aes.c 10 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336
  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 "intreadwrite.h"
  25. typedef union {
  26. uint64_t u64[2];
  27. uint32_t u32[4];
  28. uint8_t u8x4[4][4];
  29. uint8_t u8[16];
  30. } av_aes_block;
  31. typedef struct AVAES {
  32. // Note: round_key[16] is accessed in the init code, but this only
  33. // overwrites state, which does not matter (see also commit ba554c0).
  34. av_aes_block round_key[15];
  35. av_aes_block state[2];
  36. int rounds;
  37. } AVAES;
  38. const int av_aes_size= sizeof(AVAES);
  39. static const uint8_t rcon[10] = {
  40. 0x01, 0x02, 0x04, 0x08, 0x10, 0x20, 0x40, 0x80, 0x1b, 0x36
  41. };
  42. static uint8_t sbox[256];
  43. static uint8_t inv_sbox[256];
  44. #if CONFIG_SMALL
  45. static uint32_t enc_multbl[1][256];
  46. static uint32_t dec_multbl[1][256];
  47. #else
  48. static uint32_t enc_multbl[4][256];
  49. static uint32_t dec_multbl[4][256];
  50. #endif
  51. #if HAVE_BIGENDIAN
  52. # define ROT(x, s) ((x >> s) | (x << (32-s)))
  53. #else
  54. # define ROT(x, s) ((x << s) | (x >> (32-s)))
  55. #endif
  56. static inline void addkey(av_aes_block *dst, const av_aes_block *src,
  57. const av_aes_block *round_key)
  58. {
  59. dst->u64[0] = src->u64[0] ^ round_key->u64[0];
  60. dst->u64[1] = src->u64[1] ^ round_key->u64[1];
  61. }
  62. static inline void addkey_s(av_aes_block *dst, const uint8_t *src,
  63. const av_aes_block *round_key)
  64. {
  65. dst->u64[0] = AV_RN64(src) ^ round_key->u64[0];
  66. dst->u64[1] = AV_RN64(src + 8) ^ round_key->u64[1];
  67. }
  68. static inline void addkey_d(uint8_t *dst, const av_aes_block *src,
  69. const av_aes_block *round_key)
  70. {
  71. AV_WN64(dst, src->u64[0] ^ round_key->u64[0]);
  72. AV_WN64(dst + 8, src->u64[1] ^ round_key->u64[1]);
  73. }
  74. static void subshift(av_aes_block s0[2], int s, const uint8_t *box)
  75. {
  76. av_aes_block *s1 = (av_aes_block *) (s0[0].u8 - s);
  77. av_aes_block *s3 = (av_aes_block *) (s0[0].u8 + s);
  78. s0[0].u8[ 0] = box[s0[1].u8[ 0]];
  79. s0[0].u8[ 4] = box[s0[1].u8[ 4]];
  80. s0[0].u8[ 8] = box[s0[1].u8[ 8]];
  81. s0[0].u8[12] = box[s0[1].u8[12]];
  82. s1[0].u8[ 3] = box[s1[1].u8[ 7]];
  83. s1[0].u8[ 7] = box[s1[1].u8[11]];
  84. s1[0].u8[11] = box[s1[1].u8[15]];
  85. s1[0].u8[15] = box[s1[1].u8[ 3]];
  86. s0[0].u8[ 2] = box[s0[1].u8[10]];
  87. s0[0].u8[10] = box[s0[1].u8[ 2]];
  88. s0[0].u8[ 6] = box[s0[1].u8[14]];
  89. s0[0].u8[14] = box[s0[1].u8[ 6]];
  90. s3[0].u8[ 1] = box[s3[1].u8[13]];
  91. s3[0].u8[13] = box[s3[1].u8[ 9]];
  92. s3[0].u8[ 9] = box[s3[1].u8[ 5]];
  93. s3[0].u8[ 5] = box[s3[1].u8[ 1]];
  94. }
  95. static inline int mix_core(uint32_t multbl[][256], int a, int b, int c, int d){
  96. #if CONFIG_SMALL
  97. return multbl[0][a] ^ ROT(multbl[0][b], 8) ^ ROT(multbl[0][c], 16) ^ ROT(multbl[0][d], 24);
  98. #else
  99. return multbl[0][a] ^ multbl[1][b] ^ multbl[2][c] ^ multbl[3][d];
  100. #endif
  101. }
  102. static inline void mix(av_aes_block state[2], uint32_t multbl[][256], int s1, int s3){
  103. uint8_t (*src)[4] = state[1].u8x4;
