[PATCH] Use __sync_add_and_fetch and __sync_sub_and_fetch when they are available

[Mike Hommey]

From: Mike Hommey <mh at glandium.org>

These functions may be available as inlines or as libgcc functions. In the
former case, a __GCC_HAVE_SYNC_COMPARE_AND_SWAP_n macro is defined. But we
still want to use these functions in the latter case, when we don't have
our own implementation.
---
 configure.ac                        |   34 ++++++++++++++++++++++++++++++++++
 include/jemalloc/internal/atomic.h  |   18 +++++++++++++++---
 include/jemalloc/jemalloc_defs.h.in |   16 ++++++++++++++++
 3 files changed, 65 insertions(+), 3 deletions(-)

diff --git a/configure.ac b/configure.ac
index 478ae9d..5999a33 100644
--- a/configure.ac
+++ b/configure.ac
@@ -887,6 +887,40 @@ if test "x${je_cv_osatomic}" = "xyes" ; then
 fi
 
 dnl ============================================================================
+dnl Check whether __sync_{add,sub}_and_fetch() are available despite
+dnl __GCC_HAVE_SYNC_COMPARE_AND_SWAP_n macros being undefined.
+
+AC_DEFUN([JE_SYNC_COMPARE_AND_SWAP_CHECK],[
+  AC_CACHE_CHECK([whether to force $1-bit __sync_{add,sub}_and_fetch()],
+               [je_cv_sync_compare_and_swap_$2],
+               [AC_COMPILE_IFELSE([AC_LANG_PROGRAM([
+                                                    #include <stdint.h>
+                                                   ],
+                                                   [
+                                                    #ifndef __GCC_HAVE_SYNC_COMPARE_AND_SWAP_$2
+                                                    {
+                                                       uint$1_t x$1 = 0;
+                                                       __sync_add_and_fetch(&x$1, 42);
+                                                       __sync_sub_and_fetch(&x$1, 1);
+                                                    }
+                                                    #else
+                                                    #error __GCC_HAVE_SYNC_COMPARE_AND_SWAP_$2 is defined, no need to force
+                                                    #endif
+                                                   ])],
+                               [je_cv_sync_compare_and_swap_$2=yes],
+                               [je_cv_sync_compare_and_swap_$2=no])])
+
+  if test "x${je_cv_sync_compare_and_swap_$2}" = "xyes" ; then
+    AC_DEFINE([JE_FORCE_SYNC_COMPARE_AND_SWAP_$2], [ ])
+  fi
+])
+
+if test "x${je_cv_osatomic}" != "xyes" ; then
+  JE_SYNC_COMPARE_AND_SWAP_CHECK(32, 4)
+  JE_SYNC_COMPARE_AND_SWAP_CHECK(64, 8)
+fi
+
+dnl ============================================================================
 dnl Check for spinlock(3) operations as provided on Darwin.
 
 JE_COMPILABLE([Darwin OSSpin*()], [
diff --git a/include/jemalloc/internal/atomic.h b/include/jemalloc/internal/atomic.h
index 7a9cb61..d8f6ca5 100644
--- a/include/jemalloc/internal/atomic.h
+++ b/include/jemalloc/internal/atomic.h
@@ -87,6 +87,20 @@ atomic_sub_uint64(uint64_t *p, uint64_t x)
 
 	return (x);
 }
+#elif (defined(JE_FORCE_SYNC_COMPARE_AND_SWAP_8))
+JEMALLOC_INLINE uint64_t
+atomic_add_uint64(uint64_t *p, uint64_t x)
+{
+
+	return (__sync_add_and_fetch(p, x));
+}
+
+JEMALLOC_INLINE uint64_t
+atomic_sub_uint64(uint64_t *p, uint64_t x)
+{
+
+	return (__sync_sub_and_fetch(p, x));
+}
 #else
 #  if (LG_SIZEOF_PTR == 3)
 #    error "Missing implementation for 64-bit atomic operations"
@@ -150,9 +164,7 @@ atomic_sub_uint32(uint32_t *p, uint32_t x)
 
 	return (x);
 }
-#elif (defined __SH4__ || defined __mips__) && (__GNUC__ > 4 ||		\
-    (__GNUC__ == 4 && (__GNUC_MINOR__ > 1 || (__GNUC_MINOR__ == 1 &&	\
-    __GNUC_PATCHLEVEL__ > 1))))
+#elif (defined(JE_FORCE_SYNC_COMPARE_AND_SWAP_4))
 JEMALLOC_INLINE uint32_t
 atomic_add_uint32(uint32_t *p, uint32_t x)
 {
diff --git a/include/jemalloc/jemalloc_defs.h.in b/include/jemalloc/jemalloc_defs.h.in
index f150413..e4bfa04 100644
--- a/include/jemalloc/jemalloc_defs.h.in
+++ b/include/jemalloc/jemalloc_defs.h.in
@@ -54,6 +54,22 @@
 #undef JEMALLOC_OSATOMIC
 
 /*
+ * Defined if __sync_add_and_fetch(uint32_t *, uint32_t) and
+ * __sync_sub_and_fetch(uint32_t *, uint32_t) are available, despite
+ * __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4 not being defined (which means the
+ * functions are defined in libgcc instead of being inlines)
+ */
+#undef JE_FORCE_SYNC_COMPARE_AND_SWAP_4
+
+/*
+ * Defined if __sync_add_and_fetch(uint64_t *, uint64_t) and
+ * __sync_sub_and_fetch(uint64_t *, uint64_t) are available, despite
+ * __GCC_HAVE_SYNC_COMPARE_AND_SWAP_8 not being defined (which means the
+ * functions are defined in libgcc instead of being inlines)
+ */
+#undef JE_FORCE_SYNC_COMPARE_AND_SWAP_8
+
+/*
  * Defined if OSSpin*() functions are available, as provided by Darwin, and
  * documented in the spinlock(3) manual page.
  */
-- 
1.7.9.1