1 /* Software floating-point emulation. 2 Definitions for IEEE Extended Precision. 3 Copyright (C) 1999-2016 Free Software Foundation, Inc. 4 This file is part of the GNU C Library. 5 Contributed by Jakub Jelinek (jj@ultra.linux.cz). 6 7 The GNU C Library is free software; you can redistribute it and/or 8 modify it under the terms of the GNU Lesser General Public 9 License as published by the Free Software Foundation; either 10 version 2.1 of the License, or (at your option) any later version. 11 12 In addition to the permissions in the GNU Lesser General Public 13 License, the Free Software Foundation gives you unlimited 14 permission to link the compiled version of this file into 15 combinations with other programs, and to distribute those 16 combinations without any restriction coming from the use of this 17 file. (The Lesser General Public License restrictions do apply in 18 other respects; for example, they cover modification of the file, 19 and distribution when not linked into a combine executable.) 20 21 The GNU C Library is distributed in the hope that it will be useful, 22 but WITHOUT ANY WARRANTY; without even the implied warranty of 23 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU 24 Lesser General Public License for more details. 25 26 You should have received a copy of the GNU Lesser General Public 27 License along with the GNU C Library; if not, see 28 <http://www.gnu.org/licenses/>. */ 29 30 #ifndef SOFT_FP_EXTENDED_H 31 #define SOFT_FP_EXTENDED_H 1 32 33 #if _FP_W_TYPE_SIZE < 32 34 # error "Here's a nickel, kid. Go buy yourself a real computer." 35 #endif 36 37 #if _FP_W_TYPE_SIZE < 64 38 # define _FP_FRACTBITS_E (4*_FP_W_TYPE_SIZE) 39 # define _FP_FRACTBITS_DW_E (8*_FP_W_TYPE_SIZE) 40 #else 41 # define _FP_FRACTBITS_E (2*_FP_W_TYPE_SIZE) 42 # define _FP_FRACTBITS_DW_E (4*_FP_W_TYPE_SIZE) 43 #endif 44 45 #define _FP_FRACBITS_E 64 46 #define _FP_FRACXBITS_E (_FP_FRACTBITS_E - _FP_FRACBITS_E) 47 #define _FP_WFRACBITS_E (_FP_WORKBITS + _FP_FRACBITS_E) 48 #define _FP_WFRACXBITS_E (_FP_FRACTBITS_E - _FP_WFRACBITS_E) 49 #define _FP_EXPBITS_E 15 50 #define _FP_EXPBIAS_E 16383 51 #define _FP_EXPMAX_E 32767 52 53 #define _FP_QNANBIT_E \ 54 ((_FP_W_TYPE) 1 << (_FP_FRACBITS_E-2) % _FP_W_TYPE_SIZE) 55 #define _FP_QNANBIT_SH_E \ 56 ((_FP_W_TYPE) 1 << (_FP_FRACBITS_E-2+_FP_WORKBITS) % _FP_W_TYPE_SIZE) 57 #define _FP_IMPLBIT_E \ 58 ((_FP_W_TYPE) 1 << (_FP_FRACBITS_E-1) % _FP_W_TYPE_SIZE) 