jdk-24/src/hotspot/share/gc/g1/g1RemSet.cpp

/*
 * Copyright (c) 2001, 2019, 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/g1/g1BarrierSet.hpp"
#include "gc/g1/g1BlockOffsetTable.inline.hpp"
#include "gc/g1/g1CardTable.inline.hpp"
#include "gc/g1/g1CollectedHeap.inline.hpp"
#include "gc/g1/g1ConcurrentRefine.hpp"
#include "gc/g1/g1DirtyCardQueue.hpp"
#include "gc/g1/g1FromCardCache.hpp"
#include "gc/g1/g1GCPhaseTimes.hpp"
#include "gc/g1/g1HotCardCache.hpp"
#include "gc/g1/g1OopClosures.inline.hpp"
#include "gc/g1/g1RootClosures.hpp"
#include "gc/g1/g1RemSet.hpp"
#include "gc/g1/g1SharedDirtyCardQueue.hpp"
#include "gc/g1/heapRegion.inline.hpp"
#include "gc/g1/heapRegionManager.inline.hpp"
#include "gc/g1/heapRegionRemSet.inline.hpp"
#include "gc/g1/sparsePRT.hpp"
#include "gc/shared/gcTraceTime.inline.hpp"
#include "gc/shared/suspendibleThreadSet.hpp"
#include "jfr/jfrEvents.hpp"
#include "memory/iterator.hpp"
#include "memory/resourceArea.hpp"
#include "oops/access.inline.hpp"
#include "oops/oop.inline.hpp"
#include "runtime/os.hpp"
#include "utilities/align.hpp"
#include "utilities/globalDefinitions.hpp"
#include "utilities/stack.inline.hpp"
#include "utilities/ticks.hpp"

// Collects information about the overall heap root scan progress during an evacuation.
//
// Scanning the remembered sets works by first merging all sources of cards to be
// scanned (log buffers, hcc, remembered sets) into a single data structure to remove
// duplicates and simplify work distribution.
//
// During the following card scanning we not only scan this combined set of cards, but
// also remember that these were completely scanned. The following evacuation passes
// do not scan these cards again, and so need to be preserved across increments.
//
// The representation for all the cards to scan is the card table: cards can have
// one of three states during GC:
// - clean: these cards will not be scanned in this pass
// - dirty: these cards will be scanned in this pass
// - scanned: these cards have already been scanned in a previous pass
//
// After all evacuation is done, we reset the card table to clean.
//
// Work distribution occurs on "chunk" basis, i.e. contiguous ranges of cards. As an
// additional optimization, during card merging we remember which regions and which
// chunks actually contain cards to be scanned. Threads iterate only across these
// regions, and only compete for chunks containing any cards.
//
// Within these chunks, a worker scans the card table on "blocks" of cards, i.e.
// contiguous ranges of dirty cards to be scanned. These blocks are converted to actual
// memory ranges and then passed on to actual scanning.
class G1RemSetScanState : public CHeapObj<mtGC> {
  class G1DirtyRegions;

  size_t _max_regions;

  // Has this region that is part of the regions in the collection set been processed yet.
  typedef bool G1RemsetIterState;

  G1RemsetIterState volatile* _collection_set_iter_state;

  // Card table iteration claim for each heap region, from 0 (completely unscanned)
  // to (>=) HeapRegion::CardsPerRegion (completely scanned).
  uint volatile* _card_table_scan_state;

  // Random power of two number of cards we want to claim per thread. This corresponds
  // to a 64k of memory work chunk area for every thread.
  // We use the same claim size as Parallel GC. No particular measurements have been
  // performed to determine an optimal number.
  static const uint CardsPerChunk = 128;

  uint _scan_chunks_per_region;
  bool* _region_scan_chunks;
  uint8_t _scan_chunks_shift;
public:
  uint scan_chunk_size() const { return (uint)1 << _scan_chunks_shift; }

  // Returns whether the chunk corresponding to the given region/card in region contain a
  // dirty card, i.e. actually needs scanning.
  bool chunk_needs_scan(uint const region_idx, uint const card_in_region) const {
    size_t const idx = (size_t)region_idx * _scan_chunks_per_region + (card_in_region >> _scan_chunks_shift);
    assert(idx < (_max_regions * _scan_chunks_per_region), "Index " SIZE_FORMAT " out of bounds " SIZE_FORMAT,
           idx, _max_regions * _scan_chunks_per_region);
    return _region_scan_chunks[idx];
  }

private:
  // The complete set of regions which card table needs to be cleared at the end of GC because
  // we scribbled all over them.
  G1DirtyRegions* _all_dirty_regions;
  // The set of regions which card table needs to be scanned for new dirty cards
  // in the current evacuation pass.
  G1DirtyRegions* _next_dirty_regions;

  // Set of (unique) regions that can be added to concurrently.
  class G1DirtyRegions : public CHeapObj<mtGC> {
    uint* _buffer;
    uint _cur_idx;
    size_t _max_regions;

    bool* _contains;

  public:
    G1DirtyRegions(size_t max_regions) :
      _buffer(NEW_C_HEAP_ARRAY(uint, max_regions, mtGC)),
      _cur_idx(0),
      _max_regions(max_regions),
      _contains(NEW_C_HEAP_ARRAY(bool, max_regions, mtGC)) {

      reset();
    }

    static size_t chunk_size() { return M; }

    ~G1DirtyRegions() {
      FREE_C_HEAP_ARRAY(uint, _buffer);
      FREE_C_HEAP_ARRAY(bool, _contains);
    }

    void reset() {
      _cur_idx = 0;
      ::memset(_contains, false, _max_regions * sizeof(bool));
    }

    uint size() const { return _cur_idx; }

    uint at(uint idx) const {
      assert(idx < _cur_idx, "Index %u beyond valid regions", idx);
      return _buffer[idx];
    }

    void add_dirty_region(uint region) {
      if (_contains[region]) {
        return;
      }

      bool marked_as_dirty = Atomic::cmpxchg(true, &_contains[region], false) == false;
      if (marked_as_dirty) {
        uint allocated = Atomic::add(1u, &_cur_idx) - 1;
        _buffer[allocated] = region;
      }
    }

    // Creates the union of this and the other G1DirtyRegions.
    void merge(const G1DirtyRegions* other) {
      for (uint i = 0; i < other->size(); i++) {
        uint region = other->at(i);
        if (!_contains[region]) {
          _buffer[_cur_idx++] = region;
          _contains[region] = true;
        }
      }
    }
  };

  // Returns whether the given region contains cards we need to scan. The remembered
  // set and other sources may contain cards that
  // - are in uncommitted regions
  // - are located in the collection set
  // - are located in free regions
  // as we do not clean up remembered sets before merging heap roots.
  bool contains_cards_to_process(uint const region_idx) const {
    HeapRegion* hr = G1CollectedHeap::heap()->region_at_or_null(region_idx);
    return (hr != NULL && !hr->in_collection_set() && hr->is_old_or_humongous_or_archive());
  }

  class G1MergeCardSetClosure : public HeapRegionClosure {
    G1RemSetScanState* _scan_state;
    G1CardTable* _ct;

    uint _merged_sparse;
    uint _merged_fine;
    uint _merged_coarse;

    // Returns if the region contains cards we need to scan. If so, remember that
    // region in the current set of dirty regions.
    bool remember_if_interesting(uint const region_idx) {
      if (!_scan_state->contains_cards_to_process(region_idx)) {
        return false;
      }
      _scan_state->add_dirty_region(region_idx);
      return true;
    }
  public:
    G1MergeCardSetClosure(G1RemSetScanState* scan_state) :
      _scan_state(scan_state),
      _ct(G1CollectedHeap::heap()->card_table()),
      _merged_sparse(0),
      _merged_fine(0),
      _merged_coarse(0) { }

    void next_coarse_prt(uint const region_idx) {
      if (!remember_if_interesting(region_idx)) {
        return;
      }

      _merged_coarse++;

      size_t region_base_idx = (size_t)region_idx << HeapRegion::LogCardsPerRegion;
      _ct->mark_region_dirty(region_base_idx, HeapRegion::CardsPerRegion);
      _scan_state->set_chunk_region_dirty(region_base_idx);
    }

    void next_fine_prt(uint const region_idx, BitMap* bm) {
      if (!remember_if_interesting(region_idx)) {
        return;
      }

