1 /* ===-------- Intrin.h ---------------------------------------------------=== 2 * 3 * Permission is hereby granted, free of charge, to any person obtaining a copy 4 * of this software and associated documentation files (the "Software"), to deal 5 * in the Software without restriction, including without limitation the rights 6 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell 7 * copies of the Software, and to permit persons to whom the Software is 8 * furnished to do so, subject to the following conditions: 9 * 10 * The above copyright notice and this permission notice shall be included in 11 * all copies or substantial portions of the Software. 12 * 13 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR 14 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, 15 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE 16 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER 17 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, 18 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN 19 * THE SOFTWARE. 20 * 21 *===-----------------------------------------------------------------------=== 22 */ 23 24 /* Only include this if we're compiling for the windows platform. */ 25 #ifndef _MSC_VER 26 #include_next <Intrin.h> 27 #else 28 29 #ifndef __INTRIN_H 30 #define __INTRIN_H 31 32 /* First include the standard intrinsics. */ 33 #if defined(__i386__) || defined(__x86_64__) 34 #include <x86intrin.h> 35 #endif 36 37 /* For the definition of jmp_buf. */ 38 #if __STDC_HOSTED__ 39 #include <setjmp.h> 40 #endif 41 42 /* Define the default attributes for the functions in this file. */ 43 #define __DEFAULT_FN_ATTRS __attribute__((__always_inline__, __nodebug__)) 44 45 #ifdef __cplusplus 46 extern "C" { 47 #endif 48 49 #if defined(__MMX__) 50 /* And the random ones that aren't in those files. */ 51 __m64 _m_from_float(float); 52 float _m_to_float(__m64); 53 #endif 54 55 /* Other assorted instruction intrinsics. */ 56 void __addfsbyte(unsigned long, unsigned char); 57 void __addfsdword(unsigned long, unsigned long); 58 void __addfsword(unsigned long, unsigned short); 59 void __code_seg(const char *); 60 static __inline__ 61 void __cpuid(int[4], int); 62 static __inline__ 63 void __cpuidex(int[4], int, int); 64 void __debugbreak(void); 65 __int64 __emul(int, int); 66 unsigned __int64 __emulu(unsigned int, unsigned int); 67 void __cdecl __fastfail(unsigned int); 68 unsigned int __getcallerseflags(void); 69 static __inline__ 70 void __halt(void); 71 unsigned char __inbyte(unsigned short); 72 void __inbytestring(unsigned short, unsigned char *, unsigned long); 73 void __incfsbyte(unsigned long); 74 void __incfsdword(unsigned long); 75 void __incfsword(unsigned long); 76 unsigned long __indword(unsigned short); 77 void __indwordstring(unsigned short, unsigned long *, unsigned long); 78 void __int2c(void); 79 void __invlpg(void *); 80 unsigned short __inword(unsigned short); 81 void __inwordstring(unsigned short, unsigned short *, unsigned long); 82 void __lidt(void *); 83 unsigned __int64 __ll_lshift(unsigned __int64, int); 84 __int64 __ll_rshift(__int64, int); 85 void __llwpcb(void *); 86 unsigned char __lwpins32(unsigned int, unsigned int, unsigned int); 87 void __lwpval32(unsigned int, unsigned int, unsigned int); 88 unsigned int __lzcnt(unsigned int); 89 unsigned short __lzcnt16(unsigned short); 90 static __inline__ 91 void __movsb(unsigned char *, unsigned char const *, size_t); 92 static __inline__ 93 void __movsd(unsigned long *, unsigned long const *, size_t); 94 static __inline__ 95 void __movsw(unsigned short *, unsigned short const *, size_t); 96 void __nop(void); 97 void __nvreg_restore_fence(void); 98 void __nvreg_save_fence(void); 99 void __outbyte(unsigned short, unsigned char); 100 void __outbytestring(unsigned short, unsigned char *, unsigned long); 101 void __outdword(unsigned short, unsigned long); 102 void __outdwordstring(unsigned short, unsigned long *, unsigned long); 103 void __outword(unsigned short, unsigned short); 104 void __outwordstring(unsigned short, unsigned short *, unsigned long); 105 static __inline__ 106 unsigned int __popcnt(unsigned int); 107 static __inline__ 108 unsigned short __popcnt16(unsigned short); 109 unsigned long __readcr0(void); 110 unsigned long __readcr2(void); 111 static __inline__ 112 unsigned long __readcr3(void); 113 unsigned long __readcr4(void); 114 unsigned long __readcr8(void); 115 unsigned int __readdr(unsigned int); 116 #ifdef __i386__ 117 static __inline__ 118 unsigned char __readfsbyte(unsigned long); 119 static __inline__ 120 unsigned long __readfsdword(unsigned long); 121 static __inline__ 122 unsigned __int64 __readfsqword(unsigned long); 123 static __inline__ 124 unsigned short __readfsword(unsigned long); 125 #endif 126 static __inline__ 127 unsigned __int64 __readmsr(unsigned long); 128 unsigned __int64 __readpmc(unsigned long); 129 unsigned long __segmentlimit(unsigned long); 130 void __sidt(void *); 131 void *__slwpcb(void); 132 static __inline__ 133 void __stosb(unsigned char *, unsigned