xref: /openbmc/qemu/fpu/softfloat-parts.c.inc (revision 6eb169b8)
1/*
2 * QEMU float support
3 *
4 * The code in this source file is derived from release 2a of the SoftFloat
5 * IEC/IEEE Floating-point Arithmetic Package. Those parts of the code (and
6 * some later contributions) are provided under that license, as detailed below.
7 * It has subsequently been modified by contributors to the QEMU Project,
8 * so some portions are provided under:
9 *  the SoftFloat-2a license
10 *  the BSD license
11 *  GPL-v2-or-later
12 *
13 * Any future contributions to this file after December 1st 2014 will be
14 * taken to be licensed under the Softfloat-2a license unless specifically
15 * indicated otherwise.
16 */
17
18static void partsN(return_nan)(FloatPartsN *a, float_status *s)
19{
20    switch (a->cls) {
21    case float_class_snan:
22        float_raise(float_flag_invalid, s);
23        if (s->default_nan_mode) {
24            parts_default_nan(a, s);
25        } else {
26            parts_silence_nan(a, s);
27        }
28        break;
29    case float_class_qnan:
30        if (s->default_nan_mode) {
31            parts_default_nan(a, s);
32        }
33        break;
34    default:
35        g_assert_not_reached();
36    }
37}
38
39static FloatPartsN *partsN(pick_nan)(FloatPartsN *a, FloatPartsN *b,
40                                     float_status *s)
41{
42    if (is_snan(a->cls) || is_snan(b->cls)) {
43        float_raise(float_flag_invalid, s);
44    }
45
46    if (s->default_nan_mode) {
47        parts_default_nan(a, s);
48    } else {
49        int cmp = frac_cmp(a, b);
50        if (cmp == 0) {
51            cmp = a->sign < b->sign;
52        }
53
54        if (pickNaN(a->cls, b->cls, cmp > 0, s)) {
55            a = b;
56        }
57        if (is_snan(a->cls)) {
58            parts_silence_nan(a, s);
59        }
60    }
61    return a;
62}
63
64static FloatPartsN *partsN(pick_nan_muladd)(FloatPartsN *a, FloatPartsN *b,
65                                            FloatPartsN *c, float_status *s,
66                                            int ab_mask, int abc_mask)
67{
68    int which;
69
70    if (unlikely(abc_mask & float_cmask_snan)) {
71        float_raise(float_flag_invalid, s);
72    }
73
74    which = pickNaNMulAdd(a->cls, b->cls, c->cls,
75                          ab_mask == float_cmask_infzero, s);
76
77    if (s->default_nan_mode || which == 3) {
78        /*
79         * Note that this check is after pickNaNMulAdd so that function
80         * has an opportunity to set the Invalid flag for infzero.
81         */
82        parts_default_nan(a, s);
83        return a;
84    }
85
86    switch (which) {
87    case 0:
88        break;
89    case 1:
90        a = b;
91        break;
92    case 2:
93        a = c;
94        break;
95    default:
96        g_assert_not_reached();
97    }
98    if (is_snan(a->cls)) {
99        parts_silence_nan(a, s);
100    }
101    return a;
102}
103
104/*
105 * Canonicalize the FloatParts structure.  Determine the class,
106 * unbias the exponent, and normalize the fraction.
107 */
108static void partsN(canonicalize)(FloatPartsN *p, float_status *status,
109                                 const FloatFmt *fmt)
110{
111    if (unlikely(p->exp == 0)) {
112        if (likely(frac_eqz(p))) {
113            p->cls = float_class_zero;
114        } else if (status->flush_inputs_to_zero) {
115            float_raise(float_flag_input_denormal, status);
116            p->cls = float_class_zero;
117            frac_clear(p);
118        } else {
119            int shift = frac_normalize(p);
120            p->cls = float_class_normal;
121            p->exp = fmt->frac_shift - fmt->exp_bias - shift + 1;
122        }
123    } else if (likely(p->exp < fmt->exp_max) || fmt->arm_althp) {
124        p->cls = float_class_normal;
125        p->exp -= fmt->exp_bias;
126        frac_shl(p, fmt->frac_shift);
127        p->frac_hi |= DECOMPOSED_IMPLICIT_BIT;
128    } else if (likely(frac_eqz(p))) {
129        p->cls = float_class_inf;
130    } else {
131        frac_shl(p, fmt->frac_shift);
132        p->cls = (parts_is_snan_frac(p->frac_hi, status)
133                  ? float_class_snan : float_class_qnan);
134    }
135}
136
137/*
138 * Round and uncanonicalize a floating-point number by parts. There
139 * are FRAC_SHIFT bits that may require rounding at the bottom of the
140 * fraction; these bits will be removed. The exponent will be biased
141 * by EXP_BIAS and must be bounded by [EXP_MAX-1, 0].
