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 Libav.
8 * Libav 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 * Libav 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 Libav; if not, write to the Free Software
20 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
33 typedef struct AVAES {
34 // Note: round_key[16] is accessed in the init code, but this only
35 // overwrites state, which does not matter (see also r7471).
36 av_aes_block round_key[15];
37 av_aes_block state[2];
41 const int av_aes_size= sizeof(AVAES);
43 static const uint8_t rcon[10] = {
44 0x01, 0x02, 0x04, 0x08, 0x10, 0x20, 0x40, 0x80, 0x1b, 0x36
47 static uint8_t sbox[256];
48 static uint8_t inv_sbox[256];
50 static uint32_t enc_multbl[1][256];
51 static uint32_t dec_multbl[1][256];
53 static uint32_t enc_multbl[4][256];
54 static uint32_t dec_multbl[4][256];
57 static inline void addkey(av_aes_block *dst, const av_aes_block *src,
58 const av_aes_block *round_key)
60 dst->u64[0] = src->u64[0] ^ round_key->u64[0];
61 dst->u64[1] = src->u64[1] ^ round_key->u64[1];
64 static void subshift(av_aes_block s0[2], int s, const uint8_t *box)
66 av_aes_block *s1 = (av_aes_block *) (s0[0].u8 - s);
67 av_aes_block *s3 = (av_aes_block *) (s0[0].u8 + s);
69 s0[0].u8[ 0] = box[s0[1].u8[ 0]];
70 s0[0].u8[ 4] = box[s0[1].u8[ 4]];
71 s0[0].u8[ 8] = box[s0[1].u8[ 8]];
72 s0[0].u8[12] = box[s0[1].u8[12]];
73 s1[0].u8[ 3] = box[s1[1].u8[ 7]];
74 s1[0].u8[ 7] = box[s1[1].u8[11]];
75 s1[0].u8[11] = box[s1[1].u8[15]];
76 s1[0].u8[15] = box[s1[1].u8[ 3]];
77 s0[0].u8[ 2] = box[s0[1].u8[10]];
78 s0[0].u8[10] = box[s0[1].u8[ 2]];
79 s0[0].u8[ 6] = box[s0[1].u8[14]];
80 s0[0].u8[14] = box[s0[1].u8[ 6]];
81 s3[0].u8[ 1] = box[s3[1].u8[13]];
82 s3[0].u8[13] = box[s3[1].u8[ 9]];
83 s3[0].u8[ 9] = box[s3[1].u8[ 5]];
84 s3[0].u8[ 5] = box[s3[1].u8[ 1]];
87 static inline int mix_core(uint32_t multbl[][256], int a, int b, int c, int d){
89 #define ROT(x,s) ((x<<s)|(x>>(32-s)))
90 return multbl[0][a] ^ ROT(multbl[0][b], 8) ^ ROT(multbl[0][c], 16) ^ ROT(multbl[0][d], 24);
92 return multbl[0][a] ^ multbl[1][b] ^ multbl[2][c] ^ multbl[3][d];
96 static inline void mix(av_aes_block state[2], uint32_t multbl[][256], int s1, int s3){
97 uint8_t (*src)[4] = state[1].u8x4;
98 state[0].u32[0] = mix_core(multbl, src[0][0], src[s1 ][1], src[2][2], src[s3 ][3]);
99 state[0].u32[1] = mix_core(multbl, src[1][0], src[s3-1][1], src[3][2], src[s1-1][3]);
100 state[0].u32[2] = mix_core(multbl, src[2][0], src[s3 ][1], src[0][2], src[s1 ][3]);
101 state[0].u32[3] = mix_core(multbl, src[3][0], src[s1-1][1], src[1][2], src[s3-1][3]);
104 static inline void crypt(AVAES *a, int s, const uint8_t *sbox,
105 uint32_t multbl[][256])
109 for (r = a->rounds - 1; r > 0; r--) {
110 mix(a->state, multbl, 3 - s, 1 + s);
111 addkey(&a->state[1], &a->state[0], &a->round_key[r]);
114 subshift(&a->state[0], s, sbox);
117 void av_aes_crypt(AVAES *a, uint8_t *dst_, const uint8_t *src_,
118 int count, uint8_t *iv_, int decrypt)
120 av_aes_block *dst = (av_aes_block *) dst_;
121 const av_aes_block *src = (const av_aes_block *) src_;
122 av_aes_block *iv = (av_aes_block *) iv_;
125 addkey(&a->state[1], src, &a->round_key[a->rounds]);
127 crypt(a, 0, inv_sbox, dec_multbl);
