Mercurial > libavutil.hg
annotate aes.c @ 205:0723fdb8d7a8 libavutil
check the last? entry written for detecting already initalized tables
author | michael |
---|---|
date | Mon, 15 Jan 2007 01:05:40 +0000 |
parents | b7ac7733b2f4 |
children | cd7c6ff2132c |
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 #else | |
70 ((uint32_t *)(state))[i] = multbl[0][state[i][0]] ^ multbl[1][state[i][1]] | |
71 ^multbl[2][state[i][2]] ^ multbl[3][state[i][3]]; | |
72 #endif | |
164 | 73 } |
74 | |
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
|
75 static inline void crypt(AVAES *a, int s, uint8_t *sbox, uint32_t *multbl){ |
164 | 76 int t, r; |
77 | |
188 | 78 for(r=a->rounds; r>1; r--){ |
189 | 79 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
|
80 SUBSHIFT3x((a->state[0]+1+s)) |
185 | 81 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
|
82 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
|
83 mix(a->state, multbl); |
164 | 84 } |
189 | 85 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
|
86 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
|
87 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
|
88 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
|
89 SUBSHIFT1((a->state[0]+3-s), sbox) |
189 | 90 addkey(a->state, a->round_key[0]); |
164 | 91 } |
92 | |
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
|
93 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
|
94 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
|
95 } |
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 |
164 | 97 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
|
98 crypt(a, 2, sbox, enc_multbl); |
178 | 99 } |
100 | |
195 | 101 static void init_multbl2(uint8_t tbl[1024], int c[4], uint8_t *log8, uint8_t *alog8, uint8_t *sbox){ |
200 | 102 int i, j; |
183 | 103 for(i=0; i<1024; i++){ |
204
b7ac7733b2f4
replace / by >> this makes the .o ~150 bytes smaller
michael
parents:
203
diff
changeset
|
104 int x= sbox[i>>2]; |
183 | 105 if(x) tbl[i]= alog8[ log8[x] + log8[c[i&3]] ]; |
106 } | |
199 | 107 #ifndef CONFIG_SMALL |
200 | 108 for(j=256; j<1024; j++) |
109 for(i=0; i<4; i++) | |
110 tbl[4*j+i]= tbl[4*j + ((i-1)&3) - 1024]; | |
199 | 111 #endif |
183 | 112 } |
113 | |
164 | 114 // this is based on the reference AES code by Paulo Barreto and Vincent Rijmen |
189 | 115 AVAES *av_aes_init(uint8_t *key, int key_bits, int decrypt) { |
166 | 116 AVAES *a; |
164 | 117 int i, j, t, rconpointer = 0; |
118 uint8_t tk[8][4]; | |
204
b7ac7733b2f4
replace / by >> this makes the .o ~150 bytes smaller
michael
parents:
203
diff
changeset
|
119 int KC= key_bits>>5; |
167 | 120 int rounds= KC + 6; |
169 | 121 uint8_t log8[256]; |
122 uint8_t alog8[512]; | |
164 | 123 |
205
0723fdb8d7a8
check the last? entry written for detecting already initalized tables
michael
parents:
204
diff
changeset
|
124 if(!enc_multbl[4][1023]){ |
164 | 125 j=1; |
126 for(i=0; i<255; i++){ | |
127 alog8[i]= | |
128 alog8[i+255]= j; | |
129 log8[j]= i; | |
130 j^= j+j; | |
131 if(j>255) j^= 0x11B; | |
132 } | |
133 for(i=0; i<256; i++){ | |
134 j= i ? alog8[255-log8[i]] : 0; | |
165 | 135 j ^= (j<<1) ^ (j<<2) ^ (j<<3) ^ (j<<4); |
136 j = (j ^ (j>>8) ^ 99) & 255; | |
164 | 137 inv_sbox[j]= i; |
138 sbox [i]= j; | |
139 } | |
184 | 140 init_multbl2(dec_multbl[0], (int[4]){0xe, 0x9, 0xd, 0xb}, log8, alog8, inv_sbox); |
