xref: /freebsd/lib/libc/gdtoa/_hdtoa.c (revision 8ab2f5ecc596131f6ca790d6ae35540c06ed7985)
1 /*-
2  * Copyright (c) 2004 David Schultz <das@FreeBSD.ORG>
3  * All rights reserved.
4  *
5  * Redistribution and use in source and binary forms, with or without
6  * modification, are permitted provided that the following conditions
7  * are met:
8  * 1. Redistributions of source code must retain the above copyright
9  *    notice, this list of conditions and the following disclaimer.
10  * 2. Redistributions in binary form must reproduce the above copyright
11  *    notice, this list of conditions and the following disclaimer in the
12  *    documentation and/or other materials provided with the distribution.
13  *
14  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
15  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
16  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
17  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
18  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
19  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
20  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
21  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
22  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
23  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
24  * SUCH DAMAGE.
25  */
26 
27 #include <sys/cdefs.h>
28 __FBSDID("$FreeBSD$");
29 
30 #include <float.h>
31 #include <inttypes.h>
32 #include <limits.h>
33 #include <math.h>
34 #include <stdlib.h>
35 #include "fpmath.h"
36 #include "gdtoaimp.h"
37 
38 /* Strings values used by dtoa() */
39 #define	INFSTR	"Infinity"
40 #define	NANSTR	"NaN"
41 
42 #define	DBL_BIAS	(DBL_MAX_EXP - 1)
43 #define	LDBL_BIAS	(LDBL_MAX_EXP - 1)
44 
45 #ifdef	LDBL_IMPLICIT_NBIT
46 #define	LDBL_NBIT_ADJ	0
47 #else
48 #define	LDBL_NBIT_ADJ	1
49 #endif
50 
51 /*
52  * Efficiently compute the log2 of an integer.  Uses a combination of
53  * arcane tricks found in fortune and arcane tricks not (yet) in
54  * fortune.  This routine behaves similarly to fls(9).
55  */
56 static int
57 log2_32(uint32_t n)
58 {
59 
60         n |= (n >> 1);
61         n |= (n >> 2);
62         n |= (n >> 4);
63         n |= (n >> 8);
64         n |= (n >> 16);
65 
66 	n = (n & 0x55555555) + ((n & 0xaaaaaaaa) >> 1);
67 	n = (n & 0x33333333) + ((n & 0xcccccccc) >> 2);
68 	n = (n & 0x0f0f0f0f) + ((n & 0xf0f0f0f0) >> 4);
69 	n = (n & 0x00ff00ff) + ((n & 0xff00ff00) >> 8);
70 	n = (n & 0x0000ffff) + ((n & 0xffff0000) >> 16);
71 	return (n - 1);
72 }
73 
74 #if (LDBL_MANH_SIZE > 32 || LDBL_MANL_SIZE > 32)
75 
76 static int
77 log2_64(uint64_t n)
78 {
79 
80 	if (n >> 32 != 0)
81 		return (log2_32((uint32_t)(n >> 32)) + 32);
82 	else
83 		return (log2_32((uint32_t)n));
84 }
85 
86 #endif	/* (LDBL_MANH_SIZE > 32 || LDBL_MANL_SIZE > 32) */
87 
88 /*
89  * Round up the given digit string.  If the digit string is fff...f,
90  * this procedure sets it to 100...0 and returns 1 to indicate that
91  * the exponent needs to be bumped.  Otherwise, 0 is returned.
92  */
93 static int
94 roundup(char *s0, int ndigits)
95 {
96 	char *s;
97 
98 	for (s = s0 + ndigits - 1; *s == 0xf; s--) {
99 		if (s == s0) {
100 			*s = 1;
101 			return (1);
102 		}
103 		++*s;
104 	}
105 	++*s;
106 	return (0);
107 }
108 
109 /*
110  * Round the given digit string to ndigits digits according to the
111  * current rounding mode.  Note that this could produce a string whose
112  * value is not representable in the corresponding floating-point
113  * type.  The exponent pointed to by decpt is adjusted if necessary.
