aboutsummaryrefslogtreecommitdiffstats
path: root/target/linux/ar71xx/files/drivers/net/ethernet/atheros/ag71xx/ag71xx_main.c
Commit message (Collapse)AuthorAgeFilesLines
* ar71xx: fix QCA955X SGMII link lossDavid Bauer2018-09-031-0/+79
| | | | | | | | | | | | | | | | | | | | The QCA955X is affected by a hardware bug which causes link-loss of the SGMII link between SoC and PHY. This happens on change of link-state or speed. It is not really known what causes this bug. It definitely occurs when using a AR8033 Gigabit Ethernet PHY. Qualcomm solves this Bug in a similar fashion. We need to apply the fix on a per-device base via platform-data as performing the fixup work will break connectivity in case the SGMII interface is connected to a Switch. This bug was first proposed to be fixed by Sven Eckelmann in 2016. https://patchwork.ozlabs.org/patch/604782/ Based-on-patch-by: Sven Eckelmann <sven.eckelmann@open-mesh.com> Signed-off-by: David Bauer <mail@david-bauer.net> (cherry picked from commit f4f99ec9737c653815268f2efad0210caaa32e2d)
* ar71xx: remove pdata->fifo_cfg*Felix Fietkau2018-03-041-3/+3
| | | | | | The values are the same for all chipsets that use it Signed-off-by: Felix Fietkau <nbd@nbd.name>
* ag71xx: Move timestamp struct member outside of struct.Rosen Penev2018-02-281-2/+3
| | | | | | | | | | | | | | | With this change, the timestamp variable is only used in ag71xx_check_dma_stuck. Small tx speedup. Based on a Qualcomm commit. ag->timestamp = jiffies was not replaced with netif_trans_update(dev) because of this quote: It should be noted that after this series several instances of netif_trans_update() are useless (if they occur in .ndo_start_xmit and driver doesn't set LLTX flag -- stack already did an update). From: http://lists.openwall.net/netdev/2016/05/03/87 Signed-off-by: Rosen Penev <rosenp@gmail.com>
* ar71xx: use global timestamp for hang checkFelix Fietkau2018-02-051-4/+4
| | | | | | | Shrink the size of struct ag71xx_buf to 8 bytes, which improves cache footprint Signed-off-by: Felix Fietkau <nbd@nbd.name>
* ar71xx: only access device stats in tx handler if packets were processedFelix Fietkau2018-02-051-3/+3
| | | | Signed-off-by: Felix Fietkau <nbd@nbd.name>
* ar71xx: allocate rx/tx descriptor/buffers in one chunkFelix Fietkau2018-02-051-47/+38
| | | | | | Reduces false sharing due to cache aliases Signed-off-by: Felix Fietkau <nbd@nbd.name>
* ar71xx: remove obsolete '< KERNEL_VERSION(4, 2, 0)' checkGabor Juhos2018-01-091-7/+0
| | | | | | | | | Because the oldest supported kernel version on the ar71xx target is 4.4, the condition that the kernel version is smaller than 4.2.0 is always false. Remove the obsolete check from ag71xx_main.c to clean up the code a bit. Signed-off-by: Gabor Juhos <juhosg@freemail.hu>
* ag71xx: Fix rx ring buffer stall on small packets flood on qca956x and qca953x.Vittorio Gambaletta2017-12-291-0/+16
| | | | | | Backported from Code Aurora QSDK Signed-off-by: Vittorio Gambaletta <openwrt@vittgam.net>
* Revert "ag71xx: Switch from driver to kernel macro for NAPI_WEIGHT."Rosen Penev2017-12-081-1/+1
| | | | | | The motivation for this was misguided. It turns out tuning the NAPI weight could be useful for testing purposes. Therefore reverting. Signed-off-by: Rosen Penev <rosenp@gmail.com>
* ag71xx: Switch from driver to kernel macro for NAPI_WEIGHT.Rosen Penev2017-11-061-1/+1
| | | | | | NAPI_POLL_WEIGHT was introduced in the kernel for exactly this purpose 5 years ago. Signed-off-by: Rosen Penev <rosenp@gmail.com>
* Revert "ar71xx: Add GRO support to ag71xx"Felix Fietkau2017-10-171-3/+3
| | | | | | | | | | This reverts commit 13e5e473699b92f171205e0f5c57c9ebe7922492. This commit causes a severe regression in LAN->WAN routing performance for several devices. This appears to be caused by the extra requirement to validate the SKB checksum early in the rx path, which the ethernet hardware does not do Signed-off-by: Felix Fietkau <nbd@nbd.name>
* ar71xx: Add GRO support to ag71xxRosen Penev2017-09-111-3/+3
| | | | | | On a TL-WN710N, this patch increases iperf performance from ~92.5 to ~93.5 mbps. Keep in mind the WN710N is a 100mbps device. I expect greater numbers from gigabit devices. Signed-off-by: Rosen Penev <rosenp@gmail.com>
* ar71xx: fix ethernet driver fast reset issue causing tx timeoutsFelix Fietkau2016-09-031-1/+2
| | | | Signed-off-by: Felix Fietkau <nbd@nbd.name>
* ar71xx: add software transmit timestamp supportStephen Walker2016-07-151-0/+2
| | | | | | Add software transmit timestamp and ethtool (-T) timestamp support Signed-off-by: Stephen Walker <stephendwalker+github@gmail.com>
* ar71xx: prevent spurious ethernet resets from dma hang check false positivesFelix Fietkau2016-07-021-5/+11
| | | | Signed-off-by: Felix Fietkau <nbd@nbd.name>
* ar71xx: rename ethernet pdata->builtin_switch to use_flow_controlFelix Fietkau2016-06-271-1/+1
| | | | Signed-off-by: Felix Fietkau <nbd@nbd.name>
* Revert "ar71xx: prevent spurious ethernet resets from dma hang check false ↵Felix Fietkau2016-06-261-5/+4
| | | | | | positives" This reverts commit 3d58d7f053eae83b59baf5c8b8a58da78653f605.
