[Intel-gfx] [PATCH 1/2] dma-buf/sync-file: fix logic error in new fence merge code

[Intel-gfx] [PATCH 1/2] dma-buf/sync-file: fix logic error in new fence merge code

[Christian König]

When the array is empty because everything is signaled we can't use
add_fence() to add something because that would filter the signaled
fence again.

Signed-off-by: Christian König <christian.koenig@amd.com>
Fixes: 519f490db07e ("dma-buf/sync-file: fix warning about fence containers")
---
 drivers/dma-buf/sync_file.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/dma-buf/sync_file.c b/drivers/dma-buf/sync_file.c
index b8dea4ec123b..514d213261df 100644
--- a/drivers/dma-buf/sync_file.c
+++ b/drivers/dma-buf/sync_file.c
@@ -262,7 +262,7 @@ static struct sync_file *sync_file_merge(const char *name, struct sync_file *a,
 	}
 
 	if (index == 0)
-		add_fence(fences, &index, dma_fence_get_stub());
+		fences[index++] = dma_fence_get_stub();
 
 	if (num_fences > index) {
 		struct dma_fence **tmp;
-- 
2.25.1

[Intel-gfx] [PATCH 2/2] dma-buf: handle empty dma_fence_arrays gracefully

[Christian König]

A bug inside the new sync-file merge code created empty dma_fence_array instances.

Warn about that and handle those without crashing.

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

diff --git a/drivers/dma-buf/dma-fence-array.c b/drivers/dma-buf/dma-fence-array.c
index 52b85d292383..5c8a7084577b 100644
--- a/drivers/dma-buf/dma-fence-array.c
+++ b/drivers/dma-buf/dma-fence-array.c
@@ -159,6 +159,8 @@ struct dma_fence_array *dma_fence_array_create(int num_fences,
 	struct dma_fence_array *array;
 	size_t size = sizeof(*array);
 
+	WARN_ON(!num_fences || !fences);
+
 	/* Allocate the callback structures behind the array. */
 	size += num_fences * sizeof(struct dma_fence_array_cb);
 	array = kzalloc(size, GFP_KERNEL);
@@ -231,6 +233,9 @@ struct dma_fence *dma_fence_array_first(struct dma_fence *head)
 	if (!array)
 		return head;
 
+	if (!array->num_fences)
+		return NULL;
+
 	return array->fences[0];
 }
 EXPORT_SYMBOL(dma_fence_array_first);
-- 
2.25.1

Re: [Intel-gfx] [PATCH 2/2] dma-buf: handle empty dma_fence_arrays gracefully

[Daniel Vetter]

On Tue, Mar 29, 2022 at 09:00:01AM +0200, Christian König wrote:
> A bug inside the new sync-file merge code created empty dma_fence_array instances.
> 
> Warn about that and handle those without crashing.
> 
> Signed-off-by: Christian König <christian.koenig@amd.com>

> ---
>  drivers/dma-buf/dma-fence-array.c | 5 +++++
>  1 file changed, 5 insertions(+)
> 
> diff --git a/drivers/dma-buf/dma-fence-array.c b/drivers/dma-buf/dma-fence-array.c
> index 52b85d292383..5c8a7084577b 100644
> --- a/drivers/dma-buf/dma-fence-array.c
> +++ b/drivers/dma-buf/dma-fence-array.c
> @@ -159,6 +159,8 @@ struct dma_fence_array *dma_fence_array_create(int num_fences,
>  	struct dma_fence_array *array;
>  	size_t size = sizeof(*array);
>  
> +	WARN_ON(!num_fences || !fences);

WARN_ON and then dying randomly is kinda not nice, I'd wrap this in an

if (WARN_ON)
	return NULL;

with that: Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>

> +
>  	/* Allocate the callback structures behind the array. */
>  	size += num_fences * sizeof(struct dma_fence_array_cb);
>  	array = kzalloc(size, GFP_KERNEL);
> @@ -231,6 +233,9 @@ struct dma_fence *dma_fence_array_first(struct dma_fence *head)
>  	if (!array)
>  		return head;
>  
> +	if (!array->num_fences)
> +		return NULL;
> +
>  	return array->fences[0];
>  }
>  EXPORT_SYMBOL(dma_fence_array_first);
> -- 
> 2.25.1
> 

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch

Re: [Intel-gfx] [PATCH 2/2] dma-buf: handle empty dma_fence_arrays gracefully

[Daniel Vetter]

