xref: /netbsd/sys/arch/hppa/spmath/dfrem.c (revision c4a72b64)
1 /*	$NetBSD: dfrem.c,v 1.1 2002/06/05 01:04:24 fredette Exp $	*/
2 
3 /*	$OpenBSD: dfrem.c,v 1.4 2001/03/29 03:58:17 mickey Exp $	*/
4 
5 /*
6  * Copyright 1996 1995 by Open Software Foundation, Inc.
7  *              All Rights Reserved
8  *
9  * Permission to use, copy, modify, and distribute this software and
10  * its documentation for any purpose and without fee is hereby granted,
11  * provided that the above copyright notice appears in all copies and
12  * that both the copyright notice and this permission notice appear in
13  * supporting documentation.
14  *
15  * OSF DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE
16  * INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
17  * FOR A PARTICULAR PURPOSE.
18  *
19  * IN NO EVENT SHALL OSF BE LIABLE FOR ANY SPECIAL, INDIRECT, OR
20  * CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM
21  * LOSS OF USE, DATA OR PROFITS, WHETHER IN ACTION OF CONTRACT,
22  * NEGLIGENCE, OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION
23  * WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
24  *
25  */
26 /*
27  * pmk1.1
28  */
29 /*
30  * (c) Copyright 1986 HEWLETT-PACKARD COMPANY
31  *
32  * To anyone who acknowledges that this file is provided "AS IS"
33  * without any express or implied warranty:
34  *     permission to use, copy, modify, and distribute this file
35  * for any purpose is hereby granted without fee, provided that
36  * the above copyright notice and this notice appears in all
37  * copies, and that the name of Hewlett-Packard Company not be
38  * used in advertising or publicity pertaining to distribution
39  * of the software without specific, written prior permission.
40  * Hewlett-Packard Company makes no representations about the
41  * suitability of this software for any purpose.
42  */
43 
44 
45 #include "../spmath/float.h"
46 #include "../spmath/dbl_float.h"
47 
48 /*
49  *  Double Precision Floating-point Remainder
50  */
51 int
52 dbl_frem(srcptr1,srcptr2,dstptr,status)
53 
54 dbl_floating_point *srcptr1, *srcptr2, *dstptr;
55 unsigned int *status;
56 {
57 	register unsigned int opnd1p1, opnd1p2, opnd2p1, opnd2p2;
58 	register unsigned int resultp1, resultp2;
59 	register int opnd1_exponent, opnd2_exponent, dest_exponent, stepcount;
60 	register int roundup = FALSE;
61 
62 	Dbl_copyfromptr(srcptr1,opnd1p1,opnd1p2);
63 	Dbl_copyfromptr(srcptr2,opnd2p1,opnd2p2);
64 	/*
65 	 * check first operand for NaN's or infinity
66 	 */
67 	if ((opnd1_exponent = Dbl_exponent(opnd1p1)) == DBL_INFINITY_EXPONENT) {
68 		if (Dbl_iszero_mantissa(opnd1p1,opnd1p2)) {
69 			if (Dbl_isnotnan(opnd2p1,opnd2p2)) {
70 				/* invalid since first operand is infinity */
71 				if (Is_invalidtrap_enabled())
72 					return(INVALIDEXCEPTION);
73 				Set_invalidflag();
74 				Dbl_makequietnan(resultp1,resultp2);
75 				Dbl_copytoptr(resultp1,resultp2,dstptr);
76 				return(NOEXCEPTION);
77 			}
78 		}
79 		else {
80 			/*
81 			 * is NaN; signaling or quiet?
82 			 */
83 			if (Dbl_isone_signaling(opnd1p1)) {
84 				/* trap if INVALIDTRAP enabled */
85 				if (Is_invalidtrap_enabled())
86 					return(INVALIDEXCEPTION);
87 				/* make NaN quiet */
88 				Set_invalidflag();
89 				Dbl_set_quiet(opnd1p1);
90 			}
91 			/*
92 			 * is second operand a signaling NaN?