142 */
143static void partsN(uncanon)(FloatPartsN *p, float_status *s,
144                            const FloatFmt *fmt)
145{
146    const int exp_max = fmt->exp_max;
147    const int frac_shift = fmt->frac_shift;
148    const uint64_t frac_lsb = fmt->frac_lsb;
149    const uint64_t frac_lsbm1 = fmt->frac_lsbm1;
150    const uint64_t round_mask = fmt->round_mask;
151    const uint64_t roundeven_mask = fmt->roundeven_mask;
152    uint64_t inc;
153    bool overflow_norm;
154    int exp, flags = 0;
155
156    if (unlikely(p->cls != float_class_normal)) {
157        switch (p->cls) {
158        case float_class_zero:
159            p->exp = 0;
160            frac_clear(p);
161            return;
162        case float_class_inf:
163            g_assert(!fmt->arm_althp);
164            p->exp = fmt->exp_max;
165            frac_clear(p);
166            return;
167        case float_class_qnan:
168        case float_class_snan:
169            g_assert(!fmt->arm_althp);
170            p->exp = fmt->exp_max;
171            frac_shr(p, fmt->frac_shift);
172            return;
173        default:
174            break;
175        }
176        g_assert_not_reached();
177    }
178
179    overflow_norm = false;
180    switch (s->float_rounding_mode) {
181    case float_round_nearest_even:
182        inc = ((p->frac_lo & roundeven_mask) != frac_lsbm1 ? frac_lsbm1 : 0);
183        break;
184    case float_round_ties_away:
185        inc = frac_lsbm1;
186        break;
187    case float_round_to_zero:
188        overflow_norm = true;
189        inc = 0;
190        break;
191    case float_round_up:
192        inc = p->sign ? 0 : round_mask;
193        overflow_norm = p->sign;
194        break;
195    case float_round_down:
196        inc = p->sign ? round_mask : 0;
197        overflow_norm = !p->sign;
198        break;
199    case float_round_to_odd:
200        overflow_norm = true;
201        /* fall through */
202    case float_round_to_odd_inf:
203        inc = p->frac_lo & frac_lsb ? 0 : round_mask;
204        break;
205    default:
206        g_assert_not_reached();
207    }
208
209    exp = p->exp + fmt->exp_bias;
210    if (likely(exp > 0)) {
211        if (p->frac_lo & round_mask) {
212            flags |= float_flag_inexact;
213            if (frac_addi(p, p, inc)) {
214                frac_shr(p, 1);
215                p->frac_hi |= DECOMPOSED_IMPLICIT_BIT;
216                exp++;
217            }
218        }
219        frac_shr(p, frac_shift);
220
221        if (fmt->arm_althp) {
222            /* ARM Alt HP eschews Inf and NaN for a wider exponent.  */
223            if (unlikely(exp > exp_max)) {
224                /* Overflow.  Return the maximum normal.  */
225                flags = float_flag_invalid;
226                exp = exp_max;
227                frac_allones(p);
228            }
229        } else if (unlikely(exp >= exp_max)) {
230            flags |= float_flag_overflow | float_flag_inexact;
231            if (overflow_norm) {
232                exp = exp_max - 1;
233                frac_allones(p);
234            } else {
235                p->cls = float_class_inf;
236                exp = exp_max;
237                frac_clear(p);
238            }
239        }
240    } else if (s->flush_to_zero) {
241        flags |= float_flag_output_denormal;
242        p->cls = float_class_zero;
243        exp = 0;
244        frac_clear(p);
245    } else {
246        bool is_tiny = s->tininess_before_rounding || exp < 0;
247
248        if (!is_tiny) {
249            FloatPartsN discard;
250            is_tiny = !frac_addi(&discard, p, inc);
251        }
252
253        frac_shrjam(p, 1 - exp);
254
255        if (p->frac_lo & round_mask) {
256            /* Need to recompute round-to-even/round-to-odd. */
257            switch (s->float_rounding_mode) {
258            case float_round_nearest_even:
259                inc = ((p->frac_lo & roundeven_mask) != frac_lsbm1
260                       ? frac_lsbm1 : 0);
261                break;
262            case float_round_to_odd:
263            case float_round_to_odd_inf:
264                inc = p->frac_lo & frac_lsb ? 0 : round_mask;
265                break;
266            default:
267                break;
268            }
269            flags |= float_flag_inexact;
270            frac_addi(p, p, inc);
271        }
272
273        exp = (p->frac_hi & DECOMPOSED_IMPLICIT_BIT) != 0;
274        frac_shr(p, frac_shift);
275
276        if (is_tiny && (flags & float_flag_inexact)) {
277            flags |= float_flag_underflow;
278        }
279        if (exp == 0 && frac_eqz(p)) {
280            p->cls = float_class_zero;
281        }
282    }
283    p->exp = exp;
284    float_raise(flags, s);
285}
286
287/*
288 * Returns the result of adding or subtracting the values of the
289 * floating-point values `a' and `b'. The operation is performed
290 * according to the IEC/IEEE Standard for Binary Floating-Point
291 * Arithmetic.
