blake/blake_large.c

   1 /* blake_large.c */
   2 /*
   3     This file is part of the AVR-Crypto-Lib.
   4     Copyright (C) 2009  Daniel Otte (daniel.otte@rub.de)
   5
   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.
  10
  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.
  15
  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/>.
  18 */
  19 /*
  20  * \file    blake_large.c
  21  * \author  Daniel Otte
  22  * \email   daniel.otte@rub.de
  23  * \date    2009-05-08
  24  * \license GPLv3 or later
  25  *
  26  */
  27
  28 #include <stdint.h>
  29 #include <string.h>
  30 #include <avr/pgmspace.h>
  31 #include "memxor.h"
  32 #include "blake_large.h"
  33 #include "blake_common.h"
  34
  35 uint64_t pgm_read_qword(void* p){
  36         union{
  37                 uint64_t v64;
  38                 uint32_t v32[2];
  39         }r;
  40         r.v32[0] = pgm_read_dword(p);
  41         r.v32[1] = pgm_read_dword((uint8_t*)p+4);
  42         return r.v64;
  43 }
  44
  45 static
  46 uint64_t blake_c[] PROGMEM = {
  47    0x243F6A8885A308D3LL, 0x13198A2E03707344LL,
  48    0xA4093822299F31D0LL, 0x082EFA98EC4E6C89LL,
  49    0x452821E638D01377LL, 0xBE5466CF34E90C6CLL,
  50    0xC0AC29B7C97C50DDLL, 0x3F84D5B5B5470917LL,
  51    0x9216D5D98979FB1BLL, 0xD1310BA698DFB5ACLL,
  52    0x2FFD72DBD01ADFB7LL, 0xB8E1AFED6A267E96LL,
  53    0xBA7C9045F12C7F99LL, 0x24A19947B3916CF7LL,
  54    0x0801F2E2858EFC16LL, 0x636920D871574E69LL
  55 };
  56
  57
  58
  59 #define ROTL64(a, n) (((a)<<(n))|((a)>>(64-(n))))
  60 #define ROTR64(a, n) (((a)>>(n))|((a)<<(64-(n))))
  61 #define CHANGE_ENDIAN32(a) (((a)<<24)| \
  62                             ((0x0000ff00&(a))<<8)| \
  63                                                     ((0x00ff0000&(a))>>8)| \
  64                                                     (a)>>24 )
  65
  66 void blake_large_expand(uint64_t* v, const blake_large_ctx_t* ctx){
  67         uint8_t i;
  68         memcpy(v, ctx->h, 8*8);
  69         for(i=0; i<8; ++i){
  70                 v[8+i] = pgm_read_qword(&(blake_c[i]));
  71         }
  72         memxor((uint8_t*)v+8, ctx->s, 4*8);
  73
  74 }
  75
  76 void blake_large_changeendian(void* dest, const void* src){
  77         uint8_t i;
  78         uint32_t tmp;
  79         for(i=0; i<32; i+=2){
  80                 tmp = CHANGE_ENDIAN32(((uint32_t*)src)[i]);
  81                 ((uint32_t*)dest)[i] = CHANGE_ENDIAN32(((uint32_t*)src)[i+1]);
  82                 ((uint32_t*)dest)[i+1] = tmp;
  83         }
  84 }
  85
  86 void blake_large_compress(uint64_t* v,const void* m){
  87         uint8_t r,i;
  88         uint8_t a,b,c,d, s0, s1;
  89         for(r=0; r<14; ++r){
  90                 for(i=0; i<8; ++i){
  91                         a = pgm_read_byte(blake_index_lut+4*i+0);
  92                         b = pgm_read_byte(blake_index_lut+4*i+1);
  93                         c = pgm_read_byte(blake_index_lut+4*i+2);
  94                         d = pgm_read_byte(blake_index_lut+4*i+3);
  95                         s0 = pgm_read_byte(blake_sigma+16*r+2*i+0);
