diff options
Diffstat (limited to 'arch/x86_64/lib/memcpy.S')
| -rw-r--r-- | arch/x86_64/lib/memcpy.S | 69 |
1 files changed, 39 insertions, 30 deletions
diff --git a/arch/x86_64/lib/memcpy.S b/arch/x86_64/lib/memcpy.S index 5554948b5554..967b22fa7d07 100644 --- a/arch/x86_64/lib/memcpy.S +++ b/arch/x86_64/lib/memcpy.S @@ -1,6 +1,10 @@ /* Copyright 2002 Andi Kleen */ - #include <asm/cpufeature.h> +#include <linux/config.h> +#include <linux/linkage.h> +#include <asm/dwarf2.h> +#include <asm/cpufeature.h> + /* * memcpy - Copy a memory block. * @@ -13,12 +17,26 @@ * rax original destination */ - .globl __memcpy - .globl memcpy - .p2align 4 -__memcpy: -memcpy: + ALIGN +memcpy_c: + CFI_STARTPROC + movq %rdi,%rax + movl %edx,%ecx + shrl $3,%ecx + andl $7,%edx + rep movsq + movl %edx,%ecx + rep movsb + ret + CFI_ENDPROC +ENDPROC(memcpy_c) + +ENTRY(__memcpy) +ENTRY(memcpy) + CFI_STARTPROC pushq %rbx + CFI_ADJUST_CFA_OFFSET 8 + CFI_REL_OFFSET rbx, 0 movq %rdi,%rax movl %edx,%ecx @@ -86,36 +104,27 @@ memcpy: .Lende: popq %rbx + CFI_ADJUST_CFA_OFFSET -8 + CFI_RESTORE rbx ret .Lfinal: + CFI_ENDPROC +ENDPROC(memcpy) +ENDPROC(__memcpy) /* Some CPUs run faster using the string copy instructions. It is also a lot simpler. Use this when possible */ + .section .altinstr_replacement,"ax" +1: .byte 0xeb /* jmp <disp8> */ + .byte (memcpy_c - memcpy) - (2f - 1b) /* offset */ +2: + .previous .section .altinstructions,"a" .align 8 - .quad memcpy - .quad memcpy_c - .byte X86_FEATURE_REP_GOOD - .byte .Lfinal-memcpy - .byte memcpy_c_end-memcpy_c - .previous - - .section .altinstr_replacement,"ax" - /* rdi destination - * rsi source - * rdx count - */ -memcpy_c: - movq %rdi,%rax - movl %edx,%ecx - shrl $3,%ecx - andl $7,%edx - rep - movsq - movl %edx,%ecx - rep - movsb - ret -memcpy_c_end: + .quad memcpy + .quad 1b + .byte X86_FEATURE_REP_GOOD + .byte .Lfinal - memcpy + .byte 2b - 1b .previous |