diff mbox series

[v5,04/40] drm/sched: Add enqueue credit limit

Message ID 20250519175348.11924-5-robdclark@gmail.com
State New
Headers show
Series drm/msm: sparse / "VM_BIND" support | expand

Commit Message

Rob Clark May 19, 2025, 5:51 p.m. UTC 
From: Rob Clark <robdclark@chromium.org>

Similar to the existing credit limit mechanism, but applying to jobs
enqueued to the scheduler but not yet run.

The use case is to put an upper bound on preallocated, and potentially
unneeded, pgtable pages.  When this limit is exceeded, pushing new jobs
will block until the count drops below the limit.

Cc: Philipp Stanner <phasta@mailbox.org>
Cc: Danilo Krummrich <dakr@kernel.org>
Signed-off-by: Rob Clark <robdclark@chromium.org>
---
 drivers/gpu/drm/scheduler/sched_entity.c | 19 +++++++++++++++++--
 drivers/gpu/drm/scheduler/sched_main.c   |  3 +++
 include/drm/gpu_scheduler.h              | 24 +++++++++++++++++++++++-
 3 files changed, 43 insertions(+), 3 deletions(-)
diff mbox series

Patch

diff --git a/drivers/gpu/drm/scheduler/sched_entity.c b/drivers/gpu/drm/scheduler/sched_entity.c
index bd39db7bb240..8e6b12563348 100644
--- a/drivers/gpu/drm/scheduler/sched_entity.c
+++ b/drivers/gpu/drm/scheduler/sched_entity.c
@@ -579,12 +579,25 @@  void drm_sched_entity_select_rq(struct drm_sched_entity *entity)
  * fence sequence number this function should be called with drm_sched_job_arm()
  * under common lock for the struct drm_sched_entity that was set up for
  * @sched_job in drm_sched_job_init().
+ *
+ * If enqueue_credit_limit is used, this can return -ERESTARTSYS if the system
+ * call is interrupted.
  */
-void drm_sched_entity_push_job(struct drm_sched_job *sched_job)
+int drm_sched_entity_push_job(struct drm_sched_job *sched_job)
 {
 	struct drm_sched_entity *entity = sched_job->entity;
+	struct drm_gpu_scheduler *sched = sched_job->sched;
 	bool first;
 	ktime_t submit_ts;
+	int ret;
+
+	ret = wait_event_interruptible(
+			sched->job_scheduled,
+			atomic_read(&sched->enqueue_credit_count) <=
+			sched->enqueue_credit_limit);
+	if (ret)
+		return ret;
+	atomic_add(sched_job->enqueue_credits, &sched->enqueue_credit_count);
 
 	trace_drm_sched_job(sched_job, entity);
 	atomic_inc(entity->rq->sched->score);
@@ -609,7 +622,7 @@  void drm_sched_entity_push_job(struct drm_sched_job *sched_job)
 			spin_unlock(&entity->lock);
 
 			DRM_ERROR("Trying to push to a killed entity\n");
-			return;
+			return -EINVAL;
 		}
 
 		rq = entity->rq;
@@ -626,5 +639,7 @@  void drm_sched_entity_push_job(struct drm_sched_job *sched_job)
 
 		drm_sched_wakeup(sched);
 	}
+
+	return 0;
 }
 EXPORT_SYMBOL(drm_sched_entity_push_job);
diff --git a/drivers/gpu/drm/scheduler/sched_main.c b/drivers/gpu/drm/scheduler/sched_main.c
index cda1216adfa4..5f812253656a 100644
--- a/drivers/gpu/drm/scheduler/sched_main.c
+++ b/drivers/gpu/drm/scheduler/sched_main.c
@@ -1221,6 +1221,7 @@  static void drm_sched_run_job_work(struct work_struct *w)
 
 	trace_drm_run_job(sched_job, entity);
 	fence = sched->ops->run_job(sched_job);
+	atomic_sub(sched_job->enqueue_credits, &sched->enqueue_credit_count);
 	complete_all(&entity->entity_idle);
 	drm_sched_fence_scheduled(s_fence, fence);
 
