1 /* md5sum.c - Calculate RFC 1321 md5 hash and sha1 hash. 2 * 3 * Copyright 2012 Rob Landley <rob (at) landley.net> 4 * 5 * See http://refspecs.linuxfoundation.org/LSB_4.1.0/LSB-Core-generic/LSB-Core-generic/md5sum.html 6 * and http://www.ietf.org/rfc/rfc1321.txt 7 * 8 * They're combined this way to share infrastructure, and because md5sum is 9 * and LSB standard command (but sha1sum and newer hashes are a good idea, 10 * see http://valerieaurora.org/hash.html). 11 * 12 * We optionally use openssl (or equivalent) to access assembly optimized 13 * versions of these functions, but provide a built-in version to reduce 14 * required dependencies. 15 * 16 * coreutils supports --status but not -s, busybox supports -s but not --status 17 18 USE_MD5SUM(NEWTOY(md5sum, "bc(check)s(status)[!bc]", TOYFLAG_USR|TOYFLAG_BIN)) 19 USE_SHA1SUM(NEWTOY(sha1sum, "bc(check)s(status)[!bc]", TOYFLAG_USR|TOYFLAG_BIN)) 20 USE_SHA224SUM(OLDTOY(sha224sum, sha1sum, TOYFLAG_USR|TOYFLAG_BIN)) 21 USE_SHA256SUM(OLDTOY(sha256sum, sha1sum, TOYFLAG_USR|TOYFLAG_BIN)) 22 USE_SHA384SUM(OLDTOY(sha384sum, sha1sum, TOYFLAG_USR|TOYFLAG_BIN)) 23 USE_SHA512SUM(OLDTOY(sha512sum, sha1sum, TOYFLAG_USR|TOYFLAG_BIN)) 24 25 config MD5SUM 26 bool "md5sum" 27 default y 28 help 29 usage: md5sum [-bcs] [FILE]... 30 31 Calculate md5 hash for each input file, reading from stdin if none. 32 Output one hash (32 hex digits) for each input file, followed by filename. 33 34 -b Brief (hash only, no filename) 35 -c Check each line of each FILE is the same hash+filename we'd output 36 -s No output, exit status 0 if all hashes match, 1 otherwise 37 38 config SHA1SUM 39 bool "sha1sum" 40 default y 41 help 42 usage: sha?sum [-bcs] [FILE]... 43 44 Calculate sha hash for each input file, reading from stdin if none. Output 45 one hash (40 hex digits for sha1, 56 for sha224, 64 for sha256, 96 for sha384, 46 and 128 for sha512) for each input file, followed by filename. 47 48 -b Brief (hash only, no filename) 49 -c Check each line of each FILE is the same hash+filename we'd output 50 -s No output, exit status 0 if all hashes match, 1 otherwise 51 52 config SHA224SUM 53 bool "sha224sum" 54 default y 55 depends on TOYBOX_LIBCRYPTO 56 help 57 See sha1sum 58 59 config SHA256SUM 60 bool "sha256sum" 61 default y 62 depends on TOYBOX_LIBCRYPTO 63 help 64 See sha1sum 65 66 config SHA384SUM 67 bool "sha384sum" 68 default y 69 depends on TOYBOX_LIBCRYPTO 70 help 71 See sha1sum 72 73 config SHA512SUM 74 bool "sha512sum" 75 default y 76 depends on TOYBOX_LIBCRYPTO 77 help 78 See sha1sum 79 */ 80 81 #define FORCE_FLAGS 82 #define FOR_md5sum 83 #include "toys.h" 84 85 #if CFG_TOYBOX_LIBCRYPTO 86 #include <openssl/md5.h> 87 #include <openssl/sha.h> 88 #else 89 typedef int SHA512_CTX; 90 #endif 91 92 GLOBALS( 93 int sawline; 94 95 // Crypto variables blanked after summing 96 unsigned state[5]; 97 unsigned oldstate[5]; 98 uint64_t count; 99 union { 100 char c[64]; 101 unsigned i[16]; 102 } buffer; 103 ) 104 105 #define