ARC: add smp barriers around atomics per Documentation/atomic_ops.txt

 - arch_spin_lock/unlock were lacking the ACQUIRE/RELEASE barriers
   Since ARCv2 only provides load/load, store/store and all/all, we need
   the full barrier

 - LLOCK/SCOND based atomics, bitops, cmpxchg, which return modified
   values were lacking the explicit smp barriers.

 - Non LLOCK/SCOND varaints don't need the explicit barriers since that
   is implicity provided by the spin locks used to implement the
   critical section (the spin lock barriers in turn are also fixed in
   this commit as explained above

Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: stable@vger.kernel.org
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Vineet Gupta <vgupta@synopsys.com>

1 files changed, 32 insertions, 0 deletions

diff --git a/arch/arc/include/asm/spinlock.h b/arch/arc/include/asm/spinlock.h
index b6a8c2dfbe6e..e1651df6a93d 100644
--- a/arch/arc/include/asm/spinlock.h
+++ b/arch/arc/include/asm/spinlock.h
@@ -22,24 +22,46 @@ static inline void arch_spin_lock(arch_spinlock_t *lock)
 {
 	unsigned int tmp = __ARCH_SPIN_LOCK_LOCKED__;
 
+	/*
+	 * This smp_mb() is technically superfluous, we only need the one
+	 * after the lock for providing the ACQUIRE semantics.
+	 * However doing the "right" thing was regressing hackbench
+	 * so keeping this, pending further investigation
+	 */
+	smp_mb();
+
 	__asm__ __volatile__(
 	"1:	ex  %0, [%1]		\n"
 	"	breq  %0, %2, 1b	\n"
 	: "+&r" (tmp)
 	: "r"(&(lock->slock)), "ir"(__ARCH_SPIN_LOCK_LOCKED__)
 	: "memory");
+
+	/*
+	 * ACQUIRE barrier to ensure load/store after taking the lock
+	 * don't "bleed-up" out of the critical section (leak-in is allowed)
+	 * http://www.spinics.net/lists/kernel/msg2010409.html
+	 *
+	 * ARCv2 only has load-load, store-store and all-all barrier
+	 * thus need the full all-all barrier
+	 */
+	smp_mb();
 }
 
 static inline int arch_spin_trylock(arch_spinlock_t *lock)
 {
 	unsigned int tmp = __ARCH_SPIN_LOCK_LOCKED__;
 
+	smp_mb();
+
 	__asm__ __volatile__(
 	"1:	ex  %0, [%1]		\n"
 	: "+r" (tmp)
 	: "r"(&(lock->slock))
 	: "memory");
 
+	smp_mb();
+
 	return (tmp == __ARCH_SPIN_LOCK_UNLOCKED__);
 }
 
@@ -47,12 +69,22 @@ static inline void arch_spin_unlock(arch_spinlock_t *lock)
 {
 	unsigned int tmp = __ARCH_SPIN_LOCK_UNLOCKED__;
 
+	/*
+	 * RELEASE barrier: given the instructions avail on ARCv2, full barrier
+	 * is the only option
+	 */
+	smp_mb();
+
 	__asm__ __volatile__(
 	"	ex  %0, [%1]		\n"
 	: "+r" (tmp)
 	: "r"(&(lock->slock))
 	: "memory");
 
+	/*
+	 * superfluous, but keeping for now - see pairing version in
+	 * arch_spin_lock above
+	 */
 	smp_mb();
 }