forked from torvalds/linux
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathsoftfloat.c
3435 lines (3164 loc) · 116 KB
/
softfloat.c
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
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
/*
===============================================================================
This C source file is part of the SoftFloat IEC/IEEE Floating-point
Arithmetic Package, Release 2.
Written by John R. Hauser. This work was made possible in part by the
International Computer Science Institute, located at Suite 600, 1947 Center
Street, Berkeley, California 94704. Funding was partially provided by the
National Science Foundation under grant MIP-9311980. The original version
of this code was written as part of a project to build a fixed-point vector
processor in collaboration with the University of California at Berkeley,
overseen by Profs. Nelson Morgan and John Wawrzynek. More information
is available through the web page
http://www.jhauser.us/arithmetic/SoftFloat-2b/SoftFloat-source.txt
THIS SOFTWARE IS DISTRIBUTED AS IS, FOR FREE. Although reasonable effort
has been made to avoid it, THIS SOFTWARE MAY CONTAIN FAULTS THAT WILL AT
TIMES RESULT IN INCORRECT BEHAVIOR. USE OF THIS SOFTWARE IS RESTRICTED TO
PERSONS AND ORGANIZATIONS WHO CAN AND WILL TAKE FULL RESPONSIBILITY FOR ANY
AND ALL LOSSES, COSTS, OR OTHER PROBLEMS ARISING FROM ITS USE.
Derivative works are acceptable, even for commercial purposes, so long as
(1) they include prominent notice that the work is derivative, and (2) they
include prominent notice akin to these three paragraphs for those parts of
this code that are retained.
===============================================================================
*/
#include <asm/div64.h>
#include "fpa11.h"
//#include "milieu.h"
//#include "softfloat.h"
/*
-------------------------------------------------------------------------------
Primitive arithmetic functions, including multi-word arithmetic, and
division and square root approximations. (Can be specialized to target if
desired.)
-------------------------------------------------------------------------------
*/
#include "softfloat-macros"
/*
-------------------------------------------------------------------------------
Functions and definitions to determine: (1) whether tininess for underflow
is detected before or after rounding by default, (2) what (if anything)
happens when exceptions are raised, (3) how signaling NaNs are distinguished
from quiet NaNs, (4) the default generated quiet NaNs, and (5) how NaNs
are propagated from function inputs to output. These details are target-
specific.
-------------------------------------------------------------------------------
*/
#include "softfloat-specialize"
/*
-------------------------------------------------------------------------------
Takes a 64-bit fixed-point value `absZ' with binary point between bits 6
and 7, and returns the properly rounded 32-bit integer corresponding to the
input. If `zSign' is nonzero, the input is negated before being converted
to an integer. Bit 63 of `absZ' must be zero. Ordinarily, the fixed-point
input is simply rounded to an integer, with the inexact exception raised if
the input cannot be represented exactly as an integer. If the fixed-point
input is too large, however, the invalid exception is raised and the largest
positive or negative integer is returned.
-------------------------------------------------------------------------------
*/
static int32 roundAndPackInt32( struct roundingData *roundData, flag zSign, bits64 absZ )
{
int8 roundingMode;
flag roundNearestEven;
int8 roundIncrement, roundBits;
int32 z;
roundingMode = roundData->mode;
roundNearestEven = ( roundingMode == float_round_nearest_even );
roundIncrement = 0x40;
if ( ! roundNearestEven ) {
if ( roundingMode == float_round_to_zero ) {
roundIncrement = 0;
}
else {
roundIncrement = 0x7F;
if ( zSign ) {
if ( roundingMode == float_round_up ) roundIncrement = 0;
}
else {
if ( roundingMode == float_round_down ) roundIncrement = 0;
}
}
}
roundBits = absZ & 0x7F;
absZ = ( absZ + roundIncrement )>>7;
absZ &= ~ ( ( ( roundBits ^ 0x40 ) == 0 ) & roundNearestEven );
z = absZ;
if ( zSign ) z = - z;
if ( ( absZ>>32 ) || ( z && ( ( z < 0 ) ^ zSign ) ) ) {
roundData->exception |= float_flag_invalid;
return zSign ? 0x80000000 : 0x7FFFFFFF;
}
if ( roundBits ) roundData->exception |= float_flag_inexact;
return z;
}
/*
-------------------------------------------------------------------------------
Returns the fraction bits of the single-precision floating-point value `a'.
