-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathsoftfloat.c
8089 lines (7225 loc) · 271 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
/*
* QEMU float support
*
* The code in this source file is derived from release 2a of the SoftFloat
* IEC/IEEE Floating-point Arithmetic Package. Those parts of the code (and
* some later contributions) are provided under that license, as detailed below.
* It has subsequently been modified by contributors to the QEMU Project,
* so some portions are provided under:
* the SoftFloat-2a license
* the BSD license
* GPL-v2-or-later
*
* Any future contributions to this file after December 1st 2014 will be
* taken to be licensed under the Softfloat-2a license unless specifically
* indicated otherwise.
*/
/*
===============================================================================
This C source file is part of the SoftFloat IEC/IEEE Floating-point
Arithmetic Package, Release 2a.
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://HTTP.CS.Berkeley.EDU/~jhauser/
arithmetic/SoftFloat.html'.
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 four paragraphs for those parts of
this code that are retained.
===============================================================================
*/
/* BSD licensing:
* Copyright (c) 2006, Fabrice Bellard
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions are met:
*
* 1. Redistributions of source code must retain the above copyright notice,
* this list of conditions and the following disclaimer.
*
* 2. Redistributions in binary form must reproduce the above copyright notice,
* this list of conditions and the following disclaimer in the documentation
* and/or other materials provided with the distribution.
*
* 3. Neither the name of the copyright holder nor the names of its contributors
* may be used to endorse or promote products derived from this software without
* specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
* AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE
* LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
* THE POSSIBILITY OF SUCH DAMAGE.
*/
/* Portions of this work are licensed under the terms of the GNU GPL,
* version 2 or later. See the COPYING file in the top-level directory.
*/
/* softfloat (and in particular the code in softfloat-specialize.h) is
* target-dependent and needs the TARGET_* macros.
*/
#include "qemu/osdep.h"
#include <math.h>
#include "qemu/bitops.h"
#include "fpu/softfloat.h"
/* We only need stdlib for abort() */
/*----------------------------------------------------------------------------
| Primitive arithmetic functions, including multi-word arithmetic, and
| division and square root approximations. (Can be specialized to target if
| desired.)
*----------------------------------------------------------------------------*/
#include "fpu/softfloat-macros.h"
/*
* Hardfloat
*
* Fast emulation of guest FP instructions is challenging for two reasons.
* First, FP instruction semantics are similar but not identical, particularly
* when handling NaNs. Second, emulating at reasonable speed the guest FP
* exception flags is not trivial: reading the host's flags register with a
* feclearexcept & fetestexcept pair is slow [slightly slower than soft-fp],
* and trapping on every FP exception is not fast nor pleasant to work with.
*
* We address these challenges by leveraging the host FPU for a subset of the
* operations. To do this we expand on the idea presented in this paper:
*
* Guo, Yu-Chuan, et al. "Translating the ARM Neon and VFP instructions in a
* binary translator." Software: Practice and Experience 46.12 (2016):1591-1615.
*
* The idea is thus to leverage the host FPU to (1) compute FP operations
* and (2) identify whether FP exceptions occurred while avoiding
* expensive exception flag register accesses.
*
* An important optimization shown in the paper is that given that exception
* flags are rarely cleared by the guest, we can avoid recomputing some flags.
* This is particularly useful for the inexact flag, which is very frequently
* raised in floating-point workloads.
*
* We optimize the code further by deferring to soft-fp whenever FP exception
* detection might get hairy. Two examples: (1) when at least one operand is
* denormal/inf/NaN; (2) when operands are not guaranteed to lead to a 0 result
* and the result is < the minimum normal.
