• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
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 #include <openssl/sha.h>
58 
59 #include <string.h>
60 
61 #include <openssl/mem.h>
62 
63 #include "internal.h"
64 #include "../../internal.h"
65 
66 
67 // The 32-bit hash algorithms share a common byte-order neutral collector and
68 // padding function implementations that operate on unaligned data,
69 // ../digest/md32_common.h. SHA-512 is the only 64-bit hash algorithm, as of
70 // this writing, so there is no need for a common collector/padding
71 // implementation yet.
72 
SHA384_Init(SHA512_CTX * sha)73 int SHA384_Init(SHA512_CTX *sha) {
74   sha->h[0] = UINT64_C(0xcbbb9d5dc1059ed8);
75   sha->h[1] = UINT64_C(0x629a292a367cd507);
76   sha->h[2] = UINT64_C(0x9159015a3070dd17);
77   sha->h[3] = UINT64_C(0x152fecd8f70e5939);
78   sha->h[4] = UINT64_C(0x67332667ffc00b31);
79   sha->h[5] = UINT64_C(0x8eb44a8768581511);
80   sha->h[6] = UINT64_C(0xdb0c2e0d64f98fa7);
81   sha->h[7] = UINT64_C(0x47b5481dbefa4fa4);
82 
83   sha->Nl = 0;
84   sha->Nh = 0;
85   sha->num = 0;
86   sha->md_len = SHA384_DIGEST_LENGTH;
87   return 1;
88 }
89 
90 
SHA512_Init(SHA512_CTX * sha)91 int SHA512_Init(SHA512_CTX *sha) {
92   sha->h[0] = UINT64_C(0x6a09e667f3bcc908);
93   sha->h[1] = UINT64_C(0xbb67ae8584caa73b);
94   sha->h[2] = UINT64_C(0x3c6ef372fe94f82b);
95   sha->h[3] = UINT64_C(0xa54ff53a5f1d36f1);
96   sha->h[4] = UINT64_C(0x510e527fade682d1);
97   sha->h[5] = UINT64_C(0x9b05688c2b3e6c1f);
98   sha->h[6] = UINT64_C(0x1f83d9abfb41bd6b);
99   sha->h[7] = UINT64_C(0x5be0cd19137e2179);
100 
101   sha->Nl = 0;
102   sha->Nh = 0;
103   sha->num = 0;
104   sha->md_len = SHA512_DIGEST_LENGTH;
105   return 1;
106 }
107 
SHA512_256_Init(SHA512_CTX * sha)108 int SHA512_256_Init(SHA512_CTX *sha) {
109   sha->h[0] = UINT64_C(0x22312194fc2bf72c);
110   sha->h[1] = UINT64_C(0x9f555fa3c84c64c2);
111   sha->h[2] = UINT64_C(0x2393b86b6f53b151);
112   sha->h[3] = UINT64_C(0x963877195940eabd);
113   sha->h[4] = UINT64_C(0x96283ee2a88effe3);
114   sha->h[5] = UINT64_C(0xbe5e1e2553863992);
115   sha->h[6] = UINT64_C(0x2b0199fc2c85b8aa);
116   sha->h[7] = UINT64_C(0x0eb72ddc81c52ca2);
117 
118   sha->Nl = 0;
119   sha->Nh = 0;
120   sha->num = 0;
121   sha->md_len = SHA512_256_DIGEST_LENGTH;
122   return 1;
123 }
124 
SHA384(const uint8_t * data,size_t len,uint8_t out[SHA384_DIGEST_LENGTH])125 uint8_t *SHA384(const uint8_t *data, size_t len,
126                 uint8_t out[SHA384_DIGEST_LENGTH]) {
127   SHA512_CTX ctx;
128   SHA384_Init(&ctx);
129   SHA384_Update(&ctx, data, len);
130   SHA384_Final(out, &ctx);
131   OPENSSL_cleanse(&ctx, sizeof(ctx));
132   return out;
133 }
134 
SHA512(const uint8_t * data,size_t len,uint8_t out[SHA512_DIGEST_LENGTH])135 uint8_t *SHA512(const uint8_t *data, size_t len,
136                 uint8_t out[SHA512_DIGEST_LENGTH]) {
137   SHA512_CTX ctx;
138   SHA512_Init(&ctx);
139   SHA512_Update(&ctx, data, len);
140   SHA512_Final(out, &ctx);
141   OPENSSL_cleanse(&ctx, sizeof(ctx));
142   return out;
143 }
144 
SHA512_256(const uint8_t * data,size_t len,uint8_t out[SHA512_256_DIGEST_LENGTH])145 uint8_t *SHA512_256(const uint8_t *data, size_t len,
146                     uint8_t out[SHA512_256_DIGEST_LENGTH]) {
147   SHA512_CTX ctx;
148   SHA512_256_Init(&ctx);
149   SHA512_Update(&ctx, data, len);
150   SHA512_Final(out, &ctx);
151   OPENSSL_cleanse(&ctx, sizeof(ctx));
152   return out;
153 }
154 
155 #if !defined(SHA512_ASM)
156 static void sha512_block_data_order(uint64_t *state, const uint8_t *in,
157                                     size_t num_blocks);
158 #endif
159 
160 
SHA384_Final(uint8_t out[SHA384_DIGEST_LENGTH],SHA512_CTX * sha)161 int SHA384_Final(uint8_t out[SHA384_DIGEST_LENGTH], SHA512_CTX *sha) {
162   // |SHA384_Init| sets |sha->md_len| to |SHA384_DIGEST_LENGTH|, so this has a
163   // |smaller output.
