VeraCrypt
aboutsummaryrefslogtreecommitdiff
path: root/src/Crypto/SerpentFast_simd.cpp
blob: 6e5fd49bb06dd345ae8e85693ad5bc112e4546d9 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
/*
* Serpent (SIMD)
* (C) 2009,2013 Jack Lloyd
*
* Botan is released under the Simplified BSD License (see license.txt)
*/

#include "SerpentFast.h"
#include "SerpentFast_sbox.h"
#if !defined(_UEFI)
#include <memory.h>
#include <stdlib.h>
#endif
#include "cpu.h"
#include "misc.h"

#if CRYPTOPP_BOOL_SSE2_INTRINSICS_AVAILABLE

/**
* This class is not a general purpose SIMD type, and only offers
* instructions needed for evaluation of specific crypto primitives.
* For example it does not currently have equality operators of any
* kind.
*/
class SIMD_4x32
{
public:

    SIMD_4x32() // zero initialized
        {
        ::memset(&m_reg, 0, sizeof(m_reg));
        }

    explicit SIMD_4x32(const unsigned __int32 B[4])
        {
        m_reg = _mm_loadu_si128(reinterpret_cast<const __m128i*>(B));
        }

    SIMD_4x32(unsigned __int32 B0, unsigned __int32 B1, unsigned __int32 B2, unsigned __int32 B3)
        {
        m_reg = _mm_set_epi32(B0, B1, B2, B3);
        }

    explicit SIMD_4x32(unsigned __int32 B)
        {
        m_reg = _mm_set1_epi32(B);
        }

    static SIMD_4x32 load_le(const void* in)
        {
        return SIMD_4x32(_mm_loadu_si128(reinterpret_cast<const __m128i*>(in)));
        }

    static SIMD_4x32 load_be(const void* in)
        {
        return load_le(in).bswap();
        }

    void store_le(unsigned __int8 out[]) const
        {
        _mm_storeu_si128(reinterpret_cast<__m128i*>(out), m_reg);
        }

    void store_be(unsigned __int8 out[]) const
        {
        bswap().store_le(out);
        }

    void rotate_left(size_t rot)
        {
        m_reg = _mm_or_si128(_mm_slli_epi32(m_reg, static_cast<int>(rot)),
                            _mm_srli_epi32(m_reg, static_cast<int>(32-rot)));

        }

    void rotate_right(size_t rot)
        {
        rotate_left(32 - rot);
        }

    void operator+=(const SIMD_4x32& other)
        {
        m_reg = _mm_add_epi32(m_reg, other.m_reg);
        }

    SIMD_4x32 operator+(const SIMD_4x32& other) const
        {
        return SIMD_4x32(_mm_add_epi32(m_reg, other.m_reg));
        }

    void operator-=(const SIMD_4x32& other)
        {
        m_reg = _mm_sub_epi32(m_reg, other.m_reg);
        }

    SIMD_4x32 operator-(const SIMD_4x32& other) const
        {
        return SIMD_4x32(_mm_sub_epi32(m_reg, other.m_reg));
        }

    void operator^=(const SIMD_4x32& other)
        {
        m_reg = _mm_xor_si128(m_reg, other.m_reg);
        }

    SIMD_4x32 operator^(const SIMD_4x32& other) const
        {
        return SIMD_4x32(_mm_xor_si128(m_reg, other.m_reg));
        }

    void operator|=(const SIMD_4x32& other)
        {
        m_reg = _mm_or_si128(m_reg, other.m_reg);
        }

    SIMD_4x32 operator&(const SIMD_4x32& other)
        {
        return SIMD_4x32(_mm_and_si128(m_reg, other.m_reg));
        }

    void operator&=(const SIMD_4x32& other)
        {
        m_reg = _mm_and_si128(m_reg, other.m_reg);
        }

    SIMD_4x32 operator<<(size_t shift) const
        {
        return SIMD_4x32(_mm_slli_epi32(m_reg, static_cast<int>(shift)));
        }

