1 /*
   2  * Copyright (c) 2001, 2014, 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 #include "precompiled.hpp"
  26 #include "memory/allocation.hpp"
  27 #include "memory/allocation.inline.hpp"
  28 #include "runtime/os.hpp"
  29 #include "utilities/workgroup.hpp"
  30 
  31 PRAGMA_FORMAT_MUTE_WARNINGS_FOR_GCC
  32 
  33 // Definitions of WorkGang methods.
  34 
  35 AbstractWorkGang::AbstractWorkGang(const char* name,
  36                                    bool  are_GC_task_threads,
  37                                    bool  are_ConcurrentGC_threads) :
  38   _name(name),
  39   _are_GC_task_threads(are_GC_task_threads),
  40   _are_ConcurrentGC_threads(are_ConcurrentGC_threads) {
  41 
  42   assert(!(are_GC_task_threads && are_ConcurrentGC_threads),
  43          "They cannot both be STW GC and Concurrent threads" );
  44 
  45   // Other initialization.
  46   _monitor = new Monitor(/* priority */       Mutex::leaf,
  47                          /* name */           "WorkGroup monitor",
  48                          /* allow_vm_block */ are_GC_task_threads);
  49   assert(monitor() != NULL, "Failed to allocate monitor");
  50   _terminate = false;
  51   _task = NULL;
  52   _sequence_number = 0;
  53   _started_workers = 0;
  54   _finished_workers = 0;
  55 }
  56 
  57 WorkGang::WorkGang(const char* name,
  58                    uint        workers,
  59                    bool        are_GC_task_threads,
  60                    bool        are_ConcurrentGC_threads) :
  61   AbstractWorkGang(name, are_GC_task_threads, are_ConcurrentGC_threads) {
  62   _total_workers = workers;
  63 }
  64 
  65 GangWorker* WorkGang::allocate_worker(uint which) {
  66   GangWorker* new_worker = new GangWorker(this, which);
  67   return new_worker;
  68 }
  69 
  70 // The current implementation will exit if the allocation
  71 // of any worker fails.  Still, return a boolean so that
  72 // a future implementation can possibly do a partial
  73 // initialization of the workers and report such to the
  74 // caller.
  75 bool WorkGang::initialize_workers() {
  76 
  77   if (TraceWorkGang) {
  78     tty->print_cr("Constructing work gang %s with %d threads",
  79                   name(),
  80                   total_workers());
  81   }
  82   _gang_workers = NEW_C_HEAP_ARRAY(GangWorker*, total_workers(), mtInternal);
  83   if (gang_workers() == NULL) {
  84     vm_exit_out_of_memory(0, OOM_MALLOC_ERROR, "Cannot create GangWorker array.");
  85     return false;
  86   }
  87   os::ThreadType worker_type;
  88   if (are_ConcurrentGC_threads()) {
  89     worker_type = os::cgc_thread;
  90   } else {
  91     worker_type = os::pgc_thread;
  92   }
  93   for (uint worker = 0; worker < total_workers(); worker += 1) {
  94     GangWorker* new_worker = allocate_worker(worker);
  95     assert(new_worker != NULL, "Failed to allocate GangWorker");
  96     _gang_workers[worker] = new_worker;
  97     if (new_worker == NULL || !os::create_thread(new_worker, worker_type)) {
  98       vm_exit_out_of_memory(0, OOM_MALLOC_ERROR,
  99               "Cannot create worker GC thread. Out of system resources.");
 100       return false;
 101     }
 102     if (!DisableStartThread) {
 103       os::start_thread(new_worker);
 104     }
 105   }
 106   return true;
 107 }
 108 
 109 AbstractWorkGang::~AbstractWorkGang() {
 110   if (TraceWorkGang) {
 111     tty->print_cr("Destructing work gang %s", name());
 112   }
 113   stop();   // stop all the workers
 114   for (uint worker = 0; worker < total_workers(); worker += 1) {
 115     delete gang_worker(worker);
 116   }
 117   delete gang_workers();
 118   delete monitor();
 119 }
 120 
 121 GangWorker* AbstractWorkGang::gang_worker(uint i) const {
 122   // Array index bounds checking.
