1 /* 2 * Copyright (c) 2018, 2019, Oracle and/or its affiliates. All rights reserved. 3 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER. 4 * 5 * This code is free software; you can redistribute it and/or modify it 6 * under the terms of the GNU General Public License version 2 only, as 7 * published by the Free Software Foundation. 8 * 9 * This code is distributed in the hope that it will be useful, but WITHOUT 10 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or 11 * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License 12 * version 2 for more details (a copy is included in the LICENSE file that 13 * accompanied this code). 14 * 15 * You should have received a copy of the GNU General Public License version 16 * 2 along with this work; if not, write to the Free Software Foundation, 17 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA. 18 * 19 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA 20 * or visit www.oracle.com if you need additional information or have any 21 * questions. 22 * 23 */ 24 25 #ifndef SHARE_UTILITIES_CONCURRENTHASHTABLE_INLINE_HPP 26 #define SHARE_UTILITIES_CONCURRENTHASHTABLE_INLINE_HPP 27 28 #include "memory/allocation.inline.hpp" 29 #include "runtime/atomic.hpp" 30 #include "runtime/orderAccess.hpp" 31 #include "runtime/prefetch.inline.hpp" 32 #include "utilities/concurrentHashTable.hpp" 33 #include "utilities/globalCounter.inline.hpp" 34 #include "utilities/numberSeq.hpp" 35 #include "utilities/spinYield.hpp" 36 37 // 2^30 = 1G buckets 38 #define SIZE_BIG_LOG2 30 39 // 2^5 = 32 buckets 40 #define SIZE_SMALL_LOG2 5 41 42 // Number from spinYield.hpp. In some loops SpinYield would be unfair. 43 #define SPINPAUSES_PER_YIELD 8192 44 45 #ifdef ASSERT 46 #ifdef _LP64 47 // Two low bits are not usable. 48 static const void* POISON_PTR = (void*)UCONST64(0xfbadbadbadbadbac); 49 #else 50 // Two low bits are not usable. 51 static const void* POISON_PTR = (void*)0xffbadbac; 52 #endif 53 #endif 54 55 // Node 56 template <typename CONFIG, MEMFLAGS F> 57 inline typename ConcurrentHashTable<CONFIG, F>::Node* 58 ConcurrentHashTable<CONFIG, F>:: 59 Node::next() const 60 { 61 return OrderAccess::load_acquire(&_next); 62 } 63 64 // Bucket 65 template <typename CONFIG, MEMFLAGS F> 66 inline typename ConcurrentHashTable<CONFIG, F>::Node* 67 ConcurrentHashTable<CONFIG, F>:: 68 Bucket::first_raw() const 69 { 70 return OrderAccess::load_acquire(&_first); 71 } 72 73 template <typename CONFIG, MEMFLAGS F> 74 inline void ConcurrentHashTable<CONFIG, F>:: 75 Bucket::release_assign_node_ptr( 76 typename ConcurrentHashTable<CONFIG, F>::Node* const volatile * dst, 77 typename ConcurrentHashTable<CONFIG, F>::Node* node) const 78 { 79 // Due to this assert this methods is not static. 80 assert(is_locked(), "Must be locked."); 81 Node** tmp = (Node**)dst; 82 OrderAccess::release_store(tmp, clear_set_state(node, *dst)); 83 } 84 85 template <typename CONFIG, MEMFLAGS F> 86 inline typename ConcurrentHashTable<CONFIG, F>::Node* 87 ConcurrentHashTable<CONFIG, F>:: 88 Bucket::first() const 89 { 90 // We strip the states bit before returning the ptr. 91 return clear_state(OrderAccess::load_acquire(&_first)); 92 } 93 94 template <typename CONFIG, MEMFLAGS F> 95 inline bool ConcurrentHashTable<CONFIG, F>:: 96 Bucket::have_redirect() const 97 { 98 return is_state(first_raw(), STATE_REDIRECT_BIT); 99 } 100 101 template <typename CONFIG, MEMFLAGS F> 102 inline bool ConcurrentHashTable<CONFIG, F>:: 103 Bucket::is_locked() const 104 { 105 return is_state(first_raw(), STATE_LOCK_BIT); 106 } 107 108 template <typename CONFIG, MEMFLAGS F> 109 inline void ConcurrentHashTable<CONFIG, F>:: 110 Bucket::lock() 111 { 112 int i = 0; 113 // SpinYield would be unfair here 114 while (!this->trylock()) { 115 if ((++i) == SPINPAUSES_PER_YIELD) { 116 // On contemporary OS yielding will give CPU to another runnable thread if 117 // there is no CPU available. 118 os::naked_yield(); 119 i = 0; 120 } else { 121 SpinPause(); 122 } 123 } 124 } 125 126 template <typename CONFIG, MEMFLAGS F> 127 inline void ConcurrentHashTable<CONFIG, F>:: 128 Bucket::release_assign_last_node_next( 129 typename ConcurrentHashTable<CONFIG, F>::Node* node) 130 { 131 assert(is_locked(), "Must be locked."); 132 Node* const volatile * ret = first_ptr(); 133 while (clear_state(*ret) != NULL) { 134 ret = clear_state(*ret)->next_ptr(); 135 } 136 release_assign_node_ptr(ret, node); 137 } 138 139 template <typename CONFIG, MEMFLAGS F> 140 inline bool ConcurrentHashTable<CONFIG, F>:: 141 Bucket::cas_first(typename ConcurrentHashTable<CONFIG, F>::Node* node, 142 typename ConcurrentHashTable<CONFIG, F>::Node* expect 143 ) 144 { 145 if (is_locked()) { 146 return false; 147 } 148 if (Atomic::cmpxchg(node, &_first, expect) == expect) { 149 return true; 150 } 151 return false; 152 } 153 154 template <typename CONFIG, MEMFLAGS F> 155 inline bool ConcurrentHashTable<CONFIG, F>:: 156 Bucket::trylock() 157 { 158 if (is_locked()) { 159 return false; 160 } 161 // We will expect a clean first pointer. 