libxvid: add missing include of libavutil/mathematics.h
[libav.git] / libavutil / aes.c
CommitLineData
3306dfd5 1/*
034f5fba 2 * copyright (c) 2007 Michael Niedermayer <michaelni@gmx.at>
3306dfd5 3 *
2eb697a4
DB
4 * some optimization ideas from aes128.c by Reimar Doeffinger
5 *
2912e87a 6 * This file is part of Libav.
3306dfd5 7 *
2912e87a 8 * Libav is free software; you can redistribute it and/or
3306dfd5
MN
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 *
2912e87a 13 * Libav is distributed in the hope that it will be useful,
3306dfd5
MN
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
2912e87a 19 * License along with Libav; if not, write to the Free Software
3306dfd5
MN
20 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
21 */
22
23#include "common.h"
3306dfd5
MN
24#include "aes.h"
25
dd025f25
RD
26typedef union {
27 uint64_t u64[2];
28 uint32_t u32[4];
29 uint8_t u8x4[4][4];
30 uint8_t u8[16];
31} av_aes_block;
32
4bfe0644 33typedef struct AVAES {
ef0d7a0d
RD
34 // Note: round_key[16] is accessed in the init code, but this only
35 // overwrites state, which does not matter (see also r7471).
dd025f25
RD
36 av_aes_block round_key[15];
37 av_aes_block state[2];
3306dfd5 38 int rounds;
4bfe0644 39} AVAES;
3306dfd5 40
6572e1a4
MN
41const int av_aes_size= sizeof(AVAES);
42
a6e923c9
MN
43static const uint8_t rcon[10] = {
44 0x01, 0x02, 0x04, 0x08, 0x10, 0x20, 0x40, 0x80, 0x1b, 0x36
3306dfd5
MN
45};
46
3306dfd5
MN
47static uint8_t sbox[256];
48static uint8_t inv_sbox[256];
b250f9c6 49#if CONFIG_SMALL
cd0968c4
MN
50static uint32_t enc_multbl[1][256];
51static uint32_t dec_multbl[1][256];
52#else
53static uint32_t enc_multbl[4][256];
54static uint32_t dec_multbl[4][256];
55#endif
3306dfd5 56
8f175810
MR
57#if HAVE_BIGENDIAN
58# define ROT(x, s) ((x >> s) | (x << (32-s)))
59#else
60# define ROT(x, s) ((x << s) | (x >> (32-s)))
61#endif
5d20f19b 62
4bfe0644
MR
63static inline void addkey(av_aes_block *dst, const av_aes_block *src,
64 const av_aes_block *round_key)
65{
dd025f25
RD
66 dst->u64[0] = src->u64[0] ^ round_key->u64[0];
67 dst->u64[1] = src->u64[1] ^ round_key->u64[1];
3306dfd5
MN
68}
69
4bfe0644
MR
70static void subshift(av_aes_block s0[2], int s, const uint8_t *box)
71{
72 av_aes_block *s1 = (av_aes_block *) (s0[0].u8 - s);
73 av_aes_block *s3 = (av_aes_block *) (s0[0].u8 + s);
74
75 s0[0].u8[ 0] = box[s0[1].u8[ 0]];
76 s0[0].u8[ 4] = box[s0[1].u8[ 4]];
77 s0[0].u8[ 8] = box[s0[1].u8[ 8]];
78 s0[0].u8[12] = box[s0[1].u8[12]];
79 s1[0].u8[ 3] = box[s1[1].u8[ 7]];
80 s1[0].u8[ 7] = box[s1[1].u8[11]];
81 s1[0].u8[11] = box[s1[1].u8[15]];
