1 /* 2 * Copyright (c) 2016 Thomas Pornin <pornin@bolet.org> 3 * 4 * Modified for OpenBSD by Thomas Pornin and Mike Belopuhov. 5 * 6 * Permission is hereby granted, free of charge, to any person obtaining 7 * a copy of this software and associated documentation files (the 8 * "Software"), to deal in the Software without restriction, including 9 * without limitation the rights to use, copy, modify, merge, publish, 10 * distribute, sublicense, and/or sell copies of the Software, and to 11 * permit persons to whom the Software is furnished to do so, subject to 12 * the following conditions: 13 * 14 * The above copyright notice and this permission notice shall be 15 * included in all copies or substantial portions of the Software. 16 * 17 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, 18 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF 19 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND 20 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS 21 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN 22 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN 23 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE 24 * SOFTWARE. 25 */ 26 27 #include <sys/types.h> 28 #include <sys/systm.h> 29 #include <sys/stdint.h> 30 31 #include "aes.h" 32 33 static inline void 34 enc32le(void *dst, uint32_t x) 35 { 36 unsigned char *buf = dst; 37 38 buf[0] = (unsigned char)x; 39 buf[1] = (unsigned char)(x >> 8); 40 buf[2] = (unsigned char)(x >> 16); 41 buf[3] = (unsigned char)(x >> 24); 42 } 43 44 static inline uint32_t 45 dec32le(const void *src) 46 { 47 const unsigned char *buf = src; 48 49 return (uint32_t)buf[0] 50 | ((uint32_t)buf[1] << 8) 51 | ((uint32_t)buf[2] << 16) 52 | ((uint32_t)buf[3] << 24); 53 } 54 55 /* 56 * This constant-time implementation is "bitsliced": the 128-bit state is 57 * split over eight 32-bit words q* in the following way: 58 * 59 * -- Input block consists in 16 bytes: 60 * a00 a10 a20 a30 a01 a11 a21 a31 a02 a12 a22 a32 a03 a13 a23 a33 61 * In the terminology of FIPS 197, this is a 4x4 matrix which is read 62 * column by column. 63 * 64 * -- Each byte is split into eight bits which are distributed over the 65 * eight words, at the same rank. Thus, for a byte x at rank k, bit 0 66 * (least significant) of x will be at rank k in q0 (if that bit is b, 67 * then it contributes "b << k" to the value of q0), bit 1 of x will be 68 * at rank k in q1, and so on. 69 * 70 * -- Ranks given to bits are in "row order" and are either all even, or 71 * all odd. Two independent AES states are thus interleaved, one using 72 * the even ranks, the other the odd ranks. Row order means: 73 * a00 a01 a02 a03 a10 a11 a12 a13 a20 a21 a22 a23 a30 a31 a32 a33 74 * 75 * Converting input bytes from two AES blocks to bitslice representation 76 * is done in the following way: 77 * -- Decode first block into the four words q0 q2 q4 q6, in that order, 78 * using little-endian convention. 