Pixeleration, stripes, blackscreen at nyan_blaze when using external monitor |
||||||||||
Issue descriptionCustomer started to experience issues after upgrading to 57 Reproduced partly in QD5: https://drive.google.com/open?id=0B-g52zibXA02Tm1NUVJaYm9uZFU Google Chrome Version 57.0.2987.146 Platform Version 9202.64.0 (Official Build) stable-channel nyan_blaze What steps will reproduce the problem? (1)Connect external monitor via HDMI to nyan_blaze (2)open any youtube video (3)close the lid, iterate several times What is the expected result? device goes to docker mode, gui is available at external monitor What happens instead? stripes at external monitor (4)start suspend_stress_test -c 1000 (5)open and close lid several times What is the expected result? devices sleeps and wakes up when test completes, showing video on both internal and external monitor What happens instead? stripes at external monitor when lid closed, black screen on internal monitor and normal video on external monitor when lid is opened. In addition to above, customer experiencing image pixeleration (e.g. at gmail). https://drive.google.com/open?id=0B-g52zibXA02VHE4WVVQRENZOEk In customer log https://drive.google.com/a/google.com/file/d/0B-g52zibXA02ZU1CSXR0LWhPZ1E/view?usp=sharing I see 2017-05-04T08:14:55.230708+02:00 ERR tcsd[780]: TSS: Socket send connection error: Broken pipe. In my logs https://drive.google.com/a/google.com/file/d/0B-g52zibXA02TWIyU01tbk9HN2c/view?usp=sharing while repro in QD5: 2017-05-05T22:44:38.418124+00:00 WARNING kernel: [ 760.617772] ------------[ cut here ]------------ 2017-05-05T22:44:38.418156+00:00 WARNING kernel: [ 760.617804] WARNING: at /mnt/host/source/src/third_party/kernel/v3.10/drivers/gpu/drm/tegra/dpaux.c:232 tegra_dpaux_transfer+0x1ac/0x41c() 2017-05-05T22:44:38.418164+00:00 NOTICE kernel: [ 760.617814] wait dpaux transfer complete timeout. 2017-05-05T22:44:38.418170+00:00 NOTICE kernel: [ 760.617825] Modules linked in: ip6t_REJECT uinput snd_soc_tegra30_i2s snd_soc_tegra_pcm snd_hda_codec_hdmi snd_soc_tegra_max98090 snd_soc_tegra_utils snd_soc_tegra30_ahub snd_hda_tegra snd_hda_controller snd_hda_codec snd_soc_max98090 i2c_dev rfcomm fuse zram(C) nf_conntrack_ipv6 nf_defrag_ipv6 ip6table_filter ip6_tables snd_seq_midi snd_seq_midi_event snd_rawmidi snd_seq snd_seq_device mwifiex_sdio mwifiex btmrvl_sdio cfg80211 btmrvl bluetooth uvcvideo videobuf2_vmalloc videobuf2_memops videobuf2_core joydev 2017-05-05T22:44:38.418176+00:00 NOTICE kernel: [ 760.618026] CPU: 2 PID: 1281 Comm: DrmThread Tainted: G WC 3.10.18 #1 2017-05-05T22:44:38.418181+00:00 NOTICE kernel: [ 760.618050] [<c020cf9c>] (unwind_backtrace+0x0/0x110) from [<c020a08c>] (show_stack+0x20/0x24) 2017-05-05T22:44:38.418186+00:00 NOTICE kernel: [ 760.618065] [<c020a08c>] (show_stack+0x20/0x24) from [<c074e510>] (dump_stack+0x20/0x28) 2017-05-05T22:44:38.418191+00:00 NOTICE kernel: [ 760.618082] [<c074e510>] (dump_stack+0x20/0x28) from [<c0225960>] (warn_slowpath_common+0x5c/0x7c) 2017-05-05T22:44:38.418197+00:00 NOTICE kernel: [ 760.618095] [<c0225960>] (warn_slowpath_common+0x5c/0x7c) from [<c02259cc>] (warn_slowpath_fmt+0x4c/0x6c) 2017-05-05T22:44:38.418203+00:00 NOTICE kernel: [ 760.618110] [<c02259cc>] (warn_slowpath_fmt+0x4c/0x6c) from [<c04f35a4>] (tegra_dpaux_transfer+0x1ac/0x41c) 2017-05-05T22:44:38.418209+00:00 NOTICE kernel: [ 760.618124] [<c04f35a4>] (tegra_dpaux_transfer+0x1ac/0x41c) from [<c04c7c94>] (drm_dp_i2c_xfer+0xb8/0x238) 2017-05-05T22:44:38.418214+00:00 NOTICE kernel: [ 760.618137] [<c04c7c94>] (drm_dp_i2c_xfer+0xb8/0x238) from [<c059a268>] (__i2c_transfer+0x54/0x84) 2017-05-05T22:44:38.418220+00:00 NOTICE kernel: [ 760.618149] [<c059a268>] (__i2c_transfer+0x54/0x84) from [<c059ad68>] (i2c_transfer+0x7c/0xa0) 2017-05-05T22:44:38.418251+00:00 NOTICE kernel: [ 760.618163] [<c059ad68>] (i2c_transfer+0x7c/0xa0) from [<c04dc7a0>] (drm_do_probe_ddc_edid+0xbc/0x138) 2017-05-05T22:44:38.418256+00:00 NOTICE kernel: [ 760.618174] [<c04dc7a0>] (drm_do_probe_ddc_edid+0xbc/0x138) from [<c04dc860>] (drm_probe_ddc+0x44/0x7c) 2017-05-05T22:44:38.418260+00:00 NOTICE kernel: [ 760.618186] [<c04dc860>] (drm_probe_ddc+0x44/0x7c) from [<c04dd4a8>] (drm_get_edid+0x28/0x26c) 2017-05-05T22:44:38.418263+00:00 NOTICE kernel: [ 760.618199] [<c04dd4a8>] (drm_get_edid+0x28/0x26c) from [<c04eb0f4>] (tegra_connector_get_modes+0x54/0xbc) 2017-05-05T22:44:38.418269+00:00 NOTICE kernel: [ 760.618212] [<c04eb0f4>] (tegra_connector_get_modes+0x54/0xbc) from [<c04c7230>] (drm_helper_probe_single_connector_modes+0x138/0x310) 2017-05-05T22:44:38.418275+00:00 NOTICE kernel: [ 760.618223] [<c04c7230>] (drm_helper_probe_single_connector_modes+0x138/0x310) from [<c04d8448>] (drm_mode_getconnector+0xe8/0x398) 2017-05-05T22:44:38.418280+00:00 NOTICE kernel: [ 760.618235] [<c04d8448>] (drm_mode_getconnector+0xe8/0x398) from [<c04cbce8>] (drm_ioctl+0x348/0x4d8) 2017-05-05T22:44:38.418284+00:00 NOTICE kernel: [ 760.618248] [<c04cbce8>] (drm_ioctl+0x348/0x4d8) from [<c031288c>] (do_vfs_ioctl+0x4a8/0x5bc) 2017-05-05T22:44:38.418287+00:00 NOTICE kernel: [ 760.618260] [<c031288c>] (do_vfs_ioctl+0x4a8/0x5bc) from [<c0312a10>] (SyS_ioctl+0x70/0xa8) 2017-05-05T22:44:38.418290+00:00 NOTICE kernel: [ 760.618273] [<c0312a10>] (SyS_ioctl+0x70/0xa8) from [<c020618c>] (__sys_trace_return+0x0/0x14) 2017-05-05T22:44:38.418293+00:00 WARNING kernel: [ 760.618284] ---[ end trace afa0f0d706e1e779 ]--- 2017-05-05T22:44:38.668031+00:00 WARNING kernel: [ 760.866752] ------------[ cut here ]------------ and [1266:1281:0505/160125.246001:VERBOSE1:drm_display.cc(102)] DRM configuring: device=/sys/devices/platform/tegra-drm/drm/card1 crtc=5 connector=12 origin=0,0 size=0x0 [1266:1281:0505/160125.263841:VERBOSE1:drm_display.cc(102)] DRM configuring: device=/sys/devices/platform/tegra-drm/drm/card1 crtc=9 connector=14 origin=0,0 size=1920x1200 [1266:1281:0505/160125.412811:VERBOSE1:drm_util.cc(204)] Unable to get cursor width capability: Invalid argument [1266:1281:0505/160125.583501:ERROR:drm_display.cc(179)] Failed to set color correction for display: crtc_id = 5 [1266:1281:0505/160125.583721:ERROR:drm_display.cc(179)] Failed to set color correction for display: crtc_id = 9
,
May 6 2017
+Stephane have you come across anything quite like this? The pixelation looks strange, like we are rescaling multiple times somehow, not sure if we use any hardware scaling mechanisms on these Tegra systems? The stretching looks like we have a stride mismatch somewhere, or the monitor otherwise out of sync...
