kmp_atomic.h 91 KB
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767
/*
 * kmp_atomic.h - ATOMIC header file
 */

//===----------------------------------------------------------------------===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//

#ifndef KMP_ATOMIC_H
#define KMP_ATOMIC_H

#include "kmp_lock.h"
#include "kmp_os.h"

#if OMPT_SUPPORT
#include "ompt-specific.h"
#endif

// C++ build port.
// Intel compiler does not support _Complex datatype on win.
// Intel compiler supports _Complex datatype on lin and mac.
// On the other side, there is a problem of stack alignment on lin_32 and mac_32
// if the rhs is cmplx80 or cmplx128 typedef'ed datatype.
// The decision is: to use compiler supported _Complex type on lin and mac,
//                  to use typedef'ed types on win.
// Condition for WIN64 was modified in anticipation of 10.1 build compiler.

#if defined(__cplusplus) && (KMP_OS_WINDOWS)
// create shortcuts for c99 complex types

// Visual Studio cannot have function parameters that have the
// align __declspec attribute, so we must remove it. (Compiler Error C2719)
#if KMP_COMPILER_MSVC
#undef KMP_DO_ALIGN
#define KMP_DO_ALIGN(alignment) /* Nothing */
#endif

#if defined(_MSC_VER) && (_MSC_VER < 1600) && defined(_DEBUG)
// Workaround for the problem of _DebugHeapTag unresolved external.
// This problem prevented to use our static debug library for C tests
// compiled with /MDd option (the library itself built with /MTd),
#undef _DEBUG
#define _DEBUG_TEMPORARILY_UNSET_
#endif

#include <complex>

template <typename type_lhs, typename type_rhs>
std::complex<type_lhs> __kmp_lhs_div_rhs(const std::complex<type_lhs> &lhs,
                                         const std::complex<type_rhs> &rhs) {
  type_lhs a = lhs.real();
  type_lhs b = lhs.imag();
  type_rhs c = rhs.real();
  type_rhs d = rhs.imag();
  type_rhs den = c * c + d * d;
  type_rhs r = (a * c + b * d);
  type_rhs i = (b * c - a * d);
  std::complex<type_lhs> ret(r / den, i / den);
  return ret;
}

// complex8
struct __kmp_cmplx64_t : std::complex<double> {

  __kmp_cmplx64_t() : std::complex<double>() {}

  __kmp_cmplx64_t(const std::complex<double> &cd) : std::complex<double>(cd) {}

  void operator/=(const __kmp_cmplx64_t &rhs) {
    std::complex<double> lhs = *this;
    *this = __kmp_lhs_div_rhs(lhs, rhs);
  }

  __kmp_cmplx64_t operator/(const __kmp_cmplx64_t &rhs) {
    std::complex<double> lhs = *this;
    return __kmp_lhs_div_rhs(lhs, rhs);
  }
};
typedef struct __kmp_cmplx64_t kmp_cmplx64;

// complex4
struct __kmp_cmplx32_t : std::complex<float> {

  __kmp_cmplx32_t() : std::complex<float>() {}

  __kmp_cmplx32_t(const std::complex<float> &cf) : std::complex<float>(cf) {}

  __kmp_cmplx32_t operator+(const __kmp_cmplx32_t &b) {
    std::complex<float> lhs = *this;
    std::complex<float> rhs = b;
    return (lhs + rhs);
  }
  __kmp_cmplx32_t operator-(const __kmp_cmplx32_t &b) {
    std::complex<float> lhs = *this;
    std::complex<float> rhs = b;
    return (lhs - rhs);
  }
  __kmp_cmplx32_t operator*(const __kmp_cmplx32_t &b) {
    std::complex<float> lhs = *this;
    std::complex<float> rhs = b;
    return (lhs * rhs);
  }

  __kmp_cmplx32_t operator+(const kmp_cmplx64 &b) {
    kmp_cmplx64 t = kmp_cmplx64(*this) + b;
    std::complex<double> d(t);
    std::complex<float> f(d);
    __kmp_cmplx32_t r(f);
    return r;
  }
  __kmp_cmplx32_t operator-(const kmp_cmplx64 &b) {
    kmp_cmplx64 t = kmp_cmplx64(*this) - b;
    std::complex<double> d(t);
    std::complex<float> f(d);
    __kmp_cmplx32_t r(f);
    return r;
  }
  __kmp_cmplx32_t operator*(const kmp_cmplx64 &b) {
    kmp_cmplx64 t = kmp_cmplx64(*this) * b;
    std::complex<double> d(t);
    std::complex<float> f(d);
    __kmp_cmplx32_t r(f);
    return r;
  }

  void operator/=(const __kmp_cmplx32_t &rhs) {
    std::complex<float> lhs = *this;
    *this = __kmp_lhs_div_rhs(lhs, rhs);
  }

  __kmp_cmplx32_t operator/(const __kmp_cmplx32_t &rhs) {
    std::complex<float> lhs = *this;
    return __kmp_lhs_div_rhs(lhs, rhs);
  }

  void operator/=(const kmp_cmplx64 &rhs) {
    std::complex<float> lhs = *this;
    *this = __kmp_lhs_div_rhs(lhs, rhs);
  }

  __kmp_cmplx32_t operator/(const kmp_cmplx64 &rhs) {
    std::complex<float> lhs = *this;
    return __kmp_lhs_div_rhs(lhs, rhs);
  }
};
typedef struct __kmp_cmplx32_t kmp_cmplx32;

// complex10
struct KMP_DO_ALIGN(16) __kmp_cmplx80_t : std::complex<long double> {

  __kmp_cmplx80_t() : std::complex<long double>() {}

  __kmp_cmplx80_t(const std::complex<long double> &cld)
      : std::complex<long double>(cld) {}

  void operator/=(const __kmp_cmplx80_t &rhs) {
    std::complex<long double> lhs = *this;
    *this = __kmp_lhs_div_rhs(lhs, rhs);
  }

  __kmp_cmplx80_t operator/(const __kmp_cmplx80_t &rhs) {
    std::complex<long double> lhs = *this;
    return __kmp_lhs_div_rhs(lhs, rhs);
  }
};
typedef KMP_DO_ALIGN(16) struct __kmp_cmplx80_t kmp_cmplx80;

// complex16
#if KMP_HAVE_QUAD
struct __kmp_cmplx128_t : std::complex<_Quad> {

  __kmp_cmplx128_t() : std::complex<_Quad>() {}

  __kmp_cmplx128_t(const std::complex<_Quad> &cq) : std::complex<_Quad>(cq) {}

  void operator/=(const __kmp_cmplx128_t &rhs) {
    std::complex<_Quad> lhs = *this;
    *this = __kmp_lhs_div_rhs(lhs, rhs);
  }

  __kmp_cmplx128_t operator/(const __kmp_cmplx128_t &rhs) {
    std::complex<_Quad> lhs = *this;
    return __kmp_lhs_div_rhs(lhs, rhs);
  }
};
typedef struct __kmp_cmplx128_t kmp_cmplx128;
#endif /* KMP_HAVE_QUAD */

#ifdef _DEBUG_TEMPORARILY_UNSET_
#undef _DEBUG_TEMPORARILY_UNSET_
// Set it back now
#define _DEBUG 1
#endif

#else
// create shortcuts for c99 complex types
typedef float _Complex kmp_cmplx32;
typedef double _Complex kmp_cmplx64;
typedef long double _Complex kmp_cmplx80;
#if KMP_HAVE_QUAD
typedef _Quad _Complex kmp_cmplx128;
#endif
#endif

// Compiler 12.0 changed alignment of 16 and 32-byte arguments (like _Quad
// and kmp_cmplx128) on IA-32 architecture. The following aligned structures
// are implemented to support the old alignment in 10.1, 11.0, 11.1 and
// introduce the new alignment in 12.0. See CQ88405.
#if KMP_ARCH_X86 && KMP_HAVE_QUAD

// 4-byte aligned structures for backward compatibility.

#pragma pack(push, 4)

struct KMP_DO_ALIGN(4) Quad_a4_t {
  _Quad q;

  Quad_a4_t() : q() {}
  Quad_a4_t(const _Quad &cq) : q(cq) {}

  Quad_a4_t operator+(const Quad_a4_t &b) {
    _Quad lhs = (*this).q;
    _Quad rhs = b.q;
    return (Quad_a4_t)(lhs + rhs);
  }

  Quad_a4_t operator-(const Quad_a4_t &b) {
    _Quad lhs = (*this).q;
    _Quad rhs = b.q;
    return (Quad_a4_t)(lhs - rhs);
  }
  Quad_a4_t operator*(const Quad_a4_t &b) {
    _Quad lhs = (*this).q;
    _Quad rhs = b.q;
    return (Quad_a4_t)(lhs * rhs);
  }

  Quad_a4_t operator/(const Quad_a4_t &b) {
    _Quad lhs = (*this).q;
    _Quad rhs = b.q;
    return (Quad_a4_t)(lhs / rhs);
  }
};

struct KMP_DO_ALIGN(4) kmp_cmplx128_a4_t {
  kmp_cmplx128 q;

  kmp_cmplx128_a4_t() : q() {}

  kmp_cmplx128_a4_t(const kmp_cmplx128 &c128) : q(c128) {}

  kmp_cmplx128_a4_t operator+(const kmp_cmplx128_a4_t &b) {
    kmp_cmplx128 lhs = (*this).q;
    kmp_cmplx128 rhs = b.q;
    return (kmp_cmplx128_a4_t)(lhs + rhs);
  }
  kmp_cmplx128_a4_t operator-(const kmp_cmplx128_a4_t &b) {
    kmp_cmplx128 lhs = (*this).q;
    kmp_cmplx128 rhs = b.q;
    return (kmp_cmplx128_a4_t)(lhs - rhs);
  }
  kmp_cmplx128_a4_t operator*(const kmp_cmplx128_a4_t &b) {
    kmp_cmplx128 lhs = (*this).q;
    kmp_cmplx128 rhs = b.q;
    return (kmp_cmplx128_a4_t)(lhs * rhs);
  }

  kmp_cmplx128_a4_t operator/(const kmp_cmplx128_a4_t &b) {
    kmp_cmplx128 lhs = (*this).q;
    kmp_cmplx128 rhs = b.q;
    return (kmp_cmplx128_a4_t)(lhs / rhs);
  }
};

#pragma pack(pop)

// New 16-byte aligned structures for 12.0 compiler.
struct KMP_DO_ALIGN(16) Quad_a16_t {
  _Quad q;

  Quad_a16_t() : q() {}
  Quad_a16_t(const _Quad &cq) : q(cq) {}

  Quad_a16_t operator+(const Quad_a16_t &b) {
    _Quad lhs = (*this).q;
    _Quad rhs = b.q;
    return (Quad_a16_t)(lhs + rhs);
  }

  Quad_a16_t operator-(const Quad_a16_t &b) {
    _Quad lhs = (*this).q;
    _Quad rhs = b.q;
    return (Quad_a16_t)(lhs - rhs);
  }
  Quad_a16_t operator*(const Quad_a16_t &b) {
    _Quad lhs = (*this).q;
    _Quad rhs = b.q;
    return (Quad_a16_t)(lhs * rhs);
  }

  Quad_a16_t operator/(const Quad_a16_t &b) {
    _Quad lhs = (*this).q;
    _Quad rhs = b.q;
    return (Quad_a16_t)(lhs / rhs);
  }
};

struct KMP_DO_ALIGN(16) kmp_cmplx128_a16_t {
  kmp_cmplx128 q;

  kmp_cmplx128_a16_t() : q() {}

  kmp_cmplx128_a16_t(const kmp_cmplx128 &c128) : q(c128) {}

  kmp_cmplx128_a16_t operator+(const kmp_cmplx128_a16_t &b) {
    kmp_cmplx128 lhs = (*this).q;
    kmp_cmplx128 rhs = b.q;
    return (kmp_cmplx128_a16_t)(lhs + rhs);
  }
  kmp_cmplx128_a16_t operator-(const kmp_cmplx128_a16_t &b) {
    kmp_cmplx128 lhs = (*this).q;
    kmp_cmplx128 rhs = b.q;
    return (kmp_cmplx128_a16_t)(lhs - rhs);
  }
  kmp_cmplx128_a16_t operator*(const kmp_cmplx128_a16_t &b) {
    kmp_cmplx128 lhs = (*this).q;
    kmp_cmplx128 rhs = b.q;
    return (kmp_cmplx128_a16_t)(lhs * rhs);
  }

  kmp_cmplx128_a16_t operator/(const kmp_cmplx128_a16_t &b) {
    kmp_cmplx128 lhs = (*this).q;
    kmp_cmplx128 rhs = b.q;
    return (kmp_cmplx128_a16_t)(lhs / rhs);
  }
};

#endif

#if (KMP_ARCH_X86)
#define QUAD_LEGACY Quad_a4_t
#define CPLX128_LEG kmp_cmplx128_a4_t
#else
#define QUAD_LEGACY _Quad
#define CPLX128_LEG kmp_cmplx128
#endif

#ifdef __cplusplus
extern "C" {
#endif

extern int __kmp_atomic_mode;