      _merged_fine++;

      size_t const region_base_idx = (size_t)region_idx << HeapRegion::LogCardsPerRegion;
      BitMap::idx_t cur = bm->get_next_one_offset(0);
      while (cur != bm->size()) {
        _ct->mark_clean_as_dirty(region_base_idx + cur);
        _scan_state->set_chunk_dirty(region_base_idx + cur);
        cur = bm->get_next_one_offset(cur + 1);
      }
    }

    void next_sparse_prt(uint const region_idx, SparsePRTEntry::card_elem_t* cards, uint const num_cards) {
      if (!remember_if_interesting(region_idx)) {
        return;
      }

      _merged_sparse++;

      size_t const region_base_idx = (size_t)region_idx << HeapRegion::LogCardsPerRegion;
      for (uint i = 0; i < num_cards; i++) {
        size_t card_idx = region_base_idx + cards[i];
        _ct->mark_clean_as_dirty(card_idx);
        _scan_state->set_chunk_dirty(card_idx);
      }
    }

    virtual bool do_heap_region(HeapRegion* r) {
      assert(r->in_collection_set() || r->is_starts_humongous(), "must be");

      HeapRegionRemSet* rem_set = r->rem_set();
      if (!rem_set->is_empty()) {
        rem_set->iterate_prts(*this);
      }

      return false;
    }

    size_t merged_sparse() const { return _merged_sparse; }
    size_t merged_fine() const { return _merged_fine; }
    size_t merged_coarse() const { return _merged_coarse; }
  };

  // Visitor for the remembered sets of humongous candidate regions to merge their
  // remembered set into the card table.
  class G1FlushHumongousCandidateRemSets : public HeapRegionClosure {
    G1MergeCardSetClosure _cl;

  public:
    G1FlushHumongousCandidateRemSets(G1RemSetScanState* scan_state) : _cl(scan_state) { }

    virtual bool do_heap_region(HeapRegion* r) {
      G1CollectedHeap* g1h = G1CollectedHeap::heap();

      if (!r->is_starts_humongous() ||
          !g1h->region_attr(r->hrm_index()).is_humongous() ||
          r->rem_set()->is_empty()) {
        return false;
      }

      guarantee(r->rem_set()->occupancy_less_or_equal_than(G1RSetSparseRegionEntries),
                "Found a not-small remembered set here. This is inconsistent with previous assumptions.");

      _cl.do_heap_region(r);

      // We should only clear the card based remembered set here as we will not
      // implicitly rebuild anything else during eager reclaim. Note that at the moment
      // (and probably never) we do not enter this path if there are other kind of
      // remembered sets for this region.
      r->rem_set()->clear_locked(true /* only_cardset */);
      // Clear_locked() above sets the state to Empty. However we want to continue
      // collecting remembered set entries for humongous regions that were not
      // reclaimed.
      r->rem_set()->set_state_complete();
#ifdef ASSERT
      G1HeapRegionAttr region_attr = g1h->region_attr(r->hrm_index());
      assert(region_attr.needs_remset_update(), "must be");
#endif
      assert(r->rem_set()->is_empty(), "At this point any humongous candidate remembered set must be empty.");

      return false;
    }

    size_t merged_sparse() const { return _cl.merged_sparse(); }
    size_t merged_fine() const { return _cl.merged_fine(); }
    size_t merged_coarse() const { return _cl.merged_coarse(); }
  };

  // Visitor for the log buffer entries to merge them into the card table.
  class G1MergeLogBufferCardsClosure : public G1CardTableEntryClosure {
    G1RemSetScanState* _scan_state;
    G1CardTable* _ct;

    size_t _cards_dirty;
    size_t _cards_skipped;
  public:
    G1MergeLogBufferCardsClosure(G1CollectedHeap* g1h, G1RemSetScanState* scan_state) :
      _scan_state(scan_state), _ct(g1h->card_table()), _cards_dirty(0), _cards_skipped(0)
    {}

    bool do_card_ptr(CardValue* card_ptr, uint worker_i) {
      // The only time we care about recording cards that
      // contain references that point into the collection set
      // is during RSet updating within an evacuation pause.
      // In this case worker_id should be the id of a GC worker thread.
      assert(SafepointSynchronize::is_at_safepoint(), "not during an evacuation pause");

      uint const region_idx = _ct->region_idx_for(card_ptr);

      // The second clause must come after - the log buffers might contain cards to uncommited
      // regions.
      // This code may count duplicate entries in the log buffers (even if rare) multiple
      // times.
      if (_scan_state->contains_cards_to_process(region_idx) && (*card_ptr == G1CardTable::dirty_card_val())) {
        _scan_state->add_dirty_region(region_idx);
        _scan_state->set_chunk_dirty(_ct->index_for_cardvalue(card_ptr));
        _cards_dirty++;
      } else {
        // We may have had dirty cards in the (initial) collection set (or the
        // young regions which are always in the initial collection set). We do
        // not fix their cards here: we already added these regions to the set of
        // regions to clear the card table at the end during the prepare() phase.
        _cards_skipped++;
      }
      return true;
    }

    size_t cards_dirty() const { return _cards_dirty; }
    size_t cards_skipped() const { return _cards_skipped; }
  };

  class G1MergeHeapRootsTask : public AbstractGangTask {
    HeapRegionClaimer _hr_claimer;
    G1RemSetScanState* _scan_state;
    bool _remembered_set_only;

    G1GCPhaseTimes::GCParPhases _merge_phase;

    volatile bool _fast_reclaim_handled;

  public:
    G1MergeHeapRootsTask(G1RemSetScanState* scan_state, uint num_workers, bool remembered_set_only, G1GCPhaseTimes::GCParPhases merge_phase) :
      AbstractGangTask("G1 Merge Heap Roots"),
      _hr_claimer(num_workers),
      _scan_state(scan_state),
      _remembered_set_only(remembered_set_only),
      _merge_phase(merge_phase),
      _fast_reclaim_handled(false) { }

    virtual void work(uint worker_id) {
      G1CollectedHeap* g1h = G1CollectedHeap::heap();
      G1GCPhaseTimes* p = g1h->phase_times();

      // We schedule flushing the remembered sets of humongous fast reclaim candidates
      // onto the card table first to allow the remaining parallelized tasks hide it.
      if (!_remembered_set_only &&
          p->fast_reclaim_humongous_candidates() > 0 &&
          !_fast_reclaim_handled &&
          !Atomic::cmpxchg(true, &_fast_reclaim_handled, false)) {

        G1FlushHumongousCandidateRemSets cl(_scan_state);
        g1h->heap_region_iterate(&cl);

        p->record_or_add_thread_work_item(_merge_phase, worker_id, cl.merged_sparse(), G1GCPhaseTimes::MergeRSMergedSparse);
        p->record_or_add_thread_work_item(_merge_phase, worker_id, cl.merged_fine(), G1GCPhaseTimes::MergeRSMergedFine);
        p->record_or_add_thread_work_item(_merge_phase, worker_id, cl.merged_coarse(), G1GCPhaseTimes::MergeRSMergedCoarse);
      }

      // Merge remembered sets of current candidates.
      {
        G1GCParPhaseTimesTracker x(p, _merge_phase, worker_id, !_remembered_set_only /* must_record */);
        G1MergeCardSetClosure cl(_scan_state);
        g1h->collection_set_iterate_increment_from(&cl, &_hr_claimer, worker_id);

        p->record_or_add_thread_work_item(_merge_phase, worker_id, cl.merged_sparse(), G1GCPhaseTimes::MergeRSMergedSparse);
        p->record_or_add_thread_work_item(_merge_phase, worker_id, cl.merged_fine(), G1GCPhaseTimes::MergeRSMergedFine);
        p->record_or_add_thread_work_item(_merge_phase, worker_id, cl.merged_coarse(), G1GCPhaseTimes::MergeRSMergedCoarse);
      }