 123   GangWorker* result = NULL;
 124   assert(gang_workers() != NULL, "No workers for indexing");
 125   assert(((i >= 0) && (i < total_workers())), "Worker index out of bounds");
 126   result = _gang_workers[i];
 127   assert(result != NULL, "Indexing to null worker");
 128   return result;
 129 }
 130 
 131 void WorkGang::run_task(AbstractGangTask* task) {
 132   run_task(task, total_workers());
 133 }
 134 
 135 void WorkGang::run_task(AbstractGangTask* task, uint no_of_parallel_workers) {
 136   task->set_for_termination(no_of_parallel_workers);
 137 
 138   // This thread is executed by the VM thread which does not block
 139   // on ordinary MutexLocker's.
 140   MutexLockerEx ml(monitor(), Mutex::_no_safepoint_check_flag);
 141   if (TraceWorkGang) {
 142     tty->print_cr("Running work gang %s task %s", name(), task->name());
 143   }
 144   // Tell all the workers to run a task.
 145   assert(task != NULL, "Running a null task");
 146   // Initialize.
 147   _task = task;
 148   _sequence_number += 1;
 149   _started_workers = 0;
 150   _finished_workers = 0;
 151   // Tell the workers to get to work.
 152   monitor()->notify_all();
 153   // Wait for them to be finished
 154   while (finished_workers() < no_of_parallel_workers) {
 155     if (TraceWorkGang) {
 156       tty->print_cr("Waiting in work gang %s: %d/%d finished sequence %d",
 157                     name(), finished_workers(), no_of_parallel_workers,
 158                     _sequence_number);
 159     }
 160     monitor()->wait(/* no_safepoint_check */ true);
 161   }
 162   _task = NULL;
 163   if (TraceWorkGang) {
 164     tty->print_cr("\nFinished work gang %s: %d/%d sequence %d",
 165                   name(), finished_workers(), no_of_parallel_workers,
 166                   _sequence_number);
 167     Thread* me = Thread::current();
 168     tty->print_cr("  T: 0x%x  VM_thread: %d", me, me->is_VM_thread());
 169   }
 170 }
 171 
 172 void FlexibleWorkGang::run_task(AbstractGangTask* task) {
 173   // If active_workers() is passed, _finished_workers
 174   // must only be incremented for workers that find non_null
 175   // work (as opposed to all those that just check that the
 176   // task is not null).
 177   WorkGang::run_task(task, (uint) active_workers());
 178 }
 179 
 180 void AbstractWorkGang::stop() {
 181   // Tell all workers to terminate, then wait for them to become inactive.
 182   MutexLockerEx ml(monitor(), Mutex::_no_safepoint_check_flag);
 183   if (TraceWorkGang) {
 184     tty->print_cr("Stopping work gang %s task %s", name(), task()->name());
 185   }
 186   _task = NULL;
 187   _terminate = true;
 188   monitor()->notify_all();
 189   while (finished_workers() < active_workers()) {
 190     if (TraceWorkGang) {
 191       tty->print_cr("Waiting in work gang %s: %d/%d finished",
 192                     name(), finished_workers(), active_workers());
 193     }
 194     monitor()->wait(/* no_safepoint_check */ true);
 195   }
 196 }
 197 
 198 void AbstractWorkGang::internal_worker_poll(WorkData* data) const {
 199   assert(monitor()->owned_by_self(), "worker_poll is an internal method");
 200   assert(data != NULL, "worker data is null");
 201   data->set_terminate(terminate());
 202   data->set_task(task());
 203   data->set_sequence_number(sequence_number());
 204 }
 205 
 206 void AbstractWorkGang::internal_note_start() {
 207   assert(monitor()->owned_by_self(), "note_finish is an internal method");
 208   _started_workers += 1;
 209 }
 210 
 211 void AbstractWorkGang::internal_note_finish() {
 212   assert(monitor()->owned_by_self(), "note_finish is an internal method");
 213   _finished_workers += 1;
 214 }
 215 
 216 void AbstractWorkGang::print_worker_threads_on(outputStream* st) const {
 217   uint    num_thr = total_workers();
 218   for (uint i = 0; i < num_thr; i++) {
 219     gang_worker(i)->print_on(st);
 220     st->cr();
 221   }
 222 }
 223 
 224 void AbstractWorkGang::threads_do(ThreadClosure* tc) const {
 225   assert(tc != NULL, "Null ThreadClosure");
 226   uint num_thr = total_workers();
 227   for (uint i = 0; i < num_thr; i++) {
 228     tc->do_thread(gang_worker(i));
 229   }
 230 }
 231 
 232 // GangWorker methods.
