lib/sha256.c

   1 /* sha256.c - Functions to compute SHA256 and SHA224 message digest of files or
   2    memory blocks according to the NIST specification FIPS-180-2.
   3
   4    Copyright (C) 2005-2006, 2008-2012 Free Software Foundation, Inc.
   5
   6    This program is free software: you can redistribute it and/or modify
   7    it under the terms of the GNU General Public License as published by
   8    the Free Software Foundation, either version 3 of the License, or
   9    (at your option) any later version.
  10
  11    This program is distributed in the hope that it will be useful,
  12    but WITHOUT ANY WARRANTY; without even the implied warranty of
  13    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  14    GNU General Public License for more details.
  15
  16    You should have received a copy of the GNU General Public License
  17    along with this program.  If not, see <http://www.gnu.org/licenses/>.  */
  18
  19 /* Written by David Madore, considerably copypasting from
  20    Scott G. Miller's sha1.c
  21 */
  22
  23 #include <config.h>
  24
  25 #include "sha256.h"
  26
  27 #include <stdalign.h>
  28 #include <stdint.h>
  29 #include <stdlib.h>
  30 #include <string.h>
  31
  32 #if USE_UNLOCKED_IO
  33 # include "unlocked-io.h"
  34 #endif
  35
  36 #ifdef WORDS_BIGENDIAN
  37 # define SWAP(n) (n)
  38 #else
  39 # define SWAP(n) \
  40     (((n) << 24) | (((n) & 0xff00) << 8) | (((n) >> 8) & 0xff00) | ((n) >> 24))
  41 #endif
  42
  43 #define BLOCKSIZE 32768
  44 #if BLOCKSIZE % 64 != 0
  45 # error "invalid BLOCKSIZE"
  46 #endif
  47
  48 /* This array contains the bytes used to pad the buffer to the next
  49    64-byte boundary.  */
  50 static const unsigned char fillbuf[64] = { 0x80, 0 /* , 0, 0, ...  */ };
  51
  52
  53 /*
  54   Takes a pointer to a 256 bit block of data (eight 32 bit ints) and
  55   initializes it to the start constants of the SHA256 algorithm.  This
  56   must be called before using hash in the call to sha256_hash
  57 */
  58 void
  59 sha256_init_ctx (struct sha256_ctx *ctx)
  60 {
  61   ctx->state[0] = 0x6a09e667UL;
  62   ctx->state[1] = 0xbb67ae85UL;
  63   ctx->state[2] = 0x3c6ef372UL;
  64   ctx->state[3] = 0xa54ff53aUL;
  65   ctx->state[4] = 0x510e527fUL;
  66   ctx->state[5] = 0x9b05688cUL;
  67   ctx->state[6] = 0x1f83d9abUL;
  68   ctx->state[7] = 0x5be0cd19UL;
  69
  70   ctx->total[0] = ctx->total[1] = 0;
  71   ctx->buflen = 0;
  72 }
  73
  74 void
  75 sha224_init_ctx (struct sha256_ctx *ctx)
  76 {
  77   ctx->state[0] = 0xc1059ed8UL;
  78   ctx->state[1] = 0x367cd507UL;
  79   ctx->state[2] = 0x3070dd17UL;
  80   ctx->state[3] = 0xf70e5939UL;
  81   ctx->state[4] = 0xffc00b31UL;
  82   ctx->state[5] = 0x68581511UL;
  83   ctx->state[6] = 0x64f98fa7UL;
  84   ctx->state[7] = 0xbefa4fa4UL;
  85
  86   ctx->total[0] = ctx->total[1] = 0;
  87   ctx->buflen = 0;
  88 }
  89
  90 /* Copy the value from v into the memory location pointed to by *cp,
  91    If your architecture allows unaligned access this is equivalent to
  92    * (uint32_t *) cp = v  */
  93 static void
  94 set_uint32 (char *cp, uint32_t v)
  95 {
  96   memcpy (cp, &v, sizeof v);
  97 }
  98
  99 /* Put result from CTX in first 32 bytes following RESBUF.  The result
 100    must be in little endian byte order.  */
 101 void *
 102 sha256_read_ctx (const struct sha256_ctx *ctx, void *resbuf)
 103 {
 104   int i;
 105   char *r = resbuf;
 106
 107   for (i = 0; i < 8; i++)
 108     set_uint32 (r + i * sizeof ctx->state[0], SWAP (ctx->state[i]));
 109
 110   return resbuf;
 111 }
 112
 113 void *
 114 sha224_read_ctx (const struct sha256_ctx *ctx, void *resbuf)
 115 {
 116   int i;
 117   char *r = resbuf;
 118
 119   for (i = 0; i < 7; i++)
 120     set_uint32 (r + i * sizeof ctx->state[0], SWAP (ctx->state[i]));
 121
 122   return resbuf;
 123 }
 124
 125 /* Process the remaining bytes in the internal buffer and the usual
 126    prolog according to the standard and write the result to RESBUF.  */
 127 static void
 128 sha256_conclude_ctx (struct sha256_ctx *ctx)
 129 {
 130   /* Take yet unprocessed bytes into account.  */
 131   size_t bytes = ctx->buflen;
 132   size_t size = (bytes < 56) ? 64 / 4 : 64 * 2 / 4;
 133
 134   /* Now count remaining bytes.  */
 135   ctx->total[0] += bytes;
 136   if (ctx->total[0] < bytes)
 137     ++ctx->total[1];
 138
 139   /* Put the 64-bit file length in *bits* at the end of the buffer.