// Atomic locks can easily become contended, so we use queuing locks for them.
typedef kmp_queuing_lock_t kmp_atomic_lock_t;

static inline void __kmp_acquire_atomic_lock(kmp_atomic_lock_t *lck,
                                             kmp_int32 gtid) {
#if OMPT_SUPPORT && OMPT_OPTIONAL
  if (ompt_enabled.ompt_callback_mutex_acquire) {
    ompt_callbacks.ompt_callback(ompt_callback_mutex_acquire)(
        ompt_mutex_atomic, 0, kmp_mutex_impl_queuing, (ompt_wait_id_t)(uintptr_t)lck,
        OMPT_GET_RETURN_ADDRESS(0));
  }
#endif

  __kmp_acquire_queuing_lock(lck, gtid);

#if OMPT_SUPPORT && OMPT_OPTIONAL
  if (ompt_enabled.ompt_callback_mutex_acquired) {
    ompt_callbacks.ompt_callback(ompt_callback_mutex_acquired)(
        ompt_mutex_atomic, (ompt_wait_id_t)(uintptr_t)lck, OMPT_GET_RETURN_ADDRESS(0));
  }
#endif
}

static inline int __kmp_test_atomic_lock(kmp_atomic_lock_t *lck,
                                         kmp_int32 gtid) {
  return __kmp_test_queuing_lock(lck, gtid);
}

static inline void __kmp_release_atomic_lock(kmp_atomic_lock_t *lck,
                                             kmp_int32 gtid) {
  __kmp_release_queuing_lock(lck, gtid);
#if OMPT_SUPPORT && OMPT_OPTIONAL
  if (ompt_enabled.ompt_callback_mutex_released) {
    ompt_callbacks.ompt_callback(ompt_callback_mutex_released)(
        ompt_mutex_atomic, (ompt_wait_id_t)(uintptr_t)lck, OMPT_GET_RETURN_ADDRESS(0));
  }
#endif
}

static inline void __kmp_init_atomic_lock(kmp_atomic_lock_t *lck) {
  __kmp_init_queuing_lock(lck);
}

static inline void __kmp_destroy_atomic_lock(kmp_atomic_lock_t *lck) {
  __kmp_destroy_queuing_lock(lck);
}

// Global Locks
extern kmp_atomic_lock_t __kmp_atomic_lock; /* Control access to all user coded
                                               atomics in Gnu compat mode   */
extern kmp_atomic_lock_t __kmp_atomic_lock_1i; /* Control access to all user
                                                  coded atomics for 1-byte fixed
                                                  data types */
extern kmp_atomic_lock_t __kmp_atomic_lock_2i; /* Control access to all user
                                                  coded atomics for 2-byte fixed
                                                  data types */
extern kmp_atomic_lock_t __kmp_atomic_lock_4i; /* Control access to all user
                                                  coded atomics for 4-byte fixed
                                                  data types */
extern kmp_atomic_lock_t __kmp_atomic_lock_4r; /* Control access to all user
                                                  coded atomics for kmp_real32
                                                  data type    */
extern kmp_atomic_lock_t __kmp_atomic_lock_8i; /* Control access to all user
                                                  coded atomics for 8-byte fixed
                                                  data types */
extern kmp_atomic_lock_t __kmp_atomic_lock_8r; /* Control access to all user
                                                  coded atomics for kmp_real64
                                                  data type    */
extern kmp_atomic_lock_t
    __kmp_atomic_lock_8c; /* Control access to all user coded atomics for
                             complex byte data type  */
extern kmp_atomic_lock_t
    __kmp_atomic_lock_10r; /* Control access to all user coded atomics for long
                              double data type   */
extern kmp_atomic_lock_t __kmp_atomic_lock_16r; /* Control access to all user
                                                   coded atomics for _Quad data
                                                   type         */
extern kmp_atomic_lock_t __kmp_atomic_lock_16c; /* Control access to all user
                                                   coded atomics for double
                                                   complex data type*/
extern kmp_atomic_lock_t
    __kmp_atomic_lock_20c; /* Control access to all user coded atomics for long
                              double complex type*/
extern kmp_atomic_lock_t __kmp_atomic_lock_32c; /* Control access to all user
                                                   coded atomics for _Quad
                                                   complex data type */

//  Below routines for atomic UPDATE are listed

// 1-byte
void __kmpc_atomic_fixed1_add(ident_t *id_ref, int gtid, char *lhs, char rhs);
void __kmpc_atomic_fixed1_andb(ident_t *id_ref, int gtid, char *lhs, char rhs);
void __kmpc_atomic_fixed1_div(ident_t *id_ref, int gtid, char *lhs, char rhs);
void __kmpc_atomic_fixed1u_div(ident_t *id_ref, int gtid, unsigned char *lhs,
                               unsigned char rhs);
void __kmpc_atomic_fixed1_mul(ident_t *id_ref, int gtid, char *lhs, char rhs);
void __kmpc_atomic_fixed1_orb(ident_t *id_ref, int gtid, char *lhs, char rhs);
void __kmpc_atomic_fixed1_shl(ident_t *id_ref, int gtid, char *lhs, char rhs);
void __kmpc_atomic_fixed1_shr(ident_t *id_ref, int gtid, char *lhs, char rhs);
void __kmpc_atomic_fixed1u_shr(ident_t *id_ref, int gtid, unsigned char *lhs,
                               unsigned char rhs);
void __kmpc_atomic_fixed1_sub(ident_t *id_ref, int gtid, char *lhs, char rhs);
void __kmpc_atomic_fixed1_xor(ident_t *id_ref, int gtid, char *lhs, char rhs);
// 2-byte
void __kmpc_atomic_fixed2_add(ident_t *id_ref, int gtid, short *lhs, short rhs);
void __kmpc_atomic_fixed2_andb(ident_t *id_ref, int gtid, short *lhs,
                               short rhs);
void __kmpc_atomic_fixed2_div(ident_t *id_ref, int gtid, short *lhs, short rhs);
void __kmpc_atomic_fixed2u_div(ident_t *id_ref, int gtid, unsigned short *lhs,
                               unsigned short rhs);
void __kmpc_atomic_fixed2_mul(ident_t *id_ref, int gtid, short *lhs, short rhs);
void __kmpc_atomic_fixed2_orb(ident_t *id_ref, int gtid, short *lhs, short rhs);
void __kmpc_atomic_fixed2_shl(ident_t *id_ref, int gtid, short *lhs, short rhs);
void __kmpc_atomic_fixed2_shr(ident_t *id_ref, int gtid, short *lhs, short rhs);
void __kmpc_atomic_fixed2u_shr(ident_t *id_ref, int gtid, unsigned short *lhs,
                               unsigned short rhs);
void __kmpc_atomic_fixed2_sub(ident_t *id_ref, int gtid, short *lhs, short rhs);
void __kmpc_atomic_fixed2_xor(ident_t *id_ref, int gtid, short *lhs, short rhs);
// 4-byte add / sub fixed
void __kmpc_atomic_fixed4_add(ident_t *id_ref, int gtid, kmp_int32 *lhs,
                              kmp_int32 rhs);
void __kmpc_atomic_fixed4_sub(ident_t *id_ref, int gtid, kmp_int32 *lhs,
                              kmp_int32 rhs);
// 4-byte add / sub float
void __kmpc_atomic_float4_add(ident_t *id_ref, int gtid, kmp_real32 *lhs,
                              kmp_real32 rhs);
void __kmpc_atomic_float4_sub(ident_t *id_ref, int gtid, kmp_real32 *lhs,
                              kmp_real32 rhs);
// 8-byte add / sub fixed
void __kmpc_atomic_fixed8_add(ident_t *id_ref, int gtid, kmp_int64 *lhs,
                              kmp_int64 rhs);
void __kmpc_atomic_fixed8_sub(ident_t *id_ref, int gtid, kmp_int64 *lhs,
                              kmp_int64 rhs);
// 8-byte add / sub float
void __kmpc_atomic_float8_add(ident_t *id_ref, int gtid, kmp_real64 *lhs,
                              kmp_real64 rhs);
void __kmpc_atomic_float8_sub(ident_t *id_ref, int gtid, kmp_real64 *lhs,
                              kmp_real64 rhs);
// 4-byte fixed
void __kmpc_atomic_fixed4_andb(ident_t *id_ref, int gtid, kmp_int32 *lhs,
                               kmp_int32 rhs);
void __kmpc_atomic_fixed4_div(ident_t *id_ref, int gtid, kmp_int32 *lhs,
                              kmp_int32 rhs);
void __kmpc_atomic_fixed4u_div(ident_t *id_ref, int gtid, kmp_uint32 *lhs,
                               kmp_uint32 rhs);
void __kmpc_atomic_fixed4_mul(ident_t *id_ref, int gtid, kmp_int32 *lhs,
                              kmp_int32 rhs);
void __kmpc_atomic_fixed4_orb(ident_t *id_ref, int gtid, kmp_int32 *lhs,
                              kmp_int32 rhs);
void __kmpc_atomic_fixed4_shl(ident_t *id_ref, int gtid, kmp_int32 *lhs,
                              kmp_int32 rhs);
void __kmpc_atomic_fixed4_shr(ident_t *id_ref, int gtid, kmp_int32 *lhs,
                              kmp_int32 rhs);
void __kmpc_atomic_fixed4u_shr(ident_t *id_ref, int gtid, kmp_uint32 *lhs,
                               kmp_uint32 rhs);
void __kmpc_atomic_fixed4_xor(ident_t *id_ref, int gtid, kmp_int32 *lhs,
                              kmp_int32 rhs);
// 8-byte fixed
void __kmpc_atomic_fixed8_andb(ident_t *id_ref, int gtid, kmp_int64 *lhs,
                               kmp_int64 rhs);
void __kmpc_atomic_fixed8_div(ident_t *id_ref, int gtid, kmp_int64 *lhs,
                              kmp_int64 rhs);
void __kmpc_atomic_fixed8u_div(ident_t *id_ref, int gtid, kmp_uint64 *lhs,
                               kmp_uint64 rhs);
void __kmpc_atomic_fixed8_mul(ident_t *id_ref, int gtid, kmp_int64 *lhs,
                              kmp_int64 rhs);
void __kmpc_atomic_fixed8_orb(ident_t *id_ref, int gtid, kmp_int64 *lhs,
                              kmp_int64 rhs);
void __kmpc_atomic_fixed8_shl(ident_t *id_ref, int gtid, kmp_int64 *lhs,
                              kmp_int64 rhs);
void __kmpc_atomic_fixed8_shr(ident_t *id_ref, int gtid, kmp_int64 *lhs,
                              kmp_int64 rhs);
void __kmpc_atomic_fixed8u_shr(ident_t *id_ref, int gtid, kmp_uint64 *lhs,
                               kmp_uint64 rhs);
void __kmpc_atomic_fixed8_xor(ident_t *id_ref, int gtid, kmp_int64 *lhs,
                              kmp_int64 rhs);
// 4-byte float
void __kmpc_atomic_float4_div(ident_t *id_ref, int gtid, kmp_real32 *lhs,
                              kmp_real32 rhs);
void __kmpc_atomic_float4_mul(ident_t *id_ref, int gtid, kmp_real32 *lhs,
                              kmp_real32 rhs);
// 8-byte float
void __kmpc_atomic_float8_div(ident_t *id_ref, int gtid, kmp_real64 *lhs,
                              kmp_real64 rhs);
void __kmpc_atomic_float8_mul(ident_t *id_ref, int gtid, kmp_real64 *lhs,
                              kmp_real64 rhs);
// 1-, 2-, 4-, 8-byte logical (&&, ||)
void __kmpc_atomic_fixed1_andl(ident_t *id_ref, int gtid, char *lhs, char rhs);
void __kmpc_atomic_fixed1_orl(ident_t *id_ref, int gtid, char *lhs, char rhs);
void __kmpc_atomic_fixed2_andl(ident_t *id_ref, int gtid, short *lhs,
                               short rhs);
void __kmpc_atomic_fixed2_orl(ident_t *id_ref, int gtid, short *lhs, short rhs);
void __kmpc_atomic_fixed4_andl(ident_t *id_ref, int gtid, kmp_int32 *lhs,
                               kmp_int32 rhs);
void __kmpc_atomic_fixed4_orl(ident_t *id_ref, int gtid, kmp_int32 *lhs,
                              kmp_int32 rhs);
void __kmpc_atomic_fixed8_andl(ident_t *id_ref, int gtid, kmp_int64 *lhs,
                               kmp_int64 rhs);
void __kmpc_atomic_fixed8_orl(ident_t *id_ref, int gtid, kmp_int64 *lhs,
                              kmp_int64 rhs);
// MIN / MAX
void __kmpc_atomic_fixed1_max(ident_t *id_ref, int gtid, char *lhs, char rhs);
void __kmpc_atomic_fixed1_min(ident_t *id_ref, int gtid, char *lhs, char rhs);
void __kmpc_atomic_fixed2_max(ident_t *id_ref, int gtid, short *lhs, short rhs);
void __kmpc_atomic_fixed2_min(ident_t *id_ref, int gtid, short *lhs, short rhs);
void __kmpc_atomic_fixed4_max(ident_t *id_ref, int gtid, kmp_int32 *lhs,
                              kmp_int32 rhs);
void __kmpc_atomic_fixed4_min(ident_t *id_ref, int gtid, kmp_int32 *lhs,
                              kmp_int32 rhs);
void __kmpc_atomic_fixed8_max(ident_t *id_ref, int gtid, kmp_int64 *lhs,
                              kmp_int64 rhs);
void __kmpc_atomic_fixed8_min(ident_t *id_ref, int gtid, kmp_int64 *lhs,
                              kmp_int64 rhs);
void __kmpc_atomic_float4_max(ident_t *id_ref, int gtid, kmp_real32 *lhs,
                              kmp_real32 rhs);
void __kmpc_atomic_float4_min(ident_t *id_ref, int gtid, kmp_real32 *lhs,
                              kmp_real32 rhs);
void __kmpc_atomic_float8_max(ident_t *id_ref, int gtid, kmp_real64 *lhs,
                              kmp_real64 rhs);
void __kmpc_atomic_float8_min(ident_t *id_ref, int gtid, kmp_real64 *lhs,
                              kmp_real64 rhs);
#if KMP_HAVE_QUAD
void __kmpc_atomic_float16_max(ident_t *id_ref, int gtid, QUAD_LEGACY *lhs,
                               QUAD_LEGACY rhs);
void __kmpc_atomic_float16_min(ident_t *id_ref, int gtid, QUAD_LEGACY *lhs,
                               QUAD_LEGACY rhs);
#if (KMP_ARCH_X86)
// Routines with 16-byte arguments aligned to 16-byte boundary; IA-32
// architecture only
void __kmpc_atomic_float16_max_a16(ident_t *id_ref, int gtid, Quad_a16_t *lhs,
                                   Quad_a16_t rhs);
void __kmpc_atomic_float16_min_a16(ident_t *id_ref, int gtid, Quad_a16_t *lhs,
                                   Quad_a16_t rhs);
#endif
#endif
// .NEQV. (same as xor)
void __kmpc_atomic_fixed1_neqv(ident_t *id_ref, int gtid, char *lhs, char rhs);
void __kmpc_atomic_fixed2_neqv(ident_t *id_ref, int gtid, short *lhs,
                               short rhs);
void __kmpc_atomic_fixed4_neqv(ident_t *id_ref, int gtid, kmp_int32 *lhs,
                               kmp_int32 rhs);
void __kmpc_atomic_fixed8_neqv(ident_t *id_ref, int gtid, kmp_int64 *lhs,
                               kmp_int64 rhs);
// .EQV. (same as ~xor)
void __kmpc_atomic_fixed1_eqv(ident_t *id_ref, int gtid, char *lhs, char rhs);
void __kmpc_atomic_fixed2_eqv(ident_t *id_ref, int gtid, short *lhs, short rhs);
void __kmpc_atomic_fixed4_eqv(ident_t *id_ref, int gtid, kmp_int32 *lhs,
                              kmp_int32 rhs);
void __kmpc_atomic_fixed8_eqv(ident_t *id_ref, int gtid, kmp_int64 *lhs,
                              kmp_int64 rhs);
// long double type
void __kmpc_atomic_float10_add(ident_t *id_ref, int gtid, long double *lhs,
                               long double rhs);
void __kmpc_atomic_float10_sub(ident_t *id_ref, int gtid, long double *lhs,
                               long double rhs);
void __kmpc_atomic_float10_mul(ident_t *id_ref, int gtid, long double *lhs,
                               long double rhs);
void __kmpc_atomic_float10_div(ident_t *id_ref, int gtid, long double *lhs,
                               long double rhs);
// _Quad type
#if KMP_HAVE_QUAD
void __kmpc_atomic_float16_add(ident_t *id_ref, int gtid, QUAD_LEGACY *lhs,
                               QUAD_LEGACY rhs);
void __kmpc_atomic_float16_sub(ident_t *id_ref, int gtid, QUAD_LEGACY *lhs,
                               QUAD_LEGACY rhs);
void __kmpc_atomic_float16_mul(ident_t *id_ref, int gtid, QUAD_LEGACY *lhs,
                               QUAD_LEGACY rhs);
void __kmpc_atomic_float16_div(ident_t *id_ref, int gtid, QUAD_LEGACY *lhs,
                               QUAD_LEGACY rhs);
#if (KMP_ARCH_X86)
// Routines with 16-byte arguments aligned to 16-byte boundary
void __kmpc_atomic_float16_add_a16(ident_t *id_ref, int gtid, Quad_a16_t *lhs,
                                   Quad_a16_t rhs);
void __kmpc_atomic_float16_sub_a16(ident_t *id_ref, int gtid, Quad_a16_t *lhs,
                                   Quad_a16_t rhs);
void __kmpc_atomic_float16_mul_a16(ident_t *id_ref, int gtid, Quad_a16_t *lhs,
                                   Quad_a16_t rhs);
void __kmpc_atomic_float16_div_a16(ident_t *id_ref, int gtid, Quad_a16_t *lhs,
                                   Quad_a16_t rhs);
#endif
#endif
// routines for complex types
void __kmpc_atomic_cmplx4_add(ident_t *id_ref, int gtid, kmp_cmplx32 *lhs,
                              kmp_cmplx32 rhs);
void __kmpc_atomic_cmplx4_sub(ident_t *id_ref, int gtid, kmp_cmplx32 *lhs,
                              kmp_cmplx32 rhs);
void __kmpc_atomic_cmplx4_mul(ident_t *id_ref, int gtid, kmp_cmplx32 *lhs,
                              kmp_cmplx32 rhs);
void __kmpc_atomic_cmplx4_div(ident_t *id_ref, int gtid, kmp_cmplx32 *lhs,
                              kmp_cmplx32 rhs);
void __kmpc_atomic_cmplx8_add(ident_t *id_ref, int gtid, kmp_cmplx64 *lhs,
                              kmp_cmplx64 rhs);
void __kmpc_atomic_cmplx8_sub(ident_t *id_ref, int gtid, kmp_cmplx64 *lhs,
                              kmp_cmplx64 rhs);
void __kmpc_atomic_cmplx8_mul(ident_t *id_ref, int gtid, kmp_cmplx64 *lhs,
                              kmp_cmplx64 rhs);
void __kmpc_atomic_cmplx8_div(ident_t *id_ref, int gtid, kmp_cmplx64 *lhs,
                              kmp_cmplx64 rhs);
void __kmpc_atomic_cmplx10_add(ident_t *id_ref, int gtid, kmp_cmplx80 *lhs,
                               kmp_cmplx80 rhs);
void __kmpc_atomic_cmplx10_sub(ident_t *id_ref, int gtid, kmp_cmplx80 *lhs,
                               kmp_cmplx80 rhs);
void __kmpc_atomic_cmplx10_mul(ident_t *id_ref, int gtid, kmp_cmplx80 *lhs,
                               kmp_cmplx80 rhs);
void __kmpc_atomic_cmplx10_div(ident_t *id_ref, int gtid, kmp_cmplx80 *lhs,
                               kmp_cmplx80 rhs);
#if KMP_HAVE_QUAD
void __kmpc_atomic_cmplx16_add(ident_t *id_ref, int gtid, CPLX128_LEG *lhs,
                               CPLX128_LEG rhs);
void __kmpc_atomic_cmplx16_sub(ident_t *id_ref, int gtid, CPLX128_LEG *lhs,
                               CPLX128_LEG rhs);
void __kmpc_atomic_cmplx16_mul(ident_t *id_ref, int gtid, CPLX128_LEG *lhs,
                               CPLX128_LEG rhs);
void __kmpc_atomic_cmplx16_div(ident_t *id_ref, int gtid, CPLX128_LEG *lhs,
                               CPLX128_LEG rhs);
#if (KMP_ARCH_X86)
// Routines with 16-byte arguments aligned to 16-byte boundary
void __kmpc_atomic_cmplx16_add_a16(ident_t *id_ref, int gtid,
                                   kmp_cmplx128_a16_t *lhs,
                                   kmp_cmplx128_a16_t rhs);
void __kmpc_atomic_cmplx16_sub_a16(ident_t *id_ref, int gtid,
                                   kmp_cmplx128_a16_t *lhs,
                                   kmp_cmplx128_a16_t rhs);
void __kmpc_atomic_cmplx16_mul_a16(ident_t *id_ref, int gtid,
                                   kmp_cmplx128_a16_t *lhs,
                                   kmp_cmplx128_a16_t rhs);
void __kmpc_atomic_cmplx16_div_a16(ident_t *id_ref, int gtid,
                                   kmp_cmplx128_a16_t *lhs,
                                   kmp_cmplx128_a16_t rhs);
#endif
#endif

