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/memxor.h"
  32 #include "blake_large.h"
  33 #include "blake_common.h"
  34
  35 static
  36 uint64_t pgm_read_qword(void* p){
  37         union{
  38                 uint64_t v64;
  39                 uint32_t v32[2];
  40         }r;
  41         r.v32[0] = pgm_read_dword(p);
  42         r.v32[1] = pgm_read_dword((uint8_t*)p+4);
  43         return r.v64;
  44 }
  45
  46 static
  47 uint64_t blake_c[] PROGMEM = {
  48    0x243F6A8885A308D3LL, 0x13198A2E03707344LL,
  49    0xA4093822299F31D0LL, 0x082EFA98EC4E6C89LL,
  50    0x452821E638D01377LL, 0xBE5466CF34E90C6CLL,
  51    0xC0AC29B7C97C50DDLL, 0x3F84D5B5B5470917LL,
  52    0x9216D5D98979FB1BLL, 0xD1310BA698DFB5ACLL,
  53    0x2FFD72DBD01ADFB7LL, 0xB8E1AFED6A267E96LL,
  54    0xBA7C9045F12C7F99LL, 0x24A19947B3916CF7LL,
  55    0x0801F2E2858EFC16LL, 0x636920D871574E69LL
  56 };
  57
  58
  59
  60 #define ROTL64(a, n) (((a)<<(n))|((a)>>(64-(n))))
  61 #define ROTR64(a, n) (((a)>>(n))|((a)<<(64-(n))))
  62 #define CHANGE_ENDIAN32(a) (((a)<<24)| \
  63                             ((0x0000ff00&(a))<<8)| \
  64                                                     ((0x00ff0000&(a))>>8)| \
  65                                                     (a)>>24 )
  66
  67 static
  68 void blake_large_expand(uint64_t* v, const blake_large_ctx_t* ctx){
  69         uint8_t i;
  70         memcpy(v, ctx->h, 8*8);
  71         for(i=0; i<8; ++i){
  72                 v[8+i] = pgm_read_qword(&(blake_c[i]));
  73         }
  74         memxor((uint8_t*)v+8, ctx->s, 4*8);
  75
  76 }
  77
  78 static
  79 void blake_large_changeendian(void* dest, const void* src){
  80         uint8_t i;
  81         uint32_t tmp;
  82         for(i=0; i<32; i+=2){
  83                 tmp = CHANGE_ENDIAN32(((uint32_t*)src)[i]);
  84                 ((uint32_t*)dest)[i] = CHANGE_ENDIAN32(((uint32_t*)src)[i+1]);
  85                 ((uint32_t*)dest)[i+1] = tmp;
  86         }
  87 }
  88
  89 static
  90 void blake_large_compress(uint64_t* v,const void* m){
  91         uint8_t r,i;
  92         uint8_t a,b,c,d, s0, s1;
  93         for(r=0; r<14; ++r){
  94                 for(i=0; i<8; ++i){
  95                         a = pgm_read_byte(blake_index_lut+4*i+0);
  96                         b = pgm_read_byte(blake_index_lut+4*i+1);
  97                         c = pgm_read_byte(blake_index_lut+4*i+2);
  98                         d = pgm_read_byte(blake_index_lut+4*i+3);
  99                         s0 = pgm_read_byte(blake_sigma+16*r+2*i+0);
 100                         s1 = pgm_read_byte(blake_sigma+16*r+2*i+1);
 101                         v[a] += v[b] + (((uint64_t*)m)[s0] ^ pgm_read_qword(&(blake_c[s1])));
 102                         v[d]  = ROTR64(v[d]^v[a], 32);
 103                         v[c] += v[d];
 104                         v[b]  = ROTR64(v[b]^v[c], 25);
 105                         v[a] += v[b] + (((uint64_t*)m)[s1] ^ pgm_read_qword(&(blake_c[s0])));
 106                         v[d]  = ROTR64(v[d]^v[a], 16);
 107                         v[c] += v[d];
 108                         v[b]  = ROTR64(v[b]^v[c], 11);
 109                 }
 110         }
 111 }
 112
 113 static
 114 void blake_large_collapse(blake_large_ctx_t* ctx, uint64_t* v){
 115         uint8_t i;
 116         for(i=0; i<8; ++i){
 117                 ctx->h[i] ^= ctx->s[i%4] ^ v[i] ^ v[8+i];
 118         }
 119 }
 120
 121 void blake_large_nextBlock(blake_large_ctx_t* ctx, const void* msg){
 122         uint64_t v[16];
 123         uint64_t m[16];
 124         union {
 125                 uint64_t v64;
 126                 uint32_t v32[2];
 127         }ctr;
 128         blake_large_expand(v,ctx);
 129         ctx->counter++;
 130         ctr.v64 = ctx->counter*1024;
 131         v[12] ^= ctr.v64;
 132         v[13] ^= ctr.v64;
 133         blake_large_changeendian(m, msg);
 134         blake_large_compress(v, m);
 135         blake_large_collapse(ctx, v);
 136 }
 137
 138 void blake_large_lastBlock(blake_large_ctx_t* ctx, const void* msg, uint16_t length_b){
 139         while(length_b>=BLAKE_LARGE_BLOCKSIZE){
 140                 blake_large_nextBlock(ctx, msg);
 141                 msg = (uint8_t*)msg + BLAKE_LARGE_BLOCKSIZE_B;
 142                 length_b -= BLAKE_LARGE_BLOCKSIZE;
 143         }
 144         uint8_t buffer[128];