 233 
 234 GangWorker::GangWorker(AbstractWorkGang* gang, uint id) {
 235   _gang = gang;
 236   set_id(id);
 237   set_name("Gang worker#%d (%s)", id, gang->name());
 238 }
 239 
 240 void GangWorker::run() {
 241   initialize();
 242   loop();
 243 }
 244 
 245 void GangWorker::initialize() {
 246   this->initialize_thread_local_storage();
 247   this->record_stack_base_and_size();
 248   assert(_gang != NULL, "No gang to run in");
 249   os::set_priority(this, NearMaxPriority);
 250   if (TraceWorkGang) {
 251     tty->print_cr("Running gang worker for gang %s id %d",
 252                   gang()->name(), id());
 253   }
 254   // The VM thread should not execute here because MutexLocker's are used
 255   // as (opposed to MutexLockerEx's).
 256   assert(!Thread::current()->is_VM_thread(), "VM thread should not be part"
 257          " of a work gang");
 258 }
 259 
 260 void GangWorker::loop() {
 261   int previous_sequence_number = 0;
 262   Monitor* gang_monitor = gang()->monitor();
 263   for ( ; /* !terminate() */; ) {
 264     WorkData data;
 265     int part;  // Initialized below.
 266     {
 267       // Grab the gang mutex.
 268       MutexLocker ml(gang_monitor);
 269       // Wait for something to do.
 270       // Polling outside the while { wait } avoids missed notifies
 271       // in the outer loop.
 272       gang()->internal_worker_poll(&data);
 273       if (TraceWorkGang) {
 274         tty->print("Polled outside for work in gang %s worker %d",
 275                    gang()->name(), id());
 276         tty->print("  terminate: %s",
 277                    data.terminate() ? "true" : "false");
 278         tty->print("  sequence: %d (prev: %d)",
 279                    data.sequence_number(), previous_sequence_number);
 280         if (data.task() != NULL) {
 281           tty->print("  task: %s", data.task()->name());
 282         } else {
 283           tty->print("  task: NULL");
 284         }
 285         tty->cr();
 286       }
 287       for ( ; /* break or return */; ) {
 288         // Terminate if requested.
 289         if (data.terminate()) {
 290           gang()->internal_note_finish();
 291           gang_monitor->notify_all();
 292           return;
 293         }
 294         // Check for new work.
 295         if ((data.task() != NULL) &&
 296             (data.sequence_number() != previous_sequence_number)) {
 297           if (gang()->needs_more_workers()) {
 298             gang()->internal_note_start();
 299             gang_monitor->notify_all();
 300             part = gang()->started_workers() - 1;
 301             break;
 302           }
 303         }
 304         // Nothing to do.
 305         gang_monitor->wait(/* no_safepoint_check */ true);
 306         gang()->internal_worker_poll(&data);
 307         if (TraceWorkGang) {
 308           tty->print("Polled inside for work in gang %s worker %d",
 309                      gang()->name(), id());
 310           tty->print("  terminate: %s",
 311                      data.terminate() ? "true" : "false");
 312           tty->print("  sequence: %d (prev: %d)",
 313                      data.sequence_number(), previous_sequence_number);
 314           if (data.task() != NULL) {
 315             tty->print("  task: %s", data.task()->name());
 316           } else {
 317             tty->print("  task: NULL");
 318           }
 319           tty->cr();
 320         }
 321       }
 322       // Drop gang mutex.