* ar71xx: prevent spurious ethernet resets from dma hang check false positivesFelix Fietkau2016-06-261-4/+5
| | | | Signed-off-by: Felix Fietkau <nbd@nbd.name>
* ar71xx: enable flow control for ethernet MACs with built-in switchFelix Fietkau2016-06-261-1/+4
| | | | | | | | Should fix LAN speed issues on some devices. This is an updated version of the previously reverted commit with the same name. It improves the check for MACs connected to a built-in switch Signed-off-by: Felix Fietkau <nbd@nbd.name>
* Revert "ar71xx: enable flow control for ethernet MACs with built-in switch"Felix Fietkau2016-06-141-9/+1
| | | | | | | | This reverts commit 1beb5bec6453abb220c5be4ddb9e6dc02d08d735. It was found to cause the WAN port to fail on some AR934x devices Signed-off-by: Felix Fietkau <nbd@nbd.name>
* ar71xx: enable flow control for ethernet MACs with built-in switchFelix Fietkau2016-06-131-1/+9
| | | | | | Should fix LAN speed issues on some devices Signed-off-by: Felix Fietkau <nbd@nbd.name>
* ar71xx: reset ethernet tx ring on fast reset to prevent packet loss / irq issuesFelix Fietkau2016-02-221-3/+11
| | | | | | Signed-off-by: Felix Fietkau <nbd@openwrt.org> SVN-Revision: 48753
* ag71xx: store ring size order instead of ring size to avoid div/modFelix Fietkau2016-02-111-34/+43
| | | | | | Signed-off-by: Felix Fietkau <nbd@openwrt.org> SVN-Revision: 48693
* ag71xx: optimize icache footprint of the receive poll functionFelix Fietkau2016-02-111-3/+10
| | | | | | Signed-off-by: Felix Fietkau <nbd@openwrt.org> SVN-Revision: 48691
* ar71xx: use page fragment API in the ethernet driverFelix Fietkau2016-01-311-10/+24
| | | | | | Signed-off-by: Felix Fietkau <nbd@openwrt.org> SVN-Revision: 48578
* ar71xx: fix ethernet MAC reset on DMA hangFelix Fietkau2016-01-131-38/+55
| | | | | | | | | Fully reset the chip like on a full up/down, but without the PHY statemachine restart. Signed-off-by: Felix Fietkau <nbd@openwrt.org> SVN-Revision: 48228
* ar71xx: extend ethernet DMA stuck check to all ar724x (and newer) chipsFelix Fietkau2016-01-131-1/+1
| | | | | | Signed-off-by: Felix Fietkau <nbd@openwrt.org> SVN-Revision: 48227
* ar71xx: clean up ethernet tx queue after reset, wake queues when done - ↵Felix Fietkau2015-12-141-3/+8
| | | | | | | | fixes hangs reported in #18922 Signed-off-by: Felix Fietkau <nbd@openwrt.org> SVN-Revision: 47892
* ar71xx: ag71xx remove IRQF_DISABLEDJohn Crispin2015-07-091-1/+1
| | | | | | | | | | no-op since 2.6.35 removed in Kernel 4.1 see https://lwn.net/Articles/380931/ Signed-off-by: Dirk Neukirchen <dirkneukirchen@web.de> SVN-Revision: 46280
* ar71xx: return limit as number of processed frames when re-scheduling NAPI pollFelix Fietkau2015-06-141-1/+1
| | | | | | Signed-off-by: Felix Fietkau <nbd@openwrt.org> SVN-Revision: 45971
* ar71xx: improve ethernet driver cache footprint by removing an unnecessary ↵Felix Fietkau2014-12-091-22/+25
| | | | | | | | pointer Signed-off-by: Felix Fietkau <nbd@openwrt.org> SVN-Revision: 43588
* ag71xx: replace delay with sleep callsJohn Crispin2014-12-071-4/+4
| | | | | | | | | | | | I don't see that we're in an atomic context so there's no need to busy-wait. Therefore replace the delay with sleep calls. See also Documentation/timers/timers-howto.txt. It states: "In general, use of mdelay is discouraged and code should be refactored to allow for the use of msleep." Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com> SVN-Revision: 43539
* ar71xx: ack completed tx descriptors only after the full frame has been ↵Felix Fietkau2014-09-101-10/+16
| | | | | | | | completed Signed-off-by: Felix Fietkau <nbd@openwrt.org> SVN-Revision: 42457
* ar71xx: re-enable descriptor splitting on ar716x after the previous bugfixes ↵Felix Fietkau2014-09-061-2/+0