// OpenMP 4.0: x = expr binop x for non-commutative operations.
// Supported only on IA-32 architecture and Intel(R) 64
#if KMP_ARCH_X86 || KMP_ARCH_X86_64

void __kmpc_atomic_fixed1_sub_rev(ident_t *id_ref, int gtid, char *lhs,
                                  char rhs);
void __kmpc_atomic_fixed1_div_rev(ident_t *id_ref, int gtid, char *lhs,
                                  char rhs);
void __kmpc_atomic_fixed1u_div_rev(ident_t *id_ref, int gtid,
                                   unsigned char *lhs, unsigned char rhs);
void __kmpc_atomic_fixed1_shl_rev(ident_t *id_ref, int gtid, char *lhs,
                                  char rhs);
void __kmpc_atomic_fixed1_shr_rev(ident_t *id_ref, int gtid, char *lhs,
                                  char rhs);
void __kmpc_atomic_fixed1u_shr_rev(ident_t *id_ref, int gtid,
                                   unsigned char *lhs, unsigned char rhs);
void __kmpc_atomic_fixed2_sub_rev(ident_t *id_ref, int gtid, short *lhs,
                                  short rhs);
void __kmpc_atomic_fixed2_div_rev(ident_t *id_ref, int gtid, short *lhs,
                                  short rhs);
void __kmpc_atomic_fixed2u_div_rev(ident_t *id_ref, int gtid,
                                   unsigned short *lhs, unsigned short rhs);
void __kmpc_atomic_fixed2_shl_rev(ident_t *id_ref, int gtid, short *lhs,
                                  short rhs);
void __kmpc_atomic_fixed2_shr_rev(ident_t *id_ref, int gtid, short *lhs,
                                  short rhs);
void __kmpc_atomic_fixed2u_shr_rev(ident_t *id_ref, int gtid,
                                   unsigned short *lhs, unsigned short rhs);
void __kmpc_atomic_fixed4_sub_rev(ident_t *id_ref, int gtid, kmp_int32 *lhs,
                                  kmp_int32 rhs);
void __kmpc_atomic_fixed4_div_rev(ident_t *id_ref, int gtid, kmp_int32 *lhs,
                                  kmp_int32 rhs);
void __kmpc_atomic_fixed4u_div_rev(ident_t *id_ref, int gtid, kmp_uint32 *lhs,
                                   kmp_uint32 rhs);
void __kmpc_atomic_fixed4_shl_rev(ident_t *id_ref, int gtid, kmp_int32 *lhs,
                                  kmp_int32 rhs);
void __kmpc_atomic_fixed4_shr_rev(ident_t *id_ref, int gtid, kmp_int32 *lhs,
                                  kmp_int32 rhs);
void __kmpc_atomic_fixed4u_shr_rev(ident_t *id_ref, int gtid, kmp_uint32 *lhs,
                                   kmp_uint32 rhs);
void __kmpc_atomic_fixed8_sub_rev(ident_t *id_ref, int gtid, kmp_int64 *lhs,
                                  kmp_int64 rhs);
void __kmpc_atomic_fixed8_div_rev(ident_t *id_ref, int gtid, kmp_int64 *lhs,
                                  kmp_int64 rhs);
void __kmpc_atomic_fixed8u_div_rev(ident_t *id_ref, int gtid, kmp_uint64 *lhs,
                                   kmp_uint64 rhs);
void __kmpc_atomic_fixed8_shl_rev(ident_t *id_ref, int gtid, kmp_int64 *lhs,
                                  kmp_int64 rhs);
void __kmpc_atomic_fixed8_shr_rev(ident_t *id_ref, int gtid, kmp_int64 *lhs,
                                  kmp_int64 rhs);
void __kmpc_atomic_fixed8u_shr_rev(ident_t *id_ref, int gtid, kmp_uint64 *lhs,
                                   kmp_uint64 rhs);
void __kmpc_atomic_float4_sub_rev(ident_t *id_ref, int gtid, float *lhs,
                                  float rhs);
void __kmpc_atomic_float4_div_rev(ident_t *id_ref, int gtid, float *lhs,
                                  float rhs);
void __kmpc_atomic_float8_sub_rev(ident_t *id_ref, int gtid, double *lhs,
                                  double rhs);
void __kmpc_atomic_float8_div_rev(ident_t *id_ref, int gtid, double *lhs,
                                  double rhs);
void __kmpc_atomic_float10_sub_rev(ident_t *id_ref, int gtid, long double *lhs,
                                   long double rhs);
void __kmpc_atomic_float10_div_rev(ident_t *id_ref, int gtid, long double *lhs,
                                   long double rhs);
#if KMP_HAVE_QUAD
void __kmpc_atomic_float16_sub_rev(ident_t *id_ref, int gtid, QUAD_LEGACY *lhs,
                                   QUAD_LEGACY rhs);
void __kmpc_atomic_float16_div_rev(ident_t *id_ref, int gtid, QUAD_LEGACY *lhs,
                                   QUAD_LEGACY rhs);
#endif
void __kmpc_atomic_cmplx4_sub_rev(ident_t *id_ref, int gtid, kmp_cmplx32 *lhs,
                                  kmp_cmplx32 rhs);
void __kmpc_atomic_cmplx4_div_rev(ident_t *id_ref, int gtid, kmp_cmplx32 *lhs,
                                  kmp_cmplx32 rhs);
void __kmpc_atomic_cmplx8_sub_rev(ident_t *id_ref, int gtid, kmp_cmplx64 *lhs,
                                  kmp_cmplx64 rhs);
void __kmpc_atomic_cmplx8_div_rev(ident_t *id_ref, int gtid, kmp_cmplx64 *lhs,
                                  kmp_cmplx64 rhs);
void __kmpc_atomic_cmplx10_sub_rev(ident_t *id_ref, int gtid, kmp_cmplx80 *lhs,
                                   kmp_cmplx80 rhs);
void __kmpc_atomic_cmplx10_div_rev(ident_t *id_ref, int gtid, kmp_cmplx80 *lhs,
                                   kmp_cmplx80 rhs);
#if KMP_HAVE_QUAD
void __kmpc_atomic_cmplx16_sub_rev(ident_t *id_ref, int gtid, CPLX128_LEG *lhs,
                                   CPLX128_LEG rhs);
void __kmpc_atomic_cmplx16_div_rev(ident_t *id_ref, int gtid, CPLX128_LEG *lhs,
                                   CPLX128_LEG rhs);
#if (KMP_ARCH_X86)
// Routines with 16-byte arguments aligned to 16-byte boundary
void __kmpc_atomic_float16_sub_a16_rev(ident_t *id_ref, int gtid,
                                       Quad_a16_t *lhs, Quad_a16_t rhs);
void __kmpc_atomic_float16_div_a16_rev(ident_t *id_ref, int gtid,
                                       Quad_a16_t *lhs, Quad_a16_t rhs);
void __kmpc_atomic_cmplx16_sub_a16_rev(ident_t *id_ref, int gtid,
                                       kmp_cmplx128_a16_t *lhs,
                                       kmp_cmplx128_a16_t rhs);
void __kmpc_atomic_cmplx16_div_a16_rev(ident_t *id_ref, int gtid,
                                       kmp_cmplx128_a16_t *lhs,
                                       kmp_cmplx128_a16_t rhs);
#endif
#endif // KMP_HAVE_QUAD

