aboutsummaryrefslogtreecommitdiffstats
path: root/target/linux/brcm2708/patches-4.4/0281-drm-vc4-Add-support-for-scaling-of-display-planes.patch
blob: 45af0d98ef1103b01003edf26b55016dc53aca71 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
From 446bf09980764ce077b74a85af9f141e6192d3dc Mon Sep 17 00:00:00 2001
From: Eric Anholt <eric@anholt.net>
Date: Tue, 20 Oct 2015 16:06:57 +0100
Subject: [PATCH] drm/vc4: Add support for scaling of display planes.

This implements a simple policy for choosing scaling modes
(trapezoidal for decimation, PPF for magnification), and a single PPF
filter (Mitchell/Netravali's recommendation).

Signed-off-by: Eric Anholt <eric@anholt.net>
(cherry picked from commit 21af94cf1a4c2d3450ab7fead58e6e2291ab92a9)
---
 drivers/gpu/drm/vc4/vc4_drv.h   |   4 +
 drivers/gpu/drm/vc4/vc4_hvs.c   |  84 +++++++++++++
 drivers/gpu/drm/vc4/vc4_plane.c | 253 +++++++++++++++++++++++++++++++++++++---
 drivers/gpu/drm/vc4/vc4_regs.h  |  46 ++++++++
 4 files changed, 374 insertions(+), 13 deletions(-)

--- a/drivers/gpu/drm/vc4/vc4_drv.h
+++ b/drivers/gpu/drm/vc4/vc4_drv.h
@@ -156,7 +156,11 @@ struct vc4_hvs {
 	 * list.  Units are dwords.
 	 */
 	struct drm_mm dlist_mm;
+	/* Memory manager for the LBM memory used by HVS scaling. */
+	struct drm_mm lbm_mm;
 	spinlock_t mm_lock;
+
+	struct drm_mm_node mitchell_netravali_filter;
 };
 
 struct vc4_plane {
--- a/drivers/gpu/drm/vc4/vc4_hvs.c
+++ b/drivers/gpu/drm/vc4/vc4_hvs.c
@@ -100,12 +100,76 @@ int vc4_hvs_debugfs_regs(struct seq_file
 }
 #endif
 
+/* The filter kernel is composed of dwords each containing 3 9-bit
+ * signed integers packed next to each other.
+ */
+#define VC4_INT_TO_COEFF(coeff) (coeff & 0x1ff)
+#define VC4_PPF_FILTER_WORD(c0, c1, c2)				\
+	((((c0) & 0x1ff) << 0) |				\
+	 (((c1) & 0x1ff) << 9) |				\
+	 (((c2) & 0x1ff) << 18))
+
+/* The whole filter kernel is arranged as the coefficients 0-16 going
+ * up, then a pad, then 17-31 going down and reversed within the
+ * dwords.  This means that a linear phase kernel (where it's
+ * symmetrical at the boundary between 15 and 16) has the last 5
+ * dwords matching the first 5, but reversed.
+ */
+#define VC4_LINEAR_PHASE_KERNEL(c0, c1, c2, c3, c4, c5, c6, c7, c8,	\
+				c9, c10, c11, c12, c13, c14, c15)	\
+	{VC4_PPF_FILTER_WORD(c0, c1, c2),				\
+	 VC4_PPF_FILTER_WORD(c3, c4, c5),				\
+	 VC4_PPF_FILTER_WORD(c6, c7, c8),				\
+	 VC4_PPF_FILTER_WORD(c9, c10, c11),				\
+	 VC4_PPF_FILTER_WORD(c12, c13, c14),				\
+	 VC4_PPF_FILTER_WORD(c15, c15, 0)}
+
+#define VC4_LINEAR_PHASE_KERNEL_DWORDS 6
+#define VC4_KERNEL_DWORDS (VC4_LINEAR_PHASE_KERNEL_DWORDS * 2 - 1)
+
+/* Recommended B=1/3, C=1/3 filter choice from Mitchell/Netravali.
+ * http://www.cs.utexas.edu/~fussell/courses/cs384g/lectures/mitchell/Mitchell.pdf
+ */
+static const u32 mitchell_netravali_1_3_1_3_kernel[] =
+	VC4_LINEAR_PHASE_KERNEL(0, -2, -6, -8, -10, -8, -3, 2, 18,
+				50, 82, 119, 155, 187, 213, 227);
+
+static int vc4_hvs_upload_linear_kernel(struct vc4_hvs *hvs,
+					struct drm_mm_node *space,
+					const u32 *kernel)
+{
+	int ret, i;
+	u32 __iomem *dst_kernel;
+
+	ret = drm_mm_insert_node(&hvs->dlist_mm, space, VC4_KERNEL_DWORDS, 1,
+				 0);
+	if (ret) {
+		DRM_ERROR("Failed to allocate space for filter kernel: %d\n",
+			  ret);
+		return ret;
+	}
+
+	dst_kernel = hvs->dlist + space->start;
+
+	for (i = 0; i < VC4_KERNEL_DWORDS; i++) {
+		if (i < VC4_LINEAR_PHASE_KERNEL_DWORDS)
+			writel(kernel[i], &dst_kernel[i]);
+		else {
+			writel(kernel[VC4_KERNEL_DWORDS - i - 1],
+			       &dst_kernel[i]);
+		}
+	}
+
+	return 0;
+}
+
 static int vc4_hvs_bind(struct device *dev, struct device *master, void *data)
 {
 	struct platform_device *pdev = to_platform_device(dev);
 	struct drm_device *drm = dev_get_drvdata(master);
 	struct vc4_dev *vc4 = drm->dev_private;
 	struct vc4_hvs *hvs = NULL;
+	int ret;
 
