[PATCH] metag: fix memory barriers

[PATCH] metag: fix memory barriers

[Mikulas Patocka]

Volatile access doesn't really imply the compiler barrier. Volatile access
is only ordered with respect to other volatile accesses, it isn't ordered
with respect to general memory accesses. Gcc may reorder memory accesses
around volatile access, as we can see in this simple example (if we
compile it with optimization, both increments of *b will be collapsed to
just one):

void fn(volatile int *a, long *b)
{
	(*b)++;
	*a = 10;
	(*b)++;
}

Consequently, we need the compiler barrier after a write to the volatile
variable, to make sure that the compiler doesn't reorder the volatile
write with something else.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Cc: stable@vger.kernel.org

---
 arch/metag/include/asm/barrier.h |    3 +++
 1 file changed, 3 insertions(+)

Index: linux-3.15-rc4/arch/metag/include/asm/barrier.h
===================================================================
--- linux-3.15-rc4.orig/arch/metag/include/asm/barrier.h	2014-05-08 16:23:17.000000000 +0200
+++ linux-3.15-rc4/arch/metag/include/asm/barrier.h	2014-05-08 16:24:01.000000000 +0200
@@ -15,6 +15,7 @@ static inline void wr_fence(void)
 	volatile int *flushptr = (volatile int *) LINSYSEVENT_WR_FENCE;
 	barrier();
 	*flushptr = 0;
+	barrier();
 }
 
 #else /* CONFIG_METAG_META21 */
@@ -35,6 +36,7 @@ static inline void wr_fence(void)
 	*flushptr = 0;
 	*flushptr = 0;
 	*flushptr = 0;
+	barrier();
 }
 
 #endif /* !CONFIG_METAG_META21 */
@@ -68,6 +70,7 @@ static inline void fence(void)
 	volatile int *flushptr = (volatile int *) LINSYSEVENT_WR_ATOMIC_UNLOCK;
 	barrier();
 	*flushptr = 0;
+	barrier();
 }
 #define smp_mb()        fence()
 #define smp_rmb()       fence()

Re: [PATCH] metag: fix memory barriers

[Peter Zijlstra]

[-- Attachment #1: Type: text/plain, Size: 870 bytes --]

On Thu, May 08, 2014 at 03:51:37PM -0400, Mikulas Patocka wrote:
> Volatile access doesn't really imply the compiler barrier. Volatile access
> is only ordered with respect to other volatile accesses, it isn't ordered
> with respect to general memory accesses. Gcc may reorder memory accesses
> around volatile access, as we can see in this simple example (if we
> compile it with optimization, both increments of *b will be collapsed to
> just one):
> 
> void fn(volatile int *a, long *b)
> {
> 	(*b)++;
> 	*a = 10;
> 	(*b)++;
> }
> 
> Consequently, we need the compiler barrier after a write to the volatile
> variable, to make sure that the compiler doesn't reorder the volatile
> write with something else.
> 
> Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
> Cc: stable@vger.kernel.org

Acked-by: Peter Zijlstra <peterz@infradead.org>

[-- Attachment #2: Type: application/pgp-signature, Size: 836 bytes --]

Re: [PATCH] metag: fix memory barriers

[James Hogan]

[-- Attachment #1: Type: text/plain, Size: 1928 bytes --]

On 08/05/14 20:51, Mikulas Patocka wrote:
> Volatile access doesn't really imply the compiler barrier. Volatile access
> is only ordered with respect to other volatile accesses, it isn't ordered
> with respect to general memory accesses. Gcc may reorder memory accesses
> around volatile access, as we can see in this simple example (if we
> compile it with optimization, both increments of *b will be collapsed to
> just one):
> 
> void fn(volatile int *a, long *b)
> {
> 	(*b)++;
> 	*a = 10;
> 	(*b)++;
> }
> 
> Consequently, we need the compiler barrier after a write to the volatile
> variable, to make sure that the compiler doesn't reorder the volatile
> write with something else.
> 
> Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
> Cc: stable@vger.kernel.org

Applied for v3.15, thanks!

Cheers
James

> 
> ---
>  arch/metag/include/asm/barrier.h |    3 +++
>  1 file changed, 3 insertions(+)
> 
> Index: linux-3.15-rc4/arch/metag/include/asm/barrier.h
> ===================================================================
> --- linux-3.15-rc4.orig/arch/metag/include/asm/barrier.h	2014-05-08 16:23:17.000000000 +0200
> +++ linux-3.15-rc4/arch/metag/include/asm/barrier.h	2014-05-08 16:24:01.000000000 +0200
> @@ -15,6 +15,7 @@ static inline void wr_fence(void)
>  	volatile int *flushptr = (volatile int *) LINSYSEVENT_WR_FENCE;
>  	barrier();
>  	*flushptr = 0;
> +	barrier();
>  }
>  
>  #else /* CONFIG_METAG_META21 */
> @@ -35,6 +36,7 @@ static inline void wr_fence(void)
>  	*flushptr = 0;
>  	*flushptr = 0;
>  	*flushptr = 0;
> +	barrier();
>  }
>  
>  #endif /* !CONFIG_METAG_META21 */
> @@ -68,6 +70,7 @@ static inline void fence(void)
>  	volatile int *flushptr = (volatile int *) LINSYSEVENT_WR_ATOMIC_UNLOCK;
>  	barrier();
>  	*flushptr = 0;
> +	barrier();
>  }
>  #define smp_mb()        fence()
>  #define smp_rmb()       fence()
> 


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]