@@ -1257,6 +1258,7 @@  int drm_sched_init(struct drm_gpu_scheduler *sched, const struct drm_sched_init_
 
 	sched->ops = args->ops;
 	sched->credit_limit = args->credit_limit;
+	sched->enqueue_credit_limit = args->enqueue_credit_limit;
 	sched->name = args->name;
 	sched->timeout = args->timeout;
 	sched->hang_limit = args->hang_limit;
@@ -1312,6 +1314,7 @@  int drm_sched_init(struct drm_gpu_scheduler *sched, const struct drm_sched_init_
 	INIT_LIST_HEAD(&sched->pending_list);
 	spin_lock_init(&sched->job_list_lock);
 	atomic_set(&sched->credit_count, 0);
+	atomic_set(&sched->enqueue_credit_count, 0);
 	INIT_DELAYED_WORK(&sched->work_tdr, drm_sched_job_timedout);
 	INIT_WORK(&sched->work_run_job, drm_sched_run_job_work);
 	INIT_WORK(&sched->work_free_job, drm_sched_free_job_work);
diff --git a/include/drm/gpu_scheduler.h b/include/drm/gpu_scheduler.h
index da64232c989d..8ec5000f81e1 100644
--- a/include/drm/gpu_scheduler.h
+++ b/include/drm/gpu_scheduler.h
@@ -366,6 +366,19 @@  struct drm_sched_job {
 
 	enum drm_sched_priority		s_priority;
 	u32				credits;
+	/**
+	 * @enqueue_credits: the number of enqueue credits this job
+	 * contributes to the drm_gpu_scheduler.enqueue_credit_count.
+	 *
+	 * The (optional) @enqueue_credits should be set before calling
+	 * drm_sched_entity_push_job().  When sum of all the jobs pushed
+	 * to the entity, but not yet having their run_job() callback
+	 * called exceeds @drm_gpu_scheduler.enqueue_credit_limit, the
+	 * drm_sched_entity_push_job() will block until the count drops
+	 * back below the limit, providing a way to throttle the number
+	 * of queued, but not yet run, jobs.
+	 */
+	u32				enqueue_credits;
 	/** @last_dependency: tracks @dependencies as they signal */
 	unsigned int			last_dependency;
 	atomic_t			karma;
@@ -485,6 +498,10 @@  struct drm_sched_backend_ops {
  * @ops: backend operations provided by the driver.
  * @credit_limit: the credit limit of this scheduler
  * @credit_count: the current credit count of this scheduler
+ * @enqueue_credit_limit: the credit limit of jobs pushed to scheduler and not
+ *                        yet run
+ * @enqueue_credit_count: the current crdit count of jobs pushed to scheduler
+ *                        but not yet run
  * @timeout: the time after which a job is removed from the scheduler.
  * @name: name of the ring for which this scheduler is being used.
  * @num_rqs: Number of run-queues. This is at most DRM_SCHED_PRIORITY_COUNT,
@@ -518,6 +535,8 @@  struct drm_gpu_scheduler {
 	const struct drm_sched_backend_ops	*ops;
 	u32				credit_limit;
 	atomic_t			credit_count;
+	u32				enqueue_credit_limit;
+	atomic_t			enqueue_credit_count;
 	long				timeout;
 	const char			*name;
 	u32                             num_rqs;
@@ -550,6 +569,8 @@  struct drm_gpu_scheduler {
  * @num_rqs: Number of run-queues. This may be at most DRM_SCHED_PRIORITY_COUNT,
  *	     as there's usually one run-queue per priority, but may be less.
  * @credit_limit: the number of credits this scheduler can hold from all jobs
+ * @enqueue_credit_limit: the number of credits that can be enqueued before
+ *                        drm_sched_entity_push_job() blocks
  * @hang_limit: number of times to allow a job to hang before dropping it.
  *		This mechanism is DEPRECATED. Set it to 0.
  * @timeout: timeout value in jiffies for submitted jobs.
@@ -564,6 +585,7 @@  struct drm_sched_init_args {
 	struct workqueue_struct *timeout_wq;
 	u32 num_rqs;
 	u32 credit_limit;
+	u32 enqueue_credit_limit;
 	unsigned int hang_limit;
 	long timeout;
 	atomic_t *score;
@@ -600,7 +622,7 @@  int drm_sched_job_init(struct drm_sched_job *job,
 		       struct drm_sched_entity *entity,
 		       u32 credits, void *owner);
 void drm_sched_job_arm(struct drm_sched_job *job);
-void drm_sched_entity_push_job(struct drm_sched_job *sched_job);
+int drm_sched_entity_push_job(struct drm_sched_job *sched_job);
 int drm_sched_job_add_dependency(struct drm_sched_job *job,
 				 struct dma_fence *fence);
 int drm_sched_job_add_syncobj_dependency(struct drm_sched_job *job,