1 /* $FreeBSD$ */ 2 /* $KAME: rijndael-alg-fst.c,v 1.7 2001/05/27 00:23:23 itojun Exp $ */ 3 4 /* 5 * rijndael-alg-fst.c v2.3 April '2000 6 * 7 * Optimised ANSI C code 8 * 9 * authors: v1.0: Antoon Bosselaers 10 * v2.0: Vincent Rijmen 11 * v2.3: Paulo Barreto 12 * 13 * This code is placed in the public domain. 14 */ 15 16 #include <sys/cdefs.h> 17 #include <sys/types.h> 18 #ifdef _KERNEL 19 #include <sys/systm.h> 20 #else 21 #include <string.h> 22 #endif 23 #include <crypto/rijndael/rijndael-alg-fst.h> 24 #include <crypto/rijndael/rijndael_local.h> 25 26 #include <crypto/rijndael/boxes-fst.dat> 27 28 int rijndaelKeySched(word8 k[MAXKC][4], word8 W[MAXROUNDS+1][4][4], int ROUNDS) { 29 /* Calculate the necessary round keys 30 * The number of calculations depends on keyBits and blockBits 31 */ 32 int j, r, t, rconpointer = 0; 33 union { 34 word8 x8[MAXKC][4]; 35 word32 x32[MAXKC]; 36 } xtk; 37 #define tk xtk.x8 38 int KC = ROUNDS - 6; 39 40 for (j = KC-1; j >= 0; j--) { 41 *((word32*)tk[j]) = *((word32*)k[j]); 42 } 43 r = 0; 44 t = 0; 45 /* copy values into round key array */ 46 for (j = 0; (j < KC) && (r < ROUNDS + 1); ) { 47 for (; (j < KC) && (t < 4); j++, t++) { 48 *((word32*)W[r][t]) = *((word32*)tk[j]); 49 } 50 if (t == 4) { 51 r++; 52 t = 0; 53 } 54 } 55 56 while (r < ROUNDS + 1) { /* while not enough round key material calculated */ 57 /* calculate new values */ 58 tk[0][0] ^= S[tk[KC-1][1]]; 59 tk[0][1] ^= S[tk[KC-1][2]]; 60 tk[0][2] ^= S[tk[KC-1][3]]; 61 tk[0][3] ^= S[tk[KC-1][0]]; 62 tk[0][0] ^= rcon[rconpointer++]; 63 64 if (KC != 8) { 65 for (j = 1; j < KC; j++) { 66 *((word32*)tk[j]) ^= *((word32*)tk[j-1]); 67 } 68 } else { 69 for (j = 1; j < KC/2; j++) { 70 *((word32*)tk[j]) ^= *((word32*)tk[j-1]); 71 } 72 tk[KC/2][0] ^= S[tk[KC/2 - 1][0]]; 73 tk[KC/2][1] ^= S[tk[KC/2 - 1][1]]; 74 tk[KC/2][2] ^= S[tk[KC/2 - 1][2]]; 75 tk[KC/2][3] ^= S[tk[KC/2 - 1][3]]; 76 for (j = KC/2 + 1; j < KC; j++) { 77 *((word32*)tk[j]) ^= *((word32*)tk[j-1]); 78 } 79 } 80 /* copy values into round key array */ 81 for (j = 0; (j < KC) && (r < ROUNDS + 1); ) { 82 for (; (j < KC) && (t < 4); j++, t++) { 83 *((word32*)W[r][t]) = *((word32*)tk[j]); 84 } 85 if (t == 4) { 86 r++; 87 t = 0; 88 } 89 } 90 } 91 return 0; 92 #undef tk 93 } 94 95 int rijndaelKeyEncToDec(word8 W[MAXROUNDS+1][4][4], int ROUNDS) { 96 int r; 97 word8 *w; 98 99 for (r = 1; r < ROUNDS; r++) { 100 w = W[r][0]; 101 *((word32*)w) = 102 *((const word32*)U1[w[0]]) 103 ^ *((const word32*)U2[w[1]]) 104 ^ *((const word32*)U3[w[2]]) 105 ^ *((const word32*)U4[w[3]]); 106 107 w = W[r][1]; 108 *((word32*)w) = 109 *((const word32*)U1[w[0]]) 110 ^ *((const word32*)U2[w[1]]) 111 ^ *((const word32*)U3[w[2]]) 112 ^ *((const word32*)U4[w[3]]); 113 114 w = W[r][2]; 115 *((word32*)w) = 116 *((const word32*)U1[w[0]]) 117 ^ *((const word32*)U2[w[1]]) 118 ^ *((const word32*)U3[w[2]]) 119 ^ *((const word32*)U4[w[3]]); 120 121 w = W[r][3]; 122 *((word32*)w) = 123 *((const word32*)U1[w[0]]) 124 ^ *((const word32*)U2[w[1]]) 125 ^ *((const word32*)U3[w[2]]) 126 ^ *((const word32*)U4[w[3]]); 127 } 128 return 0; 129 } 130 131 /** 132 * Encrypt a single block. 133 */ 134 int rijndaelEncrypt(word8 in[16], word8 out[16], word8 rk[MAXROUNDS+1][4][4], int ROUNDS) { 135 int r; 136 union { 137 word8 x8[16]; 138 word32 x32[4]; 139 } xa, xb; 140 #define a xa.x8 141 #define b xb.x8 142 union { 143 word8 x8[4][4]; 144 word32 x32[4]; 145 } xtemp; 146 #define temp xtemp.x8 147 148 memcpy(a, in, sizeof a); 149 150 *((word32*)temp[0]) = *((word32*)(a )) ^ *((word32*)rk[0][0]); 151 *((word32*)temp[1]) = *((word32*)(a+ 4)) ^ *((word32*)rk[0][1]); 152 *((word32*)temp[2]) = *((word32*)(a+ 8)) ^ *((word32*)rk[0][2]); 153 *((word32*)temp[3]) = *((word32*)(a+12)) ^ *((word32*)rk[0][3]); 154 *((word32*)(b )) = *((const word32*)T1[temp[0][0]]) 155 ^ *((const word32*)T2[temp[1][1]]) 156 ^ *((const word32*)T3[temp[2][2]]) 157 ^ *((const word32*)T4[temp[3][3]]); 158 *((word32*)(b + 4)) = *((const word32*)T1[temp[1][0]]) 159 ^ *((const word32*)T2[temp[2][1]]) 160 ^ *((const word32*)T3[temp[3][2]]) 161 ^ *((const word32*)T4[temp[0][3]]); 162 *((word32*)(b + 8)) = *((const word32*)T1[temp[2][0]]) 163 ^ *((const word32*)T2[temp[3][1]]) 164 ^ *((const word32*)T3[temp[0][2]]) 165 ^ *((const word32*)T4[temp[1][3]]); 166 *((word32*)(b +12)) = *((const word32*)T1[temp[3][0]]) 167 ^ *((const word32*)T2[temp[0][1]]) 168 ^ *((const word32*)T3[temp[1][2]]) 169 ^ *((const word32*)T4[temp[2][3]]); 170 for (r = 1; r < ROUNDS-1; r++) { 171 *((word32*)temp[0]) = *((word32*)(b )) ^ *((word32*)rk[r][0]); 172 *((word32*)temp[1]) = *((word32*)(b+ 4)) ^ *((word32*)rk[r][1]); 173 *((word32*)temp[2]) = *((word32*)(b+ 8)) ^ *((word32*)rk[r][2]); 174 *((word32*)temp[3]) = *((word32*)(b+12)) ^ *((word32*)rk[r][3]); 175 176 *((word32*)(b )) = *((const word32*)T1[temp[0][0]]) 177 ^ *((const word32*)T2[temp[1][1]]) 178 ^ *((const word32*)T3[temp[2][2]]) 179 ^ *((const word32*)T4[temp[3][3]]); 180 *((word32*)(b + 4)) = *((const word32*)T1[temp[1][0]]) 181 ^ *((const word32*)T2[temp[2][1]]) 182 ^ *((const word32*)T3[temp[3][2]]) 183 ^ *((const word32*)T4[temp[0][3]]); 184 *((word32*)(b + 8)) = *((const word32*)T1[temp[2][0]]) 185 ^ *((const word32*)T2[temp[3][1]]) 186 ^ *((const word32*)T3[temp[0][2]]) 187 ^ *((const word32*)T4[temp[1][3]]); 188 *((word32*)(b +12)) = *((const word32*)T1[temp[3][0]]) 189 ^ *((const word32*)T2[temp[0][1]]) 190 ^ *((const word32*)T3[temp[1][2]]) 191 ^ *((const word32*)T4[temp[2][3]]); 192 } 193 /* last round is special */ 194 *((word32*)temp[0]) = *((word32*)(b )) ^ *((word32*)rk[ROUNDS-1][0]); 195 *((word32*)temp[1]) = *((word32*)(b+ 4)) ^ *((word32*)rk[ROUNDS-1][1]); 196 *((word32*)temp[2]) = *((word32*)(b+ 8)) ^ *((word32*)rk[ROUNDS-1][2]); 197 *((word32*)temp[3]) = *((word32*)(b+12)) ^ *((word32*)rk[ROUNDS-1][3]); 198 b[ 0] = T1[temp[0][0]][1]; 199 b[ 1] = T1[temp[1][1]][1]; 200 b[ 2] = T1[temp[2][2]][1]; 201 b[ 3] = T1[temp[3][3]][1]; 202 b[ 4] = T1[temp[1][0]][1]; 203 b[ 5] = T1[temp[2][1]][1]; 204 b[ 6] = T1[temp[3][2]][1]; 205 b[ 7] = T1[temp[0][3]][1]; 206 b[ 8] = T1[temp[2][0]][1]; 207 b[ 9] = T1[temp[3][1]][1]; 208 b[10] = T1[temp[0][2]][1]; 209 b[11] = T1[temp[1][3]][1]; 210 b[12] = T1[temp[3][0]][1]; 211 b[13] = T1[temp[0][1]][1]; 212 b[14] = T1[temp[1][2]][1]; 213 b[15] = T1[temp[2][3]][1]; 214 *((word32*)(b )) ^= *((word32*)rk[ROUNDS][0]); 215 *((word32*)(b+ 4)) ^= *((word32*)rk[ROUNDS][1]); 216 *((word32*)(b+ 8)) ^= *((word32*)rk[ROUNDS][2]); 217 *((word32*)(b+12)) ^= *((word32*)rk[ROUNDS][3]); 218 219 memcpy(out, b, sizeof b /* XXX out */); 220 221 return 0; 222 #undef a 223 #undef b 224 #undef temp 225 } 226 227 #ifdef INTERMEDIATE_VALUE_KAT 228 /** 229 * Encrypt only a certain number of rounds. 230 * Only used in the Intermediate Value Known Answer Test. 231 */ 232 int rijndaelEncryptRound(word8 a[4][4], word8 rk[MAXROUNDS+1][4][4], int ROUNDS, int rounds) { 233 int r; 234 word8 temp[4][4]; 235 236 /* make number of rounds sane */ 237 if (rounds > ROUNDS) { 238 rounds = ROUNDS; 239 } 240 241 *((word32*)a[0]) = *((word32*)a[0]) ^ *((word32*)rk[0][0]); 242 *((word32*)a[1]) = *((word32*)a[1]) ^ *((word32*)rk[0][1]); 243 *((word32*)a[2]) = *((word32*)a[2]) ^ *((word32*)rk[0][2]); 244 *((word32*)a[3]) = *((word32*)a[3]) ^ *((word32*)rk[0][3]); 245 246 for (r = 1; (r <= rounds) && (r < ROUNDS); r++) { 247 *((word32*)temp[0]) = *((word32*)T1[a[0][0]]) 248 ^ *((word32*)T2[a[1][1]]) 249 ^ *((word32*)T3[a[2][2]]) 250 ^ *((word32*)T4[a[3][3]]); 251 *((word32*)temp[1]) = *((word32*)T1[a[1][0]]) 252 ^ *((word32*)T2[a[2][1]]) 253 ^ *((word32*)T3[a[3][2]]) 254 ^ *((word32*)T4[a[0][3]]); 255 *((word32*)temp[2]) = *((word32*)T1[a[2][0]]) 256 ^ *((word32*)T2[a[3][1]]) 