 	hvs = devm_kzalloc(&pdev->dev, sizeof(*hvs), GFP_KERNEL);
 	if (!hvs)
@@ -130,6 +194,22 @@ static int vc4_hvs_bind(struct device *d
 		    HVS_BOOTLOADER_DLIST_END,
 		    (SCALER_DLIST_SIZE >> 2) - HVS_BOOTLOADER_DLIST_END);
 
+	/* Set up the HVS LBM memory manager.  We could have some more
+	 * complicated data structure that allowed reuse of LBM areas
+	 * between planes when they don't overlap on the screen, but
+	 * for now we just allocate globally.
+	 */
+	drm_mm_init(&hvs->lbm_mm, 0, 96 * 1024);
+
+	/* Upload filter kernels.  We only have the one for now, so we
+	 * keep it around for the lifetime of the driver.
+	 */
+	ret = vc4_hvs_upload_linear_kernel(hvs,
+					   &hvs->mitchell_netravali_filter,
+					   mitchell_netravali_1_3_1_3_kernel);
+	if (ret)
+		return ret;
+
 	vc4->hvs = hvs;
 	return 0;
 }
@@ -140,7 +220,11 @@ static void vc4_hvs_unbind(struct device
 	struct drm_device *drm = dev_get_drvdata(master);
 	struct vc4_dev *vc4 = drm->dev_private;
 
+	if (vc4->hvs->mitchell_netravali_filter.allocated)
+		drm_mm_remove_node(&vc4->hvs->mitchell_netravali_filter);
+
 	drm_mm_takedown(&vc4->hvs->dlist_mm);
+	drm_mm_takedown(&vc4->hvs->lbm_mm);
 
 	vc4->hvs = NULL;
 }
--- a/drivers/gpu/drm/vc4/vc4_plane.c
+++ b/drivers/gpu/drm/vc4/vc4_plane.c
@@ -24,6 +24,12 @@
 #include "drm_fb_cma_helper.h"
 #include "drm_plane_helper.h"
 
+enum vc4_scaling_mode {
+	VC4_SCALING_NONE,
+	VC4_SCALING_TPZ,
+	VC4_SCALING_PPF,
+};
+
 struct vc4_plane_state {
 	struct drm_plane_state base;
 	/* System memory copy of the display list for this element, computed
@@ -47,13 +53,19 @@ struct vc4_plane_state {
 
