softfloat.h revision 1.3
1 2 /* 3 =============================================================================== 4 5 This C header file is part of the SoftFloat IEC/IEEE Floating-point 6 Arithmetic Package, Release 2a. 7 8 Written by John R. Hauser. This work was made possible in part by the 9 International Computer Science Institute, located at Suite 600, 1947 Center 10 Street, Berkeley, California 94704. Funding was partially provided by the 11 National Science Foundation under grant MIP-9311980. The original version 12 of this code was written as part of a project to build a fixed-point vector 13 processor in collaboration with the University of California at Berkeley, 14 overseen by Profs. Nelson Morgan and John Wawrzynek. More information 15 is available through the Web page `http://HTTP.CS.Berkeley.EDU/~jhauser/ 16 arithmetic/SoftFloat.html'. 17 18 THIS SOFTWARE IS DISTRIBUTED AS IS, FOR FREE. Although reasonable effort 19 has been made to avoid it, THIS SOFTWARE MAY CONTAIN FAULTS THAT WILL AT 20 TIMES RESULT IN INCORRECT BEHAVIOR. USE OF THIS SOFTWARE IS RESTRICTED TO 21 PERSONS AND ORGANIZATIONS WHO CAN AND WILL TAKE FULL RESPONSIBILITY FOR ANY 22 AND ALL LOSSES, COSTS, OR OTHER PROBLEMS ARISING FROM ITS USE. 23 24 Derivative works are acceptable, even for commercial purposes, so long as 25 (1) they include prominent notice that the work is derivative, and (2) they 26 include prominent notice akin to these four paragraphs for those parts of 27 this code that are retained. 28 29 =============================================================================== 30 */ 31 32 /* 33 ------------------------------------------------------------------------------- 34 The macro `FLOATX80' must be defined to enable the extended double-precision 35 floating-point format `floatx80'. If this macro is not defined, the 36 `floatx80' type will not be defined, and none of the functions that either 37 input or output the `floatx80' type will be defined. The same applies to 38 the `FLOAT128' macro and the quadruple-precision format `float128'. 39 ------------------------------------------------------------------------------- 40 */ 41 /* #define FLOATX80 */ 42 /* #define FLOAT128 */ 43 44 /* 45 ------------------------------------------------------------------------------- 46 Software IEC/IEEE floating-point types. 47 ------------------------------------------------------------------------------- 48 */ 49 typedef unsigned int float32; 50 typedef unsigned long long float64; 51 #ifdef FLOATX80 52 typedef struct { 53 unsigned short high; 54 unsigned long long low; 55 } floatx80; 56 #endif 57 #ifdef FLOAT128 58 typedef struct { 59 unsigned long long high, low; 60 } float128; 61 #endif 62 63 /* 64 ------------------------------------------------------------------------------- 65 Software IEC/IEEE floating-point underflow tininess-detection mode. 66 ------------------------------------------------------------------------------- 67 */ 68 extern int float_detect_tininess; 69 enum { 70 float_tininess_after_rounding = 0, 71 float_tininess_before_rounding = 1 72 }; 73 74 /* 75 ------------------------------------------------------------------------------- 76 Software IEC/IEEE floating-point rounding mode. 77 ------------------------------------------------------------------------------- 78 */ 79 extern int float_rounding_mode; 80 enum { 81 float_round_nearest_even = 0, 82 float_round_to_zero = 1, 83 float_round_down = 2, 84 float_round_up = 3 85 }; 86 87 /* 88 ------------------------------------------------------------------------------- 89 Software IEC/IEEE floating-point exception flags. 90 ------------------------------------------------------------------------------- 91 */ 92 extern int float_exception_flags; 93 enum { 94 float_flag_inexact = 1, 95 float_flag_underflow = 2, 96 float_flag_overflow = 4, 97 float_flag_divbyzero = 8, 98 float_flag_invalid = 16 99 }; 100 101 /* 102 ------------------------------------------------------------------------------- 103 Routine to raise any or all of the software IEC/IEEE floating-point 104 exception flags. 