79 * -- Decode second block into the four words q1 q3 q5 q7, in that order, 80 * using little-endian convention. 81 * -- Call aes_ct_ortho(). 82 * 83 * Converting back to bytes is done by using the reverse operations. Note 84 * that aes_ct_ortho() is its own inverse. 85 */ 86 87 /* 88 * The AES S-box, as a bitsliced constant-time version. The input array 89 * consists in eight 32-bit words; 32 S-box instances are computed in 90 * parallel. Bits 0 to 7 of each S-box input (bit 0 is least significant) 91 * are spread over the words 0 to 7, at the same rank. 92 */ 93 static void 94 aes_ct_bitslice_Sbox(uint32_t *q) 95 { 96 /* 97 * This S-box implementation is a straightforward translation of 98 * the circuit described by Boyar and Peralta in "A new 99 * combinational logic minimization technique with applications 100 * to cryptology" (https://eprint.iacr.org/2009/191.pdf). 101 * 102 * Note that variables x* (input) and s* (output) are numbered 103 * in "reverse" order (x0 is the high bit, x7 is the low bit). 104 */ 105 106 uint32_t x0, x1, x2, x3, x4, x5, x6, x7; 107 uint32_t y1, y2, y3, y4, y5, y6, y7, y8, y9; 108 uint32_t y10, y11, y12, y13, y14, y15, y16, y17, y18, y19; 109 uint32_t y20, y21; 110 uint32_t z0, z1, z2, z3, z4, z5, z6, z7, z8, z9; 111 uint32_t z10, z11, z12, z13, z14, z15, z16, z17; 112 uint32_t t0, t1, t2, t3, t4, t5, t6, t7, t8, t9; 113 uint32_t t10, t11, t12, t13, t14, t15, t16, t17, t18, t19; 114 uint32_t t20, t21, t22, t23, t24, t25, t26, t27, t28, t29; 115 uint32_t t30, t31, t32, t33, t34, t35, t36, t37, t38, t39; 116 uint32_t t40, t41, t42, t43, t44, t45, t46, t47, t48, t49; 117 uint32_t t50, t51, t52, t53, t54, t55, t56, t57, t58, t59; 118 uint32_t t60, t61, t62, t63, t64, t65, t66, t67; 119 uint32_t s0, s1, s2, s3, s4, s5, s6, s7; 120 121 x0 = q[7]; 122 x1 = q[6]; 123 x2 = q[5]; 124 x3 = q[4]; 125 x4 = q[3]; 126 x5 = q[2]; 127 x6 = q[1]; 128 x7 = q[0]; 129 130 /* 131 * Top linear transformation. 132 */ 133 y14 = x3 ^ x5; 134 y13 = x0 ^ x6; 135 y9 = x0 ^ x3; 136 y8 = x0 ^ x5; 137 t0 = x1 ^ x2; 138 y1 = t0 ^ x7; 139 y4 = y1 ^ x3; 140 y12 = y13 ^ y14; 141 y2 = y1 ^ x0; 142 y5 = y1 ^ x6; 143 y3 = y5 ^ y8; 144 t1 = x4 ^ y12; 145 y15 = t1 ^ x5; 146 y20 = t1 ^ x1; 147 y6 = y15 ^ x7; 148 y10 = y15 ^ t0; 149 y11 = y20 ^ y9; 150 y7 = x7 ^ y11; 151 y17 = y10 ^ y11; 152 y19 = y10 ^ y8; 153 y16 = t0 ^ y11; 154 y21 = y13 ^ y16; 155 y18 = x0 ^ y16; 156 157 /* 158 * Non-linear section. 