AMDGPU error: "[drm:amdgpu_dm_atomic_commit_tail [amdgpu]] *ERROR* Waiting for fences timed out!"

[Michel Dänzer]

On 2021-07-08 6:13 p.m., Michel Dänzer wrote:
> On 2021-06-29 12:36 p.m., Michel Dänzer wrote:
>> On 2021-06-28 7:16 p.m., Deucher, Alexander wrote:
>>>
>>> Thanks for narrowing this down.  There is new PCO SDMA firmware available (attached).  Can you try it?
>>
>> Sure, I'll try it, thanks.
> 
> So far, so good; no hang in a week. I'll try the rest of the new firmware as well now, will follow up if there's a hang again.

Unfortunately, I hit a hang again[0] with the new firmware. I'm now back to testing with the old SDMA firmware.

BTW, since the symptoms include GPU page faults, something might be going wrong with GPU page table updates via SDMA.


[0] Triggered by furiously resizing the gtk4-demo "OpenGL Transitions and Effects" window, which in a Wayland session I can otherwise recommend as a jaw-dropping experience. :)

-- 
Earthling Michel Dänzer               |               https://redhat.com
Libre software enthusiast             |             Mesa and X developer