xref: /openbmc/qemu/include/fpu/softfloat.h (revision 7acafcfa)
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 
18 /*
19 ===============================================================================
20 This C header file is part of the SoftFloat IEC/IEEE Floating-point
21 Arithmetic Package, Release 2a.
22 
23 Written by John R. Hauser.  This work was made possible in part by the
24 International Computer Science Institute, located at Suite 600, 1947 Center
25 Street, Berkeley, California 94704.  Funding was partially provided by the
26 National Science Foundation under grant MIP-9311980.  The original version
27 of this code was written as part of a project to build a fixed-point vector
28 processor in collaboration with the University of California at Berkeley,
29 overseen by Profs. Nelson Morgan and John Wawrzynek.  More information
30 is available through the Web page `http://HTTP.CS.Berkeley.EDU/~jhauser/
31 arithmetic/SoftFloat.html'.
32 
33 THIS SOFTWARE IS DISTRIBUTED AS IS, FOR FREE.  Although reasonable effort
34 has been made to avoid it, THIS SOFTWARE MAY CONTAIN FAULTS THAT WILL AT
35 TIMES RESULT IN INCORRECT BEHAVIOR.  USE OF THIS SOFTWARE IS RESTRICTED TO
36 PERSONS AND ORGANIZATIONS WHO CAN AND WILL TAKE FULL RESPONSIBILITY FOR ANY
37 AND ALL LOSSES, COSTS, OR OTHER PROBLEMS ARISING FROM ITS USE.
38 
39 Derivative works are acceptable, even for commercial purposes, so long as
40 (1) they include prominent notice that the work is derivative, and (2) they
41 include prominent notice akin to these four paragraphs for those parts of
42 this code that are retained.
43 
44 ===============================================================================
45 */
46 
47 /* BSD licensing:
48  * Copyright (c) 2006, Fabrice Bellard
49  * All rights reserved.
50  *
51  * Redistribution and use in source and binary forms, with or without
52  * modification, are permitted provided that the following conditions are met:
53  *
54  * 1. Redistributions of source code must retain the above copyright notice,
55  * this list of conditions and the following disclaimer.
56  *
57  * 2. Redistributions in binary form must reproduce the above copyright notice,
58  * this list of conditions and the following disclaimer in the documentation
59  * and/or other materials provided with the distribution.
60  *
61  * 3. Neither the name of the copyright holder nor the names of its contributors
62  * may be used to endorse or promote products derived from this software without
63  * specific prior written permission.
64  *
65  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
66  * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
67  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
68  * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE
69  * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
70  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
71  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
72  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
73  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
74  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
75  * THE POSSIBILITY OF SUCH DAMAGE.
76  */
77 
78 /* Portions of this work are licensed under the terms of the GNU GPL,
79  * version 2 or later. See the COPYING file in the top-level directory.
80  */
81 
82 #ifndef SOFTFLOAT_H
83 #define SOFTFLOAT_H
84 
85 /*----------------------------------------------------------------------------
86 | Software IEC/IEEE floating-point ordering relations
87 *----------------------------------------------------------------------------*/
88 
89 typedef enum {
90     float_relation_less      = -1,
91     float_relation_equal     =  0,
92     float_relation_greater   =  1,
93     float_relation_unordered =  2
94 } FloatRelation;
95 
96 #include "fpu/softfloat-types.h"
97 #include "fpu/softfloat-helpers.h"
98 
99 /*----------------------------------------------------------------------------
100 | Routine to raise any or all of the software IEC/IEEE floating-point
101 | exception flags.
102 *----------------------------------------------------------------------------*/
103 void float_raise(uint8_t flags, float_status *status);
104 
105 /*----------------------------------------------------------------------------
106 | If `a' is denormal and we are in flush-to-zero mode then set the
107 | input-denormal exception and return zero. Otherwise just return the value.
108 *----------------------------------------------------------------------------*/
109 float16 float16_squash_input_denormal(float16 a, float_status *status);
110 float32 float32_squash_input_denormal(float32 a, float_status *status);
111 float64 float64_squash_input_denormal(float64 a, float_status *status);
112 
113 /*----------------------------------------------------------------------------
114 | Options to indicate which negations to perform in float*_muladd()
115 | Using these differs from negating an input or output before calling
116 | the muladd function in that this means that a NaN doesn't have its
117 | sign bit inverted before it is propagated.
118 | We also support halving the result before rounding, as a special
119 | case to support the ARM fused-sqrt-step instruction FRSQRTS.
120 *----------------------------------------------------------------------------*/
121 enum {
122     float_muladd_negate_c = 1,
123     float_muladd_negate_product = 2,
124     float_muladd_negate_result = 4,
125     float_muladd_halve_result = 8,
126 };
127 
128 /*----------------------------------------------------------------------------
129 | Software IEC/IEEE integer-to-floating-point conversion routines.