char, size_t); 134 static __inline__ 135 void __stosd(unsigned long *, unsigned long, size_t); 136 static __inline__ 137 void __stosw(unsigned short *, unsigned short, size_t); 138 void __svm_clgi(void); 139 void __svm_invlpga(void *, int); 140 void __svm_skinit(int); 141 void __svm_stgi(void); 142 void __svm_vmload(size_t); 143 void __svm_vmrun(size_t); 144 void __svm_vmsave(size_t); 145 void __ud2(void); 146 unsigned __int64 __ull_rshift(unsigned __int64, int); 147 void __vmx_off(void); 148 void __vmx_vmptrst(unsigned __int64 *); 149 void __wbinvd(void); 150 void __writecr0(unsigned int); 151 static __inline__ 152 void __writecr3(unsigned int); 153 void __writecr4(unsigned int); 154 void __writecr8(unsigned int); 155 void __writedr(unsigned int, unsigned int); 156 void __writefsbyte(unsigned long, unsigned char); 157 void __writefsdword(unsigned long, unsigned long); 158 void __writefsqword(unsigned long, unsigned __int64); 159 void __writefsword(unsigned long, unsigned short); 160 void __writemsr(unsigned long, unsigned __int64); 161 static __inline__ 162 void *_AddressOfReturnAddress(void); 163 static __inline__ 164 unsigned char _BitScanForward(unsigned long *_Index, unsigned long _Mask); 165 static __inline__ 166 unsigned char _BitScanReverse(unsigned long *_Index, unsigned long _Mask); 167 static __inline__ 168 unsigned char _bittest(long const *, long); 169 static __inline__ 170 unsigned char _bittestandcomplement(long *, long); 171 static __inline__ 172 unsigned char _bittestandreset(long *, long); 173 static __inline__ 174 unsigned char _bittestandset(long *, long); 175 unsigned __int64 __cdecl _byteswap_uint64(unsigned __int64); 176 unsigned long __cdecl _byteswap_ulong(unsigned long); 177 unsigned short __cdecl _byteswap_ushort(unsigned short); 178 void __cdecl _disable(void); 179 void __cdecl _enable(void); 180 long _InterlockedAddLargeStatistic(__int64 volatile *_Addend, long _Value); 181 static __inline__ 182 long _InterlockedAnd(long volatile *_Value, long _Mask); 183 static __inline__ 184 short _InterlockedAnd16(short volatile *_Value, short _Mask); 185 static __inline__ 186 char _InterlockedAnd8(char volatile *_Value, char _Mask); 187 unsigned char _interlockedbittestandreset(long volatile *, long); 188 static __inline__ 189 unsigned char _interlockedbittestandset(long volatile *, long); 190 static __inline__ 191 long __cdecl _InterlockedCompareExchange(long volatile *_Destination, 192 long _Exchange, long _Comparand); 193 long _InterlockedCompareExchange_HLEAcquire(long volatile *, long, long); 194 long _InterlockedCompareExchange_HLERelease(long volatile *, long, long); 195 static __inline__ 196 short _InterlockedCompareExchange16(short volatile *_Destination, 197 short _Exchange, short _Comparand); 198 static __inline__ 199 __int64 _InterlockedCompareExchange64(__int64 volatile *_Destination, 200 __int64 _Exchange, __int64 _Comparand); 201 __int64 _InterlockedcompareExchange64_HLEAcquire(__int64 volatile *, __int64, 202 __int64); 203 __int64 _InterlockedCompareExchange64_HLERelease(__int64 volatile *, __int64, 204 __int64); 205 static __inline__ 206 char _InterlockedCompareExchange8(char volatile *_Destination, char _Exchange, 207 char _Comparand); 208 void *_InterlockedCompareExchangePointer_HLEAcquire(void *volatile *, void *, 209 void *); 210 void *_InterlockedCompareExchangePointer_HLERelease(void *volatile *, void *, 211 void *); 212 static __inline__ 213 long __cdecl _InterlockedDecrement(long volatile *_Addend); 214 static __inline__ 215 short _InterlockedDecrement16(short volatile *_Addend); 216 long _InterlockedExchange(long volatile *_Target, long _Value); 217 static __inline__ 218 short _InterlockedExchange16(short volatile *_Target, short _Value); 219 static __inline__ 220 char _InterlockedExchange8(char volatile *_Target, char _Value); 221 static __inline__ 222 long __cdecl _InterlockedExchangeAdd(long volatile *_Addend, long _Value); 223 long _InterlockedExchangeAdd_HLEAcquire(long volatile *, long); 224 long _InterlockedExchangeAdd_HLERelease(long volatile *, long); 225 static __inline__ 226 short _InterlockedExchangeAdd16(short volatile *_Addend, short _Value); 227 __int64 _InterlockedExchangeAdd64_HLEAcquire(__int64 volatile *, __int64); 228 __int64 _InterlockedExchangeAdd64_HLERelease(__int64 volatile *, __int64); 229 static __inline__ 230 char _InterlockedExchangeAdd8(char volatile *_Addend, char _Value); 231 static __inline__ 232 long __cdecl _InterlockedIncrement(long volatile *_Addend); 233 static __inline__ 234 short _InterlockedIncrement16(short volatile *_Addend); 235 static __inline__ 236 long _InterlockedOr(long volatile *_Value, long _Mask); 237 static __inline__ 238 short _InterlockedOr16(short volatile *_Value, short _Mask); 239 static __inline__ 240 char _InterlockedOr8(char volatile *_Value, char _Mask); 241 static __inline__ 242 long _InterlockedXor(long volatile *_Value, long _Mask); 243 static __inline__ 244 short _InterlockedXor16(short volatile *_Value, short _Mask); 245 static __inline__ 246 char _InterlockedXor8(char