X-Git-Url: https://git.cryptolib.org/?a=blobdiff_plain;f=bmw%2Fbmw_large_speed.c;h=c20887cb2a1d0e0a7b5333e855e3ccbb6db76a34;hb=7390f9235d6bc08b7fe34a5f43a04bd3b58f6ea6;hp=fcd4a13cbf22ef208ce2abfaa37c3bf58048a84f;hpb=e0fbef3e36b457e05c8b7146a75c1d95e02b60ba;p=arm-crypto-lib.git diff --git a/bmw/bmw_large_speed.c b/bmw/bmw_large_speed.c index fcd4a13..c20887c 100644 --- a/bmw/bmw_large_speed.c +++ b/bmw/bmw_large_speed.c @@ -34,9 +34,6 @@ #define ROTL64(a,n) (((a)<<(n))|((a)>>(64-(n)))) #define ROTR64(a,n) (((a)>>(n))|((a)<<(64-(n)))) -#define TWEAK 1 -#define BUG24 0 -#define F0_HACK 0 #define DEBUG 0 @@ -107,258 +104,113 @@ #define R64_6(x) (ROTR64((x), 21)) #define R64_7(x) (ROTR64((x), 11)) -/* -#define K 0x0555555555555555LL -#define MASK 0xFFFFFFFFFFFFFFFFLL -static -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, - 28LL*K, 29LL*K, 30LL*K, 31LL*K }; -*/ -/* the same as above but precomputed to avoid compiler warnings */ -static const -uint64_t k_lut[] = { - 0x5555555555555550LL, 0x5aaaaaaaaaaaaaa5LL, 0x5ffffffffffffffaLL, - 0x655555555555554fLL, 0x6aaaaaaaaaaaaaa4LL, 0x6ffffffffffffff9LL, - 0x755555555555554eLL, 0x7aaaaaaaaaaaaaa3LL, 0x7ffffffffffffff8LL, - 0x855555555555554dLL, 0x8aaaaaaaaaaaaaa2LL, 0x8ffffffffffffff7LL, - 0x955555555555554cLL, 0x9aaaaaaaaaaaaaa1LL, 0x9ffffffffffffff6LL, - 0xa55555555555554bLL }; - -static -uint64_t bmw_large_expand1(uint8_t j, const uint64_t* q, const void* m, const void* h){ - uint64_t r; - /* r = 0x0555555555555555LL*(j+16); */ - r = ( ROTL64(((uint64_t*)m)[(j)&0xf], ((j+ 0)&0xf)+1) - + ROTL64(((uint64_t*)m)[(j+3)&0xf], ((j+ 3)&0xf)+1) - + k_lut[j] - - ROTL64(((uint64_t*)m)[(j+10)&0xf],((j+10)&0xf)+1) - ) ^ ((uint64_t*)h)[(j+7)&0xf]; - r += S64_1(q[j+ 0]) + S64_2(q[j+ 1]) + S64_3(q[j+ 2]) + S64_0(q[j+ 3]) + - S64_1(q[j+ 4]) + S64_2(q[j+ 5]) + S64_3(q[j+ 6]) + S64_0(q[j+ 7]) + - S64_1(q[j+ 8]) + S64_2(q[j+ 9]) + S64_3(q[j+10]) + S64_0(q[j+11]) + - S64_1(q[j+12]) + S64_2(q[j+13]) + S64_3(q[j+14]) + S64_0(q[j+15]); - - return r; -} - -static -uint64_t bmw_large_expand2(uint8_t j, const uint64_t* q, const void* m, const void* h){ - uint64_t r=0; - r = ( ROTL64(((uint64_t*)m)[(j)&0xf], ((j+ 0)&0xf)+1) - + ROTL64(((uint64_t*)m)[(j+3)&0xf], ((j+ 3)&0xf)+1) - + k_lut[j] - - ROTL64(((uint64_t*)m)[(j+10)&0xf],((j+10)&0xf)+1) - ) ^ ((uint64_t*)h)[(j+7)&0xf]; - r += (q[j+ 0]) + R64_1(q[j+ 1]) + (q[j+ 2]) + R64_2(q[j+ 3]) + - (q[j+ 4]) + R64_3(q[j+ 5]) + (q[j+ 6]) + R64_4(q[j+ 7]) + - (q[j+ 8]) + R64_5(q[j+ 