1 /*
2 * Public Domain poly1305 from Andrew Moon
3 * poly1305-donna-unrolled.c from https://github.com/floodyberry/poly1305-donna
4 */
5
6 #include "../system.h"
7
8 #include "poly1305.h"
9
10 #define mul32x32_64(a,b) ((uint64_t)(a) * (b))
11
12 #define U8TO32_LE(p) \
13 (((uint32_t)((p)[0])) | \
14 ((uint32_t)((p)[1]) << 8) | \
15 ((uint32_t)((p)[2]) << 16) | \
16 ((uint32_t)((p)[3]) << 24))
17
18 #define U32TO8_LE(p, v) \
19 do { \
20 (p)[0] = (uint8_t)((v)); \
21 (p)[1] = (uint8_t)((v) >> 8); \
22 (p)[2] = (uint8_t)((v) >> 16); \
23 (p)[3] = (uint8_t)((v) >> 24); \
24 } while (0)
25
26 void
poly1305_auth(unsigned char out[POLY1305_TAGLEN],const unsigned char * m,size_t inlen,const unsigned char key[POLY1305_KEYLEN])27 poly1305_auth(unsigned char out[POLY1305_TAGLEN], const unsigned char *m, size_t inlen, const unsigned char key[POLY1305_KEYLEN]) {
28 uint32_t t0, t1, t2, t3;
29 uint32_t h0, h1, h2, h3, h4;
30 uint32_t r0, r1, r2, r3, r4;
31 uint32_t s1, s2, s3, s4;
32 uint32_t b, nb;
33 size_t j;
34 uint64_t t[5];
35 uint64_t f0, f1, f2, f3;
36 uint32_t g0, g1, g2, g3, g4;
37 uint64_t c;
38 unsigned char mp[16];
39
40 /* clamp key */
41 t0 = U8TO32_LE(key + 0);
42 t1 = U8TO32_LE(key + 4);
43 t2 = U8TO32_LE(key + 8);
44 t3 = U8TO32_LE(key + 12);
45
46 /* precompute multipliers */
47 r0 = t0 & 0x3ffffff;
48 t0 >>= 26;
49 t0 |= t1 << 6;
50 r1 = t0 & 0x3ffff03;
51 t1 >>= 20;
52 t1 |= t2 << 12;
53 r2 = t1 & 0x3ffc0ff;
54 t2 >>= 14;
55 t2 |= t3 << 18;
56 r3 = t2 & 0x3f03fff;
57 t3 >>= 8;
58 r4 = t3 & 0x00fffff;
59
60 s1 = r1 * 5;
61 s2 = r2 * 5;
62 s3 = r3 * 5;
63 s4 = r4 * 5;
64
65 /* init state */
66 h0 = 0;
67 h1 = 0;
68 h2 = 0;
69 h3 = 0;
70 h4 = 0;
71
72 /* full blocks */
73 if(inlen < 16) {
74 goto poly1305_donna_atmost15bytes;
75 }
76
77 poly1305_donna_16bytes:
78 m += 16;
79 inlen -= 16;
80
81 t0 = U8TO32_LE(m - 16);
82 t1 = U8TO32_LE(m - 12);
83 t2 = U8TO32_LE(m - 8);
84 t3 = U8TO32_LE(m - 4);
85
86 h0 += t0 & 0x3ffffff;
87 h1 += ((((uint64_t) t1 << 32) | t0) >> 26) & 0x3ffffff;
88 h2 += ((((uint64_t) t2 << 32) | t1) >> 20) & 0x3ffffff;
89 h3 += ((((uint64_t) t3 << 32) | t2) >> 14) & 0x3ffffff;
90 h4 += (t3 >> 8) | (1 << 24);
91
92 poly1305_donna_mul:
93 t[0] = mul32x32_64(h0, r0) + mul32x32_64(h1, s4) + mul32x32_64(h2, s3) + mul32x32_64(h3, s2) + mul32x32_64(h4, s1);
94 t[1] = mul32x32_64(h0, r1) + mul32x32_64(h1, r0) + mul32x32_64(h2, s4) + mul32x32_64(h3, s3) + mul32x32_64(h4, s2);
95 t[2] = mul32x32_64(h0, r2) + mul32x32_64(h1, r1) + mul32x32_64(h2, r0) + mul32x32_64(h3, s4) + mul32x32_64(h4, s3);