volatile *_Value, char _Mask); 247 void __cdecl _invpcid(unsigned int, void *); 248 static __inline__ 249 unsigned long __cdecl _lrotl(unsigned long, int); 250 static __inline__ 251 unsigned long __cdecl _lrotr(unsigned long, int); 252 static __inline__ 253 static __inline__ 254 void _ReadBarrier(void); 255 static __inline__ 256 void _ReadWriteBarrier(void); 257 static __inline__ 258 void *_ReturnAddress(void); 259 unsigned int _rorx_u32(unsigned int, const unsigned int); 260 static __inline__ 261 unsigned int __cdecl _rotl(unsigned int _Value, int _Shift); 262 static __inline__ 263 unsigned short _rotl16(unsigned short _Value, unsigned char _Shift); 264 static __inline__ 265 unsigned __int64 __cdecl _rotl64(unsigned __int64 _Value, int _Shift); 266 static __inline__ 267 unsigned char _rotl8(unsigned char _Value, unsigned char _Shift); 268 static __inline__ 269 unsigned int __cdecl _rotr(unsigned int _Value, int _Shift); 270 static __inline__ 271 unsigned short _rotr16(unsigned short _Value, unsigned char _Shift); 272 static __inline__ 273 unsigned __int64 __cdecl _rotr64(unsigned __int64 _Value, int _Shift); 274 static __inline__ 275 unsigned char _rotr8(unsigned char _Value, unsigned char _Shift); 276 int _sarx_i32(int, unsigned int); 277 #if __STDC_HOSTED__ 278 int __cdecl _setjmp(jmp_buf); 279 #endif 280 unsigned int _shlx_u32(unsigned int, unsigned int); 281 unsigned int _shrx_u32(unsigned int, unsigned int); 282 void _Store_HLERelease(long volatile *, long); 283 void _Store64_HLERelease(__int64 volatile *, __int64); 284 void _StorePointer_HLERelease(void *volatile *, void *); 285 static __inline__ 286 void _WriteBarrier(void); 287 unsigned __int32 xbegin(void); 288 void _xend(void); 289 static __inline__ 290 #define _XCR_XFEATURE_ENABLED_MASK 0 291 unsigned __int64 __cdecl _xgetbv(unsigned int); 292 void __cdecl _xsetbv(unsigned int, unsigned __int64); 293 294 /* These additional intrinsics are turned on in x64/amd64/x86_64 mode. */ 295 #ifdef __x86_64__ 296 void __addgsbyte(unsigned long, unsigned char); 297 void __addgsdword(unsigned long, unsigned long); 298 void __addgsqword(unsigned long, unsigned __int64); 299 void __addgsword(unsigned long, unsigned short); 300 static __inline__ 301 void __faststorefence(void); 302 void __incgsbyte(unsigned long); 303 void __incgsdword(unsigned long); 304 void __incgsqword(unsigned long); 305 void __incgsword(unsigned long); 306 unsigned char __lwpins64(unsigned __int64, unsigned int, unsigned int); 307 void __lwpval64(unsigned __int64, unsigned int, unsigned int); 308 unsigned __int64 __lzcnt64(unsigned __int64); 309 static __inline__ 310 void __movsq(unsigned long long *, unsigned long long const *, size_t); 311 __int64 __mulh(__int64, __int64); 312 static __inline__ 313 unsigned __int64 __popcnt64(unsigned __int64); 314 static __inline__ 315 unsigned char __readgsbyte(unsigned long); 316 static __inline__ 317 unsigned long __readgsdword(unsigned long); 318 static __inline__ 319 unsigned __int64 __readgsqword(unsigned long); 320 unsigned short __readgsword(unsigned long); 321 unsigned __int64 __shiftleft128(unsigned __int64 _LowPart, 322 unsigned __int64 _HighPart, 323 unsigned char _Shift); 324 unsigned __int64 __shiftright128(unsigned __int64 _LowPart, 325 unsigned __int64 _HighPart, 326 unsigned char _Shift); 327 static __inline__ 328 void __stosq(unsigned __int64 *, unsigned __int64, size_t); 329 unsigned char __vmx_on(unsigned __int64 *); 330 unsigned char __vmx_vmclear(unsigned __int64 *); 331 unsigned char __vmx_vmlaunch(void); 332 unsigned char __vmx_vmptrld(unsigned __int64 *); 333 unsigned char __vmx_vmread(size_t, size_t *); 334 unsigned char __vmx_vmresume(void); 335 unsigned char __vmx_vmwrite(size_t, size_t); 336 void __writegsbyte(unsigned long, unsigned char); 337 void __writegsdword(unsigned long, unsigned long); 338 void __writegsqword(unsigned long, unsigned __int64); 339 void __writegsword(unsigned long, unsigned short); 340 static __inline__ 341 unsigned char _BitScanForward64(unsigned long *_Index, unsigned __int64 _Mask); 342 static __inline__ 343 unsigned char _BitScanReverse64(unsigned long *_Index, unsigned __int64 _Mask); 344 static __inline__ 345 unsigned char _bittest64(__int64 const *, __int64); 346 static __inline__ 347 unsigned char _bittestandcomplement64(__int64 *, __int64); 348 static __inline__ 349 unsigned char _bittestandreset64(__int64 *, __int64); 350 static __inline__ 351 unsigned char _bittestandset64(__int64 *, __int64); 352 unsigned __int64 __cdecl _byteswap_uint64(unsigned __int64); 353 long _InterlockedAnd_np(long volatile *_Value, long _Mask); 354 short _InterlockedAnd16_np(short volatile *_Value, short _Mask); 355 __int64 _InterlockedAnd64_np(__int64 volatile *_Value, __int64 _Mask); 356 char _InterlockedAnd8_np(char volatile *_Value, char _Mask); 357 unsigned char _interlockedbittestandreset64(__int64 volatile *, __int64); 358 static __inline__ 359 unsigned char _interlockedbittestandset64(__int64 volatile *, __int64); 360 long _InterlockedCompareExchange_np(long