 140      Use set_uint32 rather than a simple assignment, to avoid risk of
 141      unaligned access.  */
 142   set_uint32 ((char *) &ctx->buffer[size - 2],
 143               SWAP ((ctx->total[1] << 3) | (ctx->total[0] >> 29)));
 144   set_uint32 ((char *) &ctx->buffer[size - 1],
 145               SWAP (ctx->total[0] << 3));
 146
 147   memcpy (&((char *) ctx->buffer)[bytes], fillbuf, (size - 2) * 4 - bytes);
 148
 149   /* Process last bytes.  */
 150   sha256_process_block (ctx->buffer, size * 4, ctx);
 151 }
 152
 153 void *
 154 sha256_finish_ctx (struct sha256_ctx *ctx, void *resbuf)
 155 {
 156   sha256_conclude_ctx (ctx);
 157   return sha256_read_ctx (ctx, resbuf);
 158 }
 159
 160 void *
 161 sha224_finish_ctx (struct sha256_ctx *ctx, void *resbuf)
 162 {
 163   sha256_conclude_ctx (ctx);
 164   return sha224_read_ctx (ctx, resbuf);
 165 }
 166
 167 /* Compute SHA256 message digest for bytes read from STREAM.  The
 168    resulting message digest number will be written into the 32 bytes
 169    beginning at RESBLOCK.  */
 170 int
 171 sha256_stream (FILE *stream, void *resblock)
 172 {
 173   struct sha256_ctx ctx;
 174   size_t sum;
 175
 176   char *buffer = malloc (BLOCKSIZE + 72);
 177   if (!buffer)
 178     return 1;
 179
 180   /* Initialize the computation context.  */
 181   sha256_init_ctx (&ctx);
 182
 183   /* Iterate over full file contents.  */
 184   while (1)
 185     {
 186       /* We read the file in blocks of BLOCKSIZE bytes.  One call of the
 187          computation function processes the whole buffer so that with the
 188          next round of the loop another block can be read.  */
 189       size_t n;
 190       sum = 0;
 191
 192       /* Read block.  Take care for partial reads.  */
 193       while (1)
 194         {
 195           n = fread (buffer + sum, 1, BLOCKSIZE - sum, stream);
 196
 197           sum += n;
 198
 199           if (sum == BLOCKSIZE)
 200             break;
 201
 202           if (n == 0)
 203             {
 204               /* Check for the error flag IFF N == 0, so that we don't
 205                  exit the loop after a partial read due to e.g., EAGAIN
 206                  or EWOULDBLOCK.  */
 207               if (ferror (stream))
 208                 {
 209                   free (buffer);
 210                   return 1;
 211                 }
 212               goto process_partial_block;
 213             }
 214
 215           /* We've read at least one byte, so ignore errors.  But always
 216              check for EOF, since feof may be true even though N > 0.