 	/* Clipped coordinates of the plane on the display. */
 	int crtc_x, crtc_y, crtc_w, crtc_h;
-	/* Clipped size of the area scanned from in the FB. */
-	u32 src_w, src_h;
+	/* Clipped area being scanned from in the FB. */
+	u32 src_x, src_y, src_w, src_h;
+
+	enum vc4_scaling_mode x_scaling, y_scaling;
+	bool is_unity;
 
 	/* Offset to start scanning out from the start of the plane's
 	 * BO.
 	 */
 	u32 offset;
+
+	/* Our allocation in LBM for temporary storage during scaling. */
+	struct drm_mm_node lbm;
 };
 
 static inline struct vc4_plane_state *
@@ -106,6 +118,16 @@ static const struct hvs_format *vc4_get_
 	return NULL;
 }
 
+static enum vc4_scaling_mode vc4_get_scaling_mode(u32 src, u32 dst)
+{
+	if (dst > src)
+		return VC4_SCALING_PPF;
+	else if (dst < src)
+		return VC4_SCALING_TPZ;
+	else
+		return VC4_SCALING_NONE;
+}
+
 static bool plane_enabled(struct drm_plane_state *state)
 {
 	return state->fb && state->crtc;
@@ -122,6 +144,8 @@ static struct drm_plane_state *vc4_plane
 	if (!vc4_state)
 		return NULL;
 
+	memset(&vc4_state->lbm, 0, sizeof(vc4_state->lbm));
+
 	__drm_atomic_helper_plane_duplicate_state(plane, &vc4_state->base);
 
 	if (vc4_state->dlist) {
@@ -141,8 +165,17 @@ static struct drm_plane_state *vc4_plane
 static void vc4_plane_destroy_state(struct drm_plane *plane,
 				    struct drm_plane_state *state)
 {
+	struct vc4_dev *vc4 = to_vc4_dev(plane->dev);
 	struct vc4_plane_state *vc4_state = to_vc4_plane_state(state);
 
+	if (vc4_state->lbm.allocated) {
+		unsigned long irqflags;
+
+		spin_lock_irqsave(&vc4->hvs->mm_lock, irqflags);
+		drm_mm_remove_node(&vc4_state->lbm);
+		spin_unlock_irqrestore(&vc4->hvs->mm_lock, irqflags);
+	}
+
 	kfree(vc4_state->dlist);
 	__drm_atomic_helper_plane_destroy_state(plane, &vc4_state->base);
 	kfree(state);
@@ -181,23 +214,60 @@ static void vc4_dlist_write(struct vc4_p
 	vc4_state->dlist[vc4_state->dlist_count++] = val;
 }
 
+/* Returns the scl0/scl1 field based on whether the dimensions need to
+ * be up/down/non-scaled.
+ *
+ * This is a replication of a table from the spec.
+ */
+static u32 vc4_get_scl_field(struct drm_plane_state *state)
+{
+	struct vc4_plane_state *vc4_state = to_vc4_plane_state(state);
+
+	switch (vc4_state->x_scaling << 2 | vc4_state->y_scaling) {
+	case VC4_SCALING_PPF << 2 | VC4_SCALING_PPF:
+		return SCALER_CTL0_SCL_H_PPF_V_PPF;
+	case VC4_SCALING_TPZ << 2 | VC4_SCALING_PPF:
+		return SCALER_CTL0_SCL_H_TPZ_V_PPF;
+	case VC4_SCALING_PPF << 2 | VC4_SCALING_TPZ:
+		return SCALER_CTL0_SCL_H_PPF_V_TPZ;
+	case VC4_SCALING_TPZ << 2 | VC4_SCALING_TPZ:
+		return SCALER_CTL0_SCL_H_TPZ_V_TPZ;
+	case VC4_SCALING_PPF << 2 | VC4_SCALING_NONE:
+		return SCALER_CTL0_SCL_H_PPF_V_NONE;
+	case VC4_SCALING_NONE << 2 | VC4_SCALING_PPF:
+		return SCALER_CTL0_SCL_H_NONE_V_PPF;
+	case VC4_SCALING_NONE << 2 | VC4_SCALING_TPZ:
+		return SCALER_CTL0_SCL_H_NONE_V_TPZ;
+	case VC4_SCALING_TPZ << 2 | VC4_SCALING_NONE:
+		return SCALER_CTL0_SCL_H_TPZ_V_NONE;
+	default:
+	case VC4_SCALING_NONE << 2 | VC4_SCALING_NONE:
+		/* The unity case is independently handled by
+		 * SCALER_CTL0_UNITY.
+		 */
+		return 0;
+	}
+}
+
 static int vc4_plane_setup_clipping_and_scaling(struct drm_plane_state *state)
 {
+	struct drm_plane *plane = state->plane;
 	struct vc4_plane_state *vc4_state = to_vc4_plane_state(state);
 	struct drm_framebuffer *fb = state->fb;
+	u32 subpixel_src_mask = (1 << 16) - 1;
 