93 			 */
94 			else if (Dbl_is_signalingnan(opnd2p1)) {
95 				/* trap if INVALIDTRAP enabled */
96 				if (Is_invalidtrap_enabled())
97 					return(INVALIDEXCEPTION);
98 				/* make NaN quiet */
99 				Set_invalidflag();
100 				Dbl_set_quiet(opnd2p1);
101 				Dbl_copytoptr(opnd2p1,opnd2p2,dstptr);
102 				return(NOEXCEPTION);
103 			}
104 			/*
105 			 * return quiet NaN
106 			 */
107 			Dbl_copytoptr(opnd1p1,opnd1p2,dstptr);
108 			return(NOEXCEPTION);
109 		}
110 	}
111 	/*
112 	 * check second operand for NaN's or infinity
113 	 */
114 	if ((opnd2_exponent = Dbl_exponent(opnd2p1)) == DBL_INFINITY_EXPONENT) {
115 		if (Dbl_iszero_mantissa(opnd2p1,opnd2p2)) {
116 			/*
117 			 * return first operand
118 			 */
119 			Dbl_copytoptr(opnd1p1,opnd1p2,dstptr);
120 			return(NOEXCEPTION);
121 		}
122 		/*
123 		 * is NaN; signaling or quiet?
124 		 */
125 		if (Dbl_isone_signaling(opnd2p1)) {
126 			/* trap if INVALIDTRAP enabled */
127 			if (Is_invalidtrap_enabled()) return(INVALIDEXCEPTION);
128 			/* make NaN quiet */
129 			Set_invalidflag();
130 			Dbl_set_quiet(opnd2p1);
131 		}
132 		/*
133 		 * return quiet NaN
134 		 */
135 		Dbl_copytoptr(opnd2p1,opnd2p2,dstptr);
136 		return(NOEXCEPTION);
137 	}
138 	/*
139 	 * check second operand for zero
140 	 */
141 	if (Dbl_iszero_exponentmantissa(opnd2p1,opnd2p2)) {
142 		/* invalid since second operand is zero */
143 		if (Is_invalidtrap_enabled()) return(INVALIDEXCEPTION);
144 		Set_invalidflag();
145 		Dbl_makequietnan(resultp1,resultp2);
146 		Dbl_copytoptr(resultp1,resultp2,dstptr);
147 		return(NOEXCEPTION);
148 	}
149 
150 	/*
151 	 * get sign of result
152 	 */
153 	resultp1 = opnd1p1;
154 
155 	/*
156 	 * check for denormalized operands
157 	 */
158 	if (opnd1_exponent == 0) {
159 		/* check for zero */
160 		if (Dbl_iszero_mantissa(opnd1p1,opnd1p2)) {
161 			Dbl_copytoptr(opnd1p1,opnd1p2,dstptr);
162 			return(NOEXCEPTION);
163 		}
164 		/* normalize, then continue */
165 		opnd1_exponent = 1;
166 		Dbl_normalize(opnd1p1,opnd1p2,opnd1_exponent);
167 	}
168 	else {
169 		Dbl_clear_signexponent_set_hidden(opnd1p1);
170 	}
171 	if (opnd2_exponent == 0) {
172 		/* normalize, then continue */
173 		opnd2_exponent = 1;
174 		Dbl_normalize(opnd2p1,opnd2p2,opnd2_exponent);
175 	}
176 	else {
177 		Dbl_clear_signexponent_set_hidden(opnd2p1);
178 	}
179 
180 	/* find result exponent and divide step loop count */
181 	dest_exponent = opnd2_exponent - 1;
182 	stepcount = opnd1_exponent - opnd2_exponent;
183 
184 	/*
185 	 * check for opnd1/opnd2 < 1
186 	 */
187 	if (stepcount < 0) {
188 		/*
189 		 * check for opnd1/opnd2 > 1/2
190 		 *
191 		 * In this case n will round to 1, so
192 		 *    r = opnd1 - opnd2
193 		 */
194 		if (stepcount == -1 &&
195 		    Dbl_isgreaterthan(opnd1p1,opnd1p2,opnd2p1,opnd2p2)) {
196 			/* set sign */
197 			Dbl_allp1(resultp1) = ~Dbl_allp1(resultp1);
198 			/* align opnd2 with opnd1 */
199 			Dbl_leftshiftby1(opnd2p1,opnd2p2);
200 			Dbl_subtract(opnd2p1,opnd2p2,opnd1p1,opnd1p2,
201 			 opnd2p1,opnd2p2);
202 			/* now normalize */
203 			while (Dbl_iszero_hidden(opnd2p1)) {
204 				Dbl_leftshiftby1(opnd2p1,opnd2p2);
205 				dest_exponent--;
206 			}
207 			Dbl_set_exponentmantissa(resultp1,resultp2,opnd2p1,opnd2p2);
208 			goto testforunderflow;
209 		}
210 		/*
211 		 * opnd1/opnd2 <= 1/2
212 		 *
213 		 * In this case n will round to zero, so
214 		 *    r = opnd1
215 		 */
216 		Dbl_set_exponentmantissa(resultp1,resultp2,opnd1p1,opnd1p2);
217 		dest_exponent = opnd1_exponent;
218 		goto testforunderflow;
219 	}
220 
221 	/*
222 	 * Generate result
223 	 *
224 	 * Do iterative subtract until remainder is less than operand 2.