9]) + (q[j+10]) + R64_6(q[j+11]) + - (q[j+12]) + R64_7(q[j+13]) + S64_4(q[j+14]) + S64_5(q[j+15]); - - return r; -} - -#if F0_HACK==2 -/* to understand this implementation take a look at f0-opt-table.txt */ -static uint16_t hack_table[5] = { 0x0311, 0xDDB3, 0x2A79, 0x07AA, 0x51C2 }; -static uint8_t offset_table[5] = { 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){ - uint16_t hack_reg; - uint8_t i,j,c; - uint64_t(*s[])(uint64_t)={ bmw_large_s0, bmw_large_s1, bmw_large_s2, - bmw_large_s3, bmw_large_s4 }; - for(i=0; i<16; ++i){ - ((uint64_t*)h)[i] ^= ((uint64_t*)m)[i]; - } - dump_x(h, 16, 'T'); - memset(q, 0, 8*16); - c=4; - do{ - i=15; - j = offset_table[c]; - hack_reg = hack_table[c]; - do{ - if(hack_reg&1){ - q[i]-= h[j&15]; - }else{ - q[i]+= h[j&15]; - } - --j; - hack_reg>>= 1; - }while(i--!=0); - }while(c--!=0); - dump_x(q, 16, 'W'); - for(i=0; i<16; ++i){ - q[i] = s[i%5](q[i]); - } -#if TWEAK - for(i=0; i<16; ++i){ - ((uint64_t*)h)[i] ^= ((uint64_t*)m)[i]; - } - for(i=0; i<16; ++i){ - q[i] += h[(i+1)&0xf]; - } -#endif /* TWEAK */ -} -#endif /* F0_HACK==2 */ - -#if F0_HACK==1 -static -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, - 0<<1, ( 1<<1)+1, ( 8<<1)+0, (10<<1)+1, (13<<1)+0, - 1<<1, ( 2<<1)+0, ( 9<<1)+0, (11<<1)+1, (14<<1)+1, - 3<<1, ( 2<<1)+1, (10<<1)+0, (12<<1)+1, (15<<1)+0, - 4<<1, ( 0<<1)+1, ( 3<<1)+1, (11<<1)+1, (13<<1)+0, - 1<<1, ( 4<<1)+1, ( 5<<1)+1, (12<<1)+1, (14<<1)+1, - 2<<1, ( 5<<1)+1, ( 6<<1)+1, (13<<1)+0, (15<<1)+1, - 0<<1, ( 3<<1)+1, ( 6<<1)+0, ( 7<<1)+1, (14<<1)+0, - 8<<1, ( 1<<1)+1, ( 4<<1)+1, ( 7<<1)+1, (15<<1)+0, - 8<<1, ( 0<<1)+1, ( 2<<1)+1, ( 5<<1)+1, ( 9<<1)+0, - 1<<1, ( 3<<1)+0, ( 6<<1)+1, ( 9<<1)+1, (10<<1)+0, - 2<<1, ( 4<<1)+0, ( 7<<1)+0, (10<<1)+0, (11<<1)+0, - 3<<1, ( 5<<1)+1, ( 8<<1)+0, (11<<1)+1, (12<<1)+1, - 12<<1, ( 4<<1)+1, ( 6<<1)+1, ( 9<<1)+1, (13<<1)+0 -}; - -static -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 }; - for(i=0; i<16; ++i){ - ((uint64_t*)h)[i] ^= ((uint64_t*)m)[i]; - } - dump_x(h, 16, 'T'); -// memset(q, 0, 4*16); - for(i=0; i<5*16; ++i){ - v = pgm_read_byte(f0_lut+i); - sign = v&1; - v >>=1; - if(i==l){ - j++; - l+=5; - q[j] = h[v]; - continue; - } - if(sign){ - q[j] -= h[v]; - }else{ - q[j] += h[v]; - } - } +#include "f1_autogen_large.i" + +static inline +void bmw_large_f0(uint64_t* q, uint64_t* h, const uint64_t* m){ + uint64_t t[16]; + uint64_t tr0, tr1, tr2; + 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[ 5] = h[ 5] ^ m[ 5]; + t[ 6] = h[ 6] ^ m[ 6]; + t[ 7] = h[ 7] ^ m[ 7]; + t[ 8] = h[ 8] ^ m[ 8]; + t[ 9] = h[ 9] ^ m[ 9]; + t[10] = h[10] ^ m[10]; + t[11] = h[11] ^ m[11]; + t[12] = h[12] ^ m[12]; + t[13] = h[13] ^ m[13]; + t[14] = h[14] ^ m[14]; + t[15] = h[15] ^ m[15]; + // dump_x(t, 16, 'T'); + /* + q[ 0] = (t[ 5] - t[ 7] + t[10] + t[13] + t[14]); + q[ 1] = (t[ 6] - t[ 8] + t[11] + t[14] - t[15]); + q[ 2] = (t[ 0] + t[ 7] + t[ 9] - t[12] + t[15]); + q[ 3] = (t[ 0] - t[ 1] + t[ 8] - t[10] + t[13]); + q[ 4] = (t[ 1] + t[ 2] + t[ 9] - t[11] - t[14]); + q[ 5] = (t[ 3] - t[ 2] + t[10] - t[12] + t[15]); + q[ 6] = (t[ 4] - t[ 0] - t[ 3] - t[11] + t[13]); + q[ 7] = (t[ 1] - t[ 4] - t[ 5] - t[12] - t[14]); + q[ 8] = (t[ 2] - t[ 5] - t[ 6] + t[13] - t[15]); + q[ 9] = (t[ 0] - t[ 3] + t[ 6] - t[ 7] + t[14]); + q[10] = (t[ 8] - t[ 1] - t[ 4] - t[ 7] + t[15]); + q[11] = (t[ 8] - t[ 0] - t[ 2] - t[ 5] + t[ 9]); + q[12] = (t[ 1] + t[ 3] - t[ 6] - t[ 9] + t[10]); + q[13] = (t[ 2] + t[ 4] + t[ 7] + t[10] + t[11]); + q[14] = (t[ 3] - t[ 5] + t[ 8] - t[11] - t[12]); + q[15] = (t[12] - t[ 4] - t[ 6] - t[ 9] + t[13]); + */ + q[ 0] = +t[ 5] +t[10] +t[13] +(tr1=-t[ 7]+t[14]) ; + q[ 3] = +t[ 8] +t[13] +t[ 0] +(tr2=-t[ 1]-t[10]) ; + q[ 6] = -t[11] +t[13] -t[ 0] -t[ 3] +t[ 4] ; + q[ 9] = +t[ 0] +(tr0=-t[ 3]+t[ 6]) +(tr1) ; + q[12] = -t[ 9] -(tr0) -(tr2) ; + q[15] = -t[ 4] +(tr0=-t[ 9]+t[12]) +(tr1=-t[ 6]+t[13]) ; + q[ 2] = +t[ 7] +t[15] +t[ 0] -(tr0) ; + q[ 5] = +t[10] +(tr0=-t[ 2]+t[15]) +(tr2=+t[ 3]-t[12]) ; + q[ 8] = -t[ 5] -(tr0) +(tr1) ; + q[11] = -t[ 0] -t[ 2] +t[ 9] +(tr0=-t[ 5]+t[ 8]) ; + q[14] = -t[11] +(tr0) +(tr2) ; + q[ 1] = +t[ 6] +(tr0=+t[11]+t[14]) +(tr1=-t[ 8]-t[15]) ; + q[ 4] = +t[ 9] +t[ 1] +t[ 2] -(tr0) ; + q[ 7] = -t[12] -t[14] +t[ 1] -t[ 4] -t[ 5] ; + q[10] = -t[ 1] +(tr0=-t[ 4]-t[ 7]) -(tr1) ; + q[13] = +t[ 2] +t[10] +t[11] -(tr0) ; dump_x(q, 16, 'W'); - for(i=0; i<16; ++i){ - q[i] = s[i%5](q[i]); - } -#if TWEAK - for(i=0; i<16; ++i){ - ((uint64_t*)h)[i] ^= ((uint64_t*)m)[i]; - } - for(i=0; i<16; ++i){ - q[i] += h[(i+1)&0xf]; - } -#endif /* TWEAK */ -} -#endif /* F0_HACK==1 */ - -#if F0_HACK==0 -static -void bmw_large_f0(uint64_t* q, const uint64_t* h, const void* m){ - uint8_t i; - for(i=0; i<16; ++i){ - ((uint64_t*)h)[i] ^= ((uint64_t*)m)[i]; - } -// dump_x(t, 16, 'T'); - q[ 0] = (h[ 5] - h[ 7] + h[10] + h[13] + h[14]); - q[ 1] = (h[ 6] - h[ 8] + h[11] + h[14] - h[15]); - q[ 2] = (h[ 0] + h[ 7] + h[ 9] - h[12] + h[15]); - q[ 3] = (h[ 0] - h[ 1] + h[ 8] - h[10] + h[13]); - q[ 