 	vc4_state->offset = fb->offsets[0];
 
-	if (state->crtc_w << 16 != state->src_w ||
-	    state->crtc_h << 16 != state->src_h) {
-		/* We don't support scaling yet, which involves
-		 * allocating the LBM memory for scaling temporary
-		 * storage, and putting filter kernels in the HVS
-		 * context.
-		 */
+	/* We don't support subpixel source positioning for scaling. */
+	if ((state->src_x & subpixel_src_mask) ||
+	    (state->src_y & subpixel_src_mask) ||
+	    (state->src_w & subpixel_src_mask) ||
+	    (state->src_h & subpixel_src_mask)) {
 		return -EINVAL;
 	}
 
+	vc4_state->src_x = state->src_x >> 16;
+	vc4_state->src_y = state->src_y >> 16;
 	vc4_state->src_w = state->src_w >> 16;
 	vc4_state->src_h = state->src_h >> 16;
 
@@ -206,6 +276,23 @@ static int vc4_plane_setup_clipping_and_
 	vc4_state->crtc_w = state->crtc_w;
 	vc4_state->crtc_h = state->crtc_h;
 
+	vc4_state->x_scaling = vc4_get_scaling_mode(vc4_state->src_w,
+						    vc4_state->crtc_w);
+	vc4_state->y_scaling = vc4_get_scaling_mode(vc4_state->src_h,
+						    vc4_state->crtc_h);
+	vc4_state->is_unity = (vc4_state->x_scaling == VC4_SCALING_NONE &&
+			       vc4_state->y_scaling == VC4_SCALING_NONE);
+
+	/* No configuring scaling on the cursor plane, since it gets
+	   non-vblank-synced updates, and scaling requires requires
+	   LBM changes which have to be vblank-synced.
+	 */
+	if (plane->type == DRM_PLANE_TYPE_CURSOR && !vc4_state->is_unity)
+		return -EINVAL;
+
+	/* Clamp the on-screen start x/y to 0.  The hardware doesn't
+	 * support negative y, and negative x wastes bandwidth.
+	 */
 	if (vc4_state->crtc_x < 0) {
 		vc4_state->offset += (drm_format_plane_cpp(fb->pixel_format,
 							   0) *
@@ -223,6 +310,87 @@ static int vc4_plane_setup_clipping_and_
 	return 0;
 }
 
