct radeon_encoder_lvds *lvds = radeon_encoder->enc_priv; + panel_pwr_delay = lvds->panel_pwr_delay; + if (lvds->bl_dev) + backlight_level = lvds->backlight_level; + } + } + + /* macs (and possibly some x86 oem systems?) wire up LVDS strangely + * Taken from radeonfb. + */ + if ((rdev->mode_info.connector_table == CT_IBOOK) || + (rdev->mode_info.connector_table == CT_POWERBOOK_EXTERNAL) || + (rdev->mode_info.connector_table == CT_POWERBOOK_INTERNAL) || + (rdev->mode_info.connector_table == CT_POWERBOOK_VGA)) + is_mac = true; + + switch (mode) { + case DRM_MODE_DPMS_ON: + disp_pwr_man = RREG32(RADEON_DISP_PWR_MAN); + disp_pwr_man |= RADEON_AUTO_PWRUP_EN; + WREG32(RADEON_DISP_PWR_MAN, disp_pwr_man); + lvds_pll_cntl = RREG32(RADEON_LVDS_PLL_CNTL); + lvds_pll_cntl |= RADEON_LVDS_PLL_EN; + WREG32(RADEON_LVDS_PLL_CNTL, lvds_pll_cntl); + udelay(1000); + + lvds_pll_cntl = RREG32(RADEON_LVDS_PLL_CNTL); + lvds_pll_cntl &= ~RADEON_LVDS_PLL_RESET; + WREG32(RADEON_LVDS_PLL_CNTL, lvds_pll_cntl); + + lvds_gen_cntl &= ~(RADEON_LVDS_DISPLAY_DIS | + RADEON_LVDS_BL_MOD_LEVEL_MASK); + lvds_gen_cntl |= (RADEON_LVDS_ON | RADEON_LVDS_EN | + RADEON_LVDS_DIGON | RADEON_LVDS_BLON | + (backlight_level << RADEON_LVDS_BL_MOD_LEVEL_SHIFT)); + if (is_mac) + lvds_gen_cntl |= RADEON_LVDS_BL_MOD_EN; + udelay(panel_pwr_delay * 1000); + WREG32(RADEON_LVDS_GEN_CNTL, lvds_gen_cntl); + break; + case DRM_MODE_DPMS_STANDBY: + case DRM_MODE_DPMS_SUSPEND: + case DRM_MODE_DPMS_OFF: + pixclks_cntl = RREG32_PLL(RADEON_PIXCLKS_CNTL); + WREG32_PLL_P(RADEON_PIXCLKS_CNTL, 0, ~RADEON_PIXCLK_LVDS_ALWAYS_ONb); + lvds_gen_cntl |= RADEON_LVDS_DISPLAY_DIS; + if (is_mac) { + lvds_gen_cntl &= ~RADEON_LVDS_BL_MOD_EN; + WREG32(RADEON_LVDS_GEN_CNTL, lvds_gen_cntl); + lvds_gen_cntl &= ~(RADEON_LVDS_ON | RADEON_LVDS_EN); + } else { + WREG32(RADEON_LVDS_GEN_CNTL, lvds_gen_cntl); + lvds_gen_cntl &= ~(RADEON_LVDS_ON | RADEON_LVDS_BLON | RADEON_LVDS_EN | RADEON_LVDS_DIGON); + } + udelay(panel_pwr_delay * 1000); + WREG32(RADEON_LVDS_GEN_CNTL, lvds_gen_cntl); + WREG32_PLL(RADEON_PIXCLKS_CNTL, pixclks_cntl); + udelay(panel_pwr_delay * 1000); + break; + } + + if (rdev->is_atom_bios) + radeon_atombios_encoder_dpms_scratch_regs(encoder, (mode == DRM_MODE_DPMS_ON) ? true : false); + else + radeon_combios_encoder_dpms_scratch_regs(encoder, (mode == DRM_MODE_DPMS_ON) ? true : false); + +} + +static void radeon_legacy_lvds_dpms(struct drm_encoder *encoder, int mode) +{ + struct radeon_device *rdev = encoder->dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + DRM_DEBUG("\n"); + + if (radeon_encoder->enc_priv) { + if (rdev->is_atom_bios) { + struct radeon_encoder_atom_dig *lvds = radeon_encoder->enc_priv; + lvds->dpms_mode = mode; + } else { + struct radeon_encoder_lvds *lvds = radeon_encoder->enc_priv; + lvds->dpms_mode = mode; + } + } + + radeon_legacy_lvds_update(encoder, mode); +} + +static void radeon_legacy_lvds_prepare(struct drm_encoder *encoder) +{ + struct radeon_device *rdev = encoder->dev->dev_private; + + if (rdev->is_atom_bios) + radeon_atom_output_lock(encoder, true); + else + radeon_combios_output_lock(encoder, true); + radeon_legacy_lvds_dpms(encoder, DRM_MODE_DPMS_OFF); +} + +static void radeon_legacy_lvds_commit(struct drm_encoder *encoder) +{ + struct radeon_device *rdev = encoder->dev->dev_private; + + radeon_legacy_lvds_dpms(encoder, DRM_MODE_DPMS_ON); + if (rdev->is_atom_bios) + radeon_atom_output_lock(encoder, false); + else + radeon_combios_output_lock(encoder, false); +} + +static void radeon_legacy_lvds_mode_set(struct drm_encoder *encoder, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_crtc *radeon_crtc = to_radeon_crtc(encoder->crtc); + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + uint32_t lvds_pll_cntl, lvds_gen_cntl, lvds_ss_gen_cntl; + + DRM_DEBUG_KMS("\n"); + + lvds_pll_cntl = RREG32(RADEON_LVDS_PLL_CNTL); + lvds_pll_cntl &= ~RADEON_LVDS_PLL_EN; + + lvds_ss_gen_cntl = RREG32(RADEON_LVDS_SS_GEN_CNTL); + if (rdev->is_atom_bios) { + /* LVDS_GEN_CNTL parameters are computed in LVDSEncoderControl + * need to call that on resume to set up the reg properly. + */ + radeon_encoder->pixel_clock = adjusted_mode->clock; + atombios_digital_setup(encoder, PANEL_ENCODER_ACTION_ENABLE); + lvds_gen_cntl = RREG32(RADEON_LVDS_GEN_CNTL); + } else { + struct radeon_encoder_lvds *lvds = (struct radeon_encoder_lvds *)radeon_encoder->enc_priv; + if (lvds) { + DRM_DEBUG_KMS("bios LVDS_GEN_CNTL: 0x%x\n", lvds->lvds_gen_cntl); + lvds_gen_cntl = lvds->lvds_gen_cntl; + lvds_ss_gen_cntl &= ~((0xf << RADEON_LVDS_PWRSEQ_DELAY1_SHIFT) | + (0xf << RADEON_LVDS_PWRSEQ_DELAY2_SHIFT)); + lvds_ss_gen_cntl |= ((lvds->panel_digon_delay << RADEON_LVDS_PWRSEQ_DELAY1_SHIFT) | + (lvds->panel_blon_delay << RADEON_LVDS_PWRSEQ_DELAY2_SHIFT)); + } else + lvds_gen_cntl = RREG32(RADEON_LVDS_GEN_CNTL); + } + lvds_gen_cntl |= RADEON_LVDS_DISPLAY_DIS; + lvds_gen_cntl &= ~(RADEON_LVDS_ON | + RADEON_LVDS_BLON | + RADEON_LVDS_EN | + RADEON_LVDS_RST_FM); + + if (ASIC_IS_R300(rdev)) + lvds_pll_cntl &= ~(R300_LVDS_SRC_SEL_MASK); + + if (radeon_crtc->crtc_id == 0) { + if (ASIC_IS_R300(rdev)) { + if (radeon_encoder->rmx_type != RMX_OFF) + lvds_pll_cntl |= R300_LVDS_SRC_SEL_RMX; + } else + lvds_gen_cntl &= ~RADEON_LVDS_SEL_CRTC2; + } else { + if (ASIC_IS_R300(rdev)) + lvds_pll_cntl |= R300_LVDS_SRC_SEL_CRTC2; + else + lvds_gen_cntl |= RADEON_LVDS_SEL_CRTC2; + } + + WREG32(RADEON_LVDS_GEN_CNTL, lvds_gen_cntl); + WREG32(RADEON_LVDS_PLL_CNTL, lvds_pll_cntl); + WREG32(RADEON_LVDS_SS_GEN_CNTL, lvds_ss_gen_cntl); + + if (rdev->family == CHIP_RV410) + WREG32(RADEON_CLOCK_CNTL_INDEX, 0); + + if (rdev->is_atom_bios) + radeon_atombios_encoder_crtc_scratch_regs(encoder, radeon_crtc->crtc_id); + else + radeon_combios_encoder_crtc_scratch_regs(encoder, radeon_crtc->crtc_id); +} + +static bool radeon_legacy_mode_fixup(struct drm_encoder *encoder, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + + /* set the active encoder to connector routing */ + radeon_encoder_set_active_device(encoder); + drm_mode_set_crtcinfo(adjusted_mode, 0); + + /* get the native mode for LVDS */ + if (radeon_encoder->active_device & (ATOM_DEVICE_LCD_SUPPORT)) + radeon_panel_mode_fixup(encoder, adjusted_mode); + + return true; +} + +static const struct drm_encoder_helper_funcs radeon_legacy_lvds_helper_funcs = { + .dpms = radeon_legacy_lvds_dpms, + .mode_fixup = radeon_legacy_mode_fixup, + .prepare = radeon_legacy_lvds_prepare, + .mode_set = radeon_legacy_lvds_mode_set, + .commit = radeon_legacy_lvds_commit, + .disable = radeon_legacy_encoder_disable, +}; + +#if defined(CONFIG_BACKLIGHT_CLASS_DEVICE) || defined(CONFIG_BACKLIGHT_CLASS_DEVICE_MODULE) + +#define MAX_RADEON_LEVEL 0xFF + +struct radeon_backlight_privdata { + struct radeon_encoder *encoder; + uint8_t negative; +}; + +static uint8_t radeon_legacy_lvds_level(struct backlight_device *bd) +{ + struct radeon_backlight_privdata *pdata = bl_get_data(bd); + uint8_t level; + + /* Convert brightness to hardware level */ + if (bd->props.brightness < 0) + level = 0; + else if (bd->props.brightness > MAX_RADEON_LEVEL) + level = MAX_RADEON_LEVEL; + else + level = bd->props.brightness; + + if (pdata->negative) + level = MAX_RADEON_LEVEL - level; + + return level; +} + +static int radeon_legacy_backlight_update_status(struct backlight_device *bd) +{ + struct radeon_backlight_privdata *pdata = bl_get_data(bd); + struct radeon_encoder *radeon_encoder = pdata->encoder; + struct drm_device *dev = radeon_encoder->base.dev; + struct radeon_device *rdev = dev->dev_private; + int dpms_mode = DRM_MODE_DPMS_ON; + + if (radeon_encoder->enc_priv) { + if (rdev->is_atom_bios) { + struct radeon_encoder_atom_dig *lvds = radeon_encoder->enc_priv; + dpms_mode = lvds->dpms_mode; + lvds->backlight_level = radeon_legacy_lvds_level(bd); + } else { + struct radeon_encoder_lvds *lvds = radeon_encoder->enc_priv; + dpms_mode = lvds->dpms_mode; + lvds->backlight_level = radeon_legacy_lvds_level(bd); + } + } + + if (bd->props.brightness > 0) + radeon_legacy_lvds_update(&radeon_encoder->base, dpms_mode); + else + radeon_legacy_lvds_update(&radeon_encoder->base, DRM_MODE_DPMS_OFF); + + return 0; +} + +static int radeon_legacy_backlight_get_brightness(struct backlight_device *bd) +{ + struct radeon_backlight_privdata *pdata = bl_get_data(bd); + struct radeon_encoder *radeon_encoder = pdata->encoder; + struct drm_device *dev = radeon_encoder->base.dev; + struct radeon_device *rdev = dev->dev_private; + uint8_t backlight_level; + + backlight_level = (RREG32(RADEON_LVDS_GEN_CNTL) >> + RADEON_LVDS_BL_MOD_LEVEL_SHIFT) & 0xff; + + return pdata->negative ? MAX_RADEON_LEVEL - backlight_level : backlight_level; +} + +static const struct backlight_ops radeon_backlight_ops = { + .get_brightness = radeon_legacy_backlight_get_brightness, + .update_status = radeon_legacy_backlight_update_status, +}; + +void radeon_legacy_backlight_init(struct radeon_encoder *radeon_encoder, + struct drm_connector *drm_connector) +{ + struct drm_device *dev = radeon_encoder->base.dev; + struct radeon_device *rdev = dev->dev_private; + struct backlight_device *bd; + struct backlight_properties props; + struct radeon_backlight_privdata *pdata; + uint8_t backlight_level; + + if (!radeon_encoder->enc_priv) + return; + +#ifdef CONFIG_PMAC_BACKLIGHT + if (!pmac_has_backlight_type("ati") && + !pmac_has_backlight_type("mnca")) + return; +#endif + + pdata = kmalloc(sizeof(struct radeon_backlight_privdata), GFP_KERNEL); + if (!pdata) { + DRM_ERROR("Memory allocation failed\n"); + goto error; + } + + props.max_brightness = MAX_RADEON_LEVEL; + props.type = BACKLIGHT_RAW; + bd = backlight_device_register("radeon_bl", &drm_connector->kdev, + pdata, &radeon_backlight_ops, &props); + if (IS_ERR(bd)) { + DRM_ERROR("Backlight registration failed\n"); + goto error; + } + + pdata->encoder = radeon_encoder; + + backlight_level = (RREG32(RADEON_LVDS_GEN_CNTL) >> + RADEON_LVDS_BL_MOD_LEVEL_SHIFT) & 0xff; + + /* First, try to detect backlight level sense based on the assumption + * that firmware set it up at full brightness + */ + if (backlight_level == 0) + pdata->negative = true; + else if (backlight_level == 0xff) + pdata->negative = false; + else { + /* XXX hack... maybe some day we can figure out in what direction + * backlight should work on a given panel? + */ + pdata->negative = (rdev->family != CHIP_RV200 && + rdev->family != CHIP_RV250 && + rdev->family != CHIP_RV280 && + rdev->family != CHIP_RV350); + +#ifdef CONFIG_PMAC_BACKLIGHT + pdata->negative = (pdata->negative || + of_machine_is_compatible("PowerBook4,3") || + of_machine_is_compatible("PowerBook6,3") || + of_machine_is_compatible("PowerBook6,5")); +#endif + } + + if (rdev->is_atom_bios) { + struct radeon_encoder_atom_dig *lvds = radeon_encoder->enc_priv; + lvds->bl_dev = bd; + } else { + struct radeon_encoder_lvds *lvds = radeon_encoder->enc_priv; + lvds->bl_dev = bd; + } + + bd->props.brightness = radeon_legacy_backlight_get_brightness(bd); + bd->props.power = FB_BLANK_UNBLANK; + backlight_update_status(bd); + + DRM_INFO("radeon legacy LVDS backlight initialized\n"); + + return; + +error: + kfree(pdata); + return; +} + +static void radeon_legacy_backlight_exit(struct radeon_encoder *radeon_encoder) +{ + struct drm_device *dev = radeon_encoder->base.dev; + struct radeon_device *rdev = dev->dev_private; + struct backlight_device *bd = NULL; + + if (!radeon_encoder->enc_priv) + return; + + if (rdev->is_atom_bios) { + struct radeon_encoder_atom_dig *lvds = radeon_encoder->enc_priv; + bd = lvds->bl_dev; + lvds->bl_dev = NULL; + } else { + struct radeon_encoder_lvds *lvds = radeon_encoder->enc_priv; + bd = lvds->bl_dev; + lvds->bl_dev = NULL; + } + + if (bd) { + struct radeon_legacy_backlight_privdata *pdata; + + pdata = bl_get_data(bd); + backlight_device_unregister(bd); + kfree(pdata); + + DRM_INFO("radeon legacy LVDS backlight unloaded\n"); + } +} + +#else /* !CONFIG_BACKLIGHT_CLASS_DEVICE */ + +void radeon_legacy_backlight_init(struct radeon_encoder *encoder) +{ +} + +static void radeon_legacy_backlight_exit(struct radeon_encoder *encoder) +{ +} + +#endif + + +static void radeon_lvds_enc_destroy(struct drm_encoder *encoder) +{ + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + + if (radeon_encoder->enc_priv) { + radeon_legacy_backlight_exit(radeon_encoder); + kfree(radeon_encoder->enc_priv); + } + drm_encoder_cleanup(encoder); + kfree(radeon_encoder); +} + +static const struct drm_encoder_funcs radeon_legacy_lvds_enc_funcs = { + .destroy = radeon_lvds_enc_destroy, +}; + +static void radeon_legacy_primary_dac_dpms(struct drm_encoder *encoder, int mode) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + uint32_t crtc_ext_cntl = RREG32(RADEON_CRTC_EXT_CNTL); + uint32_t dac_cntl = RREG32(RADEON_DAC_CNTL); + uint32_t dac_macro_cntl = RREG32(RADEON_DAC_MACRO_CNTL); + + DRM_DEBUG_KMS("\n"); + + switch (mode) { + case DRM_MODE_DPMS_ON: + crtc_ext_cntl |= RADEON_CRTC_CRT_ON; + dac_cntl &= ~RADEON_DAC_PDWN; + dac_macro_cntl &= ~(RADEON_DAC_PDWN_R | + RADEON_DAC_PDWN_G | + RADEON_DAC_PDWN_B); + break; + case DRM_MODE_DPMS_STANDBY: + case DRM_MODE_DPMS_SUSPEND: + case DRM_MODE_DPMS_OFF: + crtc_ext_cntl &= ~RADEON_CRTC_CRT_ON; + dac_cntl |= RADEON_DAC_PDWN; + dac_macro_cntl |= (RADEON_DAC_PDWN_R | + RADEON_DAC_PDWN_G | + RADEON_DAC_PDWN_B); + break; + } + + WREG32(RADEON_CRTC_EXT_CNTL, crtc_ext_cntl); + WREG32(RADEON_DAC_CNTL, dac_cntl); + WREG32(RADEON_DAC_MACRO_CNTL, dac_macro_cntl); + + if (rdev->is_atom_bios) + radeon_atombios_encoder_dpms_scratch_regs(encoder, (mode == DRM_MODE_DPMS_ON) ? true : false); + else + radeon_combios_encoder_dpms_scratch_regs(encoder, (mode == DRM_MODE_DPMS_ON) ? true : false); + +} + +static void radeon_legacy_primary_dac_prepare(struct drm_encoder *encoder) +{ + struct radeon_device *rdev = encoder->dev->dev_private; + + if (rdev->is_atom_bios) + radeon_atom_output_lock(encoder, true); + else + radeon_combios_output_lock(encoder, true); + radeon_legacy_primary_dac_dpms(encoder, DRM_MODE_DPMS_OFF); +} + +static void radeon_legacy_primary_dac_commit(struct drm_encoder *encoder) +{ + struct radeon_device *rdev = encoder->dev->dev_private; + + radeon_legacy_primary_dac_dpms(encoder, DRM_MODE_DPMS_ON); + + if (rdev->is_atom_bios) + radeon_atom_output_lock(encoder, false); + else + radeon_combios_output_lock(encoder, false); +} + +static void radeon_legacy_primary_dac_mode_set(struct drm_encoder *encoder, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_crtc *radeon_crtc = to_radeon_crtc(encoder->crtc); + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + uint32_t disp_output_cntl, dac_cntl, dac2_cntl, dac_macro_cntl; + + DRM_DEBUG_KMS("\n"); + + if (radeon_crtc->crtc_id == 0) { + if (rdev->family == CHIP_R200 || ASIC_IS_R300(rdev)) { + disp_output_cntl = RREG32(RADEON_DISP_OUTPUT_CNTL) & + ~(RADEON_DISP_DAC_SOURCE_MASK); + WREG32(RADEON_DISP_OUTPUT_CNTL, disp_output_cntl); + } else { + dac2_cntl = RREG32(RADEON_DAC_CNTL2) & ~(RADEON_DAC2_DAC_CLK_SEL); + WREG32(RADEON_DAC_CNTL2, dac2_cntl); + } + } else { + if (rdev->family == CHIP_R200 || ASIC_IS_R300(rdev)) { + disp_output_cntl = RREG32(RADEON_DISP_OUTPUT_CNTL) & + ~(RADEON_DISP_DAC_SOURCE_MASK); + disp_output_cntl |= RADEON_DISP_DAC_SOURCE_CRTC2; + WREG32(RADEON_DISP_OUTPUT_CNTL, disp_output_cntl); + } else { + dac2_cntl = RREG32(RADEON_DAC_CNTL2) | RADEON_DAC2_DAC_CLK_SEL; + WREG32(RADEON_DAC_CNTL2, dac2_cntl); + } + } + + dac_cntl = (RADEON_DAC_MASK_ALL | + RADEON_DAC_VGA_ADR_EN | + /* TODO 6-bits */ + RADEON_DAC_8BIT_EN); + + WREG32_P(RADEON_DAC_CNTL, + dac_cntl, + RADEON_DAC_RANGE_CNTL | + RADEON_DAC_BLANKING); + + if (radeon_encoder->enc_priv) { + struct radeon_encoder_primary_dac *p_dac = (struct radeon_encoder_primary_dac *)radeon_encoder->enc_priv; + dac_macro_cntl = p_dac->ps2_pdac_adj; + } else + dac_macro_cntl = RREG32(RADEON_DAC_MACRO_CNTL); + dac_macro_cntl |= RADEON_DAC_PDWN_R | RADEON_DAC_PDWN_G | RADEON_DAC_PDWN_B; + WREG32(RADEON_DAC_MACRO_CNTL, dac_macro_cntl); + + if (rdev->is_atom_bios) + radeon_atombios_encoder_crtc_scratch_regs(encoder, radeon_crtc->crtc_id); + else + radeon_combios_encoder_crtc_scratch_regs(encoder, radeon_crtc->crtc_id); +} + +static enum drm_connector_status radeon_legacy_primary_dac_detect(struct drm_encoder *encoder, + struct drm_connector *connector) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + uint32_t vclk_ecp_cntl, crtc_ext_cntl; + uint32_t dac_ext_cntl, dac_cntl, dac_macro_cntl, tmp; + enum drm_connector_status found = connector_status_disconnected; + bool color = true; + + /* save the regs we need */ + vclk_ecp_cntl = RREG32_PLL(RADEON_VCLK_ECP_CNTL); + crtc_ext_cntl = RREG32(RADEON_CRTC_EXT_CNTL); + dac_ext_cntl = RREG32(RADEON_DAC_EXT_CNTL); + dac_cntl = RREG32(RADEON_DAC_CNTL); + dac_macro_cntl = RREG32(RADEON_DAC_MACRO_CNTL); + + tmp = vclk_ecp_cntl & + ~(RADEON_PIXCLK_ALWAYS_ONb | RADEON_PIXCLK_DAC_ALWAYS_ONb); + WREG32_PLL(RADEON_VCLK_ECP_CNTL, tmp); + + tmp = crtc_ext_cntl | RADEON_CRTC_CRT_ON; + WREG32(RADEON_CRTC_EXT_CNTL, tmp); + + tmp = RADEON_DAC_FORCE_BLANK_OFF_EN | + RADEON_DAC_FORCE_DATA_EN; + + if (color) + tmp |= RADEON_DAC_FORCE_DATA_SEL_RGB; + else + tmp |= RADEON_DAC_FORCE_DATA_SEL_G; + + if (ASIC_IS_R300(rdev)) + tmp |= (0x1b6 << RADEON_DAC_FORCE_DATA_SHIFT); + else + tmp |= (0x180 << RADEON_DAC_FORCE_DATA_SHIFT); + + WREG32(RADEON_DAC_EXT_CNTL, tmp); + + tmp = dac_cntl & ~(RADEON_DAC_RANGE_CNTL_MASK | RADEON_DAC_PDWN); + tmp |= RADEON_DAC_RANGE_CNTL_PS2 | RADEON_DAC_CMP_EN; + WREG32(RADEON_DAC_CNTL, tmp); + + tmp &= ~(RADEON_DAC_PDWN_R | + RADEON_DAC_PDWN_G | + RADEON_DAC_PDWN_B); + + WREG32(RADEON_DAC_MACRO_CNTL, tmp); + + udelay(2000); + + if (RREG32(RADEON_DAC_CNTL) & RADEON_DAC_CMP_OUTPUT) + found = connector_status_connected; + + /* restore the regs we used */ + WREG32(RADEON_DAC_CNTL, dac_cntl); + WREG32(RADEON_DAC_MACRO_CNTL, dac_macro_cntl); + WREG32(RADEON_DAC_EXT_CNTL, dac_ext_cntl); + WREG32(RADEON_CRTC_EXT_CNTL, crtc_ext_cntl); + WREG32_PLL(RADEON_VCLK_ECP_CNTL, vclk_ecp_cntl); + + return found; +} + +static const struct drm_encoder_helper_funcs radeon_legacy_primary_dac_helper_funcs = { + .dpms = radeon_legacy_primary_dac_dpms, + .mode_fixup = radeon_legacy_mode_fixup, + .prepare = radeon_legacy_primary_dac_prepare, + .mode_set = radeon_legacy_primary_dac_mode_set, + .commit = radeon_legacy_primary_dac_commit, + .detect = radeon_legacy_primary_dac_detect, + .disable = radeon_legacy_encoder_disable, +}; + + +static const struct drm_encoder_funcs radeon_legacy_primary_dac_enc_funcs = { + .destroy = radeon_enc_destroy, +}; + +static void radeon_legacy_tmds_int_dpms(struct drm_encoder *encoder, int mode) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + uint32_t fp_gen_cntl = RREG32(RADEON_FP_GEN_CNTL); + DRM_DEBUG_KMS("\n"); + + switch (mode) { + case DRM_MODE_DPMS_ON: + fp_gen_cntl |= (RADEON_FP_FPON | RADEON_FP_TMDS_EN); + break; + case DRM_MODE_DPMS_STANDBY: + case DRM_MODE_DPMS_SUSPEND: + case DRM_MODE_DPMS_OFF: + fp_gen_cntl &= ~(RADEON_FP_FPON | RADEON_FP_TMDS_EN); + break; + } + + WREG32(RADEON_FP_GEN_CNTL, fp_gen_cntl); + + if (rdev->is_atom_bios) + radeon_atombios_encoder_dpms_scratch_regs(encoder, (mode == DRM_MODE_DPMS_ON) ? true : false); + else + radeon_combios_encoder_dpms_scratch_regs(encoder, (mode == DRM_MODE_DPMS_ON) ? true : false); + +} + +static void radeon_legacy_tmds_int_prepare(struct drm_encoder *encoder) +{ + struct radeon_device *rdev = encoder->dev->dev_private; + + if (rdev->is_atom_bios) + radeon_atom_output_lock(encoder, true); + else + radeon_combios_output_lock(encoder, true); + radeon_legacy_tmds_int_dpms(encoder, DRM_MODE_DPMS_OFF); +} + +static void radeon_legacy_tmds_int_commit(struct drm_encoder *encoder) +{ + struct radeon_device *rdev = encoder->dev->dev_private; + + radeon_legacy_tmds_int_dpms(encoder, DRM_MODE_DPMS_ON); + + if (rdev->is_atom_bios) + radeon_atom_output_lock(encoder, true); + else + radeon_combios_output_lock(encoder, true); +} + +static void radeon_legacy_tmds_int_mode_set(struct drm_encoder *encoder, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_crtc *radeon_crtc = to_radeon_crtc(encoder->crtc); + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + uint32_t tmp, tmds_pll_cntl, tmds_transmitter_cntl, fp_gen_cntl; + int i; + + DRM_DEBUG_KMS("\n"); + + tmp = tmds_pll_cntl = RREG32(RADEON_TMDS_PLL_CNTL); + tmp &= 0xfffff; + if (rdev->family == CHIP_RV280) { + /* bit 22 of TMDS_PLL_CNTL is read-back inverted */ + tmp ^= (1 << 22); + tmds_pll_cntl ^= (1 << 22); + } + + if (radeon_encoder->enc_priv) { + struct radeon_encoder_int_tmds *tmds = (struct radeon_encoder_int_tmds *)radeon_encoder->enc_priv; + + for (i = 0; i < 4; i++) { + if (tmds->tmds_pll[i].freq == 0) + break; + if ((uint32_t)(mode->clock / 10) < tmds->tmds_pll[i].freq) { + tmp = tmds->tmds_pll[i].value ; + break; + } + } + } + + if (ASIC_IS_R300(rdev) || (rdev->family == CHIP_RV280)) { + if (tmp & 0xfff00000) + tmds_pll_cntl = tmp; + else { + tmds_pll_cntl &= 0xfff00000; + tmds_pll_cntl |= tmp; + } + } else + tmds_pll_cntl = tmp; + + tmds_transmitter_cntl = RREG32(RADEON_TMDS_TRANSMITTER_CNTL) & + ~(RADEON_TMDS_TRANSMITTER_PLLRST); + + if (rdev->family == CHIP_R200 || + rdev->family == CHIP_R100 || + ASIC_IS_R300(rdev)) + tmds_transmitter_cntl &= ~(RADEON_TMDS_TRANSMITTER_PLLEN); + else /* RV chips got this bit reversed */ + tmds_transmitter_cntl |= RADEON_TMDS_TRANSMITTER_PLLEN; + + fp_gen_cntl = (RREG32(RADEON_FP_GEN_CNTL) | + (RADEON_FP_CRTC_DONT_SHADOW_VPAR | + RADEON_FP_CRTC_DONT_SHADOW_HEND)); + + fp_gen_cntl &= ~(RADEON_FP_FPON | RADEON_FP_TMDS_EN); + + fp_gen_cntl &= ~(RADEON_FP_RMX_HVSYNC_CONTROL_EN | + RADEON_FP_DFP_SYNC_SEL | + RADEON_FP_CRT_SYNC_SEL | + RADEON_FP_CRTC_LOCK_8DOT | + RADEON_FP_USE_SHADOW_EN | + RADEON_FP_CRTC_USE_SHADOW_VEND | + RADEON_FP_CRT_SYNC_ALT); + + if (1) /* FIXME rgbBits == 8 */ + fp_gen_cntl |= RADEON_FP_PANEL_FORMAT; /* 24 bit format */ + else + fp_gen_cntl &= ~RADEON_FP_PANEL_FORMAT;/* 18 bit format */ + + if (radeon_crtc->crtc_id == 0) { + if (ASIC_IS_R300(rdev) || rdev->family == CHIP_R200) { + fp_gen_cntl &= ~R200_FP_SOURCE_SEL_MASK; + if (radeon_encoder->rmx_type != RMX_OFF) + fp_gen_cntl |= R200_FP_SOURCE_SEL_RMX; + else + fp_gen_cntl |= R200_FP_SOURCE_SEL_CRTC1; + } else + fp_gen_cntl &= ~RADEON_FP_SEL_CRTC2; + } else { + if (ASIC_IS_R300(rdev) || rdev->family == CHIP_R200) { + fp_gen_cntl &= ~R200_FP_SOURCE_SEL_MASK; + fp_gen_cntl |= R200_FP_SOURCE_SEL_CRTC2; + } else + fp_gen_cntl |= RADEON_FP_SEL_CRTC2; + } + + WREG32(RADEON_TMDS_PLL_CNTL, tmds_pll_cntl); + WREG32(RADEON_TMDS_TRANSMITTER_CNTL, tmds_transmitter_cntl); + WREG32(RADEON_FP_GEN_CNTL, fp_gen_cntl); + + if (rdev->is_atom_bios) + radeon_atombios_encoder_crtc_scratch_regs(encoder, radeon_crtc->crtc_id); + else + radeon_combios_encoder_crtc_scratch_regs(encoder, radeon_crtc->crtc_id); +} + +static const struct drm_encoder_helper_funcs radeon_legacy_tmds_int_helper_funcs = { + .dpms = radeon_legacy_tmds_int_dpms, + .mode_fixup = radeon_legacy_mode_fixup, + .prepare = radeon_legacy_tmds_int_prepare, + .mode_set = radeon_legacy_tmds_int_mode_set, + .commit = radeon_legacy_tmds_int_commit, + .disable = radeon_legacy_encoder_disable, +}; + + +static const struct drm_encoder_funcs radeon_legacy_tmds_int_enc_funcs = { + .destroy = radeon_enc_destroy, +}; + +static void radeon_legacy_tmds_ext_dpms(struct drm_encoder *encoder, int mode) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + uint32_t fp2_gen_cntl = RREG32(RADEON_FP2_GEN_CNTL); + DRM_DEBUG_KMS("\n"); + + switch (mode) { + case DRM_MODE_DPMS_ON: + fp2_gen_cntl &= ~RADEON_FP2_BLANK_EN; + fp2_gen_cntl |= (RADEON_FP2_ON | RADEON_FP2_DVO_EN); + break; + case DRM_MODE_DPMS_STANDBY: + case DRM_MODE_DPMS_SUSPEND: + case DRM_MODE_DPMS_OFF: + fp2_gen_cntl |= RADEON_FP2_BLANK_EN; + fp2_gen_cntl &= ~(RADEON_FP2_ON | RADEON_FP2_DVO_EN); + break; + } + + WREG32(RADEON_FP2_GEN_CNTL, fp2_gen_cntl); + + if (rdev->is_atom_bios) + radeon_atombios_encoder_dpms_scratch_regs(encoder, (mode == DRM_MODE_DPMS_ON) ? true : false); + else + radeon_combios_encoder_dpms_scratch_regs(encoder, (mode == DRM_MODE_DPMS_ON) ? true : false); + +} + +static void radeon_legacy_tmds_ext_prepare(struct drm_encoder *encoder) +{ + struct radeon_device *rdev = encoder->dev->dev_private; + + if (rdev->is_atom_bios) + radeon_atom_output_lock(encoder, true); + else + radeon_combios_output_lock(encoder, true); + radeon_legacy_tmds_ext_dpms(encoder, DRM_MODE_DPMS_OFF); +} + +static void radeon_legacy_tmds_ext_commit(struct drm_encoder *encoder) +{ + struct radeon_device *rdev = encoder->dev->dev_private; + radeon_legacy_tmds_ext_dpms(encoder, DRM_MODE_DPMS_ON); + + if (rdev->is_atom_bios) + radeon_atom_output_lock(encoder, false); + else + radeon_combios_output_lock(encoder, false); +} + +static void radeon_legacy_tmds_ext_mode_set(struct drm_encoder *encoder, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_crtc *radeon_crtc = to_radeon_crtc(encoder->crtc); + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + uint32_t fp2_gen_cntl; + + DRM_DEBUG_KMS("\n"); + + if (rdev->is_atom_bios) { + radeon_encoder->pixel_clock = adjusted_mode->clock; + atombios_dvo_setup(encoder, ATOM_ENABLE); + fp2_gen_cntl = RREG32(RADEON_FP2_GEN_CNTL); + } else { + fp2_gen_cntl = RREG32(RADEON_FP2_GEN_CNTL); + + if (1) /* FIXME rgbBits == 8 */ + fp2_gen_cntl |= RADEON_FP2_PANEL_FORMAT; /* 24 bit format, */ + else + fp2_gen_cntl &= ~RADEON_FP2_PANEL_FORMAT;/* 18 bit format, */ + + fp2_gen_cntl &= ~(RADEON_FP2_ON | + RADEON_FP2_DVO_EN | + RADEON_FP2_DVO_RATE_SEL_SDR); + + /* XXX: these are oem specific */ + if (ASIC_IS_R300(rdev)) { + if ((dev->pdev->device == 0x4850) && + (dev->pdev->subsystem_vendor == 0x1028) && + (dev->pdev->subsystem_device == 0x2001)) /* Dell Inspiron 8600 */ + fp2_gen_cntl |= R300_FP2_DVO_CLOCK_MODE_SINGLE; + else + fp2_gen_cntl |= RADEON_FP2_PAD_FLOP_EN | R300_FP2_DVO_CLOCK_MODE_SINGLE; + + /*if (mode->clock > 165000) + fp2_gen_cntl |= R300_FP2_DVO_DUAL_CHANNEL_EN;*/ + } + if (!radeon_combios_external_tmds_setup(encoder)) + radeon_external_tmds_setup(encoder); + } + + if (radeon_crtc->crtc_id == 0) { + if ((rdev->family == CHIP_R200) || ASIC_IS_R300(rdev)) { + fp2_gen_cntl &= ~R200_FP2_SOURCE_SEL_MASK; + if (radeon_encoder->rmx_type != RMX_OFF) + fp2_gen_cntl |= R200_FP2_SOURCE_SEL_RMX; + else + fp2_gen_cntl |= R200_FP2_SOURCE_SEL_CRTC1; + } else + fp2_gen_cntl &= ~RADEON_FP2_SRC_SEL_CRTC2; + } else { + if ((rdev->family == CHIP_R200) || ASIC_IS_R300(rdev)) { + fp2_gen_cntl &= ~R200_FP2_SOURCE_SEL_MASK; + fp2_gen_cntl |= R200_FP2_SOURCE_SEL_CRTC2; + } else + fp2_gen_cntl |= RADEON_FP2_SRC_SEL_CRTC2; + } + + WREG32(RADEON_FP2_GEN_CNTL, fp2_gen_cntl); + + if (rdev->is_atom_bios) + radeon_atombios_encoder_crtc_scratch_regs(encoder, radeon_crtc->crtc_id); + else + radeon_combios_encoder_crtc_scratch_regs(encoder, radeon_crtc->crtc_id); +} + +static void radeon_ext_tmds_enc_destroy(struct drm_encoder *encoder) +{ + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct radeon_encoder_ext_tmds *tmds = radeon_encoder->enc_priv; + if (tmds) { + if (tmds->i2c_bus) + radeon_i2c_destroy(tmds->i2c_bus); + } + kfree(radeon_encoder->enc_priv); + drm_encoder_cleanup(encoder); + kfree(radeon_encoder); +} + +static const struct drm_encoder_helper_funcs radeon_legacy_tmds_ext_helper_funcs = { + .dpms = radeon_legacy_tmds_ext_dpms, + .mode_fixup = radeon_legacy_mode_fixup, + .prepare = radeon_legacy_tmds_ext_prepare, + .mode_set = radeon_legacy_tmds_ext_mode_set, + .commit = radeon_legacy_tmds_ext_commit, + .disable = radeon_legacy_encoder_disable, +}; + + +static const struct drm_encoder_funcs radeon_legacy_tmds_ext_enc_funcs = { + .destroy = radeon_ext_tmds_enc_destroy, +}; + +static void radeon_legacy_tv_dac_dpms(struct drm_encoder *encoder, int mode) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + uint32_t fp2_gen_cntl = 0, crtc2_gen_cntl = 0, tv_dac_cntl = 0; + uint32_t tv_master_cntl = 0; + bool is_tv; + DRM_DEBUG_KMS("\n"); + + is_tv = radeon_encoder->active_device & ATOM_DEVICE_TV_SUPPORT ? true : false; + + if (rdev->family == CHIP_R200) + fp2_gen_cntl = RREG32(RADEON_FP2_GEN_CNTL); + else { + if (is_tv) + tv_master_cntl = RREG32(RADEON_TV_MASTER_CNTL); + else + crtc2_gen_cntl = RREG32(RADEON_CRTC2_GEN_CNTL); + tv_dac_cntl = RREG32(RADEON_TV_DAC_CNTL); + } + + switch (mode) { + case DRM_MODE_DPMS_ON: + if (rdev->family == CHIP_R200) { + fp2_gen_cntl |= (RADEON_FP2_ON | RADEON_FP2_DVO_EN); + } else { + if (is_tv) + tv_master_cntl |= RADEON_TV_ON; + else + crtc2_gen_cntl |= RADEON_CRTC2_CRT2_ON; + + if (rdev->family == CHIP_R420 || + rdev->family == CHIP_R423 || + rdev->family == CHIP_RV410) + tv_dac_cntl &= ~(R420_TV_DAC_RDACPD | + R420_TV_DAC_GDACPD | + R420_TV_DAC_BDACPD | + RADEON_TV_DAC_BGSLEEP); + else + tv_dac_cntl &= ~(RADEON_TV_DAC_RDACPD | + RADEON_TV_DAC_GDACPD | + RADEON_TV_DAC_BDACPD | + RADEON_TV_DAC_BGSLEEP); + } + break; + case DRM_MODE_DPMS_STANDBY: + case DRM_MODE_DPMS_SUSPEND: + case DRM_MODE_DPMS_OFF: + if (rdev->family == CHIP_R200) + fp2_gen_cntl &= ~(RADEON_FP2_ON | RADEON_FP2_DVO_EN); + else { + if (is_tv) + tv_master_cntl &= ~RADEON_TV_ON; + else + crtc2_gen_cntl &= ~RADEON_CRTC2_CRT2_ON; + + if (rdev->family == CHIP_R420 || + rdev->family == CHIP_R423 || + rdev->family == CHIP_RV410) + tv_dac_cntl |= (R420_TV_DAC_RDACPD | + R420_TV_DAC_GDACPD | + R420_TV_DAC_BDACPD | + RADEON_TV_DAC_BGSLEEP); + else + tv_dac_cntl |= (RADEON_TV_DAC_RDACPD | + RADEON_TV_DAC_GDACPD | + RADEON_TV_DAC_BDACPD | + RADEON_TV_DAC_BGSLEEP); + } + break; + } + + if (rdev->family == CHIP_R200) { + WREG32(RADEON_FP2_GEN_CNTL, fp2_gen_cntl); + } else { + if (is_tv) + WREG32(RADEON_TV_MASTER_CNTL, tv_master_cntl); + else + WREG32(RADEON_CRTC2_GEN_CNTL, crtc2_gen_cntl); + WREG32(RADEON_TV_DAC_CNTL, tv_dac_cntl); + } + + if (rdev->is_atom_bios) + radeon_atombios_encoder_dpms_scratch_regs(encoder, (mode == DRM_MODE_DPMS_ON) ? true : false); + else + radeon_combios_encoder_dpms_scratch_regs(encoder, (mode == DRM_MODE_DPMS_ON) ? true : false); + +} + +static void radeon_legacy_tv_dac_prepare(struct drm_encoder *encoder) +{ + struct radeon_device *rdev = encoder->dev->dev_private; + + if (rdev->is_atom_bios) + radeon_atom_output_lock(encoder, true); + else + radeon_combios_output_lock(encoder, true); + radeon_legacy_tv_dac_dpms(encoder, DRM_MODE_DPMS_OFF); +} + +static void radeon_legacy_tv_dac_commit(struct drm_encoder *encoder) +{ + struct radeon_device *rdev = encoder->dev->dev_private; + + radeon_legacy_tv_dac_dpms(encoder, DRM_MODE_DPMS_ON); + + if (rdev->is_atom_bios) + radeon_atom_output_lock(encoder, true); + else + radeon_combios_output_lock(encoder, true); +} + +static void radeon_legacy_tv_dac_mode_set(struct drm_encoder *encoder, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_crtc *radeon_crtc = to_radeon_crtc(encoder->crtc); + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct radeon_encoder_tv_dac *tv_dac = radeon_encoder->enc_priv; + uint32_t tv_dac_cntl, gpiopad_a = 0, dac2_cntl, disp_output_cntl = 0; + uint32_t disp_hw_debug = 0, fp2_gen_cntl = 0, disp_tv_out_cntl = 0; + bool is_tv = false; + + DRM_DEBUG_KMS("\n"); + + is_tv = radeon_encoder->active_device & ATOM_DEVICE_TV_SUPPORT ? true : false; + + if (rdev->family != CHIP_R200) { + tv_dac_cntl = RREG32(RADEON_TV_DAC_CNTL); + if (rdev->family == CHIP_R420 || + rdev->family == CHIP_R423 || + rdev->family == CHIP_RV410) { + tv_dac_cntl &= ~(RADEON_TV_DAC_STD_MASK | + RADEON_TV_DAC_BGADJ_MASK | + R420_TV_DAC_DACADJ_MASK | + R420_TV_DAC_RDACPD | + R420_TV_DAC_GDACPD | + R420_TV_DAC_BDACPD | + R420_TV_DAC_TVENABLE); + } else { + tv_dac_cntl &= ~(RADEON_TV_DAC_STD_MASK | + RADEON_TV_DAC_BGADJ_MASK | + RADEON_TV_DAC_DACADJ_MASK | + RADEON_TV_DAC_RDACPD | + RADEON_TV_DAC_GDACPD | + RADEON_TV_DAC_BDACPD); + } + + tv_dac_cntl |= RADEON_TV_DAC_NBLANK | RADEON_TV_DAC_NHOLD; + + if (is_tv) { + if (tv_dac->tv_std == TV_STD_NTSC || + tv_dac->tv_std == TV_STD_NTSC_J || + tv_dac->tv_std == TV_STD_PAL_M || + tv_dac->tv_std == TV_STD_PAL_60) + tv_dac_cntl |= tv_dac->ntsc_tvdac_adj; + else + tv_dac_cntl |= tv_dac->pal_tvdac_adj; + + if (tv_dac->tv_std == TV_STD_NTSC || + tv_dac->tv_std == TV_STD_NTSC_J) + tv_dac_cntl |= RADEON_TV_DAC_STD_NTSC; + else + tv_dac_cntl |= RADEON_TV_DAC_STD_PAL; + } else + tv_dac_cntl |= (RADEON_TV_DAC_STD_PS2 | + tv_dac->ps2_tvdac_adj); + + WREG32(RADEON_TV_DAC_CNTL, tv_dac_cntl); + } + + if (ASIC_IS_R300(rdev)) { + gpiopad_a = RREG32(RADEON_GPIOPAD_A) | 1; + disp_output_cntl = RREG32(RADEON_DISP_OUTPUT_CNTL); + } else if (rdev->family != CHIP_R200) + disp_hw_debug = RREG32(RADEON_DISP_HW_DEBUG); + else if (rdev->family == CHIP_R200) + fp2_gen_cntl = RREG32(RADEON_FP2_GEN_CNTL); + + if (rdev->family >= CHIP_R200) + disp_tv_out_cntl = RREG32(RADEON_DISP_TV_OUT_CNTL); + + if (is_tv) { + uint32_t dac_cntl; + + dac_cntl = RREG32(RADEON_DAC_CNTL); + dac_cntl &= ~RADEON_DAC_TVO_EN; + WREG32(RADEON_DAC_CNTL, dac_cntl); + + if (ASIC_IS_R300(rdev)) + gpiopad_a = RREG32(RADEON_GPIOPAD_A) & ~1; + + dac2_cntl = RREG32(RADEON_DAC_CNTL2) & ~RADEON_DAC2_DAC2_CLK_SEL; + if (radeon_crtc->crtc_id == 0) { + if (ASIC_IS_R300(rdev)) { + disp_output_cntl &= ~RADEON_DISP_TVDAC_SOURCE_MASK; + disp_output_cntl |= (RADEON_DISP_TVDAC_SOURCE_CRTC | + RADEON_DISP_TV_SOURCE_CRTC); + } + if (rdev->family >= CHIP_R200) { + disp_tv_out_cntl &= ~RADEON_DISP_TV_PATH_SRC_CRTC2; + } else { + disp_hw_debug |= RADEON_CRT2_DISP1_SEL; + } + } else { + if (ASIC_IS_R300(rdev)) { + disp_output_cntl &= ~RADEON_DISP_TVDAC_SOURCE_MASK; + disp_output_cntl |= RADEON_DISP_TV_SOURCE_CRTC; + } + if (rdev->family >= CHIP_R200) { + disp_tv_out_cntl |= RADEON_DISP_TV_PATH_SRC_CRTC2; + } else { + disp_hw_debug &= ~RADEON_CRT2_DISP1_SEL; + } + } + WREG32(RADEON_DAC_CNTL2, dac2_cntl); + } else { + + dac2_cntl = RREG32(RADEON_DAC_CNTL2) | RADEON_DAC2_DAC2_CLK_SEL; + + if (radeon_crtc->crtc_id == 0) { + if (ASIC_IS_R300(rdev)) { + disp_output_cntl &= ~RADEON_DISP_TVDAC_SOURCE_MASK; + disp_output_cntl |= RADEON_DISP_TVDAC_SOURCE_CRTC; + } else if (rdev->family == CHIP_R200) { + fp2_gen_cntl &= ~(R200_FP2_SOURCE_SEL_MASK | + RADEON_FP2_DVO_RATE_SEL_SDR); + } else + disp_hw_debug |= RADEON_CRT2_DISP1_SEL; + } else { + if (ASIC_IS_R300(rdev)) { + disp_output_cntl &= ~RADEON_DISP_TVDAC_SOURCE_MASK; + disp_output_cntl |= RADEON_DISP_TVDAC_SOURCE_CRTC2; + } else if (rdev->family == CHIP_R200) { + fp2_gen_cntl &= ~(R200_FP2_SOURCE_SEL_MASK | + RADEON_FP2_DVO_RATE_SEL_SDR); + fp2_gen_cntl |= R200_FP2_SOURCE_SEL_CRTC2; + } else + disp_hw_debug &= ~RADEON_CRT2_DISP1_SEL; + } + WREG32(RADEON_DAC_CNTL2, dac2_cntl); + } + + if (ASIC_IS_R300(rdev)) { + WREG32_P(RADEON_GPIOPAD_A, gpiopad_a, ~1); + WREG32(RADEON_DISP_OUTPUT_CNTL, disp_output_cntl); + } else if (rdev->family != CHIP_R200) + WREG32(RADEON_DISP_HW_DEBUG, disp_hw_debug); + else if (rdev->family == CHIP_R200) + WREG32(RADEON_FP2_GEN_CNTL, fp2_gen_cntl); + + if (rdev->family >= CHIP_R200) + WREG32(RADEON_DISP_TV_OUT_CNTL, disp_tv_out_cntl); + + if (is_tv) + radeon_legacy_tv_mode_set(encoder, mode, adjusted_mode); + + if (rdev->is_atom_bios) + radeon_atombios_encoder_crtc_scratch_regs(encoder, radeon_crtc->crtc_id); + else + radeon_combios_encoder_crtc_scratch_regs(encoder, radeon_crtc->crtc_id); + +} + +static bool r300_legacy_tv_detect(struct drm_encoder *encoder, + struct drm_connector *connector) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + uint32_t crtc2_gen_cntl, tv_dac_cntl, dac_cntl2, dac_ext_cntl; + uint32_t disp_output_cntl, gpiopad_a, tmp; + bool found = false; + + /* save regs needed */ + gpiopad_a = RREG32(RADEON_GPIOPAD_A); + dac_cntl2 = RREG32(RADEON_DAC_CNTL2); + crtc2_gen_cntl = RREG32(RADEON_CRTC2_GEN_CNTL); + dac_ext_cntl = RREG32(RADEON_DAC_EXT_CNTL); + tv_dac_cntl = RREG32(RADEON_TV_DAC_CNTL); + disp_output_cntl = RREG32(RADEON_DISP_OUTPUT_CNTL); + + WREG32_P(RADEON_GPIOPAD_A, 0, ~1); + + WREG32(RADEON_DAC_CNTL2, RADEON_DAC2_DAC2_CLK_SEL); + + WREG32(RADEON_CRTC2_GEN_CNTL, + RADEON_CRTC2_CRT2_ON | RADEON_CRTC2_VSYNC_TRISTAT); + + tmp = disp_output_cntl & ~RADEON_DISP_TVDAC_SOURCE_MASK; + tmp |= RADEON_DISP_TVDAC_SOURCE_CRTC2; + WREG32(RADEON_DISP_OUTPUT_CNTL, tmp); + + WREG32(RADEON_DAC_EXT_CNTL, + RADEON_DAC2_FORCE_BLANK_OFF_EN | + RADEON_DAC2_FORCE_DATA_EN | + RADEON_DAC_FORCE_DATA_SEL_RGB | + (0xec << RADEON_DAC_FORCE_DATA_SHIFT)); + + WREG32(RADEON_TV_DAC_CNTL, + RADEON_TV_DAC_STD_NTSC | + (8 << RADEON_TV_DAC_BGADJ_SHIFT) | + (6 << RADEON_TV_DAC_DACADJ_SHIFT)); + + RREG32(RADEON_TV_DAC_CNTL); + mdelay(4); + + WREG32(RADEON_TV_DAC_CNTL, + RADEON_TV_DAC_NBLANK | + RADEON_TV_DAC_NHOLD | + RADEON_TV_MONITOR_DETECT_EN | + RADEON_TV_DAC_STD_NTSC | + (8 << RADEON_TV_DAC_BGADJ_SHIFT) | + (6 << RADEON_TV_DAC_DACADJ_SHIFT)); + + RREG32(RADEON_TV_DAC_CNTL); + mdelay(6); + + tmp = RREG32(RADEON_TV_DAC_CNTL); + if ((tmp & RADEON_TV_DAC_GDACDET) != 0) { + found = true; + DRM_DEBUG_KMS("S-video TV connection detected\n"); + } else if ((tmp & RADEON_TV_DAC_BDACDET) != 0) { + found = true; + DRM_DEBUG_KMS("Composite TV connection detected\n"); + } + + WREG32(RADEON_TV_DAC_CNTL, tv_dac_cntl); + WREG32(RADEON_DAC_EXT_CNTL, dac_ext_cntl); + WREG32(RADEON_CRTC2_GEN_CNTL, crtc2_gen_cntl); + WREG32(RADEON_DISP_OUTPUT_CNTL, disp_output_cntl); + WREG32(RADEON_DAC_CNTL2, dac_cntl2); + WREG32_P(RADEON_GPIOPAD_A, gpiopad_a, ~1); + return found; +} + +static bool radeon_legacy_tv_detect(struct drm_encoder *encoder, + struct drm_connector *connector) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + uint32_t tv_dac_cntl, dac_cntl2; + uint32_t config_cntl, tv_pre_dac_mux_cntl, tv_master_cntl, tmp; + bool found = false; + + if (ASIC_IS_R300(rdev)) + return r300_legacy_tv_detect(encoder, connector); + + dac_cntl2 = RREG32(RADEON_DAC_CNTL2); + tv_master_cntl = RREG32(RADEON_TV_MASTER_CNTL); + tv_dac_cntl = RREG32(RADEON_TV_DAC_CNTL); + config_cntl = RREG32(RADEON_CONFIG_CNTL); + tv_pre_dac_mux_cntl = RREG32(RADEON_TV_PRE_DAC_MUX_CNTL); + + tmp = dac_cntl2 & ~RADEON_DAC2_DAC2_CLK_SEL; + WREG32(RADEON_DAC_CNTL2, tmp); + + tmp = tv_master_cntl | RADEON_TV_ON; + tmp &= ~(RADEON_TV_ASYNC_RST | + RADEON_RESTART_PHASE_FIX | + RADEON_CRT_FIFO_CE_EN | + RADEON_TV_FIFO_CE_EN | + RADEON_RE_SYNC_NOW_SEL_MASK); + tmp |= RADEON_TV_FIFO_ASYNC_RST | RADEON_CRT_ASYNC_RST; + WREG32(RADEON_TV_MASTER_CNTL, tmp); + + tmp = RADEON_TV_DAC_NBLANK | RADEON_TV_DAC_NHOLD | + RADEON_TV_MONITOR_DETECT_EN | RADEON_TV_DAC_STD_NTSC | + (8 << RADEON_TV_DAC_BGADJ_SHIFT); + + if (config_cntl & RADEON_CFG_ATI_REV_ID_MASK) + tmp |= (4 << RADEON_TV_DAC_DACADJ_SHIFT); + else + tmp |= (8 << RADEON_TV_DAC_DACADJ_SHIFT); + WREG32(RADEON_TV_DAC_CNTL, tmp); + + tmp = RADEON_C_GRN_EN | RADEON_CMP_BLU_EN | + RADEON_RED_MX_FORCE_DAC_DATA | + RADEON_GRN_MX_FORCE_DAC_DATA | + RADEON_BLU_MX_FORCE_DAC_DATA | + (0x109 << RADEON_TV_FORCE_DAC_DATA_SHIFT); + WREG32(RADEON_TV_PRE_DAC_MUX_CNTL, tmp); + + mdelay(3); + tmp = RREG32(RADEON_TV_DAC_CNTL); + if (tmp & RADEON_TV_DAC_GDACDET) { + found = true; + DRM_DEBUG_KMS("S-video TV connection detected\n"); + } else if ((tmp & RADEON_TV_DAC_BDACDET) != 0) { + found = true; + DRM_DEBUG_KMS("Composite TV connection detected\n"); + } + + WREG32(RADEON_TV_PRE_DAC_MUX_CNTL, tv_pre_dac_mux_cntl); + WREG32(RADEON_TV_DAC_CNTL, tv_dac_cntl); + WREG32(RADEON_TV_MASTER_CNTL, tv_master_cntl); + WREG32(RADEON_DAC_CNTL2, dac_cntl2); + return found; +} + +static enum drm_connector_status radeon_legacy_tv_dac_detect(struct drm_encoder *encoder, + struct drm_connector *connector) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + uint32_t crtc2_gen_cntl, tv_dac_cntl, dac_cntl2, dac_ext_cntl; + uint32_t disp_hw_debug, disp_output_cntl, gpiopad_a, pixclks_cntl, tmp; + enum drm_connector_status found = connector_status_disconnected; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct radeon_encoder_tv_dac *tv_dac = radeon_encoder->enc_priv; + bool color = true; + struct drm_crtc *crtc; + + /* find out if crtc2 is in use or if this encoder is using it */ + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + if ((radeon_crtc->crtc_id == 1) && crtc->enabled) { + if (encoder->crtc != crtc) { + return connector_status_disconnected; + } + } + } + + if (connector->connector_type == DRM_MODE_CONNECTOR_SVIDEO || + connector->connector_type == DRM_MODE_CONNECTOR_Composite || + connector->connector_type == DRM_MODE_CONNECTOR_9PinDIN) { + bool tv_detect; + + if (radeon_encoder->active_device && !(radeon_encoder->active_device & ATOM_DEVICE_TV_SUPPORT)) + return connector_status_disconnected; + + tv_detect = radeon_legacy_tv_detect(encoder, connector); + if (tv_detect && tv_dac) + found = connector_status_connected; + return found; + } + + /* don't probe if the encoder is being used for something else not CRT related */ + if (radeon_encoder->active_device && !(radeon_encoder->active_device & ATOM_DEVICE_CRT_SUPPORT)) { + DRM_INFO("not detecting due to %08x\n", radeon_encoder->active_device); + return connector_status_disconnected; + } + + /* save the regs we need */ + pixclks_cntl = RREG32_PLL(RADEON_PIXCLKS_CNTL); + gpiopad_a = ASIC_IS_R300(rdev) ? RREG32(RADEON_GPIOPAD_A) : 0; + disp_output_cntl = ASIC_IS_R300(rdev) ? RREG32(RADEON_DISP_OUTPUT_CNTL) : 0; + disp_hw_debug = ASIC_IS_R300(rdev) ? 0 : RREG32(RADEON_DISP_HW_DEBUG); + crtc2_gen_cntl = RREG32(RADEON_CRTC2_GEN_CNTL); + tv_dac_cntl = RREG32(RADEON_TV_DAC_CNTL); + dac_ext_cntl = RREG32(RADEON_DAC_EXT_CNTL); + dac_cntl2 = RREG32(RADEON_DAC_CNTL2); + + tmp = pixclks_cntl & ~(RADEON_PIX2CLK_ALWAYS_ONb + | RADEON_PIX2CLK_DAC_ALWAYS_ONb); + WREG32_PLL(RADEON_PIXCLKS_CNTL, tmp); + + if (ASIC_IS_R300(rdev)) + WREG32_P(RADEON_GPIOPAD_A, 1, ~1); + + tmp = crtc2_gen_cntl & ~RADEON_CRTC2_PIX_WIDTH_MASK; + tmp |= RADEON_CRTC2_CRT2_ON | + (2 << RADEON_CRTC2_PIX_WIDTH_SHIFT); + + WREG32(RADEON_CRTC2_GEN_CNTL, tmp); + + if (ASIC_IS_R300(rdev)) { + tmp = disp_output_cntl & ~RADEON_DISP_TVDAC_SOURCE_MASK; + tmp |= RADEON_DISP_TVDAC_SOURCE_CRTC2; + WREG32(RADEON_DISP_OUTPUT_CNTL, tmp); + } else { + tmp = disp_hw_debug & ~RADEON_CRT2_DISP1_SEL; + WREG32(RADEON_DISP_HW_DEBUG, tmp); + } + + tmp = RADEON_TV_DAC_NBLANK | + RADEON_TV_DAC_NHOLD | + RADEON_TV_MONITOR_DETECT_EN | + RADEON_TV_DAC_STD_PS2; + + WREG32(RADEON_TV_DAC_CNTL, tmp); + + tmp = RADEON_DAC2_FORCE_BLANK_OFF_EN | + RADEON_DAC2_FORCE_DATA_EN; + + if (color) + tmp |= RADEON_DAC_FORCE_DATA_SEL_RGB; + else + tmp |= RADEON_DAC_FORCE_DATA_SEL_G; + + if (ASIC_IS_R300(rdev)) + tmp |= (0x1b6 << RADEON_DAC_FORCE_DATA_SHIFT); + else + tmp |= (0x180 << RADEON_DAC_FORCE_DATA_SHIFT); + + WREG32(RADEON_DAC_EXT_CNTL, tmp); + + tmp = dac_cntl2 | RADEON_DAC2_DAC2_CLK_SEL | RADEON_DAC2_CMP_EN; + WREG32(RADEON_DAC_CNTL2, tmp); + + udelay(10000); + + if (ASIC_IS_R300(rdev)) { + if (RREG32(RADEON_DAC_CNTL2) & RADEON_DAC2_CMP_OUT_B) + found = connector_status_connected; + } else { + if (RREG32(RADEON_DAC_CNTL2) & RADEON_DAC2_CMP_OUTPUT) + found = connector_status_connected; + } + + /* restore regs we used */ + WREG32(RADEON_DAC_CNTL2, dac_cntl2); + WREG32(RADEON_DAC_EXT_CNTL, dac_ext_cntl); + WREG32(RADEON_TV_DAC_CNTL, tv_dac_cntl); + WREG32(RADEON_CRTC2_GEN_CNTL, crtc2_gen_cntl); + + if (ASIC_IS_R300(rdev)) { + WREG32(RADEON_DISP_OUTPUT_CNTL, disp_output_cntl); + WREG32_P(RADEON_GPIOPAD_A, gpiopad_a, ~1); + } else { + WREG32(RADEON_DISP_HW_DEBUG, disp_hw_debug); + } + WREG32_PLL(RADEON_PIXCLKS_CNTL, pixclks_cntl); + + return found; + +} + +static const struct drm_encoder_helper_funcs radeon_legacy_tv_dac_helper_funcs = { + .dpms = radeon_legacy_tv_dac_dpms, + .mode_fixup = radeon_legacy_mode_fixup, + .prepare = radeon_legacy_tv_dac_prepare, + .mode_set = radeon_legacy_tv_dac_mode_set, + .commit = radeon_legacy_tv_dac_commit, + .detect = radeon_legacy_tv_dac_detect, + .disable = radeon_legacy_encoder_disable, +}; + + +static const struct drm_encoder_funcs radeon_legacy_tv_dac_enc_funcs = { + .destroy = radeon_enc_destroy, +}; + + +static struct radeon_encoder_int_tmds *radeon_legacy_get_tmds_info(struct radeon_encoder *encoder) +{ + struct drm_device *dev = encoder->base.dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder_int_tmds *tmds = NULL; + bool ret; + + tmds = kzalloc(sizeof(struct radeon_encoder_int_tmds), GFP_KERNEL); + + if (!tmds) + return NULL; + + if (rdev->is_atom_bios) + ret = radeon_atombios_get_tmds_info(encoder, tmds); + else + ret = radeon_legacy_get_tmds_info_from_combios(encoder, tmds); + + if (ret == false) + radeon_legacy_get_tmds_info_from_table(encoder, tmds); + + return tmds; +} + +static struct radeon_encoder_ext_tmds *radeon_legacy_get_ext_tmds_info(struct radeon_encoder *encoder) +{ + struct drm_device *dev = encoder->base.dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder_ext_tmds *tmds = NULL; + bool ret; + + if (rdev->is_atom_bios) + return NULL; + + tmds = kzalloc(sizeof(struct radeon_encoder_ext_tmds), GFP_KERNEL); + + if (!tmds) + return NULL; + + ret = radeon_legacy_get_ext_tmds_info_from_combios(encoder, tmds); + + if (ret == false) + radeon_legacy_get_ext_tmds_info_from_table(encoder, tmds); + + return tmds; +} + +void +radeon_add_legacy_encoder(struct drm_device *dev, uint32_t encoder_enum, uint32_t supported_device) +{ + struct radeon_device *rdev = dev->dev_private; + struct drm_encoder *encoder; + struct radeon_encoder *radeon_encoder; + + /* see if we already added it */ + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + radeon_encoder = to_radeon_encoder(encoder); + if (radeon_encoder->encoder_enum == encoder_enum) { + radeon_encoder->devices |= supported_device; + return; + } + + } + + /* add a new one */ + radeon_encoder = kzalloc(sizeof(struct radeon_encoder), GFP_KERNEL); + if (!radeon_encoder) + return; + + encoder = &radeon_encoder->base; + if (rdev->flags & RADEON_SINGLE_CRTC) + encoder->possible_crtcs = 0x1; + else + encoder->possible_crtcs = 0x3; + + radeon_encoder->enc_priv = NULL; + + radeon_encoder->encoder_enum = encoder_enum; + radeon_encoder->encoder_id = (encoder_enum & OBJECT_ID_MASK) >> OBJECT_ID_SHIFT; + radeon_encoder->devices = supported_device; + radeon_encoder->rmx_type = RMX_OFF; + + switch (radeon_encoder->encoder_id) { + case ENCODER_OBJECT_ID_INTERNAL_LVDS: + encoder->possible_crtcs = 0x1; + drm_encoder_init(dev, encoder, &radeon_legacy_lvds_enc_funcs, DRM_MODE_ENCODER_LVDS); + drm_encoder_helper_add(encoder, &radeon_legacy_lvds_helper_funcs); + if (rdev->is_atom_bios) + radeon_encoder->enc_priv = radeon_atombios_get_lvds_info(radeon_encoder); + else + radeon_encoder->enc_priv = radeon_combios_get_lvds_info(radeon_encoder); + radeon_encoder->rmx_type = RMX_FULL; + break; + case ENCODER_OBJECT_ID_INTERNAL_TMDS1: + drm_encoder_init(dev, encoder, &radeon_legacy_tmds_int_enc_funcs, DRM_MODE_ENCODER_TMDS); + drm_encoder_helper_add(encoder, &radeon_legacy_tmds_int_helper_funcs); + radeon_encoder->enc_priv = radeon_legacy_get_tmds_info(radeon_encoder); + break; + case ENCODER_OBJECT_ID_INTERNAL_DAC1: + drm_encoder_init(dev, encoder, &radeon_legacy_primary_dac_enc_funcs, DRM_MODE_ENCODER_DAC); + drm_encoder_helper_add(encoder, &radeon_legacy_primary_dac_helper_funcs); + if (rdev->is_atom_bios) + radeon_encoder->enc_priv = radeon_atombios_get_primary_dac_info(radeon_encoder); + else + radeon_encoder->enc_priv = radeon_combios_get_primary_dac_info(radeon_encoder); + break; + case ENCODER_OBJECT_ID_INTERNAL_DAC2: + drm_encoder_init(dev, encoder, &radeon_legacy_tv_dac_enc_funcs, DRM_MODE_ENCODER_TVDAC); + drm_encoder_helper_add(encoder, &radeon_legacy_tv_dac_helper_funcs); + if (rdev->is_atom_bios) + radeon_encoder->enc_priv = radeon_atombios_get_tv_dac_info(radeon_encoder); + else + radeon_encoder->enc_priv = radeon_combios_get_tv_dac_info(radeon_encoder); + break; + case ENCODER_OBJECT_ID_INTERNAL_DVO1: + drm_encoder_init(dev, encoder, &radeon_legacy_tmds_ext_enc_funcs, DRM_MODE_ENCODER_TMDS); + drm_encoder_helper_add(encoder, &radeon_legacy_tmds_ext_helper_funcs); + if (!rdev->is_atom_bios) + radeon_encoder->enc_priv = radeon_legacy_get_ext_tmds_info(radeon_encoder); + break; + } +} diff --git a/drivers/gpu/drm/radeon/radeon_legacy_tv.c b/drivers/gpu/drm/radeon/radeon_legacy_tv.c new file mode 100644 index 00000000..c7b6cb42 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_legacy_tv.c @@ -0,0 +1,923 @@ +#include "drmP.h" +#include "drm_crtc_helper.h" +#include "radeon.h" + +/* + * Integrated TV out support based on the GATOS code by + * Federico Ulivi + */ + + +/* + * Limits of h/v positions (hPos & vPos) + */ +#define MAX_H_POSITION 5 /* Range: [-5..5], negative is on the left, 0 is default, positive is on the right */ +#define MAX_V_POSITION 5 /* Range: [-5..5], negative is up, 0 is default, positive is down */ + +/* + * Unit for hPos (in TV clock periods) + */ +#define H_POS_UNIT 10 + +/* + * Indexes in h. code timing table for horizontal line position adjustment + */ +#define H_TABLE_POS1 6 +#define H_TABLE_POS2 8 + +/* + * Limits of hor. size (hSize) + */ +#define MAX_H_SIZE 5 /* Range: [-5..5], negative is smaller, positive is larger */ + +/* tv standard constants */ +#define NTSC_TV_CLOCK_T 233 +#define NTSC_TV_VFTOTAL 1 +#define NTSC_TV_LINES_PER_FRAME 525 +#define NTSC_TV_ZERO_H_SIZE 479166 +#define NTSC_TV_H_SIZE_UNIT 9478 + +#define PAL_TV_CLOCK_T 188 +#define PAL_TV_VFTOTAL 3 +#define PAL_TV_LINES_PER_FRAME 625 +#define PAL_TV_ZERO_H_SIZE 473200 +#define PAL_TV_H_SIZE_UNIT 9360 + +/* tv pll setting for 27 mhz ref clk */ +#define NTSC_TV_PLL_M_27 22 +#define NTSC_TV_PLL_N_27 175 +#define NTSC_TV_PLL_P_27 5 + +#define PAL_TV_PLL_M_27 113 +#define PAL_TV_PLL_N_27 668 +#define PAL_TV_PLL_P_27 3 + +/* tv pll setting for 14 mhz ref clk */ +#define NTSC_TV_PLL_M_14 33 +#define NTSC_TV_PLL_N_14 693 +#define NTSC_TV_PLL_P_14 7 + +#define PAL_TV_PLL_M_14 19 +#define PAL_TV_PLL_N_14 353 +#define PAL_TV_PLL_P_14 5 + +#define VERT_LEAD_IN_LINES 2 +#define FRAC_BITS 0xe +#define FRAC_MASK 0x3fff + +struct radeon_tv_mode_constants { + uint16_t hor_resolution; + uint16_t ver_resolution; + enum radeon_tv_std standard; + uint16_t hor_total; + uint16_t ver_total; + uint16_t hor_start; + uint16_t hor_syncstart; + uint16_t ver_syncstart; + unsigned def_restart; + uint16_t crtcPLL_N; + uint8_t crtcPLL_M; + uint8_t crtcPLL_post_div; + unsigned pix_to_tv; +}; + +static const uint16_t hor_timing_NTSC[MAX_H_CODE_TIMING_LEN] = { + 0x0007, + 0x003f, + 0x0263, + 0x0a24, + 0x2a6b, + 0x0a36, + 0x126d, /* H_TABLE_POS1 */ + 0x1bfe, + 0x1a8f, /* H_TABLE_POS2 */ + 0x1ec7, + 0x3863, + 0x1bfe, + 0x1bfe, + 0x1a2a, + 0x1e95, + 0x0e31, + 0x201b, + 0 +}; + +static const uint16_t vert_timing_NTSC[MAX_V_CODE_TIMING_LEN] = { + 0x2001, + 0x200d, + 0x1006, + 0x0c06, + 0x1006, + 0x1818, + 0x21e3, + 0x1006, + 0x0c06, + 0x1006, + 0x1817, + 0x21d4, + 0x0002, + 0 +}; + +static const uint16_t hor_timing_PAL[MAX_H_CODE_TIMING_LEN] = { + 0x0007, + 0x0058, + 0x027c, + 0x0a31, + 0x2a77, + 0x0a95, + 0x124f, /* H_TABLE_POS1 */ + 0x1bfe, + 0x1b22, /* H_TABLE_POS2 */ + 0x1ef9, + 0x387c, + 0x1bfe, + 0x1bfe, + 0x1b31, + 0x1eb5, + 0x0e43, + 0x201b, + 0 +}; + +static const uint16_t vert_timing_PAL[MAX_V_CODE_TIMING_LEN] = { + 0x2001, + 0x200c, + 0x1005, + 0x0c05, + 0x1005, + 0x1401, + 0x1821, + 0x2240, + 0x1005, + 0x0c05, + 0x1005, + 0x1401, + 0x1822, + 0x2230, + 0x0002, + 0 +}; + +/********************************************************************** + * + * availableModes + * + * Table of all allowed modes for tv output + * + **********************************************************************/ +static const struct radeon_tv_mode_constants available_tv_modes[] = { + { /* NTSC timing for 27 Mhz ref clk */ + 800, /* horResolution */ + 600, /* verResolution */ + TV_STD_NTSC, /* standard */ + 990, /* horTotal */ + 740, /* verTotal */ + 813, /* horStart */ + 824, /* horSyncStart */ + 632, /* verSyncStart */ + 625592, /* defRestart */ + 592, /* crtcPLL_N */ + 91, /* crtcPLL_M */ + 4, /* crtcPLL_postDiv */ + 1022, /* pixToTV */ + }, + { /* PAL timing for 27 Mhz ref clk */ + 800, /* horResolution */ + 600, /* verResolution */ + TV_STD_PAL, /* standard */ + 1144, /* horTotal */ + 706, /* verTotal */ + 812, /* horStart */ + 824, /* horSyncStart */ + 669, /* verSyncStart */ + 696700, /* defRestart */ + 1382, /* crtcPLL_N */ + 231, /* crtcPLL_M */ + 4, /* crtcPLL_postDiv */ + 759, /* pixToTV */ + }, + { /* NTSC timing for 14 Mhz ref clk */ + 800, /* horResolution */ + 600, /* verResolution */ + TV_STD_NTSC, /* standard */ + 1018, /* horTotal */ + 727, /* verTotal */ + 813, /* horStart */ + 840, /* horSyncStart */ + 633, /* verSyncStart */ + 630627, /* defRestart */ + 347, /* crtcPLL_N */ + 14, /* crtcPLL_M */ + 8, /* crtcPLL_postDiv */ + 1022, /* pixToTV */ + }, + { /* PAL timing for 14 Mhz ref clk */ + 800, /* horResolution */ + 600, /* verResolution */ + TV_STD_PAL, /* standard */ + 1131, /* horTotal */ + 742, /* verTotal */ + 813, /* horStart */ + 840, /* horSyncStart */ + 633, /* verSyncStart */ + 708369, /* defRestart */ + 211, /* crtcPLL_N */ + 9, /* crtcPLL_M */ + 8, /* crtcPLL_postDiv */ + 759, /* pixToTV */ + }, +}; + +#define N_AVAILABLE_MODES ARRAY_SIZE(available_tv_modes) + +static const struct radeon_tv_mode_constants *radeon_legacy_tv_get_std_mode(struct radeon_encoder *radeon_encoder, + uint16_t *pll_ref_freq) +{ + struct drm_device *dev = radeon_encoder->base.dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_crtc *radeon_crtc; + struct radeon_encoder_tv_dac *tv_dac = radeon_encoder->enc_priv; + const struct radeon_tv_mode_constants *const_ptr; + struct radeon_pll *pll; + + radeon_crtc = to_radeon_crtc(radeon_encoder->base.crtc); + if (radeon_crtc->crtc_id == 1) + pll = &rdev->clock.p2pll; + else + pll = &rdev->clock.p1pll; + + if (pll_ref_freq) + *pll_ref_freq = pll->reference_freq; + + if (tv_dac->tv_std == TV_STD_NTSC || + tv_dac->tv_std == TV_STD_NTSC_J || + tv_dac->tv_std == TV_STD_PAL_M) { + if (pll->reference_freq == 2700) + const_ptr = &available_tv_modes[0]; + else + const_ptr = &available_tv_modes[2]; + } else { + if (pll->reference_freq == 2700) + const_ptr = &available_tv_modes[1]; + else + const_ptr = &available_tv_modes[3]; + } + return const_ptr; +} + +static long YCOEF_value[5] = { 2, 2, 0, 4, 0 }; +static long YCOEF_EN_value[5] = { 1, 1, 0, 1, 0 }; +static long SLOPE_value[5] = { 1, 2, 2, 4, 8 }; +static long SLOPE_limit[5] = { 6, 5, 4, 3, 2 }; + +static void radeon_wait_pll_lock(struct drm_encoder *encoder, unsigned n_tests, + unsigned n_wait_loops, unsigned cnt_threshold) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + uint32_t save_pll_test; + unsigned int i, j; + + WREG32(RADEON_TEST_DEBUG_MUX, (RREG32(RADEON_TEST_DEBUG_MUX) & 0xffff60ff) | 0x100); + save_pll_test = RREG32_PLL(RADEON_PLL_TEST_CNTL); + WREG32_PLL(RADEON_PLL_TEST_CNTL, save_pll_test & ~RADEON_PLL_MASK_READ_B); + + WREG8(RADEON_CLOCK_CNTL_INDEX, RADEON_PLL_TEST_CNTL); + for (i = 0; i < n_tests; i++) { + WREG8(RADEON_CLOCK_CNTL_DATA + 3, 0); + for (j = 0; j < n_wait_loops; j++) + if (RREG8(RADEON_CLOCK_CNTL_DATA + 3) >= cnt_threshold) + break; + } + WREG32_PLL(RADEON_PLL_TEST_CNTL, save_pll_test); + WREG32(RADEON_TEST_DEBUG_MUX, RREG32(RADEON_TEST_DEBUG_MUX) & 0xffffe0ff); +} + + +static void radeon_legacy_tv_write_fifo(struct radeon_encoder *radeon_encoder, + uint16_t addr, uint32_t value) +{ + struct drm_device *dev = radeon_encoder->base.dev; + struct radeon_device *rdev = dev->dev_private; + uint32_t tmp; + int i = 0; + + WREG32(RADEON_TV_HOST_WRITE_DATA, value); + + WREG32(RADEON_TV_HOST_RD_WT_CNTL, addr); + WREG32(RADEON_TV_HOST_RD_WT_CNTL, addr | RADEON_HOST_FIFO_WT); + + do { + tmp = RREG32(RADEON_TV_HOST_RD_WT_CNTL); + if ((tmp & RADEON_HOST_FIFO_WT_ACK) == 0) + break; + i++; + } while (i < 10000); + WREG32(RADEON_TV_HOST_RD_WT_CNTL, 0); +} + +#if 0 /* included for completeness */ +static uint32_t radeon_legacy_tv_read_fifo(struct radeon_encoder *radeon_encoder, uint16_t addr) +{ + struct drm_device *dev = radeon_encoder->base.dev; + struct radeon_device *rdev = dev->dev_private; + uint32_t tmp; + int i = 0; + + WREG32(RADEON_TV_HOST_RD_WT_CNTL, addr); + WREG32(RADEON_TV_HOST_RD_WT_CNTL, addr | RADEON_HOST_FIFO_RD); + + do { + tmp = RREG32(RADEON_TV_HOST_RD_WT_CNTL); + if ((tmp & RADEON_HOST_FIFO_RD_ACK) == 0) + break; + i++; + } while (i < 10000); + WREG32(RADEON_TV_HOST_RD_WT_CNTL, 0); + return RREG32(RADEON_TV_HOST_READ_DATA); +} +#endif + +static uint16_t radeon_get_htiming_tables_addr(uint32_t tv_uv_adr) +{ + uint16_t h_table; + + switch ((tv_uv_adr & RADEON_HCODE_TABLE_SEL_MASK) >> RADEON_HCODE_TABLE_SEL_SHIFT) { + case 0: + h_table = RADEON_TV_MAX_FIFO_ADDR_INTERNAL; + break; + case 1: + h_table = ((tv_uv_adr & RADEON_TABLE1_BOT_ADR_MASK) >> RADEON_TABLE1_BOT_ADR_SHIFT) * 2; + break; + case 2: + h_table = ((tv_uv_adr & RADEON_TABLE3_TOP_ADR_MASK) >> RADEON_TABLE3_TOP_ADR_SHIFT) * 2; + break; + default: + h_table = 0; + break; + } + return h_table; +} + +static uint16_t radeon_get_vtiming_tables_addr(uint32_t tv_uv_adr) +{ + uint16_t v_table; + + switch ((tv_uv_adr & RADEON_VCODE_TABLE_SEL_MASK) >> RADEON_VCODE_TABLE_SEL_SHIFT) { + case 0: + v_table = ((tv_uv_adr & RADEON_MAX_UV_ADR_MASK) >> RADEON_MAX_UV_ADR_SHIFT) * 2 + 1; + break; + case 1: + v_table = ((tv_uv_adr & RADEON_TABLE1_BOT_ADR_MASK) >> RADEON_TABLE1_BOT_ADR_SHIFT) * 2 + 1; + break; + case 2: + v_table = ((tv_uv_adr & RADEON_TABLE3_TOP_ADR_MASK) >> RADEON_TABLE3_TOP_ADR_SHIFT) * 2 + 1; + break; + default: + v_table = 0; + break; + } + return v_table; +} + +static void radeon_restore_tv_timing_tables(struct radeon_encoder *radeon_encoder) +{ + struct drm_device *dev = radeon_encoder->base.dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder_tv_dac *tv_dac = radeon_encoder->enc_priv; + uint16_t h_table, v_table; + uint32_t tmp; + int i; + + WREG32(RADEON_TV_UV_ADR, tv_dac->tv.tv_uv_adr); + h_table = radeon_get_htiming_tables_addr(tv_dac->tv.tv_uv_adr); + v_table = radeon_get_vtiming_tables_addr(tv_dac->tv.tv_uv_adr); + + for (i = 0; i < MAX_H_CODE_TIMING_LEN; i += 2, h_table--) { + tmp = ((uint32_t)tv_dac->tv.h_code_timing[i] << 14) | ((uint32_t)tv_dac->tv.h_code_timing[i+1]); + radeon_legacy_tv_write_fifo(radeon_encoder, h_table, tmp); + if (tv_dac->tv.h_code_timing[i] == 0 || tv_dac->tv.h_code_timing[i + 1] == 0) + break; + } + for (i = 0; i < MAX_V_CODE_TIMING_LEN; i += 2, v_table++) { + tmp = ((uint32_t)tv_dac->tv.v_code_timing[i+1] << 14) | ((uint32_t)tv_dac->tv.v_code_timing[i]); + radeon_legacy_tv_write_fifo(radeon_encoder, v_table, tmp); + if (tv_dac->tv.v_code_timing[i] == 0 || tv_dac->tv.v_code_timing[i + 1] == 0) + break; + } +} + +static void radeon_legacy_write_tv_restarts(struct radeon_encoder *radeon_encoder) +{ + struct drm_device *dev = radeon_encoder->base.dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder_tv_dac *tv_dac = radeon_encoder->enc_priv; + WREG32(RADEON_TV_FRESTART, tv_dac->tv.frestart); + WREG32(RADEON_TV_HRESTART, tv_dac->tv.hrestart); + WREG32(RADEON_TV_VRESTART, tv_dac->tv.vrestart); +} + +static bool radeon_legacy_tv_init_restarts(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct radeon_encoder_tv_dac *tv_dac = radeon_encoder->enc_priv; + struct radeon_crtc *radeon_crtc; + int restart; + unsigned int h_total, v_total, f_total; + int v_offset, h_offset; + u16 p1, p2, h_inc; + bool h_changed; + const struct radeon_tv_mode_constants *const_ptr; + struct radeon_pll *pll; + + radeon_crtc = to_radeon_crtc(radeon_encoder->base.crtc); + if (radeon_crtc->crtc_id == 1) + pll = &rdev->clock.p2pll; + else + pll = &rdev->clock.p1pll; + + const_ptr = radeon_legacy_tv_get_std_mode(radeon_encoder, NULL); + if (!const_ptr) + return false; + + h_total = const_ptr->hor_total; + v_total = const_ptr->ver_total; + + if (tv_dac->tv_std == TV_STD_NTSC || + tv_dac->tv_std == TV_STD_NTSC_J || + tv_dac->tv_std == TV_STD_PAL_M || + tv_dac->tv_std == TV_STD_PAL_60) + f_total = NTSC_TV_VFTOTAL + 1; + else + f_total = PAL_TV_VFTOTAL + 1; + + /* adjust positions 1&2 in hor. cod timing table */ + h_offset = tv_dac->h_pos * H_POS_UNIT; + + if (tv_dac->tv_std == TV_STD_NTSC || + tv_dac->tv_std == TV_STD_NTSC_J || + tv_dac->tv_std == TV_STD_PAL_M) { + h_offset -= 50; + p1 = hor_timing_NTSC[H_TABLE_POS1]; + p2 = hor_timing_NTSC[H_TABLE_POS2]; + } else { + p1 = hor_timing_PAL[H_TABLE_POS1]; + p2 = hor_timing_PAL[H_TABLE_POS2]; + } + + p1 = (u16)((int)p1 + h_offset); + p2 = (u16)((int)p2 - h_offset); + + h_changed = (p1 != tv_dac->tv.h_code_timing[H_TABLE_POS1] || + p2 != tv_dac->tv.h_code_timing[H_TABLE_POS2]); + + tv_dac->tv.h_code_timing[H_TABLE_POS1] = p1; + tv_dac->tv.h_code_timing[H_TABLE_POS2] = p2; + + /* Convert hOffset from n. of TV clock periods to n. of CRTC clock periods (CRTC pixels) */ + h_offset = (h_offset * (int)(const_ptr->pix_to_tv)) / 1000; + + /* adjust restart */ + restart = const_ptr->def_restart; + + /* + * convert v_pos TV lines to n. of CRTC pixels + */ + if (tv_dac->tv_std == TV_STD_NTSC || + tv_dac->tv_std == TV_STD_NTSC_J || + tv_dac->tv_std == TV_STD_PAL_M || + tv_dac->tv_std == TV_STD_PAL_60) + v_offset = ((int)(v_total * h_total) * 2 * tv_dac->v_pos) / (int)(NTSC_TV_LINES_PER_FRAME); + else + v_offset = ((int)(v_total * h_total) * 2 * tv_dac->v_pos) / (int)(PAL_TV_LINES_PER_FRAME); + + restart -= v_offset + h_offset; + + DRM_DEBUG_KMS("compute_restarts: def = %u h = %d v = %d, p1 = %04x, p2 = %04x, restart = %d\n", + const_ptr->def_restart, tv_dac->h_pos, tv_dac->v_pos, p1, p2, restart); + + tv_dac->tv.hrestart = restart % h_total; + restart /= h_total; + tv_dac->tv.vrestart = restart % v_total; + restart /= v_total; + tv_dac->tv.frestart = restart % f_total; + + DRM_DEBUG_KMS("compute_restart: F/H/V=%u,%u,%u\n", + (unsigned)tv_dac->tv.frestart, + (unsigned)tv_dac->tv.vrestart, + (unsigned)tv_dac->tv.hrestart); + + /* compute h_inc from hsize */ + if (tv_dac->tv_std == TV_STD_NTSC || + tv_dac->tv_std == TV_STD_NTSC_J || + tv_dac->tv_std == TV_STD_PAL_M) + h_inc = (u16)((int)(const_ptr->hor_resolution * 4096 * NTSC_TV_CLOCK_T) / + (tv_dac->h_size * (int)(NTSC_TV_H_SIZE_UNIT) + (int)(NTSC_TV_ZERO_H_SIZE))); + else + h_inc = (u16)((int)(const_ptr->hor_resolution * 4096 * PAL_TV_CLOCK_T) / + (tv_dac->h_size * (int)(PAL_TV_H_SIZE_UNIT) + (int)(PAL_TV_ZERO_H_SIZE))); + + tv_dac->tv.timing_cntl = (tv_dac->tv.timing_cntl & ~RADEON_H_INC_MASK) | + ((u32)h_inc << RADEON_H_INC_SHIFT); + + DRM_DEBUG_KMS("compute_restart: h_size = %d h_inc = %d\n", tv_dac->h_size, h_inc); + + return h_changed; +} + +void radeon_legacy_tv_mode_set(struct drm_encoder *encoder, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct radeon_encoder_tv_dac *tv_dac = radeon_encoder->enc_priv; + const struct radeon_tv_mode_constants *const_ptr; + struct radeon_crtc *radeon_crtc; + int i; + uint16_t pll_ref_freq; + uint32_t vert_space, flicker_removal, tmp; + uint32_t tv_master_cntl, tv_rgb_cntl, tv_dac_cntl; + uint32_t tv_modulator_cntl1, tv_modulator_cntl2; + uint32_t tv_vscaler_cntl1, tv_vscaler_cntl2; + uint32_t tv_pll_cntl, tv_pll_cntl1, tv_ftotal; + uint32_t tv_y_fall_cntl, tv_y_rise_cntl, tv_y_saw_tooth_cntl; + uint32_t m, n, p; + const uint16_t *hor_timing; + const uint16_t *vert_timing; + + const_ptr = radeon_legacy_tv_get_std_mode(radeon_encoder, &pll_ref_freq); + if (!const_ptr) + return; + + radeon_crtc = to_radeon_crtc(encoder->crtc); + + tv_master_cntl = (RADEON_VIN_ASYNC_RST | + RADEON_CRT_FIFO_CE_EN | + RADEON_TV_FIFO_CE_EN | + RADEON_TV_ON); + + if (!ASIC_IS_R300(rdev)) + tv_master_cntl |= RADEON_TVCLK_ALWAYS_ONb; + + if (tv_dac->tv_std == TV_STD_NTSC || + tv_dac->tv_std == TV_STD_NTSC_J) + tv_master_cntl |= RADEON_RESTART_PHASE_FIX; + + tv_modulator_cntl1 = (RADEON_SLEW_RATE_LIMIT | + RADEON_SYNC_TIP_LEVEL | + RADEON_YFLT_EN | + RADEON_UVFLT_EN | + (6 << RADEON_CY_FILT_BLEND_SHIFT)); + + if (tv_dac->tv_std == TV_STD_NTSC || + tv_dac->tv_std == TV_STD_NTSC_J) { + tv_modulator_cntl1 |= (0x46 << RADEON_SET_UP_LEVEL_SHIFT) | + (0x3b << RADEON_BLANK_LEVEL_SHIFT); + tv_modulator_cntl2 = (-111 & RADEON_TV_U_BURST_LEVEL_MASK) | + ((0 & RADEON_TV_V_BURST_LEVEL_MASK) << RADEON_TV_V_BURST_LEVEL_SHIFT); + } else if (tv_dac->tv_std == TV_STD_SCART_PAL) { + tv_modulator_cntl1 |= RADEON_ALT_PHASE_EN; + tv_modulator_cntl2 = (0 & RADEON_TV_U_BURST_LEVEL_MASK) | + ((0 & RADEON_TV_V_BURST_LEVEL_MASK) << RADEON_TV_V_BURST_LEVEL_SHIFT); + } else { + tv_modulator_cntl1 |= RADEON_ALT_PHASE_EN | + (0x3b << RADEON_SET_UP_LEVEL_SHIFT) | + (0x3b << RADEON_BLANK_LEVEL_SHIFT); + tv_modulator_cntl2 = (-78 & RADEON_TV_U_BURST_LEVEL_MASK) | + ((62 & RADEON_TV_V_BURST_LEVEL_MASK) << RADEON_TV_V_BURST_LEVEL_SHIFT); + } + + + tv_rgb_cntl = (RADEON_RGB_DITHER_EN + | RADEON_TVOUT_SCALE_EN + | (0x0b << RADEON_UVRAM_READ_MARGIN_SHIFT) + | (0x07 << RADEON_FIFORAM_FFMACRO_READ_MARGIN_SHIFT) + | RADEON_RGB_ATTEN_SEL(0x3) + | RADEON_RGB_ATTEN_VAL(0xc)); + + if (radeon_crtc->crtc_id == 1) + tv_rgb_cntl |= RADEON_RGB_SRC_SEL_CRTC2; + else { + if (radeon_crtc->rmx_type != RMX_OFF) + tv_rgb_cntl |= RADEON_RGB_SRC_SEL_RMX; + else + tv_rgb_cntl |= RADEON_RGB_SRC_SEL_CRTC1; + } + + if (tv_dac->tv_std == TV_STD_NTSC || + tv_dac->tv_std == TV_STD_NTSC_J || + tv_dac->tv_std == TV_STD_PAL_M || + tv_dac->tv_std == TV_STD_PAL_60) + vert_space = const_ptr->ver_total * 2 * 10000 / NTSC_TV_LINES_PER_FRAME; + else + vert_space = const_ptr->ver_total * 2 * 10000 / PAL_TV_LINES_PER_FRAME; + + tmp = RREG32(RADEON_TV_VSCALER_CNTL1); + tmp &= 0xe3ff0000; + tmp |= (vert_space * (1 << FRAC_BITS) / 10000); + tv_vscaler_cntl1 = tmp; + + if (pll_ref_freq == 2700) + tv_vscaler_cntl1 |= RADEON_RESTART_FIELD; + + if (const_ptr->hor_resolution == 1024) + tv_vscaler_cntl1 |= (4 << RADEON_Y_DEL_W_SIG_SHIFT); + else + tv_vscaler_cntl1 |= (2 << RADEON_Y_DEL_W_SIG_SHIFT); + + /* scale up for int divide */ + tmp = const_ptr->ver_total * 2 * 1000; + if (tv_dac->tv_std == TV_STD_NTSC || + tv_dac->tv_std == TV_STD_NTSC_J || + tv_dac->tv_std == TV_STD_PAL_M || + tv_dac->tv_std == TV_STD_PAL_60) { + tmp /= NTSC_TV_LINES_PER_FRAME; + } else { + tmp /= PAL_TV_LINES_PER_FRAME; + } + flicker_removal = (tmp + 500) / 1000; + + if (flicker_removal < 3) + flicker_removal = 3; + for (i = 0; i < ARRAY_SIZE(SLOPE_limit); ++i) { + if (flicker_removal == SLOPE_limit[i]) + break; + } + + tv_y_saw_tooth_cntl = (vert_space * SLOPE_value[i] * (1 << (FRAC_BITS - 1)) + + 5001) / 10000 / 8 | ((SLOPE_value[i] * + (1 << (FRAC_BITS - 1)) / 8) << 16); + tv_y_fall_cntl = + (YCOEF_EN_value[i] << 17) | ((YCOEF_value[i] * (1 << 8) / 8) << 24) | + RADEON_Y_FALL_PING_PONG | (272 * SLOPE_value[i] / 8) * (1 << (FRAC_BITS - 1)) / + 1024; + tv_y_rise_cntl = RADEON_Y_RISE_PING_PONG| + (flicker_removal * 1024 - 272) * SLOPE_value[i] / 8 * (1 << (FRAC_BITS - 1)) / 1024; + + tv_vscaler_cntl2 = RREG32(RADEON_TV_VSCALER_CNTL2) & 0x00fffff0; + tv_vscaler_cntl2 |= (0x10 << 24) | + RADEON_DITHER_MODE | + RADEON_Y_OUTPUT_DITHER_EN | + RADEON_UV_OUTPUT_DITHER_EN | + RADEON_UV_TO_BUF_DITHER_EN; + + tmp = (tv_vscaler_cntl1 >> RADEON_UV_INC_SHIFT) & RADEON_UV_INC_MASK; + tmp = ((16384 * 256 * 10) / tmp + 5) / 10; + tmp = (tmp << RADEON_UV_OUTPUT_POST_SCALE_SHIFT) | 0x000b0000; + tv_dac->tv.timing_cntl = tmp; + + if (tv_dac->tv_std == TV_STD_NTSC || + tv_dac->tv_std == TV_STD_NTSC_J || + tv_dac->tv_std == TV_STD_PAL_M || + tv_dac->tv_std == TV_STD_PAL_60) + tv_dac_cntl = tv_dac->ntsc_tvdac_adj; + else + tv_dac_cntl = tv_dac->pal_tvdac_adj; + + tv_dac_cntl |= RADEON_TV_DAC_NBLANK | RADEON_TV_DAC_NHOLD; + + if (tv_dac->tv_std == TV_STD_NTSC || + tv_dac->tv_std == TV_STD_NTSC_J) + tv_dac_cntl |= RADEON_TV_DAC_STD_NTSC; + else + tv_dac_cntl |= RADEON_TV_DAC_STD_PAL; + + if (tv_dac->tv_std == TV_STD_NTSC || + tv_dac->tv_std == TV_STD_NTSC_J) { + if (pll_ref_freq == 2700) { + m = NTSC_TV_PLL_M_27; + n = NTSC_TV_PLL_N_27; + p = NTSC_TV_PLL_P_27; + } else { + m = NTSC_TV_PLL_M_14; + n = NTSC_TV_PLL_N_14; + p = NTSC_TV_PLL_P_14; + } + } else { + if (pll_ref_freq == 2700) { + m = PAL_TV_PLL_M_27; + n = PAL_TV_PLL_N_27; + p = PAL_TV_PLL_P_27; + } else { + m = PAL_TV_PLL_M_14; + n = PAL_TV_PLL_N_14; + p = PAL_TV_PLL_P_14; + } + } + + tv_pll_cntl = (m & RADEON_TV_M0LO_MASK) | + (((m >> 8) & RADEON_TV_M0HI_MASK) << RADEON_TV_M0HI_SHIFT) | + ((n & RADEON_TV_N0LO_MASK) << RADEON_TV_N0LO_SHIFT) | + (((n >> 9) & RADEON_TV_N0HI_MASK) << RADEON_TV_N0HI_SHIFT) | + ((p & RADEON_TV_P_MASK) << RADEON_TV_P_SHIFT); + + tv_pll_cntl1 = (((4 & RADEON_TVPCP_MASK) << RADEON_TVPCP_SHIFT) | + ((4 & RADEON_TVPVG_MASK) << RADEON_TVPVG_SHIFT) | + ((1 & RADEON_TVPDC_MASK) << RADEON_TVPDC_SHIFT) | + RADEON_TVCLK_SRC_SEL_TVPLL | + RADEON_TVPLL_TEST_DIS); + + tv_dac->tv.tv_uv_adr = 0xc8; + + if (tv_dac->tv_std == TV_STD_NTSC || + tv_dac->tv_std == TV_STD_NTSC_J || + tv_dac->tv_std == TV_STD_PAL_M || + tv_dac->tv_std == TV_STD_PAL_60) { + tv_ftotal = NTSC_TV_VFTOTAL; + hor_timing = hor_timing_NTSC; + vert_timing = vert_timing_NTSC; + } else { + hor_timing = hor_timing_PAL; + vert_timing = vert_timing_PAL; + tv_ftotal = PAL_TV_VFTOTAL; + } + + for (i = 0; i < MAX_H_CODE_TIMING_LEN; i++) { + if ((tv_dac->tv.h_code_timing[i] = hor_timing[i]) == 0) + break; + } + + for (i = 0; i < MAX_V_CODE_TIMING_LEN; i++) { + if ((tv_dac->tv.v_code_timing[i] = vert_timing[i]) == 0) + break; + } + + radeon_legacy_tv_init_restarts(encoder); + + /* play with DAC_CNTL */ + /* play with GPIOPAD_A */ + /* DISP_OUTPUT_CNTL */ + /* use reference freq */ + + /* program the TV registers */ + WREG32(RADEON_TV_MASTER_CNTL, (tv_master_cntl | RADEON_TV_ASYNC_RST | + RADEON_CRT_ASYNC_RST | RADEON_TV_FIFO_ASYNC_RST)); + + tmp = RREG32(RADEON_TV_DAC_CNTL); + tmp &= ~RADEON_TV_DAC_NBLANK; + tmp |= RADEON_TV_DAC_BGSLEEP | + RADEON_TV_DAC_RDACPD | + RADEON_TV_DAC_GDACPD | + RADEON_TV_DAC_BDACPD; + WREG32(RADEON_TV_DAC_CNTL, tmp); + + /* TV PLL */ + WREG32_PLL_P(RADEON_TV_PLL_CNTL1, 0, ~RADEON_TVCLK_SRC_SEL_TVPLL); + WREG32_PLL(RADEON_TV_PLL_CNTL, tv_pll_cntl); + WREG32_PLL_P(RADEON_TV_PLL_CNTL1, RADEON_TVPLL_RESET, ~RADEON_TVPLL_RESET); + + radeon_wait_pll_lock(encoder, 200, 800, 135); + + WREG32_PLL_P(RADEON_TV_PLL_CNTL1, 0, ~RADEON_TVPLL_RESET); + + radeon_wait_pll_lock(encoder, 300, 160, 27); + radeon_wait_pll_lock(encoder, 200, 800, 135); + + WREG32_PLL_P(RADEON_TV_PLL_CNTL1, 0, ~0xf); + WREG32_PLL_P(RADEON_TV_PLL_CNTL1, RADEON_TVCLK_SRC_SEL_TVPLL, ~RADEON_TVCLK_SRC_SEL_TVPLL); + + WREG32_PLL_P(RADEON_TV_PLL_CNTL1, (1 << RADEON_TVPDC_SHIFT), ~RADEON_TVPDC_MASK); + WREG32_PLL_P(RADEON_TV_PLL_CNTL1, 0, ~RADEON_TVPLL_SLEEP); + + /* TV HV */ + WREG32(RADEON_TV_RGB_CNTL, tv_rgb_cntl); + WREG32(RADEON_TV_HTOTAL, const_ptr->hor_total - 1); + WREG32(RADEON_TV_HDISP, const_ptr->hor_resolution - 1); + WREG32(RADEON_TV_HSTART, const_ptr->hor_start); + + WREG32(RADEON_TV_VTOTAL, const_ptr->ver_total - 1); + WREG32(RADEON_TV_VDISP, const_ptr->ver_resolution - 1); + WREG32(RADEON_TV_FTOTAL, tv_ftotal); + WREG32(RADEON_TV_VSCALER_CNTL1, tv_vscaler_cntl1); + WREG32(RADEON_TV_VSCALER_CNTL2, tv_vscaler_cntl2); + + WREG32(RADEON_TV_Y_FALL_CNTL, tv_y_fall_cntl); + WREG32(RADEON_TV_Y_RISE_CNTL, tv_y_rise_cntl); + WREG32(RADEON_TV_Y_SAW_TOOTH_CNTL, tv_y_saw_tooth_cntl); + + WREG32(RADEON_TV_MASTER_CNTL, (tv_master_cntl | RADEON_TV_ASYNC_RST | + RADEON_CRT_ASYNC_RST)); + + /* TV restarts */ + radeon_legacy_write_tv_restarts(radeon_encoder); + + /* tv timings */ + radeon_restore_tv_timing_tables(radeon_encoder); + + WREG32(RADEON_TV_MASTER_CNTL, (tv_master_cntl | RADEON_TV_ASYNC_RST)); + + /* tv std */ + WREG32(RADEON_TV_SYNC_CNTL, (RADEON_SYNC_PUB | RADEON_TV_SYNC_IO_DRIVE)); + WREG32(RADEON_TV_TIMING_CNTL, tv_dac->tv.timing_cntl); + WREG32(RADEON_TV_MODULATOR_CNTL1, tv_modulator_cntl1); + WREG32(RADEON_TV_MODULATOR_CNTL2, tv_modulator_cntl2); + WREG32(RADEON_TV_PRE_DAC_MUX_CNTL, (RADEON_Y_RED_EN | + RADEON_C_GRN_EN | + RADEON_CMP_BLU_EN | + RADEON_DAC_DITHER_EN)); + + WREG32(RADEON_TV_CRC_CNTL, 0); + + WREG32(RADEON_TV_MASTER_CNTL, tv_master_cntl); + + WREG32(RADEON_TV_GAIN_LIMIT_SETTINGS, ((0x17f << RADEON_UV_GAIN_LIMIT_SHIFT) | + (0x5ff << RADEON_Y_GAIN_LIMIT_SHIFT))); + WREG32(RADEON_TV_LINEAR_GAIN_SETTINGS, ((0x100 << RADEON_UV_GAIN_SHIFT) | + (0x100 << RADEON_Y_GAIN_SHIFT))); + + WREG32(RADEON_TV_DAC_CNTL, tv_dac_cntl); + +} + +void radeon_legacy_tv_adjust_crtc_reg(struct drm_encoder *encoder, + uint32_t *h_total_disp, uint32_t *h_sync_strt_wid, + uint32_t *v_total_disp, uint32_t *v_sync_strt_wid) +{ + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + const struct radeon_tv_mode_constants *const_ptr; + uint32_t tmp; + + const_ptr = radeon_legacy_tv_get_std_mode(radeon_encoder, NULL); + if (!const_ptr) + return; + + *h_total_disp = (((const_ptr->hor_resolution / 8) - 1) << RADEON_CRTC_H_DISP_SHIFT) | + (((const_ptr->hor_total / 8) - 1) << RADEON_CRTC_H_TOTAL_SHIFT); + + tmp = *h_sync_strt_wid; + tmp &= ~(RADEON_CRTC_H_SYNC_STRT_PIX | RADEON_CRTC_H_SYNC_STRT_CHAR); + tmp |= (((const_ptr->hor_syncstart / 8) - 1) << RADEON_CRTC_H_SYNC_STRT_CHAR_SHIFT) | + (const_ptr->hor_syncstart & 7); + *h_sync_strt_wid = tmp; + + *v_total_disp = ((const_ptr->ver_resolution - 1) << RADEON_CRTC_V_DISP_SHIFT) | + ((const_ptr->ver_total - 1) << RADEON_CRTC_V_TOTAL_SHIFT); + + tmp = *v_sync_strt_wid; + tmp &= ~RADEON_CRTC_V_SYNC_STRT; + tmp |= ((const_ptr->ver_syncstart - 1) << RADEON_CRTC_V_SYNC_STRT_SHIFT); + *v_sync_strt_wid = tmp; +} + +static inline int get_post_div(int value) +{ + int post_div; + switch (value) { + case 1: post_div = 0; break; + case 2: post_div = 1; break; + case 3: post_div = 4; break; + case 4: post_div = 2; break; + case 6: post_div = 6; break; + case 8: post_div = 3; break; + case 12: post_div = 7; break; + case 16: + default: post_div = 5; break; + } + return post_div; +} + +void radeon_legacy_tv_adjust_pll1(struct drm_encoder *encoder, + uint32_t *htotal_cntl, uint32_t *ppll_ref_div, + uint32_t *ppll_div_3, uint32_t *pixclks_cntl) +{ + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + const struct radeon_tv_mode_constants *const_ptr; + + const_ptr = radeon_legacy_tv_get_std_mode(radeon_encoder, NULL); + if (!const_ptr) + return; + + *htotal_cntl = (const_ptr->hor_total & 0x7) | RADEON_HTOT_CNTL_VGA_EN; + + *ppll_ref_div = const_ptr->crtcPLL_M; + + *ppll_div_3 = (const_ptr->crtcPLL_N & 0x7ff) | (get_post_div(const_ptr->crtcPLL_post_div) << 16); + *pixclks_cntl &= ~(RADEON_PIX2CLK_SRC_SEL_MASK | RADEON_PIXCLK_TV_SRC_SEL); + *pixclks_cntl |= RADEON_PIX2CLK_SRC_SEL_P2PLLCLK; +} + +void radeon_legacy_tv_adjust_pll2(struct drm_encoder *encoder, + uint32_t *htotal2_cntl, uint32_t *p2pll_ref_div, + uint32_t *p2pll_div_0, uint32_t *pixclks_cntl) +{ + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + const struct radeon_tv_mode_constants *const_ptr; + + const_ptr = radeon_legacy_tv_get_std_mode(radeon_encoder, NULL); + if (!const_ptr) + return; + + *htotal2_cntl = (const_ptr->hor_total & 0x7); + + *p2pll_ref_div = const_ptr->crtcPLL_M; + + *p2pll_div_0 = (const_ptr->crtcPLL_N & 0x7ff) | (get_post_div(const_ptr->crtcPLL_post_div) << 16); + *pixclks_cntl &= ~RADEON_PIX2CLK_SRC_SEL_MASK; + *pixclks_cntl |= RADEON_PIX2CLK_SRC_SEL_P2PLLCLK | RADEON_PIXCLK_TV_SRC_SEL; +} + diff --git a/drivers/gpu/drm/radeon/radeon_mem.c b/drivers/gpu/drm/radeon/radeon_mem.c new file mode 100644 index 00000000..ed95155c --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_mem.c @@ -0,0 +1,302 @@ +/* radeon_mem.c -- Simple GART/fb memory manager for radeon -*- linux-c -*- */ +/* + * Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + * + * The Weather Channel (TM) funded Tungsten Graphics to develop the + * initial release of the Radeon 8500 driver under the XFree86 license. + * This notice must be preserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Keith Whitwell + */ + +#include "drmP.h" +#include "drm.h" +#include "radeon_drm.h" +#include "radeon_drv.h" + +/* Very simple allocator for GART memory, working on a static range + * already mapped into each client's address space. + */ + +static struct mem_block *split_block(struct mem_block *p, int start, int size, + struct drm_file *file_priv) +{ + /* Maybe cut off the start of an existing block */ + if (start > p->start) { + struct mem_block *newblock = kmalloc(sizeof(*newblock), + GFP_KERNEL); + if (!newblock) + goto out; + newblock->start = start; + newblock->size = p->size - (start - p->start); + newblock->file_priv = NULL; + newblock->next = p->next; + newblock->prev = p; + p->next->prev = newblock; + p->next = newblock; + p->size -= newblock->size; + p = newblock; + } + + /* Maybe cut off the end of an existing block */ + if (size < p->size) { + struct mem_block *newblock = kmalloc(sizeof(*newblock), + GFP_KERNEL); + if (!newblock) + goto out; + newblock->start = start + size; + newblock->size = p->size - size; + newblock->file_priv = NULL; + newblock->next = p->next; + newblock->prev = p; + p->next->prev = newblock; + p->next = newblock; + p->size = size; + } + + out: + /* Our block is in the middle */ + p->file_priv = file_priv; + return p; +} + +static struct mem_block *alloc_block(struct mem_block *heap, int size, + int align2, struct drm_file *file_priv) +{ + struct mem_block *p; + int mask = (1 << align2) - 1; + + list_for_each(p, heap) { + int start = (p->start + mask) & ~mask; + if (p->file_priv == NULL && start + size <= p->start + p->size) + return split_block(p, start, size, file_priv); + } + + return NULL; +} + +static struct mem_block *find_block(struct mem_block *heap, int start) +{ + struct mem_block *p; + + list_for_each(p, heap) + if (p->start == start) + return p; + + return NULL; +} + +static void free_block(struct mem_block *p) +{ + p->file_priv = NULL; + + /* Assumes a single contiguous range. Needs a special file_priv in + * 'heap' to stop it being subsumed. + */ + if (p->next->file_priv == NULL) { + struct mem_block *q = p->next; + p->size += q->size; + p->next = q->next; + p->next->prev = p; + kfree(q); + } + + if (p->prev->file_priv == NULL) { + struct mem_block *q = p->prev; + q->size += p->size; + q->next = p->next; + q->next->prev = q; + kfree(p); + } +} + +/* Initialize. How to check for an uninitialized heap? + */ +static int init_heap(struct mem_block **heap, int start, int size) +{ + struct mem_block *blocks = kmalloc(sizeof(*blocks), GFP_KERNEL); + + if (!blocks) + return -ENOMEM; + + *heap = kmalloc(sizeof(**heap), GFP_KERNEL); + if (!*heap) { + kfree(blocks); + return -ENOMEM; + } + + blocks->start = start; + blocks->size = size; + blocks->file_priv = NULL; + blocks->next = blocks->prev = *heap; + + memset(*heap, 0, sizeof(**heap)); + (*heap)->file_priv = (struct drm_file *) - 1; + (*heap)->next = (*heap)->prev = blocks; + return 0; +} + +/* Free all blocks associated with the releasing file. + */ +void radeon_mem_release(struct drm_file *file_priv, struct mem_block *heap) +{ + struct mem_block *p; + + if (!heap || !heap->next) + return; + + list_for_each(p, heap) { + if (p->file_priv == file_priv) + p->file_priv = NULL; + } + + /* Assumes a single contiguous range. Needs a special file_priv in + * 'heap' to stop it being subsumed. + */ + list_for_each(p, heap) { + while (p->file_priv == NULL && p->next->file_priv == NULL) { + struct mem_block *q = p->next; + p->size += q->size; + p->next = q->next; + p->next->prev = p; + kfree(q); + } + } +} + +/* Shutdown. + */ +void radeon_mem_takedown(struct mem_block **heap) +{ + struct mem_block *p; + + if (!*heap) + return; + + for (p = (*heap)->next; p != *heap;) { + struct mem_block *q = p; + p = p->next; + kfree(q); + } + + kfree(*heap); + *heap = NULL; +} + +/* IOCTL HANDLERS */ + +static struct mem_block **get_heap(drm_radeon_private_t * dev_priv, int region) +{ + switch (region) { + case RADEON_MEM_REGION_GART: + return &dev_priv->gart_heap; + case RADEON_MEM_REGION_FB: + return &dev_priv->fb_heap; + default: + return NULL; + } +} + +int radeon_mem_alloc(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + drm_radeon_mem_alloc_t *alloc = data; + struct mem_block *block, **heap; + + if (!dev_priv) { + DRM_ERROR("called with no initialization\n"); + return -EINVAL; + } + + heap = get_heap(dev_priv, alloc->region); + if (!heap || !*heap) + return -EFAULT; + + /* Make things easier on ourselves: all allocations at least + * 4k aligned. + */ + if (alloc->alignment < 12) + alloc->alignment = 12; + + block = alloc_block(*heap, alloc->size, alloc->alignment, file_priv); + + if (!block) + return -ENOMEM; + + if (DRM_COPY_TO_USER(alloc->region_offset, &block->start, + sizeof(int))) { + DRM_ERROR("copy_to_user\n"); + return -EFAULT; + } + + return 0; +} + +int radeon_mem_free(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + drm_radeon_mem_free_t *memfree = data; + struct mem_block *block, **heap; + + if (!dev_priv) { + DRM_ERROR("called with no initialization\n"); + return -EINVAL; + } + + heap = get_heap(dev_priv, memfree->region); + if (!heap || !*heap) + return -EFAULT; + + block = find_block(*heap, memfree->region_offset); + if (!block) + return -EFAULT; + + if (block->file_priv != file_priv) + return -EPERM; + + free_block(block); + return 0; +} + +int radeon_mem_init_heap(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + drm_radeon_mem_init_heap_t *initheap = data; + struct mem_block **heap; + + if (!dev_priv) { + DRM_ERROR("called with no initialization\n"); + return -EINVAL; + } + + heap = get_heap(dev_priv, initheap->region); + if (!heap) + return -EFAULT; + + if (*heap) { + DRM_ERROR("heap already initialized?"); + return -EFAULT; + } + + return init_heap(heap, initheap->start, initheap->size); +} diff --git a/drivers/gpu/drm/radeon/radeon_mode.h b/drivers/gpu/drm/radeon/radeon_mode.h new file mode 100644 index 00000000..ed0178f0 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_mode.h @@ -0,0 +1,700 @@ +/* + * Copyright 2000 ATI Technologies Inc., Markham, Ontario, and + * VA Linux Systems Inc., Fremont, California. + * Copyright 2008 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Original Authors: + * Kevin E. Martin, Rickard E. Faith, Alan Hourihane + * + * Kernel port Author: Dave Airlie + */ + +#ifndef RADEON_MODE_H +#define RADEON_MODE_H + +#include +#include +#include +#include +#include +#include +#include +#include + +struct radeon_bo; +struct radeon_device; + +#define to_radeon_crtc(x) container_of(x, struct radeon_crtc, base) +#define to_radeon_connector(x) container_of(x, struct radeon_connector, base) +#define to_radeon_encoder(x) container_of(x, struct radeon_encoder, base) +#define to_radeon_framebuffer(x) container_of(x, struct radeon_framebuffer, base) + +enum radeon_rmx_type { + RMX_OFF, + RMX_FULL, + RMX_CENTER, + RMX_ASPECT +}; + +enum radeon_tv_std { + TV_STD_NTSC, + TV_STD_PAL, + TV_STD_PAL_M, + TV_STD_PAL_60, + TV_STD_NTSC_J, + TV_STD_SCART_PAL, + TV_STD_SECAM, + TV_STD_PAL_CN, + TV_STD_PAL_N, +}; + +enum radeon_underscan_type { + UNDERSCAN_OFF, + UNDERSCAN_ON, + UNDERSCAN_AUTO, +}; + +enum radeon_hpd_id { + RADEON_HPD_1 = 0, + RADEON_HPD_2, + RADEON_HPD_3, + RADEON_HPD_4, + RADEON_HPD_5, + RADEON_HPD_6, + RADEON_HPD_NONE = 0xff, +}; + +#define RADEON_MAX_I2C_BUS 16 + +/* radeon gpio-based i2c + * 1. "mask" reg and bits + * grabs the gpio pins for software use + * 0=not held 1=held + * 2. "a" reg and bits + * output pin value + * 0=low 1=high + * 3. "en" reg and bits + * sets the pin direction + * 0=input 1=output + * 4. "y" reg and bits + * input pin value + * 0=low 1=high + */ +struct radeon_i2c_bus_rec { + bool valid; + /* id used by atom */ + uint8_t i2c_id; + /* id used by atom */ + enum radeon_hpd_id hpd; + /* can be used with hw i2c engine */ + bool hw_capable; + /* uses multi-media i2c engine */ + bool mm_i2c; + /* regs and bits */ + uint32_t mask_clk_reg; + uint32_t mask_data_reg; + uint32_t a_clk_reg; + uint32_t a_data_reg; + uint32_t en_clk_reg; + uint32_t en_data_reg; + uint32_t y_clk_reg; + uint32_t y_data_reg; + uint32_t mask_clk_mask; + uint32_t mask_data_mask; + uint32_t a_clk_mask; + uint32_t a_data_mask; + uint32_t en_clk_mask; + uint32_t en_data_mask; + uint32_t y_clk_mask; + uint32_t y_data_mask; +}; + +struct radeon_tmds_pll { + uint32_t freq; + uint32_t value; +}; + +#define RADEON_MAX_BIOS_CONNECTOR 16 + +/* pll flags */ +#define RADEON_PLL_USE_BIOS_DIVS (1 << 0) +#define RADEON_PLL_NO_ODD_POST_DIV (1 << 1) +#define RADEON_PLL_USE_REF_DIV (1 << 2) +#define RADEON_PLL_LEGACY (1 << 3) +#define RADEON_PLL_PREFER_LOW_REF_DIV (1 << 4) +#define RADEON_PLL_PREFER_HIGH_REF_DIV (1 << 5) +#define RADEON_PLL_PREFER_LOW_FB_DIV (1 << 6) +#define RADEON_PLL_PREFER_HIGH_FB_DIV (1 << 7) +#define RADEON_PLL_PREFER_LOW_POST_DIV (1 << 8) +#define RADEON_PLL_PREFER_HIGH_POST_DIV (1 << 9) +#define RADEON_PLL_USE_FRAC_FB_DIV (1 << 10) +#define RADEON_PLL_PREFER_CLOSEST_LOWER (1 << 11) +#define RADEON_PLL_USE_POST_DIV (1 << 12) +#define RADEON_PLL_IS_LCD (1 << 13) +#define RADEON_PLL_PREFER_MINM_OVER_MAXP (1 << 14) + +struct radeon_pll { + /* reference frequency */ + uint32_t reference_freq; + + /* fixed dividers */ + uint32_t reference_div; + uint32_t post_div; + + /* pll in/out limits */ + uint32_t pll_in_min; + uint32_t pll_in_max; + uint32_t pll_out_min; + uint32_t pll_out_max; + uint32_t lcd_pll_out_min; + uint32_t lcd_pll_out_max; + uint32_t best_vco; + + /* divider limits */ + uint32_t min_ref_div; + uint32_t max_ref_div; + uint32_t min_post_div; + uint32_t max_post_div; + uint32_t min_feedback_div; + uint32_t max_feedback_div; + uint32_t min_frac_feedback_div; + uint32_t max_frac_feedback_div; + + /* flags for the current clock */ + uint32_t flags; + + /* pll id */ + uint32_t id; +}; + +struct radeon_i2c_chan { + struct i2c_adapter adapter; + struct drm_device *dev; + union { + struct i2c_algo_bit_data bit; + struct i2c_algo_dp_aux_data dp; + } algo; + struct radeon_i2c_bus_rec rec; +}; + +/* mostly for macs, but really any system without connector tables */ +enum radeon_connector_table { + CT_NONE = 0, + CT_GENERIC, + CT_IBOOK, + CT_POWERBOOK_EXTERNAL, + CT_POWERBOOK_INTERNAL, + CT_POWERBOOK_VGA, + CT_MINI_EXTERNAL, + CT_MINI_INTERNAL, + CT_IMAC_G5_ISIGHT, + CT_EMAC, + CT_RN50_POWER, + CT_MAC_X800, + CT_MAC_G5_9600, +}; + +enum radeon_dvo_chip { + DVO_SIL164, + DVO_SIL1178, +}; + +struct radeon_fbdev; + +struct radeon_mode_info { + struct atom_context *atom_context; + struct card_info *atom_card_info; + enum radeon_connector_table connector_table; + bool mode_config_initialized; + struct radeon_crtc *crtcs[6]; + /* DVI-I properties */ + struct drm_property *coherent_mode_property; + /* DAC enable load detect */ + struct drm_property *load_detect_property; + /* TV standard */ + struct drm_property *tv_std_property; + /* legacy TMDS PLL detect */ + struct drm_property *tmds_pll_property; + /* underscan */ + struct drm_property *underscan_property; + struct drm_property *underscan_hborder_property; + struct drm_property *underscan_vborder_property; + /* hardcoded DFP edid from BIOS */ + struct edid *bios_hardcoded_edid; + int bios_hardcoded_edid_size; + + /* pointer to fbdev info structure */ + struct radeon_fbdev *rfbdev; +}; + +#define MAX_H_CODE_TIMING_LEN 32 +#define MAX_V_CODE_TIMING_LEN 32 + +/* need to store these as reading + back code tables is excessive */ +struct radeon_tv_regs { + uint32_t tv_uv_adr; + uint32_t timing_cntl; + uint32_t hrestart; + uint32_t vrestart; + uint32_t frestart; + uint16_t h_code_timing[MAX_H_CODE_TIMING_LEN]; + uint16_t v_code_timing[MAX_V_CODE_TIMING_LEN]; +}; + +struct radeon_crtc { + struct drm_crtc base; + int crtc_id; + u16 lut_r[256], lut_g[256], lut_b[256]; + bool enabled; + bool can_tile; + uint32_t crtc_offset; + struct drm_gem_object *cursor_bo; + uint64_t cursor_addr; + int cursor_width; + int cursor_height; + uint32_t legacy_display_base_addr; + uint32_t legacy_cursor_offset; + enum radeon_rmx_type rmx_type; + u8 h_border; + u8 v_border; + fixed20_12 vsc; + fixed20_12 hsc; + struct drm_display_mode native_mode; + int pll_id; + /* page flipping */ + struct radeon_unpin_work *unpin_work; + int deferred_flip_completion; +}; + +struct radeon_encoder_primary_dac { + /* legacy primary dac */ + uint32_t ps2_pdac_adj; +}; + +struct radeon_encoder_lvds { + /* legacy lvds */ + uint16_t panel_vcc_delay; + uint8_t panel_pwr_delay; + uint8_t panel_digon_delay; + uint8_t panel_blon_delay; + uint16_t panel_ref_divider; + uint8_t panel_post_divider; + uint16_t panel_fb_divider; + bool use_bios_dividers; + uint32_t lvds_gen_cntl; + /* panel mode */ + struct drm_display_mode native_mode; + struct backlight_device *bl_dev; + int dpms_mode; + uint8_t backlight_level; +}; + +struct radeon_encoder_tv_dac { + /* legacy tv dac */ + uint32_t ps2_tvdac_adj; + uint32_t ntsc_tvdac_adj; + uint32_t pal_tvdac_adj; + + int h_pos; + int v_pos; + int h_size; + int supported_tv_stds; + bool tv_on; + enum radeon_tv_std tv_std; + struct radeon_tv_regs tv; +}; + +struct radeon_encoder_int_tmds { + /* legacy int tmds */ + struct radeon_tmds_pll tmds_pll[4]; +}; + +struct radeon_encoder_ext_tmds { + /* tmds over dvo */ + struct radeon_i2c_chan *i2c_bus; + uint8_t slave_addr; + enum radeon_dvo_chip dvo_chip; +}; + +/* spread spectrum */ +struct radeon_atom_ss { + uint16_t percentage; + uint8_t type; + uint16_t step; + uint8_t delay; + uint8_t range; + uint8_t refdiv; + /* asic_ss */ + uint16_t rate; + uint16_t amount; +}; + +struct radeon_encoder_atom_dig { + bool linkb; + /* atom dig */ + bool coherent_mode; + int dig_encoder; /* -1 disabled, 0 DIGA, 1 DIGB, etc. */ + /* atom lvds/edp */ + uint32_t lcd_misc; + uint16_t panel_pwr_delay; + uint32_t lcd_ss_id; + /* panel mode */ + struct drm_display_mode native_mode; + struct backlight_device *bl_dev; + int dpms_mode; + uint8_t backlight_level; +}; + +struct radeon_encoder_atom_dac { + enum radeon_tv_std tv_std; +}; + +struct radeon_encoder { + struct drm_encoder base; + uint32_t encoder_enum; + uint32_t encoder_id; + uint32_t devices; + uint32_t active_device; + uint32_t flags; + uint32_t pixel_clock; + enum radeon_rmx_type rmx_type; + enum radeon_underscan_type underscan_type; + uint32_t underscan_hborder; + uint32_t underscan_vborder; + struct drm_display_mode native_mode; + void *enc_priv; + int audio_polling_active; + int hdmi_offset; + int hdmi_config_offset; + int hdmi_audio_workaround; + int hdmi_buffer_status; + bool is_ext_encoder; + u16 caps; +}; + +struct radeon_connector_atom_dig { + uint32_t igp_lane_info; + /* displayport */ + struct radeon_i2c_chan *dp_i2c_bus; + u8 dpcd[8]; + u8 dp_sink_type; + int dp_clock; + int dp_lane_count; + bool edp_on; +}; + +struct radeon_gpio_rec { + bool valid; + u8 id; + u32 reg; + u32 mask; +}; + +struct radeon_hpd { + enum radeon_hpd_id hpd; + u8 plugged_state; + struct radeon_gpio_rec gpio; +}; + +struct radeon_router { + u32 router_id; + struct radeon_i2c_bus_rec i2c_info; + u8 i2c_addr; + /* i2c mux */ + bool ddc_valid; + u8 ddc_mux_type; + u8 ddc_mux_control_pin; + u8 ddc_mux_state; + /* clock/data mux */ + bool cd_valid; + u8 cd_mux_type; + u8 cd_mux_control_pin; + u8 cd_mux_state; +}; + +struct radeon_connector { + struct drm_connector base; + uint32_t connector_id; + uint32_t devices; + struct radeon_i2c_chan *ddc_bus; + /* some systems have an hdmi and vga port with a shared ddc line */ + bool shared_ddc; + /* for some Radeon chip families we apply an additional EDID header + check as part of the DDC probe */ + bool requires_extended_probe; + bool use_digital; + /* we need to mind the EDID between detect + and get modes due to analog/digital/tvencoder */ + struct edid *edid; + void *con_priv; + bool dac_load_detect; + bool detected_by_load; /* if the connection status was determined by load */ + uint16_t connector_object_id; + struct radeon_hpd hpd; + struct radeon_router router; + struct radeon_i2c_chan *router_bus; +}; + +struct radeon_framebuffer { + struct drm_framebuffer base; + struct drm_gem_object *obj; +}; + + +extern enum radeon_tv_std +radeon_combios_get_tv_info(struct radeon_device *rdev); +extern enum radeon_tv_std +radeon_atombios_get_tv_info(struct radeon_device *rdev); + +extern struct drm_connector * +radeon_get_connector_for_encoder(struct drm_encoder *encoder); + +extern bool radeon_encoder_is_dp_bridge(struct drm_encoder *encoder); +extern bool radeon_connector_encoder_is_dp_bridge(struct drm_connector *connector); +extern bool radeon_connector_encoder_is_hbr2(struct drm_connector *connector); +extern bool radeon_connector_is_dp12_capable(struct drm_connector *connector); + +extern void radeon_connector_hotplug(struct drm_connector *connector); +extern int radeon_dp_mode_valid_helper(struct drm_connector *connector, + struct drm_display_mode *mode); +extern void radeon_dp_set_link_config(struct drm_connector *connector, + struct drm_display_mode *mode); +extern void radeon_dp_link_train(struct drm_encoder *encoder, + struct drm_connector *connector); +extern bool radeon_dp_needs_link_train(struct radeon_connector *radeon_connector); +extern u8 radeon_dp_getsinktype(struct radeon_connector *radeon_connector); +extern bool radeon_dp_getdpcd(struct radeon_connector *radeon_connector); +extern void atombios_dig_encoder_setup(struct drm_encoder *encoder, int action, int panel_mode); +extern void radeon_atom_encoder_init(struct radeon_device *rdev); +extern void atombios_dig_transmitter_setup(struct drm_encoder *encoder, + int action, uint8_t lane_num, + uint8_t lane_set); +extern void radeon_atom_ext_encoder_setup_ddc(struct drm_encoder *encoder); +extern struct drm_encoder *radeon_atom_get_external_encoder(struct drm_encoder *encoder); +extern int radeon_dp_i2c_aux_ch(struct i2c_adapter *adapter, int mode, + u8 write_byte, u8 *read_byte); + +extern void radeon_i2c_init(struct radeon_device *rdev); +extern void radeon_i2c_fini(struct radeon_device *rdev); +extern void radeon_combios_i2c_init(struct radeon_device *rdev); +extern void radeon_atombios_i2c_init(struct radeon_device *rdev); +extern void radeon_i2c_add(struct radeon_device *rdev, + struct radeon_i2c_bus_rec *rec, + const char *name); +extern struct radeon_i2c_chan *radeon_i2c_lookup(struct radeon_device *rdev, + struct radeon_i2c_bus_rec *i2c_bus); +extern struct radeon_i2c_chan *radeon_i2c_create_dp(struct drm_device *dev, + struct radeon_i2c_bus_rec *rec, + const char *name); +extern struct radeon_i2c_chan *radeon_i2c_create(struct drm_device *dev, + struct radeon_i2c_bus_rec *rec, + const char *name); +extern void radeon_i2c_destroy(struct radeon_i2c_chan *i2c); +extern void radeon_i2c_get_byte(struct radeon_i2c_chan *i2c_bus, + u8 slave_addr, + u8 addr, + u8 *val); +extern void radeon_i2c_put_byte(struct radeon_i2c_chan *i2c, + u8 slave_addr, + u8 addr, + u8 val); +extern void radeon_router_select_ddc_port(struct radeon_connector *radeon_connector); +extern void radeon_router_select_cd_port(struct radeon_connector *radeon_connector); +extern bool radeon_ddc_probe(struct radeon_connector *radeon_connector, + bool requires_extended_probe); +extern int radeon_ddc_get_modes(struct radeon_connector *radeon_connector); + +extern struct drm_encoder *radeon_best_encoder(struct drm_connector *connector); + +extern bool radeon_atombios_get_ppll_ss_info(struct radeon_device *rdev, + struct radeon_atom_ss *ss, + int id); +extern bool radeon_atombios_get_asic_ss_info(struct radeon_device *rdev, + struct radeon_atom_ss *ss, + int id, u32 clock); + +extern void radeon_compute_pll_legacy(struct radeon_pll *pll, + uint64_t freq, + uint32_t *dot_clock_p, + uint32_t *fb_div_p, + uint32_t *frac_fb_div_p, + uint32_t *ref_div_p, + uint32_t *post_div_p); + +extern void radeon_compute_pll_avivo(struct radeon_pll *pll, + u32 freq, + u32 *dot_clock_p, + u32 *fb_div_p, + u32 *frac_fb_div_p, + u32 *ref_div_p, + u32 *post_div_p); + +extern void radeon_setup_encoder_clones(struct drm_device *dev); + +struct drm_encoder *radeon_encoder_legacy_lvds_add(struct drm_device *dev, int bios_index); +struct drm_encoder *radeon_encoder_legacy_primary_dac_add(struct drm_device *dev, int bios_index, int with_tv); +struct drm_encoder *radeon_encoder_legacy_tv_dac_add(struct drm_device *dev, int bios_index, int with_tv); +struct drm_encoder *radeon_encoder_legacy_tmds_int_add(struct drm_device *dev, int bios_index); +struct drm_encoder *radeon_encoder_legacy_tmds_ext_add(struct drm_device *dev, int bios_index); +extern void atombios_dvo_setup(struct drm_encoder *encoder, int action); +extern void atombios_digital_setup(struct drm_encoder *encoder, int action); +extern int atombios_get_encoder_mode(struct drm_encoder *encoder); +extern bool atombios_set_edp_panel_power(struct drm_connector *connector, int action); +extern void radeon_encoder_set_active_device(struct drm_encoder *encoder); + +extern void radeon_crtc_load_lut(struct drm_crtc *crtc); +extern int atombios_crtc_set_base(struct drm_crtc *crtc, int x, int y, + struct drm_framebuffer *old_fb); +extern int atombios_crtc_set_base_atomic(struct drm_crtc *crtc, + struct drm_framebuffer *fb, + int x, int y, + enum mode_set_atomic state); +extern int atombios_crtc_mode_set(struct drm_crtc *crtc, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode, + int x, int y, + struct drm_framebuffer *old_fb); +extern void atombios_crtc_dpms(struct drm_crtc *crtc, int mode); + +extern int radeon_crtc_set_base(struct drm_crtc *crtc, int x, int y, + struct drm_framebuffer *old_fb); +extern int radeon_crtc_set_base_atomic(struct drm_crtc *crtc, + struct drm_framebuffer *fb, + int x, int y, + enum mode_set_atomic state); +extern int radeon_crtc_do_set_base(struct drm_crtc *crtc, + struct drm_framebuffer *fb, + int x, int y, int atomic); +extern int radeon_crtc_cursor_set(struct drm_crtc *crtc, + struct drm_file *file_priv, + uint32_t handle, + uint32_t width, + uint32_t height); +extern int radeon_crtc_cursor_move(struct drm_crtc *crtc, + int x, int y); + +extern int radeon_get_crtc_scanoutpos(struct drm_device *dev, int crtc, + int *vpos, int *hpos); + +extern bool radeon_combios_check_hardcoded_edid(struct radeon_device *rdev); +extern struct edid * +radeon_bios_get_hardcoded_edid(struct radeon_device *rdev); +extern bool radeon_atom_get_clock_info(struct drm_device *dev); +extern bool radeon_combios_get_clock_info(struct drm_device *dev); +extern struct radeon_encoder_atom_dig * +radeon_atombios_get_lvds_info(struct radeon_encoder *encoder); +extern bool radeon_atombios_get_tmds_info(struct radeon_encoder *encoder, + struct radeon_encoder_int_tmds *tmds); +extern bool radeon_legacy_get_tmds_info_from_combios(struct radeon_encoder *encoder, + struct radeon_encoder_int_tmds *tmds); +extern bool radeon_legacy_get_tmds_info_from_table(struct radeon_encoder *encoder, + struct radeon_encoder_int_tmds *tmds); +extern bool radeon_legacy_get_ext_tmds_info_from_combios(struct radeon_encoder *encoder, + struct radeon_encoder_ext_tmds *tmds); +extern bool radeon_legacy_get_ext_tmds_info_from_table(struct radeon_encoder *encoder, + struct radeon_encoder_ext_tmds *tmds); +extern struct radeon_encoder_primary_dac * +radeon_atombios_get_primary_dac_info(struct radeon_encoder *encoder); +extern struct radeon_encoder_tv_dac * +radeon_atombios_get_tv_dac_info(struct radeon_encoder *encoder); +extern struct radeon_encoder_lvds * +radeon_combios_get_lvds_info(struct radeon_encoder *encoder); +extern void radeon_combios_get_ext_tmds_info(struct radeon_encoder *encoder); +extern struct radeon_encoder_tv_dac * +radeon_combios_get_tv_dac_info(struct radeon_encoder *encoder); +extern struct radeon_encoder_primary_dac * +radeon_combios_get_primary_dac_info(struct radeon_encoder *encoder); +extern bool radeon_combios_external_tmds_setup(struct drm_encoder *encoder); +extern void radeon_external_tmds_setup(struct drm_encoder *encoder); +extern void radeon_combios_output_lock(struct drm_encoder *encoder, bool lock); +extern void radeon_combios_initialize_bios_scratch_regs(struct drm_device *dev); +extern void radeon_atom_output_lock(struct drm_encoder *encoder, bool lock); +extern void radeon_atom_initialize_bios_scratch_regs(struct drm_device *dev); +extern void radeon_save_bios_scratch_regs(struct radeon_device *rdev); +extern void radeon_restore_bios_scratch_regs(struct radeon_device *rdev); +extern void +radeon_atombios_encoder_crtc_scratch_regs(struct drm_encoder *encoder, int crtc); +extern void +radeon_atombios_encoder_dpms_scratch_regs(struct drm_encoder *encoder, bool on); +extern void +radeon_combios_encoder_crtc_scratch_regs(struct drm_encoder *encoder, int crtc); +extern void +radeon_combios_encoder_dpms_scratch_regs(struct drm_encoder *encoder, bool on); +extern void radeon_crtc_fb_gamma_set(struct drm_crtc *crtc, u16 red, u16 green, + u16 blue, int regno); +extern void radeon_crtc_fb_gamma_get(struct drm_crtc *crtc, u16 *red, u16 *green, + u16 *blue, int regno); +void radeon_framebuffer_init(struct drm_device *dev, + struct radeon_framebuffer *rfb, + struct drm_mode_fb_cmd *mode_cmd, + struct drm_gem_object *obj); + +int radeonfb_remove(struct drm_device *dev, struct drm_framebuffer *fb); +bool radeon_get_legacy_connector_info_from_bios(struct drm_device *dev); +bool radeon_get_legacy_connector_info_from_table(struct drm_device *dev); +void radeon_atombios_init_crtc(struct drm_device *dev, + struct radeon_crtc *radeon_crtc); +void radeon_legacy_init_crtc(struct drm_device *dev, + struct radeon_crtc *radeon_crtc); + +void radeon_get_clock_info(struct drm_device *dev); + +extern bool radeon_get_atom_connector_info_from_object_table(struct drm_device *dev); +extern bool radeon_get_atom_connector_info_from_supported_devices_table(struct drm_device *dev); + +void radeon_enc_destroy(struct drm_encoder *encoder); +void radeon_copy_fb(struct drm_device *dev, struct drm_gem_object *dst_obj); +void radeon_combios_asic_init(struct drm_device *dev); +bool radeon_crtc_scaling_mode_fixup(struct drm_crtc *crtc, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode); +void radeon_panel_mode_fixup(struct drm_encoder *encoder, + struct drm_display_mode *adjusted_mode); +void atom_rv515_force_tv_scaler(struct radeon_device *rdev, struct radeon_crtc *radeon_crtc); + +/* legacy tv */ +void radeon_legacy_tv_adjust_crtc_reg(struct drm_encoder *encoder, + uint32_t *h_total_disp, uint32_t *h_sync_strt_wid, + uint32_t *v_total_disp, uint32_t *v_sync_strt_wid); +void radeon_legacy_tv_adjust_pll1(struct drm_encoder *encoder, + uint32_t *htotal_cntl, uint32_t *ppll_ref_div, + uint32_t *ppll_div_3, uint32_t *pixclks_cntl); +void radeon_legacy_tv_adjust_pll2(struct drm_encoder *encoder, + uint32_t *htotal2_cntl, uint32_t *p2pll_ref_div, + uint32_t *p2pll_div_0, uint32_t *pixclks_cntl); +void radeon_legacy_tv_mode_set(struct drm_encoder *encoder, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode); + +/* fbdev layer */ +int radeon_fbdev_init(struct radeon_device *rdev); +void radeon_fbdev_fini(struct radeon_device *rdev); +void radeon_fbdev_set_suspend(struct radeon_device *rdev, int state); +int radeon_fbdev_total_size(struct radeon_device *rdev); +bool radeon_fbdev_robj_is_fb(struct radeon_device *rdev, struct radeon_bo *robj); + +void radeon_fb_output_poll_changed(struct radeon_device *rdev); + +void radeon_crtc_handle_flip(struct radeon_device *rdev, int crtc_id); + +int radeon_align_pitch(struct radeon_device *rdev, int width, int bpp, bool tiled); +#endif diff --git a/drivers/gpu/drm/radeon/radeon_object.c b/drivers/gpu/drm/radeon/radeon_object.c new file mode 100644 index 00000000..35da1b47 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_object.c @@ -0,0 +1,533 @@ +/* + * Copyright 2009 Jerome Glisse. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + */ +/* + * Authors: + * Jerome Glisse + * Thomas Hellstrom + * Dave Airlie + */ +#include +#include +#include +#include "radeon_drm.h" +#include "radeon.h" +#include "radeon_trace.h" + + +int radeon_ttm_init(struct radeon_device *rdev); +void radeon_ttm_fini(struct radeon_device *rdev); +static void radeon_bo_clear_surface_reg(struct radeon_bo *bo); + +/* + * To exclude mutual BO access we rely on bo_reserve exclusion, as all + * function are calling it. + */ + +static void radeon_ttm_bo_destroy(struct ttm_buffer_object *tbo) +{ + struct radeon_bo *bo; + + bo = container_of(tbo, struct radeon_bo, tbo); + mutex_lock(&bo->rdev->gem.mutex); + list_del_init(&bo->list); + mutex_unlock(&bo->rdev->gem.mutex); + radeon_bo_clear_surface_reg(bo); + drm_gem_object_release(&bo->gem_base); + kfree(bo); +} + +bool radeon_ttm_bo_is_radeon_bo(struct ttm_buffer_object *bo) +{ + if (bo->destroy == &radeon_ttm_bo_destroy) + return true; + return false; +} + +void radeon_ttm_placement_from_domain(struct radeon_bo *rbo, u32 domain) +{ + u32 c = 0; + + rbo->placement.fpfn = 0; + rbo->placement.lpfn = 0; + rbo->placement.placement = rbo->placements; + rbo->placement.busy_placement = rbo->placements; + if (domain & RADEON_GEM_DOMAIN_VRAM) + rbo->placements[c++] = TTM_PL_FLAG_WC | TTM_PL_FLAG_UNCACHED | + TTM_PL_FLAG_VRAM; + if (domain & RADEON_GEM_DOMAIN_GTT) + rbo->placements[c++] = TTM_PL_MASK_CACHING | TTM_PL_FLAG_TT; + if (domain & RADEON_GEM_DOMAIN_CPU) + rbo->placements[c++] = TTM_PL_MASK_CACHING | TTM_PL_FLAG_SYSTEM; + if (!c) + rbo->placements[c++] = TTM_PL_MASK_CACHING | TTM_PL_FLAG_SYSTEM; + rbo->placement.num_placement = c; + rbo->placement.num_busy_placement = c; +} + +int radeon_bo_create(struct radeon_device *rdev, + unsigned long size, int byte_align, bool kernel, u32 domain, + struct radeon_bo **bo_ptr) +{ + struct radeon_bo *bo; + enum ttm_bo_type type; + unsigned long page_align = roundup(byte_align, PAGE_SIZE) >> PAGE_SHIFT; + unsigned long max_size = 0; + int r; + + size = ALIGN(size, PAGE_SIZE); + + if (unlikely(rdev->mman.bdev.dev_mapping == NULL)) { + rdev->mman.bdev.dev_mapping = rdev->ddev->dev_mapping; + } + if (kernel) { + type = ttm_bo_type_kernel; + } else { + type = ttm_bo_type_device; + } + *bo_ptr = NULL; + + /* maximun bo size is the minimun btw visible vram and gtt size */ + max_size = min(rdev->mc.visible_vram_size, rdev->mc.gtt_size); + if ((page_align << PAGE_SHIFT) >= max_size) { + printk(KERN_WARNING "%s:%d alloc size %ldM bigger than %ldMb limit\n", + __func__, __LINE__, page_align >> (20 - PAGE_SHIFT), max_size >> 20); + return -ENOMEM; + } + +retry: + bo = kzalloc(sizeof(struct radeon_bo), GFP_KERNEL); + if (bo == NULL) + return -ENOMEM; + r = drm_gem_object_init(rdev->ddev, &bo->gem_base, size); + if (unlikely(r)) { + kfree(bo); + return r; + } + bo->rdev = rdev; + bo->gem_base.driver_private = NULL; + bo->surface_reg = -1; + INIT_LIST_HEAD(&bo->list); + radeon_ttm_placement_from_domain(bo, domain); + /* Kernel allocation are uninterruptible */ + mutex_lock(&rdev->vram_mutex); + r = ttm_bo_init(&rdev->mman.bdev, &bo->tbo, size, type, + &bo->placement, page_align, 0, !kernel, NULL, size, + &radeon_ttm_bo_destroy); + mutex_unlock(&rdev->vram_mutex); + if (unlikely(r != 0)) { + if (r != -ERESTARTSYS) { + if (domain == RADEON_GEM_DOMAIN_VRAM) { + domain |= RADEON_GEM_DOMAIN_GTT; + goto retry; + } + dev_err(rdev->dev, + "object_init failed for (%lu, 0x%08X)\n", + size, domain); + } + return r; + } + *bo_ptr = bo; + + trace_radeon_bo_create(bo); + + return 0; +} + +int radeon_bo_kmap(struct radeon_bo *bo, void **ptr) +{ + bool is_iomem; + int r; + + if (bo->kptr) { + if (ptr) { + *ptr = bo->kptr; + } + return 0; + } + r = ttm_bo_kmap(&bo->tbo, 0, bo->tbo.num_pages, &bo->kmap); + if (r) { + return r; + } + bo->kptr = ttm_kmap_obj_virtual(&bo->kmap, &is_iomem); + if (ptr) { + *ptr = bo->kptr; + } + radeon_bo_check_tiling(bo, 0, 0); + return 0; +} + +void radeon_bo_kunmap(struct radeon_bo *bo) +{ + if (bo->kptr == NULL) + return; + bo->kptr = NULL; + radeon_bo_check_tiling(bo, 0, 0); + ttm_bo_kunmap(&bo->kmap); +} + +void radeon_bo_unref(struct radeon_bo **bo) +{ + struct ttm_buffer_object *tbo; + struct radeon_device *rdev; + + if ((*bo) == NULL) + return; + rdev = (*bo)->rdev; + tbo = &((*bo)->tbo); + mutex_lock(&rdev->vram_mutex); + ttm_bo_unref(&tbo); + mutex_unlock(&rdev->vram_mutex); + if (tbo == NULL) + *bo = NULL; +} + +int radeon_bo_pin_restricted(struct radeon_bo *bo, u32 domain, u64 max_offset, + u64 *gpu_addr) +{ + int r, i; + + if (bo->pin_count) { + bo->pin_count++; + if (gpu_addr) + *gpu_addr = radeon_bo_gpu_offset(bo); + WARN_ON_ONCE(max_offset != 0); + return 0; + } + radeon_ttm_placement_from_domain(bo, domain); + if (domain == RADEON_GEM_DOMAIN_VRAM) { + /* force to pin into visible video ram */ + bo->placement.lpfn = bo->rdev->mc.visible_vram_size >> PAGE_SHIFT; + } + if (max_offset) { + u64 lpfn = max_offset >> PAGE_SHIFT; + + if (!bo->placement.lpfn) + bo->placement.lpfn = bo->rdev->mc.gtt_size >> PAGE_SHIFT; + + if (lpfn < bo->placement.lpfn) + bo->placement.lpfn = lpfn; + } + for (i = 0; i < bo->placement.num_placement; i++) + bo->placements[i] |= TTM_PL_FLAG_NO_EVICT; + r = ttm_bo_validate(&bo->tbo, &bo->placement, false, false, false); + if (likely(r == 0)) { + bo->pin_count = 1; + if (gpu_addr != NULL) + *gpu_addr = radeon_bo_gpu_offset(bo); + } + if (unlikely(r != 0)) + dev_err(bo->rdev->dev, "%p pin failed\n", bo); + return r; +} + +int radeon_bo_pin(struct radeon_bo *bo, u32 domain, u64 *gpu_addr) +{ + return radeon_bo_pin_restricted(bo, domain, 0, gpu_addr); +} + +int radeon_bo_unpin(struct radeon_bo *bo) +{ + int r, i; + + if (!bo->pin_count) { + dev_warn(bo->rdev->dev, "%p unpin not necessary\n", bo); + return 0; + } + bo->pin_count--; + if (bo->pin_count) + return 0; + for (i = 0; i < bo->placement.num_placement; i++) + bo->placements[i] &= ~TTM_PL_FLAG_NO_EVICT; + r = ttm_bo_validate(&bo->tbo, &bo->placement, false, false, false); + if (unlikely(r != 0)) + dev_err(bo->rdev->dev, "%p validate failed for unpin\n", bo); + return r; +} + +int radeon_bo_evict_vram(struct radeon_device *rdev) +{ + /* late 2.6.33 fix IGP hibernate - we need pm ops to do this correct */ + if (0 && (rdev->flags & RADEON_IS_IGP)) { + if (rdev->mc.igp_sideport_enabled == false) + /* Useless to evict on IGP chips */ + return 0; + } + return ttm_bo_evict_mm(&rdev->mman.bdev, TTM_PL_VRAM); +} + +void radeon_bo_force_delete(struct radeon_device *rdev) +{ + struct radeon_bo *bo, *n; + + if (list_empty(&rdev->gem.objects)) { + return; + } + dev_err(rdev->dev, "Userspace still has active objects !\n"); + list_for_each_entry_safe(bo, n, &rdev->gem.objects, list) { + mutex_lock(&rdev->ddev->struct_mutex); + dev_err(rdev->dev, "%p %p %lu %lu force free\n", + &bo->gem_base, bo, (unsigned long)bo->gem_base.size, + *((unsigned long *)&bo->gem_base.refcount)); + mutex_lock(&bo->rdev->gem.mutex); + list_del_init(&bo->list); + mutex_unlock(&bo->rdev->gem.mutex); + /* this should unref the ttm bo */ + drm_gem_object_unreference(&bo->gem_base); + mutex_unlock(&rdev->ddev->struct_mutex); + } +} + +int radeon_bo_init(struct radeon_device *rdev) +{ + /* Add an MTRR for the VRAM */ + rdev->mc.vram_mtrr = mtrr_add(rdev->mc.aper_base, rdev->mc.aper_size, + MTRR_TYPE_WRCOMB, 1); + DRM_INFO("Detected VRAM RAM=%lluM, BAR=%lluM\n", + rdev->mc.mc_vram_size >> 20, + (unsigned long long)rdev->mc.aper_size >> 20); + DRM_INFO("RAM width %dbits %cDR\n", + rdev->mc.vram_width, rdev->mc.vram_is_ddr ? 'D' : 'S'); + return radeon_ttm_init(rdev); +} + +void radeon_bo_fini(struct radeon_device *rdev) +{ + radeon_ttm_fini(rdev); +} + +void radeon_bo_list_add_object(struct radeon_bo_list *lobj, + struct list_head *head) +{ + if (lobj->wdomain) { + list_add(&lobj->tv.head, head); + } else { + list_add_tail(&lobj->tv.head, head); + } +} + +int radeon_bo_list_validate(struct list_head *head) +{ + struct radeon_bo_list *lobj; + struct radeon_bo *bo; + u32 domain; + int r; + + r = ttm_eu_reserve_buffers(head); + if (unlikely(r != 0)) { + return r; + } + list_for_each_entry(lobj, head, tv.head) { + bo = lobj->bo; + if (!bo->pin_count) { + domain = lobj->wdomain ? lobj->wdomain : lobj->rdomain; + + retry: + radeon_ttm_placement_from_domain(bo, domain); + r = ttm_bo_validate(&bo->tbo, &bo->placement, + true, false, false); + if (unlikely(r)) { + if (r != -ERESTARTSYS && domain == RADEON_GEM_DOMAIN_VRAM) { + domain |= RADEON_GEM_DOMAIN_GTT; + goto retry; + } + return r; + } + } + lobj->gpu_offset = radeon_bo_gpu_offset(bo); + lobj->tiling_flags = bo->tiling_flags; + } + return 0; +} + +int radeon_bo_fbdev_mmap(struct radeon_bo *bo, + struct vm_area_struct *vma) +{ + return ttm_fbdev_mmap(vma, &bo->tbo); +} + +int radeon_bo_get_surface_reg(struct radeon_bo *bo) +{ + struct radeon_device *rdev = bo->rdev; + struct radeon_surface_reg *reg; + struct radeon_bo *old_object; + int steal; + int i; + + BUG_ON(!atomic_read(&bo->tbo.reserved)); + + if (!bo->tiling_flags) + return 0; + + if (bo->surface_reg >= 0) { + reg = &rdev->surface_regs[bo->surface_reg]; + i = bo->surface_reg; + goto out; + } + + steal = -1; + for (i = 0; i < RADEON_GEM_MAX_SURFACES; i++) { + + reg = &rdev->surface_regs[i]; + if (!reg->bo) + break; + + old_object = reg->bo; + if (old_object->pin_count == 0) + steal = i; + } + + /* if we are all out */ + if (i == RADEON_GEM_MAX_SURFACES) { + if (steal == -1) + return -ENOMEM; + /* find someone with a surface reg and nuke their BO */ + reg = &rdev->surface_regs[steal]; + old_object = reg->bo; + /* blow away the mapping */ + DRM_DEBUG("stealing surface reg %d from %p\n", steal, old_object); + ttm_bo_unmap_virtual(&old_object->tbo); + old_object->surface_reg = -1; + i = steal; + } + + bo->surface_reg = i; + reg->bo = bo; + +out: + radeon_set_surface_reg(rdev, i, bo->tiling_flags, bo->pitch, + bo->tbo.mem.start << PAGE_SHIFT, + bo->tbo.num_pages << PAGE_SHIFT); + return 0; +} + +static void radeon_bo_clear_surface_reg(struct radeon_bo *bo) +{ + struct radeon_device *rdev = bo->rdev; + struct radeon_surface_reg *reg; + + if (bo->surface_reg == -1) + return; + + reg = &rdev->surface_regs[bo->surface_reg]; + radeon_clear_surface_reg(rdev, bo->surface_reg); + + reg->bo = NULL; + bo->surface_reg = -1; +} + +int radeon_bo_set_tiling_flags(struct radeon_bo *bo, + uint32_t tiling_flags, uint32_t pitch) +{ + int r; + + r = radeon_bo_reserve(bo, false); + if (unlikely(r != 0)) + return r; + bo->tiling_flags = tiling_flags; + bo->pitch = pitch; + radeon_bo_unreserve(bo); + return 0; +} + +void radeon_bo_get_tiling_flags(struct radeon_bo *bo, + uint32_t *tiling_flags, + uint32_t *pitch) +{ + BUG_ON(!atomic_read(&bo->tbo.reserved)); + if (tiling_flags) + *tiling_flags = bo->tiling_flags; + if (pitch) + *pitch = bo->pitch; +} + +int radeon_bo_check_tiling(struct radeon_bo *bo, bool has_moved, + bool force_drop) +{ + BUG_ON(!atomic_read(&bo->tbo.reserved)); + + if (!