| | | | | | | | (#13072) Signed-off-by: Felix Fietkau <nbd@openwrt.org> SVN-Revision: 42429
* ar71xx: fix off-by-one error in packet descriptor splitting (patch from #13072)Felix Fietkau2014-09-061-1/+6
| | | | | | Signed-off-by: Felix Fietkau <nbd@openwrt.org> SVN-Revision: 42428
* ar71xx: disable ethernet descriptor splitting for now, as it seems to cause ↵Felix Fietkau2014-08-301-0/+2
| | | | | | | | tx hangs in some setups Signed-off-by: Felix Fietkau <nbd@openwrt.org> SVN-Revision: 42333
* ar71xx: split packets into multiple descriptors on ar716xFelix Fietkau2014-08-291-25/+92
| | | | | | | | | | | This improves performance when doing concurrent rx/tx on a single ethernet MAC, e.g. when routing between VLANs. Fixes #13072 Signed-off-by: Felix Fietkau <nbd@openwrt.org> SVN-Revision: 42328
* ar71xx: ag71xx: increase calculated max frame length valueGabor Juhos2014-01-111-1/+3
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The r39147 commit introduces a regression: at lease on some routers with ar8216 switch large packets get lost if 802.1q tagged port is used on the interface connected to the aforementioned switch. The r39147 changes code in the way so interface is set to accept packets no longer than max ethernet frame length for a given mtu. Unfortunately ar8216 has a feature: it sends two additional bytes as a packet header and those this header needs to be added to the max frame length. Otherwise long enough packets get lost. The problem only manuifests itself if interface is used in vlan tagged mode. If interface is untagged then ar8216's header fits into space used by 802.1q tag and not packets are lost. Include two additional bytes in the max frame length calculation to fix the issue. This patch is tested and works with Trendnet TEW-632BRP. Signed-off-by Nikolay Martynov <mar.kolya@gmail.com> Patchwork: http://patchwork.openwrt.org/patch/4656/ [juhosg: - simplify the patch to include the additional bytes of the switch header unconditionally, - change subject and update commit message] Signed-off-by: Gabor Juhos <juhosg@openwrt.org> SVN-Revision: 39219
* ar71xx: ag71xx: calculate max frame len register value from the MTUGabor Juhos2013-12-201-3/+15
| | | | | | | | | | | | | | | Set the MAX_FRAME_LEN register to zero in ag71xx_hw_init() and write the correct value into that from the ag71xx_open() and ag71xx_fast_reset() functions. Also recalculate the RX buffer size based on the actual maximum frame length value to optimize memory allocation. Additionaly, disallow to change the MTU value while the interface it running. Signed-off-by: Gabor Juhos <juhosg@openwrt.org> SVN-Revision: 39147
* ar71xx: ag71xx: add ag71xx_max_frame_len() helperGabor Juhos2013-12-201-2/+8
| | | | | | Signed-off-by: Gabor Juhos <juhosg@openwrt.org> SVN-Revision: 39146
* ar71xx: ag71xx: get max_frame_len and desc_pktlen_mask from platform dataGabor Juhos2013-12-201-2/+5
| | | | | | | | This will allow to use SoC specific values for both. Signed-off-by: Gabor Juhos <juhosg@openwrt.org> SVN-Revision: 39145
* ar71xx: ag71xx: store descriptor packet length mask in ag71xx structGabor Juhos2013-12-201-2/+4
| | | | | | | | | | | | | | | The currently used bitmask is not correct for all SoCs. Introduce a new field in struct ag71xx and store the bitmask in that. Use the current value for now, it will be adjusted for each SoCs in further patches. Aslo use the new field directly in the ag71xx_rx_packets and ag71xx_hard_start_xmit() functions and remove the ag71xx_desc_pktlen() helper. Signed-off-by: Gabor Juhos <juhosg@openwrt.org> SVN-Revision: 39144
* ar71xx: ag71xx: compute the RX buffer size from the maximum frame sizeGabor Juhos2013-12-171-1/+1
