xref: /openbmc/qemu/include/fpu/softfloat.h (revision c46184a9)
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 static inline void float_raise(uint8_t flags, float_status *status)
104 {
105     status->float_exception_flags |= flags;
106 }
107 
108 /*----------------------------------------------------------------------------
109 | If `a' is denormal and we are in flush-to-zero mode then set the
110 | input-denormal exception and return zero. Otherwise just return the value.
111 *----------------------------------------------------------------------------*/
112 float16 float16_squash_input_denormal(float16 a, float_status *status);
113 float32 float32_squash_input_denormal(float32 a, float_status *status);
114 float64 float64_squash_input_denormal(float64 a, float_status *status);
115 bfloat16 bfloat16_squash_input_denormal(bfloat16 a, float_status *status);
116 
117 /*----------------------------------------------------------------------------
118 | Options to indicate which negations to perform in float*_muladd()
119 | Using these differs from negating an input or output before calling
120 | the muladd function in that this means that a NaN doesn't have its
121 | sign bit inverted before it is propagated.
122 | We also support halving the result before rounding, as a special
123 | case to support the ARM fused-sqrt-step instruction FRSQRTS.
124 *----------------------------------------------------------------------------*/
125 enum {
126     float_muladd_negate_c = 1,
127     float_muladd_negate_product = 2,
128     float_muladd_negate_result = 4,
129     float_muladd_halve_result = 8,
130 };
131 
132 /*----------------------------------------------------------------------------
133 | Software IEC/IEEE integer-to-floating-point conversion routines.
134 *----------------------------------------------------------------------------*/
135 
136 float16 int16_to_float16_scalbn(int16_t a, int, float_status *status);
137 float16 int32_to_float16_scalbn(int32_t a, int, float_status *status);
138 float16 int64_to_float16_scalbn(int64_t a, int, float_status *status);
139 float16 uint16_to_float16_scalbn(uint16_t a, int, float_status *status);
140 float16 uint32_to_float16_scalbn(uint32_t a, int, float_status *status);
141 float16 uint64_to_float16_scalbn(uint64_t a, int, float_status *status);
142 
143 float16 int8_to_float16(int8_t a, float_status *status);
144 float16 int16_to_float16(int16_t a, float_status *status);
145 float16 int32_to_float16(int32_t a, float_status *status);
146 float16 int64_to_float16(int64_t a, float_status *status);
147 float16 uint8_to_float16(uint8_t a, float_status *status);
148 float16 uint16_to_float16(uint16_t a, float_status *status);
149 float16 uint32_to_float16(uint32_t a, float_status *status);
150 float16 uint64_to_float16(uint64_t a, float_status *status);
151 
152 float32 int16_to_float32_scalbn(int16_t, int, float_status *status);
153 float32 int32_to_float32_scalbn(int32_t, int, float_status *status);
154 float32 int64_to_float32_scalbn(int64_t, int, float_status *status);
155 float32 uint16_to_float32_scalbn(uint16_t, int, float_status *status);
156 float32 uint32_to_float32_scalbn(uint32_t, int, float_status *status);
157 float32 uint64_to_float32_scalbn(uint64_t, int, float_status *status);
158 
159 float32 int16_to_float32(int16_t, float_status *status);
160 float32 int32_to_float32(int32_t, float_status *status);
161 float32 int64_to_float32(int64_t, float_status *status);
162 float32 uint16_to_float32(uint16_t, float_status *status);
163 float32 uint32_to_float32(uint32_t, float_status *status);
164 float32 uint64_to_float32(uint64_t, float_status *status);
165 
166 float64 int16_to_float64_scalbn(int16_t, int, float_status *status);
167 float64 int32_to_float64_scalbn(int32_t, int, float_status *status);
168 float64 int64_to_float64_scalbn(int64_t, int, float_status *status);
169 float64 uint16_to_float64_scalbn(uint16_t, int, float_status *status);
170 float64 uint32_to_float64_scalbn(uint32_t, int, float_status *status);
171 float64 uint64_to_float64_scalbn(uint64_t, int, float_status *status);
172 
173 float64 int16_to_float64(int16_t, float_status *status);
174 float64 int32_to_float64(int32_t, float_status *status);
175 float64 int64_to_float64(int64_t, float_status *status);
176 float64 uint16_to_float64(uint16_t, float_status *status);
177 float64 uint32_to_float64(uint32_t, float_status *status);
178 float64 uint64_to_float64(uint64_t, float_status *status);
179 
180 floatx80 int32_to_floatx80(int32_t, float_status *status);
181 floatx80 int64_to_floatx80(int64_t, float_status *status);
182 
183 float128 int32_to_float128(int32_t, float_status *status);
184 float128 int64_to_float128(int64_t, float_status *status);
185 float128 uint64_to_float128(uint64_t, float_status *status);
186 
187 /*----------------------------------------------------------------------------
188 | Software half-precision conversion routines.
189 *----------------------------------------------------------------------------*/
190 
191 float16 float32_to_float16(float32, bool ieee, float_status *status);
192 float32 float16_to_float32(float16, bool ieee, float_status *status);
193 float16 float64_to_float16(float64 a, bool ieee, float_status *status);
194 float64 float16_to_float64(float16 a, bool ieee, float_status *status);
195 
196 int8_t  float16_to_int8_scalbn(float16, FloatRoundMode, int,
197                                float_status *status);
198 int16_t float16_to_int16_scalbn(float16, FloatRoundMode, int, float_status *);
199 int32_t float16_to_int32_scalbn(float16, FloatRoundMode, int, float_status *);
200 int64_t float16_to_int64_scalbn(float16, FloatRoundMode, int, float_status *);
201 
202 int8_t  float16_to_int8(float16, float_status *status);
203 int16_t float16_to_int16(float16, float_status *status);
204 int32_t float16_to_int32(float16, float_status *status);
205 int64_t float16_to_int64(float16, float_status *status);
206 
207 int16_t float16_to_int16_round_to_zero(float16, float_status *status);
208 int32_t float16_to_int32_round_to_zero(float16, float_status *status);
209 int64_t float16_to_int64_round_to_zero(float16, float_status *status);
210 
211 uint8_t float16_to_uint8_scalbn(float16 a, FloatRoundMode,
212                                 int, float_status *status);
213 uint16_t float16_to_uint16_scalbn(float16 a, FloatRoundMode,
214                                   int, float_status *status);
215 uint32_t float16_to_uint32_scalbn(float16 a, FloatRoundMode,
216                                   int, float_status *status);
217 uint64_t float16_to_uint64_scalbn(float16 a, FloatRoundMode,
218                                   int, float_status *status);
219 
220 uint8_t  float16_to_uint8(float16 a, float_status *status);
221 uint16_t float16_to_uint16(float16 a, float_status *status);
222 uint32_t float16_to_uint32(float16 a, float_status *status);
223 uint64_t float16_to_uint64(float16 a, float_status *status);
224 
225 uint16_t float16_to_uint16_round_to_zero(float16 a, float_status *status);
226 uint32_t float16_to_uint32_round_to_zero(float16 a, float_status *status);
227 uint64_t float16_to_uint64_round_to_zero(float16 a, float_status *status);
228 
229 /*----------------------------------------------------------------------------
230 | Software half-precision operations.
