[PATCH 3/5] drm/v3d: Create a struct to store the GPU stats
Chema Casanova
jmcasanova at igalia.com
Mon Apr 15 15:27:21 UTC 2024
El 3/4/24 a las 22:24, Maíra Canal escribió:
> This will make it easier to instantiate the GPU stats variables and it
> will create a structure where we can store all the variables that refer
> to GPU stats.
>
> Signed-off-by: Maíra Canal <mcanal at igalia.com>
> ---
> drivers/gpu/drm/v3d/v3d_drv.c | 14 ++++++--------
> drivers/gpu/drm/v3d/v3d_drv.h | 18 ++++++++++--------
> drivers/gpu/drm/v3d/v3d_gem.c | 4 +---
> drivers/gpu/drm/v3d/v3d_sched.c | 20 ++++++++++++--------
> drivers/gpu/drm/v3d/v3d_sysfs.c | 10 ++++++----
> 5 files changed, 35 insertions(+), 31 deletions(-)
>
> diff --git a/drivers/gpu/drm/v3d/v3d_drv.c b/drivers/gpu/drm/v3d/v3d_drv.c
> index 3debf37e7d9b..cbb62be18aa5 100644
> --- a/drivers/gpu/drm/v3d/v3d_drv.c
> +++ b/drivers/gpu/drm/v3d/v3d_drv.c
> @@ -115,14 +115,11 @@ v3d_open(struct drm_device *dev, struct drm_file *file)
> v3d_priv->v3d = v3d;
>
> for (i = 0; i < V3D_MAX_QUEUES; i++) {
> - v3d_priv->enabled_ns[i] = 0;
> - v3d_priv->start_ns[i] = 0;
> - v3d_priv->jobs_sent[i] = 0;
> -
> sched = &v3d->queue[i].sched;
> drm_sched_entity_init(&v3d_priv->sched_entity[i],
> DRM_SCHED_PRIORITY_NORMAL, &sched,
> 1, NULL);
> + memset(&v3d_priv->stats[i], 0, sizeof(v3d_priv->stats[i]));
> }
>
> v3d_perfmon_open_file(v3d_priv);
> @@ -151,20 +148,21 @@ static void v3d_show_fdinfo(struct drm_printer *p, struct drm_file *file)
> enum v3d_queue queue;
>
> for (queue = 0; queue < V3D_MAX_QUEUES; queue++) {
> + struct v3d_stats *stats = &file_priv->stats[queue];
> +
> /* Note that, in case of a GPU reset, the time spent during an
> * attempt of executing the job is not computed in the runtime.
> */
> drm_printf(p, "drm-engine-%s: \t%llu ns\n",
> v3d_queue_to_string(queue),
> - file_priv->start_ns[queue] ? file_priv->enabled_ns[queue]
> - + timestamp - file_priv->start_ns[queue]
> - : file_priv->enabled_ns[queue]);
> + stats->start_ns ? stats->enabled_ns + timestamp - stats->start_ns
> + : stats->enabled_ns);
>
> /* Note that we only count jobs that completed. Therefore, jobs
> * that were resubmitted due to a GPU reset are not computed.
> */
> drm_printf(p, "v3d-jobs-%s: \t%llu jobs\n",
> - v3d_queue_to_string(queue), file_priv->jobs_sent[queue]);
> + v3d_queue_to_string(queue), stats->jobs_sent);
> }
> }
>
> diff --git a/drivers/gpu/drm/v3d/v3d_drv.h b/drivers/gpu/drm/v3d/v3d_drv.h
> index ee3545226d7f..0117593976ed 100644
> --- a/drivers/gpu/drm/v3d/v3d_drv.h
> +++ b/drivers/gpu/drm/v3d/v3d_drv.h
> @@ -36,15 +36,20 @@ static inline char *v3d_queue_to_string(enum v3d_queue queue)
> return "UNKNOWN";
> }
>
> +struct v3d_stats {
> + u64 start_ns;
> + u64 enabled_ns;
> + u64 jobs_sent;
As currently, we are only accounting jobs that have been completed. What
do you think of renaming jobs_sent to jobs_completed?