257 ^ *((word32*)T3[a[0][2]]) 258 ^ *((word32*)T4[a[1][3]]); 259 *((word32*)temp[3]) = *((word32*)T1[a[3][0]]) 260 ^ *((word32*)T2[a[0][1]]) 261 ^ *((word32*)T3[a[1][2]]) 262 ^ *((word32*)T4[a[2][3]]); 263 *((word32*)a[0]) = *((word32*)temp[0]) ^ *((word32*)rk[r][0]); 264 *((word32*)a[1]) = *((word32*)temp[1]) ^ *((word32*)rk[r][1]); 265 *((word32*)a[2]) = *((word32*)temp[2]) ^ *((word32*)rk[r][2]); 266 *((word32*)a[3]) = *((word32*)temp[3]) ^ *((word32*)rk[r][3]); 267 } 268 if (rounds == ROUNDS) { 269 /* last round is special */ 270 temp[0][0] = T1[a[0][0]][1]; 271 temp[0][1] = T1[a[1][1]][1]; 272 temp[0][2] = T1[a[2][2]][1]; 273 temp[0][3] = T1[a[3][3]][1]; 274 temp[1][0] = T1[a[1][0]][1]; 275 temp[1][1] = T1[a[2][1]][1]; 276 temp[1][2] = T1[a[3][2]][1]; 277 temp[1][3] = T1[a[0][3]][1]; 278 temp[2][0] = T1[a[2][0]][1]; 279 temp[2][1] = T1[a[3][1]][1]; 280 temp[2][2] = T1[a[0][2]][1]; 281 temp[2][3] = T1[a[1][3]][1]; 282 temp[3][0] = T1[a[3][0]][1]; 283 temp[3][1] = T1[a[0][1]][1]; 284 temp[3][2] = T1[a[1][2]][1]; 285 temp[3][3] = T1[a[2][3]][1]; 286 *((word32*)a[0]) = *((word32*)temp[0]) ^ *((word32*)rk[ROUNDS][0]); 287 *((word32*)a[1]) = *((word32*)temp[1]) ^ *((word32*)rk[ROUNDS][1]); 288 *((word32*)a[2]) = *((word32*)temp[2]) ^ *((word32*)rk[ROUNDS][2]); 289 *((word32*)a[3]) = *((word32*)temp[3]) ^ *((word32*)rk[ROUNDS][3]); 290 } 291 292 return 0; 293 } 294 #endif /* INTERMEDIATE_VALUE_KAT */ 295 296 /** 297 * Decrypt a single block. 298 */ 299 int rijndaelDecrypt(word8 in[16], word8 out[16], word8 rk[MAXROUNDS+1][4][4], int ROUNDS) { 300 int r; 301 union { 302 word8 x8[16]; 303 word32 x32[4]; 304 } xa, xb; 305 #define a xa.x8 306 #define b xb.x8 307 union { 308 word8 x8[4][4]; 309 word32 x32[4]; 310 } xtemp; 311 #define temp xtemp.x8 312 313 memcpy(a, in, sizeof a); 314 315 *((word32*)temp[0]) = *((word32*)(a )) ^ *((word32*)rk[ROUNDS][0]); 316 *((word32*)temp[1]) = *((word32*)(a+ 4)) ^ *((word32*)rk[ROUNDS][1]); 317 *((word32*)temp[2]) = *((word32*)(a+ 8)) ^ *((word32*)rk[ROUNDS][2]); 318 *((word32*)temp[3]) = *((word32*)(a+12)) ^ *((word32*)rk[ROUNDS][3]); 319 320 *((word32*)(b )) = *((const word32*)T5[temp[0][0]]) 321 ^ *((const word32*)T6[temp[3][1]]) 322 ^ *((const word32*)T7[temp[2][2]]) 323 ^ *((const word32*)T8[temp[1][3]]); 324 *((word32*)(b+ 4)) = *((const word32*)T5[temp[1][0]]) 325 ^ *((const word32*)T6[temp[0][1]]) 326 ^ *((const word32*)T7[temp[3][2]]) 327 ^ *((const word32*)T8[temp[2][3]]); 328 *((word32*)(b+ 8)) = *((const word32*)T5[temp[2][0]]) 329 ^ *((const word32*)T6[temp[1][1]]) 330 ^ *((const