159 */ 160 t2 = y12 & y15; 161 t3 = y3 & y6; 162 t4 = t3 ^ t2; 163 t5 = y4 & x7; 164 t6 = t5 ^ t2; 165 t7 = y13 & y16; 166 t8 = y5 & y1; 167 t9 = t8 ^ t7; 168 t10 = y2 & y7; 169 t11 = t10 ^ t7; 170 t12 = y9 & y11; 171 t13 = y14 & y17; 172 t14 = t13 ^ t12; 173 t15 = y8 & y10; 174 t16 = t15 ^ t12; 175 t17 = t4 ^ t14; 176 t18 = t6 ^ t16; 177 t19 = t9 ^ t14; 178 t20 = t11 ^ t16; 179 t21 = t17 ^ y20; 180 t22 = t18 ^ y19; 181 t23 = t19 ^ y21; 182 t24 = t20 ^ y18; 183 184 t25 = t21 ^ t22; 185 t26 = t21 & t23; 186 t27 = t24 ^ t26; 187 t28 = t25 & t27; 188 t29 = t28 ^ t22; 189 t30 = t23 ^ t24; 190 t31 = t22 ^ t26; 191 t32 = t31 & t30; 192 t33 = t32 ^ t24; 193 t34 = t23 ^ t33; 194 t35 = t27 ^ t33; 195 t36 = t24 & t35; 196 t37 = t36 ^ t34; 197 t38 = t27 ^ t36; 198 t39 = t29 & t38; 199 t40 = t25 ^ t39; 200 201 t41 = t40 ^ t37; 202 t42 = t29 ^ t33; 203 t43 = t29 ^ t40; 204 t44 = t33 ^ t37; 205 t45 = t42 ^ t41; 206 z0 = t44 & y15; 207 z1 = t37 & y6; 208 z2 = t33 & x7; 209 z3 = t43 & y16; 210 z4 = t40 & y1; 211 z5 = t29 & y7; 212 z6 = t42 & y11; 213 z7 = t45 & y17; 214 z8 = t41 & y10; 215 z9 = t44 & y12; 216 z10 = t37 & y3; 217 z11 = t33 & y4; 218 z12 = t43 & y13; 219 z13 = t40 & y5; 220 z14 = t29 & y2; 221 z15 = t42 & y9; 222 z16 = t45 & y14; 223 z17 = t41 & y8; 224 225 /* 226 * Bottom linear transformation. 227 */ 228 t46 = z15 ^ z16; 229 t47 = z10 ^ z11; 230 t48 = z5 ^ z13; 231 t49 = z9 ^ z10; 232 t50 = z2 ^ z12; 233 t51 = z2 ^ z5; 234 t52 = z7 ^ z8; 235 t53 = z0 ^ z3; 236 t54 = z6 ^ z7; 237 t55 = z16 ^ z17; 238 t56 = z12 ^ t48; 239 t57 = t50 ^ t53; 240 t58 = z4 ^ t46; 241 t59 = z3 ^ t54; 242 t60 = t46 ^ t57; 243 t61 = z14 ^ t57; 244 t62 = t52 ^ t58; 245 t63 = t49 ^ t58; 246 t64 = z4 ^ t59; 247 t65 = t61 ^ t62; 248 t66 = z1 ^ t63; 249 s0 = t59 ^ t63; 250 s6 = t56 ^ ~t62; 251 s7 = t48 ^ ~t60; 252 t67 = t64 ^ t65; 253 s3 = t53 ^ t66; 254 s4 = t51 ^ t66; 255 s5 = t47 ^ t65; 256 s1 = t64 ^ ~s3; 257 s2 = t55 ^ ~t67; 258 259 q[7] = s0; 260 q[6] = s1; 261 q[5] = s2; 262 q[4] = s3; 263 q[3] = s4; 264 q[2] = s5; 265 q[1] = s6; 266 q[0] = s7; 267 } 268 269 /* 270 * Perform bytewise orthogonalization of eight 32-bit words. Bytes 271 * of q0..q7 are spread over all words: for a byte x that occurs 272 * at rank i in q[j] (byte x uses bits 8*i to 8*i+7 in q[j]), the bit 273 * of rank k in x (0 <= k <= 7) goes to q[k] at rank 8*i+j. 274 * 275 * This operation is an involution. 276 */ 277 static void 278 aes_ct_ortho(uint32_t *q) 279 { 280 #define SWAPN(cl, ch, s, x, y) do { \ 281 uint32_t a, b; \ 282 a = (x); \ 283 b = (y); \ 284 (x) = (a & (uint32_t)cl) | ((b & (uint32_t)cl) << (s)); \ 285 (y) = ((a & (uint32_t)ch) >> (s)) | (b & (uint32_t)ch); \ 286 } while (0) 287 288 #define SWAP2(x, y) SWAPN(0x55555555, 0xAAAAAAAA, 1, x, y) 289 #define SWAP4(x, y) SWAPN(0x33333333, 0xCCCCCCCC, 2, x, y) 290 #define SWAP8(x, y) SWAPN(0x0F0F0F0F, 0xF0F0F0F0, 4, x, y) 291 292 SWAP2(q[0], q[1]); 293 SWAP2(q[2], q[3]); 294 SWAP2(q[4], q[5]); 295 SWAP2(q[6], q[7]); 296 297 SWAP4(q[0], q[2]); 298 SWAP4(q[1], q[3]); 299 SWAP4(q[4], q[6]); 300 SWAP4(q[5], q[7]); 301 302 SWAP8(q[0], q[4]); 303 SWAP8(q[1], q[5]); 304 SWAP8(q[2], q[6]); 305 