130 *----------------------------------------------------------------------------*/
131 
132 float16 int16_to_float16_scalbn(int16_t a, int, float_status *status);
133 float16 int32_to_float16_scalbn(int32_t a, int, float_status *status);
134 float16 int64_to_float16_scalbn(int64_t a, int, float_status *status);
135 float16 uint16_to_float16_scalbn(uint16_t a, int, float_status *status);
136 float16 uint32_to_float16_scalbn(uint32_t a, int, float_status *status);
137 float16 uint64_to_float16_scalbn(uint64_t a, int, float_status *status);
138 
139 float16 int16_to_float16(int16_t a, float_status *status);
140 float16 int32_to_float16(int32_t a, float_status *status);
141 float16 int64_to_float16(int64_t a, float_status *status);
142 float16 uint16_to_float16(uint16_t a, float_status *status);
143 float16 uint32_to_float16(uint32_t a, float_status *status);
144 float16 uint64_to_float16(uint64_t a, float_status *status);
145 
146 float32 int16_to_float32_scalbn(int16_t, int, float_status *status);
147 float32 int32_to_float32_scalbn(int32_t, int, float_status *status);
148 float32 int64_to_float32_scalbn(int64_t, int, float_status *status);
149 float32 uint16_to_float32_scalbn(uint16_t, int, float_status *status);
150 float32 uint32_to_float32_scalbn(uint32_t, int, float_status *status);
151 float32 uint64_to_float32_scalbn(uint64_t, int, float_status *status);
152 
153 float32 int16_to_float32(int16_t, float_status *status);
154 float32 int32_to_float32(int32_t, float_status *status);
155 float32 int64_to_float32(int64_t, float_status *status);
156 float32 uint16_to_float32(uint16_t, float_status *status);
157 float32 uint32_to_float32(uint32_t, float_status *status);
158 float32 uint64_to_float32(uint64_t, float_status *status);
159 
160 float64 int16_to_float64_scalbn(int16_t, int, float_status *status);
161 float64 int32_to_float64_scalbn(int32_t, int, float_status *status);
162 float64 int64_to_float64_scalbn(int64_t, int, float_status *status);
163 float64 uint16_to_float64_scalbn(uint16_t, int, float_status *status);
164 float64 uint32_to_float64_scalbn(uint32_t, int, float_status *status);
165 float64 uint64_to_float64_scalbn(uint64_t, int, float_status *status);
166 
167 float64 int16_to_float64(int16_t, float_status *status);
168 float64 int32_to_float64(int32_t, float_status *status);
169 float64 int64_to_float64(int64_t, float_status *status);
170 float64 uint16_to_float64(uint16_t, float_status *status);
171 float64 uint32_to_float64(uint32_t, float_status *status);
172 float64 uint64_to_float64(uint64_t, float_status *status);
173 
174 floatx80 int32_to_floatx80(int32_t, float_status *status);
175 floatx80 int64_to_floatx80(int64_t, float_status *status);
176 
177 float128 int32_to_float128(int32_t, float_status *status);
178 float128 int64_to_float128(int64_t, float_status *status);
179 float128 uint64_to_float128(uint64_t, float_status *status);
180 
181 /*----------------------------------------------------------------------------
182 | Software half-precision conversion routines.
183 *----------------------------------------------------------------------------*/
184 
185 float16 float32_to_float16(float32, bool ieee, float_status *status);
186 float32 float16_to_float32(float16, bool ieee, float_status *status);
187 float16 float64_to_float16(float64 a, bool ieee, float_status *status);
188 float64 float16_to_float64(float16 a, bool ieee, float_status *status);
189 
190 int16_t float16_to_int16_scalbn(float16, FloatRoundMode, int, float_status *);
191 int32_t float16_to_int32_scalbn(float16, FloatRoundMode, int, float_status *);
192 int64_t float16_to_int64_scalbn(float16, FloatRoundMode, int, float_status *);
193 
194 int16_t float16_to_int16(float16, float_status *status);
195 int32_t float16_to_int32(float16, float_status *status);
196 int64_t float16_to_int64(float16, float_status *status);
197 
198 int16_t float16_to_int16_round_to_zero(float16, float_status *status);
199 int32_t float16_to_int32_round_to_zero(float16, float_status *status);
200 int64_t float16_to_int64_round_to_zero(float16, float_status *status);
201 
202 uint16_t float16_to_uint16_scalbn(float16 a, FloatRoundMode,
203                                   int, float_status *status);
204 uint32_t float16_to_uint32_scalbn(float16 a, FloatRoundMode,
205                                   int, float_status *status);
206 uint64_t float16_to_uint64_scalbn(float16 a, FloatRoundMode,
207                                   int, float_status *status);
208 
209 uint16_t float16_to_uint16(float16 a, float_status *status);
210 uint32_t float16_to_uint32(float16 a, float_status *status);
211 uint64_t float16_to_uint64(float16 a, float_status *status);
212 
213 uint16_t float16_to_uint16_round_to_zero(float16 a, float_status *status);
214 uint32_t float16_to_uint32_round_to_zero(float16 a, float_status *status);
215 uint64_t float16_to_uint64_round_to_zero(float16 a, float_status *status);
216 
217 /*----------------------------------------------------------------------------
218 | Software half-precision operations.
219 *----------------------------------------------------------------------------*/
220 
221 float16 float16_round_to_int(float16, float_status *status);
222 float16 float16_add(float16, float16, float_status *status);
223 float16 float16_sub(float16, float16, float_status *status);
224 float16 float16_mul(float16, float16, float_status *status);
225 float16 float16_muladd(float16, float16, float16, int, float_status *status);
226 float16 float16_div(float16, float16, float_status *status);
227 float16 float16_scalbn(float16, int, float_status *status);
228 float16 float16_min(float16, float16, float_status *status);
229 float16 float16_max(float16, float16, float_status *status);
230 float16 float16_minnum(float16, float16, float_status *status);
231 float16 float16_maxnum(float16, float16, float_status *status);
232 float16 float16_minnummag(float16, float16, float_status *status);
233 float16 float16_maxnummag(float16, float16, float_status *status);
234 float16 float16_sqrt(float16, float_status *status);
235 FloatRelation float16_compare(float16, float16, float_status *status);
236 FloatRelation float16_compare_quiet(float16, float16, float_status *status);
237 
238 bool float16_is_quiet_nan(float16, float_status *status);
239 bool float16_is_signaling_nan(float16, float_status *status);
240 float16 float16_silence_nan(float16, float_status *status);
241 
242 static inline bool float16_is_any_nan(float16 a)
243 {
244     return ((float16_val(a) & ~0x8000) > 0x7c00);
245 }
246 
247 static inline bool float16_is_neg(float16 a)
248 {
249     return float16_val(a) >> 15;
250 }
251 
252 static inline bool float16_is_infinity(float16 a)
253 {
254     return (float16_val(a) & 0x7fff) == 0x7c00;
255 }
256 
257 static inline bool float16_is_zero(float16 a)
258 {
259     return (float16_val(a) & 0x7fff) == 0;
260 }
261 
262 static inline bool float16_is_zero_or_denormal(float16 a)
263 {
264     return (float16_val(a) & 0x7c00) == 0;
265 }
266 
267 static inline float16 float16_abs(float16 a)
268 {
269     /* Note that abs does *not* handle NaN specially, nor does
270      * it flush denormal inputs to zero.