*/
#define GEN_INPUT_FLUSH__NOCHECK(name, soft_t) \
static inline void name(soft_t *a, float_status *s) \
{ \
if (unlikely(soft_t ## _is_denormal(*a))) { \
*a = soft_t ## _set_sign(soft_t ## _zero, \
soft_t ## _is_neg(*a)); \
s->float_exception_flags |= float_flag_input_denormal; \
} \
}
GEN_INPUT_FLUSH__NOCHECK(float32_input_flush__nocheck, float32)
GEN_INPUT_FLUSH__NOCHECK(float64_input_flush__nocheck, float64)
#undef GEN_INPUT_FLUSH__NOCHECK
#define GEN_INPUT_FLUSH1(name, soft_t) \
static inline void name(soft_t *a, float_status *s) \
{ \
if (likely(!s->flush_inputs_to_zero)) { \
return; \
} \
soft_t ## _input_flush__nocheck(a, s); \
}
GEN_INPUT_FLUSH1(float32_input_flush1, float32)
GEN_INPUT_FLUSH1(float64_input_flush1, float64)
#undef GEN_INPUT_FLUSH1
#define GEN_INPUT_FLUSH2(name, soft_t) \
static inline void name(soft_t *a, soft_t *b, float_status *s) \
{ \
if (likely(!s->flush_inputs_to_zero)) { \
return; \
} \
soft_t ## _input_flush__nocheck(a, s); \
soft_t ## _input_flush__nocheck(b, s); \
}
GEN_INPUT_FLUSH2(float32_input_flush2, float32)
GEN_INPUT_FLUSH2(float64_input_flush2, float64)
#undef GEN_INPUT_FLUSH2
#define GEN_INPUT_FLUSH3(name, soft_t) \
static inline void name(soft_t *a, soft_t *b, soft_t *c, float_status *s) \
{ \
if (likely(!s->flush_inputs_to_zero)) { \
return; \
} \
soft_t ## _input_flush__nocheck(a, s); \
soft_t ## _input_flush__nocheck(b, s); \
soft_t ## _input_flush__nocheck(c, s); \
}
GEN_INPUT_FLUSH3(float32_input_flush3, float32)
GEN_INPUT_FLUSH3(float64_input_flush3, float64)
#undef GEN_INPUT_FLUSH3
/*
* Choose whether to use fpclassify or float32/64_* primitives in the generated
* hardfloat functions. Each combination of number of inputs and float size
* gets its own value.
*/
#if defined(__x86_64__)
# define QEMU_HARDFLOAT_1F32_USE_FP 0
# define QEMU_HARDFLOAT_1F64_USE_FP 1
# define QEMU_HARDFLOAT_2F32_USE_FP 0
# define QEMU_HARDFLOAT_2F64_USE_FP 1
# define QEMU_HARDFLOAT_3F32_USE_FP 0
# define QEMU_HARDFLOAT_3F64_USE_FP 1
#else
# define QEMU_HARDFLOAT_1F32_USE_FP 0
# define QEMU_HARDFLOAT_1F64_USE_FP 0
# define QEMU_HARDFLOAT_2F32_USE_FP 0
# define QEMU_HARDFLOAT_2F64_USE_FP 0
# define QEMU_HARDFLOAT_3F32_USE_FP 0
# define QEMU_HARDFLOAT_3F64_USE_FP 0
#endif
/*
* QEMU_HARDFLOAT_USE_ISINF chooses whether to use isinf() over
* float{32,64}_is_infinity when !USE_FP.
* On x86_64/aarch64, using the former over the latter can yield a ~6% speedup.
* On power64 however, using isinf() reduces fp-bench performance by up to 50%.
*/
#if defined(__x86_64__) || defined(__aarch64__)
# define QEMU_HARDFLOAT_USE_ISINF 1
#else
# define QEMU_HARDFLOAT_USE_ISINF 0
#endif
/*
* Some targets clear the FP flags before most FP operations. This prevents
* the use of hardfloat, since hardfloat relies on the inexact flag being
* already set.