> +};
> +
> struct v3d_queue_state {
> struct drm_gpu_scheduler sched;
>
> u64 fence_context;
> u64 emit_seqno;
>
> - u64 start_ns;
> - u64 enabled_ns;
> - u64 jobs_sent;
> + /* Stores the GPU stats for this queue in the global context. */
> + struct v3d_stats stats;
> };
>
> /* Performance monitor object. The perform lifetime is controlled by userspace
> @@ -188,11 +193,8 @@ struct v3d_file_priv {
>
> struct drm_sched_entity sched_entity[V3D_MAX_QUEUES];
>
> - u64 start_ns[V3D_MAX_QUEUES];
> -
> - u64 enabled_ns[V3D_MAX_QUEUES];
> -
> - u64 jobs_sent[V3D_MAX_QUEUES];
> + /* Stores the GPU stats for a specific queue for this fd. */
> + struct v3d_stats stats[V3D_MAX_QUEUES];
> };
>
> struct v3d_bo {
> diff --git a/drivers/gpu/drm/v3d/v3d_gem.c b/drivers/gpu/drm/v3d/v3d_gem.c
> index afc565078c78..d14589d3ae6c 100644
> --- a/drivers/gpu/drm/v3d/v3d_gem.c
> +++ b/drivers/gpu/drm/v3d/v3d_gem.c
> @@ -248,9 +248,7 @@ v3d_gem_init(struct drm_device *dev)
>
> for (i = 0; i < V3D_MAX_QUEUES; i++) {
> v3d->queue[i].fence_context = dma_fence_context_alloc(1);
> - v3d->queue[i].start_ns = 0;
> - v3d->queue[i].enabled_ns = 0;
> - v3d->queue[i].jobs_sent = 0;
> + memset(&v3d->queue[i].stats, 0, sizeof(v3d->queue[i].stats));
> }
>
> spin_lock_init(&v3d->mm_lock);
> diff --git a/drivers/gpu/drm/v3d/v3d_sched.c b/drivers/gpu/drm/v3d/v3d_sched.c
> index 8ca61bcd4b1c..ea5f5a84b55b 100644
> --- a/drivers/gpu/drm/v3d/v3d_sched.c
> +++ b/drivers/gpu/drm/v3d/v3d_sched.c
> @@ -110,10 +110,12 @@ v3d_job_start_stats(struct v3d_job *job, enum v3d_queue queue)
> {
> struct v3d_dev *v3d = job->v3d;
> struct v3d_file_priv *file = job->file->driver_priv;
> + struct v3d_stats *global_stats = &v3d->queue[queue].stats;
> + struct v3d_stats *local_stats = &file->stats[queue];
> u64 now = local_clock();
>
> - file->start_ns[queue] = now;
> - v3d->queue[queue].start_ns = now;
> + local_stats->start_ns = now;
> + global_stats->start_ns = now;
> }
>
> void
> @@ -121,15 +123,17 @@ v3d_job_update_stats(struct v3d_job *job, enum v3d_queue queue)
> {
> struct v3d_dev *v3d = job->v3d;
> struct v3d_file_priv *file = job->file->driver_priv;
> + struct v3d_stats *global_stats = &v3d->queue[queue].stats;
> + struct v3d_stats *local_stats = &file->stats[queue];
> u64 now = local_clock();
>
> - file->enabled_ns[queue] += now - file->start_ns[queue];
> - file->jobs_sent[queue]++;
> - file->start_ns[queue] = 0;
> + local_stats->enabled_ns += now - local_stats->start_ns;
> + local_stats->jobs_sent++;
> + local_stats->start_ns = 0;
>
> - v3d->queue[queue].enabled_ns += now - v3d->queue[queue].start_ns;
> - v3d->queue[queue].jobs_sent++;
> - v3d->queue[queue].start_ns = 0;
> + global_stats->enabled_ns += now - global_stats->start_ns;
> + global_stats->jobs_sent++;
> + global_stats->start_ns = 0;
> }
>
> static struct dma_fence *v3d_bin_job_run(struct drm_sched_job *sched_job)
> diff --git a/drivers/gpu/drm/v3d/v3d_sysfs.c b/drivers/gpu/drm/v3d/v3d_sysfs.c
> index d106845ba890..1eb5f3de6937 100644
> --- a/drivers/gpu/drm/v3d/v3d_sysfs.c
> +++ b/drivers/gpu/drm/v3d/v3d_sysfs.c
> @@ -21,8 +21,10 @@ gpu_stats_show(struct device *dev, struct device_attribute *attr, char *buf)
> len += sysfs_emit(buf, "queue\ttimestamp\tjobs\truntime\n");
>
> for (queue = 0; queue < V3D_MAX_QUEUES; queue++) {
> - if (v3d->queue[queue].start_ns)
> - active_runtime = timestamp - v3d->queue[queue].start_ns;
> + struct v3d_stats *stats = &v3d->queue[queue].stats;
> +
> + if (stats->start_ns)
> + active_runtime = timestamp - stats->start_ns;
> else
> active_runtime = 0;
>
> @@ -39,8 +41,8 @@ gpu_stats_show(struct device *dev, struct device_attribute *attr, char *buf)
> len += sysfs_emit_at(buf, len, "%s\t%llu\t%llu\t%llu\n",
> v3d_queue_to_string(queue),
> timestamp,
> - v3d->queue[queue].jobs_sent,
> - v3d->queue[queue].enabled_ns + active_runtime);
> + stats->jobs_sent,
> + stats->enabled_ns + active_runtime);
> }
>
> return len;
Not related to this review, but as a general doubt, I miss with our current
implementation the possibility of the user checking if a queue is running
or not when the stats are checked. That information is stored in the struct
as (start_ns !=0) but is not exposed, and it could be interesting
information.
It would explain why you have running time (enabled_ns>0) but
jobs_sent/completed is still 0, that is possible as you have one job
running.
It would be useful also if you want to check the average enabled/time
per job.
As we should consider the number of jobs completed + currently running
to calculate the average.
In any case, this patch is:
Reviewed-by: Jose Maria Casanova Crespo <jmcasanova at igalia.com>
More information about the dri-devel
mailing list