Line data Source code
1 : // SPDX-License-Identifier: GPL-2.0
2 : /*
3 : * SHA1 routine optimized to do word accesses rather than byte accesses,
4 : * and to avoid unnecessary copies into the context array.
5 : *
6 : * This was based on the git SHA1 implementation.
7 : */
8 :
9 : #include <linux/kernel.h>
10 : #include <linux/export.h>
11 : #include <linux/bitops.h>
12 : #include <crypto/sha1.h>
13 : #include <asm/unaligned.h>
14 :
15 : /*
16 : * If you have 32 registers or more, the compiler can (and should)
17 : * try to change the array[] accesses into registers. However, on
18 : * machines with less than ~25 registers, that won't really work,
19 : * and at least gcc will make an unholy mess of it.
20 : *
21 : * So to avoid that mess which just slows things down, we force
22 : * the stores to memory to actually happen (we might be better off
23 : * with a 'W(t)=(val);asm("":"+m" (W(t))' there instead, as
24 : * suggested by Artur Skawina - that will also make gcc unable to
25 : * try to do the silly "optimize away loads" part because it won't
26 : * see what the value will be).
27 : *
28 : * Ben Herrenschmidt reports that on PPC, the C version comes close
29 : * to the optimized asm with this (ie on PPC you don't want that
30 : * 'volatile', since there are lots of registers).
31 : *
32 : * On ARM we get the best code generation by forcing a full memory barrier
33 : * between each SHA_ROUND, otherwise gcc happily get wild with spilling and
34 : * the stack frame size simply explode and performance goes down the drain.
35 : */
36 :
37 : #ifdef CONFIG_X86
38 : #define setW(x, val) (*(volatile __u32 *)&W(x) = (val))
39 : #elif defined(CONFIG_ARM)
40 : #define setW(x, val) do { W(x) = (val); __asm__("":::"memory"); } while (0)
41 : #else
42 : #define setW(x, val) (W(x) = (val))
43 : #endif
44 :
45 : /* This "rolls" over the 512-bit array */
46 : #define W(x) (array[(x)&15])
47 :
48 : /*
49 : * Where do we get the source from? The first 16 iterations get it from
50 : * the input data, the next mix it from the 512-bit array.
51 : */
52 : #define SHA_SRC(t) get_unaligned_be32((__u32 *)data + t)
53 : #define SHA_MIX(t) rol32(W(t+13) ^ W(t+8) ^ W(t+2) ^ W(t), 1)
54 :
55 : #define SHA_ROUND(t, input, fn, constant, A, B, C, D, E) do { \
56 : __u32 TEMP = input(t); setW(t, TEMP); \
57 : E += TEMP + rol32(A,5) + (fn) + (constant); \
58 : B = ror32(B, 2); } while (0)
59 :
60 : #define T_0_15(t, A, B, C, D, E) SHA_ROUND(t, SHA_SRC, (((C^D)&B)^D) , 0x5a827999, A, B, C, D, E )
61 : #define T_16_19(t, A, B, C, D, E) SHA_ROUND(t, SHA_MIX, (((C^D)&B)^D) , 0x5a827999, A, B, C, D, E )
62 : #define T_20_39(t, A, B, C, D, E) SHA_ROUND(t, SHA_MIX, (B^C^D) , 0x6ed9eba1, A, B, C, D, E )
63 : #define T_40_59(t, A, B, C, D, E) SHA_ROUND(t, SHA_MIX, ((B&C)+(D&(B^C))) , 0x8f1bbcdc, A, B, C, D, E )
64 : #define T_60_79(t, A, B, C, D, E) SHA_ROUND(t, SHA_MIX, (B^C^D) , 0xca62c1d6, A, B, C, D, E )
65 :
66 : /**
67 : * sha1_transform - single block SHA1 transform (deprecated)
68 : *
69 : * @digest: 160 bit digest to update
70 : * @data: 512 bits of data to hash
71 : * @array: 16 words of workspace (see note)
72 : *
73 : * This function executes SHA-1's internal compression function. It updates the
74 : * 160-bit internal state (@digest) with a single 512-bit data block (@data).
75 : *
76 : * Don't use this function. SHA-1 is no longer considered secure. And even if
77 : * you do have to use SHA-1, this isn't the correct way to hash something with
78 : * SHA-1 as this doesn't handle padding and finalization.
79 : *
80 : * Note: If the hash is security sensitive, the caller should be sure
81 : * to clear the workspace. This is left to the caller to avoid
82 : * unnecessary clears between chained hashing operations.
