[PATCH v2 4/6] drm/v3d: Hold v3d_stats references in each job

[Maíra Canal <mcanal@igalia.com>]

From: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>

Have each job hold its own references to the per-fd and global stats
objects. This eliminates the need for `queue_lock` protection in the
stats update path, since the job's stats pointers are guaranteed to
remain valid for the job's entire lifetime regardless of file descriptor
closure.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Co-developed-by: Maíra Canal <mcanal@igalia.com>
Signed-off-by: Maíra Canal <mcanal@igalia.com>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
---
 drivers/gpu/drm/v3d/v3d_drv.h    |  6 ++++-
 drivers/gpu/drm/v3d/v3d_irq.c    |  2 +-
 drivers/gpu/drm/v3d/v3d_sched.c  | 57 ++++++++++++++++------------------------
 drivers/gpu/drm/v3d/v3d_submit.c |  6 +++++
 4 files changed, 34 insertions(+), 37 deletions(-)

diff --git a/drivers/gpu/drm/v3d/v3d_drv.h b/drivers/gpu/drm/v3d/v3d_drv.h
index 03fa2d174b1ca8b5a98a72c4addaa1f977d11174..72c3f40715dae6e86e0c8356cb997cdf1cf03fae 100644
--- a/drivers/gpu/drm/v3d/v3d_drv.h
+++ b/drivers/gpu/drm/v3d/v3d_drv.h
@@ -326,6 +326,10 @@ struct v3d_job {
 	 */
 	struct v3d_file_priv *file_priv;
 
+	/* Pointers to this job's per-fd and global queue stats. */
+	struct v3d_stats *client_stats;
+	struct v3d_stats *global_stats;
+
 	/* Callback for the freeing of the job on refcount going to 0. */
 	void (*free)(struct kref *ref);
 };
@@ -607,7 +611,7 @@ void v3d_performance_query_info_free(struct v3d_performance_query_info *query_in
 				     unsigned int count);
 struct v3d_stats *v3d_stats_alloc(void);
 void v3d_stats_release(struct kref *refcount);
-void v3d_job_update_stats(struct v3d_job *job, enum v3d_queue q);
+void v3d_job_update_stats(struct v3d_job *job);
 int v3d_sched_init(struct v3d_dev *v3d);
 void v3d_sched_fini(struct v3d_dev *v3d);
 
diff --git a/drivers/gpu/drm/v3d/v3d_irq.c b/drivers/gpu/drm/v3d/v3d_irq.c
index 1249f6e64b979fe29cf2b9bfc43b39aa755f71ce..c28e74ab5442857031b48bcbd4e43eb48c1e0f07 100644
--- a/drivers/gpu/drm/v3d/v3d_irq.c
+++ b/drivers/gpu/drm/v3d/v3d_irq.c
@@ -92,7 +92,7 @@ v3d_irq_signal_fence(struct v3d_dev *v3d, enum v3d_queue q,
 	struct v3d_queue_state *queue = &v3d->queue[q];
 	struct v3d_fence *fence = to_v3d_fence(queue->active_job->irq_fence);
 
-	v3d_job_update_stats(queue->active_job, q);
+	v3d_job_update_stats(queue->active_job);
 	trace_irq(&v3d->drm, fence->seqno);
 
 	queue->active_job = NULL;
diff --git a/drivers/gpu/drm/v3d/v3d_sched.c b/drivers/gpu/drm/v3d/v3d_sched.c
index 787f21337b2a03a923342fe4f5f927e9214082c4..5c387a152e33f5ccbca6a9af97675f050a2a701f 100644
--- a/drivers/gpu/drm/v3d/v3d_sched.c
+++ b/drivers/gpu/drm/v3d/v3d_sched.c
@@ -158,24 +158,21 @@ v3d_switch_perfmon(struct v3d_dev *v3d, struct v3d_job *job)
 }
 
 static void
-v3d_job_start_stats(struct v3d_job *job, enum v3d_queue queue)
+v3d_stats_start(struct v3d_stats *stats, u64 now)
+{
+	raw_write_seqcount_begin(&stats->lock);
+	stats->start_ns = now;
+	raw_write_seqcount_end(&stats->lock);
+}
+
+static void
+v3d_job_start_stats(struct v3d_job *job)
 {
-	struct v3d_dev *v3d = job->v3d;
-	struct v3d_file_priv *file = job->file_priv;
-	struct v3d_stats *global_stats = v3d->queue[queue].stats;
-	struct v3d_stats *local_stats = file->stats[queue];
 	u64 now = local_clock();
 
 	preempt_disable();
-
-	raw_write_seqcount_begin(&local_stats->lock);
-	local_stats->start_ns = now;
-	raw_write_seqcount_end(&local_stats->lock);
-
-	raw_write_seqcount_begin(&global_stats->lock);
-	global_stats->start_ns = now;
-	raw_write_seqcount_end(&global_stats->lock);
-
+	v3d_stats_start(job->client_stats, now);
+	v3d_stats_start(job->global_stats, now);
 	preempt_enable();
 }
 
