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
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
|
/*
* Fast C2P (Chunky-to-Planar) Conversion
*
* Copyright (C) 2003-2008 Geert Uytterhoeven
*
* NOTES:
* - This code was inspired by Scout's C2P tutorial
* - It assumes to run on a big endian system
*
* This file is subject to the terms and conditions of the GNU General Public
* License. See the file COPYING in the main directory of this archive
* for more details.
*/
#include <linux/build_bug.h>
/*
* Basic transpose step
*/
static inline void _transp(u32 d[], unsigned int i1, unsigned int i2,
unsigned int shift, u32 mask)
{
u32 t = (d[i1] ^ (d[i2] >> shift)) & mask;
d[i1] ^= t;
d[i2] ^= t << shift;
}
static __always_inline u32 get_mask(unsigned int n)
{
switch (n) {
case 1:
return 0x55555555;
case 2:
return 0x33333333;
case 4:
return 0x0f0f0f0f;
case 8:
return 0x00ff00ff;
case 16:
return 0x0000ffff;
}
BUILD_BUG();
return 0;
}
/*
* Transpose operations on 8 32-bit words
*/
static __always_inline void transp8(u32 d[], unsigned int n, unsigned int m)
{
u32 mask = get_mask(n);
switch (m) {
case 1:
/* First n x 1 block */
_transp(d, 0, 1, n, mask);
/* Second n x 1 block */
_transp(d, 2, 3, n, mask);
/* Third n x 1 block */
_transp(d, 4, 5, n, mask);
/* Fourth n x 1 block */
_transp(d, 6, 7, n, mask);
return;
case 2:
/* First n x 2 block */
_transp(d, 0, 2, n, mask);
_transp(d, 1, 3, n, mask);
/* Second n x 2 block */
_transp(d, 4, 6, n, mask);
_transp(d, 5, 7, n, mask);
return;
case 4:
/* Single n x 4 block */
_transp(d, 0, 4, n, mask);
_transp(d, 1, 5, n, mask);
_transp(d, 2, 6, n, mask);
_transp(d, 3, 7, n, mask);
return;
}
BUILD_BUG();
}
/*
* Transpose operations on 4 32-bit words
*/
static __always_inline void transp4(u32 d[], unsigned int n, unsigned int m)
{
u32 mask = get_mask(n);
switch (m) {
case 1:
/* First n x 1 block */
_transp(d, 0, 1, n, mask);
/* Second n x 1 block */
_transp(d, 2, 3, n, mask);
return;
case 2:
/* Single n x 2 block */
_transp(d, 0, 2, n, mask);
_transp(d, 1, 3, n, mask);
return;
}
BUILD_BUG();
}
/*
* Transpose operations on 4 32-bit words (reverse order)
*/
static __always_inline void transp4x(u32 d[], unsigned int n, unsigned int m)
{
u32 mask = get_mask(n);
switch (m) {
case 2:
/* Single n x 2 block */
_transp(d, 2, 0, n, mask);
_transp(d, 3, 1, n, mask);
return;
}
BUILD_BUG();
}
/*
* Compose two values, using a bitmask as decision value
* This is equivalent to (a & mask) | (b & ~mask)
*/
static inline u32 comp(u32 a, u32 b, u32 mask)
{
return ((a ^ b) & mask) ^ b;
}
|
