[PATCH 1/3] drm/scheduler: track GPU active time per entity

Andrey Grodzovsky andrey.grodzovsky at amd.com
Thu Sep 8 18:33:07 UTC 2022


On 2022-09-08 14:10, Lucas Stach wrote:
> Track the accumulated time that jobs from this entity were active
> on the GPU. This allows drivers using the scheduler to trivially
> implement the DRM fdinfo when the hardware doesn't provide more
> specific information than signalling job completion anyways.
>
> Signed-off-by: Lucas Stach <l.stach at pengutronix.de>
> ---
>   drivers/gpu/drm/scheduler/sched_main.c | 6 ++++++
>   include/drm/gpu_scheduler.h            | 7 +++++++
>   2 files changed, 13 insertions(+)
>
> diff --git a/drivers/gpu/drm/scheduler/sched_main.c b/drivers/gpu/drm/scheduler/sched_main.c
> index 76fd2904c7c6..24c77a6a157f 100644
> --- a/drivers/gpu/drm/scheduler/sched_main.c
> +++ b/drivers/gpu/drm/scheduler/sched_main.c
> @@ -847,6 +847,12 @@ drm_sched_get_cleanup_job(struct drm_gpu_scheduler *sched)
>   
>   	spin_unlock(&sched->job_list_lock);
>   
> +	if (job) {
> +		job->entity->elapsed_ns += ktime_to_ns(
> +			ktime_sub(job->s_fence->finished.timestamp,
> +				  job->s_fence->scheduled.timestamp));
> +	}
> +
>   	return job;


Looks like you making as assumption that drm_sched_entity will always be 
allocated using kzalloc ? Isn't it a bit dangerous assumption ?

Andrey


>   }
>   
> diff --git a/include/drm/gpu_scheduler.h b/include/drm/gpu_scheduler.h
> index addb135eeea6..573bef640664 100644
> --- a/include/drm/gpu_scheduler.h
> +++ b/include/drm/gpu_scheduler.h
> @@ -196,6 +196,13 @@ struct drm_sched_entity {
>   	 * drm_sched_entity_fini().
>   	 */
>   	struct completion		entity_idle;
> +	/**
> +	 * @elapsed_ns
> +	 *
> +	 * Records the amount of time where jobs from this entity were active
> +	 * on the GPU.
> +	 */
> +	uint64_t elapsed_ns;
>   };
>   
>   /**


More information about the etnaviv mailing list