114  */
115 static void
116 dorounding(char *s0, int ndigits, int sign, int *decpt)
117 {
118 	int adjust = 0;	/* do we need to adjust the exponent? */
119 
120 	switch (FLT_ROUNDS) {
121 	case 0:		/* toward zero */
122 	default:	/* implementation-defined */
123 		break;
124 	case 1:		/* to nearest, halfway rounds to even */
125 		if ((s0[ndigits] > 8) ||
126 		    (s0[ndigits] == 8 && s0[ndigits - 1] & 1))
127 			adjust = roundup(s0, ndigits);
128 		break;
129 	case 2:		/* toward +inf */
130 		if (sign == 0)
131 			adjust = roundup(s0, ndigits);
132 		break;
133 	case 3:		/* toward -inf */
134 		if (sign != 0)
135 			adjust = roundup(s0, ndigits);
136 		break;
137 	}
138 
139 	if (adjust)
140 		*decpt += 4;
141 }
142 
143 /*
144  * This procedure converts a double-precision number in IEEE format
145  * into a string of hexadecimal digits and an exponent of 2.  Its
146  * behavior is bug-for-bug compatible with dtoa() in mode 2, with the
147  * following exceptions:
148  *
149  * - An ndigits < 0 causes it to use as many digits as necessary to
150  *   represent the number exactly.
151  * - The additional xdigs argument should point to either the string
152  *   "0123456789ABCDEF" or the string "0123456789abcdef", depending on
153  *   which case is desired.
154  * - This routine does not repeat dtoa's mistake of setting decpt
155  *   to 9999 in the case of an infinity or NaN.  INT_MAX is used
156  *   for this purpose instead.
157  *
158  * Note that the C99 standard does not specify what the leading digit
159  * should be for non-zero numbers.  For instance, 0x1.3p3 is the same
160  * as 0x2.6p2 is the same as 0x4.cp3.  This implementation chooses the
161  * first digit so that subsequent digits are aligned on nibble
162  * boundaries (before rounding).
163  *
164  * Inputs:	d, xdigs, ndigits
165  * Outputs:	decpt, sign, rve
166  */
167 char *
168 __hdtoa(double d, const char *xdigs, int ndigits, int *decpt, int *sign,
169     char **rve)
170 {
171 	union IEEEd2bits u;
172 	char *s, *s0;
173 	int bufsize;
174 	int impnbit;	/* implicit normalization bit */
175 	int pos;
176 	int shift;	/* for subnormals, # of shifts required to normalize */
177 	int sigfigs;	/* number of significant hex figures in result */
178 
179 	u.d = d;
180 	*sign = u.bits.sign;
181 
182 	switch (fpclassify(d)) {
183 	case FP_NORMAL:
184 		sigfigs = (DBL_MANT_DIG + 3) / 4;
185 		impnbit = 1 << ((DBL_MANT_DIG - 1) % 4);
186 		*decpt = u.bits.exp - DBL_BIAS + 1 -
187 		    ((DBL_MANT_DIG - 1) % 4);
188 		break;
189 	case FP_ZERO:
190 		*decpt = 1;
191 		return (nrv_alloc("0", rve, 1));
192 	case FP_SUBNORMAL:
193 		/*
194 		 * The position of the highest-order bit tells us by
195 		 * how much to adjust the exponent (decpt).  The
196 		 * adjustment is raised to the next nibble boundary
197 		 * since we will later choose the leftmost hexadecimal
198 		 * digit so that all subsequent digits align on nibble
199 		 * boundaries.
200 		 */
201 		if (u.bits.manh != 0) {
202 			pos = log2_32(u.bits.manh);
203 			shift = DBL_MANH_SIZE - pos;
204 		} else {
205 			pos = log2_32(u.bits.manl);
206 			shift = DBL_MANH_SIZE + DBL_MANL_SIZE - pos;
207 		}
208 		sigfigs = (3 + DBL_MANT_DIG - shift) / 4;
209 		impnbit = 0;
210 		*decpt = DBL_MIN_EXP - ((shift + 3) & ~(4 - 1));
211 		break;
212 	case FP_INFINITE:
213 		*decpt = INT_MAX;
214 		return (nrv_alloc(INFSTR, rve, sizeof(INFSTR) - 1));
215 	case FP_NAN:
216 		*decpt = INT_MAX;
217 		return (nrv_alloc(NANSTR, rve, sizeof(NANSTR) - 1));
218 	default:
219 		abort();
220 	}
221 
222 	/* FP_NORMAL or FP_SUBNORMAL */
223 
224 	if (ndigits == 0)		/* dtoa() compatibility */
225 		ndigits = 1;
226 
227 	/*
228 	 * For simplicity, we generate all the digits even if the
229 	 * caller has requested fewer.
230 	 */
231 	bufsize = (sigfigs > ndigits) ? sigfigs : ndigits;
232 	s0 = rv_alloc(bufsize);
233 
234 	/*
235 	 * We work from right to left, first adding any requested zero
236 	 * padding, then the least significant portion of the
237 	 * mantissa, followed by the most significant.  The buffer is
238 	 * filled with the byte values 0x0 through 0xf, which are
239 	 * converted to xdigs[0x0] through xdigs[0xf] after the
240 	 * rounding phase.
241 	 */
242 	for (s = s0 + bufsize - 1; s > s0 + sigfigs - 1; s--)
243 		*s = 0;
244 	for (; s > s0 + sigfigs - (DBL_MANL_SIZE / 4) - 1 && s > s0; s--) {
245 		*s = u.bits.manl & 0xf;
246 		u.bits.manl >>= 4;
247 	}
248 	for (; s > s0; s--) {
249 		*s = u.bits.manh & 0xf;
250 		u.bits.manh >>= 4;
251 	}
252 
253 	/*
254 	 * At this point, we have snarfed all the bits in the
255 	 * mantissa, with the possible exception of the highest-order
256 	 * (partial) nibble, which is dealt with by the next
257 	 * statement.  That nibble is usually in manh, but it could be
258 	 * in manl instead for small subnormals.  We also tack on the
259 	 * implicit normalization bit if appropriate.
260 	 */
261 	*s = u.bits.manh | u.bits.manl | impnbit;
262 
263 	/* If ndigits < 0, we are expected to auto-size the precision. */
264 	if (ndigits < 0) {
265 		for (ndigits = sigfigs; s0[ndigits - 1] == 0; ndigits--)
266 			;
267 	}
268 
269 	if (sigfigs > ndigits && s0[ndigits] != 0)
270 		dorounding(s0, ndigits, u.bits.sign, decpt);
271 
272 	s = s0 + ndigits;
273 	if (rve != NULL)
274 		*rve = s;
275 	*s-- = '\0';
276 	for (; s >= s0; s--)
277 		*s = xdigs[(unsigned int)*s];
278 
279 	return (s0);
280 }
281 
282 #if (LDBL_MANT_DIG > DBL_MANT_DIG)
283 
284 /*
285  * This is the long double version of __hdtoa().
286  *
287  * On architectures that have an explicit integer bit, unnormals and
288  * pseudo-denormals cause problems in the conversion routine, so they
289  * are ``fixed'' by effectively toggling the integer bit.  Although
290  * this is not correct behavior, the hardware will not produce these
291  * formats externally.
292  */
293 char *
294 __hldtoa(long double e, const char *xdigs, int ndigits, int *decpt, int *sign,
295     char **rve)
296 {
297 	union IEEEl2bits u;
298 	char *s, *s0;
299 	int bufsize;
300 	int impnbit;	/* implicit normalization bit */
301 	int pos;
302 	int shift;	/* for subnormals, # of shifts required to normalize */
303 	int sigfigs;	/* number of significant hex figures in result */
304 
305 	u.e = e;
306 	*sign = u.bits.sign;
307 
308 	switch (fpclassify(e)) {
309 	case FP_NORMAL:
310 		sigfigs = (LDBL_MANT_DIG + 3) / 4;
311 		impnbit = 1 << ((LDBL_MANT_DIG - 1) % 4);
312 		*decpt = u.bits.exp - LDBL_BIAS + 1 -
313 		    ((LDBL_MANT_DIG - 1) % 4);
314 		break;
315 	case FP_ZERO:
316 		*decpt = 1;
317 		return (nrv_alloc("0", rve, 1));
318 	case FP_SUBNORMAL:
319 		/*
320 		 * The position of the highest-order bit tells us by
321 		 * how much to adjust the exponent (decpt).  The
322 		 * adjustment is raised to the next nibble boundary
323 		 * since we will later choose the leftmost hexadecimal
324 		 * digit so that all subsequent digits align on nibble
325 		 * boundaries.
