]> git.saurik.com Git - apple/ipsec.git/blob - ipsec-tools/racoon/Crypto/rijndael-alg-fst.c
ipsec-34.0.1.tar.gz
[apple/ipsec.git] / ipsec-tools / racoon / Crypto / rijndael-alg-fst.c
1 /* $KAME: rijndael-alg-fst.c,v 1.9 2001/06/19 15:21:05 itojun Exp $ */
2
3 /*
4 * rijndael-alg-fst.c v2.3 April '2000
5 *
6 * Optimised ANSI C code
7 *
8 * authors: v1.0: Antoon Bosselaers
9 * v2.0: Vincent Rijmen
10 * v2.3: Paulo Barreto
11 *
12 * This code is placed in the public domain.
13 */
14
15 #include <sys/cdefs.h>
16 #include <sys/types.h>
17 #ifdef _KERNEL
18 #include <sys/systm.h>
19 #else
20 #include <string.h>
21 #endif
22 #include <rijndael-alg-fst.h>
23 #include <rijndael_local.h>
24
25 #include "boxes-fst.dat"
26
27 #include <err.h>
28 #define bcopy(a, b, c) memcpy((b), (a), (c))
29 #define bzero(a, b) memset((a), 0, (b))
30 #define panic(a) err(1, (a))
31
32 int rijndaelKeySched(word8 k[MAXKC][4], word8 W[MAXROUNDS+1][4][4], int ROUNDS) {
33 /* Calculate the necessary round keys
34 * The number of calculations depends on keyBits and blockBits
35 */
36 int j, r, t, rconpointer = 0;
37 union {
38 word8 x8[MAXKC][4];
39 word32 x32[MAXKC];
40 } xtk;
41 #define tk xtk.x8
42 int KC = ROUNDS - 6;
43
44 for (j = KC-1; j >= 0; j--) {
45 *((word32*)tk[j]) = *((word32*)k[j]);
46 }
47 r = 0;
48 t = 0;
49 /* copy values into round key array */
50 for (j = 0; (j < KC) && (r < ROUNDS + 1); ) {
51 for (; (j < KC) && (t < 4); j++, t++) {
52 *((word32*)W[r][t]) = *((word32*)tk[j]);
53 }
54 if (t == 4) {
55 r++;
56 t = 0;
57 }
58 }
59
60 while (r < ROUNDS + 1) { /* while not enough round key material calculated */
61 /* calculate new values */
62 tk[0][0] ^= S[tk[KC-1][1]];
63 tk[0][1] ^= S[tk[KC-1][2]];
64 tk[0][2] ^= S[tk[KC-1][3]];
65 tk[0][3] ^= S[tk[KC-1][0]];
66 tk[0][0] ^= rcon[rconpointer++];
67
68 if (KC != 8) {
69 for (j = 1; j < KC; j++) {
70 *((word32*)tk[j]) ^= *((word32*)tk[j-1]);
71 }
72 } else {
73 for (j = 1; j < KC/2; j++) {
74 *((word32*)tk[j]) ^= *((word32*)tk[j-1]);
75 }
76 tk[KC/2][0] ^= S[tk[KC/2 - 1][0]];
77 tk[KC/2][1] ^= S[tk[KC/2 - 1][1]];
78 tk[KC/2][2] ^= S[tk[KC/2 - 1][2]];
79 tk[KC/2][3] ^= S[tk[KC/2 - 1][3]];
80 for (j = KC/2 + 1; j < KC; j++) {
81 *((word32*)tk[j]) ^= *((word32*)tk[j-1]);
82 }
83 }
84 /* copy values into round key array */
85 for (j = 0; (j < KC) && (r < ROUNDS + 1); ) {
86 for (; (j < KC) && (t < 4); j++, t++) {
87 *((word32*)W[r][t]) = *((word32*)tk[j]);
88 }
89 if (t == 4) {
90 r++;
91 t = 0;
92 }
93 }
94 }
95 return 0;
96 #undef tk
97 }
98
99 int rijndaelKeyEncToDec(word8 W[MAXROUNDS+1][4][4], int ROUNDS) {
100 int r;
101 word8 *w;
102
103 for (r = 1; r < ROUNDS; r++) {
104 w = W[r][0];
105 *((word32*)w) =
106 *((const word32*)U1[w[0]])
107 ^ *((const word32*)U2[w[1]])
108 ^ *((const word32*)U3[w[2]])
109 ^ *((const word32*)U4[w[3]]);
110
111 w = W[r][1];
112 *((word32*)w) =
113 *((const word32*)U1[w[0]])
114 ^ *((const word32*)U2[w[1]])
115 ^ *((const word32*)U3[w[2]])
116 ^ *((const word32*)U4[w[3]]);
117
118 w = W[r][2];
119 *((word32*)w) =
120 *((const word32*)U1[w[0]])
121 ^ *((const word32*)U2[w[1]])
122 ^ *((const word32*)U3[w[2]])
123 ^ *((const word32*)U4[w[3]]);
124
125 w = W[r][3];
126 *((word32*)w) =
127 *((const word32*)U1[w[0]])
128 ^ *((const word32*)U2[w[1]])
129 ^ *((const word32*)U3[w[2]])
130 ^ *((const word32*)U4[w[3]]);
131 }
132 return 0;
133 }
134
135 /**
136 * Encrypt a single block.
