view src/share/vm/gc_implementation/g1/concurrentZFThread.cpp @ 452:00b023ae2d78

6722113: CMS: Incorrect overflow handling during precleaning of Reference lists Summary: When we encounter marking stack overflow during precleaning of Reference lists, we were using the overflow list mechanism, which can cause problems on account of mutating the mark word of the header because of conflicts with mutator accesses and updates of that field. Instead we should use the usual mechanism for overflow handling in concurrent phases, namely dirtying of the card on which the overflowed object lies. Since precleaning effectively does a form of discovered list processing, albeit with discovery enabled, we needed to adjust some code to be correct in the face of interleaved processing and discovery. Reviewed-by: apetrusenko, jcoomes
author ysr
date Thu, 20 Nov 2008 12:27:41 -0800
parents 37f87013dfd8
children e1fdf4fd34dc
line wrap: on
line source

/*
 * Copyright 2001-2007 Sun Microsystems, Inc.  All Rights Reserved.
 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
 *
 * This code is free software; you can redistribute it and/or modify it
 * under the terms of the GNU General Public License version 2 only, as
 * published by the Free Software Foundation.
 *
 * This code is distributed in the hope that it will be useful, but WITHOUT
 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
 * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
 * version 2 for more details (a copy is included in the LICENSE file that
 * accompanied this code).
 *
 * You should have received a copy of the GNU General Public License version
 * 2 along with this work; if not, write to the Free Software Foundation,
 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
 *
 * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara,
 * CA 95054 USA or visit www.sun.com if you need additional information or
 * have any questions.
 *
 */

#include "incls/_precompiled.incl"
#include "incls/_concurrentZFThread.cpp.incl"

// ======= Concurrent Zero-Fill Thread ========

// The CM thread is created when the G1 garbage collector is used

int ConcurrentZFThread::_region_allocs = 0;
int ConcurrentZFThread::_sync_zfs = 0;
int ConcurrentZFThread::_zf_waits = 0;
int ConcurrentZFThread::_regions_filled = 0;

ConcurrentZFThread::ConcurrentZFThread() :
  ConcurrentGCThread(),
  _co_tracker(G1ZFGroup)
{
  create_and_start();
}

void ConcurrentZFThread::wait_for_ZF_completed(HeapRegion* hr) {
  assert(ZF_mon->owned_by_self(), "Precondition.");
  note_zf_wait();
  while (hr->zero_fill_state() == HeapRegion::ZeroFilling) {
    ZF_mon->wait(Mutex::_no_safepoint_check_flag);
  }
}

void ConcurrentZFThread::processHeapRegion(HeapRegion* hr) {
  assert(!Universe::heap()->is_gc_active(),
         "This should not happen during GC.");
  assert(hr != NULL, "Precondition");
  // These are unlocked reads, but if this test is successful, then no
  // other thread will attempt this zero filling.  Only a GC thread can
  // modify the ZF state of a region whose state is zero-filling, and this
  // should only happen while the ZF thread is locking out GC.
  if (hr->zero_fill_state() == HeapRegion::ZeroFilling
      && hr->zero_filler() == Thread::current()) {
    assert(hr->top() == hr->bottom(), "better be empty!");
    assert(!hr->isHumongous(), "Only free regions on unclean list.");
    Copy::fill_to_words(hr->bottom(), hr->capacity()/HeapWordSize);
    note_region_filled();
  }
}

void ConcurrentZFThread::run() {
  initialize_in_thread();
  Thread* thr_self = Thread::current();
  _vtime_start = os::elapsedVTime();
  wait_for_universe_init();
  _co_tracker.enable();
  _co_tracker.start();

  G1CollectedHeap* g1 = G1CollectedHeap::heap();
  _sts.join();
  while (!_should_terminate) {
    _sts.leave();

    {
      MutexLockerEx x(ZF_mon, Mutex::_no_safepoint_check_flag);

      // This local variable will hold a region being zero-filled.  This
      // region will neither be on the unclean or zero-filled lists, and
      // will not be available for allocation; thus, we might have an
      // allocation fail, causing a full GC, because of this, but this is a
      // price we will pay.  (In future, we might want to make the fact
      // that there's a region being zero-filled apparent to the G1 heap,
      // which could then wait for it in this extreme case...)
      HeapRegion* to_fill;

      while (!g1->should_zf()
             || (to_fill = g1->pop_unclean_region_list_locked()) == NULL)
        ZF_mon->wait(Mutex::_no_safepoint_check_flag);
      while (to_fill->zero_fill_state() == HeapRegion::ZeroFilling)
        ZF_mon->wait(Mutex::_no_safepoint_check_flag);

      // So now to_fill is non-NULL and is not ZeroFilling.  It might be
      // Allocated or ZeroFilled.  (The latter could happen if this thread
      // starts the zero-filling of a region, but a GC intervenes and
      // pushes new regions needing on the front of the filling on the
      // front of the list.)

      switch (to_fill->zero_fill_state()) {
      case HeapRegion::Allocated:
        to_fill = NULL;
        break;

      case HeapRegion::NotZeroFilled:
        to_fill->set_zero_fill_in_progress(thr_self);

        ZF_mon->unlock();
        _sts.join();
        processHeapRegion(to_fill);
        _sts.leave();
        ZF_mon->lock_without_safepoint_check();

        if (to_fill->zero_fill_state() == HeapRegion::ZeroFilling
            && to_fill->zero_filler() == thr_self) {
          to_fill->set_zero_fill_complete();
          (void)g1->put_free_region_on_list_locked(to_fill);
        }
        break;

      case HeapRegion::ZeroFilled:
        (void)g1->put_free_region_on_list_locked(to_fill);
        break;

      case HeapRegion::ZeroFilling:
        ShouldNotReachHere();
        break;
      }
    }
    _vtime_accum = (os::elapsedVTime() - _vtime_start);
    _sts.join();

    _co_tracker.update();
  }
  _co_tracker.update(false);
  _sts.leave();

  assert(_should_terminate, "just checking");
  terminate();
}

bool ConcurrentZFThread::offer_yield() {
  if (_sts.should_yield()) {
    _sts.yield("Concurrent ZF");
    return true;
  } else {
    return false;
  }
}

void ConcurrentZFThread::stop() {
  // it is ok to take late safepoints here, if needed
  MutexLockerEx mu(Terminator_lock);
  _should_terminate = true;
  while (!_has_terminated) {
    Terminator_lock->wait();
  }
}

void ConcurrentZFThread::print() {
  gclog_or_tty->print("\"Concurrent ZF Thread\" ");
  Thread::print();
  gclog_or_tty->cr();
}


double ConcurrentZFThread::_vtime_accum;

void ConcurrentZFThread::print_summary_info() {
  gclog_or_tty->print("\nConcurrent Zero-Filling:\n");
  gclog_or_tty->print("  Filled %d regions, used %5.2fs.\n",
                      _regions_filled,
                      vtime_accum());
  gclog_or_tty->print("  Of %d region allocs, %d (%5.2f%%) required sync ZF,\n",
                      _region_allocs, _sync_zfs,
                      (_region_allocs > 0 ?
                       (float)_sync_zfs/(float)_region_allocs*100.0 :
                       0.0));
  gclog_or_tty->print("     and %d (%5.2f%%) required a ZF wait.\n",
                      _zf_waits,
                      (_region_allocs > 0 ?
                       (float)_zf_waits/(float)_region_allocs*100.0 :
                       0.0));

}