1 /* $OpenBSD: rijndael.c,v 1.7 2001/02/04 15:32:24 stevesk Exp $ */ 2 3 /* This is an independent implementation of the encryption algorithm: */ 4 /* */ 5 /* RIJNDAEL by Joan Daemen and Vincent Rijmen */ 6 /* */ 7 /* which is a candidate algorithm in the Advanced Encryption Standard */ 8 /* programme of the US National Institute of Standards and Technology. */ 9 /* */ 10 /* Copyright in this implementation is held by Dr B R Gladman but I */ 11 /* hereby give permission for its free direct or derivative use subject */ 12 /* to acknowledgment of its origin and compliance with any conditions */ 13 /* that the originators of the algorithm place on its exploitation. */ 14 /* */ 15 /* Dr Brian Gladman (gladman@seven77.demon.co.uk) 14th January 1999 */ 16 17 /* Timing data for Rijndael (rijndael.c) 18 19 Algorithm: rijndael (rijndael.c) 20 21 128 bit key: 22 Key Setup: 305/1389 cycles (encrypt/decrypt) 23 Encrypt: 374 cycles = 68.4 mbits/sec 24 Decrypt: 352 cycles = 72.7 mbits/sec 25 Mean: 363 cycles = 70.5 mbits/sec 26 27 192 bit key: 28 Key Setup: 277/1595 cycles (encrypt/decrypt) 29 Encrypt: 439 cycles = 58.3 mbits/sec 30 Decrypt: 425 cycles = 60.2 mbits/sec 31 Mean: 432 cycles = 59.3 mbits/sec 32 33 256 bit key: 34 Key Setup: 374/1960 cycles (encrypt/decrypt) 35 Encrypt: 502 cycles = 51.0 mbits/sec 36 Decrypt: 498 cycles = 51.4 mbits/sec 37 Mean: 500 cycles = 51.2 mbits/sec 38 39 */ 40 41 #include <sys/types.h> 42 #include "rijndael.h" 43 44 void gen_tabs __P((void)); 45 46 /* 3. Basic macros for speeding up generic operations */ 47 48 /* Circular rotate of 32 bit values */ 49 50 #define rotr(x,n) (((x) >> ((int)(n))) | ((x) << (32 - (int)(n)))) 51 #define rotl(x,n) (((x) << ((int)(n))) | ((x) >> (32 - (int)(n)))) 52 53 /* Invert byte order in a 32 bit variable */ 54 55 #define bswap(x) ((rotl(x, 8) & 0x00ff00ff) | (rotr(x, 8) & 0xff00ff00)) 56 57 /* Extract byte from a 32 bit quantity (little endian notation) */ 58 59 #define byte(x,n) ((u1byte)((x) >> (8 * n))) 60 61 #if BYTE_ORDER != LITTLE_ENDIAN 62 #define BYTE_SWAP 63 #endif 64 65 #ifdef BYTE_SWAP 66 #define io_swap(x) bswap(x) 67 #else 68 #define io_swap(x) (x) 69 #endif 70 71 #define LARGE_TABLES 72 73 u1byte pow_tab[256]; 74 u1byte log_tab[256]; 75 u1byte sbx_tab[256]; 76 u1byte isb_tab[256]; 77 u4byte rco_tab[ 10]; 78 u4byte ft_tab[4][256]; 79 u4byte it_tab[4][256]; 80 81 #ifdef LARGE_TABLES 82 u4byte fl_tab[4][256]; 83 u4byte il_tab[4][256]; 84 #endif 85 86 u4byte tab_gen = 0; 87 88 #define ff_mult(a,b) (a && b ? pow_tab[(log_tab[a] + log_tab[b]) % 255] : 0) 89 90 #define f_rn(bo, bi, n, k) \ 91 bo[n] = ft_tab[0][byte(bi[n],0)] ^ \ 92 ft_tab[1][byte(bi[(n + 1) & 3],1)] ^ \ 93 ft_tab[2][byte(bi[(n + 2) & 3],2)] ^ \ 94 ft_tab[3][byte(bi[(n + 3) & 3],3)] ^ *(k + n) 95 96 #define i_rn(bo, bi, n, k) \ 97 bo[n] = it_tab[0][byte(bi[n],0)] ^ \ 98 it_tab[1][byte(bi[(n + 3) & 3],1)] ^ \ 99 it_tab[2][byte(bi[(n + 2) & 3],2)] ^ \ 100 