137 */
138 int rijndaelEncrypt(word8 in[16], word8 out[16], word8 rk[MAXROUNDS+1][4][4], int ROUNDS) {
139 int r;
140 union {
141 word8 x8[16];
142 word32 x32[4];
143 } xa, xb;
144 #define a xa.x8
145 #define b xb.x8
146 union {
147 word8 x8[4][4];
148 word32 x32[4];
149 } xtemp;
150 #define temp xtemp.x8
151
152 memcpy(a, in, sizeof a);
153
154 *((word32*)temp[0]) = *((word32*)(a )) ^ *((word32*)rk[0][0]);
155 *((word32*)temp[1]) = *((word32*)(a+ 4)) ^ *((word32*)rk[0][1]);
156 *((word32*)temp[2]) = *((word32*)(a+ 8)) ^ *((word32*)rk[0][2]);
157 *((word32*)temp[3]) = *((word32*)(a+12)) ^ *((word32*)rk[0][3]);
158 *((word32*)(b )) = *((const word32*)T1[temp[0][0]])
159 ^ *((const word32*)T2[temp[1][1]])
160 ^ *((const word32*)T3[temp[2][2]])
161 ^ *((const word32*)T4[temp[3][3]]);
162 *((word32*)(b + 4)) = *((const word32*)T1[temp[1][0]])
163 ^ *((const word32*)T2[temp[2][1]])
164 ^ *((const word32*)T3[temp[3][2]])
165 ^ *((const word32*)T4[temp[0][3]]);
166 *((word32*)(b + 8)) = *((const word32*)T1[temp[2][0]])
167 ^ *((const word32*)T2[temp[3][1]])
168 ^ *((const word32*)T3[temp[0][2]])
169 ^ *((const word32*)T4[temp[1][3]]);
170 *((word32*)(b +12)) = *((const word32*)T1[temp[3][0]])
171 ^ *((const word32*)T2[temp[0][1]])
172 ^ *((const word32*)T3[temp[1][2]])
173 ^ *((const word32*)T4[temp[2][3]]);
174 for (r = 1; r < ROUNDS-1; r++) {
175 *((word32*)temp[0]) = *((word32*)(b )) ^ *((word32*)rk[r][0]);
176 *((word32*)temp[1]) = *((word32*)(b+ 4)) ^ *((word32*)rk[r][1]);
177 *((word32*)temp[2]) = *((word32*)(b+ 8)) ^ *((word32*)rk[r][2]);
178 *((word32*)temp[3]) = *((word32*)(b+12)) ^ *((word32*)rk[r][3]);
179
180 *((word32*)(b )) = *((const word32*)T1[temp[0][0]])
181 ^ *((const word32*)T2[temp[1][1]])
182 ^ *((const word32*)T3[temp[2][2]])
183 ^ *((const word32*)T4[temp[3][3]]);
184 *((word32*)(b + 4)) = *((const word32*)T1[temp[1][0]])
185 ^ *((const word32*)T2[temp[2][1]])
186 ^ *((const word32*)T3[temp[3][2]])
187 ^ *((const word32*)T4[temp[0][3]]);
188 *((word32*)(b + 8)) = *((const word32*)T1[temp[2][0]])
189 ^ *((const word32*)T2[temp[3][1]])
190 ^ *((const word32*)T3[temp[0][2]])
191 ^ *((const word32*)T4[temp[1][3]]);
192 *((word32*)(b +12)) = *((const word32*)T1[temp[3][0]])
193 ^ *((const word32*)T2[temp[0][1]])
194 ^ *((const word32*)T3[temp[1][2]])
195 ^ *((const word32*)T4[temp[2][3]]);
196 }
197 /* last round is special */
198 *((word32*)temp[0]) = *((word32*)(b )) ^ *((word32*)rk[ROUNDS-1][0]);
