| 1 |
/* crypto/sha/sha512.c */ |
| 2 |
/* ==================================================================== |
| 3 |
* Copyright (c) 2004 The OpenSSL Project. All rights reserved |
| 4 |
* according to the OpenSSL license [found in ../../LICENSE]. |
| 5 |
* ==================================================================== |
| 6 |
*/ |
| 7 |
#include <openssl/opensslconf.h> |
| 8 |
#if !defined(OPENSSL_NO_SHA) && !defined(OPENSSL_NO_SHA512) |
| 9 |
/*- |
| 10 |
* IMPLEMENTATION NOTES. |
| 11 |
* |
| 12 |
* As you might have noticed 32-bit hash algorithms: |
| 13 |
* |
| 14 |
* - permit SHA_LONG to be wider than 32-bit (case on CRAY); |
| 15 |
* - optimized versions implement two transform functions: one operating |
| 16 |
* on [aligned] data in host byte order and one - on data in input |
| 17 |
* stream byte order; |
| 18 |
* - share common byte-order neutral collector and padding function |
| 19 |
* implementations, ../md32_common.h; |
| 20 |
* |
| 21 |
* Neither of the above applies to this SHA-512 implementations. Reasons |
| 22 |
* [in reverse order] are: |
| 23 |
* |
| 24 |
* - it's the only 64-bit hash algorithm for the moment of this writing, |
| 25 |
* there is no need for common collector/padding implementation [yet]; |
| 26 |
* - by supporting only one transform function [which operates on |
| 27 |
* *aligned* data in input stream byte order, big-endian in this case] |
| 28 |
* we minimize burden of maintenance in two ways: a) collector/padding |
| 29 |
* function is simpler; b) only one transform function to stare at; |
| 30 |
* - SHA_LONG64 is required to be exactly 64-bit in order to be able to |
| 31 |
* apply a number of optimizations to mitigate potential performance |
| 32 |
* penalties caused by previous design decision; |
| 33 |
* |
| 34 |
* Caveat lector. |
| 35 |
* |
| 36 |
* Implementation relies on the fact that "long long" is 64-bit on |
| 37 |
* both 32- and 64-bit platforms. If some compiler vendor comes up |
| 38 |
* with 128-bit long long, adjustment to sha.h would be required. |
| 39 |
* As this implementation relies on 64-bit integer type, it's totally |
| 40 |
* inappropriate for platforms which don't support it, most notably |
| 41 |
* 16-bit platforms. |
| 42 |
* <[email protected]> |
| 43 |
*/ |
| 44 |
# include <stdlib.h> |
| 45 |
# include <string.h> |
| 46 |
|
| 47 |
# include <openssl/crypto.h> |
| 48 |
# include <openssl/sha.h> |
| 49 |
# include <openssl/opensslv.h> |
| 50 |
|
| 51 |
# include "cryptlib.h" |
| 52 |
|
| 53 |
const char SHA512_version[] = "SHA-512" OPENSSL_VERSION_PTEXT; |
| 54 |
|
| 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__) || \ |
| 59 |
defined(SHA512_ASM) |
| 60 |
# define SHA512_BLOCK_CAN_MANAGE_UNALIGNED_DATA |
| 61 |
# endif |
| 62 |
|
| 63 |
fips_md_init_ctx(SHA384, SHA512) |
| 64 |
{ |
| 65 |
c->h[0] = U64(0xcbbb9d5dc1059ed8); |
| 66 |
c->h[1] = U64(0x629a292a367cd507); |
| 67 |
c->h[2] = U64(0x9159015a3070dd17); |
| 68 |
c->h[3] = U64(0x152fecd8f70e5939); |
| 69 |
c->h[4] = U64(0x67332667ffc00b31); |
| 70 |
c->h[5] = U64(0x8eb44a8768581511); |
| 71 |
c->h[6] = U64(0xdb0c2e0d64f98fa7); |
| 72 |