162 Node* tmp = first(); 163 if (Atomic::cmpxchg(set_state(tmp, STATE_LOCK_BIT), &_first, tmp) == tmp) { 164 return true; 165 } 166 return false; 167 } 168 169 template <typename CONFIG, MEMFLAGS F> 170 inline void ConcurrentHashTable<CONFIG, F>:: 171 Bucket::unlock() 172 { 173 assert(is_locked(), "Must be locked."); 174 assert(!have_redirect(), 175 "Unlocking a bucket after it has reached terminal state."); 176 OrderAccess::release_store(&_first, clear_state(first())); 177 } 178 179 template <typename CONFIG, MEMFLAGS F> 180 inline void ConcurrentHashTable<CONFIG, F>:: 181 Bucket::redirect() 182 { 183 assert(is_locked(), "Must be locked."); 184 OrderAccess::release_store(&_first, set_state(_first, STATE_REDIRECT_BIT)); 185 } 186 187 // InternalTable 188 template <typename CONFIG, MEMFLAGS F> 189 inline ConcurrentHashTable<CONFIG, F>:: 190 InternalTable::InternalTable(size_t log2_size) 191 : _log2_size(log2_size), _size(((size_t)1ul) << _log2_size), 192 _hash_mask(~(~((size_t)0) << _log2_size)) 193 { 194 assert(_log2_size >= SIZE_SMALL_LOG2 && _log2_size <= SIZE_BIG_LOG2, 195 "Bad size"); 196 _buckets = NEW_C_HEAP_ARRAY(Bucket, _size, F); 197 // Use placement new for each element instead of new[] which could use more 198 // memory than allocated. 199 for (size_t i = 0; i < _size; ++i) { 200 new (_buckets + i) Bucket(); 201 } 202 } 203 204 template <typename CONFIG, MEMFLAGS F> 205 inline ConcurrentHashTable<CONFIG, F>:: 206 InternalTable::~InternalTable() 207 { 208 FREE_C_HEAP_ARRAY(Bucket, _buckets); 209 } 210 211 // ScopedCS 212 template <typename CONFIG, MEMFLAGS F> 213 inline ConcurrentHashTable<CONFIG, F>:: 214 ScopedCS::ScopedCS(Thread* thread, ConcurrentHashTable<CONFIG, F>* cht) 215 : _thread(thread), 216 _cht(cht), 217 _cs_context(GlobalCounter::critical_section_begin(_thread)) 218 { 219 // This version is published now. 220 if (OrderAccess::load_acquire(&_cht->_invisible_epoch) != NULL) { 221 OrderAccess::release_store_fence(&_cht->_invisible_epoch, (Thread*)NULL); 222 } 223 } 224 225 template <typename CONFIG, MEMFLAGS F> 226 inline ConcurrentHashTable<CONFIG, F>:: 227 ScopedCS::~ScopedCS() 228 { 229 GlobalCounter::critical_section_end(_thread, _cs_context); 230 } 231 232 template <typename CONFIG, MEMFLAGS F> 233 template <typename LOOKUP_FUNC> 234 inline typename CONFIG::Value* ConcurrentHashTable<CONFIG, F>:: 235 MultiGetHandle::get(LOOKUP_FUNC& lookup_f, bool* grow_hint) 236 { 237 return ScopedCS::_cht->internal_get(ScopedCS::_thread, lookup_f, grow_hint); 238 } 239 240 // HaveDeletables 241 template <typename CONFIG, MEMFLAGS F> 242 template <typename EVALUATE_FUNC> 243 inline bool ConcurrentHashTable<CONFIG, F>:: 244 HaveDeletables<true, EVALUATE_FUNC>::have_deletable(Bucket* bucket, 245 EVALUATE_FUNC& eval_f, 246 Bucket* prefetch_bucket) 247 { 248 // Instantiated for pointer type (true), so we can use prefetch. 249 // When visiting all Nodes doing this prefetch give around 30%. 250 Node* pref = prefetch_bucket != NULL ? prefetch_bucket->first() : NULL; 251 for (Node* next = bucket->first(); next != NULL ; next = next->next()) { 252 if (pref != NULL) { 253 Prefetch::read(*pref->value(), 0); 254 pref = pref->next(); 255 } 256 // Read next() Node* once. May be racing with a thread moving the next 257 // pointers. 258 Node* next_pref = next->next(); 259 if (next_pref != NULL) { 260 Prefetch::read(*next_pref->value(), 0); 261 } 262 if (eval_f(next->value())) { 263 return true; 264 } 265 } 266 return false; 267 } 268 269 template <typename CONFIG, MEMFLAGS F> 270 template <bool b, typename EVALUATE_FUNC> 271 inline bool ConcurrentHashTable<CONFIG, F>:: 272 HaveDeletables<b, EVALUATE_FUNC>::have_deletable(Bucket* bucket, 273 EVALUATE_FUNC& eval_f, 274 Bucket* preb) 275 { 276 for (Node* next = bucket->first(); next != NULL ; next = next->next()) { 277 if (eval_f(next->value())) { 278 return true; 279 } 280 } 281 return false; 282 } 283 284 // ConcurrentHashTable 285 template <typename CONFIG, MEMFLAGS F> 286 inline void ConcurrentHashTable<CONFIG, F>:: 287 write_synchonize_on_visible_epoch(Thread* thread) 288 { 289 assert(_resize_lock_owner == thread, "Re-size lock not held"); 290 OrderAccess::fence(); // Prevent below load from floating up. 291 // If no reader saw this version we can skip write_synchronize. 292 if (OrderAccess::load_acquire(&_invisible_epoch) == thread) { 293 return; 294 } 295 assert(_invisible_epoch == NULL, "Two thread doing bulk operations"); 296 // We set this/next version that we are synchronizing for to not published. 297 // A reader will zero this flag if it reads this/next version. 298 OrderAccess::release_store(&_invisible_epoch, thread); 299 GlobalCounter::write_synchronize(); 300 } 301 302 template <typename CONFIG, MEMFLAGS F> 303 inline bool ConcurrentHashTable<CONFIG, F>:: 304 try_resize_lock(Thread* locker) 305 { 306 if (_resize_lock->try_lock()) { 307 if (_resize_lock_owner != NULL) { 308 assert(locker != _resize_lock_owner, "Already own lock"); 309 // We got mutex but internal state is locked. 310 _resize_lock->unlock(); 311 return false; 312 } 313 } else { 314 return false; 315 } 316 _invisible_epoch = 0; 317 _resize_lock_owner = locker; 318 return true; 319 } 320 321 template <typename CONFIG, MEMFLAGS F> 322 inline void ConcurrentHashTable<CONFIG, F>:: 323 lock_resize_lock(Thread* locker) 324 { 325 size_t i = 0; 326 // If lock is hold by some other thread, the chances that it is return quick 327 // is low. So we will prefer yielding. 