11#include <botan/twofish.h>
12#include <botan/loadstor.h>
13#include <botan/rotate.h>
19inline void TF_E(uint32_t A, uint32_t B, uint32_t& C, uint32_t& D,
20 uint32_t RK1, uint32_t RK2,
21 const secure_vector<uint32_t>& SB)
38inline void TF_D(uint32_t A, uint32_t B, uint32_t& C, uint32_t& D,
39 uint32_t RK1, uint32_t RK2,
40 const secure_vector<uint32_t>& SB)
68 uint32_t A0, B0, C0, D0;
69 uint32_t A1, B1, C1, D1;
70 load_le(in, A0, B0, C0, D0, A1, B1, C1, D1);
81 for(
size_t k = 8; k != 40; k += 4)
83 TF_E(A0, B0, C0, D0, m_RK[k+0], m_RK[k+1], m_SB);
84 TF_E(A1, B1, C1, D1, m_RK[k+0], m_RK[k+1], m_SB);
86 TF_E(C0, D0, A0, B0, m_RK[k+2], m_RK[k+3], m_SB);
87 TF_E(C1, D1, A1, B1, m_RK[k+2], m_RK[k+3], m_SB);
99 store_le(out, C0, D0, A0, B0, C1, D1, A1, B1);
116 for(
size_t k = 8; k != 40; k += 4)
118 TF_E(A, B, C, D, m_RK[k ], m_RK[k+1], m_SB);
119 TF_E(C, D, A, B, m_RK[k+2], m_RK[k+3], m_SB);
140 uint32_t A0, B0, C0, D0;
141 uint32_t A1, B1, C1, D1;
142 load_le(in, A0, B0, C0, D0, A1, B1, C1, D1);
153 for(
size_t k = 40; k != 8; k -= 4)
155 TF_D(A0, B0, C0, D0, m_RK[k-2], m_RK[k-1], m_SB);
156 TF_D(A1, B1, C1, D1, m_RK[k-2], m_RK[k-1], m_SB);
158 TF_D(C0, D0, A0, B0, m_RK[k-4], m_RK[k-3], m_SB);
159 TF_D(C1, D1, A1, B1, m_RK[k-4], m_RK[k-3], m_SB);
171 store_le(out, C0, D0, A0, B0, C1, D1, A1, B1);
188 for(
size_t k = 40; k != 8; k -= 4)
190 TF_D(A, B, C, D, m_RK[k-2], m_RK[k-1], m_SB);
191 TF_D(C, D, A, B, m_RK[k-4], m_RK[k-3], m_SB);
206void Twofish::key_schedule(
const uint8_t key[],
size_t length)
213 for(
size_t i = 0; i != length; ++i)
220 uint8_t
X = POLY_TO_EXP[key[i] - 1];
222 uint8_t RS1 = RS[(4*i ) % 32];
223 uint8_t RS2 = RS[(4*i+1) % 32];
224 uint8_t RS3 = RS[(4*i+2) % 32];
225 uint8_t RS4 = RS[(4*i+3) % 32];
227 S[4*(i/8) ] ^= EXP_TO_POLY[(
X + POLY_TO_EXP[RS1 - 1]) % 255];
228 S[4*(i/8)+1] ^= EXP_TO_POLY[(
X + POLY_TO_EXP[RS2 - 1]) % 255];
229 S[4*(i/8)+2] ^= EXP_TO_POLY[(
X + POLY_TO_EXP[RS3 - 1]) % 255];
230 S[4*(i/8)+3] ^= EXP_TO_POLY[(
X + POLY_TO_EXP[RS4 - 1]) % 255];
236 for(
size_t i = 0; i != 256; ++i)
238 m_SB[ i] = MDS0[Q0[Q0[i]^S[ 0]]^S[ 4]];
239 m_SB[256+i] = MDS1[Q0[Q1[i]^S[ 1]]^S[ 5]];
240 m_SB[512+i] = MDS2[Q1[Q0[i]^S[ 2]]^S[ 6]];
241 m_SB[768+i] = MDS3[Q1[Q1[i]^S[ 3]]^S[ 7]];
244 for(
size_t i = 0; i < 40; i += 2)
246 uint32_t
X = MDS0[Q0[Q0[i ]^key[ 8]]^key[ 0]] ^
247 MDS1[Q0[Q1[i ]^key[ 9]]^key[ 1]] ^
248 MDS2[Q1[Q0[i ]^key[10]]^key[ 2]] ^
