xref: /freebsd/crypto/openssh/poly1305.c (revision 640235e2c2ba32947f7c59d168437ffa1280f1e6)
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 "includes.h"
9 
10 #include <sys/types.h>
11 #ifdef HAVE_STDINT_H
12 # include <stdint.h>
13 #endif
14 
15 #include "poly1305.h"
16 
17 #define mul32x32_64(a,b) ((uint64_t)(a) * (b))
18 
19 #define U8TO32_LE(p) \
20 	(((uint32_t)((p)[0])) | \
21 	 ((uint32_t)((p)[1]) <<  8) | \
22 	 ((uint32_t)((p)[2]) << 16) | \
23 	 ((uint32_t)((p)[3]) << 24))
24 
25 #define U32TO8_LE(p, v) \
26 	do { \
27 		(p)[0] = (uint8_t)((v)); \
28 		(p)[1] = (uint8_t)((v) >>  8); \
29 		(p)[2] = (uint8_t)((v) >> 16); \
30 		(p)[3] = (uint8_t)((v) >> 24); \
31 	} while (0)
32 
33 void
34 poly1305_auth(unsigned char out[POLY1305_TAGLEN], const unsigned char *m, size_t inlen, const unsigned char key[POLY1305_KEYLEN]) {
35 	uint32_t t0,t1,t2,t3;
36 	uint32_t h0,h1,h2,h3,h4;
37 	uint32_t r0,r1,r2,r3,r4;
38 	uint32_t s1,s2,s3,s4;
39 	uint32_t b, nb;
40 	size_t j;
41 	uint64_t t[5];
42 	uint64_t f0,f1,f2,f3;
43 	uint32_t g0,g1,g2,g3,g4;
44 	uint64_t c;
45 	unsigned char mp[16];
46 
47 	/* clamp key */
48 	t0 = U8TO32_LE(key+0);
49 	t1 = U8TO32_LE(key+4);
50 	t2 = U8TO32_LE(key+8);
51 	t3 = U8TO32_LE(key+12);
52 
53 	/* precompute multipliers */
54 	r0 = t0 & 0x3ffffff; t0 >>= 26; t0 |= t1 << 6;
55 	r1 = t0 & 0x3ffff03; t1 >>= 20; t1 |= t2 << 12;
56 	r2 = t1 & 0x3ffc0ff; t2 >>= 14; t2 |= t3 << 18;
57 	r3 = t2 & 0x3f03fff; 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) goto poly1305_donna_atmost15bytes;
74 poly1305_donna_16bytes:
75 	m += 16;
76 	inlen -= 16;
77 
78 	t0 = U8TO32_LE(m-16);
79 	t1 = U8TO32_LE(m-12);
80 	t2 = U8TO32_LE(m-8);
81 	t3 = U8TO32_LE(m-4);
82 
83 	h0 += t0 & 0x3ffffff;
84 	h1 += ((((uint64_t)t1 << 32) | t0) >> 26) & 0x3ffffff;
85 	h2 += ((((uint64_t)t2 << 32) | t1) >> 20) & 0x3ffffff;
86 	h3 += ((((uint64_t)t3 << 32) | t2) >> 14) & 0x3ffffff;
87 	h4 += (t3 >> 8) | (1 << 24);
88 
89 
90 poly1305_donna_mul:
91 	t[0]  = mul32x32_64(h0,r0) + mul32x32_64(h1,s4) + mul32x32_64(h2,s3) + mul32x32_64(h3,s2) + mul32x32_64(h4,s1);
92 	t[1]  = mul32x32_64(h0,r1) + mul32x32_64(h1,r0) + mul32x32_64(h2,s4) + mul32x32_64(h3,s3) + mul32x32_64(h4,s2);
93 	t[2]  = mul32x32_64(h0,r2) + mul32x32_64(h1,r1) + mul32x32_64(h2,r0) + mul32x32_64(h3,s4) + mul32x32_64(h4,s3);
94 	t[3]  = mul32x32_64(h0,r3) + mul32x32_64(h1,r2) + mul32x32_64(h2,r1) + mul32x32_64(h3,r0) + mul32x32_64(h4,s4);