      // Apply closure to log entries in the HCC.
      if (!_remembered_set_only && G1HotCardCache::default_use_cache()) {
        assert(_merge_phase == G1GCPhaseTimes::MergeRS, "Wrong merge phase");
        G1GCParPhaseTimesTracker x(p, G1GCPhaseTimes::MergeHCC, worker_id);
        G1MergeLogBufferCardsClosure cl(g1h, _scan_state);
        g1h->iterate_hcc_closure(&cl, worker_id);
      }

      // Now apply the closure to all remaining log entries.
      if (!_remembered_set_only) {
        assert(_merge_phase == G1GCPhaseTimes::MergeRS, "Wrong merge phase");
        G1GCParPhaseTimesTracker x(p, G1GCPhaseTimes::MergeLB, worker_id);

        G1MergeLogBufferCardsClosure cl(g1h, _scan_state);
        g1h->iterate_dirty_card_closure(&cl, worker_id);

        p->record_thread_work_item(G1GCPhaseTimes::MergeLB, worker_id, cl.cards_dirty(), G1GCPhaseTimes::MergeLBDirtyCards);
        p->record_thread_work_item(G1GCPhaseTimes::MergeLB, worker_id, cl.cards_skipped(), G1GCPhaseTimes::MergeLBSkippedCards);
      }
    }
  };

  // Creates a snapshot of the current _top values at the start of collection to
  // filter out card marks that we do not want to scan.
  class G1ResetScanTopClosure : public HeapRegionClosure {
    G1RemSetScanState* _scan_state;

  public:
    G1ResetScanTopClosure(G1RemSetScanState* scan_state) : _scan_state(scan_state) { }

    virtual bool do_heap_region(HeapRegion* r) {
      uint hrm_index = r->hrm_index();
      if (r->in_collection_set()) {
        // Young regions had their card table marked as young at their allocation;
        // we need to make sure that these marks are cleared at the end of GC, *but*
        // they should not be scanned for cards.
        // So directly add them to the "all_dirty_regions".
        // Same for regions in the (initial) collection set: they may contain cards from
        // the log buffers, make sure they are cleaned.
        _scan_state->add_all_dirty_region(hrm_index);
       } else if (r->is_old_or_humongous_or_archive()) {
        _scan_state->set_scan_top(hrm_index, r->top());
       }
       return false;
     }
  };
  // For each region, contains the maximum top() value to be used during this garbage
  // collection. Subsumes common checks like filtering out everything but old and
  // humongous regions outside the collection set.
  // This is valid because we are not interested in scanning stray remembered set
  // entries from free or archive regions.
  HeapWord** _scan_top;

  class G1ClearCardTableTask : public AbstractGangTask {
    G1CollectedHeap* _g1h;
    G1DirtyRegions* _regions;
    uint _chunk_length;

    uint volatile _cur_dirty_regions;

    G1RemSetScanState* _scan_state;

  public:
    G1ClearCardTableTask(G1CollectedHeap* g1h,
                         G1DirtyRegions* regions,
                         uint chunk_length,
                         G1RemSetScanState* scan_state) :
      AbstractGangTask("G1 Clear Card Table Task"),
      _g1h(g1h),
      _regions(regions),
      _chunk_length(chunk_length),
      _cur_dirty_regions(0),
      _scan_state(scan_state) {

      assert(chunk_length > 0, "must be");
    }

    static uint chunk_size() { return M; }

    void work(uint worker_id) {
      while (_cur_dirty_regions < _regions->size()) {
        uint next = Atomic::add(_chunk_length, &_cur_dirty_regions) - _chunk_length;
        uint max = MIN2(next + _chunk_length, _regions->size());

        for (uint i = next; i < max; i++) {
          HeapRegion* r = _g1h->region_at(_regions->at(i));
          if (!r->is_survivor()) {
            r->clear_cardtable();
          }
        }
      }
    }
  };

  // Clear the card table of "dirty" regions.
  void clear_card_table(WorkGang* workers) {
    uint num_regions = _all_dirty_regions->size();

    if (num_regions == 0) {
      return;
    }

    uint const num_chunks = (uint)(align_up((size_t)num_regions << HeapRegion::LogCardsPerRegion, G1ClearCardTableTask::chunk_size()) / G1ClearCardTableTask::chunk_size());
    uint const num_workers = MIN2(num_chunks, workers->active_workers());
    uint const chunk_length = G1ClearCardTableTask::chunk_size() / (uint)HeapRegion::CardsPerRegion;

    // Iterate over the dirty cards region list.
    G1ClearCardTableTask cl(G1CollectedHeap::heap(), _all_dirty_regions, chunk_length, this);

    log_debug(gc, ergo)("Running %s using %u workers for %u "
                        "units of work for %u regions.",
                        cl.name(), num_workers, num_chunks, num_regions);
    workers->run_task(&cl, num_workers);

#ifndef PRODUCT
    G1CollectedHeap::heap()->verifier()->verify_card_table_cleanup();
#endif
  }

public:
  G1RemSetScanState() :
    _max_regions(0),
    _collection_set_iter_state(NULL),
    _card_table_scan_state(NULL),
    _scan_chunks_per_region((uint)(HeapRegion::CardsPerRegion / CardsPerChunk)),
    _region_scan_chunks(NULL),
    _scan_chunks_shift(0),
    _all_dirty_regions(NULL),
    _next_dirty_regions(NULL),
    _scan_top(NULL) {
  }

  ~G1RemSetScanState() {
    FREE_C_HEAP_ARRAY(G1RemsetIterState, _collection_set_iter_state);
    FREE_C_HEAP_ARRAY(uint, _card_table_scan_state);
    FREE_C_HEAP_ARRAY(bool, _region_scan_chunks);
    FREE_C_HEAP_ARRAY(HeapWord*, _scan_top);
  }

  void initialize(size_t max_regions) {
    assert(_collection_set_iter_state == NULL, "Must not be initialized twice");
    _max_regions = max_regions;
    _collection_set_iter_state = NEW_C_HEAP_ARRAY(G1RemsetIterState, max_regions, mtGC);
    _card_table_scan_state = NEW_C_HEAP_ARRAY(uint, max_regions, mtGC);
    _region_scan_chunks = NEW_C_HEAP_ARRAY(bool, max_regions * _scan_chunks_per_region, mtGC);

    _scan_chunks_shift = (uint8_t)log2_intptr(HeapRegion::CardsPerRegion / _scan_chunks_per_region);
    _scan_top = NEW_C_HEAP_ARRAY(HeapWord*, max_regions, mtGC);
  }

  void prepare() {
    for (size_t i = 0; i < _max_regions; i++) {
      _collection_set_iter_state[i] = false;
      clear_scan_top((uint)i);
    }

    _all_dirty_regions = new G1DirtyRegions(_max_regions);

    G1ResetScanTopClosure cl(this);
    G1CollectedHeap::heap()->heap_region_iterate(&cl);

    _next_dirty_regions = new G1DirtyRegions(_max_regions);
  }

  void print_merge_heap_roots_stats() {
    size_t num_scan_chunks = 0;
    for (uint i = 0; i < _max_regions * _scan_chunks_per_region; i++) {
      if (_region_scan_chunks[i]) {
        num_scan_chunks++;
      }
    }
    size_t num_visited_cards = num_scan_chunks * CardsPerChunk;
    size_t total_dirty_region_cards = _next_dirty_regions->size() * HeapRegion::CardsPerRegion;

    G1CollectedHeap* g1h = G1CollectedHeap::heap();
    size_t total_old_region_cards =
      (g1h->num_regions() - (g1h->num_free_regions() - g1h->collection_set()->cur_length())) * HeapRegion::CardsPerRegion;

    log_debug(gc,remset)("Visited cards " SIZE_FORMAT " Total dirty " SIZE_FORMAT " (%.2lf%%) Total old " SIZE_FORMAT " (%.2lf%%)",
                         num_visited_cards,
                         total_dirty_region_cards,
                         percent_of(num_visited_cards, total_dirty_region_cards),
                         total_old_region_cards,
                         percent_of(num_visited_cards, total_old_region_cards));
  }

  void merge_heap_roots(WorkGang* workers, bool remembered_set_only, G1GCPhaseTimes::GCParPhases merge_phase) {
    {
      _all_dirty_regions->merge(_next_dirty_regions);
      _next_dirty_regions->reset();
      for (size_t i = 0; i < _max_regions; i++) {
        _card_table_scan_state[i] = 0;
      }

