MdePkg: Clean up source files

[mirror_edk2.git] / MdePkg / Library / BaseMemoryLibOptDxe / X64 / CopyMem.S

#\r
# ConvertAsm.py: Automatically generated from CopyMem.asm\r
#\r
#------------------------------------------------------------------------------\r
#\r
# Copyright (c) 2006 - 2018, Intel Corporation. All rights reserved.<BR>\r
# 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.S\r
#\r
# Abstract:\r
#\r
#   CopyMem function\r
#\r
# Notes:\r
#\r
#------------------------------------------------------------------------------\r
\r
#------------------------------------------------------------------------------\r
#  VOID *\r
#  EFIAPI\r
#  InternalMemCopyMem (\r
#    IN VOID   *Destination,\r
#    IN VOID   *Source,\r
#    IN UINTN  Count\r
#    )\r
#------------------------------------------------------------------------------\r
ASM_GLOBAL ASM_PFX(InternalMemCopyMem)\r
ASM_PFX(InternalMemCopyMem):\r
    pushq   %rsi\r
    pushq   %rdi\r
    movq    %rdx, %rsi                  # rsi <- Source\r
    movq    %rcx, %rdi                  # rdi <- Destination\r
    leaq    -1(%rsi,%r8,), %r9          # r9 <- Last byte of Source\r
    cmpq    %rdi, %rsi\r
    movq    %rdi, %rax                  # rax <- Destination as return value\r
    jae     L0                          # Copy forward if Source > Destination\r
    cmpq    %rdi, %r9                   # Overlapped?\r
    jae     L_CopyBackward              # Copy backward if overlapped\r
L0:\r
    xorq    %rcx, %rcx\r
    subq    %rdi, %rcx                  # rcx <- -rdi\r
    andq    $15, %rcx                   # rcx + rsi should be 16 bytes aligned\r
    jz      L1                          # skip if rcx == 0\r
    cmpq    %r8, %rcx\r
    cmova   %r8, %rcx\r
    subq    %rcx, %r8\r
    rep     movsb\r
L1:\r
    movq    %r8,  %rcx\r
    andq    $15, %r8\r
    shrq    $4, %rcx                    # rcx <- # of DQwords to copy\r
    jz      L_CopyBytes\r
    movdqu  %xmm0, 0x18(%rsp)           # save xmm0 on stack\r
L2:\r
    movdqu  (%rsi), %xmm0               # rsi may not be 16-byte aligned\r
    movntdq %xmm0, (%rdi)               # rdi should be 16-byte aligned\r
    addq    $16, %rsi\r
    addq    $16, %rdi\r
    loop    L2\r
    mfence\r
    movdqa  0x18(%rsp), %xmm0            # restore xmm0\r
    jmp     L_CopyBytes                  # copy remaining bytes\r
L_CopyBackward:\r
    movq    %r9, %rsi                   # rsi <- Last byte of Source\r
    leaq     -1(%rdi, %r8,), %rdi       # rdi <- Last byte of Destination\r
    std\r
L_CopyBytes:\r
    movq    %r8, %rcx\r
    rep     movsb\r
    cld\r
    popq    %rdi\r
    popq    %rsi\r
    ret\r