Make bn/asm/x86_64-gcc.c gcc4 savvy. +r is likely to be initially

introduced for a reason [like bug in initial gcc port], but proposed
=&r is treated correctly by senior 3.2, so we can assume it's safe now.
PR: 1031

crypto/bn/asm/x86_64-gcc.c

@@ -186,7 +186,7 @@ BN_ULONG bn_add_words (BN_ULONG *rp, BN_ULONG *ap, BN_ULONG *bp,int n)
 	"	leaq	1(%2),%2	\n"
 	"	loop	1b		\n"
 	"	sbbq	%0,%0		\n"
-		: "+a"(ret),"+c"(n),"+r"(i)
+		: "=&a"(ret),"+c"(n),"=&r"(i)
 		: "r"(rp),"r"(ap),"r"(bp)
 		: "cc"
 	);
@@ -209,7 +209,7 @@ BN_ULONG bn_sub_words (BN_ULONG *rp, BN_ULONG *ap, BN_ULONG *bp,int n)
 	"	leaq	1(%2),%2	\n"
 	"	loop	1b		\n"
 	"	sbbq	%0,%0		\n"
-		: "+a"(ret),"+c"(n),"+r"(i)
+		: "=&a"(ret),"+c"(n),"=&r"(i)
 		: "r"(rp),"r"(ap),"r"(bp)
 		: "cc"
 	);