diff options
| author | Jack Lloyd <[email protected]> | 2017-10-11 17:02:20 -0400 |
|---|---|---|
| committer | Jack Lloyd <[email protected]> | 2017-10-12 11:13:11 -0400 |
| commit | 175f09ffd806f2f19cd509017a67ae1384f29ae1 (patch) | |
| tree | 6194884467e4720dd79797cd106a45d60211f35f | |
| parent | 40b3f979723b2b3dfb5c44047d7f786a73fd7f6f (diff) | |
Add compile-time rotation functions
The problem with asm rol/ror is the compiler can't schedule effectively.
But we only need asm in the case when the rotation is variable, so distinguish
the two cases. If a compile time constant, then static_assert that the rotation
is in the correct range and do the straightforward expression knowing the compiler
will probably do the right thing. Otherwise do a tricky expression that both
GCC and Clang happen to have recognize. Avoid the reduction case; instead
require that the rotation be in range (this reverts 2b37c13dcf).
Remove the asm rotations (making this branch illnamed), because now both Clang
and GCC will create a roll without any extra help.
Remove the reduction/mask by the word size for the variable case. The compiler
can't optimize that it out well, but it's easy to ensure it is valid in the callers,
especially now that the variable input cases are easy to grep for.
36 files changed, 716 insertions, 677 deletions
diff --git a/src/lib/block/aes/aes.cpp b/src/lib/block/aes/aes.cpp index 71a8c6a44..1893ab4a0 100644 --- a/src/lib/block/aes/aes.cpp +++ b/src/lib/block/aes/aes.cpp @@ -114,9 +114,9 @@ const std::vector<uint32_t>& AES_TE() const uint32_t x = make_uint32(xtime(s), s, s, xtime3(s)); TE[i] = x; - TE[i+256] = rotate_right(x, 8); - TE[i+512] = rotate_right(x, 16); - TE[i+768] = rotate_right(x, 24); + TE[i+256] = rotr< 8>(x); + TE[i+512] = rotr<16>(x); + TE[i+768] = rotr<24>(x); } return TE; }; @@ -135,9 +135,9 @@ const std::vector<uint32_t>& AES_TD() const uint32_t x = make_uint32(xtime14(s), xtime9(s), xtime13(s), xtime11(s)); TD[i] = x; - TD[i+256] = rotate_right(x, 8); - TD[i+512] = rotate_right(x, 16); - TD[i+768] = rotate_right(x, 24); + TD[i+256] = rotr< 8>(x); + TD[i+512] = rotr<16>(x); + TD[i+768] = rotr<24>(x); } return TD; }; @@ -188,24 +188,24 @@ void aes_encrypt_n(const uint8_t in[], uint8_t out[], */ uint32_t B0 = TE[get_byte(0, T0)] ^ - rotate_right(TE[get_byte(1, T1)], 8) ^ - rotate_right(TE[get_byte(2, T2)], 16) ^ - rotate_right(TE[get_byte(3, T3)], 24) ^ EK[4]; + rotr< 8>(TE[get_byte(1, T1)]) ^ + rotr<16>(TE[get_byte(2, T2)]) ^ + rotr<24>(TE[get_byte(3, T3)]) ^ EK[4]; uint32_t B1 = TE[get_byte(0, T1)] ^ - rotate_right(TE[get_byte(1, T2)], 8) ^ - rotate_right(TE[get_byte(2, T3)], 16) ^ - rotate_right(TE[get_byte(3, T0)], 24) ^ EK[5]; + rotr< 8>(TE[get_byte(1, T2)]) ^ + rotr<16>(TE[get_byte(2, T3)]) ^ + rotr<24>(TE[get_byte(3, T0)]) ^ EK[5]; uint32_t B2 = TE[get_byte(0, T2)] ^ - rotate_right(TE[get_byte(1, T3)], 8) ^ - rotate_right(TE[get_byte(2, T0)], 16) ^ - rotate_right(TE[get_byte(3, T1)], 24) ^ EK[6]; + rotr< 8>(TE[get_byte(1, T3)]) ^ + rotr<16>(TE[get_byte(2, T0)]) ^ + rotr<24>(TE[get_byte(3, T1)]) ^ EK[6]; uint32_t B3 = TE[get_byte(0, T3)] ^ - rotate_right(TE[get_byte(1, T0)], 8) ^ - rotate_right(TE[get_byte(2, T1)], 16) ^ - rotate_right(TE[get_byte(3, T2)], 24) ^ EK[7]; + rotr< 8>(TE[get_byte(1, T0)]) ^ + rotr<16>(TE[get_byte(2, T1)]) ^ + rotr<24>(TE[get_byte(3, T2)]) ^ EK[7]; for(size_t r = 2*4; r < EK.size(); r += 2*4) { @@ -276,24 +276,24 @@ void aes_decrypt_n(const uint8_t in[], uint8_t out[], size_t blocks, T0 ^= Z; uint32_t B0 = TD[get_byte(0, T0)] ^ - rotate_right(TD[get_byte(1, T3)], 8) ^ - rotate_right(TD[get_byte(2, T2)], 16) ^ - rotate_right(TD[get_byte(3, T1)], 24) ^ DK[4]; + rotr< 8>(TD[get_byte(1, T3)]) ^ + rotr<16>(TD[get_byte(2, T2)]) ^ + rotr<24>(TD[get_byte(3, T1)]) ^ DK[4]; uint32_t B1 = TD[get_byte(0, T1)] ^ - rotate_right(TD[get_byte(1, T0)], 8) ^ - rotate_right(TD[get_byte(2, T3)], 16) ^ - rotate_right(TD[get_byte(3, T2)], 24) ^ DK[5]; + rotr< 8>(TD[get_byte(1, T0)]) ^ + rotr<16>(TD[get_byte(2, T3)]) ^ + rotr<24>(TD[get_byte(3, T2)]) ^ DK[5]; uint32_t B2 = TD[get_byte(0, T2)] ^ - rotate_right(TD[get_byte(1, T1)], 8) ^ - rotate_right(TD[get_byte(2, T0)], 16) ^ - rotate_right(TD[get_byte(3, T3)], 24) ^ DK[6]; + rotr< 8>(TD[get_byte(1, T1)]) ^ + rotr<16>(TD[get_byte(2, T0)]) ^ + rotr<24>(TD[get_byte(3, T3)]) ^ DK[6]; uint32_t B3 = TD[get_byte(0, T3)] ^ - rotate_right(TD[get_byte(1, T2)], 8) ^ - rotate_right(TD[get_byte(2, T1)], 16) ^ - rotate_right(TD[get_byte(3, T0)], 24) ^ DK[7]; + rotr< 8>(TD[get_byte(1, T2)]) ^ + rotr<16>(TD[get_byte(2, T1)]) ^ + rotr<24>(TD[get_byte(3, T0)]) ^ DK[7]; for(size_t r = 2*4; r < DK.size(); r += 2*4) { diff --git a/src/lib/block/aria/aria.cpp b/src/lib/block/aria/aria.cpp index 5b449722a..1583dd7d3 100644 --- a/src/lib/block/aria/aria.cpp +++ b/src/lib/block/aria/aria.cpp @@ -183,7 +183,7 @@ inline void ARIA_FO(uint32_t& T0, uint32_t& T1, uint32_t& T2, uint32_t& T3) T1 ^= T2; T1 = ((T1 << 8) & 0xFF00FF00) | ((T1 >> 8) & 0x00FF00FF); - T2 = rotate_right(T2, 16); + T2 = rotr<16>(T2); T3 = reverse_bytes(T3); T1 ^= T2; @@ -205,7 +205,7 @@ inline void ARIA_FE(uint32_t& T0, uint32_t& T1, uint32_t& T2, uint32_t& T3) T1 ^= T2; T3 = ((T3 << 8) & 0xFF00FF00) | ((T3 >> 8) & 0x00FF00FF); - T0 = rotate_right(T0, 16); + T0 = rotr<16>(T0); T1 = reverse_bytes(T1); T1 ^= T2; @@ -411,9 +411,9 @@ void key_schedule(secure_vector<uint32_t>& ERK, { for(size_t j = 0; j != 4; ++j) { - DRK[i+j] = rotate_right(DRK[i+j], 8) ^ - rotate_right(DRK[i+j], 16) ^ - rotate_right(DRK[i+j], 24); + DRK[i+j] = rotr<8>(DRK[i+j]) ^ + rotr<16>(DRK[i+j]) ^ + rotr<24>(DRK[i+j]); } DRK[i+1] ^= DRK[i+2]; DRK[i+2] ^= DRK[i+3]; @@ -421,7 +421,7 @@ void key_schedule(secure_vector<uint32_t>& ERK, DRK[i+2] ^= DRK[i+0]; DRK[i+1] ^= DRK[i+2]; DRK[i+1] = ((DRK[i+1] << 8) & 0xFF00FF00) | ((DRK[i+1] >> 8) & 0x00FF00FF); - DRK[i+2] = rotate_right(DRK[i+2], 16); + DRK[i+2] = rotr<16>(DRK[i+2]); DRK[i+3] = reverse_bytes(DRK[i+3]); DRK[i+1] ^= DRK[i+2]; DRK[i+2] ^= DRK[i+3]; diff --git a/src/lib/block/camellia/camellia.cpp b/src/lib/block/camellia/camellia.cpp index ea84fa313..89db6f8b9 100644 --- a/src/lib/block/camellia/camellia.cpp +++ b/src/lib/block/camellia/camellia.cpp @@ -577,12 +577,12 @@ uint64_t F_SLOW(uint64_t v, uint64_t K) const uint64_t x = v ^ K; const uint8_t t1 = SBOX[get_byte(0, x)]; - const uint8_t t2 = rotate_left(SBOX[get_byte(1, x)], 1); - const uint8_t t3 = rotate_left(SBOX[get_byte(2, x)], 7); - const uint8_t t4 = SBOX[rotate_left(get_byte(3, x), 1)]; - const uint8_t t5 = rotate_left(SBOX[get_byte(4, x)], 1); - const uint8_t t6 = rotate_left(SBOX[get_byte(5, x)], 7); - const uint8_t t7 = SBOX[rotate_left(get_byte(6, x), 1)]; + const uint8_t t2 = rotl<1>(SBOX[get_byte(1, x)]); + const uint8_t t3 = rotl<7>(SBOX[get_byte(2, x)]); + const uint8_t t4 = SBOX[rotl<1>(get_byte(3, x))]; + const uint8_t t5 = rotl<1>(SBOX[get_byte(4, x)]); + const uint8_t t6 = rotl<7>(SBOX[get_byte(5, x)]); + const uint8_t t7 = SBOX[rotl<1>(get_byte(6, x))]; const uint8_t t8 = SBOX[get_byte(7, x)]; const uint8_t y1 = t1 ^ t3 ^ t4 ^ t6 ^ t7 ^ t8; @@ -619,7 +619,7 @@ inline uint64_t FL(uint64_t v, uint64_t K) const uint32_t k1 = static_cast<uint32_t>(K >> 32); const uint32_t k2 = static_cast<uint32_t>(K & 0xFFFFFFFF); - x2 ^= rotate_left(x1 & k1, 1); + x2 ^= rotl<1>(x1 & k1); x1 ^= (x2 | k2); return ((static_cast<uint64_t>(x1) << 32) | x2); @@ -634,7 +634,7 @@ inline uint64_t FLINV(uint64_t v, uint64_t K) const uint32_t k2 = static_cast<uint32_t>(K & 0xFFFFFFFF); x1 ^= (x2 | k2); - x2 ^= rotate_left(x1 & k1, 1); + x2 ^= rotl<1>(x1 & k1); return ((static_cast<uint64_t>(x1) << 32) | x2); } diff --git a/src/lib/block/cast/cast128.cpp b/src/lib/block/cast/cast128.cpp index d6ded8cd2..d54d0614e 100644 --- a/src/lib/block/cast/cast128.cpp +++ b/src/lib/block/cast/cast128.cpp @@ -18,7 +18,7 @@ namespace { */ inline uint32_t R1(uint32_t R, uint32_t MK, uint8_t RK) { - uint32_t T = rotate_left(MK + R, RK); + const uint32_t T = rotl_var(MK + R, RK); return (CAST_SBOX1[get_byte(0, T)] ^ CAST_SBOX2[get_byte(1, T)]) - CAST_SBOX3[get_byte(2, T)] + CAST_SBOX4[get_byte(3, T)]; } @@ -28,7 +28,7 @@ inline uint32_t R1(uint32_t R, uint32_t MK, uint8_t RK) */ inline uint32_t R2(uint32_t R, uint32_t MK, uint8_t RK) { - uint32_t T = rotate_left(MK ^ R, RK); + const uint32_t T = rotl_var(MK ^ R, RK); return (CAST_SBOX1[get_byte(0, T)] - CAST_SBOX2[get_byte(1, T)] + CAST_SBOX3[get_byte(2, T)]) ^ CAST_SBOX4[get_byte(3, T)]; } @@ -38,7 +38,7 @@ inline uint32_t R2(uint32_t R, uint32_t MK, uint8_t RK) */ inline uint32_t R3(uint32_t R, uint32_t MK, uint8_t RK) { - uint32_t T = rotate_left(MK - R, RK); + const uint32_t T = rotl_var(MK - R, RK); return ((CAST_SBOX1[get_byte(0, T)] + CAST_SBOX2[get_byte(1, T)]) ^ CAST_SBOX3[get_byte(2, T)]) - CAST_SBOX4[get_byte(3, T)]; } diff --git a/src/lib/block/cast/cast256.cpp b/src/lib/block/cast/cast256.cpp index a4a7dbd36..b4aa49166 100644 --- a/src/lib/block/cast/cast256.cpp +++ b/src/lib/block/cast/cast256.cpp @@ -16,31 +16,31 @@ namespace { /* * CAST-256 Round Type 1 */ -void round1(uint32_t& out, uint32_t in, uint32_t mask, uint32_t rot) +void round1(uint32_t& out, uint32_t in, uint32_t MK, uint32_t RK) { - uint32_t temp = rotate_left(mask + in, rot); - out ^= (CAST_SBOX1[get_byte(0, temp)] ^ CAST_SBOX2[get_byte(1, temp)]) - - CAST_SBOX3[get_byte(2, temp)] + CAST_SBOX4[get_byte(3, temp)]; + const uint32_t T = rotl_var(MK + in, RK); + out ^= (CAST_SBOX1[get_byte(0, T)] ^ CAST_SBOX2[get_byte(1, T)]) - + CAST_SBOX3[get_byte(2, T)] + CAST_SBOX4[get_byte(3, T)]; } /* * CAST-256 Round Type 2 */ -void round2(uint32_t& out, uint32_t in, uint32_t mask, uint32_t rot) +void round2(uint32_t& out, uint32_t in, uint32_t MK, uint32_t RK) { - uint32_t temp = rotate_left(mask ^ in, rot); - out ^= (CAST_SBOX1[get_byte(0, temp)] - CAST_SBOX2[get_byte(1, temp)] + - CAST_SBOX3[get_byte(2, temp)]) ^ CAST_SBOX4[get_byte(3, temp)]; + const uint32_t T = rotl_var(MK ^ in, RK); + out ^= (CAST_SBOX1[get_byte(0, T)] - CAST_SBOX2[get_byte(1, T)] + + CAST_SBOX3[get_byte(2, T)]) ^ CAST_SBOX4[get_byte(3, T)]; } /* * CAST-256 Round Type 3 */ -void round3(uint32_t& out, uint32_t in, uint32_t mask, uint32_t rot) +void round3(uint32_t& out, uint32_t in, uint32_t MK, uint32_t RK) { - uint32_t temp = rotate_left(mask - in, rot); - out ^= ((CAST_SBOX1[get_byte(0, temp)] + CAST_SBOX2[get_byte(1, temp)]) ^ - CAST_SBOX3[get_byte(2, temp)]) - CAST_SBOX4[get_byte(3, temp)]; + const uint32_t T = rotl_var(MK - in, RK); + out ^= ((CAST_SBOX1[get_byte(0, T)] + CAST_SBOX2[get_byte(1, T)]) ^ + CAST_SBOX3[get_byte(2, T)]) - CAST_SBOX4[get_byte(3, T)]; } } diff --git a/src/lib/block/des/des.cpp b/src/lib/block/des/des.cpp index 44f315047..15c2adb66 100644 --- a/src/lib/block/des/des.cpp +++ b/src/lib/block/des/des.cpp @@ -91,16 +91,16 @@ void des_encrypt(uint32_t& L, uint32_t& R, { uint32_t T0, T1; - T0 = rotate_right(R, 4) ^ round_key[2*i]; - T1 = R ^ round_key[2*i + 1]; + T0 = rotr<4>(R) ^ round_key[2*i]; + T1 = R ^ round_key[2*i + 1]; L ^= DES_SPBOX1[get_byte(0, T0)] ^ DES_SPBOX2[get_byte(0, T1)] ^ DES_SPBOX3[get_byte(1, T0)] ^ DES_SPBOX4[get_byte(1, T1)] ^ DES_SPBOX5[get_byte(2, T0)] ^ DES_SPBOX6[get_byte(2, T1)] ^ DES_SPBOX7[get_byte(3, T0)] ^ DES_SPBOX8[get_byte(3, T1)]; - T0 = rotate_right(L, 4) ^ round_key[2*i + 2]; - T1 = L ^ round_key[2*i + 3]; + T0 = rotr<4>(L) ^ round_key[2*i + 2]; + T1 = L ^ round_key[2*i + 3]; R ^= DES_SPBOX1[get_byte(0, T0)] ^ DES_SPBOX2[get_byte(0, T1)] ^ DES_SPBOX3[get_byte(1, T0)] ^ DES_SPBOX4[get_byte(1, T1)] ^ @@ -119,16 +119,16 @@ void des_decrypt(uint32_t& L, uint32_t& R, { uint32_t T0, T1; - T0 = rotate_right(R, 4) ^ round_key[2*i - 2]; - T1 = R ^ round_key[2*i - 1]; + T0 = rotr<4>(R) ^ round_key[2*i - 2]; + T1 = R ^ round_key[2*i - 1]; L ^= DES_SPBOX1[get_byte(0, T0)] ^ DES_SPBOX2[get_byte(0, T1)] ^ DES_SPBOX3[get_byte(1, T0)] ^ DES_SPBOX4[get_byte(1, T1)] ^ DES_SPBOX5[get_byte(2, T0)] ^ DES_SPBOX6[get_byte(2, T1)] ^ DES_SPBOX7[get_byte(3, T0)] ^ DES_SPBOX8[get_byte(3, T1)]; - T0 = rotate_right(L, 4) ^ round_key[2*i - 4]; - T1 = L ^ round_key[2*i - 3]; + T0 = rotr<4>(L) ^ round_key[2*i - 4]; + T1 = L ^ round_key[2*i - 3]; R ^= DES_SPBOX1[get_byte(0, T0)] ^ DES_SPBOX2[get_byte(0, T1)] ^ DES_SPBOX3[get_byte(1, T0)] ^ DES_SPBOX4[get_byte(1, T1)] ^ @@ -160,7 +160,7 @@ void DES::encrypt_n(const uint8_t in[], uint8_t out[], size_t blocks) const (DES_FPTAB1[get_byte(2, L)] << 1) | (DES_FPTAB2[get_byte(3, L)] << 1) | (DES_FPTAB1[get_byte(0, R)] << 4) | (DES_FPTAB1[get_byte(1, R)] << 2) | (DES_FPTAB1[get_byte(2, R)] ) | (DES_FPTAB2[get_byte(3, R)] ); - T = rotate_left(T, 32); + T = rotl<32>(T); store_be(T, out + 8*i); } @@ -188,7 +188,7 @@ void DES::decrypt_n(const uint8_t in[], uint8_t out[], size_t blocks) const (DES_FPTAB1[get_byte(0, R)] << 4) | (DES_FPTAB1[get_byte(1, R)] << 2) | (DES_FPTAB1[get_byte(2, R)] ) | (DES_FPTAB2[get_byte(3, R)] ); - T = rotate_left(T, 32); + T = rotl<32>(T); store_be(T, out + BLOCK_SIZE*i); } @@ -232,7 +232,7 @@ void TripleDES::encrypt_n(const uint8_t in[], uint8_t out[], size_t blocks) cons (DES_FPTAB1[get_byte(0, R)] << 4) | (DES_FPTAB1[get_byte(1, R)] << 2) | (DES_FPTAB1[get_byte(2, R)] ) | (DES_FPTAB2[get_byte(3, R)] ); - T = rotate_left(T, 32); + T = rotl<32>(T); store_be(T, out); @@ -265,7 +265,7 @@ void TripleDES::decrypt_n(const uint8_t in[], uint8_t out[], size_t blocks) cons (DES_FPTAB1[get_byte(0, R)] << 4) | (DES_FPTAB1[get_byte(1, R)] << 2) | (DES_FPTAB1[get_byte(2, R)] ) | (DES_FPTAB2[get_byte(3, R)] ); - T = rotate_left(T, 32); + T = rotl<32>(T); store_be(T, out); diff --git a/src/lib/block/gost_28147/gost_28147.cpp b/src/lib/block/gost_28147/gost_28147.cpp index f73ac5910..ffe9b5d66 100644 --- a/src/lib/block/gost_28147/gost_28147.cpp +++ b/src/lib/block/gost_28147/gost_28147.cpp @@ -12,11 +12,17 @@ namespace Botan { uint8_t GOST_28147_89_Params::sbox_entry(size_t row, size_t col) const { - uint8_t x = m_sboxes[4 * col + (row / 2)]; - + const uint8_t x = m_sboxes[4 * col + (row / 2)]; return (row % 2 == 0) ? (x >> 4) : (x & 0x0F); } +uint8_t GOST_28147_89_Params::sbox_pair(size_t row, size_t col) const + { + const uint8_t x = m_sboxes[4 * (col % 16) + row]; + const uint8_t y = m_sboxes[4 * (col / 16) + row]; + return (x >> 4) | (y << 4); + } + GOST_28147_89_Params::GOST_28147_89_Params(const std::string& n) : m_name(n) { // Encoded in the packed fromat from RFC 4357 @@ -53,13 +59,14 @@ GOST_28147_89_Params::GOST_28147_89_Params(const std::string& n) : m_name(n) GOST_28147_89::GOST_28147_89(const GOST_28147_89_Params& param) : m_SBOX(1024) { // Convert the parallel 4x4 sboxes into larger word-based sboxes - for(size_t i = 0; i != 4; ++i) - for(size_t j = 0; j != 256; ++j) - { - const uint32_t T = (param.sbox_entry(2*i , j % 16)) | - (param.sbox_entry(2*i+1, j / 16) << 4); - m_SBOX[256*i+j] = rotate_left(T, (11+8*i) % 32); - } + + for(size_t i = 0; i != 256; ++i) + { + m_SBOX[i ] = rotl<11, uint32_t>(param.sbox_pair(0, i)); + m_SBOX[i+256] = rotl<19, uint32_t>(param.sbox_pair(1, i)); + m_SBOX[i+512] = rotl<27, uint32_t>(param.sbox_pair(2, i)); + m_SBOX[i+768] = rotl< 3, uint32_t>(param.sbox_pair(3, i)); + } } std::string GOST_28147_89::name() const diff --git a/src/lib/block/gost_28147/gost_28147.h b/src/lib/block/gost_28147/gost_28147.h index 09581191e..34e45779b 100644 --- a/src/lib/block/gost_28147/gost_28147.h +++ b/src/lib/block/gost_28147/gost_28147.h @@ -34,6 +34,12 @@ class BOTAN_PUBLIC_API(2,0) GOST_28147_89_Params final std::string param_name() const { return m_name; } /** + * Return a representation used for building larger tables + * For internal use + */ + uint8_t sbox_pair(size_t row, size_t col) const; + + /** * Default GOST parameters are the ones given in GOST R 34.11 for * testing purposes; these sboxes are also used by Crypto++, and, * at least according to Wikipedia, the Central Bank of Russian diff --git a/src/lib/block/kasumi/kasumi.cpp b/src/lib/block/kasumi/kasumi.cpp index ed2524e0b..a9b5d8274 100644 --- a/src/lib/block/kasumi/kasumi.cpp +++ b/src/lib/block/kasumi/kasumi.cpp @@ -121,8 +121,8 @@ void KASUMI::encrypt_n(const uint8_t in[], uint8_t out[], size_t blocks) const { const uint16_t* K = &m_EK[8*j]; - uint16_t R = B1 ^ (rotate_left(B0, 1) & K[0]); - uint16_t L = B0 ^ (rotate_left(R, 1) | K[1]); + uint16_t R = B1 ^ (rotl<1>(B0) & K[0]); + uint16_t L = B0 ^ (rotl<1>(R) | K[1]); L = FI(L ^ K[ 2], K[ 3]) ^ R; R = FI(R ^ K[ 4], K[ 5]) ^ L; @@ -135,8 +135,8 @@ void KASUMI::encrypt_n(const uint8_t in[], uint8_t out[], size_t blocks) const L = FI(L ^ K[12], K[13]) ^ R; R = FI(R ^ K[14], K[15]) ^ L; - R ^= (rotate_left(L, 1) & K[8]); - L ^= (rotate_left(R, 1) | K[9]); + R ^= (rotl<1>(L) & K[8]); + L ^= (rotl<1>(R) | K[9]); B0 ^= L; B1 ^= R; @@ -171,14 +171,14 @@ void KASUMI::decrypt_n(const uint8_t in[], uint8_t out[], size_t blocks) const R = FI(R ^ K[12], K[13]) ^ L; L = FI(L ^ K[14], K[15]) ^ R; - L ^= (rotate_left(R, 1) & K[8]); - R ^= (rotate_left(L, 1) | K[9]); + L ^= (rotl<1>(R) & K[8]); + R ^= (rotl<1>(L) | K[9]); R = B0 ^= R; L = B1 ^= L; - L ^= (rotate_left(R, 1) & K[0]); - R ^= (rotate_left(L, 1) | K[1]); + L ^= (rotl<1>(R) & K[0]); + R ^= (rotl<1>(L) | K[1]); R = FI(R ^ K[2], K[3]) ^ L; L = FI(L ^ K[4], K[5]) ^ R; @@ -214,13 +214,13 @@ void KASUMI::key_schedule(const uint8_t key[], size_t) for(size_t i = 0; i != 8; ++i) { - m_EK[8*i ] = rotate_left(K[(i+0) % 8 ], 2); - m_EK[8*i+1] = rotate_left(K[(i+2) % 8 + 8], 1); - m_EK[8*i+2] = rotate_left(K[(i+1) % 8 ], 5); + m_EK[8*i ] = rotl<2>(K[(i+0) % 8]); + m_EK[8*i+1] = rotl<1>(K[(i+2) % 8 + 8]); + m_EK[8*i+2] = rotl<5>(K[(i+1) % 8]); m_EK[8*i+3] = K[(i+4) % 8 + 8]; - m_EK[8*i+4] = rotate_left(K[(i+5) % 8 ], 8); + m_EK[8*i+4] = rotl<8>(K[(i+5) % 8]); m_EK[8*i+5] = K[(i+3) % 8 + 8]; - m_EK[8*i+6] = rotate_left(K[(i+6) % 8 ], 13); + m_EK[8*i+6] = rotl<13>(K[(i+6) % 8]); m_EK[8*i+7] = K[(i+7) % 8 + 8]; } } diff --git a/src/lib/block/noekeon/noekeon.cpp b/src/lib/block/noekeon/noekeon.cpp index c82badd4c..a7f60a0fd 100644 --- a/src/lib/block/noekeon/noekeon.cpp +++ b/src/lib/block/noekeon/noekeon.cpp @@ -21,7 +21,7 @@ inline void theta(uint32_t& A0, uint32_t& A1, const uint32_t EK[4]) { uint32_t T = A0 ^ A2; - T ^= rotate_left(T, 8) ^ rotate_right(T, 8); + T ^= rotl<8>(T) ^ rotr<8>(T); A1 ^= T; A3 ^= T; @@ -31,7 +31,7 @@ inline void theta(uint32_t& A0, uint32_t& A1, A3 ^= EK[3]; T = A1 ^ A3; - T ^= rotate_left(T, 8) ^ rotate_right(T, 8); + T ^= rotl<8>(T) ^ rotr<8>(T); A0 ^= T; A2 ^= T; } @@ -43,12 +43,12 @@ inline void theta(uint32_t& A0, uint32_t& A1, uint32_t& A2, uint32_t& A3) { uint32_t T = A0 ^ A2; - T ^= rotate_left(T, 8) ^ rotate_right(T, 8); + T ^= rotl<8>(T) ^ rotr<8>(T); A1 ^= T; A3 ^= T; T = A1 ^ A3; - T ^= rotate_left(T, 8) ^ rotate_right(T, 8); + T ^= rotl<8>(T) ^ rotr<8>(T); A0 ^= T; A2 ^= T; } @@ -135,15 +135,15 @@ void Noekeon::encrypt_n(const uint8_t in[], uint8_t out[], size_t blocks) const A0 ^= RC[j]; theta(A0, A1, A2, A3, m_EK.data()); - A1 = rotate_left(A1, 1); - A2 = rotate_left(A2, 5); - A3 = rotate_left(A3, 2); + A1 = rotl<1>(A1); + A2 = rotl<5>(A2); + A3 = rotl<2>(A3); gamma(A0, A1, A2, A3); - A1 = rotate_right(A1, 1); - A2 = rotate_right(A2, 5); - A3 = rotate_right(A3, 2); + A1 = rotr<1>(A1); + A2 = rotr<5>(A2); + A3 = rotr<2>(A3); } A0 ^= RC[16]; @@ -186,15 +186,15 @@ void Noekeon::decrypt_n(const uint8_t in[], uint8_t out[], size_t blocks) const theta(A0, A1, A2, A3, m_DK.data()); A0 ^= RC[j]; - A1 = rotate_left(A1, 1); - A2 = rotate_left(A2, 5); - A3 = rotate_left(A3, 2); + A1 = rotl<1>(A1); + A2 = rotl<5>(A2); + A3 = rotl<2>(A3); gamma(A0, A1, A2, A3); - A1 = rotate_right(A1, 1); - A2 = rotate_right(A2, 5); - A3 = rotate_right(A3, 2); + A1 = rotr<1>(A1); + A2 = rotr<5>(A2); + A3 = rotr<2>(A3); } theta(A0, A1, A2, A3, m_DK.data()); @@ -222,15 +222,15 @@ void Noekeon::key_schedule(const uint8_t key[], size_t) A0 ^= RC[i]; theta(A0, A1, A2, A3); - A1 = rotate_left(A1, 1); - A2 = rotate_left(A2, 5); - A3 = rotate_left(A3, 2); + A1 = rotl<1>(A1); + A2 = rotl<5>(A2); + A3 = rotl<2>(A3); gamma(A0, A1, A2, A3); - A1 = rotate_right(A1, 1); - A2 = rotate_right(A2, 5); - A3 = rotate_right(A3, 2); + A1 = rotr<1>(A1); + A2 = rotr<5>(A2); + A3 = rotr<2>(A3); } A0 ^= RC[16]; diff --git a/src/lib/block/noekeon/noekeon_simd/noekeon_simd.cpp b/src/lib/block/noekeon/noekeon_simd/noekeon_simd.cpp index a77ba7b8c..f9a696d29 100644 --- a/src/lib/block/noekeon/noekeon_simd/noekeon_simd.cpp +++ b/src/lib/block/noekeon/noekeon_simd/noekeon_simd.cpp @@ -16,12 +16,7 @@ namespace Botan { #define NOK_SIMD_THETA(A0, A1, A2, A3, K0, K1, K2, K3) \ do { \ SIMD_32 T = A0 ^ A2; \ - SIMD_32 T_l8 = T; \ - SIMD_32 T_r8 = T; \ - T_l8.rotate_left(8); \ - T_r8.rotate_right(8); \ - T ^= T_l8; \ - T ^= T_r8; \ + T ^= T.rotl<8>() ^ T.rotr<8>(); \ A1 ^= T; \ A3 ^= T; \ \ @@ -31,12 +26,7 @@ namespace Botan { A3 ^= K3; \ \ T = A1 ^ A3; \ - T_l8 = T; \ - T_r8 = T; \ - T_l8.rotate_left(8); \ - T_r8.rotate_right(8); \ - T ^= T_l8; \ - T ^= T_r8; \ + T ^= T.rotl<8>() ^ T.rotr<8>(); \ A0 ^= T; \ A2 ^= T; \ } while(0) @@ -83,15 +73,15 @@ void Noekeon::simd_encrypt_4(const uint8_t in[], uint8_t out[]) const NOK_SIMD_THETA(A0, A1, A2, A3, K0, K1, K2, K3); - A1.rotate_left(1); - A2.rotate_left(5); - A3.rotate_left(2); + A1 = A1.rotl<1>(); + A2 = A2.rotl<5>(); + A3 = A3.rotl<2>(); NOK_SIMD_GAMMA(A0, A1, A2, A3); - A1.rotate_right(1); - A2.rotate_right(5); - A3.rotate_right(2); + A1 = A1.rotr<1>(); + A2 = A2.rotr<5>(); + A3 = A3.rotr<2>(); } A0 ^= SIMD_32::splat(RC[16]); @@ -128,15 +118,15 @@ void Noekeon::simd_decrypt_4(const uint8_t in[], uint8_t out[]) const A0 ^= SIMD_32::splat(RC[16-i]); - A1.rotate_left(1); - A2.rotate_left(5); - A3.rotate_left(2); + A1 = A1.rotl<1>(); + A2 = A2.rotl<5>(); + A3 = A3.rotl<2>(); NOK_SIMD_GAMMA(A0, A1, A2, A3); - A1.rotate_right(1); - A2.rotate_right(5); - A3.rotate_right(2); + A1 = A1.rotr<1>(); + A2 = A2.rotr<5>(); + A3 = A3.rotr<2>(); } NOK_SIMD_THETA(A0, A1, A2, A3, K0, K1, K2, K3); diff --git a/src/lib/block/serpent/serpent.cpp b/src/lib/block/serpent/serpent.cpp index 93af81231..6e1d79766 100644 --- a/src/lib/block/serpent/serpent.cpp +++ b/src/lib/block/serpent/serpent.cpp @@ -22,11 +22,11 @@ namespace { */ inline void transform(uint32_t& B0, uint32_t& B1, uint32_t& B2, uint32_t& B3) { - B0 = rotate_left(B0, 13); B2 = rotate_left(B2, 3); - B1 ^= B0 ^ B2; B3 ^= B2 ^ (B0 << 3); - B1 = rotate_left(B1, 1); B3 = rotate_left(B3, 7); - B0 ^= B1 ^ B3; B2 ^= B3 ^ (B1 << 7); - B0 = rotate_left(B0, 5); B2 = rotate_left(B2, 22); + B0 = rotl<13>(B0); B2 = rotl<3>(B2); + B1 ^= B0 ^ B2; B3 ^= B2 ^ (B0 << 3); + B1 = rotl<1>(B1); B3 = rotl<7>(B3); + B0 ^= B1 ^ B3; B2 ^= B3 ^ (B1 << 7); + B0 = rotl<5>(B0); B2 = rotl<22>(B2); } /* @@ -34,11 +34,11 @@ inline void transform(uint32_t& B0, uint32_t& B1, uint32_t& B2, uint32_t& B3) */ inline void i_transform(uint32_t& B0, uint32_t& B1, uint32_t& B2, uint32_t& B3) { - B2 = rotate_right(B2, 22); B0 = rotate_right(B0, 5); - B2 ^= B3 ^ (B1 << 7); B0 ^= B1 ^ B3; - B3 = rotate_right(B3, 7); B1 = rotate_right(B1, 1); - B3 ^= B2 ^ (B0 << 3); B1 ^= B0 ^ B2; - B2 = rotate_right(B2, 3); B0 = rotate_right(B0, 13); + B2 = rotr<22>(B2); B0 = rotr<5>(B0); + B2 ^= B3 ^ (B1 << 7); B0 ^= B1 ^ B3; + B3 = rotr<7>(B3); B1 = rotr<1>(B1); + B3 ^= B2 ^ (B0 << 3); B1 ^= B0 ^ B2; + B2 = rotr<3>(B2); B0 = rotr<13>(B0); } } @@ -192,7 +192,7 @@ void Serpent::key_schedule(const uint8_t key[], size_t length) for(size_t i = 8; i != 140; ++i) { uint32_t wi = W[i-8] ^ W[i-5] ^ W[i-3] ^ W[i-1] ^ PHI ^ uint32_t(i-8); - W[i] = rotate_left(wi, 11); + W[i] = rotl<11>(wi); } SBoxE1(W[ 20],W[ 21],W[ 22],W[ 23]); diff --git a/src/lib/block/serpent/serpent_simd/serpent_simd.cpp b/src/lib/block/serpent/serpent_simd/serpent_simd.cpp index 94b3cf9ad..b184b0d4a 100644 --- a/src/lib/block/serpent/serpent_simd/serpent_simd.cpp +++ b/src/lib/block/serpent/serpent_simd/serpent_simd.cpp @@ -24,30 +24,30 @@ namespace Botan { */ #define transform(B0, B1, B2, B3) \ do { \ - B0.rotate_left(13); \ - B2.rotate_left(3); \ + B0 = B0.rotl<13>(); \ + B2 = B2.rotl<3>(); \ B1 ^= B0 ^ B2; \ B3 ^= B2 ^ (B0 << 3); \ - B1.rotate_left(1); \ - B3.rotate_left(7); \ + B1 = B1.rotl<1>(); \ + B3 = B3.rotl<7>(); \ B0 ^= B1 ^ B3; \ B2 ^= B3 ^ (B1 << 7); \ - B0.rotate_left(5); \ - B2.rotate_left(22); \ + B0 = B0.rotl<5>(); \ + B2 = B2.rotl<22>(); \ } while(0); #define i_transform(B0, B1, B2, B3) \ do { \ - B2.rotate_right(22); \ - B0.rotate_right(5); \ + B2 = B2.rotr<22>(); \ + B0 = B0.rotr<5>(); \ B2 ^= B3 ^ (B1 << 7); \ B0 ^= B1 ^ B3; \ - B3.rotate_right(7); \ - B1.rotate_right(1); \ + B3 = B3.rotr<7>(); \ + B1 = B1.rotr<1>(); \ B3 ^= B2 ^ (B0 << 3); \ B1 ^= B0 ^ B2; \ - B2.rotate_right(3); \ - B0.rotate_right(13); \ + B2 = B2.rotr<3>(); \ + B0 = B0.rotr<13>(); \ } while(0); /* diff --git a/src/lib/block/shacal2/shacal2.cpp b/src/lib/block/shacal2/shacal2.cpp index 30ad711db..12c87c426 100644 --- a/src/lib/block/shacal2/shacal2.cpp +++ b/src/lib/block/shacal2/shacal2.cpp @@ -17,8 +17,8 @@ inline void SHACAL2_Fwd(uint32_t A, uint32_t B, uint32_t C, uint32_t& D, uint32_t E, uint32_t F, uint32_t G, uint32_t& H, uint32_t RK) { - const uint32_t A_rho = rotate_right(A, 2) ^ rotate_right(A, 13) ^ rotate_right(A, 22); - const uint32_t E_rho = rotate_right(E, 6) ^ rotate_right(E, 11) ^ rotate_right(E, 25); + const uint32_t A_rho = rotr<2>(A) ^ rotr<13>(A) ^ rotr<22>(A); + const uint32_t E_rho = rotr<6>(E) ^ rotr<11>(E) ^ rotr<25>(E); H += E_rho + ((E & F) ^ (~E & G)) + RK; D += H; @@ -29,8 +29,8 @@ inline void SHACAL2_Rev(uint32_t A, uint32_t B, uint32_t C, uint32_t& D, uint32_t E, uint32_t F, uint32_t G, uint32_t& H, uint32_t RK) { - const uint32_t A_rho = rotate_right(A, 2) ^ rotate_right(A, 13) ^ rotate_right(A, 22); - const uint32_t E_rho = rotate_right(E, 6) ^ rotate_right(E, 11) ^ rotate_right(E, 25); + const uint32_t A_rho = rotr<2>(A) ^ rotr<13>(A) ^ rotr<22>(A); + const uint32_t E_rho = rotr<6>(E) ^ rotr<11>(E) ^ rotr<25>(E); H -= A_rho + ((A & B) | ((A | B) & C)); D -= H; @@ -175,8 +175,8 @@ void SHACAL2::key_schedule(const uint8_t key[], size_t len) for(size_t i = 16; i != 64; ++i) { - const uint32_t sigma0_15 = rotate_right(m_RK[i-15], 7) ^ rotate_right(m_RK[i-15], 18) ^ (m_RK[i-15] >> 3); - const uint32_t sigma1_2 = rotate_right(m_RK[i-2], 17) ^ rotate_right(m_RK[i-2], 19) ^ (m_RK[i-2] >> 10); + const uint32_t sigma0_15 = rotr<7>(m_RK[i-15]) ^ rotr<18>(m_RK[i-15]) ^ (m_RK[i-15] >> 3); + const uint32_t sigma1_2 = rotr<17>(m_RK[i-2]) ^ rotr<19>(m_RK[i-2]) ^ (m_RK[i-2] >> 10); m_RK[i] = m_RK[i-16] + sigma0_15 + m_RK[i-7] + sigma1_2; } diff --git a/src/lib/block/shacal2/shacal2_simd/shacal2_simd.cpp b/src/lib/block/shacal2/shacal2_simd/shacal2_simd.cpp index a4324c8fb..bdcac1482 100644 --- a/src/lib/block/shacal2/shacal2_simd/shacal2_simd.cpp +++ b/src/lib/block/shacal2/shacal2_simd/shacal2_simd.cpp @@ -17,9 +17,9 @@ void SHACAL2_Fwd(const SIMD_32& A, const SIMD_32& B, const SIMD_32& C, SIMD_32& const SIMD_32& E, const SIMD_32& F, const SIMD_32& G, SIMD_32& H, uint32_t RK) { - H += E.rho(6,11,25) + ((E & F) ^ (~E & G)) + SIMD_32::splat(RK); + H += E.rho<6,11,25>() + ((E & F) ^ (~E & G)) + SIMD_32::splat(RK); D += H; - H += A.rho(2,13,22) + ((A & B) | ((A | B) & C)); + H += A.rho<2,13,22>() + ((A & B) | ((A | B) & C)); } inline @@ -27,9 +27,9 @@ void SHACAL2_Rev(const SIMD_32& A, const SIMD_32& B, const SIMD_32& C, SIMD_32& const SIMD_32& E, const SIMD_32& F, const SIMD_32& G, SIMD_32& H, uint32_t RK) { - H -= A.rho(2,13,22) + ((A & B) | ((A | B) & C)); + H -= A.rho<2,13,22>() + ((A & B) | ((A | B) & C)); D -= H; - H -= E.rho(6,11,25) + ((E & F) ^ (~E & G)) + SIMD_32::splat(RK); + H -= E.rho<6,11,25>() + ((E & F) ^ (~E & G)) + SIMD_32::splat(RK); } } diff --git a/src/lib/block/sm4/sm4.cpp b/src/lib/block/sm4/sm4.cpp index 979491566..42c865faf 100644 --- a/src/lib/block/sm4/sm4.cpp +++ b/src/lib/block/sm4/sm4.cpp @@ -46,7 +46,7 @@ inline uint32_t T(uint32_t b) const uint32_t t = make_uint32(SBOX[b0], SBOX[b1], SBOX[b2], SBOX[b3]); // L linear transform - return t ^ rotate_left(t, 2) ^ rotate_left(t, 10) ^ rotate_left(t, 18) ^ rotate_left(t, 24); + return t ^ rotl<2>(t) ^ rotl<10>(t) ^ rotl<18>(t) ^ rotl<24>(t); } // Variant of T for key schedule @@ -59,7 +59,7 @@ inline uint32_t Tp(uint32_t b) const uint32_t t = make_uint32(SBOX[b0], SBOX[b1], SBOX[b2], SBOX[b3]); // L' linear transform - return t ^ rotate_left(t, 13) ^ rotate_left(t, 23); + return t ^ rotl<13>(t) ^ rotl<23>(t); } } diff --git a/src/lib/block/threefish/threefish.cpp b/src/lib/block/threefish/threefish.cpp index 99ce135d5..937a673fd 100644 --- a/src/lib/block/threefish/threefish.cpp +++ b/src/lib/block/threefish/threefish.cpp @@ -17,10 +17,10 @@ namespace Botan { X1 += X5; \ X2 += X6; \ X3 += X7; \ - X4 = rotate_left(X4, ROT1); \ - X5 = rotate_left(X5, ROT2); \ - X6 = rotate_left(X6, ROT3); \ - X7 = rotate_left(X7, ROT4); \ + X4 = rotl<ROT1>(X4); \ + X5 = rotl<ROT2>(X5); \ + X6 = rotl<ROT3>(X6); \ + X7 = rotl<ROT4>(X7); \ X4 ^= X0; \ X5 ^= X1; \ X6 ^= X2; \ @@ -177,10 +177,10 @@ void Threefish_512::decrypt_n(const uint8_t in[], uint8_t out[], size_t blocks) X5 ^= X1; \ X6 ^= X2; \ X7 ^= X3; \ - X4 = rotate_right(X4, ROT1); \ - X5 = rotate_right(X5, ROT2); \ - X6 = rotate_right(X6, ROT3); \ - X7 = rotate_right(X7, ROT4); \ + X4 = rotr<ROT1>(X4); \ + X5 = rotr<ROT2>(X5); \ + X6 = rotr<ROT3>(X6); \ + X7 = rotr<ROT4>(X7); \ X0 -= X4; \ X1 -= X5; \ X2 -= X6; \ diff --git a/src/lib/block/twofish/twofish.cpp b/src/lib/block/twofish/twofish.cpp index 51ef01ea9..3a09af8da 100644 --- a/src/lib/block/twofish/twofish.cpp +++ b/src/lib/block/twofish/twofish.cpp @@ -41,8 +41,8 @@ void Twofish::encrypt_n(const uint8_t in[], uint8_t out[], size_t blocks) const Y += X + m_RK[2*j + 9]; X += m_RK[2*j + 8]; - C = rotate_right(C ^ X, 1); - D = rotate_left(D, 1) ^ Y; + C = rotr<1>(C ^ X); + D = rotl<1>(D) ^ Y; X = m_SB[ get_byte(3, C)] ^ m_SB[256+get_byte(2, C)] ^ m_SB[512+get_byte(1, C)] ^ m_SB[768+get_byte(0, C)]; @@ -52,8 +52,8 @@ void Twofish::encrypt_n(const uint8_t in[], uint8_t out[], size_t blocks) const Y += X + m_RK[2*j + 11]; X += m_RK[2*j + 10]; - A = rotate_right(A ^ X, 1); - B = rotate_left(B, 1) ^ Y; + A = rotr<1>(A ^ X); + B = rotl<1>(B) ^ Y; } C ^= m_RK[4]; @@ -92,8 +92,8 @@ void Twofish::decrypt_n(const uint8_t in[], uint8_t out[], size_t blocks) const Y += X + m_RK[39 - 2*j]; X += m_RK[38 - 2*j]; - C = rotate_left(C, 1) ^ X; - D = rotate_right(D ^ Y, 1); + C = rotl<1>(C) ^ X; + D = rotr<1>(D ^ Y); X = m_SB[ get_byte(3, C)] ^ m_SB[256+get_byte(2, C)] ^ m_SB[512+get_byte(1, C)] ^ m_SB[768+get_byte(0, C)]; @@ -103,8 +103,8 @@ void Twofish::decrypt_n(const uint8_t in[], uint8_t out[], size_t blocks) const Y += X + m_RK[37 - 2*j]; X += m_RK[36 - 2*j]; - A = rotate_left(A, 1) ^ X; - B = rotate_right(B ^ Y, 1); + A = rotl<1>(A) ^ X; + B = rotr<1>(B ^ Y); } C ^= m_RK[0]; @@ -167,11 +167,11 @@ void Twofish::key_schedule(const uint8_t key[], size_t length) MDS1[Q0[Q1[i+1]^key[13]]^key[ 5]] ^ MDS2[Q1[Q0[i+1]^key[14]]^key[ 6]] ^ MDS3[Q1[Q1[i+1]^key[15]]^key[ 7]]; - Y = rotate_left(Y, 8); + Y = rotl<8>(Y); X += Y; Y += X; m_RK[i] = X; - m_RK[i+1] = rotate_left(Y, 9); + m_RK[i+1] = rotl<9>(Y); } } else if(length == 24) @@ -194,11 +194,11 @@ void Twofish::key_schedule(const uint8_t key[], size_t length) MDS1[Q0[Q1[Q1[i+1]^key[21]]^key[13]]^key[ 5]] ^ MDS2[Q1[Q0[Q0[i+1]^key[22]]^key[14]]^key[ 6]] ^ MDS3[Q1[Q1[Q0[i+1]^key[23]]^key[15]]^key[ 7]]; - Y = rotate_left(Y, 8); + Y = rotl<8>(Y); X += Y; Y += X; m_RK[i] = X; - m_RK[i+1] = rotate_left(Y, 9); + m_RK[i+1] = rotl<9>(Y); } } else if(length == 32) @@ -221,11 +221,11 @@ void Twofish::key_schedule(const uint8_t key[], size_t length) MDS1[Q0[Q1[Q1[Q0[i+1]^key[29]]^key[21]]^key[13]]^key[ 5]] ^ MDS2[Q1[Q0[Q0[Q0[i+1]^key[30]]^key[22]]^key[14]]^key[ 6]] ^ MDS3[Q1[Q1[Q0[Q1[i+1]^key[31]]^key[23]]^key[15]]^key[ 7]]; - Y = rotate_left(Y, 8); + Y = rotl<8>(Y); X += Y; Y += X; m_RK[i] = X; - m_RK[i+1] = rotate_left(Y, 9); + m_RK[i+1] = rotl<9>(Y); } } } diff --git a/src/lib/hash/blake2/blake2b.cpp b/src/lib/hash/blake2/blake2b.cpp index 85171b16b..79a30de3d 100644 --- a/src/lib/hash/blake2/blake2b.cpp +++ b/src/lib/hash/blake2/blake2b.cpp @@ -93,13 +93,13 @@ void Blake2b::compress(bool lastblock) #define G(r, i, a, b, c, d) \ do { \ a = a + b + m[blake2b_sigma[r][2 * i + 0]]; \ - d = rotate_right<uint64_t>(d ^ a, 32); \ + d = rotr<32>(d ^ a); \ c = c + d; \ - b = rotate_right<uint64_t>(b ^ c, 24); \ + b = rotr<24>(b ^ c); \ a = a + b + m[blake2b_sigma[r][2 * i + 1]]; \ - d = rotate_right<uint64_t>(d ^ a, 16); \ + d = rotr<16>(d ^ a); \ c = c + d; \ - b = rotate_right<uint64_t>(b ^ c, 63); \ + b = rotr<63>(b ^ c); \ } while(0) #define ROUND(r) \ diff --git a/src/lib/hash/md4/md4.cpp b/src/lib/hash/md4/md4.cpp index 79f3a2d13..1c048a6b3 100644 --- a/src/lib/hash/md4/md4.cpp +++ b/src/lib/hash/md4/md4.cpp @@ -16,31 +16,55 @@ std::unique_ptr<HashFunction> MD4::copy_state() const namespace { -/* -* MD4 FF Function -*/ -inline void FF(uint32_t& A, uint32_t B, uint32_t C, uint32_t D, uint32_t M, uint8_t S) +inline void FF4(uint32_t& A, uint32_t& B, uint32_t& C, uint32_t& D, + uint32_t M0, uint32_t M1, uint32_t M2, uint32_t M3) + { - A += (D ^ (B & (C ^ D))) + M; - A = rotate_left(A, S); + A += (D ^ (B & (C ^ D))) + M0; + A = rotl<3>(A); + + D += (C ^ (A & (B ^ C))) + M1; + D = rotl<7>(D); + + C += (B ^ (D & (A ^ B))) + M2; + C = rotl<11>(C); + + B += (A ^ (C & (D ^ A))) + M3; + B = rotl<19>(B); } -/* -* MD4 GG Function -*/ -inline void GG(uint32_t& A, uint32_t B, uint32_t C, uint32_t D, uint32_t M, uint8_t S) +inline void GG4(uint32_t& A, uint32_t& B, uint32_t& C, uint32_t& D, + uint32_t M0, uint32_t M1, uint32_t M2, uint32_t M3) + { - A += ((B & C) | (D & (B | C))) + M + 0x5A827999; - A = rotate_left(A, S); + A += ((B & C) | (D & (B | C))) + M0 + 0x5A827999; + A = rotl<3>(A); + + D += ((A & B) | (C & (A | B))) + M1 + 0x5A827999; + D = rotl<5>(D); + + C += ((D & A) | (B & (D | A))) + M2 + 0x5A827999; + C = rotl<9>(C); + + B += ((C & D) | (A & (C | D))) + M3 + 0x5A827999; + B = rotl<13>(B); } -/* -* MD4 HH Function -*/ -inline void HH(uint32_t& A, uint32_t B, uint32_t C, uint32_t D, uint32_t M, uint8_t S) +inline void HH4(uint32_t& A, uint32_t& B, uint32_t& C, uint32_t& D, + uint32_t M0, uint32_t M1, uint32_t M2, uint32_t M3) + { - A += (B ^ C ^ D) + M + 0x6ED9EBA1; - A = rotate_left(A, S); + A += (B ^ C ^ D) + M0 + 0x6ED9EBA1; + A = rotl<3>(A); + + D += (A ^ B ^ C) + M1 + 0x6ED9EBA1; + D = rotl<9>(D); + + C += (A ^ B ^ D) + M2 + 0x6ED9EBA1; + C = rotl<11>(C); + + B += (A ^ C ^ D) + M3 + 0x6ED9EBA1; + B = rotl<15>(B); } } @@ -54,34 +78,37 @@ void MD4::compress_n(const uint8_t input[], size_t blocks) for(size_t i = 0; i != blocks; ++i) { - load_le(m_M.data(), input, m_M.size()); - - FF(A,B,C,D,m_M[ 0], 3); FF(D,A,B,C,m_M[ 1], 7); - FF(C,D,A,B,m_M[ 2],11); FF(B,C,D,A,m_M[ 3],19); - FF(A,B,C,D,m_M[ 4], 3); FF(D,A,B,C,m_M[ 5], 7); - FF(C,D,A,B,m_M[ 6],11); FF(B,C,D,A,m_M[ 7],19); - FF(A,B,C,D,m_M[ 8], 3); FF(D,A,B,C,m_M[ 9], 7); - FF(C,D,A,B,m_M[10],11); FF(B,C,D,A,m_M[11],19); - FF(A,B,C,D,m_M[12], 3); FF(D,A,B,C,m_M[13], 7); - FF(C,D,A,B,m_M[14],11); FF(B,C,D,A,m_M[15],19); - - GG(A,B,C,D,m_M[ 0], 3); GG(D,A,B,C,m_M[ 4], 5); - GG(C,D,A,B,m_M[ 8], 9); GG(B,C,D,A,m_M[12],13); - GG(A,B,C,D,m_M[ 1], 3); GG(D,A,B,C,m_M[ 5], 5); - GG(C,D,A,B,m_M[ 9], 9); GG(B,C,D,A,m_M[13],13); - GG(A,B,C,D,m_M[ 2], 3); GG(D,A,B,C,m_M[ 6], 5); - GG(C,D,A,B,m_M[10], 9); GG(B,C,D,A,m_M[14],13); - GG(A,B,C,D,m_M[ 3], 3); GG(D,A,B,C,m_M[ 7], 5); - GG(C,D,A,B,m_M[11], 9); GG(B,C,D,A,m_M[15],13); - - HH(A,B,C,D,m_M[ 0], 3); HH(D,A,B,C,m_M[ 8], 9); - HH(C,D,A,B,m_M[ 4],11); HH(B,C,D,A,m_M[12],15); - HH(A,B,C,D,m_M[ 2], 3); HH(D,A,B,C,m_M[10], 9); - HH(C,D,A,B,m_M[ 