(bo->tiling_flags & RADEON_TILING_SURFACE)) + return 0; + + if (force_drop) { + radeon_bo_clear_surface_reg(bo); + return 0; + } + + if (bo->tbo.mem.mem_type != TTM_PL_VRAM) { + if (!has_moved) + return 0; + + if (bo->surface_reg >= 0) + radeon_bo_clear_surface_reg(bo); + return 0; + } + + if ((bo->surface_reg >= 0) && !has_moved) + return 0; + + return radeon_bo_get_surface_reg(bo); +} + +void radeon_bo_move_notify(struct ttm_buffer_object *bo, + struct ttm_mem_reg *mem) +{ + struct radeon_bo *rbo; + if (!radeon_ttm_bo_is_radeon_bo(bo)) + return; + rbo = container_of(bo, struct radeon_bo, tbo); + radeon_bo_check_tiling(rbo, 0, 1); +} + +int radeon_bo_fault_reserve_notify(struct ttm_buffer_object *bo) +{ + struct radeon_device *rdev; + struct radeon_bo *rbo; + unsigned long offset, size; + int r; + + if (!radeon_ttm_bo_is_radeon_bo(bo)) + return 0; + rbo = container_of(bo, struct radeon_bo, tbo); + radeon_bo_check_tiling(rbo, 0, 0); + rdev = rbo->rdev; + if (bo->mem.mem_type == TTM_PL_VRAM) { + size = bo->mem.num_pages << PAGE_SHIFT; + offset = bo->mem.start << PAGE_SHIFT; + if ((offset + size) > rdev->mc.visible_vram_size) { + /* hurrah the memory is not visible ! */ + radeon_ttm_placement_from_domain(rbo, RADEON_GEM_DOMAIN_VRAM); + rbo->placement.lpfn = rdev->mc.visible_vram_size >> PAGE_SHIFT; + r = ttm_bo_validate(bo, &rbo->placement, false, true, false); + if (unlikely(r != 0)) + return r; + offset = bo->mem.start << PAGE_SHIFT; + /* this should not happen */ + if ((offset + size) > rdev->mc.visible_vram_size) + return -EINVAL; + } + } + return 0; +} diff --git a/drivers/gpu/drm/radeon/radeon_object.h b/drivers/gpu/drm/radeon/radeon_object.h new file mode 100644 index 00000000..7199c6ab --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_object.h @@ -0,0 +1,169 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#ifndef __RADEON_OBJECT_H__ +#define __RADEON_OBJECT_H__ + +#include +#include "radeon.h" + +/** + * radeon_mem_type_to_domain - return domain corresponding to mem_type + * @mem_type: ttm memory type + * + * Returns corresponding domain of the ttm mem_type + */ +static inline unsigned radeon_mem_type_to_domain(u32 mem_type) +{ + switch (mem_type) { + case TTM_PL_VRAM: + return RADEON_GEM_DOMAIN_VRAM; + case TTM_PL_TT: + return RADEON_GEM_DOMAIN_GTT; + case TTM_PL_SYSTEM: + return RADEON_GEM_DOMAIN_CPU; + default: + break; + } + return 0; +} + +/** + * radeon_bo_reserve - reserve bo + * @bo: bo structure + * @no_wait: don't sleep while trying to reserve (return -EBUSY) + * + * Returns: + * -EBUSY: buffer is busy and @no_wait is true + * -ERESTARTSYS: A wait for the buffer to become unreserved was interrupted by + * a signal. Release all buffer reservations and return to user-space. + */ +static inline int radeon_bo_reserve(struct radeon_bo *bo, bool no_wait) +{ + int r; + + r = ttm_bo_reserve(&bo->tbo, true, no_wait, false, 0); + if (unlikely(r != 0)) { + if (r != -ERESTARTSYS) + dev_err(bo->rdev->dev, "%p reserve failed\n", bo); + return r; + } + return 0; +} + +static inline void radeon_bo_unreserve(struct radeon_bo *bo) +{ + ttm_bo_unreserve(&bo->tbo); +} + +/** + * radeon_bo_gpu_offset - return GPU offset of bo + * @bo: radeon object for which we query the offset + * + * Returns current GPU offset of the object. + * + * Note: object should either be pinned or reserved when calling this + * function, it might be useful to add check for this for debugging. + */ +static inline u64 radeon_bo_gpu_offset(struct radeon_bo *bo) +{ + return bo->tbo.offset; +} + +static inline unsigned long radeon_bo_size(struct radeon_bo *bo) +{ + return bo->tbo.num_pages << PAGE_SHIFT; +} + +static inline bool radeon_bo_is_reserved(struct radeon_bo *bo) +{ + return !!atomic_read(&bo->tbo.reserved); +} + +/** + * radeon_bo_mmap_offset - return mmap offset of bo + * @bo: radeon object for which we query the offset + * + * Returns mmap offset of the object. + * + * Note: addr_space_offset is constant after ttm bo init thus isn't protected + * by any lock. + */ +static inline u64 radeon_bo_mmap_offset(struct radeon_bo *bo) +{ + return bo->tbo.addr_space_offset; +} + +static inline int radeon_bo_wait(struct radeon_bo *bo, u32 *mem_type, + bool no_wait) +{ + int r; + + r = ttm_bo_reserve(&bo->tbo, true, no_wait, false, 0); + if (unlikely(r != 0)) + return r; + spin_lock(&bo->tbo.bdev->fence_lock); + if (mem_type) + *mem_type = bo->tbo.mem.mem_type; + if (bo->tbo.sync_obj) + r = ttm_bo_wait(&bo->tbo, true, true, no_wait); + spin_unlock(&bo->tbo.bdev->fence_lock); + ttm_bo_unreserve(&bo->tbo); + return r; +} + +extern int radeon_bo_create(struct radeon_device *rdev, + unsigned long size, int byte_align, + bool kernel, u32 domain, + struct radeon_bo **bo_ptr); +extern int radeon_bo_kmap(struct radeon_bo *bo, void **ptr); +extern void radeon_bo_kunmap(struct radeon_bo *bo); +extern void radeon_bo_unref(struct radeon_bo **bo); +extern int radeon_bo_pin(struct radeon_bo *bo, u32 domain, u64 *gpu_addr); +extern int radeon_bo_pin_restricted(struct radeon_bo *bo, u32 domain, + u64 max_offset, u64 *gpu_addr); +extern int radeon_bo_unpin(struct radeon_bo *bo); +extern int radeon_bo_evict_vram(struct radeon_device *rdev); +extern void radeon_bo_force_delete(struct radeon_device *rdev); +extern int radeon_bo_init(struct radeon_device *rdev); +extern void radeon_bo_fini(struct radeon_device *rdev); +extern void radeon_bo_list_add_object(struct radeon_bo_list *lobj, + struct list_head *head); +extern int radeon_bo_list_validate(struct list_head *head); +extern int radeon_bo_fbdev_mmap(struct radeon_bo *bo, + struct vm_area_struct *vma); +extern int radeon_bo_set_tiling_flags(struct radeon_bo *bo, + u32 tiling_flags, u32 pitch); +extern void radeon_bo_get_tiling_flags(struct radeon_bo *bo, + u32 *tiling_flags, u32 *pitch); +extern int radeon_bo_check_tiling(struct radeon_bo *bo, bool has_moved, + bool force_drop); +extern void radeon_bo_move_notify(struct ttm_buffer_object *bo, + struct ttm_mem_reg *mem); +extern int radeon_bo_fault_reserve_notify(struct ttm_buffer_object *bo); +extern int radeon_bo_get_surface_reg(struct radeon_bo *bo); +#endif diff --git a/drivers/gpu/drm/radeon/radeon_pm.c b/drivers/gpu/drm/radeon/radeon_pm.c new file mode 100644 index 00000000..6fabe89f --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_pm.c @@ -0,0 +1,868 @@ +/* + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Rafał Miłecki + * Alex Deucher + */ +#include "drmP.h" +#include "radeon.h" +#include "avivod.h" +#include "atom.h" +#ifdef CONFIG_ACPI +#include +#endif +#include +#include +#include + +#define RADEON_IDLE_LOOP_MS 100 +#define RADEON_RECLOCK_DELAY_MS 200 +#define RADEON_WAIT_VBLANK_TIMEOUT 200 +#define RADEON_WAIT_IDLE_TIMEOUT 200 + +static const char *radeon_pm_state_type_name[5] = { + "Default", + "Powersave", + "Battery", + "Balanced", + "Performance", +}; + +static void radeon_dynpm_idle_work_handler(struct work_struct *work); +static int radeon_debugfs_pm_init(struct radeon_device *rdev); +static bool radeon_pm_in_vbl(struct radeon_device *rdev); +static bool radeon_pm_debug_check_in_vbl(struct radeon_device *rdev, bool finish); +static void radeon_pm_update_profile(struct radeon_device *rdev); +static void radeon_pm_set_clocks(struct radeon_device *rdev); + +#define ACPI_AC_CLASS "ac_adapter" + +#ifdef CONFIG_ACPI +static int radeon_acpi_event(struct notifier_block *nb, + unsigned long val, + void *data) +{ + struct radeon_device *rdev = container_of(nb, struct radeon_device, acpi_nb); + struct acpi_bus_event *entry = (struct acpi_bus_event *)data; + + if (strcmp(entry->device_class, ACPI_AC_CLASS) == 0) { + if (power_supply_is_system_supplied() > 0) + DRM_DEBUG_DRIVER("pm: AC\n"); + else + DRM_DEBUG_DRIVER("pm: DC\n"); + + if (rdev->pm.pm_method == PM_METHOD_PROFILE) { + if (rdev->pm.profile == PM_PROFILE_AUTO) { + mutex_lock(&rdev->pm.mutex); + radeon_pm_update_profile(rdev); + radeon_pm_set_clocks(rdev); + mutex_unlock(&rdev->pm.mutex); + } + } + } + + return NOTIFY_OK; +} +#endif + +static void radeon_pm_update_profile(struct radeon_device *rdev) +{ + switch (rdev->pm.profile) { + case PM_PROFILE_DEFAULT: + rdev->pm.profile_index = PM_PROFILE_DEFAULT_IDX; + break; + case PM_PROFILE_AUTO: + if (power_supply_is_system_supplied() > 0) { + if (rdev->pm.active_crtc_count > 1) + rdev->pm.profile_index = PM_PROFILE_HIGH_MH_IDX; + else + rdev->pm.profile_index = PM_PROFILE_HIGH_SH_IDX; + } else { + if (rdev->pm.active_crtc_count > 1) + rdev->pm.profile_index = PM_PROFILE_MID_MH_IDX; + else + rdev->pm.profile_index = PM_PROFILE_MID_SH_IDX; + } + break; + case PM_PROFILE_LOW: + if (rdev->pm.active_crtc_count > 1) + rdev->pm.profile_index = PM_PROFILE_LOW_MH_IDX; + else + rdev->pm.profile_index = PM_PROFILE_LOW_SH_IDX; + break; + case PM_PROFILE_MID: + if (rdev->pm.active_crtc_count > 1) + rdev->pm.profile_index = PM_PROFILE_MID_MH_IDX; + else + rdev->pm.profile_index = PM_PROFILE_MID_SH_IDX; + break; + case PM_PROFILE_HIGH: + if (rdev->pm.active_crtc_count > 1) + rdev->pm.profile_index = PM_PROFILE_HIGH_MH_IDX; + else + rdev->pm.profile_index = PM_PROFILE_HIGH_SH_IDX; + break; + } + + if (rdev->pm.active_crtc_count == 0) { + rdev->pm.requested_power_state_index = + rdev->pm.profiles[rdev->pm.profile_index].dpms_off_ps_idx; + rdev->pm.requested_clock_mode_index = + rdev->pm.profiles[rdev->pm.profile_index].dpms_off_cm_idx; + } else { + rdev->pm.requested_power_state_index = + rdev->pm.profiles[rdev->pm.profile_index].dpms_on_ps_idx; + rdev->pm.requested_clock_mode_index = + rdev->pm.profiles[rdev->pm.profile_index].dpms_on_cm_idx; + } +} + +static void radeon_unmap_vram_bos(struct radeon_device *rdev) +{ + struct radeon_bo *bo, *n; + + if (list_empty(&rdev->gem.objects)) + return; + + list_for_each_entry_safe(bo, n, &rdev->gem.objects, list) { + if (bo->tbo.mem.mem_type == TTM_PL_VRAM) + ttm_bo_unmap_virtual(&bo->tbo); + } +} + +static void radeon_sync_with_vblank(struct radeon_device *rdev) +{ + if (rdev->pm.active_crtcs) { + rdev->pm.vblank_sync = false; + wait_event_timeout( + rdev->irq.vblank_queue, rdev->pm.vblank_sync, + msecs_to_jiffies(RADEON_WAIT_VBLANK_TIMEOUT)); + } +} + +static void radeon_set_power_state(struct radeon_device *rdev) +{ + u32 sclk, mclk; + bool misc_after = false; + + if ((rdev->pm.requested_clock_mode_index == rdev->pm.current_clock_mode_index) && + (rdev->pm.requested_power_state_index == rdev->pm.current_power_state_index)) + return; + + if (radeon_gui_idle(rdev)) { + sclk = rdev->pm.power_state[rdev->pm.requested_power_state_index]. + clock_info[rdev->pm.requested_clock_mode_index].sclk; + if (sclk > rdev->pm.default_sclk) + sclk = rdev->pm.default_sclk; + + mclk = rdev->pm.power_state[rdev->pm.requested_power_state_index]. + clock_info[rdev->pm.requested_clock_mode_index].mclk; + if (mclk > rdev->pm.default_mclk) + mclk = rdev->pm.default_mclk; + + /* upvolt before raising clocks, downvolt after lowering clocks */ + if (sclk < rdev->pm.current_sclk) + misc_after = true; + + radeon_sync_with_vblank(rdev); + + if (rdev->pm.pm_method == PM_METHOD_DYNPM) { + if (!radeon_pm_in_vbl(rdev)) + return; + } + + radeon_pm_prepare(rdev); + + if (!misc_after) + /* voltage, pcie lanes, etc.*/ + radeon_pm_misc(rdev); + + /* set engine clock */ + if (sclk != rdev->pm.current_sclk) { + radeon_pm_debug_check_in_vbl(rdev, false); + radeon_set_engine_clock(rdev, sclk); + radeon_pm_debug_check_in_vbl(rdev, true); + rdev->pm.current_sclk = sclk; + DRM_DEBUG_DRIVER("Setting: e: %d\n", sclk); + } + + /* set memory clock */ + if (rdev->asic->set_memory_clock && (mclk != rdev->pm.current_mclk)) { + radeon_pm_debug_check_in_vbl(rdev, false); + radeon_set_memory_clock(rdev, mclk); + radeon_pm_debug_check_in_vbl(rdev, true); + rdev->pm.current_mclk = mclk; + DRM_DEBUG_DRIVER("Setting: m: %d\n", mclk); + } + + if (misc_after) + /* voltage, pcie lanes, etc.*/ + radeon_pm_misc(rdev); + + radeon_pm_finish(rdev); + + rdev->pm.current_power_state_index = rdev->pm.requested_power_state_index; + rdev->pm.current_clock_mode_index = rdev->pm.requested_clock_mode_index; + } else + DRM_DEBUG_DRIVER("pm: GUI not idle!!!\n"); +} + +static void radeon_pm_set_clocks(struct radeon_device *rdev) +{ + int i; + + /* no need to take locks, etc. if nothing's going to change */ + if ((rdev->pm.requested_clock_mode_index == rdev->pm.current_clock_mode_index) && + (rdev->pm.requested_power_state_index == rdev->pm.current_power_state_index)) + return; + + mutex_lock(&rdev->ddev->struct_mutex); + mutex_lock(&rdev->vram_mutex); + mutex_lock(&rdev->cp.mutex); + + /* gui idle int has issues on older chips it seems */ + if (rdev->family >= CHIP_R600) { + if (rdev->irq.installed) { + /* wait for GPU idle */ + rdev->pm.gui_idle = false; + rdev->irq.gui_idle = true; + radeon_irq_set(rdev); + wait_event_interruptible_timeout( + rdev->irq.idle_queue, rdev->pm.gui_idle, + msecs_to_jiffies(RADEON_WAIT_IDLE_TIMEOUT)); + rdev->irq.gui_idle = false; + radeon_irq_set(rdev); + } + } else { + if (rdev->cp.ready) { + struct radeon_fence *fence; + radeon_ring_alloc(rdev, 64); + radeon_fence_create(rdev, &fence); + radeon_fence_emit(rdev, fence); + radeon_ring_commit(rdev); + radeon_fence_wait(fence, false); + radeon_fence_unref(&fence); + } + } + radeon_unmap_vram_bos(rdev); + + if (rdev->irq.installed) { + for (i = 0; i < rdev->num_crtc; i++) { + if (rdev->pm.active_crtcs & (1 << i)) { + rdev->pm.req_vblank |= (1 << i); + drm_vblank_get(rdev->ddev, i); + } + } + } + + radeon_set_power_state(rdev); + + if (rdev->irq.installed) { + for (i = 0; i < rdev->num_crtc; i++) { + if (rdev->pm.req_vblank & (1 << i)) { + rdev->pm.req_vblank &= ~(1 << i); + drm_vblank_put(rdev->ddev, i); + } + } + } + + /* update display watermarks based on new power state */ + radeon_update_bandwidth_info(rdev); + if (rdev->pm.active_crtc_count) + radeon_bandwidth_update(rdev); + + rdev->pm.dynpm_planned_action = DYNPM_ACTION_NONE; + + mutex_unlock(&rdev->cp.mutex); + mutex_unlock(&rdev->vram_mutex); + mutex_unlock(&rdev->ddev->struct_mutex); +} + +static void radeon_pm_print_states(struct radeon_device *rdev) +{ + int i, j; + struct radeon_power_state *power_state; + struct radeon_pm_clock_info *clock_info; + + DRM_DEBUG_DRIVER("%d Power State(s)\n", rdev->pm.num_power_states); + for (i = 0; i < rdev->pm.num_power_states; i++) { + power_state = &rdev->pm.power_state[i]; + DRM_DEBUG_DRIVER("State %d: %s\n", i, + radeon_pm_state_type_name[power_state->type]); + if (i == rdev->pm.default_power_state_index) + DRM_DEBUG_DRIVER("\tDefault"); + if ((rdev->flags & RADEON_IS_PCIE) && !(rdev->flags & RADEON_IS_IGP)) + DRM_DEBUG_DRIVER("\t%d PCIE Lanes\n", power_state->pcie_lanes); + if (power_state->flags & RADEON_PM_STATE_SINGLE_DISPLAY_ONLY) + DRM_DEBUG_DRIVER("\tSingle display only\n"); + DRM_DEBUG_DRIVER("\t%d Clock Mode(s)\n", power_state->num_clock_modes); + for (j = 0; j < power_state->num_clock_modes; j++) { + clock_info = &(power_state->clock_info[j]); + if (rdev->flags & RADEON_IS_IGP) + DRM_DEBUG_DRIVER("\t\t%d e: %d%s\n", + j, + clock_info->sclk * 10, + clock_info->flags & RADEON_PM_MODE_NO_DISPLAY ? "\tNo display only" : ""); + else + DRM_DEBUG_DRIVER("\t\t%d e: %d\tm: %d\tv: %d%s\n", + j, + clock_info->sclk * 10, + clock_info->mclk * 10, + clock_info->voltage.voltage, + clock_info->flags & RADEON_PM_MODE_NO_DISPLAY ? "\tNo display only" : ""); + } + } +} + +static ssize_t radeon_get_pm_profile(struct device *dev, + struct device_attribute *attr, + char *buf) +{ + struct drm_device *ddev = pci_get_drvdata(to_pci_dev(dev)); + struct radeon_device *rdev = ddev->dev_private; + int cp = rdev->pm.profile; + + return snprintf(buf, PAGE_SIZE, "%s\n", + (cp == PM_PROFILE_AUTO) ? "auto" : + (cp == PM_PROFILE_LOW) ? "low" : + (cp == PM_PROFILE_MID) ? "mid" : + (cp == PM_PROFILE_HIGH) ? "high" : "default"); +} + +static ssize_t radeon_set_pm_profile(struct device *dev, + struct device_attribute *attr, + const char *buf, + size_t count) +{ + struct drm_device *ddev = pci_get_drvdata(to_pci_dev(dev)); + struct radeon_device *rdev = ddev->dev_private; + + mutex_lock(&rdev->pm.mutex); + if (rdev->pm.pm_method == PM_METHOD_PROFILE) { + if (strncmp("default", buf, strlen("default")) == 0) + rdev->pm.profile = PM_PROFILE_DEFAULT; + else if (strncmp("auto", buf, strlen("auto")) == 0) + rdev->pm.profile = PM_PROFILE_AUTO; + else if (strncmp("low", buf, strlen("low")) == 0) + rdev->pm.profile = PM_PROFILE_LOW; + else if (strncmp("mid", buf, strlen("mid")) == 0) + rdev->pm.profile = PM_PROFILE_MID; + else if (strncmp("high", buf, strlen("high")) == 0) + rdev->pm.profile = PM_PROFILE_HIGH; + else { + count = -EINVAL; + goto fail; + } + radeon_pm_update_profile(rdev); + radeon_pm_set_clocks(rdev); + } else + count = -EINVAL; + +fail: + mutex_unlock(&rdev->pm.mutex); + + return count; +} + +static ssize_t radeon_get_pm_method(struct device *dev, + struct device_attribute *attr, + char *buf) +{ + struct drm_device *ddev = pci_get_drvdata(to_pci_dev(dev)); + struct radeon_device *rdev = ddev->dev_private; + int pm = rdev->pm.pm_method; + + return snprintf(buf, PAGE_SIZE, "%s\n", + (pm == PM_METHOD_DYNPM) ? "dynpm" : "profile"); +} + +static ssize_t radeon_set_pm_method(struct device *dev, + struct device_attribute *attr, + const char *buf, + size_t count) +{ + struct drm_device *ddev = pci_get_drvdata(to_pci_dev(dev)); + struct radeon_device *rdev = ddev->dev_private; + + + if (strncmp("dynpm", buf, strlen("dynpm")) == 0) { + mutex_lock(&rdev->pm.mutex); + rdev->pm.pm_method = PM_METHOD_DYNPM; + rdev->pm.dynpm_state = DYNPM_STATE_PAUSED; + rdev->pm.dynpm_planned_action = DYNPM_ACTION_DEFAULT; + mutex_unlock(&rdev->pm.mutex); + } else if (strncmp("profile", buf, strlen("profile")) == 0) { + mutex_lock(&rdev->pm.mutex); + /* disable dynpm */ + rdev->pm.dynpm_state = DYNPM_STATE_DISABLED; + rdev->pm.dynpm_planned_action = DYNPM_ACTION_NONE; + rdev->pm.pm_method = PM_METHOD_PROFILE; + mutex_unlock(&rdev->pm.mutex); + cancel_delayed_work_sync(&rdev->pm.dynpm_idle_work); + } else { + count = -EINVAL; + goto fail; + } + radeon_pm_compute_clocks(rdev); +fail: + return count; +} + +static DEVICE_ATTR(power_profile, S_IRUGO | S_IWUSR, radeon_get_pm_profile, radeon_set_pm_profile); +static DEVICE_ATTR(power_method, S_IRUGO | S_IWUSR, radeon_get_pm_method, radeon_set_pm_method); + +static ssize_t radeon_hwmon_show_temp(struct device *dev, + struct device_attribute *attr, + char *buf) +{ + struct drm_device *ddev = pci_get_drvdata(to_pci_dev(dev)); + struct radeon_device *rdev = ddev->dev_private; + int temp; + + switch (rdev->pm.int_thermal_type) { + case THERMAL_TYPE_RV6XX: + temp = rv6xx_get_temp(rdev); + break; + case THERMAL_TYPE_RV770: + temp = rv770_get_temp(rdev); + break; + case THERMAL_TYPE_EVERGREEN: + case THERMAL_TYPE_NI: + temp = evergreen_get_temp(rdev); + break; + case THERMAL_TYPE_SUMO: + temp = sumo_get_temp(rdev); + break; + default: + temp = 0; + break; + } + + return snprintf(buf, PAGE_SIZE, "%d\n", temp); +} + +static ssize_t radeon_hwmon_show_name(struct device *dev, + struct device_attribute *attr, + char *buf) +{ + return sprintf(buf, "radeon\n"); +} + +static SENSOR_DEVICE_ATTR(temp1_input, S_IRUGO, radeon_hwmon_show_temp, NULL, 0); +static SENSOR_DEVICE_ATTR(name, S_IRUGO, radeon_hwmon_show_name, NULL, 0); + +static struct attribute *hwmon_attributes[] = { + &sensor_dev_attr_temp1_input.dev_attr.attr, + &sensor_dev_attr_name.dev_attr.attr, + NULL +}; + +static const struct attribute_group hwmon_attrgroup = { + .attrs = hwmon_attributes, +}; + +static int radeon_hwmon_init(struct radeon_device *rdev) +{ + int err = 0; + + rdev->pm.int_hwmon_dev = NULL; + + switch (rdev->pm.int_thermal_type) { + case THERMAL_TYPE_RV6XX: + case THERMAL_TYPE_RV770: + case THERMAL_TYPE_EVERGREEN: + case THERMAL_TYPE_NI: + case THERMAL_TYPE_SUMO: + rdev->pm.int_hwmon_dev = hwmon_device_register(rdev->dev); + if (IS_ERR(rdev->pm.int_hwmon_dev)) { + err = PTR_ERR(rdev->pm.int_hwmon_dev); + dev_err(rdev->dev, + "Unable to register hwmon device: %d\n", err); + break; + } + dev_set_drvdata(rdev->pm.int_hwmon_dev, rdev->ddev); + err = sysfs_create_group(&rdev->pm.int_hwmon_dev->kobj, + &hwmon_attrgroup); + if (err) { + dev_err(rdev->dev, + "Unable to create hwmon sysfs file: %d\n", err); + hwmon_device_unregister(rdev->dev); + } + break; + default: + break; + } + + return err; +} + +static void radeon_hwmon_fini(struct radeon_device *rdev) +{ + if (rdev->pm.int_hwmon_dev) { + sysfs_remove_group(&rdev->pm.int_hwmon_dev->kobj, &hwmon_attrgroup); + hwmon_device_unregister(rdev->pm.int_hwmon_dev); + } +} + +void radeon_pm_suspend(struct radeon_device *rdev) +{ + mutex_lock(&rdev->pm.mutex); + if (rdev->pm.pm_method == PM_METHOD_DYNPM) { + if (rdev->pm.dynpm_state == DYNPM_STATE_ACTIVE) + rdev->pm.dynpm_state = DYNPM_STATE_SUSPENDED; + } + mutex_unlock(&rdev->pm.mutex); + + cancel_delayed_work_sync(&rdev->pm.dynpm_idle_work); +} + +void radeon_pm_resume(struct radeon_device *rdev) +{ + /* set up the default clocks if the MC ucode is loaded */ + if (ASIC_IS_DCE5(rdev) && rdev->mc_fw) { + if (rdev->pm.default_vddc) + radeon_atom_set_voltage(rdev, rdev->pm.default_vddc, + SET_VOLTAGE_TYPE_ASIC_VDDC); + if (rdev->pm.default_vddci) + radeon_atom_set_voltage(rdev, rdev->pm.default_vddci, + SET_VOLTAGE_TYPE_ASIC_VDDCI); + if (rdev->pm.default_sclk) + radeon_set_engine_clock(rdev, rdev->pm.default_sclk); + if (rdev->pm.default_mclk) + radeon_set_memory_clock(rdev, rdev->pm.default_mclk); + } + /* asic init will reset the default power state */ + mutex_lock(&rdev->pm.mutex); + rdev->pm.current_power_state_index = rdev->pm.default_power_state_index; + rdev->pm.current_clock_mode_index = 0; + rdev->pm.current_sclk = rdev->pm.default_sclk; + rdev->pm.current_mclk = rdev->pm.default_mclk; + rdev->pm.current_vddc = rdev->pm.power_state[rdev->pm.default_power_state_index].clock_info[0].voltage.voltage; + rdev->pm.current_vddci = rdev->pm.power_state[rdev->pm.default_power_state_index].clock_info[0].voltage.vddci; + if (rdev->pm.pm_method == PM_METHOD_DYNPM + && rdev->pm.dynpm_state == DYNPM_STATE_SUSPENDED) { + rdev->pm.dynpm_state = DYNPM_STATE_ACTIVE; + schedule_delayed_work(&rdev->pm.dynpm_idle_work, + msecs_to_jiffies(RADEON_IDLE_LOOP_MS)); + } + mutex_unlock(&rdev->pm.mutex); + radeon_pm_compute_clocks(rdev); +} + +int radeon_pm_init(struct radeon_device *rdev) +{ + int ret; + + /* default to profile method */ + rdev->pm.pm_method = PM_METHOD_PROFILE; + rdev->pm.profile = PM_PROFILE_DEFAULT; + rdev->pm.dynpm_state = DYNPM_STATE_DISABLED; + rdev->pm.dynpm_planned_action = DYNPM_ACTION_NONE; + rdev->pm.dynpm_can_upclock = true; + rdev->pm.dynpm_can_downclock = true; + rdev->pm.default_sclk = rdev->clock.default_sclk; + rdev->pm.default_mclk = rdev->clock.default_mclk; + rdev->pm.current_sclk = rdev->clock.default_sclk; + rdev->pm.current_mclk = rdev->clock.default_mclk; + rdev->pm.int_thermal_type = THERMAL_TYPE_NONE; + + if (rdev->bios) { + if (rdev->is_atom_bios) + radeon_atombios_get_power_modes(rdev); + else + radeon_combios_get_power_modes(rdev); + radeon_pm_print_states(rdev); + radeon_pm_init_profile(rdev); + /* set up the default clocks if the MC ucode is loaded */ + if (ASIC_IS_DCE5(rdev) && rdev->mc_fw) { + if (rdev->pm.default_vddc) + radeon_atom_set_voltage(rdev, rdev->pm.default_vddc, + SET_VOLTAGE_TYPE_ASIC_VDDC); + if (rdev->pm.default_vddci) + radeon_atom_set_voltage(rdev, rdev->pm.default_vddci, + SET_VOLTAGE_TYPE_ASIC_VDDCI); + if (rdev->pm.default_sclk) + radeon_set_engine_clock(rdev, rdev->pm.default_sclk); + if (rdev->pm.default_mclk) + radeon_set_memory_clock(rdev, rdev->pm.default_mclk); + } + } + + /* set up the internal thermal sensor if applicable */ + ret = radeon_hwmon_init(rdev); + if (ret) + return ret; + + INIT_DELAYED_WORK(&rdev->pm.dynpm_idle_work, radeon_dynpm_idle_work_handler); + + if (rdev->pm.num_power_states > 1) { + /* where's the best place to put these? */ + ret = device_create_file(rdev->dev, &dev_attr_power_profile); + if (ret) + DRM_ERROR("failed to create device file for power profile\n"); + ret = device_create_file(rdev->dev, &dev_attr_power_method); + if (ret) + DRM_ERROR("failed to create device file for power method\n"); + +#ifdef CONFIG_ACPI + rdev->acpi_nb.notifier_call = radeon_acpi_event; + register_acpi_notifier(&rdev->acpi_nb); +#endif + if (radeon_debugfs_pm_init(rdev)) { + DRM_ERROR("Failed to register debugfs file for PM!\n"); + } + + DRM_INFO("radeon: power management initialized\n"); + } + + return 0; +} + +void radeon_pm_fini(struct radeon_device *rdev) +{ + if (rdev->pm.num_power_states > 1) { + mutex_lock(&rdev->pm.mutex); + if (rdev->pm.pm_method == PM_METHOD_PROFILE) { + rdev->pm.profile = PM_PROFILE_DEFAULT; + radeon_pm_update_profile(rdev); + radeon_pm_set_clocks(rdev); + } else if (rdev->pm.pm_method == PM_METHOD_DYNPM) { + /* reset default clocks */ + rdev->pm.dynpm_state = DYNPM_STATE_DISABLED; + rdev->pm.dynpm_planned_action = DYNPM_ACTION_DEFAULT; + radeon_pm_set_clocks(rdev); + } + mutex_unlock(&rdev->pm.mutex); + + cancel_delayed_work_sync(&rdev->pm.dynpm_idle_work); + + device_remove_file(rdev->dev, &dev_attr_power_profile); + device_remove_file(rdev->dev, &dev_attr_power_method); +#ifdef CONFIG_ACPI + unregister_acpi_notifier(&rdev->acpi_nb); +#endif + } + + if (rdev->pm.power_state) + kfree(rdev->pm.power_state); + + radeon_hwmon_fini(rdev); +} + +void radeon_pm_compute_clocks(struct radeon_device *rdev) +{ + struct drm_device *ddev = rdev->ddev; + struct drm_crtc *crtc; + struct radeon_crtc *radeon_crtc; + + if (rdev->pm.num_power_states < 2) + return; + + mutex_lock(&rdev->pm.mutex); + + rdev->pm.active_crtcs = 0; + rdev->pm.active_crtc_count = 0; + list_for_each_entry(crtc, + &ddev->mode_config.crtc_list, head) { + radeon_crtc = to_radeon_crtc(crtc); + if (radeon_crtc->enabled) { + rdev->pm.active_crtcs |= (1 << radeon_crtc->crtc_id); + rdev->pm.active_crtc_count++; + } + } + + if (rdev->pm.pm_method == PM_METHOD_PROFILE) { + radeon_pm_update_profile(rdev); + radeon_pm_set_clocks(rdev); + } else if (rdev->pm.pm_method == PM_METHOD_DYNPM) { + if (rdev->pm.dynpm_state != DYNPM_STATE_DISABLED) { + if (rdev->pm.active_crtc_count > 1) { + if (rdev->pm.dynpm_state == DYNPM_STATE_ACTIVE) { + cancel_delayed_work(&rdev->pm.dynpm_idle_work); + + rdev->pm.dynpm_state = DYNPM_STATE_PAUSED; + rdev->pm.dynpm_planned_action = DYNPM_ACTION_DEFAULT; + radeon_pm_get_dynpm_state(rdev); + radeon_pm_set_clocks(rdev); + + DRM_DEBUG_DRIVER("radeon: dynamic power management deactivated\n"); + } + } else if (rdev->pm.active_crtc_count == 1) { + /* TODO: Increase clocks if needed for current mode */ + + if (rdev->pm.dynpm_state == DYNPM_STATE_MINIMUM) { + rdev->pm.dynpm_state = DYNPM_STATE_ACTIVE; + rdev->pm.dynpm_planned_action = DYNPM_ACTION_UPCLOCK; + radeon_pm_get_dynpm_state(rdev); + radeon_pm_set_clocks(rdev); + + schedule_delayed_work(&rdev->pm.dynpm_idle_work, + msecs_to_jiffies(RADEON_IDLE_LOOP_MS)); + } else if (rdev->pm.dynpm_state == DYNPM_STATE_PAUSED) { + rdev->pm.dynpm_state = DYNPM_STATE_ACTIVE; + schedule_delayed_work(&rdev->pm.dynpm_idle_work, + msecs_to_jiffies(RADEON_IDLE_LOOP_MS)); + DRM_DEBUG_DRIVER("radeon: dynamic power management activated\n"); + } + } else { /* count == 0 */ + if (rdev->pm.dynpm_state != DYNPM_STATE_MINIMUM) { + cancel_delayed_work(&rdev->pm.dynpm_idle_work); + + rdev->pm.dynpm_state = DYNPM_STATE_MINIMUM; + rdev->pm.dynpm_planned_action = DYNPM_ACTION_MINIMUM; + radeon_pm_get_dynpm_state(rdev); + radeon_pm_set_clocks(rdev); + } + } + } + } + + mutex_unlock(&rdev->pm.mutex); +} + +static bool radeon_pm_in_vbl(struct radeon_device *rdev) +{ + int crtc, vpos, hpos, vbl_status; + bool in_vbl = true; + + /* Iterate over all active crtc's. All crtc's must be in vblank, + * otherwise return in_vbl == false. + */ + for (crtc = 0; (crtc < rdev->num_crtc) && in_vbl; crtc++) { + if (rdev->pm.active_crtcs & (1 << crtc)) { + vbl_status = radeon_get_crtc_scanoutpos(rdev->ddev, crtc, &vpos, &hpos); + if ((vbl_status & DRM_SCANOUTPOS_VALID) && + !(vbl_status & DRM_SCANOUTPOS_INVBL)) + in_vbl = false; + } + } + + return in_vbl; +} + +static bool radeon_pm_debug_check_in_vbl(struct radeon_device *rdev, bool finish) +{ + u32 stat_crtc = 0; + bool in_vbl = radeon_pm_in_vbl(rdev); + + if (in_vbl == false) + DRM_DEBUG_DRIVER("not in vbl for pm change %08x at %s\n", stat_crtc, + finish ? "exit" : "entry"); + return in_vbl; +} + +static void radeon_dynpm_idle_work_handler(struct work_struct *work) +{ + struct radeon_device *rdev; + int resched; + rdev = container_of(work, struct radeon_device, + pm.dynpm_idle_work.work); + + resched = ttm_bo_lock_delayed_workqueue(&rdev->mman.bdev); + mutex_lock(&rdev->pm.mutex); + if (rdev->pm.dynpm_state == DYNPM_STATE_ACTIVE) { + unsigned long irq_flags; + int not_processed = 0; + + read_lock_irqsave(&rdev->fence_drv.lock, irq_flags); + if (!list_empty(&rdev->fence_drv.emited)) { + struct list_head *ptr; + list_for_each(ptr, &rdev->fence_drv.emited) { + /* count up to 3, that's enought info */ + if (++not_processed >= 3) + break; + } + } + read_unlock_irqrestore(&rdev->fence_drv.lock, irq_flags); + + if (not_processed >= 3) { /* should upclock */ + if (rdev->pm.dynpm_planned_action == DYNPM_ACTION_DOWNCLOCK) { + rdev->pm.dynpm_planned_action = DYNPM_ACTION_NONE; + } else if (rdev->pm.dynpm_planned_action == DYNPM_ACTION_NONE && + rdev->pm.dynpm_can_upclock) { + rdev->pm.dynpm_planned_action = + DYNPM_ACTION_UPCLOCK; + rdev->pm.dynpm_action_timeout = jiffies + + msecs_to_jiffies(RADEON_RECLOCK_DELAY_MS); + } + } else if (not_processed == 0) { /* should downclock */ + if (rdev->pm.dynpm_planned_action == DYNPM_ACTION_UPCLOCK) { + rdev->pm.dynpm_planned_action = DYNPM_ACTION_NONE; + } else if (rdev->pm.dynpm_planned_action == DYNPM_ACTION_NONE && + rdev->pm.dynpm_can_downclock) { + rdev->pm.dynpm_planned_action = + DYNPM_ACTION_DOWNCLOCK; + rdev->pm.dynpm_action_timeout = jiffies + + msecs_to_jiffies(RADEON_RECLOCK_DELAY_MS); + } + } + + /* Note, radeon_pm_set_clocks is called with static_switch set + * to false since we want to wait for vbl to avoid flicker. + */ + if (rdev->pm.dynpm_planned_action != DYNPM_ACTION_NONE && + jiffies > rdev->pm.dynpm_action_timeout) { + radeon_pm_get_dynpm_state(rdev); + radeon_pm_set_clocks(rdev); + } + + schedule_delayed_work(&rdev->pm.dynpm_idle_work, + msecs_to_jiffies(RADEON_IDLE_LOOP_MS)); + } + mutex_unlock(&rdev->pm.mutex); + ttm_bo_unlock_delayed_workqueue(&rdev->mman.bdev, resched); +} + +/* + * Debugfs info + */ +#if defined(CONFIG_DEBUG_FS) + +static int radeon_debugfs_pm_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + struct radeon_device *rdev = dev->dev_private; + + seq_printf(m, "default engine clock: %u0 kHz\n", rdev->pm.default_sclk); + seq_printf(m, "current engine clock: %u0 kHz\n", radeon_get_engine_clock(rdev)); + seq_printf(m, "default memory clock: %u0 kHz\n", rdev->pm.default_mclk); + if (rdev->asic->get_memory_clock) + seq_printf(m, "current memory clock: %u0 kHz\n", radeon_get_memory_clock(rdev)); + if (rdev->pm.current_vddc) + seq_printf(m, "voltage: %u mV\n", rdev->pm.current_vddc); + if (rdev->asic->get_pcie_lanes) + seq_printf(m, "PCIE lanes: %d\n", radeon_get_pcie_lanes(rdev)); + + return 0; +} + +static struct drm_info_list radeon_pm_info_list[] = { + {"radeon_pm_info", radeon_debugfs_pm_info, 0, NULL}, +}; +#endif + +static int radeon_debugfs_pm_init(struct radeon_device *rdev) +{ +#if defined(CONFIG_DEBUG_FS) + return radeon_debugfs_add_files(rdev, radeon_pm_info_list, ARRAY_SIZE(radeon_pm_info_list)); +#else + return 0; +#endif +} diff --git a/drivers/gpu/drm/radeon/radeon_reg.h b/drivers/gpu/drm/radeon/radeon_reg.h new file mode 100644 index 00000000..bc44a3d3 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_reg.h @@ -0,0 +1,3706 @@ +/* + * Copyright 2000 ATI Technologies Inc., Markham, Ontario, and + * VA Linux Systems Inc., Fremont, California. + * + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation on the rights to use, copy, modify, merge, + * publish, distribute, sublicense, and/or sell copies of the Software, + * and to permit persons to whom the Software is furnished to do so, + * subject to the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial + * portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NON-INFRINGEMENT. IN NO EVENT SHALL ATI, VA LINUX SYSTEMS AND/OR + * THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + */ + +/* + * Authors: + * Kevin E. Martin + * Rickard E. Faith + * Alan Hourihane + * + * References: + * + * !!!! FIXME !!!! + * RAGE 128 VR/ RAGE 128 GL Register Reference Manual (Technical + * Reference Manual P/N RRG-G04100-C Rev. 0.04), ATI Technologies: April + * 1999. + * + * !!!! FIXME !!!! + * RAGE 128 Software Development Manual (Technical Reference Manual P/N + * SDK-G04000 Rev. 0.01), ATI Technologies: June 1999. + * + */ + +/* !!!! FIXME !!!! NOTE: THIS FILE HAS BEEN CONVERTED FROM r128_reg.h + * AND CONTAINS REGISTERS AND REGISTER DEFINITIONS THAT ARE NOT CORRECT + * ON THE RADEON. A FULL AUDIT OF THIS CODE IS NEEDED! */ +#ifndef _RADEON_REG_H_ +#define _RADEON_REG_H_ + +#include "r300_reg.h" +#include "r500_reg.h" +#include "r600_reg.h" +#include "evergreen_reg.h" +#include "ni_reg.h" + +#define RADEON_MC_AGP_LOCATION 0x014c +#define RADEON_MC_AGP_START_MASK 0x0000FFFF +#define RADEON_MC_AGP_START_SHIFT 0 +#define RADEON_MC_AGP_TOP_MASK 0xFFFF0000 +#define RADEON_MC_AGP_TOP_SHIFT 16 +#define RADEON_MC_FB_LOCATION 0x0148 +#define RADEON_MC_FB_START_MASK 0x0000FFFF +#define RADEON_MC_FB_START_SHIFT 0 +#define RADEON_MC_FB_TOP_MASK 0xFFFF0000 +#define RADEON_MC_FB_TOP_SHIFT 16 +#define RADEON_AGP_BASE_2 0x015c /* r200+ only */ +#define RADEON_AGP_BASE 0x0170 + +#define ATI_DATATYPE_VQ 0 +#define ATI_DATATYPE_CI4 1 +#define ATI_DATATYPE_CI8 2 +#define ATI_DATATYPE_ARGB1555 3 +#define ATI_DATATYPE_RGB565 4 +#define ATI_DATATYPE_RGB888 5 +#define ATI_DATATYPE_ARGB8888 6 +#define ATI_DATATYPE_RGB332 7 +#define ATI_DATATYPE_Y8 8 +#define ATI_DATATYPE_RGB8 9 +#define ATI_DATATYPE_CI16 10 +#define ATI_DATATYPE_VYUY_422 11 +#define ATI_DATATYPE_YVYU_422 12 +#define ATI_DATATYPE_AYUV_444 14 +#define ATI_DATATYPE_ARGB4444 15 + + /* Registers for 2D/Video/Overlay */ +#define RADEON_ADAPTER_ID 0x0f2c /* PCI */ +#define RADEON_AGP_BASE 0x0170 +#define RADEON_AGP_CNTL 0x0174 +# define RADEON_AGP_APER_SIZE_256MB (0x00 << 0) +# define RADEON_AGP_APER_SIZE_128MB (0x20 << 0) +# define RADEON_AGP_APER_SIZE_64MB (0x30 << 0) +# define RADEON_AGP_APER_SIZE_32MB (0x38 << 0) +# define RADEON_AGP_APER_SIZE_16MB (0x3c << 0) +# define RADEON_AGP_APER_SIZE_8MB (0x3e << 0) +# define RADEON_AGP_APER_SIZE_4MB (0x3f << 0) +# define RADEON_AGP_APER_SIZE_MASK (0x3f << 0) +#define RADEON_STATUS_PCI_CONFIG 0x06 +# define RADEON_CAP_LIST 0x100000 +#define RADEON_CAPABILITIES_PTR_PCI_CONFIG 0x34 /* offset in PCI config*/ +# define RADEON_CAP_PTR_MASK 0xfc /* mask off reserved bits of CAP_PTR */ +# define RADEON_CAP_ID_NULL 0x00 /* End of capability list */ +# define RADEON_CAP_ID_AGP 0x02 /* AGP capability ID */ +# define RADEON_CAP_ID_EXP 0x10 /* PCI Express */ +#define RADEON_AGP_COMMAND 0x0f60 /* PCI */ +#define RADEON_AGP_COMMAND_PCI_CONFIG 0x0060 /* offset in PCI config*/ +# define RADEON_AGP_ENABLE (1<<8) +#define RADEON_AGP_PLL_CNTL 0x000b /* PLL */ +#define RADEON_AGP_STATUS 0x0f5c /* PCI */ +# define RADEON_AGP_1X_MODE 0x01 +# define RADEON_AGP_2X_MODE 0x02 +# define RADEON_AGP_4X_MODE 0x04 +# define RADEON_AGP_FW_MODE 0x10 +# define RADEON_AGP_MODE_MASK 0x17 +# define RADEON_AGPv3_MODE 0x08 +# define RADEON_AGPv3_4X_MODE 0x01 +# define RADEON_AGPv3_8X_MODE 0x02 +#define RADEON_ATTRDR 0x03c1 /* VGA */ +#define RADEON_ATTRDW 0x03c0 /* VGA */ +#define RADEON_ATTRX 0x03c0 /* VGA */ +#define RADEON_AUX_SC_CNTL 0x1660 +# define RADEON_AUX1_SC_EN (1 << 0) +# define RADEON_AUX1_SC_MODE_OR (0 << 1) +# define RADEON_AUX1_SC_MODE_NAND (1 << 1) +# define RADEON_AUX2_SC_EN (1 << 2) +# define RADEON_AUX2_SC_MODE_OR (0 << 3) +# define RADEON_AUX2_SC_MODE_NAND (1 << 3) +# define RADEON_AUX3_SC_EN (1 << 4) +# define RADEON_AUX3_SC_MODE_OR (0 << 5) +# define RADEON_AUX3_SC_MODE_NAND (1 << 5) +#define RADEON_AUX1_SC_BOTTOM 0x1670 +#define RADEON_AUX1_SC_LEFT 0x1664 +#define RADEON_AUX1_SC_RIGHT 0x1668 +#define RADEON_AUX1_SC_TOP 0x166c +#define RADEON_AUX2_SC_BOTTOM 0x1680 +#define RADEON_AUX2_SC_LEFT 0x1674 +#define RADEON_AUX2_SC_RIGHT 0x1678 +#define RADEON_AUX2_SC_TOP 0x167c +#define RADEON_AUX3_SC_BOTTOM 0x1690 +#define RADEON_AUX3_SC_LEFT 0x1684 +#define RADEON_AUX3_SC_RIGHT 0x1688 +#define RADEON_AUX3_SC_TOP 0x168c +#define RADEON_AUX_WINDOW_HORZ_CNTL 0x02d8 +#define RADEON_AUX_WINDOW_VERT_CNTL 0x02dc + +#define RADEON_BASE_CODE 0x0f0b +#define RADEON_BIOS_0_SCRATCH 0x0010 +# define RADEON_FP_PANEL_SCALABLE (1 << 16) +# define RADEON_FP_PANEL_SCALE_EN (1 << 17) +# define RADEON_FP_CHIP_SCALE_EN (1 << 18) +# define RADEON_DRIVER_BRIGHTNESS_EN (1 << 26) +# define RADEON_DISPLAY_ROT_MASK (3 << 28) +# define RADEON_DISPLAY_ROT_00 (0 << 28) +# define RADEON_DISPLAY_ROT_90 (1 << 28) +# define RADEON_DISPLAY_ROT_180 (2 << 28) +# define RADEON_DISPLAY_ROT_270 (3 << 28) +#define RADEON_BIOS_1_SCRATCH 0x0014 +#define RADEON_BIOS_2_SCRATCH 0x0018 +#define RADEON_BIOS_3_SCRATCH 0x001c +#define RADEON_BIOS_4_SCRATCH 0x0020 +# define RADEON_CRT1_ATTACHED_MASK (3 << 0) +# define RADEON_CRT1_ATTACHED_MONO (1 << 0) +# define RADEON_CRT1_ATTACHED_COLOR (2 << 0) +# define RADEON_LCD1_ATTACHED (1 << 2) +# define RADEON_DFP1_ATTACHED (1 << 3) +# define RADEON_TV1_ATTACHED_MASK (3 << 4) +# define RADEON_TV1_ATTACHED_COMP (1 << 4) +# define RADEON_TV1_ATTACHED_SVIDEO (2 << 4) +# define RADEON_CRT2_ATTACHED_MASK (3 << 8) +# define RADEON_CRT2_ATTACHED_MONO (1 << 8) +# define RADEON_CRT2_ATTACHED_COLOR (2 << 8) +# define RADEON_DFP2_ATTACHED (1 << 11) +#define RADEON_BIOS_5_SCRATCH 0x0024 +# define RADEON_LCD1_ON (1 << 0) +# define RADEON_CRT1_ON (1 << 1) +# define RADEON_TV1_ON (1 << 2) +# define RADEON_DFP1_ON (1 << 3) +# define RADEON_CRT2_ON (1 << 5) +# define RADEON_CV1_ON (1 << 6) +# define RADEON_DFP2_ON (1 << 7) +# define RADEON_LCD1_CRTC_MASK (1 << 8) +# define RADEON_LCD1_CRTC_SHIFT 8 +# define RADEON_CRT1_CRTC_MASK (1 << 9) +# define RADEON_CRT1_CRTC_SHIFT 9 +# define RADEON_TV1_CRTC_MASK (1 << 10) +# define RADEON_TV1_CRTC_SHIFT 10 +# define RADEON_DFP1_CRTC_MASK (1 << 11) +# define RADEON_DFP1_CRTC_SHIFT 11 +# define RADEON_CRT2_CRTC_MASK (1 << 12) +# define RADEON_CRT2_CRTC_SHIFT 12 +# define RADEON_CV1_CRTC_MASK (1 << 13) +# define RADEON_CV1_CRTC_SHIFT 13 +# define RADEON_DFP2_CRTC_MASK (1 << 14) +# define RADEON_DFP2_CRTC_SHIFT 14 +# define RADEON_ACC_REQ_LCD1 (1 << 16) +# define RADEON_ACC_REQ_CRT1 (1 << 17) +# define RADEON_ACC_REQ_TV1 (1 << 18) +# define RADEON_ACC_REQ_DFP1 (1 << 19) +# define RADEON_ACC_REQ_CRT2 (1 << 21) +# define RADEON_ACC_REQ_TV2 (1 << 22) +# define RADEON_ACC_REQ_DFP2 (1 << 23) +#define RADEON_BIOS_6_SCRATCH 0x0028 +# define RADEON_ACC_MODE_CHANGE (1 << 2) +# define RADEON_EXT_DESKTOP_MODE (1 << 3) +# define RADEON_LCD_DPMS_ON (1 << 20) +# define RADEON_CRT_DPMS_ON (1 << 21) +# define RADEON_TV_DPMS_ON (1 << 22) +# define RADEON_DFP_DPMS_ON (1 << 23) +# define RADEON_DPMS_MASK (3 << 24) +# define RADEON_DPMS_ON (0 << 24) +# define RADEON_DPMS_STANDBY (1 << 24) +# define RADEON_DPMS_SUSPEND (2 << 24) +# define RADEON_DPMS_OFF (3 << 24) +# define RADEON_SCREEN_BLANKING (1 << 26) +# define RADEON_DRIVER_CRITICAL (1 << 27) +# define RADEON_DISPLAY_SWITCHING_DIS (1 << 30) +#define RADEON_BIOS_7_SCRATCH 0x002c +# define RADEON_SYS_HOTKEY (1 << 10) +# define RADEON_DRV_LOADED (1 << 12) +#define RADEON_BIOS_ROM 0x0f30 /* PCI */ +#define RADEON_BIST 0x0f0f /* PCI */ +#define RADEON_BRUSH_DATA0 0x1480 +#define RADEON_BRUSH_DATA1 0x1484 +#define RADEON_BRUSH_DATA10 0x14a8 +#define RADEON_BRUSH_DATA11 0x14ac +#define RADEON_BRUSH_DATA12 0x14b0 +#define RADEON_BRUSH_DATA13 0x14b4 +#define RADEON_BRUSH_DATA14 0x14b8 +#define RADEON_BRUSH_DATA15 0x14bc +#define RADEON_BRUSH_DATA16 0x14c0 +#define RADEON_BRUSH_DATA17 0x14c4 +#define RADEON_BRUSH_DATA18 0x14c8 +#define RADEON_BRUSH_DATA19 0x14cc +#define RADEON_BRUSH_DATA2 0x1488 +#define RADEON_BRUSH_DATA20 0x14d0 +#define RADEON_BRUSH_DATA21 0x14d4 +#define RADEON_BRUSH_DATA22 0x14d8 +#define RADEON_BRUSH_DATA23 0x14dc +#define RADEON_BRUSH_DATA24 0x14e0 +#define RADEON_BRUSH_DATA25 0x14e4 +#define RADEON_BRUSH_DATA26 0x14e8 +#define RADEON_BRUSH_DATA27 0x14ec +#define RADEON_BRUSH_DATA28 0x14f0 +#define RADEON_BRUSH_DATA29 0x14f4 +#define RADEON_BRUSH_DATA3 0x148c +#define RADEON_BRUSH_DATA30 0x14f8 +#define RADEON_BRUSH_DATA31 0x14fc +#define RADEON_BRUSH_DATA32 0x1500 +#define RADEON_BRUSH_DATA33 0x1504 +#define RADEON_BRUSH_DATA34 0x1508 +#define RADEON_BRUSH_DATA35 0x150c +#define RADEON_BRUSH_DATA36 0x1510 +#define RADEON_BRUSH_DATA37 0x1514 +#define RADEON_BRUSH_DATA38 0x1518 +#define RADEON_BRUSH_DATA39 0x151c +#define RADEON_BRUSH_DATA4 0x1490 +#define RADEON_BRUSH_DATA40 0x1520 +#define RADEON_BRUSH_DATA41 0x1524 +#define RADEON_BRUSH_DATA42 0x1528 +#define RADEON_BRUSH_DATA43 0x152c +#define RADEON_BRUSH_DATA44 0x1530 +#define RADEON_BRUSH_DATA45 0x1534 +#define RADEON_BRUSH_DATA46 0x1538 +#define RADEON_BRUSH_DATA47 0x153c +#define RADEON_BRUSH_DATA48 0x1540 +#define RADEON_BRUSH_DATA49 0x1544 +#define RADEON_BRUSH_DATA5 0x1494 +#define RADEON_BRUSH_DATA50 0x1548 +#define RADEON_BRUSH_DATA51 0x154c +#define RADEON_BRUSH_DATA52 0x1550 +#define RADEON_BRUSH_DATA53 0x1554 +#define RADEON_BRUSH_DATA54 0x1558 +#define RADEON_BRUSH_DATA55 0x155c +#define RADEON_BRUSH_DATA56 0x1560 +#define RADEON_BRUSH_DATA57 0x1564 +#define RADEON_BRUSH_DATA58 0x1568 +#define RADEON_BRUSH_DATA59 0x156c +#define RADEON_BRUSH_DATA6 0x1498 +#define RADEON_BRUSH_DATA60 0x1570 +#define RADEON_BRUSH_DATA61 0x1574 +#define RADEON_BRUSH_DATA62 0x1578 +#define RADEON_BRUSH_DATA63 0x157c +#define RADEON_BRUSH_DATA7 0x149c +#define RADEON_BRUSH_DATA8 0x14a0 +#define RADEON_BRUSH_DATA9 0x14a4 +#define RADEON_BRUSH_SCALE 0x1470 +#define RADEON_BRUSH_Y_X 0x1474 +#define RADEON_BUS_CNTL 0x0030 +# define RADEON_BUS_MASTER_DIS (1 << 6) +# define RADEON_BUS_BIOS_DIS_ROM (1 << 12) +# define RS600_BUS_MASTER_DIS (1 << 14) +# define RS600_MSI_REARM (1 << 20) /* rs600/rs690/rs740 */ +# define RADEON_BUS_RD_DISCARD_EN (1 << 24) +# define RADEON_BUS_RD_ABORT_EN (1 << 25) +# define RADEON_BUS_MSTR_DISCONNECT_EN (1 << 28) +# define RADEON_BUS_WRT_BURST (1 << 29) +# define RADEON_BUS_READ_BURST (1 << 30) +#define RADEON_BUS_CNTL1 0x0034 +# define RADEON_BUS_WAIT_ON_LOCK_EN (1 << 4) +#define RV370_BUS_CNTL 0x004c +# define RV370_BUS_BIOS_DIS_ROM (1 << 2) +/* rv370/rv380, rv410, r423/r430/r480, r5xx */ +#define RADEON_MSI_REARM_EN 0x0160 +# define RV370_MSI_REARM_EN (1 << 0) + +/* #define RADEON_PCIE_INDEX 0x0030 */ +/* #define RADEON_PCIE_DATA 0x0034 */ +#define RADEON_PCIE_LC_LINK_WIDTH_CNTL 0xa2 /* PCIE */ +# define RADEON_PCIE_LC_LINK_WIDTH_SHIFT 0 +# define RADEON_PCIE_LC_LINK_WIDTH_MASK 0x7 +# define RADEON_PCIE_LC_LINK_WIDTH_X0 0 +# define RADEON_PCIE_LC_LINK_WIDTH_X1 1 +# define RADEON_PCIE_LC_LINK_WIDTH_X2 2 +# define RADEON_PCIE_LC_LINK_WIDTH_X4 3 +# define RADEON_PCIE_LC_LINK_WIDTH_X8 4 +# define RADEON_PCIE_LC_LINK_WIDTH_X12 5 +# define RADEON_PCIE_LC_LINK_WIDTH_X16 6 +# define RADEON_PCIE_LC_LINK_WIDTH_RD_SHIFT 4 +# define RADEON_PCIE_LC_LINK_WIDTH_RD_MASK 0x70 +# define RADEON_PCIE_LC_RECONFIG_NOW (1 << 8) +# define RADEON_PCIE_LC_RECONFIG_LATER (1 << 9) +# define RADEON_PCIE_LC_SHORT_RECONFIG_EN (1 << 10) +# define R600_PCIE_LC_RECONFIG_ARC_MISSING_ESCAPE (1 << 7) +# define R600_PCIE_LC_RENEGOTIATION_SUPPORT (1 << 9) +# define R600_PCIE_LC_RENEGOTIATE_EN (1 << 10) +# define R600_PCIE_LC_SHORT_RECONFIG_EN (1 << 11) +# define R600_PCIE_LC_UPCONFIGURE_SUPPORT (1 << 12) +# define R600_PCIE_LC_UPCONFIGURE_DIS (1 << 13) + +#define R600_TARGET_AND_CURRENT_PROFILE_INDEX 0x70c +#define R700_TARGET_AND_CURRENT_PROFILE_INDEX 0x66c + +#define RADEON_CACHE_CNTL 0x1724 +#define RADEON_CACHE_LINE 0x0f0c /* PCI */ +#define RADEON_CAPABILITIES_ID 0x0f50 /* PCI */ +#define RADEON_CAPABILITIES_PTR 0x0f34 /* PCI */ +#define RADEON_CLK_PIN_CNTL 0x0001 /* PLL */ +# define RADEON_DONT_USE_XTALIN (1 << 4) +# define RADEON_SCLK_DYN_START_CNTL (1 << 15) +#define RADEON_CLOCK_CNTL_DATA 0x000c +#define RADEON_CLOCK_CNTL_INDEX 0x0008 +# define RADEON_PLL_WR_EN (1 << 7) +# define RADEON_PLL_DIV_SEL (3 << 8) +# define RADEON_PLL2_DIV_SEL_MASK (~(3 << 8)) +#define RADEON_CLK_PWRMGT_CNTL 0x0014 +# define RADEON_ENGIN_DYNCLK_MODE (1 << 12) +# define RADEON_ACTIVE_HILO_LAT_MASK (3 << 13) +# define RADEON_ACTIVE_HILO_LAT_SHIFT 13 +# define RADEON_DISP_DYN_STOP_LAT_MASK (1 << 12) +# define RADEON_MC_BUSY (1 << 16) +# define RADEON_DLL_READY (1 << 19) +# define RADEON_CG_NO1_DEBUG_0 (1 << 24) +# define RADEON_CG_NO1_DEBUG_MASK (0x1f << 24) +# define RADEON_DYN_STOP_MODE_MASK (7 << 21) +# define RADEON_TVPLL_PWRMGT_OFF (1 << 30) +# define RADEON_TVCLK_TURNOFF (1 << 31) +#define RADEON_PLL_PWRMGT_CNTL 0x0015 /* PLL */ +# define RADEON_PM_MODE_SEL (1 << 13) +# define RADEON_TCL_BYPASS_DISABLE (1 << 20) +#define RADEON_CLR_CMP_CLR_3D 0x1a24 +#define RADEON_CLR_CMP_CLR_DST 0x15c8 +#define RADEON_CLR_CMP_CLR_SRC 0x15c4 +#define RADEON_CLR_CMP_CNTL 0x15c0 +# define RADEON_SRC_CMP_EQ_COLOR (4 << 0) +# define RADEON_SRC_CMP_NEQ_COLOR (5 << 0) +# define RADEON_CLR_CMP_SRC_SOURCE (1 << 24) +#define RADEON_CLR_CMP_MASK 0x15cc +# define RADEON_CLR_CMP_MSK 0xffffffff +#define RADEON_CLR_CMP_MASK_3D 0x1A28 +#define RADEON_COMMAND 0x0f04 /* PCI */ +#define RADEON_COMPOSITE_SHADOW_ID 0x1a0c +#define RADEON_CONFIG_APER_0_BASE 0x0100 +#define RADEON_CONFIG_APER_1_BASE 0x0104 +#define RADEON_CONFIG_APER_SIZE 0x0108 +#define RADEON_CONFIG_BONDS 0x00e8 +#define RADEON_CONFIG_CNTL 0x00e0 +# define RADEON_CFG_VGA_RAM_EN (1 << 8) +# define RADEON_CFG_VGA_IO_DIS (1 << 9) +# define RADEON_CFG_ATI_REV_A11 (0 << 16) +# define RADEON_CFG_ATI_REV_A12 (1 << 16) +# define RADEON_CFG_ATI_REV_A13 (2 << 16) +# define RADEON_CFG_ATI_REV_ID_MASK (0xf << 16) +#define RADEON_CONFIG_MEMSIZE 0x00f8 +#define RADEON_CONFIG_MEMSIZE_EMBEDDED 0x0114 +#define RADEON_CONFIG_REG_1_BASE 0x010c +#define RADEON_CONFIG_REG_APER_SIZE 0x0110 +#define RADEON_CONFIG_XSTRAP 0x00e4 +#define RADEON_CONSTANT_COLOR_C 0x1d34 +# define RADEON_CONSTANT_COLOR_MASK 0x00ffffff +# define RADEON_CONSTANT_COLOR_ONE 0x00ffffff +# define RADEON_CONSTANT_COLOR_ZERO 0x00000000 +#define RADEON_CRC_CMDFIFO_ADDR 0x0740 +#define RADEON_CRC_CMDFIFO_DOUT 0x0744 +#define RADEON_GRPH_BUFFER_CNTL 0x02f0 +# define RADEON_GRPH_START_REQ_MASK (0x7f) +# define RADEON_GRPH_START_REQ_SHIFT 0 +# define RADEON_GRPH_STOP_REQ_MASK (0x7f<<8) +# define RADEON_GRPH_STOP_REQ_SHIFT 8 +# define RADEON_GRPH_CRITICAL_POINT_MASK (0x7f<<16) +# define RADEON_GRPH_CRITICAL_POINT_SHIFT 16 +# define RADEON_GRPH_CRITICAL_CNTL (1<<28) +# define RADEON_GRPH_BUFFER_SIZE (1<<29) +# define RADEON_GRPH_CRITICAL_AT_SOF (1<<30) +# define RADEON_GRPH_STOP_CNTL (1<<31) +#define RADEON_GRPH2_BUFFER_CNTL 0x03f0 +# define RADEON_GRPH2_START_REQ_MASK (0x7f) +# define RADEON_GRPH2_START_REQ_SHIFT 0 +# define RADEON_GRPH2_STOP_REQ_MASK (0x7f<<8) +# define RADEON_GRPH2_STOP_REQ_SHIFT 8 +# define RADEON_GRPH2_CRITICAL_POINT_MASK (0x7f<<16) +# define RADEON_GRPH2_CRITICAL_POINT_SHIFT 16 +# define RADEON_GRPH2_CRITICAL_CNTL (1<<28) +# define RADEON_GRPH2_BUFFER_SIZE (1<<29) +# define RADEON_GRPH2_CRITICAL_AT_SOF (1<<30) +# define RADEON_GRPH2_STOP_CNTL (1<<31) +#define RADEON_CRTC_CRNT_FRAME 0x0214 +#define RADEON_CRTC_EXT_CNTL 0x0054 +# define RADEON_CRTC_VGA_XOVERSCAN (1 << 0) +# define RADEON_VGA_ATI_LINEAR (1 << 3) +# define RADEON_XCRT_CNT_EN (1 << 6) +# define RADEON_CRTC_HSYNC_DIS (1 << 8) +# define RADEON_CRTC_VSYNC_DIS (1 << 9) +# define RADEON_CRTC_DISPLAY_DIS (1 << 10) +# define RADEON_CRTC_SYNC_TRISTAT (1 << 11) +# define RADEON_CRTC_CRT_ON (1 << 15) +#define RADEON_CRTC_EXT_CNTL_DPMS_BYTE 0x0055 +# define RADEON_CRTC_HSYNC_DIS_BYTE (1 << 0) +# define RADEON_CRTC_VSYNC_DIS_BYTE (1 << 1) +# define RADEON_CRTC_DISPLAY_DIS_BYTE (1 << 2) +#define RADEON_CRTC_GEN_CNTL 0x0050 +# define RADEON_CRTC_DBL_SCAN_EN (1 << 0) +# define RADEON_CRTC_INTERLACE_EN (1 << 1) +# define RADEON_CRTC_CSYNC_EN (1 << 4) +# define RADEON_CRTC_ICON_EN (1 << 15) +# define RADEON_CRTC_CUR_EN (1 << 16) +# define RADEON_CRTC_VSTAT_MODE_MASK (3 << 17) +# define RADEON_CRTC_CUR_MODE_MASK (7 << 20) +# define RADEON_CRTC_CUR_MODE_SHIFT 20 +# define RADEON_CRTC_CUR_MODE_MONO 0 +# define RADEON_CRTC_CUR_MODE_24BPP 2 +# define RADEON_CRTC_EXT_DISP_EN (1 << 24) +# define RADEON_CRTC_EN (1 << 25) +# define RADEON_CRTC_DISP_REQ_EN_B (1 << 26) +#define RADEON_CRTC2_GEN_CNTL 0x03f8 +# define RADEON_CRTC2_DBL_SCAN_EN (1 << 0) +# define RADEON_CRTC2_INTERLACE_EN (1 << 1) +# define RADEON_CRTC2_SYNC_TRISTAT (1 << 4) +# define RADEON_CRTC2_HSYNC_TRISTAT (1 << 5) +# define RADEON_CRTC2_VSYNC_TRISTAT (1 << 6) +# define RADEON_CRTC2_CRT2_ON (1 << 7) +# define RADEON_CRTC2_PIX_WIDTH_SHIFT 8 +# define RADEON_CRTC2_PIX_WIDTH_MASK (0xf << 8) +# define RADEON_CRTC2_ICON_EN (1 << 15) +# define RADEON_CRTC2_CUR_EN (1 << 16) +# define RADEON_CRTC2_CUR_MODE_MASK (7 << 20) +# define RADEON_CRTC2_DISP_DIS (1 << 23) +# define RADEON_CRTC2_EN (1 << 25) +# define RADEON_CRTC2_DISP_REQ_EN_B (1 << 26) +# define RADEON_CRTC2_CSYNC_EN (1 << 27) +# define RADEON_CRTC2_HSYNC_DIS (1 << 28) +# define RADEON_CRTC2_VSYNC_DIS (1 << 29) +#define RADEON_CRTC_MORE_CNTL 0x27c +# define RADEON_CRTC_AUTO_HORZ_CENTER_EN (1<<2) +# define RADEON_CRTC_AUTO_VERT_CENTER_EN (1<<3) +# define RADEON_CRTC_H_CUTOFF_ACTIVE_EN (1<<4) +# define RADEON_CRTC_V_CUTOFF_ACTIVE_EN (1<<5) +#define RADEON_CRTC_GUI_TRIG_VLINE 0x0218 +#define RADEON_CRTC_H_SYNC_STRT_WID 0x0204 +# define RADEON_CRTC_H_SYNC_STRT_PIX (0x07 << 0) +# define RADEON_CRTC_H_SYNC_STRT_CHAR (0x3ff << 3) +# define RADEON_CRTC_H_SYNC_STRT_CHAR_SHIFT 3 +# define RADEON_CRTC_H_SYNC_WID (0x3f << 16) +# define RADEON_CRTC_H_SYNC_WID_SHIFT 16 +# define RADEON_CRTC_H_SYNC_POL (1 << 23) +#define RADEON_CRTC2_H_SYNC_STRT_WID 0x0304 +# define RADEON_CRTC2_H_SYNC_STRT_PIX (0x07 << 0) +# define RADEON_CRTC2_H_SYNC_STRT_CHAR (0x3ff << 3) +# define RADEON_CRTC2_H_SYNC_STRT_CHAR_SHIFT 3 +# define RADEON_CRTC2_H_SYNC_WID (0x3f << 16) +# define RADEON_CRTC2_H_SYNC_WID_SHIFT 16 +# define RADEON_CRTC2_H_SYNC_POL (1 << 23) +#define RADEON_CRTC_H_TOTAL_DISP 0x0200 +# define RADEON_CRTC_H_TOTAL (0x03ff << 0) +# define RADEON_CRTC_H_TOTAL_SHIFT 0 +# define RADEON_CRTC_H_DISP (0x01ff << 16) +# define RADEON_CRTC_H_DISP_SHIFT 16 +#define RADEON_CRTC2_H_TOTAL_DISP 0x0300 +# define RADEON_CRTC2_H_TOTAL (0x03ff << 0) +# define RADEON_CRTC2_H_TOTAL_SHIFT 0 +# define RADEON_CRTC2_H_DISP (0x01ff << 16) +# define RADEON_CRTC2_H_DISP_SHIFT 16 + +#define RADEON_CRTC_OFFSET_RIGHT 0x0220 +#define RADEON_CRTC_OFFSET 0x0224 +# define RADEON_CRTC_OFFSET__GUI_TRIG_OFFSET (1<<30) +# define RADEON_CRTC_OFFSET__OFFSET_LOCK (1<<31) + +#define RADEON_CRTC2_OFFSET 0x0324 +# define RADEON_CRTC2_OFFSET__GUI_TRIG_OFFSET (1<<30) +# define RADEON_CRTC2_OFFSET__OFFSET_LOCK (1<<31) +#define RADEON_CRTC_OFFSET_CNTL 0x0228 +# define RADEON_CRTC_TILE_LINE_SHIFT 0 +# define RADEON_CRTC_TILE_LINE_RIGHT_SHIFT 4 +# define R300_CRTC_X_Y_MODE_EN_RIGHT (1 << 6) +# define R300_CRTC_MICRO_TILE_BUFFER_RIGHT_MASK (3 << 7) +# define R300_CRTC_MICRO_TILE_BUFFER_RIGHT_AUTO (0 << 7) +# define R300_CRTC_MICRO_TILE_BUFFER_RIGHT_SINGLE (1 << 7) +# define R300_CRTC_MICRO_TILE_BUFFER_RIGHT_DOUBLE (2 << 7) +# define R300_CRTC_MICRO_TILE_BUFFER_RIGHT_DIS (3 << 7) +# define R300_CRTC_X_Y_MODE_EN (1 << 9) +# define R300_CRTC_MICRO_TILE_BUFFER_MASK (3 << 10) +# define R300_CRTC_MICRO_TILE_BUFFER_AUTO (0 << 10) +# define R300_CRTC_MICRO_TILE_BUFFER_SINGLE (1 << 10) +# define R300_CRTC_MICRO_TILE_BUFFER_DOUBLE (2 << 10) +# define R300_CRTC_MICRO_TILE_BUFFER_DIS (3 << 10) +# define R300_CRTC_MICRO_TILE_EN_RIGHT (1 << 12) +# define R300_CRTC_MICRO_TILE_EN (1 << 13) +# define R300_CRTC_MACRO_TILE_EN_RIGHT (1 << 14) +# define R300_CRTC_MACRO_TILE_EN (1 << 15) +# define RADEON_CRTC_TILE_EN_RIGHT (1 << 14) +# define RADEON_CRTC_TILE_EN (1 << 15) +# define RADEON_CRTC_OFFSET_FLIP_CNTL (1 << 16) +# define RADEON_CRTC_STEREO_OFFSET_EN (1 << 17) +# define RADEON_CRTC_GUI_TRIG_OFFSET_LEFT_EN (1 << 28) +# define RADEON_CRTC_GUI_TRIG_OFFSET_RIGHT_EN (1 << 29) + +#define R300_CRTC_TILE_X0_Y0 0x0350 +#define R300_CRTC2_TILE_X0_Y0 0x0358 + +#define RADEON_CRTC2_OFFSET_CNTL 0x0328 +# define RADEON_CRTC2_OFFSET_FLIP_CNTL (1 << 16) +# define RADEON_CRTC2_TILE_EN (1 << 15) +#define RADEON_CRTC_PITCH 0x022c +# define RADEON_CRTC_PITCH__SHIFT 0 +# define RADEON_CRTC_PITCH__RIGHT_SHIFT 16 + +#define RADEON_CRTC2_PITCH 0x032c +#define RADEON_CRTC_STATUS 0x005c +# define RADEON_CRTC_VBLANK_SAVE (1 << 1) +# define RADEON_CRTC_VBLANK_SAVE_CLEAR (1 << 1) +#define RADEON_CRTC2_STATUS 0x03fc +# define RADEON_CRTC2_VBLANK_SAVE (1 << 1) +# define RADEON_CRTC2_VBLANK_SAVE_CLEAR (1 << 1) +#define RADEON_CRTC_V_SYNC_STRT_WID 0x020c +# define RADEON_CRTC_V_SYNC_STRT (0x7ff << 0) +# define RADEON_CRTC_V_SYNC_STRT_SHIFT 0 +# define RADEON_CRTC_V_SYNC_WID (0x1f << 16) +# define RADEON_CRTC_V_SYNC_WID_SHIFT 16 +# define RADEON_CRTC_V_SYNC_POL (1 << 23) +#define RADEON_CRTC2_V_SYNC_STRT_WID 0x030c +# define RADEON_CRTC2_V_SYNC_STRT (0x7ff << 0) +# define RADEON_CRTC2_V_SYNC_STRT_SHIFT 0 +# define RADEON_CRTC2_V_SYNC_WID (0x1f << 16) +# define RADEON_CRTC2_V_SYNC_WID_SHIFT 16 +# define RADEON_CRTC2_V_SYNC_POL (1 << 23) +#define RADEON_CRTC_V_TOTAL_DISP 0x0208 +# define RADEON_CRTC_V_TOTAL (0x07ff << 0) +# define RADEON_CRTC_V_TOTAL_SHIFT 0 +# define RADEON_CRTC_V_DISP (0x07ff << 16) +# define RADEON_CRTC_V_DISP_SHIFT 16 +#define RADEON_CRTC2_V_TOTAL_DISP 0x0308 +# define RADEON_CRTC2_V_TOTAL (0x07ff << 0) +# define RADEON_CRTC2_V_TOTAL_SHIFT 0 +# define RADEON_CRTC2_V_DISP (0x07ff << 16) +# define RADEON_CRTC2_V_DISP_SHIFT 16 +#define RADEON_CRTC_VLINE_CRNT_VLINE 0x0210 +# define RADEON_CRTC_CRNT_VLINE_MASK (0x7ff << 16) +#define RADEON_CRTC2_CRNT_FRAME 0x0314 +#define RADEON_CRTC2_GUI_TRIG_VLINE 0x0318 +#define RADEON_CRTC2_VLINE_CRNT_VLINE 0x0310 +#define RADEON_CRTC8_DATA 0x03d5 /* VGA, 0x3b5 */ +#define RADEON_CRTC8_IDX 0x03d4 /* VGA, 0x3b4 */ +#define RADEON_CUR_CLR0 0x026c +#define RADEON_CUR_CLR1 0x0270 +#define RADEON_CUR_HORZ_VERT_OFF 0x0268 +#define RADEON_CUR_HORZ_VERT_POSN 0x0264 +#define RADEON_CUR_OFFSET 0x0260 +# define RADEON_CUR_LOCK (1 << 31) +#define RADEON_CUR2_CLR0 0x036c +#define RADEON_CUR2_CLR1 0x0370 +#define RADEON_CUR2_HORZ_VERT_OFF 0x0368 +#define RADEON_CUR2_HORZ_VERT_POSN 0x0364 +#define RADEON_CUR2_OFFSET 0x0360 +# define RADEON_CUR2_LOCK (1 << 31) + +#define RADEON_DAC_CNTL 0x0058 +# define RADEON_DAC_RANGE_CNTL (3 << 0) +# define RADEON_DAC_RANGE_CNTL_PS2 (2 << 0) +# define RADEON_DAC_RANGE_CNTL_MASK 0x03 +# define RADEON_DAC_BLANKING (1 << 2) +# define RADEON_DAC_CMP_EN (1 << 3) +# define RADEON_DAC_CMP_OUTPUT (1 << 7) +# define RADEON_DAC_8BIT_EN (1 << 8) +# define RADEON_DAC_TVO_EN (1 << 10) +# define RADEON_DAC_VGA_ADR_EN (1 << 13) +# define RADEON_DAC_PDWN (1 << 15) +# define RADEON_DAC_MASK_ALL (0xff << 24) +#define RADEON_DAC_CNTL2 0x007c +# define RADEON_DAC2_TV_CLK_SEL (0 << 1) +# define RADEON_DAC2_DAC_CLK_SEL (1 << 0) +# define RADEON_DAC2_DAC2_CLK_SEL (1 << 1) +# define RADEON_DAC2_PALETTE_ACC_CTL (1 << 5) +# define RADEON_DAC2_CMP_EN (1 << 7) +# define RADEON_DAC2_CMP_OUT_R (1 << 8) +# define RADEON_DAC2_CMP_OUT_G (1 << 9) +# define RADEON_DAC2_CMP_OUT_B (1 << 10) +# define RADEON_DAC2_CMP_OUTPUT (1 << 11) +#define RADEON_DAC_EXT_CNTL 0x0280 +# define RADEON_DAC2_FORCE_BLANK_OFF_EN (1 << 0) +# define RADEON_DAC2_FORCE_DATA_EN (1 << 1) +# define RADEON_DAC_FORCE_BLANK_OFF_EN (1 << 4) +# define RADEON_DAC_FORCE_DATA_EN (1 << 5) +# define RADEON_DAC_FORCE_DATA_SEL_MASK (3 << 6) +# define RADEON_DAC_FORCE_DATA_SEL_R (0 << 6) +# define RADEON_DAC_FORCE_DATA_SEL_G (1 << 6) +# define RADEON_DAC_FORCE_DATA_SEL_B (2 << 6) +# define RADEON_DAC_FORCE_DATA_SEL_RGB (3 << 6) +# define RADEON_DAC_FORCE_DATA_MASK 0x0003ff00 +# define RADEON_DAC_FORCE_DATA_SHIFT 8 +#define RADEON_DAC_MACRO_CNTL 0x0d04 +# define RADEON_DAC_PDWN_R (1 << 16) +# define RADEON_DAC_PDWN_G (1 << 17) +# define RADEON_DAC_PDWN_B (1 << 18) +#define RADEON_DISP_PWR_MAN 0x0d08 +# define RADEON_DISP_PWR_MAN_D3_CRTC_EN (1 << 0) +# define RADEON_DISP_PWR_MAN_D3_CRTC2_EN (1 << 4) +# define RADEON_DISP_PWR_MAN_DPMS_ON (0 << 8) +# define RADEON_DISP_PWR_MAN_DPMS_STANDBY (1 << 8) +# define RADEON_DISP_PWR_MAN_DPMS_SUSPEND (2 << 8) +# define RADEON_DISP_PWR_MAN_DPMS_OFF (3 << 8) +# define RADEON_DISP_D3_RST (1 << 16) +# define RADEON_DISP_D3_REG_RST (1 << 17) +# define RADEON_DISP_D3_GRPH_RST (1 << 18) +# define RADEON_DISP_D3_SUBPIC_RST (1 << 19) +# define RADEON_DISP_D3_OV0_RST (1 << 20) +# define RADEON_DISP_D1D2_GRPH_RST (1 << 21) +# define RADEON_DISP_D1D2_SUBPIC_RST (1 << 22) +# define RADEON_DISP_D1D2_OV0_RST (1 << 23) +# define RADEON_DIG_TMDS_ENABLE_RST (1 << 24) +# define RADEON_TV_ENABLE_RST (1 << 25) +# define RADEON_AUTO_PWRUP_EN (1 << 26) +#define RADEON_TV_DAC_CNTL 0x088c +# define RADEON_TV_DAC_NBLANK (1 << 0) +# define RADEON_TV_DAC_NHOLD (1 << 1) +# define RADEON_TV_DAC_PEDESTAL (1 << 2) +# define RADEON_TV_MONITOR_DETECT_EN (1 << 4) +# define RADEON_TV_DAC_CMPOUT (1 << 5) +# define RADEON_TV_DAC_STD_MASK (3 << 8) +# define RADEON_TV_DAC_STD_PAL (0 << 8) +# define RADEON_TV_DAC_STD_NTSC (1 << 8) +# define RADEON_TV_DAC_STD_PS2 (2 << 8) +# define RADEON_TV_DAC_STD_RS343 (3 << 8) +# define RADEON_TV_DAC_BGSLEEP (1 << 6) +# define RADEON_TV_DAC_BGADJ_MASK (0xf << 16) +# define RADEON_TV_DAC_BGADJ_SHIFT 16 +# define RADEON_TV_DAC_DACADJ_MASK (0xf << 20) +# define RADEON_TV_DAC_DACADJ_SHIFT 20 +# define RADEON_TV_DAC_RDACPD (1 << 24) +# define RADEON_TV_DAC_GDACPD (1 << 25) +# define RADEON_TV_DAC_BDACPD (1 << 26) +# define RADEON_TV_DAC_RDACDET (1 << 29) +# define RADEON_TV_DAC_GDACDET (1 << 30) +# define RADEON_TV_DAC_BDACDET (1 << 31) +# define R420_TV_DAC_DACADJ_MASK (0x1f << 20) +# define R420_TV_DAC_RDACPD (1 << 25) +# define R420_TV_DAC_GDACPD (1 << 26) +# define R420_TV_DAC_BDACPD (1 << 27) +# define R420_TV_DAC_TVENABLE (1 << 28) +#define RADEON_DISP_HW_DEBUG 0x0d14 +# define RADEON_CRT2_DISP1_SEL (1 << 5) +#define RADEON_DISP_OUTPUT_CNTL 0x0d64 +# define RADEON_DISP_DAC_SOURCE_MASK 0x03 +# define RADEON_DISP_DAC2_SOURCE_MASK 0x0c +# define RADEON_DISP_DAC_SOURCE_CRTC2 0x01 +# define RADEON_DISP_DAC_SOURCE_RMX 0x02 +# define RADEON_DISP_DAC_SOURCE_LTU 0x03 +# define RADEON_DISP_DAC2_SOURCE_CRTC2 0x04 +# define RADEON_DISP_TVDAC_SOURCE_MASK (0x03 << 2) +# define RADEON_DISP_TVDAC_SOURCE_CRTC 0x0 +# define RADEON_DISP_TVDAC_SOURCE_CRTC2 (0x01 << 2) +# define RADEON_DISP_TVDAC_SOURCE_RMX (0x02 << 2) +# define RADEON_DISP_TVDAC_SOURCE_LTU (0x03 << 2) +# define RADEON_DISP_TRANS_MATRIX_MASK (0x03 << 4) +# define RADEON_DISP_TRANS_MATRIX_ALPHA_MSB (0x00 << 4) +# define RADEON_DISP_TRANS_MATRIX_GRAPHICS (0x01 << 4) +# define RADEON_DISP_TRANS_MATRIX_VIDEO (0x02 << 4) +# define RADEON_DISP_TV_SOURCE_CRTC (1 << 16) /* crtc1 or crtc2 */ +# define RADEON_DISP_TV_SOURCE_LTU (0 << 16) /* linear transform unit */ +#define RADEON_DISP_TV_OUT_CNTL 0x0d6c +# define RADEON_DISP_TV_PATH_SRC_CRTC2 (1 << 16) +# define RADEON_DISP_TV_PATH_SRC_CRTC1 (0 << 16) +#define RADEON_DAC_CRC_SIG 0x02cc +#define RADEON_DAC_DATA 0x03c9 /* VGA */ +#define RADEON_DAC_MASK 0x03c6 /* VGA */ +#define RADEON_DAC_R_INDEX 0x03c7 /* VGA */ +#define RADEON_DAC_W_INDEX 0x03c8 /* VGA */ +#define RADEON_DDA_CONFIG 0x02e0 +#define RADEON_DDA_ON_OFF 0x02e4 +#define RADEON_DEFAULT_OFFSET 0x16e0 +#define RADEON_DEFAULT_PITCH 0x16e4 +#define RADEON_DEFAULT_SC_BOTTOM_RIGHT 0x16e8 +# define RADEON_DEFAULT_SC_RIGHT_MAX (0x1fff << 0) +# define RADEON_DEFAULT_SC_BOTTOM_MAX (0x1fff << 16) +#define RADEON_DESTINATION_3D_CLR_CMP_VAL 0x1820 +#define RADEON_DESTINATION_3D_CLR_CMP_MSK 0x1824 +#define RADEON_DEVICE_ID 0x0f02 /* PCI */ +#define RADEON_DISP_MISC_CNTL 0x0d00 +# define RADEON_SOFT_RESET_GRPH_PP (1 << 0) +#define RADEON_DISP_MERGE_CNTL 0x0d60 +# define RADEON_DISP_ALPHA_MODE_MASK 0x03 +# define RADEON_DISP_ALPHA_MODE_KEY 0 +# define RADEON_DISP_ALPHA_MODE_PER_PIXEL 1 +# define RADEON_DISP_ALPHA_MODE_GLOBAL 2 +# define RADEON_DISP_RGB_OFFSET_EN (1 << 8) +# define RADEON_DISP_GRPH_ALPHA_MASK (0xff << 16) +# define RADEON_DISP_OV0_ALPHA_MASK (0xff << 24) +# define RADEON_DISP_LIN_TRANS_BYPASS (0x01 << 9) +#define RADEON_DISP2_MERGE_CNTL 0x0d68 +# define RADEON_DISP2_RGB_OFFSET_EN (1 << 8) +#define RADEON_DISP_LIN_TRANS_GRPH_A 0x0d80 +#define RADEON_DISP_LIN_TRANS_GRPH_B 0x0d84 +#define RADEON_DISP_LIN_TRANS_GRPH_C 0x0d88 +#define RADEON_DISP_LIN_TRANS_GRPH_D 0x0d8c +#define RADEON_DISP_LIN_TRANS_GRPH_E 0x0d90 +#define RADEON_DISP_LIN_TRANS_GRPH_F 0x0d98 +#define RADEON_DP_BRUSH_BKGD_CLR 0x1478 +#define RADEON_DP_BRUSH_FRGD_CLR 0x147c +#define RADEON_DP_CNTL 0x16c0 +# define RADEON_DST_X_LEFT_TO_RIGHT (1 << 0) +# define RADEON_DST_Y_TOP_TO_BOTTOM (1 << 1) +# define RADEON_DP_DST_TILE_LINEAR (0 << 3) +# define RADEON_DP_DST_TILE_MACRO (1 << 3) +# define RADEON_DP_DST_TILE_MICRO (2 << 3) +# define RADEON_DP_DST_TILE_BOTH (3 << 3) +#define RADEON_DP_CNTL_XDIR_YDIR_YMAJOR 0x16d0 +# define RADEON_DST_Y_MAJOR (1 << 2) +# define RADEON_DST_Y_DIR_TOP_TO_BOTTOM (1 << 15) +# define RADEON_DST_X_DIR_LEFT_TO_RIGHT (1 << 31) +#define RADEON_DP_DATATYPE 0x16c4 +# define RADEON_HOST_BIG_ENDIAN_EN (1 << 29) +#define RADEON_DP_GUI_MASTER_CNTL 0x146c +# define RADEON_GMC_SRC_PITCH_OFFSET_CNTL (1 << 0) +# define RADEON_GMC_DST_PITCH_OFFSET_CNTL (1 << 1) +# define RADEON_GMC_SRC_CLIPPING (1 << 2) +# define RADEON_GMC_DST_CLIPPING (1 << 3) +# define RADEON_GMC_BRUSH_DATATYPE_MASK (0x0f << 4) +# define RADEON_GMC_BRUSH_8X8_MONO_FG_BG (0 << 4) +# define RADEON_GMC_BRUSH_8X8_MONO_FG_LA (1 << 4) +# define RADEON_GMC_BRUSH_1X8_MONO_FG_BG (4 << 4) +# define RADEON_GMC_BRUSH_1X8_MONO_FG_LA (5 << 4) +# define RADEON_GMC_BRUSH_32x1_MONO_FG_BG (6 << 4) +# define RADEON_GMC_BRUSH_32x1_MONO_FG_LA (7 << 4) +# define RADEON_GMC_BRUSH_32x32_MONO_FG_BG (8 << 4) +# define RADEON_GMC_BRUSH_32x32_MONO_FG_LA (9 << 4) +# define RADEON_GMC_BRUSH_8x8_COLOR (10 << 4) +# define RADEON_GMC_BRUSH_1X8_COLOR (12 << 4) +# define RADEON_GMC_BRUSH_SOLID_COLOR (13 << 4) +# define RADEON_GMC_BRUSH_NONE (15 << 4) +# define RADEON_GMC_DST_8BPP_CI (2 << 8) +# define RADEON_GMC_DST_15BPP (3 << 8) +# define RADEON_GMC_DST_16BPP (4 << 8) +# define RADEON_GMC_DST_24BPP (5 << 8) +# define RADEON_GMC_DST_32BPP (6 << 8) +# define RADEON_GMC_DST_8BPP_RGB (7 << 8) +# define RADEON_GMC_DST_Y8 (8 << 8) +# define RADEON_GMC_DST_RGB8 (9 << 8) +# define RADEON_GMC_DST_VYUY (11 << 8) +# define RADEON_GMC_DST_YVYU (12 << 8) +# define RADEON_GMC_DST_AYUV444 (14 << 8) +# define RADEON_GMC_DST_ARGB4444 (15 << 8) +# define RADEON_GMC_DST_DATATYPE_MASK (0x0f << 8) +# define RADEON_GMC_DST_DATATYPE_SHIFT 8 +# define RADEON_GMC_SRC_DATATYPE_MASK (3 << 12) +# define RADEON_GMC_SRC_DATATYPE_MONO_FG_BG (0 << 12) +# define RADEON_GMC_SRC_DATATYPE_MONO_FG_LA (1 << 12) +# define RADEON_GMC_SRC_DATATYPE_COLOR (3 << 12) +# define RADEON_GMC_BYTE_PIX_ORDER (1 << 14) +# define RADEON_GMC_BYTE_MSB_TO_LSB (0 << 14) +# define RADEON_GMC_BYTE_LSB_TO_MSB (1 << 14) +# define RADEON_GMC_CONVERSION_TEMP (1 << 15) +# define RADEON_GMC_CONVERSION_TEMP_6500 (0 << 15) +# define RADEON_GMC_CONVERSION_TEMP_9300 (1 << 15) +# define RADEON_GMC_ROP3_MASK (0xff << 16) +# define RADEON_DP_SRC_SOURCE_MASK (7 << 24) +# define RADEON_DP_SRC_SOURCE_MEMORY (2 << 24) +# define RADEON_DP_SRC_SOURCE_HOST_DATA (3 << 24) +# define RADEON_GMC_3D_FCN_EN (1 << 27) +# define RADEON_GMC_CLR_CMP_CNTL_DIS (1 << 28) +# define RADEON_GMC_AUX_CLIP_DIS (1 << 29) +# define RADEON_GMC_WR_MSK_DIS (1 << 30) +# define RADEON_GMC_LD_BRUSH_Y_X (1 << 31) +# define RADEON_ROP3_ZERO 0x00000000 +# define RADEON_ROP3_DSa 0x00880000 +# define RADEON_ROP3_SDna 0x00440000 +# define RADEON_ROP3_S 0x00cc0000 +# define RADEON_ROP3_DSna 0x00220000 +# define RADEON_ROP3_D 0x00aa0000 +# define RADEON_ROP3_DSx 0x00660000 +# define RADEON_ROP3_DSo 0x00ee0000 +# define RADEON_ROP3_DSon 0x00110000 +# define RADEON_ROP3_DSxn 0x00990000 +# define RADEON_ROP3_Dn 0x00550000 +# define RADEON_ROP3_SDno 0x00dd0000 +# define RADEON_ROP3_Sn 0x00330000 +# define RADEON_ROP3_DSno 0x00bb0000 +# define RADEON_ROP3_DSan 0x00770000 +# define RADEON_ROP3_ONE 0x00ff0000 +# define RADEON_ROP3_DPa 0x00a00000 +# define RADEON_ROP3_PDna 0x00500000 +# define RADEON_ROP3_P 0x00f00000 +# define RADEON_ROP3_DPna 0x000a0000 +# define RADEON_ROP3_D 0x00aa0000 +# define RADEON_ROP3_DPx 0x005a0000 +# define RADEON_ROP3_DPo 0x00fa0000 +# define RADEON_ROP3_DPon 0x00050000 +# define RADEON_ROP3_PDxn 0x00a50000 +# define RADEON_ROP3_PDno 0x00f50000 +# define RADEON_ROP3_Pn 0x000f0000 +# define RADEON_ROP3_DPno 0x00af0000 +# define RADEON_ROP3_DPan 0x005f0000 +#define RADEON_DP_GUI_MASTER_CNTL_C 0x1c84 +#define RADEON_DP_MIX 0x16c8 +#define RADEON_DP_SRC_BKGD_CLR 0x15dc +#define RADEON_DP_SRC_FRGD_CLR 0x15d8 +#define RADEON_DP_WRITE_MASK 0x16cc +#define RADEON_DST_BRES_DEC 0x1630 +#define RADEON_DST_BRES_ERR 0x1628 +#define RADEON_DST_BRES_INC 0x162c +#define RADEON_DST_BRES_LNTH 0x1634 +#define RADEON_DST_BRES_LNTH_SUB 0x1638 +#define RADEON_DST_HEIGHT 0x1410 +#define RADEON_DST_HEIGHT_WIDTH 0x143c +#define RADEON_DST_HEIGHT_WIDTH_8 0x158c +#define RADEON_DST_HEIGHT_WIDTH_BW 0x15b4 +#define RADEON_DST_HEIGHT_Y 0x15a0 +#define RADEON_DST_LINE_START 0x1600 +#define RADEON_DST_LINE_END 0x1604 +#define RADEON_DST_LINE_PATCOUNT 0x1608 +# define RADEON_BRES_CNTL_SHIFT 8 +#define RADEON_DST_OFFSET 0x1404 +#define RADEON_DST_PITCH 0x1408 +#define RADEON_DST_PITCH_OFFSET 0x142c +#define RADEON_DST_PITCH_OFFSET_C 0x1c80 +# define RADEON_PITCH_SHIFT 21 +# define RADEON_DST_TILE_LINEAR (0 << 30) +# define RADEON_DST_TILE_MACRO (1 << 30) +# define RADEON_DST_TILE_MICRO (2 << 30) +# define RADEON_DST_TILE_BOTH (3 << 30) +#define RADEON_DST_WIDTH 0x140c +#define RADEON_DST_WIDTH_HEIGHT 0x1598 +#define RADEON_DST_WIDTH_X 0x1588 +#define RADEON_DST_WIDTH_X_INCY 0x159c +#define RADEON_DST_X 0x141c +#define RADEON_DST_X_SUB 0x15a4 +#define RADEON_DST_X_Y 0x1594 +#define RADEON_DST_Y 0x1420 +#define RADEON_DST_Y_SUB 0x15a8 +#define RADEON_DST_Y_X 0x1438 + +#define RADEON_FCP_CNTL 0x0910 +# define RADEON_FCP0_SRC_PCICLK 0 +# define RADEON_FCP0_SRC_PCLK 1 +# define RADEON_FCP0_SRC_PCLKb 2 +# define RADEON_FCP0_SRC_HREF 3 +# define RADEON_FCP0_SRC_GND 4 +# define RADEON_FCP0_SRC_HREFb 5 +#define RADEON_FLUSH_1 0x1704 +#define RADEON_FLUSH_2 0x1708 +#define RADEON_FLUSH_3 0x170c +#define RADEON_FLUSH_4 0x1710 +#define RADEON_FLUSH_5 0x1714 +#define RADEON_FLUSH_6 0x1718 +#define RADEON_FLUSH_7 0x171c +#define RADEON_FOG_3D_TABLE_START 0x1810 +#define RADEON_FOG_3D_TABLE_END 0x1814 +#define RADEON_FOG_3D_TABLE_DENSITY 0x181c +#define RADEON_FOG_TABLE_INDEX 0x1a14 +#define RADEON_FOG_TABLE_DATA 0x1a18 +#define RADEON_FP_CRTC_H_TOTAL_DISP 0x0250 +#define RADEON_FP_CRTC_V_TOTAL_DISP 0x0254 +# define RADEON_FP_CRTC_H_TOTAL_MASK 0x000003ff +# define RADEON_FP_CRTC_H_DISP_MASK 0x01ff0000 +# define RADEON_FP_CRTC_V_TOTAL_MASK 0x00000fff +# define RADEON_FP_CRTC_V_DISP_MASK 0x0fff0000 +# define RADEON_FP_H_SYNC_STRT_CHAR_MASK 0x00001ff8 +# define RADEON_FP_H_SYNC_WID_MASK 0x003f0000 +# define RADEON_FP_V_SYNC_STRT_MASK 0x00000fff +# define RADEON_FP_V_SYNC_WID_MASK 0x001f0000 +# define RADEON_FP_CRTC_H_TOTAL_SHIFT 0x00000000 +# define RADEON_FP_CRTC_H_DISP_SHIFT 0x00000010 +# define RADEON_FP_CRTC_V_TOTAL_SHIFT 0x00000000 +# define RADEON_FP_CRTC_V_DISP_SHIFT 0x00000010 +# define RADEON_FP_H_SYNC_STRT_CHAR_SHIFT 0x00000003 +# define RADEON_FP_H_SYNC_WID_SHIFT 0x00000010 +# define RADEON_FP_V_SYNC_STRT_SHIFT 0x00000000 +# define RADEON_FP_V_SYNC_WID_SHIFT 0x00000010 +#define RADEON_FP_GEN_CNTL 0x0284 +# define RADEON_FP_FPON (1 << 0) +# define RADEON_FP_BLANK_EN (1 << 1) +# define RADEON_FP_TMDS_EN (1 << 2) +# define RADEON_FP_PANEL_FORMAT (1 << 3) +# define RADEON_FP_EN_TMDS (1 << 7) +# define RADEON_FP_DETECT_SENSE (1 << 8) +# define RADEON_FP_DETECT_INT_POL (1 << 9) +# define R200_FP_SOURCE_SEL_MASK (3 << 10) +# define R200_FP_SOURCE_SEL_CRTC1 (0 << 10) +# define R200_FP_SOURCE_SEL_CRTC2 (1 << 10) +# define R200_FP_SOURCE_SEL_RMX (2 << 10) +# define R200_FP_SOURCE_SEL_TRANS (3 << 10) +# define RADEON_FP_SEL_CRTC1 (0 << 13) +# define RADEON_FP_SEL_CRTC2 (1 << 13) +# define R300_HPD_SEL(x) ((x) << 13) +# define RADEON_FP_CRTC_DONT_SHADOW_HPAR (1 << 15) +# define RADEON_FP_CRTC_DONT_SHADOW_VPAR (1 << 16) +# define RADEON_FP_CRTC_DONT_SHADOW_HEND (1 << 17) +# define RADEON_FP_CRTC_USE_SHADOW_VEND (1 << 18) +# define RADEON_FP_RMX_HVSYNC_CONTROL_EN (1 << 20) +# define RADEON_FP_DFP_SYNC_SEL (1 << 21) +# define RADEON_FP_CRTC_LOCK_8DOT (1 << 22) +# define RADEON_FP_CRT_SYNC_SEL (1 << 23) +# define RADEON_FP_USE_SHADOW_EN (1 << 24) +# define RADEON_FP_CRT_SYNC_ALT (1 << 26) +#define RADEON_FP2_GEN_CNTL 0x0288 +# define RADEON_FP2_BLANK_EN (1 << 1) +# define RADEON_FP2_ON (1 << 2) +# define RADEON_FP2_PANEL_FORMAT (1 << 3) +# define RADEON_FP2_DETECT_SENSE (1 << 8) +# define RADEON_FP2_DETECT_INT_POL (1 << 9) +# define R200_FP2_SOURCE_SEL_MASK (3 << 10) +# define R200_FP2_SOURCE_SEL_CRTC1 (0 << 10) +# define R200_FP2_SOURCE_SEL_CRTC2 (1 << 10) +# define R200_FP2_SOURCE_SEL_RMX (2 << 10) +# define R200_FP2_SOURCE_SEL_TRANS_UNIT (3 << 10) +# define RADEON_FP2_SRC_SEL_MASK (3 << 13) +# define RADEON_FP2_SRC_SEL_CRTC2 (1 << 13) +# define RADEON_FP2_FP_POL (1 << 16) +# define RADEON_FP2_LP_POL (1 << 17) +# define RADEON_FP2_SCK_POL (1 << 18) +# define RADEON_FP2_LCD_CNTL_MASK (7 << 19) +# define RADEON_FP2_PAD_FLOP_EN (1 << 22) +# define RADEON_FP2_CRC_EN (1 << 23) +# define RADEON_FP2_CRC_READ_EN (1 << 24) +# define RADEON_FP2_DVO_EN (1 << 25) +# define RADEON_FP2_DVO_RATE_SEL_SDR (1 << 26) +# define R200_FP2_DVO_RATE_SEL_SDR (1 << 27) +# define R300_FP2_DVO_CLOCK_MODE_SINGLE (1 << 28) +# define R300_FP2_DVO_DUAL_CHANNEL_EN (1 << 29) +#define RADEON_FP_H_SYNC_STRT_WID 0x02c4 +#define RADEON_FP_H2_SYNC_STRT_WID 0x03c4 +#define RADEON_FP_HORZ_STRETCH 0x028c +#define RADEON_FP_HORZ2_STRETCH 0x038c +# define RADEON_HORZ_STRETCH_RATIO_MASK 0xffff +# define RADEON_HORZ_STRETCH_RATIO_MAX 4096 +# define RADEON_HORZ_PANEL_SIZE (0x1ff << 16) +# define RADEON_HORZ_PANEL_SHIFT 16 +# define RADEON_HORZ_STRETCH_PIXREP (0 << 25) +# define RADEON_HORZ_STRETCH_BLEND (1 << 26) +# define RADEON_HORZ_STRETCH_ENABLE (1 << 25) +# define RADEON_HORZ_AUTO_RATIO (1 << 27) +# define RADEON_HORZ_FP_LOOP_STRETCH (0x7 << 28) +# define RADEON_HORZ_AUTO_RATIO_INC (1 << 31) +#define RADEON_FP_HORZ_VERT_ACTIVE 0x0278 +#define RADEON_FP_V_SYNC_STRT_WID 0x02c8 +#define RADEON_FP_VERT_STRETCH 0x0290 +#define RADEON_FP_V2_SYNC_STRT_WID 0x03c8 +#define RADEON_FP_VERT2_STRETCH 0x0390 +# define RADEON_VERT_PANEL_SIZE (0xfff << 12) +# define RADEON_VERT_PANEL_SHIFT 12 +# define RADEON_VERT_STRETCH_RATIO_MASK 0xfff +# define RADEON_VERT_STRETCH_RATIO_SHIFT 0 +# define RADEON_VERT_STRETCH_RATIO_MAX 4096 +# define RADEON_VERT_STRETCH_ENABLE (1 << 25) +# define RADEON_VERT_STRETCH_LINEREP (0 << 26) +# define RADEON_VERT_STRETCH_BLEND (1 << 26) +# define RADEON_VERT_AUTO_RATIO_EN (1 << 27) +# define RADEON_VERT_AUTO_RATIO_INC (1 << 31) +# define RADEON_VERT_STRETCH_RESERVED 0x71000000 +#define RS400_FP_2ND_GEN_CNTL 0x0384 +# define RS400_FP_2ND_ON (1 << 0) +# define RS400_FP_2ND_BLANK_EN (1 << 1) +# define RS400_TMDS_2ND_EN (1 << 2) +# define RS400_PANEL_FORMAT_2ND (1 << 3) +# define RS400_FP_2ND_EN_TMDS (1 << 7) +# define RS400_FP_2ND_DETECT_SENSE (1 << 8) +# define RS400_FP_2ND_SOURCE_SEL_MASK (3 << 10) +# define RS400_FP_2ND_SOURCE_SEL_CRTC1 (0 << 10) +# define RS400_FP_2ND_SOURCE_SEL_CRTC2 (1 << 10) +# define RS400_FP_2ND_SOURCE_SEL_RMX (2 << 10) +# define RS400_FP_2ND_DETECT_EN (1 << 12) +# define RS400_HPD_2ND_SEL (1 << 13) +#define RS400_FP2_2_GEN_CNTL 0x0388 +# define RS400_FP2_2_BLANK_EN (1 << 1) +# define RS400_FP2_2_ON (1 << 2) +# define RS400_FP2_2_PANEL_FORMAT (1 << 3) +# define RS400_FP2_2_DETECT_SENSE (1 << 8) +# define RS400_FP2_2_SOURCE_SEL_MASK (3 << 10) +# define RS400_FP2_2_SOURCE_SEL_CRTC1 (0 << 10) +# define RS400_FP2_2_SOURCE_SEL_CRTC2 (1 << 10) +# define RS400_FP2_2_SOURCE_SEL_RMX (2 << 10) +# define RS400_FP2_2_DVO2_EN (1 << 25) +#define RS400_TMDS2_CNTL 0x0394 +#define RS400_TMDS2_TRANSMITTER_CNTL 0x03a4 +# define RS400_TMDS2_PLLEN (1 << 0) +# define RS400_TMDS2_PLLRST (1 << 1) + +#define RADEON_GEN_INT_CNTL 0x0040 +# define RADEON_CRTC_VBLANK_MASK (1 << 0) +# define RADEON_FP_DETECT_MASK (1 << 4) +# define RADEON_CRTC2_VBLANK_MASK (1 << 9) +# define RADEON_FP2_DETECT_MASK (1 << 10) +# define RADEON_GUI_IDLE_MASK (1 << 19) +# define RADEON_SW_INT_ENABLE (1 << 25) +#define RADEON_GEN_INT_STATUS 0x0044 +# define AVIVO_DISPLAY_INT_STATUS (1 << 0) +# define RADEON_CRTC_VBLANK_STAT (1 << 0) +# define RADEON_CRTC_VBLANK_STAT_ACK (1 << 0) +# define RADEON_FP_DETECT_STAT (1 << 4) +# define RADEON_FP_DETECT_STAT_ACK (1 << 4) +# define RADEON_CRTC2_VBLANK_STAT (1 << 9) +# define RADEON_CRTC2_VBLANK_STAT_ACK (1 << 9) +# define RADEON_FP2_DETECT_STAT (1 << 10) +# define RADEON_FP2_DETECT_STAT_ACK (1 << 10) +# define RADEON_GUI_IDLE_STAT (1 << 19) +# define RADEON_GUI_IDLE_STAT_ACK (1 << 19) +# define RADEON_SW_INT_FIRE (1 << 26) +# define RADEON_SW_INT_TEST (1 << 25) +# define RADEON_SW_INT_TEST_ACK (1 << 25) +#define RADEON_GENENB 0x03c3 /* VGA */ +#define RADEON_GENFC_RD 0x03ca /* VGA */ +#define RADEON_GENFC_WT 0x03da /* VGA, 0x03ba */ +#define RADEON_GENMO_RD 0x03cc /* VGA */ +#define RADEON_GENMO_WT 0x03c2 /* VGA */ +#define RADEON_GENS0 0x03c2 /* VGA */ +#define RADEON_GENS1 0x03da /* VGA, 0x03ba */ +#define RADEON_GPIO_MONID 0x0068 /* DDC interface via I2C */ /* DDC3 */ +#define RADEON_GPIO_MONIDB 0x006c +#define RADEON_GPIO_CRT2_DDC 0x006c +#define RADEON_GPIO_DVI_DDC 0x0064 /* DDC2 */ +#define RADEON_GPIO_VGA_DDC 0x0060 /* DDC1 */ +# define RADEON_GPIO_A_0 (1 << 0) +# define RADEON_GPIO_A_1 (1 << 1) +# define RADEON_GPIO_Y_0 (1 << 8) +# define RADEON_GPIO_Y_1 (1 << 9) +# define RADEON_GPIO_Y_SHIFT_0 8 +# define RADEON_GPIO_Y_SHIFT_1 9 +# define RADEON_GPIO_EN_0 (1 << 16) +# define RADEON_GPIO_EN_1 (1 << 17) +# define RADEON_GPIO_MASK_0 (1 << 24) /*??*/ +# define RADEON_GPIO_MASK_1 (1 << 25) /*??*/ +#define RADEON_GRPH8_DATA 0x03cf /* VGA */ +#define RADEON_GRPH8_IDX 0x03ce /* VGA */ +#define RADEON_GUI_SCRATCH_REG0 0x15e0 +#define RADEON_GUI_SCRATCH_REG1 0x15e4 +#define RADEON_GUI_SCRATCH_REG2 0x15e8 +#define RADEON_GUI_SCRATCH_REG3 0x15ec +#define RADEON_GUI_SCRATCH_REG4 0x15f0 +#define RADEON_GUI_SCRATCH_REG5 0x15f4 + +#define RADEON_HEADER 0x0f0e /* PCI */ +#define RADEON_HOST_DATA0 0x17c0 +#define RADEON_HOST_DATA1 0x17c4 +#define RADEON_HOST_DATA2 0x17c8 +#define RADEON_HOST_DATA3 0x17cc +#define RADEON_HOST_DATA4 0x17d0 +#define RADEON_HOST_DATA5 0x17d4 +#define RADEON_HOST_DATA6 0x17d8 +#define RADEON_HOST_DATA7 0x17dc +#define RADEON_HOST_DATA_LAST 0x17e0 +#define RADEON_HOST_PATH_CNTL 0x0130 +# define RADEON_HP_LIN_RD_CACHE_DIS (1 << 24) +# define RADEON_HDP_READ_BUFFER_INVALIDATE (1 << 27) +# define RADEON_HDP_SOFT_RESET (1 << 26) +# define RADEON_HDP_APER_CNTL (1 << 23) +#define RADEON_HTOTAL_CNTL 0x0009 /* PLL */ +# define RADEON_HTOT_CNTL_VGA_EN (1 << 28) +#define RADEON_HTOTAL2_CNTL 0x002e /* PLL */ + + /* Multimedia I2C bus */ +#define RADEON_I2C_CNTL_0 0x0090 +# define RADEON_I2C_DONE (1 << 0) +# define RADEON_I2C_NACK (1 << 1) +# define RADEON_I2C_HALT (1 << 2) +# define RADEON_I2C_SOFT_RST (1 << 5) +# define RADEON_I2C_DRIVE_EN (1 << 6) +# define RADEON_I2C_DRIVE_SEL (1 << 7) +# define RADEON_I2C_START (1 << 8) +# define RADEON_I2C_STOP (1 << 9) +# define RADEON_I2C_RECEIVE (1 << 10) +# define RADEON_I2C_ABORT (1 << 11) +# define RADEON_I2C_GO (1 << 12) +# define RADEON_I2C_PRESCALE_SHIFT 16 +#define RADEON_I2C_CNTL_1 0x0094 +# define RADEON_I2C_DATA_COUNT_SHIFT 0 +# define RADEON_I2C_ADDR_COUNT_SHIFT 4 +# define RADEON_I2C_INTRA_BYTE_DELAY_SHIFT 8 +# define RADEON_I2C_SEL (1 << 16) +# define RADEON_I2C_EN (1 << 17) +# define RADEON_I2C_TIME_LIMIT_SHIFT 24 +#define RADEON_I2C_DATA 0x0098 + +#define RADEON_DVI_I2C_CNTL_0 0x02e0 +# define R200_DVI_I2C_PIN_SEL(x) ((x) << 3) +# define R200_SEL_DDC1 0 /* depends on asic */ +# define R200_SEL_DDC2 1 /* depends on asic */ +# define R200_SEL_DDC3 2 /* depends on asic */ +# define RADEON_SW_WANTS_TO_USE_DVI_I2C (1 << 13) +# define RADEON_SW_CAN_USE_DVI_I2C (1 << 13) +# define RADEON_SW_DONE_USING_DVI_I2C (1 << 14) +# define RADEON_HW_NEEDS_DVI_I2C (1 << 14) +# define RADEON_ABORT_HW_DVI_I2C (1 << 15) +# define RADEON_HW_USING_DVI_I2C (1 << 15) +#define RADEON_DVI_I2C_CNTL_1 0x02e4 +#define RADEON_DVI_I2C_DATA 0x02e8 + +#define RADEON_INTERRUPT_LINE 0x0f3c /* PCI */ +#define RADEON_INTERRUPT_PIN 0x0f3d /* PCI */ +#define RADEON_IO_BASE 0x0f14 /* PCI */ + +#define RADEON_LATENCY 0x0f0d /* PCI */ +#define RADEON_LEAD_BRES_DEC 0x1608 +#define RADEON_LEAD_BRES_LNTH 0x161c +#define RADEON_LEAD_BRES_LNTH_SUB 0x1624 +#define RADEON_LVDS_GEN_CNTL 0x02d0 +# define RADEON_LVDS_ON (1 << 0) +# define RADEON_LVDS_DISPLAY_DIS (1 << 1) +# define RADEON_LVDS_PANEL_TYPE (1 << 2) +# define RADEON_LVDS_PANEL_FORMAT (1 << 3) +# define RADEON_LVDS_NO_FM (0 << 4) +# define RADEON_LVDS_2_GREY (1 << 4) +# define RADEON_LVDS_4_GREY (2 << 4) +# define RADEON_LVDS_RST_FM (1 << 6) +# define RADEON_LVDS_EN (1 << 7) +# define RADEON_LVDS_BL_MOD_LEVEL_SHIFT 8 +# define RADEON_LVDS_BL_MOD_LEVEL_MASK (0xff << 8) +# define RADEON_LVDS_BL_MOD_EN (1 << 16) +# define RADEON_LVDS_BL_CLK_SEL (1 << 17) +# define RADEON_LVDS_DIGON (1 << 18) +# define RADEON_LVDS_BLON (1 << 19) +# define RADEON_LVDS_FP_POL_LOW (1 << 20) +# define RADEON_LVDS_LP_POL_LOW (1 << 21) +# define RADEON_LVDS_DTM_POL_LOW (1 << 22) +# define RADEON_LVDS_SEL_CRTC2 (1 << 23) +# define RADEON_LVDS_FPDI_EN (1 << 27) +# define RADEON_LVDS_HSYNC_DELAY_SHIFT 28 +#define RADEON_LVDS_PLL_CNTL 0x02d4 +# define RADEON_HSYNC_DELAY_SHIFT 28 +# define RADEON_HSYNC_DELAY_MASK (0xf << 28) +# define RADEON_LVDS_PLL_EN (1 << 16) +# define RADEON_LVDS_PLL_RESET (1 << 17) +# define R300_LVDS_SRC_SEL_MASK (3 << 18) +# define R300_LVDS_SRC_SEL_CRTC1 (0 << 18) +# define R300_LVDS_SRC_SEL_CRTC2 (1 << 18) +# define R300_LVDS_SRC_SEL_RMX (2 << 18) +#define RADEON_LVDS_SS_GEN_CNTL 0x02ec +# define RADEON_LVDS_PWRSEQ_DELAY1_SHIFT 16 +# define RADEON_LVDS_PWRSEQ_DELAY2_SHIFT 20 + +#define RADEON_MAX_LATENCY 0x0f3f /* PCI */ +#define RADEON_DISPLAY_BASE_ADDR 0x23c +#define RADEON_DISPLAY2_BASE_ADDR 0x33c +#define RADEON_OV0_BASE_ADDR 0x43c +#define RADEON_NB_TOM 0x15c +#define R300_MC_INIT_MISC_LAT_TIMER 0x180 +# define R300_MC_DISP0R_INIT_LAT_SHIFT 8 +# define R300_MC_DISP0R_INIT_LAT_MASK 0xf +# define R300_MC_DISP1R_INIT_LAT_SHIFT 12 +# define R300_MC_DISP1R_INIT_LAT_MASK 0xf +#define RADEON_MCLK_CNTL 0x0012 /* PLL */ +# define RADEON_MCLKA_SRC_SEL_MASK 0x7 +# define RADEON_FORCEON_MCLKA (1 << 16) +# define RADEON_FORCEON_MCLKB (1 << 17) +# define RADEON_FORCEON_YCLKA (1 << 18) +# define RADEON_FORCEON_YCLKB (1 << 19) +# define RADEON_FORCEON_MC (1 << 20) +# define RADEON_FORCEON_AIC (1 << 21) +# define R300_DISABLE_MC_MCLKA (1 << 21) +# define R300_DISABLE_MC_MCLKB (1 << 21) +#define RADEON_MCLK_MISC 0x001f /* PLL */ +# define RADEON_MC_MCLK_MAX_DYN_STOP_LAT (1 << 12) +# define RADEON_IO_MCLK_MAX_DYN_STOP_LAT (1 << 13) +# define RADEON_MC_MCLK_DYN_ENABLE (1 << 14) +# define RADEON_IO_MCLK_DYN_ENABLE (1 << 15) + +#define RADEON_GPIOPAD_MASK 0x0198 +#define RADEON_GPIOPAD_A 0x019c +#define RADEON_GPIOPAD_EN 0x01a0 +#define RADEON_GPIOPAD_Y 0x01a4 +#define RADEON_MDGPIO_MASK 0x01a8 +#define RADEON_MDGPIO_A 0x01ac +#define RADEON_MDGPIO_EN 0x01b0 +#define RADEON_MDGPIO_Y 0x01b4 + +#define RADEON_MEM_ADDR_CONFIG 0x0148 +#define RADEON_MEM_BASE 0x0f10 /* PCI */ +#define RADEON_MEM_CNTL 0x0140 +# define RADEON_MEM_NUM_CHANNELS_MASK 0x01 +# define RADEON_MEM_USE_B_CH_ONLY (1 << 1) +# define RV100_HALF_MODE (1 << 3) +# define R300_MEM_NUM_CHANNELS_MASK 0x03 +# define R300_MEM_USE_CD_CH_ONLY (1 << 2) +#define RADEON_MEM_TIMING_CNTL 0x0144 /* EXT_MEM_CNTL */ +#define RADEON_MEM_INIT_LAT_TIMER 0x0154 +#define RADEON_MEM_INTF_CNTL 0x014c +#define RADEON_MEM_SDRAM_MODE_REG 0x0158 +# define RADEON_SDRAM_MODE_MASK 0xffff0000 +# define RADEON_B3MEM_RESET_MASK 0x6fffffff +# define RADEON_MEM_CFG_TYPE_DDR (1 << 30) +#define RADEON_MEM_STR_CNTL 0x0150 +# define RADEON_MEM_PWRUP_COMPL_A (1 << 0) +# define RADEON_MEM_PWRUP_COMPL_B (1 << 1) +# define R300_MEM_PWRUP_COMPL_C (1 << 2) +# define R300_MEM_PWRUP_COMPL_D (1 << 3) +# define RADEON_MEM_PWRUP_COMPLETE 0x03 +# define R300_MEM_PWRUP_COMPLETE 0x0f +#define RADEON_MC_STATUS 0x0150 +# define RADEON_MC_IDLE (1 << 2) +# define R300_MC_IDLE (1 << 4) +#define RADEON_MEM_VGA_RP_SEL 0x003c +#define RADEON_MEM_VGA_WP_SEL 0x0038 +#define RADEON_MIN_GRANT 0x0f3e /* PCI */ +#define RADEON_MM_DATA 0x0004 +#define RADEON_MM_INDEX 0x0000 +# define RADEON_MM_APER (1 << 31) +#define RADEON_MPLL_CNTL 0x000e /* PLL */ +#define RADEON_MPP_TB_CONFIG 0x01c0 /* ? */ +#define RADEON_MPP_GP_CONFIG 0x01c8 /* ? */ +#define RADEON_SEPROM_CNTL1 0x01c0 +# define RADEON_SCK_PRESCALE_SHIFT 24 +# define RADEON_SCK_PRESCALE_MASK (0xff << 24) +#define R300_MC_IND_INDEX 0x01f8 +# define R300_MC_IND_ADDR_MASK 0x3f +# define R300_MC_IND_WR_EN (1 << 8) +#define R300_MC_IND_DATA 0x01fc +#define R300_MC_READ_CNTL_AB 0x017c +# define R300_MEM_RBS_POSITION_A_MASK 0x03 +#define R300_MC_READ_CNTL_CD_mcind 0x24 +# define R300_MEM_RBS_POSITION_C_MASK 0x03 + +#define RADEON_N_VIF_COUNT 0x0248 + +#define RADEON_OV0_AUTO_FLIP_CNTL 0x0470 +# define RADEON_OV0_AUTO_FLIP_CNTL_SOFT_BUF_NUM 0x00000007 +# define RADEON_OV0_AUTO_FLIP_CNTL_SOFT_REPEAT_FIELD 0x00000008 +# define RADEON_OV0_AUTO_FLIP_CNTL_SOFT_BUF_ODD 0x00000010 +# define RADEON_OV0_AUTO_FLIP_CNTL_IGNORE_REPEAT_FIELD 0x00000020 +# define RADEON_OV0_AUTO_FLIP_CNTL_SOFT_EOF_TOGGLE 0x00000040 +# define RADEON_OV0_AUTO_FLIP_CNTL_VID_PORT_SELECT 0x00000300 +# define RADEON_OV0_AUTO_FLIP_CNTL_P1_FIRST_LINE_EVEN 0x00010000 +# define RADEON_OV0_AUTO_FLIP_CNTL_SHIFT_EVEN_DOWN 0x00040000 +# define RADEON_OV0_AUTO_FLIP_CNTL_SHIFT_ODD_DOWN 0x00080000 +# define RADEON_OV0_AUTO_FLIP_CNTL_FIELD_POL_SOURCE 0x00800000 + +#define RADEON_OV0_COLOUR_CNTL 0x04E0 +#define RADEON_OV0_DEINTERLACE_PATTERN 0x0474 +#define RADEON_OV0_EXCLUSIVE_HORZ 0x0408 +# define RADEON_EXCL_HORZ_START_MASK 0x000000ff +# define RADEON_EXCL_HORZ_END_MASK 0x0000ff00 +# define RADEON_EXCL_HORZ_BACK_PORCH_MASK 0x00ff0000 +# define RADEON_EXCL_HORZ_EXCLUSIVE_EN 0x80000000 +#define RADEON_OV0_EXCLUSIVE_VERT 0x040C +# define RADEON_EXCL_VERT_START_MASK 0x000003ff +# define RADEON_EXCL_VERT_END_MASK 0x03ff0000 +#define RADEON_OV0_FILTER_CNTL 0x04A0 +# define RADEON_FILTER_PROGRAMMABLE_COEF 0x0 +# define RADEON_FILTER_HC_COEF_HORZ_Y 0x1 +# define RADEON_FILTER_HC_COEF_HORZ_UV 0x2 +# define RADEON_FILTER_HC_COEF_VERT_Y 0x4 +# define RADEON_FILTER_HC_COEF_VERT_UV 0x8 +# define RADEON_FILTER_HARDCODED_COEF 0xf +# define RADEON_FILTER_COEF_MASK 0xf + +#define RADEON_OV0_FOUR_TAP_COEF_0 0x04B0 +#define RADEON_OV0_FOUR_TAP_COEF_1 0x04B4 +#define RADEON_OV0_FOUR_TAP_COEF_2 0x04B8 +#define RADEON_OV0_FOUR_TAP_COEF_3 0x04BC +#define RADEON_OV0_FOUR_TAP_COEF_4 0x04C0 +#define RADEON_OV0_FLAG_CNTL 0x04DC +#define RADEON_OV0_GAMMA_000_00F 0x0d40 +#define RADEON_OV0_GAMMA_010_01F 0x0d44 +#define RADEON_OV0_GAMMA_020_03F 0x0d48 +#define RADEON_OV0_GAMMA_040_07F 0x0d4c +#define RADEON_OV0_GAMMA_080_0BF 0x0e00 +#define RADEON_OV0_GAMMA_0C0_0FF 0x0e04 +#define RADEON_OV0_GAMMA_100_13F 0x0e08 +#define RADEON_OV0_GAMMA_140_17F 0x0e0c +#define RADEON_OV0_GAMMA_180_1BF 0x0e10 +#define RADEON_OV0_GAMMA_1C0_1FF 0x0e14 +#define RADEON_OV0_GAMMA_200_23F 0x0e18 +#define RADEON_OV0_GAMMA_240_27F 0x0e1c +#define RADEON_OV0_GAMMA_280_2BF 0x0e20 +#define RADEON_OV0_GAMMA_2C0_2FF 0x0e24 +#define RADEON_OV0_GAMMA_300_33F 0x0e28 +#define RADEON_OV0_GAMMA_340_37F 0x0e2c +#define RADEON_OV0_GAMMA_380_3BF 0x0d50 +#define RADEON_OV0_GAMMA_3C0_3FF 0x0d54 +#define RADEON_OV0_GRAPHICS_KEY_CLR_LOW 0x04EC +#define RADEON_OV0_GRAPHICS_KEY_CLR_HIGH 0x04F0 +#define RADEON_OV0_H_INC 0x0480 +#define RADEON_OV0_KEY_CNTL 0x04F4 +# define RADEON_VIDEO_KEY_FN_MASK 0x00000003L +# define RADEON_VIDEO_KEY_FN_FALSE 0x00000000L +# define RADEON_VIDEO_KEY_FN_TRUE 0x00000001L +# define RADEON_VIDEO_KEY_FN_EQ 0x00000002L +# define RADEON_VIDEO_KEY_FN_NE 0x00000003L +# define RADEON_GRAPHIC_KEY_FN_MASK 0x00000030L +# define RADEON_GRAPHIC_KEY_FN_FALSE 0x00000000L +# define RADEON_GRAPHIC_KEY_FN_TRUE 0x00000010L +# define RADEON_GRAPHIC_KEY_FN_EQ 0x00000020L +# define RADEON_GRAPHIC_KEY_FN_NE 0x00000030L +# define RADEON_CMP_MIX_MASK 0x00000100L +# define RADEON_CMP_MIX_OR 0x00000000L +# define RADEON_CMP_MIX_AND 0x00000100L +#define RADEON_OV0_LIN_TRANS_A 0x0d20 +#define RADEON_OV0_LIN_TRANS_B 0x0d24 +#define RADEON_OV0_LIN_TRANS_C 0x0d28 +#define RADEON_OV0_LIN_TRANS_D 0x0d2c +#define RADEON_OV0_LIN_TRANS_E 0x0d30 +#define RADEON_OV0_LIN_TRANS_F 0x0d34 +#define RADEON_OV0_P1_BLANK_LINES_AT_TOP 0x0430 +# define RADEON_P1_BLNK_LN_AT_TOP_M1_MASK 0x00000fffL +# define RADEON_P1_ACTIVE_LINES_M1 0x0fff0000L +#define RADEON_OV0_P1_H_ACCUM_INIT 0x0488 +#define RADEON_OV0_P1_V_ACCUM_INIT 0x0428 +# define RADEON_OV0_P1_MAX_LN_IN_PER_LN_OUT 0x00000003L +# define RADEON_OV0_P1_V_ACCUM_INIT_MASK 0x01ff8000L +#define RADEON_OV0_P1_X_START_END 0x0494 +#define RADEON_OV0_P2_X_START_END 0x0498 +#define RADEON_OV0_P23_BLANK_LINES_AT_TOP 0x0434 +# define RADEON_P23_BLNK_LN_AT_TOP_M1_MASK 0x000007ffL +# define RADEON_P23_ACTIVE_LINES_M1 0x07ff0000L +#define RADEON_OV0_P23_H_ACCUM_INIT 0x048C +#define RADEON_OV0_P23_V_ACCUM_INIT 0x042C +#define RADEON_OV0_P3_X_START_END 0x049C +#define RADEON_OV0_REG_LOAD_CNTL 0x0410 +# define RADEON_REG_LD_CTL_LOCK 0x00000001L +# define RADEON_REG_LD_CTL_VBLANK_DURING_LOCK 0x00000002L +# define RADEON_REG_LD_CTL_STALL_GUI_UNTIL_FLIP 0x00000004L +# define RADEON_REG_LD_CTL_LOCK_READBACK 0x00000008L +# define RADEON_REG_LD_CTL_FLIP_READBACK 0x00000010L +#define RADEON_OV0_SCALE_CNTL 0x0420 +# define RADEON_SCALER_HORZ_PICK_NEAREST 0x00000004L +# define RADEON_SCALER_VERT_PICK_NEAREST 0x00000008L +# define RADEON_SCALER_SIGNED_UV 0x00000010L +# define RADEON_SCALER_GAMMA_SEL_MASK 0x00000060L +# define RADEON_SCALER_GAMMA_SEL_BRIGHT 0x00000000L +# define RADEON_SCALER_GAMMA_SEL_G22 0x00000020L +# define RADEON_SCALER_GAMMA_SEL_G18 0x00000040L +# define RADEON_SCALER_GAMMA_SEL_G14 0x00000060L +# define RADEON_SCALER_COMCORE_SHIFT_UP_ONE 0x00000080L +# define RADEON_SCALER_SURFAC_FORMAT 0x00000f00L +# define RADEON_SCALER_SOURCE_15BPP 0x00000300L +# define RADEON_SCALER_SOURCE_16BPP 0x00000400L +# define RADEON_SCALER_SOURCE_32BPP 0x00000600L +# define RADEON_SCALER_SOURCE_YUV9 0x00000900L +# define RADEON_SCALER_SOURCE_YUV12 0x00000A00L +# define RADEON_SCALER_SOURCE_VYUY422 0x00000B00L +# define RADEON_SCALER_SOURCE_YVYU422 0x00000C00L +# define RADEON_SCALER_ADAPTIVE_DEINT 0x00001000L +# define RADEON_SCALER_TEMPORAL_DEINT 0x00002000L +# define RADEON_SCALER_CRTC_SEL 0x00004000L +# define RADEON_SCALER_SMART_SWITCH 0x00008000L +# define RADEON_SCALER_BURST_PER_PLANE 0x007F0000L +# define RADEON_SCALER_DOUBLE_BUFFER 0x01000000L +# define RADEON_SCALER_DIS_LIMIT 0x08000000L +# define RADEON_SCALER_LIN_TRANS_BYPASS 0x10000000L +# define RADEON_SCALER_INT_EMU 0x20000000L +# define RADEON_SCALER_ENABLE 0x40000000L +# define RADEON_SCALER_SOFT_RESET 0x80000000L +#define RADEON_OV0_STEP_BY 0x0484 +#define RADEON_OV0_TEST 0x04F8 +#define RADEON_OV0_V_INC 0x0424 +#define RADEON_OV0_VID_BUF_PITCH0_VALUE 0x0460 +#define RADEON_OV0_VID_BUF_PITCH1_VALUE 0x0464 +#define RADEON_OV0_VID_BUF0_BASE_ADRS 0x0440 +# define RADEON_VIF_BUF0_PITCH_SEL 0x00000001L +# define RADEON_VIF_BUF0_TILE_ADRS 0x00000002L +# define RADEON_VIF_BUF0_BASE_ADRS_MASK 0x03fffff0L +# define RADEON_VIF_BUF0_1ST_LINE_LSBS_MASK 0x48000000L +#define RADEON_OV0_VID_BUF1_BASE_ADRS 0x0444 +# define RADEON_VIF_BUF1_PITCH_SEL 0x00000001L +# define RADEON_VIF_BUF1_TILE_ADRS 0x00000002L +# define RADEON_VIF_BUF1_BASE_ADRS_MASK 0x03fffff0L +# define RADEON_VIF_BUF1_1ST_LINE_LSBS_MASK 0x48000000L +#define RADEON_OV0_VID_BUF2_BASE_ADRS 0x0448 +# define RADEON_VIF_BUF2_PITCH_SEL 0x00000001L +# define RADEON_VIF_BUF2_TILE_ADRS 0x00000002L +# define RADEON_VIF_BUF2_BASE_ADRS_MASK 0x03fffff0L +# define RADEON_VIF_BUF2_1ST_LINE_LSBS_MASK 0x48000000L +#define RADEON_OV0_VID_BUF3_BASE_ADRS 0x044C +#define RADEON_OV0_VID_BUF4_BASE_ADRS 0x0450 +#define RADEON_OV0_VID_BUF5_BASE_ADRS 0x0454 +#define RADEON_OV0_VIDEO_KEY_CLR_HIGH 0x04E8 +#define RADEON_OV0_VIDEO_KEY_CLR_LOW 0x04E4 +#define RADEON_OV0_Y_X_START 0x0400 +#define RADEON_OV0_Y_X_END 0x0404 +#define RADEON_OV1_Y_X_START 0x0600 +#define RADEON_OV1_Y_X_END 0x0604 +#define RADEON_OVR_CLR 0x0230 +#define RADEON_OVR_WID_LEFT_RIGHT 0x0234 +#define RADEON_OVR_WID_TOP_BOTTOM 0x0238 +#define RADEON_OVR2_CLR 0x0330 +#define RADEON_OVR2_WID_LEFT_RIGHT 0x0334 +#define RADEON_OVR2_WID_TOP_BOTTOM 0x0338 + +/* first capture unit */ + +#define RADEON_CAP0_BUF0_OFFSET 0x0920 +#define RADEON_CAP0_BUF1_OFFSET 0x0924 +#define RADEON_CAP0_BUF0_EVEN_OFFSET 0x0928 +#define RADEON_CAP0_BUF1_EVEN_OFFSET 0x092C + +#define RADEON_CAP0_BUF_PITCH 0x0930 +#define RADEON_CAP0_V_WINDOW 0x0934 +#define RADEON_CAP0_H_WINDOW 0x0938 +#define RADEON_CAP0_VBI0_OFFSET 0x093C +#define RADEON_CAP0_VBI1_OFFSET 0x0940 +#define RADEON_CAP0_VBI_V_WINDOW 0x0944 +#define RADEON_CAP0_VBI_H_WINDOW 0x0948 +#define RADEON_CAP0_PORT_MODE_CNTL 0x094C +#define RADEON_CAP0_TRIG_CNTL 0x0950 +#define RADEON_CAP0_DEBUG 0x0954 +#define RADEON_CAP0_CONFIG 0x0958 +# define RADEON_CAP0_CONFIG_CONTINUOS 0x00000001 +# define RADEON_CAP0_CONFIG_START_FIELD_EVEN 0x00000002 +# define RADEON_CAP0_CONFIG_START_BUF_GET 0x00000004 +# define RADEON_CAP0_CONFIG_START_BUF_SET 0x00000008 +# define RADEON_CAP0_CONFIG_BUF_TYPE_ALT 0x00000010 +# define RADEON_CAP0_CONFIG_BUF_TYPE_FRAME 0x00000020 +# define RADEON_CAP0_CONFIG_ONESHOT_MODE_FRAME 0x00000040 +# define RADEON_CAP0_CONFIG_BUF_MODE_DOUBLE 0x00000080 +# define RADEON_CAP0_CONFIG_BUF_MODE_TRIPLE 0x00000100 +# define RADEON_CAP0_CONFIG_MIRROR_EN 0x00000200 +# define RADEON_CAP0_CONFIG_ONESHOT_MIRROR_EN 0x00000400 +# define RADEON_CAP0_CONFIG_VIDEO_SIGNED_UV 0x00000800 +# define RADEON_CAP0_CONFIG_ANC_DECODE_EN 0x00001000 +# define RADEON_CAP0_CONFIG_VBI_EN 0x00002000 +# define RADEON_CAP0_CONFIG_SOFT_PULL_DOWN_EN 0x00004000 +# define RADEON_CAP0_CONFIG_VIP_EXTEND_FLAG_EN 0x00008000 +# define RADEON_CAP0_CONFIG_FAKE_FIELD_EN 0x00010000 +# define RADEON_CAP0_CONFIG_ODD_ONE_MORE_LINE 0x00020000 +# define RADEON_CAP0_CONFIG_EVEN_ONE_MORE_LINE 0x00040000 +# define RADEON_CAP0_CONFIG_HORZ_DIVIDE_2 0x00080000 +# define RADEON_CAP0_CONFIG_HORZ_DIVIDE_4 0x00100000 +# define RADEON_CAP0_CONFIG_VERT_DIVIDE_2 0x00200000 +# define RADEON_CAP0_CONFIG_VERT_DIVIDE_4 0x00400000 +# define RADEON_CAP0_CONFIG_FORMAT_BROOKTREE 0x00000000 +# define RADEON_CAP0_CONFIG_FORMAT_CCIR656 0x00800000 +# define RADEON_CAP0_CONFIG_FORMAT_ZV 0x01000000 +# define RADEON_CAP0_CONFIG_FORMAT_VIP 0x01800000 +# define RADEON_CAP0_CONFIG_FORMAT_TRANSPORT 0x02000000 +# define RADEON_CAP0_CONFIG_HORZ_DECIMATOR 0x04000000 +# define RADEON_CAP0_CONFIG_VIDEO_IN_YVYU422 0x00000000 +# define RADEON_CAP0_CONFIG_VIDEO_IN_VYUY422 0x20000000 +# define RADEON_CAP0_CONFIG_VBI_DIVIDE_2 0x40000000 +# define RADEON_CAP0_CONFIG_VBI_DIVIDE_4 0x80000000 +#define RADEON_CAP0_ANC_ODD_OFFSET 0x095C +#define RADEON_CAP0_ANC_EVEN_OFFSET 0x0960 +#define RADEON_CAP0_ANC_H_WINDOW 0x0964 +#define RADEON_CAP0_VIDEO_SYNC_TEST 0x0968 +#define RADEON_CAP0_ONESHOT_BUF_OFFSET 0x096C +#define RADEON_CAP0_BUF_STATUS 0x0970 +/* #define RADEON_CAP0_DWNSC_XRATIO 0x0978 */ +/* #define RADEON_CAP0_XSHARPNESS 0x097C */ +#define RADEON_CAP0_VBI2_OFFSET 0x0980 +#define RADEON_CAP0_VBI3_OFFSET 0x0984 +#define RADEON_CAP0_ANC2_OFFSET 0x0988 +#define RADEON_CAP0_ANC3_OFFSET 0x098C +#define RADEON_VID_BUFFER_CONTROL 0x0900 + +/* second capture unit */ + +#define RADEON_CAP1_BUF0_OFFSET 0x0990 +#define RADEON_CAP1_BUF1_OFFSET 0x0994 +#define RADEON_CAP1_BUF0_EVEN_OFFSET 0x0998 +#define RADEON_CAP1_BUF1_EVEN_OFFSET 0x099C + +#define RADEON_CAP1_BUF_PITCH 0x09A0 +#define RADEON_CAP1_V_WINDOW 0x09A4 +#define RADEON_CAP1_H_WINDOW 0x09A8 +#define RADEON_CAP1_VBI_ODD_OFFSET 0x09AC +#define RADEON_CAP1_VBI_EVEN_OFFSET 0x09B0 +#define RADEON_CAP1_VBI_V_WINDOW 0x09B4 +#define RADEON_CAP1_VBI_H_WINDOW 0x09B8 +#define RADEON_CAP1_PORT_MODE_CNTL 0x09BC +#define RADEON_CAP1_TRIG_CNTL 0x09C0 +#define RADEON_CAP1_DEBUG 0x09C4 +#define RADEON_CAP1_CONFIG 0x09C8 +#define RADEON_CAP1_ANC_ODD_OFFSET 0x09CC +#define RADEON_CAP1_ANC_EVEN_OFFSET 0x09D0 +#define RADEON_CAP1_ANC_H_WINDOW 0x09D4 +#define RADEON_CAP1_VIDEO_SYNC_TEST 0x09D8 +#define RADEON_CAP1_ONESHOT_BUF_OFFSET 0x09DC +#define RADEON_CAP1_BUF_STATUS 0x09E0 +#define RADEON_CAP1_DWNSC_XRATIO 0x09E8 +#define RADEON_CAP1_XSHARPNESS 0x09EC + +/* misc multimedia registers */ + +#define RADEON_IDCT_RUNS 0x1F80 +#define RADEON_IDCT_LEVELS 0x1F84 +#define RADEON_IDCT_CONTROL 0x1FBC +#define RADEON_IDCT_AUTH_CONTROL 0x1F88 +#define RADEON_IDCT_AUTH 0x1F8C + +#define RADEON_P2PLL_CNTL 0x002a /* P2PLL */ +# define RADEON_P2PLL_RESET (1 << 0) +# define RADEON_P2PLL_SLEEP (1 << 1) +# define RADEON_P2PLL_PVG_MASK (7 << 11) +# define RADEON_P2PLL_PVG_SHIFT 11 +# define RADEON_P2PLL_ATOMIC_UPDATE_EN (1 << 16) +# define RADEON_P2PLL_VGA_ATOMIC_UPDATE_EN (1 << 17) +# define RADEON_P2PLL_ATOMIC_UPDATE_VSYNC (1 << 18) +#define RADEON_P2PLL_DIV_0 0x002c +# define RADEON_P2PLL_FB0_DIV_MASK 0x07ff +# define RADEON_P2PLL_POST0_DIV_MASK 0x00070000 +#define RADEON_P2PLL_REF_DIV 0x002B /* PLL */ +# define RADEON_P2PLL_REF_DIV_MASK 0x03ff +# define RADEON_P2PLL_ATOMIC_UPDATE_R (1 << 15) /* same as _W */ +# define RADEON_P2PLL_ATOMIC_UPDATE_W (1 << 15) /* same as _R */ +# define R300_PPLL_REF_DIV_ACC_MASK (0x3ff << 18) +# define R300_PPLL_REF_DIV_ACC_SHIFT 18 +#define RADEON_PALETTE_DATA 0x00b4 +#define RADEON_PALETTE_30_DATA 0x00b8 +#define RADEON_PALETTE_INDEX 0x00b0 +#define RADEON_PCI_GART_PAGE 0x017c +#define RADEON_PIXCLKS_CNTL 0x002d +# define RADEON_PIX2CLK_SRC_SEL_MASK 0x03 +# define RADEON_PIX2CLK_SRC_SEL_CPUCLK 0x00 +# define RADEON_PIX2CLK_SRC_SEL_PSCANCLK 0x01 +# define RADEON_PIX2CLK_SRC_SEL_BYTECLK 0x02 +# define RADEON_PIX2CLK_SRC_SEL_P2PLLCLK 0x03 +# define RADEON_PIX2CLK_ALWAYS_ONb (1<<6) +# define RADEON_PIX2CLK_DAC_ALWAYS_ONb (1<<7) +# define RADEON_PIXCLK_TV_SRC_SEL (1 << 8) +# define RADEON_DISP_TVOUT_PIXCLK_TV_ALWAYS_ONb (1 << 9) +# define R300_DVOCLK_ALWAYS_ONb (1 << 10) +# define RADEON_PIXCLK_BLEND_ALWAYS_ONb (1 << 11) +# define RADEON_PIXCLK_GV_ALWAYS_ONb (1 << 12) +# define RADEON_PIXCLK_DIG_TMDS_ALWAYS_ONb (1 << 13) +# define R300_PIXCLK_DVO_ALWAYS_ONb (1 << 13) +# define RADEON_PIXCLK_LVDS_ALWAYS_ONb (1 << 14) +# define RADEON_PIXCLK_TMDS_ALWAYS_ONb (1 << 15) +# define R300_PIXCLK_TRANS_ALWAYS_ONb (1 << 16) +# define R300_PIXCLK_TVO_ALWAYS_ONb (1 << 17) +# define R300_P2G2CLK_ALWAYS_ONb (1 << 18) +# define R300_P2G2CLK_DAC_ALWAYS_ONb (1 << 19) +# define R300_DISP_DAC_PIXCLK_DAC2_BLANK_OFF (1 << 23) +#define RADEON_PLANE_3D_MASK_C 0x1d44 +#define RADEON_PLL_TEST_CNTL 0x0013 /* PLL */ +# define RADEON_PLL_MASK_READ_B (1 << 9) +#define RADEON_PMI_CAP_ID 0x0f5c /* PCI */ +#define RADEON_PMI_DATA 0x0f63 /* PCI */ +#define RADEON_PMI_NXT_CAP_PTR 0x0f5d /* PCI */ +#define RADEON_PMI_PMC_REG 0x0f5e /* PCI */ +#define RADEON_PMI_PMCSR_REG 0x0f60 /* PCI */ +#define RADEON_PMI_REGISTER 0x0f5c /* PCI */ +#define RADEON_PPLL_CNTL 0x0002 /* PLL */ +# define RADEON_PPLL_RESET (1 << 0) +# define RADEON_PPLL_SLEEP (1 << 1) +# define RADEON_PPLL_PVG_MASK (7 << 11) +# define RADEON_PPLL_PVG_SHIFT 11 +# define RADEON_PPLL_ATOMIC_UPDATE_EN (1 << 16) +# define RADEON_PPLL_VGA_ATOMIC_UPDATE_EN (1 << 17) +# define RADEON_PPLL_ATOMIC_UPDATE_VSYNC (1 << 18) +#define RADEON_PPLL_DIV_0 0x0004 /* PLL */ +#define RADEON_PPLL_DIV_1 0x0005 /* PLL */ +#define RADEON_PPLL_DIV_2 0x0006 /* PLL */ +#define RADEON_PPLL_DIV_3 0x0007 /* PLL */ +# define RADEON_PPLL_FB3_DIV_MASK 0x07ff +# define RADEON_PPLL_POST3_DIV_MASK 0x00070000 +#define RADEON_PPLL_REF_DIV 0x0003 /* PLL */ +# define RADEON_PPLL_REF_DIV_MASK 0x03ff +# define RADEON_PPLL_ATOMIC_UPDATE_R (1 << 15) /* same as _W */ +# define RADEON_PPLL_ATOMIC_UPDATE_W (1 << 15) /* same as _R */ +#define RADEON_PWR_MNGMT_CNTL_STATUS 0x0f60 /* PCI */ + +#define RADEON_RBBM_GUICNTL 0x172c +# define RADEON_HOST_DATA_SWAP_NONE (0 << 0) +# define RADEON_HOST_DATA_SWAP_16BIT (1 << 0) +# define RADEON_HOST_DATA_SWAP_32BIT (2 << 0) +# define RADEON_HOST_DATA_SWAP_HDW (3 << 0) +#define RADEON_RBBM_SOFT_RESET 0x00f0 +# define RADEON_SOFT_RESET_CP (1 << 0) +# define RADEON_SOFT_RESET_HI (1 << 1) +# define RADEON_SOFT_RESET_SE (1 << 2) +# define RADEON_SOFT_RESET_RE (1 << 3) +# define RADEON_SOFT_RESET_PP (1 << 4) +# define RADEON_SOFT_RESET_E2 (1 << 5) +# define RADEON_SOFT_RESET_RB (1 << 6) +# define RADEON_SOFT_RESET_HDP (1 << 7) +#define RADEON_RBBM_STATUS 0x0e40 +# define RADEON_RBBM_FIFOCNT_MASK 0x007f +# define RADEON_RBBM_ACTIVE (1 << 31) +#define RADEON_RB2D_DSTCACHE_CTLSTAT 0x342c +# define RADEON_RB2D_DC_FLUSH (3 << 0) +# define RADEON_RB2D_DC_FREE (3 << 2) +# define RADEON_RB2D_DC_FLUSH_ALL 0xf +# define RADEON_RB2D_DC_BUSY (1 << 31) +#define RADEON_RB2D_DSTCACHE_MODE 0x3428 +#define RADEON_DSTCACHE_CTLSTAT 0x1714 + +#define RADEON_RB3D_ZCACHE_MODE 0x3250 +#define RADEON_RB3D_ZCACHE_CTLSTAT 0x3254 +# define RADEON_RB3D_ZC_FLUSH_ALL 0x5 +#define RADEON_RB3D_DSTCACHE_MODE 0x3258 +# define RADEON_RB3D_DC_CACHE_ENABLE (0) +# define RADEON_RB3D_DC_2D_CACHE_DISABLE (1) +# define RADEON_RB3D_DC_3D_CACHE_DISABLE (2) +# define RADEON_RB3D_DC_CACHE_DISABLE (3) +# define RADEON_RB3D_DC_2D_CACHE_LINESIZE_128 (1 << 2) +# define RADEON_RB3D_DC_3D_CACHE_LINESIZE_128 (2 << 2) +# define RADEON_RB3D_DC_2D_CACHE_AUTOFLUSH (1 << 8) +# define RADEON_RB3D_DC_3D_CACHE_AUTOFLUSH (2 << 8) +# define R200_RB3D_DC_2D_CACHE_AUTOFREE (1 << 10) +# define R200_RB3D_DC_3D_CACHE_AUTOFREE (2 << 10) +# define RADEON_RB3D_DC_FORCE_RMW (1 << 16) +# define RADEON_RB3D_DC_DISABLE_RI_FILL (1 << 24) +# define RADEON_RB3D_DC_DISABLE_RI_READ (1 << 25) + +#define RADEON_RB3D_DSTCACHE_CTLSTAT 0x325C +# define RADEON_RB3D_DC_FLUSH (3 << 0) +# define RADEON_RB3D_DC_FREE (3 << 2) +# define RADEON_RB3D_DC_FLUSH_ALL 0xf +# define RADEON_RB3D_DC_BUSY (1 << 31) + +#define RADEON_REG_BASE 0x0f18 /* PCI */ +#define RADEON_REGPROG_INF 0x0f09 /* PCI */ +#define RADEON_REVISION_ID 0x0f08 /* PCI */ + +#define RADEON_SC_BOTTOM 0x164c +#define RADEON_SC_BOTTOM_RIGHT 0x16f0 +#define RADEON_SC_BOTTOM_RIGHT_C 0x1c8c +#define RADEON_SC_LEFT 0x1640 +#define RADEON_SC_RIGHT 0x1644 +#define RADEON_SC_TOP 0x1648 +#define RADEON_SC_TOP_LEFT 0x16ec +#define RADEON_SC_TOP_LEFT_C 0x1c88 +# define RADEON_SC_SIGN_MASK_LO 0x8000 +# define RADEON_SC_SIGN_MASK_HI 0x80000000 +#define RADEON_M_SPLL_REF_FB_DIV 0x000a /* PLL */ +# define RADEON_M_SPLL_REF_DIV_SHIFT 0 +# define RADEON_M_SPLL_REF_DIV_MASK 0xff +# define RADEON_MPLL_FB_DIV_SHIFT 8 +# define RADEON_MPLL_FB_DIV_MASK 0xff +# define RADEON_SPLL_FB_DIV_SHIFT 16 +# define RADEON_SPLL_FB_DIV_MASK 0xff +#define RADEON_SPLL_CNTL 0x000c /* PLL */ +# define RADEON_SPLL_SLEEP (1 << 0) +# define RADEON_SPLL_RESET (1 << 1) +# define RADEON_SPLL_PCP_MASK 0x7 +# define RADEON_SPLL_PCP_SHIFT 8 +# define RADEON_SPLL_PVG_MASK 0x7 +# define RADEON_SPLL_PVG_SHIFT 11 +# define RADEON_SPLL_PDC_MASK 0x3 +# define RADEON_SPLL_PDC_SHIFT 14 +#define RADEON_SCLK_CNTL 0x000d /* PLL */ +# define RADEON_SCLK_SRC_SEL_MASK 0x0007 +# define RADEON_DYN_STOP_LAT_MASK 0x00007ff8 +# define RADEON_CP_MAX_DYN_STOP_LAT 0x0008 +# define RADEON_SCLK_FORCEON_MASK 0xffff8000 +# define RADEON_SCLK_FORCE_DISP2 (1<<15) +# define RADEON_SCLK_FORCE_CP (1<<16) +# define RADEON_SCLK_FORCE_HDP (1<<17) +# define RADEON_SCLK_FORCE_DISP1 (1<<18) +# define RADEON_SCLK_FORCE_TOP (1<<19) +# define RADEON_SCLK_FORCE_E2 (1<<20) +# define RADEON_SCLK_FORCE_SE (1<<21) +# define RADEON_SCLK_FORCE_IDCT (1<<22) +# define RADEON_SCLK_FORCE_VIP (1<<23) +# define RADEON_SCLK_FORCE_RE (1<<24) +# define RADEON_SCLK_FORCE_PB (1<<25) +# define RADEON_SCLK_FORCE_TAM (1<<26) +# define RADEON_SCLK_FORCE_TDM (1<<27) +# define RADEON_SCLK_FORCE_RB (1<<28) +# define RADEON_SCLK_FORCE_TV_SCLK (1<<29) +# define RADEON_SCLK_FORCE_SUBPIC (1<<30) +# define RADEON_SCLK_FORCE_OV0 (1<<31) +# define R300_SCLK_FORCE_VAP (1<<21) +# define R300_SCLK_FORCE_SR (1<<25) +# define R300_SCLK_FORCE_PX (1<<26) +# define R300_SCLK_FORCE_TX (1<<27) +# define R300_SCLK_FORCE_US (1<<28) +# define R300_SCLK_FORCE_SU (1<<30) +#define R300_SCLK_CNTL2 0x1e /* PLL */ +# define R300_SCLK_TCL_MAX_DYN_STOP_LAT (1<<10) +# define R300_SCLK_GA_MAX_DYN_STOP_LAT (1<<11) +# define R300_SCLK_CBA_MAX_DYN_STOP_LAT (1<<12) +# define R300_SCLK_FORCE_TCL (1<<13) +# define R300_SCLK_FORCE_CBA (1<<14) +# define R300_SCLK_FORCE_GA (1<<15) +#define RADEON_SCLK_MORE_CNTL 0x0035 /* PLL */ +# define RADEON_SCLK_MORE_MAX_DYN_STOP_LAT 0x0007 +# define RADEON_SCLK_MORE_FORCEON 0x0700 +#define RADEON_SDRAM_MODE_REG 0x0158 +#define RADEON_SEQ8_DATA 0x03c5 /* VGA */ +#define RADEON_SEQ8_IDX 0x03c4 /* VGA */ +#define RADEON_SNAPSHOT_F_COUNT 0x0244 +#define RADEON_SNAPSHOT_VH_COUNTS 0x0240 +#define RADEON_SNAPSHOT_VIF_COUNT 0x024c +#define RADEON_SRC_OFFSET 0x15ac +#define RADEON_SRC_PITCH 0x15b0 +#define RADEON_SRC_PITCH_OFFSET 0x1428 +#define RADEON_SRC_SC_BOTTOM 0x165c +#define RADEON_SRC_SC_BOTTOM_RIGHT 0x16f4 +#define RADEON_SRC_SC_RIGHT 0x1654 +#define RADEON_SRC_X 0x1414 +#define RADEON_SRC_X_Y 0x1590 +#define RADEON_SRC_Y 0x1418 +#define RADEON_SRC_Y_X 0x1434 +#define RADEON_STATUS 0x0f06 /* PCI */ +#define RADEON_SUBPIC_CNTL 0x0540 /* ? */ +#define RADEON_SUB_CLASS 0x0f0a /* PCI */ +#define RADEON_SURFACE_CNTL 0x0b00 +# define RADEON_SURF_TRANSLATION_DIS (1 << 8) +# define RADEON_NONSURF_AP0_SWP_16BPP (1 << 20) +# define RADEON_NONSURF_AP0_SWP_32BPP (1 << 21) +# define RADEON_NONSURF_AP1_SWP_16BPP (1 << 22) +# define RADEON_NONSURF_AP1_SWP_32BPP (1 << 23) +#define RADEON_SURFACE0_INFO 0x0b0c +# define RADEON_SURF_TILE_COLOR_MACRO (0 << 16) +# define RADEON_SURF_TILE_COLOR_BOTH (1 << 16) +# define RADEON_SURF_TILE_DEPTH_32BPP (2 << 16) +# define RADEON_SURF_TILE_DEPTH_16BPP (3 << 16) +# define R200_SURF_TILE_NONE (0 << 16) +# define R200_SURF_TILE_COLOR_MACRO (1 << 16) +# define R200_SURF_TILE_COLOR_MICRO (2 << 16) +# define R200_SURF_TILE_COLOR_BOTH (3 << 16) +# define R200_SURF_TILE_DEPTH_32BPP (4 << 16) +# define R200_SURF_TILE_DEPTH_16BPP (5 << 16) +# define R300_SURF_TILE_NONE (0 << 16) +# define R300_SURF_TILE_COLOR_MACRO (1 << 16) +# define R300_SURF_TILE_DEPTH_32BPP (2 << 16) +# define RADEON_SURF_AP0_SWP_16BPP (1 << 20) +# define RADEON_SURF_AP0_SWP_32BPP (1 << 21) +# define RADEON_SURF_AP1_SWP_16BPP (1 << 22) +# define RADEON_SURF_AP1_SWP_32BPP (1 << 23) +#define RADEON_SURFACE0_LOWER_BOUND 0x0b04 +#define RADEON_SURFACE0_UPPER_BOUND 0x0b08 +#define RADEON_SURFACE1_INFO 0x0b1c +#define RADEON_SURFACE1_LOWER_BOUND 0x0b14 +#define RADEON_SURFACE1_UPPER_BOUND 0x0b18 +#define RADEON_SURFACE2_INFO 0x0b2c +#define RADEON_SURFACE2_LOWER_BOUND 0x0b24 +#define RADEON_SURFACE2_UPPER_BOUND 0x0b28 +#define RADEON_SURFACE3_INFO 0x0b3c +#define RADEON_SURFACE3_LOWER_BOUND 0x0b34 +#define RADEON_SURFACE3_UPPER_BOUND 0x0b38 +#define RADEON_SURFACE4_INFO 0x0b4c +#define RADEON_SURFACE4_LOWER_BOUND 0x0b44 +#define RADEON_SURFACE4_UPPER_BOUND 0x0b48 +#define RADEON_SURFACE5_INFO 0x0b5c +#define RADEON_SURFACE5_LOWER_BOUND 0x0b54 +#define RADEON_SURFACE5_UPPER_BOUND 0x0b58 +#define RADEON_SURFACE6_INFO 0x0b6c +#define RADEON_SURFACE6_LOWER_BOUND 0x0b64 +#define RADEON_SURFACE6_UPPER_BOUND 0x0b68 +#define RADEON_SURFACE7_INFO 0x0b7c +#define RADEON_SURFACE7_LOWER_BOUND 0x0b74 +#define RADEON_SURFACE7_UPPER_BOUND 0x0b78 +#define RADEON_SW_SEMAPHORE 0x013c + +#define RADEON_TEST_DEBUG_CNTL 0x0120 +#define RADEON_TEST_DEBUG_CNTL__TEST_DEBUG_OUT_EN 0x00000001 + +#define RADEON_TEST_DEBUG_MUX 0x0124 +#define RADEON_TEST_DEBUG_OUT 0x012c +#define RADEON_TMDS_PLL_CNTL 0x02a8 +#define RADEON_TMDS_TRANSMITTER_CNTL 0x02a4 +# define RADEON_TMDS_TRANSMITTER_PLLEN 1 +# define RADEON_TMDS_TRANSMITTER_PLLRST 2 +#define RADEON_TRAIL_BRES_DEC 0x1614 +#define RADEON_TRAIL_BRES_ERR 0x160c +#define RADEON_TRAIL_BRES_INC 0x1610 +#define RADEON_TRAIL_X 0x1618 +#define RADEON_TRAIL_X_SUB 0x1620 + +#define RADEON_VCLK_ECP_CNTL 0x0008 /* PLL */ +# define RADEON_VCLK_SRC_SEL_MASK 0x03 +# define RADEON_VCLK_SRC_SEL_CPUCLK 0x00 +# define RADEON_VCLK_SRC_SEL_PSCANCLK 0x01 +# define RADEON_VCLK_SRC_SEL_BYTECLK 0x02 +# define RADEON_VCLK_SRC_SEL_PPLLCLK 0x03 +# define RADEON_PIXCLK_ALWAYS_ONb (1<<6) +# define RADEON_PIXCLK_DAC_ALWAYS_ONb (1<<7) +# define R300_DISP_DAC_PIXCLK_DAC_BLANK_OFF (1<<23) + +#define RADEON_VENDOR_ID 0x0f00 /* PCI */ +#define RADEON_VGA_DDA_CONFIG 0x02e8 +#define RADEON_VGA_DDA_ON_OFF 0x02ec +#define RADEON_VID_BUFFER_CONTROL 0x0900 +#define RADEON_VIDEOMUX_CNTL 0x0190 + +/* VIP bus */ +#define RADEON_VIPH_CH0_DATA 0x0c00 +#define RADEON_VIPH_CH1_DATA 0x0c04 +#define RADEON_VIPH_CH2_DATA 0x0c08 +#define RADEON_VIPH_CH3_DATA 0x0c0c +#define RADEON_VIPH_CH0_ADDR 0x0c10 +#define RADEON_VIPH_CH1_ADDR 0x0c14 +#define RADEON_VIPH_CH2_ADDR 0x0c18 +#define RADEON_VIPH_CH3_ADDR 0x0c1c +#define RADEON_VIPH_CH0_SBCNT 0x0c20 +#define RADEON_VIPH_CH1_SBCNT 0x0c24 +#define RADEON_VIPH_CH2_SBCNT 0x0c28 +#define RADEON_VIPH_CH3_SBCNT 0x0c2c +#define RADEON_VIPH_CH0_ABCNT 0x0c30 +#define RADEON_VIPH_CH1_ABCNT 0x0c34 +#define RADEON_VIPH_CH2_ABCNT 0x0c38 +#define RADEON_VIPH_CH3_ABCNT 0x0c3c +#define RADEON_VIPH_CONTROL 0x0c40 +# define RADEON_VIP_BUSY 0 +# define RADEON_VIP_IDLE 1 +# define RADEON_VIP_RESET 2 +# define RADEON_VIPH_EN (1 << 21) +#define RADEON_VIPH_DV_LAT 0x0c44 +#define RADEON_VIPH_BM_CHUNK 0x0c48 +#define RADEON_VIPH_DV_INT 0x0c4c +#define RADEON_VIPH_TIMEOUT_STAT 0x0c50 +#define RADEON_VIPH_TIMEOUT_STAT__VIPH_REG_STAT 0x00000010 +#define RADEON_VIPH_TIMEOUT_STAT__VIPH_REG_AK 0x00000010 +#define RADEON_VIPH_TIMEOUT_STAT__VIPH_REGR_DIS 0x01000000 + +#define RADEON_VIPH_REG_DATA 0x0084 +#define RADEON_VIPH_REG_ADDR 0x0080 + + +#define RADEON_WAIT_UNTIL 0x1720 +# define RADEON_WAIT_CRTC_PFLIP (1 << 0) +# define RADEON_WAIT_RE_CRTC_VLINE (1 << 1) +# define RADEON_WAIT_FE_CRTC_VLINE (1 << 2) +# define RADEON_WAIT_CRTC_VLINE (1 << 3) +# define RADEON_WAIT_DMA_VID_IDLE (1 << 8) +# define RADEON_WAIT_DMA_GUI_IDLE (1 << 9) +# define RADEON_WAIT_CMDFIFO (1 << 10) /* wait for CMDFIFO_ENTRIES */ +# define RADEON_WAIT_OV0_FLIP (1 << 11) +# define RADEON_WAIT_AGP_FLUSH (1 << 13) +# define RADEON_WAIT_2D_IDLE (1 << 14) +# define RADEON_WAIT_3D_IDLE (1 << 15) +# define RADEON_WAIT_2D_IDLECLEAN (1 << 16) +# define RADEON_WAIT_3D_IDLECLEAN (1 << 17) +# define RADEON_WAIT_HOST_IDLECLEAN (1 << 18) +# define RADEON_CMDFIFO_ENTRIES_SHIFT 10 +# define RADEON_CMDFIFO_ENTRIES_MASK 0x7f +# define RADEON_WAIT_VAP_IDLE (1 << 28) +# define RADEON_WAIT_BOTH_CRTC_PFLIP (1 << 30) +# define RADEON_ENG_DISPLAY_SELECT_CRTC0 (0 << 31) +# define RADEON_ENG_DISPLAY_SELECT_CRTC1 (1 << 31) + +#define RADEON_X_MPLL_REF_FB_DIV 0x000a /* PLL */ +#define RADEON_XCLK_CNTL 0x000d /* PLL */ +#define RADEON_XDLL_CNTL 0x000c /* PLL */ +#define RADEON_XPLL_CNTL 0x000b /* PLL */ + + + + /* Registers for 3D/TCL */ +#define RADEON_PP_BORDER_COLOR_0 0x1d40 +#define RADEON_PP_BORDER_COLOR_1 0x1d44 +#define RADEON_PP_BORDER_COLOR_2 0x1d48 +#define RADEON_PP_CNTL 0x1c38 +# define RADEON_STIPPLE_ENABLE (1 << 0) +# define RADEON_SCISSOR_ENABLE (1 << 1) +# define RADEON_PATTERN_ENABLE (1 << 2) +# define RADEON_SHADOW_ENABLE (1 << 3) +# define RADEON_TEX_ENABLE_MASK (0xf << 4) +# define RADEON_TEX_0_ENABLE (1 << 4) +# define RADEON_TEX_1_ENABLE (1 << 5) +# define RADEON_TEX_2_ENABLE (1 << 6) +# define RADEON_TEX_3_ENABLE (1 << 7) +# define RADEON_TEX_BLEND_ENABLE_MASK (0xf << 12) +# define RADEON_TEX_BLEND_0_ENABLE (1 << 12) +# define RADEON_TEX_BLEND_1_ENABLE (1 << 13) +# define RADEON_TEX_BLEND_2_ENABLE (1 << 14) +# define RADEON_TEX_BLEND_3_ENABLE (1 << 15) +# define RADEON_PLANAR_YUV_ENABLE (1 << 20) +# define RADEON_SPECULAR_ENABLE (1 << 21) +# define RADEON_FOG_ENABLE (1 << 22) +# define RADEON_ALPHA_TEST_ENABLE (1 << 23) +# define RADEON_ANTI_ALIAS_NONE (0 << 24) +# define RADEON_ANTI_ALIAS_LINE (1 << 24) +# define RADEON_ANTI_ALIAS_POLY (2 << 24) +# define RADEON_ANTI_ALIAS_LINE_POLY (3 << 24) +# define RADEON_BUMP_MAP_ENABLE (1 << 26) +# define RADEON_BUMPED_MAP_T0 (0 << 27) +# define RADEON_BUMPED_MAP_T1 (1 << 27) +# define RADEON_BUMPED_MAP_T2 (2 << 27) +# define RADEON_TEX_3D_ENABLE_0 (1 << 29) +# define RADEON_TEX_3D_ENABLE_1 (1 << 30) +# define RADEON_MC_ENABLE (1 << 31) +#define RADEON_PP_FOG_COLOR 0x1c18 +# define RADEON_FOG_COLOR_MASK 0x00ffffff +# define RADEON_FOG_VERTEX (0 << 24) +# define RADEON_FOG_TABLE (1 << 24) +# define RADEON_FOG_USE_DEPTH (0 << 25) +# define RADEON_FOG_USE_DIFFUSE_ALPHA (2 << 25) +# define RADEON_FOG_USE_SPEC_ALPHA (3 << 25) +#define RADEON_PP_LUM_MATRIX 0x1d00 +#define RADEON_PP_MISC 0x1c14 +# define RADEON_REF_ALPHA_MASK 0x000000ff +# define RADEON_ALPHA_TEST_FAIL (0 << 8) +# define RADEON_ALPHA_TEST_LESS (1 << 8) +# define RADEON_ALPHA_TEST_LEQUAL (2 << 8) +# define RADEON_ALPHA_TEST_EQUAL (3 << 8) +# define RADEON_ALPHA_TEST_GEQUAL (4 << 8) +# define RADEON_ALPHA_TEST_GREATER (5 << 8) +# define RADEON_ALPHA_TEST_NEQUAL (6 << 8) +# define RADEON_ALPHA_TEST_PASS (7 << 8) +# define RADEON_ALPHA_TEST_OP_MASK (7 << 8) +# define RADEON_CHROMA_FUNC_FAIL (0 << 16) +# define RADEON_CHROMA_FUNC_PASS (1 << 16) +# define RADEON_CHROMA_FUNC_NEQUAL (2 << 16) +# define RADEON_CHROMA_FUNC_EQUAL (3 << 16) +# define RADEON_CHROMA_KEY_NEAREST (0 << 18) +# define RADEON_CHROMA_KEY_ZERO (1 << 18) +# define RADEON_SHADOW_ID_AUTO_INC (1 << 20) +# define RADEON_SHADOW_FUNC_EQUAL (0 << 21) +# define RADEON_SHADOW_FUNC_NEQUAL (1 << 21) +# define RADEON_SHADOW_PASS_1 (0 << 22) +# define RADEON_SHADOW_PASS_2 (1 << 22) +# define RADEON_RIGHT_HAND_CUBE_D3D (0 << 24) +# define RADEON_RIGHT_HAND_CUBE_OGL (1 << 24) +#define RADEON_PP_ROT_MATRIX_0 0x1d58 +#define RADEON_PP_ROT_MATRIX_1 0x1d5c +#define RADEON_PP_TXFILTER_0 0x1c54 +#define RADEON_PP_TXFILTER_1 0x1c6c +#define RADEON_PP_TXFILTER_2 0x1c84 +# define RADEON_MAG_FILTER_NEAREST (0 << 0) +# define RADEON_MAG_FILTER_LINEAR (1 << 0) +# define RADEON_MAG_FILTER_MASK (1 << 0) +# define RADEON_MIN_FILTER_NEAREST (0 << 1) +# define RADEON_MIN_FILTER_LINEAR (1 << 1) +# define RADEON_MIN_FILTER_NEAREST_MIP_NEAREST (2 << 1) +# define RADEON_MIN_FILTER_NEAREST_MIP_LINEAR (3 << 1) +# define RADEON_MIN_FILTER_LINEAR_MIP_NEAREST (6 << 1) +# define RADEON_MIN_FILTER_LINEAR_MIP_LINEAR (7 << 1) +# define RADEON_MIN_FILTER_ANISO_NEAREST (8 << 1) +# define RADEON_MIN_FILTER_ANISO_LINEAR (9 << 1) +# define RADEON_MIN_FILTER_ANISO_NEAREST_MIP_NEAREST (10 << 1) +# define RADEON_MIN_FILTER_ANISO_NEAREST_MIP_LINEAR (11 << 1) +# define RADEON_MIN_FILTER_MASK (15 << 1) +# define RADEON_MAX_ANISO_1_TO_1 (0 << 5) +# define RADEON_MAX_ANISO_2_TO_1 (1 << 5) +# define RADEON_MAX_ANISO_4_TO_1 (2 << 5) +# define RADEON_MAX_ANISO_8_TO_1 (3 << 5) +# define RADEON_MAX_ANISO_16_TO_1 (4 << 5) +# define RADEON_MAX_ANISO_MASK (7 << 5) +# define RADEON_LOD_BIAS_MASK (0xff << 8) +# define RADEON_LOD_BIAS_SHIFT 8 +# define RADEON_MAX_MIP_LEVEL_MASK (0x0f << 16) +# define RADEON_MAX_MIP_LEVEL_SHIFT 16 +# define RADEON_YUV_TO_RGB (1 << 20) +# define RADEON_YUV_TEMPERATURE_COOL (0 << 21) +# define RADEON_YUV_TEMPERATURE_HOT (1 << 21) +# define RADEON_YUV_TEMPERATURE_MASK (1 << 21) +# define RADEON_WRAPEN_S (1 << 22) +# define RADEON_CLAMP_S_WRAP (0 << 23) +# define RADEON_CLAMP_S_MIRROR (1 << 23) +# define RADEON_CLAMP_S_CLAMP_LAST (2 << 23) +# define RADEON_CLAMP_S_MIRROR_CLAMP_LAST (3 << 23) +# define RADEON_CLAMP_S_CLAMP_BORDER (4 << 23) +# define RADEON_CLAMP_S_MIRROR_CLAMP_BORDER (5 << 23) +# define RADEON_CLAMP_S_CLAMP_GL (6 << 23) +# define RADEON_CLAMP_S_MIRROR_CLAMP_GL (7 << 23) +# define RADEON_CLAMP_S_MASK (7 << 23) +# define RADEON_WRAPEN_T (1 << 26) +# define RADEON_CLAMP_T_WRAP (0 << 27) +# define RADEON_CLAMP_T_MIRROR (1 << 27) +# define RADEON_CLAMP_T_CLAMP_LAST (2 << 27) +# define RADEON_CLAMP_T_MIRROR_CLAMP_LAST (3 << 27) +# define RADEON_CLAMP_T_CLAMP_BORDER (4 << 27) +# define RADEON_CLAMP_T_MIRROR_CLAMP_BORDER (5 << 27) +# define RADEON_CLAMP_T_CLAMP_GL (6 << 27) +# define RADEON_CLAMP_T_MIRROR_CLAMP_GL (7 << 27) +# define RADEON_CLAMP_T_MASK (7 << 27) +# define RADEON_BORDER_MODE_OGL (0 << 31) +# define RADEON_BORDER_MODE_D3D (1 << 31) +#define RADEON_PP_TXFORMAT_0 0x1c58 +#define RADEON_PP_TXFORMAT_1 0x1c70 +#define RADEON_PP_TXFORMAT_2 0x1c88 +# define RADEON_TXFORMAT_I8 (0 << 0) +# define RADEON_TXFORMAT_AI88 (1 << 0) +# define RADEON_TXFORMAT_RGB332 (2 << 0) +# define RADEON_TXFORMAT_ARGB1555 (3 << 0) +# define RADEON_TXFORMAT_RGB565 (4 << 0) +# define RADEON_TXFORMAT_ARGB4444 (5 << 0) +# define RADEON_TXFORMAT_ARGB8888 (6 << 0) +# define RADEON_TXFORMAT_RGBA8888 (7 << 0) +# define RADEON_TXFORMAT_Y8 (8 << 0) +# define RADEON_TXFORMAT_VYUY422 (10 << 0) +# define RADEON_TXFORMAT_YVYU422 (11 << 0) +# define RADEON_TXFORMAT_DXT1 (12 << 0) +# define RADEON_TXFORMAT_DXT23 (14 << 0) +# define RADEON_TXFORMAT_DXT45 (15 << 0) +# define RADEON_TXFORMAT_SHADOW16 (16 << 0) +# define RADEON_TXFORMAT_SHADOW32 (17 << 0) +# define RADEON_TXFORMAT_DUDV88 (18 << 0) +# define RADEON_TXFORMAT_LDUDV655 (19 << 0) +# define RADEON_TXFORMAT_LDUDUV8888 (20 << 0) +# define RADEON_TXFORMAT_FORMAT_MASK (31 << 0) +# define RADEON_TXFORMAT_FORMAT_SHIFT 0 +# define RADEON_TXFORMAT_APPLE_YUV_MODE (1 << 5) +# define RADEON_TXFORMAT_ALPHA_IN_MAP (1 << 6) +# define RADEON_TXFORMAT_NON_POWER2 (1 << 7) +# define RADEON_TXFORMAT_WIDTH_MASK (15 << 8) +# define RADEON_TXFORMAT_WIDTH_SHIFT 8 +# define RADEON_TXFORMAT_HEIGHT_MASK (15 << 12) +# define RADEON_TXFORMAT_HEIGHT_SHIFT 12 +# define RADEON_TXFORMAT_F5_WIDTH_MASK (15 << 16) +# define RADEON_TXFORMAT_F5_WIDTH_SHIFT 16 +# define RADEON_TXFORMAT_F5_HEIGHT_MASK (15 << 20) +# define RADEON_TXFORMAT_F5_HEIGHT_SHIFT 20 +# define RADEON_TXFORMAT_ST_ROUTE_STQ0 (0 << 24) +# define RADEON_TXFORMAT_ST_ROUTE_MASK (3 << 24) +# define RADEON_TXFORMAT_ST_ROUTE_STQ1 (1 << 24) +# define RADEON_TXFORMAT_ST_ROUTE_STQ2 (2 << 24) +# define RADEON_TXFORMAT_ENDIAN_NO_SWAP (0 << 26) +# define RADEON_TXFORMAT_ENDIAN_16BPP_SWAP (1 << 26) +# define RADEON_TXFORMAT_ENDIAN_32BPP_SWAP (2 << 26) +# define RADEON_TXFORMAT_ENDIAN_HALFDW_SWAP (3 << 26) +# define RADEON_TXFORMAT_ALPHA_MASK_ENABLE (1 << 28) +# define RADEON_TXFORMAT_CHROMA_KEY_ENABLE (1 << 29) +# define RADEON_TXFORMAT_CUBIC_MAP_ENABLE (1 << 30) +# define RADEON_TXFORMAT_PERSPECTIVE_ENABLE (1 << 31) +#define RADEON_PP_CUBIC_FACES_0 0x1d24 +#define RADEON_PP_CUBIC_FACES_1 