On Tue, Mar 29, 2022 at 10:48:10AM +0200, Daniel Vetter wrote:
> On Tue, Mar 29, 2022 at 09:00:01AM +0200, Christian König wrote:
> > A bug inside the new sync-file merge code created empty dma_fence_array instances.
> > 
> > Warn about that and handle those without crashing.
> > 
> > Signed-off-by: Christian König <christian.koenig@amd.com>
> 
> > ---
> >  drivers/dma-buf/dma-fence-array.c | 5 +++++
> >  1 file changed, 5 insertions(+)
> > 
> > diff --git a/drivers/dma-buf/dma-fence-array.c b/drivers/dma-buf/dma-fence-array.c
> > index 52b85d292383..5c8a7084577b 100644
> > --- a/drivers/dma-buf/dma-fence-array.c
> > +++ b/drivers/dma-buf/dma-fence-array.c
> > @@ -159,6 +159,8 @@ struct dma_fence_array *dma_fence_array_create(int num_fences,
> >  	struct dma_fence_array *array;
> >  	size_t size = sizeof(*array);
> >  
> > +	WARN_ON(!num_fences || !fences);
> 
> WARN_ON and then dying randomly is kinda not nice, I'd wrap this in an
> 
> if (WARN_ON)
> 	return NULL;
> 
> with that: Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>

Uh strike that, you handle it gracefully with the check below.

> 
> > +
> >  	/* Allocate the callback structures behind the array. */
> >  	size += num_fences * sizeof(struct dma_fence_array_cb);
> >  	array = kzalloc(size, GFP_KERNEL);
> > @@ -231,6 +233,9 @@ struct dma_fence *dma_fence_array_first(struct dma_fence *head)
> >  	if (!array)
> >  		return head;
> >  

Maybe add a comment here that this is just defensive programming, like

	/* No fences isn't allowed and splats in create, but be defensive */

Either way Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> I guess,
coffee not quite yet working.
-Daniel

> > +	if (!array->num_fences)
> > +		return NULL;
> > +
> >  	return array->fences[0];
> >  }
> >  EXPORT_SYMBOL(dma_fence_array_first);
> > -- 
> > 2.25.1
> > 
> 
> -- 
> Daniel Vetter
> Software Engineer, Intel Corporation
> http://blog.ffwll.ch

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch

Re: [Intel-gfx] [Linaro-mm-sig] [PATCH 1/2] dma-buf/sync-file: fix logic error in new fence merge code

[Thomas Hellström]

For the series,

Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>


On 3/29/22 09:00, Christian König wrote:
> When the array is empty because everything is signaled we can't use
> add_fence() to add something because that would filter the signaled
> fence again.
>
> Signed-off-by: Christian König <christian.koenig@amd.com>
> Fixes: 519f490db07e ("dma-buf/sync-file: fix warning about fence containers")
> ---
>   drivers/dma-buf/sync_file.c | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/dma-buf/sync_file.c b/drivers/dma-buf/sync_file.c
> index b8dea4ec123b..514d213261df 100644
> --- a/drivers/dma-buf/sync_file.c
> +++ b/drivers/dma-buf/sync_file.c
> @@ -262,7 +262,7 @@ static struct sync_file *sync_file_merge(const char *name, struct sync_file *a,
>   	}
>   
>   	if (index == 0)
> -		add_fence(fences, &index, dma_fence_get_stub());
> +		fences[index++] = dma_fence_get_stub();
>   
>   	if (num_fences > index) {
>   		struct dma_fence **tmp;

Re: [Intel-gfx] [PATCH 1/2] dma-buf/sync-file: fix logic error in new fence merge code

[Daniel Vetter]

On Tue, Mar 29, 2022 at 09:00:00AM +0200, Christian König wrote:
> When the array is empty because everything is signaled we can't use
> add_fence() to add something because that would filter the signaled
> fence again.
> 
> Signed-off-by: Christian König <christian.koenig@amd.com>
> Fixes: 519f490db07e ("dma-buf/sync-file: fix warning about fence containers")
> ---
>  drivers/dma-buf/sync_file.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/dma-buf/sync_file.c b/drivers/dma-buf/sync_file.c
> index b8dea4ec123b..514d213261df 100644
> --- a/drivers/dma-buf/sync_file.c
> +++ b/drivers/dma-buf/sync_file.c
> @@ -262,7 +262,7 @@ static struct sync_file *sync_file_merge(const char *name, struct sync_file *a,
>  	}
>  
>  	if (index == 0)
> -		add_fence(fences, &index, dma_fence_get_stub());
> +		fences[index++] = dma_fence_get_stub();