      ::memset(_region_scan_chunks, false, _max_regions * _scan_chunks_per_region * sizeof(*_region_scan_chunks));
    }

    size_t const increment_length = G1CollectedHeap::heap()->collection_set()->increment_length();

    uint const num_workers = !remembered_set_only ? workers->active_workers() :
                                                    MIN2(workers->active_workers(), (uint)increment_length);

    {
      G1MergeHeapRootsTask cl(this, num_workers, remembered_set_only, merge_phase);
      log_debug(gc, ergo)("Running %s using %u workers for " SIZE_FORMAT " regions",
                          cl.name(), num_workers, increment_length);
      workers->run_task(&cl, num_workers);
    }

    if (log_is_enabled(Debug, gc, remset)) {
      print_merge_heap_roots_stats();
    }
  }

  void set_chunk_region_dirty(size_t const region_card_idx) {
    size_t chunk_idx = region_card_idx >> _scan_chunks_shift;
    for (uint i = 0; i < _scan_chunks_per_region; i++) {
      _region_scan_chunks[chunk_idx++] = true;
    }
  }

  void set_chunk_dirty(size_t const card_idx) {
    assert((card_idx >> _scan_chunks_shift) < (_max_regions * _scan_chunks_per_region),
           "Trying to access index " SIZE_FORMAT " out of bounds " SIZE_FORMAT,
           card_idx >> _scan_chunks_shift, _max_regions * _scan_chunks_per_region);
    size_t const chunk_idx = card_idx >> _scan_chunks_shift;
    if (!_region_scan_chunks[chunk_idx]) {
      _region_scan_chunks[chunk_idx] = true;
    }
  }

  void cleanup(WorkGang* workers) {
    _all_dirty_regions->merge(_next_dirty_regions);

    clear_card_table(workers);

    delete _all_dirty_regions;
    _all_dirty_regions = NULL;

    delete _next_dirty_regions;
    _next_dirty_regions = NULL;
  }

  void iterate_dirty_regions_from(HeapRegionClosure* cl, uint worker_id) {
    uint num_regions = _next_dirty_regions->size();

    if (num_regions == 0) {
      return;
    }

    G1CollectedHeap* g1h = G1CollectedHeap::heap();

    WorkGang* workers = g1h->workers();
    uint const max_workers = workers->active_workers();

    uint const start_pos = num_regions * worker_id / max_workers;
    uint cur = start_pos;

    do {
      bool result = cl->do_heap_region(g1h->region_at(_next_dirty_regions->at(cur)));
      guarantee(!result, "Not allowed to ask for early termination.");
      cur++;
      if (cur == _next_dirty_regions->size()) {
        cur = 0;
      }
    } while (cur != start_pos);
  }

  // Attempt to claim the given region in the collection set for iteration. Returns true
  // if this call caused the transition from Unclaimed to Claimed.
  inline bool claim_collection_set_region(uint region) {
    assert(region < _max_regions, "Tried to access invalid region %u", region);
    if (_collection_set_iter_state[region]) {
      return false;
    }
    return !Atomic::cmpxchg(true, &_collection_set_iter_state[region], false);
  }

  bool has_cards_to_scan(uint region) {
    assert(region < _max_regions, "Tried to access invalid region %u", region);
    return _card_table_scan_state[region] < HeapRegion::CardsPerRegion;
  }

  uint claim_cards_to_scan(uint region, uint increment) {
    assert(region < _max_regions, "Tried to access invalid region %u", region);
    return Atomic::add(increment, &_card_table_scan_state[region]) - increment;
  }

  void add_dirty_region(uint const region) {
#ifdef ASSERT
   HeapRegion* hr = G1CollectedHeap::heap()->region_at(region);
   assert(!hr->in_collection_set() && hr->is_old_or_humongous_or_archive(),
          "Region %u is not suitable for scanning, is %sin collection set or %s",
          hr->hrm_index(), hr->in_collection_set() ? "" : "not ", hr->get_short_type_str());
#endif
    _next_dirty_regions->add_dirty_region(region);
  }

  void add_all_dirty_region(uint region) {
#ifdef ASSERT
    HeapRegion* hr = G1CollectedHeap::heap()->region_at(region);
    assert(hr->in_collection_set(),
           "Only add young regions to all dirty regions directly but %u is %s",
           hr->hrm_index(), hr->get_short_type_str());
#endif
    _all_dirty_regions->add_dirty_region(region);
  }

  void set_scan_top(uint region_idx, HeapWord* value) {
    _scan_top[region_idx] = value;
  }

  HeapWord* scan_top(uint region_idx) const {
    return _scan_top[region_idx];
  }

  void clear_scan_top(uint region_idx) {
    set_scan_top(region_idx, NULL);
  }
};

G1RemSet::G1RemSet(G1CollectedHeap* g1h,
                   G1CardTable* ct,
                   G1HotCardCache* hot_card_cache) :
  _scan_state(new G1RemSetScanState()),
  _prev_period_summary(),
  _g1h(g1h),
  _num_conc_refined_cards(0),
  _ct(ct),
  _g1p(_g1h->policy()),
  _hot_card_cache(hot_card_cache) {
}

G1RemSet::~G1RemSet() {
  delete _scan_state;
}

uint G1RemSet::num_par_rem_sets() {
  return G1DirtyCardQueueSet::num_par_ids() + G1ConcurrentRefine::max_num_threads() + MAX2(ConcGCThreads, ParallelGCThreads);
}

void G1RemSet::initialize(size_t capacity, uint max_regions) {
  G1FromCardCache::initialize(num_par_rem_sets(), max_regions);
  _scan_state->initialize(max_regions);
}

// Helper class to scan and detect ranges of cards that need to be scanned on the
// card table.
class G1CardTableScanner : public StackObj {
public:
  typedef CardTable::CardValue CardValue;

private:
  CardValue* const _base_addr;

  CardValue* _cur_addr;
  CardValue* const _end_addr;

  static const size_t ToScanMask = G1CardTable::g1_card_already_scanned;
  static const size_t ExpandedToScanMask = G1CardTable::WordAlreadyScanned;

  bool cur_addr_aligned() const {
    return ((uintptr_t)_cur_addr) % sizeof(size_t) == 0;
  }

  bool cur_card_is_dirty() const {
    CardValue value = *_cur_addr;
    return (value & ToScanMask) == 0;
  }

  bool cur_word_of_cards_contains_any_dirty_card() const {
    assert(cur_addr_aligned(), "Current address should be aligned");
    size_t const value = *(size_t*)_cur_addr;
    return (~value & ExpandedToScanMask) != 0;
  }

  bool cur_word_of_cards_all_dirty_cards() const {
    size_t const value = *(size_t*)_cur_addr;
    return value == G1CardTable::WordAllDirty;
  }

  size_t get_and_advance_pos() {
    _cur_addr++;
    return pointer_delta(_cur_addr, _base_addr, sizeof(CardValue)) - 1;
  }

public:
  G1CardTableScanner(CardValue* start_card, size_t size) :
    _base_addr(start_card),
    _cur_addr(start_card),
    _end_addr(start_card + size) {

    assert(is_aligned(start_card, sizeof(size_t)), "Unaligned start addr " PTR_FORMAT, p2i(start_card));
    assert(is_aligned(size, sizeof(size_t)), "Unaligned size " SIZE_FORMAT, size);
  }

  size_t find_next_dirty() {
    while (!cur_addr_aligned()) {
      if (cur_card_is_dirty()) {
        return get_and_advance_pos();
      }
      _cur_addr++;
    }

    assert(cur_addr_aligned(), "Current address should be aligned now.");
    while (_cur_addr != _end_addr) {
      if (cur_word_of_cards_contains_any_dirty_card()) {
        for (size_t i = 0; i < sizeof(size_t); i++) {
          if (cur_card_is_dirty()) {
            return get_and_advance_pos();
          }
          _cur_addr++;
        }
        assert(false, "Should not reach here given we detected a dirty card in the word.");
      }
      _cur_addr += sizeof(size_t);
    }
    return get_and_advance_pos();
  }

  size_t find_next_non_dirty() {
    assert(_cur_addr <= _end_addr, "Not allowed to search for marks after area.");

    while (!cur_addr_aligned()) {
      if (!cur_card_is_dirty()) {
        return get_and_advance_pos();
      }
      _cur_addr++;
    }

    assert(cur_addr_aligned(), "Current address should be aligned now.");
    while (_cur_addr != _end_addr) {
      if (!cur_word_of_cards_all_dirty_cards()) {
        for (size_t i = 0; i < sizeof(size_t); i++) {
          if (!cur_card_is_dirty()) {
            return get_and_advance_pos();
          }
          _cur_addr++;
        }
        assert(false, "Should not reach here given we detected a non-dirty card in the word.");
      }
      _cur_addr += sizeof(size_t);
    }
    return get_and_advance_pos();
  }
};

