Mercurial > libavutil.hg
annotate aes.c @ 197:a0010ccdc817 libavutil
revert simplification (broke 192 bit keys)
| author | michael |
|---|---|
| date | Sun, 14 Jan 2007 21:13:14 +0000 |
| parents | b1ad10c8f362 |
| children | 2677cc2d1ff7 |
| rev | line source |
|---|---|
| 164 | 1 /* |
|
194
3d5efe2de0ae
seems i didnt copy anything from aes128.c so this should be more correct ?
michael
parents:
193
diff
changeset
|
2 * copyright (c) 2007 Michael Niedermayer <michaelni@gmx.at> |
| 164 | 3 * |
| 4 * This file is part of FFmpeg. | |
| 5 * | |
| 6 * FFmpeg is free software; you can redistribute it and/or | |
| 7 * modify it under the terms of the GNU Lesser General Public | |
| 8 * License as published by the Free Software Foundation; either | |
| 9 * version 2.1 of the License, or (at your option) any later version. | |
| 10 * | |
| 11 * FFmpeg is distributed in the hope that it will be useful, | |
| 12 * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
| 13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | |
| 14 * Lesser General Public License for more details. | |
| 15 * | |
| 16 * You should have received a copy of the GNU Lesser General Public | |
| 17 * License along with FFmpeg; if not, write to the Free Software | |
| 18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA | |
|
194
3d5efe2de0ae
seems i didnt copy anything from aes128.c so this should be more correct ?
michael
parents:
193
diff
changeset
|
19 * |
|
3d5efe2de0ae
seems i didnt copy anything from aes128.c so this should be more correct ?
michael
parents:
193
diff
changeset
|
20 * some optimization ideas from aes128.c by Reimar Doeffinger |
| 164 | 21 */ |
| 22 | |
| 23 #include "common.h" | |
| 24 #include "log.h" | |
| 25 #include "aes.h" | |
| 26 | |
| 27 typedef struct AVAES{ | |
| 189 | 28 uint8_t round_key[15][4][4]; |
| 164 | 29 uint8_t state[4][4]; |
| 30 int rounds; | |
| 31 }AVAES; | |
| 32 | |
| 192 | 33 static const uint8_t rcon[10] = { |
| 34 0x01, 0x02, 0x04, 0x08, 0x10, 0x20, 0x40, 0x80, 0x1b, 0x36 | |
| 164 | 35 }; |
| 36 | |
| 37 static uint8_t sbox[256]; | |
| 38 static uint8_t inv_sbox[256]; | |
| 169 | 39 #ifdef CONFIG_SMALL |
| 40 static uint32_t enc_multbl[1][256]; | |
| 41 static uint32_t dec_multbl[1][256]; | |
| 42 #else | |
| 43 static uint32_t enc_multbl[4][256]; | |
| 44 static uint32_t dec_multbl[4][256]; | |
| 45 #endif | |
| 164 | 46 |
| 168 | 47 static inline void addkey(uint64_t state[2], uint64_t round_key[2]){ |
| 48 state[0] ^= round_key[0]; | |
| 49 state[1] ^= round_key[1]; | |
| 164 | 50 } |
| 51 | |
| 52 #define SUBSHIFT0(s, box) s[0]=box[s[ 0]]; s[ 4]=box[s[ 4]]; s[ 8]=box[s[ 8]]; s[12]=box[s[12]]; | |
