1 /*
2 * Based on the git SHA1 Implementation.
3 *
4 * Copyright (C) 2009-2015, Linus Torvalds and others.
5 *
6 * SHA1 routine optimized to do word accesses rather than byte accesses,
7 * and to avoid unnecessary copies into the context array.
8 *
9 * This was initially based on the Mozilla SHA1 implementation, although
10 * none of the original Mozilla code remains.
11 *
12 * This library is free software; you can redistribute it and/or
13 * modify it under the terms of the GNU Lesser General Public
14 * License as published by the Free Software Foundation, version 2.1
15 * exclusively.
16 *
17 * This library is distributed in the hope that it will be useful,
18 * but WITHOUT ANY WARRANTY; without even the implied warranty of
19 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
20 * Lesser General Public License for more details.
21 *
22 * You should have received a copy of the GNU Lesser General Public
23 * License along with this library; if not, write to the Free Software
24 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
25 */
26
27 /* this is only to get definitions for memcpy(), ntohl() and htonl() */
28 #include <string.h>
29 #include <inttypes.h>
30 #include <arpa/inet.h>
31
32 #include <import/sha1.h>
33
34 /*
35 * Performance might be improved if the CPU architecture is OK with
36 * unaligned 32-bit loads and a fast ntohl() is available.
37 * Otherwise fall back to byte loads and shifts which is portable,
38 * and is faster on architectures with memory alignment issues.
39 */
40
41 #if defined(__i386__) || defined(__x86_64__) || \
42 defined(__ppc__) || defined(__ppc64__) || \
43 defined(__powerpc__) || defined(__powerpc64__) || \
44 defined(__s390__) || defined(__s390x__)
45
46 #define get_be32(p) ntohl(*(unsigned int *)(p))
47 #define put_be32(p, v) do { *(unsigned int *)(p) = htonl(v); } while (0)
48
49 #else
50
get_be32(const void * ptr)51 static inline uint32_t get_be32(const void *ptr)
52 {
53 const unsigned char *p = ptr;
54 return (uint32_t)p[0] << 24 |
55 (uint32_t)p[1] << 16 |
56 (uint32_t)p[2] << 8 |
57 (uint32_t)p[3] << 0;
58 }
59
put_be32(void * ptr,uint32_t value)60 static inline void put_be32(void *ptr, uint32_t value)
61 {
62 unsigned char *p = ptr;
63 p[0] = value >> 24;
64 p[1] = value >> 16;
65 p[2] = value >> 8;
66 p[3] = value >> 0;
67 }
68
69 #endif
70
71 #if defined(__GNUC__) && (defined(__i386__) || defined(__x86_64__))
72
73 /*
74 * Force usage of rol or ror by selecting the one with the smaller constant.
75 * It _can_ generate slightly smaller code (a constant of 1 is special), but
76 * perhaps more importantly it's possibly faster on any uarch that does a
77 * rotate with a loop.
78 */
79
80 #define SHA_ASM(op, x, n) ({ unsigned int __res; __asm__(op " %1,%0":"=r" (__res):"i" (n), "0" (x)); __res; })
81 #define SHA_ROL(x,n) SHA_ASM("rol", x, n)
82 #define SHA_ROR(x,n) SHA_ASM("ror", x, n)
83
84 #else
85
86 #define SHA_ROT(X,l,r) (((X) << (l)) | ((X) >> (r)))
87 #define SHA_ROL(X,n) SHA_ROT(X,n,32-(n))
88 #define SHA_ROR(X,n) SHA_ROT(X,32-(n),n)
89
90 #endif
91
92 /*
93 * If you have 32 registers or more, the compiler can (and should)
94 * try to change the array[] accesses into registers. However, on
95 * machines with less than ~25 registers, that won't really work,
96 * and at least gcc will make an unholy mess of it.
97 *
98 * So to avoid that mess which just slows things down, we force
99 * the stores to memory to actually happen (we might be better off
100 * with a 'W(t)=(val);asm("":"+m" (W(t))' there instead, as
101 * suggested by Artur Skawina - that will also make gcc unable to
102 * try to do the silly "optimize away loads" part because it won't
103 * see what the value will be).
104 *
105 * Ben Herrenschmidt reports that on PPC, the C version comes close
106 * to the optimized asm with this (ie on PPC you don't want that
107 * 'volatile', since there are lots of registers).
108 *
109 * On ARM we get the best code generation by forcing a full memory barrier
110 * between each SHA_ROUND, otherwise gcc happily get wild with spilling and
111 * the stack frame size simply explode and performance goes down the drain.
