From: bg Date: Thu, 13 May 2010 17:10:05 +0000 (+0200) Subject: updated bmw standard implementation X-Git-Url: https://git.cryptolib.org/?a=commitdiff_plain;h=21fd4f1cb10ca32766e816210224008fe3c31705;p=arm-crypto-lib.git updated bmw standard implementation --- diff --git a/bmw/bmw_small.c b/bmw/bmw_small.c index 036ac25..74c6b9c 100644 --- a/bmw/bmw_small.c +++ b/bmw/bmw_small.c @@ -43,7 +43,7 @@ # define BUG24 1 #endif -#define F0_HACK 2 +#define F0_HACK 0 #define DEBUG 0 @@ -86,110 +86,37 @@ #define dump_x(a,b,c) #endif -static -uint32_t bmw_small_s0(uint32_t x){ - uint32_t r; - r = SHR32(x, 1) - ^ SHL32(x, 3) - ^ ROTL32(x, 4) - ^ ROTR32(x, 13); - return r; -} - -static -uint32_t bmw_small_s1(uint32_t x){ - uint32_t r; - r = SHR32(x, 1) - ^ SHL32(x, 2) - ^ ROTL32(x, 8) - ^ ROTR32(x, 9); - return r; -} - -static -uint32_t bmw_small_s2(uint32_t x){ - uint32_t r; - r = SHR32(x, 2) - ^ SHL32(x, 1) - ^ ROTL32(x, 12) - ^ ROTR32(x, 7); - return r; -} - -static -uint32_t bmw_small_s3(uint32_t x){ - uint32_t r; - r = SHR32(x, 2) - ^ SHL32(x, 2) - ^ ROTL32(x, 15) - ^ ROTR32(x, 3); - return r; -} - -static -uint32_t bmw_small_s4(uint32_t x){ - uint32_t r; - r = SHR32(x, 1) - ^ x; - return r; -} - -static -uint32_t bmw_small_s5(uint32_t x){ - uint32_t r; - r = SHR32(x, 2) - ^ x; - return r; -} - -static -uint32_t bmw_small_r1(uint32_t x){ - uint32_t r; - r = ROTL32(x, 3); - return r; -} - -static -uint32_t bmw_small_r2(uint32_t x){ - uint32_t r; - r = ROTL32(x, 7); - return r; -} - -static -uint32_t bmw_small_r3(uint32_t x){ - uint32_t r; - r = ROTL32(x, 13); - return r; -} - -static -uint32_t bmw_small_r4(uint32_t x){ - uint32_t r; - r = ROTL32(x, 16); - return r; -} - -static -uint32_t bmw_small_r5(uint32_t x){ - uint32_t r; - r = ROTR32(x, 13); - return r; -} - -static -uint32_t bmw_small_r6(uint32_t x){ - uint32_t r; - r = ROTR32(x, 9); - return r; -} - -static -uint32_t bmw_small_r7(uint32_t x){ - uint32_t r; - r = ROTR32(x, 5); - return r; -} +#define S32_0(x) ( (SHR32((x), 1)) ^ \ + (SHL32((x), 3)) ^ \ + (ROTL32((x), 4)) ^ \ + (ROTR32((x), 13)) ) + +#define S32_1(x) ( (SHR32((x), 1)) ^ \ + (SHL32((x), 2)) ^ \ + (ROTL32((x), 8)) ^ \ + (ROTR32((x), 9)) ) + +#define S32_2(x) ( (SHR32((x), 2)) ^ \ + (SHL32((x), 1)) ^ \ + (ROTL32((x), 12)) ^ \ + (ROTR32((x), 7)) ) + +#define S32_3(x) ( (SHR32((x), 2)) ^ \ + (SHL32((x), 2)) ^ \ + (ROTL32((x), 15)) ^ \ + (ROTR32((x), 3)) ) + +#define S32_4(x) ( (SHR32((x), 1)) ^ (x)) + +#define S32_5(x) ( (SHR32((x), 2)) ^ (x)) + +#define