| 53 #define SUBSHIFT1(s, box) t=s[0]; s[0]=box[s[ 4]]; s[ 4]=box[s[ 8]]; s[ 8]=box[s[12]]; s[12]=box[t]; | |
| 54 #define SUBSHIFT2(s, box) t=s[0]; s[0]=box[s[ 8]]; s[ 8]=box[ t]; t=s[ 4]; s[ 4]=box[s[12]]; s[12]=box[t]; | |
| 55 #define SUBSHIFT3(s, box) t=s[0]; s[0]=box[s[12]]; s[12]=box[s[ 8]]; s[ 8]=box[s[ 4]]; s[ 4]=box[t]; | |
| 56 | |
| 185 | 57 #define SUBSHIFT1x(s) t=s[0]; s[0]=s[ 4]; s[ 4]=s[ 8]; s[ 8]=s[12]; s[12]=t; |
| 58 #define SUBSHIFT2x(s) t=s[0]; s[0]=s[ 8]; s[ 8]= t; t=s[ 4]; s[ 4]=s[12]; s[12]=t; | |
| 59 #define SUBSHIFT3x(s) t=s[0]; s[0]=s[12]; s[12]=s[ 8]; s[ 8]=s[ 4]; s[ 4]=t; | |
| 183 | 60 |
| 181 | 61 #define ROT(x,s) ((x<<s)|(x>>(32-s))) |
| 164 | 62 |
| 169 | 63 static inline void mix(uint8_t state[4][4], uint32_t multbl[4][256]){ |
| 64 int i; | |
| 65 for(i=0; i<4; i++) | |
| 66 #ifdef CONFIG_SMALL | |
| 67 ((uint32_t *)(state))[i] = multbl[0][state[i][0]] ^ ROT(multbl[0][state[i][1]], 8) | |
| 68 ^ROT(multbl[0][state[i][2]],16) ^ ROT(multbl[0][state[i][3]],24); | |
| 69 | |
| 70 #else | |
| 71 ((uint32_t *)(state))[i] = multbl[0][state[i][0]] ^ multbl[1][state[i][1]] | |
| 72 ^multbl[2][state[i][2]] ^ multbl[3][state[i][3]]; | |
| 73 #endif | |
| 164 | 74 } |
| 75 | |
|
191
a2a3c80706e5
merge encrypt and decrypt so the source is simpler and the compiler can choose with inlining if it wants speed or small size
michael
parents:
190
diff
changeset
|
76 static inline void crypt(AVAES *a, int s, uint8_t *sbox, uint32_t *multbl){ |
| 164 | 77 int t, r; |
| 78 | |
| 188 | 79 for(r=a->rounds; r>1; r--){ |
| 189 | 80 addkey(a->state, a->round_key[r]); |
|
191
a2a3c80706e5
merge encrypt and decrypt so the source is simpler and the compiler can choose with inlining if it wants speed or small size
michael
parents:
190
diff
changeset
|
81 SUBSHIFT3x((a->state[0]+1+s)) |
| 185 | 82 SUBSHIFT2x((a->state[0]+2)) |
|
191
a2a3c80706e5
merge encrypt and decrypt so the source is simpler and the compiler can choose with inlining if it wants speed or small size
michael
parents:
190
diff
changeset
|
83 SUBSHIFT1x((a->state[0]+3-s)) |
|
a2a3c80706e5
merge encrypt and decrypt so the source is simpler and the compiler can choose with inlining if it wants speed or small size
michael
parents:
190
diff
changeset
|
84 mix(a->state, multbl); |
| 164 | 85 } |
| 189 | 86 addkey(a->state, a->round_key[1]); |
|
191
a2a3c80706e5
merge encrypt and decrypt so the source is simpler and the compiler can choose with inlining if it wants speed or small size
michael
parents:
190
diff
changeset
|
87 SUBSHIFT0((a->state[0]+0 ), sbox) |
|
a2a3c80706e5
merge encrypt and decrypt so the source is simpler and the compiler can choose with inlining if it wants speed or small size
