Bug 104211 - [hsw] GPU HANG in a qt app
Summary: [hsw] GPU HANG in a qt app
Status: RESOLVED WONTFIX
Alias: None
Product: Mesa
Classification: Unclassified
Component: Drivers/DRI/i965 (show other bugs)
Version: unspecified
Hardware: x86-64 (AMD64) Linux (All)
: medium normal
Assignee: Intel 3D Bugs Mailing List
QA Contact: Intel 3D Bugs Mailing List
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2017-12-11 11:53 UTC by Andrzej Szombierski
Modified: 2017-12-14 20:37 UTC (History)
1 user (show)

See Also:
i915 platform:
i915 features:


Attachments
data from /sys/class/drm/card0/error (3.09 MB, text/plain)
2017-12-11 11:53 UTC, Andrzej Szombierski
Details

Description Andrzej Szombierski 2017-12-11 11:53:27 UTC
Created attachment 136079 [details]
data from /sys/class/drm/card0/error

I was testing a qt-based app. The UI froze for several seconds, then resumed without an issue. In dmesg, I've found the following:

[958940.926878] [drm] stuck on render ring
[958940.927417] [drm] GPU HANG: ecode 7:0:0x87dd7d10, in app [17354], reason: Ring hung, action: reset
[958940.927418] [drm] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace.
[958940.927419] [drm] Please file a _new_ bug report on bugs.freedesktop.org against DRI -> DRM/Intel
[958940.927419] [drm] drm/i915 developers can then reassign to the right component if it's not a kernel issue.
[958940.927420] [drm] The gpu crash dump is required to analyze gpu hangs, so please always attach it.
[958940.927420] [drm] GPU crash dump saved to /sys/class/drm/card0/error
[958940.965440] drm/i915: Resetting chip after gpu hang

Distribution: Ubuntu 16.04.3 LTS
Kernel: 4.4.0-101-generic #124-Ubuntu SMP x86_64
X.Org X Server 1.18.4
Some version information from glxinfo:
    Vendor: Intel Open Source Technology Center (0x8086)
    Device: Mesa DRI Intel(R) Haswell Desktop  (0x412)
    Version: 17.0.7
GPU: Integrated into CPU, Intel(R) Core(TM) i3-4330 CPU
Comment 1 Elizabeth 2017-12-13 22:24:02 UTC
Hello Andrzej,
Any special reason to stay with 16.04? Do you have any steps to trigger the hang again? If possible could you try with a newer kernel version, 4.13 or up? Could you please try Xorg 1.19 and Mesa 17.3?
Thank you.
Comment 2 Andrzej Szombierski 2017-12-14 09:03:07 UTC
It's a work computer, I'm staying with 16.04 because "it works" and updating usually introduces unnecessary problems.

Anyway, this bug has occurred only once. I doubt that I would have any luck attempting to reproduce it. 

I have one guess - could this "Ring hung" be simply a result of the GPU having a large workload spike? I haven't verified this, but the app that I was testing could have tried to render an excessive amount of geometry during one frame.


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.