0x1d28 +#define RADEON_PP_CUBIC_FACES_2 0x1d2c +# define RADEON_FACE_WIDTH_1_SHIFT 0 +# define RADEON_FACE_HEIGHT_1_SHIFT 4 +# define RADEON_FACE_WIDTH_1_MASK (0xf << 0) +# define RADEON_FACE_HEIGHT_1_MASK (0xf << 4) +# define RADEON_FACE_WIDTH_2_SHIFT 8 +# define RADEON_FACE_HEIGHT_2_SHIFT 12 +# define RADEON_FACE_WIDTH_2_MASK (0xf << 8) +# define RADEON_FACE_HEIGHT_2_MASK (0xf << 12) +# define RADEON_FACE_WIDTH_3_SHIFT 16 +# define RADEON_FACE_HEIGHT_3_SHIFT 20 +# define RADEON_FACE_WIDTH_3_MASK (0xf << 16) +# define RADEON_FACE_HEIGHT_3_MASK (0xf << 20) +# define RADEON_FACE_WIDTH_4_SHIFT 24 +# define RADEON_FACE_HEIGHT_4_SHIFT 28 +# define RADEON_FACE_WIDTH_4_MASK (0xf << 24) +# define RADEON_FACE_HEIGHT_4_MASK (0xf << 28) + +#define RADEON_PP_TXOFFSET_0 0x1c5c +#define RADEON_PP_TXOFFSET_1 0x1c74 +#define RADEON_PP_TXOFFSET_2 0x1c8c +# define RADEON_TXO_ENDIAN_NO_SWAP (0 << 0) +# define RADEON_TXO_ENDIAN_BYTE_SWAP (1 << 0) +# define RADEON_TXO_ENDIAN_WORD_SWAP (2 << 0) +# define RADEON_TXO_ENDIAN_HALFDW_SWAP (3 << 0) +# define RADEON_TXO_MACRO_LINEAR (0 << 2) +# define RADEON_TXO_MACRO_TILE (1 << 2) +# define RADEON_TXO_MICRO_LINEAR (0 << 3) +# define RADEON_TXO_MICRO_TILE_X2 (1 << 3) +# define RADEON_TXO_MICRO_TILE_OPT (2 << 3) +# define RADEON_TXO_OFFSET_MASK 0xffffffe0 +# define RADEON_TXO_OFFSET_SHIFT 5 + +#define RADEON_PP_CUBIC_OFFSET_T0_0 0x1dd0 /* bits [31:5] */ +#define RADEON_PP_CUBIC_OFFSET_T0_1 0x1dd4 +#define RADEON_PP_CUBIC_OFFSET_T0_2 0x1dd8 +#define RADEON_PP_CUBIC_OFFSET_T0_3 0x1ddc +#define RADEON_PP_CUBIC_OFFSET_T0_4 0x1de0 +#define RADEON_PP_CUBIC_OFFSET_T1_0 0x1e00 +#define RADEON_PP_CUBIC_OFFSET_T1_1 0x1e04 +#define RADEON_PP_CUBIC_OFFSET_T1_2 0x1e08 +#define RADEON_PP_CUBIC_OFFSET_T1_3 0x1e0c +#define RADEON_PP_CUBIC_OFFSET_T1_4 0x1e10 +#define RADEON_PP_CUBIC_OFFSET_T2_0 0x1e14 +#define RADEON_PP_CUBIC_OFFSET_T2_1 0x1e18 +#define RADEON_PP_CUBIC_OFFSET_T2_2 0x1e1c +#define RADEON_PP_CUBIC_OFFSET_T2_3 0x1e20 +#define RADEON_PP_CUBIC_OFFSET_T2_4 0x1e24 + +#define RADEON_PP_TEX_SIZE_0 0x1d04 /* NPOT */ +#define RADEON_PP_TEX_SIZE_1 0x1d0c +#define RADEON_PP_TEX_SIZE_2 0x1d14 +# define RADEON_TEX_USIZE_MASK (0x7ff << 0) +# define RADEON_TEX_USIZE_SHIFT 0 +# define RADEON_TEX_VSIZE_MASK (0x7ff << 16) +# define RADEON_TEX_VSIZE_SHIFT 16 +# define RADEON_SIGNED_RGB_MASK (1 << 30) +# define RADEON_SIGNED_RGB_SHIFT 30 +# define RADEON_SIGNED_ALPHA_MASK (1 << 31) +# define RADEON_SIGNED_ALPHA_SHIFT 31 +#define RADEON_PP_TEX_PITCH_0 0x1d08 /* NPOT */ +#define RADEON_PP_TEX_PITCH_1 0x1d10 /* NPOT */ +#define RADEON_PP_TEX_PITCH_2 0x1d18 /* NPOT */ +/* note: bits 13-5: 32 byte aligned stride of texture map */ + +#define RADEON_PP_TXCBLEND_0 0x1c60 +#define RADEON_PP_TXCBLEND_1 0x1c78 +#define RADEON_PP_TXCBLEND_2 0x1c90 +# define RADEON_COLOR_ARG_A_SHIFT 0 +# define RADEON_COLOR_ARG_A_MASK (0x1f << 0) +# define RADEON_COLOR_ARG_A_ZERO (0 << 0) +# define RADEON_COLOR_ARG_A_CURRENT_COLOR (2 << 0) +# define RADEON_COLOR_ARG_A_CURRENT_ALPHA (3 << 0) +# define RADEON_COLOR_ARG_A_DIFFUSE_COLOR (4 << 0) +# define RADEON_COLOR_ARG_A_DIFFUSE_ALPHA (5 << 0) +# define RADEON_COLOR_ARG_A_SPECULAR_COLOR (6 << 0) +# define RADEON_COLOR_ARG_A_SPECULAR_ALPHA (7 << 0) +# define RADEON_COLOR_ARG_A_TFACTOR_COLOR (8 << 0) +# define RADEON_COLOR_ARG_A_TFACTOR_ALPHA (9 << 0) +# define RADEON_COLOR_ARG_A_T0_COLOR (10 << 0) +# define RADEON_COLOR_ARG_A_T0_ALPHA (11 << 0) +# define RADEON_COLOR_ARG_A_T1_COLOR (12 << 0) +# define RADEON_COLOR_ARG_A_T1_ALPHA (13 << 0) +# define RADEON_COLOR_ARG_A_T2_COLOR (14 << 0) +# define RADEON_COLOR_ARG_A_T2_ALPHA (15 << 0) +# define RADEON_COLOR_ARG_A_T3_COLOR (16 << 0) +# define RADEON_COLOR_ARG_A_T3_ALPHA (17 << 0) +# define RADEON_COLOR_ARG_B_SHIFT 5 +# define RADEON_COLOR_ARG_B_MASK (0x1f << 5) +# define RADEON_COLOR_ARG_B_ZERO (0 << 5) +# define RADEON_COLOR_ARG_B_CURRENT_COLOR (2 << 5) +# define RADEON_COLOR_ARG_B_CURRENT_ALPHA (3 << 5) +# define RADEON_COLOR_ARG_B_DIFFUSE_COLOR (4 << 5) +# define RADEON_COLOR_ARG_B_DIFFUSE_ALPHA (5 << 5) +# define RADEON_COLOR_ARG_B_SPECULAR_COLOR (6 << 5) +# define RADEON_COLOR_ARG_B_SPECULAR_ALPHA (7 << 5) +# define RADEON_COLOR_ARG_B_TFACTOR_COLOR (8 << 5) +# define RADEON_COLOR_ARG_B_TFACTOR_ALPHA (9 << 5) +# define RADEON_COLOR_ARG_B_T0_COLOR (10 << 5) +# define RADEON_COLOR_ARG_B_T0_ALPHA (11 << 5) +# define RADEON_COLOR_ARG_B_T1_COLOR (12 << 5) +# define RADEON_COLOR_ARG_B_T1_ALPHA (13 << 5) +# define RADEON_COLOR_ARG_B_T2_COLOR (14 << 5) +# define RADEON_COLOR_ARG_B_T2_ALPHA (15 << 5) +# define RADEON_COLOR_ARG_B_T3_COLOR (16 << 5) +# define RADEON_COLOR_ARG_B_T3_ALPHA (17 << 5) +# define RADEON_COLOR_ARG_C_SHIFT 10 +# define RADEON_COLOR_ARG_C_MASK (0x1f << 10) +# define RADEON_COLOR_ARG_C_ZERO (0 << 10) +# define RADEON_COLOR_ARG_C_CURRENT_COLOR (2 << 10) +# define RADEON_COLOR_ARG_C_CURRENT_ALPHA (3 << 10) +# define RADEON_COLOR_ARG_C_DIFFUSE_COLOR (4 << 10) +# define RADEON_COLOR_ARG_C_DIFFUSE_ALPHA (5 << 10) +# define RADEON_COLOR_ARG_C_SPECULAR_COLOR (6 << 10) +# define RADEON_COLOR_ARG_C_SPECULAR_ALPHA (7 << 10) +# define RADEON_COLOR_ARG_C_TFACTOR_COLOR (8 << 10) +# define RADEON_COLOR_ARG_C_TFACTOR_ALPHA (9 << 10) +# define RADEON_COLOR_ARG_C_T0_COLOR (10 << 10) +# define RADEON_COLOR_ARG_C_T0_ALPHA (11 << 10) +# define RADEON_COLOR_ARG_C_T1_COLOR (12 << 10) +# define RADEON_COLOR_ARG_C_T1_ALPHA (13 << 10) +# define RADEON_COLOR_ARG_C_T2_COLOR (14 << 10) +# define RADEON_COLOR_ARG_C_T2_ALPHA (15 << 10) +# define RADEON_COLOR_ARG_C_T3_COLOR (16 << 10) +# define RADEON_COLOR_ARG_C_T3_ALPHA (17 << 10) +# define RADEON_COMP_ARG_A (1 << 15) +# define RADEON_COMP_ARG_A_SHIFT 15 +# define RADEON_COMP_ARG_B (1 << 16) +# define RADEON_COMP_ARG_B_SHIFT 16 +# define RADEON_COMP_ARG_C (1 << 17) +# define RADEON_COMP_ARG_C_SHIFT 17 +# define RADEON_BLEND_CTL_MASK (7 << 18) +# define RADEON_BLEND_CTL_ADD (0 << 18) +# define RADEON_BLEND_CTL_SUBTRACT (1 << 18) +# define RADEON_BLEND_CTL_ADDSIGNED (2 << 18) +# define RADEON_BLEND_CTL_BLEND (3 << 18) +# define RADEON_BLEND_CTL_DOT3 (4 << 18) +# define RADEON_SCALE_SHIFT 21 +# define RADEON_SCALE_MASK (3 << 21) +# define RADEON_SCALE_1X (0 << 21) +# define RADEON_SCALE_2X (1 << 21) +# define RADEON_SCALE_4X (2 << 21) +# define RADEON_CLAMP_TX (1 << 23) +# define RADEON_T0_EQ_TCUR (1 << 24) +# define RADEON_T1_EQ_TCUR (1 << 25) +# define RADEON_T2_EQ_TCUR (1 << 26) +# define RADEON_T3_EQ_TCUR (1 << 27) +# define RADEON_COLOR_ARG_MASK 0x1f +# define RADEON_COMP_ARG_SHIFT 15 +#define RADEON_PP_TXABLEND_0 0x1c64 +#define RADEON_PP_TXABLEND_1 0x1c7c +#define RADEON_PP_TXABLEND_2 0x1c94 +# define RADEON_ALPHA_ARG_A_SHIFT 0 +# define RADEON_ALPHA_ARG_A_MASK (0xf << 0) +# define RADEON_ALPHA_ARG_A_ZERO (0 << 0) +# define RADEON_ALPHA_ARG_A_CURRENT_ALPHA (1 << 0) +# define RADEON_ALPHA_ARG_A_DIFFUSE_ALPHA (2 << 0) +# define RADEON_ALPHA_ARG_A_SPECULAR_ALPHA (3 << 0) +# define RADEON_ALPHA_ARG_A_TFACTOR_ALPHA (4 << 0) +# define RADEON_ALPHA_ARG_A_T0_ALPHA (5 << 0) +# define RADEON_ALPHA_ARG_A_T1_ALPHA (6 << 0) +# define RADEON_ALPHA_ARG_A_T2_ALPHA (7 << 0) +# define RADEON_ALPHA_ARG_A_T3_ALPHA (8 << 0) +# define RADEON_ALPHA_ARG_B_SHIFT 4 +# define RADEON_ALPHA_ARG_B_MASK (0xf << 4) +# define RADEON_ALPHA_ARG_B_ZERO (0 << 4) +# define RADEON_ALPHA_ARG_B_CURRENT_ALPHA (1 << 4) +# define RADEON_ALPHA_ARG_B_DIFFUSE_ALPHA (2 << 4) +# define RADEON_ALPHA_ARG_B_SPECULAR_ALPHA (3 << 4) +# define RADEON_ALPHA_ARG_B_TFACTOR_ALPHA (4 << 4) +# define RADEON_ALPHA_ARG_B_T0_ALPHA (5 << 4) +# define RADEON_ALPHA_ARG_B_T1_ALPHA (6 << 4) +# define RADEON_ALPHA_ARG_B_T2_ALPHA (7 << 4) +# define RADEON_ALPHA_ARG_B_T3_ALPHA (8 << 4) +# define RADEON_ALPHA_ARG_C_SHIFT 8 +# define RADEON_ALPHA_ARG_C_MASK (0xf << 8) +# define RADEON_ALPHA_ARG_C_ZERO (0 << 8) +# define RADEON_ALPHA_ARG_C_CURRENT_ALPHA (1 << 8) +# define RADEON_ALPHA_ARG_C_DIFFUSE_ALPHA (2 << 8) +# define RADEON_ALPHA_ARG_C_SPECULAR_ALPHA (3 << 8) +# define RADEON_ALPHA_ARG_C_TFACTOR_ALPHA (4 << 8) +# define RADEON_ALPHA_ARG_C_T0_ALPHA (5 << 8) +# define RADEON_ALPHA_ARG_C_T1_ALPHA (6 << 8) +# define RADEON_ALPHA_ARG_C_T2_ALPHA (7 << 8) +# define RADEON_ALPHA_ARG_C_T3_ALPHA (8 << 8) +# define RADEON_DOT_ALPHA_DONT_REPLICATE (1 << 9) +# define RADEON_ALPHA_ARG_MASK 0xf + +#define RADEON_PP_TFACTOR_0 0x1c68 +#define RADEON_PP_TFACTOR_1 0x1c80 +#define RADEON_PP_TFACTOR_2 0x1c98 + +#define RADEON_RB3D_BLENDCNTL 0x1c20 +# define RADEON_COMB_FCN_MASK (3 << 12) +# define RADEON_COMB_FCN_ADD_CLAMP (0 << 12) +# define RADEON_COMB_FCN_ADD_NOCLAMP (1 << 12) +# define RADEON_COMB_FCN_SUB_CLAMP (2 << 12) +# define RADEON_COMB_FCN_SUB_NOCLAMP (3 << 12) +# define RADEON_SRC_BLEND_GL_ZERO (32 << 16) +# define RADEON_SRC_BLEND_GL_ONE (33 << 16) +# define RADEON_SRC_BLEND_GL_SRC_COLOR (34 << 16) +# define RADEON_SRC_BLEND_GL_ONE_MINUS_SRC_COLOR (35 << 16) +# define RADEON_SRC_BLEND_GL_DST_COLOR (36 << 16) +# define RADEON_SRC_BLEND_GL_ONE_MINUS_DST_COLOR (37 << 16) +# define RADEON_SRC_BLEND_GL_SRC_ALPHA (38 << 16) +# define RADEON_SRC_BLEND_GL_ONE_MINUS_SRC_ALPHA (39 << 16) +# define RADEON_SRC_BLEND_GL_DST_ALPHA (40 << 16) +# define RADEON_SRC_BLEND_GL_ONE_MINUS_DST_ALPHA (41 << 16) +# define RADEON_SRC_BLEND_GL_SRC_ALPHA_SATURATE (42 << 16) +# define RADEON_SRC_BLEND_MASK (63 << 16) +# define RADEON_DST_BLEND_GL_ZERO (32 << 24) +# define RADEON_DST_BLEND_GL_ONE (33 << 24) +# define RADEON_DST_BLEND_GL_SRC_COLOR (34 << 24) +# define RADEON_DST_BLEND_GL_ONE_MINUS_SRC_COLOR (35 << 24) +# define RADEON_DST_BLEND_GL_DST_COLOR (36 << 24) +# define RADEON_DST_BLEND_GL_ONE_MINUS_DST_COLOR (37 << 24) +# define RADEON_DST_BLEND_GL_SRC_ALPHA (38 << 24) +# define RADEON_DST_BLEND_GL_ONE_MINUS_SRC_ALPHA (39 << 24) +# define RADEON_DST_BLEND_GL_DST_ALPHA (40 << 24) +# define RADEON_DST_BLEND_GL_ONE_MINUS_DST_ALPHA (41 << 24) +# define RADEON_DST_BLEND_MASK (63 << 24) +#define RADEON_RB3D_CNTL 0x1c3c +# define RADEON_ALPHA_BLEND_ENABLE (1 << 0) +# define RADEON_PLANE_MASK_ENABLE (1 << 1) +# define RADEON_DITHER_ENABLE (1 << 2) +# define RADEON_ROUND_ENABLE (1 << 3) +# define RADEON_SCALE_DITHER_ENABLE (1 << 4) +# define RADEON_DITHER_INIT (1 << 5) +# define RADEON_ROP_ENABLE (1 << 6) +# define RADEON_STENCIL_ENABLE (1 << 7) +# define RADEON_Z_ENABLE (1 << 8) +# define RADEON_DEPTHXY_OFFSET_ENABLE (1 << 9) +# define RADEON_RB3D_COLOR_FORMAT_SHIFT 10 + +# define RADEON_COLOR_FORMAT_ARGB1555 3 +# define RADEON_COLOR_FORMAT_RGB565 4 +# define RADEON_COLOR_FORMAT_ARGB8888 6 +# define RADEON_COLOR_FORMAT_RGB332 7 +# define RADEON_COLOR_FORMAT_Y8 8 +# define RADEON_COLOR_FORMAT_RGB8 9 +# define RADEON_COLOR_FORMAT_YUV422_VYUY 11 +# define RADEON_COLOR_FORMAT_YUV422_YVYU 12 +# define RADEON_COLOR_FORMAT_aYUV444 14 +# define RADEON_COLOR_FORMAT_ARGB4444 15 + +# define RADEON_CLRCMP_FLIP_ENABLE (1 << 14) +#define RADEON_RB3D_COLOROFFSET 0x1c40 +# define RADEON_COLOROFFSET_MASK 0xfffffff0 +#define RADEON_RB3D_COLORPITCH 0x1c48 +# define RADEON_COLORPITCH_MASK 0x000001ff8 +# define RADEON_COLOR_TILE_ENABLE (1 << 16) +# define RADEON_COLOR_MICROTILE_ENABLE (1 << 17) +# define RADEON_COLOR_ENDIAN_NO_SWAP (0 << 18) +# define RADEON_COLOR_ENDIAN_WORD_SWAP (1 << 18) +# define RADEON_COLOR_ENDIAN_DWORD_SWAP (2 << 18) +#define RADEON_RB3D_DEPTHOFFSET 0x1c24 +#define RADEON_RB3D_DEPTHPITCH 0x1c28 +# define RADEON_DEPTHPITCH_MASK 0x00001ff8 +# define RADEON_DEPTH_ENDIAN_NO_SWAP (0 << 18) +# define RADEON_DEPTH_ENDIAN_WORD_SWAP (1 << 18) +# define RADEON_DEPTH_ENDIAN_DWORD_SWAP (2 << 18) +#define RADEON_RB3D_PLANEMASK 0x1d84 +#define RADEON_RB3D_ROPCNTL 0x1d80 +# define RADEON_ROP_MASK (15 << 8) +# define RADEON_ROP_CLEAR (0 << 8) +# define RADEON_ROP_NOR (1 << 8) +# define RADEON_ROP_AND_INVERTED (2 << 8) +# define RADEON_ROP_COPY_INVERTED (3 << 8) +# define RADEON_ROP_AND_REVERSE (4 << 8) +# define RADEON_ROP_INVERT (5 << 8) +# define RADEON_ROP_XOR (6 << 8) +# define RADEON_ROP_NAND (7 << 8) +# define RADEON_ROP_AND (8 << 8) +# define RADEON_ROP_EQUIV (9 << 8) +# define RADEON_ROP_NOOP (10 << 8) +# define RADEON_ROP_OR_INVERTED (11 << 8) +# define RADEON_ROP_COPY (12 << 8) +# define RADEON_ROP_OR_REVERSE (13 << 8) +# define RADEON_ROP_OR (14 << 8) +# define RADEON_ROP_SET (15 << 8) +#define RADEON_RB3D_STENCILREFMASK 0x1d7c +# define RADEON_STENCIL_REF_SHIFT 0 +# define RADEON_STENCIL_REF_MASK (0xff << 0) +# define RADEON_STENCIL_MASK_SHIFT 16 +# define RADEON_STENCIL_VALUE_MASK (0xff << 16) +# define RADEON_STENCIL_WRITEMASK_SHIFT 24 +# define RADEON_STENCIL_WRITE_MASK (0xff << 24) +#define RADEON_RB3D_ZSTENCILCNTL 0x1c2c +# define RADEON_DEPTH_FORMAT_MASK (0xf << 0) +# define RADEON_DEPTH_FORMAT_16BIT_INT_Z (0 << 0) +# define RADEON_DEPTH_FORMAT_24BIT_INT_Z (2 << 0) +# define RADEON_DEPTH_FORMAT_24BIT_FLOAT_Z (3 << 0) +# define RADEON_DEPTH_FORMAT_32BIT_INT_Z (4 << 0) +# define RADEON_DEPTH_FORMAT_32BIT_FLOAT_Z (5 << 0) +# define RADEON_DEPTH_FORMAT_16BIT_FLOAT_W (7 << 0) +# define RADEON_DEPTH_FORMAT_24BIT_FLOAT_W (9 << 0) +# define RADEON_DEPTH_FORMAT_32BIT_FLOAT_W (11 << 0) +# define RADEON_Z_TEST_NEVER (0 << 4) +# define RADEON_Z_TEST_LESS (1 << 4) +# define RADEON_Z_TEST_LEQUAL (2 << 4) +# define RADEON_Z_TEST_EQUAL (3 << 4) +# define RADEON_Z_TEST_GEQUAL (4 << 4) +# define RADEON_Z_TEST_GREATER (5 << 4) +# define RADEON_Z_TEST_NEQUAL (6 << 4) +# define RADEON_Z_TEST_ALWAYS (7 << 4) +# define RADEON_Z_TEST_MASK (7 << 4) +# define RADEON_STENCIL_TEST_NEVER (0 << 12) +# define RADEON_STENCIL_TEST_LESS (1 << 12) +# define RADEON_STENCIL_TEST_LEQUAL (2 << 12) +# define RADEON_STENCIL_TEST_EQUAL (3 << 12) +# define RADEON_STENCIL_TEST_GEQUAL (4 << 12) +# define RADEON_STENCIL_TEST_GREATER (5 << 12) +# define RADEON_STENCIL_TEST_NEQUAL (6 << 12) +# define RADEON_STENCIL_TEST_ALWAYS (7 << 12) +# define RADEON_STENCIL_TEST_MASK (0x7 << 12) +# define RADEON_STENCIL_FAIL_KEEP (0 << 16) +# define RADEON_STENCIL_FAIL_ZERO (1 << 16) +# define RADEON_STENCIL_FAIL_REPLACE (2 << 16) +# define RADEON_STENCIL_FAIL_INC (3 << 16) +# define RADEON_STENCIL_FAIL_DEC (4 << 16) +# define RADEON_STENCIL_FAIL_INVERT (5 << 16) +# define RADEON_STENCIL_FAIL_MASK (0x7 << 16) +# define RADEON_STENCIL_ZPASS_KEEP (0 << 20) +# define RADEON_STENCIL_ZPASS_ZERO (1 << 20) +# define RADEON_STENCIL_ZPASS_REPLACE (2 << 20) +# define RADEON_STENCIL_ZPASS_INC (3 << 20) +# define RADEON_STENCIL_ZPASS_DEC (4 << 20) +# define RADEON_STENCIL_ZPASS_INVERT (5 << 20) +# define RADEON_STENCIL_ZPASS_MASK (0x7 << 20) +# define RADEON_STENCIL_ZFAIL_KEEP (0 << 24) +# define RADEON_STENCIL_ZFAIL_ZERO (1 << 24) +# define RADEON_STENCIL_ZFAIL_REPLACE (2 << 24) +# define RADEON_STENCIL_ZFAIL_INC (3 << 24) +# define RADEON_STENCIL_ZFAIL_DEC (4 << 24) +# define RADEON_STENCIL_ZFAIL_INVERT (5 << 24) +# define RADEON_STENCIL_ZFAIL_MASK (0x7 << 24) +# define RADEON_Z_COMPRESSION_ENABLE (1 << 28) +# define RADEON_FORCE_Z_DIRTY (1 << 29) +# define RADEON_Z_WRITE_ENABLE (1 << 30) +#define RADEON_RE_LINE_PATTERN 0x1cd0 +# define RADEON_LINE_PATTERN_MASK 0x0000ffff +# define RADEON_LINE_REPEAT_COUNT_SHIFT 16 +# define RADEON_LINE_PATTERN_START_SHIFT 24 +# define RADEON_LINE_PATTERN_LITTLE_BIT_ORDER (0 << 28) +# define RADEON_LINE_PATTERN_BIG_BIT_ORDER (1 << 28) +# define RADEON_LINE_PATTERN_AUTO_RESET (1 << 29) +#define RADEON_RE_LINE_STATE 0x1cd4 +# define RADEON_LINE_CURRENT_PTR_SHIFT 0 +# define RADEON_LINE_CURRENT_COUNT_SHIFT 8 +#define RADEON_RE_MISC 0x26c4 +# define RADEON_STIPPLE_COORD_MASK 0x1f +# define RADEON_STIPPLE_X_OFFSET_SHIFT 0 +# define RADEON_STIPPLE_X_OFFSET_MASK (0x1f << 0) +# define RADEON_STIPPLE_Y_OFFSET_SHIFT 8 +# define RADEON_STIPPLE_Y_OFFSET_MASK (0x1f << 8) +# define RADEON_STIPPLE_LITTLE_BIT_ORDER (0 << 16) +# define RADEON_STIPPLE_BIG_BIT_ORDER (1 << 16) +#define RADEON_RE_SOLID_COLOR 0x1c1c +#define RADEON_RE_TOP_LEFT 0x26c0 +# define RADEON_RE_LEFT_SHIFT 0 +# define RADEON_RE_TOP_SHIFT 16 +#define RADEON_RE_WIDTH_HEIGHT 0x1c44 +# define RADEON_RE_WIDTH_SHIFT 0 +# define RADEON_RE_HEIGHT_SHIFT 16 + +#define RADEON_RB3D_ZPASS_DATA 0x3290 +#define RADEON_RB3D_ZPASS_ADDR 0x3294 + +#define RADEON_SE_CNTL 0x1c4c +# define RADEON_FFACE_CULL_CW (0 << 0) +# define RADEON_FFACE_CULL_CCW (1 << 0) +# define RADEON_FFACE_CULL_DIR_MASK (1 << 0) +# define RADEON_BFACE_CULL (0 << 1) +# define RADEON_BFACE_SOLID (3 << 1) +# define RADEON_FFACE_CULL (0 << 3) +# define RADEON_FFACE_SOLID (3 << 3) +# define RADEON_FFACE_CULL_MASK (3 << 3) +# define RADEON_BADVTX_CULL_DISABLE (1 << 5) +# define RADEON_FLAT_SHADE_VTX_0 (0 << 6) +# define RADEON_FLAT_SHADE_VTX_1 (1 << 6) +# define RADEON_FLAT_SHADE_VTX_2 (2 << 6) +# define RADEON_FLAT_SHADE_VTX_LAST (3 << 6) +# define RADEON_DIFFUSE_SHADE_SOLID (0 << 8) +# define RADEON_DIFFUSE_SHADE_FLAT (1 << 8) +# define RADEON_DIFFUSE_SHADE_GOURAUD (2 << 8) +# define RADEON_DIFFUSE_SHADE_MASK (3 << 8) +# define RADEON_ALPHA_SHADE_SOLID (0 << 10) +# define RADEON_ALPHA_SHADE_FLAT (1 << 10) +# define RADEON_ALPHA_SHADE_GOURAUD (2 << 10) +# define RADEON_ALPHA_SHADE_MASK (3 << 10) +# define RADEON_SPECULAR_SHADE_SOLID (0 << 12) +# define RADEON_SPECULAR_SHADE_FLAT (1 << 12) +# define RADEON_SPECULAR_SHADE_GOURAUD (2 << 12) +# define RADEON_SPECULAR_SHADE_MASK (3 << 12) +# define RADEON_FOG_SHADE_SOLID (0 << 14) +# define RADEON_FOG_SHADE_FLAT (1 << 14) +# define RADEON_FOG_SHADE_GOURAUD (2 << 14) +# define RADEON_FOG_SHADE_MASK (3 << 14) +# define RADEON_ZBIAS_ENABLE_POINT (1 << 16) +# define RADEON_ZBIAS_ENABLE_LINE (1 << 17) +# define RADEON_ZBIAS_ENABLE_TRI (1 << 18) +# define RADEON_WIDELINE_ENABLE (1 << 20) +# define RADEON_VPORT_XY_XFORM_ENABLE (1 << 24) +# define RADEON_VPORT_Z_XFORM_ENABLE (1 << 25) +# define RADEON_VTX_PIX_CENTER_D3D (0 << 27) +# define RADEON_VTX_PIX_CENTER_OGL (1 << 27) +# define RADEON_ROUND_MODE_TRUNC (0 << 28) +# define RADEON_ROUND_MODE_ROUND (1 << 28) +# define RADEON_ROUND_MODE_ROUND_EVEN (2 << 28) +# define RADEON_ROUND_MODE_ROUND_ODD (3 << 28) +# define RADEON_ROUND_PREC_16TH_PIX (0 << 30) +# define RADEON_ROUND_PREC_8TH_PIX (1 << 30) +# define RADEON_ROUND_PREC_4TH_PIX (2 << 30) +# define RADEON_ROUND_PREC_HALF_PIX (3 << 30) +#define R200_RE_CNTL 0x1c50 +# define R200_STIPPLE_ENABLE 0x1 +# define R200_SCISSOR_ENABLE 0x2 +# define R200_PATTERN_ENABLE 0x4 +# define R200_PERSPECTIVE_ENABLE 0x8 +# define R200_POINT_SMOOTH 0x20 +# define R200_VTX_STQ0_D3D 0x00010000 +# define R200_VTX_STQ1_D3D 0x00040000 +# define R200_VTX_STQ2_D3D 0x00100000 +# define R200_VTX_STQ3_D3D 0x00400000 +# define R200_VTX_STQ4_D3D 0x01000000 +# define R200_VTX_STQ5_D3D 0x04000000 +#define RADEON_SE_CNTL_STATUS 0x2140 +# define RADEON_VC_NO_SWAP (0 << 0) +# define RADEON_VC_16BIT_SWAP (1 << 0) +# define RADEON_VC_32BIT_SWAP (2 << 0) +# define RADEON_VC_HALF_DWORD_SWAP (3 << 0) +# define RADEON_TCL_BYPASS (1 << 8) +#define RADEON_SE_COORD_FMT 0x1c50 +# define RADEON_VTX_XY_PRE_MULT_1_OVER_W0 (1 << 0) +# define RADEON_VTX_Z_PRE_MULT_1_OVER_W0 (1 << 1) +# define RADEON_VTX_ST0_NONPARAMETRIC (1 << 8) +# define RADEON_VTX_ST1_NONPARAMETRIC (1 << 9) +# define RADEON_VTX_ST2_NONPARAMETRIC (1 << 10) +# define RADEON_VTX_ST3_NONPARAMETRIC (1 << 11) +# define RADEON_VTX_W0_NORMALIZE (1 << 12) +# define RADEON_VTX_W0_IS_NOT_1_OVER_W0 (1 << 16) +# define RADEON_VTX_ST0_PRE_MULT_1_OVER_W0 (1 << 17) +# define RADEON_VTX_ST1_PRE_MULT_1_OVER_W0 (1 << 19) +# define RADEON_VTX_ST2_PRE_MULT_1_OVER_W0 (1 << 21) +# define RADEON_VTX_ST3_PRE_MULT_1_OVER_W0 (1 << 23) +# define RADEON_TEX1_W_ROUTING_USE_W0 (0 << 26) +# define RADEON_TEX1_W_ROUTING_USE_Q1 (1 << 26) +#define RADEON_SE_LINE_WIDTH 0x1db8 +#define RADEON_SE_TCL_LIGHT_MODEL_CTL 0x226c +# define RADEON_LIGHTING_ENABLE (1 << 0) +# define RADEON_LIGHT_IN_MODELSPACE (1 << 1) +# define RADEON_LOCAL_VIEWER (1 << 2) +# define RADEON_NORMALIZE_NORMALS (1 << 3) +# define RADEON_RESCALE_NORMALS (1 << 4) +# define RADEON_SPECULAR_LIGHTS (1 << 5) +# define RADEON_DIFFUSE_SPECULAR_COMBINE (1 << 6) +# define RADEON_LIGHT_ALPHA (1 << 7) +# define RADEON_LOCAL_LIGHT_VEC_GL (1 << 8) +# define RADEON_LIGHT_NO_NORMAL_AMBIENT_ONLY (1 << 9) +# define RADEON_LM_SOURCE_STATE_PREMULT 0 +# define RADEON_LM_SOURCE_STATE_MULT 1 +# define RADEON_LM_SOURCE_VERTEX_DIFFUSE 2 +# define RADEON_LM_SOURCE_VERTEX_SPECULAR 3 +# define RADEON_EMISSIVE_SOURCE_SHIFT 16 +# define RADEON_AMBIENT_SOURCE_SHIFT 18 +# define RADEON_DIFFUSE_SOURCE_SHIFT 20 +# define RADEON_SPECULAR_SOURCE_SHIFT 22 +#define RADEON_SE_TCL_MATERIAL_AMBIENT_RED 0x2220 +#define RADEON_SE_TCL_MATERIAL_AMBIENT_GREEN 0x2224 +#define RADEON_SE_TCL_MATERIAL_AMBIENT_BLUE 0x2228 +#define RADEON_SE_TCL_MATERIAL_AMBIENT_ALPHA 0x222c +#define RADEON_SE_TCL_MATERIAL_DIFFUSE_RED 0x2230 +#define RADEON_SE_TCL_MATERIAL_DIFFUSE_GREEN 0x2234 +#define RADEON_SE_TCL_MATERIAL_DIFFUSE_BLUE 0x2238 +#define RADEON_SE_TCL_MATERIAL_DIFFUSE_ALPHA 0x223c +#define RADEON_SE_TCL_MATERIAL_EMMISSIVE_RED 0x2210 +#define RADEON_SE_TCL_MATERIAL_EMMISSIVE_GREEN 0x2214 +#define RADEON_SE_TCL_MATERIAL_EMMISSIVE_BLUE 0x2218 +#define RADEON_SE_TCL_MATERIAL_EMMISSIVE_ALPHA 0x221c +#define RADEON_SE_TCL_MATERIAL_SPECULAR_RED 0x2240 +#define RADEON_SE_TCL_MATERIAL_SPECULAR_GREEN 0x2244 +#define RADEON_SE_TCL_MATERIAL_SPECULAR_BLUE 0x2248 +#define RADEON_SE_TCL_MATERIAL_SPECULAR_ALPHA 0x224c +#define RADEON_SE_TCL_MATRIX_SELECT_0 0x225c +# define RADEON_MODELVIEW_0_SHIFT 0 +# define RADEON_MODELVIEW_1_SHIFT 4 +# define RADEON_MODELVIEW_2_SHIFT 8 +# define RADEON_MODELVIEW_3_SHIFT 12 +# define RADEON_IT_MODELVIEW_0_SHIFT 16 +# define RADEON_IT_MODELVIEW_1_SHIFT 20 +# define RADEON_IT_MODELVIEW_2_SHIFT 24 +# define RADEON_IT_MODELVIEW_3_SHIFT 28 +#define RADEON_SE_TCL_MATRIX_SELECT_1 0x2260 +# define RADEON_MODELPROJECT_0_SHIFT 0 +# define RADEON_MODELPROJECT_1_SHIFT 4 +# define RADEON_MODELPROJECT_2_SHIFT 8 +# define RADEON_MODELPROJECT_3_SHIFT 12 +# define RADEON_TEXMAT_0_SHIFT 16 +# define RADEON_TEXMAT_1_SHIFT 20 +# define RADEON_TEXMAT_2_SHIFT 24 +# define RADEON_TEXMAT_3_SHIFT 28 + + +#define RADEON_SE_TCL_OUTPUT_VTX_FMT 0x2254 +# define RADEON_TCL_VTX_W0 (1 << 0) +# define RADEON_TCL_VTX_FP_DIFFUSE (1 << 1) +# define RADEON_TCL_VTX_FP_ALPHA (1 << 2) +# define RADEON_TCL_VTX_PK_DIFFUSE (1 << 3) +# define RADEON_TCL_VTX_FP_SPEC (1 << 4) +# define RADEON_TCL_VTX_FP_FOG (1 << 5) +# define RADEON_TCL_VTX_PK_SPEC (1 << 6) +# define RADEON_TCL_VTX_ST0 (1 << 7) +# define RADEON_TCL_VTX_ST1 (1 << 8) +# define RADEON_TCL_VTX_Q1 (1 << 9) +# define RADEON_TCL_VTX_ST2 (1 << 10) +# define RADEON_TCL_VTX_Q2 (1 << 11) +# define RADEON_TCL_VTX_ST3 (1 << 12) +# define RADEON_TCL_VTX_Q3 (1 << 13) +# define RADEON_TCL_VTX_Q0 (1 << 14) +# define RADEON_TCL_VTX_WEIGHT_COUNT_SHIFT 15 +# define RADEON_TCL_VTX_NORM0 (1 << 18) +# define RADEON_TCL_VTX_XY1 (1 << 27) +# define RADEON_TCL_VTX_Z1 (1 << 28) +# define RADEON_TCL_VTX_W1 (1 << 29) +# define RADEON_TCL_VTX_NORM1 (1 << 30) +# define RADEON_TCL_VTX_Z0 (1 << 31) + +#define RADEON_SE_TCL_OUTPUT_VTX_SEL 0x2258 +# define RADEON_TCL_COMPUTE_XYZW (1 << 0) +# define RADEON_TCL_COMPUTE_DIFFUSE (1 << 1) +# define RADEON_TCL_COMPUTE_SPECULAR (1 << 2) +# define RADEON_TCL_FORCE_NAN_IF_COLOR_NAN (1 << 3) +# define RADEON_TCL_FORCE_INORDER_PROC (1 << 4) +# define RADEON_TCL_TEX_INPUT_TEX_0 0 +# define RADEON_TCL_TEX_INPUT_TEX_1 1 +# define RADEON_TCL_TEX_INPUT_TEX_2 2 +# define RADEON_TCL_TEX_INPUT_TEX_3 3 +# define RADEON_TCL_TEX_COMPUTED_TEX_0 8 +# define RADEON_TCL_TEX_COMPUTED_TEX_1 9 +# define RADEON_TCL_TEX_COMPUTED_TEX_2 10 +# define RADEON_TCL_TEX_COMPUTED_TEX_3 11 +# define RADEON_TCL_TEX_0_OUTPUT_SHIFT 16 +# define RADEON_TCL_TEX_1_OUTPUT_SHIFT 20 +# define RADEON_TCL_TEX_2_OUTPUT_SHIFT 24 +# define RADEON_TCL_TEX_3_OUTPUT_SHIFT 28 + +#define RADEON_SE_TCL_PER_LIGHT_CTL_0 0x2270 +# define RADEON_LIGHT_0_ENABLE (1 << 0) +# define RADEON_LIGHT_0_ENABLE_AMBIENT (1 << 1) +# define RADEON_LIGHT_0_ENABLE_SPECULAR (1 << 2) +# define RADEON_LIGHT_0_IS_LOCAL (1 << 3) +# define RADEON_LIGHT_0_IS_SPOT (1 << 4) +# define RADEON_LIGHT_0_DUAL_CONE (1 << 5) +# define RADEON_LIGHT_0_ENABLE_RANGE_ATTEN (1 << 6) +# define RADEON_LIGHT_0_CONSTANT_RANGE_ATTEN (1 << 7) +# define RADEON_LIGHT_0_SHIFT 0 +# define RADEON_LIGHT_1_ENABLE (1 << 16) +# define RADEON_LIGHT_1_ENABLE_AMBIENT (1 << 17) +# define RADEON_LIGHT_1_ENABLE_SPECULAR (1 << 18) +# define RADEON_LIGHT_1_IS_LOCAL (1 << 19) +# define RADEON_LIGHT_1_IS_SPOT (1 << 20) +# define RADEON_LIGHT_1_DUAL_CONE (1 << 21) +# define RADEON_LIGHT_1_ENABLE_RANGE_ATTEN (1 << 22) +# define RADEON_LIGHT_1_CONSTANT_RANGE_ATTEN (1 << 23) +# define RADEON_LIGHT_1_SHIFT 16 +#define RADEON_SE_TCL_PER_LIGHT_CTL_1 0x2274 +# define RADEON_LIGHT_2_SHIFT 0 +# define RADEON_LIGHT_3_SHIFT 16 +#define RADEON_SE_TCL_PER_LIGHT_CTL_2 0x2278 +# define RADEON_LIGHT_4_SHIFT 0 +# define RADEON_LIGHT_5_SHIFT 16 +#define RADEON_SE_TCL_PER_LIGHT_CTL_3 0x227c +# define RADEON_LIGHT_6_SHIFT 0 +# define RADEON_LIGHT_7_SHIFT 16 + +#define RADEON_SE_TCL_SHININESS 0x2250 + +#define RADEON_SE_TCL_TEXTURE_PROC_CTL 0x2268 +# define RADEON_TEXGEN_TEXMAT_0_ENABLE (1 << 0) +# define RADEON_TEXGEN_TEXMAT_1_ENABLE (1 << 1) +# define RADEON_TEXGEN_TEXMAT_2_ENABLE (1 << 2) +# define RADEON_TEXGEN_TEXMAT_3_ENABLE (1 << 3) +# define RADEON_TEXMAT_0_ENABLE (1 << 4) +# define RADEON_TEXMAT_1_ENABLE (1 << 5) +# define RADEON_TEXMAT_2_ENABLE (1 << 6) +# define RADEON_TEXMAT_3_ENABLE (1 << 7) +# define RADEON_TEXGEN_INPUT_MASK 0xf +# define RADEON_TEXGEN_INPUT_TEXCOORD_0 0 +# define RADEON_TEXGEN_INPUT_TEXCOORD_1 1 +# define RADEON_TEXGEN_INPUT_TEXCOORD_2 2 +# define RADEON_TEXGEN_INPUT_TEXCOORD_3 3 +# define RADEON_TEXGEN_INPUT_OBJ 4 +# define RADEON_TEXGEN_INPUT_EYE 5 +# define RADEON_TEXGEN_INPUT_EYE_NORMAL 6 +# define RADEON_TEXGEN_INPUT_EYE_REFLECT 7 +# define RADEON_TEXGEN_INPUT_EYE_NORMALIZED 8 +# define RADEON_TEXGEN_0_INPUT_SHIFT 16 +# define RADEON_TEXGEN_1_INPUT_SHIFT 20 +# define RADEON_TEXGEN_2_INPUT_SHIFT 24 +# define RADEON_TEXGEN_3_INPUT_SHIFT 28 + +#define RADEON_SE_TCL_UCP_VERT_BLEND_CTL 0x2264 +# define RADEON_UCP_IN_CLIP_SPACE (1 << 0) +# define RADEON_UCP_IN_MODEL_SPACE (1 << 1) +# define RADEON_UCP_ENABLE_0 (1 << 2) +# define RADEON_UCP_ENABLE_1 (1 << 3) +# define RADEON_UCP_ENABLE_2 (1 << 4) +# define RADEON_UCP_ENABLE_3 (1 << 5) +# define RADEON_UCP_ENABLE_4 (1 << 6) +# define RADEON_UCP_ENABLE_5 (1 << 7) +# define RADEON_TCL_FOG_MASK (3 << 8) +# define RADEON_TCL_FOG_DISABLE (0 << 8) +# define RADEON_TCL_FOG_EXP (1 << 8) +# define RADEON_TCL_FOG_EXP2 (2 << 8) +# define RADEON_TCL_FOG_LINEAR (3 << 8) +# define RADEON_RNG_BASED_FOG (1 << 10) +# define RADEON_LIGHT_TWOSIDE (1 << 11) +# define RADEON_BLEND_OP_COUNT_MASK (7 << 12) +# define RADEON_BLEND_OP_COUNT_SHIFT 12 +# define RADEON_POSITION_BLEND_OP_ENABLE (1 << 16) +# define RADEON_NORMAL_BLEND_OP_ENABLE (1 << 17) +# define RADEON_VERTEX_BLEND_SRC_0_PRIMARY (1 << 18) +# define RADEON_VERTEX_BLEND_SRC_0_SECONDARY (1 << 18) +# define RADEON_VERTEX_BLEND_SRC_1_PRIMARY (1 << 19) +# define RADEON_VERTEX_BLEND_SRC_1_SECONDARY (1 << 19) +# define RADEON_VERTEX_BLEND_SRC_2_PRIMARY (1 << 20) +# define RADEON_VERTEX_BLEND_SRC_2_SECONDARY (1 << 20) +# define RADEON_VERTEX_BLEND_SRC_3_PRIMARY (1 << 21) +# define RADEON_VERTEX_BLEND_SRC_3_SECONDARY (1 << 21) +# define RADEON_VERTEX_BLEND_WGT_MINUS_ONE (1 << 22) +# define RADEON_CULL_FRONT_IS_CW (0 << 28) +# define RADEON_CULL_FRONT_IS_CCW (1 << 28) +# define RADEON_CULL_FRONT (1 << 29) +# define RADEON_CULL_BACK (1 << 30) +# define RADEON_FORCE_W_TO_ONE (1 << 31) + +#define RADEON_SE_VPORT_XSCALE 0x1d98 +#define RADEON_SE_VPORT_XOFFSET 0x1d9c +#define RADEON_SE_VPORT_YSCALE 0x1da0 +#define RADEON_SE_VPORT_YOFFSET 0x1da4 +#define RADEON_SE_VPORT_ZSCALE 0x1da8 +#define RADEON_SE_VPORT_ZOFFSET 0x1dac +#define RADEON_SE_ZBIAS_FACTOR 0x1db0 +#define RADEON_SE_ZBIAS_CONSTANT 0x1db4 + +#define RADEON_SE_VTX_FMT 0x2080 +# define RADEON_SE_VTX_FMT_XY 0x00000000 +# define RADEON_SE_VTX_FMT_W0 0x00000001 +# define RADEON_SE_VTX_FMT_FPCOLOR 0x00000002 +# define RADEON_SE_VTX_FMT_FPALPHA 0x00000004 +# define RADEON_SE_VTX_FMT_PKCOLOR 0x00000008 +# define RADEON_SE_VTX_FMT_FPSPEC 0x00000010 +# define RADEON_SE_VTX_FMT_FPFOG 0x00000020 +# define RADEON_SE_VTX_FMT_PKSPEC 0x00000040 +# define RADEON_SE_VTX_FMT_ST0 0x00000080 +# define RADEON_SE_VTX_FMT_ST1 0x00000100 +# define RADEON_SE_VTX_FMT_Q1 0x00000200 +# define RADEON_SE_VTX_FMT_ST2 0x00000400 +# define RADEON_SE_VTX_FMT_Q2 0x00000800 +# define RADEON_SE_VTX_FMT_ST3 0x00001000 +# define RADEON_SE_VTX_FMT_Q3 0x00002000 +# define RADEON_SE_VTX_FMT_Q0 0x00004000 +# define RADEON_SE_VTX_FMT_BLND_WEIGHT_CNT_MASK 0x00038000 +# define RADEON_SE_VTX_FMT_N0 0x00040000 +# define RADEON_SE_VTX_FMT_XY1 0x08000000 +# define RADEON_SE_VTX_FMT_Z1 0x10000000 +# define RADEON_SE_VTX_FMT_W1 0x20000000 +# define RADEON_SE_VTX_FMT_N1 0x40000000 +# define RADEON_SE_VTX_FMT_Z 0x80000000 + +#define RADEON_SE_VF_CNTL 0x2084 +# define RADEON_VF_PRIM_TYPE_POINT_LIST 1 +# define RADEON_VF_PRIM_TYPE_LINE_LIST 2 +# define RADEON_VF_PRIM_TYPE_LINE_STRIP 3 +# define RADEON_VF_PRIM_TYPE_TRIANGLE_LIST 4 +# define RADEON_VF_PRIM_TYPE_TRIANGLE_FAN 5 +# define RADEON_VF_PRIM_TYPE_TRIANGLE_STRIP 6 +# define RADEON_VF_PRIM_TYPE_TRIANGLE_FLAG 7 +# define RADEON_VF_PRIM_TYPE_RECTANGLE_LIST 8 +# define RADEON_VF_PRIM_TYPE_POINT_LIST_3 9 +# define RADEON_VF_PRIM_TYPE_LINE_LIST_3 10 +# define RADEON_VF_PRIM_TYPE_SPIRIT_LIST 11 +# define RADEON_VF_PRIM_TYPE_LINE_LOOP 12 +# define RADEON_VF_PRIM_TYPE_QUAD_LIST 13 +# define RADEON_VF_PRIM_TYPE_QUAD_STRIP 14 +# define RADEON_VF_PRIM_TYPE_POLYGON 15 +# define RADEON_VF_PRIM_WALK_STATE (0<<4) +# define RADEON_VF_PRIM_WALK_INDEX (1<<4) +# define RADEON_VF_PRIM_WALK_LIST (2<<4) +# define RADEON_VF_PRIM_WALK_DATA (3<<4) +# define RADEON_VF_COLOR_ORDER_RGBA (1<<6) +# define RADEON_VF_RADEON_MODE (1<<8) +# define RADEON_VF_TCL_OUTPUT_CTL_ENA (1<<9) +# define RADEON_VF_PROG_STREAM_ENA (1<<10) +# define RADEON_VF_INDEX_SIZE_SHIFT 11 +# define RADEON_VF_NUM_VERTICES_SHIFT 16 + +#define RADEON_SE_PORT_DATA0 0x2000 + +#define R200_SE_VAP_CNTL 0x2080 +# define R200_VAP_TCL_ENABLE 0x00000001 +# define R200_VAP_SINGLE_BUF_STATE_ENABLE 0x00000010 +# define R200_VAP_FORCE_W_TO_ONE 0x00010000 +# define R200_VAP_D3D_TEX_DEFAULT 0x00020000 +# define R200_VAP_VF_MAX_VTX_NUM__SHIFT 18 +# define R200_VAP_VF_MAX_VTX_NUM (9 << 18) +# define R200_VAP_DX_CLIP_SPACE_DEF 0x00400000 +#define R200_VF_MAX_VTX_INDX 0x210c +#define R200_VF_MIN_VTX_INDX 0x2110 +#define R200_SE_VTE_CNTL 0x20b0 +# define R200_VPORT_X_SCALE_ENA 0x00000001 +# define R200_VPORT_X_OFFSET_ENA 0x00000002 +# define R200_VPORT_Y_SCALE_ENA 0x00000004 +# define R200_VPORT_Y_OFFSET_ENA 0x00000008 +# define R200_VPORT_Z_SCALE_ENA 0x00000010 +# define R200_VPORT_Z_OFFSET_ENA 0x00000020 +# define R200_VTX_XY_FMT 0x00000100 +# define R200_VTX_Z_FMT 0x00000200 +# define R200_VTX_W0_FMT 0x00000400 +# define R200_VTX_W0_NORMALIZE 0x00000800 +# define R200_VTX_ST_DENORMALIZED 0x00001000 +#define R200_SE_VAP_CNTL_STATUS 0x2140 +# define R200_VC_NO_SWAP (0 << 0) +# define R200_VC_16BIT_SWAP (1 << 0) +# define R200_VC_32BIT_SWAP (2 << 0) +#define R200_PP_TXFILTER_0 0x2c00 +#define R200_PP_TXFILTER_1 0x2c20 +#define R200_PP_TXFILTER_2 0x2c40 +#define R200_PP_TXFILTER_3 0x2c60 +#define R200_PP_TXFILTER_4 0x2c80 +#define R200_PP_TXFILTER_5 0x2ca0 +# define R200_MAG_FILTER_NEAREST (0 << 0) +# define R200_MAG_FILTER_LINEAR (1 << 0) +# define R200_MAG_FILTER_MASK (1 << 0) +# define R200_MIN_FILTER_NEAREST (0 << 1) +# define R200_MIN_FILTER_LINEAR (1 << 1) +# define R200_MIN_FILTER_NEAREST_MIP_NEAREST (2 << 1) +# define R200_MIN_FILTER_NEAREST_MIP_LINEAR (3 << 1) +# define R200_MIN_FILTER_LINEAR_MIP_NEAREST (6 << 1) +# define R200_MIN_FILTER_LINEAR_MIP_LINEAR (7 << 1) +# define R200_MIN_FILTER_ANISO_NEAREST (8 << 1) +# define R200_MIN_FILTER_ANISO_LINEAR (9 << 1) +# define R200_MIN_FILTER_ANISO_NEAREST_MIP_NEAREST (10 << 1) +# define R200_MIN_FILTER_ANISO_NEAREST_MIP_LINEAR (11 << 1) +# define R200_MIN_FILTER_MASK (15 << 1) +# define R200_MAX_ANISO_1_TO_1 (0 << 5) +# define R200_MAX_ANISO_2_TO_1 (1 << 5) +# define R200_MAX_ANISO_4_TO_1 (2 << 5) +# define R200_MAX_ANISO_8_TO_1 (3 << 5) +# define R200_MAX_ANISO_16_TO_1 (4 << 5) +# define R200_MAX_ANISO_MASK (7 << 5) +# define R200_MAX_MIP_LEVEL_MASK (0x0f << 16) +# define R200_MAX_MIP_LEVEL_SHIFT 16 +# define R200_YUV_TO_RGB (1 << 20) +# define R200_YUV_TEMPERATURE_COOL (0 << 21) +# define R200_YUV_TEMPERATURE_HOT (1 << 21) +# define R200_YUV_TEMPERATURE_MASK (1 << 21) +# define R200_WRAPEN_S (1 << 22) +# define R200_CLAMP_S_WRAP (0 << 23) +# define R200_CLAMP_S_MIRROR (1 << 23) +# define R200_CLAMP_S_CLAMP_LAST (2 << 23) +# define R200_CLAMP_S_MIRROR_CLAMP_LAST (3 << 23) +# define R200_CLAMP_S_CLAMP_BORDER (4 << 23) +# define R200_CLAMP_S_MIRROR_CLAMP_BORDER (5 << 23) +# define R200_CLAMP_S_CLAMP_GL (6 << 23) +# define R200_CLAMP_S_MIRROR_CLAMP_GL (7 << 23) +# define R200_CLAMP_S_MASK (7 << 23) +# define R200_WRAPEN_T (1 << 26) +# define R200_CLAMP_T_WRAP (0 << 27) +# define R200_CLAMP_T_MIRROR (1 << 27) +# define R200_CLAMP_T_CLAMP_LAST (2 << 27) +# define R200_CLAMP_T_MIRROR_CLAMP_LAST (3 << 27) +# define R200_CLAMP_T_CLAMP_BORDER (4 << 27) +# define R200_CLAMP_T_MIRROR_CLAMP_BORDER (5 << 27) +# define R200_CLAMP_T_CLAMP_GL (6 << 27) +# define R200_CLAMP_T_MIRROR_CLAMP_GL (7 << 27) +# define R200_CLAMP_T_MASK (7 << 27) +# define R200_KILL_LT_ZERO (1 << 30) +# define R200_BORDER_MODE_OGL (0 << 31) +# define R200_BORDER_MODE_D3D (1 << 31) +#define R200_PP_TXFORMAT_0 0x2c04 +#define R200_PP_TXFORMAT_1 0x2c24 +#define R200_PP_TXFORMAT_2 0x2c44 +#define R200_PP_TXFORMAT_3 0x2c64 +#define R200_PP_TXFORMAT_4 0x2c84 +#define R200_PP_TXFORMAT_5 0x2ca4 +# define R200_TXFORMAT_I8 (0 << 0) +# define R200_TXFORMAT_AI88 (1 << 0) +# define R200_TXFORMAT_RGB332 (2 << 0) +# define R200_TXFORMAT_ARGB1555 (3 << 0) +# define R200_TXFORMAT_RGB565 (4 << 0) +# define R200_TXFORMAT_ARGB4444 (5 << 0) +# define R200_TXFORMAT_ARGB8888 (6 << 0) +# define R200_TXFORMAT_RGBA8888 (7 << 0) +# define R200_TXFORMAT_Y8 (8 << 0) +# define R200_TXFORMAT_AVYU4444 (9 << 0) +# define R200_TXFORMAT_VYUY422 (10 << 0) +# define R200_TXFORMAT_YVYU422 (11 << 0) +# define R200_TXFORMAT_DXT1 (12 << 0) +# define R200_TXFORMAT_DXT23 (14 << 0) +# define R200_TXFORMAT_DXT45 (15 << 0) +# define R200_TXFORMAT_DVDU88 (18 << 0) +# define R200_TXFORMAT_LDVDU655 (19 << 0) +# define R200_TXFORMAT_LDVDU8888 (20 << 0) +# define R200_TXFORMAT_GR1616 (21 << 0) +# define R200_TXFORMAT_ABGR8888 (22 << 0) +# define R200_TXFORMAT_BGR111110 (23 << 0) +# define R200_TXFORMAT_FORMAT_MASK (31 << 0) +# define R200_TXFORMAT_FORMAT_SHIFT 0 +# define R200_TXFORMAT_ALPHA_IN_MAP (1 << 6) +# define R200_TXFORMAT_NON_POWER2 (1 << 7) +# define R200_TXFORMAT_WIDTH_MASK (15 << 8) +# define R200_TXFORMAT_WIDTH_SHIFT 8 +# define R200_TXFORMAT_HEIGHT_MASK (15 << 12) +# define R200_TXFORMAT_HEIGHT_SHIFT 12 +# define R200_TXFORMAT_F5_WIDTH_MASK (15 << 16) /* cube face 5 */ +# define R200_TXFORMAT_F5_WIDTH_SHIFT 16 +# define R200_TXFORMAT_F5_HEIGHT_MASK (15 << 20) +# define R200_TXFORMAT_F5_HEIGHT_SHIFT 20 +# define R200_TXFORMAT_ST_ROUTE_STQ0 (0 << 24) +# define R200_TXFORMAT_ST_ROUTE_STQ1 (1 << 24) +# define R200_TXFORMAT_ST_ROUTE_STQ2 (2 << 24) +# define R200_TXFORMAT_ST_ROUTE_STQ3 (3 << 24) +# define R200_TXFORMAT_ST_ROUTE_STQ4 (4 << 24) +# define R200_TXFORMAT_ST_ROUTE_STQ5 (5 << 24) +# define R200_TXFORMAT_ST_ROUTE_MASK (7 << 24) +# define R200_TXFORMAT_ST_ROUTE_SHIFT 24 +# define R200_TXFORMAT_LOOKUP_DISABLE (1 << 27) +# define R200_TXFORMAT_ALPHA_MASK_ENABLE (1 << 28) +# define R200_TXFORMAT_CHROMA_KEY_ENABLE (1 << 29) +# define R200_TXFORMAT_CUBIC_MAP_ENABLE (1 << 30) +#define R200_PP_TXFORMAT_X_0 0x2c08 +#define R200_PP_TXFORMAT_X_1 0x2c28 +#define R200_PP_TXFORMAT_X_2 0x2c48 +#define R200_PP_TXFORMAT_X_3 0x2c68 +#define R200_PP_TXFORMAT_X_4 0x2c88 +#define R200_PP_TXFORMAT_X_5 0x2ca8 + +#define R200_PP_TXSIZE_0 0x2c0c /* NPOT only */ +#define R200_PP_TXSIZE_1 0x2c2c /* NPOT only */ +#define R200_PP_TXSIZE_2 0x2c4c /* NPOT only */ +#define R200_PP_TXSIZE_3 0x2c6c /* NPOT only */ +#define R200_PP_TXSIZE_4 0x2c8c /* NPOT only */ +#define R200_PP_TXSIZE_5 0x2cac /* NPOT only */ + +#define R200_PP_TXPITCH_0 0x2c10 /* NPOT only */ +#define R200_PP_TXPITCH_1 0x2c30 /* NPOT only */ +#define R200_PP_TXPITCH_2 0x2c50 /* NPOT only */ +#define R200_PP_TXPITCH_3 0x2c70 /* NPOT only */ +#define R200_PP_TXPITCH_4 0x2c90 /* NPOT only */ +#define R200_PP_TXPITCH_5 0x2cb0 /* NPOT only */ + +#define R200_PP_CUBIC_FACES_0 0x2c18 +#define R200_PP_CUBIC_FACES_1 0x2c38 +#define R200_PP_CUBIC_FACES_2 0x2c58 +#define R200_PP_CUBIC_FACES_3 0x2c78 +#define R200_PP_CUBIC_FACES_4 0x2c98 +#define R200_PP_CUBIC_FACES_5 0x2cb8 + +#define R200_PP_TXOFFSET_0 0x2d00 +# define R200_TXO_ENDIAN_NO_SWAP (0 << 0) +# define R200_TXO_ENDIAN_BYTE_SWAP (1 << 0) +# define R200_TXO_ENDIAN_WORD_SWAP (2 << 0) +# define R200_TXO_ENDIAN_HALFDW_SWAP (3 << 0) +# define R200_TXO_MACRO_LINEAR (0 << 2) +# define R200_TXO_MACRO_TILE (1 << 2) +# define R200_TXO_MICRO_LINEAR (0 << 3) +# define R200_TXO_MICRO_TILE (1 << 3) +# define R200_TXO_OFFSET_MASK 0xffffffe0 +# define R200_TXO_OFFSET_SHIFT 5 +#define R200_PP_CUBIC_OFFSET_F1_0 0x2d04 +#define R200_PP_CUBIC_OFFSET_F2_0 0x2d08 +#define R200_PP_CUBIC_OFFSET_F3_0 0x2d0c +#define R200_PP_CUBIC_OFFSET_F4_0 0x2d10 +#define R200_PP_CUBIC_OFFSET_F5_0 0x2d14 + +#define R200_PP_TXOFFSET_1 0x2d18 +#define R200_PP_CUBIC_OFFSET_F1_1 0x2d1c +#define R200_PP_CUBIC_OFFSET_F2_1 0x2d20 +#define R200_PP_CUBIC_OFFSET_F3_1 0x2d24 +#define R200_PP_CUBIC_OFFSET_F4_1 0x2d28 +#define R200_PP_CUBIC_OFFSET_F5_1 0x2d2c + +#define R200_PP_TXOFFSET_2 0x2d30 +#define R200_PP_CUBIC_OFFSET_F1_2 0x2d34 +#define R200_PP_CUBIC_OFFSET_F2_2 0x2d38 +#define R200_PP_CUBIC_OFFSET_F3_2 0x2d3c +#define R200_PP_CUBIC_OFFSET_F4_2 0x2d40 +#define R200_PP_CUBIC_OFFSET_F5_2 0x2d44 + +#define R200_PP_TXOFFSET_3 0x2d48 +#define R200_PP_CUBIC_OFFSET_F1_3 0x2d4c +#define R200_PP_CUBIC_OFFSET_F2_3 0x2d50 +#define R200_PP_CUBIC_OFFSET_F3_3 0x2d54 +#define R200_PP_CUBIC_OFFSET_F4_3 0x2d58 +#define R200_PP_CUBIC_OFFSET_F5_3 0x2d5c +#define R200_PP_TXOFFSET_4 0x2d60 +#define R200_PP_CUBIC_OFFSET_F1_4 0x2d64 +#define R200_PP_CUBIC_OFFSET_F2_4 0x2d68 +#define R200_PP_CUBIC_OFFSET_F3_4 0x2d6c +#define R200_PP_CUBIC_OFFSET_F4_4 0x2d70 +#define R200_PP_CUBIC_OFFSET_F5_4 0x2d74 +#define R200_PP_TXOFFSET_5 0x2d78 +#define R200_PP_CUBIC_OFFSET_F1_5 0x2d7c +#define R200_PP_CUBIC_OFFSET_F2_5 0x2d80 +#define R200_PP_CUBIC_OFFSET_F3_5 0x2d84 +#define R200_PP_CUBIC_OFFSET_F4_5 0x2d88 +#define R200_PP_CUBIC_OFFSET_F5_5 0x2d8c + +#define R200_PP_TFACTOR_0 0x2ee0 +#define R200_PP_TFACTOR_1 0x2ee4 +#define R200_PP_TFACTOR_2 0x2ee8 +#define R200_PP_TFACTOR_3 0x2eec +#define R200_PP_TFACTOR_4 0x2ef0 +#define R200_PP_TFACTOR_5 0x2ef4 + +#define R200_PP_TXCBLEND_0 0x2f00 +# define R200_TXC_ARG_A_ZERO (0) +# define R200_TXC_ARG_A_CURRENT_COLOR (2) +# define R200_TXC_ARG_A_CURRENT_ALPHA (3) +# define R200_TXC_ARG_A_DIFFUSE_COLOR (4) +# define R200_TXC_ARG_A_DIFFUSE_ALPHA (5) +# define R200_TXC_ARG_A_SPECULAR_COLOR (6) +# define R200_TXC_ARG_A_SPECULAR_ALPHA (7) +# define R200_TXC_ARG_A_TFACTOR_COLOR (8) +# define R200_TXC_ARG_A_TFACTOR_ALPHA (9) +# define R200_TXC_ARG_A_R0_COLOR (10) +# define R200_TXC_ARG_A_R0_ALPHA (11) +# define R200_TXC_ARG_A_R1_COLOR (12) +# define R200_TXC_ARG_A_R1_ALPHA (13) +# define R200_TXC_ARG_A_R2_COLOR (14) +# define R200_TXC_ARG_A_R2_ALPHA (15) +# define R200_TXC_ARG_A_R3_COLOR (16) +# define R200_TXC_ARG_A_R3_ALPHA (17) +# define R200_TXC_ARG_A_R4_COLOR (18) +# define R200_TXC_ARG_A_R4_ALPHA (19) +# define R200_TXC_ARG_A_R5_COLOR (20) +# define R200_TXC_ARG_A_R5_ALPHA (21) +# define R200_TXC_ARG_A_TFACTOR1_COLOR (26) +# define R200_TXC_ARG_A_TFACTOR1_ALPHA (27) +# define R200_TXC_ARG_A_MASK (31 << 0) +# define R200_TXC_ARG_A_SHIFT 0 +# define R200_TXC_ARG_B_ZERO (0 << 5) +# define R200_TXC_ARG_B_CURRENT_COLOR (2 << 5) +# define R200_TXC_ARG_B_CURRENT_ALPHA (3 << 5) +# define R200_TXC_ARG_B_DIFFUSE_COLOR (4 << 5) +# define R200_TXC_ARG_B_DIFFUSE_ALPHA (5 << 5) +# define R200_TXC_ARG_B_SPECULAR_COLOR (6 << 5) +# define R200_TXC_ARG_B_SPECULAR_ALPHA (7 << 5) +# define R200_TXC_ARG_B_TFACTOR_COLOR (8 << 5) +# define R200_TXC_ARG_B_TFACTOR_ALPHA (9 << 5) +# define R200_TXC_ARG_B_R0_COLOR (10 << 5) +# define R200_TXC_ARG_B_R0_ALPHA (11 << 5) +# define R200_TXC_ARG_B_R1_COLOR (12 << 5) +# define R200_TXC_ARG_B_R1_ALPHA (13 << 5) +# define R200_TXC_ARG_B_R2_COLOR (14 << 5) +# define R200_TXC_ARG_B_R2_ALPHA (15 << 5) +# define R200_TXC_ARG_B_R3_COLOR (16 << 5) +# define R200_TXC_ARG_B_R3_ALPHA (17 << 5) +# define R200_TXC_ARG_B_R4_COLOR (18 << 5) +# define R200_TXC_ARG_B_R4_ALPHA (19 << 5) +# define R200_TXC_ARG_B_R5_COLOR (20 << 5) +# define R200_TXC_ARG_B_R5_ALPHA (21 << 5) +# define R200_TXC_ARG_B_TFACTOR1_COLOR (26 << 5) +# define R200_TXC_ARG_B_TFACTOR1_ALPHA (27 << 5) +# define R200_TXC_ARG_B_MASK (31 << 5) +# define R200_TXC_ARG_B_SHIFT 5 +# define R200_TXC_ARG_C_ZERO (0 << 10) +# define R200_TXC_ARG_C_CURRENT_COLOR (2 << 10) +# define R200_TXC_ARG_C_CURRENT_ALPHA (3 << 10) +# define R200_TXC_ARG_C_DIFFUSE_COLOR (4 << 10) +# define R200_TXC_ARG_C_DIFFUSE_ALPHA (5 << 10) +# define R200_TXC_ARG_C_SPECULAR_COLOR (6 << 10) +# define R200_TXC_ARG_C_SPECULAR_ALPHA (7 << 10) +# define R200_TXC_ARG_C_TFACTOR_COLOR (8 << 10) +# define R200_TXC_ARG_C_TFACTOR_ALPHA (9 << 10) +# define R200_TXC_ARG_C_R0_COLOR (10 << 10) +# define R200_TXC_ARG_C_R0_ALPHA (11 << 10) +# define R200_TXC_ARG_C_R1_COLOR (12 << 10) +# define R200_TXC_ARG_C_R1_ALPHA (13 << 10) +# define R200_TXC_ARG_C_R2_COLOR (14 << 10) +# define R200_TXC_ARG_C_R2_ALPHA (15 << 10) +# define R200_TXC_ARG_C_R3_COLOR (16 << 10) +# define R200_TXC_ARG_C_R3_ALPHA (17 << 10) +# define R200_TXC_ARG_C_R4_COLOR (18 << 10) +# define R200_TXC_ARG_C_R4_ALPHA (19 << 10) +# define R200_TXC_ARG_C_R5_COLOR (20 << 10) +# define R200_TXC_ARG_C_R5_ALPHA (21 << 10) +# define R200_TXC_ARG_C_TFACTOR1_COLOR (26 << 10) +# define R200_TXC_ARG_C_TFACTOR1_ALPHA (27 << 10) +# define R200_TXC_ARG_C_MASK (31 << 10) +# define R200_TXC_ARG_C_SHIFT 10 +# define R200_TXC_COMP_ARG_A (1 << 16) +# define R200_TXC_COMP_ARG_A_SHIFT (16) +# define R200_TXC_BIAS_ARG_A (1 << 17) +# define R200_TXC_SCALE_ARG_A (1 << 18) +# define R200_TXC_NEG_ARG_A (1 << 19) +# define R200_TXC_COMP_ARG_B (1 << 20) +# define R200_TXC_COMP_ARG_B_SHIFT (20) +# define R200_TXC_BIAS_ARG_B (1 << 21) +# define R200_TXC_SCALE_ARG_B (1 << 22) +# define R200_TXC_NEG_ARG_B (1 << 23) +# define R200_TXC_COMP_ARG_C (1 << 24) +# define R200_TXC_COMP_ARG_C_SHIFT (24) +# define R200_TXC_BIAS_ARG_C (1 << 25) +# define R200_TXC_SCALE_ARG_C (1 << 26) +# define R200_TXC_NEG_ARG_C (1 << 27) +# define R200_TXC_OP_MADD (0 << 28) +# define R200_TXC_OP_CND0 (2 << 28) +# define R200_TXC_OP_LERP (3 << 28) +# define R200_TXC_OP_DOT3 (4 << 28) +# define R200_TXC_OP_DOT4 (5 << 28) +# define R200_TXC_OP_CONDITIONAL (6 << 28) +# define R200_TXC_OP_DOT2_ADD (7 << 28) +# define R200_TXC_OP_MASK (7 << 28) +#define R200_PP_TXCBLEND2_0 0x2f04 +# define R200_TXC_TFACTOR_SEL_SHIFT 0 +# define R200_TXC_TFACTOR_SEL_MASK 0x7 +# define R200_TXC_TFACTOR1_SEL_SHIFT 4 +# define R200_TXC_TFACTOR1_SEL_MASK (0x7 << 4) +# define R200_TXC_SCALE_SHIFT 8 +# define R200_TXC_SCALE_MASK (7 << 8) +# define R200_TXC_SCALE_1X (0 << 8) +# define R200_TXC_SCALE_2X (1 << 8) +# define R200_TXC_SCALE_4X (2 << 8) +# define R200_TXC_SCALE_8X (3 << 8) +# define R200_TXC_SCALE_INV2 (5 << 8) +# define R200_TXC_SCALE_INV4 (6 << 8) +# define R200_TXC_SCALE_INV8 (7 << 8) +# define R200_TXC_CLAMP_SHIFT 12 +# define R200_TXC_CLAMP_MASK (3 << 12) +# define R200_TXC_CLAMP_WRAP (0 << 12) +# define R200_TXC_CLAMP_0_1 (1 << 12) +# define R200_TXC_CLAMP_8_8 (2 << 12) +# define R200_TXC_OUTPUT_REG_MASK (7 << 16) +# define R200_TXC_OUTPUT_REG_NONE (0 << 16) +# define R200_TXC_OUTPUT_REG_R0 (1 << 16) +# define R200_TXC_OUTPUT_REG_R1 (2 << 16) +# define R200_TXC_OUTPUT_REG_R2 (3 << 16) +# define R200_TXC_OUTPUT_REG_R3 (4 << 16) +# define R200_TXC_OUTPUT_REG_R4 (5 << 16) +# define R200_TXC_OUTPUT_REG_R5 (6 << 16) +# define R200_TXC_OUTPUT_MASK_MASK (7 << 20) +# define R200_TXC_OUTPUT_MASK_RGB (0 << 20) +# define R200_TXC_OUTPUT_MASK_RG (1 << 20) +# define R200_TXC_OUTPUT_MASK_RB (2 << 20) +# define R200_TXC_OUTPUT_MASK_R (3 << 20) +# define R200_TXC_OUTPUT_MASK_GB (4 << 20) +# define R200_TXC_OUTPUT_MASK_G (5 << 20) +# define R200_TXC_OUTPUT_MASK_B (6 << 20) +# define R200_TXC_OUTPUT_MASK_NONE (7 << 20) +# define R200_TXC_REPL_NORMAL 0 +# define R200_TXC_REPL_RED 1 +# define R200_TXC_REPL_GREEN 2 +# define R200_TXC_REPL_BLUE 3 +# define R200_TXC_REPL_ARG_A_SHIFT 26 +# define R200_TXC_REPL_ARG_A_MASK (3 << 26) +# define R200_TXC_REPL_ARG_B_SHIFT 28 +# define R200_TXC_REPL_ARG_B_MASK (3 << 28) +# define R200_TXC_REPL_ARG_C_SHIFT 30 +# define R200_TXC_REPL_ARG_C_MASK (3 << 30) +#define R200_PP_TXABLEND_0 0x2f08 +# define R200_TXA_ARG_A_ZERO (0) +# define R200_TXA_ARG_A_CURRENT_ALPHA (2) /* guess */ +# define R200_TXA_ARG_A_CURRENT_BLUE (3) /* guess */ +# define R200_TXA_ARG_A_DIFFUSE_ALPHA (4) +# define R200_TXA_ARG_A_DIFFUSE_BLUE (5) +# define R200_TXA_ARG_A_SPECULAR_ALPHA (6) +# define R200_TXA_ARG_A_SPECULAR_BLUE (7) +# define R200_TXA_ARG_A_TFACTOR_ALPHA (8) +# define R200_TXA_ARG_A_TFACTOR_BLUE (9) +# define R200_TXA_ARG_A_R0_ALPHA (10) +# define R200_TXA_ARG_A_R0_BLUE (11) +# define R200_TXA_ARG_A_R1_ALPHA (12) +# define R200_TXA_ARG_A_R1_BLUE (13) +# define R200_TXA_ARG_A_R2_ALPHA (14) +# define R200_TXA_ARG_A_R2_BLUE (15) +# define R200_TXA_ARG_A_R3_ALPHA (16) +# define R200_TXA_ARG_A_R3_BLUE (17) +# define R200_TXA_ARG_A_R4_ALPHA (18) +# define R200_TXA_ARG_A_R4_BLUE (19) +# define R200_TXA_ARG_A_R5_ALPHA (20) +# define R200_TXA_ARG_A_R5_BLUE (21) +# define R200_TXA_ARG_A_TFACTOR1_ALPHA (26) +# define R200_TXA_ARG_A_TFACTOR1_BLUE (27) +# define R200_TXA_ARG_A_MASK (31 << 0) +# define R200_TXA_ARG_A_SHIFT 0 +# define R200_TXA_ARG_B_ZERO (0 << 5) +# define R200_TXA_ARG_B_CURRENT_ALPHA (2 << 5) /* guess */ +# define R200_TXA_ARG_B_CURRENT_BLUE (3 << 5) /* guess */ +# define R200_TXA_ARG_B_DIFFUSE_ALPHA (4 << 5) +# define R200_TXA_ARG_B_DIFFUSE_BLUE (5 << 5) +# define R200_TXA_ARG_B_SPECULAR_ALPHA (6 << 5) +# define R200_TXA_ARG_B_SPECULAR_BLUE (7 << 5) +# define R200_TXA_ARG_B_TFACTOR_ALPHA (8 << 5) +# define R200_TXA_ARG_B_TFACTOR_BLUE (9 << 5) +# define R200_TXA_ARG_B_R0_ALPHA (10 << 5) +# define R200_TXA_ARG_B_R0_BLUE (11 << 5) +# define R200_TXA_ARG_B_R1_ALPHA (12 << 5) +# define R200_TXA_ARG_B_R1_BLUE (13 << 5) +# define R200_TXA_ARG_B_R2_ALPHA (14 << 5) +# define R200_TXA_ARG_B_R2_BLUE (15 << 5) +# define R200_TXA_ARG_B_R3_ALPHA (16 << 5) +# define R200_TXA_ARG_B_R3_BLUE (17 << 5) +# define R200_TXA_ARG_B_R4_ALPHA (18 << 5) +# define R200_TXA_ARG_B_R4_BLUE (19 << 5) +# define R200_TXA_ARG_B_R5_ALPHA (20 << 5) +# define R200_TXA_ARG_B_R5_BLUE (21 << 5) +# define R200_TXA_ARG_B_TFACTOR1_ALPHA (26 << 5) +# define R200_TXA_ARG_B_TFACTOR1_BLUE (27 << 5) +# define R200_TXA_ARG_B_MASK (31 << 5) +# define R200_TXA_ARG_B_SHIFT 5 +# define R200_TXA_ARG_C_ZERO (0 << 10) +# define R200_TXA_ARG_C_CURRENT_ALPHA (2 << 10) /* guess */ +# define R200_TXA_ARG_C_CURRENT_BLUE (3 << 10) /* guess */ +# define R200_TXA_ARG_C_DIFFUSE_ALPHA (4 << 10) +# define R200_TXA_ARG_C_DIFFUSE_BLUE (5 << 10) +# define R200_TXA_ARG_C_SPECULAR_ALPHA (6 << 10) +# define R200_TXA_ARG_C_SPECULAR_BLUE (7 << 10) +# define R200_TXA_ARG_C_TFACTOR_ALPHA (8 << 10) +# define R200_TXA_ARG_C_TFACTOR_BLUE (9 << 10) +# define R200_TXA_ARG_C_R0_ALPHA (10 << 10) +# define R200_TXA_ARG_C_R0_BLUE (11 << 10) +# define R200_TXA_ARG_C_R1_ALPHA (12 << 10) +# define R200_TXA_ARG_C_R1_BLUE (13 << 10) +# define R200_TXA_ARG_C_R2_ALPHA (14 << 10) +# define R200_TXA_ARG_C_R2_BLUE (15 << 10) +# define R200_TXA_ARG_C_R3_ALPHA (16 << 10) +# define R200_TXA_ARG_C_R3_BLUE (17 << 10) +# define R200_TXA_ARG_C_R4_ALPHA (18 << 10) +# define R200_TXA_ARG_C_R4_BLUE (19 << 10) +# define R200_TXA_ARG_C_R5_ALPHA (20 << 10) +# define R200_TXA_ARG_C_R5_BLUE (21 << 10) +# define R200_TXA_ARG_C_TFACTOR1_ALPHA (26 << 10) +# define R200_TXA_ARG_C_TFACTOR1_BLUE (27 << 10) +# define R200_TXA_ARG_C_MASK (31 << 10) +# define R200_TXA_ARG_C_SHIFT 10 +# define R200_TXA_COMP_ARG_A (1 << 16) +# define R200_TXA_COMP_ARG_A_SHIFT (16) +# define R200_TXA_BIAS_ARG_A (1 << 17) +# define R200_TXA_SCALE_ARG_A (1 << 18) +# define R200_TXA_NEG_ARG_A (1 << 19) +# define R200_TXA_COMP_ARG_B (1 << 20) +# define R200_TXA_COMP_ARG_B_SHIFT (20) +# define R200_TXA_BIAS_ARG_B (1 << 21) +# define R200_TXA_SCALE_ARG_B (1 << 22) +# define R200_TXA_NEG_ARG_B (1 << 23) +# define R200_TXA_COMP_ARG_C (1 << 24) +# define R200_TXA_COMP_ARG_C_SHIFT (24) +# define R200_TXA_BIAS_ARG_C (1 << 25) +# define R200_TXA_SCALE_ARG_C (1 << 26) +# define R200_TXA_NEG_ARG_C (1 << 27) +# define R200_TXA_OP_MADD (0 << 28) +# define R200_TXA_OP_CND0 (2 << 28) +# define R200_TXA_OP_LERP (3 << 28) +# define R200_TXA_OP_CONDITIONAL (6 << 28) +# define R200_TXA_OP_MASK (7 << 28) +#define R200_PP_TXABLEND2_0 0x2f0c +# define R200_TXA_TFACTOR_SEL_SHIFT 0 +# define R200_TXA_TFACTOR_SEL_MASK 0x7 +# define R200_TXA_TFACTOR1_SEL_SHIFT 4 +# define R200_TXA_TFACTOR1_SEL_MASK (0x7 << 4) +# define R200_TXA_SCALE_SHIFT 8 +# define R200_TXA_SCALE_MASK (7 << 8) +# define R200_TXA_SCALE_1X (0 << 8) +# define R200_TXA_SCALE_2X (1 << 8) +# define R200_TXA_SCALE_4X (2 << 8) +# define R200_TXA_SCALE_8X (3 << 8) +# define R200_TXA_SCALE_INV2 (5 << 8) +# define R200_TXA_SCALE_INV4 (6 << 8) +# define R200_TXA_SCALE_INV8 (7 << 8) +# define R200_TXA_CLAMP_SHIFT 12 +# define R200_TXA_CLAMP_MASK (3 << 12) +# define R200_TXA_CLAMP_WRAP (0 << 12) +# define R200_TXA_CLAMP_0_1 (1 << 12) +# define R200_TXA_CLAMP_8_8 (2 << 12) +# define R200_TXA_OUTPUT_REG_MASK (7 << 16) +# define R200_TXA_OUTPUT_REG_NONE (0 << 16) +# define R200_TXA_OUTPUT_REG_R0 (1 << 16) +# define R200_TXA_OUTPUT_REG_R1 (2 << 16) +# define R200_TXA_OUTPUT_REG_R2 (3 << 16) +# define R200_TXA_OUTPUT_REG_R3 (4 << 16) +# define R200_TXA_OUTPUT_REG_R4 (5 << 16) +# define R200_TXA_OUTPUT_REG_R5 (6 << 16) +# define R200_TXA_DOT_ALPHA (1 << 20) +# define R200_TXA_REPL_NORMAL 0 +# define R200_TXA_REPL_RED 1 +# define R200_TXA_REPL_GREEN 2 +# define R200_TXA_REPL_ARG_A_SHIFT 26 +# define R200_TXA_REPL_ARG_A_MASK (3 << 26) +# define R200_TXA_REPL_ARG_B_SHIFT 28 +# define R200_TXA_REPL_ARG_B_MASK (3 << 28) +# define R200_TXA_REPL_ARG_C_SHIFT 30 +# define R200_TXA_REPL_ARG_C_MASK (3 << 30) + +#define R200_SE_VTX_FMT_0 0x2088 +# define R200_VTX_XY 0 /* always have xy */ +# define R200_VTX_Z0 (1<<0) +# define R200_VTX_W0 (1<<1) +# define R200_VTX_WEIGHT_COUNT_SHIFT (2) +# define R200_VTX_PV_MATRIX_SEL (1<<5) +# define R200_VTX_N0 (1<<6) +# define R200_VTX_POINT_SIZE (1<<7) +# define R200_VTX_DISCRETE_FOG (1<<8) +# define R200_VTX_SHININESS_0 (1<<9) +# define R200_VTX_SHININESS_1 (1<<10) +# define R200_VTX_COLOR_NOT_PRESENT 0 +# define R200_VTX_PK_RGBA 1 +# define R200_VTX_FP_RGB 2 +# define R200_VTX_FP_RGBA 3 +# define R200_VTX_COLOR_MASK 3 +# define R200_VTX_COLOR_0_SHIFT 11 +# define R200_VTX_COLOR_1_SHIFT 13 +# define R200_VTX_COLOR_2_SHIFT 15 +# define R200_VTX_COLOR_3_SHIFT 17 +# define R200_VTX_COLOR_4_SHIFT 19 +# define R200_VTX_COLOR_5_SHIFT 21 +# define R200_VTX_COLOR_6_SHIFT 23 +# define R200_VTX_COLOR_7_SHIFT 25 +# define R200_VTX_XY1 (1<<28) +# define R200_VTX_Z1 (1<<29) +# define R200_VTX_W1 (1<<30) +# define R200_VTX_N1 (1<<31) +#define R200_SE_VTX_FMT_1 0x208c +# define R200_VTX_TEX0_COMP_CNT_SHIFT 0 +# define R200_VTX_TEX1_COMP_CNT_SHIFT 3 +# define R200_VTX_TEX2_COMP_CNT_SHIFT 6 +# define R200_VTX_TEX3_COMP_CNT_SHIFT 9 +# define R200_VTX_TEX4_COMP_CNT_SHIFT 12 +# define R200_VTX_TEX5_COMP_CNT_SHIFT 15 + +#define R200_SE_TCL_OUTPUT_VTX_FMT_0 0x2090 +#define R200_SE_TCL_OUTPUT_VTX_FMT_1 0x2094 +#define R200_SE_TCL_OUTPUT_VTX_COMP_SEL 0x2250 +# define R200_OUTPUT_XYZW (1<<0) +# define R200_OUTPUT_COLOR_0 (1<<8) +# define R200_OUTPUT_COLOR_1 (1<<9) +# define R200_OUTPUT_TEX_0 (1<<16) +# define R200_OUTPUT_TEX_1 (1<<17) +# define R200_OUTPUT_TEX_2 (1<<18) +# define R200_OUTPUT_TEX_3 (1<<19) +# define R200_OUTPUT_TEX_4 (1<<20) +# define R200_OUTPUT_TEX_5 (1<<21) +# define R200_OUTPUT_TEX_MASK (0x3f<<16) +# define R200_OUTPUT_DISCRETE_FOG (1<<24) +# define R200_OUTPUT_PT_SIZE (1<<25) +# define R200_FORCE_INORDER_PROC (1<<31) +#define R200_PP_CNTL_X 0x2cc4 +#define R200_PP_TXMULTI_CTL_0 0x2c1c +#define R200_PP_TXMULTI_CTL_1 0x2c3c +#define R200_PP_TXMULTI_CTL_2 0x2c5c +#define R200_PP_TXMULTI_CTL_3 0x2c7c +#define R200_PP_TXMULTI_CTL_4 0x2c9c +#define R200_PP_TXMULTI_CTL_5 0x2cbc +#define R200_SE_VTX_STATE_CNTL 0x2180 +# define R200_UPDATE_USER_COLOR_0_ENA_MASK (1<<16) + + /* Registers for CP and Microcode Engine */ +#define RADEON_CP_ME_RAM_ADDR 0x07d4 +#define RADEON_CP_ME_RAM_RADDR 0x07d8 +#define RADEON_CP_ME_RAM_DATAH 0x07dc +#define RADEON_CP_ME_RAM_DATAL 0x07e0 + +#define RADEON_CP_RB_BASE 0x0700 +#define RADEON_CP_RB_CNTL 0x0704 +# define RADEON_RB_BUFSZ_SHIFT 0 +# define RADEON_RB_BUFSZ_MASK (0x3f << 0) +# define RADEON_RB_BLKSZ_SHIFT 8 +# define RADEON_RB_BLKSZ_MASK (0x3f << 8) +# define RADEON_BUF_SWAP_32BIT (1 << 17) +# define RADEON_MAX_FETCH_SHIFT 18 +# define RADEON_MAX_FETCH_MASK (0x3 << 18) +# define RADEON_RB_NO_UPDATE (1 << 27) +# define RADEON_RB_RPTR_WR_ENA (1 << 31) +#define RADEON_CP_RB_RPTR_ADDR 0x070c +#define RADEON_CP_RB_RPTR 0x0710 +#define RADEON_CP_RB_WPTR 0x0714 +#define RADEON_CP_RB_RPTR_WR 0x071c + +#define RADEON_SCRATCH_UMSK 0x0770 +#define RADEON_SCRATCH_ADDR 0x0774 + +#define R600_CP_RB_BASE 0xc100 +#define R600_CP_RB_CNTL 0xc104 +# define R600_RB_BUFSZ(x) ((x) << 0) +# define R600_RB_BLKSZ(x) ((x) << 8) +# define R600_RB_NO_UPDATE (1 << 27) +# define R600_RB_RPTR_WR_ENA (1 << 31) +#define R600_CP_RB_RPTR_WR 0xc108 +#define R600_CP_RB_RPTR_ADDR 0xc10c +#define R600_CP_RB_RPTR_ADDR_HI 0xc110 +#define R600_CP_RB_WPTR 0xc114 +#define R600_CP_RB_WPTR_ADDR 0xc118 +#define R600_CP_RB_WPTR_ADDR_HI 0xc11c +#define R600_CP_RB_RPTR 0x8700 +#define R600_CP_RB_WPTR_DELAY 0x8704 + +#define RADEON_CP_IB_BASE 0x0738 +#define RADEON_CP_IB_BUFSZ 0x073c + +#define RADEON_CP_CSQ_CNTL 0x0740 +# define RADEON_CSQ_CNT_PRIMARY_MASK (0xff << 0) +# define RADEON_CSQ_PRIDIS_INDDIS (0 << 28) +# define RADEON_CSQ_PRIPIO_INDDIS (1 << 28) +# define RADEON_CSQ_PRIBM_INDDIS (2 << 28) +# define RADEON_CSQ_PRIPIO_INDBM (3 << 28) +# define RADEON_CSQ_PRIBM_INDBM (4 << 28) +# define RADEON_CSQ_PRIPIO_INDPIO (15 << 28) + +#define R300_CP_RESYNC_ADDR 0x778 +#define R300_CP_RESYNC_DATA 0x77c + +#define RADEON_CP_CSQ_STAT 0x07f8 +# define RADEON_CSQ_RPTR_PRIMARY_MASK (0xff << 0) +# define RADEON_CSQ_WPTR_PRIMARY_MASK (0xff << 8) +# define RADEON_CSQ_RPTR_INDIRECT_MASK (0xff << 16) +# define RADEON_CSQ_WPTR_INDIRECT_MASK (0xff << 24) +#define RADEON_CP_CSQ2_STAT 0x07fc +#define RADEON_CP_CSQ_ADDR 0x07f0 +#define RADEON_CP_CSQ_DATA 0x07f4 +#define RADEON_CP_CSQ_APER_PRIMARY 0x1000 +#define RADEON_CP_CSQ_APER_INDIRECT 0x1300 + +#define RADEON_CP_RB_WPTR_DELAY 0x0718 +# define RADEON_PRE_WRITE_TIMER_SHIFT 0 +# define RADEON_PRE_WRITE_LIMIT_SHIFT 23 +#define RADEON_CP_CSQ_MODE 0x0744 +# define RADEON_INDIRECT2_START_SHIFT 0 +# define RADEON_INDIRECT2_START_MASK (0x7f << 0) +# define RADEON_INDIRECT1_START_SHIFT 8 +# define RADEON_INDIRECT1_START_MASK (0x7f << 8) + +#define RADEON_AIC_CNTL 0x01d0 +# define RADEON_PCIGART_TRANSLATE_EN (1 << 0) +# define RADEON_DIS_OUT_OF_PCI_GART_ACCESS (1 << 1) +# define RS400_MSI_REARM (1 << 3) /* rs400/rs480 */ +#define RADEON_AIC_LO_ADDR 0x01dc +#define RADEON_AIC_PT_BASE 0x01d8 +#define RADEON_AIC_HI_ADDR 0x01e0 + + + + /* Constants */ +/* #define RADEON_LAST_FRAME_REG RADEON_GUI_SCRATCH_REG0 */ +/* efine RADEON_LAST_CLEAR_REG RADEON_GUI_SCRATCH_REG2 */ + + + + /* CP packet types */ +#define RADEON_CP_PACKET0 0x00000000 +#define RADEON_CP_PACKET1 0x40000000 +#define RADEON_CP_PACKET2 0x80000000 +#define RADEON_CP_PACKET3 0xC0000000 +# define RADEON_CP_PACKET_MASK 0xC0000000 +# define RADEON_CP_PACKET_COUNT_MASK 0x3fff0000 +# define RADEON_CP_PACKET_MAX_DWORDS (1 << 12) +# define RADEON_CP_PACKET0_REG_MASK 0x000007ff +# define R300_CP_PACKET0_REG_MASK 0x00001fff +# define R600_CP_PACKET0_REG_MASK 0x0000ffff +# define RADEON_CP_PACKET1_REG0_MASK 0x000007ff +# define RADEON_CP_PACKET1_REG1_MASK 0x003ff800 + +#define RADEON_CP_PACKET0_ONE_REG_WR 0x00008000 + +#define RADEON_CP_PACKET3_NOP 0xC0001000 +#define RADEON_CP_PACKET3_NEXT_CHAR 0xC0001900 +#define RADEON_CP_PACKET3_PLY_NEXTSCAN 0xC0001D00 +#define RADEON_CP_PACKET3_SET_SCISSORS 0xC0001E00 +#define RADEON_CP_PACKET3_3D_RNDR_GEN_INDX_PRIM 0xC0002300 +#define RADEON_CP_PACKET3_LOAD_MICROCODE 0xC0002400 +#define RADEON_CP_PACKET3_WAIT_FOR_IDLE 0xC0002600 +#define RADEON_CP_PACKET3_3D_DRAW_VBUF 0xC0002800 +#define RADEON_CP_PACKET3_3D_DRAW_IMMD 0xC0002900 +#define RADEON_CP_PACKET3_3D_DRAW_INDX 0xC0002A00 +#define RADEON_CP_PACKET3_LOAD_PALETTE 0xC0002C00 +#define R200_CP_PACKET3_3D_DRAW_IMMD_2 0xc0003500 +#define RADEON_CP_PACKET3_3D_LOAD_VBPNTR 0xC0002F00 +#define RADEON_CP_PACKET3_CNTL_PAINT 0xC0009100 +#define RADEON_CP_PACKET3_CNTL_BITBLT 0xC0009200 +#define RADEON_CP_PACKET3_CNTL_SMALLTEXT 0xC0009300 +#define RADEON_CP_PACKET3_CNTL_HOSTDATA_BLT 0xC0009400 +#define RADEON_CP_PACKET3_CNTL_POLYLINE 0xC0009500 +#define RADEON_CP_PACKET3_CNTL_POLYSCANLINES 0xC0009800 +#define RADEON_CP_PACKET3_CNTL_PAINT_MULTI 0xC0009A00 +#define RADEON_CP_PACKET3_CNTL_BITBLT_MULTI 0xC0009B00 +#define RADEON_CP_PACKET3_CNTL_TRANS_BITBLT 0xC0009C00 + + +#define RADEON_CP_VC_FRMT_XY 0x00000000 +#define RADEON_CP_VC_FRMT_W0 0x00000001 +#define RADEON_CP_VC_FRMT_FPCOLOR 0x00000002 +#define RADEON_CP_VC_FRMT_FPALPHA 0x00000004 +#define RADEON_CP_VC_FRMT_PKCOLOR 0x00000008 +#define RADEON_CP_VC_FRMT_FPSPEC 0x00000010 +#define RADEON_CP_VC_FRMT_FPFOG 0x00000020 +#define RADEON_CP_VC_FRMT_PKSPEC 0x00000040 +#define RADEON_CP_VC_FRMT_ST0 0x00000080 +#define RADEON_CP_VC_FRMT_ST1 0x00000100 +#define RADEON_CP_VC_FRMT_Q1 0x00000200 +#define RADEON_CP_VC_FRMT_ST2 0x00000400 +#define RADEON_CP_VC_FRMT_Q2 0x00000800 +#define RADEON_CP_VC_FRMT_ST3 0x00001000 +#define RADEON_CP_VC_FRMT_Q3 0x00002000 +#define RADEON_CP_VC_FRMT_Q0 0x00004000 +#define RADEON_CP_VC_FRMT_BLND_WEIGHT_CNT_MASK 0x00038000 +#define RADEON_CP_VC_FRMT_N0 0x00040000 +#define RADEON_CP_VC_FRMT_XY1 0x08000000 +#define RADEON_CP_VC_FRMT_Z1 0x10000000 +#define RADEON_CP_VC_FRMT_W1 0x20000000 +#define RADEON_CP_VC_FRMT_N1 0x40000000 +#define RADEON_CP_VC_FRMT_Z 0x80000000 + +#define RADEON_CP_VC_CNTL_PRIM_TYPE_NONE 0x00000000 +#define RADEON_CP_VC_CNTL_PRIM_TYPE_POINT 0x00000001 +#define RADEON_CP_VC_CNTL_PRIM_TYPE_LINE 0x00000002 +#define RADEON_CP_VC_CNTL_PRIM_TYPE_LINE_STRIP 0x00000003 +#define RADEON_CP_VC_CNTL_PRIM_TYPE_TRI_LIST 0x00000004 +#define RADEON_CP_VC_CNTL_PRIM_TYPE_TRI_FAN 0x00000005 +#define RADEON_CP_VC_CNTL_PRIM_TYPE_TRI_STRIP 0x00000006 +#define RADEON_CP_VC_CNTL_PRIM_TYPE_TRI_TYPE_2 0x00000007 +#define RADEON_CP_VC_CNTL_PRIM_TYPE_RECT_LIST 0x00000008 +#define RADEON_CP_VC_CNTL_PRIM_TYPE_3VRT_POINT_LIST 0x00000009 +#define RADEON_CP_VC_CNTL_PRIM_TYPE_3VRT_LINE_LIST 0x0000000a +#define RADEON_CP_VC_CNTL_PRIM_WALK_IND 0x00000010 +#define RADEON_CP_VC_CNTL_PRIM_WALK_LIST 0x00000020 +#define RADEON_CP_VC_CNTL_PRIM_WALK_RING 0x00000030 +#define RADEON_CP_VC_CNTL_COLOR_ORDER_BGRA 0x00000000 +#define RADEON_CP_VC_CNTL_COLOR_ORDER_RGBA 0x00000040 +#define RADEON_CP_VC_CNTL_MAOS_ENABLE 0x00000080 +#define RADEON_CP_VC_CNTL_VTX_FMT_NON_RADEON_MODE 0x00000000 +#define RADEON_CP_VC_CNTL_VTX_FMT_RADEON_MODE 0x00000100 +#define RADEON_CP_VC_CNTL_TCL_DISABLE 0x00000000 +#define RADEON_CP_VC_CNTL_TCL_ENABLE 0x00000200 +#define RADEON_CP_VC_CNTL_NUM_SHIFT 16 + +#define RADEON_VS_MATRIX_0_ADDR 0 +#define RADEON_VS_MATRIX_1_ADDR 4 +#define RADEON_VS_MATRIX_2_ADDR 8 +#define RADEON_VS_MATRIX_3_ADDR 12 +#define RADEON_VS_MATRIX_4_ADDR 16 +#define RADEON_VS_MATRIX_5_ADDR 20 +#define RADEON_VS_MATRIX_6_ADDR 24 +#define RADEON_VS_MATRIX_7_ADDR 28 +#define RADEON_VS_MATRIX_8_ADDR 32 +#define RADEON_VS_MATRIX_9_ADDR 36 +#define RADEON_VS_MATRIX_10_ADDR 40 +#define RADEON_VS_MATRIX_11_ADDR 44 +#define RADEON_VS_MATRIX_12_ADDR 48 +#define RADEON_VS_MATRIX_13_ADDR 52 +#define RADEON_VS_MATRIX_14_ADDR 56 +#define RADEON_VS_MATRIX_15_ADDR 60 +#define RADEON_VS_LIGHT_AMBIENT_ADDR 64 +#define RADEON_VS_LIGHT_DIFFUSE_ADDR 72 +#define RADEON_VS_LIGHT_SPECULAR_ADDR 80 +#define RADEON_VS_LIGHT_DIRPOS_ADDR 88 +#define RADEON_VS_LIGHT_HWVSPOT_ADDR 96 +#define RADEON_VS_LIGHT_ATTENUATION_ADDR 104 +#define RADEON_VS_MATRIX_EYE2CLIP_ADDR 112 +#define RADEON_VS_UCP_ADDR 116 +#define RADEON_VS_GLOBAL_AMBIENT_ADDR 122 +#define RADEON_VS_FOG_PARAM_ADDR 123 +#define RADEON_VS_EYE_VECTOR_ADDR 124 + +#define RADEON_SS_LIGHT_DCD_ADDR 0 +#define RADEON_SS_LIGHT_SPOT_EXPONENT_ADDR 8 +#define RADEON_SS_LIGHT_SPOT_CUTOFF_ADDR 16 +#define RADEON_SS_LIGHT_SPECULAR_THRESH_ADDR 24 +#define RADEON_SS_LIGHT_RANGE_CUTOFF_ADDR 32 +#define RADEON_SS_VERT_GUARD_CLIP_ADJ_ADDR 48 +#define RADEON_SS_VERT_GUARD_DISCARD_ADJ_ADDR 49 +#define RADEON_SS_HORZ_GUARD_CLIP_ADJ_ADDR 50 +#define RADEON_SS_HORZ_GUARD_DISCARD_ADJ_ADDR 51 +#define RADEON_SS_SHININESS 60 + +#define RADEON_TV_MASTER_CNTL 0x0800 +# define RADEON_TV_ASYNC_RST (1 << 0) +# define RADEON_CRT_ASYNC_RST (1 << 1) +# define RADEON_RESTART_PHASE_FIX (1 << 3) +# define RADEON_TV_FIFO_ASYNC_RST (1 << 4) +# define RADEON_VIN_ASYNC_RST (1 << 5) +# define RADEON_AUD_ASYNC_RST (1 << 6) +# define RADEON_DVS_ASYNC_RST (1 << 7) +# define RADEON_CRT_FIFO_CE_EN (1 << 9) +# define RADEON_TV_FIFO_CE_EN (1 << 10) +# define RADEON_RE_SYNC_NOW_SEL_MASK (3 << 14) +# define RADEON_TVCLK_ALWAYS_ONb (1 << 30) +# define RADEON_TV_ON (1 << 31) +#define RADEON_TV_PRE_DAC_MUX_CNTL 0x0888 +# define RADEON_Y_RED_EN (1 << 0) +# define RADEON_C_GRN_EN (1 << 1) +# define RADEON_CMP_BLU_EN (1 << 2) +# define RADEON_DAC_DITHER_EN (1 << 3) +# define RADEON_RED_MX_FORCE_DAC_DATA (6 << 4) +# define RADEON_GRN_MX_FORCE_DAC_DATA (6 << 8) +# define RADEON_BLU_MX_FORCE_DAC_DATA (6 << 12) +# define RADEON_TV_FORCE_DAC_DATA_SHIFT 16 +#define RADEON_TV_RGB_CNTL 0x0804 +# define RADEON_SWITCH_TO_BLUE (1 << 4) +# define RADEON_RGB_DITHER_EN (1 << 5) +# define RADEON_RGB_SRC_SEL_MASK (3 << 8) +# define RADEON_RGB_SRC_SEL_CRTC1 (0 << 8) +# define RADEON_RGB_SRC_SEL_RMX (1 << 8) +# define RADEON_RGB_SRC_SEL_CRTC2 (2 << 8) +# define RADEON_RGB_CONVERT_BY_PASS (1 << 10) +# define RADEON_UVRAM_READ_MARGIN_SHIFT 16 +# define RADEON_FIFORAM_FFMACRO_READ_MARGIN_SHIFT 20 +# define RADEON_RGB_ATTEN_SEL(x) ((x) << 24) +# define RADEON_TVOUT_SCALE_EN (1 << 26) +# define RADEON_RGB_ATTEN_VAL(x) ((x) << 28) +#define RADEON_TV_SYNC_CNTL 0x0808 +# define RADEON_SYNC_OE (1 << 0) +# define RADEON_SYNC_OUT (1 << 1) +# define RADEON_SYNC_IN (1 << 2) +# define RADEON_SYNC_PUB (1 << 3) +# define RADEON_SYNC_PD (1 << 4) +# define RADEON_TV_SYNC_IO_DRIVE (1 << 5) +#define RADEON_TV_HTOTAL 0x080c +#define RADEON_TV_HDISP 0x0810 +#define RADEON_TV_HSTART 0x0818 +#define RADEON_TV_HCOUNT 0x081C +#define RADEON_TV_VTOTAL 0x0820 +#define RADEON_TV_VDISP 0x0824 +#define RADEON_TV_VCOUNT 0x0828 +#define RADEON_TV_FTOTAL 0x082c +#define RADEON_TV_FCOUNT 0x0830 +#define RADEON_TV_FRESTART 0x0834 +#define RADEON_TV_HRESTART 0x0838 +#define RADEON_TV_VRESTART 0x083c +#define RADEON_TV_HOST_READ_DATA 0x0840 +#define RADEON_TV_HOST_WRITE_DATA 0x0844 +#define RADEON_TV_HOST_RD_WT_CNTL 0x0848 +# define RADEON_HOST_FIFO_RD (1 << 12) +# define RADEON_HOST_FIFO_RD_ACK (1 << 13) +# define RADEON_HOST_FIFO_WT (1 << 14) +# define RADEON_HOST_FIFO_WT_ACK (1 << 15) +#define RADEON_TV_VSCALER_CNTL1 0x084c +# define RADEON_UV_INC_MASK 0xffff +# define RADEON_UV_INC_SHIFT 0 +# define RADEON_Y_W_EN (1 << 24) +# define RADEON_RESTART_FIELD (1 << 29) /* restart on field 0 */ +# define RADEON_Y_DEL_W_SIG_SHIFT 26 +#define RADEON_TV_TIMING_CNTL 0x0850 +# define RADEON_H_INC_MASK 0xfff +# define RADEON_H_INC_SHIFT 0 +# define RADEON_REQ_Y_FIRST (1 << 19) +# define RADEON_FORCE_BURST_ALWAYS (1 << 21) +# define RADEON_UV_POST_SCALE_BYPASS (1 << 23) +# define RADEON_UV_OUTPUT_POST_SCALE_SHIFT 24 +#define RADEON_TV_VSCALER_CNTL2 0x0854 +# define RADEON_DITHER_MODE (1 << 0) +# define RADEON_Y_OUTPUT_DITHER_EN (1 << 1) +# define RADEON_UV_OUTPUT_DITHER_EN (1 << 2) +# define RADEON_UV_TO_BUF_DITHER_EN (1 << 3) +#define RADEON_TV_Y_FALL_CNTL 0x0858 +# define RADEON_Y_FALL_PING_PONG (1 << 16) +# define RADEON_Y_COEF_EN (1 << 17) +#define RADEON_TV_Y_RISE_CNTL 0x085c +# define RADEON_Y_RISE_PING_PONG (1 << 16) +#define RADEON_TV_Y_SAW_TOOTH_CNTL 0x0860 +#define RADEON_TV_UPSAMP_AND_GAIN_CNTL 0x0864 +# define RADEON_YUPSAMP_EN (1 << 0) +# define RADEON_UVUPSAMP_EN (1 << 2) +#define RADEON_TV_GAIN_LIMIT_SETTINGS 0x0868 +# define RADEON_Y_GAIN_LIMIT_SHIFT 0 +# define RADEON_UV_GAIN_LIMIT_SHIFT 16 +#define RADEON_TV_LINEAR_GAIN_SETTINGS 0x086c +# define RADEON_Y_GAIN_SHIFT 0 +# define RADEON_UV_GAIN_SHIFT 16 +#define RADEON_TV_MODULATOR_CNTL1 0x0870 +# define RADEON_YFLT_EN (1 << 2) +# define RADEON_UVFLT_EN (1 << 3) +# define RADEON_ALT_PHASE_EN (1 << 6) +# define RADEON_SYNC_TIP_LEVEL (1 << 7) +# define RADEON_BLANK_LEVEL_SHIFT 8 +# define RADEON_SET_UP_LEVEL_SHIFT 16 +# define RADEON_SLEW_RATE_LIMIT (1 << 23) +# define RADEON_CY_FILT_BLEND_SHIFT 28 +#define RADEON_TV_MODULATOR_CNTL2 0x0874 +# define RADEON_TV_U_BURST_LEVEL_MASK 0x1ff +# define RADEON_TV_V_BURST_LEVEL_MASK 0x1ff +# define RADEON_TV_V_BURST_LEVEL_SHIFT 16 +#define RADEON_TV_CRC_CNTL 0x0890 +#define RADEON_TV_UV_ADR 0x08ac +# define RADEON_MAX_UV_ADR_MASK 0x000000ff +# define RADEON_MAX_UV_ADR_SHIFT 0 +# define RADEON_TABLE1_BOT_ADR_MASK 0x0000ff00 +# define RADEON_TABLE1_BOT_ADR_SHIFT 8 +# define RADEON_TABLE3_TOP_ADR_MASK 0x00ff0000 +# define RADEON_TABLE3_TOP_ADR_SHIFT 16 +# define RADEON_HCODE_TABLE_SEL_MASK 0x06000000 +# define RADEON_HCODE_TABLE_SEL_SHIFT 25 +# define RADEON_VCODE_TABLE_SEL_MASK 0x18000000 +# define RADEON_VCODE_TABLE_SEL_SHIFT 27 +# define RADEON_TV_MAX_FIFO_ADDR 0x1a7 +# define RADEON_TV_MAX_FIFO_ADDR_INTERNAL 0x1ff +#define RADEON_TV_PLL_FINE_CNTL 0x0020 /* PLL */ +#define RADEON_TV_PLL_CNTL 0x0021 /* PLL */ +# define RADEON_TV_M0LO_MASK 0xff +# define RADEON_TV_M0HI_MASK 0x7 +# define RADEON_TV_M0HI_SHIFT 18 +# define RADEON_TV_N0LO_MASK 0x1ff +# define RADEON_TV_N0LO_SHIFT 8 +# define RADEON_TV_N0HI_MASK 0x3 +# define RADEON_TV_N0HI_SHIFT 21 +# define RADEON_TV_P_MASK 0xf +# define RADEON_TV_P_SHIFT 24 +# define RADEON_TV_SLIP_EN (1 << 23) +# define RADEON_TV_DTO_EN (1 << 28) +#define RADEON_TV_PLL_CNTL1 0x0022 /* PLL */ +# define RADEON_TVPLL_RESET (1 << 1) +# define RADEON_TVPLL_SLEEP (1 << 3) +# define RADEON_TVPLL_REFCLK_SEL (1 << 4) +# define RADEON_TVPCP_SHIFT 8 +# define RADEON_TVPCP_MASK (7 << 8) +# define RADEON_TVPVG_SHIFT 11 +# define RADEON_TVPVG_MASK (7 << 11) +# define RADEON_TVPDC_SHIFT 14 +# define RADEON_TVPDC_MASK (3 << 14) +# define RADEON_TVPLL_TEST_DIS (1 << 31) +# define RADEON_TVCLK_SRC_SEL_TVPLL (1 << 30) + +#define RS400_DISP2_REQ_CNTL1 0xe30 +# define RS400_DISP2_START_REQ_LEVEL_SHIFT 0 +# define RS400_DISP2_START_REQ_LEVEL_MASK 0x3ff +# define RS400_DISP2_STOP_REQ_LEVEL_SHIFT 12 +# define RS400_DISP2_STOP_REQ_LEVEL_MASK 0x3ff +# define RS400_DISP2_ALLOW_FID_LEVEL_SHIFT 22 +# define RS400_DISP2_ALLOW_FID_LEVEL_MASK 0x3ff +#define RS400_DISP2_REQ_CNTL2 0xe34 +# define RS400_DISP2_CRITICAL_POINT_START_SHIFT 12 +# define RS400_DISP2_CRITICAL_POINT_START_MASK 0x3ff +# define RS400_DISP2_CRITICAL_POINT_STOP_SHIFT 22 +# define RS400_DISP2_CRITICAL_POINT_STOP_MASK 0x3ff +#define RS400_DMIF_MEM_CNTL1 0xe38 +# define RS400_DISP2_START_ADR_SHIFT 0 +# define RS400_DISP2_START_ADR_MASK 0x3ff +# define RS400_DISP1_CRITICAL_POINT_START_SHIFT 12 +# define RS400_DISP1_CRITICAL_POINT_START_MASK 0x3ff +# define RS400_DISP1_CRITICAL_POINT_STOP_SHIFT 22 +# define RS400_DISP1_CRITICAL_POINT_STOP_MASK 0x3ff +#define RS400_DISP1_REQ_CNTL1 0xe3c +# define RS400_DISP1_START_REQ_LEVEL_SHIFT 0 +# define RS400_DISP1_START_REQ_LEVEL_MASK 0x3ff +# define RS400_DISP1_STOP_REQ_LEVEL_SHIFT 12 +# define RS400_DISP1_STOP_REQ_LEVEL_MASK 0x3ff +# define RS400_DISP1_ALLOW_FID_LEVEL_SHIFT 22 +# define RS400_DISP1_ALLOW_FID_LEVEL_MASK 0x3ff + +#define RADEON_PCIE_INDEX 0x0030 +#define RADEON_PCIE_DATA 0x0034 +#define RADEON_PCIE_TX_GART_CNTL 0x10 +# define RADEON_PCIE_TX_GART_EN (1 << 0) +# define RADEON_PCIE_TX_GART_UNMAPPED_ACCESS_PASS_THRU (0 << 1) +# define RADEON_PCIE_TX_GART_UNMAPPED_ACCESS_CLAMP_LO (1 << 1) +# define RADEON_PCIE_TX_GART_UNMAPPED_ACCESS_DISCARD (3 << 1) +# define RADEON_PCIE_TX_GART_MODE_32_128_CACHE (0 << 3) +# define RADEON_PCIE_TX_GART_MODE_8_4_128_CACHE (1 << 3) +# define RADEON_PCIE_TX_GART_CHK_RW_VALID_EN (1 << 5) +# define RADEON_PCIE_TX_GART_INVALIDATE_TLB (1 << 8) +#define RADEON_PCIE_TX_DISCARD_RD_ADDR_LO 0x11 +#define RADEON_PCIE_TX_DISCARD_RD_ADDR_HI 0x12 +#define RADEON_PCIE_TX_GART_BASE 0x13 +#define RADEON_PCIE_TX_GART_START_LO 0x14 +#define RADEON_PCIE_TX_GART_START_HI 0x15 +#define RADEON_PCIE_TX_GART_END_LO 0x16 +#define RADEON_PCIE_TX_GART_END_HI 0x17 +#define RADEON_PCIE_TX_GART_ERROR 0x18 + +#define RADEON_SCRATCH_REG0 0x15e0 +#define RADEON_SCRATCH_REG1 0x15e4 +#define RADEON_SCRATCH_REG2 0x15e8 +#define RADEON_SCRATCH_REG3 0x15ec +#define RADEON_SCRATCH_REG4 0x15f0 +#define RADEON_SCRATCH_REG5 0x15f4 + +#define RV530_GB_PIPE_SELECT2 0x4124 + +#endif diff --git a/drivers/gpu/drm/radeon/radeon_ring.c b/drivers/gpu/drm/radeon/radeon_ring.c new file mode 100644 index 00000000..08c0233d --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_ring.c @@ -0,0 +1,465 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#include +#include +#include "drmP.h" +#include "radeon_drm.h" +#include "radeon_reg.h" +#include "radeon.h" +#include "atom.h" + +int radeon_debugfs_ib_init(struct radeon_device *rdev); + +void radeon_ib_bogus_cleanup(struct radeon_device *rdev) +{ + struct radeon_ib *ib, *n; + + list_for_each_entry_safe(ib, n, &rdev->ib_pool.bogus_ib, list) { + list_del(&ib->list); + vfree(ib->ptr); + kfree(ib); + } +} + +void radeon_ib_bogus_add(struct radeon_device *rdev, struct radeon_ib *ib) +{ + struct radeon_ib *bib; + + bib = kmalloc(sizeof(*bib), GFP_KERNEL); + if (bib == NULL) + return; + bib->ptr = vmalloc(ib->length_dw * 4); + if (bib->ptr == NULL) { + kfree(bib); + return; + } + memcpy(bib->ptr, ib->ptr, ib->length_dw * 4); + bib->length_dw = ib->length_dw; + mutex_lock(&rdev->ib_pool.mutex); + list_add_tail(&bib->list, &rdev->ib_pool.bogus_ib); + mutex_unlock(&rdev->ib_pool.mutex); +} + +/* + * IB. + */ +int radeon_ib_get(struct radeon_device *rdev, struct radeon_ib **ib) +{ + struct radeon_fence *fence; + struct radeon_ib *nib; + int r = 0, i, c; + + *ib = NULL; + r = radeon_fence_create(rdev, &fence); + if (r) { + dev_err(rdev->dev, "failed to create fence for new IB\n"); + return r; + } + mutex_lock(&rdev->ib_pool.mutex); + for (i = rdev->ib_pool.head_id, c = 0, nib = NULL; c < RADEON_IB_POOL_SIZE; c++, i++) { + i &= (RADEON_IB_POOL_SIZE - 1); + if (rdev->ib_pool.ibs[i].free) { + nib = &rdev->ib_pool.ibs[i]; + break; + } + } + if (nib == NULL) { + /* This should never happen, it means we allocated all + * IB and haven't scheduled one yet, return EBUSY to + * userspace hoping that on ioctl recall we get better + * luck + */ + dev_err(rdev->dev, "no free indirect buffer !\n"); + mutex_unlock(&rdev->ib_pool.mutex); + radeon_fence_unref(&fence); + return -EBUSY; + } + rdev->ib_pool.head_id = (nib->idx + 1) & (RADEON_IB_POOL_SIZE - 1); + nib->free = false; + if (nib->fence) { + mutex_unlock(&rdev->ib_pool.mutex); + r = radeon_fence_wait(nib->fence, false); + if (r) { + dev_err(rdev->dev, "error waiting fence of IB(%u:0x%016lX:%u)\n", + nib->idx, (unsigned long)nib->gpu_addr, nib->length_dw); + mutex_lock(&rdev->ib_pool.mutex); + nib->free = true; + mutex_unlock(&rdev->ib_pool.mutex); + radeon_fence_unref(&fence); + return r; + } + mutex_lock(&rdev->ib_pool.mutex); + } + radeon_fence_unref(&nib->fence); + nib->fence = fence; + nib->length_dw = 0; + mutex_unlock(&rdev->ib_pool.mutex); + *ib = nib; + return 0; +} + +void radeon_ib_free(struct radeon_device *rdev, struct radeon_ib **ib) +{ + struct radeon_ib *tmp = *ib; + + *ib = NULL; + if (tmp == NULL) { + return; + } + if (!tmp->fence->emited) + radeon_fence_unref(&tmp->fence); + mutex_lock(&rdev->ib_pool.mutex); + tmp->free = true; + mutex_unlock(&rdev->ib_pool.mutex); +} + +int radeon_ib_schedule(struct radeon_device *rdev, struct radeon_ib *ib) +{ + int r = 0; + + if (!ib->length_dw || !rdev->cp.ready) { + /* TODO: Nothings in the ib we should report. */ + DRM_ERROR("radeon: couldn't schedule IB(%u).\n", ib->idx); + return -EINVAL; + } + + /* 64 dwords should be enough for fence too */ + r = radeon_ring_lock(rdev, 64); + if (r) { + DRM_ERROR("radeon: scheduling IB failed (%d).\n", r); + return r; + } + radeon_ring_ib_execute(rdev, ib); + radeon_fence_emit(rdev, ib->fence); + mutex_lock(&rdev->ib_pool.mutex); + /* once scheduled IB is considered free and protected by the fence */ + ib->free = true; + mutex_unlock(&rdev->ib_pool.mutex); + radeon_ring_unlock_commit(rdev); + return 0; +} + +int radeon_ib_pool_init(struct radeon_device *rdev) +{ + void *ptr; + uint64_t gpu_addr; + int i; + int r = 0; + + if (rdev->ib_pool.robj) + return 0; + INIT_LIST_HEAD(&rdev->ib_pool.bogus_ib); + /* Allocate 1M object buffer */ + r = radeon_bo_create(rdev, RADEON_IB_POOL_SIZE*64*1024, + PAGE_SIZE, true, RADEON_GEM_DOMAIN_GTT, + &rdev->ib_pool.robj); + if (r) { + DRM_ERROR("radeon: failed to ib pool (%d).\n", r); + return r; + } + r = radeon_bo_reserve(rdev->ib_pool.robj, false); + if (unlikely(r != 0)) + return r; + r = radeon_bo_pin(rdev->ib_pool.robj, RADEON_GEM_DOMAIN_GTT, &gpu_addr); + if (r) { + radeon_bo_unreserve(rdev->ib_pool.robj); + DRM_ERROR("radeon: failed to pin ib pool (%d).\n", r); + return r; + } + r = radeon_bo_kmap(rdev->ib_pool.robj, &ptr); + radeon_bo_unreserve(rdev->ib_pool.robj); + if (r) { + DRM_ERROR("radeon: failed to map ib pool (%d).\n", r); + return r; + } + for (i = 0; i < RADEON_IB_POOL_SIZE; i++) { + unsigned offset; + + offset = i * 64 * 1024; + rdev->ib_pool.ibs[i].gpu_addr = gpu_addr + offset; + rdev->ib_pool.ibs[i].ptr = ptr + offset; + rdev->ib_pool.ibs[i].idx = i; + rdev->ib_pool.ibs[i].length_dw = 0; + rdev->ib_pool.ibs[i].free = true; + } + rdev->ib_pool.head_id = 0; + rdev->ib_pool.ready = true; + DRM_INFO("radeon: ib pool ready.\n"); + if (radeon_debugfs_ib_init(rdev)) { + DRM_ERROR("Failed to register debugfs file for IB !\n"); + } + return r; +} + +void radeon_ib_pool_fini(struct radeon_device *rdev) +{ + int r; + struct radeon_bo *robj; + + if (!rdev->ib_pool.ready) { + return; + } + mutex_lock(&rdev->ib_pool.mutex); + radeon_ib_bogus_cleanup(rdev); + robj = rdev->ib_pool.robj; + rdev->ib_pool.robj = NULL; + mutex_unlock(&rdev->ib_pool.mutex); + + if (robj) { + r = radeon_bo_reserve(robj, false); + if (likely(r == 0)) { + radeon_bo_kunmap(robj); + radeon_bo_unpin(robj); + radeon_bo_unreserve(robj); + } + radeon_bo_unref(&robj); + } +} + + +/* + * Ring. + */ +void radeon_ring_free_size(struct radeon_device *rdev) +{ + if (rdev->wb.enabled) + rdev->cp.rptr = le32_to_cpu(rdev->wb.wb[RADEON_WB_CP_RPTR_OFFSET/4]); + else { + if (rdev->family >= CHIP_R600) + rdev->cp.rptr = RREG32(R600_CP_RB_RPTR); + else + rdev->cp.rptr = RREG32(RADEON_CP_RB_RPTR); + } + /* This works because ring_size is a power of 2 */ + rdev->cp.ring_free_dw = (rdev->cp.rptr + (rdev->cp.ring_size / 4)); + rdev->cp.ring_free_dw -= rdev->cp.wptr; + rdev->cp.ring_free_dw &= rdev->cp.ptr_mask; + if (!rdev->cp.ring_free_dw) { + rdev->cp.ring_free_dw = rdev->cp.ring_size / 4; + } +} + +int radeon_ring_alloc(struct radeon_device *rdev, unsigned ndw) +{ + int r; + + /* Align requested size with padding so unlock_commit can + * pad safely */ + ndw = (ndw + rdev->cp.align_mask) & ~rdev->cp.align_mask; + while (ndw > (rdev->cp.ring_free_dw - 1)) { + radeon_ring_free_size(rdev); + if (ndw < rdev->cp.ring_free_dw) { + break; + } + r = radeon_fence_wait_next(rdev); + if (r) + return r; + } + rdev->cp.count_dw = ndw; + rdev->cp.wptr_old = rdev->cp.wptr; + return 0; +} + +int radeon_ring_lock(struct radeon_device *rdev, unsigned ndw) +{ + int r; + + mutex_lock(&rdev->cp.mutex); + r = radeon_ring_alloc(rdev, ndw); + if (r) { + mutex_unlock(&rdev->cp.mutex); + return r; + } + return 0; +} + +void radeon_ring_commit(struct radeon_device *rdev) +{ + unsigned count_dw_pad; + unsigned i; + + /* We pad to match fetch size */ + count_dw_pad = (rdev->cp.align_mask + 1) - + (rdev->cp.wptr & rdev->cp.align_mask); + for (i = 0; i < count_dw_pad; i++) { + radeon_ring_write(rdev, 2 << 30); + } + DRM_MEMORYBARRIER(); + radeon_cp_commit(rdev); +} + +void radeon_ring_unlock_commit(struct radeon_device *rdev) +{ + radeon_ring_commit(rdev); + mutex_unlock(&rdev->cp.mutex); +} + +void radeon_ring_unlock_undo(struct radeon_device *rdev) +{ + rdev->cp.wptr = rdev->cp.wptr_old; + mutex_unlock(&rdev->cp.mutex); +} + +int radeon_ring_init(struct radeon_device *rdev, unsigned ring_size) +{ + int r; + + rdev->cp.ring_size = ring_size; + /* Allocate ring buffer */ + if (rdev->cp.ring_obj == NULL) { + r = radeon_bo_create(rdev, rdev->cp.ring_size, PAGE_SIZE, true, + RADEON_GEM_DOMAIN_GTT, + &rdev->cp.ring_obj); + if (r) { + dev_err(rdev->dev, "(%d) ring create failed\n", r); + return r; + } + r = radeon_bo_reserve(rdev->cp.ring_obj, false); + if (unlikely(r != 0)) + return r; + r = radeon_bo_pin(rdev->cp.ring_obj, RADEON_GEM_DOMAIN_GTT, + &rdev->cp.gpu_addr); + if (r) { + radeon_bo_unreserve(rdev->cp.ring_obj); + dev_err(rdev->dev, "(%d) ring pin failed\n", r); + return r; + } + r = radeon_bo_kmap(rdev->cp.ring_obj, + (void **)&rdev->cp.ring); + radeon_bo_unreserve(rdev->cp.ring_obj); + if (r) { + dev_err(rdev->dev, "(%d) ring map failed\n", r); + return r; + } + } + rdev->cp.ptr_mask = (rdev->cp.ring_size / 4) - 1; + rdev->cp.ring_free_dw = rdev->cp.ring_size / 4; + return 0; +} + +void radeon_ring_fini(struct radeon_device *rdev) +{ + int r; + struct radeon_bo *ring_obj; + + mutex_lock(&rdev->cp.mutex); + ring_obj = rdev->cp.ring_obj; + rdev->cp.ring = NULL; + rdev->cp.ring_obj = NULL; + mutex_unlock(&rdev->cp.mutex); + + if (ring_obj) { + r = radeon_bo_reserve(ring_obj, false); + if (likely(r == 0)) { + radeon_bo_kunmap(ring_obj); + radeon_bo_unpin(ring_obj); + radeon_bo_unreserve(ring_obj); + } + radeon_bo_unref(&ring_obj); + } +} + + +/* + * Debugfs info + */ +#if defined(CONFIG_DEBUG_FS) +static int radeon_debugfs_ib_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct radeon_ib *ib = node->info_ent->data; + unsigned i; + + if (ib == NULL) { + return 0; + } + seq_printf(m, "IB %04u\n", ib->idx); + seq_printf(m, "IB fence %p\n", ib->fence); + seq_printf(m, "IB size %05u dwords\n", ib->length_dw); + for (i = 0; i < ib->length_dw; i++) { + seq_printf(m, "[%05u]=0x%08X\n", i, ib->ptr[i]); + } + return 0; +} + +static int radeon_debugfs_ib_bogus_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct radeon_device *rdev = node->info_ent->data; + struct radeon_ib *ib; + unsigned i; + + mutex_lock(&rdev->ib_pool.mutex); + if (list_empty(&rdev->ib_pool.bogus_ib)) { + mutex_unlock(&rdev->ib_pool.mutex); + seq_printf(m, "no bogus IB recorded\n"); + return 0; + } + ib = list_first_entry(&rdev->ib_pool.bogus_ib, struct radeon_ib, list); + list_del_init(&ib->list); + mutex_unlock(&rdev->ib_pool.mutex); + seq_printf(m, "IB size %05u dwords\n", ib->length_dw); + for (i = 0; i < ib->length_dw; i++) { + seq_printf(m, "[%05u]=0x%08X\n", i, ib->ptr[i]); + } + vfree(ib->ptr); + kfree(ib); + return 0; +} + +static struct drm_info_list radeon_debugfs_ib_list[RADEON_IB_POOL_SIZE]; +static char radeon_debugfs_ib_names[RADEON_IB_POOL_SIZE][32]; + +static struct drm_info_list radeon_debugfs_ib_bogus_info_list[] = { + {"radeon_ib_bogus", radeon_debugfs_ib_bogus_info, 0, NULL}, +}; +#endif + +int radeon_debugfs_ib_init(struct radeon_device *rdev) +{ +#if defined(CONFIG_DEBUG_FS) + unsigned i; + int r; + + radeon_debugfs_ib_bogus_info_list[0].data = rdev; + r = radeon_debugfs_add_files(rdev, radeon_debugfs_ib_bogus_info_list, 1); + if (r) + return r; + for (i = 0; i < RADEON_IB_POOL_SIZE; i++) { + sprintf(radeon_debugfs_ib_names[i], "radeon_ib_%04u", i); + radeon_debugfs_ib_list[i].name = radeon_debugfs_ib_names[i]; + radeon_debugfs_ib_list[i].show = &radeon_debugfs_ib_info; + radeon_debugfs_ib_list[i].driver_features = 0; + radeon_debugfs_ib_list[i].data = &rdev->ib_pool.ibs[i]; + } + return radeon_debugfs_add_files(rdev, radeon_debugfs_ib_list, + RADEON_IB_POOL_SIZE); +#else + return 0; +#endif +} diff --git a/drivers/gpu/drm/radeon/radeon_state.c b/drivers/gpu/drm/radeon/radeon_state.c new file mode 100644 index 00000000..92e7ea73 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_state.c @@ -0,0 +1,3261 @@ +/* radeon_state.c -- State support for Radeon -*- linux-c -*- */ +/* + * Copyright 2000 VA Linux Systems, Inc., Fremont, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Gareth Hughes + * Kevin E. Martin + */ + +#include "drmP.h" +#include "drm.h" +#include "drm_buffer.h" +#include "drm_sarea.h" +#include "radeon_drm.h" +#include "radeon_drv.h" + +/* ================================================================ + * Helper functions for client state checking and fixup + */ + +static __inline__ int radeon_check_and_fixup_offset(drm_radeon_private_t * + dev_priv, + struct drm_file * file_priv, + u32 *offset) +{ + u64 off = *offset; + u32 fb_end = dev_priv->fb_location + dev_priv->fb_size - 1; + struct drm_radeon_driver_file_fields *radeon_priv; + + /* Hrm ... the story of the offset ... So this function converts + * the various ideas of what userland clients might have for an + * offset in the card address space into an offset into the card + * address space :) So with a sane client, it should just keep + * the value intact and just do some boundary checking. However, + * not all clients are sane. Some older clients pass us 0 based + * offsets relative to the start of the framebuffer and some may + * assume the AGP aperture it appended to the framebuffer, so we + * try to detect those cases and fix them up. + * + * Note: It might be a good idea here to make sure the offset lands + * in some "allowed" area to protect things like the PCIE GART... + */ + + /* First, the best case, the offset already lands in either the + * framebuffer or the GART mapped space + */ + if (radeon_check_offset(dev_priv, off)) + return 0; + + /* Ok, that didn't happen... now check if we have a zero based + * offset that fits in the framebuffer + gart space, apply the + * magic offset we get from SETPARAM or calculated from fb_location + */ + if (off < (dev_priv->fb_size + dev_priv->gart_size)) { + radeon_priv = file_priv->driver_priv; + off += radeon_priv->radeon_fb_delta; + } + + /* Finally, assume we aimed at a GART offset if beyond the fb */ + if (off > fb_end) + off = off - fb_end - 1 + dev_priv->gart_vm_start; + + /* Now recheck and fail if out of bounds */ + if (radeon_check_offset(dev_priv, off)) { + DRM_DEBUG("offset fixed up to 0x%x\n", (unsigned int)off); + *offset = off; + return 0; + } + return -EINVAL; +} + +static __inline__ int radeon_check_and_fixup_packets(drm_radeon_private_t * + dev_priv, + struct drm_file *file_priv, + int id, struct drm_buffer *buf) +{ + u32 *data; + switch (id) { + + case RADEON_EMIT_PP_MISC: + data = drm_buffer_pointer_to_dword(buf, + (RADEON_RB3D_DEPTHOFFSET - RADEON_PP_MISC) / 4); + + if (radeon_check_and_fixup_offset(dev_priv, file_priv, data)) { + DRM_ERROR("Invalid depth buffer offset\n"); + return -EINVAL; + } + dev_priv->have_z_offset = 1; + break; + + case RADEON_EMIT_PP_CNTL: + data = drm_buffer_pointer_to_dword(buf, + (RADEON_RB3D_COLOROFFSET - RADEON_PP_CNTL) / 4); + + if (radeon_check_and_fixup_offset(dev_priv, file_priv, data)) { + DRM_ERROR("Invalid colour buffer offset\n"); + return -EINVAL; + } + break; + + case R200_EMIT_PP_TXOFFSET_0: + case R200_EMIT_PP_TXOFFSET_1: + case R200_EMIT_PP_TXOFFSET_2: + case R200_EMIT_PP_TXOFFSET_3: + case R200_EMIT_PP_TXOFFSET_4: + case R200_EMIT_PP_TXOFFSET_5: + data = drm_buffer_pointer_to_dword(buf, 0); + if (radeon_check_and_fixup_offset(dev_priv, file_priv, data)) { + DRM_ERROR("Invalid R200 texture offset\n"); + return -EINVAL; + } + break; + + case RADEON_EMIT_PP_TXFILTER_0: + case RADEON_EMIT_PP_TXFILTER_1: + case RADEON_EMIT_PP_TXFILTER_2: + data = drm_buffer_pointer_to_dword(buf, + (RADEON_PP_TXOFFSET_0 - RADEON_PP_TXFILTER_0) / 4); + if (radeon_check_and_fixup_offset(dev_priv, file_priv, data)) { + DRM_ERROR("Invalid R100 texture offset\n"); + return -EINVAL; + } + break; + + case R200_EMIT_PP_CUBIC_OFFSETS_0: + case R200_EMIT_PP_CUBIC_OFFSETS_1: + case R200_EMIT_PP_CUBIC_OFFSETS_2: + case R200_EMIT_PP_CUBIC_OFFSETS_3: + case R200_EMIT_PP_CUBIC_OFFSETS_4: + case R200_EMIT_PP_CUBIC_OFFSETS_5:{ + int i; + for (i = 0; i < 5; i++) { + data = drm_buffer_pointer_to_dword(buf, i); + if (radeon_check_and_fixup_offset(dev_priv, + file_priv, + data)) { + DRM_ERROR + ("Invalid R200 cubic texture offset\n"); + return -EINVAL; + } + } + break; + } + + case RADEON_EMIT_PP_CUBIC_OFFSETS_T0: + case RADEON_EMIT_PP_CUBIC_OFFSETS_T1: + case RADEON_EMIT_PP_CUBIC_OFFSETS_T2:{ + int i; + for (i = 0; i < 5; i++) { + data = drm_buffer_pointer_to_dword(buf, i); + if (radeon_check_and_fixup_offset(dev_priv, + file_priv, + data)) { + DRM_ERROR + ("Invalid R100 cubic texture offset\n"); + return -EINVAL; + } + } + } + break; + + case R200_EMIT_VAP_CTL:{ + RING_LOCALS; + BEGIN_RING(2); + OUT_RING_REG(RADEON_SE_TCL_STATE_FLUSH, 0); + ADVANCE_RING(); + } + break; + + case RADEON_EMIT_RB3D_COLORPITCH: + case RADEON_EMIT_RE_LINE_PATTERN: + case RADEON_EMIT_SE_LINE_WIDTH: + case RADEON_EMIT_PP_LUM_MATRIX: + case RADEON_EMIT_PP_ROT_MATRIX_0: + case RADEON_EMIT_RB3D_STENCILREFMASK: + case RADEON_EMIT_SE_VPORT_XSCALE: + case RADEON_EMIT_SE_CNTL: + case RADEON_EMIT_SE_CNTL_STATUS: + case RADEON_EMIT_RE_MISC: + case RADEON_EMIT_PP_BORDER_COLOR_0: + case RADEON_EMIT_PP_BORDER_COLOR_1: + case RADEON_EMIT_PP_BORDER_COLOR_2: + case RADEON_EMIT_SE_ZBIAS_FACTOR: + case RADEON_EMIT_SE_TCL_OUTPUT_VTX_FMT: + case RADEON_EMIT_SE_TCL_MATERIAL_EMMISSIVE_RED: + case R200_EMIT_PP_TXCBLEND_0: + case R200_EMIT_PP_TXCBLEND_1: + case R200_EMIT_PP_TXCBLEND_2: + case R200_EMIT_PP_TXCBLEND_3: + case R200_EMIT_PP_TXCBLEND_4: + case R200_EMIT_PP_TXCBLEND_5: + case R200_EMIT_PP_TXCBLEND_6: + case R200_EMIT_PP_TXCBLEND_7: + case R200_EMIT_TCL_LIGHT_MODEL_CTL_0: + case R200_EMIT_TFACTOR_0: + case R200_EMIT_VTX_FMT_0: + case R200_EMIT_MATRIX_SELECT_0: + case R200_EMIT_TEX_PROC_CTL_2: + case R200_EMIT_TCL_UCP_VERT_BLEND_CTL: + case R200_EMIT_PP_TXFILTER_0: + case R200_EMIT_PP_TXFILTER_1: + case R200_EMIT_PP_TXFILTER_2: + case R200_EMIT_PP_TXFILTER_3: + case R200_EMIT_PP_TXFILTER_4: + case R200_EMIT_PP_TXFILTER_5: + case R200_EMIT_VTE_CNTL: + case R200_EMIT_OUTPUT_VTX_COMP_SEL: + case R200_EMIT_PP_TAM_DEBUG3: + case R200_EMIT_PP_CNTL_X: + case R200_EMIT_RB3D_DEPTHXY_OFFSET: + case R200_EMIT_RE_AUX_SCISSOR_CNTL: + case R200_EMIT_RE_SCISSOR_TL_0: + case R200_EMIT_RE_SCISSOR_TL_1: + case R200_EMIT_RE_SCISSOR_TL_2: + case R200_EMIT_SE_VAP_CNTL_STATUS: + case R200_EMIT_SE_VTX_STATE_CNTL: + case R200_EMIT_RE_POINTSIZE: + case R200_EMIT_TCL_INPUT_VTX_VECTOR_ADDR_0: + case R200_EMIT_PP_CUBIC_FACES_0: + case R200_EMIT_PP_CUBIC_FACES_1: + case R200_EMIT_PP_CUBIC_FACES_2: + case R200_EMIT_PP_CUBIC_FACES_3: + case R200_EMIT_PP_CUBIC_FACES_4: + case R200_EMIT_PP_CUBIC_FACES_5: + case RADEON_EMIT_PP_TEX_SIZE_0: + case RADEON_EMIT_PP_TEX_SIZE_1: + case RADEON_EMIT_PP_TEX_SIZE_2: + case R200_EMIT_RB3D_BLENDCOLOR: + case R200_EMIT_TCL_POINT_SPRITE_CNTL: + case RADEON_EMIT_PP_CUBIC_FACES_0: + case RADEON_EMIT_PP_CUBIC_FACES_1: + case RADEON_EMIT_PP_CUBIC_FACES_2: + case R200_EMIT_PP_TRI_PERF_CNTL: + case R200_EMIT_PP_AFS_0: + case R200_EMIT_PP_AFS_1: + case R200_EMIT_ATF_TFACTOR: + case R200_EMIT_PP_TXCTLALL_0: + case R200_EMIT_PP_TXCTLALL_1: + case R200_EMIT_PP_TXCTLALL_2: + case R200_EMIT_PP_TXCTLALL_3: + case R200_EMIT_PP_TXCTLALL_4: + case R200_EMIT_PP_TXCTLALL_5: + case R200_EMIT_VAP_PVS_CNTL: + /* These packets don't contain memory offsets */ + break; + + default: + DRM_ERROR("Unknown state packet ID %d\n", id); + return -EINVAL; + } + + return 0; +} + +static __inline__ int radeon_check_and_fixup_packet3(drm_radeon_private_t * + dev_priv, + struct drm_file *file_priv, + drm_radeon_kcmd_buffer_t * + cmdbuf, + unsigned int *cmdsz) +{ + u32 *cmd = drm_buffer_pointer_to_dword(cmdbuf->buffer, 0); + u32 offset, narrays; + int count, i, k; + + count = ((*cmd & RADEON_CP_PACKET_COUNT_MASK) >> 16); + *cmdsz = 2 + count; + + if ((*cmd & 0xc0000000) != RADEON_CP_PACKET3) { + DRM_ERROR("Not a type 3 packet\n"); + return -EINVAL; + } + + if (4 * *cmdsz > drm_buffer_unprocessed(cmdbuf->buffer)) { + DRM_ERROR("Packet size larger than size of data provided\n"); + return -EINVAL; + } + + switch (*cmd & 0xff00) { + /* XXX Are there old drivers needing other packets? */ + + case RADEON_3D_DRAW_IMMD: + case RADEON_3D_DRAW_VBUF: + case RADEON_3D_DRAW_INDX: + case RADEON_WAIT_FOR_IDLE: + case RADEON_CP_NOP: + case RADEON_3D_CLEAR_ZMASK: +/* case RADEON_CP_NEXT_CHAR: + case RADEON_CP_PLY_NEXTSCAN: + case RADEON_CP_SET_SCISSORS: */ /* probably safe but will never need them? */ + /* these packets are safe */ + break; + + case RADEON_CP_3D_DRAW_IMMD_2: + case RADEON_CP_3D_DRAW_VBUF_2: + case RADEON_CP_3D_DRAW_INDX_2: + case RADEON_3D_CLEAR_HIZ: + /* safe but r200 only */ + if (dev_priv->microcode_version != UCODE_R200) { + DRM_ERROR("Invalid 3d packet for r100-class chip\n"); + return -EINVAL; + } + break; + + case RADEON_3D_LOAD_VBPNTR: + + if (count > 18) { /* 12 arrays max */ + DRM_ERROR("Too large payload in 3D_LOAD_VBPNTR (count=%d)\n", + count); + return -EINVAL; + } + + /* carefully check packet contents */ + cmd = drm_buffer_pointer_to_dword(cmdbuf->buffer, 1); + + narrays = *cmd & ~0xc000; + k = 0; + i = 2; + while ((k < narrays) && (i < (count + 2))) { + i++; /* skip attribute field */ + cmd = drm_buffer_pointer_to_dword(cmdbuf->buffer, i); + if (radeon_check_and_fixup_offset(dev_priv, file_priv, + cmd)) { + DRM_ERROR + ("Invalid offset (k=%d i=%d) in 3D_LOAD_VBPNTR packet.\n", + k, i); + return -EINVAL; + } + k++; + i++; + if (k == narrays) + break; + /* have one more to process, they come in pairs */ + cmd = drm_buffer_pointer_to_dword(cmdbuf->buffer, i); + + if (radeon_check_and_fixup_offset(dev_priv, + file_priv, cmd)) + { + DRM_ERROR + ("Invalid offset (k=%d i=%d) in 3D_LOAD_VBPNTR packet.\n", + k, i); + return -EINVAL; + } + k++; + i++; + } + /* do the counts match what we expect ? */ + if ((k != narrays) || (i != (count + 2))) { + DRM_ERROR + ("Malformed 3D_LOAD_VBPNTR packet (k=%d i=%d narrays=%d count+1=%d).\n", + k, i, narrays, count + 1); + return -EINVAL; + } + break; + + case RADEON_3D_RNDR_GEN_INDX_PRIM: + if (dev_priv->microcode_version != UCODE_R100) { + DRM_ERROR("Invalid 3d packet for r200-class chip\n"); + return -EINVAL; + } + + cmd = drm_buffer_pointer_to_dword(cmdbuf->buffer, 1); + if (radeon_check_and_fixup_offset(dev_priv, file_priv, cmd)) { + DRM_ERROR("Invalid rndr_gen_indx offset\n"); + return -EINVAL; + } + break; + + case RADEON_CP_INDX_BUFFER: + if (dev_priv->microcode_version != UCODE_R200) { + DRM_ERROR("Invalid 3d packet for r100-class chip\n"); + return -EINVAL; + } + + cmd = drm_buffer_pointer_to_dword(cmdbuf->buffer, 1); + if ((*cmd & 0x8000ffff) != 0x80000810) { + DRM_ERROR("Invalid indx_buffer reg address %08X\n", *cmd); + return -EINVAL; + } + cmd = drm_buffer_pointer_to_dword(cmdbuf->buffer, 2); + if (radeon_check_and_fixup_offset(dev_priv, file_priv, cmd)) { + DRM_ERROR("Invalid indx_buffer offset is %08X\n", *cmd); + return -EINVAL; + } + break; + + case RADEON_CNTL_HOSTDATA_BLT: + case RADEON_CNTL_PAINT_MULTI: + case RADEON_CNTL_BITBLT_MULTI: + /* MSB of opcode: next DWORD GUI_CNTL */ + cmd = drm_buffer_pointer_to_dword(cmdbuf->buffer, 1); + if (*cmd & (RADEON_GMC_SRC_PITCH_OFFSET_CNTL + | RADEON_GMC_DST_PITCH_OFFSET_CNTL)) { + u32 *cmd2 = drm_buffer_pointer_to_dword(cmdbuf->buffer, 2); + offset = *cmd2 << 10; + if (radeon_check_and_fixup_offset + (dev_priv, file_priv, &offset)) { + DRM_ERROR("Invalid first packet offset\n"); + return -EINVAL; + } + *cmd2 = (*cmd2 & 0xffc00000) | offset >> 10; + } + + if ((*cmd & RADEON_GMC_SRC_PITCH_OFFSET_CNTL) && + (*cmd & RADEON_GMC_DST_PITCH_OFFSET_CNTL)) { + u32 *cmd3 = drm_buffer_pointer_to_dword(cmdbuf->buffer, 3); + offset = *cmd3 << 10; + if (radeon_check_and_fixup_offset + (dev_priv, file_priv, &offset)) { + DRM_ERROR("Invalid second packet offset\n"); + return -EINVAL; + } + *cmd3 = (*cmd3 & 0xffc00000) | offset >> 10; + } + break; + + default: + DRM_ERROR("Invalid packet type %x\n", *cmd & 0xff00); + return -EINVAL; + } + + return 0; +} + +/* ================================================================ + * CP hardware state programming functions + */ + +static __inline__ void radeon_emit_clip_rect(drm_radeon_private_t * dev_priv, + struct drm_clip_rect * box) +{ + RING_LOCALS; + + DRM_DEBUG(" box: x1=%d y1=%d x2=%d y2=%d\n", + box->x1, box->y1, box->x2, box->y2); + + BEGIN_RING(4); + OUT_RING(CP_PACKET0(RADEON_RE_TOP_LEFT, 0)); + OUT_RING((box->y1 << 16) | box->x1); + OUT_RING(CP_PACKET0(RADEON_RE_WIDTH_HEIGHT, 0)); + OUT_RING(((box->y2 - 1) << 16) | (box->x2 - 1)); + ADVANCE_RING(); +} + +/* Emit 1.1 state + */ +static int radeon_emit_state(drm_radeon_private_t * dev_priv, + struct drm_file *file_priv, + drm_radeon_context_regs_t * ctx, + drm_radeon_texture_regs_t * tex, + unsigned int dirty) +{ + RING_LOCALS; + DRM_DEBUG("dirty=0x%08x\n", dirty); + + if (dirty & RADEON_UPLOAD_CONTEXT) { + if (radeon_check_and_fixup_offset(dev_priv, file_priv, + &ctx->rb3d_depthoffset)) { + DRM_ERROR("Invalid depth buffer offset\n"); + return -EINVAL; + } + + if (radeon_check_and_fixup_offset(dev_priv, file_priv, + &ctx->rb3d_coloroffset)) { + DRM_ERROR("Invalid depth buffer offset\n"); + return -EINVAL; + } + + BEGIN_RING(14); + OUT_RING(CP_PACKET0(RADEON_PP_MISC, 6)); + OUT_RING(ctx->pp_misc); + OUT_RING(ctx->pp_fog_color); + OUT_RING(ctx->re_solid_color); + OUT_RING(ctx->rb3d_blendcntl); + OUT_RING(ctx->rb3d_depthoffset); + OUT_RING(ctx->rb3d_depthpitch); + OUT_RING(ctx->rb3d_zstencilcntl); + OUT_RING(CP_PACKET0(RADEON_PP_CNTL, 2)); + OUT_RING(ctx->pp_cntl); + OUT_RING(ctx->rb3d_cntl); + OUT_RING(ctx->rb3d_coloroffset); + OUT_RING(CP_PACKET0(RADEON_RB3D_COLORPITCH, 0)); + OUT_RING(ctx->rb3d_colorpitch); + ADVANCE_RING(); + } + + if (dirty & RADEON_UPLOAD_VERTFMT) { + BEGIN_RING(2); + OUT_RING(CP_PACKET0(RADEON_SE_COORD_FMT, 0)); + OUT_RING(ctx->se_coord_fmt); + ADVANCE_RING(); + } + + if (dirty & RADEON_UPLOAD_LINE) { + BEGIN_RING(5); + OUT_RING(CP_PACKET0(RADEON_RE_LINE_PATTERN, 1)); + OUT_RING(ctx->re_line_pattern); + OUT_RING(ctx->re_line_state); + OUT_RING(CP_PACKET0(RADEON_SE_LINE_WIDTH, 0)); + OUT_RING(ctx->se_line_width); + ADVANCE_RING(); + } + + if (dirty & RADEON_UPLOAD_BUMPMAP) { + BEGIN_RING(5); + OUT_RING(CP_PACKET0(RADEON_PP_LUM_MATRIX, 0)); + OUT_RING(ctx->pp_lum_matrix); + OUT_RING(CP_PACKET0(RADEON_PP_ROT_MATRIX_0, 1)); + OUT_RING(ctx->pp_rot_matrix_0); + OUT_RING(ctx->pp_rot_matrix_1); + ADVANCE_RING(); + } + + if (dirty & RADEON_UPLOAD_MASKS) { + BEGIN_RING(4); + OUT_RING(CP_PACKET0(RADEON_RB3D_STENCILREFMASK, 2)); + OUT_RING(ctx->rb3d_stencilrefmask); + OUT_RING(ctx->rb3d_ropcntl); + OUT_RING(ctx->rb3d_planemask); + ADVANCE_RING(); + } + + if (dirty & RADEON_UPLOAD_VIEWPORT) { + BEGIN_RING(7); + OUT_RING(CP_PACKET0(RADEON_SE_VPORT_XSCALE, 5)); + OUT_RING(ctx->se_vport_xscale); + OUT_RING(ctx->se_vport_xoffset); + OUT_RING(ctx->se_vport_yscale); + OUT_RING(ctx->se_vport_yoffset); + OUT_RING(ctx->se_vport_zscale); + OUT_RING(ctx->se_vport_zoffset); + ADVANCE_RING(); + } + + if (dirty & RADEON_UPLOAD_SETUP) { + BEGIN_RING(4); + OUT_RING(CP_PACKET0(RADEON_SE_CNTL, 0)); + OUT_RING(ctx->se_cntl); + OUT_RING(CP_PACKET0(RADEON_SE_CNTL_STATUS, 0)); + OUT_RING(ctx->se_cntl_status); + ADVANCE_RING(); + } + + if (dirty & RADEON_UPLOAD_MISC) { + BEGIN_RING(2); + OUT_RING(CP_PACKET0(RADEON_RE_MISC, 0)); + OUT_RING(ctx->re_misc); + ADVANCE_RING(); + } + + if (dirty & RADEON_UPLOAD_TEX0) { + if (radeon_check_and_fixup_offset(dev_priv, file_priv, + &tex[0].pp_txoffset)) { + DRM_ERROR("Invalid texture offset for unit 0\n"); + return -EINVAL; + } + + BEGIN_RING(9); + OUT_RING(CP_PACKET0(RADEON_PP_TXFILTER_0, 5)); + OUT_RING(tex[0].pp_txfilter); + OUT_RING(tex[0].pp_txformat); + OUT_RING(tex[0].pp_txoffset); + OUT_RING(tex[0].pp_txcblend); + OUT_RING(tex[0].pp_txablend); + OUT_RING(tex[0].pp_tfactor); + OUT_RING(CP_PACKET0(RADEON_PP_BORDER_COLOR_0, 0)); + OUT_RING(tex[0].pp_border_color); + ADVANCE_RING(); + } + + if (dirty & RADEON_UPLOAD_TEX1) { + if (radeon_check_and_fixup_offset(dev_priv, file_priv, + &tex[1].pp_txoffset)) { + DRM_ERROR("Invalid texture offset for unit 1\n"); + return -EINVAL; + } + + BEGIN_RING(9); + OUT_RING(CP_PACKET0(RADEON_PP_TXFILTER_1, 5)); + OUT_RING(tex[1].pp_txfilter); + OUT_RING(tex[1].pp_txformat); + OUT_RING(tex[1].pp_txoffset); + OUT_RING(tex[1].pp_txcblend); + OUT_RING(tex[1].pp_txablend); + OUT_RING(tex[1].pp_tfactor); + OUT_RING(CP_PACKET0(RADEON_PP_BORDER_COLOR_1, 0)); + OUT_RING(tex[1].pp_border_color); + ADVANCE_RING(); + } + + if (dirty & RADEON_UPLOAD_TEX2) { + if (radeon_check_and_fixup_offset(dev_priv, file_priv, + &tex[2].pp_txoffset)) { + DRM_ERROR("Invalid texture offset for unit 2\n"); + return -EINVAL; + } + + BEGIN_RING(9); + OUT_RING(CP_PACKET0(RADEON_PP_TXFILTER_2, 5)); + OUT_RING(tex[2].pp_txfilter); + OUT_RING(tex[2].pp_txformat); + OUT_RING(tex[2].pp_txoffset); + OUT_RING(tex[2].pp_txcblend); + OUT_RING(tex[2].pp_txablend); + OUT_RING(tex[2].pp_tfactor); + OUT_RING(CP_PACKET0(RADEON_PP_BORDER_COLOR_2, 0)); + OUT_RING(tex[2].pp_border_color); + ADVANCE_RING(); + } + + return 0; +} + +/* Emit 1.2 state + */ +static int radeon_emit_state2(drm_radeon_private_t * dev_priv, + struct drm_file *file_priv, + drm_radeon_state_t * state) +{ + RING_LOCALS; + + if (state->dirty & RADEON_UPLOAD_ZBIAS) { + BEGIN_RING(3); + OUT_RING(CP_PACKET0(RADEON_SE_ZBIAS_FACTOR, 1)); + OUT_RING(state->context2.se_zbias_factor); + OUT_RING(state->context2.se_zbias_constant); + ADVANCE_RING(); + } + + return radeon_emit_state(dev_priv, file_priv, &state->context, + state->tex, state->dirty); +} + +/* New (1.3) state mechanism. 3 commands (packet, scalar, vector) in + * 1.3 cmdbuffers allow all previous state to be updated as well as + * the tcl scalar and vector areas. + */ +static struct { + int start; + int len; + const char *name; +} packet[RADEON_MAX_STATE_PACKETS] = { + {RADEON_PP_MISC, 7, "RADEON_PP_MISC"}, + {RADEON_PP_CNTL, 3, "RADEON_PP_CNTL"}, + {RADEON_RB3D_COLORPITCH, 1, "RADEON_RB3D_COLORPITCH"}, + {RADEON_RE_LINE_PATTERN, 2, "RADEON_RE_LINE_PATTERN"}, + {RADEON_SE_LINE_WIDTH, 1, "RADEON_SE_LINE_WIDTH"}, + {RADEON_PP_LUM_MATRIX, 1, "RADEON_PP_LUM_MATRIX"}, + {RADEON_PP_ROT_MATRIX_0, 2, "RADEON_PP_ROT_MATRIX_0"}, + {RADEON_RB3D_STENCILREFMASK, 3, "RADEON_RB3D_STENCILREFMASK"}, + {RADEON_SE_VPORT_XSCALE, 6, "RADEON_SE_VPORT_XSCALE"}, + {RADEON_SE_CNTL, 2, "RADEON_SE_CNTL"}, + {RADEON_SE_CNTL_STATUS, 1, "RADEON_SE_CNTL_STATUS"}, + {RADEON_RE_MISC, 1, "RADEON_RE_MISC"}, + {RADEON_PP_TXFILTER_0, 6, "RADEON_PP_TXFILTER_0"}, + {RADEON_PP_BORDER_COLOR_0, 1, "RADEON_PP_BORDER_COLOR_0"}, + {RADEON_PP_TXFILTER_1, 6, "RADEON_PP_TXFILTER_1"}, + {RADEON_PP_BORDER_COLOR_1, 1, "RADEON_PP_BORDER_COLOR_1"}, + {RADEON_PP_TXFILTER_2, 6, "RADEON_PP_TXFILTER_2"}, + {RADEON_PP_BORDER_COLOR_2, 1, "RADEON_PP_BORDER_COLOR_2"}, + {RADEON_SE_ZBIAS_FACTOR, 2, "RADEON_SE_ZBIAS_FACTOR"}, + {RADEON_SE_TCL_OUTPUT_VTX_FMT, 11, "RADEON_SE_TCL_OUTPUT_VTX_FMT"}, + {RADEON_SE_TCL_MATERIAL_EMMISSIVE_RED, 17, + "RADEON_SE_TCL_MATERIAL_EMMISSIVE_RED"}, + {R200_PP_TXCBLEND_0, 4, "R200_PP_TXCBLEND_0"}, + {R200_PP_TXCBLEND_1, 4, "R200_PP_TXCBLEND_1"}, + {R200_PP_TXCBLEND_2, 4, "R200_PP_TXCBLEND_2"}, + {R200_PP_TXCBLEND_3, 4, "R200_PP_TXCBLEND_3"}, + {R200_PP_TXCBLEND_4, 4, "R200_PP_TXCBLEND_4"}, + {R200_PP_TXCBLEND_5, 4, "R200_PP_TXCBLEND_5"}, + {R200_PP_TXCBLEND_6, 4, "R200_PP_TXCBLEND_6"}, + {R200_PP_TXCBLEND_7, 4, "R200_PP_TXCBLEND_7"}, + {R200_SE_TCL_LIGHT_MODEL_CTL_0, 6, "R200_SE_TCL_LIGHT_MODEL_CTL_0"}, + {R200_PP_TFACTOR_0, 6, "R200_PP_TFACTOR_0"}, + {R200_SE_VTX_FMT_0, 4, "R200_SE_VTX_FMT_0"}, + {R200_SE_VAP_CNTL, 1, "R200_SE_VAP_CNTL"}, + {R200_SE_TCL_MATRIX_SEL_0, 5, "R200_SE_TCL_MATRIX_SEL_0"}, + {R200_SE_TCL_TEX_PROC_CTL_2, 5, "R200_SE_TCL_TEX_PROC_CTL_2"}, + {R200_SE_TCL_UCP_VERT_BLEND_CTL, 1, "R200_SE_TCL_UCP_VERT_BLEND_CTL"}, + {R200_PP_TXFILTER_0, 6, "R200_PP_TXFILTER_0"}, + {R200_PP_TXFILTER_1, 6, "R200_PP_TXFILTER_1"}, + {R200_PP_TXFILTER_2, 6, "R200_PP_TXFILTER_2"}, + {R200_PP_TXFILTER_3, 6, "R200_PP_TXFILTER_3"}, + {R200_PP_TXFILTER_4, 6, "R200_PP_TXFILTER_4"}, + {R200_PP_TXFILTER_5, 6, "R200_PP_TXFILTER_5"}, + {R200_PP_TXOFFSET_0, 1, "R200_PP_TXOFFSET_0"}, + {R200_PP_TXOFFSET_1, 1, "R200_PP_TXOFFSET_1"}, + {R200_PP_TXOFFSET_2, 1, "R200_PP_TXOFFSET_2"}, + {R200_PP_TXOFFSET_3, 1, "R200_PP_TXOFFSET_3"}, + {R200_PP_TXOFFSET_4, 1, "R200_PP_TXOFFSET_4"}, + {R200_PP_TXOFFSET_5, 1, "R200_PP_TXOFFSET_5"}, + {R200_SE_VTE_CNTL, 1, "R200_SE_VTE_CNTL"}, + {R200_SE_TCL_OUTPUT_VTX_COMP_SEL, 1, + "R200_SE_TCL_OUTPUT_VTX_COMP_SEL"}, + {R200_PP_TAM_DEBUG3, 1, "R200_PP_TAM_DEBUG3"}, + {R200_PP_CNTL_X, 1, "R200_PP_CNTL_X"}, + {R200_RB3D_DEPTHXY_OFFSET, 1, "R200_RB3D_DEPTHXY_OFFSET"}, + {R200_RE_AUX_SCISSOR_CNTL, 1, "R200_RE_AUX_SCISSOR_CNTL"}, + {R200_RE_SCISSOR_TL_0, 2, "R200_RE_SCISSOR_TL_0"}, + {R200_RE_SCISSOR_TL_1, 2, "R200_RE_SCISSOR_TL_1"}, + {R200_RE_SCISSOR_TL_2, 2, "R200_RE_SCISSOR_TL_2"}, + {R200_SE_VAP_CNTL_STATUS, 1, "R200_SE_VAP_CNTL_STATUS"}, + {R200_SE_VTX_STATE_CNTL, 1, "R200_SE_VTX_STATE_CNTL"}, + {R200_RE_POINTSIZE, 1, "R200_RE_POINTSIZE"}, + {R200_SE_TCL_INPUT_VTX_VECTOR_ADDR_0, 4, + "R200_SE_TCL_INPUT_VTX_VECTOR_ADDR_0"}, + {R200_PP_CUBIC_FACES_0, 1, "R200_PP_CUBIC_FACES_0"}, /* 61 */ + {R200_PP_CUBIC_OFFSET_F1_0, 5, "R200_PP_CUBIC_OFFSET_F1_0"}, /* 62 */ + {R200_PP_CUBIC_FACES_1, 1, "R200_PP_CUBIC_FACES_1"}, + {R200_PP_CUBIC_OFFSET_F1_1, 5, "R200_PP_CUBIC_OFFSET_F1_1"}, + {R200_PP_CUBIC_FACES_2, 1, "R200_PP_CUBIC_FACES_2"}, + {R200_PP_CUBIC_OFFSET_F1_2, 5, "R200_PP_CUBIC_OFFSET_F1_2"}, + {R200_PP_CUBIC_FACES_3, 1, "R200_PP_CUBIC_FACES_3"}, + {R200_PP_CUBIC_OFFSET_F1_3, 5, "R200_PP_CUBIC_OFFSET_F1_3"}, + {R200_PP_CUBIC_FACES_4, 1, "R200_PP_CUBIC_FACES_4"}, + {R200_PP_CUBIC_OFFSET_F1_4, 5, "R200_PP_CUBIC_OFFSET_F1_4"}, + {R200_PP_CUBIC_FACES_5, 1, "R200_PP_CUBIC_FACES_5"}, + {R200_PP_CUBIC_OFFSET_F1_5, 5, "R200_PP_CUBIC_OFFSET_F1_5"}, + {RADEON_PP_TEX_SIZE_0, 2, "RADEON_PP_TEX_SIZE_0"}, + {RADEON_PP_TEX_SIZE_1, 2, "RADEON_PP_TEX_SIZE_1"}, + {RADEON_PP_TEX_SIZE_2, 2, "RADEON_PP_TEX_SIZE_2"}, + {R200_RB3D_BLENDCOLOR, 3, "R200_RB3D_BLENDCOLOR"}, + {R200_SE_TCL_POINT_SPRITE_CNTL, 1, "R200_SE_TCL_POINT_SPRITE_CNTL"}, + {RADEON_PP_CUBIC_FACES_0, 1, "RADEON_PP_CUBIC_FACES_0"}, + {RADEON_PP_CUBIC_OFFSET_T0_0, 5, "RADEON_PP_CUBIC_OFFSET_T0_0"}, + {RADEON_PP_CUBIC_FACES_1, 1, "RADEON_PP_CUBIC_FACES_1"}, + {RADEON_PP_CUBIC_OFFSET_T1_0, 5, "RADEON_PP_CUBIC_OFFSET_T1_0"}, + {RADEON_PP_CUBIC_FACES_2, 1, "RADEON_PP_CUBIC_FACES_2"}, + {RADEON_PP_CUBIC_OFFSET_T2_0, 5, "RADEON_PP_CUBIC_OFFSET_T2_0"}, + {R200_PP_TRI_PERF, 2, "R200_PP_TRI_PERF"}, + {R200_PP_AFS_0, 32, "R200_PP_AFS_0"}, /* 85 */ + {R200_PP_AFS_1, 32, "R200_PP_AFS_1"}, + {R200_PP_TFACTOR_0, 8, "R200_ATF_TFACTOR"}, + {R200_PP_TXFILTER_0, 8, "R200_PP_TXCTLALL_0"}, + {R200_PP_TXFILTER_1, 8, "R200_PP_TXCTLALL_1"}, + {R200_PP_TXFILTER_2, 8, "R200_PP_TXCTLALL_2"}, + {R200_PP_TXFILTER_3, 8, "R200_PP_TXCTLALL_3"}, + {R200_PP_TXFILTER_4, 8, "R200_PP_TXCTLALL_4"}, + {R200_PP_TXFILTER_5, 8, "R200_PP_TXCTLALL_5"}, + {R200_VAP_PVS_CNTL_1, 2, "R200_VAP_PVS_CNTL"}, +}; + +/* ================================================================ + * Performance monitoring functions + */ + +static void radeon_clear_box(drm_radeon_private_t * dev_priv, + struct drm_radeon_master_private *master_priv, + int x, int y, int w, int h, int r, int g, int b) +{ + u32 color; + RING_LOCALS; + + x += master_priv->sarea_priv->boxes[0].x1; + y += master_priv->sarea_priv->boxes[0].y1; + + switch (dev_priv->color_fmt) { + case RADEON_COLOR_FORMAT_RGB565: + color = (((r & 0xf8) << 8) | + ((g & 0xfc) << 3) | ((b & 0xf8) >> 3)); + break; + case RADEON_COLOR_FORMAT_ARGB8888: + default: + color = (((0xff) << 24) | (r << 16) | (g << 8) | b); + break; + } + + BEGIN_RING(4); + RADEON_WAIT_UNTIL_3D_IDLE(); + OUT_RING(CP_PACKET0(RADEON_DP_WRITE_MASK, 0)); + OUT_RING(0xffffffff); + ADVANCE_RING(); + + BEGIN_RING(6); + + OUT_RING(CP_PACKET3(RADEON_CNTL_PAINT_MULTI, 4)); + OUT_RING(RADEON_GMC_DST_PITCH_OFFSET_CNTL | + RADEON_GMC_BRUSH_SOLID_COLOR | + (dev_priv->color_fmt << 8) | + RADEON_GMC_SRC_DATATYPE_COLOR | + RADEON_ROP3_P | RADEON_GMC_CLR_CMP_CNTL_DIS); + + if (master_priv->sarea_priv->pfCurrentPage == 1) { + OUT_RING(dev_priv->front_pitch_offset); + } else { + OUT_RING(dev_priv->back_pitch_offset); + } + + OUT_RING(color); + + OUT_RING((x << 16) | y); + OUT_RING((w << 16) | h); + + ADVANCE_RING(); +} + +static void radeon_cp_performance_boxes(drm_radeon_private_t *dev_priv, struct drm_radeon_master_private *master_priv) +{ + /* Collapse various things into a wait flag -- trying to + * guess if userspase slept -- better just to have them tell us. + */ + if (dev_priv->stats.last_frame_reads > 1 || + dev_priv->stats.last_clear_reads > dev_priv->stats.clears) { + dev_priv->stats.boxes |= RADEON_BOX_WAIT_IDLE; + } + + if (dev_priv->stats.freelist_loops) { + dev_priv->stats.boxes |= RADEON_BOX_WAIT_IDLE; + } + + /* Purple box for page flipping + */ + if (dev_priv->stats.boxes & RADEON_BOX_FLIP) + radeon_clear_box(dev_priv, master_priv, 4, 4, 8, 8, 255, 0, 255); + + /* Red box if we have to wait for idle at any point + */ + if (dev_priv->stats.boxes & RADEON_BOX_WAIT_IDLE) + radeon_clear_box(dev_priv, master_priv, 16, 4, 8, 8, 255, 0, 0); + + /* Blue box: lost context? + */ + + /* Yellow box for texture swaps + */ + if (dev_priv->stats.boxes & RADEON_BOX_TEXTURE_LOAD) + radeon_clear_box(dev_priv, master_priv, 40, 4, 8, 8, 255, 255, 0); + + /* Green box if hardware never idles (as far as we can tell) + */ + if (!(dev_priv->stats.boxes & RADEON_BOX_DMA_IDLE)) + radeon_clear_box(dev_priv, master_priv, 64, 4, 8, 8, 0, 255, 0); + + /* Draw bars indicating number of buffers allocated + * (not a great measure, easily confused) + */ + if (dev_priv->stats.requested_bufs) { + if (dev_priv->stats.requested_bufs > 100) + dev_priv->stats.requested_bufs = 100; + + radeon_clear_box(dev_priv, master_priv, 4, 16, + dev_priv->stats.requested_bufs, 4, + 196, 128, 128); + } + + memset(&dev_priv->stats, 0, sizeof(dev_priv->stats)); + +} + +/* ================================================================ + * CP command dispatch functions + */ + +static void radeon_cp_dispatch_clear(struct drm_device * dev, + struct drm_master *master, + drm_radeon_clear_t * clear, + drm_radeon_clear_rect_t * depth_boxes) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + struct drm_radeon_master_private *master_priv = master->driver_priv; + drm_radeon_sarea_t *sarea_priv = master_priv->sarea_priv; + drm_radeon_depth_clear_t *depth_clear = &dev_priv->depth_clear; + int nbox = sarea_priv->nbox; + struct drm_clip_rect *pbox = sarea_priv->boxes; + unsigned int flags = clear->flags; + u32 rb3d_cntl = 0, rb3d_stencilrefmask = 0; + int i; + RING_LOCALS; + DRM_DEBUG("flags = 0x%x\n", flags); + + dev_priv->stats.clears++; + + if (sarea_priv->pfCurrentPage == 1) { + unsigned int tmp = flags; + + flags &= ~(RADEON_FRONT | RADEON_BACK); + if (tmp & RADEON_FRONT) + flags |= RADEON_BACK; + if (tmp & RADEON_BACK) + flags |= RADEON_FRONT; + } + if (flags & (RADEON_DEPTH|RADEON_STENCIL)) { + if (!dev_priv->have_z_offset) { + printk_once(KERN_ERR "radeon: illegal depth clear request. Buggy mesa detected - please update.\n"); + flags &= ~(RADEON_DEPTH | RADEON_STENCIL); + } + } + + if (flags & (RADEON_FRONT | RADEON_BACK)) { + + BEGIN_RING(4); + + /* Ensure the 3D stream is idle before doing a + * 2D fill to clear the front or back buffer. + */ + RADEON_WAIT_UNTIL_3D_IDLE(); + + OUT_RING(CP_PACKET0(RADEON_DP_WRITE_MASK, 0)); + OUT_RING(clear->color_mask); + + ADVANCE_RING(); + + /* Make sure we restore the 3D state next time. + */ + sarea_priv->ctx_owner = 0; + + for (i = 0; i < nbox; i++) { + int x = pbox[i].x1; + int y = pbox[i].y1; + int w = pbox[i].x2 - x; + int h = pbox[i].y2 - y; + + DRM_DEBUG("%d,%d-%d,%d flags 0x%x\n", + x, y, w, h, flags); + + if (flags & RADEON_FRONT) { + BEGIN_RING(6); + + OUT_RING(CP_PACKET3 + (RADEON_CNTL_PAINT_MULTI, 4)); + OUT_RING(RADEON_GMC_DST_PITCH_OFFSET_CNTL | + RADEON_GMC_BRUSH_SOLID_COLOR | + (dev_priv-> + color_fmt << 8) | + RADEON_GMC_SRC_DATATYPE_COLOR | + RADEON_ROP3_P | + RADEON_GMC_CLR_CMP_CNTL_DIS); + + OUT_RING(dev_priv->front_pitch_offset); + OUT_RING(clear->clear_color); + + OUT_RING((x << 16) | y); + OUT_RING((w << 16) | h); + + ADVANCE_RING(); + } + + if (flags & RADEON_BACK) { + BEGIN_RING(6); + + OUT_RING(CP_PACKET3 + (RADEON_CNTL_PAINT_MULTI, 4)); + OUT_RING(RADEON_GMC_DST_PITCH_OFFSET_CNTL | + RADEON_GMC_BRUSH_SOLID_COLOR | + (dev_priv-> + color_fmt << 8) | + RADEON_GMC_SRC_DATATYPE_COLOR | + RADEON_ROP3_P | + RADEON_GMC_CLR_CMP_CNTL_DIS); + + OUT_RING(dev_priv->back_pitch_offset); + OUT_RING(clear->clear_color); + + OUT_RING((x << 16) | y); + OUT_RING((w << 16) | h); + + ADVANCE_RING(); + } + } + } + + /* hyper z clear */ + /* no docs available, based on reverse engineering by Stephane Marchesin */ + if ((flags & (RADEON_DEPTH | RADEON_STENCIL)) + && (flags & RADEON_CLEAR_FASTZ)) { + + int i; + int depthpixperline = + dev_priv->depth_fmt == + RADEON_DEPTH_FORMAT_16BIT_INT_Z ? (dev_priv->depth_pitch / + 2) : (dev_priv-> + depth_pitch / 4); + + u32 clearmask; + + u32 tempRB3D_DEPTHCLEARVALUE = clear->clear_depth | + ((clear->depth_mask & 0xff) << 24); + + /* Make sure we restore the 3D state next time. + * we haven't touched any "normal" state - still need this? + */ + sarea_priv->ctx_owner = 0; + + if ((dev_priv->flags & RADEON_HAS_HIERZ) + && (flags & RADEON_USE_HIERZ)) { + /* FIXME : reverse engineer that for Rx00 cards */ + /* FIXME : the mask supposedly contains low-res z values. So can't set + just to the max (0xff? or actually 0x3fff?), need to take z clear + value into account? */ + /* pattern seems to work for r100, though get slight + rendering errors with glxgears. If hierz is not enabled for r100, + only 4 bits which indicate clear (15,16,31,32, all zero) matter, the + other ones are ignored, and the same clear mask can be used. That's + very different behaviour than R200 which needs different clear mask + and different number of tiles to clear if hierz is enabled or not !?! + */ + clearmask = (0xff << 22) | (0xff << 6) | 0x003f003f; + } else { + /* clear mask : chooses the clearing pattern. + rv250: could be used to clear only parts of macrotiles + (but that would get really complicated...)? + bit 0 and 1 (either or both of them ?!?!) are used to + not clear tile (or maybe one of the bits indicates if the tile is + compressed or not), bit 2 and 3 to not clear tile 1,...,. + Pattern is as follows: + | 0,1 | 4,5 | 8,9 |12,13|16,17|20,21|24,25|28,29| + bits ------------------------------------------------- + | 2,3 | 6,7 |10,11|14,15|18,19|22,23|26,27|30,31| + rv100: clearmask covers 2x8 4x1 tiles, but one clear still + covers 256 pixels ?!? + */ + clearmask = 0x0; + } + + BEGIN_RING(8); + RADEON_WAIT_UNTIL_2D_IDLE(); + OUT_RING_REG(RADEON_RB3D_DEPTHCLEARVALUE, + tempRB3D_DEPTHCLEARVALUE); + /* what offset is this exactly ? */ + OUT_RING_REG(RADEON_RB3D_ZMASKOFFSET, 0); + /* need ctlstat, otherwise get some strange black flickering */ + OUT_RING_REG(RADEON_RB3D_ZCACHE_CTLSTAT, + RADEON_RB3D_ZC_FLUSH_ALL); + ADVANCE_RING(); + + for (i = 0; i < nbox; i++) { + int tileoffset, nrtilesx, nrtilesy, j; + /* it looks like r200 needs rv-style clears, at least if hierz is not enabled? */ + if ((dev_priv->flags & RADEON_HAS_HIERZ) + && !