// Helper class to claim dirty chunks within the card table.
class G1CardTableChunkClaimer {
  G1RemSetScanState* _scan_state;
  uint _region_idx;
  uint _cur_claim;

public:
  G1CardTableChunkClaimer(G1RemSetScanState* scan_state, uint region_idx) :
    _scan_state(scan_state),
    _region_idx(region_idx),
    _cur_claim(0) {
    guarantee(size() <= HeapRegion::CardsPerRegion, "Should not claim more space than possible.");
  }

  bool has_next() {
    while (true) {
      _cur_claim = _scan_state->claim_cards_to_scan(_region_idx, size());
      if (_cur_claim >= HeapRegion::CardsPerRegion) {
        return false;
      }
      if (_scan_state->chunk_needs_scan(_region_idx, _cur_claim)) {
        return true;
      }
    }
  }

  uint value() const { return _cur_claim; }
  uint size() const { return _scan_state->scan_chunk_size(); }
};

// Scans a heap region for dirty cards.
class G1ScanHRForRegionClosure : public HeapRegionClosure {
  G1CollectedHeap* _g1h;
  G1CardTable* _ct;
  G1BlockOffsetTable* _bot;

  G1ParScanThreadState* _pss;

  G1RemSetScanState* _scan_state;

  G1GCPhaseTimes::GCParPhases _phase;

  uint   _worker_id;

  size_t _cards_scanned;
  size_t _blocks_scanned;
  size_t _chunks_claimed;

  Tickspan _rem_set_root_scan_time;
  Tickspan _rem_set_trim_partially_time;

  void scan_memregion(uint region_idx_for_card, MemRegion mr) {
    HeapRegion* const card_region = _g1h->region_at(region_idx_for_card);
    G1ScanCardClosure card_cl(_g1h, _pss);
    card_region->oops_on_card_seq_iterate_careful<true>(mr, &card_cl);
    _pss->trim_queue_partially();
  }

  void do_claimed_block(uint const region_idx_for_card, size_t const first_card, size_t const num_cards) {
    HeapWord* const card_start = _bot->address_for_index_raw(first_card);
#ifdef ASSERT
    HeapRegion* hr = _g1h->region_at_or_null(region_idx_for_card);
    assert(hr == NULL || hr->is_in_reserved(card_start),
             "Card start " PTR_FORMAT " to scan outside of region %u", p2i(card_start), _g1h->region_at(region_idx_for_card)->hrm_index());
#endif
    HeapWord* const top = _scan_state->scan_top(region_idx_for_card);
    if (card_start >= top) {
      return;
    }

    MemRegion mr(card_start, MIN2(card_start + ((size_t)num_cards << BOTConstants::LogN_words), top));
    scan_memregion(region_idx_for_card, mr);

    _cards_scanned += num_cards;
  }

  ALWAYSINLINE void do_card_block(uint const region_idx, size_t const first_card, size_t const num_cards) {
    _ct->mark_as_scanned(first_card, num_cards);
    do_claimed_block(region_idx, first_card, num_cards);
    _blocks_scanned++;
  }

   void scan_heap_roots(HeapRegion* r) {
    EventGCPhaseParallel event;
    uint const region_idx = r->hrm_index();

    ResourceMark rm;

    G1CardTableChunkClaimer claim(_scan_state, region_idx);

    while (claim.has_next()) {
      size_t const region_card_base_idx = ((size_t)region_idx << HeapRegion::LogCardsPerRegion) + claim.value();
      CardTable::CardValue* const base_addr = _ct->byte_for_index(region_card_base_idx);

      G1CardTableScanner scan(base_addr, claim.size());

      size_t first_scan_idx = scan.find_next_dirty();
      while (first_scan_idx != claim.size()) {
        assert(*_ct->byte_for_index(region_card_base_idx + first_scan_idx) <= 0x1, "is %d at region %u idx " SIZE_FORMAT, *_ct->byte_for_index(region_card_base_idx + first_scan_idx), region_idx, first_scan_idx);

        size_t const last_scan_idx = scan.find_next_non_dirty();
        size_t const len = last_scan_idx - first_scan_idx;

        do_card_block(region_idx, region_card_base_idx + first_scan_idx, len);

        if (last_scan_idx == claim.size()) {
          break;
        }

        first_scan_idx = scan.find_next_dirty();
      }
      _chunks_claimed++;
    }

    event.commit(GCId::current(), _worker_id, G1GCPhaseTimes::phase_name(G1GCPhaseTimes::ScanHR));
  }

public:
  G1ScanHRForRegionClosure(G1RemSetScanState* scan_state,
                           G1ParScanThreadState* pss,
                           uint worker_id,
                           G1GCPhaseTimes::GCParPhases phase) :
    _g1h(G1CollectedHeap::heap()),
    _ct(_g1h->card_table()),
    _bot(_g1h->bot()),
    _pss(pss),
    _scan_state(scan_state),
    _phase(phase),
    _worker_id(worker_id),
    _cards_scanned(0),
    _blocks_scanned(0),
    _chunks_claimed(0),
    _rem_set_root_scan_time(),
    _rem_set_trim_partially_time() {
  }

  bool do_heap_region(HeapRegion* r) {
    assert(!r->in_collection_set() && r->is_old_or_humongous_or_archive(),
           "Should only be called on old gen non-collection set regions but region %u is not.",
           r->hrm_index());
    uint const region_idx = r->hrm_index();

    if (_scan_state->has_cards_to_scan(region_idx)) {
      G1EvacPhaseWithTrimTimeTracker timer(_pss, _rem_set_root_scan_time, _rem_set_trim_partially_time);
      scan_heap_roots(r);
    }
    return false;
  }

  Tickspan rem_set_root_scan_time() const { return _rem_set_root_scan_time; }
  Tickspan rem_set_trim_partially_time() const { return _rem_set_trim_partially_time; }

  size_t cards_scanned() const { return _cards_scanned; }
  size_t blocks_scanned() const { return _blocks_scanned; }
  size_t chunks_claimed() const { return _chunks_claimed; }
};

void G1RemSet::scan_heap_roots(G1ParScanThreadState* pss,
                            uint worker_id,
                            G1GCPhaseTimes::GCParPhases scan_phase,
                            G1GCPhaseTimes::GCParPhases objcopy_phase) {
  G1ScanHRForRegionClosure cl(_scan_state, pss, worker_id, scan_phase);
  _scan_state->iterate_dirty_regions_from(&cl, worker_id);

  G1GCPhaseTimes* p = _g1p->phase_times();

  p->record_or_add_time_secs(objcopy_phase, worker_id, cl.rem_set_trim_partially_time().seconds());

  p->record_or_add_time_secs(scan_phase, worker_id, cl.rem_set_root_scan_time().seconds());
  p->record_or_add_thread_work_item(scan_phase, worker_id, cl.cards_scanned(), G1GCPhaseTimes::ScanHRScannedCards);
  p->record_or_add_thread_work_item(scan_phase, worker_id, cl.blocks_scanned(), G1GCPhaseTimes::ScanHRScannedBlocks);
  p->record_or_add_thread_work_item(scan_phase, worker_id, cl.chunks_claimed(), G1GCPhaseTimes::ScanHRClaimedChunks);
}