231 *----------------------------------------------------------------------------*/
232 
233 float16 float16_round_to_int(float16, float_status *status);
234 float16 float16_add(float16, float16, float_status *status);
235 float16 float16_sub(float16, float16, float_status *status);
236 float16 float16_mul(float16, float16, float_status *status);
237 float16 float16_muladd(float16, float16, float16, int, float_status *status);
238 float16 float16_div(float16, float16, float_status *status);
239 float16 float16_scalbn(float16, int, float_status *status);
240 float16 float16_min(float16, float16, float_status *status);
241 float16 float16_max(float16, float16, float_status *status);
242 float16 float16_minnum(float16, float16, float_status *status);
243 float16 float16_maxnum(float16, float16, float_status *status);
244 float16 float16_minnummag(float16, float16, float_status *status);
245 float16 float16_maxnummag(float16, float16, float_status *status);
246 float16 float16_sqrt(float16, float_status *status);
247 FloatRelation float16_compare(float16, float16, float_status *status);
248 FloatRelation float16_compare_quiet(float16, float16, float_status *status);
249 
250 bool float16_is_quiet_nan(float16, float_status *status);
251 bool float16_is_signaling_nan(float16, float_status *status);
252 float16 float16_silence_nan(float16, float_status *status);
253 
254 static inline bool float16_is_any_nan(float16 a)
255 {
256     return ((float16_val(a) & ~0x8000) > 0x7c00);
257 }
258 
259 static inline bool float16_is_neg(float16 a)
260 {
261     return float16_val(a) >> 15;
262 }
263 
264 static inline bool float16_is_infinity(float16 a)
265 {
266     return (float16_val(a) & 0x7fff) == 0x7c00;
267 }
268 
269 static inline bool float16_is_zero(float16 a)
270 {
271     return (float16_val(a) & 0x7fff) == 0;
272 }
273 
274 static inline bool float16_is_zero_or_denormal(float16 a)
275 {
276     return (float16_val(a) & 0x7c00) == 0;
277 }
278 
279 static inline bool float16_is_normal(float16 a)
280 {
281     return (((float16_val(a) >> 10) + 1) & 0x1f) >= 2;
282 }
283 
284 static inline float16 float16_abs(float16 a)
285 {
286     /* Note that abs does *not* handle NaN specially, nor does
287      * it flush denormal inputs to zero.
288      */
289     return make_float16(float16_val(a) & 0x7fff);
290 }
291 
292 static inline float16 float16_chs(float16 a)
293 {
294     /* Note that chs does *not* handle NaN specially, nor does
295      * it flush denormal inputs to zero.
296      */
297     return make_float16(float16_val(a) ^ 0x8000);
298 }
299 
300 static inline float16 float16_set_sign(float16 a, int sign)
301 {
302     return make_float16((float16_val(a) & 0x7fff) | (sign << 15));
303 }
304 
305 static inline bool float16_eq(float16 a, float16 b, float_status *s)
306 {
307     return float16_compare(a, b, s) == float_relation_equal;
308 }
309 
310 static inline bool float16_le(float16 a, float16 b, float_status *s)
311 {
312     return float16_compare(a, b, s) <= float_relation_equal;
313 }
314 
315 static inline bool float16_lt(float16 a, float16 b, float_status *s)
316 {
317     return float16_compare(a, b, s) < float_relation_equal;
318 }
319 
320 static inline bool float16_unordered(float16 a, float16 b, float_status *s)
321 {
322     return float16_compare(a, b, s) == float_relation_unordered;
323 }
324 
325 static inline bool float16_eq_quiet(float16 a, float16 b, float_status *s)
326 {
327     return float16_compare_quiet(a, b, s) == float_relation_equal;
328 }
329 
330 static inline bool float16_le_quiet(float16 a, float16 b, float_status *s)
331 {
332     return float16_compare_quiet(a, b, s) <= float_relation_equal;
333 }
334 
335 static inline bool float16_lt_quiet(float16 a, float16 b, float_status *s)
336 {
337     return float16_compare_quiet(a, b, s) < float_relation_equal;
338 }
339 
340 static inline bool float16_unordered_quiet(float16 a, float16 b,
341                                            float_status *s)
342 {
343     return float16_compare_quiet(a, b, s) == float_relation_unordered;
344 }
345 
346 #define float16_zero make_float16(0)
347 #define float16_half make_float16(0x3800)
348 #define float16_one make_float16(0x3c00)
349 #define float16_one_point_five make_float16(0x3e00)
350 #define float16_two make_float16(0x4000)
351 #define float16_three make_float16(0x4200)
352 #define float16_infinity make_float16(0x7c00)
353 
354 /*----------------------------------------------------------------------------
355 | Software bfloat16 conversion routines.
356 *----------------------------------------------------------------------------*/
357 
358 bfloat16 bfloat16_round_to_int(bfloat16, float_status *status);
359 bfloat16 float32_to_bfloat16(float32, float_status *status);
360 float32 bfloat16_to_float32(bfloat16, float_status *status);
361 bfloat16 float64_to_bfloat16(float64 a, float_status *status);
362 float64 bfloat16_to_float64(bfloat16 a, float_status *status);
363 
364 int16_t bfloat16_to_int16_scalbn(bfloat16, FloatRoundMode,
365                                  int, float_status *status);
366 int32_t bfloat16_to_int32_scalbn(bfloat16, FloatRoundMode,
367                                  int, float_status *status);
368 int64_t bfloat16_to_int64_scalbn(bfloat16, FloatRoundMode,
369                                  int, float_status *status);
370 
371 int16_t bfloat16_to_int16(bfloat16, float_status *status);
372 int32_t bfloat16_to_int32(bfloat16, float_status *status);
373 int64_t bfloat16_to_int64(bfloat16, float_status *status);
374 
375 int16_t bfloat16_to_int16_round_to_zero(bfloat16, float_status *status);
376 int32_t bfloat16_to_int32_round_to_zero(bfloat16, float_status *status);
377 int64_t bfloat16_to_int64_round_to_zero(bfloat16, float_status *status);
378 
379 uint16_t bfloat16_to_uint16_scalbn(bfloat16 a, FloatRoundMode,
380                                    int, float_status *status);
381 uint32_t bfloat16_to_uint32_scalbn(bfloat16 a, FloatRoundMode,
382                                    int, float_status *status);
383 uint64_t bfloat16_to_uint64_scalbn(bfloat16 a, FloatRoundMode,
384                                    int, float_status *status);
385 
386 uint16_t bfloat16_to_uint16(bfloat16 a, float_status *status);
387 uint32_t bfloat16_to_uint32(bfloat16 a, float_status *status);
388 uint64_t bfloat16_to_uint64(bfloat16 a, float_status *status);
389 
390 uint16_t bfloat16_to_uint16_round_to_zero(bfloat16 a, float_status *status);
391 uint32_t bfloat16_to_uint32_round_to_zero(bfloat16 a, float_status *status);
392 uint64_t bfloat16_to_uint64_round_to_zero(bfloat16 a, float_status *status);
393 
394 bfloat16 int16_to_bfloat16_scalbn(int16_t a, int, float_status *status);
395 bfloat16 int32_to_bfloat16_scalbn(int32_t a, int, float_status *status);
396 bfloat16 int64_to_bfloat16_scalbn(int64_t a, int, float_status *status);
397 bfloat16 uint16_to_bfloat16_scalbn(uint16_t a, int, float_status *status);
398 bfloat16 uint32_to_bfloat16_scalbn(uint32_t a, int, float_status *status);
399 bfloat16 uint64_to_bfloat16_scalbn(uint64_t a, int, float_status *status);
400 
401 bfloat16 int16_to_bfloat16(int16_t a, float_status *status);
402 bfloat16 int32_to_bfloat16(int32_t a, float_status *status);
403 bfloat16 int64_to_bfloat16(int64_t a, float_status *status);
404 bfloat16 uint16_to_bfloat16(uint16_t a, float_status *status);
405 bfloat16 uint32_to_bfloat16(uint32_t a, float_status *status);
406 bfloat16 uint64_to_bfloat16(uint64_t a, float_status *status);
407 
408 /*----------------------------------------------------------------------------
409 | Software bfloat16 operations.
