xref: /freebsd/lib/libc/riscv/softfloat/softfloat.h (revision 59c8e88e72633afbc47a4ace0d2170d00d51f7dc)
1 /*	$NetBSD: softfloat.h,v 1.6 2002/05/12 13:12:46 bjh21 Exp $	*/
2 
3 /* This is a derivative work. */
4 
5 /*
6 ===============================================================================
7 
8 This C header file is part of the SoftFloat IEC/IEEE Floating-point
9 Arithmetic Package, Release 2a.
10 
11 Written by John R. Hauser.  This work was made possible in part by the
12 International Computer Science Institute, located at Suite 600, 1947 Center
13 Street, Berkeley, California 94704.  Funding was partially provided by the
14 National Science Foundation under grant MIP-9311980.  The original version
15 of this code was written as part of a project to build a fixed-point vector
16 processor in collaboration with the University of California at Berkeley,
17 overseen by Profs. Nelson Morgan and John Wawrzynek.  More information
18 is available through the Web page `http://HTTP.CS.Berkeley.EDU/~jhauser/
19 arithmetic/SoftFloat.html'.
20 
21 THIS SOFTWARE IS DISTRIBUTED AS IS, FOR FREE.  Although reasonable effort
22 has been made to avoid it, THIS SOFTWARE MAY CONTAIN FAULTS THAT WILL AT
23 TIMES RESULT IN INCORRECT BEHAVIOR.  USE OF THIS SOFTWARE IS RESTRICTED TO
24 PERSONS AND ORGANIZATIONS WHO CAN AND WILL TAKE FULL RESPONSIBILITY FOR ANY
25 AND ALL LOSSES, COSTS, OR OTHER PROBLEMS ARISING FROM ITS USE.
26 
27 Derivative works are acceptable, even for commercial purposes, so long as
28 (1) they include prominent notice that the work is derivative, and (2) they
29 include prominent notice akin to these four paragraphs for those parts of
30 this code that are retained.
31 
32 ===============================================================================
33 */
34 
35 /*
36 -------------------------------------------------------------------------------
37 The macro `FLOATX80' must be defined to enable the extended double-precision
38 floating-point format `floatx80'.  If this macro is not defined, the
39 `floatx80' type will not be defined, and none of the functions that either
40 input or output the `floatx80' type will be defined.  The same applies to
41 the `FLOAT128' macro and the quadruple-precision format `float128'.
42 -------------------------------------------------------------------------------
43 */
44 /* #define FLOATX80 */
45 /* #define FLOAT128 */
46 
47 #include <fenv.h>
48 
49 /*
50 -------------------------------------------------------------------------------
51 Software IEC/IEEE floating-point types.
52 -------------------------------------------------------------------------------
53 */
54 typedef unsigned int float32;
55 typedef unsigned long long float64;
56 #ifdef FLOATX80
57 typedef struct {
58     unsigned short high;
59     unsigned long long low;
60 } floatx80;
61 #endif
62 #ifdef FLOAT128
63 typedef struct {
64     unsigned long long high, low;
65 } float128;
66 #endif
67 
68 /*
69 -------------------------------------------------------------------------------
70 Software IEC/IEEE floating-point underflow tininess-detection mode.
71 -------------------------------------------------------------------------------
72 */
73 #ifndef SOFTFLOAT_FOR_GCC
74 extern int float_detect_tininess;
75 #endif
76 enum {
77     float_tininess_after_rounding  = 0,
78     float_tininess_before_rounding = 1
79 };
80 
81 /*
82 -------------------------------------------------------------------------------
83 Software IEC/IEEE floating-point rounding mode.
84 -------------------------------------------------------------------------------
85 */
86 extern int float_rounding_mode;
87 enum {
88     float_round_nearest_even = FE_TONEAREST,
89     float_round_to_zero      = FE_TOWARDZERO,
90     float_round_down         = FE_DOWNWARD,
91     float_round_up           = FE_UPWARD
92 };
93 
94 /*
95 -------------------------------------------------------------------------------
96 Software IEC/IEEE floating-point exception flags.
97 -------------------------------------------------------------------------------
98 */
99 extern int float_exception_flags;
100 extern int float_exception_mask;
101 enum {
102     float_flag_inexact   = FE_INEXACT,
103     float_flag_underflow = FE_UNDERFLOW,
104     float_flag_overflow  = FE_OVERFLOW,
105     float_flag_divbyzero = FE_DIVBYZERO,
106     float_flag_invalid   = FE_INVALID
107 };
108 
109 /*
110 -------------------------------------------------------------------------------
111 Routine to raise any or all of the software IEC/IEEE floating-point
112 exception flags.