-------------------------------------------------------------------------------
*/
INLINE bits32 extractFloat32Frac( float32 a )
{
return a & 0x007FFFFF;
}
/*
-------------------------------------------------------------------------------
Returns the exponent bits of the single-precision floating-point value `a'.
-------------------------------------------------------------------------------
*/
INLINE int16 extractFloat32Exp( float32 a )
{
return ( a>>23 ) & 0xFF;
}
/*
-------------------------------------------------------------------------------
Returns the sign bit of the single-precision floating-point value `a'.
-------------------------------------------------------------------------------
*/
#if 0 /* in softfloat.h */
INLINE flag extractFloat32Sign( float32 a )
{
return a>>31;
}
#endif
/*
-------------------------------------------------------------------------------
Normalizes the subnormal single-precision floating-point value represented
by the denormalized significand `aSig'. The normalized exponent and
significand are stored at the locations pointed to by `zExpPtr' and
`zSigPtr', respectively.
-------------------------------------------------------------------------------
*/
static void
normalizeFloat32Subnormal( bits32 aSig, int16 *zExpPtr, bits32 *zSigPtr )
{
int8 shiftCount;
shiftCount = countLeadingZeros32( aSig ) - 8;
*zSigPtr = aSig<<shiftCount;
*zExpPtr = 1 - shiftCount;
}
/*
-------------------------------------------------------------------------------
Packs the sign `zSign', exponent `zExp', and significand `zSig' into a
single-precision floating-point value, returning the result. After being
shifted into the proper positions, the three fields are simply added
together to form the result. This means that any integer portion of `zSig'
will be added into the exponent. Since a properly normalized significand
will have an integer portion equal to 1, the `zExp' input should be 1 less
than the desired result exponent whenever `zSig' is a complete, normalized
significand.
-------------------------------------------------------------------------------
*/
INLINE float32 packFloat32( flag zSign, int16 zExp, bits32 zSig )
{
#if 0
float32 f;
__asm__("@ packFloat32 \n\
mov %0, %1, asl #31 \n\
orr %0, %2, asl #23 \n\
orr %0, %3"
: /* no outputs */
: "g" (f), "g" (zSign), "g" (zExp), "g" (zSig)
: "cc");
return f;
#else
return ( ( (bits32) zSign )<<31 ) + ( ( (bits32) zExp )<<23 ) + zSig;
#endif
}
/*
-------------------------------------------------------------------------------
Takes an abstract floating-point value having sign `zSign', exponent `zExp',
and significand `zSig', and returns the proper single-precision floating-
point value corresponding to the abstract input. Ordinarily, the abstract
value is simply rounded and packed into the single-precision format, with
the inexact exception raised if the abstract input cannot be represented
exactly. If the abstract value is too large, however, the overflow and
inexact exceptions are raised and an infinity or maximal finite value is
returned. If the abstract value is too small, the input value is rounded to
a subnormal number, and the underflow and inexact exceptions are raised if
the abstract input cannot be represented exactly as a subnormal single-
precision floating-point number.
The input significand `zSig' has its binary point between bits 30
and 29, which is 7 bits to the left of the usual location. This shifted
significand must be normalized or smaller. If `zSig' is not normalized,
`zExp' must be 0; in that case, the result returned is a subnormal number,
and it must not require rounding. In the usual case that `zSig' is
normalized, `zExp' must be 1 less than the ``true'' floating-point exponent.
The handling of underflow and overflow follows the IEC/IEEE Standard for
Binary Floating-point Arithmetic.
