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: christian.koenig@amd.com, pstanner@redhat.com, dakr@kernel.org
Subject: [RFC PATCH v2 2/4] drm/sched: Add several job helpers to avoid drivers touching scheduler state
Date: Fri,  3 Oct 2025 13:11:54 -0700	[thread overview]
Message-ID: <20251003201156.1995113-3-matthew.brost@intel.com> (raw)
In-Reply-To: <20251003201156.1995113-1-matthew.brost@intel.com>

Add helpers to find first pending job, pending job count, and a jobs
signaled state. Expected to used driver side on recovery and debug
flows.

Signed-off-by: Matthew Brost <matthew.brost@intel.com>
---
 include/drm/gpu_scheduler.h | 52 +++++++++++++++++++++++++++++++++++++
 1 file changed, 52 insertions(+)

diff --git a/include/drm/gpu_scheduler.h b/include/drm/gpu_scheduler.h
index bb49d8b715eb..e1e264c2043e 100644
--- a/include/drm/gpu_scheduler.h
+++ b/include/drm/gpu_scheduler.h
@@ -762,4 +762,56 @@ static inline void
 		list_for_each_entry((__job), &(__sched)->pending_list, list)	\
 			for_each_if(!(__entity) || (__job)->entity == (__entity))
 
+/**
+ * drm_sched_first_pending_job() - DRM scheduler first pending job
+ * @sched: DRM scheduler
+ *
+ * Find DRM scheduler first pending job, drivers should only call this function
+ * when scheduler is stopped or job can immediately disappear resulting in a
+ * UAF.
+ *
+ * Return: First pending job or NULL
+ */
+static inline struct drm_sched_job *
+drm_sched_first_pending_job(struct drm_gpu_scheduler *sched)
+{
+	WARN_ON(!READ_ONCE(sched->pause_submit));
+	guard(spinlock)(&sched->job_list_lock);
+	return list_first_entry_or_null(&sched->pending_list,
+					struct drm_sched_job, list);
+}
+
+/**
+ * drm_sched_pending_job_count() - DRM scheduler pending job count
+ * @sched: DRM scheduler
+ *
+ * Determine DRM scheduler pending job count. If scheduler if not stopped, this
+ * value can immediately change, thus drivers must guard against using to count
+ * to anything memory unsafe. Use with caution.
+ *
+ * Return: Number of pending jobs
+ */
+static inline int drm_sched_pending_job_count(struct drm_gpu_scheduler *sched)
+{
+	guard(spinlock)(&sched->job_list_lock);
+	return list_count_nodes(&sched->pending_list);
+}
+
+/**
+ * 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.
+ *
+ * Return: True if job is signaled, False otherwise
+ */
+static inline bool drm_sched_job_is_signaled(struct drm_sched_job *job)
+{
+	struct drm_sched_fence *s_fence = job->s_fence;
+
+	WARN_ON(!READ_ONCE(job->sched->pause_submit));
+	return s_fence->parent && dma_fence_is_signaled(s_fence->parent);
+}
+
 #endif
-- 
2.34.1


  parent reply	other threads:[~2025-10-03 20:12 UTC|newest]

Thread overview: 11+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-10-03 20:11 [RFC PATCH v2 0/4] Fix DRM scheduler layering violations in Xe Matthew Brost
2025-10-03 20:11 ` [RFC PATCH v2 1/4] drm/sched: Add pending job list iterator Matthew Brost
2025-10-06  9:19   ` Jani Nikula
2025-10-06 13:17     ` Matthew Brost
2025-10-03 20:11 ` Matthew Brost [this message]
2025-10-03 20:11 ` [RFC PATCH v2 3/4] drm/xe: Add dedicated message lock Matthew Brost
2025-10-03 20:11 ` [RFC PATCH v2 4/4] drm/xe: Stop abusing DRM scheduler internals Matthew Brost
2025-10-03 20:17 ` ✗ CI.checkpatch: warning for Fix DRM scheduler layering violations in Xe (rev2) Patchwork
2025-10-03 20:19 ` ✓ CI.KUnit: success " Patchwork
2025-10-03 20:54 ` ✓ Xe.CI.BAT: " Patchwork
2025-10-03 22:52 ` ✓ Xe.CI.Full: " 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=20251003201156.1995113-3-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=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