Home | History | Annotate | Line # | Download | only in generic
      1 /* mpn_remove -- divide out all multiples of odd mpn number from another mpn
      2    number.
      3 
      4    Contributed to the GNU project by Torbjorn Granlund.
      5 
      6    THE FUNCTION IN THIS FILE IS INTERNAL WITH A MUTABLE INTERFACE.  IT IS ONLY
      7    SAFE TO REACH IT THROUGH DOCUMENTED INTERFACES.  IN FACT, IT IS ALMOST
      8    GUARANTEED THAT IT WILL CHANGE OR DISAPPEAR IN A FUTURE GMP RELEASE.
      9 
     10 Copyright 2009, 2012-2014, 2017 Free Software Foundation, Inc.
     11 
     12 This file is part of the GNU MP Library.
     13 
     14 The GNU MP Library is free software; you can redistribute it and/or modify
     15 it under the terms of either:
     16 
     17   * the GNU Lesser General Public License as published by the Free
     18     Software Foundation; either version 3 of the License, or (at your
     19     option) any later version.
     20 
     21 or
     22 
     23   * the GNU General Public License as published by the Free Software
     24     Foundation; either version 2 of the License, or (at your option) any
     25     later version.
     26 
     27 or both in parallel, as here.
     28 
     29 The GNU MP Library is distributed in the hope that it will be useful, but
     30 WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
     31 or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
     32 for more details.
     33 
     34 You should have received copies of the GNU General Public License and the
     35 GNU Lesser General Public License along with the GNU MP Library.  If not,
     36 see https://www.gnu.org/licenses/.  */
     37 
     38 #include "gmp-impl.h"
     39 
     40 #if GMP_LIMB_BITS > 50
     41 #define LOG 50
     42 #else
     43 #define LOG GMP_LIMB_BITS
     44 #endif
     45 
     46 
     47 /* Input: U = {up,un}, V = {vp,vn} must be odd, cap
     48    Ouput  W = {wp,*wn} allocation need is exactly *wn
     49 
     50    Set W = U / V^k, where k is the largest integer <= cap such that the
     51    division yields an integer.
     52 
     53    FIXME: We currently allow any operand overlap.  This is quite non mpn-ish
     54    and might be changed, since it cost significant temporary space.
     55    * If we require W to have space for un + 1 limbs, we could save qp or qp2
     56      (but we will still need to copy things into wp 50% of the time).
     57    * If we allow ourselves to clobber U, we could save the other of qp and qp2,
     58      and the initial COPY (but also here we would need un + 1 limbs).
     59 */
     60 
     61 /* FIXME: We need to wrap mpn_bdiv_qr due to the itch interface.  This need
     62    indicates a flaw in the current itch mechanism: Which operands not greater
     63    than un,un will incur the worst itch?  We need a parallel foo_maxitch set
     64    of functions.  */
     65 static void
     66 mpn_bdiv_qr_wrap (mp_ptr qp, mp_ptr rp,
     67 		  mp_srcptr np, mp_size_t nn,
     68 		  mp_srcptr dp, mp_size_t dn)
     69 {
     70   mp_ptr scratch_out;
     71   TMP_DECL;
     72 
     73   TMP_MARK;
     74   scratch_out = TMP_ALLOC_LIMBS (mpn_bdiv_qr_itch (nn, dn));
     75   mpn_bdiv_qr (qp, rp, np, nn, dp, dn, scratch_out);
     76 
     77   TMP_FREE;
     78 }
     79 
     80 mp_bitcnt_t
     81 mpn_remove (mp_ptr wp, mp_size_t *wn,
     82 	    mp_srcptr up, mp_size_t un, mp_srcptr vp, mp_size_t vn,
     83 	    mp_bitcnt_t cap)
     84 {
     85   mp_srcptr pwpsp[LOG];
     86   mp_size_t pwpsn[LOG];
     87   mp_size_t npowers;
     88   mp_ptr tp, qp, np, qp2;
     89   mp_srcptr pp;
     90   mp_size_t pn, nn, qn, i;
     91   mp_bitcnt_t pwr;
     92   TMP_DECL;
     93 
     94   ASSERT (un > 0);
     95   ASSERT (vn > 0);
     96   ASSERT (vp[0] % 2 != 0);	/* 2-adic division wants odd numbers */
     97   ASSERT (vn > 1 || vp[0] > 1);	/* else we would loop indefinitely */
     98 
     99   TMP_MARK;
    100 
    101   TMP_ALLOC_LIMBS_3 (qp, un + 1,	/* quotient, alternating */
    102 		     qp2, un + 1,	/* quotient, alternating */
    103 		     tp, (un + 1 + vn) / 2); /* remainder */
    104   pp = vp;
    105   pn = vn;
    106 
    107   MPN_COPY (qp, up, un);
    108   qn = un;
    109 
    110   npowers = 0;
    111   while (qn >= pn)
    112     {
    113       qp[qn] = 0;
    114       mpn_bdiv_qr_wrap (qp2, tp, qp, qn + 1, pp, pn);
    115       if (!mpn_zero_p (tp, pn))
    116 	{
    117 	  if (mpn_cmp (tp, pp, pn) != 0)
    118 	    break;		/* could not divide by V^npowers */
    119 	}
    120 
    121       MP_PTR_SWAP (qp, qp2);
    122       qn = qn - pn;
    123       mpn_neg (qp, qp, qn+1);
    124 
    125       qn += qp[qn] != 0;
    126 
    127       pwpsp[npowers] = pp;
    128       pwpsn[npowers] = pn;
    129       ++npowers;
    130 
    131       if (((mp_bitcnt_t) 2 << npowers) - 1 > cap)
    132 	break;
    133 
    134       nn = 2 * pn - 1;		/* next power will be at least this large */
    135       if (nn > qn)
    136 	break;			/* next power would be overlarge */
    137 
    138       if (npowers == 1)		/* Alloc once, but only if it's needed */
    139 	np = TMP_ALLOC_LIMBS (qn + LOG);	/* powers of V */
    140       else
    141 	np += pn;
    142 
    143       mpn_sqr (np, pp, pn);
    144       pn = nn + (np[nn] != 0);
    145       pp = np;
    146     }
    147 
    148   pwr = ((mp_bitcnt_t) 1 << npowers) - 1;
    149 
    150   for (i = npowers; --i >= 0;)
    151     {
    152       pn = pwpsn[i];
    153       if (qn < pn)
    154 	continue;
    155 
    156       if (pwr + ((mp_bitcnt_t) 1 << i) > cap)
    157 	continue;		/* V^i would bring us past cap */
    158 
    159       qp[qn] = 0;
    160       mpn_bdiv_qr_wrap (qp2, tp, qp, qn + 1, pwpsp[i], pn);
    161       if (!mpn_zero_p (tp, pn))
    162 	{
    163 	  if (mpn_cmp (tp, pwpsp[i], pn) != 0)
    164 	    continue;		/* could not divide by V^i */
    165 	}
    166 
    167       MP_PTR_SWAP (qp, qp2);
    168       qn = qn - pn;
    169       mpn_neg (qp, qp, qn+1);
    170 
    171       qn += qp[qn] != 0;
    172 
    173       pwr += (mp_bitcnt_t) 1 << i;
    174     }
    175 
    176   MPN_COPY (wp, qp, qn);
    177   *wn = qn;
    178 
    179   TMP_FREE;
    180 
    181   return pwr;
    182 }
    183