-------------------------------------------------------------------------------
*/
static float32 roundAndPackFloat32( struct roundingData *roundData, flag zSign, int16 zExp, bits32 zSig )
{
int8 roundingMode;
flag roundNearestEven;
int8 roundIncrement, roundBits;
flag isTiny;
roundingMode = roundData->mode;
roundNearestEven = ( roundingMode == float_round_nearest_even );
roundIncrement = 0x40;
if ( ! roundNearestEven ) {
if ( roundingMode == float_round_to_zero ) {
roundIncrement = 0;
}
else {
roundIncrement = 0x7F;
if ( zSign ) {
if ( roundingMode == float_round_up ) roundIncrement = 0;
}
else {
if ( roundingMode == float_round_down ) roundIncrement = 0;
}
}
}
roundBits = zSig & 0x7F;
if ( 0xFD <= (bits16) zExp ) {
if ( ( 0xFD < zExp )
|| ( ( zExp == 0xFD )
&& ( (sbits32) ( zSig + roundIncrement ) < 0 ) )
) {
roundData->exception |= float_flag_overflow | float_flag_inexact;
return packFloat32( zSign, 0xFF, 0 ) - ( roundIncrement == 0 );
}
if ( zExp < 0 ) {
isTiny =
( float_detect_tininess == float_tininess_before_rounding )
|| ( zExp < -1 )
|| ( zSig + roundIncrement < 0x80000000 );
shift32RightJamming( zSig, - zExp, &zSig );
zExp = 0;
roundBits = zSig & 0x7F;
if ( isTiny && roundBits ) roundData->exception |= float_flag_underflow;
}
}
if ( roundBits ) roundData->exception |= float_flag_inexact;
zSig = ( zSig + roundIncrement )>>7;
zSig &= ~ ( ( ( roundBits ^ 0x40 ) == 0 ) & roundNearestEven );
if ( zSig == 0 ) zExp = 0;
return packFloat32( zSign, zExp, zSig );
}
/*
-------------------------------------------------------------------------------
Takes an abstract floating-point value having sign `zSign', exponent `zExp',
and significand `zSig', and returns the proper single-precision floating-
point value corresponding to the abstract input. This routine is just like
`roundAndPackFloat32' except that `zSig' does not have to be normalized in
any way. In all cases, `zExp' must be 1 less than the ``true'' floating-
point exponent.
-------------------------------------------------------------------------------
*/
static float32
normalizeRoundAndPackFloat32( struct roundingData *roundData, flag zSign, int16 zExp, bits32 zSig )
{
int8 shiftCount;
shiftCount = countLeadingZeros32( zSig ) - 1;
return roundAndPackFloat32( roundData, zSign, zExp - shiftCount, zSig<<shiftCount );
}
/*
-------------------------------------------------------------------------------
Returns the fraction bits of the double-precision floating-point value `a'.
-------------------------------------------------------------------------------
*/
INLINE bits64 extractFloat64Frac( float64 a )
{
return a & LIT64( 0x000FFFFFFFFFFFFF );
}
/*
-------------------------------------------------------------------------------
Returns the exponent bits of the double-precision floating-point value `a'.
-------------------------------------------------------------------------------
*/
INLINE int16 extractFloat64Exp( float64 a )
{
return ( a>>52 ) & 0x7FF;
}
/*
-------------------------------------------------------------------------------
Returns the sign bit of the double-precision floating-point value `a'.
-------------------------------------------------------------------------------
*/
#if 0 /* in softfloat.h */
INLINE flag extractFloat64Sign( float64 a )
{
return a>>63;
}
#endif
/*
-------------------------------------------------------------------------------
Normalizes the subnormal double-precision floating-point value represented
by the denormalized significand `aSig'. The normalized exponent and
significand are stored at the locations pointed to by `zExpPtr' and
`zSigPtr', respectively.
-------------------------------------------------------------------------------
*/
static void
normalizeFloat64Subnormal( bits64 aSig, int16 *zExpPtr, bits64 *zSigPtr )
{
int8 shiftCount;
shiftCount = countLeadingZeros64( aSig ) - 11;
*zSigPtr = aSig<<shiftCount;
*zExpPtr = 1 - shiftCount;
}
/*
-------------------------------------------------------------------------------
Packs the sign `zSign', exponent `zExp', and significand `zSig' into a
double-precision floating-point value, returning the result. After being
shifted into the proper positions, the three fields are simply added
together to form the result. This means that any integer portion of `zSig'
will be added into the exponent. Since a properly normalized significand
will have an integer portion equal to 1, the `zExp' input should be 1 less
than the desired result exponent whenever `zSig' is a complete, normalized
significand.
-------------------------------------------------------------------------------
*/
INLINE float64 packFloat64( flag zSign, int16 zExp, bits64 zSig )
{
return ( ( (bits64) zSign )<<63 ) + ( ( (bits64) zExp )<<52 ) + zSig;
}
/*
-------------------------------------------------------------------------------
Takes an abstract floating-point value having sign `zSign', exponent `zExp',
and significand `zSig', and returns the proper double-precision floating-
point value corresponding to the abstract input. Ordinarily, the abstract
value is simply rounded and packed into the double-precision format, with
the inexact exception raised if the abstract input cannot be represented
exactly. If the abstract value is too large, however, the overflow and
inexact exceptions are raised and an infinity or maximal finite value is
returned. If the abstract value is too small, the input value is rounded to
a subnormal number, and the underflow and inexact exceptions are raised if
the abstract input cannot be represented exactly as a subnormal double-
precision floating-point number.
