You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

353 lines
11KB

  1. /*
  2. * copyright (c) 2007 Michael Niedermayer <michaelni@gmx.at>
  3. *
  4. * some optimization ideas from aes128.c by Reimar Doeffinger
  5. *
  6. * This file is part of FFmpeg.
  7. *
  8. * FFmpeg is free software; you can redistribute it and/or
  9. * modify it under the terms of the GNU Lesser General Public
  10. * License as published by the Free Software Foundation; either
  11. * version 2.1 of the License, or (at your option) any later version.
  12. *
  13. * FFmpeg is distributed in the hope that it will be useful,
  14. * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15. * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  16. * Lesser General Public License for more details.
  17. *
  18. * You should have received a copy of the GNU Lesser General Public
  19. * License along with FFmpeg; if not, write to the Free Software
  20. * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  21. */
  22. #include "common.h"
  23. #include "aes.h"
  24. #include "aes_internal.h"
  25. #include "intreadwrite.h"
  26. #include "timer.h"
  27. const int av_aes_size= sizeof(AVAES);
  28. struct AVAES *av_aes_alloc(void)
  29. {
  30. return av_mallocz(sizeof(struct AVAES));
  31. }
  32. static const uint8_t rcon[10] = {
  33. 0x01, 0x02, 0x04, 0x08, 0x10, 0x20, 0x40, 0x80, 0x1b, 0x36
  34. };
  35. static uint8_t sbox[256];
  36. static uint8_t inv_sbox[256];
  37. #if CONFIG_SMALL
  38. static uint32_t enc_multbl[1][256];
  39. static uint32_t dec_multbl[1][256];
  40. #else
  41. static uint32_t enc_multbl[4][256];
  42. static uint32_t dec_multbl[4][256];
  43. #endif
  44. #if HAVE_BIGENDIAN
  45. # define ROT(x, s) (((x) >> (s)) | ((x) << (32-(s))))
  46. #else
  47. # define ROT(x, s) (((x) << (s)) | ((x) >> (32-(s))))
  48. #endif
  49. static inline void addkey(av_aes_block *dst, const av_aes_block *src,
  50. const av_aes_block *round_key)
  51. {
  52. dst->u64[0] = src->u64[0] ^ round_key->u64[0];
  53. dst->u64[1] = src->u64[1] ^ round_key->u64[1];
  54. }
  55. static inline void addkey_s(av_aes_block *dst, const uint8_t *src,
  56. const av_aes_block *round_key)
  57. {
  58. dst->u64[0] = AV_RN64(src) ^ round_key->u64[0];
  59. dst->u64[1] = AV_RN64(src + 8) ^ round_key->u64[1];
  60. }
  61. static inline void addkey_d(uint8_t *dst, const av_aes_block *src,
  62. const av_aes_block *round_key)
  63. {
  64. AV_WN64(dst, src->u64[0] ^ round_key->u64[0]);
  65. AV_WN64(dst + 8, src->u64[1] ^ round_key->u64[1]);
  66. }
  67. static void subshift(av_aes_block s0[2], int s, const uint8_t *box)
  68. {
  69. av_aes_block *s1 = (av_aes_block *) (s0[0].u8 - s);
  70. av_aes_block *s3 = (av_aes_block *) (s0[0].u8 + s);