 323     }
 324     if (TraceWorkGang) {
 325       tty->print("Work for work gang %s id %d task %s part %d",
 326                  gang()->name(), id(), data.task()->name(), part);
 327     }
 328     assert(data.task() != NULL, "Got null task");
 329     data.task()->work(part);
 330     {
 331       if (TraceWorkGang) {
 332         tty->print("Finish for work gang %s id %d task %s part %d",
 333                    gang()->name(), id(), data.task()->name(), part);
 334       }
 335       // Grab the gang mutex.
 336       MutexLocker ml(gang_monitor);
 337       gang()->internal_note_finish();
 338       // Tell the gang you are done.
 339       gang_monitor->notify_all();
 340       // Drop the gang mutex.
 341     }
 342     previous_sequence_number = data.sequence_number();
 343   }
 344 }
 345 
 346 bool GangWorker::is_GC_task_thread() const {
 347   return gang()->are_GC_task_threads();
 348 }
 349 
 350 bool GangWorker::is_ConcurrentGC_thread() const {
 351   return gang()->are_ConcurrentGC_threads();
 352 }
 353 
 354 void GangWorker::print_on(outputStream* st) const {
 355   st->print("\"%s\" ", name());
 356   Thread::print_on(st);
 357   st->cr();
 358 }
 359 
 360 // Printing methods
 361 
 362 const char* AbstractWorkGang::name() const {
 363   return _name;
 364 }
 365 
 366 #ifndef PRODUCT
 367 
 368 const char* AbstractGangTask::name() const {
 369   return _name;
 370 }
 371 
 372 #endif /* PRODUCT */
 373 
 374 // FlexibleWorkGang
 375 
 376 
 377 // *** WorkGangBarrierSync
 378 
 379 WorkGangBarrierSync::WorkGangBarrierSync()
 380   : _monitor(Mutex::safepoint, "work gang barrier sync", true),
 381     _n_workers(0), _n_completed(0), _should_reset(false), _aborted(false) {
 382 }
 383 
 384 WorkGangBarrierSync::WorkGangBarrierSync(uint n_workers, const char* name)
 385   : _monitor(Mutex::safepoint, name, true),
 386     _n_workers(n_workers), _n_completed(0), _should_reset(false), _aborted(false) {
 387 }
 388 
 389 void WorkGangBarrierSync::set_n_workers(uint n_workers) {
 390   _n_workers    = n_workers;
 391   _n_completed  = 0;
 392   _should_reset = false;
 393   _aborted      = false;
 394 }
 395 
 396 bool WorkGangBarrierSync::enter() {
 397   MutexLockerEx x(monitor(), Mutex::_no_safepoint_check_flag);
 398   if (should_reset()) {
 399     // The should_reset() was set and we are the first worker to enter
 400     // the sync barrier. We will zero the n_completed() count which
 401     // effectively resets the barrier.
 402     zero_completed();
 403     set_should_reset(false);
 404   }
 405   inc_completed();
 406   if (n_completed() == n_workers()) {
 407     // At this point we would like to reset the barrier to be ready in
 408     // case it is used again. However, we cannot set n_completed() to
 409     // 0, even after the notify_all(), given that some other workers
 410     // might still be waiting for n_completed() to become ==
 411     // n_workers(). So, if we set n_completed() to 0, those workers
 412     // will get stuck (as they will wake up, see that n_completed() !=
 413     // n_workers() and go back to sleep). Instead, we raise the
 414     // should_reset() flag and the barrier will be reset the first
 415     // time a worker enters it again.
 416     set_should_reset(true);
 417     monitor()->notify_all();
 418   } else {
 419     while (n_completed() != n_workers() && !aborted()) {
 420       monitor()->wait(/* no_safepoint_check */ true);
 421     }
 422   }
 423   return !aborted();
 424 }
 425 
 426 void WorkGangBarrierSync::abort() {
 427   MutexLockerEx x(monitor(), Mutex::_no_safepoint_check_flag);
 428   set_aborted();
 429   monitor()->notify_all();
 430 }
 431 
 432 // SubTasksDone functions.