271      */
272     return make_float16(float16_val(a) & 0x7fff);
273 }
274 
275 static inline float16 float16_chs(float16 a)
276 {
277     /* Note that chs does *not* handle NaN specially, nor does
278      * it flush denormal inputs to zero.
279      */
280     return make_float16(float16_val(a) ^ 0x8000);
281 }
282 
283 static inline float16 float16_set_sign(float16 a, int sign)
284 {
285     return make_float16((float16_val(a) & 0x7fff) | (sign << 15));
286 }
287 
288 #define float16_zero make_float16(0)
289 #define float16_half make_float16(0x3800)
290 #define float16_one make_float16(0x3c00)
291 #define float16_one_point_five make_float16(0x3e00)
292 #define float16_two make_float16(0x4000)
293 #define float16_three make_float16(0x4200)
294 #define float16_infinity make_float16(0x7c00)
295 
296 /*----------------------------------------------------------------------------
297 | The pattern for a default generated half-precision NaN.
298 *----------------------------------------------------------------------------*/
299 float16 float16_default_nan(float_status *status);
300 
301 /*----------------------------------------------------------------------------
302 | Software IEC/IEEE single-precision conversion routines.
303 *----------------------------------------------------------------------------*/
304 
305 int16_t float32_to_int16_scalbn(float32, FloatRoundMode, int, float_status *);
306 int32_t float32_to_int32_scalbn(float32, FloatRoundMode, int, float_status *);
307 int64_t float32_to_int64_scalbn(float32, FloatRoundMode, int, float_status *);
308 
309 int16_t float32_to_int16(float32, float_status *status);
310 int32_t float32_to_int32(float32, float_status *status);
311 int64_t float32_to_int64(float32, float_status *status);
312 
313 int16_t float32_to_int16_round_to_zero(float32, float_status *status);
314 int32_t float32_to_int32_round_to_zero(float32, float_status *status);
315 int64_t float32_to_int64_round_to_zero(float32, float_status *status);
316 
317 uint16_t float32_to_uint16_scalbn(float32, FloatRoundMode, int, float_status *);
318 uint32_t float32_to_uint32_scalbn(float32, FloatRoundMode, int, float_status *);
319 uint64_t float32_to_uint64_scalbn(float32, FloatRoundMode, int, float_status *);
320 
321 uint16_t float32_to_uint16(float32, float_status *status);
322 uint32_t float32_to_uint32(float32, float_status *status);
323 uint64_t float32_to_uint64(float32, float_status *status);
324 
325 uint16_t float32_to_uint16_round_to_zero(float32, float_status *status);
326 uint32_t float32_to_uint32_round_to_zero(float32, float_status *status);
327 uint64_t float32_to_uint64_round_to_zero(float32, float_status *status);
328 
329 float64 float32_to_float64(float32, float_status *status);
330 floatx80 float32_to_floatx80(float32, float_status *status);
331 float128 float32_to_float128(float32, float_status *status);
332 
333 /*----------------------------------------------------------------------------
334 | Software IEC/IEEE single-precision operations.
335 *----------------------------------------------------------------------------*/
336 float32 float32_round_to_int(float32, float_status *status);
337 float32 float32_add(float32, float32, float_status *status);
338 float32 float32_sub(float32, float32, float_status *status);
339 float32 float32_mul(float32, float32, float_status *status);
340 float32 float32_div(float32, float32, float_status *status);
341 float32 float32_rem(float32, float32, float_status *status);
342 float32 float32_muladd(float32, float32, float32, int, float_status *status);
343 float32 float32_sqrt(float32, float_status *status);
344 float32 float32_exp2(float32, float_status *status);
345 float32 float32_log2(float32, float_status *status);
346 FloatRelation float32_compare(float32, float32, float_status *status);
347 FloatRelation float32_compare_quiet(float32, float32, float_status *status);
348 float32 float32_min(float32, float32, float_status *status);
349 float32 float32_max(float32, float32, float_status *status);
350 float32 float32_minnum(float32, float32, float_status *status);
351 float32 float32_maxnum(float32, float32, float_status *status);
352 float32 float32_minnummag(float32, float32, float_status *status);
353 float32 float32_maxnummag(float32, float32, float_status *status);
354 bool float32_is_quiet_nan(float32, float_status *status);
355 bool float32_is_signaling_nan(float32, float_status *status);
356 float32 float32_silence_nan(float32, float_status *status);
357 float32 float32_scalbn(float32, int, float_status *status);
358 
359 static inline float32 float32_abs(float32 a)
360 {
361     /* Note that abs does *not* handle NaN specially, nor does
362      * it flush denormal inputs to zero.
363      */
364     return make_float32(float32_val(a) & 0x7fffffff);
365 }
366 
367 static inline float32 float32_chs(float32 a)
368 {
369     /* Note that chs does *not* handle NaN specially, nor does
370      * it flush denormal inputs to zero.