 145         uint64_t v[16];
 146         uint64_t ctr;
 147         ctr = ctx->counter*1024+length_b;
 148         memset(buffer, 0, 128);
 149         memcpy(buffer, msg, (length_b+7)/8);
 150         buffer[length_b/8] |= 0x80 >> (length_b&0x7);
 151         blake_large_changeendian(buffer, buffer);
 152         blake_large_expand(v, ctx);
 153         if(length_b>1024-128-2){
 154                 v[12] ^= ctr;
 155                 v[13] ^= ctr;
 156                 blake_large_compress(v, buffer);
 157                 blake_large_collapse(ctx, v);
 158                 memset(buffer, 0, 128-8);
 159                 blake_large_expand(v, ctx);
 160         } else {
 161                 if(length_b){
 162                         v[12] ^= ctr;
 163                         v[13] ^= ctr;
 164                 }
 165         }
 166         if(ctx->appendone)
 167                 buffer[128-16-8] |= 0x01;
 168         *((uint64_t*)(&(buffer[128-8]))) = ctr;
 169         blake_large_compress(v, buffer);
 170         blake_large_collapse(ctx, v);
 171
 172 }
 173
 174 uint64_t blake64_iv[] PROGMEM = {
 175     0x6A09E667F3BCC908LL, 0xBB67AE8584CAA73BLL,
 176     0x3C6EF372FE94F82BLL, 0xA54FF53A5F1D36F1LL,
 177     0x510E527FADE682D1LL, 0x9B05688C2B3E6C1FLL,
 178     0x1F83D9ABFB41BD6BLL, 0x5BE0CD19137E2179LL
 179 };
 180
 181 void blake64_init(blake64_ctx_t* ctx){
 182         uint8_t i;
 183         for(i=0; i<8; ++i){
 184                 ctx->h[i] = pgm_read_qword(&(blake64_iv[i]));
 185         }
 186         memset(ctx->s, 0, 4*8);
 187         ctx->counter = 0;
 188         ctx->appendone = 1;
 189 }
 190
 191 uint64_t blake48_iv[] PROGMEM = {
 192     0xCBBB9D5DC1059ED8LL, 0x629A292A367CD507LL,
 193     0x9159015A3070DD17LL, 0x152FECD8F70E5939LL,
 194     0x67332667FFC00B31LL, 0x8EB44A8768581511LL,
 195     0xDB0C2E0D64F98FA7LL, 0x47B5481DBEFA4FA4LL
 196 };
 197
 198 void blake48_init(blake48_ctx_t* ctx){
 199         uint8_t i;
 200         for(i=0; i<8; ++i){
 201                 ctx->h[i] = pgm_read_qword(&(blake48_iv[i]));
 202         }
 203         memset(ctx->s, 0, 4*8);
 204         ctx->counter = 0;
 205         ctx->appendone = 0;
 206 }
 207
 208 void blake64_ctx2hash(void* dest, const blake64_ctx_t* ctx){
 209         uint8_t i;
 210         for(i=0; i<8; ++i){
 211                 ((uint32_t*)dest)[2*i+0] = CHANGE_ENDIAN32((ctx->h[i])>>32);
 212                 ((uint32_t*)dest)[2*i+1] = CHANGE_ENDIAN32((uint32_t)ctx->h[i]);
 213         }
 214 }
 215
 216 void blake48_ctx2hash(void* dest, const blake48_ctx_t* ctx){
 217         uint8_t i;
 218         for(i=0; i<6; ++i){
 219                 ((uint32_t*)dest)[2*i+0] = CHANGE_ENDIAN32((ctx->h[i])>>32);
 220                 ((uint32_t*)dest)[2*i+1] = CHANGE_ENDIAN32((uint32_t)ctx->h[i]);
 221         }
 222 }
 223
 224 void blake64_nextBlock(blake64_ctx_t* ctx, const void* block){
 225         blake_large_nextBlock(ctx, block);
 226 }
 227
 228 void blake48_nextBlock(blake48_ctx_t* ctx, const void* block){
 229         blake_large_nextBlock(ctx, block);
 230 }
 231
 232 void blake64_lastBlock(blake64_ctx_t* ctx, const void* block, uint16_t length_b){
 233         blake_large_lastBlock(ctx, block, length_b);
 234 }
 235
 236 void blake48_lastBlock(blake48_ctx_t* ctx, const void* block, uint16_t length_b){
 237         blake_large_lastBlock(ctx, block, length_b);
 238 }
 239
 240 void blake64(void* dest, const void* msg, uint32_t length_b){
 241         blake_large_ctx_t ctx;
 242         blake64_init(&ctx);
 243         while(length_b>=BLAKE_LARGE_BLOCKSIZE){
 244                 blake_large_nextBlock(&ctx, msg);
 245                 msg = (uint8_t*)msg + BLAKE_LARGE_BLOCKSIZE_B;
 246                 length_b -= BLAKE_LARGE_BLOCKSIZE;
 247         }
 248         blake_large_lastBlock(&ctx, msg, length_b);
 249         blake64_ctx2hash(dest, &ctx);
 250 }
 251
 252 void blake48(void* dest, const void* msg, uint32_t length_b){
 253         blake_large_ctx_t ctx;
 254         blake48_init(&ctx);
 255         while(length_b>=BLAKE_LARGE_BLOCKSIZE){
 256                 blake_large_nextBlock(&ctx, msg);
 257                 msg = (uint8_t*)msg + BLAKE_LARGE_BLOCKSIZE_B;
 258                 length_b -= BLAKE_LARGE_BLOCKSIZE;
 259         }
 260         blake_large_lastBlock(&ctx, msg, length_b);
 261         blake48_ctx2hash(dest, &ctx);
 262 }