1 /* bmw_large_speed.c */
3 This file is part of the ARM-Crypto-Lib.
4 Copyright (C) 2006-2010 Daniel Otte (daniel.otte@rub.de)
6 This program is free software: you can redistribute it and/or modify
7 it under the terms of the GNU General Public License as published by
8 the Free Software Foundation, either version 3 of the License, or
9 (at your option) any later version.
11 This program 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
14 GNU General Public License for more details.
16 You should have received a copy of the GNU General Public License
17 along with this program. If not, see <http://www.gnu.org/licenses/>.
22 * \email daniel.otte@rub.de
24 * \license GPLv3 or later
30 #include "bmw_large.h"
32 #define SHL64(a,n) ((a)<<(n))
33 #define SHR64(a,n) ((a)>>(n))
34 #define ROTL64(a,n) (((a)<<(n))|((a)>>(64-(n))))
35 #define ROTR64(a,n) (((a)>>(n))|((a)<<(64-(n))))
43 void ctx_dump(const bmw_large_ctx_t* ctx){
45 cli_putstr("\r\n==== ctx dump ====");
47 cli_putstr("\r\n h[");
50 cli_hexdump_rev(&(ctx->h[i]), 8);
52 cli_putstr("\r\n counter = ");
53 cli_hexdump(&(ctx->counter), 4);
56 void dump_x(const uint64_t* q, uint8_t elements, char x){
58 cli_putstr("\r\n==== ");
60 cli_putstr(" dump ====");
61 for(i=0; i<elements;++i){
67 cli_hexdump_rev(&(q[i]), 8);
75 #define S64_0(x) ( (SHR64((x), 1)) ^ \
80 #define S64_1(x) ( (SHR64((x), 1)) ^ \
85 #define S64_2(x) ( (SHR64((x), 2)) ^ \
90 #define S64_3(x) ( (SHR64((x), 2)) ^ \
95 #define S64_4(x) ( (SHR64((x), 1)) ^ x)
97 #define S64_5(x) ( (SHR64((x), 2)) ^ x)
99 #define R64_1(x) (ROTL64((x), 5))
100 #define R64_2(x) (ROTL64((x), 11))
101 #define R64_3(x) (ROTL64((x), 27))
102 #define R64_4(x) (ROTL64((x), 32))
103 #define R64_5(x) (ROTR64((x), 27))
104 #define R64_6(x) (ROTR64((x), 21))
105 #define R64_7(x) (ROTR64((x), 11))
108 #define K 0x0555555555555555LL
109 #define MASK 0xFFFFFFFFFFFFFFFFLL
111 uint64_t k_lut[] PROGMEM = {
112 16LL*K, 17LL*K, 18LL*K, 19LL*K,
113 20LL*K, 21LL*K, 22LL*K, 23LL*K,
114 24LL*K, 25LL*K, 26LL*K, 27LL*K,
115 28LL*K, 29LL*K, 30LL*K, 31LL*K };
117 /* the same as above but precomputed to avoid compiler warnings */
120 0x5555555555555550LL, 0x5aaaaaaaaaaaaaa5LL, 0x5ffffffffffffffaLL,
121 0x655555555555554fLL, 0x6aaaaaaaaaaaaaa4LL, 0x6ffffffffffffff9LL,
122 0x755555555555554eLL, 0x7aaaaaaaaaaaaaa3LL, 0x7ffffffffffffff8LL,
123 0x855555555555554dLL, 0x8aaaaaaaaaaaaaa2LL, 0x8ffffffffffffff7LL,