(dev_priv->microcode_version == UCODE_R200)) { + /* FIXME : figure this out for r200 (when hierz is enabled). Or + maybe r200 actually doesn't need to put the low-res z value into + the tile cache like r100, but just needs to clear the hi-level z-buffer? + Works for R100, both with hierz and without. + R100 seems to operate on 2x1 8x8 tiles, but... + odd: offset/nrtiles need to be 64 pix (4 block) aligned? Potentially + problematic with resolutions which are not 64 pix aligned? */ + tileoffset = + ((pbox[i].y1 >> 3) * depthpixperline + + pbox[i].x1) >> 6; + nrtilesx = + ((pbox[i].x2 & ~63) - + (pbox[i].x1 & ~63)) >> 4; + nrtilesy = + (pbox[i].y2 >> 3) - (pbox[i].y1 >> 3); + for (j = 0; j <= nrtilesy; j++) { + BEGIN_RING(4); + OUT_RING(CP_PACKET3 + (RADEON_3D_CLEAR_ZMASK, 2)); + /* first tile */ + OUT_RING(tileoffset * 8); + /* the number of tiles to clear */ + OUT_RING(nrtilesx + 4); + /* clear mask : chooses the clearing pattern. */ + OUT_RING(clearmask); + ADVANCE_RING(); + tileoffset += depthpixperline >> 6; + } + } else if (dev_priv->microcode_version == UCODE_R200) { + /* works for rv250. */ + /* find first macro tile (8x2 4x4 z-pixels on rv250) */ + tileoffset = + ((pbox[i].y1 >> 3) * depthpixperline + + pbox[i].x1) >> 5; + nrtilesx = + (pbox[i].x2 >> 5) - (pbox[i].x1 >> 5); + nrtilesy = + (pbox[i].y2 >> 3) - (pbox[i].y1 >> 3); + for (j = 0; j <= nrtilesy; j++) { + BEGIN_RING(4); + OUT_RING(CP_PACKET3 + (RADEON_3D_CLEAR_ZMASK, 2)); + /* first tile */ + /* judging by the first tile offset needed, could possibly + directly address/clear 4x4 tiles instead of 8x2 * 4x4 + macro tiles, though would still need clear mask for + right/bottom if truly 4x4 granularity is desired ? */ + OUT_RING(tileoffset * 16); + /* the number of tiles to clear */ + OUT_RING(nrtilesx + 1); + /* clear mask : chooses the clearing pattern. */ + OUT_RING(clearmask); + ADVANCE_RING(); + tileoffset += depthpixperline >> 5; + } + } else { /* rv 100 */ + /* rv100 might not need 64 pix alignment, who knows */ + /* offsets are, hmm, weird */ + tileoffset = + ((pbox[i].y1 >> 4) * depthpixperline + + pbox[i].x1) >> 6; + nrtilesx = + ((pbox[i].x2 & ~63) - + (pbox[i].x1 & ~63)) >> 4; + nrtilesy = + (pbox[i].y2 >> 4) - (pbox[i].y1 >> 4); + for (j = 0; j <= nrtilesy; j++) { + BEGIN_RING(4); + OUT_RING(CP_PACKET3 + (RADEON_3D_CLEAR_ZMASK, 2)); + OUT_RING(tileoffset * 128); + /* the number of tiles to clear */ + OUT_RING(nrtilesx + 4); + /* clear mask : chooses the clearing pattern. */ + OUT_RING(clearmask); + ADVANCE_RING(); + tileoffset += depthpixperline >> 6; + } + } + } + + /* TODO don't always clear all hi-level z tiles */ + if ((dev_priv->flags & RADEON_HAS_HIERZ) + && (dev_priv->microcode_version == UCODE_R200) + && (flags & RADEON_USE_HIERZ)) + /* r100 and cards without hierarchical z-buffer have no high-level z-buffer */ + /* FIXME : the mask supposedly contains low-res z values. So can't set + just to the max (0xff? or actually 0x3fff?), need to take z clear + value into account? */ + { + BEGIN_RING(4); + OUT_RING(CP_PACKET3(RADEON_3D_CLEAR_HIZ, 2)); + OUT_RING(0x0); /* First tile */ + OUT_RING(0x3cc0); + OUT_RING((0xff << 22) | (0xff << 6) | 0x003f003f); + ADVANCE_RING(); + } + } + + /* We have to clear the depth and/or stencil buffers by + * rendering a quad into just those buffers. Thus, we have to + * make sure the 3D engine is configured correctly. + */ + else if ((dev_priv->microcode_version == UCODE_R200) && + (flags & (RADEON_DEPTH | RADEON_STENCIL))) { + + int tempPP_CNTL; + int tempRE_CNTL; + int tempRB3D_CNTL; + int tempRB3D_ZSTENCILCNTL; + int tempRB3D_STENCILREFMASK; + int tempRB3D_PLANEMASK; + int tempSE_CNTL; + int tempSE_VTE_CNTL; + int tempSE_VTX_FMT_0; + int tempSE_VTX_FMT_1; + int tempSE_VAP_CNTL; + int tempRE_AUX_SCISSOR_CNTL; + + tempPP_CNTL = 0; + tempRE_CNTL = 0; + + tempRB3D_CNTL = depth_clear->rb3d_cntl; + + tempRB3D_ZSTENCILCNTL = depth_clear->rb3d_zstencilcntl; + tempRB3D_STENCILREFMASK = 0x0; + + tempSE_CNTL = depth_clear->se_cntl; + + /* Disable TCL */ + + tempSE_VAP_CNTL = ( /* SE_VAP_CNTL__FORCE_W_TO_ONE_MASK | */ + (0x9 << + SE_VAP_CNTL__VF_MAX_VTX_NUM__SHIFT)); + + tempRB3D_PLANEMASK = 0x0; + + tempRE_AUX_SCISSOR_CNTL = 0x0; + + tempSE_VTE_CNTL = + SE_VTE_CNTL__VTX_XY_FMT_MASK | SE_VTE_CNTL__VTX_Z_FMT_MASK; + + /* Vertex format (X, Y, Z, W) */ + tempSE_VTX_FMT_0 = + SE_VTX_FMT_0__VTX_Z0_PRESENT_MASK | + SE_VTX_FMT_0__VTX_W0_PRESENT_MASK; + tempSE_VTX_FMT_1 = 0x0; + + /* + * Depth buffer specific enables + */ + if (flags & RADEON_DEPTH) { + /* Enable depth buffer */ + tempRB3D_CNTL |= RADEON_Z_ENABLE; + } else { + /* Disable depth buffer */ + tempRB3D_CNTL &= ~RADEON_Z_ENABLE; + } + + /* + * Stencil buffer specific enables + */ + if (flags & RADEON_STENCIL) { + tempRB3D_CNTL |= RADEON_STENCIL_ENABLE; + tempRB3D_STENCILREFMASK = clear->depth_mask; + } else { + tempRB3D_CNTL &= ~RADEON_STENCIL_ENABLE; + tempRB3D_STENCILREFMASK = 0x00000000; + } + + if (flags & RADEON_USE_COMP_ZBUF) { + tempRB3D_ZSTENCILCNTL |= RADEON_Z_COMPRESSION_ENABLE | + RADEON_Z_DECOMPRESSION_ENABLE; + } + if (flags & RADEON_USE_HIERZ) { + tempRB3D_ZSTENCILCNTL |= RADEON_Z_HIERARCHY_ENABLE; + } + + BEGIN_RING(26); + RADEON_WAIT_UNTIL_2D_IDLE(); + + OUT_RING_REG(RADEON_PP_CNTL, tempPP_CNTL); + OUT_RING_REG(R200_RE_CNTL, tempRE_CNTL); + OUT_RING_REG(RADEON_RB3D_CNTL, tempRB3D_CNTL); + OUT_RING_REG(RADEON_RB3D_ZSTENCILCNTL, tempRB3D_ZSTENCILCNTL); + OUT_RING_REG(RADEON_RB3D_STENCILREFMASK, + tempRB3D_STENCILREFMASK); + OUT_RING_REG(RADEON_RB3D_PLANEMASK, tempRB3D_PLANEMASK); + OUT_RING_REG(RADEON_SE_CNTL, tempSE_CNTL); + OUT_RING_REG(R200_SE_VTE_CNTL, tempSE_VTE_CNTL); + OUT_RING_REG(R200_SE_VTX_FMT_0, tempSE_VTX_FMT_0); + OUT_RING_REG(R200_SE_VTX_FMT_1, tempSE_VTX_FMT_1); + OUT_RING_REG(R200_SE_VAP_CNTL, tempSE_VAP_CNTL); + OUT_RING_REG(R200_RE_AUX_SCISSOR_CNTL, tempRE_AUX_SCISSOR_CNTL); + ADVANCE_RING(); + + /* Make sure we restore the 3D state next time. + */ + sarea_priv->ctx_owner = 0; + + for (i = 0; i < nbox; i++) { + + /* Funny that this should be required -- + * sets top-left? + */ + radeon_emit_clip_rect(dev_priv, &sarea_priv->boxes[i]); + + BEGIN_RING(14); + OUT_RING(CP_PACKET3(R200_3D_DRAW_IMMD_2, 12)); + OUT_RING((RADEON_PRIM_TYPE_RECT_LIST | + RADEON_PRIM_WALK_RING | + (3 << RADEON_NUM_VERTICES_SHIFT))); + OUT_RING(depth_boxes[i].ui[CLEAR_X1]); + OUT_RING(depth_boxes[i].ui[CLEAR_Y1]); + OUT_RING(depth_boxes[i].ui[CLEAR_DEPTH]); + OUT_RING(0x3f800000); + OUT_RING(depth_boxes[i].ui[CLEAR_X1]); + OUT_RING(depth_boxes[i].ui[CLEAR_Y2]); + OUT_RING(depth_boxes[i].ui[CLEAR_DEPTH]); + OUT_RING(0x3f800000); + OUT_RING(depth_boxes[i].ui[CLEAR_X2]); + OUT_RING(depth_boxes[i].ui[CLEAR_Y2]); + OUT_RING(depth_boxes[i].ui[CLEAR_DEPTH]); + OUT_RING(0x3f800000); + ADVANCE_RING(); + } + } else if ((flags & (RADEON_DEPTH | RADEON_STENCIL))) { + + int tempRB3D_ZSTENCILCNTL = depth_clear->rb3d_zstencilcntl; + + rb3d_cntl = depth_clear->rb3d_cntl; + + if (flags & RADEON_DEPTH) { + rb3d_cntl |= RADEON_Z_ENABLE; + } else { + rb3d_cntl &= ~RADEON_Z_ENABLE; + } + + if (flags & RADEON_STENCIL) { + rb3d_cntl |= RADEON_STENCIL_ENABLE; + rb3d_stencilrefmask = clear->depth_mask; /* misnamed field */ + } else { + rb3d_cntl &= ~RADEON_STENCIL_ENABLE; + rb3d_stencilrefmask = 0x00000000; + } + + if (flags & RADEON_USE_COMP_ZBUF) { + tempRB3D_ZSTENCILCNTL |= RADEON_Z_COMPRESSION_ENABLE | + RADEON_Z_DECOMPRESSION_ENABLE; + } + if (flags & RADEON_USE_HIERZ) { + tempRB3D_ZSTENCILCNTL |= RADEON_Z_HIERARCHY_ENABLE; + } + + BEGIN_RING(13); + RADEON_WAIT_UNTIL_2D_IDLE(); + + OUT_RING(CP_PACKET0(RADEON_PP_CNTL, 1)); + OUT_RING(0x00000000); + OUT_RING(rb3d_cntl); + + OUT_RING_REG(RADEON_RB3D_ZSTENCILCNTL, tempRB3D_ZSTENCILCNTL); + OUT_RING_REG(RADEON_RB3D_STENCILREFMASK, rb3d_stencilrefmask); + OUT_RING_REG(RADEON_RB3D_PLANEMASK, 0x00000000); + OUT_RING_REG(RADEON_SE_CNTL, depth_clear->se_cntl); + ADVANCE_RING(); + + /* Make sure we restore the 3D state next time. + */ + sarea_priv->ctx_owner = 0; + + for (i = 0; i < nbox; i++) { + + /* Funny that this should be required -- + * sets top-left? + */ + radeon_emit_clip_rect(dev_priv, &sarea_priv->boxes[i]); + + BEGIN_RING(15); + + OUT_RING(CP_PACKET3(RADEON_3D_DRAW_IMMD, 13)); + OUT_RING(RADEON_VTX_Z_PRESENT | + RADEON_VTX_PKCOLOR_PRESENT); + OUT_RING((RADEON_PRIM_TYPE_RECT_LIST | + RADEON_PRIM_WALK_RING | + RADEON_MAOS_ENABLE | + RADEON_VTX_FMT_RADEON_MODE | + (3 << RADEON_NUM_VERTICES_SHIFT))); + + OUT_RING(depth_boxes[i].ui[CLEAR_X1]); + OUT_RING(depth_boxes[i].ui[CLEAR_Y1]); + OUT_RING(depth_boxes[i].ui[CLEAR_DEPTH]); + OUT_RING(0x0); + + OUT_RING(depth_boxes[i].ui[CLEAR_X1]); + OUT_RING(depth_boxes[i].ui[CLEAR_Y2]); + OUT_RING(depth_boxes[i].ui[CLEAR_DEPTH]); + OUT_RING(0x0); + + OUT_RING(depth_boxes[i].ui[CLEAR_X2]); + OUT_RING(depth_boxes[i].ui[CLEAR_Y2]); + OUT_RING(depth_boxes[i].ui[CLEAR_DEPTH]); + OUT_RING(0x0); + + ADVANCE_RING(); + } + } + + /* Increment the clear counter. The client-side 3D driver must + * wait on this value before performing the clear ioctl. We + * need this because the card's so damned fast... + */ + sarea_priv->last_clear++; + + BEGIN_RING(4); + + RADEON_CLEAR_AGE(sarea_priv->last_clear); + RADEON_WAIT_UNTIL_IDLE(); + + ADVANCE_RING(); +} + +static void radeon_cp_dispatch_swap(struct drm_device *dev, struct drm_master *master) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + struct drm_radeon_master_private *master_priv = master->driver_priv; + drm_radeon_sarea_t *sarea_priv = master_priv->sarea_priv; + int nbox = sarea_priv->nbox; + struct drm_clip_rect *pbox = sarea_priv->boxes; + int i; + RING_LOCALS; + DRM_DEBUG("\n"); + + /* Do some trivial performance monitoring... + */ + if (dev_priv->do_boxes) + radeon_cp_performance_boxes(dev_priv, master_priv); + + /* Wait for the 3D stream to idle before dispatching the bitblt. + * This will prevent data corruption between the two streams. + */ + BEGIN_RING(2); + + RADEON_WAIT_UNTIL_3D_IDLE(); + + ADVANCE_RING(); + + for (i = 0; i < nbox; i++) { + int x = pbox[i].x1; + int y = pbox[i].y1; + int w = pbox[i].x2 - x; + int h = pbox[i].y2 - y; + + DRM_DEBUG("%d,%d-%d,%d\n", x, y, w, h); + + BEGIN_RING(9); + + OUT_RING(CP_PACKET0(RADEON_DP_GUI_MASTER_CNTL, 0)); + OUT_RING(RADEON_GMC_SRC_PITCH_OFFSET_CNTL | + RADEON_GMC_DST_PITCH_OFFSET_CNTL | + RADEON_GMC_BRUSH_NONE | + (dev_priv->color_fmt << 8) | + RADEON_GMC_SRC_DATATYPE_COLOR | + RADEON_ROP3_S | + RADEON_DP_SRC_SOURCE_MEMORY | + RADEON_GMC_CLR_CMP_CNTL_DIS | RADEON_GMC_WR_MSK_DIS); + + /* Make this work even if front & back are flipped: + */ + OUT_RING(CP_PACKET0(RADEON_SRC_PITCH_OFFSET, 1)); + if (sarea_priv->pfCurrentPage == 0) { + OUT_RING(dev_priv->back_pitch_offset); + OUT_RING(dev_priv->front_pitch_offset); + } else { + OUT_RING(dev_priv->front_pitch_offset); + OUT_RING(dev_priv->back_pitch_offset); + } + + OUT_RING(CP_PACKET0(RADEON_SRC_X_Y, 2)); + OUT_RING((x << 16) | y); + OUT_RING((x << 16) | y); + OUT_RING((w << 16) | h); + + ADVANCE_RING(); + } + + /* Increment the frame counter. The client-side 3D driver must + * throttle the framerate by waiting for this value before + * performing the swapbuffer ioctl. + */ + sarea_priv->last_frame++; + + BEGIN_RING(4); + + RADEON_FRAME_AGE(sarea_priv->last_frame); + RADEON_WAIT_UNTIL_2D_IDLE(); + + ADVANCE_RING(); +} + +void radeon_cp_dispatch_flip(struct drm_device *dev, struct drm_master *master) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + struct drm_radeon_master_private *master_priv = master->driver_priv; + struct drm_sarea *sarea = (struct drm_sarea *)master_priv->sarea->handle; + int offset = (master_priv->sarea_priv->pfCurrentPage == 1) + ? dev_priv->front_offset : dev_priv->back_offset; + RING_LOCALS; + DRM_DEBUG("pfCurrentPage=%d\n", + master_priv->sarea_priv->pfCurrentPage); + + /* Do some trivial performance monitoring... + */ + if (dev_priv->do_boxes) { + dev_priv->stats.boxes |= RADEON_BOX_FLIP; + radeon_cp_performance_boxes(dev_priv, master_priv); + } + + /* Update the frame offsets for both CRTCs + */ + BEGIN_RING(6); + + RADEON_WAIT_UNTIL_3D_IDLE(); + OUT_RING_REG(RADEON_CRTC_OFFSET, + ((sarea->frame.y * dev_priv->front_pitch + + sarea->frame.x * (dev_priv->color_fmt - 2)) & ~7) + + offset); + OUT_RING_REG(RADEON_CRTC2_OFFSET, master_priv->sarea_priv->crtc2_base + + offset); + + ADVANCE_RING(); + + /* Increment the frame counter. The client-side 3D driver must + * throttle the framerate by waiting for this value before + * performing the swapbuffer ioctl. + */ + master_priv->sarea_priv->last_frame++; + master_priv->sarea_priv->pfCurrentPage = + 1 - master_priv->sarea_priv->pfCurrentPage; + + BEGIN_RING(2); + + RADEON_FRAME_AGE(master_priv->sarea_priv->last_frame); + + ADVANCE_RING(); +} + +static int bad_prim_vertex_nr(int primitive, int nr) +{ + switch (primitive & RADEON_PRIM_TYPE_MASK) { + case RADEON_PRIM_TYPE_NONE: + case RADEON_PRIM_TYPE_POINT: + return nr < 1; + case RADEON_PRIM_TYPE_LINE: + return (nr & 1) || nr == 0; + case RADEON_PRIM_TYPE_LINE_STRIP: + return nr < 2; + case RADEON_PRIM_TYPE_TRI_LIST: + case RADEON_PRIM_TYPE_3VRT_POINT_LIST: + case RADEON_PRIM_TYPE_3VRT_LINE_LIST: + case RADEON_PRIM_TYPE_RECT_LIST: + return nr % 3 || nr == 0; + case RADEON_PRIM_TYPE_TRI_FAN: + case RADEON_PRIM_TYPE_TRI_STRIP: + return nr < 3; + default: + return 1; + } +} + +typedef struct { + unsigned int start; + unsigned int finish; + unsigned int prim; + unsigned int numverts; + unsigned int offset; + unsigned int vc_format; +} drm_radeon_tcl_prim_t; + +static void radeon_cp_dispatch_vertex(struct drm_device * dev, + struct drm_file *file_priv, + struct drm_buf * buf, + drm_radeon_tcl_prim_t * prim) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + struct drm_radeon_master_private *master_priv = file_priv->master->driver_priv; + drm_radeon_sarea_t *sarea_priv = master_priv->sarea_priv; + int offset = dev_priv->gart_buffers_offset + buf->offset + prim->start; + int numverts = (int)prim->numverts; + int nbox = sarea_priv->nbox; + int i = 0; + RING_LOCALS; + + DRM_DEBUG("hwprim 0x%x vfmt 0x%x %d..%d %d verts\n", + prim->prim, + prim->vc_format, prim->start, prim->finish, prim->numverts); + + if (bad_prim_vertex_nr(prim->prim, prim->numverts)) { + DRM_ERROR("bad prim %x numverts %d\n", + prim->prim, prim->numverts); + return; + } + + do { + /* Emit the next cliprect */ + if (i < nbox) { + radeon_emit_clip_rect(dev_priv, &sarea_priv->boxes[i]); + } + + /* Emit the vertex buffer rendering commands */ + BEGIN_RING(5); + + OUT_RING(CP_PACKET3(RADEON_3D_RNDR_GEN_INDX_PRIM, 3)); + OUT_RING(offset); + OUT_RING(numverts); + OUT_RING(prim->vc_format); + OUT_RING(prim->prim | RADEON_PRIM_WALK_LIST | + RADEON_COLOR_ORDER_RGBA | + RADEON_VTX_FMT_RADEON_MODE | + (numverts << RADEON_NUM_VERTICES_SHIFT)); + + ADVANCE_RING(); + + i++; + } while (i < nbox); +} + +void radeon_cp_discard_buffer(struct drm_device *dev, struct drm_master *master, struct drm_buf *buf) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + struct drm_radeon_master_private *master_priv = master->driver_priv; + drm_radeon_buf_priv_t *buf_priv = buf->dev_private; + RING_LOCALS; + + buf_priv->age = ++master_priv->sarea_priv->last_dispatch; + + /* Emit the vertex buffer age */ + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) { + BEGIN_RING(3); + R600_DISPATCH_AGE(buf_priv->age); + ADVANCE_RING(); + } else { + BEGIN_RING(2); + RADEON_DISPATCH_AGE(buf_priv->age); + ADVANCE_RING(); + } + + buf->pending = 1; + buf->used = 0; +} + +static void radeon_cp_dispatch_indirect(struct drm_device * dev, + struct drm_buf * buf, int start, int end) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + RING_LOCALS; + DRM_DEBUG("buf=%d s=0x%x e=0x%x\n", buf->idx, start, end); + + if (start != end) { + int offset = (dev_priv->gart_buffers_offset + + buf->offset + start); + int dwords = (end - start + 3) / sizeof(u32); + + /* Indirect buffer data must be an even number of + * dwords, so if we've been given an odd number we must + * pad the data with a Type-2 CP packet. + */ + if (dwords & 1) { + u32 *data = (u32 *) + ((char *)dev->agp_buffer_map->handle + + buf->offset + start); + data[dwords++] = RADEON_CP_PACKET2; + } + + /* Fire off the indirect buffer */ + BEGIN_RING(3); + + OUT_RING(CP_PACKET0(RADEON_CP_IB_BASE, 1)); + OUT_RING(offset); + OUT_RING(dwords); + + ADVANCE_RING(); + } +} + +static void radeon_cp_dispatch_indices(struct drm_device *dev, + struct drm_master *master, + struct drm_buf * elt_buf, + drm_radeon_tcl_prim_t * prim) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + struct drm_radeon_master_private *master_priv = master->driver_priv; + drm_radeon_sarea_t *sarea_priv = master_priv->sarea_priv; + int offset = dev_priv->gart_buffers_offset + prim->offset; + u32 *data; + int dwords; + int i = 0; + int start = prim->start + RADEON_INDEX_PRIM_OFFSET; + int count = (prim->finish - start) / sizeof(u16); + int nbox = sarea_priv->nbox; + + DRM_DEBUG("hwprim 0x%x vfmt 0x%x %d..%d offset: %x nr %d\n", + prim->prim, + prim->vc_format, + prim->start, prim->finish, prim->offset, prim->numverts); + + if (bad_prim_vertex_nr(prim->prim, count)) { + DRM_ERROR("bad prim %x count %d\n", prim->prim, count); + return; + } + + if (start >= prim->finish || (prim->start & 0x7)) { + DRM_ERROR("buffer prim %d\n", prim->prim); + return; + } + + dwords = (prim->finish - prim->start + 3) / sizeof(u32); + + data = (u32 *) ((char *)dev->agp_buffer_map->handle + + elt_buf->offset + prim->start); + + data[0] = CP_PACKET3(RADEON_3D_RNDR_GEN_INDX_PRIM, dwords - 2); + data[1] = offset; + data[2] = prim->numverts; + data[3] = prim->vc_format; + data[4] = (prim->prim | + RADEON_PRIM_WALK_IND | + RADEON_COLOR_ORDER_RGBA | + RADEON_VTX_FMT_RADEON_MODE | + (count << RADEON_NUM_VERTICES_SHIFT)); + + do { + if (i < nbox) + radeon_emit_clip_rect(dev_priv, &sarea_priv->boxes[i]); + + radeon_cp_dispatch_indirect(dev, elt_buf, + prim->start, prim->finish); + + i++; + } while (i < nbox); + +} + +#define RADEON_MAX_TEXTURE_SIZE RADEON_BUFFER_SIZE + +static int radeon_cp_dispatch_texture(struct drm_device * dev, + struct drm_file *file_priv, + drm_radeon_texture_t * tex, + drm_radeon_tex_image_t * image) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + struct drm_buf *buf; + u32 format; + u32 *buffer; + const u8 __user *data; + int size, dwords, tex_width, blit_width, spitch; + u32 height; + int i; + u32 texpitch, microtile; + u32 offset, byte_offset; + RING_LOCALS; + + if (radeon_check_and_fixup_offset(dev_priv, file_priv, &tex->offset)) { + DRM_ERROR("Invalid destination offset\n"); + return -EINVAL; + } + + dev_priv->stats.boxes |= RADEON_BOX_TEXTURE_LOAD; + + /* Flush the pixel cache. This ensures no pixel data gets mixed + * up with the texture data from the host data blit, otherwise + * part of the texture image may be corrupted. + */ + BEGIN_RING(4); + RADEON_FLUSH_CACHE(); + RADEON_WAIT_UNTIL_IDLE(); + ADVANCE_RING(); + + /* The compiler won't optimize away a division by a variable, + * even if the only legal values are powers of two. Thus, we'll + * use a shift instead. + */ + switch (tex->format) { + case RADEON_TXFORMAT_ARGB8888: + case RADEON_TXFORMAT_RGBA8888: + format = RADEON_COLOR_FORMAT_ARGB8888; + tex_width = tex->width * 4; + blit_width = image->width * 4; + break; + case RADEON_TXFORMAT_AI88: + case RADEON_TXFORMAT_ARGB1555: + case RADEON_TXFORMAT_RGB565: + case RADEON_TXFORMAT_ARGB4444: + case RADEON_TXFORMAT_VYUY422: + case RADEON_TXFORMAT_YVYU422: + format = RADEON_COLOR_FORMAT_RGB565; + tex_width = tex->width * 2; + blit_width = image->width * 2; + break; + case RADEON_TXFORMAT_I8: + case RADEON_TXFORMAT_RGB332: + format = RADEON_COLOR_FORMAT_CI8; + tex_width = tex->width * 1; + blit_width = image->width * 1; + break; + default: + DRM_ERROR("invalid texture format %d\n", tex->format); + return -EINVAL; + } + spitch = blit_width >> 6; + if (spitch == 0 && image->height > 1) + return -EINVAL; + + texpitch = tex->pitch; + if ((texpitch << 22) & RADEON_DST_TILE_MICRO) { + microtile = 1; + if (tex_width < 64) { + texpitch &= ~(RADEON_DST_TILE_MICRO >> 22); + /* we got tiled coordinates, untile them */ + image->x *= 2; + } + } else + microtile = 0; + + /* this might fail for zero-sized uploads - are those illegal? */ + if (!radeon_check_offset(dev_priv, tex->offset + image->height * + blit_width - 1)) { + DRM_ERROR("Invalid final destination offset\n"); + return -EINVAL; + } + + DRM_DEBUG("tex=%dx%d blit=%d\n", tex_width, tex->height, blit_width); + + do { + DRM_DEBUG("tex: ofs=0x%x p=%d f=%d x=%hd y=%hd w=%hd h=%hd\n", + tex->offset >> 10, tex->pitch, tex->format, + image->x, image->y, image->width, image->height); + + /* Make a copy of some parameters in case we have to + * update them for a multi-pass texture blit. + */ + height = image->height; + data = (const u8 __user *)image->data; + + size = height * blit_width; + + if (size > RADEON_MAX_TEXTURE_SIZE) { + height = RADEON_MAX_TEXTURE_SIZE / blit_width; + size = height * blit_width; + } else if (size < 4 && size > 0) { + size = 4; + } else if (size == 0) { + return 0; + } + + buf = radeon_freelist_get(dev); + if (0 && !buf) { + radeon_do_cp_idle(dev_priv); + buf = radeon_freelist_get(dev); + } + if (!buf) { + DRM_DEBUG("EAGAIN\n"); + if (DRM_COPY_TO_USER(tex->image, image, sizeof(*image))) + return -EFAULT; + return -EAGAIN; + } + + /* Dispatch the indirect buffer. + */ + buffer = + (u32 *) ((char *)dev->agp_buffer_map->handle + buf->offset); + dwords = size / 4; + +#define RADEON_COPY_MT(_buf, _data, _width) \ + do { \ + if (DRM_COPY_FROM_USER(_buf, _data, (_width))) {\ + DRM_ERROR("EFAULT on pad, %d bytes\n", (_width)); \ + return -EFAULT; \ + } \ + } while(0) + + if (microtile) { + /* texture micro tiling in use, minimum texture width is thus 16 bytes. + however, we cannot use blitter directly for texture width < 64 bytes, + since minimum tex pitch is 64 bytes and we need this to match + the texture width, otherwise the blitter will tile it wrong. + Thus, tiling manually in this case. Additionally, need to special + case tex height = 1, since our actual image will have height 2 + and we need to ensure we don't read beyond the texture size + from user space. */ + if (tex->height == 1) { + if (tex_width >= 64 || tex_width <= 16) { + RADEON_COPY_MT(buffer, data, + (int)(tex_width * sizeof(u32))); + } else if (tex_width == 32) { + RADEON_COPY_MT(buffer, data, 16); + RADEON_COPY_MT(buffer + 8, + data + 16, 16); + } + } else if (tex_width >= 64 || tex_width == 16) { + RADEON_COPY_MT(buffer, data, + (int)(dwords * sizeof(u32))); + } else if (tex_width < 16) { + for (i = 0; i < tex->height; i++) { + RADEON_COPY_MT(buffer, data, tex_width); + buffer += 4; + data += tex_width; + } + } else if (tex_width == 32) { + /* TODO: make sure this works when not fitting in one buffer + (i.e. 32bytes x 2048...) */ + for (i = 0; i < tex->height; i += 2) { + RADEON_COPY_MT(buffer, data, 16); + data += 16; + RADEON_COPY_MT(buffer + 8, data, 16); + data += 16; + RADEON_COPY_MT(buffer + 4, data, 16); + data += 16; + RADEON_COPY_MT(buffer + 12, data, 16); + data += 16; + buffer += 16; + } + } + } else { + if (tex_width >= 32) { + /* Texture image width is larger than the minimum, so we + * can upload it directly. + */ + RADEON_COPY_MT(buffer, data, + (int)(dwords * sizeof(u32))); + } else { + /* Texture image width is less than the minimum, so we + * need to pad out each image scanline to the minimum + * width. + */ + for (i = 0; i < tex->height; i++) { + RADEON_COPY_MT(buffer, data, tex_width); + buffer += 8; + data += tex_width; + } + } + } + +#undef RADEON_COPY_MT + byte_offset = (image->y & ~2047) * blit_width; + buf->file_priv = file_priv; + buf->used = size; + offset = dev_priv->gart_buffers_offset + buf->offset; + BEGIN_RING(9); + OUT_RING(CP_PACKET3(RADEON_CNTL_BITBLT_MULTI, 5)); + OUT_RING(RADEON_GMC_SRC_PITCH_OFFSET_CNTL | + RADEON_GMC_DST_PITCH_OFFSET_CNTL | + RADEON_GMC_BRUSH_NONE | + (format << 8) | + RADEON_GMC_SRC_DATATYPE_COLOR | + RADEON_ROP3_S | + RADEON_DP_SRC_SOURCE_MEMORY | + RADEON_GMC_CLR_CMP_CNTL_DIS | RADEON_GMC_WR_MSK_DIS); + OUT_RING((spitch << 22) | (offset >> 10)); + OUT_RING((texpitch << 22) | ((tex->offset >> 10) + (byte_offset >> 10))); + OUT_RING(0); + OUT_RING((image->x << 16) | (image->y % 2048)); + OUT_RING((image->width << 16) | height); + RADEON_WAIT_UNTIL_2D_IDLE(); + ADVANCE_RING(); + COMMIT_RING(); + + radeon_cp_discard_buffer(dev, file_priv->master, buf); + + /* Update the input parameters for next time */ + image->y += height; + image->height -= height; + image->data = (const u8 __user *)image->data + size; + } while (image->height > 0); + + /* Flush the pixel cache after the blit completes. This ensures + * the texture data is written out to memory before rendering + * continues. + */ + BEGIN_RING(4); + RADEON_FLUSH_CACHE(); + RADEON_WAIT_UNTIL_2D_IDLE(); + ADVANCE_RING(); + COMMIT_RING(); + + return 0; +} + +static void radeon_cp_dispatch_stipple(struct drm_device * dev, u32 * stipple) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + int i; + RING_LOCALS; + DRM_DEBUG("\n"); + + BEGIN_RING(35); + + OUT_RING(CP_PACKET0(RADEON_RE_STIPPLE_ADDR, 0)); + OUT_RING(0x00000000); + + OUT_RING(CP_PACKET0_TABLE(RADEON_RE_STIPPLE_DATA, 31)); + for (i = 0; i < 32; i++) { + OUT_RING(stipple[i]); + } + + ADVANCE_RING(); +} + +static void radeon_apply_surface_regs(int surf_index, + drm_radeon_private_t *dev_priv) +{ + if (!dev_priv->mmio) + return; + + radeon_do_cp_idle(dev_priv); + + RADEON_WRITE(RADEON_SURFACE0_INFO + 16 * surf_index, + dev_priv->surfaces[surf_index].flags); + RADEON_WRITE(RADEON_SURFACE0_LOWER_BOUND + 16 * surf_index, + dev_priv->surfaces[surf_index].lower); + RADEON_WRITE(RADEON_SURFACE0_UPPER_BOUND + 16 * surf_index, + dev_priv->surfaces[surf_index].upper); +} + +/* Allocates a virtual surface + * doesn't always allocate a real surface, will stretch an existing + * surface when possible. + * + * Note that refcount can be at most 2, since during a free refcount=3 + * might mean we have to allocate a new surface which might not always + * be available. + * For example : we allocate three contiguous surfaces ABC. If B is + * freed, we suddenly need two surfaces to store A and C, which might + * not always be available. + */ +static int alloc_surface(drm_radeon_surface_alloc_t *new, + drm_radeon_private_t *dev_priv, + struct drm_file *file_priv) +{ + struct radeon_virt_surface *s; + int i; + int virt_surface_index; + uint32_t new_upper, new_lower; + + new_lower = new->address; + new_upper = new_lower + new->size - 1; + + /* sanity check */ + if ((new_lower >= new_upper) || (new->flags == 0) || (new->size == 0) || + ((new_upper & RADEON_SURF_ADDRESS_FIXED_MASK) != + RADEON_SURF_ADDRESS_FIXED_MASK) + || ((new_lower & RADEON_SURF_ADDRESS_FIXED_MASK) != 0)) + return -1; + + /* make sure there is no overlap with existing surfaces */ + for (i = 0; i < RADEON_MAX_SURFACES; i++) { + if ((dev_priv->surfaces[i].refcount != 0) && + (((new_lower >= dev_priv->surfaces[i].lower) && + (new_lower < dev_priv->surfaces[i].upper)) || + ((new_lower < dev_priv->surfaces[i].lower) && + (new_upper > dev_priv->surfaces[i].lower)))) { + return -1; + } + } + + /* find a virtual surface */ + for (i = 0; i < 2 * RADEON_MAX_SURFACES; i++) + if (dev_priv->virt_surfaces[i].file_priv == NULL) + break; + if (i == 2 * RADEON_MAX_SURFACES) { + return -1; + } + virt_surface_index = i; + + /* try to reuse an existing surface */ + for (i = 0; i < RADEON_MAX_SURFACES; i++) { + /* extend before */ + if ((dev_priv->surfaces[i].refcount == 1) && + (new->flags == dev_priv->surfaces[i].flags) && + (new_upper + 1 == dev_priv->surfaces[i].lower)) { + s = &(dev_priv->virt_surfaces[virt_surface_index]); + s->surface_index = i; + s->lower = new_lower; + s->upper = new_upper; + s->flags = new->flags; + s->file_priv = file_priv; + dev_priv->surfaces[i].refcount++; + dev_priv->surfaces[i].lower = s->lower; + radeon_apply_surface_regs(s->surface_index, dev_priv); + return virt_surface_index; + } + + /* extend after */ + if ((dev_priv->surfaces[i].refcount == 1) && + (new->flags == dev_priv->surfaces[i].flags) && + (new_lower == dev_priv->surfaces[i].upper + 1)) { + s = &(dev_priv->virt_surfaces[virt_surface_index]); + s->surface_index = i; + s->lower = new_lower; + s->upper = new_upper; + s->flags = new->flags; + s->file_priv = file_priv; + dev_priv->surfaces[i].refcount++; + dev_priv->surfaces[i].upper = s->upper; + radeon_apply_surface_regs(s->surface_index, dev_priv); + return virt_surface_index; + } + } + + /* okay, we need a new one */ + for (i = 0; i < RADEON_MAX_SURFACES; i++) { + if (dev_priv->surfaces[i].refcount == 0) { + s = &(dev_priv->virt_surfaces[virt_surface_index]); + s->surface_index = i; + s->lower = new_lower; + s->upper = new_upper; + s->flags = new->flags; + s->file_priv = file_priv; + dev_priv->surfaces[i].refcount = 1; + dev_priv->surfaces[i].lower = s->lower; + dev_priv->surfaces[i].upper = s->upper; + dev_priv->surfaces[i].flags = s->flags; + radeon_apply_surface_regs(s->surface_index, dev_priv); + return virt_surface_index; + } + } + + /* we didn't find anything */ + return -1; +} + +static int free_surface(struct drm_file *file_priv, + drm_radeon_private_t * dev_priv, + int lower) +{ + struct radeon_virt_surface *s; + int i; + /* find the virtual surface */ + for (i = 0; i < 2 * RADEON_MAX_SURFACES; i++) { + s = &(dev_priv->virt_surfaces[i]); + if (s->file_priv) { + if ((lower == s->lower) && (file_priv == s->file_priv)) + { + if (dev_priv->surfaces[s->surface_index]. + lower == s->lower) + dev_priv->surfaces[s->surface_index]. + lower = s->upper; + + if (dev_priv->surfaces[s->surface_index]. + upper == s->upper) + dev_priv->surfaces[s->surface_index]. + upper = s->lower; + + dev_priv->surfaces[s->surface_index].refcount--; + if (dev_priv->surfaces[s->surface_index]. + refcount == 0) + dev_priv->surfaces[s->surface_index]. + flags = 0; + s->file_priv = NULL; + radeon_apply_surface_regs(s->surface_index, + dev_priv); + return 0; + } + } + } + return 1; +} + +static void radeon_surfaces_release(struct drm_file *file_priv, + drm_radeon_private_t * dev_priv) +{ + int i; + for (i = 0; i < 2 * RADEON_MAX_SURFACES; i++) { + if (dev_priv->virt_surfaces[i].file_priv == file_priv) + free_surface(file_priv, dev_priv, + dev_priv->virt_surfaces[i].lower); + } +} + +/* ================================================================ + * IOCTL functions + */ +static int radeon_surface_alloc(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + drm_radeon_surface_alloc_t *alloc = data; + + if (alloc_surface(alloc, dev_priv, file_priv) == -1) + return -EINVAL; + else + return 0; +} + +static int radeon_surface_free(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + drm_radeon_surface_free_t *memfree = data; + + if (free_surface(file_priv, dev_priv, memfree->address)) + return -EINVAL; + else + return 0; +} + +static int radeon_cp_clear(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + struct drm_radeon_master_private *master_priv = file_priv->master->driver_priv; + drm_radeon_sarea_t *sarea_priv = master_priv->sarea_priv; + drm_radeon_clear_t *clear = data; + drm_radeon_clear_rect_t depth_boxes[RADEON_NR_SAREA_CLIPRECTS]; + DRM_DEBUG("\n"); + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + RING_SPACE_TEST_WITH_RETURN(dev_priv); + + if (sarea_priv->nbox > RADEON_NR_SAREA_CLIPRECTS) + sarea_priv->nbox = RADEON_NR_SAREA_CLIPRECTS; + + if (DRM_COPY_FROM_USER(&depth_boxes, clear->depth_boxes, + sarea_priv->nbox * sizeof(depth_boxes[0]))) + return -EFAULT; + + radeon_cp_dispatch_clear(dev, file_priv->master, clear, depth_boxes); + + COMMIT_RING(); + return 0; +} + +/* Not sure why this isn't set all the time: + */ +static int radeon_do_init_pageflip(struct drm_device *dev, struct drm_master *master) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + struct drm_radeon_master_private *master_priv = master->driver_priv; + RING_LOCALS; + + DRM_DEBUG("\n"); + + BEGIN_RING(6); + RADEON_WAIT_UNTIL_3D_IDLE(); + OUT_RING(CP_PACKET0(RADEON_CRTC_OFFSET_CNTL, 0)); + OUT_RING(RADEON_READ(RADEON_CRTC_OFFSET_CNTL) | + RADEON_CRTC_OFFSET_FLIP_CNTL); + OUT_RING(CP_PACKET0(RADEON_CRTC2_OFFSET_CNTL, 0)); + OUT_RING(RADEON_READ(RADEON_CRTC2_OFFSET_CNTL) | + RADEON_CRTC_OFFSET_FLIP_CNTL); + ADVANCE_RING(); + + dev_priv->page_flipping = 1; + + if (master_priv->sarea_priv->pfCurrentPage != 1) + master_priv->sarea_priv->pfCurrentPage = 0; + + return 0; +} + +/* Swapping and flipping are different operations, need different ioctls. + * They can & should be intermixed to support multiple 3d windows. + */ +static int radeon_cp_flip(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + DRM_DEBUG("\n"); + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + RING_SPACE_TEST_WITH_RETURN(dev_priv); + + if (!dev_priv->page_flipping) + radeon_do_init_pageflip(dev, file_priv->master); + + radeon_cp_dispatch_flip(dev, file_priv->master); + + COMMIT_RING(); + return 0; +} + +static int radeon_cp_swap(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + struct drm_radeon_master_private *master_priv = file_priv->master->driver_priv; + drm_radeon_sarea_t *sarea_priv = master_priv->sarea_priv; + + DRM_DEBUG("\n"); + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + RING_SPACE_TEST_WITH_RETURN(dev_priv); + + if (sarea_priv->nbox > RADEON_NR_SAREA_CLIPRECTS) + sarea_priv->nbox = RADEON_NR_SAREA_CLIPRECTS; + + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) + r600_cp_dispatch_swap(dev, file_priv); + else + radeon_cp_dispatch_swap(dev, file_priv->master); + sarea_priv->ctx_owner = 0; + + COMMIT_RING(); + return 0; +} + +static int radeon_cp_vertex(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + struct drm_radeon_master_private *master_priv = file_priv->master->driver_priv; + drm_radeon_sarea_t *sarea_priv; + struct drm_device_dma *dma = dev->dma; + struct drm_buf *buf; + drm_radeon_vertex_t *vertex = data; + drm_radeon_tcl_prim_t prim; + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + sarea_priv = master_priv->sarea_priv; + + DRM_DEBUG("pid=%d index=%d count=%d discard=%d\n", + DRM_CURRENTPID, vertex->idx, vertex->count, vertex->discard); + + if (vertex->idx < 0 || vertex->idx >= dma->buf_count) { + DRM_ERROR("buffer index %d (of %d max)\n", + vertex->idx, dma->buf_count - 1); + return -EINVAL; + } + if (vertex->prim < 0 || vertex->prim > RADEON_PRIM_TYPE_3VRT_LINE_LIST) { + DRM_ERROR("buffer prim %d\n", vertex->prim); + return -EINVAL; + } + + RING_SPACE_TEST_WITH_RETURN(dev_priv); + VB_AGE_TEST_WITH_RETURN(dev_priv); + + buf = dma->buflist[vertex->idx]; + + if (buf->file_priv != file_priv) { + DRM_ERROR("process %d using buffer owned by %p\n", + DRM_CURRENTPID, buf->file_priv); + return -EINVAL; + } + if (buf->pending) { + DRM_ERROR("sending pending buffer %d\n", vertex->idx); + return -EINVAL; + } + + /* Build up a prim_t record: + */ + if (vertex->count) { + buf->used = vertex->count; /* not used? */ + + if (sarea_priv->dirty & ~RADEON_UPLOAD_CLIPRECTS) { + if (radeon_emit_state(dev_priv, file_priv, + &sarea_priv->context_state, + sarea_priv->tex_state, + sarea_priv->dirty)) { + DRM_ERROR("radeon_emit_state failed\n"); + return -EINVAL; + } + + sarea_priv->dirty &= ~(RADEON_UPLOAD_TEX0IMAGES | + RADEON_UPLOAD_TEX1IMAGES | + RADEON_UPLOAD_TEX2IMAGES | + RADEON_REQUIRE_QUIESCENCE); + } + + prim.start = 0; + prim.finish = vertex->count; /* unused */ + prim.prim = vertex->prim; + prim.numverts = vertex->count; + prim.vc_format = sarea_priv->vc_format; + + radeon_cp_dispatch_vertex(dev, file_priv, buf, &prim); + } + + if (vertex->discard) { + radeon_cp_discard_buffer(dev, file_priv->master, buf); + } + + COMMIT_RING(); + return 0; +} + +static int radeon_cp_indices(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + struct drm_radeon_master_private *master_priv = file_priv->master->driver_priv; + drm_radeon_sarea_t *sarea_priv; + struct drm_device_dma *dma = dev->dma; + struct drm_buf *buf; + drm_radeon_indices_t *elts = data; + drm_radeon_tcl_prim_t prim; + int count; + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + sarea_priv = master_priv->sarea_priv; + + DRM_DEBUG("pid=%d index=%d start=%d end=%d discard=%d\n", + DRM_CURRENTPID, elts->idx, elts->start, elts->end, + elts->discard); + + if (elts->idx < 0 || elts->idx >= dma->buf_count) { + DRM_ERROR("buffer index %d (of %d max)\n", + elts->idx, dma->buf_count - 1); + return -EINVAL; + } + if (elts->prim < 0 || elts->prim > RADEON_PRIM_TYPE_3VRT_LINE_LIST) { + DRM_ERROR("buffer prim %d\n", elts->prim); + return -EINVAL; + } + + RING_SPACE_TEST_WITH_RETURN(dev_priv); + VB_AGE_TEST_WITH_RETURN(dev_priv); + + buf = dma->buflist[elts->idx]; + + if (buf->file_priv != file_priv) { + DRM_ERROR("process %d using buffer owned by %p\n", + DRM_CURRENTPID, buf->file_priv); + return -EINVAL; + } + if (buf->pending) { + DRM_ERROR("sending pending buffer %d\n", elts->idx); + return -EINVAL; + } + + count = (elts->end - elts->start) / sizeof(u16); + elts->start -= RADEON_INDEX_PRIM_OFFSET; + + if (elts->start & 0x7) { + DRM_ERROR("misaligned buffer 0x%x\n", elts->start); + return -EINVAL; + } + if (elts->start < buf->used) { + DRM_ERROR("no header 0x%x - 0x%x\n", elts->start, buf->used); + return -EINVAL; + } + + buf->used = elts->end; + + if (sarea_priv->dirty & ~RADEON_UPLOAD_CLIPRECTS) { + if (radeon_emit_state(dev_priv, file_priv, + &sarea_priv->context_state, + sarea_priv->tex_state, + sarea_priv->dirty)) { + DRM_ERROR("radeon_emit_state failed\n"); + return -EINVAL; + } + + sarea_priv->dirty &= ~(RADEON_UPLOAD_TEX0IMAGES | + RADEON_UPLOAD_TEX1IMAGES | + RADEON_UPLOAD_TEX2IMAGES | + RADEON_REQUIRE_QUIESCENCE); + } + + /* Build up a prim_t record: + */ + prim.start = elts->start; + prim.finish = elts->end; + prim.prim = elts->prim; + prim.offset = 0; /* offset from start of dma buffers */ + prim.numverts = RADEON_MAX_VB_VERTS; /* duh */ + prim.vc_format = sarea_priv->vc_format; + + radeon_cp_dispatch_indices(dev, file_priv->master, buf, &prim); + if (elts->discard) { + radeon_cp_discard_buffer(dev, file_priv->master, buf); + } + + COMMIT_RING(); + return 0; +} + +static int radeon_cp_texture(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + drm_radeon_texture_t *tex = data; + drm_radeon_tex_image_t image; + int ret; + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + if (tex->image == NULL) { + DRM_ERROR("null texture image!\n"); + return -EINVAL; + } + + if (DRM_COPY_FROM_USER(&image, + (drm_radeon_tex_image_t __user *) tex->image, + sizeof(image))) + return -EFAULT; + + RING_SPACE_TEST_WITH_RETURN(dev_priv); + VB_AGE_TEST_WITH_RETURN(dev_priv); + + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) + ret = r600_cp_dispatch_texture(dev, file_priv, tex, &image); + else + ret = radeon_cp_dispatch_texture(dev, file_priv, tex, &image); + + return ret; +} + +static int radeon_cp_stipple(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + drm_radeon_stipple_t *stipple = data; + u32 mask[32]; + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + if (DRM_COPY_FROM_USER(&mask, stipple->mask, 32 * sizeof(u32))) + return -EFAULT; + + RING_SPACE_TEST_WITH_RETURN(dev_priv); + + radeon_cp_dispatch_stipple(dev, mask); + + COMMIT_RING(); + return 0; +} + +static int radeon_cp_indirect(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + struct drm_device_dma *dma = dev->dma; + struct drm_buf *buf; + drm_radeon_indirect_t *indirect = data; + RING_LOCALS; + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + DRM_DEBUG("idx=%d s=%d e=%d d=%d\n", + indirect->idx, indirect->start, indirect->end, + indirect->discard); + + if (indirect->idx < 0 || indirect->idx >= dma->buf_count) { + DRM_ERROR("buffer index %d (of %d max)\n", + indirect->idx, dma->buf_count - 1); + return -EINVAL; + } + + buf = dma->buflist[indirect->idx]; + + if (buf->file_priv != file_priv) { + DRM_ERROR("process %d using buffer owned by %p\n", + DRM_CURRENTPID, buf->file_priv); + return -EINVAL; + } + if (buf->pending) { + DRM_ERROR("sending pending buffer %d\n", indirect->idx); + return -EINVAL; + } + + if (indirect->start < buf->used) { + DRM_ERROR("reusing indirect: start=0x%x actual=0x%x\n", + indirect->start, buf->used); + return -EINVAL; + } + + RING_SPACE_TEST_WITH_RETURN(dev_priv); + VB_AGE_TEST_WITH_RETURN(dev_priv); + + buf->used = indirect->end; + + /* Dispatch the indirect buffer full of commands from the + * X server. This is insecure and is thus only available to + * privileged clients. + */ + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) + r600_cp_dispatch_indirect(dev, buf, indirect->start, indirect->end); + else { + /* Wait for the 3D stream to idle before the indirect buffer + * containing 2D acceleration commands is processed. + */ + BEGIN_RING(2); + RADEON_WAIT_UNTIL_3D_IDLE(); + ADVANCE_RING(); + radeon_cp_dispatch_indirect(dev, buf, indirect->start, indirect->end); + } + + if (indirect->discard) { + radeon_cp_discard_buffer(dev, file_priv->master, buf); + } + + COMMIT_RING(); + return 0; +} + +static int radeon_cp_vertex2(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + struct drm_radeon_master_private *master_priv = file_priv->master->driver_priv; + drm_radeon_sarea_t *sarea_priv; + struct drm_device_dma *dma = dev->dma; + struct drm_buf *buf; + drm_radeon_vertex2_t *vertex = data; + int i; + unsigned char laststate; + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + sarea_priv = master_priv->sarea_priv; + + DRM_DEBUG("pid=%d index=%d discard=%d\n", + DRM_CURRENTPID, vertex->idx, vertex->discard); + + if (vertex->idx < 0 || vertex->idx >= dma->buf_count) { + DRM_ERROR("buffer index %d (of %d max)\n", + vertex->idx, dma->buf_count - 1); + return -EINVAL; + } + + RING_SPACE_TEST_WITH_RETURN(dev_priv); + VB_AGE_TEST_WITH_RETURN(dev_priv); + + buf = dma->buflist[vertex->idx]; + + if (buf->file_priv != file_priv) { + DRM_ERROR("process %d using buffer owned by %p\n", + DRM_CURRENTPID, buf->file_priv); + return -EINVAL; + } + + if (buf->pending) { + DRM_ERROR("sending pending buffer %d\n", vertex->idx); + return -EINVAL; + } + + if (sarea_priv->nbox > RADEON_NR_SAREA_CLIPRECTS) + return -EINVAL; + + for (laststate = 0xff, i = 0; i < vertex->nr_prims; i++) { + drm_radeon_prim_t prim; + drm_radeon_tcl_prim_t tclprim; + + if (DRM_COPY_FROM_USER(&prim, &vertex->prim[i], sizeof(prim))) + return -EFAULT; + + if (prim.stateidx != laststate) { + drm_radeon_state_t state; + + if (DRM_COPY_FROM_USER(&state, + &vertex->state[prim.stateidx], + sizeof(state))) + return -EFAULT; + + if (radeon_emit_state2(dev_priv, file_priv, &state)) { + DRM_ERROR("radeon_emit_state2 failed\n"); + return -EINVAL; + } + + laststate = prim.stateidx; + } + + tclprim.start = prim.start; + tclprim.finish = prim.finish; + tclprim.prim = prim.prim; + tclprim.vc_format = prim.vc_format; + + if (prim.prim & RADEON_PRIM_WALK_IND) { + tclprim.offset = prim.numverts * 64; + tclprim.numverts = RADEON_MAX_VB_VERTS; /* duh */ + + radeon_cp_dispatch_indices(dev, file_priv->master, buf, &tclprim); + } else { + tclprim.numverts = prim.numverts; + tclprim.offset = 0; /* not used */ + + radeon_cp_dispatch_vertex(dev, file_priv, buf, &tclprim); + } + + if (sarea_priv->nbox == 1) + sarea_priv->nbox = 0; + } + + if (vertex->discard) { + radeon_cp_discard_buffer(dev, file_priv->master, buf); + } + + COMMIT_RING(); + return 0; +} + +static int radeon_emit_packets(drm_radeon_private_t * dev_priv, + struct drm_file *file_priv, + drm_radeon_cmd_header_t header, + drm_radeon_kcmd_buffer_t *cmdbuf) +{ + int id = (int)header.packet.packet_id; + int sz, reg; + RING_LOCALS; + + if (id >= RADEON_MAX_STATE_PACKETS) + return -EINVAL; + + sz = packet[id].len; + reg = packet[id].start; + + if (sz * sizeof(u32) > drm_buffer_unprocessed(cmdbuf->buffer)) { + DRM_ERROR("Packet size provided larger than data provided\n"); + return -EINVAL; + } + + if (radeon_check_and_fixup_packets(dev_priv, file_priv, id, + cmdbuf->buffer)) { + DRM_ERROR("Packet verification failed\n"); + return -EINVAL; + } + + BEGIN_RING(sz + 1); + OUT_RING(CP_PACKET0(reg, (sz - 1))); + OUT_RING_DRM_BUFFER(cmdbuf->buffer, sz); + ADVANCE_RING(); + + return 0; +} + +static __inline__ int radeon_emit_scalars(drm_radeon_private_t *dev_priv, + drm_radeon_cmd_header_t header, + drm_radeon_kcmd_buffer_t *cmdbuf) +{ + int sz = header.scalars.count; + int start = header.scalars.offset; + int stride = header.scalars.stride; + RING_LOCALS; + + BEGIN_RING(3 + sz); + OUT_RING(CP_PACKET0(RADEON_SE_TCL_SCALAR_INDX_REG, 0)); + OUT_RING(start | (stride << RADEON_SCAL_INDX_DWORD_STRIDE_SHIFT)); + OUT_RING(CP_PACKET0_TABLE(RADEON_SE_TCL_SCALAR_DATA_REG, sz - 1)); + OUT_RING_DRM_BUFFER(cmdbuf->buffer, sz); + ADVANCE_RING(); + return 0; +} + +/* God this is ugly + */ +static __inline__ int radeon_emit_scalars2(drm_radeon_private_t *dev_priv, + drm_radeon_cmd_header_t header, + drm_radeon_kcmd_buffer_t *cmdbuf) +{ + int sz = header.scalars.count; + int start = ((unsigned int)header.scalars.offset) + 0x100; + int stride = header.scalars.stride; + RING_LOCALS; + + BEGIN_RING(3 + sz); + OUT_RING(CP_PACKET0(RADEON_SE_TCL_SCALAR_INDX_REG, 0)); + OUT_RING(start | (stride << RADEON_SCAL_INDX_DWORD_STRIDE_SHIFT)); + OUT_RING(CP_PACKET0_TABLE(RADEON_SE_TCL_SCALAR_DATA_REG, sz - 1)); + OUT_RING_DRM_BUFFER(cmdbuf->buffer, sz); + ADVANCE_RING(); + return 0; +} + +static __inline__ int radeon_emit_vectors(drm_radeon_private_t *dev_priv, + drm_radeon_cmd_header_t header, + drm_radeon_kcmd_buffer_t *cmdbuf) +{ + int sz = header.vectors.count; + int start = header.vectors.offset; + int stride = header.vectors.stride; + RING_LOCALS; + + BEGIN_RING(5 + sz); + OUT_RING_REG(RADEON_SE_TCL_STATE_FLUSH, 0); + OUT_RING(CP_PACKET0(RADEON_SE_TCL_VECTOR_INDX_REG, 0)); + OUT_RING(start | (stride << RADEON_VEC_INDX_OCTWORD_STRIDE_SHIFT)); + OUT_RING(CP_PACKET0_TABLE(RADEON_SE_TCL_VECTOR_DATA_REG, (sz - 1))); + OUT_RING_DRM_BUFFER(cmdbuf->buffer, sz); + ADVANCE_RING(); + + return 0; +} + +static __inline__ int radeon_emit_veclinear(drm_radeon_private_t *dev_priv, + drm_radeon_cmd_header_t header, + drm_radeon_kcmd_buffer_t *cmdbuf) +{ + int sz = header.veclinear.count * 4; + int start = header.veclinear.addr_lo | (header.veclinear.addr_hi << 8); + RING_LOCALS; + + if (!sz) + return 0; + if (sz * 4 > drm_buffer_unprocessed(cmdbuf->buffer)) + return -EINVAL; + + BEGIN_RING(5 + sz); + OUT_RING_REG(RADEON_SE_TCL_STATE_FLUSH, 0); + OUT_RING(CP_PACKET0(RADEON_SE_TCL_VECTOR_INDX_REG, 0)); + OUT_RING(start | (1 << RADEON_VEC_INDX_OCTWORD_STRIDE_SHIFT)); + OUT_RING(CP_PACKET0_TABLE(RADEON_SE_TCL_VECTOR_DATA_REG, (sz - 1))); + OUT_RING_DRM_BUFFER(cmdbuf->buffer, sz); + ADVANCE_RING(); + + return 0; +} + +static int radeon_emit_packet3(struct drm_device * dev, + struct drm_file *file_priv, + drm_radeon_kcmd_buffer_t *cmdbuf) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + unsigned int cmdsz; + int ret; + RING_LOCALS; + + DRM_DEBUG("\n"); + + if ((ret = radeon_check_and_fixup_packet3(dev_priv, file_priv, + cmdbuf, &cmdsz))) { + DRM_ERROR("Packet verification failed\n"); + return ret; + } + + BEGIN_RING(cmdsz); + OUT_RING_DRM_BUFFER(cmdbuf->buffer, cmdsz); + ADVANCE_RING(); + + return 0; +} + +static int radeon_emit_packet3_cliprect(struct drm_device *dev, + struct drm_file *file_priv, + drm_radeon_kcmd_buffer_t *cmdbuf, + int orig_nbox) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + struct drm_clip_rect box; + unsigned int cmdsz; + int ret; + struct drm_clip_rect __user *boxes = cmdbuf->boxes; + int i = 0; + RING_LOCALS; + + DRM_DEBUG("\n"); + + if ((ret = radeon_check_and_fixup_packet3(dev_priv, file_priv, + cmdbuf, &cmdsz))) { + DRM_ERROR("Packet verification failed\n"); + return ret; + } + + if (!orig_nbox) + goto out; + + do { + if (i < cmdbuf->nbox) { + if (DRM_COPY_FROM_USER(&box, &boxes[i], sizeof(box))) + return -EFAULT; + /* FIXME The second and subsequent times round + * this loop, send a WAIT_UNTIL_3D_IDLE before + * calling emit_clip_rect(). This fixes a + * lockup on fast machines when sending + * several cliprects with a cmdbuf, as when + * waving a 2D window over a 3D + * window. Something in the commands from user + * space seems to hang the card when they're + * sent several times in a row. That would be + * the correct place to fix it but this works + * around it until I can figure that out - Tim + * Smith */ + if (i) { + BEGIN_RING(2); + RADEON_WAIT_UNTIL_3D_IDLE(); + ADVANCE_RING(); + } + radeon_emit_clip_rect(dev_priv, &box); + } + + BEGIN_RING(cmdsz); + OUT_RING_DRM_BUFFER(cmdbuf->buffer, cmdsz); + ADVANCE_RING(); + + } while (++i < cmdbuf->nbox); + if (cmdbuf->nbox == 1) + cmdbuf->nbox = 0; + + return 0; + out: + drm_buffer_advance(cmdbuf->buffer, cmdsz * 4); + return 0; +} + +static int radeon_emit_wait(struct drm_device * dev, int flags) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + RING_LOCALS; + + DRM_DEBUG("%x\n", flags); + switch (flags) { + case RADEON_WAIT_2D: + BEGIN_RING(2); + RADEON_WAIT_UNTIL_2D_IDLE(); + ADVANCE_RING(); + break; + case RADEON_WAIT_3D: + BEGIN_RING(2); + RADEON_WAIT_UNTIL_3D_IDLE(); + ADVANCE_RING(); + break; + case RADEON_WAIT_2D | RADEON_WAIT_3D: + BEGIN_RING(2); + RADEON_WAIT_UNTIL_IDLE(); + ADVANCE_RING(); + break; + default: + return -EINVAL; + } + + return 0; +} + +static int radeon_cp_cmdbuf(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + struct drm_device_dma *dma = dev->dma; + struct drm_buf *buf = NULL; + drm_radeon_cmd_header_t stack_header; + int idx; + drm_radeon_kcmd_buffer_t *cmdbuf = data; + int orig_nbox; + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + RING_SPACE_TEST_WITH_RETURN(dev_priv); + VB_AGE_TEST_WITH_RETURN(dev_priv); + + if (cmdbuf->bufsz > 64 * 1024 || cmdbuf->bufsz < 0) { + return -EINVAL; + } + + /* Allocate an in-kernel area and copy in the cmdbuf. Do this to avoid + * races between checking values and using those values in other code, + * and simply to avoid a lot of function calls to copy in data. + */ + if (cmdbuf->bufsz != 0) { + int rv; + void __user *buffer = cmdbuf->buffer; + rv = drm_buffer_alloc(&cmdbuf->buffer, cmdbuf->bufsz); + if (rv) + return rv; + rv = drm_buffer_copy_from_user(cmdbuf->buffer, buffer, + cmdbuf->bufsz); + if (rv) { + drm_buffer_free(cmdbuf->buffer); + return rv; + } + } else + goto done; + + orig_nbox = cmdbuf->nbox; + + if (dev_priv->microcode_version == UCODE_R300) { + int temp; + temp = r300_do_cp_cmdbuf(dev, file_priv, cmdbuf); + + drm_buffer_free(cmdbuf->buffer); + + return temp; + } + + /* microcode_version != r300 */ + while (drm_buffer_unprocessed(cmdbuf->buffer) >= sizeof(stack_header)) { + + drm_radeon_cmd_header_t *header; + header = drm_buffer_read_object(cmdbuf->buffer, + sizeof(stack_header), &stack_header); + + switch (header->header.cmd_type) { + case RADEON_CMD_PACKET: + DRM_DEBUG("RADEON_CMD_PACKET\n"); + if (radeon_emit_packets + (dev_priv, file_priv, *header, cmdbuf)) { + DRM_ERROR("radeon_emit_packets failed\n"); + goto err; + } + break; + + case RADEON_CMD_SCALARS: + DRM_DEBUG("RADEON_CMD_SCALARS\n"); + if (radeon_emit_scalars(dev_priv, *header, cmdbuf)) { + DRM_ERROR("radeon_emit_scalars failed\n"); + goto err; + } + break; + + case RADEON_CMD_VECTORS: + DRM_DEBUG("RADEON_CMD_VECTORS\n"); + if (radeon_emit_vectors(dev_priv, *header, cmdbuf)) { + DRM_ERROR("radeon_emit_vectors failed\n"); + goto err; + } + break; + + case RADEON_CMD_DMA_DISCARD: + DRM_DEBUG("RADEON_CMD_DMA_DISCARD\n"); + idx = header->dma.buf_idx; + if (idx < 0 || idx >= dma->buf_count) { + DRM_ERROR("buffer index %d (of %d max)\n", + idx, dma->buf_count - 1); + goto err; + } + + buf = dma->buflist[idx]; + if (buf->file_priv != file_priv || buf->pending) { + DRM_ERROR("bad buffer %p %p %d\n", + buf->file_priv, file_priv, + buf->pending); + goto err; + } + + radeon_cp_discard_buffer(dev, file_priv->master, buf); + break; + + case RADEON_CMD_PACKET3: + DRM_DEBUG("RADEON_CMD_PACKET3\n"); + if (radeon_emit_packet3(dev, file_priv, cmdbuf)) { + DRM_ERROR("radeon_emit_packet3 failed\n"); + goto err; + } + break; + + case RADEON_CMD_PACKET3_CLIP: + DRM_DEBUG("RADEON_CMD_PACKET3_CLIP\n"); + if (radeon_emit_packet3_cliprect + (dev, file_priv, cmdbuf, orig_nbox)) { + DRM_ERROR("radeon_emit_packet3_clip failed\n"); + goto err; + } + break; + + case RADEON_CMD_SCALARS2: + DRM_DEBUG("RADEON_CMD_SCALARS2\n"); + if (radeon_emit_scalars2(dev_priv, *header, cmdbuf)) { + DRM_ERROR("radeon_emit_scalars2 failed\n"); + goto err; + } + break; + + case RADEON_CMD_WAIT: + DRM_DEBUG("RADEON_CMD_WAIT\n"); + if (radeon_emit_wait(dev, header->wait.flags)) { + DRM_ERROR("radeon_emit_wait failed\n"); + goto err; + } + break; + case RADEON_CMD_VECLINEAR: + DRM_DEBUG("RADEON_CMD_VECLINEAR\n"); + if (radeon_emit_veclinear(dev_priv, *header, cmdbuf)) { + DRM_ERROR("radeon_emit_veclinear failed\n"); + goto err; + } + break; + + default: + DRM_ERROR("bad cmd_type %d at byte %d\n", + header->header.cmd_type, + cmdbuf->buffer->iterator); + goto err; + } + } + + drm_buffer_free(cmdbuf->buffer); + + done: + DRM_DEBUG("DONE\n"); + COMMIT_RING(); + return 0; + + err: + drm_buffer_free(cmdbuf->buffer); + return -EINVAL; +} + +static int radeon_cp_getparam(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + drm_radeon_getparam_t *param = data; + int value; + + DRM_DEBUG("pid=%d\n", DRM_CURRENTPID); + + switch (param->param) { + case RADEON_PARAM_GART_BUFFER_OFFSET: + value = dev_priv->gart_buffers_offset; + break; + case RADEON_PARAM_LAST_FRAME: + dev_priv->stats.last_frame_reads++; + value = GET_SCRATCH(dev_priv, 0); + break; + case RADEON_PARAM_LAST_DISPATCH: + value = GET_SCRATCH(dev_priv, 1); + break; + case RADEON_PARAM_LAST_CLEAR: + dev_priv->stats.last_clear_reads++; + value = GET_SCRATCH(dev_priv, 2); + break; + case RADEON_PARAM_IRQ_NR: + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) + value = 0; + else + value = drm_dev_to_irq(dev); + break; + case RADEON_PARAM_GART_BASE: + value = dev_priv->gart_vm_start; + break; + case RADEON_PARAM_REGISTER_HANDLE: + value = dev_priv->mmio->offset; + break; + case RADEON_PARAM_STATUS_HANDLE: + value = dev_priv->ring_rptr_offset; + break; +#if BITS_PER_LONG == 32 + /* + * This ioctl() doesn't work on 64-bit platforms because hw_lock is a + * pointer which can't fit into an int-sized variable. According to + * Michel Dänzer, the ioctl() is only used on embedded platforms, so + * not supporting it shouldn't be a problem. If the same functionality + * is needed on 64-bit platforms, a new ioctl() would have to be added, + * so backwards-compatibility for the embedded platforms can be + * maintained. --davidm 4-Feb-2004. + */ + case RADEON_PARAM_SAREA_HANDLE: + /* The lock is the first dword in the sarea. */ + /* no users of this parameter */ + break; +#endif + case RADEON_PARAM_GART_TEX_HANDLE: + value = dev_priv->gart_textures_offset; + break; + case RADEON_PARAM_SCRATCH_OFFSET: + if (!dev_priv->writeback_works) + return -EINVAL; + if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R600) + value = R600_SCRATCH_REG_OFFSET; + else + value = RADEON_SCRATCH_REG_OFFSET; + break; + case RADEON_PARAM_CARD_TYPE: + if (dev_priv->flags & RADEON_IS_PCIE) + value = RADEON_CARD_PCIE; + else if (dev_priv->flags & RADEON_IS_AGP) + value = RADEON_CARD_AGP; + else + value = RADEON_CARD_PCI; + break; + case RADEON_PARAM_VBLANK_CRTC: + value = radeon_vblank_crtc_get(dev); + break; + case RADEON_PARAM_FB_LOCATION: + value = radeon_read_fb_location(dev_priv); + break; + case RADEON_PARAM_NUM_GB_PIPES: + value = dev_priv->num_gb_pipes; + break; + case RADEON_PARAM_NUM_Z_PIPES: + value = dev_priv->num_z_pipes; + break; + default: + DRM_DEBUG("Invalid parameter %d\n", param->param); + return -EINVAL; + } + + if (DRM_COPY_TO_USER(param->value, &value, sizeof(int))) { + DRM_ERROR("copy_to_user\n"); + return -EFAULT; + } + + return 0; +} + +static int radeon_cp_setparam(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + struct drm_radeon_master_private *master_priv = file_priv->master->driver_priv; + drm_radeon_setparam_t *sp = data; + struct drm_radeon_driver_file_fields *radeon_priv; + + switch (sp->param) { + case RADEON_SETPARAM_FB_LOCATION: + radeon_priv = file_priv->driver_priv; + radeon_priv->radeon_fb_delta = dev_priv->fb_location - + sp->value; + break; + case RADEON_SETPARAM_SWITCH_TILING: + if (sp->value == 0) { + DRM_DEBUG("color tiling disabled\n"); + dev_priv->front_pitch_offset &= ~RADEON_DST_TILE_MACRO; + dev_priv->back_pitch_offset &= ~RADEON_DST_TILE_MACRO; + if (master_priv->sarea_priv) + master_priv->sarea_priv->tiling_enabled = 0; + } else if (sp->value == 1) { + DRM_DEBUG("color tiling enabled\n"); + dev_priv->front_pitch_offset |= RADEON_DST_TILE_MACRO; + dev_priv->back_pitch_offset |= RADEON_DST_TILE_MACRO; + if (master_priv->sarea_priv) + master_priv->sarea_priv->tiling_enabled = 1; + } + break; + case RADEON_SETPARAM_PCIGART_LOCATION: + dev_priv->pcigart_offset = sp->value; + dev_priv->pcigart_offset_set = 1; + break; + case RADEON_SETPARAM_NEW_MEMMAP: + dev_priv->new_memmap = sp->value; + break; + case RADEON_SETPARAM_PCIGART_TABLE_SIZE: + dev_priv->gart_info.table_size = sp->value; + if (dev_priv->gart_info.table_size < RADEON_PCIGART_TABLE_SIZE) + dev_priv->gart_info.table_size = RADEON_PCIGART_TABLE_SIZE; + break; + case RADEON_SETPARAM_VBLANK_CRTC: + return radeon_vblank_crtc_set(dev, sp->value); + break; + default: + DRM_DEBUG("Invalid parameter %d\n", sp->param); + return -EINVAL; + } + + return 0; +} + +/* When a client dies: + * - Check for and clean up flipped page state + * - Free any alloced GART memory. + * - Free any alloced radeon surfaces. + * + * DRM infrastructure takes care of reclaiming dma buffers. + */ +void radeon_driver_preclose(struct drm_device *dev, struct drm_file *file_priv) +{ + if (dev->dev_private) { + drm_radeon_private_t *dev_priv = dev->dev_private; + dev_priv->page_flipping = 0; + radeon_mem_release(file_priv, dev_priv->gart_heap); + radeon_mem_release(file_priv, dev_priv->fb_heap); + radeon_surfaces_release(file_priv, dev_priv); + } +} + +void radeon_driver_lastclose(struct drm_device *dev) +{ + radeon_surfaces_release(PCIGART_FILE_PRIV, dev->dev_private); + radeon_do_release(dev); +} + +int radeon_driver_open(struct drm_device *dev, struct drm_file *file_priv) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + struct drm_radeon_driver_file_fields *radeon_priv; + + DRM_DEBUG("\n"); + radeon_priv = kmalloc(sizeof(*radeon_priv), GFP_KERNEL); + + if (!radeon_priv) + return -ENOMEM; + + file_priv->driver_priv = radeon_priv; + + if (dev_priv) + radeon_priv->radeon_fb_delta = dev_priv->fb_location; + else + radeon_priv->radeon_fb_delta = 0; + return 0; +} + +void radeon_driver_postclose(struct drm_device *dev, struct drm_file *file_priv) +{ + struct drm_radeon_driver_file_fields *radeon_priv = + file_priv->driver_priv; + + kfree(radeon_priv); +} + +struct drm_ioctl_desc radeon_ioctls[] = { + DRM_IOCTL_DEF_DRV(RADEON_CP_INIT, radeon_cp_init, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF_DRV(RADEON_CP_START, radeon_cp_start, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF_DRV(RADEON_CP_STOP, radeon_cp_stop, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF_DRV(RADEON_CP_RESET, radeon_cp_reset, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF_DRV(RADEON_CP_IDLE, radeon_cp_idle, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_CP_RESUME, radeon_cp_resume, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_RESET, radeon_engine_reset, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_FULLSCREEN, radeon_fullscreen, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_SWAP, radeon_cp_swap, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_CLEAR, radeon_cp_clear, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_VERTEX, radeon_cp_vertex, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_INDICES, radeon_cp_indices, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_TEXTURE, radeon_cp_texture, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_STIPPLE, radeon_cp_stipple, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_INDIRECT, radeon_cp_indirect, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF_DRV(RADEON_VERTEX2, radeon_cp_vertex2, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_CMDBUF, radeon_cp_cmdbuf, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_GETPARAM, radeon_cp_getparam, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_FLIP, radeon_cp_flip, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_ALLOC, radeon_mem_alloc, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_FREE, radeon_mem_free, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_INIT_HEAP, radeon_mem_init_heap, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF_DRV(RADEON_IRQ_EMIT, radeon_irq_emit, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_IRQ_WAIT, radeon_irq_wait, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_SETPARAM, radeon_cp_setparam, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_SURF_ALLOC, radeon_surface_alloc, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_SURF_FREE, radeon_surface_free, DRM_AUTH), + DRM_IOCTL_DEF_DRV(RADEON_CS, r600_cs_legacy_ioctl, DRM_AUTH) +}; + +int radeon_max_ioctl = DRM_ARRAY_SIZE(radeon_ioctls); diff --git a/drivers/gpu/drm/radeon/radeon_test.c b/drivers/gpu/drm/radeon/radeon_test.c new file mode 100644 index 00000000..dee4a0c1 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_test.c @@ -0,0 +1,218 @@ +/* + * Copyright 2009 VMware, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Michel Dänzer + */ +#include +#include +#include "radeon_reg.h" +#include "radeon.h" + + +/* Test BO GTT->VRAM and VRAM->GTT GPU copies across the whole GTT aperture */ +void radeon_test_moves(struct radeon_device *rdev) +{ + struct radeon_bo *vram_obj = NULL; + struct radeon_bo **gtt_obj = NULL; + struct radeon_fence *fence = NULL; + uint64_t gtt_addr, vram_addr; + unsigned i, n, size; + int r; + + size = 1024 * 1024; + + /* Number of tests = + * (Total GTT - IB pool - writeback page - ring buffer) / test size + */ + n = ((u32)(rdev->mc.gtt_size - RADEON_IB_POOL_SIZE*64*1024 - RADEON_GPU_PAGE_SIZE - + rdev->cp.ring_size)) / size; + + gtt_obj = kzalloc(n * sizeof(*gtt_obj), GFP_KERNEL); + if (!gtt_obj) { + DRM_ERROR("Failed to allocate %d pointers\n", n); + r = 1; + goto out_cleanup; + } + + r = radeon_bo_create(rdev, size, PAGE_SIZE, true, RADEON_GEM_DOMAIN_VRAM, + &vram_obj); + if (r) { + DRM_ERROR("Failed to create VRAM object\n"); + goto out_cleanup; + } + r = radeon_bo_reserve(vram_obj, false); + if (unlikely(r != 0)) + goto out_cleanup; + r = radeon_bo_pin(vram_obj, RADEON_GEM_DOMAIN_VRAM, &vram_addr); + if (r) { + DRM_ERROR("Failed to pin VRAM object\n"); + goto out_cleanup; + } + for (i = 0; i < n; i++) { + void *gtt_map, *vram_map; + void **gtt_start, **gtt_end; + void **vram_start, **vram_end; + + r = radeon_bo_create(rdev, size, PAGE_SIZE, true, + RADEON_GEM_DOMAIN_GTT, gtt_obj + i); + if (r) { + DRM_ERROR("Failed to create GTT object %d\n", i); + goto out_cleanup; + } + + r = radeon_bo_reserve(gtt_obj[i], false); + if (unlikely(r != 0)) + goto out_cleanup; + r = radeon_bo_pin(gtt_obj[i], RADEON_GEM_DOMAIN_GTT, >t_addr); + if (r) { + DRM_ERROR("Failed to pin GTT object %d\n", i); + goto out_cleanup; + } + + r = radeon_bo_kmap(gtt_obj[i], >t_map); + if (r) { + DRM_ERROR("Failed to map GTT object %d\n", i); + goto out_cleanup; + } + + for (gtt_start = gtt_map, gtt_end = gtt_map + size; + gtt_start < gtt_end; + gtt_start++) + *gtt_start = gtt_start; + + radeon_bo_kunmap(gtt_obj[i]); + + r = radeon_fence_create(rdev, &fence); + if (r) { + DRM_ERROR("Failed to create GTT->VRAM fence %d\n", i); + goto out_cleanup; + } + + r = radeon_copy(rdev, gtt_addr, vram_addr, size / RADEON_GPU_PAGE_SIZE, fence); + if (r) { + DRM_ERROR("Failed GTT->VRAM copy %d\n", i); + goto out_cleanup; + } + + r = radeon_fence_wait(fence, false); + if (r) { + DRM_ERROR("Failed to wait for GTT->VRAM fence %d\n", i); + goto out_cleanup; + } + + radeon_fence_unref(&fence); + + r = radeon_bo_kmap(vram_obj, &vram_map); + if (r) { + DRM_ERROR("Failed to map VRAM object after copy %d\n", i); + goto out_cleanup; + } + + for (gtt_start = gtt_map, gtt_end = gtt_map + size, + vram_start = vram_map, vram_end = vram_map + size; + vram_start < vram_end; + gtt_start++, vram_start++) { + if (*vram_start != gtt_start) { + DRM_ERROR("Incorrect GTT->VRAM copy %d: Got 0x%p, " + "expected 0x%p (GTT map 0x%p-0x%p)\n", + i, *vram_start, gtt_start, gtt_map, + gtt_end); + radeon_bo_kunmap(vram_obj); + goto out_cleanup; + } + *vram_start = vram_start; + } + + radeon_bo_kunmap(vram_obj); + + r = radeon_fence_create(rdev, &fence); + if (r) { + DRM_ERROR("Failed to create VRAM->GTT fence %d\n", i); + goto out_cleanup; + } + + r = radeon_copy(rdev, vram_addr, gtt_addr, size / RADEON_GPU_PAGE_SIZE, fence); + if (r) { + DRM_ERROR("Failed VRAM->GTT copy %d\n", i); + goto out_cleanup; + } + + r = radeon_fence_wait(fence, false); + if (r) { + DRM_ERROR("Failed to wait for VRAM->GTT fence %d\n", i); + goto out_cleanup; + } + + radeon_fence_unref(&fence); + + r = radeon_bo_kmap(gtt_obj[i], >t_map); + if (r) { + DRM_ERROR("Failed to map GTT object after copy %d\n", i); + goto out_cleanup; + } + + for (gtt_start = gtt_map, gtt_end = gtt_map + size, + vram_start = vram_map, vram_end = vram_map + size; + gtt_start < gtt_end; + gtt_start++, vram_start++) { + if (*gtt_start != vram_start) { + DRM_ERROR("Incorrect VRAM->GTT copy %d: Got 0x%p, " + "expected 0x%p (VRAM map 0x%p-0x%p)\n", + i, *gtt_start, vram_start, vram_map, + vram_end); + radeon_bo_kunmap(gtt_obj[i]); + goto out_cleanup; + } + } + + radeon_bo_kunmap(gtt_obj[i]); + + DRM_INFO("Tested GTT->VRAM and VRAM->GTT copy for GTT offset 0x%llx\n", + gtt_addr - rdev->mc.gtt_start); + } + +out_cleanup: + if (vram_obj) { + if (radeon_bo_is_reserved(vram_obj)) { + radeon_bo_unpin(vram_obj); + radeon_bo_unreserve(vram_obj); + } + radeon_bo_unref(&vram_obj); + } + if (gtt_obj) { + for (i = 0; i < n; i++) { + if (gtt_obj[i]) { + if (radeon_bo_is_reserved(gtt_obj[i])) { + radeon_bo_unpin(gtt_obj[i]); + radeon_bo_unreserve(gtt_obj[i]); + } + radeon_bo_unref(>t_obj[i]); + } + } + kfree(gtt_obj); + } + if (fence) { + radeon_fence_unref(&fence); + } + if (r) { + printk(KERN_WARNING "Error while testing BO move.\n"); + } +} diff --git a/drivers/gpu/drm/radeon/radeon_trace.h b/drivers/gpu/drm/radeon/radeon_trace.h new file mode 100644 index 00000000..eafd8160 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_trace.h @@ -0,0 +1,82 @@ +#if !defined(_RADEON_TRACE_H) || defined(TRACE_HEADER_MULTI_READ) +#define _RADEON_TRACE_H_ + +#include +#include +#include + +#include + +#undef TRACE_SYSTEM +#define TRACE_SYSTEM radeon +#define TRACE_SYSTEM_STRING __stringify(TRACE_SYSTEM) +#define TRACE_INCLUDE_FILE radeon_trace + +TRACE_EVENT(radeon_bo_create, + TP_PROTO(struct radeon_bo *bo), + TP_ARGS(bo), + TP_STRUCT__entry( + __field(struct radeon_bo *, bo) + __field(u32, pages) + ), + + TP_fast_assign( + __entry->bo = bo; + __entry->pages = bo->tbo.num_pages; + ), + TP_printk("bo=%p, pages=%u", __entry->bo, __entry->pages) +); + +DECLARE_EVENT_CLASS(radeon_fence_request, + + TP_PROTO(struct drm_device *dev, u32 seqno), + + TP_ARGS(dev, seqno), + + TP_STRUCT__entry( + __field(u32, dev) + __field(u32, seqno) + ), + + TP_fast_assign( + __entry->dev = dev->primary->index; + __entry->seqno = seqno; + ), + + TP_printk("dev=%u, seqno=%u", __entry->dev, __entry->seqno) +); + +DEFINE_EVENT(radeon_fence_request, radeon_fence_emit, + + TP_PROTO(struct drm_device *dev, u32 seqno), + + TP_ARGS(dev, seqno) +); + +DEFINE_EVENT(radeon_fence_request, radeon_fence_retire, + + TP_PROTO(struct drm_device *dev, u32 seqno), + + TP_ARGS(dev, seqno) +); + +DEFINE_EVENT(radeon_fence_request, radeon_fence_wait_begin, + + TP_PROTO(struct drm_device *dev, u32 seqno), + + TP_ARGS(dev, seqno) +); + +DEFINE_EVENT(radeon_fence_request, radeon_fence_wait_end, + + TP_PROTO(struct drm_device *dev, u32 seqno), + + TP_ARGS(dev, seqno) +); + +#endif + +/* This part must be outside protection */ +#undef TRACE_INCLUDE_PATH +#define TRACE_INCLUDE_PATH . +#include diff --git a/drivers/gpu/drm/radeon/radeon_trace_points.c b/drivers/gpu/drm/radeon/radeon_trace_points.c new file mode 100644 index 00000000..8175993d --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_trace_points.c @@ -0,0 +1,9 @@ +/* Copyright Red Hat Inc 2010. + * Author : Dave Airlie + */ +#include +#include "radeon_drm.h" +#include "radeon.h" + +#define CREATE_TRACE_POINTS +#include "radeon_trace.h" diff --git a/drivers/gpu/drm/radeon/radeon_ttm.c b/drivers/gpu/drm/radeon/radeon_ttm.c new file mode 100644 index 00000000..3e9b41b3 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_ttm.c @@ -0,0 +1,828 @@ +/* + * Copyright 2009 Jerome Glisse. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + */ +/* + * Authors: + * Jerome Glisse + * Thomas Hellstrom + * Dave Airlie + */ +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include "radeon_reg.h" +#include "radeon.h" + +#define DRM_FILE_PAGE_OFFSET (0x100000000ULL >> PAGE_SHIFT) + +static int radeon_ttm_debugfs_init(struct radeon_device *rdev); + +static struct radeon_device *radeon_get_rdev(struct ttm_bo_device *bdev) +{ + struct radeon_mman *mman; + struct radeon_device *rdev; + + mman = container_of(bdev, struct radeon_mman, bdev); + rdev = container_of(mman, struct radeon_device, mman); + return rdev; +} + + +/* + * Global memory. + */ +static int radeon_ttm_mem_global_init(struct drm_global_reference *ref) +{ + return ttm_mem_global_init(ref->object); +} + +static void radeon_ttm_mem_global_release(struct drm_global_reference *ref) +{ + ttm_mem_global_release(ref->object); +} + +static int radeon_ttm_global_init(struct radeon_device *rdev) +{ + struct drm_global_reference *global_ref; + int r; + + rdev->mman.mem_global_referenced = false; + global_ref = &rdev->mman.mem_global_ref; + global_ref->global_type = DRM_GLOBAL_TTM_MEM; + global_ref->size = sizeof(struct ttm_mem_global); + global_ref->init = &radeon_ttm_mem_global_init; + global_ref->release = &radeon_ttm_mem_global_release; + r = drm_global_item_ref(global_ref); + if (r != 0) { + DRM_ERROR("Failed setting up TTM memory accounting " + "subsystem.\n"); + return r; + } + + rdev->mman.bo_global_ref.mem_glob = + rdev->mman.mem_global_ref.object; + global_ref = &rdev->mman.bo_global_ref.ref; + global_ref->global_type = DRM_GLOBAL_TTM_BO; + global_ref->size = sizeof(struct ttm_bo_global); + global_ref->init = &ttm_bo_global_init; + global_ref->release = &ttm_bo_global_release; + r = drm_global_item_ref(global_ref); + if (r != 0) { + DRM_ERROR("Failed setting up TTM BO subsystem.\n"); + drm_global_item_unref(&rdev->mman.mem_global_ref); + return r; + } + + rdev->mman.mem_global_referenced = true; + return 0; +} + +static void radeon_ttm_global_fini(struct radeon_device *rdev) +{ + if (rdev->mman.mem_global_referenced) { + drm_global_item_unref(&rdev->mman.bo_global_ref.ref); + drm_global_item_unref(&rdev->mman.mem_global_ref); + rdev->mman.mem_global_referenced = false; + } +} + +struct ttm_backend *radeon_ttm_backend_create(struct radeon_device *rdev); + +static struct ttm_backend* +radeon_create_ttm_backend_entry(struct ttm_bo_device *bdev) +{ + struct radeon_device *rdev; + + rdev = radeon_get_rdev(bdev); +#if __OS_HAS_AGP + if (rdev->flags & RADEON_IS_AGP) { + return ttm_agp_backend_init(bdev, rdev->ddev->agp->bridge); + } else +#endif + { + return radeon_ttm_backend_create(rdev); + } +} + +static int radeon_invalidate_caches(struct ttm_bo_device *bdev, uint32_t flags) +{ + return 0; +} + +static int radeon_init_mem_type(struct ttm_bo_device *bdev, uint32_t type, + struct ttm_mem_type_manager *man) +{ + struct radeon_device *rdev; + + rdev = radeon_get_rdev(bdev); + + switch (type) { + case TTM_PL_SYSTEM: + /* System memory */ + man->flags = TTM_MEMTYPE_FLAG_MAPPABLE; + man->available_caching = TTM_PL_MASK_CACHING; + man->default_caching = TTM_PL_FLAG_CACHED; + break; + case TTM_PL_TT: + man->func = &ttm_bo_manager_func; + man->gpu_offset = rdev->mc.gtt_start; + man->available_caching = TTM_PL_MASK_CACHING; + man->default_caching = TTM_PL_FLAG_CACHED; + man->flags = TTM_MEMTYPE_FLAG_MAPPABLE | TTM_MEMTYPE_FLAG_CMA; +#if __OS_HAS_AGP + if (rdev->flags & RADEON_IS_AGP) { + if (!(drm_core_has_AGP(rdev->ddev) && rdev->ddev->agp)) { + DRM_ERROR("AGP is not enabled for memory type %u\n", + (unsigned)type); + return -EINVAL; + } + if (!rdev->ddev->agp->cant_use_aperture) + man->flags = TTM_MEMTYPE_FLAG_MAPPABLE; + man->available_caching = TTM_PL_FLAG_UNCACHED | + TTM_PL_FLAG_WC; + man->default_caching = TTM_PL_FLAG_WC; + } +#endif + break; + case TTM_PL_VRAM: + /* "On-card" video ram */ + man->func = &ttm_bo_manager_func; + man->gpu_offset = rdev->mc.vram_start; + man->flags = TTM_MEMTYPE_FLAG_FIXED | + TTM_MEMTYPE_FLAG_MAPPABLE; + man->available_caching = TTM_PL_FLAG_UNCACHED | TTM_PL_FLAG_WC; + man->default_caching = TTM_PL_FLAG_WC; + break; + default: + DRM_ERROR("Unsupported memory type %u\n", (unsigned)type); + return -EINVAL; + } + return 0; +} + +static void radeon_evict_flags(struct ttm_buffer_object *bo, + struct ttm_placement *placement) +{ + struct radeon_bo *rbo; + static u32 placements = TTM_PL_MASK_CACHING | TTM_PL_FLAG_SYSTEM; + + if (!radeon_ttm_bo_is_radeon_bo(bo)) { + placement->fpfn = 0; + placement->lpfn = 0; + placement->placement = &placements; + placement->busy_placement = &placements; + placement->num_placement = 1; + placement->num_busy_placement = 1; + return; + } + rbo = container_of(bo, struct radeon_bo, tbo); + switch (bo->mem.mem_type) { + case TTM_PL_VRAM: + if (rbo->rdev->cp.ready == false) + radeon_ttm_placement_from_domain(rbo, RADEON_GEM_DOMAIN_CPU); + else + radeon_ttm_placement_from_domain(rbo, RADEON_GEM_DOMAIN_GTT); + break; + case TTM_PL_TT: + default: + radeon_ttm_placement_from_domain(rbo, RADEON_GEM_DOMAIN_CPU); + } + *placement = rbo->placement; +} + +static int radeon_verify_access(struct ttm_buffer_object *bo, struct file *filp) +{ + return 0; +} + +static void radeon_move_null(struct ttm_buffer_object *bo, + struct ttm_mem_reg *new_mem) +{ + struct ttm_mem_reg *old_mem = &bo->mem; + + BUG_ON(old_mem->mm_node != NULL); + *old_mem = *new_mem; + new_mem->mm_node = NULL; +} + +static int radeon_move_blit(struct ttm_buffer_object *bo, + bool evict, int no_wait_reserve, bool no_wait_gpu, + struct ttm_mem_reg *new_mem, + struct ttm_mem_reg *old_mem) +{ + struct radeon_device *rdev; + uint64_t old_start, new_start; + struct radeon_fence *fence; + int r; + + rdev = radeon_get_rdev(bo->bdev); + r = radeon_fence_create(rdev, &fence); + if (unlikely(r)) { + return r; + } + old_start = old_mem->start << PAGE_SHIFT; + new_start = new_mem->start << PAGE_SHIFT; + + switch (old_mem->mem_type) { + case TTM_PL_VRAM: + old_start += rdev->mc.vram_start; + break; + case TTM_PL_TT: + old_start += rdev->mc.gtt_start; + break; + default: + DRM_ERROR("Unknown placement %d\n", old_mem->mem_type); + return -EINVAL; + } + switch (new_mem->mem_type) { + case TTM_PL_VRAM: + new_start += rdev->mc.vram_start; + break; + case TTM_PL_TT: + new_start += rdev->mc.gtt_start; + break; + default: + DRM_ERROR("Unknown placement %d\n", old_mem->mem_type); + return -EINVAL; + } + if (!rdev->cp.ready) { + DRM_ERROR("Trying to move memory with CP turned off.\n"); + return -EINVAL; + } + + BUILD_BUG_ON((PAGE_SIZE % RADEON_GPU_PAGE_SIZE) != 0); + + r = radeon_copy(rdev, old_start, new_start, + new_mem->num_pages * (PAGE_SIZE / RADEON_GPU_PAGE_SIZE), /* GPU pages */ + fence); + /* FIXME: handle copy error */ + r = ttm_bo_move_accel_cleanup(bo, (void *)fence, NULL, + evict, no_wait_reserve, no_wait_gpu, new_mem); + radeon_fence_unref(&fence); + return r; +} + +static int radeon_move_vram_ram(struct ttm_buffer_object *bo, + bool evict, bool interruptible, + bool no_wait_reserve, bool no_wait_gpu, + struct ttm_mem_reg *new_mem) +{ + struct radeon_device *rdev; + struct ttm_mem_reg *old_mem = &bo->mem; + struct ttm_mem_reg tmp_mem; + u32 placements; + struct ttm_placement placement; + int r; + + rdev = radeon_get_rdev(bo->bdev); + tmp_mem = *new_mem; + tmp_mem.mm_node = NULL; + placement.fpfn = 0; + placement.lpfn = 0; + placement.num_placement = 1; + placement.placement = &placements; + placement.num_busy_placement = 1; + placement.busy_placement = &placements; + placements = TTM_PL_MASK_CACHING | TTM_PL_FLAG_TT; + r = ttm_bo_mem_space(bo, &placement, &tmp_mem, + interruptible, no_wait_reserve, no_wait_gpu); + if (unlikely(r)) { + return r; + } + + r = ttm_tt_set_placement_caching(bo->ttm, tmp_mem.placement); + if (unlikely(r)) { + goto out_cleanup; + } + + r = ttm_tt_bind(bo->ttm, &tmp_mem); + if (unlikely(r)) { + goto out_cleanup; + } + r = radeon_move_blit(bo, true, no_wait_reserve, no_wait_gpu, &tmp_mem, old_mem); + if (unlikely(r)) { + goto out_cleanup; + } + r = ttm_bo_move_ttm(bo, true, no_wait_reserve, no_wait_gpu, new_mem); +out_cleanup: + ttm_bo_mem_put(bo, &tmp_mem); + return r; +} + +static int radeon_move_ram_vram(struct ttm_buffer_object *bo, + bool evict, bool interruptible, + bool no_wait_reserve, bool no_wait_gpu, + struct ttm_mem_reg *new_mem) +{ + struct radeon_device *rdev; + struct ttm_mem_reg *old_mem = &bo->mem; + struct ttm_mem_reg tmp_mem; + struct ttm_placement placement; + u32 placements; + int r; + + rdev = radeon_get_rdev(bo->bdev); + tmp_mem = *new_mem; + tmp_mem.mm_node = NULL; + placement.fpfn = 0; + placement.lpfn = 0; + placement.num_placement = 1; + placement.placement = &placements; + placement.num_busy_placement = 1; + placement.busy_placement = &placements; + placements = TTM_PL_MASK_CACHING | TTM_PL_FLAG_TT; + r = ttm_bo_mem_space(bo, &placement, &tmp_mem, interruptible, no_wait_reserve, no_wait_gpu); + if (unlikely(r)) { + return r; + } + r = ttm_bo_move_ttm(bo, true, no_wait_reserve, no_wait_gpu, &tmp_mem); + if (unlikely(r)) { + goto out_cleanup; + } + r = radeon_move_blit(bo, true, no_wait_reserve, no_wait_gpu, new_mem, old_mem); + if (unlikely(r)) { + goto out_cleanup; + } +out_cleanup: + ttm_bo_mem_put(bo, &tmp_mem); + return r; +} + +static int radeon_bo_move(struct ttm_buffer_object *bo, + bool evict, bool interruptible, + bool no_wait_reserve, bool no_wait_gpu, + struct ttm_mem_reg *new_mem) +{ + struct radeon_device *rdev; + struct ttm_mem_reg *old_mem = &bo->mem; + int r; + + rdev = radeon_get_rdev(bo->bdev); + if (old_mem->mem_type == TTM_PL_SYSTEM && bo->ttm == NULL) { + radeon_move_null(bo, new_mem); + return 0; + } + if ((old_mem->mem_type == TTM_PL_TT && + new_mem->mem_type == TTM_PL_SYSTEM) || + (old_mem->mem_type == TTM_PL_SYSTEM && + new_mem->mem_type == TTM_PL_TT)) { + /* bind is enough */ + radeon_move_null(bo, new_mem); + return 0; + } + if (!rdev->cp.ready || rdev->asic->copy == NULL) { + /* use memcpy */ + goto memcpy; + } + + if (old_mem->mem_type == TTM_PL_VRAM && + new_mem->mem_type == TTM_PL_SYSTEM) { + r = radeon_move_vram_ram(bo, evict, interruptible, + no_wait_reserve, no_wait_gpu, new_mem); + } else if (old_mem->mem_type == TTM_PL_SYSTEM && + new_mem->mem_type == TTM_PL_VRAM) { + r = radeon_move_ram_vram(bo, evict, interruptible, + no_wait_reserve, no_wait_gpu, new_mem); + } else { + r = radeon_move_blit(bo, evict, no_wait_reserve, no_wait_gpu, new_mem, old_mem); + } + + if (r) { +memcpy: + r = ttm_bo_move_memcpy(bo, evict, no_wait_reserve, no_wait_gpu, new_mem); + } + return r; +} + +static int radeon_ttm_io_mem_reserve(struct ttm_bo_device *bdev, struct ttm_mem_reg *mem) +{ + struct ttm_mem_type_manager *man = &bdev->man[mem->mem_type]; + struct radeon_device *rdev = radeon_get_rdev(bdev); + + mem->bus.addr = NULL; + mem->bus.offset = 0; + mem->bus.size = mem->num_pages << PAGE_SHIFT; + mem->bus.base = 0; + mem->bus.is_iomem = false; + if (!(man->flags & TTM_MEMTYPE_FLAG_MAPPABLE)) + return -EINVAL; + switch (mem->mem_type) { + case TTM_PL_SYSTEM: + /* system memory */ + return 0; + case TTM_PL_TT: +#if __OS_HAS_AGP + if (rdev->flags & RADEON_IS_AGP) { + /* RADEON_IS_AGP is set only if AGP is active */ + mem->bus.offset = mem->start << PAGE_SHIFT; + mem->bus.base = rdev->mc.agp_base; + mem->bus.is_iomem = !rdev->ddev->agp->cant_use_aperture; + } +#endif + break; + case TTM_PL_VRAM: + mem->bus.offset = mem->start << PAGE_SHIFT; + /* check if it's visible */ + if ((mem->bus.offset + mem->bus.size) > rdev->mc.visible_vram_size) + return -EINVAL; + mem->bus.base = rdev->mc.aper_base; + mem->bus.is_iomem = true; + break; + default: + return -EINVAL; + } + return 0; +} + +static void radeon_ttm_io_mem_free(struct ttm_bo_device *bdev, struct ttm_mem_reg *mem) +{ +} + +static int radeon_sync_obj_wait(void *sync_obj, void *sync_arg, + bool lazy, bool interruptible) +{ + return radeon_fence_wait((struct radeon_fence *)sync_obj, interruptible); +} + +static int radeon_sync_obj_flush(void *sync_obj, void *sync_arg) +{ + return 0; +} + +static void radeon_sync_obj_unref(void **sync_obj) +{ + radeon_fence_unref((struct radeon_fence **)sync_obj); +} + +static void *radeon_sync_obj_ref(void *sync_obj) +{ + return radeon_fence_ref((struct radeon_fence *)sync_obj); +} + +static bool radeon_sync_obj_signaled(void *sync_obj, void *sync_arg) +{ + return radeon_fence_signaled((struct radeon_fence *)sync_obj); +} + +static struct ttm_bo_driver radeon_bo_driver = { + .create_ttm_backend_entry = &radeon_create_ttm_backend_entry, + .invalidate_caches = &radeon_invalidate_caches, + .init_mem_type = &radeon_init_mem_type, + .evict_flags = &radeon_evict_flags, + .move = &radeon_bo_move, + .verify_access = &radeon_verify_access, + .sync_obj_signaled = &radeon_sync_obj_signaled, + .sync_obj_wait = &radeon_sync_obj_wait, + .sync_obj_flush = &radeon_sync_obj_flush, + .sync_obj_unref = &radeon_sync_obj_unref, + .sync_obj_ref = &radeon_sync_obj_ref, + .move_notify = &radeon_bo_move_notify, + .fault_reserve_notify = &radeon_bo_fault_reserve_notify, + .io_mem_reserve = &radeon_ttm_io_mem_reserve, + .io_mem_free = &radeon_ttm_io_mem_free, +}; + +int radeon_ttm_init(struct radeon_device *rdev) +{ + int r; + + r = radeon_ttm_global_init(rdev); + if (r) { + return r; + } + /* No others user of address space so set it to 0 */ + r = ttm_bo_device_init(&rdev->mman.bdev, + rdev->mman.bo_global_ref.ref.object, + &radeon_bo_driver, DRM_FILE_PAGE_OFFSET, + rdev->need_dma32); + if (r) { + DRM_ERROR("failed initializing buffer object driver(%d).\n", r); + return r; + } + rdev->mman.initialized = true; + r = ttm_bo_init_mm(&rdev->mman.bdev, TTM_PL_VRAM, + rdev->mc.real_vram_size >> PAGE_SHIFT); + if (r) { + DRM_ERROR("Failed initializing VRAM heap.\n"); + return r; + } + r = radeon_bo_create(rdev, 256 * 1024, PAGE_SIZE, true, + RADEON_GEM_DOMAIN_VRAM, + &rdev->stollen_vga_memory); + if (r) { + return r; + } + r = radeon_bo_reserve(rdev->stollen_vga_memory, false); + if (r) + return r; + r = radeon_bo_pin(rdev->stollen_vga_memory, RADEON_GEM_DOMAIN_VRAM, NULL); + radeon_bo_unreserve(rdev->stollen_vga_memory); + if (r) { + radeon_bo_unref(&rdev->stollen_vga_memory); + return r; + } + DRM_INFO("radeon: %uM of VRAM memory ready\n", + (unsigned)rdev->mc.real_vram_size / (1024 * 1024)); + r = ttm_bo_init_mm(&rdev->mman.bdev, TTM_PL_TT, + rdev->mc.gtt_size >> PAGE_SHIFT); + if (r) { + DRM_ERROR("Failed initializing GTT heap.\n"); + return r; + } + DRM_INFO("radeon: %uM of GTT memory ready.\n", + (unsigned)(rdev->mc.gtt_size / (1024 * 1024))); + if (unlikely(rdev->mman.bdev.dev_mapping == NULL)) { + rdev->mman.bdev.dev_mapping = rdev->ddev->dev_mapping; + } + + r = radeon_ttm_debugfs_init(rdev); + if (r) { + DRM_ERROR("Failed to init debugfs\n"); + return r; + } + return 0; +} + +void radeon_ttm_fini(struct radeon_device *rdev) +{ + int r; + + if (!rdev->mman.initialized) + return; + if (rdev->stollen_vga_memory) { + r = radeon_bo_reserve(rdev->stollen_vga_memory, false); + if (r == 0) { + radeon_bo_unpin(rdev->stollen_vga_memory); + radeon_bo_unreserve(rdev->stollen_vga_memory); + } + radeon_bo_unref(&rdev->stollen_vga_memory); + } + ttm_bo_clean_mm(&rdev->mman.bdev, TTM_PL_VRAM); + ttm_bo_clean_mm(&rdev->mman.bdev, TTM_PL_TT); + ttm_bo_device_release(&rdev->mman.bdev); + radeon_gart_fini(rdev); + radeon_ttm_global_fini(rdev); + rdev->mman.initialized = false; + DRM_INFO("radeon: ttm finalized\n"); +} + +/* this should only be called at bootup or when userspace + * isn't running */ +void radeon_ttm_set_active_vram_size(struct radeon_device *rdev, u64 size) +{ + struct ttm_mem_type_manager *man; + + if (!rdev->mman.initialized) + return; + + man = &rdev->mman.bdev.man[TTM_PL_VRAM]; + /* this just adjusts TTM size idea, which sets lpfn to the correct value */ + man->size = size >> PAGE_SHIFT; +} + +static struct vm_operations_struct radeon_ttm_vm_ops; +static const struct vm_operations_struct *ttm_vm_ops = NULL; + +static int radeon_ttm_fault(struct vm_area_struct *vma, struct vm_fault *vmf) +{ + struct ttm_buffer_object *bo; + struct radeon_device *rdev; + int r; + + bo = (struct ttm_buffer_object *)vma->vm_private_data; + if (bo == NULL) { + return VM_FAULT_NOPAGE; + } + rdev = radeon_get_rdev(bo->bdev); + mutex_lock(&rdev->vram_mutex); + r = ttm_vm_ops->fault(vma, vmf); + mutex_unlock(&rdev->vram_mutex); + return r; +} + +int radeon_mmap(struct file *filp, struct vm_area_struct *vma) +{ + struct drm_file *file_priv; + struct radeon_device *rdev; + int r; + + if (unlikely(vma->vm_pgoff < DRM_FILE_PAGE_OFFSET)) { + return drm_mmap(filp, vma); + } + + file_priv = filp->private_data; + rdev = file_priv->minor->dev->dev_private; + if (rdev == NULL) { + return -EINVAL; + } + r = ttm_bo_mmap(filp, vma, &rdev->mman.bdev); + if (unlikely(r != 0)) { + return r; + } + if (unlikely(ttm_vm_ops == NULL)) { + ttm_vm_ops = vma->vm_ops; + radeon_ttm_vm_ops = *ttm_vm_ops; + radeon_ttm_vm_ops.fault = &radeon_ttm_fault; + } + vma->vm_ops = &radeon_ttm_vm_ops; + return 0; +} + + +/* + * TTM backend functions. + */ +struct radeon_ttm_backend { + struct ttm_backend backend; + struct radeon_device *rdev; + unsigned long num_pages; + struct page **pages; + struct page *dummy_read_page; + dma_addr_t *dma_addrs; + bool populated; + bool bound; + unsigned offset; +}; + +static int radeon_ttm_backend_populate(struct ttm_backend *backend, + unsigned long num_pages, + struct page **pages, + struct page *dummy_read_page, + dma_addr_t *dma_addrs) +{ + struct radeon_ttm_backend *gtt; + + gtt = container_of(backend, struct radeon_ttm_backend, backend); + gtt->pages = pages; + gtt->dma_addrs = dma_addrs; + gtt->num_pages = num_pages; + gtt->dummy_read_page = dummy_read_page; + gtt->populated = true; + return 0; +} + +static void radeon_ttm_backend_clear(struct ttm_backend *backend) +{ + struct radeon_ttm_backend *gtt; + + gtt = container_of(backend, struct radeon_ttm_backend, backend); + gtt->pages = NULL; + gtt->dma_addrs = NULL; + gtt->num_pages = 0; + gtt->dummy_read_page = NULL; + gtt->populated = false; + gtt->bound = false; +} + + +static int radeon_ttm_backend_bind(struct ttm_backend *backend, + struct ttm_mem_reg *bo_mem) +{ + struct radeon_ttm_backend *gtt; + int r; + + gtt = container_of(backend, struct radeon_ttm_backend, backend); + gtt->offset = bo_mem->start << PAGE_SHIFT; + if (!gtt->num_pages) { + WARN(1, "nothing to bind %lu pages for mreg %p back %p!\n", + gtt->num_pages, bo_mem, backend); + } + r = radeon_gart_bind(gtt->rdev, gtt->offset, + gtt->num_pages, gtt->pages, gtt->dma_addrs); + if (r) { + DRM_ERROR("failed to bind %lu pages at 0x%08X\n", + gtt->num_pages, gtt->offset); + return r; + } + gtt->bound = true; + return 0; +} + +static int radeon_ttm_backend_unbind(struct ttm_backend *backend) +{ + struct radeon_ttm_backend *gtt; + + gtt = container_of(backend, struct radeon_ttm_backend, backend); + radeon_gart_unbind(gtt->rdev, gtt->offset, gtt->num_pages); + gtt->bound = false; + return 0; +} + +static void radeon_ttm_backend_destroy(struct ttm_backend *backend) +{ + struct radeon_ttm_backend *gtt; + + gtt = container_of(backend, struct radeon_ttm_backend, backend); + if (gtt->bound) { + radeon_ttm_backend_unbind(backend); + } + kfree(gtt); +} + +static struct ttm_backend_func radeon_backend_func = { + .populate = &radeon_ttm_backend_populate, + .clear = &radeon_ttm_backend_clear, + .bind = &radeon_ttm_backend_bind, + .unbind = &radeon_ttm_backend_unbind, + .destroy = &radeon_ttm_backend_destroy, +}; + +struct ttm_backend *radeon_ttm_backend_create(struct radeon_device *rdev) +{ + struct radeon_ttm_backend *gtt; + + gtt = kzalloc(sizeof(struct radeon_ttm_backend), GFP_KERNEL); + if (gtt == NULL) { + return NULL; + } + gtt->backend.bdev = &rdev->mman.bdev; + gtt->backend.flags = 0; + gtt->backend.func = &radeon_backend_func; + gtt->rdev = rdev; + gtt->pages = NULL; + gtt->num_pages = 0; + gtt->dummy_read_page = NULL; + gtt->populated = false; + gtt->bound = false; + return >t->backend; +} + +#define RADEON_DEBUGFS_MEM_TYPES 2 + +#if defined(CONFIG_DEBUG_FS) +static int radeon_mm_dump_table(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *)m->private; + struct drm_mm *mm = (struct drm_mm *)node->info_ent->data; + struct drm_device *dev = node->minor->dev; + struct radeon_device *rdev = dev->dev_private; + int ret; + struct ttm_bo_global *glob = rdev->mman.bdev.glob; + + spin_lock(&glob->lru_lock); + ret = drm_mm_dump_table(m, mm); + spin_unlock(&glob->lru_lock); + return ret; +} +#endif + +static int radeon_ttm_debugfs_init(struct radeon_device *rdev) +{ +#if defined(CONFIG_DEBUG_FS) + static struct drm_info_list radeon_mem_types_list[RADEON_DEBUGFS_MEM_TYPES+1]; + static char radeon_mem_types_names[RADEON_DEBUGFS_MEM_TYPES+1][32]; + unsigned i; + + for (i = 0; i < RADEON_DEBUGFS_MEM_TYPES; i++) { + if (i == 0) + sprintf(radeon_mem_types_names[i], "radeon_vram_mm"); + else + sprintf(radeon_mem_types_names[i], "radeon_gtt_mm"); + radeon_mem_types_list[i].name = radeon_mem_types_names[i]; + radeon_mem_types_list[i].show = &radeon_mm_dump_table; + radeon_mem_types_list[i].driver_features = 0; + if (i == 0) + radeon_mem_types_list[i].data = rdev->mman.bdev.man[TTM_PL_VRAM].priv; + else + radeon_mem_types_list[i].data = rdev->mman.bdev.man[TTM_PL_TT].priv; + + } + /* Add ttm page pool to debugfs */ + sprintf(radeon_mem_types_names[i], "ttm_page_pool"); + radeon_mem_types_list[i].name = radeon_mem_types_names[i]; + radeon_mem_types_list[i].show = &ttm_page_alloc_debugfs; + radeon_mem_types_list[i].driver_features = 0; + radeon_mem_types_list[i].data = NULL; + return radeon_debugfs_add_files(rdev, radeon_mem_types_list, RADEON_DEBUGFS_MEM_TYPES+1); + +#endif + return 0; +} diff --git a/drivers/gpu/drm/radeon/reg_srcs/cayman b/drivers/gpu/drm/radeon/reg_srcs/cayman new file mode 100644 index 00000000..0aa8e85a --- /dev/null +++ b/drivers/gpu/drm/radeon/reg_srcs/cayman @@ -0,0 +1,620 @@ +cayman 0x9400 +0x0000802C GRBM_GFX_INDEX +0x000088B0 VGT_VTX_VECT_EJECT_REG +0x000088C4 VGT_CACHE_INVALIDATION +0x000088D4 VGT_GS_VERTEX_REUSE +0x00008958 VGT_PRIMITIVE_TYPE +0x0000895C VGT_INDEX_TYPE +0x00008970 VGT_NUM_INDICES +0x00008974 VGT_NUM_INSTANCES +0x00008990 VGT_COMPUTE_DIM_X +0x00008994 VGT_COMPUTE_DIM_Y +0x00008998 VGT_COMPUTE_DIM_Z +0x0000899C VGT_COMPUTE_START_X +0x000089A0 VGT_COMPUTE_START_Y +0x000089A4 VGT_COMPUTE_START_Z +0x000089A8 VGT_COMPUTE_INDEX +0x000089AC VGT_COMPUTE_THREAD_GOURP_SIZE +0x000089B0 VGT_HS_OFFCHIP_PARAM +0x00008A14 PA_CL_ENHANCE +0x00008A60 PA_SC_LINE_STIPPLE_VALUE +0x00008B10 PA_SC_LINE_STIPPLE_STATE +0x00008BF0 PA_SC_ENHANCE +0x00008D8C SQ_DYN_GPR_CNTL_PS_FLUSH_REQ +0x00008D94 SQ_DYN_GPR_SIMD_LOCK_EN +0x00008C00 SQ_CONFIG +0x00008C04 SQ_GPR_RESOURCE_MGMT_1 +0x00008C10 SQ_GLOBAL_GPR_RESOURCE_MGMT_1 +0x00008C14 SQ_GLOBAL_GPR_RESOURCE_MGMT_2 +0x00008DF8 SQ_CONST_MEM_BASE +0x00008E20 SQ_STATIC_THREAD_MGMT_1 +0x00008E24 SQ_STATIC_THREAD_MGMT_2 +0x00008E28 SQ_STATIC_THREAD_MGMT_3 +0x00008E48 SQ_EX_ALLOC_TABLE_SLOTS +0x00009100 SPI_CONFIG_CNTL +0x0000913C SPI_CONFIG_CNTL_1 +0x00009508 TA_CNTL_AUX +0x00009830 DB_DEBUG +0x00009834 DB_DEBUG2 +0x00009838 DB_DEBUG3 +0x0000983C DB_DEBUG4 +0x00009854 DB_WATERMARKS +0x0000A400 TD_PS_BORDER_COLOR_INDEX +0x0000A404 TD_PS_BORDER_COLOR_RED +0x0000A408 TD_PS_BORDER_COLOR_GREEN +0x0000A40C TD_PS_BORDER_COLOR_BLUE +0x0000A410 TD_PS_BORDER_COLOR_ALPHA +0x0000A414 TD_VS_BORDER_COLOR_INDEX +0x0000A418 TD_VS_BORDER_COLOR_RED +0x0000A41C TD_VS_BORDER_COLOR_GREEN +0x0000A420 TD_VS_BORDER_COLOR_BLUE +0x0000A424 TD_VS_BORDER_COLOR_ALPHA +0x0000A428 TD_GS_BORDER_COLOR_INDEX +0x0000A42C TD_GS_BORDER_COLOR_RED +0x0000A430 TD_GS_BORDER_COLOR_GREEN +0x0000A434 TD_GS_BORDER_COLOR_BLUE +0x0000A438 TD_GS_BORDER_COLOR_ALPHA +0x0000A43C TD_HS_BORDER_COLOR_INDEX +0x0000A440 TD_HS_BORDER_COLOR_RED +0x0000A444 TD_HS_BORDER_COLOR_GREEN +0x0000A448 TD_HS_BORDER_COLOR_BLUE +0x0000A44C TD_HS_BORDER_COLOR_ALPHA +0x0000A450 TD_LS_BORDER_COLOR_INDEX +0x0000A454 TD_LS_BORDER_COLOR_RED +0x0000A458 TD_LS_BORDER_COLOR_GREEN +0x0000A45C TD_LS_BORDER_COLOR_BLUE +0x0000A460 TD_LS_BORDER_COLOR_ALPHA +0x0000A464 TD_CS_BORDER_COLOR_INDEX +0x0000A468 TD_CS_BORDER_COLOR_RED +0x0000A46C TD_CS_BORDER_COLOR_GREEN +0x0000A470 TD_CS_BORDER_COLOR_BLUE +0x0000A474 TD_CS_BORDER_COLOR_ALPHA +0x00028000 DB_RENDER_CONTROL +0x00028004 DB_COUNT_CONTROL +0x0002800C DB_RENDER_OVERRIDE +0x00028010 DB_RENDER_OVERRIDE2 +0x00028028 DB_STENCIL_CLEAR +0x0002802C DB_DEPTH_CLEAR +0x00028030 PA_SC_SCREEN_SCISSOR_TL +0x00028034 PA_SC_SCREEN_SCISSOR_BR +0x0002805C DB_DEPTH_SLICE +0x00028140 SQ_ALU_CONST_BUFFER_SIZE_PS_0 +0x00028144 SQ_ALU_CONST_BUFFER_SIZE_PS_1 +0x00028148 SQ_ALU_CONST_BUFFER_SIZE_PS_2 +0x0002814C SQ_ALU_CONST_BUFFER_SIZE_PS_3 +0x00028150 SQ_ALU_CONST_BUFFER_SIZE_PS_4 +0x00028154 SQ_ALU_CONST_BUFFER_SIZE_PS_5 +0x00028158 SQ_ALU_CONST_BUFFER_SIZE_PS_6 +0x0002815C SQ_ALU_CONST_BUFFER_SIZE_PS_7 +0x00028160 SQ_ALU_CONST_BUFFER_SIZE_PS_8 +0x00028164 SQ_ALU_CONST_BUFFER_SIZE_PS_9 +0x00028168 SQ_ALU_CONST_BUFFER_SIZE_PS_10 +0x0002816C SQ_ALU_CONST_BUFFER_SIZE_PS_11 +0x00028170 SQ_ALU_CONST_BUFFER_SIZE_PS_12 +0x00028174 SQ_ALU_CONST_BUFFER_SIZE_PS_13 +0x00028178 SQ_ALU_CONST_BUFFER_SIZE_PS_14 +0x0002817C SQ_ALU_CONST_BUFFER_SIZE_PS_15 +0x00028180 SQ_ALU_CONST_BUFFER_SIZE_VS_0 +0x00028184 SQ_ALU_CONST_BUFFER_SIZE_VS_1 +0x00028188 SQ_ALU_CONST_BUFFER_SIZE_VS_2 +0x0002818C SQ_ALU_CONST_BUFFER_SIZE_VS_3 +0x00028190 SQ_ALU_CONST_BUFFER_SIZE_VS_4 +0x00028194 SQ_ALU_CONST_BUFFER_SIZE_VS_5 +0x00028198 SQ_ALU_CONST_BUFFER_SIZE_VS_6 +0x0002819C SQ_ALU_CONST_BUFFER_SIZE_VS_7 +0x000281A0 SQ_ALU_CONST_BUFFER_SIZE_VS_8 +0x000281A4 SQ_ALU_CONST_BUFFER_SIZE_VS_9 +0x000281A8 SQ_ALU_CONST_BUFFER_SIZE_VS_10 +0x000281AC SQ_ALU_CONST_BUFFER_SIZE_VS_11 +0x000281B0 SQ_ALU_CONST_BUFFER_SIZE_VS_12 +0x000281B4 SQ_ALU_CONST_BUFFER_SIZE_VS_13 +0x000281B8 SQ_ALU_CONST_BUFFER_SIZE_VS_14 +0x000281BC SQ_ALU_CONST_BUFFER_SIZE_VS_15 +0x000281C0 SQ_ALU_CONST_BUFFER_SIZE_GS_0 +0x000281C4 SQ_ALU_CONST_BUFFER_SIZE_GS_1 +0x000281C8 SQ_ALU_CONST_BUFFER_SIZE_GS_2 +0x000281CC SQ_ALU_CONST_BUFFER_SIZE_GS_3 +0x000281D0 SQ_ALU_CONST_BUFFER_SIZE_GS_4 +0x000281D4 SQ_ALU_CONST_BUFFER_SIZE_GS_5 +0x000281D8 SQ_ALU_CONST_BUFFER_SIZE_GS_6 +0x000281DC SQ_ALU_CONST_BUFFER_SIZE_GS_7 +0x000281E0 SQ_ALU_CONST_BUFFER_SIZE_GS_8 +0x000281E4 SQ_ALU_CONST_BUFFER_SIZE_GS_9 +0x000281E8 SQ_ALU_CONST_BUFFER_SIZE_GS_10 +0x000281EC SQ_ALU_CONST_BUFFER_SIZE_GS_11 +0x000281F0 SQ_ALU_CONST_BUFFER_SIZE_GS_12 +0x000281F4 SQ_ALU_CONST_BUFFER_SIZE_GS_13 +0x000281F8 SQ_ALU_CONST_BUFFER_SIZE_GS_14 +0x000281FC SQ_ALU_CONST_BUFFER_SIZE_GS_15 +0x00028200 PA_SC_WINDOW_OFFSET +0x00028204 PA_SC_WINDOW_SCISSOR_TL +0x00028208 PA_SC_WINDOW_SCISSOR_BR +0x0002820C PA_SC_CLIPRECT_RULE +0x00028210 PA_SC_CLIPRECT_0_TL +0x00028214 PA_SC_CLIPRECT_0_BR +0x00028218 PA_SC_CLIPRECT_1_TL +0x0002821C PA_SC_CLIPRECT_1_BR +0x00028220 PA_SC_CLIPRECT_2_TL +0x00028224 PA_SC_CLIPRECT_2_BR +0x00028228 PA_SC_CLIPRECT_3_TL +0x0002822C PA_SC_CLIPRECT_3_BR +0x00028230 PA_SC_EDGERULE +0x00028234 PA_SU_HARDWARE_SCREEN_OFFSET +0x00028240 PA_SC_GENERIC_SCISSOR_TL +0x00028244 PA_SC_GENERIC_SCISSOR_BR +0x00028250 PA_SC_VPORT_SCISSOR_0_TL +0x00028254 PA_SC_VPORT_SCISSOR_0_BR +0x00028258 PA_SC_VPORT_SCISSOR_1_TL +0x0002825C PA_SC_VPORT_SCISSOR_1_BR +0x00028260 PA_SC_VPORT_SCISSOR_2_TL +0x00028264 PA_SC_VPORT_SCISSOR_2_BR +0x00028268 PA_SC_VPORT_SCISSOR_3_TL +0x0002826C PA_SC_VPORT_SCISSOR_3_BR +0x00028270 PA_SC_VPORT_SCISSOR_4_TL +0x00028274 PA_SC_VPORT_SCISSOR_4_BR +0x00028278 PA_SC_VPORT_SCISSOR_5_TL +0x0002827C PA_SC_VPORT_SCISSOR_5_BR +0x00028280 PA_SC_VPORT_SCISSOR_6_TL +0x00028284 PA_SC_VPORT_SCISSOR_6_BR +0x00028288 PA_SC_VPORT_SCISSOR_7_TL +0x0002828C PA_SC_VPORT_SCISSOR_7_BR +0x00028290 PA_SC_VPORT_SCISSOR_8_TL +0x00028294 PA_SC_VPORT_SCISSOR_8_BR +0x00028298 PA_SC_VPORT_SCISSOR_9_TL +0x0002829C PA_SC_VPORT_SCISSOR_9_BR +0x000282A0 PA_SC_VPORT_SCISSOR_10_TL +0x000282A4 PA_SC_VPORT_SCISSOR_10_BR +0x000282A8 PA_SC_VPORT_SCISSOR_11_TL +0x000282AC PA_SC_VPORT_SCISSOR_11_BR +0x000282B0 PA_SC_VPORT_SCISSOR_12_TL +0x000282B4 PA_SC_VPORT_SCISSOR_12_BR +0x000282B8 PA_SC_VPORT_SCISSOR_13_TL +0x000282BC PA_SC_VPORT_SCISSOR_13_BR +0x000282C0 PA_SC_VPORT_SCISSOR_14_TL +0x000282C4 PA_SC_VPORT_SCISSOR_14_BR +0x000282C8 PA_SC_VPORT_SCISSOR_15_TL +0x000282CC PA_SC_VPORT_SCISSOR_15_BR +0x000282D0 PA_SC_VPORT_ZMIN_0 +0x000282D4 PA_SC_VPORT_ZMAX_0 +0x000282D8 PA_SC_VPORT_ZMIN_1 +0x000282DC PA_SC_VPORT_ZMAX_1 +0x000282E0 PA_SC_VPORT_ZMIN_2 +0x000282E4 PA_SC_VPORT_ZMAX_2 +0x000282E8 PA_SC_VPORT_ZMIN_3 +0x000282EC PA_SC_VPORT_ZMAX_3 +0x000282F0 PA_SC_VPORT_ZMIN_4 +0x000282F4 PA_SC_VPORT_ZMAX_4 +0x000282F8 PA_SC_VPORT_ZMIN_5 +0x000282FC PA_SC_VPORT_ZMAX_5 +0x00028300 PA_SC_VPORT_ZMIN_6 +0x00028304 PA_SC_VPORT_ZMAX_6 +0x00028308 PA_SC_VPORT_ZMIN_7 +0x0002830C PA_SC_VPORT_ZMAX_7 +0x00028310 PA_SC_VPORT_ZMIN_8 +0x00028314 PA_SC_VPORT_ZMAX_8 +0x00028318 PA_SC_VPORT_ZMIN_9 +0x0002831C PA_SC_VPORT_ZMAX_9 +0x00028320 PA_SC_VPORT_ZMIN_10 +0x00028324 PA_SC_VPORT_ZMAX_10 +0x00028328 PA_SC_VPORT_ZMIN_11 +0x0002832C PA_SC_VPORT_ZMAX_11 +0x00028330 PA_SC_VPORT_ZMIN_12 +0x00028334 PA_SC_VPORT_ZMAX_12 +0x00028338 PA_SC_VPORT_ZMIN_13 +0x0002833C PA_SC_VPORT_ZMAX_13 +0x00028340 PA_SC_VPORT_ZMIN_14 +0x00028344 PA_SC_VPORT_ZMAX_14 +0x00028348 PA_SC_VPORT_ZMIN_15 +0x0002834C PA_SC_VPORT_ZMAX_15 +0x00028350 SX_MISC +0x00028354 SX_SURFACE_SYNC +0x00028380 SQ_VTX_SEMANTIC_0 +0x00028384 SQ_VTX_SEMANTIC_1 +0x00028388 SQ_VTX_SEMANTIC_2 +0x0002838C SQ_VTX_SEMANTIC_3 +0x00028390 SQ_VTX_SEMANTIC_4 +0x00028394 SQ_VTX_SEMANTIC_5 +0x00028398 SQ_VTX_SEMANTIC_6 +0x0002839C SQ_VTX_SEMANTIC_7 +0x000283A0 SQ_VTX_SEMANTIC_8 +0x000283A4 SQ_VTX_SEMANTIC_9 +0x000283A8 SQ_VTX_SEMANTIC_10 +0x000283AC SQ_VTX_SEMANTIC_11 +0x000283B0 SQ_VTX_SEMANTIC_12 +0x000283B4 SQ_VTX_SEMANTIC_13 +0x000283B8 SQ_VTX_SEMANTIC_14 +0x000283BC SQ_VTX_SEMANTIC_15 +0x000283C0 SQ_VTX_SEMANTIC_16 +0x000283C4 SQ_VTX_SEMANTIC_17 +0x000283C8 SQ_VTX_SEMANTIC_18 +0x000283CC SQ_VTX_SEMANTIC_19 +0x000283D0 SQ_VTX_SEMANTIC_20 +0x000283D4 SQ_VTX_SEMANTIC_21 +0x000283D8 SQ_VTX_SEMANTIC_22 +0x000283DC SQ_VTX_SEMANTIC_23 +0x000283E0 SQ_VTX_SEMANTIC_24 +0x000283E4 SQ_VTX_SEMANTIC_25 +0x000283E8 SQ_VTX_SEMANTIC_26 +0x000283EC SQ_VTX_SEMANTIC_27 +0x000283F0 SQ_VTX_SEMANTIC_28 +0x000283F4 SQ_VTX_SEMANTIC_29 +0x000283F8 SQ_VTX_SEMANTIC_30 +0x000283FC SQ_VTX_SEMANTIC_31 +0x00028400 VGT_MAX_VTX_INDX +0x00028404 VGT_MIN_VTX_INDX +0x00028408 VGT_INDX_OFFSET +0x0002840C VGT_MULTI_PRIM_IB_RESET_INDX +0x00028410 SX_ALPHA_TEST_CONTROL +0x00028414 CB_BLEND_RED +0x00028418 CB_BLEND_GREEN +0x0002841C CB_BLEND_BLUE +0x00028420 CB_BLEND_ALPHA +0x00028430 DB_STENCILREFMASK +0x00028434 DB_STENCILREFMASK_BF +0x00028438 SX_ALPHA_REF +0x0002843C PA_CL_VPORT_XSCALE_0 +0x00028440 PA_CL_VPORT_XOFFSET_0 +0x00028444 PA_CL_VPORT_YSCALE_0 +0x00028448 PA_CL_VPORT_YOFFSET_0 +0x0002844C PA_CL_VPORT_ZSCALE_0 +0x00028450 PA_CL_VPORT_ZOFFSET_0 +0x00028454 PA_CL_VPORT_XSCALE_1 +0x00028458 PA_CL_VPORT_XOFFSET_1 +0x0002845C PA_CL_VPORT_YSCALE_1 +0x00028460 PA_CL_VPORT_YOFFSET_1 +0x00028464 PA_CL_VPORT_ZSCALE_1 +0x00028468 PA_CL_VPORT_ZOFFSET_1 +0x0002846C PA_CL_VPORT_XSCALE_2 +0x00028470 PA_CL_VPORT_XOFFSET_2 +0x00028474 PA_CL_VPORT_YSCALE_2 +0x00028478 PA_CL_VPORT_YOFFSET_2 +0x0002847C PA_CL_VPORT_ZSCALE_2 +0x00028480 PA_CL_VPORT_ZOFFSET_2 +0x00028484 PA_CL_VPORT_XSCALE_3 +0x00028488 PA_CL_VPORT_XOFFSET_3 +0x0002848C PA_CL_VPORT_YSCALE_3 +0x00028490 PA_CL_VPORT_YOFFSET_3 +0x00028494 PA_CL_VPORT_ZSCALE_3 +0x00028498 PA_CL_VPORT_ZOFFSET_3 +0x0002849C PA_CL_VPORT_XSCALE_4 +0x000284A0 PA_CL_VPORT_XOFFSET_4 +0x000284A4 PA_CL_VPORT_YSCALE_4 +0x000284A8 PA_CL_VPORT_YOFFSET_4 +0x000284AC PA_CL_VPORT_ZSCALE_4 +0x000284B0 PA_CL_VPORT_ZOFFSET_4 +0x000284B4 PA_CL_VPORT_XSCALE_5 +0x000284B8 PA_CL_VPORT_XOFFSET_5 +0x000284BC PA_CL_VPORT_YSCALE_5 +0x000284C0 PA_CL_VPORT_YOFFSET_5 +0x000284C4 PA_CL_VPORT_ZSCALE_5 +0x000284C8 PA_CL_VPORT_ZOFFSET_5 +0x000284CC PA_CL_VPORT_XSCALE_6 +0x000284D0 PA_CL_VPORT_XOFFSET_6 +0x000284D4 PA_CL_VPORT_YSCALE_6 +0x000284D8 PA_CL_VPORT_YOFFSET_6 +0x000284DC PA_CL_VPORT_ZSCALE_6 +0x000284E0 PA_CL_VPORT_ZOFFSET_6 +0x000284E4 PA_CL_VPORT_XSCALE_7 +0x000284E8 PA_CL_VPORT_XOFFSET_7 +0x000284EC PA_CL_VPORT_YSCALE_7 +0x000284F0 PA_CL_VPORT_YOFFSET_7 +0x000284F4 PA_CL_VPORT_ZSCALE_7 +0x000284F8 PA_CL_VPORT_ZOFFSET_7 +0x000284FC PA_CL_VPORT_XSCALE_8 +0x00028500 PA_CL_VPORT_XOFFSET_8 +0x00028504 PA_CL_VPORT_YSCALE_8 +0x00028508 PA_CL_VPORT_YOFFSET_8 +0x0002850C PA_CL_VPORT_ZSCALE_8 +0x00028510 PA_CL_VPORT_ZOFFSET_8 +0x00028514 PA_CL_VPORT_XSCALE_9 +0x00028518 PA_CL_VPORT_XOFFSET_9 +0x0002851C PA_CL_VPORT_YSCALE_9 +0x00028520 PA_CL_VPORT_YOFFSET_9 +0x00028524 PA_CL_VPORT_ZSCALE_9 +0x00028528 PA_CL_VPORT_ZOFFSET_9 +0x0002852C PA_CL_VPORT_XSCALE_10 +0x00028530 PA_CL_VPORT_XOFFSET_10 +0x00028534 PA_CL_VPORT_YSCALE_10 +0x00028538 PA_CL_VPORT_YOFFSET_10 +0x0002853C PA_CL_VPORT_ZSCALE_10 +0x00028540 PA_CL_VPORT_ZOFFSET_10 +0x00028544 PA_CL_VPORT_XSCALE_11 +0x00028548 PA_CL_VPORT_XOFFSET_11 +0x0002854C PA_CL_VPORT_YSCALE_11 +0x00028550 PA_CL_VPORT_YOFFSET_11 +0x00028554 PA_CL_VPORT_ZSCALE_11 +0x00028558 PA_CL_VPORT_ZOFFSET_11 +0x0002855C PA_CL_VPORT_XSCALE_12 +0x00028560 PA_CL_VPORT_XOFFSET_12 +0x00028564 PA_CL_VPORT_YSCALE_12 +0x00028568 PA_CL_VPORT_YOFFSET_12 +0x0002856C PA_CL_VPORT_ZSCALE_12 +0x00028570 PA_CL_VPORT_ZOFFSET_12 +0x00028574 PA_CL_VPORT_XSCALE_13 +0x00028578 PA_CL_VPORT_XOFFSET_13 +0x0002857C PA_CL_VPORT_YSCALE_13 +0x00028580 PA_CL_VPORT_YOFFSET_13 +0x00028584 PA_CL_VPORT_ZSCALE_13 +0x00028588 PA_CL_VPORT_ZOFFSET_13 +0x0002858C PA_CL_VPORT_XSCALE_14 +0x00028590 PA_CL_VPORT_XOFFSET_14 +0x00028594 PA_CL_VPORT_YSCALE_14 +0x00028598 PA_CL_VPORT_YOFFSET_14 +0x0002859C PA_CL_VPORT_ZSCALE_14 +0x000285A0 PA_CL_VPORT_ZOFFSET_14 +0x000285A4 PA_CL_VPORT_XSCALE_15 +0x000285A8 PA_CL_VPORT_XOFFSET_15 +0x000285AC PA_CL_VPORT_YSCALE_15 +0x000285B0 PA_CL_VPORT_YOFFSET_15 +0x000285B4 PA_CL_VPORT_ZSCALE_15 +0x000285B8 PA_CL_VPORT_ZOFFSET_15 +0x000285BC PA_CL_UCP_0_X +0x000285C0 PA_CL_UCP_0_Y +0x000285C4 PA_CL_UCP_0_Z +0x000285C8 PA_CL_UCP_0_W +0x000285CC PA_CL_UCP_1_X +0x000285D0 PA_CL_UCP_1_Y +0x000285D4 PA_CL_UCP_1_Z +0x000285D8 PA_CL_UCP_1_W +0x000285DC PA_CL_UCP_2_X +0x000285E0 PA_CL_UCP_2_Y +0x000285E4 PA_CL_UCP_2_Z +0x000285E8 PA_CL_UCP_2_W +0x000285EC PA_CL_UCP_3_X +0x000285F0 PA_CL_UCP_3_Y +0x000285F4 PA_CL_UCP_3_Z +0x000285F8 PA_CL_UCP_3_W +0x000285FC PA_CL_UCP_4_X +0x00028600 PA_CL_UCP_4_Y +0x00028604 PA_CL_UCP_4_Z +0x00028608 PA_CL_UCP_4_W +0x0002860C PA_CL_UCP_5_X +0x00028610 PA_CL_UCP_5_Y +0x00028614 PA_CL_UCP_5_Z +0x00028618 PA_CL_UCP_5_W +0x0002861C SPI_VS_OUT_ID_0 +0x00028620 SPI_VS_OUT_ID_1 +0x00028624 SPI_VS_OUT_ID_2 +0x00028628 SPI_VS_OUT_ID_3 +0x0002862C SPI_VS_OUT_ID_4 +0x00028630 SPI_VS_OUT_ID_5 +0x00028634 SPI_VS_OUT_ID_6 +0x00028638 SPI_VS_OUT_ID_7 +0x0002863C SPI_VS_OUT_ID_8 +0x00028640 SPI_VS_OUT_ID_9 +0x00028644 SPI_PS_INPUT_CNTL_0 +0x00028648 SPI_PS_INPUT_CNTL_1 +0x0002864C SPI_PS_INPUT_CNTL_2 +0x00028650 SPI_PS_INPUT_CNTL_3 +0x00028654 SPI_PS_INPUT_CNTL_4 +0x00028658 SPI_PS_INPUT_CNTL_5 +0x0002865C SPI_PS_INPUT_CNTL_6 +0x00028660 SPI_PS_INPUT_CNTL_7 +0x00028664 SPI_PS_INPUT_CNTL_8 +0x00028668 SPI_PS_INPUT_CNTL_9 +0x0002866C SPI_PS_INPUT_CNTL_10 +0x00028670 SPI_PS_INPUT_CNTL_11 +0x00028674 SPI_PS_INPUT_CNTL_12 +0x00028678 SPI_PS_INPUT_CNTL_13 +0x0002867C SPI_PS_INPUT_CNTL_14 +0x00028680 SPI_PS_INPUT_CNTL_15 +0x00028684 SPI_PS_INPUT_CNTL_16 +0x00028688 SPI_PS_INPUT_CNTL_17 +0x0002868C SPI_PS_INPUT_CNTL_18 +0x00028690 SPI_PS_INPUT_CNTL_19 +0x00028694 SPI_PS_INPUT_CNTL_20 +0x00028698 SPI_PS_INPUT_CNTL_21 +0x0002869C SPI_PS_INPUT_CNTL_22 +0x000286A0 SPI_PS_INPUT_CNTL_23 +0x000286A4 SPI_PS_INPUT_CNTL_24 +0x000286A8 SPI_PS_INPUT_CNTL_25 +0x000286AC SPI_PS_INPUT_CNTL_26 +0x000286B0 SPI_PS_INPUT_CNTL_27 +0x000286B4 SPI_PS_INPUT_CNTL_28 +0x000286B8 SPI_PS_INPUT_CNTL_29 +0x000286BC SPI_PS_INPUT_CNTL_30 +0x000286C0 SPI_PS_INPUT_CNTL_31 +0x000286C4 SPI_VS_OUT_CONFIG +0x000286C8 SPI_THREAD_GROUPING +0x000286CC SPI_PS_IN_CONTROL_0 +0x000286D0 SPI_PS_IN_CONTROL_1 +0x000286D4 SPI_INTERP_CONTROL_0 +0x000286D8 SPI_INPUT_Z +0x000286DC SPI_FOG_CNTL +0x000286E0 SPI_BARYC_CNTL +0x000286E4 SPI_PS_IN_CONTROL_2 +0x000286E8 SPI_COMPUTE_INPUT_CNTL +0x000286EC SPI_COMPUTE_NUM_THREAD_X +0x000286F0 SPI_COMPUTE_NUM_THREAD_Y +0x000286F4 SPI_COMPUTE_NUM_THREAD_Z +0x000286F8 SPI_GPR_MGMT +0x000286FC SPI_LDS_MGMT +0x00028700 SPI_STACK_MGMT +0x00028704 SPI_WAVE_MGMT_1 +0x00028708 SPI_WAVE_MGMT_2 +0x00028724 GDS_ADDR_SIZE +0x00028780 CB_BLEND0_CONTROL +0x00028784 CB_BLEND1_CONTROL +0x00028788 CB_BLEND2_CONTROL +0x0002878C CB_BLEND3_CONTROL +0x00028790 CB_BLEND4_CONTROL +0x00028794 CB_BLEND5_CONTROL +0x00028798 CB_BLEND6_CONTROL +0x0002879C CB_BLEND7_CONTROL +0x000287CC CS_COPY_STATE +0x000287D0 GFX_COPY_STATE +0x000287D4 PA_CL_POINT_X_RAD +0x000287D8 PA_CL_POINT_Y_RAD +0x000287DC PA_CL_POINT_SIZE +0x000287E0 PA_CL_POINT_CULL_RAD +0x00028808 CB_COLOR_CONTROL +0x0002880C DB_SHADER_CONTROL +0x00028810 PA_CL_CLIP_CNTL +0x00028814 PA_SU_SC_MODE_CNTL +0x00028818 PA_CL_VTE_CNTL +0x0002881C PA_CL_VS_OUT_CNTL +0x00028820 PA_CL_NANINF_CNTL +0x00028824 PA_SU_LINE_STIPPLE_CNTL +0x00028828 PA_SU_LINE_STIPPLE_SCALE +0x0002882C PA_SU_PRIM_FILTER_CNTL +0x00028844 SQ_PGM_RESOURCES_PS +0x00028848 SQ_PGM_RESOURCES_2_PS +0x0002884C SQ_PGM_EXPORTS_PS +0x00028860 SQ_PGM_RESOURCES_VS +0x00028864 SQ_PGM_RESOURCES_2_VS +0x00028878 SQ_PGM_RESOURCES_GS +0x0002887C SQ_PGM_RESOURCES_2_GS +0x00028890 SQ_PGM_RESOURCES_ES +0x00028894 SQ_PGM_RESOURCES_2_ES +0x000288A8 SQ_PGM_RESOURCES_FS +0x000288BC SQ_PGM_RESOURCES_HS +0x000288C0 SQ_PGM_RESOURCES_2_HS +0x000288D4 SQ_PGM_RESOURCES_LS +0x000288D8 SQ_PGM_RESOURCES_2_LS +0x000288E8 SQ_LDS_ALLOC +0x000288EC SQ_LDS_ALLOC_PS +0x000288F0 SQ_VTX_SEMANTIC_CLEAR +0x00028A00 PA_SU_POINT_SIZE +0x00028A04 PA_SU_POINT_MINMAX +0x00028A08 PA_SU_LINE_CNTL +0x00028A0C PA_SC_LINE_STIPPLE +0x00028A10 VGT_OUTPUT_PATH_CNTL +0x00028A14 VGT_HOS_CNTL +0x00028A18 VGT_HOS_MAX_TESS_LEVEL +0x00028A1C VGT_HOS_MIN_TESS_LEVEL +0x00028A20 VGT_HOS_REUSE_DEPTH +0x00028A24 VGT_GROUP_PRIM_TYPE +0x00028A28 VGT_GROUP_FIRST_DECR +0x00028A2C VGT_GROUP_DECR +0x00028A30 VGT_GROUP_VECT_0_CNTL +0x00028A34 VGT_GROUP_VECT_1_CNTL +0x00028A38 VGT_GROUP_VECT_0_FMT_CNTL +0x00028A3C VGT_GROUP_VECT_1_FMT_CNTL +0x00028A40 VGT_GS_MODE +0x00028A48 PA_SC_MODE_CNTL_0 +0x00028A4C PA_SC_MODE_CNTL_1 +0x00028A50 VGT_ENHANCE +0x00028A54 VGT_GS_PER_ES +0x00028A58 VGT_ES_PER_GS +0x00028A5C VGT_GS_PER_VS +0x00028A6C VGT_GS_OUT_PRIM_TYPE +0x00028A70 IA_ENHANCE +0x00028A84 VGT_PRIMITIVEID_EN +0x00028A94 VGT_MULTI_PRIM_IB_RESET_EN +0x00028AA0 VGT_INSTANCE_STEP_RATE_0 +0x00028AA4 VGT_INSTANCE_STEP_RATE_1 +0x00028AA8 IA_MULTI_VGT_PARAM +0x00028AB4 VGT_REUSE_OFF +0x00028AB8 VGT_VTX_CNT_EN +0x00028ABC DB_HTILE_SURFACE +0x00028AC0 DB_SRESULTS_COMPARE_STATE0 +0x00028AC4 DB_SRESULTS_COMPARE_STATE1 +0x00028AC8 DB_PRELOAD_CONTROL +0x00028B38 VGT_GS_MAX_VERT_OUT +0x00028B54 VGT_SHADER_STAGES_EN +0x00028B58 VGT_LS_HS_CONFIG +0x00028B6C VGT_TF_PARAM +0x00028B70 DB_ALPHA_TO_MASK +0x00028B74 VGT_DISPATCH_INITIATOR +0x00028B78 PA_SU_POLY_OFFSET_DB_FMT_CNTL +0x00028B7C PA_SU_POLY_OFFSET_CLAMP +0x00028B80 PA_SU_POLY_OFFSET_FRONT_SCALE +0x00028B84 PA_SU_POLY_OFFSET_FRONT_OFFSET +0x00028B88 PA_SU_POLY_OFFSET_BACK_SCALE +0x00028B8C PA_SU_POLY_OFFSET_BACK_OFFSET +0x00028B74 VGT_GS_INSTANCE_CNT +0x00028BD4 PA_SC_CENTROID_PRIORITY_0 +0x00028BD8 PA_SC_CENTROID_PRIORITY_1 +0x00028BDC PA_SC_LINE_CNTL +0x00028BE4 PA_SU_VTX_CNTL +0x00028BE8 PA_CL_GB_VERT_CLIP_ADJ +0x00028BEC PA_CL_GB_VERT_DISC_ADJ +0x00028BF0 PA_CL_GB_HORZ_CLIP_ADJ +0x00028BF4 PA_CL_GB_HORZ_DISC_ADJ +0x00028BF8 PA_SC_AA_SAMPLE_LOCS_PIXEL_X0_Y0_0 +0x00028BFC PA_SC_AA_SAMPLE_LOCS_PIXEL_X0_Y0_1 +0x00028C00 PA_SC_AA_SAMPLE_LOCS_PIXEL_X0_Y0_2 +0x00028C04 PA_SC_AA_SAMPLE_LOCS_PIXEL_X0_Y0_3 +0x00028C08 PA_SC_AA_SAMPLE_LOCS_PIXEL_X1_Y0_0 +0x00028C0C PA_SC_AA_SAMPLE_LOCS_PIXEL_X1_Y0_1 +0x00028C10 PA_SC_AA_SAMPLE_LOCS_PIXEL_X1_Y0_2 +0x00028C14 PA_SC_AA_SAMPLE_LOCS_PIXEL_X1_Y0_3 +0x00028C18 PA_SC_AA_SAMPLE_LOCS_PIXEL_X0_Y1_0 +0x00028C1C PA_SC_AA_SAMPLE_LOCS_PIXEL_X0_Y1_1 +0x00028C20 PA_SC_AA_SAMPLE_LOCS_PIXEL_X0_Y1_2 +0x00028C24 PA_SC_AA_SAMPLE_LOCS_PIXEL_X0_Y1_3 +0x00028C28 PA_SC_AA_SAMPLE_LOCS_PIXEL_X1_Y1_0 +0x00028C2C PA_SC_AA_SAMPLE_LOCS_PIXEL_X1_Y1_1 +0x00028C30 PA_SC_AA_SAMPLE_LOCS_PIXEL_X1_Y1_2 +0x00028C34 PA_SC_AA_SAMPLE_LOCS_PIXEL_X1_Y1_3 +0x00028C38 PA_SC_AA_MASK_X0_Y0_X1_Y0 +0x00028C3C PA_SC_AA_MASK_X0_Y1_X1_Y1 +0x00028C8C CB_COLOR0_CLEAR_WORD0 +0x00028C90 CB_COLOR0_CLEAR_WORD1 +0x00028C94 CB_COLOR0_CLEAR_WORD2 +0x00028C98 CB_COLOR0_CLEAR_WORD3 +0x00028CC8 CB_COLOR1_CLEAR_WORD0 +0x00028CCC CB_COLOR1_CLEAR_WORD1 +0x00028CD0 CB_COLOR1_CLEAR_WORD2 +0x00028CD4 CB_COLOR1_CLEAR_WORD3 +0x00028D04 CB_COLOR2_CLEAR_WORD0 +0x00028D08 CB_COLOR2_CLEAR_WORD1 +0x00028D0C CB_COLOR2_CLEAR_WORD2 +0x00028D10 CB_COLOR2_CLEAR_WORD3 +0x00028D40 CB_COLOR3_CLEAR_WORD0 +0x00028D44 CB_COLOR3_CLEAR_WORD1 +0x00028D48 CB_COLOR3_CLEAR_WORD2 +0x00028D4C CB_COLOR3_CLEAR_WORD3 +0x00028D7C CB_COLOR4_CLEAR_WORD0 +0x00028D80 CB_COLOR4_CLEAR_WORD1 +0x00028D84 CB_COLOR4_CLEAR_WORD2 +0x00028D88 CB_COLOR4_CLEAR_WORD3 +0x00028DB8 CB_COLOR5_CLEAR_WORD0 +0x00028DBC CB_COLOR5_CLEAR_WORD1 +0x00028DC0 CB_COLOR5_CLEAR_WORD2 +0x00028DC4 CB_COLOR5_CLEAR_WORD3 +0x00028DF4 CB_COLOR6_CLEAR_WORD0 +0x00028DF8 CB_COLOR6_CLEAR_WORD1 +0x00028DFC CB_COLOR6_CLEAR_WORD2 +0x00028E00 CB_COLOR6_CLEAR_WORD3 +0x00028E30 CB_COLOR7_CLEAR_WORD0 +0x00028E34 CB_COLOR7_CLEAR_WORD1 +0x00028E38 CB_COLOR7_CLEAR_WORD2 +0x00028E3C CB_COLOR7_CLEAR_WORD3 +0x00028F80 SQ_ALU_CONST_BUFFER_SIZE_HS_0 +0x00028F84 SQ_ALU_CONST_BUFFER_SIZE_HS_1 +0x00028F88 SQ_ALU_CONST_BUFFER_SIZE_HS_2 +0x00028F8C SQ_ALU_CONST_BUFFER_SIZE_HS_3 +0x00028F90 SQ_ALU_CONST_BUFFER_SIZE_HS_4 +0x00028F94 SQ_ALU_CONST_BUFFER_SIZE_HS_5 +0x00028F98 SQ_ALU_CONST_BUFFER_SIZE_HS_6 +0x00028F9C SQ_ALU_CONST_BUFFER_SIZE_HS_7 +0x00028FA0 SQ_ALU_CONST_BUFFER_SIZE_HS_8 +0x00028FA4 SQ_ALU_CONST_BUFFER_SIZE_HS_9 +0x00028FA8 SQ_ALU_CONST_BUFFER_SIZE_HS_10 +0x00028FAC SQ_ALU_CONST_BUFFER_SIZE_HS_11 +0x00028FB0 SQ_ALU_CONST_BUFFER_SIZE_HS_12 +0x00028FB4 SQ_ALU_CONST_BUFFER_SIZE_HS_13 +0x00028FB8 SQ_ALU_CONST_BUFFER_SIZE_HS_14 +0x00028FBC SQ_ALU_CONST_BUFFER_SIZE_HS_15 +0x00028FC0 SQ_ALU_CONST_BUFFER_SIZE_LS_0 +0x00028FC4 SQ_ALU_CONST_BUFFER_SIZE_LS_1 +0x00028FC8 SQ_ALU_CONST_BUFFER_SIZE_LS_2 +0x00028FCC SQ_ALU_CONST_BUFFER_SIZE_LS_3 +0x00028FD0 SQ_ALU_CONST_BUFFER_SIZE_LS_4 +0x00028FD4 SQ_ALU_CONST_BUFFER_SIZE_LS_5 +0x00028FD8 SQ_ALU_CONST_BUFFER_SIZE_LS_6 +0x00028FDC SQ_ALU_CONST_BUFFER_SIZE_LS_7 +0x00028FE0 SQ_ALU_CONST_BUFFER_SIZE_LS_8 +0x00028FE4 SQ_ALU_CONST_BUFFER_SIZE_LS_9 +0x00028FE8 SQ_ALU_CONST_BUFFER_SIZE_LS_10 +0x00028FEC SQ_ALU_CONST_BUFFER_SIZE_LS_11 +0x00028FF0 SQ_ALU_CONST_BUFFER_SIZE_LS_12 +0x00028FF4 SQ_ALU_CONST_BUFFER_SIZE_LS_13 +0x00028FF8 SQ_ALU_CONST_BUFFER_SIZE_LS_14 +0x00028FFC SQ_ALU_CONST_BUFFER_SIZE_LS_15 +0x0003CFF0 SQ_VTX_BASE_VTX_LOC +0x0003CFF4 SQ_VTX_START_INST_LOC +0x0003FF00 SQ_TEX_SAMPLER_CLEAR +0x0003FF04 SQ_TEX_RESOURCE_CLEAR +0x0003FF08 SQ_LOOP_BOOL_CLEAR diff --git a/drivers/gpu/drm/radeon/reg_srcs/evergreen b/drivers/gpu/drm/radeon/reg_srcs/evergreen new file mode 100644 index 00000000..0e28cae7 --- /dev/null +++ b/drivers/gpu/drm/radeon/reg_srcs/evergreen @@ -0,0 +1,622 @@ +evergreen 0x9400 +0x0000802C GRBM_GFX_INDEX +0x00008040 WAIT_UNTIL +0x00008044 WAIT_UNTIL_POLL_CNTL +0x00008048 WAIT_UNTIL_POLL_MASK +0x0000804c WAIT_UNTIL_POLL_REFDATA +0x000088B0 VGT_VTX_VECT_EJECT_REG +0x000088C4 VGT_CACHE_INVALIDATION +0x000088D4 VGT_GS_VERTEX_REUSE +0x00008958 VGT_PRIMITIVE_TYPE +0x0000895C VGT_INDEX_TYPE +0x00008970 VGT_NUM_INDICES +0x00008974 VGT_NUM_INSTANCES +0x00008990 VGT_COMPUTE_DIM_X +0x00008994 VGT_COMPUTE_DIM_Y +0x00008998 VGT_COMPUTE_DIM_Z +0x0000899C VGT_COMPUTE_START_X +0x000089A0 VGT_COMPUTE_START_Y +0x000089A4 VGT_COMPUTE_START_Z +0x000089AC VGT_COMPUTE_THREAD_GOURP_SIZE +0x00008A14 PA_CL_ENHANCE +0x00008A60 PA_SC_LINE_STIPPLE_VALUE +0x00008B10 PA_SC_LINE_STIPPLE_STATE +0x00008BF0 PA_SC_ENHANCE +0x00008D8C SQ_DYN_GPR_CNTL_PS_FLUSH_REQ +0x00008D90 SQ_DYN_GPR_OPTIMIZATION +0x00008D94 SQ_DYN_GPR_SIMD_LOCK_EN +0x00008D98 SQ_DYN_GPR_THREAD_LIMIT +0x00008D9C SQ_DYN_GPR_LDS_LIMIT +0x00008C00 SQ_CONFIG +0x00008C04 SQ_GPR_RESOURCE_MGMT_1 +0x00008C08 SQ_GPR_RESOURCE_MGMT_2 +0x00008C0C SQ_GPR_RESOURCE_MGMT_3 +0x00008C10 SQ_GLOBAL_GPR_RESOURCE_MGMT_1 +0x00008C14 SQ_GLOBAL_GPR_RESOURCE_MGMT_2 +0x00008C18 SQ_THREAD_RESOURCE_MGMT +0x00008C1C SQ_THREAD_RESOURCE_MGMT_2 +0x00008C20 SQ_STACK_RESOURCE_MGMT_1 +0x00008C24 SQ_STACK_RESOURCE_MGMT_2 +0x00008C28 SQ_STACK_RESOURCE_MGMT_3 +0x00008DF8 SQ_CONST_MEM_BASE +0x00008E20 SQ_STATIC_THREAD_MGMT_1 +0x00008E24 SQ_STATIC_THREAD_MGMT_2 +0x00008E28 SQ_STATIC_THREAD_MGMT_3 +0x00008E2C SQ_LDS_RESOURCE_MGMT +0x00008E48 SQ_EX_ALLOC_TABLE_SLOTS +0x00009100 SPI_CONFIG_CNTL +0x0000913C SPI_CONFIG_CNTL_1 +0x00009508 TA_CNTL_AUX +0x00009700 VC_CNTL +0x00009714 VC_ENHANCE +0x00009830 DB_DEBUG +0x00009834 DB_DEBUG2 +0x00009838 DB_DEBUG3 +0x0000983C DB_DEBUG4 +0x00009854 DB_WATERMARKS +0x0000A400 TD_PS_BORDER_COLOR_INDEX +0x0000A404 TD_PS_BORDER_COLOR_RED +0x0000A408 TD_PS_BORDER_COLOR_GREEN +0x0000A40C TD_PS_BORDER_COLOR_BLUE +0x0000A410 TD_PS_BORDER_COLOR_ALPHA +0x0000A414 TD_VS_BORDER_COLOR_INDEX +0x0000A418 TD_VS_BORDER_COLOR_RED +0x0000A41C TD_VS_BORDER_COLOR_GREEN +0x0000A420 TD_VS_BORDER_COLOR_BLUE +0x0000A424 TD_VS_BORDER_COLOR_ALPHA +0x0000A428 TD_GS_BORDER_COLOR_INDEX +0x0000A42C TD_GS_BORDER_COLOR_RED +0x0000A430 TD_GS_BORDER_COLOR_GREEN +0x0000A434 TD_GS_BORDER_COLOR_BLUE +0x0000A438 TD_GS_BORDER_COLOR_ALPHA +0x0000A43C TD_HS_BORDER_COLOR_INDEX +0x0000A440 TD_HS_BORDER_COLOR_RED +0x0000A444 TD_HS_BORDER_COLOR_GREEN +0x0000A448 TD_HS_BORDER_COLOR_BLUE +0x0000A44C TD_HS_BORDER_COLOR_ALPHA +0x0000A450 TD_LS_BORDER_COLOR_INDEX +0x0000A454 TD_LS_BORDER_COLOR_RED +0x0000A458 TD_LS_BORDER_COLOR_GREEN +0x0000A45C TD_LS_BORDER_COLOR_BLUE +0x0000A460 TD_LS_BORDER_COLOR_ALPHA +0x0000A464 TD_CS_BORDER_COLOR_INDEX +0x0000A468 TD_CS_BORDER_COLOR_RED +0x0000A46C TD_CS_BORDER_COLOR_GREEN +0x0000A470 TD_CS_BORDER_COLOR_BLUE +0x0000A474 TD_CS_BORDER_COLOR_ALPHA +0x00028000 DB_RENDER_CONTROL +0x00028004 DB_COUNT_CONTROL +0x0002800C DB_RENDER_OVERRIDE +0x00028010 DB_RENDER_OVERRIDE2 +0x00028028 DB_STENCIL_CLEAR +0x0002802C DB_DEPTH_CLEAR +0x00028030 PA_SC_SCREEN_SCISSOR_TL +0x00028034 PA_SC_SCREEN_SCISSOR_BR +0x0002805C DB_DEPTH_SLICE +0x00028140 SQ_ALU_CONST_BUFFER_SIZE_PS_0 +0x00028144 SQ_ALU_CONST_BUFFER_SIZE_PS_1 +0x00028148 SQ_ALU_CONST_BUFFER_SIZE_PS_2 +0x0002814C SQ_ALU_CONST_BUFFER_SIZE_PS_3 +0x00028150 SQ_ALU_CONST_BUFFER_SIZE_PS_4 +0x00028154 SQ_ALU_CONST_BUFFER_SIZE_PS_5 +0x00028158 SQ_ALU_CONST_BUFFER_SIZE_PS_6 +0x0002815C SQ_ALU_CONST_BUFFER_SIZE_PS_7 +0x00028160 SQ_ALU_CONST_BUFFER_SIZE_PS_8 +0x00028164 SQ_ALU_CONST_BUFFER_SIZE_PS_9 +0x00028168 SQ_ALU_CONST_BUFFER_SIZE_PS_10 +0x0002816C SQ_ALU_CONST_BUFFER_SIZE_PS_11 +0x00028170 SQ_ALU_CONST_BUFFER_SIZE_PS_12 +0x00028174 SQ_ALU_CONST_BUFFER_SIZE_PS_13 +0x00028178 SQ_ALU_CONST_BUFFER_SIZE_PS_14 +0x0002817C SQ_ALU_CONST_BUFFER_SIZE_PS_15 +0x00028180 SQ_ALU_CONST_BUFFER_SIZE_VS_0 +0x00028184 SQ_ALU_CONST_BUFFER_SIZE_VS_1 +0x00028188 SQ_ALU_CONST_BUFFER_SIZE_VS_2 +0x0002818C SQ_ALU_CONST_BUFFER_SIZE_VS_3 +0x00028190 SQ_ALU_CONST_BUFFER_SIZE_VS_4 +0x00028194 SQ_ALU_CONST_BUFFER_SIZE_VS_5 +0x00028198 SQ_ALU_CONST_BUFFER_SIZE_VS_6 +0x0002819C SQ_ALU_CONST_BUFFER_SIZE_VS_7 +0x000281A0 SQ_ALU_CONST_BUFFER_SIZE_VS_8 +0x000281A4 SQ_ALU_CONST_BUFFER_SIZE_VS_9 +0x000281A8 SQ_ALU_CONST_BUFFER_SIZE_VS_10 +0x000281AC SQ_ALU_CONST_BUFFER_SIZE_VS_11 +0x000281B0 SQ_ALU_CONST_BUFFER_SIZE_VS_12 +0x000281B4 SQ_ALU_CONST_BUFFER_SIZE_VS_13 +0x000281B8 SQ_ALU_CONST_BUFFER_SIZE_VS_14 +0x000281BC SQ_ALU_CONST_BUFFER_SIZE_VS_15 +0x000281C0 SQ_ALU_CONST_BUFFER_SIZE_GS_0 +0x000281C4 SQ_ALU_CONST_BUFFER_SIZE_GS_1 +0x000281C8 SQ_ALU_CONST_BUFFER_SIZE_GS_2 +0x000281CC SQ_ALU_CONST_BUFFER_SIZE_GS_3 +0x000281D0 SQ_ALU_CONST_BUFFER_SIZE_GS_4 +0x000281D4 SQ_ALU_CONST_BUFFER_SIZE_GS_5 +0x000281D8 SQ_ALU_CONST_BUFFER_SIZE_GS_6 +0x000281DC SQ_ALU_CONST_BUFFER_SIZE_GS_7 +0x000281E0 SQ_ALU_CONST_BUFFER_SIZE_GS_8 +0x000281E4 SQ_ALU_CONST_BUFFER_SIZE_GS_9 +0x000281E8 