| // Copyright 2016 The Fuchsia Authors |
| // |
| // Use of this source code is governed by a MIT-style |
| // license that can be found in the LICENSE file or at |
| // https://opensource.org/licenses/MIT |
| |
| #include <asm.h> |
| #include <lib/code_patching.h> |
| |
| .text |
| |
| // memcpy implementation relying on Intel's Enhanced REP MOVSB optimization |
| // %rax = memcpy_erms(%rdi, %rsi, %rdx) |
| FUNCTION(memcpy_erms) |
| // Save return value. |
| mov %rdi, %rax |
| |
| mov %rdx, %rcx |
| rep movsb // while (rcx-- > 0) *rdi++ = *rsi++; /* rdi, rsi are uint8_t* */ |
| ret |
| END_FUNCTION(memcpy_erms) |
| |
| // memcpy implementation that copies 8 bytes at a time when possible |
| // %rax = memcpy_quad(%rdi, %rsi, %rdx) |
| FUNCTION(memcpy_quad) |
| // Save return value. |
| mov %rdi, %rax |
| |
| // Copy all of the 8 byte chunks we can |
| mov %rdx, %rcx |
| shr $3, %rcx |
| rep movsq // while (rcx-- > 0) { *rdi++ = *rsi++; /* rdi, rsi are uint64_t* */ } |
| |
| // Copy the rest |
| mov %rdx, %rcx |
| and $0x7, %rcx |
| rep movsb |
| ret |
| END_FUNCTION(memcpy_quad) |
| |
| FUNCTION(memcpy) |
| jmp memcpy_erms |
| APPLY_CODE_PATCH_FUNC_WITH_DEFAULT(x86_memcpy_select, memcpy, 2) |
| END_FUNCTION(memcpy) |