diff options
author | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-04-27 10:05:51 +0000 |
---|---|---|
committer | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-04-27 10:05:51 +0000 |
commit | 5d1646d90e1f2cceb9f0828f4b28318cd0ec7744 (patch) | |
tree | a94efe259b9009378be6d90eb30d2b019d95c194 /tools/arch/x86/lib/memcpy_64.S | |
parent | Initial commit. (diff) | |
download | linux-upstream.tar.xz linux-upstream.zip |
Adding upstream version 5.10.209.upstream/5.10.209upstream
Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>
Diffstat (limited to 'tools/arch/x86/lib/memcpy_64.S')
-rw-r--r-- | tools/arch/x86/lib/memcpy_64.S | 186 |
1 files changed, 186 insertions, 0 deletions
diff --git a/tools/arch/x86/lib/memcpy_64.S b/tools/arch/x86/lib/memcpy_64.S new file mode 100644 index 000000000..59cf2343f --- /dev/null +++ b/tools/arch/x86/lib/memcpy_64.S @@ -0,0 +1,186 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright 2002 Andi Kleen */ + +#include <linux/linkage.h> +#include <asm/errno.h> +#include <asm/cpufeatures.h> +#include <asm/alternative.h> +#include <asm/export.h> + +.pushsection .noinstr.text, "ax" + +/* + * We build a jump to memcpy_orig by default which gets NOPped out on + * the majority of x86 CPUs which set REP_GOOD. In addition, CPUs which + * have the enhanced REP MOVSB/STOSB feature (ERMS), change those NOPs + * to a jmp to memcpy_erms which does the REP; MOVSB mem copy. + */ + +/* + * memcpy - Copy a memory block. + * + * Input: + * rdi destination + * rsi source + * rdx count + * + * Output: + * rax original destination + */ +SYM_FUNC_START_ALIAS(__memcpy) +SYM_FUNC_START_WEAK(memcpy) + ALTERNATIVE_2 "jmp memcpy_orig", "", X86_FEATURE_REP_GOOD, \ + "jmp memcpy_erms", X86_FEATURE_ERMS + + movq %rdi, %rax + movq %rdx, %rcx + shrq $3, %rcx + andl $7, %edx + rep movsq + movl %edx, %ecx + rep movsb + RET +SYM_FUNC_END(memcpy) +SYM_FUNC_END_ALIAS(__memcpy) +EXPORT_SYMBOL(memcpy) +EXPORT_SYMBOL(__memcpy) + +/* + * memcpy_erms() - enhanced fast string memcpy. This is faster and + * simpler than memcpy. Use memcpy_erms when possible. + */ +SYM_FUNC_START_LOCAL(memcpy_erms) + movq %rdi, %rax + movq %rdx, %rcx + rep movsb + RET +SYM_FUNC_END(memcpy_erms) + +SYM_FUNC_START_LOCAL(memcpy_orig) + movq %rdi, %rax + + cmpq $0x20, %rdx + jb .Lhandle_tail + + /* + * We check whether memory false dependence could occur, + * then jump to corresponding copy mode. + */ + cmp %dil, %sil + jl .Lcopy_backward + subq $0x20, %rdx +.Lcopy_forward_loop: + subq $0x20, %rdx + + /* + * Move in blocks of 4x8 bytes: + */ + movq 0*8(%rsi), %r8 + movq 1*8(%rsi), %r9 + movq 2*8(%rsi), %r10 + movq 3*8(%rsi), %r11 + leaq 4*8(%rsi), %rsi + + movq %r8, 0*8(%rdi) + movq %r9, 1*8(%rdi) + movq %r10, 2*8(%rdi) + movq %r11, 3*8(%rdi) + leaq 4*8(%rdi), %rdi + jae .Lcopy_forward_loop + addl $0x20, %edx + jmp .Lhandle_tail + +.Lcopy_backward: + /* + * Calculate copy position to tail. + */ + addq %rdx, %rsi + addq %rdx, %rdi + subq $0x20, %rdx + /* + * At most 3 ALU operations in one cycle, + * so append NOPS in the same 16 bytes trunk. + */ + .p2align 4 +.Lcopy_backward_loop: + subq $0x20, %rdx + movq -1*8(%rsi), %r8 + movq -2*8(%rsi), %r9 + movq -3*8(%rsi), %r10 + movq -4*8(%rsi), %r11 + leaq -4*8(%rsi), %rsi + movq %r8, -1*8(%rdi) + movq %r9, -2*8(%rdi) + movq %r10, -3*8(%rdi) + movq %r11, -4*8(%rdi) + leaq -4*8(%rdi), %rdi + jae .Lcopy_backward_loop + + /* + * Calculate copy position to head. + */ + addl $0x20, %edx + subq %rdx, %rsi + subq %rdx, %rdi +.Lhandle_tail: + cmpl $16, %edx + jb .Lless_16bytes + + /* + * Move data from 16 bytes to 31 bytes. + */ + movq 0*8(%rsi), %r8 + movq 1*8(%rsi), %r9 + movq -2*8(%rsi, %rdx), %r10 + movq -1*8(%rsi, %rdx), %r11 + movq %r8, 0*8(%rdi) + movq %r9, 1*8(%rdi) + movq %r10, -2*8(%rdi, %rdx) + movq %r11, -1*8(%rdi, %rdx) + RET + .p2align 4 +.Lless_16bytes: + cmpl $8, %edx + jb .Lless_8bytes + /* + * Move data from 8 bytes to 15 bytes. + */ + movq 0*8(%rsi), %r8 + movq -1*8(%rsi, %rdx), %r9 + movq %r8, 0*8(%rdi) + movq %r9, -1*8(%rdi, %rdx) + RET + .p2align 4 +.Lless_8bytes: + cmpl $4, %edx + jb .Lless_3bytes + + /* + * Move data from 4 bytes to 7 bytes. + */ + movl (%rsi), %ecx + movl -4(%rsi, %rdx), %r8d + movl %ecx, (%rdi) + movl %r8d, -4(%rdi, %rdx) + RET + .p2align 4 +.Lless_3bytes: + subl $1, %edx + jb .Lend + /* + * Move data from 1 bytes to 3 bytes. + */ + movzbl (%rsi), %ecx + jz .Lstore_1byte + movzbq 1(%rsi), %r8 + movzbq (%rsi, %rdx), %r9 + movb %r8b, 1(%rdi) + movb %r9b, (%rdi, %rdx) +.Lstore_1byte: + movb %cl, (%rdi) + +.Lend: + RET +SYM_FUNC_END(memcpy_orig) + +.popsection |