328 SpinYield yield(1, 512); 329 do { 330 _resize_lock->lock_without_safepoint_check(); 331 // If holder of lock dropped mutex for safepoint mutex might be unlocked, 332 // and _resize_lock_owner will contain the owner. 333 if (_resize_lock_owner != NULL) { 334 assert(locker != _resize_lock_owner, "Already own lock"); 335 // We got mutex but internal state is locked. 336 _resize_lock->unlock(); 337 yield.wait(); 338 } else { 339 break; 340 } 341 } while(true); 342 _resize_lock_owner = locker; 343 _invisible_epoch = 0; 344 } 345 346 template <typename CONFIG, MEMFLAGS F> 347 inline void ConcurrentHashTable<CONFIG, F>:: 348 unlock_resize_lock(Thread* locker) 349 { 350 _invisible_epoch = 0; 351 assert(locker == _resize_lock_owner, "Not unlocked by locker."); 352 _resize_lock_owner = NULL; 353 _resize_lock->unlock(); 354 } 355 356 template <typename CONFIG, MEMFLAGS F> 357 inline void ConcurrentHashTable<CONFIG, F>:: 358 free_nodes() 359 { 360 // We assume we are not MT during freeing. 361 for (size_t node_it = 0; node_it < _table->_size; node_it++) { 362 Bucket* bucket = _table->get_buckets() + node_it; 363 Node* node = bucket->first(); 364 while (node != NULL) { 365 Node* free_node = node; 366 node = node->next(); 367 Node::destroy_node(free_node); 368 } 369 } 370 } 371 372 template <typename CONFIG, MEMFLAGS F> 373 inline typename ConcurrentHashTable<CONFIG, F>::InternalTable* 374 ConcurrentHashTable<CONFIG, F>:: 375 get_table() const 376 { 377 return OrderAccess::load_acquire(&_table); 378 } 379 380 template <typename CONFIG, MEMFLAGS F> 381 inline typename ConcurrentHashTable<CONFIG, F>::InternalTable* 382 ConcurrentHashTable<CONFIG, F>:: 383 get_new_table() const 384 { 385 return OrderAccess::load_acquire(&_new_table); 386 } 387 388 template <typename CONFIG, MEMFLAGS F> 389 inline typename ConcurrentHashTable<CONFIG, F>::InternalTable* 390 ConcurrentHashTable<CONFIG, F>:: 391 set_table_from_new() 392 { 393 InternalTable* old_table = _table; 394 // Publish the new table. 395 OrderAccess::release_store(&_table, _new_table); 396 // All must see this. 397 GlobalCounter::write_synchronize(); 398 // _new_table not read any more. 399 _new_table = NULL; 400 DEBUG_ONLY(_new_table = (InternalTable*)POISON_PTR;) 401 return old_table; 402 } 403 404 template <typename CONFIG, MEMFLAGS F> 405 inline void ConcurrentHashTable<CONFIG, F>:: 406 internal_grow_range(Thread* thread, size_t start, size_t stop) 407 { 408 assert(stop <= _table->_size, "Outside backing array"); 409 assert(_new_table != NULL, "Grow not proper setup before start"); 410 // The state is also copied here. Hence all buckets in new table will be 411 // locked. I call the siblings odd/even, where even have high bit 0 and odd 412 // have high bit 1. 413 for (size_t even_index = start; even_index < stop; even_index++) { 414 Bucket* bucket = _table->get_bucket(even_index); 415 416 bucket->lock(); 417 418 size_t odd_index = even_index + _table->_size; 419 _new_table->get_buckets()[even_index] = *bucket; 420 _new_table->get_buckets()[odd_index] = *bucket; 421 422 // Moves lockers go to new table, where they will wait until unlock() below. 423 bucket->redirect(); /* Must release stores above */ 424 425 // When this is done we have separated the nodes into corresponding buckets 426 // in new table. 427 if (!unzip_bucket(thread, _table, _new_table, even_index, odd_index)) { 428 // If bucket is empty, unzip does nothing. 429 // We must make sure readers go to new table before we poison the bucket. 430 DEBUG_ONLY(GlobalCounter::write_synchronize();) 431 } 432 433 // Unlock for writes into the new table buckets. 434 _new_table->get_bucket(even_index)->unlock(); 435 _new_table->get_bucket(odd_index)->unlock(); 436 437 DEBUG_ONLY( 438 bucket->release_assign_node_ptr( 439 _table->get_bucket(even_index)->first_ptr(), (Node*)POISON_PTR); 440 ) 441 } 442 } 443 444 template <typename CONFIG, MEMFLAGS F> 445 template <typename LOOKUP_FUNC, typename DELETE_FUNC> 446 inline bool ConcurrentHashTable<CONFIG, F>:: 447 internal_remove(Thread* thread, LOOKUP_FUNC& lookup_f, DELETE_FUNC& delete_f) 448 { 449 Bucket* bucket = get_bucket_locked(thread, lookup_f.get_hash()); 450 assert(bucket->is_locked(), "Must be locked."); 451 Node* const volatile * rem_n_prev = bucket->first_ptr(); 452 Node* rem_n = bucket->first(); 453 bool have_dead = false; 454 while (rem_n != NULL) { 455 if (lookup_f.equals(rem_n->value(), &have_dead)) { 456 bucket->release_assign_node_ptr(rem_n_prev, rem_n->next()); 457 break; 458 } else { 459 rem_n_prev = rem_n->next_ptr(); 460 rem_n = rem_n->next(); 461 } 462 } 463 464 bucket->unlock(); 465 466 if (rem_n == NULL) { 467 return false; 468 } 469 // Publish the deletion. 