326 		 */
327 #ifdef	LDBL_IMPLICIT_NBIT
328 		/* Don't trust the normalization bit to be off. */
329 		u.bits.manh &= ~(~0ULL << (LDBL_MANH_SIZE - 1));
330 #endif
331 		if (u.bits.manh != 0) {
332 #if LDBL_MANH_SIZE > 32
333 			pos = log2_64(u.bits.manh);
334 #else
335 			pos = log2_32(u.bits.manh);
336 #endif
337 			shift = LDBL_MANH_SIZE - LDBL_NBIT_ADJ - pos;
338 		} else {
339 #if LDBL_MANL_SIZE > 32
340 			pos = log2_64(u.bits.manl);
341 #else
342 			pos = log2_32(u.bits.manl);
343 #endif
344 			shift = LDBL_MANH_SIZE + LDBL_MANL_SIZE -
345 			    LDBL_NBIT_ADJ - pos;
346 		}
347 		sigfigs = (3 + LDBL_MANT_DIG - LDBL_NBIT_ADJ - shift) / 4;
348 		*decpt = LDBL_MIN_EXP + LDBL_NBIT_ADJ -
349 		    ((shift + 3) & ~(4 - 1));
350 		impnbit = 0;
351 		break;
352 	case FP_INFINITE:
353 		*decpt = INT_MAX;
354 		return (nrv_alloc(INFSTR, rve, sizeof(INFSTR) - 1));
355 	case FP_NAN:
356 		*decpt = INT_MAX;
357 		return (nrv_alloc(NANSTR, rve, sizeof(NANSTR) - 1));
358 	default:
359 		abort();
360 	}
361 
362 	/* FP_NORMAL or FP_SUBNORMAL */
363 
364 	if (ndigits == 0)		/* dtoa() compatibility */
365 		ndigits = 1;
366 
367 	/*
368 	 * For simplicity, we generate all the digits even if the
369 	 * caller has requested fewer.
370 	 */
371 	bufsize = (sigfigs > ndigits) ? sigfigs : ndigits;
372 	s0 = rv_alloc(bufsize);
373 
374 	/*
375 	 * We work from right to left, first adding any requested zero
376 	 * padding, then the least significant portion of the
377 	 * mantissa, followed by the most significant.  The buffer is
378 	 * filled with the byte values 0x0 through 0xf, which are
379 	 * converted to xdigs[0x0] through xdigs[0xf] after the
380 	 * rounding phase.
381 	 */
382 	for (s = s0 + bufsize - 1; s > s0 + sigfigs - 1; s--)
383 		*s = 0;
384 	for (; s > s0 + sigfigs - (LDBL_MANL_SIZE / 4) - 1 && s > s0; s--) {
385 		*s = u.bits.manl & 0xf;
386 		u.bits.manl >>= 4;
387 	}
388 	for (; s > s0; s--) {
389 		*s = u.bits.manh & 0xf;
390 		u.bits.manh >>= 4;
391 	}
392 
393 	/*
394 	 * At this point, we have snarfed all the bits in the
395 	 * mantissa, with the possible exception of the highest-order
396 	 * (partial) nibble, which is dealt with by the next
397 	 * statement.  That nibble is usually in manh, but it could be
398 	 * in manl instead for small subnormals.  We also tack on the
399 	 * implicit normalization bit if appropriate.
400 	 */
401 	*s = u.bits.manh | u.bits.manl | impnbit;
402 
403 	/* If ndigits < 0, we are expected to auto-size the precision. */
404 	if (ndigits < 0) {
405 		for (ndigits = sigfigs; s0[ndigits - 1] == 0; ndigits--)
406 			;
407 	}
408 
409 	if (sigfigs > ndigits && s0[ndigits] != 0)
410 		dorounding(s0, ndigits, u.bits.sign, decpt);
411 
412 	s = s0 + ndigits;
413 	if (rve != NULL)
414 		*rve = s;
415 	*s-- = '\0';
416 	for (; s >= s0; s--)
417 		*s = xdigs[(unsigned int)*s];
418 
419 	return (s0);
420 }
421 
422 #else	/* (LDBL_MANT_DIG == DBL_MANT_DIG) */
423 
424 char *
425 __hldtoa(long double e, const char *xdigs, int ndigits, int *decpt, int *sign,
426     char **rve)
427 {
428 
429 	return (__hdtoa((double)e, xdigs, ndigits, decpt, sign, rve));
430 }
431 
432 #endif	/* (LDBL_MANT_DIG == DBL_MANT_DIG) */
433