I think it's a neat approach, thanks !
/Monk
-Original Message-
From: Koenig, Christian
Sent: Friday, May 10, 2019 8:07 PM
To: Liu, Monk ; amd-gfx@lists.freedesktop.org
Subject: Re: [PATCH] drm/amdgpu: avoid duplicated tmo report on same job
Yeah, that's indeed a bit problematic. How
holds TDR work) stuck there
> and other work item never get handled ?
>
> /Monk
>
> -Original Message-
> From: Koenig, Christian
> Sent: Friday, May 10, 2019 4:58 PM
> To: Liu, Monk ; amd-gfx@lists.freedesktop.org
> Subject: Re: [PATCH] drm/amdgpu: avoid duplicate
/amdgpu: avoid duplicated tmo report on same job
Hi Monk,
yeah, that's much closer to what I had in mind. But your comments got me
thinking more about this.
What do you think about changing amdgpu_job_timedout() like this:
if (amdgpu_device_should_recover_gpu(ring->a
Hi Monk,
yeah, that's much closer to what I had in mind. But your comments got me
thinking more about this.
What do you think about changing amdgpu_job_timedout() like this:
if (amdgpu_device_should_recover_gpu(ring->adev))
amdgpu_device_gpu_recover(ring->adev, job);
+
Hi Christian,
What about this one, no timer logic change on scheduler part, only the
duplicated tmo report is muted
/Monk
-Original Message-
From: amd-gfx On Behalf Of Monk Liu
Sent: Friday, May 10, 2019 3:18 PM
To: amd-gfx@lists.freedesktop.org
Cc: Liu, Monk
Subject: [PATCH]