Re: [PATCH] drm/amdgpu: Mark contexts guilty for any reset type

[André Almeida]

Hi Christian, thank you for your comments.

Em 24/04/2023 04:03, Christian König escreveu:
> Am 24.04.23 um 03:43 schrieb André Almeida:
> > When a DRM job timeout, the GPU is probably hang and amdgpu have some
> > ways to deal with that, ranging from soft recoveries to full device
> > reset. Anyway, when userspace ask the kernel the state of the context
> > (via AMDGPU_CTX_OP_QUERY_STATE), the kernel reports that the device was
> > reset, regardless if a full reset happened or not.
> >
> > However, amdgpu only marks a context guilty in the ASIC reset path. This
> > makes the userspace report incomplete, given that on soft recovery path
> > the guilty context is not told that it's the guilty one.
> >
> > Fix this by marking the context guilty for every type of reset when a
> > job timeouts.
>
> The guilty handling is pretty much broken by design and only works 
> because we go through multiple hops of validating the entity after the 
> job has already been pushed to the hw.

I see, thanks.

> I think we should probably just remove that completely and use an 
> approach where we check the in flight submissions in the query state 
> IOCTL.

Like the DRM_IOCTL_I915_GET_RESET_STATS approach?

> See my other patch on the mailing list regarding that.

Which one, the "[PATCH 1/8] drm/scheduler: properly forward fence 
errors" series?

> Additional to that I currently didn't considered soft-recovered 
> submissions as fatal and continue accepting submissions from that 
> context, but already wanted to talk with Marek about that behavior.

Interesting. I will try to test and validate this approach to see if the 
contexts keep working as expected on soft-resets.

> Regards,
> Christian.