Mercurial > libavutil.hg
annotate aes.c @ 207:b5ff46799eea libavutil
move #include log.h into the selftesting code where it belongs
author | michael |
---|---|
date | Mon, 15 Jan 2007 01:33:56 +0000 |
parents | cd7c6ff2132c |
children | d846d66f99e9 |
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 "aes.h" | |
25 | |
26 typedef struct AVAES{ | |
189 | 27 uint8_t round_key[15][4][4]; |
164 | 28 uint8_t state[4][4]; |
29 int rounds; | |
30 }AVAES; | |
31 | |
206 | 32 const int av_aes_size= sizeof(AVAES); |
33 | |
192 | 34 static const uint8_t rcon[10] = { |
35 0x01, 0x02, 0x04, 0x08, 0x10, 0x20, 0x40, 0x80, 0x1b, 0x36 | |
164 | 36 }; |
37 | |
38 static uint8_t sbox[256]; | |
39 static uint8_t inv_sbox[256]; | |
169 | 40 #ifdef CONFIG_SMALL |
41 static uint32_t enc_multbl[1][256]; | |
42 static uint32_t dec_multbl[1][256]; | |
43 #else | |
44 static uint32_t enc_multbl[4][256]; | |
45 static uint32_t dec_multbl[4][256]; | |
46 #endif | |
164 | 47 |
168 | 48 static inline void addkey(uint64_t state[2], uint64_t round_key[2]){ |
49 state[0] ^= round_key[0]; | |
50 state[1] ^= round_key[1]; | |
164 | 51 } |
52 | |
53 #define SUBSHIFT0(s, box) s[0]=box[s[ 0]]; s[ 4]=box[s[ 4]]; s[ 8]=box[s[ 8]]; s[12]=box[s[12]]; | |
54 #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]; | |
55 #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]; | |
56 #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]; | |
57 | |
185 | 58 #define SUBSHIFT1x(s) t=s[0]; s[0]=s[ 4]; s[ 4]=s[ 8]; s[ 8]=s[12]; s[12]=t; |
59 #define SUBSHIFT2x(s) t=s[0]; s[0]=s[ 8]; s[ 8]= t; t=s[ 4]; s[ 4]=s[12]; s[12]=t; | |
60 #define SUBSHIFT3x(s) t=s[0]; s[0]=s[12]; s[12]=s[ 8]; s[ 8]=s[ 4]; s[ 4]=t; | |
183 | 61 |
181 | 62 #define ROT(x,s) ((x<<s)|(x>>(32-s))) |
164 | 63 |
169 | 64 static inline void mix(uint8_t state[4][4], uint32_t multbl[4][256]){ |
65 int i; | |
66 for(i=0; i<4; i++) | |
67 #ifdef CONFIG_SMALL | |
68 ((uint32_t *)(state))[i] = multbl[0][state[i][0]] ^ ROT(multbl[0][state[i][1]], 8) | |
69 ^ROT(multbl[0][state[i][2]],16) ^ ROT(multbl[0][state[i][3]],24); | |
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){ |
200 | 103 int i, j; |
183 | 104 for(i=0; i<1024; i++){ |
204
b7ac7733b2f4
replace / by >> this makes the .o ~150 bytes smaller
michael
parents:
203
diff
changeset
|
105 int x= sbox[i>>2]; |
183 | 106 if(x) tbl[i]= alog8[ log8[x] + log8[c[i&3]] ]; |
107 } | |
199 | 108 #ifndef CONFIG_SMALL |
200 | 109 for(j=256; j<1024; j++) |
110 for(i=0; i<4; i++) | |
111 tbl[4*j+i]= tbl[4*j + ((i-1)&3) - 1024]; | |
199 | 112 #endif |
183 | 113 } |
114 | |
164 | 115 // this is based on the reference AES code by Paulo Barreto and Vincent Rijmen |
206 | 116 int av_aes_init(AVAES *a, uint8_t *key, int key_bits, int decrypt) { |
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) |
206 | 145 return -1; |
164 | 146 |
167 | 147 a->rounds= rounds; |
166 | 148 |
164 | 149 memcpy(tk, key, KC*4); |
150 | |
174
263bbdc10c1e
simplify round_key generation by writing over the end but ensuring that theres some irrelevant stuff afterwards
michael
parents:
173
diff
changeset