292 */
293static FloatPartsN *partsN(addsub)(FloatPartsN *a, FloatPartsN *b,
294                                   float_status *s, bool subtract)
295{
296    bool b_sign = b->sign ^ subtract;
297    int ab_mask = float_cmask(a->cls) | float_cmask(b->cls);
298
299    if (a->sign != b_sign) {
300        /* Subtraction */
301        if (likely(ab_mask == float_cmask_normal)) {
302            if (parts_sub_normal(a, b)) {
303                return a;
304            }
305            /* Subtract was exact, fall through to set sign. */
306            ab_mask = float_cmask_zero;
307        }
308
309        if (ab_mask == float_cmask_zero) {
310            a->sign = s->float_rounding_mode == float_round_down;
311            return a;
312        }
313
314        if (unlikely(ab_mask & float_cmask_anynan)) {
315            goto p_nan;
316        }
317
318        if (ab_mask & float_cmask_inf) {
319            if (a->cls != float_class_inf) {
320                /* N - Inf */
321                goto return_b;
322            }
323            if (b->cls != float_class_inf) {
324                /* Inf - N */
325                return a;
326            }
327            /* Inf - Inf */
328            float_raise(float_flag_invalid, s);
329            parts_default_nan(a, s);
330            return a;
331        }
332    } else {
333        /* Addition */
334        if (likely(ab_mask == float_cmask_normal)) {
335            parts_add_normal(a, b);
336            return a;
337        }
338
339        if (ab_mask == float_cmask_zero) {
340            return a;
341        }
342
343        if (unlikely(ab_mask & float_cmask_anynan)) {
344            goto p_nan;
345        }
346
347        if (ab_mask & float_cmask_inf) {
348            a->cls = float_class_inf;
349            return a;
350        }
351    }
352
353    if (b->cls == float_class_zero) {
354        g_assert(a->cls == float_class_normal);
355        return a;
356    }
357
358    g_assert(a->cls == float_class_zero);
359    g_assert(b->cls == float_class_normal);
360 return_b:
361    b->sign = b_sign;
362    return b;
363
364 p_nan:
365    return parts_pick_nan(a, b, s);
366}
367
368/*
369 * Returns the result of multiplying the floating-point values `a' and
370 * `b'. The operation is performed according to the IEC/IEEE Standard
371 * for Binary Floating-Point Arithmetic.