6],11); HH(B,C,D,A,m_M[14],15); - HH(A,B,C,D,m_M[ 1], 3); HH(D,A,B,C,m_M[ 9], 9); - HH(C,D,A,B,m_M[ 5],11); HH(B,C,D,A,m_M[13],15); - HH(A,B,C,D,m_M[ 3], 3); HH(D,A,B,C,m_M[11], 9); - HH(C,D,A,B,m_M[ 7],11); HH(B,C,D,A,m_M[15],15); + uint32_t M00 = load_le<uint32_t>(input, 0); + uint32_t M01 = load_le<uint32_t>(input, 1); + uint32_t M02 = load_le<uint32_t>(input, 2); + uint32_t M03 = load_le<uint32_t>(input, 3); + uint32_t M04 = load_le<uint32_t>(input, 4); + uint32_t M05 = load_le<uint32_t>(input, 5); + uint32_t M06 = load_le<uint32_t>(input, 6); + uint32_t M07 = load_le<uint32_t>(input, 7); + uint32_t M08 = load_le<uint32_t>(input, 8); + uint32_t M09 = load_le<uint32_t>(input, 9); + uint32_t M10 = load_le<uint32_t>(input, 10); + uint32_t M11 = load_le<uint32_t>(input, 11); + uint32_t M12 = load_le<uint32_t>(input, 12); + uint32_t M13 = load_le<uint32_t>(input, 13); + uint32_t M14 = load_le<uint32_t>(input, 14); + uint32_t M15 = load_le<uint32_t>(input, 15); + + FF4(A, B, C, D, M00, M01, M02, M03); + FF4(A, B, C, D, M04, M05, M06, M07); + FF4(A, B, C, D, M08, M09, M10, M11); + FF4(A, B, C, D, M12, M13, M14, M15); + + GG4(A, B, C, D, M00, M04, M08, M12); + GG4(A, B, C, D, M01, M05, M09, M13); + GG4(A, B, C, D, M02, M06, M10, M14); + GG4(A, B, C, D, M03, M07, M11, M15); + + HH4(A, B, C, D, M00, M08, M04, M12); + HH4(A, B, C, D, M02, M10, M06, M14); + HH4(A, B, C, D, M01, M09, M05, M13); + HH4(A, B, C, D, M03, M11, M07, M15); A = (m_digest[0] += A); B = (m_digest[1] += B); @@ -106,7 +133,6 @@ void MD4::copy_out(uint8_t output[]) void MD4::clear() { MDx_HashFunction::clear(); - zeroise(m_M); m_digest[0] = 0x67452301; m_digest[1] = 0xEFCDAB89; m_digest[2] = 0x98BADCFE; diff --git a/src/lib/hash/md4/md4.h b/src/lib/hash/md4/md4.h index c51cb1682..1dd857bce 100644 --- a/src/lib/hash/md4/md4.h +++ b/src/lib/hash/md4/md4.h @@ -25,7 +25,7 @@ class BOTAN_PUBLIC_API(2,0) MD4 final : public MDx_HashFunction void clear() override; - MD4() : MDx_HashFunction(64, false, true), m_M(16), m_digest(4) + MD4() : MDx_HashFunction(64, false, true), m_digest(4) { clear(); } private: @@ -33,11 +33,6 @@ class BOTAN_PUBLIC_API(2,0) MD4 final : public MDx_HashFunction void copy_out(uint8_t[]) override; /** - * The message buffer - */ - secure_vector<uint32_t> m_M; - - /** * The digest value */ secure_vector<uint32_t> m_digest; diff --git a/src/lib/hash/md5/md5.cpp b/src/lib/hash/md5/md5.cpp index 174443a67..de75ab2d1 100644 --- a/src/lib/hash/md5/md5.cpp +++ b/src/lib/hash/md5/md5.cpp @@ -19,41 +19,41 @@ namespace { /* * MD5 FF Function */ -inline void FF(uint32_t& A, uint32_t B, uint32_t C, uint32_t D, uint32_t msg, - uint8_t S, uint32_t magic) +template<size_t S> +inline void FF(uint32_t& A, uint32_t B, uint32_t C, uint32_t D, uint32_t M) { - A += (D ^ (B & (C ^ D))) + msg + magic; - A = rotate_left(A, S) + B; + A += (D ^ (B & (C ^ D))) + M; + A = rotl<S>(A) + B; } /* * MD5 GG Function */ -inline void GG(uint32_t& A, uint32_t B, uint32_t C, uint32_t D, uint32_t msg, - uint8_t S, uint32_t magic) +template<size_t S> +inline void GG(uint32_t& A, uint32_t B, uint32_t C, uint32_t D, uint32_t M) { - A += (C ^ (D & (B ^ C))) + msg + magic; - A = rotate_left(A, S) + B; + A += (C ^ (D & (B ^ C))) + M; + A = rotl<S>(A) + B; } /* * MD5 HH Function */ -inline void HH(uint32_t& A, uint32_t B, uint32_t C, uint32_t D, uint32_t msg, - uint8_t S, uint32_t magic) +template<size_t S> +inline void HH(uint32_t& A, uint32_t B, uint32_t C, uint32_t D, uint32_t M) { - A += (B ^ C ^ D) + msg + magic; - A = rotate_left(A, S) + B; + A += (B ^ C ^ D) + M; + A = rotl<S>(A) + B; } /* * MD5 II Function */ -inline void II(uint32_t& A, uint32_t B, uint32_t C, uint32_t D, uint32_t msg, - uint8_t S, uint32_t magic) +template<size_t S> +inline void II(uint32_t& A, uint32_t B, uint32_t C, uint32_t D, uint32_t M) { - A += (C ^ (B | ~D)) + msg + magic; - A = rotate_left(A, S) + B; + A += (C ^ (B | ~D)) + M; + A = rotl<S>(A) + B; } } @@ -69,41 +69,41 @@ void MD5::compress_n(const uint8_t input[], size_t blocks) { load_le(m_M.data(), input, m_M.size()); - FF(A,B,C,D,m_M[ 0], 7,0xD76AA478); FF(D,A,B,C,m_M[ 1],12,0xE8C7B756); - FF(C,D,A,B,m_M[ 2],17,0x242070DB); FF(B,C,D,A,m_M[ 3],22,0xC1BDCEEE); - FF(A,B,C,D,m_M[ 4], 7,0xF57C0FAF); FF(D,A,B,C,m_M[ 5],12,0x4787C62A); - FF(C,D,A,B,m_M[ 6],17,0xA8304613); FF(B,C,D,A,m_M[ 7],22,0xFD469501); - FF(A,B,C,D,m_M[ 8], 7,0x698098D8); FF(D,A,B,C,m_M[ 9],12,0x8B44F7AF); - FF(C,D,A,B,m_M[10],17,0xFFFF5BB1); FF(B,C,D,A,m_M[11],22,0x895CD7BE); - FF(A,B,C,D,m_M[12], 7,0x6B901122); FF(D,A,B,C,m_M[13],12,0xFD987193); - FF(C,D,A,B,m_M[14],17,0xA679438E); FF(B,C,D,A,m_M[15],22,0x49B40821); - - GG(A,B,C,D,m_M[ 1], 5,0xF61E2562); GG(D,A,B,C,m_M[ 6], 9,0xC040B340); - GG(C,D,A,B,m_M[11],14,0x265E5A51); GG(B,C,D,A,m_M[ 0],20,0xE9B6C7AA); - GG(A,B,C,D,m_M[ 5], 5,0xD62F105D); GG(D,A,B,C,m_M[10], 9,0x02441453); - GG(C,D,A,B,m_M[15],14,0xD8A1E681); GG(B,C,D,A,m_M[ 4],20,0xE7D3FBC8); - GG(A,B,C,D,m_M[ 9], 5,0x21E1CDE6); GG(D,A,B,C,m_M[14], 9,0xC33707D6); - GG(C,D,A,B,m_M[ 3],14,0xF4D50D87); GG(B,C,D,A,m_M[ 8],20,0x455A14ED); - GG(A,B,C,D,m_M[13], 5,0xA9E3E905); GG(D,A,B,C,m_M[ 2], 9,0xFCEFA3F8); - GG(C,D,A,B,m_M[ 7],14,0x676F02D9); GG(B,C,D,A,m_M[12],20,0x8D2A4C8A); - - HH(A,B,C,D,m_M[ 5], 4,0xFFFA3942); HH(D,A,B,C,m_M[ 8],11,0x8771F681); - HH(C,D,A,B,m_M[11],16,0x6D9D6122); HH(B,C,D,A,m_M[14],23,0xFDE5380C); - HH(A,B,C,D,m_M[ 1], 4,0xA4BEEA44); HH(D,A,B,C,m_M[ 4],11,0x4BDECFA9); - HH(C,D,A,B,m_M[ 7],16,0xF6BB4B60); HH(B,C,D,A,m_M[10],23,0xBEBFBC70); - HH(A,B,C,D,m_M[13], 4,0x289B7EC6); HH(D,A,B,C,m_M[ 0],11,0xEAA127FA); - HH(C,D,A,B,m_M[ 3],16,0xD4EF3085); HH(B,C,D,A,m_M[ 6],23,0x04881D05); - HH(A,B,C,D,m_M[ 9], 4,0xD9D4D039); HH(D,A,B,C,m_M[12],11,0xE6DB99E5); - HH(C,D,A,B,m_M[15],16,0x1FA27CF8); HH(B,C,D,A,m_M[ 2],23,0xC4AC5665); - - II(A,B,C,D,m_M[ 0], 6,0xF4292244); II(D,A,B,C,m_M[ 7],10,0x432AFF97); - II(C,D,A,B,m_M[14],15,0xAB9423A7); II(B,C,D,A,m_M[ 5],21,0xFC93A039); - II(A,B,C,D,m_M[12], 6,0x655B59C3); II(D,A,B,C,m_M[ 3],10,0x8F0CCC92); - II(C,D,A,B,m_M[10],15,0xFFEFF47D); II(B,C,D,A,m_M[ 1],21,0x85845DD1); - II(A,B,C,D,m_M[ 8], 6,0x6FA87E4F); II(D,A,B,C,m_M[15],10,0xFE2CE6E0); - II(C,D,A,B,m_M[ 6],15,0xA3014314); II(B,C,D,A,m_M[13],21,0x4E0811A1); - II(A,B,C,D,m_M[ 4], 6,0xF7537E82); II(D,A,B,C,m_M[11],10,0xBD3AF235); - II(C,D,A,B,m_M[ 2],15,0x2AD7D2BB); II(B,C,D,A,m_M[ 9],21,0xEB86D391); + FF< 7>(A,B,C,D,m_M[ 0]+0xD76AA478); FF<12>(D,A,B,C,m_M[ 1]+0xE8C7B756); + FF<17>(C,D,A,B,m_M[ 2]+0x242070DB); FF<22>(B,C,D,A,m_M[ 3]+0xC1BDCEEE); + FF< 7>(A,B,C,D,m_M[ 4]+0xF57C0FAF); FF<12>(D,A,B,C,m_M[ 5]+0x4787C62A); + FF<17>(C,D,A,B,m_M[ 6]+0xA8304613); FF<22>(B,C,D,A,m_M[ 7]+0xFD469501); + FF< 7>(A,B,C,D,m_M[ 8]+0x698098D8); FF<12>(D,A,B,C,m_M[ 9]+0x8B44F7AF); + FF<17>(C,D,A,B,m_M[10]+0xFFFF5BB1); FF<22>(B,C,D,A,m_M[11]+0x895CD7BE); + FF< 7>(A,B,C,D,m_M[12]+0x6B901122); FF<12>(D,A,B,C,m_M[13]+0xFD987193); + FF<17>(C,D,A,B,m_M[14]+0xA679438E); FF<22>(B,C,D,A,m_M[15]+0x49B40821); + + GG< 5>(A,B,C,D,m_M[ 1]+0xF61E2562); GG< 9>(D,A,B,C,m_M[ 6]+0xC040B340); + GG<14>(C,D,A,B,m_M[11]+0x265E5A51); GG<20>(B,C,D,A,m_M[ 0]+0xE9B6C7AA); + GG< 5>(A,B,C,D,m_M[ 5]+0xD62F105D); GG< 9>(D,A,B,C,m_M[10]+0x02441453); + GG<14>(C,D,A,B,m_M[15]+0xD8A1E681); GG<20>(B,C,D,A,m_M[ 4]+0xE7D3FBC8); + GG< 5>(A,B,C,D,m_M[ 9]+0x21E1CDE6); GG< 9>(D,A,B,C,m_M[14]+0xC33707D6); + GG<14>(C,D,A,B,m_M[ 3]+0xF4D50D87); GG<20>(B,C,D,A,m_M[ 8]+0x455A14ED); + GG< 5>(A,B,C,D,m_M[13]+0xA9E3E905); GG< 9>(D,A,B,C,m_M[ 2]+0xFCEFA3F8); + GG<14>(C,D,A,B,m_M[ 7]+0x676F02D9); GG<20>(B,C,D,A,m_M[12]+0x8D2A4C8A); + + HH< 4>(A,B,C,D,m_M[ 5]+0xFFFA3942); HH<11>(D,A,B,C,m_M[ 8]+0x8771F681); + HH<16>(C,D,A,B,m_M[11]+0x6D9D6122); HH<23>(B,C,D,A,m_M[14]+0xFDE5380C); + HH< 4>(A,B,C,D,m_M[ 1]+0xA4BEEA44); HH<11>(D,A,B,C,m_M[ 4]+0x4BDECFA9); + HH<16>(C,D,A,B,m_M[ 7]+0xF6BB4B60); HH<23>(B,C,D,A,m_M[10]+0xBEBFBC70); + HH< 4>(A,B,C,D,m_M[13]+0x289B7EC6); HH<11>(D,A,B,C,m_M[ 0]+0xEAA127FA); + HH<16>(C,D,A,B,m_M[ 3]+0xD4EF3085); HH<23>(B,C,D,A,m_M[ 6]+0x04881D05); + HH< 4>(A,B,C,D,m_M[ 9]+0xD9D4D039); HH<11>(D,A,B,C,m_M[12]+0xE6DB99E5); + HH<16>(C,D,A,B,m_M[15]+0x1FA27CF8); HH<23>(B,C,D,A,m_M[ 2]+0xC4AC5665); + + II< 6>(A,B,C,D,m_M[ 0]+0xF4292244); II<10>(D,A,B,C,m_M[ 7]+0x432AFF97); + II<15>(C,D,A,B,m_M[14]+0xAB9423A7); II<21>(B,C,D,A,m_M[ 5]+0xFC93A039); + II< 6>(A,B,C,D,m_M[12]+0x655B59C3); II<10>(D,A,B,C,m_M[ 3]+0x8F0CCC92); + II<15>(C,D,A,B,m_M[10]+0xFFEFF47D); II<21>(B,C,D,A,m_M[ 1]+0x85845DD1); + II< 6>(A,B,C,D,m_M[ 8]+0x6FA87E4F); II<10>(D,A,B,C,m_M[15]+0xFE2CE6E0); + II<15>(C,D,A,B,m_M[ 6]+0xA3014314); II<21>(B,C,D,A,m_M[13]+0x4E0811A1); + II< 6>(A,B,C,D,m_M[ 4]+0xF7537E82); II<10>(D,A,B,C,m_M[11]+0xBD3AF235); + II<15>(C,D,A,B,m_M[ 2]+0x2AD7D2BB); II<21>(B,C,D,A,m_M[ 9]+0xEB86D391); A = (m_digest[0] += A); B = (m_digest[1] += B); diff --git a/src/lib/hash/rmd160/rmd160.cpp b/src/lib/hash/rmd160/rmd160.cpp index 95f96c281..4c84ff422 100644 --- a/src/lib/hash/rmd160/rmd160.cpp +++ b/src/lib/hash/rmd160/rmd160.cpp @@ -19,56 +19,61 @@ namespace { /* * RIPEMD-160 F1 Function */ +template<size_t S> inline void F1(uint32_t& A, uint32_t B, uint32_t& C, uint32_t D, uint32_t E, - uint32_t msg, uint32_t shift) + uint32_t M) { - A += (B ^ C ^ D) + msg; - A = rotate_left(A, shift) + E; - C = rotate_left(C, 10); + A += (B ^ C ^ D) + M; + A = rotl<S>(A) + E; + C = rotl<10>(C); } /* * RIPEMD-160 F2 Function */ +template<size_t S> inline void F2(uint32_t& A, uint32_t B, uint32_t& C, uint32_t D, uint32_t E, - uint32_t msg, uint32_t shift, uint32_t magic) + uint32_t M) { - A += (D ^ (B & (C ^ D))) + msg + magic; - A = rotate_left(A, shift) + E; - C = rotate_left(C, 10); + A += (D ^ (B & (C ^ D))) + M; + A = rotl<S>(A) + E; + C = rotl<10>(C); } /* * RIPEMD-160 F3 Function */ +template<size_t S> inline void F3(uint32_t& A, uint32_t B, uint32_t& C, uint32_t D, uint32_t E, - uint32_t msg, uint32_t shift, uint32_t magic) + uint32_t M) { - A += (D ^ (B | ~C)) + msg + magic; - A = rotate_left(A, shift) + E; - C = rotate_left(C, 10); + A += (D ^ (B | ~C)) + M; + A = rotl<S>(A) + E; + C = rotl<10>(C); } /* * RIPEMD-160 F4 Function */ +template<size_t S> inline void F4(uint32_t& A, uint32_t B, uint32_t& C, uint32_t D, uint32_t E, - uint32_t msg, uint32_t shift, uint32_t magic) + uint32_t M) { - A += (C ^ (D & (B ^ C))) + msg + magic; - A = rotate_left(A, shift) + E; - C = rotate_left(C, 10); + A += (C ^ (D & (B ^ C))) + M; + A = rotl<S>(A) + E; + C = rotl<10>(C); } /* * RIPEMD-160 F5 Function */ +template<size_t S> inline void F5(uint32_t& A, uint32_t B, uint32_t& C, uint32_t D, uint32_t E, - uint32_t msg, uint32_t shift, uint32_t magic) + uint32_t M) { - A += (B ^ (C | ~D)) + msg + magic; - A = rotate_left(A, shift) + E; - C = rotate_left(C, 10); + A += (B ^ (C | ~D)) + M; + A = rotl<S>(A) + E; + C = rotl<10>(C); } } @@ -79,102 +84,104 @@ inline void F5(uint32_t& A, uint32_t B, uint32_t& C, uint32_t D, uint32_t E, void RIPEMD_160::compress_n(const uint8_t input[], size_t blocks) { const uint32_t MAGIC2 = 0x5A827999, MAGIC3 = 0x6ED9EBA1, - MAGIC4 = 0x8F1BBCDC, MAGIC5 = 0xA953FD4E, - MAGIC6 = 0x50A28BE6, MAGIC7 = 0x5C4DD124, - MAGIC8 = 0x6D703EF3, MAGIC9 = 0x7A6D76E9; + MAGIC4 = 0x8F1BBCDC, MAGIC5 = 0xA953FD4E, + MAGIC6 = 0x50A28BE6, MAGIC7 = 0x5C4DD124, + MAGIC8 = 0x6D703EF3, MAGIC9 = 0x7A6D76E9; for(size_t i = 0; i != blocks; ++i) { load_le(m_M.data(), input, m_M.size()); - uint32_t A1 = m_digest[0], A2 = A1, B1 = m_digest[1], B2 = B1, - C1 = m_digest[2], C2 = C1, D1 = m_digest[3], D2 = D1, - E1 = m_digest[4], E2 = E1; - - F1(A1,B1,C1,D1,E1,m_M[ 0],11 ); F5(A2,B2,C2,D2,E2,m_M[ 5], 8,MAGIC6); - F1(E1,A1,B1,C1,D1,m_M[ 1],14 ); F5(E2,A2,B2,C2,D2,m_M[14], 9,MAGIC6); - F1(D1,E1,A1,B1,C1,m_M[ 2],15 ); F5(D2,E2,A2,B2,C2,m_M[ 7], 9,MAGIC6); - F1(C1,D1,E1,A1,B1,m_M[ 3],12 ); F5(C2,D2,E2,A2,B2,m_M[ 0],11,MAGIC6); - F1(B1,C1,D1,E1,A1,m_M[ 4], 5 ); F5(B2,C2,D2,E2,A2,m_M[ 9],13,MAGIC6); - F1(A1,B1,C1,D1,E1,m_M[ 