199 *((word32*)temp[1]) = *((word32*)(b+ 4)) ^ *((word32*)rk[ROUNDS-1][1]);
200 *((word32*)temp[2]) = *((word32*)(b+ 8)) ^ *((word32*)rk[ROUNDS-1][2]);
201 *((word32*)temp[3]) = *((word32*)(b+12)) ^ *((word32*)rk[ROUNDS-1][3]);
202 b[ 0] = T1[temp[0][0]][1];
203 b[ 1] = T1[temp[1][1]][1];
204 b[ 2] = T1[temp[2][2]][1];
205 b[ 3] = T1[temp[3][3]][1];
206 b[ 4] = T1[temp[1][0]][1];
207 b[ 5] = T1[temp[2][1]][1];
208 b[ 6] = T1[temp[3][2]][1];
209 b[ 7] = T1[temp[0][3]][1];
210 b[ 8] = T1[temp[2][0]][1];
211 b[ 9] = T1[temp[3][1]][1];
212 b[10] = T1[temp[0][2]][1];
213 b[11] = T1[temp[1][3]][1];
214 b[12] = T1[temp[3][0]][1];
215 b[13] = T1[temp[0][1]][1];
216 b[14] = T1[temp[1][2]][1];
217 b[15] = T1[temp[2][3]][1];
218 *((word32*)(b )) ^= *((word32*)rk[ROUNDS][0]);
219 *((word32*)(b+ 4)) ^= *((word32*)rk[ROUNDS][1]);
220 *((word32*)(b+ 8)) ^= *((word32*)rk[ROUNDS][2]);
221 *((word32*)(b+12)) ^= *((word32*)rk[ROUNDS][3]);
222
223 memcpy(out, b, sizeof b /* XXX out */);
224
225 return 0;
226 #undef a
227 #undef b
228 #undef temp
229 }
230
231 #ifdef INTERMEDIATE_VALUE_KAT
232 /**
233 * Encrypt only a certain number of rounds.
234 * Only used in the Intermediate Value Known Answer Test.
235 */
236 int rijndaelEncryptRound(word8 a[4][4], word8 rk[MAXROUNDS+1][4][4], int ROUNDS, int rounds) {
237 int r;
238 word8 temp[4][4];
239
240 /* make number of rounds sane */
241 if (rounds > ROUNDS) {
242 rounds = ROUNDS;
243 }
244
245 *((word32*)a[0]) = *((word32*)a[0]) ^ *((word32*)rk[0][0]);
246 *((word32*)a[1]) = *((word32*)a[1]) ^ *((word32*)rk[0][1]);
247 *((word32*)a[2]) = *((word32*)a[2]) ^ *((word32*)rk[0][2]);
248 *((word32*)a[3]) = *((word32*)a[3]) ^ *((word32*)rk[0][3]);
249
250 for (r = 1; (r <= rounds) && (r < ROUNDS); r++) {
251 *((word32*)temp[0]) = *((const word32*)T1[a[0][0]])
252 ^ *((const word32*)T2[a[1][1]])
253 ^ *((const word32*)T3[a[2][2]])
254 ^ *((const word32*)T4[a[3][3]]);
255 *((word32*)temp[1]) = *((const word32*)T1[a[1][0]])
256 ^ *((const word32*)T2[a[2][1]])
257 ^ *((const word32*)T3[a[3][2]])
258 ^ *((const word32*)T4[a[0][3]]);
259 *((word32*)temp[2]) = *((const word32*)T1[a[2][0]])
260 ^ *((const word32*)T2[a[3][1]])
261 ^ *((const word32*)T3[a[0][2]])
262 ^ *((const word32*)T4[a[1][3]]);
263 *((word32*)temp[3]) = *((const word32*)T1[a[3][0]])
264 ^ *((const word32*)T2[a[0][1]])
265 ^ *((const word32*)T3[a[1][2]])
266 ^ *((const word32*)T4[a[2][3]]);
267 *((word32*)a[0]) = *((word32*)temp[0]) ^ *((word32*)rk[r][0]);
