| 1 | /***************************************************************************** |
| 2 | * WARNING |
| 3 | * |
| 4 | * THIS CODE IS CREATED FOR EXPERIMENTATION AND EDUCATIONAL USE ONLY. |
| 5 | * |
| 6 | * USAGE OF THIS CODE IN OTHER WAYS MAY INFRINGE UPON THE INTELLECTUAL |
| 7 | * PROPERTY OF OTHER PARTIES, SUCH AS INSIDE SECURE AND HID GLOBAL, |
| 8 | * AND MAY EXPOSE YOU TO AN INFRINGEMENT ACTION FROM THOSE PARTIES. |
| 9 | * |
| 10 | * THIS CODE SHOULD NEVER BE USED TO INFRINGE PATENTS OR INTELLECTUAL PROPERTY RIGHTS. |
| 11 | * |
| 12 | ***************************************************************************** |
| 13 | * |
| 14 | * This file is part of loclass. It is a reconstructon of the cipher engine |
| 15 | * used in iClass, and RFID techology. |
| 16 | * |
| 17 | * The implementation is based on the work performed by |
| 18 | * Flavio D. Garcia, Gerhard de Koning Gans, Roel Verdult and |
| 19 | * Milosch Meriac in the paper "Dismantling IClass". |
| 20 | * |
| 21 | * Copyright (C) 2014 Martin Holst Swende |
| 22 | * |
| 23 | * This is free software: you can redistribute it and/or modify |
| 24 | * it under the terms of the GNU General Public License version 2 as published |
| 25 | * by the Free Software Foundation. |
| 26 | * |
| 27 | * This file is distributed in the hope that it will be useful, |
| 28 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 29 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 30 | * GNU General Public License for more details. |
| 31 | * |
| 32 | * You should have received a copy of the GNU General Public License |
| 33 | * along with loclass. If not, see <http://www.gnu.org/licenses/>. |
| 34 | * |
| 35 | * |
| 36 | * |
| 37 | ****************************************************************************/ |
| 38 | |
| 39 | /** |
| 40 | |
| 41 | This file contains an optimized version of the MAC-calculation algorithm. Some measurements on |
| 42 | a std laptop showed it runs in about 1/3 of the time: |
| 43 | |
| 44 | Std: 0.428962 |
| 45 | Opt: 0.151609 |
| 46 | |
| 47 | Additionally, it is self-reliant, not requiring e.g. bitstreams from the cipherutils, thus can |
| 48 | be easily dropped into a code base. |
| 49 | |
| 50 | The optimizations have been performed in the following steps: |
| 51 | * Parameters passed by reference instead of by value. |
| 52 | * Iteration instead of recursion, un-nesting recursive loops into for-loops. |
| 53 | * Handling of bytes instead of individual bits, for less shuffling and masking |
| 54 | * Less creation of "objects", structs, and instead reuse of alloc:ed memory |
| 55 | * Inlining some functions via #define:s |
| 56 | |