59 #define _FP_IMPLBIT_SH_E \ 60 ((_FP_W_TYPE) 1 << (_FP_FRACBITS_E-1+_FP_WORKBITS) % _FP_W_TYPE_SIZE) 61 #define _FP_OVERFLOW_E \ 62 ((_FP_W_TYPE) 1 << (_FP_WFRACBITS_E % _FP_W_TYPE_SIZE)) 63 64 #define _FP_WFRACBITS_DW_E (2 * _FP_WFRACBITS_E) 65 #define _FP_WFRACXBITS_DW_E (_FP_FRACTBITS_DW_E - _FP_WFRACBITS_DW_E) 66 #define _FP_HIGHBIT_DW_E \ 67 ((_FP_W_TYPE) 1 << (_FP_WFRACBITS_DW_E - 1) % _FP_W_TYPE_SIZE) 68 69 typedef float XFtype __attribute__ ((mode (XF))); 70 71 #if _FP_W_TYPE_SIZE < 64 72 73 union _FP_UNION_E 74 { 75 XFtype flt; 76 struct _FP_STRUCT_LAYOUT 77 { 78 # if __BYTE_ORDER == __BIG_ENDIAN 79 unsigned long pad1 : _FP_W_TYPE_SIZE; 80 unsigned long pad2 : (_FP_W_TYPE_SIZE - 1 - _FP_EXPBITS_E); 81 unsigned long sign : 1; 82 unsigned long exp : _FP_EXPBITS_E; 83 unsigned long frac1 : _FP_W_TYPE_SIZE; 84 unsigned long frac0 : _FP_W_TYPE_SIZE; 85 # else 86 unsigned long frac0 : _FP_W_TYPE_SIZE; 87 unsigned long frac1 : _FP_W_TYPE_SIZE; 88 unsigned exp : _FP_EXPBITS_E; 89 unsigned sign : 1; 90 # endif /* not bigendian */ 91 } bits __attribute__ ((packed)); 92 }; 93 94 95 # define FP_DECL_E(X) _FP_DECL (4, X) 96 97 # define FP_UNPACK_RAW_E(X, val) \ 98 do \ 99 { \ 100 union _FP_UNION_E FP_UNPACK_RAW_E_flo; \ 101 FP_UNPACK_RAW_E_flo.flt = (val); \ 102 \ 103 X##_f[2] = 0; \ 104 X##_f[3] = 0; \ 105 X##_f[0] = FP_UNPACK_RAW_E_flo.bits.frac0; \ 106 X##_f[1] = FP_UNPACK_RAW_E_flo.bits.frac1; \ 107 X##_f[1] &= ~_FP_IMPLBIT_E; \ 108 X##_e = FP_UNPACK_RAW_E_flo.bits.exp; \ 109 X##_s = FP_UNPACK_RAW_E_flo.bits.sign; \ 110 } \ 111 while (0) 112 113 # define FP_UNPACK_RAW_EP(X, val) \ 114 do \ 115 { \ 116 union _FP_UNION_E *FP_UNPACK_RAW_EP_flo \ 117 = (union _FP_UNION_E *) (val); \ 118 \ 119 X##_f[2] = 0; \ 120 X##_f[3] = 0; \ 121 X##_f[0] = FP_UNPACK_RAW_EP_flo->bits.frac0; \ 122 X##_f[1] = FP_UNPACK_RAW_EP_flo->bits.frac1; \ 123 X##_f[1] &= ~_FP_IMPLBIT_E; \ 124 X##_e = FP_UNPACK_RAW_EP_flo->bits.exp; \ 125 X##_s = FP_UNPACK_RAW_EP_flo->bits.sign; \ 126 } \ 127 while (0) 128 129 # define FP_PACK_RAW_E(val, X) \ 130 do \ 131 { \ 132 union _FP_UNION_E FP_PACK_RAW_E_flo; \ 133 \ 134 if (X##_e) \ 135 X##_f[1] |= _FP_IMPLBIT_E; \ 136 else \ 137 X##_f[1] &= ~(_FP_IMPLBIT_E); \ 138 FP_PACK_RAW_E_flo.bits.frac0 = X##_f[0]; \ 139 FP_PACK_RAW_E_flo.bits.frac1 = X##_f[1]; \ 140 FP_PACK_RAW_E_flo.bits.exp = X##_e; \ 141 FP_PACK_RAW_E_flo.bits.sign = X##_s; \ 142 \ 143 (val) = FP_PACK_RAW_E_flo.flt; \ 144 } \ 145 while (0) 146 147 # define FP_PACK_RAW_EP(val, X) \ 148 do \ 149 { \ 150 if (!FP_INHIBIT_RESULTS) \ 151 { \ 152 union _FP_UNION_E *FP_PACK_RAW_EP_flo \ 153 = (union _FP_UNION_E *) (val); \ 154 \ 155 if (X##_e) \ 156 X##_f[1] |= _FP_IMPLBIT_E; \ 157 else \ 158 X##_f[1] &= ~(_FP_IMPLBIT_E); \ 159 FP_PACK_RAW_EP_flo->bits.frac0 = X##_f[0]; \ 160 FP_PACK_RAW_EP_flo->bits.frac1 = X##_f[1]; \ 161 FP_PACK_RAW_EP_flo->bits.exp = X##_e; \ 162 FP_PACK_RAW_EP_flo->bits.sign = X##_s; \ 163 } \ 164 } \ 165 while (0) 166 167 # define FP_UNPACK_E(X, val) \ 168 do \ 169 { \ 170 FP_UNPACK_RAW_E (X, (val)); \ 171 _FP_UNPACK_CANONICAL (E, 4, X); \ 172 } \ 173 while (0) 174 175 # define FP_UNPACK_EP(X, val) \ 176 do \ 177 { \ 178 FP_UNPACK_RAW_EP (X, (val)); \ 179 _FP_UNPACK_CANONICAL (E, 4, X); \ 180 } \ 181 while (0) 182 183 # define FP_UNPACK_SEMIRAW_E(X, val) \ 184 do \ 185 { \ 186 FP_UNPACK_RAW_E (X, (val)); \ 187 _FP_UNPACK_SEMIRAW (E, 4, X); \ 188 } \ 189 while (0) 190 191 # define FP_UNPACK_SEMIRAW_EP(X, val) \ 192 do \ 193 { \ 194 FP_UNPACK_RAW_EP (X, (val)); \ 195 _FP_UNPACK_SEMIRAW (E, 4, X); \ 196 } \ 197 while (0) 198 199 # define FP_PACK_E(val, X) \ 200 do \ 201 { \ 202 _FP_PACK_CANONICAL (E, 4, X); \ 203 FP_PACK_RAW_E ((val), X); \ 204 } \ 205 while (0) 206 207 # define FP_PACK_EP(val, X) \ 208 do \ 209 { \ 210 _FP_PACK_CANONICAL (E, 4, X); \ 211 FP_PACK_RAW_EP ((val), X); \ 212 } \ 213 while (0) 214 215 # define FP_PACK_SEMIRAW_E(val, X) \ 216 do \ 217 { \ 218 _FP_PACK_SEMIRAW (E, 4, X); \ 219 FP_PACK_RAW_E ((val), X); \ 220 } \ 221 while (0) 222 223 # define FP_PACK_SEMIRAW_EP(val, X) \ 224 do \ 225 { \ 226 _FP_PACK_SEMIRAW (E, 4, X); \ 227 FP_PACK_RAW_EP ((val), X); \ 228 } \ 229 while (0) 230 231 # define FP_ISSIGNAN_E(X) _FP_ISSIGNAN (E, 4, X) 232 # define FP_NEG_E(R, X) _FP_NEG (E, 4, R, X) 233 # define FP_ADD_E(R, X, Y) _FP_ADD (E, 4, R, X, Y) 234 # define FP_SUB_E(R, X, Y) _FP_SUB (E, 4, R, X, Y) 235 # define FP_MUL_E(R, X, Y) _FP_MUL (E, 4, R, X, Y) 236 # define FP_DIV_E(R, X, Y) _FP_DIV (E, 4, R, X, Y) 237 # define FP_SQRT_E(R, X) _FP_SQRT (E, 4, R, X) 238 # define FP_FMA_E(R, X, Y, Z) _FP_FMA (E, 4, 8, R, X, Y, Z) 239 240 /* Square root algorithms: 241 We have just one right now, maybe Newton approximation 242 should be added for those machines where division is fast. 243 This has special _E version because standard _4 square 244 