  96                         s1 = pgm_read_byte(blake_sigma+16*r+2*i+1);
  97                         v[a] += v[b] + (((uint64_t*)m)[s0] ^ pgm_read_qword(&(blake_c[s1])));
  98                         v[d]  = ROTR64(v[d]^v[a], 32);
  99                         v[c] += v[d];
 100                         v[b]  = ROTR64(v[b]^v[c], 25);
 101                         v[a] += v[b] + (((uint64_t*)m)[s1] ^ pgm_read_qword(&(blake_c[s0])));
 102                         v[d]  = ROTR64(v[d]^v[a], 16);
 103                         v[c] += v[d];
 104                         v[b]  = ROTR64(v[b]^v[c], 11);
 105                 }
 106         }
 107 }
 108
 109 void blake_large_collapse(blake_large_ctx_t* ctx, uint64_t* v){
 110         uint8_t i;
 111         for(i=0; i<8; ++i){
 112                 ctx->h[i] ^= ctx->s[i%4] ^ v[i] ^ v[8+i];
 113         }
 114 }
 115
 116 void blake_large_nextBlock(blake_large_ctx_t* ctx, const void* msg){
 117         uint64_t v[16];
 118         uint64_t m[16];
 119         union {
 120                 uint64_t v64;
 121                 uint32_t v32[2];
 122         }ctr;
 123         blake_large_expand(v,ctx);
 124         ctx->counter++;
 125         ctr.v64 = ctx->counter*1024;
 126         v[12] ^= ctr.v64;
 127         v[13] ^= ctr.v64;
 128         blake_large_changeendian(m, msg);
 129         blake_large_compress(v, m);
 130         blake_large_collapse(ctx, v);
 131 }
 132
 133 void blake_large_lastBlock(blake_large_ctx_t* ctx, const void* msg, uint16_t length_b){
 134         while(length_b>=BLAKE_LARGE_BLOCKSIZE){
 135                 blake_large_nextBlock(ctx, msg);
 136                 msg = (uint8_t*)msg + BLAKE_LARGE_BLOCKSIZE_B;
 137                 length_b -= BLAKE_LARGE_BLOCKSIZE;
 138         }
 139         uint8_t buffer[128];
 140         uint64_t v[16];
 141         uint64_t ctr;
 142         ctr = ctx->counter*1024+length_b;
 143         memset(buffer, 0, 128);
 144         memcpy(buffer, msg, (length_b+7)/8);
 145         buffer[length_b/8] |= 0x80 >> (length_b&0x7);
 146         blake_large_changeendian(buffer, buffer);
 147         blake_large_expand(v, ctx);
 148         if(length_b>1024-128-2){
 149                 v[12] ^= ctr;
 150                 v[13] ^= ctr;
 151                 blake_large_compress(v, buffer);
 152                 blake_large_collapse(ctx, v);
 153                 memset(buffer, 0, 128-8);
 154                 blake_large_expand(v, ctx);
 155         } else {
 156                 if(length_b){
 157                         v[12] ^= ctr;
 158                         v[13] ^= ctr;
 159                 }
 160         }
 161         if(ctx->appendone)
 162                 buffer[128-16-8] |= 0x01;
 163         *((uint64_t*)(&(buffer[128-8]))) = ctr;
 164         blake_large_compress(v, buffer);
 165         blake_large_collapse(ctx, v);
 166
 167 }
 168
 169 uint64_t blake64_iv[] PROGMEM = {
 170     0x6A09E667F3BCC908LL, 0xBB67AE8584CAA73BLL,
 171     0x3C6EF372FE94F82BLL, 0xA54FF53A5F1D36F1LL,
 172     0x510E527FADE682D1LL, 0x9B05688C2B3E6C1FLL,