372 */
373static FloatPartsN *partsN(mul)(FloatPartsN *a, FloatPartsN *b,
374                                float_status *s)
375{
376    int ab_mask = float_cmask(a->cls) | float_cmask(b->cls);
377    bool sign = a->sign ^ b->sign;
378
379    if (likely(ab_mask == float_cmask_normal)) {
380        FloatPartsW tmp;
381
382        frac_mulw(&tmp, a, b);
383        frac_truncjam(a, &tmp);
384
385        a->exp += b->exp + 1;
386        if (!(a->frac_hi & DECOMPOSED_IMPLICIT_BIT)) {
387            frac_add(a, a, a);
388            a->exp -= 1;
389        }
390
391        a->sign = sign;
392        return a;
393    }
394
395    /* Inf * Zero == NaN */
396    if (unlikely(ab_mask == float_cmask_infzero)) {
397        float_raise(float_flag_invalid, s);
398        parts_default_nan(a, s);
399        return a;
400    }
401
402    if (unlikely(ab_mask & float_cmask_anynan)) {
403        return parts_pick_nan(a, b, s);
404    }
405
406    /* Multiply by 0 or Inf */
407    if (ab_mask & float_cmask_inf) {
408        a->cls = float_class_inf;
409        a->sign = sign;
410        return a;
411    }
412
413    g_assert(ab_mask & float_cmask_zero);
414    a->cls = float_class_zero;
415    a->sign = sign;
416    return a;
417}
418
419/*
420 * Returns the result of multiplying the floating-point values `a' and
421 * `b' then adding 'c', with no intermediate rounding step after the
422 * multiplication. The operation is performed according to the
423 * IEC/IEEE Standard for Binary Floating-Point Arithmetic 754-2008.
424 * The flags argument allows the caller to select negation of the
425 * addend, the intermediate product, or the final result. (The
426 * difference between this and having the caller do a separate
427 * negation is that negating externally will flip the sign bit on NaNs.)
428 *
429 * Requires A and C extracted into a double-sized structure to provide the
430 * extra space for the widening multiply.
431 */
432static FloatPartsN *partsN(muladd)(FloatPartsN *a, FloatPartsN *b,
433                                   FloatPartsN *c, int flags, float_status *s)
434{
435    int ab_mask, abc_mask;
436    FloatPartsW p_widen, c_widen;
437
438    ab_mask = float_cmask(a->cls) | float_cmask(b->cls);
439    abc_mask = float_cmask(c->cls) | ab_mask;
440
441    /*
442     * It is implementation-defined whether the cases of (0,inf,qnan)
443     * and (inf,0,qnan) raise InvalidOperation or not (and what QNaN
444     * they return if they do), so we have to hand this information
445     * off to the target-specific pick-a-NaN routine.
446     */
447    if (unlikely(abc_mask & float_cmask_anynan)) {
448        return parts_pick_nan_muladd(a, b, c, s, ab_mask, abc_mask);
449    }
450
451    if (flags & float_muladd_negate_c) {
452        c->sign ^= 1;
453    }
454
455    /* Compute the sign of the product into A. */
456    a->sign ^= b->sign;
457    if (flags & float_muladd_negate_product) {
458        a->sign ^= 1;
459    }
460
461    if (unlikely(ab_mask != float_cmask_normal)) {
462        if (unlikely(ab_mask == float_cmask_infzero)) {
463            goto d_nan;
464        }
465
466        if (ab_mask & float_cmask_inf) {
467            if (c->cls == float_class_inf && a->sign != c->sign) {
468                goto d_nan;
469            }
470            goto return_inf;
471        }
472
473        g_assert(ab_mask & float_cmask_zero);
474        if (c->cls == float_class_normal) {
475            *a = *c;
476            goto return_normal;
477        }
478        if (c->cls == float_class_zero) {
479            if (a->sign != c->sign) {
480                goto return_sub_zero;
481            }
482            goto return_zero;
483        }
484        g_assert(c->cls == float_class_inf);
485    }
486
487    if (unlikely(c->cls == float_class_inf)) {
488        a->sign = c->sign;
489        goto return_inf;
490    }
491
492    /* Perform the multiplication step. */
493    p_widen.sign = a->sign;
494    p_widen.exp = a->exp + b->exp + 1;
495    frac_mulw(&p_widen, a, b);
496    if (!(p_widen.frac_hi & DECOMPOSED_IMPLICIT_BIT)) {
497        frac_add(&p_widen, &p_widen, &p_widen);
498        p_widen.exp -= 1;
499    }
500
501    /* Perform the addition step. */
502    if (c->cls != float_class_zero) {
503        /* Zero-extend C to less significant bits. */
504        frac_widen(&c_widen, c);
505        c_widen.exp = c->exp;
506
507        if (a->sign == c->sign) {
508            parts_add_normal(&p_widen, &c_widen);
509        } else if (!parts_sub_normal(&p_widen, &c_widen)) {
510            goto return_sub_zero;
511        }
512    }
513
514    /* Narrow with sticky bit, for proper rounding later. */
515    frac_truncjam(a, &p_widen);
516    a->sign = p_widen.sign;
517    a->exp = p_widen.exp;
518
519 return_normal:
520    if (flags & float_muladd_halve_result) {
521        a->exp -= 1;
522    }
523 finish_sign:
524    if (flags & float_muladd_negate_result) {
525        a->sign ^= 1;
526    }
527    return a;
528
529 return_sub_zero:
530    a->sign = s->float_rounding_mode == float_round_down;
531 return_zero:
532    a->cls = float_class_zero;
533    goto finish_sign;
534
535 return_inf:
536    a->cls = float_class_inf;
537    goto finish_sign;
538
539 d_nan:
540    float_raise(float_flag_invalid, s);
541    parts_default_nan(a, s);
542    return a;
543}
544
545/*
546 * Returns the result of dividing the floating-point value `a' by the
547 * corresponding value `b'. The operation is performed according to
548 * the IEC/IEEE Standard for Binary Floating-Point Arithmetic.
