Re: [PATCH v3] drm/sched: Make sure we wait for all dependencies in kill_jobs_cb()

[Boris Brezillon <boris.brezillon@collabora.com>]

On Tue, 13 Jun 2023 11:28:45 +0200
Boris Brezillon <boris.brezillon@collabora.com> wrote:

> drm_sched_entity_kill_jobs_cb() logic is omitting the last fence popped
> from the dependency array that was waited upon before
> drm_sched_entity_kill() was called (drm_sched_entity::dependency field),
> so we're basically waiting for all dependencies except one.
> 
> In theory, this wait shouldn't be needed because resources should have
> their users registered to the dma_resv object, thus guaranteeing that
> future jobs wanting to access these resources wait on all the previous
> users (depending on the access type, of course). But we want to keep
> these explicit waits in the kill entity path just in case.
> 
> Let's make sure we keep all dependencies in the array in
> drm_sched_job_dependency(), so we can iterate over the array and wait
> in drm_sched_entity_kill_jobs_cb().
> 
> We also make sure we wait on drm_sched_fence::finished if we were asked
> to wait on drm_sched_fence::scheduled, but the intent was probably to
> delegate the wait to the GPU, but we want resources to be completely
> idle when killing jobs.
> 
> v3:
> - Always wait for drm_sched_fence::finished fences in
>   drm_sched_entity_kill_jobs_cb() when we see a sched_fence
> 
> v2:
> - Don't evict deps in drm_sched_job_dependency()
> 
> Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
> Suggested-by: "Christian König" <christian.koenig@amd.com>
> Cc: Frank Binns <frank.binns@imgtec.com>
> Cc: Sarah Walker <sarah.walker@imgtec.com>
> Cc: Donald Robson <donald.robson@imgtec.com>
> Cc: Luben Tuikov <luben.tuikov@amd.com>
> Cc: David Airlie <airlied@gmail.com>
> Cc: Daniel Vetter <daniel@ffwll.ch>
> Cc: Sumit Semwal <sumit.semwal@linaro.org>
> Cc: "Christian König" <christian.koenig@amd.com>
> ---
>  drivers/gpu/drm/scheduler/sched_entity.c | 28 ++++++++++++++++++++----
>  1 file changed, 24 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/gpu/drm/scheduler/sched_entity.c b/drivers/gpu/drm/scheduler/sched_entity.c
> index 68e807ae136a..bc1bc3d47f7f 100644
> --- a/drivers/gpu/drm/scheduler/sched_entity.c
> +++ b/drivers/gpu/drm/scheduler/sched_entity.c
> @@ -176,13 +176,24 @@ static void drm_sched_entity_kill_jobs_cb(struct dma_fence *f,
>  {
>  	struct drm_sched_job *job = container_of(cb, struct drm_sched_job,
>  						 finish_cb);
> +	unsigned long index;
>  	int r;
>  
>  	dma_fence_put(f);
>  
>  	/* Wait for all dependencies to avoid data corruptions */
> -	while (!xa_empty(&job->dependencies)) {
> -		f = xa_erase(&job->dependencies, job->last_dependency++);
> +	xa_for_each(&job->dependencies, index, f) {
> +		struct drm_sched_fence *s_fence = to_drm_sched_fence(f);
> +
> +		/* Make sure we wait for the finished fence here, so we can
> +		 * guarantee that any job we depend on that is still accessing
> +		 * resources is done before we signal this job finished fence
> +		 * and unblock further accesses on those resources.
> +		 */
> +		if (s_fence && f == &s_fence->scheduled)
> +			f = &s_fence->finished;

Oops. There's a use-after-free+leak bug here: when switching from
scheduled to finished fence, we need to grab a ref on the finished
fence and release the one we had on the scheduled fence.

Sending a v4 to fix that.

> +
> +		xa_erase(&job->dependencies, index);
>  		r = dma_fence_add_callback(f, &job->finish_cb,
>  					   drm_sched_entity_kill_jobs_cb);
>  		if (!r)
> @@ -415,8 +426,17 @@ static struct dma_fence *
>  drm_sched_job_dependency(struct drm_sched_job *job,
>  			 struct drm_sched_entity *entity)
>  {
> -	if (!xa_empty(&job->dependencies))
> -		return xa_erase(&job->dependencies, job->last_dependency++);
> +	struct dma_fence *f;
> +
> +	/* We keep the fence around, so we can iterate over all dependencies
> +	 * in drm_sched_entity_kill_jobs_cb() to ensure all deps are signaled
> +	 * before killing the job.
> +	 */
> +	f = xa_load(&job->dependencies, job->last_dependency);
> +	if (f) {
> +		job->last_dependency++;
> +		return dma_fence_get(f);
> +	}
>  
>  	if (job->sched->ops->prepare_job)
>  		return job->sched->ops->prepare_job(job, entity);