2 * Copyright 2004-2018 The OpenSSL Project Authors. All Rights Reserved.
4 * Licensed under the OpenSSL license (the "License"). You may not use
5 * this file except in compliance with the License. You can obtain a copy
6 * in the file LICENSE in the source distribution or at
7 * https://www.openssl.org/source/license.html
10 #include <openssl/opensslconf.h>
12 * IMPLEMENTATION NOTES.
14 * As you might have noticed 32-bit hash algorithms:
16 * - permit SHA_LONG to be wider than 32-bit
17 * - optimized versions implement two transform functions: one operating
18 * on [aligned] data in host byte order and one - on data in input
20 * - share common byte-order neutral collector and padding function
21 * implementations, ../md32_common.h;
23 * Neither of the above applies to this SHA-512 implementations. Reasons
24 * [in reverse order] are:
26 * - it's the only 64-bit hash algorithm for the moment of this writing,
27 * there is no need for common collector/padding implementation [yet];
28 * - by supporting only one transform function [which operates on
29 * *aligned* data in input stream byte order, big-endian in this case]
30 * we minimize burden of maintenance in two ways: a) collector/padding
31 * function is simpler; b) only one transform function to stare at;
32 * - SHA_LONG64 is required to be exactly 64-bit in order to be able to
33 * apply a number of optimizations to mitigate potential performance
34 * penalties caused by previous design decision;
38 * Implementation relies on the fact that "long long" is 64-bit on
39 * both 32- and 64-bit platforms. If some compiler vendor comes up
40 * with 128-bit long long, adjustment to sha.h would be required.
41 * As this implementation relies on 64-bit integer type, it's totally
42 * inappropriate for platforms which don't support it, most notably
48 #include <openssl/crypto.h>
49 #include <openssl/sha.h>
50 #include <openssl/opensslv.h>
52 #include "internal/cryptlib.h"
53 #include "internal/sha.h"
55 #if defined(__i386) || defined(__i386__) || defined(_M_IX86) || \
56 defined(__x86_64) || defined(_M_AMD64) || defined(_M_X64) || \
57 defined(__s390__) || defined(__s390x__) || \
58 defined(__aarch64__) || \
60 # define SHA512_BLOCK_CAN_MANAGE_UNALIGNED_DATA
63 int sha512_224_init(SHA512_CTX *c)
65 c->h[0] = U64(0x8c3d37c819544da2);
66 c->h[1] = U64(0x73e1996689dcd4d6);
67 c->h[2] = U64(0x1dfab7ae32ff9c82);
68 c->h[3] = U64(0x679dd514582f9fcf);
69 c->h[4] = U64(0x0f6d2b697bd44da8);
70 c->h[5] = U64(0x77e36f7304c48942);
71 c->h[6] = U64(0x3f9d85a86a1d36c8);
72 c->h[7] = U64(0x1112e6ad91d692a1);
77 c->md_len = SHA224_DIGEST_LENGTH;
81 int sha512_256_init(SHA512_CTX *c)
83 c->h[0] = U64(0x22312194fc2bf72c);
84 c->h[1] = U64(0x9f555fa3c84c64c2);
85 c->h[2] = U64(0x2393b86b6f53b151);
86 c->h[3] = U64(0x963877195940eabd);
87 c->h[4] = U64(0x96283ee2a88effe3);
88 c->h[5] = U64(0xbe5e1e2553863992);