#endif // KMP_ARCH_X86 || KMP_ARCH_X86_64

// routines for mixed types

// RHS=float8
void __kmpc_atomic_fixed1_mul_float8(ident_t *id_ref, int gtid, char *lhs,
                                     kmp_real64 rhs);
void __kmpc_atomic_fixed1_div_float8(ident_t *id_ref, int gtid, char *lhs,
                                     kmp_real64 rhs);
void __kmpc_atomic_fixed2_mul_float8(ident_t *id_ref, int gtid, short *lhs,
                                     kmp_real64 rhs);
void __kmpc_atomic_fixed2_div_float8(ident_t *id_ref, int gtid, short *lhs,
                                     kmp_real64 rhs);
void __kmpc_atomic_fixed4_mul_float8(ident_t *id_ref, int gtid, kmp_int32 *lhs,
                                     kmp_real64 rhs);
void __kmpc_atomic_fixed4_div_float8(ident_t *id_ref, int gtid, kmp_int32 *lhs,
                                     kmp_real64 rhs);
void __kmpc_atomic_fixed8_mul_float8(ident_t *id_ref, int gtid, kmp_int64 *lhs,
                                     kmp_real64 rhs);
void __kmpc_atomic_fixed8_div_float8(ident_t *id_ref, int gtid, kmp_int64 *lhs,
                                     kmp_real64 rhs);
void __kmpc_atomic_float4_add_float8(ident_t *id_ref, int gtid, kmp_real32 *lhs,
                                     kmp_real64 rhs);
void __kmpc_atomic_float4_sub_float8(ident_t *id_ref, int gtid, kmp_real32 *lhs,
                                     kmp_real64 rhs);
void __kmpc_atomic_float4_mul_float8(ident_t *id_ref, int gtid, kmp_real32 *lhs,
                                     kmp_real64 rhs);
void __kmpc_atomic_float4_div_float8(ident_t *id_ref, int gtid, kmp_real32 *lhs,
                                     kmp_real64 rhs);

// RHS=float16 (deprecated, to be removed when we are sure the compiler does not
// use them)
#if KMP_HAVE_QUAD
void __kmpc_atomic_fixed1_add_fp(ident_t *id_ref, int gtid, char *lhs,
                                 _Quad rhs);
void __kmpc_atomic_fixed1u_add_fp(ident_t *id_ref, int gtid, unsigned char *lhs,
                                  _Quad rhs);
void __kmpc_atomic_fixed1_sub_fp(ident_t *id_ref, int gtid, char *lhs,
                                 _Quad rhs);
void __kmpc_atomic_fixed1u_sub_fp(ident_t *id_ref, int gtid, unsigned char *lhs,
                                  _Quad rhs);
void __kmpc_atomic_fixed1_mul_fp(ident_t *id_ref, int gtid, char *lhs,
                                 _Quad rhs);
void __kmpc_atomic_fixed1u_mul_fp(ident_t *id_ref, int gtid, unsigned char *lhs,
                                  _Quad rhs);
void __kmpc_atomic_fixed1_div_fp(ident_t *id_ref, int gtid, char *lhs,
                                 _Quad rhs);
void __kmpc_atomic_fixed1u_div_fp(ident_t *id_ref, int gtid, unsigned char *lhs,
                                  _Quad rhs);

void __kmpc_atomic_fixed2_add_fp(ident_t *id_ref, int gtid, short *lhs,
                                 _Quad rhs);
void __kmpc_atomic_fixed2u_add_fp(ident_t *id_ref, int gtid,
                                  unsigned short *lhs, _Quad rhs);
void __kmpc_atomic_fixed2_sub_fp(ident_t *id_ref, int gtid, short *lhs,
                                 _Quad rhs);
void __kmpc_atomic_fixed2u_sub_fp(ident_t *id_ref, int gtid,
                                  unsigned short *lhs, _Quad rhs);
void __kmpc_atomic_fixed2_mul_fp(ident_t *id_ref, int gtid, short *lhs,
                                 _Quad rhs);
void __kmpc_atomic_fixed2u_mul_fp(ident_t *id_ref, int gtid,
                                  unsigned short *lhs, _Quad rhs);
void __kmpc_atomic_fixed2_div_fp(ident_t *id_ref, int gtid, short *lhs,
                                 _Quad rhs);
void __kmpc_atomic_fixed2u_div_fp(ident_t *id_ref, int gtid,
                                  unsigned short *lhs, _Quad rhs);

void __kmpc_atomic_fixed4_add_fp(ident_t *id_ref, int gtid, kmp_int32 *lhs,
                                 _Quad rhs);
void __kmpc_atomic_fixed4u_add_fp(ident_t *id_ref, int gtid, kmp_uint32 *lhs,
                                  _Quad rhs);
void __kmpc_atomic_fixed4_sub_fp(ident_t *id_ref, int gtid, kmp_int32 *lhs,
                                 _Quad rhs);
void __kmpc_atomic_fixed4u_sub_fp(ident_t *id_ref, int gtid, kmp_uint32 *lhs,
                                  _Quad rhs);
void __kmpc_atomic_fixed4_mul_fp(ident_t *id_ref, int gtid, kmp_int32 *lhs,
                                 _Quad rhs);
void __kmpc_atomic_fixed4u_mul_fp(ident_t *id_ref, int gtid, kmp_uint32 *lhs,
                                  _Quad rhs);
void __kmpc_atomic_fixed4_div_fp(ident_t *id_ref, int gtid, kmp_int32 *lhs,
                                 _Quad rhs);
void __kmpc_atomic_fixed4u_div_fp(ident_t *id_ref, int gtid, kmp_uint32 *lhs,
                                  _Quad rhs);

void __kmpc_atomic_fixed8_add_fp(ident_t *id_ref, int gtid, kmp_int64 *lhs,
                                 _Quad rhs);
void __kmpc_atomic_fixed8u_add_fp(ident_t *id_ref, int gtid, kmp_uint64 *lhs,
                                  _Quad rhs);
void __kmpc_atomic_fixed8_sub_fp(ident_t *id_ref, int gtid, kmp_int64 *lhs,
                                 _Quad rhs);
void __kmpc_atomic_fixed8u_sub_fp(ident_t *id_ref, int gtid, kmp_uint64 *lhs,
                                  _Quad rhs);
void __kmpc_atomic_fixed8_mul_fp(ident_t *id_ref, int gtid, kmp_int64 *lhs,
                                 _Quad rhs);
void __kmpc_atomic_fixed8u_mul_fp(ident_t *id_ref, int gtid, kmp_uint64 *lhs,
                                  _Quad rhs);
void __kmpc_atomic_fixed8_div_fp(ident_t *id_ref, int gtid, kmp_int64 *lhs,
                                 _Quad rhs);
void __kmpc_atomic_fixed8u_div_fp(ident_t *id_ref, int gtid, kmp_uint64 *lhs,
                                  _Quad rhs);

void __kmpc_atomic_float4_add_fp(ident_t *id_ref, int gtid, kmp_real32 *lhs,
                                 _Quad rhs);
void __kmpc_atomic_float4_sub_fp(ident_t *id_ref, int gtid, kmp_real32 *lhs,
                                 _Quad rhs);
void __kmpc_atomic_float4_mul_fp(ident_t *id_ref, int gtid, kmp_real32 *lhs,
                                 _Quad rhs);
void __kmpc_atomic_float4_div_fp(ident_t *id_ref, int gtid, kmp_real32 *lhs,
                                 _Quad rhs);

void __kmpc_atomic_float8_add_fp(ident_t *id_ref, int gtid, kmp_real64 *lhs,
                                 _Quad rhs);
void __kmpc_atomic_float8_sub_fp(ident_t *id_ref, int gtid, kmp_real64 *lhs,
                                 _Quad rhs);
void __kmpc_atomic_float8_mul_fp(ident_t *id_ref, int gtid, kmp_real64 *lhs,
                                 _Quad rhs);
void __kmpc_atomic_float8_div_fp(ident_t *id_ref, int gtid, kmp_real64 *lhs,
                                 _Quad rhs);

void __kmpc_atomic_float10_add_fp(ident_t *id_ref, int gtid, long double *lhs,
                                  _Quad rhs);
void __kmpc_atomic_float10_sub_fp(ident_t *id_ref, int gtid, long double *lhs,
                                  _Quad rhs);
void __kmpc_atomic_float10_mul_fp(ident_t *id_ref, int gtid, long double *lhs,
                                  _Quad rhs);
void __kmpc_atomic_float10_div_fp(ident_t *id_ref, int gtid, long double *lhs,
                                  _Quad rhs);

// Reverse operations
void __kmpc_atomic_fixed1_sub_rev_fp(ident_t *id_ref, int gtid, char *lhs,
                                     _Quad rhs);
void __kmpc_atomic_fixed1u_sub_rev_fp(ident_t *id_ref, int gtid,
                                      unsigned char *lhs, _Quad rhs);
void __kmpc_atomic_fixed1_div_rev_fp(ident_t *id_ref, int gtid, char *lhs,
                                     _Quad rhs);
void __kmpc_atomic_fixed1u_div_rev_fp(ident_t *id_ref, int gtid,
                                      unsigned char *lhs, _Quad rhs);
void __kmpc_atomic_fixed2_sub_rev_fp(ident_t *id_ref, int gtid, short *lhs,
                                     _Quad rhs);
void __kmpc_atomic_fixed2u_sub_rev_fp(ident_t *id_ref, int gtid,
                                      unsigned short *lhs, _Quad rhs);
void __kmpc_atomic_fixed2_div_rev_fp(ident_t *id_ref, int gtid, short *lhs,
                                     _Quad rhs);
void __kmpc_atomic_fixed2u_div_rev_fp(ident_t *id_ref, int gtid,
                                      unsigned short *lhs, _Quad rhs);
void __kmpc_atomic_fixed4_sub_rev_fp(ident_t *id_ref, int gtid, kmp_int32 *lhs,
                                     _Quad rhs);
void __kmpc_atomic_fixed4u_sub_rev_fp(ident_t *id_ref, int gtid,
                                      kmp_uint32 *lhs, _Quad rhs);
void __kmpc_atomic_fixed4_div_rev_fp(ident_t *id_ref, int gtid, kmp_int32 *lhs,
                                     _Quad rhs);
void __kmpc_atomic_fixed4u_div_rev_fp(ident_t *id_ref, int gtid,
                                      kmp_uint32 *lhs, _Quad rhs);
void __kmpc_atomic_fixed8_sub_rev_fp(ident_t *id_ref, int gtid, kmp_int64 *lhs,
                                     _Quad rhs);
void __kmpc_atomic_fixed8u_sub_rev_fp(ident_t *id_ref, int gtid,
                                      kmp_uint64 *lhs, _Quad rhs);
void __kmpc_atomic_fixed8_div_rev_fp(ident_t *id_ref, int gtid, kmp_int64 *lhs,
                                     _Quad rhs);
void __kmpc_atomic_fixed8u_div_rev_fp(ident_t *id_ref, int gtid,
                                      kmp_uint64 *lhs, _Quad rhs);
void __kmpc_atomic_float4_sub_rev_fp(ident_t *id_ref, int gtid, float *lhs,
                                     _Quad rhs);
void __kmpc_atomic_float4_div_rev_fp(ident_t *id_ref, int gtid, float *lhs,
                                     _Quad rhs);
void __kmpc_atomic_float8_sub_rev_fp(ident_t *id_ref, int gtid, double *lhs,
                                     _Quad rhs);
void __kmpc_atomic_float8_div_rev_fp(ident_t *id_ref, int gtid, double *lhs,
                                     _Quad rhs);
void __kmpc_atomic_float10_sub_rev_fp(ident_t *id_ref, int gtid,
                                      long double *lhs, _Quad rhs);
void __kmpc_atomic_float10_div_rev_fp(ident_t *id_ref, int gtid,
                                      long double *lhs, _Quad rhs);

