audk/MdePkg/Library/BaseMemoryLibSse2/Ia32/CopyMem.S

89 lines
2.5 KiB
ArmAsm

#------------------------------------------------------------------------------
#
# Copyright (c) 2006, Intel Corporation
# All rights reserved. This program and the accompanying materials
# are licensed and made available under the terms and conditions of the BSD License
# which accompanies this distribution. The full text of the license may be found at
# http://opensource.org/licenses/bsd-license.php
#
# THE PROGRAM IS DISTRIBUTED UNDER THE BSD LICENSE ON AN "AS IS" BASIS,
# WITHOUT WARRANTIES OR REPRESENTATIONS OF ANY KIND, EITHER EXPRESS OR IMPLIED.
#
# Module Name:
#
# CopyMem.asm
#
# Abstract:
#
# CopyMem function
#
# Notes:
#
#------------------------------------------------------------------------------
.686:
#.MODEL flat,C
.xmm:
.code:
#------------------------------------------------------------------------------
# VOID *
# _mem_CopyMem (
# IN VOID *Destination,
# IN VOID *Source,
# IN UINTN Count
# )
#------------------------------------------------------------------------------
.global _InternalMemCopyMem
_InternalMemCopyMem:
push %esi
push %edi
movl 16(%esp), %esi # esi <- Source
movl 12(%esp), %edi # edi <- Destination
movl 20(%esp), %edx # edx <- Count
leal -1(%edi,%edx,), %eax # eax <- End of Destination
cmpl %edi, %esi
jae L0
cmpl %esi, %eax # Overlapped?
jae @CopyBackward # Copy backward if overlapped
L0:
xorl %ecx, %ecx
subl %edi, %ecx
andl $15, %ecx # ecx + edi aligns on 16-byte boundary
jz L1
cmpl %edx, %ecx
cmova %edx, %ecx
subl %ecx, %edx # edx <- remaining bytes to copy
rep
movsb
L1:
movl %edx, %ecx
andl $15, %edx
shrl $4, %ecx # ecx <- # of DQwords to copy
jz @CopyBytes
addl $-16, %esp
movdqu %xmm0, (%esp)
L2:
movdqu (%esi), %xmm0
movntdq %xmm0, (%edi)
addl $16, %esi
addl $16, %edi
loop L2
mfence
movdqu (%esp),%xmm0
addl $16, %esp # stack cleanup
jmp @CopyBytes
@CopyBackward:
movl %eax, %edi # edi <- Last byte in Destination
leal -1(%esi,%edx,), %esi # esi <- Last byte in Source
std
@CopyBytes:
movl %edx, %ecx
rep
movsb
cld
movl 12(%esp), %eax # eax <- Destination as return value
pop %edi
pop %esi
ret