+static void vc4_write_tpz(struct vc4_plane_state *vc4_state, u32 src, u32 dst)
+{
+	u32 scale, recip;
+
+	scale = (1 << 16) * src / dst;
+
+	/* The specs note that while the reciprocal would be defined
+	 * as (1<<32)/scale, ~0 is close enough.
+	 */
+	recip = ~0 / scale;
+
+	vc4_dlist_write(vc4_state,
+			VC4_SET_FIELD(scale, SCALER_TPZ0_SCALE) |
+			VC4_SET_FIELD(0, SCALER_TPZ0_IPHASE));
+	vc4_dlist_write(vc4_state,
+			VC4_SET_FIELD(recip, SCALER_TPZ1_RECIP));
+}
+
+static void vc4_write_ppf(struct vc4_plane_state *vc4_state, u32 src, u32 dst)
+{
+	u32 scale = (1 << 16) * src / dst;
+
+	vc4_dlist_write(vc4_state,
+			SCALER_PPF_AGC |
+			VC4_SET_FIELD(scale, SCALER_PPF_SCALE) |
+			VC4_SET_FIELD(0, SCALER_PPF_IPHASE));
+}
+
+static u32 vc4_lbm_size(struct drm_plane_state *state)
+{
+	struct vc4_plane_state *vc4_state = to_vc4_plane_state(state);
+	/* This is the worst case number.  One of the two sizes will
+	 * be used depending on the scaling configuration.
+	 */
+	u32 pix_per_line = max(vc4_state->src_w, (u32)vc4_state->crtc_w);
+	u32 lbm;
+
+	if (vc4_state->is_unity)
+		return 0;
+	else if (vc4_state->y_scaling == VC4_SCALING_TPZ)
+		lbm = pix_per_line * 8;
+	else {
+		/* In special cases, this multiplier might be 12. */
+		lbm = pix_per_line * 16;
+	}
+
+	lbm = roundup(lbm, 32);
+
+	return lbm;
+}
+
+static void vc4_write_scaling_parameters(struct drm_plane_state *state)
+{
+	struct vc4_plane_state *vc4_state = to_vc4_plane_state(state);
+
+	/* Ch0 H-PPF Word 0: Scaling Parameters */
+	if (vc4_state->x_scaling == VC4_SCALING_PPF) {
+		vc4_write_ppf(vc4_state,
+			      vc4_state->src_w, vc4_state->crtc_w);
+	}
+
+	/* Ch0 V-PPF Words 0-1: Scaling Parameters, Context */
+	if (vc4_state->y_scaling == VC4_SCALING_PPF) {
+		vc4_write_ppf(vc4_state,
+			      vc4_state->src_h, vc4_state->crtc_h);
+		vc4_dlist_write(vc4_state, 0xc0c0c0c0);
+	}
+
+	/* Ch0 H-TPZ Words 0-1: Scaling Parameters, Recip */
+	if (vc4_state->x_scaling == VC4_SCALING_TPZ) {
+		vc4_write_tpz(vc4_state,
+			      vc4_state->src_w, vc4_state->crtc_w);
+	}
+
+	/* Ch0 V-TPZ Words 0-2: Scaling Parameters, Recip, Context */
+	if (vc4_state->y_scaling == VC4_SCALING_TPZ) {
+		vc4_write_tpz(vc4_state,
+			      vc4_state->src_h, vc4_state->crtc_h);
+		vc4_dlist_write(vc4_state, 0xc0c0c0c0);
+	}
+}
 
 /* Writes out a full display list for an active plane to the plane's
  * private dlist state.
@@ -230,22 +398,50 @@ static int vc4_plane_setup_clipping_and_
 static int vc4_plane_mode_set(struct drm_plane *plane,
 			      struct drm_plane_state *state)
 {
+	struct vc4_dev *vc4 = to_vc4_dev(plane->dev);
 	struct vc4_plane_state *vc4_state = to_vc4_plane_state(state);
 	struct drm_framebuffer *fb = state->fb;
 	struct drm_gem_cma_object *bo = drm_fb_cma_get_gem_obj(fb, 0);
 	u32 ctl0_offset = vc4_state->dlist_count;
 	const struct hvs_format *format = vc4_get_hvs_format(fb->pixel_format);
+	u32 scl;
+	u32 lbm_size;
+	unsigned long irqflags;
 	int ret;
 
 	ret = vc4_plane_setup_clipping_and_scaling(state);
 	if (ret)
 		return ret;
 