124 0x955555555555554cLL, 0x9aaaaaaaaaaaaaa1LL, 0x9ffffffffffffff6LL,
125 0xa55555555555554bLL };
128 uint64_t bmw_large_expand1(uint8_t j, const uint64_t* q, const void* m, const void* h){
130 /* r = 0x0555555555555555LL*(j+16); */
131 r = ( ROTL64(((uint64_t*)m)[(j)&0xf], ((j+ 0)&0xf)+1)
132 + ROTL64(((uint64_t*)m)[(j+3)&0xf], ((j+ 3)&0xf)+1)
134 - ROTL64(((uint64_t*)m)[(j+10)&0xf],((j+10)&0xf)+1)
135 ) ^ ((uint64_t*)h)[(j+7)&0xf];
136 r += S64_1(q[j+ 0]) + S64_2(q[j+ 1]) + S64_3(q[j+ 2]) + S64_0(q[j+ 3]) +
137 S64_1(q[j+ 4]) + S64_2(q[j+ 5]) + S64_3(q[j+ 6]) + S64_0(q[j+ 7]) +
138 S64_1(q[j+ 8]) + S64_2(q[j+ 9]) + S64_3(q[j+10]) + S64_0(q[j+11]) +
139 S64_1(q[j+12]) + S64_2(q[j+13]) + S64_3(q[j+14]) + S64_0(q[j+15]);
145 uint64_t bmw_large_expand2(uint8_t j, const uint64_t* q, const void* m, const void* h){
147 r = ( ROTL64(((uint64_t*)m)[(j)&0xf], ((j+ 0)&0xf)+1)
148 + ROTL64(((uint64_t*)m)[(j+3)&0xf], ((j+ 3)&0xf)+1)
150 - ROTL64(((uint64_t*)m)[(j+10)&0xf],((j+10)&0xf)+1)
151 ) ^ ((uint64_t*)h)[(j+7)&0xf];
152 r += (q[j+ 0]) + R64_1(q[j+ 1]) + (q[j+ 2]) + R64_2(q[j+ 3]) +
153 (q[j+ 4]) + R64_3(q[j+ 5]) + (q[j+ 6]) + R64_4(q[j+ 7]) +
154 (q[j+ 8]) + R64_5(q[j+ 9]) + (q[j+10]) + R64_6(q[j+11]) +
155 (q[j+12]) + R64_7(q[j+13]) + S64_4(q[j+14]) + S64_5(q[j+15]);
161 void bmw_large_f0(uint64_t* q, const uint64_t* h, const void* m){
164 ((uint64_t*)h)[i] ^= ((uint64_t*)m)[i];
166 // dump_x(t, 16, 'T');
167 q[ 0] = (h[ 5] - h[ 7] + h[10] + h[13] + h[14]);
168 q[ 1] = (h[ 6] - h[ 8] + h[11] + h[14] - h[15]);
169 q[ 2] = (h[ 0] + h[ 7] + h[ 9] - h[12] + h[15]);
170 q[ 3] = (h[ 0] - h[ 1] + h[ 8] - h[10] + h[13]);
171 q[ 4] = (h[ 1] + h[ 2] + h[ 9] - h[11] - h[14]);
172 q[ 5] = (h[ 3] - h[ 2] + h[10] - h[12] + h[15]);
173 q[ 6] = (h[ 4] - h[ 0] - h[ 3] - h[11] + h[13]);
174 q[ 7] = (h[ 1] - h[ 4] - h[ 5] - h[12] - h[14]);
175 q[ 8] = (h[ 2] - h[ 5] - h[ 6] + h[13] - h[15]);
176 q[ 9] = (h[ 0] - h[ 3] + h[ 6] - h[ 7] + h[14]);
177 q[10] = (h[ 8] - h[ 1] - h[ 4] - h[ 7] + h[15]);
178 q[11] = (h[ 8] - h[ 0] - h[ 2] - h[ 5] + h[ 9]);
179 q[12] = (h[ 1] + h[ 3] - h[ 6] - h[ 9] + h[10]);
180 q[13] = (h[ 2] + h[ 4] + h[ 7] + h[10] + h[11]);
181 q[14] = (h[ 3] - h[ 5] + h[ 8] - h[11] - h[12]);
182 q[15] = (h[12] - h[ 4] - h[ 6] - h[ 9] + h[13]);
184 q[ 0] = S64_0(q[ 0]); q[ 1] = S64_1(q[ 1]); q[ 2] = S64_2(q[ 2]); q[ 3] = S64_3(q[ 3]); q[ 4] = S64_4(q[ 4]);
