xref: /aosp_15_r20/external/boringssl/src/crypto/fipsmodule/sha/sha512.c (revision 8fb009dc861624b67b6cdb62ea21f0f22d0c584b)
1 /* Copyright (C) 1995-1998 Eric Young ([email protected])
2  * All rights reserved.
3  *
4  * This package is an SSL implementation written
5  * by Eric Young ([email protected]).
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 ([email protected]).
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 ([email protected])"
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 ([email protected])"
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 #include <openssl/sha.h>
58 
59 #include <string.h>
60 
61 #include <openssl/mem.h>
62 
63 #include "../../internal.h"
64 #include "../service_indicator/internal.h"
65 #include "internal.h"
66 
67 
68 // The 32-bit hash algorithms share a common byte-order neutral collector and
69 // padding function implementations that operate on unaligned data,
70 // ../digest/md32_common.h. SHA-512 is the only 64-bit hash algorithm, as of
71 // this writing, so there is no need for a common collector/padding
72 // implementation yet.
73 
74 static int sha512_final_impl(uint8_t *out, size_t md_len, SHA512_CTX *sha);
75 
SHA384_Init(SHA512_CTX * sha)76 int SHA384_Init(SHA512_CTX *sha) {
77   sha->h[0] = UINT64_C(0xcbbb9d5dc1059ed8);
78   sha->h[1] = UINT64_C(0x629a292a367cd507);
79   sha->h[2] = UINT64_C(0x9159015a3070dd17);
80   sha->h[3] = UINT64_C(0x152fecd8f70e5939);
81   sha->h[4] = UINT64_C(0x67332667ffc00b31);
82   sha->h[5] = UINT64_C(0x8eb44a8768581511);
83   sha->h[6] = UINT64_C(0xdb0c2e0d64f98fa7);
84   sha->h[7] = UINT64_C(0x47b5481dbefa4fa4);
85 
86   sha->Nl = 0;
87   sha->Nh = 0;
88   sha->num = 0;
89   sha->md_len = SHA384_DIGEST_LENGTH;
90   return 1;
91 }
92 
93 
SHA512_Init(SHA512_CTX * sha)94 int SHA512_Init(SHA512_CTX *sha) {
95   sha->h[0] = UINT64_C(0x6a09e667f3bcc908);
96   sha->h[1] = UINT64_C(0xbb67ae8584caa73b);
97   sha->h[2] = UINT64_C(0x3c6ef372fe94f82b);
98   sha->h[3] = UINT64_C(0xa54ff53a5f1d36f1);
99   sha->h[4] = UINT64_C(0x510e527fade682d1);
100   sha->h[5] = UINT64_C(0x9b05688c2b3e6c1f);
101   sha->h[6] = UINT64_C(0x1f83d9abfb41bd6b);
102   sha->h[7] = UINT64_C(0x5be0cd19137e2179);
103 
104   sha->Nl = 0;
105   sha->Nh = 0;
106   sha->num = 0;
107   sha->md_len = SHA512_DIGEST_LENGTH;
108   return 1;
109 }
110 
SHA512_256_Init(SHA512_CTX * sha)111 int SHA512_256_Init(SHA512_CTX *sha) {
112   sha->h[0] = UINT64_C(0x22312194fc2bf72c);
113   sha->h[1] = UINT64_C(0x9f555fa3c84c64c2);
114   sha->h[2] = UINT64_C(0x2393b86b6f53b151);
115   sha->h[3] = UINT64_C(0x963877195940eabd);
116   sha->h[4] = UINT64_C(0x96283ee2a88effe3);
117   sha->h[5] = UINT64_C(0xbe5e1e2553863992);
118   sha->h[6] = UINT64_C(0x2b0199fc2c85b8aa);
119   sha->h[7] = UINT64_C(0x0eb72ddc81c52ca2);
120 
121   sha->Nl = 0;
122   sha->Nh = 0;
123   sha->num = 0;
124   sha->md_len = SHA512_256_DIGEST_LENGTH;
125   return 1;
126 }
127 
SHA384(const uint8_t * data,size_t len,uint8_t out[SHA384_DIGEST_LENGTH])128 uint8_t *SHA384(const uint8_t *data, size_t len,
129                 uint8_t out[SHA384_DIGEST_LENGTH]) {
130   SHA512_CTX ctx;
131   SHA384_Init(&ctx);
132   SHA384_Update(&ctx, data, len);
133   SHA384_Final(out, &ctx);
134   OPENSSL_cleanse(&ctx, sizeof(ctx));
135   return out;
136 }
137 
SHA512(const uint8_t * data,size_t len,uint8_t out[SHA512_DIGEST_LENGTH])138 uint8_t *SHA512(const uint8_t *data, size_t len,
139                 uint8_t out[SHA512_DIGEST_LENGTH]) {
140   SHA512_CTX ctx;
141   SHA512_Init(&ctx);
142   SHA512_Update(&ctx, data, len);
143   SHA512_Final(out, &ctx);