SWAP8(q[3], q[7]); 306 } 307 308 static inline uint32_t 309 sub_word(uint32_t x) 310 { 311 uint32_t q[8]; 312 int i; 313 314 for (i = 0; i < 8; i ++) { 315 q[i] = x; 316 } 317 aes_ct_ortho(q); 318 aes_ct_bitslice_Sbox(q); 319 aes_ct_ortho(q); 320 return q[0]; 321 } 322 323 static const unsigned char Rcon[] = { 324 0x01, 0x02, 0x04, 0x08, 0x10, 0x20, 0x40, 0x80, 0x1B, 0x36 325 }; 326 327 /* 328 * Base key schedule code. The function sub_word() must be defined 329 * below. Subkeys are produced in little-endian convention (but not 330 * bitsliced). Key length is expressed in bytes. 331 */ 332 static unsigned 333 aes_keysched_base(uint32_t *skey, const void *key, size_t key_len) 334 { 335 unsigned num_rounds; 336 int i, j, k, nk, nkf; 337 uint32_t tmp; 338 339 switch (key_len) { 340 case 16: 341 num_rounds = 10; 342 break; 343 case 24: 344 num_rounds = 12; 345 break; 346 case 32: 347 num_rounds = 14; 348 break; 349 default: 350 return 0; 351 } 352 nk = (int)(key_len >> 2); 353 nkf = (int)((num_rounds + 1) << 2); 354 for (i = 0; i < nk; i ++) { 355 tmp = dec32le((const unsigned char *)key + (i << 2)); 356 skey[i] = tmp; 357 } 358 tmp = skey[(key_len >> 2) - 1]; 359 for (i = nk, j = 0, k = 0; i < nkf; i ++) { 360 if (j == 0) { 361 tmp = (tmp << 24) | (tmp >> 8); 362 tmp = sub_word(tmp) ^ Rcon[k]; 363 } else if (nk > 6 && j == 4) { 364 tmp = sub_word(tmp); 365 } 366 tmp ^= skey[i - nk]; 367 skey[i] = tmp; 368 if (++ j == nk) { 369 j = 0; 370 k ++; 371 } 372 } 373 return num_rounds; 374 } 375 376 /* 377 * AES key schedule, constant-time version. skey[] is filled with n+1 378 * 128-bit subkeys, where n is the number of rounds (10 to 14, depending 379 * on key size). The number of rounds is returned. If the key size is 380 * invalid (not 16, 24 or 32), then 0 is returned. 381 */ 382 unsigned 383 aes_ct_keysched(uint32_t *comp_skey, const void *key, size_t key_len) 384 { 385 uint32_t skey[60]; 386 unsigned u, num_rounds; 387 388 num_rounds = aes_keysched_base(skey, key, key_len); 389 for (u = 0; u <= num_rounds; u ++) { 390 uint32_t q[8]; 391 392 q[0] = q[1] = skey[(u << 2) + 0]; 393 q[2] = q[3] = skey[(u << 2) + 1]; 394 q[4] = q[5] = skey[(u << 2) + 2]; 395 q[6] = q[7] = skey[(u << 2) + 3]; 396 aes_ct_ortho(q); 397 comp_skey[(u << 2) + 0] = 398 (q[0] & 0x55555555) | (q[1] & 0xAAAAAAAA); 399 comp_skey[(u << 2) + 1] = 400 (q[2] & 0x55555555) | (q[3] & 0xAAAAAAAA); 401 comp_skey[(u << 2) + 2] = 402 (q[4] & 0x55555555) | (q[5] & 0xAAAAAAAA); 403 comp_skey[(u << 2) + 3] = 404 (q[6] & 0x55555555) | (q[7] & 0xAAAAAAAA); 405 } 406 return num_rounds; 407 } 408 409 /* 410 * Expand AES subkeys as produced by aes_ct_keysched(), into 411 * a larger array suitable for aes_ct_bitslice_encrypt() and 412 * aes_ct_bitslice_decrypt(). 