The input significand `zSig' has its binary point between bits 62
and 61, which is 10 bits to the left of the usual location. This shifted
significand must be normalized or smaller. If `zSig' is not normalized,
`zExp' must be 0; in that case, the result returned is a subnormal number,
and it must not require rounding. In the usual case that `zSig' is
normalized, `zExp' must be 1 less than the ``true'' floating-point exponent.
The handling of underflow and overflow follows the IEC/IEEE Standard for
Binary Floating-point Arithmetic.
-------------------------------------------------------------------------------
*/
static float64 roundAndPackFloat64( struct roundingData *roundData, flag zSign, int16 zExp, bits64 zSig )
{
int8 roundingMode;
flag roundNearestEven;
int16 roundIncrement, roundBits;
flag isTiny;
roundingMode = roundData->mode;
roundNearestEven = ( roundingMode == float_round_nearest_even );
roundIncrement = 0x200;
if ( ! roundNearestEven ) {
if ( roundingMode == float_round_to_zero ) {
roundIncrement = 0;
}
else {
roundIncrement = 0x3FF;
if ( zSign ) {
if ( roundingMode == float_round_up ) roundIncrement = 0;
}
else {
if ( roundingMode == float_round_down ) roundIncrement = 0;
}
}
}
roundBits = zSig & 0x3FF;
if ( 0x7FD <= (bits16) zExp ) {
if ( ( 0x7FD < zExp )
|| ( ( zExp == 0x7FD )
&& ( (sbits64) ( zSig + roundIncrement ) < 0 ) )
) {
//register int lr = __builtin_return_address(0);
//printk("roundAndPackFloat64 called from 0x%08x\n",lr);
roundData->exception |= float_flag_overflow | float_flag_inexact;
return packFloat64( zSign, 0x7FF, 0 ) - ( roundIncrement == 0 );
}
if ( zExp < 0 ) {
isTiny =
( float_detect_tininess == float_tininess_before_rounding )
|| ( zExp < -1 )
|| ( zSig + roundIncrement < LIT64( 0x8000000000000000 ) );
shift64RightJamming( zSig, - zExp, &zSig );
zExp = 0;
roundBits = zSig & 0x3FF;
if ( isTiny && roundBits ) roundData->exception |= float_flag_underflow;
}
}
if ( roundBits ) roundData->exception |= float_flag_inexact;
zSig = ( zSig + roundIncrement )>>10;
zSig &= ~ ( ( ( roundBits ^ 0x200 ) == 0 ) & roundNearestEven );
if ( zSig == 0 ) zExp = 0;
return packFloat64( zSign, zExp, zSig );
}
/*
-------------------------------------------------------------------------------
Takes an abstract floating-point value having sign `zSign', exponent `zExp',
and significand `zSig', and returns the proper double-precision floating-
point value corresponding to the abstract input. This routine is just like
`roundAndPackFloat64' except that `zSig' does not have to be normalized in
any way. In all cases, `zExp' must be 1 less than the ``true'' floating-
point exponent.
-------------------------------------------------------------------------------
*/
static float64
normalizeRoundAndPackFloat64( struct roundingData *roundData, flag zSign, int16 zExp, bits64 zSig )
{
int8 shiftCount;
shiftCount = countLeadingZeros64( zSig ) - 1;
return roundAndPackFloat64( roundData, zSign, zExp - shiftCount, zSig<<shiftCount );
}
#ifdef FLOATX80
/*
-------------------------------------------------------------------------------
Returns the fraction bits of the extended double-precision floating-point
value `a'.
-------------------------------------------------------------------------------
*/
INLINE bits64 extractFloatx80Frac( floatx80 a )
{
return a.low;
}
/*
-------------------------------------------------------------------------------
Returns the exponent bits of the extended double-precision floating-point
value `a'.
-------------------------------------------------------------------------------
*/
INLINE int32 extractFloatx80Exp( floatx80 a )
{
return a.high & 0x7FFF;
}
/*
-------------------------------------------------------------------------------
Returns the sign bit of the extended double-precision floating-point value
`a'.
