arch/x86/memcpy.c: Optimize code for 64bit
In 64bit movsq is available which moves memory in chunks of 8 bytes
rather than 4 bytes.
Linux uses the same code.
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Change-Id: I65f178d2ed3aae54b0c1ce739c2b4af8738b9fcc
Reviewed-on: https://review.coreboot.org/c/coreboot/+/78646
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Reviewed-by: Martin Roth <martin.roth@amd.corp-partner.google.com>
Reviewed-by: Jérémy Compostella <jeremy.compostella@intel.com>
Reviewed-by: Eric Lai <ericllai@google.com>
diff --git a/src/arch/x86/memcpy.c b/src/arch/x86/memcpy.c
index 93002cd..d96a93c 100644
--- a/src/arch/x86/memcpy.c
+++ b/src/arch/x86/memcpy.c
@@ -14,19 +14,26 @@
check_memory_region((unsigned long)dest, n, true, _RET_IP_);
#endif
- asm volatile(
#if ENV_X86_64
- "rep ; movsd\n\t"
+ asm volatile(
+ "rep ; movsq\n\t"
"mov %4,%%rcx\n\t"
+ "rep ; movsb\n\t"
+ : "=&c" (d0), "=&D" (d1), "=&S" (d2)
+ : "0" (n >> 3), "g" (n & 7), "1" (dest), "2" (src)
+ : "memory"
+ );
#else
+ asm volatile(
"rep ; movsl\n\t"
"movl %4,%%ecx\n\t"
-#endif
"rep ; movsb\n\t"
: "=&c" (d0), "=&D" (d1), "=&S" (d2)
: "0" (n >> 2), "g" (n & 3), "1" (dest), "2" (src)
: "memory"
);
+#endif
+
return dest;
}