470 GlobalCounter::write_synchronize(); 471 delete_f(rem_n->value()); 472 Node::destroy_node(rem_n); 473 JFR_ONLY(_stats_rate.remove();) 474 return true; 475 } 476 477 template <typename CONFIG, MEMFLAGS F> 478 template <typename EVALUATE_FUNC, typename DELETE_FUNC> 479 inline void ConcurrentHashTable<CONFIG, F>:: 480 do_bulk_delete_locked_for(Thread* thread, size_t start_idx, size_t stop_idx, 481 EVALUATE_FUNC& eval_f, DELETE_FUNC& del_f, bool is_mt) 482 { 483 // Here we have resize lock so table is SMR safe, and there is no new 484 // table. Can do this in parallel if we want. 485 assert((is_mt && _resize_lock_owner != NULL) || 486 (!is_mt && _resize_lock_owner == thread), "Re-size lock not held"); 487 Node* ndel[BULK_DELETE_LIMIT]; 488 InternalTable* table = get_table(); 489 assert(start_idx < stop_idx, "Must be"); 490 assert(stop_idx <= _table->_size, "Must be"); 491 // Here manual do critical section since we don't want to take the cost of 492 // locking the bucket if there is nothing to delete. But we can have 493 // concurrent single deletes. The _invisible_epoch can only be used by the 494 // owner of _resize_lock, us here. There we should not changed it in our 495 // own read-side. 496 GlobalCounter::CSContext cs_context = GlobalCounter::critical_section_begin(thread); 497 for (size_t bucket_it = start_idx; bucket_it < stop_idx; bucket_it++) { 498 Bucket* bucket = table->get_bucket(bucket_it); 499 Bucket* prefetch_bucket = (bucket_it+1) < stop_idx ? 500 table->get_bucket(bucket_it+1) : NULL; 501 502 if (!HaveDeletables<IsPointer<VALUE>::value, EVALUATE_FUNC>:: 503 have_deletable(bucket, eval_f, prefetch_bucket)) { 504 // Nothing to remove in this bucket. 505 continue; 506 } 507 508 GlobalCounter::critical_section_end(thread, cs_context); 509 // We left critical section but the bucket cannot be removed while we hold 510 // the _resize_lock. 511 bucket->lock(); 512 size_t nd = delete_check_nodes(bucket, eval_f, BULK_DELETE_LIMIT, ndel); 513 bucket->unlock(); 514 if (is_mt) { 515 GlobalCounter::write_synchronize(); 516 } else { 517 write_synchonize_on_visible_epoch(thread); 518 } 519 for (size_t node_it = 0; node_it < nd; node_it++) { 520 del_f(ndel[node_it]->value()); 521 Node::destroy_node(ndel[node_it]); 522 JFR_ONLY(_stats_rate.remove();) 523 DEBUG_ONLY(ndel[node_it] = (Node*)POISON_PTR;) 524 } 525 cs_context = GlobalCounter::critical_section_begin(thread); 526 } 527 GlobalCounter::critical_section_end(thread, cs_context); 528 } 529 530 template <typename CONFIG, MEMFLAGS F> 531 template <typename LOOKUP_FUNC> 532 inline void ConcurrentHashTable<CONFIG, F>:: 533 delete_in_bucket(Thread* thread, Bucket* bucket, LOOKUP_FUNC& lookup_f) 534 { 535 assert(bucket->is_locked(), "Must be locked."); 536 537 size_t dels = 0; 538 Node* ndel[BULK_DELETE_LIMIT]; 539 Node* const volatile * rem_n_prev = bucket->first_ptr(); 540 Node* rem_n = bucket->first(); 541 while (rem_n != NULL) { 542 bool is_dead = false; 543 lookup_f.equals(rem_n->value(), &is_dead); 544 if (is_dead) { 545 ndel[dels++] = rem_n; 546 Node* next_node = rem_n->next(); 547 bucket->release_assign_node_ptr(rem_n_prev, next_node); 548 rem_n = next_node; 549 if (dels == BULK_DELETE_LIMIT) { 550 break; 551 } 552 } else { 553 rem_n_prev = rem_n->next_ptr(); 554 rem_n = rem_n->next(); 555 } 556 } 557 if (dels > 0) { 558 GlobalCounter::write_synchronize(); 559 for (size_t node_it = 0; node_it < dels; node_it++) { 560 Node::destroy_node(ndel[node_it]); 561 JFR_ONLY(_stats_rate.remove();) 562 DEBUG_ONLY(ndel[node_it] = (Node*)POISON_PTR;) 563 } 564 } 565 } 566 567 template <typename CONFIG, MEMFLAGS F> 568 inline typename ConcurrentHashTable<CONFIG, F>::Bucket* 569 ConcurrentHashTable<CONFIG, F>:: 570 get_bucket(uintx hash) const 571 { 572 InternalTable* table = get_table(); 573 Bucket* bucket = get_bucket_in(table, hash); 574 if (bucket->have_redirect()) { 575 table = get_new_table(); 576 bucket = get_bucket_in(table, hash); 577 } 578 return bucket; 579 } 580 581 template <typename CONFIG, MEMFLAGS F> 582 inline typename ConcurrentHashTable<CONFIG, F>::Bucket* 583 ConcurrentHashTable<CONFIG, F>:: 584 get_bucket_locked(Thread* thread, const uintx hash) 585 { 586 Bucket* bucket; 587 int i = 0; 588 // SpinYield would be unfair here 589 while(true) { 590 { 591 // We need a critical section to protect the table itself. But if we fail 592 // we must leave critical section otherwise we would deadlock. 593 ScopedCS cs(thread, this); 594 bucket = get_bucket(hash); 595 if (bucket->trylock()) { 596 break; /* ends critical section */ 597 } 598 } /* ends critical section */ 599 if ((++i) == SPINPAUSES_PER_YIELD) { 600 // On contemporary OS yielding will give CPU to another runnable thread if 601 // there is no CPU available. 602 os::naked_yield(); 603 i = 0; 604 } else { 605 SpinPause(); 606 } 607 } 608 return bucket; 609 } 610 611 // Always called within critical section 612 template <typename CONFIG, MEMFLAGS F> 613 template <typename LOOKUP_FUNC> 614 typename ConcurrentHashTable<CONFIG, F>::Node* 615 ConcurrentHashTable<CONFIG, F>:: 616 get_node(const Bucket* const bucket, LOOKUP_FUNC& lookup_f, 617 bool* have_dead, size_t* loops) const 618 { 619 size_t loop_count = 0; 620 Node* node = bucket->first(); 621 while (node != NULL) { 622 bool is_dead = false; 623 ++loop_count; 624 if (lookup_f.equals(node->value(), &is_dead)) { 625 break; 626 } 627 if (is_dead && !