1 /*
2 * Public Domain poly1305 from Andrew Moon
3 * poly1305-donna-unrolled.c from https://github.com/floodyberry/poly1305-donna
4 */
5
6 /* $OpenBSD: poly1305.c,v 1.3 2013/12/19 22:57:13 djm Exp $ */
7
8 #include <libwebsockets.h>
9 #include "lws-ssh.h"
10
11 #define mul32x32_64(a,b) ((uint64_t)(a) * (b))
12
13 #define U8TO32_LE(p) \
14 (((uint32_t)((p)[0])) | \
15 ((uint32_t)((p)[1]) << 8) | \
16 ((uint32_t)((p)[2]) << 16) | \
17 ((uint32_t)((p)[3]) << 24))
18
19 #define U32TO8_LE(p, v) \
20 do { \
21 (p)[0] = (uint8_t)((v)); \
22 (p)[1] = (uint8_t)((v) >> 8); \
23 (p)[2] = (uint8_t)((v) >> 16); \
24 (p)[3] = (uint8_t)((v) >> 24); \
25 } while (0)
26
27 void
poly1305_auth(unsigned char out[POLY1305_TAGLEN],const unsigned char * m,size_t inlen,const unsigned char key[POLY1305_KEYLEN])28 poly1305_auth(unsigned char out[POLY1305_TAGLEN],
29 const unsigned char *m, size_t inlen,
30 const unsigned char key[POLY1305_KEYLEN])
31 {
32 uint32_t t0,t1,t2,t3;
33 uint32_t h0,h1,h2,h3,h4;
34 uint32_t r0,r1,r2,r3,r4;
35 uint32_t s1,s2,s3,s4;
36 uint32_t b, nb;
37 size_t j;
38 uint64_t t[5];
39 uint64_t f0,f1,f2,f3;
40 uint32_t g0,g1,g2,g3,g4;
41 uint64_t c;
42 unsigned char mp[16];
43
44 /* clamp key */
45 t0 = U8TO32_LE(key + 0);
46 t1 = U8TO32_LE(key + 4);
47 t2 = U8TO32_LE(key + 8);
48 t3 = U8TO32_LE(key + 12);
49
50 /* precompute multipliers */
51 r0 = t0 & 0x3ffffff; t0 >>= 26; t0 |= t1 << 6;
52 r1 = t0 & 0x3ffff03; t1 >>= 20; t1 |= t2 << 12;
53 r2 = t1 & 0x3ffc0ff; t2 >>= 14; t2 |= t3 << 18;
54 r3 = t2 & 0x3f03fff; t3 >>= 8;
55 r4 = t3 & 0x00fffff;
56
57 s1 = r1 * 5;
58 s2 = r2 * 5;
59 s3 = r3 * 5;
60 s4 = r4 * 5;
61
62 /* init state */
63 h0 = 0;
64 h1 = 0;
65 h2 = 0;
66 h3 = 0;
67 h4 = 0;
68
69 /* full blocks */
70 if (inlen < 16)
71 goto poly1305_donna_atmost15bytes;
72
73 poly1305_donna_16bytes:
74 m += 16;
75 inlen -= 16;
76
77 t0 = U8TO32_LE(m - 16);
78 t1 = U8TO32_LE(m - 12);
79 t2 = U8TO32_LE(m - 8);
80 t3 = U8TO32_LE(m - 4);
81
82 h0 += t0 & 0x3ffffff;
83 h1 += (uint32_t)(((((uint64_t)t1 << 32) | t0) >> 26) & 0x3ffffff);
84 h2 += (uint32_t)(((((uint64_t)t2 << 32) | t1) >> 20) & 0x3ffffff);
85 h3 += (uint32_t)(((((uint64_t)t3 << 32) | t2) >> 14) & 0x3ffffff);
86 h4 += (uint32_t)((t3 >> 8) | (1 << 24));
87
88 poly1305_donna_mul:
89 t[0] = mul32x32_64(h0,r0) + mul32x32_64(h1,s4) +
90 mul32x32_64(h2,s3) + mul32x32_64(h3,s2) +
91 mul32x32_64(h4,s1);
92 t[1] = mul32x32_64(h0,r1) + mul32x32_64(h1,r0) +
93 mul32x32_64(h2,s4) + mul32x32_64(h3,s3) +
94 mul32x32_64(h4,s2);
95 t[2] = mul32x32_64(h0,r2) + mul32x32_64(h1,r1) +
