| Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 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. | 
| Simon Arlott | 7022672 | 2007-05-11 20:42:34 +0100 | [diff] [blame] | 79 | *  Fall through to negative test if negative infinity. | 
| Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 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 | } |