1
0
Fork 0
mirror of https://codeberg.org/forgejo/forgejo.git synced 2024-11-30 09:41:11 -05:00
forgejo/vendor/github.com/klauspost/crc32/crc32_amd64p32.s
2016-11-04 08:43:11 +01:00

67 lines
1.2 KiB
ArmAsm

// Copyright 2011 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
// +build gc
#define NOSPLIT 4
#define RODATA 8
// func castagnoliSSE42(crc uint32, p []byte) uint32
TEXT ·castagnoliSSE42(SB), NOSPLIT, $0
MOVL crc+0(FP), AX // CRC value
MOVL p+4(FP), SI // data pointer
MOVL p_len+8(FP), CX // len(p)
NOTL AX
// If there's less than 8 bytes to process, we do it byte-by-byte.
CMPQ CX, $8
JL cleanup
// Process individual bytes until the input is 8-byte aligned.
startup:
MOVQ SI, BX
ANDQ $7, BX
JZ aligned
CRC32B (SI), AX
DECQ CX
INCQ SI
JMP startup
aligned:
// The input is now 8-byte aligned and we can process 8-byte chunks.
CMPQ CX, $8
JL cleanup
CRC32Q (SI), AX
ADDQ $8, SI
SUBQ $8, CX
JMP aligned
cleanup:
// We may have some bytes left over that we process one at a time.
CMPQ CX, $0
JE done
CRC32B (SI), AX
INCQ SI
DECQ CX
JMP cleanup
done:
NOTL AX
MOVL AX, ret+16(FP)
RET
// func haveSSE42() bool
TEXT ·haveSSE42(SB), NOSPLIT, $0
XORQ AX, AX
INCL AX
CPUID
SHRQ $20, CX
ANDQ $1, CX
MOVB CX, ret+0(FP)
RET