// Heap region closure to be applied to all regions in the current collection set
// increment to fix up non-card related roots.
class G1ScanCollectionSetRegionClosure : public HeapRegionClosure {
  G1ParScanThreadState* _pss;
  G1RemSetScanState* _scan_state;

  G1GCPhaseTimes::GCParPhases _scan_phase;
  G1GCPhaseTimes::GCParPhases _code_roots_phase;

  uint _worker_id;

  size_t _opt_refs_scanned;
  size_t _opt_refs_memory_used;

  Tickspan _strong_code_root_scan_time;
  Tickspan _strong_code_trim_partially_time;

  Tickspan _rem_set_opt_root_scan_time;
  Tickspan _rem_set_opt_trim_partially_time;

  void scan_opt_rem_set_roots(HeapRegion* r) {
    EventGCPhaseParallel event;

    G1OopStarChunkedList* opt_rem_set_list = _pss->oops_into_optional_region(r);

    G1ScanCardClosure scan_cl(G1CollectedHeap::heap(), _pss);
    G1ScanRSForOptionalClosure cl(G1CollectedHeap::heap(), &scan_cl);
    _opt_refs_scanned += opt_rem_set_list->oops_do(&cl, _pss->closures()->raw_strong_oops());
    _opt_refs_memory_used += opt_rem_set_list->used_memory();

    event.commit(GCId::current(), _worker_id, G1GCPhaseTimes::phase_name(_scan_phase));
  }

public:
  G1ScanCollectionSetRegionClosure(G1RemSetScanState* scan_state,
                                   G1ParScanThreadState* pss,
                                   uint worker_i,
                                   G1GCPhaseTimes::GCParPhases scan_phase,
                                   G1GCPhaseTimes::GCParPhases code_roots_phase) :
    _pss(pss),
    _scan_state(scan_state),
    _scan_phase(scan_phase),
    _code_roots_phase(code_roots_phase),
    _worker_id(worker_i),
    _opt_refs_scanned(0),
    _opt_refs_memory_used(0),
    _strong_code_root_scan_time(),
    _strong_code_trim_partially_time(),
    _rem_set_opt_root_scan_time(),
    _rem_set_opt_trim_partially_time() { }

  bool do_heap_region(HeapRegion* r) {
    uint const region_idx = r->hrm_index();

    // The individual references for the optional remembered set are per-worker, so we
    // always need to scan them.
    if (r->has_index_in_opt_cset()) {
      G1EvacPhaseWithTrimTimeTracker timer(_pss, _rem_set_opt_root_scan_time, _rem_set_opt_trim_partially_time);
      scan_opt_rem_set_roots(r);
    }

    if (_scan_state->claim_collection_set_region(region_idx)) {
      EventGCPhaseParallel event;

      G1EvacPhaseWithTrimTimeTracker timer(_pss, _strong_code_root_scan_time, _strong_code_trim_partially_time);
      // Scan the strong code root list attached to the current region
      r->strong_code_roots_do(_pss->closures()->weak_codeblobs());

      event.commit(GCId::current(), _worker_id, G1GCPhaseTimes::phase_name(_code_roots_phase));
    }

    return false;
  }

  Tickspan strong_code_root_scan_time() const { return _strong_code_root_scan_time;  }
  Tickspan strong_code_root_trim_partially_time() const { return _strong_code_trim_partially_time; }

  Tickspan rem_set_opt_root_scan_time() const { return _rem_set_opt_root_scan_time; }
  Tickspan rem_set_opt_trim_partially_time() const { return _rem_set_opt_trim_partially_time; }

  size_t opt_refs_scanned() const { return _opt_refs_scanned; }
  size_t opt_refs_memory_used() const { return _opt_refs_memory_used; }
};

void G1RemSet::scan_collection_set_regions(G1ParScanThreadState* pss,
                                           uint worker_id,
                                           G1GCPhaseTimes::GCParPhases scan_phase,
                                           G1GCPhaseTimes::GCParPhases coderoots_phase,
                                           G1GCPhaseTimes::GCParPhases objcopy_phase) {
  G1ScanCollectionSetRegionClosure cl(_scan_state, pss, worker_id, scan_phase, coderoots_phase);
  _g1h->collection_set_iterate_increment_from(&cl, worker_id);

  G1GCPhaseTimes* p = _g1h->phase_times();

  p->record_or_add_time_secs(scan_phase, worker_id, cl.rem_set_opt_root_scan_time().seconds());
  p->record_or_add_time_secs(scan_phase, worker_id, cl.rem_set_opt_trim_partially_time().seconds());

  p->record_or_add_time_secs(coderoots_phase, worker_id, cl.strong_code_root_scan_time().seconds());
  p->add_time_secs(objcopy_phase, worker_id, cl.strong_code_root_trim_partially_time().seconds());

  // At this time we record some metrics only for the evacuations after the initial one.
  if (scan_phase == G1GCPhaseTimes::OptScanHR) {
    p->record_or_add_thread_work_item(scan_phase, worker_id, cl.opt_refs_scanned(), G1GCPhaseTimes::ScanHRScannedOptRefs);
    p->record_or_add_thread_work_item(scan_phase, worker_id, cl.opt_refs_memory_used(), G1GCPhaseTimes::ScanHRUsedMemory);
  }
}

void G1RemSet::prepare_for_scan_heap_roots() {
  G1DirtyCardQueueSet& dcqs = G1BarrierSet::dirty_card_queue_set();
  dcqs.concatenate_logs();

  _scan_state->prepare();
}

void G1RemSet::merge_heap_roots(bool remembered_set_only, G1GCPhaseTimes::GCParPhases merge_phase) {
  _scan_state->merge_heap_roots(_g1h->workers(), remembered_set_only, merge_phase);
}

void G1RemSet::prepare_for_scan_heap_roots(uint region_idx) {
  _scan_state->clear_scan_top(region_idx);
}

void G1RemSet::cleanup_after_scan_heap_roots() {
  G1GCPhaseTimes* phase_times = _g1h->phase_times();

  // Set all cards back to clean.
  double start = os::elapsedTime();
  _scan_state->cleanup(_g1h->workers());
  phase_times->record_clear_ct_time((os::elapsedTime() - start) * 1000.0);
}

inline void check_card_ptr(CardTable::CardValue* card_ptr, G1CardTable* ct) {
#ifdef ASSERT
  G1CollectedHeap* g1h = G1CollectedHeap::heap();
  assert(g1h->is_in_exact(ct->addr_for(card_ptr)),
         "Card at " PTR_FORMAT " index " SIZE_FORMAT " representing heap at " PTR_FORMAT " (%u) must be in committed heap",
         p2i(card_ptr),
         ct->index_for(ct->addr_for(card_ptr)),
         p2i(ct->addr_for(card_ptr)),
         g1h->addr_to_region(ct->addr_for(card_ptr)));
#endif
}

void G1RemSet::refine_card_concurrently(CardValue* card_ptr,
                                        uint worker_i) {
  assert(!_g1h->is_gc_active(), "Only call concurrently");

  // Construct the region representing the card.
  HeapWord* start = _ct->addr_for(card_ptr);
  // And find the region containing it.
  HeapRegion* r = _g1h->heap_region_containing_or_null(start);

  // If this is a (stale) card into an uncommitted region, exit.
  if (r == NULL) {
    return;
  }

  check_card_ptr(card_ptr, _ct);

  // If the card is no longer dirty, nothing to do.
  if (*card_ptr != G1CardTable::dirty_card_val()) {
    return;
  }

  // This check is needed for some uncommon cases where we should
  // ignore the card.
  //
  // The region could be young.  Cards for young regions are
  // distinctly marked (set to g1_young_gen), so the post-barrier will
  // filter them out.  However, that marking is performed
  // concurrently.  A write to a young object could occur before the
  // card has been marked young, slipping past the filter.
  //
  // The card could be stale, because the region has been freed since
  // the card was recorded. In this case the region type could be
  // anything.  If (still) free or (reallocated) young, just ignore
  // it.  If (reallocated) old or humongous, the later card trimming
  // and additional checks in iteration may detect staleness.  At
  // worst, we end up processing a stale card unnecessarily.
  //
  // In the normal (non-stale) case, the synchronization between the
  // enqueueing of the card and processing it here will have ensured
  // we see the up-to-date region type here.
  if (!r->is_old_or_humongous_or_archive()) {
    return;
  }