@@ -190,23 +187,13 @@ v3d_stats_update(struct v3d_stats *stats, u64 now)
 }
 
 void
-v3d_job_update_stats(struct v3d_job *job, enum v3d_queue q)
+v3d_job_update_stats(struct v3d_job *job)
 {
-	struct v3d_dev *v3d = job->v3d;
-	struct v3d_queue_state *queue = &v3d->queue[q];
-	struct v3d_stats *global_stats = queue->stats;
 	u64 now = local_clock();
 
 	preempt_disable();
-
-	/* Don't update the local stats if the file context has already closed */
-	spin_lock(&queue->queue_lock);
-	if (job->file_priv)
-		v3d_stats_update(job->file_priv->stats[q], now);
-	spin_unlock(&queue->queue_lock);
-
-	v3d_stats_update(global_stats, now);
-
+	v3d_stats_update(job->client_stats, now);
+	v3d_stats_update(job->global_stats, now);
 	preempt_enable();
 }
 
@@ -250,7 +237,7 @@ static struct dma_fence *v3d_bin_job_run(struct drm_sched_job *sched_job)
 	trace_v3d_submit_cl(dev, false, to_v3d_fence(fence)->seqno,
 			    job->start, job->end);
 
-	v3d_job_start_stats(&job->base, V3D_BIN);
+	v3d_job_start_stats(&job->base);
 	v3d_switch_perfmon(v3d, &job->base);
 
 	/* Set the current and end address of the control list.
@@ -304,7 +291,7 @@ static struct dma_fence *v3d_render_job_run(struct drm_sched_job *sched_job)
 	trace_v3d_submit_cl(dev, true, to_v3d_fence(fence)->seqno,
 			    job->start, job->end);
 
-	v3d_job_start_stats(&job->base, V3D_RENDER);
+	v3d_job_start_stats(&job->base);
 	v3d_switch_perfmon(v3d, &job->base);
 
 	/* XXX: Set the QCFG */
@@ -343,7 +330,7 @@ v3d_tfu_job_run(struct drm_sched_job *sched_job)
 
 	trace_v3d_submit_tfu(dev, to_v3d_fence(fence)->seqno);
 
-	v3d_job_start_stats(&job->base, V3D_TFU);
+	v3d_job_start_stats(&job->base);
 
 	V3D_WRITE(V3D_TFU_IIA(v3d->ver), job->args.iia);
 	V3D_WRITE(V3D_TFU_IIS(v3d->ver), job->args.iis);
@@ -393,7 +380,7 @@ v3d_csd_job_run(struct drm_sched_job *sched_job)
 
 	trace_v3d_submit_csd(dev, to_v3d_fence(fence)->seqno);
 
-	v3d_job_start_stats(&job->base, V3D_CSD);
+	v3d_job_start_stats(&job->base);
 	v3d_switch_perfmon(v3d, &job->base);
 
 	csd_cfg0_reg = V3D_CSD_QUEUED_CFG0(v3d->ver);
@@ -681,13 +668,13 @@ v3d_cpu_job_run(struct drm_sched_job *sched_job)
 		return NULL;
 	}
 
-	v3d_job_start_stats(&job->base, V3D_CPU);
+	v3d_job_start_stats(&job->base);
 	trace_v3d_cpu_job_begin(&v3d->drm, job->job_type);
 
 	cpu_job_function[job->job_type](job);
 
 	trace_v3d_cpu_job_end(&v3d->drm, job->job_type);
-	v3d_job_update_stats(&job->base, V3D_CPU);
+	v3d_job_update_stats(&job->base);
 
 	/* Synchronous operation, so no fence to wait on. */
 	return NULL;
@@ -699,11 +686,11 @@ v3d_cache_clean_job_run(struct drm_sched_job *sched_job)
 	struct v3d_job *job = to_v3d_job(sched_job);
 	struct v3d_dev *v3d = job->v3d;
 
-	v3d_job_start_stats(job, V3D_CACHE_CLEAN);
+	v3d_job_start_stats(job);
 
 	v3d_clean_caches(v3d);
 
-	v3d_job_update_stats(job, V3D_CACHE_CLEAN);
+	v3d_job_update_stats(job);
 
 	/* Synchronous operation, so no fence to wait on. */
 	return NULL;
diff --git a/drivers/gpu/drm/v3d/v3d_submit.c b/drivers/gpu/drm/v3d/v3d_submit.c
index 18f2bf1fe89face6ede3de465c80b63a6635511e..8f061b6a05c6aa76ea5513407ebf3c0ce80b8256 100644
--- a/drivers/gpu/drm/v3d/v3d_submit.c
+++ b/drivers/gpu/drm/v3d/v3d_submit.c
@@ -103,6 +103,9 @@ v3d_job_free(struct kref *ref)
 	if (job->perfmon)
 		v3d_perfmon_put(job->perfmon);
 
+	v3d_stats_put(job->client_stats);
+	v3d_stats_put(job->global_stats);
+
 	kfree(job);
 }
 
@@ -203,6 +206,9 @@ v3d_job_init(struct v3d_dev *v3d, struct drm_file *file_priv,
 
 	kref_init(&job->refcount);
 
+	job->client_stats = v3d_stats_get(v3d_priv->stats[queue]);
+	job->global_stats = v3d_stats_get(v3d->queue[queue].stats);
+
 	return 0;
 
 fail_deps:

-- 
2.53.0