(*have_dead)) { 628 *have_dead = true; 629 } 630 node = node->next(); 631 } 632 if (loops != NULL) { 633 *loops = loop_count; 634 } 635 return node; 636 } 637 638 template <typename CONFIG, MEMFLAGS F> 639 inline bool ConcurrentHashTable<CONFIG, F>:: 640 unzip_bucket(Thread* thread, InternalTable* old_table, 641 InternalTable* new_table, size_t even_index, size_t odd_index) 642 { 643 Node* aux = old_table->get_bucket(even_index)->first(); 644 if (aux == NULL) { 645 // This is an empty bucket and in debug we poison first ptr in bucket. 646 // Therefore we must make sure no readers are looking at this bucket. 647 // If we don't do a write_synch here, caller must do it. 648 return false; 649 } 650 Node* delete_me = NULL; 651 Node* const volatile * even = new_table->get_bucket(even_index)->first_ptr(); 652 Node* const volatile * odd = new_table->get_bucket(odd_index)->first_ptr(); 653 while (aux != NULL) { 654 bool dead_hash = false; 655 size_t aux_hash = CONFIG::get_hash(*aux->value(), &dead_hash); 656 Node* aux_next = aux->next(); 657 if (dead_hash) { 658 delete_me = aux; 659 // This item is dead, move both list to next 660 new_table->get_bucket(odd_index)->release_assign_node_ptr(odd, 661 aux_next); 662 new_table->get_bucket(even_index)->release_assign_node_ptr(even, 663 aux_next); 664 } else { 665 size_t aux_index = bucket_idx_hash(new_table, aux_hash); 666 if (aux_index == even_index) { 667 // This is a even, so move odd to aux/even next 668 new_table->get_bucket(odd_index)->release_assign_node_ptr(odd, 669 aux_next); 670 // Keep in even list 671 even = aux->next_ptr(); 672 } else if (aux_index == odd_index) { 673 // This is a odd, so move odd to aux/odd next 674 new_table->get_bucket(even_index)->release_assign_node_ptr(even, 675 aux_next); 676 // Keep in odd list 677 odd = aux->next_ptr(); 678 } else { 679 fatal("aux_index does not match even or odd indices"); 680 } 681 } 682 aux = aux_next; 683 684 // We can only move 1 pointer otherwise a reader might be moved to the wrong 685 // chain. E.g. looking for even hash value but got moved to the odd bucket 686 // chain. 687 write_synchonize_on_visible_epoch(thread); 688 if (delete_me != NULL) { 689 Node::destroy_node(delete_me); 690 delete_me = NULL; 691 } 692 } 693 return true; 694 } 695 696 template <typename CONFIG, MEMFLAGS F> 697 inline bool ConcurrentHashTable<CONFIG, F>:: 698 internal_shrink_prolog(Thread* thread, size_t log2_size) 699 { 700 if (!try_resize_lock(thread)) { 701 return false; 702 } 703 assert(_resize_lock_owner == thread, "Re-size lock not held"); 704 if (_table->_log2_size == _log2_start_size || 705 _table->_log2_size <= log2_size) { 706 unlock_resize_lock(thread); 707 return false; 708 } 709 _new_table = new InternalTable(_table->_log2_size - 1); 710 return true; 711 } 712 713 template <typename CONFIG, MEMFLAGS F> 714 inline void ConcurrentHashTable<CONFIG, F>:: 715 internal_shrink_epilog(Thread* thread) 716 { 717 assert(_resize_lock_owner == thread, "Re-size lock not held"); 718 719 InternalTable* old_table = set_table_from_new(); 720 _size_limit_reached = false; 721 unlock_resize_lock(thread); 722 #ifdef ASSERT 723 for (size_t i = 0; i < old_table->_size; i++) { 724 assert(old_table->get_bucket(i++)->first() == POISON_PTR, 725 "No poison found"); 726 } 727 #endif 728 // ABA safe, old_table not visible to any other threads. 729 delete old_table; 730 } 731 732 template <typename CONFIG, MEMFLAGS F> 733 inline void ConcurrentHashTable<CONFIG, F>:: 734 internal_shrink_range(Thread* thread, size_t start, size_t stop) 735 { 736 // The state is also copied here. 737 // Hence all buckets in new table will be locked. 738 for (size_t bucket_it = start; bucket_it < stop; bucket_it++) { 739 size_t even_hash_index = bucket_it; // High bit 0 740 size_t odd_hash_index = bucket_it + _new_table->_size; // High bit 1 741 742 Bucket* b_old_even = _table->get_bucket(even_hash_index); 743 Bucket* b_old_odd = _table->get_bucket(odd_hash_index); 744 745 b_old_even->lock(); 746 b_old_odd->lock(); 747 748 _new_table->get_buckets()[bucket_it] = *b_old_even; 749 750 // Put chains together. 751 _new_table->get_bucket(bucket_it)-> 752 release_assign_last_node_next(*(b_old_odd->first_ptr())); 753 754 b_old_even->redirect(); 755 b_old_odd->redirect(); 756 757 write_synchonize_on_visible_epoch(thread); 758 759 // Unlock for writes into new smaller table. 760 _new_table->get_bucket(bucket_it)->unlock(); 761 762 DEBUG_ONLY(b_old_even->release_assign_node_ptr(b_old_even->first_ptr(), 763 (Node*)POISON_PTR);) 764 DEBUG_ONLY(b_old_odd->release_assign_node_ptr(b_old_odd->first_ptr(), 765 (Node*)POISON_PTR);) 766 } 767 } 768 769 template <typename CONFIG, MEMFLAGS F> 770 inline bool ConcurrentHashTable<CONFIG, F>:: 771 internal_shrink(Thread* thread, size_t log2_size) 772 { 773 if (!internal_shrink_prolog(thread, log2_size)) { 774 assert(_resize_lock_owner != thread, "Re-size lock held"); 775 return false; 776 } 777 assert(_resize_lock_owner == thread, "Should be locked by me"); 778 internal_shrink_range(thread, 0, _new_table->_size); 779 internal_shrink_epilog(thread); 780 assert(_resize_lock_owner != thread, "Re-size lock held"); 781 return true; 782 } 783 784 template <typename CONFIG, MEMFLAGS F> 785 inline bool ConcurrentHashTable<CONFIG, F>:: 