linux-media.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH] dma-buf: add some lockdep asserts to the reservation object implementation
@ 2017-12-01 11:12 Lucas Stach
  2018-01-11 10:43 ` [Linaro-mm-sig] " Lucas Stach
  0 siblings, 1 reply; 4+ messages in thread
From: Lucas Stach @ 2017-12-01 11:12 UTC (permalink / raw)
  To: Sumit Semwal; +Cc: linux-media, dri-devel, linaro-mm-sig, kernel, patchwork-lst

This adds lockdep asserts to the reservation functions which state in their
documentation that obj->lock must be held. Allows builds with PROVE_LOCKING
enabled to check that the locking requirements are met.

Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
---
 drivers/dma-buf/reservation.c | 8 ++++++++
 include/linux/reservation.h   | 2 ++
 2 files changed, 10 insertions(+)

diff --git a/drivers/dma-buf/reservation.c b/drivers/dma-buf/reservation.c
index b44d9d7db347..accd398e2ea6 100644
--- a/drivers/dma-buf/reservation.c
+++ b/drivers/dma-buf/reservation.c
@@ -71,6 +71,8 @@ int reservation_object_reserve_shared(struct reservation_object *obj)
 	struct reservation_object_list *fobj, *old;
 	u32 max;
 
+	reservation_object_assert_held(obj);
+
 	old = reservation_object_get_list(obj);
 
 	if (old && old->shared_max) {
@@ -211,6 +213,8 @@ void reservation_object_add_shared_fence(struct reservation_object *obj,
 {
 	struct reservation_object_list *old, *fobj = obj->staged;
 
+	reservation_object_assert_held(obj);
+
 	old = reservation_object_get_list(obj);
 	obj->staged = NULL;
 
@@ -236,6 +240,8 @@ void reservation_object_add_excl_fence(struct reservation_object *obj,
 	struct reservation_object_list *old;
 	u32 i = 0;
 
+	reservation_object_assert_held(obj);
+
 	old = reservation_object_get_list(obj);
 	if (old)
 		i = old->shared_count;
@@ -276,6 +282,8 @@ int reservation_object_copy_fences(struct reservation_object *dst,
 	size_t size;
 	unsigned i;
 
+	reservation_object_assert_held(dst);
+
 	rcu_read_lock();
 	src_list = rcu_dereference(src->fence);
 
diff --git a/include/linux/reservation.h b/include/linux/reservation.h
index 21fc84d82d41..55e7318800fd 100644
--- a/include/linux/reservation.h
+++ b/include/linux/reservation.h
@@ -212,6 +212,8 @@ reservation_object_unlock(struct reservation_object *obj)
 static inline struct dma_fence *
 reservation_object_get_excl(struct reservation_object *obj)
 {
+	reservation_object_assert_held(obj);
+
 	return rcu_dereference_protected(obj->fence_excl,
 					 reservation_object_held(obj));
 }
-- 
2.11.0

^ permalink raw reply related	[flat|nested] 4+ messages in thread

* Re: [Linaro-mm-sig] [PATCH] dma-buf: add some lockdep asserts to the reservation object implementation
  2017-12-01 11:12 [PATCH] dma-buf: add some lockdep asserts to the reservation object implementation Lucas Stach
@ 2018-01-11 10:43 ` Lucas Stach
  2018-01-11 10:54   ` Christian König
  0 siblings, 1 reply; 4+ messages in thread
From: Lucas Stach @ 2018-01-11 10:43 UTC (permalink / raw)
  To: Sumit Semwal; +Cc: linaro-mm-sig, patchwork-lst, kernel, dri-devel, linux-media

Did this fall through the cracks over the holidays? It really has made
my work much easier while reworking some of the reservation object
handling in etnaviv and I think it might benefit others.