144   OPENSSL_cleanse(&ctx, sizeof(ctx));
145   return out;
146 }
147 
SHA512_256(const uint8_t * data,size_t len,uint8_t out[SHA512_256_DIGEST_LENGTH])148 uint8_t *SHA512_256(const uint8_t *data, size_t len,
149                     uint8_t out[SHA512_256_DIGEST_LENGTH]) {
150   SHA512_CTX ctx;
151   SHA512_256_Init(&ctx);
152   SHA512_256_Update(&ctx, data, len);
153   SHA512_256_Final(out, &ctx);
154   OPENSSL_cleanse(&ctx, sizeof(ctx));
155   return out;
156 }
157 
158 #if !defined(SHA512_ASM)
159 static void sha512_block_data_order(uint64_t state[8], const uint8_t *in,
160                                     size_t num_blocks);
161 #endif
162 
163 
SHA384_Final(uint8_t out[SHA384_DIGEST_LENGTH],SHA512_CTX * sha)164 int SHA384_Final(uint8_t out[SHA384_DIGEST_LENGTH], SHA512_CTX *sha) {
165   // This function must be paired with |SHA384_Init|, which sets |sha->md_len|
166   // to |SHA384_DIGEST_LENGTH|.
167   assert(sha->md_len == SHA384_DIGEST_LENGTH);
168   return sha512_final_impl(out, SHA384_DIGEST_LENGTH, sha);
169 }
170 
SHA384_Update(SHA512_CTX * sha,const void * data,size_t len)171 int SHA384_Update(SHA512_CTX *sha, const void *data, size_t len) {
172   return SHA512_Update(sha, data, len);
173 }
174 
SHA512_256_Update(SHA512_CTX * sha,const void * data,size_t len)175 int SHA512_256_Update(SHA512_CTX *sha, const void *data, size_t len) {
176   return SHA512_Update(sha, data, len);
177 }
178 
SHA512_256_Final(uint8_t out[SHA512_256_DIGEST_LENGTH],SHA512_CTX * sha)179 int SHA512_256_Final(uint8_t out[SHA512_256_DIGEST_LENGTH], SHA512_CTX *sha) {
180   // This function must be paired with |SHA512_256_Init|, which sets
181   // |sha->md_len| to |SHA512_256_DIGEST_LENGTH|.
182   assert(sha->md_len == SHA512_256_DIGEST_LENGTH);
183   return sha512_final_impl(out, SHA512_256_DIGEST_LENGTH, sha);
184 }
185 
SHA512_Transform(SHA512_CTX * c,const uint8_t block[SHA512_CBLOCK])186 void SHA512_Transform(SHA512_CTX *c, const uint8_t block[SHA512_CBLOCK]) {
187   sha512_block_data_order(c->h, block, 1);
188 }
189 
SHA512_Update(SHA512_CTX * c,const void * in_data,size_t len)190 int SHA512_Update(SHA512_CTX *c, const void *in_data, size_t len) {
191   uint64_t l;
192   uint8_t *p = c->p;
193   const uint8_t *data = in_data;
194 
195   if (len == 0) {
196     return 1;
197   }
198 
199   l = (c->Nl + (((uint64_t)len) << 3)) & UINT64_C(0xffffffffffffffff);
200   if (l < c->Nl) {
201     c->Nh++;
202   }
203   if (sizeof(len) >= 8) {
204     c->Nh += (((uint64_t)len) >> 61);
205   }
206   c->Nl = l;
207 
208   if (c->num != 0) {
209     size_t n = sizeof(c->p) - c->num;
210 
211     if (len < n) {
212       OPENSSL_memcpy(p + c->num, data, len);
213       c->num += (unsigned int)len;
214       return 1;
215     } else {
216       OPENSSL_memcpy(p + c->num, data, n), c->num = 0;
217       len -= n;
218       data += n;
219       sha512_block_data_order(c->h, p, 1);
220     }
221   }
222 
223   if (len >= sizeof(c->p)) {
224     sha512_block_data_order(c->h, data, len / sizeof(c->p));
225     data += len;
226     len %= sizeof(c->p);
227     data -= len;
228   }
229 
230   if (len != 0) {
231     OPENSSL_memcpy(p, data, len);
232     c->num = (int)len;
233   }
234 
235   return 1;
236 }
237 
SHA512_Final(uint8_t out[SHA512_DIGEST_LENGTH],SHA512_CTX * sha)238 int SHA512_Final(uint8_t out[SHA512_DIGEST_LENGTH], SHA512_CTX *sha) {
239   // Ideally we would assert |sha->md_len| is |SHA512_DIGEST_LENGTH| to match
240   // the size hint, but calling code often pairs |SHA384_Init| with
241   // |SHA512_Final| and expects |sha->md_len| to carry the size over.