129 addkey(&a->state[0], &a->state[0], iv);
132 addkey(dst, &a->state[0], &a->round_key[0]);
135 addkey(&a->state[1], &a->state[1], iv);
136 crypt(a, 2, sbox, enc_multbl);
137 addkey(dst, &a->state[0], &a->round_key[0]);
146 static void init_multbl2(uint8_t tbl[1024], const int c[4],
147 const uint8_t *log8, const uint8_t *alog8,
152 for (i = 0; i < 1024; i++) {
153 int x = sbox[i >> 2];
155 tbl[i] = alog8[log8[x] + log8[c[i & 3]]];
158 for (j = 256; j < 1024; j++)
159 for (i = 0; i < 4; i++)
160 tbl[4*j + i] = tbl[4*j + ((i - 1) & 3) - 1024];
164 // this is based on the reference AES code by Paulo Barreto and Vincent Rijmen
165 int av_aes_init(AVAES *a, const uint8_t *key, int key_bits, int decrypt)
167 int i, j, t, rconpointer = 0;
169 int KC = key_bits >> 5;
174 if (!enc_multbl[FF_ARRAY_ELEMS(enc_multbl)-1][FF_ARRAY_ELEMS(enc_multbl[0])-1]) {
176 for (i = 0; i < 255; i++) {
177 alog8[i] = alog8[i + 255] = j;
183 for (i = 0; i < 256; i++) {
184 j = i ? alog8[255 - log8[i]] : 0;
185 j ^= (j << 1) ^ (j << 2) ^ (j << 3) ^ (j << 4);
186 j = (j ^ (j >> 8) ^ 99) & 255;
190 init_multbl2(dec_multbl[0], (const int[4]) { 0xe, 0x9, 0xd, 0xb },
191 log8, alog8, inv_sbox);
192 init_multbl2(enc_multbl[0], (const int[4]) { 0x2, 0x1, 0x1, 0x3 },
196 if (key_bits != 128 && key_bits != 192 && key_bits != 256)
201 memcpy(tk, key, KC * 4);
203 for (t = 0; t < (rounds + 1) * 16;) {
204 memcpy(a->round_key[0].u8 + t, tk, 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++];
211 for (j = 1; j < KC; j++) {
212 if (KC != 8 || j != KC >> 1)
213 for (i = 0; i < 4; i++)
214 tk[j][i] ^= tk[j - 1][i];
216 for (i = 0; i < 4; i++)
217 tk[j][i] ^= sbox[tk[j - 1][i]];
222 for (i = 1; i < rounds; i++) {
224 memcpy(&tmp[2], &a->round_key[i], 16);
225 subshift(&tmp[1], 0, sbox);
226 mix(tmp, dec_multbl, 1, 3);
227 memcpy(&a->round_key[i], &tmp[0], 16);
230 for (i = 0; i < (rounds + 1) >> 1; i++) {
231 for (j = 0; j < 16; j++)
232 FFSWAP(int, a->round_key[i].u8[j], a->round_key[rounds-i].u8[j]);
244 int main(int argc, char **argv)
248 uint8_t rkey[2][16] = {
250 { 0x10, 0xa5, 0x88, 0x69, 0xd7, 0x4b, 0xe5, 0xa3,
251 0x74, 0xcf, 0x86, 0x7c, 0xfb, 0x47, 0x38, 0x59 }
253 uint8_t pt[16], rpt[2][16]= {
254 { 0x6a, 0x84, 0x86, 0x7c, 0xd7, 0x7e, 0x12, 0xad,
255 0x07, 0xea, 0x1b, 0xe8, 0x95, 0xc5, 0x3f, 0xa3 },
258 uint8_t rct[2][16]= {
259 { 0x73, 0x22, 0x81, 0xc0, 0xa0, 0xaa, 0xb8, 0xf7,
260 0xa5, 0x4a, 0x0c, 0x67, 0xa0, 0xc4, 0x5e, 0xcf },
261 { 0x6d, 0x25, 0x1e, 0x69, 0x44, 0xb0, 0x51, 0xe0,
262 0x4e, 0xaa, 0x6f, 0xb4, 0xdb, 0xf7, 0x84, 0x65 }
267 av_log_set_level(AV_LOG_DEBUG);
269 for (i = 0; i < 2; i++) {
270 av_aes_init(&b, rkey[i], 128, 1);
271 av_aes_crypt(&b, temp, rct[i], 1, NULL, 1);
272 for (j = 0; j < 16; j++) {
273 if (rpt[i][j] != temp[j]) {
274 av_log(NULL, AV_LOG_ERROR, "%d %02X %02X\n",
275 j, rpt[i][j], temp[j]);
281 if (argc > 1 && !strcmp(argv[1], "-t")) {
285 av_aes_init(&ae, "PI=3.141592654..", 128, 0);
286 av_aes_init(&ad, "PI=3.141592654..", 128, 1);
287 av_lfg_init(&prng, 1);
289 for (i = 0; i < 10000; i++) {
290 for (j = 0; j < 16; j++) {
291 pt[j] = av_lfg_get(&prng);
295 av_aes_crypt(&ae, temp, pt, 1, NULL, 0);
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, 1, NULL, 1);
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]);