    SIMD_4x32 operator>>(size_t shift) const
        {
        return SIMD_4x32(_mm_srli_epi32(m_reg, static_cast<int>(shift)));
        }

    SIMD_4x32 operator~() const
        {
        return SIMD_4x32(_mm_xor_si128(m_reg, _mm_set1_epi32(0xFFFFFFFF)));
        }

    // (~reg) & other
    SIMD_4x32 andc(const SIMD_4x32& other)
        {
        return SIMD_4x32(_mm_andnot_si128(m_reg, other.m_reg));
        }

    SIMD_4x32 bswap() const
        {
        __m128i T = m_reg;

        T = _mm_shufflehi_epi16(T, _MM_SHUFFLE(2, 3, 0, 1));
        T = _mm_shufflelo_epi16(T, _MM_SHUFFLE(2, 3, 0, 1));

        return SIMD_4x32(_mm_or_si128(_mm_srli_epi16(T, 8),
                                    _mm_slli_epi16(T, 8)));
        }

    static void transpose(SIMD_4x32& B0, SIMD_4x32& B1,
                        SIMD_4x32& B2, SIMD_4x32& B3)
        {
        __m128i T0 = _mm_unpacklo_epi32(B0.m_reg, B1.m_reg);
        __m128i T1 = _mm_unpacklo_epi32(B2.m_reg, B3.m_reg);
        __m128i T2 = _mm_unpackhi_epi32(B0.m_reg, B1.m_reg);
        __m128i T3 = _mm_unpackhi_epi32(B2.m_reg, B3.m_reg);
        B0.m_reg = _mm_unpacklo_epi64(T0, T1);
        B1.m_reg = _mm_unpackhi_epi64(T0, T1);
        B2.m_reg = _mm_unpacklo_epi64(T2, T3);
        B3.m_reg = _mm_unpackhi_epi64(T2, T3);
        }

private:

    explicit SIMD_4x32(__m128i in) { m_reg = in; }

    __m128i m_reg;

};

typedef SIMD_4x32 SIMD_32;

#define key_xor(round, B0, B1, B2, B3)                             \
   do {                                                            \
      B0 ^= SIMD_32(round_key[4*round  ]);                       \
      B1 ^= SIMD_32(round_key[4*round+1]);                       \
      B2 ^= SIMD_32(round_key[4*round+2]);                       \
      B3 ^= SIMD_32(round_key[4*round+3]);                       \
   } while(0);

/*
* Serpent's linear transformations
*/
#define transform(B0, B1, B2, B3)                                  \
   do {                                                            \
      B0.rotate_left(13);                                          \
      B2.rotate_left(3);                                           \
      B1 ^= B0 ^ B2;                                               \
      B3 ^= B2 ^ (B0 << 3);                                        \
      B1.rotate_left(1);                                           \
      B3.rotate_left(7);                                           \
      B0 ^= B1 ^ B3;                                               \
      B2 ^= B3 ^ (B1 << 7);                                        \
      B0.rotate_left(5);                                           \
      B2.rotate_left(22);                                          \
   } while(0);

#define i_transform(B0, B1, B2, B3)                                \
   do {                                                            \
      B2.rotate_right(22);                                         \
      B0.rotate_right(5);                                          \
      B2 ^= B3 ^ (B1 << 7);                                        \
      B0 ^= B1 ^ B3;                                               \
      B3.rotate_right(7);                                          \
      B1.rotate_right(1);                                          \
      B3 ^= B2 ^ (B0 << 3);                                        \
      B1 ^= B0 ^ B2;                                               \
      B2.rotate_right(3);                                          \
      B0.rotate_right(13);                                         \
   } while(0);