| | | | | | | | | | | | | | | | Currently, the AG71XX_RX_PKT_SIZE value limits the received frame size to 1514/1516 bytes with/without a VLAN header respectively. However the hardware limit is controlled by the value the AG71XX_REG_MAC_MFL register which contains the value of the max_frame_len field. Compute the RX buffer size from the max_frame_len field to get rid of the 1514/1516 byte limitation. Also remove the unused AG71XX_RX_PKT_SIZE definition. Signed-off-by: Gabor Juhos <juhosg@openwrt.org> SVN-Revision: 39121
* ar71xx: ag71xx: store RX buffer size in the ag71xx structGabor Juhos2013-12-171-5/+7
| | | | | | | | This allows to change the value dynamically. Signed-off-by: Gabor Juhos <juhosg@openwrt.org> SVN-Revision: 39120
* ar71xx: ag71xx: store maximum frame length in the ag71xx structGabor Juhos2013-12-171-2/+6
| | | | | | | | | This will allow to use different values for the different SoCs. Signed-off-by: Gabor Juhos <juhosg@openwrt.org> SVN-Revision: 39117
* ar71xx: ag71xx: fix ag71xx_probe error pathGabor Juhos2013-11-111-1/+3
| | | | | | | Signed-off-by: Catalin Patulea <cat@vv.carleton.ca> Signed-off-by: Gabor Juhos <juhosg@openwrt.org> SVN-Revision: 38716
* ar71xx: ag71xx: fix a race involving netdev registrationGabor Juhos2013-11-081-13/+11
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | In particular, phy_connect before register_netdev. This is because register_netdev runs the netdev notifiers, which can race with the rest of the initialization in ag71xx_probe. In my case this manifested in two ways: 1) If ag71xx is compiled as a module and inserted after netifd has started, netifd is notified by register_netdev before the call to ag71xx_phy_connect. netifd tries to bring the interface up, which calls ag71xx_open, which in turn enters ag71xx_phy_start. This keys off ag->phy_dev (which is still NULL) and thinks this is a fixed-link board, and enters ag71xx_link_adjust. This looks at ag->speed which is not yet initialized and hits the BUG() in the switch (ag->speed) in ag71xx_link_adjust. This is the wrong code path for ag71xx_phy_start - my board has PHYs that need to be brought up with phy_start. Doing ag71xx_phy_connect before register_netdev ensures that ag->phy_dev is non-NULL before ag71xx_phy_start is ever called. 2) When ag71xx is built into the kernel, and netconsole is enabled, there is a gap in the initial burst of replayed printks right after the netdev comes up. My assumption is that netconsole is also triggered by a netdev notifier, and part of this printk burst happens before the call into ag71xx_phy_connect, so part of the burst is lost while the PHY comes up. This patch fixes the gap - all the printks before eth0 comes up are bursted in full when netconsole initializes. ag71xx_phy_connect_xxx no longer runs with a registered netdev, so the logging has been adjusted accordingly to avoid "unregistered net_device" or "eth%d" messages in dmesg. Signed-off-by: Catalin Patulea <cat@vv.carleton.ca> Signed-off-by: Gabor Juhos <juhosg@openwrt.org> SVN-Revision: 38689
* ar71xx: ethernet: cache skb->len in the tx function to avoid accessing it ↵Felix Fietkau2013-08-121-3/+5
| | | | | | | | | | again in completion Improves ethernet performance, especially during bridging Signed-off-by: Felix Fietkau <nbd@openwrt.org> SVN-Revision: 37761
* ar71xx: fix typoFelix Fietkau2013-08-121-1/+1
| | | | | | Signed-off-by: Felix Fietkau <nbd@openwrt.org> SVN-Revision: 37759
* ar71xx: ethernet: skip calls to netdev_completed_queue and netif_wake_queue ↵Felix Fietkau2013-08-121-0/+3
| | | | | | | | if no tx cleanup was done Signed-off-by: Felix Fietkau <nbd@openwrt.org> SVN-Revision: 37758
generated by cgit v1.2.3 (git 2.25.1) at 2024-07-12 23:47:31 +0000