c->h[7] = U64(0x47b5481dbefa4fa4); |
| 73 |
|
| 74 |
c->Nl = 0; |
| 75 |
c->Nh = 0; |
| 76 |
c->num = 0; |
| 77 |
c->md_len = SHA384_DIGEST_LENGTH; |
| 78 |
return 1; |
| 79 |
} |
| 80 |
|
| 81 |
fips_md_init(SHA512) |
| 82 |
{ |
| 83 |
c->h[0] = U64(0x6a09e667f3bcc908); |
| 84 |
c->h[1] = U64(0xbb67ae8584caa73b); |
| 85 |
c->h[2] = U64(0x3c6ef372fe94f82b); |
| 86 |
c->h[3] = U64(0xa54ff53a5f1d36f1); |
| 87 |
c->h[4] = U64(0x510e527fade682d1); |
| 88 |
c->h[5] = U64(0x9b05688c2b3e6c1f); |
| 89 |
c->h[6] = U64(0x1f83d9abfb41bd6b); |
| 90 |
c->h[7] = U64(0x5be0cd19137e2179); |
| 91 |
|
| 92 |
c->Nl = 0; |
| 93 |
c->Nh = 0; |
| 94 |
c->num = 0; |
| 95 |
c->md_len = SHA512_DIGEST_LENGTH; |
| 96 |
return 1; |
| 97 |
} |
| 98 |
|
| 99 |
# ifndef SHA512_ASM |
| 100 |
static |
| 101 |
# endif |
| 102 |
void sha512_block_data_order(SHA512_CTX *ctx, const void *in, size_t num); |
| 103 |
|
| 104 |
int SHA512_Final(unsigned char *md, SHA512_CTX *c) |
| 105 |
{ |
| 106 |
unsigned char *p = (unsigned char *)c->u.p; |
| 107 |
size_t n = c->num; |
| 108 |
|
| 109 |
p[n] = 0x80; /* There always is a room for one */ |
| 110 |
n++; |
| 111 |
if (n > (sizeof(c->u) - 16)) |
| 112 |
memset(p + n, 0, sizeof(c->u) - n), n = 0, |
| 113 |
sha512_block_data_order(c, p, 1); |
| 114 |
|
| 115 |
memset(p + n, 0, sizeof(c->u) - 16 - n); |
| 116 |
# ifdef B_ENDIAN |
| 117 |
c->u.d[SHA_LBLOCK - 2] = c->Nh; |
| 118 |
c->u.d[SHA_LBLOCK - 1] = c->Nl; |
| 119 |
# else |
| 120 |
p[sizeof(c->u) - 1] = (unsigned char)(c->Nl); |
| 121 |
p[sizeof(c->u) - 2] = (unsigned char)(c->Nl >> 8); |
| 122 |
p[sizeof(c->u) - 3] = (unsigned char)(c->Nl >> 16); |
| 123 |
p[sizeof(c->u) - 4] = (unsigned char)(c->Nl >> 24); |
| 124 |
p[sizeof(c->u) - 5] = (unsigned char)(c->Nl >> 32); |
| 125 |
p[sizeof(c->u) - 6] = (unsigned char)(c->Nl >> 40); |
| 126 |
p[sizeof(c->u) - 7] = (unsigned char)(c->Nl >> 48); |
| 127 |
p[sizeof(c->u) - 8] = (unsigned char)(c->Nl >> 56); |
| 128 |
p[sizeof(c->u) - 9] = (unsigned char)(c->Nh); |
| 129 |
p[sizeof(c->u) - 10] = (unsigned char)(c->Nh >> 8); |
| 130 |
p[sizeof(c->u) - 11] = (unsigned char)(c->Nh >> 16); |
| 131 |
p[sizeof(c->u) - 12] = (unsigned char)(c->Nh >> 24); |
| 132 |
p[sizeof(c->u) - 13] = (unsigned char)(c->Nh >> 32); |
| 133 |
p[sizeof(c->u) - 14] = (unsigned char)(c->Nh >> 40); |
| 134 |
p[sizeof(c->u) - 15] = (unsigned char)(c->Nh >> 48); |
| 135 |
p[sizeof(c->u) - 16] = (unsigned char)(c->Nh >> 56); |
| 136 |
# endif |
| 137 |
|
| 138 |
sha512_block_data_order(c, p, 1); |
| 139 |
|
| 140 |
if (md == 0) |
| 141 |
return 0; |
| 142 |
|
| 143 |
switch (c->md_len) { |
| 144 |
/* Let compiler decide if it's appropriate to unroll... */ |
| 145 |
case SHA384_DIGEST_LENGTH: |
| 146 |
for (n = 0; n < SHA384_DIGEST_LENGTH / 8; n++) { |
| 147 |
SHA_LONG64 t = c->h[n]; |
| 148 |
|
| 149 |
*(md++) = (unsigned char)(t >> 56); |
| 150 |
*(md++) = (unsigned char)(t >> 48); |
| 151 |
*(md++) = (unsigned char)(t >> 40); |
| 152 |
*(md++) = (unsigned char)(t >> 32); |
| 153 |
*(md++) = (unsigned char)(t >> 24); |
| 154 |
*(md++) = (unsigned char)(t >> 16); |
| 155 |
*(md++) = (unsigned char)(t >> 8); |
| 156 |
