X-Git-Url: https://git.cryptolib.org/?p=avr-crypto-lib.git;a=blobdiff_plain;f=bmw%2Fbmw_large.c;h=1bae63f2794f917d544a6287e18bc11ecff31c11;hp=5b0039ab31663e81a3864a660a1ea311f2da1e26;hb=4b5da1dc27a791b5c448274a3db09cd035b33493;hpb=2873506dc8ebc5657d24af0ad5eab1fc06ce09e9 diff --git a/bmw/bmw_large.c b/bmw/bmw_large.c index 5b0039a..1bae63f 100644 --- a/bmw/bmw_large.c +++ b/bmw/bmw_large.c @@ -1,7 +1,7 @@ /* bmw_large.c */ /* This file is part of the AVR-Crypto-Lib. - Copyright (C) 2009 Daniel Otte (daniel.otte@rub.de) + Copyright (C) 2006-2015 Daniel Otte (bg@nerilex.org) This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -19,7 +19,7 @@ /* * \file bmw_large.c * \author Daniel Otte - * \email daniel.otte@rub.de + * \email bg@nerilex.org * \date 2009-04-27 * \license GPLv3 or later * @@ -30,10 +30,10 @@ #include #include "bmw_large.h" -#define SHL64(a,n) ((a)<<(n)) -#define SHR64(a,n) ((a)>>(n)) -#define ROTL64(a,n) (((a)<<(n))|((a)>>(64-(n)))) -#define ROTR64(a,n) (((a)>>(n))|((a)<<(64-(n)))) +#define SHL64(a,n) shiftl64(a,n) +#define SHR64(a,n) shiftr64(a,n) +#define ROTL64(a,n) rotl64(a,n) +#define ROTR64(a,n) rotr64(a,n) #define TWEAK 1 #define BUG24 0 @@ -43,7 +43,7 @@ #if DEBUG #include "cli.h" - void ctx_dump(const bmw_large_ctx_t* ctx){ + void ctx_dump(const bmw_large_ctx_t *ctx){ uint8_t i; cli_putstr_P(PSTR("\r\n==== ctx dump ====")); for(i=0; i<16;++i){ @@ -56,7 +56,7 @@ cli_hexdump(&(ctx->counter), 4); } - void dump_x(const uint64_t* q, uint8_t elements, char x){ + void dump_x(const uint64_t *q, uint8_t elements, char x){ uint8_t i; cli_putstr_P(PSTR("\r\n==== ")); cli_putc(x); @@ -75,6 +75,26 @@ #define dump_x(a,b,c) #endif +static +uint64_t rotl64(uint64_t a, uint8_t r){ + return (a<>(64-r)); +} + +static +uint64_t rotr64(uint64_t a, uint8_t r){ + return (a>>r)|(a<<(64-r)); +} + +static +uint64_t shiftl64(uint64_t a, uint8_t r){ + return (a<>r); +} + static uint64_t bmw_large_s0(uint64_t x){ uint64_t r; @@ -183,7 +203,7 @@ uint64_t bmw_large_r7(uint64_t x){ #define K 0x0555555555555555LL #define MASK 0xFFFFFFFFFFFFFFFFLL static -uint64_t k_lut[] PROGMEM = { +const uint64_t k_lut[] PROGMEM = { 16LL*K, 17LL*K, 18LL*K, 19LL*K, 20LL*K, 21LL*K, 22LL*K, 23LL*K, 24LL*K, 25LL*K, 26LL*K, 27LL*K, @@ -191,7 +211,7 @@ uint64_t k_lut[] PROGMEM = { */ /* the same as above but precomputed to avoid compiler warnings */ static -uint64_t k_lut[] PROGMEM = { +const uint64_t k_lut[] PROGMEM = { 0x5555555555555550LL, 0x5aaaaaaaaaaaaaa5LL, 0x5ffffffffffffffaLL, 0x655555555555554fLL, 0x6aaaaaaaaaaaaaa4LL, 0x6ffffffffffffff9LL, 0x755555555555554eLL, 0x7aaaaaaaaaaaaaa3LL, 0x7ffffffffffffff8LL, @@ -200,7 +220,7 @@ uint64_t k_lut[] PROGMEM = { 0xa55555555555554bLL }; static -uint64_t bmw_large_expand1(uint8_t j, const uint64_t* q, const void* m, const void* h){ +uint64_t bmw_large_expand1(uint8_t j, const uint64_t *q, const void *m, const void *h){ uint64_t(*s[])(uint64_t) = {bmw_large_s1, bmw_large_s2, bmw_large_s3, bmw_large_s0}; uint64_t a = 0; union{ @@ -230,7 +250,7 @@ uint64_t bmw_large_expand1(uint8_t j, const uint64_t* q, const void* m, const vo } static -uint64_t bmw_large_expand2(uint8_t j, const uint64_t* q, const void* m, const void* h){ +uint64_t bmw_large_expand2(uint8_t j, const uint64_t *q, const void *m, const void *h){ uint64_t(*rf[])(uint64_t) = {bmw_large_r1, bmw_large_r2, bmw_large_r3, bmw_large_r4, bmw_large_r5, bmw_large_r6, bmw_large_r7}; @@ -288,12 +308,12 @@ uint64_t bmw_large_expand2(uint8_t j, const uint64_t* q, const void* m, const vo #if F0_HACK==2 /* to understand this implementation take a look at f0-opt-table.txt */ -static uint16_t hack_table[5] PROGMEM = { 0x0311, 0xDDB3, 0x2A79, 0x07AA, 0x51C2 }; -static uint8_t offset_table[5] PROGMEM = { 4+16, 6+16, 9+16, 12+16, 13+16 }; +static const uint16_t hack_table[5] PROGMEM = { 0x0311, 0xDDB3, 0x2A79, 0x07AA, 0x51C2 }; +static const uint8_t offset_table[5] PROGMEM = { 4+16, 6+16, 9+16, 12+16, 13+16 }; static -void bmw_large_f0(uint64_t* q, const uint64_t* h, const void* m){ +void bmw_large_f0(uint64_t *q, const uint64_t *h, const void *m){ uint16_t hack_reg; uint8_t i,j,c; uint64_t(*s[])(uint64_t)={ bmw_large_s0, bmw_large_s1, bmw_large_s2, @@ -335,7 +355,7 @@ void bmw_large_f0(uint64_t* q, const uint64_t* h, const void* m){ #if F0_HACK==1 static -uint8_t f0_lut[] PROGMEM ={ +const uint8_t f0_lut[] PROGMEM ={ 5<<1, ( 7<<1)+1, (10<<1)+0, (13<<1)+0, (14<<1)+0, 6<<1, ( 8<<1)+1, (11<<1)+0, (14<<1)+0, (15<<1)+1, 0<<1, ( 7<<1)+0, ( 9<<1)+0, (12<<1)+1, (15<<1)+0, @@ -355,7 +375,7 @@ uint8_t f0_lut[] PROGMEM ={ }; static -void bmw_large_f0(uint64_t* q, const uint64_t* h, const void* m){ +void bmw_large_f0(uint64_t *q, const uint64_t *h, const void *m){ uint8_t i,j=-1,v,sign,l=0; uint64_t(*s[])(uint64_t)={ bmw_large_s0, bmw_large_s1, bmw_large_s2, bmw_large_s3, bmw_large_s4 }; @@ -397,7 +417,7 @@ void bmw_large_f0(uint64_t* q, const uint64_t* h, const void* m){ #if F0_HACK==0 static -void bmw_large_f0(uint64_t* q, const uint64_t* h, const void* m){ +void bmw_large_f0(uint64_t *q, const uint64_t *h, const void *m){ uint8_t i; uint64_t(*s[])(uint64_t)={ bmw_large_s0, bmw_large_s1, bmw_large_s2, bmw_large_s3, bmw_large_s4 }; @@ -438,7 +458,7 @@ void bmw_large_f0(uint64_t* q, const uint64_t* h, const void* m){ #endif /* F0_HACK==0 */ static -void bmw_large_f1(uint64_t* q, const void* m, const uint64_t* h){ +void bmw_large_f1(uint64_t *q, const void *m, const uint64_t *h){ uint8_t i; q[16] = bmw_large_expand1(0, q, m, h); q[17] = bmw_large_expand1(1, q, m, h); @@ -448,7 +468,7 @@ void bmw_large_f1(uint64_t* q, const void* m, const uint64_t* h){ } static -void bmw_large_f2(uint64_t* h, const uint64_t* q, const void* m){ +void bmw_large_f2(uint64_t *h, const uint64_t *q, const void *m){ uint64_t xl=0, xh; uint8_t i; for(i=16;i<24;++i){ @@ -490,7 +510,7 @@ void bmw_large_f2(uint64_t* h, const uint64_t* q, const void* m){ h[15] += SHR64(xl, 2) ^ q[22] ^ q[15]; } -void bmw_large_nextBlock(bmw_large_ctx_t* ctx, const void* block){ +void bmw_large_nextBlock(bmw_large_ctx_t *ctx, const void *block){ uint64_t q[32]; dump_x(block, 16, 'M'); bmw_large_f0(q, ctx->h, block); @@ -502,38 +522,41 @@ void bmw_large_nextBlock(bmw_large_ctx_t* ctx, const void* block){ ctx_dump(ctx); } -void bmw_large_lastBlock(bmw_large_ctx_t* ctx, const void* block, uint16_t length_b){ - uint8_t buffer[128]; +void bmw_large_lastBlock(bmw_large_ctx_t *ctx, const void *block, uint16_t length_b){ + union { + uint8_t v8[128]; + uint64_t v64[ 16]; + } buffer; while(length_b >= BMW_LARGE_BLOCKSIZE){ bmw_large_nextBlock(ctx, block); length_b -= BMW_LARGE_BLOCKSIZE; block = (uint8_t*)block + BMW_LARGE_BLOCKSIZE_B; } - memset(buffer, 0, 128); - memcpy(buffer, block, (length_b+7)/8); - buffer[length_b>>3] |= 0x80 >> (length_b&0x07); + memset(buffer.v8, 0, 128); + memcpy(buffer.v8, block, (length_b+7)/8); + buffer.v8[length_b>>3] |= 0x80 >> (length_b&0x07); if(length_b+1>128*8-64){ - bmw_large_nextBlock(ctx, buffer); - memset(buffer, 0, 128-8); + bmw_large_nextBlock(ctx, buffer.v8); + memset(buffer.v8, 0, 128-8); ctx->counter -= 1; } - *((uint64_t*)&(buffer[128-8])) = (uint64_t)(ctx->counter*1024LL)+(uint64_t)length_b; - bmw_large_nextBlock(ctx, buffer); + buffer.v64[15] = (uint64_t)(ctx->counter*1024LL)+(uint64_t)length_b; + bmw_large_nextBlock(ctx, buffer.v8); #if TWEAK uint8_t i; uint64_t q[32]; - memset(buffer, 0xaa, 128); + memset(buffer.v8, 0xaa, 128); for(i=0; i<16; ++i){ - buffer[8*i] = i + 0xa0; + buffer.v8[8*i] = i + 0xa0; } - bmw_large_f0(q, (uint64_t*)buffer, ctx->h); - bmw_large_f1(q, ctx->h, (uint64_t*)buffer); - bmw_large_f2((uint64_t*)buffer, q, ctx->h); - memcpy(ctx->h, buffer, 128); + bmw_large_f0(q, buffer.v64, ctx->h); + bmw_large_f1(q, ctx->h, buffer.v64); + bmw_large_f2(buffer.v64, q, ctx->h); + memcpy(ctx->h, buffer.v8, 128); #endif } -void bmw384_init(bmw384_ctx_t* ctx){ +void bmw384_init(bmw384_ctx_t *ctx){ uint8_t i; ctx->h[0] = 0x0001020304050607LL; for(i=1; i<16; ++i){ @@ -546,7 +569,7 @@ void bmw384_init(bmw384_ctx_t* ctx){ ctx_dump(ctx); } -void bmw512_init(bmw512_ctx_t* ctx){ +void bmw512_init(bmw512_ctx_t *ctx){ uint8_t i; ctx->h[0] = 0x8081828384858687LL; for(i=1; i<16; ++i){ @@ -556,31 +579,31 @@ void bmw512_init(bmw512_ctx_t* ctx){ ctx_dump(ctx); } -void bmw384_nextBlock(bmw384_ctx_t* ctx, const void* block){ +void bmw384_nextBlock(bmw384_ctx_t *ctx, const void *block){ bmw_large_nextBlock(ctx, block); } -void bmw512_nextBlock(bmw512_ctx_t* ctx, const void* block){ +void bmw512_nextBlock(bmw512_ctx_t *ctx, const void *block){ bmw_large_nextBlock(ctx, block); } -void bmw384_lastBlock(bmw384_ctx_t* ctx, const void* block, uint16_t length_b){ +void bmw384_lastBlock(bmw384_ctx_t *ctx, const void *block, uint16_t length_b){ bmw_large_lastBlock(ctx, block, length_b); } -void bmw512_lastBlock(bmw512_ctx_t* ctx, const void* block, uint16_t length_b){ +void bmw512_lastBlock(bmw512_ctx_t *ctx, const void *block, uint16_t length_b){ bmw_large_lastBlock(ctx, block, length_b); } -void bmw384_ctx2hash(void* dest, const bmw384_ctx_t* ctx){ +void bmw384_ctx2hash(void *dest, const bmw384_ctx_t *ctx){ memcpy(dest, &(ctx->h[10]), 384/8); } -void bmw512_ctx2hash(void* dest, const bmw512_ctx_t* ctx){ +void bmw512_ctx2hash(void *dest, const bmw512_ctx_t *ctx){ memcpy(dest, &(ctx->h[8]), 512/8); } -void bmw384(void* dest, const void* msg, uint32_t length_b){ +void bmw384(void *dest, const void *msg, uint32_t length_b){ bmw_large_ctx_t ctx; bmw384_init(&ctx); while(length_b>=BMW_LARGE_BLOCKSIZE){ @@ -592,7 +615,7 @@ void bmw384(void* dest, const void* msg, uint32_t length_b){ bmw384_ctx2hash(dest, &ctx); } -void bmw512(void* dest, const void* msg, uint32_t length_b){ +void bmw512(void *dest, const void *msg, uint32_t length_b){ bmw_large_ctx_t ctx; bmw512_init(&ctx); while(length_b>=BMW_LARGE_BLOCKSIZE){