164   return SHA512_Final(out, sha);
165 }
166 
SHA384_Update(SHA512_CTX * sha,const void * data,size_t len)167 int SHA384_Update(SHA512_CTX *sha, const void *data, size_t len) {
168   return SHA512_Update(sha, data, len);
169 }
170 
SHA512_256_Update(SHA512_CTX * sha,const void * data,size_t len)171 int SHA512_256_Update(SHA512_CTX *sha, const void *data, size_t len) {
172   return SHA512_Update(sha, data, len);
173 }
174 
SHA512_256_Final(uint8_t out[SHA512_256_DIGEST_LENGTH],SHA512_CTX * sha)175 int SHA512_256_Final(uint8_t out[SHA512_256_DIGEST_LENGTH],
176                                     SHA512_CTX *sha) {
177   // |SHA512_256_Init| sets |sha->md_len| to |SHA512_256_DIGEST_LENGTH|, so this
178   // has a |smaller output.
179   return SHA512_Final(out, sha);
180 }
181 
SHA512_Transform(SHA512_CTX * c,const uint8_t block[SHA512_CBLOCK])182 void SHA512_Transform(SHA512_CTX *c, const uint8_t block[SHA512_CBLOCK]) {
183   sha512_block_data_order(c->h, block, 1);
184 }
185 
SHA512_Update(SHA512_CTX * c,const void * in_data,size_t len)186 int SHA512_Update(SHA512_CTX *c, const void *in_data, size_t len) {
187   uint64_t l;
188   uint8_t *p = c->p;
189   const uint8_t *data = in_data;
190 
191   if (len == 0) {
192     return 1;
193   }
194 
195   l = (c->Nl + (((uint64_t)len) << 3)) & UINT64_C(0xffffffffffffffff);
196   if (l < c->Nl) {
197     c->Nh++;
198   }
199   if (sizeof(len) >= 8) {
200     c->Nh += (((uint64_t)len) >> 61);
201   }
202   c->Nl = l;
203 
204   if (c->num != 0) {
205     size_t n = sizeof(c->p) - c->num;
206 
207     if (len < n) {
208       OPENSSL_memcpy(p + c->num, data, len);
209       c->num += (unsigned int)len;
210       return 1;
211     } else {
212       OPENSSL_memcpy(p + c->num, data, n), c->num = 0;
213       len -= n;
214       data += n;
215       sha512_block_data_order(c->h, p, 1);
216     }
217   }
218 
219   if (len >= sizeof(c->p)) {
220     sha512_block_data_order(c->h, data, len / sizeof(c->p));
221     data += len;
222     len %= sizeof(c->p);
223     data -= len;
224   }
225 
226   if (len != 0) {
227     OPENSSL_memcpy(p, data, len);
228     c->num = (int)len;
229   }
230 
231   return 1;
232 }
233 
SHA512_Final(uint8_t out[SHA512_DIGEST_LENGTH],SHA512_CTX * sha)234 int SHA512_Final(uint8_t out[SHA512_DIGEST_LENGTH], SHA512_CTX *sha) {
235   uint8_t *p = sha->p;
236   size_t n = sha->num;
237 
238   p[n] = 0x80;  // There always is a room for one
239   n++;
240   if (n > (sizeof(sha->p) - 16)) {
241     OPENSSL_memset(p + n, 0, sizeof(sha->p) - n);
242     n = 0;
243     sha512_block_data_order(sha->h, p, 1);
244   }
245 
246   OPENSSL_memset(p + n, 0, sizeof(sha->p) - 16 - n);
247   p[sizeof(sha->p) - 1] = (uint8_t)(sha->Nl);
248   p[sizeof(sha->p) - 2] = (uint8_t)(sha->Nl >> 8);
249   p[sizeof(sha->p) - 3] = (uint8_t)(sha->Nl >> 16);
250   p[sizeof(sha->p) - 4] = (uint8_t)(sha->Nl >> 24);
251   p[sizeof(sha->p) - 5] = (uint8_t)(sha->Nl >> 32);
252   p[sizeof(sha->p) - 6] = (uint8_t)(sha->Nl >> 40);
253   p[sizeof(sha->p) - 7] = (uint8_t)(sha->Nl >> 48);