rol(value, bits) (((value) << (bits)) | ((value) >> (32 - (bits)))) 106 107 // for(i=0; i<64; i++) md5table[i] = abs(sin(i+1))*(1<<32); But calculating 108 // that involves not just floating point but pulling in -lm (and arguing with 109 // C about whether 1<<32 is a valid thing to do on 32 bit platforms) so: 110 111 static uint32_t md5table[64] = { 112 0xd76aa478, 0xe8c7b756, 0x242070db, 0xc1bdceee, 0xf57c0faf, 0x4787c62a, 113 0xa8304613, 0xfd469501, 0x698098d8, 0x8b44f7af, 0xffff5bb1, 0x895cd7be, 114 0x6b901122, 0xfd987193, 0xa679438e, 0x49b40821, 0xf61e2562, 0xc040b340, 115 0x265e5a51, 0xe9b6c7aa, 0xd62f105d, 0x02441453, 0xd8a1e681, 0xe7d3fbc8, 116 0x21e1cde6, 0xc33707d6, 0xf4d50d87, 0x455a14ed, 0xa9e3e905, 0xfcefa3f8, 117 0x676f02d9, 0x8d2a4c8a, 0xfffa3942, 0x8771f681, 0x6d9d6122, 0xfde5380c, 118 0xa4beea44, 0x4bdecfa9, 0xf6bb4b60, 0xbebfbc70, 0x289b7ec6, 0xeaa127fa, 119 0xd4ef3085, 0x04881d05, 0xd9d4d039, 0xe6db99e5, 0x1fa27cf8, 0xc4ac5665, 120 0xf4292244, 0x432aff97, 0xab9423a7, 0xfc93a039, 0x655b59c3, 0x8f0ccc92, 121 0xffeff47d, 0x85845dd1, 0x6fa87e4f, 0xfe2ce6e0, 0xa3014314, 0x4e0811a1, 122 0xf7537e82, 0xbd3af235, 0x2ad7d2bb, 0xeb86d391 123 }; 124 125 static const uint8_t md5rot[64] = { 126 7, 12, 17, 22, 7, 12, 17, 22, 7, 12, 17, 22, 7, 12, 17, 22, 127 5, 9, 14, 20, 5, 9, 14, 20, 5, 9, 14, 20, 5, 9, 14, 20, 128 4, 11, 16, 23, 4, 11, 16, 23, 4, 11, 16, 23, 4, 11, 16, 23, 129 6, 10, 15, 21, 6, 10, 15, 21, 6, 10, 15, 21, 6, 10, 15, 21 130 }; 131 132 // Mix next 64 bytes of data into md5 hash 133 134 static void md5_transform(void) 135 { 136 unsigned x[4], *b = TT.buffer.i; 137 int i; 138 139 memcpy(x, TT.state, sizeof(x)); 140 141 for (i=0; i<64; i++) { 142 unsigned int in, temp, swap; 143 if (i<16) { 144 in = i; 145 temp = x[1]; 146 temp = (temp & x[2]) | ((~temp) & x[3]); 147 } else if (i<32) { 148 in = (1+(5*i))&15; 149 temp = x[3]; 150 temp = (x[1] & temp) | (x[2] & ~temp); 151 } else if (i<48) { 152 in = (3*i+5)&15; 153 temp = x[1] ^ x[2] ^ x[3]; 154 } else { 155 in = (7*i)&15; 156 temp = x[2] ^ (x[1] | ~x[3]); 157 } 158 temp += x[0] + b[in] + md5table[i]; 159 swap = x[3]; 160 x[3] = x[2]; 161 x[2] = x[1]; 162 x[1] += rol(temp, md5rot[i]); 163 x[0] = swap; 164 } 165 for (i=0; i<4; i++) TT.state[i] += x[i]; 166 } 167 168 // Mix next 64 bytes of data into sha1 hash. 169 170 static const unsigned rconsts[]={0x5A827999,0x6ED9EBA1,0x8F1BBCDC,0xCA62C1D6}; 171 172 static void sha1_transform(void) 173 { 174 int i, j, k, count; 175 unsigned *block = TT.buffer.i; 176 unsigned *rot[5], *temp; 177 178 // Copy context->state[] to working vars 179 for (i=0; i<5; i++) { 180 TT.oldstate[i] = TT.state[i]; 181 rot[i] = TT.state + i; 182 } 183 // 4 rounds of 20 operations each. 184 for (i=count=0; i<4; i++) { 185 for (j=0; j<20; j++) { 186 unsigned work; 187 188 work = *rot[2] ^ *rot[3]; 189 if (!i) work = (work & *rot[1]) ^ *rot[3]; 190 else { 191 if (i==2) work = ((*rot[1]|*rot[2])&*rot[3])|(*rot[1]&*rot[2]); 192 else work ^= *rot[1]; 193 } 194 195 if (!i && j<16) 196 work += block[count] = (rol(block[count],24)&0xFF00FF00) 197 | (rol(block[count],8)&0x00FF00FF); 198 else 199 work += block[count&15] = rol(block[(count+13)&15] 200 ^ block[(count+8)&15] ^ block[(count+2)&15] ^ block[count&15], 1); 201 *rot[4] += work + rol(*rot[0],5) + rconsts[i]; 202 *rot[1] = rol(*rot[1],30); 203 204 // Rotate by one for next time. 205 temp = rot[4]; 206 for (k=4; k; k--) rot[k] = rot[k-1]; 207 *rot = temp; 208 count++; 209 } 210 } 211 // Add the previous values of state[] 212 for (i=0; i<5; i++) TT.state[i] += TT.oldstate[i]; 213 } 214 215 // Fill the 64-byte working buffer and call transform() when full. 216 217 static void hash_update(char *data, unsigned int len, void (*transform)(void)) 218 { 219 unsigned int i, j; 220 221 j = TT.count & 63; 222 TT.count += len; 223 224 for (;;) { 225 // Grab next chunk of data, return if it's not enough to process a frame 226 i = 64 - j; 227 if (i>len) i = len; 228 memcpy(TT.buffer.c+j, data, i); 229 if (j+i != 64) break; 230 231 // Process a frame 232 if (IS_BIG_ENDIAN) 233 for (j=0; j<16; j++) TT.buffer.i[j] = SWAP_LE32(TT.buffer.i[j]); 234 transform(); 235 j=0; 236 data += i; 237 len -= i; 238 } 239 } 240 241 // Initialize array tersely 242 #define HASH_INIT(name, prefix) { name, (void *)prefix##_Init, \ 243 (void *)prefix##_Update, (void *)prefix##_Final, \ 244 prefix##_DIGEST_LENGTH, } 245 #define SHA1_DIGEST_LENGTH SHA_DIGEST_LENGTH 246 247 // Call the assembly optimized library code when CFG_TOYBOX_LIBCRYPTO 248 static void do_lib_hash(int fd, char *name) 249 { 250 // Largest context 251 SHA512_CTX ctx; 252 struct hash { 253 char *name; 254 int (*init)(void *); 255 int (*update)(void *, void *, size_t); 256 int (*final)(void *, void *); 257 int digest_length; 258 } algorithms[] = { 259 USE_TOYBOX_LIBCRYPTO( 260 USE_MD5SUM(HASH_INIT("md5sum", MD5),) 261 USE_SHA1SUM(HASH_INIT("sha1sum", SHA1),) 262 USE_SHA224SUM(HASH_INIT("sha224sum", SHA224),) 263 USE_SHA256SUM(HASH_INIT("sha256sum", SHA256),) 264 USE_SHA384SUM(HASH_INIT("sha384sum", SHA384),) 265 USE_SHA512SUM(HASH_INIT("sha512sum", SHA512),) 266 ) 267 }, * hash; 268 int i; 269 270 // This should never NOT match, so no need to check 271 for (i = 0; i<ARRAY_LEN(algorithms); i++) 272 if (!strcmp(toys.which->name, algorithms[i].name)) break; 273 hash = algorithms+i; 274 275 hash->init(&ctx); 276 for (;;) { 277 i = read(fd, toybuf, sizeof(toybuf)); 278 if (i<1) break; 279 hash->update(&ctx, toybuf, i); 280 } 281 hash->final(toybuf+128, &ctx); 282 283 for (i = 0; i<hash->digest_length; i++) 284 sprintf(toybuf+2*i, "%02x", toybuf[i+128]); 285 } 286 287 // Callback for loopfiles() 288 289 static void do_builtin_hash(int fd, char *name) 290 { 291 uint64_t count; 292 int i, sha1=toys.which->name[0]=='s'; 293 char buf; 294 void (*transform)(void); 295 296 /* SHA1 initialization constants (md5sum uses first 4) */ 297 TT.state[0] = 0x67452301; 298 TT.state[1] = 0xEFCDAB89; 299 TT.state[2] = 0x98BADCFE; 300 TT.state[3] = 0x10325476; 301 TT.state[4] = 0xC3D2E1F0; 302 TT.count = 0; 303 304 transform = sha1 ? sha1_transform : md5_transform; 305 for (;;) { 306 i = read(fd, toybuf, sizeof(toybuf)); 307 if (i<1) break; 308 hash_update(toybuf, i, transform); 309 } 310 311 count = TT.count << 3; 312 313 // End the message by appending a "1" bit to the data, ending with the 314 // message size (in bits, big endian), and adding enough zero bits in 315 // between to pad to the end of the next 64-byte frame. 