michael
parents:
190
diff
changeset
|
88 SUBSHIFT3((a->state[0]+1+s), sbox) |
|
a2a3c80706e5
merge encrypt and decrypt so the source is simpler and the compiler can choose with inlining if it wants speed or small size
michael
parents:
190
diff
changeset
|
89 SUBSHIFT2((a->state[0]+2 ), sbox) |
|
a2a3c80706e5
merge encrypt and decrypt so the source is simpler and the compiler can choose with inlining if it wants speed or small size
michael
parents:
190
diff
changeset
|
90 SUBSHIFT1((a->state[0]+3-s), sbox) |
| 189 | 91 addkey(a->state, a->round_key[0]); |
| 164 | 92 } |
| 93 | |
|
191
a2a3c80706e5
merge encrypt and decrypt so the source is simpler and the compiler can choose with inlining if it wants speed or small size
michael
parents:
190
diff
changeset
|
94 void av_aes_decrypt(AVAES *a){ |
|
a2a3c80706e5
merge encrypt and decrypt so the source is simpler and the compiler can choose with inlining if it wants speed or small size
michael
parents:
190
diff
changeset
|
95 crypt(a, 0, inv_sbox, dec_multbl); |
|
a2a3c80706e5
merge encrypt and decrypt so the source is simpler and the compiler can choose with inlining if it wants speed or small size
michael
parents:
190
diff
changeset
|
96 } |
|
a2a3c80706e5
merge encrypt and decrypt so the source is simpler and the compiler can choose with inlining if it wants speed or small size
michael
parents:
190
diff
changeset
|
97 |
| 164 | 98 void av_aes_encrypt(AVAES *a){ |
|
191
a2a3c80706e5
merge encrypt and decrypt so the source is simpler and the compiler can choose with inlining if it wants speed or small size
michael
parents:
190
diff
changeset
|
99 crypt(a, 2, sbox, enc_multbl); |
| 178 | 100 } |
| 101 | |
| 195 | 102 static void init_multbl2(uint8_t tbl[1024], int c[4], uint8_t *log8, uint8_t *alog8, uint8_t *sbox){ |
| 183 | 103 int i; |
| 104 for(i=0; i<1024; i++){ | |
| 105 int x= sbox[i/4]; | |
| 106 if(x) tbl[i]= alog8[ log8[x] + log8[c[i&3]] ]; | |
| 107 } | |
| 108 } | |
| 109 | |
| 110 | |
| 164 | 111 // this is based on the reference AES code by Paulo Barreto and Vincent Rijmen |
| 189 | 112 AVAES *av_aes_init(uint8_t *key, int key_bits, int decrypt) { |
| 166 | 113 AVAES *a; |
| 164 | 114 int i, j, t, rconpointer = 0; |
| 115 uint8_t tk[8][4]; | |
| 167 | 116 int KC= key_bits/32; |
| 117 int rounds= KC + 6; | |
| 169 | 118 uint8_t log8[256]; |
| 119 uint8_t alog8[512]; | |
| 164 | 120 |
| 121 if(!sbox[255]){ | |
| 122 j=1; | |
| 123 for(i=0; i<255; i++){ | |
| 124 alog8[i]= | |
| 125 alog8[i+255]= j; | |
| 126 log8[j]= i; | |
| 127 j^= j+j; | |
| 128 if(j>255) j^= 0x11B; | |
| 129 } | |
| 130 for(i=0; i<256; i++){ | |
| 131 j= i ? alog8[255-log8[i]] : 0; | |
| 165 | 132 j ^= (j<<1) ^ (j<<2) ^ (j<<3) ^ (j<<4); |
