[Intel-gfx] ✓ Fi.CI.IGT: success for drm/i915: avoid concurrent writes to aux_inv (rev10)

[Tvrtko Ursulin]

On 30/03/2022 16:37, Yang, Fei wrote:
>>>>> *Patch Details*
>>>>> *Series:*	drm/i915: avoid concurrent writes to aux_inv (rev10)
>>>>> *URL:*	https://patchwork.freedesktop.org/series/100772/
>>>>> <https://patchwork.freedesktop.org/series/100772/>
>>>>> *State:*	success
>>>>> *Details:*
>>>>> https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_22704/index.html
>>>>> <https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_22704/index.html
>>>>>>
>>>>>
>>>>>
>>>>>     CI Bug Log - changes from CI_DRM_11415_full ->
>>>>> Patchwork_22704_full
>>>>>
>>>>>
>>>>>       Summary
>>>>>
>>>>> *SUCCESS*
>>>>>
>>>>> No regressions found.
>>>>
>>>> And pushed.
>>> Thank you so much.
>>>
>>>>
>>>> Does this have any chance of fixing hangs in multiple simultaneous media workloads?
>>> I hope so. The hang fixed by this patch is pretty easy to reproduce on TGL with two parallel video decoding processes.
>>
>> Great!
>>
>> Do you want to send it to stable in this case? Would need to look how far back it would be easy to backport.
> 
> How far back? The change is only applicable for GEN12, so I believe this won't go farther than 5.4?

No need to go beyond 5.9 AFAICT since that is when d248b371f747 
("drm/i915/gen12: Invalidate aux table entries forcibly") was added. 
That would mean 5.10, 5.15, 5.16 and 5.17 are potential candidates.

We recently had https://gitlab.freedesktop.org/drm/intel/-/issues/5271 
and there is also https://gitlab.freedesktop.org/drm/intel/-/issues/5025 
so it is probably worth propagating the fix to distros via stable since 
people are hitting it.

Regards,

Tvrtko