316 // 317 // Since our input up to now has been in whole bytes, we can deal with 318 // bytes here too. 319 320 buf = 0x80; 321 do { 322 hash_update(&buf, 1, transform); 323 buf = 0; 324 } while ((TT.count & 63) != 56); 325 count = sha1 ? SWAP_BE64(count) : SWAP_LE64(count); 326 hash_update((void *)&count, 8, transform); 327 328 if (sha1) 329 for (i = 0; i < 20; i++) 330 sprintf(toybuf+2*i, "%02x", 255&(TT.state[i>>2] >> ((3-(i & 3)) * 8))); 331 else for (i=0; i<4; i++) sprintf(toybuf+8*i, "%08x", bswap_32(TT.state[i])); 332 333 // Wipe variables. Cryptographer paranoia. 334 memset(TT.state, 0, sizeof(TT)-((long)TT.state-(long)&TT)); 335 i = strlen(toybuf)+1; 336 memset(toybuf+i, 0, sizeof(toybuf)-i); 337 } 338 339 // Call builtin or lib hash function, then display output if necessary 340 static void do_hash(int fd, char *name) 341 { 342 if (CFG_TOYBOX_LIBCRYPTO) do_lib_hash(fd, name); 343 else do_builtin_hash(fd, name); 344 345 if (name) 346 printf((toys.optflags & FLAG_b) ? "%s\n" : "%s %s\n", toybuf, name); 347 } 348 349 static int do_c_line(char *line) 350 { 351 int space = 0, fail = 0; 352 char *name; 353 354 for (name = line; *name; name++) { 355 if (isspace(*name)) { 356 space++; 357 *name = 0; 358 } else if (space) break; 359 } 360 361 if (!space || !*line || !*name) error_msg("bad line %s", line); 362 else { 363 int fd = !strcmp(name, "-") ? 0 : open(name, O_RDONLY); 364 365 TT.sawline = 1; 366 if (fd==-1) { 367 perror_msg_raw(name); 368 *toybuf = 0; 369 } else do_hash(fd, 0); 370 if (strcasecmp(line, toybuf)) toys.exitval = fail = 1; 371 if (!FLAG(s)) printf("%s: %s\n", name, fail ? "FAILED" : "OK"); 372 if (fd>0) close(fd); 373 } 374 375 return 0; 376 } 377 378 // Used instead of loopfiles_line to report error on files containing no hashes. 379 static void do_c_file(char *name) 380 { 381 FILE *fp = !strcmp(name, "-") ? stdin : fopen(name, "r"); 382 383 if (!fp) { 384 perror_msg_raw(name); 385 return; 386 } 387 388 TT.sawline = 0; 389 390 for (;;) { 391 char *line = 0; 392 ssize_t len; 393 394 if ((len = getline(&line, (void *)&len, fp))<1) break; 395 if (line[len-1]=='\n') line[len-1] = 0; 396 do_c_line(line); 397 free(line); 398 } 399 if (fp!=stdin) fclose(fp); 400 401 if (!TT.sawline) error_msg("%s: no lines", name); 402 } 403 404 void md5sum_main(void) 405 { 406 char **arg; 407 408 if (FLAG(c)) for (arg = toys.optargs; *arg; arg++) do_c_file(*arg); 409 else { 410 if (FLAG(s)) error_exit("-s only with -c"); 411 loopfiles(toys.optargs, do_hash); 412 } 413 } 414 415 void sha1sum_main(void) 416 { 417 md5sum_main(); 418 } 419
