projects / thirdparty / kernel / linux.git / blame
| Commit | Line | Data |
|---|---|---|
| b2441318 | 1 | /* SPDX-License-Identifier: GPL-2.0 */ |
| 1da177e4 | 2 | /* Written 2003 by Andi Kleen, based on a kernel by Evandro Menezes */ |
| 038b0a6d | 3 | |
| 8d379dad | 4 | #include <linux/linkage.h> |
| cd4d09ec | 5 | #include <asm/cpufeatures.h> |
| 59e97e4d | 6 | #include <asm/alternative-asm.h> |
| 784d5699 | 7 | #include <asm/export.h> |
| 8d379dad | 8 | |
| 090a3f61 BP |
9 | /* |
| 10 | * Some CPUs run faster using the string copy instructions (sane microcode). | |
| 11 | * It is also a lot simpler. Use this when possible. But, don't use streaming | |
| 12 | * copy unless the CPU indicates X86_FEATURE_REP_GOOD. Could vary the | |
| 13 | * prefetch distance based on SMP/UP. | |
| 14 | */ | |
| 8d379dad | 15 | ALIGN |
| 090a3f61 | 16 | ENTRY(copy_page) |
| 090a3f61 | 17 | ALTERNATIVE "jmp copy_page_regs", "", X86_FEATURE_REP_GOOD |
| 269833bd ML |
18 | movl $4096/8, %ecx |
| 19 | rep movsq | |
| 8d379dad | 20 | ret |
| 090a3f61 | 21 | ENDPROC(copy_page) |
| 784d5699 | 22 | EXPORT_SYMBOL(copy_page) |
| 1da177e4 | 23 | |
| 090a3f61 | 24 | ENTRY(copy_page_regs) |
| 269833bd | 25 | subq $2*8, %rsp |
| 269833bd | 26 | movq %rbx, (%rsp) |
| 269833bd | 27 | movq %r12, 1*8(%rsp) |
| 7bcd3f34 | 28 | |
| 269833bd | 29 | movl $(4096/64)-5, %ecx |
| 7bcd3f34 AK |
30 | .p2align 4 |
| 31 | .Loop64: | |
| 269833bd ML |
32 | dec %rcx |
| 33 | movq 0x8*0(%rsi), %rax | |
| 34 | movq 0x8*1(%rsi), %rbx | |
| 35 | movq 0x8*2(%rsi), %rdx | |
| 36 | movq 0x8*3(%rsi), %r8 | |
| 37 | movq 0x8*4(%rsi), %r9 | |
| 38 | movq 0x8*5(%rsi), %r10 | |
| 39 | movq 0x8*6(%rsi), %r11 | |
| 40 | movq 0x8*7(%rsi), %r12 | |
| 7bcd3f34 AK |
41 | |
| 42 | prefetcht0 5*64(%rsi) | |
| 43 | ||
| 269833bd ML |
44 | movq %rax, 0x8*0(%rdi) |
| 45 | movq %rbx, 0x8*1(%rdi) | |
| 46 | movq %rdx, 0x8*2(%rdi) | |
| 47 | movq %r8, 0x8*3(%rdi) | |
| 48 | movq %r9, 0x8*4(%rdi) | |
| 49 | movq %r10, 0x8*5(%rdi) | |
| 50 | movq %r11, 0x8*6(%rdi) | |
| 51 | movq %r12, 0x8*7(%rdi) | |
| 7bcd3f34 | 52 | |
| 269833bd ML |
53 | leaq 64 (%rsi), %rsi |
| 54 | leaq 64 (%rdi), %rdi | |
| 7bcd3f34 | 55 | |
| 269833bd | 56 | jnz .Loop64 |
| 7bcd3f34 | 57 | |
| 269833bd | 58 | movl $5, %ecx |
| 7bcd3f34 AK |
59 | .p2align 4 |
| 60 | .Loop2: | |
| 269833bd ML |
61 | decl %ecx |
| 62 | ||
| 63 | movq 0x8*0(%rsi), %rax | |
| 64 | movq 0x8*1(%rsi), %rbx | |
| 65 | movq 0x8*2(%rsi), %rdx | |
| 66 | movq 0x8*3(%rsi), %r8 | |
| 67 | movq 0x8*4(%rsi), %r9 | |
| 68 | movq 0x8*5(%rsi), %r10 | |
| 69 | movq 0x8*6(%rsi), %r11 | |
| 70 | movq 0x8*7(%rsi), %r12 | |
| 71 | ||
| 72 | movq %rax, 0x8*0(%rdi) | |
| 73 | movq %rbx, 0x8*1(%rdi) | |
| 74 | movq %rdx, 0x8*2(%rdi) | |
| 75 | movq %r8, 0x8*3(%rdi) | |
| 76 | movq %r9, 0x8*4(%rdi) | |
| 77 | movq %r10, 0x8*5(%rdi) | |
| 78 | movq %r11, 0x8*6(%rdi) | |
| 79 | movq %r12, 0x8*7(%rdi) | |
| 80 | ||
| 81 | leaq 64(%rdi), %rdi | |
| 82 | leaq 64(%rsi), %rsi | |
| 7bcd3f34 AK |
83 | jnz .Loop2 |
| 84 | ||
| 269833bd | 85 | movq (%rsp), %rbx |
| 269833bd | 86 | movq 1*8(%rsp), %r12 |
| 269833bd | 87 | addq $2*8, %rsp |
| 7bcd3f34 | 88 | ret |
| 090a3f61 | 89 | ENDPROC(copy_page_regs) |