89 c->h[6] = U64(0x2b0199fc2c85b8aa);
90 c->h[7] = U64(0x0eb72ddc81c52ca2);
95 c->md_len = SHA256_DIGEST_LENGTH;
99 int SHA384_Init(SHA512_CTX *c)
101 c->h[0] = U64(0xcbbb9d5dc1059ed8);
102 c->h[1] = U64(0x629a292a367cd507);
103 c->h[2] = U64(0x9159015a3070dd17);
104 c->h[3] = U64(0x152fecd8f70e5939);
105 c->h[4] = U64(0x67332667ffc00b31);
106 c->h[5] = U64(0x8eb44a8768581511);
107 c->h[6] = U64(0xdb0c2e0d64f98fa7);
108 c->h[7] = U64(0x47b5481dbefa4fa4);
113 c->md_len = SHA384_DIGEST_LENGTH;
117 int SHA512_Init(SHA512_CTX *c)
119 c->h[0] = U64(0x6a09e667f3bcc908);
120 c->h[1] = U64(0xbb67ae8584caa73b);
121 c->h[2] = U64(0x3c6ef372fe94f82b);
122 c->h[3] = U64(0xa54ff53a5f1d36f1);
123 c->h[4] = U64(0x510e527fade682d1);
124 c->h[5] = U64(0x9b05688c2b3e6c1f);
125 c->h[6] = U64(0x1f83d9abfb41bd6b);
126 c->h[7] = U64(0x5be0cd19137e2179);
131 c->md_len = SHA512_DIGEST_LENGTH;
138 void sha512_block_data_order(SHA512_CTX *ctx, const void *in, size_t num);
140 int SHA512_Final(unsigned char *md, SHA512_CTX *c)
142 unsigned char *p = (unsigned char *)c->u.p;
145 p[n] = 0x80; /* There always is a room for one */
147 if (n > (sizeof(c->u) - 16)) {
148 memset(p + n, 0, sizeof(c->u) - n);
150 sha512_block_data_order(c, p, 1);
153 memset(p + n, 0, sizeof(c->u) - 16 - n);
155 c->u.d[SHA_LBLOCK - 2] = c->Nh;
156 c->u.d[SHA_LBLOCK - 1] = c->Nl;
158 p[sizeof(c->u) - 1] = (unsigned char)(c->Nl);
159 p[sizeof(c->u) - 2] = (unsigned char)(c->Nl >> 8);
160 p[sizeof(c->u) - 3] = (unsigned char)(c->Nl >> 16);
161 p[sizeof(c->u) - 4] = (unsigned char)(c->Nl >> 24);
162 p[sizeof(c->u) - 5] = (unsigned char)(c->Nl >> 32);
163 p[sizeof(c->u) - 6] = (unsigned char)(c->Nl >> 40);
164 p[sizeof(c->u) - 7] = (unsigned char)(c->Nl >> 48);
165 p[sizeof(c->u) - 8] = (unsigned char)(c->Nl >> 56);
166 p[sizeof(c->u) - 9] = (unsigned char)(c->Nh);
167 p[sizeof(c->u) - 10] = (unsigned char)(c->Nh >> 8);
168 p[sizeof(c->u) - 11] = (unsigned char)(c->Nh >> 16);
169 p[sizeof(c->u) - 12] = (unsigned char)(c->Nh >> 24);
170 p[sizeof(c->u) - 13] = (unsigned char)(c->Nh >> 32);
171 p[sizeof(c->u) - 14] = (unsigned char)(c->Nh >> 40);
172 p[sizeof(c->u) - 15] = (unsigned char)(c->Nh >> 48);
173 p[sizeof(c->u) - 16] = (unsigned char)(c->Nh >> 56);
176 sha512_block_data_order(c, p, 1);
182 /* Let compiler decide if it's appropriate to unroll... */
183 case SHA224_DIGEST_LENGTH:
184 for (n = 0; n < SHA224_DIGEST_LENGTH / 8; n++) {
185 SHA_LONG64 t = c->h[n];
187 *(md++) = (unsigned char)(t >> 56);
188 *(md++) = (unsigned char)(t >> 48);
189 *(md++) = (unsigned char)(t >> 40);
190 *(md++) = (unsigned char)(t >> 32);
191 *(md++) = (unsigned char)(t >> 24);
192 *(md++) = (unsigned char)(t >> 16);
193 *(md++) = (unsigned char)(t >> 8);
194 *(md++) = (unsigned char)(t);
197 * For 224 bits, there are four bytes left over that have to be
198 * processed separately.