5], 8 ); F5(A2,B2,C2,D2,E2,m_M[ 2],15,MAGIC6); - F1(E1,A1,B1,C1,D1,m_M[ 6], 7 ); F5(E2,A2,B2,C2,D2,m_M[11],15,MAGIC6); - F1(D1,E1,A1,B1,C1,m_M[ 7], 9 ); F5(D2,E2,A2,B2,C2,m_M[ 4], 5,MAGIC6); - F1(C1,D1,E1,A1,B1,m_M[ 8],11 ); F5(C2,D2,E2,A2,B2,m_M[13], 7,MAGIC6); - F1(B1,C1,D1,E1,A1,m_M[ 9],13 ); F5(B2,C2,D2,E2,A2,m_M[ 6], 7,MAGIC6); - F1(A1,B1,C1,D1,E1,m_M[10],14 ); F5(A2,B2,C2,D2,E2,m_M[15], 8,MAGIC6); - F1(E1,A1,B1,C1,D1,m_M[11],15 ); F5(E2,A2,B2,C2,D2,m_M[ 8],11,MAGIC6); - F1(D1,E1,A1,B1,C1,m_M[12], 6 ); F5(D2,E2,A2,B2,C2,m_M[ 1],14,MAGIC6); - F1(C1,D1,E1,A1,B1,m_M[13], 7 ); F5(C2,D2,E2,A2,B2,m_M[10],14,MAGIC6); - F1(B1,C1,D1,E1,A1,m_M[14], 9 ); F5(B2,C2,D2,E2,A2,m_M[ 3],12,MAGIC6); - F1(A1,B1,C1,D1,E1,m_M[15], 8 ); F5(A2,B2,C2,D2,E2,m_M[12], 6,MAGIC6); - - F2(E1,A1,B1,C1,D1,m_M[ 7], 7,MAGIC2); F4(E2,A2,B2,C2,D2,m_M[ 6], 9,MAGIC7); - F2(D1,E1,A1,B1,C1,m_M[ 4], 6,MAGIC2); F4(D2,E2,A2,B2,C2,m_M[11],13,MAGIC7); - F2(C1,D1,E1,A1,B1,m_M[13], 8,MAGIC2); F4(C2,D2,E2,A2,B2,m_M[ 3],15,MAGIC7); - F2(B1,C1,D1,E1,A1,m_M[ 1],13,MAGIC2); F4(B2,C2,D2,E2,A2,m_M[ 7], 7,MAGIC7); - F2(A1,B1,C1,D1,E1,m_M[10],11,MAGIC2); F4(A2,B2,C2,D2,E2,m_M[ 0],12,MAGIC7); - F2(E1,A1,B1,C1,D1,m_M[ 6], 9,MAGIC2); F4(E2,A2,B2,C2,D2,m_M[13], 8,MAGIC7); - F2(D1,E1,A1,B1,C1,m_M[15], 7,MAGIC2); F4(D2,E2,A2,B2,C2,m_M[ 5], 9,MAGIC7); - F2(C1,D1,E1,A1,B1,m_M[ 3],15,MAGIC2); F4(C2,D2,E2,A2,B2,m_M[10],11,MAGIC7); - F2(B1,C1,D1,E1,A1,m_M[12], 7,MAGIC2); F4(B2,C2,D2,E2,A2,m_M[14], 7,MAGIC7); - F2(A1,B1,C1,D1,E1,m_M[ 0],12,MAGIC2); F4(A2,B2,C2,D2,E2,m_M[15], 7,MAGIC7); - F2(E1,A1,B1,C1,D1,m_M[ 9],15,MAGIC2); F4(E2,A2,B2,C2,D2,m_M[ 8],12,MAGIC7); - F2(D1,E1,A1,B1,C1,m_M[ 5], 9,MAGIC2); F4(D2,E2,A2,B2,C2,m_M[12], 7,MAGIC7); - F2(C1,D1,E1,A1,B1,m_M[ 2],11,MAGIC2); F4(C2,D2,E2,A2,B2,m_M[ 4], 6,MAGIC7); - F2(B1,C1,D1,E1,A1,m_M[14], 7,MAGIC2); F4(B2,C2,D2,E2,A2,m_M[ 9],15,MAGIC7); - F2(A1,B1,C1,D1,E1,m_M[11],13,MAGIC2); F4(A2,B2,C2,D2,E2,m_M[ 1],13,MAGIC7); - F2(E1,A1,B1,C1,D1,m_M[ 8],12,MAGIC2); F4(E2,A2,B2,C2,D2,m_M[ 2],11,MAGIC7); - - F3(D1,E1,A1,B1,C1,m_M[ 3],11,MAGIC3); F3(D2,E2,A2,B2,C2,m_M[15], 9,MAGIC8); - F3(C1,D1,E1,A1,B1,m_M[10],13,MAGIC3); F3(C2,D2,E2,A2,B2,m_M[ 5], 7,MAGIC8); - F3(B1,C1,D1,E1,A1,m_M[14], 6,MAGIC3); F3(B2,C2,D2,E2,A2,m_M[ 1],15,MAGIC8); - F3(A1,B1,C1,D1,E1,m_M[ 4], 7,MAGIC3); F3(A2,B2,C2,D2,E2,m_M[ 3],11,MAGIC8); - F3(E1,A1,B1,C1,D1,m_M[ 9],14,MAGIC3); F3(E2,A2,B2,C2,D2,m_M[ 7], 8,MAGIC8); - F3(D1,E1,A1,B1,C1,m_M[15], 9,MAGIC3); F3(D2,E2,A2,B2,C2,m_M[14], 6,MAGIC8); - F3(C1,D1,E1,A1,B1,m_M[ 8],13,MAGIC3); F3(C2,D2,E2,A2,B2,m_M[ 6], 6,MAGIC8); - F3(B1,C1,D1,E1,A1,m_M[ 1],15,MAGIC3); F3(B2,C2,D2,E2,A2,m_M[ 9],14,MAGIC8); - F3(A1,B1,C1,D1,E1,m_M[ 2],14,MAGIC3); F3(A2,B2,C2,D2,E2,m_M[11],12,MAGIC8); - F3(E1,A1,B1,C1,D1,m_M[ 7], 8,MAGIC3); F3(E2,A2,B2,C2,D2,m_M[ 8],13,MAGIC8); - F3(D1,E1,A1,B1,C1,m_M[ 0],13,MAGIC3); F3(D2,E2,A2,B2,C2,m_M[12], 5,MAGIC8); - F3(C1,D1,E1,A1,B1,m_M[ 6], 6,MAGIC3); F3(C2,D2,E2,A2,B2,m_M[ 2],14,MAGIC8); - F3(B1,C1,D1,E1,A1,m_M[13], 5,MAGIC3); F3(B2,C2,D2,E2,A2,m_M[10],13,MAGIC8); - F3(A1,B1,C1,D1,E1,m_M[11],12,MAGIC3); F3(A2,B2,C2,D2,E2,m_M[ 0],13,MAGIC8); - F3(E1,A1,B1,C1,D1,m_M[ 5], 7,MAGIC3); F3(E2,A2,B2,C2,D2,m_M[ 4], 7,MAGIC8); - F3(D1,E1,A1,B1,C1,m_M[12], 5,MAGIC3); F3(D2,E2,A2,B2,C2,m_M[13], 5,MAGIC8); - - F4(C1,D1,E1,A1,B1,m_M[ 1],11,MAGIC4); F2(C2,D2,E2,A2,B2,m_M[ 8],15,MAGIC9); - F4(B1,C1,D1,E1,A1,m_M[ 9],12,MAGIC4); F2(B2,C2,D2,E2,A2,m_M[ 6], 5,MAGIC9); - F4(A1,B1,C1,D1,E1,m_M[11],14,MAGIC4); F2(A2,B2,C2,D2,E2,m_M[ 4], 8,MAGIC9); - F4(E1,A1,B1,C1,D1,m_M[10],15,MAGIC4); F2(E2,A2,B2,C2,D2,m_M[ 1],11,MAGIC9); - F4(D1,E1,A1,B1,C1,m_M[ 0],14,MAGIC4); F2(D2,E2,A2,B2,C2,m_M[ 3],14,MAGIC9); - F4(C1,D1,E1,A1,B1,m_M[ 8],15,MAGIC4); F2(C2,D2,E2,A2,B2,m_M[11],14,MAGIC9); - F4(B1,C1,D1,E1,A1,m_M[12], 9,MAGIC4); F2(B2,C2,D2,E2,A2,m_M[15], 6,MAGIC9); - F4(A1,B1,C1,D1,E1,m_M[ 4], 8,MAGIC4); F2(A2,B2,C2,D2,E2,m_M[ 0],14,MAGIC9); - F4(E1,A1,B1,C1,D1,m_M[13], 9,MAGIC4); F2(E2,A2,B2,C2,D2,m_M[ 5], 6,MAGIC9); - F4(D1,E1,A1,B1,C1,m_M[ 3],14,MAGIC4); F2(D2,E2,A2,B2,C2,m_M[12], 9,MAGIC9); - F4(C1,D1,E1,A1,B1,m_M[ 7], 5,MAGIC4); F2(C2,D2,E2,A2,B2,m_M[ 2],12,MAGIC9); - F4(B1,C1,D1,E1,A1,m_M[15], 6,MAGIC4); F2(B2,C2,D2,E2,A2,m_M[13], 9,MAGIC9); - F4(A1,B1,C1,D1,E1,m_M[14], 8,MAGIC4); F2(A2,B2,C2,D2,E2,m_M[ 9],12,MAGIC9); - F4(E1,A1,B1,C1,D1,m_M[ 5], 6,MAGIC4); F2(E2,A2,B2,C2,D2,m_M[ 7], 5,MAGIC9); - F4(D1,E1,A1,B1,C1,m_M[ 6], 5,MAGIC4); F2(D2,E2,A2,B2,C2,m_M[10],15,MAGIC9); - F4(C1,D1,E1,A1,B1,m_M[ 2],12,MAGIC4); F2(C2,D2,E2,A2,B2,m_M[14], 8,MAGIC9); - - F5(B1,C1,D1,E1,A1,m_M[ 4], 9,MAGIC5); F1(B2,C2,D2,E2,A2,m_M[12], 8 ); - F5(A1,B1,C1,D1,E1,m_M[ 0],15,MAGIC5); F1(A2,B2,C2,D2,E2,m_M[15], 5 ); - F5(E1,A1,B1,C1,D1,m_M[ 5], 5,MAGIC5); F1(E2,A2,B2,C2,D2,m_M[10],12 ); - F5(D1,E1,A1,B1,C1,m_M[ 9],11,MAGIC5); F1(D2,E2,A2,B2,C2,m_M[ 4], 9 ); - F5(C1,D1,E1,A1,B1,m_M[ 7], 6,MAGIC5); F1(C2,D2,E2,A2,B2,m_M[ 1],12 ); - F5(B1,C1,D1,E1,A1,m_M[12], 8,MAGIC5); F1(B2,C2,D2,E2,A2,m_M[ 5], 5 ); - F5(A1,B1,C1,D1,E1,m_M[ 2],13,MAGIC5); F1(A2,B2,C2,D2,E2,m_M[ 8],14 ); - F5(E1,A1,B1,C1,D1,m_M[10],12,MAGIC5); F1(E2,A2,B2,C2,D2,m_M[ 7], 6 ); - F5(D1,E1,A1,B1,C1,m_M[14], 5,MAGIC5); F1(D2,E2,A2,B2,C2,m_M[ 6], 8 ); - F5(C1,D1,E1,A1,B1,m_M[ 1],12,MAGIC5); F1(C2,D2,E2,A2,B2,m_M[ 2],13 ); - F5(B1,C1,D1,E1,A1,m_M[ 3],13,MAGIC5); F1(B2,C2,D2,E2,A2,m_M[13], 6 ); - F5(A1,B1,C1,D1,E1,m_M[ 8],14,MAGIC5); F1(A2,B2,C2,D2,E2,m_M[14], 5 ); - F5(E1,A1,B1,C1,D1,m_M[11],11,MAGIC5); F1(E2,A2,B2,C2,D2,m_M[ 0],15 ); - F5(D1,E1,A1,B1,C1,m_M[ 6], 8,MAGIC5); F1(D2,E2,A2,B2,C2,m_M[ 3],13 ); - F5(C1,D1,E1,A1,B1,m_M[15], 5,MAGIC5); F1(C2,D2,E2,A2,B2,m_M[ 9],11 ); - F5(B1,C1,D1,E1,A1,m_M[13], 6,MAGIC5); F1(B2,C2,D2,E2,A2,m_M[11],11 ); + uint32_t A1 = m_digest[0], A2 = A1, + B1 = m_digest[1], B2 = B1, + C1 = m_digest[2], C2 = C1, + D1 = m_digest[3], D2 = D1, + E1 = m_digest[4], E2 = E1; + + F1<11>(A1,B1,C1,D1,E1,m_M[ 0] ); F5< 8>(A2,B2,C2,D2,E2,m_M[ 5]+MAGIC6); + F1<14>(E1,A1,B1,C1,D1,m_M[ 1] ); F5< 9>(E2,A2,B2,C2,D2,m_M[14]+MAGIC6); + F1<15>(D1,E1,A1,B1,C1,m_M[ 2] ); F5< 9>(D2,E2,A2,B2,C2,m_M[ 7]+MAGIC6); + F1<12>(C1,D1,E1,A1,B1,m_M[ 3] ); F5<11>(C2,D2,E2,A2,B2,m_M[ 0]+MAGIC6); + F1< 5>(B1,C1,D1,E1,A1,m_M[ 4] ); F5<13>(B2,C2,D2,E2,A2,m_M[ 9]+MAGIC6); + F1< 8>(A1,B1,C1,D1,E1,m_M[ 5] ); F5<15>(A2,B2,C2,D2,E2,m_M[ 2]+MAGIC6); + F1< 7>(E1,A1,B1,C1,D1,m_M[ 6] ); F5<15>(E2,A2,B2,C2,D2,m_M[11]+MAGIC6); + F1< 9>(D1,E1,A1,B1,C1,m_M[ 7] ); F5< 5>(D2,E2,A2,B2,C2,m_M[ 4]+MAGIC6); + F1<11>(C1,D1,E1,A1,B1,m_M[ 8] ); F5< 7>(C2,D2,E2,A2,B2,m_M[13]+MAGIC6); + F1<13>(B1,C1,D1,E1,A1,m_M[ 9] ); F5< 7>(B2,C2,D2,E2,A2,m_M[ 6]+MAGIC6); + F1<14>(A1,B1,C1,D1,E1,m_M[10] ); F5< 8>(A2,B2,C2,D2,E2,m_M[15]+MAGIC6); + F1<15>(E1,A1,B1,C1,D1,m_M[11] ); F5<11>(E2,A2,B2,C2,D2,m_M[ 8]+MAGIC6); + F1< 6>(D1,E1,A1,B1,C1,m_M[12] ); F5<14>(D2,E2,A2,B2,C2,m_M[ 1]+MAGIC6); + F1< 7>(C1,D1,E1,A1,B1,m_M[13] ); F5<14>(C2,D2,E2,A2,B2,m_M[10]+MAGIC6); + F1< 9>(B1,C1,D1,E1,A1,m_M[14] ); F5<12>(B2,C2,D2,E2,A2,m_M[ 3]+MAGIC6); + F1< 8>(A1,B1,C1,D1,E1,m_M[15] ); F5< 6>(A2,B2,C2,D2,E2,m_M[12]+MAGIC6); + + F2< 7>(E1,A1,B1,C1,D1,m_M[ 7]+MAGIC2); F4< 9>(E2,A2,B2,C2,D2,m_M[ 6]+MAGIC7); + F2< 6>(D1,E1,A1,B1,C1,m_M[ 4]+MAGIC2); F4<13>(D2,E2,A2,B2,C2,m_M[11]+MAGIC7); + F2< 8>(C1,D1,E1,A1,B1,m_M[13]+MAGIC2); F4<15>(C2,D2,E2,A2,B2,m_M[ 3]+MAGIC7); + F2<13>(B1,C1,D1,E1,A1,m_M[ 1]+MAGIC2); F4< 7>(B2,C2,D2,E2,A2,m_M[ 7]+MAGIC7); + F2<11>(A1,B1,C1,D1,E1,m_M[10]+MAGIC2); F4<12>(A2,B2,C2,D2,E2,m_M[ 0]+MAGIC7); + F2< 9>(E1,A1,B1,C1,D1,m_M[ 6]+MAGIC2); F4< 8>(E2,A2,B2,C2,D2,m_M[13]+MAGIC7); + F2< 7>(D1,E1,A1,B1,C1,m_M[15]+MAGIC2); F4< 9>(D2,E2,A2,B2,C2,m_M[ 5]+MAGIC7); + F2<15>(C1,D1,E1,A1,B1,m_M[ 3]+MAGIC2); F4<11>(C2,D2,E2,A2,B2,m_M[10]+MAGIC7); + F2< 7>(B1,C1,D1,E1,A1,m_M[12]+MAGIC2); F4< 7>(B2,C2,D2,E2,A2,m_M[14]+MAGIC7); + F2<12>(A1,B1,C1,D1,E1,m_M[ 0]+MAGIC2); F4< 7>(A2,B2,C2,D2,E2,m_M[15]+MAGIC7); + F2<15>(E1,A1,B1,C1,D1,m_M[ 9]+MAGIC2); F4<12>(E2,A2,B2,C2,D2,m_M[ 8]+MAGIC7); + F2< 9>(D1,E1,A1,B1,C1,m_M[ 5]+MAGIC2); F4< 7>(D2,E2,A2,B2,C2,m_M[12]+MAGIC7); + F2<11>(C1,D1,E1,A1,B1,m_M[ 2]+MAGIC2); F4< 6>(C2,D2,E2,A2,B2,m_M[ 4]+MAGIC7); + F2< 7>(B1,C1,D1,E1,A1,m_M[14]+MAGIC2); F4<15>(B2,C2,D2,E2,A2,m_M[ 9]+MAGIC7); + F2<13>(A1,B1,C1,D1,E1,m_M[11]+MAGIC2); F4<13>(A2,B2,C2,D2,E2,m_M[ 1]+MAGIC7); + F2<12>(E1,A1,B1,C1,D1,m_M[ 8]+MAGIC2); F4<11>(E2,A2,B2,C2,D2,m_M[ 2]+MAGIC7); + + F3<11>(D1,E1,A1,B1,C1,m_M[ 3]+MAGIC3); F3< 9>(D2,E2,A2,B2,C2,m_M[15]+MAGIC8); + F3<13>(C1,D1,E1,A1,B1,m_M[10]+MAGIC3); F3< 7>(C2,D2,E2,A2,B2,m_M[ 5]+MAGIC8); + F3< 6>(B1,C1,D1,E1,A1,m_M[14]+MAGIC3); F3<15>(B2,C2,D2,E2,A2,m_M[ 1]+MAGIC8); + F3< 7>(A1,B1,C1,D1,E1,m_M[ 4]+MAGIC3); F3<11>(A2,B2,C2,D2,E2,m_M[ 3]+MAGIC8); + F3<14>(E1,A1,B1,C1,D1,m_M[ 9]+MAGIC3); F3< 8>(E2,A2,B2,C2,D2,m_M[ 7]+MAGIC8); + F3< 9>(D1,E1,A1,B1,C1,m_M[15]+MAGIC3); F3< 6>(D2,E2,A2,B2,C2,m_M[14]+MAGIC8); + F3<13>(C1,D1,E1,A1,B1,m_M[ 8]+MAGIC3); F3< 6>(C2,D2,E2,A2,B2,m_M[ 6]+MAGIC8); + F3<15>(B1,C1,D1,E1,A1,m_M[ 1]+MAGIC3); F3<14>(B2,C2,D2,E2,A2,m_M[ 9]+MAGIC8); + F3<14>(A1,B1,C1,D1,E1,m_M[ 2]+MAGIC3); F3<12>(A2,B2,C2,D2,E2,m_M[11]+MAGIC8); + F3< 8>(E1,A1,B1,C1,D1,m_M[ 7]+MAGIC3); F3<13>(E2,A2,B2,C2,D2,m_M[ 8]+MAGIC8); + F3<13>(D1,E1,A1,B1,C1,m_M[ 0]+MAGIC3); F3< 5>(D2,E2,A2,B2,C2,m_M[12]+MAGIC8); + F3< 6>(C1,D1,E1,A1,B1,m_M[ 6]+MAGIC3); F3<14>(C2,D2,E2,A2,B2,m_M[ 2]+MAGIC8); + F3< 5>(B1,C1,D1,E1,A1,m_M[13]+MAGIC3); F3<13>(B2,C2,D2,E2,A2,m_M[10]+MAGIC8); + F3<12>(A1,B1,C1,D1,E1,m_M[11]+MAGIC3); F3<13>(A2,B2,C2,D2,E2,m_M[ 0]+MAGIC8); + F3< 7>(E1,A1,B1,C1,D1,m_M[ 5]+MAGIC3); F3< 7>(E2,A2,B2,C2,D2,m_M[ 4]+MAGIC8); + F3< 5>(D1,E1,A1,B1,C1,m_M[12]+MAGIC3); F3< 5>(D2,E2,A2,B2,C2,m_M[13]+MAGIC8); + + F4<11>(C1,D1,E1,A1,B1,m_M[ 1]+MAGIC4); F2<15>(C2,D2,E2,A2,B2,m_M[ 8]+MAGIC9); + F4<12>(B1,C1,D1,E1,A1,m_M[ 9]+MAGIC4); F2< 5>(B2,C2,D2,E2,A2,m_M[ 6]+MAGIC9); + F4<14>(A1,B1,C1,D1,E1,m_M[11]+MAGIC4); F2< 8>(A2,B2,C2,D2,E2,m_M[ 4]+MAGIC9); + F4<15>(E1,A1,B1,C1,D1,m_M[10]+MAGIC4); F2<11>(E2,A2,B2,C2,D2,m_M[ 1]+MAGIC9); + F4<14>(D1,E1,A1,B1,C1,m_M[ 0]+MAGIC4); F2<14>(D2,E2,A2,B2,C2,m_M[ 3]+MAGIC9); + F4<15>(C1,D1,E1,A1,B1,m_M[ 8]+MAGIC4); F2<14>(C2,D2,E2,A2,B2,m_M[11]+MAGIC9); + F4< 9>(B1,C1,D1,E1,A1,m_M[12]+MAGIC4); F2< 6>(B2,C2,D2,E2,A2,m_M[15]+MAGIC9); + F4< 8>(A1,B1,C1,D1,E1,m_M[ 4]+MAGIC4); F2<14>(A2,B2,C2,D2,E2,m_M[ 0]+MAGIC9); + F4< 9>(E1,A1,B1,C1,D1,m_M[13]+MAGIC4); F2< 6>(E2,A2,B2,C2,D2,m_M[ 5]+MAGIC9); + F4<14>(D1,E1,A1,B1,C1,m_M[ 3]+MAGIC4); F2< 9>(D2,E2,A2,B2,C2,m_M[12]+MAGIC9); + F4< 5>(C1,D1,E1,A1,B1,m_M[ 7]+MAGIC4); F2<12>(C2,D2,E2,A2,B2,m_M[ 2]+MAGIC9); + F4< 6>(B1,C1,D1,E1,A1,m_M[15]+MAGIC4); F2< 9>(B2,C2,D2,E2,A2,m_M[13]+MAGIC9); + F4< 8>(A1,B1,C1,D1,E1,m_M[14]+MAGIC4); F2<12>(A2,B2,C2,D2,E2,m_M[ 9]+MAGIC9); + F4< 6>(E1,A1,B1,C1,D1,m_M[ 5]+MAGIC4); F2< 5>(E2,A2,B2,C2,D2,m_M[ 7]+MAGIC9); + F4< 5>(D1,E1,A1,B1,C1,m_M[ 6]+MAGIC4); F2<15>(D2,E2,A2,B2,C2,m_M[10]+MAGIC9); + F4<12>(C1,D1,E1,A1,B1,m_M[ 2]+MAGIC4); F2< 8>(C2,D2,E2,A2,B2,m_M[14]+MAGIC9); + + F5< 9>(B1,C1,D1,E1,A1,m_M[ 4]+MAGIC5); F1< 8>(B2,C2,D2,E2,A2,m_M[12] ); + F5<15>(A1,B1,C1,D1,E1,m_M[ 0]+MAGIC5); F1< 5>(A2,B2,C2,D2,E2,m_M[15] ); + F5< 5>(E1,A1,B1,C1,D1,m_M[ 5]+MAGIC5); F1<12>(E2,A2,B2,C2,D2,m_M[10] ); + F5<11>(D1,E1,A1,B1,C1,m_M[ 9]+MAGIC5); F1< 9>(D2,E2,A2,B2,C2,m_M[ 4] ); + F5< 6>(C1,D1,E1,A1,B1,m_M[ 7]+MAGIC5); F1<12>(C2,D2,E2,A2,B2,m_M[ 1] ); + F5< 8>(B1,C1,D1,E1,A1,m_M[12]+MAGIC5); F1< 5>(B2,C2,D2,E2,A2,m_M[ 5] ); + F5<13>(A1,B1,C1,D1,E1,m_M[ 2]+MAGIC5); F1<14>(A2,B2,C2,D2,E2,m_M[ 8] ); + F5<12>(E1,A1,B1,C1,D1,m_M[10]+MAGIC5); F1< 6>(E2,A2,B2,C2,D2,m_M[ 7] ); + F5< 5>(D1,E1,A1,B1,C1,m_M[14]+MAGIC5); F1< 8>(D2,E2,A2,B2,C2,m_M[ 6] ); + F5<12>(C1,D1,E1,A1,B1,m_M[ 1]+MAGIC5); F1<13>(C2,D2,E2,A2,B2,m_M[ 2] ); + F5<13>(B1,C1,D1,E1,A1,m_M[ 3]+MAGIC5); F1< 6>(B2,C2,D2,E2,A2,m_M[13] ); + F5<14>(A1,B1,C1,D1,E1,m_M[ 8]+MAGIC5); F1< 5>(A2,B2,C2,D2,E2,m_M[14] ); + F5<11>(E1,A1,B1,C1,D1,m_M[11]+MAGIC5); F1<15>(E2,A2,B2,C2,D2,m_M[ 0] ); + F5< 8>(D1,E1,A1,B1,C1,m_M[ 6]+MAGIC5); F1<13>(D2,E2,A2,B2,C2,m_M[ 