185 q[ 5] = S64_0(q[ 5]); q[ 6] = S64_1(q[ 6]); q[ 7] = S64_2(q[ 7]); q[ 8] = S64_3(q[ 8]); q[ 9] = S64_4(q[ 9]);
186 q[10] = S64_0(q[10]); q[11] = S64_1(q[11]); q[12] = S64_2(q[12]); q[13] = S64_3(q[13]); q[14] = S64_4(q[14]);
187 q[15] = S64_0(q[15]);
190 q[(i+15)&15] += ((uint64_t*)h)[i] ^= ((uint64_t*)m)[i];
196 void bmw_large_f1(uint64_t* q, const void* m, const uint64_t* h){
198 q[16] = bmw_large_expand1(0, q, m, h);
199 q[17] = bmw_large_expand1(1, q, m, h);
201 q[16+i] = bmw_large_expand2(i, q, m, h);
206 void bmw_large_f2(uint64_t* h, const uint64_t* q, const uint64_t* m){
209 xl = q[16] ^ q[17] ^ q[18] ^ q[19] ^ q[20] ^ q[21] ^ q[22] ^ q[23];
210 xh = xl ^ q[24] ^ q[25] ^ q[26] ^ q[27] ^ q[28] ^ q[29] ^ q[30] ^ q[31];
212 cli_putstr("\r\n XL = ");
213 cli_hexdump_rev(&xl, 8);
214 cli_putstr("\r\n XH = ");
215 cli_hexdump_rev(&xh, 8);
218 h[0] = (SHL64(xh, 5) ^ SHR64(q[16], 5) ^ m[ 0]) + (xl ^ q[24] ^ q[ 0]);
219 h[1] = (SHR64(xh, 7) ^ SHL64(q[17], 8) ^ m[ 1]) + (xl ^ q[25] ^ q[ 1]);
220 h[2] = (SHR64(xh, 5) ^ SHL64(q[18], 5) ^ m[ 2]) + (xl ^ q[26] ^ q[ 2]);
221 h[3] = (SHR64(xh, 1) ^ SHL64(q[19], 5) ^ m[ 3]) + (xl ^ q[27] ^ q[ 3]);
222 h[4] = (SHR64(xh, 3) ^ q[20] ^ m[ 4]) + (xl ^ q[28] ^ q[ 4]);
223 h[5] = (SHL64(xh, 6) ^ SHR64(q[21], 6) ^ m[ 5]) + (xl ^ q[29] ^ q[ 5]);
224 h[6] = (SHR64(xh, 4) ^ SHL64(q[22], 6) ^ m[ 6]) + (xl ^ q[30] ^ q[ 6]);
225 h[7] = (SHR64(xh,11) ^ SHL64(q[23], 2) ^ m[ 7]) + (xl ^ q[31] ^ q[ 7]);
227 h[ 8] = ROTL64(h[4], 9) + (xh ^ q[24] ^ m[ 8]) + (SHL64(xl, 8) ^ q[23] ^ q[ 8]);
228 h[ 9] = ROTL64(h[5], 10) + (xh ^ q[25] ^ m[ 9]) + (SHR64(xl, 6) ^ q[16] ^ q[ 9]);
229 h[10] = ROTL64(h[6], 11) + (xh ^ q[26] ^ m[10]) + (SHL64(xl, 6) ^ q[17] ^ q[10]);
230 h[11] = ROTL64(h[7], 12) + (xh ^ q[27] ^ m[11]) + (SHL64(xl, 4) ^ q[18] ^ q[11]);
231 h[12] = ROTL64(h[0], 13) + (xh ^ q[28] ^ m[12]) + (SHR64(xl, 3) ^ q[19] ^ q[12]);
232 h[13] = ROTL64(h[1], 14) + (xh ^ q[29] ^ m[13]) + (SHR64(xl, 4) ^ q[20] ^ q[13]);
233 h[14] = ROTL64(h[2], 15) + (xh ^ q[30] ^ m[14]) + (SHR64(xl, 7) ^ q[21] ^ q[14]);
234 h[15] = ROTL64(h[3], 16) + (xh ^ q[31] ^ m[15]) + (SHR64(xl, 2) ^ q[22] ^ q[15]);
237 void bmw_large_nextBlock(bmw_large_ctx_t* ctx, const void* block){
239 dump_x(block, 16, 'M');
240 bmw_large_f0(q, ctx->h, block);
242 bmw_large_f1(q, block, ctx->h);
244 bmw_large_f2(ctx->h, q, block);