254   p[sizeof(sha->p) - 8] = (uint8_t)(sha->Nl >> 56);
255   p[sizeof(sha->p) - 9] = (uint8_t)(sha->Nh);
256   p[sizeof(sha->p) - 10] = (uint8_t)(sha->Nh >> 8);
257   p[sizeof(sha->p) - 11] = (uint8_t)(sha->Nh >> 16);
258   p[sizeof(sha->p) - 12] = (uint8_t)(sha->Nh >> 24);
259   p[sizeof(sha->p) - 13] = (uint8_t)(sha->Nh >> 32);
260   p[sizeof(sha->p) - 14] = (uint8_t)(sha->Nh >> 40);
261   p[sizeof(sha->p) - 15] = (uint8_t)(sha->Nh >> 48);
262   p[sizeof(sha->p) - 16] = (uint8_t)(sha->Nh >> 56);
263 
264   sha512_block_data_order(sha->h, p, 1);
265 
266   if (out == NULL) {
267     // TODO(davidben): This NULL check is absent in other low-level hash 'final'
268     // functions and is one of the few places one can fail.
269     return 0;
270   }
271 
272   assert(sha->md_len % 8 == 0);
273   const size_t out_words = sha->md_len / 8;
274   for (size_t i = 0; i < out_words; i++) {
275     const uint64_t t = CRYPTO_bswap8(sha->h[i]);
276     memcpy(out, &t, sizeof(t));
277     out += sizeof(t);
278   }
279 
280   return 1;
281 }
282 
283 #ifndef SHA512_ASM
284 static const uint64_t K512[80] = {
285     UINT64_C(0x428a2f98d728ae22), UINT64_C(0x7137449123ef65cd),
286     UINT64_C(0xb5c0fbcfec4d3b2f), UINT64_C(0xe9b5dba58189dbbc),
287     UINT64_C(0x3956c25bf348b538), UINT64_C(0x59f111f1b605d019),
288     UINT64_C(0x923f82a4af194f9b), UINT64_C(0xab1c5ed5da6d8118),
289     UINT64_C(0xd807aa98a3030242), UINT64_C(0x12835b0145706fbe),
290     UINT64_C(0x243185be4ee4b28c), UINT64_C(0x550c7dc3d5ffb4e2),
291     UINT64_C(0x72be5d74f27b896f), UINT64_C(0x80deb1fe3b1696b1),
292     UINT64_C(0x9bdc06a725c71235), UINT64_C(0xc19bf174cf692694),
293     UINT64_C(0xe49b69c19ef14ad2), UINT64_C(0xefbe4786384f25e3),
294     UINT64_C(0x0fc19dc68b8cd5b5), UINT64_C(0x240ca1cc77ac9c65),
295     UINT64_C(0x2de92c6f592b0275), UINT64_C(0x4a7484aa6ea6e483),
296     UINT64_C(0x5cb0a9dcbd41fbd4), UINT64_C(0x76f988da831153b5),
297     UINT64_C(0x983e5152ee66dfab), UINT64_C(0xa831c66d2db43210),
298     UINT64_C(0xb00327c898fb213f), UINT64_C(0xbf597fc7beef0ee4),
299     UINT64_C(0xc6e00bf33da88fc2), UINT64_C(0xd5a79147930aa725),
300     UINT64_C(0x06ca6351e003826f), UINT64_C(0x142929670a0e6e70),
301     UINT64_C(0x27b70a8546d22ffc), UINT64_C(0x2e1b21385c26c926),
302     UINT64_C(0x4d2c6dfc5ac42aed), UINT64_C(0x53380d139d95b3df),
303     UINT64_C(0x650a73548baf63de), UINT64_C(0x766a0abb3c77b2a8),
304     UINT64_C(0x81c2c92e47edaee6), UINT64_C(0x92722c851482353b),
305     UINT64_C(0xa2bfe8a14cf10364), UINT64_C(0xa81a664bbc423001),
306     UINT64_C(0xc24b8b70d0f89791), UINT64_C(0xc76c51a30654be30),
307     UINT64_C(0xd192e819d6ef5218), UINT64_C(0xd69906245565a910),
308     UINT64_C(0xf40e35855771202a), UINT64_C(0x106aa07032bbd1b8),
309     UINT64_C(0x19a4c116b8d2d0c8), UINT64_C(0x1e376c085141ab53),
310     UINT64_C(0x2748774cdf8eeb99), UINT64_C(0x34b0bcb5e19b48a8),
311     UINT64_C(0x391c0cb3c5c95a63), UINT64_C(0x4ed8aa4ae3418acb),
312     UINT64_C(0x5b9cca4f7763e373), UINT64_C(0x682e6ff3d6b2b8a3),
313     UINT64_C(0x748f82ee5defb2fc), UINT64_C(0x78a5636f43172f60),
