X-Git-Url: https://git.cryptolib.org/?a=blobdiff_plain;ds=sidebyside;f=bmw%2Fbmw_small_speed.c;h=fe3de95ae39a4ad773901b94e7c64c3ac1a9ceb2;hb=27d15fffb7de7d44cb1d3becbd0e497a228f005e;hp=c67d6aa03985740b225e9d4bc1deb8ca7cd5f389;hpb=702e23d6807714e4da96104bc1223b1149a4da27;p=arm-crypto-lib.git diff --git a/bmw/bmw_small_speed.c b/bmw/bmw_small_speed.c index c67d6aa..fe3de95 100644 --- a/bmw/bmw_small_speed.c +++ b/bmw/bmw_small_speed.c @@ -28,7 +28,6 @@ #include #include #include "bmw_small.h" -#include "memxor.h" #define SHL32(a,n) ((a)<<(n)) #define SHR32(a,n) ((a)>>(n)) @@ -106,135 +105,83 @@ #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.i" -static inline +/* static inline */ void bmw_small_f0(uint32_t* q, uint32_t* h, const uint32_t* m){ - h[ 0] ^= m[ 0]; - h[ 1] ^= m[ 1]; - h[ 2] ^= m[ 2]; - h[ 3] ^= m[ 3]; - h[ 4] ^= m[ 4]; - h[ 5] ^= m[ 5]; - h[ 6] ^= m[ 6]; - h[ 7] ^= m[ 7]; - h[ 8] ^= m[ 8]; - h[ 9] ^= m[ 9]; - h[10] ^= m[10]; - h[11] ^= m[11]; - h[12] ^= m[12]; - h[13] ^= m[13]; - h[14] ^= m[14]; - h[15] ^= m[15]; - - dump_x(h, 16, 'T'); - q[ 0] = (h[ 5] - h[ 7] + h[10] + h[13] + h[14]); - q[ 3] = (h[ 0] - h[ 1] + h[ 8] - h[10] + h[13]); - q[ 6] = (h[ 4] - h[ 0] - h[ 3] - h[11] + h[13]); - q[ 9] = (h[ 0] - h[ 3] + h[ 6] - h[ 7] + h[14]); - q[12] = (h[ 1] + h[ 3] - h[ 6] - h[ 9] + h[10]); - q[15] = (h[12] - h[ 4] - h[ 6] - h[ 9] + h[13]); - q[ 2] = (h[ 0] + h[ 7] + h[ 9] - h[12] + h[15]); - q[ 5] = (h[ 3] - h[ 2] + h[10] - h[12] + h[15]); - q[ 8] = (h[ 2] - h[ 5] - h[ 6] + h[13] - h[15]); - q[11] = (h[ 8] - h[ 0] - h[ 2] - h[ 5] + h[ 9]); - q[14] = (h[ 3] - h[ 5] + h[ 8] - h[11] - h[12]); - q[ 1] = (h[ 6] - h[ 8] + h[11] + h[14] - h[15]); - q[ 4] = (h[ 1] + h[ 2] + h[ 9] - h[11] - h[14]); - q[ 7] = (h[ 1] - h[ 4] - h[ 5] - h[12] - h[14]); - q[10] = (h[ 8] - h[ 1] - h[ 4] - h[ 7] + h[15]); - q[13] = (h[ 2] + h[ 4] + h[ 7] + h[10] + h[11]); + uint32_t t[16]; + uint32_t tr0, tr1, tr2, tmp; + t[ 0] = h[ 0] ^ m[ 0]; + t[ 1] = h[ 1] ^ m[ 1]; + t[ 2] = h[ 2] ^ m[ 2]; + t[ 3] = h[ 3] ^ m[ 3]; + t[ 4] = h[ 4] ^ m[ 4]; + t[ 6] = h[ 6] ^ m[ 6]; + t[ 8] = h[ 8] ^ m[ 8]; + t[ 9] = h[ 9] ^ m[ 9]; + t[11] = h[11] ^ m[11]; + t[12] = h[12] ^ m[12]; + t[15] = h[15] ^ m[15]; + t[ 5] = h[ 5] ^ m[ 5]; + t[10] = h[10] ^ m[10]; + t[13] = h[13] ^ m[13]; + t[ 7] = h[ 7] ^ m[ 