word32*)T7[temp[0][2]]) 331 ^ *((const word32*)T8[temp[3][3]]); 332 *((word32*)(b+12)) = *((const word32*)T5[temp[3][0]]) 333 ^ *((const word32*)T6[temp[2][1]]) 334 ^ *((const word32*)T7[temp[1][2]]) 335 ^ *((const word32*)T8[temp[0][3]]); 336 for (r = ROUNDS-1; r > 1; r--) { 337 *((word32*)temp[0]) = *((word32*)(b )) ^ *((word32*)rk[r][0]); 338 *((word32*)temp[1]) = *((word32*)(b+ 4)) ^ *((word32*)rk[r][1]); 339 *((word32*)temp[2]) = *((word32*)(b+ 8)) ^ *((word32*)rk[r][2]); 340 *((word32*)temp[3]) = *((word32*)(b+12)) ^ *((word32*)rk[r][3]); 341 *((word32*)(b )) = *((const word32*)T5[temp[0][0]]) 342 ^ *((const word32*)T6[temp[3][1]]) 343 ^ *((const word32*)T7[temp[2][2]]) 344 ^ *((const word32*)T8[temp[1][3]]); 345 *((word32*)(b+ 4)) = *((const word32*)T5[temp[1][0]]) 346 ^ *((const word32*)T6[temp[0][1]]) 347 ^ *((const word32*)T7[temp[3][2]]) 348 ^ *((const word32*)T8[temp[2][3]]); 349 *((word32*)(b+ 8)) = *((const word32*)T5[temp[2][0]]) 350 ^ *((const word32*)T6[temp[1][1]]) 351 ^ *((const word32*)T7[temp[0][2]]) 352 ^ *((const word32*)T8[temp[3][3]]); 353 *((word32*)(b+12)) = *((const word32*)T5[temp[3][0]]) 354 ^ *((const word32*)T6[temp[2][1]]) 355 ^ *((const word32*)T7[temp[1][2]]) 356 ^ *((const word32*)T8[temp[0][3]]); 357 } 358 /* last round is special */ 359 *((word32*)temp[0]) = *((word32*)(b )) ^ *((word32*)rk[1][0]); 360 *((word32*)temp[1]) = *((word32*)(b+ 4)) ^ *((word32*)rk[1][1]); 361 *((word32*)temp[2]) = *((word32*)(b+ 8)) ^ *((word32*)rk[1][2]); 362 *((word32*)temp[3]) = *((word32*)(b+12)) ^ *((word32*)rk[1][3]); 363 b[ 0] = S5[temp[0][0]]; 364 b[ 1] = S5[temp[3][1]]; 365 b[ 2] = S5[temp[2][2]]; 366 b[ 3] = S5[temp[1][3]]; 367 b[ 4] = S5[temp[1][0]]; 368 b[ 5] = S5[temp[0][1]]; 369 b[ 6] = S5[temp[3][2]]; 370 b[ 7] = S5[temp[2][3]]; 371 b[ 8] = S5[temp[2][0]]; 372 b[ 9] = S5[temp[1][1]]; 373 b[10] = S5[temp[0][2]]; 374 b[11] = S5[temp[3][3]]; 375 b[12] = S5[temp[3][0]]; 376 b[13] = S5[temp[2][1]]; 377 b[14] = S5[temp[1][2]]; 378 b[15] = S5[temp[0][3]]; 379 *((word32*)(b )) ^= *((word32*)rk[0][0]); 380 *((word32*)(b+ 4)) ^= *((word32*)rk[0][1]); 381 *((word32*)(b+ 8)) ^= *((word32*)rk[0][2]); 382 *((word32*)(b+12)) ^= *((word32*)rk[0][3]); 383 384 memcpy(out, b, sizeof b /* XXX out */); 385 386 return 0; 387 #undef a 388 #undef b 389 #undef temp 390 } 391 392 393 #ifdef INTERMEDIATE_VALUE_KAT 394 /** 395 * Decrypt only a certain number of rounds. 396 * Only used in the Intermediate Value Known Answer Test. 397 * Operations rearranged such that the intermediate values 398 * of decryption correspond with the intermediate values 399 * of encryption. 