3] ); + F5< 5>(C1,D1,E1,A1,B1,m_M[15]+MAGIC5); F1<11>(C2,D2,E2,A2,B2,m_M[ 9] ); + F5< 6>(B1,C1,D1,E1,A1,m_M[13]+MAGIC5); F1<11>(B2,C2,D2,E2,A2,m_M[11] ); C1 = m_digest[1] + C1 + D2; m_digest[1] = m_digest[2] + D1 + E2; diff --git a/src/lib/hash/sha1/sha160.cpp b/src/lib/hash/sha1/sha160.cpp index fcca67341..8c12a4f04 100644 --- a/src/lib/hash/sha1/sha160.cpp +++ b/src/lib/hash/sha1/sha160.cpp @@ -24,8 +24,8 @@ namespace { */ inline void F1(uint32_t A, uint32_t& B, uint32_t C, uint32_t D, uint32_t& E, uint32_t msg) { - E += (D ^ (B & (C ^ D))) + msg + 0x5A827999 + rotate_left(A, 5); - B = rotate_left(B, 30); + E += (D ^ (B & (C ^ D))) + msg + 0x5A827999 + rotl<5>(A); + B = rotl<30>(B); } /* @@ -33,8 +33,8 @@ inline void F1(uint32_t A, uint32_t& B, uint32_t C, uint32_t D, uint32_t& E, uin */ inline void F2(uint32_t A, uint32_t& B, uint32_t C, uint32_t D, uint32_t& E, uint32_t msg) { - E += (B ^ C ^ D) + msg + 0x6ED9EBA1 + rotate_left(A, 5); - B = rotate_left(B, 30); + E += (B ^ C ^ D) + msg + 0x6ED9EBA1 + rotl<5>(A); + B = rotl<30>(B); } /* @@ -42,8 +42,8 @@ inline void F2(uint32_t A, uint32_t& B, uint32_t C, uint32_t D, uint32_t& E, uin */ inline void F3(uint32_t A, uint32_t& B, uint32_t C, uint32_t D, uint32_t& E, uint32_t msg) { - E += ((B & C) | ((B | C) & D)) + msg + 0x8F1BBCDC + rotate_left(A, 5); - B = rotate_left(B, 30); + E += ((B & C) | ((B | C) & D)) + msg + 0x8F1BBCDC + rotl<5>(A); + B = rotl<30>(B); } /* @@ -51,8 +51,8 @@ inline void F3(uint32_t A, uint32_t& B, uint32_t C, uint32_t D, uint32_t& E, uin */ inline void F4(uint32_t A, uint32_t& B, uint32_t C, uint32_t D, uint32_t& E, uint32_t msg) { - E += (B ^ C ^ D) + msg + 0xCA62C1D6 + rotate_left(A, 5); - B = rotate_left(B, 30); + E += (B ^ C ^ D) + msg + 0xCA62C1D6 + rotl<5>(A); + B = rotl<30>(B); } } @@ -99,14 +99,14 @@ void SHA_160::compress_n(const uint8_t input[], size_t blocks) for(size_t j = 16; j != 80; j += 8) { - m_W[j ] = rotate_left((m_W[j-3] ^ m_W[j-8] ^ m_W[j-14] ^ m_W[j-16]), 1); - m_W[j+1] = rotate_left((m_W[j-2] ^ m_W[j-7] ^ m_W[j-13] ^ m_W[j-15]), 1); - m_W[j+2] = rotate_left((m_W[j-1] ^ m_W[j-6] ^ m_W[j-12] ^ m_W[j-14]), 1); - m_W[j+3] = rotate_left((m_W[j ] ^ m_W[j-5] ^ m_W[j-11] ^ m_W[j-13]), 1); - m_W[j+4] = rotate_left((m_W[j+1] ^ m_W[j-4] ^ m_W[j-10] ^ m_W[j-12]), 1); - m_W[j+5] = rotate_left((m_W[j+2] ^ m_W[j-3] ^ m_W[j- 9] ^ m_W[j-11]), 1); - m_W[j+6] = rotate_left((m_W[j+3] ^ m_W[j-2] ^ m_W[j- 8] ^ m_W[j-10]), 1); - m_W[j+7] = rotate_left((m_W[j+4] ^ m_W[j-1] ^ m_W[j- 7] ^ m_W[j- 9]), 1); + m_W[j ] = rotl<1>(m_W[j-3] ^ m_W[j-8] ^ m_W[j-14] ^ m_W[j-16]); + m_W[j+1] = rotl<1>(m_W[j-2] ^ m_W[j-7] ^ m_W[j-13] ^ m_W[j-15]); + m_W[j+2] = rotl<1>(m_W[j-1] ^ m_W[j-6] ^ m_W[j-12] ^ m_W[j-14]); + m_W[j+3] = rotl<1>(m_W[j ] ^ m_W[j-5] ^ m_W[j-11] ^ m_W[j-13]); + m_W[j+4] = rotl<1>(m_W[j+1] ^ m_W[j-4] ^ m_W[j-10] ^ m_W[j-12]); + m_W[j+5] = rotl<1>(m_W[j+2] ^ m_W[j-3] ^ m_W[j- 9] ^ m_W[j-11]); + m_W[j+6] = rotl<1>(m_W[j+3] ^ m_W[j-2] ^ m_W[j- 8] ^ m_W[j-10]); + m_W[j+7] = rotl<1>(m_W[j+4] ^ m_W[j-1] ^ m_W[j- 7] ^ m_W[j- 9]); } F1(A, B, C, D, E, m_W[ 0]); F1(E, A, B, C, D, m_W[ 1]); diff --git a/src/lib/hash/sha1/sha1_sse2/sha1_sse2.cpp b/src/lib/hash/sha1/sha1_sse2/sha1_sse2.cpp index 8c7785051..0b7f8f837 100644 --- a/src/lib/hash/sha1/sha1_sse2/sha1_sse2.cpp +++ b/src/lib/hash/sha1/sha1_sse2/sha1_sse2.cpp @@ -113,8 +113,8 @@ W0 = W[t]..W[t+3] */ inline void F1(uint32_t A, uint32_t& B, uint32_t C, uint32_t D, uint32_t& E, uint32_t msg) { - E += (D ^ (B & (C ^ D))) + msg + rotate_left(A, 5); - B = rotate_left(B, 30); + E += (D ^ (B & (C ^ D))) + msg + rotl<5>(A); + B = rotl<30>(B); } /* @@ -122,8 +122,8 @@ inline void F1(uint32_t A, uint32_t& B, uint32_t C, uint32_t D, uint32_t& E, uin */ inline void F2(uint32_t A, uint32_t& B, uint32_t C, uint32_t D, uint32_t& E, uint32_t msg) { - E += (B ^ C ^ D) + msg + rotate_left(A, 5); - B = rotate_left(B, 30); + E += (B ^ C ^ D) + msg + rotl<5>(A); + B = rotl<30>(B); } /* @@ -131,8 +131,8 @@ inline void F2(uint32_t A, uint32_t& B, uint32_t C, uint32_t D, uint32_t& E, uin */ inline void F3(uint32_t A, uint32_t& B, uint32_t C, uint32_t D, uint32_t& E, uint32_t msg) { - E += ((B & C) | ((B | C) & D)) + msg + rotate_left(A, 5); - B = rotate_left(B, 30); + E += ((B & C) | ((B | C) & D)) + msg + rotl<5>(A); + B = rotl<30>(B); } /* @@ -140,8 +140,8 @@ inline void F3(uint32_t A, uint32_t& B, uint32_t C, uint32_t D, uint32_t& E, uin */ inline void F4(uint32_t A, uint32_t& B, uint32_t C, uint32_t D, uint32_t& E, uint32_t msg) { - E += (B ^ C ^ D) + msg + rotate_left(A, 5); - B = rotate_left(B, 30); + E += (B ^ C ^ D) + msg + rotl<5>(A); + B = rotl<30>(B); } } diff --git a/src/lib/hash/sha2_32/sha2_32.cpp b/src/lib/hash/sha2_32/sha2_32.cpp index 281e6ed2b..0710747d0 100644 --- a/src/lib/hash/sha2_32/sha2_32.cpp +++ b/src/lib/hash/sha2_32/sha2_32.cpp @@ -28,10 +28,10 @@ std::unique_ptr<HashFunction> SHA_256::copy_state() const * even though it is much faster if inlined. */ #define SHA2_32_F(A, B, C, D, E, F, G, H, M1, M2, M3, M4, magic) do { \ - uint32_t A_rho = rotate_right(A, 2) ^ rotate_right(A, 13) ^ rotate_right(A, 22); \ - uint32_t E_rho = rotate_right(E, 6) ^ rotate_right(E, 11) ^ rotate_right(E, 25); \ - uint32_t M2_sigma = rotate_right(M2, 17) ^ rotate_right(M2, 19) ^ (M2 >> 10); \ - uint32_t M4_sigma = rotate_right(M4, 7) ^ rotate_right(M4, 18) ^ (M4 >> 3); \ + uint32_t A_rho = rotr<2>(A) ^ rotr<13>(A) ^ rotr<22>(A); \ + uint32_t E_rho = rotr<6>(E) ^ rotr<11>(E) ^ rotr<25>(E); \ + uint32_t M2_sigma = rotr<17>(M2) ^ rotr<19>(M2) ^ (M2 >> 10); \ + uint32_t M4_sigma = rotr<7>(M4) ^ rotr<18>(M4) ^ (M4 >> 3); \ H += magic + E_rho + ((E & F) ^ (~E & G)) + M1; \ D += H; \ H += A_rho + ((A & B) | ((A | B) & C)); \ diff --git a/src/lib/hash/sha2_64/sha2_64.cpp b/src/lib/hash/sha2_64/sha2_64.cpp index 8e01b6b4d..45992e996 100644 --- a/src/lib/hash/sha2_64/sha2_64.cpp +++ b/src/lib/hash/sha2_64/sha2_64.cpp @@ -26,44 +26,29 @@ std::unique_ptr<HashFunction> SHA_512_256::copy_state() const namespace { -namespace SHA2_64 { - -/* -* SHA-{384,512} Rho Function -*/ -inline uint64_t rho(uint64_t X, uint32_t rot1, uint32_t rot2, uint32_t rot3) - { - return (rotate_right(X, rot1) ^ rotate_right(X, rot2) ^ - rotate_right(X, rot3)); - } - -/* -* SHA-{384,512} Sigma Function -*/ -inline uint64_t sigma(uint64_t X, uint32_t rot1, uint32_t rot2, uint32_t shift) - { - return (rotate_right(X, rot1) ^ rotate_right(X, rot2) ^ (X >> shift)); - } - /* * SHA-512 F1 Function * * Use a macro as many compilers won't inline a function this big, * even though it is much faster if inlined. */ -#define SHA2_64_F(A, B, C, D, E, F, G, H, M1, M2, M3, M4, magic) \ - do { \ - H += magic + rho(E, 14, 18, 41) + ((E & F) ^ (~E & G)) + M1; \ - D += H; \ - H += rho(A, 28, 34, 39) + ((A & B) | ((A | B) & C)); \ - M1 += sigma(M2, 19, 61, 6) + M3 + sigma(M4, 1, 8, 7); \ +#define SHA2_64_F(A, B, C, D, E, F, G, H, M1, M2, M3, M4, magic) \ + do { \ + const uint64_t E_rho = rotr<14>(E) ^ rotr<18>(E) ^ rotr<41>(E); \ + const uint64_t A_rho = rotr<28>(A) ^ rotr<34>(A) ^ rotr<39>(A); \ + const uint64_t M2_sigma = rotr<19>(M2) ^ rotr<61>(M2) ^ (M2 >> 6); \ + const uint64_t M4_sigma = rotr<1>(M4) ^ rotr<8>(M4) ^ (M4 >> 7); \ + H += magic + E_rho + ((E & F) ^ (~E & G)) + M1; \ + D += H; \ + H += A_rho + ((A & B) | ((A | B) & C)); \ + M1 += M2_sigma + M3 + M4_sigma; \ } while(0); /* * SHA-{384,512} Compression Function */ -void compress(secure_vector<uint64_t>& digest, - const uint8_t input[], size_t blocks) +void SHA64_compress(secure_vector<uint64_t>& digest, + const uint8_t input[], size_t blocks) { uint64_t A = digest[0], B = digest[1], C = digest[2], D = digest[3], E = digest[4], F = digest[5], @@ -184,21 +169,19 @@ void compress(secure_vector<uint64_t>& digest, } -} - void SHA_512_256::compress_n(const uint8_t input[], size_t blocks) { - SHA2_64::compress(m_digest, input, blocks); + SHA64_compress(m_digest, input, blocks); } void SHA_384::compress_n(const uint8_t input[], size_t blocks) { - SHA2_64::compress(m_digest, input, blocks); + SHA64_compress(m_digest, input, blocks); } void SHA_512::compress_n(const uint8_t input[], size_t blocks) { - SHA2_64::compress(m_digest, input, blocks); + SHA64_compress(m_digest, input, blocks); } void SHA_512_256::copy_out(uint8_t output[]) diff --git a/src/lib/hash/sha3/sha3.cpp b/src/lib/hash/sha3/sha3.cpp index e829c3f70..1556e5498 100644 --- a/src/lib/hash/sha3/sha3.cpp +++ b/src/lib/hash/sha3/sha3.cpp @@ -37,37 +37,37 @@ void SHA_3::permute(uint64_t A[25]) const uint64_t C3 = A[3] ^ A[8] ^ A[13] ^ A[18] ^ A[23]; const uint64_t C4 = A[4] ^ A[9] ^ A[14] ^ A[19] ^ A[24]; - const uint64_t D0 = rotate_left(C0, 1) ^ C3; - const uint64_t D1 = rotate_left(C1, 1) ^ C4; - const uint64_t D2 = rotate_left(C2, 1) ^ C0; - const uint64_t D3 = rotate_left(C3, 1) ^ C1; - const uint64_t D4 = rotate_left(C4, 1) ^ C2; + const uint64_t D0 = rotl<1>(C0) ^ C3; + const uint64_t D1 = rotl<1>(C1) ^ C4; + const uint64_t D2 = rotl<1>(C2) ^ C0; + const uint64_t D3 = rotl<1>(C3) ^ C1; + const uint64_t D4 = rotl<1>(C4) ^ C2; const uint64_t B00 = A[ 0] ^ D1; - const uint64_t B10 = rotate_left(A[ 1] ^ D2, 1); - const uint64_t B20 = rotate_left(A[ 2] ^ D3, 62); - const uint64_t B05 = rotate_left(A[ 3] ^ D4, 28); - const uint64_t B15 = rotate_left(A[ 4] ^ D0, 27); - const uint64_t B16 = rotate_left(A[ 5] ^ D1, 36); - const uint64_t B01 = rotate_left(A[ 6] ^ D2, 44); - const uint64_t B11 = rotate_left(A[ 7] ^ D3, 6); - const uint64_t B21 = rotate_left(A[ 8] ^ D4, 55); - const uint64_t B06 = rotate_left(A[ 9] ^ D0, 20); - const uint64_t B07 = rotate_left(A[10] ^ D1, 3); - const uint64_t B17 = rotate_left(A[11] ^ D2, 10); - const uint64_t B02 = rotate_left(A[12] ^ D3, 43); - const uint64_t B12 = rotate_left(A[13] ^ D4, 25); - const uint64_t B22 = rotate_left(A[14] ^ D0, 39); - const uint64_t B23 = rotate_left(A[15] ^ D1, 41); - const uint64_t B08 = rotate_left(A[16] ^ D2, 45); - const uint64_t B18 = rotate_left(A[17] ^ D3, 15); - const uint64_t B03 = rotate_left(A[18] ^ D4, 21); - const uint64_t B13 = rotate_left(A[19] ^ D0, 8); - const uint64_t B14 = rotate_left(A[20] ^ D1, 18); - const uint64_t B24 = rotate_left(A[21] ^ D2, 2); - const uint64_t B09 = rotate_left(A[22] ^ D3, 61); - const uint64_t B19 = rotate_left(A[23] ^ D4, 56); - const uint64_t B04 = rotate_left(A[24] ^ D0, 14); + const uint64_t B10 = rotl<1>(A[ 1] ^ D2); + const uint64_t B20 = rotl<62>(A[ 2] ^ D3); + const uint64_t B05 = rotl<28>(A[ 3] ^ D4); + const uint64_t B15 = rotl<27>(A[ 4] ^ D0); + const uint64_t B16 = rotl<36>(A[ 5] ^ D1); + const uint64_t B01 = rotl<44>(A[ 6] ^ D2); + const uint64_t B11 = rotl<6>(A[ 7] ^ D3); + const uint64_t B21 = rotl<55>(A[ 8] ^ D4); + const uint64_t B06 = rotl<20>(A[ 9] ^ D0); + const uint64_t B07 = rotl<3>(A[10] ^ D1); + const uint64_t B17 = rotl<10>(A[11] ^ D2); + const uint64_t B02 = rotl<43>(A[12] ^ D3); + const uint64_t B12 = rotl<25>(A[13] ^ D4); + const uint64_t B22 = rotl<39>(A[14] ^ D0); + const uint64_t B23 = rotl<41>(A[15] ^ D1); + const uint64_t B08 = rotl<45>(A[16] ^ D2); + const uint64_t B18 = rotl<15>(A[17] ^ D3); + const uint64_t B03 = rotl<21>(A[18] ^ D4); + const uint64_t B13 = rotl<8>(A[19] ^ D0); + const uint64_t B14 = rotl<18>(A[20] ^ D1); + const uint64_t B24 = rotl<2>(A[21] ^ D2); + const uint64_t B09 = rotl<61>(A[22] ^ D3); + const uint64_t B19 = rotl<56>(A[23] ^ D4); + const uint64_t B04 = rotl<14>(A[24] ^ D0); A[ 0] = B00 ^ (~B01 & B02); A[ 1] = B01 ^ (~B02 & B03); diff --git a/src/lib/hash/sm3/sm3.cpp b/src/lib/hash/sm3/sm3.cpp index aeb8f2e47..c3220d243 100644 --- a/src/lib/hash/sm3/sm3.cpp +++ b/src/lib/hash/sm3/sm3.cpp @@ -23,12 +23,12 @@ const uint32_t SM3_IV[] = { inline uint32_t P0(uint32_t X) { - return X ^ rotate_left(X, 9) ^ rotate_left(X, 17); + return X ^ rotl<9>(X) ^ rotl<17>(X); } inline uint32_t P1(uint32_t X) { - return X ^ rotate_left(X, 15) ^ rotate_left(X, 