]>
cvs.zerfleddert.de Git - proxmark3-svn/blob - armsrc/optimized_cipher.c
ee9d556817772567b5ff3c8b6c991e8c79748ff4
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, or, at your option, any later version.
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"
69 #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 #define opt_B(s) (((s->b >> 6) ^ (s->b >> 5) ^ (s->b >> 4) ^ (s->b)) & 0x1)
73 #define opt__select(x,y,r) (4 & (((r & (r << 2)) >> 5) ^ ((r & ~(r << 2)) >> 4) ^ ( (r | r << 2) >> 3)))\
74 |(2 & (((r | r << 2) >> 6) ^ ( (r | r << 2) >> 1) ^ (r >> 5) ^ r ^ ((x^y) << 1)))\
75 |(1 & (((r & ~(r << 2)) >> 4) ^ ((r & (r << 2)) >> 3) ^ r ^ x))
78 * Some background on the expression above can be found here...
79 uint8_t xopt__select(bool x, bool y, uint8_t r)
81 uint8_t r_ls2 = r << 2;
82 uint8_t r_and_ls2 = r & r_ls2;
83 uint8_t r_or_ls2 = r | r_ls2;
85 //r: r0 r1 r2 r3 r4 r5 r6 r7
86 //r_ls2: r2 r3 r4 r5 r6 r7 0 0
90 // uint8_t z0 = (r0 & r2) ^ (r1 & ~r3) ^ (r2 | r4); // <-- original
91 uint8_t z0 = (r_and_ls2 >> 5) ^ ((r & ~r_ls2) >> 4) ^ ( r_or_ls2 >> 3);
93 // uint8_t z1 = (r0 | r2) ^ ( r5 | r7) ^ r1 ^ r6 ^ x ^ y; // <-- original
94 uint8_t z1 = (r_or_ls2 >> 6) ^ ( r_or_ls2 >> 1) ^ (r >> 5) ^ r ^ ((x^y) << 1);
96 // uint8_t z2 = (r3 & ~r5) ^ (r4 & r6 ) ^ r7 ^ x; // <-- original
97 uint8_t z2 = ((r & ~r_ls2) >> 4) ^ (r_and_ls2 >> 3) ^ r ^ x;
99 return (z0 & 4) | (z1 & 2) | (z2 & 1);
103 void opt_successor(const uint8_t *k
, State
*s
, bool y
, State
*successor
) {
104 uint8_t Tt
= 1 & opt_T(s
);
106 successor
->t
= (s
->t
>> 1);
107 successor
->t
|= (Tt
^ (s
->r
>> 7 & 0x1) ^ (s
->r
>> 3 & 0x1)) << 15;
109 successor
->b
= s
->b
>> 1;
110 successor
->b
|= (opt_B(s
) ^ (s
->r
& 0x1)) << 7;
112 successor
->r
= (k
[opt__select(Tt
, y
, s
->r
)] ^ successor
->b
) + s
->l
;
113 successor
->l
= successor
->r
+ s
->r
;
116 void opt_suc(const uint8_t *k
, State
*s
, uint8_t *in
, uint8_t length
, bool add32Zeroes
) {
118 for (int i
= 0; i
< length
; i
++) {
120 head
= 1 & (in
[i
] >> 7);
121 opt_successor(k
, s
, head
, &x2
);
123 head
= 1 & (in
[i
] >> 6);
124 opt_successor(k
, &x2
, head
, s
);
126 head
= 1 & (in
[i
] >> 5);
127 opt_successor(k
, s
, head
, &x2
);
129 head
= 1 & (in
[i
] >> 4);
130 opt_successor(k
, &x2
, head
, s
);
132 head
= 1 & (in
[i
] >> 3);
133 opt_successor(k
, s
, head
, &x2
);
135 head
= 1 & (in
[i
] >> 2);
136 opt_successor(k
, &x2
, head
, s
);
138 head
= 1 & (in
[i
] >> 1);
139 opt_successor(k
, s
, head
, &x2
);
142 opt_successor(k
, &x2
, head
, s
);
144 //For tag MAC, an additional 32 zeroes
146 for(int i
= 0; i
< 16; i
++) {
147 opt_successor(k
, s
, 0, &x2
);
148 opt_successor(k
, &x2
, 0, s
);
153 void opt_output(const uint8_t *k
, State
*s
, uint8_t *buffer
) {
154 State temp
= {0, 0, 0, 0};
155 for (uint8_t times
= 0; times
< 4; times
++) {
157 bout
|= (s
->r
& 0x4) << 5;
158 opt_successor(k
, s
, 0, &temp
);
159 bout
|= (temp
.r
& 0x4) << 4;
160 opt_successor(k
, &temp
, 0, s
);
161 bout
|= (s
->r
& 0x4) << 3;
162 opt_successor(k
, s
, 0, &temp
);
163 bout
|= (temp
.r
& 0x4) << 2;
