aboutsummaryrefslogtreecommitdiff
path: root/libiberty/floatformat.c
blob: 6b9b03288e29cee2713f7232e7585e8ebf5429bd (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
/* IEEE floating point support routines, for GDB, the GNU Debugger.
   Copyright (C) 1991-2020 Free Software Foundation, Inc.

This file is part of GDB.

This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 2 of the License, or
(at your option) any later version.

This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
GNU General Public License for more details.

You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Foundation, Inc., 51 Franklin Street - Fifth Floor, Boston, MA 02110-1301, USA.  */

/* This is needed to pick up the NAN macro on some systems.  */
#ifndef _GNU_SOURCE
#define _GNU_SOURCE
#endif

#ifdef HAVE_CONFIG_H
#include "config.h"
#endif

#include <math.h>

#ifdef HAVE_STRING_H
#include <string.h>
#endif

/* On some platforms, <float.h> provides DBL_QNAN.  */
#ifdef STDC_HEADERS
#include <float.h>
#endif

#include "ansidecl.h"
#include "libiberty.h"
#include "floatformat.h"

#ifndef INFINITY
#ifdef HUGE_VAL
#define INFINITY HUGE_VAL
#else
#define INFINITY (1.0 / 0.0)
#endif
#endif

#ifndef NAN
#ifdef DBL_QNAN
#define NAN DBL_QNAN
#else
#define NAN (0.0 / 0.0)
#endif
#endif

static int mant_bits_set (const struct floatformat *, const unsigned char *);
static unsigned long get_field (const unsigned char *,
                                enum floatformat_byteorders,
                                unsigned int,
                                unsigned int,
                                unsigned int);
static int floatformat_always_valid (const struct floatformat *fmt,
                                     const void *from);

static int
floatformat_always_valid (const struct floatformat *fmt ATTRIBUTE_UNUSED,
                          const void *from ATTRIBUTE_UNUSED)
{
  return 1;
}

/* The odds that CHAR_BIT will be anything but 8 are low enough that I'm not
   going to bother with trying to muck around with whether it is defined in
   a system header, what we do if not, etc.  */
#define FLOATFORMAT_CHAR_BIT 8

/* floatformats for IEEE half, single and double, big and little endian.  */
const struct floatformat floatformat_ieee_half_big =
{
  floatformat_big, 16, 0, 1, 5, 15, 31, 6, 10,
  floatformat_intbit_no,
  "floatformat_ieee_half_big",
  floatformat_always_valid,
  NULL
};
const struct floatformat floatformat_ieee_half_little =
{
  floatformat_little, 16, 0, 1, 5, 15, 31, 6, 10,
  floatformat_intbit_no,
  "floatformat_ieee_half_little",
  floatformat_always_valid,
  NULL
};
const struct floatformat floatformat_ieee_single_big =
{
  floatformat_big, 32, 0, 1, 8, 127, 255, 9, 23,
  floatformat_intbit_no,
  "floatformat_ieee_single_big",
  floatformat_always_valid,
  NULL
};
const struct floatformat floatformat_ieee_single_little =
{
  floatformat_little, 32, 0, 1, 8, 127, 255, 9, 23,
  floatformat_intbit_no,
  "floatformat_ieee_single_little",
  floatformat_always_valid,
  NULL
};
const struct floatformat floatformat_ieee_double_big =
{
  floatformat_big, 64, 0, 1, 11, 1023, 2047, 12, 52,
  floatformat_intbit_no,
  "floatformat_ieee_double_big",
  floatformat_always_valid,
  NULL
};
const struct floatformat floatformat_ieee_double_little =
{
  floatformat_little, 64, 0, 1, 11, 1023, 2047, 12, 52,
  floatformat_intbit_no,
  "floatformat_ieee_double_little",
  floatformat_always_valid,
  NULL
};

/* floatformat for IEEE double, little endian byte order, with big endian word
   ordering, as on the ARM.  */

const struct floatformat floatformat_ieee_double_littlebyte_bigword =
{
  floatformat_littlebyte_bigword, 64, 0, 1, 11, 1023, 2047, 12, 52,
  floatformat_intbit_no,
  "floatformat_ieee_double_littlebyte_bigword",
  floatformat_always_valid,
  NULL
};