786 internal_grow_prolog(Thread* thread, size_t log2_size) 787 { 788 // This double checking of _size_limit_reached/is_max_size_reached() 789 // we only do in grow path, since grow means high load on table 790 // while shrink means low load. 791 if (is_max_size_reached()) { 792 return false; 793 } 794 if (!try_resize_lock(thread)) { 795 // Either we have an ongoing resize or an operation which doesn't want us 796 // to resize now. 797 return false; 798 } 799 if (is_max_size_reached() || _table->_log2_size >= log2_size) { 800 unlock_resize_lock(thread); 801 return false; 802 } 803 804 _new_table = new InternalTable(_table->_log2_size + 1); 805 806 if (_new_table->_log2_size == _log2_size_limit) { 807 _size_limit_reached = true; 808 } 809 810 return true; 811 } 812 813 template <typename CONFIG, MEMFLAGS F> 814 inline void ConcurrentHashTable<CONFIG, F>:: 815 internal_grow_epilog(Thread* thread) 816 { 817 assert(_resize_lock_owner == thread, "Should be locked"); 818 819 InternalTable* old_table = set_table_from_new(); 820 unlock_resize_lock(thread); 821 #ifdef ASSERT 822 for (size_t i = 0; i < old_table->_size; i++) { 823 assert(old_table->get_bucket(i++)->first() == POISON_PTR, 824 "No poison found"); 825 } 826 #endif 827 // ABA safe, old_table not visible to any other threads. 828 delete old_table; 829 } 830 831 template <typename CONFIG, MEMFLAGS F> 832 inline bool ConcurrentHashTable<CONFIG, F>:: 833 internal_grow(Thread* thread, size_t log2_size) 834 { 835 if (!internal_grow_prolog(thread, log2_size)) { 836 assert(_resize_lock_owner != thread, "Re-size lock held"); 837 return false; 838 } 839 assert(_resize_lock_owner == thread, "Should be locked by me"); 840 internal_grow_range(thread, 0, _table->_size); 841 internal_grow_epilog(thread); 842 assert(_resize_lock_owner != thread, "Re-size lock held"); 843 return true; 844 } 845 846 // Always called within critical section 847 template <typename CONFIG, MEMFLAGS F> 848 template <typename LOOKUP_FUNC> 849 inline typename CONFIG::Value* ConcurrentHashTable<CONFIG, F>:: 850 internal_get(Thread* thread, LOOKUP_FUNC& lookup_f, bool* grow_hint) 851 { 852 bool clean = false; 853 size_t loops = 0; 854 VALUE* ret = NULL; 855 856 const Bucket* bucket = get_bucket(lookup_f.get_hash()); 857 Node* node = get_node(bucket, lookup_f, &clean, &loops); 858 if (node != NULL) { 859 ret = node->value(); 860 } 861 if (grow_hint != NULL) { 862 *grow_hint = loops > _grow_hint; 863 } 864 865 return ret; 866 } 867 868 template <typename CONFIG, MEMFLAGS F> 869 template <typename LOOKUP_FUNC> 870 inline bool ConcurrentHashTable<CONFIG, F>:: 871 internal_insert(Thread* thread, LOOKUP_FUNC& lookup_f, const VALUE& value, 872 bool* grow_hint, bool* clean_hint) 873 { 874 bool ret = false; 875 bool clean = false; 876 bool locked; 877 size_t loops = 0; 878 size_t i = 0; 879 uintx hash = lookup_f.get_hash(); 880 Node* new_node = Node::create_node(value, NULL); 881 882 while (true) { 883 { 884 ScopedCS cs(thread, this); /* protected the table/bucket */ 885 Bucket* bucket = get_bucket(hash); 886 Node* first_at_start = bucket->first(); 887 Node* old = get_node(bucket, lookup_f, &clean, &loops); 888 if (old == NULL) { 889 new_node->set_next(first_at_start); 890 if (bucket->cas_first(new_node, first_at_start)) { 891 JFR_ONLY(_stats_rate.add();) 892 new_node = NULL; 893 ret = true; 894 break; /* leave critical section */ 895 } 896 // CAS failed we must leave critical section and retry. 897 locked = bucket->is_locked(); 898 } else { 899 // There is a duplicate. 900 break; /* leave critical section */ 901 } 902 } /* leave critical section */ 903 i++; 904 if (locked) { 905 os::naked_yield(); 906 } else { 907 SpinPause(); 908 } 909 } 910 911 if (new_node != NULL) { 912 // CAS failed and a duplicate was inserted, we must free this node. 913 Node::destroy_node(new_node); 914 } else if (i == 0 && clean) { 915 // We only do cleaning on fast inserts. 916 Bucket* bucket = get_bucket_locked(thread, lookup_f.get_hash()); 917 delete_in_bucket(thread, bucket, lookup_f); 918 bucket->unlock(); 919 clean = false; 920 } 921 922 if (grow_hint != NULL) { 923 *grow_hint = loops > _grow_hint; 924 } 925 926 if (clean_hint != NULL) { 927 *clean_hint = clean; 928 } 929 930 return ret; 931 } 932 933 template <typename CONFIG, MEMFLAGS F> 934 template <typename FUNC> 935 inline bool ConcurrentHashTable<CONFIG, F>:: 936 visit_nodes(Bucket* bucket, FUNC& visitor_f) 937 { 938 Node* current_node = bucket->first(); 939 while (current_node != NULL) { 940 if (!visitor_f(current_node->value())) { 941 return false; 942 } 943 current_node = current_node->next(); 944 } 945 return true; 946 } 947 948 template <typename CONFIG, MEMFLAGS F> 949 template <typename FUNC> 950 inline void ConcurrentHashTable<CONFIG, F>:: 951 do_scan_locked(Thread* thread, FUNC& scan_f) 952 { 953 assert(_resize_lock_owner == thread, "Re-size lock not held"); 954 // We can do a critical section over the entire loop but that would block 955 // updates for a long time. Instead we choose to block resizes. 