  71. s0[0].u8[ 0] = box[s0[1].u8[ 0]];
  72. s0[0].u8[ 4] = box[s0[1].u8[ 4]];
  73. s0[0].u8[ 8] = box[s0[1].u8[ 8]];
  74. s0[0].u8[12] = box[s0[1].u8[12]];
  75. s1[0].u8[ 3] = box[s1[1].u8[ 7]];
  76. s1[0].u8[ 7] = box[s1[1].u8[11]];
  77. s1[0].u8[11] = box[s1[1].u8[15]];
  78. s1[0].u8[15] = box[s1[1].u8[ 3]];
  79. s0[0].u8[ 2] = box[s0[1].u8[10]];
  80. s0[0].u8[10] = box[s0[1].u8[ 2]];
  81. s0[0].u8[ 6] = box[s0[1].u8[14]];
  82. s0[0].u8[14] = box[s0[1].u8[ 6]];
  83. s3[0].u8[ 1] = box[s3[1].u8[13]];
  84. s3[0].u8[13] = box[s3[1].u8[ 9]];
  85. s3[0].u8[ 9] = box[s3[1].u8[ 5]];
  86. s3[0].u8[ 5] = box[s3[1].u8[ 1]];
  87. }
  88. static inline int mix_core(uint32_t multbl[][256], int a, int b, int c, int d){
  89. #if CONFIG_SMALL
  90. return multbl[0][a] ^ ROT(multbl[0][b], 8) ^ ROT(multbl[0][c], 16) ^ ROT(multbl[0][d], 24);
  91. #else
  92. return multbl[0][a] ^ multbl[1][b] ^ multbl[2][c] ^ multbl[3][d];
  93. #endif
  94. }
  95. static inline void mix(av_aes_block state[2], uint32_t multbl[][256], int s1, int s3){
  96. uint8_t (*src)[4] = state[1].u8x4;
  97. state[0].u32[0] = mix_core(multbl, src[0][0], src[s1 ][1], src[2][2], src[s3 ][3]);
  98. state[0].u32[1] = mix_core(multbl, src[1][0], src[s3-1][1], src[3][2], src[s1-1][3]);
  99. state[0].u32[2] = mix_core(multbl, src[2][0], src[s3 ][1], src[0][2], src[s1 ][3]);
  100. state[0].u32[3] = mix_core(multbl, src[3][0], src[s1-1][1], src[1][2], src[s3-1][3]);
  101. }
  102. static inline void aes_crypt(AVAES *a, int s, const uint8_t *sbox,
  103. uint32_t multbl[][256])
  104. {
  105. int r;
  106. for (r = a->rounds - 1; r > 0; r--) {
  107. mix(a->state, multbl, 3 - s, 1 + s);
  108. addkey(&a->state[1], &a->state[0], &a->round_key[r]);
  109. }
  110. subshift(&a->state[0], s, sbox);
  111. }
  112. static void aes_encrypt(AVAES *a, uint8_t *dst, const uint8_t *src,
  113. int count, uint8_t *iv, int rounds)
  114. {
  115. while (count--) {
  116. addkey_s(&a->state[1], src, &a->round_key[rounds]);
  117. if (iv)
  118. addkey_s(&a->state[1], iv, &a->state[1]);
  119. aes_crypt(a, 2, sbox, enc_multbl);
  120. addkey_d(dst, &a->state[0], &a->round_key[0]);
  121. if (iv)
  122. memcpy(iv, dst, 16);
  123. src += 16;
  124. dst += 16;
  125. }
  126. }
  127. static void aes_decrypt(AVAES *a, uint8_t *dst, const uint8_t *src,
  128. int count, uint8_t *iv, int rounds)
  129. {
  130. while (count--) {
  131. addkey_s(&a->state[1], src, &a->round_key[rounds]);
  132. aes_crypt(a, 0, inv_sbox, dec_multbl);
  133. if (iv) {
  134. addkey_s(&a->state[0], iv, &a->state[0]);
  135. memcpy(iv, src, 16);
  136. }
  137. addkey_d(dst, &a->state[0], &a->round_key[0]);
  138. src += 16;
  139. dst += 16;
  140. }
  141. }