/* floatformat for VAX.  Not quite IEEE, but close enough.  */

const struct floatformat floatformat_vax_f =
{
  floatformat_vax, 32, 0, 1, 8, 129, 0, 9, 23,
  floatformat_intbit_no,
  "floatformat_vax_f",
  floatformat_always_valid,
  NULL
};
const struct floatformat floatformat_vax_d =
{
  floatformat_vax, 64, 0, 1, 8, 129, 0, 9, 55,
  floatformat_intbit_no,
  "floatformat_vax_d",
  floatformat_always_valid,
  NULL
};
const struct floatformat floatformat_vax_g =
{
  floatformat_vax, 64, 0, 1, 11, 1025, 0, 12, 52,
  floatformat_intbit_no,
  "floatformat_vax_g",
  floatformat_always_valid,
  NULL
};

static int floatformat_i387_ext_is_valid (const struct floatformat *fmt,
					  const void *from);

static int
floatformat_i387_ext_is_valid (const struct floatformat *fmt, const void *from)
{
  /* In the i387 double-extended format, if the exponent is all ones,
     then the integer bit must be set.  If the exponent is neither 0
     nor ~0, the intbit must also be set.  Only if the exponent is
     zero can it be zero, and then it must be zero.  */
  unsigned long exponent, int_bit;
  const unsigned char *ufrom = (const unsigned char *) from;

  exponent = get_field (ufrom, fmt->byteorder, fmt->totalsize,
			fmt->exp_start, fmt->exp_len);
  int_bit = get_field (ufrom, fmt->byteorder, fmt->totalsize,
		       fmt->man_start, 1);

  if ((exponent == 0) != (int_bit == 0))
    return 0;
  else
    return 1;
}

const struct floatformat floatformat_i387_ext =
{
  floatformat_little, 80, 0, 1, 15, 0x3fff, 0x7fff, 16, 64,
  floatformat_intbit_yes,
  "floatformat_i387_ext",
  floatformat_i387_ext_is_valid,
  NULL
};
const struct floatformat floatformat_m68881_ext =
{
  /* Note that the bits from 16 to 31 are unused.  */
  floatformat_big, 96, 0, 1, 15, 0x3fff, 0x7fff, 32, 64,
  floatformat_intbit_yes,
  "floatformat_m68881_ext",
  floatformat_always_valid,
  NULL
};
const struct floatformat floatformat_i960_ext =
{
  /* Note that the bits from 0 to 15 are unused.  */
  floatformat_little, 96, 16, 17, 15, 0x3fff, 0x7fff, 32, 64,
  floatformat_intbit_yes,
  "floatformat_i960_ext",
  floatformat_always_valid,
  NULL
};
const struct floatformat floatformat_m88110_ext =
{
  floatformat_big, 80, 0, 1, 15, 0x3fff, 0x7fff, 16, 64,
  floatformat_intbit_yes,
  "floatformat_m88110_ext",
  floatformat_always_valid,
  NULL
};
const struct floatformat floatformat_m88110_harris_ext =
{
  /* Harris uses raw format 128 bytes long, but the number is just an ieee
     double, and the last 64 bits are wasted. */
  floatformat_big,128, 0, 1, 11,  0x3ff,  0x7ff, 12, 52,
  floatformat_intbit_no,
  "floatformat_m88110_ext_harris",
  floatformat_always_valid,
  NULL
};
const struct floatformat floatformat_arm_ext_big =
{
  /* Bits 1 to 16 are unused.  */
  floatformat_big, 96, 0, 17, 15, 0x3fff, 0x7fff, 32, 64,
  floatformat_intbit_yes,
  "floatformat_arm_ext_big",
  floatformat_always_valid,
  NULL
};
const struct floatformat floatformat_arm_ext_littlebyte_bigword =
{
  /* Bits 1 to 16 are unused.  */
  floatformat_littlebyte_bigword, 96, 0, 17, 15, 0x3fff, 0x7fff, 32, 64,
  floatformat_intbit_yes,
  "floatformat_arm_ext_littlebyte_bigword",
  floatformat_always_valid,
  NULL
};
const struct floatformat floatformat_ia64_spill_big =
{
  floatformat_big, 128, 0, 1, 17, 65535, 0x1ffff, 18, 64,
  floatformat_intbit_yes,
  "floatformat_ia64_spill_big",
  floatformat_always_valid,
  NULL
};
const struct floatformat floatformat_ia64_spill_little =
{
  floatformat_little, 128, 0, 1, 17, 65535, 0x1ffff, 18, 64,
  floatformat_intbit_yes,
  "floatformat_ia64_spill_little",
  floatformat_always_valid,
  NULL
};
const struct floatformat floatformat_ia64_quad_big =
{
  floatformat_big, 128, 0, 1, 15, 16383, 0x7fff, 16, 112,
  floatformat_intbit_no,
  "floatformat_ia64_quad_big",
  floatformat_always_valid,
  NULL
};
const struct floatformat floatformat_ia64_quad_little =
{
  floatformat_little, 128, 0, 1, 15, 16383, 0x7fff, 16, 112,
  floatformat_intbit_no,
  "floatformat_ia64_quad_little",
  floatformat_always_valid,
  NULL
};

