poly1305.cpp
1 // Copyright (c) 2019-present The Bitcoin Core developers 2 // Distributed under the MIT software license, see the accompanying 3 // file COPYING or http://www.opensource.org/licenses/mit-license.php. 4 5 #include <crypto/common.h> 6 #include <crypto/poly1305.h> 7 8 namespace poly1305_donna { 9 10 // Based on the public domain implementation by Andrew Moon 11 // poly1305-donna-32.h from https://github.com/floodyberry/poly1305-donna 12 13 void poly1305_init(poly1305_context *st, const unsigned char key[32]) noexcept { 14 /* r &= 0xffffffc0ffffffc0ffffffc0fffffff */ 15 st->r[0] = (ReadLE32(&key[ 0]) ) & 0x3ffffff; 16 st->r[1] = (ReadLE32(&key[ 3]) >> 2) & 0x3ffff03; 17 st->r[2] = (ReadLE32(&key[ 6]) >> 4) & 0x3ffc0ff; 18 st->r[3] = (ReadLE32(&key[ 9]) >> 6) & 0x3f03fff; 19 st->r[4] = (ReadLE32(&key[12]) >> 8) & 0x00fffff; 20 21 /* h = 0 */ 22 st->h[0] = 0; 23 st->h[1] = 0; 24 st->h[2] = 0; 25 st->h[3] = 0; 26 st->h[4] = 0; 27 28 /* save pad for later */ 29 st->pad[0] = ReadLE32(&key[16]); 30 st->pad[1] = ReadLE32(&key[20]); 31 st->pad[2] = ReadLE32(&key[24]); 32 st->pad[3] = ReadLE32(&key[28]); 33 34 st->leftover = 0; 35 st->final = 0; 36 } 37 38 static void poly1305_blocks(poly1305_context *st, const unsigned char *m, size_t bytes) noexcept { 39 const uint32_t hibit = (st->final) ? 0 : (1UL << 24); /* 1 << 128 */ 40 uint32_t r0,r1,r2,r3,r4; 41 uint32_t s1,s2,s3,s4; 42 uint32_t h0,h1,h2,h3,h4; 43 uint64_t d0,d1,d2,d3,d4; 44 uint32_t c; 45 46 r0 = st->r[0]; 47 r1 = st->r[1]; 48 r2 = st->r[2]; 49 r3 = st->r[3]; 50 r4 = st->r[4]; 51 52 s1 = r1 * 5; 53 s2 = r2 * 5; 54 s3 = r3 * 5; 55 s4 = r4 * 5; 56 57 h0 = st->h[0]; 58 h1 = st->h[1]; 59 h2 = st->h[2]; 60 h3 = st->h[3]; 61 h4 = st->h[4]; 62 63 while (bytes >= POLY1305_BLOCK_SIZE) { 64 /* h += m[i] */ 65 h0 += (ReadLE32(m+ 0) ) & 0x3ffffff; 66 h1 += (ReadLE32(m+ 3) >> 2) & 0x3ffffff; 67 h2 += (ReadLE32(m+ 6) >> 4) & 0x3ffffff; 68 h3 += (ReadLE32(m+ 9) >> 6) & 0x3ffffff; 69 h4 += (ReadLE32(m+12) >> 8) | hibit; 70 71 /* h *= r */ 72 d0 = ((uint64_t)h0 * r0) + ((uint64_t)h1 * s4) + ((uint64_t)h2 * s3) + ((uint64_t)h3 * s2) + ((uint64_t)h4 * s1); 73 d1 = ((uint64_t)h0 * r1) + ((uint64_t)h1 * r0) + ((uint64_t)h2 * s4) + ((uint64_t)h3 * s3) + ((uint64_t)h4 * s2); 74 d2 = ((uint64_t)h0 * r2) + ((uint64_t)h1 * r1) + ((uint64_t)h2 * r0) + ((uint64_t)h3 * s4) + ((uint64_t)h4 * s3); 75 d3 = ((uint64_t)h0 * r3) + ((uint64_t)h1 * r2) + ((uint64_t)h2 * r1) + ((uint64_t)h3 * r0) + ((uint64_t)h4 * s4); 76 d4 = ((uint64_t)h0 * r4) + ((uint64_t)h1 * r3) + ((uint64_t)h2 * r2) + ((uint64_t)h3 * r1) + ((uint64_t)h4 * r0); 77 78 /* (partial) h %= p */ 79 c = (uint32_t)(d0 >> 26); h0 = (uint32_t)d0 & 0x3ffffff; 80 d1 += c; c = (uint32_t)(d1 >> 26); h1 = (uint32_t)d1 & 0x3ffffff; 81 d2 += c; c = (uint32_t)(d2 >> 26); h2 = (uint32_t)d2 & 0x3ffffff; 82 d3 += c; c = (uint32_t)(d3 >> 26); h3 = (uint32_t)d3 & 0x3ffffff; 83 d4 += c; c = (uint32_t)(d4 >> 26); h4 = (uint32_t)d4 & 0x3ffffff; 84 h0 += c * 5; c = (h0 >> 26); h0 = h0 & 0x3ffffff; 85 h1 += c; 86 87 m += POLY1305_BLOCK_SIZE; 88 bytes -= POLY1305_BLOCK_SIZE; 89 } 90 91 st->h[0] = h0; 92 st->h[1] = h1; 93 st->h[2] = h2; 94 st->h[3] = h3; 95 st->h[4] = h4; 96 } 97 98 void poly1305_finish(poly1305_context *st, unsigned char mac[16]) noexcept { 99 uint32_t h0,h1,h2,h3,h4,c; 100 uint32_t g0,g1,g2,g3,g4; 101 uint64_t f; 102 uint32_t mask; 103 104 /* process the remaining block */ 105 if (st->leftover) { 106 size_t i = st->leftover; 107 st->buffer[i++] = 1; 108 for (; i < POLY1305_BLOCK_SIZE; i++) { 109 st->buffer[i] = 0; 110 } 111 st->final = 1; 112 poly1305_blocks(st, st->buffer, POLY1305_BLOCK_SIZE); 113 } 114 115 /* fully carry h */ 116 h0 = st->h[0]; 117 h1 = st->h[1]; 118 h2 = st->h[2]; 119 h3 = st->h[3]; 120 h4 = st->h[4]; 121 122 c = h1 >> 26; h1 = h1 & 0x3ffffff; 123 h2 += c; c = h2 >> 26; h2 = h2 & 0x3ffffff; 124 h3 += c; c = h3 >> 26; h3 = h3 & 0x3ffffff; 125 h4 += c; c = h4 >> 26; h4 = h4 & 0x3ffffff; 126 h0 += c * 5; c = h0 >> 26; h0 = h0 & 0x3ffffff; 127 h1 += c; 128 129 /* compute h + -p */ 130 g0 = h0 + 5; c = g0 >> 26; g0 &= 0x3ffffff; 131 g1 = h1 + c; c = g1 >> 26; g1 &= 0x3ffffff; 132 g2 = h2 + c; c = g2 >> 26; g2 &= 0x3ffffff; 133 g3 = h3 + c; c = g3 >> 26; g3 &= 0x3ffffff; 134 g4 = h4 + c - (1UL << 26); 135 136 /* select h if h < p, or h + -p if h >= p */ 137 mask = (g4 >> ((sizeof(uint32_t) * 8) - 1)) - 1; 138 g0 &= mask; 139 g1 &= mask; 140 g2 &= mask; 141 g3 &= mask; 142 g4 &= mask; 143 mask = ~mask; 144 h0 = (h0 & mask) | g0; 145 h1 = (h1 & mask) | g1; 146 h2 = (h2 & mask) | g2; 147 h3 = (h3 & mask) | g3; 148 h4 = (h4 & mask) | g4; 149 150 /* h = h % (2^128) */ 151 h0 = ((h0 ) | (h1 << 26)) & 0xffffffff; 152 h1 = ((h1 >> 6) | (h2 << 20)) & 0xffffffff; 153 h2 = ((h2 >> 12) | (h3 << 14)) & 0xffffffff; 154 h3 = ((h3 >> 18) | (h4 << 8)) & 0xffffffff; 155 156 /* mac = (h + pad) % (2^128) */ 157 f = (uint64_t)h0 + st->pad[0] ; h0 = (uint32_t)f; 158 f = (uint64_t)h1 + st->pad[1] + (f >> 32); h1 = (uint32_t)f; 159 f = (uint64_t)h2 + st->pad[2] + (f >> 32); h2 = (uint32_t)f; 160 f = (uint64_t)h3 + st->pad[3] + (f >> 32); h3 = (uint32_t)f; 161 162 WriteLE32(mac + 0, h0); 163 WriteLE32(mac + 4, h1); 164 WriteLE32(mac + 8, h2); 165 WriteLE32(mac + 12, h3); 166 167 /* zero out the state */ 168 st->h[0] = 0; 169 st->h[1] = 0; 170 st->h[2] = 0; 171 st->h[3] = 0; 172 st->h[4] = 0; 173 st->r[0] = 0; 174 st->r[1] = 0; 175 st->r[2] = 0; 176 st->r[3] = 0; 177 st->r[4] = 0; 178 st->pad[0] = 0; 179 st->pad[1] = 0; 180 st->pad[2] = 0; 181 st->pad[3] = 0; 182 } 183 184 void poly1305_update(poly1305_context *st, const unsigned char *m, size_t bytes) noexcept { 185 size_t i; 186 187 /* handle leftover */ 188 if (st->leftover) { 189 size_t want = (POLY1305_BLOCK_SIZE - st->leftover); 190 if (want > bytes) { 191 want = bytes; 192 } 193 for (i = 0; i < want; i++) { 194 st->buffer[st->leftover + i] = m[i]; 195 } 196 bytes -= want; 197 m += want; 198 st->leftover += want; 199 if (st->leftover < POLY1305_BLOCK_SIZE) return; 200 poly1305_blocks(st, st->buffer, POLY1305_BLOCK_SIZE); 201 st->leftover = 0; 202 } 203 204 /* process full blocks */ 205 if (bytes >= POLY1305_BLOCK_SIZE) { 206 size_t want = (bytes & ~(POLY1305_BLOCK_SIZE - 1)); 207 poly1305_blocks(st, m, want); 208 m += want; 209 bytes -= want; 210 } 211 212 /* store leftover */ 213 if (bytes) { 214 for (i = 0; i < bytes; i++) { 215 st->buffer[st->leftover + i] = m[i]; 216 } 217 st->leftover += bytes; 218 } 219 } 220 221 } // namespace poly1305_donna