R32_1(x) (ROTL32((x), 3)) +#define R32_2(x) (ROTL32((x), 7)) +#define R32_3(x) (ROTL32((x), 13)) +#define R32_4(x) (ROTL32((x), 16)) +#define R32_5(x) (ROTR32((x), 13)) +#define R32_6(x) (ROTR32((x), 9)) +#define R32_7(x) (ROTR32((x), 5)) /* #define K 0x05555555L static @@ -211,61 +138,33 @@ uint32_t k_lut[] = { static uint32_t bmw_small_expand1(uint8_t j, const uint32_t* q, const void* m, const void* h){ - uint32_t(*s[])(uint32_t) = {bmw_small_s1, bmw_small_s2, bmw_small_s3, bmw_small_s0}; uint32_t r; - uint8_t i; /* r = 0x05555555*(j+16); */ -#if TWEAK r = ( ROTL32(((uint32_t*)m)[j&0xf], ((j+0)&0xf)+1 ) + ROTL32(((uint32_t*)m)[(j+3)&0xf], ((j+3)&0xf)+1 ) - ROTL32(((uint32_t*)m)[(j+10)&0xf], ((j+10)&0xf)+1 ) + k_lut[j] ) ^ ((uint32_t*)h)[(j+7)&0xf]; -#else - r = k_lut[j]; - r += ((uint32_t*)m)[j&0xf]; - r += ((uint32_t*)m)[(j+3)&0xf]; - r -= ((uint32_t*)m)[(j+10)&0xf]; -#endif - for(i=0; i<16; ++i){ - r += s[i%4](q[j+i]); - } + r += S32_1(q[j+ 0]) + S32_2(q[j+ 1]) + S32_3(q[j+ 2]) + S32_0(q[j+ 3]) + + S32_1(q[j+ 4]) + S32_2(q[j+ 5]) + S32_3(q[j+ 6]) + S32_0(q[j+ 7]) + + S32_1(q[j+ 8]) + S32_2(q[j+ 9]) + S32_3(q[j+10]) + S32_0(q[j+11]) + + S32_1(q[j+12]) + S32_2(q[j+13]) + S32_3(q[j+14]) + S32_0(q[j+15]); return r; } static uint32_t bmw_small_expand2(uint8_t j, const uint32_t* q, const void* m, const void* h){ - uint32_t(*rf[])(uint32_t) = {bmw_small_r1, bmw_small_r2, bmw_small_r3, - bmw_small_r4, bmw_small_r5, bmw_small_r6, - bmw_small_r7}; - uint32_t r=0; - uint8_t i; - for(i=0; i<14; i+=2){ - r += q[j+i]; - } - for(i=0; i<14; i+=2){ - r += rf[i/2](q[j+i+1]); - } -#if TWEAK - r += bmw_small_s4(q[j+14]); - r += bmw_small_s5(q[j+15]); -#else - r += bmw_small_s5(q[j+14]); - r += bmw_small_s4(q[j+15]); -#endif -#if TWEAK - r += ( ROTL32(((uint32_t*)m)[j&0xf], ((j+0)&0xf)+1 ) + uint32_t r; + r = ( ROTL32(((uint32_t*)m)[j&0xf], ((j+0)&0xf)+1 ) + ROTL32(((uint32_t*)m)[(j+3)&0xf], ((j+3)&0xf)+1 ) - ROTL32(((uint32_t*)m)[(j+10)&0xf], ((j+10)&0xf)+1 ) + k_lut[j] ) ^ ((uint32_t*)h)[(j+7)&0xf]; -#else - r += k_lut[j]; - r += ((uint32_t*)m)[j&0xf]; - r += ((uint32_t*)m)[(j+3)&0xf]; - r -= ((uint32_t*)m)[(j+10)&0xf]; -#endif + r += (q[j+ 0]) + R32_1(q[j+ 1]) + (q[j+ 2]) + R32_2(q[j+ 3]) + + (q[j+ 4]) + R32_3(q[j+ 5]) + (q[j+ 6]) + R32_4(q[j+ 7]) + + (q[j+ 8]) + R32_5(q[j+ 9]) + (q[j+10]) + R32_6(q[j+11]) + + (q[j+12]) + R32_7(q[j+13]) + S32_4(q[j+14]) + S32_5(q[j+15]); return r; } @@ -304,14 +203,12 @@ void bmw_small_f0(uint32_t* q, uint32_t* h, const void* m){ for(i=0; i<16; ++i){ q[i] = s[i%5](q[i]); } -#if TWEAK for(i=0; i<16; ++i){ ((uint32_t*)h)[i] ^= ((uint32_t*)m)[i]; } for(i=0; i<16; ++i){ q[i] += h[(i+1)&0xf]; } -#endif } #endif /* F0_HACK==2*/ @@ -366,14 +263,12 @@ void bmw_small_f0(uint32_t* q, uint32_t* h, const void* m){ for(i=0; i<16; ++i){ q[i] = s[i%5](q[i]); } -#if TWEAK for(i=0; i<16; ++i){ ((uint32_t*)h)[i] ^= ((uint32_t*)m)[i]; } for(i=0; i<16; ++i){ q[i] += h[(i+1)&0xf]; } -#endif /* TWEAK */ } #endif /* F0_HACK==1 */ @@ -381,8 +276,6 @@ void bmw_small_f0(uint32_t* q, uint32_t* h, const void* m){ static void bmw_small_f0(uint32_t* q, uint32_t* h, const void* m){ uint8_t i; - uint32_t(*s[])(uint32_t)={ bmw_small_s0, bmw_small_s1, bmw_small_s2, - bmw_small_s3, bmw_small_s4 }; for(i=0; i<16; ++i){ ((uint32_t*)h)[i] ^= ((uint32_t*)m)[i]; } @@ -404,17 +297,28 @@ void bmw_small_f0(uint32_t* q, uint32_t* h, const void* m){ q[14] = (h[ 3] - h[ 5] + h[ 8] - h[11] - h[12]); q[15] = (h[12] - h[ 4] - h[ 6] - h[ 9] + h[13]); dump_x(q, 16, 'W'); - for(i=0; i<16; ++i){ - q[i] = s[i%5](q[i]); - } -#if TWEAK + q[ 0] = S32_0(q[ 0]); + q[ 1] = S32_1(q[ 1]); + q[ 2] = S32_2(q[ 2]); + q[ 3] = S32_3(q[ 3]); + q[ 4] = S32_4(q[ 4]); + q[ 5] = S32_0(q[ 5]); + q[ 6] = S32_1(q[ 6]); + q[ 7] = S32_2(q[ 7]); + q[ 8] = S32_3(q[ 8]); + q[ 9] = S32_4(q[ 9]); + q[10] = S32_0(q[10]); + q[11] = S32_1(q[11]); + q[12] = S32_2(q[12]); + q[13] = S32_3(q[13]); + q[14] = S32_4(q[14]); + q[15] = S32_0(q[15]); for(i=0; i<16; ++i){ ((uint32_t*)h)[i] ^= ((uint32_t*)m)[i]; } for(i=0; i<16; ++i){ q[i] += h[(i+1)&0xf]; } -#endif /* TWEAK */ } #endif /* F0_HACK==0 */ diff --git a/bmw/bmw_small_speed.c b/bmw/bmw_small_speed.c index c67d6aa..d683e56 100644 --- a/bmw/bmw_small_speed.c +++ b/bmw/bmw_small_speed.c @@ -106,59 +106,8 @@ #define R32_6(x) (ROTR32((x), 9)) #define R32_7(x) (ROTR32((x), 5)) -/* -#define K 0x05555555L -static -uint32_t k_lut[] PROGMEM = { - 16L*K, 17L*K, 18L*K, 19L*K, 20L*K, 21L*K, 22L*K, 23L*K, - 24L*K, 25L*K, 26L*K, 27L*K, 28L*K, 29L*K, 30L*K, 31L*K -}; -*/ -/* same as above but precomputed to avoid compiler warnings */ -/* -static -uint32_t k_lut[] = { - 0x55555550L, 0x5aaaaaa5L, 0x5ffffffaL, - 0x6555554fL, 0x6aaaaaa4L, 0x6ffffff9L, - 0x7555554eL, 0x7aaaaaa3L, 0x7ffffff8L, - 0x8555554dL, 0x8aaaaaa2L, 0x8ffffff7L, - 0x9555554cL, 0x9aaaaaa1L, 0x9ffffff6L, - 0xa555554bL }; -*/ -/* -static inline -uint32_t bmw_small_expand1(uint8_t j, const uint32_t* q, const void* m, const void* h){ - uint32_t r; - r = ( ROTL32(((uint32_t*)m)[j], ((j+0))+1 ) - + ROTL32(((uint32_t*)m)[(j+3)], ((j+3))+1 ) - - ROTL32(((uint32_t*)m)[(j+10)], ((j+10))+1 ) - + k_lut[j] - ) ^ ((uint32_t*)h)[(j+7)]; - r += S32_1(q[j+ 0]) + S32_2(q[j+ 1]) + S32_3(q[j+ 2]) + S32_0(q[j+ 3]) + - S32_1(q[j+ 4]) + S32_2(q[j+ 5]) + S32_3(q[j+ 6]) + S32_0(q[j+ 7]) + - S32_1(q[j+ 8]) + S32_2(q[j+ 9]) + S32_3(q[j+10]) + S32_0(q[j+11]) + - S32_1(q[j+12]) + S32_2(q[j+13]) + S32_3(q[j+14]) + S32_0(q[j+15]); - - return r; -} -static inline -uint32_t bmw_small_expand2(uint8_t j, const uint32_t* q, const void* m, const void* h){ - uint32_t r; - r = ( ROTL32(((uint32_t*)m)[j&0xf], ((j+0)&0xf)+1 ) - + ROTL32(((uint32_t*)m)[(j+3)&0xf], ((j+3)&0xf)+1 ) - - ROTL32(((uint32_t*)m)[(j+10)&0xf], ((j+10)&0xf)+1 ) - + k_lut[j] - ) ^ ((uint32_t*)h)[(j+7)&0xf]; - r += (q[j+ 0]) + R32_1(q[j+ 1]) + (q[j+ 2]) + R32_2(q[j+ 3]) + - (q[j+ 4]) + R32_3(q[j+ 5]) + (q[j+ 6]) + R32_4(q[j+ 7]) + - (q[j+ 8]) + R32_5(q[j+ 9]) + (q[j+10]) + R32_6(q[j+11]) + - (q[j+12]) + R32_7(q[j+13]) + S32_4(q[j+14]) + S32_5(q[j+15]); - return r; -} -*/ -static inline -void bmw_small_f1(uint32_t* q, const void* m, const void* h); +#include "f1_autogen.c" static inline void bmw_small_f0(uint32_t* q, uint32_t* h, const uint32_t* m){ @@ -197,28 +146,6 @@ void bmw_small_f0(uint32_t* q, uint32_t* h, const uint32_t* m){ q[10] = (h[ 8] - h[ 1] - h[ 4] - h[ 7] + h[15]); q[13] = (h[ 2] + h[ 4] + h[ 7] + h[10] + h[11]); dump_x(q, 16, 'W'); -/* - q[ 0] = S32_0(q[ 0]); q[ 1] = S32_1(q[ 1]); q[ 2] = S32_2(q[ 2]); q[ 3] = S32_3(q[ 3]); q[ 4] = S32_4(q[ 4]); - q[ 5] = S32_0(q[ 5]); q[ 6] = S32_1(q[ 6]); q[ 7] = S32_2(q[ 7]); q[ 8] = S32_3(q[ 8]); q[ 9] = S32_4(q[ 9]); - q[10] = S32_0(q[10]); q[11] = S32_1(q[11]); q[12] = S32_2(q[12]); q[13] = S32_3(q[13]); q[14] = S32_4(q[14]); - q[15] = S32_0(q[15]); - q[ 0] += h[ 1] ^= m[ 1]; - q[ 1] += h[ 2] ^= m[ 2]; - q[ 2] += h[ 3] ^= m[ 3]; - q[ 3] += h[ 4] ^= m[ 4]; - q[ 4] += h[ 5] ^= m[ 5]; - q[ 5] += h[ 6] ^= m[ 6]; - q[ 6] += h[ 7] ^= m[ 7]; - q[ 7] += h[ 8] ^= m[ 8]; - q[ 8] += h[ 9] ^= m[ 9]; - q[ 9] += h[10] ^= m[10]; - q[10] += h[11] ^= m[11]; - q[11] += h[12] ^= m[12]; - q[12] += h[13] ^= m[13]; - q[13] += h[14] ^= m[14]; - q[14] += h[15] ^= m[15]; - q[15] += h[ 0] ^= m[ 0]; -*/ q[ 0] = S32_0(q[ 0]) + (h[ 1] ^= m[ 1]); q[ 1] = S32_1(q[ 1]) + (h[ 2] ^= m[ 2]); q[ 2] = S32_2(q[ 2]) + (h[ 3] ^= m[ 3]); @@ -382,5 +309,3 @@ void bmw256(void* dest, const void* msg, uint32_t length_b){ bmw256_ctx2hash(dest, &ctx); } -#include "f1_autogen.c" -