SQ_ALU_CONST_BUFFER_SIZE_GS_10 +0x000281EC SQ_ALU_CONST_BUFFER_SIZE_GS_11 +0x000281F0 SQ_ALU_CONST_BUFFER_SIZE_GS_12 +0x000281F4 SQ_ALU_CONST_BUFFER_SIZE_GS_13 +0x000281F8 SQ_ALU_CONST_BUFFER_SIZE_GS_14 +0x000281FC SQ_ALU_CONST_BUFFER_SIZE_GS_15 +0x00028200 PA_SC_WINDOW_OFFSET +0x00028204 PA_SC_WINDOW_SCISSOR_TL +0x00028208 PA_SC_WINDOW_SCISSOR_BR +0x0002820C PA_SC_CLIPRECT_RULE +0x00028210 PA_SC_CLIPRECT_0_TL +0x00028214 PA_SC_CLIPRECT_0_BR +0x00028218 PA_SC_CLIPRECT_1_TL +0x0002821C PA_SC_CLIPRECT_1_BR +0x00028220 PA_SC_CLIPRECT_2_TL +0x00028224 PA_SC_CLIPRECT_2_BR +0x00028228 PA_SC_CLIPRECT_3_TL +0x0002822C PA_SC_CLIPRECT_3_BR +0x00028230 PA_SC_EDGERULE +0x00028234 PA_SU_HARDWARE_SCREEN_OFFSET +0x00028240 PA_SC_GENERIC_SCISSOR_TL +0x00028244 PA_SC_GENERIC_SCISSOR_BR +0x00028250 PA_SC_VPORT_SCISSOR_0_TL +0x00028254 PA_SC_VPORT_SCISSOR_0_BR +0x00028258 PA_SC_VPORT_SCISSOR_1_TL +0x0002825C PA_SC_VPORT_SCISSOR_1_BR +0x00028260 PA_SC_VPORT_SCISSOR_2_TL +0x00028264 PA_SC_VPORT_SCISSOR_2_BR +0x00028268 PA_SC_VPORT_SCISSOR_3_TL +0x0002826C PA_SC_VPORT_SCISSOR_3_BR +0x00028270 PA_SC_VPORT_SCISSOR_4_TL +0x00028274 PA_SC_VPORT_SCISSOR_4_BR +0x00028278 PA_SC_VPORT_SCISSOR_5_TL +0x0002827C PA_SC_VPORT_SCISSOR_5_BR +0x00028280 PA_SC_VPORT_SCISSOR_6_TL +0x00028284 PA_SC_VPORT_SCISSOR_6_BR +0x00028288 PA_SC_VPORT_SCISSOR_7_TL +0x0002828C PA_SC_VPORT_SCISSOR_7_BR +0x00028290 PA_SC_VPORT_SCISSOR_8_TL +0x00028294 PA_SC_VPORT_SCISSOR_8_BR +0x00028298 PA_SC_VPORT_SCISSOR_9_TL +0x0002829C PA_SC_VPORT_SCISSOR_9_BR +0x000282A0 PA_SC_VPORT_SCISSOR_10_TL +0x000282A4 PA_SC_VPORT_SCISSOR_10_BR +0x000282A8 PA_SC_VPORT_SCISSOR_11_TL +0x000282AC PA_SC_VPORT_SCISSOR_11_BR +0x000282B0 PA_SC_VPORT_SCISSOR_12_TL +0x000282B4 PA_SC_VPORT_SCISSOR_12_BR +0x000282B8 PA_SC_VPORT_SCISSOR_13_TL +0x000282BC PA_SC_VPORT_SCISSOR_13_BR +0x000282C0 PA_SC_VPORT_SCISSOR_14_TL +0x000282C4 PA_SC_VPORT_SCISSOR_14_BR +0x000282C8 PA_SC_VPORT_SCISSOR_15_TL +0x000282CC PA_SC_VPORT_SCISSOR_15_BR +0x000282D0 PA_SC_VPORT_ZMIN_0 +0x000282D4 PA_SC_VPORT_ZMAX_0 +0x000282D8 PA_SC_VPORT_ZMIN_1 +0x000282DC PA_SC_VPORT_ZMAX_1 +0x000282E0 PA_SC_VPORT_ZMIN_2 +0x000282E4 PA_SC_VPORT_ZMAX_2 +0x000282E8 PA_SC_VPORT_ZMIN_3 +0x000282EC PA_SC_VPORT_ZMAX_3 +0x000282F0 PA_SC_VPORT_ZMIN_4 +0x000282F4 PA_SC_VPORT_ZMAX_4 +0x000282F8 PA_SC_VPORT_ZMIN_5 +0x000282FC PA_SC_VPORT_ZMAX_5 +0x00028300 PA_SC_VPORT_ZMIN_6 +0x00028304 PA_SC_VPORT_ZMAX_6 +0x00028308 PA_SC_VPORT_ZMIN_7 +0x0002830C PA_SC_VPORT_ZMAX_7 +0x00028310 PA_SC_VPORT_ZMIN_8 +0x00028314 PA_SC_VPORT_ZMAX_8 +0x00028318 PA_SC_VPORT_ZMIN_9 +0x0002831C PA_SC_VPORT_ZMAX_9 +0x00028320 PA_SC_VPORT_ZMIN_10 +0x00028324 PA_SC_VPORT_ZMAX_10 +0x00028328 PA_SC_VPORT_ZMIN_11 +0x0002832C PA_SC_VPORT_ZMAX_11 +0x00028330 PA_SC_VPORT_ZMIN_12 +0x00028334 PA_SC_VPORT_ZMAX_12 +0x00028338 PA_SC_VPORT_ZMIN_13 +0x0002833C PA_SC_VPORT_ZMAX_13 +0x00028340 PA_SC_VPORT_ZMIN_14 +0x00028344 PA_SC_VPORT_ZMAX_14 +0x00028348 PA_SC_VPORT_ZMIN_15 +0x0002834C PA_SC_VPORT_ZMAX_15 +0x00028350 SX_MISC +0x00028354 SX_SURFACE_SYNC +0x00028380 SQ_VTX_SEMANTIC_0 +0x00028384 SQ_VTX_SEMANTIC_1 +0x00028388 SQ_VTX_SEMANTIC_2 +0x0002838C SQ_VTX_SEMANTIC_3 +0x00028390 SQ_VTX_SEMANTIC_4 +0x00028394 SQ_VTX_SEMANTIC_5 +0x00028398 SQ_VTX_SEMANTIC_6 +0x0002839C SQ_VTX_SEMANTIC_7 +0x000283A0 SQ_VTX_SEMANTIC_8 +0x000283A4 SQ_VTX_SEMANTIC_9 +0x000283A8 SQ_VTX_SEMANTIC_10 +0x000283AC SQ_VTX_SEMANTIC_11 +0x000283B0 SQ_VTX_SEMANTIC_12 +0x000283B4 SQ_VTX_SEMANTIC_13 +0x000283B8 SQ_VTX_SEMANTIC_14 +0x000283BC SQ_VTX_SEMANTIC_15 +0x000283C0 SQ_VTX_SEMANTIC_16 +0x000283C4 SQ_VTX_SEMANTIC_17 +0x000283C8 SQ_VTX_SEMANTIC_18 +0x000283CC SQ_VTX_SEMANTIC_19 +0x000283D0 SQ_VTX_SEMANTIC_20 +0x000283D4 SQ_VTX_SEMANTIC_21 +0x000283D8 SQ_VTX_SEMANTIC_22 +0x000283DC SQ_VTX_SEMANTIC_23 +0x000283E0 SQ_VTX_SEMANTIC_24 +0x000283E4 SQ_VTX_SEMANTIC_25 +0x000283E8 SQ_VTX_SEMANTIC_26 +0x000283EC SQ_VTX_SEMANTIC_27 +0x000283F0 SQ_VTX_SEMANTIC_28 +0x000283F4 SQ_VTX_SEMANTIC_29 +0x000283F8 SQ_VTX_SEMANTIC_30 +0x000283FC SQ_VTX_SEMANTIC_31 +0x00028400 VGT_MAX_VTX_INDX +0x00028404 VGT_MIN_VTX_INDX +0x00028408 VGT_INDX_OFFSET +0x0002840C VGT_MULTI_PRIM_IB_RESET_INDX +0x00028410 SX_ALPHA_TEST_CONTROL +0x00028414 CB_BLEND_RED +0x00028418 CB_BLEND_GREEN +0x0002841C CB_BLEND_BLUE +0x00028420 CB_BLEND_ALPHA +0x00028430 DB_STENCILREFMASK +0x00028434 DB_STENCILREFMASK_BF +0x00028438 SX_ALPHA_REF +0x0002843C PA_CL_VPORT_XSCALE_0 +0x00028440 PA_CL_VPORT_XOFFSET_0 +0x00028444 PA_CL_VPORT_YSCALE_0 +0x00028448 PA_CL_VPORT_YOFFSET_0 +0x0002844C PA_CL_VPORT_ZSCALE_0 +0x00028450 PA_CL_VPORT_ZOFFSET_0 +0x00028454 PA_CL_VPORT_XSCALE_1 +0x00028458 PA_CL_VPORT_XOFFSET_1 +0x0002845C PA_CL_VPORT_YSCALE_1 +0x00028460 PA_CL_VPORT_YOFFSET_1 +0x00028464 PA_CL_VPORT_ZSCALE_1 +0x00028468 PA_CL_VPORT_ZOFFSET_1 +0x0002846C PA_CL_VPORT_XSCALE_2 +0x00028470 PA_CL_VPORT_XOFFSET_2 +0x00028474 PA_CL_VPORT_YSCALE_2 +0x00028478 PA_CL_VPORT_YOFFSET_2 +0x0002847C PA_CL_VPORT_ZSCALE_2 +0x00028480 PA_CL_VPORT_ZOFFSET_2 +0x00028484 PA_CL_VPORT_XSCALE_3 +0x00028488 PA_CL_VPORT_XOFFSET_3 +0x0002848C PA_CL_VPORT_YSCALE_3 +0x00028490 PA_CL_VPORT_YOFFSET_3 +0x00028494 PA_CL_VPORT_ZSCALE_3 +0x00028498 PA_CL_VPORT_ZOFFSET_3 +0x0002849C PA_CL_VPORT_XSCALE_4 +0x000284A0 PA_CL_VPORT_XOFFSET_4 +0x000284A4 PA_CL_VPORT_YSCALE_4 +0x000284A8 PA_CL_VPORT_YOFFSET_4 +0x000284AC PA_CL_VPORT_ZSCALE_4 +0x000284B0 PA_CL_VPORT_ZOFFSET_4 +0x000284B4 PA_CL_VPORT_XSCALE_5 +0x000284B8 PA_CL_VPORT_XOFFSET_5 +0x000284BC PA_CL_VPORT_YSCALE_5 +0x000284C0 PA_CL_VPORT_YOFFSET_5 +0x000284C4 PA_CL_VPORT_ZSCALE_5 +0x000284C8 PA_CL_VPORT_ZOFFSET_5 +0x000284CC PA_CL_VPORT_XSCALE_6 +0x000284D0 PA_CL_VPORT_XOFFSET_6 +0x000284D4 PA_CL_VPORT_YSCALE_6 +0x000284D8 PA_CL_VPORT_YOFFSET_6 +0x000284DC PA_CL_VPORT_ZSCALE_6 +0x000284E0 PA_CL_VPORT_ZOFFSET_6 +0x000284E4 PA_CL_VPORT_XSCALE_7 +0x000284E8 PA_CL_VPORT_XOFFSET_7 +0x000284EC PA_CL_VPORT_YSCALE_7 +0x000284F0 PA_CL_VPORT_YOFFSET_7 +0x000284F4 PA_CL_VPORT_ZSCALE_7 +0x000284F8 PA_CL_VPORT_ZOFFSET_7 +0x000284FC PA_CL_VPORT_XSCALE_8 +0x00028500 PA_CL_VPORT_XOFFSET_8 +0x00028504 PA_CL_VPORT_YSCALE_8 +0x00028508 PA_CL_VPORT_YOFFSET_8 +0x0002850C PA_CL_VPORT_ZSCALE_8 +0x00028510 PA_CL_VPORT_ZOFFSET_8 +0x00028514 PA_CL_VPORT_XSCALE_9 +0x00028518 PA_CL_VPORT_XOFFSET_9 +0x0002851C PA_CL_VPORT_YSCALE_9 +0x00028520 PA_CL_VPORT_YOFFSET_9 +0x00028524 PA_CL_VPORT_ZSCALE_9 +0x00028528 PA_CL_VPORT_ZOFFSET_9 +0x0002852C PA_CL_VPORT_XSCALE_10 +0x00028530 PA_CL_VPORT_XOFFSET_10 +0x00028534 PA_CL_VPORT_YSCALE_10 +0x00028538 PA_CL_VPORT_YOFFSET_10 +0x0002853C PA_CL_VPORT_ZSCALE_10 +0x00028540 PA_CL_VPORT_ZOFFSET_10 +0x00028544 PA_CL_VPORT_XSCALE_11 +0x00028548 PA_CL_VPORT_XOFFSET_11 +0x0002854C PA_CL_VPORT_YSCALE_11 +0x00028550 PA_CL_VPORT_YOFFSET_11 +0x00028554 PA_CL_VPORT_ZSCALE_11 +0x00028558 PA_CL_VPORT_ZOFFSET_11 +0x0002855C PA_CL_VPORT_XSCALE_12 +0x00028560 PA_CL_VPORT_XOFFSET_12 +0x00028564 PA_CL_VPORT_YSCALE_12 +0x00028568 PA_CL_VPORT_YOFFSET_12 +0x0002856C PA_CL_VPORT_ZSCALE_12 +0x00028570 PA_CL_VPORT_ZOFFSET_12 +0x00028574 PA_CL_VPORT_XSCALE_13 +0x00028578 PA_CL_VPORT_XOFFSET_13 +0x0002857C PA_CL_VPORT_YSCALE_13 +0x00028580 PA_CL_VPORT_YOFFSET_13 +0x00028584 PA_CL_VPORT_ZSCALE_13 +0x00028588 PA_CL_VPORT_ZOFFSET_13 +0x0002858C PA_CL_VPORT_XSCALE_14 +0x00028590 PA_CL_VPORT_XOFFSET_14 +0x00028594 PA_CL_VPORT_YSCALE_14 +0x00028598 PA_CL_VPORT_YOFFSET_14 +0x0002859C PA_CL_VPORT_ZSCALE_14 +0x000285A0 PA_CL_VPORT_ZOFFSET_14 +0x000285A4 PA_CL_VPORT_XSCALE_15 +0x000285A8 PA_CL_VPORT_XOFFSET_15 +0x000285AC PA_CL_VPORT_YSCALE_15 +0x000285B0 PA_CL_VPORT_YOFFSET_15 +0x000285B4 PA_CL_VPORT_ZSCALE_15 +0x000285B8 PA_CL_VPORT_ZOFFSET_15 +0x000285BC PA_CL_UCP_0_X +0x000285C0 PA_CL_UCP_0_Y +0x000285C4 PA_CL_UCP_0_Z +0x000285C8 PA_CL_UCP_0_W +0x000285CC PA_CL_UCP_1_X +0x000285D0 PA_CL_UCP_1_Y +0x000285D4 PA_CL_UCP_1_Z +0x000285D8 PA_CL_UCP_1_W +0x000285DC PA_CL_UCP_2_X +0x000285E0 PA_CL_UCP_2_Y +0x000285E4 PA_CL_UCP_2_Z +0x000285E8 PA_CL_UCP_2_W +0x000285EC PA_CL_UCP_3_X +0x000285F0 PA_CL_UCP_3_Y +0x000285F4 PA_CL_UCP_3_Z +0x000285F8 PA_CL_UCP_3_W +0x000285FC PA_CL_UCP_4_X +0x00028600 PA_CL_UCP_4_Y +0x00028604 PA_CL_UCP_4_Z +0x00028608 PA_CL_UCP_4_W +0x0002860C PA_CL_UCP_5_X +0x00028610 PA_CL_UCP_5_Y +0x00028614 PA_CL_UCP_5_Z +0x00028618 PA_CL_UCP_5_W +0x0002861C SPI_VS_OUT_ID_0 +0x00028620 SPI_VS_OUT_ID_1 +0x00028624 SPI_VS_OUT_ID_2 +0x00028628 SPI_VS_OUT_ID_3 +0x0002862C SPI_VS_OUT_ID_4 +0x00028630 SPI_VS_OUT_ID_5 +0x00028634 SPI_VS_OUT_ID_6 +0x00028638 SPI_VS_OUT_ID_7 +0x0002863C SPI_VS_OUT_ID_8 +0x00028640 SPI_VS_OUT_ID_9 +0x00028644 SPI_PS_INPUT_CNTL_0 +0x00028648 SPI_PS_INPUT_CNTL_1 +0x0002864C SPI_PS_INPUT_CNTL_2 +0x00028650 SPI_PS_INPUT_CNTL_3 +0x00028654 SPI_PS_INPUT_CNTL_4 +0x00028658 SPI_PS_INPUT_CNTL_5 +0x0002865C SPI_PS_INPUT_CNTL_6 +0x00028660 SPI_PS_INPUT_CNTL_7 +0x00028664 SPI_PS_INPUT_CNTL_8 +0x00028668 SPI_PS_INPUT_CNTL_9 +0x0002866C SPI_PS_INPUT_CNTL_10 +0x00028670 SPI_PS_INPUT_CNTL_11 +0x00028674 SPI_PS_INPUT_CNTL_12 +0x00028678 SPI_PS_INPUT_CNTL_13 +0x0002867C SPI_PS_INPUT_CNTL_14 +0x00028680 SPI_PS_INPUT_CNTL_15 +0x00028684 SPI_PS_INPUT_CNTL_16 +0x00028688 SPI_PS_INPUT_CNTL_17 +0x0002868C SPI_PS_INPUT_CNTL_18 +0x00028690 SPI_PS_INPUT_CNTL_19 +0x00028694 SPI_PS_INPUT_CNTL_20 +0x00028698 SPI_PS_INPUT_CNTL_21 +0x0002869C SPI_PS_INPUT_CNTL_22 +0x000286A0 SPI_PS_INPUT_CNTL_23 +0x000286A4 SPI_PS_INPUT_CNTL_24 +0x000286A8 SPI_PS_INPUT_CNTL_25 +0x000286AC SPI_PS_INPUT_CNTL_26 +0x000286B0 SPI_PS_INPUT_CNTL_27 +0x000286B4 SPI_PS_INPUT_CNTL_28 +0x000286B8 SPI_PS_INPUT_CNTL_29 +0x000286BC SPI_PS_INPUT_CNTL_30 +0x000286C0 SPI_PS_INPUT_CNTL_31 +0x000286C4 SPI_VS_OUT_CONFIG +0x000286C8 SPI_THREAD_GROUPING +0x000286CC SPI_PS_IN_CONTROL_0 +0x000286D0 SPI_PS_IN_CONTROL_1 +0x000286D4 SPI_INTERP_CONTROL_0 +0x000286D8 SPI_INPUT_Z +0x000286DC SPI_FOG_CNTL +0x000286E0 SPI_BARYC_CNTL +0x000286E4 SPI_PS_IN_CONTROL_2 +0x000286E8 SPI_COMPUTE_INPUT_CNTL +0x000286EC SPI_COMPUTE_NUM_THREAD_X +0x000286F0 SPI_COMPUTE_NUM_THREAD_Y +0x000286F4 SPI_COMPUTE_NUM_THREAD_Z +0x00028724 GDS_ADDR_SIZE +0x00028780 CB_BLEND0_CONTROL +0x00028784 CB_BLEND1_CONTROL +0x00028788 CB_BLEND2_CONTROL +0x0002878C CB_BLEND3_CONTROL +0x00028790 CB_BLEND4_CONTROL +0x00028794 CB_BLEND5_CONTROL +0x00028798 CB_BLEND6_CONTROL +0x0002879C CB_BLEND7_CONTROL +0x000287CC CS_COPY_STATE +0x000287D0 GFX_COPY_STATE +0x000287D4 PA_CL_POINT_X_RAD +0x000287D8 PA_CL_POINT_Y_RAD +0x000287DC PA_CL_POINT_SIZE +0x000287E0 PA_CL_POINT_CULL_RAD +0x00028808 CB_COLOR_CONTROL +0x0002880C DB_SHADER_CONTROL +0x00028810 PA_CL_CLIP_CNTL +0x00028814 PA_SU_SC_MODE_CNTL +0x00028818 PA_CL_VTE_CNTL +0x0002881C PA_CL_VS_OUT_CNTL +0x00028820 PA_CL_NANINF_CNTL +0x00028824 PA_SU_LINE_STIPPLE_CNTL +0x00028828 PA_SU_LINE_STIPPLE_SCALE +0x0002882C PA_SU_PRIM_FILTER_CNTL +0x00028838 SQ_DYN_GPR_RESOURCE_LIMIT_1 +0x00028844 SQ_PGM_RESOURCES_PS +0x00028848 SQ_PGM_RESOURCES_2_PS +0x0002884C SQ_PGM_EXPORTS_PS +0x00028860 SQ_PGM_RESOURCES_VS +0x00028864 SQ_PGM_RESOURCES_2_VS +0x00028878 SQ_PGM_RESOURCES_GS +0x0002887C SQ_PGM_RESOURCES_2_GS +0x00028890 SQ_PGM_RESOURCES_ES +0x00028894 SQ_PGM_RESOURCES_2_ES +0x000288A8 SQ_PGM_RESOURCES_FS +0x000288BC SQ_PGM_RESOURCES_HS +0x000288C0 SQ_PGM_RESOURCES_2_HS +0x000288D4 SQ_PGM_RESOURCES_LS +0x000288D8 SQ_PGM_RESOURCES_2_LS +0x000288E8 SQ_LDS_ALLOC +0x000288EC SQ_LDS_ALLOC_PS +0x000288F0 SQ_VTX_SEMANTIC_CLEAR +0x00028A00 PA_SU_POINT_SIZE +0x00028A04 PA_SU_POINT_MINMAX +0x00028A08 PA_SU_LINE_CNTL +0x00028A0C PA_SC_LINE_STIPPLE +0x00028A10 VGT_OUTPUT_PATH_CNTL +0x00028A14 VGT_HOS_CNTL +0x00028A18 VGT_HOS_MAX_TESS_LEVEL +0x00028A1C VGT_HOS_MIN_TESS_LEVEL +0x00028A20 VGT_HOS_REUSE_DEPTH +0x00028A24 VGT_GROUP_PRIM_TYPE +0x00028A28 VGT_GROUP_FIRST_DECR +0x00028A2C VGT_GROUP_DECR +0x00028A30 VGT_GROUP_VECT_0_CNTL +0x00028A34 VGT_GROUP_VECT_1_CNTL +0x00028A38 VGT_GROUP_VECT_0_FMT_CNTL +0x00028A3C VGT_GROUP_VECT_1_FMT_CNTL +0x00028A40 VGT_GS_MODE +0x00028A48 PA_SC_MODE_CNTL_0 +0x00028A4C PA_SC_MODE_CNTL_1 +0x00028A50 VGT_ENHANCE +0x00028A54 VGT_GS_PER_ES +0x00028A58 VGT_ES_PER_GS +0x00028A5C VGT_GS_PER_VS +0x00028A6C VGT_GS_OUT_PRIM_TYPE +0x00028A84 VGT_PRIMITIVEID_EN +0x00028A94 VGT_MULTI_PRIM_IB_RESET_EN +0x00028AA0 VGT_INSTANCE_STEP_RATE_0 +0x00028AA4 VGT_INSTANCE_STEP_RATE_1 +0x00028AB4 VGT_REUSE_OFF +0x00028AB8 VGT_VTX_CNT_EN +0x00028ABC DB_HTILE_SURFACE +0x00028AC0 DB_SRESULTS_COMPARE_STATE0 +0x00028AC4 DB_SRESULTS_COMPARE_STATE1 +0x00028AC8 DB_PRELOAD_CONTROL +0x00028B38 VGT_GS_MAX_VERT_OUT +0x00028B54 VGT_SHADER_STAGES_EN +0x00028B58 VGT_LS_HS_CONFIG +0x00028B5C VGT_LS_SIZE +0x00028B60 VGT_HS_SIZE +0x00028B64 VGT_LS_HS_ALLOC +0x00028B68 VGT_HS_PATCH_CONST +0x00028B6C VGT_TF_PARAM +0x00028B70 DB_ALPHA_TO_MASK +0x00028B74 VGT_DISPATCH_INITIATOR +0x00028B78 PA_SU_POLY_OFFSET_DB_FMT_CNTL +0x00028B7C PA_SU_POLY_OFFSET_CLAMP +0x00028B80 PA_SU_POLY_OFFSET_FRONT_SCALE +0x00028B84 PA_SU_POLY_OFFSET_FRONT_OFFSET +0x00028B88 PA_SU_POLY_OFFSET_BACK_SCALE +0x00028B8C PA_SU_POLY_OFFSET_BACK_OFFSET +0x00028B74 VGT_GS_INSTANCE_CNT +0x00028C00 PA_SC_LINE_CNTL +0x00028C08 PA_SU_VTX_CNTL +0x00028C0C PA_CL_GB_VERT_CLIP_ADJ +0x00028C10 PA_CL_GB_VERT_DISC_ADJ +0x00028C14 PA_CL_GB_HORZ_CLIP_ADJ +0x00028C18 PA_CL_GB_HORZ_DISC_ADJ +0x00028C1C PA_SC_AA_SAMPLE_LOCS_0 +0x00028C20 PA_SC_AA_SAMPLE_LOCS_1 +0x00028C24 PA_SC_AA_SAMPLE_LOCS_2 +0x00028C28 PA_SC_AA_SAMPLE_LOCS_3 +0x00028C2C PA_SC_AA_SAMPLE_LOCS_4 +0x00028C30 PA_SC_AA_SAMPLE_LOCS_5 +0x00028C34 PA_SC_AA_SAMPLE_LOCS_6 +0x00028C38 PA_SC_AA_SAMPLE_LOCS_7 +0x00028C3C PA_SC_AA_MASK +0x00028C8C CB_COLOR0_CLEAR_WORD0 +0x00028C90 CB_COLOR0_CLEAR_WORD1 +0x00028C94 CB_COLOR0_CLEAR_WORD2 +0x00028C98 CB_COLOR0_CLEAR_WORD3 +0x00028CC8 CB_COLOR1_CLEAR_WORD0 +0x00028CCC CB_COLOR1_CLEAR_WORD1 +0x00028CD0 CB_COLOR1_CLEAR_WORD2 +0x00028CD4 CB_COLOR1_CLEAR_WORD3 +0x00028D04 CB_COLOR2_CLEAR_WORD0 +0x00028D08 CB_COLOR2_CLEAR_WORD1 +0x00028D0C CB_COLOR2_CLEAR_WORD2 +0x00028D10 CB_COLOR2_CLEAR_WORD3 +0x00028D40 CB_COLOR3_CLEAR_WORD0 +0x00028D44 CB_COLOR3_CLEAR_WORD1 +0x00028D48 CB_COLOR3_CLEAR_WORD2 +0x00028D4C CB_COLOR3_CLEAR_WORD3 +0x00028D7C CB_COLOR4_CLEAR_WORD0 +0x00028D80 CB_COLOR4_CLEAR_WORD1 +0x00028D84 CB_COLOR4_CLEAR_WORD2 +0x00028D88 CB_COLOR4_CLEAR_WORD3 +0x00028DB8 CB_COLOR5_CLEAR_WORD0 +0x00028DBC CB_COLOR5_CLEAR_WORD1 +0x00028DC0 CB_COLOR5_CLEAR_WORD2 +0x00028DC4 CB_COLOR5_CLEAR_WORD3 +0x00028DF4 CB_COLOR6_CLEAR_WORD0 +0x00028DF8 CB_COLOR6_CLEAR_WORD1 +0x00028DFC CB_COLOR6_CLEAR_WORD2 +0x00028E00 CB_COLOR6_CLEAR_WORD3 +0x00028E30 CB_COLOR7_CLEAR_WORD0 +0x00028E34 CB_COLOR7_CLEAR_WORD1 +0x00028E38 CB_COLOR7_CLEAR_WORD2 +0x00028E3C CB_COLOR7_CLEAR_WORD3 +0x00028F80 SQ_ALU_CONST_BUFFER_SIZE_HS_0 +0x00028F84 SQ_ALU_CONST_BUFFER_SIZE_HS_1 +0x00028F88 SQ_ALU_CONST_BUFFER_SIZE_HS_2 +0x00028F8C SQ_ALU_CONST_BUFFER_SIZE_HS_3 +0x00028F90 SQ_ALU_CONST_BUFFER_SIZE_HS_4 +0x00028F94 SQ_ALU_CONST_BUFFER_SIZE_HS_5 +0x00028F98 SQ_ALU_CONST_BUFFER_SIZE_HS_6 +0x00028F9C SQ_ALU_CONST_BUFFER_SIZE_HS_7 +0x00028FA0 SQ_ALU_CONST_BUFFER_SIZE_HS_8 +0x00028FA4 SQ_ALU_CONST_BUFFER_SIZE_HS_9 +0x00028FA8 SQ_ALU_CONST_BUFFER_SIZE_HS_10 +0x00028FAC SQ_ALU_CONST_BUFFER_SIZE_HS_11 +0x00028FB0 SQ_ALU_CONST_BUFFER_SIZE_HS_12 +0x00028FB4 SQ_ALU_CONST_BUFFER_SIZE_HS_13 +0x00028FB8 SQ_ALU_CONST_BUFFER_SIZE_HS_14 +0x00028FBC SQ_ALU_CONST_BUFFER_SIZE_HS_15 +0x00028FC0 SQ_ALU_CONST_BUFFER_SIZE_LS_0 +0x00028FC4 SQ_ALU_CONST_BUFFER_SIZE_LS_1 +0x00028FC8 SQ_ALU_CONST_BUFFER_SIZE_LS_2 +0x00028FCC SQ_ALU_CONST_BUFFER_SIZE_LS_3 +0x00028FD0 SQ_ALU_CONST_BUFFER_SIZE_LS_4 +0x00028FD4 SQ_ALU_CONST_BUFFER_SIZE_LS_5 +0x00028FD8 SQ_ALU_CONST_BUFFER_SIZE_LS_6 +0x00028FDC SQ_ALU_CONST_BUFFER_SIZE_LS_7 +0x00028FE0 SQ_ALU_CONST_BUFFER_SIZE_LS_8 +0x00028FE4 SQ_ALU_CONST_BUFFER_SIZE_LS_9 +0x00028FE8 SQ_ALU_CONST_BUFFER_SIZE_LS_10 +0x00028FEC SQ_ALU_CONST_BUFFER_SIZE_LS_11 +0x00028FF0 SQ_ALU_CONST_BUFFER_SIZE_LS_12 +0x00028FF4 SQ_ALU_CONST_BUFFER_SIZE_LS_13 +0x00028FF8 SQ_ALU_CONST_BUFFER_SIZE_LS_14 +0x00028FFC SQ_ALU_CONST_BUFFER_SIZE_LS_15 +0x0003CFF0 SQ_VTX_BASE_VTX_LOC +0x0003CFF4 SQ_VTX_START_INST_LOC +0x0003FF00 SQ_TEX_SAMPLER_CLEAR +0x0003FF04 SQ_TEX_RESOURCE_CLEAR +0x0003FF08 SQ_LOOP_BOOL_CLEAR diff --git a/drivers/gpu/drm/radeon/reg_srcs/r100 b/drivers/gpu/drm/radeon/reg_srcs/r100 new file mode 100644 index 00000000..f7ee062f --- /dev/null +++ b/drivers/gpu/drm/radeon/reg_srcs/r100 @@ -0,0 +1,105 @@ +r100 0x3294 +0x1434 SRC_Y_X +0x1438 DST_Y_X +0x143C DST_HEIGHT_WIDTH +0x146C DP_GUI_MASTER_CNTL +0x1474 BRUSH_Y_X +0x1478 DP_BRUSH_BKGD_CLR +0x147C DP_BRUSH_FRGD_CLR +0x1480 BRUSH_DATA0 +0x1484 BRUSH_DATA1 +0x1598 DST_WIDTH_HEIGHT +0x15C0 CLR_CMP_CNTL +0x15C4 CLR_CMP_CLR_SRC +0x15C8 CLR_CMP_CLR_DST +0x15CC CLR_CMP_MSK +0x15D8 DP_SRC_FRGD_CLR +0x15DC DP_SRC_BKGD_CLR +0x1600 DST_LINE_START +0x1604 DST_LINE_END +0x1608 DST_LINE_PATCOUNT +0x16C0 DP_CNTL +0x16CC DP_WRITE_MSK +0x16D0 DP_CNTL_XDIR_YDIR_YMAJOR +0x16E8 DEFAULT_SC_BOTTOM_RIGHT +0x16EC SC_TOP_LEFT +0x16F0 SC_BOTTOM_RIGHT +0x16F4 SRC_SC_BOTTOM_RIGHT +0x1714 DSTCACHE_CTLSTAT +0x1720 WAIT_UNTIL +0x172C RBBM_GUICNTL +0x1810 FOG_3D_TABLE_START +0x1814 FOG_3D_TABLE_END +0x1a14 FOG_TABLE_INDEX +0x1a18 FOG_TABLE_DATA +0x1c14 PP_MISC +0x1c18 PP_FOG_COLOR +0x1c1c RE_SOLID_COLOR +0x1c20 RB3D_BLENDCNTL +0x1c4c SE_CNTL +0x1c50 SE_COORD_FMT +0x1c60 PP_TXCBLEND_0 +0x1c64 PP_TXABLEND_0 +0x1c68 PP_TFACTOR_0 +0x1c78 PP_TXCBLEND_1 +0x1c7c PP_TXABLEND_1 +0x1c80 PP_TFACTOR_1 +0x1c90 PP_TXCBLEND_2 +0x1c94 PP_TXABLEND_2 +0x1c98 PP_TFACTOR_2 +0x1cc8 RE_STIPPLE_ADDR +0x1ccc RE_STIPPLE_DATA +0x1cd0 RE_LINE_PATTERN +0x1cd4 RE_LINE_STATE +0x1d40 PP_BORDER_COLOR0 +0x1d44 PP_BORDER_COLOR1 +0x1d48 PP_BORDER_COLOR2 +0x1d7c RB3D_STENCILREFMASK +0x1d80 RB3D_ROPCNTL +0x1d84 RB3D_PLANEMASK +0x1d98 VAP_VPORT_XSCALE +0x1d9C VAP_VPORT_XOFFSET +0x1da0 VAP_VPORT_YSCALE +0x1da4 VAP_VPORT_YOFFSET +0x1da8 VAP_VPORT_ZSCALE +0x1dac VAP_VPORT_ZOFFSET +0x1db0 SE_ZBIAS_FACTOR +0x1db4 SE_ZBIAS_CONSTANT +0x1db8 SE_LINE_WIDTH +0x2140 SE_CNTL_STATUS +0x2200 SE_TCL_VECTOR_INDX_REG +0x2204 SE_TCL_VECTOR_DATA_REG +0x2208 SE_TCL_SCALAR_INDX_REG +0x220c SE_TCL_SCALAR_DATA_REG +0x2210 SE_TCL_MATERIAL_EMISSIVE_RED +0x2214 SE_TCL_MATERIAL_EMISSIVE_GREEN +0x2218 SE_TCL_MATERIAL_EMISSIVE_BLUE +0x221c SE_TCL_MATERIAL_EMISSIVE_ALPHA +0x2220 SE_TCL_MATERIAL_AMBIENT_RED +0x2224 SE_TCL_MATERIAL_AMBIENT_GREEN +0x2228 SE_TCL_MATERIAL_AMBIENT_BLUE +0x222c SE_TCL_MATERIAL_AMBIENT_ALPHA +0x2230 SE_TCL_MATERIAL_DIFFUSE_RED +0x2234 SE_TCL_MATERIAL_DIFFUSE_GREEN +0x2238 SE_TCL_MATERIAL_DIFFUSE_BLUE +0x223c SE_TCL_MATERIAL_DIFFUSE_ALPHA +0x2240 SE_TCL_MATERIAL_SPECULAR_RED +0x2244 SE_TCL_MATERIAL_SPECULAR_GREEN +0x2248 SE_TCL_MATERIAL_SPECULAR_BLUE +0x224c SE_TCL_MATERIAL_SPECULAR_ALPHA +0x2250 SE_TCL_SHININESS +0x2254 SE_TCL_OUTPUT_VTX_FMT +0x2258 SE_TCL_OUTPUT_VTX_SEL +0x225c SE_TCL_MATRIX_SELECT_0 +0x2260 SE_TCL_MATRIX_SELECT_1 +0x2264 SE_TCL_UCP_VERT_BLEND_CNTL +0x2268 SE_TCL_TEXTURE_PROC_CTL +0x226c SE_TCL_LIGHT_MODEL_CTL +0x2270 SE_TCL_PER_LIGHT_CTL_0 +0x2274 SE_TCL_PER_LIGHT_CTL_1 +0x2278 SE_TCL_PER_LIGHT_CTL_2 +0x227c SE_TCL_PER_LIGHT_CTL_3 +0x2284 SE_TCL_STATE_FLUSH +0x26c0 RE_TOP_LEFT +0x26c4 RE_MISC +0x3290 RB3D_ZPASS_DATA diff --git a/drivers/gpu/drm/radeon/reg_srcs/r200 b/drivers/gpu/drm/radeon/reg_srcs/r200 new file mode 100644 index 00000000..c29ac434 --- /dev/null +++ b/drivers/gpu/drm/radeon/reg_srcs/r200 @@ -0,0 +1,186 @@ +r200 0x3294 +0x1434 SRC_Y_X +0x1438 DST_Y_X +0x143C DST_HEIGHT_WIDTH +0x146C DP_GUI_MASTER_CNTL +0x1474 BRUSH_Y_X +0x1478 DP_BRUSH_BKGD_CLR +0x147C DP_BRUSH_FRGD_CLR +0x1480 BRUSH_DATA0 +0x1484 BRUSH_DATA1 +0x1598 DST_WIDTH_HEIGHT +0x15C0 CLR_CMP_CNTL +0x15C4 CLR_CMP_CLR_SRC +0x15C8 CLR_CMP_CLR_DST +0x15CC CLR_CMP_MSK +0x15D8 DP_SRC_FRGD_CLR +0x15DC DP_SRC_BKGD_CLR +0x1600 DST_LINE_START +0x1604 DST_LINE_END +0x1608 DST_LINE_PATCOUNT +0x16C0 DP_CNTL +0x16CC DP_WRITE_MSK +0x16D0 DP_CNTL_XDIR_YDIR_YMAJOR +0x16E8 DEFAULT_SC_BOTTOM_RIGHT +0x16EC SC_TOP_LEFT +0x16F0 SC_BOTTOM_RIGHT +0x16F4 SRC_SC_BOTTOM_RIGHT +0x1714 DSTCACHE_CTLSTAT +0x1720 WAIT_UNTIL +0x172C RBBM_GUICNTL +0x1c14 PP_MISC +0x1c18 PP_FOG_COLOR +0x1c1c RE_SOLID_COLOR +0x1c20 RB3D_BLENDCNTL +0x1c4c SE_CNTL +0x1c50 RE_CNTL +0x1cc8 RE_STIPPLE_ADDR +0x1ccc RE_STIPPLE_DATA +0x1cd0 RE_LINE_PATTERN +0x1cd4 RE_LINE_STATE +0x1cd8 RE_SCISSOR_TL_0 +0x1cdc RE_SCISSOR_BR_0 +0x1ce0 RE_SCISSOR_TL_1 +0x1ce4 RE_SCISSOR_BR_1 +0x1ce8 RE_SCISSOR_TL_2 +0x1cec RE_SCISSOR_BR_2 +0x1d60 RB3D_DEPTHXY_OFFSET +0x1d7c RB3D_STENCILREFMASK +0x1d80 RB3D_ROPCNTL +0x1d84 RB3D_PLANEMASK +0x1d98 VAP_VPORT_XSCALE +0x1d9c VAP_VPORT_XOFFSET +0x1da0 VAP_VPORT_YSCALE +0x1da4 VAP_VPORT_YOFFSET +0x1da8 VAP_VPORT_ZSCALE +0x1dac VAP_VPORT_ZOFFSET +0x1db0 SE_ZBIAS_FACTOR +0x1db4 SE_ZBIAS_CONSTANT +0x1db8 SE_LINE_WIDTH +0x2080 SE_VAP_CNTL +0x2090 SE_TCL_OUTPUT_VTX_FMT_0 +0x2094 SE_TCL_OUTPUT_VTX_FMT_1 +0x20b0 SE_VTE_CNTL +0x2140 SE_CNTL_STATUS +0x2180 SE_VTX_STATE_CNTL +0x2200 SE_TCL_VECTOR_INDX_REG +0x2204 SE_TCL_VECTOR_DATA_REG +0x2208 SE_TCL_SCALAR_INDX_REG +0x220c SE_TCL_SCALAR_DATA_REG +0x2230 SE_TCL_MATRIX_SEL_0 +0x2234 SE_TCL_MATRIX_SEL_1 +0x2238 SE_TCL_MATRIX_SEL_2 +0x223c SE_TCL_MATRIX_SEL_3 +0x2240 SE_TCL_MATRIX_SEL_4 +0x2250 SE_TCL_OUTPUT_VTX_COMP_SEL +0x2254 SE_TCL_INPUT_VTX_VECTOR_ADDR_0 +0x2258 SE_TCL_INPUT_VTX_VECTOR_ADDR_1 +0x225c SE_TCL_INPUT_VTX_VECTOR_ADDR_2 +0x2260 SE_TCL_INPUT_VTX_VECTOR_ADDR_3 +0x2268 SE_TCL_LIGHT_MODEL_CTL_0 +0x226c SE_TCL_LIGHT_MODEL_CTL_1 +0x2270 SE_TCL_PER_LIGHT_CTL_0 +0x2274 SE_TCL_PER_LIGHT_CTL_1 +0x2278 SE_TCL_PER_LIGHT_CTL_2 +0x227c SE_TCL_PER_LIGHT_CTL_3 +0x2284 VAP_PVS_STATE_FLUSH_REG +0x22a8 SE_TCL_TEX_PROC_CTL_2 +0x22ac SE_TCL_TEX_PROC_CTL_3 +0x22b0 SE_TCL_TEX_PROC_CTL_0 +0x22b4 SE_TCL_TEX_PROC_CTL_1 +0x22b8 SE_TCL_TEX_CYL_WRAP_CTL +0x22c0 SE_TCL_UCP_VERT_BLEND_CNTL +0x22c4 SE_TCL_POINT_SPRITE_CNTL +0x22d0 SE_PVS_CNTL +0x22d4 SE_PVS_CONST_CNTL +0x2648 RE_POINTSIZE +0x26c0 RE_TOP_LEFT +0x26c4 RE_MISC +0x26f0 RE_AUX_SCISSOR_CNTL +0x2c14 PP_BORDER_COLOR_0 +0x2c34 PP_BORDER_COLOR_1 +0x2c54 PP_BORDER_COLOR_2 +0x2c74 PP_BORDER_COLOR_3 +0x2c94 PP_BORDER_COLOR_4 +0x2cb4 PP_BORDER_COLOR_5 +0x2cc4 PP_CNTL_X +0x2cf8 PP_TRI_PERF +0x2cfc PP_PERF_CNTL +0x2d9c PP_TAM_DEBUG3 +0x2ee0 PP_TFACTOR_0 +0x2ee4 PP_TFACTOR_1 +0x2ee8 PP_TFACTOR_2 +0x2eec PP_TFACTOR_3 +0x2ef0 PP_TFACTOR_4 +0x2ef4 PP_TFACTOR_5 +0x2ef8 PP_TFACTOR_6 +0x2efc PP_TFACTOR_7 +0x2f00 PP_TXCBLEND_0 +0x2f04 PP_TXCBLEND2_0 +0x2f08 PP_TXABLEND_0 +0x2f0c PP_TXABLEND2_0 +0x2f10 PP_TXCBLEND_1 +0x2f14 PP_TXCBLEND2_1 +0x2f18 PP_TXABLEND_1 +0x2f1c PP_TXABLEND2_1 +0x2f20 PP_TXCBLEND_2 +0x2f24 PP_TXCBLEND2_2 +0x2f28 PP_TXABLEND_2 +0x2f2c PP_TXABLEND2_2 +0x2f30 PP_TXCBLEND_3 +0x2f34 PP_TXCBLEND2_3 +0x2f38 PP_TXABLEND_3 +0x2f3c PP_TXABLEND2_3 +0x2f40 PP_TXCBLEND_4 +0x2f44 PP_TXCBLEND2_4 +0x2f48 PP_TXABLEND_4 +0x2f4c PP_TXABLEND2_4 +0x2f50 PP_TXCBLEND_5 +0x2f54 PP_TXCBLEND2_5 +0x2f58 PP_TXABLEND_5 +0x2f5c PP_TXABLEND2_5 +0x2f60 PP_TXCBLEND_6 +0x2f64 PP_TXCBLEND2_6 +0x2f68 PP_TXABLEND_6 +0x2f6c PP_TXABLEND2_6 +0x2f70 PP_TXCBLEND_7 +0x2f74 PP_TXCBLEND2_7 +0x2f78 PP_TXABLEND_7 +0x2f7c PP_TXABLEND2_7 +0x2f80 PP_TXCBLEND_8 +0x2f84 PP_TXCBLEND2_8 +0x2f88 PP_TXABLEND_8 +0x2f8c PP_TXABLEND2_8 +0x2f90 PP_TXCBLEND_9 +0x2f94 PP_TXCBLEND2_9 +0x2f98 PP_TXABLEND_9 +0x2f9c PP_TXABLEND2_9 +0x2fa0 PP_TXCBLEND_10 +0x2fa4 PP_TXCBLEND2_10 +0x2fa8 PP_TXABLEND_10 +0x2fac PP_TXABLEND2_10 +0x2fb0 PP_TXCBLEND_11 +0x2fb4 PP_TXCBLEND2_11 +0x2fb8 PP_TXABLEND_11 +0x2fbc PP_TXABLEND2_11 +0x2fc0 PP_TXCBLEND_12 +0x2fc4 PP_TXCBLEND2_12 +0x2fc8 PP_TXABLEND_12 +0x2fcc PP_TXABLEND2_12 +0x2fd0 PP_TXCBLEND_13 +0x2fd4 PP_TXCBLEND2_13 +0x2fd8 PP_TXABLEND_13 +0x2fdc PP_TXABLEND2_13 +0x2fe0 PP_TXCBLEND_14 +0x2fe4 PP_TXCBLEND2_14 +0x2fe8 PP_TXABLEND_14 +0x2fec PP_TXABLEND2_14 +0x2ff0 PP_TXCBLEND_15 +0x2ff4 PP_TXCBLEND2_15 +0x2ff8 PP_TXABLEND_15 +0x2ffc PP_TXABLEND2_15 +0x3218 RB3D_BLENCOLOR +0x321c RB3D_ABLENDCNTL +0x3220 RB3D_CBLENDCNTL +0x3290 RB3D_ZPASS_DATA + diff --git a/drivers/gpu/drm/radeon/reg_srcs/r300 b/drivers/gpu/drm/radeon/reg_srcs/r300 new file mode 100644 index 00000000..e8a1786b --- /dev/null +++ b/drivers/gpu/drm/radeon/reg_srcs/r300 @@ -0,0 +1,714 @@ +r300 0x4f60 +0x1434 SRC_Y_X +0x1438 DST_Y_X +0x143C DST_HEIGHT_WIDTH +0x146C DP_GUI_MASTER_CNTL +0x1474 BRUSH_Y_X +0x1478 DP_BRUSH_BKGD_CLR +0x147C DP_BRUSH_FRGD_CLR +0x1480 BRUSH_DATA0 +0x1484 BRUSH_DATA1 +0x1598 DST_WIDTH_HEIGHT +0x15C0 CLR_CMP_CNTL +0x15C4 CLR_CMP_CLR_SRC +0x15C8 CLR_CMP_CLR_DST +0x15CC CLR_CMP_MSK +0x15D8 DP_SRC_FRGD_CLR +0x15DC DP_SRC_BKGD_CLR +0x1600 DST_LINE_START +0x1604 DST_LINE_END +0x1608 DST_LINE_PATCOUNT +0x16C0 DP_CNTL +0x16CC DP_WRITE_MSK +0x16D0 DP_CNTL_XDIR_YDIR_YMAJOR +0x16E8 DEFAULT_SC_BOTTOM_RIGHT +0x16EC SC_TOP_LEFT +0x16F0 SC_BOTTOM_RIGHT +0x16F4 SRC_SC_BOTTOM_RIGHT +0x1714 DSTCACHE_CTLSTAT +0x1720 WAIT_UNTIL +0x172C RBBM_GUICNTL +0x1D98 VAP_VPORT_XSCALE +0x1D9C VAP_VPORT_XOFFSET +0x1DA0 VAP_VPORT_YSCALE +0x1DA4 VAP_VPORT_YOFFSET +0x1DA8 VAP_VPORT_ZSCALE +0x1DAC VAP_VPORT_ZOFFSET +0x2080 VAP_CNTL +0x2090 VAP_OUT_VTX_FMT_0 +0x2094 VAP_OUT_VTX_FMT_1 +0x20B0 VAP_VTE_CNTL +0x2138 VAP_VF_MIN_VTX_INDX +0x2140 VAP_CNTL_STATUS +0x2150 VAP_PROG_STREAM_CNTL_0 +0x2154 VAP_PROG_STREAM_CNTL_1 +0x2158 VAP_PROG_STREAM_CNTL_2 +0x215C VAP_PROG_STREAM_CNTL_3 +0x2160 VAP_PROG_STREAM_CNTL_4 +0x2164 VAP_PROG_STREAM_CNTL_5 +0x2168 VAP_PROG_STREAM_CNTL_6 +0x216C VAP_PROG_STREAM_CNTL_7 +0x2180 VAP_VTX_STATE_CNTL +0x2184 VAP_VSM_VTX_ASSM +0x2188 VAP_VTX_STATE_IND_REG_0 +0x218C VAP_VTX_STATE_IND_REG_1 +0x2190 VAP_VTX_STATE_IND_REG_2 +0x2194 VAP_VTX_STATE_IND_REG_3 +0x2198 VAP_VTX_STATE_IND_REG_4 +0x219C VAP_VTX_STATE_IND_REG_5 +0x21A0 VAP_VTX_STATE_IND_REG_6 +0x21A4 VAP_VTX_STATE_IND_REG_7 +0x21A8 VAP_VTX_STATE_IND_REG_8 +0x21AC VAP_VTX_STATE_IND_REG_9 +0x21B0 VAP_VTX_STATE_IND_REG_10 +0x21B4 VAP_VTX_STATE_IND_REG_11 +0x21B8 VAP_VTX_STATE_IND_REG_12 +0x21BC VAP_VTX_STATE_IND_REG_13 +0x21C0 VAP_VTX_STATE_IND_REG_14 +0x21C4 VAP_VTX_STATE_IND_REG_15 +0x21DC VAP_PSC_SGN_NORM_CNTL +0x21E0 VAP_PROG_STREAM_CNTL_EXT_0 +0x21E4 VAP_PROG_STREAM_CNTL_EXT_1 +0x21E8 VAP_PROG_STREAM_CNTL_EXT_2 +0x21EC VAP_PROG_STREAM_CNTL_EXT_3 +0x21F0 VAP_PROG_STREAM_CNTL_EXT_4 +0x21F4 VAP_PROG_STREAM_CNTL_EXT_5 +0x21F8 VAP_PROG_STREAM_CNTL_EXT_6 +0x21FC VAP_PROG_STREAM_CNTL_EXT_7 +0x2200 VAP_PVS_VECTOR_INDX_REG +0x2204 VAP_PVS_VECTOR_DATA_REG +0x2208 VAP_PVS_VECTOR_DATA_REG_128 +0x221C VAP_CLIP_CNTL +0x2220 VAP_GB_VERT_CLIP_ADJ +0x2224 VAP_GB_VERT_DISC_ADJ +0x2228 VAP_GB_HORZ_CLIP_ADJ +0x222C VAP_GB_HORZ_DISC_ADJ +0x2230 VAP_PVS_FLOW_CNTL_ADDRS_0 +0x2234 VAP_PVS_FLOW_CNTL_ADDRS_1 +0x2238 VAP_PVS_FLOW_CNTL_ADDRS_2 +0x223C VAP_PVS_FLOW_CNTL_ADDRS_3 +0x2240 VAP_PVS_FLOW_CNTL_ADDRS_4 +0x2244 VAP_PVS_FLOW_CNTL_ADDRS_5 +0x2248 VAP_PVS_FLOW_CNTL_ADDRS_6 +0x224C VAP_PVS_FLOW_CNTL_ADDRS_7 +0x2250 VAP_PVS_FLOW_CNTL_ADDRS_8 +0x2254 VAP_PVS_FLOW_CNTL_ADDRS_9 +0x2258 VAP_PVS_FLOW_CNTL_ADDRS_10 +0x225C VAP_PVS_FLOW_CNTL_ADDRS_11 +0x2260 VAP_PVS_FLOW_CNTL_ADDRS_12 +0x2264 VAP_PVS_FLOW_CNTL_ADDRS_13 +0x2268 VAP_PVS_FLOW_CNTL_ADDRS_14 +0x226C VAP_PVS_FLOW_CNTL_ADDRS_15 +0x2284 VAP_PVS_STATE_FLUSH_REG +0x2288 VAP_PVS_VTX_TIMEOUT_REG +0x2290 VAP_PVS_FLOW_CNTL_LOOP_INDEX_0 +0x2294 VAP_PVS_FLOW_CNTL_LOOP_INDEX_1 +0x2298 VAP_PVS_FLOW_CNTL_LOOP_INDEX_2 +0x229C VAP_PVS_FLOW_CNTL_LOOP_INDEX_3 +0x22A0 VAP_PVS_FLOW_CNTL_LOOP_INDEX_4 +0x22A4 VAP_PVS_FLOW_CNTL_LOOP_INDEX_5 +0x22A8 VAP_PVS_FLOW_CNTL_LOOP_INDEX_6 +0x22AC VAP_PVS_FLOW_CNTL_LOOP_INDEX_7 +0x22B0 VAP_PVS_FLOW_CNTL_LOOP_INDEX_8 +0x22B4 VAP_PVS_FLOW_CNTL_LOOP_INDEX_9 +0x22B8 VAP_PVS_FLOW_CNTL_LOOP_INDEX_10 +0x22BC VAP_PVS_FLOW_CNTL_LOOP_INDEX_11 +0x22C0 VAP_PVS_FLOW_CNTL_LOOP_INDEX_12 +0x22C4 VAP_PVS_FLOW_CNTL_LOOP_INDEX_13 +0x22C8 VAP_PVS_FLOW_CNTL_LOOP_INDEX_14 +0x22CC VAP_PVS_FLOW_CNTL_LOOP_INDEX_15 +0x22D0 VAP_PVS_CODE_CNTL_0 +0x22D4 VAP_PVS_CONST_CNTL +0x22D8 VAP_PVS_CODE_CNTL_1 +0x22DC VAP_PVS_FLOW_CNTL_OPC +0x342C RB2D_DSTCACHE_CTLSTAT +0x4000 GB_VAP_RASTER_VTX_FMT_0 +0x4004 GB_VAP_RASTER_VTX_FMT_1 +0x4008 GB_ENABLE +0x4010 GB_MSPOS0 +0x4014 GB_MSPOS1 +0x401C GB_SELECT +0x4020 GB_AA_CONFIG +0x4024 GB_FIFO_SIZE +0x4100 TX_INVALTAGS +0x4200 GA_POINT_S0 +0x4204 GA_POINT_T0 +0x4208 GA_POINT_S1 +0x420C GA_POINT_T1 +0x4214 GA_TRIANGLE_STIPPLE +0x421C GA_POINT_SIZE +0x4230 GA_POINT_MINMAX +0x4234 GA_LINE_CNTL +0x4238 GA_LINE_STIPPLE_CONFIG +0x4260 GA_LINE_STIPPLE_VALUE +0x4264 GA_LINE_S0 +0x4268 GA_LINE_S1 +0x4278 GA_COLOR_CONTROL +0x427C GA_SOLID_RG +0x4280 GA_SOLID_BA +0x4288 GA_POLY_MODE +0x428C GA_ROUND_MODE +0x4290 GA_OFFSET +0x4294 GA_FOG_SCALE +0x4298 GA_FOG_OFFSET +0x42A0 SU_TEX_WRAP +0x42A4 SU_POLY_OFFSET_FRONT_SCALE +0x42A8 SU_POLY_OFFSET_FRONT_OFFSET +0x42AC SU_POLY_OFFSET_BACK_SCALE +0x42B0 SU_POLY_OFFSET_BACK_OFFSET +0x42B4 SU_POLY_OFFSET_ENABLE +0x42B8 SU_CULL_MODE +0x42C0 SU_DEPTH_SCALE +0x42C4 SU_DEPTH_OFFSET +0x42C8 SU_REG_DEST +0x4300 RS_COUNT +0x4304 RS_INST_COUNT +0x4310 RS_IP_0 +0x4314 RS_IP_1 +0x4318 RS_IP_2 +0x431C RS_IP_3 +0x4320 RS_IP_4 +0x4324 RS_IP_5 +0x4328 RS_IP_6 +0x432C RS_IP_7 +0x4330 RS_INST_0 +0x4334 RS_INST_1 +0x4338 RS_INST_2 +0x433C RS_INST_3 +0x4340 RS_INST_4 +0x4344 RS_INST_5 +0x4348 RS_INST_6 +0x434C RS_INST_7 +0x4350 RS_INST_8 +0x4354 RS_INST_9 +0x4358 RS_INST_10 +0x435C RS_INST_11 +0x4360 RS_INST_12 +0x4364 RS_INST_13 +0x4368 RS_INST_14 +0x436C RS_INST_15 +0x43A8 SC_EDGERULE +0x43B0 SC_CLIP_0_A +0x43B4 SC_CLIP_0_B +0x43B8 SC_CLIP_1_A +0x43BC SC_CLIP_1_B +0x43C0 SC_CLIP_2_A +0x43C4 SC_CLIP_2_B +0x43C8 SC_CLIP_3_A +0x43CC SC_CLIP_3_B +0x43D0 SC_CLIP_RULE +0x43E0 SC_SCISSOR0 +0x43E8 SC_SCREENDOOR +0x4440 TX_FILTER1_0 +0x4444 TX_FILTER1_1 +0x4448 TX_FILTER1_2 +0x444C TX_FILTER1_3 +0x4450 TX_FILTER1_4 +0x4454 TX_FILTER1_5 +0x4458 TX_FILTER1_6 +0x445C TX_FILTER1_7 +0x4460 TX_FILTER1_8 +0x4464 TX_FILTER1_9 +0x4468 TX_FILTER1_10 +0x446C TX_FILTER1_11 +0x4470 TX_FILTER1_12 +0x4474 TX_FILTER1_13 +0x4478 TX_FILTER1_14 +0x447C TX_FILTER1_15 +0x4580 TX_CHROMA_KEY_0 +0x4584 TX_CHROMA_KEY_1 +0x4588 TX_CHROMA_KEY_2 +0x458C TX_CHROMA_KEY_3 +0x4590 TX_CHROMA_KEY_4 +0x4594 TX_CHROMA_KEY_5 +0x4598 TX_CHROMA_KEY_6 +0x459C TX_CHROMA_KEY_7 +0x45A0 TX_CHROMA_KEY_8 +0x45A4 TX_CHROMA_KEY_9 +0x45A8 TX_CHROMA_KEY_10 +0x45AC TX_CHROMA_KEY_11 +0x45B0 TX_CHROMA_KEY_12 +0x45B4 TX_CHROMA_KEY_13 +0x45B8 TX_CHROMA_KEY_14 +0x45BC TX_CHROMA_KEY_15 +0x45C0 TX_BORDER_COLOR_0 +0x45C4 TX_BORDER_COLOR_1 +0x45C8 TX_BORDER_COLOR_2 +0x45CC TX_BORDER_COLOR_3 +0x45D0 TX_BORDER_COLOR_4 +0x45D4 TX_BORDER_COLOR_5 +0x45D8 TX_BORDER_COLOR_6 +0x45DC TX_BORDER_COLOR_7 +0x45E0 TX_BORDER_COLOR_8 +0x45E4 TX_BORDER_COLOR_9 +0x45E8 TX_BORDER_COLOR_10 +0x45EC TX_BORDER_COLOR_11 +0x45F0 TX_BORDER_COLOR_12 +0x45F4 TX_BORDER_COLOR_13 +0x45F8 TX_BORDER_COLOR_14 +0x45FC TX_BORDER_COLOR_15 +0x4600 US_CONFIG +0x4604 US_PIXSIZE +0x4608 US_CODE_OFFSET +0x460C US_RESET +0x4610 US_CODE_ADDR_0 +0x4614 US_CODE_ADDR_1 +0x4618 US_CODE_ADDR_2 +0x461C US_CODE_ADDR_3 +0x4620 US_TEX_INST_0 +0x4624 US_TEX_INST_1 +0x4628 US_TEX_INST_2 +0x462C US_TEX_INST_3 +0x4630 US_TEX_INST_4 +0x4634 US_TEX_INST_5 +0x4638 US_TEX_INST_6 +0x463C US_TEX_INST_7 +0x4640 US_TEX_INST_8 +0x4644 US_TEX_INST_9 +0x4648 US_TEX_INST_10 +0x464C US_TEX_INST_11 +0x4650 US_TEX_INST_12 +0x4654 US_TEX_INST_13 +0x4658 US_TEX_INST_14 +0x465C US_TEX_INST_15 +0x4660 US_TEX_INST_16 +0x4664 US_TEX_INST_17 +0x4668 US_TEX_INST_18 +0x466C US_TEX_INST_19 +0x4670 US_TEX_INST_20 +0x4674 US_TEX_INST_21 +0x4678 US_TEX_INST_22 +0x467C US_TEX_INST_23 +0x4680 US_TEX_INST_24 +0x4684 US_TEX_INST_25 +0x4688 US_TEX_INST_26 +0x468C US_TEX_INST_27 +0x4690 US_TEX_INST_28 +0x4694 US_TEX_INST_29 +0x4698 US_TEX_INST_30 +0x469C US_TEX_INST_31 +0x46A4 US_OUT_FMT_0 +0x46A8 US_OUT_FMT_1 +0x46AC US_OUT_FMT_2 +0x46B0 US_OUT_FMT_3 +0x46B4 US_W_FMT +0x46C0 US_ALU_RGB_ADDR_0 +0x46C4 US_ALU_RGB_ADDR_1 +0x46C8 US_ALU_RGB_ADDR_2 +0x46CC US_ALU_RGB_ADDR_3 +0x46D0 US_ALU_RGB_ADDR_4 +0x46D4 US_ALU_RGB_ADDR_5 +0x46D8 US_ALU_RGB_ADDR_6 +0x46DC US_ALU_RGB_ADDR_7 +0x46E0 US_ALU_RGB_ADDR_8 +0x46E4 US_ALU_RGB_ADDR_9 +0x46E8 US_ALU_RGB_ADDR_10 +0x46EC US_ALU_RGB_ADDR_11 +0x46F0 US_ALU_RGB_ADDR_12 +0x46F4 US_ALU_RGB_ADDR_13 +0x46F8 US_ALU_RGB_ADDR_14 +0x46FC US_ALU_RGB_ADDR_15 +0x4700 US_ALU_RGB_ADDR_16 +0x4704 US_ALU_RGB_ADDR_17 +0x4708 US_ALU_RGB_ADDR_18 +0x470C US_ALU_RGB_ADDR_19 +0x4710 US_ALU_RGB_ADDR_20 +0x4714 US_ALU_RGB_ADDR_21 +0x4718 US_ALU_RGB_ADDR_22 +0x471C US_ALU_RGB_ADDR_23 +0x4720 US_ALU_RGB_ADDR_24 +0x4724 US_ALU_RGB_ADDR_25 +0x4728 US_ALU_RGB_ADDR_26 +0x472C US_ALU_RGB_ADDR_27 +0x4730 US_ALU_RGB_ADDR_28 +0x4734 US_ALU_RGB_ADDR_29 +0x4738 US_ALU_RGB_ADDR_30 +0x473C US_ALU_RGB_ADDR_31 +0x4740 US_ALU_RGB_ADDR_32 +0x4744 US_ALU_RGB_ADDR_33 +0x4748 US_ALU_RGB_ADDR_34 +0x474C US_ALU_RGB_ADDR_35 +0x4750 US_ALU_RGB_ADDR_36 +0x4754 US_ALU_RGB_ADDR_37 +0x4758 US_ALU_RGB_ADDR_38 +0x475C US_ALU_RGB_ADDR_39 +0x4760 US_ALU_RGB_ADDR_40 +0x4764 US_ALU_RGB_ADDR_41 +0x4768 US_ALU_RGB_ADDR_42 +0x476C US_ALU_RGB_ADDR_43 +0x4770 US_ALU_RGB_ADDR_44 +0x4774 US_ALU_RGB_ADDR_45 +0x4778 US_ALU_RGB_ADDR_46 +0x477C US_ALU_RGB_ADDR_47 +0x4780 US_ALU_RGB_ADDR_48 +0x4784 US_ALU_RGB_ADDR_49 +0x4788 US_ALU_RGB_ADDR_50 +0x478C US_ALU_RGB_ADDR_51 +0x4790 US_ALU_RGB_ADDR_52 +0x4794 US_ALU_RGB_ADDR_53 +0x4798 US_ALU_RGB_ADDR_54 +0x479C US_ALU_RGB_ADDR_55 +0x47A0 US_ALU_RGB_ADDR_56 +0x47A4 US_ALU_RGB_ADDR_57 +0x47A8 US_ALU_RGB_ADDR_58 +0x47AC US_ALU_RGB_ADDR_59 +0x47B0 US_ALU_RGB_ADDR_60 +0x47B4 US_ALU_RGB_ADDR_61 +0x47B8 US_ALU_RGB_ADDR_62 +0x47BC US_ALU_RGB_ADDR_63 +0x47C0 US_ALU_ALPHA_ADDR_0 +0x47C4 US_ALU_ALPHA_ADDR_1 +0x47C8 US_ALU_ALPHA_ADDR_2 +0x47CC US_ALU_ALPHA_ADDR_3 +0x47D0 US_ALU_ALPHA_ADDR_4 +0x47D4 US_ALU_ALPHA_ADDR_5 +0x47D8 US_ALU_ALPHA_ADDR_6 +0x47DC US_ALU_ALPHA_ADDR_7 +0x47E0 US_ALU_ALPHA_ADDR_8 +0x47E4 US_ALU_ALPHA_ADDR_9 +0x47E8 US_ALU_ALPHA_ADDR_10 +0x47EC US_ALU_ALPHA_ADDR_11 +0x47F0 US_ALU_ALPHA_ADDR_12 +0x47F4 US_ALU_ALPHA_ADDR_13 +0x47F8 US_ALU_ALPHA_ADDR_14 +0x47FC US_ALU_ALPHA_ADDR_15 +0x4800 US_ALU_ALPHA_ADDR_16 +0x4804 US_ALU_ALPHA_ADDR_17 +0x4808 US_ALU_ALPHA_ADDR_18 +0x480C US_ALU_ALPHA_ADDR_19 +0x4810 US_ALU_ALPHA_ADDR_20 +0x4814 US_ALU_ALPHA_ADDR_21 +0x4818 US_ALU_ALPHA_ADDR_22 +0x481C US_ALU_ALPHA_ADDR_23 +0x4820 US_ALU_ALPHA_ADDR_24 +0x4824 US_ALU_ALPHA_ADDR_25 +0x4828 US_ALU_ALPHA_ADDR_26 +0x482C US_ALU_ALPHA_ADDR_27 +0x4830 US_ALU_ALPHA_ADDR_28 +0x4834 US_ALU_ALPHA_ADDR_29 +0x4838 US_ALU_ALPHA_ADDR_30 +0x483C US_ALU_ALPHA_ADDR_31 +0x4840 US_ALU_ALPHA_ADDR_32 +0x4844 US_ALU_ALPHA_ADDR_33 +0x4848 US_ALU_ALPHA_ADDR_34 +0x484C US_ALU_ALPHA_ADDR_35 +0x4850 US_ALU_ALPHA_ADDR_36 +0x4854 US_ALU_ALPHA_ADDR_37 +0x4858 US_ALU_ALPHA_ADDR_38 +0x485C US_ALU_ALPHA_ADDR_39 +0x4860 US_ALU_ALPHA_ADDR_40 +0x4864 US_ALU_ALPHA_ADDR_41 +0x4868 US_ALU_ALPHA_ADDR_42 +0x486C US_ALU_ALPHA_ADDR_43 +0x4870 US_ALU_ALPHA_ADDR_44 +0x4874 US_ALU_ALPHA_ADDR_45 +0x4878 US_ALU_ALPHA_ADDR_46 +0x487C US_ALU_ALPHA_ADDR_47 +0x4880 US_ALU_ALPHA_ADDR_48 +0x4884 US_ALU_ALPHA_ADDR_49 +0x4888 US_ALU_ALPHA_ADDR_50 +0x488C US_ALU_ALPHA_ADDR_51 +0x4890 US_ALU_ALPHA_ADDR_52 +0x4894 US_ALU_ALPHA_ADDR_53 +0x4898 US_ALU_ALPHA_ADDR_54 +0x489C US_ALU_ALPHA_ADDR_55 +0x48A0 US_ALU_ALPHA_ADDR_56 +0x48A4 US_ALU_ALPHA_ADDR_57 +0x48A8 US_ALU_ALPHA_ADDR_58 +0x48AC US_ALU_ALPHA_ADDR_59 +0x48B0 US_ALU_ALPHA_ADDR_60 +0x48B4 US_ALU_ALPHA_ADDR_61 +0x48B8 US_ALU_ALPHA_ADDR_62 +0x48BC US_ALU_ALPHA_ADDR_63 +0x48C0 US_ALU_RGB_INST_0 +0x48C4 US_ALU_RGB_INST_1 +0x48C8 US_ALU_RGB_INST_2 +0x48CC US_ALU_RGB_INST_3 +0x48D0 US_ALU_RGB_INST_4 +0x48D4 US_ALU_RGB_INST_5 +0x48D8 US_ALU_RGB_INST_6 +0x48DC US_ALU_RGB_INST_7 +0x48E0 US_ALU_RGB_INST_8 +0x48E4 US_ALU_RGB_INST_9 +0x48E8 US_ALU_RGB_INST_10 +0x48EC US_ALU_RGB_INST_11 +0x48F0 US_ALU_RGB_INST_12 +0x48F4 US_ALU_RGB_INST_13 +0x48F8 US_ALU_RGB_INST_14 +0x48FC US_ALU_RGB_INST_15 +0x4900 US_ALU_RGB_INST_16 +0x4904 US_ALU_RGB_INST_17 +0x4908 US_ALU_RGB_INST_18 +0x490C US_ALU_RGB_INST_19 +0x4910 US_ALU_RGB_INST_20 +0x4914 US_ALU_RGB_INST_21 +0x4918 US_ALU_RGB_INST_22 +0x491C US_ALU_RGB_INST_23 +0x4920 US_ALU_RGB_INST_24 +0x4924 US_ALU_RGB_INST_25 +0x4928 US_ALU_RGB_INST_26 +0x492C US_ALU_RGB_INST_27 +0x4930 US_ALU_RGB_INST_28 +0x4934 US_ALU_RGB_INST_29 +0x4938 US_ALU_RGB_INST_30 +0x493C US_ALU_RGB_INST_31 +0x4940 US_ALU_RGB_INST_32 +0x4944 US_ALU_RGB_INST_33 +0x4948 US_ALU_RGB_INST_34 +0x494C US_ALU_RGB_INST_35 +0x4950 US_ALU_RGB_INST_36 +0x4954 US_ALU_RGB_INST_37 +0x4958 US_ALU_RGB_INST_38 +0x495C US_ALU_RGB_INST_39 +0x4960 US_ALU_RGB_INST_40 +0x4964 US_ALU_RGB_INST_41 +0x4968 US_ALU_RGB_INST_42 +0x496C US_ALU_RGB_INST_43 +0x4970 US_ALU_RGB_INST_44 +0x4974 US_ALU_RGB_INST_45 +0x4978 US_ALU_RGB_INST_46 +0x497C US_ALU_RGB_INST_47 +0x4980 US_ALU_RGB_INST_48 +0x4984 US_ALU_RGB_INST_49 +0x4988 US_ALU_RGB_INST_50 +0x498C US_ALU_RGB_INST_51 +0x4990 US_ALU_RGB_INST_52 +0x4994 US_ALU_RGB_INST_53 +0x4998 US_ALU_RGB_INST_54 +0x499C US_ALU_RGB_INST_55 +0x49A0 US_ALU_RGB_INST_56 +0x49A4 US_ALU_RGB_INST_57 +0x49A8 US_ALU_RGB_INST_58 +0x49AC US_ALU_RGB_INST_59 +0x49B0 US_ALU_RGB_INST_60 +0x49B4 US_ALU_RGB_INST_61 +0x49B8 US_ALU_RGB_INST_62 +0x49BC US_ALU_RGB_INST_63 +0x49C0 US_ALU_ALPHA_INST_0 +0x49C4 US_ALU_ALPHA_INST_1 +0x49C8 US_ALU_ALPHA_INST_2 +0x49CC US_ALU_ALPHA_INST_3 +0x49D0 US_ALU_ALPHA_INST_4 +0x49D4 US_ALU_ALPHA_INST_5 +0x49D8 US_ALU_ALPHA_INST_6 +0x49DC US_ALU_ALPHA_INST_7 +0x49E0 US_ALU_ALPHA_INST_8 +0x49E4 US_ALU_ALPHA_INST_9 +0x49E8 US_ALU_ALPHA_INST_10 +0x49EC US_ALU_ALPHA_INST_11 +0x49F0 US_ALU_ALPHA_INST_12 +0x49F4 US_ALU_ALPHA_INST_13 +0x49F8 US_ALU_ALPHA_INST_14 +0x49FC US_ALU_ALPHA_INST_15 +0x4A00 US_ALU_ALPHA_INST_16 +0x4A04 US_ALU_ALPHA_INST_17 +0x4A08 US_ALU_ALPHA_INST_18 +0x4A0C US_ALU_ALPHA_INST_19 +0x4A10 US_ALU_ALPHA_INST_20 +0x4A14 US_ALU_ALPHA_INST_21 +0x4A18 US_ALU_ALPHA_INST_22 +0x4A1C US_ALU_ALPHA_INST_23 +0x4A20 US_ALU_ALPHA_INST_24 +0x4A24 US_ALU_ALPHA_INST_25 +0x4A28 US_ALU_ALPHA_INST_26 +0x4A2C US_ALU_ALPHA_INST_27 +0x4A30 US_ALU_ALPHA_INST_28 +0x4A34 US_ALU_ALPHA_INST_29 +0x4A38 US_ALU_ALPHA_INST_30 +0x4A3C US_ALU_ALPHA_INST_31 +0x4A40 US_ALU_ALPHA_INST_32 +0x4A44 US_ALU_ALPHA_INST_33 +0x4A48 US_ALU_ALPHA_INST_34 +0x4A4C US_ALU_ALPHA_INST_35 +0x4A50 US_ALU_ALPHA_INST_36 +0x4A54 US_ALU_ALPHA_INST_37 +0x4A58 US_ALU_ALPHA_INST_38 +0x4A5C US_ALU_ALPHA_INST_39 +0x4A60 US_ALU_ALPHA_INST_40 +0x4A64 US_ALU_ALPHA_INST_41 +0x4A68 US_ALU_ALPHA_INST_42 +0x4A6C US_ALU_ALPHA_INST_43 +0x4A70 US_ALU_ALPHA_INST_44 +0x4A74 US_ALU_ALPHA_INST_45 +0x4A78 US_ALU_ALPHA_INST_46 +0x4A7C US_ALU_ALPHA_INST_47 +0x4A80 US_ALU_ALPHA_INST_48 +0x4A84 US_ALU_ALPHA_INST_49 +0x4A88 US_ALU_ALPHA_INST_50 +0x4A8C US_ALU_ALPHA_INST_51 +0x4A90 US_ALU_ALPHA_INST_52 +0x4A94 US_ALU_ALPHA_INST_53 +0x4A98 US_ALU_ALPHA_INST_54 +0x4A9C US_ALU_ALPHA_INST_55 +0x4AA0 US_ALU_ALPHA_INST_56 +0x4AA4 US_ALU_ALPHA_INST_57 +0x4AA8 US_ALU_ALPHA_INST_58 +0x4AAC US_ALU_ALPHA_INST_59 +0x4AB0 US_ALU_ALPHA_INST_60 +0x4AB4 US_ALU_ALPHA_INST_61 +0x4AB8 US_ALU_ALPHA_INST_62 +0x4ABC US_ALU_ALPHA_INST_63 +0x4BC0 FG_FOG_BLEND +0x4BC4 FG_FOG_FACTOR +0x4BC8 FG_FOG_COLOR_R +0x4BCC FG_FOG_COLOR_G +0x4BD0 FG_FOG_COLOR_B +0x4BD4 FG_ALPHA_FUNC +0x4BD8 FG_DEPTH_SRC +0x4C00 US_ALU_CONST_R_0 +0x4C04 US_ALU_CONST_G_0 +0x4C08 US_ALU_CONST_B_0 +0x4C0C US_ALU_CONST_A_0 +0x4C10 US_ALU_CONST_R_1 +0x4C14 US_ALU_CONST_G_1 +0x4C18 US_ALU_CONST_B_1 +0x4C1C US_ALU_CONST_A_1 +0x4C20 US_ALU_CONST_R_2 +0x4C24 US_ALU_CONST_G_2 +0x4C28 US_ALU_CONST_B_2 +0x4C2C US_ALU_CONST_A_2 +0x4C30 US_ALU_CONST_R_3 +0x4C34 US_ALU_CONST_G_3 +0x4C38 US_ALU_CONST_B_3 +0x4C3C US_ALU_CONST_A_3 +0x4C40 US_ALU_CONST_R_4 +0x4C44 US_ALU_CONST_G_4 +0x4C48 US_ALU_CONST_B_4 +0x4C4C US_ALU_CONST_A_4 +0x4C50 US_ALU_CONST_R_5 +0x4C54 US_ALU_CONST_G_5 +0x4C58 US_ALU_CONST_B_5 +0x4C5C US_ALU_CONST_A_5 +0x4C60 US_ALU_CONST_R_6 +0x4C64 US_ALU_CONST_G_6 +0x4C68 US_ALU_CONST_B_6 +0x4C6C US_ALU_CONST_A_6 +0x4C70 US_ALU_CONST_R_7 +0x4C74 US_ALU_CONST_G_7 +0x4C78 US_ALU_CONST_B_7 +0x4C7C US_ALU_CONST_A_7 +0x4C80 US_ALU_CONST_R_8 +0x4C84 US_ALU_CONST_G_8 +0x4C88 US_ALU_CONST_B_8 +0x4C8C US_ALU_CONST_A_8 +0x4C90 US_ALU_CONST_R_9 +0x4C94 US_ALU_CONST_G_9 +0x4C98 US_ALU_CONST_B_9 +0x4C9C US_ALU_CONST_A_9 +0x4CA0 US_ALU_CONST_R_10 +0x4CA4 US_ALU_CONST_G_10 +0x4CA8 US_ALU_CONST_B_10 +0x4CAC US_ALU_CONST_A_10 +0x4CB0 US_ALU_CONST_R_11 +0x4CB4 US_ALU_CONST_G_11 +0x4CB8 US_ALU_CONST_B_11 +0x4CBC US_ALU_CONST_A_11 +0x4CC0 US_ALU_CONST_R_12 +0x4CC4 US_ALU_CONST_G_12 +0x4CC8 US_ALU_CONST_B_12 +0x4CCC US_ALU_CONST_A_12 +0x4CD0 US_ALU_CONST_R_13 +0x4CD4 US_ALU_CONST_G_13 +0x4CD8 US_ALU_CONST_B_13 +0x4CDC US_ALU_CONST_A_13 +0x4CE0 US_ALU_CONST_R_14 +0x4CE4 US_ALU_CONST_G_14 +0x4CE8 US_ALU_CONST_B_14 +0x4CEC US_ALU_CONST_A_14 +0x4CF0 US_ALU_CONST_R_15 +0x4CF4 US_ALU_CONST_G_15 +0x4CF8 US_ALU_CONST_B_15 +0x4CFC US_ALU_CONST_A_15 +0x4D00 US_ALU_CONST_R_16 +0x4D04 US_ALU_CONST_G_16 +0x4D08 US_ALU_CONST_B_16 +0x4D0C US_ALU_CONST_A_16 +0x4D10 US_ALU_CONST_R_17 +0x4D14 US_ALU_CONST_G_17 +0x4D18 US_ALU_CONST_B_17 +0x4D1C US_ALU_CONST_A_17 +0x4D20 US_ALU_CONST_R_18 +0x4D24 US_ALU_CONST_G_18 +0x4D28 US_ALU_CONST_B_18 +0x4D2C US_ALU_CONST_A_18 +0x4D30 US_ALU_CONST_R_19 +0x4D34 US_ALU_CONST_G_19 +0x4D38 US_ALU_CONST_B_19 +0x4D3C US_ALU_CONST_A_19 +0x4D40 US_ALU_CONST_R_20 +0x4D44 US_ALU_CONST_G_20 +0x4D48 US_ALU_CONST_B_20 +0x4D4C US_ALU_CONST_A_20 +0x4D50 US_ALU_CONST_R_21 +0x4D54 US_ALU_CONST_G_21 +0x4D58 US_ALU_CONST_B_21 +0x4D5C US_ALU_CONST_A_21 +0x4D60 US_ALU_CONST_R_22 +0x4D64 US_ALU_CONST_G_22 +0x4D68 US_ALU_CONST_B_22 +0x4D6C US_ALU_CONST_A_22 +0x4D70 US_ALU_CONST_R_23 +0x4D74 US_ALU_CONST_G_23 +0x4D78 US_ALU_CONST_B_23 +0x4D7C US_ALU_CONST_A_23 +0x4D80 US_ALU_CONST_R_24 +0x4D84 US_ALU_CONST_G_24 +0x4D88 US_ALU_CONST_B_24 +0x4D8C US_ALU_CONST_A_24 +0x4D90 US_ALU_CONST_R_25 +0x4D94 US_ALU_CONST_G_25 +0x4D98 US_ALU_CONST_B_25 +0x4D9C US_ALU_CONST_A_25 +0x4DA0 US_ALU_CONST_R_26 +0x4DA4 US_ALU_CONST_G_26 +0x4DA8 US_ALU_CONST_B_26 +0x4DAC US_ALU_CONST_A_26 +0x4DB0 US_ALU_CONST_R_27 +0x4DB4 US_ALU_CONST_G_27 +0x4DB8 US_ALU_CONST_B_27 +0x4DBC US_ALU_CONST_A_27 +0x4DC0 US_ALU_CONST_R_28 +0x4DC4 US_ALU_CONST_G_28 +0x4DC8 US_ALU_CONST_B_28 +0x4DCC US_ALU_CONST_A_28 +0x4DD0 US_ALU_CONST_R_29 +0x4DD4 US_ALU_CONST_G_29 +0x4DD8 US_ALU_CONST_B_29 +0x4DDC US_ALU_CONST_A_29 +0x4DE0 US_ALU_CONST_R_30 +0x4DE4 US_ALU_CONST_G_30 +0x4DE8 US_ALU_CONST_B_30 +0x4DEC US_ALU_CONST_A_30 +0x4DF0 US_ALU_CONST_R_31 +0x4DF4 US_ALU_CONST_G_31 +0x4DF8 US_ALU_CONST_B_31 +0x4DFC US_ALU_CONST_A_31 +0x4E08 RB3D_ABLENDCNTL_R3 +0x4E10 RB3D_CONSTANT_COLOR +0x4E14 RB3D_COLOR_CLEAR_VALUE +0x4E18 RB3D_ROPCNTL_R3 +0x4E1C RB3D_CLRCMP_FLIPE_R3 +0x4E20 RB3D_CLRCMP_CLR_R3 +0x4E24 RB3D_CLRCMP_MSK_R3 +0x4E48 RB3D_DEBUG_CTL +0x4E4C RB3D_DSTCACHE_CTLSTAT_R3 +0x4E50 RB3D_DITHER_CTL +0x4E54 RB3D_CMASK_OFFSET0 +0x4E58 RB3D_CMASK_OFFSET1 +0x4E5C RB3D_CMASK_OFFSET2 +0x4E60 RB3D_CMASK_OFFSET3 +0x4E64 RB3D_CMASK_PITCH0 +0x4E68 RB3D_CMASK_PITCH1 +0x4E6C RB3D_CMASK_PITCH2 +0x4E70 RB3D_CMASK_PITCH3 +0x4E74 RB3D_CMASK_WRINDEX +0x4E78 RB3D_CMASK_DWORD +0x4E7C RB3D_CMASK_RDINDEX +0x4EA0 RB3D_DISCARD_SRC_PIXEL_LTE_THRESHOLD +0x4EA4 RB3D_DISCARD_SRC_PIXEL_GTE_THRESHOLD +0x4F04 ZB_ZSTENCILCNTL +0x4F08 ZB_STENCILREFMASK +0x4F14 ZB_ZTOP +0x4F18 ZB_ZCACHE_CTLSTAT +0x4F28 ZB_DEPTHCLEARVALUE +0x4F58 ZB_ZPASS_DATA diff --git a/drivers/gpu/drm/radeon/reg_srcs/r420 b/drivers/gpu/drm/radeon/reg_srcs/r420 new file mode 100644 index 00000000..722074e2 --- /dev/null +++ b/drivers/gpu/drm/radeon/reg_srcs/r420 @@ -0,0 +1,780 @@ +r420 0x4f60 +0x1434 SRC_Y_X +0x1438 DST_Y_X +0x143C DST_HEIGHT_WIDTH +0x146C DP_GUI_MASTER_CNTL +0x1474 BRUSH_Y_X +0x1478 DP_BRUSH_BKGD_CLR +0x147C DP_BRUSH_FRGD_CLR +0x1480 BRUSH_DATA0 +0x1484 BRUSH_DATA1 +0x1598 DST_WIDTH_HEIGHT +0x15C0 CLR_CMP_CNTL +0x15C4 CLR_CMP_CLR_SRC +0x15C8 CLR_CMP_CLR_DST +0x15CC CLR_CMP_MSK +0x15D8 DP_SRC_FRGD_CLR +0x15DC DP_SRC_BKGD_CLR +0x1600 DST_LINE_START +0x1604 DST_LINE_END +0x1608 DST_LINE_PATCOUNT +0x16C0 DP_CNTL +0x16CC DP_WRITE_MSK +0x16D0 DP_CNTL_XDIR_YDIR_YMAJOR +0x16E8 DEFAULT_SC_BOTTOM_RIGHT +0x16EC SC_TOP_LEFT +0x16F0 SC_BOTTOM_RIGHT +0x16F4 SRC_SC_BOTTOM_RIGHT +0x1714 DSTCACHE_CTLSTAT +0x1720 WAIT_UNTIL +0x172C RBBM_GUICNTL +0x1D98 VAP_VPORT_XSCALE +0x1D9C VAP_VPORT_XOFFSET +0x1DA0 VAP_VPORT_YSCALE +0x1DA4 VAP_VPORT_YOFFSET +0x1DA8 VAP_VPORT_ZSCALE +0x1DAC VAP_VPORT_ZOFFSET +0x2080 VAP_CNTL +0x2090 VAP_OUT_VTX_FMT_0 +0x2094 VAP_OUT_VTX_FMT_1 +0x20B0 VAP_VTE_CNTL +0x2138 VAP_VF_MIN_VTX_INDX +0x2140 VAP_CNTL_STATUS +0x2150 VAP_PROG_STREAM_CNTL_0 +0x2154 VAP_PROG_STREAM_CNTL_1 +0x2158 VAP_PROG_STREAM_CNTL_2 +0x215C VAP_PROG_STREAM_CNTL_3 +0x2160 VAP_PROG_STREAM_CNTL_4 +0x2164 VAP_PROG_STREAM_CNTL_5 +0x2168 VAP_PROG_STREAM_CNTL_6 +0x216C VAP_PROG_STREAM_CNTL_7 +0x2180 VAP_VTX_STATE_CNTL +0x2184 VAP_VSM_VTX_ASSM +0x2188 VAP_VTX_STATE_IND_REG_0 +0x218C VAP_VTX_STATE_IND_REG_1 +0x2190 VAP_VTX_STATE_IND_REG_2 +0x2194 VAP_VTX_STATE_IND_REG_3 +0x2198 VAP_VTX_STATE_IND_REG_4 +0x219C VAP_VTX_STATE_IND_REG_5 +0x21A0 VAP_VTX_STATE_IND_REG_6 +0x21A4 VAP_VTX_STATE_IND_REG_7 +0x21A8 VAP_VTX_STATE_IND_REG_8 +0x21AC VAP_VTX_STATE_IND_REG_9 +0x21B0 VAP_VTX_STATE_IND_REG_10 +0x21B4 VAP_VTX_STATE_IND_REG_11 +0x21B8 VAP_VTX_STATE_IND_REG_12 +0x21BC VAP_VTX_STATE_IND_REG_13 +0x21C0 VAP_VTX_STATE_IND_REG_14 +0x21C4 VAP_VTX_STATE_IND_REG_15 +0x21DC VAP_PSC_SGN_NORM_CNTL +0x21E0 VAP_PROG_STREAM_CNTL_EXT_0 +0x21E4 VAP_PROG_STREAM_CNTL_EXT_1 +0x21E8 VAP_PROG_STREAM_CNTL_EXT_2 +0x21EC VAP_PROG_STREAM_CNTL_EXT_3 +0x21F0 VAP_PROG_STREAM_CNTL_EXT_4 +0x21F4 VAP_PROG_STREAM_CNTL_EXT_5 +0x21F8 VAP_PROG_STREAM_CNTL_EXT_6 +0x21FC VAP_PROG_STREAM_CNTL_EXT_7 +0x2200 VAP_PVS_VECTOR_INDX_REG +0x2204 VAP_PVS_VECTOR_DATA_REG +0x2208 VAP_PVS_VECTOR_DATA_REG_128 +0x221C VAP_CLIP_CNTL +0x2220 VAP_GB_VERT_CLIP_ADJ +0x2224 VAP_GB_VERT_DISC_ADJ +0x2228 VAP_GB_HORZ_CLIP_ADJ +0x222C VAP_GB_HORZ_DISC_ADJ +0x2230 VAP_PVS_FLOW_CNTL_ADDRS_0 +0x2234 VAP_PVS_FLOW_CNTL_ADDRS_1 +0x2238 VAP_PVS_FLOW_CNTL_ADDRS_2 +0x223C VAP_PVS_FLOW_CNTL_ADDRS_3 +0x2240 VAP_PVS_FLOW_CNTL_ADDRS_4 +0x2244 VAP_PVS_FLOW_CNTL_ADDRS_5 +0x2248 VAP_PVS_FLOW_CNTL_ADDRS_6 +0x224C VAP_PVS_FLOW_CNTL_ADDRS_7 +0x2250 VAP_PVS_FLOW_CNTL_ADDRS_8 +0x2254 VAP_PVS_FLOW_CNTL_ADDRS_9 +0x2258 VAP_PVS_FLOW_CNTL_ADDRS_10 +0x225C VAP_PVS_FLOW_CNTL_ADDRS_11 +0x2260 VAP_PVS_FLOW_CNTL_ADDRS_12 +0x2264 VAP_PVS_FLOW_CNTL_ADDRS_13 +0x2268 VAP_PVS_FLOW_CNTL_ADDRS_14 +0x226C VAP_PVS_FLOW_CNTL_ADDRS_15 +0x2284 VAP_PVS_STATE_FLUSH_REG +0x2288 VAP_PVS_VTX_TIMEOUT_REG +0x2290 VAP_PVS_FLOW_CNTL_LOOP_INDEX_0 +0x2294 VAP_PVS_FLOW_CNTL_LOOP_INDEX_1 +0x2298 VAP_PVS_FLOW_CNTL_LOOP_INDEX_2 +0x229C VAP_PVS_FLOW_CNTL_LOOP_INDEX_3 +0x22A0 VAP_PVS_FLOW_CNTL_LOOP_INDEX_4 +0x22A4 VAP_PVS_FLOW_CNTL_LOOP_INDEX_5 +0x22A8 VAP_PVS_FLOW_CNTL_LOOP_INDEX_6 +0x22AC VAP_PVS_FLOW_CNTL_LOOP_INDEX_7 +0x22B0 VAP_PVS_FLOW_CNTL_LOOP_INDEX_8 +0x22B4 VAP_PVS_FLOW_CNTL_LOOP_INDEX_9 +0x22B8 VAP_PVS_FLOW_CNTL_LOOP_INDEX_10 +0x22BC VAP_PVS_FLOW_CNTL_LOOP_INDEX_11 +0x22C0 VAP_PVS_FLOW_CNTL_LOOP_INDEX_12 +0x22C4 VAP_PVS_FLOW_CNTL_LOOP_INDEX_13 +0x22C8 VAP_PVS_FLOW_CNTL_LOOP_INDEX_14 +0x22CC VAP_PVS_FLOW_CNTL_LOOP_INDEX_15 +0x22D0 VAP_PVS_CODE_CNTL_0 +0x22D4 VAP_PVS_CONST_CNTL +0x22D8 VAP_PVS_CODE_CNTL_1 +0x22DC VAP_PVS_FLOW_CNTL_OPC +0x342C RB2D_DSTCACHE_CTLSTAT +0x4000 GB_VAP_RASTER_VTX_FMT_0 +0x4004 GB_VAP_RASTER_VTX_FMT_1 +0x4008 GB_ENABLE +0x4010 GB_MSPOS0 +0x4014 GB_MSPOS1 +0x401C GB_SELECT +0x4020 GB_AA_CONFIG +0x4024 GB_FIFO_SIZE +0x4100 TX_INVALTAGS +0x4200 GA_POINT_S0 +0x4204 GA_POINT_T0 +0x4208 GA_POINT_S1 +0x420C GA_POINT_T1 +0x4214 GA_TRIANGLE_STIPPLE +0x421C GA_POINT_SIZE +0x4230 GA_POINT_MINMAX +0x4234 GA_LINE_CNTL +0x4238 GA_LINE_STIPPLE_CONFIG +0x4260 GA_LINE_STIPPLE_VALUE +0x4264 GA_LINE_S0 +0x4268 GA_LINE_S1 +0x4278 GA_COLOR_CONTROL +0x427C GA_SOLID_RG +0x4280 GA_SOLID_BA +0x4288 GA_POLY_MODE +0x428C GA_ROUND_MODE +0x4290 GA_OFFSET +0x4294 GA_FOG_SCALE +0x4298 GA_FOG_OFFSET +0x42A0 SU_TEX_WRAP +0x42A4 SU_POLY_OFFSET_FRONT_SCALE +0x42A8 SU_POLY_OFFSET_FRONT_OFFSET +0x42AC SU_POLY_OFFSET_BACK_SCALE +0x42B0 SU_POLY_OFFSET_BACK_OFFSET +0x42B4 SU_POLY_OFFSET_ENABLE +0x42B8 SU_CULL_MODE +0x42C0 SU_DEPTH_SCALE +0x42C4 SU_DEPTH_OFFSET +0x42C8 SU_REG_DEST +0x4300 RS_COUNT +0x4304 RS_INST_COUNT +0x4310 RS_IP_0 +0x4314 RS_IP_1 +0x4318 RS_IP_2 +0x431C RS_IP_3 +0x4320 RS_IP_4 +0x4324 RS_IP_5 +0x4328 RS_IP_6 +0x432C RS_IP_7 +0x4330 RS_INST_0 +0x4334 RS_INST_1 +0x4338 RS_INST_2 +0x433C RS_INST_3 +0x4340 RS_INST_4 +0x4344 RS_INST_5 +0x4348 RS_INST_6 +0x434C RS_INST_7 +0x4350 RS_INST_8 +0x4354 RS_INST_9 +0x4358 RS_INST_10 +0x435C RS_INST_11 +0x4360 RS_INST_12 +0x4364 RS_INST_13 +0x4368 RS_INST_14 +0x436C RS_INST_15 +0x43A8 SC_EDGERULE +0x43B0 SC_CLIP_0_A +0x43B4 SC_CLIP_0_B +0x43B8 SC_CLIP_1_A +0x43BC SC_CLIP_1_B +0x43C0 SC_CLIP_2_A +0x43C4 SC_CLIP_2_B +0x43C8 SC_CLIP_3_A +0x43CC SC_CLIP_3_B +0x43D0 SC_CLIP_RULE +0x43E0 SC_SCISSOR0 +0x43E8 SC_SCREENDOOR +0x4440 TX_FILTER1_0 +0x4444 TX_FILTER1_1 +0x4448 TX_FILTER1_2 +0x444C TX_FILTER1_3 +0x4450 TX_FILTER1_4 +0x4454 TX_FILTER1_5 +0x4458 TX_FILTER1_6 +0x445C TX_FILTER1_7 +0x4460 TX_FILTER1_8 +0x4464 TX_FILTER1_9 +0x4468 TX_FILTER1_10 +0x446C TX_FILTER1_11 +0x4470 TX_FILTER1_12 +0x4474 TX_FILTER1_13 +0x4478 TX_FILTER1_14 +0x447C TX_FILTER1_15 +0x4580 TX_CHROMA_KEY_0 +0x4584 TX_CHROMA_KEY_1 +0x4588 TX_CHROMA_KEY_2 +0x458C TX_CHROMA_KEY_3 +0x4590 TX_CHROMA_KEY_4 +0x4594 TX_CHROMA_KEY_5 +0x4598 TX_CHROMA_KEY_6 +0x459C TX_CHROMA_KEY_7 +0x45A0 TX_CHROMA_KEY_8 +0x45A4 TX_CHROMA_KEY_9 +0x45A8 TX_CHROMA_KEY_10 +0x45AC TX_CHROMA_KEY_11 +0x45B0 TX_CHROMA_KEY_12 +0x45B4 TX_CHROMA_KEY_13 +0x45B8 TX_CHROMA_KEY_14 +0x45BC TX_CHROMA_KEY_15 +0x45C0 TX_BORDER_COLOR_0 +0x45C4 TX_BORDER_COLOR_1 +0x45C8 TX_BORDER_COLOR_2 +0x45CC TX_BORDER_COLOR_3 +0x45D0 TX_BORDER_COLOR_4 +0x45D4 TX_BORDER_COLOR_5 +0x45D8 TX_BORDER_COLOR_6 +0x45DC TX_BORDER_COLOR_7 +0x45E0 TX_BORDER_COLOR_8 +0x45E4 TX_BORDER_COLOR_9 +0x45E8 TX_BORDER_COLOR_10 +0x45EC TX_BORDER_COLOR_11 +0x45F0 TX_BORDER_COLOR_12 +0x45F4 TX_BORDER_COLOR_13 +0x45F8 TX_BORDER_COLOR_14 +0x45FC TX_BORDER_COLOR_15 +0x4600 US_CONFIG +0x4604 US_PIXSIZE +0x4608 US_CODE_OFFSET +0x460C US_RESET +0x4610 US_CODE_ADDR_0 +0x4614 US_CODE_ADDR_1 +0x4618 US_CODE_ADDR_2 +0x461C US_CODE_ADDR_3 +0x4620 US_TEX_INST_0 +0x4624 US_TEX_INST_1 +0x4628 US_TEX_INST_2 +0x462C US_TEX_INST_3 +0x4630 US_TEX_INST_4 +0x4634 US_TEX_INST_5 +0x4638 US_TEX_INST_6 +0x463C US_TEX_INST_7 +0x4640 US_TEX_INST_8 +0x4644 US_TEX_INST_9 +0x4648 US_TEX_INST_10 +0x464C US_TEX_INST_11 +0x4650 US_TEX_INST_12 +0x4654 US_TEX_INST_13 +0x4658 US_TEX_INST_14 +0x465C US_TEX_INST_15 +0x4660 US_TEX_INST_16 +0x4664 US_TEX_INST_17 +0x4668 US_TEX_INST_18 +0x466C US_TEX_INST_19 +0x4670 US_TEX_INST_20 +0x4674 US_TEX_INST_21 +0x4678 US_TEX_INST_22 +0x467C US_TEX_INST_23 +0x4680 US_TEX_INST_24 +0x4684 US_TEX_INST_25 +0x4688 US_TEX_INST_26 +0x468C US_TEX_INST_27 +0x4690 US_TEX_INST_28 +0x4694 US_TEX_INST_29 +0x4698 US_TEX_INST_30 +0x469C US_TEX_INST_31 +0x46A4 US_OUT_FMT_0 +0x46A8 US_OUT_FMT_1 +0x46AC US_OUT_FMT_2 +0x46B0 US_OUT_FMT_3 +0x46B4 US_W_FMT +0x46B8 US_CODE_BANK +0x46BC US_CODE_EXT +0x46C0 US_ALU_RGB_ADDR_0 +0x46C4 US_ALU_RGB_ADDR_1 +0x46C8 US_ALU_RGB_ADDR_2 +0x46CC US_ALU_RGB_ADDR_3 +0x46D0 US_ALU_RGB_ADDR_4 +0x46D4 US_ALU_RGB_ADDR_5 +0x46D8 US_ALU_RGB_ADDR_6 +0x46DC US_ALU_RGB_ADDR_7 +0x46E0 US_ALU_RGB_ADDR_8 +0x46E4 US_ALU_RGB_ADDR_9 +0x46E8 US_ALU_RGB_ADDR_10 +0x46EC US_ALU_RGB_ADDR_11 +0x46F0 US_ALU_RGB_ADDR_12 +0x46F4 US_ALU_RGB_ADDR_13 +0x46F8 US_ALU_RGB_ADDR_14 +0x46FC US_ALU_RGB_ADDR_15 +0x4700 US_ALU_RGB_ADDR_16 +0x4704 US_ALU_RGB_ADDR_17 +0x4708 US_ALU_RGB_ADDR_18 +0x470C US_ALU_RGB_ADDR_19 +0x4710 US_ALU_RGB_ADDR_20 +0x4714 US_ALU_RGB_ADDR_21 +0x4718 US_ALU_RGB_ADDR_22 +0x471C US_ALU_RGB_ADDR_23 +0x4720 US_ALU_RGB_ADDR_24 +0x4724 US_ALU_RGB_ADDR_25 +0x4728 US_ALU_RGB_ADDR_26 +0x472C US_ALU_RGB_ADDR_27 +0x4730 US_ALU_RGB_ADDR_28 +0x4734 US_ALU_RGB_ADDR_29 +0x4738 US_ALU_RGB_ADDR_30 +0x473C US_ALU_RGB_ADDR_31 +0x4740 US_ALU_RGB_ADDR_32 +0x4744 US_ALU_RGB_ADDR_33 +0x4748 US_ALU_RGB_ADDR_34 +0x474C US_ALU_RGB_ADDR_35 +0x4750 US_ALU_RGB_ADDR_36 +0x4754 US_ALU_RGB_ADDR_37 +0x4758 US_ALU_RGB_ADDR_38 +0x475C US_ALU_RGB_ADDR_39 +0x4760 US_ALU_RGB_ADDR_40 +0x4764 US_ALU_RGB_ADDR_41 +0x4768 US_ALU_RGB_ADDR_42 +0x476C US_ALU_RGB_ADDR_43 +0x4770 US_ALU_RGB_ADDR_44 +0x4774 US_ALU_RGB_ADDR_45 +0x4778 US_ALU_RGB_ADDR_46 +0x477C US_ALU_RGB_ADDR_47 +0x4780 US_ALU_RGB_ADDR_48 +0x4784 US_ALU_RGB_ADDR_49 +0x4788 US_ALU_RGB_ADDR_50 +0x478C US_ALU_RGB_ADDR_51 +0x4790 US_ALU_RGB_ADDR_52 +0x4794 US_ALU_RGB_ADDR_53 +0x4798 US_ALU_RGB_ADDR_54 +0x479C US_ALU_RGB_ADDR_55 +0x47A0 US_ALU_RGB_ADDR_56 +0x47A4 US_ALU_RGB_ADDR_57 +0x47A8 US_ALU_RGB_ADDR_58 +0x47AC US_ALU_RGB_ADDR_59 +0x47B0 US_ALU_RGB_ADDR_60 +0x47B4 US_ALU_RGB_ADDR_61 +0x47B8 US_ALU_RGB_ADDR_62 +0x47BC US_ALU_RGB_ADDR_63 +0x47C0 US_ALU_ALPHA_ADDR_0 +0x47C4 US_ALU_ALPHA_ADDR_1 +0x47C8 US_ALU_ALPHA_ADDR_2 +0x47CC US_ALU_ALPHA_ADDR_3 +0x47D0 US_ALU_ALPHA_ADDR_4 +0x47D4 US_ALU_ALPHA_ADDR_5 +0x47D8 US_ALU_ALPHA_ADDR_6 +0x47DC US_ALU_ALPHA_ADDR_7 +0x47E0 US_ALU_ALPHA_ADDR_8 +0x47E4 US_ALU_ALPHA_ADDR_9 +0x47E8 US_ALU_ALPHA_ADDR_10 +0x47EC US_ALU_ALPHA_ADDR_11 +0x47F0 US_ALU_ALPHA_ADDR_12 +0x47F4 US_ALU_ALPHA_ADDR_13 +0x47F8 US_ALU_ALPHA_ADDR_14 +0x47FC US_ALU_ALPHA_ADDR_15 +0x4800 US_ALU_ALPHA_ADDR_16 +0x4804 US_ALU_ALPHA_ADDR_17 +0x4808 US_ALU_ALPHA_ADDR_18 +0x480C US_ALU_ALPHA_ADDR_19 +0x4810 US_ALU_ALPHA_ADDR_20 +0x4814 US_ALU_ALPHA_ADDR_21 +0x4818 US_ALU_ALPHA_ADDR_22 +0x481C US_ALU_ALPHA_ADDR_23 +0x4820 US_ALU_ALPHA_ADDR_24 +0x4824 US_ALU_ALPHA_ADDR_25 +0x4828 US_ALU_ALPHA_ADDR_26 +0x482C US_ALU_ALPHA_ADDR_27 +0x4830 US_ALU_ALPHA_ADDR_28 +0x4834 US_ALU_ALPHA_ADDR_29 +0x4838 US_ALU_ALPHA_ADDR_30 +0x483C US_ALU_ALPHA_ADDR_31 +0x4840 US_ALU_ALPHA_ADDR_32 +0x4844 US_ALU_ALPHA_ADDR_33 +0x4848 US_ALU_ALPHA_ADDR_34 +0x484C US_ALU_ALPHA_ADDR_35 +0x4850 US_ALU_ALPHA_ADDR_36 +0x4854 US_ALU_ALPHA_ADDR_37 +0x4858 US_ALU_ALPHA_ADDR_38 +0x485C US_ALU_ALPHA_ADDR_39 +0x4860 US_ALU_ALPHA_ADDR_40 +0x4864 US_ALU_ALPHA_ADDR_41 +0x4868 US_ALU_ALPHA_ADDR_42 +0x486C US_ALU_ALPHA_ADDR_43 +0x4870 US_ALU_ALPHA_ADDR_44 +0x4874 US_ALU_ALPHA_ADDR_45 +0x4878 US_ALU_ALPHA_ADDR_46 +0x487C US_ALU_ALPHA_ADDR_47 +0x4880 US_ALU_ALPHA_ADDR_48 +0x4884 US_ALU_ALPHA_ADDR_49 +0x4888 US_ALU_ALPHA_ADDR_50 +0x488C US_ALU_ALPHA_ADDR_51 +0x4890 US_ALU_ALPHA_ADDR_52 +0x4894 US_ALU_ALPHA_ADDR_53 +0x4898 US_ALU_ALPHA_ADDR_54 +0x489C US_ALU_ALPHA_ADDR_55 +0x48A0 US_ALU_ALPHA_ADDR_56 +0x48A4 US_ALU_ALPHA_ADDR_57 +0x48A8 US_ALU_ALPHA_ADDR_58 +0x48AC US_ALU_ALPHA_ADDR_59 +0x48B0 US_ALU_ALPHA_ADDR_60 +0x48B4 US_ALU_ALPHA_ADDR_61 +0x48B8 US_ALU_ALPHA_ADDR_62 +0x48BC US_ALU_ALPHA_ADDR_63 +0x48C0 US_ALU_RGB_INST_0 +0x48C4 US_ALU_RGB_INST_1 +0x48C8 US_ALU_RGB_INST_2 +0x48CC US_ALU_RGB_INST_3 +0x48D0 US_ALU_RGB_INST_4 +0x48D4 US_ALU_RGB_INST_5 +0x48D8 US_ALU_RGB_INST_6 +0x48DC US_ALU_RGB_INST_7 +0x48E0 US_ALU_RGB_INST_8 +0x48E4 US_ALU_RGB_INST_9 +0x48E8 US_ALU_RGB_INST_10 +0x48EC US_ALU_RGB_INST_11 +0x48F0 US_ALU_RGB_INST_12 +0x48F4 US_ALU_RGB_INST_13 +0x48F8 US_ALU_RGB_INST_14 +0x48FC US_ALU_RGB_INST_15 +0x4900 US_ALU_RGB_INST_16 +0x4904 US_ALU_RGB_INST_17 +0x4908 US_ALU_RGB_INST_18 +0x490C US_ALU_RGB_INST_19 +0x4910 US_ALU_RGB_INST_20 +0x4914 US_ALU_RGB_INST_21 +0x4918 US_ALU_RGB_INST_22 +0x491C US_ALU_RGB_INST_23 +0x4920 US_ALU_RGB_INST_24 +0x4924 US_ALU_RGB_INST_25 +0x4928 US_ALU_RGB_INST_26 +0x492C US_ALU_RGB_INST_27 +0x4930 US_ALU_RGB_INST_28 +0x4934 US_ALU_RGB_INST_29 +0x4938 US_ALU_RGB_INST_30 +0x493C US_ALU_RGB_INST_31 +0x4940 US_ALU_RGB_INST_32 +0x4944 US_ALU_RGB_INST_33 +0x4948 US_ALU_RGB_INST_34 +0x494C US_ALU_RGB_INST_35 +0x4950 US_ALU_RGB_INST_36 +0x4954 US_ALU_RGB_INST_37 +0x4958 US_ALU_RGB_INST_38 +0x495C US_ALU_RGB_INST_39 +0x4960 US_ALU_RGB_INST_40 +0x4964 US_ALU_RGB_INST_41 +0x4968 US_ALU_RGB_INST_42 +0x496C US_ALU_RGB_INST_43 +0x4970 US_ALU_RGB_INST_44 +0x4974 US_ALU_RGB_INST_45 +0x4978 US_ALU_RGB_INST_46 +0x497C US_ALU_RGB_INST_47 +0x4980 US_ALU_RGB_INST_48 +0x4984 US_ALU_RGB_INST_49 +0x4988 US_ALU_RGB_INST_50 +0x498C US_ALU_RGB_INST_51 +0x4990 US_ALU_RGB_INST_52 +0x4994 US_ALU_RGB_INST_53 +0x4998 US_ALU_RGB_INST_54 +0x499C US_ALU_RGB_INST_55 +0x49A0 US_ALU_RGB_INST_56 +0x49A4 US_ALU_RGB_INST_57 +0x49A8 US_ALU_RGB_INST_58 +0x49AC US_ALU_RGB_INST_59 +0x49B0 US_ALU_RGB_INST_60 +0x49B4 US_ALU_RGB_INST_61 +0x49B8 US_ALU_RGB_INST_62 +0x49BC US_ALU_RGB_INST_63 +0x49C0 US_ALU_ALPHA_INST_0 +0x49C4 US_ALU_ALPHA_INST_1 +0x49C8 US_ALU_ALPHA_INST_2 +0x49CC US_ALU_ALPHA_INST_3 +0x49D0 US_ALU_ALPHA_INST_4 +0x49D4 US_ALU_ALPHA_INST_5 +0x49D8 US_ALU_ALPHA_INST_6 +0x49DC US_ALU_ALPHA_INST_7 +0x49E0 US_ALU_ALPHA_INST_8 +0x49E4 US_ALU_ALPHA_INST_9 +0x49E8 US_ALU_ALPHA_INST_10 +0x49EC US_ALU_ALPHA_INST_11 +0x49F0 US_ALU_ALPHA_INST_12 +0x49F4 US_ALU_ALPHA_INST_13 +0x49F8 US_ALU_ALPHA_INST_14 +0x49FC US_ALU_ALPHA_INST_15 +0x4A00 US_ALU_ALPHA_INST_16 +0x4A04 US_ALU_ALPHA_INST_17 +0x4A08 US_ALU_ALPHA_INST_18 +0x4A0C US_ALU_ALPHA_INST_19 +0x4A10 US_ALU_ALPHA_INST_20 +0x4A14 US_ALU_ALPHA_INST_21 +0x4A18 US_ALU_ALPHA_INST_22 +0x4A1C US_ALU_ALPHA_INST_23 +0x4A20 US_ALU_ALPHA_INST_24 +0x4A24 US_ALU_ALPHA_INST_25 +0x4A28 US_ALU_ALPHA_INST_26 +0x4A2C US_ALU_ALPHA_INST_27 +0x4A30 US_ALU_ALPHA_INST_28 +0x4A34 US_ALU_ALPHA_INST_29 +0x4A38 US_ALU_ALPHA_INST_30 +0x4A3C US_ALU_ALPHA_INST_31 +0x4A40 US_ALU_ALPHA_INST_32 +0x4A44 US_ALU_ALPHA_INST_33 +0x4A48 US_ALU_ALPHA_INST_34 +0x4A4C US_ALU_ALPHA_INST_35 +0x4A50 US_ALU_ALPHA_INST_36 +0x4A54 US_ALU_ALPHA_INST_37 +0x4A58 US_ALU_ALPHA_INST_38 +0x4A5C US_ALU_ALPHA_INST_39 +0x4A60 US_ALU_ALPHA_INST_40 +0x4A64 US_ALU_ALPHA_INST_41 +0x4A68 US_ALU_ALPHA_INST_42 +0x4A6C US_ALU_ALPHA_INST_43 +0x4A70 US_ALU_ALPHA_INST_44 +0x4A74 US_ALU_ALPHA_INST_45 +0x4A78 US_ALU_ALPHA_INST_46 +0x4A7C US_ALU_ALPHA_INST_47 +0x4A80 US_ALU_ALPHA_INST_48 +0x4A84 US_ALU_ALPHA_INST_49 +0x4A88 US_ALU_ALPHA_INST_50 +0x4A8C US_ALU_ALPHA_INST_51 +0x4A90 US_ALU_ALPHA_INST_52 +0x4A94 US_ALU_ALPHA_INST_53 +0x4A98 US_ALU_ALPHA_INST_54 +0x4A9C US_ALU_ALPHA_INST_55 +0x4AA0 US_ALU_ALPHA_INST_56 +0x4AA4 US_ALU_ALPHA_INST_57 +0x4AA8 US_ALU_ALPHA_INST_58 +0x4AAC US_ALU_ALPHA_INST_59 +0x4AB0 US_ALU_ALPHA_INST_60 +0x4AB4 US_ALU_ALPHA_INST_61 +0x4AB8 US_ALU_ALPHA_INST_62 +0x4ABC US_ALU_ALPHA_INST_63 +0x4AC0 US_ALU_EXT_ADDR_0 +0x4AC4 US_ALU_EXT_ADDR_1 +0x4AC8 US_ALU_EXT_ADDR_2 +0x4ACC US_ALU_EXT_ADDR_3 +0x4AD0 US_ALU_EXT_ADDR_4 +0x4AD4 US_ALU_EXT_ADDR_5 +0x4AD8 US_ALU_EXT_ADDR_6 +0x4ADC US_ALU_EXT_ADDR_7 +0x4AE0 US_ALU_EXT_ADDR_8 +0x4AE4 US_ALU_EXT_ADDR_9 +0x4AE8 US_ALU_EXT_ADDR_10 +0x4AEC US_ALU_EXT_ADDR_11 +0x4AF0 US_ALU_EXT_ADDR_12 +0x4AF4 US_ALU_EXT_ADDR_13 +0x4AF8 US_ALU_EXT_ADDR_14 +0x4AFC US_ALU_EXT_ADDR_15 +0x4B00 US_ALU_EXT_ADDR_16 +0x4B04 US_ALU_EXT_ADDR_17 +0x4B08 US_ALU_EXT_ADDR_18 +0x4B0C US_ALU_EXT_ADDR_19 +0x4B10 US_ALU_EXT_ADDR_20 +0x4B14 US_ALU_EXT_ADDR_21 +0x4B18 US_ALU_EXT_ADDR_22 +0x4B1C US_ALU_EXT_ADDR_23 +0x4B20 US_ALU_EXT_ADDR_24 +0x4B24 US_ALU_EXT_ADDR_25 +0x4B28 US_ALU_EXT_ADDR_26 +0x4B2C US_ALU_EXT_ADDR_27 +0x4B30 US_ALU_EXT_ADDR_28 +0x4B34 US_ALU_EXT_ADDR_29 +0x4B38 US_ALU_EXT_ADDR_30 +0x4B3C US_ALU_EXT_ADDR_31 +0x4B40 US_ALU_EXT_ADDR_32 +0x4B44 US_ALU_EXT_ADDR_33 +0x4B48 US_ALU_EXT_ADDR_34 +0x4B4C US_ALU_EXT_ADDR_35 +0x4B50 US_ALU_EXT_ADDR_36 +0x4B54 US_ALU_EXT_ADDR_37 +0x4B58 US_ALU_EXT_ADDR_38 +0x4B5C US_ALU_EXT_ADDR_39 +0x4B60 US_ALU_EXT_ADDR_40 +0x4B64 US_ALU_EXT_ADDR_41 +0x4B68 US_ALU_EXT_ADDR_42 +0x4B6C US_ALU_EXT_ADDR_43 +0x4B70 US_ALU_EXT_ADDR_44 +0x4B74 US_ALU_EXT_ADDR_45 +0x4B78 US_ALU_EXT_ADDR_46 +0x4B7C US_ALU_EXT_ADDR_47 +0x4B80 US_ALU_EXT_ADDR_48 +0x4B84 US_ALU_EXT_ADDR_49 +0x4B88 US_ALU_EXT_ADDR_50 +0x4B8C US_ALU_EXT_ADDR_51 +0x4B90 US_ALU_EXT_ADDR_52 +0x4B94 US_ALU_EXT_ADDR_53 +0x4B98 US_ALU_EXT_ADDR_54 +0x4B9C US_ALU_EXT_ADDR_55 +0x4BA0 US_ALU_EXT_ADDR_56 +0x4BA4 US_ALU_EXT_ADDR_57 +0x4BA8 US_ALU_EXT_ADDR_58 +0x4BAC US_ALU_EXT_ADDR_59 +0x4BB0 US_ALU_EXT_ADDR_60 +0x4BB4 US_ALU_EXT_ADDR_61 +0x4BB8 US_ALU_EXT_ADDR_62 +0x4BBC US_ALU_EXT_ADDR_63 +0x4BC0 FG_FOG_BLEND +0x4BC4 FG_FOG_FACTOR +0x4BC8 FG_FOG_COLOR_R +0x4BCC FG_FOG_COLOR_G +0x4BD0 FG_FOG_COLOR_B +0x4BD4 FG_ALPHA_FUNC +0x4BD8 FG_DEPTH_SRC +0x4C00 US_ALU_CONST_R_0 +0x4C04 US_ALU_CONST_G_0 +0x4C08 US_ALU_CONST_B_0 +0x4C0C US_ALU_CONST_A_0 +0x4C10 US_ALU_CONST_R_1 +0x4C14 US_ALU_CONST_G_1 +0x4C18 US_ALU_CONST_B_1 +0x4C1C US_ALU_CONST_A_1 +0x4C20 US_ALU_CONST_R_2 +0x4C24 US_ALU_CONST_G_2 +0x4C28 US_ALU_CONST_B_2 +0x4C2C US_ALU_CONST_A_2 +0x4C30 US_ALU_CONST_R_3 +0x4C34 US_ALU_CONST_G_3 +0x4C38 US_ALU_CONST_B_3 +0x4C3C US_ALU_CONST_A_3 +0x4C40 US_ALU_CONST_R_4 +0x4C44 US_ALU_CONST_G_4 +0x4C48 US_ALU_CONST_B_4 +0x4C4C US_ALU_CONST_A_4 +0x4C50 US_ALU_CONST_R_5 +0x4C54 US_ALU_CONST_G_5 +0x4C58 US_ALU_CONST_B_5 +0x4C5C US_ALU_CONST_A_5 +0x4C60 US_ALU_CONST_R_6 +0x4C64 US_ALU_CONST_G_6 +0x4C68 US_ALU_CONST_B_6 +0x4C6C US_ALU_CONST_A_6 +0x4C70 US_ALU_CONST_R_7 +0x4C74 US_ALU_CONST_G_7 +0x4C78 US_ALU_CONST_B_7 +0x4C7C US_ALU_CONST_A_7 +0x4C80 US_ALU_CONST_R_8 +0x4C84 US_ALU_CONST_G_8 +0x4C88 US_ALU_CONST_B_8 +0x4C8C US_ALU_CONST_A_8 +0x4C90 US_ALU_CONST_R_9 +0x4C94 US_ALU_CONST_G_9 +0x4C98 US_ALU_CONST_B_9 +0x4C9C US_ALU_CONST_A_9 +0x4CA0 US_ALU_CONST_R_10 +0x4CA4 US_ALU_CONST_G_10 +0x4CA8 US_ALU_CONST_B_10 +0x4CAC US_ALU_CONST_A_10 +0x4CB0 US_ALU_CONST_R_11 +0x4CB4 US_ALU_CONST_G_11 +0x4CB8 US_ALU_CONST_B_11 +0x4CBC US_ALU_CONST_A_11 +0x4CC0 US_ALU_CONST_R_12 +0x4CC4 US_ALU_CONST_G_12 +0x4CC8 US_ALU_CONST_B_12 +0x4CCC US_ALU_CONST_A_12 +0x4CD0 US_ALU_CONST_R_13 +0x4CD4 US_ALU_CONST_G_13 +0x4CD8 US_ALU_CONST_B_13 +0x4CDC US_ALU_CONST_A_13 +0x4CE0 US_ALU_CONST_R_14 +0x4CE4 US_ALU_CONST_G_14 +0x4CE8 US_ALU_CONST_B_14 +0x4CEC US_ALU_CONST_A_14 +0x4CF0 US_ALU_CONST_R_15 +0x4CF4 US_ALU_CONST_G_15 +0x4CF8 US_ALU_CONST_B_15 +0x4CFC US_ALU_CONST_A_15 +0x4D00 US_ALU_CONST_R_16 +0x4D04 US_ALU_CONST_G_16 +0x4D08 US_ALU_CONST_B_16 +0x4D0C US_ALU_CONST_A_16 +0x4D10 US_ALU_CONST_R_17 +0x4D14 US_ALU_CONST_G_17 +0x4D18 US_ALU_CONST_B_17 +0x4D1C US_ALU_CONST_A_17 +0x4D20 US_ALU_CONST_R_18 +0x4D24 US_ALU_CONST_G_18 +0x4D28 US_ALU_CONST_B_18 +0x4D2C US_ALU_CONST_A_18 +0x4D30 US_ALU_CONST_R_19 +0x4D34 US_ALU_CONST_G_19 +0x4D38 US_ALU_CONST_B_19 +0x4D3C US_ALU_CONST_A_19 +0x4D40 US_ALU_CONST_R_20 +0x4D44 US_ALU_CONST_G_20 +0x4D48 US_ALU_CONST_B_20 +0x4D4C US_ALU_CONST_A_20 +0x4D50 US_ALU_CONST_R_21 +0x4D54 US_ALU_CONST_G_21 +0x4D58 US_ALU_CONST_B_21 +0x4D5C US_ALU_CONST_A_21 +0x4D60 US_ALU_CONST_R_22 +0x4D64 US_ALU_CONST_G_22 +0x4D68 US_ALU_CONST_B_22 +0x4D6C US_ALU_CONST_A_22 +0x4D70 US_ALU_CONST_R_23 +0x4D74 US_ALU_CONST_G_23 +0x4D78 US_ALU_CONST_B_23 +0x4D7C US_ALU_CONST_A_23 +0x4D80 US_ALU_CONST_R_24 +0x4D84 US_ALU_CONST_G_24 +0x4D88 US_ALU_CONST_B_24 +0x4D8C US_ALU_CONST_A_24 +0x4D90 US_ALU_CONST_R_25 +0x4D94 US_ALU_CONST_G_25 +0x4D98 US_ALU_CONST_B_25 +0x4D9C US_ALU_CONST_A_25 +0x4DA0 US_ALU_CONST_R_26 +0x4DA4 US_ALU_CONST_G_26 +0x4DA8 US_ALU_CONST_B_26 +0x4DAC US_ALU_CONST_A_26 +0x4DB0 US_ALU_CONST_R_27 +0x4DB4 US_ALU_CONST_G_27 +0x4DB8 US_ALU_CONST_B_27 +0x4DBC US_ALU_CONST_A_27 +0x4DC0 US_ALU_CONST_R_28 +0x4DC4 US_ALU_CONST_G_28 +0x4DC8 US_ALU_CONST_B_28 +0x4DCC US_ALU_CONST_A_28 +0x4DD0 US_ALU_CONST_R_29 +0x4DD4 US_ALU_CONST_G_29 +0x4DD8 US_ALU_CONST_B_29 +0x4DDC US_ALU_CONST_A_29 +0x4DE0 US_ALU_CONST_R_30 +0x4DE4 US_ALU_CONST_G_30 +0x4DE8 US_ALU_CONST_B_30 +0x4DEC US_ALU_CONST_A_30 +0x4DF0 US_ALU_CONST_R_31 +0x4DF4 US_ALU_CONST_G_31 +0x4DF8 US_ALU_CONST_B_31 +0x4DFC US_ALU_CONST_A_31 +0x4E08 RB3D_ABLENDCNTL_R3 +0x4E10 RB3D_CONSTANT_COLOR +0x4E14 RB3D_COLOR_CLEAR_VALUE +0x4E18 RB3D_ROPCNTL_R3 +0x4E1C RB3D_CLRCMP_FLIPE_R3 +0x4E20 RB3D_CLRCMP_CLR_R3 +0x4E24 RB3D_CLRCMP_MSK_R3 +0x4E48 RB3D_DEBUG_CTL +0x4E4C RB3D_DSTCACHE_CTLSTAT_R3 +0x4E50 RB3D_DITHER_CTL +0x4E54 RB3D_CMASK_OFFSET0 +0x4E58 RB3D_CMASK_OFFSET1 +0x4E5C RB3D_CMASK_OFFSET2 +0x4E60 RB3D_CMASK_OFFSET3 +0x4E64 RB3D_CMASK_PITCH0 +0x4E68 RB3D_CMASK_PITCH1 +0x4E6C RB3D_CMASK_PITCH2 +0x4E70 RB3D_CMASK_PITCH3 +0x4E74 RB3D_CMASK_WRINDEX +0x4E78 RB3D_CMASK_DWORD +0x4E7C RB3D_CMASK_RDINDEX +0x4EA0 RB3D_DISCARD_SRC_PIXEL_LTE_THRESHOLD +0x4EA4 RB3D_DISCARD_SRC_PIXEL_GTE_THRESHOLD +0x4F04 ZB_ZSTENCILCNTL +0x4F08 ZB_STENCILREFMASK +0x4F14 ZB_ZTOP +0x4F18 ZB_ZCACHE_CTLSTAT +0x4F28 ZB_DEPTHCLEARVALUE +0x4F58 ZB_ZPASS_DATA diff --git a/drivers/gpu/drm/radeon/reg_srcs/r600 b/drivers/gpu/drm/radeon/reg_srcs/r600 new file mode 100644 index 00000000..ea49752e --- /dev/null +++ b/drivers/gpu/drm/radeon/reg_srcs/r600 @@ -0,0 +1,762 @@ +r600 0x9400 +0x000287A0 R7xx_CB_SHADER_CONTROL +0x00028230 R7xx_PA_SC_EDGERULE +0x000286C8 R7xx_SPI_THREAD_GROUPING +0x00008D8C R7xx_SQ_DYN_GPR_CNTL_PS_FLUSH_REQ +0x000088C4 VGT_CACHE_INVALIDATION +0x00028A50 VGT_ENHANCE +0x000088CC VGT_ES_PER_GS +0x00028A2C VGT_GROUP_DECR +0x00028A28 VGT_GROUP_FIRST_DECR +0x00028A24 VGT_GROUP_PRIM_TYPE +0x00028A30 VGT_GROUP_VECT_0_CNTL +0x00028A38 VGT_GROUP_VECT_0_FMT_CNTL +0x00028A34 VGT_GROUP_VECT_1_CNTL +0x00028A3C VGT_GROUP_VECT_1_FMT_CNTL +0x00028A40 VGT_GS_MODE +0x00028A6C VGT_GS_OUT_PRIM_TYPE +0x000088C8 VGT_GS_PER_ES +0x000088E8 VGT_GS_PER_VS +0x000088D4 VGT_GS_VERTEX_REUSE +0x00028A14 VGT_HOS_CNTL +0x00028A18 VGT_HOS_MAX_TESS_LEVEL +0x00028A1C VGT_HOS_MIN_TESS_LEVEL +0x00028A20 VGT_HOS_REUSE_DEPTH +0x0000895C VGT_INDEX_TYPE +0x00028408 VGT_INDX_OFFSET +0x00028AA0 VGT_INSTANCE_STEP_RATE_0 +0x00028AA4 VGT_INSTANCE_STEP_RATE_1 +0x00028400 VGT_MAX_VTX_INDX +0x00028404 VGT_MIN_VTX_INDX +0x00028A94 VGT_MULTI_PRIM_IB_RESET_EN +0x0002840C VGT_MULTI_PRIM_IB_RESET_INDX +0x00008970 VGT_NUM_INDICES +0x00008974 VGT_NUM_INSTANCES +0x00028A10 VGT_OUTPUT_PATH_CNTL +0x00028A84 VGT_PRIMITIVEID_EN +0x00008958 VGT_PRIMITIVE_TYPE +0x00028AB4 VGT_REUSE_OFF +0x00028AB8 VGT_VTX_CNT_EN +0x000088B0 VGT_VTX_VECT_EJECT_REG +0x00028810 PA_CL_CLIP_CNTL +0x00008A14 PA_CL_ENHANCE +0x00028C14 PA_CL_GB_HORZ_CLIP_ADJ +0x00028C18 PA_CL_GB_HORZ_DISC_ADJ +0x00028C0C PA_CL_GB_VERT_CLIP_ADJ +0x00028C10 PA_CL_GB_VERT_DISC_ADJ +0x00028820 PA_CL_NANINF_CNTL +0x00028E1C PA_CL_POINT_CULL_RAD +0x00028E18 PA_CL_POINT_SIZE +0x00028E10 PA_CL_POINT_X_RAD +0x00028E14 PA_CL_POINT_Y_RAD +0x00028E2C PA_CL_UCP_0_W +0x00028E3C PA_CL_UCP_1_W +0x00028E4C PA_CL_UCP_2_W +0x00028E5C PA_CL_UCP_3_W +0x00028E6C PA_CL_UCP_4_W +0x00028E7C PA_CL_UCP_5_W +0x00028E20 PA_CL_UCP_0_X +0x00028E30 PA_CL_UCP_1_X +0x00028E40 PA_CL_UCP_2_X +0x00028E50 PA_CL_UCP_3_X +0x00028E60 PA_CL_UCP_4_X +0x00028E70 PA_CL_UCP_5_X +0x00028E24 PA_CL_UCP_0_Y +0x00028E34 PA_CL_UCP_1_Y +0x00028E44 PA_CL_UCP_2_Y +0x00028E54 PA_CL_UCP_3_Y +0x00028E64 PA_CL_UCP_4_Y +0x00028E74 PA_CL_UCP_5_Y +0x00028E28 PA_CL_UCP_0_Z +0x00028E38 PA_CL_UCP_1_Z +0x00028E48 PA_CL_UCP_2_Z +0x00028E58 PA_CL_UCP_3_Z +0x00028E68 PA_CL_UCP_4_Z +0x00028E78 PA_CL_UCP_5_Z +0x00028440 PA_CL_VPORT_XOFFSET_0 +0x00028458 PA_CL_VPORT_XOFFSET_1 +0x00028470 PA_CL_VPORT_XOFFSET_2 +0x00028488 PA_CL_VPORT_XOFFSET_3 +0x000284A0 PA_CL_VPORT_XOFFSET_4 +0x000284B8 PA_CL_VPORT_XOFFSET_5 +0x000284D0 PA_CL_VPORT_XOFFSET_6 +0x000284E8 PA_CL_VPORT_XOFFSET_7 +0x00028500 PA_CL_VPORT_XOFFSET_8 +0x00028518 PA_CL_VPORT_XOFFSET_9 +0x00028530 PA_CL_VPORT_XOFFSET_10 +0x00028548 PA_CL_VPORT_XOFFSET_11 +0x00028560 PA_CL_VPORT_XOFFSET_12 +0x00028578 PA_CL_VPORT_XOFFSET_13 +0x00028590 PA_CL_VPORT_XOFFSET_14 +0x000285A8 PA_CL_VPORT_XOFFSET_15 +0x0002843C PA_CL_VPORT_XSCALE_0 +0x00028454 PA_CL_VPORT_XSCALE_1 +0x0002846C PA_CL_VPORT_XSCALE_2 +0x00028484 PA_CL_VPORT_XSCALE_3 +0x0002849C PA_CL_VPORT_XSCALE_4 +0x000284B4 PA_CL_VPORT_XSCALE_5 +0x000284CC PA_CL_VPORT_XSCALE_6 +0x000284E4 PA_CL_VPORT_XSCALE_7 +0x000284FC PA_CL_VPORT_XSCALE_8 +0x00028514 PA_CL_VPORT_XSCALE_9 +0x0002852C PA_CL_VPORT_XSCALE_10 +0x00028544 PA_CL_VPORT_XSCALE_11 +0x0002855C PA_CL_VPORT_XSCALE_12 +0x00028574 PA_CL_VPORT_XSCALE_13 +0x0002858C PA_CL_VPORT_XSCALE_14 +0x000285A4 PA_CL_VPORT_XSCALE_15 +0x00028448 PA_CL_VPORT_YOFFSET_0 +0x00028460 PA_CL_VPORT_YOFFSET_1 +0x00028478 PA_CL_VPORT_YOFFSET_2 +0x00028490 PA_CL_VPORT_YOFFSET_3 +0x000284A8 PA_CL_VPORT_YOFFSET_4 +0x000284C0 PA_CL_VPORT_YOFFSET_5 +0x000284D8 PA_CL_VPORT_YOFFSET_6 +0x000284F0 PA_CL_VPORT_YOFFSET_7 +0x00028508 PA_CL_VPORT_YOFFSET_8 +0x00028520 PA_CL_VPORT_YOFFSET_9 +0x00028538 PA_CL_VPORT_YOFFSET_10 +0x00028550 PA_CL_VPORT_YOFFSET_11 +0x00028568 PA_CL_VPORT_YOFFSET_12 +0x00028580 PA_CL_VPORT_YOFFSET_13 +0x00028598 PA_CL_VPORT_YOFFSET_14 +0x000285B0 PA_CL_VPORT_YOFFSET_15 +0x00028444 PA_CL_VPORT_YSCALE_0 +0x0002845C PA_CL_VPORT_YSCALE_1 +0x00028474 PA_CL_VPORT_YSCALE_2 +0x0002848C PA_CL_VPORT_YSCALE_3 +0x000284A4 PA_CL_VPORT_YSCALE_4 +0x000284BC PA_CL_VPORT_YSCALE_5 +0x000284D4 PA_CL_VPORT_YSCALE_6 +0x000284EC PA_CL_VPORT_YSCALE_7 +0x00028504 PA_CL_VPORT_YSCALE_8 +0x0002851C PA_CL_VPORT_YSCALE_9 +0x00028534 PA_CL_VPORT_YSCALE_10 +0x0002854C PA_CL_VPORT_YSCALE_11 +0x00028564 PA_CL_VPORT_YSCALE_12 +0x0002857C PA_CL_VPORT_YSCALE_13 +0x00028594 PA_CL_VPORT_YSCALE_14 +0x000285AC PA_CL_VPORT_YSCALE_15 +0x00028450 PA_CL_VPORT_ZOFFSET_0 +0x00028468 PA_CL_VPORT_ZOFFSET_1 +0x00028480 PA_CL_VPORT_ZOFFSET_2 +0x00028498 PA_CL_VPORT_ZOFFSET_3 +0x000284B0 PA_CL_VPORT_ZOFFSET_4 +0x000284C8 PA_CL_VPORT_ZOFFSET_5 +0x000284E0 PA_CL_VPORT_ZOFFSET_6 +0x000284F8 PA_CL_VPORT_ZOFFSET_7 +0x00028510 PA_CL_VPORT_ZOFFSET_8 +0x00028528 PA_CL_VPORT_ZOFFSET_9 +0x00028540 PA_CL_VPORT_ZOFFSET_10 +0x00028558 PA_CL_VPORT_ZOFFSET_11 +0x00028570 PA_CL_VPORT_ZOFFSET_12 +0x00028588 PA_CL_VPORT_ZOFFSET_13 +0x000285A0 PA_CL_VPORT_ZOFFSET_14 +0x000285B8 PA_CL_VPORT_ZOFFSET_15 +0x0002844C PA_CL_VPORT_ZSCALE_0 +0x00028464 PA_CL_VPORT_ZSCALE_1 +0x0002847C PA_CL_VPORT_ZSCALE_2 +0x00028494 PA_CL_VPORT_ZSCALE_3 +0x000284AC PA_CL_VPORT_ZSCALE_4 +0x000284C4 PA_CL_VPORT_ZSCALE_5 +0x000284DC PA_CL_VPORT_ZSCALE_6 +0x000284F4 PA_CL_VPORT_ZSCALE_7 +0x0002850C PA_CL_VPORT_ZSCALE_8 +0x00028524 PA_CL_VPORT_ZSCALE_9 +0x0002853C PA_CL_VPORT_ZSCALE_10 +0x00028554 PA_CL_VPORT_ZSCALE_11 +0x0002856C PA_CL_VPORT_ZSCALE_12 +0x00028584 PA_CL_VPORT_ZSCALE_13 +0x0002859C PA_CL_VPORT_ZSCALE_14 +0x000285B4 PA_CL_VPORT_ZSCALE_15 +0x0002881C PA_CL_VS_OUT_CNTL +0x00028818 PA_CL_VTE_CNTL +0x00028C48 PA_SC_AA_MASK +0x00008B40 PA_SC_AA_SAMPLE_LOCS_2S +0x00008B44 PA_SC_AA_SAMPLE_LOCS_4S +0x00008B48 PA_SC_AA_SAMPLE_LOCS_8S_WD0 +0x00008B4C PA_SC_AA_SAMPLE_LOCS_8S_WD1 +0x00028C20 PA_SC_AA_SAMPLE_LOCS_8S_WD1_MCTX +0x00028C1C PA_SC_AA_SAMPLE_LOCS_MCTX +0x00028214 PA_SC_CLIPRECT_0_BR +0x0002821C PA_SC_CLIPRECT_1_BR +0x00028224 PA_SC_CLIPRECT_2_BR +0x0002822C PA_SC_CLIPRECT_3_BR +0x00028210 PA_SC_CLIPRECT_0_TL +0x00028218 PA_SC_CLIPRECT_1_TL +0x00028220 PA_SC_CLIPRECT_2_TL +0x00028228 PA_SC_CLIPRECT_3_TL +0x0002820C PA_SC_CLIPRECT_RULE +0x00008BF0 PA_SC_ENHANCE +0x00028244 PA_SC_GENERIC_SCISSOR_BR +0x00028240 PA_SC_GENERIC_SCISSOR_TL +0x00028C00 PA_SC_LINE_CNTL +0x00028A0C PA_SC_LINE_STIPPLE +0x00008B10 PA_SC_LINE_STIPPLE_STATE +0x00028A4C PA_SC_MODE_CNTL +0x00028A48 PA_SC_MPASS_PS_CNTL +0x00008B20 PA_SC_MULTI_CHIP_CNTL +0x00028034 PA_SC_SCREEN_SCISSOR_BR +0x00028030 PA_SC_SCREEN_SCISSOR_TL +0x00028254 PA_SC_VPORT_SCISSOR_0_BR +0x0002825C PA_SC_VPORT_SCISSOR_1_BR +0x00028264 PA_SC_VPORT_SCISSOR_2_BR +0x0002826C PA_SC_VPORT_SCISSOR_3_BR +0x00028274 PA_SC_VPORT_SCISSOR_4_BR +0x0002827C PA_SC_VPORT_SCISSOR_5_BR +0x00028284 PA_SC_VPORT_SCISSOR_6_BR +0x0002828C PA_SC_VPORT_SCISSOR_7_BR +0x00028294 PA_SC_VPORT_SCISSOR_8_BR +0x0002829C PA_SC_VPORT_SCISSOR_9_BR +0x000282A4 PA_SC_VPORT_SCISSOR_10_BR +0x000282AC PA_SC_VPORT_SCISSOR_11_BR +0x000282B4 PA_SC_VPORT_SCISSOR_12_BR +0x000282BC PA_SC_VPORT_SCISSOR_13_BR +0x000282C4 PA_SC_VPORT_SCISSOR_14_BR +0x000282CC PA_SC_VPORT_SCISSOR_15_BR +0x00028250 PA_SC_VPORT_SCISSOR_0_TL +0x00028258 PA_SC_VPORT_SCISSOR_1_TL +0x00028260 PA_SC_VPORT_SCISSOR_2_TL +0x00028268 PA_SC_VPORT_SCISSOR_3_TL +0x00028270 PA_SC_VPORT_SCISSOR_4_TL +0x00028278 PA_SC_VPORT_SCISSOR_5_TL +0x00028280 PA_SC_VPORT_SCISSOR_6_TL +0x00028288 PA_SC_VPORT_SCISSOR_7_TL +0x00028290 PA_SC_VPORT_SCISSOR_8_TL +0x00028298 PA_SC_VPORT_SCISSOR_9_TL +0x000282A0 PA_SC_VPORT_SCISSOR_10_TL +0x000282A8 PA_SC_VPORT_SCISSOR_11_TL +0x000282B0 PA_SC_VPORT_SCISSOR_12_TL +0x000282B8 PA_SC_VPORT_SCISSOR_13_TL +0x000282C0 PA_SC_VPORT_SCISSOR_14_TL +0x000282C8 PA_SC_VPORT_SCISSOR_15_TL +0x000282D4 PA_SC_VPORT_ZMAX_0 +0x000282DC PA_SC_VPORT_ZMAX_1 +0x000282E4 PA_SC_VPORT_ZMAX_2 +0x000282EC PA_SC_VPORT_ZMAX_3 +0x000282F4 PA_SC_VPORT_ZMAX_4 +0x000282FC PA_SC_VPORT_ZMAX_5 +0x00028304 PA_SC_VPORT_ZMAX_6 +0x0002830C PA_SC_VPORT_ZMAX_7 +0x00028314 PA_SC_VPORT_ZMAX_8 +0x0002831C PA_SC_VPORT_ZMAX_9 +0x00028324 PA_SC_VPORT_ZMAX_10 +0x0002832C PA_SC_VPORT_ZMAX_11 +0x00028334 PA_SC_VPORT_ZMAX_12 +0x0002833C PA_SC_VPORT_ZMAX_13 +0x00028344 PA_SC_VPORT_ZMAX_14 +0x0002834C PA_SC_VPORT_ZMAX_15 +0x000282D0 PA_SC_VPORT_ZMIN_0 +0x000282D8 PA_SC_VPORT_ZMIN_1 +0x000282E0 PA_SC_VPORT_ZMIN_2 +0x000282E8 PA_SC_VPORT_ZMIN_3 +0x000282F0 PA_SC_VPORT_ZMIN_4 +0x000282F8 PA_SC_VPORT_ZMIN_5 +0x00028300 PA_SC_VPORT_ZMIN_6 +0x00028308 PA_SC_VPORT_ZMIN_7 +0x00028310 PA_SC_VPORT_ZMIN_8 +0x00028318 PA_SC_VPORT_ZMIN_9 +0x00028320 PA_SC_VPORT_ZMIN_10 +0x00028328 PA_SC_VPORT_ZMIN_11 +0x00028330 PA_SC_VPORT_ZMIN_12 +0x00028338 PA_SC_VPORT_ZMIN_13 +0x00028340 PA_SC_VPORT_ZMIN_14 +0x00028348 PA_SC_VPORT_ZMIN_15 +0x00028200 PA_SC_WINDOW_OFFSET +0x00028208 PA_SC_WINDOW_SCISSOR_BR +0x00028204 PA_SC_WINDOW_SCISSOR_TL +0x00028A08 PA_SU_LINE_CNTL +0x00028A04 PA_SU_POINT_MINMAX +0x00028A00 PA_SU_POINT_SIZE +0x00028E0C PA_SU_POLY_OFFSET_BACK_OFFSET +0x00028E08 PA_SU_POLY_OFFSET_BACK_SCALE +0x00028DFC PA_SU_POLY_OFFSET_CLAMP +0x00028DF8 PA_SU_POLY_OFFSET_DB_FMT_CNTL +0x00028E04 PA_SU_POLY_OFFSET_FRONT_OFFSET +0x00028E00 PA_SU_POLY_OFFSET_FRONT_SCALE +0x00028814 PA_SU_SC_MODE_CNTL +0x00028C08 PA_SU_VTX_CNTL +0x00008C04 SQ_GPR_RESOURCE_MGMT_1 +0x00008C08 SQ_GPR_RESOURCE_MGMT_2 +0x00008C10 SQ_STACK_RESOURCE_MGMT_1 +0x00008C14 SQ_STACK_RESOURCE_MGMT_2 +0x00008C0C SQ_THREAD_RESOURCE_MGMT +0x00028380 SQ_VTX_SEMANTIC_0 +0x00028384 SQ_VTX_SEMANTIC_1 +0x00028388 SQ_VTX_SEMANTIC_2 +0x0002838C SQ_VTX_SEMANTIC_3 +0x00028390 SQ_VTX_SEMANTIC_4 +0x00028394 SQ_VTX_SEMANTIC_5 +0x00028398 SQ_VTX_SEMANTIC_6 +0x0002839C SQ_VTX_SEMANTIC_7 +0x000283A0 SQ_VTX_SEMANTIC_8 +0x000283A4 SQ_VTX_SEMANTIC_9 +0x000283A8 SQ_VTX_SEMANTIC_10 +0x000283AC SQ_VTX_SEMANTIC_11 +0x000283B0 SQ_VTX_SEMANTIC_12 +0x000283B4 SQ_VTX_SEMANTIC_13 +0x000283B8 SQ_VTX_SEMANTIC_14 +0x000283BC SQ_VTX_SEMANTIC_15 +0x000283C0 SQ_VTX_SEMANTIC_16 +0x000283C4 SQ_VTX_SEMANTIC_17 +0x000283C8 SQ_VTX_SEMANTIC_18 +0x000283CC SQ_VTX_SEMANTIC_19 +0x000283D0 SQ_VTX_SEMANTIC_20 +0x000283D4 SQ_VTX_SEMANTIC_21 +0x000283D8 SQ_VTX_SEMANTIC_22 +0x000283DC SQ_VTX_SEMANTIC_23 +0x000283E0 SQ_VTX_SEMANTIC_24 +0x000283E4 SQ_VTX_SEMANTIC_25 +0x000283E8 SQ_VTX_SEMANTIC_26 +0x000283EC SQ_VTX_SEMANTIC_27 +0x000283F0 SQ_VTX_SEMANTIC_28 +0x000283F4 SQ_VTX_SEMANTIC_29 +0x000283F8 SQ_VTX_SEMANTIC_30 +0x000283FC SQ_VTX_SEMANTIC_31 +0x000288E0 SQ_VTX_SEMANTIC_CLEAR +0x0003CFF4 SQ_VTX_START_INST_LOC +0x000281C0 SQ_ALU_CONST_BUFFER_SIZE_GS_0 +0x000281C4 SQ_ALU_CONST_BUFFER_SIZE_GS_1 +0x000281C8 SQ_ALU_CONST_BUFFER_SIZE_GS_2 +0x000281CC SQ_ALU_CONST_BUFFER_SIZE_GS_3 +0x000281D0 SQ_ALU_CONST_BUFFER_SIZE_GS_4 +0x000281D4 SQ_ALU_CONST_BUFFER_SIZE_GS_5 +0x000281D8 SQ_ALU_CONST_BUFFER_SIZE_GS_6 +0x000281DC SQ_ALU_CONST_BUFFER_SIZE_GS_7 +0x000281E0 SQ_ALU_CONST_BUFFER_SIZE_GS_8 +0x000281E4 SQ_ALU_CONST_BUFFER_SIZE_GS_9 +0x000281E8 SQ_ALU_CONST_BUFFER_SIZE_GS_10 +0x000281EC SQ_ALU_CONST_BUFFER_SIZE_GS_11 +0x000281F0 SQ_ALU_CONST_BUFFER_SIZE_GS_12 +0x000281F4 SQ_ALU_CONST_BUFFER_SIZE_GS_13 +0x000281F8 SQ_ALU_CONST_BUFFER_SIZE_GS_14 +0x000281FC SQ_ALU_CONST_BUFFER_SIZE_GS_15 +0x00028140 SQ_ALU_CONST_BUFFER_SIZE_PS_0 +0x00028144 SQ_ALU_CONST_BUFFER_SIZE_PS_1 +0x00028148 SQ_ALU_CONST_BUFFER_SIZE_PS_2 +0x0002814C SQ_ALU_CONST_BUFFER_SIZE_PS_3 +0x00028150 SQ_ALU_CONST_BUFFER_SIZE_PS_4 +0x00028154 SQ_ALU_CONST_BUFFER_SIZE_PS_5 +0x00028158 SQ_ALU_CONST_BUFFER_SIZE_PS_6 +0x0002815C SQ_ALU_CONST_BUFFER_SIZE_PS_7 +0x00028160 SQ_ALU_CONST_BUFFER_SIZE_PS_8 +0x00028164 SQ_ALU_CONST_BUFFER_SIZE_PS_9 +0x00028168 SQ_ALU_CONST_BUFFER_SIZE_PS_10 +0x0002816C SQ_ALU_CONST_BUFFER_SIZE_PS_11 +0x00028170 SQ_ALU_CONST_BUFFER_SIZE_PS_12 +0x00028174 SQ_ALU_CONST_BUFFER_SIZE_PS_13 +0x00028178 SQ_ALU_CONST_BUFFER_SIZE_PS_14 +0x0002817C SQ_ALU_CONST_BUFFER_SIZE_PS_15 +0x00028180 SQ_ALU_CONST_BUFFER_SIZE_VS_0 +0x00028184 SQ_ALU_CONST_BUFFER_SIZE_VS_1 +0x00028188 SQ_ALU_CONST_BUFFER_SIZE_VS_2 +0x0002818C SQ_ALU_CONST_BUFFER_SIZE_VS_3 +0x00028190 SQ_ALU_CONST_BUFFER_SIZE_VS_4 +0x00028194 SQ_ALU_CONST_BUFFER_SIZE_VS_5 +0x00028198 SQ_ALU_CONST_BUFFER_SIZE_VS_6 +0x0002819C SQ_ALU_CONST_BUFFER_SIZE_VS_7 +0x000281A0 SQ_ALU_CONST_BUFFER_SIZE_VS_8 +0x000281A4 SQ_ALU_CONST_BUFFER_SIZE_VS_9 +0x000281A8 SQ_ALU_CONST_BUFFER_SIZE_VS_10 +0x000281AC SQ_ALU_CONST_BUFFER_SIZE_VS_11 +0x000281B0 SQ_ALU_CONST_BUFFER_SIZE_VS_12 +0x000281B4 SQ_ALU_CONST_BUFFER_SIZE_VS_13 +0x000281B8 SQ_ALU_CONST_BUFFER_SIZE_VS_14 +0x000281BC SQ_ALU_CONST_BUFFER_SIZE_VS_15 +0x000288D8 SQ_PGM_CF_OFFSET_ES +0x000288DC SQ_PGM_CF_OFFSET_FS +0x000288D4 SQ_PGM_CF_OFFSET_GS +0x000288CC SQ_PGM_CF_OFFSET_PS +0x000288D0 SQ_PGM_CF_OFFSET_VS +0x00028854 SQ_PGM_EXPORTS_PS +0x00028890 SQ_PGM_RESOURCES_ES +0x000288A4 SQ_PGM_RESOURCES_FS +0x0002887C SQ_PGM_RESOURCES_GS +0x00028850 SQ_PGM_RESOURCES_PS +0x00028868 SQ_PGM_RESOURCES_VS +0x00009100 SPI_CONFIG_CNTL +0x0000913C SPI_CONFIG_CNTL_1 +0x000286DC SPI_FOG_CNTL +0x000286E4 SPI_FOG_FUNC_BIAS +0x000286E0 SPI_FOG_FUNC_SCALE +0x000286D8 SPI_INPUT_Z +0x000286D4 SPI_INTERP_CONTROL_0 +0x00028644 SPI_PS_INPUT_CNTL_0 +0x00028648 SPI_PS_INPUT_CNTL_1 +0x0002864C SPI_PS_INPUT_CNTL_2 +0x00028650 SPI_PS_INPUT_CNTL_3 +0x00028654 SPI_PS_INPUT_CNTL_4 +0x00028658 SPI_PS_INPUT_CNTL_5 +0x0002865C SPI_PS_INPUT_CNTL_6 +0x00028660 SPI_PS_INPUT_CNTL_7 +0x00028664 SPI_PS_INPUT_CNTL_8 +0x00028668 SPI_PS_INPUT_CNTL_9 +0x0002866C SPI_PS_INPUT_CNTL_10 +0x00028670 SPI_PS_INPUT_CNTL_11 +0x00028674 SPI_PS_INPUT_CNTL_12 +0x00028678 SPI_PS_INPUT_CNTL_13 +0x0002867C SPI_PS_INPUT_CNTL_14 +0x00028680 SPI_PS_INPUT_CNTL_15 +0x00028684 SPI_PS_INPUT_CNTL_16 +0x00028688 SPI_PS_INPUT_CNTL_17 +0x0002868C SPI_PS_INPUT_CNTL_18 +0x00028690 SPI_PS_INPUT_CNTL_19 +0x00028694 SPI_PS_INPUT_CNTL_20 +0x00028698 SPI_PS_INPUT_CNTL_21 +0x0002869C SPI_PS_INPUT_CNTL_22 +0x000286A0 SPI_PS_INPUT_CNTL_23 +0x000286A4 SPI_PS_INPUT_CNTL_24 +0x000286A8 SPI_PS_INPUT_CNTL_25 +0x000286AC SPI_PS_INPUT_CNTL_26 +0x000286B0 SPI_PS_INPUT_CNTL_27 +0x000286B4 SPI_PS_INPUT_CNTL_28 +0x000286B8 SPI_PS_INPUT_CNTL_29 +0x000286BC SPI_PS_INPUT_CNTL_30 +0x000286C0 SPI_PS_INPUT_CNTL_31 +0x000286CC SPI_PS_IN_CONTROL_0 +0x000286D0 SPI_PS_IN_CONTROL_1 +0x000286C4 SPI_VS_OUT_CONFIG +0x00028614 SPI_VS_OUT_ID_0 +0x00028618 SPI_VS_OUT_ID_1 +0x0002861C SPI_VS_OUT_ID_2 +0x00028620 SPI_VS_OUT_ID_3 +0x00028624 SPI_VS_OUT_ID_4 +0x00028628 SPI_VS_OUT_ID_5 +0x0002862C SPI_VS_OUT_ID_6 +0x00028630 SPI_VS_OUT_ID_7 +0x00028634 SPI_VS_OUT_ID_8 +0x00028638 SPI_VS_OUT_ID_9 +0x00028438 SX_ALPHA_REF +0x00028410 SX_ALPHA_TEST_CONTROL +0x00028350 SX_MISC +0x00009604 TC_INVALIDATE +0x00009400 TD_FILTER4 +0x00009404 TD_FILTER4_1 +0x00009408 TD_FILTER4_2 +0x0000940C TD_FILTER4_3 +0x00009410 TD_FILTER4_4 +0x00009414 TD_FILTER4_5 +0x00009418 TD_FILTER4_6 +0x0000941C TD_FILTER4_7 +0x00009420 TD_FILTER4_8 +0x00009424 TD_FILTER4_9 +0x00009428 TD_FILTER4_10 +0x0000942C TD_FILTER4_11 +0x00009430 TD_FILTER4_12 +0x00009434 TD_FILTER4_13 +0x00009438 TD_FILTER4_14 +0x0000943C TD_FILTER4_15 +0x00009440 TD_FILTER4_16 +0x00009444 TD_FILTER4_17 +0x00009448 TD_FILTER4_18 +0x0000944C TD_FILTER4_19 +0x00009450 TD_FILTER4_20 +0x00009454 TD_FILTER4_21 +0x00009458 TD_FILTER4_22 +0x0000945C TD_FILTER4_23 +0x00009460 TD_FILTER4_24 +0x00009464 TD_FILTER4_25 +0x00009468 TD_FILTER4_26 +0x0000946C TD_FILTER4_27 +0x00009470 TD_FILTER4_28 +0x00009474 TD_FILTER4_29 +0x00009478 TD_FILTER4_30 +0x0000947C TD_FILTER4_31 +0x00009480 TD_FILTER4_32 +0x00009484 TD_FILTER4_33 +0x00009488 TD_FILTER4_34 +0x0000948C TD_FILTER4_35 +0x0000A80C TD_GS_SAMPLER0_BORDER_ALPHA +0x0000A81C TD_GS_SAMPLER1_BORDER_ALPHA +0x0000A82C TD_GS_SAMPLER2_BORDER_ALPHA +0x0000A83C TD_GS_SAMPLER3_BORDER_ALPHA +0x0000A84C TD_GS_SAMPLER4_BORDER_ALPHA +0x0000A85C TD_GS_SAMPLER5_BORDER_ALPHA +0x0000A86C TD_GS_SAMPLER6_BORDER_ALPHA +0x0000A87C TD_GS_SAMPLER7_BORDER_ALPHA +0x0000A88C TD_GS_SAMPLER8_BORDER_ALPHA +0x0000A89C TD_GS_SAMPLER9_BORDER_ALPHA +0x0000A8AC TD_GS_SAMPLER10_BORDER_ALPHA +0x0000A8BC TD_GS_SAMPLER11_BORDER_ALPHA +0x0000A8CC TD_GS_SAMPLER12_BORDER_ALPHA +0x0000A8DC TD_GS_SAMPLER13_BORDER_ALPHA +0x0000A8EC TD_GS_SAMPLER14_BORDER_ALPHA +0x0000A8FC TD_GS_SAMPLER15_BORDER_ALPHA +0x0000A90C TD_GS_SAMPLER16_BORDER_ALPHA +0x0000A91C TD_GS_SAMPLER17_BORDER_ALPHA +0x0000A808 TD_GS_SAMPLER0_BORDER_BLUE +0x0000A818 TD_GS_SAMPLER1_BORDER_BLUE +0x0000A828 TD_GS_SAMPLER2_BORDER_BLUE +0x0000A838 TD_GS_SAMPLER3_BORDER_BLUE +0x0000A848 TD_GS_SAMPLER4_BORDER_BLUE +0x0000A858 TD_GS_SAMPLER5_BORDER_BLUE +0x0000A868 TD_GS_SAMPLER6_BORDER_BLUE +0x0000A878 TD_GS_SAMPLER7_BORDER_BLUE +0x0000A888 TD_GS_SAMPLER8_BORDER_BLUE +0x0000A898 TD_GS_SAMPLER9_BORDER_BLUE +0x0000A8A8 TD_GS_SAMPLER10_BORDER_BLUE +0x0000A8B8 TD_GS_SAMPLER11_BORDER_BLUE +0x0000A8C8 TD_GS_SAMPLER12_BORDER_BLUE +0x0000A8D8 TD_GS_SAMPLER13_BORDER_BLUE +0x0000A8E8 TD_GS_SAMPLER14_BORDER_BLUE +0x0000A8F8 TD_GS_SAMPLER15_BORDER_BLUE +0x0000A908 TD_GS_SAMPLER16_BORDER_BLUE +0x0000A918 TD_GS_SAMPLER17_BORDER_BLUE +0x0000A804 TD_GS_SAMPLER0_BORDER_GREEN +0x0000A814 TD_GS_SAMPLER1_BORDER_GREEN +0x0000A824 TD_GS_SAMPLER2_BORDER_GREEN +0x0000A834 TD_GS_SAMPLER3_BORDER_GREEN +0x0000A844 TD_GS_SAMPLER4_BORDER_GREEN +0x0000A854 TD_GS_SAMPLER5_BORDER_GREEN +0x0000A864 TD_GS_SAMPLER6_BORDER_GREEN +0x0000A874 TD_GS_SAMPLER7_BORDER_GREEN +0x0000A884 TD_GS_SAMPLER8_BORDER_GREEN +0x0000A894 TD_GS_SAMPLER9_BORDER_GREEN +0x0000A8A4 TD_GS_SAMPLER10_BORDER_GREEN +0x0000A8B4 TD_GS_SAMPLER11_BORDER_GREEN +0x0000A8C4 TD_GS_SAMPLER12_BORDER_GREEN +0x0000A8D4 TD_GS_SAMPLER13_BORDER_GREEN +0x0000A8E4 TD_GS_SAMPLER14_BORDER_GREEN +0x0000A8F4 TD_GS_SAMPLER15_BORDER_GREEN +0x0000A904 TD_GS_SAMPLER16_BORDER_GREEN +0x0000A914 TD_GS_SAMPLER17_BORDER_GREEN +0x0000A800 TD_GS_SAMPLER0_BORDER_RED +0x0000A810 TD_GS_SAMPLER1_BORDER_RED +0x0000A820 TD_GS_SAMPLER2_BORDER_RED +0x0000A830 TD_GS_SAMPLER3_BORDER_RED +0x0000A840 TD_GS_SAMPLER4_BORDER_RED +0x0000A850 TD_GS_SAMPLER5_BORDER_RED +0x0000A860 TD_GS_SAMPLER6_BORDER_RED +0x0000A870 TD_GS_SAMPLER7_BORDER_RED +0x0000A880 TD_GS_SAMPLER8_BORDER_RED +0x0000A890 TD_GS_SAMPLER9_BORDER_RED +0x0000A8A0 TD_GS_SAMPLER10_BORDER_RED +0x0000A8B0 TD_GS_SAMPLER11_BORDER_RED +0x0000A8C0 TD_GS_SAMPLER12_BORDER_RED +0x0000A8D0 TD_GS_SAMPLER13_BORDER_RED +0x0000A8E0 TD_GS_SAMPLER14_BORDER_RED +0x0000A8F0 TD_GS_SAMPLER15_BORDER_RED +0x0000A900 TD_GS_SAMPLER16_BORDER_RED +0x0000A910 TD_GS_SAMPLER17_BORDER_RED +0x0000A40C TD_PS_SAMPLER0_BORDER_ALPHA +0x0000A41C TD_PS_SAMPLER1_BORDER_ALPHA +0x0000A42C TD_PS_SAMPLER2_BORDER_ALPHA +0x0000A43C TD_PS_SAMPLER3_BORDER_ALPHA +0x0000A44C TD_PS_SAMPLER4_BORDER_ALPHA +0x0000A45C TD_PS_SAMPLER5_BORDER_ALPHA +0x0000A46C TD_PS_SAMPLER6_BORDER_ALPHA +0x0000A47C TD_PS_SAMPLER7_BORDER_ALPHA +0x0000A48C TD_PS_SAMPLER8_BORDER_ALPHA +0x0000A49C TD_PS_SAMPLER9_BORDER_ALPHA +0x0000A4AC TD_PS_SAMPLER10_BORDER_ALPHA +0x0000A4BC TD_PS_SAMPLER11_BORDER_ALPHA +0x0000A4CC TD_PS_SAMPLER12_BORDER_ALPHA +0x0000A4DC TD_PS_SAMPLER13_BORDER_ALPHA +0x0000A4EC TD_PS_SAMPLER14_BORDER_ALPHA +0x0000A4FC TD_PS_SAMPLER15_BORDER_ALPHA +0x0000A50C TD_PS_SAMPLER16_BORDER_ALPHA +0x0000A51C TD_PS_SAMPLER17_BORDER_ALPHA +0x0000A408 TD_PS_SAMPLER0_BORDER_BLUE +0x0000A418 TD_PS_SAMPLER1_BORDER_BLUE +0x0000A428 TD_PS_SAMPLER2_BORDER_BLUE +0x0000A438 TD_PS_SAMPLER3_BORDER_BLUE +0x0000A448 TD_PS_SAMPLER4_BORDER_BLUE +0x0000A458 TD_PS_SAMPLER5_BORDER_BLUE +0x0000A468 TD_PS_SAMPLER6_BORDER_BLUE +0x0000A478 TD_PS_SAMPLER7_BORDER_BLUE +0x0000A488 TD_PS_SAMPLER8_BORDER_BLUE +0x0000A498 TD_PS_SAMPLER9_BORDER_BLUE +0x0000A4A8 TD_PS_SAMPLER10_BORDER_BLUE +0x0000A4B8 TD_PS_SAMPLER11_BORDER_BLUE +0x0000A4C8 TD_PS_SAMPLER12_BORDER_BLUE +0x0000A4D8 TD_PS_SAMPLER13_BORDER_BLUE +0x0000A4E8 TD_PS_SAMPLER14_BORDER_BLUE +0x0000A4F8 TD_PS_SAMPLER15_BORDER_BLUE +0x0000A508 TD_PS_SAMPLER16_BORDER_BLUE +0x0000A518 TD_PS_SAMPLER17_BORDER_BLUE +0x0000A404 TD_PS_SAMPLER0_BORDER_GREEN +0x0000A414 TD_PS_SAMPLER1_BORDER_GREEN +0x0000A424 TD_PS_SAMPLER2_BORDER_GREEN +0x0000A434 TD_PS_SAMPLER3_BORDER_GREEN +0x0000A444 TD_PS_SAMPLER4_BORDER_GREEN +0x0000A454 TD_PS_SAMPLER5_BORDER_GREEN +0x0000A464 TD_PS_SAMPLER6_BORDER_GREEN +0x0000A474 TD_PS_SAMPLER7_BORDER_GREEN +0x0000A484 TD_PS_SAMPLER8_BORDER_GREEN +0x0000A494 TD_PS_SAMPLER9_BORDER_GREEN +0x0000A4A4 TD_PS_SAMPLER10_BORDER_GREEN +0x0000A4B4 TD_PS_SAMPLER11_BORDER_GREEN +0x0000A4C4 TD_PS_SAMPLER12_BORDER_GREEN +0x0000A4D4 TD_PS_SAMPLER13_BORDER_GREEN +0x0000A4E4 TD_PS_SAMPLER14_BORDER_GREEN +0x0000A4F4 TD_PS_SAMPLER15_BORDER_GREEN +0x0000A504 TD_PS_SAMPLER16_BORDER_GREEN +0x0000A514 TD_PS_SAMPLER17_BORDER_GREEN +0x0000A400 TD_PS_SAMPLER0_BORDER_RED +0x0000A410 TD_PS_SAMPLER1_BORDER_RED +0x0000A420 TD_PS_SAMPLER2_BORDER_RED +0x0000A430 TD_PS_SAMPLER3_BORDER_RED +0x0000A440 TD_PS_SAMPLER4_BORDER_RED +0x0000A450 TD_PS_SAMPLER5_BORDER_RED +0x0000A460 TD_PS_SAMPLER6_BORDER_RED +0x0000A470 TD_PS_SAMPLER7_BORDER_RED +0x0000A480 TD_PS_SAMPLER8_BORDER_RED +0x0000A490 TD_PS_SAMPLER9_BORDER_RED +0x0000A4A0 TD_PS_SAMPLER10_BORDER_RED +0x0000A4B0 TD_PS_SAMPLER11_BORDER_RED +0x0000A4C0 TD_PS_SAMPLER12_BORDER_RED +0x0000A4D0 TD_PS_SAMPLER13_BORDER_RED +0x0000A4E0 TD_PS_SAMPLER14_BORDER_RED +0x0000A4F0 TD_PS_SAMPLER15_BORDER_RED +0x0000A500 TD_PS_SAMPLER16_BORDER_RED +0x0000A510 TD_PS_SAMPLER17_BORDER_RED +0x0000AA00 TD_PS_SAMPLER0_CLEARTYPE_KERNEL +0x0000AA04 TD_PS_SAMPLER1_CLEARTYPE_KERNEL +0x0000AA08 TD_PS_SAMPLER2_CLEARTYPE_KERNEL +0x0000AA0C TD_PS_SAMPLER3_CLEARTYPE_KERNEL +0x0000AA10 TD_PS_SAMPLER4_CLEARTYPE_KERNEL +0x0000AA14 TD_PS_SAMPLER5_CLEARTYPE_KERNEL +0x0000AA18 TD_PS_SAMPLER6_CLEARTYPE_KERNEL +0x0000AA1C TD_PS_SAMPLER7_CLEARTYPE_KERNEL +0x0000AA20 TD_PS_SAMPLER8_CLEARTYPE_KERNEL +0x0000AA24 TD_PS_SAMPLER9_CLEARTYPE_KERNEL +0x0000AA28 TD_PS_SAMPLER10_CLEARTYPE_KERNEL +0x0000AA2C TD_PS_SAMPLER11_CLEARTYPE_KERNEL +0x0000AA30 TD_PS_SAMPLER12_CLEARTYPE_KERNEL +0x0000AA34 TD_PS_SAMPLER13_CLEARTYPE_KERNEL +0x0000AA38 TD_PS_SAMPLER14_CLEARTYPE_KERNEL +0x0000AA3C TD_PS_SAMPLER15_CLEARTYPE_KERNEL +0x0000AA40 TD_PS_SAMPLER16_CLEARTYPE_KERNEL +0x0000AA44 TD_PS_SAMPLER17_CLEARTYPE_KERNEL +0x0000A60C TD_VS_SAMPLER0_BORDER_ALPHA +0x0000A61C TD_VS_SAMPLER1_BORDER_ALPHA +0x0000A62C TD_VS_SAMPLER2_BORDER_ALPHA +0x0000A63C TD_VS_SAMPLER3_BORDER_ALPHA +0x0000A64C TD_VS_SAMPLER4_BORDER_ALPHA +0x0000A65C TD_VS_SAMPLER5_BORDER_ALPHA +0x0000A66C TD_VS_SAMPLER6_BORDER_ALPHA +0x0000A67C TD_VS_SAMPLER7_BORDER_ALPHA +0x0000A68C TD_VS_SAMPLER8_BORDER_ALPHA +0x0000A69C TD_VS_SAMPLER9_BORDER_ALPHA +0x0000A6AC TD_VS_SAMPLER10_BORDER_ALPHA +0x0000A6BC TD_VS_SAMPLER11_BORDER_ALPHA +0x0000A6CC TD_VS_SAMPLER12_BORDER_ALPHA +0x0000A6DC TD_VS_SAMPLER13_BORDER_ALPHA +0x0000A6EC TD_VS_SAMPLER14_BORDER_ALPHA +0x0000A6FC TD_VS_SAMPLER15_BORDER_ALPHA +0x0000A70C TD_VS_SAMPLER16_BORDER_ALPHA +0x0000A71C TD_VS_SAMPLER17_BORDER_ALPHA +0x0000A608 TD_VS_SAMPLER0_BORDER_BLUE +0x0000A618 TD_VS_SAMPLER1_BORDER_BLUE +0x0000A628 TD_VS_SAMPLER2_BORDER_BLUE +0x0000A638 TD_VS_SAMPLER3_BORDER_BLUE +0x0000A648 TD_VS_SAMPLER4_BORDER_BLUE +0x0000A658 TD_VS_SAMPLER5_BORDER_BLUE +0x0000A668 TD_VS_SAMPLER6_BORDER_BLUE +0x0000A678 TD_VS_SAMPLER7_BORDER_BLUE +0x0000A688 TD_VS_SAMPLER8_BORDER_BLUE +0x0000A698 TD_VS_SAMPLER9_BORDER_BLUE +0x0000A6A8 TD_VS_SAMPLER10_BORDER_BLUE +0x0000A6B8 TD_VS_SAMPLER11_BORDER_BLUE +0x0000A6C8 TD_VS_SAMPLER12_BORDER_BLUE +0x0000A6D8 TD_VS_SAMPLER13_BORDER_BLUE +0x0000A6E8 TD_VS_SAMPLER14_BORDER_BLUE +0x0000A6F8 TD_VS_SAMPLER15_BORDER_BLUE +0x0000A708 TD_VS_SAMPLER16_BORDER_BLUE +0x0000A718 TD_VS_SAMPLER17_BORDER_BLUE +0x0000A604 TD_VS_SAMPLER0_BORDER_GREEN +0x0000A614 TD_VS_SAMPLER1_BORDER_GREEN +0x0000A624 TD_VS_SAMPLER2_BORDER_GREEN +0x0000A634 TD_VS_SAMPLER3_BORDER_GREEN +0x0000A644 TD_VS_SAMPLER4_BORDER_GREEN +0x0000A654 TD_VS_SAMPLER5_BORDER_GREEN +0x0000A664 TD_VS_SAMPLER6_BORDER_GREEN +0x0000A674 TD_VS_SAMPLER7_BORDER_GREEN +0x0000A684 TD_VS_SAMPLER8_BORDER_GREEN +0x0000A694 TD_VS_SAMPLER9_BORDER_GREEN +0x0000A6A4 TD_VS_SAMPLER10_BORDER_GREEN +0x0000A6B4 TD_VS_SAMPLER11_BORDER_GREEN +0x0000A6C4 TD_VS_SAMPLER12_BORDER_GREEN +0x0000A6D4 TD_VS_SAMPLER13_BORDER_GREEN +0x0000A6E4 TD_VS_SAMPLER14_BORDER_GREEN +0x0000A6F4 TD_VS_SAMPLER15_BORDER_GREEN +0x0000A704 TD_VS_SAMPLER16_BORDER_GREEN +0x0000A714 TD_VS_SAMPLER17_BORDER_GREEN +0x0000A600 TD_VS_SAMPLER0_BORDER_RED +0x0000A610 TD_VS_SAMPLER1_BORDER_RED +0x0000A620 TD_VS_SAMPLER2_BORDER_RED +0x0000A630 TD_VS_SAMPLER3_BORDER_RED +0x0000A640 TD_VS_SAMPLER4_BORDER_RED +0x0000A650 TD_VS_SAMPLER5_BORDER_RED +0x0000A660 TD_VS_SAMPLER6_BORDER_RED +0x0000A670 TD_VS_SAMPLER7_BORDER_RED +0x0000A680 TD_VS_SAMPLER8_BORDER_RED +0x0000A690 TD_VS_SAMPLER9_BORDER_RED +0x0000A6A0 TD_VS_SAMPLER10_BORDER_RED +0x0000A6B0 TD_VS_SAMPLER11_BORDER_RED +0x0000A6C0 TD_VS_SAMPLER12_BORDER_RED +0x0000A6D0 TD_VS_SAMPLER13_BORDER_RED +0x0000A6E0 TD_VS_SAMPLER14_BORDER_RED +0x0000A6F0 TD_VS_SAMPLER15_BORDER_RED +0x0000A700 TD_VS_SAMPLER16_BORDER_RED +0x0000A710 TD_VS_SAMPLER17_BORDER_RED +0x00009508 TA_CNTL_AUX +0x0002802C DB_DEPTH_CLEAR +0x00028D24 DB_HTILE_SURFACE +0x00028D34 DB_PREFETCH_LIMIT +0x00028D30 DB_PRELOAD_CONTROL +0x00028D0C DB_RENDER_CONTROL +0x00028D10 DB_RENDER_OVERRIDE +0x0002880C DB_SHADER_CONTROL +0x00028D28 DB_SRESULTS_COMPARE_STATE0 +0x00028D2C DB_SRESULTS_COMPARE_STATE1 +0x00028430 DB_STENCILREFMASK +0x00028434 DB_STENCILREFMASK_BF +0x00028028 DB_STENCIL_CLEAR +0x00028780 CB_BLEND0_CONTROL +0x00028784 CB_BLEND1_CONTROL +0x00028788 CB_BLEND2_CONTROL +0x0002878C CB_BLEND3_CONTROL +0x00028790 CB_BLEND4_CONTROL +0x00028794 CB_BLEND5_CONTROL +0x00028798 CB_BLEND6_CONTROL +0x0002879C CB_BLEND7_CONTROL +0x00028804 CB_BLEND_CONTROL +0x00028420 CB_BLEND_ALPHA +0x0002841C CB_BLEND_BLUE +0x00028418 CB_BLEND_GREEN +0x00028414 CB_BLEND_RED +0x0002812C CB_CLEAR_ALPHA +0x00028128 CB_CLEAR_BLUE +0x00028124 CB_CLEAR_GREEN +0x00028120 CB_CLEAR_RED +0x00028C30 CB_CLRCMP_CONTROL +0x00028C38 CB_CLRCMP_DST +0x00028C3C CB_CLRCMP_MSK +0x00028C34 CB_CLRCMP_SRC +0x00028100 CB_COLOR0_MASK +0x00028104 CB_COLOR1_MASK +0x00028108 CB_COLOR2_MASK +0x0002810C CB_COLOR3_MASK +0x00028110 CB_COLOR4_MASK +0x00028114 CB_COLOR5_MASK +0x00028118 CB_COLOR6_MASK +0x0002811C CB_COLOR7_MASK +0x00028080 CB_COLOR0_VIEW +0x00028084 CB_COLOR1_VIEW +0x00028088 CB_COLOR2_VIEW +0x0002808C CB_COLOR3_VIEW +0x00028090 CB_COLOR4_VIEW +0x00028094 CB_COLOR5_VIEW +0x00028098 CB_COLOR6_VIEW +0x0002809C CB_COLOR7_VIEW +0x00028808 CB_COLOR_CONTROL +0x0002842C CB_FOG_BLUE +0x00028428 CB_FOG_GREEN +0x00028424 CB_FOG_RED +0x00008040 WAIT_UNTIL +0x00009714 VC_ENHANCE +0x00009830 DB_DEBUG +0x00009838 DB_WATERMARKS +0x00028D44 DB_ALPHA_TO_MASK +0x00009700 VC_CNTL diff --git a/drivers/gpu/drm/radeon/reg_srcs/rn50 b/drivers/gpu/drm/radeon/reg_srcs/rn50 new file mode 100644 index 00000000..2687b630 --- /dev/null +++ b/drivers/gpu/drm/radeon/reg_srcs/rn50 @@ -0,0 +1,30 @@ +rn50 0x3294 +0x1434 SRC_Y_X +0x1438 DST_Y_X +0x143C DST_HEIGHT_WIDTH +0x146C DP_GUI_MASTER_CNTL +0x1474 BRUSH_Y_X +0x1478 DP_BRUSH_BKGD_CLR +0x147C DP_BRUSH_FRGD_CLR +0x1480 BRUSH_DATA0 +0x1484 BRUSH_DATA1 +0x1598 DST_WIDTH_HEIGHT +0x15C0 CLR_CMP_CNTL +0x15C4 CLR_CMP_CLR_SRC +0x15C8 CLR_CMP_CLR_DST +0x15CC CLR_CMP_MSK +0x15D8 DP_SRC_FRGD_CLR +0x15DC DP_SRC_BKGD_CLR +0x1600 DST_LINE_START +0x1604 DST_LINE_END +0x1608 DST_LINE_PATCOUNT +0x16C0 DP_CNTL +0x16CC DP_WRITE_MSK +0x16D0 DP_CNTL_XDIR_YDIR_YMAJOR +0x16E8 DEFAULT_SC_BOTTOM_RIGHT +0x16EC SC_TOP_LEFT +0x16F0 SC_BOTTOM_RIGHT +0x16F4 SRC_SC_BOTTOM_RIGHT +0x1714 DSTCACHE_CTLSTAT +0x1720 WAIT_UNTIL +0x172C RBBM_GUICNTL diff --git a/drivers/gpu/drm/radeon/reg_srcs/rs600 b/drivers/gpu/drm/radeon/reg_srcs/rs600 new file mode 100644 index 00000000..d9f62866 --- /dev/null +++ b/drivers/gpu/drm/radeon/reg_srcs/rs600 @@ -0,0 +1,780 @@ +rs600 0x6d40 +0x1434 SRC_Y_X +0x1438 DST_Y_X +0x143C DST_HEIGHT_WIDTH +0x146C DP_GUI_MASTER_CNTL +0x1474 BRUSH_Y_X +0x1478 DP_BRUSH_BKGD_CLR +0x147C DP_BRUSH_FRGD_CLR +0x1480 BRUSH_DATA0 +0x1484 BRUSH_DATA1 +0x1598 DST_WIDTH_HEIGHT +0x15C0 CLR_CMP_CNTL +0x15C4 CLR_CMP_CLR_SRC +0x15C8 CLR_CMP_CLR_DST +0x15CC CLR_CMP_MSK +0x15D8 DP_SRC_FRGD_CLR +0x15DC DP_SRC_BKGD_CLR +0x1600 DST_LINE_START +0x1604 DST_LINE_END +0x1608 DST_LINE_PATCOUNT +0x16C0 DP_CNTL +0x16CC DP_WRITE_MSK +0x16D0 DP_CNTL_XDIR_YDIR_YMAJOR +0x16E8 DEFAULT_SC_BOTTOM_RIGHT +0x16EC SC_TOP_LEFT +0x16F0 SC_BOTTOM_RIGHT +0x16F4 SRC_SC_BOTTOM_RIGHT +0x1714 DSTCACHE_CTLSTAT +0x1720 WAIT_UNTIL +0x172C RBBM_GUICNTL +0x1D98 VAP_VPORT_XSCALE +0x1D9C VAP_VPORT_XOFFSET +0x1DA0 VAP_VPORT_YSCALE +0x1DA4 VAP_VPORT_YOFFSET +0x1DA8 VAP_VPORT_ZSCALE +0x1DAC VAP_VPORT_ZOFFSET +0x2080 VAP_CNTL +0x2090 VAP_OUT_VTX_FMT_0 +0x2094 VAP_OUT_VTX_FMT_1 +0x20B0 VAP_VTE_CNTL +0x2138 VAP_VF_MIN_VTX_INDX +0x2140 VAP_CNTL_STATUS +0x2150 VAP_PROG_STREAM_CNTL_0 +0x2154 VAP_PROG_STREAM_CNTL_1 +0x2158 VAP_PROG_STREAM_CNTL_2 +0x215C VAP_PROG_STREAM_CNTL_3 +0x2160 VAP_PROG_STREAM_CNTL_4 +0x2164 VAP_PROG_STREAM_CNTL_5 +0x2168 VAP_PROG_STREAM_CNTL_6 +0x216C VAP_PROG_STREAM_CNTL_7 +0x2180 VAP_VTX_STATE_CNTL +0x2184 VAP_VSM_VTX_ASSM +0x2188 VAP_VTX_STATE_IND_REG_0 +0x218C VAP_VTX_STATE_IND_REG_1 +0x2190 VAP_VTX_STATE_IND_REG_2 +0x2194 VAP_VTX_STATE_IND_REG_3 +0x2198 VAP_VTX_STATE_IND_REG_4 +0x219C VAP_VTX_STATE_IND_REG_5 +0x21A0 VAP_VTX_STATE_IND_REG_6 +0x21A4 VAP_VTX_STATE_IND_REG_7 +0x21A8 VAP_VTX_STATE_IND_REG_8 +0x21AC VAP_VTX_STATE_IND_REG_9 +0x21B0 VAP_VTX_STATE_IND_REG_10 +0x21B4 VAP_VTX_STATE_IND_REG_11 +0x21B8 VAP_VTX_STATE_IND_REG_12 +0x21BC VAP_VTX_STATE_IND_REG_13 +0x21C0 VAP_VTX_STATE_IND_REG_14 +0x21C4 VAP_VTX_STATE_IND_REG_15 +0x21DC VAP_PSC_SGN_NORM_CNTL +0x21E0 VAP_PROG_STREAM_CNTL_EXT_0 +0x21E4 VAP_PROG_STREAM_CNTL_EXT_1 +0x21E8 VAP_PROG_STREAM_CNTL_EXT_2 +0x21EC VAP_PROG_STREAM_CNTL_EXT_3 +0x21F0 VAP_PROG_STREAM_CNTL_EXT_4 +0x21F4 VAP_PROG_STREAM_CNTL_EXT_5 +0x21F8 VAP_PROG_STREAM_CNTL_EXT_6 +0x21FC VAP_PROG_STREAM_CNTL_EXT_7 +0x2200 VAP_PVS_VECTOR_INDX_REG +0x2204 VAP_PVS_VECTOR_DATA_REG +0x2208 VAP_PVS_VECTOR_DATA_REG_128 +0x221C VAP_CLIP_CNTL +0x2220 VAP_GB_VERT_CLIP_ADJ +0x2224 VAP_GB_VERT_DISC_ADJ +0x2228 VAP_GB_HORZ_CLIP_ADJ +0x222C VAP_GB_HORZ_DISC_ADJ +0x2230 VAP_PVS_FLOW_CNTL_ADDRS_0 +0x2234 VAP_PVS_FLOW_CNTL_ADDRS_1 +0x2238 VAP_PVS_FLOW_CNTL_ADDRS_2 +0x223C VAP_PVS_FLOW_CNTL_ADDRS_3 +0x2240 VAP_PVS_FLOW_CNTL_ADDRS_4 +0x2244 VAP_PVS_FLOW_CNTL_ADDRS_5 +0x2248 VAP_PVS_FLOW_CNTL_ADDRS_6 +0x224C VAP_PVS_FLOW_CNTL_ADDRS_7 +0x2250 VAP_PVS_FLOW_CNTL_ADDRS_8 +0x2254 VAP_PVS_FLOW_CNTL_ADDRS_9 +0x2258 VAP_PVS_FLOW_CNTL_ADDRS_10 +0x225C VAP_PVS_FLOW_CNTL_ADDRS_11 +0x2260 VAP_PVS_FLOW_CNTL_ADDRS_12 +0x2264 VAP_PVS_FLOW_CNTL_ADDRS_13 +0x2268 VAP_PVS_FLOW_CNTL_ADDRS_14 +0x226C VAP_PVS_FLOW_CNTL_ADDRS_15 +0x2284 VAP_PVS_STATE_FLUSH_REG +0x2288 VAP_PVS_VTX_TIMEOUT_REG +0x2290 VAP_PVS_FLOW_CNTL_LOOP_INDEX_0 +0x2294 VAP_PVS_FLOW_CNTL_LOOP_INDEX_1 +0x2298 VAP_PVS_FLOW_CNTL_LOOP_INDEX_2 +0x229C VAP_PVS_FLOW_CNTL_LOOP_INDEX_3 +0x22A0 VAP_PVS_FLOW_CNTL_LOOP_INDEX_4 +0x22A4 VAP_PVS_FLOW_CNTL_LOOP_INDEX_5 +0x22A8 VAP_PVS_FLOW_CNTL_LOOP_INDEX_6 +0x22AC VAP_PVS_FLOW_CNTL_LOOP_INDEX_7 +0x22B0 VAP_PVS_FLOW_CNTL_LOOP_INDEX_8 +0x22B4 VAP_PVS_FLOW_CNTL_LOOP_INDEX_9 +0x22B8 VAP_PVS_FLOW_CNTL_LOOP_INDEX_10 +0x22BC VAP_PVS_FLOW_CNTL_LOOP_INDEX_11 +0x22C0 VAP_PVS_FLOW_CNTL_LOOP_INDEX_12 +0x22C4 VAP_PVS_FLOW_CNTL_LOOP_INDEX_13 +0x22C8 VAP_PVS_FLOW_CNTL_LOOP_INDEX_14 +0x22CC VAP_PVS_FLOW_CNTL_LOOP_INDEX_15 +0x22D0 VAP_PVS_CODE_CNTL_0 +0x22D4 VAP_PVS_CONST_CNTL +0x22D8 VAP_PVS_CODE_CNTL_1 +0x22DC VAP_PVS_FLOW_CNTL_OPC +0x342C RB2D_DSTCACHE_CTLSTAT +0x4000 GB_VAP_RASTER_VTX_FMT_0 +0x4004 GB_VAP_RASTER_VTX_FMT_1 +0x4008 GB_ENABLE +0x4010 GB_MSPOS0 +0x4014 GB_MSPOS1 +0x401C GB_SELECT +0x4020 GB_AA_CONFIG +0x4024 GB_FIFO_SIZE +0x4100 TX_INVALTAGS +0x4200 GA_POINT_S0 +0x4204 GA_POINT_T0 +0x4208 GA_POINT_S1 +0x420C GA_POINT_T1 +0x4214 GA_TRIANGLE_STIPPLE +0x421C GA_POINT_SIZE +0x4230 GA_POINT_MINMAX +0x4234 GA_LINE_CNTL +0x4238 GA_LINE_STIPPLE_CONFIG +0x4260 GA_LINE_STIPPLE_VALUE +0x4264 GA_LINE_S0 +0x4268 GA_LINE_S1 +0x4278 GA_COLOR_CONTROL +0x427C GA_SOLID_RG +0x4280 GA_SOLID_BA +0x4288 GA_POLY_MODE +0x428C GA_ROUND_MODE +0x4290 GA_OFFSET +0x4294 GA_FOG_SCALE +0x4298 GA_FOG_OFFSET +0x42A0 SU_TEX_WRAP +0x42A4 SU_POLY_OFFSET_FRONT_SCALE +0x42A8 SU_POLY_OFFSET_FRONT_OFFSET +0x42AC SU_POLY_OFFSET_BACK_SCALE +0x42B0 SU_POLY_OFFSET_BACK_OFFSET +0x42B4 SU_POLY_OFFSET_ENABLE +0x42B8 SU_CULL_MODE +0x42C0 SU_DEPTH_SCALE +0x42C4 SU_DEPTH_OFFSET +0x42C8 SU_REG_DEST +0x4300 RS_COUNT +0x4304 RS_INST_COUNT +0x4310 RS_IP_0 +0x4314 RS_IP_1 +0x4318 RS_IP_2 +0x431C RS_IP_3 +0x4320 RS_IP_4 +0x4324 RS_IP_5 +0x4328 RS_IP_6 +0x432C RS_IP_7 +0x4330 RS_INST_0 +0x4334 RS_INST_1 +0x4338 RS_INST_2 +0x433C RS_INST_3 +0x4340 RS_INST_4 +0x4344 RS_INST_5 +0x4348 RS_INST_6 +0x434C RS_INST_7 +0x4350 RS_INST_8 +0x4354 RS_INST_9 +0x4358 RS_INST_10 +0x435C RS_INST_11 +0x4360 RS_INST_12 +0x4364 RS_INST_13 +0x4368 RS_INST_14 +0x436C RS_INST_15 +0x43A8 SC_EDGERULE +0x43B0 SC_CLIP_0_A +0x43B4 SC_CLIP_0_B +0x43B8 SC_CLIP_1_A +0x43BC SC_CLIP_1_B +0x43C0 SC_CLIP_2_A +0x43C4 SC_CLIP_2_B +0x43C8 SC_CLIP_3_A +0x43CC SC_CLIP_3_B +0x43D0 SC_CLIP_RULE +0x43E0 SC_SCISSOR0 +0x43E8 SC_SCREENDOOR +0x4440 TX_FILTER1_0 +0x4444 TX_FILTER1_1 +0x4448 TX_FILTER1_2 +0x444C TX_FILTER1_3 +0x4450 TX_FILTER1_4 +0x4454 TX_FILTER1_5 +0x4458 TX_FILTER1_6 +0x445C TX_FILTER1_7 +0x4460 TX_FILTER1_8 +0x4464 TX_FILTER1_9 +0x4468 TX_FILTER1_10 +0x446C TX_FILTER1_11 +0x4470 TX_FILTER1_12 +0x4474 TX_FILTER1_13 +0x4478 TX_FILTER1_14 +0x447C TX_FILTER1_15 +0x4580 TX_CHROMA_KEY_0 +0x4584 TX_CHROMA_KEY_1 +0x4588 TX_CHROMA_KEY_2 +0x458C TX_CHROMA_KEY_3 +0x4590 TX_CHROMA_KEY_4 +0x4594 TX_CHROMA_KEY_5 +0x4598 TX_CHROMA_KEY_6 +0x459C TX_CHROMA_KEY_7 +0x45A0 TX_CHROMA_KEY_8 +0x45A4 TX_CHROMA_KEY_9 +0x45A8 TX_CHROMA_KEY_10 +0x45AC TX_CHROMA_KEY_11 +0x45B0 TX_CHROMA_KEY_12 +0x45B4 TX_CHROMA_KEY_13 +0x45B8 TX_CHROMA_KEY_14 +0x45BC TX_CHROMA_KEY_15 +0x45C0 TX_BORDER_COLOR_0 +0x45C4 TX_BORDER_COLOR_1 +0x45C8 TX_BORDER_COLOR_2 +0x45CC TX_BORDER_COLOR_3 +0x45D0 TX_BORDER_COLOR_4 +0x45D4 TX_BORDER_COLOR_5 +0x45D8 TX_BORDER_COLOR_6 +0x45DC TX_BORDER_COLOR_7 +0x45E0 TX_BORDER_COLOR_8 +0x45E4 TX_BORDER_COLOR_9 +0x45E8 TX_BORDER_COLOR_10 +0x45EC TX_BORDER_COLOR_11 +0x45F0 TX_BORDER_COLOR_12 +0x45F4 TX_BORDER_COLOR_13 +0x45F8 TX_BORDER_COLOR_14 +0x45FC TX_BORDER_COLOR_15 +0x4600 US_CONFIG +0x4604 US_PIXSIZE +0x4608 US_CODE_OFFSET +0x460C US_RESET +0x4610 US_CODE_ADDR_0 +0x4614 US_CODE_ADDR_1 +0x4618 US_CODE_ADDR_2 +0x461C US_CODE_ADDR_3 +0x4620 US_TEX_INST_0 +0x4624 US_TEX_INST_1 +0x4628 US_TEX_INST_2 +0x462C US_TEX_INST_3 +0x4630 US_TEX_INST_4 +0x4634 US_TEX_INST_5 +0x4638 US_TEX_INST_6 +0x463C US_TEX_INST_7 +0x4640 US_TEX_INST_8 +0x4644 US_TEX_INST_9 +0x4648 US_TEX_INST_10 +0x464C US_TEX_INST_11 +0x4650 US_TEX_INST_12 +0x4654 US_TEX_INST_13 +0x4658 US_TEX_INST_14 +0x465C US_TEX_INST_15 +0x4660 US_TEX_INST_16 +0x4664 US_TEX_INST_17 +0x4668 US_TEX_INST_18 +0x466C US_TEX_INST_19 +0x4670 US_TEX_INST_20 +0x4674 US_TEX_INST_21 +0x4678 US_TEX_INST_22 +0x467C US_TEX_INST_23 +0x4680 US_TEX_INST_24 +0x4684 US_TEX_INST_25 +0x4688 US_TEX_INST_26 +0x468C US_TEX_INST_27 +0x4690 US_TEX_INST_28 +0x4694 US_TEX_INST_29 +0x4698 US_TEX_INST_30 +0x469C US_TEX_INST_31 +0x46A4 US_OUT_FMT_0 +0x46A8 US_OUT_FMT_1 +0x46AC US_OUT_FMT_2 +0x46B0 US_OUT_FMT_3 +0x46B4 US_W_FMT +0x46B8 US_CODE_BANK +0x46BC US_CODE_EXT +0x46C0 US_ALU_RGB_ADDR_0 +0x46C4 US_ALU_RGB_ADDR_1 +0x46C8 US_ALU_RGB_ADDR_2 +0x46CC US_ALU_RGB_ADDR_3 +0x46D0 US_ALU_RGB_ADDR_4 +0x46D4 US_ALU_RGB_ADDR_5 +0x46D8 US_ALU_RGB_ADDR_6 +0x46DC US_ALU_RGB_ADDR_7 +0x46E0 US_ALU_RGB_ADDR_8 +0x46E4 US_ALU_RGB_ADDR_9 +0x46E8 US_ALU_RGB_ADDR_10 +0x46EC US_ALU_RGB_ADDR_11 +0x46F0 US_ALU_RGB_ADDR_12 +0x46F4 US_ALU_RGB_ADDR_13 +0x46F8 US_ALU_RGB_ADDR_14 +0x46FC US_ALU_RGB_ADDR_15 +0x4700 US_ALU_RGB_ADDR_16 +0x4704 US_ALU_RGB_ADDR_17 +0x4708 US_ALU_RGB_ADDR_18 +0x470C US_ALU_RGB_ADDR_19 +0x4710 US_ALU_RGB_ADDR_20 +0x4714 US_ALU_RGB_ADDR_21 +0x4718 US_ALU_RGB_ADDR_22 +0x471C US_ALU_RGB_ADDR_23 +0x4720 US_ALU_RGB_ADDR_24 +0x4724 US_ALU_RGB_ADDR_25 +0x4728 US_ALU_RGB_ADDR_26 +0x472C US_ALU_RGB_ADDR_27 +0x4730 US_ALU_RGB_ADDR_28 +0x4734 US_ALU_RGB_ADDR_29 +0x4738 US_ALU_RGB_ADDR_30 +0x473C US_ALU_RGB_ADDR_31 +0x4740 US_ALU_RGB_ADDR_32 +0x4744 US_ALU_RGB_ADDR_33 +0x4748 US_ALU_RGB_ADDR_34 +0x474C US_ALU_RGB_ADDR_35 +0x4750 US_ALU_RGB_ADDR_36 +0x4754 US_ALU_RGB_ADDR_37 +0x4758 US_ALU_RGB_ADDR_38 +0x475C US_ALU_RGB_ADDR_39 +0x4760 US_ALU_RGB_ADDR_40 +0x4764 US_ALU_RGB_ADDR_41 +0x4768 US_ALU_RGB_ADDR_42 +0x476C US_ALU_RGB_ADDR_43 +0x4770 US_ALU_RGB_ADDR_44 +0x4774 US_ALU_RGB_ADDR_45 +0x4778 US_ALU_RGB_ADDR_46 +0x477C US_ALU_RGB_ADDR_47 +0x4780 US_ALU_RGB_ADDR_48 +0x4784 US_ALU_RGB_ADDR_49 +0x4788 US_ALU_RGB_ADDR_50 +0x478C US_ALU_RGB_ADDR_51 +0x4790 US_ALU_RGB_ADDR_52 +0x4794 US_ALU_RGB_ADDR_53 +0x4798 US_ALU_RGB_ADDR_54 +0x479C US_ALU_RGB_ADDR_55 +0x47A0 US_ALU_RGB_ADDR_56 +0x47A4 US_ALU_RGB_ADDR_57 +0x47A8 US_ALU_RGB_ADDR_58 +0x47AC US_ALU_RGB_ADDR_59 +0x47B0 US_ALU_RGB_ADDR_60 +0x47B4 US_ALU_RGB_ADDR_61 +0x47B8 US_ALU_RGB_ADDR_62 +0x47BC US_ALU_RGB_ADDR_63 +0x47C0 US_ALU_ALPHA_ADDR_0 +0x47C4 US_ALU_ALPHA_ADDR_1 +0x47C8 US_ALU_ALPHA_ADDR_2 +0x47CC US_ALU_ALPHA_ADDR_3 +0x47D0 US_ALU_ALPHA_ADDR_4 +0x47D4 US_ALU_ALPHA_ADDR_5 +0x47D8 US_ALU_ALPHA_ADDR_6 +0x47DC US_ALU_ALPHA_ADDR_7 +0x47E0 US_ALU_ALPHA_ADDR_8 +0x47E4 US_ALU_ALPHA_ADDR_9 +0x47E8 US_ALU_ALPHA_ADDR_10 +0x47EC US_ALU_ALPHA_ADDR_11 +0x47F0 US_ALU_ALPHA_ADDR_12 +0x47F4 US_ALU_ALPHA_ADDR_13 +0x47F8 US_ALU_ALPHA_ADDR_14 +0x47FC US_ALU_ALPHA_ADDR_15 +0x4800 US_ALU_ALPHA_ADDR_16 +0x4804 US_ALU_ALPHA_ADDR_17 +0x4808 US_ALU_ALPHA_ADDR_18 +0x480C US_ALU_ALPHA_ADDR_19 +0x4810 US_ALU_ALPHA_ADDR_20 +0x4814 US_ALU_ALPHA_ADDR_21 +0x4818 US_ALU_ALPHA_ADDR_22 +0x481C US_ALU_ALPHA_ADDR_23 +0x4820 US_ALU_ALPHA_ADDR_24 +0x4824 US_ALU_ALPHA_ADDR_25 +0x4828 US_ALU_ALPHA_ADDR_26 +0x482C US_ALU_ALPHA_ADDR_27 +0x4830 US_ALU_ALPHA_ADDR_28 +0x4834 US_ALU_ALPHA_ADDR_29 +0x4838 US_ALU_ALPHA_ADDR_30 +0x483C US_ALU_ALPHA_ADDR_31 +0x4840 US_ALU_ALPHA_ADDR_32 +0x4844 US_ALU_ALPHA_ADDR_33 +0x4848 US_ALU_ALPHA_ADDR_34 +0x484C US_ALU_ALPHA_ADDR_35 +0x4850 US_ALU_ALPHA_ADDR_36 +0x4854 US_ALU_ALPHA_ADDR_37 +0x4858 US_ALU_ALPHA_ADDR_38 +0x485C US_ALU_ALPHA_ADDR_39 +0x4860 US_ALU_ALPHA_ADDR_40 +0x4864 US_ALU_ALPHA_ADDR_41 +0x4868 US_ALU_ALPHA_ADDR_42 +0x486C US_ALU_ALPHA_ADDR_43 +0x4870 US_ALU_ALPHA_ADDR_44 +0x4874 US_ALU_ALPHA_ADDR_45 +0x4878 US_ALU_ALPHA_ADDR_46 +0x487C US_ALU_ALPHA_ADDR_47 +0x4880 US_ALU_ALPHA_ADDR_48 +0x4884 US_ALU_ALPHA_ADDR_49 +0x4888 US_ALU_ALPHA_ADDR_50 +0x488C US_ALU_ALPHA_ADDR_51 +0x4890 US_ALU_ALPHA_ADDR_52 +0x4894 US_ALU_ALPHA_ADDR_53 +0x4898 US_ALU_ALPHA_ADDR_54 +0x489C US_ALU_ALPHA_ADDR_55 +0x48A0 US_ALU_ALPHA_ADDR_56 +0x48A4 US_ALU_ALPHA_ADDR_57 +0x48A8 US_ALU_ALPHA_ADDR_58 +0x48AC US_ALU_ALPHA_ADDR_59 +0x48B0 US_ALU_ALPHA_ADDR_60 +0x48B4 US_ALU_ALPHA_ADDR_61 +0x48B8 US_ALU_ALPHA_ADDR_62 +0x48BC US_ALU_ALPHA_ADDR_63 +0x48C0 US_ALU_RGB_INST_0 +0x48C4 US_ALU_RGB_INST_1 +0x48C8 US_ALU_RGB_INST_2 +0x48CC US_ALU_RGB_INST_3 +0x48D0 US_ALU_RGB_INST_4 +0x48D4 US_ALU_RGB_INST_5 +0x48D8 US_ALU_RGB_INST_6 +0x48DC US_ALU_RGB_INST_7 +0x48E0 US_ALU_RGB_INST_8 +0x48E4 US_ALU_RGB_INST_9 +0x48E8 US_ALU_RGB_INST_10 +0x48EC US_ALU_RGB_INST_11 +0x48F0 US_ALU_RGB_INST_12 +0x48F4 US_ALU_RGB_INST_13 +0x48F8 US_ALU_RGB_INST_14 +0x48FC US_ALU_RGB_INST_15 +0x4900 US_ALU_RGB_INST_16 +0x4904 US_ALU_RGB_INST_17 +0x4908 US_ALU_RGB_INST_18 +0x490C US_ALU_RGB_INST_19 +0x4910 US_ALU_RGB_INST_20 +0x4914 US_ALU_RGB_INST_21 +0x4918 US_ALU_RGB_INST_22 +0x491C US_ALU_RGB_INST_23 +0x4920 US_ALU_RGB_INST_24 +0x4924 US_ALU_RGB_INST_25 +0x4928 US_ALU_RGB_INST_26 +0x492C US_ALU_RGB_INST_27 +0x4930 US_ALU_RGB_INST_28 +0x4934 US_ALU_RGB_INST_29 +0x4938 US_ALU_RGB_INST_30 +0x493C US_ALU_RGB_INST_31 +0x4940 US_ALU_RGB_INST_32 +0x4944 US_ALU_RGB_INST_33 +0x4948 US_ALU_RGB_INST_34 +0x494C US_ALU_RGB_INST_35 +0x4950 US_ALU_RGB_INST_36 +0x4954 US_ALU_RGB_INST_37 +0x4958 US_ALU_RGB_INST_38 +0x495C US_ALU_RGB_INST_39 +0x4960 US_ALU_RGB_INST_40 +0x4964 US_ALU_RGB_INST_41 +0x4968 US_ALU_RGB_INST_42 +0x496C US_ALU_RGB_INST_43 +0x4970 US_ALU_RGB_INST_44 +0x4974 US_ALU_RGB_INST_45 +0x4978 US_ALU_RGB_INST_46 +0x497C US_ALU_RGB_INST_47 +0x4980 US_ALU_RGB_INST_48 +0x4984 US_ALU_RGB_INST_49 +0x4988 US_ALU_RGB_INST_50 +0x498C US_ALU_RGB_INST_51 +0x4990 US_ALU_RGB_INST_52 +0x4994 US_ALU_RGB_INST_53 +0x4998 US_ALU_RGB_INST_54 +0x499C US_ALU_RGB_INST_55 +0x49A0 US_ALU_RGB_INST_56 +0x49A4 US_ALU_RGB_INST_57 +0x49A8 US_ALU_RGB_INST_58 +0x49AC US_ALU_RGB_INST_59 +0x49B0 US_ALU_RGB_INST_60 +0x49B4 US_ALU_RGB_INST_61 +0x49B8 US_ALU_RGB_INST_62 +0x49BC US_ALU_RGB_INST_63 +0x49C0 US_ALU_ALPHA_INST_0 +0x49C4 US_ALU_ALPHA_INST_1 +0x49C8 US_ALU_ALPHA_INST_2 +0x49CC US_ALU_ALPHA_INST_3 +0x49D0 US_ALU_ALPHA_INST_4 +0x49D4 US_ALU_ALPHA_INST_5 +0x49D8 US_ALU_ALPHA_INST_6 +0x49DC US_ALU_ALPHA_INST_7 +0x49E0 US_ALU_ALPHA_INST_8 +0x49E4 US_ALU_ALPHA_INST_9 +0x49E8 US_ALU_ALPHA_INST_10 +0x49EC US_ALU_ALPHA_INST_11 +0x49F0 US_ALU_ALPHA_INST_12 +0x49F4 US_ALU_ALPHA_INST_13 +0x49F8 US_ALU_ALPHA_INST_14 +0x49FC US_ALU_ALPHA_INST_15 +0x4A00 US_ALU_ALPHA_INST_16 +0x4A04 US_ALU_ALPHA_INST_17 +0x4A08 US_ALU_ALPHA_INST_18 +0x4A0C US_ALU_ALPHA_INST_19 +0x4A10 US_ALU_ALPHA_INST_20 +0x4A14 US_ALU_ALPHA_INST_21 +0x4A18 US_ALU_ALPHA_INST_22 +0x4A1C US_ALU_ALPHA_INST_23 +0x4A20 US_ALU_ALPHA_INST_24 +0x4A24 US_ALU_ALPHA_INST_25 +0x4A28 US_ALU_ALPHA_INST_26 +0x4A2C US_ALU_ALPHA_INST_27 +0x4A30 US_ALU_ALPHA_INST_28 +0x4A34 US_ALU_ALPHA_INST_29 +0x4A38 US_ALU_ALPHA_INST_30 +0x4A3C US_ALU_ALPHA_INST_31 +0x4A40 US_ALU_ALPHA_INST_32 +0x4A44 US_ALU_ALPHA_INST_33 +0x4A48 US_ALU_ALPHA_INST_34 +0x4A4C US_ALU_ALPHA_INST_35 +0x4A50 US_ALU_ALPHA_INST_36 +0x4A54 US_ALU_ALPHA_INST_37 +0x4A58 US_ALU_ALPHA_INST_38 +0x4A5C US_ALU_ALPHA_INST_39 +0x4A60 US_ALU_ALPHA_INST_40 +0x4A64 US_ALU_ALPHA_INST_41 +0x4A68 US_ALU_ALPHA_INST_42 +0x4A6C US_ALU_ALPHA_INST_43 +0x4A70 US_ALU_ALPHA_INST_44 +0x4A74 US_ALU_ALPHA_INST_45 +0x4A78 US_ALU_ALPHA_INST_46 +0x4A7C US_ALU_ALPHA_INST_47 +0x4A80 US_ALU_ALPHA_INST_48 +0x4A84 US_ALU_ALPHA_INST_49 +0x4A88 US_ALU_ALPHA_INST_50 +0x4A8C US_ALU_ALPHA_INST_51 +0x4A90 US_ALU_ALPHA_INST_52 +0x4A94 US_ALU_ALPHA_INST_53 +0x4A98 US_ALU_ALPHA_INST_54 +0x4A9C US_ALU_ALPHA_INST_55 +0x4AA0 US_ALU_ALPHA_INST_56 +0x4AA4 US_ALU_ALPHA_INST_57 +0x4AA8 US_ALU_ALPHA_INST_58 +0x4AAC US_ALU_ALPHA_INST_59 +0x4AB0 US_ALU_ALPHA_INST_60 +0x4AB4 US_ALU_ALPHA_INST_61 +0x4AB8 US_ALU_ALPHA_INST_62 +0x4ABC US_ALU_ALPHA_INST_63 +0x4AC0 US_ALU_EXT_ADDR_0 +0x4AC4 US_ALU_EXT_ADDR_1 +0x4AC8 US_ALU_EXT_ADDR_2 +0x4ACC US_ALU_EXT_ADDR_3 +0x4AD0 US_ALU_EXT_ADDR_4 +0x4AD4 US_ALU_EXT_ADDR_5 +0x4AD8 US_ALU_EXT_ADDR_6 +0x4ADC US_ALU_EXT_ADDR_7 +0x4AE0 US_ALU_EXT_ADDR_8 +0x4AE4 US_ALU_EXT_ADDR_9 +0x4AE8 US_ALU_EXT_ADDR_10 +0x4AEC US_ALU_EXT_ADDR_11 +0x4AF0 US_ALU_EXT_ADDR_12 +0x4AF4 US_ALU_EXT_ADDR_13 +0x4AF8 US_ALU_EXT_ADDR_14 +0x4AFC US_ALU_EXT_ADDR_15 +0x4B00 US_ALU_EXT_ADDR_16 +0x4B04 US_ALU_EXT_ADDR_17 +0x4B08 US_ALU_EXT_ADDR_18 +0x4B0C US_ALU_EXT_ADDR_19 +0x4B10 US_ALU_EXT_ADDR_20 +0x4B14 US_ALU_EXT_ADDR_21 +0x4B18 US_ALU_EXT_ADDR_22 +0x4B1C US_ALU_EXT_ADDR_23 +0x4B20 US_ALU_EXT_ADDR_24 +0x4B24 US_ALU_EXT_ADDR_25 +0x4B28 US_ALU_EXT_ADDR_26 +0x4B2C US_ALU_EXT_ADDR_27 +0x4B30 US_ALU_EXT_ADDR_28 +0x4B34 US_ALU_EXT_ADDR_29 +0x4B38 US_ALU_EXT_ADDR_30 +0x4B3C US_ALU_EXT_ADDR_31 +0x4B40 US_ALU_EXT_ADDR_32 +0x4B44 US_ALU_EXT_ADDR_33 +0x4B48 US_ALU_EXT_ADDR_34 +0x4B4C US_ALU_EXT_ADDR_35 +0x4B50 US_ALU_EXT_ADDR_36 +0x4B54 US_ALU_EXT_ADDR_37 +0x4B58 US_ALU_EXT_ADDR_38 +0x4B5C US_ALU_EXT_ADDR_39 +0x4B60 US_ALU_EXT_ADDR_40 +0x4B64 US_ALU_EXT_ADDR_41 +0x4B68 US_ALU_EXT_ADDR_42 +0x4B6C US_ALU_EXT_ADDR_43 +0x4B70 US_ALU_EXT_ADDR_44 +0x4B74 US_ALU_EXT_ADDR_45 +0x4B78 US_ALU_EXT_ADDR_46 +0x4B7C US_ALU_EXT_ADDR_47 +0x4B80 US_ALU_EXT_ADDR_48 +0x4B84 US_ALU_EXT_ADDR_49 +0x4B88 US_ALU_EXT_ADDR_50 +0x4B8C US_ALU_EXT_ADDR_51 +0x4B90 US_ALU_EXT_ADDR_52 +0x4B94 US_ALU_EXT_ADDR_53 +0x4B98 US_ALU_EXT_ADDR_54 +0x4B9C US_ALU_EXT_ADDR_55 +0x4BA0 US_ALU_EXT_ADDR_56 +0x4BA4 US_ALU_EXT_ADDR_57 +0x4BA8 US_ALU_EXT_ADDR_58 +0x4BAC US_ALU_EXT_ADDR_59 +0x4BB0 US_ALU_EXT_ADDR_60 +0x4BB4 US_ALU_EXT_ADDR_61 +0x4BB8 US_ALU_EXT_ADDR_62 +0x4BBC US_ALU_EXT_ADDR_63 +0x4BC0 FG_FOG_BLEND +0x4BC4 FG_FOG_FACTOR +0x4BC8 FG_FOG_COLOR_R +0x4BCC FG_FOG_COLOR_G +0x4BD0 FG_FOG_COLOR_B +0x4BD4 FG_ALPHA_FUNC +0x4BD8 FG_DEPTH_SRC +0x4C00 US_ALU_CONST_R_0 +0x4C04 US_ALU_CONST_G_0 +0x4C08 US_ALU_CONST_B_0 +0x4C0C US_ALU_CONST_A_0 +0x4C10 US_ALU_CONST_R_1 +0x4C14 US_ALU_CONST_G_1 +0x4C18 US_ALU_CONST_B_1 +0x4C1C US_ALU_CONST_A_1 +0x4C20 US_ALU_CONST_R_2 +0x4C24 US_ALU_CONST_G_2 +0x4C28 US_ALU_CONST_B_2 +0x4C2C US_ALU_CONST_A_2 +0x4C30 US_ALU_CONST_R_3 +0x4C34 US_ALU_CONST_G_3 +0x4C38 US_ALU_CONST_B_3 +0x4C3C US_ALU_CONST_A_3 +0x4C40 US_ALU_CONST_R_4 +0x4C44 US_ALU_CONST_G_4 +0x4C48 US_ALU_CONST_B_4 +0x4C4C US_ALU_CONST_A_4 +0x4C50 US_ALU_CONST_R_5 +0x4C54 US_ALU_CONST_G_5 +0x4C58 US_ALU_CONST_B_5 +0x4C5C US_ALU_CONST_A_5 +0x4C60 US_ALU_CONST_R_6 +0x4C64 US_ALU_CONST_G_6 +0x4C68 US_ALU_CONST_B_6 +0x4C6C US_ALU_CONST_A_6 +0x4C70 US_ALU_CONST_R_7 +0x4C74 US_ALU_CONST_G_7 +0x4C78 US_ALU_CONST_B_7 +0x4C7C US_ALU_CONST_A_7 +0x4C80 US_ALU_CONST_R_8 +0x4C84 US_ALU_CONST_G_8 +0x4C88 US_ALU_CONST_B_8 +0x4C8C US_ALU_CONST_A_8 +0x4C90 US_ALU_CONST_R_9 +0x4C94 US_ALU_CONST_G_9 +0x4C98 US_ALU_CONST_B_9 +0x4C9C US_ALU_CONST_A_9 +0x4CA0 US_ALU_CONST_R_10 +0x4CA4 US_ALU_CONST_G_10 +0x4CA8 US_ALU_CONST_B_10 +0x4CAC US_ALU_CONST_A_10 +0x4CB0 US_ALU_CONST_R_11 +0x4CB4 US_ALU_CONST_G_11 +0x4CB8 US_ALU_CONST_B_11 +0x4CBC US_ALU_CONST_A_11 +0x4CC0 US_ALU_CONST_R_12 +0x4CC4 US_ALU_CONST_G_12 +0x4CC8 US_ALU_CONST_B_12 +0x4CCC US_ALU_CONST_A_12 +0x4CD0 US_ALU_CONST_R_13 +0x4CD4 US_ALU_CONST_G_13 +0x4CD8 US_ALU_CONST_B_13 +0x4CDC US_ALU_CONST_A_13 +0x4CE0 US_ALU_CONST_R_14 +0x4CE4 US_ALU_CONST_G_14 +0x4CE8 US_ALU_CONST_B_14 +0x4CEC US_ALU_CONST_A_14 +0x4CF0 US_ALU_CONST_R_15 +0x4CF4 US_ALU_CONST_G_15 +0x4CF8 US_ALU_CONST_B_15 +0x4CFC US_ALU_CONST_A_15 +0x4D00 US_ALU_CONST_R_16 +0x4D04 US_ALU_CONST_G_16 +0x4D08 US_ALU_CONST_B_16 +0x4D0C US_ALU_CONST_A_16 +0x4D10 US_ALU_CONST_R_17 +0x4D14 US_ALU_CONST_G_17 +0x4D18 US_ALU_CONST_B_17 +0x4D1C US_ALU_CONST_A_17 +0x4D20 US_ALU_CONST_R_18 +0x4D24 US_ALU_CONST_G_18 +0x4D28 US_ALU_CONST_B_18 +0x4D2C US_ALU_CONST_A_18 +0x4D30 US_ALU_CONST_R_19 +0x4D34 US_ALU_CONST_G_19 +0x4D38 US_ALU_CONST_B_19 +0x4D3C US_ALU_CONST_A_19 +0x4D40 US_ALU_CONST_R_20 +0x4D44 US_ALU_CONST_G_20 +0x4D48 US_ALU_CONST_B_20 +0x4D4C US_ALU_CONST_A_20 +0x4D50 US_ALU_CONST_R_21 +0x4D54 US_ALU_CONST_G_21 +0x4D58 US_ALU_CONST_B_21 +0x4D5C US_ALU_CONST_A_21 +0x4D60 US_ALU_CONST_R_22 +0x4D64 US_ALU_CONST_G_22 +0x4D68 US_ALU_CONST_B_22 +0x4D6C US_ALU_CONST_A_22 +0x4D70 US_ALU_CONST_R_23 +0x4D74 US_ALU_CONST_G_23 +0x4D78 US_ALU_CONST_B_23 +0x4D7C US_ALU_CONST_A_23 +0x4D80 US_ALU_CONST_R_24 +0x4D84 US_ALU_CONST_G_24 +0x4D88 US_ALU_CONST_B_24 +0x4D8C US_ALU_CONST_A_24 +0x4D90 US_ALU_CONST_R_25 +0x4D94 US_ALU_CONST_G_25 +0x4D98 US_ALU_CONST_B_25 +0x4D9C US_ALU_CONST_A_25 +0x4DA0 US_ALU_CONST_R_26 +0x4DA4 US_ALU_CONST_G_26 +0x4DA8 US_ALU_CONST_B_26 +0x4DAC US_ALU_CONST_A_26 +0x4DB0 US_ALU_CONST_R_27 +0x4DB4 US_ALU_CONST_G_27 +0x4DB8 US_ALU_CONST_B_27 +0x4DBC US_ALU_CONST_A_27 +0x4DC0 US_ALU_CONST_R_28 +0x4DC4 US_ALU_CONST_G_28 +0x4DC8 US_ALU_CONST_B_28 +0x4DCC US_ALU_CONST_A_28 +0x4DD0 US_ALU_CONST_R_29 +0x4DD4 US_ALU_CONST_G_29 +0x4DD8 US_ALU_CONST_B_29 +0x4DDC US_ALU_CONST_A_29 +0x4DE0 US_ALU_CONST_R_30 +0x4DE4 US_ALU_CONST_G_30 +0x4DE8 US_ALU_CONST_B_30 +0x4DEC US_ALU_CONST_A_30 +0x4DF0 US_ALU_CONST_R_31 +0x4DF4 US_ALU_CONST_G_31 +0x4DF8 US_ALU_CONST_B_31 +0x4DFC US_ALU_CONST_A_31 +0x4E08 RB3D_ABLENDCNTL_R3 +0x4E10 RB3D_CONSTANT_COLOR +0x4E14 RB3D_COLOR_CLEAR_VALUE +0x4E18 RB3D_ROPCNTL_R3 +0x4E1C RB3D_CLRCMP_FLIPE_R3 +0x4E20 RB3D_CLRCMP_CLR_R3 +0x4E24 RB3D_CLRCMP_MSK_R3 +0x4E48 RB3D_DEBUG_CTL +0x4E4C RB3D_DSTCACHE_CTLSTAT_R3 +0x4E50 RB3D_DITHER_CTL +0x4E54 RB3D_CMASK_OFFSET0 +0x4E58 RB3D_CMASK_OFFSET1 +0x4E5C RB3D_CMASK_OFFSET2 +0x4E60 RB3D_CMASK_OFFSET3 +0x4E64 RB3D_CMASK_PITCH0 +0x4E68 RB3D_CMASK_PITCH1 +0x4E6C RB3D_CMASK_PITCH2 +0x4E70 RB3D_CMASK_PITCH3 +0x4E74 RB3D_CMASK_WRINDEX +0x4E78 RB3D_CMASK_DWORD +0x4E7C RB3D_CMASK_RDINDEX +0x4EA0 RB3D_DISCARD_SRC_PIXEL_LTE_THRESHOLD +0x4EA4 RB3D_DISCARD_SRC_PIXEL_GTE_THRESHOLD +0x4F04 ZB_ZSTENCILCNTL +0x4F08 ZB_STENCILREFMASK +0x4F14 ZB_ZTOP +0x4F18 ZB_ZCACHE_CTLSTAT +0x4F28 ZB_DEPTHCLEARVALUE +0x4F58 ZB_ZPASS_DATA diff --git a/drivers/gpu/drm/radeon/reg_srcs/rv515 b/drivers/gpu/drm/radeon/reg_srcs/rv515 new file mode 100644 index 00000000..911a8fbd --- /dev/null +++ b/drivers/gpu/drm/radeon/reg_srcs/rv515 @@ -0,0 +1,494 @@ +rv515 0x6d40 +0x1434 SRC_Y_X +0x1438 DST_Y_X +0x143C DST_HEIGHT_WIDTH +0x146C DP_GUI_MASTER_CNTL +0x1474 BRUSH_Y_X +0x1478 DP_BRUSH_BKGD_CLR +0x147C DP_BRUSH_FRGD_CLR +0x1480 BRUSH_DATA0 +0x1484 BRUSH_DATA1 +0x1598 DST_WIDTH_HEIGHT +0x15C0 CLR_CMP_CNTL +0x15C4 CLR_CMP_CLR_SRC +0x15C8 CLR_CMP_CLR_DST +0x15CC CLR_CMP_MSK +0x15D8 DP_SRC_FRGD_CLR +0x15DC DP_SRC_BKGD_CLR +0x1600 DST_LINE_START +0x1604 DST_LINE_END +0x1608 DST_LINE_PATCOUNT +0x16C0 DP_CNTL +0x16CC DP_WRITE_MSK +0x16D0 DP_CNTL_XDIR_YDIR_YMAJOR +0x16E8 DEFAULT_SC_BOTTOM_RIGHT +0x16EC SC_TOP_LEFT +0x16F0 SC_BOTTOM_RIGHT +0x16F4 SRC_SC_BOTTOM_RIGHT +0x1714 DSTCACHE_CTLSTAT +0x1720 WAIT_UNTIL +0x172C RBBM_GUICNTL +0x1D98 VAP_VPORT_XSCALE +0x1D9C VAP_VPORT_XOFFSET +0x1DA0 VAP_VPORT_YSCALE +0x1DA4 VAP_VPORT_YOFFSET +0x1DA8 VAP_VPORT_ZSCALE +0x1DAC VAP_VPORT_ZOFFSET +0x2080 VAP_CNTL +0x208C VAP_INDEX_OFFSET +0x2090 VAP_OUT_VTX_FMT_0 +0x2094 VAP_OUT_VTX_FMT_1 +0x20B0 VAP_VTE_CNTL +0x2138 VAP_VF_MIN_VTX_INDX +0x2140 VAP_CNTL_STATUS +0x2150 VAP_PROG_STREAM_CNTL_0 +0x2154 VAP_PROG_STREAM_CNTL_1 +0x2158 VAP_PROG_STREAM_CNTL_2 +0x215C VAP_PROG_STREAM_CNTL_3 +0x2160 VAP_PROG_STREAM_CNTL_4 +0x2164 VAP_PROG_STREAM_CNTL_5 +0x2168 VAP_PROG_STREAM_CNTL_6 +0x216C VAP_PROG_STREAM_CNTL_7 +0x2180 VAP_VTX_STATE_CNTL +0x2184 VAP_VSM_VTX_ASSM +0x2188 VAP_VTX_STATE_IND_REG_0 +0x218C VAP_VTX_STATE_IND_REG_1 +0x2190 VAP_VTX_STATE_IND_REG_2 +0x2194 VAP_VTX_STATE_IND_REG_3 +0x2198 VAP_VTX_STATE_IND_REG_4 +0x219C VAP_VTX_STATE_IND_REG_5 +0x21A0 VAP_VTX_STATE_IND_REG_6 +0x21A4 VAP_VTX_STATE_IND_REG_7 +0x21A8 VAP_VTX_STATE_IND_REG_8 +0x21AC VAP_VTX_STATE_IND_REG_9 +0x21B0 VAP_VTX_STATE_IND_REG_10 +0x21B4 VAP_VTX_STATE_IND_REG_11 +0x21B8 VAP_VTX_STATE_IND_REG_12 +0x21BC VAP_VTX_STATE_IND_REG_13 +0x21C0 VAP_VTX_STATE_IND_REG_14 +0x21C4 VAP_VTX_STATE_IND_REG_15 +0x21DC VAP_PSC_SGN_NORM_CNTL +0x21E0 VAP_PROG_STREAM_CNTL_EXT_0 +0x21E4 VAP_PROG_STREAM_CNTL_EXT_1 +0x21E8 VAP_PROG_STREAM_CNTL_EXT_2 +0x21EC VAP_PROG_STREAM_CNTL_EXT_3 +0x21F0 VAP_PROG_STREAM_CNTL_EXT_4 +0x21F4 VAP_PROG_STREAM_CNTL_EXT_5 +0x21F8 VAP_PROG_STREAM_CNTL_EXT_6 +0x21FC VAP_PROG_STREAM_CNTL_EXT_7 +0x2200 VAP_PVS_VECTOR_INDX_REG +0x2204 VAP_PVS_VECTOR_DATA_REG +0x2208 VAP_PVS_VECTOR_DATA_REG_128 +0x2218 VAP_TEX_TO_COLOR_CNTL +0x221C VAP_CLIP_CNTL +0x2220 VAP_GB_VERT_CLIP_ADJ +0x2224 VAP_GB_VERT_DISC_ADJ +0x2228 VAP_GB_HORZ_CLIP_ADJ +0x222C VAP_GB_HORZ_DISC_ADJ +0x2230 VAP_PVS_FLOW_CNTL_ADDRS_0 +0x2234 VAP_PVS_FLOW_CNTL_ADDRS_1 +0x2238 VAP_PVS_FLOW_CNTL_ADDRS_2 +0x223C VAP_PVS_FLOW_CNTL_ADDRS_3 +0x2240 VAP_PVS_FLOW_CNTL_ADDRS_4 +0x2244 VAP_PVS_FLOW_CNTL_ADDRS_5 +0x2248 VAP_PVS_FLOW_CNTL_ADDRS_6 +0x224C VAP_PVS_FLOW_CNTL_ADDRS_7 +0x2250 VAP_PVS_FLOW_CNTL_ADDRS_8 +0x2254 VAP_PVS_FLOW_CNTL_ADDRS_9 +0x2258 VAP_PVS_FLOW_CNTL_ADDRS_10 +0x225C VAP_PVS_FLOW_CNTL_ADDRS_11 +0x2260 VAP_PVS_FLOW_CNTL_ADDRS_12 +0x2264 VAP_PVS_FLOW_CNTL_ADDRS_13 +0x2268 VAP_PVS_FLOW_CNTL_ADDRS_14 +0x226C VAP_PVS_FLOW_CNTL_ADDRS_15 +0x2284 VAP_PVS_STATE_FLUSH_REG +0x2288 VAP_PVS_VTX_TIMEOUT_REG +0x2290 VAP_PVS_FLOW_CNTL_LOOP_INDEX_0 +0x2294 VAP_PVS_FLOW_CNTL_LOOP_INDEX_1 +0x2298 VAP_PVS_FLOW_CNTL_LOOP_INDEX_2 +0x229C VAP_PVS_FLOW_CNTL_LOOP_INDEX_3 +0x22A0 VAP_PVS_FLOW_CNTL_LOOP_INDEX_4 +0x22A4 VAP_PVS_FLOW_CNTL_LOOP_INDEX_5 +0x22A8 VAP_PVS_FLOW_CNTL_LOOP_INDEX_6 +0x22AC VAP_PVS_FLOW_CNTL_LOOP_INDEX_7 +0x22B0 VAP_PVS_FLOW_CNTL_LOOP_INDEX_8 +0x22B4 VAP_PVS_FLOW_CNTL_LOOP_INDEX_9 +0x22B8 VAP_PVS_FLOW_CNTL_LOOP_INDEX_10 +0x22BC VAP_PVS_FLOW_CNTL_LOOP_INDEX_11 +0x22C0 VAP_PVS_FLOW_CNTL_LOOP_INDEX_12 +0x22C4 VAP_PVS_FLOW_CNTL_LOOP_INDEX_13 +0x22C8 VAP_PVS_FLOW_CNTL_LOOP_INDEX_14 +0x22CC VAP_PVS_FLOW_CNTL_LOOP_INDEX_15 +0x22D0 VAP_PVS_CODE_CNTL_0 +0x22D4 VAP_PVS_CONST_CNTL +0x22D8 VAP_PVS_CODE_CNTL_1 +0x22DC VAP_PVS_FLOW_CNTL_OPC +0x2500 VAP_PVS_FLOW_CNTL_ADDRS_LW_0 +0x2504 VAP_PVS_FLOW_CNTL_ADDRS_UW_0 +0x2508 VAP_PVS_FLOW_CNTL_ADDRS_LW_1 +0x250C VAP_PVS_FLOW_CNTL_ADDRS_UW_1 +0x2510 VAP_PVS_FLOW_CNTL_ADDRS_LW_2 +0x2514 VAP_PVS_FLOW_CNTL_ADDRS_UW_2 +0x2518 VAP_PVS_FLOW_CNTL_ADDRS_LW_3 +0x251C VAP_PVS_FLOW_CNTL_ADDRS_UW_3 +0x2520 VAP_PVS_FLOW_CNTL_ADDRS_LW_4 +0x2524 VAP_PVS_FLOW_CNTL_ADDRS_UW_4 +0x2528 VAP_PVS_FLOW_CNTL_ADDRS_LW_5 +0x252C VAP_PVS_FLOW_CNTL_ADDRS_UW_5 +0x2530 VAP_PVS_FLOW_CNTL_ADDRS_LW_6 +0x2534 VAP_PVS_FLOW_CNTL_ADDRS_UW_6 +0x2538 VAP_PVS_FLOW_CNTL_ADDRS_LW_7 +0x253C VAP_PVS_FLOW_CNTL_ADDRS_UW_7 +0x2540 VAP_PVS_FLOW_CNTL_ADDRS_LW_8 +0x2544 VAP_PVS_FLOW_CNTL_ADDRS_UW_8 +0x2548 VAP_PVS_FLOW_CNTL_ADDRS_LW_9 +0x254C VAP_PVS_FLOW_CNTL_ADDRS_UW_9 +0x2550 VAP_PVS_FLOW_CNTL_ADDRS_LW_10 +0x2554 VAP_PVS_FLOW_CNTL_ADDRS_UW_10 +0x2558 VAP_PVS_FLOW_CNTL_ADDRS_LW_11 +0x255C VAP_PVS_FLOW_CNTL_ADDRS_UW_11 +0x2560 VAP_PVS_FLOW_CNTL_ADDRS_LW_12 +0x2564 VAP_PVS_FLOW_CNTL_ADDRS_UW_12 +0x2568 VAP_PVS_FLOW_CNTL_ADDRS_LW_13 +0x256C VAP_PVS_FLOW_CNTL_ADDRS_UW_13 +0x2570 VAP_PVS_FLOW_CNTL_ADDRS_LW_14 +0x2574 VAP_PVS_FLOW_CNTL_ADDRS_UW_14 +0x2578 VAP_PVS_FLOW_CNTL_ADDRS_LW_15 +0x257C VAP_PVS_FLOW_CNTL_ADDRS_UW_15 +0x342C RB2D_DSTCACHE_CTLSTAT +0x4000 GB_VAP_RASTER_VTX_FMT_0 +0x4004 GB_VAP_RASTER_VTX_FMT_1 +0x4008 GB_ENABLE +0x4010 GB_MSPOS0 +0x4014 GB_MSPOS1 +0x401C GB_SELECT +0x4020 GB_AA_CONFIG +0x4024 GB_FIFO_SIZE +0x4100 TX_INVALTAGS +0x4114 SU_TEX_WRAP_PS3 +0x4118 PS3_ENABLE +0x411c PS3_VTX_FMT +0x4120 PS3_TEX_SOURCE +0x4200 GA_POINT_S0 +0x4204 GA_POINT_T0 +0x4208 GA_POINT_S1 +0x420C GA_POINT_T1 +0x4214 GA_TRIANGLE_STIPPLE +0x421C GA_POINT_SIZE +0x4230 GA_POINT_MINMAX +0x4234 GA_LINE_CNTL +0x4238 GA_LINE_STIPPLE_CONFIG +0x4258 GA_COLOR_CONTROL_PS3 +0x4260 GA_LINE_STIPPLE_VALUE +0x4264 GA_LINE_S0 +0x4268 GA_LINE_S1 +0x4278 GA_COLOR_CONTROL +0x427C GA_SOLID_RG +0x4280 GA_SOLID_BA +0x4288 GA_POLY_MODE +0x428C GA_ROUND_MODE +0x4290 GA_OFFSET +0x4294 GA_FOG_SCALE +0x4298 GA_FOG_OFFSET +0x42A0 SU_TEX_WRAP +0x42A4 SU_POLY_OFFSET_FRONT_SCALE +0x42A8 SU_POLY_OFFSET_FRONT_OFFSET +0x42AC SU_POLY_OFFSET_BACK_SCALE +0x42B0 SU_POLY_OFFSET_BACK_OFFSET +0x42B4 SU_POLY_OFFSET_ENABLE +0x42B8 SU_CULL_MODE +0x42C0 SU_DEPTH_SCALE +0x42C4 SU_DEPTH_OFFSET +0x42C8 SU_REG_DEST +0x4300 RS_COUNT +0x4304 RS_INST_COUNT +0x4074 RS_IP_0 +0x4078 RS_IP_1 +0x407C RS_IP_2 +0x4080 RS_IP_3 +0x4084 RS_IP_4 +0x4088 RS_IP_5 +0x408C RS_IP_6 +0x4090 RS_IP_7 +0x4094 RS_IP_8 +0x4098 RS_IP_9 +0x409C RS_IP_10 +0x40A0 RS_IP_11 +0x40A4 RS_IP_12 +0x40A8 RS_IP_13 +0x40AC RS_IP_14 +0x40B0 RS_IP_15 +0x4320 RS_INST_0 +0x4324 RS_INST_1 +0x4328 RS_INST_2 +0x432C RS_INST_3 +0x4330 RS_INST_4 +0x4334 RS_INST_5 +0x4338 RS_INST_6 +0x433C RS_INST_7 +0x4340 RS_INST_8 +0x4344 RS_INST_9 +0x4348 RS_INST_10 +0x434C RS_INST_11 +0x4350 RS_INST_12 +0x4354 RS_INST_13 +0x4358 RS_INST_14 +0x435C RS_INST_15 +0x43A8 SC_EDGERULE +0x43B0 SC_CLIP_0_A +0x43B4 SC_CLIP_0_B +0x43B8 SC_CLIP_1_A +0x43BC SC_CLIP_1_B +0x43C0 SC_CLIP_2_A +0x43C4 SC_CLIP_2_B +0x43C8 SC_CLIP_3_A +0x43CC SC_CLIP_3_B +0x43D0 SC_CLIP_RULE +0x43E0 SC_SCISSOR0 +0x43E8 SC_SCREENDOOR +0x4440 TX_FILTER1_0 +0x4444 TX_FILTER1_1 +0x4448 TX_FILTER1_2 +0x444C TX_FILTER1_3 +0x4450 TX_FILTER1_4 +0x4454 TX_FILTER1_5 +0x4458 TX_FILTER1_6 +0x445C TX_FILTER1_7 +0x4460 TX_FILTER1_8 +0x4464 TX_FILTER1_9 +0x4468 TX_FILTER1_10 +0x446C TX_FILTER1_11 +0x4470 TX_FILTER1_12 +0x4474 TX_FILTER1_13 +0x4478 TX_FILTER1_14 +0x447C TX_FILTER1_15 +0x4580 TX_CHROMA_KEY_0 +0x4584 TX_CHROMA_KEY_1 +0x4588 TX_CHROMA_KEY_2 +0x458C TX_CHROMA_KEY_3 +0x4590 TX_CHROMA_KEY_4 +0x4594 TX_CHROMA_KEY_5 +0x4598 TX_CHROMA_KEY_6 +0x459C TX_CHROMA_KEY_7 +0x45A0 TX_CHROMA_KEY_8 +0x45A4 TX_CHROMA_KEY_9 +0x45A8 TX_CHROMA_KEY_10 +0x45AC TX_CHROMA_KEY_11 +0x45B0 TX_CHROMA_KEY_12 +0x45B4 TX_CHROMA_KEY_13 +0x45B8 TX_CHROMA_KEY_14 +0x45BC TX_CHROMA_KEY_15 +0x45C0 TX_BORDER_COLOR_0 +0x45C4 TX_BORDER_COLOR_1 +0x45C8 TX_BORDER_COLOR_2 +0x45CC TX_BORDER_COLOR_3 +0x45D0 TX_BORDER_COLOR_4 +0x45D4 TX_BORDER_COLOR_5 +0x45D8 TX_BORDER_COLOR_6 +0x45DC TX_BORDER_COLOR_7 +0x45E0 TX_BORDER_COLOR_8 +0x45E4 TX_BORDER_COLOR_9 +0x45E8 TX_BORDER_COLOR_10 +0x45EC TX_BORDER_COLOR_11 +0x45F0 TX_BORDER_COLOR_12 +0x45F4 TX_BORDER_COLOR_13 +0x45F8 TX_BORDER_COLOR_14 +0x45FC TX_BORDER_COLOR_15 +0x4250 GA_US_VECTOR_INDEX +0x4254 GA_US_VECTOR_DATA +0x4600 US_CONFIG +0x4604 US_PIXSIZE +0x4620 US_FC_BOOL_CONST +0x4624 US_FC_CTRL +0x4630 US_CODE_ADDR +0x4634 US_CODE_RANGE +0x4638 US_CODE_OFFSET +0x4640 US_FORMAT0_0 +0x4644 US_FORMAT0_1 +0x4648 US_FORMAT0_2 +0x464C US_FORMAT0_3 +0x4650 US_FORMAT0_4 +0x4654 US_FORMAT0_5 +0x4658 US_FORMAT0_6 +0x465C US_FORMAT0_7 +0x4660 US_FORMAT0_8 +0x4664 US_FORMAT0_9 +0x4668 US_FORMAT0_10 +0x466C US_FORMAT0_11 +0x4670 US_FORMAT0_12 +0x4674 US_FORMAT0_13 +0x4678 US_FORMAT0_14 +0x467C US_FORMAT0_15 +0x46A4 US_OUT_FMT_0 +0x46A8 US_OUT_FMT_1 +0x46AC US_OUT_FMT_2 +0x46B0 US_OUT_FMT_3 +0x46B4 US_W_FMT +0x4BC0 FG_FOG_BLEND +0x4BC4 FG_FOG_FACTOR +0x4BC8 FG_FOG_COLOR_R +0x4BCC FG_FOG_COLOR_G +0x4BD0 FG_FOG_COLOR_B +0x4BD4 FG_ALPHA_FUNC +0x4BD8 FG_DEPTH_SRC +0x4BE0 FG_ALPHA_VALUE +0x4C00 US_ALU_CONST_R_0 +0x4C04 US_ALU_CONST_G_0 +0x4C08 US_ALU_CONST_B_0 +0x4C0C US_ALU_CONST_A_0 +0x4C10 US_ALU_CONST_R_1 +0x4C14 US_ALU_CONST_G_1 +0x4C18 US_ALU_CONST_B_1 +0x4C1C US_ALU_CONST_A_1 +0x4C20 US_ALU_CONST_R_2 +0x4C24 US_ALU_CONST_G_2 +0x4C28 US_ALU_CONST_B_2 +0x4C2C US_ALU_CONST_A_2 +0x4C30 US_ALU_CONST_R_3 +0x4C34 US_ALU_CONST_G_3 +0x4C38 US_ALU_CONST_B_3 +0x4C3C US_ALU_CONST_A_3 +0x4C40 US_ALU_CONST_R_4 +0x4C44 US_ALU_CONST_G_4 +0x4C48 US_ALU_CONST_B_4 +0x4C4C US_ALU_CONST_A_4 +0x4C50 US_ALU_CONST_R_5 +0x4C54 US_ALU_CONST_G_5 +0x4C58 US_ALU_CONST_B_5 +0x4C5C US_ALU_CONST_A_5 +0x4C60 US_ALU_CONST_R_6 +0x4C64 US_ALU_CONST_G_6 +0x4C68 US_ALU_CONST_B_6 +0x4C6C US_ALU_CONST_A_6 +0x4C70 US_ALU_CONST_R_7 +0x4C74 US_ALU_CONST_G_7 +0x4C78 US_ALU_CONST_B_7 +0x4C7C US_ALU_CONST_A_7 +0x4C80 US_ALU_CONST_R_8 +0x4C84 US_ALU_CONST_G_8 +0x4C88 US_ALU_CONST_B_8 +0x4C8C US_ALU_CONST_A_8 +0x4C90 US_ALU_CONST_R_9 +0x4C94 US_ALU_CONST_G_9 +0x4C98 US_ALU_CONST_B_9 +0x4C9C US_ALU_CONST_A_9 +0x4CA0 US_ALU_CONST_R_10 +0x4CA4 US_ALU_CONST_G_10 +0x4CA8 US_ALU_CONST_B_10 +0x4CAC US_ALU_CONST_A_10 +0x4CB0 US_ALU_CONST_R_11 +0x4CB4 US_ALU_CONST_G_11 +0x4CB8 US_ALU_CONST_B_11 +0x4CBC US_ALU_CONST_A_11 +0x4CC0 US_ALU_CONST_R_12 +0x4CC4 US_ALU_CONST_G_12 +0x4CC8 US_ALU_CONST_B_12 +0x4CCC US_ALU_CONST_A_12 +0x4CD0 US_ALU_CONST_R_13 +0x4CD4 US_ALU_CONST_G_13 +0x4CD8 US_ALU_CONST_B_13 +0x4CDC US_ALU_CONST_A_13 +0x4CE0 US_ALU_CONST_R_14 +0x4CE4 US_ALU_CONST_G_14 +0x4CE8 US_ALU_CONST_B_14 +0x4CEC US_ALU_CONST_A_14 +0x4CF0 US_ALU_CONST_R_15 +0x4CF4 US_ALU_CONST_G_15 +0x4CF8 US_ALU_CONST_B_15 +0x4CFC US_ALU_CONST_A_15 +0x4D00 US_ALU_CONST_R_16 +0x4D04 US_ALU_CONST_G_16 +0x4D08 US_ALU_CONST_B_16 +0x4D0C US_ALU_CONST_A_16 +0x4D10 US_ALU_CONST_R_17 +0x4D14 US_ALU_CONST_G_17 +0x4D18 US_ALU_CONST_B_17 +0x4D1C US_ALU_CONST_A_17 +0x4D20 US_ALU_CONST_R_18 +0x4D24 US_ALU_CONST_G_18 +0x4D28 US_ALU_CONST_B_18 +0x4D2C US_ALU_CONST_A_18 +0x4D30 US_ALU_CONST_R_19 +0x4D34 US_ALU_CONST_G_19 +0x4D38 US_ALU_CONST_B_19 +0x4D3C US_ALU_CONST_A_19 +0x4D40 US_ALU_CONST_R_20 +0x4D44 US_ALU_CONST_G_20 +0x4D48 US_ALU_CONST_B_20 +0x4D4C US_ALU_CONST_A_20 +0x4D50 US_ALU_CONST_R_21 +0x4D54 US_ALU_CONST_G_21 +0x4D58 US_ALU_CONST_B_21 +0x4D5C US_ALU_CONST_A_21 +0x4D60 US_ALU_CONST_R_22 +0x4D64 US_ALU_CONST_G_22 +0x4D68 US_ALU_CONST_B_22 +0x4D6C US_ALU_CONST_A_22 +0x4D70 US_ALU_CONST_R_23 +0x4D74 US_ALU_CONST_G_23 +0x4D78 US_ALU_CONST_B_23 +0x4D7C US_ALU_CONST_A_23 +0x4D80 US_ALU_CONST_R_24 +0x4D84 US_ALU_CONST_G_24 +0x4D88 US_ALU_CONST_B_24 +0x4D8C US_ALU_CONST_A_24 +0x4D90 US_ALU_CONST_R_25 +0x4D94 US_ALU_CONST_G_25 +0x4D98 US_ALU_CONST_B_25 +0x4D9C US_ALU_CONST_A_25 +0x4DA0 US_ALU_CONST_R_26 +0x4DA4 US_ALU_CONST_G_26 +0x4DA8 US_ALU_CONST_B_26 +0x4DAC US_ALU_CONST_A_26 +0x4DB0 US_ALU_CONST_R_27 +0x4DB4 US_ALU_CONST_G_27 +0x4DB8 US_ALU_CONST_B_27 +0x4DBC US_ALU_CONST_A_27 +0x4DC0 US_ALU_CONST_R_28 +0x4DC4 US_ALU_CONST_G_28 +0x4DC8 US_ALU_CONST_B_28 +0x4DCC US_ALU_CONST_A_28 +0x4DD0 US_ALU_CONST_R_29 +0x4DD4 US_ALU_CONST_G_29 +0x4DD8 US_ALU_CONST_B_29 +0x4DDC US_ALU_CONST_A_29 +0x4DE0 US_ALU_CONST_R_30 +0x4DE4 US_ALU_CONST_G_30 +0x4DE8 US_ALU_CONST_B_30 +0x4DEC US_ALU_CONST_A_30 +0x4DF0 US_ALU_CONST_R_31 +0x4DF4 US_ALU_CONST_G_31 +0x4DF8 US_ALU_CONST_B_31 +0x4DFC US_ALU_CONST_A_31 +0x4E08 RB3D_ABLENDCNTL_R3 +0x4E10 RB3D_CONSTANT_COLOR +0x4E14 RB3D_COLOR_CLEAR_VALUE +0x4E18 RB3D_ROPCNTL_R3 +0x4E1C RB3D_CLRCMP_FLIPE_R3 +0x4E20 RB3D_CLRCMP_CLR_R3 +0x4E24 RB3D_CLRCMP_MSK_R3 +0x4E48 RB3D_DEBUG_CTL +0x4E4C RB3D_DSTCACHE_CTLSTAT_R3 +0x4E50 RB3D_DITHER_CTL +0x4E54 RB3D_CMASK_OFFSET0 +0x4E58 RB3D_CMASK_OFFSET1 +0x4E5C RB3D_CMASK_OFFSET2 +0x4E60 RB3D_CMASK_OFFSET3 +0x4E64 RB3D_CMASK_PITCH0 +0x4E68 RB3D_CMASK_PITCH1 +0x4E6C RB3D_CMASK_PITCH2 +0x4E70 RB3D_CMASK_PITCH3 +0x4E74 RB3D_CMASK_WRINDEX +0x4E78 RB3D_CMASK_DWORD +0x4E7C RB3D_CMASK_RDINDEX +0x4EA0 RB3D_DISCARD_SRC_PIXEL_LTE_THRESHOLD +0x4EA4 RB3D_DISCARD_SRC_PIXEL_GTE_THRESHOLD +0x4EF8 RB3D_CONSTANT_COLOR_AR +0x4EFC RB3D_CONSTANT_COLOR_GB +0x4F04 ZB_ZSTENCILCNTL +0x4F08 ZB_STENCILREFMASK +0x4F14 ZB_ZTOP +0x4F18 ZB_ZCACHE_CTLSTAT +0x4F58 ZB_ZPASS_DATA +0x4F28 ZB_DEPTHCLEARVALUE +0x4FD4 ZB_STENCILREFMASK_BF diff --git a/drivers/gpu/drm/radeon/rs100d.h b/drivers/gpu/drm/radeon/rs100d.h new file mode 100644 index 00000000..48a913a0 --- /dev/null +++ b/drivers/gpu/drm/radeon/rs100d.h @@ -0,0 +1,40 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#ifndef __RS100D_H__ +#define __RS100D_H__ + +/* Registers */ +#define R_00015C_NB_TOM 0x00015C +#define S_00015C_MC_FB_START(x) (((x) & 0xFFFF) << 0) +#define G_00015C_MC_FB_START(x) (((x) >> 0) & 0xFFFF) +#define C_00015C_MC_FB_START 0xFFFF0000 +#define S_00015C_MC_FB_TOP(x) (((x) & 0xFFFF) << 16) +#define G_00015C_MC_FB_TOP(x) (((x) >> 16) & 0xFFFF) +#define C_00015C_MC_FB_TOP 0x0000FFFF + +#endif diff --git a/drivers/gpu/drm/radeon/rs400.c b/drivers/gpu/drm/radeon/rs400.c new file mode 100644 index 00000000..aa6a66ee --- /dev/null +++ b/drivers/gpu/drm/radeon/rs400.c @@ -0,0 +1,542 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#include +#include +#include +#include "radeon.h" +#include "radeon_asic.h" +#include "rs400d.h" + +/* This files gather functions specifics to : rs400,rs480 */ +static int rs400_debugfs_pcie_gart_info_init(struct radeon_device *rdev); + +void rs400_gart_adjust_size(struct radeon_device *rdev) +{ + /* Check gart size */ + switch (rdev->mc.gtt_size/(1024*1024)) { + case 32: + case 64: + case 128: + case 256: + case 512: + case 1024: + case 2048: + break; + default: + DRM_ERROR("Unable to use IGP GART size %uM\n", + (unsigned)(rdev->mc.gtt_size >> 20)); + DRM_ERROR("Valid GART size for IGP are 32M,64M,128M,256M,512M,1G,2G\n"); + DRM_ERROR("Forcing to 32M GART size\n"); + rdev->mc.gtt_size = 32 * 1024 * 1024; + return; + } +} + +void rs400_gart_tlb_flush(struct radeon_device *rdev) +{ + uint32_t tmp; + unsigned int timeout = rdev->usec_timeout; + + WREG32_MC(RS480_GART_CACHE_CNTRL, RS480_GART_CACHE_INVALIDATE); + do { + tmp = RREG32_MC(RS480_GART_CACHE_CNTRL); + if ((tmp & RS480_GART_CACHE_INVALIDATE) == 0) + break; + DRM_UDELAY(1); + timeout--; + } while (timeout > 0); + WREG32_MC(RS480_GART_CACHE_CNTRL, 0); +} + +int rs400_gart_init(struct radeon_device *rdev) +{ + int r; + + if (rdev->gart.table.ram.ptr) { + WARN(1, "RS400 GART already initialized\n"); + return 0; + } + /* Check gart size */ + switch(rdev->mc.gtt_size / (1024 * 1024)) { + case 32: + case 64: + case 128: + case 256: + case 512: + case 1024: + case 2048: + break; + default: + return -EINVAL; + } + /* Initialize common gart structure */ + r = radeon_gart_init(rdev); + if (r) + return r; + if (rs400_debugfs_pcie_gart_info_init(rdev)) + DRM_ERROR("Failed to register debugfs file for RS400 GART !