_ Git - BearSSL/blob - src/int/i31_montmul.c

   1 /*
   2  * Copyright (c) 2016 Thomas Pornin <pornin@bolet.org>
   3  *
   4  * Permission is hereby granted, free of charge, to any person obtaining
   5  * a copy of this software and associated documentation files (the
   6  * "Software"), to deal in the Software without restriction, including
   7  * without limitation the rights to use, copy, modify, merge, publish,
   8  * distribute, sublicense, and/or sell copies of the Software, and to
   9  * permit persons to whom the Software is furnished to do so, subject to
  10  * the following conditions:
  11  *
  12  * The above copyright notice and this permission notice shall be
  13  * included in all copies or substantial portions of the Software.
  14  *
  15  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
  16  * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
  17  * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
  18  * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
  19  * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
  20  * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
  21  * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
  22  * SOFTWARE.
  23  */
  24
  25 #include "inner.h"
  26
  27 /* see inner.h */
  28 void
  29 br_i31_montymul(uint32_t *d, const uint32_t *x, const uint32_t *y,
  30         const uint32_t *m, uint32_t m0i)
  31 {
  32         size_t len, len4, u, v;
  33         uint64_t dh;
  34
  35         len = (m[0] + 31) >> 5;
  36         len4 = len & ~(size_t)3;
  37         br_i32_zero(d, m[0]);
  38         dh = 0;
  39         for (u = 0; u < len; u ++) {
  40                 uint32_t f, xu;
  41                 uint64_t r, zh;
  42
  43                 xu = x[u + 1];
  44                 f = ((d[1] + x[u + 1] * y[1]) * m0i) & 0x7FFFFFFF;
  45
  46                 r = 0;
  47                 for (v = 0; v < len4; v += 4) {
  48                         uint64_t z;
  49
  50                         z = (uint64_t)d[v + 1] + MUL31(xu, y[v + 1])
  51                                 + MUL31(f, m[v + 1]) + r;
  52                         r = z >> 31;
  53                         d[v + 0] = (uint32_t)z & 0x7FFFFFFF;
  54                         z = (uint64_t)d[v + 2] + MUL31(xu, y[v + 2])
  55                                 + MUL31(f, m[v + 2]) + r;
  56                         r = z >> 31;
  57                         d[v + 1] = (uint32_t)z & 0x7FFFFFFF;
  58                         z = (uint64_t)d[v + 3] + MUL31(xu, y[v + 3])
  59                                 + MUL31(f, m[v + 3]) + r;
  60                         r = z >> 31;
  61                         d[v + 2] = (uint32_t)z & 0x7FFFFFFF;
  62                         z = (uint64_t)d[v + 4] + MUL31(xu, y[v + 4])
  63                                 + MUL31(f, m[v + 4]) + r;
  64                         r = z >> 31;
  65                         d[v + 3] = (uint32_t)z & 0x7FFFFFFF;
  66                 }
  67                 for (; v < len; v ++) {
  68                         uint64_t z;
  69
  70                         z = (uint64_t)d[v + 1] + MUL31(xu, y[v + 1])
  71                                 + MUL31(f, m[v + 1]) + r;
  72                         r = z >> 31;
  73                         d[v] = (uint32_t)z & 0x7FFFFFFF;
  74                 }
  75
  76                 zh = dh + r;
  77                 d[len] = (uint32_t)zh & 0x7FFFFFFF;
  78                 dh = zh >> 31;
  79         }
  80
  81         /*
  82          * We must write back the bit length because it was overwritten in
  83          * the loop (not overwriting it would require a test in the loop,
  84          * which would yield bigger and slower code).
  85          */
  86         d[0] = m[0];
  87
  88         /*
  89          * d[] may still be greater than m[] at that point; notably, the
  90          * 'dh' word may be non-zero.
  91          */
  92         br_i31_sub(d, m, NEQ(dh, 0) | NOT(br_i31_sub(d, m, 0)));
  93 }