371      */
372     return make_float32(float32_val(a) ^ 0x80000000);
373 }
374 
375 static inline bool float32_is_infinity(float32 a)
376 {
377     return (float32_val(a) & 0x7fffffff) == 0x7f800000;
378 }
379 
380 static inline bool float32_is_neg(float32 a)
381 {
382     return float32_val(a) >> 31;
383 }
384 
385 static inline bool float32_is_zero(float32 a)
386 {
387     return (float32_val(a) & 0x7fffffff) == 0;
388 }
389 
390 static inline bool float32_is_any_nan(float32 a)
391 {
392     return ((float32_val(a) & ~(1 << 31)) > 0x7f800000UL);
393 }
394 
395 static inline bool float32_is_zero_or_denormal(float32 a)
396 {
397     return (float32_val(a) & 0x7f800000) == 0;
398 }
399 
400 static inline bool float32_is_normal(float32 a)
401 {
402     return (((float32_val(a) >> 23) + 1) & 0xff) >= 2;
403 }
404 
405 static inline bool float32_is_denormal(float32 a)
406 {
407     return float32_is_zero_or_denormal(a) && !float32_is_zero(a);
408 }
409 
410 static inline bool float32_is_zero_or_normal(float32 a)
411 {
412     return float32_is_normal(a) || float32_is_zero(a);
413 }
414 
415 static inline float32 float32_set_sign(float32 a, int sign)
416 {
417     return make_float32((float32_val(a) & 0x7fffffff) | (sign << 31));
418 }
419 
420 static inline bool float32_eq(float32 a, float32 b, float_status *s)
421 {
422     return float32_compare(a, b, s) == float_relation_equal;
423 }
424 
425 static inline bool float32_le(float32 a, float32 b, float_status *s)
426 {
427     return float32_compare(a, b, s) <= float_relation_equal;
428 }
429 
430 static inline bool float32_lt(float32 a, float32 b, float_status *s)
431 {
432     return float32_compare(a, b, s) < float_relation_equal;
433 }
434 
435 static inline bool float32_unordered(float32 a, float32 b, float_status *s)
436 {
437     return float32_compare(a, b, s) == float_relation_unordered;
438 }
439 
440 static inline bool float32_eq_quiet(float32 a, float32 b, float_status *s)
441 {
442     return float32_compare_quiet(a, b, s) == float_relation_equal;
443 }
444 
445 static inline bool float32_le_quiet(float32 a, float32 b, float_status *s)
446 {
447     return float32_compare_quiet(a, b, s) <= float_relation_equal;
448 }
449 
450 static inline bool float32_lt_quiet(float32 a, float32 b, float_status *s)
451 {
452     return float32_compare_quiet(a, b, s) < float_relation_equal;
453 }
454 
455 static inline bool float32_unordered_quiet(float32 a, float32 b,
456                                            float_status *s)
457 {
458     return float32_compare_quiet(a, b, s) == float_relation_unordered;
459 }
460 
461 #define float32_zero make_float32(0)
462 #define float32_half make_float32(0x3f000000)
463 #define float32_one make_float32(0x3f800000)
464 #define float32_one_point_five make_float32(0x3fc00000)
465 #define float32_two make_float32(0x40000000)
466 #define float32_three make_float32(0x40400000)
467 #define float32_infinity make_float32(0x7f800000)
468 
469 /*----------------------------------------------------------------------------
470 | Packs the sign `zSign', exponent `zExp', and significand `zSig' into a
471 | single-precision floating-point value, returning the result.  After being
472 | shifted into the proper positions, the three fields are simply added
473 | together to form the result.  This means that any integer portion of `zSig'
474 | will be added into the exponent.  Since a properly normalized significand
475 | will have an integer portion equal to 1, the `zExp' input should be 1 less
476 | than the desired result exponent whenever `zSig' is a complete, normalized
477 | significand.
478 *----------------------------------------------------------------------------*/
479 
480 static inline float32 packFloat32(bool zSign, int zExp, uint32_t zSig)
481 {
482     return make_float32(
483           (((uint32_t)zSign) << 31) + (((uint32_t)zExp) << 23) + zSig);
484 }
485 
486 /*----------------------------------------------------------------------------
487 | The pattern for a default generated single-precision NaN.
488 *----------------------------------------------------------------------------*/
489 float32 float32_default_nan(float_status *status);
490 
491 /*----------------------------------------------------------------------------
492 | Software IEC/IEEE double-precision conversion routines.
493 *----------------------------------------------------------------------------*/
494 
495 int16_t float64_to_int16_scalbn(float64, FloatRoundMode, int, float_status *);
496 int32_t float64_to_int32_scalbn(float64, FloatRoundMode, int, float_status *);
497 int64_t float64_to_int64_scalbn(float64, FloatRoundMode, int, float_status *);
498 
499 int16_t float64_to_int16(float64, float_status *status);
500 int32_t float64_to_int32(float64, float_status *status);
501 int64_t float64_to_int64(float64, float_status *status);
502 
503 int16_t float64_to_int16_round_to_zero(float64, float_status *status);
504 int32_t float64_to_int32_round_to_zero(float64, float_status *status);
505 int64_t float64_to_int64_round_to_zero(float64, float_status *status);
506 
507 uint16_t float64_to_uint16_scalbn(float64, FloatRoundMode, int, float_status *);
508 uint32_t float64_to_uint32_scalbn(float64, FloatRoundMode, int, float_status *);
509 uint64_t float64_to_uint64_scalbn(float64, FloatRoundMode, int, float_status *);
510 
511 uint16_t float64_to_uint16(float64, float_status *status);
512 uint32_t float64_to_uint32(float64, float_status *status);
513 uint64_t float64_to_uint64(float64, float_status *status);
514 
515 uint16_t float64_to_uint16_round_to_zero(float64, float_status *status);
516 uint32_t float64_to_uint32_round_to_zero(float64, float_status *status);
517 uint64_t float64_to_uint64_round_to_zero(float64, float_status *status);
518 
519 float32 float64_to_float32(float64, float_status *status);
520 floatx80 float64_to_floatx80(float64, float_status *status);
521 float128 float64_to_float128(float64, float_status *status);
522 
523 /*----------------------------------------------------------------------------
524 | Software IEC/IEEE double-precision operations.