201 SHA_LONG64 t = c->h[SHA224_DIGEST_LENGTH / 8];
203 *(md++) = (unsigned char)(t >> 56);
204 *(md++) = (unsigned char)(t >> 48);
205 *(md++) = (unsigned char)(t >> 40);
206 *(md++) = (unsigned char)(t >> 32);
209 case SHA256_DIGEST_LENGTH:
210 for (n = 0; n < SHA256_DIGEST_LENGTH / 8; n++) {
211 SHA_LONG64 t = c->h[n];
213 *(md++) = (unsigned char)(t >> 56);
214 *(md++) = (unsigned char)(t >> 48);
215 *(md++) = (unsigned char)(t >> 40);
216 *(md++) = (unsigned char)(t >> 32);
217 *(md++) = (unsigned char)(t >> 24);
218 *(md++) = (unsigned char)(t >> 16);
219 *(md++) = (unsigned char)(t >> 8);
220 *(md++) = (unsigned char)(t);
223 case SHA384_DIGEST_LENGTH:
224 for (n = 0; n < SHA384_DIGEST_LENGTH / 8; n++) {
225 SHA_LONG64 t = c->h[n];
227 *(md++) = (unsigned char)(t >> 56);
228 *(md++) = (unsigned char)(t >> 48);
229 *(md++) = (unsigned char)(t >> 40);
230 *(md++) = (unsigned char)(t >> 32);
231 *(md++) = (unsigned char)(t >> 24);
232 *(md++) = (unsigned char)(t >> 16);
233 *(md++) = (unsigned char)(t >> 8);
234 *(md++) = (unsigned char)(t);
237 case SHA512_DIGEST_LENGTH:
238 for (n = 0; n < SHA512_DIGEST_LENGTH / 8; n++) {
239 SHA_LONG64 t = c->h[n];
241 *(md++) = (unsigned char)(t >> 56);
242 *(md++) = (unsigned char)(t >> 48);
243 *(md++) = (unsigned char)(t >> 40);
244 *(md++) = (unsigned char)(t >> 32);
245 *(md++) = (unsigned char)(t >> 24);
246 *(md++) = (unsigned char)(t >> 16);
247 *(md++) = (unsigned char)(t >> 8);
248 *(md++) = (unsigned char)(t);
251 /* ... as well as make sure md_len is not abused. */
259 int SHA384_Final(unsigned char *md, SHA512_CTX *c)
261 return SHA512_Final(md, c);
264 int SHA512_Update(SHA512_CTX *c, const void *_data, size_t len)
267 unsigned char *p = c->u.p;
268 const unsigned char *data = (const unsigned char *)_data;
273 l = (c->Nl + (((SHA_LONG64) len) << 3)) & U64(0xffffffffffffffff);
276 if (sizeof(len) >= 8)
277 c->Nh += (((SHA_LONG64) len) >> 61);
281 size_t n = sizeof(c->u) - c->num;
284 memcpy(p + c->num, data, len), c->num += (unsigned int)len;
287 memcpy(p + c->num, data, n), c->num = 0;
289 sha512_block_data_order(c, p, 1);
293 if (len >= sizeof(c->u)) {
294 #ifndef SHA512_BLOCK_CAN_MANAGE_UNALIGNED_DATA
295 if ((size_t)data % sizeof(c->u.d[0]) != 0)
296 while (len >= sizeof(c->u))
297 memcpy(p, data, sizeof(c->u)),
298 sha512_block_data_order(c, p, 1),
299 len -= sizeof(c->u), data += sizeof(c->u);
302 sha512_block_data_order(c, data, len / sizeof(c->u)),
303 data += len, len %= sizeof(c->u), data -= len;
307 memcpy(p, data, len), c->num = (int)len;
312 int SHA384_Update(SHA512_CTX *c, const void *data, size_t len)
314 return SHA512_Update(c, data, len);
317 void SHA512_Transform(SHA512_CTX *c, const unsigned char *data)
319 #ifndef SHA512_BLOCK_CAN_MANAGE_UNALIGNED_DATA
320 if ((size_t)data % sizeof(c->u.d[0]) != 0)