413 */ 414 void 415 aes_ct_skey_expand(uint32_t *skey, 416 unsigned num_rounds, const uint32_t *comp_skey) 417 { 418 unsigned u, v, n; 419 420 n = (num_rounds + 1) << 2; 421 for (u = 0, v = 0; u < n; u ++, v += 2) { 422 uint32_t x, y; 423 424 x = y = comp_skey[u]; 425 x &= 0x55555555; 426 skey[v + 0] = x | (x << 1); 427 y &= 0xAAAAAAAA; 428 skey[v + 1] = y | (y >> 1); 429 } 430 } 431 432 static inline void 433 add_round_key(uint32_t *q, const uint32_t *sk) 434 { 435 q[0] ^= sk[0]; 436 q[1] ^= sk[1]; 437 q[2] ^= sk[2]; 438 q[3] ^= sk[3]; 439 q[4] ^= sk[4]; 440 q[5] ^= sk[5]; 441 q[6] ^= sk[6]; 442 q[7] ^= sk[7]; 443 } 444 445 static inline void 446 shift_rows(uint32_t *q) 447 { 448 int i; 449 450 for (i = 0; i < 8; i ++) { 451 uint32_t x; 452 453 x = q[i]; 454 q[i] = (x & 0x000000FF) 455 | ((x & 0x0000FC00) >> 2) | ((x & 0x00000300) << 6) 456 | ((x & 0x00F00000) >> 4) | ((x & 0x000F0000) << 4) 457 | ((x & 0xC0000000) >> 6) | ((x & 0x3F000000) << 2); 458 } 459 } 460 461 static inline uint32_t 462 rotr16(uint32_t x) 463 { 464 return (x << 16) | (x >> 16); 465 } 466 467 static inline void 468 mix_columns(uint32_t *q) 469 { 470 uint32_t q0, q1, q2, q3, q4, q5, q6, q7; 471 uint32_t r0, r1, r2, r3, r4, r5, r6, r7; 472 473 q0 = q[0]; 474 q1 = q[1]; 475 q2 = q[2]; 476 q3 = q[3]; 477 q4 = q[4]; 478 q5 = q[5]; 479 q6 = q[6]; 480 q7 = q[7]; 481 r0 = (q0 >> 8) | (q0 << 24); 482 r1 = (q1 >> 8) | (q1 << 24); 483 r2 = (q2 >> 8) | (q2 << 24); 484 r3 = (q3 >> 8) | (q3 << 24); 485 r4 = (q4 >> 8) | (q4 << 24); 486 r5 = (q5 >> 8) | (q5 << 24); 487 r6 = (q6 >> 8) | (q6 << 24); 488 r7 = (q7 >> 8) | (q7 << 24); 489 490 q[0] = q7 ^ r7 ^ r0 ^ rotr16(q0 ^ r0); 491 q[1] = q0 ^ r0 ^ q7 ^ r7 ^ r1 ^ rotr16(q1 ^ r1); 492 q[2] = q1 ^ r1 ^ r2 ^ rotr16(q2 ^ r2); 493 q[3] = q2 ^ r2 ^ q7 ^ r7 ^ r3 ^ rotr16(q3 ^ r3); 494 q[4] = q3 ^ r3 ^ q7 ^ r7 ^ r4 ^ rotr16(q4 ^ r4); 495 q[5] = q4 ^ r4 ^ r5 ^ rotr16(q5 ^ r5); 496 q[6] = q5 ^ r5 ^ r6 ^ rotr16(q6 ^ r6); 497 q[7] = q6 ^ r6 ^ r7 ^ rotr16(q7 ^ r7); 498 } 499 500 /* 501 * Compute AES encryption on bitsliced data. Since input is stored on 502 * eight 32-bit words, two block encryptions are actually performed 503 * in parallel. 504 */ 505 void 506 aes_ct_bitslice_encrypt(unsigned num_rounds, 507 const uint32_t *skey, uint32_t *q) 508 { 509 unsigned u; 510 511 add_round_key(q, skey); 512 for (u = 1; u < num_rounds; u ++) { 513 aes_ct_bitslice_Sbox(q); 514 shift_rows(q); 515 mix_columns(q); 516 add_round_key(q, skey + (u << 3)); 517 } 518 aes_ct_bitslice_Sbox(q); 519 shift_rows(q); 520 add_round_key(q, skey + (num_rounds << 3)); 521 } 522 523 /* 524 * Like aes_ct_bitslice_Sbox(), but for the inverse S-box. 525 */ 526 void 527 aes_ct_bitslice_invSbox(uint32_t *q) 528 { 529 /* 530 * AES S-box is: 531 * S(x) = A(I(x)) ^ 0x63 532 * where I() is inversion in GF(256), and A() is a linear 533 * transform (0 is formally defined to be its own inverse). 