On Fri, 2025-05-30 at 11:01 -0300, Maíra Canal wrote:
> When a CL/CSD job times out, we check if the GPU has made any
> progress
> since the last timeout. If so, instead of resetting the hardware, we
> skip
> the reset and allow the timer to be rearmed. This gives long-running
> jobs
> a chance to complete.
>
> Use the DRM_GPU_SCHED_STAT_NO_HANG status to skip the reset and re-
> arm
> the timer.
>
> Signed-off-by: Maíra Canal <mca...@igalia.com>
> ---
> drivers/gpu/drm/v3d/v3d_sched.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/gpu/drm/v3d/v3d_sched.c
> b/drivers/gpu/drm/v3d/v3d_sched.c
> index
> e1997387831541fb053e472672004cf511c25558..fbb09a8aff3740b5cd59573b5f2
> e26b2ee352dfb 100644
> --- a/drivers/gpu/drm/v3d/v3d_sched.c
> +++ b/drivers/gpu/drm/v3d/v3d_sched.c
> @@ -761,7 +761,7 @@ v3d_cl_job_timedout(struct drm_sched_job
> *sched_job, enum v3d_queue q,
> if (*timedout_ctca != ctca || *timedout_ctra != ctra) {
> *timedout_ctca = ctca;
> *timedout_ctra = ctra;
> - return DRM_GPU_SCHED_STAT_RESET;
> + return DRM_GPU_SCHED_STAT_NO_HANG;
> }
>
> return v3d_gpu_reset_for_timeout(v3d, sched_job);
> @@ -805,7 +805,7 @@ v3d_csd_job_timedout(struct drm_sched_job
> *sched_job)
> */
> if (job->timedout_batches != batches) {
> job->timedout_batches = batches;
> - return DRM_GPU_SCHED_STAT_RESET;
> + return DRM_GPU_SCHED_STAT_NO_HANG;
> }
Wait a second, help me out here quickly. You already added workaround
stuff where you manipulate the scheduler's pending_list, as you state
in the cover letter. That code here [1].
Don't you have to remove the very same code in this series again to
still have correct behavior in your driver?
As I see it, all drm branches end up in Linus's tree ultimately. So I'd
think about potential branch-races in case you didn't already.
P.
[1]
https://lore.kernel.org/dri-devel/20250430210643.57924-1-mca...@igalia.com/T/
>
> return v3d_gpu_reset_for_timeout(v3d, sched_job);
>