105 ------------------------------------------------------------------------------- 106 */ 107 void float_raise( int ); 108 109 /* 110 ------------------------------------------------------------------------------- 111 Software IEC/IEEE integer-to-floating-point conversion routines. 112 ------------------------------------------------------------------------------- 113 */ 114 float32 int32_to_float32( int ); 115 float64 int32_to_float64( int ); 116 #ifdef FLOATX80 117 floatx80 int32_to_floatx80( int ); 118 #endif 119 #ifdef FLOAT128 120 float128 int32_to_float128( int ); 121 #endif 122 #ifndef SOFTFLOAT_FOR_GCC /* __floatdi?f is in libgcc2.c */ 123 float32 int64_to_float32( long long ); 124 float64 int64_to_float64( long long ); 125 #ifdef FLOATX80 126 floatx80 int64_to_floatx80( long long ); 127 #endif 128 #ifdef FLOAT128 129 float128 int64_to_float128( long long ); 130 #endif 131 #endif 132 133 /* 134 ------------------------------------------------------------------------------- 135 Software IEC/IEEE single-precision conversion routines. 136 ------------------------------------------------------------------------------- 137 */ 138 int float32_to_int32( float32 ); 139 int float32_to_int32_round_to_zero( float32 ); 140 #if defined(SOFTFLOAT_FOR_GCC) && defined(SOFTFLOAT_NEED_FIXUNS) 141 unsigned int float32_to_uint32_round_to_zero( float32 ); 142 #endif 143 #ifndef SOFTFLOAT_FOR_GCC /* __fix?fdi provided by libgcc2.c */ 144 long long float32_to_int64( float32 ); 145 long long float32_to_int64_round_to_zero( float32 ); 146 #endif 147 float64 float32_to_float64( float32 ); 148 #ifdef FLOATX80 149 floatx80 float32_to_floatx80( float32 ); 150 #endif 151 #ifdef FLOAT128 152 float128 float32_to_float128( float32 ); 153 #endif 154 155 /* 156 ------------------------------------------------------------------------------- 157 Software IEC/IEEE single-precision operations. 158 ------------------------------------------------------------------------------- 159 */ 160 float32 float32_round_to_int( float32 ); 161 float32 float32_add( float32, float32 ); 162 float32 float32_sub( float32, float32 ); 163 float32 float32_mul( float32, float32 ); 164 float32 float32_div( float32, float32 ); 165 float32 float32_rem( float32, float32 ); 166 float32 float32_sqrt( float32 ); 167 int float32_eq( float32, float32 ); 168 int float32_le( float32, float32 ); 169 int float32_lt( float32, float32 ); 170 int float32_eq_signaling( float32, float32 ); 171 int float32_le_quiet( float32, float32 ); 172 int float32_lt_quiet( float32, float32 ); 173 #ifndef SOFTFLOAT_FOR_GCC 174 int float32_is_signaling_nan( float32 ); 175 #endif 176 177 /* 178 ------------------------------------------------------------------------------- 179 Software IEC/IEEE double-precision conversion routines. 180 ------------------------------------------------------------------------------- 181 */ 182 int float64_to_int32( float64 ); 183 int float64_to_int32_round_to_zero( float64 ); 184 #if defined(SOFTFLOAT_FOR_GCC) && defined(SOFTFLOAT_NEED_FIXUNS) 185 unsigned int float64_to_uint32_round_to_zero( float64 ); 186 #endif 187 #ifndef SOFTFLOAT_FOR_GCC /* __fix?fdi provided by libgcc2.c */ 188 long long float64_to_int64( float64 ); 189 long long float64_to_int64_round_to_zero( float64 ); 190 #endif 191 float32 float64_to_float32( float64 ); 192 #ifdef FLOATX80 193 floatx80 float64_to_floatx80( float64 ); 194 #endif 195 #ifdef FLOAT128 196 float128 float64_to_float128( float64 ); 197 #endif 198 199 /* 200 ------------------------------------------------------------------------------- 201 Software IEC/IEEE double-precision operations. 