Regards,
Lucas

Am Freitag, den 01.12.2017, 12:12 +0100 schrieb Lucas Stach:
> This adds lockdep asserts to the reservation functions which state in their
> documentation that obj->lock must be held. Allows builds with PROVE_LOCKING
> enabled to check that the locking requirements are met.
> 
> > Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
> ---
>  drivers/dma-buf/reservation.c | 8 ++++++++
>  include/linux/reservation.h   | 2 ++
>  2 files changed, 10 insertions(+)
> 
> diff --git a/drivers/dma-buf/reservation.c b/drivers/dma-buf/reservation.c
> index b44d9d7db347..accd398e2ea6 100644
> --- a/drivers/dma-buf/reservation.c
> +++ b/drivers/dma-buf/reservation.c
> @@ -71,6 +71,8 @@ int reservation_object_reserve_shared(struct reservation_object *obj)
> >  	struct reservation_object_list *fobj, *old;
> >  	u32 max;
>  
> > +	reservation_object_assert_held(obj);
> +
> >  	old = reservation_object_get_list(obj);
>  
> >  	if (old && old->shared_max) {
> @@ -211,6 +213,8 @@ void reservation_object_add_shared_fence(struct reservation_object *obj,
>  {
> >  	struct reservation_object_list *old, *fobj = obj->staged;
>  
> > +	reservation_object_assert_held(obj);
> +
> >  	old = reservation_object_get_list(obj);
> >  	obj->staged = NULL;
>  
> @@ -236,6 +240,8 @@ void reservation_object_add_excl_fence(struct reservation_object *obj,
> >  	struct reservation_object_list *old;
> >  	u32 i = 0;
>  
> > +	reservation_object_assert_held(obj);
> +
> >  	old = reservation_object_get_list(obj);
> >  	if (old)
> >  		i = old->shared_count;
> @@ -276,6 +282,8 @@ int reservation_object_copy_fences(struct reservation_object *dst,
> >  	size_t size;
> >  	unsigned i;
>  
> > +	reservation_object_assert_held(dst);
> +
> >  	rcu_read_lock();
> >  	src_list = rcu_dereference(src->fence);
>  
> diff --git a/include/linux/reservation.h b/include/linux/reservation.h
> index 21fc84d82d41..55e7318800fd 100644
> --- a/include/linux/reservation.h
> +++ b/include/linux/reservation.h
> @@ -212,6 +212,8 @@ reservation_object_unlock(struct reservation_object *obj)
>  static inline struct dma_fence *
>  reservation_object_get_excl(struct reservation_object *obj)
>  {
> > +	reservation_object_assert_held(obj);
> +
> >  	return rcu_dereference_protected(obj->fence_excl,
> >  					 reservation_object_held(obj));
>  }

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [Linaro-mm-sig] [PATCH] dma-buf: add some lockdep asserts to the reservation object implementation
  2018-01-11 10:43 ` [Linaro-mm-sig] " Lucas Stach
@ 2018-01-11 10:54   ` Christian König
  2018-01-11 11:13     ` Lucas Stach
  0 siblings, 1 reply; 4+ messages in thread
From: Christian König @ 2018-01-11 10:54 UTC (permalink / raw)
  To: Lucas Stach, Sumit Semwal
  Cc: linaro-mm-sig, linux-media, dri-devel, kernel, patchwork-lst

Yeah, somehow missed that one.

The patch looks mostly good, except for reservation_object_get_excl().

For that one an RCU protection is usually sufficient, so annotating it 
with reservation_object_assert_held() sounds incorrect to me.

Regards,
Christian.

