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 #include <machine/ieeefp.h> 45 46 /* 47 ------------------------------------------------------------------------------- 48 Software IEC/IEEE floating-point types. 49 ------------------------------------------------------------------------------- 50 */ 51 typedef unsigned int float32; 52 typedef unsigned long long float64; 53 #ifdef FLOATX80 54 typedef struct { 55 unsigned short high; 56 unsigned long long low; 57 } floatx80; 58 #endif 59 #ifdef FLOAT128 60 typedef struct { 61 unsigned long long high, low; 62 } float128; 63 #endif 64 65 /* 66 ------------------------------------------------------------------------------- 67 Software IEC/IEEE floating-point underflow tininess-detection mode. 68 ------------------------------------------------------------------------------- 69 */ 70 extern int8 float_detect_tininess; 71 enum { 72 float_tininess_after_rounding = 0, 73 float_tininess_before_rounding = 1 74 }; 75 76 /* 77 ------------------------------------------------------------------------------- 78 Software IEC/IEEE floating-point rounding mode. 79 ------------------------------------------------------------------------------- 80 */ 81 extern int8 float_rounding_mode; 82 enum { 83 float_round_nearest_even = FP_RN, 84 float_round_to_zero = FP_RZ, 85 float_round_down = FP_RM, 86 float_round_up = FP_RP 87 }; 88 89 /* 90 ------------------------------------------------------------------------------- 91 Software IEC/IEEE floating-point exception flags. 92 ------------------------------------------------------------------------------- 93 */ 94 extern int8 float_exception_flags; 95 enum { 96 float_flag_inexact = FP_X_IMP, 97 float_flag_underflow = FP_X_UFL, 98 float_flag_overflow = FP_X_OFL, 99 float_flag_divbyzero = FP_X_DZ, 100 float_flag_invalid = FP_X_INV 101 }; 102 103 /* 104 ------------------------------------------------------------------------------- 105 Routine to raise any or all of the software IEC/IEEE floating-point 106 exception flags. 107 ------------------------------------------------------------------------------- 108 */ 109 void float_raise( int8 ); 110 111 /* 112 ------------------------------------------------------------------------------- 113 Software IEC/IEEE integer-to-floating-point conversion routines. 114 ------------------------------------------------------------------------------- 115 */ 116 float32 int32_to_float32( int ); 117 float64 int32_to_float64( int ); 118 #ifdef FLOATX80 119 floatx80 int32_to_floatx80( int ); 120 #endif 121 #ifdef FLOAT128 122 float128 int32_to_float128( int ); 123 #endif 124 float32 int64_to_float32( long long ); 125 float64 int64_to_float64( long long ); 126 #ifdef FLOATX80 127 floatx80 int64_to_floatx80( long long ); 128 #endif 129 #ifdef FLOAT128 130 float128 int64_to_float128( long long ); 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 unsigned int float32_to_uint32_round_to_zero( float32 ); 141 long long float32_to_int64( float32 ); 142 long long float32_to_int64_round_to_zero( float32 ); 143 float64 float32_to_float64( float32 ); 144 #ifdef FLOATX80 145 floatx80 float32_to_floatx80( float32 ); 146 #endif 147 #ifdef FLOAT128 148 float128 float32_to_float128( float32 ); 149 #endif 150 151 /* 152 ------------------------------------------------------------------------------- 153 Software IEC/IEEE single-precision operations. 154 ------------------------------------------------------------------------------- 155 */ 156 float32 float32_round_to_int( float32 ); 157 float32 float32_add( float32, float32 ); 158 float32 float32_sub( float32, float32 ); 159 float32 float32_mul( float32, float32 ); 160 float32 float32_div( float32, float32 ); 161 float32 float32_rem( float32, float32 ); 162 float32 float32_sqrt( float32 ); 163 flag float32_eq( float32, float32 ); 164 flag float32_le( float32, float32 ); 165 flag float32_lt( float32, float32 ); 166 flag float32_eq_signaling( float32, float32 ); 167 flag float32_le_quiet( float32, float32 ); 168 flag float32_lt_quiet( float32, float32 ); 169 flag float32_is_signaling_nan( float32 ); 170 171 /* 172 ------------------------------------------------------------------------------- 173 Software IEC/IEEE double-precision conversion routines. 