410 *----------------------------------------------------------------------------*/
411 
412 bfloat16 bfloat16_add(bfloat16, bfloat16, float_status *status);
413 bfloat16 bfloat16_sub(bfloat16, bfloat16, float_status *status);
414 bfloat16 bfloat16_mul(bfloat16, bfloat16, float_status *status);
415 bfloat16 bfloat16_div(bfloat16, bfloat16, float_status *status);
416 bfloat16 bfloat16_muladd(bfloat16, bfloat16, bfloat16, int,
417                          float_status *status);
418 float16 bfloat16_scalbn(bfloat16, int, float_status *status);
419 bfloat16 bfloat16_min(bfloat16, bfloat16, float_status *status);
420 bfloat16 bfloat16_max(bfloat16, bfloat16, float_status *status);
421 bfloat16 bfloat16_minnum(bfloat16, bfloat16, float_status *status);
422 bfloat16 bfloat16_maxnum(bfloat16, bfloat16, float_status *status);
423 bfloat16 bfloat16_minnummag(bfloat16, bfloat16, float_status *status);
424 bfloat16 bfloat16_maxnummag(bfloat16, bfloat16, float_status *status);
425 bfloat16 bfloat16_sqrt(bfloat16, float_status *status);
426 FloatRelation bfloat16_compare(bfloat16, bfloat16, float_status *status);
427 FloatRelation bfloat16_compare_quiet(bfloat16, bfloat16, float_status *status);
428 
429 bool bfloat16_is_quiet_nan(bfloat16, float_status *status);
430 bool bfloat16_is_signaling_nan(bfloat16, float_status *status);
431 bfloat16 bfloat16_silence_nan(bfloat16, float_status *status);
432 bfloat16 bfloat16_default_nan(float_status *status);
433 
434 static inline bool bfloat16_is_any_nan(bfloat16 a)
435 {
436     return ((a & ~0x8000) > 0x7F80);
437 }
438 
439 static inline bool bfloat16_is_neg(bfloat16 a)
440 {
441     return a >> 15;
442 }
443 
444 static inline bool bfloat16_is_infinity(bfloat16 a)
445 {
446     return (a & 0x7fff) == 0x7F80;
447 }
448 
449 static inline bool bfloat16_is_zero(bfloat16 a)
450 {
451     return (a & 0x7fff) == 0;
452 }
453 
454 static inline bool bfloat16_is_zero_or_denormal(bfloat16 a)
455 {
456     return (a & 0x7F80) == 0;
457 }
458 
459 static inline bool bfloat16_is_normal(bfloat16 a)
460 {
461     return (((a >> 7) + 1) & 0xff) >= 2;
462 }
463 
464 static inline bfloat16 bfloat16_abs(bfloat16 a)
465 {
466     /* Note that abs does *not* handle NaN specially, nor does
467      * it flush denormal inputs to zero.
468      */
469     return a & 0x7fff;
470 }
471 
472 static inline bfloat16 bfloat16_chs(bfloat16 a)
473 {
474     /* Note that chs does *not* handle NaN specially, nor does
475      * it flush denormal inputs to zero.
476      */
477     return a ^ 0x8000;
478 }
479 
480 static inline bfloat16 bfloat16_set_sign(bfloat16 a, int sign)
481 {
482     return (a & 0x7fff) | (sign << 15);
483 }
484 
485 static inline bool bfloat16_eq(bfloat16 a, bfloat16 b, float_status *s)
486 {
487     return bfloat16_compare(a, b, s) == float_relation_equal;
488 }
489 
490 static inline bool bfloat16_le(bfloat16 a, bfloat16 b, float_status *s)
491 {
492     return bfloat16_compare(a, b, s) <= float_relation_equal;
493 }
494 
495 static inline bool bfloat16_lt(bfloat16 a, bfloat16 b, float_status *s)
496 {
497     return bfloat16_compare(a, b, s) < float_relation_equal;
498 }
499 
500 static inline bool bfloat16_unordered(bfloat16 a, bfloat16 b, float_status *s)
501 {
502     return bfloat16_compare(a, b, s) == float_relation_unordered;
503 }
504 
505 static inline bool bfloat16_eq_quiet(bfloat16 a, bfloat16 b, float_status *s)
506 {
507     return bfloat16_compare_quiet(a, b, s) == float_relation_equal;
508 }
509 
510 static inline bool bfloat16_le_quiet(bfloat16 a, bfloat16 b, float_status *s)
511 {
512     return bfloat16_compare_quiet(a, b, s) <= float_relation_equal;
513 }
514 
515 static inline bool bfloat16_lt_quiet(bfloat16 a, bfloat16 b, float_status *s)
516 {
517     return bfloat16_compare_quiet(a, b, s) < float_relation_equal;
518 }
519 
520 static inline bool bfloat16_unordered_quiet(bfloat16 a, bfloat16 b,
521                                            float_status *s)
522 {
523     return bfloat16_compare_quiet(a, b, s) == float_relation_unordered;
524 }
525 
526 #define bfloat16_zero 0
527 #define bfloat16_half 0x3f00
528 #define bfloat16_one 0x3f80
529 #define bfloat16_one_point_five 0x3fc0
530 #define bfloat16_two 0x4000
531 #define bfloat16_three 0x4040
532 #define bfloat16_infinity 0x7f80
533 
534 /*----------------------------------------------------------------------------
535 | The pattern for a default generated half-precision NaN.