Am 11.01.2018 um 11:43 schrieb Lucas Stach:
> Did this fall through the cracks over the holidays? It really has made
> my work much easier while reworking some of the reservation object
> handling in etnaviv and I think it might benefit others.
>
> Regards,
> Lucas
>
> Am Freitag, den 01.12.2017, 12:12 +0100 schrieb Lucas Stach:
>> This adds lockdep asserts to the reservation functions which state in their
>> documentation that obj->lock must be held. Allows builds with PROVE_LOCKING
>> enabled to check that the locking requirements are met.
>>
>>> Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
>> ---
>>   drivers/dma-buf/reservation.c | 8 ++++++++
>>   include/linux/reservation.h   | 2 ++
>>   2 files changed, 10 insertions(+)
>>
>> diff --git a/drivers/dma-buf/reservation.c b/drivers/dma-buf/reservation.c
>> index b44d9d7db347..accd398e2ea6 100644
>> --- a/drivers/dma-buf/reservation.c
>> +++ b/drivers/dma-buf/reservation.c
>> @@ -71,6 +71,8 @@ int reservation_object_reserve_shared(struct reservation_object *obj)
>>>   	struct reservation_object_list *fobj, *old;
>>>   	u32 max;
>>   
>>> +	reservation_object_assert_held(obj);
>> +
>>>   	old = reservation_object_get_list(obj);
>>   
>>>   	if (old && old->shared_max) {
>> @@ -211,6 +213,8 @@ void reservation_object_add_shared_fence(struct reservation_object *obj,
>>   {
>>>   	struct reservation_object_list *old, *fobj = obj->staged;
>>   
>>> +	reservation_object_assert_held(obj);
>> +
>>>   	old = reservation_object_get_list(obj);
>>>   	obj->staged = NULL;
>>   
>> @@ -236,6 +240,8 @@ void reservation_object_add_excl_fence(struct reservation_object *obj,
>>>   	struct reservation_object_list *old;
>>>   	u32 i = 0;
>>   
>>> +	reservation_object_assert_held(obj);
>> +
>>>   	old = reservation_object_get_list(obj);
>>>   	if (old)
>>>   		i = old->shared_count;
>> @@ -276,6 +282,8 @@ int reservation_object_copy_fences(struct reservation_object *dst,
>>>   	size_t size;
>>>   	unsigned i;
>>   
>>> +	reservation_object_assert_held(dst);
>> +
>>>   	rcu_read_lock();
>>>   	src_list = rcu_dereference(src->fence);
>>   
>> diff --git a/include/linux/reservation.h b/include/linux/reservation.h
>> index 21fc84d82d41..55e7318800fd 100644
>> --- a/include/linux/reservation.h
>> +++ b/include/linux/reservation.h
>> @@ -212,6 +212,8 @@ reservation_object_unlock(struct reservation_object *obj)
>>   static inline struct dma_fence *
>>   reservation_object_get_excl(struct reservation_object *obj)
>>   {
>>> +	reservation_object_assert_held(obj);
>> +
>>>   	return rcu_dereference_protected(obj->fence_excl,
>>>   					 reservation_object_held(obj));
>>   }
> _______________________________________________
> dri-devel mailing list
> dri-devel@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/dri-devel

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [Linaro-mm-sig] [PATCH] dma-buf: add some lockdep asserts to the reservation object implementation
  2018-01-11 10:54   ` Christian König
@ 2018-01-11 11:13     ` Lucas Stach
  0 siblings, 0 replies; 4+ messages in thread
From: Lucas Stach @ 2018-01-11 11:13 UTC (permalink / raw)
  To: christian.koenig, Sumit Semwal
  Cc: linaro-mm-sig, linux-media, dri-devel, kernel, patchwork-lst

Am Donnerstag, den 11.01.2018, 11:54 +0100 schrieb Christian König:
> Yeah, somehow missed that one.
> 
> The patch looks mostly good, except for reservation_object_get_excl().
> 
> For that one an RCU protection is usually sufficient, so annotating it 
> with reservation_object_assert_held() sounds incorrect to me.

Ah, you are correct. I was confused about this one as
reservation_object_get_excl_rcu() exists and and the doc
above reservation_object_get_excl() states "The obj->lock must be
held.", which is misleading for the read-only case.

I'll send a v2 with that fixed.