400 */ 401 int rijndaelDecryptRound(word8 a[4][4], word8 rk[MAXROUNDS+1][4][4], int ROUNDS, int rounds) { 402 int r, i; 403 word8 temp[4], shift; 404 405 /* make number of rounds sane */ 406 if (rounds > ROUNDS) { 407 rounds = ROUNDS; 408 } 409 /* first round is special: */ 410 *(word32 *)a[0] ^= *(word32 *)rk[ROUNDS][0]; 411 *(word32 *)a[1] ^= *(word32 *)rk[ROUNDS][1]; 412 *(word32 *)a[2] ^= *(word32 *)rk[ROUNDS][2]; 413 *(word32 *)a[3] ^= *(word32 *)rk[ROUNDS][3]; 414 for (i = 0; i < 4; i++) { 415 a[i][0] = Si[a[i][0]]; 416 a[i][1] = Si[a[i][1]]; 417 a[i][2] = Si[a[i][2]]; 418 a[i][3] = Si[a[i][3]]; 419 } 420 for (i = 1; i < 4; i++) { 421 shift = (4 - i) & 3; 422 temp[0] = a[(0 + shift) & 3][i]; 423 temp[1] = a[(1 + shift) & 3][i]; 424 temp[2] = a[(2 + shift) & 3][i]; 425 temp[3] = a[(3 + shift) & 3][i]; 426 a[0][i] = temp[0]; 427 a[1][i] = temp[1]; 428 a[2][i] = temp[2]; 429 a[3][i] = temp[3]; 430 } 431 /* ROUNDS-1 ordinary rounds */ 432 for (r = ROUNDS-1; r > rounds; r--) { 433 *(word32 *)a[0] ^= *(word32 *)rk[r][0]; 434 *(word32 *)a[1] ^= *(word32 *)rk[r][1]; 435 *(word32 *)a[2] ^= *(word32 *)rk[r][2]; 436 *(word32 *)a[3] ^= *(word32 *)rk[r][3]; 437 438 *((word32*)a[0]) = 439 *((word32*)U1[a[0][0]]) 440 ^ *((word32*)U2[a[0][1]]) 441 ^ *((word32*)U3[a[0][2]]) 442 ^ *((word32*)U4[a[0][3]]); 443 444 *((word32*)a[1]) = 445 *((word32*)U1[a[1][0]]) 446 ^ *((word32*)U2[a[1][1]]) 447 ^ *((word32*)U3[a[1][2]]) 448 ^ *((word32*)U4[a[1][3]]); 449 450 *((word32*)a[2]) = 451 *((word32*)U1[a[2][0]]) 452 ^ *((word32*)U2[a[2][1]]) 453 ^ *((word32*)U3[a[2][2]]) 454 ^ *((word32*)U4[a[2][3]]); 455 456 *((word32*)a[3]) = 457 *((word32*)U1[a[3][0]]) 458 ^ *((word32*)U2[a[3][1]]) 459 ^ *((word32*)U3[a[3][2]]) 460 ^ *((word32*)U4[a[3][3]]); 461 for (i = 0; i < 4; i++) { 462 a[i][0] = Si[a[i][0]]; 463 a[i][1] = Si[a[i][1]]; 464 a[i][2] = Si[a[i][2]]; 465 a[i][3] = Si[a[i][3]]; 466 } 467 for (i = 1; i < 4; i++) { 468 shift = (4 - i) & 3; 469 temp[0] = a[(0 + shift) & 3][i]; 470 temp[1] = a[(1 + shift) & 3][i]; 471 temp[2] = a[(2 + shift) & 3][i]; 472 temp[3] = a[(3 + shift) & 3][i]; 473 a[0][i] = temp[0]; 474 a[1][i] = temp[1]; 475 a[2][i] = temp[2]; 476 a[3][i] = temp[3]; 477 } 478 } 479 if (rounds == 0) { 480 /* End with the extra key addition */ 481 *(word32 *)a[0] ^= *(word32 *)rk[0][0]; 482 *(word32 *)a[1] ^= *(word32 *)rk[0][1]; 483 *(word32 *)a[2] ^= *(word32 *)rk[0][2]; 484 *(word32 *)a[3] ^= *(word32 *)rk[0][3]; 485 } 486 return 0; 487 } 488 #endif /* INTERMEDIATE_VALUE_KAT */ 489