82 s1[0].u8[15] = box[s1[1].u8[ 3]];
83 s0[0].u8[ 2] = box[s0[1].u8[10]];
84 s0[0].u8[10] = box[s0[1].u8[ 2]];
85 s0[0].u8[ 6] = box[s0[1].u8[14]];
86 s0[0].u8[14] = box[s0[1].u8[ 6]];
87 s3[0].u8[ 1] = box[s3[1].u8[13]];
88 s3[0].u8[13] = box[s3[1].u8[ 9]];
89 s3[0].u8[ 9] = box[s3[1].u8[ 5]];
90 s3[0].u8[ 5] = box[s3[1].u8[ 1]];
1aa72a70 91}
0458e790 92
17c78b08 93static inline int mix_core(uint32_t multbl[][256], int a, int b, int c, int d){
b250f9c6 94#if CONFIG_SMALL
97f6d154 95 return multbl[0][a] ^ ROT(multbl[0][b], 8) ^ ROT(multbl[0][c], 16) ^ ROT(multbl[0][d], 24);
cd0968c4 96#else
97f6d154 97 return multbl[0][a] ^ multbl[1][b] ^ multbl[2][c] ^ multbl[3][d];
cd0968c4 98#endif
3306dfd5 99}
298dd6b7 100
dd025f25 101static inline void mix(av_aes_block state[2], uint32_t multbl[][256], int s1, int s3){
19c039ee
RD
102 uint8_t (*src)[4] = state[1].u8x4;
103 state[0].u32[0] = mix_core(multbl, src[0][0], src[s1 ][1], src[2][2], src[s3 ][3]);
104 state[0].u32[1] = mix_core(multbl, src[1][0], src[s3-1][1], src[3][2], src[s1-1][3]);
105 state[0].u32[2] = mix_core(multbl, src[2][0], src[s3 ][1], src[0][2], src[s1 ][3]);
106 state[0].u32[3] = mix_core(multbl, src[3][0], src[s1-1][1], src[1][2], src[s3-1][3]);
1aa72a70 107}
3306dfd5 108
4bfe0644
MR
109static inline void crypt(AVAES *a, int s, const uint8_t *sbox,
110 uint32_t multbl[][256])
111{
14bc3849 112 int r;
3306dfd5 113
4bfe0644
MR
114 for (r = a->rounds - 1; r > 0; r--) {
115 mix(a->state, multbl, 3 - s, 1 + s);
dd025f25 116 addkey(&a->state[1], &a->state[0], &a->round_key[r]);
3306dfd5 117 }
4bfe0644 118
dd025f25 119 subshift(&a->state[0], s, sbox);
3306dfd5
MN
120}
121
4bfe0644
MR
122void av_aes_crypt(AVAES *a, uint8_t *dst_, const uint8_t *src_,
123 int count, uint8_t *iv_, int decrypt)
124{
125 av_aes_block *dst = (av_aes_block *) dst_;
126 const av_aes_block *src = (const av_aes_block *) src_;
127 av_aes_block *iv = (av_aes_block *) iv_;
128
129 while (count--) {
dd025f25 130 addkey(&a->state[1], src, &a->round_key[a->rounds]);
4bfe0644 131 if (decrypt) {
97e3458c 132 crypt(a, 0, inv_sbox, dec_multbl);
4bfe0644 133 if (iv) {
dd025f25 134 addkey(&a->state[0], &a->state[0], iv);
af2ea724 135 *iv = *src;
97e3458c 136 }
dd025f25 137 addkey(dst, &a->state[0], &a->round_key[0]);
4bfe0644
MR
138 } else {
139 if (iv)
140 addkey(&a->state[1], &a->state[1], iv);
141 crypt(a, 2, sbox, enc_multbl);
dd025f25 142 addkey(dst, &a->state[0], &a->round_key[0]);
4bfe0644 143 if (iv)
af2ea724 144 *iv = *dst;
97e3458c 145 }
dd025f25
RD
146 src++;
147 dst++;
97e3458c 148 }
ff8f6594
MN
149}
150
5d20f19b 151static void init_multbl2(uint32_t tbl[][256], const int c[4],
