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
|
/* SPDX-License-Identifier: GPL-2.0
*
* Copyright (C) 2015 Google Inc. All Rights Reserved.
* Copyright (C) 2015-2018 Jason A. Donenfeld <Jason@zx2c4.com>. All Rights Reserved.
*
* Original author: Peter Schwabe <peter@cryptojedi.org>
*/
#include <linux/kernel.h>
#include <linux/string.h>
typedef struct { uint64_t v[5]; } fe25519;
asmlinkage void x25519_x86_64_work_cswap(fe25519 *, uint64_t);
asmlinkage void x25519_x86_64_mul(fe25519 *out, const fe25519 *a, const fe25519 *b);
asmlinkage void x25519_x86_64_square(fe25519 *out, const fe25519 *a);
asmlinkage void x25519_x86_64_freeze(fe25519 *);
asmlinkage void x25519_x86_64_ladderstep(fe25519 *work);
enum { CURVE25519_POINT_SIZE = 32 };
static __always_inline void normalize_secret(u8 secret[CURVE25519_POINT_SIZE])
{
secret[0] &= 248;
secret[31] &= 127;
secret[31] |= 64;
}
static void fe25519_setint(fe25519 *r, unsigned v)
{
r->v[0] = v;
r->v[1] = 0;
r->v[2] = 0;
r->v[3] = 0;
r->v[4] = 0;
}
// Assumes input x being reduced below 2^255
static void fe25519_pack(unsigned char r[32], const fe25519 *x)
{
fe25519 t;
t = *x;
x25519_x86_64_freeze(&t);
r[0] = (uint8_t)(t.v[0] & 0xff);
r[1] = (uint8_t)((t.v[0] >> 8) & 0xff);
r[2] = (uint8_t)((t.v[0] >> 16) & 0xff);
r[3] = (uint8_t)((t.v[0] >> 24) & 0xff);
r[4] = (uint8_t)((t.v[0] >> 32) & 0xff);
r[5] = (uint8_t)((t.v[0] >> 40) & 0xff);
r[6] = (uint8_t)((t.v[0] >> 48));
r[6] ^= (uint8_t)((t.v[1] << 3) & 0xf8);
r[7] = (uint8_t)((t.v[1] >> 5) & 0xff);
r[8] = (uint8_t)((t.v[1] >> 13) & 0xff);
r[9] = (uint8_t)((t.v[1] >> 21) & 0xff);
r[10] = (uint8_t)((t.v[1] >> 29) & 0xff);
r[11] = (uint8_t)((t.v[1] >> 37) & 0xff);
r[12] = (uint8_t)((t.v[1] >> 45));
r[12] ^= (uint8_t)((t.v[2] << 6) & 0xc0);
r[13] = (uint8_t)((t.v[2] >> 2) & 0xff);
r[14] = (uint8_t)((t.v[2] >> 10) & 0xff);
r[15] = (uint8_t)((t.v[2] >> 18) & 0xff);
r[16] = (uint8_t)((t.v[2] >> 26) & 0xff);
r[17] = (uint8_t)((t.v[2] >> 34) & 0xff);
r[18] = (uint8_t)((t.v[2] >> 42) & 0xff);
r[19] = (uint8_t)((t.v[2] >> 50));
r[19] ^= (uint8_t)((t.v[3] << 1) & 0xfe);
r[20] = (uint8_t)((t.v[3] >> 7) & 0xff);
r[21] = (uint8_t)((t.v[3] >> 15) & 0xff);
r[22] = (uint8_t)((t.v[3] >> 23) & 0xff);
r[23] = (uint8_t)((t.v[3] >> 31) & 0xff);
r[24] = (uint8_t)((t.v[3] >> 39) & 0xff);
r[25] = (uint8_t)((t.v[3] >> 47));
r[25] ^= (uint8_t)((t.v[4] << 4) & 0xf0);
r[26] = (uint8_t)((t.v[4] >> 4) & 0xff);
r[27] = (uint8_t)((t.v[4] >> 12) & 0xff);
r[28] = (uint8_t)((t.v[4] >> 20) & 0xff);
r[29] = (uint8_t)((t.v[4] >> 28) & 0xff);
r[30] = (uint8_t)((t.v[4] >> 36) & 0xff);
r[31] = (uint8_t)((t.v[4] >> 44));
}
static void fe25519_unpack(fe25519 *r, const uint8_t x[32])
{
r->v[0] = x[0];
r->v[0] += (uint64_t)x[1] << 8;
r->v[0] += (uint64_t)x[2] << 16;
r->v[0] += (uint64_t)x[3] << 24;
r->v[0] += (uint64_t)x[4] << 32;
r->v[0] += (uint64_t)x[5] << 40;
r->v[0] += ((uint64_t)x[6] & 7) << 48;
r->v[1] = x[6] >> 3;
r->v[1] += (uint64_t)x[7] << 5;
r->v[1] += (uint64_t)x[8] << 13;
r->v[1] += (uint64_t)x[9] << 21;
r->v[1] += (uint64_t)x[10] << 29;
r->v[1] += (uint64_t)x[11] << 37;
r->v[1] += ((uint64_t)x[12] & 63) << 45;
r->v[2] = x[12] >> 6;
r->v[2] += (uint64_t)x[13] << 2;
r->v[2] += (uint64_t)x[14] << 10;
r->v[2] += (uint64_t)x[15] << 18;
r->v[2] += (uint64_t)x[16] << 26;
r->v[2] += (uint64_t)x[17] << 34;
r->v[2] += (uint64_t)x[18] << 42;
r->v[2] += ((uint64_t)x[19] & 1) << 50;
r->v[3] = x[19] >> 1;
