arch,tile: Convert smp_mb__*()

Implement the new smp_mb__* ops as per the old ones.

Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Acked-by: Chris Metcalf <cmetcalf@tilera.com>
Link: http://lkml.kernel.org/n/tip-euuabnf5a3u23fy4fq8m3jcg@git.kernel.org
Cc: Akinobu Mita <akinobu.mita@gmail.com>
Cc: Chen Gang <gang.chen@asianux.com>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
diff --git a/arch/tile/include/asm/barrier.h b/arch/tile/include/asm/barrier.h
index b5a05d0..96a42ae 100644
--- a/arch/tile/include/asm/barrier.h
+++ b/arch/tile/include/asm/barrier.h
@@ -72,6 +72,20 @@
 #define mb()		fast_mb()
 #define iob()		fast_iob()
 
+#ifndef __tilegx__ /* 32 bit */
+/*
+ * We need to barrier before modifying the word, since the _atomic_xxx()
+ * routines just tns the lock and then read/modify/write of the word.
+ * But after the word is updated, the routine issues an "mf" before returning,
+ * and since it's a function call, we don't even need a compiler barrier.
+ */
+#define smp_mb__before_atomic()	smp_mb()
+#define smp_mb__after_atomic()	do { } while (0)
+#else /* 64 bit */
+#define smp_mb__before_atomic()	smp_mb()
+#define smp_mb__after_atomic()	smp_mb()
+#endif
+
 #include <asm-generic/barrier.h>
 
 #endif /* !__ASSEMBLY__ */