*/
#if defined(TARGET_PPC) || defined(__FAST_MATH__)
# if defined(__FAST_MATH__)
# warning disabling hardfloat due to -ffast-math: hardfloat requires an exact \
IEEE implementation
# endif
# define QEMU_NO_HARDFLOAT 1
# define QEMU_SOFTFLOAT_ATTR QEMU_FLATTEN
#else
# define QEMU_NO_HARDFLOAT 0
# define QEMU_SOFTFLOAT_ATTR QEMU_FLATTEN __attribute__((noinline))
#endif
static inline bool can_use_fpu(const float_status *s)
{
if (QEMU_NO_HARDFLOAT) {
return false;
}
return likely(s->float_exception_flags & float_flag_inexact &&
s->float_rounding_mode == float_round_nearest_even);
}
/*
* Hardfloat generation functions. Each operation can have two flavors:
* either using softfloat primitives (e.g. float32_is_zero_or_normal) for
* most condition checks, or native ones (e.g. fpclassify).
*
* The flavor is chosen by the callers. Instead of using macros, we rely on the
* compiler to propagate constants and inline everything into the callers.
*
* We only generate functions for operations with two inputs, since only
* these are common enough to justify consolidating them into common code.
*/
typedef union {
float32 s;
float h;
} union_float32;
typedef union {
float64 s;
double h;
} union_float64;
typedef bool (*f32_check_fn)(union_float32 a, union_float32 b);
typedef bool (*f64_check_fn)(union_float64 a, union_float64 b);
typedef float32 (*soft_f32_op2_fn)(float32 a, float32 b, float_status *s);
typedef float64 (*soft_f64_op2_fn)(float64 a, float64 b, float_status *s);
typedef float (*hard_f32_op2_fn)(float a, float b);
typedef double (*hard_f64_op2_fn)(double a, double b);
/* 2-input is-zero-or-normal */
static inline bool f32_is_zon2(union_float32 a, union_float32 b)
{
if (QEMU_HARDFLOAT_2F32_USE_FP) {
/*
* Not using a temp variable for consecutive fpclassify calls ends up
* generating faster code.
*/
return (fpclassify(a.h) == FP_NORMAL || fpclassify(a.h) == FP_ZERO) &&
(fpclassify(b.h) == FP_NORMAL || fpclassify(b.h) == FP_ZERO);
}
return float32_is_zero_or_normal(a.s) &&
float32_is_zero_or_normal(b.s);
}
static inline bool f64_is_zon2(union_float64 a, union_float64 b)
{
if (QEMU_HARDFLOAT_2F64_USE_FP) {
return (fpclassify(a.h) == FP_NORMAL || fpclassify(a.h) == FP_ZERO) &&
(fpclassify(b.h) == FP_NORMAL || fpclassify(b.h) == FP_ZERO);
}
return float64_is_zero_or_normal(a.s) &&
float64_is_zero_or_normal(b.s);
}
/* 3-input is-zero-or-normal */
static inline
bool f32_is_zon3(union_float32 a, union_float32 b, union_float32 c)
{
if (QEMU_HARDFLOAT_3F32_USE_FP) {
return (fpclassify(a.h) == FP_NORMAL || fpclassify(a.h) == FP_ZERO) &&
(fpclassify(b.h) == FP_NORMAL || fpclassify(b.h) == FP_ZERO) &&
(fpclassify(c.h) == FP_NORMAL || fpclassify(c.h) == FP_ZERO);
}
return float32_is_zero_or_normal(a.s) &&
float32_is_zero_or_normal(b.s) &&
float32_is_zero_or_normal(c.s);
}
static inline
bool f64_is_zon3(union_float64 a, union_float64 b, union_float64 c)
{
if (QEMU_HARDFLOAT_3F64_USE_FP) {
return (fpclassify(a.h) == FP_NORMAL || fpclassify(a.h) == FP_ZERO) &&
(fpclassify(b.h) == FP_NORMAL || fpclassify(b.h) == FP_ZERO) &&
(fpclassify(c.h) == FP_NORMAL || fpclassify(c.h) == FP_ZERO);
}
return float64_is_zero_or_normal(a.s) &&
float64_is_zero_or_normal(b.s) &&
float64_is_zero_or_normal(c.s);
