[OPTIM] GCC4's builtin_expect() is suboptimal
GCC4 is stupid (unbelievable news!).
When some code uses __builtin_expect(x != 0, 1), it really performs
the check of x != 0 then tests that the result is not zero! This is
a double check when only one was expected. Some performance drops
of 10% in the HTTP parser code have been observed due to this bug.
GCC 3.4 is fine though.
A solution consists in expecting that the tested value is 1. In
this case, it emits the correct code, but it's still not optimal
it seems. Finally the best solution is to ignore likely() and to
pray for the compiler to emit correct code. However, we still have
to fix unlikely() to remove the test there too, and to fix all
code which passed pointers overthere to pass integers instead.
diff --git a/include/common/ebtree.h b/include/common/ebtree.h
index 854666a..1a5ce86 100644
--- a/include/common/ebtree.h
+++ b/include/common/ebtree.h
@@ -1,6 +1,6 @@
/*
* Elastic Binary Trees - generic macros and structures.
- * (C) 2002-2007 - Willy Tarreau <w@1wt.eu>
+ * (C) 2002-2008 - Willy Tarreau <w@1wt.eu>
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
@@ -307,13 +307,23 @@
* this in inline functions, because the code reordering it causes very often
* has a negative impact on the calling functions.
*/
-#if __GNUC__ < 3 && !defined(__builtin_expect)
+#if !defined(likely)
+#if __GNUC__ < 3
#define __builtin_expect(x,y) (x)
-#endif
-
-#ifndef likely
+#define likely(x) (x)
+#define unlikely(x) (x)
+#elif __GNUC__ < 4
+/* gcc 3.x does the best job at this */
#define likely(x) (__builtin_expect((x) != 0, 1))
#define unlikely(x) (__builtin_expect((x) != 0, 0))
+#else
+/* GCC 4.x is stupid, it performs the comparison then compares it to 1,
+ * so we cheat in a dirty way to prevent it from doing this. This will
+ * only work with ints and booleans though.
+ */
+#define likely(x) (x)
+#define unlikely(x) (__builtin_expect((x), 0))
+#endif
#endif
/* Support passing function parameters in registers. For this, the