Intel-XE Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: Matthew Brost <matthew.brost@intel.com>
To: intel-xe@lists.freedesktop.org, dri-devel@lists.freedesktop.org
Cc: niranjana.vishwanathapura@intel.com, christian.koenig@amd.com,
	pstanner@redhat.com, dakr@kernel.org
Subject: [PATCH v9 1/2] drm/sched: Add several job helpers to avoid drivers touching scheduler state
Date: Tue,  9 Dec 2025 12:00:38 -0800	[thread overview]
Message-ID: <20251209200039.1366764-2-matthew.brost@intel.com> (raw)
In-Reply-To: <20251209200039.1366764-1-matthew.brost@intel.com>

In the past, drivers used to reach into scheduler internals—this must
end because it makes it difficult to change scheduler internals, as
driver-side code must also be updated.

Add helpers to check if the scheduler is stopped and to query a job’s
signaled state to avoid reaching into scheduler internals. These are
expected to be used driver-side in recovery and debug flows.

Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
---

v4:
 - Reorder patch to first in series (Niranjana)
 - Also check parent fence for signaling (Niranjana)
v8:
 - Fixup kernel doc (Philipp)
 - Move chnage log (Philipp)
 - Better commit (Philipp)
 - Drop inlines (Philipp)
---
 drivers/gpu/drm/scheduler/sched_main.c | 36 ++++++++++++++++++++++++--
 include/drm/gpu_scheduler.h            |  2 ++
 2 files changed, 36 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/scheduler/sched_main.c b/drivers/gpu/drm/scheduler/sched_main.c
index 1d4f1b822e7b..5f08719a35f5 100644
--- a/drivers/gpu/drm/scheduler/sched_main.c
+++ b/drivers/gpu/drm/scheduler/sched_main.c
@@ -344,7 +344,7 @@ drm_sched_rq_select_entity_fifo(struct drm_gpu_scheduler *sched,
  */
 static void drm_sched_run_job_queue(struct drm_gpu_scheduler *sched)
 {
-	if (!READ_ONCE(sched->pause_submit))
+	if (!drm_sched_is_stopped(sched))
 		queue_work(sched->submit_wq, &sched->work_run_job);
 }
 
@@ -354,7 +354,7 @@ static void drm_sched_run_job_queue(struct drm_gpu_scheduler *sched)
  */
 static void drm_sched_run_free_queue(struct drm_gpu_scheduler *sched)
 {
-	if (!READ_ONCE(sched->pause_submit))
+	if (!drm_sched_is_stopped(sched))
 		queue_work(sched->submit_wq, &sched->work_free_job);
 }
 
@@ -1567,3 +1567,35 @@ void drm_sched_wqueue_start(struct drm_gpu_scheduler *sched)
 	queue_work(sched->submit_wq, &sched->work_free_job);
 }
 EXPORT_SYMBOL(drm_sched_wqueue_start);
+
+/**
+ * drm_sched_is_stopped() - Checks whether drm_sched is stopped
+ * @sched: DRM scheduler
+ *
+ * Return: true if sched is stopped, false otherwise
+ */
+bool drm_sched_is_stopped(struct drm_gpu_scheduler *sched)
+{
+	return READ_ONCE(sched->pause_submit);
+}
+EXPORT_SYMBOL(drm_sched_is_stopped);
+
+/**
+ * drm_sched_job_is_signaled() - DRM scheduler job is signaled
+ * @job: DRM scheduler job
+ *
+ * Determine if DRM scheduler job is signaled. DRM scheduler should be stopped
+ * to obtain a stable snapshot of state. Both parent fence (hardware fence) and
+ * finished fence (software fence) are checked to determine signaling state.
+ *
+ * Return: true if job is signaled, false otherwise
+ */
+bool drm_sched_job_is_signaled(struct drm_sched_job *job)
+{
+	struct drm_sched_fence *s_fence = job->s_fence;
+
+	WARN_ON(!drm_sched_is_stopped(job->sched));
+	return (s_fence->parent && dma_fence_is_signaled(s_fence->parent)) ||
+		dma_fence_is_signaled(&s_fence->finished);
+}
+EXPORT_SYMBOL(drm_sched_job_is_signaled);
diff --git a/include/drm/gpu_scheduler.h b/include/drm/gpu_scheduler.h
index fb88301b3c45..86b6075ce799 100644
--- a/include/drm/gpu_scheduler.h
+++ b/include/drm/gpu_scheduler.h
@@ -645,6 +645,7 @@ void drm_sched_stop(struct drm_gpu_scheduler *sched, struct drm_sched_job *bad);
 void drm_sched_start(struct drm_gpu_scheduler *sched, int errno);
 void drm_sched_resubmit_jobs(struct drm_gpu_scheduler *sched);
 void drm_sched_fault(struct drm_gpu_scheduler *sched);
+bool drm_sched_is_stopped(struct drm_gpu_scheduler *sched);
 
 struct drm_gpu_scheduler *
 drm_sched_pick_best(struct drm_gpu_scheduler **sched_list,
@@ -674,6 +675,7 @@ bool drm_sched_job_has_dependency(struct drm_sched_job *job,
 				  struct dma_fence *fence);
 void drm_sched_job_cleanup(struct drm_sched_job *job);
 void drm_sched_increase_karma(struct drm_sched_job *bad);
+bool drm_sched_job_is_signaled(struct drm_sched_job *job);
 
 static inline bool drm_sched_invalidate_job(struct drm_sched_job *s_job,
 					    int threshold)
-- 
2.34.1


  reply	other threads:[~2025-12-09 20:00 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-12-09 20:00 [PATCH v9 0/2] Add DRM sched support functions to avoid driver side layering violations Matthew Brost
2025-12-09 20:00 ` Matthew Brost [this message]
2025-12-09 20:00 ` [PATCH v9 2/2] drm/sched: Add pending job list iterator Matthew Brost
2025-12-10 10:38 ` ✗ CI.checkpatch: warning for Add DRM sched support functions to avoid driver side layering violations (rev2) Patchwork
2025-12-10 10:39 ` ✓ CI.KUnit: success " Patchwork
2025-12-10 11:50 ` ✓ Xe.CI.BAT: " Patchwork
2025-12-10 16:33 ` ✗ Xe.CI.Full: failure " Patchwork

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20251209200039.1366764-2-matthew.brost@intel.com \
    --to=matthew.brost@intel.com \
    --cc=christian.koenig@amd.com \
    --cc=dakr@kernel.org \
    --cc=dri-devel@lists.freedesktop.org \
    --cc=intel-xe@lists.freedesktop.org \
    --cc=niranjana.vishwanathapura@intel.com \
    --cc=pstanner@redhat.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox