Coverage Report

Created: 2025-11-11 06:30

next uncovered line (L), next uncovered region (R), next uncovered branch (B)
/src/openssh/smult_curve25519_ref.c
Line
Count
Source
1
/* $OpenBSD: smult_curve25519_ref.c,v 1.2 2013/11/02 22:02:14 markus Exp $ */
2
/*
3
version 20081011
4
Matthew Dempsky
5
Public domain.
6
Derived from public domain code by D. J. Bernstein.
7
*/
8
9
int crypto_scalarmult_curve25519(unsigned char *, const unsigned char *, const unsigned char *);
10
11
static void add(unsigned int out[32],const unsigned int a[32],const unsigned int b[32])
12
4.01M
{
13
4.01M
  unsigned int j;
14
4.01M
  unsigned int u;
15
4.01M
  u = 0;
16
128M
  for (j = 0;j < 31;++j) { u += a[j] + b[j]; out[j] = u & 255; u >>= 8; }
17
4.01M
  u += a[31] + b[31]; out[31] = u;
18
4.01M
}
19
20
static void sub(unsigned int out[32],const unsigned int a[32],const unsigned int b[32])
21
4.01M
{
22
4.01M
  unsigned int j;
23
4.01M
  unsigned int u;
24
4.01M
  u = 218;
25
128M
  for (j = 0;j < 31;++j) {
26
124M
    u += a[j] + 65280 - b[j];
27
124M
    out[j] = u & 255;
28
124M
    u >>= 8;
29
124M
  }
30
4.01M
  u += a[31] - b[31];
31
4.01M
  out[31] = u;
32
4.01M
}
33
34
static void squeeze(unsigned int a[32])
35
10.0M
{
36
10.0M
  unsigned int j;
37
10.0M
  unsigned int u;
38
10.0M
  u = 0;
39
322M
  for (j = 0;j < 31;++j) { u += a[j]; a[j] = u & 255; u >>= 8; }
40
10.0M
  u += a[31]; a[31] = u & 127;
41
10.0M
  u = 19 * (u >> 7);
42
322M
  for (j = 0;j < 31;++j) { u += a[j]; a[j] = u & 255; u >>= 8; }
43
10.0M
  u += a[31]; a[31] = u;
44
10.0M
}
45
46
static const unsigned int minusp[32] = {
47
 19, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 128
48
} ;
49
50
static void freeze(unsigned int a[32])
51
3.93k
{
52
3.93k
  unsigned int aorig[32];
53
3.93k
  unsigned int j;
54
3.93k
  unsigned int negative;
55
56
129k
  for (j = 0;j < 32;++j) aorig[j] = a[j];
57
3.93k
  add(a,a,minusp);
58
3.93k
  negative = -((a[31] >> 7) & 1);
59
129k
  for (j = 0;j < 32;++j) a[j] ^= negative & (aorig[j] ^ a[j]);
60
3.93k
}
61
62
static void mult(unsigned int out[32],const unsigned int a[32],const unsigned int b[32])
63
5.06M
{
64
5.06M
  unsigned int i;
65
5.06M
  unsigned int j;
66
5.06M
  unsigned int u;
67
68
167M
  for (i = 0;i < 32;++i) {
69
161M
    u = 0;
70
2.83G
    for (j = 0;j <= i;++j) u += a[j] * b[i - j];
71
2.67G
    for (j = i + 1;j < 32;++j) u += 38 * a[j] * b[i + 32 - j];
72
161M
    out[i] = u;
73
161M
  }
74
5.06M
  squeeze(out);
75
5.06M
}
76
77
static void mult121665(unsigned int out[32],const unsigned int a[32])
78
1.00M
{
79
1.00M
  unsigned int j;
80
1.00M
  unsigned int u;
81
82
1.00M
  u = 0;
83
32.0M
  for (j = 0;j < 31;++j) { u += 121665 * a[j]; out[j] = u & 255; u >>= 8; }
84
1.00M
  u += 121665 * a[31]; out[31] = u & 127;
85
1.00M
  u = 19 * (u >> 7);
86
32.0M
  for (j = 0;j < 31;++j) { u += out[j]; out[j] = u & 255; u >>= 8; }
87
1.00M
  u += out[j]; out[j] = u;
88
1.00M
}
89
90
static void square(unsigned int out[32],const unsigned int a[32])
91
5.01M
{
92
5.01M
  unsigned int i;
93
5.01M
  unsigned int j;
94
5.01M
  unsigned int u;
95
96
165M
  for (i = 0;i < 32;++i) {
97
160M
    u = 0;
98
1.44G
    for (j = 0;j < i - j;++j) u += a[j] * a[i - j];
99
1.36G
    for (j = i + 1;j < i + 32 - j;++j) u += 38 * a[j] * a[i + 32 - j];
100
160M
    u *= 2;
101
160M
    if ((i & 1) == 0) {
102
80.1M
      u += a[i / 2] * a[i / 2];
103
80.1M
      u += 38 * a[i / 2 + 16] * a[i / 2 + 16];
104
80.1M
    }
105
160M
    out[i] = u;
106
160M
  }
107
5.01M
  squeeze(out);
108
5.01M
}
109
110
static void select(unsigned int p[64],unsigned int q[64],const unsigned int r[64],const unsigned int s[64],unsigned int b)
111
2.00M
{
112
2.00M
  unsigned int j;
113
2.00M
  unsigned int t;
114
2.00M
  unsigned int bminus1;
115
116
2.00M
  bminus1 = b - 1;
117
130M
  for (j = 0;j < 64;++j) {
118
128M
    t = bminus1 & (r[j] ^ s[j]);
119
128M
    p[j] = s[j] ^ t;
120
128M
    q[j] = r[j] ^ t;
121
128M
  }
122
2.00M
}
123
124
static void mainloop(unsigned int work[64],const unsigned char e[32])
125
3.93k
{
126
3.93k
  unsigned int xzm1[64];
127
3.93k
  unsigned int xzm[64];
128
3.93k
  unsigned int xzmb[64];
129
3.93k
  unsigned int xzm1b[64];
130
3.93k
  unsigned int xznb[64];
131
3.93k
  unsigned int xzn1b[64];
132
3.93k
  unsigned int a0[64];
133
3.93k
  unsigned int a1[64];
134
3.93k
  unsigned int b0[64];
135
3.93k
  unsigned int b1[64];
136
3.93k
  unsigned int c1[64];
137
3.93k
  unsigned int r[32];
138
3.93k
  unsigned int s[32];
139
3.93k
  unsigned int t[32];
140
3.93k
  unsigned int u[32];
141
3.93k
  unsigned int j;
142
3.93k
  unsigned int b;
143
3.93k
  int pos;
144
145
129k
  for (j = 0;j < 32;++j) xzm1[j] = work[j];
146
3.93k
  xzm1[32] = 1;
147
125k
  for (j = 33;j < 64;++j) xzm1[j] = 0;
148
149
3.93k
  xzm[0] = 1;
150
251k
  for (j = 1;j < 64;++j) xzm[j] = 0;
151
152
1.00M
  for (pos = 254;pos >= 0;--pos) {
153
1.00M
    b = e[pos / 8] >> (pos & 7);
154
1.00M
    b &= 1;
155
1.00M
    select(xzmb,xzm1b,xzm,xzm1,b);
156
1.00M
    add(a0,xzmb,xzmb + 32);
157
1.00M
    sub(a0 + 32,xzmb,xzmb + 32);
158
1.00M
    add(a1,xzm1b,xzm1b + 32);
159
1.00M
    sub(a1 + 32,xzm1b,xzm1b + 32);
160
1.00M
    square(b0,a0);
161
1.00M
    square(b0 + 32,a0 + 32);
162
1.00M
    mult(b1,a1,a0 + 32);
163
1.00M
    mult(b1 + 32,a1 + 32,a0);
164
1.00M
    add(c1,b1,b1 + 32);
165
1.00M
    sub(c1 + 32,b1,b1 + 32);
166
1.00M
    square(r,c1 + 32);
167
1.00M
    sub(s,b0,b0 + 32);
168
1.00M
    mult121665(t,s);
169
1.00M
    add(u,t,b0);
170
1.00M
    mult(xznb,b0,b0 + 32);
171
1.00M
    mult(xznb + 32,s,u);
172
1.00M
    square(xzn1b,c1);
173
1.00M
    mult(xzn1b + 32,r,work);
174
1.00M
    select(xzm,xzm1,xznb,xzn1b,b);
175
1.00M
  }
176
177
255k
  for (j = 0;j < 64;++j) work[j] = xzm[j];
178
3.93k
}
179
180
static void recip(unsigned int out[32],const unsigned int z[32])
181
3.93k
{
182
3.93k
  unsigned int z2[32];
183
3.93k
  unsigned int z9[32];
184
3.93k
  unsigned int z11[32];
185
3.93k
  unsigned int z2_5_0[32];
186
3.93k
  unsigned int z2_10_0[32];
187
3.93k