}
static inline bool f32_is_inf(union_float32 a)
{
if (QEMU_HARDFLOAT_USE_ISINF) {
return isinf(a.h);
}
return float32_is_infinity(a.s);
}
static inline bool f64_is_inf(union_float64 a)
{
if (QEMU_HARDFLOAT_USE_ISINF) {
return isinf(a.h);
}
return float64_is_infinity(a.s);
}
/* Note: @fast_test and @post can be NULL */
static inline float32
float32_gen2(float32 xa, float32 xb, float_status *s,
hard_f32_op2_fn hard, soft_f32_op2_fn soft,
f32_check_fn pre, f32_check_fn post,
f32_check_fn fast_test, soft_f32_op2_fn fast_op)
{
union_float32 ua, ub, ur;
ua.s = xa;
ub.s = xb;
if (unlikely(!can_use_fpu(s))) {
goto soft;
}
float32_input_flush2(&ua.s, &ub.s, s);
if (unlikely(!pre(ua, ub))) {
goto soft;
}
if (fast_test && fast_test(ua, ub)) {
return fast_op(ua.s, ub.s, s);
}
ur.h = hard(ua.h, ub.h);
if (unlikely(f32_is_inf(ur))) {
s->float_exception_flags |= float_flag_overflow;
} else if (unlikely(fabsf(ur.h) <= FLT_MIN)) {
if (post == NULL || post(ua, ub)) {
goto soft;
}
}
return ur.s;
soft:
return soft(ua.s, ub.s, s);
}
static inline float64
float64_gen2(float64 xa, float64 xb, float_status *s,
hard_f64_op2_fn hard, soft_f64_op2_fn soft,
f64_check_fn pre, f64_check_fn post,
f64_check_fn fast_test, soft_f64_op2_fn fast_op)
{
union_float64 ua, ub, ur;
ua.s = xa;
ub.s = xb;
if (unlikely(!can_use_fpu(s))) {
goto soft;
}
float64_input_flush2(&ua.s, &ub.s, s);
if (unlikely(!pre(ua, ub))) {
goto soft;
}
if (fast_test && fast_test(ua, ub)) {
return fast_op(ua.s, ub.s, s);
}
ur.h = hard(ua.h, ub.h);
if (unlikely(f64_is_inf(ur))) {
s->float_exception_flags |= float_flag_overflow;
} else if (unlikely(fabs(ur.h) <= DBL_MIN)) {
if (post == NULL || post(ua, ub)) {
goto soft;
}
}
return ur.s;
soft:
return soft(ua.s, ub.s, s);
}
/*----------------------------------------------------------------------------
| Returns the fraction bits of the single-precision floating-point value `a'.
*----------------------------------------------------------------------------*/
static inline uint32_t extractFloat32Frac(float32 a)
{
return float32_val(a) & 0x007FFFFF;
}
/*----------------------------------------------------------------------------
| Returns the exponent bits of the single-precision floating-point value `a'.
*----------------------------------------------------------------------------*/
static inline int extractFloat32Exp(float32 a)
{
return (float32_val(a) >> 23) & 0xFF;
}
/*----------------------------------------------------------------------------
| Returns the sign bit of the single-precision floating-point value `a'.
*----------------------------------------------------------------------------*/
static inline flag extractFloat32Sign(float32 a)
{
return float32_val(a) >> 31;
}
/*----------------------------------------------------------------------------
| Returns the fraction bits of the double-precision floating-point value `a'.
*----------------------------------------------------------------------------*/
static inline uint64_t extractFloat64Frac(float64 a)
{
return float64_val(a) & LIT64(0x000FFFFFFFFFFFFF);
}
/*----------------------------------------------------------------------------
| Returns the exponent bits of the double-precision floating-point value `a'.
