summaryrefslogtreecommitdiff
path: root/third_party/heimdal/lib/hcrypto/x25519/fe_25_5/fe.h
blob: f216669e4e0d1f5b604f562364e9075f8b0b8fb1 (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
/*
 Ignores top bit of h.
 */

void
fe25519_frombytes(fe25519 h, const unsigned char *s)
{
    int64_t h0 = load_4(s);
    int64_t h1 = load_3(s + 4) << 6;
    int64_t h2 = load_3(s + 7) << 5;
    int64_t h3 = load_3(s + 10) << 3;
    int64_t h4 = load_3(s + 13) << 2;
    int64_t h5 = load_4(s + 16);
    int64_t h6 = load_3(s + 20) << 7;
    int64_t h7 = load_3(s + 23) << 5;
    int64_t h8 = load_3(s + 26) << 4;
    int64_t h9 = (load_3(s + 29) & 8388607) << 2;

    int64_t carry0;
    int64_t carry1;
    int64_t carry2;
    int64_t carry3;
    int64_t carry4;
    int64_t carry5;
    int64_t carry6;
    int64_t carry7;
    int64_t carry8;
    int64_t carry9;

    carry9 = (h9 + (int64_t)(1L << 24)) >> 25;
    h0 += carry9 * 19;
    h9 -= carry9 * ((uint64_t) 1L << 25);
    carry1 = (h1 + (int64_t)(1L << 24)) >> 25;
    h2 += carry1;
    h1 -= carry1 * ((uint64_t) 1L << 25);
    carry3 = (h3 + (int64_t)(1L << 24)) >> 25;
    h4 += carry3;
    h3 -= carry3 * ((uint64_t) 1L << 25);
    carry5 = (h5 + (int64_t)(1L << 24)) >> 25;
    h6 += carry5;
    h5 -= carry5 * ((uint64_t) 1L << 25);
    carry7 = (h7 + (int64_t)(1L << 24)) >> 25;
    h8 += carry7;
    h7 -= carry7 * ((uint64_t) 1L << 25);

    carry0 = (h0 + (int64_t)(1L << 25)) >> 26;
    h1 += carry0;
    h0 -= carry0 * ((uint64_t) 1L << 26);
    carry2 = (h2 + (int64_t)(1L << 25)) >> 26;
    h3 += carry2;
    h2 -= carry2 * ((uint64_t) 1L << 26);
    carry4 = (h4 + (int64_t)(1L << 25)) >> 26;
    h5 += carry4;
    h4 -= carry4 * ((uint64_t) 1L << 26);
    carry6 = (h6 + (int64_t)(1L << 25)) >> 26;
    h7 += carry6;
    h6 -= carry6 * ((uint64_t) 1L << 26);
    carry8 = (h8 + (int64_t)(1L << 25)) >> 26;
    h9 += carry8;
    h8 -= carry8 * ((uint64_t) 1L << 26);

    h[0] = (int32_t) h0;
    h[1] = (int32_t) h1;
    h[2] = (int32_t) h2;
    h[3] = (int32_t) h3;
    h[4] = (int32_t) h4;
    h[5] = (int32_t) h5;
    h[6] = (int32_t) h6;
    h[7] = (int32_t) h7;
    h[8] = (int32_t) h8;
    h[9] = (int32_t) h9;
}

/*
 Preconditions:
 |h| bounded by 1.1*2^26,1.1*2^25,1.1*2^26,1.1*2^25,etc.

 Write p=2^255-19; q=floor(h/p).
 Basic claim: q = floor(2^(-255)(h + 19 2^(-25)h9 + 2^(-1))).

 Proof:
 Have |h|<=p so |q|<=1 so |19^2 2^(-255) q|<1/4.
 Also have |h-2^230 h9|<2^231 so |19 2^(-255)(h-2^230 h9)|<1/4.

 Write y=2^(-1)-19^2 2^(-255)q-19 2^(-255)(h-2^230 h9).
 Then 0<y<1.

 Write r=h-pq.
 Have 0<=r<=p-1=2^255-20.
 Thus 0<=r+19(2^-255)r<r+19(2^-255)2^255<=2^255-1.

 Write x=r+19(2^-255)r+y.
 Then 0<x<2^255 so floor(2^(-255)x) = 0 so floor(q+2^(-255)x) = q.