*(md++) = (unsigned char)(t); |
| 157 |
} |
| 158 |
break; |
| 159 |
case SHA512_DIGEST_LENGTH: |
| 160 |
for (n = 0; n < SHA512_DIGEST_LENGTH / 8; n++) { |
| 161 |
SHA_LONG64 t = c->h[n]; |
| 162 |
|
| 163 |
*(md++) = (unsigned char)(t >> 56); |
| 164 |
*(md++) = (unsigned char)(t >> 48); |
| 165 |
*(md++) = (unsigned char)(t >> 40); |
| 166 |
*(md++) = (unsigned char)(t >> 32); |
| 167 |
*(md++) = (unsigned char)(t >> 24); |
| 168 |
*(md++) = (unsigned char)(t >> 16); |
| 169 |
*(md++) = (unsigned char)(t >> 8); |
| 170 |
*(md++) = (unsigned char)(t); |
| 171 |
} |
| 172 |
break; |
| 173 |
/* ... as well as make sure md_len is not abused. */ |
| 174 |
default: |
| 175 |
return 0; |
| 176 |
} |
| 177 |
|
| 178 |
return 1; |
| 179 |
} |
| 180 |
|
| 181 |
int SHA384_Final(unsigned char *md, SHA512_CTX *c) |
| 182 |
{ |
| 183 |
return SHA512_Final(md, c); |
| 184 |
} |
| 185 |
|
| 186 |
int SHA512_Update(SHA512_CTX *c, const void *_data, size_t len) |
| 187 |
{ |
| 188 |
SHA_LONG64 l; |
| 189 |
unsigned char *p = c->u.p; |
| 190 |
const unsigned char *data = (const unsigned char *)_data; |
| 191 |
|
| 192 |
if (len == 0) |
| 193 |
return 1; |
| 194 |
|
| 195 |
l = (c->Nl + (((SHA_LONG64) len) << 3)) & U64(0xffffffffffffffff); |
| 196 |
if (l < c->Nl) |
| 197 |
c->Nh++; |
| 198 |
if (sizeof(len) >= 8) |
| 199 |
c->Nh += (((SHA_LONG64) len) >> 61); |
| 200 |
c->Nl = l; |
| 201 |
|
| 202 |
if (c->num != 0) { |
| 203 |
size_t n = sizeof(c->u) - c->num; |
| 204 |
|
| 205 |
if (len < n) { |
| 206 |
memcpy(p + c->num, data, len), c->num += (unsigned int)len; |
| 207 |
return 1; |
| 208 |
} else { |
| 209 |
memcpy(p + c->num, data, n), c->num = 0; |
| 210 |
len -= n, data += n; |
| 211 |
sha512_block_data_order(c, p, 1); |
| 212 |
} |
| 213 |
} |
| 214 |
|
| 215 |
if (len >= sizeof(c->u)) { |
| 216 |
# ifndef SHA512_BLOCK_CAN_MANAGE_UNALIGNED_DATA |
| 217 |
if ((size_t)data % sizeof(c->u.d[0]) != 0) |
| 218 |
while (len >= sizeof(c->u)) |
| 219 |
memcpy(p, data, sizeof(c->u)), |
| 220 |
sha512_block_data_order(c, p, 1), |
| 221 |
len -= sizeof(c->u), data += sizeof(c->u); |
| 222 |
else |
| 223 |
# endif |
| 224 |
sha512_block_data_order(c, data, len / sizeof(c->u)), |
| 225 |
data += len, len %= sizeof(c->u), data -= len; |
| 226 |
} |
| 227 |
|
| 228 |
if (len != 0) |
| 229 |
memcpy(p, data, len), c->num = (int)len; |
| 230 |
|
| 231 |
return 1; |
| 232 |
} |
| 233 |
|
| 234 |
int SHA384_Update(SHA512_CTX *c, const void *data, size_t len) |
| 235 |
{ |
| 236 |
return SHA512_Update(c, data, len); |
| 237 |
} |
| 238 |
|
| 239 |
void SHA512_Transform(SHA512_CTX *c, const unsigned char *data) |
| 240 |
{ |
| 241 |
# ifndef SHA512_BLOCK_CAN_MANAGE_UNALIGNED_DATA |
| 242 |
if ((size_t)data % sizeof(c->u.d[0]) != 0) |
| 243 |
memcpy(c->u.p, data, sizeof(c->u.p)), data = c->u.p; |
| 244 |
# endif |
| 245 |
sha512_block_data_order(c, data, 1); |
| 246 |
} |
| 247 |
|
| 248 |
unsigned char *SHA384(const unsigned char *d, size_t n, unsigned char *md) |
| 249 |
{ |
| 250 |
SHA512_CTX c; |
| 251 |
static unsigned char m[SHA384_DIGEST_LENGTH]; |
| 252 |
|
| 253 |
if (md == NULL) |
| 254 |
md = m; |
| 255 |
SHA384_Init(&c); |