static int
floatformat_ibm_long_double_is_valid (const struct floatformat *fmt,
				      const void *from)
{
  const unsigned char *ufrom = (const unsigned char *) from;
  const struct floatformat *hfmt = fmt->split_half;
  long top_exp, bot_exp;
  int top_nan = 0;

  top_exp = get_field (ufrom, hfmt->byteorder, hfmt->totalsize,
		       hfmt->exp_start, hfmt->exp_len);
  bot_exp = get_field (ufrom + 8, hfmt->byteorder, hfmt->totalsize,
		       hfmt->exp_start, hfmt->exp_len);

  if ((unsigned long) top_exp == hfmt->exp_nan)
    top_nan = mant_bits_set (hfmt, ufrom);

  /* A NaN is valid with any low part.  */
  if (top_nan)
    return 1;

  /* An infinity, zero or denormal requires low part 0 (positive or
     negative).  */
  if ((unsigned long) top_exp == hfmt->exp_nan || top_exp == 0)
    {
      if (bot_exp != 0)
	return 0;

      return !mant_bits_set (hfmt, ufrom + 8);
    }

  /* The top part is now a finite normal value.  The long double value
     is the sum of the two parts, and the top part must equal the
     result of rounding the long double value to nearest double.  Thus
     the bottom part must be <= 0.5ulp of the top part in absolute
     value, and if it is < 0.5ulp then the long double is definitely
     valid.  */
  if (bot_exp < top_exp - 53)
    return 1;
  if (bot_exp > top_exp - 53 && bot_exp != 0)
    return 0;
  if (bot_exp == 0)
    {
      /* The bottom part is 0 or denormal.  Determine which, and if
	 denormal the first two set bits.  */
      int first_bit = -1, second_bit = -1, cur_bit;
      for (cur_bit = 0; (unsigned int) cur_bit < hfmt->man_len; cur_bit++)
	if (get_field (ufrom + 8, hfmt->byteorder, hfmt->totalsize,
		       hfmt->man_start + cur_bit, 1))
	  {
	    if (first_bit == -1)
	      first_bit = cur_bit;
	    else
	      {
		second_bit = cur_bit;
		break;
	      }
	  }
      /* Bottom part 0 is OK.  */
      if (first_bit == -1)
	return 1;
      /* The real exponent of the bottom part is -first_bit.  */
      if (-first_bit < top_exp - 53)
	return 1;
      if (-first_bit > top_exp - 53)
	return 0;
      /* The bottom part is at least 0.5ulp of the top part.  For this
	 to be OK, the bottom part must be exactly 0.5ulp (i.e. no
	 more bits set) and the top part must have last bit 0.  */
      if (second_bit != -1)
	return 0;
      return !get_field (ufrom, hfmt->byteorder, hfmt->totalsize,
			 hfmt->man_start + hfmt->man_len - 1, 1);
    }
  else
    {
      /* The bottom part is at least 0.5ulp of the top part.  For this
	 to be OK, it must be exactly 0.5ulp (i.e. no explicit bits
	 set) and the top part must have last bit 0.  */
      if (get_field (ufrom, hfmt->byteorder, hfmt->totalsize,
		     hfmt->man_start + hfmt->man_len - 1, 1))
	return 0;
      return !mant_bits_set (hfmt, ufrom + 8);
    }
}

const struct floatformat floatformat_ibm_long_double_big =
{
  floatformat_big, 128, 0, 1, 11, 1023, 2047, 12, 52,
  floatformat_intbit_no,
  "floatformat_ibm_long_double_big",
  floatformat_ibm_long_double_is_valid,
  &floatformat_ieee_double_big
};