*----------------------------------------------------------------------------*/
static inline int extractFloat64Exp(float64 a)
{
return (float64_val(a) >> 52) & 0x7FF;
}
/*----------------------------------------------------------------------------
| Returns the sign bit of the double-precision floating-point value `a'.
*----------------------------------------------------------------------------*/
static inline flag extractFloat64Sign(float64 a)
{
return float64_val(a) >> 63;
}
/*
* Classify a floating point number. Everything above float_class_qnan
* is a NaN so cls >= float_class_qnan is any NaN.
*/
typedef enum __attribute__ ((__packed__)) {
float_class_unclassified,
float_class_zero,
float_class_normal,
float_class_inf,
float_class_qnan, /* all NaNs from here */
float_class_snan,
} FloatClass;
/* Simple helpers for checking if, or what kind of, NaN we have */
static inline __attribute__((unused)) bool is_nan(FloatClass c)
{
return unlikely(c >= float_class_qnan);
}
static inline __attribute__((unused)) bool is_snan(FloatClass c)
{
return c == float_class_snan;
}
static inline __attribute__((unused)) bool is_qnan(FloatClass c)
{
return c == float_class_qnan;
}
/*
* Structure holding all of the decomposed parts of a float. The
* exponent is unbiased and the fraction is normalized. All
* calculations are done with a 64 bit fraction and then rounded as
* appropriate for the final format.
*
* Thanks to the packed FloatClass a decent compiler should be able to
* fit the whole structure into registers and avoid using the stack
* for parameter passing.
*/
typedef struct {
uint64_t frac;
int32_t exp;
FloatClass cls;
bool sign;
} FloatParts;
#define DECOMPOSED_BINARY_POINT (64 - 2)
#define DECOMPOSED_IMPLICIT_BIT (1ull << DECOMPOSED_BINARY_POINT)
#define DECOMPOSED_OVERFLOW_BIT (DECOMPOSED_IMPLICIT_BIT << 1)
/* Structure holding all of the relevant parameters for a format.
* exp_size: the size of the exponent field
* exp_bias: the offset applied to the exponent field
* exp_max: the maximum normalised exponent
* frac_size: the size of the fraction field
* frac_shift: shift to normalise the fraction with DECOMPOSED_BINARY_POINT
* The following are computed based the size of fraction
* frac_lsb: least significant bit of fraction
* frac_lsbm1: the bit below the least significant bit (for rounding)
* round_mask/roundeven_mask: masks used for rounding
* The following optional modifiers are available:
* arm_althp: handle ARM Alternative Half Precision
*/
typedef struct {
int exp_size;
int exp_bias;
int exp_max;
int frac_size;
int frac_shift;
uint64_t frac_lsb;
uint64_t frac_lsbm1;
uint64_t round_mask;
uint64_t roundeven_mask;
bool arm_althp;
} FloatFmt;
/* Expand fields based on the size of exponent and fraction */
#define FLOAT_PARAMS(E, F) \
.exp_size = E, \
.exp_bias = ((1 << E) - 1) >> 1, \
.exp_max = (1 << E) - 1, \
.frac_size = F, \
.frac_shift = DECOMPOSED_BINARY_POINT - F, \
.frac_lsb = 1ull << (DECOMPOSED_BINARY_POINT - F), \