 433 
 434 SubTasksDone::SubTasksDone(uint n) :
 435   _n_tasks(n), _n_threads(1), _tasks(NULL) {
 436   _tasks = NEW_C_HEAP_ARRAY(uint, n, mtInternal);
 437   guarantee(_tasks != NULL, "alloc failure");
 438   clear();
 439 }
 440 
 441 bool SubTasksDone::valid() {
 442   return _tasks != NULL;
 443 }
 444 
 445 void SubTasksDone::set_n_threads(uint t) {
 446   assert(_claimed == 0 || _threads_completed == _n_threads,
 447          "should not be called while tasks are being processed!");
 448   _n_threads = (t == 0 ? 1 : t);
 449 }
 450 
 451 void SubTasksDone::clear() {
 452   for (uint i = 0; i < _n_tasks; i++) {
 453     _tasks[i] = 0;
 454   }
 455   _threads_completed = 0;
 456 #ifdef ASSERT
 457   _claimed = 0;
 458 #endif
 459 }
 460 
 461 bool SubTasksDone::is_task_claimed(uint t) {
 462   assert(0 <= t && t < _n_tasks, "bad task id.");
 463   uint old = _tasks[t];
 464   if (old == 0) {
 465     old = Atomic::cmpxchg(1, &_tasks[t], 0);
 466   }
 467   assert(_tasks[t] == 1, "What else?");
 468   bool res = old != 0;
 469 #ifdef ASSERT
 470   if (!res) {
 471     assert(_claimed < _n_tasks, "Too many tasks claimed; missing clear?");
 472     Atomic::inc((volatile jint*) &_claimed);
 473   }
 474 #endif
 475   return res;
 476 }
 477 
 478 void SubTasksDone::all_tasks_completed() {
 479   jint observed = _threads_completed;
 480   jint old;
 481   do {
 482     old = observed;
 483     observed = Atomic::cmpxchg(old+1, &_threads_completed, old);
 484   } while (observed != old);
 485   // If this was the last thread checking in, clear the tasks.
 486   if (observed+1 == (jint)_n_threads) clear();
 487 }
 488 
 489 
 490 SubTasksDone::~SubTasksDone() {
 491   if (_tasks != NULL) FREE_C_HEAP_ARRAY(jint, _tasks, mtInternal);
 492 }
 493 
 494 // *** SequentialSubTasksDone
 495 
 496 void SequentialSubTasksDone::clear() {
 497   _n_tasks   = _n_claimed   = 0;
 498   _n_threads = _n_completed = 0;
 499 }
 500 
 501 bool SequentialSubTasksDone::valid() {
 502   return _n_threads > 0;
 503 }
 504 
 505 bool SequentialSubTasksDone::is_task_claimed(uint& t) {
 506   uint* n_claimed_ptr = &_n_claimed;
 507   t = *n_claimed_ptr;
 508   while (t < _n_tasks) {
 509     jint res = Atomic::cmpxchg(t+1, n_claimed_ptr, t);
 510     if (res == (jint)t) {
 511       return false;
 512     }
 513     t = *n_claimed_ptr;
 514   }
 515   return true;
 516 }
 517 
 518 bool SequentialSubTasksDone::all_tasks_completed() {
 519   uint* n_completed_ptr = &_n_completed;
 520   uint  complete        = *n_completed_ptr;
 521   while (true) {
 522     uint res = Atomic::cmpxchg(complete+1, n_completed_ptr, complete);
 523     if (res == complete) {
 524       break;
 525     }
 526     complete = res;
 527   }
 528   if (complete+1 == _n_threads) {
 529     clear();
 530     return true;
 531   }
 532   return false;
 533 }
 534 
 535 bool FreeIdSet::_stat_init = false;
 536 FreeIdSet* FreeIdSet::_sets[NSets];
 537 bool FreeIdSet::_safepoint;
 538 
 539 FreeIdSet::FreeIdSet(int sz, Monitor* mon) :
 540   _sz(sz), _mon(mon), _hd(0), _waiters(0), _index(-1), _claimed(0)
 541 {
 542   _ids = NEW_C_HEAP_ARRAY(int, sz, mtInternal);
 543   for (int i = 0; i < sz; i++) _ids[i] = i+1;
 544   _ids[sz-1] = end_of_list; // end of list.