#if (!defined (DEBUG) || !defined (TC_WINDOWS_DRIVER))
/*
* SIMD Serpent Encryption of 4 blocks in parallel
*/
extern "C" void serpent_simd_encrypt_blocks_4(const unsigned __int8 in[], unsigned __int8 out[], unsigned __int32* round_key)
{
   SIMD_32 B0 = SIMD_32::load_le(in);
   SIMD_32 B1 = SIMD_32::load_le(in + 16);
   SIMD_32 B2 = SIMD_32::load_le(in + 32);
   SIMD_32 B3 = SIMD_32::load_le(in + 48);

   SIMD_32::transpose(B0, B1, B2, B3);

   key_xor( 0,B0,B1,B2,B3); SBoxE1(SIMD_32,B0,B1,B2,B3); transform(B0,B1,B2,B3);
   key_xor( 1,B0,B1,B2,B3); SBoxE2(SIMD_32,B0,B1,B2,B3); transform(B0,B1,B2,B3);
   key_xor( 2,B0,B1,B2,B3); SBoxE3(SIMD_32,B0,B1,B2,B3); transform(B0,B1,B2,B3);
   key_xor( 3,B0,B1,B2,B3); SBoxE4(SIMD_32,B0,B1,B2,B3); transform(B0,B1,B2,B3);
   key_xor( 4,B0,B1,B2,B3); SBoxE5(SIMD_32,B0,B1,B2,B3); transform(B0,B1,B2,B3);
   key_xor( 5,B0,B1,B2,B3); SBoxE6(SIMD_32,B0,B1,B2,B3); transform(B0,B1,B2,B3);
   key_xor( 6,B0,B1,B2,B3); SBoxE7(SIMD_32,B0,B1,B2,B3); transform(B0,B1,B2,B3);
   key_xor( 7,B0,B1,B2,B3); SBoxE8(SIMD_32,B0,B1,B2,B3); transform(B0,B1,B2,B3);

   key_xor( 8,B0,B1,B2,B3); SBoxE1(SIMD_32,B0,B1,B2,B3); transform(B0,B1,B2,B3);
   key_xor( 9,B0,B1,B2,B3); SBoxE2(SIMD_32,B0,B1,B2,B3); transform(B0,B1,B2,B3);
   key_xor(10,B0,B1,B2,B3); SBoxE3(SIMD_32,B0,B1,B2,B3); transform(B0,B1,B2,B3);
   key_xor(11,B0,B1,B2,B3); SBoxE4(SIMD_32,B0,B1,B2,B3); transform(B0,B1,B2,B3);
   key_xor(12,B0,B1,B2,B3); SBoxE5(SIMD_32,B0,B1,B2,B3); transform(B0,B1,B2,B3);
   key_xor(13,B0,B1,B2,B3); SBoxE6(SIMD_32,B0,B1,B2,B3); transform(B0,B1,B2,B3);
   key_xor(14,B0,B1,B2,B3); SBoxE7(SIMD_32,B0,B1,B2,B3); transform(B0,B1,B2,B3);
   key_xor(15,B0,B1,B2,B3); SBoxE8(SIMD_32,B0,B1,B2,B3); transform(B0,B1,B2,B3);

   key_xor(16,B0,B1,B2,B3); SBoxE1(SIMD_32,B0,B1,B2,B3); transform(B0,B1,B2,B3);
   key_xor(17,B0,B1,B2,B3); SBoxE2(SIMD_32,B0,B1,B2,B3); transform(B0,B1,B2,B3);
   key_xor(18,B0,B1,B2,B3); SBoxE3(SIMD_32,B0,B1,B2,B3); transform(B0,B1,B2,B3);
   key_xor(19,B0,B1,B2,B3); SBoxE4(SIMD_32,B0,B1,B2,B3); transform(B0,B1,B2,B3);
   key_xor(20,B0,B1,B2,B3); SBoxE5(SIMD_32,B0,B1,B2,B3); transform(B0,B1,B2,B3);
   key_xor(21,B0,B1,B2,B3); SBoxE6(SIMD_32,B0,B1,B2,B3); transform(B0,B1,B2,B3);
   key_xor(22,B0,B1,B2,B3); SBoxE7(SIMD_32,B0,B1,B2,B3); transform(B0,B1,B2,B3);
   key_xor(23,B0,B1,B2,B3); SBoxE8(SIMD_32,B0,B1,B2,B3); transform(B0,B1,B2,B3);