534 * Since inversion is an involution, the inverse S-box can be 535 * computed from the S-box as: 536 * iS(x) = B(S(B(x ^ 0x63)) ^ 0x63) 537 * where B() is the inverse of A(). Indeed, for any y in GF(256): 538 * iS(S(y)) = B(A(I(B(A(I(y)) ^ 0x63 ^ 0x63))) ^ 0x63 ^ 0x63) = y 539 * 540 * Note: we reuse the implementation of the forward S-box, 541 * instead of duplicating it here, so that total code size is 542 * lower. By merging the B() transforms into the S-box circuit 543 * we could make faster CBC decryption, but CBC decryption is 544 * already quite faster than CBC encryption because we can 545 * process two blocks in parallel. 546 */ 547 uint32_t q0, q1, q2, q3, q4, q5, q6, q7; 548 549 q0 = ~q[0]; 550 q1 = ~q[1]; 551 q2 = q[2]; 552 q3 = q[3]; 553 q4 = q[4]; 554 q5 = ~q[5]; 555 q6 = ~q[6]; 556 q7 = q[7]; 557 q[7] = q1 ^ q4 ^ q6; 558 q[6] = q0 ^ q3 ^ q5; 559 q[5] = q7 ^ q2 ^ q4; 560 q[4] = q6 ^ q1 ^ q3; 561 q[3] = q5 ^ q0 ^ q2; 562 q[2] = q4 ^ q7 ^ q1; 563 q[1] = q3 ^ q6 ^ q0; 564 q[0] = q2 ^ q5 ^ q7; 565 566 aes_ct_bitslice_Sbox(q); 567 568 q0 = ~q[0]; 569 q1 = ~q[1]; 570 q2 = q[2]; 571 q3 = q[3]; 572 q4 = q[4]; 573 q5 = ~q[5]; 574 q6 = ~q[6]; 575 q7 = q[7]; 576 q[7] = q1 ^ q4 ^ q6; 577 q[6] = q0 ^ q3 ^ q5; 578 q[5] = q7 ^ q2 ^ q4; 579 q[4] = q6 ^ q1 ^ q3; 580 q[3] = q5 ^ q0 ^ q2; 581 q[2] = q4 ^ q7 ^ q1; 582 q[1] = q3 ^ q6 ^ q0; 583 q[0] = q2 ^ q5 ^ q7; 584 } 585 586 static inline void 587 inv_shift_rows(uint32_t *q) 588 { 589 int i; 590 591 for (i = 0; i < 8; i ++) { 592 uint32_t x; 593 594 x = q[i]; 595 q[i] = (x & 0x000000FF) 596 | ((x & 0x00003F00) << 2) | ((x & 0x0000C000) >> 6) 597 | ((x & 0x000F0000) << 4) | ((x & 0x00F00000) >> 4) 598 | ((x & 0x03000000) << 6) | ((x & 0xFC000000) >> 2); 599 } 600 } 601 602 static void 603 inv_mix_columns(uint32_t *q) 604 { 605 uint32_t q0, q1, q2, q3, q4, q5, q6, q7; 606 uint32_t r0, r1, r2, r3, r4, r5, r6, r7; 607 608 q0 = q[0]; 609 q1 = q[1]; 610 q2 = q[2]; 611 q3 = q[3]; 612 q4 = q[4]; 613 q5 = q[5]; 614 q6 = q[6]; 615 q7 = q[7]; 616 r0 = (q0 >> 8) | (q0 << 24); 617 r1 = (q1 >> 8) | (q1 << 24); 618 r2 = (q2 >> 8) | (q2 << 24); 619 r3 = (q3 >> 8) | (q3 << 24); 620 r4 = (q4 >> 8) | (q4 << 24); 621 r5 = (q5 >> 8) | (q5 << 24); 622 r6 = (q6 >> 8) | (q6 << 24); 623 r7 = (q7 >> 8) | (q7 << 24); 624 625 q[0] = q5 ^ q6 ^ q7 ^ r0 ^ r5 ^ r7 ^ rotr16(q0 ^ q5 ^ q6 ^ r0 ^ r5); 626 q[1] = q0 ^ q5 ^ r0 ^ r1 ^ r5 ^ r6 ^ r7 ^ rotr16(q1 ^ q5 ^ q7 ^ r1 ^ r5 ^ r6); 627 q[2] = q0 ^ q1 ^ q6 ^ r1 ^ r2 ^ r6 ^ r7 ^ rotr16(q0 ^ q2 ^ q6 ^ r2 ^ r6 ^ r7); 628 q[3] = q0 ^ q1 ^ q2 ^ q5 ^ q6 ^ r0 ^ r2 ^ r3 ^ r5 ^ rotr16(q0 ^ q1 ^ q3 ^ q5 ^ q6 ^ q7 ^ r0 ^ r3 ^ r5 ^ r7); 629 q[4] = q1 ^ q2 ^ q3 ^ q5 ^ r1 ^ r3 ^ r4 ^ r5 ^ r6 ^ r7 ^ rotr16(q1 ^ q2 ^ q4 ^ q5 ^ q7 ^ r1 ^ r4 ^ r5 ^ r6); 630 q[5] = q2 ^ q3 ^ q4 ^ q6 ^ r2 ^ r4 ^ r5 ^ r6 ^ r7 ^ rotr16(q2 ^ q3 ^ q5 ^ q6 ^ r2 ^ r5 ^ r6 ^ r7); 631 q[6] = q3 ^ q4 ^ q5 ^ q7 ^ r3 ^ r5 ^ r6 ^ r7 ^ rotr16(q3 ^ q4 ^ q6 ^ q7 ^ r3 ^ r6 ^ r7); 632 q[7] = q4 ^ q5 ^ q6 ^ r4 ^ r6 ^ r7 ^ rotr16(q4 ^ q5 ^ q7 ^ r4 ^ r7); 633 } 634 635 /* 636 * Compute AES decryption on bitsliced data. Since input is stored on 637 * eight 32-bit words, two block decryptions are actually performed 638 * in parallel. 