[PATCH] drm/amdgpu: Do gpu recovery when no job is running

[Christian König]

Am 09.09.20 um 10:56 schrieb Liu ChengZhe:
> In function flr_work, do gpu recovery when no job is running
> instead of when some job is running. Because if there is job
> in list, amdgpu_job_timedout will do the gpu recovery.

Maybe write something like "Fix the logic by inverting it".

With that the patch is Acked-by: Christian König <christian.koenig at amd.com>

>
> Signed-off-by: Liu ChengZhe <ChengZhe.Liu at amd.com>
> ---
>   drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c | 3 ++-
>   drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c | 2 +-
>   2 files changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c b/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c
> index 9c07014d9bd6..f5ce9a9f4cf5 100644
> --- a/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c
> +++ b/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c
> @@ -262,7 +262,8 @@ static void xgpu_ai_mailbox_flr_work(struct work_struct *work)
>   
>   	/* Trigger recovery for world switch failure if no TDR */
>   	if (amdgpu_device_should_recover_gpu(adev)
> -		&& (amdgpu_device_has_job_running(adev) || adev->sdma_timeout == MAX_SCHEDULE_TIMEOUT))
> +		&& (!amdgpu_device_has_job_running(adev) ||
> +		adev->sdma_timeout == MAX_SCHEDULE_TIMEOUT))
>   		amdgpu_device_gpu_recover(adev, NULL);
>   }
>   
> diff --git a/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c b/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c
> index 9c23abf9b140..666ed99cc14b 100644
> --- a/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c
> +++ b/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c
> @@ -283,7 +283,7 @@ static void xgpu_nv_mailbox_flr_work(struct work_struct *work)
>   
>   	/* Trigger recovery for world switch failure if no TDR */
>   	if (amdgpu_device_should_recover_gpu(adev)
> -		&& (amdgpu_device_has_job_running(adev) ||
> +		&& (!amdgpu_device_has_job_running(adev) ||
>   		adev->sdma_timeout == MAX_SCHEDULE_TIMEOUT ||
>   		adev->gfx_timeout == MAX_SCHEDULE_TIMEOUT ||
>   		adev->compute_timeout == MAX_SCHEDULE_TIMEOUT ||