Annotation of src/usr.bin/ssh/poly1305.c, Revision 1.1
1.1 ! djm 1: /*
! 2: * Public Domain poly1305 from Andrew M.
! 3: * poly1305-donna-unrolled.c from https://github.com/floodyberry/poly1305-donna
! 4: */
! 5:
! 6: #include <sys/types.h>
! 7: #include <stdint.h>
! 8:
! 9: #include "poly1305.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
! 28: poly1305_auth(unsigned char out[POLY1305_TAGLEN], const unsigned char *m, size_t inlen, const unsigned char key[POLY1305_KEYLEN]) {
! 29: uint32_t t0,t1,t2,t3;
! 30: uint32_t h0,h1,h2,h3,h4;
! 31: uint32_t r0,r1,r2,r3,r4;
! 32: uint32_t s1,s2,s3,s4;
! 33: uint32_t b, nb;
! 34: size_t j;
! 35: uint64_t t[5];
! 36: uint64_t f0,f1,f2,f3;
! 37: uint32_t g0,g1,g2,g3,g4;
! 38: uint64_t c;
! 39: unsigned char mp[16];
! 40:
! 41: /* clamp key */
! 42: t0 = U8TO32_LE(key+0);
! 43: t1 = U8TO32_LE(key+4);
! 44: t2 = U8TO32_LE(key+8);
! 45: t3 = U8TO32_LE(key+12);
! 46:
! 47: /* precompute multipliers */
! 48: r0 = t0 & 0x3ffffff; t0 >>= 26; t0 |= t1 << 6;
! 49: r1 = t0 & 0x3ffff03; t1 >>= 20; t1 |= t2 << 12;
! 50: r2 = t1 & 0x3ffc0ff; t2 >>= 14; t2 |= t3 << 18;
! 51: r3 = t2 & 0x3f03fff; t3 >>= 8;
! 52: r4 = t3 & 0x00fffff;
! 53:
! 54: s1 = r1 * 5;
! 55: s2 = r2 * 5;
! 56: s3 = r3 * 5;
! 57: s4 = r4 * 5;
! 58:
! 59: /* init state */
! 60: h0 = 0;
! 61: h1 = 0;
! 62: h2 = 0;
! 63: h3 = 0;
! 64: h4 = 0;
! 65:
! 66: /* full blocks */
! 67: if (inlen < 16) goto poly1305_donna_atmost15bytes;
! 68: poly1305_donna_16bytes:
! 69: m += 16;
! 70: inlen -= 16;
! 71:
! 72: t0 = U8TO32_LE(m-16);
! 73: t1 = U8TO32_LE(m-12);
! 74: t2 = U8TO32_LE(m-8);
! 75: t3 = U8TO32_LE(m-4);
! 76:
! 77: h0 += t0 & 0x3ffffff;
! 78: h1 += ((((uint64_t)t1 << 32) | t0) >> 26) & 0x3ffffff;
! 79: h2 += ((((uint64_t)t2 << 32) | t1) >> 20) & 0x3ffffff;
! 80: h3 += ((((uint64_t)t3 << 32) | t2) >> 14) & 0x3ffffff;
! 81: h4 += (t3 >> 8) | (1 << 24);
! 82:
! 83:
! 84: poly1305_donna_mul:
! 85: t[0] = mul32x32_64(h0,r0) + mul32x32_64(h1,s4) + mul32x32_64(h2,s3) + mul32x32_64(h3,s2) + mul32x32_64(h4,s1);
! 86: t[1] = mul32x32_64(h0,r1) + mul32x32_64(h1,r0) + mul32x32_64(h2,s4) + mul32x32_64(h3,s3) + mul32x32_64(h4,s2);
! 87: t[2] = mul32x32_64(h0,r2) + mul32x32_64(h1,r1) + mul32x32_64(h2,r0) + mul32x32_64(h3,s4) + mul32x32_64(h4,s3);
! 88: t[3] = mul32x32_64(h0,r3) + mul32x32_64(h1,r2) + mul32x32_64(h2,r1) + mul32x32_64(h3,r0) + mul32x32_64(h4,s4);
! 89: t[4] = mul32x32_64(h0,r4) + mul32x32_64(h1,r3) + mul32x32_64(h2,r2) + mul32x32_64(h3,r1) + mul32x32_64(h4,r0);