  142. void av_aes_crypt(AVAES *a, uint8_t *dst, const uint8_t *src,
  143. int count, uint8_t *iv, int decrypt)
  144. {
  145. a->crypt(a, dst, src, count, iv, a->rounds);
  146. }
  147. static void init_multbl2(uint32_t tbl[][256], const int c[4],
  148. const uint8_t *log8, const uint8_t *alog8,
  149. const uint8_t *sbox)
  150. {
  151. int i;
  152. for (i = 0; i < 256; i++) {
  153. int x = sbox[i];
  154. if (x) {
  155. int k, l, m, n;
  156. x = log8[x];
  157. k = alog8[x + log8[c[0]]];
  158. l = alog8[x + log8[c[1]]];
  159. m = alog8[x + log8[c[2]]];
  160. n = alog8[x + log8[c[3]]];
  161. tbl[0][i] = AV_NE(MKBETAG(k,l,m,n), MKTAG(k,l,m,n));
  162. #if !CONFIG_SMALL
  163. tbl[1][i] = ROT(tbl[0][i], 8);
  164. tbl[2][i] = ROT(tbl[0][i], 16);
  165. tbl[3][i] = ROT(tbl[0][i], 24);
  166. #endif
  167. }
  168. }
  169. }
  170. // this is based on the reference AES code by Paulo Barreto and Vincent Rijmen
  171. int av_aes_init(AVAES *a, const uint8_t *key, int key_bits, int decrypt)
  172. {
  173. int i, j, t, rconpointer = 0;
  174. uint8_t tk[8][4];
  175. int KC = key_bits >> 5;
  176. int rounds = KC + 6;
  177. uint8_t log8[256];
  178. uint8_t alog8[512];
  179. a->crypt = decrypt ? aes_decrypt : aes_encrypt;
  180. if (!enc_multbl[FF_ARRAY_ELEMS(enc_multbl)-1][FF_ARRAY_ELEMS(enc_multbl[0])-1]) {
  181. j = 1;
  182. for (i = 0; i < 255; i++) {
  183. alog8[i] = alog8[i + 255] = j;
  184. log8[j] = i;
  185. j ^= j + j;
  186. if (j > 255)
  187. j ^= 0x11B;
  188. }
  189. for (i = 0; i < 256; i++) {
  190. j = i ? alog8[255 - log8[i]] : 0;
  191. j ^= (j << 1) ^ (j << 2) ^ (j << 3) ^ (j << 4);
  192. j = (j ^ (j >> 8) ^ 99) & 255;
  193. inv_sbox[j] = i;
  194. sbox[i] = j;
  195. }
  196. init_multbl2(dec_multbl, (const int[4]) { 0xe, 0x9, 0xd, 0xb },
  197. log8, alog8, inv_sbox);
  198. init_multbl2(enc_multbl, (const int[4]) { 0x2, 0x1, 0x1, 0x3 },
  199. log8, alog8, sbox);
  200. }
  201. if (key_bits != 128 && key_bits != 192 && key_bits != 256)
  202. return AVERROR(EINVAL);
  203. a->rounds = rounds;
  204. memcpy(tk, key, KC * 4);
  205. memcpy(a->round_key[0].u8, key, KC * 4);
  206. for (t = KC * 4; t < (rounds + 1) * 16; t += KC * 4) {
  207. for (i = 0; i < 4; i++)
  208. tk[0][i] ^= sbox[tk[KC - 1][(i + 1) & 3]];
  209. tk[0][0] ^= rcon[rconpointer++];
  210. for (j = 1; j < KC; j++) {
  211. if (KC != 8 || j != KC >> 1)
  212. for (i = 0; i < 4; i++)
  213. tk[j][i] ^= tk[j - 1][i];
  214. else
  215. for (i = 0; i < 4; i++)
  216. tk[j][i] ^= sbox[tk[j - 1][i]];
  217. }
  218. memcpy(a->round_key[0].u8 + t, tk, KC * 4);
  219. }
  220. if (decrypt) {
  221. for (i = 1; i < rounds; i++) {
  222. av_aes_block tmp[3];
  223. tmp[2] = a->round_key[i];