202 ------------------------------------------------------------------------------- 203 */ 204 float64 float64_round_to_int( float64 ); 205 float64 float64_add( float64, float64 ); 206 float64 float64_sub( float64, float64 ); 207 float64 float64_mul( float64, float64 ); 208 float64 float64_div( float64, float64 ); 209 float64 float64_rem( float64, float64 ); 210 float64 float64_sqrt( float64 ); 211 int float64_eq( float64, float64 ); 212 int float64_le( float64, float64 ); 213 int float64_lt( float64, float64 ); 214 int float64_eq_signaling( float64, float64 ); 215 int float64_le_quiet( float64, float64 ); 216 int float64_lt_quiet( float64, float64 ); 217 #ifndef SOFTFLOAT_FOR_GCC 218 int float64_is_signaling_nan( float64 ); 219 #endif 220 221 #ifdef FLOATX80 222 223 /* 224 ------------------------------------------------------------------------------- 225 Software IEC/IEEE extended double-precision conversion routines. 226 ------------------------------------------------------------------------------- 227 */ 228 int floatx80_to_int32( floatx80 ); 229 int floatx80_to_int32_round_to_zero( floatx80 ); 230 long long floatx80_to_int64( floatx80 ); 231 long long floatx80_to_int64_round_to_zero( floatx80 ); 232 float32 floatx80_to_float32( floatx80 ); 233 float64 floatx80_to_float64( floatx80 ); 234 #ifdef FLOAT128 235 float128 floatx80_to_float128( floatx80 ); 236 #endif 237 238 /* 239 ------------------------------------------------------------------------------- 240 Software IEC/IEEE extended double-precision rounding precision. Valid 241 values are 32, 64, and 80. 242 ------------------------------------------------------------------------------- 243 */ 244 extern int floatx80_rounding_precision; 245 246 /* 247 ------------------------------------------------------------------------------- 248 Software IEC/IEEE extended double-precision operations. 249 ------------------------------------------------------------------------------- 250 */ 251 floatx80 floatx80_round_to_int( floatx80 ); 252 floatx80 floatx80_add( floatx80, floatx80 ); 253 floatx80 floatx80_sub( floatx80, floatx80 ); 254 floatx80 floatx80_mul( floatx80, floatx80 ); 255 floatx80 floatx80_div( floatx80, floatx80 ); 256 floatx80 floatx80_rem( floatx80, floatx80 ); 257 floatx80 floatx80_sqrt( floatx80 ); 258 int floatx80_eq( floatx80, floatx80 ); 259 int floatx80_le( floatx80, floatx80 ); 260 int floatx80_lt( floatx80, floatx80 ); 261 int floatx80_eq_signaling( floatx80, floatx80 ); 262 int floatx80_le_quiet( floatx80, floatx80 ); 263 int floatx80_lt_quiet( floatx80, floatx80 ); 264 int floatx80_is_signaling_nan( floatx80 ); 265 266 #endif 267 268 #ifdef FLOAT128 269 270 /* 271 ------------------------------------------------------------------------------- 272 Software IEC/IEEE quadruple-precision conversion routines. 273 ------------------------------------------------------------------------------- 274 */ 275 int float128_to_int32( float128 ); 276 int float128_to_int32_round_to_zero( float128 ); 277 long long float128_to_int64( float128 ); 278 long long float128_to_int64_round_to_zero( float128 ); 279 float32 float128_to_float32( float128 ); 280 float64 float128_to_float64( float128 ); 281 #ifdef FLOATX80 282 floatx80 float128_to_floatx80( float128 ); 283 #endif 284 285 /* 286 ------------------------------------------------------------------------------- 287 Software IEC/IEEE quadruple-precision operations. 288 ------------------------------------------------------------------------------- 289 */ 290 float128 float128_round_to_int( float128 ); 291 float128 float128_add( float128, float128 ); 292 float128 float128_sub( float128, float128 ); 293 float128 float128_mul( float128, float128 ); 294 float128 float128_div( float128, float128 ); 295 float128 float128_rem( float128, float128 ); 296 float128 float128_sqrt( float128 ); 297 int float128_eq( float128, float128 ); 298 int float128_le( float128, float128 ); 299 int float128_lt( float128, float128 ); 300 int float128_eq_signaling( float128, float128 ); 301 int float128_le_quiet( float128, float128 ); 302 int float128_lt_quiet( float128, float128 ); 303 int float128_is_signaling_nan( float128 ); 304 305 #endif 306 307
Indexes created Mon Sep 22 21:09:42 GMT 2025