7]; + t[14] = h[14] ^ m[14]; + + dump_x(t, 16, 'T'); + /* + q[ 0] = (t[ 5] - t[ 7] + t[10] + t[13] + t[14]); + q[ 3] = (t[ 0] - t[ 1] + t[ 8] - t[10] + t[13]); + q[ 6] = (t[ 4] - t[ 0] - t[ 3] - t[11] + t[13]); + q[ 9] = (t[ 0] - t[ 3] + t[ 6] - t[ 7] + t[14]); + q[12] = (t[ 1] + t[ 3] - t[ 6] - t[ 9] + t[10]); + q[15] = (t[12] - t[ 4] - t[ 6] - t[ 9] + t[13]); + q[ 2] = (t[ 0] + t[ 7] + t[ 9] - t[12] + t[15]); + q[ 5] = (t[ 3] - t[ 2] + t[10] - t[12] + t[15]); + q[ 8] = (t[ 2] - t[ 5] - t[ 6] + t[13] - t[15]); + q[11] = (t[ 8] - t[ 0] - t[ 2] - t[ 5] + t[ 9]); + q[14] = (t[ 3] - t[ 5] + t[ 8] - t[11] - t[12]); + q[ 1] = (t[ 6] - t[ 8] + t[11] + t[14] - t[15]); + q[ 4] = (t[ 1] + t[ 2] + t[ 9] - t[11] - t[14]); + q[ 7] = (t[ 1] - t[ 4] - t[ 5] - t[12] - t[14]); + q[10] = (t[ 8] - t[ 1] - t[ 4] - t[ 7] + t[15]); + q[13] = (t[ 2] + t[ 4] + t[ 7] + t[10] + t[11]); + */ + tmp = +t[ 5] +t[10] +t[13] +(tr1=-t[ 7]+t[14]) ; + q[ 0] = S32_0(tmp) + h[ 1]; + tmp = +t[ 8] +t[13] +t[ 0] -(tr2=+t[ 1]+t[10]) ; + q[ 3] = S32_3(tmp) + h[ 4]; + tmp = -t[11] +t[13] -t[ 0] -t[ 3] +t[ 4] ; + q[ 6] = S32_1(tmp) + h[ 7]; + tmp = +t[ 0] +(tr0=-t[ 3]+t[ 6]) +(tr1) ; + q[ 9] = S32_4(tmp) + h[10]; + tmp = -t[ 9] -(tr0) +(tr2) ; + q[12] = S32_2(tmp) + h[13]; + tmp = -t[ 4] +(tr0=-t[ 9]+t[12]) +(tr1=-t[ 6]+t[13]) ; + q[15] = S32_0(tmp) + h[ 0]; + tmp = +t[ 7] +t[15] +t[ 0] -(tr0) ; + q[ 2] = S32_2(tmp) + h[ 3]; + tmp = +t[10] +(tr0=-t[ 2]+t[15]) +(tr2=+t[ 3]-t[12]) ; + q[ 5] = S32_0(tmp) + h[ 6]; + tmp = -t[ 5] -(tr0) +(tr1) ; + q[ 8] = S32_3(tmp) + h[ 9]; + tmp = -t[ 0] -t[ 2] +t[ 9] +(tr0=-t[ 5]+t[ 8]) ; + q[11] = S32_1(tmp) + h[12]; + tmp = -t[11] +(tr0) +(tr2) ; + q[14] = S32_4(tmp) + h[15]; + tmp = +t[ 6] +(tr0=+t[11]+t[14]) -(tr1=+t[ 8]+t[15]) ; + q[ 1] = S32_1(tmp) + h[ 2]; + tmp = +t[ 9] +t[ 1] +t[ 2] -(tr0) ; + q[ 4] = S32_4(tmp) + h[ 5]; + tmp = -t[12] -t[14] +t[ 1] -t[ 4] -t[ 5] ; + q[ 7] = S32_2(tmp) + h[ 8]; + tmp = -t[ 1] -(tr0=+t[ 4]+t[ 7]) +(tr1) ; + q[10] = S32_0(tmp) + h[11]; + tmp = +t[ 2] +t[10] +t[11] +(tr0) ; + q[13] = S32_3(tmp) + h[14]; + 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]); - q[ 3] = S32_3(q[ 3]) + (h[ 4] ^= m[ 4]); - q[ 4] = S32_4(q[ 4]) + (h[ 5] ^= m[ 5]); - q[ 5] = S32_0(q[ 5]) + (h[ 6] ^= m[ 6]); - q[ 6] = S32_1(q[ 6]) + (h[ 7] ^= m[ 7]); - q[ 7] = S32_2(q[ 