164 opt_successor(k
, &temp
, 0, s
);
165 bout
|= (s
->r
& 0x4) << 1;
166 opt_successor(k
, s
, 0, &temp
);
167 bout
|= (temp
.r
& 0x4) ;
168 opt_successor(k
, &temp
, 0, s
);
169 bout
|= (s
->r
& 0x4) >> 1;
170 opt_successor(k
, s
, 0, &temp
);
171 bout
|= (temp
.r
& 0x4) >> 2;
172 opt_successor(k
, &temp
, 0, s
);
173 buffer
[times
] = bout
;
177 void opt_MAC(uint8_t *k
, uint8_t *input
, uint8_t *out
) {
179 ((k
[0] ^ 0x4c) + 0xEC) & 0xFF,// l
180 ((k
[0] ^ 0x4c) + 0x21) & 0xFF,// r
185 opt_suc(k
, &_init
, input
, 12, false);
187 opt_output(k
, &_init
, out
);
190 uint8_t rev_byte(uint8_t b
) {
191 b
= (b
& 0xF0) >> 4 | (b
& 0x0F) << 4;
192 b
= (b
& 0xCC) >> 2 | (b
& 0x33) << 2;
193 b
= (b
& 0xAA) >> 1 | (b
& 0x55) << 1;
197 void opt_reverse_arraybytecpy(uint8_t *dest
, uint8_t *src
, size_t len
) {
198 for (size_t i
= 0; i
< len
; i
++) {
199 dest
[i
] = rev_byte(src
[i
]);
203 void opt_doReaderMAC(uint8_t *cc_nr_p
, uint8_t *div_key_p
, uint8_t mac
[4]) {
204 static uint8_t cc_nr
[12];
205 opt_reverse_arraybytecpy(cc_nr
, cc_nr_p
, 12);
206 uint8_t dest
[] = {0, 0, 0, 0, 0, 0, 0, 0};
207 opt_MAC(div_key_p
, cc_nr
, dest
);
208 //The output MAC must also be reversed
209 opt_reverse_arraybytecpy(mac
, dest
, 4);
213 void opt_doTagMAC(uint8_t *cc_p
, const uint8_t *div_key_p
, uint8_t mac
[4]) {
214 static uint8_t cc_nr
[8+4+4];
215 opt_reverse_arraybytecpy(cc_nr
, cc_p
, 12);
217 ((div_key_p
[0] ^ 0x4c) + 0xEC) & 0xFF,// l
218 ((div_key_p
[0] ^ 0x4c) + 0x21) & 0xFF,// r
222 opt_suc(div_key_p
, &_init
,cc_nr
, 12, true);
223 uint8_t dest
[] = {0, 0, 0, 0};
224 opt_output(div_key_p
, &_init
, dest
);
225 //The output MAC must also be reversed
226 opt_reverse_arraybytecpy(mac
, dest
, 4);
231 * The tag MAC can be divided (both can, but no point in dividing the reader mac) into
232 * two functions, since the first 8 bytes are known, we can pre-calculate the state
233 * reached after feeding CC to the cipher.
236 * @return the cipher state
238 State
opt_doTagMAC_1(uint8_t *cc_p
, const uint8_t *div_key_p
) {
239 static uint8_t cc_nr
[8];
240 opt_reverse_arraybytecpy(cc_nr
, cc_p
, 8);
242 ((div_key_p
[0] ^ 0x4c) + 0xEC) & 0xFF,// l
243 ((div_key_p
[0] ^ 0x4c) + 0x21) & 0xFF,// r
247 opt_suc(div_key_p
, &_init
, cc_nr
, 8, false);
252 * The second part of the tag MAC calculation, since the CC is already calculated into the state,
253 * this function is fed only the NR, and internally feeds the remaining 32 0-bits to generate the tag
255 * @param _init - precalculated cipher state
256 * @param nr - the reader challenge
257 * @param mac - where to store the MAC
258 * @param div_key_p - the key to use
260 void opt_doTagMAC_2(State _init
, uint8_t *nr
, uint8_t mac
[4], const uint8_t *div_key_p
) {
261 static uint8_t _nr
[4];
262 opt_reverse_arraybytecpy(_nr
, nr
, 4);
263 opt_suc(div_key_p
, &_init
, _nr
, 4, true);
264 //opt_suc(div_key_p, &_init,nr, 4, false);
265 uint8_t dest
[] = {0, 0, 0, 0};
266 opt_output(div_key_p
, &_init
, dest
);
267 //The output MAC must also be reversed
268 opt_reverse_arraybytecpy(mac
, dest
, 4);