242   //
243   // TODO(davidben): Add an assert and fix code to match them up.
244   return sha512_final_impl(out, sha->md_len, sha);
245 }
246 
sha512_final_impl(uint8_t * out,size_t md_len,SHA512_CTX * sha)247 static int sha512_final_impl(uint8_t *out, size_t md_len, SHA512_CTX *sha) {
248   uint8_t *p = sha->p;
249   size_t n = sha->num;
250 
251   p[n] = 0x80;  // There always is a room for one
252   n++;
253   if (n > (sizeof(sha->p) - 16)) {
254     OPENSSL_memset(p + n, 0, sizeof(sha->p) - n);
255     n = 0;
256     sha512_block_data_order(sha->h, p, 1);
257   }
258 
259   OPENSSL_memset(p + n, 0, sizeof(sha->p) - 16 - n);
260   CRYPTO_store_u64_be(p + sizeof(sha->p) - 16, sha->Nh);
261   CRYPTO_store_u64_be(p + sizeof(sha->p) - 8, sha->Nl);
262 
263   sha512_block_data_order(sha->h, p, 1);
264 
265   if (out == NULL) {
266     // TODO(davidben): This NULL check is absent in other low-level hash 'final'
267     // functions and is one of the few places one can fail.
268     return 0;
269   }
270 
271   assert(md_len % 8 == 0);
272   const size_t out_words = md_len / 8;
273   for (size_t i = 0; i < out_words; i++) {
274     CRYPTO_store_u64_be(out, sha->h[i]);
275     out += 8;
276   }
277 
278   FIPS_service_indicator_update_state();
279   return 1;
280 }
281 
282 #if !defined(SHA512_ASM)
283 
284 #if !defined(SHA512_ASM_NOHW)
285 static const uint64_t K512[80] = {
286     UINT64_C(0x428a2f98d728ae22), UINT64_C(0x7137449123ef65cd),
287     UINT64_C(0xb5c0fbcfec4d3b2f), UINT64_C(0xe9b5dba58189dbbc),
288     UINT64_C(0x3956c25bf348b538), UINT64_C(0x59f111f1b605d019),
289     UINT64_C(0x923f82a4af194f9b), UINT64_C(0xab1c5ed5da6d8118),
290     UINT64_C(0xd807aa98a3030242), UINT64_C(0x12835b0145706fbe),
291     UINT64_C(0x243185be4ee4b28c), UINT64_C(0x550c7dc3d5ffb4e2),
292     UINT64_C(0x72be5d74f27b896f), UINT64_C(0x80deb1fe3b1696b1),
293     UINT64_C(0x9bdc06a725c71235), UINT64_C(0xc19bf174cf692694),
294     UINT64_C(0xe49b69c19ef14ad2), UINT64_C(0xefbe4786384f25e3),
295     UINT64_C(0x0fc19dc68b8cd5b5), UINT64_C(0x240ca1cc77ac9c65),
296     UINT64_C(0x2de92c6f592b0275), UINT64_C(0x4a7484aa6ea6e483),
297     UINT64_C(0x5cb0a9dcbd41fbd4), UINT64_C(0x76f988da831153b5),
298     UINT64_C(0x983e5152ee66dfab), UINT64_C(0xa831c66d2db43210),
299     UINT64_C(0xb00327c898fb213f), UINT64_C(0xbf597fc7beef0ee4),
300     UINT64_C(0xc6e00bf33da88fc2), UINT64_C(0xd5a79147930aa725),
301     UINT64_C(0x06ca6351e003826f), UINT64_C(0x142929670a0e6e70),