| 57 | As a consequence, this implementation is less generic. Also, I haven't bothered documenting this. |
| 58 | For a thorough documentation, check out the MAC-calculation within cipher.c instead. |
| 59 | |
| 60 | -- MHS 2015 |
| 61 | **/ |
| 62 | |
| 63 | #include "optimized_cipher.h" |
| 64 | #include <stdlib.h> |
| 65 | #include <string.h> |
| 66 | #include <stdbool.h> |
| 67 | #include <stdint.h> |
| 68 | |
| 69 | |
| 70 | #define opt_T(s) (0x1 & ((s->t >> 15) ^ (s->t >> 14)^ (s->t >> 10)^ (s->t >> 8)^ (s->t >> 5)^ (s->t >> 4)^ (s->t >> 1)^ s->t)) |
| 71 | |
| 72 | #define opt_B(s) (((s->b >> 6) ^ (s->b >> 5) ^ (s->b >> 4) ^ (s->b)) & 0x1) |
| 73 | |
| 74 | #define opt__select(x,y,r) (4 & (((r & (r << 2)) >> 5) ^ ((r & ~(r << 2)) >> 4) ^ ( (r | r << 2) >> 3)))\ |
| 75 | |(2 & (((r | r << 2) >> 6) ^ ( (r | r << 2) >> 1) ^ (r >> 5) ^ r ^ ((x^y) << 1)))\ |
| 76 | |(1 & (((r & ~(r << 2)) >> 4) ^ ((r & (r << 2)) >> 3) ^ r ^ x)) |
| 77 | |
| 78 | /* |
| 79 | * Some background on the expression above can be found here... |
| 80 | uint8_t xopt__select(bool x, bool y, uint8_t r) |
| 81 | { |
| 82 | uint8_t r_ls2 = r << 2; |
| 83 | uint8_t r_and_ls2 = r & r_ls2; |
| 84 | uint8_t r_or_ls2 = r | r_ls2; |
| 85 | |
| 86 | //r: r0 r1 r2 r3 r4 r5 r6 r7 |
| 87 | //r_ls2: r2 r3 r4 r5 r6 r7 0 0 |
| 88 | // z0 |
| 89 | // z1 |
| 90 | |
| 91 | // uint8_t z0 = (r0 & r2) ^ (r1 & ~r3) ^ (r2 | r4); // <-- original |
| 92 | uint8_t z0 = (r_and_ls2 >> 5) ^ ((r & ~r_ls2) >> 4) ^ ( r_or_ls2 >> 3); |
| 93 | |
| 94 | // uint8_t z1 = (r0 | r2) ^ ( r5 | r7) ^ r1 ^ r6 ^ x ^ y; // <-- original |
| 95 | uint8_t z1 = (r_or_ls2 >> 6) ^ ( r_or_ls2 >> 1) ^ (r >> 5) ^ r ^ ((x^y) << 1); |
| 96 | |
| 97 | // uint8_t z2 = (r3 & ~r5) ^ (r4 & r6 ) ^ r7 ^ x; // <-- original |
| 98 | uint8_t z2 = ((r & ~r_ls2) >> 4) ^ (r_and_ls2 >> 3) ^ r ^ x; |
| 99 | |
| 100 | return (z0 & 4) | (z1 & 2) | (z2 & 1); |
| 101 | } |
| 102 | */ |
| 103 | |
| 104 | void opt_successor(const uint8_t* k, State *s, bool y, State* successor) |
| 105 | { |
| 106 | |
| 107 | uint8_t Tt = 1 & opt_T(s); |
| 108 | |
| 109 | successor->t = (s->t >> 1); |
| 110 | successor->t |= (Tt ^ (s->r >> 7 & 0x1) ^ (s->r >> 3 & 0x1)) << 15; |
| 111 | |
| 112 | successor->b = s->b >> 1; |
| 113 | successor->b |= (opt_B(s) ^ (s->r & 0x1)) << 7; |
| 114 | |
| 115 | successor->r = (k[opt__select(Tt,y,s->r)] ^ successor->b) + s->l ; |
| 116 | successor->l = successor->r+s->r; |
| 117 | |
| 118 | } |
| 119 | |
| 120 | void opt_suc(const uint8_t* k,State* s, uint8_t *in, uint8_t length, bool add32Zeroes) |
| 121 | { |
| 122 | State x2; |
| 123 | int i; |
| 124 | uint8_t head = 0; |
| 125 | for(i =0 ; i < length ; i++) |
| 126 | { |
| 127 | head = 1 & (in[i] >> 7); |
| 128 | opt_successor(k,s,head,&x2); |