249 void bmw_large_lastBlock(bmw_large_ctx_t* ctx, const void* block, uint16_t length_b){
251 while(length_b >= BMW_LARGE_BLOCKSIZE){
252 bmw_large_nextBlock(ctx, block);
253 length_b -= BMW_LARGE_BLOCKSIZE;
254 block = (uint8_t*)block + BMW_LARGE_BLOCKSIZE_B;
256 memset(buffer, 0, 128);
257 memcpy(buffer, block, (length_b+7)/8);
258 buffer[length_b>>3] |= 0x80 >> (length_b&0x07);
259 if(length_b+1>128*8-64){
260 bmw_large_nextBlock(ctx, buffer);
261 memset(buffer, 0, 128-8);
264 *((uint64_t*)&(buffer[128-8])) = (uint64_t)(ctx->counter*1024LL)+(uint64_t)length_b;
265 bmw_large_nextBlock(ctx, buffer);
268 memset(buffer, 0xaa, 128);
270 buffer[8*i] = i + 0xa0;
272 bmw_large_f0(q, (uint64_t*)buffer, ctx->h);
273 bmw_large_f1(q, ctx->h, (uint64_t*)buffer);
274 bmw_large_f2((uint64_t*)buffer, q, ctx->h);
275 memcpy(ctx->h, buffer, 128);
278 void bmw384_init(bmw384_ctx_t* ctx){
280 ctx->h[0] = 0x0001020304050607LL;
282 ctx->h[i] = ctx->h[i-1]+ 0x0808080808080808LL;
285 ctx->h[6] = 0x3031323324353637LL;
291 void bmw512_init(bmw512_ctx_t* ctx){
293 ctx->h[0] = 0x8081828384858687LL;
295 ctx->h[i] = ctx->h[i-1]+ 0x0808080808080808LL;
301 void bmw384_nextBlock(bmw384_ctx_t* ctx, const void* block){
302 bmw_large_nextBlock(ctx, block);
305 void bmw512_nextBlock(bmw512_ctx_t* ctx, const void* block){
306 bmw_large_nextBlock(ctx, block);
309 void bmw384_lastBlock(bmw384_ctx_t* ctx, const void* block, uint16_t length_b){
310 bmw_large_lastBlock(ctx, block, length_b);
313 void bmw512_lastBlock(bmw512_ctx_t* ctx, const void* block, uint16_t length_b){
314 bmw_large_lastBlock(ctx, block, length_b);
317 void bmw384_ctx2hash(void* dest, const bmw384_ctx_t* ctx){
318 memcpy(dest, &(ctx->h[10]), 384/8);
321 void bmw512_ctx2hash(void* dest, const bmw512_ctx_t* ctx){
322 memcpy(dest, &(ctx->h[8]), 512/8);
325 void bmw384(void* dest, const void* msg, uint32_t length_b){
328 while(length_b>=BMW_LARGE_BLOCKSIZE){
329 bmw_large_nextBlock(&ctx, msg);
330 length_b -= BMW_LARGE_BLOCKSIZE;
331 msg = (uint8_t*)msg + BMW_LARGE_BLOCKSIZE_B;
333 bmw_large_lastBlock(&ctx, msg, length_b);
334 bmw384_ctx2hash(dest, &ctx);
337 void bmw512(void* dest, const void* msg, uint32_t length_b){
340 while(length_b>=BMW_LARGE_BLOCKSIZE){
341 bmw_large_nextBlock(&ctx, msg);
342 length_b -= BMW_LARGE_BLOCKSIZE;
343 msg = (uint8_t*)msg + BMW_LARGE_BLOCKSIZE_B;
345 bmw_large_lastBlock(&ctx, msg, length_b);
346 bmw512_ctx2hash(dest, &ctx);