956 InternalTable* table = get_table(); 957 for (size_t bucket_it = 0; bucket_it < table->_size; bucket_it++) { 958 ScopedCS cs(thread, this); 959 if (!visit_nodes(table->get_bucket(bucket_it), scan_f)) { 960 break; /* ends critical section */ 961 } 962 } /* ends critical section */ 963 } 964 965 template <typename CONFIG, MEMFLAGS F> 966 template <typename EVALUATE_FUNC> 967 inline size_t ConcurrentHashTable<CONFIG, F>:: 968 delete_check_nodes(Bucket* bucket, EVALUATE_FUNC& eval_f, 969 size_t num_del, Node** ndel) 970 { 971 size_t dels = 0; 972 Node* const volatile * rem_n_prev = bucket->first_ptr(); 973 Node* rem_n = bucket->first(); 974 while (rem_n != NULL) { 975 if (eval_f(rem_n->value())) { 976 ndel[dels++] = rem_n; 977 Node* next_node = rem_n->next(); 978 bucket->release_assign_node_ptr(rem_n_prev, next_node); 979 rem_n = next_node; 980 if (dels == num_del) { 981 break; 982 } 983 } else { 984 rem_n_prev = rem_n->next_ptr(); 985 rem_n = rem_n->next(); 986 } 987 } 988 return dels; 989 } 990 991 // Constructor 992 template <typename CONFIG, MEMFLAGS F> 993 inline ConcurrentHashTable<CONFIG, F>:: 994 ConcurrentHashTable(size_t log2size, size_t log2size_limit, size_t grow_hint) 995 : _new_table(NULL), _log2_size_limit(log2size_limit), 996 _log2_start_size(log2size), _grow_hint(grow_hint), 997 _size_limit_reached(false), _resize_lock_owner(NULL), 998 _invisible_epoch(0) 999 { 1000 _stats_rate = TableRateStatistics(); 1001 _resize_lock = 1002 new Mutex(Mutex::leaf, "ConcurrentHashTable", true, 1003 Mutex::_safepoint_check_never); 1004 _table = new InternalTable(log2size); 1005 assert(log2size_limit >= log2size, "bad ergo"); 1006 _size_limit_reached = _table->_log2_size == _log2_size_limit; 1007 } 1008 1009 template <typename CONFIG, MEMFLAGS F> 1010 inline ConcurrentHashTable<CONFIG, F>:: 1011 ~ConcurrentHashTable() 1012 { 1013 delete _resize_lock; 1014 free_nodes(); 1015 delete _table; 1016 } 1017 1018 template <typename CONFIG, MEMFLAGS F> 1019 inline size_t ConcurrentHashTable<CONFIG, F>:: 1020 get_size_log2(Thread* thread) 1021 { 1022 ScopedCS cs(thread, this); 1023 return _table->_log2_size; 1024 } 1025 1026 template <typename CONFIG, MEMFLAGS F> 1027 inline bool ConcurrentHashTable<CONFIG, F>:: 1028 shrink(Thread* thread, size_t size_limit_log2) 1029 { 1030 size_t tmp = size_limit_log2 == 0 ? _log2_start_size : size_limit_log2; 1031 bool ret = internal_shrink(thread, tmp); 1032 return ret; 1033 } 1034 1035 template <typename CONFIG, MEMFLAGS F> 1036 inline bool ConcurrentHashTable<CONFIG, F>:: 1037 grow(Thread* thread, size_t size_limit_log2) 1038 { 1039 size_t tmp = size_limit_log2 == 0 ? _log2_size_limit : size_limit_log2; 1040 return internal_grow(thread, tmp); 1041 } 1042 1043 template <typename CONFIG, MEMFLAGS F> 1044 template <typename LOOKUP_FUNC, typename FOUND_FUNC> 1045 inline bool ConcurrentHashTable<CONFIG, F>:: 1046 get(Thread* thread, LOOKUP_FUNC& lookup_f, FOUND_FUNC& found_f, bool* grow_hint) 1047 { 1048 bool ret = false; 1049 ScopedCS cs(thread, this); 1050 VALUE* val = internal_get(thread, lookup_f, grow_hint); 1051 if (val != NULL) { 1052 found_f(val); 1053 ret = true; 1054 } 1055 return ret; 1056 } 1057 1058 template <typename CONFIG, MEMFLAGS F> 1059 inline bool ConcurrentHashTable<CONFIG, F>:: 1060 unsafe_insert(const VALUE& value) { 1061 bool dead_hash = false; 1062 size_t hash = CONFIG::get_hash(value, &dead_hash); 1063 if (dead_hash) { 1064 return false; 1065 } 1066 // This is an unsafe operation. 1067 InternalTable* table = get_table(); 1068 Bucket* bucket = get_bucket_in(table, hash); 1069 assert(!bucket->have_redirect() && !bucket->is_locked(), "bad"); 1070 Node* new_node = Node::create_node(value, bucket->first()); 1071 if (!bucket->cas_first(new_node, bucket->first())) { 1072 assert(false, "bad"); 1073 } 1074 JFR_ONLY(_stats_rate.add();) 1075 return true; 1076 } 1077 1078 template <typename CONFIG, MEMFLAGS F> 1079 template <typename SCAN_FUNC> 1080 inline bool ConcurrentHashTable<CONFIG, F>:: 1081 try_scan(Thread* thread, SCAN_FUNC& scan_f) 1082 { 1083 if (!try_resize_lock(thread)) { 1084 return false; 1085 } 1086 do_scan_locked(thread, scan_f); 1087 unlock_resize_lock(thread); 1088 return true; 1089 } 1090 1091 template <typename CONFIG, MEMFLAGS F> 1092 template <typename SCAN_FUNC> 1093 inline void ConcurrentHashTable<CONFIG, F>:: 1094 do_scan(Thread* thread, SCAN_FUNC& scan_f) 1095 { 1096 assert(!SafepointSynchronize::is_at_safepoint(), 1097 "must be outside a safepoint"); 1098 assert(_resize_lock_owner != thread, "Re-size lock held"); 1099 lock_resize_lock(thread); 1100 do_scan_locked(thread, scan_f); 1101 unlock_resize_lock(thread); 1102 assert(_resize_lock_owner != thread, "Re-size lock held"); 1103 } 1104 1105 template <typename CONFIG, MEMFLAGS F> 1106 template <typename SCAN_FUNC> 1107 inline void ConcurrentHashTable<CONFIG, F>:: 1108 do_safepoint_scan(SCAN_FUNC& scan_f) 1109 { 1110 // We only allow this method to be used during a safepoint. 1111 assert(SafepointSynchronize::is_at_safepoint(), 1112 "must only be called in a safepoint"); 1113 assert(Thread::current()->is_VM_thread(), 1114 "should be in vm thread"); 1115 1116 // Here we skip protection, 1117 // thus no other thread may use this table at the same time. 1118 InternalTable* table = get_table(); 1119 for (size_t bucket_it = 0; bucket_it < table->_size; bucket_it++) { 1120 Bucket* bucket = table->get_bucket(bucket_it); 1121 // If bucket have a redirect the items will be in the new table. 