! 90:
! 91: h0 = (uint32_t)t[0] & 0x3ffffff; c = (t[0] >> 26);
! 92: t[1] += c; h1 = (uint32_t)t[1] & 0x3ffffff; b = (uint32_t)(t[1] >> 26);
! 93: t[2] += b; h2 = (uint32_t)t[2] & 0x3ffffff; b = (uint32_t)(t[2] >> 26);
! 94: t[3] += b; h3 = (uint32_t)t[3] & 0x3ffffff; b = (uint32_t)(t[3] >> 26);
! 95: t[4] += b; h4 = (uint32_t)t[4] & 0x3ffffff; b = (uint32_t)(t[4] >> 26);
! 96: h0 += b * 5;
! 97:
! 98: if (inlen >= 16) goto poly1305_donna_16bytes;
! 99:
! 100: /* final bytes */
! 101: poly1305_donna_atmost15bytes:
! 102: if (!inlen) goto poly1305_donna_finish;
! 103:
! 104: for (j = 0; j < inlen; j++) mp[j] = m[j];
! 105: mp[j++] = 1;
! 106: for (; j < 16; j++) mp[j] = 0;
! 107: inlen = 0;
! 108:
! 109: t0 = U8TO32_LE(mp+0);
! 110: t1 = U8TO32_LE(mp+4);
! 111: t2 = U8TO32_LE(mp+8);
! 112: t3 = U8TO32_LE(mp+12);
! 113:
! 114: h0 += t0 & 0x3ffffff;
! 115: h1 += ((((uint64_t)t1 << 32) | t0) >> 26) & 0x3ffffff;
! 116: h2 += ((((uint64_t)t2 << 32) | t1) >> 20) & 0x3ffffff;
! 117: h3 += ((((uint64_t)t3 << 32) | t2) >> 14) & 0x3ffffff;
! 118: h4 += (t3 >> 8);
! 119:
! 120: goto poly1305_donna_mul;
! 121:
! 122: poly1305_donna_finish:
! 123: b = h0 >> 26; h0 = h0 & 0x3ffffff;
! 124: h1 += b; b = h1 >> 26; h1 = h1 & 0x3ffffff;
! 125: h2 += b; b = h2 >> 26; h2 = h2 & 0x3ffffff;
! 126: h3 += b; b = h3 >> 26; h3 = h3 & 0x3ffffff;
! 127: h4 += b; b = h4 >> 26; h4 = h4 & 0x3ffffff;
! 128: h0 += b * 5; b = h0 >> 26; h0 = h0 & 0x3ffffff;
! 129: h1 += b;
! 130:
! 131: g0 = h0 + 5; b = g0 >> 26; g0 &= 0x3ffffff;
! 132: g1 = h1 + b; b = g1 >> 26; g1 &= 0x3ffffff;
! 133: g2 = h2 + b; b = g2 >> 26; g2 &= 0x3ffffff;
! 134: g3 = h3 + b; b = g3 >> 26; g3 &= 0x3ffffff;
! 135: g4 = h4 + b - (1 << 26);
! 136:
! 137: b = (g4 >> 31) - 1;
! 138: nb = ~b;
! 139: h0 = (h0 & nb) | (g0 & b);
! 140: h1 = (h1 & nb) | (g1 & b);
! 141: h2 = (h2 & nb) | (g2 & b);
! 142: h3 = (h3 & nb) | (g3 & b);
! 143: h4 = (h4 & nb) | (g4 & b);
! 144:
! 145: f0 = ((h0 ) | (h1 << 26)) + (uint64_t)U8TO32_LE(&key[16]);
! 146: f1 = ((h1 >> 6) | (h2 << 20)) + (uint64_t)U8TO32_LE(&key[20]);
! 147: f2 = ((h2 >> 12) | (h3 << 14)) + (uint64_t)U8TO32_LE(&key[24]);
! 148: f3 = ((h3 >> 18) | (h4 << 8)) + (uint64_t)U8TO32_LE(&key[28]);
! 149:
! 150: U32TO8_LE(&out[ 0], f0); f1 += (f0 >> 32);
! 151: U32TO8_LE(&out[ 4], f1); f2 += (f1 >> 32);
! 152: U32TO8_LE(&out[ 8], f2); f3 += (f2 >> 32);
! 153: U32TO8_LE(&out[12], f3);
! 154: }