302     UINT64_C(0x27b70a8546d22ffc), UINT64_C(0x2e1b21385c26c926),
303     UINT64_C(0x4d2c6dfc5ac42aed), UINT64_C(0x53380d139d95b3df),
304     UINT64_C(0x650a73548baf63de), UINT64_C(0x766a0abb3c77b2a8),
305     UINT64_C(0x81c2c92e47edaee6), UINT64_C(0x92722c851482353b),
306     UINT64_C(0xa2bfe8a14cf10364), UINT64_C(0xa81a664bbc423001),
307     UINT64_C(0xc24b8b70d0f89791), UINT64_C(0xc76c51a30654be30),
308     UINT64_C(0xd192e819d6ef5218), UINT64_C(0xd69906245565a910),
309     UINT64_C(0xf40e35855771202a), UINT64_C(0x106aa07032bbd1b8),
310     UINT64_C(0x19a4c116b8d2d0c8), UINT64_C(0x1e376c085141ab53),
311     UINT64_C(0x2748774cdf8eeb99), UINT64_C(0x34b0bcb5e19b48a8),
312     UINT64_C(0x391c0cb3c5c95a63), UINT64_C(0x4ed8aa4ae3418acb),
313     UINT64_C(0x5b9cca4f7763e373), UINT64_C(0x682e6ff3d6b2b8a3),
314     UINT64_C(0x748f82ee5defb2fc), UINT64_C(0x78a5636f43172f60),
315     UINT64_C(0x84c87814a1f0ab72), UINT64_C(0x8cc702081a6439ec),
316     UINT64_C(0x90befffa23631e28), UINT64_C(0xa4506cebde82bde9),
317     UINT64_C(0xbef9a3f7b2c67915), UINT64_C(0xc67178f2e372532b),
318     UINT64_C(0xca273eceea26619c), UINT64_C(0xd186b8c721c0c207),
319     UINT64_C(0xeada7dd6cde0eb1e), UINT64_C(0xf57d4f7fee6ed178),
320     UINT64_C(0x06f067aa72176fba), UINT64_C(0x0a637dc5a2c898a6),
321     UINT64_C(0x113f9804bef90dae), UINT64_C(0x1b710b35131c471b),
322     UINT64_C(0x28db77f523047d84), UINT64_C(0x32caab7b40c72493),
323     UINT64_C(0x3c9ebe0a15c9bebc), UINT64_C(0x431d67c49c100d4c),
324     UINT64_C(0x4cc5d4becb3e42b6), UINT64_C(0x597f299cfc657e2a),
325     UINT64_C(0x5fcb6fab3ad6faec), UINT64_C(0x6c44198c4a475817),
326 };
327 
328 #define Sigma0(x)                                        \
329   (CRYPTO_rotr_u64((x), 28) ^ CRYPTO_rotr_u64((x), 34) ^ \
330    CRYPTO_rotr_u64((x), 39))
331 #define Sigma1(x)                                        \
332   (CRYPTO_rotr_u64((x), 14) ^ CRYPTO_rotr_u64((x), 18) ^ \
333    CRYPTO_rotr_u64((x), 41))
334 #define sigma0(x) \
335   (CRYPTO_rotr_u64((x), 1) ^ CRYPTO_rotr_u64((x), 8) ^ ((x) >> 7))
336 #define sigma1(x) \
337   (CRYPTO_rotr_u64((x), 19) ^ CRYPTO_rotr_u64((x), 61) ^ ((x) >> 6))
338 
339 #define Ch(x, y, z) (((x) & (y)) ^ ((~(x)) & (z)))
340 #define Maj(x, y, z) (((x) & (y)) ^ ((x) & (z)) ^ ((y) & (z)))
341 
342 
343 #if defined(__i386) || defined(__i386__) || defined(_M_IX86)
344 // This code should give better results on 32-bit CPU with less than
345 // ~24 registers, both size and performance wise...