1122 // We must visit them there since the new table will contain any 1123 // concurrent inserts done after this bucket was resized. 1124 // If the bucket don't have redirect flag all items is in this table. 1125 if (!bucket->have_redirect()) { 1126 if(!visit_nodes(bucket, scan_f)) { 1127 return; 1128 } 1129 } else { 1130 assert(bucket->is_locked(), "Bucket must be locked."); 1131 } 1132 } 1133 // If there is a paused resize we also need to visit the already resized items. 1134 table = get_new_table(); 1135 if (table == NULL) { 1136 return; 1137 } 1138 DEBUG_ONLY(if (table == POISON_PTR) { return; }) 1139 for (size_t bucket_it = 0; bucket_it < table->_size; bucket_it++) { 1140 Bucket* bucket = table->get_bucket(bucket_it); 1141 assert(!bucket->is_locked(), "Bucket must be unlocked."); 1142 if (!visit_nodes(bucket, scan_f)) { 1143 return; 1144 } 1145 } 1146 } 1147 1148 template <typename CONFIG, MEMFLAGS F> 1149 template <typename EVALUATE_FUNC, typename DELETE_FUNC> 1150 inline bool ConcurrentHashTable<CONFIG, F>:: 1151 try_bulk_delete(Thread* thread, EVALUATE_FUNC& eval_f, DELETE_FUNC& del_f) 1152 { 1153 if (!try_resize_lock(thread)) { 1154 return false; 1155 } 1156 do_bulk_delete_locked(thread, eval_f, del_f); 1157 unlock_resize_lock(thread); 1158 assert(_resize_lock_owner != thread, "Re-size lock held"); 1159 return true; 1160 } 1161 1162 template <typename CONFIG, MEMFLAGS F> 1163 template <typename EVALUATE_FUNC, typename DELETE_FUNC> 1164 inline void ConcurrentHashTable<CONFIG, F>:: 1165 bulk_delete(Thread* thread, EVALUATE_FUNC& eval_f, DELETE_FUNC& del_f) 1166 { 1167 assert(!SafepointSynchronize::is_at_safepoint(), 1168 "must be outside a safepoint"); 1169 lock_resize_lock(thread); 1170 do_bulk_delete_locked(thread, eval_f, del_f); 1171 unlock_resize_lock(thread); 1172 } 1173 1174 template <typename CONFIG, MEMFLAGS F> 1175 template <typename VALUE_SIZE_FUNC> 1176 inline TableStatistics ConcurrentHashTable<CONFIG, F>:: 1177 statistics_calculate(Thread* thread, VALUE_SIZE_FUNC& vs_f) 1178 { 1179 NumberSeq summary; 1180 size_t literal_bytes = 0; 1181 InternalTable* table = get_table(); 1182 for (size_t bucket_it = 0; bucket_it < table->_size; bucket_it++) { 1183 ScopedCS cs(thread, this); 1184 size_t count = 0; 1185 Bucket* bucket = table->get_bucket(bucket_it); 1186 if (bucket->have_redirect() || bucket->is_locked()) { 1187 continue; 1188 } 1189 Node* current_node = bucket->first(); 1190 while (current_node != NULL) { 1191 ++count; 1192 literal_bytes += vs_f(current_node->value()); 1193 current_node = current_node->next(); 1194 } 1195 summary.add((double)count); 1196 } 1197 1198 return TableStatistics(_stats_rate, summary, literal_bytes, sizeof(Bucket), sizeof(Node)); 1199 } 1200 1201 template <typename CONFIG, MEMFLAGS F> 1202 template <typename VALUE_SIZE_FUNC> 1203 inline TableStatistics ConcurrentHashTable<CONFIG, F>:: 1204 statistics_get(Thread* thread, VALUE_SIZE_FUNC& vs_f, TableStatistics old) 1205 { 1206 if (!try_resize_lock(thread)) { 1207 return old; 1208 } 1209 1210 TableStatistics ts = statistics_calculate(thread, vs_f); 1211 unlock_resize_lock(thread); 1212 1213 return ts; 1214 } 1215 1216 template <typename CONFIG, MEMFLAGS F> 1217 template <typename VALUE_SIZE_FUNC> 1218 inline void ConcurrentHashTable<CONFIG, F>:: 1219 statistics_to(Thread* thread, VALUE_SIZE_FUNC& vs_f, 1220 outputStream* st, const char* table_name) 1221 { 1222 if (!try_resize_lock(thread)) { 1223 st->print_cr("statistics unavailable at this moment"); 1224 return; 1225 } 1226 1227 TableStatistics ts = statistics_calculate(thread, vs_f); 1228 unlock_resize_lock(thread); 1229 1230 ts.print(st, table_name); 1231 } 1232 1233 template <typename CONFIG, MEMFLAGS F> 1234 inline bool ConcurrentHashTable<CONFIG, F>:: 1235 try_move_nodes_to(Thread* thread, ConcurrentHashTable<CONFIG, F>* to_cht) 1236 { 1237 if (!try_resize_lock(thread)) { 1238 return false; 1239 } 1240 assert(_new_table == NULL || _new_table == POISON_PTR, "Must be NULL"); 1241 for (size_t bucket_it = 0; bucket_it < _table->_size; bucket_it++) { 1242 Bucket* bucket = _table->get_bucket(bucket_it); 1243 assert(!bucket->have_redirect() && !bucket->is_locked(), "Table must be uncontended"); 1244 while (bucket->first() != NULL) { 1245 Node* move_node = bucket->first(); 1246 bool ok = bucket->cas_first(move_node->next(), move_node); 1247 assert(ok, "Uncontended cas must work"); 1248 bool dead_hash = false; 1249 size_t insert_hash = CONFIG::get_hash(*move_node->value(), &dead_hash); 1250 if (!dead_hash) { 1251 Bucket* insert_bucket = to_cht->get_bucket(insert_hash); 1252 assert(!bucket->have_redirect() && !bucket->is_locked(), "Not bit should be present"); 1253 move_node->set_next(insert_bucket->first()); 1254 ok = insert_bucket->cas_first(move_node, insert_bucket->first()); 1255 assert(ok, "Uncontended cas must work"); 1256 } 1257 } 1258 } 1259 unlock_resize_lock(thread); 1260 return true; 1261 } 1262 1263 #endif // SHARE_UTILITIES_CONCURRENTHASHTABLE_INLINE_HPP