root would not work (it has to start normally with the 245 second word and not the first), but as we have to do it 246 anyway, we optimize it by doing most of the calculations 247 in two UWtype registers instead of four. */ 248 249 # define _FP_SQRT_MEAT_E(R, S, T, X, q) \ 250 do \ 251 { \ 252 (q) = (_FP_W_TYPE) 1 << (_FP_W_TYPE_SIZE - 1); \ 253 _FP_FRAC_SRL_4 (X, (_FP_WORKBITS)); \ 254 while (q) \ 255 { \ 256 T##_f[1] = S##_f[1] + (q); \ 257 if (T##_f[1] <= X##_f[1]) \ 258 { \ 259 S##_f[1] = T##_f[1] + (q); \ 260 X##_f[1] -= T##_f[1]; \ 261 R##_f[1] += (q); \ 262 } \ 263 _FP_FRAC_SLL_2 (X, 1); \ 264 (q) >>= 1; \ 265 } \ 266 (q) = (_FP_W_TYPE) 1 << (_FP_W_TYPE_SIZE - 1); \ 267 while (q) \ 268 { \ 269 T##_f[0] = S##_f[0] + (q); \ 270 T##_f[1] = S##_f[1]; \ 271 if (T##_f[1] < X##_f[1] \ 272 || (T##_f[1] == X##_f[1] \ 273 && T##_f[0] <= X##_f[0])) \ 274 { \ 275 S##_f[0] = T##_f[0] + (q); \ 276 S##_f[1] += (T##_f[0] > S##_f[0]); \ 277 _FP_FRAC_DEC_2 (X, T); \ 278 R##_f[0] += (q); \ 279 } \ 280 _FP_FRAC_SLL_2 (X, 1); \ 281 (q) >>= 1; \ 282 } \ 283 _FP_FRAC_SLL_4 (R, (_FP_WORKBITS)); \ 284 if (X##_f[0] | X##_f[1]) \ 285 { \ 286 if (S##_f[1] < X##_f[1] \ 287 || (S##_f[1] == X##_f[1] \ 288 && S##_f[0] < X##_f[0])) \ 289 R##_f[0] |= _FP_WORK_ROUND; \ 290 R##_f[0] |= _FP_WORK_STICKY; \ 291 } \ 292 } \ 293 while (0) 294 295 # define FP_CMP_E(r, X, Y, un, ex) _FP_CMP (E, 4, (r), X, Y, (un), (ex)) 296 # define FP_CMP_EQ_E(r, X, Y, ex) _FP_CMP_EQ (E, 4, (r), X, Y, (ex)) 297 # define FP_CMP_UNORD_E(r, X, Y, ex) _FP_CMP_UNORD (E, 4, (r), X, Y, (ex)) 298 299 # define FP_TO_INT_E(r, X, rsz, rsg) _FP_TO_INT (E, 4, (r), X, (rsz), (rsg)) 300 # define FP_TO_INT_ROUND_E(r, X, rsz, rsg) \ 301 _FP_TO_INT_ROUND (E, 4, (r), X, (rsz), (rsg)) 302 # define FP_FROM_INT_E(X, r, rs, rt) _FP_FROM_INT (E, 4, X, (r), (rs), rt) 303 304 # define _FP_FRAC_HIGH_E(X) (X##_f[2]) 305 # define _FP_FRAC_HIGH_RAW_E(X) (X##_f[1]) 306 307 # define _FP_FRAC_HIGH_DW_E(X) (X##_f[4]) 308 309 #else /* not _FP_W_TYPE_SIZE < 64 */ 310 union _FP_UNION_E 311 { 312 XFtype flt; 313 struct _FP_STRUCT_LAYOUT 314 { 315 # if __BYTE_ORDER == __BIG_ENDIAN 316 _FP_W_TYPE pad : (_FP_W_TYPE_SIZE - 1 - _FP_EXPBITS_E); 317 unsigned sign : 1; 318 unsigned exp : _FP_EXPBITS_E; 319 _FP_W_TYPE frac : _FP_W_TYPE_SIZE; 320 # else 321 _FP_W_TYPE frac : _FP_W_TYPE_SIZE; 322 unsigned exp : _FP_EXPBITS_E; 323 unsigned sign : 1; 324 # endif 325 } bits; 326 }; 