268 *((word32*)a[1]) = *((word32*)temp[1]) ^ *((word32*)rk[r][1]);
269 *((word32*)a[2]) = *((word32*)temp[2]) ^ *((word32*)rk[r][2]);
270 *((word32*)a[3]) = *((word32*)temp[3]) ^ *((word32*)rk[r][3]);
271 }
272 if (rounds == ROUNDS) {
273 /* last round is special */
274 temp[0][0] = T1[a[0][0]][1];
275 temp[0][1] = T1[a[1][1]][1];
276 temp[0][2] = T1[a[2][2]][1];
277 temp[0][3] = T1[a[3][3]][1];
278 temp[1][0] = T1[a[1][0]][1];
279 temp[1][1] = T1[a[2][1]][1];
280 temp[1][2] = T1[a[3][2]][1];
281 temp[1][3] = T1[a[0][3]][1];
282 temp[2][0] = T1[a[2][0]][1];
283 temp[2][1] = T1[a[3][1]][1];
284 temp[2][2] = T1[a[0][2]][1];
285 temp[2][3] = T1[a[1][3]][1];
286 temp[3][0] = T1[a[3][0]][1];
287 temp[3][1] = T1[a[0][1]][1];
288 temp[3][2] = T1[a[1][2]][1];
289 temp[3][3] = T1[a[2][3]][1];
290 *((word32*)a[0]) = *((word32*)temp[0]) ^ *((word32*)rk[ROUNDS][0]);
291 *((word32*)a[1]) = *((word32*)temp[1]) ^ *((word32*)rk[ROUNDS][1]);
292 *((word32*)a[2]) = *((word32*)temp[2]) ^ *((word32*)rk[ROUNDS][2]);
293 *((word32*)a[3]) = *((word32*)temp[3]) ^ *((word32*)rk[ROUNDS][3]);
294 }
295
296 return 0;
297 }
298 #endif /* INTERMEDIATE_VALUE_KAT */
299
300 /**
301 * Decrypt a single block.
302 */
303 int rijndaelDecrypt(word8 in[16], word8 out[16], word8 rk[MAXROUNDS+1][4][4], int ROUNDS) {
304 int r;
305 union {
306 word8 x8[16];
307 word32 x32[4];
308 } xa, xb;
309 #define a xa.x8
310 #define b xb.x8
311 union {
312 word8 x8[4][4];
313 word32 x32[4];
314 } xtemp;
315 #define temp xtemp.x8
316
317 memcpy(a, in, sizeof a);
318
319 *((word32*)temp[0]) = *((word32*)(a )) ^ *((word32*)rk[ROUNDS][0]);
320 *((word32*)temp[1]) = *((word32*)(a+ 4)) ^ *((word32*)rk[ROUNDS][1]);
321 *((word32*)temp[2]) = *((word32*)(a+ 8)) ^ *((word32*)rk[ROUNDS][2]);
322 *((word32*)temp[3]) = *((word32*)(a+12)) ^ *((word32*)rk[ROUNDS][3]);
323
324 *((word32*)(b )) = *((const word32*)T5[temp[0][0]])
325 ^ *((const word32*)T6[temp[3][1]])
326 ^ *((const word32*)T7[temp[2][2]])
327 ^ *((const word32*)T8[temp[1][3]]);
328 *((word32*)(b+ 4)) = *((const word32*)T5[temp[1][0]])
329 ^ *((const word32*)T6[temp[0][1]])
330 ^ *((const word32*)T7[temp[3][2]])
331 ^ *((const word32*)T8[temp[2][3]]);
332 *((word32*)(b+ 8)) = *((const word32*)T5[temp[2][0]])
333 ^ *((const word32*)T6[temp[1][1]])
334 ^ *((const word32*)T7[temp[0][2]])
335 ^ *((const word32*)T8[temp[3][3]]);
336 *((word32*)(b+12)) = *((const word32*)T5[temp[3][0]])
337 ^ *((const word32*)T6[temp[2][1]])
