diff options
Diffstat (limited to 'MdePkg/Library/BaseMemoryLibSse2/X64/CopyMem.S')
| -rw-r--r-- | MdePkg/Library/BaseMemoryLibSse2/X64/CopyMem.S | 59 |
1 files changed, 29 insertions, 30 deletions
diff --git a/MdePkg/Library/BaseMemoryLibSse2/X64/CopyMem.S b/MdePkg/Library/BaseMemoryLibSse2/X64/CopyMem.S index e5cfa7eadf..8f8400e8ac 100644 --- a/MdePkg/Library/BaseMemoryLibSse2/X64/CopyMem.S +++ b/MdePkg/Library/BaseMemoryLibSse2/X64/CopyMem.S @@ -3,7 +3,7 @@ #
#------------------------------------------------------------------------------
#
-# Copyright (c) 2006 - 2008, Intel Corporation
+# Copyright (c) 2006 - 2009, 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
@@ -34,51 +34,50 @@ # IN UINTN Count
# )
#------------------------------------------------------------------------------
-.intel_syntax noprefix
ASM_GLOBAL ASM_PFX(InternalMemCopyMem)
ASM_PFX(InternalMemCopyMem):
- push rsi
- push rdi
- mov rsi, rdx # rsi <- Source
- mov rdi, rcx # rdi <- Destination
- lea r9, [rsi + r8 - 1] # r9 <- Last byte of Source
- cmp rsi, rdi
- mov rax, rdi # rax <- Destination as return value
+ pushq %rsi
+ pushq %rdi
+ movq %rdx, %rsi # rsi <- Source
+ movq %rcx, %rdi # rdi <- Destination
+ leaq -1(%rsi, %r8,), %r9 # r9 <- Last byte of Source
+ cmpq %rdi, %rsi
+ movq %rdi, %rax # rax <- Destination as return value
jae L0 # Copy forward if Source > Destination
- cmp r9, rdi # Overlapped?
+ cmpq %rdi, %r9 # Overlapped?
jae L_CopyBackward # Copy backward if overlapped
L0:
- xor rcx, rcx
- sub rcx, rdi # rcx <- -rdi
- and rcx, 15 # rcx + rsi should be 16 bytes aligned
+ xorq %rcx, %rcx
+ subq %rdi, %rcx # rcx <- -rdi
+ andq $15, %rcx # rcx + rsi should be 16 bytes aligned
jz L1 # skip if rcx == 0
- cmp rcx, r8
- cmova rcx, r8
- sub r8, rcx
+ cmpq %r8, %rcx
+ cmova %r8, %rcx
+ subq %rcx, %r8
rep movsb
L1:
- mov rcx, r8
- and r8, 15
- shr rcx, 4 # rcx <- # of DQwords to copy
+ movq %r8, %rcx
+ andq $15, %r8
+ shrq $4, %rcx # rcx <- # of DQwords to copy
jz L_CopyBytes
- movdqa [rsp + 0x18], xmm0 # save xmm0 on stack
+ movdqa %xmm0, 0x18(%rsp) # save xmm0 on stack
L2:
- movdqu xmm0, [rsi] # rsi may not be 16-byte aligned
- movntdq [rdi], xmm0 # rdi should be 16-byte aligned
- add rsi, 16
- add rdi, 16
+ movdqu (%rsi), %xmm0 # rsi may not be 16-byte aligned
+ movntdq %xmm0, (%rdi) # rdi should be 16-byte aligned
+ addq $16, %rsi
+ addq $16, %rdi
loop L2
mfence
- movdqa xmm0, [rsp + 0x18] # restore xmm0
+ movdqa 0x18(%rsp), %xmm0 # restore xmm0
jmp L_CopyBytes # copy remaining bytes
L_CopyBackward:
- mov rsi, r9 # rsi <- Last byte of Source
- lea rdi, [rdi + r8 - 1] # rdi <- Last byte of Destination
+ movq %r9, %rsi # rsi <- Last byte of Source
+ leaq -1(%rdi, %r8,), %rdi # rdi <- Last byte of Destination
std
L_CopyBytes:
- mov rcx, r8
+ movq %r8, %rcx
rep movsb
cld
- pop rdi
- pop rsi
+ popq %rdi
+ popq %rsi
ret
|