[Bug 213145] AMDGPU resets, timesout and crashes after "*ERROR* Waiting for fences timed out!"

bugzilla-daemon at kernel.org bugzilla-daemon at kernel.org
Sun Oct 30 16:57:19 UTC 2022


https://bugzilla.kernel.org/show_bug.cgi?id=213145

--- Comment #27 from nvaert1986 (nvaert1986 at hotmail.com) ---
(In reply to rv1sr from comment #24)
> Do you guys by any chance use KWin?
> 
> Had experienced this exact issue on a daily basis (kernel 5.19 + amdgpu),
> especially while running Firefox or Vivaldi.
> 
> After setting the following environment variable in /etc/environment two
> weeks ago, the issue no longer persists.
> 
> KWIN_DRM_NO_DIRECT_SCANOUT=1

I tried this for a couple of days, but after a few days Xorg still crashed
unfortunately. It does seem to be less frequent though.

plasmashell[1280]: amdgpu: amdgpu_cs_query_fence_status failed.
kwin_x11[1255]: amdgpu: amdgpu_cs_query_fence_status failed.
plasmashell[447733]: amdgpu: amdgpu_cs_query_fence_status failed.
plasmashell[447733]: Crash Annotation GraphicsCriticalError: |[0][GFX1-]: GFX:
RenderThread detected a device reset in PostUpdate (t=5437.93) [GFX1-]: GFX:
RenderThrea>
plasmashell[1280]: amdgpu: The CS has been cancelled because the context is
lost.
plasmashell[1280]: amdgpu: amdgpu_cs_query_fence_status failed.
plasmashell[1280]: amdgpu: The CS has been cancelled because the context is
lost.
plasmashell[447733]: ATTENTION: default value of option mesa_glthread
overridden by environment.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.


More information about the dri-devel mailing list