  104. state[0].u32[0] = mix_core(multbl, src[0][0], src[s1 ][1], src[2][2], src[s3 ][3]);
  105. state[0].u32[1] = mix_core(multbl, src[1][0], src[s3-1][1], src[3][2], src[s1-1][3]);
  106. state[0].u32[2] = mix_core(multbl, src[2][0], src[s3 ][1], src[0][2], src[s1 ][3]);
  107. state[0].u32[3] = mix_core(multbl, src[3][0], src[s1-1][1], src[1][2], src[s3-1][3]);
  108. }
  109. static inline void crypt(AVAES *a, int s, const uint8_t *sbox,
  110. uint32_t multbl[][256])
  111. {
  112. int r;
  113. for (r = a->rounds - 1; r > 0; r--) {
  114. mix(a->state, multbl, 3 - s, 1 + s);
  115. addkey(&a->state[1], &a->state[0], &a->round_key[r]);
  116. }
  117. subshift(&a->state[0], s, sbox);
  118. }
  119. void av_aes_crypt(AVAES *a, uint8_t *dst, const uint8_t *src,
  120. int count, uint8_t *iv, int decrypt)
  121. {
  122. while (count--) {
  123. addkey_s(&a->state[1], src, &a->round_key[a->rounds]);
  124. if (decrypt) {
  125. crypt(a, 0, inv_sbox, dec_multbl);
  126. if (iv) {
  127. addkey_s(&a->state[0], iv, &a->state[0]);
  128. memcpy(iv, src, 16);
  129. }
  130. addkey_d(dst, &a->state[0], &a->round_key[0]);
  131. } else {
  132. if (iv)
  133. addkey_s(&a->state[1], iv, &a->state[1]);
  134. crypt(a, 2, sbox, enc_multbl);
  135. addkey_d(dst, &a->state[0], &a->round_key[0]);
  136. if (iv)
  137. memcpy(iv, dst, 16);
  138. }
  139. src += 16;
  140. dst += 16;
  141. }
  142. }
  143. static void init_multbl2(uint32_t tbl[][256], const int c[4],
  144. const uint8_t *log8, const uint8_t *alog8,
  145. const uint8_t *sbox)
  146. {
  147. int i;
  148. for (i = 0; i < 256; i++) {
  149. int x = sbox[i];
  150. if (x) {
  151. int k, l, m, n;
  152. x = log8[x];
  153. k = alog8[x + log8[c[0]]];
  154. l = alog8[x + log8[c[1]]];
  155. m = alog8[x + log8[c[2]]];
  156. n = alog8[x + log8[c[3]]];
  157. tbl[0][i] = AV_NE(MKBETAG(k,l,m,n), MKTAG(k,l,m,n));
  158. #if !CONFIG_SMALL
  159. tbl[1][i] = ROT(tbl[0][i], 8);
  160. tbl[2][i] = ROT(tbl[0][i], 16);
  161. tbl[3][i] = ROT(tbl[0][i], 24);
  162. #endif
  163. }
  164. }
  165. }
  166. // this is based on the reference AES code by Paulo Barreto and Vincent Rijmen
  167. int av_aes_init(AVAES *a, const uint8_t *key, int key_bits, int decrypt)
  168. {
  169. int i, j, t, rconpointer = 0;
  170. uint8_t tk[8][4];
  171. int KC = key_bits >> 5;
  172. int rounds = KC + 6;
  173. uint8_t log8[256];
  174. uint8_t alog8[512];
  175. if (!enc_multbl[FF_ARRAY_ELEMS(enc_multbl)-1][FF_ARRAY_ELEMS(enc_multbl[0])-1]) {
  176. j = 1;
  177. for (i = 0; i < 255; i++) {
  178. alog8[i] = alog8[i + 255] = j;
  179. log8[j] = i;
  180. j ^= j + j;
  181. if (j > 255)
  182. j ^= 0x11B;
  183. }
  184. for (i = 0; i < 256; i++) {
  185. j = i ? alog8[255 - log8[i]] : 0;
  186. j ^= (j << 1) ^ (j << 2) ^ (j << 3) ^ (j << 4);
  187. j = (j ^ (j >> 8) ^ 99) & 255;
  188. inv_sbox[j] = i;
  189. sbox[i] = j;
  190. }
  191. init_multbl2(dec_multbl, (const int[4]) { 0xe, 0x9, 0xd, 0xb },
  192. log8, alog8, inv_sbox);
  193. init_multbl2(enc_multbl, (const int[4]) { 0x2, 0x1, 0x1, 0x3 },
  194. log8, alog8, sbox);
  195. }