.frac_lsbm1 = 1ull << ((DECOMPOSED_BINARY_POINT - F) - 1), \
.round_mask = (1ull << (DECOMPOSED_BINARY_POINT - F)) - 1, \
.roundeven_mask = (2ull << (DECOMPOSED_BINARY_POINT - F)) - 1
static const FloatFmt float16_params = {
FLOAT_PARAMS(5, 10)
};
static const FloatFmt float16_params_ahp = {
FLOAT_PARAMS(5, 10),
.arm_althp = true
};
static const FloatFmt float32_params = {
FLOAT_PARAMS(8, 23)
};
static const FloatFmt float64_params = {
FLOAT_PARAMS(11, 52)
};
/* Unpack a float to parts, but do not canonicalize. */
static inline FloatParts unpack_raw(FloatFmt fmt, uint64_t raw)
{
const int sign_pos = fmt.frac_size + fmt.exp_size;
return (FloatParts) {
.cls = float_class_unclassified,
.sign = extract64(raw, sign_pos, 1),
.exp = extract64(raw, fmt.frac_size, fmt.exp_size),
.frac = extract64(raw, 0, fmt.frac_size),
};
}
static inline FloatParts float16_unpack_raw(float16 f)
{
return unpack_raw(float16_params, f);
}
static inline FloatParts float32_unpack_raw(float32 f)
{
return unpack_raw(float32_params, f);
}
static inline FloatParts float64_unpack_raw(float64 f)
{
return unpack_raw(float64_params, f);
}
/* Pack a float from parts, but do not canonicalize. */
static inline uint64_t pack_raw(FloatFmt fmt, FloatParts p)
{
const int sign_pos = fmt.frac_size + fmt.exp_size;
uint64_t ret = deposit64(p.frac, fmt.frac_size, fmt.exp_size, p.exp);
return deposit64(ret, sign_pos, 1, p.sign);
}
static inline float16 float16_pack_raw(FloatParts p)
{
return make_float16(pack_raw(float16_params, p));
}
static inline float32 float32_pack_raw(FloatParts p)
{
return make_float32(pack_raw(float32_params, p));
}
static inline float64 float64_pack_raw(FloatParts p)
{
return make_float64(pack_raw(float64_params, p));
}
/*----------------------------------------------------------------------------
| 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.h"
/* Canonicalize EXP and FRAC, setting CLS. */
static FloatParts sf_canonicalize(FloatParts part, const FloatFmt *parm,
float_status *status)
{
if (part.exp == parm->exp_max && !parm->arm_althp) {
if (part.frac == 0) {
part.cls = float_class_inf;
} else {
part.frac <<= parm->frac_shift;
part.cls = (parts_is_snan_frac(part.frac, status)
? float_class_snan : float_class_qnan);
}
} else if (part.exp == 0) {
if (likely(part.frac == 0)) {
part.cls = float_class_zero;
} else if (status->flush_inputs_to_zero) {
float_raise(float_flag_input_denormal, status);
part.cls = float_class_zero;
part.frac = 0;
} else {
int shift = clz64(part.frac) - 1;
part.cls = float_class_normal;
part.exp = parm->frac_shift - parm->exp_bias - shift + 1;
part.frac <<= shift;
}
} else {
part.cls = float_class_normal;
part.exp -= parm->exp_bias;
part.frac = DECOMPOSED_IMPLICIT_BIT + (part.frac << parm->frac_shift);
}
return part;
}
/* Round and uncanonicalize a floating-point number by parts. There
* are FRAC_SHIFT bits that may require rounding at the bottom of the
* fraction; these bits will be removed. The exponent will be biased
* by EXP_BIAS and must be bounded by [EXP_MAX-1, 0].
