1 /* $OpenBSD: poly1305.c,v 1.4 2023/07/17 05:26:38 djm Exp $ */ 2 /* 3 * Public Domain poly1305 from Andrew Moon 4 * poly1305-donna-unrolled.c from https://github.com/floodyberry/poly1305-donna 5 */ 6 7 #include <sys/types.h> 8 #include <stdint.h> 9 10 #include "poly1305.h" 11 12 #define mul32x32_64(a,b) ((uint64_t)(a) * (b)) 13 14 #define U8TO32_LE(p) \ 15 (((uint32_t)((p)[0])) | \ 16 ((uint32_t)((p)[1]) << 8) | \ 17 ((uint32_t)((p)[2]) << 16) | \ 18 ((uint32_t)((p)[3]) << 24)) 19 20 #define U32TO8_LE(p, v) \ 21 do { \ 22 (p)[0] = (uint8_t)((v)); \ 23 (p)[1] = (uint8_t)((v) >> 8); \ 24 (p)[2] = (uint8_t)((v) >> 16); \ 25 (p)[3] = (uint8_t)((v) >> 24); \ 26 } while (0) 27 28 void 29 poly1305_auth(unsigned char out[POLY1305_TAGLEN], const unsigned char *m, size_t inlen, const unsigned char key[POLY1305_KEYLEN]) { 30 uint32_t t0,t1,t2,t3; 31 uint32_t h0,h1,h2,h3,h4; 32 uint32_t r0,r1,r2,r3,r4; 33 uint32_t s1,s2,s3,s4; 34 uint32_t b, nb; 35 size_t j; 36 uint64_t t[5]; 37 uint64_t f0,f1,f2,f3; 38 uint32_t g0,g1,g2,g3,g4; 39 uint64_t c; 40 unsigned char mp[16]; 41 42 /* clamp key */ 43 t0 = U8TO32_LE(key+0); 44 t1 = U8TO32_LE(key+4); 45 t2 = U8TO32_LE(key+8); 46 t3 = U8TO32_LE(key+12); 47 48 /* precompute multipliers */ 49 r0 = t0 & 0x3ffffff; t0 >>= 26; t0 |= t1 << 6; 50 r1 = t0 & 0x3ffff03; t1 >>= 20; t1 |= t2 << 12; 51 r2 = t1 & 0x3ffc0ff; t2 >>= 14; t2 |= t3 << 18; 52 r3 = t2 & 0x3f03fff; t3 >>= 8; 53 r4 = t3 & 0x00fffff; 54 55 s1 = r1 * 5; 56 s2 = r2 * 5; 57 s3 = r3 * 5; 58 s4 = r4 * 5; 59 60 /* init state */ 61 h0 = 0; 62 h1 = 0; 63 h2 = 0; 64 h3 = 0; 65 h4 = 0; 66 67 /* full blocks */ 68 if (inlen < 16) goto poly1305_donna_atmost15bytes; 69 poly1305_donna_16bytes: 70 m += 16; 71 inlen -= 16; 72 73 t0 = U8TO32_LE(m-16); 74 t1 = U8TO32_LE(m-12); 75 t2 = U8TO32_LE(m-8); 76 t3 = U8TO32_LE(m-4); 77 78 h0 += t0 & 0x3ffffff; 79 h1 += ((((uint64_t)t1 << 32) | t0) >> 26) & 0x3ffffff; 80 h2 += ((((uint64_t)t2 << 32) | t1) >> 20) & 0x3ffffff; 81 h3 += ((((uint64_t)t3 << 32) | t2) >> 14) & 0x3ffffff; 82 h4 += (t3 >> 8) | (1 << 24); 83 84 85 poly1305_donna_mul: 86 t[0] = mul32x32_64(h0,r0) + mul32x32_64(h1,s4) + mul32x32_64(h2,s3) + mul32x32_64(h3,s2) + mul32x32_64(h4,s1); 87 t[1] = mul32x32_64(h0,r1) + mul32x32_64(h1,r0) + mul32x32_64(h2,s4) + mul32x32_64(h3,s3) + mul32x32_64(h4,s2); 88 t[2] = mul32x32_64(h0,r2) + mul32x32_64(h1,r1) + mul32x32_64(h2,r0) + mul32x32_64(h3,s4) + mul32x32_64(h4,s3); 89 t[3] = mul32x32_64(h0,r3) + mul32x32_64(h1,r2) + mul32x32_64(h2,r1) + mul32x32_64(h3,r0) + mul32x32_64(h4,s4); 90 t[4] = mul32x32_64(h0,r4) + mul32x32_64(h1,r3) + mul32x32_64(h2,r2) + mul32x32_64(h3,r1) + mul32x32_64(h4,r0); 91 92 h0 = (uint32_t)t[0] & 0x3ffffff; c = (t[0] >> 26); 93 t[1] += c; h1 = (uint32_t)t[1] & 0x3ffffff; b = (uint32_t)(t[1] >> 26); 94 t[2] += b; h2 = (uint32_t)t[2] & 0x3ffffff; b = (uint32_t)(t[2] >> 26); 95 t[3] += b; h3 = (uint32_t)t[3] & 0x3ffffff; b = (uint32_t)(t[3] >> 26); 96 t[4] += b; h4 = (uint32_t)t[4] & 0x3ffffff; b = (uint32_t)(t[4] >> 26); 97 h0 += b * 5; 98 99 if (inlen >= 16) goto poly1305_donna_16bytes; 100 101 /* final bytes */ 102 poly1305_donna_atmost15bytes: 103 if (!inlen) goto poly1305_donna_finish; 104 105 for (j = 0; j < inlen; j++) mp[j] = m[j]; 106 mp[j++] = 1; 107 for (; j < 16; j++) mp[j] = 0; 108 inlen = 0; 109 110 t0 = U8TO32_LE(mp+0); 111 t1 = U8TO32_LE(mp+4); 112 t2 = U8TO32_LE(mp+8); 113 t3 = U8TO32_LE(mp+12); 114 115 h0 += t0 & 0x3ffffff; 116 h1 += ((((uint64_t)t1 << 32) | t0) >> 26) & 0x3ffffff; 117 h2 += ((((uint64_t)t2 << 32) | t1) >> 20) & 0x3ffffff; 118 h3 += ((((uint64_t)t3 << 32) | t2) >> 14) & 0x3ffffff; 119 h4 += (t3 >> 8); 120 121 goto poly1305_donna_mul; 122 123 poly1305_donna_finish: 124 b = h0 >> 26; h0 = h0 & 0x3ffffff; 125 h1 += b; b = h1 >> 26; h1 = h1 & 0x3ffffff; 126 h2 += b; b = h2 >> 26; h2 = h2 & 0x3ffffff; 127 h3 += b; b = h3 >> 26; h3 = h3 & 0x3ffffff; 128 h4 += b; b = h4 >> 26; h4 = h4 & 0x3ffffff; 129 h0 += b * 5; b = h0 >> 26; h0 = h0 & 0x3ffffff; 130 h1 += b; 131 132 g0 = h0 + 5; b = g0 >> 26; g0 &= 0x3ffffff; 133 g1 = h1 + b; b = g1 >> 26; g1 &= 0x3ffffff; 134 g2 = h2 + b; b = g2 >> 26; g2 &= 0x3ffffff; 135 g3 = h3 + b; b = g3 >> 26; g3 &= 0x3ffffff; 136 g4 = h4 + b - (1 << 26); 137 138 b = (g4 >> 31) - 1; 139 nb = ~b; 140 h0 = (h0 & nb) | (g0 & b); 141 h1 = (h1 & nb) | (g1 & b); 142 h2 = (h2 & nb) | (g2 & b); 143 h3 = (h3 & nb) | (g3 & b); 144 h4 = (h4 & nb) | (g4 & b); 145 146 f0 = ((h0 ) | (h1 << 26)) + (uint64_t)U8TO32_LE(&key[16]); 147 f1 = ((h1 >> 6) | (h2 << 20)) + (uint64_t)U8TO32_LE(&key[20]); 148 f2 = ((h2 >> 12) | (h3 << 14)) + (uint64_t)U8TO32_LE(&key[24]); 149 f3 = ((h3 >> 18) | (h4 << 8)) + (uint64_t)U8TO32_LE(&key[28]); 150 151 U32TO8_LE(&out[ 0], f0); f1 += (f0 >> 32); 152 U32TO8_LE(&out[ 4], f1); f2 += (f1 >> 32); 153 U32TO8_LE(&out[ 8], f2); f3 += (f2 >> 32); 154 U32TO8_LE(&out[12], f3); 155 } 156