525 *----------------------------------------------------------------------------*/
526 float64 float64_round_to_int(float64, float_status *status);
527 float64 float64_add(float64, float64, float_status *status);
528 float64 float64_sub(float64, float64, float_status *status);
529 float64 float64_mul(float64, float64, float_status *status);
530 float64 float64_div(float64, float64, float_status *status);
531 float64 float64_rem(float64, float64, float_status *status);
532 float64 float64_muladd(float64, float64, float64, int, float_status *status);
533 float64 float64_sqrt(float64, float_status *status);
534 float64 float64_log2(float64, float_status *status);
535 FloatRelation float64_compare(float64, float64, float_status *status);
536 FloatRelation float64_compare_quiet(float64, float64, float_status *status);
537 float64 float64_min(float64, float64, float_status *status);
538 float64 float64_max(float64, float64, float_status *status);
539 float64 float64_minnum(float64, float64, float_status *status);
540 float64 float64_maxnum(float64, float64, float_status *status);
541 float64 float64_minnummag(float64, float64, float_status *status);
542 float64 float64_maxnummag(float64, float64, float_status *status);
543 bool float64_is_quiet_nan(float64 a, float_status *status);
544 bool float64_is_signaling_nan(float64, float_status *status);
545 float64 float64_silence_nan(float64, float_status *status);
546 float64 float64_scalbn(float64, int, float_status *status);
547 
548 static inline float64 float64_abs(float64 a)
549 {
550     /* Note that abs does *not* handle NaN specially, nor does
551      * it flush denormal inputs to zero.
552      */
553     return make_float64(float64_val(a) & 0x7fffffffffffffffLL);
554 }
555 
556 static inline float64 float64_chs(float64 a)
557 {
558     /* Note that chs does *not* handle NaN specially, nor does
559      * it flush denormal inputs to zero.
560      */
561     return make_float64(float64_val(a) ^ 0x8000000000000000LL);
562 }
563 
564 static inline bool float64_is_infinity(float64 a)
565 {
566     return (float64_val(a) & 0x7fffffffffffffffLL ) == 0x7ff0000000000000LL;
567 }
568 
569 static inline bool float64_is_neg(float64 a)
570 {
571     return float64_val(a) >> 63;
572 }
573 
574 static inline bool float64_is_zero(float64 a)
575 {
576     return (float64_val(a) & 0x7fffffffffffffffLL) == 0;
577 }
578 
579 static inline bool float64_is_any_nan(float64 a)
580 {
581     return ((float64_val(a) & ~(1ULL << 63)) > 0x7ff0000000000000ULL);
582 }
583 
584 static inline bool float64_is_zero_or_denormal(float64 a)
585 {
586     return (float64_val(a) & 0x7ff0000000000000LL) == 0;
587 }
588 
589 static inline bool float64_is_normal(float64 a)
590 {
591     return (((float64_val(a) >> 52) + 1) & 0x7ff) >= 2;
592 }
593 
594 static inline bool float64_is_denormal(float64 a)
595 {
596     return float64_is_zero_or_denormal(a) && !float64_is_zero(a);
597 }
598 
599 static inline bool float64_is_zero_or_normal(float64 a)
600 {
601     return float64_is_normal(a) || float64_is_zero(a);
602 }
603 
604 static inline float64 float64_set_sign(float64 a, int sign)
605 {
606     return make_float64((float64_val(a) & 0x7fffffffffffffffULL)
607                         | ((int64_t)sign << 63));
608 }
609 
610 static inline bool float64_eq(float64 a, float64 b, float_status *s)
611 {
612     return float64_compare(a, b, s) == float_relation_equal;
613 }
614 
615 static inline bool float64_le(float64 a, float64 b, float_status *s)
616 {
617     return float64_compare(a, b, s) <= float_relation_equal;
618 }
619 
620 static inline bool float64_lt(float64 a, float64 b, float_status *s)
621 {
622     return float64_compare(a, b, s) < float_relation_equal;
623 }
624 
625 static inline bool float64_unordered(float64 a, float64 b, float_status *s)
626 {
627     return float64_compare(a, b, s) == float_relation_unordered;
628 }
629 
630 static inline bool float64_eq_quiet(float64 a, float64 b, float_status *s)
631 {
632     return float64_compare_quiet(a, b, s) == float_relation_equal;
633 }
634 
635 static inline bool float64_le_quiet(float64 a, float64 b, float_status *s)
636 {
637     return float64_compare_quiet(a, b, s) <= float_relation_equal;
638 }
639 
640 static inline bool float64_lt_quiet(float64 a, float64 b, float_status *s)
641 {
642     return float64_compare_quiet(a, b, s) < float_relation_equal;
643 }
644 
645 static inline bool float64_unordered_quiet(float64 a, float64 b,
646                                            float_status *s)
647 {
648     return float64_compare_quiet(a, b, s) == float_relation_unordered;
649 }
650 
651 #define float64_zero make_float64(0)
652 #define float64_half make_float64(0x3fe0000000000000LL)
653 #define float64_one make_float64(0x3ff0000000000000LL)
654 #define float64_one_point_five make_float64(0x3FF8000000000000ULL)
655 #define float64_two make_float64(0x4000000000000000ULL)
656 #define float64_three make_float64(0x4008000000000000ULL)
657 #define float64_ln2 make_float64(0x3fe62e42fefa39efLL)
658 #define float64_infinity make_float64(0x7ff0000000000000LL)
659 
660 /*----------------------------------------------------------------------------
661 | The pattern for a default generated double-precision NaN.
662 *----------------------------------------------------------------------------*/
663 float64 float64_default_nan(float_status *status);
664 
665 /*----------------------------------------------------------------------------
666 | Software IEC/IEEE extended double-precision conversion routines.