| 256 |
SHA512_Update(&c, d, n); |
| 257 |
SHA512_Final(md, &c); |
| 258 |
OPENSSL_cleanse(&c, sizeof(c)); |
| 259 |
return (md); |
| 260 |
} |
| 261 |
|
| 262 |
unsigned char *SHA512(const unsigned char *d, size_t n, unsigned char *md) |
| 263 |
{ |
| 264 |
SHA512_CTX c; |
| 265 |
static unsigned char m[SHA512_DIGEST_LENGTH]; |
| 266 |
|
| 267 |
if (md == NULL) |
| 268 |
md = m; |
| 269 |
SHA512_Init(&c); |
| 270 |
SHA512_Update(&c, d, n); |
| 271 |
SHA512_Final(md, &c); |
| 272 |
OPENSSL_cleanse(&c, sizeof(c)); |
| 273 |
return (md); |
| 274 |
} |
| 275 |
|
| 276 |
# ifndef SHA512_ASM |
| 277 |
static const SHA_LONG64 K512[80] = { |
| 278 |
U64(0x428a2f98d728ae22), U64(0x7137449123ef65cd), |
| 279 |
U64(0xb5c0fbcfec4d3b2f), U64(0xe9b5dba58189dbbc), |
| 280 |
U64(0x3956c25bf348b538), U64(0x59f111f1b605d019), |
| 281 |
U64(0x923f82a4af194f9b), U64(0xab1c5ed5da6d8118), |
| 282 |
U64(0xd807aa98a3030242), U64(0x12835b0145706fbe), |
| 283 |
U64(0x243185be4ee4b28c), U64(0x550c7dc3d5ffb4e2), |
| 284 |
U64(0x72be5d74f27b896f), U64(0x80deb1fe3b1696b1), |
| 285 |
U64(0x9bdc06a725c71235), U64(0xc19bf174cf692694), |
| 286 |
U64(0xe49b69c19ef14ad2), U64(0xefbe4786384f25e3), |
| 287 |
U64(0x0fc19dc68b8cd5b5), U64(0x240ca1cc77ac9c65), |
| 288 |
U64(0x2de92c6f592b0275), U64(0x4a7484aa6ea6e483), |
| 289 |
U64(0x5cb0a9dcbd41fbd4), U64(0x76f988da831153b5), |
| 290 |
U64(0x983e5152ee66dfab), U64(0xa831c66d2db43210), |
| 291 |
U64(0xb00327c898fb213f), U64(0xbf597fc7beef0ee4), |
| 292 |
U64(0xc6e00bf33da88fc2), U64(0xd5a79147930aa725), |
| 293 |
U64(0x06ca6351e003826f), U64(0x142929670a0e6e70), |
| 294 |
U64(0x27b70a8546d22ffc), U64(0x2e1b21385c26c926), |
| 295 |
U64(0x4d2c6dfc5ac42aed), U64(0x53380d139d95b3df), |
| 296 |
U64(0x650a73548baf63de), U64(0x766a0abb3c77b2a8), |
| 297 |
U64(0x81c2c92e47edaee6), U64(0x92722c851482353b), |
| 298 |
U64(0xa2bfe8a14cf10364), U64(0xa81a664bbc423001), |
| 299 |
U64(0xc24b8b70d0f89791), U64(0xc76c51a30654be30), |
| 300 |
U64(0xd192e819d6ef5218), U64(0xd69906245565a910), |
| 301 |
U64(0xf40e35855771202a), U64(0x106aa07032bbd1b8), |
| 302 |
U64(0x19a4c116b8d2d0c8), U64(0x1e376c085141ab53), |
| 303 |
U64(0x2748774cdf8eeb99), U64(0x34b0bcb5e19b48a8), |
| 304 |
U64(0x391c0cb3c5c95a63), U64(0x4ed8aa4ae3418acb), |
| 305 |
U64(0x5b9cca4f7763e373), U64(0x682e6ff3d6b2b8a3), |
| 306 |
U64(0x748f82ee5defb2fc), U64(0x78a5636f43172f60), |
| 307 |
U64(0x84c87814a1f0ab72), U64(0x8cc702081a6439ec), |
| 308 |
U64(0x90befffa23631e28), U64(0xa4506cebde82bde9), |
| 309 |
U64(0xbef9a3f7b2c67915), U64(0xc67178f2e372532b), |
| 310 |
U64(0xca273eceea26619c), U64(0xd186b8c721c0c207), |
| 311 |
U64(0xeada7dd6cde0eb1e), U64(0xf57d4f7fee6ed178), |
| 312 |
U64(0x06f067aa72176fba), U64(0x0a637dc5a2c898a6), |
| 313 |
U64(0x113f9804bef90dae), U64(0x1b710b35131c471b), |
| 314 |
U64(0x28db77f523047d84), U64(0x32caab7b40c72493), |
| 315 |
U64(0x3c9ebe0a15c9bebc), U64(0x431d67c49c100d4c), |
| 316 |
U64(0x4cc5d4becb3e42b6), U64(0x597f299cfc657e2a), |
| 317 |
U64(0x5fcb6fab3ad6faec), U64(0x6c44198c4a475817) |
| 318 |
}; |
| 319 |
|
| 320 |
# ifndef PEDANTIC |
| 321 |