  // The result from the hot card cache insert call is either:
  //   * pointer to the current card
  //     (implying that the current card is not 'hot'),
  //   * null
  //     (meaning we had inserted the card ptr into the "hot" card cache,
  //     which had some headroom),
  //   * a pointer to a "hot" card that was evicted from the "hot" cache.
  //

  if (_hot_card_cache->use_cache()) {
    assert(!SafepointSynchronize::is_at_safepoint(), "sanity");

    const CardValue* orig_card_ptr = card_ptr;
    card_ptr = _hot_card_cache->insert(card_ptr);
    if (card_ptr == NULL) {
      // There was no eviction. Nothing to do.
      return;
    } else if (card_ptr != orig_card_ptr) {
      // Original card was inserted and an old card was evicted.
      start = _ct->addr_for(card_ptr);
      r = _g1h->heap_region_containing(start);

      // Check whether the region formerly in the cache should be
      // ignored, as discussed earlier for the original card.  The
      // region could have been freed while in the cache.
      if (!r->is_old_or_humongous_or_archive()) {
        return;
      }
    } // Else we still have the original card.
  }

  // Trim the region designated by the card to what's been allocated
  // in the region.  The card could be stale, or the card could cover
  // (part of) an object at the end of the allocated space and extend
  // beyond the end of allocation.

  // Non-humongous objects are only allocated in the old-gen during
  // GC, so if region is old then top is stable.  Humongous object
  // allocation sets top last; if top has not yet been set, this is
  // a stale card and we'll end up with an empty intersection.  If
  // this is not a stale card, the synchronization between the
  // enqueuing of the card and processing it here will have ensured
  // we see the up-to-date top here.
  HeapWord* scan_limit = r->top();

  if (scan_limit <= start) {
    // If the trimmed region is empty, the card must be stale.
    return;
  }

  // Okay to clean and process the card now.  There are still some
  // stale card cases that may be detected by iteration and dealt with
  // as iteration failure.
  *const_cast<volatile CardValue*>(card_ptr) = G1CardTable::clean_card_val();

  // This fence serves two purposes.  First, the card must be cleaned
  // before processing the contents.  Second, we can't proceed with
  // processing until after the read of top, for synchronization with
  // possibly concurrent humongous object allocation.  It's okay that
  // reading top and reading type were racy wrto each other.  We need
  // both set, in any order, to proceed.
  OrderAccess::fence();

  // Don't use addr_for(card_ptr + 1) which can ask for
  // a card beyond the heap.
  HeapWord* end = start + G1CardTable::card_size_in_words;
  MemRegion dirty_region(start, MIN2(scan_limit, end));
  assert(!dirty_region.is_empty(), "sanity");

  G1ConcurrentRefineOopClosure conc_refine_cl(_g1h, worker_i);
  if (r->oops_on_card_seq_iterate_careful<false>(dirty_region, &conc_refine_cl)) {
    _num_conc_refined_cards++; // Unsynchronized update, only used for logging.
    return;
  }

  // If unable to process the card then we encountered an unparsable
  // part of the heap (e.g. a partially allocated object, so only
  // temporarily a problem) while processing a stale card.  Despite
  // the card being stale, we can't simply ignore it, because we've
  // already marked the card cleaned, so taken responsibility for
  // ensuring the card gets scanned.
  //
  // However, the card might have gotten re-dirtied and re-enqueued
  // while we worked.  (In fact, it's pretty likely.)
  if (*card_ptr == G1CardTable::dirty_card_val()) {
    return;
  }

  // Re-dirty the card and enqueue in the *shared* queue.  Can't use
  // the thread-local queue, because that might be the queue that is
  // being processed by us; we could be a Java thread conscripted to
  // perform refinement on our queue's current buffer.
  *card_ptr = G1CardTable::dirty_card_val();
  G1BarrierSet::shared_dirty_card_queue().enqueue(card_ptr);
}

void G1RemSet::print_periodic_summary_info(const char* header, uint period_count) {
  if ((G1SummarizeRSetStatsPeriod > 0) && log_is_enabled(Trace, gc, remset) &&
      (period_count % G1SummarizeRSetStatsPeriod == 0)) {

    G1RemSetSummary current(this);
    _prev_period_summary.subtract_from(&current);

    Log(gc, remset) log;
    log.trace("%s", header);
    ResourceMark rm;
    LogStream ls(log.trace());
    _prev_period_summary.print_on(&ls);

    _prev_period_summary.set(&current);
  }
}

void G1RemSet::print_summary_info() {
  Log(gc, remset, exit) log;
  if (log.is_trace()) {
    log.trace(" Cumulative RS summary");
    G1RemSetSummary current(this);
    ResourceMark rm;
    LogStream ls(log.trace());
    current.print_on(&ls);
  }
}

class G1RebuildRemSetTask: public AbstractGangTask {
  // Aggregate the counting data that was constructed concurrently
  // with marking.
  class G1RebuildRemSetHeapRegionClosure : public HeapRegionClosure {
    G1ConcurrentMark* _cm;
    G1RebuildRemSetClosure _update_cl;

    // Applies _update_cl to the references of the given object, limiting objArrays
    // to the given MemRegion. Returns the amount of words actually scanned.
    size_t scan_for_references(oop const obj, MemRegion mr) {
      size_t const obj_size = obj->size();
      // All non-objArrays and objArrays completely within the mr
      // can be scanned without passing the mr.
      if (!obj->is_objArray() || mr.contains(MemRegion((HeapWord*)obj, obj_size))) {
        obj->oop_iterate(&_update_cl);
        return obj_size;
      }
      // This path is for objArrays crossing the given MemRegion. Only scan the
      // area within the MemRegion.
      obj->oop_iterate(&_update_cl, mr);
      return mr.intersection(MemRegion((HeapWord*)obj, obj_size)).word_size();
    }

    // A humongous object is live (with respect to the scanning) either
    // a) it is marked on the bitmap as such
    // b) its TARS is larger than TAMS, i.e. has been allocated during marking.
    bool is_humongous_live(oop const humongous_obj, const G1CMBitMap* const bitmap, HeapWord* tams, HeapWord* tars) const {
      return bitmap->is_marked(humongous_obj) || (tars > tams);
    }

    // Iterator over the live objects within the given MemRegion.
    class LiveObjIterator : public StackObj {
      const G1CMBitMap* const _bitmap;
      const HeapWord* _tams;
      const MemRegion _mr;
      HeapWord* _current;

      bool is_below_tams() const {
        return _current < _tams;
      }

      bool is_live(HeapWord* obj) const {
        return !is_below_tams() || _bitmap->is_marked(obj);
      }

      HeapWord* bitmap_limit() const {
        return MIN2(const_cast<HeapWord*>(_tams), _mr.end());
      }

      void move_if_below_tams() {
        if (is_below_tams() && has_next()) {
          _current = _bitmap->get_next_marked_addr(_current, bitmap_limit());
        }
      }
    public:
      LiveObjIterator(const G1CMBitMap* const bitmap, const HeapWord* tams, const MemRegion mr, HeapWord* first_oop_into_mr) :
          _bitmap(bitmap),
          _tams(tams),
          _mr(mr),
          _current(first_oop_into_mr) {

        assert(_current <= _mr.start(),
               "First oop " PTR_FORMAT " should extend into mr [" PTR_FORMAT ", " PTR_FORMAT ")",
               p2i(first_oop_into_mr), p2i(mr.start()), p2i(mr.end()));