321 memcpy(c->u.p, data, sizeof(c->u.p)), data = c->u.p;
323 sha512_block_data_order(c, data, 1);
326 unsigned char *SHA384(const unsigned char *d, size_t n, unsigned char *md)
329 static unsigned char m[SHA384_DIGEST_LENGTH];
334 SHA512_Update(&c, d, n);
335 SHA512_Final(md, &c);
336 OPENSSL_cleanse(&c, sizeof(c));
340 unsigned char *SHA512(const unsigned char *d, size_t n, unsigned char *md)
343 static unsigned char m[SHA512_DIGEST_LENGTH];
348 SHA512_Update(&c, d, n);
349 SHA512_Final(md, &c);
350 OPENSSL_cleanse(&c, sizeof(c));
355 static const SHA_LONG64 K512[80] = {
356 U64(0x428a2f98d728ae22), U64(0x7137449123ef65cd),
357 U64(0xb5c0fbcfec4d3b2f), U64(0xe9b5dba58189dbbc),
358 U64(0x3956c25bf348b538), U64(0x59f111f1b605d019),
359 U64(0x923f82a4af194f9b), U64(0xab1c5ed5da6d8118),
360 U64(0xd807aa98a3030242), U64(0x12835b0145706fbe),
361 U64(0x243185be4ee4b28c), U64(0x550c7dc3d5ffb4e2),
362 U64(0x72be5d74f27b896f), U64(0x80deb1fe3b1696b1),
363 U64(0x9bdc06a725c71235), U64(0xc19bf174cf692694),
364 U64(0xe49b69c19ef14ad2), U64(0xefbe4786384f25e3),
365 U64(0x0fc19dc68b8cd5b5), U64(0x240ca1cc77ac9c65),
366 U64(0x2de92c6f592b0275), U64(0x4a7484aa6ea6e483),
367 U64(0x5cb0a9dcbd41fbd4), U64(0x76f988da831153b5),
368 U64(0x983e5152ee66dfab), U64(0xa831c66d2db43210),
369 U64(0xb00327c898fb213f), U64(0xbf597fc7beef0ee4),
370 U64(0xc6e00bf33da88fc2), U64(0xd5a79147930aa725),
371 U64(0x06ca6351e003826f), U64(0x142929670a0e6e70),
372 U64(0x27b70a8546d22ffc), U64(0x2e1b21385c26c926),
373 U64(0x4d2c6dfc5ac42aed), U64(0x53380d139d95b3df),
374 U64(0x650a73548baf63de), U64(0x766a0abb3c77b2a8),
375 U64(0x81c2c92e47edaee6), U64(0x92722c851482353b),
376 U64(0xa2bfe8a14cf10364), U64(0xa81a664bbc423001),
377 U64(0xc24b8b70d0f89791), U64(0xc76c51a30654be30),
378 U64(0xd192e819d6ef5218), U64(0xd69906245565a910),
379 U64(0xf40e35855771202a), U64(0x106aa07032bbd1b8),
380 U64(0x19a4c116b8d2d0c8), U64(0x1e376c085141ab53),
381 U64(0x2748774cdf8eeb99), U64(0x34b0bcb5e19b48a8),
382 U64(0x391c0cb3c5c95a63), U64(0x4ed8aa4ae3418acb),
383 U64(0x5b9cca4f7763e373), U64(0x682e6ff3d6b2b8a3),
384 U64(0x748f82ee5defb2fc), U64(0x78a5636f43172f60),
385 U64(0x84c87814a1f0ab72), U64(0x8cc702081a6439ec),
386 U64(0x90befffa23631e28), U64(0xa4506cebde82bde9),
387 U64(0xbef9a3f7b2c67915), U64(0xc67178f2e372532b),
388 U64(0xca273eceea26619c), U64(0xd186b8c721c0c207),
389 U64(0xeada7dd6cde0eb1e), U64(0xf57d4f7fee6ed178),
390 U64(0x06f067aa72176fba), U64(0x0a637dc5a2c898a6),
391 U64(0x113f9804bef90dae), U64(0x1b710b35131c471b),
392 U64(0x28db77f523047d84), U64(0x32caab7b40c72493),
393 U64(0x3c9ebe0a15c9bebc), U64(0x431d67c49c100d4c),
394 U64(0x4cc5d4becb3e42b6), U64(0x597f299cfc657e2a),
395 U64(0x5fcb6fab3ad6faec), U64(0x6c44198c4a475817)
399 # if defined(__GNUC__) && __GNUC__>=2 && \