 173     0x1F83D9ABFB41BD6BLL, 0x5BE0CD19137E2179LL
 174 };
 175
 176 void blake64_init(blake64_ctx_t* ctx){
 177         uint8_t i;
 178         for(i=0; i<8; ++i){
 179                 ctx->h[i] = pgm_read_qword(&(blake64_iv[i]));
 180         }
 181         memset(ctx->s, 0, 4*8);
 182         ctx->counter = 0;
 183         ctx->appendone = 1;
 184 }
 185
 186 uint64_t blake48_iv[] PROGMEM = {
 187     0xCBBB9D5DC1059ED8LL, 0x629A292A367CD507LL,
 188     0x9159015A3070DD17LL, 0x152FECD8F70E5939LL,
 189     0x67332667FFC00B31LL, 0x8EB44A8768581511LL,
 190     0xDB0C2E0D64F98FA7LL, 0x47B5481DBEFA4FA4LL
 191 };
 192
 193 void blake48_init(blake48_ctx_t* ctx){
 194         uint8_t i;
 195         for(i=0; i<8; ++i){
 196                 ctx->h[i] = pgm_read_qword(&(blake48_iv[i]));
 197         }
 198         memset(ctx->s, 0, 4*8);
 199         ctx->counter = 0;
 200         ctx->appendone = 0;
 201 }
 202
 203 void blake64_ctx2hash(void* dest, const blake64_ctx_t* ctx){
 204         uint8_t i;
 205         for(i=0; i<8; ++i){
 206                 ((uint32_t*)dest)[2*i+0] = CHANGE_ENDIAN32((ctx->h[i])>>32);
 207                 ((uint32_t*)dest)[2*i+1] = CHANGE_ENDIAN32((uint32_t)ctx->h[i]);
 208         }
 209 }
 210
 211 void blake48_ctx2hash(void* dest, const blake48_ctx_t* ctx){
 212         uint8_t i;
 213         for(i=0; i<6; ++i){
 214                 ((uint32_t*)dest)[2*i+0] = CHANGE_ENDIAN32((ctx->h[i])>>32);
 215                 ((uint32_t*)dest)[2*i+1] = CHANGE_ENDIAN32((uint32_t)ctx->h[i]);
 216         }
 217 }
 218
 219 void blake64_nextBlock(blake64_ctx_t* ctx, const void* block){
 220         blake_large_nextBlock(ctx, block);
 221 }
 222
 223 void blake48_nextBlock(blake48_ctx_t* ctx, const void* block){
 224         blake_large_nextBlock(ctx, block);
 225 }
 226
 227 void blake64_lastBlock(blake64_ctx_t* ctx, const void* block, uint16_t length_b){
 228         blake_large_lastBlock(ctx, block, length_b);
 229 }
 230
 231 void blake48_lastBlock(blake48_ctx_t* ctx, const void* block, uint16_t length_b){
 232         blake_large_lastBlock(ctx, block, length_b);
 233 }
 234
 235 void blake64(void* dest, const void* msg, uint32_t length_b){
 236         blake_large_ctx_t ctx;
 237         blake64_init(&ctx);
 238         while(length_b>=BLAKE_LARGE_BLOCKSIZE){
 239                 blake_large_nextBlock(&ctx, msg);
 240                 msg = (uint8_t*)msg + BLAKE_LARGE_BLOCKSIZE_B;
 241                 length_b -= BLAKE_LARGE_BLOCKSIZE;
 242         }
 243         blake_large_lastBlock(&ctx, msg, length_b);
 244         blake64_ctx2hash(dest, &ctx);
 245 }
 246
 247 void blake48(void* dest, const void* msg, uint32_t length_b){
 248         blake_large_ctx_t ctx;
 249         blake48_init(&ctx);
 250         while(length_b>=BLAKE_LARGE_BLOCKSIZE){
 251                 blake_large_nextBlock(&ctx, msg);
 252                 msg = (uint8_t*)msg + BLAKE_LARGE_BLOCKSIZE_B;
 253                 length_b -= BLAKE_LARGE_BLOCKSIZE;
 254         }
 255         blake_large_lastBlock(&ctx, msg, length_b);
 256         blake48_ctx2hash(dest, &ctx);
 257 }