1/* Copyright (C) 1995-1998 Eric Young (eay@cryptsoft.com)
2 * All rights reserved.
3 *
4 * This package is an SSL implementation written
5 * by Eric Young (eay@cryptsoft.com).
6 * The implementation was written so as to conform with Netscapes SSL.
7 *
8 * This library is free for commercial and non-commercial use as long as
9 * the following conditions are aheared to. The following conditions
10 * apply to all code found in this distribution, be it the RC4, RSA,
11 * lhash, DES, etc., code; not just the SSL code. The SSL documentation
12 * included with this distribution is covered by the same copyright terms
13 * except that the holder is Tim Hudson (tjh@cryptsoft.com).
14 *
15 * Copyright remains Eric Young's, and as such any Copyright notices in
16 * the code are not to be removed.
17 * If this package is used in a product, Eric Young should be given attribution
18 * as the author of the parts of the library used.
19 * This can be in the form of a textual message at program startup or
20 * in documentation (online or textual) provided with the package.
21 *
22 * Redistribution and use in source and binary forms, with or without
23 * modification, are permitted provided that the following conditions
24 * are met:
25 * 1. Redistributions of source code must retain the copyright
26 * notice, this list of conditions and the following disclaimer.
27 * 2. Redistributions in binary form must reproduce the above copyright
28 * notice, this list of conditions and the following disclaimer in the
29 * documentation and/or other materials provided with the distribution.
30 * 3. All advertising materials mentioning features or use of this software
31 * must display the following acknowledgement:
32 * "This product includes cryptographic software written by
33 * Eric Young (eay@cryptsoft.com)"
34 * The word 'cryptographic' can be left out if the rouines from the library
35 * being used are not cryptographic related :-).
36 * 4. If you include any Windows specific code (or a derivative thereof) from
37 * the apps directory (application code) you must include an acknowledgement:
38 * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)"
39 *
40 * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND
41 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
42 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
43 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
44 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
45 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
46 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
47 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
48 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
49 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
50 * SUCH DAMAGE.
51 *
52 * The licence and distribution terms for any publically available version or
53 * derivative of this code cannot be changed. i.e. this code cannot simply be
54 * copied and put under another distribution licence
55 * [including the GNU Public Licence.] */
56
57// Altivec-optimized SHA1 in C. This is tested on ppc64le only.
58//
59// References:
60// https://software.intel.com/en-us/articles/improving-the-performance-of-the-secure-hash-algorithm-1
61// http://arctic.org/~dean/crypto/sha1.html
62//
63// This code used the generic SHA-1 from OpenSSL as a basis and AltiVec
64// optimisations were added on top.
65
66#include <openssl/sha.h>
67
68#if defined(OPENSSL_PPC64LE)
69
70#include <altivec.h>
71
72void sha1_block_data_order(uint32_t *state, const uint8_t *data, size_t num);
73
74static uint32_t rotate(uint32_t a, int n) { return (a << n) | (a >> (32 - n)); }
75
76typedef vector unsigned int vec_uint32_t;
77typedef vector unsigned char vec_uint8_t;
78
79// Vector constants
80static const vec_uint8_t k_swap_endianness = {3, 2, 1, 0, 7, 6, 5, 4,
81 11, 10, 9, 8, 15, 14, 13, 12};
82
83// Shift amounts for byte and bit shifts and rotations
84static const vec_uint8_t k_4_bytes = {32, 32, 32, 32, 32, 32, 32, 32,
85 32, 32, 32, 32, 32, 32, 32, 32};
86static const vec_uint8_t k_12_bytes = {96, 96, 96, 96, 96, 96, 96, 96,
87 96, 96, 96, 96, 96, 96, 96, 96};
88
89#define K_00_19 0x5a827999UL
90#define K_20_39 0x6ed9eba1UL
91#define K_40_59 0x8f1bbcdcUL
92#define K_60_79 0xca62c1d6UL
93
94// Vector versions of the above.
95static const vec_uint32_t K_00_19_x_4 = {K_00_19, K_00_19, K_00_19, K_00_19};
96static const vec_uint32_t K_20_39_x_4 = {K_20_39, K_20_39, K_20_39, K_20_39};
97static const vec_uint32_t K_40_59_x_4 = {K_40_59, K_40_59, K_40_59, K_40_59};
98static const vec_uint32_t K_60_79_x_4 = {K_60_79, K_60_79, K_60_79, K_60_79};
99
100// vector message scheduling: compute message schedule for round i..i+3 where i
101// is divisible by 4. We return the schedule w[i..i+3] as a vector. In
102// addition, we also precompute sum w[i..+3] and an additive constant K. This
103// is done to offload some computation of f() in the integer execution units.
