Bug 309235 - X Freezing & Crashing when using UXA accel with OpenGL Rendering
Summary: X Freezing & Crashing when using UXA accel with OpenGL Rendering
Status: RESOLVED UPSTREAM
Alias: None
Product: kwin
Classification: Plasma
Component: general (other bugs)
Version First Reported In: 4.9.0
Platform: Arch Linux Linux
: NOR critical
Target Milestone: ---
Assignee: KWin default assignee
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2012-10-30 05:30 UTC by Axel FIlisola
Modified: 2012-10-30 05:37 UTC (History)
0 users

See Also:
Latest Commit:
Version Fixed In:
Sentry Crash Report:


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Axel FIlisola 2012-10-30 05:30:31 UTC
I actually use chipset i915 with Intel Graphics (Ivy Bridge) (xf86-video-intel-2.20.12-1) configured with UXA (stock) acceleration.
When using OpenGL for desktop effects rendering there are graphical curroptions, such as tearing and while line borders around some dock (avant-window-navigator) icons. 
After some time, it produces GPU hangs (as read in /var/log/erros.log) lasting sometimes seconds, sometimes freezing X completly and in some rare occations, crashing (interrupting the signal to the monitor, and without the options to switch to console mode or do any hot key command at all) leaving as only option to reset.
This problem is not presented when using SNA acceleration, or using XRender for rendering. 

Reproducible: Sometimes

Steps to Reproduce:
1.Set "UXA" as acceleration method and OpenGL for desktop rendering
2. Just work, borwse the web or any other task
3. The screen starts to freeze and erros.log show the "GPU hang" legend
Actual Results:  
The sme, Xfreezes or crash

Expected Results:  
Keep working smoothly

/var/log/errors.log output::
Oct 28 20:48:15 wonderland kernel: [  310.600146] [drm:i915_hangcheck_hung] *ERROR* Hangcheck timer elapsed... GPU hung
Oct 28 20:48:21 wonderland kernel: [  316.746913] [drm:i915_hangcheck_hung] *ERROR* Hangcheck timer elapsed... GPU hung
Oct 28 20:48:27 wonderland kernel: [  322.757235] [drm:i915_hangcheck_hung] *ERROR* Hangcheck timer elapsed... GPU hung
Oct 28 20:48:33 wonderland kernel: [  328.814145] [drm:i915_hangcheck_hung] *ERROR* Hangcheck timer elapsed... GPU hung
Oct 28 20:48:39 wonderland kernel: [  334.867729] [drm:i915_hangcheck_hung] *ERROR* Hangcheck timer elapsed... GPU hung
Oct 28 20:48:45 wonderland kernel: [  341.124319] [drm:i915_hangcheck_hung] *ERROR* Hangcheck timer elapsed... GPU hung
Oct 28 20:48:51 wonderland kernel: [  347.174575] [drm:i915_hangcheck_hung] *ERROR* Hangcheck timer elapsed... GPU hung
Oct 28 20:49:24 wonderland kernel: [  379.632300] [drm:i915_hangcheck_hung] *ERROR* Hangcheck timer elapsed... GPU hung
Oct 28 20:49:30 wonderland kernel: [  385.685883] [drm:i915_hangcheck_hung] *ERROR* Hangcheck timer elapsed... GPU hung
Oct 28 20:49:36 wonderland kernel: [  391.925834] [drm:i915_hangcheck_hung] *ERROR* Hangcheck timer elapsed... GPU hung
Oct 28 20:49:42 wonderland kernel: [  397.949466] [drm:i915_hangcheck_hung] *ERROR* Hangcheck timer elapsed... GPU hung
Oct 28 20:49:48 wonderland kernel: [  404.069608] [drm:i915_hangcheck_hung] *ERROR* Hangcheck timer elapsed... GPU hung
Oct 28 20:49:54 wonderland kernel: [  410.186423] [drm:i915_hangcheck_hung] *ERROR* Hangcheck timer elapsed... GPU hung
Comment 1 Martin Flöser 2012-10-30 05:37:01 UTC
I'm sorry but this is a driver issue. From the client side it should not be possible to hung the GPU. There are overall a few reports with IvyBridge, you could try downgrading to Mesa 8.0 again in case you run Mesa 9.0.