1 /* SPDX-License-Identifier: GPL-2.0 */ !! 1 /*
2 /* memmove.S: Simple memmove implementation. !! 2 * arch/alpha/lib/memmove.S
3 * 3 *
4 * Copyright (C) 1997, 2004 David S. Miller (d !! 4 * Barely optimized memmove routine for Alpha EV5.
5 * Copyright (C) 1996, 1997, 1998, 1999 Jakub !! 5 *
>> 6 * This is hand-massaged output from the original memcpy.c. We defer to
>> 7 * memcpy whenever possible; the backwards copy loops are not unrolled.
6 */ 8 */
>> 9
>> 10 .set noat
>> 11 .set noreorder
>> 12 .text
7 13
8 #include <linux/export.h> !! 14 .align 4
9 #include <linux/linkage.h> !! 15 .globl memmove
>> 16 .ent memmove
>> 17 memmove:
>> 18 addq $16,$18,$4
>> 19 addq $17,$18,$5
>> 20 cmpule $4,$17,$1 /* dest + n <= src */
>> 21 cmpule $5,$16,$2 /* dest >= src + n */
>> 22
>> 23 bis $1,$2,$1
>> 24 mov $16,$0
>> 25 xor $16,$17,$2
>> 26 bne $1,memcpy
>> 27
>> 28 and $2,7,$2 /* Test for src/dest co-alignment. */
>> 29 and $16,7,$1
>> 30 cmpule $16,$17,$3
>> 31 bne $3,$memmove_up /* dest < src */
>> 32
>> 33 and $4,7,$1
>> 34 bne $2,$misaligned_dn
>> 35 unop
>> 36 beq $1,$skip_aligned_byte_loop_head_dn
>> 37
>> 38 $aligned_byte_loop_head_dn:
>> 39 lda $4,-1($4)
>> 40 lda $5,-1($5)
>> 41 unop
>> 42 ble $18,$egress
>> 43
>> 44 ldq_u $3,0($5)
>> 45 ldq_u $2,0($4)
>> 46 lda $18,-1($18)
>> 47 extbl $3,$5,$1
>> 48
>> 49 insbl $1,$4,$1
>> 50 mskbl $2,$4,$2
>> 51 bis $1,$2,$1
>> 52 and $4,7,$6
>> 53
>> 54 stq_u $1,0($4)
>> 55 bne $6,$aligned_byte_loop_head_dn
>> 56
>> 57 $skip_aligned_byte_loop_head_dn:
>> 58 lda $18,-8($18)
>> 59 blt $18,$skip_aligned_word_loop_dn
>> 60
>> 61 $aligned_word_loop_dn:
>> 62 ldq $1,-8($5)
>> 63 nop
>> 64 lda $5,-8($5)
>> 65 lda $18,-8($18)
>> 66
>> 67 stq $1,-8($4)
>> 68 nop
>> 69 lda $4,-8($4)
>> 70 bge $18,$aligned_word_loop_dn
>> 71
>> 72 $skip_aligned_word_loop_dn:
>> 73 lda $18,8($18)
>> 74 bgt $18,$byte_loop_tail_dn
>> 75 unop
>> 76 ret $31,($26),1
>> 77
>> 78 .align 4
>> 79 $misaligned_dn:
>> 80 nop
>> 81 fnop
>> 82 unop
>> 83 beq $18,$egress
>> 84
>> 85 $byte_loop_tail_dn:
>> 86 ldq_u $3,-1($5)
>> 87 ldq_u $2,-1($4)
>> 88 lda $5,-1($5)
>> 89 lda $4,-1($4)
>> 90
>> 91 lda $18,-1($18)
>> 92 extbl $3,$5,$1
>> 93 insbl $1,$4,$1
>> 94 mskbl $2,$4,$2
>> 95
>> 96 bis $1,$2,$1
>> 97 stq_u $1,0($4)
>> 98 bgt $18,$byte_loop_tail_dn
>> 99 br $egress
>> 100
>> 101 $memmove_up:
>> 102 mov $16,$4
>> 103 mov $17,$5
>> 104 bne $2,$misaligned_up
>> 105 beq $1,$skip_aligned_byte_loop_head_up
>> 106
>> 107 $aligned_byte_loop_head_up:
>> 108 unop