const struct floatformat floatformat_ibm_long_double_little =
{
  floatformat_little, 128, 0, 1, 11, 1023, 2047, 12, 52,
  floatformat_intbit_no,
  "floatformat_ibm_long_double_little",
  floatformat_ibm_long_double_is_valid,
  &floatformat_ieee_double_little
};

const struct floatformat floatformat_bfloat16_big =
{
  floatformat_big, 16, 0, 1, 8, 127, 255, 9, 7,
  floatformat_intbit_no,
  "floatformat_bfloat16_big",
  floatformat_always_valid,
  NULL
};

const struct floatformat floatformat_bfloat16_little =
{
  floatformat_little, 16, 0, 1, 8, 127, 255, 9, 7,
  floatformat_intbit_no,
  "floatformat_bfloat16_little",
  floatformat_always_valid,
  NULL
};

#ifndef min
#define min(a, b) ((a) < (b) ? (a) : (b))
#endif

/* Return 1 if any bits are explicitly set in the mantissa of UFROM,
   format FMT, 0 otherwise.  */
static int
mant_bits_set (const struct floatformat *fmt, const unsigned char *ufrom)
{
  unsigned int mant_bits, mant_off;
  int mant_bits_left;

  mant_off = fmt->man_start;
  mant_bits_left = fmt->man_len;
  while (mant_bits_left > 0)
    {
      mant_bits = min (mant_bits_left, 32);

      if (get_field (ufrom, fmt->byteorder, fmt->totalsize,
		     mant_off, mant_bits) != 0)
	return 1;

      mant_off += mant_bits;
      mant_bits_left -= mant_bits;
    }
  return 0;
}

/* Extract a field which starts at START and is LEN bits long.  DATA and
   TOTAL_LEN are the thing we are extracting it from, in byteorder ORDER.  */
static unsigned long
get_field (const unsigned char *data, enum floatformat_byteorders order,
           unsigned int total_len, unsigned int start, unsigned int len)
{
  unsigned long result = 0;
  unsigned int cur_byte;
  int lo_bit, hi_bit, cur_bitshift = 0;
  int nextbyte = (order == floatformat_little) ? 1 : -1;

  /* Start is in big-endian bit order!  Fix that first.  */
  start = total_len - (start + len);

  /* Start at the least significant part of the field.  */
  if (order == floatformat_little)
    cur_byte = start / FLOATFORMAT_CHAR_BIT;
  else
    cur_byte = (total_len - start - 1) / FLOATFORMAT_CHAR_BIT;

  lo_bit = start % FLOATFORMAT_CHAR_BIT;
  hi_bit = min (lo_bit + len, FLOATFORMAT_CHAR_BIT);
  
  do
    {
      unsigned int shifted = *(data + cur_byte) >> lo_bit;
      unsigned int bits = hi_bit - lo_bit;
      unsigned int mask = (1 << bits) - 1;
      result |= (shifted & mask) << cur_bitshift;
      len -= bits;
      cur_bitshift += bits;
      cur_byte += nextbyte;
      lo_bit = 0;
      hi_bit = min (len, FLOATFORMAT_CHAR_BIT);
    }
  while (len != 0);

  return result;
}
  
/* Convert from FMT to a double.
   FROM is the address of the extended float.
   Store the double in *TO.  */

void
floatformat_to_double (const struct floatformat *fmt,
                       const void *from, double *to)
{
  const unsigned char *ufrom = (const unsigned char *) from;
  double dto;
  long exponent;
  unsigned long mant;
  unsigned int mant_bits, mant_off;
  int mant_bits_left;

  /* Split values are not handled specially, since the top half has
     the correctly rounded double value (in the only supported case of
     split values).  */

  exponent = get_field (ufrom, fmt->byteorder, fmt->totalsize,
			fmt->exp_start, fmt->exp_len);

  /* If the exponent indicates a NaN, we don't have information to
     decide what to do.  So we handle it like IEEE, except that we
     don't try to preserve the type of NaN.  FIXME.  */
  if ((unsigned long) exponent == fmt->exp_nan)
    {
      int nan = mant_bits_set (fmt, ufrom);

      /* On certain systems (such as GNU/Linux), the use of the
	 INFINITY macro below may generate a warning that cannot be
	 silenced due to a bug in GCC (PR preprocessor/11931).  The
	 preprocessor fails to recognise the __extension__ keyword in
	 conjunction with the GNU/C99 extension for hexadecimal
	 floating point constants and will issue a warning when
	 compiling with -pedantic.  */
      if (nan)
	dto = NAN;
      else
	dto = INFINITY;

      if (get_field (ufrom, fmt->byteorder, fmt->totalsize, fmt->sign_start, 1))
	dto = -dto;