104//
105// Byte shifting code below may not be correct for big-endian systems.
106static vec_uint32_t sched_00_15(vec_uint32_t *pre_added, const void *data,
107 vec_uint32_t k) {
108 const vector unsigned char unaligned_data =
109 vec_vsx_ld(0, (const unsigned char*) data);
110 const vec_uint32_t v = (vec_uint32_t) unaligned_data;
111 const vec_uint32_t w = vec_perm(v, v, k_swap_endianness);
112 vec_st(w + k, 0, pre_added);
113 return w;
114}
115
116// Compute w[i..i+3] using these steps for i in [16, 20, 24, 28]
117//
118// w'[i ] = (w[i-3] ^ w[i-8] ^ w[i-14] ^ w[i-16]) <<< 1
119// w'[i+1] = (w[i-2] ^ w[i-7] ^ w[i-13] ^ w[i-15]) <<< 1
120// w'[i+2] = (w[i-1] ^ w[i-6] ^ w[i-12] ^ w[i-14]) <<< 1
121// w'[i+3] = ( 0 ^ w[i-5] ^ w[i-11] ^ w[i-13]) <<< 1
122//
123// w[ i] = w'[ i]
124// w[i+1] = w'[i+1]
125// w[i+2] = w'[i+2]
126// w[i+3] = w'[i+3] ^ (w'[i] <<< 1)
127static vec_uint32_t sched_16_31(vec_uint32_t *pre_added, vec_uint32_t minus_4,
128 vec_uint32_t minus_8, vec_uint32_t minus_12,
129 vec_uint32_t minus_16, vec_uint32_t k) {
130 const vec_uint32_t minus_3 = vec_sro(minus_4, k_4_bytes);
131 const vec_uint32_t minus_14 = vec_sld((minus_12), (minus_16), 8);
132 const vec_uint32_t k_1_bit = vec_splat_u32(1);
133 const vec_uint32_t w_prime =
134 vec_rl(minus_3 ^ minus_8 ^ minus_14 ^ minus_16, k_1_bit);
135 const vec_uint32_t w =
136 w_prime ^ vec_rl(vec_slo(w_prime, k_12_bytes), k_1_bit);
137 vec_st(w + k, 0, pre_added);
138 return w;
139}
140
141// Compute w[i..i+3] using this relation for i in [32, 36, 40 ... 76]
142// w[i] = (w[i-6] ^ w[i-16] ^ w[i-28] ^ w[i-32]), 2) <<< 2
143static vec_uint32_t sched_32_79(vec_uint32_t *pre_added, vec_uint32_t minus_4,
144 vec_uint32_t minus_8, vec_uint32_t minus_16,
145 vec_uint32_t minus_28, vec_uint32_t minus_32,
146 vec_uint32_t k) {
147 const vec_uint32_t minus_6 = vec_sld(minus_4, minus_8, 8);
148 const vec_uint32_t k_2_bits = vec_splat_u32(2);
149 const vec_uint32_t w =
150 vec_rl(minus_6 ^ minus_16 ^ minus_28 ^ minus_32, k_2_bits);
151 vec_st(w + k, 0, pre_added);
152 return w;
153}
154
155// As pointed out by Wei Dai <weidai@eskimo.com>, F() below can be simplified
156// to the code in F_00_19. Wei attributes these optimisations to Peter
157// Gutmann's SHS code, and he attributes it to Rich Schroeppel. #define
158// F(x,y,z) (((x) & (y)) | ((~(x)) & (z))) I've just become aware of another
159// tweak to be made, again from Wei Dai, in F_40_59, (x&a)|(y&a) -> (x|y)&a
160#define F_00_19(b, c, d) ((((c) ^ (d)) & (b)) ^ (d))
161#define F_20_39(b, c, d) ((b) ^ (c) ^ (d))
162#define F_40_59(b, c, d) (((b) & (c)) | (((b) | (c)) & (d)))
163#define F_60_79(b, c, d) F_20_39(b, c, d)
164
165// We pre-added the K constants during message scheduling.
