[v1] dma-fence: Remove return code of dma_fence_signal() et al.

[PATCH 3/6] drm/gpu/xe: Ignore dma_fenc_signal() return code

Posted by Philipp Stanner 5 days, 9 hours ago

The return code of dma_fence_signal() is not really useful as there is
nothing reasonable to do if a fence was already signaled. That return
code shall be removed from the kernel.

Ignore dma_fence_signal()'s return code.

Suggested-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Philipp Stanner <phasta@kernel.org>
---
 drivers/gpu/drm/xe/xe_hw_fence.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/drivers/gpu/drm/xe/xe_hw_fence.c b/drivers/gpu/drm/xe/xe_hw_fence.c
index b2a0c46dfcd4..959b30dde724 100644
--- a/drivers/gpu/drm/xe/xe_hw_fence.c
+++ b/drivers/gpu/drm/xe/xe_hw_fence.c
@@ -85,7 +85,6 @@ void xe_hw_fence_irq_finish(struct xe_hw_fence_irq *irq)
 {
 	struct xe_hw_fence *fence, *next;
 	unsigned long flags;
-	int err;
 	bool tmp;
 
 	if (XE_WARN_ON(!list_empty(&irq->pending))) {
@@ -93,9 +92,9 @@ 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);
-			err = dma_fence_signal_locked(&fence->dma);
+			XE_WARN_ON(dma_fence_test_signaled_flag(&fence->dma));
+			dma_fence_signal_locked(&fence->dma);
 			dma_fence_put(&fence->dma);
-			XE_WARN_ON(err);
 		}
 		spin_unlock_irqrestore(&irq->lock, flags);
 		dma_fence_end_signalling(tmp);
-- 
2.49.0

Re: [PATCH 3/6] drm/gpu/xe: Ignore dma_fenc_signal() return code

Posted by Andi Shyti 4 days, 23 hours ago

Hi Philipp,

in the subject /dma_fenc_signal/dma_fence_signal/

> @@ -85,7 +85,6 @@ void xe_hw_fence_irq_finish(struct xe_hw_fence_irq *irq)
>  {
>  	struct xe_hw_fence *fence, *next;
>  	unsigned long flags;
> -	int err;
>  	bool tmp;
>  
>  	if (XE_WARN_ON(!list_empty(&irq->pending))) {
> @@ -93,9 +92,9 @@ 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);
> -			err = dma_fence_signal_locked(&fence->dma);

why don't we do

XE_WARN_ON(dma_fence_signal_locked(..))

instead?

Andi

> +			XE_WARN_ON(dma_fence_test_signaled_flag(&fence->dma));
> +			dma_fence_signal_locked(&fence->dma);
>  			dma_fence_put(&fence->dma);
> -			XE_WARN_ON(err);
>  		}
>  		spin_unlock_irqrestore(&irq->lock, flags);
>  		dma_fence_end_signalling(tmp);
> -- 
> 2.49.0
>

Re: [PATCH 3/6] drm/gpu/xe: Ignore dma_fenc_signal() return code

Posted by Matthew Brost 4 days, 22 hours ago

On Wed, Nov 26, 2025 at 11:56:57PM +0100, Andi Shyti wrote:
> Hi Philipp,
> 
> in the subject /dma_fenc_signal/dma_fence_signal/
> 
> > @@ -85,7 +85,6 @@ void xe_hw_fence_irq_finish(struct xe_hw_fence_irq *irq)
> >  {
> >  	struct xe_hw_fence *fence, *next;
> >  	unsigned long flags;
> > -	int err;
> >  	bool tmp;
> >  
> >  	if (XE_WARN_ON(!list_empty(&irq->pending))) {
> > @@ -93,9 +92,9 @@ 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);
> > -			err = dma_fence_signal_locked(&fence->dma);
> 
> why don't we do
> 
> XE_WARN_ON(dma_fence_signal_locked(..))
> 

IIRC the above statement can compile out. So the patch looks correct to me.