sha512_block_data_order_nohw(uint64_t state[8],const uint8_t * in,size_t num)346 static void sha512_block_data_order_nohw(uint64_t state[8], const uint8_t *in,
347                                          size_t num) {
348   uint64_t A, E, T;
349   uint64_t X[9 + 80], *F;
350   int i;
351 
352   while (num--) {
353     F = X + 80;
354     A = state[0];
355     F[1] = state[1];
356     F[2] = state[2];
357     F[3] = state[3];
358     E = state[4];
359     F[5] = state[5];
360     F[6] = state[6];
361     F[7] = state[7];
362 
363     for (i = 0; i < 16; i++, F--) {
364       T = CRYPTO_load_u64_be(in + i * 8);
365       F[0] = A;
366       F[4] = E;
367       F[8] = T;
368       T += F[7] + Sigma1(E) + Ch(E, F[5], F[6]) + K512[i];
369       E = F[3] + T;
370       A = T + Sigma0(A) + Maj(A, F[1], F[2]);
371     }
372 
373     for (; i < 80; i++, F--) {
374       T = sigma0(F[8 + 16 - 1]);
375       T += sigma1(F[8 + 16 - 14]);
376       T += F[8 + 16] + F[8 + 16 - 9];
377 
378       F[0] = A;
379       F[4] = E;
380       F[8] = T;
381       T += F[7] + Sigma1(E) + Ch(E, F[5], F[6]) + K512[i];
382       E = F[3] + T;
383       A = T + Sigma0(A) + Maj(A, F[1], F[2]);
384     }
385 
386     state[0] += A;
387     state[1] += F[1];
388     state[2] += F[2];
389     state[3] += F[3];
390     state[4] += E;
391     state[5] += F[5];
392     state[6] += F[6];
393     state[7] += F[7];
394 
395     in += 16 * 8;
396   }
397 }
398 
399 #else
400 
401 #define ROUND_00_15(i, a, b, c, d, e, f, g, h)   \
402   do {                                           \
403     T1 += h + Sigma1(e) + Ch(e, f, g) + K512[i]; \
404     h = Sigma0(a) + Maj(a, b, c);                \
405     d += T1;                                     \
406     h += T1;                                     \
407   } while (0)
408 
409 #define ROUND_16_80(i, j, a, b, c, d, e, f, g, h, X)   \
410   do {                                                 \
411     s0 = X[(j + 1) & 0x0f];                            \
412     s0 = sigma0(s0);                                   \
413     s1 = X[(j + 14) & 0x0f];                           \
414     s1 = sigma1(s1);                                   \
415     T1 = X[(j) & 0x0f] += s0 + s1 + X[(j + 9) & 0x0f]; \
416     ROUND_00_15(i + j, a, b, c, d, e, f, g, h);        \
417   } while (0)
418 
sha512_block_data_order_nohw(uint64_t state[8],const uint8_t * in,size_t num)419 static void sha512_block_data_order_nohw(uint64_t state[8], const uint8_t *in,
420                                          size_t num) {
421   uint64_t a, b, c, d, e, f, g, h, s0, s1, T1;
422   uint64_t X[16];
423   int i;
424 
425   while (num--) {
426 
427     a = state[0];
428     b = state[1];
429     c = state[2];
430     d = state[3];
431     e = state[4];
432     f = state[5];
433     g = state[6];
434     h = state[7];
435 
436     T1 = X[0] = CRYPTO_load_u64_be(in);
437     ROUND_00_15(0, a, b, c, d, e, f, g, h);
438     T1 = X[1] = CRYPTO_load_u64_be(in + 8);
439     ROUND_00_15(1, h, a, b, c, d, e, f, g);
440     T1 = X[2] = CRYPTO_load_u64_be(in + 2 * 8);
441     ROUND_00_15(2, g, h, a, b, c, d, e, f);
442     T1 = X[3] = CRYPTO_load_u64_be(in + 3 * 8);
443     ROUND_00_15(3, f, g, h, a, b, c, d, e);
444     T1 = X[4] = CRYPTO_load_u64_be(in + 4 * 8);
445     ROUND_00_15(4, e, f, g, h, a, b, c, d);
446     T1 = X[5] = CRYPTO_load_u64_be(in + 5 * 8);
