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) shiftl64(a,n)
33 #define SHR64(a,n) shiftr64(a,n)
34 #define ROTL64(a,n) rotl64(a,n)
35 #define ROTR64(a,n) rotr64(a,n)
45 void ctx_dump(const bmw_large_ctx_t* ctx){
47 cli_putstr("\r\n==== ctx dump ====");
49 cli_putstr("\r\n h[");
52 cli_hexdump_rev(&(ctx->h[i]), 8);
54 cli_putstr("\r\n counter = ");
55 cli_hexdump(&(ctx->counter), 4);
58 void dump_x(const uint64_t* q, uint8_t elements, char x){
60 cli_putstr("\r\n==== ");
62 cli_putstr(" dump ====");
63 for(i=0; i<elements;++i){
69 cli_hexdump_rev(&(q[i]), 8);
78 uint64_t rotl64(uint64_t a, uint8_t r){
79 return (a<<r)|(a>>(64-r));
83 uint64_t rotr64(uint64_t a, uint8_t r){
84 return (a>>r)|(a<<(64-r));
88 uint64_t shiftl64(uint64_t a, uint8_t r){
93 uint64_t shiftr64(uint64_t a, uint8_t r){
98 uint64_t bmw_large_s0(uint64_t x){
108 uint64_t bmw_large_s1(uint64_t x){
118 uint64_t bmw_large_s2(uint64_t x){
128 uint64_t bmw_large_s3(uint64_t x){
138 uint64_t bmw_large_s4(uint64_t x){
146 uint64_t bmw_large_s5(uint64_t x){
154 uint64_t bmw_large_r1(uint64_t x){
161 uint64_t bmw_large_r2(uint64_t x){
168 uint64_t bmw_large_r3(uint64_t x){
175 uint64_t bmw_large_r4(uint64_t x){
182 uint64_t bmw_large_r5(uint64_t x){
184 r = ROTR64(x, 64-37);
189 uint64_t bmw_large_r6(uint64_t x){
191 r = ROTR64(x, 64-43);
196 uint64_t bmw_large_r7(uint64_t x){
198 r = ROTR64(x, 64-53);
202 #define K 0x0555555555555555LL
203 #define MASK 0xFFFFFFFFFFFFFFFFLL
205 uint64_t k_lut[] PROGMEM = {
206 16LL*K, 17LL*K, 18LL*K, 19LL*K,
207 20LL*K, 21LL*K, 22LL*K, 23LL*K,
208 24LL*K, 25LL*K, 26LL*K, 27LL*K,
209 28LL*K, 29LL*K, 30LL*K, 31LL*K };
211 /* the same as above but precomputed to avoid compiler warnings */
214 0x5555555555555550LL, 0x5aaaaaaaaaaaaaa5LL, 0x5ffffffffffffffaLL,
215 0x655555555555554fLL, 0x6aaaaaaaaaaaaaa4LL, 0x6ffffffffffffff9LL,
216 0x755555555555554eLL, 0x7aaaaaaaaaaaaaa3LL, 0x7ffffffffffffff8LL,
217 0x855555555555554dLL, 0x8aaaaaaaaaaaaaa2LL, 0x8ffffffffffffff7LL,
218 0x955555555555554cLL, 0x9aaaaaaaaaaaaaa1LL, 0x9ffffffffffffff6LL,
219 0xa55555555555554bLL };
222 uint64_t bmw_large_expand1(uint8_t j, const uint64_t* q, const void* m, const void* h){
223 uint64_t(*s[])(uint64_t) = {bmw_large_s1, bmw_large_s2, bmw_large_s3, bmw_large_s0};
230 /* r = 0x0555555555555555LL*(j+16); */
236 a += ( ROTL64(((uint64_t*)m)[(j)&0xf], ((j+ 0)&0xf)+1)
237 + ROTL64(((uint64_t*)m)[(j+3)&0xf], ((j+ 3)&0xf)+1)
239 - ROTL64(((uint64_t*)m)[(j+10)&0xf],((j+10)&0xf)+1)
240 ) ^ ((uint64_t*)h)[(j+7)&0xf];
242 a += ((uint64_t*)m)[j&0xf];
243 a += ((uint64_t*)m)[(j+3)&0xf];
244 a -= ((uint64_t*)m)[(j+10)&0xf];