549 */
550static FloatPartsN *partsN(div)(FloatPartsN *a, FloatPartsN *b,
551                                float_status *s)
552{
553    int ab_mask = float_cmask(a->cls) | float_cmask(b->cls);
554    bool sign = a->sign ^ b->sign;
555
556    if (likely(ab_mask == float_cmask_normal)) {
557        a->sign = sign;
558        a->exp -= b->exp + frac_div(a, b);
559        return a;
560    }
561
562    /* 0/0 or Inf/Inf => NaN */
563    if (unlikely(ab_mask == float_cmask_zero) ||
564        unlikely(ab_mask == float_cmask_inf)) {
565        float_raise(float_flag_invalid, s);
566        parts_default_nan(a, s);
567        return a;
568    }
569
570    /* All the NaN cases */
571    if (unlikely(ab_mask & float_cmask_anynan)) {
572        return parts_pick_nan(a, b, s);
573    }
574
575    a->sign = sign;
576
577    /* Inf / X */
578    if (a->cls == float_class_inf) {
579        return a;
580    }
581
582    /* 0 / X */
583    if (a->cls == float_class_zero) {
584        return a;
585    }
586
587    /* X / Inf */
588    if (b->cls == float_class_inf) {
589        a->cls = float_class_zero;
590        return a;
591    }
592
593    /* X / 0 => Inf */
594    g_assert(b->cls == float_class_zero);
595    float_raise(float_flag_divbyzero, s);
596    a->cls = float_class_inf;
597    return a;
598}
599
600/*
601 * Rounds the floating-point value `a' to an integer, and returns the
602 * result as a floating-point value. The operation is performed
603 * according to the IEC/IEEE Standard for Binary Floating-Point
604 * Arithmetic.
605 *
606 * parts_round_to_int_normal is an internal helper function for
607 * normal numbers only, returning true for inexact but not directly
608 * raising float_flag_inexact.
609 */
610static bool partsN(round_to_int_normal)(FloatPartsN *a, FloatRoundMode rmode,
611                                        int scale, int frac_size)
612{
613    uint64_t frac_lsb, frac_lsbm1, rnd_even_mask, rnd_mask, inc;
614    int shift_adj;
615
616    scale = MIN(MAX(scale, -0x10000), 0x10000);
617    a->exp += scale;
618
619    if (a->exp < 0) {
620        bool one;
621
622        /* All fractional */
623        switch (rmode) {
624        case float_round_nearest_even:
625            one = false;
626            if (a->exp == -1) {
627                FloatPartsN tmp;
628                /* Shift left one, discarding DECOMPOSED_IMPLICIT_BIT */
629                frac_add(&tmp, a, a);
630                /* Anything remaining means frac > 0.5. */
631                one = !frac_eqz(&tmp);
632            }
633            break;
634        case float_round_ties_away:
635            one = a->exp == -1;
636            break;
637        case float_round_to_zero:
638            one = false;
639            break;
640        case float_round_up:
641            one = !a->sign;
642            break;
643        case float_round_down:
644            one = a->sign;
645            break;
646        case float_round_to_odd:
647            one = true;
648            break;
649        default:
650            g_assert_not_reached();
651        }
652
653        frac_clear(a);
654        a->exp = 0;
655        if (one) {
656            a->frac_hi = DECOMPOSED_IMPLICIT_BIT;
657        } else {
658            a->cls = float_class_zero;
659        }
660        return true;
661    }
662
663    if (a->exp >= frac_size) {
664        /* All integral */
665        return false;
666    }
667
668    if (N > 64 && a->exp < N - 64) {
669        /*
670         * Rounding is not in the low word -- shift lsb to bit 2,
671         * which leaves room for sticky and rounding bit.