| 129 | |
| 130 | head = 1 & (in[i] >> 6); |
| 131 | opt_successor(k,&x2,head,s); |
| 132 | |
| 133 | head = 1 & (in[i] >> 5); |
| 134 | opt_successor(k,s,head,&x2); |
| 135 | |
| 136 | head = 1 & (in[i] >> 4); |
| 137 | opt_successor(k,&x2,head,s); |
| 138 | |
| 139 | head = 1 & (in[i] >> 3); |
| 140 | opt_successor(k,s,head,&x2); |
| 141 | |
| 142 | head = 1 & (in[i] >> 2); |
| 143 | opt_successor(k,&x2,head,s); |
| 144 | |
| 145 | head = 1 & (in[i] >> 1); |
| 146 | opt_successor(k,s,head,&x2); |
| 147 | |
| 148 | head = 1 & in[i]; |
| 149 | opt_successor(k,&x2,head,s); |
| 150 | |
| 151 | } |
| 152 | //For tag MAC, an additional 32 zeroes |
| 153 | if(add32Zeroes) |
| 154 | for(i =0 ; i < 16 ; i++) |
| 155 | { |
| 156 | opt_successor(k,s,0,&x2); |
| 157 | opt_successor(k,&x2,0,s); |
| 158 | } |
| 159 | } |
| 160 | |
| 161 | void opt_output(const uint8_t* k,State* s, uint8_t *buffer) |
| 162 | { |
| 163 | uint8_t times = 0; |
| 164 | uint8_t bout = 0; |
| 165 | State temp = {0,0,0,0}; |
| 166 | for( ; times < 4 ; times++) |
| 167 | { |
| 168 | bout =0; |
| 169 | bout |= (s->r & 0x4) << 5; |
| 170 | opt_successor(k,s,0,&temp); |
| 171 | bout |= (temp.r & 0x4) << 4; |
| 172 | opt_successor(k,&temp,0,s); |
| 173 | bout |= (s->r & 0x4) << 3; |
| 174 | opt_successor(k,s,0,&temp); |
| 175 | bout |= (temp.r & 0x4) << 2; |
| 176 | opt_successor(k,&temp,0,s); |
| 177 | bout |= (s->r & 0x4) << 1; |
| 178 | opt_successor(k,s,0,&temp); |
| 179 | bout |= (temp.r & 0x4) ; |
| 180 | opt_successor(k,&temp,0,s); |
| 181 | bout |= (s->r & 0x4) >> 1; |
| 182 | opt_successor(k,s,0,&temp); |
| 183 | bout |= (temp.r & 0x4) >> 2; |
| 184 | opt_successor(k,&temp,0,s); |
| 185 | buffer[times] = bout; |
| 186 | } |
| 187 | |
| 188 | } |
| 189 | |
| 190 | void opt_MAC(uint8_t* k, uint8_t* input, uint8_t* out) |
| 191 | { |
| 192 | State _init = { |
| 193 | ((k[0] ^ 0x4c) + 0xEC) & 0xFF,// l |
| 194 | ((k[0] ^ 0x4c) + 0x21) & 0xFF,// r |
| 195 | 0x4c, // b |
| 196 | 0xE012 // t |
| 197 | }; |
| 198 | |
| 199 | opt_suc(k,&_init,input,12, false); |
| 200 | //printf("\noutp "); |
| 201 | opt_output(k,&_init, out); |
| 202 | } |
| 203 | uint8_t rev_byte(uint8_t b) { |
| 204 | b = (b & 0xF0) >> 4 | (b & 0x0F) << 4; |
| 205 | b = (b & 0xCC) >> 2 | (b & 0x33) << 2; |
| 206 | b = (b & 0xAA) >> 1 | (b & 0x55) << 1; |
| 207 | return b; |
| 208 | } |
| 209 | void opt_reverse_arraybytecpy(uint8_t* dest, uint8_t *src, size_t len) |
| 210 | { |
| 211 | uint8_t i; |
| 212 | for( i =0; i< len ; i++) |
| 213 | dest[i] = rev_byte(src[i]); |
| 214 | } |
| 215 | |
| 216 | void opt_doReaderMAC(uint8_t *cc_nr_p, uint8_t *div_key_p, uint8_t mac[4]) |
| 217 | { |
| 218 | static uint8_t cc_nr[12]; |