447     ROUND_00_15(5, d, e, f, g, h, a, b, c);
448     T1 = X[6] = CRYPTO_load_u64_be(in + 6 * 8);
449     ROUND_00_15(6, c, d, e, f, g, h, a, b);
450     T1 = X[7] = CRYPTO_load_u64_be(in + 7 * 8);
451     ROUND_00_15(7, b, c, d, e, f, g, h, a);
452     T1 = X[8] = CRYPTO_load_u64_be(in + 8 * 8);
453     ROUND_00_15(8, a, b, c, d, e, f, g, h);
454     T1 = X[9] = CRYPTO_load_u64_be(in + 9 * 8);
455     ROUND_00_15(9, h, a, b, c, d, e, f, g);
456     T1 = X[10] = CRYPTO_load_u64_be(in + 10 * 8);
457     ROUND_00_15(10, g, h, a, b, c, d, e, f);
458     T1 = X[11] = CRYPTO_load_u64_be(in + 11 * 8);
459     ROUND_00_15(11, f, g, h, a, b, c, d, e);
460     T1 = X[12] = CRYPTO_load_u64_be(in + 12 * 8);
461     ROUND_00_15(12, e, f, g, h, a, b, c, d);
462     T1 = X[13] = CRYPTO_load_u64_be(in + 13 * 8);
463     ROUND_00_15(13, d, e, f, g, h, a, b, c);
464     T1 = X[14] = CRYPTO_load_u64_be(in + 14 * 8);
465     ROUND_00_15(14, c, d, e, f, g, h, a, b);
466     T1 = X[15] = CRYPTO_load_u64_be(in + 15 * 8);
467     ROUND_00_15(15, b, c, d, e, f, g, h, a);
468 
469     for (i = 16; i < 80; i += 16) {
470       ROUND_16_80(i, 0, a, b, c, d, e, f, g, h, X);
471       ROUND_16_80(i, 1, h, a, b, c, d, e, f, g, X);
472       ROUND_16_80(i, 2, g, h, a, b, c, d, e, f, X);
473       ROUND_16_80(i, 3, f, g, h, a, b, c, d, e, X);
474       ROUND_16_80(i, 4, e, f, g, h, a, b, c, d, X);
475       ROUND_16_80(i, 5, d, e, f, g, h, a, b, c, X);
476       ROUND_16_80(i, 6, c, d, e, f, g, h, a, b, X);
477       ROUND_16_80(i, 7, b, c, d, e, f, g, h, a, X);
478       ROUND_16_80(i, 8, a, b, c, d, e, f, g, h, X);
479       ROUND_16_80(i, 9, h, a, b, c, d, e, f, g, X);
480       ROUND_16_80(i, 10, g, h, a, b, c, d, e, f, X);
481       ROUND_16_80(i, 11, f, g, h, a, b, c, d, e, X);
482       ROUND_16_80(i, 12, e, f, g, h, a, b, c, d, X);
483       ROUND_16_80(i, 13, d, e, f, g, h, a, b, c, X);
484       ROUND_16_80(i, 14, c, d, e, f, g, h, a, b, X);
485       ROUND_16_80(i, 15, b, c, d, e, f, g, h, a, X);
486     }
487 
488     state[0] += a;
489     state[1] += b;
490     state[2] += c;
491     state[3] += d;
492     state[4] += e;
493     state[5] += f;
494     state[6] += g;
495     state[7] += h;
496 
497     in += 16 * 8;
498   }
499 }
500 
501 #endif
502 
503 #endif  // !SHA512_ASM_NOHW
504 
sha512_block_data_order(uint64_t state[8],const uint8_t * data,size_t num)505 static void sha512_block_data_order(uint64_t state[8], const uint8_t *data,
506                                     size_t num) {
507 #if defined(SHA512_ASM_HW)
508   if (sha512_hw_capable()) {
509     sha512_block_data_order_hw(state, data, num);
510     return;
511   }
512 #endif
513 #if defined(SHA512_ASM_AVX)
514   if (sha512_avx_capable()) {
515     sha512_block_data_order_avx(state, data, num);
516     return;
517   }
518 #endif
519 #if defined(SHA512_ASM_SSSE3)
520   if (sha512_ssse3_capable()) {
521     sha512_block_data_order_ssse3(state, data, num);
522     return;
523   }
524 #endif
525 #if defined(SHA512_ASM_NEON)
526   if (CRYPTO_is_NEON_capable()) {
527     sha512_block_data_order_neon(state, data, num);
528     return;
529   }
530 #endif
531   sha512_block_data_order_nohw(state, data, num);
532 }
533 
534 #endif  // !SHA512_ASM
535 
536 #undef Sigma0
537 #undef Sigma1
538 #undef sigma0
539 #undef sigma1
540 #undef Ch
541 #undef Maj
542 #undef ROUND_00_15
543 #undef ROUND_16_80
544