96 mul32x32_64(h2,r0) + mul32x32_64(h3,s4) +
97 mul32x32_64(h4,s3);
98 t[3] = mul32x32_64(h0,r3) + mul32x32_64(h1,r2) +
99 mul32x32_64(h2,r1) + mul32x32_64(h3,r0) +
100 mul32x32_64(h4,s4);
101 t[4] = mul32x32_64(h0,r4) + mul32x32_64(h1,r3) +
102 mul32x32_64(h2,r2) + mul32x32_64(h3,r1) +
103 mul32x32_64(h4,r0);
104
105 h0 = (uint32_t)t[0] & 0x3ffffff; c = (t[0] >> 26);
106 t[1] += c; h1 = (uint32_t)t[1] & 0x3ffffff; b = (uint32_t)(t[1] >> 26);
107 t[2] += b; h2 = (uint32_t)t[2] & 0x3ffffff; b = (uint32_t)(t[2] >> 26);
108 t[3] += b; h3 = (uint32_t)t[3] & 0x3ffffff; b = (uint32_t)(t[3] >> 26);
109 t[4] += b; h4 = (uint32_t)t[4] & 0x3ffffff; b = (uint32_t)(t[4] >> 26);
110 h0 += b * 5;
111
112 if (inlen >= 16)
113 goto poly1305_donna_16bytes;
114
115 /* final bytes */
116 poly1305_donna_atmost15bytes:
117 if (!inlen)
118 goto poly1305_donna_finish;
119
120 for (j = 0; j < inlen; j++)
121 mp[j] = m[j];
122 mp[j++] = 1;
123 for (; j < 16; j++)
124 mp[j] = 0;
125 inlen = 0;
126
127 t0 = U8TO32_LE(mp + 0);
128 t1 = U8TO32_LE(mp + 4);
129 t2 = U8TO32_LE(mp + 8);
130 t3 = U8TO32_LE(mp + 12);
131
132 h0 += t0 & 0x3ffffff;
133 h1 += (uint32_t)(((((uint64_t)t1 << 32) | t0) >> 26) & 0x3ffffff);
134 h2 += (uint32_t)(((((uint64_t)t2 << 32) | t1) >> 20) & 0x3ffffff);
135 h3 += (uint32_t)(((((uint64_t)t3 << 32) | t2) >> 14) & 0x3ffffff);
136 h4 += (uint32_t)(t3 >> 8);
137
138 goto poly1305_donna_mul;
139
140 poly1305_donna_finish:
141 b = h0 >> 26; h0 = h0 & 0x3ffffff;
142 h1 += b; b = h1 >> 26; h1 = h1 & 0x3ffffff;
143 h2 += b; b = h2 >> 26; h2 = h2 & 0x3ffffff;
144 h3 += b; b = h3 >> 26; h3 = h3 & 0x3ffffff;
145 h4 += b; b = h4 >> 26; h4 = h4 & 0x3ffffff;
146 h0 += b * 5; b = h0 >> 26; h0 = h0 & 0x3ffffff;
147 h1 += b;
148
149 g0 = h0 + 5; b = g0 >> 26; g0 &= 0x3ffffff;
150 g1 = h1 + b; b = g1 >> 26; g1 &= 0x3ffffff;
151 g2 = h2 + b; b = g2 >> 26; g2 &= 0x3ffffff;
152 g3 = h3 + b; b = g3 >> 26; g3 &= 0x3ffffff;
153 g4 = h4 + b - (1 << 26);
154
155 b = (g4 >> 31) - 1;
156 nb = ~b;
157 h0 = (h0 & nb) | (g0 & b);
158 h1 = (h1 & nb) | (g1 & b);
159 h2 = (h2 & nb) | (g2 & b);
160 h3 = (h3 & nb) | (g3 & b);
161 h4 = (h4 & nb) | (g4 & b);
162
163 f0 = ((h0 ) | (h1 << 26)) + (uint64_t)U8TO32_LE(&key[16]);
164 f1 = ((h1 >> 6) | (h2 << 20)) + (uint64_t)U8TO32_LE(&key[20]);
165 f2 = ((h2 >> 12) | (h3 << 14)) + (uint64_t)U8TO32_LE(&key[24]);
166 f3 = ((h3 >> 18) | (h4 << 8)) + (uint64_t)U8TO32_LE(&key[28]);
167
168 U32TO8_LE(&out[ 0], f0); f1 += (f0 >> 32);
169 U32TO8_LE(&out[ 4], f1); f2 += (f1 >> 32);
170 U32TO8_LE(&out[ 8], f2); f3 += (f2 >> 32);
171 U32TO8_LE(&out[12], f3);
172 }
173