  unsigned int z2_20_0[32];
188
3.93k
  unsigned int z2_50_0[32];
189
3.93k
  unsigned int z2_100_0[32];
190
3.93k
  unsigned int t0[32];
191
3.93k
  unsigned int t1[32];
192
3.93k
  int i;
193
194
  /* 2 */ square(z2,z);
195
3.93k
  /* 4 */ square(t1,z2);
196
3.93k
  /* 8 */ square(t0,t1);
197
3.93k
  /* 9 */ mult(z9,t0,z);
198
3.93k
  /* 11 */ mult(z11,z9,z2);
199
3.93k
  /* 22 */ square(t0,z11);
200
3.93k
  /* 2^5 - 2^0 = 31 */ mult(z2_5_0,t0,z9);
201
202
  /* 2^6 - 2^1 */ square(t0,z2_5_0);
203
3.93k
  /* 2^7 - 2^2 */ square(t1,t0);
204
3.93k
  /* 2^8 - 2^3 */ square(t0,t1);
205
3.93k
  /* 2^9 - 2^4 */ square(t1,t0);
206
3.93k
  /* 2^10 - 2^5 */ square(t0,t1);
207
3.93k
  /* 2^10 - 2^0 */ mult(z2_10_0,t0,z2_5_0);
208
209
  /* 2^11 - 2^1 */ square(t0,z2_10_0);
210
3.93k
  /* 2^12 - 2^2 */ square(t1,t0);
211
19.6k
  /* 2^20 - 2^10 */ for (i = 2;i < 10;i += 2) { square(t0,t1); square(t1,t0); }
212
3.93k
  /* 2^20 - 2^0 */ mult(z2_20_0,t1,z2_10_0);
213
214
  /* 2^21 - 2^1 */ square(t0,z2_20_0);
215
3.93k
  /* 2^22 - 2^2 */ square(t1,t0);
216
39.3k
  /* 2^40 - 2^20 */ for (i = 2;i < 20;i += 2) { square(t0,t1); square(t1,t0); }
217
3.93k
  /* 2^40 - 2^0 */ mult(t0,t1,z2_20_0);
218
219
  /* 2^41 - 2^1 */ square(t1,t0);
220
3.93k
  /* 2^42 - 2^2 */ square(t0,t1);
221
19.6k
  /* 2^50 - 2^10 */ for (i = 2;i < 10;i += 2) { square(t1,t0); square(t0,t1); }
222
3.93k
  /* 2^50 - 2^0 */ mult(z2_50_0,t0,z2_10_0);
223
224
  /* 2^51 - 2^1 */ square(t0,z2_50_0);
225
3.93k
  /* 2^52 - 2^2 */ square(t1,t0);
226
98.3k
  /* 2^100 - 2^50 */ for (i = 2;i < 50;i += 2) { square(t0,t1); square(t1,t0); }
227
3.93k
  /* 2^100 - 2^0 */ mult(z2_100_0,t1,z2_50_0);
228
229
  /* 2^101 - 2^1 */ square(t1,z2_100_0);
230
3.93k
  /* 2^102 - 2^2 */ square(t0,t1);
231
196k
  /* 2^200 - 2^100 */ for (i = 2;i < 100;i += 2) { square(t1,t0); square(t0,t1); }
232
3.93k
  /* 2^200 - 2^0 */ mult(t1,t0,z2_100_0);
233
234
  /* 2^201 - 2^1 */ square(t0,t1);
235
3.93k
  /* 2^202 - 2^2 */ square(t1,t0);
236
98.3k
  /* 2^250 - 2^50 */ for (i = 2;i < 50;i += 2) { square(t0,t1); square(t1,t0); }
237
3.93k
  /* 2^250 - 2^0 */ mult(t0,t1,z2_50_0);
238
239
  /* 2^251 - 2^1 */ square(t1,t0);
240
3.93k
  /* 2^252 - 2^2 */ square(t0,t1);
241
3.93k
  /* 2^253 - 2^3 */ square(t1,t0);
242
3.93k
  /* 2^254 - 2^4 */ square(t0,t1);
243
3.93k
  /* 2^255 - 2^5 */ square(t1,t0);
244
3.93k
  /* 2^255 - 21 */ mult(out,t1,z11);
245
3.93k
}
246
247
int crypto_scalarmult_curve25519(unsigned char *q,
248
  const unsigned char *n,
249
  const unsigned char *p)
250
3.93k
{
251
3.93k
  unsigned int work[96];
252
3.93k
  unsigned char e[32];
253
3.93k
  unsigned int i;
254
129k
  for (i = 0;i < 32;++i) e[i] = n[i];
255
3.93k
  e[0] &= 248;
256
3.93k
  e[31] &= 127;
257
3.93k
  e[31] |= 64;
258
129k
  for (i = 0;i < 32;++i) work[i] = p[i];
259
3.93k
  mainloop(work,e);
260
3.93k
  recip(work + 32,work + 32);
261
3.93k
  mult(work + 64,work,work + 32);
262
3.93k
  freeze(work + 64);
263
129k
  for (i = 0;i < 32;++i) q[i] = work[64 + i];
264
3.93k
  return 0;
265
3.93k
}