314     UINT64_C(0x84c87814a1f0ab72), UINT64_C(0x8cc702081a6439ec),
315     UINT64_C(0x90befffa23631e28), UINT64_C(0xa4506cebde82bde9),
316     UINT64_C(0xbef9a3f7b2c67915), UINT64_C(0xc67178f2e372532b),
317     UINT64_C(0xca273eceea26619c), UINT64_C(0xd186b8c721c0c207),
318     UINT64_C(0xeada7dd6cde0eb1e), UINT64_C(0xf57d4f7fee6ed178),
319     UINT64_C(0x06f067aa72176fba), UINT64_C(0x0a637dc5a2c898a6),
320     UINT64_C(0x113f9804bef90dae), UINT64_C(0x1b710b35131c471b),
321     UINT64_C(0x28db77f523047d84), UINT64_C(0x32caab7b40c72493),
322     UINT64_C(0x3c9ebe0a15c9bebc), UINT64_C(0x431d67c49c100d4c),
323     UINT64_C(0x4cc5d4becb3e42b6), UINT64_C(0x597f299cfc657e2a),
324     UINT64_C(0x5fcb6fab3ad6faec), UINT64_C(0x6c44198c4a475817),
325 };
326 
327 #if defined(__GNUC__) && __GNUC__ >= 2 && !defined(OPENSSL_NO_ASM)
328 #if defined(__x86_64) || defined(__x86_64__)
329 #define ROTR(a, n)                                              \
330   ({                                                            \
331     uint64_t ret;                                               \
332     __asm__("rorq %1, %0" : "=r"(ret) : "J"(n), "0"(a) : "cc"); \
333     ret;                                                        \
334   })
335 #elif(defined(_ARCH_PPC) && defined(__64BIT__)) || defined(_ARCH_PPC64)
336 #define ROTR(a, n)                                             \
337   ({                                                           \
338     uint64_t ret;                                              \
339     __asm__("rotrdi %0, %1, %2" : "=r"(ret) : "r"(a), "K"(n)); \
340     ret;                                                       \
341   })
342 #elif defined(__aarch64__)
343 #define ROTR(a, n)                                          \
344   ({                                                        \
345     uint64_t ret;                                           \
346     __asm__("ror %0, %1, %2" : "=r"(ret) : "r"(a), "I"(n)); \
347     ret;                                                    \
348   })
349 #endif
350 #elif defined(_MSC_VER) && defined(_WIN64)
351 #pragma intrinsic(_rotr64)
352 #define ROTR(a, n) _rotr64((a), n)
353 #endif
354 
355 #ifndef ROTR
356 #define ROTR(x, s) (((x) >> s) | (x) << (64 - s))
357 #endif
358 
load_u64_be(const void * ptr)359 static inline uint64_t load_u64_be(const void *ptr) {
360   uint64_t ret;
361   OPENSSL_memcpy(&ret, ptr, sizeof(ret));
362   return CRYPTO_bswap8(ret);
363 }
364 
365 #define Sigma0(x) (ROTR((x), 28) ^ ROTR((x), 34) ^ ROTR((x), 39))
366 #define Sigma1(x) (ROTR((x), 14) ^ ROTR((x), 18) ^ ROTR((x), 41))
367 #define sigma0(x) (ROTR((x), 1) ^ ROTR((x), 8) ^ ((x) >> 7))
368 #define sigma1(x) (ROTR((x), 19) ^ ROTR((x), 61) ^ ((x) >> 6))
369 
370 #define Ch(x, y, z) (((x) & (y)) ^ ((~(x)) & (z)))
371 #define Maj(x, y, z) (((x) & (y)) ^ ((x) & (z)) ^ ((y) & (z)))
372 
373 
374 #if defined(__i386) || defined(__i386__) || defined(_M_IX86)
375 // This code should give better results on 32-bit CPU with less than
376 // ~24 registers, both size and performance wise...