113 -------------------------------------------------------------------------------
114 */
115 void float_raise( int );
116 
117 /*
118 -------------------------------------------------------------------------------
119 Software IEC/IEEE integer-to-floating-point conversion routines.
120 -------------------------------------------------------------------------------
121 */
122 float32 int32_to_float32( int );
123 float64 int32_to_float64( int );
124 #ifdef FLOATX80
125 floatx80 int32_to_floatx80( int );
126 #endif
127 #ifdef FLOAT128
128 float128 int32_to_float128( int );
129 #endif
130 #ifndef SOFTFLOAT_FOR_GCC /* __floatdi?f is in libgcc2.c */
131 float32 int64_to_float32( long long );
132 float64 int64_to_float64( long long );
133 #ifdef FLOATX80
134 floatx80 int64_to_floatx80( long long );
135 #endif
136 #ifdef FLOAT128
137 float128 int64_to_float128( long long );
138 #endif
139 #endif
140 
141 /*
142 -------------------------------------------------------------------------------
143 Software IEC/IEEE single-precision conversion routines.
144 -------------------------------------------------------------------------------
145 */
146 int float32_to_int32( float32 );
147 int float32_to_int32_round_to_zero( float32 );
148 #if defined(SOFTFLOAT_FOR_GCC) && defined(SOFTFLOAT_NEED_FIXUNS)
149 unsigned int float32_to_uint32_round_to_zero( float32 );
150 #endif
151 #ifndef SOFTFLOAT_FOR_GCC /* __fix?fdi provided by libgcc2.c */
152 long long float32_to_int64( float32 );
153 long long float32_to_int64_round_to_zero( float32 );
154 #endif
155 float64 float32_to_float64( float32 );
156 #ifdef FLOATX80
157 floatx80 float32_to_floatx80( float32 );
158 #endif
159 #ifdef FLOAT128
160 float128 float32_to_float128( float32 );
161 #endif
162 
163 /*
164 -------------------------------------------------------------------------------
165 Software IEC/IEEE single-precision operations.
166 -------------------------------------------------------------------------------
167 */
168 float32 float32_round_to_int( float32 );
169 float32 float32_add( float32, float32 );
170 float32 float32_sub( float32, float32 );
171 float32 float32_mul( float32, float32 );
172 float32 float32_div( float32, float32 );
173 float32 float32_rem( float32, float32 );
174 float32 float32_sqrt( float32 );
175 int float32_eq( float32, float32 );
176 int float32_le( float32, float32 );
177 int float32_lt( float32, float32 );
178 int float32_eq_signaling( float32, float32 );
179 int float32_le_quiet( float32, float32 );
180 int float32_lt_quiet( float32, float32 );
181 #ifndef SOFTFLOAT_FOR_GCC
182 int float32_is_signaling_nan( float32 );
183 #endif
184 
185 /*
186 -------------------------------------------------------------------------------
187 Software IEC/IEEE double-precision conversion routines.
188 -------------------------------------------------------------------------------
189 */
190 int float64_to_int32( float64 );
191 int float64_to_int32_round_to_zero( float64 );
192 #if defined(SOFTFLOAT_FOR_GCC) && defined(SOFTFLOAT_NEED_FIXUNS)
193 unsigned int float64_to_uint32_round_to_zero( float64 );
194 #endif
195 #ifndef SOFTFLOAT_FOR_GCC /* __fix?fdi provided by libgcc2.c */
196 long long float64_to_int64( float64 );
197 long long float64_to_int64_round_to_zero( float64 );
198 #endif
199 float32 float64_to_float32( float64 );
200 #ifdef FLOATX80
201 floatx80 float64_to_floatx80( float64 );
202 #endif
203 #ifdef FLOAT128
204 float128 float64_to_float128( float64 );
205 #endif
206 
207 /*
208 -------------------------------------------------------------------------------
