view src/share/vm/gc_implementation/parallelScavenge/psPermGen.cpp @ 2149:7e37af9d69ef

7011379: G1: overly long concurrent marking cycles Summary: This changeset introduces filtering of SATB buffers at the point when they are about to be enqueued. If this filtering clears enough entries on each buffer, the buffer can then be re-used and not enqueued. This cuts down the number of SATB buffers that need to be processed by the concurrent marking threads. Reviewed-by: johnc, ysr
author tonyp
date Wed, 19 Jan 2011 09:35:17 -0500
parents f95d63e2154a
children eda9eb483d29
line wrap: on
line source

/*
 * Copyright (c) 2001, 2010, Oracle and/or its affiliates. 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 Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
 * or visit www.oracle.com if you need additional information or have any
 * questions.
 *
 */

#include "precompiled.hpp"
#include "gc_implementation/parallelScavenge/parallelScavengeHeap.hpp"
#include "gc_implementation/parallelScavenge/psMarkSweepDecorator.hpp"
#include "gc_implementation/parallelScavenge/psParallelCompact.hpp"
#include "gc_implementation/parallelScavenge/psPermGen.hpp"
#include "gc_implementation/shared/gcUtil.hpp"
#include "gc_implementation/shared/markSweep.inline.hpp"
#include "oops/markOop.inline.hpp"

PSPermGen::PSPermGen(ReservedSpace rs, size_t alignment,
                     size_t initial_size, size_t min_size, size_t max_size,
                     const char* gen_name, int level) :
  PSOldGen(rs, alignment, initial_size, min_size, max_size, gen_name, level),
  _last_used(0)
{
  assert(object_mark_sweep() != NULL, "Sanity");

  object_mark_sweep()->set_allowed_dead_ratio(PermMarkSweepDeadRatio);
  _avg_size = new AdaptivePaddedAverage(AdaptivePermSizeWeight,
                                        PermGenPadding);
}

HeapWord* PSPermGen::allocate_permanent(size_t size) {
  assert_locked_or_safepoint(Heap_lock);
  HeapWord* obj = allocate_noexpand(size, false);

  if (obj == NULL) {
    obj = expand_and_allocate(size, false);
  }

  return obj;
}

void PSPermGen::compute_new_size(size_t used_before_collection) {
  // Update our padded average of objects allocated in perm
  // gen between collections.
  assert(used_before_collection >= _last_used,
                                "negative allocation amount since last GC?");

  const size_t alloc_since_last_gc = used_before_collection - _last_used;
  _avg_size->sample(alloc_since_last_gc);

  const size_t current_live = used_in_bytes();
  // Stash away the current amount live for the next call to this method.
  _last_used = current_live;

  // We have different alignment constraints than the rest of the heap.
  const size_t alignment = MAX2(MinPermHeapExpansion,
                                virtual_space()->alignment());

  // Compute the desired size:
  //  The free space is the newly computed padded average,
  //  so the desired size is what's live + the free space.
  size_t desired_size = current_live + (size_t)_avg_size->padded_average();
  desired_size = align_size_up(desired_size, alignment);

  // ...and no larger or smaller than our max and min allowed.
  desired_size = MAX2(MIN2(desired_size, _max_gen_size), _min_gen_size);
  assert(desired_size <= _max_gen_size, "just checking");

  const size_t size_before = _virtual_space->committed_size();

  if (desired_size == size_before) {
    // no change, we're done
    return;
  }

  {
    // We'll be growing or shrinking the heap:  in either case,
    // we need to hold a lock.
    MutexLocker x(ExpandHeap_lock);
    if (desired_size > size_before) {
      const size_t change_bytes = desired_size - size_before;
      const size_t aligned_change_bytes =
        align_size_up(change_bytes, alignment);
      expand_by(aligned_change_bytes);
    } else {
      // Shrinking
      const size_t change_bytes =
        size_before - desired_size;
      const size_t aligned_change_bytes = align_size_down(change_bytes, alignment);
      shrink(aligned_change_bytes);
    }
  }

  // While this code isn't controlled by AdaptiveSizePolicy, it's
  // convenient to see all resizing decsions under the same flag.
  if (PrintAdaptiveSizePolicy) {
    ParallelScavengeHeap* heap = (ParallelScavengeHeap*)Universe::heap();
    assert(heap->kind() == CollectedHeap::ParallelScavengeHeap, "Sanity");

    gclog_or_tty->print_cr("AdaptiveSizePolicy::perm generation size: "
                           "collection: %d "
                           "(" SIZE_FORMAT ") -> (" SIZE_FORMAT ") ",
                           heap->total_collections(),
                           size_before, _virtual_space->committed_size());
  }
}



void PSPermGen::move_and_update(ParCompactionManager* cm) {
  PSParallelCompact::move_and_update(cm, PSParallelCompact::perm_space_id);
}

void PSPermGen::precompact() {
  // Reset start array first.
  _start_array.reset();
  object_mark_sweep()->precompact();
}