#endif // KMP_HAVE_QUAD

// RHS=cmplx8
void __kmpc_atomic_cmplx4_add_cmplx8(ident_t *id_ref, int gtid,
                                     kmp_cmplx32 *lhs, kmp_cmplx64 rhs);
void __kmpc_atomic_cmplx4_sub_cmplx8(ident_t *id_ref, int gtid,
                                     kmp_cmplx32 *lhs, kmp_cmplx64 rhs);
void __kmpc_atomic_cmplx4_mul_cmplx8(ident_t *id_ref, int gtid,
                                     kmp_cmplx32 *lhs, kmp_cmplx64 rhs);
void __kmpc_atomic_cmplx4_div_cmplx8(ident_t *id_ref, int gtid,
                                     kmp_cmplx32 *lhs, kmp_cmplx64 rhs);

// generic atomic routines
void __kmpc_atomic_1(ident_t *id_ref, int gtid, void *lhs, void *rhs,
                     void (*f)(void *, void *, void *));
void __kmpc_atomic_2(ident_t *id_ref, int gtid, void *lhs, void *rhs,
                     void (*f)(void *, void *, void *));
void __kmpc_atomic_4(ident_t *id_ref, int gtid, void *lhs, void *rhs,
                     void (*f)(void *, void *, void *));
void __kmpc_atomic_8(ident_t *id_ref, int gtid, void *lhs, void *rhs,
                     void (*f)(void *, void *, void *));
void __kmpc_atomic_10(ident_t *id_ref, int gtid, void *lhs, void *rhs,
                      void (*f)(void *, void *, void *));
void __kmpc_atomic_16(ident_t *id_ref, int gtid, void *lhs, void *rhs,
                      void (*f)(void *, void *, void *));
void __kmpc_atomic_20(ident_t *id_ref, int gtid, void *lhs, void *rhs,
                      void (*f)(void *, void *, void *));
void __kmpc_atomic_32(ident_t *id_ref, int gtid, void *lhs, void *rhs,
                      void (*f)(void *, void *, void *));

// READ, WRITE, CAPTURE are supported only on IA-32 architecture and Intel(R) 64
#if KMP_ARCH_X86 || KMP_ARCH_X86_64

//  Below routines for atomic READ are listed
char __kmpc_atomic_fixed1_rd(ident_t *id_ref, int gtid, char *loc);
short __kmpc_atomic_fixed2_rd(ident_t *id_ref, int gtid, short *loc);
kmp_int32 __kmpc_atomic_fixed4_rd(ident_t *id_ref, int gtid, kmp_int32 *loc);
kmp_int64 __kmpc_atomic_fixed8_rd(ident_t *id_ref, int gtid, kmp_int64 *loc);
kmp_real32 __kmpc_atomic_float4_rd(ident_t *id_ref, int gtid, kmp_real32 *loc);
kmp_real64 __kmpc_atomic_float8_rd(ident_t *id_ref, int gtid, kmp_real64 *loc);
long double __kmpc_atomic_float10_rd(ident_t *id_ref, int gtid,
                                     long double *loc);
#if KMP_HAVE_QUAD
QUAD_LEGACY __kmpc_atomic_float16_rd(ident_t *id_ref, int gtid,
                                     QUAD_LEGACY *loc);
#endif
// Fix for CQ220361: cmplx4 READ will return void on Windows* OS; read value
// will be returned through an additional parameter
#if (KMP_OS_WINDOWS)
void __kmpc_atomic_cmplx4_rd(kmp_cmplx32 *out, ident_t *id_ref, int gtid,
                             kmp_cmplx32 *loc);
#else
kmp_cmplx32 __kmpc_atomic_cmplx4_rd(ident_t *id_ref, int gtid,
                                    kmp_cmplx32 *loc);
#endif
kmp_cmplx64 __kmpc_atomic_cmplx8_rd(ident_t *id_ref, int gtid,
                                    kmp_cmplx64 *loc);
kmp_cmplx80 __kmpc_atomic_cmplx10_rd(ident_t *id_ref, int gtid,
                                     kmp_cmplx80 *loc);
#if KMP_HAVE_QUAD
CPLX128_LEG __kmpc_atomic_cmplx16_rd(ident_t *id_ref, int gtid,
                                     CPLX128_LEG *loc);
#if (KMP_ARCH_X86)
// Routines with 16-byte arguments aligned to 16-byte boundary
Quad_a16_t __kmpc_atomic_float16_a16_rd(ident_t *id_ref, int gtid,
                                        Quad_a16_t *loc);
kmp_cmplx128_a16_t __kmpc_atomic_cmplx16_a16_rd(ident_t *id_ref, int gtid,
                                                kmp_cmplx128_a16_t *loc);
#endif
#endif

//  Below routines for atomic WRITE are listed
void __kmpc_atomic_fixed1_wr(ident_t *id_ref, int gtid, char *lhs, char rhs);
void __kmpc_atomic_fixed2_wr(ident_t *id_ref, int gtid, short *lhs, short rhs);
void __kmpc_atomic_fixed4_wr(ident_t *id_ref, int gtid, kmp_int32 *lhs,
                             kmp_int32 rhs);
void __kmpc_atomic_fixed8_wr(ident_t *id_ref, int gtid, kmp_int64 *lhs,
                             kmp_int64 rhs);
void __kmpc_atomic_float4_wr(ident_t *id_ref, int gtid, kmp_real32 *lhs,
                             kmp_real32 rhs);
void __kmpc_atomic_float8_wr(ident_t *id_ref, int gtid, kmp_real64 *lhs,
                             kmp_real64 rhs);
void __kmpc_atomic_float10_wr(ident_t *id_ref, int gtid, long double *lhs,
                              long double rhs);
#if KMP_HAVE_QUAD
void __kmpc_atomic_float16_wr(ident_t *id_ref, int gtid, QUAD_LEGACY *lhs,
                              QUAD_LEGACY rhs);
#endif
void __kmpc_atomic_cmplx4_wr(ident_t *id_ref, int gtid, kmp_cmplx32 *lhs,
                             kmp_cmplx32 rhs);
void __kmpc_atomic_cmplx8_wr(ident_t *id_ref, int gtid, kmp_cmplx64 *lhs,
                             kmp_cmplx64 rhs);
void __kmpc_atomic_cmplx10_wr(ident_t *id_ref, int gtid, kmp_cmplx80 *lhs,
                              kmp_cmplx80 rhs);
#if KMP_HAVE_QUAD
void __kmpc_atomic_cmplx16_wr(ident_t *id_ref, int gtid, CPLX128_LEG *lhs,
                              CPLX128_LEG rhs);
#if (KMP_ARCH_X86)
// Routines with 16-byte arguments aligned to 16-byte boundary
void __kmpc_atomic_float16_a16_wr(ident_t *id_ref, int gtid, Quad_a16_t *lhs,
                                  Quad_a16_t rhs);
void __kmpc_atomic_cmplx16_a16_wr(ident_t *id_ref, int gtid,
                                  kmp_cmplx128_a16_t *lhs,
                                  kmp_cmplx128_a16_t rhs);
#endif
#endif

