diff options
Diffstat (limited to 'arch/x86/lib/memmove_64.S')
| -rw-r--r-- | arch/x86/lib/memmove_64.S | 34 |
1 files changed, 30 insertions, 4 deletions
diff --git a/arch/x86/lib/memmove_64.S b/arch/x86/lib/memmove_64.S index 0ecb8433e5a..65268a6104f 100644 --- a/arch/x86/lib/memmove_64.S +++ b/arch/x86/lib/memmove_64.S @@ -8,6 +8,8 @@ #define _STRING_C #include <linux/linkage.h> #include <asm/dwarf2.h> +#include <asm/cpufeature.h> +#include <asm/alternative-asm.h> #undef memmove @@ -24,15 +26,21 @@ */ ENTRY(memmove) CFI_STARTPROC - /* Handle more 32bytes in loop */ + + /* Handle more 32 bytes in loop */ mov %rdi, %rax cmp $0x20, %rdx jb 1f /* Decide forward/backward copy mode */ cmp %rdi, %rsi - jb 2f + jge .Lmemmove_begin_forward + mov %rsi, %r8 + add %rdx, %r8 + cmp %rdi, %r8 + jg 2f +.Lmemmove_begin_forward: /* * movsq instruction have many startup latency * so we handle small size by general register. @@ -48,7 +56,7 @@ ENTRY(memmove) 3: sub $0x20, %rdx /* - * We gobble 32byts forward in each loop. + * We gobble 32 bytes forward in each loop. */ 5: sub $0x20, %rdx @@ -78,6 +86,8 @@ ENTRY(memmove) rep movsq movq %r11, (%r10) jmp 13f +.Lmemmove_end_forward: + /* * Handle data backward by movsq. */ @@ -112,7 +122,7 @@ ENTRY(memmove) addq %rdx, %rdi subq $0x20, %rdx /* - * We gobble 32byts backward in each loop. + * We gobble 32 bytes backward in each loop. */ 8: subq $0x20, %rdx @@ -194,4 +204,20 @@ ENTRY(memmove) 13: retq CFI_ENDPROC + + .section .altinstr_replacement,"ax" +.Lmemmove_begin_forward_efs: + /* Forward moving data. */ + movq %rdx, %rcx + rep movsb + retq +.Lmemmove_end_forward_efs: + .previous + + .section .altinstructions,"a" + altinstruction_entry .Lmemmove_begin_forward, \ + .Lmemmove_begin_forward_efs,X86_FEATURE_ERMS, \ + .Lmemmove_end_forward-.Lmemmove_begin_forward, \ + .Lmemmove_end_forward_efs-.Lmemmove_begin_forward_efs + .previous ENDPROC(memmove) |