338 ^ *((const word32*)T7[temp[1][2]])
339 ^ *((const word32*)T8[temp[0][3]]);
340 for (r = ROUNDS-1; r > 1; r--) {
341 *((word32*)temp[0]) = *((word32*)(b )) ^ *((word32*)rk[r][0]);
342 *((word32*)temp[1]) = *((word32*)(b+ 4)) ^ *((word32*)rk[r][1]);
343 *((word32*)temp[2]) = *((word32*)(b+ 8)) ^ *((word32*)rk[r][2]);
344 *((word32*)temp[3]) = *((word32*)(b+12)) ^ *((word32*)rk[r][3]);
345 *((word32*)(b )) = *((const word32*)T5[temp[0][0]])
346 ^ *((const word32*)T6[temp[3][1]])
347 ^ *((const word32*)T7[temp[2][2]])
348 ^ *((const word32*)T8[temp[1][3]]);
349 *((word32*)(b+ 4)) = *((const word32*)T5[temp[1][0]])
350 ^ *((const word32*)T6[temp[0][1]])
351 ^ *((const word32*)T7[temp[3][2]])
352 ^ *((const word32*)T8[temp[2][3]]);
353 *((word32*)(b+ 8)) = *((const word32*)T5[temp[2][0]])
354 ^ *((const word32*)T6[temp[1][1]])
355 ^ *((const word32*)T7[temp[0][2]])
356 ^ *((const word32*)T8[temp[3][3]]);
357 *((word32*)(b+12)) = *((const word32*)T5[temp[3][0]])
358 ^ *((const word32*)T6[temp[2][1]])
359 ^ *((const word32*)T7[temp[1][2]])
360 ^ *((const word32*)T8[temp[0][3]]);
361 }
362 /* last round is special */
363 *((word32*)temp[0]) = *((word32*)(b )) ^ *((word32*)rk[1][0]);
364 *((word32*)temp[1]) = *((word32*)(b+ 4)) ^ *((word32*)rk[1][1]);
365 *((word32*)temp[2]) = *((word32*)(b+ 8)) ^ *((word32*)rk[1][2]);
366 *((word32*)temp[3]) = *((word32*)(b+12)) ^ *((word32*)rk[1][3]);
367 b[ 0] = S5[temp[0][0]];
368 b[ 1] = S5[temp[3][1]];
369 b[ 2] = S5[temp[2][2]];
370 b[ 3] = S5[temp[1][3]];
371 b[ 4] = S5[temp[1][0]];
372 b[ 5] = S5[temp[0][1]];
373 b[ 6] = S5[temp[3][2]];
374 b[ 7] = S5[temp[2][3]];
375 b[ 8] = S5[temp[2][0]];
376 b[ 9] = S5[temp[1][1]];
377 b[10] = S5[temp[0][2]];
378 b[11] = S5[temp[3][3]];
379 b[12] = S5[temp[3][0]];
380 b[13] = S5[temp[2][1]];
381 b[14] = S5[temp[1][2]];
382 b[15] = S5[temp[0][3]];
383 *((word32*)(b )) ^= *((word32*)rk[0][0]);
384 *((word32*)(b+ 4)) ^= *((word32*)rk[0][1]);
385 *((word32*)(b+ 8)) ^= *((word32*)rk[0][2]);
386 *((word32*)(b+12)) ^= *((word32*)rk[0][3]);
387
388 memcpy(out, b, sizeof b /* XXX out */);
389
390 return 0;
391 #undef a
392 #undef b
393 #undef temp
394 }
395
396
397 #ifdef INTERMEDIATE_VALUE_KAT
398 /**
399 * Decrypt only a certain number of rounds.
400 * Only used in the Intermediate Value Known Answer Test.
401 * Operations rearranged such that the intermediate values
402 * of decryption correspond with the intermediate values
403 * of encryption.