174 ------------------------------------------------------------------------------- 175 */ 176 int float64_to_int32( float64 ); 177 int float64_to_int32_round_to_zero( float64 ); 178 unsigned int float64_to_uint32_round_to_zero( float64 ); 179 long long float64_to_int64( float64 ); 180 long long float64_to_int64_round_to_zero( float64 ); 181 float32 float64_to_float32( float64 ); 182 #ifdef FLOATX80 183 floatx80 float64_to_floatx80( float64 ); 184 #endif 185 #ifdef FLOAT128 186 float128 float64_to_float128( float64 ); 187 #endif 188 189 /* 190 ------------------------------------------------------------------------------- 191 Software IEC/IEEE double-precision operations. 192 ------------------------------------------------------------------------------- 193 */ 194 float64 float64_round_to_int( float64 ); 195 float64 float64_add( float64, float64 ); 196 float64 float64_sub( float64, float64 ); 197 float64 float64_mul( float64, float64 ); 198 float64 float64_div( float64, float64 ); 199 float64 float64_rem( float64, float64 ); 200 float64 float64_sqrt( float64 ); 201 flag float64_eq( float64, float64 ); 202 flag float64_le( float64, float64 ); 203 flag float64_lt( float64, float64 ); 204 flag float64_eq_signaling( float64, float64 ); 205 flag float64_le_quiet( float64, float64 ); 206 flag float64_lt_quiet( float64, float64 ); 207 flag float64_is_signaling_nan( float64 ); 208 209 #ifdef FLOATX80 210 211 /* 212 ------------------------------------------------------------------------------- 213 Software IEC/IEEE extended double-precision conversion routines. 214 ------------------------------------------------------------------------------- 215 */ 216 int floatx80_to_int32( floatx80 ); 217 int floatx80_to_int32_round_to_zero( floatx80 ); 218 long long floatx80_to_int64( floatx80 ); 219 long long floatx80_to_int64_round_to_zero( floatx80 ); 220 float32 floatx80_to_float32( floatx80 ); 221 float64 floatx80_to_float64( floatx80 ); 222 #ifdef FLOAT128 223 float128 floatx80_to_float128( floatx80 ); 224 #endif 225 226 /* 227 ------------------------------------------------------------------------------- 228 Software IEC/IEEE extended double-precision rounding precision. Valid 229 values are 32, 64, and 80. 230 ------------------------------------------------------------------------------- 231 */ 232 extern int floatx80_rounding_precision; 233 234 /* 235 ------------------------------------------------------------------------------- 236 Software IEC/IEEE extended double-precision operations. 237 ------------------------------------------------------------------------------- 238 */ 239 floatx80 floatx80_round_to_int( floatx80 ); 240 floatx80 floatx80_add( floatx80, floatx80 ); 241 floatx80 floatx80_sub( floatx80, floatx80 ); 242 floatx80 floatx80_mul( floatx80, floatx80 ); 243 floatx80 floatx80_div( floatx80, floatx80 ); 244 floatx80 floatx80_rem( floatx80, floatx80 ); 245 floatx80 floatx80_sqrt( floatx80 ); 246 flag floatx80_eq( floatx80, floatx80 ); 247 flag floatx80_le( floatx80, floatx80 ); 248 flag floatx80_lt( floatx80, floatx80 ); 249 flag floatx80_eq_signaling( floatx80, floatx80 ); 250 flag floatx80_le_quiet( floatx80, floatx80 ); 251 flag floatx80_lt_quiet( floatx80, floatx80 ); 252 flag floatx80_is_signaling_nan( floatx80 ); 253 254 #endif 255 256 #ifdef FLOAT128 257 258 /* 259 ------------------------------------------------------------------------------- 260 Software IEC/IEEE quadruple-precision conversion routines. 261 ------------------------------------------------------------------------------- 262 */ 263 int float128_to_int32( float128 ); 264 int float128_to_int32_round_to_zero( float128 ); 265 long long float128_to_int64( float128 ); 266 long long float128_to_int64_round_to_zero( float128 ); 267 float32 float128_to_float32( float128 ); 268 float64 float128_to_float64( float128 ); 269 #ifdef FLOATX80 270 floatx80 float128_to_floatx80( float128 ); 271 #endif 272 273 /* 274 ------------------------------------------------------------------------------- 275 Software IEC/IEEE quadruple-precision operations. 276 ------------------------------------------------------------------------------- 277 */ 278 float128 float128_round_to_int( float128 ); 279 float128 float128_add( float128, float128 ); 280 float128 float128_sub( float128, float128 ); 281 float128 float128_mul( float128, float128 ); 282 float128 float128_div( float128, float128 ); 283 float128 float128_rem( float128, float128 ); 284 float128 float128_sqrt( float128 ); 285 flag float128_eq( float128, float128 ); 286 flag float128_le( float128, float128 ); 287 flag float128_lt( float128, float128 ); 288 flag float128_eq_signaling( float128, float128 ); 289 flag float128_le_quiet( float128, float128 ); 290 flag float128_lt_quiet( float128, float128 ); 291 flag float128_is_signaling_nan( float128 ); 292 293 #endif 294