327 328 # define FP_DECL_E(X) _FP_DECL (2, X) 329 330 # define FP_UNPACK_RAW_E(X, val) \ 331 do \ 332 { \ 333 union _FP_UNION_E FP_UNPACK_RAW_E_flo; \ 334 FP_UNPACK_RAW_E_flo.flt = (val); \ 335 \ 336 X##_f0 = FP_UNPACK_RAW_E_flo.bits.frac; \ 337 X##_f0 &= ~_FP_IMPLBIT_E; \ 338 X##_f1 = 0; \ 339 X##_e = FP_UNPACK_RAW_E_flo.bits.exp; \ 340 X##_s = FP_UNPACK_RAW_E_flo.bits.sign; \ 341 } \ 342 while (0) 343 344 # define FP_UNPACK_RAW_EP(X, val) \ 345 do \ 346 { \ 347 union _FP_UNION_E *FP_UNPACK_RAW_EP_flo \ 348 = (union _FP_UNION_E *) (val); \ 349 \ 350 X##_f0 = FP_UNPACK_RAW_EP_flo->bits.frac; \ 351 X##_f0 &= ~_FP_IMPLBIT_E; \ 352 X##_f1 = 0; \ 353 X##_e = FP_UNPACK_RAW_EP_flo->bits.exp; \ 354 X##_s = FP_UNPACK_RAW_EP_flo->bits.sign; \ 355 } \ 356 while (0) 357 358 # define FP_PACK_RAW_E(val, X) \ 359 do \ 360 { \ 361 union _FP_UNION_E FP_PACK_RAW_E_flo; \ 362 \ 363 if (X##_e) \ 364 X##_f0 |= _FP_IMPLBIT_E; \ 365 else \ 366 X##_f0 &= ~(_FP_IMPLBIT_E); \ 367 FP_PACK_RAW_E_flo.bits.frac = X##_f0; \ 368 FP_PACK_RAW_E_flo.bits.exp = X##_e; \ 369 FP_PACK_RAW_E_flo.bits.sign = X##_s; \ 370 \ 371 (val) = FP_PACK_RAW_E_flo.flt; \ 372 } \ 373 while (0) 374 375 # define FP_PACK_RAW_EP(fs, val, X) \ 376 do \ 377 { \ 378 if (!FP_INHIBIT_RESULTS) \ 379 { \ 380 union _FP_UNION_E *FP_PACK_RAW_EP_flo \ 381 = (union _FP_UNION_E *) (val); \ 382 \ 383 if (X##_e) \ 384 X##_f0 |= _FP_IMPLBIT_E; \ 385 else \ 386 X##_f0 &= ~(_FP_IMPLBIT_E); \ 387 FP_PACK_RAW_EP_flo->bits.frac = X##_f0; \ 388 FP_PACK_RAW_EP_flo->bits.exp = X##_e; \ 389 FP_PACK_RAW_EP_flo->bits.sign = X##_s; \ 390 } \ 391 } \ 392 while (0) 393 394 395 # define FP_UNPACK_E(X, val) \ 396 do \ 397 { \ 398 FP_UNPACK_RAW_E (X, (val)); \ 399 _FP_UNPACK_CANONICAL (E, 2, X); \ 400 } \ 401 while (0) 402 403 # define FP_UNPACK_EP(X, val) \ 404 do \ 405 { \ 406 FP_UNPACK_RAW_EP (X, (val)); \ 407 _FP_UNPACK_CANONICAL (E, 2, X); \ 408 } \ 409 while (0) 410 411 # define FP_UNPACK_SEMIRAW_E(X, val) \ 412 do \ 413 { \ 414 FP_UNPACK_RAW_E (X, (val)); \ 415 _FP_UNPACK_SEMIRAW (E, 2, X); \ 416 } \ 417 while (0) 418 419 # define FP_UNPACK_SEMIRAW_EP(X, val) \ 420 do \ 421 { \ 422 FP_UNPACK_RAW_EP (X, (val)); \ 423 _FP_UNPACK_SEMIRAW (E, 2, X); \ 424 } \ 425 while (0) 426 427 # define FP_PACK_E(val, X) \ 428 do \ 429 { \ 430 _FP_PACK_CANONICAL (E, 2, X); \ 431 FP_PACK_RAW_E ((val), X); \ 432 } \ 433 while (0) 434 435 # define FP_PACK_EP(val, X) \ 436 do \ 437 { \ 