# if defined(__GNUC__) && __GNUC__>=2 && !defined(OPENSSL_NO_ASM) && !defined(OPENSSL_NO_INLINE_ASM) |
| 322 |
# if defined(__x86_64) || defined(__x86_64__) |
| 323 |
# define ROTR(a,n) ({ SHA_LONG64 ret; \ |
| 324 |
asm ("rorq %1,%0" \ |
| 325 |
: "=r"(ret) \ |
| 326 |
: "J"(n),"0"(a) \ |
| 327 |
: "cc"); ret; }) |
| 328 |
# if !defined(B_ENDIAN) |
| 329 |
# define PULL64(x) ({ SHA_LONG64 ret=*((const SHA_LONG64 *)(&(x))); \ |
| 330 |
asm ("bswapq %0" \ |
| 331 |
: "=r"(ret) \ |
| 332 |
: "0"(ret)); ret; }) |
| 333 |
# endif |
| 334 |
# elif (defined(__i386) || defined(__i386__)) && !defined(B_ENDIAN) |
| 335 |
# if defined(I386_ONLY) |
| 336 |
# define PULL64(x) ({ const unsigned int *p=(const unsigned int *)(&(x));\ |
| 337 |
unsigned int hi=p[0],lo=p[1]; \ |
| 338 |
asm("xchgb %%ah,%%al;xchgb %%dh,%%dl;"\ |
| 339 |
"roll $16,%%eax; roll $16,%%edx; "\ |
| 340 |
"xchgb %%ah,%%al;xchgb %%dh,%%dl;" \ |
| 341 |
: "=a"(lo),"=d"(hi) \ |
| 342 |
: "0"(lo),"1"(hi) : "cc"); \ |
| 343 |
((SHA_LONG64)hi)<<32|lo; }) |
| 344 |
# else |
| 345 |
# define PULL64(x) ({ const unsigned int *p=(const unsigned int *)(&(x));\ |
| 346 |
unsigned int hi=p[0],lo=p[1]; \ |
| 347 |
asm ("bswapl %0; bswapl %1;" \ |
| 348 |
: "=r"(lo),"=r"(hi) \ |
| 349 |
: "0"(lo),"1"(hi)); \ |
| 350 |
((SHA_LONG64)hi)<<32|lo; }) |
| 351 |
# endif |
| 352 |
# elif (defined(_ARCH_PPC) && defined(__64BIT__)) || defined(_ARCH_PPC64) |
| 353 |
# define ROTR(a,n) ({ SHA_LONG64 ret; \ |
| 354 |
asm ("rotrdi %0,%1,%2" \ |
| 355 |
: "=r"(ret) \ |
| 356 |
: "r"(a),"K"(n)); ret; }) |
| 357 |
# elif defined(__aarch64__) |
| 358 |
# define ROTR(a,n) ({ SHA_LONG64 ret; \ |
| 359 |
asm ("ror %0,%1,%2" \ |
| 360 |
: "=r"(ret) \ |
| 361 |
: "r"(a),"I"(n)); ret; }) |
| 362 |
# if defined(__BYTE_ORDER__) && defined(__ORDER_LITTLE_ENDIAN__) && \ |
| 363 |
__BYTE_ORDER__==__ORDER_LITTLE_ENDIAN__ |
| 364 |
# define PULL64(x) ({ SHA_LONG64 ret; \ |
| 365 |
asm ("rev %0,%1" \ |
| 366 |
: "=r"(ret) \ |
| 367 |
: "r"(*((const SHA_LONG64 *)(&(x))))); ret; }) |
| 368 |
# endif |
| 369 |
# endif |
| 370 |
# elif defined(_MSC_VER) |
| 371 |
# if defined(_WIN64) /* applies to both IA-64 and AMD64 */ |
| 372 |
# pragma intrinsic(_rotr64) |
| 373 |
# define ROTR(a,n) _rotr64((a),n) |
| 374 |
# endif |
| 375 |
# if defined(_M_IX86) && !defined(OPENSSL_NO_ASM) && !defined(OPENSSL_NO_INLINE_ASM) |
| 376 |
# if defined(I386_ONLY) |
| 377 |
static SHA_LONG64 __fastcall __pull64be(const void *x) |
| 378 |
{ |
| 379 |
_asm mov edx,[ecx + 0] |
| 380 |
_asm mov eax,[ecx + 4] |
| 381 |
_asm xchg dh, dl |
| 382 |
_asm xchg ah, al |
| 383 |
_asm rol edx, 16 _asm rol eax, 16 _asm xchg dh, dl _asm xchg ah, al} |
| 384 |
# else |
| 385 |
static SHA_LONG64 __fastcall __pull64be(const void *x) |
| 386 |
{ |
| 387 |
_asm mov edx,[ecx + 0] |
| 388 |
_asm mov eax,[ecx + 4] |
| 389 |
_asm bswap edx _asm bswap eax} |
| 390 |
# endif |
| 391 |
# define PULL64(x) __pull64be(&(x)) |
| 392 |
# if _MSC_VER<=1200 |
| 393 |
# pragma inline_depth(0) |
| 394 |
# endif |
| 395 |
# endif |
| 396 |
# endif |
| 397 |
# endif |
| 398 |
# ifndef PULL64 |
| 399 |
# define B(x,j) (((SHA_LONG64)(*(((const unsigned char *)(&x))+j)))<<((7-j)*8)) |
| 400 |