//  Below routines for atomic CAPTURE are listed

// 1-byte
char __kmpc_atomic_fixed1_add_cpt(ident_t *id_ref, int gtid, char *lhs,
                                  char rhs, int flag);
char __kmpc_atomic_fixed1_andb_cpt(ident_t *id_ref, int gtid, char *lhs,
                                   char rhs, int flag);
char __kmpc_atomic_fixed1_div_cpt(ident_t *id_ref, int gtid, char *lhs,
                                  char rhs, int flag);
unsigned char __kmpc_atomic_fixed1u_div_cpt(ident_t *id_ref, int gtid,
                                            unsigned char *lhs,
                                            unsigned char rhs, int flag);
char __kmpc_atomic_fixed1_mul_cpt(ident_t *id_ref, int gtid, char *lhs,
                                  char rhs, int flag);
char __kmpc_atomic_fixed1_orb_cpt(ident_t *id_ref, int gtid, char *lhs,
                                  char rhs, int flag);
char __kmpc_atomic_fixed1_shl_cpt(ident_t *id_ref, int gtid, char *lhs,
                                  char rhs, int flag);
char __kmpc_atomic_fixed1_shr_cpt(ident_t *id_ref, int gtid, char *lhs,
                                  char rhs, int flag);
unsigned char __kmpc_atomic_fixed1u_shr_cpt(ident_t *id_ref, int gtid,
                                            unsigned char *lhs,
                                            unsigned char rhs, int flag);
char __kmpc_atomic_fixed1_sub_cpt(ident_t *id_ref, int gtid, char *lhs,
                                  char rhs, int flag);
char __kmpc_atomic_fixed1_xor_cpt(ident_t *id_ref, int gtid, char *lhs,
                                  char rhs, int flag);
// 2-byte
short __kmpc_atomic_fixed2_add_cpt(ident_t *id_ref, int gtid, short *lhs,
                                   short rhs, int flag);
short __kmpc_atomic_fixed2_andb_cpt(ident_t *id_ref, int gtid, short *lhs,
                                    short rhs, int flag);
short __kmpc_atomic_fixed2_div_cpt(ident_t *id_ref, int gtid, short *lhs,
                                   short rhs, int flag);
unsigned short __kmpc_atomic_fixed2u_div_cpt(ident_t *id_ref, int gtid,
                                             unsigned short *lhs,
                                             unsigned short rhs, int flag);
short __kmpc_atomic_fixed2_mul_cpt(ident_t *id_ref, int gtid, short *lhs,
                                   short rhs, int flag);
short __kmpc_atomic_fixed2_orb_cpt(ident_t *id_ref, int gtid, short *lhs,
                                   short rhs, int flag);
short __kmpc_atomic_fixed2_shl_cpt(ident_t *id_ref, int gtid, short *lhs,
                                   short rhs, int flag);
short __kmpc_atomic_fixed2_shr_cpt(ident_t *id_ref, int gtid, short *lhs,
                                   short rhs, int flag);
unsigned short __kmpc_atomic_fixed2u_shr_cpt(ident_t *id_ref, int gtid,
                                             unsigned short *lhs,
                                             unsigned short rhs, int flag);
short __kmpc_atomic_fixed2_sub_cpt(ident_t *id_ref, int gtid, short *lhs,
                                   short rhs, int flag);
short __kmpc_atomic_fixed2_xor_cpt(ident_t *id_ref, int gtid, short *lhs,
                                   short rhs, int flag);
// 4-byte add / sub fixed
kmp_int32 __kmpc_atomic_fixed4_add_cpt(ident_t *id_ref, int gtid,
                                       kmp_int32 *lhs, kmp_int32 rhs, int flag);
kmp_int32 __kmpc_atomic_fixed4_sub_cpt(ident_t *id_ref, int gtid,
                                       kmp_int32 *lhs, kmp_int32 rhs, int flag);
// 4-byte add / sub float
kmp_real32 __kmpc_atomic_float4_add_cpt(ident_t *id_ref, int gtid,
                                        kmp_real32 *lhs, kmp_real32 rhs,
                                        int flag);
kmp_real32 __kmpc_atomic_float4_sub_cpt(ident_t *id_ref, int gtid,
                                        kmp_real32 *lhs, kmp_real32 rhs,
                                        int flag);
// 8-byte add / sub fixed
kmp_int64 __kmpc_atomic_fixed8_add_cpt(ident_t *id_ref, int gtid,
                                       kmp_int64 *lhs, kmp_int64 rhs, int flag);
kmp_int64 __kmpc_atomic_fixed8_sub_cpt(ident_t *id_ref, int gtid,
                                       kmp_int64 *lhs, kmp_int64 rhs, int flag);
// 8-byte add / sub float
kmp_real64 __kmpc_atomic_float8_add_cpt(ident_t *id_ref, int gtid,
                                        kmp_real64 *lhs, kmp_real64 rhs,
                                        int flag);
kmp_real64 __kmpc_atomic_float8_sub_cpt(ident_t *id_ref, int gtid,
                                        kmp_real64 *lhs, kmp_real64 rhs,
                                        int flag);
// 4-byte fixed
kmp_int32 __kmpc_atomic_fixed4_andb_cpt(ident_t *id_ref, int gtid,
                                        kmp_int32 *lhs, kmp_int32 rhs,
                                        int flag);
kmp_int32 __kmpc_atomic_fixed4_div_cpt(ident_t *id_ref, int gtid,
                                       kmp_int32 *lhs, kmp_int32 rhs, int flag);
kmp_uint32 __kmpc_atomic_fixed4u_div_cpt(ident_t *id_ref, int gtid,
                                         kmp_uint32 *lhs, kmp_uint32 rhs,
                                         int flag);
kmp_int32 __kmpc_atomic_fixed4_mul_cpt(ident_t *id_ref, int gtid,
                                       kmp_int32 *lhs, kmp_int32 rhs, int flag);
kmp_int32 __kmpc_atomic_fixed4_orb_cpt(ident_t *id_ref, int gtid,
                                       kmp_int32 *lhs, kmp_int32 rhs, int flag);
kmp_int32 __kmpc_atomic_fixed4_shl_cpt(ident_t *id_ref, int gtid,
                                       kmp_int32 *lhs, kmp_int32 rhs, int flag);
kmp_int32 __kmpc_atomic_fixed4_shr_cpt(ident_t *id_ref, int gtid,
                                       kmp_int32 *lhs, kmp_int32 rhs, int flag);
kmp_uint32 __kmpc_atomic_fixed4u_shr_cpt(ident_t *id_ref, int gtid,
                                         kmp_uint32 *lhs, kmp_uint32 rhs,
                                         int flag);
kmp_int32 __kmpc_atomic_fixed4_xor_cpt(ident_t *id_ref, int gtid,
                                       kmp_int32 *lhs, kmp_int32 rhs, int flag);
// 8-byte fixed
kmp_int64 __kmpc_atomic_fixed8_andb_cpt(ident_t *id_ref, int gtid,
                                        kmp_int64 *lhs, kmp_int64 rhs,
                                        int flag);
kmp_int64 __kmpc_atomic_fixed8_div_cpt(ident_t *id_ref, int gtid,
                                       kmp_int64 *lhs, kmp_int64 rhs, int flag);
kmp_uint64 __kmpc_atomic_fixed8u_div_cpt(ident_t *id_ref, int gtid,
                                         kmp_uint64 *lhs, kmp_uint64 rhs,
                                         int flag);
kmp_int64 __kmpc_atomic_fixed8_mul_cpt(ident_t *id_ref, int gtid,
                                       kmp_int64 *lhs, kmp_int64 rhs, int flag);
kmp_int64 __kmpc_atomic_fixed8_orb_cpt(ident_t *id_ref, int gtid,
                                       kmp_int64 *lhs, kmp_int64 rhs, int flag);
kmp_int64 __kmpc_atomic_fixed8_shl_cpt(ident_t *id_ref, int gtid,
                                       kmp_int64 *lhs, kmp_int64 rhs, int flag);
kmp_int64 __kmpc_atomic_fixed8_shr_cpt(ident_t *id_ref, int gtid,
                                       kmp_int64 *lhs, kmp_int64 rhs, int flag);
kmp_uint64 __kmpc_atomic_fixed8u_shr_cpt(ident_t *id_ref, int gtid,
                                         kmp_uint64 *lhs, kmp_uint64 rhs,
                                         int flag);
kmp_int64 __kmpc_atomic_fixed8_xor_cpt(ident_t *id_ref, int gtid,
                                       kmp_int64 *lhs, kmp_int64 rhs, int flag);
// 4-byte float
kmp_real32 __kmpc_atomic_float4_div_cpt(ident_t *id_ref, int gtid,
                                        kmp_real32 *lhs, kmp_real32 rhs,
                                        int flag);
kmp_real32 __kmpc_atomic_float4_mul_cpt(ident_t *id_ref, int gtid,
                                        kmp_real32 *lhs, kmp_real32 rhs,
                                        int flag);
// 8-byte float
kmp_real64 __kmpc_atomic_float8_div_cpt(ident_t *id_ref, int gtid,
                                        kmp_real64 *lhs, kmp_real64 rhs,
                                        int flag);
kmp_real64 __kmpc_atomic_float8_mul_cpt(ident_t *id_ref, int gtid,
                                        kmp_real64 *lhs, kmp_real64 rhs,
                                        int flag);
// 1-, 2-, 4-, 8-byte logical (&&, ||)
char __kmpc_atomic_fixed1_andl_cpt(ident_t *id_ref, int gtid, char *lhs,
                                   char rhs, int flag);
char __kmpc_atomic_fixed1_orl_cpt(ident_t *id_ref, int gtid, char *lhs,
                                  char rhs, int flag);
short __kmpc_atomic_fixed2_andl_cpt(ident_t *id_ref, int gtid, short *lhs,
                                    short rhs, int flag);
short __kmpc_atomic_fixed2_orl_cpt(ident_t *id_ref, int gtid, short *lhs,
                                   short rhs, int flag);
kmp_int32 __kmpc_atomic_fixed4_andl_cpt(ident_t *id_ref, int gtid,
                                        kmp_int32 *lhs, kmp_int32 rhs,
                                        int flag);
kmp_int32 __kmpc_atomic_fixed4_orl_cpt(ident_t *id_ref, int gtid,
                                       kmp_int32 *lhs, kmp_int32 rhs, int flag);
kmp_int64 __kmpc_atomic_fixed8_andl_cpt(ident_t *id_ref, int gtid,
                                        kmp_int64 *lhs, kmp_int64 rhs,
                                        int flag);
kmp_int64 __kmpc_atomic_fixed8_orl_cpt(ident_t *id_ref, int gtid,
                                       kmp_int64 *lhs, kmp_int64 rhs, int flag);
// MIN / MAX
char __kmpc_atomic_fixed1_max_cpt(ident_t *id_ref, int gtid, char *lhs,
                                  char rhs, int flag);
char __kmpc_atomic_fixed1_min_cpt(ident_t *id_ref, int gtid, char *lhs,
                                  char rhs, int flag);
short __kmpc_atomic_fixed2_max_cpt(ident_t *id_ref, int gtid, short *lhs,
                                   short rhs, int flag);
short __kmpc_atomic_fixed2_min_cpt(ident_t *id_ref, int gtid, short *lhs,
                                   short rhs, int flag);
kmp_int32 __kmpc_atomic_fixed4_max_cpt(ident_t *id_ref, int gtid,
                                       kmp_int32 *lhs, kmp_int32 rhs, int flag);
kmp_int32 __kmpc_atomic_fixed4_min_cpt(ident_t *id_ref, int gtid,
                                       kmp_int32 *lhs, kmp_int32 rhs, int flag);
kmp_int64 __kmpc_atomic_fixed8_max_cpt(ident_t *id_ref, int gtid,
                                       kmp_int64 *lhs, kmp_int64 rhs, int flag);
kmp_int64 __kmpc_atomic_fixed8_min_cpt(ident_t *id_ref, int gtid,
                                       kmp_int64 *lhs, kmp_int64 rhs, int flag);
kmp_real32 __kmpc_atomic_float4_max_cpt(ident_t *id_ref, int gtid,
                                        kmp_real32 *lhs, kmp_real32 rhs,
                                        int flag);
kmp_real32 __kmpc_atomic_float4_min_cpt(ident_t *id_ref, int gtid,
                                        kmp_real32 *lhs, kmp_real32 rhs,
                                        int flag);
kmp_real64 __kmpc_atomic_float8_max_cpt(ident_t *id_ref, int gtid,
                                        kmp_real64 *lhs, kmp_real64 rhs,
                                        int flag);
kmp_real64 __kmpc_atomic_float8_min_cpt(ident_t *id_ref, int gtid,
                                        kmp_real64 *lhs, kmp_real64 rhs,
                                        int flag);
#if KMP_HAVE_QUAD
QUAD_LEGACY __kmpc_atomic_float16_max_cpt(ident_t *id_ref, int gtid,
                                          QUAD_LEGACY *lhs, QUAD_LEGACY rhs,
                                          int flag);
QUAD_LEGACY __kmpc_atomic_float16_min_cpt(ident_t *id_ref, int gtid,
                                          QUAD_LEGACY *lhs, QUAD_LEGACY rhs,
                                          int flag);
#endif
// .NEQV. (same as xor)
char __kmpc_atomic_fixed1_neqv_cpt(ident_t *id_ref, int gtid, char *lhs,
                                   char rhs, int flag);
short __kmpc_atomic_fixed2_neqv_cpt(ident_t *id_ref, int gtid, short *lhs,
                                    short rhs, int flag);
kmp_int32 __kmpc_atomic_fixed4_neqv_cpt(ident_t *id_ref, int gtid,
                                        kmp_int32 *lhs, kmp_int32 rhs,
                                        int flag);
kmp_int64 __kmpc_atomic_fixed8_neqv_cpt(ident_t *id_ref, int gtid,
                                        kmp_int64 *lhs, kmp_int64 rhs,
                                        int flag);
// .EQV. (same as ~xor)
char __kmpc_atomic_fixed1_eqv_cpt(ident_t *id_ref, int gtid, char *lhs,
                                  char rhs, int flag);
short __kmpc_atomic_fixed2_eqv_cpt(ident_t *id_ref, int gtid, short *lhs,
                                   short rhs, int flag);
kmp_int32 __kmpc_atomic_fixed4_eqv_cpt(ident_t *id_ref, int gtid,
                                       kmp_int32 *lhs, kmp_int32 rhs, int flag);
kmp_int64 __kmpc_atomic_fixed8_eqv_cpt(ident_t *id_ref, int gtid,
                                       kmp_int64 *lhs, kmp_int64 rhs, int flag);
// long double type
long double __kmpc_atomic_float10_add_cpt(ident_t *id_ref, int gtid,
                                          long double *lhs, long double rhs,
                                          int flag);
long double __kmpc_atomic_float10_sub_cpt(ident_t *id_ref, int gtid,
                                          long double *lhs, long double rhs,
                                          int flag);
long double __kmpc_atomic_float10_mul_cpt(ident_t *id_ref, int gtid,
                                          long double *lhs, long double rhs,
                                          int flag);
long double __kmpc_atomic_float10_div_cpt(ident_t *id_ref, int gtid,
                                          long double *lhs, long double rhs,
                                          int flag);
#if KMP_HAVE_QUAD
// _Quad type
QUAD_LEGACY __kmpc_atomic_float16_add_cpt(ident_t *id_ref, int gtid,
                                          QUAD_LEGACY *lhs, QUAD_LEGACY rhs,
                                          int flag);
QUAD_LEGACY __kmpc_atomic_float16_sub_cpt(ident_t *id_ref, int gtid,
                                          QUAD_LEGACY *lhs, QUAD_LEGACY rhs,
                                          int flag);
QUAD_LEGACY __kmpc_atomic_float16_mul_cpt(ident_t *id_ref, int gtid,
                                          QUAD_LEGACY *lhs, QUAD_LEGACY rhs,
                                          int flag);
QUAD_LEGACY __kmpc_atomic_float16_div_cpt(ident_t *id_ref, int gtid,
                                          QUAD_LEGACY *lhs, QUAD_LEGACY rhs,
                                          int flag);
#endif
// routines for complex types
// Workaround for cmplx4 routines - return void; captured value is returned via
// the argument
void __kmpc_atomic_cmplx4_add_cpt(ident_t *id_ref, int gtid, kmp_cmplx32 *lhs,
                                  kmp_cmplx32 rhs, kmp_cmplx32 *out, int flag);
void __kmpc_atomic_cmplx4_sub_cpt(ident_t *id_ref, int gtid, kmp_cmplx32 *lhs,
                                  kmp_cmplx32 rhs, kmp_cmplx32 *out, int flag);
void __kmpc_atomic_cmplx4_mul_cpt(ident_t *id_ref, int gtid, kmp_cmplx32 *lhs,
                                  kmp_cmplx32 rhs, kmp_cmplx32 *out, int flag);
void __kmpc_atomic_cmplx4_div_cpt(ident_t *id_ref, int gtid, kmp_cmplx32 *lhs,
                                  kmp_cmplx32 rhs, kmp_cmplx32 *out, int flag);

