[patch 2.6.17-rc5 2/2] i386 memcpy: minor optimization

[Chuck Ebbert <76306.1226@compuserve.com>]

Skip an additional instruction if memcpy size is divisible by four.

Signed-off-by: Chuck Ebbert <76306.1226@compuserve.com>

--- 2.6.17-rc5-32.orig/include/asm-i386/string.h
+++ 2.6.17-rc5-32/include/asm-i386/string.h
@@ -203,20 +203,20 @@ return __res;
 
 static __always_inline void * __memcpy(void * to, const void * from, size_t n)
 {
-int d0, d1, d2;
+int d0, d1, d2, d3;
 __asm__ __volatile__(
 	"rep ; movsl\n\t"
-	"movl %4,%%ecx\n\t"
-	"andl $3,%%ecx\n\t"
-#if 1	/* want to pay 2 byte penalty for a chance to skip microcoded rep? */
+	"andl $3,%3\n\t"
+#if 1	/* want to pay 2 byte penalty for a chance to skip a mov and a microcoded rep? */
 	"jz 1f\n\t"
 #endif
+	"movl %3,%%ecx\n\t"
 	"rep ; movsb\n\t"
 	"1:"
-	: "=&c" (d0), "=&D" (d1), "=&S" (d2)
-	: "0" (n/4), "g" (n), "1" ((long) to), "2" ((long) from)
+	: "=&c" (d0), "=&D" (d1), "=&S" (d2), "=&g" (d3)
+	: "0" (n/4), "1" ((long) to), "2" ((long) from), "3" (n)
 	: "memory");
-return (to);
+return to;
 }
 
 /*
-- 
Chuck