Regards,
Lucas

> Regards,
> Christian.
> 
> Am 11.01.2018 um 11:43 schrieb Lucas Stach:
> > Did this fall through the cracks over the holidays? It really has made
> > my work much easier while reworking some of the reservation object
> > handling in etnaviv and I think it might benefit others.
> > 
> > Regards,
> > Lucas
> > 
> > Am Freitag, den 01.12.2017, 12:12 +0100 schrieb Lucas Stach:
> > > This adds lockdep asserts to the reservation functions which state in their
> > > documentation that obj->lock must be held. Allows builds with PROVE_LOCKING
> > > enabled to check that the locking requirements are met.
> > > 
> > > > Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
> > > 
> > > ---
> > >   drivers/dma-buf/reservation.c | 8 ++++++++
> > >   include/linux/reservation.h   | 2 ++
> > >   2 files changed, 10 insertions(+)
> > > 
> > > diff --git a/drivers/dma-buf/reservation.c b/drivers/dma-buf/reservation.c
> > > index b44d9d7db347..accd398e2ea6 100644
> > > --- a/drivers/dma-buf/reservation.c
> > > +++ b/drivers/dma-buf/reservation.c
> > > @@ -71,6 +71,8 @@ int reservation_object_reserve_shared(struct reservation_object *obj)
> > > > > > > >   	struct reservation_object_list *fobj, *old;
> > > >   	u32 max;
> > > 
> > >   
> > > > +	reservation_object_assert_held(obj);
> > > 
> > > +
> > > >   	old = reservation_object_get_list(obj);
> > > 
> > >   
> > > >   	if (old && old->shared_max) {
> > > 
> > > @@ -211,6 +213,8 @@ void reservation_object_add_shared_fence(struct reservation_object *obj,
> > >   {
> > > >   	struct reservation_object_list *old, *fobj = obj->staged;
> > > 
> > >   
> > > > +	reservation_object_assert_held(obj);
> > > 
> > > +
> > > > > > > >   	old = reservation_object_get_list(obj);
> > > >   	obj->staged = NULL;
> > > 
> > >   
> > > @@ -236,6 +240,8 @@ void reservation_object_add_excl_fence(struct reservation_object *obj,
> > > > > > > >   	struct reservation_object_list *old;
> > > >   	u32 i = 0;
> > > 
> > >   
> > > > +	reservation_object_assert_held(obj);
> > > 
> > > +
> > > > > > > >   	old = reservation_object_get_list(obj);
> > > > > > > >   	if (old)
> > > >   		i = old->shared_count;
> > > 
> > > @@ -276,6 +282,8 @@ int reservation_object_copy_fences(struct reservation_object *dst,
> > > > > > > >   	size_t size;
> > > >   	unsigned i;
> > > 
> > >   
> > > > +	reservation_object_assert_held(dst);
> > > 
> > > +
> > > > > > > >   	rcu_read_lock();
> > > >   	src_list = rcu_dereference(src->fence);
> > > 
> > >   
> > > diff --git a/include/linux/reservation.h b/include/linux/reservation.h
> > > index 21fc84d82d41..55e7318800fd 100644
> > > --- a/include/linux/reservation.h
> > > +++ b/include/linux/reservation.h
> > > @@ -212,6 +212,8 @@ reservation_object_unlock(struct reservation_object *obj)
> > >   static inline struct dma_fence *
> > >   reservation_object_get_excl(struct reservation_object *obj)
> > >   {
> > > > +	reservation_object_assert_held(obj);
> > > 
> > > +
> > > > > > > >   	return rcu_dereference_protected(obj->fence_excl,
> > > >   					 reservation_object_held(obj));
> > > 
> > >   }
> > 
> > _______________________________________________
> > dri-devel mailing list
> > dri-devel@lists.freedesktop.org
> > https://lists.freedesktop.org/mailman/listinfo/dri-devel
> 
> 

^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2018-01-12 10:27 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2017-12-01 11:12 [PATCH] dma-buf: add some lockdep asserts to the reservation object implementation Lucas Stach
2018-01-11 10:43 ` [Linaro-mm-sig] " Lucas Stach
2018-01-11 10:54   ` Christian König
2018-01-11 11:13     ` Lucas Stach

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).