672         */
673        shift_adj = (N - 1) - (a->exp + 2);
674        frac_shrjam(a, shift_adj);
675        frac_lsb = 1 << 2;
676    } else {
677        shift_adj = 0;
678        frac_lsb = DECOMPOSED_IMPLICIT_BIT >> (a->exp & 63);
679    }
680
681    frac_lsbm1 = frac_lsb >> 1;
682    rnd_mask = frac_lsb - 1;
683    rnd_even_mask = rnd_mask | frac_lsb;
684
685    if (!(a->frac_lo & rnd_mask)) {
686        /* Fractional bits already clear, undo the shift above. */
687        frac_shl(a, shift_adj);
688        return false;
689    }
690
691    switch (rmode) {
692    case float_round_nearest_even:
693        inc = ((a->frac_lo & rnd_even_mask) != frac_lsbm1 ? frac_lsbm1 : 0);
694        break;
695    case float_round_ties_away:
696        inc = frac_lsbm1;
697        break;
698    case float_round_to_zero:
699        inc = 0;
700        break;
701    case float_round_up:
702        inc = a->sign ? 0 : rnd_mask;
703        break;
704    case float_round_down:
705        inc = a->sign ? rnd_mask : 0;
706        break;
707    case float_round_to_odd:
708        inc = a->frac_lo & frac_lsb ? 0 : rnd_mask;
709        break;
710    default:
711        g_assert_not_reached();
712    }
713
714    if (shift_adj == 0) {
715        if (frac_addi(a, a, inc)) {
716            frac_shr(a, 1);
717            a->frac_hi |= DECOMPOSED_IMPLICIT_BIT;
718            a->exp++;
719        }
720        a->frac_lo &= ~rnd_mask;
721    } else {
722        frac_addi(a, a, inc);
723        a->frac_lo &= ~rnd_mask;
724        /* Be careful shifting back, not to overflow */
725        frac_shl(a, shift_adj - 1);
726        if (a->frac_hi & DECOMPOSED_IMPLICIT_BIT) {
727            a->exp++;
728        } else {
729            frac_add(a, a, a);
730        }
731    }
732    return true;
733}
734
735static void partsN(round_to_int)(FloatPartsN *a, FloatRoundMode rmode,
736                                 int scale, float_status *s,
737                                 const FloatFmt *fmt)
738{
739    switch (a->cls) {
740    case float_class_qnan:
741    case float_class_snan:
742        parts_return_nan(a, s);
743        break;
744    case float_class_zero:
745    case float_class_inf:
746        break;
747    case float_class_normal:
748        if (parts_round_to_int_normal(a, rmode, scale, fmt->frac_size)) {
749            float_raise(float_flag_inexact, s);
750        }
751        break;
752    default:
753        g_assert_not_reached();
754    }
755}
756
757/*
758 * Returns the result of converting the floating-point value `a' to
759 * the two's complement integer format. The conversion is performed
760 * according to the IEC/IEEE Standard for Binary Floating-Point
761 * Arithmetic---which means in particular that the conversion is
762 * rounded according to the current rounding mode. If `a' is a NaN,
763 * the largest positive integer is returned. Otherwise, if the
764 * conversion overflows, the largest integer with the same sign as `a'
765 * is returned.