95 	t[4]  = mul32x32_64(h0,r4) + mul32x32_64(h1,r3) + mul32x32_64(h2,r2) + mul32x32_64(h3,r1) + mul32x32_64(h4,r0);
96 
97 	                h0 = (uint32_t)t[0] & 0x3ffffff; c =           (t[0] >> 26);
98 	t[1] += c;      h1 = (uint32_t)t[1] & 0x3ffffff; b = (uint32_t)(t[1] >> 26);
99 	t[2] += b;      h2 = (uint32_t)t[2] & 0x3ffffff; b = (uint32_t)(t[2] >> 26);
100 	t[3] += b;      h3 = (uint32_t)t[3] & 0x3ffffff; b = (uint32_t)(t[3] >> 26);
101 	t[4] += b;      h4 = (uint32_t)t[4] & 0x3ffffff; b = (uint32_t)(t[4] >> 26);
102 	h0 += b * 5;
103 
104 	if (inlen >= 16) goto poly1305_donna_16bytes;
105 
106 	/* final bytes */
107 poly1305_donna_atmost15bytes:
108 	if (!inlen) goto poly1305_donna_finish;
109 
110 	for (j = 0; j < inlen; j++) mp[j] = m[j];
111 	mp[j++] = 1;
112 	for (; j < 16; j++)	mp[j] = 0;
113 	inlen = 0;
114 
115 	t0 = U8TO32_LE(mp+0);
116 	t1 = U8TO32_LE(mp+4);
117 	t2 = U8TO32_LE(mp+8);
118 	t3 = U8TO32_LE(mp+12);
119 
120 	h0 += t0 & 0x3ffffff;
121 	h1 += ((((uint64_t)t1 << 32) | t0) >> 26) & 0x3ffffff;
122 	h2 += ((((uint64_t)t2 << 32) | t1) >> 20) & 0x3ffffff;
123 	h3 += ((((uint64_t)t3 << 32) | t2) >> 14) & 0x3ffffff;
124 	h4 += (t3 >> 8);
125 
126 	goto poly1305_donna_mul;
127 
128 poly1305_donna_finish:
129 	             b = h0 >> 26; h0 = h0 & 0x3ffffff;
130 	h1 +=     b; b = h1 >> 26; h1 = h1 & 0x3ffffff;
131 	h2 +=     b; b = h2 >> 26; h2 = h2 & 0x3ffffff;
132 	h3 +=     b; b = h3 >> 26; h3 = h3 & 0x3ffffff;
133 	h4 +=     b; b = h4 >> 26; h4 = h4 & 0x3ffffff;
134 	h0 += b * 5; b = h0 >> 26; h0 = h0 & 0x3ffffff;
135 	h1 +=     b;
136 
137 	g0 = h0 + 5; b = g0 >> 26; g0 &= 0x3ffffff;
138 	g1 = h1 + b; b = g1 >> 26; g1 &= 0x3ffffff;
139 	g2 = h2 + b; b = g2 >> 26; g2 &= 0x3ffffff;
140 	g3 = h3 + b; b = g3 >> 26; g3 &= 0x3ffffff;
141 	g4 = h4 + b - (1 << 26);
142 
143 	b = (g4 >> 31) - 1;
144 	nb = ~b;
145 	h0 = (h0 & nb) | (g0 & b);
146 	h1 = (h1 & nb) | (g1 & b);
147 	h2 = (h2 & nb) | (g2 & b);
148 	h3 = (h3 & nb) | (g3 & b);
149 	h4 = (h4 & nb) | (g4 & b);
150 
151 	f0 = ((h0      ) | (h1 << 26)) + (uint64_t)U8TO32_LE(&key[16]);
152 	f1 = ((h1 >>  6) | (h2 << 20)) + (uint64_t)U8TO32_LE(&key[20]);
153 	f2 = ((h2 >> 12) | (h3 << 14)) + (uint64_t)U8TO32_LE(&key[24]);
154 	f3 = ((h3 >> 18) | (h4 <<  8)) + (uint64_t)U8TO32_LE(&key[28]);
155 
156 	U32TO8_LE(&out[ 0], f0); f1 += (f0 >> 32);
157 	U32TO8_LE(&out[ 4], f1); f2 += (f1 >> 32);
158 	U32TO8_LE(&out[ 8], f2); f3 += (f2 >> 32);
159 	U32TO8_LE(&out[12], f3);
160 }
161