96 t[3] = mul32x32_64(h0, r3) + mul32x32_64(h1, r2) + mul32x32_64(h2, r1) + mul32x32_64(h3, r0) + mul32x32_64(h4, s4);
97 t[4] = mul32x32_64(h0, r4) + mul32x32_64(h1, r3) + mul32x32_64(h2, r2) + mul32x32_64(h3, r1) + mul32x32_64(h4, r0);
98
99 h0 = (uint32_t) t[0] & 0x3ffffff;
100 c = (t[0] >> 26);
101 t[1] += c;
102 h1 = (uint32_t) t[1] & 0x3ffffff;
103 b = (uint32_t)(t[1] >> 26);
104 t[2] += b;
105 h2 = (uint32_t) t[2] & 0x3ffffff;
106 b = (uint32_t)(t[2] >> 26);
107 t[3] += b;
108 h3 = (uint32_t) t[3] & 0x3ffffff;
109 b = (uint32_t)(t[3] >> 26);
110 t[4] += b;
111 h4 = (uint32_t) t[4] & 0x3ffffff;
112 b = (uint32_t)(t[4] >> 26);
113 h0 += b * 5;
114
115 if(inlen >= 16) {
116 goto poly1305_donna_16bytes;
117 }
118
119 /* final bytes */
120 poly1305_donna_atmost15bytes:
121
122 if(!inlen) {
123 goto poly1305_donna_finish;
124 }
125
126 for(j = 0; j < inlen; j++) {
127 mp[j] = m[j];
128 }
129
130 mp[j++] = 1;
131
132 for(; j < 16; j++) {
133 mp[j] = 0;
134 }
135
136 inlen = 0;
137
138 t0 = U8TO32_LE(mp + 0);
139 t1 = U8TO32_LE(mp + 4);
140 t2 = U8TO32_LE(mp + 8);
141 t3 = U8TO32_LE(mp + 12);
142
143 h0 += t0 & 0x3ffffff;
144 h1 += ((((uint64_t) t1 << 32) | t0) >> 26) & 0x3ffffff;
145 h2 += ((((uint64_t) t2 << 32) | t1) >> 20) & 0x3ffffff;
146 h3 += ((((uint64_t) t3 << 32) | t2) >> 14) & 0x3ffffff;
147 h4 += (t3 >> 8);
148
149 goto poly1305_donna_mul;
150
151 poly1305_donna_finish:
152 b = h0 >> 26;
153 h0 = h0 & 0x3ffffff;
154 h1 += b;
155 b = h1 >> 26;
156 h1 = h1 & 0x3ffffff;
157 h2 += b;
158 b = h2 >> 26;
159 h2 = h2 & 0x3ffffff;
160 h3 += b;
161 b = h3 >> 26;
162 h3 = h3 & 0x3ffffff;
163 h4 += b;
164 b = h4 >> 26;
165 h4 = h4 & 0x3ffffff;
166 h0 += b * 5;
167 b = h0 >> 26;
168 h0 = h0 & 0x3ffffff;
169 h1 += b;
170
171 g0 = h0 + 5;
172 b = g0 >> 26;
173 g0 &= 0x3ffffff;
174 g1 = h1 + b;
175 b = g1 >> 26;
176 g1 &= 0x3ffffff;
177 g2 = h2 + b;
178 b = g2 >> 26;
179 g2 &= 0x3ffffff;
180 g3 = h3 + b;
181 b = g3 >> 26;
182 g3 &= 0x3ffffff;
183 g4 = h4 + b - (1 << 26);
184
185 b = (g4 >> 31) - 1;
186 nb = ~b;
187 h0 = (h0 & nb) | (g0 & b);
188 h1 = (h1 & nb) | (g1 & b);
189 h2 = (h2 & nb) | (g2 & b);
190 h3 = (h3 & nb) | (g3 & b);
191 h4 = (h4 & nb) | (g4 & b);
192
193 f0 = ((h0) | (h1 << 26)) + (uint64_t) U8TO32_LE(&key[16]);
194 f1 = ((h1 >> 6) | (h2 << 20)) + (uint64_t) U8TO32_LE(&key[20]);
195 f2 = ((h2 >> 12) | (h3 << 14)) + (uint64_t) U8TO32_LE(&key[24]);
196 f3 = ((h3 >> 18) | (h4 << 8)) + (uint64_t) U8TO32_LE(&key[28]);
197
198 U32TO8_LE(&out[0], f0);
199 f1 += (f0 >> 32);
200 U32TO8_LE(&out[4], f1);
201 f2 += (f1 >> 32);
202 U32TO8_LE(&out[8], f2);
203 f3 += (f2 >> 32);
204 U32TO8_LE(&out[12], f3);
205 }
206