,
May 6 2017
I have never seen it, no. Kristian, do you have cycles to look?
,
May 17 2017
Customer asked for updates, disabling GPU helped for a while but issue came back with GPU disabled
,
May 17 2017
Ping Kristian? Was that the wrong email?
,
May 17 2017
I usually use hoegsberg@google.com. We don't have a nyan here in PDX, but Igor is shipping one. I'll take a look when it gets here.
,
May 31 2017
This looks like it's fixed in R59. I can reproduce with R58, but R59 and R60 look fine.
,
May 31 2017
A few notes for now: minigbm commit 6b41fb55deae7d8ae00c33f7ce849d6fd6525d77 fixes the stripes. The stripes look like a tiling mismatch and usually show up briefly when closing the lid. I'm not convinced the commit is right. I suspect that maybe it just broke tiling support and now all buffers are linear - which means that the stripe artifacts go away, but also that we never use tiling (bad for perf/power). There's another issue here that I haven't bisected yet, which is that when you close the lid, sometimes the external monitor freezes. Mouse cursor doesn't move and system is unresponsive until lid open. That issue is still present in R60. This issue combines with the stripey artifact to leave the external monitor looking like the second picture permanently. I haven't been able to reproduce anything that looks like the pixelation in the first photo.
,
Jun 2 2017
Fix for the freezing here: https://chromium-review.googlesource.com/c/522885/
,
Jun 7 2017
Which releases do we want this backported to?
,
Jun 8 2017
The following revision refers to this bug: https://chromium.googlesource.com/chromiumos/third_party/kernel/+/c820bbfff2d39e0a6f902f3dc3ae1b6cb4bf5368 commit c820bbfff2d39e0a6f902f3dc3ae1b6cb4bf5368 Author: Kristian H. Kristensen <hoegsberg@google.com> Date: Wed Jun 07 23:59:04 2017 CHROMIUM: drm/tegra: Send pageflip event also when disabling crtc We always need to send a pageflip event back in response to a pageflip ioctl, even if the crtc gets disabled and the pageflip never happen. Userspace relies on the pageflip event to drive it's compositing loop and if we drop events the compositor gets stuck. BUG= chromium:719097 TEST=Connect external monitor. Run animation to keep the compositor pageflipping, repeatedly close and open lid until external monitor freezes. Change-Id: I88da347f67a2ac504f2d27e7ef7b7ab440d756f6 Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com> Reviewed-on: https://chromium-review.googlesource.com/522885 Commit-Ready: Kristian H. Kristensen <hoegsberg@chromium.org> Tested-by: Kristian H. Kristensen <hoegsberg@chromium.org> Reviewed-by: Stéphane Marchesin <marcheu@chromium.org> [modify] https://crrev.com/c820bbfff2d39e0a6f902f3dc3ae1b6cb4bf5368/drivers/gpu/drm/tegra/dc.c
,
Nov 17 2017
,
Nov 20 2017
,
Jan 22 2018
,
Jan 23 2018
|
||||||||||
►
Sign in to add a comment |
||||||||||
Comment 1 by marchuk@chromium.org
, May 5 2017