MdePkg/Library/BaseMemoryLibSse2/X64/CopyMem.S

   1 #
   2 # ConvertAsm.py: Automatically generated from CopyMem.asm
   3 #
   4 #------------------------------------------------------------------------------
   5 #
   6 # Copyright (c) 2006 - 2008, Intel Corporation
   7 # All rights reserved. This program and the accompanying materials
   8 # are licensed and made available under the terms and conditions of the BSD License
   9 # which accompanies this distribution.  The full text of the license may be found at
  10 # http://opensource.org/licenses/bsd-license.php
  11 #
  12 # THE PROGRAM IS DISTRIBUTED UNDER THE BSD LICENSE ON AN "AS IS" BASIS,
  13 # WITHOUT WARRANTIES OR REPRESENTATIONS OF ANY KIND, EITHER EXPRESS OR IMPLIED.
  14 #
  15 # Module Name:
  16 #
  17 #   CopyMem.S
  18 #
  19 # Abstract:
  20 #
  21 #   CopyMem function
  22 #
  23 # Notes:
  24 #
  25 #------------------------------------------------------------------------------
  26
  27
  28 #------------------------------------------------------------------------------
  29 #  VOID *
  30 #  EFIAPI
  31 #  InternalMemCopyMem (
  32 #    IN VOID   *Destination,
  33 #    IN VOID   *Source,
  34 #    IN UINTN  Count
  35 #    )
  36 #------------------------------------------------------------------------------
  37 .intel_syntax noprefix
  38 .globl ASM_PFX(InternalMemCopyMem)
  39 ASM_PFX(InternalMemCopyMem):
  40     push    rsi
  41     push    rdi
  42     mov     rsi, rdx                    # rsi <- Source
  43     mov     rdi, rcx                    # rdi <- Destination
  44     lea     r9, [rsi + r8 - 1]          # r9 <- Last byte of Source
  45     cmp     rsi, rdi
  46     mov     rax, rdi                    # rax <- Destination as return value
  47     jae     L0                          # Copy forward if Source > Destination
  48     cmp     r9, rdi                     # Overlapped?
  49     jae     L_CopyBackward              # Copy backward if overlapped
  50 L0:
  51     xor     rcx, rcx
  52     sub     rcx, rdi                    # rcx <- -rdi
  53     and     rcx, 15                     # rcx + rsi should be 16 bytes aligned
  54     jz      L1                          # skip if rcx == 0
  55     cmp     rcx, r8
  56     cmova   rcx, r8
  57     sub     r8, rcx
  58     rep     movsb
  59 L1:
  60     mov     rcx, r8
  61     and     r8, 15
  62     shr     rcx, 4                      # rcx <- # of DQwords to copy
  63     jz      L_CopyBytes
  64     movdqa  [rsp + 0x18], xmm0          # save xmm0 on stack
  65 L2:
  66     movdqu  xmm0, [rsi]                 # rsi may not be 16-byte aligned
  67     movntdq [rdi], xmm0                 # rdi should be 16-byte aligned
  68     add     rsi, 16
  69     add     rdi, 16
  70     loop    L2
  71     mfence
  72     movdqa  xmm0, [rsp + 0x18]          # restore xmm0
  73     jmp     L_CopyBytes                 # copy remaining bytes
  74 L_CopyBackward:
  75     mov     rsi, r9                     # rsi <- Last byte of Source
  76     lea     rdi, [rdi + r8 - 1]         # rdi <- Last byte of Destination
  77     std
  78 L_CopyBytes:
  79     mov     rcx, r8
  80     rep     movsb
  81     cld
  82     pop     rdi
  83     pop     rsi
  84     ret