\n"); + rdev->gart.table_size = rdev->gart.num_gpu_pages * 4; + return radeon_gart_table_ram_alloc(rdev); +} + +int rs400_gart_enable(struct radeon_device *rdev) +{ + uint32_t size_reg; + uint32_t tmp; + + radeon_gart_restore(rdev); + tmp = RREG32_MC(RS690_AIC_CTRL_SCRATCH); + tmp |= RS690_DIS_OUT_OF_PCI_GART_ACCESS; + WREG32_MC(RS690_AIC_CTRL_SCRATCH, tmp); + /* Check gart size */ + switch(rdev->mc.gtt_size / (1024 * 1024)) { + case 32: + size_reg = RS480_VA_SIZE_32MB; + break; + case 64: + size_reg = RS480_VA_SIZE_64MB; + break; + case 128: + size_reg = RS480_VA_SIZE_128MB; + break; + case 256: + size_reg = RS480_VA_SIZE_256MB; + break; + case 512: + size_reg = RS480_VA_SIZE_512MB; + break; + case 1024: + size_reg = RS480_VA_SIZE_1GB; + break; + case 2048: + size_reg = RS480_VA_SIZE_2GB; + break; + default: + return -EINVAL; + } + /* It should be fine to program it to max value */ + if (rdev->family == CHIP_RS690 || (rdev->family == CHIP_RS740)) { + WREG32_MC(RS690_MCCFG_AGP_BASE, 0xFFFFFFFF); + WREG32_MC(RS690_MCCFG_AGP_BASE_2, 0); + } else { + WREG32(RADEON_AGP_BASE, 0xFFFFFFFF); + WREG32(RS480_AGP_BASE_2, 0); + } + tmp = REG_SET(RS690_MC_AGP_TOP, rdev->mc.gtt_end >> 16); + tmp |= REG_SET(RS690_MC_AGP_START, rdev->mc.gtt_start >> 16); + if ((rdev->family == CHIP_RS690) || (rdev->family == CHIP_RS740)) { + WREG32_MC(RS690_MCCFG_AGP_LOCATION, tmp); + tmp = RREG32(RADEON_BUS_CNTL) & ~RS600_BUS_MASTER_DIS; + WREG32(RADEON_BUS_CNTL, tmp); + } else { + WREG32(RADEON_MC_AGP_LOCATION, tmp); + tmp = RREG32(RADEON_BUS_CNTL) & ~RADEON_BUS_MASTER_DIS; + WREG32(RADEON_BUS_CNTL, tmp); + } + /* Table should be in 32bits address space so ignore bits above. */ + tmp = (u32)rdev->gart.table_addr & 0xfffff000; + tmp |= (upper_32_bits(rdev->gart.table_addr) & 0xff) << 4; + + WREG32_MC(RS480_GART_BASE, tmp); + /* TODO: more tweaking here */ + WREG32_MC(RS480_GART_FEATURE_ID, + (RS480_TLB_ENABLE | + RS480_GTW_LAC_EN | RS480_1LEVEL_GART)); + /* Disable snooping */ + WREG32_MC(RS480_AGP_MODE_CNTL, + (1 << RS480_REQ_TYPE_SNOOP_SHIFT) | RS480_REQ_TYPE_SNOOP_DIS); + /* Disable AGP mode */ + /* FIXME: according to doc we should set HIDE_MMCFG_BAR=0, + * AGPMODE30=0 & AGP30ENHANCED=0 in NB_CNTL */ + if ((rdev->family == CHIP_RS690) || (rdev->family == CHIP_RS740)) { + WREG32_MC(RS480_MC_MISC_CNTL, + (RS480_GART_INDEX_REG_EN | RS690_BLOCK_GFX_D3_EN)); + } else { + WREG32_MC(RS480_MC_MISC_CNTL, RS480_GART_INDEX_REG_EN); + } + /* Enable gart */ + WREG32_MC(RS480_AGP_ADDRESS_SPACE_SIZE, (RS480_GART_EN | size_reg)); + rs400_gart_tlb_flush(rdev); + rdev->gart.ready = true; + return 0; +} + +void rs400_gart_disable(struct radeon_device *rdev) +{ + uint32_t tmp; + + tmp = RREG32_MC(RS690_AIC_CTRL_SCRATCH); + tmp |= RS690_DIS_OUT_OF_PCI_GART_ACCESS; + WREG32_MC(RS690_AIC_CTRL_SCRATCH, tmp); + WREG32_MC(RS480_AGP_ADDRESS_SPACE_SIZE, 0); +} + +void rs400_gart_fini(struct radeon_device *rdev) +{ + radeon_gart_fini(rdev); + rs400_gart_disable(rdev); + radeon_gart_table_ram_free(rdev); +} + +#define RS400_PTE_WRITEABLE (1 << 2) +#define RS400_PTE_READABLE (1 << 3) + +int rs400_gart_set_page(struct radeon_device *rdev, int i, uint64_t addr) +{ + uint32_t entry; + + if (i < 0 || i > rdev->gart.num_gpu_pages) { + return -EINVAL; + } + + entry = (lower_32_bits(addr) & PAGE_MASK) | + ((upper_32_bits(addr) & 0xff) << 4) | + RS400_PTE_WRITEABLE | RS400_PTE_READABLE; + entry = cpu_to_le32(entry); + rdev->gart.table.ram.ptr[i] = entry; + return 0; +} + +int rs400_mc_wait_for_idle(struct radeon_device *rdev) +{ + unsigned i; + uint32_t tmp; + + for (i = 0; i < rdev->usec_timeout; i++) { + /* read MC_STATUS */ + tmp = RREG32(RADEON_MC_STATUS); + if (tmp & RADEON_MC_IDLE) { + return 0; + } + DRM_UDELAY(1); + } + return -1; +} + +void rs400_gpu_init(struct radeon_device *rdev) +{ + /* FIXME: is this correct ? */ + r420_pipes_init(rdev); + if (rs400_mc_wait_for_idle(rdev)) { + printk(KERN_WARNING "rs400: Failed to wait MC idle while " + "programming pipes. Bad things might happen. %08x\n", RREG32(RADEON_MC_STATUS)); + } +} + +void rs400_mc_init(struct radeon_device *rdev) +{ + u64 base; + + rs400_gart_adjust_size(rdev); + rdev->mc.igp_sideport_enabled = radeon_combios_sideport_present(rdev); + /* DDR for all card after R300 & IGP */ + rdev->mc.vram_is_ddr = true; + rdev->mc.vram_width = 128; + r100_vram_init_sizes(rdev); + base = (RREG32(RADEON_NB_TOM) & 0xffff) << 16; + radeon_vram_location(rdev, &rdev->mc, base); + rdev->mc.gtt_base_align = rdev->mc.gtt_size - 1; + radeon_gtt_location(rdev, &rdev->mc); + radeon_update_bandwidth_info(rdev); +} + +uint32_t rs400_mc_rreg(struct radeon_device *rdev, uint32_t reg) +{ + uint32_t r; + + WREG32(RS480_NB_MC_INDEX, reg & 0xff); + r = RREG32(RS480_NB_MC_DATA); + WREG32(RS480_NB_MC_INDEX, 0xff); + return r; +} + +void rs400_mc_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v) +{ + WREG32(RS480_NB_MC_INDEX, ((reg) & 0xff) | RS480_NB_MC_IND_WR_EN); + WREG32(RS480_NB_MC_DATA, (v)); + WREG32(RS480_NB_MC_INDEX, 0xff); +} + +#if defined(CONFIG_DEBUG_FS) +static int rs400_debugfs_gart_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + struct radeon_device *rdev = dev->dev_private; + uint32_t tmp; + + tmp = RREG32(RADEON_HOST_PATH_CNTL); + seq_printf(m, "HOST_PATH_CNTL 0x%08x\n", tmp); + tmp = RREG32(RADEON_BUS_CNTL); + seq_printf(m, "BUS_CNTL 0x%08x\n", tmp); + tmp = RREG32_MC(RS690_AIC_CTRL_SCRATCH); + seq_printf(m, "AIC_CTRL_SCRATCH 0x%08x\n", tmp); + if (rdev->family == CHIP_RS690 || (rdev->family == CHIP_RS740)) { + tmp = RREG32_MC(RS690_MCCFG_AGP_BASE); + seq_printf(m, "MCCFG_AGP_BASE 0x%08x\n", tmp); + tmp = RREG32_MC(RS690_MCCFG_AGP_BASE_2); + seq_printf(m, "MCCFG_AGP_BASE_2 0x%08x\n", tmp); + tmp = RREG32_MC(RS690_MCCFG_AGP_LOCATION); + seq_printf(m, "MCCFG_AGP_LOCATION 0x%08x\n", tmp); + tmp = RREG32_MC(RS690_MCCFG_FB_LOCATION); + seq_printf(m, "MCCFG_FB_LOCATION 0x%08x\n", tmp); + tmp = RREG32(RS690_HDP_FB_LOCATION); + seq_printf(m, "HDP_FB_LOCATION 0x%08x\n", tmp); + } else { + tmp = RREG32(RADEON_AGP_BASE); + seq_printf(m, "AGP_BASE 0x%08x\n", tmp); + tmp = RREG32(RS480_AGP_BASE_2); + seq_printf(m, "AGP_BASE_2 0x%08x\n", tmp); + tmp = RREG32(RADEON_MC_AGP_LOCATION); + seq_printf(m, "MC_AGP_LOCATION 0x%08x\n", tmp); + } + tmp = RREG32_MC(RS480_GART_BASE); + seq_printf(m, "GART_BASE 0x%08x\n", tmp); + tmp = RREG32_MC(RS480_GART_FEATURE_ID); + seq_printf(m, "GART_FEATURE_ID 0x%08x\n", tmp); + tmp = RREG32_MC(RS480_AGP_MODE_CNTL); + seq_printf(m, "AGP_MODE_CONTROL 0x%08x\n", tmp); + tmp = RREG32_MC(RS480_MC_MISC_CNTL); + seq_printf(m, "MC_MISC_CNTL 0x%08x\n", tmp); + tmp = RREG32_MC(0x5F); + seq_printf(m, "MC_MISC_UMA_CNTL 0x%08x\n", tmp); + tmp = RREG32_MC(RS480_AGP_ADDRESS_SPACE_SIZE); + seq_printf(m, "AGP_ADDRESS_SPACE_SIZE 0x%08x\n", tmp); + tmp = RREG32_MC(RS480_GART_CACHE_CNTRL); + seq_printf(m, "GART_CACHE_CNTRL 0x%08x\n", tmp); + tmp = RREG32_MC(0x3B); + seq_printf(m, "MC_GART_ERROR_ADDRESS 0x%08x\n", tmp); + tmp = RREG32_MC(0x3C); + seq_printf(m, "MC_GART_ERROR_ADDRESS_HI 0x%08x\n", tmp); + tmp = RREG32_MC(0x30); + seq_printf(m, "GART_ERROR_0 0x%08x\n", tmp); + tmp = RREG32_MC(0x31); + seq_printf(m, "GART_ERROR_1 0x%08x\n", tmp); + tmp = RREG32_MC(0x32); + seq_printf(m, "GART_ERROR_2 0x%08x\n", tmp); + tmp = RREG32_MC(0x33); + seq_printf(m, "GART_ERROR_3 0x%08x\n", tmp); + tmp = RREG32_MC(0x34); + seq_printf(m, "GART_ERROR_4 0x%08x\n", tmp); + tmp = RREG32_MC(0x35); + seq_printf(m, "GART_ERROR_5 0x%08x\n", tmp); + tmp = RREG32_MC(0x36); + seq_printf(m, "GART_ERROR_6 0x%08x\n", tmp); + tmp = RREG32_MC(0x37); + seq_printf(m, "GART_ERROR_7 0x%08x\n", tmp); + return 0; +} + +static struct drm_info_list rs400_gart_info_list[] = { + {"rs400_gart_info", rs400_debugfs_gart_info, 0, NULL}, +}; +#endif + +static int rs400_debugfs_pcie_gart_info_init(struct radeon_device *rdev) +{ +#if defined(CONFIG_DEBUG_FS) + return radeon_debugfs_add_files(rdev, rs400_gart_info_list, 1); +#else + return 0; +#endif +} + +void rs400_mc_program(struct radeon_device *rdev) +{ + struct r100_mc_save save; + + /* Stops all mc clients */ + r100_mc_stop(rdev, &save); + + /* Wait for mc idle */ + if (rs400_mc_wait_for_idle(rdev)) + dev_warn(rdev->dev, "rs400: Wait MC idle timeout before updating MC.\n"); + WREG32(R_000148_MC_FB_LOCATION, + S_000148_MC_FB_START(rdev->mc.vram_start >> 16) | + S_000148_MC_FB_TOP(rdev->mc.vram_end >> 16)); + + r100_mc_resume(rdev, &save); +} + +static int rs400_startup(struct radeon_device *rdev) +{ + int r; + + r100_set_common_regs(rdev); + + rs400_mc_program(rdev); + /* Resume clock */ + r300_clock_startup(rdev); + /* Initialize GPU configuration (# pipes, ...) */ + rs400_gpu_init(rdev); + r100_enable_bm(rdev); + /* Initialize GART (initialize after TTM so we can allocate + * memory through TTM but finalize after TTM) */ + r = rs400_gart_enable(rdev); + if (r) + return r; + + /* allocate wb buffer */ + r = radeon_wb_init(rdev); + if (r) + return r; + + /* Enable IRQ */ + r100_irq_set(rdev); + rdev->config.r300.hdp_cntl = RREG32(RADEON_HOST_PATH_CNTL); + /* 1M ring buffer */ + r = r100_cp_init(rdev, 1024 * 1024); + if (r) { + dev_err(rdev->dev, "failed initializing CP (%d).\n", r); + return r; + } + r = r100_ib_init(rdev); + if (r) { + dev_err(rdev->dev, "failed initializing IB (%d).\n", r); + return r; + } + return 0; +} + +int rs400_resume(struct radeon_device *rdev) +{ + /* Make sur GART are not working */ + rs400_gart_disable(rdev); + /* Resume clock before doing reset */ + r300_clock_startup(rdev); + /* setup MC before calling post tables */ + rs400_mc_program(rdev); + /* Reset gpu before posting otherwise ATOM will enter infinite loop */ + if (radeon_asic_reset(rdev)) { + dev_warn(rdev->dev, "GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n", + RREG32(R_000E40_RBBM_STATUS), + RREG32(R_0007C0_CP_STAT)); + } + /* post */ + radeon_combios_asic_init(rdev->ddev); + /* Resume clock after posting */ + r300_clock_startup(rdev); + /* Initialize surface registers */ + radeon_surface_init(rdev); + return rs400_startup(rdev); +} + +int rs400_suspend(struct radeon_device *rdev) +{ + r100_cp_disable(rdev); + radeon_wb_disable(rdev); + r100_irq_disable(rdev); + rs400_gart_disable(rdev); + return 0; +} + +void rs400_fini(struct radeon_device *rdev) +{ + r100_cp_fini(rdev); + radeon_wb_fini(rdev); + r100_ib_fini(rdev); + radeon_gem_fini(rdev); + rs400_gart_fini(rdev); + radeon_irq_kms_fini(rdev); + radeon_fence_driver_fini(rdev); + radeon_bo_fini(rdev); + radeon_atombios_fini(rdev); + kfree(rdev->bios); + rdev->bios = NULL; +} + +int rs400_init(struct radeon_device *rdev) +{ + int r; + + /* Disable VGA */ + r100_vga_render_disable(rdev); + /* Initialize scratch registers */ + radeon_scratch_init(rdev); + /* Initialize surface registers */ + radeon_surface_init(rdev); + /* TODO: disable VGA need to use VGA request */ + /* restore some register to sane defaults */ + r100_restore_sanity(rdev); + /* BIOS*/ + if (!radeon_get_bios(rdev)) { + if (ASIC_IS_AVIVO(rdev)) + return -EINVAL; + } + if (rdev->is_atom_bios) { + dev_err(rdev->dev, "Expecting combios for RS400/RS480 GPU\n"); + return -EINVAL; + } else { + r = radeon_combios_init(rdev); + if (r) + return r; + } + /* Reset gpu before posting otherwise ATOM will enter infinite loop */ + if (radeon_asic_reset(rdev)) { + dev_warn(rdev->dev, + "GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n", + RREG32(R_000E40_RBBM_STATUS), + RREG32(R_0007C0_CP_STAT)); + } + /* check if cards are posted or not */ + if (radeon_boot_test_post_card(rdev) == false) + return -EINVAL; + + /* Initialize clocks */ + radeon_get_clock_info(rdev->ddev); + /* initialize memory controller */ + rs400_mc_init(rdev); + /* Fence driver */ + r = radeon_fence_driver_init(rdev); + if (r) + return r; + r = radeon_irq_kms_init(rdev); + if (r) + return r; + /* Memory manager */ + r = radeon_bo_init(rdev); + if (r) + return r; + r = rs400_gart_init(rdev); + if (r) + return r; + r300_set_reg_safe(rdev); + rdev->accel_working = true; + r = rs400_startup(rdev); + if (r) { + /* Somethings want wront with the accel init stop accel */ + dev_err(rdev->dev, "Disabling GPU acceleration\n"); + r100_cp_fini(rdev); + radeon_wb_fini(rdev); + r100_ib_fini(rdev); + rs400_gart_fini(rdev); + radeon_irq_kms_fini(rdev); + rdev->accel_working = false; + } + return 0; +} diff --git a/drivers/gpu/drm/radeon/rs400d.h b/drivers/gpu/drm/radeon/rs400d.h new file mode 100644 index 00000000..6d8bac58 --- /dev/null +++ b/drivers/gpu/drm/radeon/rs400d.h @@ -0,0 +1,160 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#ifndef __RS400D_H__ +#define __RS400D_H__ + +/* Registers */ +#define R_000148_MC_FB_LOCATION 0x000148 +#define S_000148_MC_FB_START(x) (((x) & 0xFFFF) << 0) +#define G_000148_MC_FB_START(x) (((x) >> 0) & 0xFFFF) +#define C_000148_MC_FB_START 0xFFFF0000 +#define S_000148_MC_FB_TOP(x) (((x) & 0xFFFF) << 16) +#define G_000148_MC_FB_TOP(x) (((x) >> 16) & 0xFFFF) +#define C_000148_MC_FB_TOP 0x0000FFFF +#define R_00015C_NB_TOM 0x00015C +#define S_00015C_MC_FB_START(x) (((x) & 0xFFFF) << 0) +#define G_00015C_MC_FB_START(x) (((x) >> 0) & 0xFFFF) +#define C_00015C_MC_FB_START 0xFFFF0000 +#define S_00015C_MC_FB_TOP(x) (((x) & 0xFFFF) << 16) +#define G_00015C_MC_FB_TOP(x) (((x) >> 16) & 0xFFFF) +#define C_00015C_MC_FB_TOP 0x0000FFFF +#define R_0007C0_CP_STAT 0x0007C0 +#define S_0007C0_MRU_BUSY(x) (((x) & 0x1) << 0) +#define G_0007C0_MRU_BUSY(x) (((x) >> 0) & 0x1) +#define C_0007C0_MRU_BUSY 0xFFFFFFFE +#define S_0007C0_MWU_BUSY(x) (((x) & 0x1) << 1) +#define G_0007C0_MWU_BUSY(x) (((x) >> 1) & 0x1) +#define C_0007C0_MWU_BUSY 0xFFFFFFFD +#define S_0007C0_RSIU_BUSY(x) (((x) & 0x1) << 2) +#define G_0007C0_RSIU_BUSY(x) (((x) >> 2) & 0x1) +#define C_0007C0_RSIU_BUSY 0xFFFFFFFB +#define S_0007C0_RCIU_BUSY(x) (((x) & 0x1) << 3) +#define G_0007C0_RCIU_BUSY(x) (((x) >> 3) & 0x1) +#define C_0007C0_RCIU_BUSY 0xFFFFFFF7 +#define S_0007C0_CSF_PRIMARY_BUSY(x) (((x) & 0x1) << 9) +#define G_0007C0_CSF_PRIMARY_BUSY(x) (((x) >> 9) & 0x1) +#define C_0007C0_CSF_PRIMARY_BUSY 0xFFFFFDFF +#define S_0007C0_CSF_INDIRECT_BUSY(x) (((x) & 0x1) << 10) +#define G_0007C0_CSF_INDIRECT_BUSY(x) (((x) >> 10) & 0x1) +#define C_0007C0_CSF_INDIRECT_BUSY 0xFFFFFBFF +#define S_0007C0_CSQ_PRIMARY_BUSY(x) (((x) & 0x1) << 11) +#define G_0007C0_CSQ_PRIMARY_BUSY(x) (((x) >> 11) & 0x1) +#define C_0007C0_CSQ_PRIMARY_BUSY 0xFFFFF7FF +#define S_0007C0_CSQ_INDIRECT_BUSY(x) (((x) & 0x1) << 12) +#define G_0007C0_CSQ_INDIRECT_BUSY(x) (((x) >> 12) & 0x1) +#define C_0007C0_CSQ_INDIRECT_BUSY 0xFFFFEFFF +#define S_0007C0_CSI_BUSY(x) (((x) & 0x1) << 13) +#define G_0007C0_CSI_BUSY(x) (((x) >> 13) & 0x1) +#define C_0007C0_CSI_BUSY 0xFFFFDFFF +#define S_0007C0_CSF_INDIRECT2_BUSY(x) (((x) & 0x1) << 14) +#define G_0007C0_CSF_INDIRECT2_BUSY(x) (((x) >> 14) & 0x1) +#define C_0007C0_CSF_INDIRECT2_BUSY 0xFFFFBFFF +#define S_0007C0_CSQ_INDIRECT2_BUSY(x) (((x) & 0x1) << 15) +#define G_0007C0_CSQ_INDIRECT2_BUSY(x) (((x) >> 15) & 0x1) +#define C_0007C0_CSQ_INDIRECT2_BUSY 0xFFFF7FFF +#define S_0007C0_GUIDMA_BUSY(x) (((x) & 0x1) << 28) +#define G_0007C0_GUIDMA_BUSY(x) (((x) >> 28) & 0x1) +#define C_0007C0_GUIDMA_BUSY 0xEFFFFFFF +#define S_0007C0_VIDDMA_BUSY(x) (((x) & 0x1) << 29) +#define G_0007C0_VIDDMA_BUSY(x) (((x) >> 29) & 0x1) +#define C_0007C0_VIDDMA_BUSY 0xDFFFFFFF +#define S_0007C0_CMDSTRM_BUSY(x) (((x) & 0x1) << 30) +#define G_0007C0_CMDSTRM_BUSY(x) (((x) >> 30) & 0x1) +#define C_0007C0_CMDSTRM_BUSY 0xBFFFFFFF +#define S_0007C0_CP_BUSY(x) (((x) & 0x1) << 31) +#define G_0007C0_CP_BUSY(x) (((x) >> 31) & 0x1) +#define C_0007C0_CP_BUSY 0x7FFFFFFF +#define R_000E40_RBBM_STATUS 0x000E40 +#define S_000E40_CMDFIFO_AVAIL(x) (((x) & 0x7F) << 0) +#define G_000E40_CMDFIFO_AVAIL(x) (((x) >> 0) & 0x7F) +#define C_000E40_CMDFIFO_AVAIL 0xFFFFFF80 +#define S_000E40_HIRQ_ON_RBB(x) (((x) & 0x1) << 8) +#define G_000E40_HIRQ_ON_RBB(x) (((x) >> 8) & 0x1) +#define C_000E40_HIRQ_ON_RBB 0xFFFFFEFF +#define S_000E40_CPRQ_ON_RBB(x) (((x) & 0x1) << 9) +#define G_000E40_CPRQ_ON_RBB(x) (((x) >> 9) & 0x1) +#define C_000E40_CPRQ_ON_RBB 0xFFFFFDFF +#define S_000E40_CFRQ_ON_RBB(x) (((x) & 0x1) << 10) +#define G_000E40_CFRQ_ON_RBB(x) (((x) >> 10) & 0x1) +#define C_000E40_CFRQ_ON_RBB 0xFFFFFBFF +#define S_000E40_HIRQ_IN_RTBUF(x) (((x) & 0x1) << 11) +#define G_000E40_HIRQ_IN_RTBUF(x) (((x) >> 11) & 0x1) +#define C_000E40_HIRQ_IN_RTBUF 0xFFFFF7FF +#define S_000E40_CPRQ_IN_RTBUF(x) (((x) & 0x1) << 12) +#define G_000E40_CPRQ_IN_RTBUF(x) (((x) >> 12) & 0x1) +#define C_000E40_CPRQ_IN_RTBUF 0xFFFFEFFF +#define S_000E40_CFRQ_IN_RTBUF(x) (((x) & 0x1) << 13) +#define G_000E40_CFRQ_IN_RTBUF(x) (((x) >> 13) & 0x1) +#define C_000E40_CFRQ_IN_RTBUF 0xFFFFDFFF +#define S_000E40_CF_PIPE_BUSY(x) (((x) & 0x1) << 14) +#define G_000E40_CF_PIPE_BUSY(x) (((x) >> 14) & 0x1) +#define C_000E40_CF_PIPE_BUSY 0xFFFFBFFF +#define S_000E40_ENG_EV_BUSY(x) (((x) & 0x1) << 15) +#define G_000E40_ENG_EV_BUSY(x) (((x) >> 15) & 0x1) +#define C_000E40_ENG_EV_BUSY 0xFFFF7FFF +#define S_000E40_CP_CMDSTRM_BUSY(x) (((x) & 0x1) << 16) +#define G_000E40_CP_CMDSTRM_BUSY(x) (((x) >> 16) & 0x1) +#define C_000E40_CP_CMDSTRM_BUSY 0xFFFEFFFF +#define S_000E40_E2_BUSY(x) (((x) & 0x1) << 17) +#define G_000E40_E2_BUSY(x) (((x) >> 17) & 0x1) +#define C_000E40_E2_BUSY 0xFFFDFFFF +#define S_000E40_RB2D_BUSY(x) (((x) & 0x1) << 18) +#define G_000E40_RB2D_BUSY(x) (((x) >> 18) & 0x1) +#define C_000E40_RB2D_BUSY 0xFFFBFFFF +#define S_000E40_RB3D_BUSY(x) (((x) & 0x1) << 19) +#define G_000E40_RB3D_BUSY(x) (((x) >> 19) & 0x1) +#define C_000E40_RB3D_BUSY 0xFFF7FFFF +#define S_000E40_VAP_BUSY(x) (((x) & 0x1) << 20) +#define G_000E40_VAP_BUSY(x) (((x) >> 20) & 0x1) +#define C_000E40_VAP_BUSY 0xFFEFFFFF +#define S_000E40_RE_BUSY(x) (((x) & 0x1) << 21) +#define G_000E40_RE_BUSY(x) (((x) >> 21) & 0x1) +#define C_000E40_RE_BUSY 0xFFDFFFFF +#define S_000E40_TAM_BUSY(x) (((x) & 0x1) << 22) +#define G_000E40_TAM_BUSY(x) (((x) >> 22) & 0x1) +#define C_000E40_TAM_BUSY 0xFFBFFFFF +#define S_000E40_TDM_BUSY(x) (((x) & 0x1) << 23) +#define G_000E40_TDM_BUSY(x) (((x) >> 23) & 0x1) +#define C_000E40_TDM_BUSY 0xFF7FFFFF +#define S_000E40_PB_BUSY(x) (((x) & 0x1) << 24) +#define G_000E40_PB_BUSY(x) (((x) >> 24) & 0x1) +#define C_000E40_PB_BUSY 0xFEFFFFFF +#define S_000E40_TIM_BUSY(x) (((x) & 0x1) << 25) +#define G_000E40_TIM_BUSY(x) (((x) >> 25) & 0x1) +#define C_000E40_TIM_BUSY 0xFDFFFFFF +#define S_000E40_GA_BUSY(x) (((x) & 0x1) << 26) +#define G_000E40_GA_BUSY(x) (((x) >> 26) & 0x1) +#define C_000E40_GA_BUSY 0xFBFFFFFF +#define S_000E40_CBA2D_BUSY(x) (((x) & 0x1) << 27) +#define G_000E40_CBA2D_BUSY(x) (((x) >> 27) & 0x1) +#define C_000E40_CBA2D_BUSY 0xF7FFFFFF +#define S_000E40_GUI_ACTIVE(x) (((x) & 0x1) << 31) +#define G_000E40_GUI_ACTIVE(x) (((x) >> 31) & 0x1) +#define C_000E40_GUI_ACTIVE 0x7FFFFFFF + +#endif diff --git a/drivers/gpu/drm/radeon/rs600.c b/drivers/gpu/drm/radeon/rs600.c new file mode 100644 index 00000000..2026c2d5 --- /dev/null +++ b/drivers/gpu/drm/radeon/rs600.c @@ -0,0 +1,995 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +/* RS600 / Radeon X1250/X1270 integrated GPU + * + * This file gather function specific to RS600 which is the IGP of + * the X1250/X1270 family supporting intel CPU (while RS690/RS740 + * is the X1250/X1270 supporting AMD CPU). The display engine are + * the avivo one, bios is an atombios, 3D block are the one of the + * R4XX family. The GART is different from the RS400 one and is very + * close to the one of the R600 family (R600 likely being an evolution + * of the RS600 GART block). + */ +#include "drmP.h" +#include "radeon.h" +#include "radeon_asic.h" +#include "atom.h" +#include "rs600d.h" + +#include "rs600_reg_safe.h" + +void rs600_gpu_init(struct radeon_device *rdev); +int rs600_mc_wait_for_idle(struct radeon_device *rdev); + +void rs600_pre_page_flip(struct radeon_device *rdev, int crtc) +{ + /* enable the pflip int */ + radeon_irq_kms_pflip_irq_get(rdev, crtc); +} + +void rs600_post_page_flip(struct radeon_device *rdev, int crtc) +{ + /* disable the pflip int */ + radeon_irq_kms_pflip_irq_put(rdev, crtc); +} + +u32 rs600_page_flip(struct radeon_device *rdev, int crtc_id, u64 crtc_base) +{ + struct radeon_crtc *radeon_crtc = rdev->mode_info.crtcs[crtc_id]; + u32 tmp = RREG32(AVIVO_D1GRPH_UPDATE + radeon_crtc->crtc_offset); + int i; + + /* Lock the graphics update lock */ + tmp |= AVIVO_D1GRPH_UPDATE_LOCK; + WREG32(AVIVO_D1GRPH_UPDATE + radeon_crtc->crtc_offset, tmp); + + /* update the scanout addresses */ + WREG32(AVIVO_D1GRPH_SECONDARY_SURFACE_ADDRESS + radeon_crtc->crtc_offset, + (u32)crtc_base); + WREG32(AVIVO_D1GRPH_PRIMARY_SURFACE_ADDRESS + radeon_crtc->crtc_offset, + (u32)crtc_base); + + /* Wait for update_pending to go high. */ + for (i = 0; i < rdev->usec_timeout; i++) { + if (RREG32(AVIVO_D1GRPH_UPDATE + radeon_crtc->crtc_offset) & AVIVO_D1GRPH_SURFACE_UPDATE_PENDING) + break; + udelay(1); + } + DRM_DEBUG("Update pending now high. Unlocking vupdate_lock.\n"); + + /* Unlock the lock, so double-buffering can take place inside vblank */ + tmp &= ~AVIVO_D1GRPH_UPDATE_LOCK; + WREG32(AVIVO_D1GRPH_UPDATE + radeon_crtc->crtc_offset, tmp); + + /* Return current update_pending status: */ + return RREG32(AVIVO_D1GRPH_UPDATE + radeon_crtc->crtc_offset) & AVIVO_D1GRPH_SURFACE_UPDATE_PENDING; +} + +void rs600_pm_misc(struct radeon_device *rdev) +{ + int requested_index = rdev->pm.requested_power_state_index; + struct radeon_power_state *ps = &rdev->pm.power_state[requested_index]; + struct radeon_voltage *voltage = &ps->clock_info[0].voltage; + u32 tmp, dyn_pwrmgt_sclk_length, dyn_sclk_vol_cntl; + u32 hdp_dyn_cntl, /*mc_host_dyn_cntl,*/ dyn_backbias_cntl; + + if ((voltage->type == VOLTAGE_GPIO) && (voltage->gpio.valid)) { + if (ps->misc & ATOM_PM_MISCINFO_VOLTAGE_DROP_SUPPORT) { + tmp = RREG32(voltage->gpio.reg); + if (voltage->active_high) + tmp |= voltage->gpio.mask; + else + tmp &= ~(voltage->gpio.mask); + WREG32(voltage->gpio.reg, tmp); + if (voltage->delay) + udelay(voltage->delay); + } else { + tmp = RREG32(voltage->gpio.reg); + if (voltage->active_high) + tmp &= ~voltage->gpio.mask; + else + tmp |= voltage->gpio.mask; + WREG32(voltage->gpio.reg, tmp); + if (voltage->delay) + udelay(voltage->delay); + } + } else if (voltage->type == VOLTAGE_VDDC) + radeon_atom_set_voltage(rdev, voltage->vddc_id, SET_VOLTAGE_TYPE_ASIC_VDDC); + + dyn_pwrmgt_sclk_length = RREG32_PLL(DYN_PWRMGT_SCLK_LENGTH); + dyn_pwrmgt_sclk_length &= ~REDUCED_POWER_SCLK_HILEN(0xf); + dyn_pwrmgt_sclk_length &= ~REDUCED_POWER_SCLK_LOLEN(0xf); + if (ps->misc & ATOM_PM_MISCINFO_ASIC_REDUCED_SPEED_SCLK_EN) { + if (ps->misc & ATOM_PM_MISCINFO_DYNAMIC_CLOCK_DIVIDER_BY_2) { + dyn_pwrmgt_sclk_length |= REDUCED_POWER_SCLK_HILEN(2); + dyn_pwrmgt_sclk_length |= REDUCED_POWER_SCLK_LOLEN(2); + } else if (ps->misc & ATOM_PM_MISCINFO_DYNAMIC_CLOCK_DIVIDER_BY_4) { + dyn_pwrmgt_sclk_length |= REDUCED_POWER_SCLK_HILEN(4); + dyn_pwrmgt_sclk_length |= REDUCED_POWER_SCLK_LOLEN(4); + } + } else { + dyn_pwrmgt_sclk_length |= REDUCED_POWER_SCLK_HILEN(1); + dyn_pwrmgt_sclk_length |= REDUCED_POWER_SCLK_LOLEN(1); + } + WREG32_PLL(DYN_PWRMGT_SCLK_LENGTH, dyn_pwrmgt_sclk_length); + + dyn_sclk_vol_cntl = RREG32_PLL(DYN_SCLK_VOL_CNTL); + if (ps->misc & ATOM_PM_MISCINFO_ASIC_DYNAMIC_VOLTAGE_EN) { + dyn_sclk_vol_cntl |= IO_CG_VOLTAGE_DROP; + if (voltage->delay) { + dyn_sclk_vol_cntl |= VOLTAGE_DROP_SYNC; + dyn_sclk_vol_cntl |= VOLTAGE_DELAY_SEL(voltage->delay); + } else + dyn_sclk_vol_cntl &= ~VOLTAGE_DROP_SYNC; + } else + dyn_sclk_vol_cntl &= ~IO_CG_VOLTAGE_DROP; + WREG32_PLL(DYN_SCLK_VOL_CNTL, dyn_sclk_vol_cntl); + + hdp_dyn_cntl = RREG32_PLL(HDP_DYN_CNTL); + if (ps->misc & ATOM_PM_MISCINFO_DYNAMIC_HDP_BLOCK_EN) + hdp_dyn_cntl &= ~HDP_FORCEON; + else + hdp_dyn_cntl |= HDP_FORCEON; + WREG32_PLL(HDP_DYN_CNTL, hdp_dyn_cntl); +#if 0 + /* mc_host_dyn seems to cause hangs from time to time */ + mc_host_dyn_cntl = RREG32_PLL(MC_HOST_DYN_CNTL); + if (ps->misc & ATOM_PM_MISCINFO_DYNAMIC_MC_HOST_BLOCK_EN) + mc_host_dyn_cntl &= ~MC_HOST_FORCEON; + else + mc_host_dyn_cntl |= MC_HOST_FORCEON; + WREG32_PLL(MC_HOST_DYN_CNTL, mc_host_dyn_cntl); +#endif + dyn_backbias_cntl = RREG32_PLL(DYN_BACKBIAS_CNTL); + if (ps->misc & ATOM_PM_MISCINFO2_DYNAMIC_BACK_BIAS_EN) + dyn_backbias_cntl |= IO_CG_BACKBIAS_EN; + else + dyn_backbias_cntl &= ~IO_CG_BACKBIAS_EN; + WREG32_PLL(DYN_BACKBIAS_CNTL, dyn_backbias_cntl); + + /* set pcie lanes */ + if ((rdev->flags & RADEON_IS_PCIE) && + !(rdev->flags & RADEON_IS_IGP) && + rdev->asic->set_pcie_lanes && + (ps->pcie_lanes != + rdev->pm.power_state[rdev->pm.current_power_state_index].pcie_lanes)) { + radeon_set_pcie_lanes(rdev, + ps->pcie_lanes); + DRM_DEBUG("Setting: p: %d\n", ps->pcie_lanes); + } +} + +void rs600_pm_prepare(struct radeon_device *rdev) +{ + struct drm_device *ddev = rdev->ddev; + struct drm_crtc *crtc; + struct radeon_crtc *radeon_crtc; + u32 tmp; + + /* disable any active CRTCs */ + list_for_each_entry(crtc, &ddev->mode_config.crtc_list, head) { + radeon_crtc = to_radeon_crtc(crtc); + if (radeon_crtc->enabled) { + tmp = RREG32(AVIVO_D1CRTC_CONTROL + radeon_crtc->crtc_offset); + tmp |= AVIVO_CRTC_DISP_READ_REQUEST_DISABLE; + WREG32(AVIVO_D1CRTC_CONTROL + radeon_crtc->crtc_offset, tmp); + } + } +} + +void rs600_pm_finish(struct radeon_device *rdev) +{ + struct drm_device *ddev = rdev->ddev; + struct drm_crtc *crtc; + struct radeon_crtc *radeon_crtc; + u32 tmp; + + /* enable any active CRTCs */ + list_for_each_entry(crtc, &ddev->mode_config.crtc_list, head) { + radeon_crtc = to_radeon_crtc(crtc); + if (radeon_crtc->enabled) { + tmp = RREG32(AVIVO_D1CRTC_CONTROL + radeon_crtc->crtc_offset); + tmp &= ~AVIVO_CRTC_DISP_READ_REQUEST_DISABLE; + WREG32(AVIVO_D1CRTC_CONTROL + radeon_crtc->crtc_offset, tmp); + } + } +} + +/* hpd for digital panel detect/disconnect */ +bool rs600_hpd_sense(struct radeon_device *rdev, enum radeon_hpd_id hpd) +{ + u32 tmp; + bool connected = false; + + switch (hpd) { + case RADEON_HPD_1: + tmp = RREG32(R_007D04_DC_HOT_PLUG_DETECT1_INT_STATUS); + if (G_007D04_DC_HOT_PLUG_DETECT1_SENSE(tmp)) + connected = true; + break; + case RADEON_HPD_2: + tmp = RREG32(R_007D14_DC_HOT_PLUG_DETECT2_INT_STATUS); + if (G_007D14_DC_HOT_PLUG_DETECT2_SENSE(tmp)) + connected = true; + break; + default: + break; + } + return connected; +} + +void rs600_hpd_set_polarity(struct radeon_device *rdev, + enum radeon_hpd_id hpd) +{ + u32 tmp; + bool connected = rs600_hpd_sense(rdev, hpd); + + switch (hpd) { + case RADEON_HPD_1: + tmp = RREG32(R_007D08_DC_HOT_PLUG_DETECT1_INT_CONTROL); + if (connected) + tmp &= ~S_007D08_DC_HOT_PLUG_DETECT1_INT_POLARITY(1); + else + tmp |= S_007D08_DC_HOT_PLUG_DETECT1_INT_POLARITY(1); + WREG32(R_007D08_DC_HOT_PLUG_DETECT1_INT_CONTROL, tmp); + break; + case RADEON_HPD_2: + tmp = RREG32(R_007D18_DC_HOT_PLUG_DETECT2_INT_CONTROL); + if (connected) + tmp &= ~S_007D18_DC_HOT_PLUG_DETECT2_INT_POLARITY(1); + else + tmp |= S_007D18_DC_HOT_PLUG_DETECT2_INT_POLARITY(1); + WREG32(R_007D18_DC_HOT_PLUG_DETECT2_INT_CONTROL, tmp); + break; + default: + break; + } +} + +void rs600_hpd_init(struct radeon_device *rdev) +{ + struct drm_device *dev = rdev->ddev; + struct drm_connector *connector; + + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + switch (radeon_connector->hpd.hpd) { + case RADEON_HPD_1: + WREG32(R_007D00_DC_HOT_PLUG_DETECT1_CONTROL, + S_007D00_DC_HOT_PLUG_DETECT1_EN(1)); + rdev->irq.hpd[0] = true; + break; + case RADEON_HPD_2: + WREG32(R_007D10_DC_HOT_PLUG_DETECT2_CONTROL, + S_007D10_DC_HOT_PLUG_DETECT2_EN(1)); + rdev->irq.hpd[1] = true; + break; + default: + break; + } + radeon_hpd_set_polarity(rdev, radeon_connector->hpd.hpd); + } + if (rdev->irq.installed) + rs600_irq_set(rdev); +} + +void rs600_hpd_fini(struct radeon_device *rdev) +{ + struct drm_device *dev = rdev->ddev; + struct drm_connector *connector; + + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + switch (radeon_connector->hpd.hpd) { + case RADEON_HPD_1: + WREG32(R_007D00_DC_HOT_PLUG_DETECT1_CONTROL, + S_007D00_DC_HOT_PLUG_DETECT1_EN(0)); + rdev->irq.hpd[0] = false; + break; + case RADEON_HPD_2: + WREG32(R_007D10_DC_HOT_PLUG_DETECT2_CONTROL, + S_007D10_DC_HOT_PLUG_DETECT2_EN(0)); + rdev->irq.hpd[1] = false; + break; + default: + break; + } + } +} + +void rs600_bm_disable(struct radeon_device *rdev) +{ + u16 tmp; + + /* disable bus mastering */ + pci_read_config_word(rdev->pdev, 0x4, &tmp); + pci_write_config_word(rdev->pdev, 0x4, tmp & 0xFFFB); + mdelay(1); +} + +int rs600_asic_reset(struct radeon_device *rdev) +{ + struct rv515_mc_save save; + u32 status, tmp; + int ret = 0; + + status = RREG32(R_000E40_RBBM_STATUS); + if (!G_000E40_GUI_ACTIVE(status)) { + return 0; + } + /* Stops all mc clients */ + rv515_mc_stop(rdev, &save); + status = RREG32(R_000E40_RBBM_STATUS); + dev_info(rdev->dev, "(%s:%d) RBBM_STATUS=0x%08X\n", __func__, __LINE__, status); + /* stop CP */ + WREG32(RADEON_CP_CSQ_CNTL, 0); + tmp = RREG32(RADEON_CP_RB_CNTL); + WREG32(RADEON_CP_RB_CNTL, tmp | RADEON_RB_RPTR_WR_ENA); + WREG32(RADEON_CP_RB_RPTR_WR, 0); + WREG32(RADEON_CP_RB_WPTR, 0); + WREG32(RADEON_CP_RB_CNTL, tmp); + pci_save_state(rdev->pdev); + /* disable bus mastering */ + rs600_bm_disable(rdev); + /* reset GA+VAP */ + WREG32(R_0000F0_RBBM_SOFT_RESET, S_0000F0_SOFT_RESET_VAP(1) | + S_0000F0_SOFT_RESET_GA(1)); + RREG32(R_0000F0_RBBM_SOFT_RESET); + mdelay(500); + WREG32(R_0000F0_RBBM_SOFT_RESET, 0); + mdelay(1); + status = RREG32(R_000E40_RBBM_STATUS); + dev_info(rdev->dev, "(%s:%d) RBBM_STATUS=0x%08X\n", __func__, __LINE__, status); + /* reset CP */ + WREG32(R_0000F0_RBBM_SOFT_RESET, S_0000F0_SOFT_RESET_CP(1)); + RREG32(R_0000F0_RBBM_SOFT_RESET); + mdelay(500); + WREG32(R_0000F0_RBBM_SOFT_RESET, 0); + mdelay(1); + status = RREG32(R_000E40_RBBM_STATUS); + dev_info(rdev->dev, "(%s:%d) RBBM_STATUS=0x%08X\n", __func__, __LINE__, status); + /* reset MC */ + WREG32(R_0000F0_RBBM_SOFT_RESET, S_0000F0_SOFT_RESET_MC(1)); + RREG32(R_0000F0_RBBM_SOFT_RESET); + mdelay(500); + WREG32(R_0000F0_RBBM_SOFT_RESET, 0); + mdelay(1); + status = RREG32(R_000E40_RBBM_STATUS); + dev_info(rdev->dev, "(%s:%d) RBBM_STATUS=0x%08X\n", __func__, __LINE__, status); + /* restore PCI & busmastering */ + pci_restore_state(rdev->pdev); + /* Check if GPU is idle */ + if (G_000E40_GA_BUSY(status) || G_000E40_VAP_BUSY(status)) { + dev_err(rdev->dev, "failed to reset GPU\n"); + rdev->gpu_lockup = true; + ret = -1; + } else + dev_info(rdev->dev, "GPU reset succeed\n"); + rv515_mc_resume(rdev, &save); + return ret; +} + +/* + * GART. + */ +void rs600_gart_tlb_flush(struct radeon_device *rdev) +{ + uint32_t tmp; + + tmp = RREG32_MC(R_000100_MC_PT0_CNTL); + tmp &= C_000100_INVALIDATE_ALL_L1_TLBS & C_000100_INVALIDATE_L2_CACHE; + WREG32_MC(R_000100_MC_PT0_CNTL, tmp); + + tmp = RREG32_MC(R_000100_MC_PT0_CNTL); + tmp |= S_000100_INVALIDATE_ALL_L1_TLBS(1) | S_000100_INVALIDATE_L2_CACHE(1); + WREG32_MC(R_000100_MC_PT0_CNTL, tmp); + + tmp = RREG32_MC(R_000100_MC_PT0_CNTL); + tmp &= C_000100_INVALIDATE_ALL_L1_TLBS & C_000100_INVALIDATE_L2_CACHE; + WREG32_MC(R_000100_MC_PT0_CNTL, tmp); + tmp = RREG32_MC(R_000100_MC_PT0_CNTL); +} + +int rs600_gart_init(struct radeon_device *rdev) +{ + int r; + + if (rdev->gart.table.vram.robj) { + WARN(1, "RS600 GART already initialized\n"); + return 0; + } + /* Initialize common gart structure */ + r = radeon_gart_init(rdev); + if (r) { + return r; + } + rdev->gart.table_size = rdev->gart.num_gpu_pages * 8; + return radeon_gart_table_vram_alloc(rdev); +} + +static int rs600_gart_enable(struct radeon_device *rdev) +{ + u32 tmp; + int r, i; + + if (rdev->gart.table.vram.robj == NULL) { + dev_err(rdev->dev, "No VRAM object for PCIE GART.\n"); + return -EINVAL; + } + r = radeon_gart_table_vram_pin(rdev); + if (r) + return r; + radeon_gart_restore(rdev); + /* Enable bus master */ + tmp = RREG32(RADEON_BUS_CNTL) & ~RS600_BUS_MASTER_DIS; + WREG32(RADEON_BUS_CNTL, tmp); + /* FIXME: setup default page */ + WREG32_MC(R_000100_MC_PT0_CNTL, + (S_000100_EFFECTIVE_L2_CACHE_SIZE(6) | + S_000100_EFFECTIVE_L2_QUEUE_SIZE(6))); + + for (i = 0; i < 19; i++) { + WREG32_MC(R_00016C_MC_PT0_CLIENT0_CNTL + i, + S_00016C_ENABLE_TRANSLATION_MODE_OVERRIDE(1) | + S_00016C_SYSTEM_ACCESS_MODE_MASK( + V_00016C_SYSTEM_ACCESS_MODE_NOT_IN_SYS) | + S_00016C_SYSTEM_APERTURE_UNMAPPED_ACCESS( + V_00016C_SYSTEM_APERTURE_UNMAPPED_PASSTHROUGH) | + S_00016C_EFFECTIVE_L1_CACHE_SIZE(3) | + S_00016C_ENABLE_FRAGMENT_PROCESSING(1) | + S_00016C_EFFECTIVE_L1_QUEUE_SIZE(3)); + } + /* enable first context */ + WREG32_MC(R_000102_MC_PT0_CONTEXT0_CNTL, + S_000102_ENABLE_PAGE_TABLE(1) | + S_000102_PAGE_TABLE_DEPTH(V_000102_PAGE_TABLE_FLAT)); + + /* disable all other contexts */ + for (i = 1; i < 8; i++) + WREG32_MC(R_000102_MC_PT0_CONTEXT0_CNTL + i, 0); + + /* setup the page table */ + WREG32_MC(R_00012C_MC_PT0_CONTEXT0_FLAT_BASE_ADDR, + rdev->gart.table_addr); + WREG32_MC(R_00013C_MC_PT0_CONTEXT0_FLAT_START_ADDR, rdev->mc.gtt_start); + WREG32_MC(R_00014C_MC_PT0_CONTEXT0_FLAT_END_ADDR, rdev->mc.gtt_end); + WREG32_MC(R_00011C_MC_PT0_CONTEXT0_DEFAULT_READ_ADDR, 0); + + /* System context maps to VRAM space */ + WREG32_MC(R_000112_MC_PT0_SYSTEM_APERTURE_LOW_ADDR, rdev->mc.vram_start); + WREG32_MC(R_000114_MC_PT0_SYSTEM_APERTURE_HIGH_ADDR, rdev->mc.vram_end); + + /* enable page tables */ + tmp = RREG32_MC(R_000100_MC_PT0_CNTL); + WREG32_MC(R_000100_MC_PT0_CNTL, (tmp | S_000100_ENABLE_PT(1))); + tmp = RREG32_MC(R_000009_MC_CNTL1); + WREG32_MC(R_000009_MC_CNTL1, (tmp | S_000009_ENABLE_PAGE_TABLES(1))); + rs600_gart_tlb_flush(rdev); + rdev->gart.ready = true; + return 0; +} + +void rs600_gart_disable(struct radeon_device *rdev) +{ + u32 tmp; + int r; + + /* FIXME: disable out of gart access */ + WREG32_MC(R_000100_MC_PT0_CNTL, 0); + tmp = RREG32_MC(R_000009_MC_CNTL1); + WREG32_MC(R_000009_MC_CNTL1, tmp & C_000009_ENABLE_PAGE_TABLES); + if (rdev->gart.table.vram.robj) { + r = radeon_bo_reserve(rdev->gart.table.vram.robj, false); + if (r == 0) { + radeon_bo_kunmap(rdev->gart.table.vram.robj); + radeon_bo_unpin(rdev->gart.table.vram.robj); + radeon_bo_unreserve(rdev->gart.table.vram.robj); + } + } +} + +void rs600_gart_fini(struct radeon_device *rdev) +{ + radeon_gart_fini(rdev); + rs600_gart_disable(rdev); + radeon_gart_table_vram_free(rdev); +} + +#define R600_PTE_VALID (1 << 0) +#define R600_PTE_SYSTEM (1 << 1) +#define R600_PTE_SNOOPED (1 << 2) +#define R600_PTE_READABLE (1 << 5) +#define R600_PTE_WRITEABLE (1 << 6) + +int rs600_gart_set_page(struct radeon_device *rdev, int i, uint64_t addr) +{ + void __iomem *ptr = (void *)rdev->gart.table.vram.ptr; + + if (i < 0 || i > rdev->gart.num_gpu_pages) { + return -EINVAL; + } + addr = addr & 0xFFFFFFFFFFFFF000ULL; + addr |= R600_PTE_VALID | R600_PTE_SYSTEM | R600_PTE_SNOOPED; + addr |= R600_PTE_READABLE | R600_PTE_WRITEABLE; + writeq(addr, ((void __iomem *)ptr) + (i * 8)); + return 0; +} + +int rs600_irq_set(struct radeon_device *rdev) +{ + uint32_t tmp = 0; + uint32_t mode_int = 0; + u32 hpd1 = RREG32(R_007D08_DC_HOT_PLUG_DETECT1_INT_CONTROL) & + ~S_007D08_DC_HOT_PLUG_DETECT1_INT_EN(1); + u32 hpd2 = RREG32(R_007D18_DC_HOT_PLUG_DETECT2_INT_CONTROL) & + ~S_007D18_DC_HOT_PLUG_DETECT2_INT_EN(1); + + if (!rdev->irq.installed) { + WARN(1, "Can't enable IRQ/MSI because no handler is installed\n"); + WREG32(R_000040_GEN_INT_CNTL, 0); + return -EINVAL; + } + if (rdev->irq.sw_int) { + tmp |= S_000040_SW_INT_EN(1); + } + if (rdev->irq.gui_idle) { + tmp |= S_000040_GUI_IDLE(1); + } + if (rdev->irq.crtc_vblank_int[0] || + rdev->irq.pflip[0]) { + mode_int |= S_006540_D1MODE_VBLANK_INT_MASK(1); + } + if (rdev->irq.crtc_vblank_int[1] || + rdev->irq.pflip[1]) { + mode_int |= S_006540_D2MODE_VBLANK_INT_MASK(1); + } + if (rdev->irq.hpd[0]) { + hpd1 |= S_007D08_DC_HOT_PLUG_DETECT1_INT_EN(1); + } + if (rdev->irq.hpd[1]) { + hpd2 |= S_007D18_DC_HOT_PLUG_DETECT2_INT_EN(1); + } + WREG32(R_000040_GEN_INT_CNTL, tmp); + WREG32(R_006540_DxMODE_INT_MASK, mode_int); + WREG32(R_007D08_DC_HOT_PLUG_DETECT1_INT_CONTROL, hpd1); + WREG32(R_007D18_DC_HOT_PLUG_DETECT2_INT_CONTROL, hpd2); + return 0; +} + +static inline u32 rs600_irq_ack(struct radeon_device *rdev) +{ + uint32_t irqs = RREG32(R_000044_GEN_INT_STATUS); + uint32_t irq_mask = S_000044_SW_INT(1); + u32 tmp; + + /* the interrupt works, but the status bit is permanently asserted */ + if (rdev->irq.gui_idle && radeon_gui_idle(rdev)) { + if (!rdev->irq.gui_idle_acked) + irq_mask |= S_000044_GUI_IDLE_STAT(1); + } + + if (G_000044_DISPLAY_INT_STAT(irqs)) { + rdev->irq.stat_regs.r500.disp_int = RREG32(R_007EDC_DISP_INTERRUPT_STATUS); + if (G_007EDC_LB_D1_VBLANK_INTERRUPT(rdev->irq.stat_regs.r500.disp_int)) { + WREG32(R_006534_D1MODE_VBLANK_STATUS, + S_006534_D1MODE_VBLANK_ACK(1)); + } + if (G_007EDC_LB_D2_VBLANK_INTERRUPT(rdev->irq.stat_regs.r500.disp_int)) { + WREG32(R_006D34_D2MODE_VBLANK_STATUS, + S_006D34_D2MODE_VBLANK_ACK(1)); + } + if (G_007EDC_DC_HOT_PLUG_DETECT1_INTERRUPT(rdev->irq.stat_regs.r500.disp_int)) { + tmp = RREG32(R_007D08_DC_HOT_PLUG_DETECT1_INT_CONTROL); + tmp |= S_007D08_DC_HOT_PLUG_DETECT1_INT_ACK(1); + WREG32(R_007D08_DC_HOT_PLUG_DETECT1_INT_CONTROL, tmp); + } + if (G_007EDC_DC_HOT_PLUG_DETECT2_INTERRUPT(rdev->irq.stat_regs.r500.disp_int)) { + tmp = RREG32(R_007D18_DC_HOT_PLUG_DETECT2_INT_CONTROL); + tmp |= S_007D18_DC_HOT_PLUG_DETECT2_INT_ACK(1); + WREG32(R_007D18_DC_HOT_PLUG_DETECT2_INT_CONTROL, tmp); + } + } else { + rdev->irq.stat_regs.r500.disp_int = 0; + } + + if (irqs) { + WREG32(R_000044_GEN_INT_STATUS, irqs); + } + return irqs & irq_mask; +} + +void rs600_irq_disable(struct radeon_device *rdev) +{ + WREG32(R_000040_GEN_INT_CNTL, 0); + WREG32(R_006540_DxMODE_INT_MASK, 0); + /* Wait and acknowledge irq */ + mdelay(1); + rs600_irq_ack(rdev); +} + +int rs600_irq_process(struct radeon_device *rdev) +{ + u32 status, msi_rearm; + bool queue_hotplug = false; + + /* reset gui idle ack. the status bit is broken */ + rdev->irq.gui_idle_acked = false; + + status = rs600_irq_ack(rdev); + if (!status && !rdev->irq.stat_regs.r500.disp_int) { + return IRQ_NONE; + } + while (status || rdev->irq.stat_regs.r500.disp_int) { + /* SW interrupt */ + if (G_000044_SW_INT(status)) { + radeon_fence_process(rdev); + } + /* GUI idle */ + if (G_000040_GUI_IDLE(status)) { + rdev->irq.gui_idle_acked = true; + rdev->pm.gui_idle = true; + wake_up(&rdev->irq.idle_queue); + } + /* Vertical blank interrupts */ + if (G_007EDC_LB_D1_VBLANK_INTERRUPT(rdev->irq.stat_regs.r500.disp_int)) { + if (rdev->irq.crtc_vblank_int[0]) { + drm_handle_vblank(rdev->ddev, 0); + rdev->pm.vblank_sync = true; + wake_up(&rdev->irq.vblank_queue); + } + if (rdev->irq.pflip[0]) + radeon_crtc_handle_flip(rdev, 0); + } + if (G_007EDC_LB_D2_VBLANK_INTERRUPT(rdev->irq.stat_regs.r500.disp_int)) { + if (rdev->irq.crtc_vblank_int[1]) { + drm_handle_vblank(rdev->ddev, 1); + rdev->pm.vblank_sync = true; + wake_up(&rdev->irq.vblank_queue); + } + if (rdev->irq.pflip[1]) + radeon_crtc_handle_flip(rdev, 1); + } + if (G_007EDC_DC_HOT_PLUG_DETECT1_INTERRUPT(rdev->irq.stat_regs.r500.disp_int)) { + queue_hotplug = true; + DRM_DEBUG("HPD1\n"); + } + if (G_007EDC_DC_HOT_PLUG_DETECT2_INTERRUPT(rdev->irq.stat_regs.r500.disp_int)) { + queue_hotplug = true; + DRM_DEBUG("HPD2\n"); + } + status = rs600_irq_ack(rdev); + } + /* reset gui idle ack. the status bit is broken */ + rdev->irq.gui_idle_acked = false; + if (queue_hotplug) + schedule_work(&rdev->hotplug_work); + if (rdev->msi_enabled) { + switch (rdev->family) { + case CHIP_RS600: + case CHIP_RS690: + case CHIP_RS740: + msi_rearm = RREG32(RADEON_BUS_CNTL) & ~RS600_MSI_REARM; + WREG32(RADEON_BUS_CNTL, msi_rearm); + WREG32(RADEON_BUS_CNTL, msi_rearm | RS600_MSI_REARM); + break; + default: + WREG32(RADEON_MSI_REARM_EN, RV370_MSI_REARM_EN); + break; + } + } + return IRQ_HANDLED; +} + +u32 rs600_get_vblank_counter(struct radeon_device *rdev, int crtc) +{ + if (crtc == 0) + return RREG32(R_0060A4_D1CRTC_STATUS_FRAME_COUNT); + else + return RREG32(R_0068A4_D2CRTC_STATUS_FRAME_COUNT); +} + +int rs600_mc_wait_for_idle(struct radeon_device *rdev) +{ + unsigned i; + + for (i = 0; i < rdev->usec_timeout; i++) { + if (G_000000_MC_IDLE(RREG32_MC(R_000000_MC_STATUS))) + return 0; + udelay(1); + } + return -1; +} + +void rs600_gpu_init(struct radeon_device *rdev) +{ + r420_pipes_init(rdev); + /* Wait for mc idle */ + if (rs600_mc_wait_for_idle(rdev)) + dev_warn(rdev->dev, "Wait MC idle timeout before updating MC.\n"); +} + +void rs600_mc_init(struct radeon_device *rdev) +{ + u64 base; + + rdev->mc.aper_base = pci_resource_start(rdev->pdev, 0); + rdev->mc.aper_size = pci_resource_len(rdev->pdev, 0); + rdev->mc.vram_is_ddr = true; + rdev->mc.vram_width = 128; + rdev->mc.real_vram_size = RREG32(RADEON_CONFIG_MEMSIZE); + rdev->mc.mc_vram_size = rdev->mc.real_vram_size; + rdev->mc.visible_vram_size = rdev->mc.aper_size; + rdev->mc.igp_sideport_enabled = radeon_atombios_sideport_present(rdev); + base = RREG32_MC(R_000004_MC_FB_LOCATION); + base = G_000004_MC_FB_START(base) << 16; + radeon_vram_location(rdev, &rdev->mc, base); + rdev->mc.gtt_base_align = 0; + radeon_gtt_location(rdev, &rdev->mc); + radeon_update_bandwidth_info(rdev); +} + +void rs600_bandwidth_update(struct radeon_device *rdev) +{ + struct drm_display_mode *mode0 = NULL; + struct drm_display_mode *mode1 = NULL; + u32 d1mode_priority_a_cnt, d2mode_priority_a_cnt; + /* FIXME: implement full support */ + + radeon_update_display_priority(rdev); + + if (rdev->mode_info.crtcs[0]->base.enabled) + mode0 = &rdev->mode_info.crtcs[0]->base.mode; + if (rdev->mode_info.crtcs[1]->base.enabled) + mode1 = &rdev->mode_info.crtcs[1]->base.mode; + + rs690_line_buffer_adjust(rdev, mode0, mode1); + + if (rdev->disp_priority == 2) { + d1mode_priority_a_cnt = RREG32(R_006548_D1MODE_PRIORITY_A_CNT); + d2mode_priority_a_cnt = RREG32(R_006D48_D2MODE_PRIORITY_A_CNT); + d1mode_priority_a_cnt |= S_006548_D1MODE_PRIORITY_A_ALWAYS_ON(1); + d2mode_priority_a_cnt |= S_006D48_D2MODE_PRIORITY_A_ALWAYS_ON(1); + WREG32(R_006548_D1MODE_PRIORITY_A_CNT, d1mode_priority_a_cnt); + WREG32(R_00654C_D1MODE_PRIORITY_B_CNT, d1mode_priority_a_cnt); + WREG32(R_006D48_D2MODE_PRIORITY_A_CNT, d2mode_priority_a_cnt); + WREG32(R_006D4C_D2MODE_PRIORITY_B_CNT, d2mode_priority_a_cnt); + } +} + +uint32_t rs600_mc_rreg(struct radeon_device *rdev, uint32_t reg) +{ + WREG32(R_000070_MC_IND_INDEX, S_000070_MC_IND_ADDR(reg) | + S_000070_MC_IND_CITF_ARB0(1)); + return RREG32(R_000074_MC_IND_DATA); +} + +void rs600_mc_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v) +{ + WREG32(R_000070_MC_IND_INDEX, S_000070_MC_IND_ADDR(reg) | + S_000070_MC_IND_CITF_ARB0(1) | S_000070_MC_IND_WR_EN(1)); + WREG32(R_000074_MC_IND_DATA, v); +} + +void rs600_debugfs(struct radeon_device *rdev) +{ + if (r100_debugfs_rbbm_init(rdev)) + DRM_ERROR("Failed to register debugfs file for RBBM !\n"); +} + +void rs600_set_safe_registers(struct radeon_device *rdev) +{ + rdev->config.r300.reg_safe_bm = rs600_reg_safe_bm; + rdev->config.r300.reg_safe_bm_size = ARRAY_SIZE(rs600_reg_safe_bm); +} + +static void rs600_mc_program(struct radeon_device *rdev) +{ + struct rv515_mc_save save; + + /* Stops all mc clients */ + rv515_mc_stop(rdev, &save); + + /* Wait for mc idle */ + if (rs600_mc_wait_for_idle(rdev)) + dev_warn(rdev->dev, "Wait MC idle timeout before updating MC.\n"); + + /* FIXME: What does AGP means for such chipset ? */ + WREG32_MC(R_000005_MC_AGP_LOCATION, 0x0FFFFFFF); + WREG32_MC(R_000006_AGP_BASE, 0); + WREG32_MC(R_000007_AGP_BASE_2, 0); + /* Program MC */ + WREG32_MC(R_000004_MC_FB_LOCATION, + S_000004_MC_FB_START(rdev->mc.vram_start >> 16) | + S_000004_MC_FB_TOP(rdev->mc.vram_end >> 16)); + WREG32(R_000134_HDP_FB_LOCATION, + S_000134_HDP_FB_START(rdev->mc.vram_start >> 16)); + + rv515_mc_resume(rdev, &save); +} + +static int rs600_startup(struct radeon_device *rdev) +{ + int r; + + rs600_mc_program(rdev); + /* Resume clock */ + rv515_clock_startup(rdev); + /* Initialize GPU configuration (# pipes, ...) */ + rs600_gpu_init(rdev); + /* Initialize GART (initialize after TTM so we can allocate + * memory through TTM but finalize after TTM) */ + r = rs600_gart_enable(rdev); + if (r) + return r; + + /* allocate wb buffer */ + r = radeon_wb_init(rdev); + if (r) + return r; + + /* Enable IRQ */ + rs600_irq_set(rdev); + rdev->config.r300.hdp_cntl = RREG32(RADEON_HOST_PATH_CNTL); + /* 1M ring buffer */ + r = r100_cp_init(rdev, 1024 * 1024); + if (r) { + dev_err(rdev->dev, "failed initializing CP (%d).\n", r); + return r; + } + r = r100_ib_init(rdev); + if (r) { + dev_err(rdev->dev, "failed initializing IB (%d).\n", r); + return r; + } + + r = r600_audio_init(rdev); + if (r) { + dev_err(rdev->dev, "failed initializing audio\n"); + return r; + } + + return 0; +} + +int rs600_resume(struct radeon_device *rdev) +{ + /* Make sur GART are not working */ + rs600_gart_disable(rdev); + /* Resume clock before doing reset */ + rv515_clock_startup(rdev); + /* Reset gpu before posting otherwise ATOM will enter infinite loop */ + if (radeon_asic_reset(rdev)) { + dev_warn(rdev->dev, "GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n", + RREG32(R_000E40_RBBM_STATUS), + RREG32(R_0007C0_CP_STAT)); + } + /* post */ + atom_asic_init(rdev->mode_info.atom_context); + /* Resume clock after posting */ + rv515_clock_startup(rdev); + /* Initialize surface registers */ + radeon_surface_init(rdev); + return rs600_startup(rdev); +} + +int rs600_suspend(struct radeon_device *rdev) +{ + r600_audio_fini(rdev); + r100_cp_disable(rdev); + radeon_wb_disable(rdev); + rs600_irq_disable(rdev); + rs600_gart_disable(rdev); + return 0; +} + +void rs600_fini(struct radeon_device *rdev) +{ + r600_audio_fini(rdev); + r100_cp_fini(rdev); + radeon_wb_fini(rdev); + r100_ib_fini(rdev); + radeon_gem_fini(rdev); + rs600_gart_fini(rdev); + radeon_irq_kms_fini(rdev); + radeon_fence_driver_fini(rdev); + radeon_bo_fini(rdev); + radeon_atombios_fini(rdev); + kfree(rdev->bios); + rdev->bios = NULL; +} + +int rs600_init(struct radeon_device *rdev) +{ + int r; + + /* Disable VGA */ + rv515_vga_render_disable(rdev); + /* Initialize scratch registers */ + radeon_scratch_init(rdev); + /* Initialize surface registers */ + radeon_surface_init(rdev); + /* restore some register to sane defaults */ + r100_restore_sanity(rdev); + /* BIOS */ + if (!radeon_get_bios(rdev)) { + if (ASIC_IS_AVIVO(rdev)) + return -EINVAL; + } + if (rdev->is_atom_bios) { + r = radeon_atombios_init(rdev); + if (r) + return r; + } else { + dev_err(rdev->dev, "Expecting atombios for RS600 GPU\n"); + return -EINVAL; + } + /* Reset gpu before posting otherwise ATOM will enter infinite loop */ + if (radeon_asic_reset(rdev)) { + dev_warn(rdev->dev, + "GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n", + RREG32(R_000E40_RBBM_STATUS), + RREG32(R_0007C0_CP_STAT)); + } + /* check if cards are posted or not */ + if (radeon_boot_test_post_card(rdev) == false) + return -EINVAL; + + /* Initialize clocks */ + radeon_get_clock_info(rdev->ddev); + /* initialize memory controller */ + rs600_mc_init(rdev); + rs600_debugfs(rdev); + /* Fence driver */ + r = radeon_fence_driver_init(rdev); + if (r) + return r; + r = radeon_irq_kms_init(rdev); + if (r) + return r; + /* Memory manager */ + r = radeon_bo_init(rdev); + if (r) + return r; + r = rs600_gart_init(rdev); + if (r) + return r; + rs600_set_safe_registers(rdev); + rdev->accel_working = true; + r = rs600_startup(rdev); + if (r) { + /* Somethings want wront with the accel init stop accel */ + dev_err(rdev->dev, "Disabling GPU acceleration\n"); + r100_cp_fini(rdev); + radeon_wb_fini(rdev); + r100_ib_fini(rdev); + rs600_gart_fini(rdev); + radeon_irq_kms_fini(rdev); + rdev->accel_working = false; + } + return 0; +} diff --git a/drivers/gpu/drm/radeon/rs600d.h b/drivers/gpu/drm/radeon/rs600d.h new file mode 100644 index 00000000..a27c13ac --- /dev/null +++ b/drivers/gpu/drm/radeon/rs600d.h @@ -0,0 +1,671 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#ifndef __RS600D_H__ +#define __RS600D_H__ + +/* Registers */ +#define R_000040_GEN_INT_CNTL 0x000040 +#define S_000040_SCRATCH_INT_MASK(x) (((x) & 0x1) << 18) +#define G_000040_SCRATCH_INT_MASK(x) (((x) >> 18) & 0x1) +#define C_000040_SCRATCH_INT_MASK 0xFFFBFFFF +#define S_000040_GUI_IDLE_MASK(x) (((x) & 0x1) << 19) +#define G_000040_GUI_IDLE_MASK(x) (((x) >> 19) & 0x1) +#define C_000040_GUI_IDLE_MASK 0xFFF7FFFF +#define S_000040_DMA_VIPH1_INT_EN(x) (((x) & 0x1) << 13) +#define G_000040_DMA_VIPH1_INT_EN(x) (((x) >> 13) & 0x1) +#define C_000040_DMA_VIPH1_INT_EN 0xFFFFDFFF +#define S_000040_DMA_VIPH2_INT_EN(x) (((x) & 0x1) << 14) +#define G_000040_DMA_VIPH2_INT_EN(x) (((x) >> 14) & 0x1) +#define C_000040_DMA_VIPH2_INT_EN 0xFFFFBFFF +#define S_000040_DMA_VIPH3_INT_EN(x) (((x) & 0x1) << 15) +#define G_000040_DMA_VIPH3_INT_EN(x) (((x) >> 15) & 0x1) +#define C_000040_DMA_VIPH3_INT_EN 0xFFFF7FFF +#define S_000040_I2C_INT_EN(x) (((x) & 0x1) << 17) +#define G_000040_I2C_INT_EN(x) (((x) >> 17) & 0x1) +#define C_000040_I2C_INT_EN 0xFFFDFFFF +#define S_000040_GUI_IDLE(x) (((x) & 0x1) << 19) +#define G_000040_GUI_IDLE(x) (((x) >> 19) & 0x1) +#define C_000040_GUI_IDLE 0xFFF7FFFF +#define S_000040_VIPH_INT_EN(x) (((x) & 0x1) << 24) +#define G_000040_VIPH_INT_EN(x) (((x) >> 24) & 0x1) +#define C_000040_VIPH_INT_EN 0xFEFFFFFF +#define S_000040_SW_INT_EN(x) (((x) & 0x1) << 25) +#define G_000040_SW_INT_EN(x) (((x) >> 25) & 0x1) +#define C_000040_SW_INT_EN 0xFDFFFFFF +#define S_000040_GEYSERVILLE(x) (((x) & 0x1) << 27) +#define G_000040_GEYSERVILLE(x) (((x) >> 27) & 0x1) +#define C_000040_GEYSERVILLE 0xF7FFFFFF +#define S_000040_HDCP_AUTHORIZED_INT(x) (((x) & 0x1) << 28) +#define G_000040_HDCP_AUTHORIZED_INT(x) (((x) >> 28) & 0x1) +#define C_000040_HDCP_AUTHORIZED_INT 0xEFFFFFFF +#define S_000040_DVI_I2C_INT(x) (((x) & 0x1) << 29) +#define G_000040_DVI_I2C_INT(x) (((x) >> 29) & 0x1) +#define C_000040_DVI_I2C_INT 0xDFFFFFFF +#define S_000040_GUIDMA(x) (((x) & 0x1) << 30) +#define G_000040_GUIDMA(x) (((x) >> 30) & 0x1) +#define C_000040_GUIDMA 0xBFFFFFFF +#define S_000040_VIDDMA(x) (((x) & 0x1) << 31) +#define G_000040_VIDDMA(x) (((x) >> 31) & 0x1) +#define C_000040_VIDDMA 0x7FFFFFFF +#define R_000044_GEN_INT_STATUS 0x000044 +#define S_000044_DISPLAY_INT_STAT(x) (((x) & 0x1) << 0) +#define G_000044_DISPLAY_INT_STAT(x) (((x) >> 0) & 0x1) +#define C_000044_DISPLAY_INT_STAT 0xFFFFFFFE +#define S_000044_VGA_INT_STAT(x) (((x) & 0x1) << 1) +#define G_000044_VGA_INT_STAT(x) (((x) >> 1) & 0x1) +#define C_000044_VGA_INT_STAT 0xFFFFFFFD +#define S_000044_CAP0_INT_ACTIVE(x) (((x) & 0x1) << 8) +#define G_000044_CAP0_INT_ACTIVE(x) (((x) >> 8) & 0x1) +#define C_000044_CAP0_INT_ACTIVE 0xFFFFFEFF +#define S_000044_DMA_VIPH0_INT(x) (((x) & 0x1) << 12) +#define G_000044_DMA_VIPH0_INT(x) (((x) >> 12) & 0x1) +#define C_000044_DMA_VIPH0_INT 0xFFFFEFFF +#define S_000044_DMA_VIPH1_INT(x) (((x) & 0x1) << 13) +#define G_000044_DMA_VIPH1_INT(x) (((x) >> 13) & 0x1) +#define C_000044_DMA_VIPH1_INT 0xFFFFDFFF +#define S_000044_DMA_VIPH2_INT(x) (((x) & 0x1) << 14) +#define G_000044_DMA_VIPH2_INT(x) (((x) >> 14) & 0x1) +#define C_000044_DMA_VIPH2_INT 0xFFFFBFFF +#define S_000044_DMA_VIPH3_INT(x) (((x) & 0x1) << 15) +#define G_000044_DMA_VIPH3_INT(x) (((x) >> 15) & 0x1) +#define C_000044_DMA_VIPH3_INT 0xFFFF7FFF +#define S_000044_MC_PROBE_FAULT_STAT(x) (((x) & 0x1) << 16) +#define G_000044_MC_PROBE_FAULT_STAT(x) (((x) >> 16) & 0x1) +#define C_000044_MC_PROBE_FAULT_STAT 0xFFFEFFFF +#define S_000044_I2C_INT(x) (((x) & 0x1) << 17) +#define G_000044_I2C_INT(x) (((x) >> 17) & 0x1) +#define C_000044_I2C_INT 0xFFFDFFFF +#define S_000044_SCRATCH_INT_STAT(x) (((x) & 0x1) << 18) +#define G_000044_SCRATCH_INT_STAT(x) (((x) >> 18) & 0x1) +#define C_000044_SCRATCH_INT_STAT 0xFFFBFFFF +#define S_000044_GUI_IDLE_STAT(x) (((x) & 0x1) << 19) +#define G_000044_GUI_IDLE_STAT(x) (((x) >> 19) & 0x1) +#define C_000044_GUI_IDLE_STAT 0xFFF7FFFF +#define S_000044_ATI_OVERDRIVE_INT_STAT(x) (((x) & 0x1) << 20) +#define G_000044_ATI_OVERDRIVE_INT_STAT(x) (((x) >> 20) & 0x1) +#define C_000044_ATI_OVERDRIVE_INT_STAT 0xFFEFFFFF +#define S_000044_MC_PROTECTION_FAULT_STAT(x) (((x) & 0x1) << 21) +#define G_000044_MC_PROTECTION_FAULT_STAT(x) (((x) >> 21) & 0x1) +#define C_000044_MC_PROTECTION_FAULT_STAT 0xFFDFFFFF +#define S_000044_RBBM_READ_INT_STAT(x) (((x) & 0x1) << 22) +#define G_000044_RBBM_READ_INT_STAT(x) (((x) >> 22) & 0x1) +#define C_000044_RBBM_READ_INT_STAT 0xFFBFFFFF +#define S_000044_CB_CONTEXT_SWITCH_STAT(x) (((x) & 0x1) << 23) +#define G_000044_CB_CONTEXT_SWITCH_STAT(x) (((x) >> 23) & 0x1) +#define C_000044_CB_CONTEXT_SWITCH_STAT 0xFF7FFFFF +#define S_000044_VIPH_INT(x) (((x) & 0x1) << 24) +#define G_000044_VIPH_INT(x) (((x) >> 24) & 0x1) +#define C_000044_VIPH_INT 0xFEFFFFFF +#define S_000044_SW_INT(x) (((x) & 0x1) << 25) +#define G_000044_SW_INT(x) (((x) >> 25) & 0x1) +#define C_000044_SW_INT 0xFDFFFFFF +#define S_000044_SW_INT_SET(x) (((x) & 0x1) << 26) +#define G_000044_SW_INT_SET(x) (((x) >> 26) & 0x1) +#define C_000044_SW_INT_SET 0xFBFFFFFF +#define S_000044_IDCT_INT_STAT(x) (((x) & 0x1) << 27) +#define G_000044_IDCT_INT_STAT(x) (((x) >> 27) & 0x1) +#define C_000044_IDCT_INT_STAT 0xF7FFFFFF +#define S_000044_GUIDMA_STAT(x) (((x) & 0x1) << 30) +#define G_000044_GUIDMA_STAT(x) (((x) >> 30) & 0x1) +#define C_000044_GUIDMA_STAT 0xBFFFFFFF +#define S_000044_VIDDMA_STAT(x) (((x) & 0x1) << 31) +#define G_000044_VIDDMA_STAT(x) (((x) >> 31) & 0x1) +#define C_000044_VIDDMA_STAT 0x7FFFFFFF +#define R_00004C_BUS_CNTL 0x00004C +#define S_00004C_BUS_MASTER_DIS(x) (((x) & 0x1) << 14) +#define G_00004C_BUS_MASTER_DIS(x) (((x) >> 14) & 0x1) +#define C_00004C_BUS_MASTER_DIS 0xFFFFBFFF +#define S_00004C_BUS_MSI_REARM(x) (((x) & 0x1) << 20) +#define G_00004C_BUS_MSI_REARM(x) (((x) >> 20) & 0x1) +#define C_00004C_BUS_MSI_REARM 0xFFEFFFFF +#define R_000070_MC_IND_INDEX 0x000070 +#define S_000070_MC_IND_ADDR(x) (((x) & 0xFFFF) << 0) +#define G_000070_MC_IND_ADDR(x) (((x) >> 0) & 0xFFFF) +#define C_000070_MC_IND_ADDR 0xFFFF0000 +#define S_000070_MC_IND_SEQ_RBS_0(x) (((x) & 0x1) << 16) +#define G_000070_MC_IND_SEQ_RBS_0(x) (((x) >> 16) & 0x1) +#define C_000070_MC_IND_SEQ_RBS_0 0xFFFEFFFF +#define S_000070_MC_IND_SEQ_RBS_1(x) (((x) & 0x1) << 17) +#define G_000070_MC_IND_SEQ_RBS_1(x) (((x) >> 17) & 0x1) +#define C_000070_MC_IND_SEQ_RBS_1 0xFFFDFFFF +#define S_000070_MC_IND_SEQ_RBS_2(x) (((x) & 0x1) << 18) +#define G_000070_MC_IND_SEQ_RBS_2(x) (((x) >> 18) & 0x1) +#define C_000070_MC_IND_SEQ_RBS_2 0xFFFBFFFF +#define S_000070_MC_IND_SEQ_RBS_3(x) (((x) & 0x1) << 19) +#define G_000070_MC_IND_SEQ_RBS_3(x) (((x) >> 19) & 0x1) +#define C_000070_MC_IND_SEQ_RBS_3 0xFFF7FFFF +#define S_000070_MC_IND_AIC_RBS(x) (((x) & 0x1) << 20) +#define G_000070_MC_IND_AIC_RBS(x) (((x) >> 20) & 0x1) +#define C_000070_MC_IND_AIC_RBS 0xFFEFFFFF +#define S_000070_MC_IND_CITF_ARB0(x) (((x) & 0x1) << 21) +#define G_000070_MC_IND_CITF_ARB0(x) (((x) >> 21) & 0x1) +#define C_000070_MC_IND_CITF_ARB0 0xFFDFFFFF +#define S_000070_MC_IND_CITF_ARB1(x) (((x) & 0x1) << 22) +#define G_000070_MC_IND_CITF_ARB1(x) (((x) >> 22) & 0x1) +#define C_000070_MC_IND_CITF_ARB1 0xFFBFFFFF +#define S_000070_MC_IND_WR_EN(x) (((x) & 0x1) << 23) +#define G_000070_MC_IND_WR_EN(x) (((x) >> 23) & 0x1) +#define C_000070_MC_IND_WR_EN 0xFF7FFFFF +#define S_000070_MC_IND_RD_INV(x) (((x) & 0x1) << 24) +#define G_000070_MC_IND_RD_INV(x) (((x) >> 24) & 0x1) +#define C_000070_MC_IND_RD_INV 0xFEFFFFFF +#define R_000074_MC_IND_DATA 0x000074 +#define S_000074_MC_IND_DATA(x) (((x) & 0xFFFFFFFF) << 0) +#define G_000074_MC_IND_DATA(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_000074_MC_IND_DATA 0x00000000 +#define R_0000F0_RBBM_SOFT_RESET 0x0000F0 +#define S_0000F0_SOFT_RESET_CP(x) (((x) & 0x1) << 0) +#define G_0000F0_SOFT_RESET_CP(x) (((x) >> 0) & 0x1) +#define C_0000F0_SOFT_RESET_CP 0xFFFFFFFE +#define S_0000F0_SOFT_RESET_HI(x) (((x) & 0x1) << 1) +#define G_0000F0_SOFT_RESET_HI(x) (((x) >> 1) & 0x1) +#define C_0000F0_SOFT_RESET_HI 0xFFFFFFFD +#define S_0000F0_SOFT_RESET_VAP(x) (((x) & 0x1) << 2) +#define G_0000F0_SOFT_RESET_VAP(x) (((x) >> 2) & 0x1) +#define C_0000F0_SOFT_RESET_VAP 0xFFFFFFFB +#define S_0000F0_SOFT_RESET_RE(x) (((x) & 0x1) << 3) +#define G_0000F0_SOFT_RESET_RE(x) (((x) >> 3) & 0x1) +#define C_0000F0_SOFT_RESET_RE 0xFFFFFFF7 +#define S_0000F0_SOFT_RESET_PP(x) (((x) & 0x1) << 4) +#define G_0000F0_SOFT_RESET_PP(x) (((x) >> 4) & 0x1) +#define C_0000F0_SOFT_RESET_PP 0xFFFFFFEF +#define S_0000F0_SOFT_RESET_E2(x) (((x) & 0x1) << 5) +#define G_0000F0_SOFT_RESET_E2(x) (((x) >> 5) & 0x1) +#define C_0000F0_SOFT_RESET_E2 0xFFFFFFDF +#define S_0000F0_SOFT_RESET_RB(x) (((x) & 0x1) << 6) +#define G_0000F0_SOFT_RESET_RB(x) (((x) >> 6) & 0x1) +#define C_0000F0_SOFT_RESET_RB 0xFFFFFFBF +#define S_0000F0_SOFT_RESET_HDP(x) (((x) & 0x1) << 7) +#define G_0000F0_SOFT_RESET_HDP(x) (((x) >> 7) & 0x1) +#define C_0000F0_SOFT_RESET_HDP 0xFFFFFF7F +#define S_0000F0_SOFT_RESET_MC(x) (((x) & 0x1) << 8) +#define G_0000F0_SOFT_RESET_MC(x) (((x) >> 8) & 0x1) +#define C_0000F0_SOFT_RESET_MC 0xFFFFFEFF +#define S_0000F0_SOFT_RESET_AIC(x) (((x) & 0x1) << 9) +#define G_0000F0_SOFT_RESET_AIC(x) (((x) >> 9) & 0x1) +#define C_0000F0_SOFT_RESET_AIC 0xFFFFFDFF +#define S_0000F0_SOFT_RESET_VIP(x) (((x) & 0x1) << 10) +#define G_0000F0_SOFT_RESET_VIP(x) (((x) >> 10) & 0x1) +#define C_0000F0_SOFT_RESET_VIP 0xFFFFFBFF +#define S_0000F0_SOFT_RESET_DISP(x) (((x) & 0x1) << 11) +#define G_0000F0_SOFT_RESET_DISP(x) (((x) >> 11) & 0x1) +#define C_0000F0_SOFT_RESET_DISP 0xFFFFF7FF +#define S_0000F0_SOFT_RESET_CG(x) (((x) & 0x1) << 12) +#define G_0000F0_SOFT_RESET_CG(x) (((x) >> 12) & 0x1) +#define C_0000F0_SOFT_RESET_CG 0xFFFFEFFF +#define S_0000F0_SOFT_RESET_GA(x) (((x) & 0x1) << 13) +#define G_0000F0_SOFT_RESET_GA(x) (((x) >> 13) & 0x1) +#define C_0000F0_SOFT_RESET_GA 0xFFFFDFFF +#define S_0000F0_SOFT_RESET_IDCT(x) (((x) & 0x1) << 14) +#define G_0000F0_SOFT_RESET_IDCT(x) (((x) >> 14) & 0x1) +#define C_0000F0_SOFT_RESET_IDCT 0xFFFFBFFF +#define R_000134_HDP_FB_LOCATION 0x000134 +#define S_000134_HDP_FB_START(x) (((x) & 0xFFFF) << 0) +#define G_000134_HDP_FB_START(x) (((x) >> 0) & 0xFFFF) +#define C_000134_HDP_FB_START 0xFFFF0000 +#define R_0007C0_CP_STAT 0x0007C0 +#define S_0007C0_MRU_BUSY(x) (((x) & 0x1) << 0) +#define G_0007C0_MRU_BUSY(x) (((x) >> 0) & 0x1) +#define C_0007C0_MRU_BUSY 0xFFFFFFFE +#define S_0007C0_MWU_BUSY(x) (((x) & 0x1) << 1) +#define G_0007C0_MWU_BUSY(x) (((x) >> 1) & 0x1) +#define C_0007C0_MWU_BUSY 0xFFFFFFFD +#define S_0007C0_RSIU_BUSY(x) (((x) & 0x1) << 2) +#define G_0007C0_RSIU_BUSY(x) (((x) >> 2) & 0x1) +#define C_0007C0_RSIU_BUSY 0xFFFFFFFB +#define S_0007C0_RCIU_BUSY(x) (((x) & 0x1) << 3) +#define G_0007C0_RCIU_BUSY(x) (((x) >> 3) & 0x1) +#define C_0007C0_RCIU_BUSY 0xFFFFFFF7 +#define S_0007C0_CSF_PRIMARY_BUSY(x) (((x) & 0x1) << 9) +#define G_0007C0_CSF_PRIMARY_BUSY(x) (((x) >> 9) & 0x1) +#define C_0007C0_CSF_PRIMARY_BUSY 0xFFFFFDFF +#define S_0007C0_CSF_INDIRECT_BUSY(x) (((x) & 0x1) << 10) +#define G_0007C0_CSF_INDIRECT_BUSY(x) (((x) >> 10) & 0x1) +#define C_0007C0_CSF_INDIRECT_BUSY 0xFFFFFBFF +#define S_0007C0_CSQ_PRIMARY_BUSY(x) (((x) & 0x1) << 11) +#define G_0007C0_CSQ_PRIMARY_BUSY(x) (((x) >> 11) & 0x1) +#define C_0007C0_CSQ_PRIMARY_BUSY 0xFFFFF7FF +#define S_0007C0_CSQ_INDIRECT_BUSY(x) (((x) & 0x1) << 12) +#define G_0007C0_CSQ_INDIRECT_BUSY(x) (((x) >> 12) & 0x1) +#define C_0007C0_CSQ_INDIRECT_BUSY 0xFFFFEFFF +#define S_0007C0_CSI_BUSY(x) (((x) & 0x1) << 13) +#define G_0007C0_CSI_BUSY(x) (((x) >> 13) & 0x1) +#define C_0007C0_CSI_BUSY 0xFFFFDFFF +#define S_0007C0_CSF_INDIRECT2_BUSY(x) (((x) & 0x1) << 14) +#define G_0007C0_CSF_INDIRECT2_BUSY(x) (((x) >> 14) & 0x1) +#define C_0007C0_CSF_INDIRECT2_BUSY 0xFFFFBFFF +#define S_0007C0_CSQ_INDIRECT2_BUSY(x) (((x) & 0x1) << 15) +#define G_0007C0_CSQ_INDIRECT2_BUSY(x) (((x) >> 15) & 0x1) +#define C_0007C0_CSQ_INDIRECT2_BUSY 0xFFFF7FFF +#define S_0007C0_GUIDMA_BUSY(x) (((x) & 0x1) << 28) +#define G_0007C0_GUIDMA_BUSY(x) (((x) >> 28) & 0x1) +#define C_0007C0_GUIDMA_BUSY 0xEFFFFFFF +#define S_0007C0_VIDDMA_BUSY(x) (((x) & 0x1) << 29) +#define G_0007C0_VIDDMA_BUSY(x) (((x) >> 29) & 0x1) +#define C_0007C0_VIDDMA_BUSY 0xDFFFFFFF +#define S_0007C0_CMDSTRM_BUSY(x) (((x) & 0x1) << 30) +#define G_0007C0_CMDSTRM_BUSY(x) (((x) >> 30) & 0x1) +#define C_0007C0_CMDSTRM_BUSY 0xBFFFFFFF +#define S_0007C0_CP_BUSY(x) (((x) & 0x1) << 31) +#define G_0007C0_CP_BUSY(x) (((x) >> 31) & 0x1) +#define C_0007C0_CP_BUSY 0x7FFFFFFF +#define R_000E40_RBBM_STATUS 0x000E40 +#define S_000E40_CMDFIFO_AVAIL(x) (((x) & 0x7F) << 0) +#define G_000E40_CMDFIFO_AVAIL(x) (((x) >> 0) & 0x7F) +#define C_000E40_CMDFIFO_AVAIL 0xFFFFFF80 +#define S_000E40_HIRQ_ON_RBB(x) (((x) & 0x1) << 8) +#define G_000E40_HIRQ_ON_RBB(x) (((x) >> 8) & 0x1) +#define C_000E40_HIRQ_ON_RBB 0xFFFFFEFF +#define S_000E40_CPRQ_ON_RBB(x) (((x) & 0x1) << 9) +#define G_000E40_CPRQ_ON_RBB(x) (((x) >> 9) & 0x1) +#define C_000E40_CPRQ_ON_RBB 0xFFFFFDFF +#define S_000E40_CFRQ_ON_RBB(x) (((x) & 0x1) << 10) +#define G_000E40_CFRQ_ON_RBB(x) (((x) >> 10) & 0x1) +#define C_000E40_CFRQ_ON_RBB 0xFFFFFBFF +#define S_000E40_HIRQ_IN_RTBUF(x) (((x) & 0x1) << 11) +#define G_000E40_HIRQ_IN_RTBUF(x) (((x) >> 11) & 0x1) +#define C_000E40_HIRQ_IN_RTBUF 0xFFFFF7FF +#define S_000E40_CPRQ_IN_RTBUF(x) (((x) & 0x1) << 12) +#define G_000E40_CPRQ_IN_RTBUF(x) (((x) >> 12) & 0x1) +#define C_000E40_CPRQ_IN_RTBUF 0xFFFFEFFF +#define S_000E40_CFRQ_IN_RTBUF(x) (((x) & 0x1) << 13) +#define G_000E40_CFRQ_IN_RTBUF(x) (((x) >> 13) & 0x1) +#define C_000E40_CFRQ_IN_RTBUF 0xFFFFDFFF +#define S_000E40_CF_PIPE_BUSY(x) (((x) & 0x1) << 14) +#define G_000E40_CF_PIPE_BUSY(x) (((x) >> 14) & 0x1) +#define C_000E40_CF_PIPE_BUSY 0xFFFFBFFF +#define S_000E40_ENG_EV_BUSY(x) (((x) & 0x1) << 15) +#define G_000E40_ENG_EV_BUSY(x) (((x) >> 15) & 0x1) +#define C_000E40_ENG_EV_BUSY 0xFFFF7FFF +#define S_000E40_CP_CMDSTRM_BUSY(x) (((x) & 0x1) << 16) +#define G_000E40_CP_CMDSTRM_BUSY(x) (((x) >> 16) & 0x1) +#define C_000E40_CP_CMDSTRM_BUSY 0xFFFEFFFF +#define S_000E40_E2_BUSY(x) (((x) & 0x1) << 17) +#define G_000E40_E2_BUSY(x) (((x) >> 17) & 0x1) +#define C_000E40_E2_BUSY 0xFFFDFFFF +#define S_000E40_RB2D_BUSY(x) (((x) & 0x1) << 18) +#define G_000E40_RB2D_BUSY(x) (((x) >> 18) & 0x1) +#define C_000E40_RB2D_BUSY 0xFFFBFFFF +#define S_000E40_RB3D_BUSY(x) (((x) & 0x1) << 19) +#define G_000E40_RB3D_BUSY(x) (((x) >> 19) & 0x1) +#define C_000E40_RB3D_BUSY 0xFFF7FFFF +#define S_000E40_VAP_BUSY(x) (((x) & 0x1) << 20) +#define G_000E40_VAP_BUSY(x) (((x) >> 20) & 0x1) +#define C_000E40_VAP_BUSY 0xFFEFFFFF +#define S_000E40_RE_BUSY(x) (((x) & 0x1) << 21) +#define G_000E40_RE_BUSY(x) (((x) >> 21) & 0x1) +#define C_000E40_RE_BUSY 0xFFDFFFFF +#define S_000E40_TAM_BUSY(x) (((x) & 0x1) << 22) +#define G_000E40_TAM_BUSY(x) (((x) >> 22) & 0x1) +#define C_000E40_TAM_BUSY 0xFFBFFFFF +#define S_000E40_TDM_BUSY(x) (((x) & 0x1) << 23) +#define G_000E40_TDM_BUSY(x) (((x) >> 23) & 0x1) +#define C_000E40_TDM_BUSY 0xFF7FFFFF +#define S_000E40_PB_BUSY(x) (((x) & 0x1) << 24) +#define G_000E40_PB_BUSY(x) (((x) >> 24) & 0x1) +#define C_000E40_PB_BUSY 0xFEFFFFFF +#define S_000E40_TIM_BUSY(x) (((x) & 0x1) << 25) +#define G_000E40_TIM_BUSY(x) (((x) >> 25) & 0x1) +#define C_000E40_TIM_BUSY 0xFDFFFFFF +#define S_000E40_GA_BUSY(x) (((x) & 0x1) << 26) +#define G_000E40_GA_BUSY(x) (((x) >> 26) & 0x1) +#define C_000E40_GA_BUSY 0xFBFFFFFF +#define S_000E40_CBA2D_BUSY(x) (((x) & 0x1) << 27) +#define G_000E40_CBA2D_BUSY(x) (((x) >> 27) & 0x1) +#define C_000E40_CBA2D_BUSY 0xF7FFFFFF +#define S_000E40_GUI_ACTIVE(x) (((x) & 0x1) << 31) +#define G_000E40_GUI_ACTIVE(x) (((x) >> 31) & 0x1) +#define C_000E40_GUI_ACTIVE 0x7FFFFFFF +#define R_0060A4_D1CRTC_STATUS_FRAME_COUNT 0x0060A4 +#define S_0060A4_D1CRTC_FRAME_COUNT(x) (((x) & 0xFFFFFF) << 0) +#define G_0060A4_D1CRTC_FRAME_COUNT(x) (((x) >> 0) & 0xFFFFFF) +#define C_0060A4_D1CRTC_FRAME_COUNT 0xFF000000 +#define R_006534_D1MODE_VBLANK_STATUS 0x006534 +#define S_006534_D1MODE_VBLANK_OCCURRED(x) (((x) & 0x1) << 0) +#define G_006534_D1MODE_VBLANK_OCCURRED(x) (((x) >> 0) & 0x1) +#define C_006534_D1MODE_VBLANK_OCCURRED 0xFFFFFFFE +#define S_006534_D1MODE_VBLANK_ACK(x) (((x) & 0x1) << 4) +#define G_006534_D1MODE_VBLANK_ACK(x) (((x) >> 4) & 0x1) +#define C_006534_D1MODE_VBLANK_ACK 0xFFFFFFEF +#define S_006534_D1MODE_VBLANK_STAT(x) (((x) & 0x1) << 12) +#define G_006534_D1MODE_VBLANK_STAT(x) (((x) >> 12) & 0x1) +#define C_006534_D1MODE_VBLANK_STAT 0xFFFFEFFF +#define S_006534_D1MODE_VBLANK_INTERRUPT(x) (((x) & 0x1) << 16) +#define G_006534_D1MODE_VBLANK_INTERRUPT(x) (((x) >> 16) & 0x1) +#define C_006534_D1MODE_VBLANK_INTERRUPT 0xFFFEFFFF +#define R_006540_DxMODE_INT_MASK 0x006540 +#define S_006540_D1MODE_VBLANK_INT_MASK(x) (((x) & 0x1) << 0) +#define G_006540_D1MODE_VBLANK_INT_MASK(x) (((x) >> 0) & 0x1) +#define C_006540_D1MODE_VBLANK_INT_MASK 0xFFFFFFFE +#define S_006540_D1MODE_VLINE_INT_MASK(x) (((x) & 0x1) << 4) +#define G_006540_D1MODE_VLINE_INT_MASK(x) (((x) >> 4) & 0x1) +#define C_006540_D1MODE_VLINE_INT_MASK 0xFFFFFFEF +#define S_006540_D2MODE_VBLANK_INT_MASK(x) (((x) & 0x1) << 8) +#define G_006540_D2MODE_VBLANK_INT_MASK(x) (((x) >> 8) & 0x1) +#define C_006540_D2MODE_VBLANK_INT_MASK 0xFFFFFEFF +#define S_006540_D2MODE_VLINE_INT_MASK(x) (((x) & 0x1) << 12) +#define G_006540_D2MODE_VLINE_INT_MASK(x) (((x) >> 12) & 0x1) +#define C_006540_D2MODE_VLINE_INT_MASK 0xFFFFEFFF +#define S_006540_D1MODE_VBLANK_CP_SEL(x) (((x) & 0x1) << 30) +#define G_006540_D1MODE_VBLANK_CP_SEL(x) (((x) >> 30) & 0x1) +#define C_006540_D1MODE_VBLANK_CP_SEL 0xBFFFFFFF +#define S_006540_D2MODE_VBLANK_CP_SEL(x) (((x) & 0x1) << 31) +#define G_006540_D2MODE_VBLANK_CP_SEL(x) (((x) >> 31) & 0x1) +#define C_006540_D2MODE_VBLANK_CP_SEL 0x7FFFFFFF +#define R_0068A4_D2CRTC_STATUS_FRAME_COUNT 0x0068A4 +#define S_0068A4_D2CRTC_FRAME_COUNT(x) (((x) & 0xFFFFFF) << 0) +#define G_0068A4_D2CRTC_FRAME_COUNT(x) (((x) >> 0) & 0xFFFFFF) +#define C_0068A4_D2CRTC_FRAME_COUNT 0xFF000000 +#define R_006D34_D2MODE_VBLANK_STATUS 0x006D34 +#define S_006D34_D2MODE_VBLANK_OCCURRED(x) (((x) & 0x1) << 0) +#define G_006D34_D2MODE_VBLANK_OCCURRED(x) (((x) >> 0) & 0x1) +#define C_006D34_D2MODE_VBLANK_OCCURRED 0xFFFFFFFE +#define S_006D34_D2MODE_VBLANK_ACK(x) (((x) & 0x1) << 4) +#define G_006D34_D2MODE_VBLANK_ACK(x) (((x) >> 4) & 0x1) +#define C_006D34_D2MODE_VBLANK_ACK 0xFFFFFFEF +#define S_006D34_D2MODE_VBLANK_STAT(x) (((x) & 0x1) << 12) +#define G_006D34_D2MODE_VBLANK_STAT(x) (((x) >> 12) & 0x1) +#define C_006D34_D2MODE_VBLANK_STAT 0xFFFFEFFF +#define S_006D34_D2MODE_VBLANK_INTERRUPT(x) (((x) & 0x1) << 16) +#define G_006D34_D2MODE_VBLANK_INTERRUPT(x) (((x) >> 16) & 0x1) +#define C_006D34_D2MODE_VBLANK_INTERRUPT 0xFFFEFFFF +#define R_007EDC_DISP_INTERRUPT_STATUS 0x007EDC +#define S_007EDC_LB_D1_VBLANK_INTERRUPT(x) (((x) & 0x1) << 4) +#define G_007EDC_LB_D1_VBLANK_INTERRUPT(x) (((x) >> 4) & 0x1) +#define C_007EDC_LB_D1_VBLANK_INTERRUPT 0xFFFFFFEF +#define S_007EDC_LB_D2_VBLANK_INTERRUPT(x) (((x) & 0x1) << 5) +#define G_007EDC_LB_D2_VBLANK_INTERRUPT(x) (((x) >> 5) & 0x1) +#define C_007EDC_LB_D2_VBLANK_INTERRUPT 0xFFFFFFDF +#define S_007EDC_DACA_AUTODETECT_INTERRUPT(x) (((x) & 0x1) << 16) +#define G_007EDC_DACA_AUTODETECT_INTERRUPT(x) (((x) >> 16) & 0x1) +#define C_007EDC_DACA_AUTODETECT_INTERRUPT 0xFFFEFFFF +#define S_007EDC_DACB_AUTODETECT_INTERRUPT(x) (((x) & 0x1) << 17) +#define G_007EDC_DACB_AUTODETECT_INTERRUPT(x) (((x) >> 17) & 0x1) +#define C_007EDC_DACB_AUTODETECT_INTERRUPT 0xFFFDFFFF +#define S_007EDC_DC_HOT_PLUG_DETECT1_INTERRUPT(x) (((x) & 0x1) << 18) +#define G_007EDC_DC_HOT_PLUG_DETECT1_INTERRUPT(x) (((x) >> 18) & 0x1) +#define C_007EDC_DC_HOT_PLUG_DETECT1_INTERRUPT 0xFFFBFFFF +#define S_007EDC_DC_HOT_PLUG_DETECT2_INTERRUPT(x) (((x) & 0x1) << 19) +#define G_007EDC_DC_HOT_PLUG_DETECT2_INTERRUPT(x) (((x) >> 19) & 0x1) +#define C_007EDC_DC_HOT_PLUG_DETECT2_INTERRUPT 0xFFF7FFFF +#define R_007828_DACA_AUTODETECT_CONTROL 0x007828 +#define S_007828_DACA_AUTODETECT_MODE(x) (((x) & 0x3) << 0) +#define G_007828_DACA_AUTODETECT_MODE(x) (((x) >> 0) & 0x3) +#define C_007828_DACA_AUTODETECT_MODE 0xFFFFFFFC +#define S_007828_DACA_AUTODETECT_FRAME_TIME_COUNTER(x) (((x) & 0xff) << 8) +#define G_007828_DACA_AUTODETECT_FRAME_TIME_COUNTER(x) (((x) >> 8) & 0xff) +#define C_007828_DACA_AUTODETECT_FRAME_TIME_COUNTER 0xFFFF00FF +#define S_007828_DACA_AUTODETECT_CHECK_MASK(x) (((x) & 0x3) << 16) +#define G_007828_DACA_AUTODETECT_CHECK_MASK(x) (((x) >> 16) & 0x3) +#define C_007828_DACA_AUTODETECT_CHECK_MASK 0xFFFCFFFF +#define R_007838_DACA_AUTODETECT_INT_CONTROL 0x007838 +#define S_007838_DACA_AUTODETECT_ACK(x) (((x) & 0x1) << 0) +#define C_007838_DACA_DACA_AUTODETECT_ACK 0xFFFFFFFE +#define S_007838_DACA_AUTODETECT_INT_ENABLE(x) (((x) & 0x1) << 16) +#define G_007838_DACA_AUTODETECT_INT_ENABLE(x) (((x) >> 16) & 0x1) +#define C_007838_DACA_AUTODETECT_INT_ENABLE 0xFFFCFFFF +#define R_007A28_DACB_AUTODETECT_CONTROL 0x007A28 +#define S_007A28_DACB_AUTODETECT_MODE(x) (((x) & 0x3) << 0) +#define G_007A28_DACB_AUTODETECT_MODE(x) (((x) >> 0) & 0x3) +#define C_007A28_DACB_AUTODETECT_MODE 0xFFFFFFFC +#define S_007A28_DACB_AUTODETECT_FRAME_TIME_COUNTER(x) (((x) & 0xff) << 8) +#define G_007A28_DACB_AUTODETECT_FRAME_TIME_COUNTER(x) (((x) >> 8) & 0xff) +#define C_007A28_DACB_AUTODETECT_FRAME_TIME_COUNTER 0xFFFF00FF +#define S_007A28_DACB_AUTODETECT_CHECK_MASK(x) (((x) & 0x3) << 16) +#define G_007A28_DACB_AUTODETECT_CHECK_MASK(x) (((x) >> 16) & 0x3) +#define C_007A28_DACB_AUTODETECT_CHECK_MASK 0xFFFCFFFF +#define R_007A38_DACB_AUTODETECT_INT_CONTROL 0x007A38 +#define S_007A38_DACB_AUTODETECT_ACK(x) (((x) & 0x1) << 0) +#define C_007A38_DACB_DACA_AUTODETECT_ACK 0xFFFFFFFE +#define S_007A38_DACB_AUTODETECT_INT_ENABLE(x) (((x) & 0x1) << 16) +#define G_007A38_DACB_AUTODETECT_INT_ENABLE(x) (((x) >> 16) & 0x1) +#define C_007A38_DACB_AUTODETECT_INT_ENABLE 0xFFFCFFFF +#define R_007D00_DC_HOT_PLUG_DETECT1_CONTROL 0x007D00 +#define S_007D00_DC_HOT_PLUG_DETECT1_EN(x) (((x) & 0x1) << 0) +#define G_007D00_DC_HOT_PLUG_DETECT1_EN(x) (((x) >> 0) & 0x1) +#define C_007D00_DC_HOT_PLUG_DETECT1_EN 0xFFFFFFFE +#define R_007D04_DC_HOT_PLUG_DETECT1_INT_STATUS 0x007D04 +#define S_007D04_DC_HOT_PLUG_DETECT1_INT_STATUS(x) (((x) & 0x1) << 0) +#define G_007D04_DC_HOT_PLUG_DETECT1_INT_STATUS(x) (((x) >> 0) & 0x1) +#define C_007D04_DC_HOT_PLUG_DETECT1_INT_STATUS 0xFFFFFFFE +#define S_007D04_DC_HOT_PLUG_DETECT1_SENSE(x) (((x) & 0x1) << 1) +#define G_007D04_DC_HOT_PLUG_DETECT1_SENSE(x) (((x) >> 1) & 0x1) +#define C_007D04_DC_HOT_PLUG_DETECT1_SENSE 0xFFFFFFFD +#define R_007D08_DC_HOT_PLUG_DETECT1_INT_CONTROL 0x007D08 +#define S_007D08_DC_HOT_PLUG_DETECT1_INT_ACK(x) (((x) & 0x1) << 0) +#define C_007D08_DC_HOT_PLUG_DETECT1_INT_ACK 0xFFFFFFFE +#define S_007D08_DC_HOT_PLUG_DETECT1_INT_POLARITY(x) (((x) & 0x1) << 8) +#define G_007D08_DC_HOT_PLUG_DETECT1_INT_POLARITY(x) (((x) >> 8) & 0x1) +#define C_007D08_DC_HOT_PLUG_DETECT1_INT_POLARITY 0xFFFFFEFF +#define S_007D08_DC_HOT_PLUG_DETECT1_INT_EN(x) (((x) & 0x1) << 16) +#define G_007D08_DC_HOT_PLUG_DETECT1_INT_EN(x) (((x) >> 16) & 0x1) +#define C_007D08_DC_HOT_PLUG_DETECT1_INT_EN 0xFFFEFFFF +#define R_007D10_DC_HOT_PLUG_DETECT2_CONTROL 0x007D10 +#define S_007D10_DC_HOT_PLUG_DETECT2_EN(x) (((x) & 0x1) << 0) +#define G_007D10_DC_HOT_PLUG_DETECT2_EN(x) (((x) >> 0) & 0x1) +#define C_007D10_DC_HOT_PLUG_DETECT2_EN 0xFFFFFFFE +#define R_007D14_DC_HOT_PLUG_DETECT2_INT_STATUS 0x007D14 +#define S_007D14_DC_HOT_PLUG_DETECT2_INT_STATUS(x) (((x) & 0x1) << 0) +#define G_007D14_DC_HOT_PLUG_DETECT2_INT_STATUS(x) (((x) >> 0) & 0x1) +#define C_007D14_DC_HOT_PLUG_DETECT2_INT_STATUS 0xFFFFFFFE +#define S_007D14_DC_HOT_PLUG_DETECT2_SENSE(x) (((x) & 0x1) << 1) +#define G_007D14_DC_HOT_PLUG_DETECT2_SENSE(x) (((x) >> 1) & 0x1) +#define C_007D14_DC_HOT_PLUG_DETECT2_SENSE 0xFFFFFFFD +#define R_007D18_DC_HOT_PLUG_DETECT2_INT_CONTROL 0x007D18 +#define S_007D18_DC_HOT_PLUG_DETECT2_INT_ACK(x) (((x) & 0x1) << 0) +#define C_007D18_DC_HOT_PLUG_DETECT2_INT_ACK 0xFFFFFFFE +#define S_007D18_DC_HOT_PLUG_DETECT2_INT_POLARITY(x) (((x) & 0x1) << 8) +#define G_007D18_DC_HOT_PLUG_DETECT2_INT_POLARITY(x) (((x) >> 8) & 0x1) +#define C_007D18_DC_HOT_PLUG_DETECT2_INT_POLARITY 0xFFFFFEFF +#define S_007D18_DC_HOT_PLUG_DETECT2_INT_EN(x) (((x) & 0x1) << 16) +#define G_007D18_DC_HOT_PLUG_DETECT2_INT_EN(x) (((x) >> 16) & 0x1) +#define C_007D18_DC_HOT_PLUG_DETECT2_INT_EN 0xFFFEFFFF + +/* MC registers */ +#define R_000000_MC_STATUS 0x000000 +#define S_000000_MC_IDLE(x) (((x) & 0x1) << 0) +#define G_000000_MC_IDLE(x) (((x) >> 0) & 0x1) +#define C_000000_MC_IDLE 0xFFFFFFFE +#define R_000004_MC_FB_LOCATION 0x000004 +#define S_000004_MC_FB_START(x) (((x) & 0xFFFF) << 0) +#define G_000004_MC_FB_START(x) (((x) >> 0) & 0xFFFF) +#define C_000004_MC_FB_START 0xFFFF0000 +#define S_000004_MC_FB_TOP(x) (((x) & 0xFFFF) << 16) +#define G_000004_MC_FB_TOP(x) (((x) >> 16) & 0xFFFF) +#define C_000004_MC_FB_TOP 0x0000FFFF +#define R_000005_MC_AGP_LOCATION 0x000005 +#define S_000005_MC_AGP_START(x) (((x) & 0xFFFF) << 0) +#define G_000005_MC_AGP_START(x) (((x) >> 0) & 0xFFFF) +#define C_000005_MC_AGP_START 0xFFFF0000 +#define S_000005_MC_AGP_TOP(x) (((x) & 0xFFFF) << 16) +#define G_000005_MC_AGP_TOP(x) (((x) >> 16) & 0xFFFF) +#define C_000005_MC_AGP_TOP 0x0000FFFF +#define R_000006_AGP_BASE 0x000006 +#define S_000006_AGP_BASE_ADDR(x) (((x) & 0xFFFFFFFF) << 0) +#define G_000006_AGP_BASE_ADDR(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_000006_AGP_BASE_ADDR 0x00000000 +#define R_000007_AGP_BASE_2 0x000007 +#define S_000007_AGP_BASE_ADDR_2(x) (((x) & 0xF) << 0) +#define G_000007_AGP_BASE_ADDR_2(x) (((x) >> 0) & 0xF) +#define C_000007_AGP_BASE_ADDR_2 0xFFFFFFF0 +#define R_000009_MC_CNTL1 0x000009 +#define S_000009_ENABLE_PAGE_TABLES(x) (((x) & 0x1) << 26) +#define G_000009_ENABLE_PAGE_TABLES(x) (((x) >> 26) & 0x1) +#define C_000009_ENABLE_PAGE_TABLES 0xFBFFFFFF +/* FIXME don't know the various field size need feedback from AMD */ +#define R_000100_MC_PT0_CNTL 0x000100 +#define S_000100_ENABLE_PT(x) (((x) & 0x1) << 0) +#define G_000100_ENABLE_PT(x) (((x) >> 0) & 0x1) +#define C_000100_ENABLE_PT 0xFFFFFFFE +#define S_000100_EFFECTIVE_L2_CACHE_SIZE(x) (((x) & 0x7) << 15) +#define G_000100_EFFECTIVE_L2_CACHE_SIZE(x) (((x) >> 15) & 0x7) +#define C_000100_EFFECTIVE_L2_CACHE_SIZE 0xFFFC7FFF +#define S_000100_EFFECTIVE_L2_QUEUE_SIZE(x) (((x) & 0x7) << 21) +#define G_000100_EFFECTIVE_L2_QUEUE_SIZE(x) (((x) >> 21) & 0x7) +#define C_000100_EFFECTIVE_L2_QUEUE_SIZE 0xFF1FFFFF +#define S_000100_INVALIDATE_ALL_L1_TLBS(x) (((x) & 0x1) << 28) +#define G_000100_INVALIDATE_ALL_L1_TLBS(x) (((x) >> 28) & 0x1) +#define C_000100_INVALIDATE_ALL_L1_TLBS 0xEFFFFFFF +#define S_000100_INVALIDATE_L2_CACHE(x) (((x) & 0x1) << 29) +#define G_000100_INVALIDATE_L2_CACHE(x) (((x) >> 29) & 0x1) +#define C_000100_INVALIDATE_L2_CACHE 0xDFFFFFFF +#define R_000102_MC_PT0_CONTEXT0_CNTL 0x000102 +#define S_000102_ENABLE_PAGE_TABLE(x) (((x) & 0x1) << 0) +#define G_000102_ENABLE_PAGE_TABLE(x) (((x) >> 0) & 0x1) +#define C_000102_ENABLE_PAGE_TABLE 0xFFFFFFFE +#define S_000102_PAGE_TABLE_DEPTH(x) (((x) & 0x3) << 1) +#define G_000102_PAGE_TABLE_DEPTH(x) (((x) >> 1) & 0x3) +#define C_000102_PAGE_TABLE_DEPTH 0xFFFFFFF9 +#define V_000102_PAGE_TABLE_FLAT 0 +/* R600 documentation suggest that this should be a number of pages */ +#define R_000112_MC_PT0_SYSTEM_APERTURE_LOW_ADDR 0x000112 +#define R_000114_MC_PT0_SYSTEM_APERTURE_HIGH_ADDR 0x000114 +#define R_00011C_MC_PT0_CONTEXT0_DEFAULT_READ_ADDR 0x00011C +#define R_00012C_MC_PT0_CONTEXT0_FLAT_BASE_ADDR 0x00012C +#define R_00013C_MC_PT0_CONTEXT0_FLAT_START_ADDR 0x00013C +#define R_00014C_MC_PT0_CONTEXT0_FLAT_END_ADDR 0x00014C +#define R_00016C_MC_PT0_CLIENT0_CNTL 0x00016C +#define S_00016C_ENABLE_TRANSLATION_MODE_OVERRIDE(x) (((x) & 0x1) << 0) +#define G_00016C_ENABLE_TRANSLATION_MODE_OVERRIDE(x) (((x) >> 0) & 0x1) +#define C_00016C_ENABLE_TRANSLATION_MODE_OVERRIDE 0xFFFFFFFE +#define S_00016C_TRANSLATION_MODE_OVERRIDE(x) (((x) & 0x1) << 1) +#define G_00016C_TRANSLATION_MODE_OVERRIDE(x) (((x) >> 1) & 0x1) +#define C_00016C_TRANSLATION_MODE_OVERRIDE 0xFFFFFFFD +#define S_00016C_SYSTEM_ACCESS_MODE_MASK(x) (((x) & 0x3) << 8) +#define G_00016C_SYSTEM_ACCESS_MODE_MASK(x) (((x) >> 8) & 0x3) +#define C_00016C_SYSTEM_ACCESS_MODE_MASK 0xFFFFFCFF +#define V_00016C_SYSTEM_ACCESS_MODE_PA_ONLY 0 +#define V_00016C_SYSTEM_ACCESS_MODE_USE_SYS_MAP 1 +#define V_00016C_SYSTEM_ACCESS_MODE_IN_SYS 2 +#define V_00016C_SYSTEM_ACCESS_MODE_NOT_IN_SYS 3 +#define S_00016C_SYSTEM_APERTURE_UNMAPPED_ACCESS(x) (((x) & 0x1) << 10) +#define G_00016C_SYSTEM_APERTURE_UNMAPPED_ACCESS(x) (((x) >> 10) & 0x1) +#define C_00016C_SYSTEM_APERTURE_UNMAPPED_ACCESS 0xFFFFFBFF +#define V_00016C_SYSTEM_APERTURE_UNMAPPED_PASSTHROUGH 0 +#define V_00016C_SYSTEM_APERTURE_UNMAPPED_DEFAULT_PAGE 1 +#define S_00016C_EFFECTIVE_L1_CACHE_SIZE(x) (((x) & 0x7) << 11) +#define G_00016C_EFFECTIVE_L1_CACHE_SIZE(x) (((x) >> 11) & 0x7) +#define C_00016C_EFFECTIVE_L1_CACHE_SIZE 0xFFFFC7FF +#define S_00016C_ENABLE_FRAGMENT_PROCESSING(x) (((x) & 0x1) << 14) +#define G_00016C_ENABLE_FRAGMENT_PROCESSING(x) (((x) >> 14) & 0x1) +#define C_00016C_ENABLE_FRAGMENT_PROCESSING 0xFFFFBFFF +#define S_00016C_EFFECTIVE_L1_QUEUE_SIZE(x) (((x) & 0x7) << 15) +#define G_00016C_EFFECTIVE_L1_QUEUE_SIZE(x) (((x) >> 15) & 0x7) +#define C_00016C_EFFECTIVE_L1_QUEUE_SIZE 0xFFFC7FFF +#define S_00016C_INVALIDATE_L1_TLB(x) (((x) & 0x1) << 20) +#define G_00016C_INVALIDATE_L1_TLB(x) (((x) >> 20) & 0x1) +#define C_00016C_INVALIDATE_L1_TLB 0xFFEFFFFF + +#define R_006548_D1MODE_PRIORITY_A_CNT 0x006548 +#define S_006548_D1MODE_PRIORITY_MARK_A(x) (((x) & 0x7FFF) << 0) +#define G_006548_D1MODE_PRIORITY_MARK_A(x) (((x) >> 0) & 0x7FFF) +#define C_006548_D1MODE_PRIORITY_MARK_A 0xFFFF8000 +#define S_006548_D1MODE_PRIORITY_A_OFF(x) (((x) & 0x1) << 16) +#define G_006548_D1MODE_PRIORITY_A_OFF(x) (((x) >> 16) & 0x1) +#define C_006548_D1MODE_PRIORITY_A_OFF 0xFFFEFFFF +#define S_006548_D1MODE_PRIORITY_A_ALWAYS_ON(x) (((x) & 0x1) << 20) +#define G_006548_D1MODE_PRIORITY_A_ALWAYS_ON(x) (((x) >> 20) & 0x1) +#define C_006548_D1MODE_PRIORITY_A_ALWAYS_ON 0xFFEFFFFF +#define S_006548_D1MODE_PRIORITY_A_FORCE_MASK(x) (((x) & 0x1) << 24) +#define G_006548_D1MODE_PRIORITY_A_FORCE_MASK(x) (((x) >> 24) & 0x1) +#define C_006548_D1MODE_PRIORITY_A_FORCE_MASK 0xFEFFFFFF +#define R_00654C_D1MODE_PRIORITY_B_CNT 0x00654C +#define S_00654C_D1MODE_PRIORITY_MARK_B(x) (((x) & 0x7FFF) << 0) +#define G_00654C_D1MODE_PRIORITY_MARK_B(x) (((x) >> 0) & 0x7FFF) +#define C_00654C_D1MODE_PRIORITY_MARK_B 0xFFFF8000 +#define S_00654C_D1MODE_PRIORITY_B_OFF(x) (((x) & 0x1) << 16) +#define G_00654C_D1MODE_PRIORITY_B_OFF(x) (((x) >> 16) & 0x1) +#define C_00654C_D1MODE_PRIORITY_B_OFF 0xFFFEFFFF +#define S_00654C_D1MODE_PRIORITY_B_ALWAYS_ON(x) (((x) & 0x1) << 20) +#define G_00654C_D1MODE_PRIORITY_B_ALWAYS_ON(x) (((x) >> 20) & 0x1) +#define C_00654C_D1MODE_PRIORITY_B_ALWAYS_ON 0xFFEFFFFF +#define S_00654C_D1MODE_PRIORITY_B_FORCE_MASK(x) (((x) & 0x1) << 24) +#define G_00654C_D1MODE_PRIORITY_B_FORCE_MASK(x) (((x) >> 24) & 0x1) +#define C_00654C_D1MODE_PRIORITY_B_FORCE_MASK 0xFEFFFFFF +#define R_006D48_D2MODE_PRIORITY_A_CNT 0x006D48 +#define S_006D48_D2MODE_PRIORITY_MARK_A(x) (((x) & 0x7FFF) << 0) +#define G_006D48_D2MODE_PRIORITY_MARK_A(x) (((x) >> 0) & 0x7FFF) +#define C_006D48_D2MODE_PRIORITY_MARK_A 0xFFFF8000 +#define S_006D48_D2MODE_PRIORITY_A_OFF(x) (((x) & 0x1) << 16) +#define G_006D48_D2MODE_PRIORITY_A_OFF(x) (((x) >> 16) & 0x1) +#define C_006D48_D2MODE_PRIORITY_A_OFF 0xFFFEFFFF +#define S_006D48_D2MODE_PRIORITY_A_ALWAYS_ON(x) (((x) & 0x1) << 20) +#define G_006D48_D2MODE_PRIORITY_A_ALWAYS_ON(x) (((x) >> 20) & 0x1) +#define C_006D48_D2MODE_PRIORITY_A_ALWAYS_ON 0xFFEFFFFF +#define S_006D48_D2MODE_PRIORITY_A_FORCE_MASK(x) (((x) & 0x1) << 24) +#define G_006D48_D2MODE_PRIORITY_A_FORCE_MASK(x) (((x) >> 24) & 0x1) +#define C_006D48_D2MODE_PRIORITY_A_FORCE_MASK 0xFEFFFFFF +#define R_006D4C_D2MODE_PRIORITY_B_CNT 0x006D4C +#define S_006D4C_D2MODE_PRIORITY_MARK_B(x) (((x) & 0x7FFF) << 0) +#define G_006D4C_D2MODE_PRIORITY_MARK_B(x) (((x) >> 0) & 0x7FFF) +#define C_006D4C_D2MODE_PRIORITY_MARK_B 0xFFFF8000 +#define S_006D4C_D2MODE_PRIORITY_B_OFF(x) (((x) & 0x1) << 16) +#define G_006D4C_D2MODE_PRIORITY_B_OFF(x) (((x) >> 16) & 0x1) +#define C_006D4C_D2MODE_PRIORITY_B_OFF 0xFFFEFFFF +#define S_006D4C_D2MODE_PRIORITY_B_ALWAYS_ON(x) (((x) & 0x1) << 20) +#define G_006D4C_D2MODE_PRIORITY_B_ALWAYS_ON(x) (((x) >> 20) & 0x1) +#define C_006D4C_D2MODE_PRIORITY_B_ALWAYS_ON 0xFFEFFFFF +#define S_006D4C_D2MODE_PRIORITY_B_FORCE_MASK(x) (((x) & 0x1) << 24) +#define G_006D4C_D2MODE_PRIORITY_B_FORCE_MASK(x) (((x) >> 24) & 0x1) +#define C_006D4C_D2MODE_PRIORITY_B_FORCE_MASK 0xFEFFFFFF + +/* PLL regs */ +#define GENERAL_PWRMGT 0x8 +#define GLOBAL_PWRMGT_EN (1 << 0) +#define MOBILE_SU (1 << 2) +#define DYN_PWRMGT_SCLK_LENGTH 0xc +#define NORMAL_POWER_SCLK_HILEN(x) ((x) << 0) +#define NORMAL_POWER_SCLK_LOLEN(x) ((x) << 4) +#define REDUCED_POWER_SCLK_HILEN(x) ((x) << 8) +#define REDUCED_POWER_SCLK_LOLEN(x) ((x) << 12) +#define POWER_D1_SCLK_HILEN(x) ((x) << 16) +#define POWER_D1_SCLK_LOLEN(x) ((x) << 20) +#define STATIC_SCREEN_HILEN(x) ((x) << 24) +#define STATIC_SCREEN_LOLEN(x) ((x) << 28) +#define DYN_SCLK_VOL_CNTL 0xe +#define IO_CG_VOLTAGE_DROP (1 << 0) +#define VOLTAGE_DROP_SYNC (1 << 2) +#define VOLTAGE_DELAY_SEL(x) ((x) << 3) +#define HDP_DYN_CNTL 0x10 +#define HDP_FORCEON (1 << 0) +#define MC_HOST_DYN_CNTL 0x1e +#define MC_HOST_FORCEON (1 << 0) +#define DYN_BACKBIAS_CNTL 0x29 +#define IO_CG_BACKBIAS_EN (1 << 0) + +/* mmreg */ +#define DOUT_POWER_MANAGEMENT_CNTL 0x7ee0 +#define PWRDN_WAIT_BUSY_OFF (1 << 0) +#define PWRDN_WAIT_PWRSEQ_OFF (1 << 4) +#define PWRDN_WAIT_PPLL_OFF (1 << 8) +#define PWRUP_WAIT_PPLL_ON (1 << 12) +#define PWRUP_WAIT_MEM_INIT_DONE (1 << 16) +#define PM_ASSERT_RESET (1 << 20) +#define PM_PWRDN_PPLL (1 << 24) + +#endif diff --git a/drivers/gpu/drm/radeon/rs690.c b/drivers/gpu/drm/radeon/rs690.c new file mode 100644 index 00000000..a9049ed1 --- /dev/null +++ b/drivers/gpu/drm/radeon/rs690.c @@ -0,0 +1,765 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#include "drmP.h" +#include "radeon.h" +#include "radeon_asic.h" +#include "atom.h" +#include "rs690d.h" + +static int rs690_mc_wait_for_idle(struct radeon_device *rdev) +{ + unsigned i; + uint32_t tmp; + + for (i = 0; i < rdev->usec_timeout; i++) { + /* read MC_STATUS */ + tmp = RREG32_MC(R_000090_MC_SYSTEM_STATUS); + if (G_000090_MC_SYSTEM_IDLE(tmp)) + return 0; + udelay(1); + } + return -1; +} + +static void rs690_gpu_init(struct radeon_device *rdev) +{ + /* FIXME: is this correct ? */ + r420_pipes_init(rdev); + if (rs690_mc_wait_for_idle(rdev)) { + printk(KERN_WARNING "Failed to wait MC idle while " + "programming pipes. Bad things might happen.\n"); + } +} + +union igp_info { + struct _ATOM_INTEGRATED_SYSTEM_INFO info; + struct _ATOM_INTEGRATED_SYSTEM_INFO_V2 info_v2; +}; + +void rs690_pm_info(struct radeon_device *rdev) +{ + int index = GetIndexIntoMasterTable(DATA, IntegratedSystemInfo); + union igp_info *info; + uint16_t data_offset; + uint8_t frev, crev; + fixed20_12 tmp; + + if (atom_parse_data_header(rdev->mode_info.atom_context, index, NULL, + &frev, &crev, &data_offset)) { + info = (union igp_info *)(rdev->mode_info.atom_context->bios + data_offset); + + /* Get various system informations from bios */ + switch (crev) { + case 1: + tmp.full = dfixed_const(100); + rdev->pm.igp_sideport_mclk.full = dfixed_const(le32_to_cpu(info->info.ulBootUpMemoryClock)); + rdev->pm.igp_sideport_mclk.full = dfixed_div(rdev->pm.igp_sideport_mclk, tmp); + if (le16_to_cpu(info->info.usK8MemoryClock)) + rdev->pm.igp_system_mclk.full = dfixed_const(le16_to_cpu(info->info.usK8MemoryClock)); + else if (rdev->clock.default_mclk) { + rdev->pm.igp_system_mclk.full = dfixed_const(rdev->clock.default_mclk); + rdev->pm.igp_system_mclk.full = dfixed_div(rdev->pm.igp_system_mclk, tmp); + } else + rdev->pm.igp_system_mclk.full = dfixed_const(400); + rdev->pm.igp_ht_link_clk.full = dfixed_const(le16_to_cpu(info->info.usFSBClock)); + rdev->pm.igp_ht_link_width.full = dfixed_const(info->info.ucHTLinkWidth); + break; + case 2: + tmp.full = dfixed_const(100); + rdev->pm.igp_sideport_mclk.full = dfixed_const(le32_to_cpu(info->info_v2.ulBootUpSidePortClock)); + rdev->pm.igp_sideport_mclk.full = dfixed_div(rdev->pm.igp_sideport_mclk, tmp); + if (le32_to_cpu(info->info_v2.ulBootUpUMAClock)) + rdev->pm.igp_system_mclk.full = dfixed_const(le32_to_cpu(info->info_v2.ulBootUpUMAClock)); + else if (rdev->clock.default_mclk) + rdev->pm.igp_system_mclk.full = dfixed_const(rdev->clock.default_mclk); + else + rdev->pm.igp_system_mclk.full = dfixed_const(66700); + rdev->pm.igp_system_mclk.full = dfixed_div(rdev->pm.igp_system_mclk, tmp); + rdev->pm.igp_ht_link_clk.full = dfixed_const(le32_to_cpu(info->info_v2.ulHTLinkFreq)); + rdev->pm.igp_ht_link_clk.full = dfixed_div(rdev->pm.igp_ht_link_clk, tmp); + rdev->pm.igp_ht_link_width.full = dfixed_const(le16_to_cpu(info->info_v2.usMinHTLinkWidth)); + break; + default: + /* We assume the slower possible clock ie worst case */ + rdev->pm.igp_sideport_mclk.full = dfixed_const(200); + rdev->pm.igp_system_mclk.full = dfixed_const(200); + rdev->pm.igp_ht_link_clk.full = dfixed_const(1000); + rdev->pm.igp_ht_link_width.full = dfixed_const(8); + DRM_ERROR("No integrated system info for your GPU, using safe default\n"); + break; + } + } else { + /* We assume the slower possible clock ie worst case */ + rdev->pm.igp_sideport_mclk.full = dfixed_const(200); + rdev->pm.igp_system_mclk.full = dfixed_const(200); + rdev->pm.igp_ht_link_clk.full = dfixed_const(1000); + rdev->pm.igp_ht_link_width.full = dfixed_const(8); + DRM_ERROR("No integrated system info for your GPU, using safe default\n"); + } + /* Compute various bandwidth */ + /* k8_bandwidth = (memory_clk / 2) * 2 * 8 * 0.5 = memory_clk * 4 */ + tmp.full = dfixed_const(4); + rdev->pm.k8_bandwidth.full = dfixed_mul(rdev->pm.igp_system_mclk, tmp); + /* ht_bandwidth = ht_clk * 2 * ht_width / 8 * 0.8 + * = ht_clk * ht_width / 5 + */ + tmp.full = dfixed_const(5); + rdev->pm.ht_bandwidth.full = dfixed_mul(rdev->pm.igp_ht_link_clk, + rdev->pm.igp_ht_link_width); + rdev->pm.ht_bandwidth.full = dfixed_div(rdev->pm.ht_bandwidth, tmp); + if (tmp.full < rdev->pm.max_bandwidth.full) { + /* HT link is a limiting factor */ + rdev->pm.max_bandwidth.full = tmp.full; + } + /* sideport_bandwidth = (sideport_clk / 2) * 2 * 2 * 0.7 + * = (sideport_clk * 14) / 10 + */ + tmp.full = dfixed_const(14); + rdev->pm.sideport_bandwidth.full = dfixed_mul(rdev->pm.igp_sideport_mclk, tmp); + tmp.full = dfixed_const(10); + rdev->pm.sideport_bandwidth.full = dfixed_div(rdev->pm.sideport_bandwidth, tmp); +} + +void rs690_mc_init(struct radeon_device *rdev) +{ + u64 base; + + rs400_gart_adjust_size(rdev); + rdev->mc.vram_is_ddr = true; + rdev->mc.vram_width = 128; + rdev->mc.real_vram_size = RREG32(RADEON_CONFIG_MEMSIZE); + rdev->mc.mc_vram_size = rdev->mc.real_vram_size; + rdev->mc.aper_base = pci_resource_start(rdev->pdev, 0); + rdev->mc.aper_size = pci_resource_len(rdev->pdev, 0); + rdev->mc.visible_vram_size = rdev->mc.aper_size; + base = RREG32_MC(R_000100_MCCFG_FB_LOCATION); + base = G_000100_MC_FB_START(base) << 16; + rdev->mc.igp_sideport_enabled = radeon_atombios_sideport_present(rdev); + rs690_pm_info(rdev); + radeon_vram_location(rdev, &rdev->mc, base); + rdev->mc.gtt_base_align = rdev->mc.gtt_size - 1; + radeon_gtt_location(rdev, &rdev->mc); + radeon_update_bandwidth_info(rdev); +} + +void rs690_line_buffer_adjust(struct radeon_device *rdev, + struct drm_display_mode *mode1, + struct drm_display_mode *mode2) +{ + u32 tmp; + + /* + * Line Buffer Setup + * There is a single line buffer shared by both display controllers. + * R_006520_DC_LB_MEMORY_SPLIT controls how that line buffer is shared between + * the display controllers. The paritioning can either be done + * manually or via one of four preset allocations specified in bits 1:0: + * 0 - line buffer is divided in half and shared between crtc + * 1 - D1 gets 3/4 of the line buffer, D2 gets 1/4 + * 2 - D1 gets the whole buffer + * 3 - D1 gets 1/4 of the line buffer, D2 gets 3/4 + * Setting bit 2 of R_006520_DC_LB_MEMORY_SPLIT controls switches to manual + * allocation mode. In manual allocation mode, D1 always starts at 0, + * D1 end/2 is specified in bits 14:4; D2 allocation follows D1. + */ + tmp = RREG32(R_006520_DC_LB_MEMORY_SPLIT) & C_006520_DC_LB_MEMORY_SPLIT; + tmp &= ~C_006520_DC_LB_MEMORY_SPLIT_MODE; + /* auto */ + if (mode1 && mode2) { + if (mode1->hdisplay > mode2->hdisplay) { + if (mode1->hdisplay > 2560) + tmp |= V_006520_DC_LB_MEMORY_SPLIT_D1_3Q_D2_1Q; + else + tmp |= V_006520_DC_LB_MEMORY_SPLIT_D1HALF_D2HALF; + } else if (mode2->hdisplay > mode1->hdisplay) { + if (mode2->hdisplay > 2560) + tmp |= V_006520_DC_LB_MEMORY_SPLIT_D1_1Q_D2_3Q; + else + tmp |= V_006520_DC_LB_MEMORY_SPLIT_D1HALF_D2HALF; + } else + tmp |= V_006520_DC_LB_MEMORY_SPLIT_D1HALF_D2HALF; + } else if (mode1) { + tmp |= V_006520_DC_LB_MEMORY_SPLIT_D1_ONLY; + } else if (mode2) { + tmp |= V_006520_DC_LB_MEMORY_SPLIT_D1_1Q_D2_3Q; + } + WREG32(R_006520_DC_LB_MEMORY_SPLIT, tmp); +} + +struct rs690_watermark { + u32 lb_request_fifo_depth; + fixed20_12 num_line_pair; + fixed20_12 estimated_width; + fixed20_12 worst_case_latency; + fixed20_12 consumption_rate; + fixed20_12 active_time; + fixed20_12 dbpp; + fixed20_12 priority_mark_max; + fixed20_12 priority_mark; + fixed20_12 sclk; +}; + +void rs690_crtc_bandwidth_compute(struct radeon_device *rdev, + struct radeon_crtc *crtc, + struct rs690_watermark *wm) +{ + struct drm_display_mode *mode = &crtc->base.mode; + fixed20_12 a, b, c; + fixed20_12 pclk, request_fifo_depth, tolerable_latency, estimated_width; + fixed20_12 consumption_time, line_time, chunk_time, read_delay_latency; + + if (!crtc->base.enabled) { + /* FIXME: wouldn't it better to set priority mark to maximum */ + wm->lb_request_fifo_depth = 4; + return; + } + + if (crtc->vsc.full > dfixed_const(2)) + wm->num_line_pair.full = dfixed_const(2); + else + wm->num_line_pair.full = dfixed_const(1); + + b.full = dfixed_const(mode->crtc_hdisplay); + c.full = dfixed_const(256); + a.full = dfixed_div(b, c); + request_fifo_depth.full = dfixed_mul(a, wm->num_line_pair); + request_fifo_depth.full = dfixed_ceil(request_fifo_depth); + if (a.full < dfixed_const(4)) { + wm->lb_request_fifo_depth = 4; + } else { + wm->lb_request_fifo_depth = dfixed_trunc(request_fifo_depth); + } + + /* Determine consumption rate + * pclk = pixel clock period(ns) = 1000 / (mode.clock / 1000) + * vtaps = number of vertical taps, + * vsc = vertical scaling ratio, defined as source/destination + * hsc = horizontal scaling ration, defined as source/destination + */ + a.full = dfixed_const(mode->clock); + b.full = dfixed_const(1000); + a.full = dfixed_div(a, b); + pclk.full = dfixed_div(b, a); + if (crtc->rmx_type != RMX_OFF) { + b.full = dfixed_const(2); + if (crtc->vsc.full > b.full) + b.full = crtc->vsc.full; + b.full = dfixed_mul(b, crtc->hsc); + c.full = dfixed_const(2); + b.full = dfixed_div(b, c); + consumption_time.full = dfixed_div(pclk, b); + } else { + consumption_time.full = pclk.full; + } + a.full = dfixed_const(1); + wm->consumption_rate.full = dfixed_div(a, consumption_time); + + + /* Determine line time + * LineTime = total time for one line of displayhtotal + * LineTime = total number of horizontal pixels + * pclk = pixel clock period(ns) + */ + a.full = dfixed_const(crtc->base.mode.crtc_htotal); + line_time.full = dfixed_mul(a, pclk); + + /* Determine active time + * ActiveTime = time of active region of display within one line, + * hactive = total number of horizontal active pixels + * htotal = total number of horizontal pixels + */ + a.full = dfixed_const(crtc->base.mode.crtc_htotal); + b.full = dfixed_const(crtc->base.mode.crtc_hdisplay); + wm->active_time.full = dfixed_mul(line_time, b); + wm->active_time.full = dfixed_div(wm->active_time, a); + + /* Maximun bandwidth is the minimun bandwidth of all component */ + rdev->pm.max_bandwidth = rdev->pm.core_bandwidth; + if (rdev->mc.igp_sideport_enabled) { + if (rdev->pm.max_bandwidth.full > rdev->pm.sideport_bandwidth.full && + rdev->pm.sideport_bandwidth.full) + rdev->pm.max_bandwidth = rdev->pm.sideport_bandwidth; + read_delay_latency.full = dfixed_const(370 * 800 * 1000); + read_delay_latency.full = dfixed_div(read_delay_latency, + rdev->pm.igp_sideport_mclk); + } else { + if (rdev->pm.max_bandwidth.full > rdev->pm.k8_bandwidth.full && + rdev->pm.k8_bandwidth.full) + rdev->pm.max_bandwidth = rdev->pm.k8_bandwidth; + if (rdev->pm.max_bandwidth.full > rdev->pm.ht_bandwidth.full && + rdev->pm.ht_bandwidth.full) + rdev->pm.max_bandwidth = rdev->pm.ht_bandwidth; + read_delay_latency.full = dfixed_const(5000); + } + + /* sclk = system clocks(ns) = 1000 / max_bandwidth / 16 */ + a.full = dfixed_const(16); + rdev->pm.sclk.full = dfixed_mul(rdev->pm.max_bandwidth, a); + a.full = dfixed_const(1000); + rdev->pm.sclk.full = dfixed_div(a, rdev->pm.sclk); + /* Determine chunk time + * ChunkTime = the time it takes the DCP to send one chunk of data + * to the LB which consists of pipeline delay and inter chunk gap + * sclk = system clock(ns) + */ + a.full = dfixed_const(256 * 13); + chunk_time.full = dfixed_mul(rdev->pm.sclk, a); + a.full = dfixed_const(10); + chunk_time.full = dfixed_div(chunk_time, a); + + /* Determine the worst case latency + * NumLinePair = Number of line pairs to request(1=2 lines, 2=4 lines) + * WorstCaseLatency = worst case time from urgent to when the MC starts + * to return data + * READ_DELAY_IDLE_MAX = constant of 1us + * ChunkTime = time it takes the DCP to send one chunk of data to the LB + * which consists of pipeline delay and inter chunk gap + */ + if (dfixed_trunc(wm->num_line_pair) > 1) { + a.full = dfixed_const(3); + wm->worst_case_latency.full = dfixed_mul(a, chunk_time); + wm->worst_case_latency.full += read_delay_latency.full; + } else { + a.full = dfixed_const(2); + wm->worst_case_latency.full = dfixed_mul(a, chunk_time); + wm->worst_case_latency.full += read_delay_latency.full; + } + + /* Determine the tolerable latency + * TolerableLatency = Any given request has only 1 line time + * for the data to be returned + * LBRequestFifoDepth = Number of chunk requests the LB can + * put into the request FIFO for a display + * LineTime = total time for one line of display + * ChunkTime = the time it takes the DCP to send one chunk + * of data to the LB which consists of + * pipeline delay and inter chunk gap + */ + if ((2+wm->lb_request_fifo_depth) >= dfixed_trunc(request_fifo_depth)) { + tolerable_latency.full = line_time.full; + } else { + tolerable_latency.full = dfixed_const(wm->lb_request_fifo_depth - 2); + tolerable_latency.full = request_fifo_depth.full - tolerable_latency.full; + tolerable_latency.full = dfixed_mul(tolerable_latency, chunk_time); + tolerable_latency.full = line_time.full - tolerable_latency.full; + } + /* We assume worst case 32bits (4 bytes) */ + wm->dbpp.full = dfixed_const(4 * 8); + + /* Determine the maximum priority mark + * width = viewport width in pixels + */ + a.full = dfixed_const(16); + wm->priority_mark_max.full = dfixed_const(crtc->base.mode.crtc_hdisplay); + wm->priority_mark_max.full = dfixed_div(wm->priority_mark_max, a); + wm->priority_mark_max.full = dfixed_ceil(wm->priority_mark_max); + + /* Determine estimated width */ + estimated_width.full = tolerable_latency.full - wm->worst_case_latency.full; + estimated_width.full = dfixed_div(estimated_width, consumption_time); + if (dfixed_trunc(estimated_width) > crtc->base.mode.crtc_hdisplay) { + wm->priority_mark.full = dfixed_const(10); + } else { + a.full = dfixed_const(16); + wm->priority_mark.full = dfixed_div(estimated_width, a); + wm->priority_mark.full = dfixed_ceil(wm->priority_mark); + wm->priority_mark.full = wm->priority_mark_max.full - wm->priority_mark.full; + } +} + +void rs690_bandwidth_update(struct radeon_device *rdev) +{ + struct drm_display_mode *mode0 = NULL; + struct drm_display_mode *mode1 = NULL; + struct rs690_watermark wm0; + struct rs690_watermark wm1; + u32 tmp; + u32 d1mode_priority_a_cnt = S_006548_D1MODE_PRIORITY_A_OFF(1); + u32 d2mode_priority_a_cnt = S_006548_D1MODE_PRIORITY_A_OFF(1); + fixed20_12 priority_mark02, priority_mark12, fill_rate; + fixed20_12 a, b; + + radeon_update_display_priority(rdev); + + if (rdev->mode_info.crtcs[0]->base.enabled) + mode0 = &rdev->mode_info.crtcs[0]->base.mode; + if (rdev->mode_info.crtcs[1]->base.enabled) + mode1 = &rdev->mode_info.crtcs[1]->base.mode; + /* + * Set display0/1 priority up in the memory controller for + * modes if the user specifies HIGH for displaypriority + * option. + */ + if ((rdev->disp_priority == 2) && + ((rdev->family == CHIP_RS690) || (rdev->family == CHIP_RS740))) { + tmp = RREG32_MC(R_000104_MC_INIT_MISC_LAT_TIMER); + tmp &= C_000104_MC_DISP0R_INIT_LAT; + tmp &= C_000104_MC_DISP1R_INIT_LAT; + if (mode0) + tmp |= S_000104_MC_DISP0R_INIT_LAT(1); + if (mode1) + tmp |= S_000104_MC_DISP1R_INIT_LAT(1); + WREG32_MC(R_000104_MC_INIT_MISC_LAT_TIMER, tmp); + } + rs690_line_buffer_adjust(rdev, mode0, mode1); + + if ((rdev->family == CHIP_RS690) || (rdev->family == CHIP_RS740)) + WREG32(R_006C9C_DCP_CONTROL, 0); + if ((rdev->family == CHIP_RS780) || (rdev->family == CHIP_RS880)) + WREG32(R_006C9C_DCP_CONTROL, 2); + + rs690_crtc_bandwidth_compute(rdev, rdev->mode_info.crtcs[0], &wm0); + rs690_crtc_bandwidth_compute(rdev, rdev->mode_info.crtcs[1], &wm1); + + tmp = (wm0.lb_request_fifo_depth - 1); + tmp |= (wm1.lb_request_fifo_depth - 1) << 16; + WREG32(R_006D58_LB_MAX_REQ_OUTSTANDING, tmp); + + if (mode0 && mode1) { + if (dfixed_trunc(wm0.dbpp) > 64) + a.full = dfixed_mul(wm0.dbpp, wm0.num_line_pair); + else + a.full = wm0.num_line_pair.full; + if (dfixed_trunc(wm1.dbpp) > 64) + b.full = dfixed_mul(wm1.dbpp, wm1.num_line_pair); + else + b.full = wm1.num_line_pair.full; + a.full += b.full; + fill_rate.full = dfixed_div(wm0.sclk, a); + if (wm0.consumption_rate.full > fill_rate.full) { + b.full = wm0.consumption_rate.full - fill_rate.full; + b.full = dfixed_mul(b, wm0.active_time); + a.full = dfixed_mul(wm0.worst_case_latency, + wm0.consumption_rate); + a.full = a.full + b.full; + b.full = dfixed_const(16 * 1000); + priority_mark02.full = dfixed_div(a, b); + } else { + a.full = dfixed_mul(wm0.worst_case_latency, + wm0.consumption_rate); + b.full = dfixed_const(16 * 1000); + priority_mark02.full = dfixed_div(a, b); + } + if (wm1.consumption_rate.full > fill_rate.full) { + b.full = wm1.consumption_rate.full - fill_rate.full; + b.full = dfixed_mul(b, wm1.active_time); + a.full = dfixed_mul(wm1.worst_case_latency, + wm1.consumption_rate); + a.full = a.full + b.full; + b.full = dfixed_const(16 * 1000); + priority_mark12.full = dfixed_div(a, b); + } else { + a.full = dfixed_mul(wm1.worst_case_latency, + wm1.consumption_rate); + b.full = dfixed_const(16 * 1000); + priority_mark12.full = dfixed_div(a, b); + } + if (wm0.priority_mark.full > priority_mark02.full) + priority_mark02.full = wm0.priority_mark.full; + if (dfixed_trunc(priority_mark02) < 0) + priority_mark02.full = 0; + if (wm0.priority_mark_max.full > priority_mark02.full) + priority_mark02.full = wm0.priority_mark_max.full; + if (wm1.priority_mark.full > priority_mark12.full) + priority_mark12.full = wm1.priority_mark.full; + if (dfixed_trunc(priority_mark12) < 0) + priority_mark12.full = 0; + if (wm1.priority_mark_max.full > priority_mark12.full) + priority_mark12.full = wm1.priority_mark_max.full; + d1mode_priority_a_cnt = dfixed_trunc(priority_mark02); + d2mode_priority_a_cnt = dfixed_trunc(priority_mark12); + if (rdev->disp_priority == 2) { + d1mode_priority_a_cnt |= S_006548_D1MODE_PRIORITY_A_ALWAYS_ON(1); + d2mode_priority_a_cnt |= S_006D48_D2MODE_PRIORITY_A_ALWAYS_ON(1); + } + } else if (mode0) { + if (dfixed_trunc(wm0.dbpp) > 64) + a.full = dfixed_mul(wm0.dbpp, wm0.num_line_pair); + else + a.full = wm0.num_line_pair.full; + fill_rate.full = dfixed_div(wm0.sclk, a); + if (wm0.consumption_rate.full > fill_rate.full) { + b.full = wm0.consumption_rate.full - fill_rate.full; + b.full = dfixed_mul(b, wm0.active_time); + a.full = dfixed_mul(wm0.worst_case_latency, + wm0.consumption_rate); + a.full = a.full + b.full; + b.full = dfixed_const(16 * 1000); + priority_mark02.full = dfixed_div(a, b); + } else { + a.full = dfixed_mul(wm0.worst_case_latency, + wm0.consumption_rate); + b.full = dfixed_const(16 * 1000); + priority_mark02.full = dfixed_div(a, b); + } + if (wm0.priority_mark.full > priority_mark02.full) + priority_mark02.full = wm0.priority_mark.full; + if (dfixed_trunc(priority_mark02) < 0) + priority_mark02.full = 0; + if (wm0.priority_mark_max.full > priority_mark02.full) + priority_mark02.full = wm0.priority_mark_max.full; + d1mode_priority_a_cnt = dfixed_trunc(priority_mark02); + if (rdev->disp_priority == 2) + d1mode_priority_a_cnt |= S_006548_D1MODE_PRIORITY_A_ALWAYS_ON(1); + } else if (mode1) { + if (dfixed_trunc(wm1.dbpp) > 64) + a.full = dfixed_mul(wm1.dbpp, wm1.num_line_pair); + else + a.full = wm1.num_line_pair.full; + fill_rate.full = dfixed_div(wm1.sclk, a); + if (wm1.consumption_rate.full > fill_rate.full) { + b.full = wm1.consumption_rate.full - fill_rate.full; + b.full = dfixed_mul(b, wm1.active_time); + a.full = dfixed_mul(wm1.worst_case_latency, + wm1.consumption_rate); + a.full = a.full + b.full; + b.full = dfixed_const(16 * 1000); + priority_mark12.full = dfixed_div(a, b); + } else { + a.full = dfixed_mul(wm1.worst_case_latency, + wm1.consumption_rate); + b.full = dfixed_const(16 * 1000); + priority_mark12.full = dfixed_div(a, b); + } + if (wm1.priority_mark.full > priority_mark12.full) + priority_mark12.full = wm1.priority_mark.full; + if (dfixed_trunc(priority_mark12) < 0) + priority_mark12.full = 0; + if (wm1.priority_mark_max.full > priority_mark12.full) + priority_mark12.full = wm1.priority_mark_max.full; + d2mode_priority_a_cnt = dfixed_trunc(priority_mark12); + if (rdev->disp_priority == 2) + d2mode_priority_a_cnt |= S_006D48_D2MODE_PRIORITY_A_ALWAYS_ON(1); + } + + WREG32(R_006548_D1MODE_PRIORITY_A_CNT, d1mode_priority_a_cnt); + WREG32(R_00654C_D1MODE_PRIORITY_B_CNT, d1mode_priority_a_cnt); + WREG32(R_006D48_D2MODE_PRIORITY_A_CNT, d2mode_priority_a_cnt); + WREG32(R_006D4C_D2MODE_PRIORITY_B_CNT, d2mode_priority_a_cnt); +} + +uint32_t rs690_mc_rreg(struct radeon_device *rdev, uint32_t reg) +{ + uint32_t r; + + WREG32(R_000078_MC_INDEX, S_000078_MC_IND_ADDR(reg)); + r = RREG32(R_00007C_MC_DATA); + WREG32(R_000078_MC_INDEX, ~C_000078_MC_IND_ADDR); + return r; +} + +void rs690_mc_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v) +{ + WREG32(R_000078_MC_INDEX, S_000078_MC_IND_ADDR(reg) | + S_000078_MC_IND_WR_EN(1)); + WREG32(R_00007C_MC_DATA, v); + WREG32(R_000078_MC_INDEX, 0x7F); +} + +void rs690_mc_program(struct radeon_device *rdev) +{ + struct rv515_mc_save save; + + /* Stops all mc clients */ + rv515_mc_stop(rdev, &save); + + /* Wait for mc idle */ + if (rs690_mc_wait_for_idle(rdev)) + dev_warn(rdev->dev, "Wait MC idle timeout before updating MC.