667 *----------------------------------------------------------------------------*/
668 int32_t floatx80_to_int32(floatx80, float_status *status);
669 int32_t floatx80_to_int32_round_to_zero(floatx80, float_status *status);
670 int64_t floatx80_to_int64(floatx80, float_status *status);
671 int64_t floatx80_to_int64_round_to_zero(floatx80, float_status *status);
672 float32 floatx80_to_float32(floatx80, float_status *status);
673 float64 floatx80_to_float64(floatx80, float_status *status);
674 float128 floatx80_to_float128(floatx80, float_status *status);
675 
676 /*----------------------------------------------------------------------------
677 | The pattern for an extended double-precision inf.
678 *----------------------------------------------------------------------------*/
679 extern const floatx80 floatx80_infinity;
680 
681 /*----------------------------------------------------------------------------
682 | Software IEC/IEEE extended double-precision operations.
683 *----------------------------------------------------------------------------*/
684 floatx80 floatx80_round(floatx80 a, float_status *status);
685 floatx80 floatx80_round_to_int(floatx80, float_status *status);
686 floatx80 floatx80_add(floatx80, floatx80, float_status *status);
687 floatx80 floatx80_sub(floatx80, floatx80, float_status *status);
688 floatx80 floatx80_mul(floatx80, floatx80, float_status *status);
689 floatx80 floatx80_div(floatx80, floatx80, float_status *status);
690 floatx80 floatx80_modrem(floatx80, floatx80, bool, uint64_t *,
691                          float_status *status);
692 floatx80 floatx80_mod(floatx80, floatx80, float_status *status);
693 floatx80 floatx80_rem(floatx80, floatx80, float_status *status);
694 floatx80 floatx80_sqrt(floatx80, float_status *status);
695 FloatRelation floatx80_compare(floatx80, floatx80, float_status *status);
696 FloatRelation floatx80_compare_quiet(floatx80, floatx80, float_status *status);
697 int floatx80_is_quiet_nan(floatx80, float_status *status);
698 int floatx80_is_signaling_nan(floatx80, float_status *status);
699 floatx80 floatx80_silence_nan(floatx80, float_status *status);
700 floatx80 floatx80_scalbn(floatx80, int, float_status *status);
701 
702 static inline floatx80 floatx80_abs(floatx80 a)
703 {
704     a.high &= 0x7fff;
705     return a;
706 }
707 
708 static inline floatx80 floatx80_chs(floatx80 a)
709 {
710     a.high ^= 0x8000;
711     return a;
712 }
713 
714 static inline bool floatx80_is_infinity(floatx80 a)
715 {
716 #if defined(TARGET_M68K)
717     return (a.high & 0x7fff) == floatx80_infinity.high && !(a.low << 1);
718 #else
719     return (a.high & 0x7fff) == floatx80_infinity.high &&
720                        a.low == floatx80_infinity.low;
721 #endif
722 }
723 
724 static inline bool floatx80_is_neg(floatx80 a)
725 {
726     return a.high >> 15;
727 }
728 
729 static inline bool floatx80_is_zero(floatx80 a)
730 {
731     return (a.high & 0x7fff) == 0 && a.low == 0;
732 }
733 
734 static inline bool floatx80_is_zero_or_denormal(floatx80 a)
735 {
736     return (a.high & 0x7fff) == 0;
737 }
738 
739 static inline bool floatx80_is_any_nan(floatx80 a)
740 {
741     return ((a.high & 0x7fff) == 0x7fff) && (a.low<<1);
742 }
743 
744 static inline bool floatx80_eq(floatx80 a, floatx80 b, float_status *s)
745 {
746     return floatx80_compare(a, b, s) == float_relation_equal;
747 }
748 
749 static inline bool floatx80_le(floatx80 a, floatx80 b, float_status *s)
750 {
751     return floatx80_compare(a, b, s) <= float_relation_equal;
752 }
753 
754 static inline bool floatx80_lt(floatx80 a, floatx80 b, float_status *s)
755 {
756     return floatx80_compare(a, b, s) < float_relation_equal;
757 }
758 
759 static inline bool floatx80_unordered(floatx80 a, floatx80 b, float_status *s)
760 {
761     return floatx80_compare(a, b, s) == float_relation_unordered;
762 }
763 
764 static inline bool floatx80_eq_quiet(floatx80 a, floatx80 b, float_status *s)
765 {
766     return floatx80_compare_quiet(a, b, s) == float_relation_equal;
767 }
768 
769 static inline bool floatx80_le_quiet(floatx80 a, floatx80 b, float_status *s)
770 {
771     return floatx80_compare_quiet(a, b, s) <= float_relation_equal;
772 }
773 
774 static inline bool floatx80_lt_quiet(floatx80 a, floatx80 b, float_status *s)
775 {
776     return floatx80_compare_quiet(a, b, s) < float_relation_equal;
777 }
778 
779 static inline bool floatx80_unordered_quiet(floatx80 a, floatx80 b,
780                                            float_status *s)
781 {
782     return floatx80_compare_quiet(a, b, s) == float_relation_unordered;
783 }
784 
785 /*----------------------------------------------------------------------------
786 | Return whether the given value is an invalid floatx80 encoding.
787 | Invalid floatx80 encodings arise when the integer bit is not set, but
788 | the exponent is not zero. The only times the integer bit is permitted to
789 | be zero is in subnormal numbers and the value zero.
790 | This includes what the Intel software developer's manual calls pseudo-NaNs,
791 | pseudo-infinities and un-normal numbers. It does not include
792 | pseudo-denormals, which must still be correctly handled as inputs even
793 | if they are never generated as outputs.