 Have q+2^(-255)x = 2^(-255)(h + 19 2^(-25) h9 + 2^(-1))
 so floor(2^(-255)(h + 19 2^(-25) h9 + 2^(-1))) = q.
*/

static void
fe25519_reduce(fe25519 h, const fe25519 f)
{
    int32_t h0 = f[0];
    int32_t h1 = f[1];
    int32_t h2 = f[2];
    int32_t h3 = f[3];
    int32_t h4 = f[4];
    int32_t h5 = f[5];
    int32_t h6 = f[6];
    int32_t h7 = f[7];
    int32_t h8 = f[8];
    int32_t h9 = f[9];

    int32_t q;
    int32_t carry0, carry1, carry2, carry3, carry4, carry5, carry6, carry7, carry8, carry9;

    q = (19 * h9 + ((uint32_t) 1L << 24)) >> 25;
    q = (h0 + q) >> 26;
    q = (h1 + q) >> 25;
    q = (h2 + q) >> 26;
    q = (h3 + q) >> 25;
    q = (h4 + q) >> 26;
    q = (h5 + q) >> 25;
    q = (h6 + q) >> 26;
    q = (h7 + q) >> 25;
    q = (h8 + q) >> 26;
    q = (h9 + q) >> 25;

    /* Goal: Output h-(2^255-19)q, which is between 0 and 2^255-20. */
    h0 += 19 * q;
    /* Goal: Output h-2^255 q, which is between 0 and 2^255-20. */

    carry0 = h0 >> 26;
    h1 += carry0;
    h0 -= carry0 * ((uint32_t) 1L << 26);
    carry1 = h1 >> 25;
    h2 += carry1;
    h1 -= carry1 * ((uint32_t) 1L << 25);
    carry2 = h2 >> 26;
    h3 += carry2;
    h2 -= carry2 * ((uint32_t) 1L << 26);
    carry3 = h3 >> 25;
    h4 += carry3;
    h3 -= carry3 * ((uint32_t) 1L << 25);
    carry4 = h4 >> 26;
    h5 += carry4;
    h4 -= carry4 * ((uint32_t) 1L << 26);
    carry5 = h5 >> 25;
    h6 += carry5;
    h5 -= carry5 * ((uint32_t) 1L << 25);
    carry6 = h6 >> 26;
    h7 += carry6;
    h6 -= carry6 * ((uint32_t) 1L << 26);
    carry7 = h7 >> 25;
    h8 += carry7;
    h7 -= carry7 * ((uint32_t) 1L << 25);
    carry8 = h8 >> 26;
    h9 += carry8;
    h8 -= carry8 * ((uint32_t) 1L << 26);
    carry9 = h9 >> 25;
    h9 -= carry9 * ((uint32_t) 1L << 25);

    h[0] = h0;
    h[1] = h1;
    h[2] = h2;
    h[3] = h3;
    h[4] = h4;
    h[5] = h5;
    h[6] = h6;
    h[7] = h7;
    h[8] = h8;
    h[9] = h9;
}

/*
 Goal: Output h0+...+2^255 h10-2^255 q, which is between 0 and 2^255-20.
 Have h0+...+2^230 h9 between 0 and 2^255-1;
 evidently 2^255 h10-2^255 q = 0.

 Goal: Output h0+...+2^230 h9.
 */

void
fe25519_tobytes(unsigned char *s, const fe25519 h)
{
    fe25519 t;

    fe25519_reduce(t, h);
    s[0]  = t[0] >> 0;
    s[1]  = t[0] >> 8;
    s[2]  = t[0] >> 16;
    s[3]  = (t[0] >> 24) | (t[1] * ((uint32_t) 1 << 2));
    s[4]  = t[1] >> 6;
    s[5]  = t[1] >> 14;
    s[6]  = (t[1] >> 22) | (t[2] * ((uint32_t) 1 << 3));
    s[7]  = t[2] >> 5;
    s[8]  = t[2] >> 13;
    s[9]  = (t[2] >> 21) | (t[3] * ((uint32_t) 1 << 5));
    s[10] = t[3] >> 3;
    s[11] = t[3] >> 11;
    s[12] = (t[3] >> 19) | (t[4] * ((uint32_t) 1 << 6));
    s[13] = t[4] >> 2;
    s[14] = t[4] >> 10;
    s[15] = t[4] >> 18;
    s[16] = t[5] >> 0;
    s[17] = t[5] >> 8;
    s[18] = t[5] >> 16;
    s[19] = (t[5] >> 24) | (t[6] * ((uint32_t) 1 << 1));
    s[20] = t[6] >> 7;
    s[21] = t[6] >> 15;
    s[22] = (t[6] >> 23) | (t[7] * ((uint32_t) 1 << 3));
    s[23] = t[7] >> 5;
    s[24] = t[7] >> 13;
    s[25] = (t[7] >> 21) | (t[8] * ((uint32_t) 1 << 4));
    s[26] = t[8] >> 4;
    s[27] = t[8] >> 12;
    s[28] = (t[8] >> 20) | (t[9] * ((uint32_t) 1 << 6));
    s[29] = t[9] >> 2;
    s[30] = t[9] >> 10;
    s[31] = t[9] >> 18;
}