]>
cvs.zerfleddert.de Git - proxmark3-svn/blob - armsrc/optimized_cipher.c
444b93d09f688db68b71d152745c3a92f84c6b43
   1 /***************************************************************************** 
   4  * THIS CODE IS CREATED FOR EXPERIMENTATION AND EDUCATIONAL USE ONLY.  
   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.  
  10  * THIS CODE SHOULD NEVER BE USED TO INFRINGE PATENTS OR INTELLECTUAL PROPERTY RIGHTS.  
  12  ***************************************************************************** 
  14  * This file is part of loclass. It is a reconstructon of the cipher engine 
  15  * used in iClass, and RFID techology. 
  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". 
  21  * Copyright (C) 2014 Martin Holst Swende 
  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. 
  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. 
  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/>. 
  37  ****************************************************************************/ 
  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: 
  47   Additionally, it is self-reliant, not requiring e.g. bitstreams from the cipherutils, thus can 
  48   be easily dropped into a code base. 
  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 
  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. 
  63 #include "optimized_cipher.h" 
  72 #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)) 
  74 #define opt_B(s) (((s->b >> 6) ^ (s->b >> 5) ^ (s->b >> 4) ^ (s->b)) & 0x1) 
  76 #define opt__select(x,y,r)  (4 & (((r & (r << 2)) >> 5) ^ ((r & ~(r << 2)) >> 4) ^ ( (r | r << 2) >> 3)))\ 
  77         |(2 & (((r | r << 2) >> 6) ^ ( (r | r << 2) >> 1) ^ (r >> 5) ^ r ^ ((x^y) << 1)))\ 
  78         |(1 & (((r & ~(r << 2)) >> 4) ^ ((r & (r << 2)) >> 3) ^ r ^ x)) 
  81  * Some background on the expression above can be found here... 
  82 uint8_t xopt__select(bool x, bool y, uint8_t r) 
  84         uint8_t r_ls2 = r << 2; 
  85         uint8_t r_and_ls2 = r & r_ls2; 
  86         uint8_t r_or_ls2  = r | r_ls2; 
  88         //r:      r0 r1 r2 r3 r4 r5 r6 r7 
  89         //r_ls2:  r2 r3 r4 r5 r6 r7  0  0 
  93 //      uint8_t z0 = (r0 & r2) ^ (r1 & ~r3) ^ (r2 | r4); // <-- original 
  94         uint8_t z0 = (r_and_ls2 >> 5) ^ ((r & ~r_ls2) >> 4) ^ ( r_or_ls2 >> 3); 
  96 //      uint8_t z1 = (r0 | r2) ^ ( r5 | r7) ^ r1 ^ r6 ^ x ^ y;  // <-- original 
  97         uint8_t z1 = (r_or_ls2 >> 6) ^ ( r_or_ls2 >> 1) ^ (r >> 5) ^ r ^ ((x^y) << 1); 
  99 //      uint8_t z2 = (r3 & ~r5) ^ (r4 & r6 ) ^ r7 ^ x;  // <-- original 
 100         uint8_t z2 = ((r & ~r_ls2) >> 4) ^ (r_and_ls2 >> 3) ^ r ^ x; 
 102         return (z0 & 4) | (z1 & 2) | (z2 & 1); 
 106 void opt_successor(const uint8_t* k
, State 
*s
, bool y
, State
* successor
) 
 109         uint8_t Tt 
= 1 & opt_T(s
); 
 111         successor
->t 
= (s
->t 
>> 1); 
 112         successor
->t 
|= (Tt 
^ (s
->r 
>> 7 & 0x1) ^ (s
->r 
>> 3 & 0x1)) << 15; 
 114         successor
->b 
= s
->b 
>> 1; 
 115         successor
->b 
|= (opt_B(s
) ^ (s
->r 
& 0x1)) << 7; 
 117         successor
->r 
= (k
[opt__select(Tt
,y
,s
->r
)] ^ successor
->b
) + s
->l 
; 
 118         successor
->l 
= successor
->r
+s
->r
; 
 122 void opt_suc(const uint8_t* k
,State
* s
, uint8_t *in
, uint8_t length
, bool add32Zeroes
) 
 127         for(i 
=0 ; i 
< length  
; i
++) 
 129                 head 
= 1 & (in
[i
] >> 7); 
 130                 opt_successor(k
,s
,head
,&x2
); 
 132                 head 
= 1 & (in
[i
] >> 6); 
 133                 opt_successor(k
,&x2
,head
,s
); 
 135                 head 
= 1 & (in
[i
] >> 5); 
 136                 opt_successor(k
,s
,head
,&x2
); 
 138                 head 
= 1 & (in
[i
] >> 4); 
 139                 opt_successor(k
,&x2
,head
,s
); 
 141                 head 
= 1 & (in
[i
] >> 3); 
 142                 opt_successor(k
,s
,head
,&x2
); 
 144                 head 
= 1 & (in
[i
] >> 2); 
 145                 opt_successor(k
,&x2
,head
,s
); 
 147                 head 
= 1 & (in
[i
] >> 1); 
 148                 opt_successor(k
,s
,head
,&x2
); 
 151                 opt_successor(k
,&x2
,head
,s
); 
 154         //For tag MAC, an additional 32 zeroes 
 156                 for(i 
=0 ; i 
< 16 ; i
++) 
 158                         opt_successor(k
,s
,0,&x2
