Bug 100103 - [snb] GPU hang in compiz
Summary: [snb] GPU hang in compiz
Status: NEEDINFO
Alias: None
Product: Mesa
Classification: Unclassified
Component: Drivers/DRI/i965 (show other bugs)
Version: unspecified
Hardware: Other All
: medium normal
Assignee: Intel 3D Bugs Mailing List
QA Contact: Intel 3D Bugs Mailing List
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2017-03-07 17:25 UTC by longsi0008
Modified: 2017-05-08 14:57 UTC (History)
2 users (show)

See Also:
i915 platform:
i915 features:


Attachments
error (35.80 KB, text/plain)
2017-03-07 17:25 UTC, longsi0008
Details
/sys/class/drm/card0/error (30.14 KB, text/plain)
2017-04-07 19:38 UTC, longsi0008
Details

Note You need to log in before you can comment on or make changes to this bug.
Description longsi0008 2017-03-07 17:25:19 UTC
Created attachment 130115 [details]
error

I have been bearing this issue for a few months already. My two monitors system would randomly freeze up, mainly forcing a hard reset to resolve, though there are a few times it would recover by itself after freezing for 20 seconds. During a freeze, the display would turn black for most of the time. Sometimes, it would just log me out. It can take place as soon as 30 seconds from logging in.

I tried out LiveCD of (dev) Tails 2.11, Ubuntu 17.04 and I am able to use the system for the whole day without freezing up.

My "uname -a" is below:

Linux home 4.10.0-9-generic #11-Ubuntu SMP Mon Feb 20 13:47:35 UTC 2017 x86_64 x86_64 x86_64 GNU/Linux

and my "xserver-xorg-video-intel" is 2:2.99.917+git20160706-1ubuntu1

Below is found in syslog.
=============
Mar  8 01:06:30 home kernel: [ 9232.716641] [drm] GPU HANG: ecode 6:0:0x85fffffc, in compiz [4152], reason: Hang on render ring, action: reset
Mar  8 01:06:30 home kernel: [ 9232.716643] [drm] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace.
Mar  8 01:06:30 home kernel: [ 9232.716644] [drm] Please file a _new_ bug report on bugs.freedesktop.org against DRI -> DRM/Intel
Mar  8 01:06:30 home kernel: [ 9232.716644] [drm] drm/i915 developers can then reassign to the right component if it's not a kernel issue.
Mar  8 01:06:30 home kernel: [ 9232.716645] [drm] The gpu crash dump is required to analyze gpu hangs, so please always attach it.
Mar  8 01:06:30 home kernel: [ 9232.716645] [drm] GPU crash dump saved to /sys/class/drm/card0/error
Mar  8 01:06:30 home kernel: [ 9232.716684] drm/i915: Resetting chip after gpu hang
=============

I attached the "error" file in this report.
Comment 1 Matt Turner 2017-03-22 17:10:29 UTC
(In reply to longsi0008 from comment #0)
> I tried out LiveCD of (dev) Tails 2.11, Ubuntu 17.04 and I am able to use
> the system for the whole day without freezing up.

That is a good sign -- likely that the bug is fixed in a newer Mesa or kernel.

Can you update your system to the versions on the LiveCD and see if it resolves the problem?

Also, you haven't told us what Mesa version you have. That's very important.
Comment 2 longsi0008 2017-04-07 19:38:09 UTC
Created attachment 130751 [details]
/sys/class/drm/card0/error
Comment 3 longsi0008 2017-04-07 19:40:08 UTC
I am on Ubuntu 17.04, MESA is version 17.0.3-1ubuntu1. Below is my log, and I have attached the error file.

Apr  8 03:28:37 kernel: [  146.883333] [drm] GPU HANG: ecode 6:0:0x85fffffc, in compiz [1799], reason: Hang on render ring, action: reset
Apr  8 03:28:37 kernel: [  146.883334] [drm] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace.
Apr  8 03:28:37 kernel: [  146.883335] [drm] Please file a _new_ bug report on bugs.freedesktop.org against DRI -> DRM/Intel
Apr  8 03:28:37 kernel: [  146.883335] [drm] drm/i915 developers can then reassign to the right component if it's not a kernel issue.
Apr  8 03:28:37 kernel: [  146.883336] [drm] The gpu crash dump is required to analyze gpu hangs, so please always attach it.
Apr  8 03:28:37 kernel: [  146.883336] [drm] GPU crash dump saved to /sys/class/drm/card0/error
Apr  8 03:28:37 kernel: [  146.883374] drm/i915: Resetting chip after gpu hang
Comment 4 longsi0008 2017-05-08 10:49:45 UTC
Anyone looking into this bug? I am on Ubuntu 17.04 and the GPU hangs a few times everyday.
Comment 5 longsi0008 2017-05-08 10:51:39 UTC
May  7 13:27:54 kernel: [ 3160.595697] [drm] GPU HANG: ecode 6:0:0x85fffffc, in compiz [1684], reason: Hang on render ring, action: reset
May  7 13:27:54 kernel: [ 3160.595698] [drm] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace.
May  7 13:27:54 kernel: [ 3160.595698] [drm] Please file a _new_ bug report on bugs.freedesktop.org against DRI -> DRM/Intel
May  7 13:27:54 kernel: [ 3160.595699] [drm] drm/i915 developers can then reassign to the right component if it's not a kernel issue.
May  7 13:27:54 kernel: [ 3160.595699] [drm] The gpu crash dump is required to analyze gpu hangs, so please always attach it.
May  7 13:27:54 kernel: [ 3160.595700] [drm] GPU crash dump saved to /sys/class/drm/card0/error
May  7 13:27:54 kernel: [ 3160.595736] drm/i915: Resetting chip after gpu hang
Comment 6 Mark Janes 2017-05-08 14:57:31 UTC
It looks like your xorg-video-intel version preceeds a fix that Chris Wilson made which resolved many gpu hangs:

4acd4a7d3d2f41227022fa7581cfb85a0b124eae
Author:     Chris Wilson <chris@chris-wilson.co.uk>
sna/gen9: Emit a dummy primitive between VertexElements


I would see if modesetting driver fixes your issue.  Alternatively, you could build a newer sna and see if that helps.


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.