X-Git-Url: http://git.liburcu.org/?p=urcu.git;a=blobdiff_plain;f=include%2Furcu%2Farch%2Fgeneric.h;fp=include%2Furcu%2Farch%2Fgeneric.h;h=f9e115f14e190c6da199b6d005d9515641b4be56;hp=c02a6a9930d0b7126c24ac633b5995cfe18e6546;hb=72d24c88ee075c8368fe57ee7e8fc66d99ce7e39;hpb=865024e9245c55286122413aaf7366dc06a00f27

diff --git a/include/urcu/arch/generic.h b/include/urcu/arch/generic.h
index c02a6a9..f9e115f 100644
--- a/include/urcu/arch/generic.h
+++ b/include/urcu/arch/generic.h
@@ -31,6 +31,28 @@ extern "C" {
  * GCC builtins) as well as cmm_rmb and cmm_wmb (defaulting to cmm_mb).
  */
 
+#ifdef CONFIG_RCU_USE_ATOMIC_BUILTINS
+
+# ifndef cmm_smp_mb
+#  define cmm_smp_mb() __atomic_thread_fence(__ATOMIC_SEQ_CST)
+# endif
+
+#endif	/* CONFIG_RCU_USE_ATOMIC_BUILTINS */
+
+
+/*
+ * cmm_mb() expands to __sync_synchronize() instead of __atomic_thread_fence
+ * with SEQ_CST because the former "issues a full memory barrier" while the
+ * latter "acts as a synchronization fence between threads" which is too weak
+ * for what we want, for example with I/O devices.
+ *
+ * Even though sync_synchronize seems to be an alias for a sequential consistent
+ * atomic thread fence on every architecture on GCC and Clang, this assumption
+ * might be untrue in future.  Therefore, the definitions above are used to
+ * ensure correct behavior in the future.
+ *
+ * The above defintions are quoted from the GCC manual.
+ */
 #ifndef cmm_mb
 #define cmm_mb()    __sync_synchronize()
 #endif