 545   if (_stat_init) {
 546     for (int j = 0; j < NSets; j++) _sets[j] = NULL;
 547     _stat_init = true;
 548   }
 549   // Add to sets.  (This should happen while the system is still single-threaded.)
 550   for (int j = 0; j < NSets; j++) {
 551     if (_sets[j] == NULL) {
 552       _sets[j] = this;
 553       _index = j;
 554       break;
 555     }
 556   }
 557   guarantee(_index != -1, "Too many FreeIdSets in use!");
 558 }
 559 
 560 FreeIdSet::~FreeIdSet() {
 561   _sets[_index] = NULL;
 562   FREE_C_HEAP_ARRAY(int, _ids, mtInternal);
 563 }
 564 
 565 void FreeIdSet::set_safepoint(bool b) {
 566   _safepoint = b;
 567   if (b) {
 568     for (int j = 0; j < NSets; j++) {
 569       if (_sets[j] != NULL && _sets[j]->_waiters > 0) {
 570         Monitor* mon = _sets[j]->_mon;
 571         mon->lock_without_safepoint_check();
 572         mon->notify_all();
 573         mon->unlock();
 574       }
 575     }
 576   }
 577 }
 578 
 579 #define FID_STATS 0
 580 
 581 int FreeIdSet::claim_par_id() {
 582 #if FID_STATS
 583   thread_t tslf = thr_self();
 584   tty->print("claim_par_id[%d]: sz = %d, claimed = %d\n", tslf, _sz, _claimed);
 585 #endif
 586   MutexLockerEx x(_mon, Mutex::_no_safepoint_check_flag);
 587   while (!_safepoint && _hd == end_of_list) {
 588     _waiters++;
 589 #if FID_STATS
 590     if (_waiters > 5) {
 591       tty->print("claim_par_id waiting[%d]: %d waiters, %d claimed.\n",
 592                  tslf, _waiters, _claimed);
 593     }
 594 #endif
 595     _mon->wait(Mutex::_no_safepoint_check_flag);
 596     _waiters--;
 597   }
 598   if (_hd == end_of_list) {
 599 #if FID_STATS
 600     tty->print("claim_par_id[%d]: returning EOL.\n", tslf);
 601 #endif
 602     return -1;
 603   } else {
 604     int res = _hd;
 605     _hd = _ids[res];
 606     _ids[res] = claimed;  // For debugging.
 607     _claimed++;
 608 #if FID_STATS
 609     tty->print("claim_par_id[%d]: returning %d, claimed = %d.\n",
 610                tslf, res, _claimed);
 611 #endif
 612     return res;
 613   }
 614 }
 615 
 616 bool FreeIdSet::claim_perm_id(int i) {
 617   assert(0 <= i && i < _sz, "Out of range.");
 618   MutexLockerEx x(_mon, Mutex::_no_safepoint_check_flag);
 619   int prev = end_of_list;
 620   int cur = _hd;
 621   while (cur != end_of_list) {
 622     if (cur == i) {
 623       if (prev == end_of_list) {
 624         _hd = _ids[cur];
 625       } else {
 626         _ids[prev] = _ids[cur];
 627       }
 628       _ids[cur] = claimed;
 629       _claimed++;
 630       return true;
 631     } else {
 632       prev = cur;
 633       cur = _ids[cur];
 634     }
 635   }
 636   return false;
 637 
 638 }
 639 
 640 void FreeIdSet::release_par_id(int id) {
 641   MutexLockerEx x(_mon, Mutex::_no_safepoint_check_flag);
 642   assert(_ids[id] == claimed, "Precondition.");
 643   _ids[id] = _hd;
 644   _hd = id;
 645   _claimed--;
 646 #if FID_STATS
 647   tty->print("[%d] release_par_id(%d), waiters =%d,  claimed = %d.\n",
 648              thr_self(), id, _waiters, _claimed);
 649 #endif
 650   if (_waiters > 0)
 651     // Notify all would be safer, but this is OK, right?
 652     _mon->notify_all();
 653 }