+	/* Allocate the LBM memory that the HVS will use for temporary
+	 * storage due to our scaling/format conversion.
+	 */
+	lbm_size = vc4_lbm_size(state);
+	if (lbm_size) {
+		if (!vc4_state->lbm.allocated) {
+			spin_lock_irqsave(&vc4->hvs->mm_lock, irqflags);
+			ret = drm_mm_insert_node(&vc4->hvs->lbm_mm,
+						 &vc4_state->lbm,
+						 lbm_size, 32, 0);
+			spin_unlock_irqrestore(&vc4->hvs->mm_lock, irqflags);
+		} else {
+			WARN_ON_ONCE(lbm_size != vc4_state->lbm.size);
+		}
+	}
+
+	if (ret)
+		return ret;
+
+	scl = vc4_get_scl_field(state);
+
+	/* Control word */
 	vc4_dlist_write(vc4_state,
 			SCALER_CTL0_VALID |
 			(format->pixel_order << SCALER_CTL0_ORDER_SHIFT) |
 			(format->hvs << SCALER_CTL0_PIXEL_FORMAT_SHIFT) |
-			SCALER_CTL0_UNITY);
+			(vc4_state->is_unity ? SCALER_CTL0_UNITY : 0) |
+			VC4_SET_FIELD(scl, SCALER_CTL0_SCL0) |
+			VC4_SET_FIELD(scl, SCALER_CTL0_SCL1));
 
 	/* Position Word 0: Image Positions and Alpha Value */
 	vc4_state->pos0_offset = vc4_state->dlist_count;
@@ -254,9 +450,14 @@ static int vc4_plane_mode_set(struct drm
 			VC4_SET_FIELD(vc4_state->crtc_x, SCALER_POS0_START_X) |
 			VC4_SET_FIELD(vc4_state->crtc_y, SCALER_POS0_START_Y));
 
-	/* Position Word 1: Scaled Image Dimensions.
-	 * Skipped due to SCALER_CTL0_UNITY scaling.
-	 */
+	/* Position Word 1: Scaled Image Dimensions. */
+	if (!vc4_state->is_unity) {
+		vc4_dlist_write(vc4_state,
+				VC4_SET_FIELD(vc4_state->crtc_w,
+					      SCALER_POS1_SCL_WIDTH) |
+				VC4_SET_FIELD(vc4_state->crtc_h,
+					      SCALER_POS1_SCL_HEIGHT));
+	}
 
 	/* Position Word 2: Source Image Size, Alpha Mode */
 	vc4_state->pos2_offset = vc4_state->dlist_count;
@@ -282,6 +483,32 @@ static int vc4_plane_mode_set(struct drm
 	vc4_dlist_write(vc4_state,
 			VC4_SET_FIELD(fb->pitches[0], SCALER_SRC_PITCH));
 
+	if (!vc4_state->is_unity) {
+		/* LBM Base Address. */
+		if (vc4_state->y_scaling != VC4_SCALING_NONE)
+			vc4_dlist_write(vc4_state, vc4_state->lbm.start);
+
+		vc4_write_scaling_parameters(state);
+
+		/* If any PPF setup was done, then all the kernel
+		 * pointers get uploaded.
+		 */
+		if (vc4_state->x_scaling == VC4_SCALING_PPF ||
+		    vc4_state->y_scaling == VC4_SCALING_PPF) {
+			u32 kernel = VC4_SET_FIELD(vc4->hvs->mitchell_netravali_filter.start,
+						   SCALER_PPF_KERNEL_OFFSET);
+
+			/* HPPF plane 0 */
+			vc4_dlist_write(vc4_state, kernel);
+			/* VPPF plane 0 */
+			vc4_dlist_write(vc4_state, kernel);
+			/* HPPF plane 1 */
+			vc4_dlist_write(vc4_state, kernel);
+			/* VPPF plane 1 */
+			vc4_dlist_write(vc4_state, kernel);
+		}
+	}
+
 	vc4_state->dlist[ctl0_offset] |=
 		VC4_SET_FIELD(vc4_state->dlist_count, SCALER_CTL0_SIZE);
 