639 */ 640 void 641 aes_ct_bitslice_decrypt(unsigned num_rounds, 642 const uint32_t *skey, uint32_t *q) 643 { 644 unsigned u; 645 646 add_round_key(q, skey + (num_rounds << 3)); 647 for (u = num_rounds - 1; u > 0; u --) { 648 inv_shift_rows(q); 649 aes_ct_bitslice_invSbox(q); 650 add_round_key(q, skey + (u << 3)); 651 inv_mix_columns(q); 652 } 653 inv_shift_rows(q); 654 aes_ct_bitslice_invSbox(q); 655 add_round_key(q, skey); 656 } 657 658 659 int 660 AES_Setkey(AES_CTX *ctx, const uint8_t *key, int len) 661 { 662 ctx->num_rounds = aes_ct_keysched(ctx->sk, key, len); 663 if (ctx->num_rounds == 0) 664 return -1; 665 aes_ct_skey_expand(ctx->sk_exp, ctx->num_rounds, ctx->sk); 666 return 0; 667 } 668 669 void 670 AES_Encrypt_ECB(AES_CTX *ctx, const uint8_t *src, 671 uint8_t *dst, size_t num_blocks) 672 { 673 while (num_blocks > 0) { 674 uint32_t q[8]; 675 676 q[0] = dec32le(src); 677 q[2] = dec32le(src + 4); 678 q[4] = dec32le(src + 8); 679 q[6] = dec32le(src + 12); 680 if (num_blocks > 1) { 681 q[1] = dec32le(src + 16); 682 q[3] = dec32le(src + 20); 683 q[5] = dec32le(src + 24); 684 q[7] = dec32le(src + 28); 685 } else { 686 q[1] = 0; 687 q[3] = 0; 688 q[5] = 0; 689 q[7] = 0; 690 } 691 aes_ct_ortho(q); 692 aes_ct_bitslice_encrypt(ctx->num_rounds, ctx->sk_exp, q); 693 aes_ct_ortho(q); 694 enc32le(dst, q[0]); 695 enc32le(dst + 4, q[2]); 696 enc32le(dst + 8, q[4]); 697 enc32le(dst + 12, q[6]); 698 if (num_blocks > 1) { 699 enc32le(dst + 16, q[1]); 700 enc32le(dst + 20, q[3]); 701 enc32le(dst + 24, q[5]); 702 enc32le(dst + 28, q[7]); 703 src += 32; 704 dst += 32; 705 num_blocks -= 2; 706 } else { 707 break; 708 } 709 } 710 } 711 712 void 713 AES_Decrypt_ECB(AES_CTX *ctx, const uint8_t *src, 714 uint8_t *dst, size_t num_blocks) 715 { 716 while (num_blocks > 0) { 717 uint32_t q[8]; 718 719 q[0] = dec32le(src); 720 q[2] = dec32le(src + 4); 721 q[4] = dec32le(src + 8); 722 q[6] = dec32le(src + 12); 723 if (num_blocks > 1) { 724 q[1] = dec32le(src + 16); 725 q[3] = dec32le(src + 20); 726 q[5] = dec32le(src + 24); 727 q[7] = dec32le(src + 28); 728 } else { 729 q[1] = 0; 730 q[3] = 0; 731 q[5] = 0; 732 q[7] = 0; 733 } 734 aes_ct_ortho(q); 735 aes_ct_bitslice_decrypt(ctx->num_rounds, ctx->sk_exp, q); 736 aes_ct_ortho(q); 737 enc32le(dst, q[0]); 738 enc32le(dst + 4, q[2]); 739 enc32le(dst + 8, q[4]); 740 enc32le(dst + 12, q[6]); 741 if (num_blocks > 1) { 742 enc32le(dst + 16, q[1]); 743 enc32le(dst + 20, q[3]); 744 enc32le(dst + 24, q[5]); 745 enc32le(dst + 28, q[7]); 746 src += 32; 747 dst += 