# 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)) |
| 401 |
# endif |
| 402 |
# ifndef ROTR |
| 403 |
# define ROTR(x,s) (((x)>>s) | (x)<<(64-s)) |
| 404 |
# endif |
| 405 |
# define Sigma0(x) (ROTR((x),28) ^ ROTR((x),34) ^ ROTR((x),39)) |
| 406 |
# define Sigma1(x) (ROTR((x),14) ^ ROTR((x),18) ^ ROTR((x),41)) |
| 407 |
# define sigma0(x) (ROTR((x),1) ^ ROTR((x),8) ^ ((x)>>7)) |
| 408 |
# define sigma1(x) (ROTR((x),19) ^ ROTR((x),61) ^ ((x)>>6)) |
| 409 |
# define Ch(x,y,z) (((x) & (y)) ^ ((~(x)) & (z))) |
| 410 |
# define Maj(x,y,z) (((x) & (y)) ^ ((x) & (z)) ^ ((y) & (z))) |
| 411 |
# if defined(__i386) || defined(__i386__) || defined(_M_IX86) |
| 412 |
/* |
| 413 |
* This code should give better results on 32-bit CPU with less than |
| 414 |
* ~24 registers, both size and performance wise... |
| 415 |
*/ static void sha512_block_data_order(SHA512_CTX *ctx, const void *in, |
| 416 |
size_t num) |
| 417 |
{ |
| 418 |
const SHA_LONG64 *W = in; |
| 419 |
SHA_LONG64 A, E, T; |
| 420 |
SHA_LONG64 X[9 + 80], *F; |
| 421 |
int i; |
| 422 |
|
| 423 |
while (num--) { |
| 424 |
|
| 425 |
F = X + 80; |
| 426 |
A = ctx->h[0]; |
| 427 |
F[1] = ctx->h[1]; |
| 428 |
F[2] = ctx->h[2]; |
| 429 |
F[3] = ctx->h[3]; |
| 430 |
E = ctx->h[4]; |
| 431 |
F[5] = ctx->h[5]; |
| 432 |
F[6] = ctx->h[6]; |
| 433 |
F[7] = ctx->h[7]; |
| 434 |
|
| 435 |
for (i = 0; i < 16; i++, F--) { |
| 436 |
# ifdef B_ENDIAN |
| 437 |
T = W[i]; |
| 438 |
# else |
| 439 |
T = PULL64(W[i]); |
| 440 |
# endif |
| 441 |
F[0] = A; |
| 442 |
F[4] = E; |
| 443 |
F[8] = T; |
| 444 |
T += F[7] + Sigma1(E) + Ch(E, F[5], F[6]) + K512[i]; |
| 445 |
E = F[3] + T; |
| 446 |
A = T + Sigma0(A) + Maj(A, F[1], F[2]); |
| 447 |
} |
| 448 |
|
| 449 |
for (; i < 80; i++, F--) { |
| 450 |
T = sigma0(F[8 + 16 - 1]); |
| 451 |
T += sigma1(F[8 + 16 - 14]); |
| 452 |
T += F[8 + 16] + F[8 + 16 - 9]; |
| 453 |
|
| 454 |
F[0] = A; |
| 455 |
F[4] = E; |
| 456 |
F[8] = T; |
| 457 |
T += F[7] + Sigma1(E) + Ch(E, F[5], F[6]) + K512[i]; |
| 458 |
E = F[3] + T; |
| 459 |
A = T + Sigma0(A) + Maj(A, F[1], F[2]); |
| 460 |
} |
| 461 |
|
| 462 |
ctx->h[0] += A; |
| 463 |
ctx->h[1] += F[1]; |
| 464 |
ctx->h[2] += F[2]; |
| 465 |
ctx->h[3] += F[3]; |
| 466 |
ctx->h[4] += E; |
| 467 |
ctx->h[5] += F[5]; |
| 468 |
ctx->h[6] += F[6]; |
| 469 |
ctx->h[7] += F[7]; |
| 470 |
|
| 471 |
W += SHA_LBLOCK; |
| 472 |
} |
| 473 |
} |
| 474 |
|
| 475 |
# elif defined(OPENSSL_SMALL_FOOTPRINT) |
| 476 |
static void sha512_block_data_order(SHA512_CTX *ctx, const void *in, |
| 477 |
size_t num) |
| 478 |
{ |
| 479 |
const SHA_LONG64 *W = in; |
| 480 |
SHA_LONG64 a, b, c, d, e, f, g, h, s0, s1, T1, T2; |
| 481 |
SHA_LONG64 X[16]; |
| 482 |
int i; |
| 483 |
|
| 484 |
while (num--) { |
| 485 |
|
| 486 |
a = ctx->h[0]; |
| 487 |
b = ctx->h[1]; |
| 488 |
c = ctx->h[2]; |
| 489 |
d = ctx->h[3]; |
| 490 |
e = ctx->h[4]; |
| 491 |
f = ctx->h[5]; |
| 492 |
g = ctx->h[6]; |
| 493 |
h = ctx->h[7]; |
| 494 |
|
| 495 |
for (i = 0; i < 16; i++) { |
| 496 |
# ifdef B_ENDIAN |
| 497 |
T1 = X[i] = W[i]; |
| 498 |
# else |
| 499 |
T1 = X[i] = PULL64(W[i]); |
| 500 |
# endif |
| 501 |