   key_xor(24,B0,B1,B2,B3); SBoxE1(SIMD_32,B0,B1,B2,B3); transform(B0,B1,B2,B3);
   key_xor(25,B0,B1,B2,B3); SBoxE2(SIMD_32,B0,B1,B2,B3); transform(B0,B1,B2,B3);
   key_xor(26,B0,B1,B2,B3); SBoxE3(SIMD_32,B0,B1,B2,B3); transform(B0,B1,B2,B3);
   key_xor(27,B0,B1,B2,B3); SBoxE4(SIMD_32,B0,B1,B2,B3); transform(B0,B1,B2,B3);
   key_xor(28,B0,B1,B2,B3); SBoxE5(SIMD_32,B0,B1,B2,B3); transform(B0,B1,B2,B3);
   key_xor(29,B0,B1,B2,B3); SBoxE6(SIMD_32,B0,B1,B2,B3); transform(B0,B1,B2,B3);
   key_xor(30,B0,B1,B2,B3); SBoxE7(SIMD_32,B0,B1,B2,B3); transform(B0,B1,B2,B3);
   key_xor(31,B0,B1,B2,B3); SBoxE8(SIMD_32,B0,B1,B2,B3); key_xor(32,B0,B1,B2,B3);

   SIMD_32::transpose(B0, B1, B2, B3);

   B0.store_le(out);
   B1.store_le(out + 16);
   B2.store_le(out + 32);
   B3.store_le(out + 48);
}

/*
* SIMD Serpent Decryption of 4 blocks in parallel
*/
extern "C" void serpent_simd_decrypt_blocks_4(const unsigned __int8 in[], unsigned __int8 out[], unsigned __int32* round_key)
{
   SIMD_32 B0 = SIMD_32::load_le(in);
   SIMD_32 B1 = SIMD_32::load_le(in + 16);
   SIMD_32 B2 = SIMD_32::load_le(in + 32);
   SIMD_32 B3 = SIMD_32::load_le(in + 48);

   SIMD_32::transpose(B0, B1, B2, B3);

   key_xor(32,B0,B1,B2,B3);  SBoxD8(SIMD_32,B0,B1,B2,B3); key_xor(31,B0,B1,B2,B3);
   i_transform(B0,B1,B2,B3); SBoxD7(SIMD_32,B0,B1,B2,B3); key_xor(30,B0,B1,B2,B3);
   i_transform(B0,B1,B2,B3); SBoxD6(SIMD_32,B0,B1,B2,B3); key_xor(29,B0,B1,B2,B3);
   i_transform(B0,B1,B2,B3); SBoxD5(SIMD_32,B0,B1,B2,B3); key_xor(28,B0,B1,B2,B3);
   i_transform(B0,B1,B2,B3); SBoxD4(SIMD_32,B0,B1,B2,B3); key_xor(27,B0,B1,B2,B3);
   i_transform(B0,B1,B2,B3); SBoxD3(SIMD_32,B0,B1,B2,B3); key_xor(26,B0,B1,B2,B3);
   i_transform(B0,B1,B2,B3); SBoxD2(SIMD_32,B0,B1,B2,B3); key_xor(25,B0,B1,B2,B3);
   i_transform(B0,B1,B2,B3); SBoxD1(SIMD_32,B0,B1,B2,B3); key_xor(24,B0,B1,B2,B3);