4bfe0644
MR
152 const uint8_t *log8, const uint8_t *alog8,
153 const uint8_t *sbox)
154{
5d20f19b
MR
155 int i;
156
157 for (i = 0; i < 256; i++) {
158 int x = sbox[i];
159 if (x) {
160 int k, l, m, n;
161 x = log8[x];
162 k = alog8[x + log8[c[0]]];
163 l = alog8[x + log8[c[1]]];
164 m = alog8[x + log8[c[2]]];
165 n = alog8[x + log8[c[3]]];
166 tbl[0][i] = AV_NE(MKBETAG(k,l,m,n), MKTAG(k,l,m,n));
b250f9c6 167#if !CONFIG_SMALL
5d20f19b
MR
168 tbl[1][i] = ROT(tbl[0][i], 8);
169 tbl[2][i] = ROT(tbl[0][i], 16);
170 tbl[3][i] = ROT(tbl[0][i], 24);
7200f5d0 171#endif
5d20f19b
MR
172 }
173 }
0458e790
MN
174}
175
3306dfd5 176// this is based on the reference AES code by Paulo Barreto and Vincent Rijmen
4bfe0644
MR
177int av_aes_init(AVAES *a, const uint8_t *key, int key_bits, int decrypt)
178{
3306dfd5
MN
179 int i, j, t, rconpointer = 0;
180 uint8_t tk[8][4];
4bfe0644
MR
181 int KC = key_bits >> 5;
182 int rounds = KC + 6;
183 uint8_t log8[256];
cd0968c4 184 uint8_t alog8[512];
3306dfd5 185
4bfe0644
MR
186 if (!enc_multbl[FF_ARRAY_ELEMS(enc_multbl)-1][FF_ARRAY_ELEMS(enc_multbl[0])-1]) {
187 j = 1;
188 for (i = 0; i < 255; i++) {
189 alog8[i] = alog8[i + 255] = j;
190 log8[j] = i;
191 j ^= j + j;
192 if (j > 255)
193 j ^= 0x11B;
3306dfd5 194 }
4bfe0644
MR
195 for (i = 0; i < 256; i++) {
196 j = i ? alog8[255 - log8[i]] : 0;
197 j ^= (j << 1) ^ (j << 2) ^ (j << 3) ^ (j << 4);
198 j = (j ^ (j >> 8) ^ 99) & 255;
199 inv_sbox[j] = i;
200 sbox[i] = j;
3306dfd5 201 }
5d20f19b 202 init_multbl2(dec_multbl, (const int[4]) { 0xe, 0x9, 0xd, 0xb },
4bfe0644 203 log8, alog8, inv_sbox);
5d20f19b 204 init_multbl2(enc_multbl, (const int[4]) { 0x2, 0x1, 0x1, 0x3 },
4bfe0644 205 log8, alog8, sbox);
3306dfd5
MN
206 }
207
4bfe0644 208 if (key_bits != 128 && key_bits != 192 && key_bits != 256)
6572e1a4 209 return -1;
3306dfd5 210
4bfe0644 211 a->rounds = rounds;
eade86e2 212
4bfe0644 213 memcpy(tk, key, KC * 4);
3306dfd5 214
4bfe0644
MR
215 for (t = 0; t < (rounds + 1) * 16;) {
216 memcpy(a->round_key[0].u8 + t, tk, KC * 4);
217 t += KC * 4;
3306dfd5 218
4bfe0644
MR
219 for (i = 0; i < 4; i++)
220 tk[0][i] ^= sbox[tk[KC - 1][(i + 1) & 3]];
3306dfd5
MN
221 tk[0][0] ^= rcon[rconpointer++];
222
4bfe0644
MR
223 for (j = 1; j < KC; j++) {
224 if (KC != 8 || j != KC >> 1)
225 for (i = 0; i < 4; i++)
226 tk[j][i] ^= tk[j - 1][i];
3306dfd5 227 else
4bfe0644
MR
228 for (i = 0; i < 4; i++)
229 tk[j][i] ^= sbox[tk[j - 1][i]];
3306dfd5
MN
230 }
231 }
f12cf3d2 232
4bfe0644
MR
233 if (decrypt) {
234 for (i = 1; i < rounds; i++) {
dd025f25 235 av_aes_block tmp[3];
af2ea724 236 tmp[2] = a->round_key[i];
