[PATCH 1/3] dma-buf/fence: make timeout handling in fence_default_wait consistent

[PATCH 1/3] dma-buf/fence: make timeout handling in fence_default_wait consistent

[Christian König]

From: Christian König <christian.koenig@amd.com>

Kernel functions taking a timeout usually return 1 on success even
when they get a zero timeout.

Signen-off-by: Christian König <christian.koenig@amd.com>
---
 drivers/dma-buf/fence.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/drivers/dma-buf/fence.c b/drivers/dma-buf/fence.c
index 4d51f9e..fb915ab 100644
--- a/drivers/dma-buf/fence.c
+++ b/drivers/dma-buf/fence.c
@@ -335,18 +335,20 @@ fence_default_wait_cb(struct fence *fence, struct fence_cb *cb)
  * @timeout:	[in]	timeout value in jiffies, or MAX_SCHEDULE_TIMEOUT
  *
  * Returns -ERESTARTSYS if interrupted, 0 if the wait timed out, or the
- * remaining timeout in jiffies on success.
+ * remaining timeout in jiffies on success. If timeout is zero the value one is
+ * returned if the fence is already signaled for consistency with other
+ * functions taking a jiffies timeout.
  */
 signed long
 fence_default_wait(struct fence *fence, bool intr, signed long timeout)
 {
 	struct default_wait_cb cb;
 	unsigned long flags;
-	signed long ret = timeout;
+	signed long ret = timeout ? timeout : 1;
 	bool was_set;
 
 	if (test_bit(FENCE_FLAG_SIGNALED_BIT, &fence->flags))
-		return timeout;
+		return ret;
 
 	spin_lock_irqsave(fence->lock, flags);
 
-- 
2.5.0

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

[PATCH 2/3] dma-buf/fence: revert "don't wait when specified timeout is zero"

[Christian König]

From: Christian König <christian.koenig@amd.com>

This reverts commit 847b19a39e4c9b5e74c40f0842c48b41664cb43c.

When we don't call the wait function software signaling might never be
activated. This can cause infinite polling loops with unreliable interrupt
driven hardware.

Signed-off-by: Christian König <christian.koenig@amd.com>
---
 drivers/dma-buf/fence.c | 3 ---
 1 file changed, 3 deletions(-)

diff --git a/drivers/dma-buf/fence.c b/drivers/dma-buf/fence.c
index fb915ab..64478f9 100644
--- a/drivers/dma-buf/fence.c
+++ b/drivers/dma-buf/fence.c
@@ -159,9 +159,6 @@ fence_wait_timeout(struct fence *fence, bool intr, signed long timeout)
 	if (WARN_ON(timeout < 0))
 		return -EINVAL;
 
-	if (timeout == 0)
-		return fence_is_signaled(fence);
-
 	trace_fence_wait_start(fence);
 	ret = fence->ops->wait(fence, intr, timeout);
 	trace_fence_wait_end(fence);
-- 
2.5.0

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

[PATCH 3/3] reservation: revert "wait only with non-zero timeout specified (v3)"

[Christian König]

From: Christian König <christian.koenig@amd.com>

This reverts commit fb8b7d2b9d80e1e71f379e57355936bd2b024be9.

Otherwise signaling might never be activated on the fences. This can
result in infinite waiting with hardware which has unreliable interrupts.

Signed-off-by: Christian König <christian.koenig@amd.com>
---
 drivers/dma-buf/reservation.c | 3 ---
 1 file changed, 3 deletions(-)

diff --git a/drivers/dma-buf/reservation.c b/drivers/dma-buf/reservation.c
index 9566a62..bb97b04 100644
--- a/drivers/dma-buf/reservation.c
+++ b/drivers/dma-buf/reservation.c
@@ -381,9 +381,6 @@ long reservation_object_wait_timeout_rcu(struct reservation_object *obj,
 	unsigned seq, shared_count, i = 0;
 	long ret = timeout;
 
-	if (!timeout)
-		return reservation_object_test_signaled_rcu(obj, wait_all);
-
 retry:
 	fence = NULL;
 	shared_count = 0;
-- 
2.5.0

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: [PATCH 1/3] dma-buf/fence: make timeout handling in fence_default_wait consistent

[zhoucm1]

On 2016年10月17日 21:55, Christian König wrote:
> From: Christian König <christian.koenig@amd.com>
>
> Kernel functions taking a timeout usually return 1 on success even
> when they get a zero timeout.
>
> Signen-off-by: Christian König <christian.koenig@amd.com>

The series is Reviewed-by: Chunming Zhou <david1.zhou@amd.com>

> ---
>   drivers/dma-buf/fence.c | 8 +++++---
>   1 file changed, 5 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/dma-buf/fence.c b/drivers/dma-buf/fence.c
> index 4d51f9e..fb915ab 100644
> --- a/drivers/dma-buf/fence.c
> +++ b/drivers/dma-buf/fence.c
> @@ -335,18 +335,20 @@ fence_default_wait_cb(struct fence *fence, struct fence_cb *cb)
>    * @timeout:	[in]	timeout value in jiffies, or MAX_SCHEDULE_TIMEOUT
>    *
>    * Returns -ERESTARTSYS if interrupted, 0 if the wait timed out, or the
> - * remaining timeout in jiffies on success.
> + * remaining timeout in jiffies on success. If timeout is zero the value one is
> + * returned if the fence is already signaled for consistency with other
> + * functions taking a jiffies timeout.
>    */
>   signed long
>   fence_default_wait(struct fence *fence, bool intr, signed long timeout)
>   {
>   	struct default_wait_cb cb;
>   	unsigned long flags;
> -	signed long ret = timeout;
> +	signed long ret = timeout ? timeout : 1;
>   	bool was_set;
>   
>   	if (test_bit(FENCE_FLAG_SIGNALED_BIT, &fence->flags))
> -		return timeout;
> +		return ret;
>   
>   	spin_lock_irqsave(fence->lock, flags);
>   

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel