github.com/hxx258456/ccgo@v0.0.5-0.20230213014102-48b35f46f66f/internal/xor/xor_amd64.s (about)

     1  // Copyright 2018 The Go Authors. All rights reserved.
     2  // Use of this source code is governed by a BSD-style
     3  // license that can be found in the LICENSE file.
     4  
     5  #include "textflag.h"
     6  
     7  // func xorBytesSSE2(dst, a, b *byte, n int)
     8  TEXT ·xorBytesSSE2(SB), NOSPLIT, $0
     9  	MOVQ  dst+0(FP), BX
    10  	MOVQ  a+8(FP), SI
    11  	MOVQ  b+16(FP), CX
    12  	MOVQ  n+24(FP), DX
    13  	TESTQ $15, DX            // AND 15 & len, if not zero jump to not_aligned.
    14  	JNZ   not_aligned
    15  
    16  aligned:
    17  	MOVQ $0, AX // position in slices
    18  
    19  loop16b:
    20  	MOVOU (SI)(AX*1), X0   // XOR 16byte forwards.
    21  	MOVOU (CX)(AX*1), X1
    22  	PXOR  X1, X0
    23  	MOVOU X0, (BX)(AX*1)
    24  	ADDQ  $16, AX
    25  	CMPQ  DX, AX
    26  	JNE   loop16b
    27  	RET
    28  
    29  loop_1b:
    30  	SUBQ  $1, DX           // XOR 1byte backwards.
    31  	MOVB  (SI)(DX*1), DI
    32  	MOVB  (CX)(DX*1), AX
    33  	XORB  AX, DI
    34  	MOVB  DI, (BX)(DX*1)
    35  	TESTQ $7, DX           // AND 7 & len, if not zero jump to loop_1b.
    36  	JNZ   loop_1b
    37  	CMPQ  DX, $0           // if len is 0, ret.
    38  	JE    ret
    39  	TESTQ $15, DX          // AND 15 & len, if zero jump to aligned.
    40  	JZ    aligned
    41  
    42  not_aligned:
    43  	TESTQ $7, DX           // AND $7 & len, if not zero jump to loop_1b.
    44  	JNE   loop_1b
    45  	SUBQ  $8, DX           // XOR 8bytes backwards.
    46  	MOVQ  (SI)(DX*1), DI
    47  	MOVQ  (CX)(DX*1), AX
    48  	XORQ  AX, DI
    49  	MOVQ  DI, (BX)(DX*1)
    50  	CMPQ  DX, $16          // if len is greater or equal 16 here, it must be aligned.
    51  	JGE   aligned
    52  
    53  ret:
    54  	RET