766 */
767static int64_t partsN(float_to_sint)(FloatPartsN *p, FloatRoundMode rmode,
768                                     int scale, int64_t min, int64_t max,
769                                     float_status *s)
770{
771    int flags = 0;
772    uint64_t r;
773
774    switch (p->cls) {
775    case float_class_snan:
776    case float_class_qnan:
777        flags = float_flag_invalid;
778        r = max;
779        break;
780
781    case float_class_inf:
782        flags = float_flag_invalid;
783        r = p->sign ? min : max;
784        break;
785
786    case float_class_zero:
787        return 0;
788
789    case float_class_normal:
790        /* TODO: N - 2 is frac_size for rounding; could use input fmt. */
791        if (parts_round_to_int_normal(p, rmode, scale, N - 2)) {
792            flags = float_flag_inexact;
793        }
794
795        if (p->exp <= DECOMPOSED_BINARY_POINT) {
796            r = p->frac_hi >> (DECOMPOSED_BINARY_POINT - p->exp);
797        } else {
798            r = UINT64_MAX;
799        }
800        if (p->sign) {
801            if (r <= -(uint64_t)min) {
802                r = -r;
803            } else {
804                flags = float_flag_invalid;
805                r = min;
806            }
807        } else if (r > max) {
808            flags = float_flag_invalid;
809            r = max;
810        }
811        break;
812
813    default:
814        g_assert_not_reached();
815    }
816
817    float_raise(flags, s);
818    return r;
819}
820
821/*
822 *  Returns the result of converting the floating-point value `a' to
823 *  the unsigned integer format. The conversion is performed according
824 *  to the IEC/IEEE Standard for Binary Floating-Point
825 *  Arithmetic---which means in particular that the conversion is
826 *  rounded according to the current rounding mode. If `a' is a NaN,
827 *  the largest unsigned integer is returned. Otherwise, if the
828 *  conversion overflows, the largest unsigned integer is returned. If
829 *  the 'a' is negative, the result is rounded and zero is returned;
830 *  values that do not round to zero will raise the inexact exception
831 *  flag.
832 */
833static uint64_t partsN(float_to_uint)(FloatPartsN *p, FloatRoundMode rmode,
834                                      int scale, uint64_t max, float_status *s)
835{
836    int flags = 0;
837    uint64_t r;
838
839    switch (p->cls) {
840    case float_class_snan:
841    case float_class_qnan:
842        flags = float_flag_invalid;
843        r = max;
844        break;
845
846    case float_class_inf:
847        flags = float_flag_invalid;
848        r = p->sign ? 0 : max;
849        break;
850
851    case float_class_zero:
852        return 0;
853
854    case float_class_normal:
855        /* TODO: N - 2 is frac_size for rounding; could use input fmt. */
856        if (parts_round_to_int_normal(p, rmode, scale, N - 2)) {
857            flags = float_flag_inexact;
858            if (p->cls == float_class_zero) {
859                r = 0;
860                break;
861            }
862        }
863
864        if (p->sign) {
865            flags = float_flag_invalid;
866            r = 0;
867        } else if (p->exp > DECOMPOSED_BINARY_POINT) {
868            flags = float_flag_invalid;
869            r = max;
870        } else {
871            r = p->frac_hi >> (DECOMPOSED_BINARY_POINT - p->exp);
872            if (r > max) {
873                flags = float_flag_invalid;
874                r = max;
875            }
876        }
877        break;
878
879    default:
880        g_assert_not_reached();
881    }
882
883    float_raise(flags, s);
884    return r;
885}
886
887/*
888 * Integer to float conversions
889 *
890 * Returns the result of converting the two's complement integer `a'
891 * to the floating-point format. The conversion is performed according
892 * to the IEC/IEEE Standard for Binary Floating-Point Arithmetic.
893 */
894static void partsN(sint_to_float)(FloatPartsN *p, int64_t a,
895                                  int scale, float_status *s)
896{
897    uint64_t f = a;
898    int shift;
899
900    memset(p, 0, sizeof(*p));
901
902    if (a == 0) {
903        p->cls = float_class_zero;
904        return;
905    }
906
907    p->cls = float_class_normal;
908    if (a < 0) {
909        f = -f;
910        p->sign = true;
911    }
912    shift = clz64(f);
913    scale = MIN(MAX(scale, -0x10000), 0x10000);
914
915    p->exp = DECOMPOSED_BINARY_POINT - shift + scale;
916    p->frac_hi = f << shift;
917}
918
919/*
920 * Unsigned Integer to float conversions
921 *
922 * Returns the result of converting the unsigned integer `a' to the
923 * floating-point format. The conversion is performed according to the
924 * IEC/IEEE Standard for Binary Floating-Point Arithmetic.
925 */
926static void partsN(uint_to_float)(FloatPartsN *p, uint64_t a,
927                                  int scale, float_status *status)
928{
929    memset(p, 0, sizeof(*p));
930
931    if (a == 0) {
932        p->cls = float_class_zero;
933    } else {
934        int shift = clz64(a);
935        scale = MIN(MAX(scale, -0x10000), 0x10000);
936        p->cls = float_class_normal;
937        p->exp = DECOMPOSED_BINARY_POINT - shift + scale;
938        p->frac_hi = a << shift;
939    }
940}
941
942/*
943 * Float min/max.