*/
static FloatParts round_canonical(FloatParts p, float_status *s,
const FloatFmt *parm)
{
const uint64_t frac_lsb = parm->frac_lsb;
const uint64_t frac_lsbm1 = parm->frac_lsbm1;
const uint64_t round_mask = parm->round_mask;
const uint64_t roundeven_mask = parm->roundeven_mask;
const int exp_max = parm->exp_max;
const int frac_shift = parm->frac_shift;
uint64_t frac, inc;
int exp, flags = 0;
bool overflow_norm;
frac = p.frac;
exp = p.exp;
switch (p.cls) {
case float_class_normal:
switch (s->float_rounding_mode) {
case float_round_nearest_even:
overflow_norm = false;
inc = ((frac & roundeven_mask) != frac_lsbm1 ? frac_lsbm1 : 0);
break;
case float_round_ties_away:
overflow_norm = false;
inc = frac_lsbm1;
break;
case float_round_to_zero:
overflow_norm = true;
inc = 0;
break;
case float_round_up:
inc = p.sign ? 0 : round_mask;
overflow_norm = p.sign;
break;
case float_round_down:
inc = p.sign ? round_mask : 0;
overflow_norm = !p.sign;
break;
case float_round_to_odd:
overflow_norm = true;
inc = frac & frac_lsb ? 0 : round_mask;
break;
default:
g_assert_not_reached();
}
exp += parm->exp_bias;
if (likely(exp > 0)) {
if (frac & round_mask) {
flags |= float_flag_inexact;
frac += inc;
if (frac & DECOMPOSED_OVERFLOW_BIT) {
frac >>= 1;
exp++;
}
}
frac >>= frac_shift;
if (parm->arm_althp) {
/* ARM Alt HP eschews Inf and NaN for a wider exponent. */
if (unlikely(exp > exp_max)) {
/* Overflow. Return the maximum normal. */
flags = float_flag_invalid;
exp = exp_max;
frac = -1;
}
} else if (unlikely(exp >= exp_max)) {
flags |= float_flag_overflow | float_flag_inexact;
if (overflow_norm) {
exp = exp_max - 1;
frac = -1;
} else {
p.cls = float_class_inf;
goto do_inf;
}
}
} else if (s->flush_to_zero) {
flags |= float_flag_output_denormal;
p.cls = float_class_zero;
goto do_zero;
} else {
bool is_tiny = (s->float_detect_tininess
== float_tininess_before_rounding)
|| (exp < 0)
|| !((frac + inc) & DECOMPOSED_OVERFLOW_BIT);
shift64RightJamming(frac, 1 - exp, &frac);
if (frac & round_mask) {
/* Need to recompute round-to-even. */
switch (s->float_rounding_mode) {
case float_round_nearest_even:
inc = ((frac & roundeven_mask) != frac_lsbm1
? frac_lsbm1 : 0);
break;
case float_round_to_odd:
inc = frac & frac_lsb ? 0 : round_mask;
break;
}
flags |= float_flag_inexact;
frac += inc;
}
exp = (frac & DECOMPOSED_IMPLICIT_BIT ? 1 : 0);
frac >>= frac_shift;
if (is_tiny && (flags & float_flag_inexact)) {
flags |= float_flag_underflow;
}
if (exp == 0 && frac == 0) {
p.cls = float_class_zero;
}
}
break;
case float_class_zero:
do_zero:
exp = 0;
frac = 0;
break;
case float_class_inf:
do_inf:
assert(!parm->arm_althp);
exp = exp_max;
frac = 0;
break;
case float_class_qnan:
case float_class_snan:
assert(!parm->arm_althp);
exp = exp_max;
frac >>= parm->frac_shift;
break;
default:
g_assert_not_reached();
}
float_raise(flags, s);
p.exp = exp;
p.frac = frac;
return p;
}
/* Explicit FloatFmt version */
static FloatParts float16a_unpack_canonical(float16 f, float_status *s,
const FloatFmt *params)
{
return sf_canonicalize(float16_unpack_raw(f), params, s);
}
static FloatParts float16_unpack_canonical(float16 f, float_status *s)
{
return float16a_unpack_canonical(f, s, &float16_params);
}
static float16 float16a_round_pack_canonical(FloatParts p, float_status *s,
const FloatFmt *params)
{
return float16_pack_raw(round_canonical(p, s, params));
}
static float16 float16_round_pack_canonical(FloatParts p, float_status *s)
