projects / mirror_edk2.git / blame
| Commit | Line | Data |
|---|---|---|
| 9987eaea JJ |
1 | ;------------------------------------------------------------------------------\r |
| 2 | ;\r | |
| 3 | ; Copyright (c) 2006, Intel Corporation. All rights reserved.<BR>\r | |
| 9344f092 | 4 | ; SPDX-License-Identifier: BSD-2-Clause-Patent\r |
| 9987eaea JJ |
5 | ;\r |
| 6 | ; Module Name:\r | |
| 7 | ;\r | |
| 8 | ; CopyMem.nasm\r | |
| 9 | ;\r | |
| 10 | ; Abstract:\r | |
| 11 | ;\r | |
| 12 | ; CopyMem function\r | |
| 13 | ;\r | |
| 14 | ; Notes:\r | |
| 15 | ;\r | |
| 16 | ;------------------------------------------------------------------------------\r | |
| 17 | \r | |
| 18 | DEFAULT REL\r | |
| 19 | SECTION .text\r | |
| 20 | \r | |
| 21 | ;------------------------------------------------------------------------------\r | |
| 22 | ; VOID *\r | |
| 23 | ; EFIAPI\r | |
| 24 | ; InternalMemCopyMem (\r | |
| 25 | ; IN VOID *Destination,\r | |
| 26 | ; IN VOID *Source,\r | |
| 27 | ; IN UINTN Count\r | |
| 28 | ; );\r | |
| 29 | ;------------------------------------------------------------------------------\r | |
| 30 | global ASM_PFX(InternalMemCopyMem)\r | |
| 31 | ASM_PFX(InternalMemCopyMem):\r | |
| 32 | push rsi\r | |
| 33 | push rdi\r | |
| 34 | mov rsi, rdx ; rsi <- Source\r | |
| 35 | mov rdi, rcx ; rdi <- Destination\r | |
| 36 | lea r9, [rsi + r8 - 1] ; r9 <- Last byte of Source\r | |
| 37 | cmp rsi, rdi\r | |
| 38 | mov rax, rdi ; rax <- Destination as return value\r | |
| 39 | jae .0 ; Copy forward if Source > Destination\r | |
| 40 | cmp r9, rdi ; Overlapped?\r | |
| 41 | jae @CopyBackward ; Copy backward if overlapped\r | |
| 42 | .0:\r | |
| 43 | xor rcx, rcx\r | |
| 44 | sub rcx, rdi ; rcx <- -rdi\r | |
| 45 | and rcx, 15 ; rcx + rsi should be 16 bytes aligned\r | |
| 46 | jz .1 ; skip if rcx == 0\r | |
| 47 | cmp rcx, r8\r | |
| 48 | cmova rcx, r8\r | |
| 49 | sub r8, rcx\r | |
| 50 | rep movsb\r | |
| 51 | .1:\r | |
| 52 | mov rcx, r8\r | |
| 53 | and r8, 15\r | |
| 54 | shr rcx, 4 ; rcx <- # of DQwords to copy\r | |
| 55 | jz @CopyBytes\r | |
| 56 | movdqa [rsp + 0x18], xmm0 ; save xmm0 on stack\r | |
| 57 | .2:\r | |
| 58 | movdqu xmm0, [rsi] ; rsi may not be 16-byte aligned\r | |
| 59 | movntdq [rdi], xmm0 ; rdi should be 16-byte aligned\r | |
| 60 | add rsi, 16\r | |
| 61 | add rdi, 16\r | |
| 62 | loop .2\r | |
| 63 | mfence\r | |
| 64 | movdqa xmm0, [rsp + 0x18] ; restore xmm0\r | |
| 65 | jmp @CopyBytes ; copy remaining bytes\r | |
| 66 | @CopyBackward:\r | |
| 67 | mov rsi, r9 ; rsi <- Last byte of Source\r | |
| 68 | lea rdi, [rdi + r8 - 1] ; rdi <- Last byte of Destination\r | |
| 69 | std\r | |
| 70 | @CopyBytes:\r | |
| 71 | mov rcx, r8\r | |
| 72 | rep movsb\r | |
| 73 | cld\r | |
| 74 | pop rdi\r | |
| 75 | pop rsi\r | |
| 76 | ret\r | |
| 77 | \r |