2ad4a3bc 237 subshift(&tmp[1], 0, sbox);
298dd6b7 238 mix(tmp, dec_multbl, 1, 3);
af2ea724 239 a->round_key[i] = tmp[0];
28b51235 240 }
4bfe0644
MR
241 } else {
242 for (i = 0; i < (rounds + 1) >> 1; i++) {
af2ea724 243 FFSWAP(av_aes_block, a->round_key[i], a->round_key[rounds-i]);
f25b650a 244 }
cc858970 245 }
f12cf3d2 246
6572e1a4 247 return 0;
3306dfd5
MN
248}
249
250#ifdef TEST
744765a9 251#include <string.h>
294eaa26 252#include "lfg.h"
0d80bd2f 253#include "log.h"
3306dfd5 254
744765a9
MR
255int main(int argc, char **argv)
256{
4bfe0644 257 int i, j;
744765a9 258 AVAES b;
4bfe0644
MR
259 uint8_t rkey[2][16] = {
260 { 0 },
261 { 0x10, 0xa5, 0x88, 0x69, 0xd7, 0x4b, 0xe5, 0xa3,
262 0x74, 0xcf, 0x86, 0x7c, 0xfb, 0x47, 0x38, 0x59 }
263 };
d65dc0cd 264 uint8_t pt[16], rpt[2][16]= {
4bfe0644
MR
265 { 0x6a, 0x84, 0x86, 0x7c, 0xd7, 0x7e, 0x12, 0xad,
266 0x07, 0xea, 0x1b, 0xe8, 0x95, 0xc5, 0x3f, 0xa3 },
267 { 0 }
268 };
d65dc0cd 269 uint8_t rct[2][16]= {
4bfe0644
MR
270 { 0x73, 0x22, 0x81, 0xc0, 0xa0, 0xaa, 0xb8, 0xf7,
271 0xa5, 0x4a, 0x0c, 0x67, 0xa0, 0xc4, 0x5e, 0xcf },
272 { 0x6d, 0x25, 0x1e, 0x69, 0x44, 0xb0, 0x51, 0xe0,
273 0x4e, 0xaa, 0x6f, 0xb4, 0xdb, 0xf7, 0x84, 0x65 }
274 };
7465619a 275 uint8_t temp[16];
744765a9 276 int err = 0;
3306dfd5 277
b56e34c5 278 av_log_set_level(AV_LOG_DEBUG);
cf5773e9 279
4bfe0644 280 for (i = 0; i < 2; i++) {
6572e1a4 281 av_aes_init(&b, rkey[i], 128, 1);
ca2560df 282 av_aes_crypt(&b, temp, rct[i], 1, NULL, 1);
4bfe0644
MR
283 for (j = 0; j < 16; j++) {
284 if (rpt[i][j] != temp[j]) {
285 av_log(NULL, AV_LOG_ERROR, "%d %02X %02X\n",
286 j, rpt[i][j], temp[j]);
744765a9
MR
287 err = 1;
288 }
4bfe0644 289 }
d65dc0cd 290 }
0458e790 291
744765a9
MR
292 if (argc > 1 && !strcmp(argv[1], "-t")) {
293 AVAES ae, ad;
294 AVLFG prng;
295
296 av_aes_init(&ae, "PI=3.141592654..", 128, 0);
297 av_aes_init(&ad, "PI=3.141592654..", 128, 1);
298 av_lfg_init(&prng, 1);
299
4bfe0644
MR
300 for (i = 0; i < 10000; i++) {
301 for (j = 0; j < 16; j++) {
744765a9
MR
302 pt[j] = av_lfg_get(&prng);
303 }
304 {
305 START_TIMER;
306 av_aes_crypt(&ae, temp, pt, 1, NULL, 0);
4bfe0644
MR
307 if (!(i & (i - 1)))
308 av_log(NULL, AV_LOG_ERROR, "%02X %02X %02X %02X\n",
309 temp[0], temp[5], temp[10], temp[15]);
744765a9
MR
310 av_aes_crypt(&ad, temp, temp, 1, NULL, 1);
311 STOP_TIMER("aes");
312 }
4bfe0644
MR
313 for (j = 0; j < 16; j++) {
314 if (pt[j] != temp[j]) {
315 av_log(NULL, AV_LOG_ERROR, "%d %d %02X %02X\n",
316 i, j, pt[j], temp[j]);
744765a9 317 }
ff8f6594
MN
318 }
319 }
3306dfd5 320 }
744765a9 321 return err;
3306dfd5
MN
322}
323#endif