536 *----------------------------------------------------------------------------*/
537 float16 float16_default_nan(float_status *status);
538 
539 /*----------------------------------------------------------------------------
540 | Software IEC/IEEE single-precision conversion routines.
541 *----------------------------------------------------------------------------*/
542 
543 int16_t float32_to_int16_scalbn(float32, FloatRoundMode, int, float_status *);
544 int32_t float32_to_int32_scalbn(float32, FloatRoundMode, int, float_status *);
545 int64_t float32_to_int64_scalbn(float32, FloatRoundMode, int, float_status *);
546 
547 int16_t float32_to_int16(float32, float_status *status);
548 int32_t float32_to_int32(float32, float_status *status);
549 int64_t float32_to_int64(float32, float_status *status);
550 
551 int16_t float32_to_int16_round_to_zero(float32, float_status *status);
552 int32_t float32_to_int32_round_to_zero(float32, float_status *status);
553 int64_t float32_to_int64_round_to_zero(float32, float_status *status);
554 
555 uint16_t float32_to_uint16_scalbn(float32, FloatRoundMode, int, float_status *);
556 uint32_t float32_to_uint32_scalbn(float32, FloatRoundMode, int, float_status *);
557 uint64_t float32_to_uint64_scalbn(float32, FloatRoundMode, int, float_status *);
558 
559 uint16_t float32_to_uint16(float32, float_status *status);
560 uint32_t float32_to_uint32(float32, float_status *status);
561 uint64_t float32_to_uint64(float32, float_status *status);
562 
563 uint16_t float32_to_uint16_round_to_zero(float32, float_status *status);
564 uint32_t float32_to_uint32_round_to_zero(float32, float_status *status);
565 uint64_t float32_to_uint64_round_to_zero(float32, float_status *status);
566 
567 float64 float32_to_float64(float32, float_status *status);
568 floatx80 float32_to_floatx80(float32, float_status *status);
569 float128 float32_to_float128(float32, float_status *status);
570 
571 /*----------------------------------------------------------------------------
572 | Software IEC/IEEE single-precision operations.
573 *----------------------------------------------------------------------------*/
574 float32 float32_round_to_int(float32, float_status *status);
575 float32 float32_add(float32, float32, float_status *status);
576 float32 float32_sub(float32, float32, float_status *status);
577 float32 float32_mul(float32, float32, float_status *status);
578 float32 float32_div(float32, float32, float_status *status);
579 float32 float32_rem(float32, float32, float_status *status);
580 float32 float32_muladd(float32, float32, float32, int, float_status *status);
581 float32 float32_sqrt(float32, float_status *status);
582 float32 float32_exp2(float32, float_status *status);
583 float32 float32_log2(float32, float_status *status);
584 FloatRelation float32_compare(float32, float32, float_status *status);
585 FloatRelation float32_compare_quiet(float32, float32, float_status *status);
586 float32 float32_min(float32, float32, float_status *status);
587 float32 float32_max(float32, float32, float_status *status);
588 float32 float32_minnum(float32, float32, float_status *status);
589 float32 float32_maxnum(float32, float32, float_status *status);
590 float32 float32_minnummag(float32, float32, float_status *status);
591 float32 float32_maxnummag(float32, float32, float_status *status);
592 bool float32_is_quiet_nan(float32, float_status *status);
593 bool float32_is_signaling_nan(float32, float_status *status);
594 float32 float32_silence_nan(float32, float_status *status);
595 float32 float32_scalbn(float32, int, float_status *status);
596 
597 static inline float32 float32_abs(float32 a)
598 {
599     /* Note that abs does *not* handle NaN specially, nor does
600      * it flush denormal inputs to zero.
601      */
602     return make_float32(float32_val(a) & 0x7fffffff);
603 }
604 
605 static inline float32 float32_chs(float32 a)
606 {
607     /* Note that chs does *not* handle NaN specially, nor does
608      * it flush denormal inputs to zero.
609      */
610     return make_float32(float32_val(a) ^ 0x80000000);
611 }
612 
613 static inline bool float32_is_infinity(float32 a)
614 {
615     return (float32_val(a) & 0x7fffffff) == 0x7f800000;
616 }
617 
618 static inline bool float32_is_neg(float32 a)
619 {
620     return float32_val(a) >> 31;
621 }
622 
623 static inline bool float32_is_zero(float32 a)
624 {
625     return (float32_val(a) & 0x7fffffff) == 0;
626 }
627 
628 static inline bool float32_is_any_nan(float32 a)
629 {
630     return ((float32_val(a) & ~(1 << 31)) > 0x7f800000UL);
631 }
632 
633 static inline bool float32_is_zero_or_denormal(float32 a)
634 {
635     return (float32_val(a) & 0x7f800000) == 0;
636 }
637 
638 static inline bool float32_is_normal(float32 a)
639 {
640     return (((float32_val(a) >> 23) + 1) & 0xff) >= 2;
641 }
642 
643 static inline bool float32_is_denormal(float32 a)
644 {
645     return float32_is_zero_or_denormal(a) && !float32_is_zero(a);
646 }
647 
648 static inline bool float32_is_zero_or_normal(float32 a)
649 {
650     return float32_is_normal(a) || float32_is_zero(a);
651 }
652 
653 static inline float32 float32_set_sign(float32 a, int sign)
654 {
655     return make_float32((float32_val(a) & 0x7fffffff) | (sign << 31));
656 }
657 
658 static inline bool float32_eq(float32 a, float32 b, float_status *s)
659 {
660     return float32_compare(a, b, s) == float_relation_equal;
661 }
662 
663 static inline bool float32_le(float32 a, float32 b, float_status *s)
664 {
665     return float32_compare(a, b, s) <= float_relation_equal;
666 }
667 
668 static inline bool float32_lt(float32 a, float32 b, float_status *s)
669 {
670     return float32_compare(a, b, s) < float_relation_equal;
671 }
672 
673 static inline bool float32_unordered(float32 a, float32 b, float_status *s)
674 {
675     return float32_compare(a, b, s) == float_relation_unordered;
676 }
677 
678 static inline bool float32_eq_quiet(float32 a, float32 b, float_status *s)
679 {
680     return float32_compare_quiet(a, b, s) == float_relation_equal;
681 }
682 
683 static inline bool float32_le_quiet(float32 a, float32 b, float_status *s)
684 {
685     return float32_compare_quiet(a, b, s) <= float_relation_equal;
686 }
687 
688 static inline bool float32_lt_quiet(float32 a, float32 b, float_status *s)
689 {
690     return float32_compare_quiet(a, b, s) < float_relation_equal;
691 }
692 
693 static inline bool float32_unordered_quiet(float32 a, float32 b,
694                                            float_status *s)
695 {
696     return float32_compare_quiet(a, b, s) == float_relation_unordered;
697 }
698 
699 #define float32_zero make_float32(0)
700 #define float32_half make_float32(0x3f000000)
701 #define float32_one make_float32(0x3f800000)
702 #define float32_one_point_five make_float32(0x3fc00000)
703 #define float32_two make_float32(0x40000000)
704 #define float32_three make_float32(0x40400000)
705 #define float32_infinity make_float32(0x7f800000)
706 
707 /*----------------------------------------------------------------------------
708 | Packs the sign `zSign', exponent `zExp', and significand `zSig' into a
709 | single-precision floating-point value, returning the result.  After being
710 | shifted into the proper positions, the three fields are simply added
711 | together to form the result.  This means that any integer portion of `zSig'
712 | will be added into the exponent.  Since a properly normalized significand
713 | will have an integer portion equal to 1, the `zExp' input should be 1 less
714 | than the desired result exponent whenever `zSig' is a complete, normalized
715 | significand.
