Initial import.

[mirror_edk2.git] / MdePkg / Library / BaseMemoryLibSse2 / x64 / CopyMem.asm

;------------------------------------------------------------------------------\r
;\r
; Copyright (c) 2006, Intel Corporation\r
; All rights reserved. This program and the accompanying materials\r
; are licensed and made available under the terms and conditions of the BSD License\r
; which accompanies this distribution.  The full text of the license may be found at\r
; http://opensource.org/licenses/bsd-license.php\r
;\r
; THE PROGRAM IS DISTRIBUTED UNDER THE BSD LICENSE ON AN "AS IS" BASIS,\r
; WITHOUT WARRANTIES OR REPRESENTATIONS OF ANY KIND, EITHER EXPRESS OR IMPLIED.\r
;\r
; Module Name:\r
;\r
;   CopyMem.asm\r
;\r
; Abstract:\r
;\r
;   CopyMem function\r
;\r
; Notes:\r
;\r
;------------------------------------------------------------------------------\r
\r
    .code\r
\r
;------------------------------------------------------------------------------\r
;  VOID *\r
;  _mem_CopyMem (\r
;    IN VOID   *Destination,\r
;    IN VOID   *Source,\r
;    IN UINTN  Count\r
;    )\r
;------------------------------------------------------------------------------\r
InternalMemCopyMem  PROC    USES    rsi rdi\r
    mov     rsi, rdx                    ; rsi <- Source\r
    mov     rdi, rcx                    ; rdi <- Destination\r
    lea     r9, [rdi + r8 - 1]          ; r9 <- Last byte of Destination\r
    cmp     rsi, rdi\r
    mov     rax, rdi                    ; rax <- Destination as return value\r
    jae     @F                          ; Copy forward if Source > Destination\r
    cmp     r9, rsi                     ; Overlapped?\r
    jae     @CopyBackward               ; Copy backward if overlapped\r
@@:\r
    xor     rcx, rcx\r
    sub     rcx, rdi                    ; rcx <- -rdi\r
    and     rcx, 15                     ; rcx + rsi should be 16 bytes aligned\r
    jz      @F                          ; skip if rcx == 0\r
    cmp     rcx, r8\r
    cmova   rcx, r8\r
    sub     r8, rcx\r
    rep     movsb\r
@@:\r
    mov     rcx, r8\r
    and     r8, 15\r
    shr     rcx, 4                      ; rcx <- # of DQwords to copy\r
    jz      @CopyBytes\r
    movdqa  [rsp + 18h], xmm0           ; save xmm0 on stack\r
@@:\r
    movdqu  xmm0, [rsi]                 ; rsi may not be 16-byte aligned\r
    movntdq [rdi], xmm0                 ; rdi should be 16-byte aligned\r
    add     rsi, 16\r
    add     rdi, 16\r
    loop    @B\r
    mfence\r
    movdqa  xmm0, [rsp + 18h]           ; restore xmm0\r
    jmp     @CopyBytes                  ; copy remaining bytes\r
@CopyBackward:\r
    mov     rdi, r9                     ; rdi <- Last byte of Destination\r
    lea     rsi, [rsi + r8 - 1]         ; rsi <- Last byte of Source\r
    std\r
@CopyBytes:\r
    mov     rcx, r8\r
    rep     movsb\r
    cld\r
    ret\r
InternalMemCopyMem  ENDP\r
\r
    END\r