  196. if (key_bits != 128 && key_bits != 192 && key_bits != 256)
  197. return -1;
  198. a->rounds = rounds;
  199. memcpy(tk, key, KC * 4);
  200. memcpy(a->round_key[0].u8, key, KC * 4);
  201. for (t = KC * 4; t < (rounds + 1) * 16; t += KC * 4) {
  202. for (i = 0; i < 4; i++)
  203. tk[0][i] ^= sbox[tk[KC - 1][(i + 1) & 3]];
  204. tk[0][0] ^= rcon[rconpointer++];
  205. for (j = 1; j < KC; j++) {
  206. if (KC != 8 || j != KC >> 1)
  207. for (i = 0; i < 4; i++)
  208. tk[j][i] ^= tk[j - 1][i];
  209. else
  210. for (i = 0; i < 4; i++)
  211. tk[j][i] ^= sbox[tk[j - 1][i]];
  212. }
  213. memcpy(a->round_key[0].u8 + t, tk, KC * 4);
  214. }
  215. if (decrypt) {
  216. for (i = 1; i < rounds; i++) {
  217. av_aes_block tmp[3];
  218. tmp[2] = a->round_key[i];
  219. subshift(&tmp[1], 0, sbox);
  220. mix(tmp, dec_multbl, 1, 3);
  221. a->round_key[i] = tmp[0];
  222. }
  223. } else {
  224. for (i = 0; i < (rounds + 1) >> 1; i++) {
  225. FFSWAP(av_aes_block, a->round_key[i], a->round_key[rounds-i]);
  226. }
  227. }
  228. return 0;
  229. }
  230. #ifdef TEST
  231. // LCOV_EXCL_START
  232. #include <string.h>
  233. #include "lfg.h"
  234. #include "log.h"
  235. int main(int argc, char **argv)
  236. {
  237. int i, j;
  238. AVAES b;
  239. uint8_t rkey[2][16] = {
  240. { 0 },
  241. { 0x10, 0xa5, 0x88, 0x69, 0xd7, 0x4b, 0xe5, 0xa3,
  242. 0x74, 0xcf, 0x86, 0x7c, 0xfb, 0x47, 0x38, 0x59 }
  243. };
  244. uint8_t pt[16], rpt[2][16]= {
  245. { 0x6a, 0x84, 0x86, 0x7c, 0xd7, 0x7e, 0x12, 0xad,
  246. 0x07, 0xea, 0x1b, 0xe8, 0x95, 0xc5, 0x3f, 0xa3 },
  247. { 0 }
  248. };
  249. uint8_t rct[2][16]= {
  250. { 0x73, 0x22, 0x81, 0xc0, 0xa0, 0xaa, 0xb8, 0xf7,
  251. 0xa5, 0x4a, 0x0c, 0x67, 0xa0, 0xc4, 0x5e, 0xcf },
  252. { 0x6d, 0x25, 0x1e, 0x69, 0x44, 0xb0, 0x51, 0xe0,
  253. 0x4e, 0xaa, 0x6f, 0xb4, 0xdb, 0xf7, 0x84, 0x65 }
  254. };
  255. uint8_t temp[16];
  256. int err = 0;
  257. av_log_set_level(AV_LOG_DEBUG);
  258. for (i = 0; i < 2; i++) {
  259. av_aes_init(&b, rkey[i], 128, 1);
  260. av_aes_crypt(&b, temp, rct[i], 1, NULL, 1);
  261. for (j = 0; j < 16; j++) {
  262. if (rpt[i][j] != temp[j]) {
  263. av_log(NULL, AV_LOG_ERROR, "%d %02X %02X\n",
  264. j, rpt[i][j], temp[j]);
  265. err = 1;
  266. }
  267. }
  268. }
  269. if (argc > 1 && !strcmp(argv[1], "-t")) {
  270. AVAES ae, ad;
  271. AVLFG prng;
  272. av_aes_init(&ae, "PI=3.141592654..", 128, 0);
  273. av_aes_init(&ad, "PI=3.141592654..", 128, 1);
  274. av_lfg_init(&prng, 1);
  275. for (i = 0; i < 10000; i++) {
  276. for (j = 0; j < 16; j++) {
  277. pt[j] = av_lfg_get(&prng);
  278. }
  279. {
  280. START_TIMER;
  281. av_aes_crypt(&ae, temp, pt, 1, NULL, 0);
  282. if (!(i & (i - 1)))
  283. av_log(NULL, AV_LOG_ERROR, "%02X %02X %02X %02X\n",
  284. temp[0], temp[5], temp[10], temp[15]);
  285. av_aes_crypt(&ad, temp, temp, 1, NULL, 1);
  286. STOP_TIMER("aes");
  287. }
  288. for (j = 0; j < 16; j++) {
  289. if (pt[j] != temp[j]) {
  290. av_log(NULL, AV_LOG_ERROR, "%d %d %02X %02X\n",
  291. i, j, pt[j], temp[j]);
  292. }
  293. }
  294. }
  295. }
  296. return err;
  297. }
  298. // LCOV_EXCL_STOP
  299. #endif