T1 += h + Sigma1(e) + Ch(e, f, g) + K512[i]; |
| 502 |
T2 = Sigma0(a) + Maj(a, b, c); |
| 503 |
h = g; |
| 504 |
g = f; |
| 505 |
f = e; |
| 506 |
e = d + T1; |
| 507 |
d = c; |
| 508 |
c = b; |
| 509 |
b = a; |
| 510 |
a = T1 + T2; |
| 511 |
} |
| 512 |
|
| 513 |
for (; i < 80; i++) { |
| 514 |
s0 = X[(i + 1) & 0x0f]; |
| 515 |
s0 = sigma0(s0); |
| 516 |
s1 = X[(i + 14) & 0x0f]; |
| 517 |
s1 = sigma1(s1); |
| 518 |
|
| 519 |
T1 = X[i & 0xf] += s0 + s1 + X[(i + 9) & 0xf]; |
| 520 |
T1 += h + Sigma1(e) + Ch(e, f, g) + K512[i]; |
| 521 |
T2 = Sigma0(a) + Maj(a, b, c); |
| 522 |
h = g; |
| 523 |
g = f; |
| 524 |
f = e; |
| 525 |
e = d + T1; |
| 526 |
d = c; |
| 527 |
c = b; |
| 528 |
b = a; |
| 529 |
a = T1 + T2; |
| 530 |
} |
| 531 |
|
| 532 |
ctx->h[0] += a; |
| 533 |
ctx->h[1] += b; |
| 534 |
ctx->h[2] += c; |
| 535 |
ctx->h[3] += d; |
| 536 |
ctx->h[4] += e; |
| 537 |
ctx->h[5] += f; |
| 538 |
ctx->h[6] += g; |
| 539 |
ctx->h[7] += h; |
| 540 |
|
| 541 |
W += SHA_LBLOCK; |
| 542 |
} |
| 543 |
} |
| 544 |
|
| 545 |
# else |
| 546 |
# define ROUND_00_15(i,a,b,c,d,e,f,g,h) do { \ |
| 547 |
T1 += h + Sigma1(e) + Ch(e,f,g) + K512[i]; \ |
| 548 |
h = Sigma0(a) + Maj(a,b,c); \ |
| 549 |
d += T1; h += T1; } while (0) |
| 550 |
# define ROUND_16_80(i,j,a,b,c,d,e,f,g,h,X) do { \ |
| 551 |
s0 = X[(j+1)&0x0f]; s0 = sigma0(s0); \ |
| 552 |
s1 = X[(j+14)&0x0f]; s1 = sigma1(s1); \ |
| 553 |
T1 = X[(j)&0x0f] += s0 + s1 + X[(j+9)&0x0f]; \ |
| 554 |
ROUND_00_15(i+j,a,b,c,d,e,f,g,h); } while (0) |
| 555 |
static void sha512_block_data_order(SHA512_CTX *ctx, const void *in, |
| 556 |
size_t num) |
| 557 |
{ |
| 558 |
const SHA_LONG64 *W = in; |
| 559 |
SHA_LONG64 a, b, c, d, e, f, g, h, s0, s1, T1; |
| 560 |
SHA_LONG64 X[16]; |
| 561 |
int i; |
| 562 |
|
| 563 |
while (num--) { |
| 564 |
|
| 565 |
a = ctx->h[0]; |
| 566 |
b = ctx->h[1]; |
| 567 |
c = ctx->h[2]; |
| 568 |
d = ctx->h[3]; |
| 569 |
e = ctx->h[4]; |
| 570 |
f = ctx->h[5]; |
| 571 |
g = ctx->h[6]; |
| 572 |
h = ctx->h[7]; |
| 573 |
|
| 574 |
# ifdef B_ENDIAN |
| 575 |
T1 = X[0] = W[0]; |
| 576 |
ROUND_00_15(0, a, b, c, d, e, f, g, h); |
| 577 |
T1 = X[1] = W[1]; |
| 578 |
ROUND_00_15(1, h, a, b, c, d, e, f, g); |
| 579 |
T1 = X[2] = W[2]; |
| 580 |
ROUND_00_15(2, g, h, a, b, c, d, e, f); |
| 581 |
T1 = X[3] = W[3]; |
| 582 |
ROUND_00_15(3, f, g, h, a, b, c, d, e); |
| 583 |
T1 = X[4] = W[4]; |
| 584 |
ROUND_00_15(4, e, f, g, h, a, b, c, d); |
| 585 |
T1 = X[5] = W[5]; |
| 586 |
ROUND_00_15(5, d, e, f, g, h, a, b, c); |
| 587 |
T1 = X[6] = W[6]; |
| 588 |
ROUND_00_15(6, c, d, e, f, g, h, a, b); |
| 589 |
T1 = X[7] = W[7]; |
| 590 |
ROUND_00_15(7, b, c, d, e, f, g, h, a); |
| 591 |
T1 = X[8] = W[8]; |
| 592 |
ROUND_00_15(8, a, b, c, d, e, f, g, h); |
| 593 |
T1 = X[9] = W[9]; |
| 594 |
ROUND_00_15(9, h, a, b, c, d, e, f, g); |
| 595 |
T1 = X[10] = W[10]; |
| 596 |
ROUND_00_15(10, g, h, a, b, c, d, e, f); |
| 597 |
T1 = X[11] = W[11]; |
| 598 |
ROUND_00_15(11, f, g, h, a, b, c, d, e); |
| 599 |
T1 = X[12] = W[12]; |
| 600 |
ROUND_00_15(12, e, f, g, h, a, b, c, d); |
| 601 |
T1 = X[13] = W[13]; |
| 602 |
ROUND_00_15(13, d, e, f, g, h, a, b, c); |
| 603 |