225 	 */
226 	while (stepcount-- > 0 && (Dbl_allp1(opnd1p1) || Dbl_allp2(opnd1p2))) {
227 		if (Dbl_isnotlessthan(opnd1p1,opnd1p2,opnd2p1,opnd2p2)) {
228 			Dbl_subtract(opnd1p1,opnd1p2,opnd2p1,opnd2p2,opnd1p1,opnd1p2);
229 		}
230 		Dbl_leftshiftby1(opnd1p1,opnd1p2);
231 	}
232 	/*
233 	 * Do last subtract, then determine which way to round if remainder
234 	 * is exactly 1/2 of opnd2
235 	 */
236 	if (Dbl_isnotlessthan(opnd1p1,opnd1p2,opnd2p1,opnd2p2)) {
237 		Dbl_subtract(opnd1p1,opnd1p2,opnd2p1,opnd2p2,opnd1p1,opnd1p2);
238 		roundup = TRUE;
239 	}
240 	if (stepcount > 0 || Dbl_iszero(opnd1p1,opnd1p2)) {
241 		/* division is exact, remainder is zero */
242 		Dbl_setzero_exponentmantissa(resultp1,resultp2);
243 		Dbl_copytoptr(resultp1,resultp2,dstptr);
244 		return(NOEXCEPTION);
245 	}
246 
247 	/*
248 	 * Check for cases where opnd1/opnd2 < n
249 	 *
250 	 * In this case the result's sign will be opposite that of
251 	 * opnd1.  The mantissa also needs some correction.
252 	 */
253 	Dbl_leftshiftby1(opnd1p1,opnd1p2);
254 	if (Dbl_isgreaterthan(opnd1p1,opnd1p2,opnd2p1,opnd2p2)) {
255 		Dbl_invert_sign(resultp1);
256 		Dbl_leftshiftby1(opnd2p1,opnd2p2);
257 		Dbl_subtract(opnd2p1,opnd2p2,opnd1p1,opnd1p2,opnd1p1,opnd1p2);
258 	}
259 	/* check for remainder being exactly 1/2 of opnd2 */
260 	else if (Dbl_isequal(opnd1p1,opnd1p2,opnd2p1,opnd2p2) && roundup) {
261 		Dbl_invert_sign(resultp1);
262 	}
263 
264 	/* normalize result's mantissa */
265 	while (Dbl_iszero_hidden(opnd1p1)) {
266 		dest_exponent--;
267 		Dbl_leftshiftby1(opnd1p1,opnd1p2);
268 	}
269 	Dbl_set_exponentmantissa(resultp1,resultp2,opnd1p1,opnd1p2);
270 
271 	/*
272 	 * Test for underflow
273 	 */
274     testforunderflow:
275 	if (dest_exponent <= 0) {
276 		/* trap if UNDERFLOWTRAP enabled */
277 		if (Is_underflowtrap_enabled()) {
278 			/*
279 			 * Adjust bias of result
280 			 */
281 			Dbl_setwrapped_exponent(resultp1,dest_exponent,unfl);
282 			/* frem is always exact */
283 			Dbl_copytoptr(resultp1,resultp2,dstptr);
284 			return(UNDERFLOWEXCEPTION);
285 		}
286 		/*
287 		 * denormalize result or set to signed zero
288 		 */
289 		if (dest_exponent >= (1 - DBL_P)) {
290 			Dbl_rightshift_exponentmantissa(resultp1,resultp2,
291 			 1-dest_exponent);
292 		}
293 		else {
294 			Dbl_setzero_exponentmantissa(resultp1,resultp2);
295 		}
296 	}
297 	else Dbl_set_exponent(resultp1,dest_exponent);
298 	Dbl_copytoptr(resultp1,resultp2,dstptr);
299 	return(NOEXCEPTION);
300 }
301