4] = (h[ 1] + h[ 2] + h[ 9] - h[11] - h[14]); - q[ 5] = (h[ 3] - h[ 2] + h[10] - h[12] + h[15]); - q[ 6] = (h[ 4] - h[ 0] - h[ 3] - h[11] + h[13]); - q[ 7] = (h[ 1] - h[ 4] - h[ 5] - h[12] - h[14]); - q[ 8] = (h[ 2] - h[ 5] - h[ 6] + h[13] - h[15]); - q[ 9] = (h[ 0] - h[ 3] + h[ 6] - h[ 7] + h[14]); - q[10] = (h[ 8] - h[ 1] - h[ 4] - h[ 7] + h[15]); - q[11] = (h[ 8] - h[ 0] - h[ 2] - h[ 5] + h[ 9]); - q[12] = (h[ 1] + h[ 3] - h[ 6] - h[ 9] + h[10]); - q[13] = (h[ 2] + h[ 4] + h[ 7] + h[10] + h[11]); - 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'); - q[ 0] = S64_0(q[ 0]); q[ 1] = S64_1(q[ 1]); q[ 2] = S64_2(q[ 2]); q[ 3] = S64_3(q[ 3]); q[ 4] = S64_4(q[ 4]); - q[ 5] = S64_0(q[ 5]); q[ 6] = S64_1(q[ 6]); q[ 7] = S64_2(q[ 7]); q[ 8] = S64_3(q[ 8]); q[ 9] = S64_4(q[ 9]); - q[10] = S64_0(q[10]); q[11] = S64_1(q[11]); q[12] = S64_2(q[12]); q[13] = S64_3(q[13]); q[14] = S64_4(q[14]); - q[15] = S64_0(q[15]); - for(i=0; i<16; ++i){ - ((uint64_t*)h)[i] ^= ((uint64_t*)m)[i]; - } - for(i=0; i<16; ++i){ - q[i] += h[(i+1)&0xf]; - } + q[ 0] = S64_0(q[ 0]) + h[ 1]; + q[ 1] = S64_1(q[ 1]) + h[ 2]; + q[ 2] = S64_2(q[ 2]) + h[ 3]; + q[ 3] = S64_3(q[ 3]) + h[ 4]; + q[ 4] = S64_4(q[ 4]) + h[ 5]; + q[ 5] = S64_0(q[ 5]) + h[ 6]; + q[ 6] = S64_1(q[ 6]) + h[ 7]; + q[ 7] = S64_2(q[ 7]) + h[ 8]; + q[ 8] = S64_3(q[ 8]) + h[ 9]; + q[ 9] = S64_4(q[ 9]) + h[10]; + q[10] = S64_0(q[10]) + h[11]; + q[11] = S64_1(q[11]) + h[12]; + q[12] = S64_2(q[12]) + h[13]; + q[13] = S64_3(q[13]) + h[14]; + q[14] = S64_4(q[14]) + h[15]; + q[15] = S64_0(q[15]) + h[ 0]; } -#endif /* F0_HACK==0 */ -static -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); - for(i=2; i<16; ++i){ - q[16+i] = bmw_large_expand2(i, q, m, h); - } -} +static inline +void bmw_large_f2(uint64_t* h, const uint64_t* q, const uint64_t* m){ -static -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){ - xl ^= q[i]; - } - xh = xl; - for(i=24;i<32;++i){ - xh ^= q[i]; - } + uint64_t xl, xh; + xl = q[16] ^ q[17] ^ q[18] ^ q[19] ^ q[20] ^ q[21] ^ q[22] ^ q[23]; + xh = xl ^ q[24] ^ q[25] ^ q[26] ^ q[27] ^ q[28] ^ q[29] ^ q[30] ^ q[31]; #if DEBUG cli_putstr("\r\n XL = "); - cli_hexdump_rev(&xl, 4); + cli_hexdump_rev(&xl, 8); cli_putstr("\r\n XH = "); - cli_hexdump_rev(&xh, 4); + cli_hexdump_rev(&xh, 8); #endif - memcpy(h, m, 16*8); - h[0] ^= SHL64(xh, 5) ^ SHR64(q[16], 5); - h[1] ^= SHR64(xh, 7) ^ SHL64(q[17], 8); - h[2] ^= SHR64(xh, 5) ^ SHL64(q[18], 5); - h[3] ^= SHR64(xh, 1) ^ SHL64(q[19], 5); - h[4] ^= SHR64(xh, 