251 uint64_t bmw_large_expand2(uint8_t j, const uint64_t* q, const void* m, const void* h){
252 uint64_t(*rf[])(uint64_t) = {bmw_large_r1, bmw_large_r2, bmw_large_r3,
253 bmw_large_r4, bmw_large_r5, bmw_large_r6,
261 /* r = 0x0555555555555555LL*(j+16); */
263 for(i=0; i<14; i+=2){
266 for(i=0; i<14; i+=2){
267 a += rf[i/2](q[j+i+1]);
270 a += bmw_large_s4(q[j+14]);
271 a += bmw_large_s5(q[j+15]);
273 a += bmw_large_s5(q[j+14]);
274 a += bmw_large_s4(q[j+15]);
280 cli_putstr("\n+++++++++ expand_2 ++++++++++++");
283 t=ROTL64(((uint64_t*)m)[j], ((j+ 0)&0xf)+1);
285 t=ROTL64(((uint64_t*)m)[j], ((j+ 3)&0xf)+1);
287 t=ROTL64(((uint64_t*)m)[j], ((j+ 0)&0xf)+1);
292 a += ( ROTL64(((uint64_t*)m)[(j)&0xf], ((j+ 0)&0xf)+1)
293 + ROTL64(((uint64_t*)m)[(j+3)&0xf], ((j+ 3)&0xf)+1)
295 - ROTL64(((uint64_t*)m)[(j+10)&0xf],((j+10)&0xf)+1)
296 ) ^ ((uint64_t*)h)[(j+7)&0xf];
298 a += ((uint64_t*)m)[j&0xf];
299 a += ((uint64_t*)m)[(j+3)&0xf];
300 a -= ((uint64_t*)m)[(j+10)&0xf];
307 /* to understand this implementation take a look at f0-opt-table.txt */
308 static uint16_t hack_table[5] = { 0x0311, 0xDDB3, 0x2A79, 0x07AA, 0x51C2 };
309 static uint8_t offset_table[5] = { 4+16, 6+16, 9+16, 12+16, 13+16 };
313 void bmw_large_f0(uint64_t* q, const uint64_t* h, const void* m){
316 uint64_t(*s[])(uint64_t)={ bmw_large_s0, bmw_large_s1, bmw_large_s2,
317 bmw_large_s3, bmw_large_s4 };
319 ((uint64_t*)h)[i] ^= ((uint64_t*)m)[i];
327 hack_reg = hack_table[c];
344 ((uint64_t*)h)[i] ^= ((uint64_t*)m)[i];
347 q[i] += h[(i+1)&0xf];
351 #endif /* F0_HACK==2 */
355 uint8_t f0_lut[] PROGMEM ={
356 5<<1, ( 7<<1)+1, (10<<1)+0, (13<<1)+0, (14<<1)+0,
357 6<<1, ( 8<<1)+1, (11<<1)+0, (14<<1)+0, (15<<1)+1,
358 0<<1, ( 7<<1)+0, ( 9<<1)+0, (12<<1)+1, (15<<1)+0,
359 0<<1, ( 1<<1)+1, ( 8<<1)+0, (10<<1)+1, (13<<1)+0,
360 1<<1, ( 2<<1)+0, ( 9<<1)+0, (11<<1)+1, (14<<1)+1,
361 3<<1, ( 2<<1)+1, (10<<1)+0, (12<<1)+1, (15<<1)+0,
362 4<<1, ( 0<<1)+1, ( 3<<1)+1, (11<<1)+1, (13<<1)+0,
363 1<<1, ( 4<<1)+1, ( 5<<1)+1, (12<<1)+1, (14<<1)+1,
364 2<<1, ( 5<<1)+1, ( 6<<1)+1, (13<<1)+0, (15<<1)+1,
365 0<<1, ( 3<<1)+1, ( 6<<1)+0, ( 7<<1)+1, (14<<1)+0,
366 8<<1, ( 1<<1)+1, ( 4<<1)+1, ( 7<<1)+1, (15<<1)+0,
367 8<<1, ( 0<<1)+1, ( 2<<1)+1, ( 5<<1)+1, ( 9<<1)+0,
368 1<<1, ( 3<<1)+0, ( 6<<1)+1, ( 9<<1)+1, (10<<1)+0,
369 2<<1, ( 4<<1)+0, ( 7<<1)+0, (10<<1)+0, (11<<1)+0,
370 3<<1, ( 5<<1)+1, ( 8<<1)+0, (11<<1)+1, (12<<1)+1,
371 12<<1, ( 4<<1)+1, ( 6<<1)+1, ( 9<<1)+1, (13<<1)+0
375 void bmw_large_f0(uint64_t* q, const uint64_t* h, const void* m){
376 uint8_t i,j=-1,v,sign,l=0;
377 uint64_t(*s[])(uint64_t)={ bmw_large_s0, bmw_large_s1, bmw_large_s2,
