Allow atomic memory count update with C11 builtins

From mailing list post https://groups.google.com/forum/#!topic/redis-db/QLjiQe4D7LA

In zmalloc.c the following primitives are currently used
to synchronize access to single global variable:
__sync_add_and_fetch
__sync_sub_and_fetch

In some architectures such as powerpc these primitives are overhead
intensive. More efficient C11 __atomic builtins are available with
newer GCC versions, see
http://gcc.gnu.org/onlinedocs/gcc-4.8.2/gcc/_005f_005fatomic-Builtins.html#_005f_005fatomic-Builtins

By substituting the following  __atomic… builtins:
__atomic_add_fetch
__atomic_sub_fetch

the performance improvement on certain architectures such as powerpc can be significant,
around 10% to 15%, over the implementation using __sync builtins while there is only slight uptick on
Intel architectures because it was already enforcing Intel Strongly ordered memory semantics.

The selection of __atomic built-ins can be predicated on the definition of ATOMIC_RELAXED
which Is available on in gcc 4.8.2 and later versions.

src/zmalloc.c

@@ -67,7 +67,10 @@ void zlibc_free(void *ptr) {
 #define free(ptr) je_free(ptr)
 #endif
 
-#ifdef HAVE_ATOMIC
+#if defined(__ATOMIC_RELAXED)
+#define update_zmalloc_stat_add(__n) __atomic_add_fetch(&used_memory, (__n), __ATOMIC_RELAXED)
+#define update_zmalloc_stat_sub(__n) __atomic_sub_fetch(&used_memory, (__n), __ATOMIC_RELAXED)
+#elif defined(HAVE_ATOMIC)
 #define update_zmalloc_stat_add(__n) __sync_add_and_fetch(&used_memory, (__n))
 #define update_zmalloc_stat_sub(__n) __sync_sub_and_fetch(&used_memory, (__n))
 #else
@@ -219,7 +222,7 @@ size_t zmalloc_used_memory(void) {
     size_t um;
 
     if (zmalloc_thread_safe) {
-#ifdef HAVE_ATOMIC
+#if defined(__ATOMIC_RELAXED) || defined(HAVE_ATOMIC)
         um = update_zmalloc_stat_add(0);
 #else
         pthread_mutex_lock(&used_memory_mutex);