--- a/drivers/gpu/drm/vc4/vc4_regs.h
+++ b/drivers/gpu/drm/vc4/vc4_regs.h
@@ -536,6 +536,21 @@ enum hvs_pixel_format {
 #define SCALER_CTL0_ORDER_MASK			VC4_MASK(14, 13)
 #define SCALER_CTL0_ORDER_SHIFT			13
 
+#define SCALER_CTL0_SCL1_MASK			VC4_MASK(10, 8)
+#define SCALER_CTL0_SCL1_SHIFT			8
+
+#define SCALER_CTL0_SCL0_MASK			VC4_MASK(7, 5)
+#define SCALER_CTL0_SCL0_SHIFT			5
+
+#define SCALER_CTL0_SCL_H_PPF_V_PPF		0
+#define SCALER_CTL0_SCL_H_TPZ_V_PPF		1
+#define SCALER_CTL0_SCL_H_PPF_V_TPZ		2
+#define SCALER_CTL0_SCL_H_TPZ_V_TPZ		3
+#define SCALER_CTL0_SCL_H_PPF_V_NONE		4
+#define SCALER_CTL0_SCL_H_NONE_V_PPF		5
+#define SCALER_CTL0_SCL_H_NONE_V_TPZ		6
+#define SCALER_CTL0_SCL_H_TPZ_V_NONE		7
+
 /* Set to indicate no scaling. */
 #define SCALER_CTL0_UNITY			BIT(4)
 
@@ -551,6 +566,12 @@ enum hvs_pixel_format {
 #define SCALER_POS0_START_X_MASK		VC4_MASK(11, 0)
 #define SCALER_POS0_START_X_SHIFT		0
 
+#define SCALER_POS1_SCL_HEIGHT_MASK		VC4_MASK(27, 16)
+#define SCALER_POS1_SCL_HEIGHT_SHIFT		16
+
+#define SCALER_POS1_SCL_WIDTH_MASK		VC4_MASK(11, 0)
+#define SCALER_POS1_SCL_WIDTH_SHIFT		0
+
 #define SCALER_POS2_ALPHA_MODE_MASK		VC4_MASK(31, 30)
 #define SCALER_POS2_ALPHA_MODE_SHIFT		30
 #define SCALER_POS2_ALPHA_MODE_PIPELINE		0
@@ -564,6 +585,31 @@ enum hvs_pixel_format {
 #define SCALER_POS2_WIDTH_MASK			VC4_MASK(11, 0)
 #define SCALER_POS2_WIDTH_SHIFT			0
 
+#define SCALER_TPZ0_VERT_RECALC			BIT(31)
+#define SCALER_TPZ0_SCALE_MASK			VC4_MASK(28, 8)
+#define SCALER_TPZ0_SCALE_SHIFT			8
+#define SCALER_TPZ0_IPHASE_MASK			VC4_MASK(7, 0)
+#define SCALER_TPZ0_IPHASE_SHIFT		0
+#define SCALER_TPZ1_RECIP_MASK			VC4_MASK(15, 0)
+#define SCALER_TPZ1_RECIP_SHIFT			0
+
+/* Skips interpolating coefficients to 64 phases, so just 8 are used.
+ * Required for nearest neighbor.
+ */
+#define SCALER_PPF_NOINTERP			BIT(31)
+/* Replaes the highest valued coefficient with one that makes all 4
+ * sum to unity.
+ */
+#define SCALER_PPF_AGC				BIT(30)
+#define SCALER_PPF_SCALE_MASK			VC4_MASK(24, 8)
+#define SCALER_PPF_SCALE_SHIFT			8
+#define SCALER_PPF_IPHASE_MASK			VC4_MASK(6, 0)
+#define SCALER_PPF_IPHASE_SHIFT			0
+
+#define SCALER_PPF_KERNEL_OFFSET_MASK		VC4_MASK(13, 0)
+#define SCALER_PPF_KERNEL_OFFSET_SHIFT		0
+#define SCALER_PPF_KERNEL_UNCACHED		BIT(31)
+
 #define SCALER_SRC_PITCH_MASK			VC4_MASK(15, 0)
 #define SCALER_SRC_PITCH_SHIFT			0