kmp_cmplx64 __kmpc_atomic_cmplx8_add_cpt(ident_t *id_ref, int gtid,
                                         kmp_cmplx64 *lhs, kmp_cmplx64 rhs,
                                         int flag);
kmp_cmplx64 __kmpc_atomic_cmplx8_sub_cpt(ident_t *id_ref, int gtid,
                                         kmp_cmplx64 *lhs, kmp_cmplx64 rhs,
                                         int flag);
kmp_cmplx64 __kmpc_atomic_cmplx8_mul_cpt(ident_t *id_ref, int gtid,
                                         kmp_cmplx64 *lhs, kmp_cmplx64 rhs,
                                         int flag);
kmp_cmplx64 __kmpc_atomic_cmplx8_div_cpt(ident_t *id_ref, int gtid,
                                         kmp_cmplx64 *lhs, kmp_cmplx64 rhs,
                                         int flag);
kmp_cmplx80 __kmpc_atomic_cmplx10_add_cpt(ident_t *id_ref, int gtid,
                                          kmp_cmplx80 *lhs, kmp_cmplx80 rhs,
                                          int flag);
kmp_cmplx80 __kmpc_atomic_cmplx10_sub_cpt(ident_t *id_ref, int gtid,
                                          kmp_cmplx80 *lhs, kmp_cmplx80 rhs,
                                          int flag);
kmp_cmplx80 __kmpc_atomic_cmplx10_mul_cpt(ident_t *id_ref, int gtid,
                                          kmp_cmplx80 *lhs, kmp_cmplx80 rhs,
                                          int flag);
kmp_cmplx80 __kmpc_atomic_cmplx10_div_cpt(ident_t *id_ref, int gtid,
                                          kmp_cmplx80 *lhs, kmp_cmplx80 rhs,
                                          int flag);
#if KMP_HAVE_QUAD
CPLX128_LEG __kmpc_atomic_cmplx16_add_cpt(ident_t *id_ref, int gtid,
                                          CPLX128_LEG *lhs, CPLX128_LEG rhs,
                                          int flag);
CPLX128_LEG __kmpc_atomic_cmplx16_sub_cpt(ident_t *id_ref, int gtid,
                                          CPLX128_LEG *lhs, CPLX128_LEG rhs,
                                          int flag);
CPLX128_LEG __kmpc_atomic_cmplx16_mul_cpt(ident_t *id_ref, int gtid,
                                          CPLX128_LEG *lhs, CPLX128_LEG rhs,
                                          int flag);
CPLX128_LEG __kmpc_atomic_cmplx16_div_cpt(ident_t *id_ref, int gtid,
                                          CPLX128_LEG *lhs, CPLX128_LEG rhs,
                                          int flag);
#if (KMP_ARCH_X86)
// Routines with 16-byte arguments aligned to 16-byte boundary
Quad_a16_t __kmpc_atomic_float16_add_a16_cpt(ident_t *id_ref, int gtid,
                                             Quad_a16_t *lhs, Quad_a16_t rhs,
                                             int flag);
Quad_a16_t __kmpc_atomic_float16_sub_a16_cpt(ident_t *id_ref, int gtid,
                                             Quad_a16_t *lhs, Quad_a16_t rhs,
                                             int flag);
Quad_a16_t __kmpc_atomic_float16_mul_a16_cpt(ident_t *id_ref, int gtid,
                                             Quad_a16_t *lhs, Quad_a16_t rhs,
                                             int flag);
Quad_a16_t __kmpc_atomic_float16_div_a16_cpt(ident_t *id_ref, int gtid,
                                             Quad_a16_t *lhs, Quad_a16_t rhs,
                                             int flag);
Quad_a16_t __kmpc_atomic_float16_max_a16_cpt(ident_t *id_ref, int gtid,
                                             Quad_a16_t *lhs, Quad_a16_t rhs,
                                             int flag);
Quad_a16_t __kmpc_atomic_float16_min_a16_cpt(ident_t *id_ref, int gtid,
                                             Quad_a16_t *lhs, Quad_a16_t rhs,
                                             int flag);
kmp_cmplx128_a16_t __kmpc_atomic_cmplx16_add_a16_cpt(ident_t *id_ref, int gtid,
                                                     kmp_cmplx128_a16_t *lhs,
                                                     kmp_cmplx128_a16_t rhs,
                                                     int flag);
kmp_cmplx128_a16_t __kmpc_atomic_cmplx16_sub_a16_cpt(ident_t *id_ref, int gtid,
                                                     kmp_cmplx128_a16_t *lhs,
                                                     kmp_cmplx128_a16_t rhs,
                                                     int flag);
kmp_cmplx128_a16_t __kmpc_atomic_cmplx16_mul_a16_cpt(ident_t *id_ref, int gtid,
                                                     kmp_cmplx128_a16_t *lhs,
                                                     kmp_cmplx128_a16_t rhs,
                                                     int flag);
kmp_cmplx128_a16_t __kmpc_atomic_cmplx16_div_a16_cpt(ident_t *id_ref, int gtid,
                                                     kmp_cmplx128_a16_t *lhs,
                                                     kmp_cmplx128_a16_t rhs,
                                                     int flag);
#endif
#endif

void __kmpc_atomic_start(void);
void __kmpc_atomic_end(void);

// OpenMP 4.0: v = x = expr binop x; { v = x; x = expr binop x; } { x = expr
// binop x; v = x; }  for non-commutative operations.

char __kmpc_atomic_fixed1_sub_cpt_rev(ident_t *id_ref, int gtid, char *lhs,
                                      char rhs, int flag);
char __kmpc_atomic_fixed1_div_cpt_rev(ident_t *id_ref, int gtid, char *lhs,
                                      char rhs, int flag);
unsigned char __kmpc_atomic_fixed1u_div_cpt_rev(ident_t *id_ref, int gtid,
                                                unsigned char *lhs,
                                                unsigned char rhs, int flag);
char __kmpc_atomic_fixed1_shl_cpt_rev(ident_t *id_ref, int gtid, char *lhs,
                                      char rhs, int flag);
char __kmpc_atomic_fixed1_shr_cpt_rev(ident_t *id_ref, int gtid, char *lhs,
                                      char rhs, int flag);
unsigned char __kmpc_atomic_fixed1u_shr_cpt_rev(ident_t *id_ref, int gtid,
                                                unsigned char *lhs,
                                                unsigned char rhs, int flag);
short __kmpc_atomic_fixed2_sub_cpt_rev(ident_t *id_ref, int gtid, short *lhs,
                                       short rhs, int flag);
short __kmpc_atomic_fixed2_div_cpt_rev(ident_t *id_ref, int gtid, short *lhs,
                                       short rhs, int flag);
unsigned short __kmpc_atomic_fixed2u_div_cpt_rev(ident_t *id_ref, int gtid,
                                                 unsigned short *lhs,
                                                 unsigned short rhs, int flag);
short __kmpc_atomic_fixed2_shl_cpt_rev(ident_t *id_ref, int gtid, short *lhs,
                                       short rhs, int flag);
short __kmpc_atomic_fixed2_shr_cpt_rev(ident_t *id_ref, int gtid, short *lhs,
                                       short rhs, int flag);
unsigned short __kmpc_atomic_fixed2u_shr_cpt_rev(ident_t *id_ref, int gtid,
                                                 unsigned short *lhs,
                                                 unsigned short rhs, int flag);
kmp_int32 __kmpc_atomic_fixed4_sub_cpt_rev(ident_t *id_ref, int gtid,
                                           kmp_int32 *lhs, kmp_int32 rhs,
                                           int flag);
kmp_int32 __kmpc_atomic_fixed4_div_cpt_rev(ident_t *id_ref, int gtid,
                                           kmp_int32 *lhs, kmp_int32 rhs,
                                           int flag);
kmp_uint32 __kmpc_atomic_fixed4u_div_cpt_rev(ident_t *id_ref, int gtid,
                                             kmp_uint32 *lhs, kmp_uint32 rhs,
                                             int flag);
kmp_int32 __kmpc_atomic_fixed4_shl_cpt_rev(ident_t *id_ref, int gtid,
                                           kmp_int32 *lhs, kmp_int32 rhs,
                                           int flag);
kmp_int32 __kmpc_atomic_fixed4_shr_cpt_rev(ident_t *id_ref, int gtid,
                                           kmp_int32 *lhs, kmp_int32 rhs,
                                           int flag);
kmp_uint32 __kmpc_atomic_fixed4u_shr_cpt_rev(ident_t *id_ref, int gtid,
                                             kmp_uint32 *lhs, kmp_uint32 rhs,
                                             int flag);
kmp_int64 __kmpc_atomic_fixed8_sub_cpt_rev(ident_t *id_ref, int gtid,
                                           kmp_int64 *lhs, kmp_int64 rhs,
                                           int flag);
kmp_int64 __kmpc_atomic_fixed8_div_cpt_rev(ident_t *id_ref, int gtid,
                                           kmp_int64 *lhs, kmp_int64 rhs,
                                           int flag);
kmp_uint64 __kmpc_atomic_fixed8u_div_cpt_rev(ident_t *id_ref, int gtid,
                                             kmp_uint64 *lhs, kmp_uint64 rhs,
                                             int flag);
kmp_int64 __kmpc_atomic_fixed8_shl_cpt_rev(ident_t *id_ref, int gtid,
                                           kmp_int64 *lhs, kmp_int64 rhs,
                                           int flag);
kmp_int64 __kmpc_atomic_fixed8_shr_cpt_rev(ident_t *id_ref, int gtid,
                                           kmp_int64 *lhs, kmp_int64 rhs,
                                           int flag);
kmp_uint64 __kmpc_atomic_fixed8u_shr_cpt_rev(ident_t *id_ref, int gtid,
                                             kmp_uint64 *lhs, kmp_uint64 rhs,
                                             int flag);
float __kmpc_atomic_float4_sub_cpt_rev(ident_t *id_ref, int gtid, float *lhs,
                                       float rhs, int flag);
float __kmpc_atomic_float4_div_cpt_rev(ident_t *id_ref, int gtid, float *lhs,
                                       float rhs, int flag);
double __kmpc_atomic_float8_sub_cpt_rev(ident_t *id_ref, int gtid, double *lhs,
                                        double rhs, int flag);
double __kmpc_atomic_float8_div_cpt_rev(ident_t *id_ref, int gtid, double *lhs,
                                        double rhs, int flag);
long double __kmpc_atomic_float10_sub_cpt_rev(ident_t *id_ref, int gtid,
                                              long double *lhs, long double rhs,
                                              int flag);
long double __kmpc_atomic_float10_div_cpt_rev(ident_t *id_ref, int gtid,
                                              long double *lhs, long double rhs,
                                              int flag);
#if KMP_HAVE_QUAD
QUAD_LEGACY __kmpc_atomic_float16_sub_cpt_rev(ident_t *id_ref, int gtid,
                                              QUAD_LEGACY *lhs, QUAD_LEGACY rhs,
                                              int flag);
QUAD_LEGACY __kmpc_atomic_float16_div_cpt_rev(ident_t *id_ref, int gtid,
                                              QUAD_LEGACY *lhs, QUAD_LEGACY rhs,
                                              int flag);
#endif
// Workaround for cmplx4 routines - return void; captured value is returned via
// the argument
void __kmpc_atomic_cmplx4_sub_cpt_rev(ident_t *id_ref, int gtid,
                                      kmp_cmplx32 *lhs, kmp_cmplx32 rhs,
                                      kmp_cmplx32 *out, int flag);
void __kmpc_atomic_cmplx4_div_cpt_rev(ident_t *id_ref, int gtid,
                                      kmp_cmplx32 *lhs, kmp_cmplx32 rhs,
                                      kmp_cmplx32 *out, int flag);
kmp_cmplx64 __kmpc_atomic_cmplx8_sub_cpt_rev(ident_t *id_ref, int gtid,
                                             kmp_cmplx64 *lhs, kmp_cmplx64 rhs,
                                             int flag);
kmp_cmplx64 __kmpc_atomic_cmplx8_div_cpt_rev(ident_t *id_ref, int gtid,
                                             kmp_cmplx64 *lhs, kmp_cmplx64 rhs,
                                             int flag);
kmp_cmplx80 __kmpc_atomic_cmplx10_sub_cpt_rev(ident_t *id_ref, int gtid,
                                              kmp_cmplx80 *lhs, kmp_cmplx80 rhs,
                                              int flag);
kmp_cmplx80 __kmpc_atomic_cmplx10_div_cpt_rev(ident_t *id_ref, int gtid,
                                              kmp_cmplx80 *lhs, kmp_cmplx80 rhs,
                                              int flag);
#if KMP_HAVE_QUAD
CPLX128_LEG __kmpc_atomic_cmplx16_sub_cpt_rev(ident_t *id_ref, int gtid,
                                              CPLX128_LEG *lhs, CPLX128_LEG rhs,
                                              int flag);
CPLX128_LEG __kmpc_atomic_cmplx16_div_cpt_rev(ident_t *id_ref, int gtid,
                                              CPLX128_LEG *lhs, CPLX128_LEG rhs,
                                              int flag);
#if (KMP_ARCH_X86)
Quad_a16_t __kmpc_atomic_float16_sub_a16_cpt_rev(ident_t *id_ref, int gtid,
                                                 Quad_a16_t *lhs,
                                                 Quad_a16_t rhs, int flag);
Quad_a16_t __kmpc_atomic_float16_div_a16_cpt_rev(ident_t *id_ref, int gtid,
                                                 Quad_a16_t *lhs,
                                                 Quad_a16_t rhs, int flag);
kmp_cmplx128_a16_t
__kmpc_atomic_cmplx16_sub_a16_cpt_rev(ident_t *id_ref, int gtid,
                                      kmp_cmplx128_a16_t *lhs,
                                      kmp_cmplx128_a16_t rhs, int flag);
kmp_cmplx128_a16_t
__kmpc_atomic_cmplx16_div_a16_cpt_rev(ident_t *id_ref, int gtid,
                                      kmp_cmplx128_a16_t *lhs,
                                      kmp_cmplx128_a16_t rhs, int flag);
#endif
#endif

//   OpenMP 4.0 Capture-write (swap): {v = x; x = expr;}
char __kmpc_atomic_fixed1_swp(ident_t *id_ref, int gtid, char *lhs, char rhs);
short __kmpc_atomic_fixed2_swp(ident_t *id_ref, int gtid, short *lhs,
                               short rhs);
kmp_int32 __kmpc_atomic_fixed4_swp(ident_t *id_ref, int gtid, kmp_int32 *lhs,
                                   kmp_int32 rhs);
kmp_int64 __kmpc_atomic_fixed8_swp(ident_t *id_ref, int gtid, kmp_int64 *lhs,
                                   kmp_int64 rhs);
float __kmpc_atomic_float4_swp(ident_t *id_ref, int gtid, float *lhs,
                               float rhs);
double __kmpc_atomic_float8_swp(ident_t *id_ref, int gtid, double *lhs,
                                double rhs);
long double __kmpc_atomic_float10_swp(ident_t *id_ref, int gtid,
                                      long double *lhs, long double rhs);
#if KMP_HAVE_QUAD
QUAD_LEGACY __kmpc_atomic_float16_swp(ident_t *id_ref, int gtid,
                                      QUAD_LEGACY *lhs, QUAD_LEGACY rhs);
#endif
// !!! TODO: check if we need a workaround here
void __kmpc_atomic_cmplx4_swp(ident_t *id_ref, int gtid, kmp_cmplx32 *lhs,
                              kmp_cmplx32 rhs, kmp_cmplx32 *out);
// kmp_cmplx32   	__kmpc_atomic_cmplx4_swp(  ident_t *id_ref, int gtid,
// kmp_cmplx32 * lhs, kmp_cmplx32 rhs );

kmp_cmplx64 __kmpc_atomic_cmplx8_swp(ident_t *id_ref, int gtid,
                                     kmp_cmplx64 *lhs, kmp_cmplx64 rhs);
kmp_cmplx80 __kmpc_atomic_cmplx10_swp(ident_t *id_ref, int gtid,
                                      kmp_cmplx80 *lhs, kmp_cmplx80 rhs);
#if KMP_HAVE_QUAD
CPLX128_LEG __kmpc_atomic_cmplx16_swp(ident_t *id_ref, int gtid,
                                      CPLX128_LEG *lhs, CPLX128_LEG rhs);
#if (KMP_ARCH_X86)
Quad_a16_t __kmpc_atomic_float16_a16_swp(ident_t *id_ref, int gtid,
                                         Quad_a16_t *lhs, Quad_a16_t rhs);
kmp_cmplx128_a16_t __kmpc_atomic_cmplx16_a16_swp(ident_t *id_ref, int gtid,
                                                 kmp_cmplx128_a16_t *lhs,
                                                 kmp_cmplx128_a16_t rhs);
#endif
#endif