-------------------------------------------------------------------------------
*/
INLINE flag extractFloatx80Sign( floatx80 a )
{
return a.high>>15;
}
/*
-------------------------------------------------------------------------------
Normalizes the subnormal extended double-precision floating-point value
represented by the denormalized significand `aSig'. The normalized exponent
and significand are stored at the locations pointed to by `zExpPtr' and
`zSigPtr', respectively.
-------------------------------------------------------------------------------
*/
static void
normalizeFloatx80Subnormal( bits64 aSig, int32 *zExpPtr, bits64 *zSigPtr )
{
int8 shiftCount;
shiftCount = countLeadingZeros64( aSig );
*zSigPtr = aSig<<shiftCount;
*zExpPtr = 1 - shiftCount;
}
/*
-------------------------------------------------------------------------------
Packs the sign `zSign', exponent `zExp', and significand `zSig' into an
extended double-precision floating-point value, returning the result.
-------------------------------------------------------------------------------
*/
INLINE floatx80 packFloatx80( flag zSign, int32 zExp, bits64 zSig )
{
floatx80 z;
z.low = zSig;
z.high = ( ( (bits16) zSign )<<15 ) + zExp;
z.__padding = 0;
return z;
}
/*
-------------------------------------------------------------------------------
Takes an abstract floating-point value having sign `zSign', exponent `zExp',
and extended significand formed by the concatenation of `zSig0' and `zSig1',
and returns the proper extended double-precision floating-point value
corresponding to the abstract input. Ordinarily, the abstract value is
rounded and packed into the extended double-precision format, with the
inexact exception raised if the abstract input cannot be represented
exactly. If the abstract value is too large, however, the overflow and
inexact exceptions are raised and an infinity or maximal finite value is
returned. If the abstract value is too small, the input value is rounded to
a subnormal number, and the underflow and inexact exceptions are raised if
the abstract input cannot be represented exactly as a subnormal extended
double-precision floating-point number.
If `roundingPrecision' is 32 or 64, the result is rounded to the same
number of bits as single or double precision, respectively. Otherwise, the
result is rounded to the full precision of the extended double-precision
format.
The input significand must be normalized or smaller. If the input
significand is not normalized, `zExp' must be 0; in that case, the result
returned is a subnormal number, and it must not require rounding. The
handling of underflow and overflow follows the IEC/IEEE Standard for Binary
Floating-point Arithmetic.
-------------------------------------------------------------------------------
*/
static floatx80
roundAndPackFloatx80(
struct roundingData *roundData, flag zSign, int32 zExp, bits64 zSig0, bits64 zSig1
)
{
int8 roundingMode, roundingPrecision;
flag roundNearestEven, increment, isTiny;
int64 roundIncrement, roundMask, roundBits;
roundingMode = roundData->mode;
roundingPrecision = roundData->precision;
roundNearestEven = ( roundingMode == float_round_nearest_even );
if ( roundingPrecision == 80 ) goto precision80;
if ( roundingPrecision == 64 ) {
roundIncrement = LIT64( 0x0000000000000400 );
roundMask = LIT64( 0x00000000000007FF );
}
else if ( roundingPrecision == 32 ) {
roundIncrement = LIT64( 0x0000008000000000 );
roundMask = LIT64( 0x000000FFFFFFFFFF );
}
else {
goto precision80;
}
zSig0 |= ( zSig1 != 0 );
if ( ! roundNearestEven ) {
if ( roundingMode == float_round_to_zero ) {
roundIncrement = 0;
}
else {
roundIncrement = roundMask;