 217              Otherwise, we could end up calling fread after EOF.  */
 218           if (feof (stream))
 219             goto process_partial_block;
 220         }
 221
 222       /* Process buffer with BLOCKSIZE bytes.  Note that
 223                         BLOCKSIZE % 64 == 0
 224        */
 225       sha256_process_block (buffer, BLOCKSIZE, &ctx);
 226     }
 227
 228  process_partial_block:;
 229
 230   /* Process any remaining bytes.  */
 231   if (sum > 0)
 232     sha256_process_bytes (buffer, sum, &ctx);
 233
 234   /* Construct result in desired memory.  */
 235   sha256_finish_ctx (&ctx, resblock);
 236   free (buffer);
 237   return 0;
 238 }
 239
 240 /* FIXME: Avoid code duplication */
 241 int
 242 sha224_stream (FILE *stream, void *resblock)
 243 {
 244   struct sha256_ctx ctx;
 245   size_t sum;
 246
 247   char *buffer = malloc (BLOCKSIZE + 72);
 248   if (!buffer)
 249     return 1;
 250
 251   /* Initialize the computation context.  */
 252   sha224_init_ctx (&ctx);
 253
 254   /* Iterate over full file contents.  */
 255   while (1)
 256     {
 257       /* We read the file in blocks of BLOCKSIZE bytes.  One call of the
 258          computation function processes the whole buffer so that with the
 259          next round of the loop another block can be read.  */
 260       size_t n;
 261       sum = 0;
 262
 263       /* Read block.  Take care for partial reads.  */
 264       while (1)
 265         {
 266           n = fread (buffer + sum, 1, BLOCKSIZE - sum, stream);
 267
 268           sum += n;
 269
 270           if (sum == BLOCKSIZE)
 271             break;
 272
 273           if (n == 0)
 274             {
 275               /* Check for the error flag IFF N == 0, so that we don't
 276                  exit the loop after a partial read due to e.g., EAGAIN
 277                  or EWOULDBLOCK.  */
 278               if (ferror (stream))
 279                 {
 280                   free (buffer);
 281                   return 1;
 282                 }
 283               goto process_partial_block;
 284             }
 285
 286           /* We've read at least one byte, so ignore errors.  But always
 287              check for EOF, since feof may be true even though N > 0.
 288              Otherwise, we could end up calling fread after EOF.  */
 289           if (feof (stream))
 290             goto process_partial_block;
 291         }
 292
 293       /* Process buffer with BLOCKSIZE bytes.  Note that
 294                         BLOCKSIZE % 64 == 0
 295        */
 296       sha256_process_block (buffer, BLOCKSIZE, &ctx);
 297     }
 298
 299  process_partial_block:;
 300
 301   /* Process any remaining bytes.  */
 302   if (sum > 0)
 303     sha256_process_bytes (buffer, sum, &ctx);
 304
 305   /* Construct result in desired memory.  */
 306   sha224_finish_ctx (&ctx, resblock);
 307   free (buffer);
 308   return 0;
 309 }
 310
 311 /* Compute SHA512 message digest for LEN bytes beginning at BUFFER.  The
 312    result is always in little endian byte order, so that a byte-wise
 313    output yields to the wanted ASCII representation of the message
 314    digest.  */
 315 void *
 316 sha256_buffer (const char *buffer, size_t len, void *resblock)
 317 {
 318   struct sha256_ctx ctx;
 319
 320   /* Initialize the computation context.  */
 321   sha256_init_ctx (&ctx);
 322
 323   /* Process whole buffer but last len % 64 bytes.  */
 324   sha256_process_bytes (buffer, len, &ctx);
 325
 326   /* Put result in desired memory area.  */
 327   return sha256_finish_ctx (&ctx, resblock);
 328 }
 329
 330 void *
 331 sha224_buffer (const char *buffer, size_t len, void *resblock)
 332 {
 333   struct sha256_ctx ctx;
 334
 335   /* Initialize the computation context.  */
 336   sha224_init_ctx (&ctx);
 337