944 */
945static FloatPartsN *partsN(minmax)(FloatPartsN *a, FloatPartsN *b,
946                                   float_status *s, int flags)
947{
948    int ab_mask = float_cmask(a->cls) | float_cmask(b->cls);
949    int a_exp, b_exp, cmp;
950
951    if (unlikely(ab_mask & float_cmask_anynan)) {
952        /*
953         * For minnum/maxnum, if one operand is a QNaN, and the other
954         * operand is numerical, then return numerical argument.
955         */
956        if ((flags & minmax_isnum)
957            && !(ab_mask & float_cmask_snan)
958            && (ab_mask & ~float_cmask_qnan)) {
959            return is_nan(a->cls) ? b : a;
960        }
961        return parts_pick_nan(a, b, s);
962    }
963
964    a_exp = a->exp;
965    b_exp = b->exp;
966
967    if (unlikely(ab_mask != float_cmask_normal)) {
968        switch (a->cls) {
969        case float_class_normal:
970            break;
971        case float_class_inf:
972            a_exp = INT16_MAX;
973            break;
974        case float_class_zero:
975            a_exp = INT16_MIN;
976            break;
977        default:
978            g_assert_not_reached();
979            break;
980        }
981        switch (b->cls) {
982        case float_class_normal:
983            break;
984        case float_class_inf:
985            b_exp = INT16_MAX;
986            break;
987        case float_class_zero:
988            b_exp = INT16_MIN;
989            break;
990        default:
991            g_assert_not_reached();
992            break;
993        }
994    }
995
996    /* Compare magnitudes. */
997    cmp = a_exp - b_exp;
998    if (cmp == 0) {
999        cmp = frac_cmp(a, b);
1000    }
1001
1002    /*
1003     * Take the sign into account.
1004     * For ismag, only do this if the magnitudes are equal.
1005     */
1006    if (!(flags & minmax_ismag) || cmp == 0) {
1007        if (a->sign != b->sign) {
1008            /* For differing signs, the negative operand is less. */
1009            cmp = a->sign ? -1 : 1;
1010        } else if (a->sign) {
1011            /* For two negative operands, invert the magnitude comparison. */
1012            cmp = -cmp;
1013        }
1014    }
1015
1016    if (flags & minmax_ismin) {
1017        cmp = -cmp;
1018    }
1019    return cmp < 0 ? b : a;
1020}
1021
1022/*
1023 * Floating point compare
1024 */
1025static FloatRelation partsN(compare)(FloatPartsN *a, FloatPartsN *b,
1026                                     float_status *s, bool is_quiet)
1027{
1028    int ab_mask = float_cmask(a->cls) | float_cmask(b->cls);
1029    int cmp;
1030
1031    if (likely(ab_mask == float_cmask_normal)) {
1032        if (a->sign != b->sign) {
1033            goto a_sign;
1034        }
1035        if (a->exp != b->exp) {
1036            cmp = a->exp < b->exp ? -1 : 1;
1037        } else {
1038            cmp = frac_cmp(a, b);
1039        }
1040        if (a->sign) {
1041            cmp = -cmp;
1042        }
1043        return cmp;
1044    }
1045
1046    if (unlikely(ab_mask & float_cmask_anynan)) {
1047        if (!is_quiet || (ab_mask & float_cmask_snan)) {
1048            float_raise(float_flag_invalid, s);
1049        }
1050        return float_relation_unordered;
1051    }
1052
1053    if (ab_mask & float_cmask_zero) {
1054        if (ab_mask == float_cmask_zero) {
1055            return float_relation_equal;
1056        } else if (a->cls == float_class_zero) {
1057            goto b_sign;
1058        } else {
1059            goto a_sign;
1060        }
1061    }
1062
1063    if (ab_mask == float_cmask_inf) {
1064        if (a->sign == b->sign) {
1065            return float_relation_equal;
1066        }
1067    } else if (b->cls == float_class_inf) {
1068        goto b_sign;
1069    } else {
1070        g_assert(a->cls == float_class_inf);
1071    }
1072
1073 a_sign:
1074    return a->sign ? float_relation_less : float_relation_greater;
1075 b_sign:
1076    return b->sign ? float_relation_greater : float_relation_less;
1077}
1078