794 *----------------------------------------------------------------------------*/
795 static inline bool floatx80_invalid_encoding(floatx80 a)
796 {
797 #if defined(TARGET_M68K)
798     /*-------------------------------------------------------------------------
799     | With m68k, the explicit integer bit can be zero in the case of:
800     | - zeros                (exp == 0, mantissa == 0)
801     | - denormalized numbers (exp == 0, mantissa != 0)
802     | - unnormalized numbers (exp != 0, exp < 0x7FFF)
803     | - infinities           (exp == 0x7FFF, mantissa == 0)
804     | - not-a-numbers        (exp == 0x7FFF, mantissa != 0)
805     |
806     | For infinities and NaNs, the explicit integer bit can be either one or
807     | zero.
808     |
809     | The IEEE 754 standard does not define a zero integer bit. Such a number
810     | is an unnormalized number. Hardware does not directly support
811     | denormalized and unnormalized numbers, but implicitly supports them by
812     | trapping them as unimplemented data types, allowing efficient conversion
813     | in software.
814     |
815     | See "M68000 FAMILY PROGRAMMER’S REFERENCE MANUAL",
816     |     "1.6 FLOATING-POINT DATA TYPES"
817     *------------------------------------------------------------------------*/
818     return false;
819 #else
820     return (a.low & (1ULL << 63)) == 0 && (a.high & 0x7FFF) != 0;
821 #endif
822 }
823 
824 #define floatx80_zero make_floatx80(0x0000, 0x0000000000000000LL)
825 #define floatx80_one make_floatx80(0x3fff, 0x8000000000000000LL)
826 #define floatx80_ln2 make_floatx80(0x3ffe, 0xb17217f7d1cf79acLL)
827 #define floatx80_pi make_floatx80(0x4000, 0xc90fdaa22168c235LL)
828 #define floatx80_half make_floatx80(0x3ffe, 0x8000000000000000LL)
829 
830 /*----------------------------------------------------------------------------
831 | Returns the fraction bits of the extended double-precision floating-point
832 | value `a'.
833 *----------------------------------------------------------------------------*/
834 
835 static inline uint64_t extractFloatx80Frac(floatx80 a)
836 {
837     return a.low;
838 }
839 
840 /*----------------------------------------------------------------------------
841 | Returns the exponent bits of the extended double-precision floating-point
842 | value `a'.
843 *----------------------------------------------------------------------------*/
844 
845 static inline int32_t extractFloatx80Exp(floatx80 a)
846 {
847     return a.high & 0x7FFF;
848 }
849 
850 /*----------------------------------------------------------------------------
851 | Returns the sign bit of the extended double-precision floating-point value
852 | `a'.
853 *----------------------------------------------------------------------------*/
854 
855 static inline bool extractFloatx80Sign(floatx80 a)
856 {
857     return a.high >> 15;
858 }
859 
860 /*----------------------------------------------------------------------------
861 | Packs the sign `zSign', exponent `zExp', and significand `zSig' into an
862 | extended double-precision floating-point value, returning the result.
863 *----------------------------------------------------------------------------*/
864 
865 static inline floatx80 packFloatx80(bool zSign, int32_t zExp, uint64_t zSig)
866 {
867     floatx80 z;
868 
869     z.low = zSig;
870     z.high = (((uint16_t)zSign) << 15) + zExp;
871     return z;
872 }
873 
874 /*----------------------------------------------------------------------------
875 | Normalizes the subnormal extended double-precision floating-point value
876 | represented by the denormalized significand `aSig'.  The normalized exponent
877 | and significand are stored at the locations pointed to by `zExpPtr' and
878 | `zSigPtr', respectively.
879 *----------------------------------------------------------------------------*/
880 
881 void normalizeFloatx80Subnormal(uint64_t aSig, int32_t *zExpPtr,
882                                 uint64_t *zSigPtr);
883 
884 /*----------------------------------------------------------------------------
885 | Takes two extended double-precision floating-point values `a' and `b', one
886 | of which is a NaN, and returns the appropriate NaN result.  If either `a' or
887 | `b' is a signaling NaN, the invalid exception is raised.
888 *----------------------------------------------------------------------------*/
889 
890 floatx80 propagateFloatx80NaN(floatx80 a, floatx80 b, float_status *status);
891 
892 /*----------------------------------------------------------------------------
893 | Takes an abstract floating-point value having sign `zSign', exponent `zExp',
894 | and extended significand formed by the concatenation of `zSig0' and `zSig1',
895 | and returns the proper extended double-precision floating-point value
896 | corresponding to the abstract input.  Ordinarily, the abstract value is
897 | rounded and packed into the extended double-precision format, with the
898 | inexact exception raised if the abstract input cannot be represented
899 | exactly.  However, if the abstract value is too large, the overflow and
900 | inexact exceptions are raised and an infinity or maximal finite value is
901 | returned.  If the abstract value is too small, the input value is rounded to
902 | a subnormal number, and the underflow and inexact exceptions are raised if
903 | the abstract input cannot be represented exactly as a subnormal extended
904 | double-precision floating-point number.
905 |     If `roundingPrecision' is 32 or 64, the result is rounded to the same
906 | number of bits as single or double precision, respectively.  Otherwise, the
907 | result is rounded to the full precision of the extended double-precision
908 | format.
909 |     The input significand must be normalized or smaller.  If the input
910 | significand is not normalized, `zExp' must be 0; in that case, the result
911 | returned is a subnormal number, and it must not require rounding.  The
912 | handling of underflow and overflow follows the IEC/IEEE Standard for Binary
913 | Floating-Point Arithmetic.
914 *----------------------------------------------------------------------------*/
915 
916 floatx80 roundAndPackFloatx80(int8_t roundingPrecision, bool zSign,
917                               int32_t zExp, uint64_t zSig0, uint64_t zSig1,
918                               float_status *status);
919 
920 /*----------------------------------------------------------------------------
921 | Takes an abstract floating-point value having sign `zSign', exponent
922 | `zExp', and significand formed by the concatenation of `zSig0' and `zSig1',
923 | and returns the proper extended double-precision floating-point value
924 | corresponding to the abstract input.  This routine is just like
925 | `roundAndPackFloatx80' except that the input significand does not have to be
926 | normalized.