 338   /* Process whole buffer but last len % 64 bytes.  */
 339   sha256_process_bytes (buffer, len, &ctx);
 340
 341   /* Put result in desired memory area.  */
 342   return sha224_finish_ctx (&ctx, resblock);
 343 }
 344
 345 void
 346 sha256_process_bytes (const void *buffer, size_t len, struct sha256_ctx *ctx)
 347 {
 348   /* When we already have some bits in our internal buffer concatenate
 349      both inputs first.  */
 350   if (ctx->buflen != 0)
 351     {
 352       size_t left_over = ctx->buflen;
 353       size_t add = 128 - left_over > len ? len : 128 - left_over;
 354
 355       memcpy (&((char *) ctx->buffer)[left_over], buffer, add);
 356       ctx->buflen += add;
 357
 358       if (ctx->buflen > 64)
 359         {
 360           sha256_process_block (ctx->buffer, ctx->buflen & ~63, ctx);
 361
 362           ctx->buflen &= 63;
 363           /* The regions in the following copy operation cannot overlap.  */
 364           memcpy (ctx->buffer,
 365                   &((char *) ctx->buffer)[(left_over + add) & ~63],
 366                   ctx->buflen);
 367         }
 368
 369       buffer = (const char *) buffer + add;
 370       len -= add;
 371     }
 372
 373   /* Process available complete blocks.  */
 374   if (len >= 64)
 375     {
 376 #if !_STRING_ARCH_unaligned
 377 # define UNALIGNED_P(p) ((uintptr_t) (p) % alignof (uint32_t) != 0)
 378       if (UNALIGNED_P (buffer))
 379         while (len > 64)
 380           {
 381             sha256_process_block (memcpy (ctx->buffer, buffer, 64), 64, ctx);
 382             buffer = (const char *) buffer + 64;
 383             len -= 64;
 384           }
 385       else
 386 #endif
 387         {
 388           sha256_process_block (buffer, len & ~63, ctx);
 389           buffer = (const char *) buffer + (len & ~63);
 390           len &= 63;
 391         }
 392     }
 393
 394   /* Move remaining bytes in internal buffer.  */
 395   if (len > 0)
 396     {
 397       size_t left_over = ctx->buflen;
 398
 399       memcpy (&((char *) ctx->buffer)[left_over], buffer, len);
 400       left_over += len;
 401       if (left_over >= 64)
 402         {
 403           sha256_process_block (ctx->buffer, 64, ctx);
 404           left_over -= 64;
 405           memcpy (ctx->buffer, &ctx->buffer[16], left_over);
 406         }
 407       ctx->buflen = left_over;
 408     }
 409 }
 410
 411 /* --- Code below is the primary difference between sha1.c and sha256.c --- */
 412
 413 /* SHA256 round constants */
 414 #define K(I) sha256_round_constants[I]
 415 static const uint32_t sha256_round_constants[64] = {
 416   0x428a2f98UL, 0x71374491UL, 0xb5c0fbcfUL, 0xe9b5dba5UL,
 417   0x3956c25bUL, 0x59f111f1UL, 0x923f82a4UL, 0xab1c5ed5UL,
 418   0xd807aa98UL, 0x12835b01UL, 0x243185beUL, 0x550c7dc3UL,
 419   0x72be5d74UL, 0x80deb1feUL, 0x9bdc06a7UL, 0xc19bf174UL,
 420   0xe49b69c1UL, 0xefbe4786UL, 0x0fc19dc6UL, 0x240ca1ccUL,
 421   0x2de92c6fUL, 0x4a7484aaUL, 0x5cb0a9dcUL, 0x76f988daUL,
 422   0x983e5152UL, 0xa831c66dUL, 0xb00327c8UL, 0xbf597fc7UL,
 423   0xc6e00bf3UL, 0xd5a79147UL, 0x06ca6351UL, 0x14292967UL,
 424   0x27b70a85UL, 0x2e1b2138UL, 0x4d2c6dfcUL, 0x53380d13UL,
 425   0x650a7354UL, 0x766a0abbUL, 0x81c2c92eUL, 0x92722c85UL,
 426   0xa2bfe8a1UL, 0xa81a664bUL, 0xc24b8b70UL, 0xc76c51a3UL,
 427   0xd192e819UL, 0xd6990624UL, 0xf40e3585UL, 0x106aa070UL,
 428   0x19a4c116UL, 0x1e376c08UL, 0x2748774cUL, 0x34b0bcb5UL,
 429   0x391c0cb3UL, 0x4ed8aa4aUL, 0x5b9cca4fUL, 0x682e6ff3UL,
 430   0x748f82eeUL, 0x78a5636fUL, 0x84c87814UL, 0x8cc70208UL,
 431   0x90befffaUL, 0xa4506cebUL, 0xbef9a3f7UL, 0xc67178f2UL,
 432 };
 433
 434 /* Round functions.  */
 435 #define F2(A,B,C) ( ( A & B ) | ( C & ( A | B ) ) )
 436 #define F1(E,F,G) ( G ^ ( E & ( F ^ G ) ) )
 437
 438 /* Process LEN bytes of BUFFER, accumulating context into CTX.
