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., 675 Mass Ave, Cambridge, MA 02139, USA.
20  */
21 /*
22  * BEGIN_DESC
23  *
24  *  File:
25  *	@(#)	pa/spmath/sfsqrt.c		$Revision: $
26  *
27  *  Purpose:
28  *	Single Floating-point Square Root
29  *
30  *  External Interfaces:
31  *	sgl_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 "sgl_float.h"
44 
45 /*
46  *  Single Floating-point Square Root
47  */
48 
49 /*ARGSUSED*/
50 unsigned int
sgl_fsqrt(sgl_floating_point * srcptr,unsigned int * nullptr,sgl_floating_point * dstptr,unsigned int * status)51 sgl_fsqrt(
52     sgl_floating_point *srcptr,
53     unsigned int *nullptr,
54     sgl_floating_point *dstptr,
55     unsigned int *status)
56 {
57 	register unsigned int src, result;
58 	register int src_exponent;
59 	register unsigned int newbit, sum;
60 	register boolean guardbit = FALSE, even_exponent;
61 
62 	src = *srcptr;
63         /*
64          * check source operand for NaN or infinity
65          */
66         if ((src_exponent = Sgl_exponent(src)) == SGL_INFINITY_EXPONENT) {
67                 /*
68                  * is signaling NaN?
69                  */
70                 if (Sgl_isone_signaling(src)) {
71                         /* trap if INVALIDTRAP enabled */
72                         if (Is_invalidtrap_enabled()) return(INVALIDEXCEPTION);
73                         /* make NaN quiet */
74                         Set_invalidflag();
75                         Sgl_set_quiet(src);
76                 }
77                 /*
78                  * Return quiet NaN or positive infinity.
79 		 *  Fall thru to negative test if negative infinity.
80                  */
81 		if (Sgl_iszero_sign(src) || Sgl_isnotzero_mantissa(src)) {
82                 	*dstptr = src;
83                 	return(NOEXCEPTION);
84 		}
85         }
86 
87         /*
88          * check for zero source operand
89          */
90 	if (Sgl_iszero_exponentmantissa(src)) {
91 		*dstptr = src;
92 		return(NOEXCEPTION);
93 	}
94 
95         /*
96          * check for negative source operand
97          */
98 	if (Sgl_isone_sign(src)) {
99 		/* trap if INVALIDTRAP enabled */
100 		if (Is_invalidtrap_enabled()) return(INVALIDEXCEPTION);
101 		/* make NaN quiet */
102 		Set_invalidflag();
103 		Sgl_makequietnan(src);
104 		*dstptr = src;
105 		return(NOEXCEPTION);
106 	}
107 
108 	/*
109 	 * Generate result
110 	 */
111 	if (src_exponent > 0) {
112 		even_exponent = Sgl_hidden(src);
113 		Sgl_clear_signexponent_set_hidden(src);
114 	}
115 	else {
116 		/* normalize operand */
117 		Sgl_clear_signexponent(src);
118 		src_exponent++;
119 		Sgl_normalize(src,src_exponent);
120 		even_exponent = src_exponent & 1;
121 	}
122 	if (even_exponent) {
123 		/* exponent is even */
124 		/* Add comment here.  Explain why odd exponent needs correction */
125 		Sgl_leftshiftby1(src);
126 	}
127 	/*
128 	 * Add comment here.  Explain following algorithm.
129 	 *
130 	 * Trust me, it works.
131 	 *
132 	 */
133 	Sgl_setzero(result);
134 	newbit = 1 << SGL_P;
135 	while (newbit && Sgl_isnotzero(src)) {
136 		Sgl_addition(result,newbit,sum);
137 		if(sum <= Sgl_all(src)) {
138 			/* update result */
139 			Sgl_addition(result,(newbit<<1),result);
140 			Sgl_subtract(src,sum,src);
141 		}
142 		Sgl_rightshiftby1(newbit);
143 		Sgl_leftshiftby1(src);
144 	}
145 	/* correct exponent for pre-shift */
146 	if (even_exponent) {
147 		Sgl_rightshiftby1(result);
148 	}
149 
150 	/* check for inexact */
151 	if (Sgl_isnotzero(src)) {
152 		if (!even_exponent && Sgl_islessthan(result,src))
153 			Sgl_increment(result);
154 		guardbit = Sgl_lowmantissa(result);
155 		Sgl_rightshiftby1(result);
156 
157 		/*  now round result  */
158 		switch (Rounding_mode()) {
159 		case ROUNDPLUS:
160 		     Sgl_increment(result);
161 		     break;
162 		case ROUNDNEAREST:
163 		     /* stickybit is always true, so guardbit
164 		      * is enough to determine rounding */
165 		     if (guardbit) {
166 			Sgl_increment(result);
167 		     }
168 		     break;
169 		}
170 		/* increment result exponent by 1 if mantissa overflowed */
171 		if (Sgl_isone_hiddenoverflow(result)) src_exponent+=2;
172 
173 		if (Is_inexacttrap_enabled()) {
174 			Sgl_set_exponent(result,
175 			 ((src_exponent-SGL_BIAS)>>1)+SGL_BIAS);
176 			*dstptr = result;
177 			return(INEXACTEXCEPTION);
178 		}
179 		else Set_inexactflag();
180 	}
181 	else {
182 		Sgl_rightshiftby1(result);
183 	}
184 	Sgl_set_exponent(result,((src_exponent-SGL_BIAS)>>1)+SGL_BIAS);
185 	*dstptr = result;
186 	return(NOEXCEPTION);
187 }
188