378 bmw_large_s3, bmw_large_s4 };
380 ((uint64_t*)h)[i] ^= ((uint64_t*)m)[i];
383 // memset(q, 0, 4*16);
384 for(i=0; i<5*16; ++i){
385 v = pgm_read_byte(f0_lut+i);
406 ((uint64_t*)h)[i] ^= ((uint64_t*)m)[i];
409 q[i] += h[(i+1)&0xf];
413 #endif /* F0_HACK==1 */
417 void bmw_large_f0(uint64_t* q, const uint64_t* h, const void* m){
419 uint64_t(*s[])(uint64_t)={ bmw_large_s0, bmw_large_s1, bmw_large_s2,
420 bmw_large_s3, bmw_large_s4 };
422 ((uint64_t*)h)[i] ^= ((uint64_t*)m)[i];
424 // dump_x(t, 16, 'T');
425 q[ 0] = (h[ 5] - h[ 7] + h[10] + h[13] + h[14]);
426 q[ 1] = (h[ 6] - h[ 8] + h[11] + h[14] - h[15]);
427 q[ 2] = (h[ 0] + h[ 7] + h[ 9] - h[12] + h[15]);
428 q[ 3] = (h[ 0] - h[ 1] + h[ 8] - h[10] + h[13]);
429 q[ 4] = (h[ 1] + h[ 2] + h[ 9] - h[11] - h[14]);
430 q[ 5] = (h[ 3] - h[ 2] + h[10] - h[12] + h[15]);
431 q[ 6] = (h[ 4] - h[ 0] - h[ 3] - h[11] + h[13]);
432 q[ 7] = (h[ 1] - h[ 4] - h[ 5] - h[12] - h[14]);
433 q[ 8] = (h[ 2] - h[ 5] - h[ 6] + h[13] - h[15]);
434 q[ 9] = (h[ 0] - h[ 3] + h[ 6] - h[ 7] + h[14]);
435 q[10] = (h[ 8] - h[ 1] - h[ 4] - h[ 7] + h[15]);
436 q[11] = (h[ 8] - h[ 0] - h[ 2] - h[ 5] + h[ 9]);
437 q[12] = (h[ 1] + h[ 3] - h[ 6] - h[ 9] + h[10]);
438 q[13] = (h[ 2] + h[ 4] + h[ 7] + h[10] + h[11]);
439 q[14] = (h[ 3] - h[ 5] + h[ 8] - h[11] - h[12]);
440 q[15] = (h[12] - h[ 4] - h[ 6] - h[ 9] + h[13]);
447 ((uint64_t*)h)[i] ^= ((uint64_t*)m)[i];
450 q[i] += h[(i+1)&0xf];
455 #endif /* F0_HACK==0 */
458 void bmw_large_f1(uint64_t* q, const void* m, const uint64_t* h){
460 q[16] = bmw_large_expand1(0, q, m, h);
461 q[17] = bmw_large_expand1(1, q, m, h);
463 q[16+i] = bmw_large_expand2(i, q, m, h);
468 void bmw_large_f2(uint64_t* h, const uint64_t* q, const void* m){
479 cli_putstr("\r\n XL = ");
480 cli_hexdump_rev(&xl, 4);
481 cli_putstr("\r\n XH = ");
482 cli_hexdump_rev(&xh, 4);
485 h[0] ^= SHL64(xh, 5) ^ SHR64(q[16], 5);
486 h[1] ^= SHR64(xh, 7) ^ SHL64(q[17], 8);
487 h[2] ^= SHR64(xh, 5) ^ SHL64(q[18], 5);
488 h[3] ^= SHR64(xh, 1) ^ SHL64(q[19], 5);
489 h[4] ^= SHR64(xh, 3) ^ q[20];
490 h[5] ^= SHL64(xh, 6) ^ SHR64(q[21], 6);
491 h[6] ^= SHR64(xh, 4) ^ SHL64(q[22], 6);
492 h[7] ^= SHR64(xh,11) ^ SHL64(q[23], 2);
494 h[i] += xl ^ q[24+i] ^ q[i];
497 h[8+i] ^= xh ^ q[24+i];
498 h[8+i] += ROTL64(h[(4+i)%8],i+9);
500 h[ 8] += SHL64(xl, 8) ^ q[23] ^ q[ 8];
501 h[ 9] += SHR64(xl, 6) ^ q[16] ^ q[ 9];
502 h[10] += SHL64(xl, 6) ^ q[17] ^ q[10];
503 h[11] += SHL64(xl, 4) ^ q[18] ^ q[11];
504 h[12] += SHR64(xl, 3) ^ q[19] ^ q[12];
505 h[13] += SHR64(xl, 4) ^ q[20] ^ q[13];
506 h[14] += SHR64(xl, 7) ^ q[21] ^ q[14];
507 h[15] += SHR64(xl, 2) ^ q[22] ^ q[15];