Matt

> instead?
> 
> Andi
> 
> > +			XE_WARN_ON(dma_fence_test_signaled_flag(&fence->dma));
> > +			dma_fence_signal_locked(&fence->dma);
> >  			dma_fence_put(&fence->dma);
> > -			XE_WARN_ON(err);
> >  		}
> >  		spin_unlock_irqrestore(&irq->lock, flags);
> >  		dma_fence_end_signalling(tmp);
> > -- 
> > 2.49.0
> >

Re: [PATCH 3/6] drm/gpu/xe: Ignore dma_fenc_signal() return code

Posted by Andi Shyti 4 days, 9 hours ago

Hi Matt,

On Wed, Nov 26, 2025 at 03:56:32PM -0800, Matthew Brost wrote:
> On Wed, Nov 26, 2025 at 11:56:57PM +0100, Andi Shyti wrote:
> > > @@ -85,7 +85,6 @@ void xe_hw_fence_irq_finish(struct xe_hw_fence_irq *irq)
> > > @@ -93,9 +92,9 @@ 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);
> > > -			err = dma_fence_signal_locked(&fence->dma);
> > 
> > why don't we do
> > 
> > XE_WARN_ON(dma_fence_signal_locked(..))
> > 
> 
> IIRC the above statement can compile out. So the patch looks correct to me.

you have defined XE_WARN_ON as WARN_ON that should always
evaluate the content and, depending on the configuration, it
prints the logs or not.

What I don't like from this patch is that we end up checking
twice for the DMA_FENCE_FLAG_SIGNALED_BIT bit.

Thanks,
Andi

Re: [PATCH 3/6] drm/gpu/xe: Ignore dma_fenc_signal() return code

Posted by Philipp Stanner 4 days, 8 hours ago

On Thu, 2025-11-27 at 14:37 +0100, Andi Shyti wrote:
> Hi Matt,
> 
> On Wed, Nov 26, 2025 at 03:56:32PM -0800, Matthew Brost wrote:
> > On Wed, Nov 26, 2025 at 11:56:57PM +0100, Andi Shyti wrote:
> > > > @@ -85,7 +85,6 @@ void xe_hw_fence_irq_finish(struct xe_hw_fence_irq *irq)
> > > > @@ -93,9 +92,9 @@ 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);
> > > > -			err = dma_fence_signal_locked(&fence->dma);
> > > 
> > > why don't we do
> > > 
> > > XE_WARN_ON(dma_fence_signal_locked(..))

because it's impossible because the series is about removing the return
codes from the dma_fence_signal_* functions.

> > > 
> > 
> > IIRC the above statement can compile out. So the patch looks correct to me.
> 
> you have defined XE_WARN_ON as WARN_ON that should always
> evaluate the content and, depending on the configuration, it
> prints the logs or not.
> 
> What I don't like from this patch is that we end up checking
> twice for the DMA_FENCE_FLAG_SIGNALED_BIT bit.

Depends on what you mean by "we". The Xe code checks it only once, with
dma_fence_test_signaled_flag(). The dma_fence backend checks it yet
again, as it always does, to avoid signaling a signaled fence.

That's not racy here, however, because the fence lock is already being
held, as evidenced by the current usage of dma_fence_signal_locked().

P.

