FS#41678 - [intel-dri] Chrome gives occasional gpu hang with "ring hung" for drm with Intel graphics

Attached to Project: Arch Linux
Opened by Mike Cloaked (mcloaked) - Sunday, 24 August 2014, 18:07 GMT
Last edited by Andreas Radke (AndyRTR) - Saturday, 11 October 2014, 18:12 GMT
Task Type Bug Report
Category Packages: Extra
Status Closed
Assigned To Jan de Groot (JGC)
Andreas Radke (AndyRTR)
Laurent Carlier (lordheavy)
Architecture All
Severity Low
Priority Normal
Reported Version
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 2
Private No

Details

Description:
Running chrome in kde occasionally chrome hangs after updates the past day or two, and no further mouse input can be made. Can killall chrome to recover. Running google-chrome 36.0.1985.143-1 in kde version 4.14.0-1 with latest kernel linux 3.16.1-1

The journal contains the following lines:
Aug 24 10:01:33 home1 kernel: [drm] stuck on render ring
Aug 24 10:01:33 home1 kernel: [drm] GPU HANG: ecode 0:0xf5f7fffe, in chrome [11355], reason: Ring
hung, action: reset
Aug 24 10:01:33 home1 kernel: [drm] GPU hangs can indicate a bug anywhere in the entire gfx stack
, including userspace.
Aug 24 10:01:33 home1 kernel: [drm] Please file a _new_ bug report on bugs.freedesktop.org agains
t DRI -> DRM/Intel
Aug 24 10:01:33 home1 kernel: [drm] drm/i915 developers can then reassign to the right component
if it's not a kernel issue.
Aug 24 10:01:33 home1 kernel: [drm] The gpu crash dump is required to analyze gpu hangs, so pleas
e always attach it.
Aug 24 10:01:33 home1 kernel: [drm] GPU crash dump saved to /sys/class/drm/card0/error
Aug 24 10:01:35 home1 kernel: [drm] Enabling RC6 states: RC6 on, RC6p on, RC6pp off


Additional info:
* package version(s)
Running google-chrome 36.0.1985.143-1 in kde version 4.14.0-1 with latest kernel linux 3.16.1-1

xf86-video-intel 2.99.914-4
kdebase-workspace 4.11.11-2
mesa 10.2.6-1

libdrm 2.4.56-1

* config and/or log files etc.
Attaching /sys/class/drm/card0/error as error.log

Attaching journal log file shortly chrome-hang-bug-1.log


Steps to reproduce:
Login to KDE, start using chrome. Chrome locks up at seemingly random times with a GPU hang.
   chrome-hang-bug-1.log (159.9 KiB)
This task depends upon

Closed by  Andreas Radke (AndyRTR)
Saturday, 11 October 2014, 18:12 GMT
Reason for closing:  Fixed
Comment by Mike Cloaked (mcloaked) - Sunday, 24 August 2014, 18:11 GMT
The error log from /sys/class/drm/card0/error was larger than 2MiB so I am attaching it as a compressed gzipped file instead. The previous attachment was the journal log file.
(application/x-gzip)    error1.log.gz (336.4 KiB)
Comment by Mike Cloaked (mcloaked) - Sunday, 24 August 2014, 18:19 GMT
Graphics card is 00:02.0 VGA compatible controller: Intel Corporation Xeon E3-1200 v2/3rd Gen Core processor Graphics Controller (rev 09)
Comment by Mike Cloaked (mcloaked) - Sunday, 24 August 2014, 19:32 GMT
Reported upstream at https://bugzilla.kernel.org/show_bug.cgi?id=83141
Comment by Doug Newgard (Scimmia) - Monday, 25 August 2014, 04:13 GMT
Does this happen with Chromium? Chrome is in the AUR, so it's unsupported.
Comment by Mike Cloaked (mcloaked) - Monday, 25 August 2014, 09:01 GMT
I don't know if it happens with chromium. This may be specific to particular web sites that trigger the graphics issue when they are loaded. However I did not think to record which url I was on when the problem happened, which is about 4 times in a couple of days with many hours of web usage. I will report back again if I can add more information. However although chrome triggered the problem there is presumably a bug in the graphics code, and hopefully the journal and error log will be able to point to where the underlying buggy code lies?

These two upstream bugs might be related: https://bugs.freedesktop.org/show_bug.cgi?id=81539 and https://bugs.freedesktop.org/show_bug.cgi?id=77104
Comment by Laurent Carlier (lordheavy) - Monday, 08 September 2014, 12:36 GMT
please test with xf86-video-intel-2.99.916-1 in testing
Comment by Mike Cloaked (mcloaked) - Monday, 08 September 2014, 14:22 GMT
OK I will test when my local mirror syncs and report any hangs with some logs.
Comment by Mike Cloaked (mcloaked) - Tuesday, 09 September 2014, 09:28 GMT
I have been running my system with:

xf86-video-intel (2.99.914-4 -> 2.99.916-1)

and also with

mesa (10.2.7-1 -> 10.2.7-2)
mesa-libgl (10.2.7-1 -> 10.2.7-2)
intel-dri (10.2.7-1 -> 10.2.7-2)
linux (3.16.1-1 -> 3.16.2-1)

I have not had any repeat of the hang originally reported. I will continue to test for several days and then report back if no further occurrence.

Loading...