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
Cc: daniele.ceraolospurio@intel.com, carlos.santa@intel.com
Subject: [RFC PATCH 11/13] drm/xe: Enable deadlines on hardware fences
Date: Wed, 24 Dec 2025 17:17:32 -0800	[thread overview]
Message-ID: <20251225011734.341683-12-matthew.brost@intel.com> (raw)
In-Reply-To: <20251225011734.341683-1-matthew.brost@intel.com>

Implement the set_deadline vfunc on hardware fences, which, with GuC
submission, allows priority and frequency boosts for queues that have
fences at risk of missing a deadline.

Signed-off-by: Matthew Brost <matthew.brost@intel.com>
---
 drivers/gpu/drm/xe/xe_hw_fence.c | 13 +++++++++++++
 1 file changed, 13 insertions(+)

diff --git a/drivers/gpu/drm/xe/xe_hw_fence.c b/drivers/gpu/drm/xe/xe_hw_fence.c
index 8b884691718f..9452ea2be62f 100644
--- a/drivers/gpu/drm/xe/xe_hw_fence.c
+++ b/drivers/gpu/drm/xe/xe_hw_fence.c
@@ -66,6 +66,7 @@ static void hw_fence_irq_run_cb(struct irq_work *work)
 			if (dma_fence_is_signaled_locked(dma_fence)) {
 				trace_xe_hw_fence_signal(fence);
 				list_del_init(&fence->irq_link);
+				fence->q = NULL;
 				dma_fence_put(dma_fence);
 			}
 		}
@@ -93,6 +94,7 @@ void xe_hw_fence_irq_finish(struct xe_hw_fence_irq *irq)
 		spin_lock_irqsave(&irq->lock, flags);
 		list_for_each_entry_safe(fence, next, &irq->pending, irq_link) {
 			list_del_init(&fence->irq_link);
+			fence->q = NULL;
 			XE_WARN_ON(dma_fence_check_and_signal_locked(&fence->dma));
 			dma_fence_put(&fence->dma);
 		}
@@ -191,12 +193,23 @@ static void xe_hw_fence_release(struct dma_fence *dma_fence)
 	call_rcu(&dma_fence->rcu, fence_free);
 }
 
+static void xe_hw_fence_set_deadline(struct dma_fence *dma_fence,
+				     ktime_t deadline)
+{
+	struct xe_hw_fence *fence = to_xe_hw_fence(dma_fence);
+
+	guard(spinlock_irqsave)(dma_fence->lock);
+	if (fence->q)
+		fence->q->ops->set_deadline(fence->q, dma_fence, deadline);
+}
+
 static const struct dma_fence_ops xe_hw_fence_ops = {
 	.get_driver_name = xe_hw_fence_get_driver_name,
 	.get_timeline_name = xe_hw_fence_get_timeline_name,
 	.enable_signaling = xe_hw_fence_enable_signaling,
 	.signaled = xe_hw_fence_signaled,
 	.release = xe_hw_fence_release,
+	.set_deadline = xe_hw_fence_set_deadline,
 };
 
 /**
-- 
2.34.1


  parent reply	other threads:[~2025-12-25  1:17 UTC|newest]

Thread overview: 18+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-12-25  1:17 [RFC PATCH 00/13] Fence deadlines in Xe Matthew Brost
2025-12-25  1:17 ` [RFC PATCH 01/13] drm/xe: Add dedicated message lock Matthew Brost
2025-12-25  1:17 ` [RFC PATCH 02/13] drm/xe: Add EXEC_QUEUE_FLAG_CAP_SYS_NICE Matthew Brost
2025-12-25  1:17 ` [RFC PATCH 03/13] drm/xe: Store exec queue in hardware fence Matthew Brost
2025-12-25  1:17 ` [RFC PATCH 04/13] drm/xe: Add deadline exec queue vfuncs Matthew Brost
2025-12-25  1:17 ` [RFC PATCH 05/13] drm/xe: Export to_xe_hw_fence Matthew Brost
2025-12-25  1:17 ` [RFC PATCH 06/13] drm/xe: Add deadline manager Matthew Brost
2025-12-25  1:17 ` [RFC PATCH 07/13] drm/xe: Add deadline manager to user exec queues Matthew Brost
2025-12-25  1:17 ` [RFC PATCH 08/13] drm/xe: Stub out execlists deadline vfuncs as NOPs Matthew Brost
2025-12-25  1:17 ` [RFC PATCH 09/13] drm/xe: Make scheduler message lock IRQ-safe Matthew Brost
2025-12-25  1:17 ` [RFC PATCH 10/13] drm/xe: Implement GuC submission backend ops for deadlines Matthew Brost
2025-12-25  1:17 ` Matthew Brost [this message]
2025-12-25  1:17 ` [RFC PATCH 12/13] drm/xe: Add CONFIG_DRM_XE_DEADLINE_WINDOW Matthew Brost
2025-12-25  1:17 ` [RFC PATCH 13/13] drm/xe: Add exec queue deadline trace points Matthew Brost
2025-12-25  1:25 ` ✗ CI.checkpatch: warning for Fence deadlines in Xe Patchwork
2025-12-25  1:26 ` ✓ CI.KUnit: success " Patchwork
2025-12-25  2:01 ` ✗ Xe.CI.BAT: failure " Patchwork
2025-12-25  3:02 ` ✓ Xe.CI.Full: success " 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=20251225011734.341683-12-matthew.brost@intel.com \
    --to=matthew.brost@intel.com \
    --cc=carlos.santa@intel.com \
    --cc=daniele.ceraolospurio@intel.com \
    --cc=intel-xe@lists.freedesktop.org \
    /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