      *to = dto;

      return;
    }

  mant_bits_left = fmt->man_len;
  mant_off = fmt->man_start;
  dto = 0.0;

  /* Build the result algebraically.  Might go infinite, underflow, etc;
     who cares. */

  /* For denorms use minimum exponent.  */
  if (exponent == 0)
    exponent = 1 - fmt->exp_bias;
  else
    {
      exponent -= fmt->exp_bias;

      /* If this format uses a hidden bit, explicitly add it in now.
	 Otherwise, increment the exponent by one to account for the
	 integer bit.  */

      if (fmt->intbit == floatformat_intbit_no)
	dto = ldexp (1.0, exponent);
      else
	exponent++;
    }

  while (mant_bits_left > 0)
    {
      mant_bits = min (mant_bits_left, 32);

      mant = get_field (ufrom, fmt->byteorder, fmt->totalsize,
			 mant_off, mant_bits);

      dto += ldexp ((double) mant, exponent - mant_bits);
      exponent -= mant_bits;
      mant_off += mant_bits;
      mant_bits_left -= mant_bits;
    }

  /* Negate it if negative.  */
  if (get_field (ufrom, fmt->byteorder, fmt->totalsize, fmt->sign_start, 1))
    dto = -dto;
  *to = dto;
}

static void put_field (unsigned char *, enum floatformat_byteorders,
                       unsigned int,
                       unsigned int,
                       unsigned int,
                       unsigned long);

/* Set a field which starts at START and is LEN bits long.  DATA and
   TOTAL_LEN are the thing we are extracting it from, in byteorder ORDER.  */
static void
put_field (unsigned char *data, enum floatformat_byteorders order,
           unsigned int total_len, unsigned int start, unsigned int len,
           unsigned long stuff_to_put)
{
  unsigned int cur_byte;
  int lo_bit, hi_bit;
  int nextbyte = (order == floatformat_little) ? 1 : -1;

  /* Start is in big-endian bit order!  Fix that first.  */
  start = total_len - (start + len);

  /* Start at the least significant part of the field.  */
  if (order == floatformat_little)
    cur_byte = start / FLOATFORMAT_CHAR_BIT;
  else
    cur_byte = (total_len - start - 1) / FLOATFORMAT_CHAR_BIT;

  lo_bit = start % FLOATFORMAT_CHAR_BIT;
  hi_bit = min (lo_bit + len, FLOATFORMAT_CHAR_BIT);
  
  do
    {
      unsigned char *byte_ptr = data + cur_byte;
      unsigned int bits = hi_bit - lo_bit;
      unsigned int mask = ((1 << bits) - 1) << lo_bit;
      *byte_ptr = (*byte_ptr & ~mask) | ((stuff_to_put << lo_bit) & mask);
      stuff_to_put >>= bits;
      len -= bits;
      cur_byte += nextbyte;
      lo_bit = 0;
      hi_bit = min (len, FLOATFORMAT_CHAR_BIT);
    }
  while (len != 0);
}

/* The converse: convert the double *FROM to an extended float
   and store where TO points.  Neither FROM nor TO have any alignment
   restrictions.  */

void
floatformat_from_double (const struct floatformat *fmt,
                         const double *from, void *to)
{
  double dfrom;
  int exponent;
  double mant;
  unsigned int mant_bits, mant_off;
  int mant_bits_left;
  unsigned char *uto = (unsigned char *) to;

  dfrom = *from;
  memset (uto, 0, fmt->totalsize / FLOATFORMAT_CHAR_BIT);

  /* Split values are not handled specially, since a bottom half of
     zero is correct for any value representable as double (in the
     only supported case of split values).  */

