source/libs/gmp/gmp-src/mpn/generic/redc_2.c

   1 /* mpn_redc_2.  Set rp[] <- up[]/R^n mod mp[].  Clobber up[].
   2    mp[] is n limbs; up[] is 2n limbs.
   3
   4    THIS IS AN INTERNAL FUNCTION WITH A MUTABLE INTERFACE.  IT IS ONLY
   5    SAFE TO REACH THIS FUNCTION THROUGH DOCUMENTED INTERFACES.
   6
   7 Copyright (C) 2000-2002, 2004, 2008, 2012 Free Software Foundation, Inc.
   8
   9 This file is part of the GNU MP Library.
  10
  11 The GNU MP Library is free software; you can redistribute it and/or modify
  12 it under the terms of either:
  13
  14   * the GNU Lesser General Public License as published by the Free
  15     Software Foundation; either version 3 of the License, or (at your
  16     option) any later version.
  17
  18 or
  19
  20   * the GNU General Public License as published by the Free Software
  21     Foundation; either version 2 of the License, or (at your option) any
  22     later version.
  23
  24 or both in parallel, as here.
  25
  26 The GNU MP Library is distributed in the hope that it will be useful, but
  27 WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
  28 or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  29 for more details.
  30
  31 You should have received copies of the GNU General Public License and the
  32 GNU Lesser General Public License along with the GNU MP Library.  If not,
  33 see https://www.gnu.org/licenses/.  */
  34
  35 #include "gmp.h"
  36 #include "gmp-impl.h"
  37 #include "longlong.h"
  38
  39
  40 #if GMP_NAIL_BITS != 0
  41 you lose
  42 #endif
  43
  44 /* For testing purposes, define our own mpn_addmul_2 if there is none already
  45    available.  */
  46 #ifndef HAVE_NATIVE_mpn_addmul_2
  47 #undef mpn_addmul_2
  48 static mp_limb_t
  49 mpn_addmul_2 (mp_ptr rp, mp_srcptr up, mp_size_t n, mp_srcptr vp)
  50 {
  51   rp[n] = mpn_addmul_1 (rp, up, n, vp[0]);
  52   return mpn_addmul_1 (rp + 1, up, n, vp[1]);
  53 }
  54 #endif
  55
  56 #if defined (__GNUC__) && ! defined (NO_ASM) \
  57   && defined (__ia64) && W_TYPE_SIZE == 64
  58 #define umul2low(ph, pl, uh, ul, vh, vl) \
  59   do {                                                                  \
  60     mp_limb_t _ph, _pl;                                                 \
  61     __asm__ ("xma.hu %0 = %3, %5, f0\n\t"                               \
  62              "xma.l %1 = %3, %5, f0\n\t"                                \
  63              ";;\n\t"                                                   \
  64              "xma.l %0 = %3, %4, %0\n\t"                                \
  65              ";;\n\t"                                                   \
  66              "xma.l %0 = %2, %5, %0"                                    \
  67              : "=&f" (ph), "=&f" (pl)                                   \
  68              : "f" (uh), "f" (ul), "f" (vh), "f" (vl));                 \
  69   } while (0)
  70 #endif
  71
  72 #ifndef umul2low
  73 #define umul2low(ph, pl, uh, ul, vh, vl) \
  74   do {                                                                  \
  75     mp_limb_t _ph, _pl;                                                 \
  76     umul_ppmm (_ph, _pl, ul, vl);                                       \
  77     (ph) = _ph + (ul) * (vh) + (uh) * (vl);                             \
  78     (pl) = _pl;                                                         \
  79   } while (0)
  80 #endif
  81
  82 mp_limb_t
  83 mpn_redc_2 (mp_ptr rp, mp_ptr up, mp_srcptr mp, mp_size_t n, mp_srcptr mip)
  84 {
  85   mp_limb_t q[2];
  86   mp_size_t j;
  87   mp_limb_t upn;
  88   mp_limb_t cy;
  89
  90   ASSERT (n > 0);
  91   ASSERT_MPN (up, 2*n);
  92
  93   if ((n & 1) != 0)
  94     {
  95       up[0] = mpn_addmul_1 (up, mp, n, (up[0] * mip[0]) & GMP_NUMB_MASK);
  96       up++;
  97     }
  98
  99   for (j = n - 2; j >= 0; j -= 2)
 100     {
 101       umul2low (q[1], q[0], mip[1], mip[0], up[1], up[0]);
 102       upn = up[n];              /* mpn_addmul_2 overwrites this */
 103       up[1] = mpn_addmul_2 (up, mp, n, q);
 104       up[0] = up[n];
 105       up[n] = upn;
 106       up += 2;
 107     }
 108
 109   cy = mpn_add_n (rp, up, up - n, n);
 110   return cy;
 111 }