On Wed, 2025-12-10 at 13:57 +0100, Christian König wrote:
> The job might already be freed up here. So cache the pasid for later
> use.
>
> Signed-off-by: Christian König <[email protected]>
> Suggested-by: Philipp Stanner <[email protected]>
No! :D
You read the wrong quote.
Matthew found your issue (in the middle of my time zone's night, while
I was peacefully asleep).
> Fixes: a72002cb181f ("drm/amdgpu: Make use of drm_wedge_task_info")
I think there was a gitlab ticket about that a month ago? Could be
added to Closes:
P.
> ---
> drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 5 +++--
> 1 file changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> index 654f4844b7ad..84bb89498e12 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> @@ -6506,8 +6506,9 @@ int amdgpu_device_gpu_recover(struct amdgpu_device
> *adev,
> struct list_head device_list;
> bool job_signaled = false;
> struct amdgpu_hive_info *hive = NULL;
> - int r = 0;
> bool need_emergency_restart = false;
> + unsigned int pasid = job->pasid;
> + int r = 0;
>
> /*
> * If it reaches here because of hang/timeout and a RAS error is
> @@ -6605,7 +6606,7 @@ int amdgpu_device_gpu_recover(struct amdgpu_device
> *adev,
> struct amdgpu_task_info *ti = NULL;
>
> if (job)
> - ti = amdgpu_vm_get_task_info_pasid(adev, job->pasid);
> + ti = amdgpu_vm_get_task_info_pasid(adev, pasid);
>
> drm_dev_wedged_event(adev_to_drm(adev), DRM_WEDGE_RECOVERY_NONE,
> ti ? &ti->task : NULL);