7]) + (h[ 8] ^= m[ 8]); - q[ 8] = S32_3(q[ 8]) + (h[ 9] ^= m[ 9]); - q[ 9] = S32_4(q[ 9]) + (h[10] ^= m[10]); - q[10] = S32_0(q[10]) + (h[11] ^= m[11]); - q[11] = S32_1(q[11]) + (h[12] ^= m[12]); - q[12] = S32_2(q[12]) + (h[13] ^= m[13]); - q[13] = S32_3(q[13]) + (h[14] ^= m[14]); - q[14] = S32_4(q[14]) + (h[15] ^= m[15]); - q[15] = S32_0(q[15]) + (h[ 0] ^= m[ 0]); } static inline @@ -281,37 +228,41 @@ void bmw_small_nextBlock(bmw_small_ctx_t* ctx, const void* block){ } void bmw_small_lastBlock(bmw_small_ctx_t* ctx, const void* block, uint16_t length_b){ - uint8_t buffer[64]; + union { + uint8_t v8[64]; + uint32_t v32[16]; + uint64_t v64[ 8]; + } buffer; while(length_b >= BMW_SMALL_BLOCKSIZE){ bmw_small_nextBlock(ctx, block); length_b -= BMW_SMALL_BLOCKSIZE; block = (uint8_t*)block + BMW_SMALL_BLOCKSIZE_B; } - memset(buffer, 0, 64); - memcpy(buffer, block, (length_b+7)/8); - buffer[length_b>>3] |= 0x80 >> (length_b&0x07); + memset(buffer.v8, 0, 64); + memcpy(buffer.v8, block, (length_b+7)/8); + buffer.v8[length_b>>3] |= 0x80 >> (length_b&0x07); if(length_b+1>64*8-64){ - bmw_small_nextBlock(ctx, buffer); - memset(buffer, 0, 64-8); + bmw_small_nextBlock(ctx, buffer.v8); + memset(buffer.v8, 0, 64-8); ctx->counter -= 1; } - *((uint64_t*)&(buffer[64-8])) = (uint64_t)(ctx->counter*512LL)+(uint64_t)length_b; - bmw_small_nextBlock(ctx, buffer); + buffer.v64[7] = (uint64_t)(ctx->counter*512LL)+(uint64_t)length_b; + bmw_small_nextBlock(ctx, buffer.v8); uint8_t i; uint32_t q[32]; - memset(buffer, 0xaa, 64); + memset(buffer.v8, 0xaa, 64); for(i=0; i<16;++i){ - buffer[i*4] = i+0xa0; + buffer.v8[i*4] = i+0xa0; } -// dump_x(buffer, 16, 'A'); +// dump_x(buffer.v8, 16, 'A'); dump_x(ctx->h, 16, 'M'); - bmw_small_f0(q, (uint32_t*)buffer, ctx->h); - dump_x(buffer, 16, 'a'); + bmw_small_f0(q, buffer.v32, ctx->h); + dump_x(buffer.v8, 16, 'a'); dump_x(q, 16, 'Q'); - bmw_small_f1(q, ctx->h, (uint32_t*)buffer); + bmw_small_f1(q, ctx->h, buffer.v32); dump_x(q, 32, 'Q'); - bmw_small_f2((uint32_t*)buffer, q, ctx->h); - memcpy(ctx->h, buffer, 64); + bmw_small_f2(buffer.v32, q, ctx->h); + memcpy(ctx->h, buffer.v8, 64); } void bmw224_init(bmw224_ctx_t* ctx){ @@ -382,5 +333,3 @@ void bmw256(void* dest, const void* msg, uint32_t length_b){ bmw256_ctx2hash(dest, &ctx); } -#include "f1_autogen.c" -