23); + return X ^ rotl<15>(X) ^ rotl<23>(X); } inline uint32_t FF1(uint32_t X, uint32_t Y, uint32_t Z) @@ -47,14 +47,14 @@ inline void R1(uint32_t A, uint32_t& B, uint32_t C, uint32_t& D, uint32_t E, uint32_t& F, uint32_t G, uint32_t& H, uint32_t TJ, uint32_t Wi, uint32_t Wj) { - const uint32_t A12 = rotate_left(A, 12); - const uint32_t SS1 = rotate_left(A12 + E + TJ, 7); + const uint32_t A12 = rotl<12>(A); + const uint32_t SS1 = rotl<7>(A12 + E + TJ); const uint32_t TT1 = (A ^ B ^ C) + D + (SS1 ^ A12) + Wj; const uint32_t TT2 = (E ^ F ^ G) + H + SS1 + Wi; - B = rotate_left(B, 9); + B = rotl<9>(B); D = TT1; - F = rotate_left(F, 19); + F = rotl<19>(F); H = P0(TT2); } @@ -62,14 +62,14 @@ inline void R2(uint32_t A, uint32_t& B, uint32_t C, uint32_t& D, uint32_t E, uint32_t& F, uint32_t G, uint32_t& H, uint32_t TJ, uint32_t Wi, uint32_t Wj) { - const uint32_t A12 = rotate_left(A, 12); - const uint32_t SS1 = rotate_left(A12 + E + TJ, 7); + const uint32_t A12 = rotl<12>(A); + const uint32_t SS1 = rotl<7>(A12 + E + TJ); const uint32_t TT1 = FF1(A, B, C) + D + (SS1 ^ A12) + Wj; const uint32_t TT2 = GG1(E, F, G) + H + SS1 + Wi; - B = rotate_left(B, 9); + B = rotl<9>(B); D = TT1; - F = rotate_left(F, 19); + F = rotl<19>(F); H = P0(TT2); } @@ -105,58 +105,58 @@ void SM3::compress_n(const uint8_t input[], size_t blocks) W[15] = load_be<uint32_t>(input, 15); // Message Extension (b) - W[16] = P1(W[ 0] ^ W[ 7] ^ rotate_left(W[13], 15)) ^ rotate_left(W[ 3], 7) ^ W[10]; - W[17] = P1(W[ 1] ^ W[ 8] ^ rotate_left(W[14], 15)) ^ rotate_left(W[ 4], 7) ^ W[11]; - W[18] = P1(W[ 2] ^ W[ 9] ^ rotate_left(W[15], 15)) ^ rotate_left(W[ 5], 7) ^ W[12]; - W[19] = P1(W[ 3] ^ W[10] ^ rotate_left(W[16], 15)) ^ rotate_left(W[ 6], 7) ^ W[13]; - W[20] = P1(W[ 4] ^ W[11] ^ rotate_left(W[17], 15)) ^ rotate_left(W[ 7], 7) ^ W[14]; - W[21] = P1(W[ 5] ^ W[12] ^ rotate_left(W[18], 15)) ^ rotate_left(W[ 8], 7) ^ W[15]; - W[22] = P1(W[ 6] ^ W[13] ^ rotate_left(W[19], 15)) ^ rotate_left(W[ 9], 7) ^ W[16]; - W[23] = P1(W[ 7] ^ W[14] ^ rotate_left(W[20], 15)) ^ rotate_left(W[10], 7) ^ W[17]; - W[24] = P1(W[ 8] ^ W[15] ^ rotate_left(W[21], 15)) ^ rotate_left(W[11], 7) ^ W[18]; - W[25] = P1(W[ 9] ^ W[16] ^ rotate_left(W[22], 15)) ^ rotate_left(W[12], 7) ^ W[19]; - W[26] = P1(W[10] ^ W[17] ^ rotate_left(W[23], 15)) ^ rotate_left(W[13], 7) ^ W[20]; - W[27] = P1(W[11] ^ W[18] ^ rotate_left(W[24], 15)) ^ rotate_left(W[14], 7) ^ W[21]; - W[28] = P1(W[12] ^ W[19] ^ rotate_left(W[25], 15)) ^ rotate_left(W[15], 7) ^ W[22]; - W[29] = P1(W[13] ^ W[20] ^ rotate_left(W[26], 15)) ^ rotate_left(W[16], 7) ^ W[23]; - W[30] = P1(W[14] ^ W[21] ^ rotate_left(W[27], 15)) ^ rotate_left(W[17], 7) ^ W[24]; - W[31] = P1(W[15] ^ W[22] ^ rotate_left(W[28], 15)) ^ rotate_left(W[18], 7) ^ W[25]; - W[32] = P1(W[16] ^ W[23] ^ rotate_left(W[29], 15)) ^ rotate_left(W[19], 7) ^ W[26]; - W[33] = P1(W[17] ^ W[24] ^ rotate_left(W[30], 15)) ^ rotate_left(W[20], 7) ^ W[27]; - W[34] = P1(W[18] ^ W[25] ^ rotate_left(W[31], 15)) ^ rotate_left(W[21], 7) ^ W[28]; - W[35] = P1(W[19] ^ W[26] ^ rotate_left(W[32], 15)) ^ rotate_left(W[22], 7) ^ W[29]; - W[36] = P1(W[20] ^ W[27] ^ rotate_left(W[33], 15)) ^ rotate_left(W[23], 7) ^ W[30]; - W[37] = P1(W[21] ^ W[28] ^ rotate_left(W[34], 15)) ^ rotate_left(W[24], 7) ^ W[31]; - W[38] = P1(W[22] ^ W[29] ^ rotate_left(W[35], 15)) ^ rotate_left(W[25], 7) ^ W[32]; - W[39] = P1(W[23] ^ W[30] ^ rotate_left(W[36], 15)) ^ rotate_left(W[26], 7) ^ W[33]; - W[40] = P1(W[24] ^ W[31] ^ rotate_left(W[37], 15)) ^ rotate_left(W[27], 7) ^ W[34]; - W[41] = P1(W[25] ^ W[32] ^ rotate_left(W[38], 15)) ^ rotate_left(W[28], 7) ^ W[35]; - W[42] = P1(W[26] ^ W[33] ^ rotate_left(W[39], 15)) ^ rotate_left(W[29], 7) ^ W[36]; - W[43] = P1(W[27] ^ W[34] ^ rotate_left(W[40], 15)) ^ rotate_left(W[30], 7) ^ W[37]; - W[44] = P1(W[28] ^ W[35] ^ rotate_left(W[41], 15)) ^ rotate_left(W[31], 7) ^ W[38]; - W[45] = P1(W[29] ^ W[36] ^ rotate_left(W[42], 15)) ^ rotate_left(W[32], 7) ^ W[39]; - W[46] = P1(W[30] ^ W[37] ^ rotate_left(W[43], 15)) ^ rotate_left(W[33], 7) ^ W[40]; - W[47] = P1(W[31] ^ W[38] ^ rotate_left(W[44], 15)) ^ rotate_left(W[34], 7) ^ W[41]; - W[48] = P1(W[32] ^ W[39] ^ rotate_left(W[45], 15)) ^ rotate_left(W[35], 7) ^ W[42]; - W[49] = P1(W[33] ^ W[40] ^ rotate_left(W[46], 15)) ^ rotate_left(W[36], 7) ^ W[43]; - W[50] = P1(W[34] ^ W[41] ^ rotate_left(W[47], 15)) ^ rotate_left(W[37], 7) ^ W[44]; - W[51] = P1(W[35] ^ W[42] ^ rotate_left(W[48], 15)) ^ rotate_left(W[38], 7) ^ W[45]; - W[52] = P1(W[36] ^ W[43] ^ rotate_left(W[49], 15)) ^ rotate_left(W[39], 7) ^ W[46]; - W[53] = P1(W[37] ^ W[44] ^ rotate_left(W[50], 15)) ^ rotate_left(W[40], 7) ^ W[47]; - W[54] = P1(W[38] ^ W[45] ^ rotate_left(W[51], 15)) ^ rotate_left(W[41], 7) ^ W[48]; - W[55] = P1(W[39] ^ W[46] ^ rotate_left(W[52], 15)) ^ rotate_left(W[42], 7) ^ W[49]; - W[56] = P1(W[40] ^ W[47] ^ rotate_left(W[53], 15)) ^ rotate_left(W[43], 7) ^ W[50]; - W[57] = P1(W[41] ^ W[48] ^ rotate_left(W[54], 15)) ^ rotate_left(W[44], 7) ^ W[51]; - W[58] = P1(W[42] ^ W[49] ^ rotate_left(W[55], 15)) ^ rotate_left(W[45], 7) ^ W[52]; - W[59] = P1(W[43] ^ W[50] ^ rotate_left(W[56], 15)) ^ rotate_left(W[46], 7) ^ W[53]; - W[60] = P1(W[44] ^ W[51] ^ rotate_left(W[57], 15)) ^ rotate_left(W[47], 7) ^ W[54]; - W[61] = P1(W[45] ^ W[52] ^ rotate_left(W[58], 15)) ^ rotate_left(W[48], 7) ^ W[55]; - W[62] = P1(W[46] ^ W[53] ^ rotate_left(W[59], 15)) ^ rotate_left(W[49], 7) ^ W[56]; - W[63] = P1(W[47] ^ W[54] ^ rotate_left(W[60], 15)) ^ rotate_left(W[50], 7) ^ W[57]; - W[64] = P1(W[48] ^ W[55] ^ rotate_left(W[61], 15)) ^ rotate_left(W[51], 7) ^ W[58]; - W[65] = P1(W[49] ^ W[56] ^ rotate_left(W[62], 15)) ^ rotate_left(W[52], 7) ^ W[59]; - W[66] = P1(W[50] ^ W[57] ^ rotate_left(W[63], 15)) ^ rotate_left(W[53], 7) ^ W[60]; - W[67] = P1(W[51] ^ W[58] ^ rotate_left(W[64], 15)) ^ rotate_left(W[54], 7) ^ W[61]; + W[16] = P1(W[ 0] ^ W[ 7] ^ rotl<15>(W[13])) ^ rotl<7>(W[ 3]) ^ W[10]; + W[17] = P1(W[ 1] ^ W[ 8] ^ rotl<15>(W[14])) ^ rotl<7>(W[ 4]) ^ W[11]; + W[18] = P1(W[ 2] ^ W[ 9] ^ rotl<15>(W[15])) ^ rotl<7>(W[ 5]) ^ W[12]; + W[19] = P1(W[ 3] ^ W[10] ^ rotl<15>(W[16])) ^ rotl<7>(W[ 6]) ^ W[13]; + W[20] = P1(W[ 4] ^ W[11] ^ rotl<15>(W[17])) ^ rotl<7>(W[ 7]) ^ W[14]; + W[21] = P1(W[ 5] ^ W[12] ^ rotl<15>(W[18])) ^ rotl<7>(W[ 8]) ^ W[15]; + W[22] = P1(W[ 6] ^ W[13] ^ rotl<15>(W[19])) ^ rotl<7>(W[ 9]) ^ W[16]; + W[23] = P1(W[ 7] ^ W[14] ^ rotl<15>(W[20])) ^ rotl<7>(W[10]) ^ W[17]; + W[24] = P1(W[ 8] ^ W[15] ^ rotl<15>(W[21])) ^ rotl<7>(W[11]) ^ W[18]; + W[25] = P1(W[ 9] ^ W[16] ^ rotl<15>(W[22])) ^ rotl<7>(W[12]) ^ W[19]; + W[26] = P1(W[10] ^ W[17] ^ rotl<15>(W[23])) ^ rotl<7>(W[13]) ^ W[20]; + W[27] = P1(W[11] ^ W[18] ^ rotl<15>(W[24])) ^ rotl<7>(W[14]) ^ W[21]; + W[28] = P1(W[12] ^ W[19] ^ rotl<15>(W[25])) ^ rotl<7>(W[15]) ^ W[22]; + W[29] = P1(W[13] ^ W[20] ^ rotl<15>(W[26])) ^ rotl<7>(W[16]) ^ W[23]; + W[30] = P1(W[14] ^ W[21] ^ rotl<15>(W[27])) ^ rotl<7>(W[17]) ^ W[24]; + W[31] = P1(W[15] ^ W[22] ^ rotl<15>(W[28])) ^ rotl<7>(W[18]) ^ W[25]; + W[32] = P1(W[16] ^ W[23] ^ rotl<15>(W[29])) ^ rotl<7>(W[19]) ^ W[26]; + W[33] = P1(W[17] ^ W[24] ^ rotl<15>(W[30])) ^ rotl<7>(W[20]) ^ W[27]; + W[34] = P1(W[18] ^ W[25] ^ rotl<15>(W[31])) ^ rotl<7>(W[21]) ^ W[28]; + W[35] = P1(W[19] ^ W[26] ^ rotl<15>(W[32])) ^ rotl<7>(W[22]) ^ W[29]; + W[36] = P1(W[20] ^ W[27] ^ rotl<15>(W[33])) ^ rotl<7>(W[23]) ^ W[30]; + W[37] = P1(W[21] ^ W[28] ^ rotl<15>(W[34])) ^ rotl<7>(W[24]) ^ W[31]; + W[38] = P1(W[22] ^ W[29] ^ rotl<15>(W[35])) ^ rotl<7>(W[25]) ^ W[32]; + W[39] = P1(W[23] ^ W[30] ^ rotl<15>(W[36])) ^ rotl<7>(W[26]) ^ W[33]; + W[40] = P1(W[24] ^ W[31] ^ rotl<15>(W[37])) ^ rotl<7>(W[27]) ^ W[34]; + W[41] = P1(W[25] ^ W[32] ^ rotl<15>(W[38])) ^ rotl<7>(W[28]) ^ W[35]; + W[42] = P1(W[26] ^ W[33] ^ rotl<15>(W[39])) ^ rotl<7>(W[29]) ^ W[36]; + W[43] = P1(W[27] ^ W[34] ^ rotl<15>(W[40])) ^ rotl<7>(W[30]) ^ W[37]; + W[44] = P1(W[28] ^ W[35] ^ rotl<15>(W[41])) ^ rotl<7>(W[31]) ^ W[38]; + W[45] = P1(W[29] ^ W[36] ^ rotl<15>(W[42])) ^ rotl<7>(W[32]) ^ W[39]; + W[46] = P1(W[30] ^ W[37] ^ rotl<15>(W[43])) ^ rotl<7>(W[33]) ^ W[40]; + W[47] = P1(W[31] ^ W[38] ^ rotl<15>(W[44])) ^ rotl<7>(W[34]) ^ W[41]; + W[48] = P1(W[32] ^ W[39] ^ rotl<15>(W[45])) ^ rotl<7>(W[35]) ^ W[42]; + W[49] = P1(W[33] ^ W[40] ^ rotl<15>(W[46])) ^ rotl<7>(W[36]) ^ W[43]; + W[50] = P1(W[34] ^ W[41] ^ rotl<15>(W[47])) ^ rotl<7>(W[37]) ^ W[44]; + W[51] = P1(W[35] ^ W[42] ^ rotl<15>(W[48])) ^ rotl<7>(W[38]) ^ W[45]; + W[52] = P1(W[36] ^ W[43] ^ rotl<15>(W[49])) ^ rotl<7>(W[39]) ^ W[46]; + W[53] = P1(W[37] ^ W[44] ^ rotl<15>(W[50])) ^ rotl<7>(W[40]) ^ W[47]; + W[54] = P1(W[38] ^ W[45] ^ rotl<15>(W[51])) ^ rotl<7>(W[41]) ^ W[48]; + W[55] = P1(W[39] ^ W[46] ^ rotl<15>(W[52])) ^ rotl<7>(W[42]) ^ W[49]; + W[56] = P1(W[40] ^ W[47] ^ rotl<15>(W[53])) ^ rotl<7>(W[43]) ^ W[50]; + W[57] = P1(W[41] ^ W[48] ^ rotl<15>(W[54])) ^ rotl<7>(W[44]) ^ W[51]; + W[58] = P1(W[42] ^ W[49] ^ rotl<15>(W[55])) ^ rotl<7>(W[45]) ^ W[52]; + W[59] = P1(W[43] ^ W[50] ^ rotl<15>(W[56])) ^ rotl<7>(W[46]) ^ W[53]; + W[60] = P1(W[44] ^ W[51] ^ rotl<15>(W[57])) ^ rotl<7>(W[47]) ^ W[54]; + W[61] = P1(W[45] ^ W[52] ^ rotl<15>(W[58])) ^ rotl<7>(W[48]) ^ W[55]; + W[62] = P1(W[46] ^ W[53] ^ rotl<15>(W[59])) ^ rotl<7>(W[49]) ^ W[56]; + W[63] = P1(W[47] ^ W[54] ^ rotl<15>(W[60])) ^ rotl<7>(W[50]) ^ W[57]; + W[64] = P1(W[48] ^ W[55] ^ rotl<15>(W[61])) ^ rotl<7>(W[51]) ^ W[58]; + W[65] = P1(W[49] ^ W[56] ^ rotl<15>(W[62])) ^ rotl<7>(W[52]) ^ W[59]; + W[66] = P1(W[50] ^ W[57] ^ rotl<15>(W[63])) ^ rotl<7>(W[53]) ^ W[60]; + W[67] = P1(W[51] ^ W[58] ^ rotl<15>(W[64])) ^ rotl<7>(W[54]) ^ W[61]; R1(A, B, C, D, E, F, G, H, 0x79CC4519, W[ 0], W[ 0] ^ W[ 4]); R1(D, A, B, C, H, E, F, G, 0xF3988A32, W[ 1], W[ 1] ^ W[ 5]); diff --git a/src/lib/mac/siphash/siphash.cpp b/src/lib/mac/siphash/siphash.cpp index 54adcd5a5..255a35493 100644 --- a/src/lib/mac/siphash/siphash.cpp +++ b/src/lib/mac/siphash/siphash.cpp @@ -19,16 +19,16 @@ void SipRounds(uint64_t M, secure_vector<uint64_t>& V, size_t r) for(size_t i = 0; i != r; ++i) { V0 += V1; V2 += V3; - V1 = rotate_left(V1, 13); - V3 = rotate_left(V3, 16); + V1 = rotl<13>(V1); + V3 = rotl<16>(V3); V1 ^= V0; V3 ^= V2; - V0 = rotate_left(V0, 32); + V0 = rotl<32>(V0); V2 += V1; V0 += V3; - V1 = rotate_left(V1, 17); - V3 = rotate_left(V3, 21); + V1 = rotl<17>(V1); + V3 = rotl<21>(V3); V1 ^= V2; V3 ^= V0; - V2 = rotate_left(V2, 32); + V2 = rotl<32>(V2); } V0 ^= M; diff --git a/src/lib/stream/chacha/chacha.cpp b/src/lib/stream/chacha/chacha.cpp index d56f9e60a..0bbb47bcb 100644 --- a/src/lib/stream/chacha/chacha.cpp +++ b/src/lib/stream/chacha/chacha.cpp @@ -49,12 +49,12 @@ void ChaCha::chacha_x4(uint8_t output[64*4], uint32_t input[16], size_t rounds) x08 = input[ 8], x09 = input[ 9], x10 = input[10], x11 = input[11], x12 = input[12], x13 = input[13], x14 = input[14], x15 = input[15]; -#define CHACHA_QUARTER_ROUND(a, b, c, d) \ - do { \ - a += b; d ^= a; d = rotate_left(d, 16); \ - c += d; b ^= c; b = rotate_left(b, 12); \ - a += b; d ^= a; d = rotate_left(d, 8); \ - c += d; b ^= c; b = rotate_left(b, 7); \ +#define CHACHA_QUARTER_ROUND(a, b, c, d) \ + do { \ + a += b; d ^= a; d = rotl<16>(d); \ + c += d; b ^= c; b = rotl<12>(b); \ + a += b; d ^= a; d = rotl<8>(d); \ + c += d; b ^= c; b = rotl<7>(b); \ } while(0) for(size_t r = 0; r != rounds / 2; ++r) diff --git a/src/lib/stream/salsa20/salsa20.cpp b/src/lib/stream/salsa20/salsa20.cpp index 1c8846183..e27b2d2bb 100644 --- a/src/lib/stream/salsa20/salsa20.cpp +++ b/src/lib/stream/salsa20/salsa20.cpp @@ -14,10 +14,10 @@ namespace { #define SALSA20_QUARTER_ROUND(x1, x2, x3, x4) \ do { \ - x2 ^= rotate_left(x1 + x4, 7); \ - x3 ^= rotate_left(x2 + x1, 9); \ - x4 ^= rotate_left(x3 + x2, 13); \ - x1 ^= rotate_left(x4 + x3, 18); \ + x2 ^= rotl<7>(x1 + x4); \ + x3 ^= rotl<9>(x2 + x1); \ + x4 ^= rotl<13>(x3 + x2); \ + x1 ^= rotl<18>(x4 + x3); \ } while(0) /* @@ -26,9 +26,9 @@ namespace { void hsalsa20(uint32_t output[8], const uint32_t input[16]) { uint32_t x00 = input[ 0], x01 = input[ 1], x02 = input[ 2], x03 = input[ 3], - x04 = input[ 4], x05 = input[ 5], x06 = input[ 6], x07 = input[ 7], - x08 = input[ 8], x09 = input[ 9], x10 = input[10], x11 = input[11], - x12 = input[12], x13 = input[13], x14 = input[14], x15 = input[15]; + x04 = input[ 4], x05 = input[ 5], x06 = input[ 6], x07 = input[ 7], + x08 = input[ 8], x09 = input[ 9], x10 = input[10], x11 = input[11], + x12 = input[12], x13 = input[13], x14 = input[14], x15 = input[15]; for(size_t i = 0; i != 10; ++i) { @@ -59,9 +59,9 @@ void hsalsa20(uint32_t output[8], const uint32_t input[16]) void salsa20(uint8_t output[64], const uint32_t input[16]) { uint32_t x00 = input[ 0], x01 = input[ 1], x02 = input[ 2], x03 = input[ 3], - x04 = input[ 4], x05 = input[ 5], x06 = input[ 6], x07 = input[ 7], - x08 = input[ 8], x09 = input[ 9], x10 = input[10], x11 = input[11], - x12 = input[12], x13 = input[13], x14 = input[14], x15 = input[15]; + x04 = input[ 4], x05 = input[ 5], x06 = input[ 6], x07 = input[ 7], + x08 = input[ 8], x09 = input[ 9], x10 = input[10], x11 = input[11], + x12 = input[12], x13 = input[13], x14 = input[14], x15 = input[15]; for(size_t i = 0; i != 10; ++i) { diff --git a/src/lib/utils/bswap.h b/src/lib/utils/bswap.h index 23b3113ce..c1aa8b594 100644 --- a/src/lib/utils/bswap.h +++ b/src/lib/utils/bswap.h @@ -19,7 +19,7 @@ namespace Botan { */ inline uint16_t reverse_bytes(uint16_t val) { - return rotate_left(val, 8); + return rotl<8>(val); } /** @@ -64,8 +64,8 @@ inline uint32_t reverse_bytes(uint32_t val) #else // Generic implementation - return (rotate_right(val, 8) & 0xFF00FF00) | - (rotate_left (val, 8) & 0x00FF00FF); + return (rotr<8>(val) & 0xFF00FF00) | + (rotl<8>(val) & 0x00FF00FF); #endif } diff --git a/src/lib/utils/rotate.h b/src/lib/utils/rotate.h index 6aeb631b3..adc620feb 100644 --- a/src/lib/utils/rotate.h +++ b/src/lib/utils/rotate.h @@ -1,6 +1,6 @@ /* * Word Rotation Operations -* (C) 1999-2008 Jack Lloyd +* (C) 1999-2008,2017 Jack Lloyd * * Botan is released under the Simplified BSD License (see license.txt) */ @@ -13,50 +13,74 @@ namespace Botan { /** -* Bit rotation left +* Bit rotation left by a compile-time constant amount * @param input the input word -* @param rot the number of bits to rotate -* @return input rotated left by rot bits +* @return input rotated left by ROT bits */ -template<typename T> inline T rotate_left(T input, uint8_t rot) +template<size_t ROT, typename T> +inline T rotl(T input) { - rot %= 8 * sizeof(T); - return (rot == 0) ? input : static_cast<T>((input << rot) | (input >> (8*sizeof(T)-rot)));; + static_assert(ROT > 0 && ROT < 8*sizeof(T), "Invalid rotation constant"); + return static_cast<T>((input << ROT) | (input >> (8*sizeof(T) - ROT))); } /** -* Bit rotation right +* Bit rotation right by a compile-time constant amount * @param input the input word -* @param rot the number of bits to rotate -* @return input rotated right by rot bits +* @return input rotated right by ROT bits */ -template<typename T> inline T rotate_right(T input, uint8_t rot) +template<size_t ROT, typename T> +inline T rotr(T input) { - rot %= 8 * sizeof(T); - return (rot == 0) ? input : static_cast<T>((input >> rot) | (input << (8*sizeof(T)-rot))); + static_assert(ROT > 0 && ROT < 8*sizeof(T), "Invalid rotation constant"); + return static_cast<T>((input >> ROT) | (input << (8*sizeof(T) - ROT))); } -#if BOTAN_USE_GCC_INLINE_ASM - -#if defined(BOTAN_TARGET_ARCH_IS_X86_64) || defined(BOTAN_TARGET_ARCH_IS_X86_32) - -template<> -inline uint32_t rotate_left(uint32_t input, uint8_t rot) +/** +* Bit rotation left, variable rotation amount +* @param input the input word +* @param rot the number of bits to rotate, must be between 0 and sizeof(T)*8-1 +* @return input rotated left by rot bits +*/ +template<typename T> +inline T rotl_var(T input, size_t rot) { - asm("roll %1,%0" : "+r" (input) : "c" (rot)); - return input; + /* + * This is a strange way of expressing the rotation operation but it + * so happens that both GCC and Clang will recognize it and turn it + * into a rotation, which cannot be said for many other forms. See + * https://gcc.gnu.org/bugzilla/show_bug.cgi?id=82498 + */ + return static_cast<T>((input << rot) | (input >> (-rot % (8*sizeof(T))))); } -template<> -inline uint32_t rotate_right(uint32_t input, uint8_t rot) +/** +* Bit rotation right, variable rotation amount +* @param input the input word +* @param rot the number of bits to rotate, must be between 0 and sizeof(T)*8-1 +* @return input rotated right by rot bits +*/ +template<typename T> +inline T rotr_var(T input, size_t rot) { - asm("rorl %1,%0" : "+r" (input) : "c" (rot)); - return input; + return static_cast<T>((input >> rot) | (input << (-rot % (8*sizeof(T))))); } -#endif +template<typename T> +BOTAN_DEPRECATED("Use rotl<N> or rotl_var") +inline T rotate_left(T input, size_t rot) + { + // rotl_var does not reduce + return rotl_var(input, rot % (8 * sizeof(T))); + } -#endif +template<typename T> +BOTAN_DEPRECATED("Use rotr<N> or rotr_var") +inline T rotate_right(T input, size_t rot) + { + // rotr_var does not reduce + return rotr_var(input, rot % (8 * sizeof(T))); + } } diff --git a/src/lib/utils/simd/simd_32.h b/src/lib/utils/simd/simd_32.h index 2a177b388..aca7a8e9a 100644 --- a/src/lib/utils/simd/simd_32.h +++ b/src/lib/utils/simd/simd_32.h @@ -281,56 +281,57 @@ class SIMD_4x32 final /* - Return rotate_right(x, rot1) ^ rotate_right(x, rot2) ^ rotate_right(x, rot3) + * This is used for SHA-2/SHACAL2 + * Return rotr(ROT1) ^ rotr(ROT2) ^ rotr(ROT3) */ - SIMD_4x32 rho(size_t rot1, size_t rot2, size_t rot3) const + template<size_t ROT1, size_t ROT2, size_t ROT3> + SIMD_4x32 rho() const { SIMD_4x32 res; #if defined(BOTAN_SIMD_USE_SSE2) - res.m_sse = _mm_or_si128(_mm_slli_epi32(m_sse, static_cast<int>(32-rot1)), - _mm_srli_epi32(m_sse, static_cast<int>(rot1))); + res.m_sse = _mm_or_si128(_mm_slli_epi32(m_sse, static_cast<int>(32-ROT1)), + _mm_srli_epi32(m_sse, static_cast<int>(ROT1))); res.m_sse = _mm_xor_si128( res.m_sse, - _mm_or_si128(_mm_slli_epi32(m_sse, static_cast<int>(32-rot2)), - _mm_srli_epi32(m_sse, static_cast<int>(rot2)))); + _mm_or_si128(_mm_slli_epi32(m_sse, static_cast<int>(32-ROT2)), + _mm_srli_epi32(m_sse, static_cast<int>(ROT2)))); res.m_sse = _mm_xor_si128( res.m_sse, - _mm_or_si128(_mm_slli_epi32(m_sse, static_cast<int>(32-rot3)), - _mm_srli_epi32(m_sse, static_cast<int>(rot3)))); + _mm_or_si128(_mm_slli_epi32(m_sse, static_cast<int>(32-ROT3)), + _mm_srli_epi32(m_sse, static_cast<int>(ROT3)))); #elif defined(BOTAN_SIMD_USE_ALTIVEC) - const unsigned int r1 = static_cast<unsigned int>(32-rot1); - const unsigned int r2 = static_cast<unsigned int>(32-rot2); - const unsigned int r3 = static_cast<unsigned int>(32-rot3); + const unsigned int r1 = static_cast<unsigned int>(32-ROT1); + const unsigned int r2 = static_cast<unsigned int>(32-ROT2); + const unsigned int r3 = static_cast<unsigned int>(32-ROT3); res.m_vmx = vec_rl(m_vmx, (__vector unsigned int){r1, r1, r1, r1}); res.m_vmx = vec_xor(res.m_vmx, vec_rl(m_vmx, (__vector unsigned int){r2, r2, r2, r2})); res.m_vmx = vec_xor(res.m_vmx, vec_rl(m_vmx, (__vector unsigned int){r3, r3, r3, r3})); #elif defined(BOTAN_SIMD_USE_NEON) - res.m_neon = vorrq_u32(vshlq_n_u32(m_neon, static_cast<int>(32-rot1)), - vshrq_n_u32(m_neon, static_cast<int>(rot1))); + res.m_neon = vorrq_u32(vshlq_n_u32(m_neon, static_cast<int>(32-ROT1)), + vshrq_n_u32(m_neon, static_cast<int>(ROT1))); res.m_neon = veorq_u32( res.m_neon, - vorrq_u32(vshlq_n_u32(m_neon, static_cast<int>(32-rot2)), - vshrq_n_u32(m_neon, static_cast<int>(rot2)))); + vorrq_u32(vshlq_n_u32(m_neon, static_cast<int>(32-ROT2)), + vshrq_n_u32(m_neon, static_cast<int>(ROT2)))); res.m_neon = veorq_u32( res.m_neon, - vorrq_u32(vshlq_n_u32(m_neon, static_cast<int>(32-rot3)), - vshrq_n_u32(m_neon, static_cast<int>(rot3)))); + vorrq_u32(vshlq_n_u32(m_neon, static_cast<int>(32-ROT3)), + vshrq_n_u32(m_neon, static_cast<int>(ROT3)))); #else for(size_t i = 0; i != 4; ++i) { - res.m_scalar[i] = - Botan::rotate_right(m_scalar[i], rot1) ^ - Botan::rotate_right(m_scalar[i], rot2) ^ - Botan::rotate_right(m_scalar[i], rot3); + res.m_scalar[i] = Botan::rotr<ROT1>(m_scalar[i]) ^ + Botan::rotr<ROT2>(m_scalar[i]) ^ + Botan::rotr<ROT3>(m_scalar[i]); } #endif @@ -338,38 +339,42 @@ class SIMD_4x32 final } /** - * Rotate each element of SIMD register n bits left + * Left rotation by a compile time constant */ - void rotate_left(size_t rot) + template<size_t ROT> + SIMD_4x32 rotl() const { + static_assert(ROT > 0 && ROT < 32, "Invalid rotation constant"); + #if defined(BOTAN_SIMD_USE_SSE2) - m_sse = _mm_or_si128(_mm_slli_epi32(m_sse, static_cast<int>(rot)), - _mm_srli_epi32(m_sse, static_cast<int>(32-rot))); + return SIMD_4x32(_mm_or_si128(_mm_slli_epi32(m_sse, static_cast<int>(ROT)), + _mm_srli_epi32(m_sse, static_cast<int>(32-ROT)))); #elif defined(BOTAN_SIMD_USE_ALTIVEC) - const unsigned int r = static_cast<unsigned int>(rot); - m_vmx = vec_rl(m_vmx, (__vector unsigned int){r, r, r, r}); + const unsigned int r = static_cast<unsigned int>(ROT); + return SIMD_4x32(vec_rl(m_vmx, (__vector unsigned int){r, r, r, r})); #elif defined(BOTAN_SIMD_USE_NEON) - m_neon = vorrq_u32(vshlq_n_u32(m_neon, static_cast<int>(rot)), - vshrq_n_u32(m_neon, static_cast<int>(32-rot))); + return SIMD_4x32(vorrq_u32(vshlq_n_u32(m_neon, static_cast<int>(ROT)), + vshrq_n_u32(m_neon, static_cast<int>(32-ROT)))); #else - m_scalar[0] = Botan::rotate_left(m_scalar[0], rot); - m_scalar[1] = Botan::rotate_left(m_scalar[1], rot); - m_scalar[2] = Botan::rotate_left(m_scalar[2], rot); - m_scalar[3] = Botan::rotate_left(m_scalar[3], rot); + return SIMD_4x32(Botan::rotl<ROT>(m_scalar[0]), + Botan::rotl<ROT>(m_scalar[1]), + Botan::rotl<ROT>(m_scalar[2]), + Botan::rotl<ROT>(m_scalar[3])); #endif } /** - * Rotate each element of SIMD register n bits right + * Right rotation by a compile time constant */ - void rotate_right(size_t rot) + template<size_t ROT> + SIMD_4x32 rotr() const { - rotate_left(32 - rot); + return this->rotl<32-ROT>(); } /** @@ -596,13 +601,11 @@ class SIMD_4x32 final //return SIMD_4x32(vrev64q_u32(m_neon)); // FIXME this is really slow - SIMD_4x32 ror8(m_neon); - ror8.rotate_right(8); - SIMD_4x32 rol8(m_neon); - rol8.rotate_left(8); + SIMD_4x32 ror8 = this->rotr<8>(); + SIMD_4x32 rol8 = this->rotl<8>(); - SIMD_4x32 mask1 = SIMD_4x32::splat(0xFF00FF00); - SIMD_4x32 mask2 = SIMD_4x32::splat(0x00FF00FF); + const SIMD_4x32 mask1 = SIMD_4x32::splat(0xFF00FF00); + const SIMD_4x32 mask2 = SIMD_4x32::splat(0x00FF00FF); return (ror8 & mask1) | (rol8 & mask2); #else // scalar diff --git a/src/tests/test_simd.cpp b/src/tests/test_simd.cpp index 6da743514..1f192c6b6 100644 --- a/src/tests/test_simd.cpp +++ b/src/tests/test_simd.cpp @@ -42,23 +42,21 @@ class SIMD_32_Tests final : public Test const Botan::SIMD_4x32 input(pat1, pat2, pat3, pat4); - Botan::SIMD_4x32 rol = input; - rol.rotate_left(3); - - test_eq(result, "rotate_left", rol, - Botan::rotate_left(pat1, 3), - Botan::rotate_left(pat2, 3), - Botan::rotate_left(pat3, 3), - Botan::rotate_left(pat4, 3)); - - Botan::SIMD_4x32 ror = input; - ror.rotate_right(9); - - test_eq(result, "rotate_right", ror, - Botan::rotate_right(pat1, 9), - Botan::rotate_right(pat2, 9), - Botan::rotate_right(pat3, 9), - Botan::rotate_right(pat4, 9)); + Botan::SIMD_4x32 rol = input.rotl<3>(); + + test_eq(result, "rotl", rol, + Botan::rotl<3>(pat1), + Botan::rotl<3>(pat2), + Botan::rotl<3>(pat3), + Botan::rotl<3>(pat4)); + + Botan::SIMD_4x32 ror = input.rotr<9>(); + + test_eq(result, "rotr", ror, + Botan::rotr<9>(pat1), + Botan::rotr<9>(pat2), + Botan::rotr<9>(pat3), + Botan::rotr<9>(pat4)); Botan::SIMD_4x32 add = input + splat; test_eq(result, "add +", add, pat1 + pat1, pat2 + pat1, pat3 + pat1, pat4 + pat1); |