git.oblomov.eu Git - linux-2.6/blob - arch/parisc/math-emu/dfsqrt.c

   1 /*
   2  * Linux/PA-RISC Project (http://www.parisc-linux.org/)
   3  *
   4  * Floating-point emulation code
   5  *  Copyright (C) 2001 Hewlett-Packard (Paul Bame) <bame@debian.org>
   6  *
   7  *    This program is free software; you can redistribute it and/or modify
   8  *    it under the terms of the GNU General Public License as published by
   9  *    the Free Software Foundation; either version 2, or (at your option)
  10  *    any later version.
  11  *
  12  *    This program is distributed in the hope that it will be useful,
  13  *    but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  *    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15  *    GNU General Public License for more details.
  16  *
  17  *    You should have received a copy of the GNU General Public License
  18  *    along with this program; if not, write to the Free Software
  19  *    Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  20  */
  21 /*
  22  * BEGIN_DESC
  23  *
  24  *  File:
  25  *      @(#)    pa/spmath/dfsqrt.c              $Revision: 1.1 $
  26  *
  27  *  Purpose:
  28  *      Double Floating-point Square Root
  29  *
  30  *  External Interfaces:
  31  *      dbl_fsqrt(srcptr,nullptr,dstptr,status)
  32  *
  33  *  Internal Interfaces:
  34  *
  35  *  Theory:
  36  *      <<please update with a overview of the operation of this file>>
  37  *
  38  * END_DESC
  39 */
  40
  41
  42 #include "float.h"
  43 #include "dbl_float.h"
  44
  45 /*
  46  *  Double Floating-point Square Root
  47  */
  48
  49 /*ARGSUSED*/
  50 unsigned int
  51 dbl_fsqrt(
  52             dbl_floating_point *srcptr,
  53             unsigned int *nullptr,
  54             dbl_floating_point *dstptr,
  55             unsigned int *status)
  56 {
  57         register unsigned int srcp1, srcp2, resultp1, resultp2;
  58         register unsigned int newbitp1, newbitp2, sump1, sump2;
  59         register int src_exponent;
  60         register boolean guardbit = FALSE, even_exponent;
  61
  62         Dbl_copyfromptr(srcptr,srcp1,srcp2);
  63         /*
  64          * check source operand for NaN or infinity
  65          */
  66         if ((src_exponent = Dbl_exponent(srcp1)) == DBL_INFINITY_EXPONENT) {
  67                 /*
  68                  * is signaling NaN?
  69                  */
  70                 if (Dbl_isone_signaling(srcp1)) {
  71                         /* trap if INVALIDTRAP enabled */
  72                         if (Is_invalidtrap_enabled()) return(INVALIDEXCEPTION);
  73                         /* make NaN quiet */
  74                         Set_invalidflag();
  75                         Dbl_set_quiet(srcp1);
  76                 }
  77                 /*
  78                  * Return quiet NaN or positive infinity.
  79                  *  Fall through to negative test if negative infinity.
  80                  */
  81                 if (Dbl_iszero_sign(srcp1) ||
  82                     Dbl_isnotzero_mantissa(srcp1,srcp2)) {
  83                         Dbl_copytoptr(srcp1,srcp2,dstptr);
  84                         return(NOEXCEPTION);
  85                 }
  86         }
  87
  88         /*
  89          * check for zero source operand
  90          */
  91         if (Dbl_iszero_exponentmantissa(srcp1,srcp2)) {
  92                 Dbl_copytoptr(srcp1,srcp2,dstptr);
  93                 return(NOEXCEPTION);
  94         }
  95
  96         /*
  97          * check for negative source operand
  98          */
  99         if (Dbl_isone_sign(srcp1)) {
 100                 /* trap if INVALIDTRAP enabled */
 101                 if (Is_invalidtrap_enabled()) return(INVALIDEXCEPTION);
 102                 /* make NaN quiet */
 103                 Set_invalidflag();
 104                 Dbl_makequietnan(srcp1,srcp2);
 105                 Dbl_copytoptr(srcp1,srcp2,dstptr);
 106                 return(NOEXCEPTION);
 107         }
 108
 109         /*
 110          * Generate result
 111          */
 112         if (src_exponent > 0) {
 113                 even_exponent = Dbl_hidden(srcp1);
 114                 Dbl_clear_signexponent_set_hidden(srcp1);
 115         }
 116         else {
 117                 /* normalize operand */
 118                 Dbl_clear_signexponent(srcp1);
 119                 src_exponent++;
 120                 Dbl_normalize(srcp1,srcp2,src_exponent);
 121                 even_exponent = src_exponent & 1;
 122         }
 123         if (even_exponent) {
 124                 /* exponent is even */
 125                 /* Add comment here.  Explain why odd exponent needs correction */
 126                 Dbl_leftshiftby1(srcp1,srcp2);
 127         }
 128         /*
 129          * Add comment here.  Explain following algorithm.