sha512_block_data_order(uint64_t * state,const uint8_t * in,size_t num)377 static void sha512_block_data_order(uint64_t *state, const uint8_t *in,
378                                     size_t num) {
379   uint64_t A, E, T;
380   uint64_t X[9 + 80], *F;
381   int i;
382 
383   while (num--) {
384     F = X + 80;
385     A = state[0];
386     F[1] = state[1];
387     F[2] = state[2];
388     F[3] = state[3];
389     E = state[4];
390     F[5] = state[5];
391     F[6] = state[6];
392     F[7] = state[7];
393 
394     for (i = 0; i < 16; i++, F--) {
395       T = load_u64_be(in + i * 8);
396       F[0] = A;
397       F[4] = E;
398       F[8] = T;
399       T += F[7] + Sigma1(E) + Ch(E, F[5], F[6]) + K512[i];
400       E = F[3] + T;
401       A = T + Sigma0(A) + Maj(A, F[1], F[2]);
402     }
403 
404     for (; i < 80; i++, F--) {
405       T = sigma0(F[8 + 16 - 1]);
406       T += sigma1(F[8 + 16 - 14]);
407       T += F[8 + 16] + F[8 + 16 - 9];
408 
409       F[0] = A;
410       F[4] = E;
411       F[8] = T;
412       T += F[7] + Sigma1(E) + Ch(E, F[5], F[6]) + K512[i];
413       E = F[3] + T;
414       A = T + Sigma0(A) + Maj(A, F[1], F[2]);
415     }
416 
417     state[0] += A;
418     state[1] += F[1];
419     state[2] += F[2];
420     state[3] += F[3];
421     state[4] += E;
422     state[5] += F[5];
423     state[6] += F[6];
424     state[7] += F[7];
425 
426     in += 16 * 8;
427   }
428 }
429 
430 #else
431 
432 #define ROUND_00_15(i, a, b, c, d, e, f, g, h)   \
433   do {                                           \
434     T1 += h + Sigma1(e) + Ch(e, f, g) + K512[i]; \
435     h = Sigma0(a) + Maj(a, b, c);                \
436     d += T1;                                     \
437     h += T1;                                     \
438   } while (0)
439 
440 #define ROUND_16_80(i, j, a, b, c, d, e, f, g, h, X)   \
441   do {                                                 \
442     s0 = X[(j + 1) & 0x0f];                            \
443     s0 = sigma0(s0);                                   \
444     s1 = X[(j + 14) & 0x0f];                           \
445     s1 = sigma1(s1);                                   \
446     T1 = X[(j) & 0x0f] += s0 + s1 + X[(j + 9) & 0x0f]; \
447     ROUND_00_15(i + j, a, b, c, d, e, f, g, h);        \
448   } while (0)
449 
sha512_block_data_order(uint64_t * state,const uint8_t * in,size_t num)450 static void sha512_block_data_order(uint64_t *state, const uint8_t *in,
451                                     size_t num) {
452   uint64_t a, b, c, d, e, f, g, h, s0, s1, T1;
453   uint64_t X[16];
454   int i;
455 
456   while (num--) {
457 
458     a = state[0];
459     b = state[1];
460     c = state[2];
461     d = state[3];
462     e = state[4];
463     f = state[5];
464     g = state[6];
465     h = state[7];
466 
467     T1 = X[0] = load_u64_be(in);