716 *----------------------------------------------------------------------------*/
717 
718 static inline float32 packFloat32(bool zSign, int zExp, uint32_t zSig)
719 {
720     return make_float32(
721           (((uint32_t)zSign) << 31) + (((uint32_t)zExp) << 23) + zSig);
722 }
723 
724 /*----------------------------------------------------------------------------
725 | The pattern for a default generated single-precision NaN.
726 *----------------------------------------------------------------------------*/
727 float32 float32_default_nan(float_status *status);
728 
729 /*----------------------------------------------------------------------------
730 | Software IEC/IEEE double-precision conversion routines.
731 *----------------------------------------------------------------------------*/
732 
733 int16_t float64_to_int16_scalbn(float64, FloatRoundMode, int, float_status *);
734 int32_t float64_to_int32_scalbn(float64, FloatRoundMode, int, float_status *);
735 int64_t float64_to_int64_scalbn(float64, FloatRoundMode, int, float_status *);
736 
737 int16_t float64_to_int16(float64, float_status *status);
738 int32_t float64_to_int32(float64, float_status *status);
739 int64_t float64_to_int64(float64, float_status *status);
740 
741 int16_t float64_to_int16_round_to_zero(float64, float_status *status);
742 int32_t float64_to_int32_round_to_zero(float64, float_status *status);
743 int64_t float64_to_int64_round_to_zero(float64, float_status *status);
744 
745 uint16_t float64_to_uint16_scalbn(float64, FloatRoundMode, int, float_status *);
746 uint32_t float64_to_uint32_scalbn(float64, FloatRoundMode, int, float_status *);
747 uint64_t float64_to_uint64_scalbn(float64, FloatRoundMode, int, float_status *);
748 
749 uint16_t float64_to_uint16(float64, float_status *status);
750 uint32_t float64_to_uint32(float64, float_status *status);
751 uint64_t float64_to_uint64(float64, float_status *status);
752 
753 uint16_t float64_to_uint16_round_to_zero(float64, float_status *status);
754 uint32_t float64_to_uint32_round_to_zero(float64, float_status *status);
755 uint64_t float64_to_uint64_round_to_zero(float64, float_status *status);
756 
757 float32 float64_to_float32(float64, float_status *status);
758 floatx80 float64_to_floatx80(float64, float_status *status);
759 float128 float64_to_float128(float64, float_status *status);
760 
761 /*----------------------------------------------------------------------------
762 | Software IEC/IEEE double-precision operations.
763 *----------------------------------------------------------------------------*/
764 float64 float64_round_to_int(float64, float_status *status);
765 float64 float64_add(float64, float64, float_status *status);
766 float64 float64_sub(float64, float64, float_status *status);
767 float64 float64_mul(float64, float64, float_status *status);
768 float64 float64_div(float64, float64, float_status *status);
769 float64 float64_rem(float64, float64, float_status *status);
770 float64 float64_muladd(float64, float64, float64, int, float_status *status);
771 float64 float64_sqrt(float64, float_status *status);
772 float64 float64_log2(float64, float_status *status);
773 FloatRelation float64_compare(float64, float64, float_status *status);
774 FloatRelation float64_compare_quiet(float64, float64, float_status *status);
775 float64 float64_min(float64, float64, float_status *status);
776 float64 float64_max(float64, float64, float_status *status);
777 float64 float64_minnum(float64, float64, float_status *status);
778 float64 float64_maxnum(float64, float64, float_status *status);
779 float64 float64_minnummag(float64, float64, float_status *status);
780 float64 float64_maxnummag(float64, float64, float_status *status);
781 bool float64_is_quiet_nan(float64 a, float_status *status);
782 bool float64_is_signaling_nan(float64, float_status *status);
783 float64 float64_silence_nan(float64, float_status *status);
784 float64 float64_scalbn(float64, int, float_status *status);
785 
786 static inline float64 float64_abs(float64 a)
787 {
788     /* Note that abs does *not* handle NaN specially, nor does
789      * it flush denormal inputs to zero.
790      */
791     return make_float64(float64_val(a) & 0x7fffffffffffffffLL);
792 }
793 
794 static inline float64 float64_chs(float64 a)
795 {
796     /* Note that chs does *not* handle NaN specially, nor does
797      * it flush denormal inputs to zero.
798      */
799     return make_float64(float64_val(a) ^ 0x8000000000000000LL);
800 }
801 
802 static inline bool float64_is_infinity(float64 a)
803 {
804     return (float64_val(a) & 0x7fffffffffffffffLL ) == 0x7ff0000000000000LL;
805 }
806 
807 static inline bool float64_is_neg(float64 a)
808 {
809     return float64_val(a) >> 63;
810 }
811 
812 static inline bool float64_is_zero(float64 a)
813 {
814     return (float64_val(a) & 0x7fffffffffffffffLL) == 0;
815 }
816 
817 static inline bool float64_is_any_nan(float64 a)
818 {
819     return ((float64_val(a) & ~(1ULL << 63)) > 0x7ff0000000000000ULL);
820 }
821 
822 static inline bool float64_is_zero_or_denormal(float64 a)
823 {
824     return (float64_val(a) & 0x7ff0000000000000LL) == 0;
825 }
826 
827 static inline bool float64_is_normal(float64 a)
828 {
829     return (((float64_val(a) >> 52) + 1) & 0x7ff) >= 2;
830 }
831 
832 static inline bool float64_is_denormal(float64 a)
833 {
834     return float64_is_zero_or_denormal(a) && !float64_is_zero(a);
835 }
836 
837 static inline bool float64_is_zero_or_normal(float64 a)
838 {
839     return float64_is_normal(a) || float64_is_zero(a);
840 }
841 
842 static inline float64 float64_set_sign(float64 a, int sign)
843 {
844     return make_float64((float64_val(a) & 0x7fffffffffffffffULL)
845                         | ((int64_t)sign << 63));
846 }
847 
848 static inline bool float64_eq(float64 a, float64 b, float_status *s)
849 {
850     return float64_compare(a, b, s) == float_relation_equal;
851 }
852 
853 static inline bool float64_le(float64 a, float64 b, float_status *s)
854 {
855     return float64_compare(a, b, s) <= float_relation_equal;
856 }
857 
858 static inline bool float64_lt(float64 a, float64 b, float_status *s)
859 {
860     return float64_compare(a, b, s) < float_relation_equal;
861 }
862 
863 static inline bool float64_unordered(float64 a, float64 b, float_status *s)
864 {
865     return float64_compare(a, b, s) == float_relation_unordered;
866 }
867 
868 static inline bool float64_eq_quiet(float64 a, float64 b, float_status *s)
869 {
870     return float64_compare_quiet(a, b, s) == float_relation_equal;
871 }
872 
873 static inline bool float64_le_quiet(float64 a, float64 b, float_status *s)
874 {
875     return float64_compare_quiet(a, b, s) <= float_relation_equal;
876 }
877 
878 static inline bool float64_lt_quiet(float64 a, float64 b, float_status *s)
879 {
880     return float64_compare_quiet(a, b, s) < float_relation_equal;
881 }
882 
883 static inline bool float64_unordered_quiet(float64 a, float64 b,
884                                            float_status *s)
885 {
886     return float64_compare_quiet(a, b, s) == float_relation_unordered;
887 }
888 
889 #define float64_zero make_float64(0)
890 #define float64_half make_float64(0x3fe0000000000000LL)
891 #define float64_one make_float64(0x3ff0000000000000LL)
892 #define float64_one_point_five make_float64(0x3FF8000000000000ULL)
893 #define float64_two make_float64(0x4000000000000000ULL)
894 #define float64_three make_float64(0x4008000000000000ULL)
895 #define float64_ln2 make_float64(0x3fe62e42fefa39efLL)
896 #define float64_infinity make_float64(0x7ff0000000000000LL)
897 
898 /*----------------------------------------------------------------------------
899 | The pattern for a default generated double-precision NaN.