\n"); + /* Program MC, should be a 32bits limited address space */ + WREG32_MC(R_000100_MCCFG_FB_LOCATION, + S_000100_MC_FB_START(rdev->mc.vram_start >> 16) | + S_000100_MC_FB_TOP(rdev->mc.vram_end >> 16)); + WREG32(R_000134_HDP_FB_LOCATION, + S_000134_HDP_FB_START(rdev->mc.vram_start >> 16)); + + rv515_mc_resume(rdev, &save); +} + +static int rs690_startup(struct radeon_device *rdev) +{ + int r; + + rs690_mc_program(rdev); + /* Resume clock */ + rv515_clock_startup(rdev); + /* Initialize GPU configuration (# pipes, ...) */ + rs690_gpu_init(rdev); + /* Initialize GART (initialize after TTM so we can allocate + * memory through TTM but finalize after TTM) */ + r = rs400_gart_enable(rdev); + if (r) + return r; + + /* allocate wb buffer */ + r = radeon_wb_init(rdev); + if (r) + return r; + + /* Enable IRQ */ + rs600_irq_set(rdev); + rdev->config.r300.hdp_cntl = RREG32(RADEON_HOST_PATH_CNTL); + /* 1M ring buffer */ + r = r100_cp_init(rdev, 1024 * 1024); + if (r) { + dev_err(rdev->dev, "failed initializing CP (%d).\n", r); + return r; + } + r = r100_ib_init(rdev); + if (r) { + dev_err(rdev->dev, "failed initializing IB (%d).\n", r); + return r; + } + + r = r600_audio_init(rdev); + if (r) { + dev_err(rdev->dev, "failed initializing audio\n"); + return r; + } + + return 0; +} + +int rs690_resume(struct radeon_device *rdev) +{ + /* Make sur GART are not working */ + rs400_gart_disable(rdev); + /* Resume clock before doing reset */ + rv515_clock_startup(rdev); + /* Reset gpu before posting otherwise ATOM will enter infinite loop */ + if (radeon_asic_reset(rdev)) { + dev_warn(rdev->dev, "GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n", + RREG32(R_000E40_RBBM_STATUS), + RREG32(R_0007C0_CP_STAT)); + } + /* post */ + atom_asic_init(rdev->mode_info.atom_context); + /* Resume clock after posting */ + rv515_clock_startup(rdev); + /* Initialize surface registers */ + radeon_surface_init(rdev); + return rs690_startup(rdev); +} + +int rs690_suspend(struct radeon_device *rdev) +{ + r600_audio_fini(rdev); + r100_cp_disable(rdev); + radeon_wb_disable(rdev); + rs600_irq_disable(rdev); + rs400_gart_disable(rdev); + return 0; +} + +void rs690_fini(struct radeon_device *rdev) +{ + r600_audio_fini(rdev); + r100_cp_fini(rdev); + radeon_wb_fini(rdev); + r100_ib_fini(rdev); + radeon_gem_fini(rdev); + rs400_gart_fini(rdev); + radeon_irq_kms_fini(rdev); + radeon_fence_driver_fini(rdev); + radeon_bo_fini(rdev); + radeon_atombios_fini(rdev); + kfree(rdev->bios); + rdev->bios = NULL; +} + +int rs690_init(struct radeon_device *rdev) +{ + int r; + + /* Disable VGA */ + rv515_vga_render_disable(rdev); + /* Initialize scratch registers */ + radeon_scratch_init(rdev); + /* Initialize surface registers */ + radeon_surface_init(rdev); + /* restore some register to sane defaults */ + r100_restore_sanity(rdev); + /* TODO: disable VGA need to use VGA request */ + /* BIOS*/ + if (!radeon_get_bios(rdev)) { + if (ASIC_IS_AVIVO(rdev)) + return -EINVAL; + } + if (rdev->is_atom_bios) { + r = radeon_atombios_init(rdev); + if (r) + return r; + } else { + dev_err(rdev->dev, "Expecting atombios for RV515 GPU\n"); + return -EINVAL; + } + /* Reset gpu before posting otherwise ATOM will enter infinite loop */ + if (radeon_asic_reset(rdev)) { + dev_warn(rdev->dev, + "GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n", + RREG32(R_000E40_RBBM_STATUS), + RREG32(R_0007C0_CP_STAT)); + } + /* check if cards are posted or not */ + if (radeon_boot_test_post_card(rdev) == false) + return -EINVAL; + + /* Initialize clocks */ + radeon_get_clock_info(rdev->ddev); + /* initialize memory controller */ + rs690_mc_init(rdev); + rv515_debugfs(rdev); + /* Fence driver */ + r = radeon_fence_driver_init(rdev); + if (r) + return r; + r = radeon_irq_kms_init(rdev); + if (r) + return r; + /* Memory manager */ + r = radeon_bo_init(rdev); + if (r) + return r; + r = rs400_gart_init(rdev); + if (r) + return r; + rs600_set_safe_registers(rdev); + rdev->accel_working = true; + r = rs690_startup(rdev); + if (r) { + /* Somethings want wront with the accel init stop accel */ + dev_err(rdev->dev, "Disabling GPU acceleration\n"); + r100_cp_fini(rdev); + radeon_wb_fini(rdev); + r100_ib_fini(rdev); + rs400_gart_fini(rdev); + radeon_irq_kms_fini(rdev); + rdev->accel_working = false; + } + return 0; +} diff --git a/drivers/gpu/drm/radeon/rs690d.h b/drivers/gpu/drm/radeon/rs690d.h new file mode 100644 index 00000000..36e6398a --- /dev/null +++ b/drivers/gpu/drm/radeon/rs690d.h @@ -0,0 +1,310 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#ifndef __RS690D_H__ +#define __RS690D_H__ + +/* Registers */ +#define R_000078_MC_INDEX 0x000078 +#define S_000078_MC_IND_ADDR(x) (((x) & 0x1FF) << 0) +#define G_000078_MC_IND_ADDR(x) (((x) >> 0) & 0x1FF) +#define C_000078_MC_IND_ADDR 0xFFFFFE00 +#define S_000078_MC_IND_WR_EN(x) (((x) & 0x1) << 9) +#define G_000078_MC_IND_WR_EN(x) (((x) >> 9) & 0x1) +#define C_000078_MC_IND_WR_EN 0xFFFFFDFF +#define R_00007C_MC_DATA 0x00007C +#define S_00007C_MC_DATA(x) (((x) & 0xFFFFFFFF) << 0) +#define G_00007C_MC_DATA(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_00007C_MC_DATA 0x00000000 +#define R_0000F8_CONFIG_MEMSIZE 0x0000F8 +#define S_0000F8_CONFIG_MEMSIZE(x) (((x) & 0xFFFFFFFF) << 0) +#define G_0000F8_CONFIG_MEMSIZE(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_0000F8_CONFIG_MEMSIZE 0x00000000 +#define R_000134_HDP_FB_LOCATION 0x000134 +#define S_000134_HDP_FB_START(x) (((x) & 0xFFFF) << 0) +#define G_000134_HDP_FB_START(x) (((x) >> 0) & 0xFFFF) +#define C_000134_HDP_FB_START 0xFFFF0000 +#define R_0007C0_CP_STAT 0x0007C0 +#define S_0007C0_MRU_BUSY(x) (((x) & 0x1) << 0) +#define G_0007C0_MRU_BUSY(x) (((x) >> 0) & 0x1) +#define C_0007C0_MRU_BUSY 0xFFFFFFFE +#define S_0007C0_MWU_BUSY(x) (((x) & 0x1) << 1) +#define G_0007C0_MWU_BUSY(x) (((x) >> 1) & 0x1) +#define C_0007C0_MWU_BUSY 0xFFFFFFFD +#define S_0007C0_RSIU_BUSY(x) (((x) & 0x1) << 2) +#define G_0007C0_RSIU_BUSY(x) (((x) >> 2) & 0x1) +#define C_0007C0_RSIU_BUSY 0xFFFFFFFB +#define S_0007C0_RCIU_BUSY(x) (((x) & 0x1) << 3) +#define G_0007C0_RCIU_BUSY(x) (((x) >> 3) & 0x1) +#define C_0007C0_RCIU_BUSY 0xFFFFFFF7 +#define S_0007C0_CSF_PRIMARY_BUSY(x) (((x) & 0x1) << 9) +#define G_0007C0_CSF_PRIMARY_BUSY(x) (((x) >> 9) & 0x1) +#define C_0007C0_CSF_PRIMARY_BUSY 0xFFFFFDFF +#define S_0007C0_CSF_INDIRECT_BUSY(x) (((x) & 0x1) << 10) +#define G_0007C0_CSF_INDIRECT_BUSY(x) (((x) >> 10) & 0x1) +#define C_0007C0_CSF_INDIRECT_BUSY 0xFFFFFBFF +#define S_0007C0_CSQ_PRIMARY_BUSY(x) (((x) & 0x1) << 11) +#define G_0007C0_CSQ_PRIMARY_BUSY(x) (((x) >> 11) & 0x1) +#define C_0007C0_CSQ_PRIMARY_BUSY 0xFFFFF7FF +#define S_0007C0_CSQ_INDIRECT_BUSY(x) (((x) & 0x1) << 12) +#define G_0007C0_CSQ_INDIRECT_BUSY(x) (((x) >> 12) & 0x1) +#define C_0007C0_CSQ_INDIRECT_BUSY 0xFFFFEFFF +#define S_0007C0_CSI_BUSY(x) (((x) & 0x1) << 13) +#define G_0007C0_CSI_BUSY(x) (((x) >> 13) & 0x1) +#define C_0007C0_CSI_BUSY 0xFFFFDFFF +#define S_0007C0_CSF_INDIRECT2_BUSY(x) (((x) & 0x1) << 14) +#define G_0007C0_CSF_INDIRECT2_BUSY(x) (((x) >> 14) & 0x1) +#define C_0007C0_CSF_INDIRECT2_BUSY 0xFFFFBFFF +#define S_0007C0_CSQ_INDIRECT2_BUSY(x) (((x) & 0x1) << 15) +#define G_0007C0_CSQ_INDIRECT2_BUSY(x) (((x) >> 15) & 0x1) +#define C_0007C0_CSQ_INDIRECT2_BUSY 0xFFFF7FFF +#define S_0007C0_GUIDMA_BUSY(x) (((x) & 0x1) << 28) +#define G_0007C0_GUIDMA_BUSY(x) (((x) >> 28) & 0x1) +#define C_0007C0_GUIDMA_BUSY 0xEFFFFFFF +#define S_0007C0_VIDDMA_BUSY(x) (((x) & 0x1) << 29) +#define G_0007C0_VIDDMA_BUSY(x) (((x) >> 29) & 0x1) +#define C_0007C0_VIDDMA_BUSY 0xDFFFFFFF +#define S_0007C0_CMDSTRM_BUSY(x) (((x) & 0x1) << 30) +#define G_0007C0_CMDSTRM_BUSY(x) (((x) >> 30) & 0x1) +#define C_0007C0_CMDSTRM_BUSY 0xBFFFFFFF +#define S_0007C0_CP_BUSY(x) (((x) & 0x1) << 31) +#define G_0007C0_CP_BUSY(x) (((x) >> 31) & 0x1) +#define C_0007C0_CP_BUSY 0x7FFFFFFF +#define R_000E40_RBBM_STATUS 0x000E40 +#define S_000E40_CMDFIFO_AVAIL(x) (((x) & 0x7F) << 0) +#define G_000E40_CMDFIFO_AVAIL(x) (((x) >> 0) & 0x7F) +#define C_000E40_CMDFIFO_AVAIL 0xFFFFFF80 +#define S_000E40_HIRQ_ON_RBB(x) (((x) & 0x1) << 8) +#define G_000E40_HIRQ_ON_RBB(x) (((x) >> 8) & 0x1) +#define C_000E40_HIRQ_ON_RBB 0xFFFFFEFF +#define S_000E40_CPRQ_ON_RBB(x) (((x) & 0x1) << 9) +#define G_000E40_CPRQ_ON_RBB(x) (((x) >> 9) & 0x1) +#define C_000E40_CPRQ_ON_RBB 0xFFFFFDFF +#define S_000E40_CFRQ_ON_RBB(x) (((x) & 0x1) << 10) +#define G_000E40_CFRQ_ON_RBB(x) (((x) >> 10) & 0x1) +#define C_000E40_CFRQ_ON_RBB 0xFFFFFBFF +#define S_000E40_HIRQ_IN_RTBUF(x) (((x) & 0x1) << 11) +#define G_000E40_HIRQ_IN_RTBUF(x) (((x) >> 11) & 0x1) +#define C_000E40_HIRQ_IN_RTBUF 0xFFFFF7FF +#define S_000E40_CPRQ_IN_RTBUF(x) (((x) & 0x1) << 12) +#define G_000E40_CPRQ_IN_RTBUF(x) (((x) >> 12) & 0x1) +#define C_000E40_CPRQ_IN_RTBUF 0xFFFFEFFF +#define S_000E40_CFRQ_IN_RTBUF(x) (((x) & 0x1) << 13) +#define G_000E40_CFRQ_IN_RTBUF(x) (((x) >> 13) & 0x1) +#define C_000E40_CFRQ_IN_RTBUF 0xFFFFDFFF +#define S_000E40_CF_PIPE_BUSY(x) (((x) & 0x1) << 14) +#define G_000E40_CF_PIPE_BUSY(x) (((x) >> 14) & 0x1) +#define C_000E40_CF_PIPE_BUSY 0xFFFFBFFF +#define S_000E40_ENG_EV_BUSY(x) (((x) & 0x1) << 15) +#define G_000E40_ENG_EV_BUSY(x) (((x) >> 15) & 0x1) +#define C_000E40_ENG_EV_BUSY 0xFFFF7FFF +#define S_000E40_CP_CMDSTRM_BUSY(x) (((x) & 0x1) << 16) +#define G_000E40_CP_CMDSTRM_BUSY(x) (((x) >> 16) & 0x1) +#define C_000E40_CP_CMDSTRM_BUSY 0xFFFEFFFF +#define S_000E40_E2_BUSY(x) (((x) & 0x1) << 17) +#define G_000E40_E2_BUSY(x) (((x) >> 17) & 0x1) +#define C_000E40_E2_BUSY 0xFFFDFFFF +#define S_000E40_RB2D_BUSY(x) (((x) & 0x1) << 18) +#define G_000E40_RB2D_BUSY(x) (((x) >> 18) & 0x1) +#define C_000E40_RB2D_BUSY 0xFFFBFFFF +#define S_000E40_RB3D_BUSY(x) (((x) & 0x1) << 19) +#define G_000E40_RB3D_BUSY(x) (((x) >> 19) & 0x1) +#define C_000E40_RB3D_BUSY 0xFFF7FFFF +#define S_000E40_VAP_BUSY(x) (((x) & 0x1) << 20) +#define G_000E40_VAP_BUSY(x) (((x) >> 20) & 0x1) +#define C_000E40_VAP_BUSY 0xFFEFFFFF +#define S_000E40_RE_BUSY(x) (((x) & 0x1) << 21) +#define G_000E40_RE_BUSY(x) (((x) >> 21) & 0x1) +#define C_000E40_RE_BUSY 0xFFDFFFFF +#define S_000E40_TAM_BUSY(x) (((x) & 0x1) << 22) +#define G_000E40_TAM_BUSY(x) (((x) >> 22) & 0x1) +#define C_000E40_TAM_BUSY 0xFFBFFFFF +#define S_000E40_TDM_BUSY(x) (((x) & 0x1) << 23) +#define G_000E40_TDM_BUSY(x) (((x) >> 23) & 0x1) +#define C_000E40_TDM_BUSY 0xFF7FFFFF +#define S_000E40_PB_BUSY(x) (((x) & 0x1) << 24) +#define G_000E40_PB_BUSY(x) (((x) >> 24) & 0x1) +#define C_000E40_PB_BUSY 0xFEFFFFFF +#define S_000E40_TIM_BUSY(x) (((x) & 0x1) << 25) +#define G_000E40_TIM_BUSY(x) (((x) >> 25) & 0x1) +#define C_000E40_TIM_BUSY 0xFDFFFFFF +#define S_000E40_GA_BUSY(x) (((x) & 0x1) << 26) +#define G_000E40_GA_BUSY(x) (((x) >> 26) & 0x1) +#define C_000E40_GA_BUSY 0xFBFFFFFF +#define S_000E40_CBA2D_BUSY(x) (((x) & 0x1) << 27) +#define G_000E40_CBA2D_BUSY(x) (((x) >> 27) & 0x1) +#define C_000E40_CBA2D_BUSY 0xF7FFFFFF +#define S_000E40_GUI_ACTIVE(x) (((x) & 0x1) << 31) +#define G_000E40_GUI_ACTIVE(x) (((x) >> 31) & 0x1) +#define C_000E40_GUI_ACTIVE 0x7FFFFFFF +#define R_006520_DC_LB_MEMORY_SPLIT 0x006520 +#define S_006520_DC_LB_MEMORY_SPLIT(x) (((x) & 0x3) << 0) +#define G_006520_DC_LB_MEMORY_SPLIT(x) (((x) >> 0) & 0x3) +#define C_006520_DC_LB_MEMORY_SPLIT 0xFFFFFFFC +#define S_006520_DC_LB_MEMORY_SPLIT_MODE(x) (((x) & 0x1) << 2) +#define G_006520_DC_LB_MEMORY_SPLIT_MODE(x) (((x) >> 2) & 0x1) +#define C_006520_DC_LB_MEMORY_SPLIT_MODE 0xFFFFFFFB +#define V_006520_DC_LB_MEMORY_SPLIT_D1HALF_D2HALF 0 +#define V_006520_DC_LB_MEMORY_SPLIT_D1_3Q_D2_1Q 1 +#define V_006520_DC_LB_MEMORY_SPLIT_D1_ONLY 2 +#define V_006520_DC_LB_MEMORY_SPLIT_D1_1Q_D2_3Q 3 +#define S_006520_DC_LB_DISP1_END_ADR(x) (((x) & 0x7FF) << 4) +#define G_006520_DC_LB_DISP1_END_ADR(x) (((x) >> 4) & 0x7FF) +#define C_006520_DC_LB_DISP1_END_ADR 0xFFFF800F +#define R_006548_D1MODE_PRIORITY_A_CNT 0x006548 +#define S_006548_D1MODE_PRIORITY_MARK_A(x) (((x) & 0x7FFF) << 0) +#define G_006548_D1MODE_PRIORITY_MARK_A(x) (((x) >> 0) & 0x7FFF) +#define C_006548_D1MODE_PRIORITY_MARK_A 0xFFFF8000 +#define S_006548_D1MODE_PRIORITY_A_OFF(x) (((x) & 0x1) << 16) +#define G_006548_D1MODE_PRIORITY_A_OFF(x) (((x) >> 16) & 0x1) +#define C_006548_D1MODE_PRIORITY_A_OFF 0xFFFEFFFF +#define S_006548_D1MODE_PRIORITY_A_ALWAYS_ON(x) (((x) & 0x1) << 20) +#define G_006548_D1MODE_PRIORITY_A_ALWAYS_ON(x) (((x) >> 20) & 0x1) +#define C_006548_D1MODE_PRIORITY_A_ALWAYS_ON 0xFFEFFFFF +#define S_006548_D1MODE_PRIORITY_A_FORCE_MASK(x) (((x) & 0x1) << 24) +#define G_006548_D1MODE_PRIORITY_A_FORCE_MASK(x) (((x) >> 24) & 0x1) +#define C_006548_D1MODE_PRIORITY_A_FORCE_MASK 0xFEFFFFFF +#define R_00654C_D1MODE_PRIORITY_B_CNT 0x00654C +#define S_00654C_D1MODE_PRIORITY_MARK_B(x) (((x) & 0x7FFF) << 0) +#define G_00654C_D1MODE_PRIORITY_MARK_B(x) (((x) >> 0) & 0x7FFF) +#define C_00654C_D1MODE_PRIORITY_MARK_B 0xFFFF8000 +#define S_00654C_D1MODE_PRIORITY_B_OFF(x) (((x) & 0x1) << 16) +#define G_00654C_D1MODE_PRIORITY_B_OFF(x) (((x) >> 16) & 0x1) +#define C_00654C_D1MODE_PRIORITY_B_OFF 0xFFFEFFFF +#define S_00654C_D1MODE_PRIORITY_B_ALWAYS_ON(x) (((x) & 0x1) << 20) +#define G_00654C_D1MODE_PRIORITY_B_ALWAYS_ON(x) (((x) >> 20) & 0x1) +#define C_00654C_D1MODE_PRIORITY_B_ALWAYS_ON 0xFFEFFFFF +#define S_00654C_D1MODE_PRIORITY_B_FORCE_MASK(x) (((x) & 0x1) << 24) +#define G_00654C_D1MODE_PRIORITY_B_FORCE_MASK(x) (((x) >> 24) & 0x1) +#define C_00654C_D1MODE_PRIORITY_B_FORCE_MASK 0xFEFFFFFF +#define R_006C9C_DCP_CONTROL 0x006C9C +#define R_006D48_D2MODE_PRIORITY_A_CNT 0x006D48 +#define S_006D48_D2MODE_PRIORITY_MARK_A(x) (((x) & 0x7FFF) << 0) +#define G_006D48_D2MODE_PRIORITY_MARK_A(x) (((x) >> 0) & 0x7FFF) +#define C_006D48_D2MODE_PRIORITY_MARK_A 0xFFFF8000 +#define S_006D48_D2MODE_PRIORITY_A_OFF(x) (((x) & 0x1) << 16) +#define G_006D48_D2MODE_PRIORITY_A_OFF(x) (((x) >> 16) & 0x1) +#define C_006D48_D2MODE_PRIORITY_A_OFF 0xFFFEFFFF +#define S_006D48_D2MODE_PRIORITY_A_ALWAYS_ON(x) (((x) & 0x1) << 20) +#define G_006D48_D2MODE_PRIORITY_A_ALWAYS_ON(x) (((x) >> 20) & 0x1) +#define C_006D48_D2MODE_PRIORITY_A_ALWAYS_ON 0xFFEFFFFF +#define S_006D48_D2MODE_PRIORITY_A_FORCE_MASK(x) (((x) & 0x1) << 24) +#define G_006D48_D2MODE_PRIORITY_A_FORCE_MASK(x) (((x) >> 24) & 0x1) +#define C_006D48_D2MODE_PRIORITY_A_FORCE_MASK 0xFEFFFFFF +#define R_006D4C_D2MODE_PRIORITY_B_CNT 0x006D4C +#define S_006D4C_D2MODE_PRIORITY_MARK_B(x) (((x) & 0x7FFF) << 0) +#define G_006D4C_D2MODE_PRIORITY_MARK_B(x) (((x) >> 0) & 0x7FFF) +#define C_006D4C_D2MODE_PRIORITY_MARK_B 0xFFFF8000 +#define S_006D4C_D2MODE_PRIORITY_B_OFF(x) (((x) & 0x1) << 16) +#define G_006D4C_D2MODE_PRIORITY_B_OFF(x) (((x) >> 16) & 0x1) +#define C_006D4C_D2MODE_PRIORITY_B_OFF 0xFFFEFFFF +#define S_006D4C_D2MODE_PRIORITY_B_ALWAYS_ON(x) (((x) & 0x1) << 20) +#define G_006D4C_D2MODE_PRIORITY_B_ALWAYS_ON(x) (((x) >> 20) & 0x1) +#define C_006D4C_D2MODE_PRIORITY_B_ALWAYS_ON 0xFFEFFFFF +#define S_006D4C_D2MODE_PRIORITY_B_FORCE_MASK(x) (((x) & 0x1) << 24) +#define G_006D4C_D2MODE_PRIORITY_B_FORCE_MASK(x) (((x) >> 24) & 0x1) +#define C_006D4C_D2MODE_PRIORITY_B_FORCE_MASK 0xFEFFFFFF +#define R_006D58_LB_MAX_REQ_OUTSTANDING 0x006D58 +#define S_006D58_LB_D1_MAX_REQ_OUTSTANDING(x) (((x) & 0xF) << 0) +#define G_006D58_LB_D1_MAX_REQ_OUTSTANDING(x) (((x) >> 0) & 0xF) +#define C_006D58_LB_D1_MAX_REQ_OUTSTANDING 0xFFFFFFF0 +#define S_006D58_LB_D2_MAX_REQ_OUTSTANDING(x) (((x) & 0xF) << 16) +#define G_006D58_LB_D2_MAX_REQ_OUTSTANDING(x) (((x) >> 16) & 0xF) +#define C_006D58_LB_D2_MAX_REQ_OUTSTANDING 0xFFF0FFFF + + +#define R_000090_MC_SYSTEM_STATUS 0x000090 +#define S_000090_MC_SYSTEM_IDLE(x) (((x) & 0x1) << 0) +#define G_000090_MC_SYSTEM_IDLE(x) (((x) >> 0) & 0x1) +#define C_000090_MC_SYSTEM_IDLE 0xFFFFFFFE +#define S_000090_MC_SEQUENCER_IDLE(x) (((x) & 0x1) << 1) +#define G_000090_MC_SEQUENCER_IDLE(x) (((x) >> 1) & 0x1) +#define C_000090_MC_SEQUENCER_IDLE 0xFFFFFFFD +#define S_000090_MC_ARBITER_IDLE(x) (((x) & 0x1) << 2) +#define G_000090_MC_ARBITER_IDLE(x) (((x) >> 2) & 0x1) +#define C_000090_MC_ARBITER_IDLE 0xFFFFFFFB +#define S_000090_MC_SELECT_PM(x) (((x) & 0x1) << 3) +#define G_000090_MC_SELECT_PM(x) (((x) >> 3) & 0x1) +#define C_000090_MC_SELECT_PM 0xFFFFFFF7 +#define S_000090_RESERVED4(x) (((x) & 0xF) << 4) +#define G_000090_RESERVED4(x) (((x) >> 4) & 0xF) +#define C_000090_RESERVED4 0xFFFFFF0F +#define S_000090_RESERVED8(x) (((x) & 0xF) << 8) +#define G_000090_RESERVED8(x) (((x) >> 8) & 0xF) +#define C_000090_RESERVED8 0xFFFFF0FF +#define S_000090_RESERVED12(x) (((x) & 0xF) << 12) +#define G_000090_RESERVED12(x) (((x) >> 12) & 0xF) +#define C_000090_RESERVED12 0xFFFF0FFF +#define S_000090_MCA_INIT_EXECUTED(x) (((x) & 0x1) << 16) +#define G_000090_MCA_INIT_EXECUTED(x) (((x) >> 16) & 0x1) +#define C_000090_MCA_INIT_EXECUTED 0xFFFEFFFF +#define S_000090_MCA_IDLE(x) (((x) & 0x1) << 17) +#define G_000090_MCA_IDLE(x) (((x) >> 17) & 0x1) +#define C_000090_MCA_IDLE 0xFFFDFFFF +#define S_000090_MCA_SEQ_IDLE(x) (((x) & 0x1) << 18) +#define G_000090_MCA_SEQ_IDLE(x) (((x) >> 18) & 0x1) +#define C_000090_MCA_SEQ_IDLE 0xFFFBFFFF +#define S_000090_MCA_ARB_IDLE(x) (((x) & 0x1) << 19) +#define G_000090_MCA_ARB_IDLE(x) (((x) >> 19) & 0x1) +#define C_000090_MCA_ARB_IDLE 0xFFF7FFFF +#define S_000090_RESERVED20(x) (((x) & 0xFFF) << 20) +#define G_000090_RESERVED20(x) (((x) >> 20) & 0xFFF) +#define C_000090_RESERVED20 0x000FFFFF +#define R_000100_MCCFG_FB_LOCATION 0x000100 +#define S_000100_MC_FB_START(x) (((x) & 0xFFFF) << 0) +#define G_000100_MC_FB_START(x) (((x) >> 0) & 0xFFFF) +#define C_000100_MC_FB_START 0xFFFF0000 +#define S_000100_MC_FB_TOP(x) (((x) & 0xFFFF) << 16) +#define G_000100_MC_FB_TOP(x) (((x) >> 16) & 0xFFFF) +#define C_000100_MC_FB_TOP 0x0000FFFF +#define R_000104_MC_INIT_MISC_LAT_TIMER 0x000104 +#define S_000104_MC_CPR_INIT_LAT(x) (((x) & 0xF) << 0) +#define G_000104_MC_CPR_INIT_LAT(x) (((x) >> 0) & 0xF) +#define C_000104_MC_CPR_INIT_LAT 0xFFFFFFF0 +#define S_000104_MC_VF_INIT_LAT(x) (((x) & 0xF) << 4) +#define G_000104_MC_VF_INIT_LAT(x) (((x) >> 4) & 0xF) +#define C_000104_MC_VF_INIT_LAT 0xFFFFFF0F +#define S_000104_MC_DISP0R_INIT_LAT(x) (((x) & 0xF) << 8) +#define G_000104_MC_DISP0R_INIT_LAT(x) (((x) >> 8) & 0xF) +#define C_000104_MC_DISP0R_INIT_LAT 0xFFFFF0FF +#define S_000104_MC_DISP1R_INIT_LAT(x) (((x) & 0xF) << 12) +#define G_000104_MC_DISP1R_INIT_LAT(x) (((x) >> 12) & 0xF) +#define C_000104_MC_DISP1R_INIT_LAT 0xFFFF0FFF +#define S_000104_MC_FIXED_INIT_LAT(x) (((x) & 0xF) << 16) +#define G_000104_MC_FIXED_INIT_LAT(x) (((x) >> 16) & 0xF) +#define C_000104_MC_FIXED_INIT_LAT 0xFFF0FFFF +#define S_000104_MC_E2R_INIT_LAT(x) (((x) & 0xF) << 20) +#define G_000104_MC_E2R_INIT_LAT(x) (((x) >> 20) & 0xF) +#define C_000104_MC_E2R_INIT_LAT 0xFF0FFFFF +#define S_000104_SAME_PAGE_PRIO(x) (((x) & 0xF) << 24) +#define G_000104_SAME_PAGE_PRIO(x) (((x) >> 24) & 0xF) +#define C_000104_SAME_PAGE_PRIO 0xF0FFFFFF +#define S_000104_MC_GLOBW_INIT_LAT(x) (((x) & 0xF) << 28) +#define G_000104_MC_GLOBW_INIT_LAT(x) (((x) >> 28) & 0xF) +#define C_000104_MC_GLOBW_INIT_LAT 0x0FFFFFFF + +#endif diff --git a/drivers/gpu/drm/radeon/rv200d.h b/drivers/gpu/drm/radeon/rv200d.h new file mode 100644 index 00000000..c5b39833 --- /dev/null +++ b/drivers/gpu/drm/radeon/rv200d.h @@ -0,0 +1,36 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#ifndef __RV200D_H__ +#define __RV200D_H__ + +#define R_00015C_AGP_BASE_2 0x00015C +#define S_00015C_AGP_BASE_ADDR_2(x) (((x) & 0xF) << 0) +#define G_00015C_AGP_BASE_ADDR_2(x) (((x) >> 0) & 0xF) +#define C_00015C_AGP_BASE_ADDR_2 0xFFFFFFF0 + +#endif diff --git a/drivers/gpu/drm/radeon/rv250d.h b/drivers/gpu/drm/radeon/rv250d.h new file mode 100644 index 00000000..e5a70b06 --- /dev/null +++ b/drivers/gpu/drm/radeon/rv250d.h @@ -0,0 +1,123 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#ifndef __RV250D_H__ +#define __RV250D_H__ + +#define R_00000D_SCLK_CNTL_M6 0x00000D +#define S_00000D_SCLK_SRC_SEL(x) (((x) & 0x7) << 0) +#define G_00000D_SCLK_SRC_SEL(x) (((x) >> 0) & 0x7) +#define C_00000D_SCLK_SRC_SEL 0xFFFFFFF8 +#define S_00000D_CP_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 3) +#define G_00000D_CP_MAX_DYN_STOP_LAT(x) (((x) >> 3) & 0x1) +#define C_00000D_CP_MAX_DYN_STOP_LAT 0xFFFFFFF7 +#define S_00000D_HDP_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 4) +#define G_00000D_HDP_MAX_DYN_STOP_LAT(x) (((x) >> 4) & 0x1) +#define C_00000D_HDP_MAX_DYN_STOP_LAT 0xFFFFFFEF +#define S_00000D_TV_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 5) +#define G_00000D_TV_MAX_DYN_STOP_LAT(x) (((x) >> 5) & 0x1) +#define C_00000D_TV_MAX_DYN_STOP_LAT 0xFFFFFFDF +#define S_00000D_E2_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 6) +#define G_00000D_E2_MAX_DYN_STOP_LAT(x) (((x) >> 6) & 0x1) +#define C_00000D_E2_MAX_DYN_STOP_LAT 0xFFFFFFBF +#define S_00000D_SE_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 7) +#define G_00000D_SE_MAX_DYN_STOP_LAT(x) (((x) >> 7) & 0x1) +#define C_00000D_SE_MAX_DYN_STOP_LAT 0xFFFFFF7F +#define S_00000D_IDCT_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 8) +#define G_00000D_IDCT_MAX_DYN_STOP_LAT(x) (((x) >> 8) & 0x1) +#define C_00000D_IDCT_MAX_DYN_STOP_LAT 0xFFFFFEFF +#define S_00000D_VIP_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 9) +#define G_00000D_VIP_MAX_DYN_STOP_LAT(x) (((x) >> 9) & 0x1) +#define C_00000D_VIP_MAX_DYN_STOP_LAT 0xFFFFFDFF +#define S_00000D_RE_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 10) +#define G_00000D_RE_MAX_DYN_STOP_LAT(x) (((x) >> 10) & 0x1) +#define C_00000D_RE_MAX_DYN_STOP_LAT 0xFFFFFBFF +#define S_00000D_PB_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 11) +#define G_00000D_PB_MAX_DYN_STOP_LAT(x) (((x) >> 11) & 0x1) +#define C_00000D_PB_MAX_DYN_STOP_LAT 0xFFFFF7FF +#define S_00000D_TAM_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 12) +#define G_00000D_TAM_MAX_DYN_STOP_LAT(x) (((x) >> 12) & 0x1) +#define C_00000D_TAM_MAX_DYN_STOP_LAT 0xFFFFEFFF +#define S_00000D_TDM_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 13) +#define G_00000D_TDM_MAX_DYN_STOP_LAT(x) (((x) >> 13) & 0x1) +#define C_00000D_TDM_MAX_DYN_STOP_LAT 0xFFFFDFFF +#define S_00000D_RB_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 14) +#define G_00000D_RB_MAX_DYN_STOP_LAT(x) (((x) >> 14) & 0x1) +#define C_00000D_RB_MAX_DYN_STOP_LAT 0xFFFFBFFF +#define S_00000D_FORCE_DISP2(x) (((x) & 0x1) << 15) +#define G_00000D_FORCE_DISP2(x) (((x) >> 15) & 0x1) +#define C_00000D_FORCE_DISP2 0xFFFF7FFF +#define S_00000D_FORCE_CP(x) (((x) & 0x1) << 16) +#define G_00000D_FORCE_CP(x) (((x) >> 16) & 0x1) +#define C_00000D_FORCE_CP 0xFFFEFFFF +#define S_00000D_FORCE_HDP(x) (((x) & 0x1) << 17) +#define G_00000D_FORCE_HDP(x) (((x) >> 17) & 0x1) +#define C_00000D_FORCE_HDP 0xFFFDFFFF +#define S_00000D_FORCE_DISP1(x) (((x) & 0x1) << 18) +#define G_00000D_FORCE_DISP1(x) (((x) >> 18) & 0x1) +#define C_00000D_FORCE_DISP1 0xFFFBFFFF +#define S_00000D_FORCE_TOP(x) (((x) & 0x1) << 19) +#define G_00000D_FORCE_TOP(x) (((x) >> 19) & 0x1) +#define C_00000D_FORCE_TOP 0xFFF7FFFF +#define S_00000D_FORCE_E2(x) (((x) & 0x1) << 20) +#define G_00000D_FORCE_E2(x) (((x) >> 20) & 0x1) +#define C_00000D_FORCE_E2 0xFFEFFFFF +#define S_00000D_FORCE_SE(x) (((x) & 0x1) << 21) +#define G_00000D_FORCE_SE(x) (((x) >> 21) & 0x1) +#define C_00000D_FORCE_SE 0xFFDFFFFF +#define S_00000D_FORCE_IDCT(x) (((x) & 0x1) << 22) +#define G_00000D_FORCE_IDCT(x) (((x) >> 22) & 0x1) +#define C_00000D_FORCE_IDCT 0xFFBFFFFF +#define S_00000D_FORCE_VIP(x) (((x) & 0x1) << 23) +#define G_00000D_FORCE_VIP(x) (((x) >> 23) & 0x1) +#define C_00000D_FORCE_VIP 0xFF7FFFFF +#define S_00000D_FORCE_RE(x) (((x) & 0x1) << 24) +#define G_00000D_FORCE_RE(x) (((x) >> 24) & 0x1) +#define C_00000D_FORCE_RE 0xFEFFFFFF +#define S_00000D_FORCE_PB(x) (((x) & 0x1) << 25) +#define G_00000D_FORCE_PB(x) (((x) >> 25) & 0x1) +#define C_00000D_FORCE_PB 0xFDFFFFFF +#define S_00000D_FORCE_TAM(x) (((x) & 0x1) << 26) +#define G_00000D_FORCE_TAM(x) (((x) >> 26) & 0x1) +#define C_00000D_FORCE_TAM 0xFBFFFFFF +#define S_00000D_FORCE_TDM(x) (((x) & 0x1) << 27) +#define G_00000D_FORCE_TDM(x) (((x) >> 27) & 0x1) +#define C_00000D_FORCE_TDM 0xF7FFFFFF +#define S_00000D_FORCE_RB(x) (((x) & 0x1) << 28) +#define G_00000D_FORCE_RB(x) (((x) >> 28) & 0x1) +#define C_00000D_FORCE_RB 0xEFFFFFFF +#define S_00000D_FORCE_TV_SCLK(x) (((x) & 0x1) << 29) +#define G_00000D_FORCE_TV_SCLK(x) (((x) >> 29) & 0x1) +#define C_00000D_FORCE_TV_SCLK 0xDFFFFFFF +#define S_00000D_FORCE_SUBPIC(x) (((x) & 0x1) << 30) +#define G_00000D_FORCE_SUBPIC(x) (((x) >> 30) & 0x1) +#define C_00000D_FORCE_SUBPIC 0xBFFFFFFF +#define S_00000D_FORCE_OV0(x) (((x) & 0x1) << 31) +#define G_00000D_FORCE_OV0(x) (((x) >> 31) & 0x1) +#define C_00000D_FORCE_OV0 0x7FFFFFFF + +#endif diff --git a/drivers/gpu/drm/radeon/rv350d.h b/drivers/gpu/drm/radeon/rv350d.h new file mode 100644 index 00000000..c75c5ed9 --- /dev/null +++ b/drivers/gpu/drm/radeon/rv350d.h @@ -0,0 +1,52 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#ifndef __RV350D_H__ +#define __RV350D_H__ + +/* RV350, RV380 registers */ +/* #define R_00000D_SCLK_CNTL 0x00000D */ +#define S_00000D_FORCE_VAP(x) (((x) & 0x1) << 21) +#define G_00000D_FORCE_VAP(x) (((x) >> 21) & 0x1) +#define C_00000D_FORCE_VAP 0xFFDFFFFF +#define S_00000D_FORCE_SR(x) (((x) & 0x1) << 25) +#define G_00000D_FORCE_SR(x) (((x) >> 25) & 0x1) +#define C_00000D_FORCE_SR 0xFDFFFFFF +#define S_00000D_FORCE_PX(x) (((x) & 0x1) << 26) +#define G_00000D_FORCE_PX(x) (((x) >> 26) & 0x1) +#define C_00000D_FORCE_PX 0xFBFFFFFF +#define S_00000D_FORCE_TX(x) (((x) & 0x1) << 27) +#define G_00000D_FORCE_TX(x) (((x) >> 27) & 0x1) +#define C_00000D_FORCE_TX 0xF7FFFFFF +#define S_00000D_FORCE_US(x) (((x) & 0x1) << 28) +#define G_00000D_FORCE_US(x) (((x) >> 28) & 0x1) +#define C_00000D_FORCE_US 0xEFFFFFFF +#define S_00000D_FORCE_SU(x) (((x) & 0x1) << 30) +#define G_00000D_FORCE_SU(x) (((x) >> 30) & 0x1) +#define C_00000D_FORCE_SU 0xBFFFFFFF + +#endif diff --git a/drivers/gpu/drm/radeon/rv515.c b/drivers/gpu/drm/radeon/rv515.c new file mode 100644 index 00000000..6613ee9e --- /dev/null +++ b/drivers/gpu/drm/radeon/rv515.c @@ -0,0 +1,1104 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#include +#include +#include "drmP.h" +#include "rv515d.h" +#include "radeon.h" +#include "radeon_asic.h" +#include "atom.h" +#include "rv515_reg_safe.h" + +/* This files gather functions specifics to: rv515 */ +int rv515_debugfs_pipes_info_init(struct radeon_device *rdev); +int rv515_debugfs_ga_info_init(struct radeon_device *rdev); +void rv515_gpu_init(struct radeon_device *rdev); +int rv515_mc_wait_for_idle(struct radeon_device *rdev); + +void rv515_debugfs(struct radeon_device *rdev) +{ + if (r100_debugfs_rbbm_init(rdev)) { + DRM_ERROR("Failed to register debugfs file for RBBM !\n"); + } + if (rv515_debugfs_pipes_info_init(rdev)) { + DRM_ERROR("Failed to register debugfs file for pipes !\n"); + } + if (rv515_debugfs_ga_info_init(rdev)) { + DRM_ERROR("Failed to register debugfs file for pipes !\n"); + } +} + +void rv515_ring_start(struct radeon_device *rdev) +{ + int r; + + r = radeon_ring_lock(rdev, 64); + if (r) { + return; + } + radeon_ring_write(rdev, PACKET0(ISYNC_CNTL, 0)); + radeon_ring_write(rdev, + ISYNC_ANY2D_IDLE3D | + ISYNC_ANY3D_IDLE2D | + ISYNC_WAIT_IDLEGUI | + ISYNC_CPSCRATCH_IDLEGUI); + radeon_ring_write(rdev, PACKET0(WAIT_UNTIL, 0)); + radeon_ring_write(rdev, WAIT_2D_IDLECLEAN | WAIT_3D_IDLECLEAN); + radeon_ring_write(rdev, PACKET0(R300_DST_PIPE_CONFIG, 0)); + radeon_ring_write(rdev, R300_PIPE_AUTO_CONFIG); + radeon_ring_write(rdev, PACKET0(GB_SELECT, 0)); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, PACKET0(GB_ENABLE, 0)); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, PACKET0(R500_SU_REG_DEST, 0)); + radeon_ring_write(rdev, (1 << rdev->num_gb_pipes) - 1); + radeon_ring_write(rdev, PACKET0(VAP_INDEX_OFFSET, 0)); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, PACKET0(RB3D_DSTCACHE_CTLSTAT, 0)); + radeon_ring_write(rdev, RB3D_DC_FLUSH | RB3D_DC_FREE); + radeon_ring_write(rdev, PACKET0(ZB_ZCACHE_CTLSTAT, 0)); + radeon_ring_write(rdev, ZC_FLUSH | ZC_FREE); + radeon_ring_write(rdev, PACKET0(WAIT_UNTIL, 0)); + radeon_ring_write(rdev, WAIT_2D_IDLECLEAN | WAIT_3D_IDLECLEAN); + radeon_ring_write(rdev, PACKET0(GB_AA_CONFIG, 0)); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, PACKET0(RB3D_DSTCACHE_CTLSTAT, 0)); + radeon_ring_write(rdev, RB3D_DC_FLUSH | RB3D_DC_FREE); + radeon_ring_write(rdev, PACKET0(ZB_ZCACHE_CTLSTAT, 0)); + radeon_ring_write(rdev, ZC_FLUSH | ZC_FREE); + radeon_ring_write(rdev, PACKET0(GB_MSPOS0, 0)); + radeon_ring_write(rdev, + ((6 << MS_X0_SHIFT) | + (6 << MS_Y0_SHIFT) | + (6 << MS_X1_SHIFT) | + (6 << MS_Y1_SHIFT) | + (6 << MS_X2_SHIFT) | + (6 << MS_Y2_SHIFT) | + (6 << MSBD0_Y_SHIFT) | + (6 << MSBD0_X_SHIFT))); + radeon_ring_write(rdev, PACKET0(GB_MSPOS1, 0)); + radeon_ring_write(rdev, + ((6 << MS_X3_SHIFT) | + (6 << MS_Y3_SHIFT) | + (6 << MS_X4_SHIFT) | + (6 << MS_Y4_SHIFT) | + (6 << MS_X5_SHIFT) | + (6 << MS_Y5_SHIFT) | + (6 << MSBD1_SHIFT))); + radeon_ring_write(rdev, PACKET0(GA_ENHANCE, 0)); + radeon_ring_write(rdev, GA_DEADLOCK_CNTL | GA_FASTSYNC_CNTL); + radeon_ring_write(rdev, PACKET0(GA_POLY_MODE, 0)); + radeon_ring_write(rdev, FRONT_PTYPE_TRIANGE | BACK_PTYPE_TRIANGE); + radeon_ring_write(rdev, PACKET0(GA_ROUND_MODE, 0)); + radeon_ring_write(rdev, GEOMETRY_ROUND_NEAREST | COLOR_ROUND_NEAREST); + radeon_ring_write(rdev, PACKET0(0x20C8, 0)); + radeon_ring_write(rdev, 0); + radeon_ring_unlock_commit(rdev); +} + +int rv515_mc_wait_for_idle(struct radeon_device *rdev) +{ + unsigned i; + uint32_t tmp; + + for (i = 0; i < rdev->usec_timeout; i++) { + /* read MC_STATUS */ + tmp = RREG32_MC(MC_STATUS); + if (tmp & MC_STATUS_IDLE) { + return 0; + } + DRM_UDELAY(1); + } + return -1; +} + +void rv515_vga_render_disable(struct radeon_device *rdev) +{ + WREG32(R_000300_VGA_RENDER_CONTROL, + RREG32(R_000300_VGA_RENDER_CONTROL) & C_000300_VGA_VSTATUS_CNTL); +} + +void rv515_gpu_init(struct radeon_device *rdev) +{ + unsigned pipe_select_current, gb_pipe_select, tmp; + + if (r100_gui_wait_for_idle(rdev)) { + printk(KERN_WARNING "Failed to wait GUI idle while " + "reseting GPU. Bad things might happen.\n"); + } + rv515_vga_render_disable(rdev); + r420_pipes_init(rdev); + gb_pipe_select = RREG32(R400_GB_PIPE_SELECT); + tmp = RREG32(R300_DST_PIPE_CONFIG); + pipe_select_current = (tmp >> 2) & 3; + tmp = (1 << pipe_select_current) | + (((gb_pipe_select >> 8) & 0xF) << 4); + WREG32_PLL(0x000D, tmp); + if (r100_gui_wait_for_idle(rdev)) { + printk(KERN_WARNING "Failed to wait GUI idle while " + "reseting GPU. Bad things might happen.\n"); + } + if (rv515_mc_wait_for_idle(rdev)) { + printk(KERN_WARNING "Failed to wait MC idle while " + "programming pipes. Bad things might happen.\n"); + } +} + +static void rv515_vram_get_type(struct radeon_device *rdev) +{ + uint32_t tmp; + + rdev->mc.vram_width = 128; + rdev->mc.vram_is_ddr = true; + tmp = RREG32_MC(RV515_MC_CNTL) & MEM_NUM_CHANNELS_MASK; + switch (tmp) { + case 0: + rdev->mc.vram_width = 64; + break; + case 1: + rdev->mc.vram_width = 128; + break; + default: + rdev->mc.vram_width = 128; + break; + } +} + +void rv515_mc_init(struct radeon_device *rdev) +{ + + rv515_vram_get_type(rdev); + r100_vram_init_sizes(rdev); + radeon_vram_location(rdev, &rdev->mc, 0); + rdev->mc.gtt_base_align = 0; + if (!(rdev->flags & RADEON_IS_AGP)) + radeon_gtt_location(rdev, &rdev->mc); + radeon_update_bandwidth_info(rdev); +} + +uint32_t rv515_mc_rreg(struct radeon_device *rdev, uint32_t reg) +{ + uint32_t r; + + WREG32(MC_IND_INDEX, 0x7f0000 | (reg & 0xffff)); + r = RREG32(MC_IND_DATA); + WREG32(MC_IND_INDEX, 0); + return r; +} + +void rv515_mc_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v) +{ + WREG32(MC_IND_INDEX, 0xff0000 | ((reg) & 0xffff)); + WREG32(MC_IND_DATA, (v)); + WREG32(MC_IND_INDEX, 0); +} + +#if defined(CONFIG_DEBUG_FS) +static int rv515_debugfs_pipes_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + struct radeon_device *rdev = dev->dev_private; + uint32_t tmp; + + tmp = RREG32(GB_PIPE_SELECT); + seq_printf(m, "GB_PIPE_SELECT 0x%08x\n", tmp); + tmp = RREG32(SU_REG_DEST); + seq_printf(m, "SU_REG_DEST 0x%08x\n", tmp); + tmp = RREG32(GB_TILE_CONFIG); + seq_printf(m, "GB_TILE_CONFIG 0x%08x\n", tmp); + tmp = RREG32(DST_PIPE_CONFIG); + seq_printf(m, "DST_PIPE_CONFIG 0x%08x\n", tmp); + return 0; +} + +static int rv515_debugfs_ga_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + struct radeon_device *rdev = dev->dev_private; + uint32_t tmp; + + tmp = RREG32(0x2140); + seq_printf(m, "VAP_CNTL_STATUS 0x%08x\n", tmp); + radeon_asic_reset(rdev); + tmp = RREG32(0x425C); + seq_printf(m, "GA_IDLE 0x%08x\n", tmp); + return 0; +} + +static struct drm_info_list rv515_pipes_info_list[] = { + {"rv515_pipes_info", rv515_debugfs_pipes_info, 0, NULL}, +}; + +static struct drm_info_list rv515_ga_info_list[] = { + {"rv515_ga_info", rv515_debugfs_ga_info, 0, NULL}, +}; +#endif + +int rv515_debugfs_pipes_info_init(struct radeon_device *rdev) +{ +#if defined(CONFIG_DEBUG_FS) + return radeon_debugfs_add_files(rdev, rv515_pipes_info_list, 1); +#else + return 0; +#endif +} + +int rv515_debugfs_ga_info_init(struct radeon_device *rdev) +{ +#if defined(CONFIG_DEBUG_FS) + return radeon_debugfs_add_files(rdev, rv515_ga_info_list, 1); +#else + return 0; +#endif +} + +void rv515_mc_stop(struct radeon_device *rdev, struct rv515_mc_save *save) +{ + save->d1vga_control = RREG32(R_000330_D1VGA_CONTROL); + save->d2vga_control = RREG32(R_000338_D2VGA_CONTROL); + save->vga_render_control = RREG32(R_000300_VGA_RENDER_CONTROL); + save->vga_hdp_control = RREG32(R_000328_VGA_HDP_CONTROL); + save->d1crtc_control = RREG32(R_006080_D1CRTC_CONTROL); + save->d2crtc_control = RREG32(R_006880_D2CRTC_CONTROL); + + /* Stop all video */ + WREG32(R_0068E8_D2CRTC_UPDATE_LOCK, 0); + WREG32(R_000300_VGA_RENDER_CONTROL, 0); + WREG32(R_0060E8_D1CRTC_UPDATE_LOCK, 1); + WREG32(R_0068E8_D2CRTC_UPDATE_LOCK, 1); + WREG32(R_006080_D1CRTC_CONTROL, 0); + WREG32(R_006880_D2CRTC_CONTROL, 0); + WREG32(R_0060E8_D1CRTC_UPDATE_LOCK, 0); + WREG32(R_0068E8_D2CRTC_UPDATE_LOCK, 0); + WREG32(R_000330_D1VGA_CONTROL, 0); + WREG32(R_000338_D2VGA_CONTROL, 0); +} + +void rv515_mc_resume(struct radeon_device *rdev, struct rv515_mc_save *save) +{ + WREG32(R_006110_D1GRPH_PRIMARY_SURFACE_ADDRESS, rdev->mc.vram_start); + WREG32(R_006118_D1GRPH_SECONDARY_SURFACE_ADDRESS, rdev->mc.vram_start); + WREG32(R_006910_D2GRPH_PRIMARY_SURFACE_ADDRESS, rdev->mc.vram_start); + WREG32(R_006918_D2GRPH_SECONDARY_SURFACE_ADDRESS, rdev->mc.vram_start); + WREG32(R_000310_VGA_MEMORY_BASE_ADDRESS, rdev->mc.vram_start); + /* Unlock host access */ + WREG32(R_000328_VGA_HDP_CONTROL, save->vga_hdp_control); + mdelay(1); + /* Restore video state */ + WREG32(R_000330_D1VGA_CONTROL, save->d1vga_control); + WREG32(R_000338_D2VGA_CONTROL, save->d2vga_control); + WREG32(R_0060E8_D1CRTC_UPDATE_LOCK, 1); + WREG32(R_0068E8_D2CRTC_UPDATE_LOCK, 1); + WREG32(R_006080_D1CRTC_CONTROL, save->d1crtc_control); + WREG32(R_006880_D2CRTC_CONTROL, save->d2crtc_control); + WREG32(R_0060E8_D1CRTC_UPDATE_LOCK, 0); + WREG32(R_0068E8_D2CRTC_UPDATE_LOCK, 0); + WREG32(R_000300_VGA_RENDER_CONTROL, save->vga_render_control); +} + +void rv515_mc_program(struct radeon_device *rdev) +{ + struct rv515_mc_save save; + + /* Stops all mc clients */ + rv515_mc_stop(rdev, &save); + + /* Wait for mc idle */ + if (rv515_mc_wait_for_idle(rdev)) + dev_warn(rdev->dev, "Wait MC idle timeout before updating MC.\n"); + /* Write VRAM size in case we are limiting it */ + WREG32(R_0000F8_CONFIG_MEMSIZE, rdev->mc.real_vram_size); + /* Program MC, should be a 32bits limited address space */ + WREG32_MC(R_000001_MC_FB_LOCATION, + S_000001_MC_FB_START(rdev->mc.vram_start >> 16) | + S_000001_MC_FB_TOP(rdev->mc.vram_end >> 16)); + WREG32(R_000134_HDP_FB_LOCATION, + S_000134_HDP_FB_START(rdev->mc.vram_start >> 16)); + if (rdev->flags & RADEON_IS_AGP) { + WREG32_MC(R_000002_MC_AGP_LOCATION, + S_000002_MC_AGP_START(rdev->mc.gtt_start >> 16) | + S_000002_MC_AGP_TOP(rdev->mc.gtt_end >> 16)); + WREG32_MC(R_000003_MC_AGP_BASE, lower_32_bits(rdev->mc.agp_base)); + WREG32_MC(R_000004_MC_AGP_BASE_2, + S_000004_AGP_BASE_ADDR_2(upper_32_bits(rdev->mc.agp_base))); + } else { + WREG32_MC(R_000002_MC_AGP_LOCATION, 0xFFFFFFFF); + WREG32_MC(R_000003_MC_AGP_BASE, 0); + WREG32_MC(R_000004_MC_AGP_BASE_2, 0); + } + + rv515_mc_resume(rdev, &save); +} + +void rv515_clock_startup(struct radeon_device *rdev) +{ + if (radeon_dynclks != -1 && radeon_dynclks) + radeon_atom_set_clock_gating(rdev, 1); + /* We need to force on some of the block */ + WREG32_PLL(R_00000F_CP_DYN_CNTL, + RREG32_PLL(R_00000F_CP_DYN_CNTL) | S_00000F_CP_FORCEON(1)); + WREG32_PLL(R_000011_E2_DYN_CNTL, + RREG32_PLL(R_000011_E2_DYN_CNTL) | S_000011_E2_FORCEON(1)); + WREG32_PLL(R_000013_IDCT_DYN_CNTL, + RREG32_PLL(R_000013_IDCT_DYN_CNTL) | S_000013_IDCT_FORCEON(1)); +} + +static int rv515_startup(struct radeon_device *rdev) +{ + int r; + + rv515_mc_program(rdev); + /* Resume clock */ + rv515_clock_startup(rdev); + /* Initialize GPU configuration (# pipes, ...) */ + rv515_gpu_init(rdev); + /* Initialize GART (initialize after TTM so we can allocate + * memory through TTM but finalize after TTM) */ + if (rdev->flags & RADEON_IS_PCIE) { + r = rv370_pcie_gart_enable(rdev); + if (r) + return r; + } + + /* allocate wb buffer */ + r = radeon_wb_init(rdev); + if (r) + return r; + + /* Enable IRQ */ + rs600_irq_set(rdev); + rdev->config.r300.hdp_cntl = RREG32(RADEON_HOST_PATH_CNTL); + /* 1M ring buffer */ + r = r100_cp_init(rdev, 1024 * 1024); + if (r) { + dev_err(rdev->dev, "failed initializing CP (%d).\n", r); + return r; + } + r = r100_ib_init(rdev); + if (r) { + dev_err(rdev->dev, "failed initializing IB (%d).\n", r); + return r; + } + return 0; +} + +int rv515_resume(struct radeon_device *rdev) +{ + /* Make sur GART are not working */ + if (rdev->flags & RADEON_IS_PCIE) + rv370_pcie_gart_disable(rdev); + /* Resume clock before doing reset */ + rv515_clock_startup(rdev); + /* Reset gpu before posting otherwise ATOM will enter infinite loop */ + if (radeon_asic_reset(rdev)) { + dev_warn(rdev->dev, "GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n", + RREG32(R_000E40_RBBM_STATUS), + RREG32(R_0007C0_CP_STAT)); + } + /* post */ + atom_asic_init(rdev->mode_info.atom_context); + /* Resume clock after posting */ + rv515_clock_startup(rdev); + /* Initialize surface registers */ + radeon_surface_init(rdev); + return rv515_startup(rdev); +} + +int rv515_suspend(struct radeon_device *rdev) +{ + r100_cp_disable(rdev); + radeon_wb_disable(rdev); + rs600_irq_disable(rdev); + if (rdev->flags & RADEON_IS_PCIE) + rv370_pcie_gart_disable(rdev); + return 0; +} + +void rv515_set_safe_registers(struct radeon_device *rdev) +{ + rdev->config.r300.reg_safe_bm = rv515_reg_safe_bm; + rdev->config.r300.reg_safe_bm_size = ARRAY_SIZE(rv515_reg_safe_bm); +} + +void rv515_fini(struct radeon_device *rdev) +{ + r100_cp_fini(rdev); + radeon_wb_fini(rdev); + r100_ib_fini(rdev); + radeon_gem_fini(rdev); + rv370_pcie_gart_fini(rdev); + radeon_agp_fini(rdev); + radeon_irq_kms_fini(rdev); + radeon_fence_driver_fini(rdev); + radeon_bo_fini(rdev); + radeon_atombios_fini(rdev); + kfree(rdev->bios); + rdev->bios = NULL; +} + +int rv515_init(struct radeon_device *rdev) +{ + int r; + + /* Initialize scratch registers */ + radeon_scratch_init(rdev); + /* Initialize surface registers */ + radeon_surface_init(rdev); + /* TODO: disable VGA need to use VGA request */ + /* restore some register to sane defaults */ + r100_restore_sanity(rdev); + /* BIOS*/ + if (!radeon_get_bios(rdev)) { + if (ASIC_IS_AVIVO(rdev)) + return -EINVAL; + } + if (rdev->is_atom_bios) { + r = radeon_atombios_init(rdev); + if (r) + return r; + } else { + dev_err(rdev->dev, "Expecting atombios for RV515 GPU\n"); + return -EINVAL; + } + /* Reset gpu before posting otherwise ATOM will enter infinite loop */ + if (radeon_asic_reset(rdev)) { + dev_warn(rdev->dev, + "GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n", + RREG32(R_000E40_RBBM_STATUS), + RREG32(R_0007C0_CP_STAT)); + } + /* check if cards are posted or not */ + if (radeon_boot_test_post_card(rdev) == false) + return -EINVAL; + /* Initialize clocks */ + radeon_get_clock_info(rdev->ddev); + /* initialize AGP */ + if (rdev->flags & RADEON_IS_AGP) { + r = radeon_agp_init(rdev); + if (r) { + radeon_agp_disable(rdev); + } + } + /* initialize memory controller */ + rv515_mc_init(rdev); + rv515_debugfs(rdev); + /* Fence driver */ + r = radeon_fence_driver_init(rdev); + if (r) + return r; + r = radeon_irq_kms_init(rdev); + if (r) + return r; + /* Memory manager */ + r = radeon_bo_init(rdev); + if (r) + return r; + r = rv370_pcie_gart_init(rdev); + if (r) + return r; + rv515_set_safe_registers(rdev); + rdev->accel_working = true; + r = rv515_startup(rdev); + if (r) { + /* Somethings want wront with the accel init stop accel */ + dev_err(rdev->dev, "Disabling GPU acceleration\n"); + r100_cp_fini(rdev); + radeon_wb_fini(rdev); + r100_ib_fini(rdev); + radeon_irq_kms_fini(rdev); + rv370_pcie_gart_fini(rdev); + radeon_agp_fini(rdev); + rdev->accel_working = false; + } + return 0; +} + +void atom_rv515_force_tv_scaler(struct radeon_device *rdev, struct radeon_crtc *crtc) +{ + int index_reg = 0x6578 + crtc->crtc_offset; + int data_reg = 0x657c + crtc->crtc_offset; + + WREG32(0x659C + crtc->crtc_offset, 0x0); + WREG32(0x6594 + crtc->crtc_offset, 0x705); + WREG32(0x65A4 + crtc->crtc_offset, 0x10001); + WREG32(0x65D8 + crtc->crtc_offset, 0x0); + WREG32(0x65B0 + crtc->crtc_offset, 0x0); + WREG32(0x65C0 + crtc->crtc_offset, 0x0); + WREG32(0x65D4 + crtc->crtc_offset, 0x0); + WREG32(index_reg, 0x0); + WREG32(data_reg, 0x841880A8); + WREG32(index_reg, 0x1); + WREG32(data_reg, 0x84208680); + WREG32(index_reg, 0x2); + WREG32(data_reg, 0xBFF880B0); + WREG32(index_reg, 0x100); + WREG32(data_reg, 0x83D88088); + WREG32(index_reg, 0x101); + WREG32(data_reg, 0x84608680); + WREG32(index_reg, 0x102); + WREG32(data_reg, 0xBFF080D0); + WREG32(index_reg, 0x200); + WREG32(data_reg, 0x83988068); + WREG32(index_reg, 0x201); + WREG32(data_reg, 0x84A08680); + WREG32(index_reg, 0x202); + WREG32(data_reg, 0xBFF080F8); + WREG32(index_reg, 0x300); + WREG32(data_reg, 0x83588058); + WREG32(index_reg, 0x301); + WREG32(data_reg, 0x84E08660); + WREG32(index_reg, 0x302); + WREG32(data_reg, 0xBFF88120); + WREG32(index_reg, 0x400); + WREG32(data_reg, 0x83188040); + WREG32(index_reg, 0x401); + WREG32(data_reg, 0x85008660); + WREG32(index_reg, 0x402); + WREG32(data_reg, 0xBFF88150); + WREG32(index_reg, 0x500); + WREG32(data_reg, 0x82D88030); + WREG32(index_reg, 0x501); + WREG32(data_reg, 0x85408640); + WREG32(index_reg, 0x502); + WREG32(data_reg, 0xBFF88180); + WREG32(index_reg, 0x600); + WREG32(data_reg, 0x82A08018); + WREG32(index_reg, 0x601); + WREG32(data_reg, 0x85808620); + WREG32(index_reg, 0x602); + WREG32(data_reg, 0xBFF081B8); + WREG32(index_reg, 0x700); + WREG32(data_reg, 0x82608010); + WREG32(index_reg, 0x701); + WREG32(data_reg, 0x85A08600); + WREG32(index_reg, 0x702); + WREG32(data_reg, 0x800081F0); + WREG32(index_reg, 0x800); + WREG32(data_reg, 0x8228BFF8); + WREG32(index_reg, 0x801); + WREG32(data_reg, 0x85E085E0); + WREG32(index_reg, 0x802); + WREG32(data_reg, 0xBFF88228); + WREG32(index_reg, 0x10000); + WREG32(data_reg, 0x82A8BF00); + WREG32(index_reg, 0x10001); + WREG32(data_reg, 0x82A08CC0); + WREG32(index_reg, 0x10002); + WREG32(data_reg, 0x8008BEF8); + WREG32(index_reg, 0x10100); + WREG32(data_reg, 0x81F0BF28); + WREG32(index_reg, 0x10101); + WREG32(data_reg, 0x83608CA0); + WREG32(index_reg, 0x10102); + WREG32(data_reg, 0x8018BED0); + WREG32(index_reg, 0x10200); + WREG32(data_reg, 0x8148BF38); + WREG32(index_reg, 0x10201); + WREG32(data_reg, 0x84408C80); + WREG32(index_reg, 0x10202); + WREG32(data_reg, 0x8008BEB8); + WREG32(index_reg, 0x10300); + WREG32(data_reg, 0x80B0BF78); + WREG32(index_reg, 0x10301); + WREG32(data_reg, 0x85008C20); + WREG32(index_reg, 0x10302); + WREG32(data_reg, 0x8020BEA0); + WREG32(index_reg, 0x10400); + WREG32(data_reg, 0x8028BF90); + WREG32(index_reg, 0x10401); + WREG32(data_reg, 0x85E08BC0); + WREG32(index_reg, 0x10402); + WREG32(data_reg, 0x8018BE90); + WREG32(index_reg, 0x10500); + WREG32(data_reg, 0xBFB8BFB0); + WREG32(index_reg, 0x10501); + WREG32(data_reg, 0x86C08B40); + WREG32(index_reg, 0x10502); + WREG32(data_reg, 0x8010BE90); + WREG32(index_reg, 0x10600); + WREG32(data_reg, 0xBF58BFC8); + WREG32(index_reg, 0x10601); + WREG32(data_reg, 0x87A08AA0); + WREG32(index_reg, 0x10602); + WREG32(data_reg, 0x8010BE98); + WREG32(index_reg, 0x10700); + WREG32(data_reg, 0xBF10BFF0); + WREG32(index_reg, 0x10701); + WREG32(data_reg, 0x886089E0); + WREG32(index_reg, 0x10702); + WREG32(data_reg, 0x8018BEB0); + WREG32(index_reg, 0x10800); + WREG32(data_reg, 0xBED8BFE8); + WREG32(index_reg, 0x10801); + WREG32(data_reg, 0x89408940); + WREG32(index_reg, 0x10802); + WREG32(data_reg, 0xBFE8BED8); + WREG32(index_reg, 0x20000); + WREG32(data_reg, 0x80008000); + WREG32(index_reg, 0x20001); + WREG32(data_reg, 0x90008000); + WREG32(index_reg, 0x20002); + WREG32(data_reg, 0x80008000); + WREG32(index_reg, 0x20003); + WREG32(data_reg, 0x80008000); + WREG32(index_reg, 0x20100); + WREG32(data_reg, 0x80108000); + WREG32(index_reg, 0x20101); + WREG32(data_reg, 0x8FE0BF70); + WREG32(index_reg, 0x20102); + WREG32(data_reg, 0xBFE880C0); + WREG32(index_reg, 0x20103); + WREG32(data_reg, 0x80008000); + WREG32(index_reg, 0x20200); + WREG32(data_reg, 0x8018BFF8); + WREG32(index_reg, 0x20201); + WREG32(data_reg, 0x8F80BF08); + WREG32(index_reg, 0x20202); + WREG32(data_reg, 0xBFD081A0); + WREG32(index_reg, 0x20203); + WREG32(data_reg, 0xBFF88000); + WREG32(index_reg, 0x20300); + WREG32(data_reg, 0x80188000); + WREG32(index_reg, 0x20301); + WREG32(data_reg, 0x8EE0BEC0); + WREG32(index_reg, 0x20302); + WREG32(data_reg, 0xBFB082A0); + WREG32(index_reg, 0x20303); + WREG32(data_reg, 0x80008000); + WREG32(index_reg, 0x20400); + WREG32(data_reg, 0x80188000); + WREG32(index_reg, 0x20401); + WREG32(data_reg, 0x8E00BEA0); + WREG32(index_reg, 0x20402); + WREG32(data_reg, 0xBF8883C0); + WREG32(index_reg, 0x20403); + WREG32(data_reg, 0x80008000); + WREG32(index_reg, 0x20500); + WREG32(data_reg, 0x80188000); + WREG32(index_reg, 0x20501); + WREG32(data_reg, 0x8D00BE90); + WREG32(index_reg, 0x20502); + WREG32(data_reg, 0xBF588500); + WREG32(index_reg, 0x20503); + WREG32(data_reg, 0x80008008); + WREG32(index_reg, 0x20600); + WREG32(data_reg, 0x80188000); + WREG32(index_reg, 0x20601); + WREG32(data_reg, 0x8BC0BE98); + WREG32(index_reg, 0x20602); + WREG32(data_reg, 0xBF308660); + WREG32(index_reg, 0x20603); + WREG32(data_reg, 0x80008008); + WREG32(index_reg, 0x20700); + WREG32(data_reg, 0x80108000); + WREG32(index_reg, 0x20701); + WREG32(data_reg, 0x8A80BEB0); + WREG32(index_reg, 0x20702); + WREG32(data_reg, 0xBF0087C0); + WREG32(index_reg, 0x20703); + WREG32(data_reg, 0x80008008); + WREG32(index_reg, 0x20800); + WREG32(data_reg, 0x80108000); + WREG32(index_reg, 0x20801); + WREG32(data_reg, 0x8920BED0); + WREG32(index_reg, 0x20802); + WREG32(data_reg, 0xBED08920); + WREG32(index_reg, 0x20803); + WREG32(data_reg, 0x80008010); + WREG32(index_reg, 0x30000); + WREG32(data_reg, 0x90008000); + WREG32(index_reg, 0x30001); + WREG32(data_reg, 0x80008000); + WREG32(index_reg, 0x30100); + WREG32(data_reg, 0x8FE0BF90); + WREG32(index_reg, 0x30101); + WREG32(data_reg, 0xBFF880A0); + WREG32(index_reg, 0x30200); + WREG32(data_reg, 0x8F60BF40); + WREG32(index_reg, 0x30201); + WREG32(data_reg, 0xBFE88180); + WREG32(index_reg, 0x30300); + WREG32(data_reg, 0x8EC0BF00); + WREG32(index_reg, 0x30301); + WREG32(data_reg, 0xBFC88280); + WREG32(index_reg, 0x30400); + WREG32(data_reg, 0x8DE0BEE0); + WREG32(index_reg, 0x30401); + WREG32(data_reg, 0xBFA083A0); + WREG32(index_reg, 0x30500); + WREG32(data_reg, 0x8CE0BED0); + WREG32(index_reg, 0x30501); + WREG32(data_reg, 0xBF7884E0); + WREG32(index_reg, 0x30600); + WREG32(data_reg, 0x8BA0BED8); + WREG32(index_reg, 0x30601); + WREG32(data_reg, 0xBF508640); + WREG32(index_reg, 0x30700); + WREG32(data_reg, 0x8A60BEE8); + WREG32(index_reg, 0x30701); + WREG32(data_reg, 0xBF2087A0); + WREG32(index_reg, 0x30800); + WREG32(data_reg, 0x8900BF00); + WREG32(index_reg, 0x30801); + WREG32(data_reg, 0xBF008900); +} + +struct rv515_watermark { + u32 lb_request_fifo_depth; + fixed20_12 num_line_pair; + fixed20_12 estimated_width; + fixed20_12 worst_case_latency; + fixed20_12 consumption_rate; + fixed20_12 active_time; + fixed20_12 dbpp; + fixed20_12 priority_mark_max; + fixed20_12 priority_mark; + fixed20_12 sclk; +}; + +void rv515_crtc_bandwidth_compute(struct radeon_device *rdev, + struct radeon_crtc *crtc, + struct rv515_watermark *wm) +{ + struct drm_display_mode *mode = &crtc->base.mode; + fixed20_12 a, b, c; + fixed20_12 pclk, request_fifo_depth, tolerable_latency, estimated_width; + fixed20_12 consumption_time, line_time, chunk_time, read_delay_latency; + + if (!crtc->base.enabled) { + /* FIXME: wouldn't it better to set priority mark to maximum */ + wm->lb_request_fifo_depth = 4; + return; + } + + if (crtc->vsc.full > dfixed_const(2)) + wm->num_line_pair.full = dfixed_const(2); + else + wm->num_line_pair.full = dfixed_const(1); + + b.full = dfixed_const(mode->crtc_hdisplay); + c.full = dfixed_const(256); + a.full = dfixed_div(b, c); + request_fifo_depth.full = dfixed_mul(a, wm->num_line_pair); + request_fifo_depth.full = dfixed_ceil(request_fifo_depth); + if (a.full < dfixed_const(4)) { + wm->lb_request_fifo_depth = 4; + } else { + wm->lb_request_fifo_depth = dfixed_trunc(request_fifo_depth); + } + + /* Determine consumption rate + * pclk = pixel clock period(ns) = 1000 / (mode.clock / 1000) + * vtaps = number of vertical taps, + * vsc = vertical scaling ratio, defined as source/destination + * hsc = horizontal scaling ration, defined as source/destination + */ + a.full = dfixed_const(mode->clock); + b.full = dfixed_const(1000); + a.full = dfixed_div(a, b); + pclk.full = dfixed_div(b, a); + if (crtc->rmx_type != RMX_OFF) { + b.full = dfixed_const(2); + if (crtc->vsc.full > b.full) + b.full = crtc->vsc.full; + b.full = dfixed_mul(b, crtc->hsc); + c.full = dfixed_const(2); + b.full = dfixed_div(b, c); + consumption_time.full = dfixed_div(pclk, b); + } else { + consumption_time.full = pclk.full; + } + a.full = dfixed_const(1); + wm->consumption_rate.full = dfixed_div(a, consumption_time); + + + /* Determine line time + * LineTime = total time for one line of displayhtotal + * LineTime = total number of horizontal pixels + * pclk = pixel clock period(ns) + */ + a.full = dfixed_const(crtc->base.mode.crtc_htotal); + line_time.full = dfixed_mul(a, pclk); + + /* Determine active time + * ActiveTime = time of active region of display within one line, + * hactive = total number of horizontal active pixels + * htotal = total number of horizontal pixels + */ + a.full = dfixed_const(crtc->base.mode.crtc_htotal); + b.full = dfixed_const(crtc->base.mode.crtc_hdisplay); + wm->active_time.full = dfixed_mul(line_time, b); + wm->active_time.full = dfixed_div(wm->active_time, a); + + /* Determine chunk time + * ChunkTime = the time it takes the DCP to send one chunk of data + * to the LB which consists of pipeline delay and inter chunk gap + * sclk = system clock(Mhz) + */ + a.full = dfixed_const(600 * 1000); + chunk_time.full = dfixed_div(a, rdev->pm.sclk); + read_delay_latency.full = dfixed_const(1000); + + /* Determine the worst case latency + * NumLinePair = Number of line pairs to request(1=2 lines, 2=4 lines) + * WorstCaseLatency = worst case time from urgent to when the MC starts + * to return data + * READ_DELAY_IDLE_MAX = constant of 1us + * ChunkTime = time it takes the DCP to send one chunk of data to the LB + * which consists of pipeline delay and inter chunk gap + */ + if (dfixed_trunc(wm->num_line_pair) > 1) { + a.full = dfixed_const(3); + wm->worst_case_latency.full = dfixed_mul(a, chunk_time); + wm->worst_case_latency.full += read_delay_latency.full; + } else { + wm->worst_case_latency.full = chunk_time.full + read_delay_latency.full; + } + + /* Determine the tolerable latency + * TolerableLatency = Any given request has only 1 line time + * for the data to be returned + * LBRequestFifoDepth = Number of chunk requests the LB can + * put into the request FIFO for a display + * LineTime = total time for one line of display + * ChunkTime = the time it takes the DCP to send one chunk + * of data to the LB which consists of + * pipeline delay and inter chunk gap + */ + if ((2+wm->lb_request_fifo_depth) >= dfixed_trunc(request_fifo_depth)) { + tolerable_latency.full = line_time.full; + } else { + tolerable_latency.full = dfixed_const(wm->lb_request_fifo_depth - 2); + tolerable_latency.full = request_fifo_depth.full - tolerable_latency.full; + tolerable_latency.full = dfixed_mul(tolerable_latency, chunk_time); + tolerable_latency.full = line_time.full - tolerable_latency.full; + } + /* We assume worst case 32bits (4 bytes) */ + wm->dbpp.full = dfixed_const(2 * 16); + + /* Determine the maximum priority mark + * width = viewport width in pixels + */ + a.full = dfixed_const(16); + wm->priority_mark_max.full = dfixed_const(crtc->base.mode.crtc_hdisplay); + wm->priority_mark_max.full = dfixed_div(wm->priority_mark_max, a); + wm->priority_mark_max.full = dfixed_ceil(wm->priority_mark_max); + + /* Determine estimated width */ + estimated_width.full = tolerable_latency.full - wm->worst_case_latency.full; + estimated_width.full = dfixed_div(estimated_width, consumption_time); + if (dfixed_trunc(estimated_width) > crtc->base.mode.crtc_hdisplay) { + wm->priority_mark.full = wm->priority_mark_max.full; + } else { + a.full = dfixed_const(16); + wm->priority_mark.full = dfixed_div(estimated_width, a); + wm->priority_mark.full = dfixed_ceil(wm->priority_mark); + wm->priority_mark.full = wm->priority_mark_max.full - wm->priority_mark.full; + } +} + +void rv515_bandwidth_avivo_update(struct radeon_device *rdev) +{ + struct drm_display_mode *mode0 = NULL; + struct drm_display_mode *mode1 = NULL; + struct rv515_watermark wm0; + struct rv515_watermark wm1; + u32 tmp; + u32 d1mode_priority_a_cnt = MODE_PRIORITY_OFF; + u32 d2mode_priority_a_cnt = MODE_PRIORITY_OFF; + fixed20_12 priority_mark02, priority_mark12, fill_rate; + fixed20_12 a, b; + + if (rdev->mode_info.crtcs[0]->base.enabled) + mode0 = &rdev->mode_info.crtcs[0]->base.mode; + if (rdev->mode_info.crtcs[1]->base.enabled) + mode1 = &rdev->mode_info.crtcs[1]->base.mode; + rs690_line_buffer_adjust(rdev, mode0, mode1); + + rv515_crtc_bandwidth_compute(rdev, rdev->mode_info.crtcs[0], &wm0); + rv515_crtc_bandwidth_compute(rdev, rdev->mode_info.crtcs[1], &wm1); + + tmp = wm0.lb_request_fifo_depth; + tmp |= wm1.lb_request_fifo_depth << 16; + WREG32(LB_MAX_REQ_OUTSTANDING, tmp); + + if (mode0 && mode1) { + if (dfixed_trunc(wm0.dbpp) > 64) + a.full = dfixed_div(wm0.dbpp, wm0.num_line_pair); + else + a.full = wm0.num_line_pair.full; + if (dfixed_trunc(wm1.dbpp) > 64) + b.full = dfixed_div(wm1.dbpp, wm1.num_line_pair); + else + b.full = wm1.num_line_pair.full; + a.full += b.full; + fill_rate.full = dfixed_div(wm0.sclk, a); + if (wm0.consumption_rate.full > fill_rate.full) { + b.full = wm0.consumption_rate.full - fill_rate.full; + b.full = dfixed_mul(b, wm0.active_time); + a.full = dfixed_const(16); + b.full = dfixed_div(b, a); + a.full = dfixed_mul(wm0.worst_case_latency, + wm0.consumption_rate); + priority_mark02.full = a.full + b.full; + } else { + a.full = dfixed_mul(wm0.worst_case_latency, + wm0.consumption_rate); + b.full = dfixed_const(16 * 1000); + priority_mark02.full = dfixed_div(a, b); + } + if (wm1.consumption_rate.full > fill_rate.full) { + b.full = wm1.consumption_rate.full - fill_rate.full; + b.full = dfixed_mul(b, wm1.active_time); + a.full = dfixed_const(16); + b.full = dfixed_div(b, a); + a.full = dfixed_mul(wm1.worst_case_latency, + wm1.consumption_rate); + priority_mark12.full = a.full + b.full; + } else { + a.full = dfixed_mul(wm1.worst_case_latency, + wm1.consumption_rate); + b.full = dfixed_const(16 * 1000); + priority_mark12.full = dfixed_div(a, b); + } + if (wm0.priority_mark.full > priority_mark02.full) + priority_mark02.full = wm0.priority_mark.full; + if (dfixed_trunc(priority_mark02) < 0) + priority_mark02.full = 0; + if (wm0.priority_mark_max.full > priority_mark02.full) + priority_mark02.full = wm0.priority_mark_max.full; + if (wm1.priority_mark.full > priority_mark12.full) + priority_mark12.full = wm1.priority_mark.full; + if (dfixed_trunc(priority_mark12) < 0) + priority_mark12.full = 0; + if (wm1.priority_mark_max.full > priority_mark12.full) + priority_mark12.full = wm1.priority_mark_max.full; + d1mode_priority_a_cnt = dfixed_trunc(priority_mark02); + d2mode_priority_a_cnt = dfixed_trunc(priority_mark12); + if (rdev->disp_priority == 2) { + d1mode_priority_a_cnt |= MODE_PRIORITY_ALWAYS_ON; + d2mode_priority_a_cnt |= MODE_PRIORITY_ALWAYS_ON; + } + } else if (mode0) { + if (dfixed_trunc(wm0.dbpp) > 64) + a.full = dfixed_div(wm0.dbpp, wm0.num_line_pair); + else + a.full = wm0.num_line_pair.full; + fill_rate.full = dfixed_div(wm0.sclk, a); + if (wm0.consumption_rate.full > fill_rate.full) { + b.full = wm0.consumption_rate.full - fill_rate.full; + b.full = dfixed_mul(b, wm0.active_time); + a.full = dfixed_const(16); + b.full = dfixed_div(b, a); + a.full = dfixed_mul(wm0.worst_case_latency, + wm0.consumption_rate); + priority_mark02.full = a.full + b.full; + } else { + a.full = dfixed_mul(wm0.worst_case_latency, + wm0.consumption_rate); + b.full = dfixed_const(16); + priority_mark02.full = dfixed_div(a, b); + } + if (wm0.priority_mark.full > priority_mark02.full) + priority_mark02.full = wm0.priority_mark.full; + if (dfixed_trunc(priority_mark02) < 0) + priority_mark02.full = 0; + if (wm0.priority_mark_max.full > priority_mark02.full) + priority_mark02.full = wm0.priority_mark_max.full; + d1mode_priority_a_cnt = dfixed_trunc(priority_mark02); + if (rdev->disp_priority == 2) + d1mode_priority_a_cnt |= MODE_PRIORITY_ALWAYS_ON; + } else if (mode1) { + if (dfixed_trunc(wm1.dbpp) > 64) + a.full = dfixed_div(wm1.dbpp, wm1.num_line_pair); + else + a.full = wm1.num_line_pair.full; + fill_rate.full = dfixed_div(wm1.sclk, a); + if (wm1.consumption_rate.full > fill_rate.full) { + b.full = wm1.consumption_rate.full - fill_rate.full; + b.full = dfixed_mul(b, wm1.active_time); + a.full = dfixed_const(16); + b.full = dfixed_div(b, a); + a.full = dfixed_mul(wm1.worst_case_latency, + wm1.consumption_rate); + priority_mark12.full = a.full + b.full; + } else { + a.full = dfixed_mul(wm1.worst_case_latency, + wm1.consumption_rate); + b.full = dfixed_const(16 * 1000); + priority_mark12.full = dfixed_div(a, b); + } + if (wm1.priority_mark.full > priority_mark12.full) + priority_mark12.full = wm1.priority_mark.full; + if (dfixed_trunc(priority_mark12) < 0) + priority_mark12.full = 0; + if (wm1.priority_mark_max.full > priority_mark12.full) + priority_mark12.full = wm1.priority_mark_max.full; + d2mode_priority_a_cnt = dfixed_trunc(priority_mark12); + if (rdev->disp_priority == 2) + d2mode_priority_a_cnt |= MODE_PRIORITY_ALWAYS_ON; + } + + WREG32(D1MODE_PRIORITY_A_CNT, d1mode_priority_a_cnt); + WREG32(D1MODE_PRIORITY_B_CNT, d1mode_priority_a_cnt); + WREG32(D2MODE_PRIORITY_A_CNT, d2mode_priority_a_cnt); + WREG32(D2MODE_PRIORITY_B_CNT, d2mode_priority_a_cnt); +} + +void rv515_bandwidth_update(struct radeon_device *rdev) +{ + uint32_t tmp; + struct drm_display_mode *mode0 = NULL; + struct drm_display_mode *mode1 = NULL; + + radeon_update_display_priority(rdev); + + if (rdev->mode_info.crtcs[0]->base.enabled) + mode0 = &rdev->mode_info.crtcs[0]->base.mode; + if (rdev->mode_info.crtcs[1]->base.enabled) + mode1 = &rdev->mode_info.crtcs[1]->base.mode; + /* + * Set display0/1 priority up in the memory controller for + * modes if the user specifies HIGH for displaypriority + * option. + */ + if ((rdev->disp_priority == 2) && + (rdev->family == CHIP_RV515)) { + tmp = RREG32_MC(MC_MISC_LAT_TIMER); + tmp &= ~MC_DISP1R_INIT_LAT_MASK; + tmp &= ~MC_DISP0R_INIT_LAT_MASK; + if (mode1) + tmp |= (1 << MC_DISP1R_INIT_LAT_SHIFT); + if (mode0) + tmp |= (1 << MC_DISP0R_INIT_LAT_SHIFT); + WREG32_MC(MC_MISC_LAT_TIMER, tmp); + } + rv515_bandwidth_avivo_update(rdev); +} diff --git a/drivers/gpu/drm/radeon/rv515d.h b/drivers/gpu/drm/radeon/rv515d.h new file mode 100644 index 00000000..590309a7 --- /dev/null +++ b/drivers/gpu/drm/radeon/rv515d.h @@ -0,0 +1,649 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#ifndef __RV515D_H__ +#define __RV515D_H__ + +/* + * RV515 registers + */ +#define PCIE_INDEX 0x0030 +#define PCIE_DATA 0x0034 +#define MC_IND_INDEX 0x0070 +#define MC_IND_WR_EN (1 << 24) +#define MC_IND_DATA 0x0074 +#define RBBM_SOFT_RESET 0x00F0 +#define CONFIG_MEMSIZE 0x00F8 +#define HDP_FB_LOCATION 0x0134 +#define CP_CSQ_CNTL 0x0740 +#define CP_CSQ_MODE 0x0744 +#define CP_CSQ_ADDR 0x07F0 +#define CP_CSQ_DATA 0x07F4 +#define CP_CSQ_STAT 0x07F8 +#define CP_CSQ2_STAT 0x07FC +#define RBBM_STATUS 0x0E40 +#define DST_PIPE_CONFIG 0x170C +#define WAIT_UNTIL 0x1720 +#define WAIT_2D_IDLE (1 << 14) +#define WAIT_3D_IDLE (1 << 15) +#define WAIT_2D_IDLECLEAN (1 << 16) +#define WAIT_3D_IDLECLEAN (1 << 17) +#define ISYNC_CNTL 0x1724 +#define ISYNC_ANY2D_IDLE3D (1 << 0) +#define ISYNC_ANY3D_IDLE2D (1 << 1) +#define ISYNC_TRIG2D_IDLE3D (1 << 2) +#define ISYNC_TRIG3D_IDLE2D (1 << 3) +#define ISYNC_WAIT_IDLEGUI (1 << 4) +#define ISYNC_CPSCRATCH_IDLEGUI (1 << 5) +#define VAP_INDEX_OFFSET 0x208C +#define VAP_PVS_STATE_FLUSH_REG 0x2284 +#define GB_ENABLE 0x4008 +#define GB_MSPOS0 0x4010 +#define MS_X0_SHIFT 0 +#define MS_Y0_SHIFT 4 +#define MS_X1_SHIFT 8 +#define MS_Y1_SHIFT 12 +#define MS_X2_SHIFT 16 +#define MS_Y2_SHIFT 20 +#define MSBD0_Y_SHIFT 24 +#define MSBD0_X_SHIFT 28 +#define GB_MSPOS1 0x4014 +#define MS_X3_SHIFT 0 +#define MS_Y3_SHIFT 4 +#define MS_X4_SHIFT 8 +#define MS_Y4_SHIFT 12 +#define MS_X5_SHIFT 16 +#define MS_Y5_SHIFT 20 +#define MSBD1_SHIFT 24 +#define GB_TILE_CONFIG 0x4018 +#define ENABLE_TILING (1 << 0) +#define PIPE_COUNT_MASK 0x0000000E +#define PIPE_COUNT_SHIFT 1 +#define TILE_SIZE_8 (0 << 4) +#define TILE_SIZE_16 (1 << 4) +#define TILE_SIZE_32 (2 << 4) +#define SUBPIXEL_1_12 (0 << 16) +#define SUBPIXEL_1_16 (1 << 16) +#define GB_SELECT 0x401C +#define GB_AA_CONFIG 0x4020 +#define GB_PIPE_SELECT 0x402C +#define GA_ENHANCE 0x4274 +#define GA_DEADLOCK_CNTL (1 << 0) +#define GA_FASTSYNC_CNTL (1 << 1) +#define GA_POLY_MODE 0x4288 +#define FRONT_PTYPE_POINT (0 << 4) +#define FRONT_PTYPE_LINE (1 << 4) +#define FRONT_PTYPE_TRIANGE (2 << 4) +#define BACK_PTYPE_POINT (0 << 7) +#define BACK_PTYPE_LINE (1 << 7) +#define BACK_PTYPE_TRIANGE (2 << 7) +#define GA_ROUND_MODE 0x428C +#define GEOMETRY_ROUND_TRUNC (0 << 0) +#define GEOMETRY_ROUND_NEAREST (1 << 0) +#define COLOR_ROUND_TRUNC (0 << 2) +#define COLOR_ROUND_NEAREST (1 << 2) +#define SU_REG_DEST 0x42C8 +#define RB3D_DSTCACHE_CTLSTAT 0x4E4C +#define RB3D_DC_FLUSH (2 << 0) +#define RB3D_DC_FREE (2 << 2) +#define RB3D_DC_FINISH (1 << 4) +#define ZB_ZCACHE_CTLSTAT 0x4F18 +#define ZC_FLUSH (1 << 0) +#define ZC_FREE (1 << 1) +#define DC_LB_MEMORY_SPLIT 0x6520 +#define DC_LB_MEMORY_SPLIT_MASK 0x00000003 +#define DC_LB_MEMORY_SPLIT_SHIFT 0 +#define DC_LB_MEMORY_SPLIT_D1HALF_D2HALF 0 +#define DC_LB_MEMORY_SPLIT_D1_3Q_D2_1Q 1 +#define DC_LB_MEMORY_SPLIT_D1_ONLY 2 +#define DC_LB_MEMORY_SPLIT_D1_1Q_D2_3Q 3 +#define DC_LB_MEMORY_SPLIT_SHIFT_MODE (1 << 2) +#define DC_LB_DISP1_END_ADR_SHIFT 4 +#define DC_LB_DISP1_END_ADR_MASK 0x00007FF0 +#define D1MODE_PRIORITY_A_CNT 0x6548 +#define MODE_PRIORITY_MARK_MASK 0x00007FFF +#define MODE_PRIORITY_OFF (1 << 16) +#define MODE_PRIORITY_ALWAYS_ON (1 << 20) +#define MODE_PRIORITY_FORCE_MASK (1 << 24) +#define D1MODE_PRIORITY_B_CNT 0x654C +#define LB_MAX_REQ_OUTSTANDING 0x6D58 +#define LB_D1_MAX_REQ_OUTSTANDING_MASK 0x0000000F +#define LB_D1_MAX_REQ_OUTSTANDING_SHIFT 0 +#define LB_D2_MAX_REQ_OUTSTANDING_MASK 0x000F0000 +#define LB_D2_MAX_REQ_OUTSTANDING_SHIFT 16 +#define D2MODE_PRIORITY_A_CNT 0x6D48 +#define D2MODE_PRIORITY_B_CNT 0x6D4C + +/* ix[MC] registers */ +#define MC_FB_LOCATION 0x01 +#define MC_FB_START_MASK 0x0000FFFF +#define MC_FB_START_SHIFT 0 +#define MC_FB_TOP_MASK 0xFFFF0000 +#define MC_FB_TOP_SHIFT 16 +#define MC_AGP_LOCATION 0x02 +#define MC_AGP_START_MASK 0x0000FFFF +#define MC_AGP_START_SHIFT 0 +#define MC_AGP_TOP_MASK 0xFFFF0000 +#define MC_AGP_TOP_SHIFT 16 +#define MC_AGP_BASE 0x03 +#define MC_AGP_BASE_2 0x04 +#define MC_CNTL 0x5 +#define MEM_NUM_CHANNELS_MASK 0x00000003 +#define MC_STATUS 0x08 +#define MC_STATUS_IDLE (1 << 4) +#define MC_MISC_LAT_TIMER 0x09 +#define MC_CPR_INIT_LAT_MASK 0x0000000F +#define MC_VF_INIT_LAT_MASK 0x000000F0 +#define MC_DISP0R_INIT_LAT_MASK 0x00000F00 +#define MC_DISP0R_INIT_LAT_SHIFT 8 +#define MC_DISP1R_INIT_LAT_MASK 0x0000F000 +#define MC_DISP1R_INIT_LAT_SHIFT 12 +#define MC_FIXED_INIT_LAT_MASK 0x000F0000 +#define MC_E2R_INIT_LAT_MASK 0x00F00000 +#define SAME_PAGE_PRIO_MASK 0x0F000000 +#define MC_GLOBW_INIT_LAT_MASK 0xF0000000 + + +/* + * PM4 packet + */ +#define CP_PACKET0 0x00000000 +#define PACKET0_BASE_INDEX_SHIFT 0 +#define PACKET0_BASE_INDEX_MASK (0x1ffff << 0) +#define PACKET0_COUNT_SHIFT 16 +#define PACKET0_COUNT_MASK (0x3fff << 16) +#define CP_PACKET1 0x40000000 +#define CP_PACKET2 0x80000000 +#define PACKET2_PAD_SHIFT 0 +#define PACKET2_PAD_MASK (0x3fffffff << 0) +#define CP_PACKET3 0xC0000000 +#define PACKET3_IT_OPCODE_SHIFT 8 +#define PACKET3_IT_OPCODE_MASK (0xff << 8) +#define PACKET3_COUNT_SHIFT 16 +#define PACKET3_COUNT_MASK (0x3fff << 16) +/* PACKET3 op code */ +#define PACKET3_NOP 0x10 +#define PACKET3_3D_DRAW_VBUF 0x28 +#define PACKET3_3D_DRAW_IMMD 0x29 +#define PACKET3_3D_DRAW_INDX 0x2A +#define PACKET3_3D_LOAD_VBPNTR 0x2F +#define PACKET3_INDX_BUFFER 0x33 +#define PACKET3_3D_DRAW_VBUF_2 0x34 +#define PACKET3_3D_DRAW_IMMD_2 0x35 +#define PACKET3_3D_DRAW_INDX_2 0x36 +#define PACKET3_BITBLT_MULTI 0x9B + +#define PACKET0(reg, n) (CP_PACKET0 | \ + REG_SET(PACKET0_BASE_INDEX, (reg) >> 2) | \ + REG_SET(PACKET0_COUNT, (n))) +#define PACKET2(v) (CP_PACKET2 | REG_SET(PACKET2_PAD, (v))) +#define PACKET3(op, n) (CP_PACKET3 | \ + REG_SET(PACKET3_IT_OPCODE, (op)) | \ + REG_SET(PACKET3_COUNT, (n))) + +#define PACKET_TYPE0 0 +#define PACKET_TYPE1 1 +#define PACKET_TYPE2 2 +#define PACKET_TYPE3 3 + +#define CP_PACKET_GET_TYPE(h) (((h) >> 30) & 3) +#define CP_PACKET_GET_COUNT(h) (((h) >> 16) & 0x3FFF) +#define CP_PACKET0_GET_REG(h) (((h) & 0x1FFF) << 2) +#define CP_PACKET0_GET_ONE_REG_WR(h) (((h) >> 15) & 1) +#define CP_PACKET3_GET_OPCODE(h) (((h) >> 8) & 0xFF) + +/* Registers */ +#define R_0000F0_RBBM_SOFT_RESET 0x0000F0 +#define S_0000F0_SOFT_RESET_CP(x) (((x) & 0x1) << 0) +#define G_0000F0_SOFT_RESET_CP(x) (((x) >> 0) & 0x1) +#define C_0000F0_SOFT_RESET_CP 0xFFFFFFFE +#define S_0000F0_SOFT_RESET_HI(x) (((x) & 0x1) << 1) +#define G_0000F0_SOFT_RESET_HI(x) (((x) >> 1) & 0x1) +#define C_0000F0_SOFT_RESET_HI 0xFFFFFFFD +#define S_0000F0_SOFT_RESET_VAP(x) (((x) & 0x1) << 2) +#define G_0000F0_SOFT_RESET_VAP(x) (((x) >> 2) & 0x1) +#define C_0000F0_SOFT_RESET_VAP 0xFFFFFFFB +#define S_0000F0_SOFT_RESET_RE(x) (((x) & 0x1) << 3) +#define G_0000F0_SOFT_RESET_RE(x) (((x) >> 3) & 0x1) +#define C_0000F0_SOFT_RESET_RE 0xFFFFFFF7 +#define S_0000F0_SOFT_RESET_PP(x) (((x) & 0x1) << 4) +#define G_0000F0_SOFT_RESET_PP(x) (((x) >> 4) & 0x1) +#define C_0000F0_SOFT_RESET_PP 0xFFFFFFEF +#define S_0000F0_SOFT_RESET_E2(x) (((x) & 0x1) << 5) +#define G_0000F0_SOFT_RESET_E2(x) (((x) >> 5) & 0x1) +#define C_0000F0_SOFT_RESET_E2 0xFFFFFFDF +#define S_0000F0_SOFT_RESET_RB(x) (((x) & 0x1) << 6) +#define G_0000F0_SOFT_RESET_RB(x) (((x) >> 6) & 0x1) +#define C_0000F0_SOFT_RESET_RB 0xFFFFFFBF +#define S_0000F0_SOFT_RESET_HDP(x) (((x) & 0x1) << 7) +#define G_0000F0_SOFT_RESET_HDP(x) (((x) >> 7) & 0x1) +#define C_0000F0_SOFT_RESET_HDP 0xFFFFFF7F +#define S_0000F0_SOFT_RESET_MC(x) (((x) & 0x1) << 8) +#define G_0000F0_SOFT_RESET_MC(x) (((x) >> 8) & 0x1) +#define C_0000F0_SOFT_RESET_MC 0xFFFFFEFF +#define S_0000F0_SOFT_RESET_AIC(x) (((x) & 0x1) << 9) +#define G_0000F0_SOFT_RESET_AIC(x) (((x) >> 9) & 0x1) +#define C_0000F0_SOFT_RESET_AIC 0xFFFFFDFF +#define S_0000F0_SOFT_RESET_VIP(x) (((x) & 0x1) << 10) +#define G_0000F0_SOFT_RESET_VIP(x) (((x) >> 10) & 0x1) +#define C_0000F0_SOFT_RESET_VIP 0xFFFFFBFF +#define S_0000F0_SOFT_RESET_DISP(x) (((x) & 0x1) << 11) +#define G_0000F0_SOFT_RESET_DISP(x) (((x) >> 11) & 0x1) +#define C_0000F0_SOFT_RESET_DISP 0xFFFFF7FF +#define S_0000F0_SOFT_RESET_CG(x) (((x) & 0x1) << 12) +#define G_0000F0_SOFT_RESET_CG(x) (((x) >> 12) & 0x1) +#define C_0000F0_SOFT_RESET_CG 0xFFFFEFFF +#define S_0000F0_SOFT_RESET_GA(x) (((x) & 0x1) << 13) +#define G_0000F0_SOFT_RESET_GA(x) (((x) >> 13) & 0x1) +#define C_0000F0_SOFT_RESET_GA 0xFFFFDFFF +#define S_0000F0_SOFT_RESET_IDCT(x) (((x) & 0x1) << 14) +#define G_0000F0_SOFT_RESET_IDCT(x) (((x) >> 14) & 0x1) +#define C_0000F0_SOFT_RESET_IDCT 0xFFFFBFFF +#define R_0000F8_CONFIG_MEMSIZE 0x0000F8 +#define S_0000F8_CONFIG_MEMSIZE(x) (((x) & 0xFFFFFFFF) << 0) +#define G_0000F8_CONFIG_MEMSIZE(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_0000F8_CONFIG_MEMSIZE 0x00000000 +#define R_000134_HDP_FB_LOCATION 0x000134 +#define S_000134_HDP_FB_START(x) (((x) & 0xFFFF) << 0) +#define G_000134_HDP_FB_START(x) (((x) >> 0) & 0xFFFF) +#define C_000134_HDP_FB_START 0xFFFF0000 +#define R_000300_VGA_RENDER_CONTROL 0x000300 +#define S_000300_VGA_BLINK_RATE(x) (((x) & 0x1F) << 0) +#define G_000300_VGA_BLINK_RATE(x) (((x) >> 0) & 0x1F) +#define C_000300_VGA_BLINK_RATE 0xFFFFFFE0 +#define S_000300_VGA_BLINK_MODE(x) (((x) & 0x3) << 5) +#define G_000300_VGA_BLINK_MODE(x) (((x) >> 5) & 0x3) +#define C_000300_VGA_BLINK_MODE 0xFFFFFF9F +#define S_000300_VGA_CURSOR_BLINK_INVERT(x) (((x) & 0x1) << 7) +#define G_000300_VGA_CURSOR_BLINK_INVERT(x) (((x) >> 7) & 0x1) +#define C_000300_VGA_CURSOR_BLINK_INVERT 0xFFFFFF7F +#define S_000300_VGA_EXTD_ADDR_COUNT_ENABLE(x) (((x) & 0x1) << 8) +#define G_000300_VGA_EXTD_ADDR_COUNT_ENABLE(x) (((x) >> 8) & 0x1) +#define C_000300_VGA_EXTD_ADDR_COUNT_ENABLE 0xFFFFFEFF +#define S_000300_VGA_VSTATUS_CNTL(x) (((x) & 0x3) << 16) +#define G_000300_VGA_VSTATUS_CNTL(x) (((x) >> 16) & 0x3) +#define C_000300_VGA_VSTATUS_CNTL 0xFFFCFFFF +#define S_000300_VGA_LOCK_8DOT(x) (((x) & 0x1) << 24) +#define G_000300_VGA_LOCK_8DOT(x) (((x) >> 24) & 0x1) +#define C_000300_VGA_LOCK_8DOT 0xFEFFFFFF +#define S_000300_VGAREG_LINECMP_COMPATIBILITY_SEL(x) (((x) & 0x1) << 25) +#define G_000300_VGAREG_LINECMP_COMPATIBILITY_SEL(x) (((x) >> 25) & 0x1) +#define C_000300_VGAREG_LINECMP_COMPATIBILITY_SEL 0xFDFFFFFF +#define R_000310_VGA_MEMORY_BASE_ADDRESS 0x000310 +#define S_000310_VGA_MEMORY_BASE_ADDRESS(x) (((x) & 0xFFFFFFFF) << 0) +#define G_000310_VGA_MEMORY_BASE_ADDRESS(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_000310_VGA_MEMORY_BASE_ADDRESS 0x00000000 +#define R_000328_VGA_HDP_CONTROL 0x000328 +#define S_000328_VGA_MEM_PAGE_SELECT_EN(x) (((x) & 0x1) << 0) +#define G_000328_VGA_MEM_PAGE_SELECT_EN(x) (((x) >> 0) & 0x1) +#define C_000328_VGA_MEM_PAGE_SELECT_EN 0xFFFFFFFE +#define S_000328_VGA_RBBM_LOCK_DISABLE(x) (((x) & 0x1) << 8) +#define G_000328_VGA_RBBM_LOCK_DISABLE(x) (((x) >> 8) & 0x1) +#define C_000328_VGA_RBBM_LOCK_DISABLE 0xFFFFFEFF +#define S_000328_VGA_SOFT_RESET(x) (((x) & 0x1) << 16) +#define G_000328_VGA_SOFT_RESET(x) (((x) >> 16) & 0x1) +#define C_000328_VGA_SOFT_RESET 0xFFFEFFFF +#define S_000328_VGA_TEST_RESET_CONTROL(x) (((x) & 0x1) << 24) +#define G_000328_VGA_TEST_RESET_CONTROL(x) (((x) >> 24) & 0x1) +#define C_000328_VGA_TEST_RESET_CONTROL 0xFEFFFFFF +#define R_000330_D1VGA_CONTROL 0x000330 +#define S_000330_D1VGA_MODE_ENABLE(x) (((x) & 0x1) << 0) +#define G_000330_D1VGA_MODE_ENABLE(x) (((x) >> 0) & 0x1) +#define C_000330_D1VGA_MODE_ENABLE 0xFFFFFFFE +#define S_000330_D1VGA_TIMING_SELECT(x) (((x) & 0x1) << 8) +#define G_000330_D1VGA_TIMING_SELECT(x) (((x) >> 8) & 0x1) +#define C_000330_D1VGA_TIMING_SELECT 0xFFFFFEFF +#define S_000330_D1VGA_SYNC_POLARITY_SELECT(x) (((x) & 0x1) << 9) +#define G_000330_D1VGA_SYNC_POLARITY_SELECT(x) (((x) >> 9) & 0x1) +#define C_000330_D1VGA_SYNC_POLARITY_SELECT 0xFFFFFDFF +#define S_000330_D1VGA_OVERSCAN_TIMING_SELECT(x) (((x) & 0x1) << 10) +#define G_000330_D1VGA_OVERSCAN_TIMING_SELECT(x) (((x) >> 10) & 0x1) +#define C_000330_D1VGA_OVERSCAN_TIMING_SELECT 0xFFFFFBFF +#define S_000330_D1VGA_OVERSCAN_COLOR_EN(x) (((x) & 0x1) << 16) +#define G_000330_D1VGA_OVERSCAN_COLOR_EN(x) (((x) >> 16) & 0x1) +#define C_000330_D1VGA_OVERSCAN_COLOR_EN 0xFFFEFFFF +#define S_000330_D1VGA_ROTATE(x) (((x) & 0x3) << 24) +#define G_000330_D1VGA_ROTATE(x) (((x) >> 24) & 0x3) +#define C_000330_D1VGA_ROTATE 0xFCFFFFFF +#define R_000338_D2VGA_CONTROL 0x000338 +#define S_000338_D2VGA_MODE_ENABLE(x) (((x) & 0x1) << 0) +#define G_000338_D2VGA_MODE_ENABLE(x) (((x) >> 0) & 0x1) +#define C_000338_D2VGA_MODE_ENABLE 0xFFFFFFFE +#define S_000338_D2VGA_TIMING_SELECT(x) (((x) & 0x1) << 8) +#define G_000338_D2VGA_TIMING_SELECT(x) (((x) >> 8) & 0x1) +#define C_000338_D2VGA_TIMING_SELECT 0xFFFFFEFF +#define S_000338_D2VGA_SYNC_POLARITY_SELECT(x) (((x) & 0x1) << 9) +#define G_000338_D2VGA_SYNC_POLARITY_SELECT(x) (((x) >> 9) & 0x1) +#define C_000338_D2VGA_SYNC_POLARITY_SELECT 0xFFFFFDFF +#define S_000338_D2VGA_OVERSCAN_TIMING_SELECT(x) (((x) & 0x1) << 10) +#define G_000338_D2VGA_OVERSCAN_TIMING_SELECT(x) (((x) >> 10) & 0x1) +#define C_000338_D2VGA_OVERSCAN_TIMING_SELECT 0xFFFFFBFF +#define S_000338_D2VGA_OVERSCAN_COLOR_EN(x) (((x) & 0x1) << 16) +#define G_000338_D2VGA_OVERSCAN_COLOR_EN(x) (((x) >> 16) & 0x1) +#define C_000338_D2VGA_OVERSCAN_COLOR_EN 0xFFFEFFFF +#define S_000338_D2VGA_ROTATE(x) (((x) & 0x3) << 24) +#define G_000338_D2VGA_ROTATE(x) (((x) >> 24) & 0x3) +#define C_000338_D2VGA_ROTATE 0xFCFFFFFF +#define R_0007C0_CP_STAT 0x0007C0 +#define S_0007C0_MRU_BUSY(x) (((x) & 0x1) << 0) +#define G_0007C0_MRU_BUSY(x) (((x) >> 0) & 0x1) +#define C_0007C0_MRU_BUSY 0xFFFFFFFE +#define S_0007C0_MWU_BUSY(x) (((x) & 0x1) << 1) +#define G_0007C0_MWU_BUSY(x) (((x) >> 1) & 0x1) +#define C_0007C0_MWU_BUSY 0xFFFFFFFD +#define S_0007C0_RSIU_BUSY(x) (((x) & 0x1) << 2) +#define G_0007C0_RSIU_BUSY(x) (((x) >> 2) & 0x1) +#define C_0007C0_RSIU_BUSY 0xFFFFFFFB +#define S_0007C0_RCIU_BUSY(x) (((x) & 0x1) << 3) +#define G_0007C0_RCIU_BUSY(x) (((x) >> 3) & 0x1) +#define C_0007C0_RCIU_BUSY 0xFFFFFFF7 +#define S_0007C0_CSF_PRIMARY_BUSY(x) (((x) & 0x1) << 9) +#define G_0007C0_CSF_PRIMARY_BUSY(x) (((x) >> 9) & 0x1) +#define C_0007C0_CSF_PRIMARY_BUSY 0xFFFFFDFF +#define S_0007C0_CSF_INDIRECT_BUSY(x) (((x) & 0x1) << 10) +#define G_0007C0_CSF_INDIRECT_BUSY(x) (((x) >> 10) & 0x1) +#define C_0007C0_CSF_INDIRECT_BUSY 0xFFFFFBFF +#define S_0007C0_CSQ_PRIMARY_BUSY(x) (((x) & 0x1) << 11) +#define G_0007C0_CSQ_PRIMARY_BUSY(x) (((x) >> 11) & 0x1) +#define C_0007C0_CSQ_PRIMARY_BUSY 0xFFFFF7FF +#define S_0007C0_CSQ_INDIRECT_BUSY(x) (((x) & 0x1) << 12) +#define G_0007C0_CSQ_INDIRECT_BUSY(x) (((x) >> 12) & 0x1) +#define C_0007C0_CSQ_INDIRECT_BUSY 0xFFFFEFFF +#define S_0007C0_CSI_BUSY(x) (((x) & 0x1) << 13) +#define G_0007C0_CSI_BUSY(x) (((x) >> 13) & 0x1) +#define C_0007C0_CSI_BUSY 0xFFFFDFFF +#define S_0007C0_CSF_INDIRECT2_BUSY(x) (((x) & 0x1) << 14) +#define G_0007C0_CSF_INDIRECT2_BUSY(x) (((x) >> 14) & 0x1) +#define C_0007C0_CSF_INDIRECT2_BUSY 0xFFFFBFFF +#define S_0007C0_CSQ_INDIRECT2_BUSY(x) (((x) & 0x1) << 15) +#define G_0007C0_CSQ_INDIRECT2_BUSY(x) (((x) >> 15) & 0x1) +#define C_0007C0_CSQ_INDIRECT2_BUSY 0xFFFF7FFF +#define S_0007C0_GUIDMA_BUSY(x) (((x) & 0x1) << 28) +#define G_0007C0_GUIDMA_BUSY(x) (((x) >> 28) & 0x1) +#define C_0007C0_GUIDMA_BUSY 0xEFFFFFFF +#define S_0007C0_VIDDMA_BUSY(x) (((x) & 0x1) << 29) +#define G_0007C0_VIDDMA_BUSY(x) (((x) >> 29) & 0x1) +#define C_0007C0_VIDDMA_BUSY 0xDFFFFFFF +#define S_0007C0_CMDSTRM_BUSY(x) (((x) & 0x1) << 30) +#define G_0007C0_CMDSTRM_BUSY(x) (((x) >> 30) & 0x1) +#define C_0007C0_CMDSTRM_BUSY 0xBFFFFFFF +#define S_0007C0_CP_BUSY(x) (((x) & 0x1) << 31) +#define G_0007C0_CP_BUSY(x) (((x) >> 31) & 0x1) +#define C_0007C0_CP_BUSY 0x7FFFFFFF +#define R_000E40_RBBM_STATUS 0x000E40 +#define S_000E40_CMDFIFO_AVAIL(x) (((x) & 0x7F) << 0) +#define G_000E40_CMDFIFO_AVAIL(x) (((x) >> 0) & 0x7F) +#define C_000E40_CMDFIFO_AVAIL 0xFFFFFF80 +#define S_000E40_HIRQ_ON_RBB(x) (((x) & 0x1) << 8) +#define G_000E40_HIRQ_ON_RBB(x) (((x) >> 8) & 0x1) +#define C_000E40_HIRQ_ON_RBB 0xFFFFFEFF +#define S_000E40_CPRQ_ON_RBB(x) (((x) & 0x1) << 9) +#define G_000E40_CPRQ_ON_RBB(x) (((x) >> 9) & 0x1) +#define C_000E40_CPRQ_ON_RBB 0xFFFFFDFF +#define S_000E40_CFRQ_ON_RBB(x) (((x) & 0x1) << 10) +#define G_000E40_CFRQ_ON_RBB(x) (((x) >> 10) & 0x1) +#define C_000E40_CFRQ_ON_RBB 0xFFFFFBFF +#define S_000E40_HIRQ_IN_RTBUF(x) (((x) & 0x1) << 11) +#define G_000E40_HIRQ_IN_RTBUF(x) (((x) >> 11) & 0x1) +#define C_000E40_HIRQ_IN_RTBUF 0xFFFFF7FF +#define S_000E40_CPRQ_IN_RTBUF(x) (((x) & 0x1) << 12) +#define G_000E40_CPRQ_IN_RTBUF(x) (((x) >> 12) & 0x1) +#define C_000E40_CPRQ_IN_RTBUF 0xFFFFEFFF +#define S_000E40_CFRQ_IN_RTBUF(x) (((x) & 0x1) << 13) +#define G_000E40_CFRQ_IN_RTBUF(x) (((x) >> 13) & 0x1) +#define C_000E40_CFRQ_IN_RTBUF 0xFFFFDFFF +#define S_000E40_CF_PIPE_BUSY(x) (((x) & 0x1) << 14) +#define G_000E40_CF_PIPE_BUSY(x) (((x) >> 14) & 0x1) +#define C_000E40_CF_PIPE_BUSY 0xFFFFBFFF +#define S_000E40_ENG_EV_BUSY(x) (((x) & 0x1) << 15) +#define G_000E40_ENG_EV_BUSY(x) (((x) >> 15) & 0x1) +#define C_000E40_ENG_EV_BUSY 0xFFFF7FFF +#define S_000E40_CP_CMDSTRM_BUSY(x) (((x) & 0x1) << 16) +#define G_000E40_CP_CMDSTRM_BUSY(x) (((x) >> 16) & 0x1) +#define C_000E40_CP_CMDSTRM_BUSY 0xFFFEFFFF +#define S_000E40_E2_BUSY(x) (((x) & 0x1) << 17) +#define G_000E40_E2_BUSY(x) (((x) >> 17) & 0x1) +#define C_000E40_E2_BUSY 0xFFFDFFFF +#define S_000E40_RB2D_BUSY(x) (((x) & 0x1) << 18) +#define G_000E40_RB2D_BUSY(x) (((x) >> 18) & 0x1) +#define C_000E40_RB2D_BUSY 0xFFFBFFFF +#define S_000E40_RB3D_BUSY(x) (((x) & 0x1) << 19) +#define G_000E40_RB3D_BUSY(x) (((x) >> 19) & 0x1) +#define C_000E40_RB3D_BUSY 0xFFF7FFFF +#define S_000E40_VAP_BUSY(x) (((x) & 0x1) << 20) +#define G_000E40_VAP_BUSY(x) (((x) >> 20) & 0x1) +#define C_000E40_VAP_BUSY 0xFFEFFFFF +#define S_000E40_RE_BUSY(x) (((x) & 0x1) << 21) +#define G_000E40_RE_BUSY(x) (((x) >> 21) & 0x1) +#define C_000E40_RE_BUSY 0xFFDFFFFF +#define S_000E40_TAM_BUSY(x) (((x) & 0x1) << 22) +#define G_000E40_TAM_BUSY(x) (((x) >> 22) & 0x1) +#define C_000E40_TAM_BUSY 0xFFBFFFFF +#define S_000E40_TDM_BUSY(x) (((x) & 0x1) << 23) +#define G_000E40_TDM_BUSY(x) (((x) >> 23) & 0x1) +#define C_000E40_TDM_BUSY 0xFF7FFFFF +#define S_000E40_PB_BUSY(x) (((x) & 0x1) << 24) +#define G_000E40_PB_BUSY(x) (((x) >> 24) & 0x1) +#define C_000E40_PB_BUSY 0xFEFFFFFF +#define S_000E40_TIM_BUSY(x) (((x) & 0x1) << 25) +#define G_000E40_TIM_BUSY(x) (((x) >> 25) & 0x1) +#define C_000E40_TIM_BUSY 0xFDFFFFFF +#define S_000E40_GA_BUSY(x) (((x) & 0x1) << 26) +#define G_000E40_GA_BUSY(x) (((x) >> 26) & 0x1) +#define C_000E40_GA_BUSY 0xFBFFFFFF +#define S_000E40_CBA2D_BUSY(x) (((x) & 0x1) << 27) +#define G_000E40_CBA2D_BUSY(x) (((x) >> 27) & 0x1) +#define C_000E40_CBA2D_BUSY 0xF7FFFFFF +#define S_000E40_RBBM_HIBUSY(x) (((x) & 0x1) << 28) +#define G_000E40_RBBM_HIBUSY(x) (((x) >> 28) & 0x1) +#define C_000E40_RBBM_HIBUSY 0xEFFFFFFF +#define S_000E40_SKID_CFBUSY(x) (((x) & 0x1) << 29) +#define G_000E40_SKID_CFBUSY(x) (((x) >> 29) & 0x1) +#define C_000E40_SKID_CFBUSY 0xDFFFFFFF +#define S_000E40_VAP_VF_BUSY(x) (((x) & 0x1) << 30) +#define G_000E40_VAP_VF_BUSY(x) (((x) >> 30) & 0x1) +#define C_000E40_VAP_VF_BUSY 0xBFFFFFFF +#define S_000E40_GUI_ACTIVE(x) (((x) & 0x1) << 31) +#define G_000E40_GUI_ACTIVE(x) (((x) >> 31) & 0x1) +#define C_000E40_GUI_ACTIVE 0x7FFFFFFF +#define R_006080_D1CRTC_CONTROL 0x006080 +#define S_006080_D1CRTC_MASTER_EN(x) (((x) & 0x1) << 0) +#define G_006080_D1CRTC_MASTER_EN(x) (((x) >> 0) & 0x1) +#define C_006080_D1CRTC_MASTER_EN 0xFFFFFFFE +#define S_006080_D1CRTC_SYNC_RESET_SEL(x) (((x) & 0x1) << 4) +#define G_006080_D1CRTC_SYNC_RESET_SEL(x) (((x) >> 4) & 0x1) +#define C_006080_D1CRTC_SYNC_RESET_SEL 0xFFFFFFEF +#define S_006080_D1CRTC_DISABLE_POINT_CNTL(x) (((x) & 0x3) << 8) +#define G_006080_D1CRTC_DISABLE_POINT_CNTL(x) (((x) >> 8) & 0x3) +#define C_006080_D1CRTC_DISABLE_POINT_CNTL 0xFFFFFCFF +#define S_006080_D1CRTC_CURRENT_MASTER_EN_STATE(x) (((x) & 0x1) << 16) +#define G_006080_D1CRTC_CURRENT_MASTER_EN_STATE(x) (((x) >> 16) & 0x1) +#define C_006080_D1CRTC_CURRENT_MASTER_EN_STATE 0xFFFEFFFF +#define S_006080_D1CRTC_DISP_READ_REQUEST_DISABLE(x) (((x) & 0x1) << 24) +#define G_006080_D1CRTC_DISP_READ_REQUEST_DISABLE(x) (((x) >> 24) & 0x1) +#define C_006080_D1CRTC_DISP_READ_REQUEST_DISABLE 0xFEFFFFFF +#define R_0060E8_D1CRTC_UPDATE_LOCK 0x0060E8 +#define S_0060E8_D1CRTC_UPDATE_LOCK(x) (((x) & 0x1) << 0) +#define G_0060E8_D1CRTC_UPDATE_LOCK(x) (((x) >> 0) & 0x1) +#define C_0060E8_D1CRTC_UPDATE_LOCK 0xFFFFFFFE +#define R_006110_D1GRPH_PRIMARY_SURFACE_ADDRESS 0x006110 +#define S_006110_D1GRPH_PRIMARY_SURFACE_ADDRESS(x) (((x) & 0xFFFFFFFF) << 0) +#define G_006110_D1GRPH_PRIMARY_SURFACE_ADDRESS(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_006110_D1GRPH_PRIMARY_SURFACE_ADDRESS 0x00000000 +#define R_006118_D1GRPH_SECONDARY_SURFACE_ADDRESS 0x006118 +#define S_006118_D1GRPH_SECONDARY_SURFACE_ADDRESS(x) (((x) & 0xFFFFFFFF) << 0) +#define G_006118_D1GRPH_SECONDARY_SURFACE_ADDRESS(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_006118_D1GRPH_SECONDARY_SURFACE_ADDRESS 0x00000000 +#define R_006880_D2CRTC_CONTROL 0x006880 +#define S_006880_D2CRTC_MASTER_EN(x) (((x) & 0x1) << 0) +#define G_006880_D2CRTC_MASTER_EN(x) (((x) >> 0) & 0x1) +#define C_006880_D2CRTC_MASTER_EN 0xFFFFFFFE +#define S_006880_D2CRTC_SYNC_RESET_SEL(x) (((x) & 0x1) << 4) +#define G_006880_D2CRTC_SYNC_RESET_SEL(x) (((x) >> 4) & 0x1) +#define C_006880_D2CRTC_SYNC_RESET_SEL 0xFFFFFFEF +#define S_006880_D2CRTC_DISABLE_POINT_CNTL(x) (((x) & 0x3) << 8) +#define G_006880_D2CRTC_DISABLE_POINT_CNTL(x) (((x) >> 8) & 0x3) +#define C_006880_D2CRTC_DISABLE_POINT_CNTL 0xFFFFFCFF +#define S_006880_D2CRTC_CURRENT_MASTER_EN_STATE(x) (((x) & 0x1) << 16) +#define G_006880_D2CRTC_CURRENT_MASTER_EN_STATE(x) (((x) >> 16) & 0x1) +#define C_006880_D2CRTC_CURRENT_MASTER_EN_STATE 0xFFFEFFFF +#define S_006880_D2CRTC_DISP_READ_REQUEST_DISABLE(x) (((x) & 0x1) << 24) +#define G_006880_D2CRTC_DISP_READ_REQUEST_DISABLE(x) (((x) >> 24) & 0x1) +#define C_006880_D2CRTC_DISP_READ_REQUEST_DISABLE 0xFEFFFFFF +#define R_0068E8_D2CRTC_UPDATE_LOCK 0x0068E8 +#define S_0068E8_D2CRTC_UPDATE_LOCK(x) (((x) & 0x1) << 0) +#define G_0068E8_D2CRTC_UPDATE_LOCK(x) (((x) >> 0) & 0x1) +#define C_0068E8_D2CRTC_UPDATE_LOCK 0xFFFFFFFE +#define R_006910_D2GRPH_PRIMARY_SURFACE_ADDRESS 0x006910 +#define S_006910_D2GRPH_PRIMARY_SURFACE_ADDRESS(x) (((x) & 0xFFFFFFFF) << 0) +#define G_006910_D2GRPH_PRIMARY_SURFACE_ADDRESS(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_006910_D2GRPH_PRIMARY_SURFACE_ADDRESS 0x00000000 +#define R_006918_D2GRPH_SECONDARY_SURFACE_ADDRESS 0x006918 +#define S_006918_D2GRPH_SECONDARY_SURFACE_ADDRESS(x) (((x) & 0xFFFFFFFF) << 0) +#define G_006918_D2GRPH_SECONDARY_SURFACE_ADDRESS(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_006918_D2GRPH_SECONDARY_SURFACE_ADDRESS 0x00000000 + + +#define R_000001_MC_FB_LOCATION 0x000001 +#define S_000001_MC_FB_START(x) (((x) & 0xFFFF) << 0) +#define G_000001_MC_FB_START(x) (((x) >> 0) & 0xFFFF) +#define C_000001_MC_FB_START 0xFFFF0000 +#define S_000001_MC_FB_TOP(x) (((x) & 0xFFFF) << 16) +#define G_000001_MC_FB_TOP(x) (((x) >> 16) & 0xFFFF) +#define C_000001_MC_FB_TOP 0x0000FFFF +#define R_000002_MC_AGP_LOCATION 0x000002 +#define S_000002_MC_AGP_START(x) (((x) & 0xFFFF) << 0) +#define G_000002_MC_AGP_START(x) (((x) >> 0) & 0xFFFF) +#define C_000002_MC_AGP_START 0xFFFF0000 +#define S_000002_MC_AGP_TOP(x) (((x) & 0xFFFF) << 16) +#define G_000002_MC_AGP_TOP(x) (((x) >> 16) & 0xFFFF) +#define C_000002_MC_AGP_TOP 0x0000FFFF +#define R_000003_MC_AGP_BASE 0x000003 +#define S_000003_AGP_BASE_ADDR(x) (((x) & 0xFFFFFFFF) << 0) +#define G_000003_AGP_BASE_ADDR(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_000003_AGP_BASE_ADDR 0x00000000 +#define R_000004_MC_AGP_BASE_2 0x000004 +#define S_000004_AGP_BASE_ADDR_2(x) (((x) & 0xF) << 0) +#define G_000004_AGP_BASE_ADDR_2(x) (((x) >> 0) & 0xF) +#define C_000004_AGP_BASE_ADDR_2 0xFFFFFFF0 + + +#define R_00000F_CP_DYN_CNTL 0x00000F +#define S_00000F_CP_FORCEON(x) (((x) & 0x1) << 0) +#define G_00000F_CP_FORCEON(x) (((x) >> 0) & 0x1) +#define C_00000F_CP_FORCEON 0xFFFFFFFE +#define S_00000F_CP_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 1) +#define G_00000F_CP_MAX_DYN_STOP_LAT(x) (((x) >> 1) & 0x1) +#define C_00000F_CP_MAX_DYN_STOP_LAT 0xFFFFFFFD +#define S_00000F_CP_CLOCK_STATUS(x) (((x) & 0x1) << 2) +#define G_00000F_CP_CLOCK_STATUS(x) (((x) >> 2) & 0x1) +#define C_00000F_CP_CLOCK_STATUS 0xFFFFFFFB +#define S_00000F_CP_PROG_SHUTOFF(x) (((x) & 0x1) << 3) +#define G_00000F_CP_PROG_SHUTOFF(x) (((x) >> 3) & 0x1) +#define C_00000F_CP_PROG_SHUTOFF 0xFFFFFFF7 +#define S_00000F_CP_PROG_DELAY_VALUE(x) (((x) & 0xFF) << 4) +#define G_00000F_CP_PROG_DELAY_VALUE(x) (((x) >> 4) & 0xFF) +#define C_00000F_CP_PROG_DELAY_VALUE 0xFFFFF00F +#define S_00000F_CP_LOWER_POWER_IDLE(x) (((x) & 0xFF) << 12) +#define G_00000F_CP_LOWER_POWER_IDLE(x) (((x) >> 12) & 0xFF) +#define C_00000F_CP_LOWER_POWER_IDLE 0xFFF00FFF +#define S_00000F_CP_LOWER_POWER_IGNORE(x) (((x) & 0x1) << 20) +#define G_00000F_CP_LOWER_POWER_IGNORE(x) (((x) >> 20) & 0x1) +#define C_00000F_CP_LOWER_POWER_IGNORE 0xFFEFFFFF +#define S_00000F_CP_NORMAL_POWER_IGNORE(x) (((x) & 0x1) << 21) +#define G_00000F_CP_NORMAL_POWER_IGNORE(x) (((x) >> 21) & 0x1) +#define C_00000F_CP_NORMAL_POWER_IGNORE 0xFFDFFFFF +#define S_00000F_SPARE(x) (((x) & 0x3) << 22) +#define G_00000F_SPARE(x) (((x) >> 22) & 0x3) +#define C_00000F_SPARE 0xFF3FFFFF +#define S_00000F_CP_NORMAL_POWER_BUSY(x) (((x) & 0xFF) << 24) +#define G_00000F_CP_NORMAL_POWER_BUSY(x) (((x) >> 24) & 0xFF) +#define C_00000F_CP_NORMAL_POWER_BUSY 0x00FFFFFF +#define R_000011_E2_DYN_CNTL 0x000011 +#define S_000011_E2_FORCEON(x) (((x) & 0x1) << 0) +#define G_000011_E2_FORCEON(x) (((x) >> 0) & 0x1) +#define C_000011_E2_FORCEON 0xFFFFFFFE +#define S_000011_E2_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 1) +#define G_000011_E2_MAX_DYN_STOP_LAT(x) (((x) >> 1) & 0x1) +#define C_000011_E2_MAX_DYN_STOP_LAT 0xFFFFFFFD +#define S_000011_E2_CLOCK_STATUS(x) (((x) & 0x1) << 2) +#define G_000011_E2_CLOCK_STATUS(x) (((x) >> 2) & 0x1) +#define C_000011_E2_CLOCK_STATUS 0xFFFFFFFB +#define S_000011_E2_PROG_SHUTOFF(x) (((x) & 0x1) << 3) +#define G_000011_E2_PROG_SHUTOFF(x) (((x) >> 3) & 0x1) +#define C_000011_E2_PROG_SHUTOFF 0xFFFFFFF7 +#define S_000011_E2_PROG_DELAY_VALUE(x) (((x) & 0xFF) << 4) +#define G_000011_E2_PROG_DELAY_VALUE(x) (((x) >> 4) & 0xFF) +#define C_000011_E2_PROG_DELAY_VALUE 0xFFFFF00F +#define S_000011_E2_LOWER_POWER_IDLE(x) (((x) & 0xFF) << 12) +#define G_000011_E2_LOWER_POWER_IDLE(x) (((x) >> 12) & 0xFF) +#define C_000011_E2_LOWER_POWER_IDLE 0xFFF00FFF +#define S_000011_E2_LOWER_POWER_IGNORE(x) (((x) & 0x1) << 20) +#define G_000011_E2_LOWER_POWER_IGNORE(x) (((x) >> 20) & 0x1) +#define C_000011_E2_LOWER_POWER_IGNORE 0xFFEFFFFF +#define S_000011_E2_NORMAL_POWER_IGNORE(x) (((x) & 0x1) << 21) +#define G_000011_E2_NORMAL_POWER_IGNORE(x) (((x) >> 21) & 0x1) +#define C_000011_E2_NORMAL_POWER_IGNORE 0xFFDFFFFF +#define S_000011_SPARE(x) (((x) & 0x3) << 22) +#define G_000011_SPARE(x) (((x) >> 22) & 0x3) +#define C_000011_SPARE 0xFF3FFFFF +#define S_000011_E2_NORMAL_POWER_BUSY(x) (((x) & 0xFF) << 24) +#define G_000011_E2_NORMAL_POWER_BUSY(x) (((x) >> 24) & 0xFF) +#define C_000011_E2_NORMAL_POWER_BUSY 0x00FFFFFF +#define R_000013_IDCT_DYN_CNTL 0x000013 +#define S_000013_IDCT_FORCEON(x) (((x) & 0x1) << 0) +#define G_000013_IDCT_FORCEON(x) (((x) >> 0) & 0x1) +#define C_000013_IDCT_FORCEON 0xFFFFFFFE +#define S_000013_IDCT_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 1) +#define G_000013_IDCT_MAX_DYN_STOP_LAT(x) (((x) >> 1) & 0x1) +#define C_000013_IDCT_MAX_DYN_STOP_LAT 0xFFFFFFFD +#define S_000013_IDCT_CLOCK_STATUS(x) (((x) & 0x1) << 2) +#define G_000013_IDCT_CLOCK_STATUS(x) (((x) >> 2) & 0x1) +#define C_000013_IDCT_CLOCK_STATUS 0xFFFFFFFB +#define S_000013_IDCT_PROG_SHUTOFF(x) (((x) & 0x1) << 3) +#define G_000013_IDCT_PROG_SHUTOFF(x) (((x) >> 3) & 0x1) +#define C_000013_IDCT_PROG_SHUTOFF 0xFFFFFFF7 +#define S_000013_IDCT_PROG_DELAY_VALUE(x) (((x) & 0xFF) << 4) +#define G_000013_IDCT_PROG_DELAY_VALUE(x) (((x) >> 4) & 0xFF) +#define C_000013_IDCT_PROG_DELAY_VALUE 0xFFFFF00F +#define S_000013_IDCT_LOWER_POWER_IDLE(x) (((x) & 0xFF) << 12) +#define G_000013_IDCT_LOWER_POWER_IDLE(x) (((x) >> 12) & 0xFF) +#define C_000013_IDCT_LOWER_POWER_IDLE 0xFFF00FFF +#define S_000013_IDCT_LOWER_POWER_IGNORE(x) (((x) & 0x1) << 20) +#define G_000013_IDCT_LOWER_POWER_IGNORE(x) (((x) >> 20) & 0x1) +#define C_000013_IDCT_LOWER_POWER_IGNORE 0xFFEFFFFF +#define S_000013_IDCT_NORMAL_POWER_IGNORE(x) (((x) & 0x1) << 21) +#define G_000013_IDCT_NORMAL_POWER_IGNORE(x) (((x) >> 21) & 0x1) +#define C_000013_IDCT_NORMAL_POWER_IGNORE 0xFFDFFFFF +#define S_000013_SPARE(x) (((x) & 0x3) << 22) +#define G_000013_SPARE(x) (((x) >> 22) & 0x3) +#define C_000013_SPARE 0xFF3FFFFF +#define S_000013_IDCT_NORMAL_POWER_BUSY(x) (((x) & 0xFF) << 24) +#define G_000013_IDCT_NORMAL_POWER_BUSY(x) (((x) >> 24) & 0xFF) +#define C_000013_IDCT_NORMAL_POWER_BUSY 0x00FFFFFF + +#endif diff --git a/drivers/gpu/drm/radeon/rv770.c b/drivers/gpu/drm/radeon/rv770.c new file mode 100644 index 00000000..51d20aa6 --- /dev/null +++ b/drivers/gpu/drm/radeon/rv770.c @@ -0,0 +1,1413 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#include +#include +#include +#include "drmP.h" +#include "radeon.h" +#include "radeon_asic.h" +#include "radeon_drm.h" +#include "rv770d.h" +#include "atom.h" +#include "avivod.h" + +#define R700_PFP_UCODE_SIZE 848 +#define R700_PM4_UCODE_SIZE 1360 + +static void rv770_gpu_init(struct radeon_device *rdev); +void rv770_fini(struct radeon_device *rdev); +static void rv770_pcie_gen2_enable(struct radeon_device *rdev); + +u32 rv770_page_flip(struct radeon_device *rdev, int crtc_id, u64 crtc_base) +{ + struct radeon_crtc *radeon_crtc = rdev->mode_info.crtcs[crtc_id]; + u32 tmp = RREG32(AVIVO_D1GRPH_UPDATE + radeon_crtc->crtc_offset); + int i; + + /* Lock the graphics update lock */ + tmp |= AVIVO_D1GRPH_UPDATE_LOCK; + WREG32(AVIVO_D1GRPH_UPDATE + radeon_crtc->crtc_offset, tmp); + + /* update the scanout addresses */ + if (radeon_crtc->crtc_id) { + WREG32(D2GRPH_SECONDARY_SURFACE_ADDRESS_HIGH, upper_32_bits(crtc_base)); + WREG32(D2GRPH_PRIMARY_SURFACE_ADDRESS_HIGH, upper_32_bits(crtc_base)); + } else { + WREG32(D1GRPH_SECONDARY_SURFACE_ADDRESS_HIGH, upper_32_bits(crtc_base)); + WREG32(D1GRPH_PRIMARY_SURFACE_ADDRESS_HIGH, upper_32_bits(crtc_base)); + } + WREG32(D1GRPH_SECONDARY_SURFACE_ADDRESS + radeon_crtc->crtc_offset, + (u32)crtc_base); + WREG32(D1GRPH_PRIMARY_SURFACE_ADDRESS + radeon_crtc->crtc_offset, + (u32)crtc_base); + + /* Wait for update_pending to go high. */ + for (i = 0; i < rdev->usec_timeout; i++) { + if (RREG32(AVIVO_D1GRPH_UPDATE + radeon_crtc->crtc_offset) & AVIVO_D1GRPH_SURFACE_UPDATE_PENDING) + break; + udelay(1); + } + DRM_DEBUG("Update pending now high. Unlocking vupdate_lock.\n"); + + /* Unlock the lock, so double-buffering can take place inside vblank */ + tmp &= ~AVIVO_D1GRPH_UPDATE_LOCK; + WREG32(AVIVO_D1GRPH_UPDATE + radeon_crtc->crtc_offset, tmp); + + /* Return current update_pending status: */ + return RREG32(AVIVO_D1GRPH_UPDATE + radeon_crtc->crtc_offset) & AVIVO_D1GRPH_SURFACE_UPDATE_PENDING; +} + +/* get temperature in millidegrees */ +int rv770_get_temp(struct radeon_device *rdev) +{ + u32 temp = (RREG32(CG_MULT_THERMAL_STATUS) & ASIC_T_MASK) >> + ASIC_T_SHIFT; + int actual_temp; + + if (temp & 0x400) + actual_temp = -256; + else if (temp & 0x200) + actual_temp = 255; + else if (temp & 0x100) { + actual_temp = temp & 0x1ff; + actual_temp |= ~0x1ff; + } else + actual_temp = temp & 0xff; + + return (actual_temp * 1000) / 2; +} + +void rv770_pm_misc(struct radeon_device *rdev) +{ + int req_ps_idx = rdev->pm.requested_power_state_index; + int req_cm_idx = rdev->pm.requested_clock_mode_index; + struct radeon_power_state *ps = &rdev->pm.power_state[req_ps_idx]; + struct radeon_voltage *voltage = &ps->clock_info[req_cm_idx].voltage; + + if ((voltage->type == VOLTAGE_SW) && voltage->voltage) { + /* 0xff01 is a flag rather then an actual voltage */ + if (voltage->voltage == 0xff01) + return; + if (voltage->voltage != rdev->pm.current_vddc) { + radeon_atom_set_voltage(rdev, voltage->voltage, SET_VOLTAGE_TYPE_ASIC_VDDC); + rdev->pm.current_vddc = voltage->voltage; + DRM_DEBUG("Setting: v: %d\n", voltage->voltage); + } + } +} + +/* + * GART + */ +int rv770_pcie_gart_enable(struct radeon_device *rdev) +{ + u32 tmp; + int r, i; + + if (rdev->gart.table.vram.robj == NULL) { + dev_err(rdev->dev, "No VRAM object for PCIE GART.\n"); + return -EINVAL; + } + r = radeon_gart_table_vram_pin(rdev); + if (r) + return r; + radeon_gart_restore(rdev); + /* Setup L2 cache */ + WREG32(VM_L2_CNTL, ENABLE_L2_CACHE | ENABLE_L2_FRAGMENT_PROCESSING | + ENABLE_L2_PTE_CACHE_LRU_UPDATE_BY_WRITE | + EFFECTIVE_L2_QUEUE_SIZE(7)); + WREG32(VM_L2_CNTL2, 0); + WREG32(VM_L2_CNTL3, BANK_SELECT(0) | CACHE_UPDATE_MODE(2)); + /* Setup TLB control */ + tmp = ENABLE_L1_TLB | ENABLE_L1_FRAGMENT_PROCESSING | + SYSTEM_ACCESS_MODE_NOT_IN_SYS | + SYSTEM_APERTURE_UNMAPPED_ACCESS_PASS_THRU | + EFFECTIVE_L1_TLB_SIZE(5) | EFFECTIVE_L1_QUEUE_SIZE(5); + WREG32(MC_VM_MD_L1_TLB0_CNTL, tmp); + WREG32(MC_VM_MD_L1_TLB1_CNTL, tmp); + WREG32(MC_VM_MD_L1_TLB2_CNTL, tmp); + if (rdev->family == CHIP_RV740) + WREG32(MC_VM_MD_L1_TLB3_CNTL, tmp); + WREG32(MC_VM_MB_L1_TLB0_CNTL, tmp); + WREG32(MC_VM_MB_L1_TLB1_CNTL, tmp); + WREG32(MC_VM_MB_L1_TLB2_CNTL, tmp); + WREG32(MC_VM_MB_L1_TLB3_CNTL, tmp); + WREG32(VM_CONTEXT0_PAGE_TABLE_START_ADDR, rdev->mc.gtt_start >> 12); + WREG32(VM_CONTEXT0_PAGE_TABLE_END_ADDR, rdev->mc.gtt_end >> 12); + WREG32(VM_CONTEXT0_PAGE_TABLE_BASE_ADDR, rdev->gart.table_addr >> 12); + WREG32(VM_CONTEXT0_CNTL, ENABLE_CONTEXT | PAGE_TABLE_DEPTH(0) | + RANGE_PROTECTION_FAULT_ENABLE_DEFAULT); + WREG32(VM_CONTEXT0_PROTECTION_FAULT_DEFAULT_ADDR, + (u32)(rdev->dummy_page.addr >> 12)); + for (i = 1; i < 7; i++) + WREG32(VM_CONTEXT0_CNTL + (i * 4), 0); + + r600_pcie_gart_tlb_flush(rdev); + rdev->gart.ready = true; + return 0; +} + +void rv770_pcie_gart_disable(struct radeon_device *rdev) +{ + u32 tmp; + int i, r; + + /* Disable all tables */ + for (i = 0; i < 7; i++) + WREG32(VM_CONTEXT0_CNTL + (i * 4), 0); + + /* Setup L2 cache */ + WREG32(VM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING | + EFFECTIVE_L2_QUEUE_SIZE(7)); + WREG32(VM_L2_CNTL2, 0); + WREG32(VM_L2_CNTL3, BANK_SELECT(0) | CACHE_UPDATE_MODE(2)); + /* Setup TLB control */ + tmp = EFFECTIVE_L1_TLB_SIZE(5) | EFFECTIVE_L1_QUEUE_SIZE(5); + WREG32(MC_VM_MD_L1_TLB0_CNTL, tmp); + WREG32(MC_VM_MD_L1_TLB1_CNTL, tmp); + WREG32(MC_VM_MD_L1_TLB2_CNTL, tmp); + WREG32(MC_VM_MB_L1_TLB0_CNTL, tmp); + WREG32(MC_VM_MB_L1_TLB1_CNTL, tmp); + WREG32(MC_VM_MB_L1_TLB2_CNTL, tmp); + WREG32(MC_VM_MB_L1_TLB3_CNTL, tmp); + if (rdev->gart.table.vram.robj) { + r = radeon_bo_reserve(rdev->gart.table.vram.robj, false); + if (likely(r == 0)) { + radeon_bo_kunmap(rdev->gart.table.vram.robj); + radeon_bo_unpin(rdev->gart.table.vram.robj); + radeon_bo_unreserve(rdev->gart.table.vram.robj); + } + } +} + +void rv770_pcie_gart_fini(struct radeon_device *rdev) +{ + radeon_gart_fini(rdev); + rv770_pcie_gart_disable(rdev); + radeon_gart_table_vram_free(rdev); +} + + +void rv770_agp_enable(struct radeon_device *rdev) +{ + u32 tmp; + int i; + + /* Setup L2 cache */ + WREG32(VM_L2_CNTL, ENABLE_L2_CACHE | ENABLE_L2_FRAGMENT_PROCESSING | + ENABLE_L2_PTE_CACHE_LRU_UPDATE_BY_WRITE | + EFFECTIVE_L2_QUEUE_SIZE(7)); + WREG32(VM_L2_CNTL2, 0); + WREG32(VM_L2_CNTL3, BANK_SELECT(0) | CACHE_UPDATE_MODE(2)); + /* Setup TLB control */ + tmp = ENABLE_L1_TLB | ENABLE_L1_FRAGMENT_PROCESSING | + SYSTEM_ACCESS_MODE_NOT_IN_SYS | + SYSTEM_APERTURE_UNMAPPED_ACCESS_PASS_THRU | + EFFECTIVE_L1_TLB_SIZE(5) | EFFECTIVE_L1_QUEUE_SIZE(5); + WREG32(MC_VM_MD_L1_TLB0_CNTL, tmp); + WREG32(MC_VM_MD_L1_TLB1_CNTL, tmp); + WREG32(MC_VM_MD_L1_TLB2_CNTL, tmp); + WREG32(MC_VM_MB_L1_TLB0_CNTL, tmp); + WREG32(MC_VM_MB_L1_TLB1_CNTL, tmp); + WREG32(MC_VM_MB_L1_TLB2_CNTL, tmp); + WREG32(MC_VM_MB_L1_TLB3_CNTL, tmp); + for (i = 0; i < 7; i++) + WREG32(VM_CONTEXT0_CNTL + (i * 4), 0); +} + +static void rv770_mc_program(struct radeon_device *rdev) +{ + struct rv515_mc_save save; + u32 tmp; + int i, j; + + /* Initialize HDP */ + for (i = 0, j = 0; i < 32; i++, j += 0x18) { + WREG32((0x2c14 + j), 0x00000000); + WREG32((0x2c18 + j), 0x00000000); + WREG32((0x2c1c + j), 0x00000000); + WREG32((0x2c20 + j), 0x00000000); + WREG32((0x2c24 + j), 0x00000000); + } + /* r7xx hw bug. Read from HDP_DEBUG1 rather + * than writing to HDP_REG_COHERENCY_FLUSH_CNTL + */ + tmp = RREG32(HDP_DEBUG1); + + rv515_mc_stop(rdev, &save); + if (r600_mc_wait_for_idle(rdev)) { + dev_warn(rdev->dev, "Wait for MC idle timedout !\n"); + } + /* Lockout access through VGA aperture*/ + WREG32(VGA_HDP_CONTROL, VGA_MEMORY_DISABLE); + /* Update configuration */ + if (rdev->flags & RADEON_IS_AGP) { + if (rdev->mc.vram_start < rdev->mc.gtt_start) { + /* VRAM before AGP */ + WREG32(MC_VM_SYSTEM_APERTURE_LOW_ADDR, + rdev->mc.vram_start >> 12); + WREG32(MC_VM_SYSTEM_APERTURE_HIGH_ADDR, + rdev->mc.gtt_end >> 12); + } else { + /* VRAM after AGP */ + WREG32(MC_VM_SYSTEM_APERTURE_LOW_ADDR, + rdev->mc.gtt_start >> 12); + WREG32(MC_VM_SYSTEM_APERTURE_HIGH_ADDR, + rdev->mc.vram_end >> 12); + } + } else { + WREG32(MC_VM_SYSTEM_APERTURE_LOW_ADDR, + rdev->mc.vram_start >> 12); + WREG32(MC_VM_SYSTEM_APERTURE_HIGH_ADDR, + rdev->mc.vram_end >> 12); + } + WREG32(MC_VM_SYSTEM_APERTURE_DEFAULT_ADDR, 0); + tmp = ((rdev->mc.vram_end >> 24) & 0xFFFF) << 16; + tmp |= ((rdev->mc.vram_start >> 24) & 0xFFFF); + WREG32(MC_VM_FB_LOCATION, tmp); + WREG32(HDP_NONSURFACE_BASE, (rdev->mc.vram_start >> 8)); + WREG32(HDP_NONSURFACE_INFO, (2 << 7)); + WREG32(HDP_NONSURFACE_SIZE, 0x3FFFFFFF); + if (rdev->flags & RADEON_IS_AGP) { + WREG32(MC_VM_AGP_TOP, rdev->mc.gtt_end >> 16); + WREG32(MC_VM_AGP_BOT, rdev->mc.gtt_start >> 16); + WREG32(MC_VM_AGP_BASE, rdev->mc.agp_base >> 22); + } else { + WREG32(MC_VM_AGP_BASE, 0); + WREG32(MC_VM_AGP_TOP, 0x0FFFFFFF); + WREG32(MC_VM_AGP_BOT, 0x0FFFFFFF); + } + if (r600_mc_wait_for_idle(rdev)) { + dev_warn(rdev->dev, "Wait for MC idle timedout !\n"); + } + rv515_mc_resume(rdev, &save); + /* we need to own VRAM, so turn off the VGA renderer here + * to stop it overwriting our objects */ + rv515_vga_render_disable(rdev); +} + + +/* + * CP. + */ +void r700_cp_stop(struct radeon_device *rdev) +{ + radeon_ttm_set_active_vram_size(rdev, rdev->mc.visible_vram_size); + WREG32(CP_ME_CNTL, (CP_ME_HALT | CP_PFP_HALT)); + WREG32(SCRATCH_UMSK, 0); +} + +static int rv770_cp_load_microcode(struct radeon_device *rdev) +{ + const __be32 *fw_data; + int i; + + if (!rdev->me_fw || !rdev->pfp_fw) + return -EINVAL; + + r700_cp_stop(rdev); + WREG32(CP_RB_CNTL, +#ifdef __BIG_ENDIAN + BUF_SWAP_32BIT | +#endif + RB_NO_UPDATE | RB_BLKSZ(15) | RB_BUFSZ(3)); + + /* Reset cp */ + WREG32(GRBM_SOFT_RESET, SOFT_RESET_CP); + RREG32(GRBM_SOFT_RESET); + mdelay(15); + WREG32(GRBM_SOFT_RESET, 0); + + fw_data = (const __be32 *)rdev->pfp_fw->data; + WREG32(CP_PFP_UCODE_ADDR, 0); + for (i = 0; i < R700_PFP_UCODE_SIZE; i++) + WREG32(CP_PFP_UCODE_DATA, be32_to_cpup(fw_data++)); + WREG32(CP_PFP_UCODE_ADDR, 0); + + fw_data = (const __be32 *)rdev->me_fw->data; + WREG32(CP_ME_RAM_WADDR, 0); + for (i = 0; i < R700_PM4_UCODE_SIZE; i++) + WREG32(CP_ME_RAM_DATA, be32_to_cpup(fw_data++)); + + WREG32(CP_PFP_UCODE_ADDR, 0); + WREG32(CP_ME_RAM_WADDR, 0); + WREG32(CP_ME_RAM_RADDR, 0); + return 0; +} + +void r700_cp_fini(struct radeon_device *rdev) +{ + r700_cp_stop(rdev); + radeon_ring_fini(rdev); +} + +/* + * Core functions + */ +static u32 r700_get_tile_pipe_to_backend_map(struct radeon_device *rdev, + u32 num_tile_pipes, + u32 num_backends, + u32 backend_disable_mask) +{ + u32 backend_map = 0; + u32 enabled_backends_mask; + u32 enabled_backends_count; + u32 cur_pipe; + u32 swizzle_pipe[R7XX_MAX_PIPES]; + u32 cur_backend; + u32 i; + bool force_no_swizzle; + + if (num_tile_pipes > R7XX_MAX_PIPES) + num_tile_pipes = R7XX_MAX_PIPES; + if (num_tile_pipes < 1) + num_tile_pipes = 1; + if (num_backends > R7XX_MAX_BACKENDS) + num_backends = R7XX_MAX_BACKENDS; + if (num_backends < 1) + num_backends = 1; + + enabled_backends_mask = 0; + enabled_backends_count = 0; + for (i = 0; i < R7XX_MAX_BACKENDS; ++i) { + if (((backend_disable_mask >> i) & 1) == 0) { + enabled_backends_mask |= (1 << i); + ++enabled_backends_count; + } + if (enabled_backends_count == num_backends) + break; + } + + if (enabled_backends_count == 0) { + enabled_backends_mask = 1; + enabled_backends_count = 1; + } + + if (enabled_backends_count != num_backends) + num_backends = enabled_backends_count; + + switch (rdev->family) { + case CHIP_RV770: + case CHIP_RV730: + force_no_swizzle = false; + break; + case CHIP_RV710: + case CHIP_RV740: + default: + force_no_swizzle = true; + break; + } + + memset((uint8_t *)&swizzle_pipe[0], 0, sizeof(u32) * R7XX_MAX_PIPES); + switch (num_tile_pipes) { + case 1: + swizzle_pipe[0] = 0; + break; + case 2: + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 1; + break; + case 3: + if (force_no_swizzle) { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 1; + swizzle_pipe[2] = 2; + } else { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 2; + swizzle_pipe[2] = 1; + } + break; + case 4: + if (force_no_swizzle) { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 1; + swizzle_pipe[2] = 2; + swizzle_pipe[3] = 3; + } else { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 2; + swizzle_pipe[2] = 3; + swizzle_pipe[3] = 1; + } + break; + case 5: + if (force_no_swizzle) { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 1; + swizzle_pipe[2] = 2; + swizzle_pipe[3] = 3; + swizzle_pipe[4] = 4; + } else { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 2; + swizzle_pipe[2] = 4; + swizzle_pipe[3] = 1; + swizzle_pipe[4] = 3; + } + break; + case 6: + if (force_no_swizzle) { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 1; + swizzle_pipe[2] = 2; + swizzle_pipe[3] = 3; + swizzle_pipe[4] = 4; + swizzle_pipe[5] = 5; + } else { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 2; + swizzle_pipe[2] = 4; + swizzle_pipe[3] = 5; + swizzle_pipe[4] = 3; + swizzle_pipe[5] = 1; + } + break; + case 7: + if (force_no_swizzle) { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 1; + swizzle_pipe[2] = 2; + swizzle_pipe[3] = 3; + swizzle_pipe[4] = 4; + swizzle_pipe[5] = 5; + swizzle_pipe[6] = 6; + } else { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 2; + swizzle_pipe[2] = 4; + swizzle_pipe[3] = 6; + swizzle_pipe[4] = 3; + swizzle_pipe[5] = 1; + swizzle_pipe[6] = 5; + } + break; + case 8: + if (force_no_swizzle) { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 1; + swizzle_pipe[2] = 2; + swizzle_pipe[3] = 3; + swizzle_pipe[4] = 4; + swizzle_pipe[5] = 5; + swizzle_pipe[6] = 6; + swizzle_pipe[7] = 7; + } else { + swizzle_pipe[0] = 0; + swizzle_pipe[1] = 2; + swizzle_pipe[2] = 4; + swizzle_pipe[3] = 6; + swizzle_pipe[4] = 3; + swizzle_pipe[5] = 1; + swizzle_pipe[6] = 7; + swizzle_pipe[7] = 5; + } + break; + } + + cur_backend = 0; + for (cur_pipe = 0; cur_pipe < num_tile_pipes; ++cur_pipe) { + while (((1 << cur_backend) & enabled_backends_mask) == 0) + cur_backend = (cur_backend + 1) % R7XX_MAX_BACKENDS; + + backend_map |= (u32)(((cur_backend & 3) << (swizzle_pipe[cur_pipe] * 2))); + + cur_backend = (cur_backend + 1) % R7XX_MAX_BACKENDS; + } + + return backend_map; +} + +static void rv770_gpu_init(struct radeon_device *rdev) +{ + int i, j, num_qd_pipes; + u32 ta_aux_cntl; + u32 sx_debug_1; + u32 smx_dc_ctl0; + u32 db_debug3; + u32 num_gs_verts_per_thread; + u32 vgt_gs_per_es; + u32 gs_prim_buffer_depth = 0; + u32 sq_ms_fifo_sizes; + u32 sq_config; + u32 sq_thread_resource_mgmt; + u32 hdp_host_path_cntl; + u32 sq_dyn_gpr_size_simd_ab_0; + u32 backend_map; + u32 gb_tiling_config = 0; + u32 cc_rb_backend_disable = 0; + u32 cc_gc_shader_pipe_config = 0; + u32 mc_arb_ramcfg; + u32 db_debug4; + + /* setup chip specs */ + switch (rdev->family) { + case CHIP_RV770: + rdev->config.rv770.max_pipes = 4; + rdev->config.rv770.max_tile_pipes = 8; + rdev->config.rv770.max_simds = 10; + rdev->config.rv770.max_backends = 4; + rdev->config.rv770.max_gprs = 256; + rdev->config.rv770.max_threads = 248; + rdev->config.rv770.max_stack_entries = 512; + rdev->config.rv770.max_hw_contexts = 8; + rdev->config.rv770.max_gs_threads = 16 * 2; + rdev->config.rv770.sx_max_export_size = 128; + rdev->config.rv770.sx_max_export_pos_size = 16; + rdev->config.rv770.sx_max_export_smx_size = 112; + rdev->config.rv770.sq_num_cf_insts = 2; + + rdev->config.rv770.sx_num_of_sets = 7; + rdev->config.rv770.sc_prim_fifo_size = 0xF9; + rdev->config.rv770.sc_hiz_tile_fifo_size = 0x30; + rdev->config.rv770.sc_earlyz_tile_fifo_fize = 0x130; + break; + case CHIP_RV730: + rdev->config.rv770.max_pipes = 2; + rdev->config.rv770.max_tile_pipes = 4; + rdev->config.rv770.max_simds = 8; + rdev->config.rv770.max_backends = 2; + rdev->config.rv770.max_gprs = 128; + rdev->config.rv770.max_threads = 248; + rdev->config.rv770.max_stack_entries = 256; + rdev->config.rv770.max_hw_contexts = 8; + rdev->config.rv770.max_gs_threads = 16 * 2; + rdev->config.rv770.sx_max_export_size = 256; + rdev->config.rv770.sx_max_export_pos_size = 32; + rdev->config.rv770.sx_max_export_smx_size = 224; + rdev->config.rv770.sq_num_cf_insts = 2; + + rdev->config.rv770.sx_num_of_sets = 7; + rdev->config.rv770.sc_prim_fifo_size = 0xf9; + rdev->config.rv770.sc_hiz_tile_fifo_size = 0x30; + rdev->config.rv770.sc_earlyz_tile_fifo_fize = 0x130; + if (rdev->config.rv770.sx_max_export_pos_size > 16) { + rdev->config.rv770.sx_max_export_pos_size -= 16; + rdev->config.rv770.sx_max_export_smx_size += 16; + } + break; + case CHIP_RV710: + rdev->config.rv770.max_pipes = 2; + rdev->config.rv770.max_tile_pipes = 2; + rdev->config.rv770.max_simds = 2; + rdev->config.rv770.max_backends = 1; + rdev->config.rv770.max_gprs = 256; + rdev->config.rv770.max_threads = 192; + rdev->config.rv770.max_stack_entries = 256; + rdev->config.rv770.max_hw_contexts = 4; + rdev->config.rv770.max_gs_threads = 8 * 2; + rdev->config.rv770.sx_max_export_size = 128; + rdev->config.rv770.sx_max_export_pos_size = 16; + rdev->config.rv770.sx_max_export_smx_size = 112; + rdev->config.rv770.sq_num_cf_insts = 1; + + rdev->config.rv770.sx_num_of_sets = 7; + rdev->config.rv770.sc_prim_fifo_size = 0x40; + rdev->config.rv770.sc_hiz_tile_fifo_size = 0x30; + rdev->config.rv770.sc_earlyz_tile_fifo_fize = 0x130; + break; + case CHIP_RV740: + rdev->config.rv770.max_pipes = 4; + rdev->config.rv770.max_tile_pipes = 4; + rdev->config.rv770.max_simds = 8; + rdev->config.rv770.max_backends = 4; + rdev->config.rv770.max_gprs = 256; + rdev->config.rv770.max_threads = 248; + rdev->config.rv770.max_stack_entries = 512; + rdev->config.rv770.max_hw_contexts = 8; + rdev->config.rv770.max_gs_threads = 16 * 2; + rdev->config.rv770.sx_max_export_size = 256; + rdev->config.rv770.sx_max_export_pos_size = 32; + rdev->config.rv770.sx_max_export_smx_size = 224; + rdev->config.rv770.sq_num_cf_insts = 2; + + rdev->config.rv770.sx_num_of_sets = 7; + rdev->config.rv770.sc_prim_fifo_size = 0x100; + rdev->config.rv770.sc_hiz_tile_fifo_size = 0x30; + rdev->config.rv770.sc_earlyz_tile_fifo_fize = 0x130; + + if (rdev->config.rv770.sx_max_export_pos_size > 16) { + rdev->config.rv770.sx_max_export_pos_size -= 16; + rdev->config.rv770.sx_max_export_smx_size += 16; + } + break; + default: + break; + } + + /* Initialize HDP */ + j = 0; + for (i = 0; i < 32; i++) { + WREG32((0x2c14 + j), 0x00000000); + WREG32((0x2c18 + j), 0x00000000); + WREG32((0x2c1c + j), 0x00000000); + WREG32((0x2c20 + j), 0x00000000); + WREG32((0x2c24 + j), 0x00000000); + j += 0x18; + } + + WREG32(GRBM_CNTL, GRBM_READ_TIMEOUT(0xff)); + + /* setup tiling, simd, pipe config */ + mc_arb_ramcfg = RREG32(MC_ARB_RAMCFG); + + switch (rdev->config.rv770.max_tile_pipes) { + case 1: + default: + gb_tiling_config |= PIPE_TILING(0); + break; + case 2: + gb_tiling_config |= PIPE_TILING(1); + break; + case 4: + gb_tiling_config |= PIPE_TILING(2); + break; + case 8: + gb_tiling_config |= PIPE_TILING(3); + break; + } + rdev->config.rv770.tiling_npipes = rdev->config.rv770.max_tile_pipes; + + if (rdev->family == CHIP_RV770) + gb_tiling_config |= BANK_TILING(1); + else + gb_tiling_config |= BANK_TILING((mc_arb_ramcfg & NOOFBANK_MASK) >> NOOFBANK_SHIFT); + rdev->config.rv770.tiling_nbanks = 4 << ((gb_tiling_config >> 4) & 0x3); + gb_tiling_config |= GROUP_SIZE((mc_arb_ramcfg & BURSTLENGTH_MASK) >> BURSTLENGTH_SHIFT); + if ((mc_arb_ramcfg & BURSTLENGTH_MASK) >> BURSTLENGTH_SHIFT) + rdev->config.rv770.tiling_group_size = 512; + else + rdev->config.rv770.tiling_group_size = 256; + if (((mc_arb_ramcfg & NOOFROWS_MASK) >> NOOFROWS_SHIFT) > 3) { + gb_tiling_config |= ROW_TILING(3); + gb_tiling_config |= SAMPLE_SPLIT(3); + } else { + gb_tiling_config |= + ROW_TILING(((mc_arb_ramcfg & NOOFROWS_MASK) >> NOOFROWS_SHIFT)); + gb_tiling_config |= + SAMPLE_SPLIT(((mc_arb_ramcfg & NOOFROWS_MASK) >> NOOFROWS_SHIFT)); + } + + gb_tiling_config |= BANK_SWAPS(1); + + cc_rb_backend_disable = RREG32(CC_RB_BACKEND_DISABLE) & 0x00ff0000; + cc_rb_backend_disable |= + BACKEND_DISABLE((R7XX_MAX_BACKENDS_MASK << rdev->config.rv770.max_backends) & R7XX_MAX_BACKENDS_MASK); + + cc_gc_shader_pipe_config = RREG32(CC_GC_SHADER_PIPE_CONFIG) & 0xffffff00; + cc_gc_shader_pipe_config |= + INACTIVE_QD_PIPES((R7XX_MAX_PIPES_MASK << rdev->config.rv770.max_pipes) & R7XX_MAX_PIPES_MASK); + cc_gc_shader_pipe_config |= + INACTIVE_SIMDS((R7XX_MAX_SIMDS_MASK << rdev->config.rv770.max_simds) & R7XX_MAX_SIMDS_MASK); + + if (rdev->family == CHIP_RV740) + backend_map = 0x28; + else + backend_map = r700_get_tile_pipe_to_backend_map(rdev, + rdev->config.rv770.max_tile_pipes, + (R7XX_MAX_BACKENDS - + r600_count_pipe_bits((cc_rb_backend_disable & + R7XX_MAX_BACKENDS_MASK) >> 16)), + (cc_rb_backend_disable >> 16)); + + rdev->config.rv770.tile_config = gb_tiling_config; + gb_tiling_config |= BACKEND_MAP(backend_map); + + WREG32(GB_TILING_CONFIG, gb_tiling_config); + WREG32(DCP_TILING_CONFIG, (gb_tiling_config & 0xffff)); + WREG32(HDP_TILING_CONFIG, (gb_tiling_config & 0xffff)); + + WREG32(CC_RB_BACKEND_DISABLE, cc_rb_backend_disable); + WREG32(CC_GC_SHADER_PIPE_CONFIG, cc_gc_shader_pipe_config); + WREG32(GC_USER_SHADER_PIPE_CONFIG, cc_gc_shader_pipe_config); + WREG32(CC_SYS_RB_BACKEND_DISABLE, cc_rb_backend_disable); + + WREG32(CGTS_SYS_TCC_DISABLE, 0); + WREG32(CGTS_TCC_DISABLE, 0); + WREG32(CGTS_USER_SYS_TCC_DISABLE, 0); + WREG32(CGTS_USER_TCC_DISABLE, 0); + + num_qd_pipes = + R7XX_MAX_PIPES - r600_count_pipe_bits((cc_gc_shader_pipe_config & INACTIVE_QD_PIPES_MASK) >> 8); + WREG32(VGT_OUT_DEALLOC_CNTL, (num_qd_pipes * 4) & DEALLOC_DIST_MASK); + WREG32(VGT_VERTEX_REUSE_BLOCK_CNTL, ((num_qd_pipes * 4) - 2) & VTX_REUSE_DEPTH_MASK); + + /* set HW defaults for 3D engine */ + WREG32(CP_QUEUE_THRESHOLDS, (ROQ_IB1_START(0x16) | + ROQ_IB2_START(0x2b))); + + WREG32(CP_MEQ_THRESHOLDS, STQ_SPLIT(0x30)); + + ta_aux_cntl = RREG32(TA_CNTL_AUX); + WREG32(TA_CNTL_AUX, ta_aux_cntl | DISABLE_CUBE_ANISO); + + sx_debug_1 = RREG32(SX_DEBUG_1); + sx_debug_1 |= ENABLE_NEW_SMX_ADDRESS; + WREG32(SX_DEBUG_1, sx_debug_1); + + smx_dc_ctl0 = RREG32(SMX_DC_CTL0); + smx_dc_ctl0 &= ~CACHE_DEPTH(0x1ff); + smx_dc_ctl0 |= CACHE_DEPTH((rdev->config.rv770.sx_num_of_sets * 64) - 1); + WREG32(SMX_DC_CTL0, smx_dc_ctl0); + + if (rdev->family != CHIP_RV740) + WREG32(SMX_EVENT_CTL, (ES_FLUSH_CTL(4) | + GS_FLUSH_CTL(4) | + ACK_FLUSH_CTL(3) | + SYNC_FLUSH_CTL)); + + db_debug3 = RREG32(DB_DEBUG3); + db_debug3 &= ~DB_CLK_OFF_DELAY(0x1f); + switch (rdev->family) { + case CHIP_RV770: + case CHIP_RV740: + db_debug3 |= DB_CLK_OFF_DELAY(0x1f); + break; + case CHIP_RV710: + case CHIP_RV730: + default: + db_debug3 |= DB_CLK_OFF_DELAY(2); + break; + } + WREG32(DB_DEBUG3, db_debug3); + + if (rdev->family != CHIP_RV770) { + db_debug4 = RREG32(DB_DEBUG4); + db_debug4 |= DISABLE_TILE_COVERED_FOR_PS_ITER; + WREG32(DB_DEBUG4, db_debug4); + } + + WREG32(SX_EXPORT_BUFFER_SIZES, (COLOR_BUFFER_SIZE((rdev->config.rv770.sx_max_export_size / 4) - 1) | + POSITION_BUFFER_SIZE((rdev->config.rv770.sx_max_export_pos_size / 4) - 1) | + SMX_BUFFER_SIZE((rdev->config.rv770.sx_max_export_smx_size / 4) - 1))); + + WREG32(PA_SC_FIFO_SIZE, (SC_PRIM_FIFO_SIZE(rdev->config.rv770.sc_prim_fifo_size) | + SC_HIZ_TILE_FIFO_SIZE(rdev->config.rv770.sc_hiz_tile_fifo_size) | + SC_EARLYZ_TILE_FIFO_SIZE(rdev->config.rv770.sc_earlyz_tile_fifo_fize))); + + WREG32(PA_SC_MULTI_CHIP_CNTL, 0); + + WREG32(VGT_NUM_INSTANCES, 1); + + WREG32(SPI_CONFIG_CNTL, GPR_WRITE_PRIORITY(0)); + + WREG32(SPI_CONFIG_CNTL_1, VTX_DONE_DELAY(4)); + + WREG32(CP_PERFMON_CNTL, 0); + + sq_ms_fifo_sizes = (CACHE_FIFO_SIZE(16 * rdev->config.rv770.sq_num_cf_insts) | + DONE_FIFO_HIWATER(0xe0) | + ALU_UPDATE_FIFO_HIWATER(0x8)); + switch (rdev->family) { + case CHIP_RV770: + case CHIP_RV730: + case CHIP_RV710: + sq_ms_fifo_sizes |= FETCH_FIFO_HIWATER(0x1); + break; + case CHIP_RV740: + default: + sq_ms_fifo_sizes |= FETCH_FIFO_HIWATER(0x4); + break; + } + WREG32(SQ_MS_FIFO_SIZES, sq_ms_fifo_sizes); + + /* SQ_CONFIG, SQ_GPR_RESOURCE_MGMT, SQ_THREAD_RESOURCE_MGMT, SQ_STACK_RESOURCE_MGMT + * should be adjusted as needed by the 2D/3D drivers. This just sets default values + */ + sq_config = RREG32(SQ_CONFIG); + sq_config &= ~(PS_PRIO(3) | + VS_PRIO(3) | + GS_PRIO(3) | + ES_PRIO(3)); + sq_config |= (DX9_CONSTS | + VC_ENABLE | + EXPORT_SRC_C | + PS_PRIO(0) | + VS_PRIO(1) | + GS_PRIO(2) | + ES_PRIO(3)); + if (rdev->family == CHIP_RV710) + /* no vertex cache */ + sq_config &= ~VC_ENABLE; + + WREG32(SQ_CONFIG, sq_config); + + WREG32(SQ_GPR_RESOURCE_MGMT_1, (NUM_PS_GPRS((rdev->config.rv770.max_gprs * 24)/64) | + NUM_VS_GPRS((rdev->config.rv770.max_gprs * 24)/64) | + NUM_CLAUSE_TEMP_GPRS(((rdev->config.rv770.max_gprs * 24)/64)/2))); + + WREG32(SQ_GPR_RESOURCE_MGMT_2, (NUM_GS_GPRS((rdev->config.rv770.max_gprs * 7)/64) | + NUM_ES_GPRS((rdev->config.rv770.max_gprs * 7)/64))); + + sq_thread_resource_mgmt = (NUM_PS_THREADS((rdev->config.rv770.max_threads * 4)/8) | + NUM_VS_THREADS((rdev->config.rv770.max_threads * 2)/8) | + NUM_ES_THREADS((rdev->config.rv770.max_threads * 1)/8)); + if (((rdev->config.rv770.max_threads * 1) / 8) > rdev->config.rv770.max_gs_threads) + sq_thread_resource_mgmt |= NUM_GS_THREADS(rdev->config.rv770.max_gs_threads); + else + sq_thread_resource_mgmt |= NUM_GS_THREADS((rdev->config.rv770.max_gs_threads * 1)/8); + WREG32(SQ_THREAD_RESOURCE_MGMT, sq_thread_resource_mgmt); + + WREG32(SQ_STACK_RESOURCE_MGMT_1, (NUM_PS_STACK_ENTRIES((rdev->config.rv770.max_stack_entries * 1)/4) | + NUM_VS_STACK_ENTRIES((rdev->config.rv770.max_stack_entries * 1)/4))); + + WREG32(SQ_STACK_RESOURCE_MGMT_2, (NUM_GS_STACK_ENTRIES((rdev->config.rv770.max_stack_entries * 1)/4) | + NUM_ES_STACK_ENTRIES((rdev->config.rv770.max_stack_entries * 1)/4))); + + sq_dyn_gpr_size_simd_ab_0 = (SIMDA_RING0((rdev->config.rv770.max_gprs * 38)/64) | + SIMDA_RING1((rdev->config.rv770.max_gprs * 38)/64) | + SIMDB_RING0((rdev->config.rv770.max_gprs * 38)/64) | + SIMDB_RING1((rdev->config.rv770.max_gprs * 38)/64)); + + WREG32(SQ_DYN_GPR_SIZE_SIMD_AB_0, sq_dyn_gpr_size_simd_ab_0); + WREG32(SQ_DYN_GPR_SIZE_SIMD_AB_1, sq_dyn_gpr_size_simd_ab_0); + WREG32(SQ_DYN_GPR_SIZE_SIMD_AB_2, sq_dyn_gpr_size_simd_ab_0); + WREG32(SQ_DYN_GPR_SIZE_SIMD_AB_3, sq_dyn_gpr_size_simd_ab_0); + WREG32(SQ_DYN_GPR_SIZE_SIMD_AB_4, sq_dyn_gpr_size_simd_ab_0); + WREG32(SQ_DYN_GPR_SIZE_SIMD_AB_5, sq_dyn_gpr_size_simd_ab_0); + WREG32(SQ_DYN_GPR_SIZE_SIMD_AB_6, sq_dyn_gpr_size_simd_ab_0); + WREG32(SQ_DYN_GPR_SIZE_SIMD_AB_7, sq_dyn_gpr_size_simd_ab_0); + + WREG32(PA_SC_FORCE_EOV_MAX_CNTS, (FORCE_EOV_MAX_CLK_CNT(4095) | + FORCE_EOV_MAX_REZ_CNT(255))); + + if (rdev->family == CHIP_RV710) + WREG32(VGT_CACHE_INVALIDATION, (CACHE_INVALIDATION(TC_ONLY) | + AUTO_INVLD_EN(ES_AND_GS_AUTO))); + else + WREG32(VGT_CACHE_INVALIDATION, (CACHE_INVALIDATION(VC_AND_TC) | + AUTO_INVLD_EN(ES_AND_GS_AUTO))); + + switch (rdev->family) { + case CHIP_RV770: + case CHIP_RV730: + case CHIP_RV740: + gs_prim_buffer_depth = 384; + break; + case CHIP_RV710: + gs_prim_buffer_depth = 128; + break; + default: + break; + } + + num_gs_verts_per_thread = rdev->config.rv770.max_pipes * 16; + vgt_gs_per_es = gs_prim_buffer_depth + num_gs_verts_per_thread; + /* Max value for this is 256 */ + if (vgt_gs_per_es > 256) + vgt_gs_per_es = 256; + + WREG32(VGT_ES_PER_GS, 128); + WREG32(VGT_GS_PER_ES, vgt_gs_per_es); + WREG32(VGT_GS_PER_VS, 2); + + /* more default values. 