if ( zSign ) {
if ( roundingMode == float_round_up ) roundIncrement = 0;
}
else {
if ( roundingMode == float_round_down ) roundIncrement = 0;
}
}
}
roundBits = zSig0 & roundMask;
if ( 0x7FFD <= (bits32) ( zExp - 1 ) ) {
if ( ( 0x7FFE < zExp )
|| ( ( zExp == 0x7FFE ) && ( zSig0 + roundIncrement < zSig0 ) )
) {
goto overflow;
}
if ( zExp <= 0 ) {
isTiny =
( float_detect_tininess == float_tininess_before_rounding )
|| ( zExp < 0 )
|| ( zSig0 <= zSig0 + roundIncrement );
shift64RightJamming( zSig0, 1 - zExp, &zSig0 );
zExp = 0;
roundBits = zSig0 & roundMask;
if ( isTiny && roundBits ) roundData->exception |= float_flag_underflow;
if ( roundBits ) roundData->exception |= float_flag_inexact;
zSig0 += roundIncrement;
if ( (sbits64) zSig0 < 0 ) zExp = 1;
roundIncrement = roundMask + 1;
if ( roundNearestEven && ( roundBits<<1 == roundIncrement ) ) {
roundMask |= roundIncrement;
}
zSig0 &= ~ roundMask;
return packFloatx80( zSign, zExp, zSig0 );
}
}
if ( roundBits ) roundData->exception |= float_flag_inexact;
zSig0 += roundIncrement;
if ( zSig0 < roundIncrement ) {
++zExp;
zSig0 = LIT64( 0x8000000000000000 );
}
roundIncrement = roundMask + 1;
if ( roundNearestEven && ( roundBits<<1 == roundIncrement ) ) {
roundMask |= roundIncrement;
}
zSig0 &= ~ roundMask;
if ( zSig0 == 0 ) zExp = 0;
return packFloatx80( zSign, zExp, zSig0 );
precision80:
increment = ( (sbits64) zSig1 < 0 );
if ( ! roundNearestEven ) {
if ( roundingMode == float_round_to_zero ) {
increment = 0;
}
else {
if ( zSign ) {
increment = ( roundingMode == float_round_down ) && zSig1;
}
else {
increment = ( roundingMode == float_round_up ) && zSig1;
}
}
}
if ( 0x7FFD <= (bits32) ( zExp - 1 ) ) {
if ( ( 0x7FFE < zExp )
|| ( ( zExp == 0x7FFE )
&& ( zSig0 == LIT64( 0xFFFFFFFFFFFFFFFF ) )
&& increment
)
) {
roundMask = 0;
overflow:
roundData->exception |= float_flag_overflow | float_flag_inexact;
if ( ( roundingMode == float_round_to_zero )
|| ( zSign && ( roundingMode == float_round_up ) )
|| ( ! zSign && ( roundingMode == float_round_down ) )
) {
return packFloatx80( zSign, 0x7FFE, ~ roundMask );
}
return packFloatx80( zSign, 0x7FFF, LIT64( 0x8000000000000000 ) );
}
if ( zExp <= 0 ) {
isTiny =
( float_detect_tininess == float_tininess_before_rounding )
|| ( zExp < 0 )
|| ! increment
|| ( zSig0 < LIT64( 0xFFFFFFFFFFFFFFFF ) );
shift64ExtraRightJamming( zSig0, zSig1, 1 - zExp, &zSig0, &zSig1 );
zExp = 0;
if ( isTiny && zSig1 ) roundData->exception |= float_flag_underflow;
if ( zSig1 ) roundData->exception |= float_flag_inexact;
if ( roundNearestEven ) {
increment = ( (sbits64) zSig1 < 0 );
}
else {
if ( zSign ) {
increment = ( roundingMode == float_round_down ) && zSig1;
}
else {
increment = ( roundingMode == float_round_up ) && zSig1;
}
}
if ( increment ) {
++zSig0;
zSig0 &= ~ ( ( zSig1 + zSig1 == 0 ) & roundNearestEven );
if ( (sbits64) zSig0 < 0 ) zExp = 1;
}
return packFloatx80( zSign, zExp, zSig0 );
}
}
if ( zSig1 ) roundData->exception |= float_flag_inexact;
if ( increment ) {
++zSig0;
if ( zSig0 == 0 ) {
++zExp;
zSig0 = LIT64( 0x8000000000000000 );
}
else {
zSig0 &= ~ ( ( zSig1 + zSig1 == 0 ) & roundNearestEven );
}
}
else {
if ( zSig0 == 0 ) zExp = 0;
}
return packFloatx80( zSign, zExp, zSig0 );
}
/*
-------------------------------------------------------------------------------
Takes an abstract floating-point value having sign `zSign', exponent
`zExp', and significand formed by the concatenation of `zSig0' and `zSig1',
and returns the proper extended double-precision floating-point value
corresponding to the abstract input. This routine is just like
`roundAndPackFloatx80' except that the input significand does not have to be
normalized.