Re: [PATCH 3/6] drm/gpu/xe: Ignore dma_fenc_signal() return code

Posted by Andi Shyti 4 days, 5 hours ago

On Thu, Nov 27, 2025 at 02:51:39PM +0100, Philipp Stanner wrote:
> On Thu, 2025-11-27 at 14:37 +0100, Andi Shyti wrote:
> > On Wed, Nov 26, 2025 at 03:56:32PM -0800, Matthew Brost wrote:
> > > On Wed, Nov 26, 2025 at 11:56:57PM +0100, Andi Shyti wrote:
> > > > > @@ -85,7 +85,6 @@ void xe_hw_fence_irq_finish(struct xe_hw_fence_irq *irq)
> > > > > @@ -93,9 +92,9 @@ 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);
> > > > > -			err = dma_fence_signal_locked(&fence->dma);
> > > > 
> > > > why don't we do
> > > > 
> > > > XE_WARN_ON(dma_fence_signal_locked(..))
> 
> because it's impossible because the series is about removing the return
> codes from the dma_fence_signal_* functions.

oh yes, the last patch. Sorry, I went on reviewing and lost the
final target from sight.

> > > > 
> > > 
> > > IIRC the above statement can compile out. So the patch looks correct to me.
> > 
> > you have defined XE_WARN_ON as WARN_ON that should always
> > evaluate the content and, depending on the configuration, it
> > prints the logs or not.
> > 
> > What I don't like from this patch is that we end up checking
> > twice for the DMA_FENCE_FLAG_SIGNALED_BIT bit.
> 
> Depends on what you mean by "we". The Xe code checks it only once, with
> dma_fence_test_signaled_flag(). The dma_fence backend checks it yet
> again, as it always does, to avoid signaling a signaled fence.
> 
> That's not racy here, however, because the fence lock is already being
> held, as evidenced by the current usage of dma_fence_signal_locked().

I haven't said it's racy, I just didn't like that we are testing
for the DMA_FENCE_FLAG_SIGNALED_BIT twice. On the other hand,
with dma_fence_signal_locked() being void, I wouldn't know how to
do it better. So that I guess it's fine.

Andi

Re: [PATCH 3/6] drm/gpu/xe: Ignore dma_fenc_signal() return code

Posted by Matthew Brost 5 days, 5 hours ago

On Wed, Nov 26, 2025 at 02:19:12PM +0100, Philipp Stanner wrote:
> The return code of dma_fence_signal() is not really useful as there is
> nothing reasonable to do if a fence was already signaled. That return
> code shall be removed from the kernel.
> 
> Ignore dma_fence_signal()'s return code.
> 
> Suggested-by: Christian König <christian.koenig@amd.com>
> Signed-off-by: Philipp Stanner <phasta@kernel.org>
> ---
>  drivers/gpu/drm/xe/xe_hw_fence.c | 5 ++---
>  1 file changed, 2 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/gpu/drm/xe/xe_hw_fence.c b/drivers/gpu/drm/xe/xe_hw_fence.c
> index b2a0c46dfcd4..959b30dde724 100644
> --- a/drivers/gpu/drm/xe/xe_hw_fence.c
> +++ b/drivers/gpu/drm/xe/xe_hw_fence.c
> @@ -85,7 +85,6 @@ void xe_hw_fence_irq_finish(struct xe_hw_fence_irq *irq)
>  {
>  	struct xe_hw_fence *fence, *next;
>  	unsigned long flags;
> -	int err;
>  	bool tmp;
>  
>  	if (XE_WARN_ON(!list_empty(&irq->pending))) {
> @@ -93,9 +92,9 @@ 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);
> -			err = dma_fence_signal_locked(&fence->dma);
> +			XE_WARN_ON(dma_fence_test_signaled_flag(&fence->dma));
> +			dma_fence_signal_locked(&fence->dma);

If you also want fix Xe to use dma_fence_test_signaled_flag in all
places where we manually check DMA_FENCE_FLAG_SIGNALED_BIT, I'm not
going to complain. Ofc I can also do this in follow if patch when patch
#1 merges too.

Anyways this patch LGTM:
Reviewed-by: Matthew Brost <matthew.brost@intel.com>

>  			dma_fence_put(&fence->dma);
> -			XE_WARN_ON(err);
>  		}
>  		spin_unlock_irqrestore(&irq->lock, flags);
>  		dma_fence_end_signalling(tmp);
> -- 
> 2.49.0
>