From: Guo Ren <guoren@xxxxxxxxxxxxxxxxx>
Optimize the performance of cmpxchg by using more fine-grained
acquire/release barriers.
Signed-off-by: Guo Ren <guoren@xxxxxxxxxxxxxxxxx>
Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
Cc: Arnd Bergmann <arnd@xxxxxxxx>
Cc: Paul E. McKenney <paulmck@xxxxxxxxxx>
---
arch/csky/include/asm/cmpxchg.h | 27 +++++++++++++++++----------
1 file changed, 17 insertions(+), 10 deletions(-)
diff --git a/arch/csky/include/asm/cmpxchg.h b/arch/csky/include/asm/cmpxchg.h
index 8922453..dabc8e4 100644
--- a/arch/csky/include/asm/cmpxchg.h
+++ b/arch/csky/include/asm/cmpxchg.h
@@ -3,12 +3,12 @@
#ifndef __ASM_CSKY_CMPXCHG_H
#define __ASM_CSKY_CMPXCHG_H
-#ifdef CONFIG_CPU_HAS_LDSTEX
+#ifdef CONFIG_SMP
#include <asm/barrier.h>
extern void __bad_xchg(void);
-#define __xchg(new, ptr, size) \
+#define __xchg_relaxed(new, ptr, size) \
({ \
__typeof__(ptr) __ptr = (ptr); \
__typeof__(new) __new = (new); \
@@ -16,7 +16,6 @@ extern void __bad_xchg(void);
unsigned long tmp; \
switch (size) { \
case 4: \
- smp_mb(); \
asm volatile ( \
"1: ldex.w %0, (%3) \n" \
" mov %1, %2 \n" \
@@ -25,7 +24,6 @@ extern void __bad_xchg(void);
: "=&r" (__ret), "=&r" (tmp) \
: "r" (__new), "r"(__ptr) \
:); \
- smp_mb(); \
break; \
default: \
__bad_xchg(); \
@@ -33,9 +31,10 @@ extern void __bad_xchg(void);
__ret; \
})
-#define xchg(ptr, x) (__xchg((x), (ptr), sizeof(*(ptr))))
+#define xchg_relaxed(ptr, x) \
+ (__xchg_relaxed((x), (ptr), sizeof(*(ptr))))
-#define __cmpxchg(ptr, old, new, size) \
+#define __cmpxchg_relaxed(ptr, old, new, size) \
({ \
__typeof__(ptr) __ptr = (ptr); \
__typeof__(new) __new = (new); \
@@ -44,7 +43,6 @@ extern void __bad_xchg(void);
__typeof__(*(ptr)) __ret; \
switch (size) { \
case 4: \
- smp_mb(); \
asm volatile ( \
"1: ldex.w %0, (%3) \n" \
" cmpne %0, %4 \n" \
@@ -56,7 +54,6 @@ extern void __bad_xchg(void);
: "=&r" (__ret), "=&r" (__tmp) \
: "r" (__new), "r"(__ptr), "r"(__old) \
:); \
- smp_mb(); \
break; \
default: \
__bad_xchg(); \
@@ -64,8 +61,18 @@ extern void __bad_xchg(void);
__ret; \
})
-#define cmpxchg(ptr, o, n) \
- (__cmpxchg((ptr), (o), (n), sizeof(*(ptr))))
+#define cmpxchg_relaxed(ptr, o, n) \
+ (__cmpxchg_relaxed((ptr), (o), (n), sizeof(*(ptr))))
+
+#define cmpxchg(ptr, o, n) \
+({ \
+ __typeof__(*(ptr)) __ret; \
+ __smp_release_fence(); \
+ __ret = cmpxchg_relaxed(ptr, o, n); \
+ __smp_acquire_fence(); \
+ __ret; \
+})
+
#else
#include <asm-generic/cmpxchg.h>
#endif
--
2.7.4
