Bug 91651

Created attachment 117714 [details]
/sys/class/drm/card0/error

I'm watching a video in VLC (which uses hw accelerated video decoding), and quite often the computer gets stuck/unresponsive when I use chrome at the same time.

Software versions (all archlinux packages from the standard repository, architecture is AMD64):

  - vlc 2.2.1-6 
  - libva 1.6.0-1
  - libva-intel-driver 1.6.0-1
  - chromium 44.0.2403.155-1
  - linux 4.1.4-1


VLC Log, where it says to use libva:

[00007f6e4e26db68] core decoder warning: early picture skipped
libva info: VA-API version 0.38.0
libva info: va_getDriverName() returns 0
libva info: Trying to open /usr/lib/dri/i965_drv_video.so
libva info: Found init function __vaDriverInit_0_38
libva info: va_openDriver() returns 0
[00007f6e4e26db68] avcodec decoder: Using Intel i965 driver for Intel(R) Ironlake Mobile - 1.6.0 for hardware decoding.

[00007f5f8c078958] core video output warning: picture is too late to be displayed (missing 4053 ms)
[00007f5f8c078958] core video output warning: picture is too late to be displayed (missing 4035 ms)
(... many of them, also audio messages ...)
[00007f5f62566518] avcodec decoder error: more than 5 seconds of late video -> dropping frame (computer too slow ?)
[00007f5f62566518] avcodec decoder error: more than 5 seconds of late video -> dropping frame (computer too slow ?)
[00007f5f62566518] avcodec decoder error: more than 5 seconds of late video -> dropping frame (computer too slow ?)
(... many of them ...)

<--- here, I kill VLC, everything goes back to normal

dmesg, where it tells me to file a bug:

[118556.083623] [drm] stuck on render ring
[118556.088505] [drm] GPU HANG: ecode 5:0:0x9ffdffff, in chromium [21672], reason: Ring hung, action: reset
[118556.088508] [drm] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace.
[118556.088509] [drm] Please file a _new_ bug report on bugs.freedesktop.org against DRI -> DRM/Intel
[118556.088510] [drm] drm/i915 developers can then reassign to the right component if it's not a kernel issue.
[118556.088511] [drm] The gpu crash dump is required to analyze gpu hangs, so please always attach it.
[118556.088512] [drm] GPU crash dump saved to /sys/class/drm/card0/error
[118556.092392] drm/i915: Resetting chip after gpu hang
[118562.106816] [drm] stuck on render ring
[118562.112249] [drm] GPU HANG: ecode 5:0:0x9ffdffff, in chromium [21672], reason: Ring hung, action: reset
[118562.112359] [drm:i915_set_reset_status [i915]] *ERROR* gpu hanging too fast, banning!
[118562.116673] drm/i915: Resetting chip after gpu hang
[118568.090019] [drm] stuck on render ring
[118568.096421] [drm] GPU HANG: ecode 5:0:0x9ffdffff, in chromium [23036], reason: Ring hung, action: reset
[118568.100501] drm/i915: Resetting chip after gpu hang
[118574.113055] [drm] stuck on render ring
[118574.118765] [drm] GPU HANG: ecode 5:0:0x9ffdffff, in chromium [23036], reason: Ring hung, action: reset
[118574.119423] [drm:i915_set_reset_status [i915]] *ERROR* gpu hanging too fast, banning!
[118574.123604] drm/i915: Resetting chip after gpu hang
[118580.096255] [drm] stuck on render ring
[118580.101743] [drm] GPU HANG: ecode 5:0:0x9ffdffff, in chromium [23094], reason: Ring hung, action: reset
[118580.112790] drm/i915: Resetting chip after gpu hang
[118586.119404] [drm] stuck on render ring
[118586.124394] [drm] GPU HANG: ecode 5:0:0x9ffdffff, in chromium [23094], reason: Ring hung, action: reset
[118586.124504] [drm:i915_set_reset_status [i915]] *ERROR* gpu hanging too fast, banning!
[118586.128137] drm/i915: Resetting chip after gpu hang

Video that was playing on VLC at that time (probably relevant to know the hardware decoder chosen):
  ffprobe:
    Stream #0:0(und): Video: h264 (Main) (avc1 / 0x31637661), yuv420p, 854x478
        [SAR 1:1 DAR 427:239], 760 kb/s, 29.97 fps, 29.97 tbr, 90k tbn, 59.94 tbc (default)

  http://youtu.be/-YMcqNc4nSY  (<-- you should definitely watch and listen to this concert!)