 439    It is assumed that LEN % 64 == 0.
 440    Most of this code comes from GnuPG's cipher/sha1.c.  */
 441
 442 void
 443 sha256_process_block (const void *buffer, size_t len, struct sha256_ctx *ctx)
 444 {
 445   const uint32_t *words = buffer;
 446   size_t nwords = len / sizeof (uint32_t);
 447   const uint32_t *endp = words + nwords;
 448   uint32_t x[16];
 449   uint32_t a = ctx->state[0];
 450   uint32_t b = ctx->state[1];
 451   uint32_t c = ctx->state[2];
 452   uint32_t d = ctx->state[3];
 453   uint32_t e = ctx->state[4];
 454   uint32_t f = ctx->state[5];
 455   uint32_t g = ctx->state[6];
 456   uint32_t h = ctx->state[7];
 457   uint32_t lolen = len;
 458
 459   /* First increment the byte count.  FIPS PUB 180-2 specifies the possible
 460      length of the file up to 2^64 bits.  Here we only compute the
 461      number of bytes.  Do a double word increment.  */
 462   ctx->total[0] += lolen;
 463   ctx->total[1] += (len >> 31 >> 1) + (ctx->total[0] < lolen);
 464
 465 #define rol(x, n) (((x) << (n)) | ((x) >> (32 - (n))))
 466 #define S0(x) (rol(x,25)^rol(x,14)^(x>>3))
 467 #define S1(x) (rol(x,15)^rol(x,13)^(x>>10))
 468 #define SS0(x) (rol(x,30)^rol(x,19)^rol(x,10))
 469 #define SS1(x) (rol(x,26)^rol(x,21)^rol(x,7))
 470
 471 #define M(I) ( tm =   S1(x[(I-2)&0x0f]) + x[(I-7)&0x0f] \
 472                     + S0(x[(I-15)&0x0f]) + x[I&0x0f]    \
 473                , x[I&0x0f] = tm )
 474
 475 #define R(A,B,C,D,E,F,G,H,K,M)  do { t0 = SS0(A) + F2(A,B,C); \
 476                                      t1 = H + SS1(E)  \
 477                                       + F1(E,F,G)     \
 478                                       + K             \
 479                                       + M;            \
 480                                      D += t1;  H = t0 + t1; \
 481                                } while(0)
 482
 483   while (words < endp)
 484     {
 485       uint32_t tm;
 486       uint32_t t0, t1;
 487       int t;
 488       /* FIXME: see sha1.c for a better implementation.  */
 489       for (t = 0; t < 16; t++)
 490         {
 491           x[t] = SWAP (*words);
 492           words++;
 493         }
 494
 495       R( a, b, c, d, e, f, g, h, K( 0), x[ 0] );
 496       R( h, a, b, c, d, e, f, g, K( 1), x[ 1] );
 497       R( g, h, a, b, c, d, e, f, K( 2), x[ 2] );
 498       R( f, g, h, a, b, c, d, e, K( 3), x[ 3] );
 499       R( e, f, g, h, a, b, c, d, K( 4), x[ 4] );
 500       R( d, e, f, g, h, a, b, c, K( 5), x[ 5] );
 501       R( c, d, e, f, g, h, a, b, K( 6), x[ 6] );
 502       R( b, c, d, e, f, g, h, a, K( 7), x[ 7] );
 503       R( a, b, c, d, e, f, g, h, K( 8), x[ 8] );
 504       R( h, a, b, c, d, e, f, g, K( 9), x[ 9] );
 505       R( g, h, a, b, c, d, e, f, K(10), x[10] );
 506       R( f, g, h, a, b, c, d, e, K(11), x[11] );
 507       R( e, f, g, h, a, b, c, d, K(12), x[12] );
 508       R( d, e, f, g, h, a, b, c, K(13), x[13] );
 509       R( c, d, e, f, g, h, a, b, K(14), x[14] );