900 *----------------------------------------------------------------------------*/
901 float64 float64_default_nan(float_status *status);
902 
903 /*----------------------------------------------------------------------------
904 | Software IEC/IEEE extended double-precision conversion routines.
905 *----------------------------------------------------------------------------*/
906 int32_t floatx80_to_int32(floatx80, float_status *status);
907 int32_t floatx80_to_int32_round_to_zero(floatx80, float_status *status);
908 int64_t floatx80_to_int64(floatx80, float_status *status);
909 int64_t floatx80_to_int64_round_to_zero(floatx80, float_status *status);
910 float32 floatx80_to_float32(floatx80, float_status *status);
911 float64 floatx80_to_float64(floatx80, float_status *status);
912 float128 floatx80_to_float128(floatx80, float_status *status);
913 
914 /*----------------------------------------------------------------------------
915 | The pattern for an extended double-precision inf.
916 *----------------------------------------------------------------------------*/
917 extern const floatx80 floatx80_infinity;
918 
919 /*----------------------------------------------------------------------------
920 | Software IEC/IEEE extended double-precision operations.
921 *----------------------------------------------------------------------------*/
922 floatx80 floatx80_round(floatx80 a, float_status *status);
923 floatx80 floatx80_round_to_int(floatx80, float_status *status);
924 floatx80 floatx80_add(floatx80, floatx80, float_status *status);
925 floatx80 floatx80_sub(floatx80, floatx80, float_status *status);
926 floatx80 floatx80_mul(floatx80, floatx80, float_status *status);
927 floatx80 floatx80_div(floatx80, floatx80, float_status *status);
928 floatx80 floatx80_modrem(floatx80, floatx80, bool, uint64_t *,
929                          float_status *status);
930 floatx80 floatx80_mod(floatx80, floatx80, float_status *status);
931 floatx80 floatx80_rem(floatx80, floatx80, float_status *status);
932 floatx80 floatx80_sqrt(floatx80, float_status *status);
933 FloatRelation floatx80_compare(floatx80, floatx80, float_status *status);
934 FloatRelation floatx80_compare_quiet(floatx80, floatx80, float_status *status);
935 int floatx80_is_quiet_nan(floatx80, float_status *status);
936 int floatx80_is_signaling_nan(floatx80, float_status *status);
937 floatx80 floatx80_silence_nan(floatx80, float_status *status);
938 floatx80 floatx80_scalbn(floatx80, int, float_status *status);
939 
940 static inline floatx80 floatx80_abs(floatx80 a)
941 {
942     a.high &= 0x7fff;
943     return a;
944 }
945 
946 static inline floatx80 floatx80_chs(floatx80 a)
947 {
948     a.high ^= 0x8000;
949     return a;
950 }
951 
952 static inline bool floatx80_is_infinity(floatx80 a)
953 {
954 #if defined(TARGET_M68K)
955     return (a.high & 0x7fff) == floatx80_infinity.high && !(a.low << 1);
956 #else
957     return (a.high & 0x7fff) == floatx80_infinity.high &&
958                        a.low == floatx80_infinity.low;
959 #endif
960 }
961 
962 static inline bool floatx80_is_neg(floatx80 a)
963 {
964     return a.high >> 15;
965 }
966 
967 static inline bool floatx80_is_zero(floatx80 a)
968 {
969     return (a.high & 0x7fff) == 0 && a.low == 0;
970 }
971 
972 static inline bool floatx80_is_zero_or_denormal(floatx80 a)
973 {
974     return (a.high & 0x7fff) == 0;
975 }
976 
977 static inline bool floatx80_is_any_nan(floatx80 a)
978 {
979     return ((a.high & 0x7fff) == 0x7fff) && (a.low<<1);
980 }
981 
982 static inline bool floatx80_eq(floatx80 a, floatx80 b, float_status *s)
983 {
984     return floatx80_compare(a, b, s) == float_relation_equal;
985 }
986 
987 static inline bool floatx80_le(floatx80 a, floatx80 b, float_status *s)
988 {
989     return floatx80_compare(a, b, s) <= float_relation_equal;
990 }
991 
992 static inline bool floatx80_lt(floatx80 a, floatx80 b, float_status *s)
993 {
994     return floatx80_compare(a, b, s) < float_relation_equal;
995 }
996 
997 static inline bool floatx80_unordered(floatx80 a, floatx80 b, float_status *s)
998 {
999     return floatx80_compare(a, b, s) == float_relation_unordered;
1000 }
1001 
1002 static inline bool floatx80_eq_quiet(floatx80 a, floatx80 b, float_status *s)
1003 {
1004     return floatx80_compare_quiet(a, b, s) == float_relation_equal;
1005 }
1006 
1007 static inline bool floatx80_le_quiet(floatx80 a, floatx80 b, float_status *s)
1008 {
1009     return floatx80_compare_quiet(a, b, s) <= float_relation_equal;
1010 }
1011 
1012 static inline bool floatx80_lt_quiet(floatx80 a, floatx80 b, float_status *s)
1013 {
1014     return floatx80_compare_quiet(a, b, s) < float_relation_equal;
1015 }
1016 
1017 static inline bool floatx80_unordered_quiet(floatx80 a, floatx80 b,
1018                                            float_status *s)
1019 {
1020     return floatx80_compare_quiet(a, b, s) == float_relation_unordered;
1021 }
1022 
1023 /*----------------------------------------------------------------------------
1024 | Return whether the given value is an invalid floatx80 encoding.