   i_transform(B0,B1,B2,B3); SBoxD8(SIMD_32,B0,B1,B2,B3); key_xor(23,B0,B1,B2,B3);
   i_transform(B0,B1,B2,B3); SBoxD7(SIMD_32,B0,B1,B2,B3); key_xor(22,B0,B1,B2,B3);
   i_transform(B0,B1,B2,B3); SBoxD6(SIMD_32,B0,B1,B2,B3); key_xor(21,B0,B1,B2,B3);
   i_transform(B0,B1,B2,B3); SBoxD5(SIMD_32,B0,B1,B2,B3); key_xor(20,B0,B1,B2,B3);
   i_transform(B0,B1,B2,B3); SBoxD4(SIMD_32,B0,B1,B2,B3); key_xor(19,B0,B1,B2,B3);
   i_transform(B0,B1,B2,B3); SBoxD3(SIMD_32,B0,B1,B2,B3); key_xor(18,B0,B1,B2,B3);
   i_transform(B0,B1,B2,B3); SBoxD2(SIMD_32,B0,B1,B2,B3); key_xor(17,B0,B1,B2,B3);
   i_transform(B0,B1,B2,B3); SBoxD1(SIMD_32,B0,B1,B2,B3); key_xor(16,B0,B1,B2,B3);

   i_transform(B0,B1,B2,B3); SBoxD8(SIMD_32,B0,B1,B2,B3); key_xor(15,B0,B1,B2,B3);
   i_transform(B0,B1,B2,B3); SBoxD7(SIMD_32,B0,B1,B2,B3); key_xor(14,B0,B1,B2,B3);
   i_transform(B0,B1,B2,B3); SBoxD6(SIMD_32,B0,B1,B2,B3); key_xor(13,B0,B1,B2,B3);
   i_transform(B0,B1,B2,B3); SBoxD5(SIMD_32,B0,B1,B2,B3); key_xor(12,B0,B1,B2,B3);
   i_transform(B0,B1,B2,B3); SBoxD4(SIMD_32,B0,B1,B2,B3); key_xor(11,B0,B1,B2,B3);
   i_transform(B0,B1,B2,B3); SBoxD3(SIMD_32,B0,B1,B2,B3); key_xor(10,B0,B1,B2,B3);
   i_transform(B0,B1,B2,B3); SBoxD2(SIMD_32,B0,B1,B2,B3); key_xor( 9,B0,B1,B2,B3);
   i_transform(B0,B1,B2,B3); SBoxD1(SIMD_32,B0,B1,B2,B3); key_xor( 8,B0,B1,B2,B3);

   i_transform(B0,B1,B2,B3); SBoxD8(SIMD_32,B0,B1,B2,B3); key_xor( 7,B0,B1,B2,B3);
   i_transform(B0,B1,B2,B3); SBoxD7(SIMD_32,B0,B1,B2,B3); key_xor( 6,B0,B1,B2,B3);
   i_transform(B0,B1,B2,B3); SBoxD6(SIMD_32,B0,B1,B2,B3); key_xor( 5,B0,B1,B2,B3);
   i_transform(B0,B1,B2,B3); SBoxD5(SIMD_32,B0,B1,B2,B3); key_xor( 4,B0,B1,B2,B3);
   i_transform(B0,B1,B2,B3); SBoxD4(SIMD_32,B0,B1,B2,B3); key_xor( 3,B0,B1,B2,B3);
   i_transform(B0,B1,B2,B3); SBoxD3(SIMD_32,B0,B1,B2,B3); key_xor( 2,B0,B1,B2,B3);
   i_transform(B0,B1,B2,B3); SBoxD2(SIMD_32,B0,B1,B2,B3); key_xor( 1,B0,B1,B2,B3);
   i_transform(B0,B1,B2,B3); SBoxD1(SIMD_32,B0,B1,B2,B3); key_xor( 0,B0,B1,B2,B3);

   SIMD_32::transpose(B0, B1, B2, B3);

   B0.store_le(out);
   B1.store_le(out + 16);
   B2.store_le(out + 32);
   B3.store_le(out + 48);
}
#endif
#undef key_xor
#undef transform
#undef i_transform

#endif