Faster atomic loads and stores on windows

1 files changed, 37 insertions, 6 deletions

diff --git a/include/atomic/generic-msvc.h b/include/atomic/generic-msvc.h
index 9fe2cddd7ae..56fa4f66fcd 100644
--- a/include/atomic/generic-msvc.h
+++ b/include/atomic/generic-msvc.h
@@ -61,19 +61,43 @@ static inline int64 my_atomic_add64(int64 volatile *a, int64 v)
   return (int64)InterlockedExchangeAdd64((volatile LONGLONG*)a, (LONGLONG)v);
 }
 
+
+/*
+  According to MSDN:
+
+  Simple reads and writes to properly-aligned 32-bit variables are atomic
+  operations.
+  ...
+  Simple reads and writes to properly aligned 64-bit variables are atomic on
+  64-bit Windows. Reads and writes to 64-bit values are not guaranteed to be
+  atomic on 32-bit Windows.
+
+  https://msdn.microsoft.com/en-us/library/windows/desktop/ms684122(v=vs.85).aspx
+*/
+
 static inline int32 my_atomic_load32(int32 volatile *a)
 {
-  return (int32)InterlockedCompareExchange((volatile LONG *)a, 0, 0);
+  int32 value= *a;
+  MemoryBarrier();
+  return value;
 }
 
 static inline int64 my_atomic_load64(int64 volatile *a)
 {
-  return (int64)InterlockedCompareExchange64((volatile LONGLONG *)a, 0, 0);
+#ifdef _M_X64
+  int64 value= *a;
+  MemoryBarrier();
+  return value;
+#else
+  return (int64) InterlockedCompareExchange64((volatile LONGLONG *) a, 0, 0);
+#endif
 }
 
 static inline void* my_atomic_loadptr(void * volatile *a)
 {
-  return InterlockedCompareExchangePointer(a, 0, 0);
+  void *value= *a;
+  MemoryBarrier();
+  return value;
 }
 
 static inline int32 my_atomic_fas32(int32 volatile *a, int32 v)
@@ -93,17 +117,24 @@ static inline void * my_atomic_fasptr(void * volatile *a, void * v)
 
 static inline void my_atomic_store32(int32 volatile *a, int32 v)
 {
-  (void)InterlockedExchange((volatile LONG*)a, v);
+  MemoryBarrier();
+  *a= v;
 }
 
 static inline void my_atomic_store64(int64 volatile *a, int64 v)
 {
-  (void)InterlockedExchange64((volatile LONGLONG*)a, v);
+#ifdef _M_X64
+  MemoryBarrier();
+  *a= v;
+#else
+  (void) InterlockedExchange64((volatile LONGLONG *) a, v);
+#endif
 }
 
 static inline void my_atomic_storeptr(void * volatile *a, void *v)
 {
-  (void)InterlockedExchangePointer(a, v);
+  MemoryBarrier();
+  *a= v;
 }