it_tab[3][byte(bi[(n + 1) & 3],3)] ^ *(k + n) 101 102 #ifdef LARGE_TABLES 103 104 #define ls_box(x) \ 105 ( fl_tab[0][byte(x, 0)] ^ \ 106 fl_tab[1][byte(x, 1)] ^ \ 107 fl_tab[2][byte(x, 2)] ^ \ 108 fl_tab[3][byte(x, 3)] ) 109 110 #define f_rl(bo, bi, n, k) \ 111 bo[n] = fl_tab[0][byte(bi[n],0)] ^ \ 112 fl_tab[1][byte(bi[(n + 1) & 3],1)] ^ \ 113 fl_tab[2][byte(bi[(n + 2) & 3],2)] ^ \ 114 fl_tab[3][byte(bi[(n + 3) & 3],3)] ^ *(k + n) 115 116 #define i_rl(bo, bi, n, k) \ 117 bo[n] = il_tab[0][byte(bi[n],0)] ^ \ 118 il_tab[1][byte(bi[(n + 3) & 3],1)] ^ \ 119 il_tab[2][byte(bi[(n + 2) & 3],2)] ^ \ 120 il_tab[3][byte(bi[(n + 1) & 3],3)] ^ *(k + n) 121 122 #else 123 124 #define ls_box(x) \ 125 ((u4byte)sbx_tab[byte(x, 0)] << 0) ^ \ 126 ((u4byte)sbx_tab[byte(x, 1)] << 8) ^ \ 127 ((u4byte)sbx_tab[byte(x, 2)] << 16) ^ \ 128 ((u4byte)sbx_tab[byte(x, 3)] << 24) 129 130 #define f_rl(bo, bi, n, k) \ 131 bo[n] = (u4byte)sbx_tab[byte(bi[n],0)] ^ \ 132 rotl(((u4byte)sbx_tab[byte(bi[(n + 1) & 3],1)]), 8) ^ \ 133 rotl(((u4byte)sbx_tab[byte(bi[(n + 2) & 3],2)]), 16) ^ \ 134 rotl(((u4byte)sbx_tab[byte(bi[(n + 3) & 3],3)]), 24) ^ *(k + n) 135 136 #define i_rl(bo, bi, n, k) \ 137 bo[n] = (u4byte)isb_tab[byte(bi[n],0)] ^ \ 138 rotl(((u4byte)isb_tab[byte(bi[(n + 3) & 3],1)]), 8) ^ \ 139 rotl(((u4byte)isb_tab[byte(bi[(n + 2) & 3],2)]), 16) ^ \ 140 rotl(((u4byte)isb_tab[byte(bi[(n + 1) & 3],3)]), 24) ^ *(k + n) 141 142 #endif 143 144 void 145 gen_tabs(void) 146 { 147 u4byte i, t; 148 u1byte p, q; 149 150 /* log and power tables for GF(2**8) finite field with */ 151 /* 0x11b as modular polynomial - the simplest prmitive */ 152 /* root is 0x11, used here to generate the tables */ 153 154 for(i = 0,p = 1; i < 256; ++i) { 155 pow_tab[i] = (u1byte)p; log_tab[p] = (u1byte)i; 156 157 p = p ^ (p << 1) ^ (p & 0x80 ? 0x01b : 0); 158 } 159 160 log_tab[1] = 0; p = 1; 161 162 for(i = 0; i < 10; ++i) { 163 rco_tab[i] = p; 164 165 p = (p << 1) ^ (p & 0x80 ? 0x1b : 0); 166 } 167 168 /* note that the affine byte transformation matrix in */ 169 /* rijndael specification is in big endian format with */ 170 /* bit 0 as the most significant bit. In the remainder */ 171 /* of the specification the bits are numbered from the */ 172 /* least significant end of a byte. */ 173 174 for(i = 0; i < 256; ++i) { 175 p = (i ? pow_tab[255 - log_tab[i]] : 0); q = p; 176 q = (q >> 7) | (q << 1); p ^= q; 177 q = (q >> 7) | (q << 1); p ^= q; 178 q = (q >> 7) | (q << 1); p ^= q; 179 q = (q >> 7) | (q << 1); p ^= q ^ 0x63; 180 sbx_tab[i] = (u1byte)p; isb_tab[p] = (u1byte)i; 181 } 182 183 for(i = 0; i < 256; ++i) { 184 p = sbx_tab[i]; 185 186 #ifdef LARGE_TABLES 187 188 t = p; fl_tab[0][i] = t; 189 fl_tab[1][i] = rotl(t, 8); 190 fl_tab[2][i] = rotl(t, 16); 191 fl_tab[3][i] = rotl(t, 24); 192 #endif 193 t = ((u4byte)ff_mult(2, p)) | 194 ((u4byte)p << 8) | 195 ((u4byte)p << 16) | 196 ((u4byte)ff_mult(3, p) << 24); 197 198 ft_tab[0][i] = t; 199 