>> 109 ble $18,$egress
>> 110 ldq_u $3,0($5)
>> 111 ldq_u $2,0($4)
>> 112
>> 113 lda $18,-1($18)
>> 114 extbl $3,$5,$1
>> 115 insbl $1,$4,$1
>> 116 mskbl $2,$4,$2
>> 117
>> 118 bis $1,$2,$1
>> 119 lda $5,1($5)
>> 120 stq_u $1,0($4)
>> 121 lda $4,1($4)
>> 122
>> 123 and $4,7,$6
>> 124 bne $6,$aligned_byte_loop_head_up
>> 125
>> 126 $skip_aligned_byte_loop_head_up:
>> 127 lda $18,-8($18)
>> 128 blt $18,$skip_aligned_word_loop_up
>> 129
>> 130 $aligned_word_loop_up:
>> 131 ldq $1,0($5)
>> 132 nop
>> 133 lda $5,8($5)
>> 134 lda $18,-8($18)
>> 135
>> 136 stq $1,0($4)
>> 137 nop
>> 138 lda $4,8($4)
>> 139 bge $18,$aligned_word_loop_up
>> 140
>> 141 $skip_aligned_word_loop_up:
>> 142 lda $18,8($18)
>> 143 bgt $18,$byte_loop_tail_up
>> 144 unop
>> 145 ret $31,($26),1
>> 146
>> 147 .align 4
>> 148 $misaligned_up:
>> 149 nop
>> 150 fnop
>> 151 unop
>> 152 beq $18,$egress
>> 153
>> 154 $byte_loop_tail_up:
>> 155 ldq_u $3,0($5)
>> 156 ldq_u $2,0($4)
>> 157 lda $18,-1($18)
>> 158 extbl $3,$5,$1
>> 159
>> 160 insbl $1,$4,$1
>> 161 mskbl $2,$4,$2
>> 162 bis $1,$2,$1
>> 163 stq_u $1,0($4)
>> 164
>> 165 lda $5,1($5)
>> 166 lda $4,1($4)
>> 167 nop
>> 168 bgt $18,$byte_loop_tail_up
>> 169
>> 170 $egress:
>> 171 ret $31,($26),1
>> 172 nop
>> 173 nop
>> 174 nop
10 175
11 .text !! 176 .end memmove
12 ENTRY(memmove) /* o0=dst o1=src o2=len */ <<
13 brz,pn %o2, 99f <<
14 mov %o0, %g1 <<
15 <<
16 cmp %o0, %o1 <<
17 bleu,pt %xcc, 2f <<
18 add %o1, %o2, %g7 <<
19 cmp %g7, %o0 <<
20 bleu,pt %xcc, memcpy <<
21 add %o0, %o2, %o5 <<
22 sub %g7, 1, %o1 <<
23 <<
24 sub %o5, 1, %o0 <<
25 1: ldub [%o1], %g7 <<
26 subcc %o2, 1, %o2 <<
27 sub %o1, 1, %o1 <<
28 stb %g7, [%o0] <<
29 bne,pt %icc, 1b <<
30 sub %o0, 1, %o0 <<
31 99: <<
32 retl <<
33 mov %g1, %o0 <<
34 <<
35 /* We can't just call memcpy for these <<
36 * chips the memcpy uses cache initial <<
37 * and src are close enough, those can <<
38 * before we've loaded it in. <<
39 */ <<
40 2: or %o0, %o1, %g7 <<
41 or %o2, %g7, %g7 <<
42 andcc %g7, 0x7, %g0 <<
43 bne,pn %xcc, 4f <<
44 nop <<
45 <<
46 3: ldx [%o1], %g7 <<
47 add %o1, 8, %o1 <<
48 subcc %o2, 8, %o2 <<
49 add %o0, 8, %o0 <<
50 bne,pt %icc, 3b <<
51 stx %g7, [%o0 - 0x8] <<
52 ba,a,pt %xcc, 99b <<
53 <<
54 4: ldub [%o1], %g7 <<
55 add %o1, 1, %o1 <<
56 subcc %o2, 1, %o2 <<
57 add %o0, 1, %o0 <<
58 bne,pt %icc, 4b <<
59 stb %g7, [%o0 - 0x1] <<
60 ba,a,pt %xcc, 99b <<
61 ENDPROC(memmove) <<
62 EXPORT_SYMBOL(memmove) <<
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.