166#define BODY_00_19(i, a, b, c, d, e, f) \
167 do { \
168 (f) = w[i] + (e) + rotate((a), 5) + F_00_19((b), (c), (d)); \
169 (b) = rotate((b), 30); \
170 } while (0)
171
172#define BODY_20_39(i, a, b, c, d, e, f) \
173 do { \
174 (f) = w[i] + (e) + rotate((a), 5) + F_20_39((b), (c), (d)); \
175 (b) = rotate((b), 30); \
176 } while (0)
177
178#define BODY_40_59(i, a, b, c, d, e, f) \
179 do { \
180 (f) = w[i] + (e) + rotate((a), 5) + F_40_59((b), (c), (d)); \
181 (b) = rotate((b), 30); \
182 } while (0)
183
184#define BODY_60_79(i, a, b, c, d, e, f) \
185 do { \
186 (f) = w[i] + (e) + rotate((a), 5) + F_60_79((b), (c), (d)); \
187 (b) = rotate((b), 30); \
188 } while (0)
189
190void sha1_block_data_order(uint32_t *state, const uint8_t *data, size_t num) {
191 uint32_t A, B, C, D, E, T;
192
193 A = state[0];
194 B = state[1];
195 C = state[2];
196 D = state[3];
197 E = state[4];
198
199 for (;;) {
200 vec_uint32_t vw[20];
201 const uint32_t *w = (const uint32_t *)&vw;
202
203 vec_uint32_t k = K_00_19_x_4;
204 const vec_uint32_t w0 = sched_00_15(vw + 0, data + 0, k);
205 BODY_00_19(0, A, B, C, D, E, T);
206 BODY_00_19(1, T, A, B, C, D, E);
207 BODY_00_19(2, E, T, A, B, C, D);
208 BODY_00_19(3, D, E, T, A, B, C);
209
210 const vec_uint32_t w4 = sched_00_15(vw + 1, data + 16, k);
211 BODY_00_19(4, C, D, E, T, A, B);
212 BODY_00_19(5, B, C, D, E, T, A);
213 BODY_00_19(6, A, B, C, D, E, T);
214 BODY_00_19(7, T, A, B, C, D, E);
215
216 const vec_uint32_t w8 = sched_00_15(vw + 2, data + 32, k);
217 BODY_00_19(8, E, T, A, B, C, D);
218 BODY_00_19(9, D, E, T, A, B, C);
219 BODY_00_19(10, C, D, E, T, A, B);
220 BODY_00_19(11, B, C, D, E, T, A);
221
222 const vec_uint32_t w12 = sched_00_15(vw + 3, data + 48, k);
223 BODY_00_19(12, A, B, C, D, E, T);
224 BODY_00_19(13, T, A, B, C, D, E);
225 BODY_00_19(14, E, T, A, B, C, D);
226 BODY_00_19(15, D, E, T, A, B, C);
227
228 const vec_uint32_t w16 = sched_16_31(vw + 4, w12, w8, w4, w0, k);
229 BODY_00_19(16, C, D, E, T, A, B);
230 BODY_00_19(17, B, C, D, E, T, A);
231 BODY_00_19(18, A, B, C, D, E, T);
232 BODY_00_19(19, T, A, B, C, D, E);
233
234 k = K_20_39_x_4;
235 const vec_uint32_t w20 = sched_16_31(vw + 5, w16, w12, w8, w4, k);
236 BODY_20_39(20, E, T, A, B, C, D);
237 BODY_20_39(21, D, E, T, A, B, C);
238 BODY_20_39(22, C, D, E, T, A, B);
239 BODY_20_39(23, B, C, D, E, T, A);
240
241 const vec_uint32_t w24 = sched_16_31(vw + 6, w20, w16, w12, w8, k);
242 BODY_20_39(24, A, B, C, D, E, T);
243 BODY_20_39(25, T, A, B, C, D, E);
244 BODY_20_39(26, E, T, A, B, C, D);
245 BODY_20_39(27, D, E, T, A, B, C);
246
247 const vec_uint32_t w28 = sched_16_31(vw + 7, w24, w20, w16, w12, k);
248 BODY_20_39(28, C, D, E, T, A, B);
249 BODY_20_39(29, B, C, D, E, T, A);
250 BODY_20_39(30, A, B, C, D, E, T);
251 BODY_20_39(31, T, A, B, C, D, E);
252
253 const vec_uint32_t w32 = sched_32_79(vw + 8, w28, w24, w16, w4, w0, k);
254 BODY_20_39(32, E, T, A, B, C, D);
255 BODY_20_39(33, D, E, T, A, B, C);
256 BODY_20_39(34, C, D, E, T, A, B);
257 BODY_20_39(35, B, C, D, E, T, A);
258
259 const vec_uint32_t w36 = sched_32_79(vw + 9, w32, w28, w20, w8, w4, k);
260 BODY_20_39(36, A, B, C, D, E, T);