volatile *_Destination, long _Exchange, 361 long _Comparand); 362 unsigned char _InterlockedCompareExchange128(__int64 volatile *_Destination, 363 __int64 _ExchangeHigh, 364 __int64 _ExchangeLow, 365 __int64 *_CompareandResult); 366 unsigned char _InterlockedCompareExchange128_np(__int64 volatile *_Destination, 367 __int64 _ExchangeHigh, 368 __int64 _ExchangeLow, 369 __int64 *_ComparandResult); 370 short _InterlockedCompareExchange16_np(short volatile *_Destination, 371 short _Exchange, short _Comparand); 372 __int64 _InterlockedCompareExchange64_HLEAcquire(__int64 volatile *, __int64, 373 __int64); 374 __int64 _InterlockedCompareExchange64_HLERelease(__int64 volatile *, __int64, 375 __int64); 376 __int64 _InterlockedCompareExchange64_np(__int64 volatile *_Destination, 377 __int64 _Exchange, __int64 _Comparand); 378 void *_InterlockedCompareExchangePointer(void *volatile *_Destination, 379 void *_Exchange, void *_Comparand); 380 void *_InterlockedCompareExchangePointer_np(void *volatile *_Destination, 381 void *_Exchange, void *_Comparand); 382 static __inline__ 383 __int64 _InterlockedDecrement64(__int64 volatile *_Addend); 384 static __inline__ 385 __int64 _InterlockedExchange64(__int64 volatile *_Target, __int64 _Value); 386 static __inline__ 387 __int64 _InterlockedExchangeAdd64(__int64 volatile *_Addend, __int64 _Value); 388 void *_InterlockedExchangePointer(void *volatile *_Target, void *_Value); 389 static __inline__ 390 __int64 _InterlockedIncrement64(__int64 volatile *_Addend); 391 long _InterlockedOr_np(long volatile *_Value, long _Mask); 392 short _InterlockedOr16_np(short volatile *_Value, short _Mask); 393 static __inline__ 394 __int64 _InterlockedOr64(__int64 volatile *_Value, __int64 _Mask); 395 __int64 _InterlockedOr64_np(__int64 volatile *_Value, __int64 _Mask); 396 char _InterlockedOr8_np(char volatile *_Value, char _Mask); 397 long _InterlockedXor_np(long volatile *_Value, long _Mask); 398 short _InterlockedXor16_np(short volatile *_Value, short _Mask); 399 static __inline__ 400 __int64 _InterlockedXor64(__int64 volatile *_Value, __int64 _Mask); 401 __int64 _InterlockedXor64_np(__int64 volatile *_Value, __int64 _Mask); 402 char _InterlockedXor8_np(char volatile *_Value, char _Mask); 403 static __inline__ 404 __int64 _mul128(__int64 _Multiplier, __int64 _Multiplicand, 405 __int64 *_HighProduct); 406 unsigned __int64 _rorx_u64(unsigned __int64, const unsigned int); 407 __int64 _sarx_i64(__int64, unsigned int); 408 #if __STDC_HOSTED__ 409 int __cdecl _setjmpex(jmp_buf); 410 #endif 411 unsigned __int64 _shlx_u64(unsigned __int64, unsigned int); 412 unsigned __int64 _shrx_u64(unsigned __int64, unsigned int); 413 /* 414 * Multiply two 64-bit integers and obtain a 64-bit result. 415 * The low-half is returned directly and the high half is in an out parameter. 416 */ 417 static __inline__ unsigned __int64 __DEFAULT_FN_ATTRS 418 _umul128(unsigned __int64 _Multiplier, unsigned __int64 _Multiplicand, 419 unsigned __int64 *_HighProduct) { 420 unsigned __int128 _FullProduct = 421 (unsigned __int128)_Multiplier * (unsigned __int128)_Multiplicand; 422 *_HighProduct = _FullProduct >> 64; 423 return _FullProduct; 424 } 425 static __inline__ unsigned __int64 __DEFAULT_FN_ATTRS 426 __umulh(unsigned __int64 _Multiplier, unsigned __int64 _Multiplicand) { 427 unsigned __int128 _FullProduct = 428 (unsigned __int128)_Multiplier * (unsigned __int128)_Multiplicand; 429 return _FullProduct >> 64; 430 } 431 432 #endif /* __x86_64__ */ 433 434 /*----------------------------------------------------------------------------*\ 435 |* Multiplication 436 \*----------------------------------------------------------------------------*/ 437 static __inline__ __int64 __DEFAULT_FN_ATTRS 438 __emul(int __in1, int __in2) { 439 return (__int64)__in1 * (__int64)__in2; 440 } 441 static __inline__ unsigned __int64 __DEFAULT_FN_ATTRS 442 __emulu(unsigned int __in1, unsigned int __in2) { 443 return (unsigned __int64)__in1 * (unsigned __int64)__in2; 444 } 445 /*----------------------------------------------------------------------------*\ 446 |* Bit Twiddling 447 \*----------------------------------------------------------------------------*/ 448 static __inline__ unsigned char __DEFAULT_FN_ATTRS 449 _rotl8(unsigned char _Value, unsigned char _Shift) { 450 _Shift &= 0x7; 451 return _Shift ? (_Value << _Shift) | (_Value >> (8 - _Shift)) : _Value; 452 } 453 static __inline__ unsigned char __DEFAULT_FN_ATTRS 454 _rotr8(unsigned char _Value, unsigned char _Shift) { 455 _Shift &= 0x7; 456 return _Shift ? (_Value >> _Shift) | (_Value << (8 - _Shift)) : _Value; 457 } 458 static __inline__ unsigned short __DEFAULT_FN_ATTRS 459 _rotl16(unsigned short _Value, unsigned char _Shift) { 460 _Shift &= 0xf; 461 return _Shift ? (_Value << _Shift) | (_Value >> (16 - _Shift)) : _Value; 462 } 463 static __inline__ unsigned short __DEFAULT_FN_ATTRS 464 _rotr16(unsigned short _Value, unsigned char _Shift) { 465 _Shift &= 0xf; 466 return _Shift ? (_Value >> _Shift) | (_Value << (16 - _Shift)) : _Value; 467 } 468 static __inline__ unsigned int __DEFAULT_FN_ATTRS 469 _rotl(unsigned int _Value, int _Shift) { 470 _Shift &= 0x1f; 471 return _Shift ? (_Value << _Shift) | (_Value >> (32 - _Shift)) : _Value; 472 } 473 static __inline__ unsigned int __DEFAULT_FN_ATTRS 474 _rotr(unsigned int _Value, int _Shift) { 475 _Shift &= 0x1f; 476 return _Shift ? (_Value >> _Shift) | (_Value << (32 - _Shift)) : _Value; 477 } 478 static __inline__ unsigned long __DEFAULT_FN_ATTRS 479 _lrotl(unsigned long _Value, int _Shift) { 480 _Shift &= 0x1f; 481 return _Shift ? (_Value << _Shift) | (_Value >> (32 - _Shift)) : _Value; 482 } 483 static __inline__ unsigned long __DEFAULT_FN_ATTRS 484 _lrotr(unsigned long _Value, int _Shift) { 485 _Shift &= 0x1f; 486 return _Shift ? (_Value >> _Shift) | (_Value << (32 - _Shift)) : _Value; 487 } 488 static 489 __inline__ unsigned __int64 __DEFAULT_FN_ATTRS 490 _rotl64(unsigned __int64 _Value, int _Shift) { 491 _Shift &= 0x3f; 492 return _Shift ? (_Value << _Shift) | (_Value >> (64 - _Shift)) : _Value; 493 } 494 static 495 __inline__ unsigned __int64 __DEFAULT_FN_ATTRS 496 _rotr64(unsigned __int64 _Value, int _Shift) { 497 _Shift &= 0x3f; 498 return _Shift ? (_Value >> _Shift) | (_Value << (64 - _Shift)) : _Value; 499 } 500 /*----------------------------------------------------------------------------*\ 501 |* Bit Counting and Testing 502 \*----------------------------------------------------------------------------*/ 503 static __inline__ unsigned char __DEFAULT_FN_ATTRS 504 _BitScanForward(unsigned long *_Index, unsigned long _Mask) { 505 if (!_Mask) 506 return 0; 507 *_Index = __builtin_ctzl(_Mask); 508 return 1; 509 } 510 static __inline__ unsigned char __DEFAULT_FN_ATTRS 511 _BitScanReverse(unsigned long *_Index, unsigned long _Mask) { 512 if (!_Mask) 513 return 0; 514 *_Index = 31 - __builtin_clzl(_Mask); 515 return 1; 516 } 517 static __inline__ unsigned short __DEFAULT_FN_ATTRS 518 __popcnt16(unsigned short _Value) { 519 return __builtin_popcount((int)_Value); 520 } 521 static __inline__ unsigned int __DEFAULT_FN_ATTRS 522 __popcnt(unsigned int _Value) { 523 return __builtin_popcount(_Value); 524 } 525 static __inline__ unsigned char __DEFAULT_FN_ATTRS 526 _bittest(long const *_BitBase, long _BitPos) { 527 return (*_BitBase >> _BitPos) & 1; 528 } 529 static __inline__ unsigned char __DEFAULT_FN_ATTRS 530 _bittestandcomplement(long *_BitBase, long _BitPos) { 531 unsigned char _Res = (*_BitBase >> _BitPos) & 1; 532 *_BitBase = *_BitBase ^ (1 << _BitPos); 533 return _Res; 534 } 535 static __inline__ unsigned char __DEFAULT_FN_ATTRS 536 _bittestandreset(long *_BitBase, long _BitPos) { 537 unsigned char _Res = (*_BitBase >> _BitPos) & 1; 538 *_BitBase = *_BitBase & ~(1 << _BitPos); 539 return _Res; 540 } 541 static __inline__ unsigned char __DEFAULT_FN_ATTRS 542 _bittestandset(long *_BitBase, long _BitPos) { 543 unsigned char _Res = (*_BitBase >> _BitPos) & 1; 544 *_BitBase = *_BitBase | (1 << _BitPos); 545 return _Res; 546 } 547 static __inline__ unsigned char __DEFAULT_FN_ATTRS 548 _interlockedbittestandset(long volatile *_BitBase, long _BitPos) { 549 long _PrevVal = __atomic_fetch_or(_BitBase, 1l << _BitPos, __ATOMIC_SEQ_CST); 550 return (_PrevVal >> _BitPos) & 1; 551 } 552 #ifdef __x86_64__ 553 static __inline__ unsigned char __DEFAULT_FN_ATTRS 554 _BitScanForward64(unsigned long *_Index, unsigned __int64 _Mask) { 555 if (!_Mask) 556 return 0; 557 *_Index = __builtin_ctzll(_Mask); 558 return 1; 559 } 560 static __inline__ unsigned char __DEFAULT_FN_ATTRS 561 _BitScanReverse64(unsigned long *_Index, unsigned __int64 _Mask) { 562 if (!_Mask) 563 return 0; 564 *_Index = 63 - __builtin_clzll(_Mask); 565 return 1; 566 } 567 static __inline__ 568 unsigned __int64 __DEFAULT_FN_ATTRS 569 __popcnt64(unsigned __int64 _Value) { 570 return __builtin_popcountll(_Value); 571 } 572 static __inline__ unsigned char __DEFAULT_FN_ATTRS 573 _bittest64(__int64 const *_BitBase, __int64 _BitPos) { 574 return (*_BitBase >> _BitPos) & 1; 575 } 576 static __inline__ unsigned char __DEFAULT_FN_ATTRS 577 _bittestandcomplement64(__int64 *_BitBase, __int64 _BitPos) { 578 unsigned char _Res = (*_BitBase >> _BitPos) & 1; 579 *_BitBase = *_BitBase ^ (1ll << _BitPos); 580 return _Res; 581 } 582 static __inline__ unsigned char __DEFAULT_FN_ATTRS 583 _bittestandreset64(__int64 *_BitBase, __int64 _BitPos) { 584 unsigned char _Res = (*_BitBase >> _BitPos) & 1; 585 *_BitBase = *_BitBase & ~(1ll << _BitPos); 586 return _Res; 587 } 588 static __inline__ unsigned char __DEFAULT_FN_ATTRS 589 _bittestandset64(__int64 *_BitBase, __int64 _BitPos) { 590 unsigned char _Res = (*_BitBase >> _BitPos) & 1; 591 *_BitBase = *_BitBase | (1ll << _BitPos); 592 return _Res; 593 } 594 static __inline__ unsigned char __DEFAULT_FN_ATTRS 595 _interlockedbittestandset64(__int64 volatile *_BitBase, __int64 _BitPos) { 596 long long _PrevVal = 597 __atomic_fetch_or(_BitBase, 1ll << _BitPos, __ATOMIC_SEQ_CST); 598 return (_PrevVal >> _BitPos) & 1; 599 } 600 #endif 601 /*----------------------------------------------------------------------------*\ 