[Intel-xe] [PATCH 1/2] drm/xe/sched_job: prefer dma_fence_is_later

[Thomas Hellström]

On 4/6/23 18:26, Matthew Auld wrote:
> Doesn't look like we are accounting for seqno wrap. Just use
> __dma_fence_is_later() like we already do for xe_hw_fence_signaled().
>
> Signed-off-by: Matthew Auld <matthew.auld at intel.com>
> Cc: Thomas Hellström <thomas.hellstrom at linux.intel.com>
> Cc: Matthew Brost <matthew.brost at intel.com>

Reviewed-by: Thomas Hellström <thomas.hellstrom at linux.intel.com>


> ---
>   drivers/gpu/drm/xe/xe_sched_job.c | 7 +++++--
>   1 file changed, 5 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/gpu/drm/xe/xe_sched_job.c b/drivers/gpu/drm/xe/xe_sched_job.c
> index 2985caa6097b..ed0b69ef449c 100644
> --- a/drivers/gpu/drm/xe/xe_sched_job.c
> +++ b/drivers/gpu/drm/xe/xe_sched_job.c
> @@ -229,7 +229,9 @@ bool xe_sched_job_started(struct xe_sched_job *job)
>   {
>   	struct xe_lrc *lrc = job->engine->lrc;
>   
> -	return xe_lrc_start_seqno(lrc) >= xe_sched_job_seqno(job);
> +	return !__dma_fence_is_later(xe_sched_job_seqno(job),
> +				     xe_lrc_start_seqno(lrc),
> +				     job->fence->ops);
>   }
>   
>   bool xe_sched_job_completed(struct xe_sched_job *job)
> @@ -241,7 +243,8 @@ bool xe_sched_job_completed(struct xe_sched_job *job)
>   	 * parallel handshake is done.
>   	 */
>   
> -	return xe_lrc_seqno(lrc) >= xe_sched_job_seqno(job);
> +	return !__dma_fence_is_later(xe_sched_job_seqno(job), xe_lrc_seqno(lrc),
> +				     job->fence->ops);
>   }
>   
>   void xe_sched_job_arm(struct xe_sched_job *job)