249 MDS3[Q1[Q1[i ]^key[11]]^key[ 3]];
250 uint32_t
Y = MDS0[Q0[Q0[i+1]^key[12]]^key[ 4]] ^
251 MDS1[Q0[Q1[i+1]^key[13]]^key[ 5]] ^
252 MDS2[Q1[Q0[i+1]^key[14]]^key[ 6]] ^
253 MDS3[Q1[Q1[i+1]^key[15]]^key[ 7]];
258 m_RK[i+1] = rotl<9>(
Y);
261 else if(length == 24)
263 for(
size_t i = 0; i != 256; ++i)
265 m_SB[ i] = MDS0[Q0[Q0[Q1[i]^S[ 0]]^S[ 4]]^S[ 8]];
266 m_SB[256+i] = MDS1[Q0[Q1[Q1[i]^S[ 1]]^S[ 5]]^S[ 9]];
267 m_SB[512+i] = MDS2[Q1[Q0[Q0[i]^S[ 2]]^S[ 6]]^S[10]];
268 m_SB[768+i] = MDS3[Q1[Q1[Q0[i]^S[ 3]]^S[ 7]]^S[11]];
271 for(
size_t i = 0; i < 40; i += 2)
273 uint32_t
X = MDS0[Q0[Q0[Q1[i ]^key[16]]^key[ 8]]^key[ 0]] ^
274 MDS1[Q0[Q1[Q1[i ]^key[17]]^key[ 9]]^key[ 1]] ^
275 MDS2[Q1[Q0[Q0[i ]^key[18]]^key[10]]^key[ 2]] ^
276 MDS3[Q1[Q1[Q0[i ]^key[19]]^key[11]]^key[ 3]];
277 uint32_t
Y = MDS0[Q0[Q0[Q1[i+1]^key[20]]^key[12]]^key[ 4]] ^
278 MDS1[Q0[Q1[Q1[i+1]^key[21]]^key[13]]^key[ 5]] ^
279 MDS2[Q1[Q0[Q0[i+1]^key[22]]^key[14]]^key[ 6]] ^
280 MDS3[Q1[Q1[Q0[i+1]^key[23]]^key[15]]^key[ 7]];
285 m_RK[i+1] = rotl<9>(
Y);
288 else if(length == 32)
290 for(
size_t i = 0; i != 256; ++i)
292 m_SB[ i] = MDS0[Q0[Q0[Q1[Q1[i]^S[ 0]]^S[ 4]]^S[ 8]]^S[12]];
293 m_SB[256+i] = MDS1[Q0[Q1[Q1[Q0[i]^S[ 1]]^S[ 5]]^S[ 9]]^S[13]];
294 m_SB[512+i] = MDS2[Q1[Q0[Q0[Q0[i]^S[ 2]]^S[ 6]]^S[10]]^S[14]];
295 m_SB[768+i] = MDS3[Q1[Q1[Q0[Q1[i]^S[ 3]]^S[ 7]]^S[11]]^S[15]];
298 for(
size_t i = 0; i < 40; i += 2)
300 uint32_t
X = MDS0[Q0[Q0[Q1[Q1[i ]^key[24]]^key[16]]^key[ 8]]^key[ 0]] ^
301 MDS1[Q0[Q1[Q1[Q0[i ]^key[25]]^key[17]]^key[ 9]]^key[ 1]] ^
302 MDS2[Q1[Q0[Q0[Q0[i ]^key[26]]^key[18]]^key[10]]^key[ 2]] ^
303 MDS3[Q1[Q1[Q0[Q1[i ]^key[27]]^key[19]]^key[11]]^key[ 3]];
304 uint32_t
Y = MDS0[Q0[Q0[Q1[Q1[i+1]^key[28]]^key[20]]^key[12]]^key[ 4]] ^
305 MDS1[Q0[Q1[Q1[Q0[i+1]^key[29]]^key[21]]^key[13]]^key[ 5]] ^
306 MDS2[Q1[Q0[Q0[Q0[i+1]^key[30]]^key[22]]^key[14]]^key[ 6]] ^
307 MDS3[Q1[Q1[Q0[Q1[i+1]^key[31]]^key[23]]^key[15]]^key[ 7]];
312 m_RK[i+1] = rotl<9>(
Y);
void verify_key_set(bool cond) const
void encrypt_n(const uint8_t in[], uint8_t out[], size_t blocks) const override
void decrypt_n(const uint8_t in[], uint8_t out[], size_t blocks) const override
void zap(std::vector< T, Alloc > &vec)
T load_le(const uint8_t in[], size_t off)
void store_le(uint16_t in, uint8_t out[2])
constexpr uint8_t get_byte(size_t byte_num, T input)
std::vector< T, secure_allocator< T > > secure_vector