Bug 17391

Summary: Hangcheck timer elapsed... GPU hung
Product: Drivers Reporter: Eduardo Bacchi Kienetz (eduardo)
Component: Video(DRI - Intel)Assignee: drivers_video-dri-intel (drivers_video-dri-intel)
Status: RESOLVED INVALID    
Severity: high CC: chris
Priority: P1    
Hardware: All   
OS: Linux   
Kernel Version: 2.6.35.4 Subsystem:
Regression: No Bisected commit-id:
Attachments: dmesg output with GPU hung messages

Description Eduardo Bacchi Kienetz 2010-08-30 02:32:07 UTC 
Created attachment 28391 [details]
dmesg output with GPU hung messages

Following my report on https://bugzilla.kernel.org/show_bug.cgi?id=15671, the bug still exists in 2.6.35.4 but is much less frequent, which could be due to my kernel configuration (selecting some kernels options would increase the occurrence of this bug).

I see that in this 2.6.35.4 release an extra debug message seems to have been added "[drm:init_ring_common] *ERROR* render ring head not reset to zero ctl 00000000 head 02001000 tail 00000000 start 02001000
[drm:init_ring_common] *ERROR* render ring head forced to zero ctl 00000000 head 00000000 tail 00000000 start 02001000".

Please check the attached dmesg output for the full output.
After that I need to go to a console and "reboot", cause the screen stops updating on X.
Comment 1 Chris Wilson 2010-09-04 21:03:33 UTC 
A bug in the userspace driver. Can you please attach Xorg.log and /sys/kernel/debug/dri/0/i915_error_state? [Might be best moved to bugs.freedesktop.org as well where it is more likely to get the attention of the appropriate developers.]
Comment 2 Chris Wilson 2010-12-16 15:41:53 UTC 
The render ring reset error message has been resolved. The GPU hang is due to a userspace driver bug, for which an i915_error_state is required to help debugging.