510 void bmw_large_nextBlock(bmw_large_ctx_t* ctx, const void* block){
512 dump_x(block, 16, 'M');
513 bmw_large_f0(q, ctx->h, block);
515 bmw_large_f1(q, block, ctx->h);
517 bmw_large_f2(ctx->h, q, block);
522 void bmw_large_lastBlock(bmw_large_ctx_t* ctx, const void* block, uint16_t length_b){
527 while(length_b >= BMW_LARGE_BLOCKSIZE){
528 bmw_large_nextBlock(ctx, block);
529 length_b -= BMW_LARGE_BLOCKSIZE;
530 block = (uint8_t*)block + BMW_LARGE_BLOCKSIZE_B;
532 memset(buffer.v8, 0, 128);
533 memcpy(buffer.v8, block, (length_b+7)/8);
534 buffer.v8[length_b>>3] |= 0x80 >> (length_b&0x07);
535 if(length_b+1>128*8-64){
536 bmw_large_nextBlock(ctx, buffer.v8);
537 memset(buffer.v8, 0, 128-8);
540 buffer.v64[15] = (uint64_t)(ctx->counter*1024LL)+(uint64_t)length_b;
541 bmw_large_nextBlock(ctx, buffer.v8);
545 memset(buffer.v8, 0xaa, 128);
547 buffer.v8[8*i] = i + 0xa0;
549 bmw_large_f0(q, buffer.v64, ctx->h);
550 bmw_large_f1(q, ctx->h, buffer.v64);
551 bmw_large_f2(buffer.v64, q, ctx->h);
552 memcpy(ctx->h, buffer.v8, 128);
556 void bmw384_init(bmw384_ctx_t* ctx){
558 ctx->h[0] = 0x0001020304050607LL;
560 ctx->h[i] = ctx->h[i-1]+ 0x0808080808080808LL;
563 ctx->h[6] = 0x3031323324353637LL;
569 void bmw512_init(bmw512_ctx_t* ctx){
571 ctx->h[0] = 0x8081828384858687LL;
573 ctx->h[i] = ctx->h[i-1]+ 0x0808080808080808LL;
579 void bmw384_nextBlock(bmw384_ctx_t* ctx, const void* block){
580 bmw_large_nextBlock(ctx, block);
583 void bmw512_nextBlock(bmw512_ctx_t* ctx, const void* block){
584 bmw_large_nextBlock(ctx, block);
587 void bmw384_lastBlock(bmw384_ctx_t* ctx, const void* block, uint16_t length_b){
588 bmw_large_lastBlock(ctx, block, length_b);
591 void bmw512_lastBlock(bmw512_ctx_t* ctx, const void* block, uint16_t length_b){
592 bmw_large_lastBlock(ctx, block, length_b);
595 void bmw384_ctx2hash(void* dest, const bmw384_ctx_t* ctx){
596 memcpy(dest, &(ctx->h[10]), 384/8);
599 void bmw512_ctx2hash(void* dest, const bmw512_ctx_t* ctx){
600 memcpy(dest, &(ctx->h[8]), 512/8);
603 void bmw384(void* dest, const void* msg, uint32_t length_b){
606 while(length_b>=BMW_LARGE_BLOCKSIZE){
607 bmw_large_nextBlock(&ctx, msg);
608 length_b -= BMW_LARGE_BLOCKSIZE;
609 msg = (uint8_t*)msg + BMW_LARGE_BLOCKSIZE_B;
611 bmw_large_lastBlock(&ctx, msg, length_b);
612 bmw384_ctx2hash(dest, &ctx);
615 void bmw512(void* dest, const void* msg, uint32_t length_b){
618 while(length_b>=BMW_LARGE_BLOCKSIZE){
619 bmw_large_nextBlock(&ctx, msg);
620 length_b -= BMW_LARGE_BLOCKSIZE;
621 msg = (uint8_t*)msg + BMW_LARGE_BLOCKSIZE_B;
623 bmw_large_lastBlock(&ctx, msg, length_b);
624 bmw512_ctx2hash(dest, &ctx);