2D/3D driver should adjust as needed */ + WREG32(VGT_GS_VERTEX_REUSE, 16); + WREG32(PA_SC_LINE_STIPPLE_STATE, 0); + WREG32(VGT_STRMOUT_EN, 0); + WREG32(SX_MISC, 0); + WREG32(PA_SC_MODE_CNTL, 0); + WREG32(PA_SC_EDGERULE, 0xaaaaaaaa); + WREG32(PA_SC_AA_CONFIG, 0); + WREG32(PA_SC_CLIPRECT_RULE, 0xffff); + WREG32(PA_SC_LINE_STIPPLE, 0); + WREG32(SPI_INPUT_Z, 0); + WREG32(SPI_PS_IN_CONTROL_0, NUM_INTERP(2)); + WREG32(CB_COLOR7_FRAG, 0); + + /* clear render buffer base addresses */ + WREG32(CB_COLOR0_BASE, 0); + WREG32(CB_COLOR1_BASE, 0); + WREG32(CB_COLOR2_BASE, 0); + WREG32(CB_COLOR3_BASE, 0); + WREG32(CB_COLOR4_BASE, 0); + WREG32(CB_COLOR5_BASE, 0); + WREG32(CB_COLOR6_BASE, 0); + WREG32(CB_COLOR7_BASE, 0); + + WREG32(TCP_CNTL, 0); + + hdp_host_path_cntl = RREG32(HDP_HOST_PATH_CNTL); + WREG32(HDP_HOST_PATH_CNTL, hdp_host_path_cntl); + + WREG32(PA_SC_MULTI_CHIP_CNTL, 0); + + WREG32(PA_CL_ENHANCE, (CLIP_VTX_REORDER_ENA | + NUM_CLIP_SEQ(3))); + +} + +static int rv770_vram_scratch_init(struct radeon_device *rdev) +{ + int r; + u64 gpu_addr; + + if (rdev->vram_scratch.robj == NULL) { + r = radeon_bo_create(rdev, RADEON_GPU_PAGE_SIZE, + PAGE_SIZE, true, RADEON_GEM_DOMAIN_VRAM, + &rdev->vram_scratch.robj); + if (r) { + return r; + } + } + + r = radeon_bo_reserve(rdev->vram_scratch.robj, false); + if (unlikely(r != 0)) + return r; + r = radeon_bo_pin(rdev->vram_scratch.robj, + RADEON_GEM_DOMAIN_VRAM, &gpu_addr); + if (r) { + radeon_bo_unreserve(rdev->vram_scratch.robj); + return r; + } + r = radeon_bo_kmap(rdev->vram_scratch.robj, + (void **)&rdev->vram_scratch.ptr); + if (r) + radeon_bo_unpin(rdev->vram_scratch.robj); + radeon_bo_unreserve(rdev->vram_scratch.robj); + + return r; +} + +static void rv770_vram_scratch_fini(struct radeon_device *rdev) +{ + int r; + + if (rdev->vram_scratch.robj == NULL) { + return; + } + r = radeon_bo_reserve(rdev->vram_scratch.robj, false); + if (likely(r == 0)) { + radeon_bo_kunmap(rdev->vram_scratch.robj); + radeon_bo_unpin(rdev->vram_scratch.robj); + radeon_bo_unreserve(rdev->vram_scratch.robj); + } + radeon_bo_unref(&rdev->vram_scratch.robj); +} + +void r700_vram_gtt_location(struct radeon_device *rdev, struct radeon_mc *mc) +{ + u64 size_bf, size_af; + + if (mc->mc_vram_size > 0xE0000000) { + /* leave room for at least 512M GTT */ + dev_warn(rdev->dev, "limiting VRAM\n"); + mc->real_vram_size = 0xE0000000; + mc->mc_vram_size = 0xE0000000; + } + if (rdev->flags & RADEON_IS_AGP) { + size_bf = mc->gtt_start; + size_af = 0xFFFFFFFF - mc->gtt_end + 1; + if (size_bf > size_af) { + if (mc->mc_vram_size > size_bf) { + dev_warn(rdev->dev, "limiting VRAM\n"); + mc->real_vram_size = size_bf; + mc->mc_vram_size = size_bf; + } + mc->vram_start = mc->gtt_start - mc->mc_vram_size; + } else { + if (mc->mc_vram_size > size_af) { + dev_warn(rdev->dev, "limiting VRAM\n"); + mc->real_vram_size = size_af; + mc->mc_vram_size = size_af; + } + mc->vram_start = mc->gtt_end; + } + mc->vram_end = mc->vram_start + mc->mc_vram_size - 1; + dev_info(rdev->dev, "VRAM: %lluM 0x%08llX - 0x%08llX (%lluM used)\n", + mc->mc_vram_size >> 20, mc->vram_start, + mc->vram_end, mc->real_vram_size >> 20); + } else { + radeon_vram_location(rdev, &rdev->mc, 0); + rdev->mc.gtt_base_align = 0; + radeon_gtt_location(rdev, mc); + } +} + +int rv770_mc_init(struct radeon_device *rdev) +{ + u32 tmp; + int chansize, numchan; + + /* Get VRAM informations */ + rdev->mc.vram_is_ddr = true; + tmp = RREG32(MC_ARB_RAMCFG); + if (tmp & CHANSIZE_OVERRIDE) { + chansize = 16; + } else if (tmp & CHANSIZE_MASK) { + chansize = 64; + } else { + chansize = 32; + } + tmp = RREG32(MC_SHARED_CHMAP); + switch ((tmp & NOOFCHAN_MASK) >> NOOFCHAN_SHIFT) { + case 0: + default: + numchan = 1; + break; + case 1: + numchan = 2; + break; + case 2: + numchan = 4; + break; + case 3: + numchan = 8; + break; + } + rdev->mc.vram_width = numchan * chansize; + /* Could aper size report 0 ? */ + rdev->mc.aper_base = pci_resource_start(rdev->pdev, 0); + rdev->mc.aper_size = pci_resource_len(rdev->pdev, 0); + /* Setup GPU memory space */ + rdev->mc.mc_vram_size = RREG32(CONFIG_MEMSIZE); + rdev->mc.real_vram_size = RREG32(CONFIG_MEMSIZE); + rdev->mc.visible_vram_size = rdev->mc.aper_size; + r700_vram_gtt_location(rdev, &rdev->mc); + radeon_update_bandwidth_info(rdev); + + return 0; +} + +static int rv770_startup(struct radeon_device *rdev) +{ + int r; + + /* enable pcie gen2 link */ + rv770_pcie_gen2_enable(rdev); + + if (!rdev->me_fw || !rdev->pfp_fw || !rdev->rlc_fw) { + r = r600_init_microcode(rdev); + if (r) { + DRM_ERROR("Failed to load firmware!\n"); + return r; + } + } + + rv770_mc_program(rdev); + if (rdev->flags & RADEON_IS_AGP) { + rv770_agp_enable(rdev); + } else { + r = rv770_pcie_gart_enable(rdev); + if (r) + return r; + } + r = rv770_vram_scratch_init(rdev); + if (r) + return r; + rv770_gpu_init(rdev); + r = r600_blit_init(rdev); + if (r) { + r600_blit_fini(rdev); + rdev->asic->copy = NULL; + dev_warn(rdev->dev, "failed blitter (%d) falling back to memcpy\n", r); + } + + /* allocate wb buffer */ + r = radeon_wb_init(rdev); + if (r) + return r; + + /* Enable IRQ */ + r = r600_irq_init(rdev); + if (r) { + DRM_ERROR("radeon: IH init failed (%d).\n", r); + radeon_irq_kms_fini(rdev); + return r; + } + r600_irq_set(rdev); + + r = radeon_ring_init(rdev, rdev->cp.ring_size); + if (r) + return r; + r = rv770_cp_load_microcode(rdev); + if (r) + return r; + r = r600_cp_resume(rdev); + if (r) + return r; + + return 0; +} + +int rv770_resume(struct radeon_device *rdev) +{ + int r; + + /* Do not reset GPU before posting, on rv770 hw unlike on r500 hw, + * posting will perform necessary task to bring back GPU into good + * shape. + */ + /* post card */ + atom_asic_init(rdev->mode_info.atom_context); + + r = rv770_startup(rdev); + if (r) { + DRM_ERROR("r600 startup failed on resume\n"); + return r; + } + + r = r600_ib_test(rdev); + if (r) { + DRM_ERROR("radeon: failed testing IB (%d).\n", r); + return r; + } + + r = r600_audio_init(rdev); + if (r) { + dev_err(rdev->dev, "radeon: audio init failed\n"); + return r; + } + + return r; + +} + +int rv770_suspend(struct radeon_device *rdev) +{ + int r; + + r600_audio_fini(rdev); + /* FIXME: we should wait for ring to be empty */ + r700_cp_stop(rdev); + rdev->cp.ready = false; + r600_irq_suspend(rdev); + radeon_wb_disable(rdev); + rv770_pcie_gart_disable(rdev); + /* unpin shaders bo */ + if (rdev->r600_blit.shader_obj) { + r = radeon_bo_reserve(rdev->r600_blit.shader_obj, false); + if (likely(r == 0)) { + radeon_bo_unpin(rdev->r600_blit.shader_obj); + radeon_bo_unreserve(rdev->r600_blit.shader_obj); + } + } + return 0; +} + +/* Plan is to move initialization in that function and use + * helper function so that radeon_device_init pretty much + * do nothing more than calling asic specific function. This + * should also allow to remove a bunch of callback function + * like vram_info. + */ +int rv770_init(struct radeon_device *rdev) +{ + int r; + + /* This don't do much */ + r = radeon_gem_init(rdev); + if (r) + return r; + /* Read BIOS */ + if (!radeon_get_bios(rdev)) { + if (ASIC_IS_AVIVO(rdev)) + return -EINVAL; + } + /* Must be an ATOMBIOS */ + if (!rdev->is_atom_bios) { + dev_err(rdev->dev, "Expecting atombios for R600 GPU\n"); + return -EINVAL; + } + r = radeon_atombios_init(rdev); + if (r) + return r; + /* Post card if necessary */ + if (!radeon_card_posted(rdev)) { + if (!rdev->bios) { + dev_err(rdev->dev, "Card not posted and no BIOS - ignoring\n"); + return -EINVAL; + } + DRM_INFO("GPU not posted. posting now...\n"); + atom_asic_init(rdev->mode_info.atom_context); + } + /* Initialize scratch registers */ + r600_scratch_init(rdev); + /* Initialize surface registers */ + radeon_surface_init(rdev); + /* Initialize clocks */ + radeon_get_clock_info(rdev->ddev); + /* Fence driver */ + r = radeon_fence_driver_init(rdev); + if (r) + return r; + /* initialize AGP */ + if (rdev->flags & RADEON_IS_AGP) { + r = radeon_agp_init(rdev); + if (r) + radeon_agp_disable(rdev); + } + r = rv770_mc_init(rdev); + if (r) + return r; + /* Memory manager */ + r = radeon_bo_init(rdev); + if (r) + return r; + + r = radeon_irq_kms_init(rdev); + if (r) + return r; + + rdev->cp.ring_obj = NULL; + r600_ring_init(rdev, 1024 * 1024); + + rdev->ih.ring_obj = NULL; + r600_ih_ring_init(rdev, 64 * 1024); + + r = r600_pcie_gart_init(rdev); + if (r) + return r; + + rdev->accel_working = true; + r = rv770_startup(rdev); + if (r) { + dev_err(rdev->dev, "disabling GPU acceleration\n"); + r700_cp_fini(rdev); + r600_irq_fini(rdev); + radeon_wb_fini(rdev); + radeon_irq_kms_fini(rdev); + rv770_pcie_gart_fini(rdev); + rdev->accel_working = false; + } + if (rdev->accel_working) { + r = radeon_ib_pool_init(rdev); + if (r) { + dev_err(rdev->dev, "IB initialization failed (%d).\n", r); + rdev->accel_working = false; + } else { + r = r600_ib_test(rdev); + if (r) { + dev_err(rdev->dev, "IB test failed (%d).\n", r); + rdev->accel_working = false; + } + } + } + + r = r600_audio_init(rdev); + if (r) { + dev_err(rdev->dev, "radeon: audio init failed\n"); + return r; + } + + return 0; +} + +void rv770_fini(struct radeon_device *rdev) +{ + r600_blit_fini(rdev); + r700_cp_fini(rdev); + r600_irq_fini(rdev); + radeon_wb_fini(rdev); + radeon_ib_pool_fini(rdev); + radeon_irq_kms_fini(rdev); + rv770_pcie_gart_fini(rdev); + rv770_vram_scratch_fini(rdev); + radeon_gem_fini(rdev); + radeon_fence_driver_fini(rdev); + radeon_agp_fini(rdev); + radeon_bo_fini(rdev); + radeon_atombios_fini(rdev); + kfree(rdev->bios); + rdev->bios = NULL; +} + +static void rv770_pcie_gen2_enable(struct radeon_device *rdev) +{ + u32 link_width_cntl, lanes, speed_cntl, tmp; + u16 link_cntl2; + + if (radeon_pcie_gen2 == 0) + return; + + if (rdev->flags & RADEON_IS_IGP) + return; + + if (!(rdev->flags & RADEON_IS_PCIE)) + return; + + /* x2 cards have a special sequence */ + if (ASIC_IS_X2(rdev)) + return; + + /* advertise upconfig capability */ + link_width_cntl = RREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL); + link_width_cntl &= ~LC_UPCONFIGURE_DIS; + WREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl); + link_width_cntl = RREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL); + if (link_width_cntl & LC_RENEGOTIATION_SUPPORT) { + lanes = (link_width_cntl & LC_LINK_WIDTH_RD_MASK) >> LC_LINK_WIDTH_RD_SHIFT; + link_width_cntl &= ~(LC_LINK_WIDTH_MASK | + LC_RECONFIG_ARC_MISSING_ESCAPE); + link_width_cntl |= lanes | LC_RECONFIG_NOW | + LC_RENEGOTIATE_EN | LC_UPCONFIGURE_SUPPORT; + WREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl); + } else { + link_width_cntl |= LC_UPCONFIGURE_DIS; + WREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl); + } + + speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL); + if ((speed_cntl & LC_OTHER_SIDE_EVER_SENT_GEN2) && + (speed_cntl & LC_OTHER_SIDE_SUPPORTS_GEN2)) { + + tmp = RREG32(0x541c); + WREG32(0x541c, tmp | 0x8); + WREG32(MM_CFGREGS_CNTL, MM_WR_TO_CFG_EN); + link_cntl2 = RREG16(0x4088); + link_cntl2 &= ~TARGET_LINK_SPEED_MASK; + link_cntl2 |= 0x2; + WREG16(0x4088, link_cntl2); + WREG32(MM_CFGREGS_CNTL, 0); + + speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL); + speed_cntl &= ~LC_TARGET_LINK_SPEED_OVERRIDE_EN; + WREG32_PCIE_P(PCIE_LC_SPEED_CNTL, speed_cntl); + + speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL); + speed_cntl |= LC_CLR_FAILED_SPD_CHANGE_CNT; + WREG32_PCIE_P(PCIE_LC_SPEED_CNTL, speed_cntl); + + speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL); + speed_cntl &= ~LC_CLR_FAILED_SPD_CHANGE_CNT; + WREG32_PCIE_P(PCIE_LC_SPEED_CNTL, speed_cntl); + + speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL); + speed_cntl |= LC_GEN2_EN_STRAP; + WREG32_PCIE_P(PCIE_LC_SPEED_CNTL, speed_cntl); + + } else { + link_width_cntl = RREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL); + /* XXX: only disable it if gen1 bridge vendor == 0x111d or 0x1106 */ + if (1) + link_width_cntl |= LC_UPCONFIGURE_DIS; + else + link_width_cntl &= ~LC_UPCONFIGURE_DIS; + WREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl); + } +} diff --git a/drivers/gpu/drm/radeon/rv770d.h b/drivers/gpu/drm/radeon/rv770d.h new file mode 100644 index 00000000..75380927 --- /dev/null +++ b/drivers/gpu/drm/radeon/rv770d.h @@ -0,0 +1,402 @@ +/* + * Copyright 2009 Advanced Micro Devices, Inc. + * Copyright 2009 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#ifndef RV770_H +#define RV770_H + +#define R7XX_MAX_SH_GPRS 256 +#define R7XX_MAX_TEMP_GPRS 16 +#define R7XX_MAX_SH_THREADS 256 +#define R7XX_MAX_SH_STACK_ENTRIES 4096 +#define R7XX_MAX_BACKENDS 8 +#define R7XX_MAX_BACKENDS_MASK 0xff +#define R7XX_MAX_SIMDS 16 +#define R7XX_MAX_SIMDS_MASK 0xffff +#define R7XX_MAX_PIPES 8 +#define R7XX_MAX_PIPES_MASK 0xff + +/* Registers */ +#define CB_COLOR0_BASE 0x28040 +#define CB_COLOR1_BASE 0x28044 +#define CB_COLOR2_BASE 0x28048 +#define CB_COLOR3_BASE 0x2804C +#define CB_COLOR4_BASE 0x28050 +#define CB_COLOR5_BASE 0x28054 +#define CB_COLOR6_BASE 0x28058 +#define CB_COLOR7_BASE 0x2805C +#define CB_COLOR7_FRAG 0x280FC + +#define CC_GC_SHADER_PIPE_CONFIG 0x8950 +#define CC_RB_BACKEND_DISABLE 0x98F4 +#define BACKEND_DISABLE(x) ((x) << 16) +#define CC_SYS_RB_BACKEND_DISABLE 0x3F88 + +#define CGTS_SYS_TCC_DISABLE 0x3F90 +#define CGTS_TCC_DISABLE 0x9148 +#define CGTS_USER_SYS_TCC_DISABLE 0x3F94 +#define CGTS_USER_TCC_DISABLE 0x914C + +#define CONFIG_MEMSIZE 0x5428 + +#define CP_ME_CNTL 0x86D8 +#define CP_ME_HALT (1<<28) +#define CP_PFP_HALT (1<<26) +#define CP_ME_RAM_DATA 0xC160 +#define CP_ME_RAM_RADDR 0xC158 +#define CP_ME_RAM_WADDR 0xC15C +#define CP_MEQ_THRESHOLDS 0x8764 +#define STQ_SPLIT(x) ((x) << 0) +#define CP_PERFMON_CNTL 0x87FC +#define CP_PFP_UCODE_ADDR 0xC150 +#define CP_PFP_UCODE_DATA 0xC154 +#define CP_QUEUE_THRESHOLDS 0x8760 +#define ROQ_IB1_START(x) ((x) << 0) +#define ROQ_IB2_START(x) ((x) << 8) +#define CP_RB_CNTL 0xC104 +#define RB_BUFSZ(x) ((x) << 0) +#define RB_BLKSZ(x) ((x) << 8) +#define RB_NO_UPDATE (1 << 27) +#define RB_RPTR_WR_ENA (1 << 31) +#define BUF_SWAP_32BIT (2 << 16) +#define CP_RB_RPTR 0x8700 +#define CP_RB_RPTR_ADDR 0xC10C +#define CP_RB_RPTR_ADDR_HI 0xC110 +#define CP_RB_RPTR_WR 0xC108 +#define CP_RB_WPTR 0xC114 +#define CP_RB_WPTR_ADDR 0xC118 +#define CP_RB_WPTR_ADDR_HI 0xC11C +#define CP_RB_WPTR_DELAY 0x8704 +#define CP_SEM_WAIT_TIMER 0x85BC + +#define DB_DEBUG3 0x98B0 +#define DB_CLK_OFF_DELAY(x) ((x) << 11) +#define DB_DEBUG4 0x9B8C +#define DISABLE_TILE_COVERED_FOR_PS_ITER (1 << 6) + +#define DCP_TILING_CONFIG 0x6CA0 +#define PIPE_TILING(x) ((x) << 1) +#define BANK_TILING(x) ((x) << 4) +#define GROUP_SIZE(x) ((x) << 6) +#define ROW_TILING(x) ((x) << 8) +#define BANK_SWAPS(x) ((x) << 11) +#define SAMPLE_SPLIT(x) ((x) << 14) +#define BACKEND_MAP(x) ((x) << 16) + +#define GB_TILING_CONFIG 0x98F0 + +#define GC_USER_SHADER_PIPE_CONFIG 0x8954 +#define INACTIVE_QD_PIPES(x) ((x) << 8) +#define INACTIVE_QD_PIPES_MASK 0x0000FF00 +#define INACTIVE_SIMDS(x) ((x) << 16) +#define INACTIVE_SIMDS_MASK 0x00FF0000 + +#define GRBM_CNTL 0x8000 +#define GRBM_READ_TIMEOUT(x) ((x) << 0) +#define GRBM_SOFT_RESET 0x8020 +#define SOFT_RESET_CP (1<<0) +#define GRBM_STATUS 0x8010 +#define CMDFIFO_AVAIL_MASK 0x0000000F +#define GUI_ACTIVE (1<<31) +#define GRBM_STATUS2 0x8014 + +#define CG_MULT_THERMAL_STATUS 0x740 +#define ASIC_T(x) ((x) << 16) +#define ASIC_T_MASK 0x3FF0000 +#define ASIC_T_SHIFT 16 + +#define HDP_HOST_PATH_CNTL 0x2C00 +#define HDP_NONSURFACE_BASE 0x2C04 +#define HDP_NONSURFACE_INFO 0x2C08 +#define HDP_NONSURFACE_SIZE 0x2C0C +#define HDP_REG_COHERENCY_FLUSH_CNTL 0x54A0 +#define HDP_TILING_CONFIG 0x2F3C +#define HDP_DEBUG1 0x2F34 + +#define MC_SHARED_CHMAP 0x2004 +#define NOOFCHAN_SHIFT 12 +#define NOOFCHAN_MASK 0x00003000 +#define MC_SHARED_CHREMAP 0x2008 + +#define MC_ARB_RAMCFG 0x2760 +#define NOOFBANK_SHIFT 0 +#define NOOFBANK_MASK 0x00000003 +#define NOOFRANK_SHIFT 2 +#define NOOFRANK_MASK 0x00000004 +#define NOOFROWS_SHIFT 3 +#define NOOFROWS_MASK 0x00000038 +#define NOOFCOLS_SHIFT 6 +#define NOOFCOLS_MASK 0x000000C0 +#define CHANSIZE_SHIFT 8 +#define CHANSIZE_MASK 0x00000100 +#define BURSTLENGTH_SHIFT 9 +#define BURSTLENGTH_MASK 0x00000200 +#define CHANSIZE_OVERRIDE (1 << 11) +#define MC_VM_AGP_TOP 0x2028 +#define MC_VM_AGP_BOT 0x202C +#define MC_VM_AGP_BASE 0x2030 +#define MC_VM_FB_LOCATION 0x2024 +#define MC_VM_MB_L1_TLB0_CNTL 0x2234 +#define MC_VM_MB_L1_TLB1_CNTL 0x2238 +#define MC_VM_MB_L1_TLB2_CNTL 0x223C +#define MC_VM_MB_L1_TLB3_CNTL 0x2240 +#define ENABLE_L1_TLB (1 << 0) +#define ENABLE_L1_FRAGMENT_PROCESSING (1 << 1) +#define SYSTEM_ACCESS_MODE_PA_ONLY (0 << 3) +#define SYSTEM_ACCESS_MODE_USE_SYS_MAP (1 << 3) +#define SYSTEM_ACCESS_MODE_IN_SYS (2 << 3) +#define SYSTEM_ACCESS_MODE_NOT_IN_SYS (3 << 3) +#define SYSTEM_APERTURE_UNMAPPED_ACCESS_PASS_THRU (0 << 5) +#define EFFECTIVE_L1_TLB_SIZE(x) ((x)<<15) +#define EFFECTIVE_L1_QUEUE_SIZE(x) ((x)<<18) +#define MC_VM_MD_L1_TLB0_CNTL 0x2654 +#define MC_VM_MD_L1_TLB1_CNTL 0x2658 +#define MC_VM_MD_L1_TLB2_CNTL 0x265C +#define MC_VM_MD_L1_TLB3_CNTL 0x2698 +#define MC_VM_SYSTEM_APERTURE_DEFAULT_ADDR 0x203C +#define MC_VM_SYSTEM_APERTURE_HIGH_ADDR 0x2038 +#define MC_VM_SYSTEM_APERTURE_LOW_ADDR 0x2034 + +#define PA_CL_ENHANCE 0x8A14 +#define CLIP_VTX_REORDER_ENA (1 << 0) +#define NUM_CLIP_SEQ(x) ((x) << 1) +#define PA_SC_AA_CONFIG 0x28C04 +#define PA_SC_CLIPRECT_RULE 0x2820C +#define PA_SC_EDGERULE 0x28230 +#define PA_SC_FIFO_SIZE 0x8BCC +#define SC_PRIM_FIFO_SIZE(x) ((x) << 0) +#define SC_HIZ_TILE_FIFO_SIZE(x) ((x) << 12) +#define PA_SC_FORCE_EOV_MAX_CNTS 0x8B24 +#define FORCE_EOV_MAX_CLK_CNT(x) ((x)<<0) +#define FORCE_EOV_MAX_REZ_CNT(x) ((x)<<16) +#define PA_SC_LINE_STIPPLE 0x28A0C +#define PA_SC_LINE_STIPPLE_STATE 0x8B10 +#define PA_SC_MODE_CNTL 0x28A4C +#define PA_SC_MULTI_CHIP_CNTL 0x8B20 +#define SC_EARLYZ_TILE_FIFO_SIZE(x) ((x) << 20) + +#define SCRATCH_REG0 0x8500 +#define SCRATCH_REG1 0x8504 +#define SCRATCH_REG2 0x8508 +#define SCRATCH_REG3 0x850C +#define SCRATCH_REG4 0x8510 +#define SCRATCH_REG5 0x8514 +#define SCRATCH_REG6 0x8518 +#define SCRATCH_REG7 0x851C +#define SCRATCH_UMSK 0x8540 +#define SCRATCH_ADDR 0x8544 + +#define SMX_DC_CTL0 0xA020 +#define USE_HASH_FUNCTION (1 << 0) +#define CACHE_DEPTH(x) ((x) << 1) +#define FLUSH_ALL_ON_EVENT (1 << 10) +#define STALL_ON_EVENT (1 << 11) +#define SMX_EVENT_CTL 0xA02C +#define ES_FLUSH_CTL(x) ((x) << 0) +#define GS_FLUSH_CTL(x) ((x) << 3) +#define ACK_FLUSH_CTL(x) ((x) << 6) +#define SYNC_FLUSH_CTL (1 << 8) + +#define SPI_CONFIG_CNTL 0x9100 +#define GPR_WRITE_PRIORITY(x) ((x) << 0) +#define DISABLE_INTERP_1 (1 << 5) +#define SPI_CONFIG_CNTL_1 0x913C +#define VTX_DONE_DELAY(x) ((x) << 0) +#define INTERP_ONE_PRIM_PER_ROW (1 << 4) +#define SPI_INPUT_Z 0x286D8 +#define SPI_PS_IN_CONTROL_0 0x286CC +#define NUM_INTERP(x) ((x)<<0) +#define POSITION_ENA (1<<8) +#define POSITION_CENTROID (1<<9) +#define POSITION_ADDR(x) ((x)<<10) +#define PARAM_GEN(x) ((x)<<15) +#define PARAM_GEN_ADDR(x) ((x)<<19) +#define BARYC_SAMPLE_CNTL(x) ((x)<<26) +#define PERSP_GRADIENT_ENA (1<<28) +#define LINEAR_GRADIENT_ENA (1<<29) +#define POSITION_SAMPLE (1<<30) +#define BARYC_AT_SAMPLE_ENA (1<<31) + +#define SQ_CONFIG 0x8C00 +#define VC_ENABLE (1 << 0) +#define EXPORT_SRC_C (1 << 1) +#define DX9_CONSTS (1 << 2) +#define ALU_INST_PREFER_VECTOR (1 << 3) +#define DX10_CLAMP (1 << 4) +#define CLAUSE_SEQ_PRIO(x) ((x) << 8) +#define PS_PRIO(x) ((x) << 24) +#define VS_PRIO(x) ((x) << 26) +#define GS_PRIO(x) ((x) << 28) +#define SQ_DYN_GPR_SIZE_SIMD_AB_0 0x8DB0 +#define SIMDA_RING0(x) ((x)<<0) +#define SIMDA_RING1(x) ((x)<<8) +#define SIMDB_RING0(x) ((x)<<16) +#define SIMDB_RING1(x) ((x)<<24) +#define SQ_DYN_GPR_SIZE_SIMD_AB_1 0x8DB4 +#define SQ_DYN_GPR_SIZE_SIMD_AB_2 0x8DB8 +#define SQ_DYN_GPR_SIZE_SIMD_AB_3 0x8DBC +#define SQ_DYN_GPR_SIZE_SIMD_AB_4 0x8DC0 +#define SQ_DYN_GPR_SIZE_SIMD_AB_5 0x8DC4 +#define SQ_DYN_GPR_SIZE_SIMD_AB_6 0x8DC8 +#define SQ_DYN_GPR_SIZE_SIMD_AB_7 0x8DCC +#define ES_PRIO(x) ((x) << 30) +#define SQ_GPR_RESOURCE_MGMT_1 0x8C04 +#define NUM_PS_GPRS(x) ((x) << 0) +#define NUM_VS_GPRS(x) ((x) << 16) +#define DYN_GPR_ENABLE (1 << 27) +#define NUM_CLAUSE_TEMP_GPRS(x) ((x) << 28) +#define SQ_GPR_RESOURCE_MGMT_2 0x8C08 +#define NUM_GS_GPRS(x) ((x) << 0) +#define NUM_ES_GPRS(x) ((x) << 16) +#define SQ_MS_FIFO_SIZES 0x8CF0 +#define CACHE_FIFO_SIZE(x) ((x) << 0) +#define FETCH_FIFO_HIWATER(x) ((x) << 8) +#define DONE_FIFO_HIWATER(x) ((x) << 16) +#define ALU_UPDATE_FIFO_HIWATER(x) ((x) << 24) +#define SQ_STACK_RESOURCE_MGMT_1 0x8C10 +#define NUM_PS_STACK_ENTRIES(x) ((x) << 0) +#define NUM_VS_STACK_ENTRIES(x) ((x) << 16) +#define SQ_STACK_RESOURCE_MGMT_2 0x8C14 +#define NUM_GS_STACK_ENTRIES(x) ((x) << 0) +#define NUM_ES_STACK_ENTRIES(x) ((x) << 16) +#define SQ_THREAD_RESOURCE_MGMT 0x8C0C +#define NUM_PS_THREADS(x) ((x) << 0) +#define NUM_VS_THREADS(x) ((x) << 8) +#define NUM_GS_THREADS(x) ((x) << 16) +#define NUM_ES_THREADS(x) ((x) << 24) + +#define SX_DEBUG_1 0x9058 +#define ENABLE_NEW_SMX_ADDRESS (1 << 16) +#define SX_EXPORT_BUFFER_SIZES 0x900C +#define COLOR_BUFFER_SIZE(x) ((x) << 0) +#define POSITION_BUFFER_SIZE(x) ((x) << 8) +#define SMX_BUFFER_SIZE(x) ((x) << 16) +#define SX_MISC 0x28350 + +#define TA_CNTL_AUX 0x9508 +#define DISABLE_CUBE_WRAP (1 << 0) +#define DISABLE_CUBE_ANISO (1 << 1) +#define SYNC_GRADIENT (1 << 24) +#define SYNC_WALKER (1 << 25) +#define SYNC_ALIGNER (1 << 26) +#define BILINEAR_PRECISION_6_BIT (0 << 31) +#define BILINEAR_PRECISION_8_BIT (1 << 31) + +#define TCP_CNTL 0x9610 +#define TCP_CHAN_STEER 0x9614 + +#define VGT_CACHE_INVALIDATION 0x88C4 +#define CACHE_INVALIDATION(x) ((x)<<0) +#define VC_ONLY 0 +#define TC_ONLY 1 +#define VC_AND_TC 2 +#define AUTO_INVLD_EN(x) ((x) << 6) +#define NO_AUTO 0 +#define ES_AUTO 1 +#define GS_AUTO 2 +#define ES_AND_GS_AUTO 3 +#define VGT_ES_PER_GS 0x88CC +#define VGT_GS_PER_ES 0x88C8 +#define VGT_GS_PER_VS 0x88E8 +#define VGT_GS_VERTEX_REUSE 0x88D4 +#define VGT_NUM_INSTANCES 0x8974 +#define VGT_OUT_DEALLOC_CNTL 0x28C5C +#define DEALLOC_DIST_MASK 0x0000007F +#define VGT_STRMOUT_EN 0x28AB0 +#define VGT_VERTEX_REUSE_BLOCK_CNTL 0x28C58 +#define VTX_REUSE_DEPTH_MASK 0x000000FF + +#define VM_CONTEXT0_CNTL 0x1410 +#define ENABLE_CONTEXT (1 << 0) +#define PAGE_TABLE_DEPTH(x) (((x) & 3) << 1) +#define RANGE_PROTECTION_FAULT_ENABLE_DEFAULT (1 << 4) +#define VM_CONTEXT0_PAGE_TABLE_BASE_ADDR 0x153C +#define VM_CONTEXT0_PAGE_TABLE_END_ADDR 0x157C +#define VM_CONTEXT0_PAGE_TABLE_START_ADDR 0x155C +#define VM_CONTEXT0_PROTECTION_FAULT_DEFAULT_ADDR 0x1518 +#define VM_L2_CNTL 0x1400 +#define ENABLE_L2_CACHE (1 << 0) +#define ENABLE_L2_FRAGMENT_PROCESSING (1 << 1) +#define ENABLE_L2_PTE_CACHE_LRU_UPDATE_BY_WRITE (1 << 9) +#define EFFECTIVE_L2_QUEUE_SIZE(x) (((x) & 7) << 14) +#define VM_L2_CNTL2 0x1404 +#define INVALIDATE_ALL_L1_TLBS (1 << 0) +#define INVALIDATE_L2_CACHE (1 << 1) +#define VM_L2_CNTL3 0x1408 +#define BANK_SELECT(x) ((x) << 0) +#define CACHE_UPDATE_MODE(x) ((x) << 6) +#define VM_L2_STATUS 0x140C +#define L2_BUSY (1 << 0) + +#define WAIT_UNTIL 0x8040 + +#define SRBM_STATUS 0x0E50 + +#define D1GRPH_PRIMARY_SURFACE_ADDRESS 0x6110 +#define D1GRPH_PRIMARY_SURFACE_ADDRESS_HIGH 0x6914 +#define D2GRPH_PRIMARY_SURFACE_ADDRESS_HIGH 0x6114 +#define D1GRPH_SECONDARY_SURFACE_ADDRESS 0x6118 +#define D1GRPH_SECONDARY_SURFACE_ADDRESS_HIGH 0x691c +#define D2GRPH_SECONDARY_SURFACE_ADDRESS_HIGH 0x611c + +/* PCIE link stuff */ +#define PCIE_LC_TRAINING_CNTL 0xa1 /* PCIE_P */ +#define PCIE_LC_LINK_WIDTH_CNTL 0xa2 /* PCIE_P */ +# define LC_LINK_WIDTH_SHIFT 0 +# define LC_LINK_WIDTH_MASK 0x7 +# define LC_LINK_WIDTH_X0 0 +# define LC_LINK_WIDTH_X1 1 +# define LC_LINK_WIDTH_X2 2 +# define LC_LINK_WIDTH_X4 3 +# define LC_LINK_WIDTH_X8 4 +# define LC_LINK_WIDTH_X16 6 +# define LC_LINK_WIDTH_RD_SHIFT 4 +# define LC_LINK_WIDTH_RD_MASK 0x70 +# define LC_RECONFIG_ARC_MISSING_ESCAPE (1 << 7) +# define LC_RECONFIG_NOW (1 << 8) +# define LC_RENEGOTIATION_SUPPORT (1 << 9) +# define LC_RENEGOTIATE_EN (1 << 10) +# define LC_SHORT_RECONFIG_EN (1 << 11) +# define LC_UPCONFIGURE_SUPPORT (1 << 12) +# define LC_UPCONFIGURE_DIS (1 << 13) +#define PCIE_LC_SPEED_CNTL 0xa4 /* PCIE_P */ +# define LC_GEN2_EN_STRAP (1 << 0) +# define LC_TARGET_LINK_SPEED_OVERRIDE_EN (1 << 1) +# define LC_FORCE_EN_HW_SPEED_CHANGE (1 << 5) +# define LC_FORCE_DIS_HW_SPEED_CHANGE (1 << 6) +# define LC_SPEED_CHANGE_ATTEMPTS_ALLOWED_MASK (0x3 << 8) +# define LC_SPEED_CHANGE_ATTEMPTS_ALLOWED_SHIFT 3 +# define LC_CURRENT_DATA_RATE (1 << 11) +# define LC_VOLTAGE_TIMER_SEL_MASK (0xf << 14) +# define LC_CLR_FAILED_SPD_CHANGE_CNT (1 << 21) +# define LC_OTHER_SIDE_EVER_SENT_GEN2 (1 << 23) +# define LC_OTHER_SIDE_SUPPORTS_GEN2 (1 << 24) +#define MM_CFGREGS_CNTL 0x544c +# define MM_WR_TO_CFG_EN (1 << 3) +#define LINK_CNTL2 0x88 /* F0 */ +# define TARGET_LINK_SPEED_MASK (0xf << 0) +# define SELECTABLE_DEEMPHASIS (1 << 6) + +#endif diff --git a/drivers/gpu/drm/savage/Makefile b/drivers/gpu/drm/savage/Makefile new file mode 100644 index 00000000..d8f84ac7 --- /dev/null +++ b/drivers/gpu/drm/savage/Makefile @@ -0,0 +1,9 @@ +# +# Makefile for the drm device driver. This driver provides support for the +# Direct Rendering Infrastructure (DRI) in XFree86 4.1.0 and higher. + +ccflags-y = -Iinclude/drm +savage-y := savage_drv.o savage_bci.o savage_state.o + +obj-$(CONFIG_DRM_SAVAGE)+= savage.o + diff --git a/drivers/gpu/drm/savage/savage_bci.c b/drivers/gpu/drm/savage/savage_bci.c new file mode 100644 index 00000000..cb1ee4e0 --- /dev/null +++ b/drivers/gpu/drm/savage/savage_bci.c @@ -0,0 +1,1088 @@ +/* savage_bci.c -- BCI support for Savage + * + * Copyright 2004 Felix Kuehling + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sub license, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NON-INFRINGEMENT. IN NO EVENT SHALL FELIX KUEHLING BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF + * CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + */ +#include "drmP.h" +#include "savage_drm.h" +#include "savage_drv.h" + +/* Need a long timeout for shadow status updates can take a while + * and so can waiting for events when the queue is full. */ +#define SAVAGE_DEFAULT_USEC_TIMEOUT 1000000 /* 1s */ +#define SAVAGE_EVENT_USEC_TIMEOUT 5000000 /* 5s */ +#define SAVAGE_FREELIST_DEBUG 0 + +static int savage_do_cleanup_bci(struct drm_device *dev); + +static int +savage_bci_wait_fifo_shadow(drm_savage_private_t * dev_priv, unsigned int n) +{ + uint32_t mask = dev_priv->status_used_mask; + uint32_t threshold = dev_priv->bci_threshold_hi; + uint32_t status; + int i; + +#if SAVAGE_BCI_DEBUG + if (n > dev_priv->cob_size + SAVAGE_BCI_FIFO_SIZE - threshold) + DRM_ERROR("Trying to emit %d words " + "(more than guaranteed space in COB)\n", n); +#endif + + for (i = 0; i < SAVAGE_DEFAULT_USEC_TIMEOUT; i++) { + DRM_MEMORYBARRIER(); + status = dev_priv->status_ptr[0]; + if ((status & mask) < threshold) + return 0; + DRM_UDELAY(1); + } + +#if SAVAGE_BCI_DEBUG + DRM_ERROR("failed!\n"); + DRM_INFO(" status=0x%08x, threshold=0x%08x\n", status, threshold); +#endif + return -EBUSY; +} + +static int +savage_bci_wait_fifo_s3d(drm_savage_private_t * dev_priv, unsigned int n) +{ + uint32_t maxUsed = dev_priv->cob_size + SAVAGE_BCI_FIFO_SIZE - n; + uint32_t status; + int i; + + for (i = 0; i < SAVAGE_DEFAULT_USEC_TIMEOUT; i++) { + status = SAVAGE_READ(SAVAGE_STATUS_WORD0); + if ((status & SAVAGE_FIFO_USED_MASK_S3D) <= maxUsed) + return 0; + DRM_UDELAY(1); + } + +#if SAVAGE_BCI_DEBUG + DRM_ERROR("failed!\n"); + DRM_INFO(" status=0x%08x\n", status); +#endif + return -EBUSY; +} + +static int +savage_bci_wait_fifo_s4(drm_savage_private_t * dev_priv, unsigned int n) +{ + uint32_t maxUsed = dev_priv->cob_size + SAVAGE_BCI_FIFO_SIZE - n; + uint32_t status; + int i; + + for (i = 0; i < SAVAGE_DEFAULT_USEC_TIMEOUT; i++) { + status = SAVAGE_READ(SAVAGE_ALT_STATUS_WORD0); + if ((status & SAVAGE_FIFO_USED_MASK_S4) <= maxUsed) + return 0; + DRM_UDELAY(1); + } + +#if SAVAGE_BCI_DEBUG + DRM_ERROR("failed!\n"); + DRM_INFO(" status=0x%08x\n", status); +#endif + return -EBUSY; +} + +/* + * Waiting for events. + * + * The BIOSresets the event tag to 0 on mode changes. Therefore we + * never emit 0 to the event tag. If we find a 0 event tag we know the + * BIOS stomped on it and return success assuming that the BIOS waited + * for engine idle. + * + * Note: if the Xserver uses the event tag it has to follow the same + * rule. Otherwise there may be glitches every 2^16 events. + */ +static int +savage_bci_wait_event_shadow(drm_savage_private_t * dev_priv, uint16_t e) +{ + uint32_t status; + int i; + + for (i = 0; i < SAVAGE_EVENT_USEC_TIMEOUT; i++) { + DRM_MEMORYBARRIER(); + status = dev_priv->status_ptr[1]; + if ((((status & 0xffff) - e) & 0xffff) <= 0x7fff || + (status & 0xffff) == 0) + return 0; + DRM_UDELAY(1); + } + +#if SAVAGE_BCI_DEBUG + DRM_ERROR("failed!\n"); + DRM_INFO(" status=0x%08x, e=0x%04x\n", status, e); +#endif + + return -EBUSY; +} + +static int +savage_bci_wait_event_reg(drm_savage_private_t * dev_priv, uint16_t e) +{ + uint32_t status; + int i; + + for (i = 0; i < SAVAGE_EVENT_USEC_TIMEOUT; i++) { + status = SAVAGE_READ(SAVAGE_STATUS_WORD1); + if ((((status & 0xffff) - e) & 0xffff) <= 0x7fff || + (status & 0xffff) == 0) + return 0; + DRM_UDELAY(1); + } + +#if SAVAGE_BCI_DEBUG + DRM_ERROR("failed!\n"); + DRM_INFO(" status=0x%08x, e=0x%04x\n", status, e); +#endif + + return -EBUSY; +} + +uint16_t savage_bci_emit_event(drm_savage_private_t * dev_priv, + unsigned int flags) +{ + uint16_t count; + BCI_LOCALS; + + if (dev_priv->status_ptr) { + /* coordinate with Xserver */ + count = dev_priv->status_ptr[1023]; + if (count < dev_priv->event_counter) + dev_priv->event_wrap++; + } else { + count = dev_priv->event_counter; + } + count = (count + 1) & 0xffff; + if (count == 0) { + count++; /* See the comment above savage_wait_event_*. */ + dev_priv->event_wrap++; + } + dev_priv->event_counter = count; + if (dev_priv->status_ptr) + dev_priv->status_ptr[1023] = (uint32_t) count; + + if ((flags & (SAVAGE_WAIT_2D | SAVAGE_WAIT_3D))) { + unsigned int wait_cmd = BCI_CMD_WAIT; + if ((flags & SAVAGE_WAIT_2D)) + wait_cmd |= BCI_CMD_WAIT_2D; + if ((flags & SAVAGE_WAIT_3D)) + wait_cmd |= BCI_CMD_WAIT_3D; + BEGIN_BCI(2); + BCI_WRITE(wait_cmd); + } else { + BEGIN_BCI(1); + } + BCI_WRITE(BCI_CMD_UPDATE_EVENT_TAG | (uint32_t) count); + + return count; +} + +/* + * Freelist management + */ +static int savage_freelist_init(struct drm_device * dev) +{ + drm_savage_private_t *dev_priv = dev->dev_private; + struct drm_device_dma *dma = dev->dma; + struct drm_buf *buf; + drm_savage_buf_priv_t *entry; + int i; + DRM_DEBUG("count=%d\n", dma->buf_count); + + dev_priv->head.next = &dev_priv->tail; + dev_priv->head.prev = NULL; + dev_priv->head.buf = NULL; + + dev_priv->tail.next = NULL; + dev_priv->tail.prev = &dev_priv->head; + dev_priv->tail.buf = NULL; + + for (i = 0; i < dma->buf_count; i++) { + buf = dma->buflist[i]; + entry = buf->dev_private; + + SET_AGE(&entry->age, 0, 0); + entry->buf = buf; + + entry->next = dev_priv->head.next; + entry->prev = &dev_priv->head; + dev_priv->head.next->prev = entry; + dev_priv->head.next = entry; + } + + return 0; +} + +static struct drm_buf *savage_freelist_get(struct drm_device * dev) +{ + drm_savage_private_t *dev_priv = dev->dev_private; + drm_savage_buf_priv_t *tail = dev_priv->tail.prev; + uint16_t event; + unsigned int wrap; + DRM_DEBUG("\n"); + + UPDATE_EVENT_COUNTER(); + if (dev_priv->status_ptr) + event = dev_priv->status_ptr[1] & 0xffff; + else + event = SAVAGE_READ(SAVAGE_STATUS_WORD1) & 0xffff; + wrap = dev_priv->event_wrap; + if (event > dev_priv->event_counter) + wrap--; /* hardware hasn't passed the last wrap yet */ + + DRM_DEBUG(" tail=0x%04x %d\n", tail->age.event, tail->age.wrap); + DRM_DEBUG(" head=0x%04x %d\n", event, wrap); + + if (tail->buf && (TEST_AGE(&tail->age, event, wrap) || event == 0)) { + drm_savage_buf_priv_t *next = tail->next; + drm_savage_buf_priv_t *prev = tail->prev; + prev->next = next; + next->prev = prev; + tail->next = tail->prev = NULL; + return tail->buf; + } + + DRM_DEBUG("returning NULL, tail->buf=%p!\n", tail->buf); + return NULL; +} + +void savage_freelist_put(struct drm_device * dev, struct drm_buf * buf) +{ + drm_savage_private_t *dev_priv = dev->dev_private; + drm_savage_buf_priv_t *entry = buf->dev_private, *prev, *next; + + DRM_DEBUG("age=0x%04x wrap=%d\n", entry->age.event, entry->age.wrap); + + if (entry->next != NULL || entry->prev != NULL) { + DRM_ERROR("entry already on freelist.\n"); + return; + } + + prev = &dev_priv->head; + next = prev->next; + prev->next = entry; + next->prev = entry; + entry->prev = prev; + entry->next = next; +} + +/* + * Command DMA + */ +static int savage_dma_init(drm_savage_private_t * dev_priv) +{ + unsigned int i; + + dev_priv->nr_dma_pages = dev_priv->cmd_dma->size / + (SAVAGE_DMA_PAGE_SIZE * 4); + dev_priv->dma_pages = kmalloc(sizeof(drm_savage_dma_page_t) * + dev_priv->nr_dma_pages, GFP_KERNEL); + if (dev_priv->dma_pages == NULL) + return -ENOMEM; + + for (i = 0; i < dev_priv->nr_dma_pages; ++i) { + SET_AGE(&dev_priv->dma_pages[i].age, 0, 0); + dev_priv->dma_pages[i].used = 0; + dev_priv->dma_pages[i].flushed = 0; + } + SET_AGE(&dev_priv->last_dma_age, 0, 0); + + dev_priv->first_dma_page = 0; + dev_priv->current_dma_page = 0; + + return 0; +} + +void savage_dma_reset(drm_savage_private_t * dev_priv) +{ + uint16_t event; + unsigned int wrap, i; + event = savage_bci_emit_event(dev_priv, 0); + wrap = dev_priv->event_wrap; + for (i = 0; i < dev_priv->nr_dma_pages; ++i) { + SET_AGE(&dev_priv->dma_pages[i].age, event, wrap); + dev_priv->dma_pages[i].used = 0; + dev_priv->dma_pages[i].flushed = 0; + } + SET_AGE(&dev_priv->last_dma_age, event, wrap); + dev_priv->first_dma_page = dev_priv->current_dma_page = 0; +} + +void savage_dma_wait(drm_savage_private_t * dev_priv, unsigned int page) +{ + uint16_t event; + unsigned int wrap; + + /* Faked DMA buffer pages don't age. */ + if (dev_priv->cmd_dma == &dev_priv->fake_dma) + return; + + UPDATE_EVENT_COUNTER(); + if (dev_priv->status_ptr) + event = dev_priv->status_ptr[1] & 0xffff; + else + event = SAVAGE_READ(SAVAGE_STATUS_WORD1) & 0xffff; + wrap = dev_priv->event_wrap; + if (event > dev_priv->event_counter) + wrap--; /* hardware hasn't passed the last wrap yet */ + + if (dev_priv->dma_pages[page].age.wrap > wrap || + (dev_priv->dma_pages[page].age.wrap == wrap && + dev_priv->dma_pages[page].age.event > event)) { + if (dev_priv->wait_evnt(dev_priv, + dev_priv->dma_pages[page].age.event) + < 0) + DRM_ERROR("wait_evnt failed!\n"); + } +} + +uint32_t *savage_dma_alloc(drm_savage_private_t * dev_priv, unsigned int n) +{ + unsigned int cur = dev_priv->current_dma_page; + unsigned int rest = SAVAGE_DMA_PAGE_SIZE - + dev_priv->dma_pages[cur].used; + unsigned int nr_pages = (n - rest + SAVAGE_DMA_PAGE_SIZE - 1) / + SAVAGE_DMA_PAGE_SIZE; + uint32_t *dma_ptr; + unsigned int i; + + DRM_DEBUG("cur=%u, cur->used=%u, n=%u, rest=%u, nr_pages=%u\n", + cur, dev_priv->dma_pages[cur].used, n, rest, nr_pages); + + if (cur + nr_pages < dev_priv->nr_dma_pages) { + dma_ptr = (uint32_t *) dev_priv->cmd_dma->handle + + cur * SAVAGE_DMA_PAGE_SIZE + dev_priv->dma_pages[cur].used; + if (n < rest) + rest = n; + dev_priv->dma_pages[cur].used += rest; + n -= rest; + cur++; + } else { + dev_priv->dma_flush(dev_priv); + nr_pages = + (n + SAVAGE_DMA_PAGE_SIZE - 1) / SAVAGE_DMA_PAGE_SIZE; + for (i = cur; i < dev_priv->nr_dma_pages; ++i) { + dev_priv->dma_pages[i].age = dev_priv->last_dma_age; + dev_priv->dma_pages[i].used = 0; + dev_priv->dma_pages[i].flushed = 0; + } + dma_ptr = (uint32_t *) dev_priv->cmd_dma->handle; + dev_priv->first_dma_page = cur = 0; + } + for (i = cur; nr_pages > 0; ++i, --nr_pages) { +#if SAVAGE_DMA_DEBUG + if (dev_priv->dma_pages[i].used) { + DRM_ERROR("unflushed page %u: used=%u\n", + i, dev_priv->dma_pages[i].used); + } +#endif + if (n > SAVAGE_DMA_PAGE_SIZE) + dev_priv->dma_pages[i].used = SAVAGE_DMA_PAGE_SIZE; + else + dev_priv->dma_pages[i].used = n; + n -= SAVAGE_DMA_PAGE_SIZE; + } + dev_priv->current_dma_page = --i; + + DRM_DEBUG("cur=%u, cur->used=%u, n=%u\n", + i, dev_priv->dma_pages[i].used, n); + + savage_dma_wait(dev_priv, dev_priv->current_dma_page); + + return dma_ptr; +} + +static void savage_dma_flush(drm_savage_private_t * dev_priv) +{ + unsigned int first = dev_priv->first_dma_page; + unsigned int cur = dev_priv->current_dma_page; + uint16_t event; + unsigned int wrap, pad, align, len, i; + unsigned long phys_addr; + BCI_LOCALS; + + if (first == cur && + dev_priv->dma_pages[cur].used == dev_priv->dma_pages[cur].flushed) + return; + + /* pad length to multiples of 2 entries + * align start of next DMA block to multiles of 8 entries */ + pad = -dev_priv->dma_pages[cur].used & 1; + align = -(dev_priv->dma_pages[cur].used + pad) & 7; + + DRM_DEBUG("first=%u, cur=%u, first->flushed=%u, cur->used=%u, " + "pad=%u, align=%u\n", + first, cur, dev_priv->dma_pages[first].flushed, + dev_priv->dma_pages[cur].used, pad, align); + + /* pad with noops */ + if (pad) { + uint32_t *dma_ptr = (uint32_t *) dev_priv->cmd_dma->handle + + cur * SAVAGE_DMA_PAGE_SIZE + dev_priv->dma_pages[cur].used; + dev_priv->dma_pages[cur].used += pad; + while (pad != 0) { + *dma_ptr++ = BCI_CMD_WAIT; + pad--; + } + } + + DRM_MEMORYBARRIER(); + + /* do flush ... */ + phys_addr = dev_priv->cmd_dma->offset + + (first * SAVAGE_DMA_PAGE_SIZE + + dev_priv->dma_pages[first].flushed) * 4; + len = (cur - first) * SAVAGE_DMA_PAGE_SIZE + + dev_priv->dma_pages[cur].used - dev_priv->dma_pages[first].flushed; + + DRM_DEBUG("phys_addr=%lx, len=%u\n", + phys_addr | dev_priv->dma_type, len); + + BEGIN_BCI(3); + BCI_SET_REGISTERS(SAVAGE_DMABUFADDR, 1); + BCI_WRITE(phys_addr | dev_priv->dma_type); + BCI_DMA(len); + + /* fix alignment of the start of the next block */ + dev_priv->dma_pages[cur].used += align; + + /* age DMA pages */ + event = savage_bci_emit_event(dev_priv, 0); + wrap = dev_priv->event_wrap; + for (i = first; i < cur; ++i) { + SET_AGE(&dev_priv->dma_pages[i].age, event, wrap); + dev_priv->dma_pages[i].used = 0; + dev_priv->dma_pages[i].flushed = 0; + } + /* age the current page only when it's full */ + if (dev_priv->dma_pages[cur].used == SAVAGE_DMA_PAGE_SIZE) { + SET_AGE(&dev_priv->dma_pages[cur].age, event, wrap); + dev_priv->dma_pages[cur].used = 0; + dev_priv->dma_pages[cur].flushed = 0; + /* advance to next page */ + cur++; + if (cur == dev_priv->nr_dma_pages) + cur = 0; + dev_priv->first_dma_page = dev_priv->current_dma_page = cur; + } else { + dev_priv->first_dma_page = cur; + dev_priv->dma_pages[cur].flushed = dev_priv->dma_pages[i].used; + } + SET_AGE(&dev_priv->last_dma_age, event, wrap); + + DRM_DEBUG("first=cur=%u, cur->used=%u, cur->flushed=%u\n", cur, + dev_priv->dma_pages[cur].used, + dev_priv->dma_pages[cur].flushed); +} + +static void savage_fake_dma_flush(drm_savage_private_t * dev_priv) +{ + unsigned int i, j; + BCI_LOCALS; + + if (dev_priv->first_dma_page == dev_priv->current_dma_page && + dev_priv->dma_pages[dev_priv->current_dma_page].used == 0) + return; + + DRM_DEBUG("first=%u, cur=%u, cur->used=%u\n", + dev_priv->first_dma_page, dev_priv->current_dma_page, + dev_priv->dma_pages[dev_priv->current_dma_page].used); + + for (i = dev_priv->first_dma_page; + i <= dev_priv->current_dma_page && dev_priv->dma_pages[i].used; + ++i) { + uint32_t *dma_ptr = (uint32_t *) dev_priv->cmd_dma->handle + + i * SAVAGE_DMA_PAGE_SIZE; +#if SAVAGE_DMA_DEBUG + /* Sanity check: all pages except the last one must be full. */ + if (i < dev_priv->current_dma_page && + dev_priv->dma_pages[i].used != SAVAGE_DMA_PAGE_SIZE) { + DRM_ERROR("partial DMA page %u: used=%u", + i, dev_priv->dma_pages[i].used); + } +#endif + BEGIN_BCI(dev_priv->dma_pages[i].used); + for (j = 0; j < dev_priv->dma_pages[i].used; ++j) { + BCI_WRITE(dma_ptr[j]); + } + dev_priv->dma_pages[i].used = 0; + } + + /* reset to first page */ + dev_priv->first_dma_page = dev_priv->current_dma_page = 0; +} + +int savage_driver_load(struct drm_device *dev, unsigned long chipset) +{ + drm_savage_private_t *dev_priv; + + dev_priv = kzalloc(sizeof(drm_savage_private_t), GFP_KERNEL); + if (dev_priv == NULL) + return -ENOMEM; + + dev->dev_private = (void *)dev_priv; + + dev_priv->chipset = (enum savage_family)chipset; + + return 0; +} + + +/* + * Initialize mappings. On Savage4 and SavageIX the alignment + * and size of the aperture is not suitable for automatic MTRR setup + * in drm_addmap. Therefore we add them manually before the maps are + * initialized, and tear them down on last close. + */ +int savage_driver_firstopen(struct drm_device *dev) +{ + drm_savage_private_t *dev_priv = dev->dev_private; + unsigned long mmio_base, fb_base, fb_size, aperture_base; + /* fb_rsrc and aper_rsrc aren't really used currently, but still exist + * in case we decide we need information on the BAR for BSD in the + * future. + */ + unsigned int fb_rsrc, aper_rsrc; + int ret = 0; + + dev_priv->mtrr[0].handle = -1; + dev_priv->mtrr[1].handle = -1; + dev_priv->mtrr[2].handle = -1; + if (S3_SAVAGE3D_SERIES(dev_priv->chipset)) { + fb_rsrc = 0; + fb_base = pci_resource_start(dev->pdev, 0); + fb_size = SAVAGE_FB_SIZE_S3; + mmio_base = fb_base + SAVAGE_FB_SIZE_S3; + aper_rsrc = 0; + aperture_base = fb_base + SAVAGE_APERTURE_OFFSET; + /* this should always be true */ + if (pci_resource_len(dev->pdev, 0) == 0x08000000) { + /* Don't make MMIO write-cobining! We need 3 + * MTRRs. */ + dev_priv->mtrr[0].base = fb_base; + dev_priv->mtrr[0].size = 0x01000000; + dev_priv->mtrr[0].handle = + drm_mtrr_add(dev_priv->mtrr[0].base, + dev_priv->mtrr[0].size, DRM_MTRR_WC); + dev_priv->mtrr[1].base = fb_base + 0x02000000; + dev_priv->mtrr[1].size = 0x02000000; + dev_priv->mtrr[1].handle = + drm_mtrr_add(dev_priv->mtrr[1].base, + dev_priv->mtrr[1].size, DRM_MTRR_WC); + dev_priv->mtrr[2].base = fb_base + 0x04000000; + dev_priv->mtrr[2].size = 0x04000000; + dev_priv->mtrr[2].handle = + drm_mtrr_add(dev_priv->mtrr[2].base, + dev_priv->mtrr[2].size, DRM_MTRR_WC); + } else { + DRM_ERROR("strange pci_resource_len %08llx\n", + (unsigned long long) + pci_resource_len(dev->pdev, 0)); + } + } else if (dev_priv->chipset != S3_SUPERSAVAGE && + dev_priv->chipset != S3_SAVAGE2000) { + mmio_base = pci_resource_start(dev->pdev, 0); + fb_rsrc = 1; + fb_base = pci_resource_start(dev->pdev, 1); + fb_size = SAVAGE_FB_SIZE_S4; + aper_rsrc = 1; + aperture_base = fb_base + SAVAGE_APERTURE_OFFSET; + /* this should always be true */ + if (pci_resource_len(dev->pdev, 1) == 0x08000000) { + /* Can use one MTRR to cover both fb and + * aperture. */ + dev_priv->mtrr[0].base = fb_base; + dev_priv->mtrr[0].size = 0x08000000; + dev_priv->mtrr[0].handle = + drm_mtrr_add(dev_priv->mtrr[0].base, + dev_priv->mtrr[0].size, DRM_MTRR_WC); + } else { + DRM_ERROR("strange pci_resource_len %08llx\n", + (unsigned long long) + pci_resource_len(dev->pdev, 1)); + } + } else { + mmio_base = pci_resource_start(dev->pdev, 0); + fb_rsrc = 1; + fb_base = pci_resource_start(dev->pdev, 1); + fb_size = pci_resource_len(dev->pdev, 1); + aper_rsrc = 2; + aperture_base = pci_resource_start(dev->pdev, 2); + /* Automatic MTRR setup will do the right thing. */ + } + + ret = drm_addmap(dev, mmio_base, SAVAGE_MMIO_SIZE, _DRM_REGISTERS, + _DRM_READ_ONLY, &dev_priv->mmio); + if (ret) + return ret; + + ret = drm_addmap(dev, fb_base, fb_size, _DRM_FRAME_BUFFER, + _DRM_WRITE_COMBINING, &dev_priv->fb); + if (ret) + return ret; + + ret = drm_addmap(dev, aperture_base, SAVAGE_APERTURE_SIZE, + _DRM_FRAME_BUFFER, _DRM_WRITE_COMBINING, + &dev_priv->aperture); + return ret; +} + +/* + * Delete MTRRs and free device-private data. + */ +void savage_driver_lastclose(struct drm_device *dev) +{ + drm_savage_private_t *dev_priv = dev->dev_private; + int i; + + for (i = 0; i < 3; ++i) + if (dev_priv->mtrr[i].handle >= 0) + drm_mtrr_del(dev_priv->mtrr[i].handle, + dev_priv->mtrr[i].base, + dev_priv->mtrr[i].size, DRM_MTRR_WC); +} + +int savage_driver_unload(struct drm_device *dev) +{ + drm_savage_private_t *dev_priv = dev->dev_private; + + kfree(dev_priv); + + return 0; +} + +static int savage_do_init_bci(struct drm_device * dev, drm_savage_init_t * init) +{ + drm_savage_private_t *dev_priv = dev->dev_private; + + if (init->fb_bpp != 16 && init->fb_bpp != 32) { + DRM_ERROR("invalid frame buffer bpp %d!\n", init->fb_bpp); + return -EINVAL; + } + if (init->depth_bpp != 16 && init->depth_bpp != 32) { + DRM_ERROR("invalid depth buffer bpp %d!\n", init->fb_bpp); + return -EINVAL; + } + if (init->dma_type != SAVAGE_DMA_AGP && + init->dma_type != SAVAGE_DMA_PCI) { + DRM_ERROR("invalid dma memory type %d!\n", init->dma_type); + return -EINVAL; + } + + dev_priv->cob_size = init->cob_size; + dev_priv->bci_threshold_lo = init->bci_threshold_lo; + dev_priv->bci_threshold_hi = init->bci_threshold_hi; + dev_priv->dma_type = init->dma_type; + + dev_priv->fb_bpp = init->fb_bpp; + dev_priv->front_offset = init->front_offset; + dev_priv->front_pitch = init->front_pitch; + dev_priv->back_offset = init->back_offset; + dev_priv->back_pitch = init->back_pitch; + dev_priv->depth_bpp = init->depth_bpp; + dev_priv->depth_offset = init->depth_offset; + dev_priv->depth_pitch = init->depth_pitch; + + dev_priv->texture_offset = init->texture_offset; + dev_priv->texture_size = init->texture_size; + + dev_priv->sarea = drm_getsarea(dev); + if (!dev_priv->sarea) { + DRM_ERROR("could not find sarea!\n"); + savage_do_cleanup_bci(dev); + return -EINVAL; + } + if (init->status_offset != 0) { + dev_priv->status = drm_core_findmap(dev, init->status_offset); + if (!dev_priv->status) { + DRM_ERROR("could not find shadow status region!\n"); + savage_do_cleanup_bci(dev); + return -EINVAL; + } + } else { + dev_priv->status = NULL; + } + if (dev_priv->dma_type == SAVAGE_DMA_AGP && init->buffers_offset) { + dev->agp_buffer_token = init->buffers_offset; + dev->agp_buffer_map = drm_core_findmap(dev, + init->buffers_offset); + if (!dev->agp_buffer_map) { + DRM_ERROR("could not find DMA buffer region!\n"); + savage_do_cleanup_bci(dev); + return -EINVAL; + } + drm_core_ioremap(dev->agp_buffer_map, dev); + if (!dev->agp_buffer_map) { + DRM_ERROR("failed to ioremap DMA buffer region!\n"); + savage_do_cleanup_bci(dev); + return -ENOMEM; + } + } + if (init->agp_textures_offset) { + dev_priv->agp_textures = + drm_core_findmap(dev, init->agp_textures_offset); + if (!dev_priv->agp_textures) { + DRM_ERROR("could not find agp texture region!\n"); + savage_do_cleanup_bci(dev); + return -EINVAL; + } + } else { + dev_priv->agp_textures = NULL; + } + + if (init->cmd_dma_offset) { + if (S3_SAVAGE3D_SERIES(dev_priv->chipset)) { + DRM_ERROR("command DMA not supported on " + "Savage3D/MX/IX.\n"); + savage_do_cleanup_bci(dev); + return -EINVAL; + } + if (dev->dma && dev->dma->buflist) { + DRM_ERROR("command and vertex DMA not supported " + "at the same time.\n"); + savage_do_cleanup_bci(dev); + return -EINVAL; + } + dev_priv->cmd_dma = drm_core_findmap(dev, init->cmd_dma_offset); + if (!dev_priv->cmd_dma) { + DRM_ERROR("could not find command DMA region!\n"); + savage_do_cleanup_bci(dev); + return -EINVAL; + } + if (dev_priv->dma_type == SAVAGE_DMA_AGP) { + if (dev_priv->cmd_dma->type != _DRM_AGP) { + DRM_ERROR("AGP command DMA region is not a " + "_DRM_AGP map!\n"); + savage_do_cleanup_bci(dev); + return -EINVAL; + } + drm_core_ioremap(dev_priv->cmd_dma, dev); + if (!dev_priv->cmd_dma->handle) { + DRM_ERROR("failed to ioremap command " + "DMA region!\n"); + savage_do_cleanup_bci(dev); + return -ENOMEM; + } + } else if (dev_priv->cmd_dma->type != _DRM_CONSISTENT) { + DRM_ERROR("PCI command DMA region is not a " + "_DRM_CONSISTENT map!\n"); + savage_do_cleanup_bci(dev); + return -EINVAL; + } + } else { + dev_priv->cmd_dma = NULL; + } + + dev_priv->dma_flush = savage_dma_flush; + if (!dev_priv->cmd_dma) { + DRM_DEBUG("falling back to faked command DMA.\n"); + dev_priv->fake_dma.offset = 0; + dev_priv->fake_dma.size = SAVAGE_FAKE_DMA_SIZE; + dev_priv->fake_dma.type = _DRM_SHM; + dev_priv->fake_dma.handle = kmalloc(SAVAGE_FAKE_DMA_SIZE, + GFP_KERNEL); + if (!dev_priv->fake_dma.handle) { + DRM_ERROR("could not allocate faked DMA buffer!\n"); + savage_do_cleanup_bci(dev); + return -ENOMEM; + } + dev_priv->cmd_dma = &dev_priv->fake_dma; + dev_priv->dma_flush = savage_fake_dma_flush; + } + + dev_priv->sarea_priv = + (drm_savage_sarea_t *) ((uint8_t *) dev_priv->sarea->handle + + init->sarea_priv_offset); + + /* setup bitmap descriptors */ + { + unsigned int color_tile_format; + unsigned int depth_tile_format; + unsigned int front_stride, back_stride, depth_stride; + if (dev_priv->chipset <= S3_SAVAGE4) { + color_tile_format = dev_priv->fb_bpp == 16 ? + SAVAGE_BD_TILE_16BPP : SAVAGE_BD_TILE_32BPP; + depth_tile_format = dev_priv->depth_bpp == 16 ? + SAVAGE_BD_TILE_16BPP : SAVAGE_BD_TILE_32BPP; + } else { + color_tile_format = SAVAGE_BD_TILE_DEST; + depth_tile_format = SAVAGE_BD_TILE_DEST; + } + front_stride = dev_priv->front_pitch / (dev_priv->fb_bpp / 8); + back_stride = dev_priv->back_pitch / (dev_priv->fb_bpp / 8); + depth_stride = + dev_priv->depth_pitch / (dev_priv->depth_bpp / 8); + + dev_priv->front_bd = front_stride | SAVAGE_BD_BW_DISABLE | + (dev_priv->fb_bpp << SAVAGE_BD_BPP_SHIFT) | + (color_tile_format << SAVAGE_BD_TILE_SHIFT); + + dev_priv->back_bd = back_stride | SAVAGE_BD_BW_DISABLE | + (dev_priv->fb_bpp << SAVAGE_BD_BPP_SHIFT) | + (color_tile_format << SAVAGE_BD_TILE_SHIFT); + + dev_priv->depth_bd = depth_stride | SAVAGE_BD_BW_DISABLE | + (dev_priv->depth_bpp << SAVAGE_BD_BPP_SHIFT) | + (depth_tile_format << SAVAGE_BD_TILE_SHIFT); + } + + /* setup status and bci ptr */ + dev_priv->event_counter = 0; + dev_priv->event_wrap = 0; + dev_priv->bci_ptr = (volatile uint32_t *) + ((uint8_t *) dev_priv->mmio->handle + SAVAGE_BCI_OFFSET); + if (S3_SAVAGE3D_SERIES(dev_priv->chipset)) { + dev_priv->status_used_mask = SAVAGE_FIFO_USED_MASK_S3D; + } else { + dev_priv->status_used_mask = SAVAGE_FIFO_USED_MASK_S4; + } + if (dev_priv->status != NULL) { + dev_priv->status_ptr = + (volatile uint32_t *)dev_priv->status->handle; + dev_priv->wait_fifo = savage_bci_wait_fifo_shadow; + dev_priv->wait_evnt = savage_bci_wait_event_shadow; + dev_priv->status_ptr[1023] = dev_priv->event_counter; + } else { + dev_priv->status_ptr = NULL; + if (S3_SAVAGE3D_SERIES(dev_priv->chipset)) { + dev_priv->wait_fifo = savage_bci_wait_fifo_s3d; + } else { + dev_priv->wait_fifo = savage_bci_wait_fifo_s4; + } + dev_priv->wait_evnt = savage_bci_wait_event_reg; + } + + /* cliprect functions */ + if (S3_SAVAGE3D_SERIES(dev_priv->chipset)) + dev_priv->emit_clip_rect = savage_emit_clip_rect_s3d; + else + dev_priv->emit_clip_rect = savage_emit_clip_rect_s4; + + if (savage_freelist_init(dev) < 0) { + DRM_ERROR("could not initialize freelist\n"); + savage_do_cleanup_bci(dev); + return -ENOMEM; + } + + if (savage_dma_init(dev_priv) < 0) { + DRM_ERROR("could not initialize command DMA\n"); + savage_do_cleanup_bci(dev); + return -ENOMEM; + } + + return 0; +} + +static int savage_do_cleanup_bci(struct drm_device * dev) +{ + drm_savage_private_t *dev_priv = dev->dev_private; + + if (dev_priv->cmd_dma == &dev_priv->fake_dma) { + kfree(dev_priv->fake_dma.handle); + } else if (dev_priv->cmd_dma && dev_priv->cmd_dma->handle && + dev_priv->cmd_dma->type == _DRM_AGP && + dev_priv->dma_type == SAVAGE_DMA_AGP) + drm_core_ioremapfree(dev_priv->cmd_dma, dev); + + if (dev_priv->dma_type == SAVAGE_DMA_AGP && + dev->agp_buffer_map && dev->agp_buffer_map->handle) { + drm_core_ioremapfree(dev->agp_buffer_map, dev); + /* make sure the next instance (which may be running + * in PCI mode) doesn't try to use an old + * agp_buffer_map. */ + dev->agp_buffer_map = NULL; + } + + kfree(dev_priv->dma_pages); + + return 0; +} + +static int savage_bci_init(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_savage_init_t *init = data; + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + switch (init->func) { + case SAVAGE_INIT_BCI: + return savage_do_init_bci(dev, init); + case SAVAGE_CLEANUP_BCI: + return savage_do_cleanup_bci(dev); + } + + return -EINVAL; +} + +static int savage_bci_event_emit(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_savage_private_t *dev_priv = dev->dev_private; + drm_savage_event_emit_t *event = data; + + DRM_DEBUG("\n"); + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + event->count = savage_bci_emit_event(dev_priv, event->flags); + event->count |= dev_priv->event_wrap << 16; + + return 0; +} + +static int savage_bci_event_wait(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_savage_private_t *dev_priv = dev->dev_private; + drm_savage_event_wait_t *event = data; + unsigned int event_e, hw_e; + unsigned int event_w, hw_w; + + DRM_DEBUG("\n"); + + UPDATE_EVENT_COUNTER(); + if (dev_priv->status_ptr) + hw_e = dev_priv->status_ptr[1] & 0xffff; + else + hw_e = SAVAGE_READ(SAVAGE_STATUS_WORD1) & 0xffff; + hw_w = dev_priv->event_wrap; + if (hw_e > dev_priv->event_counter) + hw_w--; /* hardware hasn't passed the last wrap yet */ + + event_e = event->count & 0xffff; + event_w = event->count >> 16; + + /* Don't need to wait if + * - event counter wrapped since the event was emitted or + * - the hardware has advanced up to or over the event to wait for. + */ + if (event_w < hw_w || (event_w == hw_w && event_e <= hw_e)) + return 0; + else + return dev_priv->wait_evnt(dev_priv, event_e); +} + +/* + * DMA buffer management + */ + +static int savage_bci_get_buffers(struct drm_device *dev, + struct drm_file *file_priv, + struct drm_dma *d) +{ + struct drm_buf *buf; + int i; + + for (i = d->granted_count; i < d->request_count; i++) { + buf = savage_freelist_get(dev); + if (!buf) + return -EAGAIN; + + buf->file_priv = file_priv; + + if (DRM_COPY_TO_USER(&d->request_indices[i], + &buf->idx, sizeof(buf->idx))) + return -EFAULT; + if (DRM_COPY_TO_USER(&d->request_sizes[i], + &buf->total, sizeof(buf->total))) + return -EFAULT; + + d->granted_count++; + } + return 0; +} + +int savage_bci_buffers(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + struct drm_device_dma *dma = dev->dma; + struct drm_dma *d = data; + int ret = 0; + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + /* Please don't send us buffers. + */ + if (d->send_count != 0) { + DRM_ERROR("Process %d trying to send %d buffers via drmDMA\n", + DRM_CURRENTPID, d->send_count); + return -EINVAL; + } + + /* We'll send you buffers. + */ + if (d->request_count < 0 || d->request_count > dma->buf_count) { + DRM_ERROR("Process %d trying to get %d buffers (of %d max)\n", + DRM_CURRENTPID, d->request_count, dma->buf_count); + return -EINVAL; + } + + d->granted_count = 0; + + if (d->request_count) { + ret = savage_bci_get_buffers(dev, file_priv, d); + } + + return ret; +} + +void savage_reclaim_buffers(struct drm_device *dev, struct drm_file *file_priv) +{ + struct drm_device_dma *dma = dev->dma; + drm_savage_private_t *dev_priv = dev->dev_private; + int i; + + if (!dma) + return; + if (!dev_priv) + return; + if (!dma->buflist) + return; + + /*i830_flush_queue(dev); */ + + for (i = 0; i < dma->buf_count; i++) { + struct drm_buf *buf = dma->buflist[i]; + drm_savage_buf_priv_t *buf_priv = buf->dev_private; + + if (buf->file_priv == file_priv && buf_priv && + buf_priv->next == NULL && buf_priv->prev == NULL) { + uint16_t event; + DRM_DEBUG("reclaimed from client\n"); + event = savage_bci_emit_event(dev_priv, SAVAGE_WAIT_3D); + SET_AGE(&buf_priv->age, event, dev_priv->event_wrap); + savage_freelist_put(dev, buf); + } + } + + drm_core_reclaim_buffers(dev, file_priv); +} + +struct drm_ioctl_desc savage_ioctls[] = { + DRM_IOCTL_DEF_DRV(SAVAGE_BCI_INIT, savage_bci_init, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY), + DRM_IOCTL_DEF_DRV(SAVAGE_BCI_CMDBUF, savage_bci_cmdbuf, DRM_AUTH), + DRM_IOCTL_DEF_DRV(SAVAGE_BCI_EVENT_EMIT, savage_bci_event_emit, DRM_AUTH), + DRM_IOCTL_DEF_DRV(SAVAGE_BCI_EVENT_WAIT, savage_bci_event_wait, DRM_AUTH), +}; + +int savage_max_ioctl = DRM_ARRAY_SIZE(savage_ioctls); diff --git a/drivers/gpu/drm/savage/savage_drv.c b/drivers/gpu/drm/savage/savage_drv.c new file mode 100644 index 00000000..6464490b --- /dev/null +++ b/drivers/gpu/drm/savage/savage_drv.c @@ -0,0 +1,87 @@ +/* savage_drv.c -- Savage driver for Linux + * + * Copyright 2004 Felix Kuehling + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sub license, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NON-INFRINGEMENT. IN NO EVENT SHALL FELIX KUEHLING BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF + * CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + */ + +#include "drmP.h" +#include "savage_drm.h" +#include "savage_drv.h" + +#include "drm_pciids.h" + +static struct pci_device_id pciidlist[] = { + savage_PCI_IDS +}; + +static struct drm_driver driver = { + .driver_features = + DRIVER_USE_AGP | DRIVER_USE_MTRR | DRIVER_HAVE_DMA | DRIVER_PCI_DMA, + .dev_priv_size = sizeof(drm_savage_buf_priv_t), + .load = savage_driver_load, + .firstopen = savage_driver_firstopen, + .lastclose = savage_driver_lastclose, + .unload = savage_driver_unload, + .reclaim_buffers = savage_reclaim_buffers, + .ioctls = savage_ioctls, + .dma_ioctl = savage_bci_buffers, + .fops = { + .owner = THIS_MODULE, + .open = drm_open, + .release = drm_release, + .unlocked_ioctl = drm_ioctl, + .mmap = drm_mmap, + .poll = drm_poll, + .fasync = drm_fasync, + .llseek = noop_llseek, + }, + + .name = DRIVER_NAME, + .desc = DRIVER_DESC, + .date = DRIVER_DATE, + .major = DRIVER_MAJOR, + .minor = DRIVER_MINOR, + .patchlevel = DRIVER_PATCHLEVEL, +}; + +static struct pci_driver savage_pci_driver = { + .name = DRIVER_NAME, + .id_table = pciidlist, +}; + +static int __init savage_init(void) +{ + driver.num_ioctls = savage_max_ioctl; + return drm_pci_init(&driver, &savage_pci_driver); +} + +static void __exit savage_exit(void) +{ + drm_pci_exit(&driver, &savage_pci_driver); +} + +module_init(savage_init); +module_exit(savage_exit); + +MODULE_AUTHOR(DRIVER_AUTHOR); +MODULE_DESCRIPTION(DRIVER_DESC); +MODULE_LICENSE("GPL and additional rights"); diff --git a/drivers/gpu/drm/savage/savage_drv.h b/drivers/gpu/drm/savage/savage_drv.h new file mode 100644 index 00000000..df2aac66 --- /dev/null +++ b/drivers/gpu/drm/savage/savage_drv.h @@ -0,0 +1,575 @@ +/* savage_drv.h -- Private header for the savage driver */ +/* + * Copyright 2004 Felix Kuehling + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sub license, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NON-INFRINGEMENT. IN NO EVENT SHALL FELIX KUEHLING BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF + * CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + */ + +#ifndef __SAVAGE_DRV_H__ +#define __SAVAGE_DRV_H__ + +#define DRIVER_AUTHOR "Felix Kuehling" + +#define DRIVER_NAME "savage" +#define DRIVER_DESC "Savage3D/MX/IX, Savage4, SuperSavage, Twister, ProSavage[DDR]" +#define DRIVER_DATE "20050313" + +#define DRIVER_MAJOR 2 +#define DRIVER_MINOR 4 +#define DRIVER_PATCHLEVEL 1 +/* Interface history: + * + * 1.x The DRM driver from the VIA/S3 code drop, basically a dummy + * 2.0 The first real DRM + * 2.1 Scissors registers managed by the DRM, 3D operations clipped by + * cliprects of the cmdbuf ioctl + * 2.2 Implemented SAVAGE_CMD_DMA_IDX and SAVAGE_CMD_VB_IDX + * 2.3 Event counters used by BCI_EVENT_EMIT/WAIT ioctls are now 32 bits + * wide and thus very long lived (unlikely to ever wrap). The size + * in the struct was 32 bits before, but only 16 bits were used + * 2.4 Implemented command DMA. Now drm_savage_init_t.cmd_dma_offset is + * actually used + */ + +typedef struct drm_savage_age { + uint16_t event; + unsigned int wrap; +} drm_savage_age_t; + +typedef struct drm_savage_buf_priv { + struct drm_savage_buf_priv *next; + struct drm_savage_buf_priv *prev; + drm_savage_age_t age; + struct drm_buf *buf; +} drm_savage_buf_priv_t; + +typedef struct drm_savage_dma_page { + drm_savage_age_t age; + unsigned int used, flushed; +} drm_savage_dma_page_t; +#define SAVAGE_DMA_PAGE_SIZE 1024 /* in dwords */ +/* Fake DMA buffer size in bytes. 4 pages. Allows a maximum command + * size of 16kbytes or 4k entries. Minimum requirement would be + * 10kbytes for 255 40-byte vertices in one drawing command. */ +#define SAVAGE_FAKE_DMA_SIZE (SAVAGE_DMA_PAGE_SIZE*4*4) + +/* interesting bits of hardware state that are saved in dev_priv */ +typedef union { + struct drm_savage_common_state { + uint32_t vbaddr; + } common; + struct { + unsigned char pad[sizeof(struct drm_savage_common_state)]; + uint32_t texctrl, texaddr; + uint32_t scstart, new_scstart; + uint32_t scend, new_scend; + } s3d; + struct { + unsigned char pad[sizeof(struct drm_savage_common_state)]; + uint32_t texdescr, texaddr0, texaddr1; + uint32_t drawctrl0, new_drawctrl0; + uint32_t drawctrl1, new_drawctrl1; + } s4; +} drm_savage_state_t; + +/* these chip tags should match the ones in the 2D driver in savage_regs.h. */ +enum savage_family { + S3_UNKNOWN = 0, + S3_SAVAGE3D, + S3_SAVAGE_MX, + S3_SAVAGE4, + S3_PROSAVAGE, + S3_TWISTER, + S3_PROSAVAGEDDR, + S3_SUPERSAVAGE, + S3_SAVAGE2000, + S3_LAST +}; + +extern struct drm_ioctl_desc savage_ioctls[]; +extern int savage_max_ioctl; + +#define S3_SAVAGE3D_SERIES(chip) ((chip>=S3_SAVAGE3D) && (chip<=S3_SAVAGE_MX)) + +#define S3_SAVAGE4_SERIES(chip) ((chip==S3_SAVAGE4) \ + || (chip==S3_PROSAVAGE) \ + || (chip==S3_TWISTER) \ + || (chip==S3_PROSAVAGEDDR)) + +#define S3_SAVAGE_MOBILE_SERIES(chip) ((chip==S3_SAVAGE_MX) || (chip==S3_SUPERSAVAGE)) + +#define S3_SAVAGE_SERIES(chip) ((chip>=S3_SAVAGE3D) && (chip<=S3_SAVAGE2000)) + +#define S3_MOBILE_TWISTER_SERIES(chip) ((chip==S3_TWISTER) \ + ||(chip==S3_PROSAVAGEDDR)) + +/* flags */ +#define SAVAGE_IS_AGP 1 + +typedef struct drm_savage_private { + drm_savage_sarea_t *sarea_priv; + + drm_savage_buf_priv_t head, tail; + + /* who am I? */ + enum savage_family chipset; + + unsigned int cob_size; + unsigned int bci_threshold_lo, bci_threshold_hi; + unsigned int dma_type; + + /* frame buffer layout */ + unsigned int fb_bpp; + unsigned int front_offset, front_pitch; + unsigned int back_offset, back_pitch; + unsigned int depth_bpp; + unsigned int depth_offset, depth_pitch; + + /* bitmap descriptors for swap and clear */ + unsigned int front_bd, back_bd, depth_bd; + + /* local textures */ + unsigned int texture_offset; + unsigned int texture_size; + + /* memory regions in physical memory */ + drm_local_map_t *sarea; + drm_local_map_t *mmio; + drm_local_map_t *fb; + drm_local_map_t *aperture; + drm_local_map_t *status; + drm_local_map_t *agp_textures; + drm_local_map_t *cmd_dma; + drm_local_map_t fake_dma; + + struct { + int handle; + unsigned long base, size; + } mtrr[3]; + + /* BCI and status-related stuff */ + volatile uint32_t *status_ptr, *bci_ptr; + uint32_t status_used_mask; + uint16_t event_counter; + unsigned int event_wrap; + + /* Savage4 command DMA */ + drm_savage_dma_page_t *dma_pages; + unsigned int nr_dma_pages, first_dma_page, current_dma_page; + drm_savage_age_t last_dma_age; + + /* saved hw state for global/local check on S3D */ + uint32_t hw_draw_ctrl, hw_zbuf_ctrl; + /* and for scissors (global, so don't emit if not changed) */ + uint32_t hw_scissors_start, hw_scissors_end; + + drm_savage_state_t state; + + /* after emitting a wait cmd Savage3D needs 63 nops before next DMA */ + unsigned int waiting; + + /* config/hardware-dependent function pointers */ + int (*wait_fifo) (struct drm_savage_private * dev_priv, unsigned int n); + int (*wait_evnt) (struct drm_savage_private * dev_priv, uint16_t e); + /* Err, there is a macro wait_event in include/linux/wait.h. + * Avoid unwanted macro expansion. */ + void (*emit_clip_rect) (struct drm_savage_private * dev_priv, + const struct drm_clip_rect * pbox); + void (*dma_flush) (struct drm_savage_private * dev_priv); +} drm_savage_private_t; + +/* ioctls */ +extern int savage_bci_cmdbuf(struct drm_device *dev, void *data, struct drm_file *file_priv); +extern int savage_bci_buffers(struct drm_device *dev, void *data, struct drm_file *file_priv); + +/* BCI functions */ +extern uint16_t savage_bci_emit_event(drm_savage_private_t * dev_priv, + unsigned int flags); +extern void savage_freelist_put(struct drm_device * dev, struct drm_buf * buf); +extern void savage_dma_reset(drm_savage_private_t * dev_priv); +extern void savage_dma_wait(drm_savage_private_t * dev_priv, unsigned int page); +extern uint32_t *savage_dma_alloc(drm_savage_private_t * dev_priv, + unsigned int n); +extern int savage_driver_load(struct drm_device *dev, unsigned long chipset); +extern int savage_driver_firstopen(struct drm_device *dev); +extern void savage_driver_lastclose(struct drm_device *dev); +extern int savage_driver_unload(struct drm_device *dev); +extern void savage_reclaim_buffers(struct drm_device *dev, + struct drm_file *file_priv); + +/* state functions */ +extern void savage_emit_clip_rect_s3d(drm_savage_private_t * dev_priv, + const struct drm_clip_rect * pbox); +extern void savage_emit_clip_rect_s4(drm_savage_private_t * dev_priv, + const struct drm_clip_rect * pbox); + +#define SAVAGE_FB_SIZE_S3 0x01000000 /* 16MB */ +#define SAVAGE_FB_SIZE_S4 0x02000000 /* 32MB */ +#define SAVAGE_MMIO_SIZE 0x00080000 /* 512kB */ +#define SAVAGE_APERTURE_OFFSET 0x02000000 /* 32MB */ +#define SAVAGE_APERTURE_SIZE 0x05000000 /* 5 tiled surfaces, 16MB each */ + +#define SAVAGE_BCI_OFFSET 0x00010000 /* offset of the BCI region + * inside the MMIO region */ +#define SAVAGE_BCI_FIFO_SIZE 32 /* number of entries in on-chip + * BCI FIFO */ + +/* + * MMIO registers + */ +#define SAVAGE_STATUS_WORD0 0x48C00 +#define SAVAGE_STATUS_WORD1 0x48C04 +#define SAVAGE_ALT_STATUS_WORD0 0x48C60 + +#define SAVAGE_FIFO_USED_MASK_S3D 0x0001ffff +#define SAVAGE_FIFO_USED_MASK_S4 0x001fffff + +/* Copied from savage_bci.h in the 2D driver with some renaming. */ + +/* Bitmap descriptors */ +#define SAVAGE_BD_STRIDE_SHIFT 0 +#define SAVAGE_BD_BPP_SHIFT 16 +#define SAVAGE_BD_TILE_SHIFT 24 +#define SAVAGE_BD_BW_DISABLE (1<<28) +/* common: */ +#define SAVAGE_BD_TILE_LINEAR 0 +/* savage4, MX, IX, 3D */ +#define SAVAGE_BD_TILE_16BPP 2 +#define SAVAGE_BD_TILE_32BPP 3 +/* twister, prosavage, DDR, supersavage, 2000 */ +#define SAVAGE_BD_TILE_DEST 1 +#define SAVAGE_BD_TILE_TEXTURE 2 +/* GBD - BCI enable */ +/* savage4, MX, IX, 3D */ +#define SAVAGE_GBD_BCI_ENABLE 8 +/* twister, prosavage, DDR, supersavage, 2000 */ +#define SAVAGE_GBD_BCI_ENABLE_TWISTER 0 + +#define SAVAGE_GBD_BIG_ENDIAN 4 +#define SAVAGE_GBD_LITTLE_ENDIAN 0 +#define SAVAGE_GBD_64 1 + +/* Global Bitmap Descriptor */ +#define SAVAGE_BCI_GLB_BD_LOW 0x8168 +#define SAVAGE_BCI_GLB_BD_HIGH 0x816C + +/* + * BCI registers + */ +/* Savage4/Twister/ProSavage 3D registers */ +#define SAVAGE_DRAWLOCALCTRL_S4 0x1e +#define SAVAGE_TEXPALADDR_S4 0x1f +#define SAVAGE_TEXCTRL0_S4 0x20 +#define SAVAGE_TEXCTRL1_S4 0x21 +#define SAVAGE_TEXADDR0_S4 0x22 +#define SAVAGE_TEXADDR1_S4 0x23 +#define SAVAGE_TEXBLEND0_S4 0x24 +#define SAVAGE_TEXBLEND1_S4 0x25 +#define SAVAGE_TEXXPRCLR_S4 0x26 /* never used */ +#define SAVAGE_TEXDESCR_S4 0x27 +#define SAVAGE_FOGTABLE_S4 0x28 +#define SAVAGE_FOGCTRL_S4 0x30 +#define SAVAGE_STENCILCTRL_S4 0x31 +#define SAVAGE_ZBUFCTRL_S4 0x32 +#define SAVAGE_ZBUFOFF_S4 0x33 +#define SAVAGE_DESTCTRL_S4 0x34 +#define SAVAGE_DRAWCTRL0_S4 0x35 +#define SAVAGE_DRAWCTRL1_S4 0x36 +#define SAVAGE_ZWATERMARK_S4 0x37 +#define SAVAGE_DESTTEXRWWATERMARK_S4 0x38 +#define SAVAGE_TEXBLENDCOLOR_S4 0x39 +/* Savage3D/MX/IX 3D registers */ +#define SAVAGE_TEXPALADDR_S3D 0x18 +#define SAVAGE_TEXXPRCLR_S3D 0x19 /* never used */ +#define SAVAGE_TEXADDR_S3D 0x1A +#define SAVAGE_TEXDESCR_S3D 0x1B +#define SAVAGE_TEXCTRL_S3D 0x1C +#define SAVAGE_FOGTABLE_S3D 0x20 +#define SAVAGE_FOGCTRL_S3D 0x30 +#define SAVAGE_DRAWCTRL_S3D 0x31 +#define SAVAGE_ZBUFCTRL_S3D 0x32 +#define SAVAGE_ZBUFOFF_S3D 0x33 +#define SAVAGE_DESTCTRL_S3D 0x34 +#define SAVAGE_SCSTART_S3D 0x35 +#define SAVAGE_SCEND_S3D 0x36 +#define SAVAGE_ZWATERMARK_S3D 0x37 +#define SAVAGE_DESTTEXRWWATERMARK_S3D 0x38 +/* common stuff */ +#define SAVAGE_VERTBUFADDR 0x3e +#define SAVAGE_BITPLANEWTMASK 0xd7 +#define SAVAGE_DMABUFADDR 0x51 + +/* texture enable bits (needed for tex addr checking) */ +#define SAVAGE_TEXCTRL_TEXEN_MASK 0x00010000 /* S3D */ +#define SAVAGE_TEXDESCR_TEX0EN_MASK 0x02000000 /* S4 */ +#define SAVAGE_TEXDESCR_TEX1EN_MASK 0x04000000 /* S4 */ + +/* Global fields in Savage4/Twister/ProSavage 3D registers: + * + * All texture registers and DrawLocalCtrl are local. All other + * registers are global. */ + +/* Global fields in Savage3D/MX/IX 3D registers: + * + * All texture registers are local. DrawCtrl and ZBufCtrl are + * partially local. All other registers are global. + * + * DrawCtrl global fields: cullMode, alphaTestCmpFunc, alphaTestEn, alphaRefVal + * ZBufCtrl global fields: zCmpFunc, zBufEn + */ +#define SAVAGE_DRAWCTRL_S3D_GLOBAL 0x03f3c00c +#define SAVAGE_ZBUFCTRL_S3D_GLOBAL 0x00000027 + +/* Masks for scissor bits (drawCtrl[01] on s4, scissorStart/End on s3d) + */ +#define SAVAGE_SCISSOR_MASK_S4 0x00fff7ff +#define SAVAGE_SCISSOR_MASK_S3D 0x07ff07ff + +/* + * BCI commands + */ +#define BCI_CMD_NOP 0x40000000 +#define BCI_CMD_RECT 0x48000000 +#define BCI_CMD_RECT_XP 0x01000000 +#define BCI_CMD_RECT_YP 0x02000000 +#define BCI_CMD_SCANLINE 0x50000000 +#define BCI_CMD_LINE 0x5C000000 +#define BCI_CMD_LINE_LAST_PIXEL 0x58000000 +#define BCI_CMD_BYTE_TEXT 0x63000000 +#define BCI_CMD_NT_BYTE_TEXT 0x67000000 +#define BCI_CMD_BIT_TEXT 0x6C000000 +#define BCI_CMD_GET_ROP(cmd) (((cmd) >> 16) & 0xFF) +#define BCI_CMD_SET_ROP(cmd, rop) ((cmd) |= ((rop & 0xFF) << 16)) +#define BCI_CMD_SEND_COLOR 0x00008000 + +#define BCI_CMD_CLIP_NONE 0x00000000 +#define BCI_CMD_CLIP_CURRENT 0x00002000 +#define BCI_CMD_CLIP_LR 0x00004000 +#define BCI_CMD_CLIP_NEW 0x00006000 + +#define BCI_CMD_DEST_GBD 0x00000000 +#define BCI_CMD_DEST_PBD 0x00000800 +#define BCI_CMD_DEST_PBD_NEW 0x00000C00 +#define BCI_CMD_DEST_SBD 0x00001000 +#define BCI_CMD_DEST_SBD_NEW 0x00001400 + +#define BCI_CMD_SRC_TRANSPARENT 0x00000200 +#define BCI_CMD_SRC_SOLID 0x00000000 +#define BCI_CMD_SRC_GBD 0x00000020 +#define BCI_CMD_SRC_COLOR 0x00000040 +#define BCI_CMD_SRC_MONO 0x00000060 +#define BCI_CMD_SRC_PBD_COLOR 0x00000080 +#define BCI_CMD_SRC_PBD_MONO 0x000000A0 +#define BCI_CMD_SRC_PBD_COLOR_NEW 0x000000C0 +#define BCI_CMD_SRC_PBD_MONO_NEW 0x000000E0 +#define BCI_CMD_SRC_SBD_COLOR 0x00000100 +#define BCI_CMD_SRC_SBD_MONO 0x00000120 +#define BCI_CMD_SRC_SBD_COLOR_NEW 0x00000140 +#define BCI_CMD_SRC_SBD_MONO_NEW 0x00000160 + +#define BCI_CMD_PAT_TRANSPARENT 0x00000010 +#define BCI_CMD_PAT_NONE 0x00000000 +#define BCI_CMD_PAT_COLOR 0x00000002 +#define BCI_CMD_PAT_MONO 0x00000003 +#define BCI_CMD_PAT_PBD_COLOR 0x00000004 +#define BCI_CMD_PAT_PBD_MONO 0x00000005 +#define BCI_CMD_PAT_PBD_COLOR_NEW 0x00000006 +#define BCI_CMD_PAT_PBD_MONO_NEW 0x00000007 +#define BCI_CMD_PAT_SBD_COLOR 0x00000008 +#define BCI_CMD_PAT_SBD_MONO 0x00000009 +#define BCI_CMD_PAT_SBD_COLOR_NEW 0x0000000A +#define BCI_CMD_PAT_SBD_MONO_NEW 0x0000000B + +#define BCI_BD_BW_DISABLE 0x10000000 +#define BCI_BD_TILE_MASK 0x03000000 +#define BCI_BD_TILE_NONE 0x00000000 +#define BCI_BD_TILE_16 0x02000000 +#define BCI_BD_TILE_32 0x03000000 +#define BCI_BD_GET_BPP(bd) (((bd) >> 16) & 0xFF) +#define BCI_BD_SET_BPP(bd, bpp) ((bd) |= (((bpp) & 0xFF) << 16)) +#define BCI_BD_GET_STRIDE(bd) ((bd) & 0xFFFF) +#define BCI_BD_SET_STRIDE(bd, st) ((bd) |= ((st) & 0xFFFF)) + +#define BCI_CMD_SET_REGISTER 0x96000000 + +#define BCI_CMD_WAIT 0xC0000000 +#define BCI_CMD_WAIT_3D 0x00010000 +#define BCI_CMD_WAIT_2D 0x00020000 + +#define BCI_CMD_UPDATE_EVENT_TAG 0x98000000 + +#define BCI_CMD_DRAW_PRIM 0x80000000 +#define BCI_CMD_DRAW_INDEXED_PRIM 0x88000000 +#define BCI_CMD_DRAW_CONT 0x01000000 +#define BCI_CMD_DRAW_TRILIST 0x00000000 +#define BCI_CMD_DRAW_TRISTRIP 0x02000000 +#define BCI_CMD_DRAW_TRIFAN 0x04000000 +#define BCI_CMD_DRAW_SKIPFLAGS 0x000000ff +#define BCI_CMD_DRAW_NO_Z 0x00000001 +#define BCI_CMD_DRAW_NO_W 0x00000002 +#define BCI_CMD_DRAW_NO_CD 0x00000004 +#define BCI_CMD_DRAW_NO_CS 0x00000008 +#define BCI_CMD_DRAW_NO_U0 0x00000010 +#define BCI_CMD_DRAW_NO_V0 0x00000020 +#define BCI_CMD_DRAW_NO_UV0 0x00000030 +#define BCI_CMD_DRAW_NO_U1 0x00000040 +#define BCI_CMD_DRAW_NO_V1 0x00000080 +#define BCI_CMD_DRAW_NO_UV1 0x000000c0 + +#define BCI_CMD_DMA 0xa8000000 + +#define BCI_W_H(w, h) ((((h) << 16) | (w)) & 0x0FFF0FFF) +#define BCI_X_Y(x, y) ((((y) << 16) | (x)) & 0x0FFF0FFF) +#define BCI_X_W(x, y) ((((w) << 16) | (x)) & 0x0FFF0FFF) +#define BCI_CLIP_LR(l, r) ((((r) << 16) | (l)) & 0x0FFF0FFF) +#define BCI_CLIP_TL(t, l) ((((t) << 16) | (l)) & 0x0FFF0FFF) +#define BCI_CLIP_BR(b, r) ((((b) << 16) | (r)) & 0x0FFF0FFF) + +#define BCI_LINE_X_Y(x, y) (((y) << 16) | ((x) & 0xFFFF)) +#define BCI_LINE_STEPS(diag, axi) (((axi) << 16) | ((diag) & 0xFFFF)) +#define BCI_LINE_MISC(maj, ym, xp, yp, err) \ + (((maj) & 0x1FFF) | \ + ((ym) ? 1<<13 : 0) | \ + ((xp) ? 1<<14 : 0) | \ + ((yp) ? 1<<15 : 0) | \ + ((err) << 16)) + +/* + * common commands + */ +#define BCI_SET_REGISTERS( first, n ) \ + BCI_WRITE(BCI_CMD_SET_REGISTER | \ + ((uint32_t)(n) & 0xff) << 16 | \ + ((uint32_t)(first) & 0xffff)) +#define DMA_SET_REGISTERS( first, n ) \ + DMA_WRITE(BCI_CMD_SET_REGISTER | \ + ((uint32_t)(n) & 0xff) << 16 | \ + ((uint32_t)(first) & 0xffff)) + +#define BCI_DRAW_PRIMITIVE(n, type, skip) \ + BCI_WRITE(BCI_CMD_DRAW_PRIM | (type) | (skip) | \ + ((n) << 16)) +#define DMA_DRAW_PRIMITIVE(n, type, skip) \ + DMA_WRITE(BCI_CMD_DRAW_PRIM | (type) | (skip) | \ + ((n) << 16)) + +#define BCI_DRAW_INDICES_S3D(n, type, i0) \ + BCI_WRITE(BCI_CMD_DRAW_INDEXED_PRIM | (type) | \ + ((n) << 16) | (i0)) + +#define BCI_DRAW_INDICES_S4(n, type, skip) \ + BCI_WRITE(BCI_CMD_DRAW_INDEXED_PRIM | (type) | \ + (skip) | ((n) << 16)) + +#define BCI_DMA(n) \ + BCI_WRITE(BCI_CMD_DMA | (((n) >> 1) - 1)) + +/* + * access to MMIO + */ +#define SAVAGE_READ(reg) DRM_READ32( dev_priv->mmio, (reg) ) +#define SAVAGE_WRITE(reg) DRM_WRITE32( dev_priv->mmio, (reg) ) + +/* + * access to the burst command interface (BCI) + */ +#define SAVAGE_BCI_DEBUG 1 + +#define BCI_LOCALS volatile uint32_t *bci_ptr; + +#define BEGIN_BCI( n ) do { \ + dev_priv->wait_fifo(dev_priv, (n)); \ + bci_ptr = dev_priv->bci_ptr; \ +} while(0) + +#define BCI_WRITE( val ) *bci_ptr++ = (uint32_t)(val) + +/* + * command DMA support + */ +#define SAVAGE_DMA_DEBUG 1 + +#define DMA_LOCALS uint32_t *dma_ptr; + +#define BEGIN_DMA( n ) do { \ + unsigned int cur = dev_priv->current_dma_page; \ + unsigned int rest = SAVAGE_DMA_PAGE_SIZE - \ + dev_priv->dma_pages[cur].used; \ + if ((n) > rest) { \ + dma_ptr = savage_dma_alloc(dev_priv, (n)); \ + } else { /* fast path for small allocations */ \ + dma_ptr = (uint32_t *)dev_priv->cmd_dma->handle + \ + cur * SAVAGE_DMA_PAGE_SIZE + \ + dev_priv->dma_pages[cur].used; \ + if (dev_priv->dma_pages[cur].used == 0) \ + savage_dma_wait(dev_priv, cur); \ + dev_priv->dma_pages[cur].used += (n); \ + } \ +} while(0) + +#define DMA_WRITE( val ) *dma_ptr++ = (uint32_t)(val) + +#define DMA_COPY(src, n) do { \ + memcpy(dma_ptr, (src), (n)*4); \ + dma_ptr += n; \ +} while(0) + +#if SAVAGE_DMA_DEBUG +#define DMA_COMMIT() do { \ + unsigned int cur = dev_priv->current_dma_page; \ + uint32_t *expected = (uint32_t *)dev_priv->cmd_dma->handle + \ + cur * SAVAGE_DMA_PAGE_SIZE + \ + dev_priv->dma_pages[cur].used; \ + if (dma_ptr != expected) { \ + DRM_ERROR("DMA allocation and use don't match: " \ + "%p != %p\n", expected, dma_ptr); \ + savage_dma_reset(dev_priv); \ + } \ +} while(0) +#else +#define DMA_COMMIT() do {/* nothing */} while(0) +#endif + +#define DMA_FLUSH() dev_priv->dma_flush(dev_priv) + +/* Buffer aging via event tag + */ + +#define UPDATE_EVENT_COUNTER( ) do { \ + if (dev_priv->status_ptr) { \ + uint16_t count; \ + /* coordinate with Xserver */ \ + count = dev_priv->status_ptr[1023]; \ + if (count < dev_priv->event_counter) \ + dev_priv->event_wrap++; \ + dev_priv->event_counter = count; \ + } \ +} while(0) + +#define SET_AGE( age, e, w ) do { \ + (age)->event = e; \ + (age)->wrap = w; \ +} while(0) + +#define TEST_AGE( age, e, w ) \ + ( (age)->wrap < (w) || ( (age)->wrap == (w) && (age)->event <= (e) ) ) + +#endif /* __SAVAGE_DRV_H__ */ diff --git a/drivers/gpu/drm/savage/savage_state.c b/drivers/gpu/drm/savage/savage_state.c new file mode 100644 index 00000000..8a3e3159 --- /dev/null +++ b/drivers/gpu/drm/savage/savage_state.c @@ -0,0 +1,1162 @@ +/* savage_state.c -- State and drawing support for Savage + * + * Copyright 2004 Felix Kuehling + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sub license, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NON-INFRINGEMENT. IN NO EVENT SHALL FELIX KUEHLING BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF + * CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION + * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + */ +#include "drmP.h" +#include "savage_drm.h" +#include "savage_drv.h" + +void savage_emit_clip_rect_s3d(drm_savage_private_t * dev_priv, + const struct drm_clip_rect * pbox) +{ + uint32_t scstart = dev_priv->state.s3d.new_scstart; + uint32_t scend = dev_priv->state.s3d.new_scend; + scstart = (scstart & ~SAVAGE_SCISSOR_MASK_S3D) | + ((uint32_t) pbox->x1 & 0x000007ff) | + (((uint32_t) pbox->y1 << 16) & 0x07ff0000); + scend = (scend & ~SAVAGE_SCISSOR_MASK_S3D) | + (((uint32_t) pbox->x2 - 1) & 0x000007ff) | + ((((uint32_t) pbox->y2 - 1) << 16) & 0x07ff0000); + if (scstart != dev_priv->state.s3d.scstart || + scend != dev_priv->state.s3d.scend) { + DMA_LOCALS; + BEGIN_DMA(4); + DMA_WRITE(BCI_CMD_WAIT | BCI_CMD_WAIT_3D); + DMA_SET_REGISTERS(SAVAGE_SCSTART_S3D, 2); + DMA_WRITE(scstart); + DMA_WRITE(scend); + dev_priv->state.s3d.scstart = scstart; + dev_priv->state.s3d.scend = scend; + dev_priv->waiting = 1; + DMA_COMMIT(); + } +} + +void savage_emit_clip_rect_s4(drm_savage_private_t * dev_priv, + const struct drm_clip_rect * pbox) +{ + uint32_t drawctrl0 = dev_priv->state.s4.new_drawctrl0; + uint32_t drawctrl1 = dev_priv->state.s4.new_drawctrl1; + drawctrl0 = (drawctrl0 & ~SAVAGE_SCISSOR_MASK_S4) | + ((uint32_t) pbox->x1 & 0x000007ff) | + (((uint32_t) pbox->y1 << 12) & 0x00fff000); + drawctrl1 = (drawctrl1 & ~SAVAGE_SCISSOR_MASK_S4) | + (((uint32_t) pbox->x2 - 1) & 0x000007ff) | + ((((uint32_t) pbox->y2 - 1) << 12) & 0x00fff000); + if (drawctrl0 != dev_priv->state.s4.drawctrl0 || + drawctrl1 != dev_priv->state.s4.drawctrl1) { + DMA_LOCALS; + BEGIN_DMA(4); + DMA_WRITE(BCI_CMD_WAIT | BCI_CMD_WAIT_3D); + DMA_SET_REGISTERS(SAVAGE_DRAWCTRL0_S4, 2); + DMA_WRITE(drawctrl0); + DMA_WRITE(drawctrl1); + dev_priv->state.s4.drawctrl0 = drawctrl0; + dev_priv->state.s4.drawctrl1 = drawctrl1; + dev_priv->waiting = 1; + DMA_COMMIT(); + } +} + +static int savage_verify_texaddr(drm_savage_private_t * dev_priv, int unit, + uint32_t addr) +{ + if ((addr & 6) != 2) { /* reserved bits */ + DRM_ERROR("bad texAddr%d %08x (reserved bits)\n", unit, addr); + return -EINVAL; + } + if (!(addr & 1)) { /* local */ + addr &= ~7; + if (addr < dev_priv->texture_offset || + addr >= dev_priv->texture_offset + dev_priv->texture_size) { + DRM_ERROR + ("bad texAddr%d %08x (local addr out of range)\n", + unit, addr); + return -EINVAL; + } + } else { /* AGP */ + if (!dev_priv->agp_textures) { + DRM_ERROR("bad texAddr%d %08x (AGP not available)\n", + unit, addr); + return -EINVAL; + } + addr &= ~7; + if (addr < dev_priv->agp_textures->offset || + addr >= (dev_priv->agp_textures->offset + + dev_priv->agp_textures->size)) { + DRM_ERROR + ("bad texAddr%d %08x (AGP addr out of range)\n", + unit, addr); + return -EINVAL; + } + } + return 0; +} + +#define SAVE_STATE(reg,where) \ + if(start <= reg && start+count > reg) \ + dev_priv->state.where = regs[reg - start] +#define SAVE_STATE_MASK(reg,where,mask) do { \ + if(start <= reg && start+count > reg) { \ + uint32_t tmp; \ + tmp = regs[reg - start]; \ + dev_priv->state.where = (tmp & (mask)) | \ + (dev_priv->state.where & ~(mask)); \ + } \ +} while (0) + +static int savage_verify_state_s3d(drm_savage_private_t * dev_priv, + unsigned int start, unsigned int count, + const uint32_t *regs) +{ + if (start < SAVAGE_TEXPALADDR_S3D || + start + count - 1 > SAVAGE_DESTTEXRWWATERMARK_S3D) { + DRM_ERROR("invalid register range (0x%04x-0x%04x)\n", + start, start + count - 1); + return -EINVAL; + } + + SAVE_STATE_MASK(SAVAGE_SCSTART_S3D, s3d.new_scstart, + ~SAVAGE_SCISSOR_MASK_S3D); + SAVE_STATE_MASK(SAVAGE_SCEND_S3D, s3d.new_scend, + ~SAVAGE_SCISSOR_MASK_S3D); + + /* if any texture regs were changed ... */ + if (start <= SAVAGE_TEXCTRL_S3D && + start + count > SAVAGE_TEXPALADDR_S3D) { + /* ... check texture state */ + SAVE_STATE(SAVAGE_TEXCTRL_S3D, s3d.texctrl); + SAVE_STATE(SAVAGE_TEXADDR_S3D, s3d.texaddr); + if (dev_priv->state.s3d.texctrl & SAVAGE_TEXCTRL_TEXEN_MASK) + return savage_verify_texaddr(dev_priv, 0, + dev_priv->state.s3d.texaddr); + } + + return 0; +} + +static int savage_verify_state_s4(drm_savage_private_t * dev_priv, + unsigned int start, unsigned int count, + const uint32_t *regs) +{ + int ret = 0; + + if (start < SAVAGE_DRAWLOCALCTRL_S4 || + start + count - 1 > SAVAGE_TEXBLENDCOLOR_S4) { + DRM_ERROR("invalid register range (0x%04x-0x%04x)\n", + start, start + count - 1); + return -EINVAL; + } + + SAVE_STATE_MASK(SAVAGE_DRAWCTRL0_S4, s4.new_drawctrl0, + ~SAVAGE_SCISSOR_MASK_S4); + SAVE_STATE_MASK(SAVAGE_DRAWCTRL1_S4, s4.new_drawctrl1, + ~SAVAGE_SCISSOR_MASK_S4); + + /* if any texture regs were changed ... */ + if (start <= SAVAGE_TEXDESCR_S4 && + start + count > SAVAGE_TEXPALADDR_S4) { + /* ... check texture state */ + SAVE_STATE(SAVAGE_TEXDESCR_S4, s4.texdescr); + SAVE_STATE(SAVAGE_TEXADDR0_S4, s4.texaddr0); + SAVE_STATE(SAVAGE_TEXADDR1_S4, s4.texaddr1); + if (dev_priv->state.s4.texdescr & SAVAGE_TEXDESCR_TEX0EN_MASK) + ret |= savage_verify_texaddr(dev_priv, 0, + dev_priv->state.s4.texaddr0); + if (dev_priv->state.s4.texdescr & SAVAGE_TEXDESCR_TEX1EN_MASK) + ret |= savage_verify_texaddr(dev_priv, 1, + dev_priv->state.s4.texaddr1); + } + + return ret; +} + +#undef SAVE_STATE +#undef SAVE_STATE_MASK + +static int savage_dispatch_state(drm_savage_private_t * dev_priv, + const drm_savage_cmd_header_t * cmd_header, + const uint32_t *regs) +{ + unsigned int count = cmd_header->state.count; + unsigned int start = cmd_header->state.start; + unsigned int count2 = 0; + unsigned int bci_size; + int ret; + DMA_LOCALS; + + if (!count) + return 0; + + if (S3_SAVAGE3D_SERIES(dev_priv->chipset)) { + ret = savage_verify_state_s3d(dev_priv, start, count, regs); + if (ret != 0) + return ret; + /* scissor regs are emitted in savage_dispatch_draw */ + if (start < SAVAGE_SCSTART_S3D) { + if (start + count > SAVAGE_SCEND_S3D + 1) + count2 = count - (SAVAGE_SCEND_S3D + 1 - start); + if (start + count > SAVAGE_SCSTART_S3D) + count = SAVAGE_SCSTART_S3D - start; + } else if (start <= SAVAGE_SCEND_S3D) { + if (start + count > SAVAGE_SCEND_S3D + 1) { + count -= SAVAGE_SCEND_S3D + 1 - start; + start = SAVAGE_SCEND_S3D + 1; + } else + return 0; + } + } else { + ret = savage_verify_state_s4(dev_priv, start, count, regs); + if (ret != 0) + return ret; + /* scissor regs are emitted in savage_dispatch_draw */ + if (start < SAVAGE_DRAWCTRL0_S4) { + if (start + count > SAVAGE_DRAWCTRL1_S4 + 1) + count2 = count - + (SAVAGE_DRAWCTRL1_S4 + 1 - start); + if (start + count > SAVAGE_DRAWCTRL0_S4) + count = SAVAGE_DRAWCTRL0_S4 - start; + } else if (start <= SAVAGE_DRAWCTRL1_S4) { + if (start + count > SAVAGE_DRAWCTRL1_S4 + 1) { + count -= SAVAGE_DRAWCTRL1_S4 + 1 - start; + start = SAVAGE_DRAWCTRL1_S4 + 1; + } else + return 0; + } + } + + bci_size = count + (count + 254) / 255 + count2 + (count2 + 254) / 255; + + if (cmd_header->state.global) { + BEGIN_DMA(bci_size + 1); + DMA_WRITE(BCI_CMD_WAIT | BCI_CMD_WAIT_3D); + dev_priv->waiting = 1; + } else { + BEGIN_DMA(bci_size); + } + + do { + while (count > 0) { + unsigned int n = count < 255 ? count : 255; + DMA_SET_REGISTERS(start, n); + DMA_COPY(regs, n); + count -= n; + start += n; + regs += n; + } + start += 2; + regs += 2; + count = count2; + count2 = 0; + } while (count); + + DMA_COMMIT(); + + return 0; +} + +static int savage_dispatch_dma_prim(drm_savage_private_t * dev_priv, + const drm_savage_cmd_header_t * cmd_header, + const struct drm_buf * dmabuf) +{ + unsigned char reorder = 0; + unsigned int prim = cmd_header->prim.prim; + unsigned int skip = cmd_header->prim.skip; + unsigned int n = cmd_header->prim.count; + unsigned int start = cmd_header->prim.start; + unsigned int i; + BCI_LOCALS; + + if (!dmabuf) { + DRM_ERROR("called without dma buffers!\n"); + return -EINVAL; + } + + if (!n) + return 0; + + switch (prim) { + case SAVAGE_PRIM_TRILIST_201: + reorder = 1; + prim = SAVAGE_PRIM_TRILIST; + case SAVAGE_PRIM_TRILIST: + if (n % 3 != 0) { + DRM_ERROR("wrong number of vertices %u in TRILIST\n", + n); + return -EINVAL; + } + break; + case SAVAGE_PRIM_TRISTRIP: + case SAVAGE_PRIM_TRIFAN: + if (n < 3) { + DRM_ERROR + ("wrong number of vertices %u in TRIFAN/STRIP\n", + n); + return -EINVAL; + } + break; + default: + DRM_ERROR("invalid primitive type %u\n", prim); + return -EINVAL; + } + + if (S3_SAVAGE3D_SERIES(dev_priv->chipset)) { + if (skip != 0) { + DRM_ERROR("invalid skip flags 0x%04x for DMA\n", skip); + return -EINVAL; + } + } else { + unsigned int size = 10 - (skip & 1) - (skip >> 1 & 1) - + (skip >> 2 & 1) - (skip >> 3 & 1) - (skip >> 4 & 1) - + (skip >> 5 & 1) - (skip >> 6 & 1) - (skip >> 7 & 1); + if (skip > SAVAGE_SKIP_ALL_S4 || size != 8) { + DRM_ERROR("invalid skip flags 0x%04x for DMA\n", skip); + return -EINVAL; + } + if (reorder) { + DRM_ERROR("TRILIST_201 used on Savage4 hardware\n"); + return -EINVAL; + } + } + + if (start + n > dmabuf->total / 32) { + DRM_ERROR("vertex indices (%u-%u) out of range (0-%u)\n", + start, start + n - 1, dmabuf->total / 32); + return -EINVAL; + } + + /* Vertex DMA doesn't work with command DMA at the same time, + * so we use BCI_... to submit commands here. Flush buffered + * faked DMA first. */ + DMA_FLUSH(); + + if (dmabuf->bus_address != dev_priv->state.common.vbaddr) { + BEGIN_BCI(2); + BCI_SET_REGISTERS(SAVAGE_VERTBUFADDR, 1); + BCI_WRITE(dmabuf->bus_address | dev_priv->dma_type); + dev_priv->state.common.vbaddr = dmabuf->bus_address; + } + if (S3_SAVAGE3D_SERIES(dev_priv->chipset) && dev_priv->waiting) { + /* Workaround for what looks like a hardware bug. If a + * WAIT_3D_IDLE was emitted some time before the + * indexed drawing command then the engine will lock + * up. There are two known workarounds: + * WAIT_IDLE_EMPTY or emit at least 63 NOPs. */ + BEGIN_BCI(63); + for (i = 0; i < 63; ++i) + BCI_WRITE(BCI_CMD_WAIT); + dev_priv->waiting = 0; + } + + prim <<= 25; + while (n != 0) { + /* Can emit up to 255 indices (85 triangles) at once. */ + unsigned int count = n > 255 ? 255 : n; + if (reorder) { + /* Need to reorder indices for correct flat + * shading while preserving the clock sense + * for correct culling. Only on Savage3D. */ + int reorder[3] = { -1, -1, -1 }; + reorder[start % 3] = 2; + + BEGIN_BCI((count + 1 + 1) / 2); + BCI_DRAW_INDICES_S3D(count, prim, start + 2); + + for (i = start + 1; i + 1 < start + count; i += 2) + BCI_WRITE((i + reorder[i % 3]) | + ((i + 1 + + reorder[(i + 1) % 3]) << 16)); + if (i < start + count) + BCI_WRITE(i + reorder[i % 3]); + } else if (S3_SAVAGE3D_SERIES(dev_priv->chipset)) { + BEGIN_BCI((count + 1 + 1) / 2); + BCI_DRAW_INDICES_S3D(count, prim, start); + + for (i = start + 1; i + 1 < start + count; i += 2) + BCI_WRITE(i | ((i + 1) << 16)); + if (i < start + count) + BCI_WRITE(i); + } else { + BEGIN_BCI((count + 2 + 1) / 2); + BCI_DRAW_INDICES_S4(count, prim, skip); + + for (i = start; i + 1 < start + count; i += 2) + BCI_WRITE(i | ((i + 1) << 16)); + if (i < start + count) + BCI_WRITE(i); + } + + start += count; + n -= count; + + prim |= BCI_CMD_DRAW_CONT; + } + + return 0; +} + +static int savage_dispatch_vb_prim(drm_savage_private_t * dev_priv, + const drm_savage_cmd_header_t * cmd_header, + const uint32_t *vtxbuf, unsigned int vb_size, + unsigned int vb_stride) +{ + unsigned char reorder = 0; + unsigned int prim = cmd_header->prim.prim; + unsigned int skip = cmd_header->prim.skip; + unsigned int n = cmd_header->prim.count; + unsigned int start = cmd_header->prim.start; + unsigned int vtx_size; + unsigned int i; + DMA_LOCALS; + + if (!n) + return 0; + + switch (prim) { + case SAVAGE_PRIM_TRILIST_201: + reorder = 1; + prim = SAVAGE_PRIM_TRILIST; + case SAVAGE_PRIM_TRILIST: + if (n % 3 != 0) { + DRM_ERROR("wrong number of vertices %u in TRILIST\n", + n); + return -EINVAL; + } + break; + case SAVAGE_PRIM_TRISTRIP: + case SAVAGE_PRIM_TRIFAN: + if (n < 3) { + DRM_ERROR + ("wrong number of vertices %u in TRIFAN/STRIP\n", + n); + return -EINVAL; + } + break; + default: + DRM_ERROR("invalid primitive type %u\n", prim); + return -EINVAL; + } + + if (S3_SAVAGE3D_SERIES(dev_priv->chipset)) { + if (skip > SAVAGE_SKIP_ALL_S3D) { + DRM_ERROR("invalid skip flags 0x%04x\n", skip); + return -EINVAL; + } + vtx_size = 8; /* full vertex */ + } else { + if (skip > SAVAGE_SKIP_ALL_S4) { + DRM_ERROR("invalid skip flags 0x%04x\n", skip); + return -EINVAL; + } + vtx_size = 10; /* full vertex */ + } + + vtx_size -= (skip & 1) + (skip >> 1 & 1) + + (skip >> 2 & 1) + (skip >> 3 & 1) + (skip >> 4 & 1) + + (skip >> 5 & 1) + (skip >> 6 & 1) + (skip >> 7 & 1); + + if (vtx_size > vb_stride) { + DRM_ERROR("vertex size greater than vb stride (%u > %u)\n", + vtx_size, vb_stride); + return -EINVAL; + } + + if (start + n > vb_size / (vb_stride * 4)) { + DRM_ERROR("vertex indices (%u-%u) out of range (0-%u)\n", + start, start + n - 1, vb_size / (vb_stride * 4)); + return -EINVAL; + } + + prim <<= 25; + while (n != 0) { + /* Can emit up to 255 vertices (85 triangles) at once. */ + unsigned int count = n > 255 ? 255 : n; + if (reorder) { + /* Need to reorder vertices for correct flat + * shading while preserving the clock sense + * for correct culling. Only on Savage3D. */ + int reorder[3] = { -1, -1, -1 }; + reorder[start % 3] = 2; + + BEGIN_DMA(count * vtx_size + 1); + DMA_DRAW_PRIMITIVE(count, prim, skip); + + for (i = start; i < start + count; ++i) { + unsigned int j = i + reorder[i % 3]; + DMA_COPY(&vtxbuf[vb_stride * j], vtx_size); + } + + DMA_COMMIT(); + } else { + BEGIN_DMA(count * vtx_size + 1); + DMA_DRAW_PRIMITIVE(count, prim, skip); + + if (vb_stride == vtx_size) { + DMA_COPY(&vtxbuf[vb_stride * start], + vtx_size * count); + } else { + for (i = start; i < start + count; ++i) { + DMA_COPY(&vtxbuf [vb_stride * i], + vtx_size); + } + } + + DMA_COMMIT(); + } + + start += count; + n -= count; + + prim |= BCI_CMD_DRAW_CONT; + } + + return 0; +} + +static int savage_dispatch_dma_idx(drm_savage_private_t * dev_priv, + const drm_savage_cmd_header_t * cmd_header, + const uint16_t *idx, + const struct drm_buf * dmabuf) +{ + unsigned char reorder = 0; + unsigned int prim = cmd_header->idx.prim; + unsigned int skip = cmd_header->idx.skip; + unsigned int n = cmd_header->idx.count; + unsigned int i; + BCI_LOCALS; + + if (!dmabuf) { + DRM_ERROR("called without dma buffers!\n"); + return -EINVAL; + } + + if (!n) + return 0; + + switch (prim) { + case SAVAGE_PRIM_TRILIST_201: + reorder = 1; + prim = SAVAGE_PRIM_TRILIST; + case SAVAGE_PRIM_TRILIST: + if (n % 3 != 0) { + DRM_ERROR("wrong number of indices %u in TRILIST\n", n); + return -EINVAL; + } + break; + case SAVAGE_PRIM_TRISTRIP: + case SAVAGE_PRIM_TRIFAN: + if (n < 3) { + DRM_ERROR + ("wrong number of indices %u in TRIFAN/STRIP\n", n); + return -EINVAL; + } + break; + default: + DRM_ERROR("invalid primitive type %u\n", prim); + return -EINVAL; + } + + if (S3_SAVAGE3D_SERIES(dev_priv->chipset)) { + if (skip != 0) { + DRM_ERROR("invalid skip flags 0x%04x for DMA\n", skip); + return -EINVAL; + } + } else { + unsigned int size = 10 - (skip & 1) - (skip >> 1 & 1) - + (skip >> 2 & 1) - (skip >> 3 & 1) - (skip >> 4 & 1) - + (skip >> 5 & 1) - (skip >> 6 & 1) - (skip >> 7 & 1); + if (skip > SAVAGE_SKIP_ALL_S4 || size != 8) { + DRM_ERROR("invalid skip flags 0x%04x for DMA\n", skip); + return -EINVAL; + } + if (reorder) { + DRM_ERROR("TRILIST_201 used on Savage4 hardware\n"); + return -EINVAL; + } + } + + /* Vertex DMA doesn't work with command DMA at the same time, + * so we use BCI_... to submit commands here. Flush buffered + * faked DMA first. */ + DMA_FLUSH(); + + if (dmabuf->bus_address != dev_priv->state.common.vbaddr) { + BEGIN_BCI(2); + BCI_SET_REGISTERS(SAVAGE_VERTBUFADDR, 1); + BCI_WRITE(dmabuf->bus_address | dev_priv->dma_type); + dev_priv->state.common.vbaddr = dmabuf->bus_address; + } + if (S3_SAVAGE3D_SERIES(dev_priv->chipset) && dev_priv->waiting) { + /* Workaround for what looks like a hardware bug. If a + * WAIT_3D_IDLE was emitted some time before the + * indexed drawing command then the engine will lock + * up. There are two known workarounds: + * WAIT_IDLE_EMPTY or emit at least 63 NOPs. */ + BEGIN_BCI(63); + for (i = 0; i < 63; ++i) + BCI_WRITE(BCI_CMD_WAIT); + dev_priv->waiting = 0; + } + + prim <<= 25; + while (n != 0) { + /* Can emit up to 255 indices (85 triangles) at once. */ + unsigned int count = n > 255 ? 255 : n; + + /* check indices */ + for (i = 0; i < count; ++i) { + if (idx[i] > dmabuf->total / 32) { + DRM_ERROR("idx[%u]=%u out of range (0-%u)\n", + i, idx[i], dmabuf->total / 32); + return -EINVAL; + } + } + + if (reorder) { + /* Need to reorder indices for correct flat + * shading while preserving the clock sense + * for correct culling. Only on Savage3D. */ + int reorder[3] = { 2, -1, -1 }; + + BEGIN_BCI((count + 1 + 1) / 2); + BCI_DRAW_INDICES_S3D(count, prim, idx[2]); + + for (i = 1; i + 1 < count; i += 2) + BCI_WRITE(idx[i + reorder[i % 3]] | + (idx[i + 1 + + reorder[(i + 1) % 3]] << 16)); + if (i < count) + BCI_WRITE(idx[i + reorder[i % 3]]); + } else if (S3_SAVAGE3D_SERIES(dev_priv->chipset)) { + BEGIN_BCI((count + 1 + 1) / 2); + BCI_DRAW_INDICES_S3D(count, prim, idx[0]); + + for (i = 1; i + 1 < count; i += 2) + BCI_WRITE(idx[i] | (idx[i + 1] << 16)); + if (i < count) + BCI_WRITE(idx[i]); + } else { + BEGIN_BCI((count + 2 + 1) / 2); + BCI_DRAW_INDICES_S4(count, prim, skip); + + for (i = 0; i + 1 < count; i += 2) + BCI_WRITE(idx[i] | (idx[i + 1] << 16)); + if (i < count) + BCI_WRITE(idx[i]); + } + + idx += count; + n -= count; + + prim |= BCI_CMD_DRAW_CONT; + } + + return 0; +} + +static int savage_dispatch_vb_idx(drm_savage_private_t * dev_priv, + const drm_savage_cmd_header_t * cmd_header, + const uint16_t *idx, + const uint32_t *vtxbuf, + unsigned int vb_size, unsigned int vb_stride) +{ + unsigned char reorder = 0; + unsigned int prim = cmd_header->idx.prim; + unsigned int skip = cmd_header->idx.skip; + unsigned int n = cmd_header->idx.count; + unsigned int vtx_size; + unsigned int i; + DMA_LOCALS; + + if (!n) + return 0; + + switch (prim) { + case SAVAGE_PRIM_TRILIST_201: + reorder = 1; + prim = SAVAGE_PRIM_TRILIST; + case SAVAGE_PRIM_TRILIST: + if (n % 3 != 0) { + DRM_ERROR("wrong number of indices %u in TRILIST\n", n); + return -EINVAL; + } + break; + case SAVAGE_PRIM_TRISTRIP: + case SAVAGE_PRIM_TRIFAN: + if (n < 3) { + DRM_ERROR + ("wrong number of indices %u in TRIFAN/STRIP\n", n); + return -EINVAL; + } + break; + default: + DRM_ERROR("invalid primitive type %u\n", prim); + return -EINVAL; + } + + if (S3_SAVAGE3D_SERIES(dev_priv->chipset)) { + if (skip > SAVAGE_SKIP_ALL_S3D) { + DRM_ERROR("invalid skip flags 0x%04x\n", skip); + return -EINVAL; + } + vtx_size = 8; /* full vertex */ + } else { + if (skip > SAVAGE_SKIP_ALL_S4) { + DRM_ERROR("invalid skip flags 0x%04x\n", skip); + return -EINVAL; + } + vtx_size = 10; /* full vertex */ + } + + vtx_size -= (skip & 1) + (skip >> 1 & 1) + + (skip >> 2 & 1) + (skip >> 3 & 1) + (skip >> 4 & 1) + + (skip >> 5 & 1) + (skip >> 6 & 1) + (skip >> 7 & 1); + + if (vtx_size > vb_stride) { + DRM_ERROR("vertex size greater than vb stride (%u > %u)\n", + vtx_size, vb_stride); + return -EINVAL; + } + + prim <<= 25; + while (n != 0) { + /* Can emit up to 255 vertices (85 triangles) at once. */ + unsigned int count = n > 255 ? 255 : n; + + /* Check indices */ + for (i = 0; i < count; ++i) { + if (idx[i] > vb_size / (vb_stride * 4)) { + DRM_ERROR("idx[%u]=%u out of range (0-%u)\n", + i, idx[i], vb_size / (vb_stride * 4)); + return -EINVAL; + } + } + + if (reorder) { + /* Need to reorder vertices for correct flat + * shading while preserving the clock sense + * for correct culling. Only on Savage3D. */ + int reorder[3] = { 2, -1, -1 }; + + BEGIN_DMA(count * vtx_size + 1); + DMA_DRAW_PRIMITIVE(count, prim, skip); + + for (i = 0; i < count; ++i) { + unsigned int j = idx[i + reorder[i % 3]]; + DMA_COPY(&vtxbuf[vb_stride * j], vtx_size); + } + + DMA_COMMIT(); + } else { + BEGIN_DMA(count * vtx_size + 1); + DMA_DRAW_PRIMITIVE(count, prim, skip); + + for (i = 0; i < count; ++i) { + unsigned int j = idx[i]; + DMA_COPY(&vtxbuf[vb_stride * j], vtx_size); + } + + DMA_COMMIT(); + } + + idx += count; + n -= count; + + prim |= BCI_CMD_DRAW_CONT; + } + + return 0; +} + +static int savage_dispatch_clear(drm_savage_private_t * dev_priv, + const drm_savage_cmd_header_t * cmd_header, + const drm_savage_cmd_header_t *data, + unsigned int nbox, + const struct drm_clip_rect *boxes) +{ + unsigned int flags = cmd_header->clear0.flags; + unsigned int clear_cmd; + unsigned int i, nbufs; + DMA_LOCALS; + + if (nbox == 0) + return 0; + + clear_cmd = BCI_CMD_RECT | BCI_CMD_RECT_XP | BCI_CMD_RECT_YP | + BCI_CMD_SEND_COLOR | BCI_CMD_DEST_PBD_NEW; + BCI_CMD_SET_ROP(clear_cmd, 0xCC); + + nbufs = ((flags & SAVAGE_FRONT) ? 1 : 0) + + ((flags & SAVAGE_BACK) ? 1 : 0) + ((flags & SAVAGE_DEPTH) ? 1 : 0); + if (nbufs == 0) + return 0; + + if (data->clear1.mask != 0xffffffff) { + /* set mask */ + BEGIN_DMA(2); + DMA_SET_REGISTERS(SAVAGE_BITPLANEWTMASK, 1); + DMA_WRITE(data->clear1.mask); + DMA_COMMIT(); + } + for (i = 0; i < nbox; ++i) { + unsigned int x, y, w, h; + unsigned int buf; + x = boxes[i].x1, y = boxes[i].y1; + w = boxes[i].x2 - boxes[i].x1; + h = boxes[i].y2 - boxes[i].y1; + BEGIN_DMA(nbufs * 6); + for (buf = SAVAGE_FRONT; buf <= SAVAGE_DEPTH; buf <<= 1) { + if (!(flags & buf)) + continue; + DMA_WRITE(clear_cmd); + switch (buf) { + case SAVAGE_FRONT: + DMA_WRITE(dev_priv->front_offset); + DMA_WRITE(dev_priv->front_bd); + break; + case SAVAGE_BACK: + DMA_WRITE(dev_priv->back_offset); + DMA_WRITE(dev_priv->back_bd); + break; + case SAVAGE_DEPTH: + DMA_WRITE(dev_priv->depth_offset); + DMA_WRITE(dev_priv->depth_bd); + break; + } + DMA_WRITE(data->clear1.value); + DMA_WRITE(BCI_X_Y(x, y)); + DMA_WRITE(BCI_W_H(w, h)); + } + DMA_COMMIT(); + } + if (data->clear1.mask != 0xffffffff) { + /* reset mask */ + BEGIN_DMA(2); + DMA_SET_REGISTERS(SAVAGE_BITPLANEWTMASK, 1); + DMA_WRITE(0xffffffff); + DMA_COMMIT(); + } + + return 0; +} + +static int savage_dispatch_swap(drm_savage_private_t * dev_priv, + unsigned int nbox, const struct drm_clip_rect *boxes) +{ + unsigned int swap_cmd; + unsigned int i; + DMA_LOCALS; + + if (nbox == 0) + return 0; + + swap_cmd = BCI_CMD_RECT | BCI_CMD_RECT_XP | BCI_CMD_RECT_YP | + BCI_CMD_SRC_PBD_COLOR_NEW | BCI_CMD_DEST_GBD; + BCI_CMD_SET_ROP(swap_cmd, 0xCC); + + for (i = 0; i < nbox; ++i) { + BEGIN_DMA(6); + DMA_WRITE(swap_cmd); + DMA_WRITE(dev_priv->back_offset); + DMA_WRITE(dev_priv->back_bd); + DMA_WRITE(BCI_X_Y(boxes[i].x1, boxes[i].y1)); + DMA_WRITE(BCI_X_Y(boxes[i].x1, boxes[i].y1)); + DMA_WRITE(BCI_W_H(boxes[i].x2 - boxes[i].x1, + boxes[i].y2 - boxes[i].y1)); + DMA_COMMIT(); + } + + return 0; +} + +static int savage_dispatch_draw(drm_savage_private_t * dev_priv, + const drm_savage_cmd_header_t *start, + const drm_savage_cmd_header_t *end, + const struct drm_buf * dmabuf, + const unsigned int *vtxbuf, + unsigned int vb_size, unsigned int vb_stride, + unsigned int nbox, + const struct drm_clip_rect *boxes) +{ + unsigned int i, j; + int ret; + + for (i = 0; i < nbox; ++i) { + const drm_savage_cmd_header_t *cmdbuf; + dev_priv->emit_clip_rect(dev_priv, &boxes[i]); + + cmdbuf = start; + while (cmdbuf < end) { + drm_savage_cmd_header_t cmd_header; + cmd_header = *cmdbuf; + cmdbuf++; + switch (cmd_header.cmd.cmd) { + case SAVAGE_CMD_DMA_PRIM: + ret = savage_dispatch_dma_prim( + dev_priv, &cmd_header, dmabuf); + break; + case SAVAGE_CMD_VB_PRIM: + ret = savage_dispatch_vb_prim( + dev_priv, &cmd_header, + vtxbuf, vb_size, vb_stride); + break; + case SAVAGE_CMD_DMA_IDX: + j = (cmd_header.idx.count + 3) / 4; + /* j was check in savage_bci_cmdbuf */ + ret = savage_dispatch_dma_idx(dev_priv, + &cmd_header, (const uint16_t *)cmdbuf, + dmabuf); + cmdbuf += j; + break; + case SAVAGE_CMD_VB_IDX: + j = (cmd_header.idx.count + 3) / 4; + /* j was check in savage_bci_cmdbuf */ + ret = savage_dispatch_vb_idx(dev_priv, + &cmd_header, (const uint16_t *)cmdbuf, + (const uint32_t *)vtxbuf, vb_size, + vb_stride); + cmdbuf += j; + break; + default: + /* What's the best return code? EFAULT? */ + DRM_ERROR("IMPLEMENTATION ERROR: " + "non-drawing-command %d\n", + cmd_header.cmd.cmd); + return -EINVAL; + } + + if (ret != 0) + return ret; + } + } + + return 0; +} + +int savage_bci_cmdbuf(struct drm_device *dev, void *data, struct drm_file *file_priv) +{ + drm_savage_private_t *dev_priv = dev->dev_private; + struct drm_device_dma *dma = dev->dma; + struct drm_buf *dmabuf; + drm_savage_cmdbuf_t *cmdbuf = data; + drm_savage_cmd_header_t *kcmd_addr = NULL; + drm_savage_cmd_header_t *first_draw_cmd; + unsigned int *kvb_addr = NULL; + struct drm_clip_rect *kbox_addr = NULL; + unsigned int i, j; + int ret = 0; + + DRM_DEBUG("\n"); + + LOCK_TEST_WITH_RETURN(dev, file_priv); + + if (dma && dma->buflist) { + if (cmdbuf->dma_idx > dma->buf_count) { + DRM_ERROR + ("vertex buffer index %u out of range (0-%u)\n", + cmdbuf->dma_idx, dma->buf_count - 1); + return -EINVAL; + } + dmabuf = dma->buflist[cmdbuf->dma_idx]; + } else { + dmabuf = NULL; + } + + /* Copy the user buffers into kernel temporary areas. This hasn't been + * a performance loss compared to VERIFYAREA_READ/ + * COPY_FROM_USER_UNCHECKED when done in other drivers, and is correct + * for locking on FreeBSD. + */ + if (cmdbuf->size) { + kcmd_addr = kmalloc(cmdbuf->size * 8, GFP_KERNEL); + if (kcmd_addr == NULL) + return -ENOMEM; + + if (DRM_COPY_FROM_USER(kcmd_addr, cmdbuf->cmd_addr, + cmdbuf->size * 8)) + { + kfree(kcmd_addr); + return -EFAULT; + } + cmdbuf->cmd_addr = kcmd_addr; + } + if (cmdbuf->vb_size) { + kvb_addr = kmalloc(cmdbuf->vb_size, GFP_KERNEL); + if (kvb_addr == NULL) { + ret = -ENOMEM; + goto done; + } + + if (DRM_COPY_FROM_USER(kvb_addr, cmdbuf->vb_addr, + cmdbuf->vb_size)) { + ret = -EFAULT; + goto done; + } + cmdbuf->vb_addr = kvb_addr; + } + if (cmdbuf->nbox) { + kbox_addr = kmalloc(cmdbuf->nbox * sizeof(struct drm_clip_rect), + GFP_KERNEL); + if (kbox_addr == NULL) { + ret = -ENOMEM; + goto done; + } + + if (DRM_COPY_FROM_USER(kbox_addr, cmdbuf->box_addr, + cmdbuf->nbox * sizeof(struct drm_clip_rect))) { + ret = -EFAULT; + goto done; + } + cmdbuf->box_addr = kbox_addr; + } + + /* Make sure writes to DMA buffers are finished before sending + * DMA commands to the graphics hardware. */ + DRM_MEMORYBARRIER(); + + /* Coming from user space. Don't know if the Xserver has + * emitted wait commands. Assuming the worst. */ + dev_priv->waiting = 1; + + i = 0; + first_draw_cmd = NULL; + while (i < cmdbuf->size) { + drm_savage_cmd_header_t cmd_header; + cmd_header = *(drm_savage_cmd_header_t *)cmdbuf->cmd_addr; + cmdbuf->cmd_addr++; + i++; + + /* Group drawing commands with same state to minimize + * iterations over clip rects. */ + j = 0; + switch (cmd_header.cmd.cmd) { + case SAVAGE_CMD_DMA_IDX: + case SAVAGE_CMD_VB_IDX: + j = (cmd_header.idx.count + 3) / 4; + if (i + j > cmdbuf->size) { + DRM_ERROR("indexed drawing command extends " + "beyond end of command buffer\n"); + DMA_FLUSH(); + return -EINVAL; + } + /* fall through */ + case SAVAGE_CMD_DMA_PRIM: + case SAVAGE_CMD_VB_PRIM: + if (!first_draw_cmd) + first_draw_cmd = cmdbuf->cmd_addr - 1; + cmdbuf->cmd_addr += j; + i += j; + break; + default: + if (first_draw_cmd) { + ret = savage_dispatch_draw( + dev_priv, first_draw_cmd, + cmdbuf->cmd_addr - 1, + dmabuf, cmdbuf->vb_addr, cmdbuf->vb_size, + cmdbuf->vb_stride, + cmdbuf->nbox, cmdbuf->box_addr); + if (ret != 0) + return ret; + first_draw_cmd = NULL; + } + } + if (first_draw_cmd) + continue; + + switch (cmd_header.cmd.cmd) { + case SAVAGE_CMD_STATE: + j = (cmd_header.state.count + 1) / 2; + if (i + j > cmdbuf->size) { + DRM_ERROR("command SAVAGE_CMD_STATE extends " + "beyond end of command buffer\n"); + DMA_FLUSH(); + ret = -EINVAL; + goto done; + } + ret = savage_dispatch_state(dev_priv, &cmd_header, + (const uint32_t *)cmdbuf->cmd_addr); + cmdbuf->cmd_addr += j; + i += j; + break; + case SAVAGE_CMD_CLEAR: + if (i + 1 > cmdbuf->size) { + DRM_ERROR("command SAVAGE_CMD_CLEAR extends " + "beyond end of command buffer\n"); + DMA_FLUSH(); + ret = -EINVAL; + goto done; + } + ret = savage_dispatch_clear(dev_priv, &cmd_header, + cmdbuf->cmd_addr, + cmdbuf->nbox, + cmdbuf->box_addr); + cmdbuf->cmd_addr++; + i++; + break; + case SAVAGE_CMD_SWAP: + ret = savage_dispatch_swap(dev_priv, cmdbuf->nbox, + cmdbuf->box_addr); + break; + default: + DRM_ERROR("invalid command 0x%x\n", + cmd_header.cmd.cmd); + DMA_FLUSH(); + ret = -EINVAL; + goto done; + } + + if (ret != 0) { + DMA_FLUSH(); + goto done; + } + } + + if (first_draw_cmd) { + ret = savage_dispatch_draw ( + dev_priv, first_draw_cmd, cmdbuf->cmd_addr, dmabuf, + cmdbuf->vb_addr, cmdbuf->vb_size, cmdbuf->vb_stride, + cmdbuf->nbox, cmdbuf->box_addr); + if (ret != 0) { + DMA_FLUSH(); + goto done; + } + } + + DMA_FLUSH(); + + if (dmabuf && cmdbuf->discard) { + drm_savage_buf_priv_t *buf_priv = dmabuf->dev_private; + uint16_t event; + event = savage_bci_emit_event(dev_priv, SAVAGE_WAIT_3D); + SET_AGE(&buf_priv->age, event, dev_priv->event_wrap); + savage_freelist_put(dev, dmabuf); + } + +done: + /* If we didn't need to allocate them, these'll be NULL */ + kfree(kcmd_addr); + kfree(kvb_addr); + kfree(kbox_addr); + + return ret; +} diff --git a/drivers/gpu/drm/sis/Makefile b/drivers/gpu/drm/sis/Makefile new file mode 100644 index 00000000..441c061c --- /dev/null +++ b/drivers/gpu/drm/sis/Makefile @@ -0,0 +1,10 @@ +# +# Makefile for the drm device driver. This driver provides support for the +# Direct Rendering Infrastructure (DRI) in XFree86 4.1.0 and higher. + +ccflags-y = -Iinclude/drm +sis-y := sis_drv.o sis_mm.o + +obj-$(CONFIG_DRM_SIS) += sis.o + + diff --git a/drivers/gpu/drm/sis/sis_drv.c b/drivers/gpu/drm/sis/sis_drv.c new file mode 100644 index 00000000..46d5be6e --- /dev/null +++ b/drivers/gpu/drm/sis/sis_drv.c @@ -0,0 +1,115 @@ +/* sis.c -- sis driver -*- linux-c -*- + * + * Copyright 1999 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + */ + +#include "drmP.h" +#include "sis_drm.h" +#include "sis_drv.h" + +#include "drm_pciids.h" + +static struct pci_device_id pciidlist[] = { + sisdrv_PCI_IDS +}; + +static int sis_driver_load(struct drm_device *dev, unsigned long chipset) +{ + drm_sis_private_t *dev_priv; + int ret; + + dev_priv = kzalloc(sizeof(drm_sis_private_t), GFP_KERNEL); + if (dev_priv == NULL) + return -ENOMEM; + + dev->dev_private = (void *)dev_priv; + dev_priv->chipset = chipset; + ret = drm_sman_init(&dev_priv->sman, 2, 12, 8); + if (ret) + kfree(dev_priv); + + return ret; +} + +static int sis_driver_unload(struct drm_device *dev) +{ + drm_sis_private_t *dev_priv = dev->dev_private; + + drm_sman_takedown(&dev_priv->sman); + kfree(dev_priv); + + return 0; +} + +static struct drm_driver driver = { + .driver_features = DRIVER_USE_AGP | DRIVER_USE_MTRR, + .load = sis_driver_load, + .unload = sis_driver_unload, + .dma_quiescent = sis_idle, + .reclaim_buffers = NULL, + .reclaim_buffers_idlelocked = sis_reclaim_buffers_locked, + .lastclose = sis_lastclose, + .ioctls = sis_ioctls, + .fops = { + .owner = THIS_MODULE, + .open = drm_open, + .release = drm_release, + .unlocked_ioctl = drm_ioctl, + .mmap = drm_mmap, + .poll = drm_poll, + .fasync = drm_fasync, + .llseek = noop_llseek, + }, + + .name = DRIVER_NAME, + .desc = DRIVER_DESC, + .date = DRIVER_DATE, + .major = DRIVER_MAJOR, + .minor = DRIVER_MINOR, + .patchlevel = DRIVER_PATCHLEVEL, +}; + +static struct pci_driver sis_pci_driver = { + .name = DRIVER_NAME, + .id_table = pciidlist, +}; + +static int __init sis_init(void) +{ + driver.num_ioctls = sis_max_ioctl; + return drm_pci_init(&driver, &sis_pci_driver); +} + +static void __exit sis_exit(void) +{ + drm_pci_exit(&driver, &sis_pci_driver); +} + +module_init(sis_init); +module_exit(sis_exit); + +MODULE_AUTHOR(DRIVER_AUTHOR); +MODULE_DESCRIPTION(DRIVER_DESC); +MODULE_LICENSE("GPL and additional rights"); diff --git a/drivers/gpu/drm/sis/sis_drv.h b/drivers/gpu/drm/sis/sis_drv.h new file mode 100644 index 00000000..ef940bad --- /dev/null +++ b/drivers/gpu/drm/sis/sis_drv.h @@ -0,0 +1,73 @@ +/* sis_drv.h -- Private header for sis driver -*- linux-c -*- */ +/* + * Copyright 1999 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California. + * All rights reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef _SIS_DRV_H_ +#define _SIS_DRV_H_ + +/* General customization: + */ + +#define DRIVER_AUTHOR "SIS, Tungsten Graphics" +#define DRIVER_NAME "sis" +#define DRIVER_DESC "SIS 300/630/540 and XGI V3XE/V5/V8" +#define DRIVER_DATE "20070626" +#define DRIVER_MAJOR 1 +#define DRIVER_MINOR 3 +#define DRIVER_PATCHLEVEL 0 + +enum sis_family { + SIS_OTHER = 0, + SIS_CHIP_315 = 1, +}; + +#include "drm_sman.h" + + +#define SIS_BASE (dev_priv->mmio) +#define SIS_READ(reg) DRM_READ32(SIS_BASE, reg); +#define SIS_WRITE(reg, val) DRM_WRITE32(SIS_BASE, reg, val); + +typedef struct drm_sis_private { + drm_local_map_t *mmio; + unsigned int idle_fault; + struct drm_sman sman; + unsigned int chipset; + int vram_initialized; + int agp_initialized; + unsigned long vram_offset; + unsigned long agp_offset; +} drm_sis_private_t; + +extern int sis_idle(struct drm_device *dev); +extern void sis_reclaim_buffers_locked(struct drm_device *dev, + struct drm_file *file_priv); +extern void sis_lastclose(struct drm_device *dev); + +extern struct drm_ioctl_desc sis_ioctls[]; +extern int sis_max_ioctl; + +#endif diff --git a/drivers/gpu/drm/sis/sis_mm.c b/drivers/gpu/drm/sis/sis_mm.c new file mode 100644 index 00000000..7fe2b634 --- /dev/null +++ b/drivers/gpu/drm/sis/sis_mm.c @@ -0,0 +1,331 @@ +/************************************************************************** + * + * Copyright 2006 Tungsten Graphics, Inc., Bismarck, ND., USA. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. + * + * + **************************************************************************/ + +/* + * Authors: + * Thomas Hellström + */ + +#include "drmP.h" +#include "sis_drm.h" +#include "sis_drv.h" + +#include