ft_tab[1][i] = rotl(t, 8); 200 ft_tab[2][i] = rotl(t, 16); 201 ft_tab[3][i] = rotl(t, 24); 202 203 p = isb_tab[i]; 204 205 #ifdef LARGE_TABLES 206 207 t = p; il_tab[0][i] = t; 208 il_tab[1][i] = rotl(t, 8); 209 il_tab[2][i] = rotl(t, 16); 210 il_tab[3][i] = rotl(t, 24); 211 #endif 212 t = ((u4byte)ff_mult(14, p)) | 213 ((u4byte)ff_mult( 9, p) << 8) | 214 ((u4byte)ff_mult(13, p) << 16) | 215 ((u4byte)ff_mult(11, p) << 24); 216 217 it_tab[0][i] = t; 218 it_tab[1][i] = rotl(t, 8); 219 it_tab[2][i] = rotl(t, 16); 220 it_tab[3][i] = rotl(t, 24); 221 } 222 223 tab_gen = 1; 224 } 225 226 #define star_x(x) (((x) & 0x7f7f7f7f) << 1) ^ ((((x) & 0x80808080) >> 7) * 0x1b) 227 228 #define imix_col(y,x) \ 229 u = star_x(x); \ 230 v = star_x(u); \ 231 w = star_x(v); \ 232 t = w ^ (x); \ 233 (y) = u ^ v ^ w; \ 234 (y) ^= rotr(u ^ t, 8) ^ \ 235 rotr(v ^ t, 16) ^ \ 236 rotr(t,24) 237 238 /* initialise the key schedule from the user supplied key */ 239 240 #define loop4(i) \ 241 { t = ls_box(rotr(t, 8)) ^ rco_tab[i]; \ 242 t ^= e_key[4 * i]; e_key[4 * i + 4] = t; \ 243 t ^= e_key[4 * i + 1]; e_key[4 * i + 5] = t; \ 244 t ^= e_key[4 * i + 2]; e_key[4 * i + 6] = t; \ 245 t ^= e_key[4 * i + 3]; e_key[4 * i + 7] = t; \ 246 } 247 248 #define loop6(i) \ 249 { t = ls_box(rotr(t, 8)) ^ rco_tab[i]; \ 250 t ^= e_key[6 * i]; e_key[6 * i + 6] = t; \ 251 t ^= e_key[6 * i + 1]; e_key[6 * i + 7] = t; \ 252 t ^= e_key[6 * i + 2]; e_key[6 * i + 8] = t; \ 253 t ^= e_key[6 * i + 3]; e_key[6 * i + 9] = t; \ 254 t ^= e_key[6 * i + 4]; e_key[6 * i + 10] = t; \ 255 t ^= e_key[6 * i + 5]; e_key[6 * i + 11] = t; \ 256 } 257 258 #define loop8(i) \ 259 { t = ls_box(rotr(t, 8)) ^ rco_tab[i]; \ 260 t ^= e_key[8 * i]; e_key[8 * i + 8] = t; \ 261 t ^= e_key[8 * i + 1]; e_key[8 * i + 9] = t; \ 262 t ^= e_key[8 * i + 2]; e_key[8 * i + 10] = t; \ 263 t ^= e_key[8 * i + 3]; e_key[8 * i + 11] = t; \ 264 t = e_key[8 * i + 4] ^ ls_box(t); \ 265 e_key[8 * i + 12] = t; \ 266 t ^= e_key[8 * i + 5]; e_key[8 * i + 13] = t; \ 267 t ^= e_key[8 * i + 6]; e_key[8 * i + 14] = t; \ 268 t ^= e_key[8 * i + 7]; e_key[8 * i + 15] = t; \ 269 } 270 271 rijndael_ctx * 272 rijndael_set_key(rijndael_ctx *ctx, const u4byte *in_key, const u4byte key_len, 273 int encrypt) 274 { 275 u4byte i, t, u, v, w; 276 u4byte *e_key = ctx->e_key; 277 u4byte *d_key = ctx->d_key; 278 279 ctx->decrypt = !encrypt; 280 281 if(!tab_gen) 282 gen_tabs(); 283 284 ctx->k_len = (key_len + 31) / 32; 285 286 e_key[0] = io_swap(in_key[0]); e_key[1] = io_swap(in_key[1]); 287 e_key[2] = io_swap(in_key[2]); e_key[3] = io_swap(in_key[3]); 288 289 switch(ctx->k_len) { 290 case 4: t = e_key[3]; 291 for(i = 0; i < 10; ++i) 292 loop4(i); 293 break; 294 295 case 6: e_key[4] = io_swap(in_key[4]); t = e_key[5] = io_swap(in_key[5]); 296 for(i = 0; i < 8; ++i) 297 loop6(i); 298 break; 299 300 case 8: e_key[4] = io_swap(in_key[4]); e_key[5] = io_swap(in_key[5]); 301 e_key[6] = io_swap(in_key[6]); t = e_key[7] = io_swap(in_key[7]); 302 for(i = 0; i < 7; ++i) 303 loop8(i); 304 break; 305 } 306 307 if (!encrypt) { 308 d_key[0] = e_key[0]; d_key[1] = e_key[1]; 309 d_key[2] = e_key[2]; d_key[3] = e_key[3]; 310 311 for(i = 4; i < 4 * ctx->k_len + 24; ++i) { 312 imix_col(d_key[i], e_key[i]); 313 } 314 } 315 316 return ctx; 317 } 318 319 /* encrypt a block of text */ 320 321 #define f_nround(bo, bi, k) \ 322 f_rn(bo, bi, 0, k); \ 323 f_rn(bo, bi, 1, k); \ 324 f_rn(bo, bi, 2, k); \ 325 f_rn(bo, bi, 3, k); \ 326 k += 4 327 328 #define f_lround(bo, bi, k) \ 329 f_rl(bo, bi, 0, k); \ 330 f_rl(bo, bi, 1, k); \ 331 f_rl(bo, bi, 2, k); \ 332 f_rl(bo, bi, 3, k) 333 334 void 335 rijndael_encrypt(rijndael_ctx *ctx, const u4byte *in_blk, u4byte *out_blk) 336 { 337 u4byte k_len = ctx->k_len; 338 u4byte *e_key = ctx->e_key; 339 u4byte b0[4], b1[4], *kp; 340 341 b0[0] = io_swap(in_blk[0]) ^ e_key[0]; 342 b0[1] = io_swap(in_blk[1]) ^ e_key[1]; 343 b0[2] = io_swap(in_blk[2]) ^ e_key[2]; 344 b0[3] = io_swap(in_blk[3]) ^ e_key[3]; 345 346 kp = e_key + 4; 347 348 if(k_len > 6) { 349 f_nround(b1, b0, kp); f_nround(b0, b1, kp); 350 } 351 352 if(k_len > 4) { 353 f_nround(b1, b0, kp); f_nround(b0, b1, kp); 354 } 355 356 f_nround(b1, b0, kp); f_nround(b0, b1, kp); 357 f_nround(b1, b0, kp); f_nround(b0, b1, kp); 358 f_nround(b1, b0, kp); f_nround(b0, b1, kp); 359 f_nround(b1, b0, kp); f_nround(b0, b1, kp); 360 f_nround(b1, b0, kp); f_lround(b0, b1, kp); 361 362 out_blk[0] = io_swap(b0[0]); out_blk[1] = io_swap(b0[1]); 363 out_blk[2] = io_swap(b0[2]); out_blk[3] = io_swap(b0[3]); 364 } 365 366 /* decrypt a block of text */ 367 368 #define i_nround(bo, bi, k) \ 369 i_rn(bo, bi, 0, k); \ 370 i_rn(bo, bi, 1, k); \ 371 i_rn(bo, bi, 2, k); \ 372 i_rn(bo, bi, 3, k); \ 373 k -= 4 374 375 #define i_lround(bo, bi, k) \ 376 i_rl(bo, bi, 0, k); \ 377 i_rl(bo, bi, 1, k); \ 378 i_rl(bo, bi, 2, k); \ 379 i_rl(bo, bi, 3, k) 380 381 void 382 rijndael_decrypt(rijndael_ctx *ctx, const u4byte *in_blk, u4byte *out_blk) 383 { 384 u4byte b0[4], b1[4], *kp; 385 u4byte k_len = ctx->k_len; 386 u4byte *e_key = ctx->e_key; 387 u4byte *d_key = ctx->d_key; 388 389 b0[0] = io_swap(in_blk[0]) ^ e_key[4 * k_len + 24]; 390 b0[1] = io_swap(in_blk[1]) ^ e_key[4 * k_len + 25]; 391 b0[2] = io_swap(in_blk[2]) ^ e_key[4 * k_len + 26]; 392 b0[3] = io_swap(in_blk[3]) ^ e_key[4 * k_len + 27]; 393 394 kp = d_key + 4 * (k_len + 5); 395 396 if(k_len > 6) { 397 i_nround(b1, b0, kp); i_nround(b0, b1, kp); 398 } 399 400 if(k_len > 4) { 401 i_nround(b1, b0, kp); i_nround(b0, b1, kp); 402 } 403 404 i_nround(b1, b0, kp); i_nround(b0, b1, kp); 405 i_nround(b1, b0, kp); i_nround(b0, b1, kp); 406 i_nround(b1, b0, kp); i_nround(b0, b1, kp); 407 i_nround(b1, b0, kp); i_nround(b0, b1, kp); 408 i_nround(b1, b0, kp); i_lround(b0, b1, kp); 409 410 out_blk[0] = io_swap(b0[0]); out_blk[1] = io_swap(b0[1]); 411 out_blk[2] = io_swap(b0[2]); out_blk[3] = io_swap(b0[3]); 412 } 413