183 | 141 init_multbl2(enc_multbl[0], (int[4]){0x2, 0x1, 0x1, 0x3}, log8, alog8, sbox); |
164 | 142 } |
143 | |
167 | 144 if(key_bits!=128 && key_bits!=192 && key_bits!=256) |
164 | 145 return NULL; |
146 | |
166 | 147 a= av_malloc(sizeof(AVAES)); |
167 | 148 a->rounds= rounds; |
166 | 149 |
164 | 150 memcpy(tk, key, KC*4); |
151 | |
174
263bbdc10c1e
simplify round_key generation by writing over the end but ensuring that theres some irrelevant stuff afterwards
michael
parents:
173
diff
changeset
|
152 for(t= 0; t < (rounds+1)*4;) { |
197 | 153 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
|
154 t+= KC; |
164 | 155 |
156 for(i = 0; i < 4; i++) | |
173 | 157 tk[0][i] ^= sbox[tk[KC-1][(i+1)&3]]; |
164 | 158 tk[0][0] ^= rcon[rconpointer++]; |
159 | |
160 for(j = 1; j < KC; j++){ | |
204
b7ac7733b2f4
replace / by >> this makes the .o ~150 bytes smaller
michael
parents:
203
diff
changeset
|
161 if(KC != 8 || j != KC>>1) |
175 | 162 for(i = 0; i < 4; i++) tk[j][i] ^= tk[j-1][i]; |
164 | 163 else |
175 | 164 for(i = 0; i < 4; i++) tk[j][i] ^= sbox[tk[j-1][i]]; |
164 | 165 } |
166 } | |
184 | 167 |
189 | 168 if(decrypt){ |
190 | 169 for(i=1; i<rounds; i++){ |
170 for(j=0; j<16; j++) | |
171 a->round_key[i][0][j]= sbox[a->round_key[i][0][j]]; | |
172 mix(a->round_key[i], dec_multbl); | |
173 } | |
197 | 174 }else{ |
204
b7ac7733b2f4
replace / by >> this makes the .o ~150 bytes smaller
michael
parents:
203
diff
changeset
|
175 for(i=0; i<(rounds+1)>>1; i++){ |
197 | 176 for(j=0; j<16; j++) |
177 FFSWAP(int, a->round_key[i][0][j], a->round_key[rounds-i][0][j]); | |
178 } | |
187 | 179 } |
184 | 180 |
164 | 181 return a; |
182 } | |
183 | |
184 #ifdef TEST | |
185 | |
186 int main(){ | |
195 | 187 int i,j; |
189 | 188 AVAES *ae= av_aes_init("PI=3.141592654..", 128, 0); |
189 AVAES *ad= av_aes_init("PI=3.141592654..", 128, 1); | |
198 | 190 uint8_t rkey[2][16]= { |
191 {0}, | |
192 {0x10, 0xa5, 0x88, 0x69, 0xd7, 0x4b, 0xe5, 0xa3, 0x74, 0xcf, 0x86, 0x7c, 0xfb, 0x47, 0x38, 0x59}}; | |
193 uint8_t pt[16], rpt[2][16]= { | |
194 {0x6a, 0x84, 0x86, 0x7c, 0xd7, 0x7e, 0x12, 0xad, 0x07, 0xea, 0x1b, 0xe8, 0x95, 0xc5, 0x3f, 0xa3}, | |
195 {0}}; | |
196 uint8_t rct[2][16]= { | |
197 {0x73, 0x22, 0x81, 0xc0, 0xa0, 0xaa, 0xb8, 0xf7, 0xa5, 0x4a, 0x0c, 0x67, 0xa0, 0xc4, 0x5e, 0xcf}, | |
198 {0x6d, 0x25, 0x1e, 0x69, 0x44, 0xb0, 0x51, 0xe0, 0x4e, 0xaa, 0x6f, 0xb4, 0xdb, 0xf7, 0x84, 0x65}}; | |
164 | 199 |
179 | 200 av_log_level= AV_LOG_DEBUG; |
201 | |
198 | 202 for(i=0; i<2; i++){ |
203 AVAES *b= av_aes_init(rkey[i], 128, 1); | |
204 memcpy(b->state, rct[i], 16); | |
205 av_aes_decrypt(b); | |
206 for(j=0; j<16; j++) | |
207 if(rpt[i][j] != b->state[0][j]) | |
208 av_log(NULL, AV_LOG_ERROR, "%d %02X %02X\n", j, rpt[i][j], b->state[0][j]); | |
209 } | |
183 | 210 |
164 | 211 for(i=0; i<10000; i++){ |
178 | 212 for(j=0; j<16; j++){ |
213 pt[j]= random(); | |
214 } | |
189 | 215 memcpy(ae->state, pt, 16); |
179 | 216 {START_TIMER |
189 | 217 av_aes_encrypt(ae); |
178 | 218 if(!(i&(i-1))) |
189 | 219 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]); |
220 memcpy(ad->state, ae->state, 16); | |
221 av_aes_decrypt(ad); | |
179 | 222 STOP_TIMER("aes")} |
178 | 223 for(j=0; j<16; j++){ |
189 | 224 if(pt[j] != ad->state[0][j]){ |
225 av_log(NULL, AV_LOG_ERROR, "%d %d %02X %02X\n", i,j, pt[j], ad->state[0][j]); | |
178 | 226 } |
227 } | |
164 | 228 } |
229 return 0; | |
230 } | |
231 #endif |