602 |* Interlocked Exchange Add 603 \*----------------------------------------------------------------------------*/ 604 static __inline__ char __DEFAULT_FN_ATTRS 605 _InterlockedExchangeAdd8(char volatile *_Addend, char _Value) { 606 return __atomic_fetch_add(_Addend, _Value, __ATOMIC_SEQ_CST); 607 } 608 static __inline__ short __DEFAULT_FN_ATTRS 609 _InterlockedExchangeAdd16(short volatile *_Addend, short _Value) { 610 return __atomic_fetch_add(_Addend, _Value, __ATOMIC_SEQ_CST); 611 } 612 #ifdef __x86_64__ 613 static __inline__ __int64 __DEFAULT_FN_ATTRS 614 _InterlockedExchangeAdd64(__int64 volatile *_Addend, __int64 _Value) { 615 return __atomic_fetch_add(_Addend, _Value, __ATOMIC_SEQ_CST); 616 } 617 #endif 618 /*----------------------------------------------------------------------------*\ 619 |* Interlocked Exchange Sub 620 \*----------------------------------------------------------------------------*/ 621 static __inline__ char __DEFAULT_FN_ATTRS 622 _InterlockedExchangeSub8(char volatile *_Subend, char _Value) { 623 return __atomic_fetch_sub(_Subend, _Value, __ATOMIC_SEQ_CST); 624 } 625 static __inline__ short __DEFAULT_FN_ATTRS 626 _InterlockedExchangeSub16(short volatile *_Subend, short _Value) { 627 return __atomic_fetch_sub(_Subend, _Value, __ATOMIC_SEQ_CST); 628 } 629 static __inline__ long __DEFAULT_FN_ATTRS 630 _InterlockedExchangeSub(long volatile *_Subend, long _Value) { 631 return __atomic_fetch_sub(_Subend, _Value, __ATOMIC_SEQ_CST); 632 } 633 #ifdef __x86_64__ 634 static __inline__ __int64 __DEFAULT_FN_ATTRS 635 _InterlockedExchangeSub64(__int64 volatile *_Subend, __int64 _Value) { 636 return __atomic_fetch_sub(_Subend, _Value, __ATOMIC_SEQ_CST); 637 } 638 #endif 639 /*----------------------------------------------------------------------------*\ 640 |* Interlocked Increment 641 \*----------------------------------------------------------------------------*/ 642 static __inline__ short __DEFAULT_FN_ATTRS 643 _InterlockedIncrement16(short volatile *_Value) { 644 return __atomic_add_fetch(_Value, 1, __ATOMIC_SEQ_CST); 645 } 646 #ifdef __x86_64__ 647 static __inline__ __int64 __DEFAULT_FN_ATTRS 648 _InterlockedIncrement64(__int64 volatile *_Value) { 649 return __atomic_add_fetch(_Value, 1, __ATOMIC_SEQ_CST); 650 } 651 #endif 652 /*----------------------------------------------------------------------------*\ 653 |* Interlocked Decrement 654 \*----------------------------------------------------------------------------*/ 655 static __inline__ short __DEFAULT_FN_ATTRS 656 _InterlockedDecrement16(short volatile *_Value) { 657 return __atomic_sub_fetch(_Value, 1, __ATOMIC_SEQ_CST); 658 } 659 #ifdef __x86_64__ 660 static __inline__ __int64 __DEFAULT_FN_ATTRS 661 _InterlockedDecrement64(__int64 volatile *_Value) { 662 return __atomic_sub_fetch(_Value, 1, __ATOMIC_SEQ_CST); 663 } 664 #endif 665 /*----------------------------------------------------------------------------*\ 666 |* Interlocked And 667 \*----------------------------------------------------------------------------*/ 668 static __inline__ char __DEFAULT_FN_ATTRS 669 _InterlockedAnd8(char volatile *_Value, char _Mask) { 670 return __atomic_and_fetch(_Value, _Mask, __ATOMIC_SEQ_CST); 671 } 672 static __inline__ short __DEFAULT_FN_ATTRS 673 _InterlockedAnd16(short volatile *_Value, short _Mask) { 674 return __atomic_and_fetch(_Value, _Mask, __ATOMIC_SEQ_CST); 675 } 676 static __inline__ long __DEFAULT_FN_ATTRS 677 _InterlockedAnd(long volatile *_Value, long _Mask) { 678 return __atomic_and_fetch(_Value, _Mask, __ATOMIC_SEQ_CST); 679 } 680 #ifdef __x86_64__ 681 static __inline__ __int64 __DEFAULT_FN_ATTRS 682 _InterlockedAnd64(__int64 volatile *_Value, __int64 _Mask) { 683 return __atomic_and_fetch(_Value, _Mask, __ATOMIC_SEQ_CST); 684 } 685 #endif 686 /*----------------------------------------------------------------------------*\ 687 |* Interlocked Or 688 \*----------------------------------------------------------------------------*/ 689 static __inline__ char __DEFAULT_FN_ATTRS 690 _InterlockedOr8(char volatile *_Value, char _Mask) { 691 return __atomic_or_fetch(_Value, _Mask, __ATOMIC_SEQ_CST); 692 } 693 static __inline__ short __DEFAULT_FN_ATTRS 694 _InterlockedOr16(short volatile *_Value, short _Mask) { 695 return __atomic_or_fetch(_Value, _Mask, __ATOMIC_SEQ_CST); 696 } 697 static __inline__ long __DEFAULT_FN_ATTRS 698 _InterlockedOr(long volatile *_Value, long _Mask) { 699 return __atomic_or_fetch(_Value, _Mask, __ATOMIC_SEQ_CST); 700 } 701 #ifdef __x86_64__ 702 static __inline__ __int64 __DEFAULT_FN_ATTRS 703 _InterlockedOr64(__int64 volatile *_Value, __int64 _Mask) { 704 return __atomic_or_fetch(_Value, _Mask, __ATOMIC_SEQ_CST); 705 } 706 #endif 707 /*----------------------------------------------------------------------------*\ 708 |* Interlocked Xor 709 \*----------------------------------------------------------------------------*/ 710 static __inline__ char __DEFAULT_FN_ATTRS 711 _InterlockedXor8(char volatile *_Value, char _Mask) { 712 return __atomic_xor_fetch(_Value, _Mask, __ATOMIC_SEQ_CST); 713 } 714 static __inline__ short __DEFAULT_FN_ATTRS 715 _InterlockedXor16(short volatile *_Value, short _Mask) { 716 return __atomic_xor_fetch(_Value, _Mask, __ATOMIC_SEQ_CST); 717 } 718 static __inline__ long __DEFAULT_FN_ATTRS 719 _InterlockedXor(long volatile *_Value, long _Mask) { 720 return __atomic_xor_fetch(_Value, _Mask, __ATOMIC_SEQ_CST); 721 } 722 #ifdef __x86_64__ 723 static __inline__ __int64 __DEFAULT_FN_ATTRS 724 _InterlockedXor64(__int64 volatile *_Value, __int64 _Mask) { 725 return __atomic_xor_fetch(_Value, _Mask, __ATOMIC_SEQ_CST); 726 } 727 #endif 728 /*----------------------------------------------------------------------------*\ 729 |* Interlocked Exchange 730 \*----------------------------------------------------------------------------*/ 731 static __inline__ char __DEFAULT_FN_ATTRS 732 _InterlockedExchange8(char volatile *_Target, char _Value) { 733 __atomic_exchange(_Target, &_Value, &_Value, __ATOMIC_SEQ_CST); 734 return _Value; 735 } 736 static __inline__ short __DEFAULT_FN_ATTRS 737 _InterlockedExchange16(short volatile *_Target, short _Value) { 738 __atomic_exchange(_Target, &_Value, &_Value, __ATOMIC_SEQ_CST); 739 return _Value; 740 } 741 #ifdef __x86_64__ 742 static __inline__ __int64 __DEFAULT_FN_ATTRS 743 _InterlockedExchange64(__int64 volatile *_Target, __int64 _Value) { 744 __atomic_exchange(_Target, &_Value, &_Value, __ATOMIC_SEQ_CST); 745 return _Value; 746 } 747 #endif 748 /*----------------------------------------------------------------------------*\ 749 |* Interlocked Compare Exchange 750 \*----------------------------------------------------------------------------*/ 751 static __inline__ char __DEFAULT_FN_ATTRS 752 _InterlockedCompareExchange8(char volatile *_Destination, 753 char _Exchange, char _Comparand) { 754 __atomic_compare_exchange(_Destination, &_Comparand, &_Exchange, 0, 755 __ATOMIC_SEQ_CST, __ATOMIC_SEQ_CST); 756 return _Comparand; 757 } 758 static __inline__ short __DEFAULT_FN_ATTRS 759 _InterlockedCompareExchange16(short volatile *_Destination, 760 short _Exchange, short _Comparand) { 761 __atomic_compare_exchange(_Destination, &_Comparand, &_Exchange, 0, 762 __ATOMIC_SEQ_CST, __ATOMIC_SEQ_CST); 763 return _Comparand; 764 } 765 static __inline__ __int64 __DEFAULT_FN_ATTRS 766 _InterlockedCompareExchange64(__int64 volatile *_Destination, 767 __int64 _Exchange, __int64 _Comparand) { 768 __atomic_compare_exchange(_Destination, &_Comparand, &_Exchange, 0, 769 __ATOMIC_SEQ_CST, __ATOMIC_SEQ_CST); 770 return _Comparand; 771 } 772 /*----------------------------------------------------------------------------*\ 773 |* Barriers 774 \*----------------------------------------------------------------------------*/ 775 static __inline__ void __DEFAULT_FN_ATTRS 776 __attribute__((__deprecated__("use other intrinsics or C++11 atomics instead"))) 777 _ReadWriteBarrier(void) { 778 __atomic_signal_fence(__ATOMIC_SEQ_CST); 779 } 780 static __inline__ void __DEFAULT_FN_ATTRS 781 __attribute__((__deprecated__("use other intrinsics or C++11 atomics instead"))) 782 _ReadBarrier(void) { 783 __atomic_signal_fence(__ATOMIC_SEQ_CST); 784 } 785 static __inline__ void __DEFAULT_FN_ATTRS 786 __attribute__((__deprecated__("use other intrinsics or C++11 atomics instead"))) 787 _WriteBarrier(void) { 788 __atomic_signal_fence(__ATOMIC_SEQ_CST); 789 } 790 #ifdef __x86_64__ 791 static __inline__ void __DEFAULT_FN_ATTRS 792 __faststorefence(void) { 793 __atomic_thread_fence(__ATOMIC_SEQ_CST); 794 } 795 #endif 796 /*----------------------------------------------------------------------------*\ 797 |* readfs, readgs 798 |* (Pointers in address space #256 and #257 are relative to the GS and FS 799 |* segment registers, respectively.) 800 \*----------------------------------------------------------------------------*/ 801 #define __ptr_to_addr_space(__addr_space_nbr, __type, __offset) \ 802 ((volatile __type __attribute__((__address_space__(__addr_space_nbr)))*) \ 803 (__offset)) 804 805 #ifdef __i386__ 806 static __inline__ unsigned char __DEFAULT_FN_ATTRS 807 __readfsbyte(unsigned long __offset) { 808 return *__ptr_to_addr_space(257, unsigned char, __offset); 809 } 810 static __inline__ unsigned __int64 __DEFAULT_FN_ATTRS 811 __readfsqword(unsigned long __offset) { 812 return *__ptr_to_addr_space(257, unsigned __int64, __offset); 813 } 814 static __inline__ unsigned short __DEFAULT_FN_ATTRS 815 __readfsword(unsigned long __offset) { 816 return *__ptr_to_addr_space(257, unsigned short, __offset); 817 } 818 #endif 819 #ifdef __x86_64__ 820 static __inline__ unsigned char __DEFAULT_FN_ATTRS 821 __readgsbyte(unsigned long __offset) { 822 return *__ptr_to_addr_space(256, unsigned char, __offset); 823 } 824 static __inline__ unsigned long __DEFAULT_FN_ATTRS 825 __readgsdword(unsigned long __offset) { 826 return *__ptr_to_addr_space(256, unsigned long, __offset); 827 } 828 static __inline__ unsigned __int64 __DEFAULT_FN_ATTRS 829 __readgsqword(unsigned long __offset) { 830 return *__ptr_to_addr_space(256, unsigned __int64, __offset); 831 } 832 static __inline__ unsigned short __DEFAULT_FN_ATTRS 833 __readgsword(unsigned long __offset) { 834 return *__ptr_to_addr_space(256, unsigned short, __offset); 835 } 836 #endif 837 #undef __ptr_to_addr_space 838 /*----------------------------------------------------------------------------*\ 839 |* movs, stos 840 \*----------------------------------------------------------------------------*/ 841 #if defined(__i386__) || defined(__x86_64__) 842 static __inline__ void __DEFAULT_FN_ATTRS 843 __movsb(unsigned char *__dst, unsigned char const *__src, size_t __n) { 844 __asm__("rep movsb" : : "D"(__dst), "S"(__src), "c"(__n) 845 : "%edi", "%esi", "%ecx"); 846 } 847 static __inline__ void __DEFAULT_FN_ATTRS 848 __movsd(unsigned long *__dst, unsigned long const *__src, size_t __n) { 849 __asm__("rep movsl" : : "D"(__dst), "S"(__src), "c"(__n) 850 : "%edi", "%esi", "%ecx"); 851 } 852 static __inline__ void __DEFAULT_FN_ATTRS 853 __movsw(unsigned short *__dst, unsigned short const *__src, size_t __n) { 854 __asm__("rep movsw" : : "D"(__dst), "S"(__src), "c"(__n) 855 : "%edi", "%esi", "%ecx"); 856 } 857 static __inline__ void __DEFAULT_FN_ATTRS 858 __stosb(unsigned char *__dst, unsigned char __x, size_t __n) { 859 __asm__("rep stosb" : : "D"(__dst), "a"(__x), "c"(__n) 860 : "%edi", "%ecx"); 861 } 862 static __inline__ void __DEFAULT_FN_ATTRS 863 __stosd(unsigned long *__dst, unsigned long __x, size_t __n) { 864 __asm__("rep stosl" : : "D"(__dst), "a"(__x), "c"(__n) 865 : "%edi", "%ecx"); 866 } 867 static __inline__ void __DEFAULT_FN_ATTRS 868 __stosw(unsigned short *__dst, unsigned short __x, size_t __n) { 869 __asm__("rep stosw" : : "D"(__dst), "a"(__x), "c"(__n) 870 : "%edi", "%ecx"); 871 } 872 #endif 873 #ifdef __x86_64__ 874 static __inline__ void __DEFAULT_FN_ATTRS 875 __movsq(unsigned long long *__dst, unsigned long long const *__src, size_t __n) { 876 __asm__("rep movsq" : : "D"(__dst), "S"(__src), "c"(__n) 877 : "%edi", "%esi", "%ecx"); 878 } 879 static __inline__ void __DEFAULT_FN_ATTRS 880 __stosq(unsigned __int64 *__dst, unsigned __int64 __x, size_t __n) { 881 __asm__("rep stosq" : : "D"(__dst), "a"(__x), "c"(__n) 882 : "%edi", "%ecx"); 883 } 884 #endif 885 886 /*----------------------------------------------------------------------------*\ 887 |* Misc 888 \*----------------------------------------------------------------------------*/ 889 static __inline__ void * __DEFAULT_FN_ATTRS 890 _AddressOfReturnAddress(void) { 891 return (void*)((char*)__builtin_frame_address(0) + sizeof(void*)); 892 } 893 static __inline__ void * __DEFAULT_FN_ATTRS 894 _ReturnAddress(void) { 895 return __builtin_return_address(0); 896 } 897 #if defined(__i386__) || defined(__x86_64__) 898 static __inline__ void __DEFAULT_FN_ATTRS 899 __cpuid(int __info[4], int __level) { 900 __asm__ ("cpuid" : "=a"(__info[0]), "=b" (__info[1]), "=c"(__info[2]), "=d"(__info[3]) 901 : "a"(__level)); 902 } 903 static __inline__ void __DEFAULT_FN_ATTRS 904 __cpuidex(int __info[4], int __level, int __ecx) { 905 __asm__ ("cpuid" : "=a"(__info[0]), "=b" (__info[1]), "=c"(__info[2]), "=d"(__info[3]) 906 : "a"(__level), "c"(__ecx)); 907 } 908 static __inline__ unsigned __int64 __cdecl __DEFAULT_FN_ATTRS 909 _xgetbv(unsigned int __xcr_no) { 910 unsigned int __eax, __edx; 911 __asm__ ("xgetbv" : "=a" (__eax), "=d" (__edx) : "c" (__xcr_no)); 912 return ((unsigned __int64)__edx << 32) | __eax; 913 } 914 static __inline__ void __DEFAULT_FN_ATTRS 915 __halt(void) { 916 __asm__ volatile ("hlt"); 917 } 918 #endif 919 920 /*----------------------------------------------------------------------------*\ 921 |* Privileged intrinsics 922 \*----------------------------------------------------------------------------*/ 923 #if defined(__i386__) || defined(__x86_64__) 924 static __inline__ unsigned __int64 __DEFAULT_FN_ATTRS 925 __readmsr(unsigned long __register) { 926 // Loads the contents of a 64-bit model specific register (MSR) specified in 927 // the ECX register into registers EDX:EAX. The EDX register is loaded with 928 // the high-order 32 bits of the MSR and the EAX register is loaded with the 929 // low-order 32 bits. If less than 64 bits are implemented in the MSR being 930 // read, the values returned to EDX:EAX in unimplemented bit locations are 931 // undefined. 932 unsigned long __edx; 933 unsigned long __eax; 934 __asm__ ("rdmsr" : "=d"(__edx), "=a"(__eax) : "c"(__register)); 935 return (((unsigned __int64)__edx) << 32) | (unsigned __int64)__eax; 936 } 937 938 static __inline__ unsigned long __DEFAULT_FN_ATTRS 939 __readcr3(void) { 940 unsigned long __cr3_val; 941 __asm__ __volatile__ ("mov %%cr3, %0" : "=q"(__cr3_val) : : "memory"); 942 return __cr3_val; 943 } 944 945 static __inline__ void __DEFAULT_FN_ATTRS 946 __writecr3(unsigned int __cr3_val) { 947 __asm__ ("mov %0, %%cr3" : : "q"(__cr3_val) : "memory"); 948 } 949 #endif 950 951 #ifdef __cplusplus 952 } 953 #endif 954 955 #undef __DEFAULT_FN_ATTRS 956 957 #endif /* __INTRIN_H */ 958 #endif /* _MSC_VER */ 959