400 !defined(OPENSSL_NO_ASM) && !defined(OPENSSL_NO_INLINE_ASM)
401 # if defined(__x86_64) || defined(__x86_64__)
402 # define ROTR(a,n) ({ SHA_LONG64 ret; \
407 # if !defined(B_ENDIAN)
408 # define PULL64(x) ({ SHA_LONG64 ret=*((const SHA_LONG64 *)(&(x))); \
413 # elif (defined(__i386) || defined(__i386__)) && !defined(B_ENDIAN)
414 # if defined(I386_ONLY)
415 # define PULL64(x) ({ const unsigned int *p=(const unsigned int *)(&(x));\
416 unsigned int hi=p[0],lo=p[1]; \
417 asm("xchgb %%ah,%%al;xchgb %%dh,%%dl;"\
418 "roll $16,%%eax; roll $16,%%edx; "\
419 "xchgb %%ah,%%al;xchgb %%dh,%%dl;"\
420 : "=a"(lo),"=d"(hi) \
421 : "0"(lo),"1"(hi) : "cc"); \
422 ((SHA_LONG64)hi)<<32|lo; })
424 # define PULL64(x) ({ const unsigned int *p=(const unsigned int *)(&(x));\
425 unsigned int hi=p[0],lo=p[1]; \
426 asm ("bswapl %0; bswapl %1;" \
427 : "=r"(lo),"=r"(hi) \
428 : "0"(lo),"1"(hi)); \
429 ((SHA_LONG64)hi)<<32|lo; })
431 # elif (defined(_ARCH_PPC) && defined(__64BIT__)) || defined(_ARCH_PPC64)
432 # define ROTR(a,n) ({ SHA_LONG64 ret; \
433 asm ("rotrdi %0,%1,%2" \
435 : "r"(a),"K"(n)); ret; })
436 # elif defined(__aarch64__)
437 # define ROTR(a,n) ({ SHA_LONG64 ret; \
438 asm ("ror %0,%1,%2" \
440 : "r"(a),"I"(n)); ret; })
441 # if defined(__BYTE_ORDER__) && defined(__ORDER_LITTLE_ENDIAN__) && \
442 __BYTE_ORDER__==__ORDER_LITTLE_ENDIAN__
443 # define PULL64(x) ({ SHA_LONG64 ret; \
446 : "r"(*((const SHA_LONG64 *)(&(x))))); ret; })
449 # elif defined(_MSC_VER)
450 # if defined(_WIN64) /* applies to both IA-64 and AMD64 */
451 # pragma intrinsic(_rotr64)
452 # define ROTR(a,n) _rotr64((a),n)
454 # if defined(_M_IX86) && !defined(OPENSSL_NO_ASM) && \
455 !defined(OPENSSL_NO_INLINE_ASM)
456 # if defined(I386_ONLY)
457 static SHA_LONG64 __fastcall __pull64be(const void *x)
459 _asm mov edx,[ecx + 0]
460 _asm mov eax,[ecx + 4]
469 static SHA_LONG64 __fastcall __pull64be(const void *x)
471 _asm mov edx,[ecx + 0]
472 _asm mov eax,[ecx + 4]
477 # define PULL64(x) __pull64be(&(x))
482 # define B(x,j) (((SHA_LONG64)(*(((const unsigned char *)(&x))+j)))<<((7-j)*8))
483 # define PULL64(x) (B(x,0)|B(x,1)|B(x,2)|B(x,3)|B(x,4)|B(x,5)|B(x,6)|B(x,7))
486 # define ROTR(x,s) (((x)>>s) | (x)<<(64-s))
488 # define Sigma0(x) (ROTR((x),28) ^ ROTR((x),34) ^ ROTR((x),39))
489 # define Sigma1(x) (ROTR((x),14) ^ ROTR((x),18) ^ ROTR((x),41))
490 # define sigma0(x) (ROTR((x),1) ^ ROTR((x),8) ^ ((x)>>7))
491 # define sigma1(x) (ROTR((x),19) ^ ROTR((x),61) ^ ((x)>>6))
492 # define Ch(x,y,z) (((x) & (y)) ^ ((~(x)) & (z)))
493 # define Maj(x,y,z) (((x) & (y)) ^ ((x) & (z)) ^ ((y) & (z)))
495 # if defined(__i386) || defined(__i386__) || defined(_M_IX86)
497 * This code should give better results on 32-bit CPU with less than
498 * ~24 registers, both size and performance wise...