| 219 | |
| 220 | opt_reverse_arraybytecpy(cc_nr, cc_nr_p,12); |
| 221 | uint8_t dest []= {0,0,0,0,0,0,0,0}; |
| 222 | opt_MAC(div_key_p,cc_nr, dest); |
| 223 | //The output MAC must also be reversed |
| 224 | opt_reverse_arraybytecpy(mac, dest,4); |
| 225 | return; |
| 226 | } |
| 227 | void opt_doTagMAC(uint8_t *cc_p, const uint8_t *div_key_p, uint8_t mac[4]) |
| 228 | { |
| 229 | static uint8_t cc_nr[8+4+4]; |
| 230 | opt_reverse_arraybytecpy(cc_nr, cc_p,12); |
| 231 | State _init = { |
| 232 | ((div_key_p[0] ^ 0x4c) + 0xEC) & 0xFF,// l |
| 233 | ((div_key_p[0] ^ 0x4c) + 0x21) & 0xFF,// r |
| 234 | 0x4c, // b |
| 235 | 0xE012 // t |
| 236 | }; |
| 237 | opt_suc(div_key_p,&_init,cc_nr, 12,true); |
| 238 | uint8_t dest []= {0,0,0,0}; |
| 239 | opt_output(div_key_p,&_init, dest); |
| 240 | //The output MAC must also be reversed |
| 241 | opt_reverse_arraybytecpy(mac, dest,4); |
| 242 | return; |
| 243 | |
| 244 | } |
| 245 | /** |
| 246 | * The tag MAC can be divided (both can, but no point in dividing the reader mac) into |
| 247 | * two functions, since the first 8 bytes are known, we can pre-calculate the state |
| 248 | * reached after feeding CC to the cipher. |
| 249 | * @param cc_p |
| 250 | * @param div_key_p |
| 251 | * @return the cipher state |
| 252 | */ |
| 253 | State opt_doTagMAC_1(uint8_t *cc_p, const uint8_t *div_key_p) |
| 254 | { |
| 255 | static uint8_t cc_nr[8]; |
| 256 | opt_reverse_arraybytecpy(cc_nr, cc_p,8); |
| 257 | State _init = { |
| 258 | ((div_key_p[0] ^ 0x4c) + 0xEC) & 0xFF,// l |
| 259 | ((div_key_p[0] ^ 0x4c) + 0x21) & 0xFF,// r |
| 260 | 0x4c, // b |
| 261 | 0xE012 // t |
| 262 | }; |
| 263 | opt_suc(div_key_p,&_init,cc_nr, 8,false); |
| 264 | return _init; |
| 265 | } |
| 266 | /** |
| 267 | * The second part of the tag MAC calculation, since the CC is already calculated into the state, |
| 268 | * this function is fed only the NR, and internally feeds the remaining 32 0-bits to generate the tag |
| 269 | * MAC response. |
| 270 | * @param _init - precalculated cipher state |
| 271 | * @param nr - the reader challenge |
| 272 | * @param mac - where to store the MAC |
| 273 | * @param div_key_p - the key to use |
| 274 | */ |
| 275 | void opt_doTagMAC_2(State _init, uint8_t* nr, uint8_t mac[4], const uint8_t* div_key_p) |
| 276 | { |
| 277 | static uint8_t _nr [4]; |
| 278 | opt_reverse_arraybytecpy(_nr, nr, 4); |
| 279 | opt_suc(div_key_p,&_init,_nr, 4, true); |
| 280 | //opt_suc(div_key_p,&_init,nr, 4, false); |
| 281 | uint8_t dest []= {0,0,0,0}; |
| 282 | opt_output(div_key_p,&_init, dest); |
| 283 | //The output MAC must also be reversed |
| 284 | opt_reverse_arraybytecpy(mac, dest,4); |
| 285 | return; |
| 286 | } |