404 */
405 int rijndaelDecryptRound(word8 a[4][4], word8 rk[MAXROUNDS+1][4][4], int ROUNDS, int rounds) {
406 int r, i;
407 word8 temp[4], shift;
408
409 /* make number of rounds sane */
410 if (rounds > ROUNDS) {
411 rounds = ROUNDS;
412 }
413 /* first round is special: */
414 *(word32 *)a[0] ^= *(word32 *)rk[ROUNDS][0];
415 *(word32 *)a[1] ^= *(word32 *)rk[ROUNDS][1];
416 *(word32 *)a[2] ^= *(word32 *)rk[ROUNDS][2];
417 *(word32 *)a[3] ^= *(word32 *)rk[ROUNDS][3];
418 for (i = 0; i < 4; i++) {
419 a[i][0] = Si[a[i][0]];
420 a[i][1] = Si[a[i][1]];
421 a[i][2] = Si[a[i][2]];
422 a[i][3] = Si[a[i][3]];
423 }
424 for (i = 1; i < 4; i++) {
425 shift = (4 - i) & 3;
426 temp[0] = a[(0 + shift) & 3][i];
427 temp[1] = a[(1 + shift) & 3][i];
428 temp[2] = a[(2 + shift) & 3][i];
429 temp[3] = a[(3 + shift) & 3][i];
430 a[0][i] = temp[0];
431 a[1][i] = temp[1];
432 a[2][i] = temp[2];
433 a[3][i] = temp[3];
434 }
435 /* ROUNDS-1 ordinary rounds */
436 for (r = ROUNDS-1; r > rounds; r--) {
437 *(word32 *)a[0] ^= *(word32 *)rk[r][0];
438 *(word32 *)a[1] ^= *(word32 *)rk[r][1];
439 *(word32 *)a[2] ^= *(word32 *)rk[r][2];
440 *(word32 *)a[3] ^= *(word32 *)rk[r][3];
441
442 *((word32*)a[0]) =
443 *((const word32*)U1[a[0][0]])
444 ^ *((const word32*)U2[a[0][1]])
445 ^ *((const word32*)U3[a[0][2]])
446 ^ *((const word32*)U4[a[0][3]]);
447
448 *((word32*)a[1]) =
449 *((const word32*)U1[a[1][0]])
450 ^ *((const word32*)U2[a[1][1]])
451 ^ *((const word32*)U3[a[1][2]])
452 ^ *((const word32*)U4[a[1][3]]);
453
454 *((word32*)a[2]) =
455 *((const word32*)U1[a[2][0]])
456 ^ *((const word32*)U2[a[2][1]])
457 ^ *((const word32*)U3[a[2][2]])
458 ^ *((const word32*)U4[a[2][3]]);
459
460 *((word32*)a[3]) =
461 *((const word32*)U1[a[3][0]])
462 ^ *((const word32*)U2[a[3][1]])
463 ^ *((const word32*)U3[a[3][2]])
464 ^ *((const word32*)U4[a[3][3]]);
465 for (i = 0; i < 4; i++) {
466 a[i][0] = Si[a[i][0]];
467 a[i][1] = Si[a[i][1]];
468 a[i][2] = Si[a[i][2]];
469 a[i][3] = Si[a[i][3]];
470 }
471 for (i = 1; i < 4; i++) {
472 shift = (4 - i) & 3;
473 temp[0] = a[(0 + shift) & 3][i];
474 temp[1] = a[(1 + shift) & 3][i];
475 temp[2] = a[(2 + shift) & 3][i];
476 temp[3] = a[(3 + shift) & 3][i];
477 a[0][i] = temp[0];
478 a[1][i] = temp[1];
479 a[2][i] = temp[2];
480 a[3][i] = temp[3];
481 }
482 }
483 if (rounds == 0) {
484 /* End with the extra key addition */
485 *(word32 *)a[0] ^= *(word32 *)rk[0][0];
486 *(word32 *)a[1] ^= *(word32 *)rk[0][1];
487 *(word32 *)a[2] ^= *(word32 *)rk[0][2];
488 *(word32 *)a[3] ^= *(word32 *)rk[0][3];
489 }
490 return 0;
491 }
492 #endif /* INTERMEDIATE_VALUE_KAT */