| 133 j = (j ^ (j>>8) ^ 99) & 255; | |
| 164 | 134 inv_sbox[j]= i; |
| 135 sbox [i]= j; | |
| 136 // av_log(NULL, AV_LOG_ERROR, "%d, ", log8[i]); | |
| 137 } | |
| 184 | 138 init_multbl2(dec_multbl[0], (int[4]){0xe, 0x9, 0xd, 0xb}, log8, alog8, inv_sbox); |
| 181 | 139 #ifndef CONFIG_SMALL |
| 184 | 140 init_multbl2(dec_multbl[1], (int[4]){0xb, 0xe, 0x9, 0xd}, log8, alog8, inv_sbox); |
| 141 init_multbl2(dec_multbl[2], (int[4]){0xd, 0xb, 0xe, 0x9}, log8, alog8, inv_sbox); | |
| 142 init_multbl2(dec_multbl[3], (int[4]){0x9, 0xd, 0xb, 0xe}, log8, alog8, inv_sbox); | |
| 181 | 143 #endif |
| 183 | 144 init_multbl2(enc_multbl[0], (int[4]){0x2, 0x1, 0x1, 0x3}, log8, alog8, sbox); |
| 181 | 145 #ifndef CONFIG_SMALL |
| 183 | 146 init_multbl2(enc_multbl[1], (int[4]){0x3, 0x2, 0x1, 0x1}, log8, alog8, sbox); |
| 147 init_multbl2(enc_multbl[2], (int[4]){0x1, 0x3, 0x2, 0x1}, log8, alog8, sbox); | |
| 148 init_multbl2(enc_multbl[3], (int[4]){0x1, 0x1, 0x3, 0x2}, log8, alog8, sbox); | |
| 181 | 149 #endif |
| 164 | 150 } |
| 151 | |
| 167 | 152 if(key_bits!=128 && key_bits!=192 && key_bits!=256) |
| 164 | 153 return NULL; |
| 154 | |
| 166 | 155 a= av_malloc(sizeof(AVAES)); |
| 167 | 156 a->rounds= rounds; |
| 166 | 157 |
| 164 | 158 memcpy(tk, key, KC*4); |
| 159 | |
|
174
263bbdc10c1e
simplify round_key generation by writing over the end but ensuring that theres some irrelevant stuff afterwards
michael
parents:
173
diff
changeset
|
160 for(t= 0; t < (rounds+1)*4;) { |
| 197 | 161 memcpy(a->round_key[0][t], tk, KC*4); |
|
174
263bbdc10c1e
simplify round_key generation by writing over the end but ensuring that theres some irrelevant stuff afterwards
michael
parents:
173
diff
changeset
|
162 t+= KC; |
| 164 | 163 |
| 164 for(i = 0; i < 4; i++) | |
| 173 | 165 tk[0][i] ^= sbox[tk[KC-1][(i+1)&3]]; |
| 164 | 166 tk[0][0] ^= rcon[rconpointer++]; |
| 167 | |
| 168 for(j = 1; j < KC; j++){ | |
| 169 if(KC != 8 || j != KC/2) | |
| 175 | 170 for(i = 0; i < 4; i++) tk[j][i] ^= tk[j-1][i]; |
| 164 | 171 else |
| 175 | 172 for(i = 0; i < 4; i++) tk[j][i] ^= sbox[tk[j-1][i]]; |
| 164 | 173 } |
| 174 } | |
| 184 | 175 |
| 189 | 176 if(decrypt){ |
| 190 | 177 for(i=1; i<rounds; i++){ |
| 178 for(j=0; j<16; j++) | |
| 179 a->round_key[i][0][j]= sbox[a->round_key[i][0][j]]; | |
| 180 mix(a->round_key[i], dec_multbl); | |
| 181 } | |
| 197 | 182 }else{ |
| 183 for(i=0; i<(rounds+1)/2; i++){ | |
| 184 for(j=0; j<16; j++) | |
| 185 FFSWAP(int, a->round_key[i][0][j], a->round_key[rounds-i][0][j]); | |
| 186 } | |
| 187 | 187 } |
| 184 | 188 |
| 164 | 189 return a; |
| 190 } | |
| 191 | |
| 192 #ifdef TEST | |
| 193 | |
| 194 int main(){ | |