  /* If negative, set the sign bit.  */
  if (dfrom < 0)
    {
      put_field (uto, fmt->byteorder, fmt->totalsize, fmt->sign_start, 1, 1);
      dfrom = -dfrom;
    }

  if (dfrom == 0)
    {
      /* 0.0.  */
      return;
    }

  if (dfrom != dfrom)
    {
      /* NaN.  */
      put_field (uto, fmt->byteorder, fmt->totalsize, fmt->exp_start,
		 fmt->exp_len, fmt->exp_nan);
      /* Be sure it's not infinity, but NaN value is irrelevant.  */
      put_field (uto, fmt->byteorder, fmt->totalsize, fmt->man_start,
		 32, 1);
      return;
    }

  if (dfrom + dfrom == dfrom)
    {
      /* This can only happen for an infinite value (or zero, which we
	 already handled above).  */
      put_field (uto, fmt->byteorder, fmt->totalsize, fmt->exp_start,
		 fmt->exp_len, fmt->exp_nan);
      return;
    }

  mant = frexp (dfrom, &exponent);
  if (exponent + fmt->exp_bias - 1 > 0)
    put_field (uto, fmt->byteorder, fmt->totalsize, fmt->exp_start,
	       fmt->exp_len, exponent + fmt->exp_bias - 1);
  else
    {
      /* Handle a denormalized number.  FIXME: What should we do for
	 non-IEEE formats?  */
      put_field (uto, fmt->byteorder, fmt->totalsize, fmt->exp_start,
		 fmt->exp_len, 0);
      mant = ldexp (mant, exponent + fmt->exp_bias - 1);
    }

  mant_bits_left = fmt->man_len;
  mant_off = fmt->man_start;
  while (mant_bits_left > 0)
    {
      unsigned long mant_long;
      mant_bits = mant_bits_left < 32 ? mant_bits_left : 32;

      mant *= 4294967296.0;
      mant_long = (unsigned long)mant;
      mant -= mant_long;

      /* If the integer bit is implicit, and we are not creating a
	 denormalized number, then we need to discard it.  */
      if ((unsigned int) mant_bits_left == fmt->man_len
	  && fmt->intbit == floatformat_intbit_no
	  && exponent + fmt->exp_bias - 1 > 0)
	{
	  mant_long &= 0x7fffffff;
	  mant_bits -= 1;
	}
      else if (mant_bits < 32)
	{
	  /* The bits we want are in the most significant MANT_BITS bits of
	     mant_long.  Move them to the least significant.  */
	  mant_long >>= 32 - mant_bits;
	}

      put_field (uto, fmt->byteorder, fmt->totalsize,
		 mant_off, mant_bits, mant_long);
      mant_off += mant_bits;
      mant_bits_left -= mant_bits;
    }
}

/* Return non-zero iff the data at FROM is a valid number in format FMT.  */

int
floatformat_is_valid (const struct floatformat *fmt, const void *from)
{
  return fmt->is_valid (fmt, from);
}


#ifdef IEEE_DEBUG

#include <stdio.h>

/* This is to be run on a host which uses IEEE floating point.  */

void
ieee_test (double n)
{
  double result;

  floatformat_to_double (&floatformat_ieee_double_little, &n, &result);
  if ((n != result && (! isnan (n) || ! isnan (result)))
      || (n < 0 && result >= 0)
      || (n >= 0 && result < 0))
    printf ("Differ(to): %.20g -> %.20g\n", n, result);

  floatformat_from_double (&floatformat_ieee_double_little, &n, &result);
  if ((n != result && (! isnan (n) || ! isnan (result)))
      || (n < 0 && result >= 0)
      || (n >= 0 && result < 0))
    printf ("Differ(from): %.20g -> %.20g\n", n, result);

#if 0
  {
    char exten[16];

    floatformat_from_double (&floatformat_m68881_ext, &n, exten);
    floatformat_to_double (&floatformat_m68881_ext, exten, &result);
    if (n != result)
      printf ("Differ(to+from): %.20g -> %.20g\n", n, result);
  }
#endif

#if IEEE_DEBUG > 1
  /* This is to be run on a host which uses 68881 format.  */
  {
    long double ex = *(long double *)exten;
    if (ex != n)
      printf ("Differ(from vs. extended): %.20g\n", n);
  }
#endif
}

int
main (void)
{
  ieee_test (0.0);
  ieee_test (0.5);
  ieee_test (1.1);
  ieee_test (256.0);
  ieee_test (0.12345);
  ieee_test (234235.78907234);
  ieee_test (-512.0);
  ieee_test (-0.004321);
  ieee_test (1.2E-70);
  ieee_test (1.2E-316);
  ieee_test (4.9406564584124654E-324);
  ieee_test (- 4.9406564584124654E-324);
  ieee_test (- 0.0);
  ieee_test (- INFINITY);
  ieee_test (- NAN);
  ieee_test (INFINITY);
  ieee_test (NAN);
  return 0;
}
#endif