* [PATCH v2] drm/radeon: restore hardware polling in fence_is_signaled to fix performance regression
@ 2026-07-30 1:56 2564278112
2026-08-05 11:43 ` Christian König
0 siblings, 1 reply; 2+ messages in thread
From: 2564278112 @ 2026-07-30 1:56 UTC (permalink / raw)
To: alexander.deucher
Cc: christian.koenig, airlied, simona, amd-gfx, dri-devel,
linux-kernel, 2564278112, Wang Jiang
From: Wang Jiang <jiangwang@kylinos.cn>
Commit 9eb00b5f5697b ("drm/radeon: delete radeon_fence_process in
is_signaled, no deadlock") removed the hardware status check from
radeon_fence_is_signaled() to fix a self-deadlock caused by
wake_up_all(&rdev->fence_queue) being called with the fence queue
lock held.
However, removing the hardware poll entirely causes significant
performance regression (e.g. glxgears FPS drop) because the signaled
check becomes purely passive — it only reads the cached last_seq
without probing the GPU, so completed GPU work is not detected in
time, causing unnecessary CPU stalls in sync-heavy workloads.
Fix this by calling radeon_fence_read() directly in
radeon_fence_is_signaled() to poll the hardware fence counter
without updating last_seq or calling wake_up_all(). This restores
timely fence detection while avoiding the original deadlock, since
last_seq update and waiter wakeup remain solely in the interrupt
handler's responsibility.
Fixes: 9eb00b5f5697b ("drm/radeon: delete radeon_fence_process in is_signaled, no deadlock")
Signed-off-by: Wang Jiang <jiangwang@kylinos.cn>
---
drivers/gpu/drm/radeon/radeon_fence.c | 12 ++++++++++++
1 file changed, 12 insertions(+)
diff --git a/drivers/gpu/drm/radeon/radeon_fence.c b/drivers/gpu/drm/radeon/radeon_fence.c
index 02a40e4750c7..047ec5a1d194 100644
--- a/drivers/gpu/drm/radeon/radeon_fence.c
+++ b/drivers/gpu/drm/radeon/radeon_fence.c
@@ -360,6 +360,18 @@ static bool radeon_fence_is_signaled(struct dma_fence *f)
if (atomic64_read(&rdev->fence_drv[ring].last_seq) >= seq)
return true;
+ /*
+ * Poll hardware directly without updating last_seq.
+ * This allows immediate detection of signaled fences (performance)
+ * while leaving last_seq update and wake_up_all() to the interrupt
+ * handler, avoiding:
+ * 1. Deadlock from calling wake_up_all() with fence lock held
+ * 2. Wake event stealing by advancing last_seq without waking waiters
+ * 3. rw_semaphore usage in atomic/irq context (unsafe on PREEMPT_RT)
+ */
+ if (radeon_fence_read(rdev, ring) >= (u32)seq)
+ return true;
+
return false;
}
--
2.25.1
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: [PATCH v2] drm/radeon: restore hardware polling in fence_is_signaled to fix performance regression
2026-07-30 1:56 [PATCH v2] drm/radeon: restore hardware polling in fence_is_signaled to fix performance regression 2564278112
@ 2026-08-05 11:43 ` Christian König
0 siblings, 0 replies; 2+ messages in thread
From: Christian König @ 2026-08-05 11:43 UTC (permalink / raw)
To: 2564278112, alexander.deucher
Cc: airlied, simona, amd-gfx, dri-devel, linux-kernel, Wang Jiang
On 7/30/26 03:56, 2564278112@qq.com wrote:
> From: Wang Jiang <jiangwang@kylinos.cn>
>
> Commit 9eb00b5f5697b ("drm/radeon: delete radeon_fence_process in
> is_signaled, no deadlock") removed the hardware status check from
> radeon_fence_is_signaled() to fix a self-deadlock caused by
> wake_up_all(&rdev->fence_queue) being called with the fence queue
> lock held.
>
> However, removing the hardware poll entirely causes significant
> performance regression (e.g. glxgears FPS drop) because the signaled
> check becomes purely passive — it only reads the cached last_seq
> without probing the GPU, so completed GPU work is not detected in
> time, causing unnecessary CPU stalls in sync-heavy workloads.
>
> Fix this by calling radeon_fence_read() directly in
> radeon_fence_is_signaled() to poll the hardware fence counter
> without updating last_seq or calling wake_up_all(). This restores
> timely fence detection while avoiding the original deadlock, since
> last_seq update and waiter wakeup remain solely in the interrupt
> handler's responsibility.
>
> Fixes: 9eb00b5f5697b ("drm/radeon: delete radeon_fence_process in is_signaled, no deadlock")
> Signed-off-by: Wang Jiang <jiangwang@kylinos.cn>
> ---
> drivers/gpu/drm/radeon/radeon_fence.c | 12 ++++++++++++
> 1 file changed, 12 insertions(+)
>
> diff --git a/drivers/gpu/drm/radeon/radeon_fence.c b/drivers/gpu/drm/radeon/radeon_fence.c
> index 02a40e4750c7..047ec5a1d194 100644
> --- a/drivers/gpu/drm/radeon/radeon_fence.c
> +++ b/drivers/gpu/drm/radeon/radeon_fence.c
> @@ -360,6 +360,18 @@ static bool radeon_fence_is_signaled(struct dma_fence *f)
> if (atomic64_read(&rdev->fence_drv[ring].last_seq) >= seq)
> return true;
>
> + /*
> + * Poll hardware directly without updating last_seq.
> + * This allows immediate detection of signaled fences (performance)
> + * while leaving last_seq update and wake_up_all() to the interrupt
> + * handler, avoiding:
> + * 1. Deadlock from calling wake_up_all() with fence lock held
> + * 2. Wake event stealing by advancing last_seq without waking waiters
> + * 3. rw_semaphore usage in atomic/irq context (unsafe on PREEMPT_RT)
> + */
> + if (radeon_fence_read(rdev, ring) >= (u32)seq)
> + return true;
> +
That check is not correct, you need to take wrap around into account.
Try using __dma_fence_is_later() instead.
Regards,
Christian.
> return false;
> }
>
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2026-08-05 11:43 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2026-07-30 1:56 [PATCH v2] drm/radeon: restore hardware polling in fence_is_signaled to fix performance regression 2564278112
2026-08-05 11:43 ` Christian König
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®