-------------------------------------------------------------------------------
*/
static floatx80
normalizeRoundAndPackFloatx80(
struct roundingData *roundData, flag zSign, int32 zExp, bits64 zSig0, bits64 zSig1
)
{
int8 shiftCount;
if ( zSig0 == 0 ) {
zSig0 = zSig1;
zSig1 = 0;
zExp -= 64;
}
shiftCount = countLeadingZeros64( zSig0 );
shortShift128Left( zSig0, zSig1, shiftCount, &zSig0, &zSig1 );
zExp -= shiftCount;
return
roundAndPackFloatx80( roundData, zSign, zExp, zSig0, zSig1 );
}
#endif
/*
-------------------------------------------------------------------------------
Returns the result of converting the 32-bit two's complement integer `a' to
the single-precision floating-point format. The conversion is performed
according to the IEC/IEEE Standard for Binary Floating-point Arithmetic.
-------------------------------------------------------------------------------
*/
float32 int32_to_float32(struct roundingData *roundData, int32 a)
{
flag zSign;
if ( a == 0 ) return 0;
if ( a == 0x80000000 ) return packFloat32( 1, 0x9E, 0 );
zSign = ( a < 0 );
return normalizeRoundAndPackFloat32( roundData, zSign, 0x9C, zSign ? - a : a );
}
/*
-------------------------------------------------------------------------------
Returns the result of converting the 32-bit two's complement integer `a' to
the double-precision floating-point format. The conversion is performed
according to the IEC/IEEE Standard for Binary Floating-point Arithmetic.
-------------------------------------------------------------------------------
*/
float64 int32_to_float64( int32 a )
{
flag aSign;
uint32 absA;
int8 shiftCount;
bits64 zSig;
if ( a == 0 ) return 0;
aSign = ( a < 0 );
absA = aSign ? - a : a;
shiftCount = countLeadingZeros32( absA ) + 21;
zSig = absA;
return packFloat64( aSign, 0x432 - shiftCount, zSig<<shiftCount );
}
#ifdef FLOATX80
/*
-------------------------------------------------------------------------------
Returns the result of converting the 32-bit two's complement integer `a'
to the extended double-precision floating-point format. The conversion
is performed according to the IEC/IEEE Standard for Binary Floating-point
Arithmetic.
-------------------------------------------------------------------------------
*/
floatx80 int32_to_floatx80( int32 a )
{
flag zSign;
uint32 absA;
int8 shiftCount;
bits64 zSig;
if ( a == 0 ) return packFloatx80( 0, 0, 0 );
zSign = ( a < 0 );
absA = zSign ? - a : a;
shiftCount = countLeadingZeros32( absA ) + 32;
zSig = absA;
return packFloatx80( zSign, 0x403E - shiftCount, zSig<<shiftCount );
}
#endif
/*
-------------------------------------------------------------------------------
Returns the result of converting the single-precision floating-point value
`a' to the 32-bit two's complement integer format. The conversion is
performed according to the IEC/IEEE Standard for Binary Floating-point
Arithmetic---which means in particular that the conversion is rounded
according to the current rounding mode. If `a' is a NaN, the largest
positive integer is returned. Otherwise, if the conversion overflows, the
largest integer with the same sign as `a' is returned.
-------------------------------------------------------------------------------
*/
int32 float32_to_int32( struct roundingData *roundData, float32 a )
{
flag aSign;
int16 aExp, shiftCount;
bits32 aSig;
bits64 zSig;
aSig = extractFloat32Frac( a );
aExp = extractFloat32Exp( a );
aSign = extractFloat32Sign( a );
if ( ( aExp == 0x7FF ) && aSig ) aSign = 0;
if ( aExp ) aSig |= 0x00800000;
shiftCount = 0xAF - aExp;
zSig = aSig;
zSig <<= 32;
if ( 0 < shiftCount ) shift64RightJamming( zSig, shiftCount, &zSig );
return roundAndPackInt32( roundData, aSign, zSig );
}
/*
-------------------------------------------------------------------------------
Returns the result of converting the single-precision floating-point value
`a' to the 32-bit two's complement integer format. The conversion is
performed according to the IEC/IEEE Standard for Binary Floating-point
Arithmetic, except that the conversion is always rounded toward zero. If
`a' is a NaN, the largest positive integer is returned. Otherwise, if the
conversion overflows, the largest integer with the same sign as `a' is
returned.