 130          *
 131          * Trust me, it works.
 132          *
 133          */
 134         Dbl_setzero(resultp1,resultp2);
 135         Dbl_allp1(newbitp1) = 1 << (DBL_P - 32);
 136         Dbl_setzero_mantissap2(newbitp2);
 137         while (Dbl_isnotzero(newbitp1,newbitp2) && Dbl_isnotzero(srcp1,srcp2)) {
 138                 Dbl_addition(resultp1,resultp2,newbitp1,newbitp2,sump1,sump2);
 139                 if(Dbl_isnotgreaterthan(sump1,sump2,srcp1,srcp2)) {
 140                         Dbl_leftshiftby1(newbitp1,newbitp2);
 141                         /* update result */
 142                         Dbl_addition(resultp1,resultp2,newbitp1,newbitp2,
 143                          resultp1,resultp2);
 144                         Dbl_subtract(srcp1,srcp2,sump1,sump2,srcp1,srcp2);
 145                         Dbl_rightshiftby2(newbitp1,newbitp2);
 146                 }
 147                 else {
 148                         Dbl_rightshiftby1(newbitp1,newbitp2);
 149                 }
 150                 Dbl_leftshiftby1(srcp1,srcp2);
 151         }
 152         /* correct exponent for pre-shift */
 153         if (even_exponent) {
 154                 Dbl_rightshiftby1(resultp1,resultp2);
 155         }
 156
 157         /* check for inexact */
 158         if (Dbl_isnotzero(srcp1,srcp2)) {
 159                 if (!even_exponent && Dbl_islessthan(resultp1,resultp2,srcp1,srcp2)) {
 160                         Dbl_increment(resultp1,resultp2);
 161                 }
 162                 guardbit = Dbl_lowmantissap2(resultp2);
 163                 Dbl_rightshiftby1(resultp1,resultp2);
 164
 165                 /*  now round result  */
 166                 switch (Rounding_mode()) {
 167                 case ROUNDPLUS:
 168                      Dbl_increment(resultp1,resultp2);
 169                      break;
 170                 case ROUNDNEAREST:
 171                      /* stickybit is always true, so guardbit
 172                       * is enough to determine rounding */
 173                      if (guardbit) {
 174                             Dbl_increment(resultp1,resultp2);
 175                      }
 176                      break;
 177                 }
 178                 /* increment result exponent by 1 if mantissa overflowed */
 179                 if (Dbl_isone_hiddenoverflow(resultp1)) src_exponent+=2;
 180
 181                 if (Is_inexacttrap_enabled()) {
 182                         Dbl_set_exponent(resultp1,
 183                          ((src_exponent-DBL_BIAS)>>1)+DBL_BIAS);
 184                         Dbl_copytoptr(resultp1,resultp2,dstptr);
 185                         return(INEXACTEXCEPTION);
 186                 }
 187                 else Set_inexactflag();
 188         }
 189         else {
 190                 Dbl_rightshiftby1(resultp1,resultp2);
 191         }
 192         Dbl_set_exponent(resultp1,((src_exponent-DBL_BIAS)>>1)+DBL_BIAS);
 193         Dbl_copytoptr(resultp1,resultp2,dstptr);
 194         return(NOEXCEPTION);
 195 }