1025 | Invalid floatx80 encodings arise when the integer bit is not set, but
1026 | the exponent is not zero. The only times the integer bit is permitted to
1027 | be zero is in subnormal numbers and the value zero.
1028 | This includes what the Intel software developer's manual calls pseudo-NaNs,
1029 | pseudo-infinities and un-normal numbers. It does not include
1030 | pseudo-denormals, which must still be correctly handled as inputs even
1031 | if they are never generated as outputs.
1032 *----------------------------------------------------------------------------*/
1033 static inline bool floatx80_invalid_encoding(floatx80 a)
1034 {
1035 #if defined(TARGET_M68K)
1036     /*-------------------------------------------------------------------------
1037     | With m68k, the explicit integer bit can be zero in the case of:
1038     | - zeros                (exp == 0, mantissa == 0)
1039     | - denormalized numbers (exp == 0, mantissa != 0)
1040     | - unnormalized numbers (exp != 0, exp < 0x7FFF)
1041     | - infinities           (exp == 0x7FFF, mantissa == 0)
1042     | - not-a-numbers        (exp == 0x7FFF, mantissa != 0)
1043     |
1044     | For infinities and NaNs, the explicit integer bit can be either one or
1045     | zero.
1046     |
1047     | The IEEE 754 standard does not define a zero integer bit. Such a number
1048     | is an unnormalized number. Hardware does not directly support
1049     | denormalized and unnormalized numbers, but implicitly supports them by
1050     | trapping them as unimplemented data types, allowing efficient conversion
1051     | in software.
1052     |
1053     | See "M68000 FAMILY PROGRAMMER’S REFERENCE MANUAL",
1054     |     "1.6 FLOATING-POINT DATA TYPES"
1055     *------------------------------------------------------------------------*/
1056     return false;
1057 #else
1058     return (a.low & (1ULL << 63)) == 0 && (a.high & 0x7FFF) != 0;
1059 #endif
1060 }
1061 
1062 #define floatx80_zero make_floatx80(0x0000, 0x0000000000000000LL)
1063 #define floatx80_zero_init make_floatx80_init(0x0000, 0x0000000000000000LL)
1064 #define floatx80_one make_floatx80(0x3fff, 0x8000000000000000LL)
1065 #define floatx80_ln2 make_floatx80(0x3ffe, 0xb17217f7d1cf79acLL)
1066 #define floatx80_pi make_floatx80(0x4000, 0xc90fdaa22168c235LL)
1067 #define floatx80_half make_floatx80(0x3ffe, 0x8000000000000000LL)
1068 
1069 /*----------------------------------------------------------------------------
1070 | Returns the fraction bits of the extended double-precision floating-point
1071 | value `a'.
1072 *----------------------------------------------------------------------------*/
1073 
1074 static inline uint64_t extractFloatx80Frac(floatx80 a)
1075 {
1076     return a.low;
1077 }
1078 
1079 /*----------------------------------------------------------------------------
1080 | Returns the exponent bits of the extended double-precision floating-point
1081 | value `a'.
1082 *----------------------------------------------------------------------------*/
1083 
1084 static inline int32_t extractFloatx80Exp(floatx80 a)
1085 {
1086     return a.high & 0x7FFF;
1087 }
1088 
1089 /*----------------------------------------------------------------------------
1090 | Returns the sign bit of the extended double-precision floating-point value
1091 | `a'.
1092 *----------------------------------------------------------------------------*/
1093 
1094 static inline bool extractFloatx80Sign(floatx80 a)
1095 {
1096     return a.high >> 15;
1097 }
1098 
1099 /*----------------------------------------------------------------------------
1100 | Packs the sign `zSign', exponent `zExp', and significand `zSig' into an
1101 | extended double-precision floating-point value, returning the result.
1102 *----------------------------------------------------------------------------*/
1103 
1104 static inline floatx80 packFloatx80(bool zSign, int32_t zExp, uint64_t zSig)
1105 {
1106     floatx80 z;
1107 
1108     z.low = zSig;
1109     z.high = (((uint16_t)zSign) << 15) + zExp;
1110     return z;
1111 }
1112 
1113 /*----------------------------------------------------------------------------
1114 | Normalizes the subnormal extended double-precision floating-point value
1115 | represented by the denormalized significand `aSig'.  The normalized exponent
1116 | and significand are stored at the locations pointed to by `zExpPtr' and
1117 | `zSigPtr', respectively.
1118 *----------------------------------------------------------------------------*/
1119 
1120 void normalizeFloatx80Subnormal(uint64_t aSig, int32_t *zExpPtr,
1121                                 uint64_t *zSigPtr);
1122 
1123 /*----------------------------------------------------------------------------
1124 | Takes two extended double-precision floating-point values `a' and `b', one
1125 | of which is a NaN, and returns the appropriate NaN result.  If either `a' or
1126 | `b' is a signaling NaN, the invalid exception is raised.
1127 *----------------------------------------------------------------------------*/
1128 
1129 floatx80 propagateFloatx80NaN(floatx80 a, floatx80 b, float_status *status);
1130 
1131 /*----------------------------------------------------------------------------
1132 | Takes an abstract floating-point value having sign `zSign', exponent `zExp',
1133 | and extended significand formed by the concatenation of `zSig0' and `zSig1',
1134 | and returns the proper extended double-precision floating-point value
1135 | corresponding to the abstract input.  Ordinarily, the abstract value is
1136 | rounded and packed into the extended double-precision format, with the
1137 | inexact exception raised if the abstract input cannot be represented
1138 | exactly.  However, if the abstract value is too large, the overflow and
1139 | inexact exceptions are raised and an infinity or maximal finite value is
1140 | returned.  If the abstract value is too small, the input value is rounded to
1141 | a subnormal number, and the underflow and inexact exceptions are raised if
1142 | the abstract input cannot be represented exactly as a subnormal extended
1143 | double-precision floating-point number.
1144 |     If `roundingPrecision' is 32 or 64, the result is rounded to the same
1145 | number of bits as single or double precision, respectively.  Otherwise, the
1146 | result is rounded to the full precision of the extended double-precision
1147 | format.
1148 |     The input significand must be normalized or smaller.  If the input
1149 | significand is not normalized, `zExp' must be 0; in that case, the result
1150 | returned is a subnormal number, and it must not require rounding.  The
1151 | handling of underflow and overflow follows the IEC/IEEE Standard for Binary
1152 | Floating-Point Arithmetic.