112 */
113
114 #if defined(__i386__) || defined(__x86_64__)
115 #define setW(x, val) (*(volatile unsigned int *)&W(x) = (val))
116 #elif defined(__GNUC__) && defined(__arm__)
117 #define setW(x, val) do { W(x) = (val); __asm__("":::"memory"); } while (0)
118 #else
119 #define setW(x, val) (W(x) = (val))
120 #endif
121
122 /* This "rolls" over the 512-bit array */
123 #define W(x) (array[(x)&15])
124
125 /*
126 * Where do we get the source from? The first 16 iterations get it from
127 * the input data, the next mix it from the 512-bit array.
128 */
129 #define SHA_SRC(t) get_be32((unsigned char *) block + (t)*4)
130 #define SHA_MIX(t) SHA_ROL(W((t)+13) ^ W((t)+8) ^ W((t)+2) ^ W(t), 1);
131
132 #define SHA_ROUND(t, input, fn, constant, A, B, C, D, E) do { \
133 unsigned int TEMP = input(t); setW(t, TEMP); \
134 E += TEMP + SHA_ROL(A,5) + (fn) + (constant); \
135 B = SHA_ROR(B, 2); } while (0)
136
137 #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 )
138 #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 )
139 #define T_20_39(t, A, B, C, D, E) SHA_ROUND(t, SHA_MIX, (B^C^D) , 0x6ed9eba1, A, B, C, D, E )
140 #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 )
141 #define T_60_79(t, A, B, C, D, E) SHA_ROUND(t, SHA_MIX, (B^C^D) , 0xca62c1d6, A, B, C, D, E )
142
blk_SHA1_Block(blk_SHA_CTX * ctx,const void * block)143 static void blk_SHA1_Block(blk_SHA_CTX *ctx, const void *block)
144 {
145 unsigned int A,B,C,D,E;
146 unsigned int array[16];
147
148 A = ctx->H[0];
149 B = ctx->H[1];
150 C = ctx->H[2];
151 D = ctx->H[3];
152 E = ctx->H[4];
153
154 /* Round 1 - iterations 0-16 take their input from 'block' */
155 T_0_15( 0, A, B, C, D, E);
156 T_0_15( 1, E, A, B, C, D);
157 T_0_15( 2, D, E, A, B, C);
158 T_0_15( 3, C, D, E, A, B);
159 T_0_15( 4, B, C, D, E, A);
160 T_0_15( 5, A, B, C, D, E);
161 T_0_15( 6, E, A, B, C, D);
162 T_0_15( 7, D, E, A, B, C);
163 T_0_15( 8, C, D, E, A, B);
164 T_0_15( 9, B, C, D, E, A);
165 T_0_15(10, A, B, C, D, E);
166 T_0_15(11, E, A, B, C, D);
167 T_0_15(12, D, E, A, B, C);
168 T_0_15(13, C, D, E, A, B);
169 T_0_15(14, B, C, D, E, A);
170 T_0_15(15, A, B, C, D, E);
171
172 /* Round 1 - tail. Input from 512-bit mixing array */
173 T_16_19(16, E, A, B, C, D);
174 T_16_19(17, D, E, A, B, C);
175 T_16_19(18, C, D, E, A, B);
176 T_16_19(19, B, C, D, E, A);
177
178 /* Round 2 */
179 T_20_39(20, A, B, C, D, E);
180 T_20_39(21, E, A, B, C, D);
181 T_20_39(22, D, E, A, B, C);
182 T_20_39(23, C, D, E, A, B);
183 T_20_39(24, B, C, D, E, A);
184 T_20_39(25, A, B, C, D, E);
185 T_20_39(26, E, A, B, C, D);
186 T_20_39(27, D, E, A, B, C);
187 T_20_39(28, C, D, E, A, B);
188 T_20_39(29, B, C, D, E, A);
189 T_20_39(30, A, B, C, D, E);
190 T_20_39(31, E, A, B, C, D);
191 T_20_39(32, D, E, A, B, C);
192 T_20_39(33, C, D, E, A, B);
193 T_20_39(34, B, C, D, E, A);
194 T_20_39(35, A, B, C, D, E);
195 T_20_39(36, E, A, B, C, D);
196 T_20_39(37, D, E, A, B, C);
197 T_20_39(38, C, D, E, A, B);
198 T_20_39(39, B, C, D, E, A);
199
200 /* Round 3 */
201 T_40_59(40, A, B, C, D, E);
202 T_40_59(41, E, A, B, C, D);