3) ^ q[20]; - h[5] ^= SHL64(xh, 6) ^ SHR64(q[21], 6); - h[6] ^= SHR64(xh, 4) ^ SHL64(q[22], 6); - h[7] ^= SHR64(xh,11) ^ SHL64(q[23], 2); - for(i=0; i<8; ++i){ - h[i] += xl ^ q[24+i] ^ q[i]; - } - for(i=0; i<8; ++i){ - h[8+i] ^= xh ^ q[24+i]; - h[8+i] += ROTL64(h[(4+i)%8],i+9); - } - h[ 8] += SHL64(xl, 8) ^ q[23] ^ q[ 8]; - h[ 9] += SHR64(xl, 6) ^ q[16] ^ q[ 9]; - h[10] += SHL64(xl, 6) ^ q[17] ^ q[10]; - h[11] += SHL64(xl, 4) ^ q[18] ^ q[11]; - h[12] += SHR64(xl, 3) ^ q[19] ^ q[12]; - h[13] += SHR64(xl, 4) ^ q[20] ^ q[13]; - h[14] += SHR64(xl, 7) ^ q[21] ^ q[14]; - h[15] += SHR64(xl, 2) ^ q[22] ^ q[15]; + + h[0] = (SHL64(xh, 5) ^ SHR64(q[16], 5) ^ m[ 0]) + (xl ^ q[24] ^ q[ 0]); + h[1] = (SHR64(xh, 7) ^ SHL64(q[17], 8) ^ m[ 1]) + (xl ^ q[25] ^ q[ 1]); + h[2] = (SHR64(xh, 5) ^ SHL64(q[18], 5) ^ m[ 2]) + (xl ^ q[26] ^ q[ 2]); + h[3] = (SHR64(xh, 1) ^ SHL64(q[19], 5) ^ m[ 3]) + (xl ^ q[27] ^ q[ 3]); + h[4] = (SHR64(xh, 3) ^ q[20] ^ m[ 4]) + (xl ^ q[28] ^ q[ 4]); + h[5] = (SHL64(xh, 6) ^ SHR64(q[21], 6) ^ m[ 5]) + (xl ^ q[29] ^ q[ 5]); + h[6] = (SHR64(xh, 4) ^ SHL64(q[22], 6) ^ m[ 6]) + (xl ^ q[30] ^ q[ 6]); + h[7] = (SHR64(xh,11) ^ SHL64(q[23], 2) ^ m[ 7]) + (xl ^ q[31] ^ q[ 7]); + + h[ 8] = ROTL64(h[4], 9) + (xh ^ q[24] ^ m[ 8]) + (SHL64(xl, 8) ^ q[23] ^ q[ 8]); + h[ 9] = ROTL64(h[5], 10) + (xh ^ q[25] ^ m[ 9]) + (SHR64(xl, 6) ^ q[16] ^ q[ 9]); + h[10] = ROTL64(h[6], 11) + (xh ^ q[26] ^ m[10]) + (SHL64(xl, 6) ^ q[17] ^ q[10]); + h[11] = ROTL64(h[7], 12) + (xh ^ q[27] ^ m[11]) + (SHL64(xl, 4) ^ q[18] ^ q[11]); + h[12] = ROTL64(h[0], 13) + (xh ^ q[28] ^ m[12]) + (SHR64(xl, 3) ^ q[19] ^ q[12]); + h[13] = ROTL64(h[1], 14) + (xh ^ q[29] ^ m[13]) + (SHR64(xl, 4) ^ q[20] ^ q[13]); + h[14] = ROTL64(h[2], 15) + (xh ^ q[30] ^ m[14]) + (SHR64(xl, 7) ^ q[21] ^ q[14]); + h[15] = ROTL64(h[3], 16) + (xh ^ q[31] ^ m[15]) + (SHR64(xl, 2) ^ q[22] ^ q[15]); } void bmw_large_nextBlock(bmw_large_ctx_t* ctx, const void* block){ @@ -374,34 +226,35 @@ void bmw_large_nextBlock(bmw_large_ctx_t* ctx, const void* block){ } void bmw_large_lastBlock(bmw_large_ctx_t* ctx, const void* block, uint16_t length_b){ - uint8_t buffer[128]; + 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); -#if TWEAK + buffer.v64[15] = (uint64_t)(ctx->counter*1024LL)+(uint64_t)length_b; + bmw_large_nextBlock(ctx, buffer.v8); 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); -#endif + 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); } void bmw384_init(bmw384_ctx_t* ctx){