501 static void sha512_block_data_order(SHA512_CTX *ctx, const void *in,
504 const SHA_LONG64 *W = in;
506 SHA_LONG64 X[9 + 80], *F;
521 for (i = 0; i < 16; i++, F--) {
530 T += F[7] + Sigma1(E) + Ch(E, F[5], F[6]) + K512[i];
532 A = T + Sigma0(A) + Maj(A, F[1], F[2]);
535 for (; i < 80; i++, F--) {
536 T = sigma0(F[8 + 16 - 1]);
537 T += sigma1(F[8 + 16 - 14]);
538 T += F[8 + 16] + F[8 + 16 - 9];
543 T += F[7] + Sigma1(E) + Ch(E, F[5], F[6]) + K512[i];
545 A = T + Sigma0(A) + Maj(A, F[1], F[2]);
561 # elif defined(OPENSSL_SMALL_FOOTPRINT)
563 static void sha512_block_data_order(SHA512_CTX *ctx, const void *in,
566 const SHA_LONG64 *W = in;
567 SHA_LONG64 a, b, c, d, e, f, g, h, s0, s1, T1, T2;
582 for (i = 0; i < 16; i++) {
586 T1 = X[i] = PULL64(W[i]);
588 T1 += h + Sigma1(e) + Ch(e, f, g) + K512[i];
589 T2 = Sigma0(a) + Maj(a, b, c);
600 for (; i < 80; i++) {
601 s0 = X[(i + 1) & 0x0f];
603 s1 = X[(i + 14) & 0x0f];
606 T1 = X[i & 0xf] += s0 + s1 + X[(i + 9) & 0xf];
607 T1 += h + Sigma1(e) + Ch(e, f, g) + K512[i];
608 T2 = Sigma0(a) + Maj(a, b, c);
633 # define ROUND_00_15(i,a,b,c,d,e,f,g,h) do { \
634 T1 += h + Sigma1(e) + Ch(e,f,g) + K512[i]; \
635 h = Sigma0(a) + Maj(a,b,c); \
636 d += T1; h += T1; } while (0)
638 # define ROUND_16_80(i,j,a,b,c,d,e,f,g,h,X) do { \
639 s0 = X[(j+1)&0x0f]; s0 = sigma0(s0); \
640 s1 = X[(j+14)&0x0f]; s1 = sigma1(s1); \
641 T1 = X[(j)&0x0f] += s0 + s1 + X[(j+9)&0x0f]; \
642 ROUND_00_15(i+j,a,b,c,d,e,f,g,h); } while (0)
644 static void sha512_block_data_order(SHA512_CTX *ctx, const void *in,
647 const SHA_LONG64 *W = in;
648 SHA_LONG64 a, b, c, d, e, f, g, h, s0, s1, T1;
665 ROUND_00_15(0, a, b, c, d, e, f, g, h);
667 ROUND_00_15(1, h, a, b, c, d, e, f, g);
669 ROUND_00_15(2, g, h, a, b, c, d, e, f);
671 ROUND_00_15(3, f, g, h, a, b, c, d, e);
673 ROUND_00_15(4, e, f, g, h, a, b, c, d);
675 ROUND_00_15(5, d, e, f, g, h, a, b, c);
677 ROUND_00_15(6, c, d, e, f, g, h, a, b);
679 ROUND_00_15(7, b, c, d, e, f, g, h, a);
681 ROUND_00_15(8, a, b, c, d, e, f, g, h);
683 ROUND_00_15(9, h, a, b, c, d, e, f, g);
685 ROUND_00_15(10, g, h, a, b, c, d, e, f);
687 ROUND_00_15(11, f, g, h, a, b, c, d, e);
689 ROUND_00_15(12, e, f, g, h, a, b, c, d);