); 
 159                         opt_successor(k
,&x2
,0,s
); 
 163 void opt_output(const uint8_t* k
,State
* s
,  uint8_t *buffer
) 
 167         State temp 
= {0,0,0,0}; 
 168         for( ; times 
< 4 ; times
++) 
 171                 bout 
|= (s
->r 
& 0x4) << 5; 
 172                 opt_successor(k
,s
,0,&temp
); 
 173                 bout 
|= (temp
.r 
& 0x4) << 4; 
 174                 opt_successor(k
,&temp
,0,s
); 
 175                 bout 
|= (s
->r 
& 0x4) << 3; 
 176                 opt_successor(k
,s
,0,&temp
); 
 177                 bout 
|= (temp
.r 
& 0x4) << 2; 
 178                 opt_successor(k
,&temp
,0,s
); 
 179                 bout 
|= (s
->r 
& 0x4) << 1; 
 180                 opt_successor(k
,s
,0,&temp
); 
 181                 bout 
|= (temp
.r 
& 0x4) ; 
 182                 opt_successor(k
,&temp
,0,s
); 
 183                 bout 
|= (s
->r 
& 0x4) >> 1; 
 184                 opt_successor(k
,s
,0,&temp
); 
 185                 bout 
|= (temp
.r 
& 0x4) >> 2; 
 186                 opt_successor(k
,&temp
,0,s
); 
 187                 buffer
[times
] = bout
; 
 192 void opt_MAC(uint8_t* k
, uint8_t* input
, uint8_t* out
) 
 195                         ((k
[0] ^ 0x4c) + 0xEC) & 0xFF,// l 
 196                         ((k
[0] ^ 0x4c) + 0x21) & 0xFF,// r 
 201         opt_suc(k
,&_init
,input
,12, false); 
 203         opt_output(k
,&_init
, out
); 
 205 uint8_t rev_byte(uint8_t b
) { 
 206         b 
= (b 
& 0xF0) >> 4 | (b 
& 0x0F) << 4; 
 207         b 
= (b 
& 0xCC) >> 2 | (b 
& 0x33) << 2; 
 208         b 
= (b 
& 0xAA) >> 1 | (b 
& 0x55) << 1; 
 211 void opt_reverse_arraybytecpy(uint8_t* dest
, uint8_t *src
, size_t len
) 
 214         for( i 
=0; i
< len 
; i
++) 
 215                 dest
[i
] = rev_byte(src
[i
]); 
 218 void opt_doReaderMAC(uint8_t *cc_nr_p
, uint8_t *div_key_p
, uint8_t mac
[4]) 
 220         static uint8_t cc_nr
[12]; 
 222         opt_reverse_arraybytecpy(cc_nr
, cc_nr_p
,12); 
 223         uint8_t dest 
[]= {0,0,0,0,0,0,0,0}; 
 224         opt_MAC(div_key_p
,cc_nr
, dest
); 
 225         //The output MAC must also be reversed 
 226         opt_reverse_arraybytecpy(mac
, dest
,4); 
 229 void opt_doTagMAC(uint8_t *cc_p
, const uint8_t *div_key_p
, uint8_t mac
[4]) 
 231         static uint8_t cc_nr
[8+4+4]; 
 232         opt_reverse_arraybytecpy(cc_nr
, cc_p
,12); 
 234                         ((div_key_p
[0] ^ 0x4c) + 0xEC) & 0xFF,// l 
 235                         ((div_key_p
[0] ^ 0x4c) + 0x21) & 0xFF,// r 
 239         opt_suc(div_key_p
,&_init
,cc_nr
, 12,true); 
 240         uint8_t dest 
[]= {0,0,0,0}; 
 241         opt_output(div_key_p
,&_init
, dest
); 
 242         //The output MAC must also be reversed 
 243         opt_reverse_arraybytecpy(mac
, dest
,4); 
 248  * The tag MAC can be divided (both can, but no point in dividing the reader mac) into 
 249  * two functions, since the first 8 bytes are known, we can pre-calculate the state 
 250  * reached after feeding CC to the cipher. 
 253  * @return the cipher state 
 255 State 
opt_doTagMAC_1(uint8_t *cc_p
, const uint8_t *div_key_p
) 
 257         static uint8_t cc_nr
[8]; 
 258         opt_reverse_arraybytecpy(cc_nr
, cc_p
,8); 
 260                         ((div_key_p
[0] ^ 0x4c) + 0xEC) & 0xFF,// l 
 261                         ((div_key_p
[0] ^ 0x4c) + 0x21) & 0xFF,// r 
 265         opt_suc(div_key_p
,&_init
,cc_nr
, 8,false); 
 269  * The second part of the tag MAC calculation, since the CC is already calculated into the state, 
 270  * this function is fed only the NR, and internally feeds the remaining 32 0-bits to generate the tag 
 272  * @param _init - precalculated cipher state 
 273  * @param nr - the reader challenge 
 274  * @param mac - where to store the MAC 
 275  * @param div_key_p - the key to use 
 277 void opt_doTagMAC_2(State _init
,  uint8_t* nr
, uint8_t mac
[4], const uint8_t* div_key_p
) 
 279         static uint8_t _nr 
[4]; 
 280         opt_reverse_arraybytecpy(_nr
, nr
, 4); 
 281         opt_suc(div_key_p
,&_init
,_nr
, 4, true); 
 282         //opt_suc(div_key_p,&_init,nr, 4, false); 
 283         uint8_t dest 
[]= {0,0,0,0}; 
 284         opt_output(div_key_p
,&_init
, dest
); 
 285         //The output MAC must also be reversed 
 286         opt_reverse_arraybytecpy(mac
, dest
,4);