203 T_40_59(42, D, E, A, B, C);
204 T_40_59(43, C, D, E, A, B);
205 T_40_59(44, B, C, D, E, A);
206 T_40_59(45, A, B, C, D, E);
207 T_40_59(46, E, A, B, C, D);
208 T_40_59(47, D, E, A, B, C);
209 T_40_59(48, C, D, E, A, B);
210 T_40_59(49, B, C, D, E, A);
211 T_40_59(50, A, B, C, D, E);
212 T_40_59(51, E, A, B, C, D);
213 T_40_59(52, D, E, A, B, C);
214 T_40_59(53, C, D, E, A, B);
215 T_40_59(54, B, C, D, E, A);
216 T_40_59(55, A, B, C, D, E);
217 T_40_59(56, E, A, B, C, D);
218 T_40_59(57, D, E, A, B, C);
219 T_40_59(58, C, D, E, A, B);
220 T_40_59(59, B, C, D, E, A);
221
222 /* Round 4 */
223 T_60_79(60, A, B, C, D, E);
224 T_60_79(61, E, A, B, C, D);
225 T_60_79(62, D, E, A, B, C);
226 T_60_79(63, C, D, E, A, B);
227 T_60_79(64, B, C, D, E, A);
228 T_60_79(65, A, B, C, D, E);
229 T_60_79(66, E, A, B, C, D);
230 T_60_79(67, D, E, A, B, C);
231 T_60_79(68, C, D, E, A, B);
232 T_60_79(69, B, C, D, E, A);
233 T_60_79(70, A, B, C, D, E);
234 T_60_79(71, E, A, B, C, D);
235 T_60_79(72, D, E, A, B, C);
236 T_60_79(73, C, D, E, A, B);
237 T_60_79(74, B, C, D, E, A);
238 T_60_79(75, A, B, C, D, E);
239 T_60_79(76, E, A, B, C, D);
240 T_60_79(77, D, E, A, B, C);
241 T_60_79(78, C, D, E, A, B);
242 T_60_79(79, B, C, D, E, A);
243
244 ctx->H[0] += A;
245 ctx->H[1] += B;
246 ctx->H[2] += C;
247 ctx->H[3] += D;
248 ctx->H[4] += E;
249 }
250
blk_SHA1_Init(blk_SHA_CTX * ctx)251 void blk_SHA1_Init(blk_SHA_CTX *ctx)
252 {
253 ctx->size = 0;
254
255 /* Initialize H with the magic constants (see FIPS180 for constants) */
256 ctx->H[0] = 0x67452301;
257 ctx->H[1] = 0xefcdab89;
258 ctx->H[2] = 0x98badcfe;
259 ctx->H[3] = 0x10325476;
260 ctx->H[4] = 0xc3d2e1f0;
261 }
262
blk_SHA1_Update(blk_SHA_CTX * ctx,const void * data,unsigned long len)263 void blk_SHA1_Update(blk_SHA_CTX *ctx, const void *data, unsigned long len)
264 {
265 unsigned int lenW = ctx->size & 63;
266
267 ctx->size += len;
268
269 /* Read the data into W and process blocks as they get full */
270 if (lenW) {
271 unsigned int left = 64 - lenW;
272 if (len < left)
273 left = len;
274 memcpy(lenW + (char *)ctx->W, data, left);
275 lenW = (lenW + left) & 63;
276 len -= left;
277 data = ((const char *)data + left);
278 if (lenW)
279 return;
280 blk_SHA1_Block(ctx, ctx->W);
281 }
282 while (len >= 64) {
283 blk_SHA1_Block(ctx, data);
284 data = ((const char *)data + 64);
285 len -= 64;
286 }
287 if (len)
288 memcpy(ctx->W, data, len);
289 }
290
blk_SHA1_Final(unsigned char hashout[20],blk_SHA_CTX * ctx)291 void blk_SHA1_Final(unsigned char hashout[20], blk_SHA_CTX *ctx)
292 {
293 static const unsigned char pad[64] = { 0x80 };
294 unsigned int padlen[2];
295 int i;
296
297 /* Pad with a binary 1 (ie 0x80), then zeroes, then length */
298 padlen[0] = htonl((uint32_t)(ctx->size >> 29));
299 padlen[1] = htonl((uint32_t)(ctx->size << 3));
300
301 i = ctx->size & 63;
302 blk_SHA1_Update(ctx, pad, 1 + (63 & (55 - i)));
303 blk_SHA1_Update(ctx, padlen, 8);
304
305 /* Output hash */
306 for (i = 0; i < 5; i++)
307 put_be32(hashout + i * 4, ctx->H[i]);
308 }
309