|
151 for(t= 0; t < (rounds+1)*4;) { |
197 | 152 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
|
153 t+= KC; |
164 | 154 |
155 for(i = 0; i < 4; i++) | |
173 | 156 tk[0][i] ^= sbox[tk[KC-1][(i+1)&3]]; |
164 | 157 tk[0][0] ^= rcon[rconpointer++]; |
158 | |
159 for(j = 1; j < KC; j++){ | |
204
b7ac7733b2f4
replace / by >> this makes the .o ~150 bytes smaller
michael
parents:
203
diff
changeset
|
160 if(KC != 8 || j != KC>>1) |
175 | 161 for(i = 0; i < 4; i++) tk[j][i] ^= tk[j-1][i]; |
164 | 162 else |
175 | 163 for(i = 0; i < 4; i++) tk[j][i] ^= sbox[tk[j-1][i]]; |
164 | 164 } |
165 } | |
184 | 166 |
189 | 167 if(decrypt){ |
190 | 168 for(i=1; i<rounds; i++){ |
169 for(j=0; j<16; j++) | |
170 a->round_key[i][0][j]= sbox[a->round_key[i][0][j]]; | |
171 mix(a->round_key[i], dec_multbl); | |
172 } | |
197 | 173 }else{ |
204
b7ac7733b2f4
replace / by >> this makes the .o ~150 bytes smaller
michael
parents:
203
diff
changeset
|
174 for(i=0; i<(rounds+1)>>1; i++){ |
197 | 175 for(j=0; j<16; j++) |
176 FFSWAP(int, a->round_key[i][0][j], a->round_key[rounds-i][0][j]); | |
177 } | |
187 | 178 } |
184 | 179 |
206 | 180 return 0; |
164 | 181 } |
182 | |
183 #ifdef TEST | |
207
b5ff46799eea
move #include log.h into the selftesting code where it belongs
michael
parents:
206
diff
changeset
|
184 #include "log.h" |
164 | 185 |
186 int main(){ | |
195 | 187 int i,j; |
206 | 188 AVAES ae, ad, b; |
198 | 189 uint8_t rkey[2][16]= { |
190 {0}, | |
191 {0x10, 0xa5, 0x88, 0x69, 0xd7, 0x4b, 0xe5, 0xa3, 0x74, 0xcf, 0x86, 0x7c, 0xfb, 0x47, 0x38, 0x59}}; | |
192 uint8_t pt[16], rpt[2][16]= { | |
193 {0x6a, 0x84, 0x86, 0x7c, 0xd7, 0x7e, 0x12, 0xad, 0x07, 0xea, 0x1b, 0xe8, 0x95, 0xc5, 0x3f, 0xa3}, | |
194 {0}}; | |
195 uint8_t rct[2][16]= { | |
196 {0x73, 0x22, 0x81, 0xc0, 0xa0, 0xaa, 0xb8, 0xf7, 0xa5, 0x4a, 0x0c, 0x67, 0xa0, 0xc4, 0x5e, 0xcf}, | |
197 {0x6d, 0x25, 0x1e, 0x69, 0x44, 0xb0, 0x51, 0xe0, 0x4e, 0xaa, 0x6f, 0xb4, 0xdb, 0xf7, 0x84, 0x65}}; | |
164 | 198 |
206 | 199 av_aes_init(&ae, "PI=3.141592654..", 128, 0); |
200 av_aes_init(&ad, "PI=3.141592654..", 128, 1); | |
179 | 201 av_log_level= AV_LOG_DEBUG; |
202 | |
198 | 203 for(i=0; i<2; i++){ |
206 | 204 av_aes_init(&b, rkey[i], 128, 1); |
205 memcpy(b.state, rct[i], 16); | |
206 av_aes_decrypt(&b); | |
198 | 207 for(j=0; j<16; j++) |
206 | 208 if(rpt[i][j] != b.state[0][j]) |
209 av_log(NULL, AV_LOG_ERROR, "%d %02X %02X\n", j, rpt[i][j], b.state[0][j]); | |
198 | 210 } |
183 | 211 |
164 | 212 for(i=0; i<10000; i++){ |
178 | 213 for(j=0; j<16; j++){ |
214 pt[j]= random(); | |
215 } | |
206 | 216 memcpy(ae.state, pt, 16); |
179 | 217 {START_TIMER |
206 | 218 av_aes_encrypt(&ae); |
178 | 219 if(!(i&(i-1))) |
206 | 220 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]); |
221 memcpy(ad.state, ae.state, 16); | |
222 av_aes_decrypt(&ad); | |
179 | 223 STOP_TIMER("aes")} |
178 | 224 for(j=0; j<16; j++){ |
206 | 225 if(pt[j] != ad.state[0][j]){ |
226 av_log(NULL, AV_LOG_ERROR, "%d %d %02X %02X\n", i,j, pt[j], ad.state[0][j]); | |
178 | 227 } |
228 } | |
164 | 229 } |
230 return 0; | |
231 } | |
232 #endif |