2 * copyright (c) 2007 Michael Niedermayer <michaelni@gmx.at>
4 * some optimization ideas from aes128.c by Reimar Doeffinger
6 * This file is part of FFmpeg.
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.
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.
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
25 #include "intreadwrite.h"
34 typedef struct AVAES {
35 // Note: round_key[16] is accessed in the init code, but this only
36 // overwrites state, which does not matter (see also commit ba554c0).
37 av_aes_block round_key[15];
38 av_aes_block state[2];
42 const int av_aes_size= sizeof(AVAES);
44 struct AVAES *av_aes_alloc(void)
46 return av_mallocz(sizeof(struct AVAES));
49 static const uint8_t rcon[10] = {
50 0x01, 0x02, 0x04, 0x08, 0x10, 0x20, 0x40, 0x80, 0x1b, 0x36
53 static uint8_t sbox[256];
54 static uint8_t inv_sbox[256];
56 static uint32_t enc_multbl[1][256];
57 static uint32_t dec_multbl[1][256];
59 static uint32_t enc_multbl[4][256];
60 static uint32_t dec_multbl[4][256];
64 # define ROT(x, s) ((x >> s) | (x << (32-s)))
66 # define ROT(x, s) ((x << s) | (x >> (32-s)))
69 static inline void addkey(av_aes_block *dst, const av_aes_block *src,
70 const av_aes_block *round_key)
72 dst->u64[0] = src->u64[0] ^ round_key->u64[0];
73 dst->u64[1] = src->u64[1] ^ round_key->u64[1];
76 static inline void addkey_s(av_aes_block *dst, const uint8_t *src,
77 const av_aes_block *round_key)
79 dst->u64[0] = AV_RN64(src) ^ round_key->u64[0];
80 dst->u64[1] = AV_RN64(src + 8) ^ round_key->u64[1];
83 static inline void addkey_d(uint8_t *dst, const av_aes_block *src,
84 const av_aes_block *round_key)
86 AV_WN64(dst, src->u64[0] ^ round_key->u64[0]);
87 AV_WN64(dst + 8, src->u64[1] ^ round_key->u64[1]);
90 static void subshift(av_aes_block s0[2], int s, const uint8_t *box)
92 av_aes_block *s1 = (av_aes_block *) (s0[0].u8 - s);
93 av_aes_block *s3 = (av_aes_block *) (s0[0].u8 + s);
95 s0[0].u8[ 0] = box[s0[1].u8[ 0]];
96 s0[0].u8[ 4] = box[s0[1].u8[ 4]];
97 s0[0].u8[ 8] = box[s0[1].u8[ 8]];
98 s0[0].u8[12] = box[s0[1].u8[12]];
99 s1[0].u8[ 3] = box[s1[1].u8[ 7]];
100 s1[0].u8[ 7] = box[s1[1].u8[11]];
101 s1[0].u8[11] = box[s1[1].u8[15]];
102 s1[0].u8[15] = box[s1[1].u8[ 3]];
103 s0[0].u8[ 2] = box[s0[1].u8[10]];
104 s0[0].u8[10] = box[s0[1].u8[ 2]];
105 s0[0].u8[ 6] = box[s0[1].u8[14]];
106 s0[0].u8[14] = box[s0[1].u8[ 6]];
107 s3[0].u8[ 1] = box[s3[1].u8[13]];
108 s3[0].u8[13] = box[s3[1].u8[ 9]];
109 s3[0].u8[ 9] = box[s3[1].u8[ 5]];
110 s3[0].u8[ 5] = box[s3[1].u8[ 1]];
113 static inline int mix_core(uint32_t multbl[][256], int a, int b, int c, int d){
115 return multbl[0][a] ^ ROT(multbl[0][b], 8) ^ ROT(multbl[0][c], 16) ^ ROT(multbl[0][d], 24);
117 return multbl[0][a] ^ multbl[1][b] ^ multbl[2][c] ^ multbl[3][d];
121 static inline void mix(av_aes_block state[2], uint32_t multbl[][256], int s1, int s3){
122 uint8_t (*src)[4] = state[1].u8x4;
123 state[0].u32[0] = mix_core(multbl, src[0][0], src[s1 ][1], src[2][2], src[s3 ][3]);
124 state[0].u32[1] = mix_core(multbl, src[1][0], src[s3-1][1], src[3][2], src[s1-1][3]);
125 state[0].u32[2] = mix_core(multbl, src[2][0], src[s3 ][1], src[0][2], src[s1 ][3]);
126 state[0].u32[3] = mix_core(multbl, src[3][0], src[s1-1][1], src[1][2], src[s3-1][3]);
129 static inline void crypt(AVAES *a, int s, const uint8_t *sbox,
130 uint32_t multbl[][256])
134 for (r = a->rounds - 1; r > 0; r--) {
135 mix(a->state, multbl, 3 - s, 1 + s);
136 addkey(&a->state[1], &a->state[0], &a->round_key[r]);
139 subshift(&a->state[0], s, sbox);
142 void av_aes_crypt(AVAES *a, uint8_t *dst, const uint8_t *src,
143 int count, uint8_t *iv, int decrypt)
146 addkey_s(&a->state[1], src, &a->round_key[a->rounds]);