        // Step to the next live object within the MemRegion if needed.
        if (is_live(_current)) {
          // Non-objArrays were scanned by the previous part of that region.
          if (_current < mr.start() && !oop(_current)->is_objArray()) {
            _current += oop(_current)->size();
            // We might have positioned _current on a non-live object. Reposition to the next
            // live one if needed.
            move_if_below_tams();
          }
        } else {
          // The object at _current can only be dead if below TAMS, so we can use the bitmap.
          // immediately.
          _current = _bitmap->get_next_marked_addr(_current, bitmap_limit());
          assert(_current == _mr.end() || is_live(_current),
                 "Current " PTR_FORMAT " should be live (%s) or beyond the end of the MemRegion (" PTR_FORMAT ")",
                 p2i(_current), BOOL_TO_STR(is_live(_current)), p2i(_mr.end()));
        }
      }

      void move_to_next() {
        _current += next()->size();
        move_if_below_tams();
      }

      oop next() const {
        oop result = oop(_current);
        assert(is_live(_current),
               "Object " PTR_FORMAT " must be live TAMS " PTR_FORMAT " below %d mr " PTR_FORMAT " " PTR_FORMAT " outside %d",
               p2i(_current), p2i(_tams), _tams > _current, p2i(_mr.start()), p2i(_mr.end()), _mr.contains(result));
        return result;
      }

      bool has_next() const {
        return _current < _mr.end();
      }
    };

    // Rebuild remembered sets in the part of the region specified by mr and hr.
    // Objects between the bottom of the region and the TAMS are checked for liveness
    // using the given bitmap. Objects between TAMS and TARS are assumed to be live.
    // Returns the number of live words between bottom and TAMS.
    size_t rebuild_rem_set_in_region(const G1CMBitMap* const bitmap,
                                     HeapWord* const top_at_mark_start,
                                     HeapWord* const top_at_rebuild_start,
                                     HeapRegion* hr,
                                     MemRegion mr) {
      size_t marked_words = 0;

      if (hr->is_humongous()) {
        oop const humongous_obj = oop(hr->humongous_start_region()->bottom());
        if (is_humongous_live(humongous_obj, bitmap, top_at_mark_start, top_at_rebuild_start)) {
          // We need to scan both [bottom, TAMS) and [TAMS, top_at_rebuild_start);
          // however in case of humongous objects it is sufficient to scan the encompassing
          // area (top_at_rebuild_start is always larger or equal to TAMS) as one of the
          // two areas will be zero sized. I.e. TAMS is either
          // the same as bottom or top(_at_rebuild_start). There is no way TAMS has a different
          // value: this would mean that TAMS points somewhere into the object.
          assert(hr->top() == top_at_mark_start || hr->top() == top_at_rebuild_start,
                 "More than one object in the humongous region?");
          humongous_obj->oop_iterate(&_update_cl, mr);
          return top_at_mark_start != hr->bottom() ? mr.intersection(MemRegion((HeapWord*)humongous_obj, humongous_obj->size())).byte_size() : 0;
        } else {
          return 0;
        }
      }

      for (LiveObjIterator it(bitmap, top_at_mark_start, mr, hr->block_start(mr.start())); it.has_next(); it.move_to_next()) {
        oop obj = it.next();
        size_t scanned_size = scan_for_references(obj, mr);
        if ((HeapWord*)obj < top_at_mark_start) {
          marked_words += scanned_size;
        }
      }

      return marked_words * HeapWordSize;
    }
public:
  G1RebuildRemSetHeapRegionClosure(G1CollectedHeap* g1h,
                                   G1ConcurrentMark* cm,
                                   uint worker_id) :
    HeapRegionClosure(),
    _cm(cm),
    _update_cl(g1h, worker_id) { }

    bool do_heap_region(HeapRegion* hr) {
      if (_cm->has_aborted()) {
        return true;
      }

      uint const region_idx = hr->hrm_index();
      DEBUG_ONLY(HeapWord* const top_at_rebuild_start_check = _cm->top_at_rebuild_start(region_idx);)
      assert(top_at_rebuild_start_check == NULL ||
             top_at_rebuild_start_check > hr->bottom(),
             "A TARS (" PTR_FORMAT ") == bottom() (" PTR_FORMAT ") indicates the old region %u is empty (%s)",
             p2i(top_at_rebuild_start_check), p2i(hr->bottom()),  region_idx, hr->get_type_str());

      size_t total_marked_bytes = 0;
      size_t const chunk_size_in_words = G1RebuildRemSetChunkSize / HeapWordSize;

      HeapWord* const top_at_mark_start = hr->prev_top_at_mark_start();

      HeapWord* cur = hr->bottom();
      while (cur < hr->end()) {
        // After every iteration (yield point) we need to check whether the region's
        // TARS changed due to e.g. eager reclaim.
        HeapWord* const top_at_rebuild_start = _cm->top_at_rebuild_start(region_idx);
        if (top_at_rebuild_start == NULL) {
          return false;
        }

        MemRegion next_chunk = MemRegion(hr->bottom(), top_at_rebuild_start).intersection(MemRegion(cur, chunk_size_in_words));
        if (next_chunk.is_empty()) {
          break;
        }

        const Ticks start = Ticks::now();
        size_t marked_bytes = rebuild_rem_set_in_region(_cm->prev_mark_bitmap(),
                                                        top_at_mark_start,
                                                        top_at_rebuild_start,
                                                        hr,
                                                        next_chunk);
        Tickspan time = Ticks::now() - start;

        log_trace(gc, remset, tracking)("Rebuilt region %u "
                                        "live " SIZE_FORMAT " "
                                        "time %.3fms "
                                        "marked bytes " SIZE_FORMAT " "
                                        "bot " PTR_FORMAT " "
                                        "TAMS " PTR_FORMAT " "
                                        "TARS " PTR_FORMAT,
                                        region_idx,
                                        _cm->liveness(region_idx) * HeapWordSize,
                                        time.seconds() * 1000.0,
                                        marked_bytes,
                                        p2i(hr->bottom()),
                                        p2i(top_at_mark_start),
                                        p2i(top_at_rebuild_start));

        if (marked_bytes > 0) {
          total_marked_bytes += marked_bytes;
        }
        cur += chunk_size_in_words;

        _cm->do_yield_check();
        if (_cm->has_aborted()) {
          return true;
        }
      }
      // In the final iteration of the loop the region might have been eagerly reclaimed.
      // Simply filter out those regions. We can not just use region type because there
      // might have already been new allocations into these regions.
      DEBUG_ONLY(HeapWord* const top_at_rebuild_start = _cm->top_at_rebuild_start(region_idx);)
      assert(top_at_rebuild_start == NULL ||
             total_marked_bytes == hr->marked_bytes(),
             "Marked bytes " SIZE_FORMAT " for region %u (%s) in [bottom, TAMS) do not match calculated marked bytes " SIZE_FORMAT " "
             "(" PTR_FORMAT " " PTR_FORMAT " " PTR_FORMAT ")",
             total_marked_bytes, hr->hrm_index(), hr->get_type_str(), hr->marked_bytes(),
             p2i(hr->bottom()), p2i(top_at_mark_start), p2i(top_at_rebuild_start));
       // Abort state may have changed after the yield check.
      return _cm->has_aborted();
    }
  };

  HeapRegionClaimer _hr_claimer;
  G1ConcurrentMark* _cm;

  uint _worker_id_offset;
public:
  G1RebuildRemSetTask(G1ConcurrentMark* cm,
                      uint n_workers,
                      uint worker_id_offset) :
      AbstractGangTask("G1 Rebuild Remembered Set"),
      _hr_claimer(n_workers),
      _cm(cm),
      _worker_id_offset(worker_id_offset) {
  }

  void work(uint worker_id) {
    SuspendibleThreadSetJoiner sts_join;

    G1CollectedHeap* g1h = G1CollectedHeap::heap();

    G1RebuildRemSetHeapRegionClosure cl(g1h, _cm, _worker_id_offset + worker_id);
    g1h->heap_region_par_iterate_from_worker_offset(&cl, &_hr_claimer, worker_id);
  }
};

void G1RemSet::rebuild_rem_set(G1ConcurrentMark* cm,
                               WorkGang* workers,
                               uint worker_id_offset) {
  uint num_workers = workers->active_workers();

  G1RebuildRemSetTask cl(cm,
                         num_workers,
                         worker_id_offset);
  workers->run_task(&cl, num_workers);
}