| b1ff428c |
1 | #\r |
| 2 | # ConvertAsm.py: Automatically generated from CopyMem.asm\r |
| 3 | #\r |
| 4 | #------------------------------------------------------------------------------\r |
| 5 | #\r |
| 373ade0e |
6 | # Copyright (c) 2006 - 2008, Intel Corporation\r |
| b1ff428c |
7 | # All rights reserved. This program and the accompanying materials\r |
| 8 | # are licensed and made available under the terms and conditions of the BSD License\r |
| 9 | # which accompanies this distribution. The full text of the license may be found at\r |
| 10 | # http://opensource.org/licenses/bsd-license.php\r |
| 11 | #\r |
| 12 | # THE PROGRAM IS DISTRIBUTED UNDER THE BSD LICENSE ON AN "AS IS" BASIS,\r |
| 13 | # WITHOUT WARRANTIES OR REPRESENTATIONS OF ANY KIND, EITHER EXPRESS OR IMPLIED.\r |
| 14 | #\r |
| 15 | # Module Name:\r |
| 16 | #\r |
| 17 | # CopyMem.S\r |
| 18 | #\r |
| 19 | # Abstract:\r |
| 20 | #\r |
| 21 | # CopyMem function\r |
| 22 | #\r |
| 23 | # Notes:\r |
| 24 | #\r |
| 25 | #------------------------------------------------------------------------------\r |
| 26 | \r |
| 27 | #------------------------------------------------------------------------------\r |
| 28 | # VOID *\r |
| 29 | # EFIAPI\r |
| 30 | # InternalMemCopyMem (\r |
| 31 | # IN VOID *Destination,\r |
| 32 | # IN VOID *Source,\r |
| 33 | # IN UINTN Count\r |
| 34 | # )\r |
| 2f3d6fb5 |
35 | #------------------------------------------------------------------------------\r |
| 36 | .intel_syntax noprefix\r |
| 37 | .globl ASM_PFX(InternalMemCopyMem)\r |
| 38 | ASM_PFX(InternalMemCopyMem):\r |
| 39 | push rsi\r |
| 40 | push rdi\r |
| 41 | mov rsi, rdx # rsi <- Source\r |
| 42 | mov rdi, rcx # rdi <- Destination\r |
| 43 | lea r9, [rsi + r8 - 1] # r9 <- Last byte of Source\r |
| 44 | cmp rsi, rdi\r |
| 45 | mov rax, rdi # rax <- Destination as return value\r |
| 46 | jae L0 # Copy forward if Source > Destination\r |
| 47 | cmp r9, rdi # Overlapped?\r |
| 48 | jae @CopyBackward # Copy backward if overlapped\r |
| 49 | L0:\r |
| 50 | xor rcx, rcx\r |
| 51 | sub rcx, rdi # rcx <- -rdi\r |
| 52 | and rcx, 15 # rcx + rsi should be 16 bytes aligned\r |
| 53 | jz L1 # skip if rcx == 0\r |
| 54 | cmp rcx, r8\r |
| 55 | cmova rcx, r8\r |
| 56 | sub r8, rcx\r |
| 57 | rep movsb\r |
| 58 | L1:\r |
| 59 | mov rcx, r8\r |
| 60 | and r8, 15\r |
| 61 | shr rcx, 4 # rcx <- # of DQwords to copy\r |
| 62 | jz L_CopyBytes\r |
| 63 | movdqa [rsp + 0x18], xmm0 # save xmm0 on stack\r |
| 64 | L2:\r |
| 65 | movdqu xmm0, [rsi] # rsi may not be 16-byte aligned\r |
| 66 | movntdq [rdi], xmm0 # rdi should be 16-byte aligned\r |
| 67 | add rsi, 16\r |
| 68 | add rdi, 16\r |
| 69 | loop L2\r |
| 70 | mfence\r |
| 71 | movdqa xmm0, [rsp + 0x18] # restore xmm0\r |
| 72 | jmp L_CopyBytes # copy remaining bytes\r |
| 73 | L_CopyBackward:\r |
| 74 | mov rsi, r9 # rsi <- Last byte of Source\r |
| 75 | lea rdi, [rdi + r8 - 1] # rdi <- Last byte of Destination\r |
| 76 | std\r |
| 77 | L_CopyBytes:\r |
| 78 | mov rcx, r8\r |
| 79 | rep movsb\r |
| 80 | cld\r |
| 81 | pop rdi\r |
| 82 | pop rsi\r |
| 83 | ret\r |
projects / mirror_edk2.git / blame