32; 748 num_blocks -= 2; 749 } else { 750 break; 751 } 752 } 753 } 754 755 void 756 AES_Encrypt(AES_CTX *ctx, const uint8_t *src, uint8_t *dst) 757 { 758 AES_Encrypt_ECB(ctx, src, dst, 1); 759 } 760 761 void 762 AES_Decrypt(AES_CTX *ctx, const uint8_t *src, uint8_t *dst) 763 { 764 AES_Decrypt_ECB(ctx, src, dst, 1); 765 } 766 767 int 768 AES_KeySetup_Encrypt(uint32_t *skey, const uint8_t *key, int len) 769 { 770 unsigned r, u; 771 uint32_t tkey[60]; 772 773 r = aes_keysched_base(tkey, key, len); 774 if (r == 0) { 775 return 0; 776 } 777 for (u = 0; u < ((r + 1) << 2); u ++) { 778 uint32_t w; 779 780 w = tkey[u]; 781 skey[u] = (w << 24) 782 | ((w & 0x0000FF00) << 8) 783 | ((w & 0x00FF0000) >> 8) 784 | (w >> 24); 785 } 786 return r; 787 } 788 789 /* 790 * Reduce value x modulo polynomial x^8+x^4+x^3+x+1. This works as 791 * long as x fits on 12 bits at most. 792 */ 793 static inline uint32_t 794 redgf256(uint32_t x) 795 { 796 uint32_t h; 797 798 h = x >> 8; 799 return (x ^ h ^ (h << 1) ^ (h << 3) ^ (h << 4)) & 0xFF; 800 } 801 802 /* 803 * Multiplication by 0x09 in GF(256). 804 */ 805 static inline uint32_t 806 mul9(uint32_t x) 807 { 808 return redgf256(x ^ (x << 3)); 809 } 810 811 /* 812 * Multiplication by 0x0B in GF(256). 813 */ 814 static inline uint32_t 815 mulb(uint32_t x) 816 { 817 return redgf256(x ^ (x << 1) ^ (x << 3)); 818 } 819 820 /* 821 * Multiplication by 0x0D in GF(256). 822 */ 823 static inline uint32_t 824 muld(uint32_t x) 825 { 826 return redgf256(x ^ (x << 2) ^ (x << 3)); 827 } 828 829 /* 830 * Multiplication by 0x0E in GF(256). 831 */ 832 static inline uint32_t 833 mule(uint32_t x) 834 { 835 return redgf256((x << 1) ^ (x << 2) ^ (x << 3)); 836 } 837 838 int 839 AES_KeySetup_Decrypt(uint32_t *skey, const uint8_t *key, int len) 840 { 841 unsigned r, u; 842 uint32_t tkey[60]; 843 844 /* 845 * Compute encryption subkeys. We get them in big-endian 846 * notation. 847 */ 848 r = AES_KeySetup_Encrypt(tkey, key, len); 849 if (r == 0) { 850 return 0; 851 } 852 853 /* 854 * Copy the subkeys in reverse order. Also, apply InvMixColumns() 855 * on the subkeys (except first and last). 856 */ 857 memcpy(skey + (r << 2), tkey, 4 * sizeof(uint32_t)); 858 memcpy(skey, tkey + (r << 2), 4 * sizeof(uint32_t)); 859 for (u = 4; u < (r << 2); u ++) { 860 uint32_t sk, sk0, sk1, sk2, sk3; 861 uint32_t tk, tk0, tk1, tk2, tk3; 862 863 sk = tkey[u]; 864 sk0 = sk >> 24; 865 sk1 = (sk >> 16) & 0xFF; 866 sk2 = (sk >> 8) & 0xFF; 867 sk3 = sk & 0xFF; 868 tk0 = mule(sk0) ^ mulb(sk1) ^ muld(sk2) ^ mul9(sk3); 869 tk1 = mul9(sk0) ^ mule(sk1) ^ mulb(sk2) ^ muld(sk3); 870 tk2 = muld(sk0) ^ mul9(sk1) ^ mule(sk2) ^ mulb(sk3); 871 tk3 = mulb(sk0) ^ muld(sk1) ^ mul9(sk2) ^ mule(sk3); 872 tk = (tk0 << 24) ^ (tk1 << 16) ^ (tk2 << 8) ^ tk3; 873 skey[((r - (u >> 2)) << 2) + (u & 3)] = tk; 874 } 875 876 return r; 877 } 878