691 ROUND_00_15(13, d, e, f, g, h, a, b, c);
693 ROUND_00_15(14, c, d, e, f, g, h, a, b);
695 ROUND_00_15(15, b, c, d, e, f, g, h, a);
697 T1 = X[0] = PULL64(W[0]);
698 ROUND_00_15(0, a, b, c, d, e, f, g, h);
699 T1 = X[1] = PULL64(W[1]);
700 ROUND_00_15(1, h, a, b, c, d, e, f, g);
701 T1 = X[2] = PULL64(W[2]);
702 ROUND_00_15(2, g, h, a, b, c, d, e, f);
703 T1 = X[3] = PULL64(W[3]);
704 ROUND_00_15(3, f, g, h, a, b, c, d, e);
705 T1 = X[4] = PULL64(W[4]);
706 ROUND_00_15(4, e, f, g, h, a, b, c, d);
707 T1 = X[5] = PULL64(W[5]);
708 ROUND_00_15(5, d, e, f, g, h, a, b, c);
709 T1 = X[6] = PULL64(W[6]);
710 ROUND_00_15(6, c, d, e, f, g, h, a, b);
711 T1 = X[7] = PULL64(W[7]);
712 ROUND_00_15(7, b, c, d, e, f, g, h, a);
713 T1 = X[8] = PULL64(W[8]);
714 ROUND_00_15(8, a, b, c, d, e, f, g, h);
715 T1 = X[9] = PULL64(W[9]);
716 ROUND_00_15(9, h, a, b, c, d, e, f, g);
717 T1 = X[10] = PULL64(W[10]);
718 ROUND_00_15(10, g, h, a, b, c, d, e, f);
719 T1 = X[11] = PULL64(W[11]);
720 ROUND_00_15(11, f, g, h, a, b, c, d, e);
721 T1 = X[12] = PULL64(W[12]);
722 ROUND_00_15(12, e, f, g, h, a, b, c, d);
723 T1 = X[13] = PULL64(W[13]);
724 ROUND_00_15(13, d, e, f, g, h, a, b, c);
725 T1 = X[14] = PULL64(W[14]);
726 ROUND_00_15(14, c, d, e, f, g, h, a, b);
727 T1 = X[15] = PULL64(W[15]);
728 ROUND_00_15(15, b, c, d, e, f, g, h, a);
731 for (i = 16; i < 80; i += 16) {
732 ROUND_16_80(i, 0, a, b, c, d, e, f, g, h, X);
733 ROUND_16_80(i, 1, h, a, b, c, d, e, f, g, X);
734 ROUND_16_80(i, 2, g, h, a, b, c, d, e, f, X);
735 ROUND_16_80(i, 3, f, g, h, a, b, c, d, e, X);
736 ROUND_16_80(i, 4, e, f, g, h, a, b, c, d, X);
737 ROUND_16_80(i, 5, d, e, f, g, h, a, b, c, X);
738 ROUND_16_80(i, 6, c, d, e, f, g, h, a, b, X);
739 ROUND_16_80(i, 7, b, c, d, e, f, g, h, a, X);
740 ROUND_16_80(i, 8, a, b, c, d, e, f, g, h, X);
741 ROUND_16_80(i, 9, h, a, b, c, d, e, f, g, X);
742 ROUND_16_80(i, 10, g, h, a, b, c, d, e, f, X);
743 ROUND_16_80(i, 11, f, g, h, a, b, c, d, e, X);
744 ROUND_16_80(i, 12, e, f, g, h, a, b, c, d, X);
745 ROUND_16_80(i, 13, d, e, f, g, h, a, b, c, X);
746 ROUND_16_80(i, 14, c, d, e, f, g, h, a, b, X);
747 ROUND_16_80(i, 15, b, c, d, e, f, g, h, a, X);
765 #endif /* SHA512_ASM */