diff options
| author | Arthur Heymans <arthur@aheymans.xyz> | 2023-10-25 08:22:49 +0200 |
|---|---|---|
| committer | Martin L Roth <gaumless@gmail.com> | 2023-11-01 15:47:03 +0000 |
| commit | fdf85614dcb6d9a0d2134f2d2c82ff85ea185b44 (patch) | |
| tree | a3ec85f5e42ef489b24db37d88ee0df0ec441501 /src/arch/x86 | |
| parent | 8bd7d6c806d2a7122b524279302688db2f93ea98 (diff) | |
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>
Diffstat (limited to 'src/arch/x86')
| -rw-r--r-- | src/arch/x86/memcpy.c | 13 |
1 files changed, 10 insertions, 3 deletions
diff --git a/src/arch/x86/memcpy.c b/src/arch/x86/memcpy.c index 93002cdd27..d96a93cd93 100644 --- a/src/arch/x86/memcpy.c +++ b/src/arch/x86/memcpy.c @@ -14,19 +14,26 @@ void *memcpy(void *dest, const void *src, size_t n) 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; } |