Uh idiot me wondered about this and somehow convinced myself that adding a
random fences instead of a singalled stub is ok. I should have checked
more carefully :-/

Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>

>  
>  	if (num_fences > index) {
>  		struct dma_fence **tmp;
> -- 
> 2.25.1
> 

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch

Re: [Intel-gfx] [PATCH 1/2] dma-buf/sync-file: fix logic error in new fence merge code

[Christian König]

Am 29.03.22 um 10:46 schrieb Daniel Vetter:
> On Tue, Mar 29, 2022 at 09:00:00AM +0200, Christian König wrote:
>> When the array is empty because everything is signaled we can't use
>> add_fence() to add something because that would filter the signaled
>> fence again.
>>
>> Signed-off-by: Christian König <christian.koenig@amd.com>
>> Fixes: 519f490db07e ("dma-buf/sync-file: fix warning about fence containers")
>> ---
>>   drivers/dma-buf/sync_file.c | 2 +-
>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/drivers/dma-buf/sync_file.c b/drivers/dma-buf/sync_file.c
>> index b8dea4ec123b..514d213261df 100644
>> --- a/drivers/dma-buf/sync_file.c
>> +++ b/drivers/dma-buf/sync_file.c
>> @@ -262,7 +262,7 @@ static struct sync_file *sync_file_merge(const char *name, struct sync_file *a,
>>   	}
>>   
>>   	if (index == 0)
>> -		add_fence(fences, &index, dma_fence_get_stub());
>> +		fences[index++] = dma_fence_get_stub();
> Uh idiot me wondered about this and somehow convinced myself that adding a
> random fences instead of a singalled stub is ok. I should have checked
> more carefully :-/

Well idiot me wanted to clean that up fully and move all of that into 
sync_file_set_fence(), but then I though "Na, that patch is for 
next-fixes, don't make it too complicate" and then changed the code 
anyway just out of habit.

Cheers,
Christian.

>
> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
>
>>   
>>   	if (num_fences > index) {
>>   		struct dma_fence **tmp;
>> -- 
>> 2.25.1
>>

Re: [Intel-gfx] [PATCH 1/2] dma-buf/sync-file: fix logic error in new fence merge code

[Das, Nirmoy]

I finally managed to find a machine and tested this series. If it is not 
too late

The series is Tested-by: Nirmoy Das <nirmoy.das@intel.com>

On 3/29/2022 9:00 AM, Christian König wrote:
> When the array is empty because everything is signaled we can't use
> add_fence() to add something because that would filter the signaled
> fence again.
>
> Signed-off-by: Christian König <christian.koenig@amd.com>
> Fixes: 519f490db07e ("dma-buf/sync-file: fix warning about fence containers")
> ---
>   drivers/dma-buf/sync_file.c | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/dma-buf/sync_file.c b/drivers/dma-buf/sync_file.c
> index b8dea4ec123b..514d213261df 100644
> --- a/drivers/dma-buf/sync_file.c
> +++ b/drivers/dma-buf/sync_file.c
> @@ -262,7 +262,7 @@ static struct sync_file *sync_file_merge(const char *name, struct sync_file *a,
>   	}
>   
>   	if (index == 0)
> -		add_fence(fences, &index, dma_fence_get_stub());
> +		fences[index++] = dma_fence_get_stub();
>   
>   	if (num_fences > index) {
>   		struct dma_fence **tmp;

[Intel-gfx] ✗ Fi.CI.BAT: failure for series starting with [1/2] dma-buf/sync-file: fix logic error in new fence merge code

[Patchwork]

[-- Attachment #1: Type: text/plain, Size: 668 bytes --]

== Series Details ==

Series: series starting with [1/2] dma-buf/sync-file: fix logic error in new fence merge code
URL   : https://patchwork.freedesktop.org/series/101923/
State : failure

== Summary ==

Applying: dma-buf/sync-file: fix logic error in new fence merge code
Using index info to reconstruct a base tree...
M	drivers/dma-buf/sync_file.c
Falling back to patching base and 3-way merge...
No changes -- Patch already applied.
Applying: dma-buf: handle empty dma_fence_arrays gracefully
Using index info to reconstruct a base tree...
M	drivers/dma-buf/dma-fence-array.c
Falling back to patching base and 3-way merge...
No changes -- Patch already applied.



[-- Attachment #2: Type: text/html, Size: 1195 bytes --]