927 *----------------------------------------------------------------------------*/
928 
929 floatx80 normalizeRoundAndPackFloatx80(int8_t roundingPrecision,
930                                        bool zSign, int32_t zExp,
931                                        uint64_t zSig0, uint64_t zSig1,
932                                        float_status *status);
933 
934 /*----------------------------------------------------------------------------
935 | The pattern for a default generated extended double-precision NaN.
936 *----------------------------------------------------------------------------*/
937 floatx80 floatx80_default_nan(float_status *status);
938 
939 /*----------------------------------------------------------------------------
940 | Software IEC/IEEE quadruple-precision conversion routines.
941 *----------------------------------------------------------------------------*/
942 int32_t float128_to_int32(float128, float_status *status);
943 int32_t float128_to_int32_round_to_zero(float128, float_status *status);
944 int64_t float128_to_int64(float128, float_status *status);
945 int64_t float128_to_int64_round_to_zero(float128, float_status *status);
946 uint64_t float128_to_uint64(float128, float_status *status);
947 uint64_t float128_to_uint64_round_to_zero(float128, float_status *status);
948 uint32_t float128_to_uint32(float128, float_status *status);
949 uint32_t float128_to_uint32_round_to_zero(float128, float_status *status);
950 float32 float128_to_float32(float128, float_status *status);
951 float64 float128_to_float64(float128, float_status *status);
952 floatx80 float128_to_floatx80(float128, float_status *status);
953 
954 /*----------------------------------------------------------------------------
955 | Software IEC/IEEE quadruple-precision operations.
956 *----------------------------------------------------------------------------*/
957 float128 float128_round_to_int(float128, float_status *status);
958 float128 float128_add(float128, float128, float_status *status);
959 float128 float128_sub(float128, float128, float_status *status);
960 float128 float128_mul(float128, float128, float_status *status);
961 float128 float128_div(float128, float128, float_status *status);
962 float128 float128_rem(float128, float128, float_status *status);
963 float128 float128_sqrt(float128, float_status *status);
964 FloatRelation float128_compare(float128, float128, float_status *status);
965 FloatRelation float128_compare_quiet(float128, float128, float_status *status);
966 bool float128_is_quiet_nan(float128, float_status *status);
967 bool float128_is_signaling_nan(float128, float_status *status);
968 float128 float128_silence_nan(float128, float_status *status);
969 float128 float128_scalbn(float128, int, float_status *status);
970 
971 static inline float128 float128_abs(float128 a)
972 {
973     a.high &= 0x7fffffffffffffffLL;
974     return a;
975 }
976 
977 static inline float128 float128_chs(float128 a)
978 {
979     a.high ^= 0x8000000000000000LL;
980     return a;
981 }
982 
983 static inline bool float128_is_infinity(float128 a)
984 {
985     return (a.high & 0x7fffffffffffffffLL) == 0x7fff000000000000LL && a.low == 0;
986 }
987 
988 static inline bool float128_is_neg(float128 a)
989 {
990     return a.high >> 63;
991 }
992 
993 static inline bool float128_is_zero(float128 a)
994 {
995     return (a.high & 0x7fffffffffffffffLL) == 0 && a.low == 0;
996 }
997 
998 static inline bool float128_is_zero_or_denormal(float128 a)
999 {
1000     return (a.high & 0x7fff000000000000LL) == 0;
1001 }
1002 
1003 static inline bool float128_is_normal(float128 a)
1004 {
1005     return (((a.high >> 48) + 1) & 0x7fff) >= 2;
1006 }
1007 
1008 static inline bool float128_is_denormal(float128 a)
1009 {
1010     return float128_is_zero_or_denormal(a) && !float128_is_zero(a);
1011 }
1012 
1013 static inline bool float128_is_any_nan(float128 a)
1014 {
1015     return ((a.high >> 48) & 0x7fff) == 0x7fff &&
1016         ((a.low != 0) || ((a.high & 0xffffffffffffLL) != 0));
1017 }
1018 
1019 static inline bool float128_eq(float128 a, float128 b, float_status *s)
1020 {
1021     return float128_compare(a, b, s) == float_relation_equal;
1022 }
1023 
1024 static inline bool float128_le(float128 a, float128 b, float_status *s)
1025 {
1026     return float128_compare(a, b, s) <= float_relation_equal;
1027 }
1028 
1029 static inline bool float128_lt(float128 a, float128 b, float_status *s)
1030 {
1031     return float128_compare(a, b, s) < float_relation_equal;
1032 }
1033 
1034 static inline bool float128_unordered(float128 a, float128 b, float_status *s)
1035 {
1036     return float128_compare(a, b, s) == float_relation_unordered;
1037 }
1038 
1039 static inline bool float128_eq_quiet(float128 a, float128 b, float_status *s)
1040 {
1041     return float128_compare_quiet(a, b, s) == float_relation_equal;
1042 }
1043 
1044 static inline bool float128_le_quiet(float128 a, float128 b, float_status *s)
1045 {
1046     return float128_compare_quiet(a, b, s) <= float_relation_equal;
1047 }
1048 
1049 static inline bool float128_lt_quiet(float128 a, float128 b, float_status *s)
1050 {
1051     return float128_compare_quiet(a, b, s) < float_relation_equal;
1052 }
1053 
1054 static inline bool float128_unordered_quiet(float128 a, float128 b,
1055                                            float_status *s)
1056 {
1057     return float128_compare_quiet(a, b, s) == float_relation_unordered;
1058 }
1059 
1060 #define float128_zero make_float128(0, 0)
1061 
1062 /*----------------------------------------------------------------------------
1063 | The pattern for a default generated quadruple-precision NaN.
1064 *----------------------------------------------------------------------------*/
1065 float128 float128_default_nan(float_status *status);
1066 
1067 #endif /* SOFTFLOAT_H */
1068