  224. subshift(&tmp[1], 0, sbox);
  225. mix(tmp, dec_multbl, 1, 3);
  226. a->round_key[i] = tmp[0];
  227. }
  228. } else {
  229. for (i = 0; i < (rounds + 1) >> 1; i++) {
  230. FFSWAP(av_aes_block, a->round_key[i], a->round_key[rounds-i]);
  231. }
  232. }
  233. return 0;
  234. }
  235. #ifdef TEST
  236. // LCOV_EXCL_START
  237. #include <string.h>
  238. #include "lfg.h"
  239. #include "log.h"
  240. int main(int argc, char **argv)
  241. {
  242. int i, j;
  243. AVAES b;
  244. uint8_t rkey[2][16] = {
  245. { 0 },
  246. { 0x10, 0xa5, 0x88, 0x69, 0xd7, 0x4b, 0xe5, 0xa3,
  247. 0x74, 0xcf, 0x86, 0x7c, 0xfb, 0x47, 0x38, 0x59 }
  248. };
  249. uint8_t pt[32], rpt[2][16]= {
  250. { 0x6a, 0x84, 0x86, 0x7c, 0xd7, 0x7e, 0x12, 0xad,
  251. 0x07, 0xea, 0x1b, 0xe8, 0x95, 0xc5, 0x3f, 0xa3 },
  252. { 0 }
  253. };
  254. uint8_t rct[2][16]= {
  255. { 0x73, 0x22, 0x81, 0xc0, 0xa0, 0xaa, 0xb8, 0xf7,
  256. 0xa5, 0x4a, 0x0c, 0x67, 0xa0, 0xc4, 0x5e, 0xcf },
  257. { 0x6d, 0x25, 0x1e, 0x69, 0x44, 0xb0, 0x51, 0xe0,
  258. 0x4e, 0xaa, 0x6f, 0xb4, 0xdb, 0xf7, 0x84, 0x65 }
  259. };
  260. uint8_t temp[32];
  261. uint8_t iv[2][16];
  262. int err = 0;
  263. av_log_set_level(AV_LOG_DEBUG);
  264. for (i = 0; i < 2; i++) {
  265. av_aes_init(&b, rkey[i], 128, 1);
  266. av_aes_crypt(&b, temp, rct[i], 1, NULL, 1);
  267. for (j = 0; j < 16; j++) {
  268. if (rpt[i][j] != temp[j]) {
  269. av_log(NULL, AV_LOG_ERROR, "%d %02X %02X\n",
  270. j, rpt[i][j], temp[j]);
  271. err = 1;
  272. }
  273. }
  274. }
  275. if (argc > 1 && !strcmp(argv[1], "-t")) {
  276. AVAES ae, ad;
  277. AVLFG prng;
  278. av_aes_init(&ae, (const uint8_t*)"PI=3.141592654..", 128, 0);
  279. av_aes_init(&ad, (const uint8_t*)"PI=3.141592654..", 128, 1);
  280. av_lfg_init(&prng, 1);
  281. for (i = 0; i < 10000; i++) {
  282. for (j = 0; j < 32; j++) {
  283. pt[j] = av_lfg_get(&prng);
  284. }
  285. for (j = 0; j < 16; j++) {
  286. iv[0][j] = iv[1][j] = av_lfg_get(&prng);
  287. }
  288. {
  289. START_TIMER;
  290. av_aes_crypt(&ae, temp, pt, 2, iv[0], 0);
  291. if (!(i & (i - 1)))
  292. av_log(NULL, AV_LOG_ERROR, "%02X %02X %02X %02X\n",
  293. temp[0], temp[5], temp[10], temp[15]);
  294. av_aes_crypt(&ad, temp, temp, 2, iv[1], 1);
  295. av_aes_crypt(&ae, temp, pt, 2, NULL, 0);
  296. if (!(i & (i - 1)))
  297. av_log(NULL, AV_LOG_ERROR, "%02X %02X %02X %02X\n",
  298. temp[0], temp[5], temp[10], temp[15]);
  299. av_aes_crypt(&ad, temp, temp, 2, NULL, 1);
  300. STOP_TIMER("aes");
  301. }
  302. for (j = 0; j < 16; j++) {
  303. if (pt[j] != temp[j]) {
  304. av_log(NULL, AV_LOG_ERROR, "%d %d %02X %02X\n",
  305. i, j, pt[j], temp[j]);
  306. }
  307. }
  308. }
  309. }
  310. return err;
  311. }
  312. // LCOV_EXCL_STOP
  313. #endif