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
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
|
/*
* SHA transform optimized for ARM
*
* Copyright: (C) 2005 by Nicolas Pitre <nico@cam.org>
* Created: September 17, 2005
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License version 2 as
* published by the Free Software Foundation.
*/
.text
.globl arm_sha_transform
/*
* void sha_transform(uint32_t *hash, const unsigned char *data, uint32_t *W);
*
* note: the "data" pointer may be unaligned.
*/
arm_sha_transform:
stmfd sp!, {r4 - r8, lr}
@ for (i = 0; i < 16; i++)
@ W[i] = ntohl(((uint32_t *)data)[i]);
#ifdef __ARMEB__
mov r4, r0
mov r0, r2
mov r2, #64
bl memcpy
mov r2, r0
mov r0, r4
#else
mov r3, r2
mov lr, #16
1: ldrb r4, [r1], #1
ldrb r5, [r1], #1
ldrb r6, [r1], #1
ldrb r7, [r1], #1
subs lr, lr, #1
orr r5, r5, r4, lsl #8
orr r6, r6, r5, lsl #8
orr r7, r7, r6, lsl #8
str r7, [r3], #4
bne 1b
#endif
@ for (i = 0; i < 64; i++)
@ W[i+16] = ror(W[i+13] ^ W[i+8] ^ W[i+2] ^ W[i], 31);
sub r3, r2, #4
mov lr, #64
2: ldr r4, [r3, #4]!
subs lr, lr, #1
ldr r5, [r3, #8]
ldr r6, [r3, #32]
ldr r7, [r3, #52]
eor r4, r4, r5
eor r4, r4, r6
eor r4, r4, r7
mov r4, r4, ror #31
str r4, [r3, #64]
bne 2b
/*
* The SHA functions are:
*
* f1(B,C,D) = (D ^ (B & (C ^ D)))
* f2(B,C,D) = (B ^ C ^ D)
* f3(B,C,D) = ((B & C) | (D & (B | C)))
*
* Then the sub-blocks are processed as follows:
*
* A' = ror(A, 27) + f(B,C,D) + E + K + *W++
* B' = A
* C' = ror(B, 2)
* D' = C
* E' = D
*
* We therefore unroll each loop 5 times to avoid register shuffling.
* Also the ror for C (and also D and E which are successivelyderived
* from it) is applied in place to cut on an additional mov insn for
* each round.
*/
.macro sha_f1, A, B, C, D, E
ldr r3, [r2], #4
eor ip, \C, \D
add \E, r1, \E, ror #2
and ip, \B, ip, ror #2
add \E, \E, \A, ror #27
eor ip, ip, \D, ror #2
add \E, \E, r3
add \E, \E, ip
.endm
.macro sha_f2, A, B, C, D, E
ldr r3, [r2], #4
add \E, r1, \E, ror #2
eor ip, \B, \C, ror #2
add \E, \E, \A, ror #27
eor ip, ip, \D, ror #2
add \E, \E, r3
add \E, \E, ip
.endm
.macro sha_f3, A, B, C, D, E
ldr r3, [r2], #4
add \E, r1, \E, ror #2
orr ip, \B, \C, ror #2
add \E, \E, \A, ror #27
and ip, ip, \D, ror #2
add \E, \E, r3
and r3, \B, \C, ror #2
orr ip, ip, r3
add \E, \E, ip
.endm
ldmia r0, {r4 - r8}
mov lr, #4
ldr r1, .L_sha_K + 0
/* adjust initial values */
mov r6, r6, ror #30
mov r7, r7, ror #30
mov r8, r8, ror #30
3: subs lr, lr, #1
sha_f1 r4, r5, r6, r7, r8
sha_f1 r8, r4, r5, r6, r7
sha_f1 r7, r8, r4, r5, r6
sha_f1 r6, r7, r8, r4, r5
sha_f1 r5, r6, r7, r8, r4
bne 3b
ldr r1, .L_sha_K + 4
mov lr, #4
4: subs lr, lr, #1
sha_f2 r4, r5, r6, r7, r8
sha_f2 r8, r4, r5, r6, r7
sha_f2 r7, r8, r4, r5, r6
sha_f2 r6, r7, r8, r4, r5
sha_f2 r5, r6, r7, r8, r4
bne 4b
ldr r1, .L_sha_K + 8
mov lr, #4
5: subs lr, lr, #1
sha_f3 r4, r5, r6, r7, r8
sha_f3 r8, r4, r5, r6, r7
sha_f3 r7, r8, r4, r5, r6
sha_f3 r6, r7, r8, r4, r5
sha_f3 r5, r6, r7, r8, r4
bne 5b
ldr r1, .L_sha_K + 12
mov lr, #4
6: subs lr, lr, #1
sha_f2 r4, r5, r6, r7, r8
sha_f2 r8, r4, r5, r6, r7
sha_f2 r7, r8, r4, r5, r6
sha_f2 r6, r7, r8, r4, r5
sha_f2 r5, r6, r7, r8, r4
bne 6b
ldmia r0, {r1, r2, r3, ip, lr}
add r4, r1, r4
add r5, r2, r5
add r6, r3, r6, ror #2
add r7, ip, r7, ror #2
add r8, lr, r8, ror #2
stmia r0, {r4 - r8}
ldmfd sp!, {r4 - r8, pc}
.L_sha_K:
.word 0x5a827999, 0x6ed9eba1, 0x8f1bbcdc, 0xca62c1d6
|