r->v[3] += (uint64_t)x[20] << 7;
r->v[3] += (uint64_t)x[21] << 15;
r->v[3] += (uint64_t)x[22] << 23;
r->v[3] += (uint64_t)x[23] << 31;
r->v[3] += (uint64_t)x[24] << 39;
r->v[3] += ((uint64_t)x[25] & 15) << 47;
r->v[4] = x[25] >> 4;
r->v[4] += (uint64_t)x[26] << 4;
r->v[4] += (uint64_t)x[27] << 12;
r->v[4] += (uint64_t)x[28] << 20;
r->v[4] += (uint64_t)x[29] << 28;
r->v[4] += (uint64_t)x[30] << 36;
r->v[4] += ((uint64_t)x[31] & 127) << 44;
}
static void fe25519_invert(fe25519 *r, const fe25519 *x)
{
fe25519 z2;
fe25519 z9;
fe25519 z11;
fe25519 z2_5_0;
fe25519 z2_10_0;
fe25519 z2_20_0;
fe25519 z2_50_0;
fe25519 z2_100_0;
fe25519 t;
int i;
/* 2 */ x25519_x86_64_square(&z2, x);
/* 4 */ x25519_x86_64_square(&t, &z2);
/* 8 */ x25519_x86_64_square(&t, &t);
/* 9 */ x25519_x86_64_mul(&z9, &t, x);
/* 11 */ x25519_x86_64_mul(&z11, &z9, &z2);
/* 22 */ x25519_x86_64_square(&t, &z11);
/* 2^5 - 2^0 = 31 */ x25519_x86_64_mul(&z2_5_0, &t, &z9);
/* 2^6 - 2^1 */ x25519_x86_64_square(&t, &z2_5_0);
/* 2^20 - 2^10 */ for (i = 1; i < 5; i++) { x25519_x86_64_square(&t, &t); }
/* 2^10 - 2^0 */ x25519_x86_64_mul(&z2_10_0, &t, &z2_5_0);
/* 2^11 - 2^1 */ x25519_x86_64_square(&t, &z2_10_0);
/* 2^20 - 2^10 */ for (i = 1; i < 10; i++) { x25519_x86_64_square(&t, &t); }
/* 2^20 - 2^0 */ x25519_x86_64_mul(&z2_20_0, &t, &z2_10_0);
/* 2^21 - 2^1 */ x25519_x86_64_square(&t, &z2_20_0);
/* 2^40 - 2^20 */ for (i = 1; i < 20; i++) { x25519_x86_64_square(&t, &t); }
/* 2^40 - 2^0 */ x25519_x86_64_mul(&t, &t, &z2_20_0);
/* 2^41 - 2^1 */ x25519_x86_64_square(&t, &t);
/* 2^50 - 2^10 */ for (i = 1; i < 10; i++) { x25519_x86_64_square(&t, &t); }
/* 2^50 - 2^0 */ x25519_x86_64_mul(&z2_50_0, &t, &z2_10_0);
/* 2^51 - 2^1 */ x25519_x86_64_square(&t, &z2_50_0);
/* 2^100 - 2^50 */ for (i = 1; i < 50; i++) { x25519_x86_64_square(&t, &t); }
/* 2^100 - 2^0 */ x25519_x86_64_mul(&z2_100_0, &t, &z2_50_0);
/* 2^101 - 2^1 */ x25519_x86_64_square(&t, &z2_100_0);
/* 2^200 - 2^100 */ for (i = 1; i < 100; i++) {
x25519_x86_64_square(&t, &t);
}
/* 2^200 - 2^0 */ x25519_x86_64_mul(&t, &t, &z2_100_0);
/* 2^201 - 2^1 */ x25519_x86_64_square(&t, &t);
/* 2^250 - 2^50 */ for (i = 1; i < 50; i++) { x25519_x86_64_square(&t, &t); }
/* 2^250 - 2^0 */ x25519_x86_64_mul(&t, &t, &z2_50_0);
/* 2^251 - 2^1 */ x25519_x86_64_square(&t, &t);
/* 2^252 - 2^2 */ x25519_x86_64_square(&t, &t);
/* 2^253 - 2^3 */ x25519_x86_64_square(&t, &t);
/* 2^254 - 2^4 */ x25519_x86_64_square(&t, &t);
/* 2^255 - 2^5 */ x25519_x86_64_square(&t, &t);
/* 2^255 - 21 */ x25519_x86_64_mul(r, &t, &z11);
}
static void mladder(fe25519 *xr, fe25519 *zr, const uint8_t s[32])
{
int i, j;
uint8_t prevbit = 0;
fe25519 work[5];
work[0] = *xr;
fe25519_setint(work + 1, 1);
fe25519_setint(work + 2, 0);
work[3] = *xr;
fe25519_setint(work + 4, 1);
j = 6;
for (i = 31; i >= 0; i--) {
while (j >= 0) {
const uint8_t bit = 1 & (s[i] >> j);
const uint64_t swap = bit ^ prevbit;
prevbit = bit;
x25519_x86_64_work_cswap(work + 1, swap);
x25519_x86_64_ladderstep(work);
j -= 1;
}
j = 7;
}
*xr = work[1];
*zr = work[2];
}
bool curve25519_amd64(u8 out[CURVE25519_POINT_SIZE], const u8 scalar[CURVE25519_POINT_SIZE], const u8 point[CURVE25519_POINT_SIZE])
{
fe25519 t;
fe25519 z;
uint8_t e[32];
memcpy(e, scalar, sizeof(e));
normalize_secret(e);
fe25519_unpack(&t, point);
mladder(&t, &z, e);
fe25519_invert(&z, &z);
x25519_x86_64_mul(&t, &t, &z);
fe25519_pack(out, &t);
return true;
}
|