1153 *----------------------------------------------------------------------------*/
1154 
1155 floatx80 roundAndPackFloatx80(FloatX80RoundPrec roundingPrecision, bool zSign,
1156                               int32_t zExp, uint64_t zSig0, uint64_t zSig1,
1157                               float_status *status);
1158 
1159 /*----------------------------------------------------------------------------
1160 | Takes an abstract floating-point value having sign `zSign', exponent
1161 | `zExp', and significand formed by the concatenation of `zSig0' and `zSig1',
1162 | and returns the proper extended double-precision floating-point value
1163 | corresponding to the abstract input.  This routine is just like
1164 | `roundAndPackFloatx80' except that the input significand does not have to be
1165 | normalized.
1166 *----------------------------------------------------------------------------*/
1167 
1168 floatx80 normalizeRoundAndPackFloatx80(FloatX80RoundPrec roundingPrecision,
1169                                        bool zSign, int32_t zExp,
1170                                        uint64_t zSig0, uint64_t zSig1,
1171                                        float_status *status);
1172 
1173 /*----------------------------------------------------------------------------
1174 | The pattern for a default generated extended double-precision NaN.
1175 *----------------------------------------------------------------------------*/
1176 floatx80 floatx80_default_nan(float_status *status);
1177 
1178 /*----------------------------------------------------------------------------
1179 | Software IEC/IEEE quadruple-precision conversion routines.
1180 *----------------------------------------------------------------------------*/
1181 int32_t float128_to_int32(float128, float_status *status);
1182 int32_t float128_to_int32_round_to_zero(float128, float_status *status);
1183 int64_t float128_to_int64(float128, float_status *status);
1184 int64_t float128_to_int64_round_to_zero(float128, float_status *status);
1185 uint64_t float128_to_uint64(float128, float_status *status);
1186 uint64_t float128_to_uint64_round_to_zero(float128, float_status *status);
1187 uint32_t float128_to_uint32(float128, float_status *status);
1188 uint32_t float128_to_uint32_round_to_zero(float128, float_status *status);
1189 float32 float128_to_float32(float128, float_status *status);
1190 float64 float128_to_float64(float128, float_status *status);
1191 floatx80 float128_to_floatx80(float128, float_status *status);
1192 
1193 /*----------------------------------------------------------------------------
1194 | Software IEC/IEEE quadruple-precision operations.
1195 *----------------------------------------------------------------------------*/
1196 float128 float128_round_to_int(float128, float_status *status);
1197 float128 float128_add(float128, float128, float_status *status);
1198 float128 float128_sub(float128, float128, float_status *status);
1199 float128 float128_mul(float128, float128, float_status *status);
1200 float128 float128_muladd(float128, float128, float128, int,
1201                          float_status *status);
1202 float128 float128_div(float128, float128, float_status *status);
1203 float128 float128_rem(float128, float128, float_status *status);
1204 float128 float128_sqrt(float128, float_status *status);
1205 FloatRelation float128_compare(float128, float128, float_status *status);
1206 FloatRelation float128_compare_quiet(float128, float128, float_status *status);
1207 float128 float128_min(float128, float128, float_status *status);
1208 float128 float128_max(float128, float128, float_status *status);
1209 float128 float128_minnum(float128, float128, float_status *status);
1210 float128 float128_maxnum(float128, float128, float_status *status);
1211 float128 float128_minnummag(float128, float128, float_status *status);
1212 float128 float128_maxnummag(float128, float128, float_status *status);
1213 bool float128_is_quiet_nan(float128, float_status *status);
1214 bool float128_is_signaling_nan(float128, float_status *status);
1215 float128 float128_silence_nan(float128, float_status *status);
1216 float128 float128_scalbn(float128, int, float_status *status);
1217 
1218 static inline float128 float128_abs(float128 a)
1219 {
1220     a.high &= 0x7fffffffffffffffLL;
1221     return a;
1222 }
1223 
1224 static inline float128 float128_chs(float128 a)
1225 {
1226     a.high ^= 0x8000000000000000LL;
1227     return a;
1228 }
1229 
1230 static inline bool float128_is_infinity(float128 a)
1231 {
1232     return (a.high & 0x7fffffffffffffffLL) == 0x7fff000000000000LL && a.low == 0;
1233 }
1234 
1235 static inline bool float128_is_neg(float128 a)
1236 {
1237     return a.high >> 63;
1238 }
1239 
1240 static inline bool float128_is_zero(float128 a)
1241 {
1242     return (a.high & 0x7fffffffffffffffLL) == 0 && a.low == 0;
1243 }
1244 
1245 static inline bool float128_is_zero_or_denormal(float128 a)
1246 {
1247     return (a.high & 0x7fff000000000000LL) == 0;
1248 }
1249 
1250 static inline bool float128_is_normal(float128 a)
1251 {
1252     return (((a.high >> 48) + 1) & 0x7fff) >= 2;
1253 }
1254 
1255 static inline bool float128_is_denormal(float128 a)
1256 {
1257     return float128_is_zero_or_denormal(a) && !float128_is_zero(a);
1258 }
1259 
1260 static inline bool float128_is_any_nan(float128 a)
1261 {
1262     return ((a.high >> 48) & 0x7fff) == 0x7fff &&
1263         ((a.low != 0) || ((a.high & 0xffffffffffffLL) != 0));
1264 }
1265 
1266 static inline bool float128_eq(float128 a, float128 b, float_status *s)
1267 {
1268     return float128_compare(a, b, s) == float_relation_equal;
1269 }
1270 
1271 static inline bool float128_le(float128 a, float128 b, float_status *s)
1272 {
1273     return float128_compare(a, b, s) <= float_relation_equal;
1274 }
1275 
1276 static inline bool float128_lt(float128 a, float128 b, float_status *s)
1277 {
1278     return float128_compare(a, b, s) < float_relation_equal;
1279 }
1280 
1281 static inline bool float128_unordered(float128 a, float128 b, float_status *s)
1282 {
1283     return float128_compare(a, b, s) == float_relation_unordered;
1284 }
1285 
1286 static inline bool float128_eq_quiet(float128 a, float128 b, float_status *s)
1287 {
1288     return float128_compare_quiet(a, b, s) == float_relation_equal;
1289 }
1290 
1291 static inline bool float128_le_quiet(float128 a, float128 b, float_status *s)
1292 {
1293     return float128_compare_quiet(a, b, s) <= float_relation_equal;
1294 }
1295 
1296 static inline bool float128_lt_quiet(float128 a, float128 b, float_status *s)
1297 {
1298     return float128_compare_quiet(a, b, s) < float_relation_equal;
1299 }
1300 
1301 static inline bool float128_unordered_quiet(float128 a, float128 b,
1302                                            float_status *s)
1303 {
1304     return float128_compare_quiet(a, b, s) == float_relation_unordered;
1305 }
1306 
1307 #define float128_zero make_float128(0, 0)
1308 
1309 /*----------------------------------------------------------------------------
1310 | The pattern for a default generated quadruple-precision NaN.
1311 *----------------------------------------------------------------------------*/
1312 float128 float128_default_nan(float_status *status);
1313 
1314 #endif /* SOFTFLOAT_H */
1315