Bug 51692

Summary:	System crash while playing concurrently VC1 and H264 videos
Product:	libva	Reporter:	Josep Torra <n770galaxy>
Component:	intel	Assignee:	haihao <haihao.xiang>
Status:	RESOLVED WORKSFORME	QA Contact:
Severity:	critical
Priority:	medium	CC:	jjardon, seanvk
Version:	unspecified
Hardware:	Other
OS:	All
Whiteboard:
i915 platform:		i915 features:

Description Josep Torra 2012-07-03 07:54:00 UTC

In system Huron River running updated Ubuntu 12.04 (32 bits) and git master for:

- libva
- intel-vaapi
- gstreamer-vaapi

Test conditions:

- Playback of multiple mpeg2, h264 and vc1 files sequentially is correct.
- Concurrent playback[1]| of mpeg2 and h264 files is fine
- Concurrent playback[1] of mpeg2 and vc1 files is ok

Failing test
- Concurrent playback[1] of vc1 and h264 clips crashes the system

[1] Concurrent playback understood as running simultaneous two gst-launch-0.10 like in the following example.

term1$ gst-launch-0.10 filesrc location=big_buck_bunny_720p_h264.mov !
qtdemux ! vaapidecode ! queue ! vaapisink

term2$ gst-launch-0.10 filesrc location=The_Magic_of_Flight_720.wmv !
asfdemux ! vaapidecode ! queue ! vaapisink

The crashes are reproducible with gstreamer-vaapi and Fluendo's VA decoder.

I've tried with multiple media from different sources and I'm always getting the crashing scenario. "The Magic of Flight" was part of Microsoft "WMV HD Show Case"

As requested the following are example media files that reproduces the issue.

H264
http://blender-mirror.kino3d.org/peach/bigbuckbunny_movies/big_buck_bunny_720p_h264.mov

VC1
http://download.microsoft.com/download/9/b/4/9b4535e2-f908-4785-97a5-ef360405e391/Elephant%27s%20Dream%20720p%202M.wmv

The VC1 clip provided doesn't seems to decode properly with gstreamer-vaapi.
With the Fluendo's VA decoder it decodes somehow much better but still with errors.

Same VC1 clip with the same Fluendo's VA decoder is properly decoded in the CedarTrail netbook running meego/libva 1.0.15-1.1/Intel GMA500-CDV-5.3.0.32L.0039.

So the VC1 decoding errors are another issue.

Comment 1 Christophe Oosterlynck 2012-07-10 07:29:22 UTC

I'm experiencing the same problem on a Huron River system with Debian Wheezy x86. All the scenarios described by Josep Torra also run fine here except when playing a vc1 clip simultaneously with a h264 clip.

Although in my case, my system doesn't crash. It just hangs for a while. I start playing a long vc1 clip: it runs smooth. Meanwhile I start playing a short h264 clip: system hangs (but mouse still reacts) + video is mostly green and doesn't advance. After a while (longer than the length of the short h264 clip though), the h264 clip stops playing and the vc1 clip catches up and plays smoothly again.

During the hang, dmesg reports hangs in the GPU. Relevant output is attached at the end of this comment. Trying to get debug info from i915_error_state results in no info:

# mount -t debugfs debugfs /sys/kernel/debug
# cat /sys/kernel/debug/dri/0/i915_error_state
cat: /sys/kernel/debug/dri/0/i915_error_state: Cannot allocate memory

dmesg output:
[ 5246.370219] [drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed... GPU hung
[ 5246.370241] [drm] capturing error event; look for more information in /debug/dri/0/i915_error_state
[ 5246.376918] [drm:i915_wait_request] *ERROR* i915_wait_request returns -11 (awaiting 3630 at 3623, next 3708)
[ 5252.716957] [drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed... GPU hung
[ 5252.717109] [drm:i915_wait_request] *ERROR* i915_wait_request returns -11 (awaiting 3813 at 3758, next 3814)
[ 5278.659069] [drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed... GPU hung
[ 5278.659129] [drm:i915_wait_request] *ERROR* i915_wait_request returns -11 (awaiting 7827 at 7821, next 7828)
[ 5285.017843] [drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed... GPU hung
[ 5285.017897] [drm:i915_wait_request] *ERROR* i915_wait_request returns -11 (awaiting 7950 at 7946, next 7951)
[ 5292.566763] [drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed... GPU hung
[ 5292.566817] [drm:i915_wait_request] *ERROR* i915_wait_request returns -11 (awaiting 8031 at 7960, next 8032)
[ 5298.969450] [drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed... GPU hung
[ 5298.969500] [drm:i915_wait_request] *ERROR* i915_wait_request returns -11 (awaiting 8107 at 8082, next 8108)
[ 5306.622228] [drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed... GPU hung
[ 5306.622276] [drm:i915_wait_request] *ERROR* i915_wait_request returns -11 (awaiting 8178 at 8113, next 8179)
[ 5313.136808] [drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed... GPU hung
[ 5313.136867] [drm:i915_wait_request] *ERROR* i915_wait_request returns -11 (awaiting 8290 at 8263, next 8291)

Comment 2 ykzhao 2013-11-29 02:33:37 UTC

Will you please try the latest intel-vaapi driver and see whether the issue still exists?

Thanks.
     Yakui

Comment 3 haihao 2015-11-18 08:25:02 UTC

No response from user for a long time so closing as WORKSFORME, Feel free to reopen this bug if you still experience the issue

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.