| 195 | 195 int i,j; |
| 189 | 196 AVAES *ae= av_aes_init("PI=3.141592654..", 128, 0); |
| 197 AVAES *ad= av_aes_init("PI=3.141592654..", 128, 1); | |
| 182 | 198 uint8_t zero[16]= {0}; |
| 199 uint8_t pt[16]= {0x6a, 0x84, 0x86, 0x7c, 0xd7, 0x7e, 0x12, 0xad, 0x07, 0xea, 0x1b, 0xe8, 0x95, 0xc5, 0x3f, 0xa3}; | |
| 200 uint8_t ct[16]= {0x73, 0x22, 0x81, 0xc0, 0xa0, 0xaa, 0xb8, 0xf7, 0xa5, 0x4a, 0x0c, 0x67, 0xa0, 0xc4, 0x5e, 0xcf}; | |
| 189 | 201 AVAES *b= av_aes_init(zero, 128, 1); |
| 183 | 202 |
| 203 /* uint8_t key[16]= {0x42, 0x78, 0xb8, 0x40, 0xfb, 0x44, 0xaa, 0xa7, 0x57, 0xc1, 0xbf, 0x04, 0xac, 0xbe, 0x1a, 0x3e}; | |
| 204 uint8_t IV[16] = {0x57, 0xf0, 0x2a, 0x5c, 0x53, 0x39, 0xda, 0xeb, 0x0a, 0x29, 0x08, 0xa0, 0x6a, 0xc6, 0x39, 0x3f}; | |
| 205 uint8_t pt[16] = {0x3c, 0x88, 0x8b, 0xbb, 0xb1, 0xa8, 0xeb, 0x9f, 0x3e, 0x9b, 0x87, 0xac, 0xaa, 0xd9, 0x86, 0xc4}; | |
| 206 // 66e2f7071c83083b8a557971918850e5 | |
| 207 uint8_t ct[16] = {0x47, 0x9c, 0x89, 0xec, 0x14, 0xbc, 0x98, 0x99, 0x4e, 0x62, 0xb2, 0xc7, 0x05, 0xb5, 0x0, 0x14e}; | |
| 208 // 175bd7832e7e60a1e92aac568a861eb7*/ | |
| 209 uint8_t ckey[16]= {0x10, 0xa5, 0x88, 0x69, 0xd7, 0x4b, 0xe5, 0xa3, 0x74, 0xcf, 0x86, 0x7c, 0xfb, 0x47, 0x38, 0x59}; | |
| 210 uint8_t cct[16] = {0x6d, 0x25, 0x1e, 0x69, 0x44, 0xb0, 0x51, 0xe0, 0x4e, 0xaa, 0x6f, 0xb4, 0xdb, 0xf7, 0x84, 0x65}; | |
| 189 | 211 AVAES *c= av_aes_init(ckey, 128, 1); |
| 164 | 212 |
| 179 | 213 av_log_level= AV_LOG_DEBUG; |
| 214 | |
| 182 | 215 memcpy(b->state, ct, 16); |
| 216 av_aes_decrypt(b); | |
| 217 for(j=0; j<16; j++) | |
| 218 if(pt[j] != b->state[0][j]){ | |
| 219 av_log(NULL, AV_LOG_ERROR, "%d %02X %02X\n", j, pt[j], b->state[0][j]); | |
| 220 } | |
| 221 | |
| 183 | 222 memcpy(c->state, cct, 16); |
| 223 av_aes_decrypt(c); | |
| 224 for(j=0; j<16; j++) | |
| 225 if(zero[j] != c->state[0][j]){ | |
| 226 av_log(NULL, AV_LOG_ERROR, "%d %02X %02X\n", j, zero[j], c->state[0][j]); | |
| 227 } | |
| 228 | |
| 164 | 229 for(i=0; i<10000; i++){ |
| 178 | 230 for(j=0; j<16; j++){ |
| 231 pt[j]= random(); | |
| 232 } | |
| 189 | 233 memcpy(ae->state, pt, 16); |
| 179 | 234 {START_TIMER |
| 189 | 235 av_aes_encrypt(ae); |
| 178 | 236 if(!(i&(i-1))) |
| 189 | 237 av_log(NULL, AV_LOG_ERROR, "%02X %02X %02X %02X\n", ae->state[0][0], ae->state[1][1], ae->state[2][2], ae->state[3][3]); |
| 238 memcpy(ad->state, ae->state, 16); | |
| 239 av_aes_decrypt(ad); | |
| 179 | 240 STOP_TIMER("aes")} |
| 178 | 241 for(j=0; j<16; j++){ |
| 189 | 242 if(pt[j] != ad->state[0][j]){ |
| 243 av_log(NULL, AV_LOG_ERROR, "%d %d %02X %02X\n", i,j, pt[j], ad->state[0][j]); | |
| 178 | 244 } |
| 245 } | |
| 164 | 246 } |
| 247 return 0; | |
| 248 } | |
| 249 #endif |