468     ROUND_00_15(0, a, b, c, d, e, f, g, h);
469     T1 = X[1] = load_u64_be(in + 8);
470     ROUND_00_15(1, h, a, b, c, d, e, f, g);
471     T1 = X[2] = load_u64_be(in + 2 * 8);
472     ROUND_00_15(2, g, h, a, b, c, d, e, f);
473     T1 = X[3] = load_u64_be(in + 3 * 8);
474     ROUND_00_15(3, f, g, h, a, b, c, d, e);
475     T1 = X[4] = load_u64_be(in + 4 * 8);
476     ROUND_00_15(4, e, f, g, h, a, b, c, d);
477     T1 = X[5] = load_u64_be(in + 5 * 8);
478     ROUND_00_15(5, d, e, f, g, h, a, b, c);
479     T1 = X[6] = load_u64_be(in + 6 * 8);
480     ROUND_00_15(6, c, d, e, f, g, h, a, b);
481     T1 = X[7] = load_u64_be(in + 7 * 8);
482     ROUND_00_15(7, b, c, d, e, f, g, h, a);
483     T1 = X[8] = load_u64_be(in + 8 * 8);
484     ROUND_00_15(8, a, b, c, d, e, f, g, h);
485     T1 = X[9] = load_u64_be(in + 9 * 8);
486     ROUND_00_15(9, h, a, b, c, d, e, f, g);
487     T1 = X[10] = load_u64_be(in + 10 * 8);
488     ROUND_00_15(10, g, h, a, b, c, d, e, f);
489     T1 = X[11] = load_u64_be(in + 11 * 8);
490     ROUND_00_15(11, f, g, h, a, b, c, d, e);
491     T1 = X[12] = load_u64_be(in + 12 * 8);
492     ROUND_00_15(12, e, f, g, h, a, b, c, d);
493     T1 = X[13] = load_u64_be(in + 13 * 8);
494     ROUND_00_15(13, d, e, f, g, h, a, b, c);
495     T1 = X[14] = load_u64_be(in + 14 * 8);
496     ROUND_00_15(14, c, d, e, f, g, h, a, b);
497     T1 = X[15] = load_u64_be(in + 15 * 8);
498     ROUND_00_15(15, b, c, d, e, f, g, h, a);
499 
500     for (i = 16; i < 80; i += 16) {
501       ROUND_16_80(i, 0, a, b, c, d, e, f, g, h, X);
502       ROUND_16_80(i, 1, h, a, b, c, d, e, f, g, X);
503       ROUND_16_80(i, 2, g, h, a, b, c, d, e, f, X);
504       ROUND_16_80(i, 3, f, g, h, a, b, c, d, e, X);
505       ROUND_16_80(i, 4, e, f, g, h, a, b, c, d, X);
506       ROUND_16_80(i, 5, d, e, f, g, h, a, b, c, X);
507       ROUND_16_80(i, 6, c, d, e, f, g, h, a, b, X);
508       ROUND_16_80(i, 7, b, c, d, e, f, g, h, a, X);
509       ROUND_16_80(i, 8, a, b, c, d, e, f, g, h, X);
510       ROUND_16_80(i, 9, h, a, b, c, d, e, f, g, X);
511       ROUND_16_80(i, 10, g, h, a, b, c, d, e, f, X);
512       ROUND_16_80(i, 11, f, g, h, a, b, c, d, e, X);
513       ROUND_16_80(i, 12, e, f, g, h, a, b, c, d, X);
514       ROUND_16_80(i, 13, d, e, f, g, h, a, b, c, X);
515       ROUND_16_80(i, 14, c, d, e, f, g, h, a, b, X);
516       ROUND_16_80(i, 15, b, c, d, e, f, g, h, a, X);
517     }
518 
519     state[0] += a;
520     state[1] += b;
521     state[2] += c;
522     state[3] += d;
523     state[4] += e;
524     state[5] += f;
525     state[6] += g;
526     state[7] += h;
527 
528     in += 16 * 8;
529   }
530 }
531 
532 #endif
533 
534 #endif  // !SHA512_ASM
535 
536 #undef ROTR
537 #undef Sigma0
538 #undef Sigma1
539 #undef sigma0
540 #undef sigma1
541 #undef Ch
542 #undef Maj
543 #undef ROUND_00_15
544 #undef ROUND_16_80
545