{
return float16a_round_pack_canonical(p, s, &float16_params);
}
static FloatParts float32_unpack_canonical(float32 f, float_status *s)
{
return sf_canonicalize(float32_unpack_raw(f), &float32_params, s);
}
static float32 float32_round_pack_canonical(FloatParts p, float_status *s)
{
return float32_pack_raw(round_canonical(p, s, &float32_params));
}
static FloatParts float64_unpack_canonical(float64 f, float_status *s)
{
return sf_canonicalize(float64_unpack_raw(f), &float64_params, s);
}
static float64 float64_round_pack_canonical(FloatParts p, float_status *s)
{
return float64_pack_raw(round_canonical(p, s, &float64_params));
}
static FloatParts return_nan(FloatParts a, float_status *s)
{
switch (a.cls) {
case float_class_snan:
s->float_exception_flags |= float_flag_invalid;
a = parts_silence_nan(a, s);
/* fall through */
case float_class_qnan:
if (s->default_nan_mode) {
return parts_default_nan(s);
}
break;
default:
g_assert_not_reached();
}
return a;
}
static FloatParts pick_nan(FloatParts a, FloatParts b, float_status *s)
{
if (is_snan(a.cls) || is_snan(b.cls)) {
s->float_exception_flags |= float_flag_invalid;
}
if (s->default_nan_mode) {
return parts_default_nan(s);
} else {
if (pickNaN(a.cls, b.cls,
a.frac > b.frac ||
(a.frac == b.frac && a.sign < b.sign))) {
a = b;
}
if (is_snan(a.cls)) {
return parts_silence_nan(a, s);
}
}
return a;
}
static FloatParts pick_nan_muladd(FloatParts a, FloatParts b, FloatParts c,
bool inf_zero, float_status *s)
{
int which;
if (is_snan(a.cls) || is_snan(b.cls) || is_snan(c.cls)) {
s->float_exception_flags |= float_flag_invalid;
}
which = pickNaNMulAdd(a.cls, b.cls, c.cls, inf_zero, s);
if (s->default_nan_mode) {
/* Note that this check is after pickNaNMulAdd so that function
* has an opportunity to set the Invalid flag.
*/
which = 3;
}
switch (which) {
case 0:
break;
case 1:
a = b;
break;
case 2:
a = c;
break;
case 3:
return parts_default_nan(s);
default:
g_assert_not_reached();
}
if (is_snan(a.cls)) {
return parts_silence_nan(a, s);
}
return a;
}
/*
* Returns the result of adding or subtracting the values of the
* floating-point values `a' and `b'. The operation is performed
* according to the IEC/IEEE Standard for Binary Floating-Point
* Arithmetic.
*/
static FloatParts addsub_floats(FloatParts a, FloatParts b, bool subtract,
float_status *s)
{
bool a_sign = a.sign;
bool b_sign = b.sign ^ subtract;
if (a_sign != b_sign) {
/* Subtraction */
if (a.cls == float_class_normal && b.cls == float_class_normal) {
if (a.exp > b.exp || (a.exp == b.exp && a.frac >= b.frac)) {
shift64RightJamming(b.frac, a.exp - b.exp, &b.frac);
a.frac = a.frac - b.frac;
} else {
shift64RightJamming(a.frac, b.exp - a.exp, &a.frac);
a.frac = b.frac - a.frac;
a.exp = b.exp;
a_sign ^= 1;
}
if (a.frac == 0) {
a.cls = float_class_zero;
a.sign = s->float_rounding_mode == float_round_down;
} else {
int shift = clz64(a.frac) - 1;
a.frac = a.frac << shift;
a.exp = a.exp - shift;
a.sign = a_sign;
}
return a;
}
if (is_nan(a.cls) || is_nan(b.cls)) {
return pick_nan(a, b, s);
}
if (a.cls == float_class_inf) {
if (b.cls == float_class_inf) {
float_raise(float_flag_invalid, s);
return parts_default_nan(s);
}
return a;
}
if (a.cls == float_class_zero && b.cls == float_class_zero) {
a.sign = s->float_rounding_mode == float_round_down;
return a;
}
if (a.cls == float_class_zero || b.cls == float_class_inf) {
b.sign = a_sign ^ 1;
return b;
}