148 crypt(a, 0, inv_sbox, dec_multbl);
150 addkey_s(&a->state[0], iv, &a->state[0]);
153 addkey_d(dst, &a->state[0], &a->round_key[0]);
156 addkey_s(&a->state[1], iv, &a->state[1]);
157 crypt(a, 2, sbox, enc_multbl);
158 addkey_d(dst, &a->state[0], &a->round_key[0]);
167 static void init_multbl2(uint32_t tbl[][256], const int c[4],
168 const uint8_t *log8, const uint8_t *alog8,
173 for (i = 0; i < 256; i++) {
178 k = alog8[x + log8[c[0]]];
179 l = alog8[x + log8[c[1]]];
180 m = alog8[x + log8[c[2]]];
181 n = alog8[x + log8[c[3]]];
182 tbl[0][i] = AV_NE(MKBETAG(k,l,m,n), MKTAG(k,l,m,n));
184 tbl[1][i] = ROT(tbl[0][i], 8);
185 tbl[2][i] = ROT(tbl[0][i], 16);
186 tbl[3][i] = ROT(tbl[0][i], 24);
192 // this is based on the reference AES code by Paulo Barreto and Vincent Rijmen
193 int av_aes_init(AVAES *a, const uint8_t *key, int key_bits, int decrypt)
195 int i, j, t, rconpointer = 0;
197 int KC = key_bits >> 5;
202 if (!enc_multbl[FF_ARRAY_ELEMS(enc_multbl)-1][FF_ARRAY_ELEMS(enc_multbl[0])-1]) {
204 for (i = 0; i < 255; i++) {
205 alog8[i] = alog8[i + 255] = j;
211 for (i = 0; i < 256; i++) {
212 j = i ? alog8[255 - log8[i]] : 0;
213 j ^= (j << 1) ^ (j << 2) ^ (j << 3) ^ (j << 4);
214 j = (j ^ (j >> 8) ^ 99) & 255;
218 init_multbl2(dec_multbl, (const int[4]) { 0xe, 0x9, 0xd, 0xb },
219 log8, alog8, inv_sbox);
220 init_multbl2(enc_multbl, (const int[4]) { 0x2, 0x1, 0x1, 0x3 },
224 if (key_bits != 128 && key_bits != 192 && key_bits != 256)
229 memcpy(tk, key, KC * 4);
230 memcpy(a->round_key[0].u8, key, KC * 4);
232 for (t = KC * 4; t < (rounds + 1) * 16; t += KC * 4) {
233 for (i = 0; i < 4; i++)
234 tk[0][i] ^= sbox[tk[KC - 1][(i + 1) & 3]];
235 tk[0][0] ^= rcon[rconpointer++];
237 for (j = 1; j < KC; j++) {
238 if (KC != 8 || j != KC >> 1)
239 for (i = 0; i < 4; i++)
240 tk[j][i] ^= tk[j - 1][i];
242 for (i = 0; i < 4; i++)
243 tk[j][i] ^= sbox[tk[j - 1][i]];
246 memcpy(a->round_key[0].u8 + t, tk, KC * 4);
250 for (i = 1; i < rounds; i++) {
252 tmp[2] = a->round_key[i];
253 subshift(&tmp[1], 0, sbox);
254 mix(tmp, dec_multbl, 1, 3);
255 a->round_key[i] = tmp[0];
258 for (i = 0; i < (rounds + 1) >> 1; i++) {
259 FFSWAP(av_aes_block, a->round_key[i], a->round_key[rounds-i]);
272 int main(int argc, char **argv)
276 uint8_t rkey[2][16] = {
278 { 0x10, 0xa5, 0x88, 0x69, 0xd7, 0x4b, 0xe5, 0xa3,
279 0x74, 0xcf, 0x86, 0x7c, 0xfb, 0x47, 0x38, 0x59 }
281 uint8_t pt[16], rpt[2][16]= {
282 { 0x6a, 0x84, 0x86, 0x7c, 0xd7, 0x7e, 0x12, 0xad,
283 0x07, 0xea, 0x1b, 0xe8, 0x95, 0xc5, 0x3f, 0xa3 },
286 uint8_t rct[2][16]= {
287 { 0x73, 0x22, 0x81, 0xc0, 0xa0, 0xaa, 0xb8, 0xf7,
288 0xa5, 0x4a, 0x0c, 0x67, 0xa0, 0xc4, 0x5e, 0xcf },
289 { 0x6d, 0x25, 0x1e, 0x69, 0x44, 0xb0, 0x51, 0xe0,
290 0x4e, 0xaa, 0x6f, 0xb4, 0xdb, 0xf7, 0x84, 0x65 }
295 av_log_set_level(AV_LOG_DEBUG);
297 for (i = 0; i < 2; i++) {
298 av_aes_init(&b, rkey[i], 128, 1);
299 av_aes_crypt(&b, temp, rct[i], 1, NULL, 1);
300 for (j = 0; j < 16; j++) {
301 if (rpt[i][j] != temp[j]) {
302 av_log(NULL, AV_LOG_ERROR, "%d %02X %02X\n",
303 j, rpt[i][j], temp[j]);
309 if (argc > 1 && !strcmp(argv[1], "-t")) {
313 av_aes_init(&ae, "PI=3.141592654..", 128, 0);
314 av_aes_init(&ad, "PI=3.141592654..", 128, 1);
315 av_lfg_init(&prng, 1);
317 for (i = 0; i < 10000; i++) {
318 for (j = 0; j < 16; j++) {
319 pt[j] = av_lfg_get(&prng);
323 av_aes_crypt(&ae, temp, pt, 1, NULL, 0);
325 av_log(NULL, AV_LOG_ERROR, "%02X %02X %02X %02X\n",
326 temp[0], temp[5], temp[10], temp[15]);
327 av_aes_crypt(&ad, temp, temp, 1, NULL, 1);
330 for (j = 0; j < 16; j++) {
331 if (pt[j] != temp[j]) {
332 av_log(NULL, AV_LOG_ERROR, "%d %d %02X %02X\n",
333 i, j, pt[j], temp[j]);