261 BODY_20_39(37, T, A, B, C, D, E);
262 BODY_20_39(38, E, T, A, B, C, D);
263 BODY_20_39(39, D, E, T, A, B, C);
264
265 k = K_40_59_x_4;
266 const vec_uint32_t w40 = sched_32_79(vw + 10, w36, w32, w24, w12, w8, k);
267 BODY_40_59(40, C, D, E, T, A, B);
268 BODY_40_59(41, B, C, D, E, T, A);
269 BODY_40_59(42, A, B, C, D, E, T);
270 BODY_40_59(43, T, A, B, C, D, E);
271
272 const vec_uint32_t w44 = sched_32_79(vw + 11, w40, w36, w28, w16, w12, k);
273 BODY_40_59(44, E, T, A, B, C, D);
274 BODY_40_59(45, D, E, T, A, B, C);
275 BODY_40_59(46, C, D, E, T, A, B);
276 BODY_40_59(47, B, C, D, E, T, A);
277
278 const vec_uint32_t w48 = sched_32_79(vw + 12, w44, w40, w32, w20, w16, k);
279 BODY_40_59(48, A, B, C, D, E, T);
280 BODY_40_59(49, T, A, B, C, D, E);
281 BODY_40_59(50, E, T, A, B, C, D);
282 BODY_40_59(51, D, E, T, A, B, C);
283
284 const vec_uint32_t w52 = sched_32_79(vw + 13, w48, w44, w36, w24, w20, k);
285 BODY_40_59(52, C, D, E, T, A, B);
286 BODY_40_59(53, B, C, D, E, T, A);
287 BODY_40_59(54, A, B, C, D, E, T);
288 BODY_40_59(55, T, A, B, C, D, E);
289
290 const vec_uint32_t w56 = sched_32_79(vw + 14, w52, w48, w40, w28, w24, k);
291 BODY_40_59(56, E, T, A, B, C, D);
292 BODY_40_59(57, D, E, T, A, B, C);
293 BODY_40_59(58, C, D, E, T, A, B);
294 BODY_40_59(59, B, C, D, E, T, A);
295
296 k = K_60_79_x_4;
297 const vec_uint32_t w60 = sched_32_79(vw + 15, w56, w52, w44, w32, w28, k);
298 BODY_60_79(60, A, B, C, D, E, T);
299 BODY_60_79(61, T, A, B, C, D, E);
300 BODY_60_79(62, E, T, A, B, C, D);
301 BODY_60_79(63, D, E, T, A, B, C);
302
303 const vec_uint32_t w64 = sched_32_79(vw + 16, w60, w56, w48, w36, w32, k);
304 BODY_60_79(64, C, D, E, T, A, B);
305 BODY_60_79(65, B, C, D, E, T, A);
306 BODY_60_79(66, A, B, C, D, E, T);
307 BODY_60_79(67, T, A, B, C, D, E);
308
309 const vec_uint32_t w68 = sched_32_79(vw + 17, w64, w60, w52, w40, w36, k);
310 BODY_60_79(68, E, T, A, B, C, D);
311 BODY_60_79(69, D, E, T, A, B, C);
312 BODY_60_79(70, C, D, E, T, A, B);
313 BODY_60_79(71, B, C, D, E, T, A);
314
315 const vec_uint32_t w72 = sched_32_79(vw + 18, w68, w64, w56, w44, w40, k);
316 BODY_60_79(72, A, B, C, D, E, T);
317 BODY_60_79(73, T, A, B, C, D, E);
318 BODY_60_79(74, E, T, A, B, C, D);
319 BODY_60_79(75, D, E, T, A, B, C);
320
321 // We don't use the last value
322 (void)sched_32_79(vw + 19, w72, w68, w60, w48, w44, k);
323 BODY_60_79(76, C, D, E, T, A, B);
324 BODY_60_79(77, B, C, D, E, T, A);
325 BODY_60_79(78, A, B, C, D, E, T);
326 BODY_60_79(79, T, A, B, C, D, E);
327
328 const uint32_t mask = 0xffffffffUL;
329 state[0] = (state[0] + E) & mask;
330 state[1] = (state[1] + T) & mask;
331 state[2] = (state[2] + A) & mask;
332 state[3] = (state[3] + B) & mask;
333 state[4] = (state[4] + C) & mask;
334
335 data += 64;
336 if (--num == 0) {
337 break;
338 }
339
340 A = state[0];
341 B = state[1];
342 C = state[2];
343 D = state[3];
344 E = state[4];
345 }
346}
347
348#endif // OPENSSL_PPC64LE
349
350#undef K_00_19
351#undef K_20_39
352#undef K_40_59
353#undef K_60_79
354#undef F_00_19
355#undef F_20_39
356#undef F_40_59
357#undef F_60_79
358#undef BODY_00_19
359#undef BODY_20_39
360#undef BODY_40_59
361#undef BODY_60_79
362