-------------------------------------------------------------------------------
*/
int32 float32_to_int32_round_to_zero( float32 a )
{
flag aSign;
int16 aExp, shiftCount;
bits32 aSig;
int32 z;
aSig = extractFloat32Frac( a );
aExp = extractFloat32Exp( a );
aSign = extractFloat32Sign( a );
shiftCount = aExp - 0x9E;
if ( 0 <= shiftCount ) {
if ( a == 0xCF000000 ) return 0x80000000;
float_raise( float_flag_invalid );
if ( ! aSign || ( ( aExp == 0xFF ) && aSig ) ) return 0x7FFFFFFF;
return 0x80000000;
}
else if ( aExp <= 0x7E ) {
if ( aExp | aSig ) float_raise( float_flag_inexact );
return 0;
}
aSig = ( aSig | 0x00800000 )<<8;
z = aSig>>( - shiftCount );
if ( (bits32) ( aSig<<( shiftCount & 31 ) ) ) {
float_raise( float_flag_inexact );
}
return aSign ? - z : z;
}
/*
-------------------------------------------------------------------------------
Returns the result of converting the single-precision floating-point value
`a' to the double-precision floating-point format. The conversion is
performed according to the IEC/IEEE Standard for Binary Floating-point
Arithmetic.
-------------------------------------------------------------------------------
*/
float64 float32_to_float64( float32 a )
{
flag aSign;
int16 aExp;
bits32 aSig;
aSig = extractFloat32Frac( a );
aExp = extractFloat32Exp( a );
aSign = extractFloat32Sign( a );
if ( aExp == 0xFF ) {
if ( aSig ) return commonNaNToFloat64( float32ToCommonNaN( a ) );
return packFloat64( aSign, 0x7FF, 0 );
}
if ( aExp == 0 ) {
if ( aSig == 0 ) return packFloat64( aSign, 0, 0 );
normalizeFloat32Subnormal( aSig, &aExp, &aSig );
--aExp;
}
return packFloat64( aSign, aExp + 0x380, ( (bits64) aSig )<<29 );
}
#ifdef FLOATX80
/*
-------------------------------------------------------------------------------
Returns the result of converting the single-precision floating-point value
`a' to the extended double-precision floating-point format. The conversion
is performed according to the IEC/IEEE Standard for Binary Floating-point
Arithmetic.
-------------------------------------------------------------------------------
*/
floatx80 float32_to_floatx80( float32 a )
{
flag aSign;
int16 aExp;
bits32 aSig;
aSig = extractFloat32Frac( a );
aExp = extractFloat32Exp( a );
aSign = extractFloat32Sign( a );
if ( aExp == 0xFF ) {
if ( aSig ) return commonNaNToFloatx80( float32ToCommonNaN( a ) );
return packFloatx80( aSign, 0x7FFF, LIT64( 0x8000000000000000 ) );
}
if ( aExp == 0 ) {
if ( aSig == 0 ) return packFloatx80( aSign, 0, 0 );
normalizeFloat32Subnormal( aSig, &aExp, &aSig );
}
aSig |= 0x00800000;
return packFloatx80( aSign, aExp + 0x3F80, ( (bits64) aSig )<<40 );
}
#endif
/*
-------------------------------------------------------------------------------
Rounds the single-precision floating-point value `a' to an integer, and
returns the result as a single-precision floating-point value. The
operation is performed according to the IEC/IEEE Standard for Binary
Floating-point Arithmetic.
-------------------------------------------------------------------------------
*/
float32 float32_round_to_int( struct roundingData *roundData, float32 a )
{
flag aSign;
int16 aExp;
bits32 lastBitMask, roundBitsMask;
int8 roundingMode;
float32 z;
aExp = extractFloat32Exp( a );
if ( 0x96 <= aExp ) {
if ( ( aExp == 0xFF ) && extractFloat32Frac( a ) ) {
return propagateFloat32NaN( a, a );
}
return a;
}
roundingMode = roundData->mode;
if ( aExp <= 0x7E ) {
if ( (bits32) ( a<<1 ) == 0 ) return a;
roundData->exception |= float_flag_inexact;
aSign = extractFloat32Sign( a );
switch ( roundingMode ) {
case float_round_nearest_even:
if ( ( aExp == 0x7E ) && extractFloat32Frac( a ) ) {
return packFloat32( aSign, 0x7F, 0 );
}
break;
case float_round_down:
return aSign ? 0xBF800000 : 0;
case float_round_up:
return aSign ? 0x80000000 : 0x3F800000;
}
return packFloat32( aSign, 0, 0 );
}