// Capture routines for mixed types (RHS=float16)
#if KMP_HAVE_QUAD

char __kmpc_atomic_fixed1_add_cpt_fp(ident_t *id_ref, int gtid, char *lhs,
                                     _Quad rhs, int flag);
char __kmpc_atomic_fixed1_sub_cpt_fp(ident_t *id_ref, int gtid, char *lhs,
                                     _Quad rhs, int flag);
char __kmpc_atomic_fixed1_mul_cpt_fp(ident_t *id_ref, int gtid, char *lhs,
                                     _Quad rhs, int flag);
char __kmpc_atomic_fixed1_div_cpt_fp(ident_t *id_ref, int gtid, char *lhs,
                                     _Quad rhs, int flag);
unsigned char __kmpc_atomic_fixed1u_add_cpt_fp(ident_t *id_ref, int gtid,
                                               unsigned char *lhs, _Quad rhs,
                                               int flag);
unsigned char __kmpc_atomic_fixed1u_sub_cpt_fp(ident_t *id_ref, int gtid,
                                               unsigned char *lhs, _Quad rhs,
                                               int flag);
unsigned char __kmpc_atomic_fixed1u_mul_cpt_fp(ident_t *id_ref, int gtid,
                                               unsigned char *lhs, _Quad rhs,
                                               int flag);
unsigned char __kmpc_atomic_fixed1u_div_cpt_fp(ident_t *id_ref, int gtid,
                                               unsigned char *lhs, _Quad rhs,
                                               int flag);

short __kmpc_atomic_fixed2_add_cpt_fp(ident_t *id_ref, int gtid, short *lhs,
                                      _Quad rhs, int flag);
short __kmpc_atomic_fixed2_sub_cpt_fp(ident_t *id_ref, int gtid, short *lhs,
                                      _Quad rhs, int flag);
short __kmpc_atomic_fixed2_mul_cpt_fp(ident_t *id_ref, int gtid, short *lhs,
                                      _Quad rhs, int flag);
short __kmpc_atomic_fixed2_div_cpt_fp(ident_t *id_ref, int gtid, short *lhs,
                                      _Quad rhs, int flag);
unsigned short __kmpc_atomic_fixed2u_add_cpt_fp(ident_t *id_ref, int gtid,
                                                unsigned short *lhs, _Quad rhs,
                                                int flag);
unsigned short __kmpc_atomic_fixed2u_sub_cpt_fp(ident_t *id_ref, int gtid,
                                                unsigned short *lhs, _Quad rhs,
                                                int flag);
unsigned short __kmpc_atomic_fixed2u_mul_cpt_fp(ident_t *id_ref, int gtid,
                                                unsigned short *lhs, _Quad rhs,
                                                int flag);
unsigned short __kmpc_atomic_fixed2u_div_cpt_fp(ident_t *id_ref, int gtid,
                                                unsigned short *lhs, _Quad rhs,
                                                int flag);

kmp_int32 __kmpc_atomic_fixed4_add_cpt_fp(ident_t *id_ref, int gtid,
                                          kmp_int32 *lhs, _Quad rhs, int flag);
kmp_int32 __kmpc_atomic_fixed4_sub_cpt_fp(ident_t *id_ref, int gtid,
                                          kmp_int32 *lhs, _Quad rhs, int flag);
kmp_int32 __kmpc_atomic_fixed4_mul_cpt_fp(ident_t *id_ref, int gtid,
                                          kmp_int32 *lhs, _Quad rhs, int flag);
kmp_int32 __kmpc_atomic_fixed4_div_cpt_fp(ident_t *id_ref, int gtid,
                                          kmp_int32 *lhs, _Quad rhs, int flag);
kmp_uint32 __kmpc_atomic_fixed4u_add_cpt_fp(ident_t *id_ref, int gtid,
                                            kmp_uint32 *lhs, _Quad rhs,
                                            int flag);
kmp_uint32 __kmpc_atomic_fixed4u_sub_cpt_fp(ident_t *id_ref, int gtid,
                                            kmp_uint32 *lhs, _Quad rhs,
                                            int flag);
kmp_uint32 __kmpc_atomic_fixed4u_mul_cpt_fp(ident_t *id_ref, int gtid,
                                            kmp_uint32 *lhs, _Quad rhs,
                                            int flag);
kmp_uint32 __kmpc_atomic_fixed4u_div_cpt_fp(ident_t *id_ref, int gtid,
                                            kmp_uint32 *lhs, _Quad rhs,
                                            int flag);

kmp_int64 __kmpc_atomic_fixed8_add_cpt_fp(ident_t *id_ref, int gtid,
                                          kmp_int64 *lhs, _Quad rhs, int flag);
kmp_int64 __kmpc_atomic_fixed8_sub_cpt_fp(ident_t *id_ref, int gtid,
                                          kmp_int64 *lhs, _Quad rhs, int flag);
kmp_int64 __kmpc_atomic_fixed8_mul_cpt_fp(ident_t *id_ref, int gtid,
                                          kmp_int64 *lhs, _Quad rhs, int flag);
kmp_int64 __kmpc_atomic_fixed8_div_cpt_fp(ident_t *id_ref, int gtid,
                                          kmp_int64 *lhs, _Quad rhs, int flag);
kmp_uint64 __kmpc_atomic_fixed8u_add_cpt_fp(ident_t *id_ref, int gtid,
                                            kmp_uint64 *lhs, _Quad rhs,
                                            int flag);
kmp_uint64 __kmpc_atomic_fixed8u_sub_cpt_fp(ident_t *id_ref, int gtid,
                                            kmp_uint64 *lhs, _Quad rhs,
                                            int flag);
kmp_uint64 __kmpc_atomic_fixed8u_mul_cpt_fp(ident_t *id_ref, int gtid,
                                            kmp_uint64 *lhs, _Quad rhs,
                                            int flag);
kmp_uint64 __kmpc_atomic_fixed8u_div_cpt_fp(ident_t *id_ref, int gtid,
                                            kmp_uint64 *lhs, _Quad rhs,
                                            int flag);

float __kmpc_atomic_float4_add_cpt_fp(ident_t *id_ref, int gtid,
                                      kmp_real32 *lhs, _Quad rhs, int flag);
float __kmpc_atomic_float4_sub_cpt_fp(ident_t *id_ref, int gtid,
                                      kmp_real32 *lhs, _Quad rhs, int flag);
float __kmpc_atomic_float4_mul_cpt_fp(ident_t *id_ref, int gtid,
                                      kmp_real32 *lhs, _Quad rhs, int flag);
float __kmpc_atomic_float4_div_cpt_fp(ident_t *id_ref, int gtid,
                                      kmp_real32 *lhs, _Quad rhs, int flag);

double __kmpc_atomic_float8_add_cpt_fp(ident_t *id_ref, int gtid,
                                       kmp_real64 *lhs, _Quad rhs, int flag);
double __kmpc_atomic_float8_sub_cpt_fp(ident_t *id_ref, int gtid,
                                       kmp_real64 *lhs, _Quad rhs, int flag);
double __kmpc_atomic_float8_mul_cpt_fp(ident_t *id_ref, int gtid,
                                       kmp_real64 *lhs, _Quad rhs, int flag);
double __kmpc_atomic_float8_div_cpt_fp(ident_t *id_ref, int gtid,
                                       kmp_real64 *lhs, _Quad rhs, int flag);

long double __kmpc_atomic_float10_add_cpt_fp(ident_t *id_ref, int gtid,
                                             long double *lhs, _Quad rhs,
                                             int flag);
long double __kmpc_atomic_float10_sub_cpt_fp(ident_t *id_ref, int gtid,
                                             long double *lhs, _Quad rhs,
                                             int flag);
long double __kmpc_atomic_float10_mul_cpt_fp(ident_t *id_ref, int gtid,
                                             long double *lhs, _Quad rhs,
                                             int flag);
long double __kmpc_atomic_float10_div_cpt_fp(ident_t *id_ref, int gtid,
                                             long double *lhs, _Quad rhs,
                                             int flag);

char __kmpc_atomic_fixed1_sub_cpt_rev_fp(ident_t *id_ref, int gtid, char *lhs,
                                         _Quad rhs, int flag);
unsigned char __kmpc_atomic_fixed1u_sub_cpt_rev_fp(ident_t *id_ref, int gtid,
                                                   unsigned char *lhs,
                                                   _Quad rhs, int flag);
char __kmpc_atomic_fixed1_div_cpt_rev_fp(ident_t *id_ref, int gtid, char *lhs,
                                         _Quad rhs, int flag);
unsigned char __kmpc_atomic_fixed1u_div_cpt_rev_fp(ident_t *id_ref, int gtid,
                                                   unsigned char *lhs,
                                                   _Quad rhs, int flag);
short __kmpc_atomic_fixed2_sub_cpt_rev_fp(ident_t *id_ref, int gtid, short *lhs,
                                          _Quad rhs, int flag);
unsigned short __kmpc_atomic_fixed2u_sub_cpt_rev_fp(ident_t *id_ref, int gtid,
                                                    unsigned short *lhs,
                                                    _Quad rhs, int flag);
short __kmpc_atomic_fixed2_div_cpt_rev_fp(ident_t *id_ref, int gtid, short *lhs,
                                          _Quad rhs, int flag);
unsigned short __kmpc_atomic_fixed2u_div_cpt_rev_fp(ident_t *id_ref, int gtid,
                                                    unsigned short *lhs,
                                                    _Quad rhs, int flag);
kmp_int32 __kmpc_atomic_fixed4_sub_cpt_rev_fp(ident_t *id_ref, int gtid,
                                              kmp_int32 *lhs, _Quad rhs,
                                              int flag);
kmp_uint32 __kmpc_atomic_fixed4u_sub_cpt_rev_fp(ident_t *id_ref, int gtid,
                                                kmp_uint32 *lhs, _Quad rhs,
                                                int flag);
kmp_int32 __kmpc_atomic_fixed4_div_cpt_rev_fp(ident_t *id_ref, int gtid,
                                              kmp_int32 *lhs, _Quad rhs,
                                              int flag);
kmp_uint32 __kmpc_atomic_fixed4u_div_cpt_rev_fp(ident_t *id_ref, int gtid,
                                                kmp_uint32 *lhs, _Quad rhs,
                                                int flag);
kmp_int64 __kmpc_atomic_fixed8_sub_cpt_rev_fp(ident_t *id_ref, int gtid,
                                              kmp_int64 *lhs, _Quad rhs,
                                              int flag);
kmp_uint64 __kmpc_atomic_fixed8u_sub_cpt_rev_fp(ident_t *id_ref, int gtid,
                                                kmp_uint64 *lhs, _Quad rhs,
                                                int flag);
kmp_int64 __kmpc_atomic_fixed8_div_cpt_rev_fp(ident_t *id_ref, int gtid,
                                              kmp_int64 *lhs, _Quad rhs,
                                              int flag);
kmp_uint64 __kmpc_atomic_fixed8u_div_cpt_rev_fp(ident_t *id_ref, int gtid,
                                                kmp_uint64 *lhs, _Quad rhs,
                                                int flag);
float __kmpc_atomic_float4_sub_cpt_rev_fp(ident_t *id_ref, int gtid, float *lhs,
                                          _Quad rhs, int flag);
float __kmpc_atomic_float4_div_cpt_rev_fp(ident_t *id_ref, int gtid, float *lhs,
                                          _Quad rhs, int flag);
double __kmpc_atomic_float8_sub_cpt_rev_fp(ident_t *id_ref, int gtid,
                                           double *lhs, _Quad rhs, int flag);
double __kmpc_atomic_float8_div_cpt_rev_fp(ident_t *id_ref, int gtid,
                                           double *lhs, _Quad rhs, int flag);
long double __kmpc_atomic_float10_sub_cpt_rev_fp(ident_t *id_ref, int gtid,
                                                 long double *lhs, _Quad rhs,
                                                 int flag);
long double __kmpc_atomic_float10_div_cpt_rev_fp(ident_t *id_ref, int gtid,
                                                 long double *lhs, _Quad rhs,
                                                 int flag);

#endif // KMP_HAVE_QUAD

// End of OpenMP 4.0 capture

#endif // KMP_ARCH_X86 || KMP_ARCH_X86_64

/* ------------------------------------------------------------------------ */

#ifdef __cplusplus
} // extern "C"
#endif

#endif /* KMP_ATOMIC_H */

// end of file