T1 = X[14] = W[14]; |
| 604 |
ROUND_00_15(14, c, d, e, f, g, h, a, b); |
| 605 |
T1 = X[15] = W[15]; |
| 606 |
ROUND_00_15(15, b, c, d, e, f, g, h, a); |
| 607 |
# else |
| 608 |
T1 = X[0] = PULL64(W[0]); |
| 609 |
ROUND_00_15(0, a, b, c, d, e, f, g, h); |
| 610 |
T1 = X[1] = PULL64(W[1]); |
| 611 |
ROUND_00_15(1, h, a, b, c, d, e, f, g); |
| 612 |
T1 = X[2] = PULL64(W[2]); |
| 613 |
ROUND_00_15(2, g, h, a, b, c, d, e, f); |
| 614 |
T1 = X[3] = PULL64(W[3]); |
| 615 |
ROUND_00_15(3, f, g, h, a, b, c, d, e); |
| 616 |
T1 = X[4] = PULL64(W[4]); |
| 617 |
ROUND_00_15(4, e, f, g, h, a, b, c, d); |
| 618 |
T1 = X[5] = PULL64(W[5]); |
| 619 |
ROUND_00_15(5, d, e, f, g, h, a, b, c); |
| 620 |
T1 = X[6] = PULL64(W[6]); |
| 621 |
ROUND_00_15(6, c, d, e, f, g, h, a, b); |
| 622 |
T1 = X[7] = PULL64(W[7]); |
| 623 |
ROUND_00_15(7, b, c, d, e, f, g, h, a); |
| 624 |
T1 = X[8] = PULL64(W[8]); |
| 625 |
ROUND_00_15(8, a, b, c, d, e, f, g, h); |
| 626 |
T1 = X[9] = PULL64(W[9]); |
| 627 |
ROUND_00_15(9, h, a, b, c, d, e, f, g); |
| 628 |
T1 = X[10] = PULL64(W[10]); |
| 629 |
ROUND_00_15(10, g, h, a, b, c, d, e, f); |
| 630 |
T1 = X[11] = PULL64(W[11]); |
| 631 |
ROUND_00_15(11, f, g, h, a, b, c, d, e); |
| 632 |
T1 = X[12] = PULL64(W[12]); |
| 633 |
ROUND_00_15(12, e, f, g, h, a, b, c, d); |
| 634 |
T1 = X[13] = PULL64(W[13]); |
| 635 |
ROUND_00_15(13, d, e, f, g, h, a, b, c); |
| 636 |
T1 = X[14] = PULL64(W[14]); |
| 637 |
ROUND_00_15(14, c, d, e, f, g, h, a, b); |
| 638 |
T1 = X[15] = PULL64(W[15]); |
| 639 |
ROUND_00_15(15, b, c, d, e, f, g, h, a); |
| 640 |
# endif |
| 641 |
|
| 642 |
for (i = 16; i < 80; i += 16) { |
| 643 |
ROUND_16_80(i, 0, a, b, c, d, e, f, g, h, X); |
| 644 |
ROUND_16_80(i, 1, h, a, b, c, d, e, f, g, X); |
| 645 |
ROUND_16_80(i, 2, g, h, a, b, c, d, e, f, X); |
| 646 |
ROUND_16_80(i, 3, f, g, h, a, b, c, d, e, X); |
| 647 |
ROUND_16_80(i, 4, e, f, g, h, a, b, c, d, X); |
| 648 |
ROUND_16_80(i, 5, d, e, f, g, h, a, b, c, X); |
| 649 |
ROUND_16_80(i, 6, c, d, e, f, g, h, a, b, X); |
| 650 |
ROUND_16_80(i, 7, b, c, d, e, f, g, h, a, X); |
| 651 |
ROUND_16_80(i, 8, a, b, c, d, e, f, g, h, X); |
| 652 |
ROUND_16_80(i, 9, h, a, b, c, d, e, f, g, X); |
| 653 |
ROUND_16_80(i, 10, g, h, a, b, c, d, e, f, X); |
| 654 |
ROUND_16_80(i, 11, f, g, h, a, b, c, d, e, X); |
| 655 |
ROUND_16_80(i, 12, e, f, g, h, a, b, c, d, X); |
| 656 |
ROUND_16_80(i, 13, d, e, f, g, h, a, b, c, X); |
| 657 |
ROUND_16_80(i, 14, c, d, e, f, g, h, a, b, X); |
| 658 |
ROUND_16_80(i, 15, b, c, d, e, f, g, h, a, X); |
| 659 |
} |
| 660 |
|
| 661 |
ctx->h[0] += a; |
| 662 |
ctx->h[1] += b; |
| 663 |
ctx->h[2] += c; |
| 664 |
ctx->h[3] += d; |
| 665 |
ctx->h[4] += e; |
| 666 |
ctx->h[5] += f; |
| 667 |
ctx->h[6] += g; |
| 668 |
ctx->h[7] += h; |
| 669 |
|
| 670 |
W += SHA_LBLOCK; |
| 671 |
} |
| 672 |
} |
| 673 |
|
| 674 |
# endif |
| 675 |
|
| 676 |
# endif /* SHA512_ASM */ |
| 677 |
|
| 678 |
#else /* !OPENSSL_NO_SHA512 */ |
| 679 |
|
| 680 |
# if defined(PEDANTIC) || defined(__DECC) || defined(OPENSSL_SYS_MACOSX) |
| 681 |
static void *dummy = &dummy; |
| 682 |
# endif |
| 683 |
|
| 684 |
#endif /* !OPENSSL_NO_SHA512 */ |