Bug 86626 - [snb] Stuck on render ring
Summary: [snb] Stuck on render ring
Status: CLOSED INVALID
Alias: None
Product: DRI
Classification: Unclassified
Component: DRM/Intel (show other bugs)
Version: unspecified
Hardware: Other All
: medium normal
Assignee: Intel GFX Bugs mailing list
QA Contact: Intel GFX Bugs mailing list
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2014-11-23 13:22 UTC by Cedric Sodhi
Modified: 2017-07-03 10:59 UTC (History)
1 user (show)

See Also:
i915 platform: SNB
i915 features: GPU hang


Attachments
dmesg after artefacts appear (59.58 KB, text/plain)
2014-11-23 13:23 UTC, Cedric Sodhi
no flags Details
device error dump after artefacts appear (304.09 KB, application/x-gzip)
2014-11-23 13:27 UTC, Cedric Sodhi
no flags Details
dmesg with 3.18 (63.00 KB, text/plain)
2014-12-06 19:48 UTC, Cedric Sodhi
no flags Details
batch dump with 3.18 (366.08 KB, application/x-gzip)
2014-12-06 19:50 UTC, Cedric Sodhi
no flags Details
X.org.log wtih 3.18 (28.28 KB, text/plain)
2014-12-06 19:51 UTC, Cedric Sodhi
no flags Details
Detect GPU page faults (1.80 KB, patch)
2014-12-09 08:15 UTC, Chris Wilson
no flags Details | Splinter Review

Description Cedric Sodhi 2014-11-23 13:22:19 UTC
On Linux slate 3.16.5-gentoo #6 SMP Sat Nov 22 11:43:01 CET 2014 x86_64 Intel(R) Pentium(R) CPU 997 @ 1.60GHz GenuineIntel GNU/Linux if frequently happens under X.org running (presumably triggered by firefox-bin) that first, large portions of the screen render black, a drop in performance, quickly followed by a deadlock (no SysReq will help). When the first signs of the issue occur, I manage to drop into TTY and kill firefox which usually stabilizes the situation. I can then grab dmesg and the error log, which you find attached. Clearly, I can't assert that they contain everything related to the freeze, since I can only obtain them if the actual freeze does not occur.

I will try to reproduce the issue with current 3.18 mainline but can not promise that I'll again be as lucky as being able to grab the logs in the narrow window before the actual freeze.
Comment 1 Cedric Sodhi 2014-11-23 13:23:38 UTC
Created attachment 109896 [details]
dmesg after artefacts appear
Comment 2 Cedric Sodhi 2014-11-23 13:27:19 UTC
Created attachment 109897 [details]
device error dump after artefacts appear
Comment 3 Daniel Vetter 2014-11-23 14:59:40 UTC
Seems to have died after an SNA batch I think. Reassigning.

Please retest with latest ddx release and attach Xorg.log.
Comment 4 Chris Wilson 2014-11-23 16:49:31 UTC
The error is not the batch, but the execution.
Comment 5 Cedric Sodhi 2014-11-25 20:17:00 UTC
I think, though without certainty, the issue is fixed in 3.18 or greater. I've been running 3.18 for a while and did not experience any problems or signs thereof.
Comment 6 Daniel Vetter 2014-11-26 08:27:04 UTC
Thanks for reporting back, tentatively closing this one.
Comment 7 Cedric Sodhi 2014-12-06 19:47:43 UTC
Unfortunally not fixed. Attaching logs for 3.18.0-rc5
Comment 8 Cedric Sodhi 2014-12-06 19:48:53 UTC
Created attachment 110513 [details]
dmesg with 3.18
Comment 9 Cedric Sodhi 2014-12-06 19:50:00 UTC
Created attachment 110514 [details]
batch dump with 3.18
Comment 10 Cedric Sodhi 2014-12-06 19:51:20 UTC
Created attachment 110515 [details]
X.org.log wtih 3.18
Comment 11 Chris Wilson 2014-12-07 08:38:59 UTC
Same issue, tries to execute a batch ends up somewhere entirely different.
Comment 12 Chris Wilson 2014-12-09 08:15:20 UTC
Created attachment 110605 [details] [review]
Detect GPU page faults

This should hopefully catch the error earlier, and so might reveal more about the issue.
Comment 13 Jani Nikula 2016-06-17 15:21:04 UTC
Timeout, closing. Please reopen if the problem persists with latest kernels.
Comment 14 Jari Tahvanainen 2017-07-03 10:59:34 UTC
Closing >1 year old resolved+invalid.


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.