438 _FP_PACK_CANONICAL (E, 2, X); \ 439 FP_PACK_RAW_EP ((val), X); \ 440 } \ 441 while (0) 442 443 # define FP_PACK_SEMIRAW_E(val, X) \ 444 do \ 445 { \ 446 _FP_PACK_SEMIRAW (E, 2, X); \ 447 FP_PACK_RAW_E ((val), X); \ 448 } \ 449 while (0) 450 451 # define FP_PACK_SEMIRAW_EP(val, X) \ 452 do \ 453 { \ 454 _FP_PACK_SEMIRAW (E, 2, X); \ 455 FP_PACK_RAW_EP ((val), X); \ 456 } \ 457 while (0) 458 459 # define FP_ISSIGNAN_E(X) _FP_ISSIGNAN (E, 2, X) 460 # define FP_NEG_E(R, X) _FP_NEG (E, 2, R, X) 461 # define FP_ADD_E(R, X, Y) _FP_ADD (E, 2, R, X, Y) 462 # define FP_SUB_E(R, X, Y) _FP_SUB (E, 2, R, X, Y) 463 # define FP_MUL_E(R, X, Y) _FP_MUL (E, 2, R, X, Y) 464 # define FP_DIV_E(R, X, Y) _FP_DIV (E, 2, R, X, Y) 465 # define FP_SQRT_E(R, X) _FP_SQRT (E, 2, R, X) 466 # define FP_FMA_E(R, X, Y, Z) _FP_FMA (E, 2, 4, R, X, Y, Z) 467 468 /* Square root algorithms: 469 We have just one right now, maybe Newton approximation 470 should be added for those machines where division is fast. 471 We optimize it by doing most of the calculations 472 in one UWtype registers instead of two, although we don't 473 have to. */ 474 # define _FP_SQRT_MEAT_E(R, S, T, X, q) \ 475 do \ 476 { \ 477 (q) = (_FP_W_TYPE) 1 << (_FP_W_TYPE_SIZE - 1); \ 478 _FP_FRAC_SRL_2 (X, (_FP_WORKBITS)); \ 479 while (q) \ 480 { \ 481 T##_f0 = S##_f0 + (q); \ 482 if (T##_f0 <= X##_f0) \ 483 { \ 484 S##_f0 = T##_f0 + (q); \ 485 X##_f0 -= T##_f0; \ 486 R##_f0 += (q); \ 487 } \ 488 _FP_FRAC_SLL_1 (X, 1); \ 489 (q) >>= 1; \ 490 } \ 491 _FP_FRAC_SLL_2 (R, (_FP_WORKBITS)); \ 492 if (X##_f0) \ 493 { \ 494 if (S##_f0 < X##_f0) \ 495 R##_f0 |= _FP_WORK_ROUND; \ 496 R##_f0 |= _FP_WORK_STICKY; \ 497 } \ 498 } \ 499 while (0) 500 501 # define FP_CMP_E(r, X, Y, un, ex) _FP_CMP (E, 2, (r), X, Y, (un), (ex)) 502 # define FP_CMP_EQ_E(r, X, Y, ex) _FP_CMP_EQ (E, 2, (r), X, Y, (ex)) 503 # define FP_CMP_UNORD_E(r, X, Y, ex) _FP_CMP_UNORD (E, 2, (r), X, Y, (ex)) 504 505 # define FP_TO_INT_E(r, X, rsz, rsg) _FP_TO_INT (E, 2, (r), X, (rsz), (rsg)) 506 # define FP_TO_INT_ROUND_E(r, X, rsz, rsg) \ 507 _FP_TO_INT_ROUND (E, 2, (r), X, (rsz), (rsg)) 508 # define FP_FROM_INT_E(X, r, rs, rt) _FP_FROM_INT (E, 2, X, (r), (rs), rt) 509 510 # define _FP_FRAC_HIGH_E(X) (X##_f1) 511 # define _FP_FRAC_HIGH_RAW_E(X) (X##_f0) 512 513 # define _FP_FRAC_HIGH_DW_E(X) (X##_f[2]) 514 515 #endif /* not _FP_W_TYPE_SIZE < 64 */ 516 517 #endif /* !SOFT_FP_EXTENDED_H */ 518