209 Software IEC/IEEE double-precision operations.
210 -------------------------------------------------------------------------------
211 */
212 float64 float64_round_to_int( float64 );
213 float64 float64_add( float64, float64 );
214 float64 float64_sub( float64, float64 );
215 float64 float64_mul( float64, float64 );
216 float64 float64_div( float64, float64 );
217 float64 float64_rem( float64, float64 );
218 float64 float64_sqrt( float64 );
219 int float64_eq( float64, float64 );
220 int float64_le( float64, float64 );
221 int float64_lt( float64, float64 );
222 int float64_eq_signaling( float64, float64 );
223 int float64_le_quiet( float64, float64 );
224 int float64_lt_quiet( float64, float64 );
225 #ifndef SOFTFLOAT_FOR_GCC
226 int float64_is_signaling_nan( float64 );
227 #endif
228 
229 #ifdef FLOATX80
230 
231 /*
232 -------------------------------------------------------------------------------
233 Software IEC/IEEE extended double-precision conversion routines.
234 -------------------------------------------------------------------------------
235 */
236 int floatx80_to_int32( floatx80 );
237 int floatx80_to_int32_round_to_zero( floatx80 );
238 long long floatx80_to_int64( floatx80 );
239 long long floatx80_to_int64_round_to_zero( floatx80 );
240 float32 floatx80_to_float32( floatx80 );
241 float64 floatx80_to_float64( floatx80 );
242 #ifdef FLOAT128
243 float128 floatx80_to_float128( floatx80 );
244 #endif
245 
246 /*
247 -------------------------------------------------------------------------------
248 Software IEC/IEEE extended double-precision rounding precision.  Valid
249 values are 32, 64, and 80.
250 -------------------------------------------------------------------------------
251 */
252 extern int floatx80_rounding_precision;
253 
254 /*
255 -------------------------------------------------------------------------------
256 Software IEC/IEEE extended double-precision operations.
257 -------------------------------------------------------------------------------
258 */
259 floatx80 floatx80_round_to_int( floatx80 );
260 floatx80 floatx80_add( floatx80, floatx80 );
261 floatx80 floatx80_sub( floatx80, floatx80 );
262 floatx80 floatx80_mul( floatx80, floatx80 );
263 floatx80 floatx80_div( floatx80, floatx80 );
264 floatx80 floatx80_rem( floatx80, floatx80 );
265 floatx80 floatx80_sqrt( floatx80 );
266 int floatx80_eq( floatx80, floatx80 );
267 int floatx80_le( floatx80, floatx80 );
268 int floatx80_lt( floatx80, floatx80 );
269 int floatx80_eq_signaling( floatx80, floatx80 );
270 int floatx80_le_quiet( floatx80, floatx80 );
271 int floatx80_lt_quiet( floatx80, floatx80 );
272 int floatx80_is_signaling_nan( floatx80 );
273 
274 #endif
275 
276 #ifdef FLOAT128
277 
278 /*
279 -------------------------------------------------------------------------------
280 Software IEC/IEEE quadruple-precision conversion routines.
281 -------------------------------------------------------------------------------
282 */
283 int float128_to_int32( float128 );
284 int float128_to_int32_round_to_zero( float128 );
285 long long float128_to_int64( float128 );
286 long long float128_to_int64_round_to_zero( float128 );
287 float32 float128_to_float32( float128 );
288 float64 float128_to_float64( float128 );
289 #ifdef FLOATX80
290 floatx80 float128_to_floatx80( float128 );
291 #endif
292 
293 /*
294 -------------------------------------------------------------------------------
295 Software IEC/IEEE quadruple-precision operations.
296 -------------------------------------------------------------------------------
297 */
298 float128 float128_round_to_int( float128 );
299 float128 float128_add( float128, float128 );
300 float128 float128_sub( float128, float128 );
301 float128 float128_mul( float128, float128 );
302 float128 float128_div( float128, float128 );
303 float128 float128_rem( float128, float128 );
304 float128 float128_sqrt( float128 );
305 int float128_eq( float128, float128 );
306 int float128_le( float128, float128 );
307 int float128_lt( float128, float128 );
308 int float128_eq_signaling( float128, float128 );
309 int float128_le_quiet( float128, float128 );
310 int float128_lt_quiet( float128, float128 );
311 int float128_is_signaling_nan( float128 );
312 
313 #endif
314 
315