 510       R( b, c, d, e, f, g, h, a, K(15), x[15] );
 511       R( a, b, c, d, e, f, g, h, K(16), M(16) );
 512       R( h, a, b, c, d, e, f, g, K(17), M(17) );
 513       R( g, h, a, b, c, d, e, f, K(18), M(18) );
 514       R( f, g, h, a, b, c, d, e, K(19), M(19) );
 515       R( e, f, g, h, a, b, c, d, K(20), M(20) );
 516       R( d, e, f, g, h, a, b, c, K(21), M(21) );
 517       R( c, d, e, f, g, h, a, b, K(22), M(22) );
 518       R( b, c, d, e, f, g, h, a, K(23), M(23) );
 519       R( a, b, c, d, e, f, g, h, K(24), M(24) );
 520       R( h, a, b, c, d, e, f, g, K(25), M(25) );
 521       R( g, h, a, b, c, d, e, f, K(26), M(26) );
 522       R( f, g, h, a, b, c, d, e, K(27), M(27) );
 523       R( e, f, g, h, a, b, c, d, K(28), M(28) );
 524       R( d, e, f, g, h, a, b, c, K(29), M(29) );
 525       R( c, d, e, f, g, h, a, b, K(30), M(30) );
 526       R( b, c, d, e, f, g, h, a, K(31), M(31) );
 527       R( a, b, c, d, e, f, g, h, K(32), M(32) );
 528       R( h, a, b, c, d, e, f, g, K(33), M(33) );
 529       R( g, h, a, b, c, d, e, f, K(34), M(34) );
 530       R( f, g, h, a, b, c, d, e, K(35), M(35) );
 531       R( e, f, g, h, a, b, c, d, K(36), M(36) );
 532       R( d, e, f, g, h, a, b, c, K(37), M(37) );
 533       R( c, d, e, f, g, h, a, b, K(38), M(38) );
 534       R( b, c, d, e, f, g, h, a, K(39), M(39) );
 535       R( a, b, c, d, e, f, g, h, K(40), M(40) );
 536       R( h, a, b, c, d, e, f, g, K(41), M(41) );
 537       R( g, h, a, b, c, d, e, f, K(42), M(42) );
 538       R( f, g, h, a, b, c, d, e, K(43), M(43) );
 539       R( e, f, g, h, a, b, c, d, K(44), M(44) );
 540       R( d, e, f, g, h, a, b, c, K(45), M(45) );
 541       R( c, d, e, f, g, h, a, b, K(46), M(46) );
 542       R( b, c, d, e, f, g, h, a, K(47), M(47) );
 543       R( a, b, c, d, e, f, g, h, K(48), M(48) );
 544       R( h, a, b, c, d, e, f, g, K(49), M(49) );
 545       R( g, h, a, b, c, d, e, f, K(50), M(50) );
 546       R( f, g, h, a, b, c, d, e, K(51), M(51) );
 547       R( e, f, g, h, a, b, c, d, K(52), M(52) );
 548       R( d, e, f, g, h, a, b, c, K(53), M(53) );
 549       R( c, d, e, f, g, h, a, b, K(54), M(54) );
 550       R( b, c, d, e, f, g, h, a, K(55), M(55) );
 551       R( a, b, c, d, e, f, g, h, K(56), M(56) );
 552       R( h, a, b, c, d, e, f, g, K(57), M(57) );
 553       R( g, h, a, b, c, d, e, f, K(58), M(58) );
 554       R( f, g, h, a, b, c, d, e, K(59), M(59) );
 555       R( e, f, g, h, a, b, c, d, K(60), M(60) );
 556       R( d, e, f, g, h, a, b, c, K(61), M(61) );
 557       R( c, d, e, f, g, h, a, b, K(62), M(62) );
 558       R( b, c, d, e, f, g, h, a, K(63), M(63) );
 559
 560       a = ctx->state[0] += a;
 561       b = ctx->state[1] += b;
 562       c = ctx->state[2] += c;
 563       d = ctx->state[3] += d;
 564       e = ctx->state[4] += e;
 565       f = ctx->state[5] += f;
 566       g = ctx->state[6] += g;
 567       h = ctx->state[7] += h;
 568     }
 569 }