include/drm/gpu_scheduler.h

38  * Setting this flag on a scheduler fence prevents pipelining of jobs depending
85  * ring, and the scheduler will alternate between entities based on
114 	 * be scheduled on any scheduler on this list.
200 	 * by the scheduler thread, can be accessed locklessly from
247  * @sched: the scheduler to which this rq belongs to.
269          * @scheduled: this fence is what will be signaled by the scheduler
275          * @finished: this fence is what will be signaled by the scheduler
299          * @sched: the scheduler instance to which the job having this struct
320  * @sched: the scheduler instance on which this job is scheduled.
323  * @credits: the number of credits this job contributes to the scheduler
325  * @id: a unique id assigned to each job scheduled on the scheduler.
327  *         limit of the scheduler then the job is marked guilty and will not
334  * should call drm_sched_entity_push_job() once it wants the scheduler
393  *	called by the scheduler
401 	 * Called when the scheduler is considering scheduling this job next, to
428 	 * 1. Stop the scheduler using drm_sched_stop(). This will park the
429 	 *    scheduler thread and cancel the timeout work, guaranteeing that
434 	 * 5. Restart the scheduler using drm_sched_start(). At that point, new
435 	 *    jobs can be queued, and the scheduler thread is unblocked
470 	 * @update_job_credits: Called when the scheduler is considering this
484  * struct drm_gpu_scheduler - scheduler instance-specific data
487  * @credit_limit: the credit limit of this scheduler
488  * @credit_count: the current credit count of this scheduler
489  * @timeout: the time after which a job is removed from the scheduler.
490  * @name: name of the ring for which this scheduler is being used.
494  * @job_scheduled: once @drm_sched_entity_do_release is called the scheduler
500  * @work_run_job: work which calls run_job op of each scheduler.
501  * @work_free_job: work which calls free_job op of each scheduler.
513  * @own_submit_wq: scheduler owns allocation of @submit_wq
516  * One scheduler is implemented for each hardware ring.