1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
|
/*
* "memset" implementation for SH4
*
* Copyright (C) 1999 Niibe Yutaka
* Copyright (c) 2009 STMicroelectronics Limited
* Author: Stuart Menefy <stuart.menefy:st.com>
*/
/*
* void *memset(void *s, int c, size_t n);
*/
#include <linux/linkage.h>
ENTRY(memset)
mov #12,r0
add r6,r4
cmp/gt r6,r0
bt/s 40f ! if it's too small, set a byte at once
mov r4,r0
and #3,r0
cmp/eq #0,r0
bt/s 2f ! It's aligned
sub r0,r6
1:
dt r0
bf/s 1b
mov.b r5,@-r4
2: ! make VVVV
extu.b r5,r5
swap.b r5,r0 ! V0
or r0,r5 ! VV
swap.w r5,r0 ! VV00
or r0,r5 ! VVVV
! Check if enough bytes need to be copied to be worth the big loop
mov #0x40, r0 ! (MT)
cmp/gt r6,r0 ! (MT) 64 > len => slow loop
bt/s 22f
mov r6,r0
! align the dst to the cache block size if necessary
mov r4, r3
mov #~(0x1f), r1
and r3, r1
cmp/eq r3, r1
bt/s 11f ! dst is already aligned
sub r1, r3 ! r3-r1 -> r3
shlr2 r3 ! number of loops
10: mov.l r5,@-r4
dt r3
bf/s 10b
add #-4, r6
11: ! dst is 32byte aligned
mov r6,r2
mov #-5,r0
shld r0,r2 ! number of loops
add #-32, r4
mov r5, r0
12:
movca.l r0,@r4
mov.l r5,@(4, r4)
mov.l r5,@(8, r4)
mov.l r5,@(12,r4)
mov.l r5,@(16,r4)
mov.l r5,@(20,r4)
add #-0x20, r6
mov.l r5,@(24,r4)
dt r2
mov.l r5,@(28,r4)
bf/s 12b
add #-32, r4
add #32, r4
mov #8, r0
cmp/ge r0, r6
bf 40f
mov r6,r0
22:
shlr2 r0
shlr r0 ! r0 = r6 >> 3
3:
dt r0
mov.l r5,@-r4 ! set 8-byte at once
bf/s 3b
mov.l r5,@-r4
!
mov #7,r0
and r0,r6
! fill bytes (length may be zero)
40: tst r6,r6
bt 5f
4:
dt r6
bf/s 4b
mov.b r5,@-r4
5:
rts
mov r4,r0
|