83 : */
84 40 : void sha1_transform(__u32 *digest, const char *data, __u32 *array)
85 : {
86 40 : __u32 A, B, C, D, E;
87 :
88 40 : A = digest[0];
89 40 : B = digest[1];
90 40 : C = digest[2];
91 40 : D = digest[3];
92 40 : E = digest[4];
93 :
94 : /* Round 1 - iterations 0-16 take their input from 'data' */
95 40 : T_0_15( 0, A, B, C, D, E);
96 40 : T_0_15( 1, E, A, B, C, D);
97 40 : T_0_15( 2, D, E, A, B, C);
98 40 : T_0_15( 3, C, D, E, A, B);
99 40 : T_0_15( 4, B, C, D, E, A);
100 40 : T_0_15( 5, A, B, C, D, E);
101 40 : T_0_15( 6, E, A, B, C, D);
102 40 : T_0_15( 7, D, E, A, B, C);
103 40 : T_0_15( 8, C, D, E, A, B);
104 40 : T_0_15( 9, B, C, D, E, A);
105 40 : T_0_15(10, A, B, C, D, E);
106 40 : T_0_15(11, E, A, B, C, D);
107 40 : T_0_15(12, D, E, A, B, C);
108 40 : T_0_15(13, C, D, E, A, B);
109 40 : T_0_15(14, B, C, D, E, A);
110 40 : T_0_15(15, A, B, C, D, E);
111 :
112 : /* Round 1 - tail. Input from 512-bit mixing array */
113 40 : T_16_19(16, E, A, B, C, D);
114 40 : T_16_19(17, D, E, A, B, C);
115 40 : T_16_19(18, C, D, E, A, B);
116 40 : T_16_19(19, B, C, D, E, A);
117 :
118 : /* Round 2 */
119 40 : T_20_39(20, A, B, C, D, E);
120 40 : T_20_39(21, E, A, B, C, D);
121 40 : T_20_39(22, D, E, A, B, C);
122 40 : T_20_39(23, C, D, E, A, B);
123 40 : T_20_39(24, B, C, D, E, A);
124 40 : T_20_39(25, A, B, C, D, E);
125 40 : T_20_39(26, E, A, B, C, D);
126 40 : T_20_39(27, D, E, A, B, C);
127 40 : T_20_39(28, C, D, E, A, B);
128 40 : T_20_39(29, B, C, D, E, A);
129 40 : T_20_39(30, A, B, C, D, E);
130 40 : T_20_39(31, E, A, B, C, D);
131 40 : T_20_39(32, D, E, A, B, C);
132 40 : T_20_39(33, C, D, E, A, B);
133 40 : T_20_39(34, B, C, D, E, A);
134 40 : T_20_39(35, A, B, C, D, E);
135 40 : T_20_39(36, E, A, B, C, D);
136 40 : T_20_39(37, D, E, A, B, C);
137 40 : T_20_39(38, C, D, E, A, B);
138 40 : T_20_39(39, B, C, D, E, A);
139 :
140 : /* Round 3 */
141 40 : T_40_59(40, A, B, C, D, E);
142 40 : T_40_59(41, E, A, B, C, D);
143 40 : T_40_59(42, D, E, A, B, C);
144 40 : T_40_59(43, C, D, E, A, B);
145 40 : T_40_59(44, B, C, D, E, A);
146 40 : T_40_59(45, A, B, C, D, E);
147 40 : T_40_59(46, E, A, B, C, D);
148 40 : T_40_59(47, D, E, A, B, C);
149 40 : T_40_59(48, C, D, E, A, B);
150 40 : T_40_59(49, B, C, D, E, A);
151 40 : T_40_59(50, A, B, C, D, E);
152 40 : T_40_59(51, E, A, B, C, D);
153 40 : T_40_59(52, D, E, A, B, C);
154 40 : T_40_59(53, C, D, E, A, B);
155 40 : T_40_59(54, B, C, D, E, A);
156 40 : T_40_59(55, A, B, C, D, E);
157 40 : T_40_59(56, E, A, B, C, D);
158 40 : T_40_59(57, D, E, A, B, C);
159 40 : T_40_59(58, C, D, E, A, B);
160 40 : T_40_59(59, B, C, D, E, A);
161 :
162 : /* Round 4 */
163 40 : T_60_79(60, A, B, C, D, E);
164 40 : T_60_79(61, E, A, B, C, D);
165 40 : T_60_79(62, D, E, A, B, C);
166 40 : T_60_79(63, C, D, E, A, B);
167 40 : T_60_79(64, B, C, D, E, A);
168 40 : T_60_79(65, A, B, C, D, E);
169 40 : T_60_79(66, E, A, B, C, D);
170 40 : T_60_79(67, D, E, A, B, C);
171 40 : T_60_79(68, C, D, E, A, B);
172 40 : T_60_79(69, B, C, D, E, A);
173 40 : T_60_79(70, A, B, C, D, E);
174 40 : T_60_79(71, E, A, B, C, D);
175 40 : T_60_79(72, D, E, A, B, C);
176 40 : T_60_79(73, C, D, E, A, B);
177 40 : T_60_79(74, B, C, D, E, A);
178 40 : T_60_79(75, A, B, C, D, E);
179 40 : T_60_79(76, E, A, B, C, D);
180 40 : T_60_79(77, D, E, A, B, C);
181 40 : T_60_79(78, C, D, E, A, B);
182 40 : T_60_79(79, B, C, D, E, A);
183 :
184 40 : digest[0] += A;
185 40 : digest[1] += B;
186 40 : digest[2] += C;
187 40 : digest[3] += D;
188 40 : digest[4] += E;
189 40 : }
190 : EXPORT_SYMBOL(sha1_transform);
191 :
192 : /**
193 : * sha1_init - initialize the vectors for a SHA1 digest
194 : * @buf: vector to initialize
195 : */
196 5 : void sha1_init(__u32 *buf)
197 : {
198 5 : buf[0] = 0x67452301;
199 5 : buf[1] = 0xefcdab89;
200 5 : buf[2] = 0x98badcfe;
201 5 : buf[3] = 0x10325476;
202 5 : buf[4] = 0xc3d2e1f0;
203 5 : }
204 : EXPORT_SYMBOL(sha1_init);
|
