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 23#include "common.h" 24#include "aes.h" 25 26typedef struct AVAES{ 27 // Note: round_key[16] is accessed in the init code, but this only 28 // overwrites state, which does not matter (see also r7471). 29 uint8_t round_key[15][4][4]; 30 uint8_t state[2][4][4]; 31 int rounds; 32}AVAES; 33 34const int av_aes_size= sizeof(AVAES); 35 36static const uint8_t rcon[10] = { 37 0x01, 0x02, 0x04, 0x08, 0x10, 0x20, 0x40, 0x80, 0x1b, 0x36 38}; 39 40static uint8_t sbox[256]; 41static uint8_t inv_sbox[256]; 42#if CONFIG_SMALL 43static uint32_t enc_multbl[1][256]; 44static uint32_t dec_multbl[1][256]; 45#else 46static uint32_t enc_multbl[4][256]; 47static uint32_t dec_multbl[4][256]; 48#endif 49 50static inline void addkey(uint64_t dst[2], const uint64_t src[2], const uint64_t round_key[2]){ 51 dst[0] = src[0] ^ round_key[0]; 52 dst[1] = src[1] ^ round_key[1]; 53} 54 55static void subshift(uint8_t s0[2][16], int s, const uint8_t *box){ 56 uint8_t (*s1)[16]= s0[0] - s; 57 uint8_t (*s3)[16]= s0[0] + s; 58 s0[0][0]=box[s0[1][ 0]]; s0[0][ 4]=box[s0[1][ 4]]; s0[0][ 8]=box[s0[1][ 8]]; s0[0][12]=box[s0[1][12]]; 59 s1[0][3]=box[s1[1][ 7]]; s1[0][ 7]=box[s1[1][11]]; s1[0][11]=box[s1[1][15]]; s1[0][15]=box[s1[1][ 3]]; 60 s0[0][2]=box[s0[1][10]]; s0[0][10]=box[s0[1][ 2]]; s0[0][ 6]=box[s0[1][14]]; s0[0][14]=box[s0[1][ 6]]; 61 s3[0][1]=box[s3[1][13]]; s3[0][13]=box[s3[1][ 9]]; s3[0][ 9]=box[s3[1][ 5]]; s3[0][ 5]=box[s3[1][ 1]]; 62} 63 64static inline int mix_core(uint32_t multbl[4][256], int a, int b, int c, int d){ 65#if CONFIG_SMALL 66#define ROT(x,s) ((x<<s)|(x>>(32-s))) 67 return multbl[0][a] ^ ROT(multbl[0][b], 8) ^ ROT(multbl[0][c], 16) ^ ROT(multbl[0][d], 24); 68#else 69 return multbl[0][a] ^ multbl[1][b] ^ multbl[2][c] ^ multbl[3][d]; 70#endif 71} 72 73static inline void mix(uint8_t state[2][4][4], uint32_t multbl[4][256], int s1, int s3){ 74 ((uint32_t *)(state))[0] = mix_core(multbl, state[1][0][0], state[1][s1 ][1], state[1][2][2], state[1][s3 ][3]); 75 ((uint32_t *)(state))[1] = mix_core(multbl, state[1][1][0], state[1][s3-1][1], state[1][3][2], state[1][s1-1][3]); 76 ((uint32_t *)(state))[2] = mix_core(multbl, state[1][2][0], state[1][s3 ][1], state[1][0][2], state[1][s1 ][3]); 77 ((uint32_t *)(state))[3] = mix_core(multbl, state[1][3][0], state[1][s1-1][1], state[1][1][2], state[1][s3-1][3]); 78} 79 80static inline void crypt(AVAES *a, int s, const uint8_t *sbox, const uint32_t *multbl){ 81 int r; 82 83 for(r=a->rounds-1; r>0; r--){ 84 mix(a->state, multbl, 3-s, 1+s); 85 addkey(a->state[1], a->state[0], a->round_key[r]); 86 } 87 subshift(a->state[0][0], s, sbox); 88} 89 90void av_aes_crypt(AVAES *a, uint8_t *dst, const uint8_t *src, int count, uint8_t *iv, int decrypt){ 91 while(count--){ 92 addkey(a->state[1], src, a->round_key[a->rounds]); 93 if(decrypt) { 94 crypt(a, 0, inv_sbox, dec_multbl); 95 if(iv){ 96 addkey(a->state[0], a->state[0], iv); 97 memcpy(iv, src, 16); 98 } 99 addkey(dst, a->state[0], a->round_key[0]); 100 }else{ 101 if(iv) addkey(a->state[1], a->state[1], iv); 102 crypt(a, 2, sbox, enc_multbl); 103 addkey(dst, a->state[0], a->round_key[0]); 104 if(iv) memcpy(iv, dst, 16); 105 } 106 src+=16; 107 dst+=16; 108 } 109} 110 111static void init_multbl2(uint8_t tbl[1024], const int c[4], const uint8_t *log8, const uint8_t *alog8, const uint8_t *sbox){ 112 int i, j; 113 for(i=0; i<1024; i++){ 114 int x= sbox[i>>2]; 115 if(x) tbl[i]= alog8[ log8[x] + log8[c[i&3]] ]; 116 } 117#if !CONFIG_SMALL 118 for(j=256; j<1024; j++) 119 for(i=0; i<4; i++) 120 tbl[4*j+i]= tbl[4*j + ((i-1)&3) - 1024]; 121#endif 122} 123 124// this is based on the reference AES code by Paulo Barreto and Vincent Rijmen 125int av_aes_init(AVAES *a, const uint8_t *key, int key_bits, int decrypt) { 126 int i, j, t, rconpointer = 0; 127 uint8_t tk[8][4]; 128 int KC= key_bits>>5; 129 int rounds= KC + 6; 130 uint8_t log8[256]; 131 uint8_t alog8[512]; 132 133 if(!enc_multbl[0][sizeof(enc_multbl)/sizeof(enc_multbl[0][0])-1]){ 134 j=1; 135 for(i=0; i<255; i++){ 136 alog8[i]= 137 alog8[i+255]= j; 138 log8[j]= i; 139 j^= j+j; 140 if(j>255) j^= 0x11B; 141 } 142 for(i=0; i<256; i++){ 143 j= i ? alog8[255-log8[i]] : 0; 144 j ^= (j<<1) ^ (j<<2) ^ (j<<3) ^ (j<<4); 145 j = (j ^ (j>>8) ^ 99) & 255; 146 inv_sbox[j]= i; 147 sbox [i]= j; 148 } 149 init_multbl2(dec_multbl[0], (const int[4]){0xe, 0x9, 0xd, 0xb}, log8, alog8, inv_sbox); 150 init_multbl2(enc_multbl[0], (const int[4]){0x2, 0x1, 0x1, 0x3}, log8, alog8, sbox); 151 } 152 153 if(key_bits!=128 && key_bits!=192 && key_bits!=256) 154 return -1; 155 156 a->rounds= rounds; 157 158 memcpy(tk, key, KC*4); 159 160 for(t= 0; t < (rounds+1)*16;) { 161 memcpy(a->round_key[0][0]+t, tk, KC*4); 162 t+= KC*4; 163 164 for(i = 0; i < 4; i++) 165 tk[0][i] ^= sbox[tk[KC-1][(i+1)&3]]; 166 tk[0][0] ^= rcon[rconpointer++]; 167 168 for(j = 1; j < KC; j++){ 169 if(KC != 8 || j != KC>>1) 170 for(i = 0; i < 4; i++) tk[j][i] ^= tk[j-1][i]; 171 else 172 for(i = 0; i < 4; i++) tk[j][i] ^= sbox[tk[j-1][i]]; 173 } 174 } 175 176 if(decrypt){ 177 for(i=1; i<rounds; i++){ 178 uint8_t tmp[3][16]; 179 memcpy(tmp[2], a->round_key[i][0], 16); 180 subshift(tmp[1], 0, sbox); 181 mix(tmp, dec_multbl, 1, 3); 182 memcpy(a->round_key[i][0], tmp[0], 16); 183 } 184 }else{ 185 for(i=0; i<(rounds+1)>>1; i++){ 186 for(j=0; j<16; j++) 187 FFSWAP(int, a->round_key[i][0][j], a->round_key[rounds-i][0][j]); 188 } 189 } 190 191 return 0; 192} 193 194#ifdef TEST 195#include "log.h" 196 197#undef random 198 199int main(void){ 200 int i,j; 201 AVAES ae, ad, b; 202 uint8_t rkey[2][16]= { 203 {0}, 204 {0x10, 0xa5, 0x88, 0x69, 0xd7, 0x4b, 0xe5, 0xa3, 0x74, 0xcf, 0x86, 0x7c, 0xfb, 0x47, 0x38, 0x59}}; 205 uint8_t pt[16], rpt[2][16]= { 206 {0x6a, 0x84, 0x86, 0x7c, 0xd7, 0x7e, 0x12, 0xad, 0x07, 0xea, 0x1b, 0xe8, 0x95, 0xc5, 0x3f, 0xa3}, 207 {0}}; 208 uint8_t rct[2][16]= { 209 {0x73, 0x22, 0x81, 0xc0, 0xa0, 0xaa, 0xb8, 0xf7, 0xa5, 0x4a, 0x0c, 0x67, 0xa0, 0xc4, 0x5e, 0xcf}, 210 {0x6d, 0x25, 0x1e, 0x69, 0x44, 0xb0, 0x51, 0xe0, 0x4e, 0xaa, 0x6f, 0xb4, 0xdb, 0xf7, 0x84, 0x65}}; 211 uint8_t temp[16]; 212 213 av_aes_init(&ae, "PI=3.141592654..", 128, 0); 214 av_aes_init(&ad, "PI=3.141592654..", 128, 1); 215 av_log_level= AV_LOG_DEBUG; 216 217 for(i=0; i<2; i++){ 218 av_aes_init(&b, rkey[i], 128, 1); 219 av_aes_crypt(&b, temp, rct[i], 1, NULL, 1); 220 for(j=0; j<16; j++) 221 if(rpt[i][j] != temp[j]) 222 av_log(NULL, AV_LOG_ERROR, "%d %02X %02X\n", j, rpt[i][j], temp[j]); 223 } 224 225 for(i=0; i<10000; i++){ 226 for(j=0; j<16; j++){ 227 pt[j]= random(); 228 } 229{START_TIMER 230 av_aes_crypt(&ae, temp, pt, 1, NULL, 0); 231 if(!(i&(i-1))) 232 av_log(NULL, AV_LOG_ERROR, "%02X %02X %02X %02X\n", temp[0], temp[5], temp[10], temp[15]); 233 av_aes_crypt(&ad, temp, temp, 1, NULL, 1); 234STOP_TIMER("aes")} 235 for(j=0; j<16; j++){ 236 if(pt[j] != temp[j]){ 237 av_log(NULL, AV_LOG_ERROR, "%d %d %02X %02X\n", i,j, pt[j], temp[j]); 238 } 239 } 240 } 241 return 0; 242} 243#endif 244